From 81328b654027b405284e80df47c974d95a196821 Mon Sep 17 00:00:00 2001 From: suguanYang Date: Tue, 16 Jun 2026 03:53:20 +0000 Subject: [PATCH 1/4] Add official library demo catalog --- .../chunks.json | 12865 +++++ .../doc_nav.json | 1581 + .../full.md | 1290 + .../images/image-1-Alibaba.jpg | Bin 0 -> 4916 bytes .../images/image-2-### (vii) .jpg | Bin 0 -> 3619 bytes .../manifest.json | 259 + .../tables/table-0 Financial Results.html | 1 + .../tables/table-1 Alibaba Revenue.html | 1 + .../tables/table-10 Board of Directors.html | 1 + .../tables/table-11 Shareholding Table.html | 1 + .../tables/table-12 Shareholder Holdings.html | 1 + .../tables/table-13 RSU Grant Summary.html | 1 + .../tables/table-14 Share Options Table.html | 1 + .../tables/table-15 RSU Activity Report.html | 1 + .../tables/table-16 RSU Grant Details.html | 1 + .../tables/table-17 Share Repurchase.html | 1 + .../tables/table-18 Notes Proceeds Use.html | 1 + .../tables/table-19 Interim Report.html | 1 + .../tables/table-2 Alibaba EBITA.html | 1 + .../tables/table-20 Alibaba Financials.html | 1 + .../tables/table-21 Income Statement.html | 1 + .../tables/table-22 Balance Sheet Data.html | 1 + .../tables/table-23 Equity Summary.html | 1 + .../tables/table-24 Equity Changes.html | 1 + .../tables/table-25 Cash Flow Statement.html | 1 + .../tables/table-26 Financial Data.html | 1 + .../tables/table-27 Alibaba Revenue.html | 1 + .../tables/table-28 Revenue Breakdown.html | 1 + .../tables/table-29 Tax Expense Data.html | 1 + .../tables/table-3 Costs & Expenses.html | 1 + .../tables/table-30 Cost Breakdown.html | 1 + .../tables/table-31 EPS Data Table.html | 1 + .../tables/table-32 Investment Summary.html | 1 + .../tables/table-33 Fair Value Table.html | 1 + .../tables/table-34 Fair Value Table.html | 1 + .../tables/table-35 Balance Change.html | 1 + .../tables/table-36 Fair Value Change.html | 1 + .../tables/table-37 Intangible Assets.html | 1 + .../table-38 Alibaba Segment Assets.html | 1 + .../table-39 Financial Liabilities.html | 1 + .../table-4 Share-based Compensation.html | 1 + .../tables/table-40 Cloud Revenue Growth.html | 1 + .../table-41 Alibaba Segment Results.html | 1 + .../tables/table-42 Financial Results.html | 1 + .../table-43 Depreciation by Segment.html | 1 + .../tables/table-44 Accounts Receivable.html | 1 + .../tables/table-45 Accounts Payable.html | 1 + .../tables/table-46 IFRS Adjustments.html | 1 + .../tables/table-47 GAAP to IFRS.html | 1 + .../tables/table-48 GAAP to IFRS.html | 1 + .../table-49 GAAP to IFRS Reconciliation.html | 1 + .../table-5 Equity Investee Results.html | 1 + .../tables/table-6 Financial Results.html | 1 + .../tables/table-7 Non-GAAP Net Income.html | 1 + .../tables/table-8 Earnings per Share.html | 1 + .../tables/table-9 Free Cash Flow.html | 1 + .../toc_hierarchies.json | 23 + .../financial-goog-10-k-2025/chunks.json | 20355 ++++++++ .../financial-goog-10-k-2025/doc_nav.json | 2169 + .../financial-goog-10-k-2025/full.md | 2198 + .../image-1-Stock Performance Comparison.jpg | Bin 0 -> 90564 bytes ...ock Performance Comparison (2020-2025).jpg | Bin 0 -> 89187 bytes .../financial-goog-10-k-2025/manifest.json | 325 + .../tables/table-0 Stock & Note Listings.html | 1 + .../tables/table-1 Filer Status.html | 1 + .../tables/table-10 R&D Expenses.html | 1 + .../tables/table-11 Sales Expenses.html | 1 + .../tables/table-12 Expense Analysis.html | 1 + .../tables/table-13 Alphabet Income.html | 1 + .../tables/table-14 Other Income_Expense.html | 1 + .../tables/table-15 Tax Data Table.html | 1 + .../tables/table-16 Cash Flow Table.html | 1 + .../table-17 Foreign Currency Risk.html | 1 + .../tables/table-18 Interest Rate Risk.html | 1 + .../tables/table-19 Financial Index.html | 1 + .../tables/table-2 Share Repurchase.html | 1 + .../tables/table-20 Loss Contingencies.html | 1 + .../tables/table-21 Balance Sheet.html | 1 + .../tables/table-22 Financial Data.html | 1 + .../tables/table-23 Comprehensive Income.html | 1 + .../tables/table-24 Stockholders' Equity.html | 1 + .../tables/table-25 Cash Flow Statement.html | 1 + .../tables/table-26 Google Revenue.html | 1 + .../tables/table-27 Revenue Breakdown.html | 1 + .../tables/table-28 Fair Value Table.html | 1 + .../tables/table-29 Fair Value Table.html | 1 + .../table-3 Alphabet Stock vs Indexes.html | 160 + .../tables/table-30 Debt Maturity.html | 1 + .../tables/table-31 Investment Summary.html | 1 + .../tables/table-32 Investment Summary.html | 1 + .../table-33 Investment Securities.html | 1 + .../table-34 Investment Securities.html | 1 + .../tables/table-35 Debt Securities P&L.html | 1 + .../table-36 Non-marketable securities.html | 1 + .../tables/table-37 Equity Securities.html | 1 + .../tables/table-38 Derivatives Table.html | 1 + .../tables/table-39 Derivatives Table.html | 1 + .../tables/table-4 Alphabet vs S&P 500.html | 160 + .../tables/table-40 Hedge Gains_Losses.html | 1 + .../tables/table-41 Hedge Effects.html | 1 + .../table-42 Derivatives Offsetting.html | 1 + .../tables/table-43 Lease Cost Table.html | 1 + .../tables/table-44 Lease Data.html | 1 + .../tables/table-45 Lease Cash Flows.html | 1 + .../tables/table-46 Lease Payments.html | 1 + .../tables/table-47 Debt Table.html | 1 + .../tables/table-48 Debt Maturity.html | 1 + .../tables/table-49 Property & Equipment.html | 1 + .../tables/table-5 Financial Results.html | 1 + .../tables/table-50 Accrued Liabilities.html | 1 + .../tables/table-51 OCI Rollforward.html | 1 + .../tables/table-52 AOCI Reclass.html | 1 + .../tables/table-53 Income Data Table.html | 1 + .../table-54 Google Segment Balances.html | 1 + .../tables/table-55 Share Repurchases.html | 1 + .../tables/table-56 EPS Breakdown.html | 1 + .../tables/table-57 EPS Calculation.html | 1 + .../tables/table-58 EPS Calculation.html | 1 + .../tables/table-59 Stock Vesting.html | 1 + .../tables/table-6 Revenue Breakdown.html | 1 + .../tables/table-60 Revenue Breakdown.html | 1 + .../tables/table-61 Income Tax Provision.html | 1 + .../table-62 Tax Rate Reconciliation.html | 1 + .../tables/table-63 Deferred Tax Assets.html | 1 + .../tables/table-64 Income Tax Payments.html | 1 + .../tables/table-65 Tax Benefits.html | 1 + .../tables/table-66 Alphabet Revenue.html | 1 + .../tables/table-67 Long-lived Assets.html | 1 + .../tables/table-68 Warranty Reserves.html | 1 + .../tables/table-69 Exhibit List.html | 1 + .../tables/table-7 Google Ad Changes.html | 1 + .../tables/table-70 CEO Signature.html | 1 + .../tables/table-71 Signatures.html | 1 + .../tables/table-8 Revenue by Region.html | 1 + .../tables/table-9 Cost of Revenues.html | 1 + .../toc_hierarchies.json | 19 + .../chunks.json | 2525 + .../doc_nav.json | 119 + .../full.md | 419 + .../manifest.json | 64 + .../tables/table-0 AI Compute Plans.html | 1 + .../chunks.json | 2115 + .../doc_nav.json | 306 + .../full.md | 420 + .../image-1-Revenue Forecast by Region.jpg | Bin 0 -> 157397 bytes .../image-10-Quarterly Percentage Growth.jpg | Bin 0 -> 20956 bytes .../image-11-Quarterly Growth Trend.jpg | Bin 0 -> 23122 bytes .../image-12-Quarterly Growth Trend.jpg | Bin 0 -> 27098 bytes .../image-13-Growth Trend 2025-2026.jpg | Bin 0 -> 23156 bytes ...-Quarterly Percentage Data (2025-2026).jpg | Bin 0 -> 22947 bytes ...-Quarterly Percentage Data (2025-2026).jpg | Bin 0 -> 22700 bytes .../image-16-Quarterly Percentage Trends.jpg | Bin 0 -> 24735 bytes .../image-17-Quarterly Growth Trend.jpg | Bin 0 -> 20182 bytes ...-Quarterly Percentage Data (2025-2026).jpg | Bin 0 -> 25843 bytes ...evenue Breakdown by Region and Quarter.jpg | Bin 0 -> 154959 bytes ...image-3-Quarterly Percentage Breakdown.jpg | Bin 0 -> 117606 bytes ...mage-4-Quarterly Financial Performance.jpg | Bin 0 -> 84470 bytes .../image-5-Quarterly Financial Growth.jpg | Bin 0 -> 78096 bytes .../image-6-Financial Growth Projection.jpg | Bin 0 -> 56358 bytes .../images/image-7-Quarterly Data Trend.jpg | Bin 0 -> 82491 bytes .../image-8-Quarterly Financial Data.jpg | Bin 0 -> 86676 bytes .../images/image-9-Quarterly Growth Trend.jpg | Bin 0 -> 24591 bytes .../manifest.json | 71 + .../tables/table-0 Revenue by Region.html | 209 + .../tables/table-1 Quarterly Revenue.html | 76 + .../tables/table-10 Quarterly Values.html | 30 + .../tables/table-11 Quarterly Values.html | 30 + .../tables/table-12 Quarterly Values.html | 30 + .../tables/table-13 Quarterly Values.html | 30 + .../tables/table-14 Quarterly Values.html | 30 + .../tables/table-15 Quarterly Values.html | 30 + .../tables/table-16 Quarterly Values.html | 30 + .../tables/table-17 Quarterly Values.html | 30 + .../tables/table-18 Quarterly Values.html | 30 + .../tables/table-19 Quarterly Values.html | 30 + .../tables/table-2 Meta Revenue Data.html | 1 + .../tables/table-20 Free Cash Flow.html | 1 + .../tables/table-3 Segment Margins.html | 56 + .../tables/table-4 Quarterly Tax Data.html | 1 + .../tables/table-5 Quarterly Values.html | 46 + .../tables/table-6 Quarterly Values.html | 46 + .../tables/table-7 Financial Data.html | 26 + .../tables/table-8 Quarterly Values.html | 46 + .../tables/table-9 Quarterly Values.html | 46 + .../chunks.json | 1083 + .../doc_nav.json | 131 + .../financial-meta-q1-2026-results/full.md | 128 + .../manifest.json | 63 + .../tables/table-0 Q1 Financial Results.html | 1 + .../tables/table-1 Q1 Financial Results.html | 1 + .../tables/table-2 Balance Sheet.html | 1 + .../tables/table-3 Cash Flow Statement.html | 1 + .../tables/table-4 Meta Q1 Revenue.html | 1 + .../tables/table-5 Revenue & Cash Flow.html | 1 + .../chunks.json | 14047 ++++++ .../doc_nav.json | 712 + .../images/image-1 Annual Report 2025.png | Bin 0 -> 524557 bytes .../images/image-2 Signature.jpeg | Bin 0 -> 23982 bytes .../image-3 Microsoft vs S&P 500 v.jpeg | Bin 0 -> 96436 bytes .../image-4 FSC Certification Labe.jpeg | Bin 0 -> 26726 bytes .../manifest.json | 96 + .../tables/table-1 Share Repurchases.html | 1 + .../tables/table-10 Sales & Marketing.html | 1 + .../tables/table-11 G&A Expenses.html | 1 + .../tables/table-12 Financial Summary.html | 1 + .../tables/table-13 Revenue Schedule.html | 1 + .../tables/table-14 Debt Commitments.html | 1 + .../tables/table-15 Risk Sensitivity.html | 1 + .../tables/table-16 Revenue & Income.html | 1 + .../tables/table-17 Comprehensive Income.html | 1 + .../tables/table-18 Balance Sheet.html | 1 + .../tables/table-19 Cash Flow Statement.html | 1 + .../tables/table-2 Dividend History.html | 1 + .../tables/table-20 Stockholders' Equity.html | 1 + .../tables/table-21 EPS Data Table.html | 1 + .../tables/table-22 Investment Income.html | 1 + .../tables/table-23 Investment Gains.html | 1 + .../tables/table-24 Investment Gains.html | 1 + .../tables/table-25 Investment Table.html | 1 + .../tables/table-26 Investment Summary.html | 1 + .../tables/table-27 Unrealized Losses.html | 1 + .../tables/table-28 Debt Securities.html | 1 + .../tables/table-29 Derivative Notional.html | 1 + .../tables/table-3 Stock Index Data.html | 1 + .../tables/table-30 Derivative Assets.html | 1 + .../tables/table-31 Derivatives by Level.html | 1 + .../tables/table-32 Hedging Gains_Losses.html | 1 + .../tables/table-33 Cash Flow Hedges.html | 1 + .../tables/table-34 Property Assets.html | 1 + .../tables/table-35 Purchase Price.html | 1 + .../tables/table-36 Intangible Assets.html | 1 + .../tables/table-37 Microsoft FY2024.html | 1 + .../tables/table-38 Goodwill by Segment.html | 1 + .../tables/table-39 Intangible Assets.html | 1 + .../tables/table-4 Microsoft Cloud KPIs.html | 1 + .../tables/table-40 Intangible Assets.html | 1 + .../tables/table-41 Revenue Forecast.html | 1 + .../tables/table-42 Debt Issuance Table.html | 1 + .../tables/table-43 Debt Maturity.html | 1 + .../tables/table-44 Income Tax Provision.html | 1 + .../tables/table-45 Income Before Taxes.html | 1 + .../tables/table-46 Tax Rate Analysis.html | 1 + .../tables/table-47 Deferred Tax Assets.html | 1 + .../tables/table-48 Tax Benefits.html | 1 + .../tables/table-49 Microsoft Revenue.html | 1 + .../tables/table-5 Microsoft Cloud KPIs.html | 1 + .../tables/table-50 Unearned Revenue.html | 1 + .../tables/table-51 Lease Costs.html | 1 + .../tables/table-52 Lease Cash Flows.html | 1 + .../tables/table-53 Lease Data.html | 1 + .../tables/table-54 Lease Payments.html | 1 + .../tables/table-55 Share Repurchases.html | 1 + .../tables/table-56 Share Repurchases.html | 1 + .../tables/table-57 Dividend History.html | 1 + .../tables/table-58 AOCI Rollforward.html | 1 + .../tables/table-59 Stock Compensation.html | 1 + .../tables/table-6 Microsoft Revenue.html | 1 + .../tables/table-60 Dividends & Rates.html | 1 + .../tables/table-61 Stock Awards.html | 1 + .../tables/table-62 Share Repurchase.html | 1 + .../table-63 Microsoft Segment Results.html | 1 + .../tables/table-64 Revenue by Region.html | 1 + .../tables/table-65 Microsoft Revenue.html | 1 + .../tables/table-66 Revenue by Region.html | 1 + .../tables/table-67 Deloitte Signature.html | 1 + .../tables/table-68 Microsoft Board.html | 1 + .../tables/table-69 Microsoft Executives.html | 1 + .../tables/table-7 Revenue Growth.html | 1 + .../table-8 Microsoft Segment Results.html | 1 + .../tables/table-9 R&D Spending 2025.html | 1 + .../chunks.json | 39384 ++++++++++++++++ .../doc_nav.json | 3684 ++ .../financial-nvda-fy26-annual-report/full.md | 4864 ++ .../images/image-1-NVIDIA Logo.jpg | Bin 0 -> 118840 bytes .../images/image-10-SCIENCE AI.jpg | Bin 0 -> 3979 bytes ...11-Acronyms in AI and Machine Learning.jpg | Bin 0 -> 17685 bytes .../images/image-12-Stacked Server Racks.jpg | Bin 0 -> 43349 bytes .../image-13-DSX Architecture Overview.jpg | Bin 0 -> 240714 bytes .../image-14-NVIDIA CEO InferenceMAX King.jpg | Bin 0 -> 235454 bytes .../image-15-NVIDIA Leads Open-Source AI.jpg | Bin 0 -> 60385 bytes ...6-Collection of decorative enamel pins.jpg | Bin 0 -> 50861 bytes ...ge-17-Nemoclaw Agentic AI Architecture.jpg | Bin 0 -> 139575 bytes .../image-18-Future Robotics Exhibition.jpg | Bin 0 -> 255944 bytes .../images/image-19-Signature.jpg | Bin 0 -> 11853 bytes .../images/image-2-AI Technology Stack.jpg | Bin 0 -> 84636 bytes .../image-20-Jensen Huang Presentation.jpg | Bin 0 -> 168019 bytes .../images/image-21-Poster Gallery Event.jpg | Bin 0 -> 103761 bytes ...nference At-Scale is Extreme Computing.jpg | Bin 0 -> 40257 bytes .../image-23-NVIDIA GTC Event Scene.jpg | Bin 0 -> 119359 bytes ...e-24-Silhouette of Man Gazing at Earth.jpg | Bin 0 -> 170159 bytes .../images/image-25-Virtual me.jpg | Bin 0 -> 5141 bytes .../images/image-26-Server Rack Hardware.jpg | Bin 0 -> 16078 bytes .../images/image-27-NVIDIA RTX 4090.jpg | Bin 0 -> 11813 bytes .../image-28-Photo of a laptop workspace.jpg | Bin 0 -> 11910 bytes ...mage-29-Automotive X-Ray Visualization.jpg | Bin 0 -> 11353 bytes .../images/image-3-EINFRASTRU.jpg | Bin 0 -> 2512 bytes .../images/image-30-Growth Over Time.jpg | Bin 0 -> 14649 bytes .../image-31-Financial Returns by Period.jpg | Bin 0 -> 25606 bytes .../images/image-32-TENCH COXE.jpg | Bin 0 -> 1151 bytes .../images/image-33-Tench Coxe.jpg | Bin 0 -> 1501 bytes .../images/image-34-Financial_.jpg | Bin 0 -> 1467 bytes .../images/image-35-Governance.jpg | Bin 0 -> 1445 bytes .../images/image-36-JOHN O. DA.jpg | Bin 0 -> 1389 bytes .../images/image-37-John O. Da.jpg | Bin 0 -> 1483 bytes .../images/image-38-Jen-Hsun H.jpg | Bin 0 -> 1561 bytes .../images/image-39-Mr. Huang .jpg | Bin 0 -> 1353 bytes .../images/image-4-ENERGY --_.jpg | Bin 0 -> 3155 bytes .../images/image-40-Mr. Huang .jpg | Bin 0 -> 1192 bytes .../images/image-41-Industry &.jpg | Bin 0 -> 1485 bytes .../images/image-42-Financial_.jpg | Bin 0 -> 1471 bytes .../images/image-43-Community .jpg | Bin 0 -> 1494 bytes .../images/image-44-Emerging T.jpg | Bin 0 -> 1442 bytes .../images/image-45-Marketing,.jpg | Bin 0 -> 1490 bytes .../images/image-46-Financial .jpg | Bin 0 -> 1194 bytes .../images/image-47-Ms. Hudson.jpg | Bin 0 -> 1502 bytes .../images/image-48-Financial_.jpg | Bin 0 -> 1587 bytes .../images/image-49-Company Bo.jpg | Bin 0 -> 1479 bytes .../images/image-5-__details_.jpg | Bin 0 -> 3730 bytes .../images/image-50-Other Curr.jpg | Bin 0 -> 1544 bytes .../images/image-51-Financial .jpg | Bin 0 -> 1353 bytes .../images/image-52-Mr. Jones .jpg | Bin 0 -> 1184 bytes .../images/image-53-Industry &.jpg | Bin 0 -> 1482 bytes .../images/image-54-Financial_.jpg | Bin 0 -> 1454 bytes .../images/image-55-Governance.jpg | Bin 0 -> 1520 bytes .../images/image-56-Emerging T.jpg | Bin 0 -> 1441 bytes .../images/image-57-### A. BRO.jpg | Bin 0 -> 1561 bytes .../images/image-58-Senior Lea.jpg | Bin 0 -> 1464 bytes .../images/image-59-Financial_.jpg | Bin 0 -> 1462 bytes .../images/image-6-CHATBOTS E.jpg | Bin 0 -> 4146 bytes .../images/image-60-Governance.jpg | Bin 0 -> 1446 bytes .../images/image-61-Company Bo.jpg | Bin 0 -> 1474 bytes .../images/image-62-AARTI SHAH.jpg | Bin 0 -> 1601 bytes .../images/image-63-Aarti Shah.jpg | Bin 0 -> 1350 bytes .../images/image-64-Dr. Shah b.jpg | Bin 0 -> 1457 bytes .../images/image-65-Industry &.jpg | Bin 0 -> 1432 bytes .../images/image-66-Governance.jpg | Bin 0 -> 1580 bytes .../images/image-67-Emerging T.jpg | Bin 0 -> 1449 bytes .../images/image-68-Marketing,.jpg | Bin 0 -> 1443 bytes .../images/image-69-Committees.jpg | Bin 0 -> 1352 bytes .../images/image-7-ROBOTAXI S.jpg | Bin 0 -> 3244 bytes .../images/image-70-#### Indep.jpg | Bin 0 -> 1166 bytes .../images/image-71-Mr. Steven.jpg | Bin 0 -> 1504 bytes .../images/image-72-Financial_.jpg | Bin 0 -> 1496 bytes ...ge-73-Compensation Structure Breakdown.jpg | Bin 0 -> 14399 bytes ...ge-74-Compensation Structure Breakdown.jpg | Bin 0 -> 14293 bytes ...age-75-Executive Compensation Timeline.jpg | Bin 0 -> 64047 bytes .../image-76-NVIDIA Salary Percentiles.jpg | Bin 0 -> 20189 bytes .../image-77-NVIDIA vs Percentile Data.jpg | Bin 0 -> 21654 bytes .../images/image-78-NEO CAP versus TSR.jpg | Bin 0 -> 62737 bytes .../images/image-79-NEO CAP vs Income.jpg | Bin 0 -> 76940 bytes .../images/image-8-ENTERPRISE.jpg | Bin 0 -> 3405 bytes .../images/image-80-Effect of .jpg | Bin 0 -> 8185 bytes .../images/image-81-### Delinq.jpg | Bin 0 -> 4973 bytes .../images/image-82-ANNUAL REP.jpg | Bin 0 -> 2618 bytes .../images/image-83-NVIDIA Stock Growth.jpg | Bin 0 -> 56955 bytes .../images/image-9-AI AGENTS .jpg | Bin 0 -> 3464 bytes .../manifest.json | 467 + .../tables/table-0 GPU Cloud Growth.html | 51 + .../tables/table-1 NVIDIA Definitions.html | 1 + .../tables/table-10 Performance Payouts.html | 1 + .../tables/table-100 Income Before Tax.html | 1 + .../tables/table-101 Tax Rate Breakdown.html | 1 + .../tables/table-102 Tax Rate Analysis.html | 1 + .../tables/table-103 Income Tax Paid.html | 1 + .../tables/table-104 Deferred Tax Assets.html | 1 + .../tables/table-105 Tax Positions.html | 1 + .../tables/table-106 Segment Revenue.html | 1 + .../tables/table-107 Income Breakdown.html | 1 + .../tables/table-108 Revenue by Region.html | 1 + .../tables/table-109 NVIDIA Revenue.html | 1 + .../tables/table-11 Proxy Voting.html | 1 + .../tables/table-110 Long-lived assets.html | 1 + .../tables/table-111 Lease Obligations.html | 1 + .../tables/table-112 Lease Cash Flows.html | 1 + .../tables/table-113 Allowance Changes.html | 1 + .../tables/table-114 Exhibit Index.html | 1 + .../tables/table-115 Signatures.html | 1 + .../tables/table-12 Proposal Voting.html | 1 + .../tables/table-13 NVIDIA Board.html | 1 + .../tables/table-14 Board Skills Matrix.html | 1 + .../tables/table-15 Director Profile.html | 1 + .../tables/table-16 Director Profile.html | 1 + .../tables/table-17 Director Profile.html | 1 + .../table-18 Leadership & Operations.html | 1 + .../tables/table-19 Director Profile.html | 1 + .../tables/table-2 Revenue Growth.html | 1 + .../table-20 Leadership & Governance.html | 1 + .../tables/table-21 Director Profile.html | 1 + .../tables/table-22 Director Profile.html | 1 + .../tables/table-23 Board Risk Oversight.html | 1 + .../tables/table-24 Board Oversight.html | 1 + .../table-25 Director Compensation.html | 1 + .../tables/table-26 RSU Holdings Table.html | 1 + .../tables/table-27 Beneficial Ownership.html | 1 + .../table-28 Compensation Breakdown.html | 26 + .../tables/table-29 Compensation Mix.html | 30 + .../tables/table-3 NVIDIA Revenue.html | 1 + .../tables/table-30 Fiscal 2026 Goals.html | 1 + .../tables/table-31 Tech Companies.html | 1 + .../tables/table-32 Percentile Values.html | 26 + .../tables/table-33 NVIDIA Value Data.html | 26 + .../tables/table-34 NEO Compensation.html | 1 + .../tables/table-35 Compensation Plans.html | 1 + .../tables/table-36 Compensation Plan.html | 1 + .../tables/table-37 Performance Payouts.html | 1 + .../tables/table-38 CEO Pay Summary.html | 1 + .../tables/table-39 EVP CFO Pay.html | 1 + .../tables/table-4 Returns by Period.html | 22 + .../tables/table-40 EVP Compensation.html | 1 + .../tables/table-41 EVP Compensation.html | 1 + .../tables/table-42 Executive Pay.html | 1 + .../tables/table-43 Non-GAAP Income.html | 1 + .../table-44 Executive Compensation.html | 1 + .../tables/table-45 PSU Awards Table.html | 1 + .../table-46 Executive Compensation.html | 1 + .../tables/table-47 Stock Awards.html | 1 + .../tables/table-48 Stock Vesting Data.html | 1 + .../tables/table-49 Unvested RSUs_PSUs.html | 1 + .../tables/table-5 Dividend & Buyback.html | 26 + .../tables/table-50 SY PSU Grants.html | 1 + .../tables/table-51 MY PSU Grants 2024.html | 1 + .../table-52 CEO Pay & Performance.html | 1 + .../tables/table-53 CEO Compensation.html | 1 + .../tables/table-54 Non-CEO NEO Pay.html | 1 + .../tables/table-55 Financial Metrics.html | 1 + .../tables/table-56 NVIDIA CEO Pay.html | 48 + .../tables/table-57 CEO Pay & Income.html | 48 + .../tables/table-58 Fee Breakdown.html | 1 + .../tables/table-59 Equity Plan Table.html | 1 + .../tables/table-6 Proxy Voting Items.html | 1 + .../tables/table-60 Nvidia Executives.html | 1 + .../tables/table-61 Stock Buyback.html | 1 + .../tables/table-62 NVIDIA Stock Growth.html | 48 + .../tables/table-63 Stock Performance.html | 1 + .../tables/table-64 Financial Results.html | 1 + .../tables/table-65 Income Statement.html | 1 + .../tables/table-66 Revenue Growth.html | 1 + .../tables/table-67 Revenue Growth.html | 1 + .../tables/table-68 Operating Expenses.html | 1 + .../tables/table-69 Other Income.html | 1 + .../tables/table-7 Board of Directors.html | 1 + .../tables/table-70 Cash & Securities.html | 1 + .../tables/table-71 Cash Flow Table.html | 1 + .../tables/table-72 Debt Maturity.html | 1 + .../tables/table-73 Stock Sale Plans.html | 1 + .../tables/table-74 Income Statement.html | 1 + .../tables/table-75 Comprehensive Income.html | 1 + .../tables/table-76 Balance Sheet.html | 1 + .../tables/table-77 Equity Changes.html | 1 + .../tables/table-78 Cash Flow Statement.html | 1 + .../tables/table-79 Cost Breakdown.html | 1 + .../tables/table-8 Board Skills Matrix.html | 1 + .../tables/table-80 Equity Awards.html | 1 + .../tables/table-81 ESPP Option Data.html | 1 + .../tables/table-82 RSU_PSU Activity.html | 1 + .../tables/table-83 EPS Data.html | 1 + .../tables/table-84 Intangible Assets.html | 1 + .../tables/table-85 Amortization Expense.html | 1 + .../tables/table-86 Debt Securities.html | 1 + .../tables/table-87 Fair Value Table.html | 1 + .../tables/table-88 Debt Securities.html | 1 + .../tables/table-89 Debt Maturity.html | 1 + .../tables/table-9 Performance Goals.html | 1 + .../tables/table-90 Equity Securities.html | 1 + .../tables/table-91 Inventory Data.html | 1 + .../tables/table-92 Property & Equipment.html | 1 + .../tables/table-93 Current Liabilities.html | 1 + .../table-94 Long-Term Liabilities.html | 1 + .../tables/table-95 Deferred Revenue.html | 1 + .../tables/table-96 Hedge Values.html | 1 + .../tables/table-97 Debt Notes Table.html | 1 + .../table-98 Restructuring Reserve.html | 1 + .../tables/table-99 Income Tax Expense.html | 1 + .../toc_hierarchies.json | 64 + .../chunks.json | 1151 + .../doc_nav.json | 132 + .../full.md | 135 + .../manifest.json | 64 + .../tables/table-0 Q1 FY27 Results.html | 1 + .../tables/table-1 Q1 FY27 Results.html | 1 + .../tables/table-2 Revenue Growth.html | 1 + .../tables/table-3 Data Center Revenue.html | 1 + .../tables/table-4 Non-GAAP Financials.html | 1 + .../tables/table-5 Financial Data.html | 1 + .../tables/table-6 Cost Data Table.html | 1 + .../tables/table-7 Q2 FY27 Outlook.html | 1 + .../chunks.json | 3912 ++ .../doc_nav.json | 317 + .../full.md | 443 + .../images/image-1-Thanks, Co.jpg | Bin 0 -> 1369 bytes .../images/image-10-Analyst, B.jpg | Bin 0 -> 1243 bytes .../images/image-11-And so for.jpg | Bin 0 -> 1368 bytes .../images/image-12-Analyst, G.jpg | Bin 0 -> 1169 bytes .../images/image-13-In terms o.jpg | Bin 0 -> 1378 bytes .../images/image-14-Analyst, T.jpg | Bin 0 -> 1208 bytes .../images/image-2-Analyst, M.jpg | Bin 0 -> 1186 bytes .../images/image-3-And as you.jpg | Bin 0 -> 1382 bytes .../images/image-4-This secon.jpg | Bin 0 -> 1377 bytes .../images/image-5-Analyst, C.jpg | Bin 0 -> 1214 bytes .../images/image-6-Let me go .jpg | Bin 0 -> 1375 bytes .../images/image-7-So, we're .jpg | Bin 0 -> 1371 bytes .../images/image-8-Hi, guys. .jpg | Bin 0 -> 1207 bytes .../images/image-9-Co-Founder.jpg | Bin 0 -> 1373 bytes .../manifest.json | 84 + .../chunks.json | 1417 + .../doc_nav.json | 224 + .../full.md | 210 + .../image-1-Financial Growth Comparison.jpg | Bin 0 -> 41505 bytes ...mage-10-Dividends vs Share Repurchases.jpg | Bin 0 -> 45689 bytes .../image-11-Abstract Green Wave Design.jpg | Bin 0 -> 183056 bytes .../image-2-Q1 FY26 vs Q1 FY27 Revenue.jpg | Bin 0 -> 39357 bytes .../image-3-Financial Growth Comparison.jpg | Bin 0 -> 39423 bytes .../images/image-4-Q1 FY26 vs Q1 FY27.jpg | Bin 0 -> 37490 bytes .../images/image-5-Q1 FY26 vs Q1 FY27.jpg | Bin 0 -> 36832 bytes .../images/image-6-GAAP Revenue Growth.jpg | Bin 0 -> 43665 bytes .../image-7-Non-GAAP Revenue Growth.jpg | Bin 0 -> 43938 bytes .../image-8-GAAP vs Non-GAAP Comparison.jpg | Bin 0 -> 49234 bytes .../image-9-Financial Growth Comparison.jpg | Bin 0 -> 36801 bytes .../manifest.json | 65 + .../tables/table-0 Revenue Forecast.html | 18 + .../tables/table-1 Revenue Forecast.html | 18 + .../tables/table-10 Q1 FY27 Results.html | 1 + .../tables/table-11 Q2 FY27 Guidance.html | 1 + .../tables/table-12 Q1 FY27 GAAP.html | 1 + .../tables/table-13 Q1 FY26 Financials.html | 1 + .../tables/table-14 Q2 FY27 Outlook.html | 1 + .../tables/table-15 Free Cash Flow.html | 1 + .../tables/table-2 Revenue Forecast.html | 18 + .../tables/table-3 Quarterly Values.html | 18 + .../tables/table-4 Quarterly Values.html | 18 + .../tables/table-5 Revenue Forecast.html | 18 + .../tables/table-6 Revenue Forecast.html | 18 + .../tables/table-7 Revenue Forecast.html | 18 + .../tables/table-8 Dividends & Buybacks.html | 21 + .../tables/table-9 Q1 FY27 Results.html | 1 + .../chunks.json | 8228 ++++ .../doc_nav.json | 804 + .../financial-nvda-q1-fy27-results/full.md | 1059 + .../images/image-1-\342\230\222 QUARTERL.jpg" | Bin 0 -> 2333 bytes .../images/image-2-The Nasdaq.jpg | Bin 0 -> 842 bytes .../images/image-3-Indicate b.jpg | Bin 0 -> 799 bytes .../images/image-4-Large acce.jpg | Bin 0 -> 789 bytes .../images/image-5-Accelerate.jpg | Bin 0 -> 798 bytes .../images/image-6-Non-accele.jpg | Bin 0 -> 796 bytes .../manifest.json | 142 + .../tables/table-0 Financial Report.html | 1 + .../tables/table-1 Financial Results.html | 1 + .../tables/table-10 Amortization Expense.html | 1 + .../tables/table-11 Debt Securities.html | 1 + .../tables/table-12 Fair Value Table.html | 1 + .../tables/table-13 Debt Securities.html | 1 + .../tables/table-14 Debt Maturity.html | 1 + .../tables/table-15 Equity Securities.html | 1 + .../tables/table-16 Inventory Data.html | 1 + .../tables/table-17 Accrued Liabilities.html | 1 + .../table-18 Long-Term Liabilities.html | 1 + .../tables/table-19 Deferred Revenue.html | 1 + .../tables/table-2 Comprehensive Income.html | 1 + .../tables/table-20 Hedge Designation.html | 1 + .../tables/table-21 Debt Notes Table.html | 1 + .../tables/table-22 Balance Changes.html | 1 + .../tables/table-23 Segment Revenue.html | 1 + .../tables/table-24 Income Statement.html | 1 + .../tables/table-25 Revenue by Region.html | 1 + .../tables/table-26 Revenue by Market.html | 1 + .../tables/table-27 Lease Obligations.html | 1 + .../tables/table-28 Lease Cash Flows.html | 1 + .../tables/table-29 Quarterly Financials.html | 1 + .../tables/table-3 Balance Sheet.html | 1 + .../tables/table-30 Revenue by Market.html | 1 + .../tables/table-31 Revenue Analysis.html | 1 + .../tables/table-32 Revenue Growth.html | 1 + .../tables/table-33 Revenue Growth.html | 1 + .../tables/table-34 Operating Expenses.html | 1 + .../tables/table-35 Other Income.html | 1 + .../tables/table-36 Cash and Securities.html | 1 + .../tables/table-37 Cash Flow Data.html | 1 + .../tables/table-38 Debt Maturity.html | 1 + .../tables/table-39 Stock Buyback.html | 1 + .../tables/table-4 Shareholders' Equity.html | 1 + .../tables/table-40 Stock Sale Plans.html | 1 + .../tables/table-41 Exhibit List.html | 1 + .../tables/table-5 Cash Flow Statement.html | 1 + .../tables/table-6 Cost Breakdown.html | 1 + .../tables/table-7 Stock Grant Activity.html | 1 + .../tables/table-8 EPS Data Table.html | 1 + .../tables/table-9 Intangible Assets.html | 1 + .../toc_hierarchies.json | 17 + .../chunks.json | 5285 +++ .../doc_nav.json | 425 + .../full.md | 527 + .../images/image-1-We will no.jpg | Bin 0 -> 1369 bytes .../images/image-10-Hi, guys. .jpg | Bin 0 -> 1203 bytes .../images/image-11-It's too e.jpg | Bin 0 -> 1370 bytes .../images/image-12-Analyst, C.jpg | Bin 0 -> 1193 bytes .../images/image-13-For the Da.jpg | Bin 0 -> 1371 bytes .../image-14-Every \342\200\223 th.jpg" | Bin 0 -> 1354 bytes .../images/image-15-Analyst, N.jpg | Bin 0 -> 1200 bytes .../images/image-16-And one of.jpg | Bin 0 -> 1368 bytes .../images/image-17-Analyst, E.jpg | Bin 0 -> 1227 bytes .../images/image-18-And then l.jpg | Bin 0 -> 1361 bytes .../images/image-19-Yeah. Than.jpg | Bin 0 -> 1183 bytes .../images/image-2-Analyst, B.jpg | Bin 0 -> 1216 bytes .../images/image-20-So, thanks.jpg | Bin 0 -> 1378 bytes .../images/image-21-Thank you .jpg | Bin 0 -> 1222 bytes .../images/image-3-I am confi.jpg | Bin 0 -> 1362 bytes .../images/image-4-Analyst, M.jpg | Bin 0 -> 1214 bytes .../images/image-5-Our ecosys.jpg | Bin 0 -> 1369 bytes .../images/image-6-Good after.jpg | Bin 0 -> 1209 bytes .../images/image-7-We're also.jpg | Bin 0 -> 1370 bytes .../images/image-8-Analyst, C.jpg | Bin 0 -> 1205 bytes .../images/image-9-With respe.jpg | Bin 0 -> 1362 bytes .../manifest.json | 92 + .../chunks.json | 1301 + .../doc_nav.json | 153 + .../full.md | 458 + .../image-1-Attention Mechanism Process.jpg | Bin 0 -> 21937 bytes ...e-10-Attention Mechanism Visualization.jpg | Bin 0 -> 25953 bytes ...ge-11-Transformer Encoder Architecture.jpg | Bin 0 -> 41229 bytes .../image-2-Attention Matrix Heatmap.jpg | Bin 0 -> 26710 bytes .../images/image-3-Word Alignment Matrix.jpg | Bin 0 -> 18131 bytes ...age-4-Transformer Architecture Diagram.jpg | Bin 0 -> 25178 bytes .../images/image-5-Heatmap Visualization.jpg | Bin 0 -> 56322 bytes .../images/image-6-Heatmap Visualization.jpg | Bin 0 -> 30651 bytes .../images/image-7-Heatmap Visualization.jpg | Bin 0 -> 22458 bytes .../images/image-8-$$ _detail.jpg | Bin 0 -> 6542 bytes .../images/image-9-E --_ G[_K.jpg | Bin 0 -> 9871 bytes .../manifest.json | 62 + .../tables/table-0 Attention Matrix.html | 69 + .../tables/table-1 EEA Agreement.html | 70 + .../tables/table-2 Constant Matrix.html | 201 + .../tables/table-3 Low Values Table.html | 139 + .../tables/table-4 Color Pattern Table.html | 158 + .../tables/table-5 BERT Embeddings.html | 1 + .../chunks.json | 2186 + .../doc_nav.json | 447 + .../full.md | 658 + .../image-1-Housing Prices Scatter Plot.jpg | Bin 0 -> 11300 bytes .../images/image-10-Movie Rank.jpg | Bin 0 -> 2883 bytes .../images/image-11-user_127 _.jpg | Bin 0 -> 2165 bytes .../images/image-12-_details_ .jpg | Bin 0 -> 1929 bytes .../images/image-13-_summary_n.jpg | Bin 0 -> 2634 bytes .../image-14-Loss Functions Comparison.jpg | Bin 0 -> 16704 bytes .../image-15-Exponential Functions Graph.jpg | Bin 0 -> 10793 bytes .../images/image-2-__details_.jpg | Bin 0 -> 9375 bytes .../images/image-3-graph LR _.jpg | Bin 0 -> 8368 bytes .../images/image-4-Image Type_ Photo.jpg | Bin 0 -> 60878 bytes .../images/image-5-Visual Data Grid.jpg | Bin 0 -> 119414 bytes .../images/image-6-Grid of nu.jpg | Bin 0 -> 3632 bytes .../images/image-7-_details_ .jpg | Bin 0 -> 4427 bytes .../images/image-8-Coffee mug.jpg | Bin 0 -> 2789 bytes .../images/image-9-Newspaper .jpg | Bin 0 -> 2763 bytes .../manifest.json | 103 + .../tables/table-0 Housing Data.html | 1 + .../tables/table-1 House Price Data.html | 94 + .../tables/table-2 Loss Functions.html | 42 + .../tables/table-3 Data Table.html | 34 + .../stem-jurafsky-transformers/chunks.json | 4362 ++ .../stem-jurafsky-transformers/doc_nav.json | 300 + .../stem-jurafsky-transformers/full.md | 1364 + ...ansformer Encoder Architecture Diagram.jpg | Bin 0 -> 33126 bytes .../image-10-Attention Mechanism Process.jpg | Bin 0 -> 91348 bytes .../images/image-11-$$ _detail.jpg | Bin 0 -> 6361 bytes .../images/image-12-__details_.jpg | Bin 0 -> 10228 bytes ...mage-13-Transformer Block Architecture.jpg | Bin 0 -> 34066 bytes ...ge-14-Language Model Head Architecture.jpg | Bin 0 -> 57576 bytes ...ge-15-Transformer Architecture Diagram.jpg | Bin 0 -> 82154 bytes ...ransformer Language Model Architecture.jpg | Bin 0 -> 109648 bytes ...ention Mechanism Matrix Multiplication.jpg | Bin 0 -> 20207 bytes ...ge-18-Pretrained Weights Decomposition.jpg | Bin 0 -> 17397 bytes .../image-19-Prefix matching and copying.jpg | Bin 0 -> 15070 bytes ...age-2-Transformer Architecture Diagram.jpg | Bin 0 -> 84211 bytes .../image-3-Self-Attention Visualization.jpg | Bin 0 -> 37010 bytes .../image-4-Self-Attention Layer Diagram.jpg | Bin 0 -> 23545 bytes ...age-5-Self-Attention Mechanism Process.jpg | Bin 0 -> 78423 bytes ...image-6-Multi-Head Attention Mechanism.jpg | Bin 0 -> 69112 bytes ...image-7-Transformer Block Architecture.jpg | Bin 0 -> 32130 bytes ...ge-8-Diagram of Token Residual Streams.jpg | Bin 0 -> 14453 bytes .../images/image-9-N x N Grid Structure.jpg | Bin 0 -> 13705 bytes .../stem-jurafsky-transformers/manifest.json | 80 + .../tables/table-0 Self-Attention.html | 106 + .../tables/table-1 Causal Attention Mask.html | 1 + .../tables/table-3 .html | 8 + .../tables/table-4 .html | 8 + .../chunks.json | 6120 +++ .../doc_nav.json | 756 + .../full.md | 4426 ++ .../images/image-1-_- Events_.jpg | Bin 0 -> 5062 bytes .../images/image-10-PROPOSITIO.jpg | Bin 0 -> 853 bytes .../images/image-11-Step Function Graph.jpg | Bin 0 -> 59546 bytes .../images/image-12-PROOF. We .jpg | Bin 0 -> 853 bytes .../images/image-13-Thus plugg.jpg | Bin 0 -> 835 bytes .../images/image-14-DEFINITION.jpg | Bin 0 -> 854 bytes .../images/image-15-$$ _- Summ.jpg | Bin 0 -> 873 bytes .../images/image-16-$$ PROPOSI.jpg | Bin 0 -> 858 bytes ...-17-Standard Normal Distribution Curve.jpg | Bin 0 -> 13394 bytes ...-18-Standard Normal Distribution Graph.jpg | Bin 0 -> 12948 bytes .../image-19-Normal Distribution Curve.jpg | Bin 0 -> 16019 bytes .../images/image-2-- $A _cup .jpg | Bin 0 -> 5538 bytes .../images/image-20-PROOF. To .jpg | Bin 0 -> 853 bytes .../images/image-21-Graph of Inequality.jpg | Bin 0 -> 15988 bytes .../image-22-Linear Inequality Region.jpg | Bin 0 -> 17366 bytes .../images/image-23-Graph of Inequality.jpg | Bin 0 -> 16580 bytes .../images/image-24-$$ _- Exam.jpg | Bin 0 -> 858 bytes .../images/image-25-Linear Function Graph.jpg | Bin 0 -> 16290 bytes .../images/image-26-(b) If $X .jpg | Bin 0 -> 838 bytes .../image-27-Scatter Plot with Trend Line.jpg | Bin 0 -> 21473 bytes .../images/image-3-- $A _cap .jpg | Bin 0 -> 4725 bytes .../images/image-4-- $A^c$ is.jpg | Bin 0 -> 5225 bytes .../images/image-5-- Can exte.jpg | Bin 0 -> 5168 bytes .../images/image-6-\342\200\242 Example1.jpg" | Bin 0 -> 4914 bytes .../images/image-7-\342\200\242 Example1.jpg" | Bin 0 -> 4892 bytes .../images/image-8-\342\200\242 Example1.jpg" | Bin 0 -> 4983 bytes .../images/image-9-\342\200\242 Example1.jpg" | Bin 0 -> 5379 bytes .../manifest.json | 146 + .../tables/table-0 Income by Gender.html | 1 + .../tables/table-1 Data Table.html | 34 + .../tables/table-10 Joint Probability.html | 1 + .../tables/table-11 Linear Data Table.html | 78 + .../tables/table-2 Symmetric Data.html | 62 + .../tables/table-3 Z and CDF Table.html | 18 + .../tables/table-4 Normal Distribution.html | 42 + .../tables/table-5 Table of Pairs.html | 1 + .../table-6 Joint Probability Table.html | 1 + .../tables/table-7 Linear Data Table.html | 34 + .../tables/table-8 Data Table.html | 14 + .../tables/table-9 Linear Data Table.html | 34 + .../toc_hierarchies.json | 54 + .../stem-statistical-learning/chunks.json | 6257 +++ .../stem-statistical-learning/doc_nav.json | 559 + .../stem-statistical-learning/full.md | 2427 + ...-1-Support Vector Margin Visualization.jpg | Bin 0 -> 37312 bytes .../images/image-10-This motiv.jpg | Bin 0 -> 4379 bytes .../images/image-11-$$ 5.1 Usi.jpg | Bin 0 -> 869 bytes .../images/image-2-Propositio.jpg | Bin 0 -> 842 bytes .../images/image-3-$$ # Chapt.jpg | Bin 0 -> 847 bytes .../images/image-4-From this .jpg | Bin 0 -> 846 bytes .../images/image-5-$$ We woul.jpg | Bin 0 -> 802 bytes .../images/image-6-Let $Z = (.jpg | Bin 0 -> 8451 bytes .../images/image-7-Z2 --_ Z3 .jpg | Bin 0 -> 7701 bytes .../images/image-8-Example 4..jpg | Bin 0 -> 7039 bytes .../images/image-9-$$ For a D.jpg | Bin 0 -> 4347 bytes .../stem-statistical-learning/manifest.json | 130 + .../toc_hierarchies.json | 79 + .../services/demo/official_library_catalog.py | 359 + apps/api/app/services/demo/source_catalog.py | 226 +- apps/api/scripts/validate_demo_documents.py | 80 + .../contract/test_demo_documents_contract.py | 31 + 748 files changed, 172997 insertions(+), 7 deletions(-) create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/full.md create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/images/image-1-Alibaba.jpg create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/images/image-2-### (vii) .jpg create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-0 Financial Results.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-1 Alibaba Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-10 Board of Directors.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-11 Shareholding Table.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-12 Shareholder Holdings.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-13 RSU Grant Summary.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-14 Share Options Table.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-15 RSU Activity Report.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-16 RSU Grant Details.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-17 Share Repurchase.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-18 Notes Proceeds Use.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-19 Interim Report.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-2 Alibaba EBITA.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-20 Alibaba Financials.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-21 Income Statement.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-22 Balance Sheet Data.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-23 Equity Summary.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-24 Equity Changes.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-25 Cash Flow Statement.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-26 Financial Data.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-27 Alibaba Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-28 Revenue Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-29 Tax Expense Data.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-3 Costs & Expenses.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-30 Cost Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-31 EPS Data Table.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-32 Investment Summary.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-33 Fair Value Table.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-34 Fair Value Table.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-35 Balance Change.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-36 Fair Value Change.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-37 Intangible Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-38 Alibaba Segment Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-39 Financial Liabilities.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-4 Share-based Compensation.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-40 Cloud Revenue Growth.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-41 Alibaba Segment Results.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-42 Financial Results.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-43 Depreciation by Segment.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-44 Accounts Receivable.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-45 Accounts Payable.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-46 IFRS Adjustments.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-47 GAAP to IFRS.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-48 GAAP to IFRS.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-49 GAAP to IFRS Reconciliation.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-5 Equity Investee Results.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-6 Financial Results.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-7 Non-GAAP Net Income.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-8 Earnings per Share.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-9 Free Cash Flow.html create mode 100644 apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/toc_hierarchies.json create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/full.md create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/images/image-1-Stock Performance Comparison.jpg create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/images/image-2-Stock Performance Comparison (2020-2025).jpg create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-0 Stock & Note Listings.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-1 Filer Status.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-10 R&D Expenses.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-11 Sales Expenses.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-12 Expense Analysis.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-13 Alphabet Income.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-14 Other Income_Expense.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-15 Tax Data Table.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-16 Cash Flow Table.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-17 Foreign Currency Risk.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-18 Interest Rate Risk.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-19 Financial Index.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-2 Share Repurchase.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-20 Loss Contingencies.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-21 Balance Sheet.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-22 Financial Data.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-23 Comprehensive Income.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-24 Stockholders' Equity.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-25 Cash Flow Statement.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-26 Google Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-27 Revenue Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-28 Fair Value Table.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-29 Fair Value Table.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-3 Alphabet Stock vs Indexes.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-30 Debt Maturity.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-31 Investment Summary.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-32 Investment Summary.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-33 Investment Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-34 Investment Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-35 Debt Securities P&L.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-36 Non-marketable securities.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-37 Equity Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-38 Derivatives Table.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-39 Derivatives Table.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-4 Alphabet vs S&P 500.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-40 Hedge Gains_Losses.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-41 Hedge Effects.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-42 Derivatives Offsetting.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-43 Lease Cost Table.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-44 Lease Data.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-45 Lease Cash Flows.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-46 Lease Payments.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-47 Debt Table.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-48 Debt Maturity.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-49 Property & Equipment.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-5 Financial Results.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-50 Accrued Liabilities.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-51 OCI Rollforward.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-52 AOCI Reclass.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-53 Income Data Table.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-54 Google Segment Balances.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-55 Share Repurchases.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-56 EPS Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-57 EPS Calculation.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-58 EPS Calculation.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-59 Stock Vesting.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-6 Revenue Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-60 Revenue Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-61 Income Tax Provision.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-62 Tax Rate Reconciliation.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-63 Deferred Tax Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-64 Income Tax Payments.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-65 Tax Benefits.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-66 Alphabet Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-67 Long-lived Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-68 Warranty Reserves.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-69 Exhibit List.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-7 Google Ad Changes.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-70 CEO Signature.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-71 Signatures.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-8 Revenue by Region.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-9 Cost of Revenues.html create mode 100644 apps/api/app/data/demo_documents/financial-goog-10-k-2025/toc_hierarchies.json create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/full.md create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/tables/table-0 AI Compute Plans.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/full.md create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-1-Revenue Forecast by Region.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-10-Quarterly Percentage Growth.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-11-Quarterly Growth Trend.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-12-Quarterly Growth Trend.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-13-Growth Trend 2025-2026.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-14-Quarterly Percentage Data (2025-2026).jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-15-Quarterly Percentage Data (2025-2026).jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-16-Quarterly Percentage Trends.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-17-Quarterly Growth Trend.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-18-Quarterly Percentage Data (2025-2026).jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-2-Revenue Breakdown by Region and Quarter.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-3-Quarterly Percentage Breakdown.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-4-Quarterly Financial Performance.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-5-Quarterly Financial Growth.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-6-Financial Growth Projection.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-7-Quarterly Data Trend.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-8-Quarterly Financial Data.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-9-Quarterly Growth Trend.jpg create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-0 Revenue by Region.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-1 Quarterly Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-10 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-11 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-12 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-13 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-14 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-15 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-16 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-17 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-18 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-19 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-2 Meta Revenue Data.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-20 Free Cash Flow.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-3 Segment Margins.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-4 Quarterly Tax Data.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-5 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-6 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-7 Financial Data.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-8 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-9 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-results/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-results/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-results/full.md create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-results/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-0 Q1 Financial Results.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-1 Q1 Financial Results.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-2 Balance Sheet.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-3 Cash Flow Statement.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-4 Meta Q1 Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-5 Revenue & Cash Flow.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/images/image-1 Annual Report 2025.png create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/images/image-2 Signature.jpeg create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/images/image-3 Microsoft vs S&P 500 v.jpeg create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/images/image-4 FSC Certification Labe.jpeg create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-1 Share Repurchases.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-10 Sales & Marketing.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-11 G&A Expenses.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-12 Financial Summary.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-13 Revenue Schedule.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-14 Debt Commitments.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-15 Risk Sensitivity.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-16 Revenue & Income.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-17 Comprehensive Income.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-18 Balance Sheet.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-19 Cash Flow Statement.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-2 Dividend History.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-20 Stockholders' Equity.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-21 EPS Data Table.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-22 Investment Income.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-23 Investment Gains.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-24 Investment Gains.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-25 Investment Table.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-26 Investment Summary.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-27 Unrealized Losses.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-28 Debt Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-29 Derivative Notional.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-3 Stock Index Data.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-30 Derivative Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-31 Derivatives by Level.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-32 Hedging Gains_Losses.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-33 Cash Flow Hedges.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-34 Property Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-35 Purchase Price.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-36 Intangible Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-37 Microsoft FY2024.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-38 Goodwill by Segment.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-39 Intangible Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-4 Microsoft Cloud KPIs.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-40 Intangible Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-41 Revenue Forecast.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-42 Debt Issuance Table.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-43 Debt Maturity.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-44 Income Tax Provision.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-45 Income Before Taxes.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-46 Tax Rate Analysis.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-47 Deferred Tax Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-48 Tax Benefits.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-49 Microsoft Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-5 Microsoft Cloud KPIs.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-50 Unearned Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-51 Lease Costs.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-52 Lease Cash Flows.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-53 Lease Data.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-54 Lease Payments.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-55 Share Repurchases.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-56 Share Repurchases.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-57 Dividend History.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-58 AOCI Rollforward.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-59 Stock Compensation.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-6 Microsoft Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-60 Dividends & Rates.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-61 Stock Awards.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-62 Share Repurchase.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-63 Microsoft Segment Results.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-64 Revenue by Region.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-65 Microsoft Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-66 Revenue by Region.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-67 Deloitte Signature.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-68 Microsoft Board.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-69 Microsoft Executives.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-7 Revenue Growth.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-8 Microsoft Segment Results.html create mode 100644 apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-9 R&D Spending 2025.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/full.md create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-1-NVIDIA Logo.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-10-SCIENCE AI.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-11-Acronyms in AI and Machine Learning.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-12-Stacked Server Racks.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-13-DSX Architecture Overview.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-14-NVIDIA CEO InferenceMAX King.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-15-NVIDIA Leads Open-Source AI.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-16-Collection of decorative enamel pins.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-17-Nemoclaw Agentic AI Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-18-Future Robotics Exhibition.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-19-Signature.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-2-AI Technology Stack.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-20-Jensen Huang Presentation.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-21-Poster Gallery Event.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-22-Inference At-Scale is Extreme Computing.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-23-NVIDIA GTC Event Scene.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-24-Silhouette of Man Gazing at Earth.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-25-Virtual me.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-26-Server Rack Hardware.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-27-NVIDIA RTX 4090.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-28-Photo of a laptop workspace.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-29-Automotive X-Ray Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-3-EINFRASTRU.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-30-Growth Over Time.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-31-Financial Returns by Period.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-32-TENCH COXE.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-33-Tench Coxe.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-34-Financial_.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-35-Governance.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-36-JOHN O. DA.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-37-John O. Da.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-38-Jen-Hsun H.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-39-Mr. Huang .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-4-ENERGY --_.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-40-Mr. Huang .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-41-Industry &.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-42-Financial_.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-43-Community .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-44-Emerging T.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-45-Marketing,.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-46-Financial .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-47-Ms. Hudson.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-48-Financial_.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-49-Company Bo.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-5-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-50-Other Curr.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-51-Financial .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-52-Mr. Jones .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-53-Industry &.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-54-Financial_.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-55-Governance.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-56-Emerging T.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-57-### A. BRO.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-58-Senior Lea.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-59-Financial_.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-6-CHATBOTS E.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-60-Governance.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-61-Company Bo.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-62-AARTI SHAH.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-63-Aarti Shah.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-64-Dr. Shah b.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-65-Industry &.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-66-Governance.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-67-Emerging T.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-68-Marketing,.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-69-Committees.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-7-ROBOTAXI S.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-70-#### Indep.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-71-Mr. Steven.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-72-Financial_.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-73-Compensation Structure Breakdown.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-74-Compensation Structure Breakdown.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-75-Executive Compensation Timeline.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-76-NVIDIA Salary Percentiles.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-77-NVIDIA vs Percentile Data.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-78-NEO CAP versus TSR.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-79-NEO CAP vs Income.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-8-ENTERPRISE.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-80-Effect of .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-81-### Delinq.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-82-ANNUAL REP.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-83-NVIDIA Stock Growth.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-9-AI AGENTS .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-0 GPU Cloud Growth.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-1 NVIDIA Definitions.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-10 Performance Payouts.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-100 Income Before Tax.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-101 Tax Rate Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-102 Tax Rate Analysis.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-103 Income Tax Paid.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-104 Deferred Tax Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-105 Tax Positions.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-106 Segment Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-107 Income Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-108 Revenue by Region.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-109 NVIDIA Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-11 Proxy Voting.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-110 Long-lived assets.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-111 Lease Obligations.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-112 Lease Cash Flows.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-113 Allowance Changes.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-114 Exhibit Index.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-115 Signatures.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-12 Proposal Voting.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-13 NVIDIA Board.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-14 Board Skills Matrix.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-15 Director Profile.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-16 Director Profile.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-17 Director Profile.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-18 Leadership & Operations.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-19 Director Profile.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-2 Revenue Growth.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-20 Leadership & Governance.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-21 Director Profile.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-22 Director Profile.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-23 Board Risk Oversight.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-24 Board Oversight.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-25 Director Compensation.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-26 RSU Holdings Table.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-27 Beneficial Ownership.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-28 Compensation Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-29 Compensation Mix.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-3 NVIDIA Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-30 Fiscal 2026 Goals.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-31 Tech Companies.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-32 Percentile Values.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-33 NVIDIA Value Data.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-34 NEO Compensation.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-35 Compensation Plans.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-36 Compensation Plan.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-37 Performance Payouts.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-38 CEO Pay Summary.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-39 EVP CFO Pay.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-4 Returns by Period.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-40 EVP Compensation.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-41 EVP Compensation.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-42 Executive Pay.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-43 Non-GAAP Income.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-44 Executive Compensation.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-45 PSU Awards Table.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-46 Executive Compensation.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-47 Stock Awards.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-48 Stock Vesting Data.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-49 Unvested RSUs_PSUs.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-5 Dividend & Buyback.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-50 SY PSU Grants.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-51 MY PSU Grants 2024.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-52 CEO Pay & Performance.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-53 CEO Compensation.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-54 Non-CEO NEO Pay.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-55 Financial Metrics.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-56 NVIDIA CEO Pay.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-57 CEO Pay & Income.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-58 Fee Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-59 Equity Plan Table.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-6 Proxy Voting Items.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-60 Nvidia Executives.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-61 Stock Buyback.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-62 NVIDIA Stock Growth.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-63 Stock Performance.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-64 Financial Results.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-65 Income Statement.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-66 Revenue Growth.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-67 Revenue Growth.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-68 Operating Expenses.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-69 Other Income.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-7 Board of Directors.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-70 Cash & Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-71 Cash Flow Table.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-72 Debt Maturity.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-73 Stock Sale Plans.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-74 Income Statement.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-75 Comprehensive Income.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-76 Balance Sheet.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-77 Equity Changes.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-78 Cash Flow Statement.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-79 Cost Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-8 Board Skills Matrix.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-80 Equity Awards.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-81 ESPP Option Data.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-82 RSU_PSU Activity.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-83 EPS Data.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-84 Intangible Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-85 Amortization Expense.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-86 Debt Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-87 Fair Value Table.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-88 Debt Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-89 Debt Maturity.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-9 Performance Goals.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-90 Equity Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-91 Inventory Data.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-92 Property & Equipment.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-93 Current Liabilities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-94 Long-Term Liabilities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-95 Deferred Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-96 Hedge Values.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-97 Debt Notes Table.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-98 Restructuring Reserve.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-99 Income Tax Expense.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/toc_hierarchies.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/full.md create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-0 Q1 FY27 Results.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-1 Q1 FY27 Results.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-2 Revenue Growth.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-3 Data Center Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-4 Non-GAAP Financials.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-5 Financial Data.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-6 Cost Data Table.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-7 Q2 FY27 Outlook.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/full.md create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-1-Thanks, Co.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-10-Analyst, B.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-11-And so for.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-12-Analyst, G.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-13-In terms o.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-14-Analyst, T.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-2-Analyst, M.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-3-And as you.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-4-This secon.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-5-Analyst, C.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-6-Let me go .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-7-So, we're .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-8-Hi, guys. .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-9-Co-Founder.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/full.md create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-1-Financial Growth Comparison.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-10-Dividends vs Share Repurchases.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-11-Abstract Green Wave Design.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-2-Q1 FY26 vs Q1 FY27 Revenue.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-3-Financial Growth Comparison.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-4-Q1 FY26 vs Q1 FY27.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-5-Q1 FY26 vs Q1 FY27.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-6-GAAP Revenue Growth.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-7-Non-GAAP Revenue Growth.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-8-GAAP vs Non-GAAP Comparison.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-9-Financial Growth Comparison.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-0 Revenue Forecast.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-1 Revenue Forecast.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-10 Q1 FY27 Results.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-11 Q2 FY27 Guidance.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-12 Q1 FY27 GAAP.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-13 Q1 FY26 Financials.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-14 Q2 FY27 Outlook.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-15 Free Cash Flow.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-2 Revenue Forecast.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-3 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-4 Quarterly Values.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-5 Revenue Forecast.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-6 Revenue Forecast.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-7 Revenue Forecast.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-8 Dividends & Buybacks.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-9 Q1 FY27 Results.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/full.md create mode 100644 "apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-1-\342\230\222 QUARTERL.jpg" create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-2-The Nasdaq.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-3-Indicate b.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-4-Large acce.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-5-Accelerate.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-6-Non-accele.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-0 Financial Report.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-1 Financial Results.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-10 Amortization Expense.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-11 Debt Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-12 Fair Value Table.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-13 Debt Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-14 Debt Maturity.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-15 Equity Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-16 Inventory Data.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-17 Accrued Liabilities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-18 Long-Term Liabilities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-19 Deferred Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-2 Comprehensive Income.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-20 Hedge Designation.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-21 Debt Notes Table.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-22 Balance Changes.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-23 Segment Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-24 Income Statement.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-25 Revenue by Region.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-26 Revenue by Market.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-27 Lease Obligations.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-28 Lease Cash Flows.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-29 Quarterly Financials.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-3 Balance Sheet.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-30 Revenue by Market.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-31 Revenue Analysis.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-32 Revenue Growth.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-33 Revenue Growth.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-34 Operating Expenses.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-35 Other Income.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-36 Cash and Securities.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-37 Cash Flow Data.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-38 Debt Maturity.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-39 Stock Buyback.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-4 Shareholders' Equity.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-40 Stock Sale Plans.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-41 Exhibit List.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-5 Cash Flow Statement.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-6 Cost Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-7 Stock Grant Activity.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-8 EPS Data Table.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-9 Intangible Assets.html create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/toc_hierarchies.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/full.md create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-1-We will no.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-10-Hi, guys. .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-11-It's too e.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-12-Analyst, C.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-13-For the Da.jpg create mode 100644 "apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-14-Every \342\200\223 th.jpg" create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-15-Analyst, N.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-16-And one of.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-17-Analyst, E.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-18-And then l.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-19-Yeah. Than.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-2-Analyst, B.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-20-So, thanks.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-21-Thank you .jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-3-I am confi.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-4-Analyst, M.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-5-Our ecosys.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-6-Good after.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-7-We're also.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-8-Analyst, C.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-9-With respe.jpg create mode 100644 apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/manifest.json create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/chunks.json create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/full.md create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-1-Attention Mechanism Process.jpg create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-10-Attention Mechanism Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-11-Transformer Encoder Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-2-Attention Matrix Heatmap.jpg create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-3-Word Alignment Matrix.jpg create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-4-Transformer Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-5-Heatmap Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-6-Heatmap Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-7-Heatmap Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-8-$$ _detail.jpg create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-9-E --_ G[_K.jpg create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/manifest.json create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-0 Attention Matrix.html create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-1 EEA Agreement.html create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-2 Constant Matrix.html create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-3 Low Values Table.html create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-4 Color Pattern Table.html create mode 100644 apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-5 BERT Embeddings.html create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/chunks.json create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/full.md create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-1-Housing Prices Scatter Plot.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-10-Movie Rank.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-11-user_127 _.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-12-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-13-_summary_n.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-14-Loss Functions Comparison.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-15-Exponential Functions Graph.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-2-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-3-graph LR _.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-4-Image Type_ Photo.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-5-Visual Data Grid.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-6-Grid of nu.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-7-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-8-Coffee mug.jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-9-Newspaper .jpg create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/manifest.json create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-0 Housing Data.html create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-1 House Price Data.html create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-2 Loss Functions.html create mode 100644 apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-3 Data Table.html create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/chunks.json create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/full.md create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-1-Transformer Encoder Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-10-Attention Mechanism Process.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-11-$$ _detail.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-12-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-13-Transformer Block Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-14-Language Model Head Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-15-Transformer Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-16-Transformer Language Model Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-17-Attention Mechanism Matrix Multiplication.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-18-Pretrained Weights Decomposition.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-19-Prefix matching and copying.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-2-Transformer Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-3-Self-Attention Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-4-Self-Attention Layer Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-5-Self-Attention Mechanism Process.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-6-Multi-Head Attention Mechanism.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-7-Transformer Block Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-8-Diagram of Token Residual Streams.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-9-N x N Grid Structure.jpg create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/manifest.json create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-0 Self-Attention.html create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-1 Causal Attention Mask.html create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-3 .html create mode 100644 apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-4 .html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/chunks.json create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/full.md create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-1-_- Events_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-10-PROPOSITIO.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-11-Step Function Graph.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-12-PROOF. We .jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-13-Thus plugg.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-14-DEFINITION.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-15-$$ _- Summ.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-16-$$ PROPOSI.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-17-Standard Normal Distribution Curve.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-18-Standard Normal Distribution Graph.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-19-Normal Distribution Curve.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-2-- $A _cup .jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-20-PROOF. To .jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-21-Graph of Inequality.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-22-Linear Inequality Region.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-23-Graph of Inequality.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-24-$$ _- Exam.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-25-Linear Function Graph.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-26-(b) If $X .jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-27-Scatter Plot with Trend Line.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-3-- $A _cap .jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-4-- $A^c$ is.jpg create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-5-- Can exte.jpg create mode 100644 "apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-6-\342\200\242 Example1.jpg" create mode 100644 "apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-7-\342\200\242 Example1.jpg" create mode 100644 "apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-8-\342\200\242 Example1.jpg" create mode 100644 "apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-9-\342\200\242 Example1.jpg" create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/manifest.json create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-0 Income by Gender.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-1 Data Table.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-10 Joint Probability.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-11 Linear Data Table.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-2 Symmetric Data.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-3 Z and CDF Table.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-4 Normal Distribution.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-5 Table of Pairs.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-6 Joint Probability Table.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-7 Linear Data Table.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-8 Data Table.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-9 Linear Data Table.html create mode 100644 apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/toc_hierarchies.json create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/chunks.json create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/full.md create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/images/image-1-Support Vector Margin Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/images/image-10-This motiv.jpg create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/images/image-11-$$ 5.1 Usi.jpg create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/images/image-2-Propositio.jpg create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/images/image-3-$$ # Chapt.jpg create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/images/image-4-From this .jpg create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/images/image-5-$$ We woul.jpg create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/images/image-6-Let $Z = (.jpg create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/images/image-7-Z2 --_ Z3 .jpg create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/images/image-8-Example 4..jpg create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/images/image-9-$$ For a D.jpg create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/manifest.json create mode 100644 apps/api/app/data/demo_documents/stem-statistical-learning/toc_hierarchies.json create mode 100644 apps/api/app/services/demo/official_library_catalog.py diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/chunks.json b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/chunks.json new file mode 100644 index 000000000..3da9fb95d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/chunks.json @@ -0,0 +1,12865 @@ +{ + "chunks": [ + { + "chunk_id": "993d58a9-eab8-5049-a6ed-e3499a6ebaeb", + "type": "image", + "content": "\nAlibaba0\n[images/image-1-Alibaba.jpg]\n", + "path": "images/image-1-Alibaba.jpg", + "metadata": { + "length": 39, + "summary": "image-1\nAlibaba0", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "images/image-1-Alibaba.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "2e84d373-7ada-5b6b-a279-354102e231d8", + "type": "text", + "content": "Alibaba0\n[images/image-1-Alibaba.jpg]\n\nAlibaba\nAlibaba Group Holding Limited\nNYSE: BABA HKEX: 9988 (HKD Counter) 89988 (RMB Counter)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf", + "metadata": { + "length": 132, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Alibaba0", + "Alibaba", + "Group", + "Holding", + "Limited", + "NYSE", + "BABA", + "HKEX", + "9988", + "HKD", + "Counter", + "89988", + "RMB" + ], + "keywords": [], + "connect_to": [ + { + "target": "993d58a9-eab8-5049-a6ed-e3499a6ebaeb", + "relation": "embeds", + "ref": "[images/image-1-Alibaba.jpg]", + "position": { + "start": 9, + "end": 37 + } + } + ] + } + }, + { + "chunk_id": "d0517fb5-8828-5bb7-904c-d961aa10e640", + "type": "text", + "content": "We have one class of Shares, and each holder of our Shares is entitled to one vote per Share. Pursuant to our Articles of Association, for so long as the Partnership Condition (as defined in our Articles of Association) is satisfied, the Alibaba Partnership has the exclusive right to nominate or, in limited situations, appoint, up to a simple majority of the members of our board of directors. These rights are categorized as a weighted voting rights structure, or WVR structure, under the Hong Kong Listing Rules. As a result, we are deemed as a company with a WVR structure.\nAlibaba Partnership will remain in place for the life of the partnership unless our Articles of Association are amended to provide otherwise by a vote of shareholders representing at least 95% of shares that vote at a shareholders meeting. The nomination rights of the Alibaba Partnership will remain in place notwithstanding a change of control or merger of our company. For further details of the Alibaba Partnership, please see the section headed “Directors, Senior Management and Employees – Directors and Senior Management – Alibaba Partnership” in our fiscal year 2025 annual report.\nShareholders and prospective investors are advised to be aware of the potential risks of investing in companies with a WVR structure, in particular, the Alibaba Partnership limits the ability of our shareholders to nominate and elect directors, and the interests of the Alibaba Partnership may conflict with the interests of our shareholders. For further information about the risks associated with our WVR structure, please see the section headed “Risk Factors — Risks Related to Our Corporate Structure” in our fiscal year 2025 annual report. Prospective investors should make the decision to invest in us only after due and careful consideration.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights", + "metadata": { + "length": 1818, + "summary": "The company has a weighted voting rights (WVR) structure where the Alibaba Partnership can nominate or appoint up to a simple majority of board members as long as the Partnership Condition is met. This structure is permanent unless amended by a 95% shareholder vote. The partnership's nomination rights survive a change of control. Shareholders face risks as the partnership limits their ability to elect directors and its interests may conflict with theirs.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "We", + "class", + "Shares", + "holder", + "entitled", + "vote", + "Share", + "Pursuant", + "Articles", + "Association", + "long", + "Partnership", + "Condition", + "defined", + "satisfied", + "Alibaba", + "exclusive", + "nominate", + "limited", + "situations", + "appoint", + "simple", + "majority", + "members", + "board", + "directors", + "These", + "rights", + "categorized", + "weighted", + "voting", + "structure", + "WVR", + "Hong", + "Kong", + "Listing", + "Rules", + "As", + "result", + "deemed", + "company", + "remain", + "place", + "life", + "partnership", + "amended", + "provide", + "shareholders", + "representing", + "95%", + "shares", + "meeting", + "The", + "nomination", + "notwithstanding", + "change", + "control", + "merger", + "For", + "details", + "section", + "headed", + "Directors", + "Senior", + "Management", + "Employees", + "fiscal", + "year", + "2025", + "annual", + "report", + "Shareholders", + "prospective", + "investors", + "advised", + "aware", + "potential", + "risks", + "investing", + "companies", + "limits", + "ability", + "elect", + "interests", + "conflict", + "information", + "Risk", + "Factors", + "Risks", + "Related", + "Our", + "Corporate", + "Structure", + "Prospective", + "make", + "decision", + "invest", + "due", + "careful", + "consideration" + ], + "keywords": [ + "Alibaba Partnership", + "Weighted Voting Rights", + "Shareholder Risks" + ], + "connect_to": [] + } + }, + { + "chunk_id": "31cde054-445f-54e6-8f0a-e5fc954a8fc7", + "type": "text", + "content": "This Interim Report contains translations of certain Renminbi (“RMB”) amounts into U.S. dollars (“US\\$”) and Hong Kong dollars (“HK\\$”) for the convenience of the reader. Unless otherwise stated, all translations of RMB into US\\$ were made at RMB7.1190 to US\\$1.00, the exchange rate on September 30, 2025 as set forth in the H.10 statistical release of the Federal Reserve Board, and all translations of RMB into HK\\$ were made at RMB0.91298 to HK\\$1.00, the middle rate on September 30, 2025 as published by the People’s Bank of China. The percentages stated in this Interim Report are calculated based on the RMB amounts and there may be minor differences due to rounding.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Exchange Rate Information", + "metadata": { + "length": 675, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "This", + "Interim", + "Report", + "translations", + "Renminbi", + "RMB", + "amounts", + "dollars", + "US", + "Hong", + "Kong", + "HK", + "convenience", + "reader", + "Unless", + "stated", + "made", + "RMB7.1190", + "1.00", + "exchange", + "rate", + "September", + "30", + "2025", + "set", + "H.10", + "statistical", + "release", + "Federal", + "Reserve", + "Board", + "RMB0.91298", + "middle", + "published", + "People", + "Bank", + "China", + "The", + "percentages", + "calculated", + "based", + "minor", + "differences", + "due", + "rounding" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d8e74222-0839-546a-b7c9-bdb4d95cfc59", + "type": "table", + "content": "
Six months ended September 30,
20242025YoY % Change
RMBRMBUS$
(in millions, except percentages and per share amounts)
Revenue479,739495,44769,5953%
Income from operations71,23540,3535,668(43)% (2)
Operating margin15%8%
Adjusted EBITDA (1)98,48862,9918,848(36)% (3)
Adjusted EBITDA margin (1)21%13%
Adjusted EBITA (1)85,59647,9176,731(44)% (3)
Adjusted EBITA margin (1)18%10%
Net income67,56962,9948,849(7)% (4)
Net income attributable to ordinary shareholders68,14364,1069,005(6)% (4)
Non-GAAP net income (1)77,20943,8626,161(43)% (3)
Diluted earnings per share (5)3.503.340.47(5)% (4)(6)
Diluted earnings per ADS (5)28.0026.733.75(5)% (4)(6)
Non-GAAP diluted earnings per share (1)(5)3.942.390.34(39)% (3)(6)
Non-GAAP diluted earnings per ADS (1)(5)31.5019.102.68(39)% (3)(6)
", + "path": "tables/table-0 Financial Results.html", + "metadata": { + "length": 1716, + "summary": "table-1\nThe table shows financial data for six months ending Sep 30, 2024 and 2025. Revenue grew 3% to RMB 495,447 million, while income from operations dropped 43%. Adjusted EBITDA fell 36% to RMB 62,991 million.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-0 Financial Results.html", + "keywords": [ + "Revenue", + "Income", + "EBITDA" + ], + "tokens": [] + } + }, + { + "chunk_id": "bd1ea156-fced-5d10-a8c6-3af98937c254", + "type": "text", + "content": "This Interim Report contains forward-looking statements. These statements are made under the “safe harbor” provisions of the U.S. Private Securities Litigation Reform Act of 1995. These forward-looking statements can be identified by terminology such as “may,” “will,” “expect,” “anticipate,” “future,” “aim,” “estimate,” “intend,” “seek,” “plan,” “believe,” “potential,” “continue,” “ongoing,” “target,” “guidance,” “is/are likely to” and similar statements. In addition, statements that are not historical facts, including statements about Alibaba’s strategies and business and operational plans, Alibaba’s beliefs, expectations and guidance regarding the growth of its business, its operating and financial results, return on investments, strategic investments and dispositions and share repurchases, and the business outlook in this Interim Report, are or contain forward-looking statements. Forward-looking statements involve inherent risks and uncertainties. A number of factors could cause actual results to differ materially from those contained in any forward-looking statement, including but not limited to: Alibaba’s ability to compete, innovate and maintain or grow its business; risks associated with sustained investments in Alibaba’s businesses; risks related to strategic transactions; fluctuations in general economic and business conditions in China and globally; uncertainties arising from competition among countries and geopolitical tensions, including national trade, investment, protectionist or other policies and\nexport control, economic or trade sanctions; changes to our shareholder return initiatives; and assumptions underlying or related to any of the foregoing. Further information regarding these and other risks is included in Alibaba’s filings with the U.S. Securities and Exchange Commission and announcements on the website of The Stock Exchange of Hong Kong Limited. All information provided in this Interim Report is as of the date of this Interim Report and are based on assumptions that we believe to be reasonable as of this date, and Alibaba does not undertake any obligation to update any forward-looking statement, except as required under applicable law.\nSix Months Ended September Summary Financial Results\n\n[tables/table-0 Financial Results.html]\n\n(1) See the sections entitled “Non-GAAP Financial Measures” and “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” for more information about the non-GAAP measures referred to within this Interim Report.\n(2) The year-over-year decrease was primarily due to the decrease in adjusted EBITA, partly offset by a one-time provision in the same period last year (See the section entitled “Six Months Ended September Other Financial Results” for more information).\n(3) The year-over-year decreases were mainly attributable to the investment in quick commerce, user experiences, and technology, partly offset by double-digit revenue growth in Alibaba China E-commerce Group, the improved operating results supported by continued growth in Cloud business, as well as enhanced operating efficiencies across various businesses.\n(4) The year-over-year decreases were mainly attributable to the decrease in income from operations, partly offset by the mark-to-market changes from our equity investments, gain from the disposal of local consumer service business of Trendyol, and the decrease in both the impairment of our investments and net exchange loss, while net income attributable to ordinary shareholders and earnings per share/ADS would further take into account the net loss (income) attributable to noncontrolling interests and (accretion) reversal of accretion of mezzanine equity. We excluded non-cash share-based compensation expense, gains/losses of investments, impairment of goodwill and intangible assets, and certain other items from our non-GAAP measurements.\n(5) Each ADS represents eight ordinary shares.\n(6) The year-over-year percentages as stated are calculated based on the exact amount and there may be minor differences from the year-over-year percentages calculated based on the RMB amounts after rounding.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Safe Harbor Statements", + "metadata": { + "length": 4155, + "summary": "This Interim Report contains forward-looking statements under safe harbor provisions, discussing Alibaba's strategies, business plans, and growth expectations. It highlights risks including competition, investments, strategic transactions, economic conditions, geopolitical tensions, and shareholder return changes. The Six Months Ended September Summary Financial Results section references non-GAAP measures, with year-over-year changes driven by factors like adjusted EBITA decrease, investments in quick commerce and technology, revenue growth in Alibaba China E-commerce Group, improved Cloud business results, and mark-to-market changes from equity investments.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "This", + "Interim", + "Report", + "forward", + "statements", + "These", + "made", + "safe", + "harbor", + "provisions", + "Private", + "Securities", + "Litigation", + "Reform", + "Act", + "1995", + "identified", + "terminology", + "expect", + "anticipate", + "future", + "aim", + "estimate", + "intend", + "seek", + "plan", + "potential", + "continue", + "ongoing", + "target", + "guidance", + "similar", + "In", + "addition", + "historical", + "facts", + "including", + "Alibaba", + "strategies", + "business", + "operational", + "plans", + "beliefs", + "expectations", + "growth", + "operating", + "financial", + "results", + "return", + "investments", + "strategic", + "dispositions", + "share", + "repurchases", + "outlook", + "Forward", + "involve", + "inherent", + "risks", + "uncertainties", + "number", + "factors", + "actual", + "differ", + "materially", + "contained", + "statement", + "limited", + "ability", + "compete", + "innovate", + "maintain", + "grow", + "sustained", + "businesses", + "related", + "transactions", + "fluctuations", + "general", + "economic", + "conditions", + "China", + "globally", + "arising", + "competition", + "countries", + "geopolitical", + "tensions", + "national", + "trade", + "investment", + "protectionist", + "policies", + "export", + "control", + "sanctions", + "shareholder", + "initiatives", + "assumptions", + "underlying", + "foregoing", + "Further", + "information", + "included", + "filings", + "Exchange", + "Commission", + "announcements", + "website", + "The", + "Stock", + "Hong", + "Kong", + "Limited", + "All", + "provided", + "date", + "based", + "reasonable", + "undertake", + "obligation", + "update", + "required", + "applicable", + "law", + "Six", + "Months", + "Ended", + "September", + "Summary", + "Financial", + "Results", + "See", + "sections", + "entitled", + "Non", + "GAAP", + "Measures", + "Reconciliations", + "Nearest", + "Comparable", + "measures", + "referred", + "year", + "decrease", + "primarily", + "due", + "adjusted", + "EBITA", + "partly", + "offset", + "time", + "provision", + "period", + "section", + "Other", + "decreases", + "attributable", + "quick", + "commerce", + "user", + "experiences", + "technology", + "double", + "digit", + "revenue", + "Group", + "improved", + "supported", + "continued", + "Cloud", + "enhanced", + "efficiencies", + "income", + "operations", + "mark", + "market", + "equity", + "gain", + "disposal", + "local", + "consumer", + "service", + "Trendyol", + "impairment", + "net", + "exchange", + "loss", + "ordinary", + "shareholders", + "earnings", + "ADS", + "account", + "noncontrolling", + "interests", + "accretion", + "reversal", + "mezzanine", + "We", + "excluded", + "cash", + "compensation", + "expense", + "gains", + "losses", + "goodwill", + "intangible", + "assets", + "items", + "measurements", + "Each", + "represents", + "shares", + "percentages", + "stated", + "calculated", + "exact", + "amount", + "minor", + "differences", + "RMB", + "amounts", + "rounding" + ], + "keywords": [ + "forward-looking", + "financial results", + "risks" + ], + "connect_to": [ + { + "target": "d8e74222-0839-546a-b7c9-bdb4d95cfc59", + "relation": "embeds", + "ref": "[tables/table-0 Financial Results.html]", + "position": { + "start": 2254, + "end": 2293 + } + } + ] + } + }, + { + "chunk_id": "12ea3f3e-43c1-57f5-8a2e-0f8602614578", + "type": "table", + "content": "
Six months ended September 30,
20242025YoY % Change
RMBRMBUS$
(in millions, except percentages)
Alibaba China E-commerce Group:
E-commerce
- Customer management152,755168,17923,62410%
- Direct sales, logistics and others(2)50,23353,3317,4926%
202,988221,51031,1169%
Quick commerce(3)27,51737,6905,29437%
China commerce wholesale11,93813,4501,88913%
Total Alibaba China E-commerce Group242,443272,65038,29912%
Alibaba International Digital Commerce Group:
International commerce retail49,30956,4637,93115%
International commerce wholesale11,65613,0771,83712%
Total Alibaba International Digital Commerce Group60,96569,5409,76814%
Cloud Intelligence Group56,15973,22210,28530%
All others(4)165,837121,56817,077(27)%
Unallocated8881,096154
Inter-segment elimination(46,553)(42,629)(5,988)
Consolidated revenue479,739495,44769,5953%
", + "path": "tables/table-1 Alibaba Revenue.html", + "metadata": { + "length": 1927, + "summary": "table-2\nAlibaba's consolidated revenue for six months ended Sep 2025 was RMB 495,447 million (US$69,595 million), up 3% YoY. China E-commerce grew 12%, International Digital Commerce 14%, and Cloud Intelligence 30%.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-1 Alibaba Revenue.html", + "keywords": [ + "Revenue", + "E-commerce", + "Growth" + ], + "tokens": [] + } + }, + { + "chunk_id": "2a795fd6-a216-5029-86a6-f0f5de762830", + "type": "table", + "content": "
Six months ended September 30,
20242025YoY % Change(3)
RMBRMBUS$
(in millions, except percentages)
Alibaba China E-commerce Group93,08048,8866,867(47)%
Alibaba International Digital Commerce Group(6,611)10314N/A
Cloud Intelligence Group4,9986,55892131%
All others(2,910)(4,785)(672)(64)%
Unallocated(2)(2,142)(1,640)(230)
Inter-segment elimination(819)(1,205)(169)
Consolidated adjusted EBITA85,59647,9176,731(44)%
Less: Non-cash share-based compensation expense(7,775)(6,076)(854)
Less: Amortization and impairment of intangible assets, and others(3,441)(1,488)(209)
Less: Provision for the shareholder class action lawsuits(3,145)--
Income from operations71,23540,3535,668(43)%
", + "path": "tables/table-2 Alibaba EBITA.html", + "metadata": { + "length": 1460, + "summary": "table-3\nAlibaba's consolidated adjusted EBITA for six months ended Sep 2025 was RMB 47,917M, down 44% YoY. Cloud Intelligence Group grew 31%, while China E-commerce dropped 47%.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-2 Alibaba EBITA.html", + "keywords": [ + "EBITA", + "Revenue", + "YoY Change" + ], + "tokens": [] + } + }, + { + "chunk_id": "2db1f9a2-1723-5ae1-ba8e-c1dbd6dc6094", + "type": "text", + "content": "Revenue for the six months ended September 30, 2025 was RMB495,447 million (US\\$69,595 million), an increase of 3% year-over-year compared to RMB479,739 million in the same period of 2024. Excluding revenue from the disposed businesses of Sun Art and Intime, revenue on a like-for-like basis would have grown by 12% year-over-year.\nThe following table sets forth a breakdown of our revenue by segment for the periods indicated:\n\n[tables/table-1 Alibaba Revenue.html]\n\n(1) To advance our “user first” strategy and enhance user experience, during the quarter ended June 30, 2025, we undertook a strategic combination of Taobao and Tmall Group, Ele.me and Fliggy into Alibaba China E-commerce Group. We simplified the financial reporting structure by reclassifying Cainiao, Amap and Digital Media and Entertainment Group (rebranded to Hujing Digital Media and Entertainment Group) into “All others”. The above presentation has been updated to conform with the new reporting structure, as reviewed by our chief operating decision maker.\n(2) Direct sales, logistics and others revenue under Alibaba China E-commerce Group primarily represents direct sales businesses of Tmall Supermarket, Tmall Global and other businesses, where revenue and cost of inventory are recorded on a gross basis within the business group, as well as revenue from logistics services and value-added services.\n(3) Quick commerce revenue represents quick commerce business revenue, including revenue generated through “Taobao Instant Commerce” and the Ele.me app. Quick commerce revenue is net of subsidies that are contra revenue.\n(4) All others include Freshippo, Cainiao, Alibaba Health, Hujing Digital Media and Entertainment Group, Amap, Intelligent Information Platform (which mainly consists of UCWeb and Quark businesses), Lingxi Games, DingTalk and other businesses. The majority of revenue within All others consists of direct sales, where revenue and cost of inventory are recorded on a gross basis, and revenue from logistics services. The decrease was primarily due to the revenue decrease as a result of the disposal of Sun Art and Intime businesses, as well as the decrease in revenue from Cainiao, partly offset by the increase in revenue from Freshippo, Alibaba Health and Amap.\nThe following table sets forth a breakdown of our adjusted EBITA by segment for the periods indicated:\n\n[tables/table-2 Alibaba EBITA.html]\n\n(1) To advance our “user first” strategy and enhance user experience, during the quarter ended June 30, 2025, we undertook a strategic combination of Taobao and Tmall Group, Ele.me and Fliggy into Alibaba China E-commerce Group. We simplified the financial reporting structure by reclassifying Cainiao, Amap and Digital Media and Entertainment Group (rebranded to Hujing Digital Media and Entertainment Group) into “All others”. The above presentation has been updated to conform with the new reporting structure, as reviewed by our chief operating decision maker.\n(2) Unallocated primarily relates to certain costs incurred by corporate functions and other miscellaneous items that are not allocated to individual segments.\n(3) For a more intuitive presentation, widening of loss in YoY% is shown in terms of negative growth rate, and narrowing of loss in YoY% is shown in terms of positive growth rate.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results", + "metadata": { + "length": 3329, + "summary": "Alibaba Group's revenue for the six months ended September 30, 2025, was RMB495,447 million (US$69,595 million), a 3% year-over-year increase from RMB479,739 million in 2024. Excluding disposed businesses (Sun Art and Intime), like-for-like revenue grew 12%. The company restructured its reporting segments, combining Taobao, Tmall, Ele.me, and Fliggy into Alibaba China E-commerce Group, and reclassifying Cainiao, Amap, and Digital Media and Entertainment Group into 'All others'. The text includes tables detailing revenue and adjusted EBITA breakdowns by segment, with notes on direct sales, quick commerce revenue, and segment performance drivers.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Revenue", + "months", + "ended", + "September", + "30", + "2025", + "RMB495", + "447", + "million", + "US", + "69", + "595", + "increase", + "3%", + "year", + "compared", + "RMB479", + "739", + "period", + "2024", + "Excluding", + "revenue", + "disposed", + "businesses", + "Sun", + "Art", + "Intime", + "basis", + "grown", + "12%", + "The", + "table", + "sets", + "breakdown", + "segment", + "periods", + "To", + "advance", + "user", + "strategy", + "enhance", + "experience", + "quarter", + "June", + "undertook", + "strategic", + "combination", + "Taobao", + "Tmall", + "Group", + "Ele", + "Fliggy", + "Alibaba", + "China", + "commerce", + "We", + "simplified", + "financial", + "reporting", + "structure", + "reclassifying", + "Cainiao", + "Amap", + "Digital", + "Media", + "Entertainment", + "rebranded", + "Hujing", + "All", + "presentation", + "updated", + "conform", + "reviewed", + "chief", + "operating", + "decision", + "maker", + "Direct", + "sales", + "logistics", + "primarily", + "represents", + "direct", + "Supermarket", + "Global", + "cost", + "inventory", + "recorded", + "gross", + "business", + "group", + "services", + "added", + "Quick", + "quick", + "including", + "generated", + "Instant", + "Commerce", + "app", + "net", + "subsidies", + "contra", + "include", + "Freshippo", + "Health", + "Intelligent", + "Information", + "Platform", + "consists", + "UCWeb", + "Quark", + "Lingxi", + "Games", + "DingTalk", + "majority", + "decrease", + "due", + "result", + "disposal", + "partly", + "offset", + "adjusted", + "EBITA", + "Unallocated", + "relates", + "costs", + "incurred", + "corporate", + "functions", + "miscellaneous", + "items", + "allocated", + "individual", + "segments", + "For", + "intuitive", + "widening", + "loss", + "YoY%", + "shown", + "terms", + "negative", + "growth", + "rate", + "narrowing", + "positive" + ], + "keywords": [ + "Revenue", + "Segments", + "EBITA" + ], + "connect_to": [ + { + "target": "12ea3f3e-43c1-57f5-8a2e-0f8602614578", + "relation": "embeds", + "ref": "[tables/table-1 Alibaba Revenue.html]", + "position": { + "start": 429, + "end": 466 + } + }, + { + "target": "2a795fd6-a216-5029-86a6-f0f5de762830", + "relation": "embeds", + "ref": "[tables/table-2 Alibaba EBITA.html]", + "position": { + "start": 2370, + "end": 2405 + } + } + ] + } + }, + { + "chunk_id": "5ab9ab8a-5525-5725-ba30-3341dc3ee505", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ee8b513d-5f91-570b-949c-9702cd2f8d95", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group/(i) Segment revenue", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cf9dd8ee-22ff-5276-84f1-b8911de9c041", + "type": "text", + "content": "Revenue from our E-commerce business in the six months ended September 30, 2025 was RMB221,510 million (US\\$31,116 million), an increase of 9% compared to RMB202,988 million in the same period of 2024.\nCustomer management revenue increased by 10% year-over-year, primarily due to the improvement of take rate.\nDirect sales, logistics and others revenue under E-commerce business in the six months ended September 30, 2025 was RMB53,331 million (US\\$7,492 million), an increase of 6% compared to RMB50,233 million in the same period of 2024, primarily driven by the increase in revenue from logistics services and value-added services, partly offset by the decrease in revenue from certain direct sales businesses.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group/(i) Segment revenue/• E-commerce Business", + "metadata": { + "length": 713, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Revenue", + "commerce", + "business", + "months", + "ended", + "September", + "30", + "2025", + "RMB221", + "510", + "million", + "US", + "31", + "116", + "increase", + "9%", + "compared", + "RMB202", + "988", + "period", + "2024", + "Customer", + "management", + "revenue", + "increased", + "10%", + "year", + "primarily", + "due", + "improvement", + "rate", + "Direct", + "sales", + "logistics", + "RMB53", + "331", + "492", + "6%", + "RMB50", + "233", + "driven", + "services", + "added", + "partly", + "offset", + "decrease", + "direct", + "businesses" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "eee7627f-5ba0-549e-83bb-1e16f9d28940", + "type": "text", + "content": "Revenue from our Quick commerce business in the six months ended September 30, 2025 was RMB37,690 million (US\\$5,294 million), an increase of 37% compared to RMB27,517 million in the same period of 2024, mainly due to order growth as a result of the rollout of “Taobao Instant Commerce” at the end of April 2025.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group/(i) Segment revenue/• Quick Commerce Business", + "metadata": { + "length": 312, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Revenue", + "Quick", + "commerce", + "business", + "months", + "ended", + "September", + "30", + "2025", + "RMB37", + "690", + "million", + "US", + "294", + "increase", + "37%", + "compared", + "RMB27", + "517", + "period", + "2024", + "due", + "order", + "growth", + "result", + "rollout", + "Taobao", + "Instant", + "Commerce", + "end", + "April" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d2aa7def-41e8-50ac-a610-9820b5f8db9e", + "type": "text", + "content": "Revenue from our China commerce wholesale business in the six months ended September 30, 2025 was RMB13,450 million (US\\$1,889 million), an increase of 13% compared to RMB11,938 million in the same period of 2024, primarily due to an increase in revenue from value-added services provided to paying members.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group/(i) Segment revenue/• China Commerce Wholesale Business", + "metadata": { + "length": 307, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Revenue", + "China", + "commerce", + "wholesale", + "business", + "months", + "ended", + "September", + "30", + "2025", + "RMB13", + "450", + "million", + "US", + "889", + "increase", + "13%", + "compared", + "RMB11", + "938", + "period", + "2024", + "primarily", + "due", + "revenue", + "added", + "services", + "provided", + "paying", + "members" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ad6c0695-d324-56dc-b57a-55076dbe60fc", + "type": "text", + "content": "Alibaba China E-commerce Group adjusted EBITA decreased by 47% to RMB48,886 million (US\\$6,867 million) in the six months ended September 30, 2025, compared to RMB93,080 million in the same period of 2024, primarily due to the investment in quick commerce, user experiences, and technology, partly offset by double-digit revenue growth in Alibaba China E-commerce Group.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group/(ii) Segment adjusted EBITA", + "metadata": { + "length": 370, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Alibaba", + "China", + "commerce", + "Group", + "adjusted", + "EBITA", + "decreased", + "47%", + "RMB48", + "886", + "million", + "US", + "867", + "months", + "ended", + "September", + "30", + "2025", + "compared", + "RMB93", + "080", + "period", + "2024", + "primarily", + "due", + "investment", + "quick", + "user", + "experiences", + "technology", + "partly", + "offset", + "double", + "digit", + "revenue", + "growth" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6544088b-bb4c-5815-9b60-5e8dd8e6ea3a", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba International Digital Commerce Group", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e13f3f94-6d80-5a02-b9b1-b0a03feff0a9", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba International Digital Commerce Group/(i) Segment revenue", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1f55f4f3-aeb8-5fc5-a123-d940b3b7f768", + "type": "text", + "content": "Revenue from our International commerce retail business in the six months ended September 30, 2025 was RMB56,463 million (US\\$7,931 million), an increase of 15% compared to RMB49,309 million in the same period of 2024, primarily driven by the increase in revenue contributed by AliExpress and other international businesses. As certain of our international businesses generate revenue in local currencies while our reporting currency is Renminbi, AIDC's revenue is affected by exchange rate fluctuations.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba International Digital Commerce Group/(i) Segment revenue/• International Commerce Retail Business", + "metadata": { + "length": 504, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Revenue", + "International", + "commerce", + "retail", + "business", + "months", + "ended", + "September", + "30", + "2025", + "RMB56", + "463", + "million", + "US", + "931", + "increase", + "15%", + "compared", + "RMB49", + "309", + "period", + "2024", + "primarily", + "driven", + "revenue", + "contributed", + "AliExpress", + "international", + "businesses", + "As", + "generate", + "local", + "currencies", + "reporting", + "currency", + "Renminbi", + "AIDC", + "affected", + "exchange", + "rate", + "fluctuations" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7f4a419e-7353-5897-9b3e-13865d071d1f", + "type": "text", + "content": "Revenue from our International commerce wholesale business in the six months ended September 30, 2025 was RMB13,077 million (US\\$1,837 million), an increase of 12% compared to RMB11,656 million in the same period of 2024, primarily due to an increase in revenue generated by cross-border related value-added services.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba International Digital Commerce Group/(i) Segment revenue/• International Commerce Wholesale Business", + "metadata": { + "length": 317, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Revenue", + "International", + "commerce", + "wholesale", + "business", + "months", + "ended", + "September", + "30", + "2025", + "RMB13", + "077", + "million", + "US", + "837", + "increase", + "12%", + "compared", + "RMB11", + "656", + "period", + "2024", + "primarily", + "due", + "revenue", + "generated", + "cross", + "border", + "related", + "added", + "services" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5718a095-76b6-5deb-b0bc-dcdb0b5aa205", + "type": "text", + "content": "Alibaba International Digital Commerce Group adjusted EBITA was a profit of RMB103 million (US\\$14 million) in the six months ended September 30, 2025, compared to a loss of RMB6,611 million in the same period of 2024, primarily due to significant improvement in AliExpress' operating efficiency, and enhanced efficiency across various businesses.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba International Digital Commerce Group/(ii) Segment adjusted EBITA", + "metadata": { + "length": 347, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Alibaba", + "International", + "Digital", + "Commerce", + "Group", + "adjusted", + "EBITA", + "profit", + "RMB103", + "million", + "US", + "14", + "months", + "ended", + "September", + "30", + "2025", + "compared", + "loss", + "RMB6", + "611", + "period", + "2024", + "primarily", + "due", + "significant", + "improvement", + "AliExpress", + "operating", + "efficiency", + "enhanced", + "businesses" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0d2966b9-f8ed-5a9c-91e8-0793d878983c", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Cloud Intelligence Group", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b556ada4-a4c3-51c6-bc96-a9fa87a6f234", + "type": "text", + "content": "Revenue from Cloud Intelligence Group was RMB73,222 million (US\\$10,285 million) in the six months ended September 30, 2025, an increase of 30% compared to RMB56,159 million in the same period of 2024. Overall revenue excluding Alibaba-consolidated subsidiaries increased by 28% year-over-year, primarily driven by public cloud revenue growth, including the increasing adoption of AI-related products.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Cloud Intelligence Group/(i) Segment revenue", + "metadata": { + "length": 401, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Revenue", + "Cloud", + "Intelligence", + "Group", + "RMB73", + "222", + "million", + "US", + "10", + "285", + "months", + "ended", + "September", + "30", + "2025", + "increase", + "30%", + "compared", + "RMB56", + "159", + "period", + "2024", + "Overall", + "revenue", + "excluding", + "Alibaba", + "consolidated", + "subsidiaries", + "increased", + "28%", + "year", + "primarily", + "driven", + "public", + "cloud", + "growth", + "including", + "increasing", + "adoption", + "AI", + "related", + "products" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cd4dd494-e5cc-5c3b-939d-9be1eb8f70bb", + "type": "text", + "content": "Cloud Intelligence Group adjusted EBITA increased by 31% to RMB6,558 million (US\\$921 million) in the six months ended September 30, 2025, compared to RMB4,998 million in the same period of 2024, primarily due to revenue growth and improving operating efficiency, partly offset by the increasing investments in customer growth and technology innovation.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Cloud Intelligence Group/(ii) Segment adjusted EBITA", + "metadata": { + "length": 353, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Cloud", + "Intelligence", + "Group", + "adjusted", + "EBITA", + "increased", + "31%", + "RMB6", + "558", + "million", + "US", + "921", + "months", + "ended", + "September", + "30", + "2025", + "compared", + "RMB4", + "998", + "period", + "2024", + "primarily", + "due", + "revenue", + "growth", + "improving", + "operating", + "efficiency", + "partly", + "offset", + "increasing", + "investments", + "customer", + "technology", + "innovation" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ab0b4c0b-7823-56b9-832c-a4500ab19ab6", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/All Others", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "25df0fe4-8ee9-5859-981b-3e914c0ce3d0", + "type": "text", + "content": "Revenue from All others segment was RMB121,568 million (US\\$17,077 million) in the six months ended September 30, 2025, a decrease of 27% compared to RMB165,837 million in the same period of 2024, primarily due to the revenue decrease as a result of the disposal of Sun Art and Intime businesses, as well as the decrease in revenue from Cainiao, partly offset by the increase in revenue from Freshippo, Alibaba Health and Amap.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/All Others/(i) Segment revenue", + "metadata": { + "length": 427, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Revenue", + "All", + "segment", + "RMB121", + "568", + "million", + "US", + "17", + "077", + "months", + "ended", + "September", + "30", + "2025", + "decrease", + "27%", + "compared", + "RMB165", + "837", + "period", + "2024", + "primarily", + "due", + "revenue", + "result", + "disposal", + "Sun", + "Art", + "Intime", + "businesses", + "Cainiao", + "partly", + "offset", + "increase", + "Freshippo", + "Alibaba", + "Health", + "Amap" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4e2788b6-2413-5828-bd82-0473726de7ae", + "type": "text", + "content": "Adjusted EBITA from All others segment in the six months ended September 30, 2025 was a loss of RMB4,785 million (US\\$672 million), compared to a loss of RMB2,910 million in the same period of 2024, primarily due to the increased investment in technology businesses, partly offset by the improved operating results of Freshippo, Hujing Digital Media and Entertainment Group and Alibaba Health.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/All Others/(ii) Segment adjusted EBITA", + "metadata": { + "length": 393, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Adjusted", + "EBITA", + "All", + "segment", + "months", + "ended", + "September", + "30", + "2025", + "loss", + "RMB4", + "785", + "million", + "US", + "672", + "compared", + "RMB2", + "910", + "period", + "2024", + "primarily", + "due", + "increased", + "investment", + "technology", + "businesses", + "partly", + "offset", + "improved", + "operating", + "results", + "Freshippo", + "Hujing", + "Digital", + "Media", + "Entertainment", + "Group", + "Alibaba", + "Health" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "eac163cf-ee25-5494-b601-6a25e833c8f6", + "type": "table", + "content": "
Six months ended September 30,% of Revenue YoY change
20242025
RMB% of RevenueRMBUS$% of Revenue
(in millions, except percentages)
Costs and expenses:
Cost of revenue290,13560.5%287,21040,34458.0%(2.5)%
Product development expenses27,5555.7%32,0964,5096.5%0.8%
Sales and marketing expenses65,16713.6%119,67416,81124.2%10.6%
General and administrative expenses23,0574.8%14,7782,0763.0%(1.8)%
Amortization and impairment of intangible assets3,4410.7%1,6332290.3%(0.4)%
Total costs and expenses409,355455,39163,969
Share-based compensation expense:
Cost of revenue1,2050.3%9131280.2%(0.1)%
Product development expenses3,5600.7%2,8624020.6%(0.1)%
Sales and marketing expenses9480.2%9581350.2%0.0%
General and administrative expenses2,5640.5%2,1373000.4%(0.1)%
Total share-based compensation expense(1)8,2776,870965
Costs and expenses excluding share-based compensation expense:
Cost of revenue288,93060.2%286,29740,21657.8%(2.4)%
Product development expenses23,9955.0%29,2344,1075.9%0.9%
Sales and marketing expenses64,21913.4%118,71616,67624.0%10.6%
General and administrative expenses20,4934.3%12,6411,7762.6%(1.7)%
Amortization and impairment of intangible assets3,4410.7%1,6332290.3%(0.4)%
Total costs and expenses excluding share-based compensation expense401,078448,52163,004
", + "path": "tables/table-3 Costs & Expenses.html", + "metadata": { + "length": 2790, + "summary": "table-4\nThe table shows costs and expenses for six months ended Sep 30, 2024 and 2025. Total costs rose from 409,355 to 455,391 RMB. Sales and marketing expenses increased significantly by 10.6% of revenue.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-3 Costs & Expenses.html", + "keywords": [ + "Cost of revenue", + "Expenses", + "YoY change" + ], + "tokens": [] + } + }, + { + "chunk_id": "1fe38c48-14bf-5579-a964-aef1869b0708", + "type": "table", + "content": "
Six months ended September 30,
20242025YoY % Change
RMBRMBUS$
(in millions, except percentages)
By type of awards:
Alibaba Group share-based awards(1)5,8774,664655(21)%
Others(2)2,4002,206310(8)%
Total share-based compensation expense(3)8,2776,870965(17)%
", + "path": "tables/table-4 Share-based Compensation.html", + "metadata": { + "length": 641, + "summary": "table-5\nAlibaba Group's share-based compensation expense decreased 17% YoY to RMB 6,870 million for six months ended Sep 2025, with Alibaba awards down 21% and others down 8%.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-4 Share-based Compensation.html", + "keywords": [ + "share-based awards", + "compensation expense", + "YoY change" + ], + "tokens": [] + } + }, + { + "chunk_id": "a99c8b09-3111-521d-817f-bd6bc8cb799a", + "type": "text", + "content": "Costs and Expenses\nThe following tables set forth a breakdown of our costs and expenses, share-based compensation expense, and costs and expenses excluding share-based compensation expense by function for the periods indicated:\n\n[tables/table-3 Costs & Expenses.html]\n\n(1) This includes both cash and non-cash share-based compensation expenses.\nCost of revenue – Cost of revenue in the six months ended September 30, 2025 was RMB287,210 million (US\\$40,344 million), or 58.0% of revenue, compared to RMB290,135 million, or 60.5% of revenue, in the same period of 2024. Without the effect of share-based compensation expense, cost of revenue as a percentage of revenue would have decreased from 60.2% in the same period of 2024 to 57.8% in the six months ended September 30, 2025, primarily due to the disposal of Sun Art and Intime businesses, decrease in scale of low margin direct sales businesses, improvement in monetization and operating efficiency, partly offset by higher logistics cost driven by the growth in our quick commerce business.\nProduct development expenses – Product development expenses in the six months ended September 30, 2025 were RMB32,096 million (US\\$4,509 million), or 6.5% of revenue, compared to RMB27,555 million, or 5.7% of revenue, in the same period of 2024. Without the effect of share-based compensation expense, product development expenses as a percentage of revenue would have increased from 5.0% in the same period of 2024 to 5.9% in the six months ended September 30, 2025, primarily attributable to our increased investment in technology development.\nSales and marketing expenses – Sales and marketing expenses in the six months ended September 30, 2025 were RMB119,674 million (US\\$16,811 million), or 24.2% of revenue, compared to RMB65,167 million, or 13.6% of revenue, in the same period of 2024. Without the effect of share-based compensation expense, sales and marketing expenses as a percentage of revenue would have increased from 13.4% in the same period of 2024 to 24.0% in the six months ended September 30, 2025, primarily attributable to the investment in user experiences of Alibaba China E-commerce Group.\nGeneral and administrative expenses – General and administrative expenses in the six months ended September 30, 2025 were RMB14,778 million (US\\$2,076 million), or 3.0% of revenue, compared to RMB23,057 million, or 4.8% of revenue, in the same period of 2024. Without the effect of share-based compensation expense, general and administrative expenses as a percentage of revenue would have decreased from 4.3% in the same period of 2024 to 2.6% in the six months ended September 30, 2025, primarily due to a one-time provision for the shareholder class action lawsuits in the same period last year and our improved cost control measures.\nShare-based compensation expense – Total share-based compensation expense included in the cost and expense items above in the six months ended September 30, 2025 was RMB6,870 million (US\\$965 million), compared to RMB8,277 million in the same period of 2024.\nThe following table sets forth our analysis of share-based compensation expense for the quarters indicated by type of share-based awards:\n\n[tables/table-4 Share-based Compensation.html]\n\nShare-based compensation expense decreased in the six months ended September 30, 2025 compared to the same period of 2024. This decrease was primarily due to the decrease in the number of Alibaba Group share-based awards granted as we have increased the proportion of long-term cash incentives granted after considering the macroeconomic environment and the general trends in the talent market.\nWe expect that our share-based compensation expense will continue to be affected by changes in the fair value of the underlying awards and the quantity of awards we grant in the future.\nAmortization and impairment of intangible assets – Amortization and impairment of intangible assets in the six months ended September 30, 2025 was RMB1,633 million (US\\$229 million), a decrease of 53% from RMB3,441 million in the same period of 2024, primarily due to full amortization of certain intangible assets.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses", + "metadata": { + "length": 4173, + "summary": "The text details Alibaba's costs and expenses for the six months ended September 30, 2025, compared to the same period in 2024. Cost of revenue decreased to 58.0% of revenue from 60.5%, driven by disposal of Sun Art and Intime, lower direct sales, and improved efficiency. Product development expenses rose to 6.5% of revenue from 5.7% due to increased technology investment. Sales and marketing expenses surged to 24.2% from 13.6%, mainly from investment in Alibaba China E-commerce Group. General and administrative expenses fell to 3.0% from 4.8%, aided by a prior-year lawsuit provision and cost controls. Total share-based compensation expense decreased to RMB6,870 million from RMB8,277 million, due to fewer awards granted. Amortization and impairment of intangible assets dropped 53% to RMB1,633 million from full amortization of certain assets.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Costs", + "Expenses", + "The", + "tables", + "set", + "breakdown", + "costs", + "expenses", + "share", + "based", + "compensation", + "expense", + "excluding", + "function", + "periods", + "This", + "includes", + "cash", + "Cost", + "revenue", + "months", + "ended", + "September", + "30", + "2025", + "RMB287", + "210", + "million", + "US", + "40", + "344", + "58.0%", + "compared", + "RMB290", + "135", + "60.5%", + "period", + "2024", + "Without", + "effect", + "cost", + "percentage", + "decreased", + "60.2%", + "57.8%", + "primarily", + "due", + "disposal", + "Sun", + "Art", + "Intime", + "businesses", + "decrease", + "scale", + "low", + "margin", + "direct", + "sales", + "improvement", + "monetization", + "operating", + "efficiency", + "partly", + "offset", + "higher", + "logistics", + "driven", + "growth", + "quick", + "commerce", + "business", + "Product", + "development", + "RMB32", + "096", + "509", + "6.5%", + "RMB27", + "555", + "5.7%", + "product", + "increased", + "5.0%", + "5.9%", + "attributable", + "investment", + "technology", + "Sales", + "marketing", + "RMB119", + "674", + "16", + "811", + "24.2%", + "RMB65", + "167", + "13.6%", + "13.4%", + "24.0%", + "user", + "experiences", + "Alibaba", + "China", + "Group", + "General", + "administrative", + "RMB14", + "778", + "076", + "3.0%", + "RMB23", + "057", + "4.8%", + "general", + "4.3%", + "2.6%", + "time", + "provision", + "shareholder", + "class", + "action", + "lawsuits", + "year", + "improved", + "control", + "measures", + "Share", + "Total", + "included", + "items", + "RMB6", + "870", + "965", + "RMB8", + "277", + "table", + "sets", + "analysis", + "quarters", + "type", + "awards", + "number", + "granted", + "proportion", + "long", + "term", + "incentives", + "macroeconomic", + "environment", + "trends", + "talent", + "market", + "We", + "expect", + "continue", + "affected", + "fair", + "underlying", + "quantity", + "grant", + "future", + "Amortization", + "impairment", + "intangible", + "assets", + "RMB1", + "633", + "229", + "53%", + "RMB3", + "441", + "full", + "amortization" + ], + "keywords": [ + "costs", + "expenses", + "share-based compensation" + ], + "connect_to": [ + { + "target": "eac163cf-ee25-5494-b601-6a25e833c8f6", + "relation": "embeds", + "ref": "[tables/table-3 Costs & Expenses.html]", + "position": { + "start": 229, + "end": 267 + } + }, + { + "target": "1fe38c48-14bf-5579-a964-aef1869b0708", + "relation": "embeds", + "ref": "[tables/table-4 Share-based Compensation.html]", + "position": { + "start": 3199, + "end": 3245 + } + } + ] + } + }, + { + "chunk_id": "5e4ae1be-44c8-5035-809b-d56c2b426842", + "type": "text", + "content": "Income from operations in the six months ended September 30, 2025 was RMB40,353 million (US\\$5,668 million), or 8% of revenue, a decrease of 43% compared to RMB71,235 million, or 15% of revenue, in the same period of 2024, primarily due to the decrease in adjusted EBITA, partly offset by a one-time provision in the same period last year.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Income from operations and operating margin", + "metadata": { + "length": 339, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Income", + "operations", + "months", + "ended", + "September", + "30", + "2025", + "RMB40", + "353", + "million", + "US", + "668", + "8%", + "revenue", + "decrease", + "43%", + "compared", + "RMB71", + "235", + "15%", + "period", + "2024", + "primarily", + "due", + "adjusted", + "EBITA", + "partly", + "offset", + "time", + "provision", + "year" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "dcb535c5-0381-5a35-8831-de3e03f5f583", + "type": "text", + "content": "Adjusted EBITDA decreased 36% year-over-year to RMB62,991 million (US\\$8,848 million) in the six months ended September 30, 2025, compared to RMB98,488 million in the same period of 2024. Adjusted EBITA decreased 44% year-over-year to RMB47,917 million (US\\$6,731 million) in the six months ended September 30, 2025, compared to RMB85,596 million in the same period of 2024, primarily attributable to the investment in quick commerce, user experiences, and technology, partly offset by double-digit revenue growth in Alibaba China E-commerce Group, the improved operating results supported by continued growth in Cloud business, as well as enhanced operating efficiencies across various businesses. A reconciliation of net income to adjusted EBITDA and adjusted EBITA is included in the section entitled “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” of this Interim Report.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Adjusted EBITDA and Adjusted EBITA", + "metadata": { + "length": 911, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Adjusted", + "EBITDA", + "decreased", + "36%", + "year", + "RMB62", + "991", + "million", + "US", + "848", + "months", + "ended", + "September", + "30", + "2025", + "compared", + "RMB98", + "488", + "period", + "2024", + "EBITA", + "44%", + "RMB47", + "917", + "731", + "RMB85", + "596", + "primarily", + "attributable", + "investment", + "quick", + "commerce", + "user", + "experiences", + "technology", + "partly", + "offset", + "double", + "digit", + "revenue", + "growth", + "Alibaba", + "China", + "Group", + "improved", + "operating", + "results", + "supported", + "continued", + "Cloud", + "business", + "enhanced", + "efficiencies", + "businesses", + "reconciliation", + "net", + "income", + "adjusted", + "included", + "section", + "entitled", + "Reconciliations", + "Non", + "GAAP", + "Measures", + "Nearest", + "Comparable", + "Interim", + "Report" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "02f885bb-a810-57b2-bf1d-ded78c771f2d", + "type": "text", + "content": "Adjusted EBITA by segment as well as a reconciliation of income from operations to adjusted EBITA are set forth in the section entitled “Six Months Ended September Segment Results” above.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Adjusted EBITA by segment", + "metadata": { + "length": 187, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Adjusted", + "EBITA", + "segment", + "reconciliation", + "income", + "operations", + "adjusted", + "set", + "section", + "entitled", + "Six", + "Months", + "Ended", + "September", + "Segment", + "Results" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9374b98e-c2a1-57f8-8d6b-fc62b8d94f52", + "type": "text", + "content": "Interest and investment income, net in the six months ended September 30, 2025 was RMB37,468 million (US\\$5,263 million), an increase of 119% compared to RMB17,129 million in the same period of 2024, primarily due to mark-to-market changes from our equity investments, gain from the disposal of local consumer service business of Trendyol, and the decrease in impairment of our investments.\nThe above-mentioned investment gains and losses were excluded from our non-GAAP net income.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Interest and investment income, net", + "metadata": { + "length": 482, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Interest", + "investment", + "income", + "net", + "months", + "ended", + "September", + "30", + "2025", + "RMB37", + "468", + "million", + "US", + "263", + "increase", + "119%", + "compared", + "RMB17", + "129", + "period", + "2024", + "primarily", + "due", + "mark", + "market", + "equity", + "investments", + "gain", + "disposal", + "local", + "consumer", + "service", + "business", + "Trendyol", + "decrease", + "impairment", + "The", + "mentioned", + "gains", + "losses", + "excluded", + "GAAP" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ab08db16-7ad2-596c-b5cf-c1604f04f57a", + "type": "text", + "content": "Other (expense) income, net in the six months ended September 30, 2025 was an income of RMB1,329 million (US\\$187 million), compared to expense of RMB1,221 million in the same period of 2024, was primarily due to the decrease in net exchange loss in the six months ended September 30, 2025 compared to the same period last year, arising from the exchange rate fluctuation between Renminbi and U.S. dollar.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Other (expense) income, net", + "metadata": { + "length": 405, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Other", + "expense", + "income", + "net", + "months", + "ended", + "September", + "30", + "2025", + "RMB1", + "329", + "million", + "US", + "187", + "compared", + "221", + "period", + "2024", + "primarily", + "due", + "decrease", + "exchange", + "loss", + "year", + "arising", + "rate", + "fluctuation", + "Renminbi", + "dollar" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "26ffbf71-d44e-5531-b1f3-20d3979d180d", + "type": "text", + "content": "Income tax expenses in the six months ended September 30, 2025 were RMB14,415 million (US\\$2,024 million), compared to RMB17,442 million in the same period of 2024.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Income tax expenses", + "metadata": { + "length": 164, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Income", + "tax", + "expenses", + "months", + "ended", + "September", + "30", + "2025", + "RMB14", + "415", + "million", + "US", + "024", + "compared", + "RMB17", + "442", + "period", + "2024" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d348d07f-85ac-51cb-bec5-639f447a2809", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Share of profit (loss) of equity method investees
- Ant Group6,3954,280601
- Others(1,334)1,126158
Impairment loss(2,157)(5)(1)
$Others^{(1)}$ (421)(2,147)(301)
Total2,4833,254457
", + "path": "tables/table-5 Equity Investee Results.html", + "metadata": { + "length": 682, + "summary": "table-6\nThe table shows share of profit from equity investees for six months ending Sep 30, 2024 and 2025. Ant Group contributed RMB 6,395M (2024) and RMB 4,280M (2025). Total profit was RMB 2,483M (2024) and RMB 3,254M (2025).", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-5 Equity Investee Results.html", + "keywords": [ + "Ant Group", + "Impairment", + "Share of Profit" + ], + "tokens": [] + } + }, + { + "chunk_id": "190ebc3a-f4b0-5153-9a2a-19b34f78b3ea", + "type": "text", + "content": "Share of results of equity method investees in the six months ended September 30, 2025 was RMB3,254 million (US\\$457 million), an increase of 31% compared to RMB2,483 million in the same period of 2024. The following table sets forth a breakdown of share of results of equity method investees for the periods indicated:\n\n[tables/table-5 Equity Investee Results.html]\n\n(1) “Others” mainly include basis differences arising from equity method investees, share-based compensation expense related to share-based awards granted to employees of our equity method investees, as well as gain or loss arising from the deemed disposal of the equity method investees.\nWe record our share of results of all equity method investees one quarter in arrears. The share of net profit of other equity method investees recorded in the six months ended September 30 2025, compared to the share of net losses in the same period last year, was primarily attributable to the overall improvement in the financial performance of our equity method investees. This was partly offset by the decrease in share of profit of Ant Group, which was mainly attributable to investments in new growth initiatives and technologies.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Share of results of equity method investees", + "metadata": { + "length": 1193, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Share", + "results", + "equity", + "method", + "investees", + "months", + "ended", + "September", + "30", + "2025", + "RMB3", + "254", + "million", + "US", + "457", + "increase", + "31%", + "compared", + "RMB2", + "483", + "period", + "2024", + "The", + "table", + "sets", + "breakdown", + "share", + "periods", + "Others", + "include", + "basis", + "differences", + "arising", + "based", + "compensation", + "expense", + "related", + "awards", + "granted", + "employees", + "gain", + "loss", + "deemed", + "disposal", + "We", + "record", + "quarter", + "arrears", + "net", + "profit", + "recorded", + "losses", + "year", + "primarily", + "attributable", + "improvement", + "financial", + "performance", + "This", + "partly", + "offset", + "decrease", + "Ant", + "Group", + "investments", + "growth", + "initiatives", + "technologies" + ], + "keywords": [], + "connect_to": [ + { + "target": "d348d07f-85ac-51cb-bec5-639f447a2809", + "relation": "embeds", + "ref": "[tables/table-5 Equity Investee Results.html]", + "position": { + "start": 321, + "end": 366 + } + } + ] + } + }, + { + "chunk_id": "87d13fbd-2507-5e4a-b875-91e2b0bf4871", + "type": "text", + "content": "Our net income in the six months ended September 30, 2025 was RMB62,994 million (US\\$8,849 million), compared to RMB67,569 million in the same period of 2024, primarily attributable to the decrease in income from operations, partly offset by the mark-to-market changes from our equity investments, gain from the disposal of local consumer service business of Trendyol and the decrease in both the impairment of our investments and net exchange loss.\nExcluding non-cash share-based compensation expense, gains/losses of investments, impairment of goodwill and intangible assets, and certain other items, non-GAAP net income in the six months ended September 30, 2025 was RMB43,862 million (US\\$6,161 million), a decrease of 43% compared to RMB77,209 million in the same period of 2024, primarily attributable to the investment in quick commerce, user experiences, and technology, partly offset by double-digit revenue growth in Alibaba China E-commerce Group, the improved operating results supported by continued growth in Cloud business, as well as enhanced operating efficiencies across various businesses. A reconciliation of net income to non-GAAP net income is included in the section entitled “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” of this Interim Report.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Net income and Non-GAAP net income", + "metadata": { + "length": 1306, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "net", + "income", + "months", + "ended", + "September", + "30", + "2025", + "RMB62", + "994", + "million", + "US", + "849", + "compared", + "RMB67", + "569", + "period", + "2024", + "primarily", + "attributable", + "decrease", + "operations", + "partly", + "offset", + "mark", + "market", + "equity", + "investments", + "gain", + "disposal", + "local", + "consumer", + "service", + "business", + "Trendyol", + "impairment", + "exchange", + "loss", + "Excluding", + "cash", + "share", + "based", + "compensation", + "expense", + "gains", + "losses", + "goodwill", + "intangible", + "assets", + "items", + "GAAP", + "RMB43", + "862", + "161", + "43%", + "RMB77", + "209", + "investment", + "quick", + "commerce", + "user", + "experiences", + "technology", + "double", + "digit", + "revenue", + "growth", + "Alibaba", + "China", + "Group", + "improved", + "operating", + "results", + "supported", + "continued", + "Cloud", + "enhanced", + "efficiencies", + "businesses", + "reconciliation", + "included", + "section", + "entitled", + "Reconciliations", + "Non", + "Measures", + "Nearest", + "Comparable", + "Interim", + "Report" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0fdde5ee-22df-50db-ad40-e1f0b07ebf8a", + "type": "text", + "content": "Net income attributable to ordinary shareholders in the six months ended September 30, 2025 was RMB64,106 million (US\\$9,005 million), compared to RMB68,143 million in the same period of 2024, primarily attributable to the decrease in income from operations, partly offset by the mark-to-market changes from our equity investments, gain from the disposal of local consumer service business of Trendyol and the decrease in both the impairment of our investments and net exchange loss.\nDiluted earnings per ADS/share and non-GAAP diluted earnings per ADS/share\nDiluted earnings per ADS in the six months ended September 30, 2025 was RMB26.73 (US\\$3.75), compared to RMB28.00 in the same period of 2024. Excluding non-cash share-based compensation expense, gains/losses of investments, impairment of goodwill and intangible assets, and certain other items, non-GAAP diluted earnings per ADS in the six months ended September 30, 2025 was RMB19.10 (US\\$2.68), a decrease of 39% compared to RMB31.50 in the same period of 2024.\nDiluted earnings per share in the six months ended September 30, 2025 was RMB3.34 (US\\$0.47 or HK\\$3.66), compared to RMB3.50 in the same period of 2024. Excluding non-cash share-based compensation expense, gains/losses of investments, impairment of goodwill and intangible assets, and certain other items, non-GAAP diluted earnings per share in the six months ended September 30, 2025 was RMB2.39 (US\\$0.34 or HK\\$2.62), a decrease of 39% compared to RMB3.94 in the same period of 2024.\nA reconciliation of diluted earnings per ADS/share to non-GAAP diluted earnings per ADS/share is included in the section entitled “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” of this Interim Report. Each ADS represents eight ordinary shares.\nCash and cash equivalents, short-term investments and other treasury investments\nAs of September 30, 2025, cash and cash equivalents, short-term investments and other treasury investments included in equity securities and other investments on the consolidated balance sheets, of which that are unrestricted for withdrawal and use, were RMB573,889 million (US\\$80,614 million), compared to RMB597,132 million as of March 31, 2025. Other treasury investments consist of fixed deposits, certificate of deposits and marketable debt securities with original maturities over one year for treasury purposes. The decrease in cash and cash equivalents, short-term investments and other treasury investments of RMB23,243 million during the six months ended September 30, 2025, was primarily due to (i) free cash flow outflow of RMB40,655 million (US\\$5,711 million), (ii) dividend payment of RMB33,621 million (US\\$4,723 million), (iii) cash used in repurchase of ordinary shares of RMB7,638 million (US\\$1,073 million), partly offset by (iv) net proceeds from issuance of convertible unsecured senior notes and payments for capped call transactions of RMB20,994 million (US\\$2,949 million), (v) net proceeds from bank borrowings of RMB17,804 million (US\\$2,501 million), (vi) net proceeds from issuance of exchangeable bonds of RMB10,986 million (US\\$1,543 million) and (vii) proceeds of RMB12,026 million (US\\$1,689 million) from the disposal of Intime and local consumer service business of Trendyol.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Net income attributable to ordinary shareholders", + "metadata": { + "length": 3285, + "summary": "Net income attributable to ordinary shareholders for the six months ended September 30, 2025 was RMB64,106 million, down from RMB68,143 million in 2024, due to decreased income from operations, partially offset by investment mark-to-market changes, Trendyol disposal gains, and lower impairment and exchange losses. Diluted earnings per ADS were RMB26.73, compared to RMB28.00 in 2024; non-GAAP diluted earnings per ADS fell 39% to RMB19.10. Diluted earnings per share were RMB3.34, versus RMB3.50 in 2024; non-GAAP diluted earnings per share also decreased 39% to RMB2.39. Cash, cash equivalents, short-term investments, and other treasury investments totaled RMB573,889 million as of September 30, 2025, down from RMB597,132 million as of March 31, 2025, primarily due to free cash flow outflow of RMB40,655 million, dividend payments of RMB33,621 million, and share repurchases of RMB7,638 million, partly offset by proceeds from convertible notes, bank borrowings, exchangeable bonds, and asset disposals.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Net", + "income", + "attributable", + "ordinary", + "shareholders", + "months", + "ended", + "September", + "30", + "2025", + "RMB64", + "106", + "million", + "US", + "005", + "compared", + "RMB68", + "143", + "period", + "2024", + "primarily", + "decrease", + "operations", + "partly", + "offset", + "mark", + "market", + "equity", + "investments", + "gain", + "disposal", + "local", + "consumer", + "service", + "business", + "Trendyol", + "impairment", + "net", + "exchange", + "loss", + "Diluted", + "earnings", + "ADS", + "share", + "GAAP", + "diluted", + "RMB26.73", + "3.75", + "RMB28.00", + "Excluding", + "cash", + "based", + "compensation", + "expense", + "gains", + "losses", + "goodwill", + "intangible", + "assets", + "items", + "RMB19.10", + "2.68", + "39%", + "RMB31.50", + "RMB3.34", + "0.47", + "HK", + "3.66", + "RMB3.50", + "RMB2.39", + "0.34", + "2.62", + "RMB3.94", + "reconciliation", + "included", + "section", + "entitled", + "Reconciliations", + "Non", + "Measures", + "Nearest", + "Comparable", + "Interim", + "Report", + "Each", + "represents", + "shares", + "Cash", + "equivalents", + "short", + "term", + "treasury", + "As", + "securities", + "consolidated", + "balance", + "sheets", + "unrestricted", + "withdrawal", + "RMB573", + "889", + "80", + "614", + "RMB597", + "132", + "March", + "31", + "Other", + "consist", + "fixed", + "deposits", + "certificate", + "marketable", + "debt", + "original", + "maturities", + "year", + "purposes", + "The", + "RMB23", + "243", + "due", + "free", + "flow", + "outflow", + "RMB40", + "655", + "711", + "ii", + "dividend", + "payment", + "RMB33", + "621", + "723", + "iii", + "repurchase", + "RMB7", + "638", + "073", + "iv", + "proceeds", + "issuance", + "convertible", + "unsecured", + "senior", + "notes", + "payments", + "capped", + "call", + "transactions", + "RMB20", + "994", + "949", + "bank", + "borrowings", + "RMB17", + "804", + "501", + "vi", + "exchangeable", + "bonds", + "RMB10", + "986", + "543", + "vii", + "RMB12", + "026", + "689", + "Intime" + ], + "keywords": [ + "Net income", + "Earnings per ADS", + "Cash investments" + ], + "connect_to": [] + } + }, + { + "chunk_id": "d701c347-b47d-5939-af4c-1753208fcd4c", + "type": "text", + "content": "During the six months ended September 30, 2025, net cash provided by operating activities was RMB30,771 million (US\\$4,322 million), a decrease of 53% compared to RMB65,074 million in the same period of 2024. Free cash flow, a non-GAAP measurement of liquidity, was an outflow of RMB40,655 million (US\\$5,711 million), compared to an inflow of RMB31,107 million in the same period of 2024. The decrease in free cash flow was mainly attributed to the increase in our cloud infrastructure expenditure and the investment in quick commerce. A reconciliation of net cash provided by operating activities to free cash flow is included in the section entitled “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” of this Interim Report.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Net cash provided by operating activities and free cash flow", + "metadata": { + "length": 760, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "During", + "months", + "ended", + "September", + "30", + "2025", + "net", + "cash", + "provided", + "operating", + "activities", + "RMB30", + "771", + "million", + "US", + "322", + "decrease", + "53%", + "compared", + "RMB65", + "074", + "period", + "2024", + "Free", + "flow", + "GAAP", + "measurement", + "liquidity", + "outflow", + "RMB40", + "655", + "711", + "inflow", + "RMB31", + "107", + "The", + "free", + "attributed", + "increase", + "cloud", + "infrastructure", + "expenditure", + "investment", + "quick", + "commerce", + "reconciliation", + "included", + "section", + "entitled", + "Reconciliations", + "Non", + "Measures", + "Nearest", + "Comparable", + "Interim", + "Report" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d8dfa2cb-781a-5c32-ae58-0d57d130776c", + "type": "text", + "content": "During the six months ended September 30, 2025, net cash used in investing activities of RMB51,324 million (US\\$7,209 million) primarily reflected capital expenditures of RMB70,177 million (US\\$9,858 million) and net increase in short-term investments and other treasury investments by RMB11,119 million (US\\$1,562 million), partly offset by net cash inflow of RMB29,710 million (US\\$4,173 million) for investment and acquisition activities.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Net cash used in investing activities", + "metadata": { + "length": 441, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "During", + "months", + "ended", + "September", + "30", + "2025", + "net", + "cash", + "investing", + "activities", + "RMB51", + "324", + "million", + "US", + "209", + "primarily", + "reflected", + "capital", + "expenditures", + "RMB70", + "177", + "858", + "increase", + "short", + "term", + "investments", + "treasury", + "RMB11", + "119", + "562", + "partly", + "offset", + "inflow", + "RMB29", + "710", + "173", + "investment", + "acquisition" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9281a011-69f6-55c1-a1db-3446e38b3a0d", + "type": "text", + "content": "During the six months ended September 30, 2025, net cash provided by financing activities of RMB8,171 million (US\\$1,148 million) primarily reflected cash provided by net proceeds from issuance of convertible unsecured senior notes and payments for capped call transactions of RMB20,994 million (US\\$2,949 million), net proceeds from bank borrowings of RMB17,804 million (US\\$2,501 million), and net proceeds from issuance of exchangeable bonds of RMB10,986 million (US\\$1,543 million), partly offset by dividend payment of RMB33,621 million (US\\$4,723 million), and cash used in repurchase of ordinary shares of RMB7,638 million (US\\$1,073 million).", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Net cash provided by financing activities", + "metadata": { + "length": 650, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "During", + "months", + "ended", + "September", + "30", + "2025", + "net", + "cash", + "provided", + "financing", + "activities", + "RMB8", + "171", + "million", + "US", + "148", + "primarily", + "reflected", + "proceeds", + "issuance", + "convertible", + "unsecured", + "senior", + "notes", + "payments", + "capped", + "call", + "transactions", + "RMB20", + "994", + "949", + "bank", + "borrowings", + "RMB17", + "804", + "501", + "exchangeable", + "bonds", + "RMB10", + "986", + "543", + "partly", + "offset", + "dividend", + "payment", + "RMB33", + "621", + "723", + "repurchase", + "ordinary", + "shares", + "RMB7", + "638", + "073" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0e128507-3b38-5c99-92ca-852f2f4b781a", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Net income67,56962,9948,849
Adjustments to reconcile net income to adjusted EBITA and adjusted EBITDA:
Interest and investment income, net(17,129)(37,468)(5,263)
Interest expense4,6154,995702
Other expense (income), net1,221(1,329)(187)
Income tax expenses17,44214,4152,024
Share of results of equity method investees(2,483)(3,254)(457)
Income from operations71,23540,3535,668
Non-cash share-based compensation expense7,7756,076854
Amortization and impairment of intangible assets, and others3,4411,488209
Provision for the shareholder class action lawsuits3,145--
Adjusted EBITA85,59647,9176,731
Depreciation and impairment of property and equipment, and operating lease cost relating to land use rights12,89215,0742,117
Adjusted EBITDA98,48862,9918,848
", + "path": "tables/table-6 Financial Results.html", + "metadata": { + "length": 1602, + "summary": "table-7\nThe table shows financial data for six months ending Sep 30, 2024 and 2025, with net income, adjusted EBITA, and EBITDA in RMB and US dollars.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-6 Financial Results.html", + "keywords": [ + "Adjusted EBITA", + "Net Income", + "EBITDA" + ], + "tokens": [] + } + }, + { + "chunk_id": "cf726d83-13c5-59df-973d-1ae0d77a2240", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Net income67,56962,9948,849
Adjustments to reconcile net income to non-GAAP net income:
Non-cash share-based compensation expense7,7756,076854
Amortization and impairment of intangible assets3,4411,633229
Provision for the shareholder class action lawsuits3,145--
Gain on deemed disposals/disposals/revaluation of investments(8,116)(29,320)(4,119)
Impairment of investments, and others5,0672,455345
Tax effects(1)(1,672)243
Non-GAAP net income77,20943,8626,161
", + "path": "tables/table-7 Non-GAAP Net Income.html", + "metadata": { + "length": 1088, + "summary": "table-8\nThe table shows net income and non-GAAP net income for six months ending September 30, 2024 and 2025, with adjustments including share-based compensation, amortization, and investment gains.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-7 Non-GAAP Net Income.html", + "keywords": [ + "Net income", + "Non-GAAP adjustments", + "Financial results" + ], + "tokens": [] + } + }, + { + "chunk_id": "fd2dc317-0fc3-572a-a24c-f37cf5f0db4d", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMBUS$
(in millions, except per share data)
Net income attributable to ordinary shareholders – basic68,14364,1069,005
Dilution effect on earnings arising from non-cash share-based awards operated by equity method investees and subsidiaries(131)(258)(36)
Adjustments for interest expense attributable to convertible unsecured senior notes9514320
Net income attributable to ordinary shareholders – diluted68,10763,9918,989
Non-GAAP adjustments to net income attributable to ordinary shareholders(1)8,521(18,250)(2,564)
Non-GAAP net income attributable to ordinary shareholders for computing non-GAAP diluted earnings per share/ADS76,62845,7416,425
Weighted average number of shares on a diluted basis for computing non-GAAP diluted earnings per share/ADS (million shares)(2)19,45919,154
Diluted earnings per share(2)(3)3.503.340.47
Non-GAAP diluted earnings per share(2)(4)3.942.390.34
Diluted earnings per ADS(2)(3)28.0026.733.75
Non-GAAP diluted earnings per ADS(2)(4)31.5019.102.68
", + "path": "tables/table-8 Earnings per Share.html", + "metadata": { + "length": 1677, + "summary": "table-9\nThe table shows basic and diluted net income attributable to ordinary shareholders for six months ending Sep 30, 2024 and 2025, with non-GAAP adjustments and diluted earnings per share/ADS in RMB and USD.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-8 Earnings per Share.html", + "keywords": [ + "Net income", + "Diluted EPS", + "Non-GAAP" + ], + "tokens": [] + } + }, + { + "chunk_id": "b7a25b9e-4b38-5c23-a976-7b75b7d25f45", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Net cash provided by operating activities65,07430,7714,322
Less: Purchase of property and equipment (excluding land use rights and construction in progress relating to office campuses)(28,916)(70,057)(9,841)
Less: Changes in the buyer protection fund deposits(5,051)(1,369)(192)
Free cash flow31,107(40,655)(5,711)
", + "path": "tables/table-9 Free Cash Flow.html", + "metadata": { + "length": 722, + "summary": "table-10\nThe table shows net cash from operations, property purchases, and buyer protection deposits for six months ending Sep 2024 and 2025, resulting in free cash flow of 31,107 RMB (2024) and -40,655 RMB (2025).", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-9 Free Cash Flow.html", + "keywords": [ + "Free cash flow", + "Operating activities", + "Capital expenditure" + ], + "tokens": [] + } + }, + { + "chunk_id": "fb07a64f-f537-5605-8b7b-49fda91cd7c3", + "type": "text", + "content": "As of September 30, 2025, we had a total of 126,661 employees, compared to 124,320 as of March 31, 2025.\nReconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures\nThe table below sets forth a reconciliation of our net income to adjusted EBITA and adjusted EBITDA for the periods indicated:\n\n[tables/table-6 Financial Results.html]\n\nReconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures (Continued)\nThe table below sets forth a reconciliation of our net income to non-GAAP net income for the periods indicated:\n\n[tables/table-7 Non-GAAP Net Income.html]\n\n(1) Tax effects primarily comprise tax effects relating to non-cash share-based compensation expense, amortization and impairment of intangible assets and certain gains and losses from investments, and others.\nReconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures (Continued)\nThe table below sets forth a reconciliation of our diluted earnings per share/ADS to non-GAAP diluted earnings per share/ADS for the periods indicated:\n\n[tables/table-8 Earnings per Share.html]\n\n(1) Non-GAAP adjustments excluding the attributions to the noncontrolling interests. See the table above for items regarding the reconciliation of net income to non-GAAP net income (before excluding the attributions to the noncontrolling interests).\n(2) Each ADS represents eight ordinary shares.\n(3) Diluted earnings per share is derived from dividing net income attributable to ordinary shareholders by the weighted average number of outstanding ordinary shares, on a diluted basis. Diluted earnings per ADS is derived from the diluted earnings per share after adjusting for the ordinary share-to-ADS ratio.\n(4) Non-GAAP diluted earnings per share is derived from dividing non-GAAP net income attributable to ordinary shareholders by the weighted average number of outstanding ordinary shares for computing non-GAAP diluted earnings per share, on a diluted basis. Non-GAAP diluted earnings per ADS is derived from the non-GAAP diluted earnings per share after adjusting for the ordinary share-to-ADS ratio.\nReconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures (Continued)\nThe table below sets forth a reconciliation of net cash provided by operating activities to free cash flow for the periods indicated:\n\n[tables/table-9 Free Cash Flow.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Employees", + "metadata": { + "length": 2404, + "summary": "The passage provides reconciliations of non-GAAP financial measures to GAAP measures for a company, including adjusted EBITA, adjusted EBITDA, non-GAAP net income, non-GAAP diluted earnings per share/ADS, and free cash flow for the six months ended September 30, 2025. It also notes employee counts: 126,661 as of September 30, 2025, compared to 124,320 as of March 31, 2025. The reconciliations are presented in tables, with footnotes explaining tax effects, ADS-to-ordinary share ratio (1:8), and adjustments for noncontrolling interests.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "As", + "September", + "30", + "2025", + "total", + "126", + "661", + "employees", + "compared", + "124", + "320", + "March", + "31", + "Reconciliations", + "Non", + "GAAP", + "Measures", + "Nearest", + "Comparable", + "The", + "table", + "sets", + "reconciliation", + "net", + "income", + "adjusted", + "EBITA", + "EBITDA", + "periods", + "Continued", + "Tax", + "effects", + "primarily", + "comprise", + "tax", + "relating", + "cash", + "share", + "based", + "compensation", + "expense", + "amortization", + "impairment", + "intangible", + "assets", + "gains", + "losses", + "investments", + "diluted", + "earnings", + "ADS", + "adjustments", + "excluding", + "attributions", + "noncontrolling", + "interests", + "See", + "items", + "Each", + "represents", + "ordinary", + "shares", + "Diluted", + "derived", + "dividing", + "attributable", + "shareholders", + "weighted", + "average", + "number", + "outstanding", + "basis", + "adjusting", + "ratio", + "computing", + "provided", + "operating", + "activities", + "free", + "flow" + ], + "keywords": [ + "Non-GAAP", + "reconciliation", + "financial measures" + ], + "connect_to": [ + { + "target": "0e128507-3b38-5c99-92ca-852f2f4b781a", + "relation": "embeds", + "ref": "[tables/table-6 Financial Results.html]", + "position": { + "start": 315, + "end": 354 + } + }, + { + "target": "cf726d83-13c5-59df-973d-1ae0d77a2240", + "relation": "embeds", + "ref": "[tables/table-7 Non-GAAP Net Income.html]", + "position": { + "start": 563, + "end": 604 + } + }, + { + "target": "fd2dc317-0fc3-572a-a24c-f37cf5f0db4d", + "relation": "embeds", + "ref": "[tables/table-8 Earnings per Share.html]", + "position": { + "start": 1063, + "end": 1103 + } + }, + { + "target": "b7a25b9e-4b38-5c23-a976-7b75b7d25f45", + "relation": "embeds", + "ref": "[tables/table-9 Free Cash Flow.html]", + "position": { + "start": 2343, + "end": 2379 + } + } + ] + } + }, + { + "chunk_id": "82c16a00-2789-5824-bd1e-8a55ab010121", + "type": "text", + "content": "To supplement our unaudited condensed consolidated financial statements, which are prepared and presented in accordance with GAAP, we use the following non-GAAP financial measures: for our consolidated results, adjusted EBITDA (including adjusted EBITDA margin), adjusted EBITA (including adjusted EBITA margin), non-GAAP net income, non-GAAP diluted earnings per share/ADS and free cash flow. For more information on these non-GAAP financial measures, please refer to the table captioned “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” in this Interim Report.\nWe believe that adjusted EBITDA, adjusted EBITA, non-GAAP net income and non-GAAP diluted earnings per share/ADS help identify underlying trends in our business that could otherwise be distorted by the effect of certain income or expenses that we include in income from operations, net income and diluted earnings per share/ADS. We believe that these non-GAAP measures provide useful information about our core operating results, enhance the overall understanding of our past performance and future prospects and allow for greater visibility with respect to key metrics used by our management in its financial and operational decision-making. We present three different income measures, namely adjusted EBITDA, adjusted EBITA and non-GAAP net income in order to provide more information and greater transparency to investors about our operating results.\nWe consider free cash flow to be a liquidity measure that provides useful information to management and investors about the amount of cash generated by our business that can be used for strategic corporate transactions, including investing in our new business initiatives, making strategic investments and acquisitions and strengthening our balance sheet.\nAdjusted EBITDA, adjusted EBITA, non-GAAP net income, non-GAAP diluted earnings per share/ADS and free cash flow should not be considered in isolation or construed as an alternative to income from operations, net income, diluted earnings per share/ADS, cash flows or any other measure of performance or as an indicator of our operating performance. These non-GAAP financial measures presented here do not have standardized meanings prescribed by U.S. GAAP and may not be comparable to similarly titled measures presented by other companies. Other companies may calculate similarly titled measures differently, limiting their usefulness as comparative measures to our data.\nAdjusted EBITDA represents net income before interest and investment income, net, interest expense, other income (expense), net, income tax expenses, share of results of equity method investees, certain non-cash expenses, consisting of share-based compensation expense, amortization and impairment of intangible assets, impairment of goodwill, depreciation and impairment of property and equipment, and operating lease cost relating to land use rights, and others (including provision in relation to matters outside the ordinary course of business), which we do not believe are reflective of our core operating performance during the periods presented.\nAdjusted EBITA represents net income before interest and investment income, net, interest expense, other income (expense), net, income tax expenses, share of results of equity method investees, certain non-cash expenses, consisting of share-based compensation expense, amortization and impairment of intangible assets, impairment of goodwill, and others (including provision in relation to matters outside the ordinary course of business), which we do not believe are reflective of our core operating performance during the periods presented.\nNon-GAAP net income represents net income before non-cash share-based compensation expense, amortization and impairment of intangible assets, gain or loss on deemed disposals/disposals/revaluation of investments, impairment of goodwill and investments, and others (including provision in relation to matters outside the ordinary course of business), and adjustments for the tax effects.\nNon-GAAP diluted earnings per share represents non-GAAP net income attributable to ordinary shareholders divided by the weighted average number of outstanding ordinary shares for computing non-GAAP diluted earnings per share on a diluted basis. Non-GAAP diluted earnings per ADS represents non-GAAP diluted earnings per share after adjusting for the ordinary share-to-ADS ratio.\nFree cash flow represents net cash provided by operating activities as presented in our unaudited condensed consolidated cash flow statement less purchases of property and equipment (excluding acquisition of land use rights and construction in progress relating to office campuses) and intangible assets (excluding those acquired through acquisitions), as well as adjustments to exclude from net cash provided by operating activities the buyer protection fund deposits from merchants on our marketplaces. We deduct certain items of cash flows from investing activities in order to provide greater transparency into cash flow from our revenue-generating business operations. We exclude “acquisition of land use rights and construction in progress relating to office campuses” because the office campuses are used by us for corporate and administrative purposes and are not directly related to our revenue-generating business operations. We also exclude buyer protection fund deposits from merchants on our marketplaces because these deposits are restricted for the purpose of compensating buyers for claims against merchants.\nThe table captioned “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” in this Interim Report has more details on the non-GAAP financial measures that are most directly comparable to GAAP financial measures and the related reconciliations between these financial measures.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Non-GAAP Financial Measures", + "metadata": { + "length": 5872, + "summary": "The passage explains the company's use of non-GAAP financial measures including adjusted EBITDA, adjusted EBITA, non-GAAP net income, non-GAAP diluted earnings per share/ADS, and free cash flow. These measures are used to supplement GAAP financial statements and help identify underlying business trends by excluding certain non-cash expenses and items not reflective of core operating performance. The definitions of each measure are provided, along with cautionary notes that they should not be considered in isolation or as alternatives to GAAP measures, and may not be comparable to similarly titled measures from other companies. Free cash flow is described as a liquidity measure for strategic corporate transactions, with specific adjustments explained.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "To", + "supplement", + "unaudited", + "condensed", + "consolidated", + "financial", + "statements", + "prepared", + "presented", + "accordance", + "GAAP", + "measures", + "results", + "adjusted", + "EBITDA", + "including", + "margin", + "EBITA", + "net", + "income", + "diluted", + "earnings", + "share", + "ADS", + "free", + "cash", + "flow", + "For", + "information", + "refer", + "table", + "captioned", + "Reconciliations", + "Non", + "Measures", + "Nearest", + "Comparable", + "Interim", + "Report", + "We", + "identify", + "underlying", + "trends", + "business", + "distorted", + "effect", + "expenses", + "include", + "operations", + "provide", + "core", + "operating", + "enhance", + "understanding", + "past", + "performance", + "future", + "prospects", + "greater", + "visibility", + "respect", + "key", + "metrics", + "management", + "operational", + "decision", + "making", + "present", + "order", + "transparency", + "investors", + "liquidity", + "measure", + "amount", + "generated", + "strategic", + "corporate", + "transactions", + "investing", + "initiatives", + "investments", + "acquisitions", + "strengthening", + "balance", + "sheet", + "Adjusted", + "considered", + "isolation", + "construed", + "alternative", + "flows", + "indicator", + "These", + "standardized", + "meanings", + "prescribed", + "comparable", + "similarly", + "titled", + "companies", + "Other", + "calculate", + "differently", + "limiting", + "usefulness", + "comparative", + "data", + "represents", + "interest", + "investment", + "expense", + "tax", + "equity", + "method", + "investees", + "consisting", + "based", + "compensation", + "amortization", + "impairment", + "intangible", + "assets", + "goodwill", + "depreciation", + "property", + "equipment", + "lease", + "cost", + "relating", + "land", + "rights", + "provision", + "relation", + "matters", + "ordinary", + "reflective", + "periods", + "gain", + "loss", + "deemed", + "disposals", + "revaluation", + "adjustments", + "effects", + "attributable", + "shareholders", + "divided", + "weighted", + "average", + "number", + "outstanding", + "shares", + "computing", + "basis", + "adjusting", + "ratio", + "Free", + "provided", + "activities", + "statement", + "purchases", + "excluding", + "acquisition", + "construction", + "progress", + "office", + "campuses", + "acquired", + "exclude", + "buyer", + "protection", + "fund", + "deposits", + "merchants", + "marketplaces", + "deduct", + "items", + "revenue", + "generating", + "administrative", + "purposes", + "directly", + "related", + "restricted", + "purpose", + "compensating", + "buyers", + "claims", + "The", + "details", + "reconciliations" + ], + "keywords": [ + "adjusted EBITDA", + "non-GAAP net income", + "free cash flow" + ], + "connect_to": [] + } + }, + { + "chunk_id": "cc3cbde1-83cc-5ae8-8fde-0f08c6849593", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "42159302-b4f8-5cfe-840d-777e319a8516", + "type": "text", + "content": "We fund our operations and strategic investments from cash generated from our operations and through debt and equity financing. We generated RMB65,074 million and RMB30,771 million (US\\$4,322 million) of cash from operating activities for the six months ended September 30, 2024 and 2025, respectively. As of September 30, 2025, we had cash and cash equivalents, short-term investments and other treasury investments of RMB573,889 million (US\\$80,614 million) that are unrestricted for withdrawal and use. Short-term investments include investments in fixed deposits with original maturities between three months and one year and certain investments in wealth management products, certificates of deposits, marketable debt securities and other investments whereby we have the intention to redeem within one year. Other treasury investments mainly include investments in fixed deposits, certificates of deposits and marketable debt securities with original maturities over one year for treasury purposes. The remaining maturities of these treasury investments held by us generally range from one to five years.\nWe believe that our current levels of cash and cash flows from operations will be sufficient to meet our anticipated cash needs for at least the next twelve months. However, we may need additional cash resources in the future if we find and wish to pursue opportunities for investment, acquisition, strategic cooperation or other similar actions, which may include investing in technology, infrastructure, including data management and analytics solutions, or related talent. If we determine that our cash requirements exceed our amounts of cash on hand or if we decide to further optimize our capital structure, we may seek to issue additional debt or equity securities or obtain credit facilities or other sources of funding.\nIn July 2025, we issued zero coupon exchangeable bonds due 2032 by reference to the ordinary shares of our subsidiary, Alibaba Health Information Technology Limited, that are listed on the Hong Kong Stock Exchange, for an aggregate principal amount of approximately HK\\$12 billion.\nIn September 2025, we issued zero coupon convertible unsecured senior notes due 2032 for an aggregate principal amount of approximately US\\$3.2 billion.\nWe have made a series of amendments to our syndicated loan facility. Subsequent to the partial repayment of US\\$830 million in January 2025, the size of the US\\$4.0 billion syndicated loan was reduced to US\\$3.17 billion. In September 2025, we amended the loan facility and reduced the pricing terms to Secured Overnight Financing Rate (“SOFR”) plus 66 basis points. Effective on November 28, 2025, the facility was restructured as a revolving credit facility with drawdowns permitted in both U.S. dollars and Hong Kong dollars, and the expiration date of the facility was extended to September 30, 2028, with an option to further extend to September 30, 2030. The interest rate of the credit facility is 66 basis points over SOFR or Hong Kong Interbank Offered Rate (“HIBOR”), and the margin will be 81 basis points for the optional extension period. Subsequently, we submitted a prepayment notice to repay the outstanding balance of US\\$3.17 billion under the revolving credit facility.\nIn September 2025, we amended the terms of a US\\$6.5 billion revolving credit facility agreement. The size of the credit facility was amended to US\\$3.33 billion and the utilization currency was also amended from U.S. dollar only to both U.S. dollar and Hong Kong dollar. The interest rate of the credit facility was adjusted to SOFR or HIBOR plus 66 basis points. The expiration date of the credit facility was extended from June 24, 2026 to September 30, 2028, with an option to further extend to September 30, 2030 and the margin will be 81 basis points for the optional extension period. We have not yet drawn down this facility.\nThe Group monitors its financial health and liquidity position by reviewing its total debts to adjusted EBITDA ratio and total debts to total capital ratio. Our total debts is defined as the sum of bank borrowings, unsecured senior notes, convertible unsecured senior notes and exchangeable bonds. Our total debts to adjusted EBITDA ratio is calculated by dividing total debts by adjusted EBITDA for the last twelve months. Our total debts to total capital ratio is calculated by dividing total debts by total capital. Total capital is calculated as total equity plus total debts. The Group's total debts to adjusted EBITDA ratio was 1.14 and 1.69 as at March 31, 2025 and September 30, 2025, respectively. The Group's total debts to total capital ratio was $17.62\\%$ and $20.35\\%$ as at March 31, 2025 and September 30, 2025, respectively.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Liquidity and Capital Resources", + "metadata": { + "length": 4736, + "summary": "The passage discusses the company's funding sources, including cash from operations and debt/equity financing. It reports cash from operations of RMB65,074 million and RMB30,771 million for the six months ended September 30, 2024 and 2025, respectively. As of September 30, 2025, unrestricted cash, short-term investments, and other treasury investments totaled RMB573,889 million. The company issued zero coupon exchangeable bonds due 2032 for HK$12 billion in July 2025 and zero coupon convertible unsecured senior notes due 2032 for US$3.2 billion in September 2025. It amended a syndicated loan facility, reducing it to US$3.17 billion and restructuring it as a revolving credit facility expiring September 30, 2028, with an option to extend to 2030. Another US$6.5 billion revolving credit facility was amended to US$3.33 billion with similar terms. The company monitors financial health via total debts to adjusted EBITDA ratio (1.14 as of March 2025, 1.69 as of September 2025) and total debts to total capital ratio (17.62% and 20.35%, respectively).", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "We", + "fund", + "operations", + "strategic", + "investments", + "cash", + "generated", + "debt", + "equity", + "financing", + "RMB65", + "074", + "million", + "RMB30", + "771", + "US", + "322", + "operating", + "activities", + "months", + "ended", + "September", + "30", + "2024", + "2025", + "As", + "equivalents", + "short", + "term", + "treasury", + "RMB573", + "889", + "80", + "614", + "unrestricted", + "withdrawal", + "Short", + "include", + "fixed", + "deposits", + "original", + "maturities", + "year", + "wealth", + "management", + "products", + "certificates", + "marketable", + "securities", + "intention", + "redeem", + "Other", + "purposes", + "The", + "remaining", + "held", + "generally", + "range", + "years", + "current", + "levels", + "flows", + "sufficient", + "meet", + "anticipated", + "twelve", + "However", + "additional", + "resources", + "future", + "find", + "pursue", + "opportunities", + "investment", + "acquisition", + "cooperation", + "similar", + "actions", + "investing", + "technology", + "infrastructure", + "including", + "data", + "analytics", + "solutions", + "related", + "talent", + "If", + "determine", + "requirements", + "exceed", + "amounts", + "hand", + "decide", + "optimize", + "capital", + "structure", + "seek", + "issue", + "obtain", + "credit", + "facilities", + "sources", + "funding", + "In", + "July", + "issued", + "coupon", + "exchangeable", + "bonds", + "due", + "2032", + "reference", + "ordinary", + "shares", + "subsidiary", + "Alibaba", + "Health", + "Information", + "Technology", + "Limited", + "listed", + "Hong", + "Kong", + "Stock", + "Exchange", + "aggregate", + "principal", + "amount", + "approximately", + "HK", + "12", + "billion", + "convertible", + "unsecured", + "senior", + "notes", + "3.2", + "made", + "series", + "amendments", + "syndicated", + "loan", + "facility", + "Subsequent", + "partial", + "repayment", + "830", + "January", + "size", + "4.0", + "reduced", + "3.17", + "amended", + "pricing", + "terms", + "Secured", + "Overnight", + "Financing", + "Rate", + "SOFR", + "66", + "basis", + "points", + "Effective", + "November", + "28", + "restructured", + "revolving", + "drawdowns", + "permitted", + "dollars", + "expiration", + "date", + "extended", + "2028", + "option", + "extend", + "2030", + "interest", + "rate", + "Interbank", + "Offered", + "HIBOR", + "margin", + "81", + "optional", + "extension", + "period", + "Subsequently", + "submitted", + "prepayment", + "notice", + "repay", + "outstanding", + "balance", + "6.5", + "agreement", + "3.33", + "utilization", + "currency", + "dollar", + "adjusted", + "June", + "24", + "2026", + "drawn", + "Group", + "monitors", + "financial", + "health", + "liquidity", + "position", + "reviewing", + "total", + "debts", + "EBITDA", + "ratio", + "Our", + "defined", + "sum", + "bank", + "borrowings", + "calculated", + "dividing", + "Total", + "1.14", + "1.69", + "March", + "31", + "17.62", + "20.35" + ], + "keywords": [ + "cash flow", + "debt financing", + "liquidity" + ], + "connect_to": [] + } + }, + { + "chunk_id": "fba5a379-4b76-5183-a470-e276b950ff12", + "type": "text", + "content": "Our significant investment consists of Ant Group. Ant Group provides comprehensive digital payment services and facilitates digital financial and value-added services for consumers and merchants, in China and across the world. As of September 30, 2025, our equity interest in Ant Group on a fully diluted basis was 33%. During the six months ended September 30, 2025, dividend received from Ant Group amounted to RMB3,293 million (US\\$463 million).\nWe did not hold any other significant investments as of September 30, 2025.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Significant Investments", + "metadata": { + "length": 524, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "significant", + "investment", + "consists", + "Ant", + "Group", + "comprehensive", + "digital", + "payment", + "services", + "facilitates", + "financial", + "added", + "consumers", + "merchants", + "China", + "world", + "As", + "September", + "30", + "2025", + "equity", + "interest", + "fully", + "diluted", + "basis", + "33%", + "During", + "months", + "ended", + "dividend", + "received", + "amounted", + "RMB3", + "293", + "million", + "US", + "463", + "We", + "hold", + "investments" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "20cc45a1-c317-54f5-9f4e-6726fc2fffd3", + "type": "text", + "content": "Our material investments, acquisitions and dispositions in the six months ended September 30, 2025 and the period through the date of this Interim Report are set forth below.\nIn May 2025, we agreed to sell 85% of the equity interest in Trendyol GO, a wholly-owned subsidiary of Trendyol that operates local service business in Türkiye (the “Disposal”). The cash consideration for the Disposal is approximately US\\$0.7 billion (RMB5 billion). The Disposal was completed during the six months ended September 30, 2025.\nSave as disclosed above, as at the date of this Interim Report, the Group did not have detailed future plans for material investments or capital assets.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Material Investments, Acquisitions and Dispositions", + "metadata": { + "length": 669, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "material", + "investments", + "acquisitions", + "dispositions", + "months", + "ended", + "September", + "30", + "2025", + "period", + "date", + "Interim", + "Report", + "set", + "In", + "May", + "agreed", + "sell", + "85%", + "equity", + "interest", + "Trendyol", + "GO", + "wholly", + "owned", + "subsidiary", + "operates", + "local", + "service", + "business", + "rkiye", + "Disposal", + "The", + "cash", + "consideration", + "approximately", + "US", + "0.7", + "billion", + "RMB5", + "completed", + "Save", + "disclosed", + "Group", + "detailed", + "future", + "plans", + "capital", + "assets" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a7c4e83d-e10f-5f5d-83f4-24914bf28e62", + "type": "text", + "content": "Certain of the Group's bank borrowings are collateralized by a pledge of certain buildings and property improvements, construction in progress and land use rights in the PRC, receivables and other treasury investments with carrying values of RMB30,213 million and RMB25,417 million, as of March 31, 2025 and September 30, 2025, respectively. In addition, certain of the Group's payables are collateralized by a pledge of certain short-term investments and other treasury investments with carrying values of RMB3,697 million and RMB4,155 million as of March 31, 2025 and September 30, 2025, respectively.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Pledge of Assets", + "metadata": { + "length": 603, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Certain", + "Group", + "bank", + "borrowings", + "collateralized", + "pledge", + "buildings", + "property", + "improvements", + "construction", + "progress", + "land", + "rights", + "PRC", + "receivables", + "treasury", + "investments", + "carrying", + "values", + "RMB30", + "213", + "million", + "RMB25", + "417", + "March", + "31", + "2025", + "September", + "30", + "In", + "addition", + "payables", + "short", + "term", + "RMB3", + "697", + "RMB4", + "155" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "47a67577-5880-5dd9-b9ca-618bcf257c68", + "type": "text", + "content": "Foreign currency risk arises from future commercial transactions, recognized assets and liabilities and net investments in foreign operations. Although we operate businesses in different countries and regions, most of our revenue-generating transactions, and a majority of our expense-related transactions, are denominated in Renminbi, which is the functional currency of our major operating subsidiaries and the reporting currency of our financial statements. When considered appropriate, we enter into hedging activities with regard to exchange rate risk.\nThe value of the Renminbi against the U.S. dollar and other currencies may fluctuate and is affected by, among other things, changes in political and economic conditions and the foreign exchange policy adopted by the governments. It is difficult to predict how market forces or PRC or U.S. government policy may impact the exchange rate between the Renminbi and the U.S. dollar in the future. There remains significant international pressure on the PRC government to adopt a more flexible currency policy, which could result in greater fluctuations of the Renminbi against the U.S. dollar.\nTo the extent that we need to convert U.S. dollars into Renminbi for our operations, appreciation of the Renminbi against the U.S. dollar would reduce the Renminbi amount we receive from the conversion. Conversely, if we decide to convert Renminbi into U.S. dollars for the purpose of making payments for dividends on our ordinary shares or ADSs, servicing our outstanding debts, or for other business purposes, appreciation of the U.S. dollar against the Renminbi would reduce the U.S. dollar amounts available to us.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Foreign Exchange Risk", + "metadata": { + "length": 1666, + "summary": "The passage discusses foreign currency risk arising from transactions and investments. Most revenue and expenses are in Renminbi, the functional currency. The Renminbi's value against the U.S. dollar fluctuates due to political, economic, and policy factors. Appreciation of the Renminbi reduces U.S. dollar conversion amounts for operations, while U.S. dollar appreciation reduces Renminbi conversion for dividends or debt payments.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Foreign", + "currency", + "risk", + "arises", + "future", + "commercial", + "transactions", + "recognized", + "assets", + "liabilities", + "net", + "investments", + "foreign", + "operations", + "Although", + "operate", + "businesses", + "countries", + "regions", + "revenue", + "generating", + "majority", + "expense", + "related", + "denominated", + "Renminbi", + "functional", + "major", + "operating", + "subsidiaries", + "reporting", + "financial", + "statements", + "When", + "considered", + "enter", + "hedging", + "activities", + "regard", + "exchange", + "rate", + "The", + "dollar", + "currencies", + "fluctuate", + "affected", + "things", + "political", + "economic", + "conditions", + "policy", + "adopted", + "governments", + "It", + "difficult", + "predict", + "market", + "forces", + "PRC", + "government", + "impact", + "There", + "remains", + "significant", + "international", + "pressure", + "adopt", + "flexible", + "result", + "greater", + "fluctuations", + "To", + "extent", + "convert", + "dollars", + "appreciation", + "reduce", + "amount", + "receive", + "conversion", + "Conversely", + "decide", + "purpose", + "making", + "payments", + "dividends", + "ordinary", + "shares", + "ADSs", + "servicing", + "outstanding", + "debts", + "business", + "purposes", + "amounts" + ], + "keywords": [ + "Renminbi", + "U.S. dollar", + "exchange rate" + ], + "connect_to": [] + } + }, + { + "chunk_id": "2aab5adc-6216-562a-9d01-2fdc27a01f50", + "type": "text", + "content": "Save as disclosed in this Interim Report, as at September 30, 2025, the Group had no material contingent liabilities.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Contingent Liabilities", + "metadata": { + "length": 117, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Save", + "disclosed", + "Interim", + "Report", + "September", + "30", + "2025", + "Group", + "material", + "contingent", + "liabilities" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3ad0c078-f679-5493-a796-114c8bd29bbe", + "type": "text", + "content": "Our capital expenditures have been incurred primarily in relation to (i) the acquisition of computer equipment and construction of data centers relating to our Cloud business and the operation of our mobile platforms and websites; (ii) the acquisition of infrastructure for logistics services and direct sales businesses; and (iii) the acquisition of land use rights and construction of corporate campuses and office facilities. In the six months ended September 30, 2024 and 2025, our capital expenditures totaled RMB29,585 million and RMB70,177 million (US\\$9,858 million), respectively.\nThe Group’s capital commitments primarily relate to capital expenditures contracted for purchase of property and equipment, including the construction of corporate campuses. Total capital commitments contracted but not provided for amounted to RMB45,321 million and RMB29,483 million (US\\$4,141 million) as of March 31, 2025 and September 30, 2025, respectively.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Capital Expenditure and Capital Commitment", + "metadata": { + "length": 952, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "capital", + "expenditures", + "incurred", + "primarily", + "relation", + "acquisition", + "computer", + "equipment", + "construction", + "data", + "centers", + "relating", + "Cloud", + "business", + "operation", + "mobile", + "platforms", + "websites", + "ii", + "infrastructure", + "logistics", + "services", + "direct", + "sales", + "businesses", + "iii", + "land", + "rights", + "corporate", + "campuses", + "office", + "facilities", + "In", + "months", + "ended", + "September", + "30", + "2024", + "2025", + "totaled", + "RMB29", + "585", + "million", + "RMB70", + "177", + "US", + "858", + "The", + "Group", + "commitments", + "relate", + "contracted", + "purchase", + "property", + "including", + "Total", + "provided", + "amounted", + "RMB45", + "321", + "483", + "141", + "March", + "31" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "24224826-6d2f-5702-b4b7-1ef7bfc87014", + "type": "text", + "content": "Alibaba Group periodically reviews its remuneration policy and compensation packages to ensure they are competitive within the industry. Furthermore, discretionary bonuses and other long-term incentives may be awarded to selected employees based on various factors including but not limited to individual performance and the overall performance of our business. We have established learning and training programs to develop our employees both personally and professionally, helping them to better realize their potential and create value, thereby supporting their long-term career success.\nThe Company’s subsidiaries in the PRC participate in a government-mandated multi-employer defined contribution plan. These plans provide employees with housing, pension, medical, maternity, work-related injury and unemployment benefits, as well as other welfare benefits. The relevant labor regulations require the Company’s subsidiaries in the PRC to make monthly contributions to the local labor and social security authorities based on the applicable benchmarks and rates stipulated by the local government. Additionally, we provide commercial health and accidental insurance for our employees. The Company’s subsidiaries also formulate their own unique benefit plans and assistance programs tailored to their specific business needs.\nAlibaba Group also makes payments to other defined contribution plans and defined benefit plans for the benefit of employees employed by subsidiaries outside of the PRC.\nShare-based awards such as restricted share units, incentive and non-statutory stock options, restricted shares and share appreciation rights may be granted to any directors, employees, service providers and consultants of Alibaba Group or affiliated companies under equity incentive plans adopted since the inception of the Company. For details of Alibaba Group's equity incentive plans, please refer to the section titled “Equity Incentive Plans”.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Remuneration Policy", + "metadata": { + "length": 1947, + "summary": "Alibaba Group reviews its remuneration policy to maintain industry competitiveness, offering discretionary bonuses and long-term incentives based on performance. It provides learning programs for employee development. PRC subsidiaries participate in government-mandated defined contribution plans covering housing, pension, medical, and other benefits, with monthly contributions to local authorities. Commercial health and accidental insurance are also provided, along with tailored benefit plans. Outside the PRC, subsidiaries make payments to defined contribution and benefit plans. Equity incentives, including restricted share units, stock options, and share appreciation rights, may be granted to directors, employees, service providers, and consultants under adopted plans.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Alibaba", + "Group", + "periodically", + "reviews", + "remuneration", + "policy", + "compensation", + "packages", + "ensure", + "competitive", + "industry", + "Furthermore", + "discretionary", + "bonuses", + "long", + "term", + "incentives", + "awarded", + "selected", + "employees", + "based", + "factors", + "including", + "limited", + "individual", + "performance", + "business", + "We", + "established", + "learning", + "training", + "programs", + "develop", + "personally", + "professionally", + "helping", + "realize", + "potential", + "create", + "supporting", + "career", + "success", + "The", + "Company", + "subsidiaries", + "PRC", + "participate", + "government", + "mandated", + "multi", + "employer", + "defined", + "contribution", + "plan", + "These", + "plans", + "provide", + "housing", + "pension", + "medical", + "maternity", + "work", + "related", + "injury", + "unemployment", + "benefits", + "welfare", + "relevant", + "labor", + "regulations", + "require", + "make", + "monthly", + "contributions", + "local", + "social", + "security", + "authorities", + "applicable", + "benchmarks", + "rates", + "stipulated", + "Additionally", + "commercial", + "health", + "accidental", + "insurance", + "formulate", + "unique", + "benefit", + "assistance", + "tailored", + "specific", + "makes", + "payments", + "employed", + "Share", + "awards", + "restricted", + "share", + "units", + "incentive", + "statutory", + "stock", + "options", + "shares", + "appreciation", + "rights", + "granted", + "directors", + "service", + "providers", + "consultants", + "affiliated", + "companies", + "equity", + "adopted", + "inception", + "For", + "details", + "refer", + "section", + "titled", + "Equity", + "Incentive", + "Plans" + ], + "keywords": [ + "remuneration", + "benefits", + "equity incentives" + ], + "connect_to": [] + } + }, + { + "chunk_id": "1b63af73-b5b8-5688-80e4-4b83c9babb76", + "type": "text", + "content": "Save as disclosed in this Interim Report, as at the date of this Interim Report, there were no significant events that might affect Alibaba Group since September 30, 2025.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Subsequent Events", + "metadata": { + "length": 171, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Save", + "disclosed", + "Interim", + "Report", + "date", + "significant", + "events", + "affect", + "Alibaba", + "Group", + "September", + "30", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e2c7d6e5-68fd-5d7c-8402-f83871e5a379", + "type": "table", + "content": "
Name(1)AgePosition/Title
Joseph C. TSAI $^{\\dagger(a)}$ 61Chairman
Eddie Yongming WU $^{\\dagger(b)}$ 50Director and Chief Executive Officer
J. Michael EVANS $^{\\dagger(a)}$ 68Director and President
Maggie Wei WU $^{\\dagger(c)}$ 57Director $^{(d)}$
Jerry YANG $^{(b)}$ 57Independent director
Wan Ling MARTELLO $^{(b)}$ 67Independent director
Weijian SHAN $^{(c)}$ 72Independent director
Irene Yun-Lien LEE $^{(a)}$ 72Independent director
Albert Kong Ping NG $^{(b)}$ 68Independent director
Kabir MISRA $^{(c)}$ 56Independent director
", + "path": "tables/table-10 Board of Directors.html", + "metadata": { + "length": 934, + "summary": "table-11\nThe table lists 10 board members of a company, including Chairman Joseph C. Tsai (61), CEO Eddie Yongming Wu (50), and President J. Michael Evans (68), along with independent directors.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-10 Board of Directors.html", + "keywords": [ + "Directors", + "Ages", + "Titles" + ], + "tokens": [] + } + }, + { + "chunk_id": "27c29d8b-8af0-5e5c-98fa-11c63b8afdbb", + "type": "text", + "content": "The following table sets forth certain information relating to our directors and chief executive officer.\n\n[tables/table-10 Board of Directors.html]\n\n$^{\\dagger}$ Director nominated by the Alibaba Partnership.\n(a) Group I directors. Current term of office will expire at our 2027 annual general meeting.\n(b) Group II directors. Current term of office will expire at our 2028 annual general meeting.\n(c) Group III directors. Current term of office will expire at our 2026 annual general meeting.\n(d) Beginning from April 1, 2025, Maggie Wu has started to serve as a non-executive director.\n(1) The business address of our directors is 26/F Tower One, Times Square, 1 Matheson Street, Causeway Bay, Hong Kong SAR., the People's Republic of China.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER", + "metadata": { + "length": 752, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "table", + "sets", + "information", + "relating", + "directors", + "chief", + "executive", + "officer", + "dagger", + "Director", + "nominated", + "Alibaba", + "Partnership", + "Group", + "Current", + "term", + "office", + "expire", + "2027", + "annual", + "general", + "meeting", + "II", + "2028", + "III", + "2026", + "Beginning", + "April", + "2025", + "Maggie", + "Wu", + "started", + "serve", + "director", + "business", + "address", + "26", + "Tower", + "One", + "Times", + "Square", + "Matheson", + "Street", + "Causeway", + "Bay", + "Hong", + "Kong", + "SAR", + "People", + "Republic", + "China" + ], + "keywords": [], + "connect_to": [ + { + "target": "e2c7d6e5-68fd-5d7c-8402-f83871e5a379", + "relation": "embeds", + "ref": "[tables/table-10 Board of Directors.html]", + "position": { + "start": 107, + "end": 148 + } + } + ] + } + }, + { + "chunk_id": "f684a444-f5a8-5c5d-82cc-7e27dd8954d1", + "type": "text", + "content": "Joseph C. TSAI (蔡崇信) joined our company in 1999 as a member of the Alibaba founding team and has served on our board of directors since our inception. He was chief financial officer until 2013, our executive vice chairman until September 2023 and currently serves as our Chairman. Joe is a founding member of the Alibaba Partnership and a board member of our affiliate Ant Group. From 1995 to 1999, Joe was a private equity investor based in Asia with Investor AB of Sweden's Wallenberg family. Prior to that, he was general counsel of Rosecliff, Inc., a management buyout firm based in New York. From 1990 to 1993, Joe was an associate attorney in the tax group of Sullivan & Cromwell LLP, a New York-based international law firm. Joe is qualified to practice law in the State of New York. Joe received his bachelor's degree in Economics and East Asian Studies from Yale College and a juris doctor degree from Yale Law School.\nEddie Yongming WU (吳泳銘) has served as our Chief Executive Officer and director since September 2023. Eddie is one of our co-founders and a member of the Alibaba Partnership. Eddie was technology director of Alibaba at the company's inception in 1999. He served as chief technology officer of Alipay from December 2004, and became business director of our monetization platform, Alimama, in November 2005 and was promoted to its general manager in December 2007. In September 2008, he became chief technology officer of Taobao, and in October 2011 he took on the role of head of Alibaba Group's search, advertising and mobile business. Eddie served as a non-executive director of Alibaba Health Information Technology Limited, a company listed on the Main Board of the Hong Kong Stock Exchange, from April 2015 to October 2021 and chairman of Alibaba Health from April 2015 to March 2020. From September 2014 to September 2019, Eddie was a special assistant to Alibaba Group's chairman. In August 2015, Eddie founded Vision Plus Capital, a venture capital firm focused on investing in the areas of advanced technologies, enterprise services and digital healthcare. Eddie graduated from the College of Information Engineering of Zhejiang University of Technology in June 1996.\nJ. Michael EVANS has been our president since August 2015 and our director since September 2014. Mike served as vice chairman of The Goldman Sachs Group, Inc. from February 2008 until his retirement in December 2013. He served as chairman of Asia operations at Goldman Sachs from 2004 to 2013 and was the global head of Growth Markets at Goldman Sachs from January 2011 to December 2013. He also co-chaired the Business Standards Committee of Goldman Sachs from 2010 to 2013. Mike joined Goldman Sachs in 1993, became a partner of the firm in 1994 and held various leadership positions within the firm’s securities business while based in New York and London, including global head of equity capital markets and global co-head of the equities division, and global co-head of the securities business. Mike is a trustee of the Asia Society and a member of the Advisory Council for the Bendheim Center for Finance at Princeton University. Mike received his bachelor’s degree in politics from Princeton University in 1981.\nMaggie Wei WU (武衛) has been our director since September 2020. Maggie joined our company in July 2007 as chief financial officer of Alibaba.com. She served as our chief financial officer from May 2013 to March 2022 and our head of strategic investments from June 2019 to March 2022. She was voted the best CFO in FinanceAsia’s annual poll for Asia’s Best Managed Companies in 2010. In 2018, she was named as one of the world’s 100 most powerful women by Forbes. Before joining Alibaba, Maggie was an audit partner at KPMG in Beijing. Maggie is a member of the Association of Chartered Certified Accountants (ACCA). She received a bachelor’s degree in accounting from Capital University of Economics and Business.\nJerry YANG (楊致遠) has served as our director since September 2014 and previously held the role from October 2005 to January 2012. Since March 2012, Jerry has been the founding partner of AME Cloud Ventures, a venture capital firm. Jerry is also a co-founder of Yahoo! Inc., where he served as Chief Yahoo! and as a member of its board of directors from March 1995 to January 2012. Jerry held the position of Yahoo!'s CEO from June 2007 to January 2009 and served as a director of Yahoo! Japan from January 1996 to January 2012. Jerry also served as an independent director of Cisco Systems, Inc. from July 2000 to November 2012 and Lenovo Group Limited, a company listed on the Main Board of the Hong Kong Stock Exchange, from November 2014 to November 2023. He is currently an independent director of Workday Inc., a company listed on the NYSE. He also serves as a director of various private companies and foundations. Jerry holds bachelor's and master's degrees in electrical engineering from Stanford University. Since 2017, he has served on Stanford's board of Trustees, including as board chair from 2021 to 2025. He was previously on Stanford's board of Trustees from 2005 to 2015, during which he also held the role of vice chair.\nWan Ling MARTELLO has been our director since September 2015. She is a founding partner of BayPine, a private equity firm based in Boston, U.S.A., a role she has held since February 2020. She is on the board of portfolio companies of BayPine. She has also served as a director of Nasdaq-listed Lovesac since November 2025. She served as the executive vice president and chief executive officer of the Asia, Oceania, and sub-Saharan Africa region for Nestlé SA from May 2015 to December 2018. She was Nestlé’s global chief financial officer from April 2012 to May 2015, and executive vice president from November 2011 to March 2012. Prior to Nestlé, Wan Ling was a senior executive at Walmart Stores Inc., a global retailer, from 2005 to 2011. Her roles included executive vice president and chief operating officer for Global eCommerce, and senior vice president, chief financial officer and strategy for Walmart International. Before Walmart, she was president, U.S.A. at NCH Marketing Services Inc. She was with the firm from 1998 to 2005. She also worked at Borden Foods Corporation and Kraft Inc. where she held various senior management positions. Wan Ling received a master’s degree in business administration (management information systems) from the University of Minnesota and a bachelor’s degree in business administration and accountancy from the University of the Philippines.\nWeijian SHAN (單偉建) has been our director since March 2022. He is the executive chairman and a co-founder of PAG, a leading private equity firm in Asia. Between 1998 and 2010, he was a partner of the private equity firm TPG and co-managing partner of TPG Asia (formerly known as Newbridge Capital). Previously, he was a managing director of JP Morgan, where he was concurrently the chief representative for China between 1993 and 1998. He was an assistant professor at the Wharton School of the University of Pennsylvania between 1987 and 1993. Shan is a Trustee of the British Museum. He is also a member of the International Advisory Council of Hong Kong Exchanges and Clearing Limited. He served as an independent director of Singapore-listed Wilmar International Limited between 2018 and 2021. He also served as a director of Nasdaq-listed iQiyi, Inc since December 2022. He holds an M.A. and a Ph.D. from the University of California, Berkeley, and an M.B.A. from the University of San Francisco. He graduated with a major in English from the Beijing Institute of Foreign Trade (currently the Beijing University of International Business and Economics).\nIrene Yun-Lien LEE (利蘊蓮) has been our director since August 2022. Irene is the executive chairman of Hysan Development Limited and serves as a member of the board of trustees of the Better Hong Kong Foundation. Irene was on the board of many listed and unlisted companies in Hong Kong, Singapore, UK and Australia. She was a member of the Australian Takeovers Panel, a member of the Advisory Council of JP Morgan Australia, and a member of the Exchange Fund Advisory Committee of the Hong Kong Monetary Authority. She was the independent non-executive chairman of Hang Seng Bank Limited, an independent non-executive director of HSBC Holding plc and The Hongkong and Shanghai Banking Corporation Limited. She was also an independent non-executive director of CLP Holdings Limited and Cathay Pacific Airways Limited, amongst others. Irene had a long career in financial services and held senior positions at Citibank in New York, London and Sydney. She was the global head of corporate finance at the Commonwealth Bank of Australia and she held other senior positions in investment banking and funds management in a number of international financial institutions. Irene received a Bachelor of Arts degree from Smith College, United States of America, and is a Barrister-at-Law in England and Wales and a member of the Honourable Society of Gray's Inn, United Kingdom. She was awarded the degree of Doctor of Social Science, honoris causa from the Chinese University of Hong Kong in November 2022.\nAlbert Kong Ping NG (吳港平) has been our director since August 2022 and chairman of our Audit Committee since December 2022. Albert currently serves as an independent non-executive director and chairman of the audit committee of a number of public companies, including Ping An Insurance (Group) Company of China, Ltd., a company listed on the Shanghai Stock Exchange and the Hong Kong Stock Exchange, China International Capital Corporation Limited, a company listed on the Shanghai Stock Exchange and the Hong Kong Stock Exchange and Shui On Land Limited, a company listed on the Hong Kong Stock Exchange. Albert worked at Ernst & Young China from April 2007 to June 2020, where he was the chairman of Ernst & Young China and a member of Ernst & Young's Global Executive Board. Prior to joining Ernst & Young, he was Greater China managing partner of Arthur Andersen, managing partner – China Operation of PricewaterhouseCoopers and managing director of Citigroup – China Investment Banking. Albert is a member of the Hong Kong Institute of Certified Public Accountants (HKICPA), Chartered Accountants of Australia and New Zealand (CAANZ), CPA Australia (CPAA) and Association of Chartered Certified Accountants (ACCA). He received a bachelor's degree in business administration and a master's degree in business administration from the Chinese University of Hong Kong.\nKabir MISRA has been our director since September 2020, redesignated as our independent director since February 2023, and is currently managing partner at RPS Ventures, a venture capital firm in Palo Alto, CA. Prior to October 2018, Kabir was a managing partner at SoftBank Investment Advisors (which manages the SoftBank Vision Fund) and SoftBank Capital. He worked with SoftBank from 2006 to 2022 (as advisor from\n2018 to 2022) and has assisted Mr. Masayoshi Son with our company, and his duties as one of our directors, since before our IPO. Kabir also represented SoftBank at various points on the boards of its investee companies, including other e-commerce and payments companies Flipkart, Paytm, Tokopedia, Coupang and BigCommerce. Prior to joining SoftBank, Kabir worked as an investment banker in the U.S. and Hong Kong. Kabir is currently also an independent director of PayActiv and Cargomatic. He received a Bachelor of Arts degree in Economics from Harvard University and a master's degree in business administration from the Stanford Graduate School of Business.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Biographical Information", + "metadata": { + "length": 11661, + "summary": "The text provides detailed biographies of 10 members of Alibaba Group's board of directors, including Joseph C. Tsai (Chairman), Eddie Yongming Wu (CEO), J. Michael Evans (President), Maggie Wei Wu, Jerry Yang, Wan Ling Martello, Weijian Shan, Irene Yun-Lien Lee, Albert Kong Ping Ng, and Kabir Misra. Each entry outlines their professional background, education, key roles at Alibaba and other companies, and board memberships.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Joseph", + "TSAI", + "崇信", + "joined", + "company", + "1999", + "member", + "Alibaba", + "founding", + "team", + "served", + "board", + "directors", + "inception", + "He", + "chief", + "financial", + "officer", + "2013", + "executive", + "vice", + "chairman", + "September", + "2023", + "serves", + "Chairman", + "Joe", + "Partnership", + "affiliate", + "Ant", + "Group", + "From", + "1995", + "private", + "equity", + "investor", + "based", + "Asia", + "Investor", + "AB", + "Sweden", + "Wallenberg", + "family", + "Prior", + "general", + "counsel", + "Rosecliff", + "Inc", + "management", + "buyout", + "firm", + "New", + "York", + "1990", + "1993", + "associate", + "attorney", + "tax", + "group", + "Sullivan", + "Cromwell", + "LLP", + "international", + "law", + "qualified", + "practice", + "State", + "received", + "bachelor", + "degree", + "Economics", + "East", + "Asian", + "Studies", + "Yale", + "College", + "juris", + "doctor", + "Law", + "School", + "Eddie", + "Yongming", + "WU", + "吳泳銘", + "Chief", + "Executive", + "Officer", + "director", + "founders", + "technology", + "Alipay", + "December", + "2004", + "business", + "monetization", + "platform", + "Alimama", + "November", + "2005", + "promoted", + "manager", + "2007", + "In", + "2008", + "Taobao", + "October", + "2011", + "role", + "head", + "search", + "advertising", + "mobile", + "Health", + "Information", + "Technology", + "Limited", + "listed", + "Main", + "Board", + "Hong", + "Kong", + "Stock", + "Exchange", + "April", + "2015", + "2021", + "March", + "2020", + "2014", + "2019", + "special", + "assistant", + "August", + "founded", + "Vision", + "Plus", + "Capital", + "venture", + "capital", + "focused", + "investing", + "areas", + "advanced", + "technologies", + "enterprise", + "services", + "digital", + "healthcare", + "graduated", + "Engineering", + "Zhejiang", + "University", + "June", + "1996", + "Michael", + "EVANS", + "president", + "Mike", + "The", + "Goldman", + "Sachs", + "February", + "retirement", + "operations", + "global", + "Growth", + "Markets", + "January", + "chaired", + "Business", + "Standards", + "Committee", + "2010", + "partner", + "1994", + "held", + "leadership", + "positions", + "securities", + "London", + "including", + "markets", + "equities", + "division", + "trustee", + "Society", + "Advisory", + "Council", + "Bendheim", + "Center", + "Finance", + "Princeton", + "politics", + "1981", + "Maggie", + "Wei", + "武衛", + "July", + "She", + "May", + "2022", + "strategic", + "investments", + "voted", + "CFO", + "FinanceAsia", + "annual", + "poll", + "Best", + "Managed", + "Companies", + "2018", + "named", + "world", + "100", + "powerful", + "women", + "Forbes", + "Before", + "joining", + "audit", + "KPMG", + "Beijing", + "Association", + "Chartered", + "Certified", + "Accountants", + "ACCA", + "accounting", + "Jerry", + "YANG", + "楊致", + "previously", + "2012", + "Since", + "AME", + "Cloud", + "Ventures", + "founder", + "Yahoo", + "position", + "CEO", + "2009", + "Japan", + "independent", + "Cisco", + "Systems", + "2000", + "Lenovo", + "Workday", + "NYSE", + "companies", + "foundations", + "holds", + "master", + "degrees", + "electrical", + "engineering", + "Stanford", + "2017", + "Trustees", + "chair", + "2025", + "Wan", + "Ling", + "MARTELLO", + "BayPine", + "Boston", + "portfolio", + "Nasdaq", + "Lovesac", + "Oceania", + "Saharan", + "Africa", + "region", + "Nestl", + "SA", + "senior", + "Walmart", + "Stores", + "retailer", + "Her", + "roles", + "included", + "operating", + "Global", + "eCommerce", + "strategy", + "International", + "NCH", + "Marketing", + "Services", + "1998", + "worked", + "Borden", + "Foods", + "Corporation", + "Kraft", + "administration", + "information", + "systems", + "Minnesota", + "accountancy", + "Philippines", + "Weijian", + "SHAN", + "單偉建", + "PAG", + "leading", + "Between", + "TPG", + "managing", + "Newbridge", + "Previously", + "JP", + "Morgan", + "concurrently", + "representative", + "China", + "professor", + "Wharton", + "Pennsylvania", + "1987", + "Shan", + "Trustee", + "British", + "Museum", + "Exchanges", + "Clearing", + "Singapore", + "Wilmar", + "iQiyi", + "Ph", + "California", + "Berkeley", + "San", + "Francisco", + "major", + "English", + "Institute", + "Foreign", + "Trade", + "Irene", + "Yun", + "Lien", + "LEE", + "利蘊蓮", + "Hysan", + "Development", + "trustees", + "Better", + "Foundation", + "unlisted", + "UK", + "Australia", + "Australian", + "Takeovers", + "Panel", + "Fund", + "Monetary", + "Authority", + "Hang", + "Seng", + "Bank", + "HSBC", + "Holding", + "plc", + "Hongkong", + "Shanghai", + "Banking", + "CLP", + "Holdings", + "Cathay", + "Pacific", + "Airways", + "long", + "career", + "Citibank", + "Sydney", + "corporate", + "finance", + "Commonwealth", + "investment", + "banking", + "funds", + "number", + "institutions", + "Bachelor", + "Arts", + "Smith", + "United", + "States", + "America", + "Barrister", + "England", + "Wales", + "Honourable", + "Gray", + "Inn", + "Kingdom", + "awarded", + "Doctor", + "Social", + "Science", + "honoris", + "causa", + "Chinese", + "Albert", + "Ping", + "NG", + "吳港平", + "Audit", + "committee", + "public", + "An", + "Insurance", + "Company", + "Ltd", + "Shui", + "On", + "Land", + "Ernst", + "Young", + "Greater", + "Arthur", + "Andersen", + "Operation", + "PricewaterhouseCoopers", + "Citigroup", + "Investment", + "Public", + "HKICPA", + "Zealand", + "CAANZ", + "CPA", + "CPAA", + "Kabir", + "MISRA", + "redesignated", + "RPS", + "Palo", + "Alto", + "CA", + "SoftBank", + "Advisors", + "manages", + "2006", + "advisor", + "assisted", + "Mr", + "Masayoshi", + "Son", + "duties", + "IPO", + "represented", + "points", + "boards", + "investee", + "commerce", + "payments", + "Flipkart", + "Paytm", + "Tokopedia", + "Coupang", + "BigCommerce", + "banker", + "PayActiv", + "Cargomatic", + "Harvard", + "Graduate" + ], + "keywords": [ + "Alibaba", + "Board of Directors", + "Biographies" + ], + "connect_to": [] + } + }, + { + "chunk_id": "d1fcf8ba-8289-5e6b-a87b-991967302ae3", + "type": "text", + "content": "Our board of directors has established an audit committee, a compensation committee, a nominating and corporate governance committee, a sustainability committee, a compliance and risk committee and a capital management committee.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees", + "metadata": { + "length": 229, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "board", + "directors", + "established", + "audit", + "committee", + "compensation", + "nominating", + "corporate", + "governance", + "sustainability", + "compliance", + "risk", + "capital", + "management" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "eb70153e-7ea5-5263-969d-332a07532a8a", + "type": "text", + "content": "Our audit committee currently consists of Albert Ng, Wan Ling Martello and Weijian Shan. Mr. Ng is the chairman of our audit committee.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Audit Committee", + "metadata": { + "length": 135, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "audit", + "committee", + "consists", + "Albert", + "Ng", + "Wan", + "Ling", + "Martello", + "Weijian", + "Shan", + "Mr", + "chairman" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "88634be9-ec1b-5fb8-8b03-e0a052ae0332", + "type": "text", + "content": "Our compensation committee currently consists of Jerry Yang, Albert Ng and Kabir Misra. Mr. Yang is the chairman of our compensation committee.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Compensation Committee", + "metadata": { + "length": 143, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "compensation", + "committee", + "consists", + "Jerry", + "Yang", + "Albert", + "Ng", + "Kabir", + "Misra", + "Mr", + "chairman" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "22ae1fa1-4a25-5b3e-8fcf-e5b86a463ed2", + "type": "text", + "content": "Our nominating and corporate governance committee currently consists of Irene Lee and Jerry Yang. Ms. Lee is the chairman of our nominating and corporate governance committee.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Nominating and Corporate Governance Committee", + "metadata": { + "length": 175, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "nominating", + "corporate", + "governance", + "committee", + "consists", + "Irene", + "Lee", + "Jerry", + "Yang", + "Ms", + "chairman" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "122df0ff-bc5d-526a-b1db-71d7b9a3586a", + "type": "text", + "content": "Our sustainability committee currently consists of Jerry Yang, Joe Tsai and Maggie Wu. Mr. Yang is the chairman of our sustainability committee.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Sustainability Committee", + "metadata": { + "length": 144, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "sustainability", + "committee", + "consists", + "Jerry", + "Yang", + "Joe", + "Tsai", + "Maggie", + "Wu", + "Mr", + "chairman" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9943eae6-89dd-5753-9c12-5db691b70a65", + "type": "text", + "content": "Our compliance and risk committee currently consists of Irene Lee, Albert Ng, Kabir Misra and J. Michael Evans. Ms. Lee is the chairman of our compliance and risk committee.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Compliance and Risk Committee", + "metadata": { + "length": 173, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "compliance", + "risk", + "committee", + "consists", + "Irene", + "Lee", + "Albert", + "Ng", + "Kabir", + "Misra", + "Michael", + "Evans", + "Ms", + "chairman" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "029a6af5-c4e0-5bb9-b85b-cd29cf1c0408", + "type": "text", + "content": "Our capital management committee currently consists of Joe Tsai, Eddie Wu, J. Michael Evans and Maggie Wu. Mr. Tsai is the chairman of our capital management committee.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Capital Management Committee", + "metadata": { + "length": 168, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "capital", + "management", + "committee", + "consists", + "Joe", + "Tsai", + "Eddie", + "Wu", + "Michael", + "Evans", + "Maggie", + "Mr", + "chairman" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1395476e-8563-522a-bfaa-57757f3eca57", + "type": "table", + "content": "
Name of director / chief executive officerCapacity / nature of interestNumber of shares/Approximate % of shareholding (3)
underlying shares (1),(2)
(in the number of Shares)(in the number of ADSs)
Joseph C. TSAI (4)Beneficial owner881,072 (L)110,134 (L)0.00%
Interest of spouse1,280,000 (L)160,000 (L)0.01%
Founder of a discretionary trust who can influence how the trustee exercises his discretion12,658,152 (L)1,582,269 (L)0.06%
Interest in controlled corporation260,924,840 (L)32,615,605 (L)1.37%
275,744,064 (L)34,468,008 (L)1.44%
Eddie Yongming WU (5)Beneficial owner18,199,752 (L)2,274,969 (L)0.10%
Interest of spouse108,000 (L)13,500 (L)0.00%
Founder of a discretionary trust who can influence how the trustee exercises his discretion12,320,000 (L)1,540,000 (L)0.06%
30,627,752 (L)3,828,469 (L)0.16%
J. Michael EVANS (6)Beneficial owner9,964,000 (L)1,245,500 (L)0.05%
Maggie Wei WU (7)Beneficial owner3,661,632 (L)457,704 (L)0.02%
Founder of a discretionary trust who can influence how the trustee exercises his discretion7,200,000 (L)900,000 (L)0.04%
10,861,632 (L)1,357,704 (L)0.06%
Jerry YANGBeneficial owner485,072 (L)60,634 (L)0.00%
Wan Ling MARTELLOBeneficial owner360,000 (L)45,000 (L)0.00%
Weijian SHANBeneficial owner144,800 (L)18,100 (L)0.00%
Irene Yun-Lien LEEBeneficial owner133,600 (L)16,700 (L)0.00%
Albert Kong Ping NGBeneficial owner133,600 (L)16,700 (L)0.00%
Kabir MISRABeneficial owner528,800 (L)66,100 (L)0.00%
", + "path": "tables/table-11 Shareholding Table.html", + "metadata": { + "length": 2561, + "summary": "table-12\nThe table lists directors and CEOs with their shareholding interests, including beneficial ownership, spouse interests, and controlled corporations, showing share and ADS counts with percentages.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-11 Shareholding Table.html", + "keywords": [ + "shareholding", + "directors", + "ADSs" + ], + "tokens": [] + } + }, + { + "chunk_id": "883f191e-f742-58a1-96f7-1f7f61f7521e", + "type": "table", + "content": "
Name of shareholderCapacity / nature of interestNumber of shares/ underlying shares (1),(2)Approximate % of shareholding (3)
(in the number of Shares)(in the number of ADSs)
JPMorgan Chase & Co. (4)Beneficial owner411,801,958 (L)51,475,245 (L)2.16%
418,739,874 (S)52,342,484 (S)2.19%
Trustee1,753,320 (L)219,165 (L)0.01%
Investment manager281,850,861 (L)35,231,358 (L)1.47%
6,527,284 (S)815,911 (S)0.04%
Person having a security interest in shares85,927,145 (L)10,740,893 (L)0.45%
Approved lending agent581,658,439 (L)72,707,305 (L)3.05%
581,658,439 (P)72,707,305 (P)3.05%
1,362,991,723 (L)170,373,966 (L)7.14%
425,267,158 (S)53,158,395 (S)2.23%
581,658,439 (P)72,707,305 (P)3.05%
Citigroup Inc. (5)Interest in controlled corporation183,474,400 (L)22,934,300 (L)0.96%
197,906,617 (S)24,738,327 (S)1.04%
Person having a security interest in shares160 (L)20 (L)0.00%
Approved lending agent832,721,084 (L)104,090,136 (L)4.36%
832,721,084 (P)104,090,136 (P)4.36%
1,016,195,644 (L)127,024,456 (L)5.32%
197,906,617 (S)24,738,327 (S)1.04%
832,721,084 (P)104,090,136 (P)4.36%
BlackRock, Inc. (6)Interest in controlled corporation983,512,344 (L)122,939,043 (L)5.15%
3,297,800 (S)412,225 (S)0.02%
", + "path": "tables/table-12 Shareholder Holdings.html", + "metadata": { + "length": 2276, + "summary": "table-13\nThe table lists major shareholders (JPMorgan, Citigroup, BlackRock) with their share counts, ADS equivalents, and approximate percentages of total shareholding, categorized by capacity (beneficial owner, trustee, etc.).", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-12 Shareholder Holdings.html", + "keywords": [ + "shareholders", + "holdings", + "percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "18b2d082-7b58-5dcc-93a5-b1c7674dfa4b", + "type": "text", + "content": "The following tables set forth:\n- the interest and short positions of our directors and chief executive officer in the shares, underlying shares and debentures of our Company or its associated corporations within the meaning of Part XV of the SFO, which were required (a) to be notified to us and the Hong Kong Stock Exchange pursuant to Divisions 7 and 8 of Part XV of the SFO (including interests and short positions which they were taken or deemed to have under such provisions of the SFO); or (b) to be recorded in the register required to be kept by us pursuant to Section 352 of the SFO; or (c) as otherwise notified to us and the Hong Kong Stock Exchange pursuant to the Model Code; and\n- the interest and short positions of persons other than our directors and chief executive officer in the Shares and underlying Shares which would fall to be disclosed to us pursuant to Divisions 2 and 3 of Part XV of the SFO or as recorded in the register required to be kept by us pursuant to Section 336 of the SFO,\nInformation about persons other than our directors and chief executive officer, including our substantial shareholders (as defined under SFO), was obtained solely from publicly available information on the website of the Hong Kong Stock Exchange or as otherwise the Company is aware, and the Company cannot guarantee the accuracy and completeness of such information.\nDirectors and Chief Executive Officer\nInterest in our Company\n\n[tables/table-11 Shareholding Table.html]\n\nNotes:\n1. The letter \"L\" stands for long position.\n2. Each ADS represents eight Shares.\n3. The calculation is based on a total of 19,088,280,284 Shares in issue as of September 30, 2025.\n4. The interests comprised (i) 804,405 Shares held by Mr. Joseph C. TSAI; (ii) 76,667 Shares underlying the outstanding RSUs granted to Mr. Joseph C. TSAI; (iii) 1,280,000 Shares held by his spouse; (iv) 12,658,152 Shares held by Joe and\nClara Tsai Foundation Limited, a company incorporated under the law of the Island of Guernsey that has granted Mr. Joseph C. TSAI a revocable proxy over these Shares and which is wholly-owned by Joe and Clara Tsai Foundation; (v) 147,385,672 Shares held by Parufam Limited, a Bahamas corporation of which Mr. Joseph C. TSAI is the sole director; and (vi) 113,539,168 Shares held by PMH Holding Limited, a British Virgin Islands corporation of which Mr. Joseph C. TSAI is the sole director.\n5. The interests comprised (i) 933,085 Shares held by Mr. Eddie Yongming WU; (ii) 1,266,667 Shares underlying the outstanding RSUs and 16,000,000 Shares underlying the outstanding share options granted to Mr. Eddie Yongming WU; (iii) 108,000 Shares held by his spouse; and (iv) 12,320,000 Shares held by a discretionary trust of which Mr. Eddie Yongming WU is a founder.\n6. The interests comprised 764,000 Shares underlying the outstanding RSUs and 9,200,000 Shares underlying the outstanding share options granted to Mr. J. Michael EVANS.\n7. The interests comprised (i) 3,613,352 Shares held by Ms. Maggie Wei WU; (ii) 48,280 Shares underlying the outstanding RSUs granted to Ms. Maggie Wei WU; and (iii) 7,200,000 Shares held by a discretionary trust of which Ms. Maggie Wei WU is a founder.\nSave as disclosed above, as of September 30, 2025, none of our directors or chief executive officer had any interest or short position in the shares, underlying shares or debentures of our Company or any of its associated corporations (within the meaning of Part XV of the SFO) which were required to be notified to us and the Hong Kong Stock Exchange pursuant to Divisions 7 and 8 of Part XV of the SFO (including interests and short positions which were taken or deemed to have taken under such provisions of the SFO), or which were recorded in the register required to be kept pursuant to Section 352 of the SFO, to be entered in the register referred to therein, or which were required, pursuant to the Model Code, to be notified to us and the Hong Kong Stock Exchange.\nSubstantial Shareholders\n\n[tables/table-12 Shareholder Holdings.html]\n\nNotes:\n1. The letter “L” stands for long position, the letter “S” stands for short position and the letter “P” stands for lending pool.\n2. Each ADS represents eight Shares. The number of ADSs is, where applicable, rounded down to the nearest whole number and for reference only.\n3. The calculation is based on a total of 19,088,280,284 Shares in issue as of September 30, 2025.\n4. According to the disclosure of interests notice filed by JPMorgan Chase & Co. regarding the relevant event dated September 26, 2025, (a) 411,801,958 Shares (long position) and 418,739,874 Shares (short position) were held by JPMorgan Chase & Co. directly; (b) 1,753,320 Shares (long position) were held by a trust of which JPMorgan Chase & Co. is a trustee; and (c) the remaining interests were held by JPMorgan Chase & Co. indirectly through certain of its controlled corporations. Among them, (i) 105,285,940 Shares (long position) and 105,155,304 Shares (short position) were held through physically settled listed derivatives; (ii) 6,797,279 Shares (long position) and 18,013,180 Shares (short position) were held through cash settled listed derivatives; (iii) 40,691,926 Shares (long position) and 48,319,704 Shares (short position) were held through physically settled unlisted derivatives; (iv) 87,425,503 Shares (long position) and\n90,603,044 Shares (short position) were held through cash settled unlisted derivatives; and (v) 84,973,906 Shares (long position) and 1,381,027 Shares (short position) were held through listed derivatives which are convertible instruments.\n5. According to the disclosure of interests notice filed by Citigroup Inc. regarding the relevant event dated September 11, 2025, 183,474,400 Shares (long position) and 197,906,617 Shares (short position) interests were held by Citigroup Inc. indirectly through certain of its controlled corporations. Among them, (i) 54,743,028 Shares (long position) and 36,327,420 Shares (short position) were held through physically settled listed derivatives; (ii) 77,818,263 Shares (long position) and 120,475,576 Shares (short position) were held through physically settled unlisted derivatives; (iii) 13,761,835 Shares (long position) and 37,337,978 Shares (short position) were held through cash settled unlisted derivatives; and (iv) 82,139 Shares (long position) were held through listed derivatives which are convertible instruments.\n6. According to the disclosure of interests notice filed by Blackrock, Inc. regarding the relevant event dated June 13, 2025, 983,512,344 Shares (long position) and 3,297,800 Shares (short position) were held by Blackrock, Inc. indirectly through certain of its controlled corporations. Among them, (i) 6,137,756 Shares (long position) and 3,297,800 Shares (short position) were held through cash settled unlisted derivatives; and (ii) 6,572,498 Shares (long position) were held through listed derivatives which are convertible instruments.\nSave as disclosed above, as of September 30, 2025, so far as known to the Company, no person (other than our directors and chief executive officer) had an interest or short position in the Shares or underlying Shares which would fall to be disclosed to us under Divisions 2 and 3 of Part XV of the SFO, or would fall to be recorded in the register required to be kept by us pursuant to Section 336 of the SFO.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DISCLOSURE OF INTERESTS", + "metadata": { + "length": 7501, + "summary": "The document details the interests and short positions in shares of the company as of September 30, 2025, for directors, the CEO, and substantial shareholders, as required under the Securities and Futures Ordinance (SFO). It includes tables and notes for directors like Joseph C. Tsai, Eddie Yongming Wu, J. Michael Evans, and Maggie Wei Wu, listing their holdings through shares, RSUs, options, trusts, and foundations. For substantial shareholders, it covers JPMorgan Chase & Co., Citigroup Inc., and Blackrock, Inc., with breakdowns of long and short positions, including derivatives. The report confirms no other disclosable interests beyond those listed.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "tables", + "set", + "interest", + "short", + "positions", + "directors", + "chief", + "executive", + "officer", + "shares", + "underlying", + "debentures", + "Company", + "corporations", + "meaning", + "Part", + "XV", + "SFO", + "required", + "notified", + "Hong", + "Kong", + "Stock", + "Exchange", + "pursuant", + "Divisions", + "including", + "interests", + "deemed", + "provisions", + "recorded", + "register", + "Section", + "352", + "Model", + "Code", + "persons", + "Shares", + "fall", + "disclosed", + "336", + "Information", + "substantial", + "shareholders", + "defined", + "obtained", + "solely", + "publicly", + "information", + "website", + "aware", + "guarantee", + "accuracy", + "completeness", + "Directors", + "Chief", + "Executive", + "Officer", + "Interest", + "Notes", + "letter", + "stands", + "long", + "position", + "Each", + "ADS", + "represents", + "calculation", + "based", + "total", + "19", + "088", + "280", + "284", + "issue", + "September", + "30", + "2025", + "comprised", + "804", + "405", + "held", + "Mr", + "Joseph", + "TSAI", + "ii", + "76", + "667", + "outstanding", + "RSUs", + "granted", + "iii", + "000", + "spouse", + "iv", + "12", + "658", + "152", + "Joe", + "Clara", + "Tsai", + "Foundation", + "Limited", + "company", + "incorporated", + "law", + "Island", + "Guernsey", + "revocable", + "proxy", + "wholly", + "owned", + "147", + "385", + "672", + "Parufam", + "Bahamas", + "corporation", + "sole", + "director", + "vi", + "113", + "539", + "168", + "PMH", + "Holding", + "British", + "Virgin", + "Islands", + "933", + "085", + "Eddie", + "Yongming", + "WU", + "266", + "16", + "share", + "options", + "108", + "320", + "discretionary", + "trust", + "founder", + "764", + "200", + "Michael", + "EVANS", + "613", + "Ms", + "Maggie", + "Wei", + "48", + "Save", + "entered", + "referred", + "Substantial", + "Shareholders", + "lending", + "pool", + "number", + "ADSs", + "applicable", + "rounded", + "nearest", + "reference", + "According", + "disclosure", + "notice", + "filed", + "JPMorgan", + "Chase", + "Co", + "relevant", + "event", + "dated", + "26", + "411", + "801", + "958", + "418", + "739", + "874", + "directly", + "753", + "trustee", + "remaining", + "indirectly", + "controlled", + "Among", + "105", + "285", + "940", + "155", + "304", + "physically", + "settled", + "listed", + "derivatives", + "797", + "279", + "18", + "013", + "180", + "cash", + "40", + "691", + "926", + "319", + "704", + "unlisted", + "87", + "425", + "503", + "90", + "603", + "044", + "84", + "973", + "906", + "381", + "027", + "convertible", + "instruments", + "Citigroup", + "Inc", + "11", + "183", + "474", + "400", + "197", + "617", + "54", + "743", + "028", + "36", + "327", + "420", + "77", + "818", + "263", + "120", + "475", + "576", + "13", + "761", + "835", + "37", + "337", + "978", + "82", + "139", + "Blackrock", + "June", + "983", + "512", + "344", + "297", + "800", + "137", + "756", + "572", + "498", + "person" + ], + "keywords": [ + "Directors", + "Substantial Shareholders", + "SFO" + ], + "connect_to": [ + { + "target": "1395476e-8563-522a-bfaa-57757f3eca57", + "relation": "embeds", + "ref": "[tables/table-11 Shareholding Table.html]", + "position": { + "start": 1444, + "end": 1485 + } + }, + { + "target": "883f191e-f742-58a1-96f7-1f7f61f7521e", + "relation": "embeds", + "ref": "[tables/table-12 Shareholder Holdings.html]", + "position": { + "start": 3996, + "end": 4039 + } + } + ] + } + }, + { + "chunk_id": "b39dda78-f6f6-52c7-814a-fcc8411ef71a", + "type": "text", + "content": "The 2014 Plan, the 2024 Plan and the 2024 Plan (Existing Shares) are the equity incentive plans of our Company in effect. For a summary of the key terms of these plans, please see “Directors, Senior Management and Employees - Equity Incentive Plans” in our fiscal year 2025 annual report.\nDuring the Reporting Period, a total of 63,594,318 Shares (equivalent to approximately 7,949,290 ADSs), representing approximately 0.3% of the weighted average number of Shares in issue (excluding treasury shares) of our Company, that may be issued in respect of all share-based awards were granted to eligible participants under the 2024 Plan of our Company.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS", + "metadata": { + "length": 648, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "2014", + "Plan", + "2024", + "Existing", + "Shares", + "equity", + "incentive", + "plans", + "Company", + "effect", + "For", + "summary", + "key", + "terms", + "Directors", + "Senior", + "Management", + "Employees", + "Equity", + "Incentive", + "Plans", + "fiscal", + "year", + "2025", + "annual", + "report", + "During", + "Reporting", + "Period", + "total", + "63", + "594", + "318", + "equivalent", + "approximately", + "949", + "290", + "ADSs", + "representing", + "0.3%", + "weighted", + "average", + "number", + "issue", + "excluding", + "treasury", + "shares", + "issued", + "respect", + "share", + "based", + "awards", + "granted", + "eligible", + "participants" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cbf659a3-f78f-5e7d-b07a-81c2c24f6057", + "type": "table", + "content": "
Name of granteeDate of grantNumber of RSUs (1), (2)Number of underlying Shares as of September 30, 2025 (2)Notes
Outstanding as of April 1, 2025Granted during the Reporting PeriodVested during the Reporting PeriodLapsed/ forfeited during the Reporting PeriodCancelled during the Reporting PeriodOutstanding as of September 30, 2025
Directors and chief executive officer of our Company
Joseph C. TSAIJuly 24, 2018 to August 16, 20191,334-1,334----Notes 3, 8
Eddie Yongming WUNovember 25, 2023160,000-40,000--120,000960,000Notes 4, 8
J. Michael EVANSJune 15, 2020 to May 20, 202310,050-4,550--5,50044,000Notes 5, 8
May 13, 2024160,00080,000-80,000640,000Notes 6, 8
Maggie Wei WUJuly 24, 2018 to May 24, 202112,801-6,766--6,03548,280Notes 3, 8
Other grantees by category
Employee participantsNovember 15, 2016 to August 18, 202444,402,154-16,110,6831,681,3791,82526,608,267212,866,136Notes 7, 9
Related entity $participants^{(11)}$ July 24, 2018 to June 8, 2022113,139-84,7672,578-25,794206,352Notes 7, 10
Total44,859,478-16,328,1001,683,9571,82526,845,596214,764,768
", + "path": "tables/table-13 RSU Grant Summary.html", + "metadata": { + "length": 2100, + "summary": "table-14\nTable shows RSU activity for directors, CEO, and employees from April to September 2025, with 44.9M outstanding, 16.3M vested, and 26.8M remaining.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-13 RSU Grant Summary.html", + "keywords": [ + "RSU", + "grants", + "shares" + ], + "tokens": [] + } + }, + { + "chunk_id": "bde3bc38-582d-502c-8329-8ae99e527f8e", + "type": "text", + "content": "The 2014 Plan (which we adopted in September 2014, amended and restated in February 2020 to reflect the Share Split and other administrative changes, and further amended and restated in May 2022 to reflect administrative changes) provides for the granting of share-based awards, including restricted share units, stock options, restricted shares and share appreciation rights, to any participant who is an employee, consultant or director of our company, our affiliates and/or certain other companies.\nNo further awards will be granted under the 2014 Plan as from September 18, 2024. Any shares authorized but unissued under the 2014 Plan will no longer be available for granting. The share-based awards already granted under the 2014 Plan will remain in full force and effect pursuant to the terms and conditions of the 2014 Plan.\nDetails of movements of RSUs under the 2014 Plan during the Reporting Period are as follows:\n\n[tables/table-13 RSU Grant Summary.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS/The 2014 Plan", + "metadata": { + "length": 1029, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "2014", + "Plan", + "adopted", + "September", + "amended", + "restated", + "February", + "2020", + "reflect", + "Share", + "Split", + "administrative", + "May", + "2022", + "granting", + "share", + "based", + "awards", + "including", + "restricted", + "units", + "stock", + "options", + "shares", + "appreciation", + "rights", + "participant", + "employee", + "consultant", + "director", + "company", + "affiliates", + "companies", + "No", + "granted", + "18", + "2024", + "Any", + "authorized", + "unissued", + "longer", + "remain", + "full", + "force", + "effect", + "pursuant", + "terms", + "conditions", + "Details", + "movements", + "RSUs", + "Reporting", + "Period" + ], + "keywords": [], + "connect_to": [ + { + "target": "cbf659a3-f78f-5e7d-b07a-81c2c24f6057", + "relation": "embeds", + "ref": "[tables/table-13 RSU Grant Summary.html]", + "position": { + "start": 926, + "end": 966 + } + } + ] + } + }, + { + "chunk_id": "3f1d825b-a6b9-57e9-b47a-3b10f61cf790", + "type": "table", + "content": "
Name of granteeDate of grantExercise price (US$)Number of share options (1)Number of underlying Shares as of September 30, 2025 (1)Notes
Outstanding as of April 1, 2025Granted during the Reporting PeriodExercised during the Reporting PeriodLapsed/ forfeited during the Reporting Period (9)Outstanding as of September 30, 2025
Directors and chief executive officer of our Company
Eddie Yongming WUNovember 25, 202378.372,000,000---2,000,00016,000,000Note 2
J. Michael EVANSJuly 31, 201579.962,000,000-1,000,000-1,000,0008,000,000Note 3, 6
May 13, 202484.60150,000---150,0001,200,000Note 4
Other grantees by category
Employee participantsMay 10, 2015 to March 12, 202223.00 to 182.482,385,667-842,6671,000,000543,0004,344,000Notes 5, 7
Related entity participant (10)March 12, 202223.00100,000-100,000---Notes 5, 8
Total6,635,667-1,942,6671,000,0003,693,00029,544,000
", + "path": "tables/table-14 Share Options Table.html", + "metadata": { + "length": 1716, + "summary": "table-15\nTable shows share option grants for directors and employees, including exercise prices, outstanding numbers, and changes during the reporting period ending September 30, 2025.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-14 Share Options Table.html", + "keywords": [ + "share options", + "grantees", + "exercise price" + ], + "tokens": [] + } + }, + { + "chunk_id": "983caa0b-f639-57e0-b3a3-e0a938eb9e32", + "type": "text", + "content": "1. The purchase price of all RSUs granted is nil.\n2. Eight Shares are issuable upon the vesting of each RSU.\n3. The RSUs vest over a period of 6 years.\n4. The RSUs vest over a period of 5 years.\n5. The RSUs vest over a period of 4 years.\n6. The RSUs vest over a period of 2 years.\n7. The RSUs granted to employee participants and related entity participants vest over a period of up to 10 years.\n8. For directors and chief executive officer, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the RSUs were vested during the Reporting Period was US\\$136.71.\n9. For employee participants, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the RSUs were vested during the Reporting Period was US\\$131.05\n10. For related entity participants, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the RSUs were vested during the Reporting Period was US\\$132.15.\n11. A related entity participant is an employee of an associated company in which our Company owns, directly or indirectly, securities or interests representing 20% or more of its voting power.\nDetails of movements of share options under the 2014 Plan during the Reporting Period are as follows:\n\n[tables/table-14 Share Options Table.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS/The 2014 Plan/Notes:", + "metadata": { + "length": 1403, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "purchase", + "price", + "RSUs", + "granted", + "nil", + "Eight", + "Shares", + "issuable", + "vesting", + "RSU", + "vest", + "period", + "years", + "employee", + "participants", + "related", + "entity", + "10", + "For", + "directors", + "chief", + "executive", + "officer", + "weighted", + "average", + "closing", + "ADSs", + "traded", + "NYSE", + "immediately", + "dates", + "vested", + "Reporting", + "Period", + "US", + "136.71", + "131.05", + "132.15", + "11", + "participant", + "company", + "Company", + "owns", + "directly", + "indirectly", + "securities", + "interests", + "representing", + "20%", + "voting", + "power", + "Details", + "movements", + "share", + "options", + "2014", + "Plan" + ], + "keywords": [], + "connect_to": [ + { + "target": "3f1d825b-a6b9-57e9-b47a-3b10f61cf790", + "relation": "embeds", + "ref": "[tables/table-14 Share Options Table.html]", + "position": { + "start": 1300, + "end": 1342 + } + } + ] + } + }, + { + "chunk_id": "927c4caa-9a64-5c73-93d7-31dfb673dd4a", + "type": "text", + "content": "1. Eight Shares are issuable upon the exercise of each share option.\n2. The share options vest over a period of 5 years and are exercisable within 10 years after date of grant.\n3. The share options vest over a period of 6 years and are exercisable within 12 years after date of grant.\n4. The share options vest over a period of 2 years and are exercisable within 8 years after date of grant.\n5. The share options vest over a period of up to 9 years and are exercisable within up to 12 years after date of grant.\n6. The closing price of ADSs traded on the NYSE immediately before the date on which the share options were exercised during the Reporting Period was US\\$123.90.\n7. For employee participants, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the share options were exercised during the Reporting Period was US\\$114.78.\n8. For related entity participant, the weighted average closing price of ADSs traded on the NYSE immediately before the date on which the share options were exercised during the Reporting Period was US\\$118.09.\n9. No share options granted were cancelled during the Reporting Period.\n10. A related entity participant is an employee of an associated company in which our Company owns, directly or indirectly, securities or interests representing 20% or more of its voting power.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS/The 2014 Plan/Notes:_2", + "metadata": { + "length": 1357, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Eight", + "Shares", + "issuable", + "exercise", + "share", + "option", + "The", + "options", + "vest", + "period", + "years", + "exercisable", + "10", + "date", + "grant", + "12", + "closing", + "price", + "ADSs", + "traded", + "NYSE", + "immediately", + "exercised", + "Reporting", + "Period", + "US", + "123.90", + "For", + "employee", + "participants", + "weighted", + "average", + "dates", + "114.78", + "related", + "entity", + "participant", + "118.09", + "No", + "granted", + "cancelled", + "company", + "Company", + "owns", + "directly", + "indirectly", + "securities", + "interests", + "representing", + "20%", + "voting", + "power" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "354b9644-5d8b-5d56-a5ca-8679ec93219f", + "type": "table", + "content": "
Name (3)Date of grantNumber of RSUs (1),(2)Number of underlying Shares as of September 30, 2025 (2)Notes
Outstanding as of April 1, 2025Granted during the Reporting PeriodVested during the Reporting PeriodLapsed/ forfeited during the Reporting Period (8)Outstanding as of September 30, 2025
Employee participantsNovember 19, 2024 and February 24, 20259,435,535-1,456,534236,2007,742,8017,742,801Notes 4, 5
May 23, 2025, June 13, 2025 and September 4, 2025-63,594,3183,209,7961,116,59959,267,92359,267,923Notes 5, 6, 7
Total9,435,53563,594,3184,666,3301,352,79967,010,72467,010,724
", + "path": "tables/table-15 RSU Activity Report.html", + "metadata": { + "length": 1032, + "summary": "table-16\nThe table shows RSU activity for employee participants from April to September 2025, with 63.6 million granted, 4.7 million vested, 1.4 million forfeited, and 67 million outstanding as of September 30, 2025.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-15 RSU Activity Report.html", + "keywords": [ + "RSUs", + "Employee Grants", + "Share Counts" + ], + "tokens": [] + } + }, + { + "chunk_id": "9cad7312-407b-5ad0-bc77-472a9ee56bfc", + "type": "table", + "content": "
Date of grantNumber of RSUs grantedNumber of underlying SharesClosing price of sharesimmediately before date of grant(HK$)Fair value of each RSU as of dateof grant (HK$)
May 23, 202553,429,59753,429,597119.10118.78
June 13, 20257,687,0917,687,091114.60114.60
September 4, 20252,477,6302,477,630134.10130.31
63,594,31863,594,318
", + "path": "tables/table-16 RSU Grant Details.html", + "metadata": { + "length": 608, + "summary": "table-17\nThe table shows RSU grants on three dates in 2025: May 23 (53,429,597 RSUs, HK$118.78 fair value), June 13 (7,687,091 RSUs, HK$114.60), and September 4 (2,477,630 RSUs, HK$130.31), totaling 63,594,318 RSUs.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-16 RSU Grant Details.html", + "keywords": [ + "RSU", + "Shares", + "Grant Date" + ], + "tokens": [] + } + }, + { + "chunk_id": "7f25c25b-f029-5311-9ea5-3f9a861544d9", + "type": "text", + "content": "The 2024 Plan was approved at our annual general meeting of shareholders held in August 2024. The purpose of the 2024 Plan is to provide incentives to attract, motivate and retain the grantees and align the interests of the grantees with those of our shareholders. It provides for the granting of restricted share units, stock options, restricted shares and share appreciation rights to any participant who is an employee, an executive director or a service provider.\nAs of April 1, 2025 and September 30, 2025, the number of share-based awards available for grant under the scheme mandate was 473,488,465 Shares (equivalent to approximately 59,186,058 ADSs) and 411,246,946 Shares (equivalent to approximately 51,405,868 ADSs), respectively.\nAs of April 1, 2025 and September 30, 2025, the number of share-based awards available for grant under the service provider sub-limit was 93,716,369 Shares (equivalent to 11,714,546 ADSs).\nDetails of movements of RSUs under the 2024 Plan during the Reporting Period are as follows:\n\n[tables/table-15 RSU Activity Report.html]\n\nNotes:\n1. The purchase price of all RSUs granted is nil.\n2. One Share is issuable upon the vesting of each RSU.\n3. No RSUs were granted under the 2024 Plan to directors during the Reporting Period. As at September 30, 2025, no directors were holding any outstanding awards under the 2024 Plan.\n4. The RSUs granted to employee participants vest over a period of up to 6 years.\n5. For employee participants, the weighted average closing price of the Shares traded on the Hong Kong Stock Exchange immediately before the dates on which the RSUs were vested during the Reporting Period was HK\\$113.70.\n6. The following grants of RSUs were made to employee participants during the Reporting Period. These RSUs vest over up to 6 years without any performance target, except that certain RSUs granted to selected participants on May 23, 2025 are subject to satisfaction of certain performance targets based on the financial or operational indicators of the relevant segments as determined by the administrator of the 2024 Plan.\n\n[tables/table-16 RSU Grant Details.html]\n\n7. In accordance with the accounting standards and policies adopted for preparing our Company's financial statements, the fair values of RSUs granted during the Reporting Period were determined with reference to the fair values of the underlying Shares on grant date.\n8. No RSUs granted were cancelled during the Reporting Period.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS/The 2024 Plan", + "metadata": { + "length": 2583, + "summary": "The 2024 Plan, approved in August 2024, aims to attract, motivate, and retain employees, executive directors, and service providers by granting restricted share units (RSUs), stock options, restricted shares, and share appreciation rights. As of April 1, 2025, 473,488,465 shares were available for grant under the scheme mandate, and 93,716,369 shares under the service provider sub-limit. By September 30, 2025, available shares under the scheme mandate decreased to 411,246,946. RSUs granted have a nil purchase price, vest over up to 6 years, and one share is issuable per RSU. No RSUs were granted to directors during the reporting period. The weighted average closing price before vesting was HK$113.70. Grants to employee participants on various dates are detailed in a table, with some RSUs granted on May 23, 2025 subject to performance targets based on financial or operational indicators. Fair values of RSUs were determined by reference to underlying share fair values on grant date. No RSUs were cancelled during the period.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "2024", + "Plan", + "approved", + "annual", + "general", + "meeting", + "shareholders", + "held", + "August", + "purpose", + "provide", + "incentives", + "attract", + "motivate", + "retain", + "grantees", + "align", + "interests", + "It", + "granting", + "restricted", + "share", + "units", + "stock", + "options", + "shares", + "appreciation", + "rights", + "participant", + "employee", + "executive", + "director", + "service", + "provider", + "As", + "April", + "2025", + "September", + "30", + "number", + "based", + "awards", + "grant", + "scheme", + "mandate", + "473", + "488", + "465", + "Shares", + "equivalent", + "approximately", + "59", + "186", + "058", + "ADSs", + "411", + "246", + "946", + "51", + "405", + "868", + "limit", + "93", + "716", + "369", + "11", + "714", + "546", + "Details", + "movements", + "RSUs", + "Reporting", + "Period", + "Notes", + "purchase", + "price", + "granted", + "nil", + "One", + "Share", + "issuable", + "vesting", + "RSU", + "No", + "directors", + "holding", + "outstanding", + "participants", + "vest", + "period", + "years", + "For", + "weighted", + "average", + "closing", + "traded", + "Hong", + "Kong", + "Stock", + "Exchange", + "immediately", + "dates", + "vested", + "HK", + "113.70", + "grants", + "made", + "These", + "performance", + "target", + "selected", + "May", + "23", + "subject", + "satisfaction", + "targets", + "financial", + "operational", + "indicators", + "relevant", + "segments", + "determined", + "administrator", + "In", + "accordance", + "accounting", + "standards", + "policies", + "adopted", + "preparing", + "Company", + "statements", + "fair", + "values", + "reference", + "underlying", + "date", + "cancelled" + ], + "keywords": [ + "RSUs", + "2024 Plan", + "Share Awards" + ], + "connect_to": [ + { + "target": "354b9644-5d8b-5d56-a5ca-8679ec93219f", + "relation": "embeds", + "ref": "[tables/table-15 RSU Activity Report.html]", + "position": { + "start": 1026, + "end": 1068 + } + }, + { + "target": "9cad7312-407b-5ad0-bc77-472a9ee56bfc", + "relation": "embeds", + "ref": "[tables/table-16 RSU Grant Details.html]", + "position": { + "start": 2091, + "end": 2131 + } + } + ] + } + }, + { + "chunk_id": "e6ee7abc-d522-51a0-881e-6df36d050dc0", + "type": "text", + "content": "The 2024 Plan (Existing Shares) was approved by our board in August 2024. The purpose of the 2024 Plan (Existing Shares) is to provide incentives to attract, motivate and retain the grantees and align the interests of the grantees with those of our shareholders. It provides for the granting of restricted share units, stock options, restricted shares and share appreciation rights to any participant who is an employee, a director or a service provider. The 2024 Plan (Existing Shares) is funded by existing Shares, therefore it does not constitute a scheme involving the new issuance of Shares as referred to in Chapter 17 of the Hong Kong Listing Rules but is subject to the applicable disclosure requirements under Rule 17.12 of the Hong Kong Listing Rules.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS/The 2024 Plan (Existing Shares)", + "metadata": { + "length": 761, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "2024", + "Plan", + "Existing", + "Shares", + "approved", + "board", + "August", + "purpose", + "provide", + "incentives", + "attract", + "motivate", + "retain", + "grantees", + "align", + "interests", + "shareholders", + "It", + "granting", + "restricted", + "share", + "units", + "stock", + "options", + "shares", + "appreciation", + "rights", + "participant", + "employee", + "director", + "service", + "provider", + "funded", + "existing", + "constitute", + "scheme", + "involving", + "issuance", + "referred", + "Chapter", + "17", + "Hong", + "Kong", + "Listing", + "Rules", + "subject", + "applicable", + "disclosure", + "requirements", + "Rule", + "17.12" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d4bbd080-2225-5e9c-9794-ad3353e3c8fe", + "type": "table", + "content": "
Month of repurchaseNumber of Shares underlying ADSs repurchased (1)Highest price paid (US$)Lowest price paid (US$)Aggregate consideration paid (US$, in millions)
April 202530,630,48016.7611.97427
May 202513,559,32016.7914.02210
June 202511,685,19215.0013.92168
July 202511,019,16015.0012.96154
August 20255,839,33615.0014.5287
Total72,733,4881,046
", + "path": "tables/table-17 Share Repurchase.html", + "metadata": { + "length": 734, + "summary": "table-18\nThe table shows monthly share repurchases from April to August 2025, totaling 72,733,488 shares for $1,046 million, with prices ranging from $11.97 to $16.79.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-17 Share Repurchase.html", + "keywords": [ + "Share repurchase", + "ADSs", + "Monthly data" + ], + "tokens": [] + } + }, + { + "chunk_id": "88e93b21-7256-53d9-a05a-f6863bcb612c", + "type": "text", + "content": "During the Reporting Period, our Company repurchased a total of 73 million Shares on the NYSE for an aggregate consideration of US\\$1.0 billion.\nDetails of the Shares repurchased on the NYSE are as follows:\n\n[tables/table-17 Share Repurchase.html]\n\nNote:\n1. Each ADS represents eight Shares.\nAs of the date of this Interim Report, all the Shares repurchased during the Reporting Period have been cancelled.\nSave as disclosed above, neither our Company nor any of its subsidiaries purchased, sold or redeemed any of our Company's securities listed on the Hong Kong Stock Exchange or the NYSE (including sale of treasury shares) during the Reporting Period. As of September 30, 2025, our Company did not hold any treasury shares as defined in the Hong Kong Listing Rules.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES", + "metadata": { + "length": 833, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "During", + "Reporting", + "Period", + "Company", + "repurchased", + "total", + "73", + "million", + "Shares", + "NYSE", + "aggregate", + "consideration", + "US", + "1.0", + "billion", + "Details", + "Note", + "Each", + "ADS", + "represents", + "As", + "date", + "Interim", + "Report", + "cancelled", + "Save", + "disclosed", + "subsidiaries", + "purchased", + "sold", + "redeemed", + "securities", + "listed", + "Hong", + "Kong", + "Stock", + "Exchange", + "including", + "sale", + "treasury", + "shares", + "September", + "30", + "2025", + "hold", + "defined", + "Listing", + "Rules" + ], + "keywords": [], + "connect_to": [ + { + "target": "d4bbd080-2225-5e9c-9794-ad3353e3c8fe", + "relation": "embeds", + "ref": "[tables/table-17 Share Repurchase.html]", + "position": { + "start": 208, + "end": 247 + } + } + ] + } + }, + { + "chunk_id": "386fa6ba-552b-5613-be13-09afca6ad8bf", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "89ad1d88-5a52-5029-bddb-4c04e8451dfe", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Corporate Governance", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8d970e60-8767-5d79-89a0-0adc2b675c6e", + "type": "text", + "content": "To the knowledge of the Company and our directors, we have complied with all applicable code provisions set out in the Corporate Governance Code as set forth in Part 2 of Appendix C1 to the Hong Kong Listing Rules for the six months ended September 30, 2025.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Corporate Governance/Compliance with the Corporate Governance Code", + "metadata": { + "length": 258, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "To", + "knowledge", + "Company", + "directors", + "complied", + "applicable", + "code", + "provisions", + "set", + "Corporate", + "Governance", + "Code", + "Part", + "Appendix", + "C1", + "Hong", + "Kong", + "Listing", + "Rules", + "months", + "ended", + "September", + "30", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b31ab0ef-8d72-5c31-b3fd-7c7b7905cc5e", + "type": "text", + "content": "We have adopted our own trading guidelines, on terms no less exacting than the required standard set out in the Model Code as set forth in Appendix C3 to the Hong Kong Listing Rules, to regulate, among others, all dealings by directors and relevant employees of securities in the Company.\nHaving made specific enquiry of all directors, all directors confirmed that they have complied with our trading guidelines during the six months ended September 30, 2025.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Corporate Governance/Compliance with the Model Code", + "metadata": { + "length": 459, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "We", + "adopted", + "trading", + "guidelines", + "terms", + "exacting", + "required", + "standard", + "set", + "Model", + "Code", + "Appendix", + "C3", + "Hong", + "Kong", + "Listing", + "Rules", + "regulate", + "dealings", + "directors", + "relevant", + "employees", + "securities", + "Company", + "Having", + "made", + "specific", + "enquiry", + "confirmed", + "complied", + "months", + "ended", + "September", + "30", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f2870f99-7335-5811-99ad-73d09ad75b2a", + "type": "text", + "content": "The audit committee oversees our accounting and financial reporting processes and the audits of our financial statements. Our audit committee is responsible for, among other things:\n- selecting, and evaluating the qualifications, performance and independence of the independent auditor;\n- pre-approving or, as permitted, approving auditing and non-auditing services permitted to be performed by the independent auditor;\n• assessing the adequacy of our internal accounting controls and audit procedures;\n- reviewing with the independent auditor any audit problems or difficulties and management's response;\n- reviewing and approving related party transactions (as defined in Form 20-F) and connected transactions (as required under the Hong Kong Listing Rules);\n- reviewing and discussing the quarterly reports, half-year reports and annual reports with management and the independent auditor;\n- establishing procedures for the receipt, retention and treatment of complaints received from our employees regarding accounting, internal accounting controls or auditing matters and the confidential, anonymous submission by our employees of concerns regarding questionable accounting or auditing matters;\n- meeting separately, periodically, with management, internal auditors and the independent auditor; and\n• reporting regularly to the full board of directors.\nOur audit committee has reviewed our unaudited condensed consolidated interim financial statements for the six months ended September 30, 2025, and has met with the independent auditor of the Company,\nPricewaterhouseCoopers. Our audit committee has also discussed the accounting policies and practices adopted by us, as well as internal control and financial reporting matters.\nOur unaudited condensed consolidated interim financial statements for the six months ended September 30, 2025 have been reviewed by the independent auditor of the Company, PricewaterhouseCoopers, in accordance with International Standard on Review Engagements 2410 “Review of Interim Financial Information Performed by the Independent Auditor of the Entity”.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Corporate Governance/Audit Committee Review", + "metadata": { + "length": 2094, + "summary": "The audit committee oversees accounting, financial reporting, and audits. Its responsibilities include selecting and evaluating the independent auditor, pre-approving audit and non-audit services, assessing internal controls, reviewing audit problems and related party transactions, and discussing quarterly, half-year, and annual reports. It also establishes procedures for employee complaints on accounting matters and meets separately with management and auditors. The committee reviewed the unaudited interim financial statements for the six months ended September 30, 2025, with PricewaterhouseCoopers, which also reviewed the statements under International Standard on Review Engagements 2410.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "audit", + "committee", + "oversees", + "accounting", + "financial", + "reporting", + "processes", + "audits", + "statements", + "Our", + "responsible", + "things", + "selecting", + "evaluating", + "qualifications", + "performance", + "independence", + "independent", + "auditor", + "pre", + "approving", + "permitted", + "auditing", + "services", + "performed", + "assessing", + "adequacy", + "internal", + "controls", + "procedures", + "reviewing", + "problems", + "difficulties", + "management", + "response", + "related", + "party", + "transactions", + "defined", + "Form", + "20", + "connected", + "required", + "Hong", + "Kong", + "Listing", + "Rules", + "discussing", + "quarterly", + "reports", + "half", + "year", + "annual", + "establishing", + "receipt", + "retention", + "treatment", + "complaints", + "received", + "employees", + "matters", + "confidential", + "anonymous", + "submission", + "concerns", + "questionable", + "meeting", + "separately", + "periodically", + "auditors", + "regularly", + "full", + "board", + "directors", + "reviewed", + "unaudited", + "condensed", + "consolidated", + "interim", + "months", + "ended", + "September", + "30", + "2025", + "met", + "Company", + "PricewaterhouseCoopers", + "discussed", + "policies", + "practices", + "adopted", + "control", + "accordance", + "International", + "Standard", + "Review", + "Engagements", + "2410", + "Interim", + "Financial", + "Information", + "Performed", + "Independent", + "Auditor", + "Entity" + ], + "keywords": [ + "audit committee", + "financial oversight", + "internal controls" + ], + "connect_to": [] + } + }, + { + "chunk_id": "d5ca1f33-d7a1-5b6c-85a1-c88f5c2ad303", + "type": "text", + "content": "Our board did not recommend the distribution of an interim dividend for the Reporting Period.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Interim Dividend", + "metadata": { + "length": 93, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Our", + "board", + "recommend", + "distribution", + "interim", + "dividend", + "Reporting", + "Period" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "87c76147-9c67-5b79-a77e-d3430c6a4d6d", + "type": "table", + "content": "
General Corporate purposesNet proceeds from the September 2025 Notes Offering (US$ in millions)Amount of net proceeds utilized during the reporting period (US$ in millions)Actual use of proceeds up to September 30, 2025 (US$ in millions)
Enhancing its cloud infrastructure2,504443443
Expanding international commerce operations626287287
Total3,130730730
", + "path": "tables/table-18 Notes Proceeds Use.html", + "metadata": { + "length": 545, + "summary": "table-19\nThe table shows $3,130M net proceeds from September 2025 Notes Offering, with $730M utilized by September 30, 2025: $443M for cloud infrastructure and $287M for international commerce.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-18 Notes Proceeds Use.html", + "keywords": [ + "Net proceeds", + "Cloud infrastructure", + "Commerce expansion" + ], + "tokens": [] + } + }, + { + "chunk_id": "acc2ea20-a2ee-5a9b-b315-4eaa53fbaffb", + "type": "text", + "content": "September 2025 Notes Offering\nIn September 2025, we completed a private offering (the “September 2025 Notes Offering”) of US\\$3.168 billion aggregate principal amount of Zero Coupon Convertible Senior Notes due 2032 (the “September 2025 Notes”). In connection with the September 2025 Notes Offering, we also entered into capped call transactions (the “Capped Call Transactions”) with certain financial institutions at the costs of US\\$184 million. The cap price of the Capped Call Transactions is initially US\\$235.46 per ADS (subject to certain adjustments), which represents a premium of 60% over the closing price of the ordinary shares on the Hong Kong Stock Exchange on the date of pricing of the Notes Offering of HK\\$143.30, which is converted and multiplied by eight, the ordinary-share-to-ADS ratio to, US\\$147.16 (the “Reference Share Price”). The September 2025 Notes have been offered to certain non-U.S. persons in offshore transaction in reliance on Regulation S under the U.S. Securities Act.\nThe Company intended to use the net proceeds from the September 2025 Notes Offering for general corporate purposes, with a strategic focus on strengthening its cloud infrastructure capabilities and international commerce business operations. Specifically, the allocation will include approximately 80% directed towards enhancing its cloud infrastructure, which encompasses scaling up data centers, upgrading technology, and improving services to meet growing demand. The remaining 20% will be invested in expanding international commerce operations, focusing on operational investments that will enable the Company to enhance its market presence and efficiency.\nThe September 2025 Notes will mature on September 15, 2032, unless earlier redeemed, repurchased or converted in accordance with their terms prior to such date. Prior to the close of business on the business day immediately preceding March 15, 2032, holders will have the right to convert all or any portion of their Notes, in multiples of US\\$1,000 principal amount, only upon satisfaction of certain conditions. On or after March 15, 2032, until the close of business on the third scheduled trading day immediately preceding the maturity date, holders may convert all or any portion of their Notes, in multiples of US\\$1,000 principal amount, at their option at any time irrespective of the foregoing conditions. Upon conversion, we retain flexibility to control any potential dilution with full discretion to settle the conversions in cash or ADSs, or a combination of both. Holders may also elect to receive ordinary shares in lieu of any ADSs upon conversion.\nThe initial conversion rate for the Notes is 5.1773 ADSs per US\\$1,000 principal amount of the September 2025 Notes, which is equivalent to an initial conversion price of approximately US\\$193.15 per ADS (or HK\\$188.08 per ordinary share, as converted and divided by eight, the ordinary-share-to-ADS ratio). The initial conversion price represents a premium of approximately 31.25% over the Reference Share Price, and the conversion rate will be subject to adjustment in some events, including but not limited to a share split, a share combination, cash distributions, certain issuances of rights, options or warrants, distribution of capital stock, certain tender or exchange offers, and certain other dilutive events.\nThe net price of each conversion share, based on the estimated net proceeds of approximately US\\$3.13 billion and 16,401,686 conversion shares (in the form of ADSs) that are issuable under the September 2025 Notes (excluding adjustment and make-whole adjustments), is expected to be approximately US\\$190.60. Upon exercise in full of the conversion rights attached to the September 2025 Notes at the initial conversion price of US\\$193.15 per ADS (equivalent to HK\\$188.08 per ordinary share, as converted and divided by eight, the ordinary-share-to-ADS ratio), a total of approximately 16,401,686 ADSs (or 131,213,488 ordinary shares) will be issued.\nFor further details of the September 2025 Notes Offering, please see the announcements of our Company dated September 11, 2025 and September 17, 2025.\nThe net proceeds raised from the September 2025 Notes Offering were used according to the intentions as disclosed in the announcement of our Company dated September 11, 2025. As of September 30, 2025, the unutilized proceeds amounted to US\\$2,400 million. A summary of the utilization of the net proceeds from the September 2025 Notes Offering as of September 30, 2025 is as follows:\n\n[tables/table-18 Notes Proceeds Use.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Use of Proceeds from the Notes Offerings", + "metadata": { + "length": 4646, + "summary": "In September 2025, the company completed a private offering of $3.168 billion in Zero Coupon Convertible Senior Notes due 2032, along with capped call transactions costing $184 million. The notes have an initial conversion rate of 5.1773 ADSs per $1,000 principal, equivalent to an initial conversion price of about $193.15 per ADS. Net proceeds of approximately $3.13 billion are intended for general corporate purposes, with 80% allocated to enhancing cloud infrastructure and 20% to expanding international commerce operations. As of September 30, 2025, unutilized proceeds totaled $2.4 billion.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "September", + "2025", + "Notes", + "Offering", + "In", + "completed", + "private", + "offering", + "US", + "3.168", + "billion", + "aggregate", + "principal", + "amount", + "Zero", + "Coupon", + "Convertible", + "Senior", + "due", + "2032", + "connection", + "entered", + "capped", + "call", + "transactions", + "Capped", + "Call", + "Transactions", + "financial", + "institutions", + "costs", + "184", + "million", + "The", + "cap", + "price", + "initially", + "235.46", + "ADS", + "subject", + "adjustments", + "represents", + "premium", + "60%", + "closing", + "ordinary", + "shares", + "Hong", + "Kong", + "Stock", + "Exchange", + "date", + "pricing", + "HK", + "143.30", + "converted", + "multiplied", + "share", + "ratio", + "147.16", + "Reference", + "Share", + "Price", + "offered", + "persons", + "offshore", + "transaction", + "reliance", + "Regulation", + "Securities", + "Act", + "Company", + "intended", + "net", + "proceeds", + "general", + "corporate", + "purposes", + "strategic", + "focus", + "strengthening", + "cloud", + "infrastructure", + "capabilities", + "international", + "commerce", + "business", + "operations", + "Specifically", + "allocation", + "include", + "approximately", + "80%", + "directed", + "enhancing", + "encompasses", + "scaling", + "data", + "centers", + "upgrading", + "technology", + "improving", + "services", + "meet", + "growing", + "demand", + "remaining", + "20%", + "invested", + "expanding", + "focusing", + "operational", + "investments", + "enable", + "enhance", + "market", + "presence", + "efficiency", + "mature", + "15", + "earlier", + "redeemed", + "repurchased", + "accordance", + "terms", + "prior", + "Prior", + "close", + "day", + "immediately", + "preceding", + "March", + "holders", + "convert", + "portion", + "multiples", + "000", + "satisfaction", + "conditions", + "On", + "scheduled", + "trading", + "maturity", + "option", + "time", + "irrespective", + "foregoing", + "Upon", + "conversion", + "retain", + "flexibility", + "control", + "potential", + "dilution", + "full", + "discretion", + "settle", + "conversions", + "cash", + "ADSs", + "combination", + "Holders", + "elect", + "receive", + "lieu", + "initial", + "rate", + "5.1773", + "equivalent", + "193.15", + "188.08", + "divided", + "31.25%", + "adjustment", + "events", + "including", + "limited", + "split", + "distributions", + "issuances", + "rights", + "options", + "warrants", + "distribution", + "capital", + "stock", + "tender", + "exchange", + "offers", + "dilutive", + "based", + "estimated", + "3.13", + "16", + "401", + "686", + "form", + "issuable", + "excluding", + "make", + "expected", + "190.60", + "exercise", + "attached", + "total", + "131", + "213", + "488", + "issued", + "For", + "details", + "announcements", + "dated", + "11", + "17", + "raised", + "intentions", + "disclosed", + "announcement", + "As", + "30", + "unutilized", + "amounted", + "400", + "summary", + "utilization" + ], + "keywords": [ + "Convertible Notes", + "Cloud Infrastructure", + "Capital Raising" + ], + "connect_to": [ + { + "target": "87c76147-9c67-5b79-a77e-d3430c6a4d6d", + "relation": "embeds", + "ref": "[tables/table-18 Notes Proceeds Use.html]", + "position": { + "start": 4543, + "end": 4584 + } + } + ] + } + }, + { + "chunk_id": "33a7dcb1-935d-5f20-9810-518072efabe7", + "type": "text", + "content": "Conventions that apply to this Interim Report\nUnless the context otherwise requires, references in this interim report to:\n- “2014 Plan” is to the 2014 Post-IPO Equity Incentive Plan, which we adopted on September 2, 2014 and amended and restated on February 12, 2020 and May 25, 2022;\n- “2024 Plan” is to the 2024 Equity Incentive Plan, which we adopted on August 22, 2024;\n- “2024 Plan (Existing Shares)” is to the 2024 Equity Incentive Plan (Existing Shares), which we adopted on August 26, 2024;\n- “ADS(s)” are to the American depositary share(s), each of which represents eight Shares;\n- “Alibaba,” “Alibaba Group,” “the Group,” “Company,” “our company,” “we,” “our” or “us” are to Alibaba Group Holding Limited, a company incorporated in the Cayman Islands with limited liability on June 28, 1999 and, where the context requires, its consolidated subsidiaries and its affiliated consolidated entities, including its variable interest entities and their subsidiaries, from time to time;\n- “Articles” or “Articles of Association” is to our Amended and Restated Articles of Association (as amended and restated from time to time), adopted on September 30, 2020 and amended on August 22, 2024;\n- “board” or “board of directors” is to our board of directors, unless otherwise stated;\n- “Corporate Governance Code” is to the Corporate Governance Code set out in Appendix C1 to the Hong Kong Listing Rules;\n- “director(s)” are to member(s) of our board, unless otherwise stated;\n- “Hong Kong Listing Rules” are to the Rules Governing the Listing of Securities on The Stock Exchange of Hong Kong Limited, as amended or supplemented from time to time;\n- “Hong Kong Stock Exchange” is to The Stock Exchange of Hong Kong Limited;\n- “Model Code” is to the Model Code for Securities Transactions by Directors of Listed Issuers set out in Appendix C3 to the Hong Kong Listing Rules;\n- “NYSE” is to the New York Stock Exchange;\n- “Reporting Period” is to the six months ended September 30, 2025;\n- “RSU(s)” are to restricted share unit(s);\n- “SEC” is to the United States Securities and Exchange Commission;\n- “SFO” is to the Securities and Futures Ordinance (Chapter 571 of the Laws of Hong Kong), as amended or supplemented from time to time;\n- “Share Split” is to the subdivision of each ordinary share into eight Shares, pursuant to which the par value of our Shares was correspondingly changed from US\\$0.000025 per Share to US\\$0.000003125 per Share, with effect from July 30, 2019; immediately after the Share Split became effective, our authorized share capital became US\\$100,000 divided into 32,000,000,000 Shares of par value US\\$0.000003125 per Share;\n- “shareholder(s)” are to holder(s) of Shares and, where the context requires, ADSs;\n- “Share(s)” or “ordinary share(s)” are to ordinary share(s) in our capital with par value of US\\$0.000003125 each;\n- “substantial shareholder(s)” has the meaning ascribed to it under the Hong Kong Listing Rules;\n- “U.S.” or “United States” is to the United States of America, its territories, its possessions and all areas subject to its jurisdiction;\n- “US\\$” or “U.S. dollars” are to the lawful currency of the United States;\n- “U.S. Exchange Act” is to the United States Securities Exchange Act of 1934, as amended, and the rules and regulations promulgated thereunder;\n- “U.S. Securities Act” is to the United States Securities Act of 1933, as amended, and the rules and regulations promulgated thereunder.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DEFINITIONS Conventions that apply to this Interim Report", + "metadata": { + "length": 3451, + "summary": "This section defines key terms and conventions used in the interim report, including references to equity incentive plans (2014 Plan, 2024 Plan, 2024 Plan Existing Shares), ADSs, Alibaba Group, Articles of Association, board of directors, Corporate Governance Code, Hong Kong Listing Rules, Model Code, NYSE, Reporting Period, RSUs, SEC, SFO, Share Split, shareholders, Shares, substantial shareholders, U.S., U.S. dollars, U.S. Exchange Act, and U.S. Securities Act.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Conventions", + "apply", + "Interim", + "Report", + "Unless", + "context", + "requires", + "references", + "interim", + "report", + "2014", + "Plan", + "Post", + "IPO", + "Equity", + "Incentive", + "adopted", + "September", + "amended", + "restated", + "February", + "12", + "2020", + "May", + "25", + "2022", + "2024", + "August", + "22", + "Existing", + "Shares", + "26", + "ADS", + "American", + "depositary", + "share", + "represents", + "Alibaba", + "Group", + "Company", + "company", + "Holding", + "Limited", + "incorporated", + "Cayman", + "Islands", + "limited", + "liability", + "June", + "28", + "1999", + "consolidated", + "subsidiaries", + "affiliated", + "entities", + "including", + "variable", + "interest", + "time", + "Articles", + "Association", + "Amended", + "Restated", + "30", + "board", + "directors", + "stated", + "Corporate", + "Governance", + "Code", + "set", + "Appendix", + "C1", + "Hong", + "Kong", + "Listing", + "Rules", + "director", + "member", + "Governing", + "Securities", + "The", + "Stock", + "Exchange", + "supplemented", + "Model", + "Transactions", + "Directors", + "Listed", + "Issuers", + "C3", + "NYSE", + "New", + "York", + "Reporting", + "Period", + "months", + "ended", + "2025", + "RSU", + "restricted", + "unit", + "SEC", + "United", + "States", + "Commission", + "SFO", + "Futures", + "Ordinance", + "Chapter", + "571", + "Laws", + "Share", + "Split", + "subdivision", + "ordinary", + "pursuant", + "par", + "correspondingly", + "changed", + "US", + "0.000025", + "0.000003125", + "effect", + "July", + "2019", + "immediately", + "effective", + "authorized", + "capital", + "100", + "000", + "divided", + "32", + "shareholder", + "holder", + "ADSs", + "substantial", + "meaning", + "ascribed", + "America", + "territories", + "possessions", + "areas", + "subject", + "jurisdiction", + "dollars", + "lawful", + "currency", + "Act", + "1934", + "rules", + "regulations", + "promulgated", + "thereunder", + "1933" + ], + "keywords": [ + "definitions", + "terminology", + "glossary" + ], + "connect_to": [] + } + }, + { + "chunk_id": "4758cb97-9722-5894-88b7-214ccb7cb5ba", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cb9c96f5-c33e-5385-8c33-17acf3ab221a", + "type": "table", + "content": "
Page
Report on Review of Interim Financial Information for the six-months ended September 30, 202545
Unaudited Condensed Consolidated Income Statements for the six months ended September 30, 2024 and 202546
Unaudited Condensed Consolidated Statements of Comprehensive Income for the six months ended September 30, 2024 and 202547
Unaudited Condensed Consolidated Balance Sheets as of March 31, 2025 and September 30, 202548
Unaudited Condensed Consolidated Statements of Changes in Shareholders’ Equity for the six months ended September 30, 2024 and 202550
Unaudited Condensed Consolidated Statements of Cash Flows for the six months ended September 30, 2024 and 202552
Notes to Unaudited Condensed Consolidated Financial Statements53
", + "path": "tables/table-19 Interim Report.html", + "metadata": { + "length": 959, + "summary": "table-20\nTable of contents for an interim financial report for the six months ended September 30, 2025, listing unaudited financial statements and notes starting on page 45.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-19 Interim Report.html", + "keywords": [ + "Financial Statements", + "Interim Report", + "2025" + ], + "tokens": [] + } + }, + { + "chunk_id": "84d877c2-cdfc-5b18-8ac9-5fed9f8befdf", + "type": "text", + "content": "[tables/table-19 Interim Report.html]\n\nReport on Review of Interim Financial Information To the Board of Directors of Alibaba Group Holding Limited (incorporated in the Cayman Islands with limited liability)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED/INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS", + "metadata": { + "length": 197, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Report", + "Review", + "Interim", + "Financial", + "Information", + "To", + "Board", + "Directors", + "Alibaba", + "Group", + "Holding", + "Limited", + "incorporated", + "Cayman", + "Islands", + "limited", + "liability" + ], + "keywords": [], + "connect_to": [ + { + "target": "cb9c96f5-c33e-5385-8c33-17acf3ab221a", + "relation": "embeds", + "ref": "[tables/table-19 Interim Report.html]", + "position": { + "start": 0, + "end": 37 + } + } + ] + } + }, + { + "chunk_id": "3a566ab5-5827-5994-82bb-8049fa9214f8", + "type": "text", + "content": "We have reviewed the interim financial information set out on pages 46 to 81, which comprises the condensed consolidated balance sheet of Alibaba Group Holding Limited (the “Company”) and its subsidiaries (together, the “Group”) as at September 30, 2025 and the condensed consolidated income statement, the condensed consolidated statement of comprehensive income, the condensed consolidated statement of changes in shareholders’ equity and the condensed consolidated statement of cash flows for the six-month period then ended, and selected explanatory notes. The Rules Governing the Listing of Securities on The Stock Exchange of Hong Kong Limited require the preparation of a report on interim financial information to be in compliance with the relevant provisions thereof and accounting principles generally accepted in the United States of America (“U.S. GAAP”). The directors of the Company are responsible for the preparation and presentation of this interim financial information in accordance with U.S. GAAP. Our responsibility is to express a conclusion on this interim financial information based on our review and to report our conclusion solely to you, as a body, in accordance with our agreed terms of engagement, and for no other purpose. We do not assume responsibility towards or accept liability to any other person for the contents of this report.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED/INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS/Introduction", + "metadata": { + "length": 1366, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "We", + "reviewed", + "interim", + "financial", + "information", + "set", + "pages", + "46", + "81", + "comprises", + "condensed", + "consolidated", + "balance", + "sheet", + "Alibaba", + "Group", + "Holding", + "Limited", + "Company", + "subsidiaries", + "September", + "30", + "2025", + "income", + "statement", + "comprehensive", + "shareholders", + "equity", + "cash", + "flows", + "month", + "period", + "ended", + "selected", + "explanatory", + "notes", + "The", + "Rules", + "Governing", + "Listing", + "Securities", + "Stock", + "Exchange", + "Hong", + "Kong", + "require", + "preparation", + "report", + "compliance", + "relevant", + "provisions", + "thereof", + "accounting", + "principles", + "generally", + "accepted", + "United", + "States", + "America", + "GAAP", + "directors", + "responsible", + "presentation", + "accordance", + "Our", + "responsibility", + "express", + "conclusion", + "based", + "review", + "solely", + "body", + "agreed", + "terms", + "engagement", + "purpose", + "assume", + "accept", + "liability", + "person", + "contents" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b9676ef9-177d-54db-b5ac-b831d2a9e634", + "type": "text", + "content": "We conducted our review in accordance with International Standard on Review Engagements 2410, “Review of Interim Financial Information Performed by the Independent Auditor of the Entity”. A review of interim financial information consists of making inquiries, primarily of persons responsible for financial and accounting matters, and applying analytical and other review procedures. A review is substantially less in scope than an audit conducted in accordance with International Standards on Auditing and consequently does not enable us to obtain assurance that we would become aware of all significant matters that might be identified in an audit. Accordingly, we do not express an audit opinion.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED/INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS/Scope of Review", + "metadata": { + "length": 699, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "We", + "conducted", + "review", + "accordance", + "International", + "Standard", + "Review", + "Engagements", + "2410", + "Interim", + "Financial", + "Information", + "Performed", + "Independent", + "Auditor", + "Entity", + "interim", + "financial", + "information", + "consists", + "making", + "inquiries", + "primarily", + "persons", + "responsible", + "accounting", + "matters", + "applying", + "analytical", + "procedures", + "substantially", + "scope", + "audit", + "Standards", + "Auditing", + "enable", + "obtain", + "assurance", + "aware", + "significant", + "identified", + "Accordingly", + "express", + "opinion" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "025455e0-0711-5ac3-a008-b7537466da15", + "type": "text", + "content": "Based on our review, nothing has come to our attention that causes us to believe that the interim financial information of the Group is not prepared, in all material respects, in accordance with U.S. GAAP.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED/INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS/Conclusion", + "metadata": { + "length": 205, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Based", + "review", + "attention", + "interim", + "financial", + "information", + "Group", + "prepared", + "material", + "respects", + "accordance", + "GAAP" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "20b56326-11ff-560d-991f-19b9c9961eb2", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMBUS$ (Note 2(a))
(in millions, except per share data)
Notes
Revenue4,16479,739495,44769,595
Cost of revenue16(290,135)(287,210)(40,344)
Product development expenses16(27,555)(32,096)(4,509)
Sales and marketing expenses16(65,167)(119,674)(16,811)
General and administrative expenses16(23,057)(14,778)(2,076)
Amortization and impairment of intangible assets(3,441)(1,633)(229)
Other gains, net85129742
Income from operations71,23540,3535,668
Interest and investment income, net17,12937,4685,263
Interest expense(4,615)(4,995)(702)
Other (expense) income, net16(1,221)1,329187
Income before income tax and share of results of equity method investees82,52874,15510,416
Income tax expenses5(17,442)(14,415)(2,024)
Share of results of equity method investees2,4833,254457
Net income67,56962,9948,849
Net loss (income) attributable to noncontrolling interests854(1,326)(187)
Net income attributable to Alibaba Group Holding Limited68,42361,6688,662
(Accretion) Reversal of accretion of mezzanine equity(280)2,438343
Net income attributable to ordinary shareholders68,14364,1069,005
Earnings per share attributable to ordinary shareholders7
Basic3.583.450.49
Diluted3.503.340.47
Earnings per ADS attributable to ordinary shareholders (one ADS equals eight ordinary shares)7
Basic28.6227.633.88
Diluted28.0026.733.75
Weighted average number of shares used in computing earnings per share (million shares)7
Basic19,04518,562
Diluted19,45919,154
", + "path": "tables/table-20 Alibaba Financials.html", + "metadata": { + "length": 3172, + "summary": "table-21\nAlibaba Group's financial results for six months ended Sep 30, 2024 and 2025, showing revenue of 495,447 million RMB and net income attributable to ordinary shareholders of 64,106 million RMB in 2025.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-20 Alibaba Financials.html", + "keywords": [ + "Revenue", + "Net Income", + "Earnings Per Share" + ], + "tokens": [] + } + }, + { + "chunk_id": "f722fa1a-9184-54f7-b15b-a327ddc38537", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMBUS$ (Note 2(a))
(in millions)
Net income67,56962,9948,849
Other comprehensive income (loss):
- Foreign currency translation:
Change in unrealized losses, net of tax(3,699)(5,081)(714)
- Share of other comprehensive income (loss) of equity method investees:
Change in unrealized gains (losses)350(40)(6)
- Interest rate swaps under hedge accounting and others:
Change in unrealized gains601
Other comprehensive loss(3,289)(5,120)(720)
Total comprehensive income64,28057,8748,129
Total comprehensive loss (income) attributable to noncontrolling interests1,119(273)(38)
Total comprehensive income attributable to ordinary shareholders65,39957,6018,091
", + "path": "tables/table-21 Income Statement.html", + "metadata": { + "length": 1397, + "summary": "table-22\nThe table shows net income of 67,569 RMB (2024) and 62,994 RMB (2025), with total comprehensive income of 64,280 RMB and 57,874 RMB respectively, including foreign currency translation losses and other items.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-21 Income Statement.html", + "keywords": [ + "comprehensive income", + "net income", + "foreign currency" + ], + "tokens": [] + } + }, + { + "chunk_id": "16a35153-a74b-531d-9c83-a89110c2420c", + "type": "table", + "content": "
As of March 31,As of September 30,
20252025
RMBRMBUS$
(in millions)(Note 2(a))
AssetsNotes
Current assets:
Cash and cash equivalents145,487135,06918,973
Short-term investments228,826193,24627,145
Restricted cash and escrow receivables43,78140,3745,671
Equity securities and other investments853,78045,2576,357
Prepayments, receivables and other assets202,175232,67332,684
Total current assets674,049646,61990,830
Equity securities and other investments8356,818411,95357,867
Prepayments, receivables and other assets83,43196,92713,615
Investments in equity method investees210,169206,86229,058
Property and equipment, net203,348246,53934,631
Intangible assets, net1020,91119,4292,729
Goodwill11255,501255,55135,897
Total assets1,804,2271,883,880264,627
Liabilities, mezzanine equity and shareholders' equity
Current liabilities:
Current bank borrowings22,56226,2883,693
Income tax payable11,6385,588785
Accrued expenses, accounts payable and other liabilities12332,537340,76947,868
Merchant deposits27425135
Deferred revenue and customer advances68,33571,24110,007
Total current liabilities435,346444,13762,388
Deferred revenue4,5364,496632
Deferred tax liabilities48,45446,8026,574
Non-current bank borrowings49,90963,5668,929
Non-current unsecured senior notes122,398120,50416,927
Non-current convertible unsecured senior notes1335,83457,4818,074
Non-current exchangeable bonds1413,7551,932
Other liabilities1217,64421,3543,000
Total liabilities714,121772,095108,456
", + "path": "tables/table-22 Balance Sheet Data.html", + "metadata": { + "length": 3350, + "summary": "table-23\nThe table shows assets, liabilities, and equity as of March 31, 2025 and September 30, 2025, with figures in RMB and US dollars. Total assets were 1,804,227 million RMB in March and 1,883,880 million in September.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-22 Balance Sheet Data.html", + "keywords": [ + "Assets", + "Liabilities", + "Financials" + ], + "tokens": [] + } + }, + { + "chunk_id": "c97e9b7d-9775-5f33-b3fe-37a1aefc264f", + "type": "table", + "content": "
As of March 31,As of September 30,
20252025
RMBRMBUS$ (Note 2(a))
(in millions)
Commitments and contingencies
Mezzanine equity11,7139,8841,388
Shareholders’ equity:
Ordinary shares, US$0.000003125 par value; 32,000,000,000 shares authorized as of March 31 and September 30, 2025; 18,474,235,708 and 18,551,943,896 shares issued and outstanding as of March 31 and September 30, 2025 respectively11
Additional paid-in capital381,379387,14754,382
Treasury shares, at cost(36,329)(36,162)(5,080)
Statutory reserves15,93616,2862,288
Accumulated other comprehensive income (loss)
Cumulative translation adjustments3,286(1,670)(234)
Unrealized gains on interest rate swaps and others10710915
Retained earnings645,478666,78493,663
Total shareholders’ equity1,009,8581,032,495145,034
Noncontrolling interests68,53569,4069,749
Total equity1,078,3931,101,901154,783
Total liabilities, mezzanine equity and equity1,804,2271,883,880264,627
", + "path": "tables/table-23 Equity Summary.html", + "metadata": { + "length": 1725, + "summary": "table-24\nThe table shows commitments, mezzanine equity, and shareholders' equity components as of March 31 and September 30, 2025, with RMB and US$ values. Total equity increased from 1,078,393 to 1,101,901 million RMB.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-23 Equity Summary.html", + "keywords": [ + "Shareholders' equity", + "mezzanine equity", + "financial data" + ], + "tokens": [] + } + }, + { + "chunk_id": "b0d44ad2-3684-5ae4-9b17-dea56f9beb54", + "type": "table", + "content": "
Ordinary sharesAccumulated othercomprehensive income (loss)
ShareAmountAdditionalpaid-incapitalRMBTreasurysharesRMBStatutoryreservesRMBCumulativetranslationadjustmentsRMBUnrealizedgains(losses) oninterestrate swapsand othersRMBRetainedearningsRMBTotalshareholders’equityRMBNoncontrollinginterestsRMBTotalequityRMB
(in millions, except per share data)
Balance as ofApril 1, 202419,469,126,9561397,999(27,684)14,7333,635(37)597,897986,544115,3271,101,871
Foreign currencytranslationadjustment,net of tax(3,541)(3,541)(158)(3,699)
Share of additionalpaid-in capitaland othercomprehensiveincome of equitymethod investees(42)33515308308
Change in fair value ofinterest rate swapsunder hedgeaccountingand others606060
Net income for theperiod68,42368,423(961)67,462
Acquisition ofsubsidiaries187187
Issuance of shares,including vesting ofRSUs and earlyexercised optionsand exercise ofshare options178,297,456
Repurchase andretirement ofordinary shares(1,027,554,280)(21,096)(8,501)(42,216)(71,813)(71,813)
Transactions withnoncontrollinginterests2,3572,357(23,852)(21,495)
Amortization ofcompensation cost5,8195,8191,9627,781
Declaration ofdividends(29,340)(29,340)(29,340)
Capped calltransactions(4,612)(4,612)(4,612)
Appropriation tostatutory reserves1,152(1,152)
Others(280)(280)(144)(424)
Balance as ofSeptember 30, 2024(in millions, except per share data)18,619,870,1321380,145(36,185)15,88542938593,612953,92592,3611,046,286
Balance as ofApril 1, 202518,474,235,7081381,379(36,329)15,9363,286107645,4781,009,85868,5351,078,393
Foreign currencytranslationadjustment,net of tax(3,643)7(3,636)(166)(3,802)
Share of additionalpaid-in capitaland othercomprehensiveincome of equitymethod investees(129)(34)(6)(169)(169)
Change in fair value ofinterest rate swapsunder hedgeaccounting and others111
Net investment hedge(1,279)(1,279)(1,279)
Net income for theperiod61,66861,66843962,107
Acquisition ofsubsidiaries22
Deconsolidation ofsubsidiaries(80)(80)
Issuance of shares,including vesting ofRSUs and earlyexercised optionsand exercise ofshare options150,441,6761,0381,0381,038
Repurchase andretirement ofordinary shares(72,733,488)(1,737)167(5,965)(7,535)(7,535)
Transactions withnoncontrollinginterests6868(315)(247)
Amortization ofcompensation cost5,3975,3971,0756,472
Declaration ofdividends(34,047)(34,047)(34,047)
Capped calltransactions(1,309)(1,309)(1,309)
Appropriation tostatutory reserves350(350)
Others2,4402,440(84)2,356
Balance as ofSeptember 30, 202518,551,943,8961387,147(36,162)16,286(1,670)109666,7841,032,49569,4061,101,901
", + "path": "tables/table-24 Equity Changes.html", + "metadata": { + "length": 6682, + "summary": "table-25\nThe table shows changes in shareholders' equity from April to September 2024 and 2025, including share issuances, repurchases, dividends, and net income, with totals in millions of RMB.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-24 Equity Changes.html", + "keywords": [ + "shareholders' equity", + "financial data", + "stock transactions" + ], + "tokens": [] + } + }, + { + "chunk_id": "e5b2d47b-9c5a-543f-a075-2113a27a1252", + "type": "text", + "content": "Hong Kong, December 4, 2025\nALIBABA GROUP HOLDING LIMITED\nUNAUDITED CONDENSED CONSOLIDATED INCOME STATEMENTS\n\n[tables/table-20 Alibaba Financials.html]\n\nThe accompanying notes form an integral part of these unaudited condensed consolidated financial statements.\nALIBABA GROUP HOLDING LIMITED\nUNAUDITED CONDENSED CONSOLIDATED STATEMENTS OF COMPREHENSIVE INCOME\n\n[tables/table-21 Income Statement.html]\n\nThe accompanying notes form an integral part of these unaudited condensed consolidated financial statements.\nALIBABA GROUP HOLDING LIMITED\nUNAUDITED CONDENSED CONSOLIDATED BALANCE SHEETS\n\n[tables/table-22 Balance Sheet Data.html]\n\nThe accompanying notes form an integral part of these unaudited condensed consolidated financial statements.\nALIBABA GROUP HOLDING LIMITED\nUNAUDITED CONDENSED CONSOLIDATED BALANCE SHEETS (CONTINUED)\n\n[tables/table-23 Equity Summary.html]\n\nThe accompanying notes form an integral part of these unaudited condensed consolidated financial statements.\nALIBABA GROUP HOLDING LIMITED\nUNAUDITED CONDENSED CONSOLIDATED STATEMENTS OF CHANGES IN SHAREHOLDERS' EQUITY\n\n[tables/table-24 Equity Changes.html]\n\nThe accompanying notes form an integral part of these unaudited condensed consolidated financial statements.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED/INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS/PricewaterhouseCoopers Certified Public Accountants", + "metadata": { + "length": 1313, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Hong", + "Kong", + "December", + "2025", + "ALIBABA", + "GROUP", + "HOLDING", + "LIMITED", + "UNAUDITED", + "CONDENSED", + "CONSOLIDATED", + "INCOME", + "STATEMENTS", + "The", + "accompanying", + "notes", + "form", + "integral", + "part", + "unaudited", + "condensed", + "consolidated", + "financial", + "statements", + "OF", + "COMPREHENSIVE", + "BALANCE", + "SHEETS", + "CONTINUED", + "CHANGES", + "IN", + "SHAREHOLDERS", + "EQUITY" + ], + "keywords": [], + "connect_to": [ + { + "target": "20b56326-11ff-560d-991f-19b9c9961eb2", + "relation": "embeds", + "ref": "[tables/table-20 Alibaba Financials.html]", + "position": { + "start": 110, + "end": 151 + } + }, + { + "target": "f722fa1a-9184-54f7-b15b-a327ddc38537", + "relation": "embeds", + "ref": "[tables/table-21 Income Statement.html]", + "position": { + "start": 361, + "end": 400 + } + }, + { + "target": "16a35153-a74b-531d-9c83-a89110c2420c", + "relation": "embeds", + "ref": "[tables/table-22 Balance Sheet Data.html]", + "position": { + "start": 590, + "end": 631 + } + }, + { + "target": "c97e9b7d-9775-5f33-b3fe-37a1aefc264f", + "relation": "embeds", + "ref": "[tables/table-23 Equity Summary.html]", + "position": { + "start": 833, + "end": 870 + } + }, + { + "target": "b0d44ad2-3684-5ae4-9b17-dea56f9beb54", + "relation": "embeds", + "ref": "[tables/table-24 Equity Changes.html]", + "position": { + "start": 1091, + "end": 1128 + } + } + ] + } + }, + { + "chunk_id": "a94bfa0c-3ded-56f1-ae92-af805378926f", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_2", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "16ede654-c979-5091-8964-4d3c4be26524", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)(Note 2(a))
Cash flows from operating activities:65,07430,7714,322
Cash flows from investing activities:
Decrease in short-term investments, net105,47012,9721,822
Increase in other treasury investments, net(113,387)(24,091)(3,384)
Settlement of forward exchange contracts, net311(111)(16)
Acquisitions of equity and debt securities, and others(3,905)(18,690)(2,625)
Disposals of equity and debt securities, and others5,77235,3904,971
Acquisitions of equity method investees(1,964)(1,179)(165)
Disposals of and distributions from equity method investees7372,426341
Acquisitions of land use rights, property and equipment(29,585)(70,177)(9,858)
Disposals of property and equipment1,47523934
Cash received (paid) for business combinations, net of cash acquired62(62)(9)
Deconsolidation and disposal of subsidiaries, net of cash proceeds11,8251,661
Loans to employees, net of repayments14913419
Net cash used in investing activities(34,865)(51,324)(7,209)
Cash flows from financing activities:
Issuance of ordinary shares11,038146
Repurchase of ordinary shares(72,889)(7,638)(1,073)
Dividend distribution(29,022)(33,621)(4,723)
Proceeds from convertible unsecured senior notes, net of debt issuance cost35,67722,3033,133
Payments for capped call transactions(4,612)(1,309)(184)
Acquisition of additional equity interests in non-wholly owned subsidiaries(19,947)(1,155)(162)
Dividends paid by non-wholly owned subsidiaries to noncontrolling interests(116)(306)(43)
Contingent consideration payments made after a business combination(149)(81)(11)
Capital injection from noncontrolling interests2,63616824
Proceeds from exchangeable bonds, net of debt issuance cost10,9861,543
Proceeds from bank borrowings and other borrowings, net of upfront fee payment for a syndicated loan22,09837,8055,310
Repayment of bank borrowings(20,041)(20,019)(2,812)
Net cash (used in) provided by financing activities(86,364)8,1711,148
Effect of exchange rate changes on cash and cash equivalents, restricted cash and escrow receivables(1,797)(1,443)(203)
Decrease in cash and cash equivalents, restricted cash and escrow receivables(57,952)(13,825)(1,942)
Cash and cash equivalents, restricted cash and escrow receivables at beginning of period286,424189,26826,586
Cash and cash equivalents, restricted cash and escrow receivables at end of period228,472175,44324,644
", + "path": "tables/table-25 Cash Flow Statement.html", + "metadata": { + "length": 4033, + "summary": "table-26\nThe table shows cash flows for six months ending Sep 30, 2024 and 2025. Operating cash flow decreased from 65,074 to 30,771. Investing activities used 34,865 in 2024 and 51,324 in 2025. Financing activities shifted from -86,364 to 8,171.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-25 Cash Flow Statement.html", + "keywords": [ + "Cash flows", + "Investing", + "Financing" + ], + "tokens": [] + } + }, + { + "chunk_id": "7a395fd5-ab2c-59c0-85fd-3099acec4397", + "type": "text", + "content": "[tables/table-25 Cash Flow Statement.html]\n\nThe accompanying notes form an integral part of these unaudited condensed consolidated financial statements.\nALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_2/UNAUDITED CONDENSED CONSOLIDATED STATEMENTS OF CASH FLOWS", + "metadata": { + "length": 302, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "accompanying", + "notes", + "form", + "integral", + "part", + "unaudited", + "condensed", + "consolidated", + "financial", + "statements", + "ALIBABA", + "GROUP", + "HOLDING", + "LIMITED", + "NOTES", + "TO", + "UNAUDITED", + "CONDENSED", + "CONSOLIDATED", + "FINANCIAL", + "STATEMENTS", + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [ + { + "target": "16ede654-c979-5091-8964-4d3c4be26524", + "relation": "embeds", + "ref": "[tables/table-25 Cash Flow Statement.html]", + "position": { + "start": 0, + "end": 42 + } + } + ] + } + }, + { + "chunk_id": "4ae780ef-abf5-59df-9696-006cb85ed1d7", + "type": "text", + "content": "Alibaba Group Holding Limited (the “Company”) is a limited liability company, which was incorporated in the Cayman Islands on June 28, 1999. The Company is a holding company and conducts its businesses primarily through its subsidiaries and variable interest entities. In these unaudited condensed consolidated financial statements, where appropriate, the term “Company” also refers to its subsidiaries as a whole. The Company provides the technology infrastructure and marketing reach to help merchants, brands, retailers and other businesses to leverage the power of new technology to engage with their users and customers and operate in a more efficient way.\nThe Company's businesses comprise Alibaba China E-commerce Group, Alibaba International Digital Commerce Group, Cloud Intelligence Group and All others. An ecosystem has developed around the Company's platforms and businesses that consists of consumers, merchants, brands, retailers, enterprises, third-party service providers, strategic alliance partners and other businesses.\nThe Company's American depositary shares (“ADSs”) are listed on the New York Stock Exchange (“NYSE”) under the symbol “BABA” and the Company’s ordinary shares are listed on the Hong Kong Stock Exchange (“HKSE”) under the stock codes “9988 (HKD Counter)” and “89988 (RMB Counter).”", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_2/1. Organization and principal activities", + "metadata": { + "length": 1320, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Alibaba", + "Group", + "Holding", + "Limited", + "Company", + "limited", + "liability", + "company", + "incorporated", + "Cayman", + "Islands", + "June", + "28", + "1999", + "The", + "holding", + "conducts", + "businesses", + "primarily", + "subsidiaries", + "variable", + "interest", + "entities", + "In", + "unaudited", + "condensed", + "consolidated", + "financial", + "statements", + "term", + "refers", + "technology", + "infrastructure", + "marketing", + "reach", + "merchants", + "brands", + "retailers", + "leverage", + "power", + "engage", + "users", + "customers", + "operate", + "efficient", + "comprise", + "China", + "commerce", + "International", + "Digital", + "Commerce", + "Cloud", + "Intelligence", + "All", + "An", + "ecosystem", + "developed", + "platforms", + "consists", + "consumers", + "enterprises", + "party", + "service", + "providers", + "strategic", + "alliance", + "partners", + "American", + "depositary", + "shares", + "ADSs", + "listed", + "New", + "York", + "Stock", + "Exchange", + "NYSE", + "symbol", + "BABA", + "ordinary", + "Hong", + "Kong", + "HKSE", + "stock", + "codes", + "9988", + "HKD", + "Counter", + "89988", + "RMB" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "460358e7-d379-519c-949f-6d5f575812d5", + "type": "text", + "content": "(a) Basis of presentation\nThe accompanying unaudited condensed consolidated financial statements are prepared in accordance with accounting principles generally accepted in the United States of America (“U.S. GAAP”) for interim financial reporting and the disclosure requirements of the Rules Governing the Listing of Securities on The Stock Exchange of Hong Kong Limited, as amended or supplemented from time to time (“Hong Kong Listing Rules”). Accordingly, they do not include all of the information and footnote disclosures required by U.S. GAAP for a complete set of financial statements. These unaudited condensed consolidated financial statements include all adjustments of a normal recurring nature necessary to a fair statement of the results for the interim periods presented. Results of operations for an interim period are not necessarily indicative of results for the entire year.\nThese unaudited condensed consolidated financial statements should be read in conjunction with the audited consolidated financial statements and notes thereto for the year ended March 31, 2025. There were no significant changes to our significant accounting policies from the audited consolidated financial statements for the preceding fiscal year, except that the accounting policies relating to derivatives and hedging (Note 2(d)) and borrowings (Note 2(e)) were updated as a result of the issuance of exchangeable bonds.\nTranslations of balances in the unaudited condensed consolidated balance sheet, unaudited condensed consolidated income statements, unaudited condensed consolidated statement of comprehensive income and unaudited condensed consolidated statement of cash flows from RMB into the United States Dollar (“US\\$”) as of and for the six months ended September 30, 2025 are solely for the convenience of the readers and are calculated at the rate of US\\$1.00=RMB7.1190, representing the exchange rate set forth in the H.10 statistical release of the Federal Reserve Board on September 30, 2025. No representation is made that the RMB amounts could have been, or could be, converted, realized or settled into US\\$ at this rate, or at any other rate.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_2/2. Summary of significant accounting policies", + "metadata": { + "length": 2158, + "summary": "The document outlines the basis of presentation for unaudited condensed consolidated financial statements prepared under U.S. GAAP and Hong Kong Listing Rules for interim periods. It notes that these statements do not include all disclosures required for a full set of financials, but include normal recurring adjustments. The statements should be read with the audited financials for the year ended March 31, 2025, with no significant accounting policy changes except updates for derivatives, hedging, and borrowings due to exchangeable bonds. RMB amounts are translated to US dollars at a rate of 7.1190 for convenience, with no guarantee of convertibility at that rate.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Basis", + "presentation", + "The", + "accompanying", + "unaudited", + "condensed", + "consolidated", + "financial", + "statements", + "prepared", + "accordance", + "accounting", + "principles", + "generally", + "accepted", + "United", + "States", + "America", + "GAAP", + "interim", + "reporting", + "disclosure", + "requirements", + "Rules", + "Governing", + "Listing", + "Securities", + "Stock", + "Exchange", + "Hong", + "Kong", + "Limited", + "amended", + "supplemented", + "time", + "Accordingly", + "include", + "information", + "footnote", + "disclosures", + "required", + "complete", + "set", + "These", + "adjustments", + "normal", + "recurring", + "nature", + "fair", + "statement", + "results", + "periods", + "presented", + "Results", + "operations", + "period", + "necessarily", + "indicative", + "entire", + "year", + "read", + "conjunction", + "audited", + "notes", + "thereto", + "ended", + "March", + "31", + "2025", + "There", + "significant", + "policies", + "preceding", + "fiscal", + "relating", + "derivatives", + "hedging", + "Note", + "borrowings", + "updated", + "result", + "issuance", + "exchangeable", + "bonds", + "Translations", + "balances", + "balance", + "sheet", + "income", + "comprehensive", + "cash", + "flows", + "RMB", + "Dollar", + "US", + "months", + "September", + "30", + "solely", + "convenience", + "readers", + "calculated", + "rate", + "1.00", + "RMB7.1190", + "representing", + "exchange", + "H.10", + "statistical", + "release", + "Federal", + "Reserve", + "Board", + "No", + "representation", + "made", + "amounts", + "converted", + "realized", + "settled" + ], + "keywords": [ + "U.S. GAAP", + "Exchangeable Bonds", + "RMB Translation" + ], + "connect_to": [] + } + }, + { + "chunk_id": "71f755ac-d123-5bff-bbc0-cdc6ab203c04", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fff0fc29-3e7a-521d-99c5-8cffaed65721", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ae585794-4405-51cc-9cdb-e7bbad89e68d", + "type": "text", + "content": "The preparation of the unaudited condensed consolidated financial statements in conformity with U.S. GAAP requires the Company to make estimates and assumptions that affect the reported amounts of assets and liabilities, disclosure of contingent assets and liabilities as of the date of the unaudited condensed consolidated financial statements and the reported amounts of revenue and expenses during the reporting period. The Company bases its estimates on historical experience and on various other assumptions that are believed to be reasonable, the results of which form the basis for making judgments about the carrying values of assets and liabilities.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)/(b) Use of estimates", + "metadata": { + "length": 658, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "preparation", + "unaudited", + "condensed", + "consolidated", + "financial", + "statements", + "conformity", + "GAAP", + "requires", + "Company", + "make", + "estimates", + "assumptions", + "affect", + "reported", + "amounts", + "assets", + "liabilities", + "disclosure", + "contingent", + "date", + "revenue", + "expenses", + "reporting", + "period", + "bases", + "historical", + "experience", + "believed", + "reasonable", + "results", + "form", + "basis", + "making", + "judgments", + "carrying", + "values" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8a7d3377-59dd-5525-8c4d-5be0b3cccf52", + "type": "table", + "content": "
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
Cash and cash equivalents and short-term investments10,6218,807
Investments in equity method investees and equity securities and other investments37,11735,335
Accounts receivable and contract assets, net of allowance18,40818,666
Amounts due from non-VIE subsidiaries of the Company53,79251,309
Property and equipment, net and intangible assets, net19,91137,527
Others40,42956,590
Total assets180,278208,234
Amounts due to non-VIE subsidiaries of the Company113,332131,635
Accrued expenses, accounts payable and other liabilities50,52160,882
Deferred revenue and customer advances19,12618,669
Total liabilities182,979211,186
Six months ended September 30,
20242025
RMBRMB
(in millions)
Revenue (i)67,73977,466
Net income2,1032,442
Net cash provided by operating activities12,1182,817
Net cash used in investing activities(28,993)(31,914)
Net cash provided by financing activities8,62527,034
", + "path": "tables/table-26 Financial Data.html", + "metadata": { + "length": 1789, + "summary": "table-27\nThe table shows financial data for a company as of March 31 and September 30, 2025, and for six months ended September 30, 2024 and 2025. Total assets were 180,278 million RMB and 208,234 million RMB, while total liabilities were 182,979 million and 211,186 million. Revenue increased from 67,739 to 77,466 million, net income from 2,103 to 2,442 million.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-26 Financial Data.html", + "keywords": [ + "assets", + "liabilities", + "revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "34fcadf7-59f0-5422-b8e2-4b1ae61f95b6", + "type": "text", + "content": "The unaudited condensed consolidated financial statements include the financial statements of the Company and its subsidiaries, which include the PRC-registered entities directly or indirectly owned by the Company (“WFOEs”) and variable interest entities (“VIEs”) over which the Company is the primary beneficiary for accounting purposes only. All transactions and balances among the Company, its subsidiaries and the VIEs have been eliminated upon consolidation. The results of subsidiaries acquired or disposed of are recorded in the unaudited condensed consolidated income statements from the effective date of acquisition or up to the effective date of disposal, as appropriate. The nature of the businesses and activities of the consolidated VIEs have not changed materially from the preceding fiscal year.\nThe following financial information of the consolidated VIEs and their subsidiaries was recorded in the accompanying unaudited condensed consolidated financial statements:\n\n[tables/table-26 Financial Data.html]\n\n(i) Revenue generated by the VIEs are primarily from cloud services, digital media and entertainment services and others.\nALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)/(c) Consolidation", + "metadata": { + "length": 1312, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "unaudited", + "condensed", + "consolidated", + "financial", + "statements", + "include", + "Company", + "subsidiaries", + "PRC", + "registered", + "entities", + "directly", + "indirectly", + "owned", + "WFOEs", + "variable", + "interest", + "VIEs", + "primary", + "beneficiary", + "accounting", + "purposes", + "All", + "transactions", + "balances", + "eliminated", + "consolidation", + "results", + "acquired", + "disposed", + "recorded", + "income", + "effective", + "date", + "acquisition", + "disposal", + "nature", + "businesses", + "activities", + "changed", + "materially", + "preceding", + "fiscal", + "year", + "information", + "accompanying", + "Revenue", + "generated", + "primarily", + "cloud", + "services", + "digital", + "media", + "entertainment", + "ALIBABA", + "GROUP", + "HOLDING", + "LIMITED", + "NOTES", + "TO", + "UNAUDITED", + "CONDENSED", + "CONSOLIDATED", + "FINANCIAL", + "STATEMENTS", + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [ + { + "target": "8a7d3377-59dd-5525-8c4d-5be0b3cccf52", + "relation": "embeds", + "ref": "[tables/table-26 Financial Data.html]", + "position": { + "start": 985, + "end": 1022 + } + } + ] + } + }, + { + "chunk_id": "732045e3-0649-58f6-b477-1bb98d34a752", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_2", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "da71b847-ac22-535a-89aa-0e79609981fc", + "type": "text", + "content": "Contracts that meet the definition of a derivative and an embedded derivative that is not closely related to the host contract are generally recognized on the unaudited condensed consolidated balance sheets as either assets or liabilities and recorded at fair value. Changes in the fair value of derivative assets or liabilities are either recognized periodically in the unaudited condensed consolidated income statements or in other comprehensive income depending on the use of the derivatives and whether they qualify for hedge accounting and are so designated as cash flow hedges, fair value hedges or net investment hedges. The capped call transactions in connection with the issuance of the convertible unsecured senior notes meet the scope exception for contracts in an entity’s own equity provided in ASC 815 “Derivatives and Hedging” and are recognized in shareholders’ equity. Other instruments involving an entity’s own equity that do not meet the scope exception for contracts in an entity’s own equity provided in ASC 815 are classified as assets or liabilities and recorded at fair value.\nTo qualify for hedge accounting, the hedge relationship is designated and formally documented at inception, detailing the particular risk management objective and strategy for the hedge (which includes the item and risk that is being hedged), the hedging instrument that is being used and how hedge effectiveness is being assessed. A hedging instrument has to be effective in accomplishing the objective of offsetting changes in the risk being hedged. The effectiveness of the hedging relationship is evaluated on a prospective and retrospective basis using qualitative and quantitative measures of correlation. Qualitative methods may include comparison of critical terms of the hedging instrument to those of the hedged item due to the hedged risk.\nQuantitative methods include a comparison of the changes in the value or discounted cash flow of the hedging instrument to those of the hedged item due to the hedged risk. A hedging relationship is considered effective if the results of the hedging instrument are within a ratio of $80\\%$ to $125\\%$ of the results of the hedged item.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_2/(d) Derivatives and hedging", + "metadata": { + "length": 2187, + "summary": "The passage discusses the accounting treatment for derivatives and embedded derivatives under ASC 815. Derivatives are recorded at fair value on the balance sheet, with changes recognized in income or other comprehensive income depending on hedge designation. Capped call transactions meeting the scope exception are recognized in shareholders' equity. Hedge accounting requires formal documentation and effectiveness testing, with a hedging relationship considered effective if results fall within an 80% to 125% ratio.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Contracts", + "meet", + "definition", + "derivative", + "embedded", + "closely", + "related", + "host", + "contract", + "generally", + "recognized", + "unaudited", + "condensed", + "consolidated", + "balance", + "sheets", + "assets", + "liabilities", + "recorded", + "fair", + "Changes", + "periodically", + "income", + "statements", + "comprehensive", + "depending", + "derivatives", + "qualify", + "hedge", + "accounting", + "designated", + "cash", + "flow", + "hedges", + "net", + "investment", + "The", + "capped", + "call", + "transactions", + "connection", + "issuance", + "convertible", + "unsecured", + "senior", + "notes", + "scope", + "exception", + "contracts", + "entity", + "equity", + "provided", + "ASC", + "815", + "Derivatives", + "Hedging", + "shareholders", + "Other", + "instruments", + "involving", + "classified", + "To", + "relationship", + "formally", + "documented", + "inception", + "detailing", + "risk", + "management", + "objective", + "strategy", + "includes", + "item", + "hedged", + "hedging", + "instrument", + "effectiveness", + "assessed", + "effective", + "accomplishing", + "offsetting", + "evaluated", + "prospective", + "retrospective", + "basis", + "qualitative", + "quantitative", + "measures", + "correlation", + "Qualitative", + "methods", + "include", + "comparison", + "critical", + "terms", + "due", + "Quantitative", + "discounted", + "considered", + "results", + "ratio", + "80", + "125" + ], + "keywords": [ + "Derivatives", + "Hedge Accounting", + "Fair Value" + ], + "connect_to": [] + } + }, + { + "chunk_id": "2d711627-5185-5101-95cb-84e9e81ce03c", + "type": "text", + "content": "Interest rate swaps designated as hedging instruments to hedge against the cash flows attributable to recognized assets or liabilities or forecasted payments may qualify as cash flow hedges. The Company entered into interest rate swap contracts to swap floating interest payments related to certain borrowings for fixed interest payments to hedge the interest rate risk associated with certain forecasted payments and obligations. All changes in the fair value of interest rate swaps that are designated and qualify as cash flow hedges are recognized in accumulated other comprehensive income. Amounts in accumulated other comprehensive income are reclassified into earnings in the same period during which the hedged forecasted transaction affects earnings. The Company has elected the optional expedients under ASC 848 “Reference Rate Reform” for certain existing interest rate swaps that are designated as cash flow hedges in the hedging relationship designation and the assessment of probability of forecasted transaction and hedge effectiveness.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_2/(d) Derivatives and hedging/Cash flow hedges", + "metadata": { + "length": 1050, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Interest", + "rate", + "swaps", + "designated", + "hedging", + "instruments", + "hedge", + "cash", + "flows", + "attributable", + "recognized", + "assets", + "liabilities", + "forecasted", + "payments", + "qualify", + "flow", + "hedges", + "The", + "Company", + "entered", + "interest", + "swap", + "contracts", + "floating", + "related", + "borrowings", + "fixed", + "risk", + "obligations", + "All", + "fair", + "accumulated", + "comprehensive", + "income", + "Amounts", + "reclassified", + "earnings", + "period", + "hedged", + "transaction", + "affects", + "elected", + "optional", + "expedients", + "ASC", + "848", + "Reference", + "Rate", + "Reform", + "existing", + "relationship", + "designation", + "assessment", + "probability", + "effectiveness" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f649f490-7a7b-5f5c-9db0-4f6fbbd0ef09", + "type": "text", + "content": "The Company uses cross currency swap (“CCS”) contracts and RMB denominated unsecured senior notes and borrowings to hedge the foreign currency risk associated with investments in net assets of certain PRC subsidiaries held by the Company which are designated as net investment hedges. The Company excludes the changes in the fair value of the CCS contracts attributable to changes other than those due to fluctuations in the spot exchange rate from the assessment of hedge effectiveness and the value of such excluded component is recognized in interest expenses in the unaudited condensed consolidated income statement over the life of the hedging instrument under a systematic and rational method. Changes in the value of the hedging instruments due to fluctuations in the spot foreign currency exchange rates designated in net investment hedges are recognized in accumulated other comprehensive income to offset the cumulative translation adjustments relating to those subsidiaries. For the six months ended September 30, 2025, the losses recognized in accumulated other comprehensive income from changes in value of the CCS contracts and the non-derivative financial instruments designated as net investment hedges amounted to RMB1,279 million.\nAmounts accumulated are reclassified from accumulated other comprehensive income and recognized in the unaudited condensed consolidated income statements upon disposal of those subsidiaries. Once the hedge becomes ineffective, hedge accounting is discontinued prospectively. The estimated fair value of the derivatives is determined based on relevant market information. These estimates are calculated with reference to the market rates using industry standard valuation techniques.\nALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_2/(d) Derivatives and hedging/Net investment hedges", + "metadata": { + "length": 1877, + "summary": "The Company uses cross currency swap contracts and RMB denominated unsecured senior notes and borrowings to hedge foreign currency risk from investments in PRC subsidiaries, designated as net investment hedges. Changes in fair value from spot exchange rate fluctuations are recognized in accumulated other comprehensive income, while excluded components are recognized in interest expenses. For the six months ended September 30, 2025, losses of RMB1,279 million were recognized in accumulated other comprehensive income. Amounts are reclassified to income upon subsidiary disposal, and hedge accounting is discontinued if ineffective. Fair values are estimated using market rates and industry standard valuation techniques.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "Company", + "cross", + "currency", + "swap", + "CCS", + "contracts", + "RMB", + "denominated", + "unsecured", + "senior", + "notes", + "borrowings", + "hedge", + "foreign", + "risk", + "investments", + "net", + "assets", + "PRC", + "subsidiaries", + "held", + "designated", + "investment", + "hedges", + "excludes", + "fair", + "attributable", + "due", + "fluctuations", + "spot", + "exchange", + "rate", + "assessment", + "effectiveness", + "excluded", + "component", + "recognized", + "interest", + "expenses", + "unaudited", + "condensed", + "consolidated", + "income", + "statement", + "life", + "hedging", + "instrument", + "systematic", + "rational", + "method", + "Changes", + "instruments", + "rates", + "accumulated", + "comprehensive", + "offset", + "cumulative", + "translation", + "adjustments", + "relating", + "For", + "months", + "ended", + "September", + "30", + "2025", + "losses", + "derivative", + "financial", + "amounted", + "RMB1", + "279", + "million", + "Amounts", + "reclassified", + "statements", + "disposal", + "Once", + "ineffective", + "accounting", + "discontinued", + "prospectively", + "estimated", + "derivatives", + "determined", + "based", + "relevant", + "market", + "information", + "These", + "estimates", + "calculated", + "reference", + "industry", + "standard", + "valuation", + "techniques", + "ALIBABA", + "GROUP", + "HOLDING", + "LIMITED", + "NOTES", + "TO", + "UNAUDITED", + "CONDENSED", + "CONSOLIDATED", + "FINANCIAL", + "STATEMENTS", + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "2024" + ], + "keywords": [ + "hedging", + "foreign currency", + "derivatives" + ], + "connect_to": [] + } + }, + { + "chunk_id": "3f4849d9-a5a4-53f5-ba20-12ce846c1646", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_3", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "33ff0b0e-e43b-507e-a7c6-4cba7e2c801a", + "type": "text", + "content": "Borrowings consist of bank borrowings, unsecured senior notes, convertible unsecured senior notes and exchangeable bonds. Bank borrowings and unsecured senior notes are recognized initially at fair value, net of upfront fees, debt discounts or premiums, debt issuance costs and other incidental fees. Upfront fees, debt discounts or premiums, debt issuance costs and other incidental fees are recorded as a reduction of the proceeds received and the related accretion is recorded as interest expense in the unaudited condensed consolidated income statements over the estimated term of the facilities using the effective interest method. Convertible unsecured senior notes are accounted for in its entirety as liabilities, and the embedded conversion feature is not required to be accounted for separately under ASC 815. Exchangeable bonds are accounted for under the fair value option with changes in fair value recorded in the unaudited condensed consolidated income statements.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_3/(e) Borrowings", + "metadata": { + "length": 979, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Borrowings", + "consist", + "bank", + "borrowings", + "unsecured", + "senior", + "notes", + "convertible", + "exchangeable", + "bonds", + "Bank", + "recognized", + "initially", + "fair", + "net", + "upfront", + "fees", + "debt", + "discounts", + "premiums", + "issuance", + "costs", + "incidental", + "Upfront", + "recorded", + "reduction", + "proceeds", + "received", + "related", + "accretion", + "interest", + "expense", + "unaudited", + "condensed", + "consolidated", + "income", + "statements", + "estimated", + "term", + "facilities", + "effective", + "method", + "Convertible", + "accounted", + "entirety", + "liabilities", + "embedded", + "conversion", + "feature", + "required", + "separately", + "ASC", + "815", + "Exchangeable", + "option" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "bd7f7eb3-fc1a-530e-a7d7-5dc398e56957", + "type": "text", + "content": "In July 2025, the Financial Accounting Standards Board (“FASB”) issued ASU 2025-05, “Financial Instruments—Credit Losses (Topic 326): Measurement of Credit Losses for Accounts Receivable and Contract Assets”. The amendments provide all entities with a practical expedient when estimating expected credit losses for current accounts receivable and current contract assets arising from transactions accounted for under Topic 606. In developing reasonable and supportable forecasts as part of estimating expected credit losses, all entities may elect a practical expedient that assumes that current conditions as of the balance sheet date do not change for the remaining life of the asset. The new guidance is required to be applied prospectively. This guidance is effective for the Company for the year ending March 31, 2027. Early adoption is permitted. The Company is evaluating the impact of the adoption of this guidance.\nIn September 2025, the FASB issued ASU 2025-06, “Intangibles—Goodwill and Other—Internal-Use Software (Subtopic 350-40): Targeted Improvements to the Accounting for Internal-Use Software”, which removes all references to prescriptive and sequential software development stages (“project stages”) throughout Subtopic 350-40. The amendments provide guidance on determining whether there is significant development uncertainty in evaluating the probable-to-complete recognition threshold. The amendments also supersede the website development costs guidance and incorporate the recognition requirements for website-specific development costs. This guidance also specifies the disclosures requirements for capitalized internal-use software costs. The new guidance is required to be applied using either the prospective transition approach, the modified transition approach or the retrospective transition approach. This guidance is effective for the Company for the year ending March 31, 2029. Early adoption is permitted. The Company is evaluating the impact of the adoption of this guidance.\nIn November 2025, the FASB issued ASU 2025-08, “Financial Instruments—Credit Losses (Topic 326): Purchased Loans”, which defines purchased seasoned loans and expands the use of the gross-up approach in ASC 326 to the purchased seasoned loans. The new guidance is required to be applied prospectively to loans that are acquired on or after the initial application date. This guidance is effective for the Company for the year ending March 31, 2028. Early adoption is permitted. The Company is evaluating the impact of the adoption of this guidance.\nIn November 2025, the FASB issued ASU 2025-09, “Derivatives and Hedging (Topic 815): Hedge Accounting Improvements”, which clarifies certain aspects of the guidance on hedge accounting and addresses several incremental hedge accounting issues arising from the global reference rate reform initiative. The amendments include: (i) expanding the hedged risks permitted to be aggregated in a group of individual forecasted transactions in a cash flow hedge and clarifying the circumstance under which a group of individual forecasted transactions can be considered to have a similar risk exposure; (ii) providing a model to facilitate the application of cash flow hedge accounting to forecasted interest payments on choose-your-rate debt instruments; (iii) expanding hedge accounting for forecasted purchases and sales of nonfinancial assets; (iv) eliminating the requirement to apply the net written option test to a compound derivative comprising a swap and a written option designated as the hedging instrument in a cash flow hedge or a fair value hedge of interest rate risk; and (v) eliminating the recognition and presentation mismatch related to a dual hedge strategy. The new guidance is required to be applied prospectively for all hedging relationships. This guidance is effective for the Company for the year ending March 31, 2028. Early adoption is permitted. The Company is evaluating the impact of the adoption of this guidance.\nALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_3/(f) Recent accounting pronouncements", + "metadata": { + "length": 4146, + "summary": "The passage discusses four FASB Accounting Standards Updates (ASUs) issued in 2025. ASU 2025-05 provides a practical expedient for estimating credit losses on current receivables and contract assets, effective for the year ending March 31, 2027. ASU 2025-06 removes prescriptive software development stages for internal-use software, effective for the year ending March 31, 2029. ASU 2025-08 defines purchased seasoned loans and expands the gross-up approach for credit losses, effective for the year ending March 31, 2028. ASU 2025-09 clarifies hedge accounting, including expanding hedged risks and eliminating certain tests, effective for the year ending March 31, 2028. All updates permit early adoption, and the Company is evaluating their impacts.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "In", + "July", + "2025", + "Financial", + "Accounting", + "Standards", + "Board", + "FASB", + "issued", + "ASU", + "05", + "Instruments", + "Credit", + "Losses", + "Topic", + "326", + "Measurement", + "Accounts", + "Receivable", + "Contract", + "Assets", + "The", + "amendments", + "provide", + "entities", + "practical", + "expedient", + "estimating", + "expected", + "credit", + "losses", + "current", + "accounts", + "receivable", + "contract", + "assets", + "arising", + "transactions", + "accounted", + "606", + "developing", + "reasonable", + "supportable", + "forecasts", + "part", + "elect", + "assumes", + "conditions", + "balance", + "sheet", + "date", + "change", + "remaining", + "life", + "asset", + "guidance", + "required", + "applied", + "prospectively", + "This", + "effective", + "Company", + "year", + "ending", + "March", + "31", + "2027", + "Early", + "adoption", + "permitted", + "evaluating", + "impact", + "September", + "06", + "Intangibles", + "Goodwill", + "Other", + "Internal", + "Use", + "Software", + "Subtopic", + "350", + "40", + "Targeted", + "Improvements", + "removes", + "references", + "prescriptive", + "sequential", + "software", + "development", + "stages", + "project", + "determining", + "significant", + "uncertainty", + "probable", + "complete", + "recognition", + "threshold", + "supersede", + "website", + "costs", + "incorporate", + "requirements", + "specific", + "specifies", + "disclosures", + "capitalized", + "internal", + "prospective", + "transition", + "approach", + "modified", + "retrospective", + "2029", + "November", + "08", + "Purchased", + "Loans", + "defines", + "purchased", + "seasoned", + "loans", + "expands", + "gross", + "ASC", + "acquired", + "initial", + "application", + "2028", + "09", + "Derivatives", + "Hedging", + "815", + "Hedge", + "clarifies", + "aspects", + "hedge", + "accounting", + "addresses", + "incremental", + "issues", + "global", + "reference", + "rate", + "reform", + "initiative", + "include", + "expanding", + "hedged", + "risks", + "aggregated", + "group", + "individual", + "forecasted", + "cash", + "flow", + "clarifying", + "circumstance", + "considered", + "similar", + "risk", + "exposure", + "ii", + "providing", + "model", + "facilitate", + "interest", + "payments", + "choose", + "debt", + "instruments", + "iii", + "purchases", + "sales", + "nonfinancial", + "iv", + "eliminating", + "requirement", + "apply", + "net", + "written", + "option", + "test", + "compound", + "derivative", + "comprising", + "swap", + "designated", + "hedging", + "instrument", + "fair", + "presentation", + "mismatch", + "related", + "dual", + "strategy", + "relationships", + "ALIBABA", + "GROUP", + "HOLDING", + "LIMITED", + "NOTES", + "TO", + "UNAUDITED", + "CONDENSED", + "CONSOLIDATED", + "FINANCIAL", + "STATEMENTS", + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024" + ], + "keywords": [ + "FASB", + "ASU", + "Accounting Standards" + ], + "connect_to": [] + } + }, + { + "chunk_id": "cbca7f98-abeb-566b-a191-6d49f65bb57b", + "type": "text", + "content": "Disposal of Trendyol GO\nIn May 2025, the Company entered into a sale and purchase agreement to sell 85% of the equity interest in Trendyol GO, a wholly-owned subsidiary of Trendyol that operates local service business in Türkiye (the \"Disposal\"). The cash consideration for the Disposal is approximately US\\$0.7 billion (RMB5 billion). The Disposal was completed during the six months ended September 30, 2025, and a gain arising from the Disposal of approximately RMB6 billion was recorded in interest and investment income, net in the unaudited condensed consolidated income statements for the six months ended September 30, 2025, primarily taking into consideration (i) the cash considerations received, (ii) the carrying values of the net assets of Trendyol GO, and (iii) the fair value of the 15% retained equity interest in Trendyol GO.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/3. Significant mergers and acquisitions, investments and dispositions", + "metadata": { + "length": 842, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Disposal", + "Trendyol", + "GO", + "In", + "May", + "2025", + "Company", + "entered", + "sale", + "purchase", + "agreement", + "sell", + "85%", + "equity", + "interest", + "wholly", + "owned", + "subsidiary", + "operates", + "local", + "service", + "business", + "rkiye", + "The", + "cash", + "consideration", + "approximately", + "US", + "0.7", + "billion", + "RMB5", + "completed", + "months", + "ended", + "September", + "30", + "gain", + "arising", + "RMB6", + "recorded", + "investment", + "income", + "net", + "unaudited", + "condensed", + "consolidated", + "statements", + "primarily", + "taking", + "considerations", + "received", + "ii", + "carrying", + "values", + "assets", + "iii", + "fair", + "15%", + "retained" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "43f47aa9-02e7-5678-934d-850767a0e84d", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_4", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_4/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "52362dc1-7657-5609-9e11-1d9f5c72ec56", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMB
(in millions)
Alibaba China E-commerce Group:
E-commerce (i)
- Customer management152,755168,179
- Direct sales, logistics and others (ii)50,23353,331
202,988221,510
Quick commerce (iii)27,51737,690
China commerce wholesale (iv)11,93813,450
Total Alibaba China E-commerce Group:242,443272,650
Alibaba International Digital Commerce Group:
International commerce retail (v)49,30956,463
International commerce wholesale (vi)11,65613,077
Total Alibaba International Digital Commerce Group60,96569,540
Cloud Intelligence Group (vii)56,15973,222
All others (viii)165,837121,568
Unallocated8881,096
Inter-segment elimination (ix)(46,553)(42,629)
Consolidated revenue479,739495,447
", + "path": "tables/table-27 Alibaba Revenue.html", + "metadata": { + "length": 1463, + "summary": "table-28\nAlibaba Group's consolidated revenue for six months ended Sep 30, 2025 was RMB 495,447 million, up from RMB 479,739 million in 2024, driven by growth in China E-commerce, International Digital Commerce, and Cloud Intelligence.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-27 Alibaba Revenue.html", + "keywords": [ + "Revenue", + "Segments", + "Comparison" + ], + "tokens": [] + } + }, + { + "chunk_id": "22f4487a-93f7-547c-ae22-a789920a9054", + "type": "text", + "content": "Revenue by segment is as follows:\n\n[tables/table-27 Alibaba Revenue.html]\n\n(i) Revenue from E-commerce is primarily generated from Alibaba China E-commerce Group and includes primarily revenue from customer management services, sales of goods and logistics services.\n(ii) Revenue from direct sales, logistics and other revenue under Alibaba China E-commerce Group primarily represents direct sales businesses of Tmall Supermarket, Tmall Global and other businesses, and primarily consists of revenue from sales of goods, as well as logistics services.\n(iii) Revenue from Quick commerce is primarily generated from through “Taobao Instant Commerce” and the Ele.me app. Quick commerce revenue is net of subsidies that are contra revenue, and includes primarily revenue from logistics services and customer management services.\n(iv) Revenue from China commerce wholesale is primarily generated from 1688.com and includes revenue from membership fees and related value-added services and customer management services.\n(v) Revenue from International commerce retail is primarily generated from AliExpress, Trendyol and Lazada and includes revenue from customer management services, logistics services and sales of goods.\n(vi) Revenue from International commerce wholesale is primarily generated from Alibaba.com and includes revenue from membership fees and related value-added services and customer management services.\n(vii) Revenue from Cloud Intelligence Group is primarily generated from the provision of cloud services, which include public cloud services and non-public cloud services.\n(viii) Revenue from All others represents revenue from businesses including Freshippo, Cainiao, Alibaba Health, Hujing Digital Media and Entertainment Group, Amap, Intelligent Information Platform (which mainly consists of UCWeb and Quark businesses), Lingxi Games, DingTalk and other businesses. Revenue within All others includes primarily revenue from sales of goods, and logistics services.\n(ix) Inter-segment elimination consists of revenue primarily from Cloud Intelligence Group and Cainiao.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_4/4. Revenue", + "metadata": { + "length": 2094, + "summary": "The passage details revenue breakdown by segment for Alibaba. Key segments include E-commerce (Alibaba China E-commerce Group with customer management, goods, logistics), Quick commerce (Taobao Instant Commerce, Ele.me), China commerce wholesale (1688.com), International commerce retail (AliExpress, Trendyol, Lazada), International commerce wholesale (Alibaba.com), Cloud Intelligence Group (cloud services), and All others (Freshippo, Cainiao, Alibaba Health, etc.). Inter-segment eliminations primarily involve Cloud Intelligence Group and Cainiao.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Revenue", + "segment", + "commerce", + "primarily", + "generated", + "Alibaba", + "China", + "Group", + "includes", + "revenue", + "customer", + "management", + "services", + "sales", + "goods", + "logistics", + "ii", + "direct", + "represents", + "businesses", + "Tmall", + "Supermarket", + "Global", + "consists", + "iii", + "Quick", + "Taobao", + "Instant", + "Commerce", + "Ele", + "app", + "net", + "subsidies", + "contra", + "iv", + "wholesale", + "1688", + "membership", + "fees", + "related", + "added", + "International", + "retail", + "AliExpress", + "Trendyol", + "Lazada", + "vi", + "vii", + "Cloud", + "Intelligence", + "provision", + "cloud", + "include", + "public", + "viii", + "All", + "including", + "Freshippo", + "Cainiao", + "Health", + "Hujing", + "Digital", + "Media", + "Entertainment", + "Amap", + "Intelligent", + "Information", + "Platform", + "UCWeb", + "Quark", + "Lingxi", + "Games", + "DingTalk", + "ix", + "Inter", + "elimination" + ], + "keywords": [ + "E-commerce", + "Cloud", + "Commerce" + ], + "connect_to": [ + { + "target": "52362dc1-7657-5609-9e11-1d9f5c72ec56", + "relation": "embeds", + "ref": "[tables/table-27 Alibaba Revenue.html]", + "position": { + "start": 35, + "end": 73 + } + } + ] + } + }, + { + "chunk_id": "627ab9d5-9aef-5225-9da5-23a2d9fbb3ea", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_5", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_5/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7cabbae6-1c04-5b35-8ee9-bd7693a7b129", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMB
(in millions)
Customer management services (i)199,821223,399
Membership fees and value-added services23,04424,035
Logistics services59,90468,018
Cloud services41,32652,826
Sales of goods135,837108,435
Other revenue (ii)19,80718,734
479,739495,447
", + "path": "tables/table-28 Revenue Breakdown.html", + "metadata": { + "length": 685, + "summary": "table-29\nThe table shows revenue for six months ending Sep 30, 2024 and 2025 in RMB millions. Key services include customer management, membership, logistics, cloud, goods sales, and other revenue, totaling 479,739 in 2024 and 495,447 in 2025.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-28 Revenue Breakdown.html", + "keywords": [ + "Revenue", + "Services", + "Six Months" + ], + "tokens": [] + } + }, + { + "chunk_id": "15a25806-8ba9-5afc-9d5b-cc1f7d352236", + "type": "text", + "content": "(x) As a result of the change in composition in reportable segments (Note 18), the Company reclassified revenue by segment. Comparative figures for the six months ended September 30, 2024 were updated to conform to the segment presentation.\nRevenue by type is as follows:\n\n[tables/table-28 Revenue Breakdown.html]\n\n(i) Customer management services mainly include cost-per-click (\"CPC\"), cost-per-thousand impressions (\"CPM\"), time-based and cost-per-sale (\"CPS\") marketing services.\n(ii) Other revenue includes revenue from self-developed online games and multiple services provided through various platforms and businesses.\nThe amount of revenue recognized for performance obligations satisfied (or partially satisfied) in prior periods for contracts with expected duration of more than one year during the six months ended September 30, 2024 and 2025 were not material.\nDeferred revenue and customer advances of the Company primarily represent service fees prepaid by merchants or customers for which the relevant services have not been provided. Substantially all of the balances of deferred revenue and customer advances are generally recognized as revenue within one year.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_5/4. Revenue (Continued)", + "metadata": { + "length": 1183, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "As", + "result", + "change", + "composition", + "reportable", + "segments", + "Note", + "18", + "Company", + "reclassified", + "revenue", + "segment", + "Comparative", + "figures", + "months", + "ended", + "September", + "30", + "2024", + "updated", + "conform", + "presentation", + "Revenue", + "type", + "Customer", + "management", + "services", + "include", + "cost", + "click", + "CPC", + "thousand", + "impressions", + "CPM", + "time", + "based", + "sale", + "CPS", + "marketing", + "ii", + "Other", + "includes", + "developed", + "online", + "games", + "multiple", + "provided", + "platforms", + "businesses", + "The", + "amount", + "recognized", + "performance", + "obligations", + "satisfied", + "partially", + "prior", + "periods", + "contracts", + "expected", + "duration", + "year", + "2025", + "material", + "Deferred", + "customer", + "advances", + "primarily", + "represent", + "service", + "fees", + "prepaid", + "merchants", + "customers", + "relevant", + "Substantially", + "balances", + "deferred", + "generally" + ], + "keywords": [], + "connect_to": [ + { + "target": "7cabbae6-1c04-5b35-8ee9-bd7693a7b129", + "relation": "embeds", + "ref": "[tables/table-28 Revenue Breakdown.html]", + "position": { + "start": 273, + "end": 313 + } + } + ] + } + }, + { + "chunk_id": "d8f11a06-45d2-563e-8415-711f6a7849b8", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_6", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_6/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "48220f87-af11-5cec-ae46-424d72de7304", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMB
(in millions)
Current income tax expense14,93913,985
Deferred taxation2,503430
17,44214,415
", + "path": "tables/table-29 Tax Expense Data.html", + "metadata": { + "length": 390, + "summary": "table-30\nThe table shows current income tax expense and deferred taxation for six-month periods ending September 30, 2024 and 2025, with totals of 17,442 and 14,415 million RMB respectively.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-29 Tax Expense Data.html", + "keywords": [ + "income tax", + "deferred taxation", + "RMB" + ], + "tokens": [] + } + }, + { + "chunk_id": "0bdf9b8a-4a46-5653-bfbf-904151570bbc", + "type": "text", + "content": "Composition of income tax expenses is as follows:\n\n[tables/table-29 Tax Expense Data.html]\n\nThe Company's effective tax rate for the six months ended September 30, 2024 and 2025 was $21\\%$ and $19\\%$ , respectively. For the interim financial reporting, the Company estimates the annual tax rate based on projected taxable income for the full year and records an income tax provision for the interim period in accordance with the guidance on accounting for income taxes in an interim period.\nThe tax status of the subsidiaries of the Company with major taxable profits is described below:\n\\- Alibaba (China) Technology Co., Ltd. (“Alibaba China”), Taobao (China) Software Co., Ltd. (“Taobao China”) and Zhejiang Tmall Technology Co., Ltd. (“Tmall China”), entities primarily engaged in the operations of the Company’s wholesale marketplaces, Taobao and Tmall, respectively, and Alibaba (Beijing) Software Services Co., Ltd (“Alibaba Beijing”) and Alibaba (China) Co., Ltd (“China Co.”), entities primarily engaged in the operations of technology, software research and development and relevant services, were qualified as High and New Technology Enterprises. For the taxation years of 2024 and 2025, Alibaba China, Taobao China, Tmall China, Alibaba Beijing and China Co. applied an EIT rate of 15% as High and New Technology Enterprises.\nMost of the remaining PRC entities of the Company are subject to EIT at 25% for the six months ended September 30, 2024 and 2025.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_6/5. Income tax expenses", + "metadata": { + "length": 1474, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Composition", + "income", + "tax", + "expenses", + "The", + "Company", + "effective", + "rate", + "months", + "ended", + "September", + "30", + "2024", + "2025", + "21", + "19", + "For", + "interim", + "financial", + "reporting", + "estimates", + "annual", + "based", + "projected", + "taxable", + "full", + "year", + "records", + "provision", + "period", + "accordance", + "guidance", + "accounting", + "taxes", + "status", + "subsidiaries", + "major", + "profits", + "Alibaba", + "China", + "Technology", + "Co", + "Ltd", + "Taobao", + "Software", + "Zhejiang", + "Tmall", + "entities", + "primarily", + "engaged", + "operations", + "wholesale", + "marketplaces", + "Beijing", + "Services", + "technology", + "software", + "research", + "development", + "relevant", + "services", + "qualified", + "High", + "New", + "Enterprises", + "taxation", + "years", + "applied", + "EIT", + "15%", + "Most", + "remaining", + "PRC", + "subject", + "25%" + ], + "keywords": [], + "connect_to": [ + { + "target": "48220f87-af11-5cec-ae46-424d72de7304", + "relation": "embeds", + "ref": "[tables/table-29 Tax Expense Data.html]", + "position": { + "start": 51, + "end": 90 + } + } + ] + } + }, + { + "chunk_id": "2c037159-55ff-540c-8c5f-8e964f78f78d", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMB
(in millions)
Cost of revenue1,205913
Product development expenses3,5602,862
Sales and marketing expenses948958
General and administrative expenses2,5642,137
8,2776,870
", + "path": "tables/table-30 Cost Breakdown.html", + "metadata": { + "length": 537, + "summary": "table-31\nThe table shows cost of revenue, product development, sales and marketing, and general administrative expenses for six months ending Sep 30, 2024 and 2025, with totals of 8,277 and 6,870 million RMB respectively.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-30 Cost Breakdown.html", + "keywords": [ + "Cost", + "Expenses", + "RMB" + ], + "tokens": [] + } + }, + { + "chunk_id": "33d5cfe0-5bd7-5df5-9788-6f8abc6845e7", + "type": "text", + "content": "Share-based compensation expense by function is as follows:\n\n[tables/table-30 Cost Breakdown.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_6/6. Share-based compensation expense by function", + "metadata": { + "length": 107, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Share", + "based", + "compensation", + "expense", + "function" + ], + "keywords": [], + "connect_to": [ + { + "target": "2c037159-55ff-540c-8c5f-8e964f78f78d", + "relation": "embeds", + "ref": "[tables/table-30 Cost Breakdown.html]", + "position": { + "start": 61, + "end": 98 + } + } + ] + } + }, + { + "chunk_id": "8b72c77d-bf9a-583a-bc99-2297b5e0c57d", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_7", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_7/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "55052ccf-a69d-50b7-8733-6e28f09d87e1", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMB
(in millions, except per share data)
Earnings per share
Numerator:
Net income attributable to ordinary shareholders for computing net income per ordinary share — basic68,14364,106
Dilution effect on earnings arising from non-cash share-based awards operated by equity method investees and subsidiaries(131)(258)
Adjustments for interest expense attributable to convertible unsecured senior notes95143
Net income attributable to ordinary shareholders for computing net income per ordinary share — diluted68,10763,991
Shares (denominator):
Weighted average number of shares used in calculating net income per ordinary share — basic (million shares)19,04518,562
Adjustments for dilutive RSUs and share options (million shares)149186
Adjustments for convertible unsecured senior notes (million shares)265406
Weighted average number of shares used in calculating net income per ordinary share — diluted (million shares)19,45919,154
Net income per ordinary share — basic (RMB)3.583.45
Net income per ordinary share — diluted (RMB)3.503.34
Earnings per ADS
Net income per ADS — basic (RMB)28.6227.63
Net income per ADS — diluted (RMB)28.0026.73
", + "path": "tables/table-31 EPS Data Table.html", + "metadata": { + "length": 1891, + "summary": "table-32\nThe table shows basic and diluted earnings per ordinary share and ADS for six months ending September 30, 2024 and 2025, with net income and share count adjustments.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-31 EPS Data Table.html", + "keywords": [ + "Earnings per share", + "Net income", + "Diluted" + ], + "tokens": [] + } + }, + { + "chunk_id": "666292ce-8f67-5389-95a3-c1bc740a99f5", + "type": "text", + "content": "Each ADS represents eight ordinary shares.\nBasic earnings per share is computed by dividing net income attributable to ordinary shareholders by the weighted average number of outstanding ordinary shares, adjusted for treasury shares. Basic earnings per ADS is derived from the basic earnings per share.\nFor the calculation of diluted earnings per share, net income attributable to ordinary shareholders for basic earnings per share is adjusted by the effect of dilutive securities, including share-based awards under the treasury stock method and convertible unsecured senior notes and exchangeable bonds under the if-converted method. Certain potentially dilutive instruments, including the exchangeable bonds and the share lending arrangement in connection with the issuance of the exchangeable bonds, and the capped call transactions in connection with the issuance of the convertible unsecured senior notes have been excluded from the computation of diluted net income per share as their inclusion is anti-dilutive. Diluted earnings per ADS is derived from the diluted earnings per share.\nThe following table sets forth the computation of basic and diluted net income per share/ADS for the following periods:\n\n[tables/table-31 EPS Data Table.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_7/7. Earnings per share∕ADS", + "metadata": { + "length": 1260, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Each", + "ADS", + "represents", + "ordinary", + "shares", + "Basic", + "earnings", + "share", + "computed", + "dividing", + "net", + "income", + "attributable", + "shareholders", + "weighted", + "average", + "number", + "outstanding", + "adjusted", + "treasury", + "derived", + "basic", + "For", + "calculation", + "diluted", + "effect", + "dilutive", + "securities", + "including", + "based", + "awards", + "stock", + "method", + "convertible", + "unsecured", + "senior", + "notes", + "exchangeable", + "bonds", + "converted", + "Certain", + "potentially", + "instruments", + "lending", + "arrangement", + "connection", + "issuance", + "capped", + "call", + "transactions", + "excluded", + "computation", + "inclusion", + "anti", + "Diluted", + "The", + "table", + "sets", + "periods" + ], + "keywords": [], + "connect_to": [ + { + "target": "55052ccf-a69d-50b7-8733-6e28f09d87e1", + "relation": "embeds", + "ref": "[tables/table-31 EPS Data Table.html]", + "position": { + "start": 1214, + "end": 1251 + } + } + ] + } + }, + { + "chunk_id": "55415a65-d598-5572-ac82-a8542542fbfe", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "46cc6b97-cdd7-5fbb-9ecc-888194681086", + "type": "table", + "content": "
As of March 31, 2025
Original costCumulative net (losses) gainsCarrying value
RMBRMBRMB
(in millions)
Equity securities:
Listed equity securities79,024(1,394)77,630
Investments in privately held companies113,646(17,479)96,167
Debt investments:
Debt securities and loan investments15,009(6,224)8,785
Other treasury investments227,93581228,016
435,614(25,016)410,598
As of September 30, 2025
Original costCumulative net gains (losses)Carrying value
RMBRMBRMB
(in millions)
Equity securities:
Listed equity securities70,8804,42475,304
Investments in privately held companies120,899(1,279)119,620
Debt investments:
Debt securities and loan investments14,892(5,173)9,719
Other treasury investments252,464103252,567
459,135(1,925)457,210
", + "path": "tables/table-32 Investment Summary.html", + "metadata": { + "length": 1661, + "summary": "table-33\nThe table compares investment portfolios as of March 31, 2025 and September 30, 2025, showing original cost, cumulative net gains/losses, and carrying value for equity and debt investments in millions of RMB.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-32 Investment Summary.html", + "keywords": [ + "Equity securities", + "Debt investments", + "Carrying value" + ], + "tokens": [] + } + }, + { + "chunk_id": "c5854108-9e0c-5d03-b5f6-43f3085da536", + "type": "text", + "content": "[tables/table-32 Investment Summary.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/8. Equity securities and other investments", + "metadata": { + "length": 43, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [ + { + "target": "46cc6b97-cdd7-5fbb-9ecc-888194681086", + "relation": "embeds", + "ref": "[tables/table-32 Investment Summary.html]", + "position": { + "start": 0, + "end": 41 + } + } + ] + } + }, + { + "chunk_id": "c5693655-31cf-5b3c-acfb-c33c49c9b468", + "type": "text", + "content": "Net unrealized gains, including impairment losses, on equity securities, recognized in interest and investment income, net were RMB4,159 million and RMB20,017 million for the six months ended September 30, 2024 and 2025, respectively.\nInvestments in privately held companies include equity investments for which the Company elected to account for using the measurement alternative, for which the carrying value as of March 31 and September 30, 2025 were RMB88,728 million and RMB112,829 million, respectively. For the equity investments accounted for using the measurement alternative, the cumulative upward adjustments as of March 31 and September 30, 2025 were RMB27,197 million and RMB45,334 million, respectively, and the cumulative impairments and downward adjustments as of March 31 and September 30, 2025 were RMB44,232 million and RMB45,741 million, respectively.\nUpward adjustments recorded in the unaudited condensed consolidated income statements on equity investments accounted for using the measurement alternative were RMB4,614 million and RMB20,241 million for the six months ended September 30, 2024 and 2025, respectively, and impairments and downward adjustments recorded in the unaudited condensed consolidated income statements were RMB4,918 million and RMB4,338 million for the same periods, respectively.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/8. Equity securities and other investments/Equity securities", + "metadata": { + "length": 1326, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Net", + "unrealized", + "gains", + "including", + "impairment", + "losses", + "equity", + "securities", + "recognized", + "interest", + "investment", + "income", + "net", + "RMB4", + "159", + "million", + "RMB20", + "017", + "months", + "ended", + "September", + "30", + "2024", + "2025", + "Investments", + "privately", + "held", + "companies", + "include", + "investments", + "Company", + "elected", + "account", + "measurement", + "alternative", + "carrying", + "March", + "31", + "RMB88", + "728", + "RMB112", + "829", + "For", + "accounted", + "cumulative", + "upward", + "adjustments", + "RMB27", + "197", + "RMB45", + "334", + "impairments", + "downward", + "RMB44", + "232", + "741", + "Upward", + "recorded", + "unaudited", + "condensed", + "consolidated", + "statements", + "614", + "241", + "918", + "338", + "periods" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5031d9ef-8428-508a-b46d-e30efb85e1c5", + "type": "text", + "content": "Debt investments include convertible and exchangeable bonds accounted for under the fair value option, for which the fair value as of March 31 and September 30, 2025 were RMB963 million and RMB759 million, respectively. The aggregate fair value of these convertible and exchangeable bonds was lower than their aggregate unpaid principal balance as of March 31 and September 30, 2025 by RMB2,420 million and RMB2,529 million, respectively. Unrealized losses recorded in the unaudited condensed consolidated income statements on these convertible and exchangeable bonds were RMB32 million and RMB158 million for the six months ended September 30, 2024 and 2025, respectively.\nALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/8. Equity securities and other investments/Debt investments", + "metadata": { + "length": 819, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Debt", + "investments", + "include", + "convertible", + "exchangeable", + "bonds", + "accounted", + "fair", + "option", + "March", + "31", + "September", + "30", + "2025", + "RMB963", + "million", + "RMB759", + "The", + "aggregate", + "lower", + "unpaid", + "principal", + "balance", + "RMB2", + "420", + "529", + "Unrealized", + "losses", + "recorded", + "unaudited", + "condensed", + "consolidated", + "income", + "statements", + "RMB32", + "RMB158", + "months", + "ended", + "2024", + "ALIBABA", + "GROUP", + "HOLDING", + "LIMITED", + "NOTES", + "TO", + "UNAUDITED", + "CONDENSED", + "CONSOLIDATED", + "FINANCIAL", + "STATEMENTS", + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "09c5ba95-a054-58fe-b00b-14d320c0805f", + "type": "text", + "content": "Debt investments (Continued)\nDebt investments also include debt investments accounted for at amortized cost, for which the allowance for credit losses as of March 31 and September 30, 2025 were RMB3,779 million and RMB2,641 million, respectively. Reversal of impairment losses recorded in the unaudited condensed consolidated income statements on these debt investments were RMB1,117 million and RMB1,103 million for the six months ended September 30, 2024 and 2025, respectively.\nAs of September 30, 2025, repayment of loans provided to shareholders of equity method investees with total principal amount of RMB5,518 million was expected to be provided substantially through the sale of collateral. Expected credit losses for these loans were assessed on an individual basis, based on the fair value of the corresponding shares pledged as collateral as of the reporting date, adjusted for selling costs as appropriate. The fair value of these collateral as of March 31 and September 30, 2025 were RMB4,325 million and RMB6,250 million, respectively. There was no commitment to lend additional funds.\nThe carrying amount of debt investments accounted for at amortized cost approximates their fair value due to the fact that the related effective interest rates approximate rates currently offered by financial institutions for similar debt instruments of comparable maturities.\nOther treasury investments mainly comprise of investments in fixed deposits, certificates of deposits and marketable debt securities with original maturities over one year for treasury purposes.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/8. Equity securities and other investments (Continued)", + "metadata": { + "length": 1572, + "summary": "The passage discusses debt investments accounted for at amortized cost, with allowances for credit losses of RMB3,779 million and RMB2,641 million as of March 31 and September 30, 2025. Reversal of impairment losses were RMB1,117 million and RMB1,103 million for the six months ended September 30, 2024 and 2025. Loans to shareholders of equity method investees totaling RMB5,518 million principal are expected to be repaid through collateral sale, with collateral fair values of RMB4,325 million and RMB6,250 million as of March 31 and September 30, 2025. Carrying amounts approximate fair value due to comparable interest rates. Other treasury investments include fixed deposits, certificates of deposit, and marketable debt securities with maturities over one year.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Debt", + "investments", + "Continued", + "include", + "debt", + "accounted", + "amortized", + "cost", + "allowance", + "credit", + "losses", + "March", + "31", + "September", + "30", + "2025", + "RMB3", + "779", + "million", + "RMB2", + "641", + "Reversal", + "impairment", + "recorded", + "unaudited", + "condensed", + "consolidated", + "income", + "statements", + "RMB1", + "117", + "103", + "months", + "ended", + "2024", + "As", + "repayment", + "loans", + "provided", + "shareholders", + "equity", + "method", + "investees", + "total", + "principal", + "amount", + "RMB5", + "518", + "expected", + "substantially", + "sale", + "collateral", + "Expected", + "assessed", + "individual", + "basis", + "based", + "fair", + "shares", + "pledged", + "reporting", + "date", + "adjusted", + "selling", + "costs", + "The", + "RMB4", + "325", + "RMB6", + "250", + "There", + "commitment", + "lend", + "additional", + "funds", + "carrying", + "approximates", + "due", + "fact", + "related", + "effective", + "interest", + "rates", + "approximate", + "offered", + "financial", + "institutions", + "similar", + "instruments", + "comparable", + "maturities", + "Other", + "treasury", + "comprise", + "fixed", + "deposits", + "certificates", + "marketable", + "securities", + "original", + "year", + "purposes" + ], + "keywords": [ + "debt investments", + "credit losses", + "collateral" + ], + "connect_to": [] + } + }, + { + "chunk_id": "dc88d61e-5448-594f-9a98-323acb562d81", + "type": "text", + "content": "Fair value is defined as the price that would be received to sell an asset or paid to transfer a liability in an orderly transaction between market participants at the measurement date. To increase the comparability of fair value measures, the following hierarchy prioritizes the inputs to valuation methodologies used to measure fair value:\nLevel 1 — Valuations based on unadjusted quoted prices for identical assets and liabilities in active markets.\nLevel 2 — Valuations based on observable inputs other than quoted prices included in Level 1, such as quoted prices for similar assets and liabilities in active markets, quoted prices for identical or similar assets and liabilities in markets that are not active, or other inputs that are observable or can be corroborated by observable market data.\nLevel 3 — Valuations based on unobservable inputs reflecting assumptions, consistent with reasonably available assumptions made by other market participants. These valuations require significant judgment.\nFair value of listed equity investments are based on quoted prices in active markets for identical assets or liabilities and, if applicable, are adjusted for the characteristic included in the equity security. The valuation of unlisted equity investments that do not have a quoted price may include the use of market and income valuation approaches and the use of estimates, which may include discount rates, investees' liquidity and financial performance, and market data of comparable companies in similar industries. Certain other financial instruments, such as interest rate swap contracts and certain option and forward agreements, are valued based on inputs derived from or corroborated by observable market data. Valuations of investments in convertible and exchangeable bonds that do not have a quoted price are generally performed using valuation models such as the binomial model with unobservable inputs including risk-free interest rate and expected volatility. The valuation of the exchangeable bonds is primarily determined based on quoted market price in the over-the-counter market. The valuation of contingent consideration is performed using an expected cash flow method with unobservable inputs including the probability to achieve the contingencies, which is assessed by the Company, in connection with the contingent consideration arrangements. Investments in privately held companies for which the Company elected to record using the measurement alternative are remeasured on a non-recurring basis, and are categorized within Level 3 under the fair value hierarchy. The values are estimated based on valuation methods using the observable transaction price at the transaction date and considering the rights and obligations of the securities and other unobservable inputs including volatility.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/9. Fair value measurement", + "metadata": { + "length": 2823, + "summary": "The passage defines fair value and describes the three-level hierarchy for valuation inputs: Level 1 uses unadjusted quoted prices in active markets for identical assets; Level 2 uses observable inputs like quoted prices for similar assets or inactive markets; Level 3 uses unobservable inputs requiring significant judgment. It details valuation methods for various instruments: listed equities use Level 1 inputs; unlisted equities may use market/income approaches with estimates; interest rate swaps and options use observable market data; convertible bonds use binomial models with unobservable inputs; exchangeable bonds rely on over-the-counter market prices; contingent consideration uses expected cash flow methods; and privately held company investments are measured on a non-recurring basis under Level 3.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Fair", + "defined", + "price", + "received", + "sell", + "asset", + "paid", + "transfer", + "liability", + "orderly", + "transaction", + "market", + "participants", + "measurement", + "date", + "To", + "increase", + "comparability", + "fair", + "measures", + "hierarchy", + "prioritizes", + "inputs", + "valuation", + "methodologies", + "measure", + "Level", + "Valuations", + "based", + "unadjusted", + "quoted", + "prices", + "identical", + "assets", + "liabilities", + "active", + "markets", + "observable", + "included", + "similar", + "corroborated", + "data", + "unobservable", + "reflecting", + "assumptions", + "consistent", + "made", + "These", + "valuations", + "require", + "significant", + "judgment", + "listed", + "equity", + "investments", + "applicable", + "adjusted", + "characteristic", + "security", + "The", + "unlisted", + "include", + "income", + "approaches", + "estimates", + "discount", + "rates", + "investees", + "liquidity", + "financial", + "performance", + "comparable", + "companies", + "industries", + "Certain", + "instruments", + "interest", + "rate", + "swap", + "contracts", + "option", + "forward", + "agreements", + "valued", + "derived", + "convertible", + "exchangeable", + "bonds", + "generally", + "performed", + "models", + "binomial", + "model", + "including", + "risk", + "free", + "expected", + "volatility", + "primarily", + "determined", + "counter", + "contingent", + "consideration", + "cash", + "flow", + "method", + "probability", + "achieve", + "contingencies", + "assessed", + "Company", + "connection", + "arrangements", + "Investments", + "privately", + "held", + "elected", + "record", + "alternative", + "remeasured", + "recurring", + "basis", + "categorized", + "values", + "estimated", + "methods", + "rights", + "obligations", + "securities" + ], + "keywords": [ + "Fair Value", + "Valuation Hierarchy", + "Financial Instruments" + ], + "connect_to": [] + } + }, + { + "chunk_id": "c3ce39e7-dcd4-578d-bf43-376005b76d6c", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_9", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_9/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fee6bc2c-a451-59ab-b938-3120a3571297", + "type": "table", + "content": "
As of March 31, 2025
Level 1RMBLevel 2RMBLevel 3RMBTotalRMB
(in millions)
Assets
Time deposits and certificate of deposits (i)357,569357,569
Wealth management products (i)83,14483,144
Marketable debt securities (i)16,12916,129
Restricted cash and escrow receivables43,78143,781
Listed equity securities (ii)(v)67,7129,91877,630
Convertible and exchangeable bonds (ii)145818963
Option agreements (iii)87814901
Deferred consideration (iii)3,0393,039
Others (vi)1,3296,0477,376
111,493468,32110,718590,532
Liabilities
Contingent consideration in relation to investments and acquisitions (iv)484484
Others (iv)3899241,313
3891,4081,797
", + "path": "tables/table-33 Fair Value Table.html", + "metadata": { + "length": 1535, + "summary": "table-34\nThe table shows assets and liabilities measured at fair value as of March 31, 2025, with totals of 590,532 million RMB for assets and 1,797 million RMB for liabilities, categorized by Level 1, 2, and 3 inputs.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-33 Fair Value Table.html", + "keywords": [ + "fair value", + "assets", + "liabilities" + ], + "tokens": [] + } + }, + { + "chunk_id": "af7f31f6-fbb8-5e7a-8a70-f54833c5d861", + "type": "table", + "content": "
As of September 30, 2025
Level 1RMBLevel 2RMBLevel 3RMBTotalRMB
(in millions)
Assets
Time deposits and certificate of deposits (i)314,794314,794
Wealth management products (i)84,78984,789
Marketable debt securities (i)46,23046,230
Restricted cash and escrow receivables40,37440,374
Listed equity securities (ii)(v)65,3279,97775,304
Convertible and exchangeable bonds (ii)139620759
Option and forward agreements (iii)3,4781,1544,632
Deferred consideration (iii)3,0523,052
Others (vi)8455,6276,472
105,701460,25210,453576,406
Liabilities
Exchangeable bonds13,75513,755
Contingent consideration in relation to investments and acquisitions (iv)397397
Others (iv)1,1229072,029
14,8771,30416,181
", + "path": "tables/table-34 Fair Value Table.html", + "metadata": { + "length": 1696, + "summary": "table-35\nTable shows assets and liabilities measured at fair value as of September 30, 2025, with totals of 576,406 million RMB for assets and 16,181 million RMB for liabilities, categorized by Level 1, 2, and 3 inputs.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-34 Fair Value Table.html", + "keywords": [ + "fair value", + "assets", + "liabilities" + ], + "tokens": [] + } + }, + { + "chunk_id": "611497b4-d515-5152-9680-11789921fbad", + "type": "text", + "content": "The following table summarizes the Company's assets and liabilities that are measured at fair value on a recurring basis and are categorized under the fair value hierarchy:\n\n[tables/table-33 Fair Value Table.html]\n\n\n[tables/table-34 Fair Value Table.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_9/9. Fair value measurement (Continued)", + "metadata": { + "length": 266, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "table", + "summarizes", + "Company", + "assets", + "liabilities", + "measured", + "fair", + "recurring", + "basis", + "categorized", + "hierarchy" + ], + "keywords": [], + "connect_to": [ + { + "target": "fee6bc2c-a451-59ab-b938-3120a3571297", + "relation": "embeds", + "ref": "[tables/table-33 Fair Value Table.html]", + "position": { + "start": 174, + "end": 213 + } + }, + { + "target": "af7f31f6-fbb8-5e7a-8a70-f54833c5d861", + "relation": "embeds", + "ref": "[tables/table-34 Fair Value Table.html]", + "position": { + "start": 216, + "end": 255 + } + } + ] + } + }, + { + "chunk_id": "66eafa59-dd45-5abd-a82b-936c1b36a809", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_10", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_10/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "bdda5c98-eded-5e74-87c1-c49e7694cea0", + "type": "table", + "content": "
Amounts RMB (in millions)
Balance as of April 1, 2025818
Additions42
Net decrease in fair value(158)
Disposal(39)
Conversion(34)
Foreign currency translation adjustments(9)
Balance as of September 30, 2025620
", + "path": "tables/table-35 Balance Change.html", + "metadata": { + "length": 432, + "summary": "table-36\nBalance decreased from 818 to 620 million RMB between April and September 2025, with additions of 42, net fair value decrease of 158, disposals of 39, conversions of 34, and currency adjustments of 9.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-35 Balance Change.html", + "keywords": [ + "balance", + "fair value", + "conversion" + ], + "tokens": [] + } + }, + { + "chunk_id": "544a0505-d888-5655-99f3-68d46fc987c5", + "type": "table", + "content": "
Amounts RMB (in millions)
Balance as of April 1, 20253,039
Net increase in fair value40
Foreign currency translation adjustments(27)
Balance as of September 30, 20253,052
", + "path": "tables/table-36 Fair Value Change.html", + "metadata": { + "length": 316, + "summary": "table-37\nBalance increased from 3,039 to 3,052 million RMB between April and September 2025, with a net fair value increase of 40 and a currency translation loss of 27.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-36 Fair Value Change.html", + "keywords": [ + "fair value", + "balance", + "currency" + ], + "tokens": [] + } + }, + { + "chunk_id": "74114f0c-6168-53b2-8e75-882026ec0b77", + "type": "text", + "content": "(i) Included in short-term investments and equity securities and other investments on the unaudited condensed consolidated balance sheets.\n(ii) Included in equity securities and other investments on the unaudited condensed consolidated balance sheets.\n(iii) Included in prepayments, receivables and other assets on the unaudited condensed consolidated balance sheets.\n(iv) Included in accrued expenses, accounts payable and other liabilities on the unaudited condensed consolidated balance sheets.\n(v) As of March 31 and September 30, 2025, listed equity securities with fair value of RMB11,921 million and RMB10,415 million were subject to contractual sale restrictions, respectively.\n(vi) Others primarily represent other investments with underlying assets measured at fair value.\nConvertible and exchangeable bonds investments categorized within Level 3 under the fair value hierarchy:\n\n[tables/table-35 Balance Change.html]\n\nDeferred consideration categorized within Level 3 under the fair value hierarchy:\n\n[tables/table-36 Fair Value Change.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_10/9. Fair value measurement (Continued)", + "metadata": { + "length": 1067, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Included", + "short", + "term", + "investments", + "equity", + "securities", + "unaudited", + "condensed", + "consolidated", + "balance", + "sheets", + "ii", + "iii", + "prepayments", + "receivables", + "assets", + "iv", + "accrued", + "expenses", + "accounts", + "payable", + "liabilities", + "As", + "March", + "31", + "September", + "30", + "2025", + "listed", + "fair", + "RMB11", + "921", + "million", + "RMB10", + "415", + "subject", + "contractual", + "sale", + "restrictions", + "vi", + "Others", + "primarily", + "represent", + "underlying", + "measured", + "Convertible", + "exchangeable", + "bonds", + "categorized", + "Level", + "hierarchy", + "Deferred", + "consideration" + ], + "keywords": [], + "connect_to": [ + { + "target": "bdda5c98-eded-5e74-87c1-c49e7694cea0", + "relation": "embeds", + "ref": "[tables/table-35 Balance Change.html]", + "position": { + "start": 890, + "end": 927 + } + }, + { + "target": "544a0505-d888-5655-99f3-68d46fc987c5", + "relation": "embeds", + "ref": "[tables/table-36 Fair Value Change.html]", + "position": { + "start": 1012, + "end": 1052 + } + } + ] + } + }, + { + "chunk_id": "4c9dcbde-b694-5545-94fb-1dc3d518a609", + "type": "table", + "content": "
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
User base and customer relationships48,56540,420
Trade names, trademarks and domain names26,93627,256
Non-compete agreements6,0305,745
Developed technology and patents4,8235,136
Licensed copyrights and others8,0018,585
94,35587,142
Less: accumulated amortization and impairment(73,444)(67,713)
Net book value20,91119,429
", + "path": "tables/table-37 Intangible Assets.html", + "metadata": { + "length": 809, + "summary": "table-38\nThe table shows intangible assets as of March 31 and September 30, 2025, including user base, trademarks, non-compete agreements, developed technology, and licensed copyrights, with total cost, accumulated amortization, and net book value.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-37 Intangible Assets.html", + "keywords": [ + "intangible assets", + "amortization", + "net book value" + ], + "tokens": [] + } + }, + { + "chunk_id": "736daa1c-7a17-57ec-997f-a4037d9df4eb", + "type": "text", + "content": "[tables/table-37 Intangible Assets.html]\n\nTotal amortization expenses recognized for the six months ended September 30, 2024 and 2025 amounted to RMB7,122 million and RMB5,111 million, respectively, including the portion relating to licensed copyrights of RMB3,689 million and RMB3,478 million which were recorded in cost of revenue for the six months ended September 30, 2024 and 2025, respectively.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_10/10. Intangible assets, net", + "metadata": { + "length": 418, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Total", + "amortization", + "expenses", + "recognized", + "months", + "ended", + "September", + "30", + "2024", + "2025", + "amounted", + "RMB7", + "122", + "million", + "RMB5", + "111", + "including", + "portion", + "relating", + "licensed", + "copyrights", + "RMB3", + "689", + "478", + "recorded", + "cost", + "revenue" + ], + "keywords": [], + "connect_to": [ + { + "target": "4c9dcbde-b694-5545-94fb-1dc3d518a609", + "relation": "embeds", + "ref": "[tables/table-37 Intangible Assets.html]", + "position": { + "start": 0, + "end": 40 + } + } + ] + } + }, + { + "chunk_id": "c5f8c9ca-00ad-50c4-9eec-5fbcc41286f2", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9b9ddceb-a453-5d04-8e13-ef22a3ff13f2", + "type": "table", + "content": "
AlibabaChinaE-commerceGroupRMBAlibabaInternationalDigitalCommerceGroupRMBCloudIntelligenceGroupRMBCainiaoSmartLogisticsNetworkLimitedRMBLocalServicesGroupRMBHujingDigital MediaandEntertainmentGroupRMBAll othersRMBTotalRMB
(in millions)
Balance as of April 1, 2025164,94520,0773,72820,38520,4478,03117,888255,501
Segment changes12,948(20,385)(20,447)(8,031)35,915
Additions435140
Foreign currency translationadjustments(42)46610
Balance as of September 30, 2025177,89320,0393,80953,810255,551
", + "path": "tables/table-38 Alibaba Segment Assets.html", + "metadata": { + "length": 1096, + "summary": "table-39\nAlibaba's segment asset balances from April to September 2025 show a total of 255,551 million RMB, with major changes including the transfer of Cainiao, Local Services, and Hujing to 'All others'.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-38 Alibaba Segment Assets.html", + "keywords": [ + "Segment Assets", + "Balance Sheet", + "Alibaba" + ], + "tokens": [] + } + }, + { + "chunk_id": "9e0b6fd9-805f-5659-919b-bfc9d0306647", + "type": "text", + "content": "Changes in the carrying amount of goodwill by segment for the six months ended September 30, 2025 were as follows:\n\n[tables/table-38 Alibaba Segment Assets.html]\n\nGross goodwill balances were RMB302,194 million and RMB302,244 million as of March 31 and September 30, 2025, respectively. Accumulated impairment losses were RMB46,693 million and RMB46,693 million as of March 31 and September 30, 2025, respectively.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/11. Goodwill", + "metadata": { + "length": 472, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Changes", + "carrying", + "amount", + "goodwill", + "segment", + "months", + "ended", + "September", + "30", + "2025", + "Gross", + "balances", + "RMB302", + "194", + "million", + "244", + "March", + "31", + "Accumulated", + "impairment", + "losses", + "RMB46", + "693" + ], + "keywords": [], + "connect_to": [ + { + "target": "9b9ddceb-a453-5d04-8e13-ef22a3ff13f2", + "relation": "embeds", + "ref": "[tables/table-38 Alibaba Segment Assets.html]", + "position": { + "start": 116, + "end": 161 + } + } + ] + } + }, + { + "chunk_id": "d273f37f-9810-5d1a-ad82-dab8a03b1289", + "type": "table", + "content": "
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
Accrued expenses, accounts payable and other liabilities1,6053,152
Bank borrowings4,4704,047
6,0757,199
", + "path": "tables/table-39 Financial Liabilities.html", + "metadata": { + "length": 426, + "summary": "table-40\nThe table shows accrued expenses, accounts payable, and bank borrowings as of March 31 and September 30, 2025, in millions of RMB, with totals of 6,075 and 7,199 respectively.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-39 Financial Liabilities.html", + "keywords": [ + "accrued expenses", + "bank borrowings", + "liabilities" + ], + "tokens": [] + } + }, + { + "chunk_id": "38860174-1536-51a8-b4aa-a3a1f97f7403", + "type": "text", + "content": "The Company enters into agreements with several financial institutions and offer supplier finance programs to the Company's suppliers. Suppliers can sell one or more of the Company's payment obligations at their sole discretion to the financial institutions to receive funds, usually at a discounted price, prior to the scheduled due dates to meet their cash flow needs. The Company's current payment terms with the majority of suppliers are up to 180 days. Generally, the Company's rights and obligations are not impacted and the original payment terms, timing or amount, remain unchanged. Except for the pledge of other treasury investments with carrying value of RMB1,500 million and nil as of March 31 and September 30, 2025, respectively, the Company did not provide assets pledged as security or other forms of guarantees under these supplier finance programs.\nThe outstanding payment obligations under these supplier finance programs are generally recorded within accrued expenses, accounts payable and other liabilities on the unaudited condensed consolidated balance sheets, except for certain arrangements in which the Company pays the discount to the financial institutions on behalf of the suppliers which are recorded within current bank borrowings. The respective balances are as follows:\n\n[tables/table-39 Financial Liabilities.html]\n\nALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/12. Supplier Finance Programs", + "metadata": { + "length": 1509, + "summary": "The Company uses supplier finance programs with financial institutions, allowing suppliers to sell payment obligations at a discount for early cash. Payment terms are up to 180 days, and the Company's rights and obligations remain unchanged. As of March 31 and September 30, 2025, no assets were pledged except for RMB1,500 million in treasury investments. Outstanding obligations are recorded in accrued expenses, accounts payable, and other liabilities, except for certain discount payments recorded as current bank borrowings.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "Company", + "enters", + "agreements", + "financial", + "institutions", + "offer", + "supplier", + "finance", + "programs", + "suppliers", + "Suppliers", + "sell", + "payment", + "obligations", + "sole", + "discretion", + "receive", + "funds", + "discounted", + "price", + "prior", + "scheduled", + "due", + "dates", + "meet", + "cash", + "flow", + "current", + "terms", + "majority", + "180", + "days", + "Generally", + "rights", + "impacted", + "original", + "timing", + "amount", + "remain", + "unchanged", + "Except", + "pledge", + "treasury", + "investments", + "carrying", + "RMB1", + "500", + "million", + "nil", + "March", + "31", + "September", + "30", + "2025", + "provide", + "assets", + "pledged", + "security", + "forms", + "guarantees", + "outstanding", + "generally", + "recorded", + "accrued", + "expenses", + "accounts", + "payable", + "liabilities", + "unaudited", + "condensed", + "consolidated", + "balance", + "sheets", + "arrangements", + "pays", + "discount", + "behalf", + "bank", + "borrowings", + "respective", + "balances", + "ALIBABA", + "GROUP", + "HOLDING", + "LIMITED", + "NOTES", + "TO", + "UNAUDITED", + "CONDENSED", + "CONSOLIDATED", + "FINANCIAL", + "STATEMENTS", + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "2024" + ], + "keywords": [ + "supplier finance", + "payment obligations", + "financial institutions" + ], + "connect_to": [ + { + "target": "d273f37f-9810-5d1a-ad82-dab8a03b1289", + "relation": "embeds", + "ref": "[tables/table-39 Financial Liabilities.html]", + "position": { + "start": 1304, + "end": 1348 + } + } + ] + } + }, + { + "chunk_id": "e9e7a86a-d932-5152-8f67-50f0d12e93fd", + "type": "text", + "content": "In May 2024, the Company issued convertible unsecured senior notes for an aggregate principal amount of US\\$5.0 billion due on June 1, 2031 (the “2024 Convertible Senior Notes”). The 2024 Convertible Senior Notes are senior unsecured obligations, and interest at an annual rate of 0.5% is payable in arrears semiannually. The 2024 Convertible Senior Notes may be converted into the Company’s ADSs, at the option of holders, at any time prior to the maturity date at an initial conversion rate of 9.5202 ADSs per US\\$1,000 principal amount.\nIn September 2025, the Company issued zero coupon convertible unsecured senior notes for an aggregate principal amount of approximately US\\$3.2 billion due on September 15, 2032 (the “2025 Convertible Senior Notes”). The 2025 Convertible Senior Notes are senior unsecured obligations. The 2025 Convertible Senior Notes may be converted into the Company’s ADSs, at the option of holders, at any time from March 15, 2032 until maturity at an initial conversion rate of 5.1773 ADSs per US\\$1,000 principal amount, and may be convertible prior to March 15, 2032 only upon satisfaction of certain conditions.\nThe initial conversion rates are subject to adjustment in some events such as dividend distribution. In addition, in the event of a fundamental change that occurs prior to the respective maturity dates or following the Company's delivery of a notice of redemption, the Company will increase the initial conversion rates respectively, which shall not exceed 12.3762 ADSs per US\\$1,000 principal amount for the 2024 Convertible Senior Notes and 6.7953 ADSs per US\\$1,000 principal amount for the 2025 Convertible Senior Notes, for a holder who elects to convert its notes in connection with such a fundamental change or such notice of redemption. Such make-whole adjustments are subject to the same adjustments as the respective initial conversion rates noted above. Upon conversion, the Company will pay or deliver, at its election, cash, ADSs, or a combination of cash and ADSs. Holders may also elect to receive ordinary shares in lieu of any ADSs deliverable upon conversion, with each ADS representing eight ordinary shares.\nAs of September 30, 2025, the adjusted conversion rate for the 2024 Convertible Senior Notes was 9.8915 ADSs per US\\$1,000 principal amount, and the adjusted conversion rate taking into account the make-whole adjustments was 12.8589 ADSs per US\\$1,000 principal amount. As of September 30, 2025, the conversion rate for the 2025 Convertible Senior Notes remained unadjusted at its initial conversion rate.\nThe Company may redeem for cash all but not part of the respective convertible senior notes in the event of certain tax law changes, or at any time if less than $10\\%$ of the aggregate principal amount of the respective convertible senior notes originally issued remains outstanding. The Company may also redeem for cash all or part of the 2024 Convertible Senior Notes and the 2025 Convertible Senior Notes on or after June 8, 2029 and September 20, 2030, respectively, provided that the Company's ADS price has been at least $130\\%$ of the then effective conversion price for a specific period of time and on the specified date. The redemption price will be equal to the principal amount of the notes being redeemed plus accrued and unpaid interest, if any, to, but excluding, the related redemption date.\nHolders have the right to require the Company to repurchase for cash all or part of the 2024 Convertible Senior Notes and the 2025 Convertible Senior Notes on June 1, 2029 and September 15, 2030, respectively, or in the event of a fundamental change, subject to certain conditions. The repurchase price will be equal to the principal amount of the notes being repurchased plus accrued and unpaid interest, if any, to, but excluding, the related repurchase date.\nAs of March 31 and September 30, 2025, the unamortized debt discounts and debt issuance costs of the 2024 Convertible Senior Notes were US\\$58 million (RMB424 million) and US\\$52 million (RMB367 million), respectively, and the fair value of the 2024 Convertible Senior Notes, based on level 2 inputs, was US\\$7,151 million (RMB51,854 million) and US\\$9,234 million (RMB65,733 million), respectively.\nAs of September 30, 2025, the unamortized debt discounts and debt issuance costs of the 2025 Convertible Senior Notes were US\\$42 million (RMB296 million), and the fair value of the 2025 Convertible Senior Notes, based on level 2 inputs, was US\\$3,626 million (RMB25,812 million).\nFor the six months ended September 30, 2024 and 2025, the effective interest rate for the 2024 Convertible Senior Notes was approximately 0.8%. For the six months ended September 30, 2025, the effective interest rate for the 2025 Convertible Senior Notes was approximately 0.3%.\nALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/13. Convertible unsecured senior notes", + "metadata": { + "length": 4953, + "summary": "The passage details two convertible senior note issuances by Alibaba Group: $5.0 billion in 2024 notes due 2031 with 0.5% interest and initial conversion rate of 9.5202 ADSs per $1,000, and $3.2 billion in 2025 zero-coupon notes due 2032 with initial conversion rate of 5.1773 ADSs per $1,000. Both are senior unsecured obligations convertible into ADSs, with adjustment provisions for dividends and fundamental changes. As of September 30, 2025, the 2024 notes had an adjusted conversion rate of 9.8915 ADSs, while the 2025 notes remained unadjusted. The company may redeem notes under certain conditions, and holders can require repurchase on specific dates or upon fundamental changes. Unamortized costs and fair values are provided for both notes as of March and September 2025, with effective interest rates of 0.8% for 2024 notes and 0.3% for 2025 notes.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "In", + "May", + "2024", + "Company", + "issued", + "convertible", + "unsecured", + "senior", + "notes", + "aggregate", + "principal", + "amount", + "US", + "5.0", + "billion", + "due", + "June", + "2031", + "Convertible", + "Senior", + "Notes", + "The", + "obligations", + "interest", + "annual", + "rate", + "0.5%", + "payable", + "arrears", + "semiannually", + "converted", + "ADSs", + "option", + "holders", + "time", + "prior", + "maturity", + "date", + "initial", + "conversion", + "9.5202", + "000", + "September", + "2025", + "coupon", + "approximately", + "3.2", + "15", + "2032", + "March", + "5.1773", + "satisfaction", + "conditions", + "rates", + "subject", + "adjustment", + "events", + "dividend", + "distribution", + "addition", + "event", + "fundamental", + "change", + "occurs", + "respective", + "dates", + "delivery", + "notice", + "redemption", + "increase", + "exceed", + "12.3762", + "6.7953", + "holder", + "elects", + "convert", + "connection", + "Such", + "make", + "adjustments", + "noted", + "Upon", + "pay", + "deliver", + "election", + "cash", + "combination", + "Holders", + "elect", + "receive", + "ordinary", + "shares", + "lieu", + "deliverable", + "ADS", + "representing", + "As", + "30", + "adjusted", + "9.8915", + "taking", + "account", + "12.8589", + "remained", + "unadjusted", + "redeem", + "part", + "tax", + "law", + "10", + "originally", + "remains", + "outstanding", + "2029", + "20", + "2030", + "provided", + "price", + "130", + "effective", + "specific", + "period", + "equal", + "redeemed", + "accrued", + "unpaid", + "excluding", + "related", + "require", + "repurchase", + "repurchased", + "31", + "unamortized", + "debt", + "discounts", + "issuance", + "costs", + "58", + "million", + "RMB424", + "52", + "RMB367", + "fair", + "based", + "level", + "inputs", + "151", + "RMB51", + "854", + "234", + "RMB65", + "733", + "42", + "RMB296", + "626", + "RMB25", + "812", + "For", + "months", + "ended", + "0.8%", + "0.3%", + "ALIBABA", + "GROUP", + "HOLDING", + "LIMITED", + "NOTES", + "TO", + "UNAUDITED", + "CONDENSED", + "CONSOLIDATED", + "FINANCIAL", + "STATEMENTS", + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER" + ], + "keywords": [ + "Convertible Senior Notes", + "Alibaba", + "ADS" + ], + "connect_to": [] + } + }, + { + "chunk_id": "60185086-2fc0-5bb8-9f03-c2cddd0c8921", + "type": "text", + "content": "In connection with the issuance of the convertible senior notes, the Company entered into capped call transactions with certain financial institutions at a cost of US\\$638 million (RMB4,612 million) and US\\$184 million (RMB1,309 million) for the 2024 Convertible Senior Notes and the 2025 Convertible Senior Notes, respectively, which are expected to reduce potential dilution and/or offset cash payments upon conversion. The cap prices of the capped call transactions for the 2024 Convertible Senior Notes and the 2025 Convertible Senior Notes are initially US\\$161.60 per ADS and US\\$235.46 per ADS, respectively. The capped prices are subject to adjustments similar to the adjustments on the conversion rates of the respective convertible senior notes. The capped call transactions may be settled in cash at the Company’s election.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/13. Convertible unsecured senior notes (Continued)", + "metadata": { + "length": 834, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "In", + "connection", + "issuance", + "convertible", + "senior", + "notes", + "Company", + "entered", + "capped", + "call", + "transactions", + "financial", + "institutions", + "cost", + "US", + "638", + "million", + "RMB4", + "612", + "184", + "RMB1", + "309", + "2024", + "Convertible", + "Senior", + "Notes", + "2025", + "expected", + "reduce", + "potential", + "dilution", + "offset", + "cash", + "payments", + "conversion", + "The", + "cap", + "prices", + "initially", + "161.60", + "ADS", + "235.46", + "subject", + "adjustments", + "similar", + "rates", + "respective", + "settled", + "election" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "499795f3-8bf6-5200-9d73-51cbae770124", + "type": "text", + "content": "In July 2025, the Company issued HKD denominated zero coupon exchangeable bonds referencing the ordinary shares of Alibaba Health Information Technology Limited (“Alibaba Health”), a subsidiary of the Company in which the Company holds approximately 64% of equity interest for an aggregate principal amount of approximately HK\\$12 billion (approximately RMB11 billion) due on July 9, 2032 (the “2025 Exchangeable Bonds”). The ordinary shares of Alibaba Health are listed on the Hong Kong Stock Exchange (“AH Shares”). The 2025 Exchangeable Bonds are unsecured and unsubordinated obligations of the Company and are listed on the Vienna MTF operated by the Vienna Stock Exchange.\nThe 2025 Exchangeable Bonds may be exchanged into the AH Shares, at the option of holders, at any time prior to the maturity date at an initial exchange rate of approximately 160,513.6 AH Shares per HK\\$1,000,000 principal amount.\nThe initial exchange rate is subject to adjustment in some events such as dividend distribution by Alibaba Health. In addition, in a relevant event such as the delisting of AH Shares occurring prior to the maturity date or following the Company's delivery of a notice of redemption, the Company will increase the initial exchange rate, which shall not exceed approximately 237,529.7 AH Shares per HK\\$1,000,000 principal amount, for a holder who elects to exchange its bonds in connection with such a relevant event or such notice of redemption. Such make-whole adjustment is subject to the same adjustments as the initial exchange rate noted above. Upon exchange, the Company will pay or deliver, at its election, cash, AH Shares, or a combination of cash and AH Shares.\nAs of September 30, 2025, the exchange rate for the 2025 Exchangeable bonds remained unadjusted at its initial exchange rate.\nThe Company may redeem for cash all but not part of the 2025 Exchangeable Bonds in the event of certain tax law changes, or at any time if less than 10% of the aggregate principal amount of the 2025 Exchangeable Bonds originally issued remains outstanding. The Company may also redeem for cash all or part of the 2025 Exchangeable Bonds on or after July 9, 2030, provided that the AH Shares price has been at least 130% of the then effective exchange price for a specific period of time. The redemption price will be equal to the principal amount of the bonds being redeemed.\nHolders have the right to require the Company to repurchase for cash all or part of the 2025 Exchangeable Bonds on July 9, 2030, or in a relevant event, subject to certain conditions. The repurchase price will be equal to the principal amount of the bonds being repurchased.\nIn connection with the issuance of the 2025 Exchangeable Bonds, the Company entered into a stock borrowing and lending arrangement with an affiliate of one of the bookrunners (the “Borrower”), pursuant to which the Company has committed to lending a certain number of AH Shares, which shall not exceed the number of AH Shares exchangeable under the 2025 Exchangeable Bonds, to the Borrower to facilitate hedging activities of certain bondholders. If a termination event occurs and the Borrower is unable to deliver the AH Shares due to legal restrictions, force majeure, or market disruption, cash settlement by the Borrower would be required. As of September 30, 2025, the fair value of the outstanding AH Shares lent under the arrangement was HK\\$8,492 million (RMB7,770 million).\nFor the six months ended September 30, 2025, losses of RMB327 million arising from changes in the fair value of the 2025 Exchangeable Bonds and the stock borrowing and lending arrangement were recorded in the unaudited condensed consolidated income statements.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/14. Exchangeable bonds", + "metadata": { + "length": 3701, + "summary": "In July 2025, the Company issued HKD-denominated zero coupon exchangeable bonds due 2032, referencing Alibaba Health shares, with an aggregate principal of HK$12 billion. The bonds are exchangeable into AH Shares at an initial rate of 160,513.6 shares per HK$1,000,000, subject to adjustments. The Company may redeem or repurchase the bonds under certain conditions. A stock borrowing and lending arrangement was entered into for hedging, with fair value of lent shares at HK$8,492 million as of September 30, 2025. Losses of RMB327 million from fair value changes were recorded for the six months ended September 30, 2025.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "In", + "July", + "2025", + "Company", + "issued", + "HKD", + "denominated", + "coupon", + "exchangeable", + "bonds", + "referencing", + "ordinary", + "shares", + "Alibaba", + "Health", + "Information", + "Technology", + "Limited", + "subsidiary", + "holds", + "approximately", + "64%", + "equity", + "interest", + "aggregate", + "principal", + "amount", + "HK", + "12", + "billion", + "RMB11", + "due", + "2032", + "Exchangeable", + "Bonds", + "The", + "listed", + "Hong", + "Kong", + "Stock", + "Exchange", + "AH", + "Shares", + "unsecured", + "unsubordinated", + "obligations", + "Vienna", + "MTF", + "operated", + "exchanged", + "option", + "holders", + "time", + "prior", + "maturity", + "date", + "initial", + "exchange", + "rate", + "160", + "513.6", + "000", + "subject", + "adjustment", + "events", + "dividend", + "distribution", + "addition", + "relevant", + "event", + "delisting", + "occurring", + "delivery", + "notice", + "redemption", + "increase", + "exceed", + "237", + "529.7", + "holder", + "elects", + "connection", + "Such", + "make", + "adjustments", + "noted", + "Upon", + "pay", + "deliver", + "election", + "cash", + "combination", + "As", + "September", + "30", + "remained", + "unadjusted", + "redeem", + "part", + "tax", + "law", + "10%", + "originally", + "remains", + "outstanding", + "2030", + "provided", + "price", + "130%", + "effective", + "specific", + "period", + "equal", + "redeemed", + "Holders", + "require", + "repurchase", + "conditions", + "repurchased", + "issuance", + "entered", + "stock", + "borrowing", + "lending", + "arrangement", + "affiliate", + "bookrunners", + "Borrower", + "pursuant", + "committed", + "number", + "facilitate", + "hedging", + "activities", + "bondholders", + "If", + "termination", + "occurs", + "unable", + "legal", + "restrictions", + "force", + "majeure", + "market", + "disruption", + "settlement", + "required", + "fair", + "lent", + "492", + "million", + "RMB7", + "770", + "For", + "months", + "ended", + "losses", + "RMB327", + "arising", + "recorded", + "unaudited", + "condensed", + "consolidated", + "income", + "statements" + ], + "keywords": [ + "Exchangeable Bonds", + "Alibaba Health", + "Hedging" + ], + "connect_to": [] + } + }, + { + "chunk_id": "62351f79-83f9-51e8-b71f-505076785868", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "523c4b2c-1636-5e72-a06e-6f7029e05a97", + "type": "text", + "content": "The Company has made a series of amendments to the syndicated loan facility. Subsequent to the partial repayment of US\\$830 million in January 2025, the size of the US\\$4.0 billion syndicated loan was reduced to US\\$3.17 billion. In September 2025, the Company amended the loan facility and reduced the pricing terms to Secured Overnight Financing Rate (“SOFR”) plus 66 basis points. Effective on November 28, 2025, the facility was restructured as a revolving credit facility with drawdowns permitted in both U.S. dollars and Hong Kong dollars, and the expiration date of the facility was extended to September 30, 2028, with an option to further extend to September 30, 2030. The interest rate of the credit facility is 66 basis points over SOFR or Hong Kong Interbank Offered Rate (“HIBOR”), and the margin will be 81 basis points for the optional extension period. Subsequently, the Company submitted a prepayment notice to repay the outstanding balance of US\\$3.17 billion under the revolving credit facility.\nIn September 2025, the Company amended the terms of a US\\$6.5 billion revolving credit facility agreement. The size of the credit facility was amended to US\\$3.33 billion and the utilization currency was also amended from U.S. dollar only to both U.S. dollar and Hong Kong dollar. The interest rate of the credit facility was adjusted to SOFR or HIBOR plus 66 basis points. The expiration date of the credit facility was extended from June 24, 2026 to September 30, 2028, with an option to further extend to September 30, 2030 and the margin will be 81 basis points for the optional extension period. The Company has not yet drawn down this facility.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/15. Bank borrowings", + "metadata": { + "length": 1665, + "summary": "The Company amended its syndicated loan facility, reducing it from $4.0 billion to $3.17 billion after a partial repayment. In September 2025, pricing was reduced to SOFR plus 66 basis points. The facility was restructured as a revolving credit facility on November 28, 2025, allowing drawdowns in USD and HKD, with the expiration extended to September 30, 2028, and an option to extend to September 30, 2030. Interest is SOFR or HIBOR plus 66 basis points, rising to 81 basis points during the optional extension. The Company then prepaid the $3.17 billion outstanding balance. Separately, a $6.5 billion revolving credit facility was amended to $3.33 billion, allowing USD and HKD utilization, with interest at SOFR or HIBOR plus 66 basis points, and expiration extended to September 30, 2028, with an optional extension to September 30, 2030 at 81 basis points. This facility has not been drawn.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "Company", + "made", + "series", + "amendments", + "syndicated", + "loan", + "facility", + "Subsequent", + "partial", + "repayment", + "US", + "830", + "million", + "January", + "2025", + "size", + "4.0", + "billion", + "reduced", + "3.17", + "In", + "September", + "amended", + "pricing", + "terms", + "Secured", + "Overnight", + "Financing", + "Rate", + "SOFR", + "66", + "basis", + "points", + "Effective", + "November", + "28", + "restructured", + "revolving", + "credit", + "drawdowns", + "permitted", + "dollars", + "Hong", + "Kong", + "expiration", + "date", + "extended", + "30", + "2028", + "option", + "extend", + "2030", + "interest", + "rate", + "Interbank", + "Offered", + "HIBOR", + "margin", + "81", + "optional", + "extension", + "period", + "Subsequently", + "submitted", + "prepayment", + "notice", + "repay", + "outstanding", + "balance", + "6.5", + "agreement", + "3.33", + "utilization", + "currency", + "dollar", + "adjusted", + "June", + "24", + "2026", + "drawn" + ], + "keywords": [ + "Loan Amendments", + "Interest Rate", + "Revolving Credit" + ], + "connect_to": [] + } + }, + { + "chunk_id": "b2603e11-998c-553b-bf44-03b283e6ed4c", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMB
(in millions)
Amounts earned by the Company
Cloud services revenue (i)4,9778,928
Marketplace software technology services fee and other amounts earned (i)3,5142,819
8,49111,747
Amounts incurred by the Company
Payment processing and escrow services fee (ii)6,3818,937
Other amounts incurred (i)1,4492,378
7,83011,315
", + "path": "tables/table-40 Cloud Revenue Growth.html", + "metadata": { + "length": 789, + "summary": "table-41\nThe table shows six-month financial data ending September 30, 2024 and 2025. Cloud services revenue rose from 4,977 to 8,928 million RMB, while marketplace fees decreased. Total amounts earned increased from 8,491 to 11,747 million RMB.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-40 Cloud Revenue Growth.html", + "keywords": [ + "cloud services", + "marketplace fees", + "payment processing" + ], + "tokens": [] + } + }, + { + "chunk_id": "976143be-850e-5b5d-9e99-cbe37c4d7d93", + "type": "text", + "content": "During the six months ended September 30, 2024 and 2025, other than disclosed elsewhere, the Company had the following material related party transactions:\nTransactions with Ant Group and its affiliates\n\n[tables/table-40 Cloud Revenue Growth.html]\n\n(i) The Company has other commercial arrangements and cost sharing arrangements with Ant Group and its affiliates on various sales and marketing, cloud, and other administrative and support services.\n(ii) The Company has a commercial agreement with Alipay.com Co., Ltd., a wholly-owned subsidiary of Ant Group (“Alipay”), whereby the Company receives payment processing and escrow services in exchange for a payment for the services fee, which was recognized in cost of revenue.\nAs of March 31 and September 30, 2025, the Company had certain amounts of cash held in accounts managed by Alipay in connection with the provision of online and mobile commerce and related services for a total amount of RMB5,863 million and RMB6,379 million, respectively, which have been classified as cash and cash equivalents on the unaudited condensed consolidated balance sheets.\nALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/16. Related party transactions", + "metadata": { + "length": 1261, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "During", + "months", + "ended", + "September", + "30", + "2024", + "2025", + "disclosed", + "Company", + "material", + "related", + "party", + "transactions", + "Transactions", + "Ant", + "Group", + "affiliates", + "The", + "commercial", + "arrangements", + "cost", + "sharing", + "sales", + "marketing", + "cloud", + "administrative", + "support", + "services", + "ii", + "agreement", + "Alipay", + "Co", + "Ltd", + "wholly", + "owned", + "subsidiary", + "receives", + "payment", + "processing", + "escrow", + "exchange", + "fee", + "recognized", + "revenue", + "As", + "March", + "31", + "amounts", + "cash", + "held", + "accounts", + "managed", + "connection", + "provision", + "online", + "mobile", + "commerce", + "total", + "amount", + "RMB5", + "863", + "million", + "RMB6", + "379", + "classified", + "equivalents", + "unaudited", + "condensed", + "consolidated", + "balance", + "sheets", + "ALIBABA", + "GROUP", + "HOLDING", + "LIMITED", + "NOTES", + "TO", + "UNAUDITED", + "CONDENSED", + "CONSOLIDATED", + "FINANCIAL", + "STATEMENTS", + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER" + ], + "keywords": [], + "connect_to": [ + { + "target": "b2603e11-998c-553b-bf44-03b283e6ed4c", + "relation": "embeds", + "ref": "[tables/table-40 Cloud Revenue Growth.html]", + "position": { + "start": 204, + "end": 247 + } + } + ] + } + }, + { + "chunk_id": "7325c87e-0d01-5420-8f36-99eb27f13ed0", + "type": "text", + "content": "Transactions with other investees\nThe Company has commercial arrangements with certain investees of the Company related to cloud services. In connection with these services provided by the Company, RMB1,058 million and RMB3,854 million were recorded in revenue in the unaudited condensed consolidated income statements for the six months ended September 30, 2024 and 2025, respectively.\nThe Company has commercial arrangements with certain investees of the Company related to marketing services. In connection with these services provided to the Company, RMB358 million and RMB864 million were recorded in cost of revenue and sales and marketing expenses in the unaudited condensed consolidated income statements for the six months ended September 30, 2024 and 2025, respectively.\nThe Company has commercial arrangements with certain investees of the Company related to logistics services. In connection with these services provided by the Company, RMB2,092 million and RMB1,680 million were recorded in revenue in the unaudited condensed consolidated income statements for the six months ended September 30, 2024 and 2025, respectively. Costs and expenses incurred in connection with these services provided to the Company of RMB7,999 million and RMB8,251 million were recorded in the unaudited condensed consolidated income statements for the same periods, respectively.\nThe Company has extended loans to certain investees for working capital and other uses in conjunction with the Company's investments. As of March 31 and September 30, 2025, the aggregate outstanding balance of these loans was RMB1,771 million and RMB1,992 million, respectively, with remaining terms of up to 5 years and interest rates of up to $10\\%$ per annum as of March 31, 2025, and remaining terms of up to 6 years and interest rates of up to $10\\%$ per annum as of September 30, 2025.\nThe Company provided a guarantee for a term loan facility of HK\\$7.7 billion in favor of Hong Kong Cingleot Investment Management Limited (“Cingleot”), a company that is partially owned by the Company, in connection with a logistics center development project at the Hong Kong International Airport. In May 2024, the loan facility was modified to a revolving loan facility and the facility amount was reduced to HK\\$6.5 billion. As of March 31 and September 30, 2025, HK\\$5,090 million (RMB4,697 million) and HK\\$5,226 million (RMB4,771 million) was drawn down by Cingleot under this facility, respectively.\nThe Company's ecosystem offers different platforms on which different enterprises operate and the Company believes that all transactions on the Company's platforms are conducted on terms determined based on normal commercial negotiation with similar unrelated parties.\nOther than the transactions disclosed above or elsewhere in the unaudited condensed consolidated financial statements, the Company has commercial arrangements with other investees and other related parties to provide and receive certain marketing, cloud and other services and products. The amounts relating to these services provided and received represent less than $1\\%$ of the Company's revenue and total costs and expenses, respectively, for the six months ended September 30, 2024 and 2025.\nIn addition, the Company has made certain acquisitions and equity investments together with related parties from time to time. The agreements for acquisitions and equity investments were entered into by the parties involved and conducted on fair value basis.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/16. Related party transactions (Continued)", + "metadata": { + "length": 3497, + "summary": "The Company has commercial arrangements with investees for cloud, marketing, and logistics services, recording revenues and costs for the six months ended September 30, 2024 and 2025. Loans to investees totaled RMB1,771 million and RMB1,992 million as of March 31 and September 30, 2025, with terms up to 6 years and interest rates up to 10% per annum. A guarantee was provided for a HK$6.5 billion revolving loan facility to Cingleot, with HK$5,090 million and HK$5,226 million drawn as of those dates. Other transactions with investees and related parties are less than 1% of revenue and costs.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Transactions", + "investees", + "The", + "Company", + "commercial", + "arrangements", + "related", + "cloud", + "services", + "In", + "connection", + "provided", + "RMB1", + "058", + "million", + "RMB3", + "854", + "recorded", + "revenue", + "unaudited", + "condensed", + "consolidated", + "income", + "statements", + "months", + "ended", + "September", + "30", + "2024", + "2025", + "marketing", + "RMB358", + "RMB864", + "cost", + "sales", + "expenses", + "logistics", + "RMB2", + "092", + "680", + "Costs", + "incurred", + "RMB7", + "999", + "RMB8", + "251", + "periods", + "extended", + "loans", + "working", + "capital", + "conjunction", + "investments", + "As", + "March", + "31", + "aggregate", + "outstanding", + "balance", + "771", + "992", + "remaining", + "terms", + "years", + "interest", + "rates", + "10", + "annum", + "guarantee", + "term", + "loan", + "facility", + "HK", + "7.7", + "billion", + "favor", + "Hong", + "Kong", + "Cingleot", + "Investment", + "Management", + "Limited", + "company", + "partially", + "owned", + "center", + "development", + "project", + "International", + "Airport", + "May", + "modified", + "revolving", + "amount", + "reduced", + "6.5", + "090", + "RMB4", + "697", + "226", + "drawn", + "ecosystem", + "offers", + "platforms", + "enterprises", + "operate", + "believes", + "transactions", + "conducted", + "determined", + "based", + "normal", + "negotiation", + "similar", + "unrelated", + "parties", + "Other", + "disclosed", + "financial", + "provide", + "receive", + "products", + "amounts", + "relating", + "received", + "represent", + "total", + "costs", + "addition", + "made", + "acquisitions", + "equity", + "time", + "agreements", + "entered", + "involved", + "fair", + "basis" + ], + "keywords": [ + "investees", + "commercial arrangements", + "loans" + ], + "connect_to": [] + } + }, + { + "chunk_id": "98ae8846-9947-5021-b1a0-e020184f7545", + "type": "text", + "content": "(a) The Company is incorporated in the Cayman Islands and considered as a foreign entity under PRC laws. Due to legal restrictions on foreign ownership and investment in, among other areas, value-added telecommunications services, which include the operations of Internet content providers, the Company operates its Internet businesses and other businesses through various contractual arrangements with VIEs that are incorporated and owned by PRC citizens or by PRC entities owned and/or controlled by PRC citizens. The VIEs hold the licenses and approvals that are essential for their business operations in the PRC and the Company has entered into various agreements with the VIEs and their equity holders such that the Company has the right to benefit from their licenses and approvals and generally has control of the VIEs. In the Company’s opinion, the current ownership structure and the contractual arrangements with the VIEs and their equity holders as well as the operations of the VIEs are in substantial compliance with all existing PRC laws, rules and regulations. However, there may be changes and other developments in PRC laws, rules and regulations. Accordingly, the Company gives no assurance that PRC government authorities will not take a view in the future that is contrary to the opinion of the Company. If the current ownership structure of the Company and its contractual arrangements with the VIEs and their equity holders were found to be in violation of any existing or future PRC laws or regulations, the Company’s ability to conduct its business could be impacted and the Company may be required to restructure its ownership structure and operations in the PRC to comply with the changes in the PRC laws which may result in deconsolidation of the VIEs.\n(b) The PRC market in which the Company operates poses certain macro-economic and regulatory risks and uncertainties. These uncertainties extend to the ability of the Company to operate or invest in online and mobile commerce or other Internet related businesses, representing the principal services provided by the Company, in the PRC. The information and technology industries are highly regulated. Restrictions are currently in place or are unclear regarding what specific segments of these industries foreign owned enterprises, like the Company, may operate. If new or more extensive restrictions were imposed on the segments in which the Company is permitted to operate, the Company could be required to sell or cease to operate or invest in some or all of its current businesses in the PRC.\n(c) Because of the Company's equity interest in and close association with Ant Group and overlapping user bases, regulatory developments, litigation or proceedings, media and other reports, whether or not true, and other events that affect Ant Group could also negatively affect customers', regulators', investors' and other third parties' perception of the Company. Changes in Ant Group's business and future prospects, or speculation of such changes, as well as additional regulatory requirements placed on Ant Group, could in turn have a material adverse effect on the Company.\n(d) The Company's sales, purchase and expense transactions are generally denominated in RMB and a significant portion of the Company's assets and liabilities are denominated in RMB. RMB is not freely convertible into foreign currencies. In the PRC, foreign exchange transactions are required by law to be transacted only by authorized financial institutions at exchange rates set by the People's Bank of China (the “PBOC”).\nRemittances in currencies other than RMB by the Company in the PRC must be processed through the PBOC or other PRC foreign exchange regulatory bodies and require certain supporting documentation in order to effect the remittance. If the foreign exchange control system prevents the Company from obtaining sufficient foreign currencies to satisfy its currency demands, the Company may not be able to pay dividends in foreign currencies and the Company’s ability to fund its business activities that are conducted in foreign currencies could be adversely affected.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/17. Risks and contingencies", + "metadata": { + "length": 4145, + "summary": "The passage outlines key risks for a Cayman Islands company operating in China through Variable Interest Entities (VIEs) due to foreign ownership restrictions in value-added telecom services. It highlights uncertainties in PRC laws that could force restructuring or deconsolidation of VIEs. Macro-economic and regulatory risks in the PRC market, especially for internet and mobile commerce, are noted, with potential new restrictions possibly requiring business sales or closures. The company's association with Ant Group poses reputational and regulatory risks. Additionally, currency risks are discussed, as RMB is not freely convertible, and foreign exchange controls could hinder dividend payments and funding of foreign currency activities.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "Company", + "incorporated", + "Cayman", + "Islands", + "considered", + "foreign", + "entity", + "PRC", + "laws", + "Due", + "legal", + "restrictions", + "ownership", + "investment", + "areas", + "added", + "telecommunications", + "services", + "include", + "operations", + "Internet", + "content", + "providers", + "operates", + "businesses", + "contractual", + "arrangements", + "VIEs", + "owned", + "citizens", + "entities", + "controlled", + "hold", + "licenses", + "approvals", + "essential", + "business", + "entered", + "agreements", + "equity", + "holders", + "benefit", + "generally", + "control", + "In", + "opinion", + "current", + "structure", + "substantial", + "compliance", + "existing", + "rules", + "regulations", + "However", + "developments", + "Accordingly", + "assurance", + "government", + "authorities", + "view", + "future", + "contrary", + "If", + "found", + "violation", + "ability", + "conduct", + "impacted", + "required", + "restructure", + "comply", + "result", + "deconsolidation", + "market", + "poses", + "macro", + "economic", + "regulatory", + "risks", + "uncertainties", + "These", + "extend", + "operate", + "invest", + "online", + "mobile", + "commerce", + "related", + "representing", + "principal", + "provided", + "information", + "technology", + "industries", + "highly", + "regulated", + "Restrictions", + "place", + "unclear", + "specific", + "segments", + "enterprises", + "extensive", + "imposed", + "permitted", + "sell", + "cease", + "Because", + "interest", + "close", + "association", + "Ant", + "Group", + "overlapping", + "user", + "bases", + "litigation", + "proceedings", + "media", + "reports", + "true", + "events", + "affect", + "negatively", + "customers", + "regulators", + "investors", + "parties", + "perception", + "Changes", + "prospects", + "speculation", + "additional", + "requirements", + "turn", + "material", + "adverse", + "effect", + "sales", + "purchase", + "expense", + "transactions", + "denominated", + "RMB", + "significant", + "portion", + "assets", + "liabilities", + "freely", + "convertible", + "currencies", + "exchange", + "law", + "transacted", + "authorized", + "financial", + "institutions", + "rates", + "set", + "People", + "Bank", + "China", + "PBOC", + "Remittances", + "processed", + "bodies", + "require", + "supporting", + "documentation", + "order", + "remittance", + "system", + "prevents", + "obtaining", + "sufficient", + "satisfy", + "currency", + "demands", + "pay", + "dividends", + "fund", + "activities", + "conducted", + "adversely", + "affected" + ], + "keywords": [ + "VIE structure", + "PRC regulatory risk", + "foreign exchange" + ], + "connect_to": [] + } + }, + { + "chunk_id": "71bb0fd2-cf31-5290-adc7-ec73ca565af6", + "type": "text", + "content": "(e) In the ordinary course of business, the Company makes strategic investments to increase the service offerings and expand capabilities. The Company continually reviews its investments to determine whether there is a decline in fair value below the carrying value. Fair value of the listed securities is subject to volatility and may be materially affected by market fluctuations.\n(f) Financial instruments that potentially subject the Company to significant concentration of credit risk consist principally of cash and cash equivalents, short-term investments, restricted cash and equity securities and other investments. As of September 30, 2025, substantially all of the Company's cash and cash equivalents, restricted cash, short-term investments and other treasury investments were held by major financial institutions located worldwide, including Chinese mainland and Hong Kong S.A.R. If the financial institutions and other issuers of financial instruments held by the Company could become insolvent or if the markets for these instruments could become illiquid as a result of a severe economic downturn, the Company could lose some or all of the value of its investments.\n(g) The Company offers trade assurance program on the international wholesale marketplaces at no charge to the wholesale buyers and sellers. If the wholesale sellers who participate in this program do not deliver the products in their stated specifications to the wholesale buyers on schedule, the Company may compensate the wholesale buyers for their losses on behalf of the wholesale sellers up to a pre-determined amount following a review of each particular case. In turn, the Company will seek a full reimbursement from the wholesale sellers for the prepaid reimbursement amount, yet the Company is exposed to a risk over the collectability of the reimbursement from the wholesale sellers. During the six months ended September 30, 2025, the Company did not incur any material losses with respect to the compensation provided under this program. Given that the maximum compensation for each wholesale seller is pre-determined based on their individual risk assessments by the Company considering their credit profile or other relevant information, the Company determined that the likelihood of material default on the payments is not probable and therefore no provisions have been made in relation to this program.\n(h) In the ordinary course of business, the Company is from time to time involved in legal proceedings and litigations and is subject to regulatory investigations. The more stringent obligations under laws and regulations will create additional operational requirements with increased compliance costs for the Company. In March 2024, the European Commission, or the EU Commission, opened formal proceeding against AliExpress to assess whether AliExpress breached the Digital Services Act. On June 18, 2025, the EU Commission issued preliminary findings in which it considers on a preliminary basis that AliExpress is in breach of its obligation to assess and mitigate risks related to the dissemination of illegal content on its platform. In September 2025, AliExpress filed observations concerning such preliminary findings. The ultimate timeline and final outcome of the investigation is currently uncertain and subject to further communications with the EU Commission and their final decision based on such further communications and their deliberations. Any potential loss associated with the investigation is not reasonably estimable at this stage. Except for the above, there are no legal proceedings and litigations that have in the recent past had, or to the Company’s knowledge, are probable to have, a material impact on the Company’s financial positions, results of operations or cash flows. The Company did not accrue any material loss contingencies in this respect as of September 30, 2025.\n(i) The Russia-Ukraine conflict has resulted in significant disruptions to supply chains, logistics and business activities in the region that has negatively affected our international commerce business and Cainiao's international logistics business. The conflict has also caused, and continues to intensify, significant geopolitical tensions in Europe and across the globe. The resulting sanctions imposed are expected to have significant impacts on the economic conditions of the countries and markets targeted by such sanctions, and may have unforeseen, unpredictable secondary effects on global energy prices, supply chains and other aspects of the global economy. The conflict may adversely affect the Company's business, financial condition and results of operations.\n(j) The United Nations and a number of countries and jurisdictions, including China, the United States and the EU, have adopted various export control and economic or trade sanction regimes. In particular, the United States government and other governments have increasingly threatened and/or imposed export control, as well as economic, trade and other sanctions on a number of China-based companies. The United States and other countries may impose other and more expansive restrictions on the supply of chips, computing power or other technologies or services to China and China-based companies, including the Company, in the future. Such restrictions may affect the Company's businesses by limiting the Company's ability to upgrade its products and services, technological capabilities, in particular AI technologies, and to maintain its competitive edge, thereby negatively affecting the Company's results of operations, financial condition and growth potential.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/17. Risks and contingencies (Continued)", + "metadata": { + "length": 5647, + "summary": "The passage outlines various risks faced by the company, including strategic investment volatility, credit risk from financial institutions, trade assurance program liabilities, legal proceedings such as the EU investigation into AliExpress under the Digital Services Act, impacts from the Russia-Ukraine conflict on supply chains and logistics, and risks from export controls and sanctions affecting technology upgrades and competitive edge.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "In", + "ordinary", + "business", + "Company", + "makes", + "strategic", + "investments", + "increase", + "service", + "offerings", + "expand", + "capabilities", + "The", + "continually", + "reviews", + "determine", + "decline", + "fair", + "carrying", + "Fair", + "listed", + "securities", + "subject", + "volatility", + "materially", + "affected", + "market", + "fluctuations", + "Financial", + "instruments", + "potentially", + "significant", + "concentration", + "credit", + "risk", + "consist", + "principally", + "cash", + "equivalents", + "short", + "term", + "restricted", + "equity", + "As", + "September", + "30", + "2025", + "substantially", + "treasury", + "held", + "major", + "financial", + "institutions", + "located", + "worldwide", + "including", + "Chinese", + "mainland", + "Hong", + "Kong", + "If", + "issuers", + "insolvent", + "markets", + "illiquid", + "result", + "severe", + "economic", + "downturn", + "lose", + "offers", + "trade", + "assurance", + "program", + "international", + "wholesale", + "marketplaces", + "charge", + "buyers", + "sellers", + "participate", + "deliver", + "products", + "stated", + "specifications", + "schedule", + "compensate", + "losses", + "behalf", + "pre", + "determined", + "amount", + "review", + "case", + "turn", + "seek", + "full", + "reimbursement", + "prepaid", + "exposed", + "collectability", + "During", + "months", + "ended", + "incur", + "material", + "respect", + "compensation", + "provided", + "Given", + "maximum", + "seller", + "based", + "individual", + "assessments", + "profile", + "relevant", + "information", + "likelihood", + "default", + "payments", + "probable", + "provisions", + "made", + "relation", + "time", + "involved", + "legal", + "proceedings", + "litigations", + "regulatory", + "investigations", + "stringent", + "obligations", + "laws", + "regulations", + "create", + "additional", + "operational", + "requirements", + "increased", + "compliance", + "costs", + "March", + "2024", + "European", + "Commission", + "EU", + "opened", + "formal", + "proceeding", + "AliExpress", + "assess", + "breached", + "Digital", + "Services", + "Act", + "On", + "June", + "18", + "issued", + "preliminary", + "findings", + "considers", + "basis", + "breach", + "obligation", + "mitigate", + "risks", + "related", + "dissemination", + "illegal", + "content", + "platform", + "filed", + "observations", + "ultimate", + "timeline", + "final", + "outcome", + "investigation", + "uncertain", + "communications", + "decision", + "deliberations", + "Any", + "potential", + "loss", + "estimable", + "stage", + "Except", + "recent", + "past", + "knowledge", + "impact", + "positions", + "results", + "operations", + "flows", + "accrue", + "contingencies", + "Russia", + "Ukraine", + "conflict", + "resulted", + "disruptions", + "supply", + "chains", + "logistics", + "activities", + "region", + "negatively", + "commerce", + "Cainiao", + "caused", + "continues", + "intensify", + "geopolitical", + "tensions", + "Europe", + "globe", + "resulting", + "sanctions", + "imposed", + "expected", + "impacts", + "conditions", + "countries", + "targeted", + "unforeseen", + "unpredictable", + "secondary", + "effects", + "global", + "energy", + "prices", + "aspects", + "economy", + "adversely", + "affect", + "condition", + "United", + "Nations", + "number", + "jurisdictions", + "China", + "States", + "adopted", + "export", + "control", + "sanction", + "regimes", + "government", + "governments", + "increasingly", + "threatened", + "companies", + "impose", + "expansive", + "restrictions", + "chips", + "computing", + "power", + "technologies", + "services", + "future", + "Such", + "businesses", + "limiting", + "ability", + "upgrade", + "technological", + "AI", + "maintain", + "competitive", + "edge", + "affecting", + "growth" + ], + "keywords": [ + "Financial Risks", + "Legal Proceedings", + "Geopolitical Risks" + ], + "connect_to": [] + } + }, + { + "chunk_id": "6c24ca06-8e05-51d9-b945-d5fcdebe76b7", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_13", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_13/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ee1f4a8d-c134-5b2e-8eb9-04c9e0eb3230", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMB
(in millions)
Alibaba China E-commerce Group
Revenue242,443272,650
Costs and expenses (i)(149,363)(223,764)
Adjusted EBITA (ii)93,08048,886
Alibaba International Digital Commerce Group
Revenue60,96569,540
Costs and expenses (i)(67,576)(69,437)
Adjusted EBITA (ii)(6,611)103
Cloud Intelligence Group
Revenue56,15973,222
Costs and expenses (i)(51,161)(66,664)
Adjusted EBITA (ii)4,9986,558
All others
Revenue165,837121,568
Costs and expenses (i)(168,747)(126,353)
Adjusted EBITA (ii)(2,910)(4,785)
", + "path": "tables/table-41 Alibaba Segment Results.html", + "metadata": { + "length": 1249, + "summary": "table-42\nFinancial data for Alibaba segments (China E-commerce, International Digital Commerce, Cloud Intelligence, Others) for six months ended Sep 2024 and 2025, showing revenue, costs, and adjusted EBITA in RMB millions.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-41 Alibaba Segment Results.html", + "keywords": [ + "Revenue", + "Adjusted EBITA", + "Segment" + ], + "tokens": [] + } + }, + { + "chunk_id": "8a68445e-21ca-5a0f-88a6-5a05c0902789", + "type": "text", + "content": "Prior to the quarter ended June 30, 2025, the Company had six reportable segments, namely Taobao and Tmall Group, Alibaba International Digital Commerce Group, Cloud Intelligence Group, Cainiao Smart Logistics Network Limited, Local Services Group, and Hujing Digital Media and Entertainment Group. Starting from the quarter ended June 30, 2025, the Company has implemented a new organizational structure, which the CODM started to review information under a new reporting structure, and segment reporting has been updated to conform to this change. Comparative figures for the six months ended September 30, 2024 were updated to conform to the segment presentation.\nSegment information is presented before elimination of inter-segment transactions. In general, revenue, cost of revenue and operating expenses are directly attributable, or are allocated, to each segment. The Company allocates costs and expenses that are not directly attributable to a specific segment, such as those that support infrastructure across different segments, to different segments mainly on the basis of usage, revenue or headcount, depending on the nature of the relevant costs and expenses. The Company does not allocate assets to its segments as the CODM does not evaluate the performance of segments using asset information.\nThe CODM regularly reviews adjusted earnings before interest, taxes and amortization (“Adjusted EBITA”) for each segment which is considered as a segment operating performance measure. The CODM uses revenue and Adjusted EBITA to assess performance for each segment and allocate resources for each segment in the annual budget and forecasting process.\nThe following table presents the information by segment for the six months ended September 30, 2024 and 2025:\n\n[tables/table-41 Alibaba Segment Results.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_13/18. Segment information", + "metadata": { + "length": 1818, + "summary": "The company updated its segment reporting from six segments to a new organizational structure starting June 30, 2025. Segments include Taobao and Tmall Group, Alibaba International Digital Commerce Group, Cloud Intelligence Group, Cainiao Smart Logistics, Local Services Group, and Hujing Digital Media. The CODM uses revenue and Adjusted EBITA to evaluate performance and allocate resources. Segment information is presented before inter-segment elimination, with costs allocated based on usage, revenue, or headcount. Assets are not allocated to segments. The table provides segment data for the six months ended September 30, 2024 and 2025.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Prior", + "quarter", + "ended", + "June", + "30", + "2025", + "Company", + "reportable", + "segments", + "Taobao", + "Tmall", + "Group", + "Alibaba", + "International", + "Digital", + "Commerce", + "Cloud", + "Intelligence", + "Cainiao", + "Smart", + "Logistics", + "Network", + "Limited", + "Local", + "Services", + "Hujing", + "Media", + "Entertainment", + "Starting", + "implemented", + "organizational", + "structure", + "CODM", + "started", + "review", + "information", + "reporting", + "segment", + "updated", + "conform", + "change", + "Comparative", + "figures", + "months", + "September", + "2024", + "presentation", + "Segment", + "presented", + "elimination", + "inter", + "transactions", + "In", + "general", + "revenue", + "cost", + "operating", + "expenses", + "directly", + "attributable", + "allocated", + "The", + "allocates", + "costs", + "specific", + "support", + "infrastructure", + "basis", + "usage", + "headcount", + "depending", + "nature", + "relevant", + "allocate", + "assets", + "evaluate", + "performance", + "asset", + "regularly", + "reviews", + "adjusted", + "earnings", + "interest", + "taxes", + "amortization", + "Adjusted", + "EBITA", + "considered", + "measure", + "assess", + "resources", + "annual", + "budget", + "forecasting", + "process", + "table", + "presents" + ], + "keywords": [ + "segments", + "Adjusted EBITA", + "organizational structure" + ], + "connect_to": [ + { + "target": "ee1f4a8d-c134-5b2e-8eb9-04c9e0eb3230", + "relation": "embeds", + "ref": "[tables/table-41 Alibaba Segment Results.html]", + "position": { + "start": 1772, + "end": 1818 + } + } + ] + } + }, + { + "chunk_id": "d8b1e398-4315-5ec4-b339-231ee4d640fc", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_14", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_14/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4543413c-2e9a-597d-961e-6c5e116f0993", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMB
(in millions)
Total segments Adjusted EBITA88,55750,762
Unallocated (iii)(2,142)(1,640)
Inter-segment elimination(819)(1,205)
Non-cash share-based compensation expense(7,775)(6,076)
Amortization and impairment of intangible assets, and others(3,441)(1,488)
Provision for the shareholder class action lawsuits(3,145)
Consolidated income from operations71,23540,353
Interest and investment income, net17,12937,468
Interest expense(4,615)(4,995)
Other (expense) income, net(1,221)1,329
Income tax expenses(17,442)(14,415)
Share of results of equity method investees2,4833,254
Consolidated net income67,56962,994
", + "path": "tables/table-42 Financial Results.html", + "metadata": { + "length": 1267, + "summary": "table-43\nThe table shows consolidated income from operations decreased from 71,235 to 40,353 million RMB, while net income dropped from 67,569 to 62,994 million RMB for the six months ended September 30, 2024 and 2025.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-42 Financial Results.html", + "keywords": [ + "EBITA", + "Net Income", + "Operations" + ], + "tokens": [] + } + }, + { + "chunk_id": "70c66016-defa-5742-83c6-c44462b7552d", + "type": "table", + "content": "
Six months ended September 30,
20242025
RMBRMB
(in millions)
Alibaba China E-commerce Group7962
Alibaba International Digital Commerce Group547370
Cloud Intelligence Group6,90611,877
All others4,0871,226
Total segments depreciation and impairment of property and equipment, and operating lease cost relating to land use rights11,61913,535
", + "path": "tables/table-43 Depreciation by Segment.html", + "metadata": { + "length": 660, + "summary": "table-44\nThe table shows depreciation and impairment of property and equipment for Alibaba segments for six months ended September 30, 2024 and 2025, with totals of 11,619 and 13,535 million RMB respectively.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-43 Depreciation by Segment.html", + "keywords": [ + "depreciation", + "segments", + "Alibaba" + ], + "tokens": [] + } + }, + { + "chunk_id": "6fa09758-cf3c-50ee-a2d4-691d01aaf544", + "type": "text", + "content": "The following table presents the reconciliation from the total segments Adjusted EBITA to the consolidated net income for the six months ended September 30, 2024 and 2025:\n\n[tables/table-42 Financial Results.html]\n\nThe following table presents the depreciation and impairment of property and equipment, and operating lease cost relating to land use rights by segment for the six months ended September 30, 2024 and 2025:\n\n[tables/table-43 Depreciation by Segment.html]\n\n(i) Segment costs and expenses primarily comprise components that are included in cost of revenue, product development expenses, sales and marketing expenses, as well as general and administrative expenses.\n(ii) Adjusted EBITA represents net income before interest and investment income, net, interest expense, other (expense) income, net, income tax expenses, share of results of equity method investees, certain non-cash expenses, consisting of share-based compensation expense, amortization and impairment of intangible assets, impairment of goodwill, and others, which the Company does not believe are reflective of the Company's core operating performance during the periods presented.\n(iii) Unallocated primarily relates to certain costs incurred by corporate functions and other miscellaneous items that are not allocated to individual segments.\nDetails of the Company's revenue by segment are set out in Note 4.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_14/18. Segment information (Continued)", + "metadata": { + "length": 1395, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "table", + "presents", + "reconciliation", + "total", + "segments", + "Adjusted", + "EBITA", + "consolidated", + "net", + "income", + "months", + "ended", + "September", + "30", + "2024", + "2025", + "depreciation", + "impairment", + "property", + "equipment", + "operating", + "lease", + "cost", + "relating", + "land", + "rights", + "segment", + "Segment", + "costs", + "expenses", + "primarily", + "comprise", + "components", + "included", + "revenue", + "product", + "development", + "sales", + "marketing", + "general", + "administrative", + "ii", + "represents", + "interest", + "investment", + "expense", + "tax", + "share", + "results", + "equity", + "method", + "investees", + "cash", + "consisting", + "based", + "compensation", + "amortization", + "intangible", + "assets", + "goodwill", + "Company", + "reflective", + "core", + "performance", + "periods", + "presented", + "iii", + "Unallocated", + "relates", + "incurred", + "corporate", + "functions", + "miscellaneous", + "items", + "allocated", + "individual", + "Details", + "set", + "Note" + ], + "keywords": [], + "connect_to": [ + { + "target": "4543413c-2e9a-597d-961e-6c5e116f0993", + "relation": "embeds", + "ref": "[tables/table-42 Financial Results.html]", + "position": { + "start": 173, + "end": 213 + } + }, + { + "target": "70c66016-defa-5742-83c6-c44462b7552d", + "relation": "embeds", + "ref": "[tables/table-43 Depreciation by Segment.html]", + "position": { + "start": 422, + "end": 468 + } + } + ] + } + }, + { + "chunk_id": "1d4d310f-7f2b-5e90-893f-3758245a3353", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00f3c68c-07ec-59ec-8d48-a994959b7460", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "99274fe1-9806-5dc6-bf00-dea24e15e8da", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/19. Aging analysis", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "80b62589-b7e3-503f-be82-c2cabc32b546", + "type": "table", + "content": "
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
0-3 months25,17225,073
3-6 months3,0783,395
6-12 months1,7752,344
Over 1 year627834
Accounts receivable, net of allowance30,65231,646
", + "path": "tables/table-44 Accounts Receivable.html", + "metadata": { + "length": 517, + "summary": "table-45\nThe table shows accounts receivable aging as of March 31 and September 30, 2025, with totals of 30,652 and 31,646 million RMB respectively, broken down by 0-3, 3-6, 6-12 months, and over 1 year.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-44 Accounts Receivable.html", + "keywords": [ + "Accounts Receivable", + "Aging", + "RMB" + ], + "tokens": [] + } + }, + { + "chunk_id": "b2790564-1fe9-562e-84bb-502b2a7c59b6", + "type": "text", + "content": "The aging analysis of the accounts receivable, net of allowance, based on billing date is as follows:\n\n[tables/table-44 Accounts Receivable.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/19. Aging analysis/Accounts receivable", + "metadata": { + "length": 161, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "aging", + "analysis", + "accounts", + "receivable", + "net", + "allowance", + "based", + "billing", + "date" + ], + "keywords": [], + "connect_to": [ + { + "target": "80b62589-b7e3-503f-be82-c2cabc32b546", + "relation": "embeds", + "ref": "[tables/table-44 Accounts Receivable.html]", + "position": { + "start": 103, + "end": 145 + } + } + ] + } + }, + { + "chunk_id": "4ad4f24e-df74-5d21-b267-84d469b43112", + "type": "table", + "content": "
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
0-3 months52,01956,073
3-6 months3,9905,696
6-12 months8461,600
Over 1 year1,3461,570
Accounts payable58,20164,939
", + "path": "tables/table-45 Accounts Payable.html", + "metadata": { + "length": 498, + "summary": "table-46\nThe table shows accounts payable by maturity as of March 31 and September 30, 2025, in millions of RMB. Total payables are 58,201 and 64,939, with most due within 3 months.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-45 Accounts Payable.html", + "keywords": [ + "Accounts Payable", + "Maturity", + "RMB" + ], + "tokens": [] + } + }, + { + "chunk_id": "f703a9a6-ffe1-598a-bf46-613acf15dd2e", + "type": "text", + "content": "The aging analysis of the accounts payable based on billing date is as follows:\n\n[tables/table-45 Accounts Payable.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/19. Aging analysis/Accounts payable", + "metadata": { + "length": 139, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "aging", + "analysis", + "accounts", + "payable", + "based", + "billing", + "date" + ], + "keywords": [], + "connect_to": [ + { + "target": "4ad4f24e-df74-5d21-b267-84d469b43112", + "relation": "embeds", + "ref": "[tables/table-45 Accounts Payable.html]", + "position": { + "start": 81, + "end": 120 + } + } + ] + } + }, + { + "chunk_id": "08afc35e-c6c7-5594-8868-8a67169f1218", + "type": "text", + "content": "The Board did not recommend the distribution of interim dividend for the six months ended September 30, 2024 and 2025.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/20. Dividends", + "metadata": { + "length": 118, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "Board", + "recommend", + "distribution", + "interim", + "dividend", + "months", + "ended", + "September", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b258f4c5-ae5e-569e-a64e-9a43407488c0", + "type": "text", + "content": "Save as disclosed in the unaudited condensed consolidated financial statements, there were no significant events that might affect the Company since September 30, 2025.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/21. Subsequent events", + "metadata": { + "length": 168, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Save", + "disclosed", + "unaudited", + "condensed", + "consolidated", + "financial", + "statements", + "significant", + "events", + "affect", + "Company", + "September", + "30", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "862bb99a-e381-54a8-93cb-7ab69138224e", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_16", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5fa7fa56-65ab-506c-b7ce-e97e85afdf22", + "type": "table", + "content": "
As of March 31, 2025
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity’s own equity (viii)Amounts under IFRS
(in millions of RMB)
Equity securities and other investments356,8185,0033,547365,368
Prepayments, receivables and other assets83,431(258)485578(1,010)3826,87490,482
Investments in equity method investees210,169(241)(226)209,702
Property and equipment, net203,3481,318204,666
Intangible assets, net20,911(1,861)42619,476
Goodwill255,501(36,284)3,180222,397
Total assets1,804,227(38,644)5,4883,321578(1,010)5,3066,8741,786,140
Deferred tax liabilities48,454(719)368(283)(78)(51)47,691
Convertible unsecured senior notes35,83416,02151,855
Other liabilities17,644(2)10,24727,889
Total liabilities714,121(719)368(283)(78)(2)10,247(51)16,021739,624
Mezzanine equity11,713(11,713)
Total shareholders’ equity1,009,858(17,402)5,1203,604656(1,008)1,4392,616(9,147)995,736
Noncontrolling interests68,535(20,523)272,74150,780
Total equity1,078,393(37,925)5,1203,604656(1,008)1,4665,357(9,147)1,046,516
Total liabilities, mezzanine equity and equity1,804,227(38,644)5,4883,321578(1,010)5,3066,8741,786,140
", + "path": "tables/table-46 IFRS Adjustments.html", + "metadata": { + "length": 3372, + "summary": "table-47\nReconciliation of U.S. GAAP to IFRS for assets, liabilities, and equity as of March 31, 2025, showing adjustments for consolidation, investments, leases, and convertible notes.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-46 IFRS Adjustments.html", + "keywords": [ + "IFRS", + "GAAP", + "reconciliation" + ], + "tokens": [] + } + }, + { + "chunk_id": "a5af7c4f-4493-5c2c-9e4b-8446c8232801", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025\n22. Reconciliations between U.S. GAAP and International Financial Reporting Standards\nThe unaudited condensed consolidated financial statements are prepared in accordance with U.S. GAAP, which differ in certain respects from International Financial Reporting Standards (“IFRS”). The effects of material differences between the unaudited condensed consolidated financial statements of the Company prepared under U.S. GAAP and IFRS are as follows:\nReconciliation of Unaudited Condensed Consolidated Balance Sheets (Extract)\n\n[tables/table-46 IFRS Adjustments.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_16/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 619, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025", + "22", + "Reconciliations", + "GAAP", + "International", + "Financial", + "Reporting", + "Standards", + "The", + "unaudited", + "condensed", + "consolidated", + "financial", + "statements", + "prepared", + "accordance", + "differ", + "respects", + "IFRS", + "effects", + "material", + "differences", + "Company", + "Reconciliation", + "Unaudited", + "Condensed", + "Consolidated", + "Balance", + "Sheets", + "Extract" + ], + "keywords": [], + "connect_to": [ + { + "target": "5fa7fa56-65ab-506c-b7ce-e97e85afdf22", + "relation": "embeds", + "ref": "[tables/table-46 IFRS Adjustments.html]", + "position": { + "start": 576, + "end": 615 + } + } + ] + } + }, + { + "chunk_id": "a7d64cc9-32b2-56a7-835e-791463e0508f", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_17", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "da3489db-58f6-5ae2-ac42-24d4999305f1", + "type": "table", + "content": "
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity's own equity (viii)Amounts under IFRS
(in millions of RMB)
Equity securities and other investments411,9533,2824,223419,458
Prepayments, receivables and other assets96,927(258)521637(974)1,0238,315106,191
Investments in equity method investees206,862(241)(784)205,837
Property and equipment, net246,539699247,238
Intangible assets, net19,429(1,688)44518,186
Goodwill255,551(36,278)3,304222,577
Total assets1,883,880(38,465)3,8033,439637(974)5,4718,3151,866,106
Deferred tax liabilities46,802(684)324(292)(170)(52)45,928
Convertible unsecured senior notes57,48134,06491,545
Exchangeable bonds13,75513,755
Other liabilities21,3548,46729,821
Total liabilities772,095(684)324(292)(170)8,467(52)34,064813,752
Mezzanine equity9,884(9,884)
Total shareholders' equity1,032,495(17,252)3,4793,731807(974)1,4322,711(25,749)1,000,680
Noncontrolling interests69,406(20,529)(15)2,81251,674
Total equity1,101,901(37,781)3,4793,731807(974)1,4175,523(25,749)1,052,354
Total liabilities, mezzanine equity and equity1,883,880(38,465)3,8033,439637(974)5,4718,3151,866,106
", + "path": "tables/table-47 GAAP to IFRS.html", + "metadata": { + "length": 3468, + "summary": "table-48\nReconciliation of U.S. GAAP to IFRS for assets, liabilities, and equity, showing adjustments for consolidation, equity securities, leases, and convertible notes.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-47 GAAP to IFRS.html", + "keywords": [ + "GAAP", + "IFRS", + "reconciliation" + ], + "tokens": [] + } + }, + { + "chunk_id": "8a988273-cbf5-5031-8898-dd4481f593d3", + "type": "text", + "content": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025\n22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued)\nReconciliation of Unaudited Condensed Consolidated Balance Sheets (Extract)\nAs of September 30, 2025\n\n[tables/table-47 GAAP to IFRS.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_17/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 356, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025", + "22", + "Reconciliations", + "GAAP", + "International", + "Financial", + "Reporting", + "Standards", + "Continued", + "Reconciliation", + "Unaudited", + "Condensed", + "Consolidated", + "Balance", + "Sheets", + "Extract", + "As", + "September" + ], + "keywords": [], + "connect_to": [ + { + "target": "da3489db-58f6-5ae2-ac42-24d4999305f1", + "relation": "embeds", + "ref": "[tables/table-47 GAAP to IFRS.html]", + "position": { + "start": 253, + "end": 288 + } + } + ] + } + }, + { + "chunk_id": "7db1b722-ad9e-572f-a8c0-99b1062d5e37", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "859bfd8e-2b98-5081-aed8-ee1cbdcf329b", + "type": "text", + "content": "", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "02532963-dfe0-58ae-9d80-c082b4c19956", + "type": "table", + "content": "
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity's own equity (viii)Amounts under IFRS
(in millions of RMB)
Revenue479,73997479,836
Cost of revenue(290,135)(22)641(235)(289,751)
Product development expenses(27,555)(10)(1)(27,566)
Sales and marketing expenses(65,167)(2)1(323)(65,491)
General and administrative expenses(23,057)(18)70(9)(23,014)
Amortization and impairment of intangible assets(3,441)16039(3,242)
Other gains, net851851
Income from operations71,235160(52)712(432)71,623
Interest and investment income, net17,1291,750(56)2261(7,689)11,217
Interest expense(4,615)(775)(434)(12)(5,836)
Other expense, net(1,221)442(779)
Income tax expenses(17,442)(35)109(227)86611(221)(16,939)
Share of results of equity method investees2,4831,15143,638
Net income67,5691251,859868818(52)(412)(162)(7,689)62,924
Net loss attributable to noncontrolling interests854(59)(3)1331451,070
Accretion of mezzanine equity(280)280
Net income attributable to ordinary shareholders68,143661,859868815(52)1(17)(7,689)63,994
", + "path": "tables/table-48 GAAP to IFRS.html", + "metadata": { + "length": 3378, + "summary": "table-49\nReconciliation of financial data from U.S. GAAP to IFRS for a company, showing adjustments for items like hyperinflation, convertible notes, and equity method investments, resulting in net income of 62,924 million RMB under IFRS.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-48 GAAP to IFRS.html", + "keywords": [ + "GAAP", + "IFRS", + "reconciliation" + ], + "tokens": [] + } + }, + { + "chunk_id": "e965f3b3-fb05-5387-b683-7bb1b8b0c0c9", + "type": "table", + "content": "
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity’s own equity (viii)Amounts under IFRS
(in millions of RMB)
Revenue495,4471495,448
Cost of revenue(287,210)(12)304(425)(287,343)
Product development expenses(32,096)(7)(15)(32,118)
Sales and marketing expenses(119,674)(2)1(154)(119,829)
General and administrative expenses(14,778)(14)36(5)(14,761)
Amortization and impairment of intangible assets(1,633)17367(1,393)
Other gains (losses), net297(506)(209)
Income from operations40,353173(35)341(1,037)39,795
Interest and investment income, net37,468(1,641)210611016(18,008)18,116
Interest expense(4,995)(395)(337)1,536(30)(4,221)
Other income, net1,32973961,732
Income tax expenses(14,415)(35)39160(31)(389)(14,770)
Share of results of equity method investees3,254(79)293,204
Net income62,994(250)(1,602)13254341,546(1,044)(18,008)43,856
Net income attributable to noncontrolling interests(1,326)(65)(3)887251(256)
Reversal of accretion of mezzanine equity2,438(2,438)
Net income attributable to ordinary shareholders64,106(315)(1,602)1325134(5)(793)(18,008)43,600
", + "path": "tables/table-49 GAAP to IFRS Reconciliation.html", + "metadata": { + "length": 3423, + "summary": "table-50\nThis table reconciles financial statement line items from U.S. GAAP to IFRS, showing adjustments for consolidation, equity securities, leases, and convertible notes, resulting in net income of RMB 43,856 million under IFRS.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "tables/table-49 GAAP to IFRS Reconciliation.html", + "keywords": [ + "GAAP", + "IFRS", + "reconciliation" + ], + "tokens": [] + } + }, + { + "chunk_id": "b3fb2f2a-e0b3-5822-8741-ce3018c26d7b", + "type": "text", + "content": "22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued)\nReconciliation of Unaudited Condensed Consolidated Income Statement (Extract)\nSix months ended September 30, 2024\n\n[tables/table-48 GAAP to IFRS.html]\n\n22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued)\nReconciliation of Unaudited Condensed Consolidated Income Statement (Extract) (Continued)\nSix months ended September 30, 2025\n\n[tables/table-49 GAAP to IFRS Reconciliation.html]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "metadata": { + "length": 646, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "22", + "Reconciliations", + "GAAP", + "International", + "Financial", + "Reporting", + "Standards", + "Continued", + "Reconciliation", + "Unaudited", + "Condensed", + "Consolidated", + "Income", + "Statement", + "Extract", + "Six", + "months", + "ended", + "September", + "30", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [ + { + "target": "02532963-dfe0-58ae-9d80-c082b4c19956", + "relation": "embeds", + "ref": "[tables/table-48 GAAP to IFRS.html]", + "position": { + "start": 213, + "end": 248 + } + }, + { + "target": "e965f3b3-fb05-5387-b683-7bb1b8b0c0c9", + "relation": "embeds", + "ref": "[tables/table-49 GAAP to IFRS Reconciliation.html]", + "position": { + "start": 475, + "end": 525 + } + } + ] + } + }, + { + "chunk_id": "7aa61892-1fa2-5b0b-85e8-45fd421af7a0", + "type": "text", + "content": "The Company consolidates an entity when the Company obtains control over the entity and deconsolidates the entity upon the loss of control. Under U.S. GAAP, control generally exists when the Company obtains a controlling financial interest over an entity, whereby the usual condition is ownership of over 50% of the voting shares. Under IFRS, de facto control exists when the Company has the practical ability to direct the relevant activities of the entity, even if the Company owns less than 50% of the voting shares.\nThe Company recognizes noncontrolling interests to reflect the portion of equity of a subsidiary that is not attributable to the Company. Under U.S. GAAP, noncontrolling interests are measured at fair value and full goodwill in relation to the acquiree is recognized in a business combination. Under IFRS, the Company can elect, on a transaction-by-transaction basis, to measure noncontrolling interests at the noncontrolling interests' proportionate share of the acquiree's net identifiable assets and partial goodwill is recognized to reflect the controlling interests only.\nThe Company recognizes an impairment loss when the Company determines that the carrying value of goodwill is not recoverable. Under U.S. GAAP, the impairment loss is measured by comparing the carrying value of the reporting unit, including goodwill, with its fair value. Under IFRS, the carrying value is compared with the recoverable amount, which is the higher of fair value less costs of disposal and value in use.\nALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025\n22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(i) Consolidation and business combinations", + "metadata": { + "length": 1758, + "summary": "The passage discusses key differences between U.S. GAAP and IFRS in three areas: consolidation criteria (control via voting shares vs. de facto control), measurement of noncontrolling interests (fair value with full goodwill vs. proportionate share with partial goodwill), and goodwill impairment testing (comparing carrying value to fair value vs. recoverable amount). It is from Alibaba Group's financial statement notes for the six months ended September 30, 2024 and 2025.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "Company", + "consolidates", + "entity", + "obtains", + "control", + "deconsolidates", + "loss", + "Under", + "GAAP", + "generally", + "exists", + "controlling", + "financial", + "interest", + "usual", + "condition", + "ownership", + "50%", + "voting", + "shares", + "IFRS", + "de", + "facto", + "practical", + "ability", + "direct", + "relevant", + "activities", + "owns", + "recognizes", + "noncontrolling", + "interests", + "reflect", + "portion", + "equity", + "subsidiary", + "attributable", + "measured", + "fair", + "full", + "goodwill", + "relation", + "acquiree", + "recognized", + "business", + "combination", + "elect", + "transaction", + "basis", + "measure", + "proportionate", + "share", + "net", + "identifiable", + "assets", + "partial", + "impairment", + "determines", + "carrying", + "recoverable", + "comparing", + "reporting", + "unit", + "including", + "compared", + "amount", + "higher", + "costs", + "disposal", + "ALIBABA", + "GROUP", + "HOLDING", + "LIMITED", + "NOTES", + "TO", + "UNAUDITED", + "CONDENSED", + "CONSOLIDATED", + "FINANCIAL", + "STATEMENTS", + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025", + "22", + "Reconciliations", + "International", + "Financial", + "Reporting", + "Standards", + "Continued" + ], + "keywords": [ + "U.S. GAAP", + "IFRS", + "Consolidation" + ], + "connect_to": [] + } + }, + { + "chunk_id": "99c8e281-71d6-517d-ba9b-f1cebcd939df", + "type": "text", + "content": "Under U.S. GAAP, the Company can elect, on an instrument-by-instrument basis, to apply the measurement alternative to record the investments in equity securities without readily determinable fair values at cost, less impairment, with subsequent adjustments for observable price changes recognized in the unaudited condensed consolidated income statements. Under IFRS, these investments are measured at fair value with changes in fair value recognized in the unaudited condensed consolidated income statements.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(ii) Equity securities without readily determinable fair value", + "metadata": { + "length": 509, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Under", + "GAAP", + "Company", + "elect", + "instrument", + "basis", + "apply", + "measurement", + "alternative", + "record", + "investments", + "equity", + "securities", + "readily", + "determinable", + "fair", + "values", + "cost", + "impairment", + "subsequent", + "adjustments", + "observable", + "price", + "recognized", + "unaudited", + "condensed", + "consolidated", + "income", + "statements", + "IFRS", + "measured" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3099eafb-6cde-5acf-8312-13d910bcc320", + "type": "text", + "content": "The Company generally applies the equity method to account for equity investments over which it has significant influence. Under U.S. GAAP, significant influence is presumed to exist for an investment in limited partnership or unincorporated entity, unless the investment is so minor that the Company has virtually no influence over the entity's operating and financial policies. Under IFRS, significant influence is presumed to exist for an investment of over $20\\%$ of the voting rights of an entity.\nThe Company records its share of the post-acquisition results of its equity method investees and adjusts for the basis differences that exist between the carrying values of the equity method investments and the Company's proportionate share of the carrying value of the investee's net assets. Adjustments are made to the financial statements of the equity method investees prepared under U.S. GAAP in order to conform to the Company's accounting policies under IFRS and to reflect the basis differences of the equity method investments under IFRS, if different from those under U.S. GAAP.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(iii) Equity method investments", + "metadata": { + "length": 1091, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "Company", + "generally", + "applies", + "equity", + "method", + "account", + "investments", + "significant", + "influence", + "Under", + "GAAP", + "presumed", + "exist", + "investment", + "limited", + "partnership", + "unincorporated", + "entity", + "minor", + "virtually", + "operating", + "financial", + "policies", + "IFRS", + "20", + "voting", + "rights", + "records", + "share", + "post", + "acquisition", + "results", + "investees", + "adjusts", + "basis", + "differences", + "carrying", + "values", + "proportionate", + "investee", + "net", + "assets", + "Adjustments", + "made", + "statements", + "prepared", + "order", + "conform", + "accounting", + "reflect" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9eebafb5-7ced-5cc1-9ca5-b6aa5fb0e5d8", + "type": "text", + "content": "The employees of the Company hold share-based awards relating to an equity method investee of the Company that were granted and will be settled by related parties or economic interest holders of the Company. Under U.S. GAAP, the cost related to these awards is recognized over the requisite service period, with subsequent changes in fair value of these awards recognized in the unaudited condensed consolidated income statements. Under IFRS, these awards are not considered as share based payments of the Company and the cost relating to these awards is not recognized.\nThe Company accounts for income tax effects of share-based awards that ordinarily give rise to tax deduction. Under U.S. GAAP, deferred taxes for these awards are measured based on share-based compensation expenses recognized in the unaudited condensed consolidated financial statements. Under IFRS, deferred taxes for these awards are measured based on future tax deduction estimated at the end of each reporting period.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(iv) Share-based awards", + "metadata": { + "length": 992, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "The", + "employees", + "Company", + "hold", + "share", + "based", + "awards", + "relating", + "equity", + "method", + "investee", + "granted", + "settled", + "related", + "parties", + "economic", + "interest", + "holders", + "Under", + "GAAP", + "cost", + "recognized", + "requisite", + "service", + "period", + "subsequent", + "fair", + "unaudited", + "condensed", + "consolidated", + "income", + "statements", + "IFRS", + "considered", + "payments", + "accounts", + "tax", + "effects", + "ordinarily", + "give", + "rise", + "deduction", + "deferred", + "taxes", + "measured", + "compensation", + "expenses", + "financial", + "future", + "estimated", + "end", + "reporting" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6dc18dc5-b646-573e-ba3a-071b29921aef", + "type": "text", + "content": "Under U.S. GAAP, the amortization of right-of-use assets and the interest expense related to lease liabilities are recorded together as lease expense and recognized in the unaudited condensed consolidated income statements on a straight-line basis. Under IFRS, the right-of-use assets are amortized on a straight-line basis while the interest expense related to lease liabilities are recognized in the unaudited condensed consolidated income statements using effective interest method.", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(v) Operating leases", + "metadata": { + "length": 485, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Under", + "GAAP", + "amortization", + "assets", + "interest", + "expense", + "related", + "lease", + "liabilities", + "recorded", + "recognized", + "unaudited", + "condensed", + "consolidated", + "income", + "statements", + "straight", + "line", + "basis", + "IFRS", + "amortized", + "effective", + "method" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "166864b7-c89b-5499-b5f4-b6aaf7063132", + "type": "text", + "content": "Equity interests issued by certain subsidiaries of the Company are redeemable. Under U.S. GAAP, redeemable equity interests are classified as mezzanine equity if the redemption is outside the Company's control and as noncontrolling interests if equity interests issued by finite-lived subsidiaries are mandatorily redeemable only upon liquidation. Under IFRS, these redeemable equity interests are generally classified as financial liabilities.\nALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025\n22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(vi) Redeemable noncontrolling interests", + "metadata": { + "length": 688, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Equity", + "interests", + "issued", + "subsidiaries", + "Company", + "redeemable", + "Under", + "GAAP", + "equity", + "classified", + "mezzanine", + "redemption", + "control", + "noncontrolling", + "finite", + "lived", + "mandatorily", + "liquidation", + "IFRS", + "generally", + "financial", + "liabilities", + "ALIBABA", + "GROUP", + "HOLDING", + "LIMITED", + "NOTES", + "TO", + "UNAUDITED", + "CONDENSED", + "CONSOLIDATED", + "FINANCIAL", + "STATEMENTS", + "FOR", + "THE", + "SIX", + "MONTHS", + "ENDED", + "SEPTEMBER", + "30", + "2024", + "2025", + "22", + "Reconciliations", + "International", + "Financial", + "Reporting", + "Standards", + "Continued" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6182ef4d-731c-53ad-80c5-26b10780a5f1", + "type": "image", + "content": "\n### (vii) Hyperinflation ### Alibaba0\n[images/image-2-### (vii) .jpg]\n", + "path": "images/image-2-### (vii) .jpg", + "metadata": { + "length": 71, + "summary": "image-2\n### (vii) Hyperinflation ### Alibaba0", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "file_path": "images/image-2-### (vii) .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "2adc00fb-4e23-542c-94c8-b940de8f14b7", + "type": "text", + "content": "Under U.S. GAAP, when the Company determines that a subsidiary is operating in a highly inflationary economy, the financial statements of this subsidiary are remeasured prospectively as if its functional currency was the functional currency of its immediate parent company. Under IFRS, the financial statements of the subsidiary operating in a highly inflationary economy are restated in terms of the measuring unit current at the end of the reporting period.\n(viii) Convertible unsecured senior notes and instruments in an entity's own equity\nCertain unsecured senior notes issued by the Company are convertible into its ordinary shares. Under U.S. GAAP, these convertible notes are accounted for in entirety as liabilities measured at amortized cost with debt discounts and debt issuance costs amortized using the effective interest method. Under IFRS, these convertible notes that the Company can settle in cash at its election are accounted for as hybrid instruments, and the Company designates the hybrid instruments as financial liabilities measured at fair value with changes in fair value recognized in the unaudited condensed consolidated income statements, whereas changes in fair value attributed to credit risk recognized separately in other comprehensive income are insignificant. Furthermore, under IFRS, if the conversion options embedded in the convertible unsecured senior notes or the exchange options embedded in the exchangeable bonds are not classified as equity, the respective instruments are classified as current liabilities when the holders have the right to convert or exchange them within twelve months after the reporting period.\nUnder U.S. GAAP, for an instrument in an entity's own equity to be classified as equity, the instrument must be both (1) indexed to the entity's stock (commonly referred to as “fixed-for-fixed” test) and (2) classified in shareholders' equity, regardless of whether the instrument is a derivative under ASC 815. Under IFRS, a derivative in an entity's own equity may nonetheless be considered an equity instrument if the derivative meets the “fixed-for-fixed” test, whereas a non-derivative that includes no contractual obligation for the entity to deliver cash, financial asset or a variable number of its own equity instrument is recognized in shareholders' equity with no subsequent measurement. Besides, differences exist in the application of the “fixed-for-fixed” test between U.S. GAAP and IFRS. For example, settlement alternatives that may permit equity classification under U.S. GAAP typically result in an asset or a liability classification under IFRS. Consequently, depending on the specific terms and conditions, an instrument in an entity's own equity may be classified as equity under U.S. GAAP but as an asset or a liability measured at fair value, with changes in fair value recognized in the unaudited condensed consolidated income statement, under IFRS, or vice versa.\n\n### (vii) Hyperinflation ### Alibaba0\n[images/image-2-### (vii) .jpg]", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(vii) Hyperinflation", + "metadata": { + "length": 3018, + "summary": "The passage compares U.S. GAAP and IFRS treatments for hyperinflationary subsidiaries, convertible unsecured senior notes, and instruments in an entity's own equity. Under U.S. GAAP, subsidiaries in highly inflationary economies are remeasured using the parent's functional currency, while IFRS requires restatement for current purchasing power. For convertible notes, U.S. GAAP treats them as liabilities at amortized cost, whereas IFRS accounts for them as hybrid instruments at fair value with credit risk changes in OCI. For equity instruments, U.S. GAAP requires both a fixed-for-fixed test and equity classification, while IFRS allows equity treatment for derivatives meeting the test and non-derivatives without cash delivery obligations, with differences in settlement alternatives leading to potential classification disparities.", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Under", + "GAAP", + "Company", + "determines", + "subsidiary", + "operating", + "highly", + "inflationary", + "economy", + "financial", + "statements", + "remeasured", + "prospectively", + "functional", + "currency", + "parent", + "company", + "IFRS", + "restated", + "terms", + "measuring", + "unit", + "current", + "end", + "reporting", + "period", + "viii", + "Convertible", + "unsecured", + "senior", + "notes", + "instruments", + "entity", + "equity", + "Certain", + "issued", + "convertible", + "ordinary", + "shares", + "accounted", + "entirety", + "liabilities", + "measured", + "amortized", + "cost", + "debt", + "discounts", + "issuance", + "costs", + "effective", + "interest", + "method", + "settle", + "cash", + "election", + "hybrid", + "designates", + "fair", + "recognized", + "unaudited", + "condensed", + "consolidated", + "income", + "attributed", + "credit", + "risk", + "separately", + "comprehensive", + "insignificant", + "Furthermore", + "conversion", + "options", + "embedded", + "exchange", + "exchangeable", + "bonds", + "classified", + "respective", + "holders", + "convert", + "twelve", + "months", + "instrument", + "indexed", + "stock", + "commonly", + "referred", + "fixed", + "test", + "shareholders", + "derivative", + "ASC", + "815", + "nonetheless", + "considered", + "meets", + "includes", + "contractual", + "obligation", + "deliver", + "asset", + "variable", + "number", + "subsequent", + "measurement", + "Besides", + "differences", + "exist", + "application", + "For", + "settlement", + "alternatives", + "permit", + "classification", + "typically", + "result", + "liability", + "Consequently", + "depending", + "specific", + "conditions", + "statement", + "vice", + "versa", + "vii", + "Hyperinflation", + "Alibaba0" + ], + "keywords": [ + "Hyperinflation", + "Convertible Notes", + "Equity Instruments" + ], + "connect_to": [ + { + "target": "6182ef4d-731c-53ad-80c5-26b10780a5f1", + "relation": "embeds", + "ref": "[images/image-2-### (vii) .jpg]", + "position": { + "start": 2987, + "end": 3018 + } + } + ] + } + }, + { + "chunk_id": "871b41b6-9016-52a2-94d2-0d36953f69a7", + "type": "text", + "content": "Alibaba Group Holding Limited", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/Alibaba", + "metadata": { + "length": 29, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Weighted Voting Rights, DIRECTORS AND CHIEF EXECUTIVE OFFICER, DISCLOSURE OF INTERESTS, EQUITY INCENTIVE PLANS, PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES, OTHER INFORMATION, DEFINITIONS Conventions that apply to this Interim Report, ALIBABA GROUP HOLDING LIMITED, ALIBABA GROUP HOLDING LIMITED_2, ALIBABA GROUP HOLDING LIMITED_3, ALIBABA GROUP HOLDING LIMITED_4, ALIBABA GROUP HOLDING LIMITED_5, ALIBABA GROUP HOLDING LIMITED_6, ALIBABA GROUP HOLDING LIMITED_7, ALIBABA GROUP HOLDING LIMITED_8, ALIBABA GROUP HOLDING LIMITED_9, ALIBABA GROUP HOLDING LIMITED_10, ALIBABA GROUP HOLDING LIMITED_11, ALIBABA GROUP HOLDING LIMITED_12, ALIBABA GROUP HOLDING LIMITED_13, ALIBABA GROUP HOLDING LIMITED_14, ALIBABA GROUP HOLDING LIMITED_15, ALIBABA GROUP HOLDING LIMITED_16, ALIBABA GROUP HOLDING LIMITED_17, ALIBABA GROUP HOLDING LIMITED_18", + "tokens": [ + "Alibaba", + "Group", + "Holding", + "Limited" + ], + "keywords": [], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/doc_nav.json b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/doc_nav.json new file mode 100644 index 000000000..a48a13255 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/doc_nav.json @@ -0,0 +1,1581 @@ +{ + "version": "1.0", + "file_name": "Alibaba Fiscal Year 2026 Interim Report.pdf", + "stats": { + "total_chunks": 216, + "text_chunks": 164, + "image_chunks": 2, + "table_chunks": 50, + "max_depth": 5 + }, + "sections": [ + { + "title": "Root", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf", + "level": 1, + "summary": "Alibaba0 [images/image-1-Alibaba.jpg] Alibaba Alibaba Group Holding Limited NYSE: BABA HKEX: 9988 (HKD Counter) 89988 (RMB Counter)", + "chunk_count": 1, + "children": [] + }, + { + "title": "Weighted Voting Rights", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights", + "level": 1, + "summary": "The company has a weighted voting rights (WVR) structure where the Alibaba Partnership can nominate or appoint up to a simple majority of board members as long as the Partnership Condition is met. This structure is permanent unless amended by a 95% shareholder vote. The partnership's nomination rights survive a change of control. Shareholders face risks as the partnership limits their ability to elect directors and its interests may conflict with theirs.", + "chunk_count": 46, + "children": [ + { + "title": "Exchange Rate Information", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Exchange Rate Information", + "level": 2, + "summary": "This Interim Report contains translations of certain Renminbi (“RMB”) amounts into U.S. dollars (“US\\$”) and Hong Kong dollars (“HK\\$”) for the convenience of the reader. Unless otherwise stated, all translations of RMB into US\\$ were made at RMB7.1190 to US\\$1.00, the exchange rate on September 30, 2025 as set forth in the H.10 statistical release of the Federal Reserve Board, and all translations of RMB into HK\\$ were made at RMB0.91298 to HK\\$1.00, the middle rate on September 30, 2025 as published by the People’s Bank of China. The percentages stated in this Interim Report are calculated based on the RMB amounts and there may be minor differences due to rounding.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Safe Harbor Statements", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Safe Harbor Statements", + "level": 2, + "summary": "This Interim Report contains forward-looking statements under safe harbor provisions, discussing Alibaba's strategies, business plans, and growth expectations. It highlights risks including competition, investments, strategic transactions, economic conditions, geopolitical tensions, and shareholder return changes. The Six Months Ended September Summary Financial Results section references non-GAAP measures, with year-over-year changes driven by factors like adjusted EBITA decrease, investments in quick commerce and technology, revenue growth in Alibaba China E-commerce Group, improved Cloud business results, and mark-to-market changes from equity investments.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Six Months Ended September Segment Results", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results", + "level": 2, + "summary": "Alibaba Group's revenue for the six months ended September 30, 2025, was RMB495,447 million (US$69,595 million), a 3% year-over-year increase from RMB479,739 million in 2024. Excluding disposed businesses (Sun Art and Intime), like-for-like revenue grew 12%. The company restructured its reporting segments, combining Taobao, Tmall, Ele.me, and Fliggy into Alibaba China E-commerce Group, and reclassifying Cainiao, Amap, and Digital Media and Entertainment Group into 'All others'. The text includes tables detailing revenue and adjusted EBITA breakdowns by segment, with notes on direct sales, quick commerce revenue, and segment performance drivers.", + "chunk_count": 18, + "children": [ + { + "title": "Alibaba China E-commerce Group", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group", + "level": 3, + "summary": "", + "chunk_count": 6, + "children": [ + { + "title": "(i) Segment revenue", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group/(i) Segment revenue", + "level": 4, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "• E-commerce Business", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group/(i) Segment revenue/• E-commerce Business", + "level": 5, + "summary": "Revenue from our E-commerce business in the six months ended September 30, 2025 was RMB221,510 million (US\\$31,116 million), an increase of 9% compared to RMB202,988 million in the same period of 2024. Customer management revenue increased by 10% year-over-year, primarily due to the improvement of take rate. Direct sales, logistics and others revenue under E-commerce business in the six months ended September 30, 2025 was RMB53,331 million (US\\$7,492 million), an increase of 6% compared to RMB50,233 million in the same period of 2024, primarily driven by the increase in revenue from logistics services and value-added services, partly offset by the decrease in revenue from certain direct sales businesses.", + "chunk_count": 1, + "children": [] + }, + { + "title": "• Quick Commerce Business", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group/(i) Segment revenue/• Quick Commerce Business", + "level": 5, + "summary": "Revenue from our Quick commerce business in the six months ended September 30, 2025 was RMB37,690 million (US\\$5,294 million), an increase of 37% compared to RMB27,517 million in the same period of 2024, mainly due to order growth as a result of the rollout of “Taobao Instant Commerce” at the end of April 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "• China Commerce Wholesale Business", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group/(i) Segment revenue/• China Commerce Wholesale Business", + "level": 5, + "summary": "Revenue from our China commerce wholesale business in the six months ended September 30, 2025 was RMB13,450 million (US\\$1,889 million), an increase of 13% compared to RMB11,938 million in the same period of 2024, primarily due to an increase in revenue from value-added services provided to paying members.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "(ii) Segment adjusted EBITA", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba China E-commerce Group/(ii) Segment adjusted EBITA", + "level": 4, + "summary": "Alibaba China E-commerce Group adjusted EBITA decreased by 47% to RMB48,886 million (US\\$6,867 million) in the six months ended September 30, 2025, compared to RMB93,080 million in the same period of 2024, primarily due to the investment in quick commerce, user experiences, and technology, partly offset by double-digit revenue growth in Alibaba China E-commerce Group.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Alibaba International Digital Commerce Group", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba International Digital Commerce Group", + "level": 3, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "(i) Segment revenue", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba International Digital Commerce Group/(i) Segment revenue", + "level": 4, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "• International Commerce Retail Business", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba International Digital Commerce Group/(i) Segment revenue/• International Commerce Retail Business", + "level": 5, + "summary": "Revenue from our International commerce retail business in the six months ended September 30, 2025 was RMB56,463 million (US\\$7,931 million), an increase of 15% compared to RMB49,309 million in the same period of 2024, primarily driven by the increase in revenue contributed by AliExpress and other international businesses. As certain of our international businesses generate revenue in local currencies while our reporting currency is Renminbi, AIDC's revenue is affected by exchange rate fluctuations.", + "chunk_count": 1, + "children": [] + }, + { + "title": "• International Commerce Wholesale Business", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba International Digital Commerce Group/(i) Segment revenue/• International Commerce Wholesale Business", + "level": 5, + "summary": "Revenue from our International commerce wholesale business in the six months ended September 30, 2025 was RMB13,077 million (US\\$1,837 million), an increase of 12% compared to RMB11,656 million in the same period of 2024, primarily due to an increase in revenue generated by cross-border related value-added services.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "(ii) Segment adjusted EBITA", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Alibaba International Digital Commerce Group/(ii) Segment adjusted EBITA", + "level": 4, + "summary": "Alibaba International Digital Commerce Group adjusted EBITA was a profit of RMB103 million (US\\$14 million) in the six months ended September 30, 2025, compared to a loss of RMB6,611 million in the same period of 2024, primarily due to significant improvement in AliExpress' operating efficiency, and enhanced efficiency across various businesses.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Cloud Intelligence Group", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Cloud Intelligence Group", + "level": 3, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "(i) Segment revenue", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Cloud Intelligence Group/(i) Segment revenue", + "level": 4, + "summary": "Revenue from Cloud Intelligence Group was RMB73,222 million (US\\$10,285 million) in the six months ended September 30, 2025, an increase of 30% compared to RMB56,159 million in the same period of 2024. Overall revenue excluding Alibaba-consolidated subsidiaries increased by 28% year-over-year, primarily driven by public cloud revenue growth, including the increasing adoption of AI-related products.", + "chunk_count": 1, + "children": [] + }, + { + "title": "(ii) Segment adjusted EBITA", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/Cloud Intelligence Group/(ii) Segment adjusted EBITA", + "level": 4, + "summary": "Cloud Intelligence Group adjusted EBITA increased by 31% to RMB6,558 million (US\\$921 million) in the six months ended September 30, 2025, compared to RMB4,998 million in the same period of 2024, primarily due to revenue growth and improving operating efficiency, partly offset by the increasing investments in customer growth and technology innovation.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "All Others", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/All Others", + "level": 3, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "(i) Segment revenue", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/All Others/(i) Segment revenue", + "level": 4, + "summary": "Revenue from All others segment was RMB121,568 million (US\\$17,077 million) in the six months ended September 30, 2025, a decrease of 27% compared to RMB165,837 million in the same period of 2024, primarily due to the revenue decrease as a result of the disposal of Sun Art and Intime businesses, as well as the decrease in revenue from Cainiao, partly offset by the increase in revenue from Freshippo, Alibaba Health and Amap.", + "chunk_count": 1, + "children": [] + }, + { + "title": "(ii) Segment adjusted EBITA", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Segment Results/All Others/(ii) Segment adjusted EBITA", + "level": 4, + "summary": "Adjusted EBITA from All others segment in the six months ended September 30, 2025 was a loss of RMB4,785 million (US\\$672 million), compared to a loss of RMB2,910 million in the same period of 2024, primarily due to the increased investment in technology businesses, partly offset by the improved operating results of Freshippo, Hujing Digital Media and Entertainment Group and Alibaba Health.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Six Months Ended September Other Financial Results Costs and Expenses", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses", + "level": 2, + "summary": "The text details Alibaba's costs and expenses for the six months ended September 30, 2025, compared to the same period in 2024. Cost of revenue decreased to 58.0% of revenue from 60.5%, driven by disposal of Sun Art and Intime, lower direct sales, and improved efficiency. Product development expenses rose to 6.5% of revenue from 5.7% due to increased technology investment. Sales and marketing expenses surged to 24.2% from 13.6%, mainly from investment in Alibaba China E-commerce Group. General and administrative expenses fell to 3.0% from 4.8%, aided by a prior-year lawsuit provision and cost controls. Total share-based compensation expense decreased to RMB6,870 million from RMB8,277 million, due to fewer awards granted. Amortization and impairment of intangible assets dropped 53% to RMB1,633 million from full amortization of certain assets.", + "chunk_count": 14, + "children": [ + { + "title": "Income from operations and operating margin", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Income from operations and operating margin", + "level": 3, + "summary": "Income from operations in the six months ended September 30, 2025 was RMB40,353 million (US\\$5,668 million), or 8% of revenue, a decrease of 43% compared to RMB71,235 million, or 15% of revenue, in the same period of 2024, primarily due to the decrease in adjusted EBITA, partly offset by a one-time provision in the same period last year.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Adjusted EBITDA and Adjusted EBITA", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Adjusted EBITDA and Adjusted EBITA", + "level": 3, + "summary": "Adjusted EBITDA decreased 36% year-over-year to RMB62,991 million (US\\$8,848 million) in the six months ended September 30, 2025, compared to RMB98,488 million in the same period of 2024. Adjusted EBITA decreased 44% year-over-year to RMB47,917 million (US\\$6,731 million) in the six months ended September 30, 2025, compared to RMB85,596 million in the same period of 2024, primarily attributable to the investment in quick commerce, user experiences, and technology, partly offset by double-digit revenue growth in Alibaba China E-commerce Group, the improved operating results supported by continued growth in Cloud business, as well as enhanced operating efficiencies across various businesses. A reconciliation of net income to adjusted EBITDA and adjusted EBITA is included in the section entitled “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” of this Interim Report.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Adjusted EBITA by segment", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Adjusted EBITA by segment", + "level": 3, + "summary": "Adjusted EBITA by segment as well as a reconciliation of income from operations to adjusted EBITA are set forth in the section entitled “Six Months Ended September Segment Results” above.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Interest and investment income, net", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Interest and investment income, net", + "level": 3, + "summary": "Interest and investment income, net in the six months ended September 30, 2025 was RMB37,468 million (US\\$5,263 million), an increase of 119% compared to RMB17,129 million in the same period of 2024, primarily due to mark-to-market changes from our equity investments, gain from the disposal of local consumer service business of Trendyol, and the decrease in impairment of our investments. The above-mentioned investment gains and losses were excluded from our non-GAAP net income.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other (expense) income, net", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Other (expense) income, net", + "level": 3, + "summary": "Other (expense) income, net in the six months ended September 30, 2025 was an income of RMB1,329 million (US\\$187 million), compared to expense of RMB1,221 million in the same period of 2024, was primarily due to the decrease in net exchange loss in the six months ended September 30, 2025 compared to the same period last year, arising from the exchange rate fluctuation between Renminbi and U.S. dollar.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Income tax expenses", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Income tax expenses", + "level": 3, + "summary": "Income tax expenses in the six months ended September 30, 2025 were RMB14,415 million (US\\$2,024 million), compared to RMB17,442 million in the same period of 2024.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Share of results of equity method investees", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Share of results of equity method investees", + "level": 3, + "summary": "Share of results of equity method investees in the six months ended September 30, 2025 was RMB3,254 million (US\\$457 million), an increase of 31% compared to RMB2,483 million in the same period of 2024. The following table sets forth a breakdown of share of results of equity method investees for the periods indicated: [tables/table-5 Equity Investee Results.html] (1) “Others” mainly include basis differences arising from equity method investees, share-based compensation expense related to share-based awards granted to employees of our equity method investees, as well as gain or loss arising from the deemed disposal of the equity method investees. We record our share of results of all equity method investees one quarter in arrears. The share of net profit of other equity method investees recorded in the six months ended September 30 2025, compared to the share of net losses in the same period last year, was primarily attributable to the overall improvement in the financial performance of our equity method investees. This was partly offset by the decrease in share of profit of Ant Group, which was mainly attributable to investments in new growth initiatives and technologies.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Net income and Non-GAAP net income", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Net income and Non-GAAP net income", + "level": 3, + "summary": "Our net income in the six months ended September 30, 2025 was RMB62,994 million (US\\$8,849 million), compared to RMB67,569 million in the same period of 2024, primarily attributable to the decrease in income from operations, partly offset by the mark-to-market changes from our equity investments, gain from the disposal of local consumer service business of Trendyol and the decrease in both the impairment of our investments and net exchange loss. Excluding non-cash share-based compensation expense, gains/losses of investments, impairment of goodwill and intangible assets, and certain other items, non-GAAP net income in the six months ended September 30, 2025 was RMB43,862 million (US\\$6,161 million), a decrease of 43% compared to RMB77,209 million in the same period of 2024, primarily attributable to the investment in quick commerce, user experiences, and technology, partly offset by double-digit revenue growth in Alibaba China E-commerce Group, the improved operating results supported by continued growth in Cloud business, as well as enhanced operating efficiencies across various businesses. A reconciliation of net income to non-GAAP net income is included in the section entitled “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” of this Interim Report.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Net income attributable to ordinary shareholders", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Net income attributable to ordinary shareholders", + "level": 3, + "summary": "Net income attributable to ordinary shareholders for the six months ended September 30, 2025 was RMB64,106 million, down from RMB68,143 million in 2024, due to decreased income from operations, partially offset by investment mark-to-market changes, Trendyol disposal gains, and lower impairment and exchange losses. Diluted earnings per ADS were RMB26.73, compared to RMB28.00 in 2024; non-GAAP diluted earnings per ADS fell 39% to RMB19.10. Diluted earnings per share were RMB3.34, versus RMB3.50 in 2024; non-GAAP diluted earnings per share also decreased 39% to RMB2.39. Cash, cash equivalents, short-term investments, and other treasury investments totaled RMB573,889 million as of September 30, 2025, down from RMB597,132 million as of March 31, 2025, primarily due to free cash flow outflow of RMB40,655 million, dividend payments of RMB33,621 million, and share repurchases of RMB7,638 million, partly offset by proceeds from convertible notes, bank borrowings, exchangeable bonds, and asset disposals.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Net cash provided by operating activities and free cash flow", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Net cash provided by operating activities and free cash flow", + "level": 3, + "summary": "During the six months ended September 30, 2025, net cash provided by operating activities was RMB30,771 million (US\\$4,322 million), a decrease of 53% compared to RMB65,074 million in the same period of 2024. Free cash flow, a non-GAAP measurement of liquidity, was an outflow of RMB40,655 million (US\\$5,711 million), compared to an inflow of RMB31,107 million in the same period of 2024. The decrease in free cash flow was mainly attributed to the increase in our cloud infrastructure expenditure and the investment in quick commerce. A reconciliation of net cash provided by operating activities to free cash flow is included in the section entitled “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” of this Interim Report.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Net cash used in investing activities", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Net cash used in investing activities", + "level": 3, + "summary": "During the six months ended September 30, 2025, net cash used in investing activities of RMB51,324 million (US\\$7,209 million) primarily reflected capital expenditures of RMB70,177 million (US\\$9,858 million) and net increase in short-term investments and other treasury investments by RMB11,119 million (US\\$1,562 million), partly offset by net cash inflow of RMB29,710 million (US\\$4,173 million) for investment and acquisition activities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Net cash provided by financing activities", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Net cash provided by financing activities", + "level": 3, + "summary": "During the six months ended September 30, 2025, net cash provided by financing activities of RMB8,171 million (US\\$1,148 million) primarily reflected cash provided by net proceeds from issuance of convertible unsecured senior notes and payments for capped call transactions of RMB20,994 million (US\\$2,949 million), net proceeds from bank borrowings of RMB17,804 million (US\\$2,501 million), and net proceeds from issuance of exchangeable bonds of RMB10,986 million (US\\$1,543 million), partly offset by dividend payment of RMB33,621 million (US\\$4,723 million), and cash used in repurchase of ordinary shares of RMB7,638 million (US\\$1,073 million).", + "chunk_count": 1, + "children": [] + }, + { + "title": "Employees", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Six Months Ended September Other Financial Results Costs and Expenses/Employees", + "level": 3, + "summary": "The passage provides reconciliations of non-GAAP financial measures to GAAP measures for a company, including adjusted EBITA, adjusted EBITDA, non-GAAP net income, non-GAAP diluted earnings per share/ADS, and free cash flow for the six months ended September 30, 2025. It also notes employee counts: 126,661 as of September 30, 2025, compared to 124,320 as of March 31, 2025. The reconciliations are presented in tables, with footnotes explaining tax effects, ADS-to-ordinary share ratio (1:8), and adjustments for noncontrolling interests.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Non-GAAP Financial Measures", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Non-GAAP Financial Measures", + "level": 2, + "summary": "The passage explains the company's use of non-GAAP financial measures including adjusted EBITDA, adjusted EBITA, non-GAAP net income, non-GAAP diluted earnings per share/ADS, and free cash flow. These measures are used to supplement GAAP financial statements and help identify underlying business trends by excluding certain non-cash expenses and items not reflective of core operating performance. The definitions of each measure are provided, along with cautionary notes that they should not be considered in isolation or as alternatives to GAAP measures, and may not be comparable to similarly titled measures from other companies. Free cash flow is described as a liquidity measure for strategic corporate transactions, with specific adjustments explained.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Financial Information", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information", + "level": 2, + "summary": "", + "chunk_count": 10, + "children": [ + { + "title": "Liquidity and Capital Resources", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Liquidity and Capital Resources", + "level": 3, + "summary": "The passage discusses the company's funding sources, including cash from operations and debt/equity financing. It reports cash from operations of RMB65,074 million and RMB30,771 million for the six months ended September 30, 2024 and 2025, respectively. As of September 30, 2025, unrestricted cash, short-term investments, and other treasury investments totaled RMB573,889 million. The company issued zero coupon exchangeable bonds due 2032 for HK$12 billion in July 2025 and zero coupon convertible unsecured senior notes due 2032 for US$3.2 billion in September 2025. It amended a syndicated loan facility, reducing it to US$3.17 billion and restructuring it as a revolving credit facility expiring September 30, 2028, with an option to extend to 2030. Another US$6.5 billion revolving credit facility was amended to US$3.33 billion with similar terms. The company monitors financial health via total debts to adjusted EBITDA ratio (1.14 as of March 2025, 1.69 as of September 2025) and total debts to total capital ratio (17.62% and 20.35%, respectively).", + "chunk_count": 1, + "children": [] + }, + { + "title": "Significant Investments", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Significant Investments", + "level": 3, + "summary": "Our significant investment consists of Ant Group. Ant Group provides comprehensive digital payment services and facilitates digital financial and value-added services for consumers and merchants, in China and across the world. As of September 30, 2025, our equity interest in Ant Group on a fully diluted basis was 33%. During the six months ended September 30, 2025, dividend received from Ant Group amounted to RMB3,293 million (US\\$463 million). We did not hold any other significant investments as of September 30, 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Material Investments, Acquisitions and Dispositions", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Material Investments, Acquisitions and Dispositions", + "level": 3, + "summary": "Our material investments, acquisitions and dispositions in the six months ended September 30, 2025 and the period through the date of this Interim Report are set forth below. In May 2025, we agreed to sell 85% of the equity interest in Trendyol GO, a wholly-owned subsidiary of Trendyol that operates local service business in Türkiye (the “Disposal”). The cash consideration for the Disposal is approximately US\\$0.7 billion (RMB5 billion). The Disposal was completed during the six months ended September 30, 2025. Save as disclosed above, as at the date of this Interim Report, the Group did not have detailed future plans for material investments or capital assets.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Pledge of Assets", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Pledge of Assets", + "level": 3, + "summary": "Certain of the Group's bank borrowings are collateralized by a pledge of certain buildings and property improvements, construction in progress and land use rights in the PRC, receivables and other treasury investments with carrying values of RMB30,213 million and RMB25,417 million, as of March 31, 2025 and September 30, 2025, respectively. In addition, certain of the Group's payables are collateralized by a pledge of certain short-term investments and other treasury investments with carrying values of RMB3,697 million and RMB4,155 million as of March 31, 2025 and September 30, 2025, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Foreign Exchange Risk", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Foreign Exchange Risk", + "level": 3, + "summary": "The passage discusses foreign currency risk arising from transactions and investments. Most revenue and expenses are in Renminbi, the functional currency. The Renminbi's value against the U.S. dollar fluctuates due to political, economic, and policy factors. Appreciation of the Renminbi reduces U.S. dollar conversion amounts for operations, while U.S. dollar appreciation reduces Renminbi conversion for dividends or debt payments.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Contingent Liabilities", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Contingent Liabilities", + "level": 3, + "summary": "Save as disclosed in this Interim Report, as at September 30, 2025, the Group had no material contingent liabilities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Capital Expenditure and Capital Commitment", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Capital Expenditure and Capital Commitment", + "level": 3, + "summary": "Our capital expenditures have been incurred primarily in relation to (i) the acquisition of computer equipment and construction of data centers relating to our Cloud business and the operation of our mobile platforms and websites; (ii) the acquisition of infrastructure for logistics services and direct sales businesses; and (iii) the acquisition of land use rights and construction of corporate campuses and office facilities. In the six months ended September 30, 2024 and 2025, our capital expenditures totaled RMB29,585 million and RMB70,177 million (US\\$9,858 million), respectively. The Group’s capital commitments primarily relate to capital expenditures contracted for purchase of property and equipment, including the construction of corporate campuses. Total capital commitments contracted but not provided for amounted to RMB45,321 million and RMB29,483 million (US\\$4,141 million) as of March 31, 2025 and September 30, 2025, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Remuneration Policy", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Remuneration Policy", + "level": 3, + "summary": "Alibaba Group reviews its remuneration policy to maintain industry competitiveness, offering discretionary bonuses and long-term incentives based on performance. It provides learning programs for employee development. PRC subsidiaries participate in government-mandated defined contribution plans covering housing, pension, medical, and other benefits, with monthly contributions to local authorities. Commercial health and accidental insurance are also provided, along with tailored benefit plans. Outside the PRC, subsidiaries make payments to defined contribution and benefit plans. Equity incentives, including restricted share units, stock options, and share appreciation rights, may be granted to directors, employees, service providers, and consultants under adopted plans.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Subsequent Events", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/Weighted Voting Rights/Other Financial Information/Subsequent Events", + "level": 3, + "summary": "Save as disclosed in this Interim Report, as at the date of this Interim Report, there were no significant events that might affect Alibaba Group since September 30, 2025.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "DIRECTORS AND CHIEF EXECUTIVE OFFICER", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER", + "level": 1, + "summary": "The following table sets forth certain information relating to our directors and chief executive officer. [tables/table-10 Board of Directors.html] $^{\\dagger}$ Director nominated by the Alibaba Partnership. (a) Group I directors. Current term of office will expire at our 2027 annual general meeting. (b) Group II directors. Current term of office will expire at our 2028 annual general meeting. (c) Group III directors. Current term of office will expire at our 2026 annual general meeting. (d) Beginning from April 1, 2025, Maggie Wu has started to serve as a non-executive director. (1) The business address of our directors is 26/F Tower One, Times Square, 1 Matheson Street, Causeway Bay, Hong Kong SAR., the People's Republic of China.", + "chunk_count": 9, + "children": [ + { + "title": "Biographical Information", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Biographical Information", + "level": 2, + "summary": "The text provides detailed biographies of 10 members of Alibaba Group's board of directors, including Joseph C. Tsai (Chairman), Eddie Yongming Wu (CEO), J. Michael Evans (President), Maggie Wei Wu, Jerry Yang, Wan Ling Martello, Weijian Shan, Irene Yun-Lien Lee, Albert Kong Ping Ng, and Kabir Misra. Each entry outlines their professional background, education, key roles at Alibaba and other companies, and board memberships.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Board Committees", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees", + "level": 2, + "summary": "Our board of directors has established an audit committee, a compensation committee, a nominating and corporate governance committee, a sustainability committee, a compliance and risk committee and a capital management committee.", + "chunk_count": 7, + "children": [ + { + "title": "Audit Committee", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Audit Committee", + "level": 3, + "summary": "Our audit committee currently consists of Albert Ng, Wan Ling Martello and Weijian Shan. Mr. Ng is the chairman of our audit committee.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Compensation Committee", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Compensation Committee", + "level": 3, + "summary": "Our compensation committee currently consists of Jerry Yang, Albert Ng and Kabir Misra. Mr. Yang is the chairman of our compensation committee.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Nominating and Corporate Governance Committee", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Nominating and Corporate Governance Committee", + "level": 3, + "summary": "Our nominating and corporate governance committee currently consists of Irene Lee and Jerry Yang. Ms. Lee is the chairman of our nominating and corporate governance committee.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Sustainability Committee", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Sustainability Committee", + "level": 3, + "summary": "Our sustainability committee currently consists of Jerry Yang, Joe Tsai and Maggie Wu. Mr. Yang is the chairman of our sustainability committee.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Compliance and Risk Committee", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Compliance and Risk Committee", + "level": 3, + "summary": "Our compliance and risk committee currently consists of Irene Lee, Albert Ng, Kabir Misra and J. Michael Evans. Ms. Lee is the chairman of our compliance and risk committee.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Capital Management Committee", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DIRECTORS AND CHIEF EXECUTIVE OFFICER/Board Committees/Capital Management Committee", + "level": 3, + "summary": "Our capital management committee currently consists of Joe Tsai, Eddie Wu, J. Michael Evans and Maggie Wu. Mr. Tsai is the chairman of our capital management committee.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "DISCLOSURE OF INTERESTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DISCLOSURE OF INTERESTS", + "level": 1, + "summary": "The document details the interests and short positions in shares of the company as of September 30, 2025, for directors, the CEO, and substantial shareholders, as required under the Securities and Futures Ordinance (SFO). It includes tables and notes for directors like Joseph C. Tsai, Eddie Yongming Wu, J. Michael Evans, and Maggie Wei Wu, listing their holdings through shares, RSUs, options, trusts, and foundations. For substantial shareholders, it covers JPMorgan Chase & Co., Citigroup Inc., and Blackrock, Inc., with breakdowns of long and short positions, including derivatives. The report confirms no other disclosable interests beyond those listed.", + "chunk_count": 1, + "children": [] + }, + { + "title": "EQUITY INCENTIVE PLANS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS", + "level": 1, + "summary": "The 2014 Plan, the 2024 Plan and the 2024 Plan (Existing Shares) are the equity incentive plans of our Company in effect. For a summary of the key terms of these plans, please see “Directors, Senior Management and Employees - Equity Incentive Plans” in our fiscal year 2025 annual report. During the Reporting Period, a total of 63,594,318 Shares (equivalent to approximately 7,949,290 ADSs), representing approximately 0.3% of the weighted average number of Shares in issue (excluding treasury shares) of our Company, that may be issued in respect of all share-based awards were granted to eligible participants under the 2024 Plan of our Company.", + "chunk_count": 6, + "children": [ + { + "title": "The 2014 Plan", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS/The 2014 Plan", + "level": 2, + "summary": "The 2014 Plan (which we adopted in September 2014, amended and restated in February 2020 to reflect the Share Split and other administrative changes, and further amended and restated in May 2022 to reflect administrative changes) provides for the granting of share-based awards, including restricted share units, stock options, restricted shares and share appreciation rights, to any participant who is an employee, consultant or director of our company, our affiliates and/or certain other companies. No further awards will be granted under the 2014 Plan as from September 18, 2024. Any shares authorized but unissued under the 2014 Plan will no longer be available for granting. The share-based awards already granted under the 2014 Plan will remain in full force and effect pursuant to the terms and conditions of the 2014 Plan. Details of movements of RSUs under the 2014 Plan during the Reporting Period are as follows: [tables/table-13 RSU Grant Summary.html]", + "chunk_count": 3, + "children": [ + { + "title": "Notes:", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS/The 2014 Plan/Notes:", + "level": 3, + "summary": "1. The purchase price of all RSUs granted is nil. 2. Eight Shares are issuable upon the vesting of each RSU. 3. The RSUs vest over a period of 6 years. 4. The RSUs vest over a period of 5 years. 5. The RSUs vest over a period of 4 years. 6. The RSUs vest over a period of 2 years. 7. The RSUs granted to employee participants and related entity participants vest over a period of up to 10 years. 8. For directors and chief executive officer, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the RSUs were vested during the Reporting Period was US\\$136.71. 9. For employee participants, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the RSUs were vested during the Reporting Period was US\\$131.05 10. For related entity participants, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the RSUs were vested during the Reporting Period was US\\$132.15. 11. A related entity participant is an employee of an associated company in which our Company owns, directly or indirectly, securities or interests representing 20% or more of its voting power. Details of movements of share options under the 2014 Plan during the Reporting Period are as follows: [tables/table-14 Share Options Table.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Notes:_2", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS/The 2014 Plan/Notes:_2", + "level": 3, + "summary": "1. Eight Shares are issuable upon the exercise of each share option. 2. The share options vest over a period of 5 years and are exercisable within 10 years after date of grant. 3. The share options vest over a period of 6 years and are exercisable within 12 years after date of grant. 4. The share options vest over a period of 2 years and are exercisable within 8 years after date of grant. 5. The share options vest over a period of up to 9 years and are exercisable within up to 12 years after date of grant. 6. The closing price of ADSs traded on the NYSE immediately before the date on which the share options were exercised during the Reporting Period was US\\$123.90. 7. For employee participants, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the share options were exercised during the Reporting Period was US\\$114.78. 8. For related entity participant, the weighted average closing price of ADSs traded on the NYSE immediately before the date on which the share options were exercised during the Reporting Period was US\\$118.09. 9. No share options granted were cancelled during the Reporting Period. 10. A related entity participant is an employee of an associated company in which our Company owns, directly or indirectly, securities or interests representing 20% or more of its voting power.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "The 2024 Plan", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS/The 2024 Plan", + "level": 2, + "summary": "The 2024 Plan, approved in August 2024, aims to attract, motivate, and retain employees, executive directors, and service providers by granting restricted share units (RSUs), stock options, restricted shares, and share appreciation rights. As of April 1, 2025, 473,488,465 shares were available for grant under the scheme mandate, and 93,716,369 shares under the service provider sub-limit. By September 30, 2025, available shares under the scheme mandate decreased to 411,246,946. RSUs granted have a nil purchase price, vest over up to 6 years, and one share is issuable per RSU. No RSUs were granted to directors during the reporting period. The weighted average closing price before vesting was HK$113.70. Grants to employee participants on various dates are detailed in a table, with some RSUs granted on May 23, 2025 subject to performance targets based on financial or operational indicators. Fair values of RSUs were determined by reference to underlying share fair values on grant date. No RSUs were cancelled during the period.", + "chunk_count": 1, + "children": [] + }, + { + "title": "The 2024 Plan (Existing Shares)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/EQUITY INCENTIVE PLANS/The 2024 Plan (Existing Shares)", + "level": 2, + "summary": "The 2024 Plan (Existing Shares) was approved by our board in August 2024. The purpose of the 2024 Plan (Existing Shares) is to provide incentives to attract, motivate and retain the grantees and align the interests of the grantees with those of our shareholders. It provides for the granting of restricted share units, stock options, restricted shares and share appreciation rights to any participant who is an employee, a director or a service provider. The 2024 Plan (Existing Shares) is funded by existing Shares, therefore it does not constitute a scheme involving the new issuance of Shares as referred to in Chapter 17 of the Hong Kong Listing Rules but is subject to the applicable disclosure requirements under Rule 17.12 of the Hong Kong Listing Rules.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES", + "level": 1, + "summary": "During the Reporting Period, our Company repurchased a total of 73 million Shares on the NYSE for an aggregate consideration of US\\$1.0 billion. Details of the Shares repurchased on the NYSE are as follows: [tables/table-17 Share Repurchase.html] Note: 1. Each ADS represents eight Shares. As of the date of this Interim Report, all the Shares repurchased during the Reporting Period have been cancelled. Save as disclosed above, neither our Company nor any of its subsidiaries purchased, sold or redeemed any of our Company's securities listed on the Hong Kong Stock Exchange or the NYSE (including sale of treasury shares) during the Reporting Period. As of September 30, 2025, our Company did not hold any treasury shares as defined in the Hong Kong Listing Rules.", + "chunk_count": 1, + "children": [] + }, + { + "title": "OTHER INFORMATION", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION", + "level": 1, + "summary": "", + "chunk_count": 7, + "children": [ + { + "title": "Corporate Governance", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Corporate Governance", + "level": 2, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "Compliance with the Corporate Governance Code", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Corporate Governance/Compliance with the Corporate Governance Code", + "level": 3, + "summary": "To the knowledge of the Company and our directors, we have complied with all applicable code provisions set out in the Corporate Governance Code as set forth in Part 2 of Appendix C1 to the Hong Kong Listing Rules for the six months ended September 30, 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Compliance with the Model Code", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Corporate Governance/Compliance with the Model Code", + "level": 3, + "summary": "We have adopted our own trading guidelines, on terms no less exacting than the required standard set out in the Model Code as set forth in Appendix C3 to the Hong Kong Listing Rules, to regulate, among others, all dealings by directors and relevant employees of securities in the Company. Having made specific enquiry of all directors, all directors confirmed that they have complied with our trading guidelines during the six months ended September 30, 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Audit Committee Review", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Corporate Governance/Audit Committee Review", + "level": 3, + "summary": "The audit committee oversees accounting, financial reporting, and audits. Its responsibilities include selecting and evaluating the independent auditor, pre-approving audit and non-audit services, assessing internal controls, reviewing audit problems and related party transactions, and discussing quarterly, half-year, and annual reports. It also establishes procedures for employee complaints on accounting matters and meets separately with management and auditors. The committee reviewed the unaudited interim financial statements for the six months ended September 30, 2025, with PricewaterhouseCoopers, which also reviewed the statements under International Standard on Review Engagements 2410.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Interim Dividend", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Interim Dividend", + "level": 2, + "summary": "Our board did not recommend the distribution of an interim dividend for the Reporting Period.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Use of Proceeds from the Notes Offerings", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/OTHER INFORMATION/Use of Proceeds from the Notes Offerings", + "level": 2, + "summary": "In September 2025, the company completed a private offering of $3.168 billion in Zero Coupon Convertible Senior Notes due 2032, along with capped call transactions costing $184 million. The notes have an initial conversion rate of 5.1773 ADSs per $1,000 principal, equivalent to an initial conversion price of about $193.15 per ADS. Net proceeds of approximately $3.13 billion are intended for general corporate purposes, with 80% allocated to enhancing cloud infrastructure and 20% to expanding international commerce operations. As of September 30, 2025, unutilized proceeds totaled $2.4 billion.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "DEFINITIONS Conventions that apply to this Interim Report", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/DEFINITIONS Conventions that apply to this Interim Report", + "level": 1, + "summary": "This section defines key terms and conventions used in the interim report, including references to equity incentive plans (2014 Plan, 2024 Plan, 2024 Plan Existing Shares), ADSs, Alibaba Group, Articles of Association, board of directors, Corporate Governance Code, Hong Kong Listing Rules, Model Code, NYSE, Reporting Period, RSUs, SEC, SFO, Share Split, shareholders, Shares, substantial shareholders, U.S., U.S. dollars, U.S. Exchange Act, and U.S. Securities Act.", + "chunk_count": 1, + "children": [] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED", + "level": 1, + "summary": "", + "chunk_count": 6, + "children": [ + { + "title": "INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED/INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS", + "level": 2, + "summary": "[tables/table-19 Interim Report.html] Report on Review of Interim Financial Information To the Board of Directors of Alibaba Group Holding Limited (incorporated in the Cayman Islands with limited liability)", + "chunk_count": 5, + "children": [ + { + "title": "Introduction", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED/INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS/Introduction", + "level": 3, + "summary": "We have reviewed the interim financial information set out on pages 46 to 81, which comprises the condensed consolidated balance sheet of Alibaba Group Holding Limited (the “Company”) and its subsidiaries (together, the “Group”) as at September 30, 2025 and the condensed consolidated income statement, the condensed consolidated statement of comprehensive income, the condensed consolidated statement of changes in shareholders’ equity and the condensed consolidated statement of cash flows for the six-month period then ended, and selected explanatory notes. The Rules Governing the Listing of Securities on The Stock Exchange of Hong Kong Limited require the preparation of a report on interim financial information to be in compliance with the relevant provisions thereof and accounting principles generally accepted in the United States of America (“U.S. GAAP”). The directors of the Company are responsible for the preparation and presentation of this interim financial information in accordance with U.S. GAAP. Our responsibility is to express a conclusion on this interim financial information based on our review and to report our conclusion solely to you, as a body, in accordance with our agreed terms of engagement, and for no other purpose. We do not assume responsibility towards or accept liability to any other person for the contents of this report.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Scope of Review", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED/INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS/Scope of Review", + "level": 3, + "summary": "We conducted our review in accordance with International Standard on Review Engagements 2410, “Review of Interim Financial Information Performed by the Independent Auditor of the Entity”. A review of interim financial information consists of making inquiries, primarily of persons responsible for financial and accounting matters, and applying analytical and other review procedures. A review is substantially less in scope than an audit conducted in accordance with International Standards on Auditing and consequently does not enable us to obtain assurance that we would become aware of all significant matters that might be identified in an audit. Accordingly, we do not express an audit opinion.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Conclusion", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED/INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS/Conclusion", + "level": 3, + "summary": "Based on our review, nothing has come to our attention that causes us to believe that the interim financial information of the Group is not prepared, in all material respects, in accordance with U.S. GAAP.", + "chunk_count": 1, + "children": [] + }, + { + "title": "PricewaterhouseCoopers Certified Public Accountants", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED/INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS/PricewaterhouseCoopers Certified Public Accountants", + "level": 3, + "summary": "Hong Kong, December 4, 2025 ALIBABA GROUP HOLDING LIMITED UNAUDITED CONDENSED CONSOLIDATED INCOME STATEMENTS [tables/table-20 Alibaba Financials.html] The accompanying notes form an integral part of these unaudited condensed consolidated financial statements. ALIBABA GROUP HOLDING LIMITED UNAUDITED CONDENSED CONSOLIDATED STATEMENTS OF COMPREHENSIVE INCOME [tables/table-21 Income Statement.html] The accompanying notes form an integral part of these unaudited condensed consolidated financial statements. ALIBABA GROUP HOLDING LIMITED UNAUDITED CONDENSED CONSOLIDATED BALANCE SHEETS [tables/table-22 Balance Sheet Data.html] The accompanying notes form an integral part of these unaudited condensed consolidated financial statements. ALIBABA GROUP HOLDING LIMITED UNAUDITED CONDENSED CONSOLIDATED BALANCE SHEETS (CONTINUED) [tables/table-23 Equity Summary.html] The accompanying notes form an integral part of these unaudited condensed consolidated financial statements. ALIBABA GROUP HOLDING LIMITED UNAUDITED CONDENSED CONSOLIDATED STATEMENTS OF CHANGES IN SHAREHOLDERS' EQUITY [tables/table-24 Equity Changes.html] The accompanying notes form an integral part of these unaudited condensed consolidated financial statements.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_2", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_2", + "level": 1, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "UNAUDITED CONDENSED CONSOLIDATED STATEMENTS OF CASH FLOWS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_2/UNAUDITED CONDENSED CONSOLIDATED STATEMENTS OF CASH FLOWS", + "level": 2, + "summary": "[tables/table-25 Cash Flow Statement.html] The accompanying notes form an integral part of these unaudited condensed consolidated financial statements. ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "1. Organization and principal activities", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_2/1. Organization and principal activities", + "level": 2, + "summary": "Alibaba Group Holding Limited (the “Company”) is a limited liability company, which was incorporated in the Cayman Islands on June 28, 1999. The Company is a holding company and conducts its businesses primarily through its subsidiaries and variable interest entities. In these unaudited condensed consolidated financial statements, where appropriate, the term “Company” also refers to its subsidiaries as a whole. The Company provides the technology infrastructure and marketing reach to help merchants, brands, retailers and other businesses to leverage the power of new technology to engage with their users and customers and operate in a more efficient way. The Company's businesses comprise Alibaba China E-commerce Group, Alibaba International Digital Commerce Group, Cloud Intelligence Group and All others. An ecosystem has developed around the Company's platforms and businesses that consists of consumers, merchants, brands, retailers, enterprises, third-party service providers, strategic alliance partners and other businesses. The Company's American depositary shares (“ADSs”) are listed on the New York Stock Exchange (“NYSE”) under the symbol “BABA” and the Company’s ordinary shares are listed on the Hong Kong Stock Exchange (“HKSE”) under the stock codes “9988 (HKD Counter)” and “89988 (RMB Counter).”", + "chunk_count": 1, + "children": [] + }, + { + "title": "2. Summary of significant accounting policies", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_2/2. Summary of significant accounting policies", + "level": 2, + "summary": "The document outlines the basis of presentation for unaudited condensed consolidated financial statements prepared under U.S. GAAP and Hong Kong Listing Rules for interim periods. It notes that these statements do not include all disclosures required for a full set of financials, but include normal recurring adjustments. The statements should be read with the audited financials for the year ended March 31, 2025, with no significant accounting policy changes except updates for derivatives, hedging, and borrowings due to exchangeable bonds. RMB amounts are translated to US dollars at a rate of 7.1190 for convenience, with no guarantee of convertibility at that rate.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_3", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3", + "level": 1, + "summary": "", + "chunk_count": 13, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "2. Summary of significant accounting policies (Continued)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)", + "level": 2, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "(b) Use of estimates", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)/(b) Use of estimates", + "level": 3, + "summary": "The preparation of the unaudited condensed consolidated financial statements in conformity with U.S. GAAP requires the Company to make estimates and assumptions that affect the reported amounts of assets and liabilities, disclosure of contingent assets and liabilities as of the date of the unaudited condensed consolidated financial statements and the reported amounts of revenue and expenses during the reporting period. The Company bases its estimates on historical experience and on various other assumptions that are believed to be reasonable, the results of which form the basis for making judgments about the carrying values of assets and liabilities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "(c) Consolidation", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)/(c) Consolidation", + "level": 3, + "summary": "The unaudited condensed consolidated financial statements include the financial statements of the Company and its subsidiaries, which include the PRC-registered entities directly or indirectly owned by the Company (“WFOEs”) and variable interest entities (“VIEs”) over which the Company is the primary beneficiary for accounting purposes only. All transactions and balances among the Company, its subsidiaries and the VIEs have been eliminated upon consolidation. The results of subsidiaries acquired or disposed of are recorded in the unaudited condensed consolidated income statements from the effective date of acquisition or up to the effective date of disposal, as appropriate. The nature of the businesses and activities of the consolidated VIEs have not changed materially from the preceding fiscal year. The following financial information of the consolidated VIEs and their subsidiaries was recorded in the accompanying unaudited condensed consolidated financial statements: [tables/table-26 Financial Data.html] (i) Revenue generated by the VIEs are primarily from cloud services, digital media and entertainment services and others. ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "2. Summary of significant accounting policies (Continued)_2", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_2", + "level": 2, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "(d) Derivatives and hedging", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_2/(d) Derivatives and hedging", + "level": 3, + "summary": "The passage discusses the accounting treatment for derivatives and embedded derivatives under ASC 815. Derivatives are recorded at fair value on the balance sheet, with changes recognized in income or other comprehensive income depending on hedge designation. Capped call transactions meeting the scope exception are recognized in shareholders' equity. Hedge accounting requires formal documentation and effectiveness testing, with a hedging relationship considered effective if results fall within an 80% to 125% ratio.", + "chunk_count": 3, + "children": [ + { + "title": "Cash flow hedges", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_2/(d) Derivatives and hedging/Cash flow hedges", + "level": 4, + "summary": "Interest rate swaps designated as hedging instruments to hedge against the cash flows attributable to recognized assets or liabilities or forecasted payments may qualify as cash flow hedges. The Company entered into interest rate swap contracts to swap floating interest payments related to certain borrowings for fixed interest payments to hedge the interest rate risk associated with certain forecasted payments and obligations. All changes in the fair value of interest rate swaps that are designated and qualify as cash flow hedges are recognized in accumulated other comprehensive income. Amounts in accumulated other comprehensive income are reclassified into earnings in the same period during which the hedged forecasted transaction affects earnings. The Company has elected the optional expedients under ASC 848 “Reference Rate Reform” for certain existing interest rate swaps that are designated as cash flow hedges in the hedging relationship designation and the assessment of probability of forecasted transaction and hedge effectiveness.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Net investment hedges", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_2/(d) Derivatives and hedging/Net investment hedges", + "level": 4, + "summary": "The Company uses cross currency swap contracts and RMB denominated unsecured senior notes and borrowings to hedge foreign currency risk from investments in PRC subsidiaries, designated as net investment hedges. Changes in fair value from spot exchange rate fluctuations are recognized in accumulated other comprehensive income, while excluded components are recognized in interest expenses. For the six months ended September 30, 2025, losses of RMB1,279 million were recognized in accumulated other comprehensive income. Amounts are reclassified to income upon subsidiary disposal, and hedge accounting is discontinued if ineffective. Fair values are estimated using market rates and industry standard valuation techniques.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "2. Summary of significant accounting policies (Continued)_3", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_3", + "level": 2, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "(e) Borrowings", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_3/(e) Borrowings", + "level": 3, + "summary": "Borrowings consist of bank borrowings, unsecured senior notes, convertible unsecured senior notes and exchangeable bonds. Bank borrowings and unsecured senior notes are recognized initially at fair value, net of upfront fees, debt discounts or premiums, debt issuance costs and other incidental fees. Upfront fees, debt discounts or premiums, debt issuance costs and other incidental fees are recorded as a reduction of the proceeds received and the related accretion is recorded as interest expense in the unaudited condensed consolidated income statements over the estimated term of the facilities using the effective interest method. Convertible unsecured senior notes are accounted for in its entirety as liabilities, and the embedded conversion feature is not required to be accounted for separately under ASC 815. Exchangeable bonds are accounted for under the fair value option with changes in fair value recorded in the unaudited condensed consolidated income statements.", + "chunk_count": 1, + "children": [] + }, + { + "title": "(f) Recent accounting pronouncements", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/2. Summary of significant accounting policies (Continued)_3/(f) Recent accounting pronouncements", + "level": 3, + "summary": "The passage discusses four FASB Accounting Standards Updates (ASUs) issued in 2025. ASU 2025-05 provides a practical expedient for estimating credit losses on current receivables and contract assets, effective for the year ending March 31, 2027. ASU 2025-06 removes prescriptive software development stages for internal-use software, effective for the year ending March 31, 2029. ASU 2025-08 defines purchased seasoned loans and expands the gross-up approach for credit losses, effective for the year ending March 31, 2028. ASU 2025-09 clarifies hedge accounting, including expanding hedged risks and eliminating certain tests, effective for the year ending March 31, 2028. All updates permit early adoption, and the Company is evaluating their impacts.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "3. Significant mergers and acquisitions, investments and dispositions", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_3/3. Significant mergers and acquisitions, investments and dispositions", + "level": 2, + "summary": "Disposal of Trendyol GO In May 2025, the Company entered into a sale and purchase agreement to sell 85% of the equity interest in Trendyol GO, a wholly-owned subsidiary of Trendyol that operates local service business in Türkiye (the \"Disposal\"). The cash consideration for the Disposal is approximately US\\$0.7 billion (RMB5 billion). The Disposal was completed during the six months ended September 30, 2025, and a gain arising from the Disposal of approximately RMB6 billion was recorded in interest and investment income, net in the unaudited condensed consolidated income statements for the six months ended September 30, 2025, primarily taking into consideration (i) the cash considerations received, (ii) the carrying values of the net assets of Trendyol GO, and (iii) the fair value of the 15% retained equity interest in Trendyol GO.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_4", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_4", + "level": 1, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_4/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "4. Revenue", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_4/4. Revenue", + "level": 2, + "summary": "The passage details revenue breakdown by segment for Alibaba. Key segments include E-commerce (Alibaba China E-commerce Group with customer management, goods, logistics), Quick commerce (Taobao Instant Commerce, Ele.me), China commerce wholesale (1688.com), International commerce retail (AliExpress, Trendyol, Lazada), International commerce wholesale (Alibaba.com), Cloud Intelligence Group (cloud services), and All others (Freshippo, Cainiao, Alibaba Health, etc.). Inter-segment eliminations primarily involve Cloud Intelligence Group and Cainiao.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_5", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_5", + "level": 1, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_5/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "4. Revenue (Continued)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_5/4. Revenue (Continued)", + "level": 2, + "summary": "(x) As a result of the change in composition in reportable segments (Note 18), the Company reclassified revenue by segment. Comparative figures for the six months ended September 30, 2024 were updated to conform to the segment presentation. Revenue by type is as follows: [tables/table-28 Revenue Breakdown.html] (i) Customer management services mainly include cost-per-click (\"CPC\"), cost-per-thousand impressions (\"CPM\"), time-based and cost-per-sale (\"CPS\") marketing services. (ii) Other revenue includes revenue from self-developed online games and multiple services provided through various platforms and businesses. The amount of revenue recognized for performance obligations satisfied (or partially satisfied) in prior periods for contracts with expected duration of more than one year during the six months ended September 30, 2024 and 2025 were not material. Deferred revenue and customer advances of the Company primarily represent service fees prepaid by merchants or customers for which the relevant services have not been provided. Substantially all of the balances of deferred revenue and customer advances are generally recognized as revenue within one year.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_6", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_6", + "level": 1, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_6/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "5. Income tax expenses", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_6/5. Income tax expenses", + "level": 2, + "summary": "Composition of income tax expenses is as follows: [tables/table-29 Tax Expense Data.html] The Company's effective tax rate for the six months ended September 30, 2024 and 2025 was $21\\%$ and $19\\%$ , respectively. For the interim financial reporting, the Company estimates the annual tax rate based on projected taxable income for the full year and records an income tax provision for the interim period in accordance with the guidance on accounting for income taxes in an interim period. The tax status of the subsidiaries of the Company with major taxable profits is described below: \\- Alibaba (China) Technology Co., Ltd. (“Alibaba China”), Taobao (China) Software Co., Ltd. (“Taobao China”) and Zhejiang Tmall Technology Co., Ltd. (“Tmall China”), entities primarily engaged in the operations of the Company’s wholesale marketplaces, Taobao and Tmall, respectively, and Alibaba (Beijing) Software Services Co., Ltd (“Alibaba Beijing”) and Alibaba (China) Co., Ltd (“China Co.”), entities primarily engaged in the operations of technology, software research and development and relevant services, were qualified as High and New Technology Enterprises. For the taxation years of 2024 and 2025, Alibaba China, Taobao China, Tmall China, Alibaba Beijing and China Co. applied an EIT rate of 15% as High and New Technology Enterprises. Most of the remaining PRC entities of the Company are subject to EIT at 25% for the six months ended September 30, 2024 and 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "6. Share-based compensation expense by function", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_6/6. Share-based compensation expense by function", + "level": 2, + "summary": "Share-based compensation expense by function is as follows: [tables/table-30 Cost Breakdown.html]", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_7", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_7", + "level": 1, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_7/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "7. Earnings per share∕ADS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_7/7. Earnings per share∕ADS", + "level": 2, + "summary": "Each ADS represents eight ordinary shares. Basic earnings per share is computed by dividing net income attributable to ordinary shareholders by the weighted average number of outstanding ordinary shares, adjusted for treasury shares. Basic earnings per ADS is derived from the basic earnings per share. For the calculation of diluted earnings per share, net income attributable to ordinary shareholders for basic earnings per share is adjusted by the effect of dilutive securities, including share-based awards under the treasury stock method and convertible unsecured senior notes and exchangeable bonds under the if-converted method. Certain potentially dilutive instruments, including the exchangeable bonds and the share lending arrangement in connection with the issuance of the exchangeable bonds, and the capped call transactions in connection with the issuance of the convertible unsecured senior notes have been excluded from the computation of diluted net income per share as their inclusion is anti-dilutive. Diluted earnings per ADS is derived from the diluted earnings per share. The following table sets forth the computation of basic and diluted net income per share/ADS for the following periods: [tables/table-31 EPS Data Table.html]", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_8", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8", + "level": 1, + "summary": "", + "chunk_count": 7, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "8. Equity securities and other investments", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/8. Equity securities and other investments", + "level": 2, + "summary": "[tables/table-32 Investment Summary.html]", + "chunk_count": 3, + "children": [ + { + "title": "Equity securities", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/8. Equity securities and other investments/Equity securities", + "level": 3, + "summary": "Net unrealized gains, including impairment losses, on equity securities, recognized in interest and investment income, net were RMB4,159 million and RMB20,017 million for the six months ended September 30, 2024 and 2025, respectively. Investments in privately held companies include equity investments for which the Company elected to account for using the measurement alternative, for which the carrying value as of March 31 and September 30, 2025 were RMB88,728 million and RMB112,829 million, respectively. For the equity investments accounted for using the measurement alternative, the cumulative upward adjustments as of March 31 and September 30, 2025 were RMB27,197 million and RMB45,334 million, respectively, and the cumulative impairments and downward adjustments as of March 31 and September 30, 2025 were RMB44,232 million and RMB45,741 million, respectively. Upward adjustments recorded in the unaudited condensed consolidated income statements on equity investments accounted for using the measurement alternative were RMB4,614 million and RMB20,241 million for the six months ended September 30, 2024 and 2025, respectively, and impairments and downward adjustments recorded in the unaudited condensed consolidated income statements were RMB4,918 million and RMB4,338 million for the same periods, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Debt investments", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/8. Equity securities and other investments/Debt investments", + "level": 3, + "summary": "Debt investments include convertible and exchangeable bonds accounted for under the fair value option, for which the fair value as of March 31 and September 30, 2025 were RMB963 million and RMB759 million, respectively. The aggregate fair value of these convertible and exchangeable bonds was lower than their aggregate unpaid principal balance as of March 31 and September 30, 2025 by RMB2,420 million and RMB2,529 million, respectively. Unrealized losses recorded in the unaudited condensed consolidated income statements on these convertible and exchangeable bonds were RMB32 million and RMB158 million for the six months ended September 30, 2024 and 2025, respectively. ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "8. Equity securities and other investments (Continued)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/8. Equity securities and other investments (Continued)", + "level": 2, + "summary": "The passage discusses debt investments accounted for at amortized cost, with allowances for credit losses of RMB3,779 million and RMB2,641 million as of March 31 and September 30, 2025. Reversal of impairment losses were RMB1,117 million and RMB1,103 million for the six months ended September 30, 2024 and 2025. Loans to shareholders of equity method investees totaling RMB5,518 million principal are expected to be repaid through collateral sale, with collateral fair values of RMB4,325 million and RMB6,250 million as of March 31 and September 30, 2025. Carrying amounts approximate fair value due to comparable interest rates. Other treasury investments include fixed deposits, certificates of deposit, and marketable debt securities with maturities over one year.", + "chunk_count": 1, + "children": [] + }, + { + "title": "9. Fair value measurement", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_8/9. Fair value measurement", + "level": 2, + "summary": "The passage defines fair value and describes the three-level hierarchy for valuation inputs: Level 1 uses unadjusted quoted prices in active markets for identical assets; Level 2 uses observable inputs like quoted prices for similar assets or inactive markets; Level 3 uses unobservable inputs requiring significant judgment. It details valuation methods for various instruments: listed equities use Level 1 inputs; unlisted equities may use market/income approaches with estimates; interest rate swaps and options use observable market data; convertible bonds use binomial models with unobservable inputs; exchangeable bonds rely on over-the-counter market prices; contingent consideration uses expected cash flow methods; and privately held company investments are measured on a non-recurring basis under Level 3.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_9", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_9", + "level": 1, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_9/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "9. Fair value measurement (Continued)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_9/9. Fair value measurement (Continued)", + "level": 2, + "summary": "The following table summarizes the Company's assets and liabilities that are measured at fair value on a recurring basis and are categorized under the fair value hierarchy: [tables/table-33 Fair Value Table.html] [tables/table-34 Fair Value Table.html]", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_10", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_10", + "level": 1, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_10/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "9. Fair value measurement (Continued)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_10/9. Fair value measurement (Continued)", + "level": 2, + "summary": "(i) Included in short-term investments and equity securities and other investments on the unaudited condensed consolidated balance sheets. (ii) Included in equity securities and other investments on the unaudited condensed consolidated balance sheets. (iii) Included in prepayments, receivables and other assets on the unaudited condensed consolidated balance sheets. (iv) Included in accrued expenses, accounts payable and other liabilities on the unaudited condensed consolidated balance sheets. (v) As of March 31 and September 30, 2025, listed equity securities with fair value of RMB11,921 million and RMB10,415 million were subject to contractual sale restrictions, respectively. (vi) Others primarily represent other investments with underlying assets measured at fair value. Convertible and exchangeable bonds investments categorized within Level 3 under the fair value hierarchy: [tables/table-35 Balance Change.html] Deferred consideration categorized within Level 3 under the fair value hierarchy: [tables/table-36 Fair Value Change.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "10. Intangible assets, net", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_10/10. Intangible assets, net", + "level": 2, + "summary": "[tables/table-37 Intangible Assets.html] Total amortization expenses recognized for the six months ended September 30, 2024 and 2025 amounted to RMB7,122 million and RMB5,111 million, respectively, including the portion relating to licensed copyrights of RMB3,689 million and RMB3,478 million which were recorded in cost of revenue for the six months ended September 30, 2024 and 2025, respectively.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_11", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11", + "level": 1, + "summary": "", + "chunk_count": 7, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "11. Goodwill", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/11. Goodwill", + "level": 2, + "summary": "Changes in the carrying amount of goodwill by segment for the six months ended September 30, 2025 were as follows: [tables/table-38 Alibaba Segment Assets.html] Gross goodwill balances were RMB302,194 million and RMB302,244 million as of March 31 and September 30, 2025, respectively. Accumulated impairment losses were RMB46,693 million and RMB46,693 million as of March 31 and September 30, 2025, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "12. Supplier Finance Programs", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/12. Supplier Finance Programs", + "level": 2, + "summary": "The Company uses supplier finance programs with financial institutions, allowing suppliers to sell payment obligations at a discount for early cash. Payment terms are up to 180 days, and the Company's rights and obligations remain unchanged. As of March 31 and September 30, 2025, no assets were pledged except for RMB1,500 million in treasury investments. Outstanding obligations are recorded in accrued expenses, accounts payable, and other liabilities, except for certain discount payments recorded as current bank borrowings.", + "chunk_count": 1, + "children": [] + }, + { + "title": "13. Convertible unsecured senior notes", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/13. Convertible unsecured senior notes", + "level": 2, + "summary": "The passage details two convertible senior note issuances by Alibaba Group: $5.0 billion in 2024 notes due 2031 with 0.5% interest and initial conversion rate of 9.5202 ADSs per $1,000, and $3.2 billion in 2025 zero-coupon notes due 2032 with initial conversion rate of 5.1773 ADSs per $1,000. Both are senior unsecured obligations convertible into ADSs, with adjustment provisions for dividends and fundamental changes. As of September 30, 2025, the 2024 notes had an adjusted conversion rate of 9.8915 ADSs, while the 2025 notes remained unadjusted. The company may redeem notes under certain conditions, and holders can require repurchase on specific dates or upon fundamental changes. Unamortized costs and fair values are provided for both notes as of March and September 2025, with effective interest rates of 0.8% for 2024 notes and 0.3% for 2025 notes.", + "chunk_count": 1, + "children": [] + }, + { + "title": "13. Convertible unsecured senior notes (Continued)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/13. Convertible unsecured senior notes (Continued)", + "level": 2, + "summary": "In connection with the issuance of the convertible senior notes, the Company entered into capped call transactions with certain financial institutions at a cost of US\\$638 million (RMB4,612 million) and US\\$184 million (RMB1,309 million) for the 2024 Convertible Senior Notes and the 2025 Convertible Senior Notes, respectively, which are expected to reduce potential dilution and/or offset cash payments upon conversion. The cap prices of the capped call transactions for the 2024 Convertible Senior Notes and the 2025 Convertible Senior Notes are initially US\\$161.60 per ADS and US\\$235.46 per ADS, respectively. The capped prices are subject to adjustments similar to the adjustments on the conversion rates of the respective convertible senior notes. The capped call transactions may be settled in cash at the Company’s election.", + "chunk_count": 1, + "children": [] + }, + { + "title": "14. Exchangeable bonds", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_11/14. Exchangeable bonds", + "level": 2, + "summary": "In July 2025, the Company issued HKD-denominated zero coupon exchangeable bonds due 2032, referencing Alibaba Health shares, with an aggregate principal of HK$12 billion. The bonds are exchangeable into AH Shares at an initial rate of 160,513.6 shares per HK$1,000,000, subject to adjustments. The Company may redeem or repurchase the bonds under certain conditions. A stock borrowing and lending arrangement was entered into for hedging, with fair value of lent shares at HK$8,492 million as of September 30, 2025. Losses of RMB327 million from fair value changes were recorded for the six months ended September 30, 2025.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_12", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12", + "level": 1, + "summary": "", + "chunk_count": 7, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "15. Bank borrowings", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/15. Bank borrowings", + "level": 2, + "summary": "The Company amended its syndicated loan facility, reducing it from $4.0 billion to $3.17 billion after a partial repayment. In September 2025, pricing was reduced to SOFR plus 66 basis points. The facility was restructured as a revolving credit facility on November 28, 2025, allowing drawdowns in USD and HKD, with the expiration extended to September 30, 2028, and an option to extend to September 30, 2030. Interest is SOFR or HIBOR plus 66 basis points, rising to 81 basis points during the optional extension. The Company then prepaid the $3.17 billion outstanding balance. Separately, a $6.5 billion revolving credit facility was amended to $3.33 billion, allowing USD and HKD utilization, with interest at SOFR or HIBOR plus 66 basis points, and expiration extended to September 30, 2028, with an optional extension to September 30, 2030 at 81 basis points. This facility has not been drawn.", + "chunk_count": 1, + "children": [] + }, + { + "title": "16. Related party transactions", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/16. Related party transactions", + "level": 2, + "summary": "During the six months ended September 30, 2024 and 2025, other than disclosed elsewhere, the Company had the following material related party transactions: Transactions with Ant Group and its affiliates [tables/table-40 Cloud Revenue Growth.html] (i) The Company has other commercial arrangements and cost sharing arrangements with Ant Group and its affiliates on various sales and marketing, cloud, and other administrative and support services. (ii) The Company has a commercial agreement with Alipay.com Co., Ltd., a wholly-owned subsidiary of Ant Group (“Alipay”), whereby the Company receives payment processing and escrow services in exchange for a payment for the services fee, which was recognized in cost of revenue. As of March 31 and September 30, 2025, the Company had certain amounts of cash held in accounts managed by Alipay in connection with the provision of online and mobile commerce and related services for a total amount of RMB5,863 million and RMB6,379 million, respectively, which have been classified as cash and cash equivalents on the unaudited condensed consolidated balance sheets. ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "16. Related party transactions (Continued)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/16. Related party transactions (Continued)", + "level": 2, + "summary": "The Company has commercial arrangements with investees for cloud, marketing, and logistics services, recording revenues and costs for the six months ended September 30, 2024 and 2025. Loans to investees totaled RMB1,771 million and RMB1,992 million as of March 31 and September 30, 2025, with terms up to 6 years and interest rates up to 10% per annum. A guarantee was provided for a HK$6.5 billion revolving loan facility to Cingleot, with HK$5,090 million and HK$5,226 million drawn as of those dates. Other transactions with investees and related parties are less than 1% of revenue and costs.", + "chunk_count": 1, + "children": [] + }, + { + "title": "17. Risks and contingencies", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/17. Risks and contingencies", + "level": 2, + "summary": "The passage outlines key risks for a Cayman Islands company operating in China through Variable Interest Entities (VIEs) due to foreign ownership restrictions in value-added telecom services. It highlights uncertainties in PRC laws that could force restructuring or deconsolidation of VIEs. Macro-economic and regulatory risks in the PRC market, especially for internet and mobile commerce, are noted, with potential new restrictions possibly requiring business sales or closures. The company's association with Ant Group poses reputational and regulatory risks. Additionally, currency risks are discussed, as RMB is not freely convertible, and foreign exchange controls could hinder dividend payments and funding of foreign currency activities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "17. Risks and contingencies (Continued)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_12/17. Risks and contingencies (Continued)", + "level": 2, + "summary": "The passage outlines various risks faced by the company, including strategic investment volatility, credit risk from financial institutions, trade assurance program liabilities, legal proceedings such as the EU investigation into AliExpress under the Digital Services Act, impacts from the Russia-Ukraine conflict on supply chains and logistics, and risks from export controls and sanctions affecting technology upgrades and competitive edge.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_13", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_13", + "level": 1, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_13/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "18. Segment information", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_13/18. Segment information", + "level": 2, + "summary": "The company updated its segment reporting from six segments to a new organizational structure starting June 30, 2025. Segments include Taobao and Tmall Group, Alibaba International Digital Commerce Group, Cloud Intelligence Group, Cainiao Smart Logistics, Local Services Group, and Hujing Digital Media. The CODM uses revenue and Adjusted EBITA to evaluate performance and allocate resources. Segment information is presented before inter-segment elimination, with costs allocated based on usage, revenue, or headcount. Assets are not allocated to segments. The table provides segment data for the six months ended September 30, 2024 and 2025.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_14", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_14", + "level": 1, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_14/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "18. Segment information (Continued)", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_14/18. Segment information (Continued)", + "level": 2, + "summary": "The following table presents the reconciliation from the total segments Adjusted EBITA to the consolidated net income for the six months ended September 30, 2024 and 2025: [tables/table-42 Financial Results.html] The following table presents the depreciation and impairment of property and equipment, and operating lease cost relating to land use rights by segment for the six months ended September 30, 2024 and 2025: [tables/table-43 Depreciation by Segment.html] (i) Segment costs and expenses primarily comprise components that are included in cost of revenue, product development expenses, sales and marketing expenses, as well as general and administrative expenses. (ii) Adjusted EBITA represents net income before interest and investment income, net, interest expense, other (expense) income, net, income tax expenses, share of results of equity method investees, certain non-cash expenses, consisting of share-based compensation expense, amortization and impairment of intangible assets, impairment of goodwill, and others, which the Company does not believe are reflective of the Company's core operating performance during the periods presented. (iii) Unallocated primarily relates to certain costs incurred by corporate functions and other miscellaneous items that are not allocated to individual segments. Details of the Company's revenue by segment are set out in Note 4.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_15", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15", + "level": 1, + "summary": "", + "chunk_count": 7, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "chunk_count": 1, + "children": [] + }, + { + "title": "19. Aging analysis", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/19. Aging analysis", + "level": 2, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Accounts receivable", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/19. Aging analysis/Accounts receivable", + "level": 3, + "summary": "The aging analysis of the accounts receivable, net of allowance, based on billing date is as follows: [tables/table-44 Accounts Receivable.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Accounts payable", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/19. Aging analysis/Accounts payable", + "level": 3, + "summary": "The aging analysis of the accounts payable based on billing date is as follows: [tables/table-45 Accounts Payable.html]", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "20. Dividends", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/20. Dividends", + "level": 2, + "summary": "The Board did not recommend the distribution of interim dividend for the six months ended September 30, 2024 and 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "21. Subsequent events", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_15/21. Subsequent events", + "level": 2, + "summary": "Save as disclosed in the unaudited condensed consolidated financial statements, there were no significant events that might affect the Company since September 30, 2025.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_16", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_16", + "level": 1, + "summary": "", + "chunk_count": 2, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_16/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 22. Reconciliations between U.S. GAAP and International Financial Reporting Standards The unaudited condensed consolidated financial statements are prepared in accordance with U.S. GAAP, which differ in certain respects from International Financial Reporting Standards (“IFRS”). The effects of material differences between the unaudited condensed consolidated financial statements of the Company prepared under U.S. GAAP and IFRS are as follows: Reconciliation of Unaudited Condensed Consolidated Balance Sheets (Extract) [tables/table-46 IFRS Adjustments.html]", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_17", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_17", + "level": 1, + "summary": "", + "chunk_count": 2, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_17/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued) Reconciliation of Unaudited Condensed Consolidated Balance Sheets (Extract) As of September 30, 2025 [tables/table-47 GAAP to IFRS.html]", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ALIBABA GROUP HOLDING LIMITED_18", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18", + "level": 1, + "summary": "", + "chunk_count": 11, + "children": [ + { + "title": "NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS", + "level": 2, + "summary": "", + "chunk_count": 10, + "children": [ + { + "title": "FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025", + "level": 3, + "summary": "22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued) Reconciliation of Unaudited Condensed Consolidated Income Statement (Extract) Six months ended September 30, 2024 [tables/table-48 GAAP to IFRS.html] 22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued) Reconciliation of Unaudited Condensed Consolidated Income Statement (Extract) (Continued) Six months ended September 30, 2025 [tables/table-49 GAAP to IFRS Reconciliation.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "(i) Consolidation and business combinations", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(i) Consolidation and business combinations", + "level": 3, + "summary": "The passage discusses key differences between U.S. GAAP and IFRS in three areas: consolidation criteria (control via voting shares vs. de facto control), measurement of noncontrolling interests (fair value with full goodwill vs. proportionate share with partial goodwill), and goodwill impairment testing (comparing carrying value to fair value vs. recoverable amount). It is from Alibaba Group's financial statement notes for the six months ended September 30, 2024 and 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "(ii) Equity securities without readily determinable fair value", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(ii) Equity securities without readily determinable fair value", + "level": 3, + "summary": "Under U.S. GAAP, the Company can elect, on an instrument-by-instrument basis, to apply the measurement alternative to record the investments in equity securities without readily determinable fair values at cost, less impairment, with subsequent adjustments for observable price changes recognized in the unaudited condensed consolidated income statements. Under IFRS, these investments are measured at fair value with changes in fair value recognized in the unaudited condensed consolidated income statements.", + "chunk_count": 1, + "children": [] + }, + { + "title": "(iii) Equity method investments", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(iii) Equity method investments", + "level": 3, + "summary": "The Company generally applies the equity method to account for equity investments over which it has significant influence. Under U.S. GAAP, significant influence is presumed to exist for an investment in limited partnership or unincorporated entity, unless the investment is so minor that the Company has virtually no influence over the entity's operating and financial policies. Under IFRS, significant influence is presumed to exist for an investment of over $20\\%$ of the voting rights of an entity. The Company records its share of the post-acquisition results of its equity method investees and adjusts for the basis differences that exist between the carrying values of the equity method investments and the Company's proportionate share of the carrying value of the investee's net assets. Adjustments are made to the financial statements of the equity method investees prepared under U.S. GAAP in order to conform to the Company's accounting policies under IFRS and to reflect the basis differences of the equity method investments under IFRS, if different from those under U.S. GAAP.", + "chunk_count": 1, + "children": [] + }, + { + "title": "(iv) Share-based awards", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(iv) Share-based awards", + "level": 3, + "summary": "The employees of the Company hold share-based awards relating to an equity method investee of the Company that were granted and will be settled by related parties or economic interest holders of the Company. Under U.S. GAAP, the cost related to these awards is recognized over the requisite service period, with subsequent changes in fair value of these awards recognized in the unaudited condensed consolidated income statements. Under IFRS, these awards are not considered as share based payments of the Company and the cost relating to these awards is not recognized. The Company accounts for income tax effects of share-based awards that ordinarily give rise to tax deduction. Under U.S. GAAP, deferred taxes for these awards are measured based on share-based compensation expenses recognized in the unaudited condensed consolidated financial statements. Under IFRS, deferred taxes for these awards are measured based on future tax deduction estimated at the end of each reporting period.", + "chunk_count": 1, + "children": [] + }, + { + "title": "(v) Operating leases", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(v) Operating leases", + "level": 3, + "summary": "Under U.S. GAAP, the amortization of right-of-use assets and the interest expense related to lease liabilities are recorded together as lease expense and recognized in the unaudited condensed consolidated income statements on a straight-line basis. Under IFRS, the right-of-use assets are amortized on a straight-line basis while the interest expense related to lease liabilities are recognized in the unaudited condensed consolidated income statements using effective interest method.", + "chunk_count": 1, + "children": [] + }, + { + "title": "(vi) Redeemable noncontrolling interests", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(vi) Redeemable noncontrolling interests", + "level": 3, + "summary": "Equity interests issued by certain subsidiaries of the Company are redeemable. Under U.S. GAAP, redeemable equity interests are classified as mezzanine equity if the redemption is outside the Company's control and as noncontrolling interests if equity interests issued by finite-lived subsidiaries are mandatorily redeemable only upon liquidation. Under IFRS, these redeemable equity interests are generally classified as financial liabilities. ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued)", + "chunk_count": 1, + "children": [] + }, + { + "title": "(vii) Hyperinflation", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/(vii) Hyperinflation", + "level": 3, + "summary": "The passage compares U.S. GAAP and IFRS treatments for hyperinflationary subsidiaries, convertible unsecured senior notes, and instruments in an entity's own equity. Under U.S. GAAP, subsidiaries in highly inflationary economies are remeasured using the parent's functional currency, while IFRS requires restatement for current purchasing power. For convertible notes, U.S. GAAP treats them as liabilities at amortized cost, whereas IFRS accounts for them as hybrid instruments at fair value with credit risk changes in OCI. For equity instruments, U.S. GAAP requires both a fixed-for-fixed test and equity classification, while IFRS allows equity treatment for derivatives meeting the test and non-derivatives without cash delivery obligations, with differences in settlement alternatives leading to potential classification disparities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Alibaba", + "path": "Fiscal%20Year%202026%20Interim%20Report.pdf/ALIBABA GROUP HOLDING LIMITED_18/NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS/Alibaba", + "level": 3, + "summary": "Alibaba Group Holding Limited", + "chunk_count": 1, + "children": [] + } + ] + } + ] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-Alibaba.jpg", + "summary": "image-1 Alibaba0" + }, + { + "path": "images/image-2-### (vii) .jpg", + "summary": "image-2 ### (vii) Hyperinflation ### Alibaba0" + } + ], + "tables": [ + { + "path": "tables/table-0 Financial Results.html", + "summary": "table-1 The table shows financial data for six months ending Sep 30, 2024 and 2025. Revenue grew 3% to RMB 495,447 million, while income from operations dropped 43%. Adjusted EBITDA fell 36% to RMB 62,991 million." + }, + { + "path": "tables/table-1 Alibaba Revenue.html", + "summary": "table-2 Alibaba's consolidated revenue for six months ended Sep 2025 was RMB 495,447 million (US$69,595 million), up 3% YoY. China E-commerce grew 12%, International Digital Commerce 14%, and Cloud Intelligence 30%." + }, + { + "path": "tables/table-2 Alibaba EBITA.html", + "summary": "table-3 Alibaba's consolidated adjusted EBITA for six months ended Sep 2025 was RMB 47,917M, down 44% YoY. Cloud Intelligence Group grew 31%, while China E-commerce dropped 47%." + }, + { + "path": "tables/table-3 Costs & Expenses.html", + "summary": "table-4 The table shows costs and expenses for six months ended Sep 30, 2024 and 2025. Total costs rose from 409,355 to 455,391 RMB. Sales and marketing expenses increased significantly by 10.6% of revenue." + }, + { + "path": "tables/table-4 Share-based Compensation.html", + "summary": "table-5 Alibaba Group's share-based compensation expense decreased 17% YoY to RMB 6,870 million for six months ended Sep 2025, with Alibaba awards down 21% and others down 8%." + }, + { + "path": "tables/table-5 Equity Investee Results.html", + "summary": "table-6 The table shows share of profit from equity investees for six months ending Sep 30, 2024 and 2025. Ant Group contributed RMB 6,395M (2024) and RMB 4,280M (2025). Total profit was RMB 2,483M (2024) and RMB 3,254M (2025)." + }, + { + "path": "tables/table-6 Financial Results.html", + "summary": "table-7 The table shows financial data for six months ending Sep 30, 2024 and 2025, with net income, adjusted EBITA, and EBITDA in RMB and US dollars." + }, + { + "path": "tables/table-7 Non-GAAP Net Income.html", + "summary": "table-8 The table shows net income and non-GAAP net income for six months ending September 30, 2024 and 2025, with adjustments including share-based compensation, amortization, and investment gains." + }, + { + "path": "tables/table-8 Earnings per Share.html", + "summary": "table-9 The table shows basic and diluted net income attributable to ordinary shareholders for six months ending Sep 30, 2024 and 2025, with non-GAAP adjustments and diluted earnings per share/ADS in RMB and USD." + }, + { + "path": "tables/table-9 Free Cash Flow.html", + "summary": "table-10 The table shows net cash from operations, property purchases, and buyer protection deposits for six months ending Sep 2024 and 2025, resulting in free cash flow of 31,107 RMB (2024) and -40,655 RMB (2025)." + }, + { + "path": "tables/table-10 Board of Directors.html", + "summary": "table-11 The table lists 10 board members of a company, including Chairman Joseph C. Tsai (61), CEO Eddie Yongming Wu (50), and President J. Michael Evans (68), along with independent directors." + }, + { + "path": "tables/table-11 Shareholding Table.html", + "summary": "table-12 The table lists directors and CEOs with their shareholding interests, including beneficial ownership, spouse interests, and controlled corporations, showing share and ADS counts with percentages." + }, + { + "path": "tables/table-12 Shareholder Holdings.html", + "summary": "table-13 The table lists major shareholders (JPMorgan, Citigroup, BlackRock) with their share counts, ADS equivalents, and approximate percentages of total shareholding, categorized by capacity (beneficial owner, trustee, etc.)." + }, + { + "path": "tables/table-13 RSU Grant Summary.html", + "summary": "table-14 Table shows RSU activity for directors, CEO, and employees from April to September 2025, with 44.9M outstanding, 16.3M vested, and 26.8M remaining." + }, + { + "path": "tables/table-14 Share Options Table.html", + "summary": "table-15 Table shows share option grants for directors and employees, including exercise prices, outstanding numbers, and changes during the reporting period ending September 30, 2025." + }, + { + "path": "tables/table-15 RSU Activity Report.html", + "summary": "table-16 The table shows RSU activity for employee participants from April to September 2025, with 63.6 million granted, 4.7 million vested, 1.4 million forfeited, and 67 million outstanding as of September 30, 2025." + }, + { + "path": "tables/table-16 RSU Grant Details.html", + "summary": "table-17 The table shows RSU grants on three dates in 2025: May 23 (53,429,597 RSUs, HK$118.78 fair value), June 13 (7,687,091 RSUs, HK$114.60), and September 4 (2,477,630 RSUs, HK$130.31), totaling 63,594,318 RSUs." + }, + { + "path": "tables/table-17 Share Repurchase.html", + "summary": "table-18 The table shows monthly share repurchases from April to August 2025, totaling 72,733,488 shares for $1,046 million, with prices ranging from $11.97 to $16.79." + }, + { + "path": "tables/table-18 Notes Proceeds Use.html", + "summary": "table-19 The table shows $3,130M net proceeds from September 2025 Notes Offering, with $730M utilized by September 30, 2025: $443M for cloud infrastructure and $287M for international commerce." + }, + { + "path": "tables/table-19 Interim Report.html", + "summary": "table-20 Table of contents for an interim financial report for the six months ended September 30, 2025, listing unaudited financial statements and notes starting on page 45." + }, + { + "path": "tables/table-20 Alibaba Financials.html", + "summary": "table-21 Alibaba Group's financial results for six months ended Sep 30, 2024 and 2025, showing revenue of 495,447 million RMB and net income attributable to ordinary shareholders of 64,106 million RMB in 2025." + }, + { + "path": "tables/table-21 Income Statement.html", + "summary": "table-22 The table shows net income of 67,569 RMB (2024) and 62,994 RMB (2025), with total comprehensive income of 64,280 RMB and 57,874 RMB respectively, including foreign currency translation losses and other items." + }, + { + "path": "tables/table-22 Balance Sheet Data.html", + "summary": "table-23 The table shows assets, liabilities, and equity as of March 31, 2025 and September 30, 2025, with figures in RMB and US dollars. Total assets were 1,804,227 million RMB in March and 1,883,880 million in September." + }, + { + "path": "tables/table-23 Equity Summary.html", + "summary": "table-24 The table shows commitments, mezzanine equity, and shareholders' equity components as of March 31 and September 30, 2025, with RMB and US$ values. Total equity increased from 1,078,393 to 1,101,901 million RMB." + }, + { + "path": "tables/table-24 Equity Changes.html", + "summary": "table-25 The table shows changes in shareholders' equity from April to September 2024 and 2025, including share issuances, repurchases, dividends, and net income, with totals in millions of RMB." + }, + { + "path": "tables/table-25 Cash Flow Statement.html", + "summary": "table-26 The table shows cash flows for six months ending Sep 30, 2024 and 2025. Operating cash flow decreased from 65,074 to 30,771. Investing activities used 34,865 in 2024 and 51,324 in 2025. Financing activities shifted from -86,364 to 8,171." + }, + { + "path": "tables/table-26 Financial Data.html", + "summary": "table-27 The table shows financial data for a company as of March 31 and September 30, 2025, and for six months ended September 30, 2024 and 2025. Total assets were 180,278 million RMB and 208,234 million RMB, while total liabilities were 182,979 million and 211,186 million. Revenue increased from 67,739 to 77,466 million, net income from 2,103 to 2,442 million." + }, + { + "path": "tables/table-27 Alibaba Revenue.html", + "summary": "table-28 Alibaba Group's consolidated revenue for six months ended Sep 30, 2025 was RMB 495,447 million, up from RMB 479,739 million in 2024, driven by growth in China E-commerce, International Digital Commerce, and Cloud Intelligence." + }, + { + "path": "tables/table-28 Revenue Breakdown.html", + "summary": "table-29 The table shows revenue for six months ending Sep 30, 2024 and 2025 in RMB millions. Key services include customer management, membership, logistics, cloud, goods sales, and other revenue, totaling 479,739 in 2024 and 495,447 in 2025." + }, + { + "path": "tables/table-29 Tax Expense Data.html", + "summary": "table-30 The table shows current income tax expense and deferred taxation for six-month periods ending September 30, 2024 and 2025, with totals of 17,442 and 14,415 million RMB respectively." + }, + { + "path": "tables/table-30 Cost Breakdown.html", + "summary": "table-31 The table shows cost of revenue, product development, sales and marketing, and general administrative expenses for six months ending Sep 30, 2024 and 2025, with totals of 8,277 and 6,870 million RMB respectively." + }, + { + "path": "tables/table-31 EPS Data Table.html", + "summary": "table-32 The table shows basic and diluted earnings per ordinary share and ADS for six months ending September 30, 2024 and 2025, with net income and share count adjustments." + }, + { + "path": "tables/table-32 Investment Summary.html", + "summary": "table-33 The table compares investment portfolios as of March 31, 2025 and September 30, 2025, showing original cost, cumulative net gains/losses, and carrying value for equity and debt investments in millions of RMB." + }, + { + "path": "tables/table-33 Fair Value Table.html", + "summary": "table-34 The table shows assets and liabilities measured at fair value as of March 31, 2025, with totals of 590,532 million RMB for assets and 1,797 million RMB for liabilities, categorized by Level 1, 2, and 3 inputs." + }, + { + "path": "tables/table-34 Fair Value Table.html", + "summary": "table-35 Table shows assets and liabilities measured at fair value as of September 30, 2025, with totals of 576,406 million RMB for assets and 16,181 million RMB for liabilities, categorized by Level 1, 2, and 3 inputs." + }, + { + "path": "tables/table-35 Balance Change.html", + "summary": "table-36 Balance decreased from 818 to 620 million RMB between April and September 2025, with additions of 42, net fair value decrease of 158, disposals of 39, conversions of 34, and currency adjustments of 9." + }, + { + "path": "tables/table-36 Fair Value Change.html", + "summary": "table-37 Balance increased from 3,039 to 3,052 million RMB between April and September 2025, with a net fair value increase of 40 and a currency translation loss of 27." + }, + { + "path": "tables/table-37 Intangible Assets.html", + "summary": "table-38 The table shows intangible assets as of March 31 and September 30, 2025, including user base, trademarks, non-compete agreements, developed technology, and licensed copyrights, with total cost, accumulated amortization, and net book value." + }, + { + "path": "tables/table-38 Alibaba Segment Assets.html", + "summary": "table-39 Alibaba's segment asset balances from April to September 2025 show a total of 255,551 million RMB, with major changes including the transfer of Cainiao, Local Services, and Hujing to 'All others'." + }, + { + "path": "tables/table-39 Financial Liabilities.html", + "summary": "table-40 The table shows accrued expenses, accounts payable, and bank borrowings as of March 31 and September 30, 2025, in millions of RMB, with totals of 6,075 and 7,199 respectively." + }, + { + "path": "tables/table-40 Cloud Revenue Growth.html", + "summary": "table-41 The table shows six-month financial data ending September 30, 2024 and 2025. Cloud services revenue rose from 4,977 to 8,928 million RMB, while marketplace fees decreased. Total amounts earned increased from 8,491 to 11,747 million RMB." + }, + { + "path": "tables/table-41 Alibaba Segment Results.html", + "summary": "table-42 Financial data for Alibaba segments (China E-commerce, International Digital Commerce, Cloud Intelligence, Others) for six months ended Sep 2024 and 2025, showing revenue, costs, and adjusted EBITA in RMB millions." + }, + { + "path": "tables/table-42 Financial Results.html", + "summary": "table-43 The table shows consolidated income from operations decreased from 71,235 to 40,353 million RMB, while net income dropped from 67,569 to 62,994 million RMB for the six months ended September 30, 2024 and 2025." + }, + { + "path": "tables/table-43 Depreciation by Segment.html", + "summary": "table-44 The table shows depreciation and impairment of property and equipment for Alibaba segments for six months ended September 30, 2024 and 2025, with totals of 11,619 and 13,535 million RMB respectively." + }, + { + "path": "tables/table-44 Accounts Receivable.html", + "summary": "table-45 The table shows accounts receivable aging as of March 31 and September 30, 2025, with totals of 30,652 and 31,646 million RMB respectively, broken down by 0-3, 3-6, 6-12 months, and over 1 year." + }, + { + "path": "tables/table-45 Accounts Payable.html", + "summary": "table-46 The table shows accounts payable by maturity as of March 31 and September 30, 2025, in millions of RMB. Total payables are 58,201 and 64,939, with most due within 3 months." + }, + { + "path": "tables/table-46 IFRS Adjustments.html", + "summary": "table-47 Reconciliation of U.S. GAAP to IFRS for assets, liabilities, and equity as of March 31, 2025, showing adjustments for consolidation, investments, leases, and convertible notes." + }, + { + "path": "tables/table-47 GAAP to IFRS.html", + "summary": "table-48 Reconciliation of U.S. GAAP to IFRS for assets, liabilities, and equity, showing adjustments for consolidation, equity securities, leases, and convertible notes." + }, + { + "path": "tables/table-48 GAAP to IFRS.html", + "summary": "table-49 Reconciliation of financial data from U.S. GAAP to IFRS for a company, showing adjustments for items like hyperinflation, convertible notes, and equity method investments, resulting in net income of 62,924 million RMB under IFRS." + }, + { + "path": "tables/table-49 GAAP to IFRS Reconciliation.html", + "summary": "table-50 This table reconciles financial statement line items from U.S. GAAP to IFRS, showing adjustments for consolidation, equity securities, leases, and convertible notes, resulting in net income of RMB 43,856 million under IFRS." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/full.md b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/full.md new file mode 100644 index 000000000..089bf2222 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/full.md @@ -0,0 +1,1290 @@ +![](images/cf85a522c9ebb0f5c0b0503aff6d37164121606686506820e69c44ac67aeef6b.jpg) + +# Alibaba + +Alibaba Group Holding Limited + +NYSE: BABA HKEX: 9988 (HKD Counter) 89988 (RMB Counter) + +## CONTENTS + +4 Management Discussion and Analysis +25 Directors and Chief Executive Officer +29 Disclosure of Interests +34 Equity Incentive Plans +38 Purchase, Sale or Redemption of our Company's Listed Securities +39 Other Information +42 Definitions +44 Financial Statements + +## Weighted Voting Rights + +We have one class of Shares, and each holder of our Shares is entitled to one vote per Share. Pursuant to our Articles of Association, for so long as the Partnership Condition (as defined in our Articles of Association) is satisfied, the Alibaba Partnership has the exclusive right to nominate or, in limited situations, appoint, up to a simple majority of the members of our board of directors. These rights are categorized as a weighted voting rights structure, or WVR structure, under the Hong Kong Listing Rules. As a result, we are deemed as a company with a WVR structure. + +Alibaba Partnership will remain in place for the life of the partnership unless our Articles of Association are amended to provide otherwise by a vote of shareholders representing at least 95% of shares that vote at a shareholders meeting. The nomination rights of the Alibaba Partnership will remain in place notwithstanding a change of control or merger of our company. For further details of the Alibaba Partnership, please see the section headed “Directors, Senior Management and Employees – Directors and Senior Management – Alibaba Partnership” in our fiscal year 2025 annual report. + +Shareholders and prospective investors are advised to be aware of the potential risks of investing in companies with a WVR structure, in particular, the Alibaba Partnership limits the ability of our shareholders to nominate and elect directors, and the interests of the Alibaba Partnership may conflict with the interests of our shareholders. For further information about the risks associated with our WVR structure, please see the section headed “Risk Factors — Risks Related to Our Corporate Structure” in our fiscal year 2025 annual report. Prospective investors should make the decision to invest in us only after due and careful consideration. + +## Exchange Rate Information + +This Interim Report contains translations of certain Renminbi (“RMB”) amounts into U.S. dollars (“US\$”) and Hong Kong dollars (“HK\$”) for the convenience of the reader. Unless otherwise stated, all translations of RMB into US\$ were made at RMB7.1190 to US\$1.00, the exchange rate on September 30, 2025 as set forth in the H.10 statistical release of the Federal Reserve Board, and all translations of RMB into HK\$ were made at RMB0.91298 to HK\$1.00, the middle rate on September 30, 2025 as published by the People’s Bank of China. The percentages stated in this Interim Report are calculated based on the RMB amounts and there may be minor differences due to rounding. + +## Safe Harbor Statements + +This Interim Report contains forward-looking statements. These statements are made under the “safe harbor” provisions of the U.S. Private Securities Litigation Reform Act of 1995. These forward-looking statements can be identified by terminology such as “may,” “will,” “expect,” “anticipate,” “future,” “aim,” “estimate,” “intend,” “seek,” “plan,” “believe,” “potential,” “continue,” “ongoing,” “target,” “guidance,” “is/are likely to” and similar statements. In addition, statements that are not historical facts, including statements about Alibaba’s strategies and business and operational plans, Alibaba’s beliefs, expectations and guidance regarding the growth of its business, its operating and financial results, return on investments, strategic investments and dispositions and share repurchases, and the business outlook in this Interim Report, are or contain forward-looking statements. Forward-looking statements involve inherent risks and uncertainties. A number of factors could cause actual results to differ materially from those contained in any forward-looking statement, including but not limited to: Alibaba’s ability to compete, innovate and maintain or grow its business; risks associated with sustained investments in Alibaba’s businesses; risks related to strategic transactions; fluctuations in general economic and business conditions in China and globally; uncertainties arising from competition among countries and geopolitical tensions, including national trade, investment, protectionist or other policies and + +export control, economic or trade sanctions; changes to our shareholder return initiatives; and assumptions underlying or related to any of the foregoing. Further information regarding these and other risks is included in Alibaba’s filings with the U.S. Securities and Exchange Commission and announcements on the website of The Stock Exchange of Hong Kong Limited. All information provided in this Interim Report is as of the date of this Interim Report and are based on assumptions that we believe to be reasonable as of this date, and Alibaba does not undertake any obligation to update any forward-looking statement, except as required under applicable law. + +Six Months Ended September Summary Financial Results + +
Six months ended September 30,
20242025YoY % Change
RMBRMBUS$
(in millions, except percentages and per share amounts)
Revenue479,739495,44769,5953%
Income from operations71,23540,3535,668(43)% (2)
Operating margin15%8%
Adjusted EBITDA (1)98,48862,9918,848(36)% (3)
Adjusted EBITDA margin (1)21%13%
Adjusted EBITA (1)85,59647,9176,731(44)% (3)
Adjusted EBITA margin (1)18%10%
Net income67,56962,9948,849(7)% (4)
Net income attributable to ordinary shareholders68,14364,1069,005(6)% (4)
Non-GAAP net income (1)77,20943,8626,161(43)% (3)
Diluted earnings per share (5)3.503.340.47(5)% (4)(6)
Diluted earnings per ADS (5)28.0026.733.75(5)% (4)(6)
Non-GAAP diluted earnings per share (1)(5)3.942.390.34(39)% (3)(6)
Non-GAAP diluted earnings per ADS (1)(5)31.5019.102.68(39)% (3)(6)
+ +(1) See the sections entitled “Non-GAAP Financial Measures” and “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” for more information about the non-GAAP measures referred to within this Interim Report. +(2) The year-over-year decrease was primarily due to the decrease in adjusted EBITA, partly offset by a one-time provision in the same period last year (See the section entitled “Six Months Ended September Other Financial Results” for more information). +(3) The year-over-year decreases were mainly attributable to the investment in quick commerce, user experiences, and technology, partly offset by double-digit revenue growth in Alibaba China E-commerce Group, the improved operating results supported by continued growth in Cloud business, as well as enhanced operating efficiencies across various businesses. +(4) The year-over-year decreases were mainly attributable to the decrease in income from operations, partly offset by the mark-to-market changes from our equity investments, gain from the disposal of local consumer service business of Trendyol, and the decrease in both the impairment of our investments and net exchange loss, while net income attributable to ordinary shareholders and earnings per share/ADS would further take into account the net loss (income) attributable to noncontrolling interests and (accretion) reversal of accretion of mezzanine equity. We excluded non-cash share-based compensation expense, gains/losses of investments, impairment of goodwill and intangible assets, and certain other items from our non-GAAP measurements. +(5) Each ADS represents eight ordinary shares. +(6) The year-over-year percentages as stated are calculated based on the exact amount and there may be minor differences from the year-over-year percentages calculated based on the RMB amounts after rounding. + +## Six Months Ended September Segment Results + +Revenue for the six months ended September 30, 2025 was RMB495,447 million (US\$69,595 million), an increase of 3% year-over-year compared to RMB479,739 million in the same period of 2024. Excluding revenue from the disposed businesses of Sun Art and Intime, revenue on a like-for-like basis would have grown by 12% year-over-year. + +The following table sets forth a breakdown of our revenue by segment for the periods indicated: + +
Six months ended September 30,
20242025YoY % Change
RMBRMBUS$
(in millions, except percentages)
Alibaba China E-commerce Group:
E-commerce
- Customer management152,755168,17923,62410%
- Direct sales, logistics and others(2)50,23353,3317,4926%
202,988221,51031,1169%
Quick commerce(3)27,51737,6905,29437%
China commerce wholesale11,93813,4501,88913%
Total Alibaba China E-commerce Group242,443272,65038,29912%
Alibaba International Digital Commerce Group:
International commerce retail49,30956,4637,93115%
International commerce wholesale11,65613,0771,83712%
Total Alibaba International Digital Commerce Group60,96569,5409,76814%
Cloud Intelligence Group56,15973,22210,28530%
All others(4)165,837121,56817,077(27)%
Unallocated8881,096154
Inter-segment elimination(46,553)(42,629)(5,988)
Consolidated revenue479,739495,44769,5953%
+ +(1) To advance our “user first” strategy and enhance user experience, during the quarter ended June 30, 2025, we undertook a strategic combination of Taobao and Tmall Group, Ele.me and Fliggy into Alibaba China E-commerce Group. We simplified the financial reporting structure by reclassifying Cainiao, Amap and Digital Media and Entertainment Group (rebranded to Hujing Digital Media and Entertainment Group) into “All others”. The above presentation has been updated to conform with the new reporting structure, as reviewed by our chief operating decision maker. +(2) Direct sales, logistics and others revenue under Alibaba China E-commerce Group primarily represents direct sales businesses of Tmall Supermarket, Tmall Global and other businesses, where revenue and cost of inventory are recorded on a gross basis within the business group, as well as revenue from logistics services and value-added services. +(3) Quick commerce revenue represents quick commerce business revenue, including revenue generated through “Taobao Instant Commerce” and the Ele.me app. Quick commerce revenue is net of subsidies that are contra revenue. +(4) All others include Freshippo, Cainiao, Alibaba Health, Hujing Digital Media and Entertainment Group, Amap, Intelligent Information Platform (which mainly consists of UCWeb and Quark businesses), Lingxi Games, DingTalk and other businesses. The majority of revenue within All others consists of direct sales, where revenue and cost of inventory are recorded on a gross basis, and revenue from logistics services. The decrease was primarily due to the revenue decrease as a result of the disposal of Sun Art and Intime businesses, as well as the decrease in revenue from Cainiao, partly offset by the increase in revenue from Freshippo, Alibaba Health and Amap. + +The following table sets forth a breakdown of our adjusted EBITA by segment for the periods indicated: + +
Six months ended September 30,
20242025YoY % Change(3)
RMBRMBUS$
(in millions, except percentages)
Alibaba China E-commerce Group93,08048,8866,867(47)%
Alibaba International Digital Commerce Group(6,611)10314N/A
Cloud Intelligence Group4,9986,55892131%
All others(2,910)(4,785)(672)(64)%
Unallocated(2)(2,142)(1,640)(230)
Inter-segment elimination(819)(1,205)(169)
Consolidated adjusted EBITA85,59647,9176,731(44)%
Less: Non-cash share-based compensation expense(7,775)(6,076)(854)
Less: Amortization and impairment of intangible assets, and others(3,441)(1,488)(209)
Less: Provision for the shareholder class action lawsuits(3,145)--
Income from operations71,23540,3535,668(43)%
+ +(1) To advance our “user first” strategy and enhance user experience, during the quarter ended June 30, 2025, we undertook a strategic combination of Taobao and Tmall Group, Ele.me and Fliggy into Alibaba China E-commerce Group. We simplified the financial reporting structure by reclassifying Cainiao, Amap and Digital Media and Entertainment Group (rebranded to Hujing Digital Media and Entertainment Group) into “All others”. The above presentation has been updated to conform with the new reporting structure, as reviewed by our chief operating decision maker. +(2) Unallocated primarily relates to certain costs incurred by corporate functions and other miscellaneous items that are not allocated to individual segments. +(3) For a more intuitive presentation, widening of loss in YoY% is shown in terms of negative growth rate, and narrowing of loss in YoY% is shown in terms of positive growth rate. + +## Alibaba China E-commerce Group + +## (i) Segment revenue + +## • E-commerce Business + +Revenue from our E-commerce business in the six months ended September 30, 2025 was RMB221,510 million (US\$31,116 million), an increase of 9% compared to RMB202,988 million in the same period of 2024. + +Customer management revenue increased by 10% year-over-year, primarily due to the improvement of take rate. + +Direct sales, logistics and others revenue under E-commerce business in the six months ended September 30, 2025 was RMB53,331 million (US\$7,492 million), an increase of 6% compared to RMB50,233 million in the same period of 2024, primarily driven by the increase in revenue from logistics services and value-added services, partly offset by the decrease in revenue from certain direct sales businesses. + +## • Quick Commerce Business + +Revenue from our Quick commerce business in the six months ended September 30, 2025 was RMB37,690 million (US\$5,294 million), an increase of 37% compared to RMB27,517 million in the same period of 2024, mainly due to order growth as a result of the rollout of “Taobao Instant Commerce” at the end of April 2025. + +## • China Commerce Wholesale Business + +Revenue from our China commerce wholesale business in the six months ended September 30, 2025 was RMB13,450 million (US\$1,889 million), an increase of 13% compared to RMB11,938 million in the same period of 2024, primarily due to an increase in revenue from value-added services provided to paying members. + +## (ii) Segment adjusted EBITA + +Alibaba China E-commerce Group adjusted EBITA decreased by 47% to RMB48,886 million (US\$6,867 million) in the six months ended September 30, 2025, compared to RMB93,080 million in the same period of 2024, primarily due to the investment in quick commerce, user experiences, and technology, partly offset by double-digit revenue growth in Alibaba China E-commerce Group. + +## Alibaba International Digital Commerce Group + +## (i) Segment revenue + +## • International Commerce Retail Business + +Revenue from our International commerce retail business in the six months ended September 30, 2025 was RMB56,463 million (US\$7,931 million), an increase of 15% compared to RMB49,309 million in the same period of 2024, primarily driven by the increase in revenue contributed by AliExpress and other international businesses. As certain of our international businesses generate revenue in local currencies while our reporting currency is Renminbi, AIDC's revenue is affected by exchange rate fluctuations. + +## • International Commerce Wholesale Business + +Revenue from our International commerce wholesale business in the six months ended September 30, 2025 was RMB13,077 million (US\$1,837 million), an increase of 12% compared to RMB11,656 million in the same period of 2024, primarily due to an increase in revenue generated by cross-border related value-added services. + +## (ii) Segment adjusted EBITA + +Alibaba International Digital Commerce Group adjusted EBITA was a profit of RMB103 million (US\$14 million) in the six months ended September 30, 2025, compared to a loss of RMB6,611 million in the same period of 2024, primarily due to significant improvement in AliExpress' operating efficiency, and enhanced efficiency across various businesses. + +## Cloud Intelligence Group + +## (i) Segment revenue + +Revenue from Cloud Intelligence Group was RMB73,222 million (US\$10,285 million) in the six months ended September 30, 2025, an increase of 30% compared to RMB56,159 million in the same period of 2024. Overall revenue excluding Alibaba-consolidated subsidiaries increased by 28% year-over-year, primarily driven by public cloud revenue growth, including the increasing adoption of AI-related products. + +## (ii) Segment adjusted EBITA + +Cloud Intelligence Group adjusted EBITA increased by 31% to RMB6,558 million (US\$921 million) in the six months ended September 30, 2025, compared to RMB4,998 million in the same period of 2024, primarily due to revenue growth and improving operating efficiency, partly offset by the increasing investments in customer growth and technology innovation. + +## All Others + +## (i) Segment revenue + +Revenue from All others segment was RMB121,568 million (US\$17,077 million) in the six months ended September 30, 2025, a decrease of 27% compared to RMB165,837 million in the same period of 2024, primarily due to the revenue decrease as a result of the disposal of Sun Art and Intime businesses, as well as the decrease in revenue from Cainiao, partly offset by the increase in revenue from Freshippo, Alibaba Health and Amap. + +## (ii) Segment adjusted EBITA + +Adjusted EBITA from All others segment in the six months ended September 30, 2025 was a loss of RMB4,785 million (US\$672 million), compared to a loss of RMB2,910 million in the same period of 2024, primarily due to the increased investment in technology businesses, partly offset by the improved operating results of Freshippo, Hujing Digital Media and Entertainment Group and Alibaba Health. + +## Six Months Ended September Other Financial Results + +## Costs and Expenses + +The following tables set forth a breakdown of our costs and expenses, share-based compensation expense, and costs and expenses excluding share-based compensation expense by function for the periods indicated: + +
Six months ended September 30,% of Revenue YoY change
20242025
RMB% of RevenueRMBUS$% of Revenue
(in millions, except percentages)
Costs and expenses:
Cost of revenue290,13560.5%287,21040,34458.0%(2.5)%
Product development expenses27,5555.7%32,0964,5096.5%0.8%
Sales and marketing expenses65,16713.6%119,67416,81124.2%10.6%
General and administrative expenses23,0574.8%14,7782,0763.0%(1.8)%
Amortization and impairment of intangible assets3,4410.7%1,6332290.3%(0.4)%
Total costs and expenses409,355455,39163,969
Share-based compensation expense:
Cost of revenue1,2050.3%9131280.2%(0.1)%
Product development expenses3,5600.7%2,8624020.6%(0.1)%
Sales and marketing expenses9480.2%9581350.2%0.0%
General and administrative expenses2,5640.5%2,1373000.4%(0.1)%
Total share-based compensation expense(1)8,2776,870965
Costs and expenses excluding share-based compensation expense:
Cost of revenue288,93060.2%286,29740,21657.8%(2.4)%
Product development expenses23,9955.0%29,2344,1075.9%0.9%
Sales and marketing expenses64,21913.4%118,71616,67624.0%10.6%
General and administrative expenses20,4934.3%12,6411,7762.6%(1.7)%
Amortization and impairment of intangible assets3,4410.7%1,6332290.3%(0.4)%
Total costs and expenses excluding share-based compensation expense401,078448,52163,004
+ +(1) This includes both cash and non-cash share-based compensation expenses. + +Cost of revenue – Cost of revenue in the six months ended September 30, 2025 was RMB287,210 million (US\$40,344 million), or 58.0% of revenue, compared to RMB290,135 million, or 60.5% of revenue, in the same period of 2024. Without the effect of share-based compensation expense, cost of revenue as a percentage of revenue would have decreased from 60.2% in the same period of 2024 to 57.8% in the six months ended September 30, 2025, primarily due to the disposal of Sun Art and Intime businesses, decrease in scale of low margin direct sales businesses, improvement in monetization and operating efficiency, partly offset by higher logistics cost driven by the growth in our quick commerce business. + +Product development expenses – Product development expenses in the six months ended September 30, 2025 were RMB32,096 million (US\$4,509 million), or 6.5% of revenue, compared to RMB27,555 million, or 5.7% of revenue, in the same period of 2024. Without the effect of share-based compensation expense, product development expenses as a percentage of revenue would have increased from 5.0% in the same period of 2024 to 5.9% in the six months ended September 30, 2025, primarily attributable to our increased investment in technology development. + +Sales and marketing expenses – Sales and marketing expenses in the six months ended September 30, 2025 were RMB119,674 million (US\$16,811 million), or 24.2% of revenue, compared to RMB65,167 million, or 13.6% of revenue, in the same period of 2024. Without the effect of share-based compensation expense, sales and marketing expenses as a percentage of revenue would have increased from 13.4% in the same period of 2024 to 24.0% in the six months ended September 30, 2025, primarily attributable to the investment in user experiences of Alibaba China E-commerce Group. + +General and administrative expenses – General and administrative expenses in the six months ended September 30, 2025 were RMB14,778 million (US\$2,076 million), or 3.0% of revenue, compared to RMB23,057 million, or 4.8% of revenue, in the same period of 2024. Without the effect of share-based compensation expense, general and administrative expenses as a percentage of revenue would have decreased from 4.3% in the same period of 2024 to 2.6% in the six months ended September 30, 2025, primarily due to a one-time provision for the shareholder class action lawsuits in the same period last year and our improved cost control measures. + +Share-based compensation expense – Total share-based compensation expense included in the cost and expense items above in the six months ended September 30, 2025 was RMB6,870 million (US\$965 million), compared to RMB8,277 million in the same period of 2024. + +The following table sets forth our analysis of share-based compensation expense for the quarters indicated by type of share-based awards: + +
Six months ended September 30,
20242025YoY % Change
RMBRMBUS$
(in millions, except percentages)
By type of awards:
Alibaba Group share-based awards(1)5,8774,664655(21)%
Others(2)2,4002,206310(8)%
Total share-based compensation expense(3)8,2776,870965(17)%
+ +Share-based compensation expense decreased in the six months ended September 30, 2025 compared to the same period of 2024. This decrease was primarily due to the decrease in the number of Alibaba Group share-based awards granted as we have increased the proportion of long-term cash incentives granted after considering the macroeconomic environment and the general trends in the talent market. + +We expect that our share-based compensation expense will continue to be affected by changes in the fair value of the underlying awards and the quantity of awards we grant in the future. + +Amortization and impairment of intangible assets – Amortization and impairment of intangible assets in the six months ended September 30, 2025 was RMB1,633 million (US\$229 million), a decrease of 53% from RMB3,441 million in the same period of 2024, primarily due to full amortization of certain intangible assets. + +## Income from operations and operating margin + +Income from operations in the six months ended September 30, 2025 was RMB40,353 million (US\$5,668 million), or 8% of revenue, a decrease of 43% compared to RMB71,235 million, or 15% of revenue, in the same period of 2024, primarily due to the decrease in adjusted EBITA, partly offset by a one-time provision in the same period last year. + +## Adjusted EBITDA and Adjusted EBITA + +Adjusted EBITDA decreased 36% year-over-year to RMB62,991 million (US\$8,848 million) in the six months ended September 30, 2025, compared to RMB98,488 million in the same period of 2024. Adjusted EBITA decreased 44% year-over-year to RMB47,917 million (US\$6,731 million) in the six months ended September 30, 2025, compared to RMB85,596 million in the same period of 2024, primarily attributable to the investment in quick commerce, user experiences, and technology, partly offset by double-digit revenue growth in Alibaba China E-commerce Group, the improved operating results supported by continued growth in Cloud business, as well as enhanced operating efficiencies across various businesses. A reconciliation of net income to adjusted EBITDA and adjusted EBITA is included in the section entitled “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” of this Interim Report. + +## Adjusted EBITA by segment + +Adjusted EBITA by segment as well as a reconciliation of income from operations to adjusted EBITA are set forth in the section entitled “Six Months Ended September Segment Results” above. + +## Interest and investment income, net + +Interest and investment income, net in the six months ended September 30, 2025 was RMB37,468 million (US\$5,263 million), an increase of 119% compared to RMB17,129 million in the same period of 2024, primarily due to mark-to-market changes from our equity investments, gain from the disposal of local consumer service business of Trendyol, and the decrease in impairment of our investments. + +The above-mentioned investment gains and losses were excluded from our non-GAAP net income. + +## Other (expense) income, net + +Other (expense) income, net in the six months ended September 30, 2025 was an income of RMB1,329 million (US\$187 million), compared to expense of RMB1,221 million in the same period of 2024, was primarily due to the decrease in net exchange loss in the six months ended September 30, 2025 compared to the same period last year, arising from the exchange rate fluctuation between Renminbi and U.S. dollar. + +## Income tax expenses + +Income tax expenses in the six months ended September 30, 2025 were RMB14,415 million (US\$2,024 million), compared to RMB17,442 million in the same period of 2024. + +## Share of results of equity method investees + +Share of results of equity method investees in the six months ended September 30, 2025 was RMB3,254 million (US\$457 million), an increase of 31% compared to RMB2,483 million in the same period of 2024. The following table sets forth a breakdown of share of results of equity method investees for the periods indicated: + +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Share of profit (loss) of equity method investees
- Ant Group6,3954,280601
- Others(1,334)1,126158
Impairment loss(2,157)(5)(1)
$Others^{(1)}$ (421)(2,147)(301)
Total2,4833,254457
+ +(1) “Others” mainly include basis differences arising from equity method investees, share-based compensation expense related to share-based awards granted to employees of our equity method investees, as well as gain or loss arising from the deemed disposal of the equity method investees. + +We record our share of results of all equity method investees one quarter in arrears. The share of net profit of other equity method investees recorded in the six months ended September 30 2025, compared to the share of net losses in the same period last year, was primarily attributable to the overall improvement in the financial performance of our equity method investees. This was partly offset by the decrease in share of profit of Ant Group, which was mainly attributable to investments in new growth initiatives and technologies. + +## Net income and Non-GAAP net income + +Our net income in the six months ended September 30, 2025 was RMB62,994 million (US\$8,849 million), compared to RMB67,569 million in the same period of 2024, primarily attributable to the decrease in income from operations, partly offset by the mark-to-market changes from our equity investments, gain from the disposal of local consumer service business of Trendyol and the decrease in both the impairment of our investments and net exchange loss. + +Excluding non-cash share-based compensation expense, gains/losses of investments, impairment of goodwill and intangible assets, and certain other items, non-GAAP net income in the six months ended September 30, 2025 was RMB43,862 million (US\$6,161 million), a decrease of 43% compared to RMB77,209 million in the same period of 2024, primarily attributable to the investment in quick commerce, user experiences, and technology, partly offset by double-digit revenue growth in Alibaba China E-commerce Group, the improved operating results supported by continued growth in Cloud business, as well as enhanced operating efficiencies across various businesses. A reconciliation of net income to non-GAAP net income is included in the section entitled “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” of this Interim Report. + +## Net income attributable to ordinary shareholders + +Net income attributable to ordinary shareholders in the six months ended September 30, 2025 was RMB64,106 million (US\$9,005 million), compared to RMB68,143 million in the same period of 2024, primarily attributable to the decrease in income from operations, partly offset by the mark-to-market changes from our equity investments, gain from the disposal of local consumer service business of Trendyol and the decrease in both the impairment of our investments and net exchange loss. + +## Diluted earnings per ADS/share and non-GAAP diluted earnings per ADS/share + +Diluted earnings per ADS in the six months ended September 30, 2025 was RMB26.73 (US\$3.75), compared to RMB28.00 in the same period of 2024. Excluding non-cash share-based compensation expense, gains/losses of investments, impairment of goodwill and intangible assets, and certain other items, non-GAAP diluted earnings per ADS in the six months ended September 30, 2025 was RMB19.10 (US\$2.68), a decrease of 39% compared to RMB31.50 in the same period of 2024. + +Diluted earnings per share in the six months ended September 30, 2025 was RMB3.34 (US\$0.47 or HK\$3.66), compared to RMB3.50 in the same period of 2024. Excluding non-cash share-based compensation expense, gains/losses of investments, impairment of goodwill and intangible assets, and certain other items, non-GAAP diluted earnings per share in the six months ended September 30, 2025 was RMB2.39 (US\$0.34 or HK\$2.62), a decrease of 39% compared to RMB3.94 in the same period of 2024. + +A reconciliation of diluted earnings per ADS/share to non-GAAP diluted earnings per ADS/share is included in the section entitled “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” of this Interim Report. Each ADS represents eight ordinary shares. + +## Cash and cash equivalents, short-term investments and other treasury investments + +As of September 30, 2025, cash and cash equivalents, short-term investments and other treasury investments included in equity securities and other investments on the consolidated balance sheets, of which that are unrestricted for withdrawal and use, were RMB573,889 million (US\$80,614 million), compared to RMB597,132 million as of March 31, 2025. Other treasury investments consist of fixed deposits, certificate of deposits and marketable debt securities with original maturities over one year for treasury purposes. The decrease in cash and cash equivalents, short-term investments and other treasury investments of RMB23,243 million during the six months ended September 30, 2025, was primarily due to (i) free cash flow outflow of RMB40,655 million (US\$5,711 million), (ii) dividend payment of RMB33,621 million (US\$4,723 million), (iii) cash used in repurchase of ordinary shares of RMB7,638 million (US\$1,073 million), partly offset by (iv) net proceeds from issuance of convertible unsecured senior notes and payments for capped call transactions of RMB20,994 million (US\$2,949 million), (v) net proceeds from bank borrowings of RMB17,804 million (US\$2,501 million), (vi) net proceeds from issuance of exchangeable bonds of RMB10,986 million (US\$1,543 million) and (vii) proceeds of RMB12,026 million (US\$1,689 million) from the disposal of Intime and local consumer service business of Trendyol. + +## Net cash provided by operating activities and free cash flow + +During the six months ended September 30, 2025, net cash provided by operating activities was RMB30,771 million (US\$4,322 million), a decrease of 53% compared to RMB65,074 million in the same period of 2024. Free cash flow, a non-GAAP measurement of liquidity, was an outflow of RMB40,655 million (US\$5,711 million), compared to an inflow of RMB31,107 million in the same period of 2024. The decrease in free cash flow was mainly attributed to the increase in our cloud infrastructure expenditure and the investment in quick commerce. A reconciliation of net cash provided by operating activities to free cash flow is included in the section entitled “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” of this Interim Report. + +## Net cash used in investing activities + +During the six months ended September 30, 2025, net cash used in investing activities of RMB51,324 million (US\$7,209 million) primarily reflected capital expenditures of RMB70,177 million (US\$9,858 million) and net increase in short-term investments and other treasury investments by RMB11,119 million (US\$1,562 million), partly offset by net cash inflow of RMB29,710 million (US\$4,173 million) for investment and acquisition activities. + +## Net cash provided by financing activities + +During the six months ended September 30, 2025, net cash provided by financing activities of RMB8,171 million (US\$1,148 million) primarily reflected cash provided by net proceeds from issuance of convertible unsecured senior notes and payments for capped call transactions of RMB20,994 million (US\$2,949 million), net proceeds from bank borrowings of RMB17,804 million (US\$2,501 million), and net proceeds from issuance of exchangeable bonds of RMB10,986 million (US\$1,543 million), partly offset by dividend payment of RMB33,621 million (US\$4,723 million), and cash used in repurchase of ordinary shares of RMB7,638 million (US\$1,073 million). + +## Employees + +As of September 30, 2025, we had a total of 126,661 employees, compared to 124,320 as of March 31, 2025. + +## Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures + +The table below sets forth a reconciliation of our net income to adjusted EBITA and adjusted EBITDA for the periods indicated: + +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Net income67,56962,9948,849
Adjustments to reconcile net income to adjusted EBITA and adjusted EBITDA:
Interest and investment income, net(17,129)(37,468)(5,263)
Interest expense4,6154,995702
Other expense (income), net1,221(1,329)(187)
Income tax expenses17,44214,4152,024
Share of results of equity method investees(2,483)(3,254)(457)
Income from operations71,23540,3535,668
Non-cash share-based compensation expense7,7756,076854
Amortization and impairment of intangible assets, and others3,4411,488209
Provision for the shareholder class action lawsuits3,145--
Adjusted EBITA85,59647,9176,731
Depreciation and impairment of property and equipment, and operating lease cost relating to land use rights12,89215,0742,117
Adjusted EBITDA98,48862,9918,848
+ +## Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures (Continued) + +The table below sets forth a reconciliation of our net income to non-GAAP net income for the periods indicated: + +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Net income67,56962,9948,849
Adjustments to reconcile net income to non-GAAP net income:
Non-cash share-based compensation expense7,7756,076854
Amortization and impairment of intangible assets3,4411,633229
Provision for the shareholder class action lawsuits3,145--
Gain on deemed disposals/disposals/revaluation of investments(8,116)(29,320)(4,119)
Impairment of investments, and others5,0672,455345
Tax effects(1)(1,672)243
Non-GAAP net income77,20943,8626,161
+ +(1) Tax effects primarily comprise tax effects relating to non-cash share-based compensation expense, amortization and impairment of intangible assets and certain gains and losses from investments, and others. + +## Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures (Continued) + +The table below sets forth a reconciliation of our diluted earnings per share/ADS to non-GAAP diluted earnings per share/ADS for the periods indicated: + +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions, except per share data)
Net income attributable to ordinary shareholders – basic68,14364,1069,005
Dilution effect on earnings arising from non-cash share-based awards operated by equity method investees and subsidiaries(131)(258)(36)
Adjustments for interest expense attributable to convertible unsecured senior notes9514320
Net income attributable to ordinary shareholders – diluted68,10763,9918,989
Non-GAAP adjustments to net income attributable to ordinary shareholders(1)8,521(18,250)(2,564)
Non-GAAP net income attributable to ordinary shareholders for computing non-GAAP diluted earnings per share/ADS76,62845,7416,425
Weighted average number of shares on a diluted basis for computing non-GAAP diluted earnings per share/ADS (million shares)(2)19,45919,154
Diluted earnings per share(2)(3)3.503.340.47
Non-GAAP diluted earnings per share(2)(4)3.942.390.34
Diluted earnings per ADS(2)(3)28.0026.733.75
Non-GAAP diluted earnings per ADS(2)(4)31.5019.102.68
+ +(1) Non-GAAP adjustments excluding the attributions to the noncontrolling interests. See the table above for items regarding the reconciliation of net income to non-GAAP net income (before excluding the attributions to the noncontrolling interests). +(2) Each ADS represents eight ordinary shares. +(3) Diluted earnings per share is derived from dividing net income attributable to ordinary shareholders by the weighted average number of outstanding ordinary shares, on a diluted basis. Diluted earnings per ADS is derived from the diluted earnings per share after adjusting for the ordinary share-to-ADS ratio. +(4) Non-GAAP diluted earnings per share is derived from dividing non-GAAP net income attributable to ordinary shareholders by the weighted average number of outstanding ordinary shares for computing non-GAAP diluted earnings per share, on a diluted basis. Non-GAAP diluted earnings per ADS is derived from the non-GAAP diluted earnings per share after adjusting for the ordinary share-to-ADS ratio. + +## Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures (Continued) + +The table below sets forth a reconciliation of net cash provided by operating activities to free cash flow for the periods indicated: + +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Net cash provided by operating activities65,07430,7714,322
Less: Purchase of property and equipment (excluding land use rights and construction in progress relating to office campuses)(28,916)(70,057)(9,841)
Less: Changes in the buyer protection fund deposits(5,051)(1,369)(192)
Free cash flow31,107(40,655)(5,711)
+ +## Non-GAAP Financial Measures + +To supplement our unaudited condensed consolidated financial statements, which are prepared and presented in accordance with GAAP, we use the following non-GAAP financial measures: for our consolidated results, adjusted EBITDA (including adjusted EBITDA margin), adjusted EBITA (including adjusted EBITA margin), non-GAAP net income, non-GAAP diluted earnings per share/ADS and free cash flow. For more information on these non-GAAP financial measures, please refer to the table captioned “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” in this Interim Report. + +We believe that adjusted EBITDA, adjusted EBITA, non-GAAP net income and non-GAAP diluted earnings per share/ADS help identify underlying trends in our business that could otherwise be distorted by the effect of certain income or expenses that we include in income from operations, net income and diluted earnings per share/ADS. We believe that these non-GAAP measures provide useful information about our core operating results, enhance the overall understanding of our past performance and future prospects and allow for greater visibility with respect to key metrics used by our management in its financial and operational decision-making. We present three different income measures, namely adjusted EBITDA, adjusted EBITA and non-GAAP net income in order to provide more information and greater transparency to investors about our operating results. + +We consider free cash flow to be a liquidity measure that provides useful information to management and investors about the amount of cash generated by our business that can be used for strategic corporate transactions, including investing in our new business initiatives, making strategic investments and acquisitions and strengthening our balance sheet. + +Adjusted EBITDA, adjusted EBITA, non-GAAP net income, non-GAAP diluted earnings per share/ADS and free cash flow should not be considered in isolation or construed as an alternative to income from operations, net income, diluted earnings per share/ADS, cash flows or any other measure of performance or as an indicator of our operating performance. These non-GAAP financial measures presented here do not have standardized meanings prescribed by U.S. GAAP and may not be comparable to similarly titled measures presented by other companies. Other companies may calculate similarly titled measures differently, limiting their usefulness as comparative measures to our data. + +Adjusted EBITDA represents net income before interest and investment income, net, interest expense, other income (expense), net, income tax expenses, share of results of equity method investees, certain non-cash expenses, consisting of share-based compensation expense, amortization and impairment of intangible assets, impairment of goodwill, depreciation and impairment of property and equipment, and operating lease cost relating to land use rights, and others (including provision in relation to matters outside the ordinary course of business), which we do not believe are reflective of our core operating performance during the periods presented. + +Adjusted EBITA represents net income before interest and investment income, net, interest expense, other income (expense), net, income tax expenses, share of results of equity method investees, certain non-cash expenses, consisting of share-based compensation expense, amortization and impairment of intangible assets, impairment of goodwill, and others (including provision in relation to matters outside the ordinary course of business), which we do not believe are reflective of our core operating performance during the periods presented. + +Non-GAAP net income represents net income before non-cash share-based compensation expense, amortization and impairment of intangible assets, gain or loss on deemed disposals/disposals/revaluation of investments, impairment of goodwill and investments, and others (including provision in relation to matters outside the ordinary course of business), and adjustments for the tax effects. + +Non-GAAP diluted earnings per share represents non-GAAP net income attributable to ordinary shareholders divided by the weighted average number of outstanding ordinary shares for computing non-GAAP diluted earnings per share on a diluted basis. Non-GAAP diluted earnings per ADS represents non-GAAP diluted earnings per share after adjusting for the ordinary share-to-ADS ratio. + +Free cash flow represents net cash provided by operating activities as presented in our unaudited condensed consolidated cash flow statement less purchases of property and equipment (excluding acquisition of land use rights and construction in progress relating to office campuses) and intangible assets (excluding those acquired through acquisitions), as well as adjustments to exclude from net cash provided by operating activities the buyer protection fund deposits from merchants on our marketplaces. We deduct certain items of cash flows from investing activities in order to provide greater transparency into cash flow from our revenue-generating business operations. We exclude “acquisition of land use rights and construction in progress relating to office campuses” because the office campuses are used by us for corporate and administrative purposes and are not directly related to our revenue-generating business operations. We also exclude buyer protection fund deposits from merchants on our marketplaces because these deposits are restricted for the purpose of compensating buyers for claims against merchants. + +The table captioned “Reconciliations of Non-GAAP Measures to the Nearest Comparable U.S. GAAP Measures” in this Interim Report has more details on the non-GAAP financial measures that are most directly comparable to GAAP financial measures and the related reconciliations between these financial measures. + +## Other Financial Information + +## Liquidity and Capital Resources + +We fund our operations and strategic investments from cash generated from our operations and through debt and equity financing. We generated RMB65,074 million and RMB30,771 million (US\$4,322 million) of cash from operating activities for the six months ended September 30, 2024 and 2025, respectively. As of September 30, 2025, we had cash and cash equivalents, short-term investments and other treasury investments of RMB573,889 million (US\$80,614 million) that are unrestricted for withdrawal and use. Short-term investments include investments in fixed deposits with original maturities between three months and one year and certain investments in wealth management products, certificates of deposits, marketable debt securities and other investments whereby we have the intention to redeem within one year. Other treasury investments mainly include investments in fixed deposits, certificates of deposits and marketable debt securities with original maturities over one year for treasury purposes. The remaining maturities of these treasury investments held by us generally range from one to five years. + +We believe that our current levels of cash and cash flows from operations will be sufficient to meet our anticipated cash needs for at least the next twelve months. However, we may need additional cash resources in the future if we find and wish to pursue opportunities for investment, acquisition, strategic cooperation or other similar actions, which may include investing in technology, infrastructure, including data management and analytics solutions, or related talent. If we determine that our cash requirements exceed our amounts of cash on hand or if we decide to further optimize our capital structure, we may seek to issue additional debt or equity securities or obtain credit facilities or other sources of funding. + +In July 2025, we issued zero coupon exchangeable bonds due 2032 by reference to the ordinary shares of our subsidiary, Alibaba Health Information Technology Limited, that are listed on the Hong Kong Stock Exchange, for an aggregate principal amount of approximately HK\$12 billion. + +In September 2025, we issued zero coupon convertible unsecured senior notes due 2032 for an aggregate principal amount of approximately US\$3.2 billion. + +We have made a series of amendments to our syndicated loan facility. Subsequent to the partial repayment of US\$830 million in January 2025, the size of the US\$4.0 billion syndicated loan was reduced to US\$3.17 billion. In September 2025, we amended the loan facility and reduced the pricing terms to Secured Overnight Financing Rate (“SOFR”) plus 66 basis points. Effective on November 28, 2025, the facility was restructured as a revolving credit facility with drawdowns permitted in both U.S. dollars and Hong Kong dollars, and the expiration date of the facility was extended to September 30, 2028, with an option to further extend to September 30, 2030. The interest rate of the credit facility is 66 basis points over SOFR or Hong Kong Interbank Offered Rate (“HIBOR”), and the margin will be 81 basis points for the optional extension period. Subsequently, we submitted a prepayment notice to repay the outstanding balance of US\$3.17 billion under the revolving credit facility. + +In September 2025, we amended the terms of a US\$6.5 billion revolving credit facility agreement. The size of the credit facility was amended to US\$3.33 billion and the utilization currency was also amended from U.S. dollar only to both U.S. dollar and Hong Kong dollar. The interest rate of the credit facility was adjusted to SOFR or HIBOR plus 66 basis points. The expiration date of the credit facility was extended from June 24, 2026 to September 30, 2028, with an option to further extend to September 30, 2030 and the margin will be 81 basis points for the optional extension period. We have not yet drawn down this facility. + +The Group monitors its financial health and liquidity position by reviewing its total debts to adjusted EBITDA ratio and total debts to total capital ratio. Our total debts is defined as the sum of bank borrowings, unsecured senior notes, convertible unsecured senior notes and exchangeable bonds. Our total debts to adjusted EBITDA ratio is calculated by dividing total debts by adjusted EBITDA for the last twelve months. Our total debts to total capital ratio is calculated by dividing total debts by total capital. Total capital is calculated as total equity plus total debts. The Group's total debts to adjusted EBITDA ratio was 1.14 and 1.69 as at March 31, 2025 and September 30, 2025, respectively. The Group's total debts to total capital ratio was $17.62\%$ and $20.35\%$ as at March 31, 2025 and September 30, 2025, respectively. + +## Significant Investments + +Our significant investment consists of Ant Group. Ant Group provides comprehensive digital payment services and facilitates digital financial and value-added services for consumers and merchants, in China and across the world. As of September 30, 2025, our equity interest in Ant Group on a fully diluted basis was 33%. During the six months ended September 30, 2025, dividend received from Ant Group amounted to RMB3,293 million (US\$463 million). + +We did not hold any other significant investments as of September 30, 2025. + +## Material Investments, Acquisitions and Dispositions + +Our material investments, acquisitions and dispositions in the six months ended September 30, 2025 and the period through the date of this Interim Report are set forth below. + +In May 2025, we agreed to sell 85% of the equity interest in Trendyol GO, a wholly-owned subsidiary of Trendyol that operates local service business in Türkiye (the “Disposal”). The cash consideration for the Disposal is approximately US\$0.7 billion (RMB5 billion). The Disposal was completed during the six months ended September 30, 2025. + +Save as disclosed above, as at the date of this Interim Report, the Group did not have detailed future plans for material investments or capital assets. + +## Pledge of Assets + +Certain of the Group's bank borrowings are collateralized by a pledge of certain buildings and property improvements, construction in progress and land use rights in the PRC, receivables and other treasury investments with carrying values of RMB30,213 million and RMB25,417 million, as of March 31, 2025 and September 30, 2025, respectively. In addition, certain of the Group's payables are collateralized by a pledge of certain short-term investments and other treasury investments with carrying values of RMB3,697 million and RMB4,155 million as of March 31, 2025 and September 30, 2025, respectively. + +## Foreign Exchange Risk + +Foreign currency risk arises from future commercial transactions, recognized assets and liabilities and net investments in foreign operations. Although we operate businesses in different countries and regions, most of our revenue-generating transactions, and a majority of our expense-related transactions, are denominated in Renminbi, which is the functional currency of our major operating subsidiaries and the reporting currency of our financial statements. When considered appropriate, we enter into hedging activities with regard to exchange rate risk. + +The value of the Renminbi against the U.S. dollar and other currencies may fluctuate and is affected by, among other things, changes in political and economic conditions and the foreign exchange policy adopted by the governments. It is difficult to predict how market forces or PRC or U.S. government policy may impact the exchange rate between the Renminbi and the U.S. dollar in the future. There remains significant international pressure on the PRC government to adopt a more flexible currency policy, which could result in greater fluctuations of the Renminbi against the U.S. dollar. + +To the extent that we need to convert U.S. dollars into Renminbi for our operations, appreciation of the Renminbi against the U.S. dollar would reduce the Renminbi amount we receive from the conversion. Conversely, if we decide to convert Renminbi into U.S. dollars for the purpose of making payments for dividends on our ordinary shares or ADSs, servicing our outstanding debts, or for other business purposes, appreciation of the U.S. dollar against the Renminbi would reduce the U.S. dollar amounts available to us. + +## Contingent Liabilities + +Save as disclosed in this Interim Report, as at September 30, 2025, the Group had no material contingent liabilities. + +## Capital Expenditure and Capital Commitment + +Our capital expenditures have been incurred primarily in relation to (i) the acquisition of computer equipment and construction of data centers relating to our Cloud business and the operation of our mobile platforms and websites; (ii) the acquisition of infrastructure for logistics services and direct sales businesses; and (iii) the acquisition of land use rights and construction of corporate campuses and office facilities. In the six months ended September 30, 2024 and 2025, our capital expenditures totaled RMB29,585 million and RMB70,177 million (US\$9,858 million), respectively. + +The Group’s capital commitments primarily relate to capital expenditures contracted for purchase of property and equipment, including the construction of corporate campuses. Total capital commitments contracted but not provided for amounted to RMB45,321 million and RMB29,483 million (US\$4,141 million) as of March 31, 2025 and September 30, 2025, respectively. + +## Remuneration Policy + +Alibaba Group periodically reviews its remuneration policy and compensation packages to ensure they are competitive within the industry. Furthermore, discretionary bonuses and other long-term incentives may be awarded to selected employees based on various factors including but not limited to individual performance and the overall performance of our business. We have established learning and training programs to develop our employees both personally and professionally, helping them to better realize their potential and create value, thereby supporting their long-term career success. + +The Company’s subsidiaries in the PRC participate in a government-mandated multi-employer defined contribution plan. These plans provide employees with housing, pension, medical, maternity, work-related injury and unemployment benefits, as well as other welfare benefits. The relevant labor regulations require the Company’s subsidiaries in the PRC to make monthly contributions to the local labor and social security authorities based on the applicable benchmarks and rates stipulated by the local government. Additionally, we provide commercial health and accidental insurance for our employees. The Company’s subsidiaries also formulate their own unique benefit plans and assistance programs tailored to their specific business needs. + +Alibaba Group also makes payments to other defined contribution plans and defined benefit plans for the benefit of employees employed by subsidiaries outside of the PRC. + +Share-based awards such as restricted share units, incentive and non-statutory stock options, restricted shares and share appreciation rights may be granted to any directors, employees, service providers and consultants of Alibaba Group or affiliated companies under equity incentive plans adopted since the inception of the Company. For details of Alibaba Group's equity incentive plans, please refer to the section titled “Equity Incentive Plans”. + +## Subsequent Events + +Save as disclosed in this Interim Report, as at the date of this Interim Report, there were no significant events that might affect Alibaba Group since September 30, 2025. + +## DIRECTORS AND CHIEF EXECUTIVE OFFICER + +The following table sets forth certain information relating to our directors and chief executive officer. + +
Name(1)AgePosition/Title
Joseph C. TSAI $^{\dagger(a)}$ 61Chairman
Eddie Yongming WU $^{\dagger(b)}$ 50Director and Chief Executive Officer
J. Michael EVANS $^{\dagger(a)}$ 68Director and President
Maggie Wei WU $^{\dagger(c)}$ 57Director $^{(d)}$
Jerry YANG $^{(b)}$ 57Independent director
Wan Ling MARTELLO $^{(b)}$ 67Independent director
Weijian SHAN $^{(c)}$ 72Independent director
Irene Yun-Lien LEE $^{(a)}$ 72Independent director
Albert Kong Ping NG $^{(b)}$ 68Independent director
Kabir MISRA $^{(c)}$ 56Independent director
+ +$^{\dagger}$ Director nominated by the Alibaba Partnership. +(a) Group I directors. Current term of office will expire at our 2027 annual general meeting. +(b) Group II directors. Current term of office will expire at our 2028 annual general meeting. +(c) Group III directors. Current term of office will expire at our 2026 annual general meeting. +(d) Beginning from April 1, 2025, Maggie Wu has started to serve as a non-executive director. +(1) The business address of our directors is 26/F Tower One, Times Square, 1 Matheson Street, Causeway Bay, Hong Kong SAR., the People's Republic of China. + +## Biographical Information + +Joseph C. TSAI (蔡崇信) joined our company in 1999 as a member of the Alibaba founding team and has served on our board of directors since our inception. He was chief financial officer until 2013, our executive vice chairman until September 2023 and currently serves as our Chairman. Joe is a founding member of the Alibaba Partnership and a board member of our affiliate Ant Group. From 1995 to 1999, Joe was a private equity investor based in Asia with Investor AB of Sweden's Wallenberg family. Prior to that, he was general counsel of Rosecliff, Inc., a management buyout firm based in New York. From 1990 to 1993, Joe was an associate attorney in the tax group of Sullivan & Cromwell LLP, a New York-based international law firm. Joe is qualified to practice law in the State of New York. Joe received his bachelor's degree in Economics and East Asian Studies from Yale College and a juris doctor degree from Yale Law School. + +Eddie Yongming WU (吳泳銘) has served as our Chief Executive Officer and director since September 2023. Eddie is one of our co-founders and a member of the Alibaba Partnership. Eddie was technology director of Alibaba at the company's inception in 1999. He served as chief technology officer of Alipay from December 2004, and became business director of our monetization platform, Alimama, in November 2005 and was promoted to its general manager in December 2007. In September 2008, he became chief technology officer of Taobao, and in October 2011 he took on the role of head of Alibaba Group's search, advertising and mobile business. Eddie served as a non-executive director of Alibaba Health Information Technology Limited, a company listed on the Main Board of the Hong Kong Stock Exchange, from April 2015 to October 2021 and chairman of Alibaba Health from April 2015 to March 2020. From September 2014 to September 2019, Eddie was a special assistant to Alibaba Group's chairman. In August 2015, Eddie founded Vision Plus Capital, a venture capital firm focused on investing in the areas of advanced technologies, enterprise services and digital healthcare. Eddie graduated from the College of Information Engineering of Zhejiang University of Technology in June 1996. + +J. Michael EVANS has been our president since August 2015 and our director since September 2014. Mike served as vice chairman of The Goldman Sachs Group, Inc. from February 2008 until his retirement in December 2013. He served as chairman of Asia operations at Goldman Sachs from 2004 to 2013 and was the global head of Growth Markets at Goldman Sachs from January 2011 to December 2013. He also co-chaired the Business Standards Committee of Goldman Sachs from 2010 to 2013. Mike joined Goldman Sachs in 1993, became a partner of the firm in 1994 and held various leadership positions within the firm’s securities business while based in New York and London, including global head of equity capital markets and global co-head of the equities division, and global co-head of the securities business. Mike is a trustee of the Asia Society and a member of the Advisory Council for the Bendheim Center for Finance at Princeton University. Mike received his bachelor’s degree in politics from Princeton University in 1981. + +Maggie Wei WU (武衛) has been our director since September 2020. Maggie joined our company in July 2007 as chief financial officer of Alibaba.com. She served as our chief financial officer from May 2013 to March 2022 and our head of strategic investments from June 2019 to March 2022. She was voted the best CFO in FinanceAsia’s annual poll for Asia’s Best Managed Companies in 2010. In 2018, she was named as one of the world’s 100 most powerful women by Forbes. Before joining Alibaba, Maggie was an audit partner at KPMG in Beijing. Maggie is a member of the Association of Chartered Certified Accountants (ACCA). She received a bachelor’s degree in accounting from Capital University of Economics and Business. + +Jerry YANG (楊致遠) has served as our director since September 2014 and previously held the role from October 2005 to January 2012. Since March 2012, Jerry has been the founding partner of AME Cloud Ventures, a venture capital firm. Jerry is also a co-founder of Yahoo! Inc., where he served as Chief Yahoo! and as a member of its board of directors from March 1995 to January 2012. Jerry held the position of Yahoo!'s CEO from June 2007 to January 2009 and served as a director of Yahoo! Japan from January 1996 to January 2012. Jerry also served as an independent director of Cisco Systems, Inc. from July 2000 to November 2012 and Lenovo Group Limited, a company listed on the Main Board of the Hong Kong Stock Exchange, from November 2014 to November 2023. He is currently an independent director of Workday Inc., a company listed on the NYSE. He also serves as a director of various private companies and foundations. Jerry holds bachelor's and master's degrees in electrical engineering from Stanford University. Since 2017, he has served on Stanford's board of Trustees, including as board chair from 2021 to 2025. He was previously on Stanford's board of Trustees from 2005 to 2015, during which he also held the role of vice chair. + +Wan Ling MARTELLO has been our director since September 2015. She is a founding partner of BayPine, a private equity firm based in Boston, U.S.A., a role she has held since February 2020. She is on the board of portfolio companies of BayPine. She has also served as a director of Nasdaq-listed Lovesac since November 2025. She served as the executive vice president and chief executive officer of the Asia, Oceania, and sub-Saharan Africa region for Nestlé SA from May 2015 to December 2018. She was Nestlé’s global chief financial officer from April 2012 to May 2015, and executive vice president from November 2011 to March 2012. Prior to Nestlé, Wan Ling was a senior executive at Walmart Stores Inc., a global retailer, from 2005 to 2011. Her roles included executive vice president and chief operating officer for Global eCommerce, and senior vice president, chief financial officer and strategy for Walmart International. Before Walmart, she was president, U.S.A. at NCH Marketing Services Inc. She was with the firm from 1998 to 2005. She also worked at Borden Foods Corporation and Kraft Inc. where she held various senior management positions. Wan Ling received a master’s degree in business administration (management information systems) from the University of Minnesota and a bachelor’s degree in business administration and accountancy from the University of the Philippines. + +Weijian SHAN (單偉建) has been our director since March 2022. He is the executive chairman and a co-founder of PAG, a leading private equity firm in Asia. Between 1998 and 2010, he was a partner of the private equity firm TPG and co-managing partner of TPG Asia (formerly known as Newbridge Capital). Previously, he was a managing director of JP Morgan, where he was concurrently the chief representative for China between 1993 and 1998. He was an assistant professor at the Wharton School of the University of Pennsylvania between 1987 and 1993. Shan is a Trustee of the British Museum. He is also a member of the International Advisory Council of Hong Kong Exchanges and Clearing Limited. He served as an independent director of Singapore-listed Wilmar International Limited between 2018 and 2021. He also served as a director of Nasdaq-listed iQiyi, Inc since December 2022. He holds an M.A. and a Ph.D. from the University of California, Berkeley, and an M.B.A. from the University of San Francisco. He graduated with a major in English from the Beijing Institute of Foreign Trade (currently the Beijing University of International Business and Economics). + +Irene Yun-Lien LEE (利蘊蓮) has been our director since August 2022. Irene is the executive chairman of Hysan Development Limited and serves as a member of the board of trustees of the Better Hong Kong Foundation. Irene was on the board of many listed and unlisted companies in Hong Kong, Singapore, UK and Australia. She was a member of the Australian Takeovers Panel, a member of the Advisory Council of JP Morgan Australia, and a member of the Exchange Fund Advisory Committee of the Hong Kong Monetary Authority. She was the independent non-executive chairman of Hang Seng Bank Limited, an independent non-executive director of HSBC Holding plc and The Hongkong and Shanghai Banking Corporation Limited. She was also an independent non-executive director of CLP Holdings Limited and Cathay Pacific Airways Limited, amongst others. Irene had a long career in financial services and held senior positions at Citibank in New York, London and Sydney. She was the global head of corporate finance at the Commonwealth Bank of Australia and she held other senior positions in investment banking and funds management in a number of international financial institutions. Irene received a Bachelor of Arts degree from Smith College, United States of America, and is a Barrister-at-Law in England and Wales and a member of the Honourable Society of Gray's Inn, United Kingdom. She was awarded the degree of Doctor of Social Science, honoris causa from the Chinese University of Hong Kong in November 2022. + +Albert Kong Ping NG (吳港平) has been our director since August 2022 and chairman of our Audit Committee since December 2022. Albert currently serves as an independent non-executive director and chairman of the audit committee of a number of public companies, including Ping An Insurance (Group) Company of China, Ltd., a company listed on the Shanghai Stock Exchange and the Hong Kong Stock Exchange, China International Capital Corporation Limited, a company listed on the Shanghai Stock Exchange and the Hong Kong Stock Exchange and Shui On Land Limited, a company listed on the Hong Kong Stock Exchange. Albert worked at Ernst & Young China from April 2007 to June 2020, where he was the chairman of Ernst & Young China and a member of Ernst & Young's Global Executive Board. Prior to joining Ernst & Young, he was Greater China managing partner of Arthur Andersen, managing partner – China Operation of PricewaterhouseCoopers and managing director of Citigroup – China Investment Banking. Albert is a member of the Hong Kong Institute of Certified Public Accountants (HKICPA), Chartered Accountants of Australia and New Zealand (CAANZ), CPA Australia (CPAA) and Association of Chartered Certified Accountants (ACCA). He received a bachelor's degree in business administration and a master's degree in business administration from the Chinese University of Hong Kong. + +Kabir MISRA has been our director since September 2020, redesignated as our independent director since February 2023, and is currently managing partner at RPS Ventures, a venture capital firm in Palo Alto, CA. Prior to October 2018, Kabir was a managing partner at SoftBank Investment Advisors (which manages the SoftBank Vision Fund) and SoftBank Capital. He worked with SoftBank from 2006 to 2022 (as advisor from + +2018 to 2022) and has assisted Mr. Masayoshi Son with our company, and his duties as one of our directors, since before our IPO. Kabir also represented SoftBank at various points on the boards of its investee companies, including other e-commerce and payments companies Flipkart, Paytm, Tokopedia, Coupang and BigCommerce. Prior to joining SoftBank, Kabir worked as an investment banker in the U.S. and Hong Kong. Kabir is currently also an independent director of PayActiv and Cargomatic. He received a Bachelor of Arts degree in Economics from Harvard University and a master's degree in business administration from the Stanford Graduate School of Business. + +## Board Committees + +Our board of directors has established an audit committee, a compensation committee, a nominating and corporate governance committee, a sustainability committee, a compliance and risk committee and a capital management committee. + +## Audit Committee + +Our audit committee currently consists of Albert Ng, Wan Ling Martello and Weijian Shan. Mr. Ng is the chairman of our audit committee. + +## Compensation Committee + +Our compensation committee currently consists of Jerry Yang, Albert Ng and Kabir Misra. Mr. Yang is the chairman of our compensation committee. + +## Nominating and Corporate Governance Committee + +Our nominating and corporate governance committee currently consists of Irene Lee and Jerry Yang. Ms. Lee is the chairman of our nominating and corporate governance committee. + +## Sustainability Committee + +Our sustainability committee currently consists of Jerry Yang, Joe Tsai and Maggie Wu. Mr. Yang is the chairman of our sustainability committee. + +## Compliance and Risk Committee + +Our compliance and risk committee currently consists of Irene Lee, Albert Ng, Kabir Misra and J. Michael Evans. Ms. Lee is the chairman of our compliance and risk committee. + +## Capital Management Committee + +Our capital management committee currently consists of Joe Tsai, Eddie Wu, J. Michael Evans and Maggie Wu. Mr. Tsai is the chairman of our capital management committee. + +## DISCLOSURE OF INTERESTS + +The following tables set forth: + +- the interest and short positions of our directors and chief executive officer in the shares, underlying shares and debentures of our Company or its associated corporations within the meaning of Part XV of the SFO, which were required (a) to be notified to us and the Hong Kong Stock Exchange pursuant to Divisions 7 and 8 of Part XV of the SFO (including interests and short positions which they were taken or deemed to have under such provisions of the SFO); or (b) to be recorded in the register required to be kept by us pursuant to Section 352 of the SFO; or (c) as otherwise notified to us and the Hong Kong Stock Exchange pursuant to the Model Code; and +- the interest and short positions of persons other than our directors and chief executive officer in the Shares and underlying Shares which would fall to be disclosed to us pursuant to Divisions 2 and 3 of Part XV of the SFO or as recorded in the register required to be kept by us pursuant to Section 336 of the SFO, + +Information about persons other than our directors and chief executive officer, including our substantial shareholders (as defined under SFO), was obtained solely from publicly available information on the website of the Hong Kong Stock Exchange or as otherwise the Company is aware, and the Company cannot guarantee the accuracy and completeness of such information. + +## Directors and Chief Executive Officer + +Interest in our Company + +
Name of director / chief executive officerCapacity / nature of interestNumber of shares/Approximate % of shareholding (3)
underlying shares (1),(2)
(in the number of Shares)(in the number of ADSs)
Joseph C. TSAI (4)Beneficial owner881,072 (L)110,134 (L)0.00%
Interest of spouse1,280,000 (L)160,000 (L)0.01%
Founder of a discretionary trust who can influence how the trustee exercises his discretion12,658,152 (L)1,582,269 (L)0.06%
Interest in controlled corporation260,924,840 (L)32,615,605 (L)1.37%
275,744,064 (L)34,468,008 (L)1.44%
Eddie Yongming WU (5)Beneficial owner18,199,752 (L)2,274,969 (L)0.10%
Interest of spouse108,000 (L)13,500 (L)0.00%
Founder of a discretionary trust who can influence how the trustee exercises his discretion12,320,000 (L)1,540,000 (L)0.06%
30,627,752 (L)3,828,469 (L)0.16%
J. Michael EVANS (6)Beneficial owner9,964,000 (L)1,245,500 (L)0.05%
Maggie Wei WU (7)Beneficial owner3,661,632 (L)457,704 (L)0.02%
Founder of a discretionary trust who can influence how the trustee exercises his discretion7,200,000 (L)900,000 (L)0.04%
10,861,632 (L)1,357,704 (L)0.06%
Jerry YANGBeneficial owner485,072 (L)60,634 (L)0.00%
Wan Ling MARTELLOBeneficial owner360,000 (L)45,000 (L)0.00%
Weijian SHANBeneficial owner144,800 (L)18,100 (L)0.00%
Irene Yun-Lien LEEBeneficial owner133,600 (L)16,700 (L)0.00%
Albert Kong Ping NGBeneficial owner133,600 (L)16,700 (L)0.00%
Kabir MISRABeneficial owner528,800 (L)66,100 (L)0.00%
+ +Notes: +1. The letter "L" stands for long position. +2. Each ADS represents eight Shares. +3. The calculation is based on a total of 19,088,280,284 Shares in issue as of September 30, 2025. +4. The interests comprised (i) 804,405 Shares held by Mr. Joseph C. TSAI; (ii) 76,667 Shares underlying the outstanding RSUs granted to Mr. Joseph C. TSAI; (iii) 1,280,000 Shares held by his spouse; (iv) 12,658,152 Shares held by Joe and + +Clara Tsai Foundation Limited, a company incorporated under the law of the Island of Guernsey that has granted Mr. Joseph C. TSAI a revocable proxy over these Shares and which is wholly-owned by Joe and Clara Tsai Foundation; (v) 147,385,672 Shares held by Parufam Limited, a Bahamas corporation of which Mr. Joseph C. TSAI is the sole director; and (vi) 113,539,168 Shares held by PMH Holding Limited, a British Virgin Islands corporation of which Mr. Joseph C. TSAI is the sole director. + +5. The interests comprised (i) 933,085 Shares held by Mr. Eddie Yongming WU; (ii) 1,266,667 Shares underlying the outstanding RSUs and 16,000,000 Shares underlying the outstanding share options granted to Mr. Eddie Yongming WU; (iii) 108,000 Shares held by his spouse; and (iv) 12,320,000 Shares held by a discretionary trust of which Mr. Eddie Yongming WU is a founder. +6. The interests comprised 764,000 Shares underlying the outstanding RSUs and 9,200,000 Shares underlying the outstanding share options granted to Mr. J. Michael EVANS. +7. The interests comprised (i) 3,613,352 Shares held by Ms. Maggie Wei WU; (ii) 48,280 Shares underlying the outstanding RSUs granted to Ms. Maggie Wei WU; and (iii) 7,200,000 Shares held by a discretionary trust of which Ms. Maggie Wei WU is a founder. + +Save as disclosed above, as of September 30, 2025, none of our directors or chief executive officer had any interest or short position in the shares, underlying shares or debentures of our Company or any of its associated corporations (within the meaning of Part XV of the SFO) which were required to be notified to us and the Hong Kong Stock Exchange pursuant to Divisions 7 and 8 of Part XV of the SFO (including interests and short positions which were taken or deemed to have taken under such provisions of the SFO), or which were recorded in the register required to be kept pursuant to Section 352 of the SFO, to be entered in the register referred to therein, or which were required, pursuant to the Model Code, to be notified to us and the Hong Kong Stock Exchange. + +Substantial Shareholders + +
Name of shareholderCapacity / nature of interestNumber of shares/ underlying shares (1),(2)Approximate % of shareholding (3)
(in the number of Shares)(in the number of ADSs)
JPMorgan Chase & Co. (4)Beneficial owner411,801,958 (L)51,475,245 (L)2.16%
418,739,874 (S)52,342,484 (S)2.19%
Trustee1,753,320 (L)219,165 (L)0.01%
Investment manager281,850,861 (L)35,231,358 (L)1.47%
6,527,284 (S)815,911 (S)0.04%
Person having a security interest in shares85,927,145 (L)10,740,893 (L)0.45%
Approved lending agent581,658,439 (L)72,707,305 (L)3.05%
581,658,439 (P)72,707,305 (P)3.05%
1,362,991,723 (L)170,373,966 (L)7.14%
425,267,158 (S)53,158,395 (S)2.23%
581,658,439 (P)72,707,305 (P)3.05%
Citigroup Inc. (5)Interest in controlled corporation183,474,400 (L)22,934,300 (L)0.96%
197,906,617 (S)24,738,327 (S)1.04%
Person having a security interest in shares160 (L)20 (L)0.00%
Approved lending agent832,721,084 (L)104,090,136 (L)4.36%
832,721,084 (P)104,090,136 (P)4.36%
1,016,195,644 (L)127,024,456 (L)5.32%
197,906,617 (S)24,738,327 (S)1.04%
832,721,084 (P)104,090,136 (P)4.36%
BlackRock, Inc. (6)Interest in controlled corporation983,512,344 (L)122,939,043 (L)5.15%
3,297,800 (S)412,225 (S)0.02%
+ +Notes: +1. The letter “L” stands for long position, the letter “S” stands for short position and the letter “P” stands for lending pool. +2. Each ADS represents eight Shares. The number of ADSs is, where applicable, rounded down to the nearest whole number and for reference only. +3. The calculation is based on a total of 19,088,280,284 Shares in issue as of September 30, 2025. +4. According to the disclosure of interests notice filed by JPMorgan Chase & Co. regarding the relevant event dated September 26, 2025, (a) 411,801,958 Shares (long position) and 418,739,874 Shares (short position) were held by JPMorgan Chase & Co. directly; (b) 1,753,320 Shares (long position) were held by a trust of which JPMorgan Chase & Co. is a trustee; and (c) the remaining interests were held by JPMorgan Chase & Co. indirectly through certain of its controlled corporations. Among them, (i) 105,285,940 Shares (long position) and 105,155,304 Shares (short position) were held through physically settled listed derivatives; (ii) 6,797,279 Shares (long position) and 18,013,180 Shares (short position) were held through cash settled listed derivatives; (iii) 40,691,926 Shares (long position) and 48,319,704 Shares (short position) were held through physically settled unlisted derivatives; (iv) 87,425,503 Shares (long position) and + +90,603,044 Shares (short position) were held through cash settled unlisted derivatives; and (v) 84,973,906 Shares (long position) and 1,381,027 Shares (short position) were held through listed derivatives which are convertible instruments. + +5. According to the disclosure of interests notice filed by Citigroup Inc. regarding the relevant event dated September 11, 2025, 183,474,400 Shares (long position) and 197,906,617 Shares (short position) interests were held by Citigroup Inc. indirectly through certain of its controlled corporations. Among them, (i) 54,743,028 Shares (long position) and 36,327,420 Shares (short position) were held through physically settled listed derivatives; (ii) 77,818,263 Shares (long position) and 120,475,576 Shares (short position) were held through physically settled unlisted derivatives; (iii) 13,761,835 Shares (long position) and 37,337,978 Shares (short position) were held through cash settled unlisted derivatives; and (iv) 82,139 Shares (long position) were held through listed derivatives which are convertible instruments. +6. According to the disclosure of interests notice filed by Blackrock, Inc. regarding the relevant event dated June 13, 2025, 983,512,344 Shares (long position) and 3,297,800 Shares (short position) were held by Blackrock, Inc. indirectly through certain of its controlled corporations. Among them, (i) 6,137,756 Shares (long position) and 3,297,800 Shares (short position) were held through cash settled unlisted derivatives; and (ii) 6,572,498 Shares (long position) were held through listed derivatives which are convertible instruments. + +Save as disclosed above, as of September 30, 2025, so far as known to the Company, no person (other than our directors and chief executive officer) had an interest or short position in the Shares or underlying Shares which would fall to be disclosed to us under Divisions 2 and 3 of Part XV of the SFO, or would fall to be recorded in the register required to be kept by us pursuant to Section 336 of the SFO. + +## EQUITY INCENTIVE PLANS + +The 2014 Plan, the 2024 Plan and the 2024 Plan (Existing Shares) are the equity incentive plans of our Company in effect. For a summary of the key terms of these plans, please see “Directors, Senior Management and Employees - Equity Incentive Plans” in our fiscal year 2025 annual report. + +During the Reporting Period, a total of 63,594,318 Shares (equivalent to approximately 7,949,290 ADSs), representing approximately 0.3% of the weighted average number of Shares in issue (excluding treasury shares) of our Company, that may be issued in respect of all share-based awards were granted to eligible participants under the 2024 Plan of our Company. + +## The 2014 Plan + +The 2014 Plan (which we adopted in September 2014, amended and restated in February 2020 to reflect the Share Split and other administrative changes, and further amended and restated in May 2022 to reflect administrative changes) provides for the granting of share-based awards, including restricted share units, stock options, restricted shares and share appreciation rights, to any participant who is an employee, consultant or director of our company, our affiliates and/or certain other companies. + +No further awards will be granted under the 2014 Plan as from September 18, 2024. Any shares authorized but unissued under the 2014 Plan will no longer be available for granting. The share-based awards already granted under the 2014 Plan will remain in full force and effect pursuant to the terms and conditions of the 2014 Plan. + +Details of movements of RSUs under the 2014 Plan during the Reporting Period are as follows: + +
Name of granteeDate of grantNumber of RSUs (1), (2)Number of underlying Shares as of September 30, 2025 (2)Notes
Outstanding as of April 1, 2025Granted during the Reporting PeriodVested during the Reporting PeriodLapsed/ forfeited during the Reporting PeriodCancelled during the Reporting PeriodOutstanding as of September 30, 2025
Directors and chief executive officer of our Company
Joseph C. TSAIJuly 24, 2018 to August 16, 20191,334-1,334----Notes 3, 8
Eddie Yongming WUNovember 25, 2023160,000-40,000--120,000960,000Notes 4, 8
J. Michael EVANSJune 15, 2020 to May 20, 202310,050-4,550--5,50044,000Notes 5, 8
May 13, 2024160,00080,000-80,000640,000Notes 6, 8
Maggie Wei WUJuly 24, 2018 to May 24, 202112,801-6,766--6,03548,280Notes 3, 8
Other grantees by category
Employee participantsNovember 15, 2016 to August 18, 202444,402,154-16,110,6831,681,3791,82526,608,267212,866,136Notes 7, 9
Related entity $participants^{(11)}$ July 24, 2018 to June 8, 2022113,139-84,7672,578-25,794206,352Notes 7, 10
Total44,859,478-16,328,1001,683,9571,82526,845,596214,764,768
+ +## Notes: + +1. The purchase price of all RSUs granted is nil. +2. Eight Shares are issuable upon the vesting of each RSU. +3. The RSUs vest over a period of 6 years. +4. The RSUs vest over a period of 5 years. +5. The RSUs vest over a period of 4 years. +6. The RSUs vest over a period of 2 years. + +7. The RSUs granted to employee participants and related entity participants vest over a period of up to 10 years. +8. For directors and chief executive officer, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the RSUs were vested during the Reporting Period was US\$136.71. +9. For employee participants, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the RSUs were vested during the Reporting Period was US\$131.05 +10. For related entity participants, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the RSUs were vested during the Reporting Period was US\$132.15. +11. A related entity participant is an employee of an associated company in which our Company owns, directly or indirectly, securities or interests representing 20% or more of its voting power. + +Details of movements of share options under the 2014 Plan during the Reporting Period are as follows: + +
Name of granteeDate of grantExercise price (US$)Number of share options (1)Number of underlying Shares as of September 30, 2025 (1)Notes
Outstanding as of April 1, 2025Granted during the Reporting PeriodExercised during the Reporting PeriodLapsed/ forfeited during the Reporting Period (9)Outstanding as of September 30, 2025
Directors and chief executive officer of our Company
Eddie Yongming WUNovember 25, 202378.372,000,000---2,000,00016,000,000Note 2
J. Michael EVANSJuly 31, 201579.962,000,000-1,000,000-1,000,0008,000,000Note 3, 6
May 13, 202484.60150,000---150,0001,200,000Note 4
Other grantees by category
Employee participantsMay 10, 2015 to March 12, 202223.00 to 182.482,385,667-842,6671,000,000543,0004,344,000Notes 5, 7
Related entity participant (10)March 12, 202223.00100,000-100,000---Notes 5, 8
Total6,635,667-1,942,6671,000,0003,693,00029,544,000
+ +## Notes: + +1. Eight Shares are issuable upon the exercise of each share option. +2. The share options vest over a period of 5 years and are exercisable within 10 years after date of grant. +3. The share options vest over a period of 6 years and are exercisable within 12 years after date of grant. +4. The share options vest over a period of 2 years and are exercisable within 8 years after date of grant. +5. The share options vest over a period of up to 9 years and are exercisable within up to 12 years after date of grant. +6. The closing price of ADSs traded on the NYSE immediately before the date on which the share options were exercised during the Reporting Period was US\$123.90. +7. For employee participants, the weighted average closing price of ADSs traded on the NYSE immediately before the dates on which the share options were exercised during the Reporting Period was US\$114.78. +8. For related entity participant, the weighted average closing price of ADSs traded on the NYSE immediately before the date on which the share options were exercised during the Reporting Period was US\$118.09. +9. No share options granted were cancelled during the Reporting Period. +10. A related entity participant is an employee of an associated company in which our Company owns, directly or indirectly, securities or interests representing 20% or more of its voting power. + +## The 2024 Plan + +The 2024 Plan was approved at our annual general meeting of shareholders held in August 2024. The purpose of the 2024 Plan is to provide incentives to attract, motivate and retain the grantees and align the interests of the grantees with those of our shareholders. It provides for the granting of restricted share units, stock options, restricted shares and share appreciation rights to any participant who is an employee, an executive director or a service provider. + +As of April 1, 2025 and September 30, 2025, the number of share-based awards available for grant under the scheme mandate was 473,488,465 Shares (equivalent to approximately 59,186,058 ADSs) and 411,246,946 Shares (equivalent to approximately 51,405,868 ADSs), respectively. + +As of April 1, 2025 and September 30, 2025, the number of share-based awards available for grant under the service provider sub-limit was 93,716,369 Shares (equivalent to 11,714,546 ADSs). + +Details of movements of RSUs under the 2024 Plan during the Reporting Period are as follows: + +
Name (3)Date of grantNumber of RSUs (1),(2)Number of underlying Shares as of September 30, 2025 (2)Notes
Outstanding as of April 1, 2025Granted during the Reporting PeriodVested during the Reporting PeriodLapsed/ forfeited during the Reporting Period (8)Outstanding as of September 30, 2025
Employee participantsNovember 19, 2024 and February 24, 20259,435,535-1,456,534236,2007,742,8017,742,801Notes 4, 5
May 23, 2025, June 13, 2025 and September 4, 2025-63,594,3183,209,7961,116,59959,267,92359,267,923Notes 5, 6, 7
Total9,435,53563,594,3184,666,3301,352,79967,010,72467,010,724
+ +Notes: +1. The purchase price of all RSUs granted is nil. +2. One Share is issuable upon the vesting of each RSU. +3. No RSUs were granted under the 2024 Plan to directors during the Reporting Period. As at September 30, 2025, no directors were holding any outstanding awards under the 2024 Plan. +4. The RSUs granted to employee participants vest over a period of up to 6 years. +5. For employee participants, the weighted average closing price of the Shares traded on the Hong Kong Stock Exchange immediately before the dates on which the RSUs were vested during the Reporting Period was HK\$113.70. +6. The following grants of RSUs were made to employee participants during the Reporting Period. These RSUs vest over up to 6 years without any performance target, except that certain RSUs granted to selected participants on May 23, 2025 are subject to satisfaction of certain performance targets based on the financial or operational indicators of the relevant segments as determined by the administrator of the 2024 Plan. + +
Date of grantNumber of RSUs grantedNumber of underlying SharesClosing price of sharesimmediately before date of grant(HK$)Fair value of each RSU as of dateof grant (HK$)
May 23, 202553,429,59753,429,597119.10118.78
June 13, 20257,687,0917,687,091114.60114.60
September 4, 20252,477,6302,477,630134.10130.31
63,594,31863,594,318
+ +7. In accordance with the accounting standards and policies adopted for preparing our Company's financial statements, the fair values of RSUs granted during the Reporting Period were determined with reference to the fair values of the underlying Shares on grant date. +8. No RSUs granted were cancelled during the Reporting Period. + +## The 2024 Plan (Existing Shares) + +The 2024 Plan (Existing Shares) was approved by our board in August 2024. The purpose of the 2024 Plan (Existing Shares) is to provide incentives to attract, motivate and retain the grantees and align the interests of the grantees with those of our shareholders. It provides for the granting of restricted share units, stock options, restricted shares and share appreciation rights to any participant who is an employee, a director or a service provider. The 2024 Plan (Existing Shares) is funded by existing Shares, therefore it does not constitute a scheme involving the new issuance of Shares as referred to in Chapter 17 of the Hong Kong Listing Rules but is subject to the applicable disclosure requirements under Rule 17.12 of the Hong Kong Listing Rules. + +## PURCHASE, SALE OR REDEMPTION OF OUR COMPANY'S LISTED SECURITIES + +During the Reporting Period, our Company repurchased a total of 73 million Shares on the NYSE for an aggregate consideration of US\$1.0 billion. + +Details of the Shares repurchased on the NYSE are as follows: + +
Month of repurchaseNumber of Shares underlying ADSs repurchased (1)Highest price paid (US$)Lowest price paid (US$)Aggregate consideration paid (US$, in millions)
April 202530,630,48016.7611.97427
May 202513,559,32016.7914.02210
June 202511,685,19215.0013.92168
July 202511,019,16015.0012.96154
August 20255,839,33615.0014.5287
Total72,733,4881,046
+ +Note: + +1. Each ADS represents eight Shares. + +As of the date of this Interim Report, all the Shares repurchased during the Reporting Period have been cancelled. + +Save as disclosed above, neither our Company nor any of its subsidiaries purchased, sold or redeemed any of our Company's securities listed on the Hong Kong Stock Exchange or the NYSE (including sale of treasury shares) during the Reporting Period. As of September 30, 2025, our Company did not hold any treasury shares as defined in the Hong Kong Listing Rules. + +## OTHER INFORMATION + +## Corporate Governance + +## Compliance with the Corporate Governance Code + +To the knowledge of the Company and our directors, we have complied with all applicable code provisions set out in the Corporate Governance Code as set forth in Part 2 of Appendix C1 to the Hong Kong Listing Rules for the six months ended September 30, 2025. + +## Compliance with the Model Code + +We have adopted our own trading guidelines, on terms no less exacting than the required standard set out in the Model Code as set forth in Appendix C3 to the Hong Kong Listing Rules, to regulate, among others, all dealings by directors and relevant employees of securities in the Company. + +Having made specific enquiry of all directors, all directors confirmed that they have complied with our trading guidelines during the six months ended September 30, 2025. + +## Audit Committee Review + +The audit committee oversees our accounting and financial reporting processes and the audits of our financial statements. Our audit committee is responsible for, among other things: + +- selecting, and evaluating the qualifications, performance and independence of the independent auditor; +- pre-approving or, as permitted, approving auditing and non-auditing services permitted to be performed by the independent auditor; +• assessing the adequacy of our internal accounting controls and audit procedures; +- reviewing with the independent auditor any audit problems or difficulties and management's response; +- reviewing and approving related party transactions (as defined in Form 20-F) and connected transactions (as required under the Hong Kong Listing Rules); +- reviewing and discussing the quarterly reports, half-year reports and annual reports with management and the independent auditor; +- establishing procedures for the receipt, retention and treatment of complaints received from our employees regarding accounting, internal accounting controls or auditing matters and the confidential, anonymous submission by our employees of concerns regarding questionable accounting or auditing matters; +- meeting separately, periodically, with management, internal auditors and the independent auditor; and +• reporting regularly to the full board of directors. + +Our audit committee has reviewed our unaudited condensed consolidated interim financial statements for the six months ended September 30, 2025, and has met with the independent auditor of the Company, + +PricewaterhouseCoopers. Our audit committee has also discussed the accounting policies and practices adopted by us, as well as internal control and financial reporting matters. + +Our unaudited condensed consolidated interim financial statements for the six months ended September 30, 2025 have been reviewed by the independent auditor of the Company, PricewaterhouseCoopers, in accordance with International Standard on Review Engagements 2410 “Review of Interim Financial Information Performed by the Independent Auditor of the Entity”. + +## Interim Dividend + +Our board did not recommend the distribution of an interim dividend for the Reporting Period. + +## Use of Proceeds from the Notes Offerings + +## September 2025 Notes Offering + +In September 2025, we completed a private offering (the “September 2025 Notes Offering”) of US\$3.168 billion aggregate principal amount of Zero Coupon Convertible Senior Notes due 2032 (the “September 2025 Notes”). In connection with the September 2025 Notes Offering, we also entered into capped call transactions (the “Capped Call Transactions”) with certain financial institutions at the costs of US\$184 million. The cap price of the Capped Call Transactions is initially US\$235.46 per ADS (subject to certain adjustments), which represents a premium of 60% over the closing price of the ordinary shares on the Hong Kong Stock Exchange on the date of pricing of the Notes Offering of HK\$143.30, which is converted and multiplied by eight, the ordinary-share-to-ADS ratio to, US\$147.16 (the “Reference Share Price”). The September 2025 Notes have been offered to certain non-U.S. persons in offshore transaction in reliance on Regulation S under the U.S. Securities Act. + +The Company intended to use the net proceeds from the September 2025 Notes Offering for general corporate purposes, with a strategic focus on strengthening its cloud infrastructure capabilities and international commerce business operations. Specifically, the allocation will include approximately 80% directed towards enhancing its cloud infrastructure, which encompasses scaling up data centers, upgrading technology, and improving services to meet growing demand. The remaining 20% will be invested in expanding international commerce operations, focusing on operational investments that will enable the Company to enhance its market presence and efficiency. + +The September 2025 Notes will mature on September 15, 2032, unless earlier redeemed, repurchased or converted in accordance with their terms prior to such date. Prior to the close of business on the business day immediately preceding March 15, 2032, holders will have the right to convert all or any portion of their Notes, in multiples of US\$1,000 principal amount, only upon satisfaction of certain conditions. On or after March 15, 2032, until the close of business on the third scheduled trading day immediately preceding the maturity date, holders may convert all or any portion of their Notes, in multiples of US\$1,000 principal amount, at their option at any time irrespective of the foregoing conditions. Upon conversion, we retain flexibility to control any potential dilution with full discretion to settle the conversions in cash or ADSs, or a combination of both. Holders may also elect to receive ordinary shares in lieu of any ADSs upon conversion. + +The initial conversion rate for the Notes is 5.1773 ADSs per US\$1,000 principal amount of the September 2025 Notes, which is equivalent to an initial conversion price of approximately US\$193.15 per ADS (or HK\$188.08 per ordinary share, as converted and divided by eight, the ordinary-share-to-ADS ratio). The initial conversion price represents a premium of approximately 31.25% over the Reference Share Price, and the conversion rate will be subject to adjustment in some events, including but not limited to a share split, a share combination, cash distributions, certain issuances of rights, options or warrants, distribution of capital stock, certain tender or exchange offers, and certain other dilutive events. + +The net price of each conversion share, based on the estimated net proceeds of approximately US\$3.13 billion and 16,401,686 conversion shares (in the form of ADSs) that are issuable under the September 2025 Notes (excluding adjustment and make-whole adjustments), is expected to be approximately US\$190.60. Upon exercise in full of the conversion rights attached to the September 2025 Notes at the initial conversion price of US\$193.15 per ADS (equivalent to HK\$188.08 per ordinary share, as converted and divided by eight, the ordinary-share-to-ADS ratio), a total of approximately 16,401,686 ADSs (or 131,213,488 ordinary shares) will be issued. + +For further details of the September 2025 Notes Offering, please see the announcements of our Company dated September 11, 2025 and September 17, 2025. + +The net proceeds raised from the September 2025 Notes Offering were used according to the intentions as disclosed in the announcement of our Company dated September 11, 2025. As of September 30, 2025, the unutilized proceeds amounted to US\$2,400 million. A summary of the utilization of the net proceeds from the September 2025 Notes Offering as of September 30, 2025 is as follows: + +
General Corporate purposesNet proceeds from the September 2025 Notes Offering (US$ in millions)Amount of net proceeds utilized during the reporting period (US$ in millions)Actual use of proceeds up to September 30, 2025 (US$ in millions)
Enhancing its cloud infrastructure2,504443443
Expanding international commerce operations626287287
Total3,130730730
+ +## DEFINITIONS + +## Conventions that apply to this Interim Report + +Unless the context otherwise requires, references in this interim report to: + +- “2014 Plan” is to the 2014 Post-IPO Equity Incentive Plan, which we adopted on September 2, 2014 and amended and restated on February 12, 2020 and May 25, 2022; +- “2024 Plan” is to the 2024 Equity Incentive Plan, which we adopted on August 22, 2024; +- “2024 Plan (Existing Shares)” is to the 2024 Equity Incentive Plan (Existing Shares), which we adopted on August 26, 2024; +- “ADS(s)” are to the American depositary share(s), each of which represents eight Shares; +- “Alibaba,” “Alibaba Group,” “the Group,” “Company,” “our company,” “we,” “our” or “us” are to Alibaba Group Holding Limited, a company incorporated in the Cayman Islands with limited liability on June 28, 1999 and, where the context requires, its consolidated subsidiaries and its affiliated consolidated entities, including its variable interest entities and their subsidiaries, from time to time; +- “Articles” or “Articles of Association” is to our Amended and Restated Articles of Association (as amended and restated from time to time), adopted on September 30, 2020 and amended on August 22, 2024; +- “board” or “board of directors” is to our board of directors, unless otherwise stated; +- “Corporate Governance Code” is to the Corporate Governance Code set out in Appendix C1 to the Hong Kong Listing Rules; +- “director(s)” are to member(s) of our board, unless otherwise stated; +- “Hong Kong Listing Rules” are to the Rules Governing the Listing of Securities on The Stock Exchange of Hong Kong Limited, as amended or supplemented from time to time; +- “Hong Kong Stock Exchange” is to The Stock Exchange of Hong Kong Limited; +- “Model Code” is to the Model Code for Securities Transactions by Directors of Listed Issuers set out in Appendix C3 to the Hong Kong Listing Rules; +- “NYSE” is to the New York Stock Exchange; +- “Reporting Period” is to the six months ended September 30, 2025; +- “RSU(s)” are to restricted share unit(s); +- “SEC” is to the United States Securities and Exchange Commission; +- “SFO” is to the Securities and Futures Ordinance (Chapter 571 of the Laws of Hong Kong), as amended or supplemented from time to time; + +- “Share Split” is to the subdivision of each ordinary share into eight Shares, pursuant to which the par value of our Shares was correspondingly changed from US\$0.000025 per Share to US\$0.000003125 per Share, with effect from July 30, 2019; immediately after the Share Split became effective, our authorized share capital became US\$100,000 divided into 32,000,000,000 Shares of par value US\$0.000003125 per Share; +- “shareholder(s)” are to holder(s) of Shares and, where the context requires, ADSs; +- “Share(s)” or “ordinary share(s)” are to ordinary share(s) in our capital with par value of US\$0.000003125 each; +- “substantial shareholder(s)” has the meaning ascribed to it under the Hong Kong Listing Rules; +- “U.S.” or “United States” is to the United States of America, its territories, its possessions and all areas subject to its jurisdiction; +- “US\$” or “U.S. dollars” are to the lawful currency of the United States; +- “U.S. Exchange Act” is to the United States Securities Exchange Act of 1934, as amended, and the rules and regulations promulgated thereunder; +- “U.S. Securities Act” is to the United States Securities Act of 1933, as amended, and the rules and regulations promulgated thereunder. + +# ALIBABA GROUP HOLDING LIMITED + +# INDEX TO CONDENSED CONSOLIDATED INTERIM FINANCIAL STATEMENTS + +
Page
Report on Review of Interim Financial Information for the six-months ended September 30, 202545
Unaudited Condensed Consolidated Income Statements for the six months ended September 30, 2024 and 202546
Unaudited Condensed Consolidated Statements of Comprehensive Income for the six months ended September 30, 2024 and 202547
Unaudited Condensed Consolidated Balance Sheets as of March 31, 2025 and September 30, 202548
Unaudited Condensed Consolidated Statements of Changes in Shareholders’ Equity for the six months ended September 30, 2024 and 202550
Unaudited Condensed Consolidated Statements of Cash Flows for the six months ended September 30, 2024 and 202552
Notes to Unaudited Condensed Consolidated Financial Statements53
+ +# Report on Review of Interim Financial Information To the Board of Directors of Alibaba Group Holding Limited (incorporated in the Cayman Islands with limited liability) + +## Introduction + +We have reviewed the interim financial information set out on pages 46 to 81, which comprises the condensed consolidated balance sheet of Alibaba Group Holding Limited (the “Company”) and its subsidiaries (together, the “Group”) as at September 30, 2025 and the condensed consolidated income statement, the condensed consolidated statement of comprehensive income, the condensed consolidated statement of changes in shareholders’ equity and the condensed consolidated statement of cash flows for the six-month period then ended, and selected explanatory notes. The Rules Governing the Listing of Securities on The Stock Exchange of Hong Kong Limited require the preparation of a report on interim financial information to be in compliance with the relevant provisions thereof and accounting principles generally accepted in the United States of America (“U.S. GAAP”). The directors of the Company are responsible for the preparation and presentation of this interim financial information in accordance with U.S. GAAP. Our responsibility is to express a conclusion on this interim financial information based on our review and to report our conclusion solely to you, as a body, in accordance with our agreed terms of engagement, and for no other purpose. We do not assume responsibility towards or accept liability to any other person for the contents of this report. + +## Scope of Review + +We conducted our review in accordance with International Standard on Review Engagements 2410, “Review of Interim Financial Information Performed by the Independent Auditor of the Entity”. A review of interim financial information consists of making inquiries, primarily of persons responsible for financial and accounting matters, and applying analytical and other review procedures. A review is substantially less in scope than an audit conducted in accordance with International Standards on Auditing and consequently does not enable us to obtain assurance that we would become aware of all significant matters that might be identified in an audit. Accordingly, we do not express an audit opinion. + +## Conclusion + +Based on our review, nothing has come to our attention that causes us to believe that the interim financial information of the Group is not prepared, in all material respects, in accordance with U.S. GAAP. + +## PricewaterhouseCoopers Certified Public Accountants + +Hong Kong, December 4, 2025 + +ALIBABA GROUP HOLDING LIMITED +UNAUDITED CONDENSED CONSOLIDATED INCOME STATEMENTS + +
Six months ended September 30,
20242025
RMBRMBUS$ (Note 2(a))
(in millions, except per share data)
Notes
Revenue4,16479,739495,44769,595
Cost of revenue16(290,135)(287,210)(40,344)
Product development expenses16(27,555)(32,096)(4,509)
Sales and marketing expenses16(65,167)(119,674)(16,811)
General and administrative expenses16(23,057)(14,778)(2,076)
Amortization and impairment of intangible assets(3,441)(1,633)(229)
Other gains, net85129742
Income from operations71,23540,3535,668
Interest and investment income, net17,12937,4685,263
Interest expense(4,615)(4,995)(702)
Other (expense) income, net16(1,221)1,329187
Income before income tax and share of results of equity method investees82,52874,15510,416
Income tax expenses5(17,442)(14,415)(2,024)
Share of results of equity method investees2,4833,254457
Net income67,56962,9948,849
Net loss (income) attributable to noncontrolling interests854(1,326)(187)
Net income attributable to Alibaba Group Holding Limited68,42361,6688,662
(Accretion) Reversal of accretion of mezzanine equity(280)2,438343
Net income attributable to ordinary shareholders68,14364,1069,005
Earnings per share attributable to ordinary shareholders7
Basic3.583.450.49
Diluted3.503.340.47
Earnings per ADS attributable to ordinary shareholders (one ADS equals eight ordinary shares)7
Basic28.6227.633.88
Diluted28.0026.733.75
Weighted average number of shares used in computing earnings per share (million shares)7
Basic19,04518,562
Diluted19,45919,154
+ +The accompanying notes form an integral part of these unaudited condensed consolidated financial statements. + +ALIBABA GROUP HOLDING LIMITED +UNAUDITED CONDENSED CONSOLIDATED STATEMENTS OF COMPREHENSIVE INCOME + +
Six months ended September 30,
20242025
RMBRMBUS$ (Note 2(a))
(in millions)
Net income67,56962,9948,849
Other comprehensive income (loss):
- Foreign currency translation:
Change in unrealized losses, net of tax(3,699)(5,081)(714)
- Share of other comprehensive income (loss) of equity method investees:
Change in unrealized gains (losses)350(40)(6)
- Interest rate swaps under hedge accounting and others:
Change in unrealized gains601
Other comprehensive loss(3,289)(5,120)(720)
Total comprehensive income64,28057,8748,129
Total comprehensive loss (income) attributable to noncontrolling interests1,119(273)(38)
Total comprehensive income attributable to ordinary shareholders65,39957,6018,091
+ +The accompanying notes form an integral part of these unaudited condensed consolidated financial statements. + +ALIBABA GROUP HOLDING LIMITED +UNAUDITED CONDENSED CONSOLIDATED BALANCE SHEETS + +
As of March 31,As of September 30,
20252025
RMBRMBUS$
(in millions)(Note 2(a))
AssetsNotes
Current assets:
Cash and cash equivalents145,487135,06918,973
Short-term investments228,826193,24627,145
Restricted cash and escrow receivables43,78140,3745,671
Equity securities and other investments853,78045,2576,357
Prepayments, receivables and other assets202,175232,67332,684
Total current assets674,049646,61990,830
Equity securities and other investments8356,818411,95357,867
Prepayments, receivables and other assets83,43196,92713,615
Investments in equity method investees210,169206,86229,058
Property and equipment, net203,348246,53934,631
Intangible assets, net1020,91119,4292,729
Goodwill11255,501255,55135,897
Total assets1,804,2271,883,880264,627
Liabilities, mezzanine equity and shareholders' equity
Current liabilities:
Current bank borrowings22,56226,2883,693
Income tax payable11,6385,588785
Accrued expenses, accounts payable and other liabilities12332,537340,76947,868
Merchant deposits27425135
Deferred revenue and customer advances68,33571,24110,007
Total current liabilities435,346444,13762,388
Deferred revenue4,5364,496632
Deferred tax liabilities48,45446,8026,574
Non-current bank borrowings49,90963,5668,929
Non-current unsecured senior notes122,398120,50416,927
Non-current convertible unsecured senior notes1335,83457,4818,074
Non-current exchangeable bonds1413,7551,932
Other liabilities1217,64421,3543,000
Total liabilities714,121772,095108,456
+ +The accompanying notes form an integral part of these unaudited condensed consolidated financial statements. + +ALIBABA GROUP HOLDING LIMITED +UNAUDITED CONDENSED CONSOLIDATED BALANCE SHEETS (CONTINUED) + +
As of March 31,As of September 30,
20252025
RMBRMBUS$ (Note 2(a))
(in millions)
Commitments and contingencies
Mezzanine equity11,7139,8841,388
Shareholders’ equity:
Ordinary shares, US$0.000003125 par value; 32,000,000,000 shares authorized as of March 31 and September 30, 2025; 18,474,235,708 and 18,551,943,896 shares issued and outstanding as of March 31 and September 30, 2025 respectively11
Additional paid-in capital381,379387,14754,382
Treasury shares, at cost(36,329)(36,162)(5,080)
Statutory reserves15,93616,2862,288
Accumulated other comprehensive income (loss)
Cumulative translation adjustments3,286(1,670)(234)
Unrealized gains on interest rate swaps and others10710915
Retained earnings645,478666,78493,663
Total shareholders’ equity1,009,8581,032,495145,034
Noncontrolling interests68,53569,4069,749
Total equity1,078,3931,101,901154,783
Total liabilities, mezzanine equity and equity1,804,2271,883,880264,627
+ +The accompanying notes form an integral part of these unaudited condensed consolidated financial statements. + +ALIBABA GROUP HOLDING LIMITED +UNAUDITED CONDENSED CONSOLIDATED STATEMENTS OF CHANGES IN SHAREHOLDERS' EQUITY + +
Ordinary sharesAccumulated othercomprehensive income (loss)
ShareAmountAdditionalpaid-incapitalRMBTreasurysharesRMBStatutoryreservesRMBCumulativetranslationadjustmentsRMBUnrealizedgains(losses) oninterestrate swapsand othersRMBRetainedearningsRMBTotalshareholders’equityRMBNoncontrollinginterestsRMBTotalequityRMB
(in millions, except per share data)
Balance as ofApril 1, 202419,469,126,9561397,999(27,684)14,7333,635(37)597,897986,544115,3271,101,871
Foreign currencytranslationadjustment,net of tax(3,541)(3,541)(158)(3,699)
Share of additionalpaid-in capitaland othercomprehensiveincome of equitymethod investees(42)33515308308
Change in fair value ofinterest rate swapsunder hedgeaccountingand others606060
Net income for theperiod68,42368,423(961)67,462
Acquisition ofsubsidiaries187187
Issuance of shares,including vesting ofRSUs and earlyexercised optionsand exercise ofshare options178,297,456
Repurchase andretirement ofordinary shares(1,027,554,280)(21,096)(8,501)(42,216)(71,813)(71,813)
Transactions withnoncontrollinginterests2,3572,357(23,852)(21,495)
Amortization ofcompensation cost5,8195,8191,9627,781
Declaration ofdividends(29,340)(29,340)(29,340)
Capped calltransactions(4,612)(4,612)(4,612)
Appropriation tostatutory reserves1,152(1,152)
Others(280)(280)(144)(424)
Balance as ofSeptember 30, 2024(in millions, except per share data)18,619,870,1321380,145(36,185)15,88542938593,612953,92592,3611,046,286
Balance as ofApril 1, 202518,474,235,7081381,379(36,329)15,9363,286107645,4781,009,85868,5351,078,393
Foreign currencytranslationadjustment,net of tax(3,643)7(3,636)(166)(3,802)
Share of additionalpaid-in capitaland othercomprehensiveincome of equitymethod investees(129)(34)(6)(169)(169)
Change in fair value ofinterest rate swapsunder hedgeaccounting and others111
Net investment hedge(1,279)(1,279)(1,279)
Net income for theperiod61,66861,66843962,107
Acquisition ofsubsidiaries22
Deconsolidation ofsubsidiaries(80)(80)
Issuance of shares,including vesting ofRSUs and earlyexercised optionsand exercise ofshare options150,441,6761,0381,0381,038
Repurchase andretirement ofordinary shares(72,733,488)(1,737)167(5,965)(7,535)(7,535)
Transactions withnoncontrollinginterests6868(315)(247)
Amortization ofcompensation cost5,3975,3971,0756,472
Declaration ofdividends(34,047)(34,047)(34,047)
Capped calltransactions(1,309)(1,309)(1,309)
Appropriation tostatutory reserves350(350)
Others2,4402,440(84)2,356
Balance as ofSeptember 30, 202518,551,943,8961387,147(36,162)16,286(1,670)109666,7841,032,49569,4061,101,901
+ +The accompanying notes form an integral part of these unaudited condensed consolidated financial statements. + +## ALIBABA GROUP HOLDING LIMITED + +## UNAUDITED CONDENSED CONSOLIDATED STATEMENTS OF CASH FLOWS + +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)(Note 2(a))
Cash flows from operating activities:65,07430,7714,322
Cash flows from investing activities:
Decrease in short-term investments, net105,47012,9721,822
Increase in other treasury investments, net(113,387)(24,091)(3,384)
Settlement of forward exchange contracts, net311(111)(16)
Acquisitions of equity and debt securities, and others(3,905)(18,690)(2,625)
Disposals of equity and debt securities, and others5,77235,3904,971
Acquisitions of equity method investees(1,964)(1,179)(165)
Disposals of and distributions from equity method investees7372,426341
Acquisitions of land use rights, property and equipment(29,585)(70,177)(9,858)
Disposals of property and equipment1,47523934
Cash received (paid) for business combinations, net of cash acquired62(62)(9)
Deconsolidation and disposal of subsidiaries, net of cash proceeds11,8251,661
Loans to employees, net of repayments14913419
Net cash used in investing activities(34,865)(51,324)(7,209)
Cash flows from financing activities:
Issuance of ordinary shares11,038146
Repurchase of ordinary shares(72,889)(7,638)(1,073)
Dividend distribution(29,022)(33,621)(4,723)
Proceeds from convertible unsecured senior notes, net of debt issuance cost35,67722,3033,133
Payments for capped call transactions(4,612)(1,309)(184)
Acquisition of additional equity interests in non-wholly owned subsidiaries(19,947)(1,155)(162)
Dividends paid by non-wholly owned subsidiaries to noncontrolling interests(116)(306)(43)
Contingent consideration payments made after a business combination(149)(81)(11)
Capital injection from noncontrolling interests2,63616824
Proceeds from exchangeable bonds, net of debt issuance cost10,9861,543
Proceeds from bank borrowings and other borrowings, net of upfront fee payment for a syndicated loan22,09837,8055,310
Repayment of bank borrowings(20,041)(20,019)(2,812)
Net cash (used in) provided by financing activities(86,364)8,1711,148
Effect of exchange rate changes on cash and cash equivalents, restricted cash and escrow receivables(1,797)(1,443)(203)
Decrease in cash and cash equivalents, restricted cash and escrow receivables(57,952)(13,825)(1,942)
Cash and cash equivalents, restricted cash and escrow receivables at beginning of period286,424189,26826,586
Cash and cash equivalents, restricted cash and escrow receivables at end of period228,472175,44324,644
+ +The accompanying notes form an integral part of these unaudited condensed consolidated financial statements. + +# ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 1. Organization and principal activities + +Alibaba Group Holding Limited (the “Company”) is a limited liability company, which was incorporated in the Cayman Islands on June 28, 1999. The Company is a holding company and conducts its businesses primarily through its subsidiaries and variable interest entities. In these unaudited condensed consolidated financial statements, where appropriate, the term “Company” also refers to its subsidiaries as a whole. The Company provides the technology infrastructure and marketing reach to help merchants, brands, retailers and other businesses to leverage the power of new technology to engage with their users and customers and operate in a more efficient way. + +The Company's businesses comprise Alibaba China E-commerce Group, Alibaba International Digital Commerce Group, Cloud Intelligence Group and All others. An ecosystem has developed around the Company's platforms and businesses that consists of consumers, merchants, brands, retailers, enterprises, third-party service providers, strategic alliance partners and other businesses. + +The Company's American depositary shares (“ADSs”) are listed on the New York Stock Exchange (“NYSE”) under the symbol “BABA” and the Company’s ordinary shares are listed on the Hong Kong Stock Exchange (“HKSE”) under the stock codes “9988 (HKD Counter)” and “89988 (RMB Counter).” + +## 2. Summary of significant accounting policies + +## (a) Basis of presentation + +The accompanying unaudited condensed consolidated financial statements are prepared in accordance with accounting principles generally accepted in the United States of America (“U.S. GAAP”) for interim financial reporting and the disclosure requirements of the Rules Governing the Listing of Securities on The Stock Exchange of Hong Kong Limited, as amended or supplemented from time to time (“Hong Kong Listing Rules”). Accordingly, they do not include all of the information and footnote disclosures required by U.S. GAAP for a complete set of financial statements. These unaudited condensed consolidated financial statements include all adjustments of a normal recurring nature necessary to a fair statement of the results for the interim periods presented. Results of operations for an interim period are not necessarily indicative of results for the entire year. + +These unaudited condensed consolidated financial statements should be read in conjunction with the audited consolidated financial statements and notes thereto for the year ended March 31, 2025. There were no significant changes to our significant accounting policies from the audited consolidated financial statements for the preceding fiscal year, except that the accounting policies relating to derivatives and hedging (Note 2(d)) and borrowings (Note 2(e)) were updated as a result of the issuance of exchangeable bonds. + +Translations of balances in the unaudited condensed consolidated balance sheet, unaudited condensed consolidated income statements, unaudited condensed consolidated statement of comprehensive income and unaudited condensed consolidated statement of cash flows from RMB into the United States Dollar (“US\$”) as of and for the six months ended September 30, 2025 are solely for the convenience of the readers and are calculated at the rate of US\$1.00=RMB7.1190, representing the exchange rate set forth in the H.10 statistical release of the Federal Reserve Board on September 30, 2025. No representation is made that the RMB amounts could have been, or could be, converted, realized or settled into US\$ at this rate, or at any other rate. + +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 2. Summary of significant accounting policies (Continued) + +## (b) Use of estimates + +The preparation of the unaudited condensed consolidated financial statements in conformity with U.S. GAAP requires the Company to make estimates and assumptions that affect the reported amounts of assets and liabilities, disclosure of contingent assets and liabilities as of the date of the unaudited condensed consolidated financial statements and the reported amounts of revenue and expenses during the reporting period. The Company bases its estimates on historical experience and on various other assumptions that are believed to be reasonable, the results of which form the basis for making judgments about the carrying values of assets and liabilities. + +## (c) Consolidation + +The unaudited condensed consolidated financial statements include the financial statements of the Company and its subsidiaries, which include the PRC-registered entities directly or indirectly owned by the Company (“WFOEs”) and variable interest entities (“VIEs”) over which the Company is the primary beneficiary for accounting purposes only. All transactions and balances among the Company, its subsidiaries and the VIEs have been eliminated upon consolidation. The results of subsidiaries acquired or disposed of are recorded in the unaudited condensed consolidated income statements from the effective date of acquisition or up to the effective date of disposal, as appropriate. The nature of the businesses and activities of the consolidated VIEs have not changed materially from the preceding fiscal year. + +The following financial information of the consolidated VIEs and their subsidiaries was recorded in the accompanying unaudited condensed consolidated financial statements: + +
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
Cash and cash equivalents and short-term investments10,6218,807
Investments in equity method investees and equity securities and other investments37,11735,335
Accounts receivable and contract assets, net of allowance18,40818,666
Amounts due from non-VIE subsidiaries of the Company53,79251,309
Property and equipment, net and intangible assets, net19,91137,527
Others40,42956,590
Total assets180,278208,234
Amounts due to non-VIE subsidiaries of the Company113,332131,635
Accrued expenses, accounts payable and other liabilities50,52160,882
Deferred revenue and customer advances19,12618,669
Total liabilities182,979211,186
Six months ended September 30,
20242025
RMBRMB
(in millions)
Revenue (i)67,73977,466
Net income2,1032,442
Net cash provided by operating activities12,1182,817
Net cash used in investing activities(28,993)(31,914)
Net cash provided by financing activities8,62527,034
+ +(i) Revenue generated by the VIEs are primarily from cloud services, digital media and entertainment services and others. + +# ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 2. Summary of significant accounting policies (Continued) + +## (d) Derivatives and hedging + +Contracts that meet the definition of a derivative and an embedded derivative that is not closely related to the host contract are generally recognized on the unaudited condensed consolidated balance sheets as either assets or liabilities and recorded at fair value. Changes in the fair value of derivative assets or liabilities are either recognized periodically in the unaudited condensed consolidated income statements or in other comprehensive income depending on the use of the derivatives and whether they qualify for hedge accounting and are so designated as cash flow hedges, fair value hedges or net investment hedges. The capped call transactions in connection with the issuance of the convertible unsecured senior notes meet the scope exception for contracts in an entity’s own equity provided in ASC 815 “Derivatives and Hedging” and are recognized in shareholders’ equity. Other instruments involving an entity’s own equity that do not meet the scope exception for contracts in an entity’s own equity provided in ASC 815 are classified as assets or liabilities and recorded at fair value. + +To qualify for hedge accounting, the hedge relationship is designated and formally documented at inception, detailing the particular risk management objective and strategy for the hedge (which includes the item and risk that is being hedged), the hedging instrument that is being used and how hedge effectiveness is being assessed. A hedging instrument has to be effective in accomplishing the objective of offsetting changes in the risk being hedged. The effectiveness of the hedging relationship is evaluated on a prospective and retrospective basis using qualitative and quantitative measures of correlation. Qualitative methods may include comparison of critical terms of the hedging instrument to those of the hedged item due to the hedged risk. + +Quantitative methods include a comparison of the changes in the value or discounted cash flow of the hedging instrument to those of the hedged item due to the hedged risk. A hedging relationship is considered effective if the results of the hedging instrument are within a ratio of $80\%$ to $125\%$ of the results of the hedged item. + +## Cash flow hedges + +Interest rate swaps designated as hedging instruments to hedge against the cash flows attributable to recognized assets or liabilities or forecasted payments may qualify as cash flow hedges. The Company entered into interest rate swap contracts to swap floating interest payments related to certain borrowings for fixed interest payments to hedge the interest rate risk associated with certain forecasted payments and obligations. All changes in the fair value of interest rate swaps that are designated and qualify as cash flow hedges are recognized in accumulated other comprehensive income. Amounts in accumulated other comprehensive income are reclassified into earnings in the same period during which the hedged forecasted transaction affects earnings. The Company has elected the optional expedients under ASC 848 “Reference Rate Reform” for certain existing interest rate swaps that are designated as cash flow hedges in the hedging relationship designation and the assessment of probability of forecasted transaction and hedge effectiveness. + +## Net investment hedges + +The Company uses cross currency swap (“CCS”) contracts and RMB denominated unsecured senior notes and borrowings to hedge the foreign currency risk associated with investments in net assets of certain PRC subsidiaries held by the Company which are designated as net investment hedges. The Company excludes the changes in the fair value of the CCS contracts attributable to changes other than those due to fluctuations in the spot exchange rate from the assessment of hedge effectiveness and the value of such excluded component is recognized in interest expenses in the unaudited condensed consolidated income statement over the life of the hedging instrument under a systematic and rational method. Changes in the value of the hedging instruments due to fluctuations in the spot foreign currency exchange rates designated in net investment hedges are recognized in accumulated other comprehensive income to offset the cumulative translation adjustments relating to those subsidiaries. For the six months ended September 30, 2025, the losses recognized in accumulated other comprehensive income from changes in value of the CCS contracts and the non-derivative financial instruments designated as net investment hedges amounted to RMB1,279 million. + +Amounts accumulated are reclassified from accumulated other comprehensive income and recognized in the unaudited condensed consolidated income statements upon disposal of those subsidiaries. Once the hedge becomes ineffective, hedge accounting is discontinued prospectively. The estimated fair value of the derivatives is determined based on relevant market information. These estimates are calculated with reference to the market rates using industry standard valuation techniques. + +# ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 2. Summary of significant accounting policies (Continued) + +## (e) Borrowings + +Borrowings consist of bank borrowings, unsecured senior notes, convertible unsecured senior notes and exchangeable bonds. Bank borrowings and unsecured senior notes are recognized initially at fair value, net of upfront fees, debt discounts or premiums, debt issuance costs and other incidental fees. Upfront fees, debt discounts or premiums, debt issuance costs and other incidental fees are recorded as a reduction of the proceeds received and the related accretion is recorded as interest expense in the unaudited condensed consolidated income statements over the estimated term of the facilities using the effective interest method. Convertible unsecured senior notes are accounted for in its entirety as liabilities, and the embedded conversion feature is not required to be accounted for separately under ASC 815. Exchangeable bonds are accounted for under the fair value option with changes in fair value recorded in the unaudited condensed consolidated income statements. + +## (f) Recent accounting pronouncements + +In July 2025, the Financial Accounting Standards Board (“FASB”) issued ASU 2025-05, “Financial Instruments—Credit Losses (Topic 326): Measurement of Credit Losses for Accounts Receivable and Contract Assets”. The amendments provide all entities with a practical expedient when estimating expected credit losses for current accounts receivable and current contract assets arising from transactions accounted for under Topic 606. In developing reasonable and supportable forecasts as part of estimating expected credit losses, all entities may elect a practical expedient that assumes that current conditions as of the balance sheet date do not change for the remaining life of the asset. The new guidance is required to be applied prospectively. This guidance is effective for the Company for the year ending March 31, 2027. Early adoption is permitted. The Company is evaluating the impact of the adoption of this guidance. + +In September 2025, the FASB issued ASU 2025-06, “Intangibles—Goodwill and Other—Internal-Use Software (Subtopic 350-40): Targeted Improvements to the Accounting for Internal-Use Software”, which removes all references to prescriptive and sequential software development stages (“project stages”) throughout Subtopic 350-40. The amendments provide guidance on determining whether there is significant development uncertainty in evaluating the probable-to-complete recognition threshold. The amendments also supersede the website development costs guidance and incorporate the recognition requirements for website-specific development costs. This guidance also specifies the disclosures requirements for capitalized internal-use software costs. The new guidance is required to be applied using either the prospective transition approach, the modified transition approach or the retrospective transition approach. This guidance is effective for the Company for the year ending March 31, 2029. Early adoption is permitted. The Company is evaluating the impact of the adoption of this guidance. + +In November 2025, the FASB issued ASU 2025-08, “Financial Instruments—Credit Losses (Topic 326): Purchased Loans”, which defines purchased seasoned loans and expands the use of the gross-up approach in ASC 326 to the purchased seasoned loans. The new guidance is required to be applied prospectively to loans that are acquired on or after the initial application date. This guidance is effective for the Company for the year ending March 31, 2028. Early adoption is permitted. The Company is evaluating the impact of the adoption of this guidance. + +In November 2025, the FASB issued ASU 2025-09, “Derivatives and Hedging (Topic 815): Hedge Accounting Improvements”, which clarifies certain aspects of the guidance on hedge accounting and addresses several incremental hedge accounting issues arising from the global reference rate reform initiative. The amendments include: (i) expanding the hedged risks permitted to be aggregated in a group of individual forecasted transactions in a cash flow hedge and clarifying the circumstance under which a group of individual forecasted transactions can be considered to have a similar risk exposure; (ii) providing a model to facilitate the application of cash flow hedge accounting to forecasted interest payments on choose-your-rate debt instruments; (iii) expanding hedge accounting for forecasted purchases and sales of nonfinancial assets; (iv) eliminating the requirement to apply the net written option test to a compound derivative comprising a swap and a written option designated as the hedging instrument in a cash flow hedge or a fair value hedge of interest rate risk; and (v) eliminating the recognition and presentation mismatch related to a dual hedge strategy. The new guidance is required to be applied prospectively for all hedging relationships. This guidance is effective for the Company for the year ending March 31, 2028. Early adoption is permitted. The Company is evaluating the impact of the adoption of this guidance. + +# ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 3. Significant mergers and acquisitions, investments and dispositions + +## Disposal of Trendyol GO + +In May 2025, the Company entered into a sale and purchase agreement to sell 85% of the equity interest in Trendyol GO, a wholly-owned subsidiary of Trendyol that operates local service business in Türkiye (the "Disposal"). The cash consideration for the Disposal is approximately US\$0.7 billion (RMB5 billion). The Disposal was completed during the six months ended September 30, 2025, and a gain arising from the Disposal of approximately RMB6 billion was recorded in interest and investment income, net in the unaudited condensed consolidated income statements for the six months ended September 30, 2025, primarily taking into consideration (i) the cash considerations received, (ii) the carrying values of the net assets of Trendyol GO, and (iii) the fair value of the 15% retained equity interest in Trendyol GO. + +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 4. Revenue + +Revenue by segment is as follows: + +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Alibaba China E-commerce Group:
E-commerce (i)
- Customer management152,755168,179
- Direct sales, logistics and others (ii)50,23353,331
202,988221,510
Quick commerce (iii)27,51737,690
China commerce wholesale (iv)11,93813,450
Total Alibaba China E-commerce Group:242,443272,650
Alibaba International Digital Commerce Group:
International commerce retail (v)49,30956,463
International commerce wholesale (vi)11,65613,077
Total Alibaba International Digital Commerce Group60,96569,540
Cloud Intelligence Group (vii)56,15973,222
All others (viii)165,837121,568
Unallocated8881,096
Inter-segment elimination (ix)(46,553)(42,629)
Consolidated revenue479,739495,447
+ +(i) Revenue from E-commerce is primarily generated from Alibaba China E-commerce Group and includes primarily revenue from customer management services, sales of goods and logistics services. +(ii) Revenue from direct sales, logistics and other revenue under Alibaba China E-commerce Group primarily represents direct sales businesses of Tmall Supermarket, Tmall Global and other businesses, and primarily consists of revenue from sales of goods, as well as logistics services. +(iii) Revenue from Quick commerce is primarily generated from through “Taobao Instant Commerce” and the Ele.me app. Quick commerce revenue is net of subsidies that are contra revenue, and includes primarily revenue from logistics services and customer management services. +(iv) Revenue from China commerce wholesale is primarily generated from 1688.com and includes revenue from membership fees and related value-added services and customer management services. +(v) Revenue from International commerce retail is primarily generated from AliExpress, Trendyol and Lazada and includes revenue from customer management services, logistics services and sales of goods. +(vi) Revenue from International commerce wholesale is primarily generated from Alibaba.com and includes revenue from membership fees and related value-added services and customer management services. +(vii) Revenue from Cloud Intelligence Group is primarily generated from the provision of cloud services, which include public cloud services and non-public cloud services. +(viii) Revenue from All others represents revenue from businesses including Freshippo, Cainiao, Alibaba Health, Hujing Digital Media and Entertainment Group, Amap, Intelligent Information Platform (which mainly consists of UCWeb and Quark businesses), Lingxi Games, DingTalk and other businesses. Revenue within All others includes primarily revenue from sales of goods, and logistics services. +(ix) Inter-segment elimination consists of revenue primarily from Cloud Intelligence Group and Cainiao. + +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 4. Revenue (Continued) + +(x) As a result of the change in composition in reportable segments (Note 18), the Company reclassified revenue by segment. Comparative figures for the six months ended September 30, 2024 were updated to conform to the segment presentation. + +Revenue by type is as follows: + +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Customer management services (i)199,821223,399
Membership fees and value-added services23,04424,035
Logistics services59,90468,018
Cloud services41,32652,826
Sales of goods135,837108,435
Other revenue (ii)19,80718,734
479,739495,447
+ +(i) Customer management services mainly include cost-per-click ("CPC"), cost-per-thousand impressions ("CPM"), time-based and cost-per-sale ("CPS") marketing services. +(ii) Other revenue includes revenue from self-developed online games and multiple services provided through various platforms and businesses. + +The amount of revenue recognized for performance obligations satisfied (or partially satisfied) in prior periods for contracts with expected duration of more than one year during the six months ended September 30, 2024 and 2025 were not material. + +Deferred revenue and customer advances of the Company primarily represent service fees prepaid by merchants or customers for which the relevant services have not been provided. Substantially all of the balances of deferred revenue and customer advances are generally recognized as revenue within one year. + +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 5. Income tax expenses + +Composition of income tax expenses is as follows: + +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Current income tax expense14,93913,985
Deferred taxation2,503430
17,44214,415
+ +The Company's effective tax rate for the six months ended September 30, 2024 and 2025 was $21\%$ and $19\%$ , respectively. For the interim financial reporting, the Company estimates the annual tax rate based on projected taxable income for the full year and records an income tax provision for the interim period in accordance with the guidance on accounting for income taxes in an interim period. + +The tax status of the subsidiaries of the Company with major taxable profits is described below: + +\- Alibaba (China) Technology Co., Ltd. (“Alibaba China”), Taobao (China) Software Co., Ltd. (“Taobao China”) and Zhejiang Tmall Technology Co., Ltd. (“Tmall China”), entities primarily engaged in the operations of the Company’s wholesale marketplaces, Taobao and Tmall, respectively, and Alibaba (Beijing) Software Services Co., Ltd (“Alibaba Beijing”) and Alibaba (China) Co., Ltd (“China Co.”), entities primarily engaged in the operations of technology, software research and development and relevant services, were qualified as High and New Technology Enterprises. For the taxation years of 2024 and 2025, Alibaba China, Taobao China, Tmall China, Alibaba Beijing and China Co. applied an EIT rate of 15% as High and New Technology Enterprises. + +Most of the remaining PRC entities of the Company are subject to EIT at 25% for the six months ended September 30, 2024 and 2025. + +## 6. Share-based compensation expense by function + +Share-based compensation expense by function is as follows: + +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Cost of revenue1,205913
Product development expenses3,5602,862
Sales and marketing expenses948958
General and administrative expenses2,5642,137
8,2776,870
+ +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 7. Earnings per share/ADS + +Each ADS represents eight ordinary shares. + +Basic earnings per share is computed by dividing net income attributable to ordinary shareholders by the weighted average number of outstanding ordinary shares, adjusted for treasury shares. Basic earnings per ADS is derived from the basic earnings per share. + +For the calculation of diluted earnings per share, net income attributable to ordinary shareholders for basic earnings per share is adjusted by the effect of dilutive securities, including share-based awards under the treasury stock method and convertible unsecured senior notes and exchangeable bonds under the if-converted method. Certain potentially dilutive instruments, including the exchangeable bonds and the share lending arrangement in connection with the issuance of the exchangeable bonds, and the capped call transactions in connection with the issuance of the convertible unsecured senior notes have been excluded from the computation of diluted net income per share as their inclusion is anti-dilutive. Diluted earnings per ADS is derived from the diluted earnings per share. + +The following table sets forth the computation of basic and diluted net income per share/ADS for the following periods: + +
Six months ended September 30,
20242025
RMBRMB
(in millions, except per share data)
Earnings per share
Numerator:
Net income attributable to ordinary shareholders for computing net income per ordinary share — basic68,14364,106
Dilution effect on earnings arising from non-cash share-based awards operated by equity method investees and subsidiaries(131)(258)
Adjustments for interest expense attributable to convertible unsecured senior notes95143
Net income attributable to ordinary shareholders for computing net income per ordinary share — diluted68,10763,991
Shares (denominator):
Weighted average number of shares used in calculating net income per ordinary share — basic (million shares)19,04518,562
Adjustments for dilutive RSUs and share options (million shares)149186
Adjustments for convertible unsecured senior notes (million shares)265406
Weighted average number of shares used in calculating net income per ordinary share — diluted (million shares)19,45919,154
Net income per ordinary share — basic (RMB)3.583.45
Net income per ordinary share — diluted (RMB)3.503.34
Earnings per ADS
Net income per ADS — basic (RMB)28.6227.63
Net income per ADS — diluted (RMB)28.0026.73
+ +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 8. Equity securities and other investments + +
As of March 31, 2025
Original costCumulative net (losses) gainsCarrying value
RMBRMBRMB
(in millions)
Equity securities:
Listed equity securities79,024(1,394)77,630
Investments in privately held companies113,646(17,479)96,167
Debt investments:
Debt securities and loan investments15,009(6,224)8,785
Other treasury investments227,93581228,016
435,614(25,016)410,598
As of September 30, 2025
Original costCumulative net gains (losses)Carrying value
RMBRMBRMB
(in millions)
Equity securities:
Listed equity securities70,8804,42475,304
Investments in privately held companies120,899(1,279)119,620
Debt investments:
Debt securities and loan investments14,892(5,173)9,719
Other treasury investments252,464103252,567
459,135(1,925)457,210
+ +## Equity securities + +Net unrealized gains, including impairment losses, on equity securities, recognized in interest and investment income, net were RMB4,159 million and RMB20,017 million for the six months ended September 30, 2024 and 2025, respectively. + +Investments in privately held companies include equity investments for which the Company elected to account for using the measurement alternative, for which the carrying value as of March 31 and September 30, 2025 were RMB88,728 million and RMB112,829 million, respectively. For the equity investments accounted for using the measurement alternative, the cumulative upward adjustments as of March 31 and September 30, 2025 were RMB27,197 million and RMB45,334 million, respectively, and the cumulative impairments and downward adjustments as of March 31 and September 30, 2025 were RMB44,232 million and RMB45,741 million, respectively. + +Upward adjustments recorded in the unaudited condensed consolidated income statements on equity investments accounted for using the measurement alternative were RMB4,614 million and RMB20,241 million for the six months ended September 30, 2024 and 2025, respectively, and impairments and downward adjustments recorded in the unaudited condensed consolidated income statements were RMB4,918 million and RMB4,338 million for the same periods, respectively. + +## Debt investments + +Debt investments include convertible and exchangeable bonds accounted for under the fair value option, for which the fair value as of March 31 and September 30, 2025 were RMB963 million and RMB759 million, respectively. The aggregate fair value of these convertible and exchangeable bonds was lower than their aggregate unpaid principal balance as of March 31 and September 30, 2025 by RMB2,420 million and RMB2,529 million, respectively. Unrealized losses recorded in the unaudited condensed consolidated income statements on these convertible and exchangeable bonds were RMB32 million and RMB158 million for the six months ended September 30, 2024 and 2025, respectively. + +# ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 8. Equity securities and other investments (Continued) + +Debt investments (Continued) + +Debt investments also include debt investments accounted for at amortized cost, for which the allowance for credit losses as of March 31 and September 30, 2025 were RMB3,779 million and RMB2,641 million, respectively. Reversal of impairment losses recorded in the unaudited condensed consolidated income statements on these debt investments were RMB1,117 million and RMB1,103 million for the six months ended September 30, 2024 and 2025, respectively. + +As of September 30, 2025, repayment of loans provided to shareholders of equity method investees with total principal amount of RMB5,518 million was expected to be provided substantially through the sale of collateral. Expected credit losses for these loans were assessed on an individual basis, based on the fair value of the corresponding shares pledged as collateral as of the reporting date, adjusted for selling costs as appropriate. The fair value of these collateral as of March 31 and September 30, 2025 were RMB4,325 million and RMB6,250 million, respectively. There was no commitment to lend additional funds. + +The carrying amount of debt investments accounted for at amortized cost approximates their fair value due to the fact that the related effective interest rates approximate rates currently offered by financial institutions for similar debt instruments of comparable maturities. + +Other treasury investments mainly comprise of investments in fixed deposits, certificates of deposits and marketable debt securities with original maturities over one year for treasury purposes. + +## 9. Fair value measurement + +Fair value is defined as the price that would be received to sell an asset or paid to transfer a liability in an orderly transaction between market participants at the measurement date. To increase the comparability of fair value measures, the following hierarchy prioritizes the inputs to valuation methodologies used to measure fair value: + +Level 1 — Valuations based on unadjusted quoted prices for identical assets and liabilities in active markets. + +Level 2 — Valuations based on observable inputs other than quoted prices included in Level 1, such as quoted prices for similar assets and liabilities in active markets, quoted prices for identical or similar assets and liabilities in markets that are not active, or other inputs that are observable or can be corroborated by observable market data. + +Level 3 — Valuations based on unobservable inputs reflecting assumptions, consistent with reasonably available assumptions made by other market participants. These valuations require significant judgment. + +Fair value of listed equity investments are based on quoted prices in active markets for identical assets or liabilities and, if applicable, are adjusted for the characteristic included in the equity security. The valuation of unlisted equity investments that do not have a quoted price may include the use of market and income valuation approaches and the use of estimates, which may include discount rates, investees' liquidity and financial performance, and market data of comparable companies in similar industries. Certain other financial instruments, such as interest rate swap contracts and certain option and forward agreements, are valued based on inputs derived from or corroborated by observable market data. Valuations of investments in convertible and exchangeable bonds that do not have a quoted price are generally performed using valuation models such as the binomial model with unobservable inputs including risk-free interest rate and expected volatility. The valuation of the exchangeable bonds is primarily determined based on quoted market price in the over-the-counter market. The valuation of contingent consideration is performed using an expected cash flow method with unobservable inputs including the probability to achieve the contingencies, which is assessed by the Company, in connection with the contingent consideration arrangements. Investments in privately held companies for which the Company elected to record using the measurement alternative are remeasured on a non-recurring basis, and are categorized within Level 3 under the fair value hierarchy. The values are estimated based on valuation methods using the observable transaction price at the transaction date and considering the rights and obligations of the securities and other unobservable inputs including volatility. + +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 9. Fair value measurement (Continued) + +The following table summarizes the Company's assets and liabilities that are measured at fair value on a recurring basis and are categorized under the fair value hierarchy: + +
As of March 31, 2025
Level 1RMBLevel 2RMBLevel 3RMBTotalRMB
(in millions)
Assets
Time deposits and certificate of deposits (i)357,569357,569
Wealth management products (i)83,14483,144
Marketable debt securities (i)16,12916,129
Restricted cash and escrow receivables43,78143,781
Listed equity securities (ii)(v)67,7129,91877,630
Convertible and exchangeable bonds (ii)145818963
Option agreements (iii)87814901
Deferred consideration (iii)3,0393,039
Others (vi)1,3296,0477,376
111,493468,32110,718590,532
Liabilities
Contingent consideration in relation to investments and acquisitions (iv)484484
Others (iv)3899241,313
3891,4081,797
+ +
As of September 30, 2025
Level 1RMBLevel 2RMBLevel 3RMBTotalRMB
(in millions)
Assets
Time deposits and certificate of deposits (i)314,794314,794
Wealth management products (i)84,78984,789
Marketable debt securities (i)46,23046,230
Restricted cash and escrow receivables40,37440,374
Listed equity securities (ii)(v)65,3279,97775,304
Convertible and exchangeable bonds (ii)139620759
Option and forward agreements (iii)3,4781,1544,632
Deferred consideration (iii)3,0523,052
Others (vi)8455,6276,472
105,701460,25210,453576,406
Liabilities
Exchangeable bonds13,75513,755
Contingent consideration in relation to investments and acquisitions (iv)397397
Others (iv)1,1229072,029
14,8771,30416,181
+ +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 9. Fair value measurement (Continued) + +(i) Included in short-term investments and equity securities and other investments on the unaudited condensed consolidated balance sheets. +(ii) Included in equity securities and other investments on the unaudited condensed consolidated balance sheets. +(iii) Included in prepayments, receivables and other assets on the unaudited condensed consolidated balance sheets. +(iv) Included in accrued expenses, accounts payable and other liabilities on the unaudited condensed consolidated balance sheets. +(v) As of March 31 and September 30, 2025, listed equity securities with fair value of RMB11,921 million and RMB10,415 million were subject to contractual sale restrictions, respectively. +(vi) Others primarily represent other investments with underlying assets measured at fair value. + +Convertible and exchangeable bonds investments categorized within Level 3 under the fair value hierarchy: + +
Amounts RMB (in millions)
Balance as of April 1, 2025818
Additions42
Net decrease in fair value(158)
Disposal(39)
Conversion(34)
Foreign currency translation adjustments(9)
Balance as of September 30, 2025620
+ +Deferred consideration categorized within Level 3 under the fair value hierarchy: + +
Amounts RMB (in millions)
Balance as of April 1, 20253,039
Net increase in fair value40
Foreign currency translation adjustments(27)
Balance as of September 30, 20253,052
+ +## 10. Intangible assets, net + +
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
User base and customer relationships48,56540,420
Trade names, trademarks and domain names26,93627,256
Non-compete agreements6,0305,745
Developed technology and patents4,8235,136
Licensed copyrights and others8,0018,585
94,35587,142
Less: accumulated amortization and impairment(73,444)(67,713)
Net book value20,91119,429
+ +Total amortization expenses recognized for the six months ended September 30, 2024 and 2025 amounted to RMB7,122 million and RMB5,111 million, respectively, including the portion relating to licensed copyrights of RMB3,689 million and RMB3,478 million which were recorded in cost of revenue for the six months ended September 30, 2024 and 2025, respectively. + +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 11. Goodwill + +Changes in the carrying amount of goodwill by segment for the six months ended September 30, 2025 were as follows: + +
AlibabaChinaE-commerceGroupRMBAlibabaInternationalDigitalCommerceGroupRMBCloudIntelligenceGroupRMBCainiaoSmartLogisticsNetworkLimitedRMBLocalServicesGroupRMBHujingDigital MediaandEntertainmentGroupRMBAll othersRMBTotalRMB
(in millions)
Balance as of April 1, 2025164,94520,0773,72820,38520,4478,03117,888255,501
Segment changes12,948(20,385)(20,447)(8,031)35,915
Additions435140
Foreign currency translationadjustments(42)46610
Balance as of September 30, 2025177,89320,0393,80953,810255,551
+ +Gross goodwill balances were RMB302,194 million and RMB302,244 million as of March 31 and September 30, 2025, respectively. Accumulated impairment losses were RMB46,693 million and RMB46,693 million as of March 31 and September 30, 2025, respectively. + +## 12. Supplier Finance Programs + +The Company enters into agreements with several financial institutions and offer supplier finance programs to the Company's suppliers. Suppliers can sell one or more of the Company's payment obligations at their sole discretion to the financial institutions to receive funds, usually at a discounted price, prior to the scheduled due dates to meet their cash flow needs. The Company's current payment terms with the majority of suppliers are up to 180 days. Generally, the Company's rights and obligations are not impacted and the original payment terms, timing or amount, remain unchanged. Except for the pledge of other treasury investments with carrying value of RMB1,500 million and nil as of March 31 and September 30, 2025, respectively, the Company did not provide assets pledged as security or other forms of guarantees under these supplier finance programs. + +The outstanding payment obligations under these supplier finance programs are generally recorded within accrued expenses, accounts payable and other liabilities on the unaudited condensed consolidated balance sheets, except for certain arrangements in which the Company pays the discount to the financial institutions on behalf of the suppliers which are recorded within current bank borrowings. The respective balances are as follows: + +
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
Accrued expenses, accounts payable and other liabilities1,6053,152
Bank borrowings4,4704,047
6,0757,199
+ +# ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 13. Convertible unsecured senior notes + +In May 2024, the Company issued convertible unsecured senior notes for an aggregate principal amount of US\$5.0 billion due on June 1, 2031 (the “2024 Convertible Senior Notes”). The 2024 Convertible Senior Notes are senior unsecured obligations, and interest at an annual rate of 0.5% is payable in arrears semiannually. The 2024 Convertible Senior Notes may be converted into the Company’s ADSs, at the option of holders, at any time prior to the maturity date at an initial conversion rate of 9.5202 ADSs per US\$1,000 principal amount. + +In September 2025, the Company issued zero coupon convertible unsecured senior notes for an aggregate principal amount of approximately US\$3.2 billion due on September 15, 2032 (the “2025 Convertible Senior Notes”). The 2025 Convertible Senior Notes are senior unsecured obligations. The 2025 Convertible Senior Notes may be converted into the Company’s ADSs, at the option of holders, at any time from March 15, 2032 until maturity at an initial conversion rate of 5.1773 ADSs per US\$1,000 principal amount, and may be convertible prior to March 15, 2032 only upon satisfaction of certain conditions. + +The initial conversion rates are subject to adjustment in some events such as dividend distribution. In addition, in the event of a fundamental change that occurs prior to the respective maturity dates or following the Company's delivery of a notice of redemption, the Company will increase the initial conversion rates respectively, which shall not exceed 12.3762 ADSs per US\$1,000 principal amount for the 2024 Convertible Senior Notes and 6.7953 ADSs per US\$1,000 principal amount for the 2025 Convertible Senior Notes, for a holder who elects to convert its notes in connection with such a fundamental change or such notice of redemption. Such make-whole adjustments are subject to the same adjustments as the respective initial conversion rates noted above. Upon conversion, the Company will pay or deliver, at its election, cash, ADSs, or a combination of cash and ADSs. Holders may also elect to receive ordinary shares in lieu of any ADSs deliverable upon conversion, with each ADS representing eight ordinary shares. + +As of September 30, 2025, the adjusted conversion rate for the 2024 Convertible Senior Notes was 9.8915 ADSs per US\$1,000 principal amount, and the adjusted conversion rate taking into account the make-whole adjustments was 12.8589 ADSs per US\$1,000 principal amount. As of September 30, 2025, the conversion rate for the 2025 Convertible Senior Notes remained unadjusted at its initial conversion rate. + +The Company may redeem for cash all but not part of the respective convertible senior notes in the event of certain tax law changes, or at any time if less than $10\%$ of the aggregate principal amount of the respective convertible senior notes originally issued remains outstanding. The Company may also redeem for cash all or part of the 2024 Convertible Senior Notes and the 2025 Convertible Senior Notes on or after June 8, 2029 and September 20, 2030, respectively, provided that the Company's ADS price has been at least $130\%$ of the then effective conversion price for a specific period of time and on the specified date. The redemption price will be equal to the principal amount of the notes being redeemed plus accrued and unpaid interest, if any, to, but excluding, the related redemption date. + +Holders have the right to require the Company to repurchase for cash all or part of the 2024 Convertible Senior Notes and the 2025 Convertible Senior Notes on June 1, 2029 and September 15, 2030, respectively, or in the event of a fundamental change, subject to certain conditions. The repurchase price will be equal to the principal amount of the notes being repurchased plus accrued and unpaid interest, if any, to, but excluding, the related repurchase date. + +As of March 31 and September 30, 2025, the unamortized debt discounts and debt issuance costs of the 2024 Convertible Senior Notes were US\$58 million (RMB424 million) and US\$52 million (RMB367 million), respectively, and the fair value of the 2024 Convertible Senior Notes, based on level 2 inputs, was US\$7,151 million (RMB51,854 million) and US\$9,234 million (RMB65,733 million), respectively. + +As of September 30, 2025, the unamortized debt discounts and debt issuance costs of the 2025 Convertible Senior Notes were US\$42 million (RMB296 million), and the fair value of the 2025 Convertible Senior Notes, based on level 2 inputs, was US\$3,626 million (RMB25,812 million). + +For the six months ended September 30, 2024 and 2025, the effective interest rate for the 2024 Convertible Senior Notes was approximately 0.8%. For the six months ended September 30, 2025, the effective interest rate for the 2025 Convertible Senior Notes was approximately 0.3%. + +# ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 13. Convertible unsecured senior notes (Continued) + +In connection with the issuance of the convertible senior notes, the Company entered into capped call transactions with certain financial institutions at a cost of US\$638 million (RMB4,612 million) and US\$184 million (RMB1,309 million) for the 2024 Convertible Senior Notes and the 2025 Convertible Senior Notes, respectively, which are expected to reduce potential dilution and/or offset cash payments upon conversion. The cap prices of the capped call transactions for the 2024 Convertible Senior Notes and the 2025 Convertible Senior Notes are initially US\$161.60 per ADS and US\$235.46 per ADS, respectively. The capped prices are subject to adjustments similar to the adjustments on the conversion rates of the respective convertible senior notes. The capped call transactions may be settled in cash at the Company’s election. + +## 14. Exchangeable bonds + +In July 2025, the Company issued HKD denominated zero coupon exchangeable bonds referencing the ordinary shares of Alibaba Health Information Technology Limited (“Alibaba Health”), a subsidiary of the Company in which the Company holds approximately 64% of equity interest for an aggregate principal amount of approximately HK\$12 billion (approximately RMB11 billion) due on July 9, 2032 (the “2025 Exchangeable Bonds”). The ordinary shares of Alibaba Health are listed on the Hong Kong Stock Exchange (“AH Shares”). The 2025 Exchangeable Bonds are unsecured and unsubordinated obligations of the Company and are listed on the Vienna MTF operated by the Vienna Stock Exchange. + +The 2025 Exchangeable Bonds may be exchanged into the AH Shares, at the option of holders, at any time prior to the maturity date at an initial exchange rate of approximately 160,513.6 AH Shares per HK\$1,000,000 principal amount. + +The initial exchange rate is subject to adjustment in some events such as dividend distribution by Alibaba Health. In addition, in a relevant event such as the delisting of AH Shares occurring prior to the maturity date or following the Company's delivery of a notice of redemption, the Company will increase the initial exchange rate, which shall not exceed approximately 237,529.7 AH Shares per HK\$1,000,000 principal amount, for a holder who elects to exchange its bonds in connection with such a relevant event or such notice of redemption. Such make-whole adjustment is subject to the same adjustments as the initial exchange rate noted above. Upon exchange, the Company will pay or deliver, at its election, cash, AH Shares, or a combination of cash and AH Shares. + +As of September 30, 2025, the exchange rate for the 2025 Exchangeable bonds remained unadjusted at its initial exchange rate. + +The Company may redeem for cash all but not part of the 2025 Exchangeable Bonds in the event of certain tax law changes, or at any time if less than 10% of the aggregate principal amount of the 2025 Exchangeable Bonds originally issued remains outstanding. The Company may also redeem for cash all or part of the 2025 Exchangeable Bonds on or after July 9, 2030, provided that the AH Shares price has been at least 130% of the then effective exchange price for a specific period of time. The redemption price will be equal to the principal amount of the bonds being redeemed. + +Holders have the right to require the Company to repurchase for cash all or part of the 2025 Exchangeable Bonds on July 9, 2030, or in a relevant event, subject to certain conditions. The repurchase price will be equal to the principal amount of the bonds being repurchased. + +In connection with the issuance of the 2025 Exchangeable Bonds, the Company entered into a stock borrowing and lending arrangement with an affiliate of one of the bookrunners (the “Borrower”), pursuant to which the Company has committed to lending a certain number of AH Shares, which shall not exceed the number of AH Shares exchangeable under the 2025 Exchangeable Bonds, to the Borrower to facilitate hedging activities of certain bondholders. If a termination event occurs and the Borrower is unable to deliver the AH Shares due to legal restrictions, force majeure, or market disruption, cash settlement by the Borrower would be required. As of September 30, 2025, the fair value of the outstanding AH Shares lent under the arrangement was HK\$8,492 million (RMB7,770 million). + +For the six months ended September 30, 2025, losses of RMB327 million arising from changes in the fair value of the 2025 Exchangeable Bonds and the stock borrowing and lending arrangement were recorded in the unaudited condensed consolidated income statements. + +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 15. Bank borrowings + +The Company has made a series of amendments to the syndicated loan facility. Subsequent to the partial repayment of US\$830 million in January 2025, the size of the US\$4.0 billion syndicated loan was reduced to US\$3.17 billion. In September 2025, the Company amended the loan facility and reduced the pricing terms to Secured Overnight Financing Rate (“SOFR”) plus 66 basis points. Effective on November 28, 2025, the facility was restructured as a revolving credit facility with drawdowns permitted in both U.S. dollars and Hong Kong dollars, and the expiration date of the facility was extended to September 30, 2028, with an option to further extend to September 30, 2030. The interest rate of the credit facility is 66 basis points over SOFR or Hong Kong Interbank Offered Rate (“HIBOR”), and the margin will be 81 basis points for the optional extension period. Subsequently, the Company submitted a prepayment notice to repay the outstanding balance of US\$3.17 billion under the revolving credit facility. + +In September 2025, the Company amended the terms of a US\$6.5 billion revolving credit facility agreement. The size of the credit facility was amended to US\$3.33 billion and the utilization currency was also amended from U.S. dollar only to both U.S. dollar and Hong Kong dollar. The interest rate of the credit facility was adjusted to SOFR or HIBOR plus 66 basis points. The expiration date of the credit facility was extended from June 24, 2026 to September 30, 2028, with an option to further extend to September 30, 2030 and the margin will be 81 basis points for the optional extension period. The Company has not yet drawn down this facility. + +## 16. Related party transactions + +During the six months ended September 30, 2024 and 2025, other than disclosed elsewhere, the Company had the following material related party transactions: + +Transactions with Ant Group and its affiliates + +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Amounts earned by the Company
Cloud services revenue (i)4,9778,928
Marketplace software technology services fee and other amounts earned (i)3,5142,819
8,49111,747
Amounts incurred by the Company
Payment processing and escrow services fee (ii)6,3818,937
Other amounts incurred (i)1,4492,378
7,83011,315
+ +(i) The Company has other commercial arrangements and cost sharing arrangements with Ant Group and its affiliates on various sales and marketing, cloud, and other administrative and support services. +(ii) The Company has a commercial agreement with Alipay.com Co., Ltd., a wholly-owned subsidiary of Ant Group (“Alipay”), whereby the Company receives payment processing and escrow services in exchange for a payment for the services fee, which was recognized in cost of revenue. + +As of March 31 and September 30, 2025, the Company had certain amounts of cash held in accounts managed by Alipay in connection with the provision of online and mobile commerce and related services for a total amount of RMB5,863 million and RMB6,379 million, respectively, which have been classified as cash and cash equivalents on the unaudited condensed consolidated balance sheets. + +# ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 16. Related party transactions (Continued) + +Transactions with other investees + +The Company has commercial arrangements with certain investees of the Company related to cloud services. In connection with these services provided by the Company, RMB1,058 million and RMB3,854 million were recorded in revenue in the unaudited condensed consolidated income statements for the six months ended September 30, 2024 and 2025, respectively. + +The Company has commercial arrangements with certain investees of the Company related to marketing services. In connection with these services provided to the Company, RMB358 million and RMB864 million were recorded in cost of revenue and sales and marketing expenses in the unaudited condensed consolidated income statements for the six months ended September 30, 2024 and 2025, respectively. + +The Company has commercial arrangements with certain investees of the Company related to logistics services. In connection with these services provided by the Company, RMB2,092 million and RMB1,680 million were recorded in revenue in the unaudited condensed consolidated income statements for the six months ended September 30, 2024 and 2025, respectively. Costs and expenses incurred in connection with these services provided to the Company of RMB7,999 million and RMB8,251 million were recorded in the unaudited condensed consolidated income statements for the same periods, respectively. + +The Company has extended loans to certain investees for working capital and other uses in conjunction with the Company's investments. As of March 31 and September 30, 2025, the aggregate outstanding balance of these loans was RMB1,771 million and RMB1,992 million, respectively, with remaining terms of up to 5 years and interest rates of up to $10\%$ per annum as of March 31, 2025, and remaining terms of up to 6 years and interest rates of up to $10\%$ per annum as of September 30, 2025. + +The Company provided a guarantee for a term loan facility of HK\$7.7 billion in favor of Hong Kong Cingleot Investment Management Limited (“Cingleot”), a company that is partially owned by the Company, in connection with a logistics center development project at the Hong Kong International Airport. In May 2024, the loan facility was modified to a revolving loan facility and the facility amount was reduced to HK\$6.5 billion. As of March 31 and September 30, 2025, HK\$5,090 million (RMB4,697 million) and HK\$5,226 million (RMB4,771 million) was drawn down by Cingleot under this facility, respectively. + +The Company's ecosystem offers different platforms on which different enterprises operate and the Company believes that all transactions on the Company's platforms are conducted on terms determined based on normal commercial negotiation with similar unrelated parties. + +Other than the transactions disclosed above or elsewhere in the unaudited condensed consolidated financial statements, the Company has commercial arrangements with other investees and other related parties to provide and receive certain marketing, cloud and other services and products. The amounts relating to these services provided and received represent less than $1\%$ of the Company's revenue and total costs and expenses, respectively, for the six months ended September 30, 2024 and 2025. + +In addition, the Company has made certain acquisitions and equity investments together with related parties from time to time. The agreements for acquisitions and equity investments were entered into by the parties involved and conducted on fair value basis. + +## 17. Risks and contingencies + +(a) The Company is incorporated in the Cayman Islands and considered as a foreign entity under PRC laws. Due to legal restrictions on foreign ownership and investment in, among other areas, value-added telecommunications services, which include the operations of Internet content providers, the Company operates its Internet businesses and other businesses through various contractual arrangements with VIEs that are incorporated and owned by PRC citizens or by PRC entities owned and/or controlled by PRC citizens. The VIEs hold the licenses and approvals that are essential for their business operations in the PRC and the Company has entered into various agreements with the VIEs and their equity holders such that the Company has the right to benefit from their licenses and approvals and generally has control of the VIEs. In the Company’s opinion, the current ownership structure and the contractual arrangements with the VIEs and their equity holders as well as the operations of the VIEs are in substantial compliance with all existing PRC laws, rules and regulations. However, there may be changes and other developments in PRC laws, rules and regulations. Accordingly, the Company gives no assurance that PRC government authorities will not take a view in the future that is contrary to the opinion of the Company. If the current ownership structure of the Company and its contractual arrangements with the VIEs and their equity holders were found to be in violation of any existing or future PRC laws or regulations, the Company’s ability to conduct its business could be impacted and the Company may be required to restructure its ownership structure and operations in the PRC to comply with the changes in the PRC laws which may result in deconsolidation of the VIEs. +(b) The PRC market in which the Company operates poses certain macro-economic and regulatory risks and uncertainties. These uncertainties extend to the ability of the Company to operate or invest in online and mobile commerce or other Internet related businesses, representing the principal services provided by the Company, in the PRC. The information and technology industries are highly regulated. Restrictions are currently in place or are unclear regarding what specific segments of these industries foreign owned enterprises, like the Company, may operate. If new or more extensive restrictions were imposed on the segments in which the Company is permitted to operate, the Company could be required to sell or cease to operate or invest in some or all of its current businesses in the PRC. +(c) Because of the Company's equity interest in and close association with Ant Group and overlapping user bases, regulatory developments, litigation or proceedings, media and other reports, whether or not true, and other events that affect Ant Group could also negatively affect customers', regulators', investors' and other third parties' perception of the Company. Changes in Ant Group's business and future prospects, or speculation of such changes, as well as additional regulatory requirements placed on Ant Group, could in turn have a material adverse effect on the Company. +(d) The Company's sales, purchase and expense transactions are generally denominated in RMB and a significant portion of the Company's assets and liabilities are denominated in RMB. RMB is not freely convertible into foreign currencies. In the PRC, foreign exchange transactions are required by law to be transacted only by authorized financial institutions at exchange rates set by the People's Bank of China (the “PBOC”). + +Remittances in currencies other than RMB by the Company in the PRC must be processed through the PBOC or other PRC foreign exchange regulatory bodies and require certain supporting documentation in order to effect the remittance. If the foreign exchange control system prevents the Company from obtaining sufficient foreign currencies to satisfy its currency demands, the Company may not be able to pay dividends in foreign currencies and the Company’s ability to fund its business activities that are conducted in foreign currencies could be adversely affected. + +## 17. Risks and contingencies (Continued) + +(e) In the ordinary course of business, the Company makes strategic investments to increase the service offerings and expand capabilities. The Company continually reviews its investments to determine whether there is a decline in fair value below the carrying value. Fair value of the listed securities is subject to volatility and may be materially affected by market fluctuations. +(f) Financial instruments that potentially subject the Company to significant concentration of credit risk consist principally of cash and cash equivalents, short-term investments, restricted cash and equity securities and other investments. As of September 30, 2025, substantially all of the Company's cash and cash equivalents, restricted cash, short-term investments and other treasury investments were held by major financial institutions located worldwide, including Chinese mainland and Hong Kong S.A.R. If the financial institutions and other issuers of financial instruments held by the Company could become insolvent or if the markets for these instruments could become illiquid as a result of a severe economic downturn, the Company could lose some or all of the value of its investments. +(g) The Company offers trade assurance program on the international wholesale marketplaces at no charge to the wholesale buyers and sellers. If the wholesale sellers who participate in this program do not deliver the products in their stated specifications to the wholesale buyers on schedule, the Company may compensate the wholesale buyers for their losses on behalf of the wholesale sellers up to a pre-determined amount following a review of each particular case. In turn, the Company will seek a full reimbursement from the wholesale sellers for the prepaid reimbursement amount, yet the Company is exposed to a risk over the collectability of the reimbursement from the wholesale sellers. During the six months ended September 30, 2025, the Company did not incur any material losses with respect to the compensation provided under this program. Given that the maximum compensation for each wholesale seller is pre-determined based on their individual risk assessments by the Company considering their credit profile or other relevant information, the Company determined that the likelihood of material default on the payments is not probable and therefore no provisions have been made in relation to this program. +(h) In the ordinary course of business, the Company is from time to time involved in legal proceedings and litigations and is subject to regulatory investigations. The more stringent obligations under laws and regulations will create additional operational requirements with increased compliance costs for the Company. In March 2024, the European Commission, or the EU Commission, opened formal proceeding against AliExpress to assess whether AliExpress breached the Digital Services Act. On June 18, 2025, the EU Commission issued preliminary findings in which it considers on a preliminary basis that AliExpress is in breach of its obligation to assess and mitigate risks related to the dissemination of illegal content on its platform. In September 2025, AliExpress filed observations concerning such preliminary findings. The ultimate timeline and final outcome of the investigation is currently uncertain and subject to further communications with the EU Commission and their final decision based on such further communications and their deliberations. Any potential loss associated with the investigation is not reasonably estimable at this stage. Except for the above, there are no legal proceedings and litigations that have in the recent past had, or to the Company’s knowledge, are probable to have, a material impact on the Company’s financial positions, results of operations or cash flows. The Company did not accrue any material loss contingencies in this respect as of September 30, 2025. +(i) The Russia-Ukraine conflict has resulted in significant disruptions to supply chains, logistics and business activities in the region that has negatively affected our international commerce business and Cainiao's international logistics business. The conflict has also caused, and continues to intensify, significant geopolitical tensions in Europe and across the globe. The resulting sanctions imposed are expected to have significant impacts on the economic conditions of the countries and markets targeted by such sanctions, and may have unforeseen, unpredictable secondary effects on global energy prices, supply chains and other aspects of the global economy. The conflict may adversely affect the Company's business, financial condition and results of operations. +(j) The United Nations and a number of countries and jurisdictions, including China, the United States and the EU, have adopted various export control and economic or trade sanction regimes. In particular, the United States government and other governments have increasingly threatened and/or imposed export control, as well as economic, trade and other sanctions on a number of China-based companies. The United States and other countries may impose other and more expansive restrictions on the supply of chips, computing power or other technologies or services to China and China-based companies, including the Company, in the future. Such restrictions may affect the Company's businesses by limiting the Company's ability to upgrade its products and services, technological capabilities, in particular AI technologies, and to maintain its competitive edge, thereby negatively affecting the Company's results of operations, financial condition and growth potential. + +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 18. Segment information + +Prior to the quarter ended June 30, 2025, the Company had six reportable segments, namely Taobao and Tmall Group, Alibaba International Digital Commerce Group, Cloud Intelligence Group, Cainiao Smart Logistics Network Limited, Local Services Group, and Hujing Digital Media and Entertainment Group. Starting from the quarter ended June 30, 2025, the Company has implemented a new organizational structure, which the CODM started to review information under a new reporting structure, and segment reporting has been updated to conform to this change. Comparative figures for the six months ended September 30, 2024 were updated to conform to the segment presentation. + +Segment information is presented before elimination of inter-segment transactions. In general, revenue, cost of revenue and operating expenses are directly attributable, or are allocated, to each segment. The Company allocates costs and expenses that are not directly attributable to a specific segment, such as those that support infrastructure across different segments, to different segments mainly on the basis of usage, revenue or headcount, depending on the nature of the relevant costs and expenses. The Company does not allocate assets to its segments as the CODM does not evaluate the performance of segments using asset information. + +The CODM regularly reviews adjusted earnings before interest, taxes and amortization (“Adjusted EBITA”) for each segment which is considered as a segment operating performance measure. The CODM uses revenue and Adjusted EBITA to assess performance for each segment and allocate resources for each segment in the annual budget and forecasting process. + +The following table presents the information by segment for the six months ended September 30, 2024 and 2025: + +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Alibaba China E-commerce Group
Revenue242,443272,650
Costs and expenses (i)(149,363)(223,764)
Adjusted EBITA (ii)93,08048,886
Alibaba International Digital Commerce Group
Revenue60,96569,540
Costs and expenses (i)(67,576)(69,437)
Adjusted EBITA (ii)(6,611)103
Cloud Intelligence Group
Revenue56,15973,222
Costs and expenses (i)(51,161)(66,664)
Adjusted EBITA (ii)4,9986,558
All others
Revenue165,837121,568
Costs and expenses (i)(168,747)(126,353)
Adjusted EBITA (ii)(2,910)(4,785)
+ +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 18. Segment information (Continued) + +The following table presents the reconciliation from the total segments Adjusted EBITA to the consolidated net income for the six months ended September 30, 2024 and 2025: + +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Total segments Adjusted EBITA88,55750,762
Unallocated (iii)(2,142)(1,640)
Inter-segment elimination(819)(1,205)
Non-cash share-based compensation expense(7,775)(6,076)
Amortization and impairment of intangible assets, and others(3,441)(1,488)
Provision for the shareholder class action lawsuits(3,145)
Consolidated income from operations71,23540,353
Interest and investment income, net17,12937,468
Interest expense(4,615)(4,995)
Other (expense) income, net(1,221)1,329
Income tax expenses(17,442)(14,415)
Share of results of equity method investees2,4833,254
Consolidated net income67,56962,994
+ +The following table presents the depreciation and impairment of property and equipment, and operating lease cost relating to land use rights by segment for the six months ended September 30, 2024 and 2025: + +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Alibaba China E-commerce Group7962
Alibaba International Digital Commerce Group547370
Cloud Intelligence Group6,90611,877
All others4,0871,226
Total segments depreciation and impairment of property and equipment, and operating lease cost relating to land use rights11,61913,535
+ +(i) Segment costs and expenses primarily comprise components that are included in cost of revenue, product development expenses, sales and marketing expenses, as well as general and administrative expenses. +(ii) Adjusted EBITA represents net income before interest and investment income, net, interest expense, other (expense) income, net, income tax expenses, share of results of equity method investees, certain non-cash expenses, consisting of share-based compensation expense, amortization and impairment of intangible assets, impairment of goodwill, and others, which the Company does not believe are reflective of the Company's core operating performance during the periods presented. +(iii) Unallocated primarily relates to certain costs incurred by corporate functions and other miscellaneous items that are not allocated to individual segments. + +Details of the Company's revenue by segment are set out in Note 4. + +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 19. Aging analysis + +## Accounts receivable + +The aging analysis of the accounts receivable, net of allowance, based on billing date is as follows: + +
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
0-3 months25,17225,073
3-6 months3,0783,395
6-12 months1,7752,344
Over 1 year627834
Accounts receivable, net of allowance30,65231,646
+ +## Accounts payable + +The aging analysis of the accounts payable based on billing date is as follows: + +
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
0-3 months52,01956,073
3-6 months3,9905,696
6-12 months8461,600
Over 1 year1,3461,570
Accounts payable58,20164,939
+ +## 20. Dividends + +The Board did not recommend the distribution of interim dividend for the six months ended September 30, 2024 and 2025. + +## 21. Subsequent events + +Save as disclosed in the unaudited condensed consolidated financial statements, there were no significant events that might affect the Company since September 30, 2025. + +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 22. Reconciliations between U.S. GAAP and International Financial Reporting Standards + +The unaudited condensed consolidated financial statements are prepared in accordance with U.S. GAAP, which differ in certain respects from International Financial Reporting Standards (“IFRS”). The effects of material differences between the unaudited condensed consolidated financial statements of the Company prepared under U.S. GAAP and IFRS are as follows: + +Reconciliation of Unaudited Condensed Consolidated Balance Sheets (Extract) + +
As of March 31, 2025
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity’s own equity (viii)Amounts under IFRS
(in millions of RMB)
Equity securities and other investments356,8185,0033,547365,368
Prepayments, receivables and other assets83,431(258)485578(1,010)3826,87490,482
Investments in equity method investees210,169(241)(226)209,702
Property and equipment, net203,3481,318204,666
Intangible assets, net20,911(1,861)42619,476
Goodwill255,501(36,284)3,180222,397
Total assets1,804,227(38,644)5,4883,321578(1,010)5,3066,8741,786,140
Deferred tax liabilities48,454(719)368(283)(78)(51)47,691
Convertible unsecured senior notes35,83416,02151,855
Other liabilities17,644(2)10,24727,889
Total liabilities714,121(719)368(283)(78)(2)10,247(51)16,021739,624
Mezzanine equity11,713(11,713)
Total shareholders’ equity1,009,858(17,402)5,1203,604656(1,008)1,4392,616(9,147)995,736
Noncontrolling interests68,535(20,523)272,74150,780
Total equity1,078,393(37,925)5,1203,604656(1,008)1,4665,357(9,147)1,046,516
Total liabilities, mezzanine equity and equity1,804,227(38,644)5,4883,321578(1,010)5,3066,8741,786,140
+ +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued) + +Reconciliation of Unaudited Condensed Consolidated Balance Sheets (Extract) + +As of September 30, 2025 + +
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity's own equity (viii)Amounts under IFRS
(in millions of RMB)
Equity securities and other investments411,9533,2824,223419,458
Prepayments, receivables and other assets96,927(258)521637(974)1,0238,315106,191
Investments in equity method investees206,862(241)(784)205,837
Property and equipment, net246,539699247,238
Intangible assets, net19,429(1,688)44518,186
Goodwill255,551(36,278)3,304222,577
Total assets1,883,880(38,465)3,8033,439637(974)5,4718,3151,866,106
Deferred tax liabilities46,802(684)324(292)(170)(52)45,928
Convertible unsecured senior notes57,48134,06491,545
Exchangeable bonds13,75513,755
Other liabilities21,3548,46729,821
Total liabilities772,095(684)324(292)(170)8,467(52)34,064813,752
Mezzanine equity9,884(9,884)
Total shareholders' equity1,032,495(17,252)3,4793,731807(974)1,4322,711(25,749)1,000,680
Noncontrolling interests69,406(20,529)(15)2,81251,674
Total equity1,101,901(37,781)3,4793,731807(974)1,4175,523(25,749)1,052,354
Total liabilities, mezzanine equity and equity1,883,880(38,465)3,8033,439637(974)5,4718,3151,866,106
+ +# ALIBABA GROUP HOLDING LIMITED + +# NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS + +# FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued) + +Reconciliation of Unaudited Condensed Consolidated Income Statement (Extract) + +Six months ended September 30, 2024 + +
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity's own equity (viii)Amounts under IFRS
(in millions of RMB)
Revenue479,73997479,836
Cost of revenue(290,135)(22)641(235)(289,751)
Product development expenses(27,555)(10)(1)(27,566)
Sales and marketing expenses(65,167)(2)1(323)(65,491)
General and administrative expenses(23,057)(18)70(9)(23,014)
Amortization and impairment of intangible assets(3,441)16039(3,242)
Other gains, net851851
Income from operations71,235160(52)712(432)71,623
Interest and investment income, net17,1291,750(56)2261(7,689)11,217
Interest expense(4,615)(775)(434)(12)(5,836)
Other expense, net(1,221)442(779)
Income tax expenses(17,442)(35)109(227)86611(221)(16,939)
Share of results of equity method investees2,4831,15143,638
Net income67,5691251,859868818(52)(412)(162)(7,689)62,924
Net loss attributable to noncontrolling interests854(59)(3)1331451,070
Accretion of mezzanine equity(280)280
Net income attributable to ordinary shareholders68,143661,859868815(52)1(17)(7,689)63,994
+ +## 22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued) + +Reconciliation of Unaudited Condensed Consolidated Income Statement (Extract) (Continued) + +Six months ended September 30, 2025 + +
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity’s own equity (viii)Amounts under IFRS
(in millions of RMB)
Revenue495,4471495,448
Cost of revenue(287,210)(12)304(425)(287,343)
Product development expenses(32,096)(7)(15)(32,118)
Sales and marketing expenses(119,674)(2)1(154)(119,829)
General and administrative expenses(14,778)(14)36(5)(14,761)
Amortization and impairment of intangible assets(1,633)17367(1,393)
Other gains (losses), net297(506)(209)
Income from operations40,353173(35)341(1,037)39,795
Interest and investment income, net37,468(1,641)210611016(18,008)18,116
Interest expense(4,995)(395)(337)1,536(30)(4,221)
Other income, net1,32973961,732
Income tax expenses(14,415)(35)39160(31)(389)(14,770)
Share of results of equity method investees3,254(79)293,204
Net income62,994(250)(1,602)13254341,546(1,044)(18,008)43,856
Net income attributable to noncontrolling interests(1,326)(65)(3)887251(256)
Reversal of accretion of mezzanine equity2,438(2,438)
Net income attributable to ordinary shareholders64,106(315)(1,602)1325134(5)(793)(18,008)43,600
+ +## (i) Consolidation and business combinations + +The Company consolidates an entity when the Company obtains control over the entity and deconsolidates the entity upon the loss of control. Under U.S. GAAP, control generally exists when the Company obtains a controlling financial interest over an entity, whereby the usual condition is ownership of over 50% of the voting shares. Under IFRS, de facto control exists when the Company has the practical ability to direct the relevant activities of the entity, even if the Company owns less than 50% of the voting shares. + +The Company recognizes noncontrolling interests to reflect the portion of equity of a subsidiary that is not attributable to the Company. Under U.S. GAAP, noncontrolling interests are measured at fair value and full goodwill in relation to the acquiree is recognized in a business combination. Under IFRS, the Company can elect, on a transaction-by-transaction basis, to measure noncontrolling interests at the noncontrolling interests' proportionate share of the acquiree's net identifiable assets and partial goodwill is recognized to reflect the controlling interests only. + +The Company recognizes an impairment loss when the Company determines that the carrying value of goodwill is not recoverable. Under U.S. GAAP, the impairment loss is measured by comparing the carrying value of the reporting unit, including goodwill, with its fair value. Under IFRS, the carrying value is compared with the recoverable amount, which is the higher of fair value less costs of disposal and value in use. + +# ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued) + +## (ii) Equity securities without readily determinable fair value + +Under U.S. GAAP, the Company can elect, on an instrument-by-instrument basis, to apply the measurement alternative to record the investments in equity securities without readily determinable fair values at cost, less impairment, with subsequent adjustments for observable price changes recognized in the unaudited condensed consolidated income statements. Under IFRS, these investments are measured at fair value with changes in fair value recognized in the unaudited condensed consolidated income statements. + +## (iii) Equity method investments + +The Company generally applies the equity method to account for equity investments over which it has significant influence. Under U.S. GAAP, significant influence is presumed to exist for an investment in limited partnership or unincorporated entity, unless the investment is so minor that the Company has virtually no influence over the entity's operating and financial policies. Under IFRS, significant influence is presumed to exist for an investment of over $20\%$ of the voting rights of an entity. + +The Company records its share of the post-acquisition results of its equity method investees and adjusts for the basis differences that exist between the carrying values of the equity method investments and the Company's proportionate share of the carrying value of the investee's net assets. Adjustments are made to the financial statements of the equity method investees prepared under U.S. GAAP in order to conform to the Company's accounting policies under IFRS and to reflect the basis differences of the equity method investments under IFRS, if different from those under U.S. GAAP. + +## (iv) Share-based awards + +The employees of the Company hold share-based awards relating to an equity method investee of the Company that were granted and will be settled by related parties or economic interest holders of the Company. Under U.S. GAAP, the cost related to these awards is recognized over the requisite service period, with subsequent changes in fair value of these awards recognized in the unaudited condensed consolidated income statements. Under IFRS, these awards are not considered as share based payments of the Company and the cost relating to these awards is not recognized. + +The Company accounts for income tax effects of share-based awards that ordinarily give rise to tax deduction. Under U.S. GAAP, deferred taxes for these awards are measured based on share-based compensation expenses recognized in the unaudited condensed consolidated financial statements. Under IFRS, deferred taxes for these awards are measured based on future tax deduction estimated at the end of each reporting period. + +## (v) Operating leases + +Under U.S. GAAP, the amortization of right-of-use assets and the interest expense related to lease liabilities are recorded together as lease expense and recognized in the unaudited condensed consolidated income statements on a straight-line basis. Under IFRS, the right-of-use assets are amortized on a straight-line basis while the interest expense related to lease liabilities are recognized in the unaudited condensed consolidated income statements using effective interest method. + +## (vi) Redeemable noncontrolling interests + +Equity interests issued by certain subsidiaries of the Company are redeemable. Under U.S. GAAP, redeemable equity interests are classified as mezzanine equity if the redemption is outside the Company's control and as noncontrolling interests if equity interests issued by finite-lived subsidiaries are mandatorily redeemable only upon liquidation. Under IFRS, these redeemable equity interests are generally classified as financial liabilities. + +# ALIBABA GROUP HOLDING LIMITED NOTES TO UNAUDITED CONDENSED CONSOLIDATED FINANCIAL STATEMENTS FOR THE SIX MONTHS ENDED SEPTEMBER 30, 2024 and 2025 + +## 22. Reconciliations between U.S. GAAP and International Financial Reporting Standards (Continued) + +## (vii) Hyperinflation + +Under U.S. GAAP, when the Company determines that a subsidiary is operating in a highly inflationary economy, the financial statements of this subsidiary are remeasured prospectively as if its functional currency was the functional currency of its immediate parent company. Under IFRS, the financial statements of the subsidiary operating in a highly inflationary economy are restated in terms of the measuring unit current at the end of the reporting period. + +## (viii) Convertible unsecured senior notes and instruments in an entity's own equity + +Certain unsecured senior notes issued by the Company are convertible into its ordinary shares. Under U.S. GAAP, these convertible notes are accounted for in entirety as liabilities measured at amortized cost with debt discounts and debt issuance costs amortized using the effective interest method. Under IFRS, these convertible notes that the Company can settle in cash at its election are accounted for as hybrid instruments, and the Company designates the hybrid instruments as financial liabilities measured at fair value with changes in fair value recognized in the unaudited condensed consolidated income statements, whereas changes in fair value attributed to credit risk recognized separately in other comprehensive income are insignificant. Furthermore, under IFRS, if the conversion options embedded in the convertible unsecured senior notes or the exchange options embedded in the exchangeable bonds are not classified as equity, the respective instruments are classified as current liabilities when the holders have the right to convert or exchange them within twelve months after the reporting period. + +Under U.S. GAAP, for an instrument in an entity's own equity to be classified as equity, the instrument must be both (1) indexed to the entity's stock (commonly referred to as “fixed-for-fixed” test) and (2) classified in shareholders' equity, regardless of whether the instrument is a derivative under ASC 815. Under IFRS, a derivative in an entity's own equity may nonetheless be considered an equity instrument if the derivative meets the “fixed-for-fixed” test, whereas a non-derivative that includes no contractual obligation for the entity to deliver cash, financial asset or a variable number of its own equity instrument is recognized in shareholders' equity with no subsequent measurement. Besides, differences exist in the application of the “fixed-for-fixed” test between U.S. GAAP and IFRS. For example, settlement alternatives that may permit equity classification under U.S. GAAP typically result in an asset or a liability classification under IFRS. Consequently, depending on the specific terms and conditions, an instrument in an entity's own equity may be classified as equity under U.S. GAAP but as an asset or a liability measured at fair value, with changes in fair value recognized in the unaudited condensed consolidated income statement, under IFRS, or vice versa. + +![](images/660e275e7fb83ccfc83880f2a2915923c0d612de14238c317aa7c59ec73b42d0.jpg) + +## Alibaba + +Alibaba Group Holding Limited \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/images/image-1-Alibaba.jpg b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/images/image-1-Alibaba.jpg new file mode 100644 index 0000000000000000000000000000000000000000..61a0f0f3da4419241fb40b25d4e33801991d5b95 GIT binary patch literal 4916 zcmbW4XH-*Lw}v-OI!H$$geC$3l-`Mop&meL=p00fMDc*sAVfh$awq}dcmVnsU zoTQ|pu9C`GbuBF|@M(QxJq;r@O)ZUIn=o;5a`JHVi1YG_Ysg5-X#Agp{ss`>03HF~ zS(s!2W&tJ^0VaA60A|$5#`HUY{|+W*7FM=n>>Qk2+>8Yc#{p(078Yh!7B)6kR>tZ$ z#&>{KfK5kjBclZ3*KZ{9^vvws{KDeW^2X-Y zHf3jbZ~x#I7Zbqp4;JJ757__U5@2vKv$C?Vvj5^@Vvb@YivTN|jQTOb^Y-k%p+d46 zcQ}MEWE9tRa>{8stczR@8{`s|*CLs0xPb(MS{k>byV+moP=+WOex}=5<3r6K-Z1BSLCveLg6!m z;Y~?>O=M+r4`opRI*90?vgyzP<$;!Uoh&^Cu)##@wOi@$=j(Ie&bL%G zT?3uKJ^4mM=;a1}=4H`0ZKcWKiF@%Y@#g7-=DgyZNf&g`q`nxosZ*reCDqDy*u_{` z5AB@2)IS?;54&>hKV@M^b&2UTa3y~!JpGuD&uY%fktoaO4<1|*+R|UG(aVU5B#D122sJj`$=pQOw=AUW!2euKu(rBKWKjr zTxJPFhA$33uV(Y>t$a}FHd$K3C^_h|u)V!a#Ep>F@_fCd%3AQJtg9hfPX@)kWjD0) z)LGSSl`F8@Q3hg&ra7lE7|22411+Vu_ctb7pb~mH#t~F&J|Yr#%m%Xgv$?$ zo2f0M{p;&i51k&GUx3`O+X$#Ef7p&HI!@zkzd@`Pu6Npg(N1>hlW!=4928bMY1>E* zz;E|P9GB8BSr_zn~dddc7y#UTe!AMlq1N`Uu>S@YKmEj#>YrC z7kdWfdV3ZbsnTjL>G>Y<`omuaw_G!*DN?QKzJJ1nwi2m5vwPeUaZ_F}#Zp4VcMa0{ zjvRLv2g;+Ih}2yPX+NM)B>L zd0Af{U0bY1Vz;kC-^3Cz9%H6gnY&pLlEG(1%mAD5; zv*+Dw!L9J!VBvL*K^%MVI0n6|G@Pop+35z>b*Ne%$Pf9Uh>dEgPAh0`X-PY?V>wF$ zA@bWdj_@%akif@uAl|eArB=ROwLE>*VQ==TIj4JLr^M4#cyoJ0y$w-U-~x2)FHLX% z*y)MOyK<{07r^AZ*N>322K3sdepEn!v9)BDr--I2ELKH=}-+dw9I4n zf`8ME9Wj=6jXoGaNjOS-R z8S|{WI@S9D`LZx?7fkj>c%^qQx_+PdOobdbrJRNiwTA8YYY%zZ8EThv`j#n$yfS8f zJ@9S^^Sws_CNA`@CcX_?hP)p0vxgaqUo{WqJiVp8Ja1_rwq@jrvAOKv<*3g%yY}#d zS`1nd_?0zdT{OW-K|0`&o>q6R8`p}ur`u(6Vk&scA+BWRVqTwxb$BtOH?fuy84#)R z!7bgv*0L0XUblPBrC8~5mNafR=Q*z36e4QV`Us4=fxPPV)gcqzsyyiK{wgV^)L?(P zn)|4)MBk-~6n6L7{*|h}dln<MR|wm-+v`y3k*cQ3r8Y_Tp`Oz7ibpXQh` z`xLY~wEg&?ZCM3(GF*o+tNz(6r2Bwa4zJJgT}8g!2OAem5EXljtcF?gZ5|n0lCIwezQK0kd5b{1{=i4OtUEYdO=I$KJ3R}l5##p)DqTgRk`$x zq=Bt6jGLCkx!*B%V~q}^`YjQJf}=Lv&zZ#VX?`;gLlCP} ze21$r4R=B%{lJdnMab>5;ywtqm}}_}t#^kUVW!il1QkH!!kycCfWsgxj&rjZCc-L~ zL-e>SCh*>goKPzdM#4RK6-Gat5#GxPf6UuZ&k`#=(~)Cr*HISpq}4E&A#|9fzC5vb zS?bvR&1uUaT12L4bINTKyCt{u^@AVa%$;N4RVDC@w}M)lGJb8Q+;nkWJO#uR=`i}U zL~j{VYnMd4OWmziqyzqzbbxEHVV$kuibQ3zf3H6iJEbOQN?q_!$6z7f*KY^683;Vs z)Uz1P`#PnZlI0bjG?f?N-!#gUCiDI-IIq1uk#gWN_I>YWw}lInSU&s2-wQV>#vMYQ z&l{%4CY-d?Q6(NyTuS?oV~dvD4^N861-A!&i+|29ko-0l?>8Xz6O}~)nSR&<)%l|pz9{ES>d zy;LbdufPf?Osey{{kFV1|3IDM7?CRO7E3<%#^xRD5n|X# z?itFT6o(~4hvp>{soYBxN=<a4ZEAx4@O3jY; zLr(V?bOjxFeBjW84hSAJ;=aI#bWdJt?*q$;e`>H+^$0vQ^VHLGQthcwfF9qH`os~G ziMomkq_EyeV>**|% zdN;q>$#+V(7`;@L`n#ciH5D4-*WDtq0w-I@l8jDxB9Z(z;T%?_kNjF?t1X5S3nm|Y zT@OrZA66V$%i2{$g_Xw5gJ_BS#Am~@$LCiN9s;}3-9lI{E{2l?MOYC7iihpLO_91nNcD`-FW zh+^Khe&~hVmuUkx$NT^C{$xUmZEJ&0F(9#A!}nZwn3X$MfU_k9mG<=r@exp|7@Uj#3AAKc;%hM8@j?q_oxX|T3Y{`xse z+L?V@8RXW=wHasa&^&tM ze1w9aW06*Je>SQM`kgNRd`mrcRfux$N6hQ>RyWko;eKhJ1A3 zdGOP4DQ;jdpJ&Is_h=L0uI^YN-d-&}t*|ush_IwNJSeI+DsTaN8vxxiKtope@9RUs z@*hjO-g71g9Qz}3YDRm(dG+T>t9u|*IUW(;hp@|bWM08YmFqe}biiohQM}vRnCK5Y zD|UrEJGCdC8VDa14DU<9VaS4b-Y;KfN%g2U2rAX5aWH(qfk#JY*?i*~D=nb8B-mZhh5aK(~%X-7?Y znOS0aRp~614!~Y=g&Dq;eV^hyGUpm`>vm9btoj0FGw3|sGZFHEJ zy$3}Yby~K!oy20ru78f*t{;>rQb2o7 zxz%w8Z_A8H3db&CFcGyxtr$q={7u z`14tW{PgR7l*EcAv!Dfot|pWYq#;CC?AzM)TUv5_0xqAUDhK8}`fy#t#;C$K`p)!h zS2IRnoBK{SJIUJ`5>IhE0YGBGybzkyvA!| lRrl5{Iw13DRgN*5^~~nzD?EXT^dMzmj2rd;Gui3i{tI2kI}!i@ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/images/image-2-### (vii) .jpg b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/images/image-2-### (vii) .jpg new file mode 100644 index 0000000000000000000000000000000000000000..61136e120ccbc93739f3a3b3161f27f7ab7e6a1c GIT binary patch literal 3619 zcmbW2XH?Tmx5oc80|?S0(j!%rqBM~bas>kfkr-N#jx-|zNc;36-v)27|@7=SW`7(RWn%~Uc&m4~(&jQ>=`Uri1fdK#* zP6lv13FrZ=r%r)Rv9N+bAT~Bub`CHn$LZ4?g1mfOU{N86n5dA5$ayKn3+E-|B}GJJ z)i26lQBqY^g2}rK@{jaA0*Wll||&BK}`w{{j1lYXab4VmNs`CLRC=G{eE@ zPDmIS1qc3-nT@rrJz{C{x)fm*e@pdMfUc)>fF9g$Bn&j{^>8rAvWGtD+A407It^4C z3%@GP*NJ81TDLPNtF2Hy*B%aitsJ5(w-yv;_1l+oJ*hD?;5S+so_a+aCid055D@Yw zEO4t0RF`DB#3S7Ouy8sk0cHPbfuYx(QJ196$D~-S_AO-g z8%+da#7<7Yei``!wFMpK}+pK!T*YcMuxYcN_I^RPN z@Q&7*W`a?+E4NW_Fz}D8D+f1-a^Wm_3X)m%tDs_Nao*yOsEe|tUE^mx&^O7{+){4O zxyf^fyU&*#=&f^hS#h(aH_unk++7Z9G(0#*-rT2K3<2PiqVQvEwia~84hL*&Yg_sX z{Hi&7INpP4IaOKPHmzcZAnloLZ`#T8lPC2hBc$R$a5!LIgDy9U(lqL`)Ycrs6EoM0 z=HRHN5!GKDM+MsgV>q@4b&DEhas)9_m5a@1)s<0UgY{mLQGrAX3~yLsWj!3mTI=th zSvKilhuIX%_N4~VqB|WPP-&y*d>4M|8b1sELtLyp4b~TylJsca?(VwQ=F;t8{Q#Ui z`mAloGm}f3WGK6!i8ul=ZqrhL-!HTuP!yWp;Z+s^X9gX1I+S~j?NhrSC@k{Y)a|9M zIBXrs^)99&vO}s>>Pkg2T!Sh*EenoD{a1>AnD;!~MK=q>6u2`kJMWn3u zV4>a0`muRvmmKUXdF#C&xwR>>;SjY#m*~sB(K5#75u|+A_b{Qp47-awH?FV}Qy~n| zzLtV4(4@a4$Pq3GUq_YR`_b#G*xm#e?(xrMm(8jFPR_y00y+5ACY8Cu5t*G87Wu?& ztRUALdO=OTk%ddH=+{!qC;^RctFeAuoVM$x-WvFR6CqR0UzdvsMuU_|JLR$K9^5xX zL#=4fr_1y>0=DE)5_$Kol;i^Jp`rZ3C;ryB5t5}MYR0GFm*7~|cnolpumLyrFikkd zkZ%WPvg1s*;>y2HNcMQj$2w_cqGG3;cPQ%zqiG`oUmvXAOjbI`)CjZVFxFVQoU9eR z8Oz~Q%b!jON}xa%Y^OG3{cR4P5~L<@;b(>qqufyTaAS-6+290h{An(Oxr8;a*vjhCYIRw?6iK^78yO3DW_?kLG6< zM+qb$9}L#z>eKHZchvXNchyL)FJ71IPz;h@qi^T_p!IC)di+`ZW+s(H4TPktEf7gM z8m;^jvP8>+4J!^@wZ=(}sVxufAPOzue!5+l`hK5KD-`_fve*X{9KoS%Ym(TWpO+L- z62CZTL}Lm-yQRfY5Nf4c5IM$>f%+d%(0c;9E<(@dQ5Rq+{xJT!MvmxouVloBU?30I|@NzQV+@vJ?Z_bZ=7mc$oPP51B`*4}; z4qxnm@KtrP`(`Dfm9!n^QCv0S(%dz{^2)|r9JR$U%NUv7#dOWgtO z4>^ztKRXqQufo>TAtO!3`Vc<*kZ+hlv(?JQrwh95nBBjR{((B|7z%?uN%q zEwnv)R^+M57aG>wtXJXdH`Q&`Wc6t`4Vj$g`~DIyoIS$q-O?XGtxbJi6GD$dVpntU zOIo^kxxh9dJ`eS|JZ9?~*AtI{;G-n8$?Fn?uGTwEqO;K%-*I6&YhP0heYhT(8GRm9 zxiJ$L@Z>R@Hv3r%?n{r)>I0%tJ}s&v^v|wTR^h#!g!F_tB)t{+y(LN3NfsL>e;}xt z{=z||HT1>{1AV8vuMpqDJfvOvOVo3R^Kk}E^Fxr$J(9qfBF6GZNwJ@=3)VyjzdJ%f4a6 zU;HnrSw`fWH_SPCscdg9$vj<%=R1WuspDihv$C?%^Ch z;4~LKMh?}yYNti3;SlVVmA~p>BaPK-M)@8s8Ch-6$mGO<eObg<+!?!c(G3a+ z%AEN5{*4-_3|L9STsP0QBRS=RNHOA-^_W;>plK_WXU{y@4AED=Fo zU+3bf--je-UiIS~Br{S4&nI@OdW%Gs)PICTf#ZKlzH8(U(2(zx%8*Ms5bdT8c?Rh4 zZRy_0A?@|niqO~0`k@JR&5wpGUw9b(c`)xB+t$W&29Y&eR`K>lL}8)`S&196QaQU>S&nMWy@-d)HHz=I5lTJ&h5KYzi;p_%$k4IC^kJcfTbJZ}U z{cJ6?Y`a-tbZdBF;H;fufWqlpUP#vm38RM|&=xVpuv>=Vn_S0$eH~xB*4&OxPlC6~ zF;IRdu;rSES$JWzPJuVQBgN3M<>KXDim1@bXCaliv^BItOeE}To*^~n7$8{YkXa&j z6o%W5fkqs^^~$Z zO=kjE<3BCjB&Yrf8whBM+!xaQc?^UkkjsKw&%ptu)rOF(xrG_d!mZ++D;Z1oDuNcI ziPwr~-g}vTWFG6`=dX6Kq3w0*X4A%Q^(jk3#07sik;OA8(?kh84|yE8MO1^Df` zgSix months ended September 30,20242025YoY % ChangeRMBRMBUS$(in millions, except percentages and per share amounts)Revenue479,739495,44769,5953%Income from operations71,23540,3535,668(43)% (2)Operating margin15%8%Adjusted EBITDA (1)98,48862,9918,848(36)% (3)Adjusted EBITDA margin (1)21%13%Adjusted EBITA (1)85,59647,9176,731(44)% (3)Adjusted EBITA margin (1)18%10%Net income67,56962,9948,849(7)% (4)Net income attributable to ordinary shareholders68,14364,1069,005(6)% (4)Non-GAAP net income (1)77,20943,8626,161(43)% (3)Diluted earnings per share (5)3.503.340.47(5)% (4)(6)Diluted earnings per ADS (5)28.0026.733.75(5)% (4)(6)Non-GAAP diluted earnings per share (1)(5)3.942.390.34(39)% (3)(6)Non-GAAP diluted earnings per ADS (1)(5)31.5019.102.68(39)% (3)(6) \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-1 Alibaba Revenue.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-1 Alibaba Revenue.html new file mode 100644 index 000000000..512fb8dfe --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-1 Alibaba Revenue.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025YoY % Change
RMBRMBUS$
(in millions, except percentages)
Alibaba China E-commerce Group:
E-commerce
- Customer management152,755168,17923,62410%
- Direct sales, logistics and others(2)50,23353,3317,4926%
202,988221,51031,1169%
Quick commerce(3)27,51737,6905,29437%
China commerce wholesale11,93813,4501,88913%
Total Alibaba China E-commerce Group242,443272,65038,29912%
Alibaba International Digital Commerce Group:
International commerce retail49,30956,4637,93115%
International commerce wholesale11,65613,0771,83712%
Total Alibaba International Digital Commerce Group60,96569,5409,76814%
Cloud Intelligence Group56,15973,22210,28530%
All others(4)165,837121,56817,077(27)%
Unallocated8881,096154
Inter-segment elimination(46,553)(42,629)(5,988)
Consolidated revenue479,739495,44769,5953%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-10 Board of Directors.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-10 Board of Directors.html new file mode 100644 index 000000000..ea46f5f65 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-10 Board of Directors.html @@ -0,0 +1 @@ +
Name(1)AgePosition/Title
Joseph C. TSAI $^{\dagger(a)}$ 61Chairman
Eddie Yongming WU $^{\dagger(b)}$ 50Director and Chief Executive Officer
J. Michael EVANS $^{\dagger(a)}$ 68Director and President
Maggie Wei WU $^{\dagger(c)}$ 57Director $^{(d)}$
Jerry YANG $^{(b)}$ 57Independent director
Wan Ling MARTELLO $^{(b)}$ 67Independent director
Weijian SHAN $^{(c)}$ 72Independent director
Irene Yun-Lien LEE $^{(a)}$ 72Independent director
Albert Kong Ping NG $^{(b)}$ 68Independent director
Kabir MISRA $^{(c)}$ 56Independent director
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-11 Shareholding Table.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-11 Shareholding Table.html new file mode 100644 index 000000000..3a8c0fc2d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-11 Shareholding Table.html @@ -0,0 +1 @@ +
Name of director / chief executive officerCapacity / nature of interestNumber of shares/Approximate % of shareholding (3)
underlying shares (1),(2)
(in the number of Shares)(in the number of ADSs)
Joseph C. TSAI (4)Beneficial owner881,072 (L)110,134 (L)0.00%
Interest of spouse1,280,000 (L)160,000 (L)0.01%
Founder of a discretionary trust who can influence how the trustee exercises his discretion12,658,152 (L)1,582,269 (L)0.06%
Interest in controlled corporation260,924,840 (L)32,615,605 (L)1.37%
275,744,064 (L)34,468,008 (L)1.44%
Eddie Yongming WU (5)Beneficial owner18,199,752 (L)2,274,969 (L)0.10%
Interest of spouse108,000 (L)13,500 (L)0.00%
Founder of a discretionary trust who can influence how the trustee exercises his discretion12,320,000 (L)1,540,000 (L)0.06%
30,627,752 (L)3,828,469 (L)0.16%
J. Michael EVANS (6)Beneficial owner9,964,000 (L)1,245,500 (L)0.05%
Maggie Wei WU (7)Beneficial owner3,661,632 (L)457,704 (L)0.02%
Founder of a discretionary trust who can influence how the trustee exercises his discretion7,200,000 (L)900,000 (L)0.04%
10,861,632 (L)1,357,704 (L)0.06%
Jerry YANGBeneficial owner485,072 (L)60,634 (L)0.00%
Wan Ling MARTELLOBeneficial owner360,000 (L)45,000 (L)0.00%
Weijian SHANBeneficial owner144,800 (L)18,100 (L)0.00%
Irene Yun-Lien LEEBeneficial owner133,600 (L)16,700 (L)0.00%
Albert Kong Ping NGBeneficial owner133,600 (L)16,700 (L)0.00%
Kabir MISRABeneficial owner528,800 (L)66,100 (L)0.00%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-12 Shareholder Holdings.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-12 Shareholder Holdings.html new file mode 100644 index 000000000..40263f88d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-12 Shareholder Holdings.html @@ -0,0 +1 @@ +
Name of shareholderCapacity / nature of interestNumber of shares/ underlying shares (1),(2)Approximate % of shareholding (3)
(in the number of Shares)(in the number of ADSs)
JPMorgan Chase & Co. (4)Beneficial owner411,801,958 (L)51,475,245 (L)2.16%
418,739,874 (S)52,342,484 (S)2.19%
Trustee1,753,320 (L)219,165 (L)0.01%
Investment manager281,850,861 (L)35,231,358 (L)1.47%
6,527,284 (S)815,911 (S)0.04%
Person having a security interest in shares85,927,145 (L)10,740,893 (L)0.45%
Approved lending agent581,658,439 (L)72,707,305 (L)3.05%
581,658,439 (P)72,707,305 (P)3.05%
1,362,991,723 (L)170,373,966 (L)7.14%
425,267,158 (S)53,158,395 (S)2.23%
581,658,439 (P)72,707,305 (P)3.05%
Citigroup Inc. (5)Interest in controlled corporation183,474,400 (L)22,934,300 (L)0.96%
197,906,617 (S)24,738,327 (S)1.04%
Person having a security interest in shares160 (L)20 (L)0.00%
Approved lending agent832,721,084 (L)104,090,136 (L)4.36%
832,721,084 (P)104,090,136 (P)4.36%
1,016,195,644 (L)127,024,456 (L)5.32%
197,906,617 (S)24,738,327 (S)1.04%
832,721,084 (P)104,090,136 (P)4.36%
BlackRock, Inc. (6)Interest in controlled corporation983,512,344 (L)122,939,043 (L)5.15%
3,297,800 (S)412,225 (S)0.02%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-13 RSU Grant Summary.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-13 RSU Grant Summary.html new file mode 100644 index 000000000..11b47edd8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-13 RSU Grant Summary.html @@ -0,0 +1 @@ +
Name of granteeDate of grantNumber of RSUs (1), (2)Number of underlying Shares as of September 30, 2025 (2)Notes
Outstanding as of April 1, 2025Granted during the Reporting PeriodVested during the Reporting PeriodLapsed/ forfeited during the Reporting PeriodCancelled during the Reporting PeriodOutstanding as of September 30, 2025
Directors and chief executive officer of our Company
Joseph C. TSAIJuly 24, 2018 to August 16, 20191,334-1,334----Notes 3, 8
Eddie Yongming WUNovember 25, 2023160,000-40,000--120,000960,000Notes 4, 8
J. Michael EVANSJune 15, 2020 to May 20, 202310,050-4,550--5,50044,000Notes 5, 8
May 13, 2024160,00080,000-80,000640,000Notes 6, 8
Maggie Wei WUJuly 24, 2018 to May 24, 202112,801-6,766--6,03548,280Notes 3, 8
Other grantees by category
Employee participantsNovember 15, 2016 to August 18, 202444,402,154-16,110,6831,681,3791,82526,608,267212,866,136Notes 7, 9
Related entity $participants^{(11)}$ July 24, 2018 to June 8, 2022113,139-84,7672,578-25,794206,352Notes 7, 10
Total44,859,478-16,328,1001,683,9571,82526,845,596214,764,768
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-14 Share Options Table.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-14 Share Options Table.html new file mode 100644 index 000000000..3d43edfea --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-14 Share Options Table.html @@ -0,0 +1 @@ +
Name of granteeDate of grantExercise price (US$)Number of share options (1)Number of underlying Shares as of September 30, 2025 (1)Notes
Outstanding as of April 1, 2025Granted during the Reporting PeriodExercised during the Reporting PeriodLapsed/ forfeited during the Reporting Period (9)Outstanding as of September 30, 2025
Directors and chief executive officer of our Company
Eddie Yongming WUNovember 25, 202378.372,000,000---2,000,00016,000,000Note 2
J. Michael EVANSJuly 31, 201579.962,000,000-1,000,000-1,000,0008,000,000Note 3, 6
May 13, 202484.60150,000---150,0001,200,000Note 4
Other grantees by category
Employee participantsMay 10, 2015 to March 12, 202223.00 to 182.482,385,667-842,6671,000,000543,0004,344,000Notes 5, 7
Related entity participant (10)March 12, 202223.00100,000-100,000---Notes 5, 8
Total6,635,667-1,942,6671,000,0003,693,00029,544,000
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-15 RSU Activity Report.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-15 RSU Activity Report.html new file mode 100644 index 000000000..b2d56655f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-15 RSU Activity Report.html @@ -0,0 +1 @@ +
Name (3)Date of grantNumber of RSUs (1),(2)Number of underlying Shares as of September 30, 2025 (2)Notes
Outstanding as of April 1, 2025Granted during the Reporting PeriodVested during the Reporting PeriodLapsed/ forfeited during the Reporting Period (8)Outstanding as of September 30, 2025
Employee participantsNovember 19, 2024 and February 24, 20259,435,535-1,456,534236,2007,742,8017,742,801Notes 4, 5
May 23, 2025, June 13, 2025 and September 4, 2025-63,594,3183,209,7961,116,59959,267,92359,267,923Notes 5, 6, 7
Total9,435,53563,594,3184,666,3301,352,79967,010,72467,010,724
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-16 RSU Grant Details.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-16 RSU Grant Details.html new file mode 100644 index 000000000..ddd416e59 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-16 RSU Grant Details.html @@ -0,0 +1 @@ +
Date of grantNumber of RSUs grantedNumber of underlying SharesClosing price of sharesimmediately before date of grant(HK$)Fair value of each RSU as of dateof grant (HK$)
May 23, 202553,429,59753,429,597119.10118.78
June 13, 20257,687,0917,687,091114.60114.60
September 4, 20252,477,6302,477,630134.10130.31
63,594,31863,594,318
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-17 Share Repurchase.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-17 Share Repurchase.html new file mode 100644 index 000000000..043c75cf5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-17 Share Repurchase.html @@ -0,0 +1 @@ +
Month of repurchaseNumber of Shares underlying ADSs repurchased (1)Highest price paid (US$)Lowest price paid (US$)Aggregate consideration paid (US$, in millions)
April 202530,630,48016.7611.97427
May 202513,559,32016.7914.02210
June 202511,685,19215.0013.92168
July 202511,019,16015.0012.96154
August 20255,839,33615.0014.5287
Total72,733,4881,046
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-18 Notes Proceeds Use.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-18 Notes Proceeds Use.html new file mode 100644 index 000000000..02d300473 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-18 Notes Proceeds Use.html @@ -0,0 +1 @@ +
General Corporate purposesNet proceeds from the September 2025 Notes Offering (US$ in millions)Amount of net proceeds utilized during the reporting period (US$ in millions)Actual use of proceeds up to September 30, 2025 (US$ in millions)
Enhancing its cloud infrastructure2,504443443
Expanding international commerce operations626287287
Total3,130730730
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-19 Interim Report.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-19 Interim Report.html new file mode 100644 index 000000000..6a974aca7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-19 Interim Report.html @@ -0,0 +1 @@ +
Page
Report on Review of Interim Financial Information for the six-months ended September 30, 202545
Unaudited Condensed Consolidated Income Statements for the six months ended September 30, 2024 and 202546
Unaudited Condensed Consolidated Statements of Comprehensive Income for the six months ended September 30, 2024 and 202547
Unaudited Condensed Consolidated Balance Sheets as of March 31, 2025 and September 30, 202548
Unaudited Condensed Consolidated Statements of Changes in Shareholders’ Equity for the six months ended September 30, 2024 and 202550
Unaudited Condensed Consolidated Statements of Cash Flows for the six months ended September 30, 2024 and 202552
Notes to Unaudited Condensed Consolidated Financial Statements53
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-2 Alibaba EBITA.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-2 Alibaba EBITA.html new file mode 100644 index 000000000..73266fa12 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-2 Alibaba EBITA.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025YoY % Change(3)
RMBRMBUS$
(in millions, except percentages)
Alibaba China E-commerce Group93,08048,8866,867(47)%
Alibaba International Digital Commerce Group(6,611)10314N/A
Cloud Intelligence Group4,9986,55892131%
All others(2,910)(4,785)(672)(64)%
Unallocated(2)(2,142)(1,640)(230)
Inter-segment elimination(819)(1,205)(169)
Consolidated adjusted EBITA85,59647,9176,731(44)%
Less: Non-cash share-based compensation expense(7,775)(6,076)(854)
Less: Amortization and impairment of intangible assets, and others(3,441)(1,488)(209)
Less: Provision for the shareholder class action lawsuits(3,145)--
Income from operations71,23540,3535,668(43)%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-20 Alibaba Financials.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-20 Alibaba Financials.html new file mode 100644 index 000000000..6853608ff --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-20 Alibaba Financials.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMBUS$ (Note 2(a))
(in millions, except per share data)
Notes
Revenue4,16479,739495,44769,595
Cost of revenue16(290,135)(287,210)(40,344)
Product development expenses16(27,555)(32,096)(4,509)
Sales and marketing expenses16(65,167)(119,674)(16,811)
General and administrative expenses16(23,057)(14,778)(2,076)
Amortization and impairment of intangible assets(3,441)(1,633)(229)
Other gains, net85129742
Income from operations71,23540,3535,668
Interest and investment income, net17,12937,4685,263
Interest expense(4,615)(4,995)(702)
Other (expense) income, net16(1,221)1,329187
Income before income tax and share of results of equity method investees82,52874,15510,416
Income tax expenses5(17,442)(14,415)(2,024)
Share of results of equity method investees2,4833,254457
Net income67,56962,9948,849
Net loss (income) attributable to noncontrolling interests854(1,326)(187)
Net income attributable to Alibaba Group Holding Limited68,42361,6688,662
(Accretion) Reversal of accretion of mezzanine equity(280)2,438343
Net income attributable to ordinary shareholders68,14364,1069,005
Earnings per share attributable to ordinary shareholders7
Basic3.583.450.49
Diluted3.503.340.47
Earnings per ADS attributable to ordinary shareholders (one ADS equals eight ordinary shares)7
Basic28.6227.633.88
Diluted28.0026.733.75
Weighted average number of shares used in computing earnings per share (million shares)7
Basic19,04518,562
Diluted19,45919,154
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-21 Income Statement.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-21 Income Statement.html new file mode 100644 index 000000000..282a955f5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-21 Income Statement.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMBUS$ (Note 2(a))
(in millions)
Net income67,56962,9948,849
Other comprehensive income (loss):
- Foreign currency translation:
Change in unrealized losses, net of tax(3,699)(5,081)(714)
- Share of other comprehensive income (loss) of equity method investees:
Change in unrealized gains (losses)350(40)(6)
- Interest rate swaps under hedge accounting and others:
Change in unrealized gains601
Other comprehensive loss(3,289)(5,120)(720)
Total comprehensive income64,28057,8748,129
Total comprehensive loss (income) attributable to noncontrolling interests1,119(273)(38)
Total comprehensive income attributable to ordinary shareholders65,39957,6018,091
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-22 Balance Sheet Data.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-22 Balance Sheet Data.html new file mode 100644 index 000000000..245ce7fed --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-22 Balance Sheet Data.html @@ -0,0 +1 @@ +
As of March 31,As of September 30,
20252025
RMBRMBUS$
(in millions)(Note 2(a))
AssetsNotes
Current assets:
Cash and cash equivalents145,487135,06918,973
Short-term investments228,826193,24627,145
Restricted cash and escrow receivables43,78140,3745,671
Equity securities and other investments853,78045,2576,357
Prepayments, receivables and other assets202,175232,67332,684
Total current assets674,049646,61990,830
Equity securities and other investments8356,818411,95357,867
Prepayments, receivables and other assets83,43196,92713,615
Investments in equity method investees210,169206,86229,058
Property and equipment, net203,348246,53934,631
Intangible assets, net1020,91119,4292,729
Goodwill11255,501255,55135,897
Total assets1,804,2271,883,880264,627
Liabilities, mezzanine equity and shareholders' equity
Current liabilities:
Current bank borrowings22,56226,2883,693
Income tax payable11,6385,588785
Accrued expenses, accounts payable and other liabilities12332,537340,76947,868
Merchant deposits27425135
Deferred revenue and customer advances68,33571,24110,007
Total current liabilities435,346444,13762,388
Deferred revenue4,5364,496632
Deferred tax liabilities48,45446,8026,574
Non-current bank borrowings49,90963,5668,929
Non-current unsecured senior notes122,398120,50416,927
Non-current convertible unsecured senior notes1335,83457,4818,074
Non-current exchangeable bonds1413,7551,932
Other liabilities1217,64421,3543,000
Total liabilities714,121772,095108,456
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-23 Equity Summary.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-23 Equity Summary.html new file mode 100644 index 000000000..3c134608b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-23 Equity Summary.html @@ -0,0 +1 @@ +
As of March 31,As of September 30,
20252025
RMBRMBUS$ (Note 2(a))
(in millions)
Commitments and contingencies
Mezzanine equity11,7139,8841,388
Shareholders’ equity:
Ordinary shares, US$0.000003125 par value; 32,000,000,000 shares authorized as of March 31 and September 30, 2025; 18,474,235,708 and 18,551,943,896 shares issued and outstanding as of March 31 and September 30, 2025 respectively11
Additional paid-in capital381,379387,14754,382
Treasury shares, at cost(36,329)(36,162)(5,080)
Statutory reserves15,93616,2862,288
Accumulated other comprehensive income (loss)
Cumulative translation adjustments3,286(1,670)(234)
Unrealized gains on interest rate swaps and others10710915
Retained earnings645,478666,78493,663
Total shareholders’ equity1,009,8581,032,495145,034
Noncontrolling interests68,53569,4069,749
Total equity1,078,3931,101,901154,783
Total liabilities, mezzanine equity and equity1,804,2271,883,880264,627
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-24 Equity Changes.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-24 Equity Changes.html new file mode 100644 index 000000000..ce168b407 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-24 Equity Changes.html @@ -0,0 +1 @@ +
Ordinary sharesAccumulated othercomprehensive income (loss)
ShareAmountAdditionalpaid-incapitalRMBTreasurysharesRMBStatutoryreservesRMBCumulativetranslationadjustmentsRMBUnrealizedgains(losses) oninterestrate swapsand othersRMBRetainedearningsRMBTotalshareholders’equityRMBNoncontrollinginterestsRMBTotalequityRMB
(in millions, except per share data)
Balance as ofApril 1, 202419,469,126,9561397,999(27,684)14,7333,635(37)597,897986,544115,3271,101,871
Foreign currencytranslationadjustment,net of tax(3,541)(3,541)(158)(3,699)
Share of additionalpaid-in capitaland othercomprehensiveincome of equitymethod investees(42)33515308308
Change in fair value ofinterest rate swapsunder hedgeaccountingand others606060
Net income for theperiod68,42368,423(961)67,462
Acquisition ofsubsidiaries187187
Issuance of shares,including vesting ofRSUs and earlyexercised optionsand exercise ofshare options178,297,456
Repurchase andretirement ofordinary shares(1,027,554,280)(21,096)(8,501)(42,216)(71,813)(71,813)
Transactions withnoncontrollinginterests2,3572,357(23,852)(21,495)
Amortization ofcompensation cost5,8195,8191,9627,781
Declaration ofdividends(29,340)(29,340)(29,340)
Capped calltransactions(4,612)(4,612)(4,612)
Appropriation tostatutory reserves1,152(1,152)
Others(280)(280)(144)(424)
Balance as ofSeptember 30, 2024(in millions, except per share data)18,619,870,1321380,145(36,185)15,88542938593,612953,92592,3611,046,286
Balance as ofApril 1, 202518,474,235,7081381,379(36,329)15,9363,286107645,4781,009,85868,5351,078,393
Foreign currencytranslationadjustment,net of tax(3,643)7(3,636)(166)(3,802)
Share of additionalpaid-in capitaland othercomprehensiveincome of equitymethod investees(129)(34)(6)(169)(169)
Change in fair value ofinterest rate swapsunder hedgeaccounting and others111
Net investment hedge(1,279)(1,279)(1,279)
Net income for theperiod61,66861,66843962,107
Acquisition ofsubsidiaries22
Deconsolidation ofsubsidiaries(80)(80)
Issuance of shares,including vesting ofRSUs and earlyexercised optionsand exercise ofshare options150,441,6761,0381,0381,038
Repurchase andretirement ofordinary shares(72,733,488)(1,737)167(5,965)(7,535)(7,535)
Transactions withnoncontrollinginterests6868(315)(247)
Amortization ofcompensation cost5,3975,3971,0756,472
Declaration ofdividends(34,047)(34,047)(34,047)
Capped calltransactions(1,309)(1,309)(1,309)
Appropriation tostatutory reserves350(350)
Others2,4402,440(84)2,356
Balance as ofSeptember 30, 202518,551,943,8961387,147(36,162)16,286(1,670)109666,7841,032,49569,4061,101,901
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-25 Cash Flow Statement.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-25 Cash Flow Statement.html new file mode 100644 index 000000000..06f8d2c9e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-25 Cash Flow Statement.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)(Note 2(a))
Cash flows from operating activities:65,07430,7714,322
Cash flows from investing activities:
Decrease in short-term investments, net105,47012,9721,822
Increase in other treasury investments, net(113,387)(24,091)(3,384)
Settlement of forward exchange contracts, net311(111)(16)
Acquisitions of equity and debt securities, and others(3,905)(18,690)(2,625)
Disposals of equity and debt securities, and others5,77235,3904,971
Acquisitions of equity method investees(1,964)(1,179)(165)
Disposals of and distributions from equity method investees7372,426341
Acquisitions of land use rights, property and equipment(29,585)(70,177)(9,858)
Disposals of property and equipment1,47523934
Cash received (paid) for business combinations, net of cash acquired62(62)(9)
Deconsolidation and disposal of subsidiaries, net of cash proceeds11,8251,661
Loans to employees, net of repayments14913419
Net cash used in investing activities(34,865)(51,324)(7,209)
Cash flows from financing activities:
Issuance of ordinary shares11,038146
Repurchase of ordinary shares(72,889)(7,638)(1,073)
Dividend distribution(29,022)(33,621)(4,723)
Proceeds from convertible unsecured senior notes, net of debt issuance cost35,67722,3033,133
Payments for capped call transactions(4,612)(1,309)(184)
Acquisition of additional equity interests in non-wholly owned subsidiaries(19,947)(1,155)(162)
Dividends paid by non-wholly owned subsidiaries to noncontrolling interests(116)(306)(43)
Contingent consideration payments made after a business combination(149)(81)(11)
Capital injection from noncontrolling interests2,63616824
Proceeds from exchangeable bonds, net of debt issuance cost10,9861,543
Proceeds from bank borrowings and other borrowings, net of upfront fee payment for a syndicated loan22,09837,8055,310
Repayment of bank borrowings(20,041)(20,019)(2,812)
Net cash (used in) provided by financing activities(86,364)8,1711,148
Effect of exchange rate changes on cash and cash equivalents, restricted cash and escrow receivables(1,797)(1,443)(203)
Decrease in cash and cash equivalents, restricted cash and escrow receivables(57,952)(13,825)(1,942)
Cash and cash equivalents, restricted cash and escrow receivables at beginning of period286,424189,26826,586
Cash and cash equivalents, restricted cash and escrow receivables at end of period228,472175,44324,644
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-26 Financial Data.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-26 Financial Data.html new file mode 100644 index 000000000..492cbd0b2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-26 Financial Data.html @@ -0,0 +1 @@ +
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
Cash and cash equivalents and short-term investments10,6218,807
Investments in equity method investees and equity securities and other investments37,11735,335
Accounts receivable and contract assets, net of allowance18,40818,666
Amounts due from non-VIE subsidiaries of the Company53,79251,309
Property and equipment, net and intangible assets, net19,91137,527
Others40,42956,590
Total assets180,278208,234
Amounts due to non-VIE subsidiaries of the Company113,332131,635
Accrued expenses, accounts payable and other liabilities50,52160,882
Deferred revenue and customer advances19,12618,669
Total liabilities182,979211,186
Six months ended September 30,
20242025
RMBRMB
(in millions)
Revenue (i)67,73977,466
Net income2,1032,442
Net cash provided by operating activities12,1182,817
Net cash used in investing activities(28,993)(31,914)
Net cash provided by financing activities8,62527,034
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-27 Alibaba Revenue.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-27 Alibaba Revenue.html new file mode 100644 index 000000000..0832687a7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-27 Alibaba Revenue.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Alibaba China E-commerce Group:
E-commerce (i)
- Customer management152,755168,179
- Direct sales, logistics and others (ii)50,23353,331
202,988221,510
Quick commerce (iii)27,51737,690
China commerce wholesale (iv)11,93813,450
Total Alibaba China E-commerce Group:242,443272,650
Alibaba International Digital Commerce Group:
International commerce retail (v)49,30956,463
International commerce wholesale (vi)11,65613,077
Total Alibaba International Digital Commerce Group60,96569,540
Cloud Intelligence Group (vii)56,15973,222
All others (viii)165,837121,568
Unallocated8881,096
Inter-segment elimination (ix)(46,553)(42,629)
Consolidated revenue479,739495,447
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-28 Revenue Breakdown.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-28 Revenue Breakdown.html new file mode 100644 index 000000000..ab9ecc6a4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-28 Revenue Breakdown.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Customer management services (i)199,821223,399
Membership fees and value-added services23,04424,035
Logistics services59,90468,018
Cloud services41,32652,826
Sales of goods135,837108,435
Other revenue (ii)19,80718,734
479,739495,447
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-29 Tax Expense Data.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-29 Tax Expense Data.html new file mode 100644 index 000000000..a6cda3180 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-29 Tax Expense Data.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Current income tax expense14,93913,985
Deferred taxation2,503430
17,44214,415
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-3 Costs & Expenses.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-3 Costs & Expenses.html new file mode 100644 index 000000000..57a67a9bd --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-3 Costs & Expenses.html @@ -0,0 +1 @@ +
Six months ended September 30,% of Revenue YoY change
20242025
RMB% of RevenueRMBUS$% of Revenue
(in millions, except percentages)
Costs and expenses:
Cost of revenue290,13560.5%287,21040,34458.0%(2.5)%
Product development expenses27,5555.7%32,0964,5096.5%0.8%
Sales and marketing expenses65,16713.6%119,67416,81124.2%10.6%
General and administrative expenses23,0574.8%14,7782,0763.0%(1.8)%
Amortization and impairment of intangible assets3,4410.7%1,6332290.3%(0.4)%
Total costs and expenses409,355455,39163,969
Share-based compensation expense:
Cost of revenue1,2050.3%9131280.2%(0.1)%
Product development expenses3,5600.7%2,8624020.6%(0.1)%
Sales and marketing expenses9480.2%9581350.2%0.0%
General and administrative expenses2,5640.5%2,1373000.4%(0.1)%
Total share-based compensation expense(1)8,2776,870965
Costs and expenses excluding share-based compensation expense:
Cost of revenue288,93060.2%286,29740,21657.8%(2.4)%
Product development expenses23,9955.0%29,2344,1075.9%0.9%
Sales and marketing expenses64,21913.4%118,71616,67624.0%10.6%
General and administrative expenses20,4934.3%12,6411,7762.6%(1.7)%
Amortization and impairment of intangible assets3,4410.7%1,6332290.3%(0.4)%
Total costs and expenses excluding share-based compensation expense401,078448,52163,004
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-30 Cost Breakdown.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-30 Cost Breakdown.html new file mode 100644 index 000000000..a935122e0 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-30 Cost Breakdown.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Cost of revenue1,205913
Product development expenses3,5602,862
Sales and marketing expenses948958
General and administrative expenses2,5642,137
8,2776,870
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-31 EPS Data Table.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-31 EPS Data Table.html new file mode 100644 index 000000000..195b1cd16 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-31 EPS Data Table.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMB
(in millions, except per share data)
Earnings per share
Numerator:
Net income attributable to ordinary shareholders for computing net income per ordinary share — basic68,14364,106
Dilution effect on earnings arising from non-cash share-based awards operated by equity method investees and subsidiaries(131)(258)
Adjustments for interest expense attributable to convertible unsecured senior notes95143
Net income attributable to ordinary shareholders for computing net income per ordinary share — diluted68,10763,991
Shares (denominator):
Weighted average number of shares used in calculating net income per ordinary share — basic (million shares)19,04518,562
Adjustments for dilutive RSUs and share options (million shares)149186
Adjustments for convertible unsecured senior notes (million shares)265406
Weighted average number of shares used in calculating net income per ordinary share — diluted (million shares)19,45919,154
Net income per ordinary share — basic (RMB)3.583.45
Net income per ordinary share — diluted (RMB)3.503.34
Earnings per ADS
Net income per ADS — basic (RMB)28.6227.63
Net income per ADS — diluted (RMB)28.0026.73
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-32 Investment Summary.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-32 Investment Summary.html new file mode 100644 index 000000000..7344192bb --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-32 Investment Summary.html @@ -0,0 +1 @@ +
As of March 31, 2025
Original costCumulative net (losses) gainsCarrying value
RMBRMBRMB
(in millions)
Equity securities:
Listed equity securities79,024(1,394)77,630
Investments in privately held companies113,646(17,479)96,167
Debt investments:
Debt securities and loan investments15,009(6,224)8,785
Other treasury investments227,93581228,016
435,614(25,016)410,598
As of September 30, 2025
Original costCumulative net gains (losses)Carrying value
RMBRMBRMB
(in millions)
Equity securities:
Listed equity securities70,8804,42475,304
Investments in privately held companies120,899(1,279)119,620
Debt investments:
Debt securities and loan investments14,892(5,173)9,719
Other treasury investments252,464103252,567
459,135(1,925)457,210
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-33 Fair Value Table.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-33 Fair Value Table.html new file mode 100644 index 000000000..e801031c4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-33 Fair Value Table.html @@ -0,0 +1 @@ +
As of March 31, 2025
Level 1RMBLevel 2RMBLevel 3RMBTotalRMB
(in millions)
Assets
Time deposits and certificate of deposits (i)357,569357,569
Wealth management products (i)83,14483,144
Marketable debt securities (i)16,12916,129
Restricted cash and escrow receivables43,78143,781
Listed equity securities (ii)(v)67,7129,91877,630
Convertible and exchangeable bonds (ii)145818963
Option agreements (iii)87814901
Deferred consideration (iii)3,0393,039
Others (vi)1,3296,0477,376
111,493468,32110,718590,532
Liabilities
Contingent consideration in relation to investments and acquisitions (iv)484484
Others (iv)3899241,313
3891,4081,797
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-34 Fair Value Table.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-34 Fair Value Table.html new file mode 100644 index 000000000..5f54f9318 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-34 Fair Value Table.html @@ -0,0 +1 @@ +
As of September 30, 2025
Level 1RMBLevel 2RMBLevel 3RMBTotalRMB
(in millions)
Assets
Time deposits and certificate of deposits (i)314,794314,794
Wealth management products (i)84,78984,789
Marketable debt securities (i)46,23046,230
Restricted cash and escrow receivables40,37440,374
Listed equity securities (ii)(v)65,3279,97775,304
Convertible and exchangeable bonds (ii)139620759
Option and forward agreements (iii)3,4781,1544,632
Deferred consideration (iii)3,0523,052
Others (vi)8455,6276,472
105,701460,25210,453576,406
Liabilities
Exchangeable bonds13,75513,755
Contingent consideration in relation to investments and acquisitions (iv)397397
Others (iv)1,1229072,029
14,8771,30416,181
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-35 Balance Change.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-35 Balance Change.html new file mode 100644 index 000000000..ddf928675 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-35 Balance Change.html @@ -0,0 +1 @@ +
Amounts RMB (in millions)
Balance as of April 1, 2025818
Additions42
Net decrease in fair value(158)
Disposal(39)
Conversion(34)
Foreign currency translation adjustments(9)
Balance as of September 30, 2025620
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-36 Fair Value Change.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-36 Fair Value Change.html new file mode 100644 index 000000000..727385dd5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-36 Fair Value Change.html @@ -0,0 +1 @@ +
Amounts RMB (in millions)
Balance as of April 1, 20253,039
Net increase in fair value40
Foreign currency translation adjustments(27)
Balance as of September 30, 20253,052
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-37 Intangible Assets.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-37 Intangible Assets.html new file mode 100644 index 000000000..0cdc839ab --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-37 Intangible Assets.html @@ -0,0 +1 @@ +
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
User base and customer relationships48,56540,420
Trade names, trademarks and domain names26,93627,256
Non-compete agreements6,0305,745
Developed technology and patents4,8235,136
Licensed copyrights and others8,0018,585
94,35587,142
Less: accumulated amortization and impairment(73,444)(67,713)
Net book value20,91119,429
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-38 Alibaba Segment Assets.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-38 Alibaba Segment Assets.html new file mode 100644 index 000000000..00eb05066 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-38 Alibaba Segment Assets.html @@ -0,0 +1 @@ +
AlibabaChinaE-commerceGroupRMBAlibabaInternationalDigitalCommerceGroupRMBCloudIntelligenceGroupRMBCainiaoSmartLogisticsNetworkLimitedRMBLocalServicesGroupRMBHujingDigital MediaandEntertainmentGroupRMBAll othersRMBTotalRMB
(in millions)
Balance as of April 1, 2025164,94520,0773,72820,38520,4478,03117,888255,501
Segment changes12,948(20,385)(20,447)(8,031)35,915
Additions435140
Foreign currency translationadjustments(42)46610
Balance as of September 30, 2025177,89320,0393,80953,810255,551
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-39 Financial Liabilities.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-39 Financial Liabilities.html new file mode 100644 index 000000000..2c03c5cc3 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-39 Financial Liabilities.html @@ -0,0 +1 @@ +
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
Accrued expenses, accounts payable and other liabilities1,6053,152
Bank borrowings4,4704,047
6,0757,199
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-4 Share-based Compensation.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-4 Share-based Compensation.html new file mode 100644 index 000000000..9911cc7be --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-4 Share-based Compensation.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025YoY % Change
RMBRMBUS$
(in millions, except percentages)
By type of awards:
Alibaba Group share-based awards(1)5,8774,664655(21)%
Others(2)2,4002,206310(8)%
Total share-based compensation expense(3)8,2776,870965(17)%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-40 Cloud Revenue Growth.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-40 Cloud Revenue Growth.html new file mode 100644 index 000000000..7879bd334 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-40 Cloud Revenue Growth.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Amounts earned by the Company
Cloud services revenue (i)4,9778,928
Marketplace software technology services fee and other amounts earned (i)3,5142,819
8,49111,747
Amounts incurred by the Company
Payment processing and escrow services fee (ii)6,3818,937
Other amounts incurred (i)1,4492,378
7,83011,315
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-41 Alibaba Segment Results.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-41 Alibaba Segment Results.html new file mode 100644 index 000000000..4ffe9259d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-41 Alibaba Segment Results.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Alibaba China E-commerce Group
Revenue242,443272,650
Costs and expenses (i)(149,363)(223,764)
Adjusted EBITA (ii)93,08048,886
Alibaba International Digital Commerce Group
Revenue60,96569,540
Costs and expenses (i)(67,576)(69,437)
Adjusted EBITA (ii)(6,611)103
Cloud Intelligence Group
Revenue56,15973,222
Costs and expenses (i)(51,161)(66,664)
Adjusted EBITA (ii)4,9986,558
All others
Revenue165,837121,568
Costs and expenses (i)(168,747)(126,353)
Adjusted EBITA (ii)(2,910)(4,785)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-42 Financial Results.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-42 Financial Results.html new file mode 100644 index 000000000..21ecf552d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-42 Financial Results.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Total segments Adjusted EBITA88,55750,762
Unallocated (iii)(2,142)(1,640)
Inter-segment elimination(819)(1,205)
Non-cash share-based compensation expense(7,775)(6,076)
Amortization and impairment of intangible assets, and others(3,441)(1,488)
Provision for the shareholder class action lawsuits(3,145)
Consolidated income from operations71,23540,353
Interest and investment income, net17,12937,468
Interest expense(4,615)(4,995)
Other (expense) income, net(1,221)1,329
Income tax expenses(17,442)(14,415)
Share of results of equity method investees2,4833,254
Consolidated net income67,56962,994
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-43 Depreciation by Segment.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-43 Depreciation by Segment.html new file mode 100644 index 000000000..aeb740f8b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-43 Depreciation by Segment.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMB
(in millions)
Alibaba China E-commerce Group7962
Alibaba International Digital Commerce Group547370
Cloud Intelligence Group6,90611,877
All others4,0871,226
Total segments depreciation and impairment of property and equipment, and operating lease cost relating to land use rights11,61913,535
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-44 Accounts Receivable.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-44 Accounts Receivable.html new file mode 100644 index 000000000..b021be1d9 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-44 Accounts Receivable.html @@ -0,0 +1 @@ +
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
0-3 months25,17225,073
3-6 months3,0783,395
6-12 months1,7752,344
Over 1 year627834
Accounts receivable, net of allowance30,65231,646
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-45 Accounts Payable.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-45 Accounts Payable.html new file mode 100644 index 000000000..28bd0e3dd --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-45 Accounts Payable.html @@ -0,0 +1 @@ +
As of March 31,As of September 30,
20252025
RMBRMB
(in millions)
0-3 months52,01956,073
3-6 months3,9905,696
6-12 months8461,600
Over 1 year1,3461,570
Accounts payable58,20164,939
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-46 IFRS Adjustments.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-46 IFRS Adjustments.html new file mode 100644 index 000000000..6b7230900 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-46 IFRS Adjustments.html @@ -0,0 +1 @@ +
As of March 31, 2025
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity’s own equity (viii)Amounts under IFRS
(in millions of RMB)
Equity securities and other investments356,8185,0033,547365,368
Prepayments, receivables and other assets83,431(258)485578(1,010)3826,87490,482
Investments in equity method investees210,169(241)(226)209,702
Property and equipment, net203,3481,318204,666
Intangible assets, net20,911(1,861)42619,476
Goodwill255,501(36,284)3,180222,397
Total assets1,804,227(38,644)5,4883,321578(1,010)5,3066,8741,786,140
Deferred tax liabilities48,454(719)368(283)(78)(51)47,691
Convertible unsecured senior notes35,83416,02151,855
Other liabilities17,644(2)10,24727,889
Total liabilities714,121(719)368(283)(78)(2)10,247(51)16,021739,624
Mezzanine equity11,713(11,713)
Total shareholders’ equity1,009,858(17,402)5,1203,604656(1,008)1,4392,616(9,147)995,736
Noncontrolling interests68,535(20,523)272,74150,780
Total equity1,078,393(37,925)5,1203,604656(1,008)1,4665,357(9,147)1,046,516
Total liabilities, mezzanine equity and equity1,804,227(38,644)5,4883,321578(1,010)5,3066,8741,786,140
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-47 GAAP to IFRS.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-47 GAAP to IFRS.html new file mode 100644 index 000000000..314f46fcc --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-47 GAAP to IFRS.html @@ -0,0 +1 @@ +
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity's own equity (viii)Amounts under IFRS
(in millions of RMB)
Equity securities and other investments411,9533,2824,223419,458
Prepayments, receivables and other assets96,927(258)521637(974)1,0238,315106,191
Investments in equity method investees206,862(241)(784)205,837
Property and equipment, net246,539699247,238
Intangible assets, net19,429(1,688)44518,186
Goodwill255,551(36,278)3,304222,577
Total assets1,883,880(38,465)3,8033,439637(974)5,4718,3151,866,106
Deferred tax liabilities46,802(684)324(292)(170)(52)45,928
Convertible unsecured senior notes57,48134,06491,545
Exchangeable bonds13,75513,755
Other liabilities21,3548,46729,821
Total liabilities772,095(684)324(292)(170)8,467(52)34,064813,752
Mezzanine equity9,884(9,884)
Total shareholders' equity1,032,495(17,252)3,4793,731807(974)1,4322,711(25,749)1,000,680
Noncontrolling interests69,406(20,529)(15)2,81251,674
Total equity1,101,901(37,781)3,4793,731807(974)1,4175,523(25,749)1,052,354
Total liabilities, mezzanine equity and equity1,883,880(38,465)3,8033,439637(974)5,4718,3151,866,106
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-48 GAAP to IFRS.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-48 GAAP to IFRS.html new file mode 100644 index 000000000..f0c1d63d0 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-48 GAAP to IFRS.html @@ -0,0 +1 @@ +
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity's own equity (viii)Amounts under IFRS
(in millions of RMB)
Revenue479,73997479,836
Cost of revenue(290,135)(22)641(235)(289,751)
Product development expenses(27,555)(10)(1)(27,566)
Sales and marketing expenses(65,167)(2)1(323)(65,491)
General and administrative expenses(23,057)(18)70(9)(23,014)
Amortization and impairment of intangible assets(3,441)16039(3,242)
Other gains, net851851
Income from operations71,235160(52)712(432)71,623
Interest and investment income, net17,1291,750(56)2261(7,689)11,217
Interest expense(4,615)(775)(434)(12)(5,836)
Other expense, net(1,221)442(779)
Income tax expenses(17,442)(35)109(227)86611(221)(16,939)
Share of results of equity method investees2,4831,15143,638
Net income67,5691251,859868818(52)(412)(162)(7,689)62,924
Net loss attributable to noncontrolling interests854(59)(3)1331451,070
Accretion of mezzanine equity(280)280
Net income attributable to ordinary shareholders68,143661,859868815(52)1(17)(7,689)63,994
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-49 GAAP to IFRS Reconciliation.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-49 GAAP to IFRS Reconciliation.html new file mode 100644 index 000000000..b4e7b73f4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-49 GAAP to IFRS Reconciliation.html @@ -0,0 +1 @@ +
Amounts as reported under U.S. GAAPConsolidation and business combinations, and others (i)Equity securities without readily determinable fair value (ii)Equity method investments (iii)Share-based awards (iv)Operating leases (v)Redeemable noncontrolling interests (vi)Hyperinflation (vii)Convertible unsecured senior notes and instruments in an entity’s own equity (viii)Amounts under IFRS
(in millions of RMB)
Revenue495,4471495,448
Cost of revenue(287,210)(12)304(425)(287,343)
Product development expenses(32,096)(7)(15)(32,118)
Sales and marketing expenses(119,674)(2)1(154)(119,829)
General and administrative expenses(14,778)(14)36(5)(14,761)
Amortization and impairment of intangible assets(1,633)17367(1,393)
Other gains (losses), net297(506)(209)
Income from operations40,353173(35)341(1,037)39,795
Interest and investment income, net37,468(1,641)210611016(18,008)18,116
Interest expense(4,995)(395)(337)1,536(30)(4,221)
Other income, net1,32973961,732
Income tax expenses(14,415)(35)39160(31)(389)(14,770)
Share of results of equity method investees3,254(79)293,204
Net income62,994(250)(1,602)13254341,546(1,044)(18,008)43,856
Net income attributable to noncontrolling interests(1,326)(65)(3)887251(256)
Reversal of accretion of mezzanine equity2,438(2,438)
Net income attributable to ordinary shareholders64,106(315)(1,602)1325134(5)(793)(18,008)43,600
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-5 Equity Investee Results.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-5 Equity Investee Results.html new file mode 100644 index 000000000..60ac84251 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-5 Equity Investee Results.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Share of profit (loss) of equity method investees
- Ant Group6,3954,280601
- Others(1,334)1,126158
Impairment loss(2,157)(5)(1)
$Others^{(1)}$ (421)(2,147)(301)
Total2,4833,254457
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-6 Financial Results.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-6 Financial Results.html new file mode 100644 index 000000000..ddce537cc --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-6 Financial Results.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Net income67,56962,9948,849
Adjustments to reconcile net income to adjusted EBITA and adjusted EBITDA:
Interest and investment income, net(17,129)(37,468)(5,263)
Interest expense4,6154,995702
Other expense (income), net1,221(1,329)(187)
Income tax expenses17,44214,4152,024
Share of results of equity method investees(2,483)(3,254)(457)
Income from operations71,23540,3535,668
Non-cash share-based compensation expense7,7756,076854
Amortization and impairment of intangible assets, and others3,4411,488209
Provision for the shareholder class action lawsuits3,145--
Adjusted EBITA85,59647,9176,731
Depreciation and impairment of property and equipment, and operating lease cost relating to land use rights12,89215,0742,117
Adjusted EBITDA98,48862,9918,848
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-7 Non-GAAP Net Income.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-7 Non-GAAP Net Income.html new file mode 100644 index 000000000..ad0549e82 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-7 Non-GAAP Net Income.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Net income67,56962,9948,849
Adjustments to reconcile net income to non-GAAP net income:
Non-cash share-based compensation expense7,7756,076854
Amortization and impairment of intangible assets3,4411,633229
Provision for the shareholder class action lawsuits3,145--
Gain on deemed disposals/disposals/revaluation of investments(8,116)(29,320)(4,119)
Impairment of investments, and others5,0672,455345
Tax effects(1)(1,672)243
Non-GAAP net income77,20943,8626,161
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-8 Earnings per Share.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-8 Earnings per Share.html new file mode 100644 index 000000000..0b4f4414d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-8 Earnings per Share.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions, except per share data)
Net income attributable to ordinary shareholders – basic68,14364,1069,005
Dilution effect on earnings arising from non-cash share-based awards operated by equity method investees and subsidiaries(131)(258)(36)
Adjustments for interest expense attributable to convertible unsecured senior notes9514320
Net income attributable to ordinary shareholders – diluted68,10763,9918,989
Non-GAAP adjustments to net income attributable to ordinary shareholders(1)8,521(18,250)(2,564)
Non-GAAP net income attributable to ordinary shareholders for computing non-GAAP diluted earnings per share/ADS76,62845,7416,425
Weighted average number of shares on a diluted basis for computing non-GAAP diluted earnings per share/ADS (million shares)(2)19,45919,154
Diluted earnings per share(2)(3)3.503.340.47
Non-GAAP diluted earnings per share(2)(4)3.942.390.34
Diluted earnings per ADS(2)(3)28.0026.733.75
Non-GAAP diluted earnings per ADS(2)(4)31.5019.102.68
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-9 Free Cash Flow.html b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-9 Free Cash Flow.html new file mode 100644 index 000000000..380d1784d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/tables/table-9 Free Cash Flow.html @@ -0,0 +1 @@ +
Six months ended September 30,
20242025
RMBRMBUS$
(in millions)
Net cash provided by operating activities65,07430,7714,322
Less: Purchase of property and equipment (excluding land use rights and construction in progress relating to office campuses)(28,916)(70,057)(9,841)
Less: Changes in the buyer protection fund deposits(5,051)(1,369)(192)
Free cash flow31,107(40,655)(5,711)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/toc_hierarchies.json b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/toc_hierarchies.json new file mode 100644 index 000000000..b38b706ca --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-alibaba-fy2026-interim-report/toc_hierarchies.json @@ -0,0 +1,23 @@ +[ + { + "toc_range": [ + 4, + 12 + ], + "scan_range": [ + 4, + 163 + ], + "toc_with_level": "| id | heading | level |\n|----|--------------------------------------------------------------------|-------|\n| 1 | 4 Management Discussion and Analysis | 1 |\n| 2 | 25 Directors and Chief Executive Officer | 1 |\n| 3 | 29 Disclosure of Interests | 1 |\n| 4 | 34 Equity Incentive Plans | 1 |\n| 5 | 38 Purchase, Sale or Redemption of our Company's Listed Securities | 1 |\n| 6 | 39 Other Information | 1 |\n| 7 | 42 Definitions | 1 |\n| 8 | 44 Financial Statements | 1 |", + "toc_tree": { + "4 Management Discussion and Analysis": {}, + "25 Directors and Chief Executive Officer": {}, + "29 Disclosure of Interests": {}, + "34 Equity Incentive Plans": {}, + "38 Purchase, Sale or Redemption of our Company's Listed Securities": {}, + "39 Other Information": {}, + "42 Definitions": {}, + "44 Financial Statements": {} + } + } +] \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/chunks.json b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/chunks.json new file mode 100644 index 000000000..a4ece90de --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/chunks.json @@ -0,0 +1,20355 @@ +{ + "chunks": [ + { + "chunk_id": "4253031f-a154-518a-80db-53af82529a33", + "type": "table", + "content": "
Title of each classTrading Symbol(s)Name of each exchange on which registered
Class A Common Stock, $0.001 par valueGOOGLNasdaq Stock Market LLC(Nasdaq Global Select Market)
Class C Capital Stock, $0.001 par valueGOOGNasdaq Stock Market LLC(Nasdaq Global Select Market)
2.375% Senior Notes due 2028Nasdaq Stock Market LLC
2.500% Senior Notes due 2029Nasdaq Stock Market LLC
2.875% Senior Notes due 2031Nasdaq Stock Market LLC
3.000% Senior Notes due 2033Nasdaq Stock Market LLC
3.125% Senior Notes due 2034Nasdaq Stock Market LLC
3.375% Senior Notes due 2037Nasdaq Stock Market LLC
3.500% Senior Notes due 2038Nasdaq Stock Market LLC
4.000% Senior Notes due 2044Nasdaq Stock Market LLC
3.875% Senior Notes due 2045Nasdaq Stock Market LLC
4.000% Senior Notes due 2054Nasdaq Stock Market LLC
4.375% Senior Notes due 2064Nasdaq Stock Market LLC
", + "path": "tables/table-0 Stock & Note Listings.html", + "metadata": { + "length": 1358, + "summary": "table-1\nTable lists Alphabet's Class A and C stocks (GOOGL, GOOG) on Nasdaq, plus multiple senior notes (2.375%-4.375%) also listed on Nasdaq.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-0 Stock & Note Listings.html", + "keywords": [ + "Stock", + "Notes", + "Nasdaq" + ], + "tokens": [] + } + }, + { + "chunk_id": "ea12468c-955a-5298-8122-7c12828f904d", + "type": "table", + "content": "
Large accelerated filerAccelerated filer
Non-accelerated filerSmaller reporting company
Emerging growth company
", + "path": "tables/table-1 Filer Status.html", + "metadata": { + "length": 264, + "summary": "table-2\nThe table shows filer statuses: Large accelerated filer is checked, while Accelerated filer, Non-accelerated filer, Smaller reporting company, and Emerging growth company are unchecked.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-1 Filer Status.html", + "keywords": [ + "Filer", + "Status", + "SEC" + ], + "tokens": [] + } + }, + { + "chunk_id": "816c2fd6-819d-572c-9926-949105ee9ee3", + "type": "text", + "content": "UNITED STATES\nSECURITIES AND EXCHANGE COMMISSION\nWashington, D.C. 20549\nFORM 10-K\n(Mark One)\nANNUAL REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934\nFor the fiscal year ended December 31, 2025\nOR\n☐ TRANSITION REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934\nFor the transition period from \\_\\_\\_\\_ to \\_\\_\\_\\_.\nCommission file number: 001-37580\nAlphabet Inc.\n(Exact name of registrant as specified in its charter)\nDelaware\n(State or other jurisdiction of incorporation or organization)\n61-1767919\n(I.R.S. Employer Identification No.)\n1600 Amphitheatre Parkway\nMountain View, CA 94043\n(Address of principal executive offices, including zip code)\n(650) 253-0000\n(Registrant's telephone number, including area code)\nSecurities registered pursuant to Section 12(b) of the Act:\n\n[tables/table-0 Stock & Note Listings.html]\n\nSecurities registered pursuant to Section 12(g) of the Act:\nTitle of each class\nNone\nIndicate by check mark whether the registrant (1) has filed all reports required to be filed by Section 13 or 15(d) of the Securities Exchange Act of 1934 during the preceding 12 months (or for such shorter period that the registrant was required to file such reports), and (2) has been subject to such filing requirements for the past 90 days. Yes ☑ No □\nIndicate by check mark whether the registrant has submitted electronically every Interactive Data File required to be submitted pursuant to Rule 405 of Regulation S-T ( $§232.405$ of this chapter) during the preceding 12 months (or for such shorter period that the registrant was required to submit such files). Yes ☒ No ☐\nIndicate by check mark whether the registrant is a large accelerated filer, an accelerated filer, a non-accelerated filer, a smaller reporting company, or an emerging growth company. See the definitions of “large accelerated filer,” “accelerated filer,” “smaller reporting company,” and \"emerging growth company\" in Rule 12b-2 of the Exchange Act.\n\n[tables/table-1 Filer Status.html]\n\nIf an emerging growth company, indicate by check mark if the registrant has elected not to use the extended transition period for complying with any new or revised financial accounting standards provided pursuant to Section 13(a) of the Exchange Act. □\nIndicate by check mark whether the registrant has filed a report on and attestation to its management's assessment of the effectiveness of its internal control over financial reporting under Section 404(b) of the Sarbanes-Oxley Act (15 U.S.C.7262(b)) by the registered public accounting firm that prepared or issued its audit report. ☒\nIf securities are registered pursuant to Section 12(b) of the Act, indicate by check mark whether the financial statements of the registrant included in the filing reflect the correction of an error to previously issued financial statements. □\nIndicate by check mark whether any of those error corrections are restatements that required a recovery analysis of incentive-based compensation received by any of the registrant's executive officers during the relevant recovery period pursuant to §240.10D-1(b). □\nIndicate by check mark whether the registrant is a shell company (as defined in Rule 12b-2 of the Exchange Act). Yes □ No ☒\nAs of June 30, 2025, the aggregate market value of shares held by non-affiliates of the registrant (based upon the closing sale prices of such shares on the Nasdaq Global Select Market on June 30, 2025) was approximately \\$1.9 trillion. For purposes of calculating the aggregate market value of shares held by non-affiliates, we have assumed that all outstanding shares are held by non-affiliates, except for shares held by each of our executive officers, directors, and 5% or greater stockholders. In the case of 5% or greater stockholders, we have not deemed such stockholders to be affiliates unless there are facts and circumstances which would indicate that such stockholders exercise any control over our company, or unless they hold 10% or more of our outstanding common stock. These assumptions should not be deemed to constitute an admission that all executive officers, directors, and 5% or greater stockholders are, in fact, affiliates of our company, or that there are not other persons who may be deemed to be affiliates of our company. Further information concerning shareholdings of our officers, directors, and principal stockholders is included or incorporated by reference in Part III, Item 12 of this Annual Report on Form 10-K.\nAs of January 28, 2026, there were 5,822 million shares of Alphabet's Class A stock outstanding, 837 million shares of Alphabet's Class B stock outstanding, and 5,438 million shares of the Alphabet's Class C stock outstanding.\nDOCUMENTS INCORPORATED BY REFERENCE\nPortions of the registrant's Proxy Statement for the 2026 Annual Meeting of Stockholders are incorporated herein by reference in Part III of this Annual Report on Form 10-K to the extent stated herein. Such proxy statement will be filed with the Securities and Exchange Commission within 120 days of the registrant's fiscal year ended December 31, 2025.\nAlphabet Inc.\nForm 10-K\nFor the Fiscal Year Ended December 31, 2025", + "path": "GOOG-10-K-2025.pdf", + "metadata": { + "length": 5249, + "summary": "This is the Form 10-K annual report filed by Alphabet Inc. with the U.S. Securities and Exchange Commission for the fiscal year ended December 31, 2025. It includes company identification details (Delaware incorporation, IRS employer ID, principal executive offices in Mountain View, CA), checkmarks indicating compliance with SEC filing requirements, and classification as a large accelerated filer. The report notes that as of June 30, 2025, the aggregate market value of shares held by non-affiliates was approximately $1.9 trillion, and as of January 28, 2026, there were 5,822 million Class A shares, 837 million Class B shares, and 5,438 million Class C shares outstanding. Portions of the proxy statement for the 2026 annual meeting are incorporated by reference in Part III.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "UNITED", + "STATES", + "SECURITIES", + "AND", + "EXCHANGE", + "COMMISSION", + "Washington", + "20549", + "FORM", + "10", + "Mark", + "One", + "ANNUAL", + "REPORT", + "PURSUANT", + "TO", + "SECTION", + "13", + "OR", + "15", + "OF", + "THE", + "ACT", + "1934", + "For", + "fiscal", + "year", + "ended", + "December", + "31", + "2025", + "TRANSITION", + "transition", + "period", + "Commission", + "file", + "number", + "001", + "37580", + "Alphabet", + "Inc", + "Exact", + "registrant", + "charter", + "Delaware", + "State", + "jurisdiction", + "incorporation", + "organization", + "61", + "1767919", + "Employer", + "Identification", + "No", + "1600", + "Amphitheatre", + "Parkway", + "Mountain", + "View", + "CA", + "94043", + "Address", + "principal", + "executive", + "offices", + "including", + "zip", + "code", + "650", + "253", + "0000", + "Registrant", + "telephone", + "area", + "Securities", + "registered", + "pursuant", + "Section", + "12", + "Act", + "Title", + "class", + "None", + "Indicate", + "check", + "mark", + "filed", + "reports", + "required", + "Exchange", + "preceding", + "months", + "shorter", + "subject", + "filing", + "requirements", + "past", + "90", + "days", + "Yes", + "submitted", + "electronically", + "Interactive", + "Data", + "File", + "Rule", + "405", + "Regulation", + "232.405", + "chapter", + "submit", + "files", + "large", + "accelerated", + "filer", + "smaller", + "reporting", + "company", + "emerging", + "growth", + "See", + "definitions", + "12b", + "If", + "elected", + "extended", + "complying", + "revised", + "financial", + "accounting", + "standards", + "provided", + "report", + "attestation", + "management", + "assessment", + "effectiveness", + "internal", + "control", + "404", + "Sarbanes", + "Oxley", + "C.7262", + "public", + "firm", + "prepared", + "issued", + "audit", + "securities", + "statements", + "included", + "reflect", + "correction", + "error", + "previously", + "corrections", + "restatements", + "recovery", + "analysis", + "incentive", + "based", + "compensation", + "received", + "officers", + "relevant", + "240.10", + "shell", + "defined", + "As", + "June", + "30", + "aggregate", + "market", + "shares", + "held", + "affiliates", + "closing", + "sale", + "prices", + "Nasdaq", + "Global", + "Select", + "Market", + "approximately", + "1.9", + "trillion", + "purposes", + "calculating", + "assumed", + "outstanding", + "directors", + "5%", + "greater", + "stockholders", + "In", + "case", + "deemed", + "facts", + "circumstances", + "exercise", + "hold", + "10%", + "common", + "stock", + "These", + "assumptions", + "constitute", + "admission", + "fact", + "persons", + "Further", + "information", + "shareholdings", + "incorporated", + "reference", + "Part", + "III", + "Item", + "Annual", + "Report", + "Form", + "January", + "28", + "2026", + "822", + "million", + "Class", + "837", + "438", + "DOCUMENTS", + "INCORPORATED", + "BY", + "REFERENCE", + "Portions", + "Proxy", + "Statement", + "Meeting", + "Stockholders", + "extent", + "stated", + "Such", + "proxy", + "statement", + "120", + "Fiscal", + "Year", + "Ended" + ], + "keywords": [ + "SEC Filing", + "Annual Report", + "Alphabet" + ], + "connect_to": [ + { + "target": "4253031f-a154-518a-80db-53af82529a33", + "relation": "embeds", + "ref": "[tables/table-0 Stock & Note Listings.html]", + "position": { + "start": 832, + "end": 875 + } + }, + { + "target": "ea12468c-955a-5298-8122-7c12828f904d", + "relation": "embeds", + "ref": "[tables/table-1 Filer Status.html]", + "position": { + "start": 1990, + "end": 2024 + } + } + ] + } + }, + { + "chunk_id": "a48980e5-7c03-5b19-b8e9-e4416b727182", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART I", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c0b5be75-51c2-521d-85c7-49ae34e32796", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a92f9964-19af-5284-8791-fd92807ffffd", + "type": "text", + "content": "As our founders Larry and Sergey wrote in the original founders' letter, \"Google is not a conventional company. We do not intend to become one.\" That unconventional spirit has been a driving force throughout our history, inspiring us to tackle big problems and invest in moonshots. It led us to be a pioneer in the development of artificial intelligence (AI) and, since 2016, be an AI-first company. We continue this work under the leadership of Alphabet and Google CEO, Sundar Pichai.\nAlphabet is a collection of businesses — the largest of which is Google. We report Google in two segments, Google Services and Google Cloud, and all non-Google businesses collectively as Other Bets. Supporting these businesses, we have centralized certain AI-related research and development focused on advanced research in AI and developing the frontier models that serve our businesses, which is reported in Alphabet-level activities. Alphabet's structure is about helping each of our businesses prosper through strong leaders and independence.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview", + "metadata": { + "length": 1032, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "As", + "founders", + "Larry", + "Sergey", + "wrote", + "original", + "letter", + "Google", + "conventional", + "company", + "We", + "intend", + "That", + "unconventional", + "spirit", + "driving", + "force", + "history", + "inspiring", + "tackle", + "big", + "problems", + "invest", + "moonshots", + "It", + "led", + "pioneer", + "development", + "artificial", + "intelligence", + "AI", + "2016", + "continue", + "work", + "leadership", + "Alphabet", + "CEO", + "Sundar", + "Pichai", + "collection", + "businesses", + "largest", + "report", + "segments", + "Services", + "Cloud", + "collectively", + "Other", + "Bets", + "Supporting", + "centralized", + "related", + "research", + "focused", + "advanced", + "developing", + "frontier", + "models", + "serve", + "reported", + "level", + "activities", + "structure", + "helping", + "prosper", + "strong", + "leaders", + "independence" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fc6d7284-d91d-5c3a-b9f8-3a1e706a060b", + "type": "text", + "content": "The Internet is one of the world's most powerful equalizers; it propels ideas, people, and businesses large and small. Our mission to organize the world's information and make it universally accessible and useful is as relevant today as it was when we were founded in 1998. Since then, we have evolved from a company that helps people find answers to a company that also helps people get things done.\nWe are focused on building an even more helpful Google for everyone, and we aspire to give everyone the tools they need to increase their knowledge, health, happiness, and success. Google Search helps people find information and make sense of the world in more natural and intuitive ways, with trillions of searches on Google every year. YouTube provides people with entertainment, information, and opportunities to learn something new and helps support the creator economy through the YouTube Partner Program. Google Cloud helps customers build for the future, improve productivity, reduce costs, and unlock new growth engines. We continually innovate and build new products and features to help our users, partners, customers, and communities and have invested more than \\$200 billion in research and development in the last five years in support of these efforts.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Access and Technology for Everyone", + "metadata": { + "length": 1267, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "Internet", + "world", + "powerful", + "equalizers", + "propels", + "ideas", + "people", + "businesses", + "large", + "small", + "Our", + "mission", + "organize", + "information", + "make", + "universally", + "accessible", + "relevant", + "today", + "founded", + "1998", + "Since", + "evolved", + "company", + "helps", + "find", + "answers", + "things", + "We", + "focused", + "building", + "helpful", + "Google", + "aspire", + "give", + "tools", + "increase", + "knowledge", + "health", + "happiness", + "success", + "Search", + "sense", + "natural", + "intuitive", + "ways", + "trillions", + "searches", + "year", + "YouTube", + "entertainment", + "opportunities", + "learn", + "support", + "creator", + "economy", + "Partner", + "Program", + "Cloud", + "customers", + "build", + "future", + "improve", + "productivity", + "reduce", + "costs", + "unlock", + "growth", + "engines", + "continually", + "innovate", + "products", + "features", + "users", + "partners", + "communities", + "invested", + "200", + "billion", + "research", + "development", + "years", + "efforts" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ff58f036-8701-548d-871f-e238853d3194", + "type": "text", + "content": "We believe AI is a profound platform shift that can bring meaningful and positive change to people and societies across the world, and to our business. We aim to build the most advanced, safe, and responsible AI through our full-stack approach, which spans AI-optimized infrastructure; world-class research, including models and tooling; and our products and platforms that bring AI to billions of people, developers, and enterprises.\nAt the foundation of our full-stack approach is our AI-optimized infrastructure — a key differentiator enabling us to power our own products, such as Search and YouTube, and support the services we provide to our Google Cloud customers. Our technical infrastructure allows us to use and offer our customers a range of AI accelerator options, including specialized Graphics Processing Units (GPUs) and our own custom-built Tensor Processing Units (TPUs), such as Ironwood, our seventh-generation TPU. We are focused on driving efficiencies in our data centers, allowing us to leverage our technical infrastructure to deliver our products and services at an increasing scale while simultaneously enabling world-class research and model development.\nOver the last decade, our research teams have pushed the boundaries of AI forward, which is displayed through Gemini 3, our most intelligent AI model yet. Designed to deliver advanced multimodal understanding, Gemini 3 represents our most capable iteration of agentic and generative coding technologies. Gemini 3 integrates enhanced reasoning capabilities to support visualizations and interactive user experiences across our product ecosystem, including Search and the Gemini app.\nAs technology continues to improve rapidly, we are focused on bringing our latest AI advances to our products and platforms. We continue to help our users access information and knowledge, express themselves, and get things done by embedding the power of generative AI and Gemini into our products and platforms. Today, all 15 of our half-billion-user products — including seven with two billion users — use our Gemini models. For our Google Cloud customers, our offerings are helping organizations stay at the forefront of innovation with solutions such as Gemini Enterprise and Gemini for Google Workspace.\nGuided by our AI principles, we believe our approach to AI must be both bold and responsible. That means developing AI in a way that maximizes the positive benefits to society while addressing its potential challenges.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Making AI Helpful for Everyone", + "metadata": { + "length": 2491, + "summary": "Google's AI strategy is a full-stack approach encompassing AI-optimized infrastructure, world-class research, and products. Key differentiators include custom TPUs like Ironwood and GPUs. Gemini 3 is their most advanced model, featuring enhanced reasoning and multimodal capabilities. AI is embedded across all major products, including Search and YouTube, and offered to Google Cloud customers. The approach is guided by principles of boldness and responsibility.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "AI", + "profound", + "platform", + "shift", + "bring", + "meaningful", + "positive", + "change", + "people", + "societies", + "world", + "business", + "aim", + "build", + "advanced", + "safe", + "responsible", + "full", + "stack", + "approach", + "spans", + "optimized", + "infrastructure", + "class", + "research", + "including", + "models", + "tooling", + "products", + "platforms", + "billions", + "developers", + "enterprises", + "At", + "foundation", + "key", + "differentiator", + "enabling", + "power", + "Search", + "YouTube", + "support", + "services", + "provide", + "Google", + "Cloud", + "customers", + "Our", + "technical", + "offer", + "range", + "accelerator", + "options", + "specialized", + "Graphics", + "Processing", + "Units", + "GPUs", + "custom", + "built", + "Tensor", + "TPUs", + "Ironwood", + "seventh", + "generation", + "TPU", + "focused", + "driving", + "efficiencies", + "data", + "centers", + "allowing", + "leverage", + "deliver", + "increasing", + "scale", + "simultaneously", + "model", + "development", + "Over", + "decade", + "teams", + "pushed", + "boundaries", + "forward", + "displayed", + "Gemini", + "intelligent", + "Designed", + "multimodal", + "understanding", + "represents", + "capable", + "iteration", + "agentic", + "generative", + "coding", + "technologies", + "integrates", + "enhanced", + "reasoning", + "capabilities", + "visualizations", + "interactive", + "user", + "experiences", + "product", + "ecosystem", + "app", + "As", + "technology", + "continues", + "improve", + "rapidly", + "bringing", + "latest", + "advances", + "continue", + "users", + "access", + "information", + "knowledge", + "express", + "things", + "embedding", + "Today", + "15", + "half", + "billion", + "For", + "offerings", + "helping", + "organizations", + "stay", + "forefront", + "innovation", + "solutions", + "Enterprise", + "Workspace", + "Guided", + "principles", + "bold", + "That", + "means", + "developing", + "maximizes", + "benefits", + "society", + "addressing", + "potential", + "challenges" + ], + "keywords": [ + "AI", + "infrastructure", + "Gemini" + ], + "connect_to": [] + } + }, + { + "chunk_id": "d4c98f29-5ab9-5346-a504-c4263b2a2423", + "type": "text", + "content": "Many companies get comfortable doing what they have always done, making only incremental changes. This incrementalism leads to irrelevance over time, especially in technology, where change tends to be revolutionary, not evolutionary.\nOur early investments in AI started out as moonshots but are now incorporated into our core products and central to future developments. In Other Bets, our fully autonomous driving technology company, Waymo, is now providing fully autonomous, paid ride-hailing services to customers in multiple cities. Isomorphic Labs is reimagining the drug discovery process from first principles, applying AI to accelerate the development of new medicines. We continue to look toward the future and to invest for the long term, most notably for the application of AI to our products and services, as well as other frontier technologies such as quantum computing.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Moonshots", + "metadata": { + "length": 883, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Many", + "companies", + "comfortable", + "making", + "incremental", + "This", + "incrementalism", + "leads", + "irrelevance", + "time", + "technology", + "change", + "revolutionary", + "evolutionary", + "Our", + "early", + "investments", + "AI", + "started", + "moonshots", + "incorporated", + "core", + "products", + "central", + "future", + "developments", + "In", + "Other", + "Bets", + "fully", + "autonomous", + "driving", + "company", + "Waymo", + "providing", + "paid", + "ride", + "hailing", + "services", + "customers", + "multiple", + "cities", + "Isomorphic", + "Labs", + "reimagining", + "drug", + "discovery", + "process", + "principles", + "applying", + "accelerate", + "development", + "medicines", + "We", + "continue", + "invest", + "long", + "term", + "notably", + "application", + "frontier", + "technologies", + "quantum", + "computing" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fe8f8933-fb80-55c2-946d-4f28d7cf2b09", + "type": "text", + "content": "We make it a priority to protect the privacy and security of our products, users, and customers, even if there are near-term financial consequences. We do this by continuously investing in building products that are secure by default; strictly upholding responsible data practices that emphasize privacy by design; and building easy-to-use settings that put people in control. We are continually enhancing these efforts over time, whether by enabling users to auto-delete their data, applying privacy technologies like on-device processing, giving people tools to control their experience, or advancing anti-malware, anti-phishing, and password security features.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Privacy and Security", + "metadata": { + "length": 663, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "make", + "priority", + "protect", + "privacy", + "security", + "products", + "users", + "customers", + "term", + "financial", + "consequences", + "continuously", + "investing", + "building", + "secure", + "default", + "strictly", + "upholding", + "responsible", + "data", + "practices", + "emphasize", + "design", + "easy", + "settings", + "put", + "people", + "control", + "continually", + "enhancing", + "efforts", + "time", + "enabling", + "auto", + "delete", + "applying", + "technologies", + "device", + "processing", + "giving", + "tools", + "experience", + "advancing", + "anti", + "malware", + "phishing", + "password", + "features" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a740b5a0-bb9e-5731-b930-c38e38561280", + "type": "text", + "content": "For reporting purposes Google comprises two segments: Google Services and Google Cloud.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google", + "metadata": { + "length": 87, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "For", + "reporting", + "purposes", + "Google", + "comprises", + "segments", + "Services", + "Cloud" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4d146748-3b1d-5ada-b9ae-dadb95d7ecc8", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google Services", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b97fe3a8-4f38-52ed-a066-56bd445a5dc8", + "type": "text", + "content": "We have always been committed to building helpful products that can improve the lives of millions of people worldwide. Our product innovations are what make our services widely used, and our brand one of the most recognized in the world. Google Services' core products and platforms include ads, Android, Chrome, devices, Gmail, Google Drive, Google Gemini, Google Maps, Google Photos, Google Play, Search, and YouTube, with broad and growing adoption by users around the world.\nOur products and services have come a long way since the company was founded more than 25 years ago. While Google Search started as a way to find web pages, organized into ten blue links, we have driven technical advancements and product innovations that have transformed Google Search into a dynamic, multimodal experience. Large language models have made it possible to express more natural language queries, vastly improving the types of questions users can ask, and the quality of results. For example, AI Overviews makes it easier to ask Google anything and get a helpful response. AI Mode allows users to ask more nuanced questions that might have previously taken multiple searches, using Gemini's advanced reasoning, thinking, and multimodal capabilities.\nThis drive to make information more accessible and helpful has led us over the years to improve the discovery and creation of digital content both on the web and through platforms like Google Play and YouTube. People are consuming many forms of digital content, including watching long and short form videos and podcasts, streaming TV, playing games, listening to music, reading books, and using apps. Working with content creators and partners, we continue to build new ways for people around the world to create and find great digital content.\nFueling all of these great digital experiences are extraordinary platforms and devices. That is why we continue to invest in platforms like our Android mobile operating system, Chrome browser, and Chrome operating system, as well as our family of devices. We see tremendous potential for devices to be helpful and make people's lives easier by combining the best of our AI, software, and hardware. This potential is reflected in our latest generation of devices, such as the new Pixel 10 series and the Pixel Watch 4. Creating products and services that people rely on every day is a journey that we are investing in for the long-term.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google Services/Serving Our Users", + "metadata": { + "length": 2424, + "summary": "Google is committed to building helpful products that improve lives worldwide, with core offerings including ads, Android, Chrome, devices, Gmail, Google Drive, Gemini, Maps, Photos, Play, Search, and YouTube. Over 25 years, Google Search evolved from simple web links to a multimodal experience using large language models, enabling natural language queries and features like AI Overviews and AI Mode. The company enhances digital content discovery and creation through platforms like Google Play and YouTube, supporting videos, podcasts, TV, games, music, books, and apps. Investments in Android, Chrome, and devices like the Pixel 10 series and Pixel Watch 4 combine AI, software, and hardware to make daily life easier.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "committed", + "building", + "helpful", + "products", + "improve", + "lives", + "millions", + "people", + "worldwide", + "Our", + "product", + "innovations", + "make", + "services", + "widely", + "brand", + "recognized", + "world", + "Google", + "Services", + "core", + "platforms", + "include", + "ads", + "Android", + "Chrome", + "devices", + "Gmail", + "Drive", + "Gemini", + "Maps", + "Photos", + "Play", + "Search", + "YouTube", + "broad", + "growing", + "adoption", + "users", + "long", + "company", + "founded", + "25", + "years", + "ago", + "While", + "started", + "find", + "web", + "pages", + "organized", + "ten", + "blue", + "links", + "driven", + "technical", + "advancements", + "transformed", + "dynamic", + "multimodal", + "experience", + "Large", + "language", + "models", + "made", + "express", + "natural", + "queries", + "vastly", + "improving", + "types", + "questions", + "quality", + "results", + "For", + "AI", + "Overviews", + "makes", + "easier", + "response", + "Mode", + "nuanced", + "previously", + "multiple", + "searches", + "advanced", + "reasoning", + "thinking", + "capabilities", + "This", + "drive", + "information", + "accessible", + "led", + "discovery", + "creation", + "digital", + "content", + "People", + "consuming", + "forms", + "including", + "watching", + "short", + "form", + "videos", + "podcasts", + "streaming", + "TV", + "playing", + "games", + "listening", + "music", + "reading", + "books", + "apps", + "Working", + "creators", + "partners", + "continue", + "build", + "ways", + "create", + "great", + "Fueling", + "experiences", + "extraordinary", + "That", + "invest", + "mobile", + "operating", + "system", + "browser", + "family", + "tremendous", + "potential", + "combining", + "software", + "hardware", + "reflected", + "latest", + "generation", + "Pixel", + "10", + "series", + "Watch", + "Creating", + "rely", + "day", + "journey", + "investing", + "term" + ], + "keywords": [ + "Google", + "AI", + "Products" + ], + "connect_to": [] + } + }, + { + "chunk_id": "37e500b7-d0ba-5c73-9f83-6e1b95b3ac2e", + "type": "text", + "content": "We have built world-class advertising technologies for advertisers, agencies, and publishers to power their digital marketing businesses. Our advertising solutions help millions of companies grow their businesses through our wide range of products across devices and formats, and we aim to ensure positive user experiences by serving the right ads at the right time and by building deep partnerships with brands and agencies. AI has been foundational to our advertising business for more than a decade. Products like Demand Gen, Performance Max, and Product Studio use the full power of our AI to help advertisers find untapped and incremental conversion opportunities.\nGoogle Services generates revenues primarily by delivering both performance and brand advertising that appears on Google Search & other properties, YouTube, and Google Network partners' properties (\"Google Network properties\"). We continue to invest in both performance and brand advertising and seek to improve the measurability of advertising so advertisers understand the effectiveness of their campaigns.\n- Performance advertising creates and delivers relevant ads that users will click on leading to direct engagement with advertisers. Performance advertising lets our advertisers connect with users while driving measurable results. Our ads tools allow performance advertisers to create simple text-based ads.\n- Brand advertising helps enhance users' awareness of and affinity for advertisers' products and services, through videos, text, images, and other interactive ads that run across various devices. We help brand advertisers deliver digital videos and other types of ads to specific audiences for their brand-building marketing campaigns.\nWe have allocated substantial resources to stopping bad advertising practices and protecting users on the web. We focus on creating the best advertising experiences for our users and advertisers in many ways, including filtering out invalid traffic, removing billions of bad ads from our systems every year, and closely monitoring the sites, apps, and videos where ads appear and blocklisting them when necessary to ensure that ads do not fund bad content.\nIn addition, Google Services generates revenues from products and services beyond advertising, including:\n- consumer subscriptions, which primarily include revenues from YouTube services, such as YouTube TV, YouTube Music and Premium, and NFL Sunday Ticket, as well as Google One, which offers access to our most capable Gemini models;\n- platforms, which primarily include revenues from Google Play sales of apps and in-app purchases; and\n• devices, which primarily include sales of the Pixel family of devices.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google Services/How We Make Money", + "metadata": { + "length": 2690, + "summary": "Google provides world-class advertising technologies for advertisers, agencies, and publishers, using AI for over a decade in products like Demand Gen and Performance Max. Revenue comes from performance and brand ads on Google Search, YouTube, and Google Network properties. Performance ads drive direct user engagement, while brand ads enhance awareness. Google invests in stopping bad advertising practices. Additional revenue sources include consumer subscriptions (YouTube services, Google One), platforms (Google Play), and devices (Pixel).", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "built", + "world", + "class", + "advertising", + "technologies", + "advertisers", + "agencies", + "publishers", + "power", + "digital", + "marketing", + "businesses", + "Our", + "solutions", + "millions", + "companies", + "grow", + "wide", + "range", + "products", + "devices", + "formats", + "aim", + "ensure", + "positive", + "user", + "experiences", + "serving", + "ads", + "time", + "building", + "deep", + "partnerships", + "brands", + "AI", + "foundational", + "business", + "decade", + "Products", + "Demand", + "Gen", + "Performance", + "Max", + "Product", + "Studio", + "full", + "find", + "untapped", + "incremental", + "conversion", + "opportunities", + "Google", + "Services", + "generates", + "revenues", + "primarily", + "delivering", + "performance", + "brand", + "appears", + "Search", + "properties", + "YouTube", + "Network", + "partners", + "continue", + "invest", + "seek", + "improve", + "measurability", + "understand", + "effectiveness", + "campaigns", + "creates", + "delivers", + "relevant", + "users", + "click", + "leading", + "direct", + "engagement", + "lets", + "connect", + "driving", + "measurable", + "results", + "tools", + "create", + "simple", + "text", + "based", + "Brand", + "helps", + "enhance", + "awareness", + "affinity", + "services", + "videos", + "images", + "interactive", + "run", + "deliver", + "types", + "specific", + "audiences", + "allocated", + "substantial", + "resources", + "stopping", + "bad", + "practices", + "protecting", + "web", + "focus", + "creating", + "ways", + "including", + "filtering", + "invalid", + "traffic", + "removing", + "billions", + "systems", + "year", + "closely", + "monitoring", + "sites", + "apps", + "blocklisting", + "fund", + "content", + "In", + "addition", + "consumer", + "subscriptions", + "include", + "TV", + "Music", + "Premium", + "NFL", + "Sunday", + "Ticket", + "One", + "offers", + "access", + "capable", + "Gemini", + "models", + "platforms", + "Play", + "sales", + "app", + "purchases", + "Pixel", + "family" + ], + "keywords": [ + "Advertising", + "AI", + "Google Services" + ], + "connect_to": [] + } + }, + { + "chunk_id": "f815d4c0-81e8-5d77-bc02-40ffdf8c9f04", + "type": "text", + "content": "Through our Google Cloud Platform and Google Workspace offerings, Google Cloud generates revenues primarily from consumption-based fees and subscriptions for infrastructure, platform, applications, and other cloud services. Customers use Google Cloud in multiple ways such as:\n- AI-optimized Infrastructure: runs on our Cloud, at the edge, or in customers' data centers. It can be used to migrate and modernize information technology (IT) systems and to train and serve various types of AI models. Our AI infrastructure delivers cost-performance for AI workloads. We offer a range of AI accelerators, including our custom TPUs and specialized GPUs, as well as AI-optimized storage offerings, and efficient AI software.\n- Developer Platform: delivers a fully managed AI development platform, through Vertex AI, for accessing, tuning, augmenting, and deploying custom models and agents, helping customers build applications with more than 200 foundation models, including our Gemini family, third-party, and open models.\n- Cybersecurity: provides AI-powered threat intelligence and cybersecurity solutions to help customers detect, analyze, protect against, and respond to a broad range of cybersecurity threats.\n- Data and Analytics: enables customers to migrate, clean, prepare, and feed data into their models. Our data platform also unifies data lakes, data warehouses, data governance, and advanced machine learning into a single platform that helps users analyze data using AI models across any cloud.\nGemini Enterprise: empowers teams to discover, create, share, and run AI agents all in one secure platform, bringing the best of Google AI to employees through an intuitive chat interface, helping to automate workflows and drive smarter business outcomes.\nGemini for Google Workspace: brings our AI-powered agents into Gmail, Docs, Sheets, and more to help users write, organize, visualize, accelerate workflows, and have more productive meetings.\n- Agents:", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google Services/Google Cloud", + "metadata": { + "length": 1963, + "summary": "Google Cloud generates revenue from consumption-based fees and subscriptions for infrastructure, platform, applications, and other cloud services. Customers use Google Cloud for AI-optimized infrastructure, a developer platform (Vertex AI) with over 200 foundation models, cybersecurity solutions, data and analytics, Gemini Enterprise for AI agents, and Gemini for Google Workspace.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Through", + "Google", + "Cloud", + "Platform", + "Workspace", + "offerings", + "generates", + "revenues", + "primarily", + "consumption", + "based", + "fees", + "subscriptions", + "infrastructure", + "platform", + "applications", + "cloud", + "services", + "Customers", + "multiple", + "ways", + "AI", + "optimized", + "Infrastructure", + "runs", + "edge", + "customers", + "data", + "centers", + "It", + "migrate", + "modernize", + "information", + "technology", + "IT", + "systems", + "train", + "serve", + "types", + "models", + "Our", + "delivers", + "cost", + "performance", + "workloads", + "We", + "offer", + "range", + "accelerators", + "including", + "custom", + "TPUs", + "specialized", + "GPUs", + "storage", + "efficient", + "software", + "Developer", + "fully", + "managed", + "development", + "Vertex", + "accessing", + "tuning", + "augmenting", + "deploying", + "agents", + "helping", + "build", + "200", + "foundation", + "Gemini", + "family", + "party", + "open", + "Cybersecurity", + "powered", + "threat", + "intelligence", + "cybersecurity", + "solutions", + "detect", + "analyze", + "protect", + "respond", + "broad", + "threats", + "Data", + "Analytics", + "enables", + "clean", + "prepare", + "feed", + "unifies", + "lakes", + "warehouses", + "governance", + "advanced", + "machine", + "learning", + "single", + "helps", + "users", + "Enterprise", + "empowers", + "teams", + "discover", + "create", + "share", + "run", + "secure", + "bringing", + "employees", + "intuitive", + "chat", + "interface", + "automate", + "workflows", + "drive", + "smarter", + "business", + "outcomes", + "brings", + "Gmail", + "Docs", + "Sheets", + "write", + "organize", + "visualize", + "accelerate", + "productive", + "meetings", + "Agents" + ], + "keywords": [ + "Google Cloud", + "AI", + "cloud services" + ], + "connect_to": [] + } + }, + { + "chunk_id": "9a7e9411-d339-5c44-a494-9da3d02708eb", + "type": "text", + "content": "Across Alphabet, we are also using technology to try to solve big problems that affect a wide variety of industries, including transportation and health technology. Alphabet's investment in the portfolio of Other Bets includes businesses that are at various stages of development, ranging from those in the research and development phase, such as X, our moonshot factory focused on developing breakthrough technologies, to those that are scaling commercialization, such as Waymo, which is expanding to more cities domestically, entering international markets, and further scaling operations.\nOther Bets operate as independent companies and some of them have their own boards with independent members and outside investors. While these early-stage businesses naturally come with considerable uncertainty, some of them are already generating revenue and making important strides in their industries. Revenues from Other Bets are generated primarily from the sale of autonomous transportation and internet services.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google Services/Other Bets", + "metadata": { + "length": 1012, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Across", + "Alphabet", + "technology", + "solve", + "big", + "problems", + "affect", + "wide", + "variety", + "industries", + "including", + "transportation", + "health", + "investment", + "portfolio", + "Other", + "Bets", + "includes", + "businesses", + "stages", + "development", + "ranging", + "research", + "phase", + "moonshot", + "factory", + "focused", + "developing", + "breakthrough", + "technologies", + "scaling", + "commercialization", + "Waymo", + "expanding", + "cities", + "domestically", + "entering", + "international", + "markets", + "operations", + "operate", + "independent", + "companies", + "boards", + "members", + "investors", + "While", + "early", + "stage", + "naturally", + "considerable", + "uncertainty", + "generating", + "revenue", + "making", + "important", + "strides", + "Revenues", + "generated", + "primarily", + "sale", + "autonomous", + "internet", + "services" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "777662ef-e768-56f2-80f7-73eefe8b2055", + "type": "text", + "content": "Our business is characterized by rapid change as well as new and disruptive technologies. We face formidable competition in every aspect of our business, including but not limited to, from:\n- general purpose search engines and information services;\n• vertical search engines and e-commerce providers for queries on topics such as those related to travel, jobs, and health, which users may navigate directly to rather than go through Google;\n• online advertising platforms and networks, including online shopping and streaming services;\n- other forms of advertising, such as billboards, magazines, newspapers, radio, and television, as our advertisers typically advertise in multiple media, both online and offline;\n• digital content and application platform providers;\n- providers of enterprise cloud services;\n• AI model developers and providers of AI products and services;\n- companies that design, manufacture, and market consumer hardware products, including businesses that have developed proprietary platforms;\n• providers of digital video services;\n- social networks, which users may rely on for product or service referrals, rather than seeking information through traditional search engines; and\n- providers of workspace communication and connectivity products.\nCompeting successfully depends heavily on our ability to continually develop and distribute innovative products and technologies to the marketplace across our businesses. For example, for advertising, competing successfully depends on attracting and retaining:\n- users, for whom other products and services are literally one click away, on the basis of the relevance of our advertising, as well as the general usefulness, security, and availability of our products and services;\n- advertisers, primarily based on our ability to generate sales leads, and ultimately customers, and to deliver their advertisements in an efficient and effective manner across a variety of distribution channels even as trends in advertising mediums and user preferences change; and\n- content providers, primarily based on the quality of our advertiser base, our ability to help these partners generate revenues from advertising, and the terms of our agreements with them.\nFor additional information about competition, see Item 1A Risk Factors of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Competition", + "metadata": { + "length": 2329, + "summary": "The passage describes the competitive landscape of a business facing rapid change and disruptive technologies. It lists various competitors including general and vertical search engines, online advertising platforms, other media, cloud services, AI developers, hardware makers, digital video services, social networks, and workspace communication providers. Success depends on continuous innovation and attracting users, advertisers, and content providers based on relevance, efficiency, and quality.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "business", + "characterized", + "rapid", + "change", + "disruptive", + "technologies", + "We", + "face", + "formidable", + "competition", + "aspect", + "including", + "limited", + "general", + "purpose", + "search", + "engines", + "information", + "services", + "vertical", + "commerce", + "providers", + "queries", + "topics", + "related", + "travel", + "jobs", + "health", + "users", + "navigate", + "directly", + "Google", + "online", + "advertising", + "platforms", + "networks", + "shopping", + "streaming", + "forms", + "billboards", + "magazines", + "newspapers", + "radio", + "television", + "advertisers", + "typically", + "advertise", + "multiple", + "media", + "offline", + "digital", + "content", + "application", + "platform", + "enterprise", + "cloud", + "AI", + "model", + "developers", + "products", + "companies", + "design", + "manufacture", + "market", + "consumer", + "hardware", + "businesses", + "developed", + "proprietary", + "video", + "social", + "rely", + "product", + "service", + "referrals", + "seeking", + "traditional", + "workspace", + "communication", + "connectivity", + "Competing", + "successfully", + "depends", + "heavily", + "ability", + "continually", + "develop", + "distribute", + "innovative", + "marketplace", + "For", + "competing", + "attracting", + "retaining", + "literally", + "click", + "basis", + "relevance", + "usefulness", + "security", + "availability", + "primarily", + "based", + "generate", + "sales", + "leads", + "ultimately", + "customers", + "deliver", + "advertisements", + "efficient", + "effective", + "manner", + "variety", + "distribution", + "channels", + "trends", + "mediums", + "user", + "preferences", + "quality", + "advertiser", + "base", + "partners", + "revenues", + "terms", + "agreements", + "additional", + "Item", + "1A", + "Risk", + "Factors", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [ + "competition", + "technology", + "advertising" + ], + "connect_to": [] + } + }, + { + "chunk_id": "bcbb40c7-e306-58ad-9cce-17ae56dc85e4", + "type": "text", + "content": "Our people are critical for our continued success, so we work hard to create an environment where employees can have fulfilling careers and perform at a high level. We offer industry-leading benefits and programs to take care of the diverse needs of our employees and their families, including opportunities for career growth and development, resources to support their financial health, and access to excellent healthcare choices. Our competitive compensation programs help us to attract and retain key talent, and we will continue to invest in recruiting talented people to technical and non-technical roles and rewarding them well. We provide a variety of high-quality training and support to managers to build and strengthen their capabilities — ranging from courses for new managers, to learning resources that help them provide feedback and manage performance, to coaching and individual support.\nAs of December 31, 2025, Alphabet had 190,820 employees. We have work councils and statutory employee representation obligations in certain countries, and we are committed to supporting protected labor rights, maintaining an open culture, and listening to our employees.\nWhen appropriate we partner with outside companies on a contractual basis to provide a specialized service or to temporarily cover a short-term need. The employees of our suppliers and staffing partners — vendors and temporary staff, respectively — and independent contractors who are self-employed, make up our extended workforce. We choose our partners and staffing agencies carefully, and review their compliance with Google's Supplier Code of Conduct.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Culture and Workforce", + "metadata": { + "length": 1629, + "summary": "Alphabet emphasizes its commitment to employees through industry-leading benefits, career growth opportunities, and competitive compensation to attract and retain talent. As of December 31, 2025, it had 190,820 employees. The company supports labor rights and maintains an open culture. It also partners with external companies for specialized services, using an extended workforce of vendors, temporary staff, and independent contractors, while ensuring compliance with its Supplier Code of Conduct.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "people", + "critical", + "continued", + "success", + "work", + "hard", + "create", + "environment", + "employees", + "fulfilling", + "careers", + "perform", + "high", + "level", + "We", + "offer", + "industry", + "leading", + "benefits", + "programs", + "care", + "diverse", + "families", + "including", + "opportunities", + "career", + "growth", + "development", + "resources", + "support", + "financial", + "health", + "access", + "excellent", + "healthcare", + "choices", + "competitive", + "compensation", + "attract", + "retain", + "key", + "talent", + "continue", + "invest", + "recruiting", + "talented", + "technical", + "roles", + "rewarding", + "provide", + "variety", + "quality", + "training", + "managers", + "build", + "strengthen", + "capabilities", + "ranging", + "courses", + "learning", + "feedback", + "manage", + "performance", + "coaching", + "individual", + "As", + "December", + "31", + "2025", + "Alphabet", + "190", + "820", + "councils", + "statutory", + "employee", + "representation", + "obligations", + "countries", + "committed", + "supporting", + "protected", + "labor", + "rights", + "maintaining", + "open", + "culture", + "listening", + "When", + "partner", + "companies", + "contractual", + "basis", + "specialized", + "service", + "temporarily", + "cover", + "short", + "term", + "The", + "suppliers", + "staffing", + "partners", + "vendors", + "temporary", + "staff", + "independent", + "contractors", + "employed", + "make", + "extended", + "workforce", + "choose", + "agencies", + "carefully", + "review", + "compliance", + "Google", + "Supplier", + "Code", + "Conduct" + ], + "keywords": [ + "employees", + "benefits", + "workforce" + ], + "connect_to": [] + } + }, + { + "chunk_id": "55cb9ade-2586-519e-a6da-4d445b380c1f", + "type": "text", + "content": "We are subject to numerous United States (US) federal, state, and local, as well as foreign laws, and regulations covering a wide variety of subjects, and the scope of this coverage continues to broaden with continuing new legal and regulatory developments in the US and internationally. Like other companies in the technology industry, we face increasingly heightened scrutiny from both US and foreign governments with respect to our compliance with laws and regulations. Many of these laws and regulations are evolving and their applicability and scope, as interpreted by the courts, remain uncertain. Particularly with regard to AI; competition; consumer protection; content moderation, including access restrictions for minors; data privacy and security; intellectual property; news publications; and sustainability and other social matters, we have seen an increase in new and evolving laws and regulations, as well as related enforcement actions and investigations, being proposed and implemented in recent years by legislative and regulatory bodies around the world. As we have seen in recent years, different laws and regulations on the same topic may not always have the same requirements (and sometimes may seem to have conflicting requirements), and even when requirements overlap, the rules are not always consistently implemented, interpreted, and enforced from jurisdiction to jurisdiction.\nOur compliance with these laws and regulations may be onerous and could, individually or in the aggregate, increase our cost of doing business, make our products and services less useful, limit our ability to pursue certain business practices or offer certain products and services (either in certain geographies or at all), cause us to change our business models and operations, affect our competitive position relative to our peers, or otherwise harm our business, reputation, financial condition, and operating results.\nFor additional information about government regulation applicable to our business, see Item 1A Risk Factors; Trends in Our Business and Financial Effect in Part II, Item 7; and Legal Matters in Note 10 of the Notes to Consolidated Financial Statements included in Part II, Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Government Regulation", + "metadata": { + "length": 2243, + "summary": "The passage discusses the company's exposure to numerous US and foreign laws and regulations, particularly in areas like AI, competition, consumer protection, data privacy, and sustainability. It notes increasing scrutiny, evolving and sometimes conflicting requirements across jurisdictions, and potential negative impacts on business operations, costs, and competitive position.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "subject", + "numerous", + "United", + "States", + "US", + "federal", + "state", + "local", + "foreign", + "laws", + "regulations", + "covering", + "wide", + "variety", + "subjects", + "scope", + "coverage", + "continues", + "broaden", + "continuing", + "legal", + "regulatory", + "developments", + "internationally", + "Like", + "companies", + "technology", + "industry", + "face", + "increasingly", + "heightened", + "scrutiny", + "governments", + "respect", + "compliance", + "Many", + "evolving", + "applicability", + "interpreted", + "courts", + "remain", + "uncertain", + "Particularly", + "regard", + "AI", + "competition", + "consumer", + "protection", + "content", + "moderation", + "including", + "access", + "restrictions", + "minors", + "data", + "privacy", + "security", + "intellectual", + "property", + "news", + "publications", + "sustainability", + "social", + "matters", + "increase", + "related", + "enforcement", + "actions", + "investigations", + "proposed", + "implemented", + "recent", + "years", + "legislative", + "bodies", + "world", + "As", + "topic", + "requirements", + "conflicting", + "overlap", + "rules", + "consistently", + "enforced", + "jurisdiction", + "Our", + "onerous", + "individually", + "aggregate", + "cost", + "business", + "make", + "products", + "services", + "limit", + "ability", + "pursue", + "practices", + "offer", + "geographies", + "change", + "models", + "operations", + "affect", + "competitive", + "position", + "relative", + "peers", + "harm", + "reputation", + "financial", + "condition", + "operating", + "results", + "For", + "additional", + "information", + "government", + "regulation", + "applicable", + "Item", + "1A", + "Risk", + "Factors", + "Trends", + "Business", + "Financial", + "Effect", + "Part", + "II", + "Legal", + "Matters", + "Note", + "10", + "Notes", + "Consolidated", + "Statements", + "included", + "Annual", + "Report", + "Form" + ], + "keywords": [ + "compliance", + "regulations", + "technology" + ], + "connect_to": [] + } + }, + { + "chunk_id": "1e0fa1f2-0f30-5654-ab1d-feba71cf4d21", + "type": "text", + "content": "We rely on intellectual property such as trademarks, copyrights, patents, and trade secrets, as well as confidentiality procedures and contractual provisions, to protect our proprietary technology and our brand. We have registered, and applied for the registration of, US and international trademarks, service marks, domain names, and copyrights. We have also filed patent applications in the US and foreign countries covering certain of our technology, and acquired patent assets to supplement our portfolio. We have licensed in the past, and expect that we may license in the future, certain of our rights to other parties. For additional information, see Item 1A Risk Factors of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Intellectual Property", + "metadata": { + "length": 714, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "rely", + "intellectual", + "property", + "trademarks", + "copyrights", + "patents", + "trade", + "secrets", + "confidentiality", + "procedures", + "contractual", + "provisions", + "protect", + "proprietary", + "technology", + "brand", + "registered", + "applied", + "registration", + "US", + "international", + "service", + "marks", + "domain", + "names", + "filed", + "patent", + "applications", + "foreign", + "countries", + "covering", + "acquired", + "assets", + "supplement", + "portfolio", + "licensed", + "past", + "expect", + "license", + "future", + "rights", + "parties", + "For", + "additional", + "information", + "Item", + "1A", + "Risk", + "Factors", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "241ab189-1167-558a-b2a6-8438c1fb8acb", + "type": "text", + "content": "Our website is located at www.abc.xyz, and our investor relations website is located at www.abc.xyz/investor. Access to our Annual Reports on Form 10-K, Quarterly Reports on Form 10-Q, Current Reports on Form 8-K, and our Proxy Statements, and any amendments to these reports, is available on our investor relations website, free of charge, after we file or furnish them with the SEC and they are available on the SEC's website at www.sec.gov.\nWe webcast our earnings calls, as well as certain events we participate in or host with members of the investment community, via our investor relations YouTube channel and website. Our investor relations website also provides notifications of news or announcements regarding our financial performance and other items that may be material or of interest to our investors, including SEC filings, investor events, press and earnings releases, and blogs. We also share Google news and product updates on Google's Keyword blog at https://www.blog.google/ and News From Google page on X at x.com/NewsFromGoogle, and our executive officers may also use certain social media channels, such as X and LinkedIn, to communicate information about earnings results and company updates, which may be of interest or material to our investors. Further, corporate governance information, including our certificate of incorporation, bylaws, corporate governance guidelines, board committee charters, and code of conduct, is also available on our investor relations website under the heading \"Governance.\" The information contained on, or that may be accessed through our websites or our executive officers' social media channels, is not incorporated by reference into this Annual Report on Form 10-K or in any other report or document we file with the SEC, and any references to our websites are intended to be inactive textual references only.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Available Information", + "metadata": { + "length": 1869, + "summary": "The passage describes the company's investor relations website (www.abc.xyz/investor) where annual and quarterly reports, proxy statements, and SEC filings are available free of charge. It also mentions webcasts of earnings calls, notifications of financial news, and corporate governance information. Additionally, it notes that executive officers may use social media like X and LinkedIn to share company updates, but clarifies that information on these platforms is not incorporated into SEC filings.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "website", + "located", + "www", + "abc", + "xyz", + "investor", + "relations", + "Access", + "Annual", + "Reports", + "Form", + "10", + "Quarterly", + "Current", + "Proxy", + "Statements", + "amendments", + "reports", + "free", + "charge", + "file", + "furnish", + "SEC", + "sec", + "gov", + "We", + "webcast", + "earnings", + "calls", + "events", + "participate", + "host", + "members", + "investment", + "community", + "YouTube", + "channel", + "notifications", + "news", + "announcements", + "financial", + "performance", + "items", + "material", + "interest", + "investors", + "including", + "filings", + "press", + "releases", + "blogs", + "share", + "Google", + "product", + "updates", + "Keyword", + "blog", + "https", + "google", + "News", + "From", + "page", + "NewsFromGoogle", + "executive", + "officers", + "social", + "media", + "channels", + "LinkedIn", + "communicate", + "information", + "results", + "company", + "Further", + "corporate", + "governance", + "certificate", + "incorporation", + "bylaws", + "guidelines", + "board", + "committee", + "charters", + "code", + "conduct", + "heading", + "Governance", + "The", + "contained", + "accessed", + "websites", + "incorporated", + "reference", + "Report", + "report", + "document", + "references", + "intended", + "inactive", + "textual" + ], + "keywords": [ + "SEC filings", + "investor website", + "social media" + ], + "connect_to": [] + } + }, + { + "chunk_id": "ff20ce5b-0b0a-5c94-a526-41a43668024a", + "type": "text", + "content": "Our operations and financial results are subject to various risks and uncertainties, including but not limited to those described below, which could harm our business, reputation, financial condition, and operating results, and may affect the trading price and price volatility of our Class A and Class C stock.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS", + "metadata": { + "length": 311, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "operations", + "financial", + "results", + "subject", + "risks", + "uncertainties", + "including", + "limited", + "harm", + "business", + "reputation", + "condition", + "operating", + "affect", + "trading", + "price", + "volatility", + "Class", + "stock" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "62e62f8a-ce07-5d93-a601-636ee71acc75", + "type": "text", + "content": "We generate a significant portion of our revenues from advertising. Reduced spending by advertisers, a loss of partners, shifts in online advertising, new and evolving advertising formats, or new or existing technologies that block ads online or affect our ability to personalize ads could harm our business.\nWe generated more than 70% of total revenues from online advertising in 2025. Many of our advertisers, companies that distribute our products and services, digital publishers, and content providers can terminate their contracts with us at any time. These partners may not continue to do business with us if we do not create more value (such as increased numbers of users or customers, new sales leads, increased brand awareness, or more effective monetization) than their available alternatives.\nWe believe AI is quickly reshaping the advertising industry, including how ads are delivered online, and we and our competitors are constantly adjusting to meet this shift and provide new and evolving advertising formats. There is no assurance that we will adapt effectively and competitively to meet this shift, and that such advertising formats, strategies, and offerings will be successful.\nChanges to our advertising policies and data privacy practices, as well as changes to other companies' advertising or data privacy practices have in the past, and may in the future, affect the advertising services that we are able to provide. In addition, technologies have been developed that make personalized ads more difficult, or that block the display of ads altogether, and some providers of online services have integrated technologies that could impair the availability and functionality of third-party digital advertising. Failing to provide superior value or deliver advertisements effectively and competitively could harm our business, reputation, financial condition, and operating results.\nExpenditures by advertisers tend to correlate with overall economic conditions. Adverse macroeconomic conditions have affected, and may in the future affect, the demand for advertising, resulting in fluctuations in the amounts our advertisers spend on advertising, which could harm our financial condition and operating results.\nOur increasing investment in new businesses, products, services, and technologies is inherently risky, and could divert management attention and harm our business, financial condition, and operating results.\nWe have invested and expect to expand our investment in new businesses, products, services, and technologies in a wide range of industries beyond online advertising. The investments that we are making across our businesses — such as building AI-optimized infrastructure, including our custom TPUs, and integrating AI capabilities into new and existing products and services — reflect our ongoing efforts to innovate and provide products and services that are helpful to users, advertisers, publishers, customers, content providers, and distribution partners. Our investments ultimately may not be commercially viable or may not result in an adequate return of capital and, in pursuing new strategies, we may incur unanticipated liabilities.\nTo meet the compute capacity demands of AI training and inference, as well as traditional cloud computing services, we are entering into significant leasing arrangements with third party operators, which may increase costs and operational complexity. We also have a number of large, long-duration commercial agreements, which could increase our liabilities and obligations in the event of nonperformance by us, our counterparties, or vendors. In such nonperformance or an industry downturn, we may incur additional liabilities, have excess capacity that we cannot easily redeploy, and not receive payments from our counterparties or customers.\nWe have invested and expect to significantly expand our investment in property and equipment, including our technical infrastructure, and we expect these assets to benefit our business over their estimated useful lives. Changes in facts and circumstances such as changes to historical asset performance, expected technology advancements, and future network deployment plans could change the period over which we expect to benefit from the asset and impact our financial condition and operating results.\nInnovations in our products and services could also result in changes to user and customer behavior and affect our revenue trends. These endeavors involve significant risks and uncertainties, including diversion of resources and management attention from current operations, different monetization models, and the use of alternative investment, governance, or compensation structures that may fail to adequately align incentives across the company or otherwise accomplish their objectives.\nWithin Google Services, we continue to invest heavily in devices, including our smartphones, home devices, and wearables, which is a highly competitive market with frequent introduction of new products and services, rapid adoption of technological advancements by competitors, increased market saturation in developed countries, short product life cycles, evolving industry standards, continual improvement in performance characteristics, and price and feature sensitivity on the part of consumers and businesses. There can be no assurance we will be able to provide devices that compete effectively.\nWithin Google Cloud, we devote significant resources to develop and deploy our enterprise-ready cloud services, including Google Cloud Platform and Google Workspace, and we are advancing our AI platforms and models to support these tools and technologies, including the development of our custom TPUs and how we deliver them to our customers. We are incurring significant and increasing costs and new liabilities, including contingent liabilities, to build and maintain infrastructure to support cloud computing services, invest in cybersecurity, and hire talent. Meanwhile, our competitors are rapidly developing and deploying cloud-based services and capacity. Pricing and delivery models, which are subject to increasing regulatory scrutiny and requirements, are competitive and constantly evolving, and we may therefore not achieve our business objectives. Further, our business with financial services, healthcare, and public sector customers may present additional risks, including regulatory compliance risks. For instance, we may be subject to government audits and cost reviews, and any failure to comply or any deficiencies found may expose us to legal, financial, and reputational risks. Evolving laws and regulations may require us to make new capital investments, build new products, and seek partners to deliver localized services in other countries, and we may not be able to meet sovereign operating requirements.\nWithin Other Bets, we are investing significantly in areas such as life sciences and transportation, among others. These investment areas face intense competition from large, experienced, and well-funded competitors, and our offerings, many of which involve the development of new and emerging technologies, may not be successful, or be able to compete effectively or operate at sufficient levels of profitability.\nIn addition, new and evolving products and services, including those that use AI, raise ethical, technological, legal, regulatory, and other challenges, which could harm our brands and demand for our products and services. Because all of these investment areas are inherently risky, no assurance can be given that such strategies and offerings will be successful or will not harm our reputation, financial condition, and operating results.\nWe face intense competition. If we do not continue to innovate and provide products and services that are useful to users, customers, and other partners, we may not remain competitive, which could harm our business, financial condition, and operating results.\nOur business environment is rapidly evolving and intensely competitive. Our businesses face changing technologies, shifting user needs, and frequent introductions of rival products and services. To compete successfully, we must accurately anticipate technology developments and deliver innovative, relevant, and useful products, services, and technologies in a timely manner. For example, with the rise of AI in recent years, we have increasingly focused our investments in building powerful AI tools and AI enhancements to our existing products and services to better cater to our users, customers, and other partners. As our businesses evolve, the competitive pressure to innovate will encompass a wider range of products and services. We must continue to invest significant resources in technical infrastructure, including the development of our custom TPUs, and research and development, including through acquisitions, in order to enhance our technology, products, and services.\nWe have many competitors in different industries. Our current and potential domestic and international competitors range from large and established companies to emerging start-ups. Some competitors have longer operating histories and well-established relationships in various sectors. They can use their experience and resources in ways that could affect our competitive position, including by making acquisitions and entering into other strategic arrangements; continuing to invest heavily in technical infrastructure, research and development, and in talent; initiating intellectual property and competition claims (whether or not meritorious); and continuing to compete for users, advertisers, customers, and content providers. Further, discrepancies in enforcement of existing laws may enable our lesser known competitors to aggressively interpret those laws without commensurate scrutiny, thereby affording them competitive advantages. Our competitors may also be able to innovate and provide products and services faster or more cost effectively than we can or may foresee the need for products and services before we do.\nWe are developing frontier generative AI models and building AI capabilities into products and services across the company, tailoring them to the evolving preferences of our users and customers. AI technology and services are highly competitive, rapidly evolving, and require significant investment, including technical infrastructure, development, and operational costs. Our ability to deploy certain AI technologies critical for our products and services and for our business strategy may depend on the availability and pricing of third-party equipment and other technical infrastructure operations costs, including network capacity, energy, and equipment costs. Additionally, other companies may develop\nAI products and technologies that are similar or superior to our technologies or more cost-effective to develop or deploy. Other companies may also have (or in the future may obtain) patents or other proprietary rights that would prevent, limit, or interfere with our ability to make, use, or sell our own AI products and services.\nOur financial condition and operating results may also suffer if our products and services are not responsive in a timely manner to the evolving needs and desires of our users, advertisers, publishers, customers, and content providers, or if we miscalculate those needs and desires and invest significantly in areas that fail to gain sufficient market traction. As new and existing technologies continue to develop, competitors and new entrants may be able to offer experiences that are, or that are perceived to be, substantially similar to or better than ours, and the consumers may change how they obtain information online, potentially reducing the utility of our existing products and services. These technologies could reduce usage of our products and services, and force us to compete in different ways and expend significant resources to develop and operate equal or better products and services. Competitors' success in providing compelling products and services or in attracting and retaining users, advertisers, publishers, customers, and content providers could harm our financial condition and operating results.\nOur revenue growth rate could decline over time, and we may experience downward pressure on our operating margin in the future.\nOur revenue growth rate could decline over time as a result of a number of factors, including changes in customer usage and demand for our existing products and increasing demand for competing technologies; changes in the devices and modalities used to access our products and services; changes in geographic mix; deceleration or declines in advertiser spending; competition; decreases in the pricing of our products and services; ongoing product and policy changes; and shifts to lower priced products and services.\nWe may experience downward pressure on our operating margin resulting from a variety of factors. These include increasing costs for many aspects of our business resulting from a higher level of investment in technical infrastructure, increasing regulations, and increasing competition. Certain of our costs and expenses are relatively fixed and may not correlate to changes in revenue, and we may also not be able to continue to drive efficiencies at the rate we have in the past. We may also face margin compression from an increase in the mix of lower-margin products and services, in particular from the continued expansion of our business into new fields, including products and services such as Google Cloud, our devices, and consumer subscription products, as well as significant investments in Other Bets. For instance, margins on our devices have had, and may continue to have, an adverse effect on our consolidated margins due to pricing pressures and higher cost of sales. Due to these factors and the evolving nature of our business, our historical revenue growth rate and historical operating margin may not be indicative of our future performance. For additional information, see Trends in Our Business and Financial Effect and Revenues and Monetization Metrics in Part II, Item 7 of this Annual Report on Form 10-K.\nOur intellectual property rights are valuable, and any inability to protect them could reduce the value of our products, services, and brands as well as affect our ability to compete.\nOur patents, trademarks, trade secrets, copyrights, and other intellectual property rights are important assets for us. Various events outside of our control pose a threat to our intellectual property rights, as well as to our products, services, and technologies. For example, effective intellectual property protection may not be available in every country in which our products and services are distributed or made available through the Internet. Also, the efforts we have taken and may take in the future to protect our proprietary rights, including obtaining copyright and patent protections for our important innovations, including AI innovations, may not be sufficient or effective. There is also the possibility that an issued patent may be deemed invalid or unenforceable.\nWe also seek to maintain certain intellectual property as trade secrets. The confidentiality of such trade secrets and other sensitive information could be compromised, which could cause us to lose the competitive advantage resulting from these trade secrets. We also face risks associated with our trademarks. For example, there is a risk that the word \"Google\" could become so commonly used that it becomes synonymous with the word \"search.\" Some courts have ruled that \"Google\" is a protectable trademark, but it is possible that other courts, particularly those outside of the US, may reach a different determination. If this happens, we could lose protection for this trademark.\nAny significant impairment of our intellectual property rights could harm our business and ability to compete. Protecting our intellectual property rights is costly and time consuming; any increase in unauthorized use could make it more expensive to do business and harm our financial condition and operating results.\nOur business depends on strong brands, and failing to maintain and enhance our brands would hurt our ability to expand our base of users, advertisers, customers, content providers, and other partners.\nOur strong brands have significantly contributed to the success of our business. Maintaining and enhancing the brands within Google Services, Google Cloud, and Other Bets increases our ability to enter new categories and launch new and innovative products and services that better serve the needs of our users, advertisers, customers, content providers, and other partners. Our brands have been, and may in the future be, harmed by a number of factors, including, among others, reputational issues, third-party content shared on our platforms, data privacy and security issues and developments, issues in delivering age-appropriate experiences to minors, and product or technical performance failures. For example, if we fail to respond appropriately to the sharing of misinformation or objectionable content on our services or products or objectionable practices by advertisers, or otherwise to adequately address user concerns, our users may lose confidence in our brands. Furthermore, failure to maintain and enhance our brands could harm our business, reputation, financial condition, and operating results. Our success will depend largely on our ability to remain a technology leader and continue to provide high-quality, trustworthy, innovative products and services that are truly useful and play a valuable role in a range of settings.\nWe face a number of manufacturing and supply chain risks that could affect our ability to supply our products and services and harm our business, financial condition, and operating results.\nWe rely on contract manufacturers to manufacture or assemble our devices as well as servers and networking equipment used in our technical infrastructure, certain components of which we may supply. We also rely on third parties to supply components and distribute our products and services. Our business could be harmed if we are not able to engage these companies with the necessary capabilities or capacity on reasonable terms, or if those we engage fail to meet their obligations (whether due to financial difficulties or other reasons), or make adverse changes in the pricing or other material terms of our arrangements with them.\nWe have experienced and may in the future experience supply shortages, price increases, quality issues, or longer lead times that could harm our operations, driven by raw material or component availability, manufacturing capacity, labor shortages, industry allocations, logistics capacity, inflation, foreign currency exchange rates, tariffs, sanctions and export controls, trade disputes and barriers, forced labor concerns, sourcing requirements, geopolitical tensions, armed conflicts, natural disasters or pandemics, the effects of climate change, power and transmission availability, and significant changes in the financial or business condition of our suppliers. Some of the components we use in our technical infrastructure and our devices are available from only one or limited sources, and we may not be able to find replacement vendors on favorable terms in the event of a supply chain disruption. A significant supply interruption that affects us or our vendors could delay critical data center or network infrastructure upgrades or expansions and delay consumer product availability.\nOur ability to scale our technical infrastructure is increasingly constrained by the availability of power, water, and land. For example, energy supply is constrained globally due to the significant increase in demand for and limited availability of energy to power AI compute. Securing this capacity involves entering into complex, long-lead-time arrangements. Additionally, manufacturing and supply of servers and network equipment for our technical infrastructure, particularly for specialized AI chips, is limited to a small number of qualified suppliers. Extended or unforeseen disruptions at these suppliers could impact our ability to meet customer demand. Failure to secure sufficient capacity in a timely manner would limit our ability to train models and serve Cloud customers.\nWe may enter into long-term contracts for materials and products that commit us to significant terms and conditions. We may face costs for materials and products that are not consumed due to market demand, technological change, excess or obsolete inventory, changed consumer preferences, quality, product recalls, and warranty issues. Certain of our competitors may negotiate more favorable contractual terms based on volume and other commitments that may provide them with competitive advantages and may affect our supply. For example, industry supply capacity for AI accelerators, including GPUs as well as our custom-built TPUs, is highly competitive and rapidly evolving. If we are unable to negotiate favorable contractual terms or our competitors claim the supply or capacity first, we may face increased costs and supply constraints, which could harm our business, financial condition, and operating results.\nOur devices have had, and in the future may have, quality issues resulting from design, manufacturing, or operations. Sometimes, these issues may be caused by components we purchase from other manufacturers or suppliers. In addition, quality issues with equipment used in our technical infrastructure could constrain our capacity to support the delivery and continued development of our products and services. If the quality of our products and services does not meet expectations, we lack the capacity to deliver them, or our products or services are defective or require a corrective action or recall, it could harm our business, reputation, financial condition, and operating results.\nWe require our suppliers and business partners to comply with laws and, where applicable, our company policies and practices, such as the Google Supplier Code of Conduct, regarding workplace and employment practices, data security, environmental compliance, and intellectual property licensing, but we do not control them or their practices. Violations of law or unethical business practices could result in supply chain disruptions, canceled orders, harm to key relationships, and damage to our reputation. Their failure to procure necessary license rights to intellectual property could affect our ability to sell our products or services and expose us to litigation or financial claims.\nInterruption to, interference with, or failure of our complex information technology and communications systems could hurt our ability to effectively provide our products and services, which could harm our reputation, financial condition, and operating results.\nThe availability of our products and services and fulfillment of our customer contracts depend on the continuing operation of our information technology and communications systems. Our systems are vulnerable to damage, interference, or interruption from modifications or upgrades, terrorist attacks, state-sponsored attacks, natural disasters or pandemics, geopolitical tensions or armed conflicts, export controls and sanctions, tariffs and non-tariff trade barriers, the effects of climate change, power and transmission availability challenges, utility outages, telecommunications failures, computer viruses, software bugs, cyber attacks, supply-chain attacks, computer denial of service attacks, phishing schemes, or other attempts to harm or access our systems. Some of our data centers are located in areas with a high risk of major earthquakes or other natural disasters. Our data centers are also subject to break-ins, sabotage, and intentional acts of vandalism, and, in some cases, to potential disruptions resulting from problems experienced by facility operators or disruptions as a result of geopolitical tensions and conflicts happening in the area. Some of our systems are not fully redundant, and disaster recovery planning cannot account for all eventualities. The occurrence of a natural disaster or pandemic, closure of a facility, or other unanticipated problems affecting our data centers could result in lengthy interruptions in our service. In addition, our products and services are highly technical and complex and have contained in the past, and may contain in the future, errors or vulnerabilities, which could result in interruptions in or failure of our services or systems. Any of these incidents could impede or prevent us from effectively offering products and providing services, which could harm our business, reputation, financial condition, and operating results.\nOur international operations expose us to additional risks that could harm our business, reputation, financial condition, and operating results.\nOur international operations are significant to our revenues and net income, and we plan to continue growing internationally. International revenues accounted for approximately 52% of consolidated revenues in 2025. In addition to risks described elsewhere in this section, our international operations expose us to other risks, including:\n- restrictions on foreign ownership and investments, and stringent foreign exchange controls that might prevent us from repatriating cash earned in countries outside the US;\n- sanctions, tariffs, import and export controls, other market access barriers, political unrest, geopolitical tensions, changes in regimes, or armed conflict (such as ongoing conflicts in the Middle East and Ukraine), any of which may affect our business continuity, increase our operating costs, limit demand for our products and services, limit our ability to source components or final products, or prevent or impede us from operating in certain jurisdictions, complying with local laws, or offering products or services;\n- longer payment cycles in some countries, increased credit risk, and higher levels of payment fraud;\n- an evolving foreign policy landscape that could harm our revenues and could subject us to litigation, new regulatory costs and challenges (including new customer requirements), uncertainty regarding regulatory outcomes, and other liabilities under local laws that may not offer due process or clear legal precedent;\n- anti-corruption laws, such as the US Foreign Corrupt Practices Act, and other local laws prohibiting certain payments to government officials, violations of which could result in civil and criminal penalties; and\n- different employee/employer relationships; different labor laws, regulations, and labor practices; and other challenges caused by distance, language, local expertise, and cultural differences, increasing the complexity of doing business in multiple jurisdictions.\nBecause we conduct business in currencies other than US dollars but report our financial results in US dollars, we have faced, and will continue to face, exposure to fluctuations in foreign currency exchange rates. Although we hedge a portion of our international currency exposure, significant fluctuations in exchange rates between the US dollar and foreign currencies have and may in the future adversely affect our revenues and earnings. Hedging programs are also inherently risky and could expose us to additional risks that could harm our financial condition and operating results.\nDisruptions in our ability to access future financing or manage our indebtedness could adversely affect our ability to execute our strategy and harm our financial condition.\nWe may from time to time access capital markets for debt or seek to enter into other forms of financing, such as leases. Any difficulty in accessing capital markets, entering into other forms of financing on favorable terms, or managing our existing indebtedness could increase our costs of financing and restrict our ability to invest in our business. Furthermore, the combination of our current and any future indebtedness, including obligations arising under leases, backstops, guarantees, and potential liabilities from large commercial agreements, could harm our financial condition and reduce our financial and business flexibility.\nWe are exposed to fluctuations in the fair values of our investments and, in some instances, our financial statements incorporate inherently subjective valuation methodologies.\nThe fair value of our debt and equity investments may in the future be, and certain investments have been in the past, negatively affected by liquidity, credit deterioration or losses, performance and financial results of the underlying entities, foreign exchange rates, changes in interest rates, the effect of new or changing regulations, the stock market in general, or other factors. As a result of these factors, the value of our investments could significantly decline, which could harm our financial condition and operating results.\nWe measure certain of our non-marketable equity and debt securities, and certain other instruments including stock-based compensation (SBC) awards of Other Bet companies, at fair value on a nonrecurring basis, which is inherently subjective and requires management judgment and estimation. Gains and losses on non-marketable equity securities are recognized in other income (expense), net (OI&E), which increases the volatility of our OI&E. The unrealized gains and losses or impairments we record from fair value remeasurements in any particular period may differ significantly from the gains and losses we ultimately realize on such investments. Changes in fair value on SBC awards are recognized primarily through operating expenses.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS/Risks Specific to our Company", + "metadata": { + "length": 29834, + "summary": "The passage outlines significant risks to the company's business, primarily from its heavy reliance on advertising revenue (over 70% in 2025), which is threatened by reduced advertiser spending, loss of partners, and new ad-blocking technologies. Intense competition, especially in AI, requires substantial investment in infrastructure like custom TPUs, with no guarantee of success. The company also faces risks from supply chain disruptions, intellectual property protection, brand reputation, international operations, and the inherent uncertainty of investing in new businesses and technologies.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "generate", + "significant", + "portion", + "revenues", + "advertising", + "Reduced", + "spending", + "advertisers", + "loss", + "partners", + "shifts", + "online", + "evolving", + "formats", + "existing", + "technologies", + "block", + "ads", + "affect", + "ability", + "personalize", + "harm", + "business", + "generated", + "70%", + "total", + "2025", + "Many", + "companies", + "distribute", + "products", + "services", + "digital", + "publishers", + "content", + "providers", + "terminate", + "contracts", + "time", + "These", + "continue", + "create", + "increased", + "numbers", + "users", + "customers", + "sales", + "leads", + "brand", + "awareness", + "effective", + "monetization", + "alternatives", + "AI", + "quickly", + "reshaping", + "industry", + "including", + "delivered", + "competitors", + "constantly", + "adjusting", + "meet", + "shift", + "provide", + "There", + "assurance", + "adapt", + "effectively", + "competitively", + "strategies", + "offerings", + "successful", + "Changes", + "policies", + "data", + "privacy", + "practices", + "past", + "future", + "In", + "addition", + "developed", + "make", + "personalized", + "difficult", + "display", + "altogether", + "integrated", + "impair", + "availability", + "functionality", + "party", + "Failing", + "superior", + "deliver", + "advertisements", + "reputation", + "financial", + "condition", + "operating", + "results", + "Expenditures", + "tend", + "correlate", + "economic", + "conditions", + "Adverse", + "macroeconomic", + "affected", + "demand", + "resulting", + "fluctuations", + "amounts", + "spend", + "Our", + "increasing", + "investment", + "businesses", + "inherently", + "risky", + "divert", + "management", + "attention", + "invested", + "expect", + "expand", + "wide", + "range", + "industries", + "The", + "investments", + "making", + "building", + "optimized", + "infrastructure", + "custom", + "TPUs", + "integrating", + "capabilities", + "reflect", + "ongoing", + "efforts", + "innovate", + "helpful", + "distribution", + "ultimately", + "commercially", + "viable", + "result", + "adequate", + "return", + "capital", + "pursuing", + "incur", + "unanticipated", + "liabilities", + "To", + "compute", + "capacity", + "demands", + "training", + "inference", + "traditional", + "cloud", + "computing", + "entering", + "leasing", + "arrangements", + "operators", + "increase", + "costs", + "operational", + "complexity", + "number", + "large", + "long", + "duration", + "commercial", + "agreements", + "obligations", + "event", + "nonperformance", + "counterparties", + "vendors", + "downturn", + "additional", + "excess", + "easily", + "redeploy", + "receive", + "payments", + "significantly", + "property", + "equipment", + "technical", + "assets", + "benefit", + "estimated", + "lives", + "facts", + "circumstances", + "historical", + "asset", + "performance", + "expected", + "technology", + "advancements", + "network", + "deployment", + "plans", + "change", + "period", + "impact", + "Innovations", + "user", + "customer", + "behavior", + "revenue", + "trends", + "endeavors", + "involve", + "risks", + "uncertainties", + "diversion", + "resources", + "current", + "operations", + "models", + "alternative", + "governance", + "compensation", + "structures", + "fail", + "adequately", + "align", + "incentives", + "company", + "accomplish", + "objectives", + "Within", + "Google", + "Services", + "invest", + "heavily", + "devices", + "smartphones", + "home", + "wearables", + "highly", + "competitive", + "market", + "frequent", + "introduction", + "rapid", + "adoption", + "technological", + "saturation", + "countries", + "short", + "product", + "life", + "cycles", + "standards", + "continual", + "improvement", + "characteristics", + "price", + "feature", + "sensitivity", + "part", + "consumers", + "compete", + "Cloud", + "devote", + "develop", + "deploy", + "enterprise", + "ready", + "Platform", + "Workspace", + "advancing", + "platforms", + "support", + "tools", + "development", + "incurring", + "contingent", + "build", + "maintain", + "cybersecurity", + "hire", + "talent", + "Meanwhile", + "rapidly", + "developing", + "deploying", + "based", + "Pricing", + "delivery", + "subject", + "regulatory", + "scrutiny", + "requirements", + "achieve", + "Further", + "healthcare", + "public", + "sector", + "present", + "compliance", + "For", + "instance", + "government", + "audits", + "cost", + "reviews", + "failure", + "comply", + "deficiencies", + "found", + "expose", + "legal", + "reputational", + "Evolving", + "laws", + "regulations", + "require", + "seek", + "localized", + "sovereign", + "Other", + "Bets", + "investing", + "areas", + "sciences", + "transportation", + "face", + "intense", + "competition", + "experienced", + "funded", + "emerging", + "operate", + "sufficient", + "levels", + "profitability", + "raise", + "ethical", + "challenges", + "brands", + "Because", + "If", + "remain", + "environment", + "intensely", + "changing", + "shifting", + "introductions", + "rival", + "successfully", + "accurately", + "anticipate", + "developments", + "innovative", + "relevant", + "timely", + "manner", + "rise", + "recent", + "years", + "increasingly", + "focused", + "powerful", + "enhancements", + "cater", + "As", + "evolve", + "pressure", + "encompass", + "wider", + "research", + "acquisitions", + "order", + "enhance", + "potential", + "domestic", + "international", + "established", + "start", + "ups", + "Some", + "longer", + "histories", + "relationships", + "sectors", + "They", + "experience", + "ways", + "position", + "strategic", + "continuing", + "initiating", + "intellectual", + "claims", + "meritorious", + "discrepancies", + "enforcement", + "enable", + "lesser", + "aggressively", + "interpret", + "commensurate", + "affording", + "advantages", + "faster", + "foresee", + "frontier", + "generative", + "tailoring", + "preferences", + "critical", + "strategy", + "depend", + "pricing", + "energy", + "Additionally", + "similar", + "obtain", + "patents", + "proprietary", + "rights", + "prevent", + "limit", + "interfere", + "sell", + "suffer", + "responsive", + "desires", + "miscalculate", + "gain", + "traction", + "entrants", + "offer", + "experiences", + "perceived", + "substantially", + "information", + "potentially", + "reducing", + "utility", + "reduce", + "usage", + "force", + "expend", + "equal", + "Competitors", + "success", + "providing", + "compelling", + "attracting", + "retaining", + "growth", + "rate", + "decline", + "downward", + "margin", + "factors", + "competing", + "modalities", + "access", + "geographic", + "mix", + "deceleration", + "declines", + "advertiser", + "decreases", + "policy", + "lower", + "priced", + "variety", + "include", + "aspects", + "higher", + "level", + "Certain", + "expenses", + "fixed", + "drive", + "efficiencies", + "compression", + "continued", + "expansion", + "fields", + "consumer", + "subscription", + "margins", + "adverse", + "effect", + "consolidated", + "due", + "pressures", + "Due", + "nature", + "indicative", + "Trends", + "Business", + "Financial", + "Effect", + "Revenues", + "Monetization", + "Metrics", + "Part", + "II", + "Item", + "Annual", + "Report", + "Form", + "10", + "valuable", + "inability", + "protect", + "trademarks", + "trade", + "secrets", + "copyrights", + "important", + "Various", + "events", + "control", + "pose", + "threat", + "protection", + "country", + "distributed", + "made", + "Internet", + "Also", + "obtaining", + "copyright", + "patent", + "protections", + "innovations", + "possibility", + "issued", + "deemed", + "invalid", + "unenforceable", + "confidentiality", + "sensitive", + "compromised", + "lose", + "advantage", + "risk", + "word", + "commonly", + "synonymous", + "search", + "courts", + "ruled", + "protectable", + "trademark", + "US", + "reach", + "determination", + "Any", + "impairment", + "Protecting", + "costly", + "consuming", + "unauthorized", + "expensive", + "depends", + "strong", + "failing", + "hurt", + "base", + "contributed", + "Maintaining", + "enhancing", + "increases", + "enter", + "categories", + "launch", + "serve", + "harmed", + "issues", + "shared", + "security", + "delivering", + "age", + "minors", + "failures", + "respond", + "appropriately", + "sharing", + "misinformation", + "objectionable", + "address", + "concerns", + "confidence", + "Furthermore", + "largely", + "leader", + "high", + "quality", + "trustworthy", + "play", + "role", + "settings", + "manufacturing", + "supply", + "chain", + "rely", + "contract", + "manufacturers", + "manufacture", + "assemble", + "servers", + "networking", + "components", + "parties", + "engage", + "reasonable", + "terms", + "difficulties", + "reasons", + "material", + "shortages", + "lead", + "times", + "driven", + "raw", + "component", + "labor", + "allocations", + "logistics", + "inflation", + "foreign", + "currency", + "exchange", + "rates", + "tariffs", + "sanctions", + "export", + "controls", + "disputes", + "barriers", + "forced", + "sourcing", + "geopolitical", + "tensions", + "armed", + "conflicts", + "natural", + "disasters", + "pandemics", + "effects", + "climate", + "power", + "transmission", + "suppliers", + "limited", + "sources", + "find", + "replacement", + "favorable", + "disruption", + "interruption", + "affects", + "delay", + "center", + "upgrades", + "expansions", + "scale", + "constrained", + "water", + "land", + "globally", + "Securing", + "involves", + "complex", + "specialized", + "chips", + "small", + "qualified", + "Extended", + "unforeseen", + "disruptions", + "Failure", + "secure", + "train", + "term", + "materials", + "commit", + "consumed", + "obsolete", + "inventory", + "changed", + "recalls", + "warranty", + "negotiate", + "contractual", + "volume", + "commitments", + "accelerators", + "GPUs", + "built", + "unable", + "claim", + "constraints", + "design", + "Sometimes", + "caused", + "purchase", + "constrain", + "expectations", + "lack", + "defective", + "corrective", + "action", + "recall", + "applicable", + "Supplier", + "Code", + "Conduct", + "workplace", + "employment", + "environmental", + "licensing", + "Violations", + "law", + "unethical", + "canceled", + "orders", + "key", + "damage", + "Their", + "procure", + "license", + "litigation", + "Interruption", + "interference", + "communications", + "systems", + "fulfillment", + "operation", + "vulnerable", + "modifications", + "terrorist", + "attacks", + "state", + "sponsored", + "tariff", + "outages", + "telecommunications", + "computer", + "viruses", + "software", + "bugs", + "cyber", + "denial", + "service", + "phishing", + "schemes", + "attempts", + "centers", + "located", + "major", + "earthquakes", + "break", + "ins", + "sabotage", + "intentional", + "acts", + "vandalism", + "cases", + "problems", + "facility", + "happening", + "area", + "fully", + "redundant", + "disaster", + "recovery", + "planning", + "account", + "eventualities", + "occurrence", + "pandemic", + "closure", + "affecting", + "lengthy", + "interruptions", + "contained", + "errors", + "vulnerabilities", + "incidents", + "impede", + "offering", + "net", + "income", + "plan", + "growing", + "internationally", + "International", + "accounted", + "approximately", + "52%", + "section", + "restrictions", + "ownership", + "stringent", + "repatriating", + "cash", + "earned", + "import", + "political", + "unrest", + "regimes", + "conflict", + "Middle", + "East", + "Ukraine", + "continuity", + "source", + "final", + "jurisdictions", + "complying", + "local", + "payment", + "credit", + "fraud", + "landscape", + "uncertainty", + "outcomes", + "process", + "clear", + "precedent", + "anti", + "corruption", + "Foreign", + "Corrupt", + "Practices", + "Act", + "prohibiting", + "officials", + "violations", + "civil", + "criminal", + "penalties", + "employee", + "employer", + "distance", + "language", + "expertise", + "cultural", + "differences", + "multiple", + "conduct", + "currencies", + "dollars", + "report", + "faced", + "exposure", + "Although", + "hedge", + "dollar", + "adversely", + "earnings", + "Hedging", + "programs", + "Disruptions", + "financing", + "manage", + "indebtedness", + "execute", + "markets", + "debt", + "forms", + "leases", + "difficulty", + "accessing", + "managing", + "restrict", + "combination", + "arising", + "backstops", + "guarantees", + "flexibility", + "exposed", + "fair", + "values", + "instances", + "statements", + "incorporate", + "subjective", + "valuation", + "methodologies", + "equity", + "negatively", + "liquidity", + "deterioration", + "losses", + "underlying", + "entities", + "interest", + "stock", + "general", + "measure", + "marketable", + "securities", + "instruments", + "SBC", + "awards", + "Bet", + "nonrecurring", + "basis", + "requires", + "judgment", + "estimation", + "Gains", + "recognized", + "expense", + "OI", + "volatility", + "unrealized", + "gains", + "impairments", + "record", + "remeasurements", + "differ", + "realize", + "primarily" + ], + "keywords": [ + "Advertising", + "Competition", + "AI" + ], + "connect_to": [] + } + }, + { + "chunk_id": "0194eebe-cf80-5ed8-80d2-add97c36fb3f", + "type": "text", + "content": "Issues in the development and use of AI may result in reputational harm and increased liability exposure.\nOur evolving AI-enabled products and services may give rise to risks related to harmful content, inaccuracies, discrimination, intellectual property infringement or misappropriation, violation of rights of publicity, defamation, data privacy, cybersecurity, minor protection, and other issues. As a result of these and other challenges associated with innovative technologies, our implementation of AI systems could subject us to competitive harm, regulatory action, legal liability (including under new and proposed legislation and regulations), new applications of existing data protection, privacy, intellectual property, and other laws, and brand or reputational harm.\nSome uses of AI will present ethical issues and may have broad effects on society. In order to implement AI responsibly and minimize unintended harmful effects, we have already devoted and will continue to invest significant resources to develop, test, and maintain our products and services, but we may not be able to identify or resolve all AI-related issues, deficiencies, and failures before they arise. Unintended consequences, uses, or customization of our AI tools and systems may negatively affect human rights, privacy, employment, or other social concerns, which may result in claims, lawsuits, brand or reputational harm, and increased regulatory scrutiny, any of which could harm our business, financial condition, and operating results.\nPeople access our products and services through a variety of platforms and devices that continue to evolve with the advancement of technology and user preferences. If manufacturers and users do not widely adopt versions of our products and services developed for these interfaces, our business could be harmed.\nWhile the modalities used to access information is evolving, people access our products and services through a growing variety of devices such as phones, laptops and tablets, video game consoles, voice-activated speakers, wearables (including virtual reality and augmented reality devices), automobiles, and television-streaming devices. Our products and services may be less popular on some interfaces. Each manufacturer or distributor may establish unique technical standards for its devices, and our products and services may not be available or may only be available with limited functionality for our users or our advertisers on these devices as a result. Some manufacturers may also elect not to include our products on their devices.\nIt is hard to predict the challenges we may encounter in adapting our products and services and developing competitive new products and services. We expect to continue to devote significant resources to creating and supporting products and services across multiple platforms and devices. Failing to attract and retain a substantial number of device manufacturers, suppliers, distributors, developers, and users, or failing to develop products and technologies that work well on new devices and platforms, could harm our business, financial condition, and operating results and ability to capture future business opportunities.\nProblematic content on our platforms, including low-quality user-generated content, web spam, content farms, and other violations of our guidelines could affect the quality of our services, which could harm our reputation and deter our current and potential users from using our products and services.\nWe, like others in the industry, face violations of our content guidelines across our platforms, including sophisticated attempts by bad actors to manipulate our hosting and advertising systems to fraudulently generate revenues, or to otherwise generate traffic that does not represent genuine user interest or intent. While we invest significantly in efforts to promote high-quality and relevant results and to detect and prevent low-quality content and invalid traffic, we have been unable and may continue to be unable to detect and prevent all such abuses or promote uniformly high-quality content. Increased use of AI in our offerings and internal systems may create new instances of problematic content and increased potential for misuse and abuse.\nMany websites violate or attempt to violate our guidelines, including by seeking to inappropriately rank higher in search results than our search engine's assessment of their relevance and utility would rank them. Such efforts have affected, and may continue to affect, the quality of content on our platforms and lead them to display false, misleading, or undesirable content. Although English-language web spam in our search results has been reduced, and web spam in most other languages is limited, we expect web spammers will continue to seek inappropriate ways to improve their rankings. Although we continue to invest in and deploy proprietary technology to detect and prevent web spam on our platforms and to evolve our policies to address emerging threats, there is no guarantee that our technology and policy enforcement will always be successful, and our users may have negative experiences that make them less likely to use our platforms. We face legal and regulatory challenges to our efforts to address low-quality content, and our ability to address it may be constrained or made more costly through added compliance requirements. We also face other challenges to the quality of our search results from low-quality and irrelevant content websites, including content farms, which are websites that generate large quantities of low-quality content in an effort to improve their search rankings. We are continually launching algorithmic changes designed to detect and prevent these efforts, but we may not always be successful. We also face other challenges on our platforms, including attempted election interference; fraud, content inappropriate for minors, misleading or deceptive information, and other types of potentially harmful content.\nIf we fail to either detect and prevent an increase in problematic content or effectively promote high-quality content, it could hurt our reputation for delivering relevant information or reduce use of our platforms, harming our financial condition and operating results. It may also subject us to litigation and regulatory actions, which could result in monetary penalties and damages and divert management's time and attention.\nData privacy and security concerns relating to our technology and our practices could harm our reputation, cause us to incur significant liability, and deter current and potential users or customers from using our products and services. Computer viruses, software bugs or defects, security breaches, and attacks on our systems could result in the improper disclosure and use of user data and interference with our users' and customers' ability to use our products and services, harming our business and reputation.\nConcerns about, including the adequacy of, our practices with regard to the collection, use, governance, disclosure, or security of personal data or other data-privacy-related matters, even if unfounded, could harm our business, reputation, financial condition, and operating results. Our policies and practices may change over time as expectations and regulations regarding privacy and data change.\nOur products and services involve the storage, handling, and transmission of proprietary and other sensitive information. Malicious software such as viruses, software bugs, theft, misuse, defects, vulnerabilities in our products and services, as well as cyber attacks, phishing schemes, and other types of security attacks, which increasingly use AI, expose us to a risk of loss or improper use and disclosure of such information, which could result in litigation and other potential liabilities, including regulatory fines and penalties, as well as reputational harm. Additionally, our products incorporate highly technical and complex technologies, and thus our technologies and software have contained, and are likely in the future to contain, undetected errors, bugs, or vulnerabilities. We continue to add new features involving AI to our offerings and internal systems, and features that rely on AI may be susceptible to unanticipated security threats as our and the market's understanding of AI-centric security risks and protection methods continue to develop. We have in the past discovered, and may in the future discover, some errors in our software code only after we have released the code. Systems and control failures, security breaches, failure to comply with our privacy policies, and inadvertent disclosure of user data could result in regulatory and legal exposure, seriously harm our reputation, brand, and business, and impair our ability to attract and retain users or customers. Such incidents have occurred in the past and may continue to occur due to the scale and nature of our products and services. While there is no guarantee that such incidents will not cause significant damage, we expect to continue to expend significant resources to maintain security protections that limit the effect of bugs, theft, misuse, and security vulnerabilities or breaches.\nWe experience cyber attacks and other attempts to gain unauthorized access to our systems on a regular basis. Cyber attacks continue to evolve in sophistication and volume, and inherently may be difficult to detect for long periods of time. The development and implementation of AI technologies may further increase our exposure to or exacerbate the risks of cyber attacks or other security incidents, particularly where such technologies are exploited by third parties to breach our or other parties' systems, including when such technologies are used to target our employees or impersonate members of senior management in order to gain unauthorized access to our systems. We have also seen, and will continue to see, industry-wide software supply chain vulnerabilities and attacks on telecommunications and other critical infrastructure, which could affect our or other parties' systems. We expect to continue to experience such incidents or vulnerabilities in the future. Our efforts to prevent security incidents and address undesirable activity on our platform may require us to spend additional resources to prepare and defend against such threats, and could also increase the risk of retaliatory attack. In addition, we face the risk of cyber attacks and data exfiltration or compromise by nation-states and state-sponsored actors. These attacks may target us or our customers, particularly our public sector customers (including federal, state, and local governments). Geopolitical tensions or armed conflicts, such as the ongoing conflict in the Middle East and Ukraine, may increase these risks.\nWe may experience security and privacy issues, whether due to employee or insider error or malfeasance, system errors, or vulnerabilities in our or other parties' systems. While we may not determine some of these issues to be material when they occur and may remedy them quickly, there is no guarantee that these issues will not ultimately result in significant legal, financial, and reputational harm, including government inquiries, enforcement actions, litigation, and negative publicity. There is also no guarantee that a series of related issues may not be determined to be material at a later date in the aggregate, even if they may not be material individually at the time of their occurrence. Because the techniques used to obtain unauthorized access to, disable, or degrade service provided by or otherwise sabotage systems change frequently and often are recognized only after being launched against a target, even taking all reasonable precautions, including those required by law, we have been unable in the past and may continue to be unable to anticipate or detect certain attacks or vulnerabilities or implement adequate preventative measures.\nFurther, if any partners with whom we share user or other customer information fail to implement adequate data-security practices, fail to comply with our terms and policies, or otherwise suffer a network or other security breach, our users' data may be improperly accessed, used, or disclosed. If an actual or perceived breach of our or our business partners' or service providers' security occurs, the market perception of the effectiveness of our security measures would be harmed, we could lose users and customers, our trade secrets or those of our business partners may be compromised, and we may be exposed to significant legal and financial risks, including legal claims (which may include class-action litigation) and regulatory actions, fines, and penalties. Any of the foregoing consequences could harm our business, reputation, financial condition, and operating results.\nWhile we have dedicated significant resources to privacy and security incident response capabilities, including dedicated worldwide incident response teams, our response process, particularly during times of a natural disaster or pandemic, may not be adequate, may fail to accurately assess the severity of an incident, may not be fast enough to prevent or limit harm, or may fail to sufficiently remediate an incident. As a result, we may suffer significant legal, reputational, or financial exposure, which could harm our business, financial condition, and operating results.\nFor additional information, see also our risk factor on privacy and data protection regulations under 'Risks Related to Laws, Regulations, and Policies' below.\nOur ongoing investments in safety, security, and content review will likely continue to identify abuse of our platforms and misuse of user data.\nIn addition to our efforts to prevent and mitigate cyber attacks, we are making significant investments in safety, security, and review efforts to combat misuse of our services and unauthorized access to user data by third parties, including investigation and review of platform applications that could access the information of users of our services. As a result of these efforts, we have in the past discovered, and may in the future discover, incidents of unnecessary access to or misuse of user data or other undesirable activity by third parties. However, we may not have discovered, and may in the future not discover, all such incidents or activity, whether as a result of our data limitations, including our lack of visibility over our encrypted services, the scale of activity on our platform, or other factors, including factors outside of our control such as a natural disaster or pandemic, and we may learn of such incidents or activity via third parties. Such incidents and activities may include the use of user data or our systems in a manner inconsistent with our terms, contracts, or policies, the existence of false or undesirable user accounts, election interference, improper ad purchases, activities that threaten people's safety on- or off-line, or instances of spamming, scraping, or spreading disinformation. While we may not determine some of these incidents to be material at the time they occurred and we may remedy them quickly, there is no guarantee that these issues will not ultimately result in significant legal, financial, and reputational harm, including government inquiries and enforcement actions, litigation, and negative publicity. There is also no guarantee that a series of related issues may not be determined to be material at a later date in the aggregate, even if they may not be material individually at the time of their occurrence.\nWe may also be unsuccessful in our efforts to enforce our policies or otherwise prevent or remediate any such incidents. Any of the foregoing developments may negatively affect user trust and engagement, harm our reputation and brands, require us to change our business practices in ways that harm our business operations, and adversely affect our business and financial results. Any such developments may also subject us to additional litigation and regulatory inquiries, which could result in monetary penalties and damages, divert management's time and attention, and lead to enhanced regulatory oversight.\nOur business depends on continued and unimpeded access to the Internet by us and our users. Internet access providers may be able to restrict, block, degrade, or charge for access to certain of our products and services, which could lead to additional expenses and the loss of users and advertisers.\nOur products and services depend on the ability of our users to access the Internet, and certain of our products require significant network capacity to work effectively. Currently, this access is provided by companies that have significant market power in the broadband and internet access marketplace, including incumbent telephone companies, cable companies, mobile communications companies, and government-owned service providers. Some of these providers have taken, or have stated that they may take, measures that could degrade, disrupt, or increase the cost of user access to certain of our products by restricting or prohibiting the use of their infrastructure to support or facilitate our offerings, by charging increased fees to us or our users to provide our offerings, or by providing our competitors preferential access. Some jurisdictions have adopted regulations prohibiting certain forms of discrimination by internet access providers; however, substantial uncertainty exists in the US and elsewhere regarding such protections. In addition, in some jurisdictions, our products and services have been subject to government-initiated restrictions or blockages. These could harm existing key relationships, including with our users, customers, advertisers, and content providers, and impair our ability to attract new ones; harm our reputation; and increase costs, thereby negatively affecting our business.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS/Risks Related to our Industry", + "metadata": { + "length": 17994, + "summary": "The passage outlines significant risks associated with developing and using AI, including reputational harm, legal liability, and regulatory scrutiny due to issues like harmful content, inaccuracies, discrimination, and intellectual property violations. It highlights challenges in implementing AI responsibly, the impact of problematic content on platforms, and the evolving landscape of device access. Data privacy and security concerns are emphasized, with risks from cyber attacks, software bugs, and unauthorized access potentially leading to litigation, fines, and loss of user trust. The text also discusses the dependence on internet access providers and the potential for restrictions or increased costs, which could harm business operations and financial results.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Issues", + "development", + "AI", + "result", + "reputational", + "harm", + "increased", + "liability", + "exposure", + "Our", + "evolving", + "enabled", + "products", + "services", + "give", + "rise", + "risks", + "related", + "harmful", + "content", + "inaccuracies", + "discrimination", + "intellectual", + "property", + "infringement", + "misappropriation", + "violation", + "rights", + "publicity", + "defamation", + "data", + "privacy", + "cybersecurity", + "minor", + "protection", + "issues", + "As", + "challenges", + "innovative", + "technologies", + "implementation", + "systems", + "subject", + "competitive", + "regulatory", + "action", + "legal", + "including", + "proposed", + "legislation", + "regulations", + "applications", + "existing", + "laws", + "brand", + "Some", + "present", + "ethical", + "broad", + "effects", + "society", + "In", + "order", + "implement", + "responsibly", + "minimize", + "unintended", + "devoted", + "continue", + "invest", + "significant", + "resources", + "develop", + "test", + "maintain", + "identify", + "resolve", + "deficiencies", + "failures", + "arise", + "Unintended", + "consequences", + "customization", + "tools", + "negatively", + "affect", + "human", + "employment", + "social", + "concerns", + "claims", + "lawsuits", + "scrutiny", + "business", + "financial", + "condition", + "operating", + "results", + "People", + "access", + "variety", + "platforms", + "devices", + "evolve", + "advancement", + "technology", + "user", + "preferences", + "If", + "manufacturers", + "users", + "widely", + "adopt", + "versions", + "developed", + "interfaces", + "harmed", + "While", + "modalities", + "information", + "people", + "growing", + "phones", + "laptops", + "tablets", + "video", + "game", + "consoles", + "voice", + "activated", + "speakers", + "wearables", + "virtual", + "reality", + "augmented", + "automobiles", + "television", + "streaming", + "popular", + "Each", + "manufacturer", + "distributor", + "establish", + "unique", + "technical", + "standards", + "limited", + "functionality", + "advertisers", + "elect", + "include", + "It", + "hard", + "predict", + "encounter", + "adapting", + "developing", + "We", + "expect", + "devote", + "creating", + "supporting", + "multiple", + "Failing", + "attract", + "retain", + "substantial", + "number", + "device", + "suppliers", + "distributors", + "developers", + "failing", + "work", + "ability", + "capture", + "future", + "opportunities", + "Problematic", + "low", + "quality", + "generated", + "web", + "spam", + "farms", + "violations", + "guidelines", + "reputation", + "deter", + "current", + "potential", + "industry", + "face", + "sophisticated", + "attempts", + "bad", + "actors", + "manipulate", + "hosting", + "advertising", + "fraudulently", + "generate", + "revenues", + "traffic", + "represent", + "genuine", + "interest", + "intent", + "significantly", + "efforts", + "promote", + "high", + "relevant", + "detect", + "prevent", + "invalid", + "unable", + "abuses", + "uniformly", + "Increased", + "offerings", + "internal", + "create", + "instances", + "problematic", + "misuse", + "abuse", + "Many", + "websites", + "violate", + "attempt", + "seeking", + "inappropriately", + "rank", + "higher", + "search", + "engine", + "assessment", + "relevance", + "utility", + "Such", + "affected", + "lead", + "display", + "false", + "misleading", + "undesirable", + "Although", + "English", + "language", + "reduced", + "languages", + "spammers", + "seek", + "inappropriate", + "ways", + "improve", + "rankings", + "deploy", + "proprietary", + "policies", + "address", + "emerging", + "threats", + "guarantee", + "policy", + "enforcement", + "successful", + "negative", + "experiences", + "make", + "constrained", + "made", + "costly", + "added", + "compliance", + "requirements", + "irrelevant", + "large", + "quantities", + "effort", + "continually", + "launching", + "algorithmic", + "designed", + "attempted", + "election", + "interference", + "fraud", + "minors", + "deceptive", + "types", + "potentially", + "fail", + "increase", + "effectively", + "hurt", + "delivering", + "reduce", + "harming", + "litigation", + "actions", + "monetary", + "penalties", + "damages", + "divert", + "management", + "time", + "attention", + "Data", + "security", + "relating", + "practices", + "incur", + "customers", + "Computer", + "viruses", + "software", + "bugs", + "defects", + "breaches", + "attacks", + "improper", + "disclosure", + "Concerns", + "adequacy", + "regard", + "collection", + "governance", + "personal", + "matters", + "unfounded", + "change", + "expectations", + "involve", + "storage", + "handling", + "transmission", + "sensitive", + "Malicious", + "theft", + "vulnerabilities", + "cyber", + "phishing", + "schemes", + "increasingly", + "expose", + "risk", + "loss", + "liabilities", + "fines", + "Additionally", + "incorporate", + "highly", + "complex", + "contained", + "undetected", + "errors", + "add", + "features", + "involving", + "rely", + "susceptible", + "unanticipated", + "market", + "understanding", + "centric", + "methods", + "past", + "discovered", + "discover", + "code", + "released", + "Systems", + "control", + "failure", + "comply", + "inadvertent", + "impair", + "incidents", + "occurred", + "occur", + "due", + "scale", + "nature", + "damage", + "expend", + "protections", + "limit", + "effect", + "experience", + "gain", + "unauthorized", + "regular", + "basis", + "Cyber", + "sophistication", + "volume", + "inherently", + "difficult", + "long", + "periods", + "The", + "exacerbate", + "exploited", + "parties", + "breach", + "target", + "employees", + "impersonate", + "members", + "senior", + "wide", + "supply", + "chain", + "telecommunications", + "critical", + "infrastructure", + "activity", + "platform", + "require", + "spend", + "additional", + "prepare", + "defend", + "retaliatory", + "attack", + "addition", + "exfiltration", + "compromise", + "nation", + "states", + "state", + "sponsored", + "These", + "public", + "sector", + "federal", + "local", + "governments", + "Geopolitical", + "tensions", + "armed", + "conflicts", + "ongoing", + "conflict", + "Middle", + "East", + "Ukraine", + "employee", + "insider", + "error", + "malfeasance", + "system", + "determine", + "material", + "remedy", + "quickly", + "ultimately", + "government", + "inquiries", + "There", + "series", + "determined", + "date", + "aggregate", + "individually", + "occurrence", + "Because", + "techniques", + "obtain", + "disable", + "degrade", + "service", + "provided", + "sabotage", + "frequently", + "recognized", + "launched", + "taking", + "reasonable", + "precautions", + "required", + "law", + "anticipate", + "adequate", + "preventative", + "measures", + "Further", + "partners", + "share", + "customer", + "terms", + "suffer", + "network", + "improperly", + "accessed", + "disclosed", + "actual", + "perceived", + "providers", + "occurs", + "perception", + "effectiveness", + "lose", + "trade", + "secrets", + "compromised", + "exposed", + "class", + "Any", + "foregoing", + "dedicated", + "incident", + "response", + "capabilities", + "worldwide", + "teams", + "process", + "times", + "natural", + "disaster", + "pandemic", + "accurately", + "assess", + "severity", + "fast", + "sufficiently", + "remediate", + "For", + "factor", + "Risks", + "Related", + "Laws", + "Regulations", + "Policies", + "investments", + "safety", + "review", + "mitigate", + "making", + "combat", + "investigation", + "unnecessary", + "However", + "limitations", + "lack", + "visibility", + "encrypted", + "factors", + "learn", + "activities", + "manner", + "inconsistent", + "contracts", + "existence", + "accounts", + "ad", + "purchases", + "threaten", + "line", + "spamming", + "scraping", + "spreading", + "disinformation", + "unsuccessful", + "enforce", + "developments", + "trust", + "engagement", + "brands", + "operations", + "adversely", + "enhanced", + "oversight", + "depends", + "continued", + "unimpeded", + "Internet", + "restrict", + "block", + "charge", + "expenses", + "depend", + "capacity", + "Currently", + "companies", + "power", + "broadband", + "internet", + "marketplace", + "incumbent", + "telephone", + "cable", + "mobile", + "communications", + "owned", + "stated", + "disrupt", + "cost", + "restricting", + "prohibiting", + "support", + "facilitate", + "charging", + "fees", + "provide", + "providing", + "competitors", + "preferential", + "jurisdictions", + "adopted", + "forms", + "uncertainty", + "exists", + "US", + "initiated", + "restrictions", + "blockages", + "key", + "relationships", + "costs", + "affecting" + ], + "keywords": [ + "AI risks", + "data privacy", + "cybersecurity" + ], + "connect_to": [] + } + }, + { + "chunk_id": "9c60953b-01fc-5d74-b822-f32702073c53", + "type": "text", + "content": "We are subject to a variety of new, existing, and changing laws and regulations worldwide that could harm our business, and will likely be subject to an even broader scope of laws and regulations as we continue to expand our business.\nWe are subject to numerous US federal and state as well as foreign laws and regulations covering a wide variety of subjects, and our introduction of new businesses, products, services, and technologies will likely continue to subject us to additional laws and regulations. In recent years, governments around the world have proposed and adopted a large number of new laws and regulations relevant to the digital economy, particularly in the areas of data privacy and security, competition, AI, and online content. The costs of compliance with these measures are high and are likely to increase in the future, including as a result of differing, and sometimes conflicting, laws and regulations.\nNew or changing laws and regulations, or interpretations or applications of existing laws and regulations in a manner inconsistent with our interpretations of such laws and regulations or our practices, have resulted in, and may continue to result in, less useful products and services, altered business models and operations, limited ability to pursue certain business practices or offer certain products and services, substantial costs, and civil or criminal liability. Examples include laws and regulations regarding:\n\\- Competition and technology platforms' business practices: Laws and regulations focused on large technology platforms, including the Digital Markets Act in the European Union (EU) and the Act on Promotion of Competition for Specified Smartphone Software in Japan; regulations and legal settlements in the US, South Korea, and elsewhere that affect Google Play's billing policies, fees, and business model; as well as litigation and new and expected regulations in a range of jurisdictions.\n\\- AI: Laws and regulations focused on the development, use, and provision of AI technologies and other digital products and services, which could result in monetary penalties or other regulatory actions. For example, the EU AI Act came into force on August 1, 2024, and will generally become fully applicable after a two-year transitional period (although certain obligations have already taken effect). The EU AI Act introduces various requirements for AI systems and models placed on the market in the EU, including specific transparency, safety, and copyright requirements for general purpose AI systems and the models on which those systems are based. Various countries, including Brazil, India, Japan, South Korea, Singapore, and Vietnam, have also enacted or are considering enacting regulations focused on AI. In the US, an increasing amount of legislative and regulatory activity regarding AI is taking place at the state level. In 2025, state legislatures considered more than 1,000 AI-related bills, including on fundamental model research and development, synthetic media, algorithmic decision-making, and many others, and took a variety of approaches to AI regulation. For instance, in 2025, California and New York passed the Transparency in Frontier Artificial Intelligence Act and the Responsible AI Safety and Education Act, respectively, each of which imposes safety and reporting obligations on developers of frontier models. At the same time, the White House's Executive Order, Removing Barriers to American Leadership in Artificial Intelligence, prioritizes deregulation, while its AI Action Plan emphasizes accelerating American innovation leadership.\n- Data privacy, collection, processing, and portability: Laws and regulations further restricting the collection, processing, or sharing of user or advertising-related data, including privacy and data protection laws; laws affecting the processing of children's data (as discussed further below), data breach notification laws; laws limiting data transfers (including data localization laws); laws limiting use of data for AI training; and laws requiring data portability.\n- Copyright and other intellectual property: Copyright and related laws, including the EU Directive on Copyright in the Digital Single Market and European Economic Area transpositions, which have introduced new licensing regimes, increase liability with respect to content uploaded by users or linked to from our platforms, or create property rights in news publications that could require payments to news agencies and publishers, which may result in other regulatory actions. The scope of the text and data mining exception is being challenged before courts in the EU, which could harm some aspects of our business.\n- Content moderation: Various laws covering content moderation and removal, and related disclosure obligations, such as the EU's Digital Services Act, Florida's Senate Bill 7072 and Texas' House Bill 20, and laws and proposed legislation in Singapore, Australia, and the United Kingdom (UK) that impose penalties for failure to remove certain types of content or require disclosure of information about the operation of our services and algorithms, which may make it harder for services like Google Search and YouTube to detect and limit low-quality, deceptive, or harmful content, or, on the other hand, may impinge on the rights of free expression and access to content. Additionally, new regulations apply to online child safety, including access and content restrictions as well as other limitations for minors, which may also conflict with rights of free expression and access to information. These regulations could result in our having to modify our products and services and monitor minors' experiences on our products and services.\n- Consumer protection: Consumer protection laws, including the EU's New Deal for Consumers, which could result in monetary penalties and create a range of new compliance obligations.\nIn addition, the applicability and scope of these and other laws and regulations, as interpreted by courts, regulators, or administrative bodies, remain uncertain and could be interpreted in ways that harm our business. For example, we rely on statutory safe harbors, like those set forth in the Digital Millennium Copyright Act and Section 230 of the Communications Decency Act in the US and the Digital Services Act in Europe, to protect against liability for various linking, caching, ranking, recommending, and hosting activities. Legislation or court rulings affecting these safe harbors may harm us and may impose significant operational challenges. There are legislative proposals and pending litigation in the US, EU, and around the world that could diminish or eliminate safe harbor protection for websites and online platforms. Our development, use, and commercialization of AI products and services (including our implementation of AI in our offerings and internal systems) could subject us to regulatory action and legal liability, including under specific legislation regulating AI, as well as new applications of existing data protection, cybersecurity, privacy, intellectual property, and other laws.\nFurther, we are subject to evolving laws, regulations, policies, and international accords relating to matters beyond our core products and services, including environmental sustainability, climate change, human capital, and employment matters. In response, we have implemented robust programs and initiatives and adopted reporting frameworks and principles that may require considerable investments. For instance, AI's energy and water demands have made efforts to reduce our emissions more complex and challenging across every level. We cannot guarantee that our initiatives will be fully realized on the timelines we expect or at all, and projects that are completed as planned may not achieve the results we anticipate.\nWe are and may continue to be subject to claims, lawsuits, regulatory and government inquiries and investigations, enforcement actions, consent orders, and other forms of regulatory scrutiny and legal liability, including competition matters, that could harm our business, reputation, financial condition, and operating results.\nWe are subject to claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and orders involving competition, intellectual property, data privacy and security, tax and related compliance, labor and employment, commercial disputes, content generated by our users, goods and services offered by advertisers or publishers using our platforms, design of our products and services, personal injury and other tort and nuisance theories, consumer protection claims, including how we moderate content on our platforms, AI, and other matters.\nThe US Department of Justice (DOJ), various states, and other plaintiffs have filed, and may continue to file in the future, several antitrust lawsuits about aspects of our business, including our advertising technologies and practices, the operation and distribution of Google Search, and the operation and distribution of Android operating system and Play Store. If we are unsuccessful in these lawsuits, we could face significant expenses to implement the remedies, and such costs and alterations could harm our business, reputation, financial condition, and operating results.\nFor example, the DOJ and a number of state Attorneys General filed a lawsuit concerning our Search and Search advertising practices and our compliance with US antitrust laws. In August 2024, the US District Court for the District of Columbia ruled against Google, and in December 2025, entered a final judgment requiring remedies, which, among other things, imposes restrictions on how we distribute our services and requires us to share certain search data with and offer syndication services to certain competitors. In January 2026, we appealed the final judgment and moved to pause implementation of certain remedies. In February 2026, the DOJ and state Attorneys General also appealed.\nFurthermore, in December 2020, a number of state Attorneys General, led by the Texas Attorney General, filed a lawsuit in the US District Court for the Eastern District of Texas concerning our advertising technology and our compliance with US antitrust laws and other laws. In January 2023, the DOJ and a number of state Attorneys General sued in the Eastern District of Virginia alleging similar antitrust violations relating to our advertising technology. In April 2025, the presiding judge issued a mixed decision in the DOJ case against us, ruling that neither our advertiser tools nor the DoubleClick and AdMeld acquisitions were anticompetitive, but that our publisher tools unfairly excluded rivals. A separate proceeding to determine remedies, the range of which varies widely, took place in September 2025 with the parties presenting differing remedy proposals. The DOJ's remedy proposal includes structural remedies that could harm our business. Closing arguments were held in November 2025, and we are awaiting a final judgment. After that judgment, we plan to appeal the adverse portion of the April 2025 decision and potentially aspects of the remedies decision. A trial in the state Attorneys General case in the Eastern District of Texas will take place after a decision on remedies is issued in the DOJ advertising technology case, and could result in remedies that could harm our business, reputation, financial condition, and operating results.\nIn addition to these regulatory proceedings, private individual and collective actions that overlap with claims pursued by regulators are pending in the US and in several other jurisdictions. Adverse results in these or similar future lawsuits may include awards of monetary damages and remedies that could harm our business, reputation, financial condition, and operating results.\nOther regulatory agencies in the US and around the world, including competition enforcers, consumer protection agencies, and data protection authorities, have challenged and may continue to challenge our business practices and compliance with laws and regulations. We are cooperating with these investigations and defending litigation or appealing decisions where appropriate.\nWe are also subject to a variety of claims including product warranty, product liability, and consumer protection claims related to product defects, among other litigation, and we may also be subject to claims involving health and safety, hazardous materials usage, other environmental effects, AI training, development, and commercialization, or service disruptions or failures. Claims have been brought, and we expect will continue to be brought, against us for defamation, negligence, breaches of contract, patent, copyright, and trademark infringement, unfair competition, unlawful activity, torts, privacy rights violations, fraud, or other legal theories based on the nature and content of information available on or via our services, the design and effect of our products and services, or due to our involvement in hosting, transmitting, marketing, branding, or providing access to content created by third parties.\nVarious laws, regulations, investigations, enforcement lawsuits, and regulatory actions have involved in the past, and may in the future result in substantial fines and penalties, injunctive relief, ongoing monitoring and auditing obligations, changes to our products and services, alterations to our business models and operations, including divestiture, and collateral related civil litigation or other adverse consequences. Any of these legal proceedings could also result in legal costs, diversion of management resources, and negative publicity, all of which could harm our business, reputation, financial condition, and operating results.\nEstimating liabilities for our pending proceedings is a complex, fact-specific, and speculative process that requires significant judgment, and the amounts we are ultimately liable for may differ from our estimates. The resolution of one or more such proceedings has resulted in, and may in the future result in, additional substantial fines, penalties, injunctions, and other sanctions that could harm our business, reputation, financial condition, and operating results.\nFor additional information about the ongoing material legal proceedings to which we are subject, see Legal Proceedings in Item 3 of this Annual Report on Form 10-K.\nPrivacy, data protection, data usage, and portability regulations are complex and rapidly evolving areas. Any failure or alleged failure to comply with these laws could harm our business, reputation, financial condition, and operating results.\nAuthorities around the world have adopted and are considering a number of legislative and regulatory proposals concerning data protection, data usage and portability, and encryption of user data. Additionally, the increasing adoption of AI technologies, which rely on the collection of large amounts of data and use of such data for training purposes, has led data protection authorities around the world to consider and adopt new and evolving interpretations of data protection laws, imposing specific obligations with respect to the processing of personal data, including required notices, consents, and opt-outs. Further, the increased risk of inadvertent disclosure of confidential information or personal data in connection with the utilization of AI technologies may result in stronger regulatory scrutiny, leading to legal and regulatory investigations and enforcement actions that could harm our business, even if unfounded. Adverse legal rulings, legislation, or regulation have resulted in, and may continue to result in, fines and orders requiring that we change our practices, which have had and could continue to have an adverse effect on how we provide services, harming our business, reputation, financial condition, and operating results. These laws and regulations are evolving and subject to interpretation, and compliance obligations could cause us to incur substantial costs or harm the quality and operations of our products and services in ways that harm our business. Examples of these laws include:\n- The EU General Data Protection Regulation and the UK General Data Protection Regulations, which apply to all of our activities conducted from an establishment in the EU or the UK, respectively, or related to products and services that we offer to EU or the UK users or customers, respectively, or the monitoring of their behavior in the EU or the UK, respectively.\n- Various US federal, US state, and foreign privacy laws related to the processing and security of personal data, including (1) comprehensive privacy laws that provide data privacy rights (including, in California, a private right of action in the event of a data breach resulting from our failure to implement and maintain reasonable security procedures and practices) and impose significant obligations on controllers and processors of consumer data; (2) laws imposing obligations on businesses that collect or disclose biometric information (including, in Colorado, Illinois, Texas, and Washington); (3) laws governing the collection and processing of children and minor's data and how companies provide age-appropriate online experiences (including, in the US, the Children's Online Privacy Protection Act of 1998; the pending Children and Teens' Online Privacy Protection Act (COPPA 2.0); similar US state laws related to children's privacy, such as the New York Child Data Protection Act; and the UK's Age-Appropriate Design Code); and (4) laws regulating internet-connected devices (such as, in California, the Internet of Things Security Law).\n- The EU's Digital Markets Act, which requires in-scope companies to obtain user consent for combining data across certain products and require search engines to share anonymized data with rival companies, among other changes; and the EU Data Act, which introduces new data portability requirements with respect to connected products (i.e., 'internet of things' products) and related services, as well as interoperability obligations on data processing services.\nFurther, we are subject to evolving laws and regulations that dictate whether, how, and under what circumstances we can transfer, process, or receive personal data, as well as ongoing enforcement actions from supervisory authorities related to cross-border transfers of personal data. The validity of various data transfer mechanisms we currently rely upon remains subject to legal, regulatory, and political developments globally, which may require us to adapt our existing arrangements.\nWe face, and may continue to face, intellectual property infringement or misappropriation, violation of rights of publicity, and other claims that could be costly to defend, result in significant damage awards or other costs (including indemnification awards), and limit our ability to use certain technologies.\nWe, like other internet, technology, and media companies, are frequently subject to litigation based on allegations of infringement or other violations of intellectual property rights, including patent, copyright, trade secrets, and trademarks. Parties have also sought broad injunctive relief against us by filing claims in US and international courts and the US International Trade Commission (ITC) for exclusion and cease-and-desist orders. In addition, patent-holding companies may frequently seek to generate income from patents they have obtained by bringing claims against us. As we continue to expand our business, intellectual property claims against us have increased and may continue to increase as we develop and acquire new products, services, and technologies, including AI technologies.\nAdverse results in any of these lawsuits may include awards of monetary damages, costly royalty or licensing agreements (if licenses are available at all), or orders limiting our ability to sell our products and services in the US or elsewhere, including by preventing us from offering certain features, functionalities, products, or services in certain jurisdictions. They may also cause us to change our business practices in ways that could result in a loss of revenues for us and otherwise harm our business.\nMany of our agreements with our customers and partners, including certain suppliers, require us to defend against certain intellectual property infringement claims and in some cases indemnify them for certain intellectual property infringement claims against them, which could result in increased costs for defending such claims or significant damages if there were an adverse ruling in any such claims. Such customers and partners may also discontinue the use of our products, services, and technologies, as a result of injunctions or otherwise, which could result in loss of revenues and harm our business. Moreover, intellectual property indemnities provided to us by our suppliers, when obtainable, may not cover all damages and losses suffered by us and our customers arising from intellectual property infringement claims. Furthermore, in connection with our divestitures, we have agreed, and may in the future agree, to provide indemnification for certain potential liabilities, including those associated with intellectual property claims. Regardless of their merits, intellectual property claims are often time consuming and expensive to litigate or settle. To the extent such claims are successful, they could harm our business, including our product and service offerings, financial condition, and operating results.\nWe could be subject to changes in tax rates, the adoption of new US or international tax legislation, or exposure to additional tax liabilities.\nWe are subject to a variety of taxes and tax collection obligations in the US and numerous foreign jurisdictions. Our effective tax rates are affected by a variety of factors, including changes in the mix of earnings in jurisdictions with different statutory tax rates, net gains and losses on hedges and related transactions under our foreign exchange risk management program, changes in our stock price for shares issued as employee compensation, changes in the valuation of our deferred tax assets or liabilities, and the application of different provisions of tax laws or changes in tax laws, regulations, or accounting principles (including changes in the interpretation of existing laws). Further, if we are unable or fail to collect taxes on behalf of customers, employees, and partners as the withholding agent, we could become liable for taxes that are levied against third parties.\nWe are subject to regular review and audit by both domestic and foreign tax authorities. As a result, we have received, and may in the future receive, assessments in multiple jurisdictions, on various tax-related assertions, such as transfer-pricing adjustments or permanent-establishment claims. Any adverse outcome of such a review or audit could harm our financial condition and operating results, require adverse changes to our business practices, or subject us to additional litigation and regulatory inquiries. In addition, the determination of our worldwide provision for income taxes and other tax liabilities requires significant judgment and often involves uncertainty. Although we believe our estimates are reasonable, the ultimate tax outcome may differ from the amounts recorded in our financial statements and may affect our financial results in the period or periods for which such determination is made.\nFurther, due to shifting economic and political conditions, tax policies, laws, or rates in various jurisdictions may be subject to significant changes in ways that could harm our financial condition and operating results. For example, various jurisdictions around the world have enacted or are considering revenue-based taxes such as digital services taxes and other targeted taxes, which could lead to inconsistent and potentially overlapping international tax regimes. The US has imposed a 1% excise tax on the fair market value of shares repurchased, which could increase in the future. Over 140 countries are negotiating changes to international tax policies led by the Organization for Economic Cooperation and Development (OECD), including a 15% global minimum tax rate. In January 2026, the OECD announced a \"Side-by-Side Safe Harbor\" that exempts US operations of US-parented companies from global minimum tax rules. Adoption of minimum tax rules outside the US could increase our effective tax rate and cash tax payments.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS/Risks Related to Laws, Regulations, and Policies", + "metadata": { + "length": 24650, + "summary": "The passage details extensive legal and regulatory risks facing a large technology company, including antitrust lawsuits (e.g., DOJ cases on Search and ad tech), evolving AI regulations (e.g., EU AI Act), data privacy laws (e.g., GDPR, state laws), content moderation rules, intellectual property claims, and tax changes. It highlights potential costs, business model alterations, and liability from global compliance challenges.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "subject", + "variety", + "existing", + "changing", + "laws", + "regulations", + "worldwide", + "harm", + "business", + "broader", + "scope", + "continue", + "expand", + "numerous", + "US", + "federal", + "state", + "foreign", + "covering", + "wide", + "subjects", + "introduction", + "businesses", + "products", + "services", + "technologies", + "additional", + "In", + "recent", + "years", + "governments", + "world", + "proposed", + "adopted", + "large", + "number", + "relevant", + "digital", + "economy", + "areas", + "data", + "privacy", + "security", + "competition", + "AI", + "online", + "content", + "The", + "costs", + "compliance", + "measures", + "high", + "increase", + "future", + "including", + "result", + "differing", + "conflicting", + "New", + "interpretations", + "applications", + "manner", + "inconsistent", + "practices", + "resulted", + "altered", + "models", + "operations", + "limited", + "ability", + "pursue", + "offer", + "substantial", + "civil", + "criminal", + "liability", + "Examples", + "include", + "Competition", + "technology", + "platforms", + "Laws", + "focused", + "Digital", + "Markets", + "Act", + "European", + "Union", + "EU", + "Promotion", + "Specified", + "Smartphone", + "Software", + "Japan", + "legal", + "settlements", + "South", + "Korea", + "affect", + "Google", + "Play", + "billing", + "policies", + "fees", + "model", + "litigation", + "expected", + "range", + "jurisdictions", + "development", + "provision", + "monetary", + "penalties", + "regulatory", + "actions", + "For", + "force", + "August", + "2024", + "generally", + "fully", + "applicable", + "year", + "transitional", + "period", + "obligations", + "effect", + "introduces", + "requirements", + "systems", + "market", + "specific", + "transparency", + "safety", + "copyright", + "general", + "purpose", + "based", + "Various", + "countries", + "Brazil", + "India", + "Singapore", + "Vietnam", + "enacted", + "enacting", + "increasing", + "amount", + "legislative", + "activity", + "taking", + "place", + "level", + "2025", + "legislatures", + "considered", + "000", + "related", + "bills", + "fundamental", + "research", + "synthetic", + "media", + "algorithmic", + "decision", + "making", + "approaches", + "regulation", + "instance", + "California", + "York", + "passed", + "Transparency", + "Frontier", + "Artificial", + "Intelligence", + "Responsible", + "Safety", + "Education", + "imposes", + "reporting", + "developers", + "frontier", + "At", + "time", + "White", + "House", + "Executive", + "Order", + "Removing", + "Barriers", + "American", + "Leadership", + "prioritizes", + "deregulation", + "Action", + "Plan", + "emphasizes", + "accelerating", + "innovation", + "leadership", + "Data", + "collection", + "processing", + "portability", + "restricting", + "sharing", + "user", + "advertising", + "protection", + "affecting", + "children", + "discussed", + "breach", + "notification", + "limiting", + "transfers", + "localization", + "training", + "requiring", + "Copyright", + "intellectual", + "property", + "Directive", + "Single", + "Market", + "Economic", + "Area", + "transpositions", + "introduced", + "licensing", + "regimes", + "respect", + "uploaded", + "users", + "linked", + "create", + "rights", + "news", + "publications", + "require", + "payments", + "agencies", + "publishers", + "text", + "mining", + "exception", + "challenged", + "courts", + "aspects", + "Content", + "moderation", + "removal", + "disclosure", + "Services", + "Florida", + "Senate", + "Bill", + "7072", + "Texas", + "20", + "legislation", + "Australia", + "United", + "Kingdom", + "UK", + "impose", + "failure", + "remove", + "types", + "information", + "operation", + "algorithms", + "make", + "harder", + "Search", + "YouTube", + "detect", + "limit", + "low", + "quality", + "deceptive", + "harmful", + "hand", + "impinge", + "free", + "expression", + "access", + "Additionally", + "apply", + "child", + "restrictions", + "limitations", + "minors", + "conflict", + "These", + "modify", + "monitor", + "experiences", + "Consumer", + "Deal", + "Consumers", + "addition", + "applicability", + "interpreted", + "regulators", + "administrative", + "bodies", + "remain", + "uncertain", + "ways", + "rely", + "statutory", + "safe", + "harbors", + "set", + "Millennium", + "Section", + "230", + "Communications", + "Decency", + "Europe", + "protect", + "linking", + "caching", + "ranking", + "recommending", + "hosting", + "activities", + "Legislation", + "court", + "rulings", + "significant", + "operational", + "challenges", + "There", + "proposals", + "pending", + "diminish", + "eliminate", + "harbor", + "websites", + "Our", + "commercialization", + "implementation", + "offerings", + "internal", + "action", + "regulating", + "cybersecurity", + "Further", + "evolving", + "international", + "accords", + "relating", + "matters", + "core", + "environmental", + "sustainability", + "climate", + "change", + "human", + "capital", + "employment", + "response", + "implemented", + "robust", + "programs", + "initiatives", + "frameworks", + "principles", + "considerable", + "investments", + "energy", + "water", + "demands", + "made", + "efforts", + "reduce", + "emissions", + "complex", + "challenging", + "guarantee", + "realized", + "timelines", + "expect", + "projects", + "completed", + "planned", + "achieve", + "results", + "anticipate", + "claims", + "lawsuits", + "government", + "inquiries", + "investigations", + "enforcement", + "consent", + "orders", + "forms", + "scrutiny", + "reputation", + "financial", + "condition", + "operating", + "proceedings", + "involving", + "tax", + "labor", + "commercial", + "disputes", + "generated", + "goods", + "offered", + "advertisers", + "design", + "personal", + "injury", + "tort", + "nuisance", + "theories", + "consumer", + "moderate", + "Department", + "Justice", + "DOJ", + "states", + "plaintiffs", + "filed", + "file", + "antitrust", + "distribution", + "Android", + "system", + "Store", + "If", + "unsuccessful", + "face", + "expenses", + "implement", + "remedies", + "alterations", + "Attorneys", + "General", + "lawsuit", + "District", + "Court", + "Columbia", + "ruled", + "December", + "entered", + "final", + "judgment", + "things", + "distribute", + "requires", + "share", + "search", + "syndication", + "competitors", + "January", + "2026", + "appealed", + "moved", + "pause", + "February", + "Furthermore", + "2020", + "led", + "Attorney", + "Eastern", + "2023", + "sued", + "Virginia", + "alleging", + "similar", + "violations", + "April", + "presiding", + "judge", + "issued", + "mixed", + "case", + "ruling", + "advertiser", + "tools", + "DoubleClick", + "AdMeld", + "acquisitions", + "anticompetitive", + "publisher", + "unfairly", + "excluded", + "rivals", + "separate", + "proceeding", + "determine", + "varies", + "widely", + "September", + "parties", + "presenting", + "remedy", + "proposal", + "includes", + "structural", + "Closing", + "arguments", + "held", + "November", + "awaiting", + "After", + "plan", + "appeal", + "adverse", + "portion", + "potentially", + "trial", + "private", + "individual", + "collective", + "overlap", + "pursued", + "Adverse", + "awards", + "damages", + "Other", + "enforcers", + "authorities", + "challenge", + "cooperating", + "defending", + "appealing", + "decisions", + "product", + "warranty", + "defects", + "health", + "hazardous", + "materials", + "usage", + "effects", + "service", + "disruptions", + "failures", + "Claims", + "brought", + "defamation", + "negligence", + "breaches", + "contract", + "patent", + "trademark", + "infringement", + "unfair", + "unlawful", + "torts", + "fraud", + "nature", + "due", + "involvement", + "transmitting", + "marketing", + "branding", + "providing", + "created", + "involved", + "past", + "fines", + "injunctive", + "relief", + "ongoing", + "monitoring", + "auditing", + "divestiture", + "collateral", + "consequences", + "Any", + "diversion", + "management", + "resources", + "negative", + "publicity", + "Estimating", + "liabilities", + "fact", + "speculative", + "process", + "amounts", + "ultimately", + "liable", + "differ", + "estimates", + "resolution", + "injunctions", + "sanctions", + "material", + "Legal", + "Proceedings", + "Item", + "Annual", + "Report", + "Form", + "10", + "Privacy", + "rapidly", + "alleged", + "comply", + "Authorities", + "encryption", + "adoption", + "purposes", + "adopt", + "imposing", + "required", + "notices", + "consents", + "opt", + "outs", + "increased", + "risk", + "inadvertent", + "confidential", + "connection", + "utilization", + "stronger", + "leading", + "unfounded", + "provide", + "harming", + "interpretation", + "incur", + "Protection", + "Regulation", + "Regulations", + "conducted", + "establishment", + "customers", + "behavior", + "comprehensive", + "event", + "resulting", + "maintain", + "reasonable", + "procedures", + "controllers", + "processors", + "collect", + "disclose", + "biometric", + "Colorado", + "Illinois", + "Washington", + "governing", + "minor", + "companies", + "age", + "Children", + "Online", + "1998", + "Teens", + "COPPA", + "2.0", + "Child", + "Age", + "Appropriate", + "Design", + "Code", + "internet", + "connected", + "devices", + "Internet", + "Things", + "Security", + "Law", + "obtain", + "combining", + "engines", + "anonymized", + "rival", + "interoperability", + "dictate", + "circumstances", + "transfer", + "receive", + "supervisory", + "cross", + "border", + "validity", + "mechanisms", + "remains", + "political", + "developments", + "globally", + "adapt", + "arrangements", + "misappropriation", + "violation", + "costly", + "defend", + "damage", + "indemnification", + "frequently", + "allegations", + "trade", + "secrets", + "trademarks", + "Parties", + "sought", + "broad", + "filing", + "International", + "Trade", + "Commission", + "ITC", + "exclusion", + "cease", + "desist", + "holding", + "seek", + "generate", + "income", + "patents", + "obtained", + "bringing", + "As", + "develop", + "acquire", + "royalty", + "agreements", + "licenses", + "sell", + "preventing", + "offering", + "features", + "functionalities", + "They", + "loss", + "revenues", + "Many", + "partners", + "suppliers", + "cases", + "indemnify", + "Such", + "discontinue", + "Moreover", + "indemnities", + "provided", + "obtainable", + "cover", + "losses", + "suffered", + "arising", + "divestitures", + "agreed", + "agree", + "potential", + "Regardless", + "merits", + "consuming", + "expensive", + "litigate", + "settle", + "To", + "extent", + "successful", + "rates", + "exposure", + "taxes", + "effective", + "affected", + "factors", + "mix", + "earnings", + "net", + "gains", + "hedges", + "transactions", + "exchange", + "program", + "stock", + "price", + "shares", + "employee", + "compensation", + "valuation", + "deferred", + "assets", + "application", + "provisions", + "accounting", + "unable", + "fail", + "behalf", + "employees", + "withholding", + "agent", + "levied", + "regular", + "review", + "audit", + "domestic", + "received", + "assessments", + "multiple", + "assertions", + "pricing", + "adjustments", + "permanent", + "outcome", + "determination", + "involves", + "uncertainty", + "Although", + "ultimate", + "recorded", + "statements", + "periods", + "shifting", + "economic", + "conditions", + "revenue", + "targeted", + "lead", + "overlapping", + "imposed", + "1%", + "excise", + "fair", + "repurchased", + "Over", + "140", + "negotiating", + "Organization", + "Cooperation", + "Development", + "OECD", + "15%", + "global", + "minimum", + "rate", + "announced", + "Side", + "Safe", + "Harbor", + "exempts", + "parented", + "rules", + "Adoption", + "cash" + ], + "keywords": [ + "Regulatory", + "Compliance", + "Litigation" + ], + "connect_to": [] + } + }, + { + "chunk_id": "3e4be579-9be2-5452-8fb7-e0c7da6c618b", + "type": "text", + "content": "We cannot guarantee that any share repurchase program or dividend program will be continuously active or fully consummated or will enhance long-term stockholder value, and share repurchases or dividends could increase the volatility of our stock prices and could diminish our cash reserves.\nWe engage in share repurchases of our Class A and Class C stock from time to time in accordance with authorizations from the Board of Directors of Alphabet. Our repurchase program does not have an expiration date and does not obligate Alphabet to repurchase any specific dollar amount or to acquire any specific number of shares. Our cash dividend program pays regular cash dividends to our Class A, Class B, and Class C stockholders. Any and all future cash dividends are subject to declaration by our Board of Directors in its sole discretion, and in accordance with the requirements of any applicable laws, rules, and regulations, including the Delaware General Corporation Law. Our cash dividend program does not require, and our Board of Directors may decide not to declare, a cash dividend each quarter, and does not obligate our Board of Directors to declare a dividend at any specific dollar amount per share. Any such decision by our Board of Directors may depend on a variety of factors that it may deem relevant, including but not limited to our earnings, liquidity, financial condition, other capital deployment opportunities, level of indebtedness, and general market conditions. Our share repurchases and dividends could affect our share trading prices, increase their volatility, reduce our cash reserves and may be suspended or terminated at any time, which may result in a decrease in the trading prices of our stock.\nThe concentration of our stock ownership limits our stockholders' ability to influence corporate matters.\nOur Class B stock has 10 votes per share, our Class A stock has one vote per share, and our Class C stock has no voting rights. As of December 31, 2025, Larry Page and Sergey Brin beneficially owned approximately 89.3% of our outstanding Class B stock, which represented approximately 52.7% of the voting power of our outstanding common stock. Through their stock ownership, Larry and Sergey have significant influence over all matters requiring stockholder approval, including the election of directors and significant corporate transactions, such as a merger or other sale of our company or our assets, for the foreseeable future. In addition, because our Class C stock carries no voting rights (except as required by applicable law), the issuance of the Class C stock, including in future stock-based acquisition transactions and to fund employee equity incentive programs, could continue Larry and Sergey's current relative voting power and their ability to elect all of our directors and to determine the outcome of most matters submitted to a vote of our stockholders. The share repurchases made pursuant to our repurchase program may also affect Larry and Sergey's relative voting power. This concentrated control limits or severely restricts other stockholders' ability to influence corporate matters and we may take actions that some of our stockholders do not view as beneficial, which could reduce the market price of our Class A stock and our Class C stock.\nProvisions in our charter documents and under Delaware law could discourage a takeover that stockholders may consider favorable.\nProvisions in Alphabet's certificate of incorporation and bylaws may have the effect of delaying or preventing a change of control or changes in our management. These provisions include the following:\n- Our Board of Directors has the right to elect directors to fill a vacancy created by the expansion of the Board of Directors or the resignation, death, or removal of a director.\n- Our stockholders may not act by written consent, which makes it difficult to take certain actions without holding a stockholders' meeting.\n- Our certificate of incorporation prohibits cumulative voting in the election of directors. This limits the ability of minority stockholders to elect director candidates.\n- Stockholders must provide advance notice to nominate individuals for election to the Board of Directors or to propose matters that can be acted upon at a stockholders' meeting. These provisions may discourage or deter a potential acquirer from conducting a solicitation of proxies to elect the acquirer's own slate of directors or otherwise attempting to obtain control of our company.\n- Our Board of Directors may issue, without stockholder approval, shares of undesignated preferred stock, which makes it possible for our Board of Directors to issue preferred stock with voting or other rights or preferences that could impede the success of any attempt to acquire us.\nAs a Delaware corporation, we are also subject to certain Delaware anti-takeover provisions. Under Delaware law, a corporation may not engage in a business combination with any holder of 15% or more of its outstanding voting stock unless the holder has held the stock for three years or, among other things, the Board of Directors has approved the transaction. Our Board of Directors could rely on Delaware law to prevent or delay an acquisition of us.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS/Risks Related to Ownership of Our Stock", + "metadata": { + "length": 5250, + "summary": "The passage discusses Alphabet's share repurchase and dividend programs, noting they are not guaranteed to enhance long-term value and may increase stock price volatility. It highlights the concentrated voting power of Larry Page and Sergey Brin through Class B stock, which limits other stockholders' influence. Additionally, it outlines charter provisions and Delaware law that could discourage takeovers, such as the ability to issue preferred stock without stockholder approval and restrictions on business combinations with large stockholders.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "guarantee", + "share", + "repurchase", + "program", + "dividend", + "continuously", + "active", + "fully", + "consummated", + "enhance", + "long", + "term", + "stockholder", + "repurchases", + "dividends", + "increase", + "volatility", + "stock", + "prices", + "diminish", + "cash", + "reserves", + "engage", + "Class", + "time", + "accordance", + "authorizations", + "Board", + "Directors", + "Alphabet", + "Our", + "expiration", + "date", + "obligate", + "specific", + "dollar", + "amount", + "acquire", + "number", + "shares", + "pays", + "regular", + "stockholders", + "Any", + "future", + "subject", + "declaration", + "sole", + "discretion", + "requirements", + "applicable", + "laws", + "rules", + "regulations", + "including", + "Delaware", + "General", + "Corporation", + "Law", + "require", + "decide", + "declare", + "quarter", + "decision", + "depend", + "variety", + "factors", + "deem", + "relevant", + "limited", + "earnings", + "liquidity", + "financial", + "condition", + "capital", + "deployment", + "opportunities", + "level", + "indebtedness", + "general", + "market", + "conditions", + "affect", + "trading", + "reduce", + "suspended", + "terminated", + "result", + "decrease", + "The", + "concentration", + "ownership", + "limits", + "ability", + "influence", + "corporate", + "matters", + "10", + "votes", + "vote", + "voting", + "rights", + "As", + "December", + "31", + "2025", + "Larry", + "Page", + "Sergey", + "Brin", + "beneficially", + "owned", + "approximately", + "89.3%", + "outstanding", + "represented", + "52.7%", + "power", + "common", + "Through", + "significant", + "requiring", + "approval", + "election", + "directors", + "transactions", + "merger", + "sale", + "company", + "assets", + "foreseeable", + "In", + "addition", + "carries", + "required", + "law", + "issuance", + "based", + "acquisition", + "fund", + "employee", + "equity", + "incentive", + "programs", + "continue", + "current", + "relative", + "elect", + "determine", + "outcome", + "submitted", + "made", + "pursuant", + "This", + "concentrated", + "control", + "severely", + "restricts", + "actions", + "view", + "beneficial", + "price", + "Provisions", + "charter", + "documents", + "discourage", + "takeover", + "favorable", + "certificate", + "incorporation", + "bylaws", + "effect", + "delaying", + "preventing", + "change", + "management", + "These", + "provisions", + "include", + "fill", + "vacancy", + "created", + "expansion", + "resignation", + "death", + "removal", + "director", + "act", + "written", + "consent", + "makes", + "difficult", + "holding", + "meeting", + "prohibits", + "cumulative", + "minority", + "candidates", + "Stockholders", + "provide", + "advance", + "notice", + "nominate", + "individuals", + "propose", + "acted", + "deter", + "potential", + "acquirer", + "conducting", + "solicitation", + "proxies", + "slate", + "attempting", + "obtain", + "issue", + "undesignated", + "preferred", + "preferences", + "impede", + "success", + "attempt", + "corporation", + "anti", + "Under", + "business", + "combination", + "holder", + "15%", + "held", + "years", + "things", + "approved", + "transaction", + "rely", + "prevent", + "delay" + ], + "keywords": [ + "share repurchase", + "voting control", + "anti-takeover" + ], + "connect_to": [] + } + }, + { + "chunk_id": "ed5ae0be-e762-53a1-a54c-85053157e430", + "type": "text", + "content": "Our operating results may fluctuate, which makes our results difficult to predict and could cause our results to fall short of expectations.\nOur operating results have fluctuated, and may in the future fluctuate, as a result of a number of factors, many outside of our control, including the cyclical nature and seasonality in our business and geopolitical events. As a result, comparing our operating results (including our expenses as a percentage of our revenues) on a period-to-period basis may not be meaningful, and our past results should not be relied on as an indication of our future performance. Consequently, our operating results in future quarters may fall below expectations.\nAcquisitions, joint ventures, investments, and divestitures could result in operating difficulties, dilution, and other consequences that could harm our business, financial condition, and operating results.\nAcquisitions, joint ventures, investments, and divestitures are important elements of our overall corporate strategy and use of capital, and these transactions could be material to our financial condition and operating results. We expect to continue to evaluate and enter into discussions regarding a wide array of such potential strategic arrangements, which could create unforeseen operating difficulties and expenditures. Some of the areas where we face risks include:\n- diversion of management time and focus from operating our business to challenges related to acquisitions and other strategic arrangements;\n- failure to obtain required approvals on a timely basis, if at all, from governmental authorities, or conditions placed upon approval, either of which could, among other things, delay or prevent us from completing a transaction, result in the payment of fees or penalties to a counterparty, or otherwise restrict our ability to\nrealize the expected financial or strategic goals of a transaction; or investigations or litigation by governmental authorities related to our acquisitions, investments, and other strategic arrangements;\n- failure to successfully integrate the acquired operations, technologies, services, and personnel (including cultural integration and retention of employees) and further develop the acquired business or technology;\n- implementation of controls (or remediation of deficiencies), procedures, and policies at the acquired company;\n- integration of the acquired company's accounting and other administrative systems, and the coordination of product, engineering, and sales and marketing functions;\n• transition of operations, users, and customers onto our existing platforms;\n- in the case of foreign acquisitions, the need to integrate operations across different cultures and languages and to address the particular economic, currency, political, and regulatory risks associated with specific countries;\n- failure to accomplish commercial, strategic, or financial objectives with respect to investments, joint ventures, and other strategic arrangements;\n- failure to realize the value of investments and joint ventures due to a lack of liquidity or an inability to identify buyers or negotiate favorable terms for intended divestitures;\n- liability for activities of the acquired company before the acquisition, including intellectual property infringement claims, data privacy and security issues, violations of laws, commercial disputes, tax liabilities, warranty claims, product liabilities, and other known and unknown liabilities; and\n- litigation or other claims in connection with the acquired company, including claims from terminated employees, customers, former stockholders, or other third parties.\nOur failure to address these risks or other problems encountered in connection with our past or future acquisitions and other strategic arrangements could cause us to fail to realize their anticipated benefits, incur unanticipated liabilities, and harm our business generally.\nOur acquisitions and other strategic arrangements could also result in dilutive issuances of our equity securities, the incurrence of debt, contingent liabilities, or amortization expenses, or impairment of goodwill or purchased long-lived assets, and restructuring charges, any of which could harm our financial condition and operating results. Also, the anticipated benefits or value of our acquisitions, investments, and other strategic arrangements may not materialize. In connection with our divestitures and certain strategic arrangements, we have agreed, and may in the future agree, to provide indemnification for certain potential liabilities, which could harm our financial condition and operating results.\nWe rely on highly skilled personnel and, if we are unable to retain or motivate key personnel, hire qualified personnel, or maintain and continue to adapt our corporate culture, we may not be able to grow or operate effectively.\nOur performance and future success depends in large part upon the continued service of key technical leads as well as members of our senior management team. For instance, Sundar Pichai is critical to the overall management of Alphabet and its subsidiaries and plays an important role in the development of our technology, maintaining our culture, and setting our strategic direction.\nOur ability to compete effectively and our future success depend on our continuing to identify, hire, develop, motivate, and retain highly skilled personnel for all areas of our organization. Competition in our industry for qualified employees, particularly AI talent, is intense, and certain of our competitors have directly targeted, and may continue to target, our employees. In addition, our compensation arrangements, such as our equity award programs, may not always be successful in attracting new employees and retaining and motivating our existing employees. Immigration policy and regulatory changes, and uncertainty regarding such policies and regulations, may also affect our ability to hire, mobilize, or retain some of our global talent. All of our executive officers and key employees are at-will employees, and we do not maintain any key-person life insurance policies.\nWe believe that our corporate culture fosters innovation, creativity, and teamwork. As our organization grows and evolves, we may need to adapt our corporate culture and work environments to ever-changing circumstances, and these changes could affect our ability to compete effectively or have an adverse effect on our corporate culture.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS/General Risks", + "metadata": { + "length": 6483, + "summary": "The passage discusses factors causing fluctuations in operating results, including cyclical nature, seasonality, and geopolitical events. It highlights risks from acquisitions, joint ventures, investments, and divestitures, such as management distraction, integration failures, regulatory issues, and potential dilution. It also emphasizes reliance on highly skilled personnel, competition for talent, and the need to maintain corporate culture for effective growth and competition.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "operating", + "results", + "fluctuate", + "makes", + "difficult", + "predict", + "fall", + "short", + "expectations", + "fluctuated", + "future", + "result", + "number", + "factors", + "control", + "including", + "cyclical", + "nature", + "seasonality", + "business", + "geopolitical", + "events", + "As", + "comparing", + "expenses", + "percentage", + "revenues", + "period", + "basis", + "meaningful", + "past", + "relied", + "indication", + "performance", + "Consequently", + "quarters", + "Acquisitions", + "joint", + "ventures", + "investments", + "divestitures", + "difficulties", + "dilution", + "consequences", + "harm", + "financial", + "condition", + "important", + "elements", + "corporate", + "strategy", + "capital", + "transactions", + "material", + "We", + "expect", + "continue", + "evaluate", + "enter", + "discussions", + "wide", + "array", + "potential", + "strategic", + "arrangements", + "create", + "unforeseen", + "expenditures", + "Some", + "areas", + "face", + "risks", + "include", + "diversion", + "management", + "time", + "focus", + "challenges", + "related", + "acquisitions", + "failure", + "obtain", + "required", + "approvals", + "timely", + "governmental", + "authorities", + "conditions", + "approval", + "things", + "delay", + "prevent", + "completing", + "transaction", + "payment", + "fees", + "penalties", + "counterparty", + "restrict", + "ability", + "realize", + "expected", + "goals", + "investigations", + "litigation", + "successfully", + "integrate", + "acquired", + "operations", + "technologies", + "services", + "personnel", + "cultural", + "integration", + "retention", + "employees", + "develop", + "technology", + "implementation", + "controls", + "remediation", + "deficiencies", + "procedures", + "policies", + "company", + "accounting", + "administrative", + "systems", + "coordination", + "product", + "engineering", + "sales", + "marketing", + "functions", + "transition", + "users", + "customers", + "existing", + "platforms", + "case", + "foreign", + "cultures", + "languages", + "address", + "economic", + "currency", + "political", + "regulatory", + "specific", + "countries", + "accomplish", + "commercial", + "objectives", + "respect", + "due", + "lack", + "liquidity", + "inability", + "identify", + "buyers", + "negotiate", + "favorable", + "terms", + "intended", + "liability", + "activities", + "acquisition", + "intellectual", + "property", + "infringement", + "claims", + "data", + "privacy", + "security", + "issues", + "violations", + "laws", + "disputes", + "tax", + "liabilities", + "warranty", + "unknown", + "connection", + "terminated", + "stockholders", + "parties", + "problems", + "encountered", + "fail", + "anticipated", + "benefits", + "incur", + "unanticipated", + "generally", + "dilutive", + "issuances", + "equity", + "securities", + "incurrence", + "debt", + "contingent", + "amortization", + "impairment", + "goodwill", + "purchased", + "long", + "lived", + "assets", + "restructuring", + "charges", + "Also", + "materialize", + "In", + "agreed", + "agree", + "provide", + "indemnification", + "rely", + "highly", + "skilled", + "unable", + "retain", + "motivate", + "key", + "hire", + "qualified", + "maintain", + "adapt", + "culture", + "grow", + "operate", + "effectively", + "success", + "depends", + "large", + "part", + "continued", + "service", + "technical", + "leads", + "members", + "senior", + "team", + "For", + "instance", + "Sundar", + "Pichai", + "critical", + "Alphabet", + "subsidiaries", + "plays", + "role", + "development", + "maintaining", + "setting", + "direction", + "compete", + "depend", + "continuing", + "organization", + "Competition", + "industry", + "AI", + "talent", + "intense", + "competitors", + "directly", + "targeted", + "target", + "addition", + "compensation", + "award", + "programs", + "successful", + "attracting", + "retaining", + "motivating", + "Immigration", + "policy", + "uncertainty", + "regulations", + "affect", + "mobilize", + "global", + "All", + "executive", + "officers", + "person", + "life", + "insurance", + "fosters", + "innovation", + "creativity", + "teamwork", + "grows", + "evolves", + "work", + "environments", + "changing", + "circumstances", + "adverse", + "effect" + ], + "keywords": [ + "Operating results", + "Acquisitions", + "Personnel" + ], + "connect_to": [] + } + }, + { + "chunk_id": "cc4c3124-4c66-5fd7-863e-f016f6ab1f29", + "type": "text", + "content": "Not applicable.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1B. UNRESOLVED STAFF COMMENTS", + "metadata": { + "length": 15, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Not", + "applicable" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8b50df7b-f7bc-5dac-a7a6-39a8a665d86f", + "type": "text", + "content": "We maintain a comprehensive process for identifying, assessing, and managing material risks from cybersecurity threats as part of our broader risk management system and processes. For example, some risks include our software supply chain and other third-party dependencies, vulnerabilities in our products and services, theft of our intellectual property, and attempts to compromise our infrastructure. We obtain input, as appropriate, for our cybersecurity risk management program on the security industry and threat trends from multiple external experts and internal threat intelligence teams. Teams of dedicated privacy, safety, and security professionals oversee cybersecurity risk management and mitigation, incident prevention, detection, and remediation. These teams comprise professionals with deep cybersecurity expertise across multiple industries and are led by our Vice President of Privacy, Safety, and Security, who has more than 20 years of experience, including roles in technology infrastructure for two other large public companies. Our executive leadership team, along with input from the above teams, are responsible for our overall enterprise risk management system and processes and regularly consider cybersecurity risks in the context of other material risks to the company.\nAs part of our cybersecurity risk management process, our incident management teams track and log privacy and security incidents across Alphabet, our vendors, and other third-party service providers to remediate and resolve any such incidents. Significant incidents are reviewed regularly by a cross-functional working group to determine whether further escalation is appropriate. Any incident assessed as potentially being or potentially becoming material is promptly escalated for further assessment, and then reported to designated members of our senior management. We consult with outside counsel as appropriate, including on materiality analysis and disclosure matters, and our senior management makes the final materiality determinations and disclosure and other compliance decisions. Our management apprises Alphabet's independent public accounting firm of relevant matters and developments.\nThe Risk and Compliance Committee has oversight responsibility for risks and incidents relating to cybersecurity threats, including compliance with disclosure requirements, cooperation with law enforcement, and related effects on financial and other risks, and it reports any findings and recommendations, as appropriate, to the full Board for consideration. Senior management regularly discusses cybersecurity risks and trends and, should they arise, any material incidents with the Risk and Compliance Committee. Internal Audit maintains a dedicated cybersecurity auditing team that independently tests our cybersecurity controls.\nOur business strategy, results of operations and financial condition have not been materially affected by risks from cybersecurity threats, including as a result of previously identified cybersecurity incidents, but we cannot provide assurance that they will not be materially affected in the future by such risks or any future material incidents. For more information on our cybersecurity related risks, see Item 1A Risk Factors of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1C. CYBERSECURITY", + "metadata": { + "length": 3296, + "summary": "The passage describes a comprehensive process for identifying, assessing, and managing material cybersecurity risks, including software supply chain vulnerabilities, IP theft, and infrastructure compromise. It details oversight by dedicated teams led by a Vice President with over 20 years of experience, incident tracking and escalation procedures, and board-level oversight via the Risk and Compliance Committee. The company states that cybersecurity threats have not materially affected its business strategy or financial condition to date, but acknowledges future risks cannot be assured.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "maintain", + "comprehensive", + "process", + "identifying", + "assessing", + "managing", + "material", + "risks", + "cybersecurity", + "threats", + "part", + "broader", + "risk", + "management", + "system", + "processes", + "For", + "include", + "software", + "supply", + "chain", + "party", + "dependencies", + "vulnerabilities", + "products", + "services", + "theft", + "intellectual", + "property", + "attempts", + "compromise", + "infrastructure", + "obtain", + "input", + "program", + "security", + "industry", + "threat", + "trends", + "multiple", + "external", + "experts", + "internal", + "intelligence", + "teams", + "Teams", + "dedicated", + "privacy", + "safety", + "professionals", + "oversee", + "mitigation", + "incident", + "prevention", + "detection", + "remediation", + "These", + "comprise", + "deep", + "expertise", + "industries", + "led", + "Vice", + "President", + "Privacy", + "Safety", + "Security", + "20", + "years", + "experience", + "including", + "roles", + "technology", + "large", + "public", + "companies", + "Our", + "executive", + "leadership", + "team", + "responsible", + "enterprise", + "regularly", + "context", + "company", + "As", + "track", + "log", + "incidents", + "Alphabet", + "vendors", + "service", + "providers", + "remediate", + "resolve", + "Significant", + "reviewed", + "cross", + "functional", + "working", + "group", + "determine", + "escalation", + "Any", + "assessed", + "potentially", + "promptly", + "escalated", + "assessment", + "reported", + "designated", + "members", + "senior", + "consult", + "counsel", + "materiality", + "analysis", + "disclosure", + "matters", + "makes", + "final", + "determinations", + "compliance", + "decisions", + "apprises", + "independent", + "accounting", + "firm", + "relevant", + "developments", + "The", + "Risk", + "Compliance", + "Committee", + "oversight", + "responsibility", + "relating", + "requirements", + "cooperation", + "law", + "enforcement", + "related", + "effects", + "financial", + "reports", + "findings", + "recommendations", + "full", + "Board", + "consideration", + "Senior", + "discusses", + "arise", + "Internal", + "Audit", + "maintains", + "auditing", + "independently", + "tests", + "controls", + "business", + "strategy", + "results", + "operations", + "condition", + "materially", + "affected", + "result", + "previously", + "identified", + "provide", + "assurance", + "future", + "information", + "Item", + "1A", + "Factors", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [ + "cybersecurity", + "risk management", + "incident response" + ], + "connect_to": [] + } + }, + { + "chunk_id": "c6e34405-c56e-5ea7-98c8-c60e37930a8a", + "type": "text", + "content": "Our headquarters are located in Mountain View, California. We own and lease office facilities and data centers around the world, primarily in Asia, Europe, and North America. We believe our existing facilities are in good condition and suitable for the conduct of our business.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 2. PROPERTIES", + "metadata": { + "length": 277, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "headquarters", + "located", + "Mountain", + "View", + "California", + "We", + "lease", + "office", + "facilities", + "data", + "centers", + "world", + "primarily", + "Asia", + "Europe", + "North", + "America", + "existing", + "good", + "condition", + "suitable", + "conduct", + "business" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0fc02f66-50e6-5618-bcba-e45190c6c37c", + "type": "text", + "content": "For a description of our material pending legal proceedings, see Legal Matters in Note 10 of the Notes to Consolidated Financial Statements included in Part II, Item 8 of this Annual Report on Form 10-K, which is incorporated herein by reference.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 3. LEGAL PROCEEDINGS", + "metadata": { + "length": 246, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "For", + "description", + "material", + "pending", + "legal", + "proceedings", + "Legal", + "Matters", + "Note", + "10", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Part", + "II", + "Item", + "Annual", + "Report", + "Form", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cc4c3124-4c66-5fd7-863e-f016f6ab1f29", + "type": "text", + "content": "Not applicable.", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 4. MINE SAFETY DISCLOSURES", + "metadata": { + "length": 15, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Not", + "applicable" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "12e026ef-e52d-5a17-b5c3-c9e5f5bb6bdf", + "type": "text", + "content": "ITEM 5. MARKET FOR REGISTRANT'S COMMON EQUITY, RELATED STOCKHOLDER MATTERS, AND ISSUER PURCHASES OF EQUITY SECURITIES\nAs of October 2, 2015, Alphabet Inc. became the successor issuer of Google Inc. pursuant to Rule 12g-3(a) under the Exchange Act. Our Class A stock has been listed on the Nasdaq Global Select Market under the symbol \"GOOG\" since August 19, 2004, and under the symbol \"GOOGL\" since April 3, 2014. Prior to August 19, 2004, there was no public market for our stock. Our Class B stock is neither listed nor traded. Our Class C stock has been listed on the Nasdaq Global Select Market under the symbol \"GOOG\" since April 3, 2014.", + "path": "GOOG-10-K-2025.pdf/PART II", + "metadata": { + "length": 643, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "ITEM", + "MARKET", + "FOR", + "REGISTRANT", + "COMMON", + "EQUITY", + "RELATED", + "STOCKHOLDER", + "MATTERS", + "AND", + "ISSUER", + "PURCHASES", + "OF", + "SECURITIES", + "As", + "October", + "2015", + "Alphabet", + "Inc", + "successor", + "issuer", + "Google", + "pursuant", + "Rule", + "12g", + "Exchange", + "Act", + "Our", + "Class", + "stock", + "listed", + "Nasdaq", + "Global", + "Select", + "Market", + "symbol", + "GOOG", + "August", + "19", + "2004", + "GOOGL", + "April", + "2014", + "Prior", + "public", + "market", + "traded" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a51305e8-f591-5d8e-8245-0a2a293b451c", + "type": "text", + "content": "As of December 31, 2025, there were approximately 5,861 and 1,611 stockholders of record of our Class A stock and Class C stock, respectively. Because many of our shares of Class A stock and Class C stock are held by brokers and other institutions on behalf of stockholders, we are unable to estimate the total number of stockholders represented by these record holders. As of December 31, 2025, there were approximately 57 stockholders of record of our Class B stock.", + "path": "GOOG-10-K-2025.pdf/PART II/Holders of Record", + "metadata": { + "length": 468, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "As", + "December", + "31", + "2025", + "approximately", + "861", + "611", + "stockholders", + "record", + "Class", + "stock", + "Because", + "shares", + "held", + "brokers", + "institutions", + "behalf", + "unable", + "estimate", + "total", + "number", + "represented", + "holders", + "57" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "abb8c718-a977-5a7c-8e77-e17a2d15b7a1", + "type": "text", + "content": "Dividend payments to stockholders of Class A, Class B, and Class C shares were \\$4.8 billion, \\$703 million, and \\$4.5 billion, respectively, for the year ended December 31, 2025. The company intends to pay quarterly cash dividends in the future, subject to review and approval by the company's Board of Directors in its sole discretion. We regularly evaluate our cash and capital structure, including the size, pace, and form of capital return to stockholders.", + "path": "GOOG-10-K-2025.pdf/PART II/Dividend Program", + "metadata": { + "length": 461, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Dividend", + "payments", + "stockholders", + "Class", + "shares", + "4.8", + "billion", + "703", + "million", + "4.5", + "year", + "ended", + "December", + "31", + "2025", + "The", + "company", + "intends", + "pay", + "quarterly", + "cash", + "dividends", + "future", + "subject", + "review", + "approval", + "Board", + "Directors", + "sole", + "discretion", + "We", + "regularly", + "evaluate", + "capital", + "structure", + "including", + "size", + "pace", + "form", + "return" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b3cd25a0-bdb7-58b3-9405-fcdccbc49bef", + "type": "table", + "content": "
PeriodTotal Number of Class A Shares Purchased (in thousands)(1)Total Number of Class C Shares Purchased (in thousands)(1)Average Price Paid per Class A Share(2)Average Price Paid per Class C Share(2)Total Number of Shares Purchased as Part of Publicly Announced Programs (in thousands)(1)Approximate Dollar Value of Shares that May Yet Be Purchased Under the Program (in millions)
October 1 - 3184113,305$ 253.20$ 255.0314,146$ 71,231
November 1 - 302095,885$ 286.00$ 281.466,094$ 69,503
December 1 - 3100$ 0.00$ 0.000$ 69,503
Total1,05019,19020,240
", + "path": "tables/table-2 Share Repurchase.html", + "metadata": { + "length": 921, + "summary": "table-3\nThe table shows monthly share repurchases from October to December, with 1,050 Class A and 19,190 Class C shares bought, totaling 20,240 shares under a publicly announced program, leaving $69,503 million available.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-2 Share Repurchase.html", + "keywords": [ + "share repurchase", + "Class A", + "Class C" + ], + "tokens": [] + } + }, + { + "chunk_id": "a97d03c5-7d76-5830-91be-b3fa8a6050c3", + "type": "text", + "content": "The following table presents information with respect to Alphabet's repurchases of Class A and Class C stock during the quarter ended December 31, 2025:\n\n[tables/table-2 Share Repurchase.html]\n\n(1) In April 2024, the company's Board of Directors authorized a \\$70.0 billion share repurchase program for its Class A and Class C shares. In April 2025, the company's Board of Directors authorized an additional \\$70.0 billion share repurchase program for its Class A and Class C shares. Repurchases are being executed from time to time, subject to general business and market conditions and other investment opportunities, through open market purchases or privately negotiated transactions, including through Rule 10b5-1 plans. The repurchase programs do not have an expiration date. For additional information related to share repurchases, see Note 11 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.\n(2) Average price paid per share includes costs associated with the repurchases.", + "path": "GOOG-10-K-2025.pdf/PART II/Issuer Purchases of Equity Securities", + "metadata": { + "length": 1099, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "presents", + "information", + "respect", + "Alphabet", + "repurchases", + "Class", + "stock", + "quarter", + "ended", + "December", + "31", + "2025", + "In", + "April", + "2024", + "company", + "Board", + "Directors", + "authorized", + "70.0", + "billion", + "share", + "repurchase", + "program", + "shares", + "additional", + "Repurchases", + "executed", + "time", + "subject", + "general", + "business", + "market", + "conditions", + "investment", + "opportunities", + "open", + "purchases", + "privately", + "negotiated", + "transactions", + "including", + "Rule", + "10b5", + "plans", + "programs", + "expiration", + "date", + "For", + "related", + "Note", + "11", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10", + "Average", + "price", + "paid", + "includes", + "costs" + ], + "keywords": [], + "connect_to": [ + { + "target": "b3cd25a0-bdb7-58b3-9405-fcdccbc49bef", + "relation": "embeds", + "ref": "[tables/table-2 Share Repurchase.html]", + "position": { + "start": 154, + "end": 192 + } + } + ] + } + }, + { + "chunk_id": "03c81db1-5030-5162-a756-87070b310e3f", + "type": "text", + "content": "The graph below matches Alphabet Inc. Class A's cumulative five-year total stockholder return on common stock with the cumulative total returns of the S&P 500 index, the NASDAQ Composite index, and the RDG Internet Composite index. The graph tracks the performance of a \\$100 investment in our common stock and in each index (with the reinvestment of all dividends) from December 31, 2020, to December 31, 2025. The returns shown are based on historical results and are not intended to suggest future performance.", + "path": "GOOG-10-K-2025.pdf/PART II/Stock Performance Graphs", + "metadata": { + "length": 513, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "graph", + "matches", + "Alphabet", + "Inc", + "Class", + "cumulative", + "year", + "total", + "stockholder", + "return", + "common", + "stock", + "returns", + "500", + "index", + "NASDAQ", + "Composite", + "RDG", + "Internet", + "tracks", + "performance", + "100", + "investment", + "reinvestment", + "dividends", + "December", + "31", + "2020", + "2025", + "shown", + "based", + "historical", + "results", + "intended", + "suggest", + "future" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "940870ef-e25f-5db3-a862-70c4db1d767c", + "type": "image", + "content": "\nThe graph below matches Alphabet Inc. Class A's cumulative five-year total stockholder return on common stock with the cumulative total returns of the S&P 500 index, the NASDAQ Composite index, and the RDG Internet Composite index. The graph tracks the performance of a \\$100 investment in our common stock and in each index (with the reinvestment of all dividends) from December 31, 2020, to December 31, 2025. The returns shown are based on historical results and are not intended to suggest future performance.
0\n[images/image-1-Stock Performance Comparison.jpg]\n", + "path": "images/image-1-Stock Performance Comparison.jpg", + "metadata": { + "length": 558, + "summary": "image-1\nChart Title: Stock Performance Comparison (implied)\nY-Axis: Dollar Value ($) ranging from $0 to $400\nX-Axis: Time range from 12/20 to 12/25\n\nData Series:\n- Alphabet Inc. Class A (Red Line): Starts at $100, peaks around $165 in late 2021, drops below $100 in late 2022, and shows a sharp upward trend ending above $350 by 12/25.\n- S&P 500 (Green Line): Starts at $100, fluctuates between roughly $100 and $170 until mid-2024, then rises to just under $200 by 12/25.\n- NASDAQ Composite (Yellow Line): Starts at $100, generally tracks below the S&P 500, dips significantly in 2022, and ends near $190 by 12/25.\n- RDG Internet Composite (Blue Line):", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "images/image-1-Stock Performance Comparison.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "0a62485f-cee1-587f-8412-7d7a1625c25e", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
DateAlphabet Inc. Class AS&P 500NASDAQ CompositeRDG Internet Composite
12/20$100$100$100$100
3/21$120$110$110$105
6/21$140$115$115$110
9/21$150$120$120$105
12/21$165$130$125$100
3/22$160$125$120$95
6/22$140$110$105$85
9/22$120$100$95$75
12/22$105$95$85$65
3/23$110$105$95$75
6/23$130$120$105$85
9/23$145$125$110$90
12/23$160$135$120$95
3/24$170$145$130$100
6/24$205$155$140$105
9/24$190$165$145$110
12/24$215$170$155$115
3/25$180$160$140$105
6/25$200$175$160$120
9/25$250$185$175$130
12/25$360$195$185$135
", + "path": "tables/table-3 Alphabet Stock vs Indexes.html", + "metadata": { + "length": 2780, + "summary": "table-4\nThe table tracks Alphabet Inc. Class A stock price alongside S&P 500, NASDAQ Composite, and RDG Internet Composite from Dec 2020 to Dec 2025, showing Alphabet's growth from $100 to $360.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-3 Alphabet Stock vs Indexes.html", + "keywords": [ + "Alphabet", + "S&P 500", + "NASDAQ" + ], + "tokens": [] + } + }, + { + "chunk_id": "1869c30a-2baa-5a14-8536-912821d272bc", + "type": "text", + "content": "ALPHABET INC. CLASS A COMMON STOCK\nAmong Alphabet Inc., the S&P 500 Index, the\nNASDAQ Composite Index, and the RDG Internet Composite Index\n\nThe graph below matches Alphabet Inc. Class A's cumulative five-year total stockholder return on common stock with the cumulative total returns of the S&P 500 index, the NASDAQ Composite index, and the RDG Internet Composite index. The graph tracks the performance of a \\$100 investment in our common stock and in each index (with the reinvestment of all dividends) from December 31, 2020, to December 31, 2025. The returns shown are based on historical results and are not intended to suggest future performance.
0\n[images/image-1-Stock Performance Comparison.jpg]\n\n
\nline chart\n\n[tables/table-3 Alphabet Stock vs Indexes.html]\n\n
\n\\*\\$100 invested on December 31, 2020, in stock or index, including reinvestment of dividends.\nCopyright $^{©}$ 2026 S&P, a division of The McGraw-Hill Companies Inc. All rights reserved.\nThe graph below matches Alphabet Inc. Class C's cumulative five-year total stockholder return on capital stock with the cumulative total returns of the S&P 500 index, the NASDAQ Composite index, and the RDG Internet Composite index. The graph tracks the performance of a \\$100 investment in our Class C capital stock and in each index (with the reinvestment of all dividends) from December 31, 2020, to December 31, 2025. The returns shown are based on historical results and are not intended to suggest future performance.", + "path": "GOOG-10-K-2025.pdf/PART II/Stock Performance Graphs/COMPARISON OF CUMULATIVE 5-YEAR TOTAL RETURN\\* ALPHABET INC. CLASS A COMMON STOCK", + "metadata": { + "length": 1563, + "summary": "The document compares the cumulative five-year total stockholder return of Alphabet Inc. Class A and Class C common stock against the S&P 500 Index, NASDAQ Composite Index, and RDG Internet Composite Index. It tracks a $100 investment from December 31, 2020, to December 31, 2025, with dividends reinvested. The returns are historical and not indicative of future performance. A line chart and supporting table provide detailed data on the performance of each investment over the period.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "ALPHABET", + "INC", + "CLASS", + "COMMON", + "STOCK", + "Among", + "Alphabet", + "Inc", + "500", + "Index", + "NASDAQ", + "Composite", + "RDG", + "Internet", + "The", + "graph", + "matches", + "Class", + "cumulative", + "year", + "total", + "stockholder", + "return", + "common", + "stock", + "returns", + "index", + "tracks", + "performance", + "100", + "investment", + "reinvestment", + "dividends", + "December", + "31", + "2020", + "2025", + "shown", + "based", + "historical", + "results", + "intended", + "suggest", + "future", + "details", + "summary", + "line", + "chart", + "invested", + "including", + "Copyright", + "2026", + "division", + "McGraw", + "Hill", + "Companies", + "All", + "rights", + "reserved", + "capital" + ], + "keywords": [ + "Alphabet Inc.", + "stock return", + "index comparison" + ], + "connect_to": [ + { + "target": "940870ef-e25f-5db3-a862-70c4db1d767c", + "relation": "embeds", + "ref": "[images/image-1-Stock Performance Comparison.jpg]", + "position": { + "start": 666, + "end": 715 + } + }, + { + "target": "0a62485f-cee1-587f-8412-7d7a1625c25e", + "relation": "embeds", + "ref": "[tables/table-3 Alphabet Stock vs Indexes.html]", + "position": { + "start": 758, + "end": 805 + } + } + ] + } + }, + { + "chunk_id": "39793db0-1fca-5f04-a5ed-d3fcb40c2632", + "type": "image", + "content": "\nThe graph below matches Alphabet Inc. Class C's cumulative five-year total stockholder return on capital stock with the cumulative total returns of the S&P 500 index, the NASDAQ Composite index, and the RDG Internet Composite index. The graph tracks the performance of a \\$100 investment in our Class C capital stock and in each index (with the reinvestment of all dividends) from December 31, 2020, to December 31, 2025. The returns shown are based on historical results and are not intended to suggest future performance.
0\n[images/image-2-Stock Performance Comparison (2020-2025).jpg]\n", + "path": "images/image-2-Stock Performance Comparison (2020-2025).jpg", + "metadata": { + "length": 568, + "summary": "image-2\nChart Title: Stock Performance Comparison (implied by legend and data)\nTime Range: December 2020 to December 2025\n\nData Series & Trends:\n- Alphabet Inc. Class C (Red Line): Starts at $100, peaks around $165 in late 2021, dips below $100 in late 2022, then rises sharply to over $350 by the end of the period.\n- S&P 500 (Green Line): Starts at $100, fluctuates between $100-$170 until mid-2024, then climbs steadily to approximately $200.\n- NASDAQ Composite (Yellow Line): Starts at $100, generally follows a similar but lower trajectory than the S&P 500, ending near $190.\n- RDG Internet Composite (Blue Line): Starts at $100, drops significantly to ~$60 in late 2022, recovers slowly to end around $", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "images/image-2-Stock Performance Comparison (2020-2025).jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "fe1a02f5-477c-5bd8-9a2a-8c34350f256e", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
DateAlphabet Inc. Class CS&P 500NASDAQ CompositeRDG Internet Composite
12/20$100$100$100$100
3/21$120$110$110$105
6/21$140$115$115$110
9/21$150$120$120$105
12/21$160$130$125$100
3/22$165$135$125$95
6/22$150$130$120$85
9/22$130$110$105$75
12/22$105$105$95$60
3/23$110$115$105$70
6/23$130$125$115$80
9/23$145$130$120$85
12/23$160$140$130$90
3/24$170$150$140$95
6/24$210$160$150$105
9/24$190$165$155$110
12/24$220$170$160$115
3/25$180$165$155$105
6/25$200$175$165$115
9/25$250$185$175$125
12/25$360$195$185$135
", + "path": "tables/table-4 Alphabet vs S&P 500.html", + "metadata": { + "length": 2782, + "summary": "table-5\nThe table tracks Alphabet Inc. Class C stock price from $100 (12/20) to $360 (12/25), compared to S&P 500, NASDAQ Composite, and RDG Internet Composite indices over the same period.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-4 Alphabet vs S&P 500.html", + "keywords": [ + "Alphabet", + "S&P 500", + "NASDAQ" + ], + "tokens": [] + } + }, + { + "chunk_id": "37f93f33-b2e2-577e-bdc1-8c5180cae55f", + "type": "text", + "content": "ALPHABET INC. CLASS C CAPITAL STOCK\nAmong Alphabet Inc., the S&P 500 Index, the\nNASDAQ Composite Index, and the RDG Internet Composite Index\n\nThe graph below matches Alphabet Inc. Class C's cumulative five-year total stockholder return on capital stock with the cumulative total returns of the S&P 500 index, the NASDAQ Composite index, and the RDG Internet Composite index. The graph tracks the performance of a \\$100 investment in our Class C capital stock and in each index (with the reinvestment of all dividends) from December 31, 2020, to December 31, 2025. The returns shown are based on historical results and are not intended to suggest future performance.
0\n[images/image-2-Stock Performance Comparison (2020-2025).jpg]\n\n
\nline chart\n\n[tables/table-4 Alphabet vs S&P 500.html]\n\n
\n\\*\\$100 invested on December 31, 2020, in stock or in index, including reinvestment of dividends.\nCopyright $^{©}$ 2026 S&P, a division of The McGraw-Hill Companies Inc. All rights reserved.", + "path": "GOOG-10-K-2025.pdf/PART II/Stock Performance Graphs/COMPARISON OF CUMULATIVE 5-YEAR TOTAL RETURN\\* ALPHABET INC. CLASS C CAPITAL STOCK", + "metadata": { + "length": 1053, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "ALPHABET", + "INC", + "CLASS", + "CAPITAL", + "STOCK", + "Among", + "Alphabet", + "Inc", + "500", + "Index", + "NASDAQ", + "Composite", + "RDG", + "Internet", + "The", + "graph", + "matches", + "Class", + "cumulative", + "year", + "total", + "stockholder", + "return", + "capital", + "stock", + "returns", + "index", + "tracks", + "performance", + "100", + "investment", + "reinvestment", + "dividends", + "December", + "31", + "2020", + "2025", + "shown", + "based", + "historical", + "results", + "intended", + "suggest", + "future", + "details", + "summary", + "line", + "chart", + "invested", + "including", + "Copyright", + "2026", + "division", + "McGraw", + "Hill", + "Companies", + "All", + "rights", + "reserved" + ], + "keywords": [], + "connect_to": [ + { + "target": "39793db0-1fca-5f04-a5ed-d3fcb40c2632", + "relation": "embeds", + "ref": "[images/image-2-Stock Performance Comparison (2020-2025).jpg]", + "position": { + "start": 677, + "end": 738 + } + }, + { + "target": "fe1a02f5-477c-5bd8-9a2a-8c34350f256e", + "relation": "embeds", + "ref": "[tables/table-4 Alphabet vs S&P 500.html]", + "position": { + "start": 781, + "end": 822 + } + } + ] + } + }, + { + "chunk_id": "9ea3153b-ec56-52f3-9f85-265ad97100f4", + "type": "text", + "content": "ITEM 7. MANAGEMENT'S DISCUSSION AND ANALYSIS OF FINANCIAL CONDITION AND RESULTS OF OPERATIONS\nPlease read the following discussion and analysis of our financial condition and results of operations together with \"Note about Forward-Looking Statements,\" Part I, Item 1 \"Business,\" Part I, Item 1A \"Risk Factors,\" and our consolidated financial statements and related notes included under Item 8 of this Annual Report on Form 10-K.\nThe following section generally discusses 2025 results compared to 2024 results. Discussion of 2024 results compared to 2023 results to the extent not included in this report can be found in Item 7 of our 2024 Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]", + "metadata": { + "length": 666, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "ITEM", + "MANAGEMENT", + "DISCUSSION", + "AND", + "ANALYSIS", + "OF", + "FINANCIAL", + "CONDITION", + "RESULTS", + "OPERATIONS", + "Please", + "read", + "discussion", + "analysis", + "financial", + "condition", + "results", + "operations", + "Note", + "Forward", + "Looking", + "Statements", + "Part", + "Item", + "Business", + "1A", + "Risk", + "Factors", + "consolidated", + "statements", + "related", + "notes", + "included", + "Annual", + "Report", + "Form", + "10", + "The", + "section", + "generally", + "discusses", + "2025", + "compared", + "2024", + "Discussion", + "2023", + "extent", + "report", + "found" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "73db955f-7e14-56dd-adc6-0ee65dc6b7b2", + "type": "text", + "content": "Alphabet is a collection of businesses — the largest of which is Google. We report Google in two segments, Google Services and Google Cloud, and all non-Google businesses collectively as Other Bets. Supporting these businesses, we have centralized certain AI-related research and development focused on advanced research in AI and developing the frontier models that serve our businesses, which is reported in Alphabet-level activities. For further details on our segments, see Part I, Item 1 Business and Note 15 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Understanding Alphabet's Financial Results", + "metadata": { + "length": 618, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Alphabet", + "collection", + "businesses", + "largest", + "Google", + "We", + "report", + "segments", + "Services", + "Cloud", + "collectively", + "Other", + "Bets", + "Supporting", + "centralized", + "AI", + "related", + "research", + "development", + "focused", + "advanced", + "developing", + "frontier", + "models", + "serve", + "reported", + "level", + "activities", + "For", + "details", + "Part", + "Item", + "Business", + "Note", + "15", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "975324c7-e7b2-5624-9ed8-a68d7e60f725", + "type": "text", + "content": "The following long-term trends have contributed to the results of our consolidated operations, and we anticipate that they will continue to affect our future results:\n\\- As we continue to grow our business and meet the evolving behaviors and needs of our users and customers, our revenue growth and mix along with our cost and margin profiles are being influenced by a number of factors, including:\nExpanded AI Offerings in our Products and Services: The continuing evolution of the online world has contributed to the growth of our business. We expect that this evolution, including user engagement with AI products and services, will continue to benefit our business and our revenues. As we continue to incorporate AI into our products and services, such as with AI Overviews and AI Mode in Search, and with enterprise AI solutions on our Google Cloud Platform, we may monetize differently than our historical consumer and enterprise offerings which could affect revenue growth rates and margin trends. When developing new products and services we generally focus first on user experience and then on monetization. At the same time, we face increasing competition, including from other developers and providers of AI products and services, which may affect our revenues.\nIncreasing Revenues Beyond Advertising: Revenues from cloud, consumer subscriptions, platforms, and devices, which may have differing characteristics than our advertising revenues, have grown over time. Certain of these revenues have been growing at a rate higher than our advertising revenues, becoming a larger percentage of our consolidated revenues, and we expect this trend to continue. The margins on these revenues vary significantly and are generally lower than the margins on our advertising revenues.\nIncreased Investment in Technical Infrastructure: We continue to invest in capital expenditures as we scale our technical infrastructure, in particular for AI, to meet the demand of our users and enterprise customers and to support research internally. We invested heavily in capital expenditures in 2025 and in 2026, we expect to significantly increase, relative to 2025, our investment in our technical infrastructure, including servers and network equipment, and data centers. The costs associated with operating our technical infrastructure - depreciation, energy, equipment, and network capacity - are expected to significantly increase as developing and serving AI offerings require more compute power than our historical consumer and enterprise offerings. While our technical infrastructure costs increase, we expect to continue to drive efficiencies in our data centers, for example, through the design of our AI models and our TPU and GPU-based technical infrastructure.\nContinued Investment in Intellectual Property through R&D and Acquisitions: We continue to make significant research and development investments in areas of strategic focus as we seek to develop new, innovative offerings, and improve our existing offerings across our businesses. Acquisitions and strategic investments remain important elements in our use of capital and contribute to the breadth and depth of our offerings, expand our expertise in engineering and other functional areas, and build strong partnerships around strategic initiatives.\nTraffic Acquisition Costs Growth and Rate Changes: We expect traffic acquisition costs (\"TAC\") paid to our distribution partners and Google Network partners to increase as our advertising revenues grow. Our overall TAC as a percentage of our advertising revenues (\"TAC rate\") has been decreasing primarily due to a revenue mix shift from Google Network properties to Google Search & other properties. Our TAC rate will continue to be affected by changes in device mix; geographic mix; partner agreement terms; partner mix; the percentage of queries channeled through paid access points; product mix; the relative revenue growth rates of advertising revenues from different channels; and revenue share terms.\n• We have raised capital through external financing in the form of debt and we may continue to seek debt or other forms of financing in the future to support our capital and operating needs.\nIn 2025, we raised capital through the issuance of debt and we expect to continue to assess the use of debt and other forms of financing in the future. We expect to continue to enter into finance leases, primarily for data centers. Additionally, in 2025, we provided credit support, such as through backstops and guarantees, to certain infrastructure related counterparties and may continue to provide additional credit support in the future.\n• We face an evolving regulatory environment, and we are subject to claims, lawsuits, investigations, and other forms of potential legal liability, which could affect our business practices and financial results.\nChanges in social, political, economic, tax, and regulatory conditions or in laws and policies governing a wide range of topics and related legal matters, including investigations, lawsuits, and regulatory actions, have resulted in fines and caused us to change our business practices. As the regulatory environment continues to evolve, we may continue to incur fines and we expect increased costs associated with compliance, modifications to our products and services, and limitations on our ability to pursue certain business practices. For additional information, see Part I, Item 1A Risk Factors and Legal Matters in Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Trends in Our Business and Financial Effect", + "metadata": { + "length": 5601, + "summary": "The passage outlines long-term trends affecting consolidated operations, including expanded AI offerings that may alter monetization and face competition; increasing non-advertising revenues from cloud and subscriptions with generally lower margins; significant capital expenditure growth for technical infrastructure, especially AI, driving higher operating costs; continued R&D and acquisition investments; rising traffic acquisition costs with a shifting rate; potential future debt financing; and an evolving regulatory environment leading to compliance costs and business practice changes.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "long", + "term", + "trends", + "contributed", + "results", + "consolidated", + "operations", + "anticipate", + "continue", + "affect", + "future", + "As", + "grow", + "business", + "meet", + "evolving", + "behaviors", + "users", + "customers", + "revenue", + "growth", + "mix", + "cost", + "margin", + "profiles", + "influenced", + "number", + "factors", + "including", + "Expanded", + "AI", + "Offerings", + "Products", + "Services", + "continuing", + "evolution", + "online", + "world", + "We", + "expect", + "user", + "engagement", + "products", + "services", + "benefit", + "revenues", + "incorporate", + "Overviews", + "Mode", + "Search", + "enterprise", + "solutions", + "Google", + "Cloud", + "Platform", + "monetize", + "differently", + "historical", + "consumer", + "offerings", + "rates", + "When", + "developing", + "generally", + "focus", + "experience", + "monetization", + "At", + "time", + "face", + "increasing", + "competition", + "developers", + "providers", + "Increasing", + "Revenues", + "Beyond", + "Advertising", + "cloud", + "subscriptions", + "platforms", + "devices", + "differing", + "characteristics", + "advertising", + "grown", + "Certain", + "growing", + "rate", + "higher", + "larger", + "percentage", + "trend", + "margins", + "vary", + "significantly", + "lower", + "Increased", + "Investment", + "Technical", + "Infrastructure", + "invest", + "capital", + "expenditures", + "scale", + "technical", + "infrastructure", + "demand", + "support", + "research", + "internally", + "invested", + "heavily", + "2025", + "2026", + "increase", + "relative", + "investment", + "servers", + "network", + "equipment", + "data", + "centers", + "costs", + "operating", + "depreciation", + "energy", + "capacity", + "expected", + "serving", + "require", + "compute", + "power", + "While", + "drive", + "efficiencies", + "design", + "models", + "TPU", + "GPU", + "based", + "Continued", + "Intellectual", + "Property", + "Acquisitions", + "make", + "significant", + "development", + "investments", + "areas", + "strategic", + "seek", + "develop", + "innovative", + "improve", + "existing", + "businesses", + "remain", + "important", + "elements", + "contribute", + "breadth", + "depth", + "expand", + "expertise", + "engineering", + "functional", + "build", + "strong", + "partnerships", + "initiatives", + "Traffic", + "Acquisition", + "Costs", + "Growth", + "Rate", + "Changes", + "traffic", + "acquisition", + "TAC", + "paid", + "distribution", + "partners", + "Network", + "Our", + "decreasing", + "primarily", + "due", + "shift", + "properties", + "affected", + "device", + "geographic", + "partner", + "agreement", + "terms", + "queries", + "channeled", + "access", + "points", + "product", + "channels", + "share", + "raised", + "external", + "financing", + "form", + "debt", + "forms", + "In", + "issuance", + "assess", + "enter", + "finance", + "leases", + "Additionally", + "provided", + "credit", + "backstops", + "guarantees", + "related", + "counterparties", + "provide", + "additional", + "regulatory", + "environment", + "subject", + "claims", + "lawsuits", + "investigations", + "potential", + "legal", + "liability", + "practices", + "financial", + "social", + "political", + "economic", + "tax", + "conditions", + "laws", + "policies", + "governing", + "wide", + "range", + "topics", + "matters", + "actions", + "resulted", + "fines", + "caused", + "change", + "continues", + "evolve", + "incur", + "increased", + "compliance", + "modifications", + "limitations", + "ability", + "pursue", + "For", + "information", + "Part", + "Item", + "1A", + "Risk", + "Factors", + "Legal", + "Matters", + "Note", + "10", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Annual", + "Report", + "Form" + ], + "keywords": [ + "AI investment", + "revenue mix", + "regulatory risks" + ], + "connect_to": [] + } + }, + { + "chunk_id": "3f6a010d-cc2c-5dd6-9ec7-449c72f4e80a", + "type": "text", + "content": "We generate revenues by delivering relevant, cost-effective online advertising; cloud-based solutions that provide enterprise customers of all sizes with infrastructure, platform services, and applications; and sales of other products and services, such as fees received for subscription-based products, apps and in-app purchases, and devices. For additional information on how we recognize revenue, see Note 1 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.\nIn addition to the long-term trends and their financial effect on our business discussed above, fluctuations in our revenues have been and may continue to be affected by a combination of factors, including:\n• changes in foreign currency exchange rates;\n• changes in pricing, such as those resulting from changes in fee structures, discounts, and customer incentives;\n- general economic conditions and various external dynamics, including geopolitical events, regulations, and other measures and their effect on advertiser, consumer, and enterprise spending;\n- new product, service, and market launches; and\n- seasonality.\nAdditionally, fluctuations in our revenues generated from advertising (\"Google advertising\"), other sources (\"Google subscriptions, platforms, and devices\"), Google Cloud, and Other Bets have been, and may continue to be, affected by other factors unique to each set of revenues, as described below.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics", + "metadata": { + "length": 1437, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "generate", + "revenues", + "delivering", + "relevant", + "cost", + "effective", + "online", + "advertising", + "cloud", + "based", + "solutions", + "provide", + "enterprise", + "customers", + "sizes", + "infrastructure", + "platform", + "services", + "applications", + "sales", + "products", + "fees", + "received", + "subscription", + "apps", + "app", + "purchases", + "devices", + "For", + "additional", + "information", + "recognize", + "revenue", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10", + "In", + "addition", + "long", + "term", + "trends", + "financial", + "effect", + "business", + "discussed", + "fluctuations", + "continue", + "affected", + "combination", + "factors", + "including", + "foreign", + "currency", + "exchange", + "rates", + "pricing", + "resulting", + "fee", + "structures", + "discounts", + "customer", + "incentives", + "general", + "economic", + "conditions", + "external", + "dynamics", + "geopolitical", + "events", + "regulations", + "measures", + "advertiser", + "consumer", + "spending", + "product", + "service", + "market", + "launches", + "seasonality", + "Additionally", + "generated", + "Google", + "sources", + "subscriptions", + "platforms", + "Cloud", + "Other", + "Bets", + "unique", + "set" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "563b357c-c855-5f6d-adf4-952742e5df67", + "type": "text", + "content": "Google Services revenues consist of Google advertising as well as Google subscriptions, platforms, and devices revenues.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics/Google Services", + "metadata": { + "length": 120, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "Services", + "revenues", + "consist", + "advertising", + "subscriptions", + "platforms", + "devices" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "eab7f293-a847-5f21-a169-bd0a037731a5", + "type": "text", + "content": "Google advertising revenues are comprised of the following:\n- Google Search & other, which includes revenues generated on Google search properties (including revenues from traffic generated by search distribution partners who use Google.com as their default search in browsers, toolbars, etc.), and other Google owned and operated properties like Gmail, Google Maps, and Google Play;\n• YouTube ads, which includes revenues generated on YouTube properties; and\n- Google Network, which includes revenues generated on Google Network properties participating in AdMob, AdSense, and Google Ad Manager.\nWe use certain metrics to track how well traffic across various properties is monetized as it relates to our advertising revenues: paid clicks and cost-per-click pertain to traffic on Google Search & other properties, while impressions and cost-per-impression pertain to traffic on our Google Network properties.\nPaid clicks represent engagement by users and include clicks on advertisements by end-users on Google search properties and other Google owned and operated properties including Gmail, Google Maps, and Google Play. Cost-per-click is defined as click-driven revenues divided by our total number of paid clicks and represents the average amount we charge advertisers for each engagement by users.\nImpressions include impressions displayed to users on Google Network properties participating primarily in AdMob, AdSense, and Google Ad Manager. Cost-per-impression is defined as impression-based and click-based revenues divided by our total number of impressions, and represents the average amount we charge advertisers for each impression displayed to users.\nAs our business evolves, we periodically review, refine, and update our methodologies for monitoring, gathering, and counting the number of paid clicks and the number of impressions, and for identifying the revenues generated by the corresponding click and impression activity.\nFluctuations in our advertising revenues, as well as the change in paid clicks and cost-per-click on Google Search & other properties and the change in impressions and cost-per-impression on Google Network properties and the correlation between these items have been, and may continue to be, affected by factors in addition to the general factors described above, such as:\n• advertiser competition for keywords;\n• changes in advertising quality, formats, delivery, or policy;\n• changes in device mix;\n- seasonal fluctuations in internet usage, advertising expenditures, and underlying business trends, such as traditional retail seasonality; and\n- traffic growth in emerging markets compared to more mature markets and across various verticals and channels.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics/Google Services/Google Advertising", + "metadata": { + "length": 2701, + "summary": "Google advertising revenues come from Google Search & other properties, YouTube ads, and Google Network. Key metrics include paid clicks and cost-per-click for Search, and impressions and cost-per-impression for Network. Revenues are affected by factors like advertiser competition, ad quality, device mix, seasonality, and emerging market growth.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "advertising", + "revenues", + "comprised", + "Search", + "includes", + "generated", + "search", + "properties", + "including", + "traffic", + "distribution", + "partners", + "default", + "browsers", + "toolbars", + "owned", + "operated", + "Gmail", + "Maps", + "Play", + "YouTube", + "ads", + "Network", + "participating", + "AdMob", + "AdSense", + "Ad", + "Manager", + "We", + "metrics", + "track", + "monetized", + "relates", + "paid", + "clicks", + "cost", + "click", + "pertain", + "impressions", + "impression", + "Paid", + "represent", + "engagement", + "users", + "include", + "advertisements", + "end", + "Cost", + "defined", + "driven", + "divided", + "total", + "number", + "represents", + "average", + "amount", + "charge", + "advertisers", + "Impressions", + "displayed", + "primarily", + "based", + "As", + "business", + "evolves", + "periodically", + "review", + "refine", + "update", + "methodologies", + "monitoring", + "gathering", + "counting", + "identifying", + "activity", + "Fluctuations", + "change", + "correlation", + "items", + "continue", + "affected", + "factors", + "addition", + "general", + "advertiser", + "competition", + "keywords", + "quality", + "formats", + "delivery", + "policy", + "device", + "mix", + "seasonal", + "fluctuations", + "internet", + "usage", + "expenditures", + "underlying", + "trends", + "traditional", + "retail", + "seasonality", + "growth", + "emerging", + "markets", + "compared", + "mature", + "verticals", + "channels" + ], + "keywords": [ + "advertising", + "paid clicks", + "impressions" + ], + "connect_to": [] + } + }, + { + "chunk_id": "0974902e-d672-55cf-ae75-1a6788144e2a", + "type": "text", + "content": "Google subscriptions, platforms, and devices revenues are comprised of the following:\n- consumer subscriptions, which primarily include revenues from YouTube services, such as YouTube TV, YouTube Music and Premium, and NFL Sunday Ticket, as well as Google One, which offers access to our most capable Gemini models;\n- platforms, which primarily include revenues from Google Play sales of apps and in-app purchases;\n• devices, which primarily include sales of the Pixel family of devices; and\n• other products and services.\nFluctuations in our Google subscriptions, platforms, and devices revenues have been, and may continue to be, affected by factors in addition to the general factors described above, such as changes in customer usage and demand, number of subscribers, and the timing of product launches.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics/Google Services/Google Subscriptions, Platforms, and Devices", + "metadata": { + "length": 808, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "subscriptions", + "platforms", + "devices", + "revenues", + "comprised", + "consumer", + "primarily", + "include", + "YouTube", + "services", + "TV", + "Music", + "Premium", + "NFL", + "Sunday", + "Ticket", + "One", + "offers", + "access", + "capable", + "Gemini", + "models", + "Play", + "sales", + "apps", + "app", + "purchases", + "Pixel", + "family", + "products", + "Fluctuations", + "continue", + "affected", + "factors", + "addition", + "general", + "customer", + "usage", + "demand", + "number", + "subscribers", + "timing", + "product", + "launches" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "64d5d750-eb27-5e0e-854a-6d4de7c26d96", + "type": "text", + "content": "Google Cloud revenues are comprised of the following:\n- Google Cloud Platform primarily generates consumption-based fees and subscriptions for infrastructure, platform, and other services. These services provide access to solutions such as AI offerings including our enterprise AI infrastructure, Vertex AI platform, and Gemini Enterprise; cybersecurity offerings; and data and analytics solutions;\n- Google Workspace includes subscriptions for cloud-based communication and collaboration tools for enterprises, such as Gmail, Docs, Calendar, Drive, and Meet, with integrated features like Gemini for Google Workspace; and\n• other enterprise services.\nFluctuations in our Google Cloud revenues have been, and may continue to be, affected by factors in addition to the general factors described above, such as changes in customer usage, demand, and supply availability.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics/Google Services/Google Cloud", + "metadata": { + "length": 868, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "Cloud", + "revenues", + "comprised", + "Platform", + "primarily", + "generates", + "consumption", + "based", + "fees", + "subscriptions", + "infrastructure", + "platform", + "services", + "These", + "provide", + "access", + "solutions", + "AI", + "offerings", + "including", + "enterprise", + "Vertex", + "Gemini", + "Enterprise", + "cybersecurity", + "data", + "analytics", + "Workspace", + "includes", + "cloud", + "communication", + "collaboration", + "tools", + "enterprises", + "Gmail", + "Docs", + "Calendar", + "Drive", + "Meet", + "integrated", + "features", + "Fluctuations", + "continue", + "affected", + "factors", + "addition", + "general", + "customer", + "usage", + "demand", + "supply", + "availability" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "70bce9c9-092f-58fa-8e74-8b766e3b8166", + "type": "text", + "content": "Revenues from Other Bets are generated primarily from the sale of autonomous transportation services and internet services.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics/Google Services/Other Bets", + "metadata": { + "length": 123, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Revenues", + "Other", + "Bets", + "generated", + "primarily", + "sale", + "autonomous", + "transportation", + "services", + "internet" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "95b151ef-1975-571e-b52d-3399f43e825c", + "type": "text", + "content": "Our cost structure has two components: cost of revenues and operating expenses. Our operating expenses include costs related to research and development, sales and marketing, and general and administrative functions. Certain of our costs and expenses, including those associated with the operation of our technical infrastructure as well as components of our operating expenses, are generally less variable in nature and may not correlate to changes in revenue. Additionally, fluctuations in employee compensation expenses may not directly correlate with changes in headcount, due to factors such as annual SBC awards that vest over time.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses", + "metadata": { + "length": 638, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "cost", + "structure", + "components", + "revenues", + "operating", + "expenses", + "include", + "costs", + "related", + "research", + "development", + "sales", + "marketing", + "general", + "administrative", + "functions", + "Certain", + "including", + "operation", + "technical", + "infrastructure", + "generally", + "variable", + "nature", + "correlate", + "revenue", + "Additionally", + "fluctuations", + "employee", + "compensation", + "directly", + "headcount", + "due", + "factors", + "annual", + "SBC", + "awards", + "vest", + "time" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d94184e9-a762-5d60-afaf-19c0a8c0807b", + "type": "text", + "content": "Cost of revenues is comprised of TAC and other costs of revenues.\n• TAC includes:\namounts paid to our distribution partners who make available our search access points and other ad-supported services. Our distribution partners include browser providers, mobile carriers, original equipment manufacturers, and software developers; and\n• amounts paid to Google Network partners primarily for ads displayed on their properties.\n• Other cost of revenues primarily includes:\ncontent acquisition costs, which are payments to content providers from whom we license video and other content for distribution, primarily related to YouTube (we pay fees to these content providers based on revenues generated, subscriber counts, or a flat fee);\n• depreciation expense, primarily related to our technical infrastructure;\n• employee compensation expenses related to our technical infrastructure and other operations such as content review and customer and product support;\ninventory and other costs related to the devices we sell; and\nother technical infrastructure operations costs, including energy, equipment, and network capacity costs.\nTAC as a percentage of revenues generated from ads placed on Google Network properties are significantly higher than TAC as a percentage of revenues generated from ads placed on Google Search & other properties, because most of the advertiser revenues from ads served on Google Network properties are paid as TAC to our Google Network partners.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses/Cost of Revenues", + "metadata": { + "length": 1471, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Cost", + "revenues", + "comprised", + "TAC", + "costs", + "includes", + "amounts", + "paid", + "distribution", + "partners", + "make", + "search", + "access", + "points", + "ad", + "supported", + "services", + "Our", + "include", + "browser", + "providers", + "mobile", + "carriers", + "original", + "equipment", + "manufacturers", + "software", + "developers", + "Google", + "Network", + "primarily", + "ads", + "displayed", + "properties", + "Other", + "cost", + "content", + "acquisition", + "payments", + "license", + "video", + "related", + "YouTube", + "pay", + "fees", + "based", + "generated", + "subscriber", + "counts", + "flat", + "fee", + "depreciation", + "expense", + "technical", + "infrastructure", + "employee", + "compensation", + "expenses", + "operations", + "review", + "customer", + "product", + "support", + "inventory", + "devices", + "sell", + "including", + "energy", + "network", + "capacity", + "percentage", + "significantly", + "higher", + "Search", + "advertiser", + "served" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a734fcb3-cde2-5c8a-aa36-ff2a8c357c4c", + "type": "text", + "content": "Operating expenses are generally incurred during our normal course of business, which we categorize as either research and development, sales and marketing, or general and administrative.\nThe main components of our research and development expenses are:\n• depreciation expense, primarily related to our technical infrastructure;\n- employee compensation expenses for engineering and technical employees responsible for research and development related to our existing and new products and services;\n• other technical infrastructure operations costs, including energy, equipment, and network capacity costs; and\n- third-party services fees primarily relating to consulting and outsourced services in support of our engineering and product development efforts.\nThe main components of our sales and marketing expenses are:\n- employee compensation expenses for employees engaged in sales and marketing, sales support, and certain customer service functions; and\n- spend relating to our advertising and promotional activities in support of our products and services.\nThe main components of our general and administrative expenses are:\n- employee compensation expenses for employees in finance, human resources, information technology, legal, and other administrative support functions;\n• expenses relating to legal and other matters, including certain fines and settlements; and\n- third-party services fees, including audit, consulting, outside legal, and other outsourced administrative services.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses/Operating Expenses", + "metadata": { + "length": 1491, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Operating", + "expenses", + "generally", + "incurred", + "normal", + "business", + "categorize", + "research", + "development", + "sales", + "marketing", + "general", + "administrative", + "The", + "main", + "components", + "depreciation", + "expense", + "primarily", + "related", + "technical", + "infrastructure", + "employee", + "compensation", + "engineering", + "employees", + "responsible", + "existing", + "products", + "services", + "operations", + "costs", + "including", + "energy", + "equipment", + "network", + "capacity", + "party", + "fees", + "relating", + "consulting", + "outsourced", + "support", + "product", + "efforts", + "engaged", + "customer", + "service", + "functions", + "spend", + "advertising", + "promotional", + "activities", + "finance", + "human", + "resources", + "information", + "technology", + "legal", + "matters", + "fines", + "settlements", + "audit" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c6f65398-6674-58e6-bcaa-ee7cc7160add", + "type": "text", + "content": "OI&E, net primarily consists of interest income (expense), the effect of foreign currency exchange gains (losses), net gains (losses) and impairment on our marketable and non-marketable securities and income (loss) and impairment from our equity method investments.\nFor additional information, including how we account for our investments and factors that can drive fluctuations in the value of our investments, see Note 1 and Note 3 of the Notes to Consolidated Financial Statements included in Item 8 as well as Item 7A Quantitative and Qualitative Disclosures About Market Risk of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses/Other Income (Expense), Net", + "metadata": { + "length": 616, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "OI", + "net", + "primarily", + "consists", + "interest", + "income", + "expense", + "effect", + "foreign", + "currency", + "exchange", + "gains", + "losses", + "impairment", + "marketable", + "securities", + "loss", + "equity", + "method", + "investments", + "For", + "additional", + "information", + "including", + "account", + "factors", + "drive", + "fluctuations", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "7A", + "Quantitative", + "Qualitative", + "Disclosures", + "About", + "Market", + "Risk", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d82a6b7b-8b0f-53dd-8627-cecbcf104c1a", + "type": "text", + "content": "Provision for income taxes represents the estimated amount of federal, state, and foreign income taxes incurred in the US and the many jurisdictions in which we operate. The provision includes the effect of reserve provisions and changes to reserves that are considered appropriate as well as the related net interest and penalties.\nFor additional information, including a reconciliation of the US federal statutory rate to our effective tax rate, see Note 14 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses/Provision for Income Taxes", + "metadata": { + "length": 564, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Provision", + "income", + "taxes", + "represents", + "estimated", + "amount", + "federal", + "state", + "foreign", + "incurred", + "US", + "jurisdictions", + "operate", + "The", + "provision", + "includes", + "effect", + "reserve", + "provisions", + "reserves", + "considered", + "related", + "net", + "interest", + "penalties", + "For", + "additional", + "information", + "including", + "reconciliation", + "statutory", + "rate", + "effective", + "tax", + "Note", + "14", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fb4f5994-9126-534f-9dcd-1d08d979a0b4", + "type": "table", + "content": "
Year Ended December 31,$ Change% Change
20242025
Consolidated revenues$ 350,018$ 402,836$ 52,81815 %
Cost of revenues$ 146,306$ 162,535$ 16,22911 %
Operating expenses$ 91,322$ 111,262$ 19,94022 %
Operating income$ 112,390$ 129,039$ 16,64915 %
Operating margin32 %32 %0 %
Other income (expense), net$ 7,425$ 29,787$ 22,362301 %
Net income$ 100,118$ 132,170$ 32,05232 %
Diluted net income per share(1)$ 8.04$ 10.81$ 2.7734 %
", + "path": "tables/table-5 Financial Results.html", + "metadata": { + "length": 980, + "summary": "table-6\nThe table shows financial data for years ended December 31, 2024 and 2025. Consolidated revenues increased 15% to $402,836, net income rose 32% to $132,170, and diluted net income per share grew 34% to $10.81.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-5 Financial Results.html", + "keywords": [ + "revenues", + "net income", + "operating income" + ], + "tokens": [] + } + }, + { + "chunk_id": "643e21e4-bb5b-5bc0-9dbc-a4d309218737", + "type": "text", + "content": "The following table summarizes consolidated financial results (in millions, except for per share information and percentages):\n\n[tables/table-5 Financial Results.html]\n\n(1) For additional information on the calculation of diluted net income per share, see Note 12 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.\n- Revenues were \\$402.8 billion, an increase of 15% year over year, primarily driven by an increase in Google Services revenues of \\$37.8 billion, or 12%, and an increase in Google Cloud revenues of \\$15.5 billion, or 36%.\n- Cost of revenues was \\$162.5 billion, an increase of 11% year over year, primarily driven by increases in TAC, content acquisition costs, and depreciation expense.\n- Operating expenses were \\$111.3 billion, an increase of 22% year over year, primarily driven by increases in employee compensation expenses, expenses related to legal and other matters, and depreciation expense.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Executive Overview", + "metadata": { + "length": 995, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "summarizes", + "consolidated", + "financial", + "results", + "millions", + "share", + "information", + "percentages", + "For", + "additional", + "calculation", + "diluted", + "net", + "income", + "Note", + "12", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10", + "Revenues", + "402.8", + "billion", + "increase", + "15%", + "year", + "primarily", + "driven", + "Google", + "Services", + "revenues", + "37.8", + "12%", + "Cloud", + "15.5", + "36%", + "Cost", + "162.5", + "11%", + "increases", + "TAC", + "content", + "acquisition", + "costs", + "depreciation", + "expense", + "Operating", + "expenses", + "111.3", + "22%", + "employee", + "compensation", + "related", + "legal", + "matters" + ], + "keywords": [], + "connect_to": [ + { + "target": "fb4f5994-9126-534f-9dcd-1d08d979a0b4", + "relation": "embeds", + "ref": "[tables/table-5 Financial Results.html]", + "position": { + "start": 128, + "end": 167 + } + } + ] + } + }, + { + "chunk_id": "a4b5d9a4-cf8c-5906-b920-b3a276af44c6", + "type": "text", + "content": "- In 2025, we entered into definitive agreements to acquire Wiz, a leading cloud security platform, for \\$32.0 billion, and Intersect, a provider of data center and energy infrastructure solutions, for \\$4.8 billion in cash plus the assumption of debt. Both acquisitions are expected to close in 2026, subject to customary closing conditions, including the receipt of regulatory approvals.\n- In 2025, we issued senior unsecured notes for net proceeds of \\$37.3 billion, to be used for general corporate purposes.\n- OI&E of \\$29.8 billion for the year ended December 31, 2025 included net gains on equity securities of \\$24.1 billion, primarily related to unrealized gains on our non-marketable equity securities.\n- Other Bets operating loss of \\$7.5 billion for the year ended December 31, 2025 included a \\$2.1 billion employee compensation charge recognized in the fourth quarter for Waymo, primarily reflected in research and development expenses, based on estimated stock valuation. In February 2026, Waymo announced an investment round of \\$16.0 billion, the significant majority of which was funded by Alphabet.\n- Changes to U.S. tax law enacted on July 4, 2025, allow, among other things, for immediate expensing of domestic research and experimentation costs and accelerated depreciation on eligible capital expenditures, the effects of which are included in operating cash flows for the year ended December 31, 2025.\n- Repurchases of Class A and Class C shares were \\$6.5 billion and \\$38.9 billion, respectively, totaling \\$45.4 billion for the year ended December 31, 2025.\n- Operating cash flow was \\$164.7 billion for the year ended December 31, 2025.\n- Capital expenditures, which primarily reflected investments in technical infrastructure, were \\$91.4 billion for the year ended December 31, 2025.\n• As of December 31, 2025, we had 190,820 employees.\nWe are monitoring ongoing developments surrounding international trade and the macroeconomic environment. As a result of volatility in international trade and financial markets, we may experience direct and indirect effects on our business, operations, and financial results. Our past results may not be indicative of our future performance, and our financial results may differ materially from historical trends.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Other Information:", + "metadata": { + "length": 2280, + "summary": "In 2025, Alphabet entered agreements to acquire Wiz for $32.0 billion and Intersect for $4.8 billion, both expected to close in 2026. It issued $37.3 billion in senior notes. Other income and expense included $24.1 billion in net gains on equity securities. Other Bets had a $7.5 billion operating loss including a $2.1 billion Waymo compensation charge. U.S. tax law changes allowed immediate expensing of R&D costs. Share repurchases totaled $45.4 billion. Operating cash flow was $164.7 billion, capital expenditures $91.4 billion, and employee count 190,820. The company noted potential impacts from international trade volatility.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "In", + "2025", + "entered", + "definitive", + "agreements", + "acquire", + "Wiz", + "leading", + "cloud", + "security", + "platform", + "32.0", + "billion", + "Intersect", + "provider", + "data", + "center", + "energy", + "infrastructure", + "solutions", + "4.8", + "cash", + "assumption", + "debt", + "Both", + "acquisitions", + "expected", + "close", + "2026", + "subject", + "customary", + "closing", + "conditions", + "including", + "receipt", + "regulatory", + "approvals", + "issued", + "senior", + "unsecured", + "notes", + "net", + "proceeds", + "37.3", + "general", + "corporate", + "purposes", + "OI", + "29.8", + "year", + "ended", + "December", + "31", + "included", + "gains", + "equity", + "securities", + "24.1", + "primarily", + "related", + "unrealized", + "marketable", + "Other", + "Bets", + "operating", + "loss", + "7.5", + "2.1", + "employee", + "compensation", + "charge", + "recognized", + "fourth", + "quarter", + "Waymo", + "reflected", + "research", + "development", + "expenses", + "based", + "estimated", + "stock", + "valuation", + "February", + "announced", + "investment", + "round", + "16.0", + "significant", + "majority", + "funded", + "Alphabet", + "Changes", + "tax", + "law", + "enacted", + "July", + "things", + "expensing", + "domestic", + "experimentation", + "costs", + "accelerated", + "depreciation", + "eligible", + "capital", + "expenditures", + "effects", + "flows", + "Repurchases", + "Class", + "shares", + "6.5", + "38.9", + "totaling", + "45.4", + "Operating", + "flow", + "164.7", + "Capital", + "investments", + "technical", + "91.4", + "As", + "190", + "820", + "employees", + "We", + "monitoring", + "ongoing", + "developments", + "surrounding", + "international", + "trade", + "macroeconomic", + "environment", + "result", + "volatility", + "financial", + "markets", + "experience", + "direct", + "indirect", + "business", + "operations", + "results", + "Our", + "past", + "indicative", + "future", + "performance", + "differ", + "materially", + "historical", + "trends" + ], + "keywords": [ + "acquisitions", + "financial results", + "tax law" + ], + "connect_to": [] + } + }, + { + "chunk_id": "047f28cf-731c-5e8f-ad68-f5104050b011", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0f78488e-2596-5d22-ab3c-7f51ebc11d1f", + "type": "table", + "content": "
Year Ended December 31,
20242025
Google Search & other$ 198,084$ 224,532
YouTube ads36,14740,367
Google Network30,35929,792
Google advertising264,590294,691
Google subscriptions, platforms, and devices40,34048,030
Google Services total304,930342,721
Google Cloud43,22958,705
Other Bets1,6481,537
Hedging gains (losses)211(127)
Total revenues$ 350,018$ 402,836
", + "path": "tables/table-6 Revenue Breakdown.html", + "metadata": { + "length": 805, + "summary": "table-7\nThe table shows Google's revenue breakdown for 2024 and 2025, with total revenues rising from $350,018 to $402,836, driven by growth in Search, YouTube, and Cloud.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-6 Revenue Breakdown.html", + "keywords": [ + "Google", + "Revenue", + "2024-2025" + ], + "tokens": [] + } + }, + { + "chunk_id": "1a95e78c-c850-5db1-84ea-1245afaf0dea", + "type": "text", + "content": "The following table presents revenues by type (in millions):\n\n[tables/table-6 Revenue Breakdown.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues", + "metadata": { + "length": 107, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "presents", + "revenues", + "type", + "millions" + ], + "keywords": [], + "connect_to": [ + { + "target": "0f78488e-2596-5d22-ab3c-7f51ebc11d1f", + "relation": "embeds", + "ref": "[tables/table-6 Revenue Breakdown.html]", + "position": { + "start": 62, + "end": 101 + } + } + ] + } + }, + { + "chunk_id": "720ac780-b2e9-5147-a0c1-bc5ca477ab8c", + "type": "text", + "content": "Google Advertising\nGoogle Search & other\nGoogle Search & other revenues increased \\$26.4 billion from 2024 to 2025. The overall growth was driven by interrelated factors including increases in search queries resulting from growth in user adoption and usage on mobile devices; growth in advertiser spending; and improvements we have made in ad formats and delivery.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Google Services", + "metadata": { + "length": 364, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "Advertising", + "Search", + "revenues", + "increased", + "26.4", + "billion", + "2024", + "2025", + "The", + "growth", + "driven", + "interrelated", + "factors", + "including", + "increases", + "search", + "queries", + "resulting", + "user", + "adoption", + "usage", + "mobile", + "devices", + "advertiser", + "spending", + "improvements", + "made", + "ad", + "formats", + "delivery" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "27539173-e19a-5dbc-a024-baaa41e4660e", + "type": "text", + "content": "YouTube ads revenues increased \\$4.2 billion from 2024 to 2025. The growth was driven by our direct response advertising products followed by our brand advertising products, both of which benefited from increased spending by our advertisers.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Google Services/YouTube ads", + "metadata": { + "length": 241, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "YouTube", + "ads", + "revenues", + "increased", + "4.2", + "billion", + "2024", + "2025", + "The", + "growth", + "driven", + "direct", + "response", + "advertising", + "products", + "brand", + "benefited", + "spending", + "advertisers" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "349b4f87-a8a7-512f-82ed-6eefb63c1933", + "type": "text", + "content": "Google Network revenues decreased \\$567 million from 2024 to 2025, primarily due to a decrease in AdSense revenues, partially offset by an increase in AdMob revenues.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Google Services/Google Network", + "metadata": { + "length": 166, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "Network", + "revenues", + "decreased", + "567", + "million", + "2024", + "2025", + "primarily", + "due", + "decrease", + "AdSense", + "partially", + "offset", + "increase", + "AdMob" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cec58956-34bb-5e45-8711-1c3d6e52246c", + "type": "table", + "content": "
Google Search & other
Paid clicks change6 %
Cost-per-click change7 %
Google Network
Impressions change(7)%
Cost-per-impression change7 %
", + "path": "tables/table-7 Google Ad Changes.html", + "metadata": { + "length": 318, + "summary": "table-8\nGoogle Search shows 6% paid clicks and 7% cost-per-click increases. Google Network shows 7% impressions decrease and 7% cost-per-impression increase.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-7 Google Ad Changes.html", + "keywords": [ + "Paid clicks", + "Cost-per-click", + "Impressions" + ], + "tokens": [] + } + }, + { + "chunk_id": "92489b58-8f66-5152-9d39-45fecf671dd5", + "type": "text", + "content": "The following table presents changes in monetization metrics for Google Search & other revenues (paid clicks and cost-per-click) and Google Network revenues (impressions and cost-per-impression), expressed as a percentage, from 2024 to 2025:\n\n[tables/table-7 Google Ad Changes.html]\n\nChanges in paid clicks and impressions are driven by a number of interrelated factors, including changes in advertiser spending; ongoing product and policy changes; and, as it relates to paid clicks, fluctuations in search queries resulting from changes in user adoption and usage, primarily on mobile devices.\nChanges in cost-per-click and cost-per-impression are driven by a number of interrelated factors including changes in device mix, geographic mix, advertiser spending, ongoing product and policy changes, product mix, property mix, and changes in foreign currency exchange rates.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Monetization Metrics", + "metadata": { + "length": 878, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "presents", + "monetization", + "metrics", + "Google", + "Search", + "revenues", + "paid", + "clicks", + "cost", + "click", + "Network", + "impressions", + "impression", + "expressed", + "percentage", + "2024", + "2025", + "Changes", + "driven", + "number", + "interrelated", + "factors", + "including", + "advertiser", + "spending", + "ongoing", + "product", + "policy", + "relates", + "fluctuations", + "search", + "queries", + "resulting", + "user", + "adoption", + "usage", + "primarily", + "mobile", + "devices", + "device", + "mix", + "geographic", + "property", + "foreign", + "currency", + "exchange", + "rates" + ], + "keywords": [], + "connect_to": [ + { + "target": "cec58956-34bb-5e45-8711-1c3d6e52246c", + "relation": "embeds", + "ref": "[tables/table-7 Google Ad Changes.html]", + "position": { + "start": 243, + "end": 282 + } + } + ] + } + }, + { + "chunk_id": "1b9b2905-e35b-5a61-9ee7-7ea7a3f2d7c7", + "type": "text", + "content": "Google subscriptions, platforms, and devices revenues increased \\$7.7 billion from 2024 to 2025. The growth was primarily driven by an increase in subscriptions revenues. This increase was primarily due to the contribution from growth in paid subscriptions across both YouTube services and Google One.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Google Subscriptions, Platforms, and Devices", + "metadata": { + "length": 301, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "subscriptions", + "platforms", + "devices", + "revenues", + "increased", + "7.7", + "billion", + "2024", + "2025", + "The", + "growth", + "primarily", + "driven", + "increase", + "This", + "due", + "contribution", + "paid", + "YouTube", + "services", + "One" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cfde9343-8ac8-514f-8ec7-5740d601e56d", + "type": "text", + "content": "Google Cloud revenues increased \\$15.5 billion from 2024 to 2025, primarily driven by growth in Google Cloud Platform largely from infrastructure and platform services.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Google Cloud", + "metadata": { + "length": 168, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "Cloud", + "revenues", + "increased", + "15.5", + "billion", + "2024", + "2025", + "primarily", + "driven", + "growth", + "Platform", + "largely", + "infrastructure", + "platform", + "services" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1e770cf0-3189-5a4c-9564-f2340d2e1b76", + "type": "table", + "content": "
Year Ended December 31,
20242025
United States49 %48 %
$EMEA^{(1)}$ 29 %29 %
$APAC^{(1)}$ 16 %17 %
$Other Americas^{(1)}$ 6 %6 %
Hedging gains (losses)0 %0 %
", + "path": "tables/table-8 Revenue by Region.html", + "metadata": { + "length": 427, + "summary": "table-9\nThe table shows revenue percentages by region for years 2024 and 2025: United States 49% to 48%, EMEA 29% both years, APAC 16% to 17%, Other Americas 6% both years, hedging gains 0%.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-8 Revenue by Region.html", + "keywords": [ + "Revenue", + "Region", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "4cc424b5-d065-5b93-b3ee-c5a2a3c2d1c3", + "type": "text", + "content": "The following table presents revenues by geography as a percentage of revenues, determined based on the addresses of our customers:\n\n[tables/table-8 Revenue by Region.html]\n\n(1) Regions represent Europe, the Middle East, and Africa (EMEA); Asia-Pacific (APAC); and Canada and Latin America (\"Other Americas\").\nFor additional information, see Note 2 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Revenues by Geography", + "metadata": { + "length": 459, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "presents", + "revenues", + "geography", + "percentage", + "determined", + "based", + "addresses", + "customers", + "Regions", + "represent", + "Europe", + "Middle", + "East", + "Africa", + "EMEA", + "Asia", + "Pacific", + "APAC", + "Canada", + "Latin", + "America", + "Other", + "Americas", + "For", + "additional", + "information", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [ + { + "target": "1e770cf0-3189-5a4c-9564-f2340d2e1b76", + "relation": "embeds", + "ref": "[tables/table-8 Revenue by Region.html]", + "position": { + "start": 133, + "end": 172 + } + } + ] + } + }, + { + "chunk_id": "5ea4d928-4b72-5fe5-8cbe-e4a4ad6558a0", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses_2", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1bb2b228-d785-5419-82f0-a07aea247efe", + "type": "table", + "content": "
Year Ended December 31,
20242025
TAC$ 54,900$ 59,926
Other cost of revenues91,406102,609
Total cost of revenues$ 146,306$ 162,535
Total cost of revenues as a percentage of revenues42 %40 %
", + "path": "tables/table-9 Cost of Revenues.html", + "metadata": { + "length": 417, + "summary": "table-10\nThe table shows TAC, other cost of revenues, and total cost of revenues for 2024 and 2025, with total cost as a percentage of revenues decreasing from 42% to 40%.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-9 Cost of Revenues.html", + "keywords": [ + "TAC", + "cost of revenues", + "percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "28b8bcf2-932d-5722-b272-47d1ff9ecc01", + "type": "text", + "content": "The following table presents cost of revenues, including TAC (in millions, except percentages):\n\n[tables/table-9 Cost of Revenues.html]\n\nCost of revenues increased \\$16.2 billion from 2024 to 2025 due to an increase in other cost of revenues and TAC of \\$11.2 billion and \\$5.0 billion, respectively.\nThe increase in TAC from 2024 to 2025 was largely due to an increase in TAC paid to distribution partners, primarily driven by growth in revenues subject to TAC. The TAC rate decreased from 20.7% to 20.3% from 2024 to 2025, primarily due to a revenue mix shift from Google Network properties to Google Search & other properties. The TAC rates on Google Search & other and Google Network revenues were substantially consistent from 2024 to 2025.\nThe increase in other cost of revenues from 2024 to 2025 was primarily due to increases in content acquisition costs, largely for YouTube, depreciation expense, and other technical infrastructure operations costs.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses_2/Cost of Revenues", + "metadata": { + "length": 967, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "presents", + "cost", + "revenues", + "including", + "TAC", + "millions", + "percentages", + "Cost", + "increased", + "16.2", + "billion", + "2024", + "2025", + "due", + "increase", + "11.2", + "5.0", + "largely", + "paid", + "distribution", + "partners", + "primarily", + "driven", + "growth", + "subject", + "rate", + "decreased", + "20.7%", + "20.3%", + "revenue", + "mix", + "shift", + "Google", + "Network", + "properties", + "Search", + "rates", + "substantially", + "consistent", + "increases", + "content", + "acquisition", + "costs", + "YouTube", + "depreciation", + "expense", + "technical", + "infrastructure", + "operations" + ], + "keywords": [], + "connect_to": [ + { + "target": "1bb2b228-d785-5419-82f0-a07aea247efe", + "relation": "embeds", + "ref": "[tables/table-9 Cost of Revenues.html]", + "position": { + "start": 97, + "end": 135 + } + } + ] + } + }, + { + "chunk_id": "8536888b-5925-55e0-9c1b-963c8a082365", + "type": "table", + "content": "
Year Ended December 31,
20242025
Research and development expenses$ 49,326$ 61,087
Research and development expenses as a percentage of revenues14 %15 %
", + "path": "tables/table-10 R&D Expenses.html", + "metadata": { + "length": 311, + "summary": "table-11\nResearch and development expenses increased from $49,326 in 2024 to $61,087 in 2025, rising from 14% to 15% of revenues.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-10 R&D Expenses.html", + "keywords": [ + "R&D", + "Revenue", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "e0805667-883c-58b1-8a19-8d9bb1c3955d", + "type": "text", + "content": "The following table presents research and development expenses (in millions, except percentages):\n\n[tables/table-10 R&D Expenses.html]\n\nResearch and development expenses increased \\$11.8 billion from 2024 to 2025, primarily driven by increases in employee compensation expenses of \\$6.9 billion and depreciation expense of \\$2.4 billion. The increase in employee compensation expenses was primarily driven by an increase in SBC expenses of \\$4.2 billion, which included an increase in a valuation-based compensation charge related to Waymo.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses_2/Research and Development", + "metadata": { + "length": 551, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "presents", + "research", + "development", + "expenses", + "millions", + "percentages", + "Research", + "increased", + "11.8", + "billion", + "2024", + "2025", + "primarily", + "driven", + "increases", + "employee", + "compensation", + "6.9", + "depreciation", + "expense", + "2.4", + "increase", + "SBC", + "4.2", + "included", + "valuation", + "based", + "charge", + "related", + "Waymo" + ], + "keywords": [], + "connect_to": [ + { + "target": "8536888b-5925-55e0-9c1b-963c8a082365", + "relation": "embeds", + "ref": "[tables/table-10 R&D Expenses.html]", + "position": { + "start": 99, + "end": 134 + } + } + ] + } + }, + { + "chunk_id": "f5b85065-17e6-543e-b18a-8fcae12fa9a4", + "type": "table", + "content": "
Year Ended December 31,
20242025
Sales and marketing expenses$ 27,808$ 28,693
Sales and marketing expenses as a percentage of revenues8 %7 %
", + "path": "tables/table-11 Sales Expenses.html", + "metadata": { + "length": 299, + "summary": "table-12\nSales and marketing expenses were $27,808 in 2024 and $28,693 in 2025, with percentages of revenue at 8% and 7% respectively.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-11 Sales Expenses.html", + "keywords": [ + "Sales expenses", + "Revenue percentage", + "2024-2025" + ], + "tokens": [] + } + }, + { + "chunk_id": "2201e9e3-8747-5cf2-8663-ddc7df327f94", + "type": "text", + "content": "The following table presents sales and marketing expenses (in millions, except percentages):\n\n[tables/table-11 Sales Expenses.html]\n\nSales and marketing expenses increased \\$885 million from 2024 to 2025, primarily driven by an increase in advertising and promotional activities of \\$1.2 billion, partially offset by a decrease in employee compensation expenses of \\$214 million.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses_2/Sales and Marketing", + "metadata": { + "length": 388, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "presents", + "sales", + "marketing", + "expenses", + "millions", + "percentages", + "Sales", + "increased", + "885", + "million", + "2024", + "2025", + "primarily", + "driven", + "increase", + "advertising", + "promotional", + "activities", + "1.2", + "billion", + "partially", + "offset", + "decrease", + "employee", + "compensation", + "214" + ], + "keywords": [], + "connect_to": [ + { + "target": "f5b85065-17e6-543e-b18a-8fcae12fa9a4", + "relation": "embeds", + "ref": "[tables/table-11 Sales Expenses.html]", + "position": { + "start": 94, + "end": 131 + } + } + ] + } + }, + { + "chunk_id": "f3ece2c5-3ab2-5ae1-8f3c-0200ce067736", + "type": "table", + "content": "
Year Ended December 31,
20242025
General and administrative expenses$ 14,188$ 21,482
General and administrative expenses as a percentage of revenues4 %5 %
", + "path": "tables/table-12 Expense Analysis.html", + "metadata": { + "length": 313, + "summary": "table-13\nGeneral and administrative expenses rose from $14,188 in 2024 to $21,482 in 2025, with the percentage of revenues increasing from 4% to 5%.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-12 Expense Analysis.html", + "keywords": [ + "General expenses", + "Revenue percentage", + "2024-2025" + ], + "tokens": [] + } + }, + { + "chunk_id": "1ace6bbf-e843-5452-bb35-9b295c92a474", + "type": "text", + "content": "The following table presents general and administrative expenses (in millions, except percentages):\n\n[tables/table-12 Expense Analysis.html]\n\nGeneral and administrative expenses increased \\$7.3 billion from 2024 to 2025, primarily driven by an increase in expenses related to legal and other matters of \\$6.2 billion, largely the result of the \\$3.5 billion EC fine accrued in the third quarter of 2025 and a \\$1.4 billion legal accrual made in the second quarter of 2025.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses_2/General and Administrative", + "metadata": { + "length": 479, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "presents", + "general", + "administrative", + "expenses", + "millions", + "percentages", + "General", + "increased", + "7.3", + "billion", + "2024", + "2025", + "primarily", + "driven", + "increase", + "related", + "legal", + "matters", + "6.2", + "largely", + "result", + "3.5", + "EC", + "fine", + "accrued", + "quarter", + "1.4", + "accrual", + "made" + ], + "keywords": [], + "connect_to": [ + { + "target": "f3ece2c5-3ab2-5ae1-8f3c-0200ce067736", + "relation": "embeds", + "ref": "[tables/table-12 Expense Analysis.html]", + "position": { + "start": 101, + "end": 140 + } + } + ] + } + }, + { + "chunk_id": "6f6395c6-f065-5073-be84-c1fa7b7304b4", + "type": "table", + "content": "
Year Ended December 31,
20242025
Operating income (loss):
Google Services$ 121,263$ 139,404
Google Cloud6,11213,910
Other Bets(4,444)(7,515)
Alphabet-level activities(1)(10,541)(16,760)
Total income from operations$ 112,390$ 129,039
", + "path": "tables/table-13 Alphabet Income.html", + "metadata": { + "length": 534, + "summary": "table-14\nThe table shows Alphabet's operating income for 2024 and 2025. Google Services earned $121,263M and $139,404M; Google Cloud $6,112M and $13,910M; total income from operations was $112,390M and $129,039M.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-13 Alphabet Income.html", + "keywords": [ + "Operating Income", + "Google Services", + "Google Cloud" + ], + "tokens": [] + } + }, + { + "chunk_id": "a0e1aed0-6c9c-5929-8964-803a30682080", + "type": "text", + "content": "We report our segment results as Google Services, Google Cloud, and Other Bets. Additionally, certain costs are not allocated to our segments because they represent Alphabet-level activities. For further details on our segments, see Part I, Item 1 Business and Note 15 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.\nThe following table presents segment operating income (loss) (in millions).\n\n[tables/table-13 Alphabet Income.html]\n\n(1) Alphabet-level activities primarily reflect expenses related to our shared AI research and development.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability", + "metadata": { + "length": 606, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "report", + "segment", + "results", + "Google", + "Services", + "Cloud", + "Other", + "Bets", + "Additionally", + "costs", + "allocated", + "segments", + "represent", + "Alphabet", + "level", + "activities", + "For", + "details", + "Part", + "Item", + "Business", + "Note", + "15", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Annual", + "Report", + "Form", + "10", + "The", + "table", + "presents", + "operating", + "income", + "loss", + "millions", + "primarily", + "reflect", + "expenses", + "related", + "shared", + "AI", + "research", + "development" + ], + "keywords": [], + "connect_to": [ + { + "target": "6f6395c6-f065-5073-be84-c1fa7b7304b4", + "relation": "embeds", + "ref": "[tables/table-13 Alphabet Income.html]", + "position": { + "start": 451, + "end": 489 + } + } + ] + } + }, + { + "chunk_id": "3f89770a-c436-53ff-ba7e-ac002394a7d3", + "type": "text", + "content": "Google Services operating income increased \\$18.1 billion from 2024 to 2025. The increase in operating income was primarily driven by an increase in revenues, partially offset by an increase in expenses related to legal and other matters, TAC, and content acquisition costs.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability/Google Services", + "metadata": { + "length": 274, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "Services", + "operating", + "income", + "increased", + "18.1", + "billion", + "2024", + "2025", + "The", + "increase", + "primarily", + "driven", + "revenues", + "partially", + "offset", + "expenses", + "related", + "legal", + "matters", + "TAC", + "content", + "acquisition", + "costs" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "648bf6e2-b7a1-5628-87e0-af2411543d68", + "type": "text", + "content": "Google Cloud operating income increased \\$7.8 billion from 2024 to 2025. The increase in operating income was primarily driven by an increase in revenues, partially offset by increases in usage costs for technical infrastructure and employee compensation expenses.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability/Google Cloud", + "metadata": { + "length": 264, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "Cloud", + "operating", + "income", + "increased", + "7.8", + "billion", + "2024", + "2025", + "The", + "increase", + "primarily", + "driven", + "revenues", + "partially", + "offset", + "increases", + "usage", + "costs", + "technical", + "infrastructure", + "employee", + "compensation", + "expenses" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2a145b46-2807-5620-a778-37e6ceb3d2ca", + "type": "text", + "content": "Other Bets operating loss increased \\$3.1 billion from 2024 to 2025. The increase in operating loss was primarily driven by an increase in employee compensation expenses largely due to an increase in a valuation-based compensation charge related to Waymo.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability/Other Bets", + "metadata": { + "length": 255, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Other", + "Bets", + "operating", + "loss", + "increased", + "3.1", + "billion", + "2024", + "2025", + "The", + "increase", + "primarily", + "driven", + "employee", + "compensation", + "expenses", + "largely", + "due", + "valuation", + "based", + "charge", + "related", + "Waymo" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4c0578e1-0f98-5df8-b3b8-2672b79407bf", + "type": "table", + "content": "
Year Ended December 31,
20242025
Interest income$ 4,482$ 4,337
Interest expense(268)(736)
Foreign currency exchange gain (loss), net(409)(382)
Gain (loss) on debt securities, net(1,043)540
Gain (loss) on equity securities, net3,71424,080
Income (loss) and impairment from equity method investments, net(188)281
Other1,1371,667
Other income (expense), net$ 7,425$ 29,787
", + "path": "tables/table-14 Other Income_Expense.html", + "metadata": { + "length": 741, + "summary": "table-15\nThe table shows other income/expense for years ended Dec 31, 2024 and 2025. Total net income increased from $7,425 to $29,787, driven by gains on equity securities.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-14 Other Income_Expense.html", + "keywords": [ + "Interest", + "Securities", + "Foreign Exchange" + ], + "tokens": [] + } + }, + { + "chunk_id": "e16837dc-82c9-57e4-b121-5d738323f2a9", + "type": "text", + "content": "The following table presents OI&E, (in millions):\n\n[tables/table-14 Other Income_Expense.html]\n\nOI&E, net increased \\$22.4 billion from 2024 to 2025, primarily due to increases in net unrealized gains on equity securities resulting from fair value adjustments on non-marketable equity securities.\nFor additional information, see Note 3 and Note 7 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability/Other Income (Expense), Net", + "metadata": { + "length": 454, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "presents", + "OI", + "millions", + "net", + "increased", + "22.4", + "billion", + "2024", + "2025", + "primarily", + "due", + "increases", + "unrealized", + "gains", + "equity", + "securities", + "resulting", + "fair", + "adjustments", + "marketable", + "For", + "additional", + "information", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [ + { + "target": "4c0578e1-0f98-5df8-b3b8-2672b79407bf", + "relation": "embeds", + "ref": "[tables/table-14 Other Income_Expense.html]", + "position": { + "start": 51, + "end": 94 + } + } + ] + } + }, + { + "chunk_id": "d65abff2-aed5-522b-b7bf-fab4e01d9039", + "type": "table", + "content": "
Year Ended December 31,
20242025
Income before provision for income taxes$ 119,815$ 158,826
Provision for income taxes$ 19,697$ 26,656
Effective tax rate16.4 %16.8 %
", + "path": "tables/table-15 Tax Data Table.html", + "metadata": { + "length": 359, + "summary": "table-16\nThe table shows income before taxes, provision for income taxes, and effective tax rates for years 2024 and 2025.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-15 Tax Data Table.html", + "keywords": [ + "income", + "tax", + "rate" + ], + "tokens": [] + } + }, + { + "chunk_id": "4a9c2bb8-8fea-56e0-b45a-3ae23e1d2656", + "type": "text", + "content": "The following table presents provision for income taxes (in millions, except effective tax rate):\n\n[tables/table-15 Tax Data Table.html]\n\nThe effective tax rate increased from 2024 to 2025. This increase was primarily due to a decrease in the US Federal Foreign Derived Intangible Income tax deduction, a non-deductible EC fine and legal settlement in the US, partially offset by changes in prior period tax positions.\nChanges to US tax law enacted on July 4, 2025, allow for immediate expensing of domestic research and experimentation costs, accelerated depreciation on eligible capital expenditures, and other tax law changes impacting 2025 with certain changes effective in 2026. These changes are reflected in our results for the year ended December 31, 2025.\nThe OECD is coordinating negotiations among more than 140 countries with the goal of achieving consensus around substantial changes to international tax policies, including the implementation of a minimum global effective tax rate of 15%. Some countries have already implemented the legislation effective January 1, 2024. This did not have a material effect on our income tax provision for the 2025 fiscal year.\nIn January 2026, the OECD introduced new guidance including a \"Side-by-Side Safe Harbor\" which, if elected, exempts U.S. domestic operations from being taxed by global minimum tax rules. However, it does not exempt foreign subsidiaries from local minimum tax requirements if implemented. As more countries enact these global minimum tax rules, our effective tax rate and cash tax payments could increase.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability/Provision for Income Taxes", + "metadata": { + "length": 1590, + "summary": "The effective tax rate increased from 2024 to 2025 due to a decrease in the US Foreign Derived Intangible Income deduction and a non-deductible EC fine, partially offset by prior period tax adjustments. US tax law changes enacted July 4, 2025, allow immediate expensing of R&D and accelerated depreciation, impacting 2025 with some changes effective in 2026. OECD global minimum tax rules (15% rate) did not materially affect 2025 provision. New OECD guidance in January 2026 introduces a safe harbor exempting US domestic operations from global minimum tax, but foreign subsidiaries may still be subject to local minimum taxes, potentially increasing future effective tax rates and cash taxes.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "presents", + "provision", + "income", + "taxes", + "millions", + "effective", + "tax", + "rate", + "increased", + "2024", + "2025", + "This", + "increase", + "primarily", + "due", + "decrease", + "US", + "Federal", + "Foreign", + "Derived", + "Intangible", + "Income", + "deduction", + "deductible", + "EC", + "fine", + "legal", + "settlement", + "partially", + "offset", + "prior", + "period", + "positions", + "Changes", + "law", + "enacted", + "July", + "expensing", + "domestic", + "research", + "experimentation", + "costs", + "accelerated", + "depreciation", + "eligible", + "capital", + "expenditures", + "impacting", + "2026", + "These", + "reflected", + "results", + "year", + "ended", + "December", + "31", + "OECD", + "coordinating", + "negotiations", + "140", + "countries", + "goal", + "achieving", + "consensus", + "substantial", + "international", + "policies", + "including", + "implementation", + "minimum", + "global", + "15%", + "Some", + "implemented", + "legislation", + "January", + "material", + "effect", + "fiscal", + "In", + "introduced", + "guidance", + "Side", + "Safe", + "Harbor", + "elected", + "exempts", + "operations", + "taxed", + "rules", + "However", + "exempt", + "foreign", + "subsidiaries", + "local", + "requirements", + "As", + "enact", + "cash", + "payments" + ], + "keywords": [ + "effective tax rate", + "US tax law", + "global minimum tax" + ], + "connect_to": [ + { + "target": "d65abff2-aed5-522b-b7bf-fab4e01d9039", + "relation": "embeds", + "ref": "[tables/table-15 Tax Data Table.html]", + "position": { + "start": 99, + "end": 136 + } + } + ] + } + }, + { + "chunk_id": "16b881e6-1bb7-5942-90a9-4c9c2a2f9f6d", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "913cba9c-1b8c-5a7a-8e2c-a6f012f0902e", + "type": "text", + "content": "As of December 31, 2025, we had \\$126.8 billion in cash, cash equivalents, and short-term marketable securities. Cash equivalents and marketable securities are comprised of time deposits, money market funds, highly liquid government bonds, corporate debt securities, mortgage-backed and asset-backed securities, and marketable equity securities.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Cash, Cash Equivalents, and Marketable Securities", + "metadata": { + "length": 345, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "As", + "December", + "31", + "2025", + "126.8", + "billion", + "cash", + "equivalents", + "short", + "term", + "marketable", + "securities", + "Cash", + "comprised", + "time", + "deposits", + "money", + "market", + "funds", + "highly", + "liquid", + "government", + "bonds", + "corporate", + "debt", + "mortgage", + "backed", + "asset", + "equity" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "af950cbb-fc50-5d6d-8a06-45b7ea21a51e", + "type": "table", + "content": "
Year Ended December 31,
20242025
Net cash provided by operating activities$ 125,299$ 164,713
Net cash used in investing activities$ (45,536)$ (120,291)
Net cash used in financing activities$ (79,733)$ (37,388)
", + "path": "tables/table-16 Cash Flow Table.html", + "metadata": { + "length": 403, + "summary": "table-17\nThe table shows net cash from operating activities increased from $125,299 in 2024 to $164,713 in 2025, while investing activities used more cash in 2025 ($120,291) than 2024 ($45,536), and financing activities used less cash in 2025 ($37,388) than 2024 ($79,733).", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-16 Cash Flow Table.html", + "keywords": [ + "cash flow", + "operating", + "investing" + ], + "tokens": [] + } + }, + { + "chunk_id": "854be2e8-66e8-5e0b-941e-cd2b21019e1d", + "type": "text", + "content": "Our principal sources of liquidity are cash, cash equivalents, and marketable securities, as well as the cash flow that we generate from operations. The primary use of capital continues to be to invest for the long-term growth of the business. We regularly evaluate our cash and capital structure, including the size, pace, and form of capital return to stockholders.\nThe following table presents cash flows (in millions):\n\n[tables/table-16 Cash Flow Table.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Sources, Uses of Cash and Related Trends", + "metadata": { + "length": 470, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "principal", + "sources", + "liquidity", + "cash", + "equivalents", + "marketable", + "securities", + "flow", + "generate", + "operations", + "The", + "primary", + "capital", + "continues", + "invest", + "long", + "term", + "growth", + "business", + "We", + "regularly", + "evaluate", + "structure", + "including", + "size", + "pace", + "form", + "return", + "stockholders", + "table", + "presents", + "flows", + "millions" + ], + "keywords": [], + "connect_to": [ + { + "target": "af950cbb-fc50-5d6d-8a06-45b7ea21a51e", + "relation": "embeds", + "ref": "[tables/table-16 Cash Flow Table.html]", + "position": { + "start": 424, + "end": 462 + } + } + ] + } + }, + { + "chunk_id": "4b442b8d-9c82-5f5b-9eaa-699aa61e5a88", + "type": "text", + "content": "Our largest source of cash provided by operations are advertising revenues generated by Google Search & other properties, YouTube properties, and Google Network properties. In Google Services, we also generate cash through consumer subscriptions, the sale of apps and in-app purchases, and devices. In Google Cloud, we generate cash through consumption-based fees and subscriptions for infrastructure, platform, applications, and other cloud services.\nOur primary uses of cash from operating activities include payments to distribution and Google Network partners, to employees for compensation, and to content providers. Other uses of cash from operating activities include payments to suppliers for devices, to tax authorities for income taxes, and other general corporate expenditures.\nNet cash provided by operating activities increased from 2024 to 2025 due to an increase in cash received from customers, partially offset by an increase in cash payments for cost of revenues and operating expenses.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Sources, Uses of Cash and Related Trends/Cash Provided by Operating Activities", + "metadata": { + "length": 1004, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "largest", + "source", + "cash", + "provided", + "operations", + "advertising", + "revenues", + "generated", + "Google", + "Search", + "properties", + "YouTube", + "Network", + "In", + "Services", + "generate", + "consumer", + "subscriptions", + "sale", + "apps", + "app", + "purchases", + "devices", + "Cloud", + "consumption", + "based", + "fees", + "infrastructure", + "platform", + "applications", + "cloud", + "services", + "primary", + "operating", + "activities", + "include", + "payments", + "distribution", + "partners", + "employees", + "compensation", + "content", + "providers", + "Other", + "suppliers", + "tax", + "authorities", + "income", + "taxes", + "general", + "corporate", + "expenditures", + "Net", + "increased", + "2024", + "2025", + "due", + "increase", + "received", + "customers", + "partially", + "offset", + "cost", + "expenses" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1d71c96c-1eec-535b-a89e-407001afdd49", + "type": "text", + "content": "Cash provided by investing activities consists primarily of maturities and sales of investments in marketable and non-marketable securities. Cash used in investing activities consists primarily of purchases of marketable and non-marketable securities, purchases of property and equipment, and payments for acquisitions.\nNet cash used in investing activities increased from 2024 to 2025, primarily due to an increase in purchases of property and equipment, driven by investments in technical infrastructure, and a decrease in maturities and sales of marketable securities.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Sources, Uses of Cash and Related Trends/Cash Used in Investing Activities", + "metadata": { + "length": 571, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Cash", + "provided", + "investing", + "activities", + "consists", + "primarily", + "maturities", + "sales", + "investments", + "marketable", + "securities", + "purchases", + "property", + "equipment", + "payments", + "acquisitions", + "Net", + "cash", + "increased", + "2024", + "2025", + "due", + "increase", + "driven", + "technical", + "infrastructure", + "decrease" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1b8f912c-7669-5ab6-8b9b-197b58c7947b", + "type": "text", + "content": "Cash provided by financing activities consists primarily of proceeds from issuance of debt and proceeds from the sale of interests in consolidated entities. Cash used in financing activities consists primarily of repurchases of stock, repayments of debt, net payments related to stock-based award activities, and dividend payments.\nNet cash used in financing activities decreased from 2024 to 2025 due to an increase in proceeds from issuance of debt and a decrease in repurchases of stock, partially offset by repayments of debt.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Sources, Uses of Cash and Related Trends/Cash Used in Financing Activities", + "metadata": { + "length": 530, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Cash", + "provided", + "financing", + "activities", + "consists", + "primarily", + "proceeds", + "issuance", + "debt", + "sale", + "interests", + "consolidated", + "entities", + "repurchases", + "stock", + "repayments", + "net", + "payments", + "related", + "based", + "award", + "dividend", + "Net", + "cash", + "decreased", + "2024", + "2025", + "due", + "increase", + "decrease", + "partially", + "offset" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fe230205-045b-5afc-84e6-52b9faaf2ef6", + "type": "text", + "content": "We expect existing cash, cash equivalents, short-term marketable securities, and cash flows from operations and financing activities to continue to be sufficient to fund our operating activities and cash commitments for investing and financing activities for at least the next 12 months, and thereafter for the foreseeable future.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Liquidity and Material Cash Requirements", + "metadata": { + "length": 330, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "expect", + "existing", + "cash", + "equivalents", + "short", + "term", + "marketable", + "securities", + "flows", + "operations", + "financing", + "activities", + "continue", + "sufficient", + "fund", + "operating", + "commitments", + "investing", + "12", + "months", + "foreseeable", + "future" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "76257d19-4010-516f-9887-f460e1e5214f", + "type": "text", + "content": "We make investments in land, buildings, and servers and network equipment through purchases of property and equipment and lease arrangements to provide capacity for the growth of our services and products.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases", + "metadata": { + "length": 205, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "make", + "investments", + "land", + "buildings", + "servers", + "network", + "equipment", + "purchases", + "property", + "lease", + "arrangements", + "provide", + "capacity", + "growth", + "services", + "products" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6d68f18b-f5f9-5744-b991-8e0e069d83ab", + "type": "text", + "content": "Our capital investments in property and equipment consist primarily of the following major categories:\n- technical infrastructure, which consists of our investments in servers and network equipment, data center land, and building construction and improvements; and\n• office facilities, ground-up development projects, and building improvements.\nAssets not yet in service are those that are not ready for their intended use, including assets in the process of construction or assembly, and consist primarily of technical infrastructure. The time frame from date of purchase to placement in service of these assets may extend from months to years. For example, our data center construction projects are generally multi-year projects with multiple phases, where we acquire land and buildings, construct buildings, and secure and install servers and network equipment.\nDuring the years ended December 31, 2024 and 2025, we spent \\$52.5 billion and \\$91.4 billion on capital expenditures, respectively. In 2026, we expect to significantly increase, relative to 2025, our investment in our technical infrastructure, including servers and network equipment, and data centers. Depreciation of our property and equipment commences when such assets are ready for their intended use. For the years ended December 31, 2024 and 2025, our depreciation on property and equipment was \\$15.3 billion and \\$21.1 billion, respectively.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Capital Expenditures", + "metadata": { + "length": 1416, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "capital", + "investments", + "property", + "equipment", + "consist", + "primarily", + "major", + "categories", + "technical", + "infrastructure", + "consists", + "servers", + "network", + "data", + "center", + "land", + "building", + "construction", + "improvements", + "office", + "facilities", + "ground", + "development", + "projects", + "Assets", + "service", + "ready", + "intended", + "including", + "assets", + "process", + "assembly", + "The", + "time", + "frame", + "date", + "purchase", + "placement", + "extend", + "months", + "years", + "For", + "generally", + "multi", + "year", + "multiple", + "phases", + "acquire", + "buildings", + "construct", + "secure", + "install", + "During", + "ended", + "December", + "31", + "2024", + "2025", + "spent", + "52.5", + "billion", + "91.4", + "expenditures", + "In", + "2026", + "expect", + "significantly", + "increase", + "relative", + "investment", + "centers", + "Depreciation", + "commences", + "depreciation", + "15.3", + "21.1" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "61bc9068-867a-5779-9c9a-7642dd9ddeb9", + "type": "text", + "content": "As of December 31, 2025, the amount of total undiscounted future lease payments under operating leases was \\$18.3 billion, of which \\$3.3 billion is short-term, and total undiscounted future lease payments under finance leases was \\$2.9 billion, of which \\$491 million is short-term.\nAs of December 31, 2025, we have entered into leases primarily related to data centers that have not yet commenced with short-term and long-term future lease payments of \\$5.8 billion and \\$52.7 billion, respectively. These leases will commence between 2026 and 2031 with non-cancelable lease terms primarily between one and 25 years.\nIn January 2026, we executed a power purchase agreement which we expect to be accounted for as a lease resulting in future payments depending on certain agreement terms of \\$9.9 billion between 2027 and 2047. If certain contractual conditions for the project are not met, we would instead make a one-time payment of approximately \\$3.5 billion and assume ownership of the power generating assets.\nFor additional information on leases, see Note 4 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Leases", + "metadata": { + "length": 1169, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "As", + "December", + "31", + "2025", + "amount", + "total", + "undiscounted", + "future", + "lease", + "payments", + "operating", + "leases", + "18.3", + "billion", + "3.3", + "short", + "term", + "finance", + "2.9", + "491", + "million", + "entered", + "primarily", + "related", + "data", + "centers", + "commenced", + "long", + "5.8", + "52.7", + "These", + "commence", + "2026", + "2031", + "cancelable", + "terms", + "25", + "years", + "In", + "January", + "executed", + "power", + "purchase", + "agreement", + "expect", + "accounted", + "resulting", + "depending", + "9.9", + "2027", + "2047", + "If", + "contractual", + "conditions", + "project", + "met", + "make", + "time", + "payment", + "approximately", + "3.5", + "assume", + "ownership", + "generating", + "assets", + "For", + "additional", + "information", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "292f223d-30c0-572c-9a38-513e3929f349", + "type": "text", + "content": "As of December 31, 2025, we had senior unsecured notes outstanding with a total carrying value of \\$48.5 billion, of which \\$2.0 billion was short-term. The associated short-term and long-term future interest payments were \\$1.8 billion and \\$35.7 billion, respectively.\nDuring 2025, we issued \\$22.5 billion of US dollar-denominated senior unsecured notes and €13.25 billion of euro-denominated senior unsecured notes for general corporate purposes, comprised of the following:\n- May 2025: We issued \\$5.0 billion of US dollar-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 4.89%, and a weighted-average maturity of approximately 24 years. We also issued €6.75 billion of euro-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 3.31%, and a weighted-average maturity of approximately 14 years.\n- November 2025: We issued \\$500 million of US dollar-denominated floating-rate senior unsecured notes and \\$17.0 billion of US dollar-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 4.92% and a weighted-average maturity of approximately 20 years. We also issued €6.5 billion of euro-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 3.44% and a weighted-average maturity of approximately 16 years.\nAs of December 31, 2025, we had \\$10.0 billion of revolving credit facilities, \\$4.0 billion expiring in April 2026 and \\$6.0 billion expiring in April 2030. No amounts have been borrowed under the credit facilities. We also have a commercial paper program of up to \\$25.0 billion, which is used for general corporate purposes. As of December 31, 2025, we had no commercial paper outstanding.\nFor additional information, see Note 6 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.\nWe use contract manufacturers for our technical infrastructure and device assembly and may supply them with components purchased directly from suppliers. Certain of these arrangements result in a portion of the cash received from and paid to contract manufacturers to be presented as financing activities on the Consolidated Statements of Cash Flows included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Financing", + "metadata": { + "length": 2279, + "summary": "As of December 31, 2025, the company had $48.5 billion in senior unsecured notes outstanding, with $2.0 billion short-term. Future interest payments were $1.8 billion short-term and $35.7 billion long-term. In 2025, it issued $22.5 billion in USD and €13.25 billion in EUR senior unsecured notes, including fixed and floating-rate notes with varying maturities and coupon rates. The company also had $10.0 billion in revolving credit facilities (expiring 2026 and 2030) with no borrowings, and a $25.0 billion commercial paper program with no outstanding balance. Additional details are in Note 6 of the financial statements.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "As", + "December", + "31", + "2025", + "senior", + "unsecured", + "notes", + "outstanding", + "total", + "carrying", + "48.5", + "billion", + "2.0", + "short", + "term", + "The", + "long", + "future", + "interest", + "payments", + "1.8", + "35.7", + "During", + "issued", + "22.5", + "US", + "dollar", + "denominated", + "13.25", + "euro", + "general", + "corporate", + "purposes", + "comprised", + "May", + "We", + "5.0", + "fixed", + "rate", + "weighted", + "average", + "coupon", + "4.89%", + "maturity", + "approximately", + "24", + "years", + "6.75", + "3.31%", + "14", + "November", + "500", + "million", + "floating", + "17.0", + "4.92%", + "20", + "6.5", + "3.44%", + "16", + "10.0", + "revolving", + "credit", + "facilities", + "4.0", + "expiring", + "April", + "2026", + "6.0", + "2030", + "No", + "amounts", + "borrowed", + "commercial", + "paper", + "program", + "25.0", + "For", + "additional", + "information", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10", + "contract", + "manufacturers", + "technical", + "infrastructure", + "device", + "assembly", + "supply", + "components", + "purchased", + "directly", + "suppliers", + "Certain", + "arrangements", + "result", + "portion", + "cash", + "received", + "paid", + "presented", + "financing", + "activities", + "Cash", + "Flows" + ], + "keywords": [ + "senior notes", + "credit facilities", + "commercial paper" + ], + "connect_to": [] + } + }, + { + "chunk_id": "86470a22-c774-52e3-96a0-5b38bfafa7cd", + "type": "text", + "content": "During 2025, we repurchased and subsequently retired 240 million shares for \\$45.4 billion.\nIn April 2024, the company's Board of Directors authorized a \\$70.0 billion share repurchase program for its Class A and Class C shares. In April 2025, the company's Board of Directors authorized an additional \\$70.0 billion share repurchase program for its Class A and Class C shares. As of December 31, 2025, \\$69.5 billion remained available for Class A and Class C share repurchases.\nFor additional information, see Note 11 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Share Repurchase Program", + "metadata": { + "length": 624, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "During", + "2025", + "repurchased", + "subsequently", + "retired", + "240", + "million", + "shares", + "45.4", + "billion", + "In", + "April", + "2024", + "company", + "Board", + "Directors", + "authorized", + "70.0", + "share", + "repurchase", + "program", + "Class", + "additional", + "As", + "December", + "31", + "69.5", + "remained", + "repurchases", + "For", + "information", + "Note", + "11", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0ad5d51d-0a0c-5bd9-a523-aad31f89a6ff", + "type": "text", + "content": "During the year ended December 31, 2025, total cash dividends were \\$4.8 billion for Class A, \\$703 million for Class B, and \\$4.5 billion for Class C shares, respectively.\nIn April 2025, the company's Board of Directors increased the quarterly cash dividend by 5% to \\$0.21 per share of outstanding Class A, Class B, and Class C shares.\nThe company has declared a quarterly cash dividend in the current quarter, and intends to pay quarterly cash dividends in the future, subject to review and approval by the company's Board of Directors in its sole discretion.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Dividend Program", + "metadata": { + "length": 562, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "During", + "year", + "ended", + "December", + "31", + "2025", + "total", + "cash", + "dividends", + "4.8", + "billion", + "Class", + "703", + "million", + "4.5", + "shares", + "In", + "April", + "company", + "Board", + "Directors", + "increased", + "quarterly", + "dividend", + "5%", + "0.21", + "share", + "outstanding", + "The", + "declared", + "current", + "quarter", + "intends", + "pay", + "future", + "subject", + "review", + "approval", + "sole", + "discretion" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8f0f45c5-c1f9-5953-9eea-e9055d958577", + "type": "text", + "content": "As of December 31, 2025, we had short-term accrued legal and regulatory fines and settlements of \\$15.6 billion. This amount primarily included EC fines, in addition to accruals related to other legal matters and regulatory fines and settlements. For additional information, see Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Accrued Legal and Regulatory", + "metadata": { + "length": 391, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "As", + "December", + "31", + "2025", + "short", + "term", + "accrued", + "legal", + "regulatory", + "fines", + "settlements", + "15.6", + "billion", + "This", + "amount", + "primarily", + "included", + "EC", + "addition", + "accruals", + "related", + "matters", + "For", + "additional", + "information", + "Note", + "10", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Item", + "Annual", + "Report", + "Form" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "64860e99-6e70-576d-832b-9cf4d1d67a2b", + "type": "text", + "content": "As of December 31, 2025, we had long-term income taxes payable of \\$9.5 billion primarily related to unrecognized tax benefits. The timing and amount of any payment related to these unrecognized tax benefits are uncertain and cannot be estimated.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Taxes", + "metadata": { + "length": 246, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "As", + "December", + "31", + "2025", + "long", + "term", + "income", + "taxes", + "payable", + "9.5", + "billion", + "primarily", + "related", + "unrecognized", + "tax", + "benefits", + "The", + "timing", + "amount", + "payment", + "uncertain", + "estimated" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "bb288ebf-330a-5175-8e00-84d51fb990a9", + "type": "text", + "content": "We have material purchase commitments and other contractual obligations primarily related to energy take-or-pay contracts, licenses (including content licenses), and technical infrastructure and inventory orders. As of December 31, 2025, the total for these commitments was \\$149.1 billion, of which \\$113.0 billion was short-term, mostly related to technical infrastructure and inventory orders. These amounts reflect commitments and obligations through open purchase orders as well as the non-cancelable portion or the minimum cancellation fee in certain agreements. For those agreements with variable terms, we do not estimate the non-cancelable obligation beyond any minimum quantities and/or pricing as of December 31, 2025. In certain instances, the amount of our contractual obligations may change based on the expected timing of order fulfillment from our suppliers. For additional information related to our content licenses, see Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.\nAs of December 31, 2025, we provided backstops in the form of financial guarantees and credit derivatives with maximum potential amount of future payments of \\$5.7 billion and \\$16.9 billion, respectively. For additional information on credit derivatives and financial guarantees, see Note 3 and Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.\nIn addition, we regularly enter into multi-year, non-cancellable power purchase agreements with third-party suppliers that do not include a minimum dollar commitment. The amounts to be paid under these agreements are based on the actual volumes to be generated and are not readily determinable.\nWe may experience increases in the costs associated with our purchase commitments and other contractual obligations as a result of ongoing developments surrounding international trade. For details on risks related to our manufacturing and supply chain and other risks, refer to Part 1, Item 1A, \"Risk Factors\" of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Purchase Commitments and Other Contractual Obligations", + "metadata": { + "length": 2101, + "summary": "As of December 31, 2025, the company had material purchase commitments and contractual obligations totaling $149.1 billion, with $113.0 billion short-term, mainly for technical infrastructure and inventory orders. These include energy take-or-pay contracts, licenses, and inventory orders, reflecting non-cancelable portions or minimum cancellation fees. Additionally, the company provided backstops via financial guarantees ($5.7 billion) and credit derivatives ($16.9 billion). Multi-year power purchase agreements without minimum dollar commitments are also entered into, with payments based on actual volumes. Risks from international trade developments may increase costs.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "material", + "purchase", + "commitments", + "contractual", + "obligations", + "primarily", + "related", + "energy", + "pay", + "contracts", + "licenses", + "including", + "content", + "technical", + "infrastructure", + "inventory", + "orders", + "As", + "December", + "31", + "2025", + "total", + "149.1", + "billion", + "113.0", + "short", + "term", + "These", + "amounts", + "reflect", + "open", + "cancelable", + "portion", + "minimum", + "cancellation", + "fee", + "agreements", + "For", + "variable", + "terms", + "estimate", + "obligation", + "quantities", + "pricing", + "In", + "instances", + "amount", + "change", + "based", + "expected", + "timing", + "order", + "fulfillment", + "suppliers", + "additional", + "information", + "Note", + "10", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "provided", + "backstops", + "form", + "financial", + "guarantees", + "credit", + "derivatives", + "maximum", + "potential", + "future", + "payments", + "5.7", + "16.9", + "addition", + "regularly", + "enter", + "multi", + "year", + "cancellable", + "power", + "party", + "include", + "dollar", + "commitment", + "The", + "paid", + "actual", + "volumes", + "generated", + "readily", + "determinable", + "experience", + "increases", + "costs", + "result", + "ongoing", + "developments", + "surrounding", + "international", + "trade", + "details", + "risks", + "manufacturing", + "supply", + "chain", + "refer", + "Part", + "1A", + "Risk", + "Factors" + ], + "keywords": [ + "purchase commitments", + "financial guarantees", + "contractual obligations" + ], + "connect_to": [] + } + }, + { + "chunk_id": "07300c9a-f552-5afb-8e43-8748d1a772c0", + "type": "text", + "content": "In March 2025, we entered into a definitive agreement to acquire Wiz, Inc. (\"Wiz\"), a leading cloud security platform, for \\$32.0 billion, subject to closing adjustments, in an all-cash transaction. The acquisition of Wiz is expected to close in 2026, subject to customary closing conditions, including the receipt of regulatory approvals.\nIn December 2025, we entered into a definitive agreement to acquire Intersect, which provides data center and energy infrastructure solutions, for \\$4.8 billion in cash, plus the assumption of debt. The acquisition of Intersect is expected to close in the first half of 2026, subject to customary closing conditions.\nFor additional information, see Note 8 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Pending Acquisitions", + "metadata": { + "length": 800, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "In", + "March", + "2025", + "entered", + "definitive", + "agreement", + "acquire", + "Wiz", + "Inc", + "leading", + "cloud", + "security", + "platform", + "32.0", + "billion", + "subject", + "closing", + "adjustments", + "cash", + "transaction", + "The", + "acquisition", + "expected", + "close", + "2026", + "customary", + "conditions", + "including", + "receipt", + "regulatory", + "approvals", + "December", + "Intersect", + "data", + "center", + "energy", + "infrastructure", + "solutions", + "4.8", + "assumption", + "debt", + "half", + "For", + "additional", + "information", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4f4ef56d-2101-53f1-be67-ae4434c34234", + "type": "text", + "content": "We prepare our consolidated financial statements in accordance with GAAP. In doing so, we have to make estimates and assumptions. Our critical accounting estimates are those estimates that involve a significant level of uncertainty at the time the estimate was made, and changes in them have had or are reasonably likely to have a material effect on our financial condition or results of operations. Accordingly, actual results could differ materially from our estimates. We base our estimates on past experience and other assumptions that we believe are reasonable under the circumstances, and we evaluate these estimates on an ongoing basis. We have reviewed our critical accounting estimates with the Audit Committee of our Board of Directors.\nFor a summary of significant accounting policies and the effect on our financial statements, see Note 1 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Critical Accounting Estimates", + "metadata": { + "length": 955, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "prepare", + "consolidated", + "financial", + "statements", + "accordance", + "GAAP", + "In", + "make", + "estimates", + "assumptions", + "Our", + "critical", + "accounting", + "involve", + "significant", + "level", + "uncertainty", + "time", + "estimate", + "made", + "material", + "effect", + "condition", + "results", + "operations", + "Accordingly", + "actual", + "differ", + "materially", + "base", + "past", + "experience", + "reasonable", + "circumstances", + "evaluate", + "ongoing", + "basis", + "reviewed", + "Audit", + "Committee", + "Board", + "Directors", + "For", + "summary", + "policies", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "738c806e-6488-593c-88b1-26923da02e79", + "type": "text", + "content": "We measure certain financial instruments at fair value on a nonrecurring basis, consisting primarily of our non-marketable equity securities. These investments are accounted for under the measurement alternative method (\"the measurement alternative\") and are measured at cost, less impairment, subject to upward and downward adjustments resulting from observable price changes for identical or similar investments of the same issuer. These adjustments require quantitative assessments of the fair value of our securities, which may require the use of unobservable inputs.\nAdjustments are determined primarily based on a market approach as of the transaction date and involve the use of estimates using the best information available, which may include cash flow projections or other available market data.\nNon-marketable equity securities are also evaluated for impairment, based on qualitative factors including the companies' financial and liquidity position and access to capital resources, among others. When indicators of impairment exist, we prepare quantitative measurements of the fair value of our equity investments using a market approach or an income approach, which requires judgment and the use of unobservable inputs, including discount rates, investee revenues and costs, and comparable market data of private and public companies, among others. When the quantitative remeasurements of fair value indicate an impairment exists, we write down the investment to its current fair value.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Critical Accounting Estimates/Fair Value Measurements of Non-Marketable Equity Securities", + "metadata": { + "length": 1499, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "measure", + "financial", + "instruments", + "fair", + "nonrecurring", + "basis", + "consisting", + "primarily", + "marketable", + "equity", + "securities", + "These", + "investments", + "accounted", + "measurement", + "alternative", + "method", + "measured", + "cost", + "impairment", + "subject", + "upward", + "downward", + "adjustments", + "resulting", + "observable", + "price", + "identical", + "similar", + "issuer", + "require", + "quantitative", + "assessments", + "unobservable", + "inputs", + "Adjustments", + "determined", + "based", + "market", + "approach", + "transaction", + "date", + "involve", + "estimates", + "information", + "include", + "cash", + "flow", + "projections", + "data", + "Non", + "evaluated", + "qualitative", + "factors", + "including", + "companies", + "liquidity", + "position", + "access", + "capital", + "resources", + "When", + "indicators", + "exist", + "prepare", + "measurements", + "income", + "requires", + "judgment", + "discount", + "rates", + "investee", + "revenues", + "costs", + "comparable", + "private", + "public", + "remeasurements", + "exists", + "write", + "investment", + "current" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "033ba763-000f-5cc0-a5ac-231cc4a8de45", + "type": "text", + "content": "We assess the reasonableness of the useful lives of our property and equipment periodically or when events indicate a change is necessary. To determine the useful lives of our technical infrastructure, we rely on multiple inputs, including historical asset performance, expected technology advancements, and our future infrastructure deployment plans. Any change in the estimated useful lives is recognized on a prospective basis.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Critical Accounting Estimates/Property and Equipment", + "metadata": { + "length": 430, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "assess", + "reasonableness", + "lives", + "property", + "equipment", + "periodically", + "events", + "change", + "To", + "determine", + "technical", + "infrastructure", + "rely", + "multiple", + "inputs", + "including", + "historical", + "asset", + "performance", + "expected", + "technology", + "advancements", + "future", + "deployment", + "plans", + "Any", + "estimated", + "recognized", + "prospective", + "basis" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4b9b0246-c96e-5723-849e-848557fb14d4", + "type": "text", + "content": "We are subject to income taxes in the US and foreign jurisdictions. Significant judgment is required in evaluating our uncertain tax positions and determining our provision for income taxes.\nRecording an uncertain tax position involves various qualitative considerations, including evaluation of comparable and resolved tax exposures, applicability of tax laws, and likelihood of settlement. We evaluate uncertain tax positions periodically, considering changes in facts and circumstances, such as new regulations or recent judicial opinions, as well as the status of audit activities by taxing authorities. Although we believe we have adequately reserved for our uncertain tax positions, no assurance can be given that the final tax outcome of these matters will not be different. To the extent that the final tax outcome of these matters is different than the amounts recorded, such differences will affect the provision for income taxes and the effective tax rate in the period in which such determination is made.\nThe provision for income taxes includes the effect of reserve provisions and changes to reserves as well as the related net interest and penalties. In addition, we are subject to the continuous examination of our income tax returns by the Internal Revenue Service (IRS) and other tax authorities which may assert assessments against us. We regularly assess the likelihood of adverse outcomes resulting from these examinations and assessments to determine the adequacy of our provision for income taxes.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Critical Accounting Estimates/Income Taxes", + "metadata": { + "length": 1520, + "summary": "The passage discusses the company's exposure to income taxes in the US and foreign jurisdictions, emphasizing the need for significant judgment in evaluating uncertain tax positions and determining the provision for income taxes. It describes qualitative considerations such as comparable tax exposures, tax laws, and settlement likelihood, and notes periodic evaluation of these positions based on new regulations, judicial opinions, and audit activities. The provision includes reserves, changes to reserves, and related net interest and penalties. The company is subject to continuous examination by tax authorities, and it regularly assesses the likelihood of adverse outcomes to ensure adequate provisioning.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "subject", + "income", + "taxes", + "US", + "foreign", + "jurisdictions", + "Significant", + "judgment", + "required", + "evaluating", + "uncertain", + "tax", + "positions", + "determining", + "provision", + "Recording", + "position", + "involves", + "qualitative", + "considerations", + "including", + "evaluation", + "comparable", + "resolved", + "exposures", + "applicability", + "laws", + "likelihood", + "settlement", + "evaluate", + "periodically", + "facts", + "circumstances", + "regulations", + "recent", + "judicial", + "opinions", + "status", + "audit", + "activities", + "taxing", + "authorities", + "Although", + "adequately", + "reserved", + "assurance", + "final", + "outcome", + "matters", + "To", + "extent", + "amounts", + "recorded", + "differences", + "affect", + "effective", + "rate", + "period", + "determination", + "made", + "The", + "includes", + "effect", + "reserve", + "provisions", + "reserves", + "related", + "net", + "interest", + "penalties", + "In", + "addition", + "continuous", + "examination", + "returns", + "Internal", + "Revenue", + "Service", + "IRS", + "assert", + "assessments", + "regularly", + "assess", + "adverse", + "outcomes", + "resulting", + "examinations", + "determine", + "adequacy" + ], + "keywords": [ + "income taxes", + "uncertain tax positions", + "tax reserves" + ], + "connect_to": [] + } + }, + { + "chunk_id": "b5a840da-e34e-5021-ae6e-c3bc1637469d", + "type": "text", + "content": "We are subject to claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders involving competition, intellectual property, data privacy and security, tax and related compliance, labor and employment, commercial disputes, content generated by our users, goods and services offered by advertisers or publishers using our platforms, personal injury, consumer protection, and other matters. Certain of these matters include speculative claims for substantial or indeterminate amounts of damages. We record a liability when we believe that it is probable that a loss has been incurred and the amount can be reasonably estimated. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the possible loss in Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.\nWe evaluate, on a regular basis, developments in our legal matters that could affect the amount of liability that has been previously accrued, and the matters and related reasonably possible losses disclosed, and make adjustments as necessary. Significant judgment is required to determine both the likelihood and the estimated amount of a loss related to such matters. Until the final resolution of such matters, there may be an exposure to loss in excess of the amount recorded, and such amounts could be material.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Critical Accounting Estimates/Loss Contingencies", + "metadata": { + "length": 1438, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "subject", + "claims", + "lawsuits", + "regulatory", + "government", + "inquiries", + "investigations", + "proceedings", + "consent", + "orders", + "involving", + "competition", + "intellectual", + "property", + "data", + "privacy", + "security", + "tax", + "related", + "compliance", + "labor", + "employment", + "commercial", + "disputes", + "content", + "generated", + "users", + "goods", + "services", + "offered", + "advertisers", + "publishers", + "platforms", + "personal", + "injury", + "consumer", + "protection", + "matters", + "Certain", + "include", + "speculative", + "substantial", + "indeterminate", + "amounts", + "damages", + "record", + "liability", + "probable", + "loss", + "incurred", + "amount", + "estimated", + "If", + "determine", + "range", + "disclose", + "Note", + "10", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "evaluate", + "regular", + "basis", + "developments", + "legal", + "affect", + "previously", + "accrued", + "losses", + "disclosed", + "make", + "adjustments", + "Significant", + "judgment", + "required", + "likelihood", + "Until", + "final", + "resolution", + "exposure", + "excess", + "recorded", + "material" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0be73f7a-bd8c-53dc-9ab2-7a4f59883226", + "type": "text", + "content": "We are exposed to financial market risks, including changes in foreign currency exchange rates, interest rates, and equity investment risks.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 7A. QUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK", + "metadata": { + "length": 140, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "exposed", + "financial", + "market", + "risks", + "including", + "foreign", + "currency", + "exchange", + "rates", + "interest", + "equity", + "investment" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "38e8b2c6-900e-5609-8bbf-b9152f8a839d", + "type": "table", + "content": "
ImpactAs of December 31,
20242025
Foreign currency risk
Foreign denominated monetary assets and liabilities(1)OI&E$ 135$ 671
Cash flow hedges of foreign currency revenue(2)AOCI$ 1,627$ 2,096
Net investment hedges of investments in foreign subsidiaries(3)AOCI$ 660$ 2,942
", + "path": "tables/table-17 Foreign Currency Risk.html", + "metadata": { + "length": 563, + "summary": "table-18\nThe table shows foreign currency risk impacts on OI&E and AOCI for 2024 and 2025, with monetary assets/liabilities, cash flow hedges, and net investment hedges.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-17 Foreign Currency Risk.html", + "keywords": [ + "Foreign currency risk", + "hedges", + "monetary assets" + ], + "tokens": [] + } + }, + { + "chunk_id": "7b1c523f-7a2e-5cff-9efb-f671df41fe9f", + "type": "text", + "content": "We transact business globally in multiple currencies. International revenues, foreign-denominated monetary assets and liabilities, and investments in foreign subsidiaries expose us to the risk of fluctuations in foreign exchange rates against the US dollar. Principal currency exposures include the Australian dollar, British pound, Canadian dollar, Euro, and Japanese yen.\nWe monitor our foreign currency exposures and hedge foreign exchange risks with derivative and non-derivative instruments, including forwards, options (including collars), cross-currency swaps, and foreign currency-denominated debt. Gains or losses on these foreign currency exposures are generally offset by corresponding gains or losses on the derivative and non-derivative instruments.\nConsidering historical trends in foreign exchange rates, we determined that it was reasonably possible that adverse changes in exchange rates of 10% could be experienced. We performed a sensitivity analysis on our foreign currency exposures to estimate the potential impact of this adverse 10% change. The estimated effects on our financial position would be as follows (in millions):\n\n[tables/table-17 Foreign Currency Risk.html]\n\n(1) After consideration of the effect of derivative contracts.\n(2) The change in accumulated other comprehensive income (AOCI) would be expected to offset a corresponding foreign currency change in forecasted hedged revenues when recognized.\n(3) The change in AOCI would be expected to offset a corresponding foreign currency translation gain or loss from our investments in foreign subsidiaries.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 7A. QUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK/Foreign Currency Exchange Risk", + "metadata": { + "length": 1595, + "summary": "The company transacts globally in multiple currencies, exposing it to foreign exchange rate fluctuations against the US dollar, with principal exposures including the Australian dollar, British pound, Canadian dollar, Euro, and Japanese yen. It monitors and hedges these risks using derivatives and non-derivatives such as forwards, options, cross-currency swaps, and foreign currency-denominated debt. A sensitivity analysis assuming a 10% adverse change in exchange rates estimates the potential impact on financial position, with offsets from derivative contracts and changes in accumulated other comprehensive income (AOCI) for forecasted hedged revenues and foreign subsidiary investments.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "transact", + "business", + "globally", + "multiple", + "currencies", + "International", + "revenues", + "foreign", + "denominated", + "monetary", + "assets", + "liabilities", + "investments", + "subsidiaries", + "expose", + "risk", + "fluctuations", + "exchange", + "rates", + "US", + "dollar", + "Principal", + "currency", + "exposures", + "include", + "Australian", + "British", + "pound", + "Canadian", + "Euro", + "Japanese", + "yen", + "monitor", + "hedge", + "risks", + "derivative", + "instruments", + "including", + "forwards", + "options", + "collars", + "cross", + "swaps", + "debt", + "Gains", + "losses", + "generally", + "offset", + "gains", + "Considering", + "historical", + "trends", + "determined", + "adverse", + "10%", + "experienced", + "performed", + "sensitivity", + "analysis", + "estimate", + "potential", + "impact", + "change", + "The", + "estimated", + "effects", + "financial", + "position", + "millions", + "After", + "consideration", + "effect", + "contracts", + "accumulated", + "comprehensive", + "income", + "AOCI", + "expected", + "forecasted", + "hedged", + "recognized", + "translation", + "gain", + "loss" + ], + "keywords": [ + "foreign exchange", + "hedging", + "sensitivity analysis" + ], + "connect_to": [ + { + "target": "38e8b2c6-900e-5609-8bbf-b9152f8a839d", + "relation": "embeds", + "ref": "[tables/table-17 Foreign Currency Risk.html]", + "position": { + "start": 1149, + "end": 1193 + } + } + ] + } + }, + { + "chunk_id": "fe1ebedb-d41a-57da-9cf8-1c6e369d4f18", + "type": "table", + "content": "
As of December 31,12-Month Average As of December 31,
2024202520242025
Risk category - interest rate$ 208$ 162$ 230$ 184
", + "path": "tables/table-18 Interest Rate Risk.html", + "metadata": { + "length": 304, + "summary": "table-19\nThe table shows interest rate risk values for 2024 and 2025, with point-in-time figures of $208 and $162, and 12-month averages of $230 and $184, respectively.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-18 Interest Rate Risk.html", + "keywords": [ + "interest rate", + "risk", + "financial" + ], + "tokens": [] + } + }, + { + "chunk_id": "2da6c060-26c4-51e9-8477-18c97df647b0", + "type": "text", + "content": "We are exposed to interest rate risk related to our investment portfolio and outstanding debt.\nOur Corporate Treasury investment strategy is to achieve a return that will allow us to preserve capital and maintain liquidity. By policy, we limit the amount of credit exposure within our investment portfolio to any one issuer. Our investments in both fixed rate and floating rate interest earning securities carry a degree of interest rate risk. Fixed rate securities may have their fair market value adversely affected due to a rise in interest rates, while floating rate securities may produce less income than predicted if interest rates fall. Unrealized gains and losses on our marketable debt securities are primarily due to interest rate fluctuations as compared to interest rates at the time of purchase. For certain fixed and floating rate debt securities, we have elected the fair value option for which changes in fair value are recorded in OI&E. We measure securities for which we have not elected the fair value option at fair value with gains and losses recorded in AOCI until the securities are sold, less any expected credit losses.\nWe use value-at-risk (VaR) analysis to determine the potential effect of fluctuations in interest rates on the value of our investment portfolio. The VaR is the expected loss in fair value, for a given confidence interval, for our investment portfolio due to adverse movements in interest rates. We use a variance/covariance VaR model with 95% confidence interval. The estimated one-day loss in fair value of our investment portfolio as of December 31, 2024 and 2025 are shown below (in millions):\n\n[tables/table-18 Interest Rate Risk.html]\n\nActual future gains and losses associated with our investment portfolio may differ materially from the sensitivity analyses performed as of December 31, 2024 and 2025 due to the inherent limitations associated with predicting the timing and amount of changes in interest rates and our actual exposures and positions. VaR analysis is not intended to represent actual losses but is used as a risk estimation.\nAdditionally, we had senior unsecured notes outstanding with a total carrying value of \\$11.9 billion and \\$48.5 billion as of December 31, 2024 and 2025, respectively. As our senior unsecured notes primarily bear interest at fixed rates and are recorded at amortized cost, interest rate fluctuations generally do not affect our consolidated financial statements. However, the fair value of the notes will fluctuate with movement in market interest rates.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 7A. QUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK/Interest Rate Risk", + "metadata": { + "length": 2574, + "summary": "The passage discusses interest rate risk related to the company's investment portfolio and outstanding debt. The Corporate Treasury strategy aims to preserve capital and maintain liquidity. Fixed and floating rate securities carry interest rate risk, with fair value changes recorded in OI&E or AOCI. Value-at-risk (VaR) analysis with a 95% confidence interval estimates potential daily losses, with figures for 2024 and 2025 shown in a table. Actual results may differ due to limitations. Additionally, senior unsecured notes with carrying values of $11.9 billion (2024) and $48.5 billion (2025) bear fixed rates and are recorded at amortized cost, so interest rate fluctuations do not affect financial statements but impact fair value.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "exposed", + "interest", + "rate", + "risk", + "related", + "investment", + "portfolio", + "outstanding", + "debt", + "Our", + "Corporate", + "Treasury", + "strategy", + "achieve", + "return", + "preserve", + "capital", + "maintain", + "liquidity", + "By", + "policy", + "limit", + "amount", + "credit", + "exposure", + "issuer", + "investments", + "fixed", + "floating", + "earning", + "securities", + "carry", + "degree", + "Fixed", + "fair", + "market", + "adversely", + "affected", + "due", + "rise", + "rates", + "produce", + "income", + "predicted", + "fall", + "Unrealized", + "gains", + "losses", + "marketable", + "primarily", + "fluctuations", + "compared", + "time", + "purchase", + "For", + "elected", + "option", + "recorded", + "OI", + "measure", + "AOCI", + "sold", + "expected", + "VaR", + "analysis", + "determine", + "potential", + "effect", + "The", + "loss", + "confidence", + "interval", + "adverse", + "movements", + "variance", + "covariance", + "model", + "95%", + "estimated", + "day", + "December", + "31", + "2024", + "2025", + "shown", + "millions", + "Actual", + "future", + "differ", + "materially", + "sensitivity", + "analyses", + "performed", + "inherent", + "limitations", + "predicting", + "timing", + "actual", + "exposures", + "positions", + "intended", + "represent", + "estimation", + "Additionally", + "senior", + "unsecured", + "notes", + "total", + "carrying", + "11.9", + "billion", + "48.5", + "As", + "bear", + "amortized", + "cost", + "generally", + "affect", + "consolidated", + "financial", + "statements", + "However", + "fluctuate", + "movement" + ], + "keywords": [ + "Interest rate risk", + "VaR analysis", + "Investment portfolio" + ], + "connect_to": [ + { + "target": "fe1ebedb-d41a-57da-9cf8-1c6e369d4f18", + "relation": "embeds", + "ref": "[tables/table-18 Interest Rate Risk.html]", + "position": { + "start": 1645, + "end": 1686 + } + } + ] + } + }, + { + "chunk_id": "93db7b0d-b3f0-5405-a27b-2b21c5492879", + "type": "text", + "content": "Our marketable and non-marketable equity securities are subject to a wide variety of market-related risks that could substantially reduce or increase the fair value of our holdings.\nOur marketable equity securities are primarily publicly traded stocks or funds and our non-marketable equity securities are primarily investments in privately held companies, some of which are in the startup or development stages.\nWe record marketable equity securities at fair value subject to market price volatility. These securities represent \\$5.1 billion and \\$6.3 billion of our investments as of December 31, 2024 and 2025, respectively. A hypothetical adverse price change of 10% on our December 31, 2025 balance would decrease the fair value of marketable equity securities by \\$631 million. From time to time, we may enter into derivatives to hedge the market price risk on certain of our marketable equity securities.\nOur non-marketable equity securities not accounted for under the equity method are primarily adjusted to fair value for observable transactions for identical or similar investments of the same issuer or impairment (referred to as the measurement alternative). The fair value measured at the time of the observable transaction is not necessarily an indication of the current fair value as of the balance sheet date. These investments, especially those that are in the early stages, are inherently risky because the technologies or products these companies have under development are typically in the early phases and may never materialize, and they may experience a decline in financial condition, which could result in a loss of a substantial part of our investment in these companies. Valuations of our equity investments in private companies are inherently more complex due to the lack of readily available market data and observable transactions at lower valuations could result in significant losses. In addition, global economic conditions could result in additional volatility. The success of our investment in any private company is also typically dependent on the likelihood of our ability to realize appreciation in the value of investments through liquidity events such as public offerings, acquisitions, private sales, or other market events. Changes in the valuation of non-marketable equity securities may not directly correlate with changes in valuation of marketable equity securities. As of December 31, 2024 and 2025, the carrying value of our non-marketable equity securities, which were accounted for under the measurement alternative, was \\$35.2 billion and \\$64.1 billion, respectively.\nThe carrying values of our equity method investments, which totaled approximately \\$2.0 billion and \\$2.5 billion as of December 31, 2024 and 2025, respectively, generally do not fluctuate based on market price changes. However, these investments could be impaired if the carrying value exceeds the fair value and is not expected to recover.\nFor additional information about our equity investments, see Note 1 and Note 3 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 7A. QUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK/Equity Investment Risk", + "metadata": { + "length": 3145, + "summary": "The passage discusses marketable and non-marketable equity securities, their risks, and valuation methods. Marketable equity securities, primarily publicly traded stocks, are recorded at fair value with market price volatility, totaling $6.3 billion as of December 31, 2025, with a hypothetical 10% adverse price change decreasing fair value by $631 million. Non-marketable equity securities, mainly investments in private companies, are adjusted to fair value based on observable transactions or impairment, with carrying values of $64.1 billion as of December 31, 2025. These investments are inherently risky due to early-stage technologies and lack of market data. Equity method investments, totaling $2.5 billion as of December 31, 2025, generally do not fluctuate with market prices but may be impaired. The passage highlights the complexity and volatility of these investments.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "marketable", + "equity", + "securities", + "subject", + "wide", + "variety", + "market", + "related", + "risks", + "substantially", + "reduce", + "increase", + "fair", + "holdings", + "primarily", + "publicly", + "traded", + "stocks", + "funds", + "investments", + "privately", + "held", + "companies", + "startup", + "development", + "stages", + "We", + "record", + "price", + "volatility", + "These", + "represent", + "5.1", + "billion", + "6.3", + "December", + "31", + "2024", + "2025", + "hypothetical", + "adverse", + "change", + "10%", + "balance", + "decrease", + "631", + "million", + "From", + "time", + "enter", + "derivatives", + "hedge", + "risk", + "accounted", + "method", + "adjusted", + "observable", + "transactions", + "identical", + "similar", + "issuer", + "impairment", + "referred", + "measurement", + "alternative", + "The", + "measured", + "transaction", + "necessarily", + "indication", + "current", + "sheet", + "date", + "early", + "inherently", + "risky", + "technologies", + "products", + "typically", + "phases", + "materialize", + "experience", + "decline", + "financial", + "condition", + "result", + "loss", + "substantial", + "part", + "investment", + "Valuations", + "private", + "complex", + "due", + "lack", + "readily", + "data", + "lower", + "valuations", + "significant", + "losses", + "In", + "addition", + "global", + "economic", + "conditions", + "additional", + "success", + "company", + "dependent", + "likelihood", + "ability", + "realize", + "appreciation", + "liquidity", + "events", + "public", + "offerings", + "acquisitions", + "sales", + "Changes", + "valuation", + "directly", + "correlate", + "As", + "carrying", + "35.2", + "64.1", + "values", + "totaled", + "approximately", + "2.0", + "2.5", + "generally", + "fluctuate", + "based", + "However", + "impaired", + "exceeds", + "expected", + "recover", + "For", + "information", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Item", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [ + "marketable", + "non-marketable", + "fair value" + ], + "connect_to": [] + } + }, + { + "chunk_id": "3bb9d496-c8de-5b50-a537-0204d9e0f225", + "type": "text", + "content": "Alphabet Inc.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA", + "metadata": { + "length": 13, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Alphabet", + "Inc" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6c8cb0fd-ab33-575d-b4ed-85db334a7fc8", + "type": "table", + "content": "
Page
Reports of Independent Registered Public Accounting Firm (PCAOB ID: 42)45
Financial Statements:
Consolidated Balance Sheets48
Consolidated Statements of Income49
Consolidated Statements of Comprehensive Income50
Consolidated Statements of Stockholders’ Equity51
Consolidated Statements of Cash Flows52
Notes to Consolidated Financial Statements53
", + "path": "tables/table-19 Financial Index.html", + "metadata": { + "length": 601, + "summary": "table-20\nThe table lists page numbers for financial statements including balance sheets, income statements, and auditor reports.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-19 Financial Index.html", + "keywords": [ + "Financial Statements", + "Page Numbers", + "SEC Filing" + ], + "tokens": [] + } + }, + { + "chunk_id": "fa32ae04-53d9-5c18-a28e-298734ec6d4b", + "type": "text", + "content": "[tables/table-19 Financial Index.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 27, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [ + { + "target": "6c8cb0fd-ab33-575d-b4ed-85db334a7fc8", + "relation": "embeds", + "ref": "[tables/table-19 Financial Index.html]", + "position": { + "start": 0, + "end": 38 + } + } + ] + } + }, + { + "chunk_id": "bda85331-7fe8-5392-8b19-16901e73b47a", + "type": "text", + "content": "To the Stockholders and the Board of Directors of Alphabet Inc.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM", + "metadata": { + "length": 63, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "To", + "Stockholders", + "Board", + "Directors", + "Alphabet", + "Inc" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ecf3ab4a-b753-5d60-9046-6adf1bb39fe8", + "type": "text", + "content": "We have audited the accompanying consolidated balance sheets of Alphabet Inc. (the Company) as of December 31, 2024 and 2025, the related consolidated statements of income, comprehensive income, stockholders' equity and cash flows for each of the three years in the period ended December 31, 2025, and the related notes and financial statement schedule listed in the Index at Item 15 (collectively referred to as the “consolidated financial statements”). In our opinion, the consolidated financial statements present fairly, in all material respects, the financial position of the Company at December 31, 2024 and 2025, and the results of its operations and its cash flows for each of the three years in the period ended December 31, 2025, in conformity with U.S. generally accepted accounting principles.\nWe also have audited, in accordance with the standards of the Public Company Accounting Oversight Board (United States) (PCAOB), the Company's internal control over financial reporting as of December 31, 2025, based on criteria established in Internal Control—Integrated Framework issued by the Committee of Sponsoring Organizations of the Treadway Commission (2013 framework), and our report dated February 4, 2026 expressed an unqualified opinion thereon.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM/Opinion on the Financial Statements", + "metadata": { + "length": 1263, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "audited", + "accompanying", + "consolidated", + "balance", + "sheets", + "Alphabet", + "Inc", + "Company", + "December", + "31", + "2024", + "2025", + "related", + "statements", + "income", + "comprehensive", + "stockholders", + "equity", + "cash", + "flows", + "years", + "period", + "ended", + "notes", + "financial", + "statement", + "schedule", + "listed", + "Index", + "Item", + "15", + "collectively", + "referred", + "In", + "opinion", + "present", + "fairly", + "material", + "respects", + "position", + "results", + "operations", + "conformity", + "generally", + "accepted", + "accounting", + "principles", + "accordance", + "standards", + "Public", + "Accounting", + "Oversight", + "Board", + "United", + "States", + "PCAOB", + "internal", + "control", + "reporting", + "based", + "criteria", + "established", + "Internal", + "Control", + "Integrated", + "Framework", + "issued", + "Committee", + "Sponsoring", + "Organizations", + "Treadway", + "Commission", + "2013", + "framework", + "report", + "dated", + "February", + "2026", + "expressed", + "unqualified", + "thereon" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "11f7596c-04a6-5e85-8821-bea66ebb47b7", + "type": "text", + "content": "These financial statements are the responsibility of the Company's management. Our responsibility is to express an opinion on the Company's financial statements based on our audits. We are a public accounting firm registered with the PCAOB and are required to be independent with respect to the Company in accordance with the U.S. federal securities laws and the applicable rules and regulations of the Securities and Exchange Commission and the PCAOB.\nWe conducted our audits in accordance with the standards of the PCAOB. Those standards require that we plan and perform the audit to obtain reasonable assurance about whether the financial statements are free of material misstatement, whether due to error or fraud. Our audits included performing procedures to assess the risks of material misstatement of the financial statements, whether due to error or fraud, and performing procedures that respond to those risks. Such procedures included examining, on a test basis, evidence regarding the amounts and disclosures in the financial statements. Our audits also included evaluating the accounting principles used and significant estimates made by management, as well as evaluating the overall presentation of the financial statements. We believe that our audits provide a reasonable basis for our opinion.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM/Basis for Opinion", + "metadata": { + "length": 1309, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "These", + "financial", + "statements", + "responsibility", + "Company", + "management", + "Our", + "express", + "opinion", + "based", + "audits", + "We", + "public", + "accounting", + "firm", + "registered", + "PCAOB", + "required", + "independent", + "respect", + "accordance", + "federal", + "securities", + "laws", + "applicable", + "rules", + "regulations", + "Securities", + "Exchange", + "Commission", + "conducted", + "standards", + "Those", + "require", + "plan", + "perform", + "audit", + "obtain", + "reasonable", + "assurance", + "free", + "material", + "misstatement", + "due", + "error", + "fraud", + "included", + "performing", + "procedures", + "assess", + "risks", + "respond", + "Such", + "examining", + "test", + "basis", + "evidence", + "amounts", + "disclosures", + "evaluating", + "principles", + "significant", + "estimates", + "made", + "presentation", + "provide" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6d802d6d-2196-541f-84d8-c77bf792525c", + "type": "table", + "content": "
Description of the MatterThe Company is subject to claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders. As described in Note 10 to the consolidated financial statements, such claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders could result in adverse consequences.Significant judgment is required to determine both the likelihood and the estimated amount of a loss related to such matters. Auditing management’s accounting for and disclosure of loss contingencies from these matters involved challenging and subjective auditor judgment in assessing the Company’s evaluation of the probability of a loss, and the estimated amount or range of loss.
How We Addressed the Matter in Our AuditWe tested relevant controls over the identified risks associated with management’s accounting for and disclosure of these matters. This included controls over management’s assessment of the probability of incurrence of a loss and whether the loss or range of loss was reasonably estimable and the development of related disclosures.Our audit procedures included, among others, gaining an understanding of previous rulings and the status of ongoing lawsuits, reviewing letters from internal and external legal counsel addressing the matters, meeting with internal legal counsel to discuss the allegations, and obtaining a representation letter from management on these matters. We also evaluated the Company’s disclosures in relation to these matters.
", + "path": "tables/table-20 Loss Contingencies.html", + "metadata": { + "length": 1625, + "summary": "table-21\nThe company faces claims and lawsuits requiring judgment on loss probability and amount. Audit tested controls, reviewed legal counsel letters, and evaluated disclosures.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-20 Loss Contingencies.html", + "keywords": [ + "loss contingencies", + "legal proceedings", + "audit" + ], + "tokens": [] + } + }, + { + "chunk_id": "50357562-293d-58bc-84ed-867e9dec0f2b", + "type": "text", + "content": "The critical audit matter communicated below is a matter arising from the current period audit of the financial statements that was communicated or required to be communicated to the audit committee and that: (1) relates to accounts or disclosures that are material to the financial statements and (2) involved our especially challenging, subjective or complex judgments. The communication of the critical audit matter does not alter in any way our opinion on the consolidated financial statements, taken as a whole, and we are not, by communicating the critical audit matter below, providing a separate opinion on the critical audit matter or on the accounts or disclosures to which it relates.\nLoss Contingencies\n\n[tables/table-20 Loss Contingencies.html]\n\n/s/ Ernst & Young LLP\nWe have served as the Company's auditor since 1999.\nSan Jose, California\nFebruary 4, 2026", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM/Critical Audit Matter", + "metadata": { + "length": 899, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "critical", + "audit", + "matter", + "communicated", + "arising", + "current", + "period", + "financial", + "statements", + "required", + "committee", + "relates", + "accounts", + "disclosures", + "material", + "involved", + "challenging", + "subjective", + "complex", + "judgments", + "communication", + "alter", + "opinion", + "consolidated", + "communicating", + "providing", + "separate", + "Loss", + "Contingencies", + "Ernst", + "Young", + "LLP", + "We", + "served", + "Company", + "auditor", + "1999", + "San", + "Jose", + "California", + "February", + "2026" + ], + "keywords": [], + "connect_to": [ + { + "target": "6d802d6d-2196-541f-84d8-c77bf792525c", + "relation": "embeds", + "ref": "[tables/table-20 Loss Contingencies.html]", + "position": { + "start": 716, + "end": 757 + } + } + ] + } + }, + { + "chunk_id": "bda85331-7fe8-5392-8b19-16901e73b47a", + "type": "text", + "content": "To the Stockholders and the Board of Directors of Alphabet Inc.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM_2", + "metadata": { + "length": 63, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "To", + "Stockholders", + "Board", + "Directors", + "Alphabet", + "Inc" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ee3dd9f9-895d-5979-82d8-f5e0b210623e", + "type": "text", + "content": "We have audited Alphabet Inc.'s internal control over financial reporting as of December 31, 2025, based on criteria established in Internal Control—Integrated Framework issued by the Committee of Sponsoring Organizations of the Treadway Commission (2013 framework) (the COSO criteria). In our opinion, Alphabet Inc. (the Company) maintained, in all material respects, effective internal control over financial reporting as of December 31, 2025, based on the COSO criteria.\nWe also have audited, in accordance with the standards of the Public Company Accounting Oversight Board (United States) (PCAOB), the 2025 consolidated financial statements of the Company and our report dated February 4, 2026 expressed an unqualified opinion thereon.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM_2/Opinion on Internal Control Over Financial Reporting", + "metadata": { + "length": 740, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "audited", + "Alphabet", + "Inc", + "internal", + "control", + "financial", + "reporting", + "December", + "31", + "2025", + "based", + "criteria", + "established", + "Internal", + "Control", + "Integrated", + "Framework", + "issued", + "Committee", + "Sponsoring", + "Organizations", + "Treadway", + "Commission", + "2013", + "framework", + "COSO", + "In", + "opinion", + "Company", + "maintained", + "material", + "respects", + "effective", + "accordance", + "standards", + "Public", + "Accounting", + "Oversight", + "Board", + "United", + "States", + "PCAOB", + "consolidated", + "statements", + "report", + "dated", + "February", + "2026", + "expressed", + "unqualified", + "thereon" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5f5995bd-122c-5543-8208-a2963863de60", + "type": "text", + "content": "The Company's management is responsible for maintaining effective internal control over financial reporting and for its assessment of the effectiveness of internal control over financial reporting included in the accompanying Management's Report on Internal Control over Financial Reporting. Our responsibility is to express an opinion on the Company's internal control over financial reporting based on our audit. We are a public accounting firm registered with the PCAOB and are required to be independent with respect to the Company in accordance with the U.S. federal securities laws and the applicable rules and regulations of the Securities and Exchange Commission and the PCAOB.\nWe conducted our audit in accordance with the standards of the PCAOB. Those standards require that we plan and perform the audit to obtain reasonable assurance about whether effective internal control over financial reporting was maintained in all material respects.\nOur audit included obtaining an understanding of internal control over financial reporting, assessing the risk that a material weakness exists, testing and evaluating the design and operating effectiveness of internal control based on the assessed risk, and performing such other procedures as we considered necessary in the circumstances. We believe that our audit provides a reasonable basis for our opinion.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM_2/Basis for Opinion", + "metadata": { + "length": 1363, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "Company", + "management", + "responsible", + "maintaining", + "effective", + "internal", + "control", + "financial", + "reporting", + "assessment", + "effectiveness", + "included", + "accompanying", + "Management", + "Report", + "Internal", + "Control", + "Financial", + "Reporting", + "Our", + "responsibility", + "express", + "opinion", + "based", + "audit", + "We", + "public", + "accounting", + "firm", + "registered", + "PCAOB", + "required", + "independent", + "respect", + "accordance", + "federal", + "securities", + "laws", + "applicable", + "rules", + "regulations", + "Securities", + "Exchange", + "Commission", + "conducted", + "standards", + "Those", + "require", + "plan", + "perform", + "obtain", + "reasonable", + "assurance", + "maintained", + "material", + "respects", + "obtaining", + "understanding", + "assessing", + "risk", + "weakness", + "exists", + "testing", + "evaluating", + "design", + "operating", + "assessed", + "performing", + "procedures", + "considered", + "circumstances", + "basis" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d6f0d8ea-10fd-50a3-bb14-da4241a8b3c7", + "type": "table", + "content": "
As of December 31,
20242025
Assets
Current assets:
Cash and cash equivalents$ 23,466$ 30,708
Marketable securities72,19196,135
Total cash, cash equivalents, and marketable securities95,657126,843
Accounts receivable, net52,34062,886
Other current assets15,71416,309
Total current assets163,711206,038
Non-marketable securities37,98268,687
Deferred income taxes17,1809,113
Property and equipment, net171,036246,597
Operating lease assets13,58815,221
Goodwill31,88533,380
Other non-current assets14,87416,245
Total assets$ 450,256$ 595,281
Liabilities and Stockholders’ Equity
Current liabilities:
Accounts payable$ 7,987$ 12,200
Accrued compensation and benefits15,06917,546
Accrued expenses and other current liabilities51,22855,557
Accrued revenue share9,80210,864
Deferred revenue5,0366,578
Total current liabilities89,122102,745
Long-term debt10,88346,547
Income taxes payable, non-current8,7829,531
Operating lease liabilities11,69112,744
Other long-term liabilities4,6948,449
Total liabilities125,172180,016
Commitments and Contingencies (Note 10)
Stockholders’ equity:
Preferred stock, $0.001 par value per share, 100 shares authorized; no shares issued and outstanding00
Class A, Class B, and Class C stock and additional paid-in capital, $0.001 par value per share: 300,000 shares authorized (Class A 180,000, Class B 60,000, Class C 60,000); 12,211 (Class A 5,835, Class B 861, Class C 5,515) and 12,088 (Class A 5,822, Class B 837, Class C 5,429) shares issued and outstanding84,80093,126
Accumulated other comprehensive income (loss)(4,800)(1,916)
Retained earnings245,084324,055
Total stockholders’ equity325,084415,265
Total liabilities and stockholders’ equity$ 450,256$ 595,281
", + "path": "tables/table-21 Balance Sheet.html", + "metadata": { + "length": 3008, + "summary": "table-22\nBalance sheet as of Dec 31, 2024 and 2025: total assets $450,256 and $595,281; total liabilities $125,172 and $180,016; stockholders' equity $325,084 and $415,265.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-21 Balance Sheet.html", + "keywords": [ + "Assets", + "Liabilities", + "Equity" + ], + "tokens": [] + } + }, + { + "chunk_id": "4a1a0d55-6b9c-54e6-917f-9a9fc8e08092", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Revenues$ 307,394$ 350,018$ 402,836
Costs and expenses:
Cost of revenues133,332146,306162,535
Research and development45,42749,32661,087
Sales and marketing27,91727,80828,693
General and administrative16,42514,18821,482
Total costs and expenses223,101237,628273,797
Income from operations84,293112,390129,039
Other income (expense), net1,4247,42529,787
Income before income taxes85,717119,815158,826
Provision for income taxes11,92219,69726,656
Net income$ 73,795$ 100,118$ 132,170
Basic net income per share (Note 12)$ 5.84$ 8.13$ 10.91
Diluted net income per share (Note 12)$ 5.80$ 8.04$ 10.81
", + "path": "tables/table-22 Financial Data.html", + "metadata": { + "length": 1349, + "summary": "table-23\nThe table shows revenues, costs, and net income for 2023-2025. Revenues grew from $307,394 to $402,836, and net income rose from $73,795 to $132,170.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-22 Financial Data.html", + "keywords": [ + "Revenue", + "Net Income", + "Expenses" + ], + "tokens": [] + } + }, + { + "chunk_id": "39d3d348-fbc9-557b-a530-851e53c82c5c", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Net income$ 73,795$ 100,118$ 132,170
Other comprehensive income (loss):
Change in foreign currency translation adjustment, net of income tax benefit (expense) of $63, $(49) and $180735(1,673)2,522
Available-for-sale investments:
Change in net unrealized gains (losses)1,344(116)1,146
Less: reclassification adjustment for net (gains) losses included in net income1,168782(169)
Net change, net of income tax benefit (expense) of $(698), $(190), and $(276)2,512666977
Cash flow hedges:
Change in net unrealized gains (losses)168775(779)
Less: reclassification adjustment for net (gains) losses included in net income(214)(166)164
Net change, net of income tax benefit (expense) of $2, $(151), and $174(46)609(615)
Other comprehensive income (loss)3,201(398)2,884
Comprehensive income$ 76,996$ 99,720$ 135,054
", + "path": "tables/table-23 Comprehensive Income.html", + "metadata": { + "length": 1516, + "summary": "table-24\nThe table shows net income, other comprehensive income (OCI) components, and total comprehensive income for 2023-2025. Net income grew from $73,795 to $132,170, while OCI fluctuated, resulting in comprehensive income of $76,996, $99,720, and $135,054 respectively.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-23 Comprehensive Income.html", + "keywords": [ + "Comprehensive income", + "Net income", + "OCI" + ], + "tokens": [] + } + }, + { + "chunk_id": "618f672e-71c9-57bf-a816-53463b91fd5b", + "type": "table", + "content": "
Class A, Class B, Class C Stock and Additional Paid-In CapitalAccumulated Other Comprehensive Income (Loss)Retained EarningsTotal Stockholders' Equity
SharesAmount
Balance as of December 31, 202212,849$ 68,184$ (7,603)$ 195,563$ 256,144
Stock issued1390000
Stock-based compensation022,5780022,578
Tax withholding related to vesting of restricted stock units and other0(10,164)09(10,155)
Repurchases of stock(528)(4,064)0(58,120)(62,184)
Net income00073,79573,795
Other comprehensive income (loss)003,20103,201
Balance as of December 31, 202312,46076,534(4,402)211,247283,379
Stock issued1300000
Stock-based compensation022,9370022,937
Tax withholding related to vesting of restricted stock units and other0(12,507)0(16)(12,523)
Repurchases of stock(379)(3,359)0(58,688)(62,047)
Dividends and dividend equivalents declared ($0.60 per share)0410(7,577)(7,536)
Sale of interest in consolidated entities01,154001,154
Net income000100,118100,118
Other comprehensive income (loss)00(398)0(398)
Balance as of December 31, 202412,21184,800(4,800)245,084325,084
Stock issued1170000
Stock-based compensation025,1300025,130
Tax withholding related to vesting of restricted stock units and other0(14,842)00(14,842)
Repurchases of stock(240)(2,514)0(42,884)(45,398)
Dividends and dividend equivalents declared ($0.83 per share)01520(10,315)(10,163)
Sale of interest in consolidated entities040000400
Net income000132,170132,170
Other comprehensive income (loss)002,88402,884
Balance as of December 31, 202512,088$ 93,126$ (1,916)$ 324,055$ 415,265
", + "path": "tables/table-24 Stockholders' Equity.html", + "metadata": { + "length": 3306, + "summary": "table-25\nThe table shows stockholders' equity from 2022 to 2025, detailing changes in shares, paid-in capital, comprehensive income, and retained earnings, with net income increasing each year.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-24 Stockholders' Equity.html", + "keywords": [ + "Stockholders' Equity", + "Stock Repurchases", + "Net Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "732473c5-c78f-5c31-b842-10239502aec9", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Operating activities
Net income$ 73,795$ 100,118$ 132,170
Adjustments:
Depreciation of property and equipment11,94615,31121,136
Stock-based compensation expense22,46022,78524,953
Deferred income taxes(7,763)(5,257)8,348
Loss (gain) on debt and equity securities, net823(2,671)(24,620)
Other4,3303,4192,108
Changes in assets and liabilities, net of effects of acquisitions:
Accounts receivable, net(7,833)(5,891)(8,779)
Income taxes, net523(2,418)(3,226)
Other assets(2,143)(1,397)(4,542)
Accounts payable664359907
Accrued expenses and other liabilities3,937(1,161)12,939
Accrued revenue share4821,059899
Deferred revenue5251,0432,420
Net cash provided by operating activities101,746125,299164,713
Investing activities
Purchases of property and equipment(32,251)(52,535)(91,447)
Purchases of marketable securities(77,858)(86,679)(103,773)
Maturities and sales of marketable securities86,672103,42883,240
Purchases of non-marketable securities(3,027)(5,034)(5,716)
Maturities and sales of non-marketable securities9478821,367
Acquisitions, net of cash acquired, and purchases of intangible assets(495)(2,931)(1,592)
Other investing activities(1,051)(2,667)(2,370)
Net cash used in investing activities(27,063)(45,536)(120,291)
Financing activities
Net payments related to stock-based award activities(9,837)(12,190)(14,167)
Repurchases of stock(61,504)(62,222)(45,709)
Dividend payments0(7,363)(10,049)
Proceeds from issuance of debt, net of costs10,79013,58964,564
Repayments of debt(11,550)(12,701)(32,427)
Proceeds from sale of interest in consolidated entities, net81,154400
Net cash used in financing activities(72,093)(79,733)(37,388)
Effect of exchange rate changes on cash and cash equivalents(421)(612)208
Net increase (decrease) in cash and cash equivalents2,169(582)7,242
Cash and cash equivalents at beginning of period21,87924,04823,466
Cash and cash equivalents at end of period$ 24,048$ 23,466$ 30,708
Supplemental disclosures of non-cash investing activities:
Purchases of property and equipment included in accrued liabilities and accounts payable$ 7,435$ 10,326$ 15,090
", + "path": "tables/table-25 Cash Flow Statement.html", + "metadata": { + "length": 3983, + "summary": "table-26\nThe table shows cash flow from operating, investing, and financing activities for 2023-2025. Net cash from operations increased from $101,746 to $164,713. Investing activities used more cash over time, while financing activities fluctuated.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-25 Cash Flow Statement.html", + "keywords": [ + "cash flow", + "operating", + "investing" + ], + "tokens": [] + } + }, + { + "chunk_id": "fb0bb8f8-07fc-5d11-9d4d-9b146212adaf", + "type": "text", + "content": "A company's internal control over financial reporting is a process designed to provide reasonable assurance regarding the reliability of financial reporting and the preparation of financial statements for external purposes in accordance with generally accepted accounting principles. A company's internal control over financial reporting includes those policies and procedures that (1) pertain to the maintenance of records that, in reasonable detail, accurately and fairly reflect the transactions and dispositions of the assets of the company; (2) provide reasonable assurance that transactions are recorded as necessary to permit preparation of financial statements in accordance with generally accepted accounting principles, and that receipts and expenditures of the company are being made only in accordance with authorizations of management and directors of the company; and (3) provide reasonable assurance regarding prevention or timely detection of unauthorized acquisition, use, or disposition of the company's assets that could have a material effect on the financial statements.\nBecause of its inherent limitations, internal control over financial reporting may not prevent or detect misstatements. Also, projections of any evaluation of effectiveness to future periods are subject to the risk that controls may become inadequate because of changes in conditions, or that the degree of compliance with the policies or procedures may deteriorate.\n/s/ Ernst & Young LLP\nSan Jose, California\nFebruary 4, 2026\nAlphabet Inc.\nCONSOLIDATED BALANCE SHEETS\n(in millions, except par value per share amounts)\n\n[tables/table-21 Balance Sheet.html]\n\nSee accompanying notes.\nAlphabet Inc.\nCONSOLIDATED STATEMENTS OF INCOME\n(in millions, except per share amounts)\n\n[tables/table-22 Financial Data.html]\n\nSee accompanying notes.\nAlphabet Inc.\nCONSOLIDATED STATEMENTS OF COMPREHENSIVE INCOME\n(in millions)\n\n[tables/table-23 Comprehensive Income.html]\n\nSee accompanying notes.\nAlphabet Inc.\nCONSOLIDATED STATEMENTS OF STOCKHOLDERS' EQUITY (in millions)\n\n[tables/table-24 Stockholders' Equity.html]\n\nSee accompanying notes.\nAlphabet Inc.\nCONSOLIDATED STATEMENTS OF CASH FLOWS\n(in millions)\n\n[tables/table-25 Cash Flow Statement.html]\n\nSee accompanying notes.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM_2/Definition and Limitations of Internal Control Over Financial Reporting", + "metadata": { + "length": 2333, + "summary": "The passage describes a company's internal control over financial reporting, designed to ensure reliable financial statements in accordance with GAAP. It includes policies for accurate record-keeping, authorized transactions, and asset protection. Inherent limitations mean controls may not prevent all misstatements, and future effectiveness is subject to change. The text also includes audit firm Ernst & Young's signature and references to Alphabet Inc.'s consolidated financial statements for December 2025, including balance sheets, income statements, and cash flows.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "company", + "internal", + "control", + "financial", + "reporting", + "process", + "designed", + "provide", + "reasonable", + "assurance", + "reliability", + "preparation", + "statements", + "external", + "purposes", + "accordance", + "generally", + "accepted", + "accounting", + "principles", + "includes", + "policies", + "procedures", + "pertain", + "maintenance", + "records", + "detail", + "accurately", + "fairly", + "reflect", + "transactions", + "dispositions", + "assets", + "recorded", + "permit", + "receipts", + "expenditures", + "made", + "authorizations", + "management", + "directors", + "prevention", + "timely", + "detection", + "unauthorized", + "acquisition", + "disposition", + "material", + "effect", + "Because", + "inherent", + "limitations", + "prevent", + "detect", + "misstatements", + "Also", + "projections", + "evaluation", + "effectiveness", + "future", + "periods", + "subject", + "risk", + "controls", + "inadequate", + "conditions", + "degree", + "compliance", + "deteriorate", + "Ernst", + "Young", + "LLP", + "San", + "Jose", + "California", + "February", + "2026", + "Alphabet", + "Inc", + "CONSOLIDATED", + "BALANCE", + "SHEETS", + "millions", + "par", + "share", + "amounts", + "See", + "accompanying", + "notes", + "STATEMENTS", + "OF", + "INCOME", + "COMPREHENSIVE", + "STOCKHOLDERS", + "EQUITY", + "CASH", + "FLOWS" + ], + "keywords": [ + "Internal Control", + "Financial Reporting", + "Audit" + ], + "connect_to": [ + { + "target": "d6f0d8ea-10fd-50a3-bb14-da4241a8b3c7", + "relation": "embeds", + "ref": "[tables/table-21 Balance Sheet.html]", + "position": { + "start": 1612, + "end": 1648 + } + }, + { + "target": "4a1a0d55-6b9c-54e6-917f-9a9fc8e08092", + "relation": "embeds", + "ref": "[tables/table-22 Financial Data.html]", + "position": { + "start": 1763, + "end": 1800 + } + }, + { + "target": "39d3d348-fbc9-557b-a530-851e53c82c5c", + "relation": "embeds", + "ref": "[tables/table-23 Comprehensive Income.html]", + "position": { + "start": 1903, + "end": 1946 + } + }, + { + "target": "618f672e-71c9-57bf-a816-53463b91fd5b", + "relation": "embeds", + "ref": "[tables/table-24 Stockholders' Equity.html]", + "position": { + "start": 2049, + "end": 2092 + } + }, + { + "target": "732473c5-c78f-5c31-b842-10239502aec9", + "relation": "embeds", + "ref": "[tables/table-25 Cash Flow Statement.html]", + "position": { + "start": 2185, + "end": 2227 + } + } + ] + } + }, + { + "chunk_id": "f62b1565-e001-5645-99d0-e8614aea8de6", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "961de481-a6cc-5370-b90e-cfa29f429427", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3e7e3f16-44b6-5661-85ad-0cf46ebd9b74", + "type": "text", + "content": "Google was incorporated in California in September 1998 and re-incorporated in the State of Delaware in August 2003. In 2015, we implemented a holding company reorganization, and as a result, Alphabet Inc. (\"Alphabet\") became the successor issuer to Google.\nWe generate revenues by delivering relevant, cost-effective online advertising; cloud-based solutions that provide enterprise customers of all sizes with infrastructure, platform services, and applications; and sales of other products and services, such as fees received for subscription-based products, apps and in-app purchases, and devices.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Nature of Operations", + "metadata": { + "length": 601, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "incorporated", + "California", + "September", + "1998", + "State", + "Delaware", + "August", + "2003", + "In", + "2015", + "implemented", + "holding", + "company", + "reorganization", + "result", + "Alphabet", + "Inc", + "successor", + "issuer", + "We", + "generate", + "revenues", + "delivering", + "relevant", + "cost", + "effective", + "online", + "advertising", + "cloud", + "based", + "solutions", + "provide", + "enterprise", + "customers", + "sizes", + "infrastructure", + "platform", + "services", + "applications", + "sales", + "products", + "fees", + "received", + "subscription", + "apps", + "app", + "purchases", + "devices" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3b953578-220e-5853-b567-745abc1ef3bc", + "type": "text", + "content": "The consolidated financial statements of Alphabet include the accounts of Alphabet and entities consolidated under the variable interest and voting models. Intercompany balances and transactions have been eliminated.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Basis of Consolidation", + "metadata": { + "length": 216, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "consolidated", + "financial", + "statements", + "Alphabet", + "include", + "accounts", + "entities", + "variable", + "interest", + "voting", + "models", + "Intercompany", + "balances", + "transactions", + "eliminated" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5ed644c3-5c58-55f0-971b-bc3ba1ec7dc4", + "type": "text", + "content": "Preparation of consolidated financial statements in conformity with GAAP requires us to make estimates and assumptions that affect the amounts reported and disclosed in the financial statements and the accompanying notes. Actual results could differ materially from these estimates due to uncertainties. On an ongoing basis, we evaluate our estimates, including those related to the allowance for credit losses; contingent liabilities; fair values of financial instruments and goodwill; income taxes; inventory; and useful lives of property and equipment, among others. We base our estimates on assumptions, both historical and forward looking, that are believed to be reasonable, and the results of which form the basis for making judgments about the carrying values of assets and liabilities.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Use of Estimates", + "metadata": { + "length": 794, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Preparation", + "consolidated", + "financial", + "statements", + "conformity", + "GAAP", + "requires", + "make", + "estimates", + "assumptions", + "affect", + "amounts", + "reported", + "disclosed", + "accompanying", + "notes", + "Actual", + "results", + "differ", + "materially", + "due", + "uncertainties", + "On", + "ongoing", + "basis", + "evaluate", + "including", + "related", + "allowance", + "credit", + "losses", + "contingent", + "liabilities", + "fair", + "values", + "instruments", + "goodwill", + "income", + "taxes", + "inventory", + "lives", + "property", + "equipment", + "We", + "base", + "historical", + "forward", + "believed", + "reasonable", + "form", + "making", + "judgments", + "carrying", + "assets" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "35d9b68b-ca08-5ce5-aedd-7e8f373eb2e5", + "type": "text", + "content": "Revenues are recognized when control of the promised goods or services is transferred to our customers, and the collectibility of an amount that we expect in exchange for those goods or services is probable. Sales and other similar taxes are excluded from revenues.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition", + "metadata": { + "length": 265, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Revenues", + "recognized", + "control", + "promised", + "goods", + "services", + "transferred", + "customers", + "collectibility", + "amount", + "expect", + "exchange", + "probable", + "Sales", + "similar", + "taxes", + "excluded", + "revenues" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4e3e9ba6-cebf-561c-8ea7-684c9ac0ad13", + "type": "text", + "content": "Google advertising revenues consist of revenues from:\n- Google Search and other properties, including revenues from traffic generated by search distribution partners who use Google.com as their default search in browsers, toolbars, etc. and other Google owned and operated properties like Gmail, Google Maps, and Google Play;\n• YouTube properties; and\n- Google Network properties, including revenues from Google Network properties participating in AdMob, AdSense, and Google Ad Manager.\nOur customers generally purchase advertising inventory through Google Ads, Google Ad Manager, Google Display & Video 360, and Google Marketing Platform, among others.\nWe offer advertising by delivering both performance and brand advertising. We recognize revenues for performance advertising when a user engages with the advertisement. For brand advertising, we recognize revenues when the ad is displayed, or a user views the ad.\nFor ads placed on Google Network properties, we evaluate whether we are the principal (i.e., report revenues on a gross basis) or agent (i.e., report revenues on a net basis). Generally, we report advertising revenues for ads placed on Google Network properties on a gross basis, that is, the amounts billed to our customers are recorded as revenues, and amounts paid to Google Network partners are recorded as cost of revenues. Where we are the principal, we control the advertising inventory before it is transferred to our customers. Our control is evidenced by our sole ability to monetize the advertising inventory before it is transferred to our customers and is further supported by us being primarily responsible to our customers and having a level of discretion in establishing pricing.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Google Advertising", + "metadata": { + "length": 1713, + "summary": "Google advertising revenues come from Google Search, YouTube, and Google Network properties. Customers purchase inventory through platforms like Google Ads and Google Ad Manager. Performance advertising revenue is recognized upon user engagement, while brand advertising revenue is recognized upon display or view. For Google Network ads, Google generally acts as principal, reporting revenue on a gross basis.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "advertising", + "revenues", + "consist", + "Search", + "properties", + "including", + "traffic", + "generated", + "search", + "distribution", + "partners", + "default", + "browsers", + "toolbars", + "owned", + "operated", + "Gmail", + "Maps", + "Play", + "YouTube", + "Network", + "participating", + "AdMob", + "AdSense", + "Ad", + "Manager", + "Our", + "customers", + "generally", + "purchase", + "inventory", + "Ads", + "Display", + "Video", + "360", + "Marketing", + "Platform", + "We", + "offer", + "delivering", + "performance", + "brand", + "recognize", + "user", + "engages", + "advertisement", + "For", + "ad", + "displayed", + "views", + "ads", + "evaluate", + "principal", + "report", + "gross", + "basis", + "agent", + "net", + "Generally", + "amounts", + "billed", + "recorded", + "paid", + "cost", + "Where", + "control", + "transferred", + "evidenced", + "sole", + "ability", + "monetize", + "supported", + "primarily", + "responsible", + "level", + "discretion", + "establishing", + "pricing" + ], + "keywords": [ + "advertising", + "revenue", + "Google" + ], + "connect_to": [] + } + }, + { + "chunk_id": "64887c68-47ef-5b79-ba80-97728d1e2a5d", + "type": "text", + "content": "Google subscriptions, platforms, and devices revenues consist of revenues from:\n- consumer subscriptions, which primarily include revenues from YouTube services, such as YouTube TV, YouTube Music and Premium, and NFL Sunday Ticket, as well as Google One, which offers access to our most capable Gemini models;\n- platforms, which primarily include revenues from Google Play sales of apps and in-app purchases;\n• devices, which primarily include sales of the Pixel family of devices; and\n• other products and services.\nSubscription revenues are recognized ratably over the period of the subscription, primarily monthly. We report revenues from Google Play sales of apps and in-app purchases on a net basis because our performance obligation is to facilitate a transaction between app developers and end users for which we earn a service fee.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Google Subscriptions, Platforms, and Devices", + "metadata": { + "length": 839, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "subscriptions", + "platforms", + "devices", + "revenues", + "consist", + "consumer", + "primarily", + "include", + "YouTube", + "services", + "TV", + "Music", + "Premium", + "NFL", + "Sunday", + "Ticket", + "One", + "offers", + "access", + "capable", + "Gemini", + "models", + "Play", + "sales", + "apps", + "app", + "purchases", + "Pixel", + "family", + "products", + "Subscription", + "recognized", + "ratably", + "period", + "subscription", + "monthly", + "We", + "report", + "net", + "basis", + "performance", + "obligation", + "facilitate", + "transaction", + "developers", + "end", + "users", + "earn", + "service", + "fee" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f5112a17-44a4-5e7f-aa59-7b78130348aa", + "type": "text", + "content": "Google Cloud revenues consist of revenues from:\n- Google Cloud Platform primarily generates consumption-based fees and subscriptions for infrastructure, platform, and other services. These services provide access to solutions such as AI offerings including our enterprise AI infrastructure, Vertex AI platform, and Gemini Enterprise; cybersecurity offerings; and data and analytics solutions;\n- Google Workspace includes subscriptions for cloud-based communication and collaboration tools for enterprises, such as Gmail, Docs, Calendar, Drive, and Meet, with integrated features like Gemini for Google Workspace; and\n• other enterprise services.\nOur cloud services are generally provided on either a consumption or subscription basis and may have contract terms longer than a year. Revenues related to cloud services provided on a consumption basis are recognized when the customer utilizes the services, based on the quantity of services consumed using the relative standalone selling price allocation. Revenues related to cloud services provided on a subscription basis are recognized ratably over the contract term as the customer receives and consumes the benefits of the cloud services.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Google Cloud", + "metadata": { + "length": 1191, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Google", + "Cloud", + "revenues", + "consist", + "Platform", + "primarily", + "generates", + "consumption", + "based", + "fees", + "subscriptions", + "infrastructure", + "platform", + "services", + "These", + "provide", + "access", + "solutions", + "AI", + "offerings", + "including", + "enterprise", + "Vertex", + "Gemini", + "Enterprise", + "cybersecurity", + "data", + "analytics", + "Workspace", + "includes", + "cloud", + "communication", + "collaboration", + "tools", + "enterprises", + "Gmail", + "Docs", + "Calendar", + "Drive", + "Meet", + "integrated", + "features", + "Our", + "generally", + "provided", + "subscription", + "basis", + "contract", + "terms", + "longer", + "year", + "Revenues", + "related", + "recognized", + "customer", + "utilizes", + "quantity", + "consumed", + "relative", + "standalone", + "selling", + "price", + "allocation", + "ratably", + "term", + "receives", + "consumes", + "benefits" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "068978eb-deee-5ab4-b790-efa367d3483e", + "type": "text", + "content": "Our contracts with customers may include multiple performance obligations. For such arrangements, we allocate revenues to each performance obligation based on its relative standalone selling price. We generally determine standalone selling prices based on observable prices of our products and services sold or priced separately in comparable circumstances to similar customers.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Arrangements with Multiple Performance Obligations", + "metadata": { + "length": 378, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "contracts", + "customers", + "include", + "multiple", + "performance", + "obligations", + "For", + "arrangements", + "allocate", + "revenues", + "obligation", + "based", + "relative", + "standalone", + "selling", + "price", + "We", + "generally", + "determine", + "prices", + "observable", + "products", + "services", + "sold", + "priced", + "separately", + "comparable", + "circumstances", + "similar" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a64f07ab-e7c7-54ae-a365-d36db8c5366e", + "type": "text", + "content": "Certain customers receive cash-based incentives or credits, which are accounted for as variable consideration. We estimate these amounts based on the expected amount to be provided to customers and reduce revenues. We believe that there will not be significant changes to our estimates of variable consideration related to customer incentives and credits.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Customer Incentives and Credits", + "metadata": { + "length": 355, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Certain", + "customers", + "receive", + "cash", + "based", + "incentives", + "credits", + "accounted", + "variable", + "consideration", + "We", + "estimate", + "amounts", + "expected", + "amount", + "provided", + "reduce", + "revenues", + "significant", + "estimates", + "related", + "customer" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f5e4b355-98db-57e6-80b9-0809d9d2cad5", + "type": "text", + "content": "We expense sales commissions when incurred when the period of the expected benefit is one year or less. We recognize an asset for certain sales commissions and amortize if the expected benefit period is greater than one year. These costs are recorded within sales and marketing expenses.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Sales Commissions", + "metadata": { + "length": 287, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "expense", + "sales", + "commissions", + "incurred", + "period", + "expected", + "benefit", + "year", + "recognize", + "asset", + "amortize", + "greater", + "These", + "costs", + "recorded", + "marketing", + "expenses" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8dfb08f1-ce1d-5cfa-93be-31a8c514a147", + "type": "text", + "content": "Cost of revenues consists of TAC and other costs of revenues.\n• TAC includes:\namounts paid to our distribution partners who make available our search access points and other ad-supported services. Our distribution partners include browser providers, mobile carriers, original equipment manufacturers, and software developers; and\n- amounts paid to Google Network partners primarily for ads displayed on their properties.\n• Other cost of revenues includes:\ncontent acquisition costs, which are payments to content providers from whom we license video and other content for distribution, primarily related to YouTube (we pay fees to these content providers based on revenues generated, subscriber counts, or a flat fee);\n• depreciation expense, primarily related to our technical infrastructure;\n• employee compensation expenses related to our technical infrastructure and other operations such as content review and customer and product support;\ninventory and other costs related to the devices we sell; and\nother technical infrastructure operations costs, including energy, equipment, and network capacity costs.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Cost of Revenues", + "metadata": { + "length": 1112, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Cost", + "revenues", + "consists", + "TAC", + "costs", + "includes", + "amounts", + "paid", + "distribution", + "partners", + "make", + "search", + "access", + "points", + "ad", + "supported", + "services", + "Our", + "include", + "browser", + "providers", + "mobile", + "carriers", + "original", + "equipment", + "manufacturers", + "software", + "developers", + "Google", + "Network", + "primarily", + "ads", + "displayed", + "properties", + "Other", + "cost", + "content", + "acquisition", + "payments", + "license", + "video", + "related", + "YouTube", + "pay", + "fees", + "based", + "generated", + "subscriber", + "counts", + "flat", + "fee", + "depreciation", + "expense", + "technical", + "infrastructure", + "employee", + "compensation", + "expenses", + "operations", + "review", + "customer", + "product", + "support", + "inventory", + "devices", + "sell", + "including", + "energy", + "network", + "capacity" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a5655c13-e1e1-5631-ba25-7526c01f759b", + "type": "text", + "content": "We expense software development costs, including costs to develop software products or the software component of products to be sold, leased, or marketed to external users, before technological feasibility is reached. Technological feasibility is typically reached shortly before the release of such products. As a result, development costs that meet the criteria for capitalization were not material for the periods presented.\nSoftware development costs also include costs to develop software to be used solely to meet internal needs and cloud-based applications used to deliver our services. We capitalize development costs related to these software applications once the preliminary project stage is complete and it is probable that the project will be completed and the software will be used to perform the function intended. Costs capitalized for developing such software applications were not material for the periods presented.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Software Development Costs", + "metadata": { + "length": 934, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "expense", + "software", + "development", + "costs", + "including", + "develop", + "products", + "component", + "sold", + "leased", + "marketed", + "external", + "users", + "technological", + "feasibility", + "reached", + "Technological", + "typically", + "shortly", + "release", + "As", + "result", + "meet", + "criteria", + "capitalization", + "material", + "periods", + "presented", + "Software", + "include", + "solely", + "internal", + "cloud", + "based", + "applications", + "deliver", + "services", + "capitalize", + "related", + "preliminary", + "project", + "stage", + "complete", + "probable", + "completed", + "perform", + "function", + "intended", + "Costs", + "capitalized", + "developing" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e8ceb96a-dab5-5dd4-8b96-98482342179c", + "type": "text", + "content": "Stock-based compensation (SBC) primarily consists of Alphabet restricted stock units (RSUs). RSUs are equity classified and measured at the fair market value of the underlying stock at the grant date. We recognize RSU expense using the straight-line attribution method over the requisite service period and account for forfeitures as they occur. RSUs are awarded dividend equivalents, which are subject to the same vesting conditions as the underlying award, and settled in Class C shares.\nFor RSUs, shares are issued on the vesting dates net of the applicable statutory income tax withholding to be paid by us on behalf of our employees. As a result, fewer shares are issued than the number of RSUs vested, and the income tax withholding is recorded as a reduction to additional paid-in capital.\nAdditionally, SBC includes other stock-based awards, such as performance stock units (PSUs) that include market conditions and awards that may be settled in cash or the stock of certain Other Bet companies. PSUs and certain awards granted by Other Bet companies are equity classified and expense is recognized over the requisite service period. Certain awards granted by Other Bet companies are liability classified and remeasured at fair value through settlement. The fair value of awards granted by Other Bet companies is based on the equity valuation of the respective Other Bet company.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Stock-Based Compensation", + "metadata": { + "length": 1387, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Stock", + "based", + "compensation", + "SBC", + "primarily", + "consists", + "Alphabet", + "restricted", + "stock", + "units", + "RSUs", + "equity", + "classified", + "measured", + "fair", + "market", + "underlying", + "grant", + "date", + "We", + "recognize", + "RSU", + "expense", + "straight", + "line", + "attribution", + "method", + "requisite", + "service", + "period", + "account", + "forfeitures", + "occur", + "awarded", + "dividend", + "equivalents", + "subject", + "vesting", + "conditions", + "award", + "settled", + "Class", + "shares", + "For", + "issued", + "dates", + "net", + "applicable", + "statutory", + "income", + "tax", + "withholding", + "paid", + "behalf", + "employees", + "As", + "result", + "fewer", + "number", + "vested", + "recorded", + "reduction", + "additional", + "capital", + "Additionally", + "includes", + "awards", + "performance", + "PSUs", + "include", + "cash", + "Other", + "Bet", + "companies", + "granted", + "recognized", + "Certain", + "liability", + "remeasured", + "settlement", + "The", + "valuation", + "respective", + "company" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2456b038-f05e-5923-b99d-aba3731bad50", + "type": "text", + "content": "We expense advertising and promotional costs in the period in which they are incurred. For the years ended December 31, 2023, 2024, and 2025, advertising and promotional expenses totaled approximately \\$8.7 billion, \\$8.7 billion, and \\$9.9 billion, respectively.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Advertising and Promotional Expenses", + "metadata": { + "length": 263, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "expense", + "advertising", + "promotional", + "costs", + "period", + "incurred", + "For", + "years", + "ended", + "December", + "31", + "2023", + "2024", + "2025", + "expenses", + "totaled", + "approximately", + "8.7", + "billion", + "9.9" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a12ef2ec-d7e2-5ae5-88f0-21a39867692c", + "type": "text", + "content": "Fair value is an exit price, representing the amount that would be received to sell an asset or paid to transfer a liability in an orderly transaction between market participants. Assets and liabilities recorded at fair value are measured and classified in accordance with a three-tier fair value hierarchy based on the observability of the inputs available in the market used to measure fair value:\nLevel 1 - Observable inputs that reflect quoted prices (unadjusted) for identical assets or liabilities in active markets.\nLevel 2 - Inputs that are based upon quoted prices for similar instruments in active markets, quoted prices for identical or similar instruments in markets that are not active, and model-based valuation techniques for which all significant inputs are observable in the market or can be derived from observable market data. Where applicable, these models project future cash flows and discount the future amounts to a present value using market-based observable inputs including interest rate curves, foreign exchange rates, and credit ratings.\nLevel 3 - Unobservable inputs that are supported by little or no market activities.\nThe fair value hierarchy requires an entity to maximize the use of observable inputs and minimize the use of unobservable inputs when measuring fair value. The determination of fair value involves the use of appropriate valuation methods and relevant inputs into valuation models.\nOur financial assets and liabilities that are measured at fair value on a recurring basis include cash equivalents, marketable securities, and derivative financial instruments. Our financial assets measured at fair value on a nonrecurring basis include non-marketable equity securities. Other financial assets and liabilities are carried at cost with fair value disclosed, if required.\nWe measure certain other instruments, and certain assets and liabilities acquired in a business combination, also at fair value on a nonrecurring basis.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Fair Value Measurements", + "metadata": { + "length": 1970, + "summary": "Fair value is an exit price for selling assets or transferring liabilities in an orderly transaction. It uses a three-tier hierarchy: Level 1 (quoted prices in active markets), Level 2 (observable inputs from similar instruments or models), and Level 3 (unobservable inputs). The hierarchy prioritizes observable inputs. Financial assets and liabilities measured at fair value include cash equivalents, marketable securities, derivatives, and non-marketable equity securities.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Fair", + "exit", + "price", + "representing", + "amount", + "received", + "sell", + "asset", + "paid", + "transfer", + "liability", + "orderly", + "transaction", + "market", + "participants", + "Assets", + "liabilities", + "recorded", + "fair", + "measured", + "classified", + "accordance", + "tier", + "hierarchy", + "based", + "observability", + "inputs", + "measure", + "Level", + "Observable", + "reflect", + "quoted", + "prices", + "unadjusted", + "identical", + "assets", + "active", + "markets", + "Inputs", + "similar", + "instruments", + "model", + "valuation", + "techniques", + "significant", + "observable", + "derived", + "data", + "Where", + "applicable", + "models", + "project", + "future", + "cash", + "flows", + "discount", + "amounts", + "present", + "including", + "interest", + "rate", + "curves", + "foreign", + "exchange", + "rates", + "credit", + "ratings", + "Unobservable", + "supported", + "activities", + "The", + "requires", + "entity", + "maximize", + "minimize", + "unobservable", + "measuring", + "determination", + "involves", + "methods", + "relevant", + "Our", + "financial", + "recurring", + "basis", + "include", + "equivalents", + "marketable", + "securities", + "derivative", + "nonrecurring", + "equity", + "Other", + "carried", + "cost", + "disclosed", + "required", + "We", + "acquired", + "business", + "combination" + ], + "keywords": [ + "Fair value", + "Hierarchy", + "Observable inputs" + ], + "connect_to": [] + } + }, + { + "chunk_id": "3d51c76c-dbd7-5142-aa28-163cc0e75b00", + "type": "text", + "content": "Our financial instruments include cash, cash equivalents, marketable and non-marketable securities, derivative financial instruments, financial guarantees, accounts receivable, and convertible notes.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments", + "metadata": { + "length": 199, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "financial", + "instruments", + "include", + "cash", + "equivalents", + "marketable", + "securities", + "derivative", + "guarantees", + "accounts", + "receivable", + "convertible", + "notes" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "75eb49a9-da90-5fd1-b14a-05efc18a33dc", + "type": "text", + "content": "We are subject to concentration of credit risk primarily from cash equivalents, marketable debt securities, derivative financial instruments, including foreign exchange contracts, accounts receivable, and convertible notes. We manage the concentration of our credit risk exposure through timely assessment of our counterparty creditworthiness, credit limits, and use of collateral management. Foreign exchange contracts are transacted with various financial institutions with high credit standing. Accounts receivable are typically unsecured and are derived from revenues earned from customers located around the world. We manage the concentration of our credit risk exposure by performing ongoing evaluations to determine customer credit and we limit the amount of credit we extend. We generally do not require collateral from our customers.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Credit Risks", + "metadata": { + "length": 842, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "subject", + "concentration", + "credit", + "risk", + "primarily", + "cash", + "equivalents", + "marketable", + "debt", + "securities", + "derivative", + "financial", + "instruments", + "including", + "foreign", + "exchange", + "contracts", + "accounts", + "receivable", + "convertible", + "notes", + "manage", + "exposure", + "timely", + "assessment", + "counterparty", + "creditworthiness", + "limits", + "collateral", + "management", + "Foreign", + "transacted", + "institutions", + "high", + "standing", + "Accounts", + "typically", + "unsecured", + "derived", + "revenues", + "earned", + "customers", + "located", + "world", + "performing", + "ongoing", + "evaluations", + "determine", + "customer", + "limit", + "amount", + "extend", + "generally", + "require" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7992fa1d-869f-54b2-a2f4-22c5d1ed87f4", + "type": "text", + "content": "We invest excess cash primarily in asset-backed and mortgage-backed securities, corporate debt securities, government bonds, money market funds, and time deposits.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Cash Equivalents", + "metadata": { + "length": 163, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "invest", + "excess", + "cash", + "primarily", + "asset", + "backed", + "mortgage", + "securities", + "corporate", + "debt", + "government", + "bonds", + "money", + "market", + "funds", + "time", + "deposits" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4b918c7d-dfce-5c4f-9266-573aee652fae", + "type": "text", + "content": "We classify all marketable debt securities that have effective maturities of three months or less from the date of purchase as cash equivalents and those with effective maturities of greater than three months as marketable securities. We determine the appropriate classification of our investments in marketable debt securities at the time of purchase and reevaluate such designation at each balance sheet date. We have classified and accounted for our marketable debt securities as available-for-sale. After consideration of our risk versus reward objectives, as well as our liquidity requirements, we may sell these debt securities prior to their effective maturities. As we view these securities as available to support current operations, we classify highly liquid securities with maturities beyond 12 months as current assets under the caption marketable securities. We carry these securities at fair value, and report the unrealized gains and losses, net of taxes, as a component of stockholders' equity, except for the changes in allowance for expected credit losses, which are recorded in OI&E. For certain marketable debt securities we have elected the fair value option, for which changes in fair value are recorded in OI&E. We determine any realized gains and losses on the sale of marketable debt securities on a specific identification method, and we record such gains and losses as a component of OI&E.\nOur investments in marketable equity securities are measured at fair value with the related gains and losses, including unrealized, recognized in OI&E.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Marketable Securities", + "metadata": { + "length": 1568, + "summary": "The passage describes the classification and accounting treatment of marketable debt securities. Securities with maturities of three months or less are cash equivalents; longer maturities are marketable securities. They are classified as available-for-sale, carried at fair value, with unrealized gains/losses (net of taxes) in stockholders' equity, except credit loss allowances in OI&E. Some securities use the fair value option with changes in OI&E. Realized gains/losses use specific identification and are in OI&E. Marketable equity securities are also at fair value with gains/losses in OI&E.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "classify", + "marketable", + "debt", + "securities", + "effective", + "maturities", + "months", + "date", + "purchase", + "cash", + "equivalents", + "greater", + "determine", + "classification", + "investments", + "time", + "reevaluate", + "designation", + "balance", + "sheet", + "classified", + "accounted", + "sale", + "After", + "consideration", + "risk", + "versus", + "reward", + "objectives", + "liquidity", + "requirements", + "sell", + "prior", + "As", + "view", + "support", + "current", + "operations", + "highly", + "liquid", + "12", + "assets", + "caption", + "carry", + "fair", + "report", + "unrealized", + "gains", + "losses", + "net", + "taxes", + "component", + "stockholders", + "equity", + "allowance", + "expected", + "credit", + "recorded", + "OI", + "For", + "elected", + "option", + "realized", + "specific", + "identification", + "method", + "record", + "Our", + "measured", + "related", + "including", + "recognized" + ], + "keywords": [ + "marketable debt", + "fair value", + "classification" + ], + "connect_to": [] + } + }, + { + "chunk_id": "63e39cae-0db9-5757-8019-d762b6ab08e7", + "type": "text", + "content": "Non-marketable securities primarily consist of equity securities. We account for non-marketable equity securities through which we exercise significant influence but do not have control over the investee under the equity method. Other non-marketable equity securities that we hold are primarily accounted for under the measurement alternative. Under the measurement alternative, the carrying value is measured at cost, less any impairment, plus or minus changes resulting from observable price changes in orderly transactions for identical or similar investments of the same issuer. Adjustments are determined primarily based on a market approach as of the transaction date and are recorded as a component of OI&E.\nNon-marketable securities that do not have effective contractual maturity dates are classified as other non-current assets.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Non-Marketable Securities", + "metadata": { + "length": 838, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Non", + "marketable", + "securities", + "primarily", + "consist", + "equity", + "We", + "account", + "exercise", + "significant", + "influence", + "control", + "investee", + "method", + "Other", + "hold", + "accounted", + "measurement", + "alternative", + "Under", + "carrying", + "measured", + "cost", + "impairment", + "minus", + "resulting", + "observable", + "price", + "orderly", + "transactions", + "identical", + "similar", + "investments", + "issuer", + "Adjustments", + "determined", + "based", + "market", + "approach", + "transaction", + "date", + "recorded", + "component", + "OI", + "effective", + "contractual", + "maturity", + "dates", + "classified", + "current", + "assets" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "749a1337-91d7-5cda-b503-cb4950c7f8b5", + "type": "text", + "content": "See Note 3 for the accounting policy pertaining to derivative financial instruments.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Derivative Financial Instruments", + "metadata": { + "length": 84, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "See", + "Note", + "accounting", + "policy", + "pertaining", + "derivative", + "financial", + "instruments" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1a01c4df-78f5-51c2-9e57-6fc03e28a16a", + "type": "text", + "content": "In certain arrangements, we provide reimbursements for costs incurred by third parties during power generation project development phases if specified trigger events occur. We recognize a noncontingent liability for the fair value of our obligation to stand ready to perform, reported in other long-term liabilities. We also recognize a contingent liability when it becomes probable that a payment will be required and the amount can be reasonably estimated.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Financial Guarantees", + "metadata": { + "length": 458, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "In", + "arrangements", + "provide", + "reimbursements", + "costs", + "incurred", + "parties", + "power", + "generation", + "project", + "development", + "phases", + "trigger", + "events", + "occur", + "We", + "recognize", + "noncontingent", + "liability", + "fair", + "obligation", + "stand", + "ready", + "perform", + "reported", + "long", + "term", + "liabilities", + "contingent", + "probable", + "payment", + "required", + "amount", + "estimated" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b2664bc1-29dc-5586-bc9e-45e4c3ac944a", + "type": "text", + "content": "Our payment terms for accounts receivable vary by the types and locations of our customers and the products or services offered. The term between invoicing and when payment is due is not significant. Additionally, accounts receivable includes amounts for services performed in advance of the right to invoice the customer.\nWe maintain an allowance for credit losses for accounts receivable, which is recorded as an offset to accounts receivable, and changes in such are classified as general and administrative expense. We assess collectibility by reviewing accounts receivable on a collective basis where similar characteristics exist and on an individual basis when we identify specific customers with known disputes or collectibility issues. With respect to current accounts receivables, we elected to assume that current conditions as of the balance sheet date do not change for the remaining life of the asset. In determining the amount of the allowance for credit losses for those assets, we adjust historical loss information to reflect current market conditions and customer-specific information to the extent that historical loss information does not reflect current conditions.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Accounts Receivable", + "metadata": { + "length": 1187, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "payment", + "terms", + "accounts", + "receivable", + "vary", + "types", + "locations", + "customers", + "products", + "services", + "offered", + "The", + "term", + "invoicing", + "due", + "significant", + "Additionally", + "includes", + "amounts", + "performed", + "advance", + "invoice", + "customer", + "We", + "maintain", + "allowance", + "credit", + "losses", + "recorded", + "offset", + "classified", + "general", + "administrative", + "expense", + "assess", + "collectibility", + "reviewing", + "collective", + "basis", + "similar", + "characteristics", + "exist", + "individual", + "identify", + "specific", + "disputes", + "issues", + "With", + "respect", + "current", + "receivables", + "elected", + "assume", + "conditions", + "balance", + "sheet", + "date", + "change", + "remaining", + "life", + "asset", + "In", + "determining", + "amount", + "assets", + "adjust", + "historical", + "loss", + "information", + "reflect", + "market", + "extent" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "985546b4-ea65-5b2c-b1c9-989d4c447e09", + "type": "text", + "content": "Our investments in convertible notes are primarily recorded at amortized cost which includes unpaid principal balances, deferred origination costs, and any related discount or premium, net of allowances for credit losses, and are included within other non-current assets.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Convertible Notes", + "metadata": { + "length": 271, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "investments", + "convertible", + "notes", + "primarily", + "recorded", + "amortized", + "cost", + "includes", + "unpaid", + "principal", + "balances", + "deferred", + "origination", + "costs", + "related", + "discount", + "premium", + "net", + "allowances", + "credit", + "losses", + "included", + "current", + "assets" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ec208930-e017-5bb1-9fc9-fd8c47329d02", + "type": "text", + "content": "Our financial instruments also include debt and equity investments in companies with which we also entered into commercial arrangements at or near the same time. For these transactions, judgment is required in assessing the substance of the arrangements, including assessing whether the components of the arrangements should be accounted for as separate transactions under the applicable GAAP, and determining the value of the components of the arrangements, including the fair value of the investments. Additionally, if our investment in such companies becomes impaired, we may need to re-evaluate the accounting for the commercial arrangement, including reducing any remaining performance obligations.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Other", + "metadata": { + "length": 703, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "financial", + "instruments", + "include", + "debt", + "equity", + "investments", + "companies", + "entered", + "commercial", + "arrangements", + "time", + "For", + "transactions", + "judgment", + "required", + "assessing", + "substance", + "including", + "components", + "accounted", + "separate", + "applicable", + "GAAP", + "determining", + "fair", + "Additionally", + "investment", + "impaired", + "evaluate", + "accounting", + "arrangement", + "reducing", + "remaining", + "performance", + "obligations" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a086f7d2-556d-5721-912d-4581c7fa584a", + "type": "text", + "content": "We periodically review our debt securities with unrealized gains and losses recorded as a component of stockholders' equity and non-marketable equity securities for impairment.\nFor debt securities in an unrealized loss position, we determine whether a credit loss exists. The credit loss is estimated by considering available information relevant to the collectibility of the security and information about past events, current conditions, and reasonable and supportable forecasts. Any credit loss is recorded as a charge to OI&E, not to exceed the amount of the unrealized loss. Unrealized losses other than the credit loss are recognized in AOCI. If we have an intent to sell, or if it is more likely than not that we will be required to sell a debt security in an unrealized loss position before recovery of its amortized cost basis, we will write down the security to its fair value and record the corresponding charge as a component of OI&E.\nFor non-marketable equity securities, including equity method investments, we consider whether impairment indicators exist by evaluating the companies' financial and liquidity position and access to capital resources, among other indicators. If the assessment indicates that the investment is impaired, we write down the investment to its fair value by recording the corresponding charge as a component of OI&E. We prepare quantitative measurements of the fair value of our equity investments using a market approach or an income approach.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Impairment of Investments", + "metadata": { + "length": 1486, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "periodically", + "review", + "debt", + "securities", + "unrealized", + "gains", + "losses", + "recorded", + "component", + "stockholders", + "equity", + "marketable", + "impairment", + "For", + "loss", + "position", + "determine", + "credit", + "exists", + "The", + "estimated", + "information", + "relevant", + "collectibility", + "security", + "past", + "events", + "current", + "conditions", + "reasonable", + "supportable", + "forecasts", + "Any", + "charge", + "OI", + "exceed", + "amount", + "Unrealized", + "recognized", + "AOCI", + "If", + "intent", + "sell", + "required", + "recovery", + "amortized", + "cost", + "basis", + "write", + "fair", + "record", + "including", + "method", + "investments", + "indicators", + "exist", + "evaluating", + "companies", + "financial", + "liquidity", + "access", + "capital", + "resources", + "assessment", + "investment", + "impaired", + "recording", + "prepare", + "quantitative", + "measurements", + "market", + "approach", + "income" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e68ee47f-a442-5aa2-88a6-61fb2eb01158", + "type": "text", + "content": "Inventory consists primarily of finished goods and is stated at the lower of cost and net realizable value. Cost is generally computed using the first-in, first-out method.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Inventory", + "metadata": { + "length": 172, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Inventory", + "consists", + "primarily", + "finished", + "goods", + "stated", + "lower", + "cost", + "net", + "realizable", + "Cost", + "generally", + "computed", + "method" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3d53afc2-36f8-576b-a07f-e4c5073bdc5e", + "type": "text", + "content": "We determine at the inception of each arrangement whether an entity in which we have made an investment or in which we have other variable interests is considered a variable interest entity (VIE). We consolidate VIEs when we are the primary beneficiary. We are the primary beneficiary of a VIE when we have the power to direct activities that most significantly affect the economic performance of the VIE and have the obligation to absorb the majority of their losses or benefits. If we are not the primary beneficiary in a VIE, we account for the investment or other variable interests in a VIE in accordance with applicable GAAP.\nPeriodically, we assess whether any changes in our interest or relationship with the entity affect our determination of whether the entity is a VIE and, if so, whether we are the primary beneficiary.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Variable Interest Entities", + "metadata": { + "length": 831, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "determine", + "inception", + "arrangement", + "entity", + "made", + "investment", + "variable", + "interests", + "considered", + "interest", + "VIE", + "consolidate", + "VIEs", + "primary", + "beneficiary", + "power", + "direct", + "activities", + "significantly", + "affect", + "economic", + "performance", + "obligation", + "absorb", + "majority", + "losses", + "benefits", + "If", + "account", + "accordance", + "applicable", + "GAAP", + "Periodically", + "assess", + "relationship", + "determination" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "eba31bed-562f-5afd-9190-2d1f52d7635f", + "type": "text", + "content": "Property and equipment is comprised of technical infrastructure, office space, corporate and other assets currently in service, and assets not yet in service. Technical infrastructure includes data center land, buildings and leasehold improvements, and servers and network equipment. Office space includes office land, buildings, and leasehold improvements. Assets not yet in service are those that are not ready for their intended use, including data center buildings and servers in the process of construction or assembly.\nProperty and equipment are stated at cost less accumulated depreciation. Depreciation commences once assets are ready for their intended use and is recorded using the straight-line method over the estimated useful lives of the assets, which we regularly evaluate for factors such as technological obsolescence and our planned use and utilization. We depreciate data center and office buildings over periods of seven to 40 years. We depreciate servers and network equipment generally over a period of six years. We depreciate corporate and other assets over periods of two to 25 years. We depreciate leasehold improvements over the shorter of the remaining lease term or the estimated useful lives of the assets. Land is not depreciated.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Property and Equipment", + "metadata": { + "length": 1261, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Property", + "equipment", + "comprised", + "technical", + "infrastructure", + "office", + "space", + "corporate", + "assets", + "service", + "Technical", + "includes", + "data", + "center", + "land", + "buildings", + "leasehold", + "improvements", + "servers", + "network", + "Office", + "Assets", + "ready", + "intended", + "including", + "process", + "construction", + "assembly", + "stated", + "cost", + "accumulated", + "depreciation", + "Depreciation", + "commences", + "recorded", + "straight", + "line", + "method", + "estimated", + "lives", + "regularly", + "evaluate", + "factors", + "technological", + "obsolescence", + "planned", + "utilization", + "We", + "depreciate", + "periods", + "40", + "years", + "generally", + "period", + "25", + "shorter", + "remaining", + "lease", + "term", + "Land", + "depreciated" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "df676685-16d7-5d29-946c-c64673c40969", + "type": "text", + "content": "We allocate goodwill to reporting units based on the expected benefit from the business combination. We evaluate our reporting units periodically, as well as when changes in our operating segments occur. For changes in reporting units, we reassign goodwill using a relative fair value allocation approach. We test our goodwill for impairment at least annually, or more frequently if events or changes in circumstances indicate that the asset may be impaired. Goodwill impairments were not material for the periods presented.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Goodwill", + "metadata": { + "length": 524, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "allocate", + "goodwill", + "reporting", + "units", + "based", + "expected", + "benefit", + "business", + "combination", + "evaluate", + "periodically", + "operating", + "segments", + "occur", + "For", + "reassign", + "relative", + "fair", + "allocation", + "approach", + "test", + "impairment", + "annually", + "frequently", + "events", + "circumstances", + "asset", + "impaired", + "Goodwill", + "impairments", + "material", + "periods", + "presented" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "75335eb4-8f37-5475-81f9-f610dc7f561d", + "type": "text", + "content": "We determine if an arrangement is a lease at inception. Our lease agreements generally contain lease and non-lease components. Payments under our lease arrangements are primarily fixed. Non-lease components primarily include payments for maintenance and utilities. We combine fixed payments for non-lease components with lease payments and account for them together as a single lease component which increases the amount of our lease assets and liabilities.\nCertain lease agreements contain variable payments, which are expensed as incurred and not included in the lease assets and liabilities. These amounts primarily include payments affected by the Consumer Price Index, and payments for maintenance and utilities.\nLease assets and liabilities are recognized at the present value of the future lease payments at the lease commencement date. The interest rate used to determine the present value of the future lease payments is our incremental borrowing rate, because the interest rate implicit in our leases is not readily determinable. Our incremental borrowing rate is estimated to approximate the interest rate on a collateralized basis with similar terms and payments, and in economic environments where the leased asset is located. Our lease terms and payments include periods under options to purchase, extend, or terminate the lease when it is reasonably certain that we will exercise that option. We generally use the base, non-cancelable, lease term when determining the lease assets and liabilities. Lease assets also include any prepaid lease payments and lease incentives.\nThe current portion of our operating lease liabilities is included in accrued expenses and other current liabilities, and the long-term portion is included in operating lease liabilities. Finance lease assets are included in property and equipment, net. Finance lease liabilities are included in accrued expenses and other current liabilities or other long-term liabilities.\nOperating lease expense (excluding variable lease costs) is recognized on a straight-line basis over the lease term. Finance lease expense is recognized on a straight-line basis over the shorter of the lease term or the useful life of the asset, and interest expense is recognized based on the incremental borrowing rate.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Leases", + "metadata": { + "length": 2284, + "summary": "The passage explains how lease arrangements are classified and accounted for. Lease and non-lease components are combined into a single lease component for fixed payments, increasing lease assets and liabilities. Variable payments are expensed as incurred. Lease assets and liabilities are recognized at present value using the incremental borrowing rate. Lease terms include options reasonably certain to be exercised. Operating lease expense is recognized straight-line, while finance lease expense is also straight-line with separate interest expense.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "determine", + "arrangement", + "lease", + "inception", + "Our", + "agreements", + "generally", + "components", + "Payments", + "arrangements", + "primarily", + "fixed", + "Non", + "include", + "payments", + "maintenance", + "utilities", + "combine", + "account", + "single", + "component", + "increases", + "amount", + "assets", + "liabilities", + "Certain", + "variable", + "expensed", + "incurred", + "included", + "These", + "amounts", + "affected", + "Consumer", + "Price", + "Index", + "Lease", + "recognized", + "present", + "future", + "commencement", + "date", + "The", + "interest", + "rate", + "incremental", + "borrowing", + "implicit", + "leases", + "readily", + "determinable", + "estimated", + "approximate", + "collateralized", + "basis", + "similar", + "terms", + "economic", + "environments", + "leased", + "asset", + "located", + "periods", + "options", + "purchase", + "extend", + "terminate", + "exercise", + "option", + "base", + "cancelable", + "term", + "determining", + "prepaid", + "incentives", + "current", + "portion", + "operating", + "accrued", + "expenses", + "long", + "Finance", + "property", + "equipment", + "net", + "Operating", + "expense", + "excluding", + "costs", + "straight", + "line", + "shorter", + "life", + "based" + ], + "keywords": [ + "Lease", + "Assets", + "Liabilities" + ], + "connect_to": [] + } + }, + { + "chunk_id": "5cb85c6e-25f2-5627-9108-e9bec62cce91", + "type": "text", + "content": "We review leases, property and equipment, and intangible assets, excluding goodwill, for impairment when events or changes in circumstances indicate the carrying amount may not be recoverable. The evaluation is performed at the lowest level of identifiable cash flows independent of other assets. We measure recoverability of these assets by comparing the carrying amounts to the future undiscounted cash flows that the assets or the asset group are expected to generate. If the carrying value of the assets or asset group is not recoverable, the impairment recognized is measured as the amount by which the carrying value exceeds its fair value.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Impairment of Long-Lived Assets", + "metadata": { + "length": 646, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "review", + "leases", + "property", + "equipment", + "intangible", + "assets", + "excluding", + "goodwill", + "impairment", + "events", + "circumstances", + "carrying", + "amount", + "recoverable", + "The", + "evaluation", + "performed", + "lowest", + "level", + "identifiable", + "cash", + "flows", + "independent", + "measure", + "recoverability", + "comparing", + "amounts", + "future", + "undiscounted", + "asset", + "group", + "expected", + "generate", + "If", + "recognized", + "measured", + "exceeds", + "fair" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "78531d53-fbe4-567b-90b3-c6c083b0ea2d", + "type": "text", + "content": "We account for income taxes using the asset and liability method, under which we recognize the amount of taxes payable or refundable for the current year and deferred tax assets and liabilities for the future tax consequences of events that have been recognized in our financial statements or tax returns. We measure current and deferred tax assets and liabilities based on provisions of enacted tax law. We evaluate the likelihood of future realization of our deferred tax assets based on all available evidence and establish a valuation allowance to reduce deferred tax assets when it is more likely than not that they will not be realized or release a valuation allowance to increase deferred tax assets when it is more likely than not that they will be realized. We have elected to account for the tax effects of the global intangible low tax income provision as a current period expense.\nWe recognize the financial statement effects of a tax position when it is more likely than not that, based on technical merits, the position will be sustained upon examination. The tax benefits of the position recognized in the financial statements are then measured based on the largest amount of benefit that is greater than 50% likely to be realized upon settlement with a taxing authority. In addition, we recognize interest and penalties related to unrecognized tax benefits as a component of the income tax provision.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Income Taxes", + "metadata": { + "length": 1416, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "account", + "income", + "taxes", + "asset", + "liability", + "method", + "recognize", + "amount", + "payable", + "refundable", + "current", + "year", + "deferred", + "tax", + "assets", + "liabilities", + "future", + "consequences", + "events", + "recognized", + "financial", + "statements", + "returns", + "measure", + "based", + "provisions", + "enacted", + "law", + "evaluate", + "likelihood", + "realization", + "evidence", + "establish", + "valuation", + "allowance", + "reduce", + "realized", + "release", + "increase", + "elected", + "effects", + "global", + "intangible", + "low", + "provision", + "period", + "expense", + "statement", + "position", + "technical", + "merits", + "sustained", + "examination", + "The", + "benefits", + "measured", + "largest", + "benefit", + "greater", + "50%", + "settlement", + "taxing", + "authority", + "In", + "addition", + "interest", + "penalties", + "related", + "unrecognized", + "component" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9f353c5f-b74b-59f2-ad13-b9214cbb950c", + "type": "text", + "content": "We include the results of operations of the businesses that we acquire as of the acquisition date. We allocate the purchase price of the acquisitions to the assets acquired and liabilities assumed based on their estimated fair values, except for revenue contracts acquired, which are recognized in accordance with our revenue recognition policy. The excess of the purchase price over the fair values of identifiable assets and liabilities is recorded as goodwill. Acquisition-related expenses are recognized separately from the business combination and are expensed as incurred.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Business Combinations", + "metadata": { + "length": 578, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "include", + "results", + "operations", + "businesses", + "acquire", + "acquisition", + "date", + "allocate", + "purchase", + "price", + "acquisitions", + "assets", + "acquired", + "liabilities", + "assumed", + "based", + "estimated", + "fair", + "values", + "revenue", + "contracts", + "recognized", + "accordance", + "recognition", + "policy", + "The", + "excess", + "identifiable", + "recorded", + "goodwill", + "Acquisition", + "related", + "expenses", + "separately", + "business", + "combination", + "expensed", + "incurred" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a38540c2-3cde-57e2-b9ec-d4c4a20c5bc9", + "type": "text", + "content": "We translate the financial statements of our international subsidiaries to US dollars using month-end exchange rates for assets and liabilities, and average rates for the period derived from month-end exchange rates for revenues, costs, and expenses. We record translation gains and losses in AOCI as a component of stockholders' equity. We reflect net foreign exchange transaction gains and losses resulting from the conversion of the transaction currency to functional currency as a component of foreign currency exchange gain (loss) in OI&E.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Foreign Currency", + "metadata": { + "length": 544, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "translate", + "financial", + "statements", + "international", + "subsidiaries", + "US", + "dollars", + "month", + "end", + "exchange", + "rates", + "assets", + "liabilities", + "average", + "period", + "derived", + "revenues", + "costs", + "expenses", + "record", + "translation", + "gains", + "losses", + "AOCI", + "component", + "stockholders", + "equity", + "reflect", + "net", + "foreign", + "transaction", + "resulting", + "conversion", + "currency", + "functional", + "gain", + "loss", + "OI" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "99f2f366-4949-549e-a261-aa1aa17aa804", + "type": "text", + "content": "In November 2024, the Financial Accounting Standards Board (FASB) issued Accounting Standards Update (ASU) 2024-03 \"Income Statement: Reporting Comprehensive Income-Expense Disaggregation Disclosures (Subtopic 220-40)\" to improve the disclosures about an entity's expenses. Upon adoption, we will be required to disclose in the notes to the financial statements a disaggregation of certain expense categories included within the relevant expense captions on the consolidated statements of income. The standard is effective for our 2027 annual period, and our interim periods beginning in 2028, with early adoption permitted. The standard can be applied either prospectively or retrospectively. We are currently assessing adoption timing, the method of adoption, and the effect that the updated standard will have on our financial statement disclosures.\nIn September 2025, the FASB issued ASU 2025-06 \"Intangibles: Goodwill and Other–Internal-Use Software (Subtopic 350-40): Targeted Improvements to the Accounting for Internal-Use Software\" to modernize the accounting for software costs under Subtopic 350-40, Intangibles–Goodwill and Other–Internal-Use Software (referred to as \"internal-use software\"). Upon adoption, we will be required to account for internal-use software under the updated capitalization criteria. The standard is effective for our interim and annual 2028 periods, with early adoption permitted. The standard can be applied either prospectively, retrospectively, or under a modified transition approach. We are currently assessing adoption timing, the method of adoption, and the effect that the updated standard will have on our consolidated financial statements.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Recently Issued Accounting Pronouncements Not Yet Adopted", + "metadata": { + "length": 1687, + "summary": "The passage discusses two recent FASB Accounting Standards Updates. ASU 2024-03, issued in November 2024, requires disaggregation of certain expense categories in financial statement notes, effective for annual 2027 and interim 2028 periods. ASU 2025-06, issued in September 2025, modernizes accounting for internal-use software costs, effective for interim and annual 2028 periods. Both standards permit early adoption and various transition methods. The entity is currently assessing adoption timing, method, and impact on financial statements.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "In", + "November", + "2024", + "Financial", + "Accounting", + "Standards", + "Board", + "FASB", + "issued", + "Update", + "ASU", + "03", + "Income", + "Statement", + "Reporting", + "Comprehensive", + "Expense", + "Disaggregation", + "Disclosures", + "Subtopic", + "220", + "40", + "improve", + "disclosures", + "entity", + "expenses", + "Upon", + "adoption", + "required", + "disclose", + "notes", + "financial", + "statements", + "disaggregation", + "expense", + "categories", + "included", + "relevant", + "captions", + "consolidated", + "income", + "The", + "standard", + "effective", + "2027", + "annual", + "period", + "interim", + "periods", + "beginning", + "2028", + "early", + "permitted", + "applied", + "prospectively", + "retrospectively", + "We", + "assessing", + "timing", + "method", + "effect", + "updated", + "statement", + "September", + "2025", + "06", + "Intangibles", + "Goodwill", + "Other", + "Internal", + "Use", + "Software", + "350", + "Targeted", + "Improvements", + "modernize", + "accounting", + "software", + "costs", + "referred", + "internal", + "account", + "capitalization", + "criteria", + "modified", + "transition", + "approach" + ], + "keywords": [ + "FASB", + "ASU", + "Disclosures" + ], + "connect_to": [] + } + }, + { + "chunk_id": "debc154e-4e56-50c5-a4bf-a99cc4ef25a7", + "type": "text", + "content": "In December 2023, the FASB issued ASU 2023-09 \"Income Taxes (Topics 740): Improvements to Income Tax Disclosures\" which expands the disclosure requirements for income taxes. We adopted this ASU for our 2025 annual period with the comparative periods updated to reflect additional disclosures. See Note 14 for the revised disclosures consistent with the new standard.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Recently Adopted Accounting Pronouncements", + "metadata": { + "length": 366, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "In", + "December", + "2023", + "FASB", + "issued", + "ASU", + "09", + "Income", + "Taxes", + "Topics", + "740", + "Improvements", + "Tax", + "Disclosures", + "expands", + "disclosure", + "requirements", + "income", + "taxes", + "We", + "adopted", + "2025", + "annual", + "period", + "comparative", + "periods", + "updated", + "reflect", + "additional", + "disclosures", + "See", + "Note", + "14", + "revised", + "consistent", + "standard" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a63f53a5-5e6e-58e8-8711-a289a7736b21", + "type": "text", + "content": "Certain amounts in prior periods have been reclassified to conform with current period presentation.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Prior Period Reclassifications", + "metadata": { + "length": 100, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Certain", + "amounts", + "prior", + "periods", + "reclassified", + "conform", + "current", + "period", + "presentation" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8f28f761-3ee9-54e3-9f0c-d00b17c2a32c", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 2. Revenues", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f7100cd2-7c6b-505a-83b2-80acc6ffa12a", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Google Search & other$ 175,033$ 198,084$ 224,532
YouTube ads31,51036,14740,367
Google Network31,31230,35929,792
Google advertising237,855264,590294,691
Google subscriptions, platforms, and devices34,68840,34048,030
Google Services total272,543304,930342,721
Google Cloud33,08843,22958,705
Other Bets1,5271,6481,537
Hedging gains (losses)236211(127)
Total revenues$ 307,394$ 350,018$ 402,836
", + "path": "tables/table-26 Google Revenue.html", + "metadata": { + "length": 972, + "summary": "table-27\nGoogle's total revenues grew from $307.4B in 2023 to $402.8B in 2025, driven by Search, YouTube, and Cloud.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-26 Google Revenue.html", + "keywords": [ + "Google", + "Revenue", + "2023-2025" + ], + "tokens": [] + } + }, + { + "chunk_id": "4598bb76-0e0d-5c14-8433-2c216f59a649", + "type": "table", + "content": "
Year Ended December 31,
202320242025
United States$ 146,28647 %$ 170,44749 %$ 194,22948 %
$EMEA^{(1)}$ 91,03830102,12729117,15229
$APAC^{(1)}$ 51,5141756,8151667,68017
Other $Americas^{(1)}$ 18,320620,418623,9026
Hedging gains (losses)23602110(127)0
Total revenues$ 307,394100 %$ 350,018100 %$ 402,836100 %
", + "path": "tables/table-27 Revenue Breakdown.html", + "metadata": { + "length": 872, + "summary": "table-28\nThe table shows total revenues by region for 2023-2025, with US leading at 48-49%, followed by EMEA, APAC, and Other Americas, including hedging gains.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-27 Revenue Breakdown.html", + "keywords": [ + "Revenue", + "Region", + "Year" + ], + "tokens": [] + } + }, + { + "chunk_id": "9a96ae1f-d96f-549e-b59b-49fd61ae4af9", + "type": "text", + "content": "The following table presents revenues disaggregated by type (in millions):\n\n[tables/table-26 Google Revenue.html]\n\nNo individual customer or groups of affiliated customers represented more than 10% of our revenues in 2023, 2024, or 2025.\nThe following table presents revenues disaggregated by geography, based on the addresses of our customers (in millions):\n\n[tables/table-27 Revenue Breakdown.html]\n\n(1) Regions represent Europe, the Middle East, and Africa (EMEA); Asia-Pacific (APAC); and Canada and Latin America (\"Other Americas\").", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 2. Revenues/Disaggregated Revenues", + "metadata": { + "length": 552, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "presents", + "revenues", + "disaggregated", + "type", + "millions", + "No", + "individual", + "customer", + "groups", + "affiliated", + "customers", + "represented", + "10%", + "2023", + "2024", + "2025", + "geography", + "based", + "addresses", + "Regions", + "represent", + "Europe", + "Middle", + "East", + "Africa", + "EMEA", + "Asia", + "Pacific", + "APAC", + "Canada", + "Latin", + "America", + "Other", + "Americas" + ], + "keywords": [], + "connect_to": [ + { + "target": "f7100cd2-7c6b-505a-83b2-80acc6ffa12a", + "relation": "embeds", + "ref": "[tables/table-26 Google Revenue.html]", + "position": { + "start": 76, + "end": 113 + } + }, + { + "target": "4598bb76-0e0d-5c14-8433-2c216f59a649", + "relation": "embeds", + "ref": "[tables/table-27 Revenue Breakdown.html]", + "position": { + "start": 360, + "end": 400 + } + } + ] + } + }, + { + "chunk_id": "51c1c3ab-da4f-5840-a21b-0370f2211cac", + "type": "text", + "content": "As of December 31, 2025, we had \\$242.8 billion of remaining performance obligations (\"revenue backlog\"), primarily related to Google Cloud. Revenue backlog represents commitments in customer contracts that have not yet been recognized as revenue. We expect to recognize just over 50% of the revenue backlog as revenues over the next 24 months with the remainder to be recognized thereafter. The estimated revenue backlog and timing of revenue recognition for these commitments is largely driven by contract duration, our ability to deliver in accordance with relevant contract terms, and when our customers utilize services. Revenue backlog includes related deferred revenue currently recorded as well as amounts that will be invoiced in future periods, and excludes contracts with an original expected term of one year or less and cancellable contracts.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 2. Revenues/Revenue Backlog", + "metadata": { + "length": 855, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "As", + "December", + "31", + "2025", + "242.8", + "billion", + "remaining", + "performance", + "obligations", + "revenue", + "backlog", + "primarily", + "related", + "Google", + "Cloud", + "Revenue", + "represents", + "commitments", + "customer", + "contracts", + "recognized", + "We", + "expect", + "recognize", + "50%", + "revenues", + "24", + "months", + "remainder", + "The", + "estimated", + "timing", + "recognition", + "largely", + "driven", + "contract", + "duration", + "ability", + "deliver", + "accordance", + "relevant", + "terms", + "customers", + "utilize", + "services", + "includes", + "deferred", + "recorded", + "amounts", + "invoiced", + "future", + "periods", + "excludes", + "original", + "expected", + "term", + "year", + "cancellable" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "73eacb3a-b71c-51cc-872d-e8afc7e7faf0", + "type": "text", + "content": "We record deferred revenues when cash payments are received or due in advance of our performance, including amounts which are refundable. Deferred revenues primarily relate to Google Cloud and Google subscriptions, platforms, and devices. Total deferred revenue as of December 31, 2024 was \\$6.0 billion, of which \\$4.6 billion was recognized as revenues for the year ended December 31, 2025. Total deferred revenue as of December 31, 2025 was \\$8.6 billion.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 2. Revenues/Deferred Revenues", + "metadata": { + "length": 458, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "record", + "deferred", + "revenues", + "cash", + "payments", + "received", + "due", + "advance", + "performance", + "including", + "amounts", + "refundable", + "Deferred", + "primarily", + "relate", + "Google", + "Cloud", + "subscriptions", + "platforms", + "devices", + "Total", + "revenue", + "December", + "31", + "2024", + "6.0", + "billion", + "4.6", + "recognized", + "year", + "ended", + "2025", + "8.6" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e320c229-571c-58bc-8fe0-69f5197e91fd", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8ec98047-1413-546f-84ae-32a037bb520e", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Fair Value Measurements", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a2c15371-b763-5e46-89a5-01a8e8108296", + "type": "table", + "content": "
As of December 31, 2024
Quoted Prices in Active Markets for Identical Assets (Level 1)Significant Other Observable Inputs (Level 2)Total
Cash$ 12,407
Cash equivalents:
Money market funds$ 8,154$ 0$ 8,154
Time deposits02,0812,081
Government bonds0746746
Corporate debt securities07878
Total cash and cash equivalents8,1542,90523,466
Marketable securities:
Marketable equity $securities^{(1)}$ 4,7081054,813
Time deposits0136136
Government bonds028,70928,709
Corporate debt securities021,11621,116
Mortgage-backed and asset-backed securities017,41717,417
Total marketable securities4,70867,48372,191
Total$ 12,862$ 70,388$ 95,657
", + "path": "tables/table-28 Fair Value Table.html", + "metadata": { + "length": 1389, + "summary": "table-29\nThe table presents the fair value hierarchy of cash, cash equivalents, and marketable securities as of December 31, 2024, totaling $95,657, with Level 1 and Level 2 classifications.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-28 Fair Value Table.html", + "keywords": [ + "fair value", + "cash equivalents", + "marketable securities" + ], + "tokens": [] + } + }, + { + "chunk_id": "e5735ebc-e049-5b2e-92eb-ad1c7594d541", + "type": "table", + "content": "
As of December 31, 2025
Quoted Prices in Active Markets for Identical Assets (Level 1)Significant Other Observable Inputs (Level 2)Total
Cash$ 15,305
Cash equivalents:
Money market funds$ 11,349$ 0$ 11,349
Time deposits03,3533,353
Government bonds0602602
Corporate debt securities09999
Total cash and cash equivalents11,3494,05430,708
Marketable securities:
Marketable equity securities4,4021,9116,313
Time deposits000
Government bonds050,54950,549
Corporate debt securities021,56521,565
Mortgage-backed and asset-backed securities017,70817,708
Total marketable securities4,40291,73396,135
Total$ 15,751$ 95,787$ 126,843
", + "path": "tables/table-29 Fair Value Table.html", + "metadata": { + "length": 1381, + "summary": "table-30\nThe table shows fair value measurements of cash, cash equivalents, and marketable securities as of December 31, 2025, totaling $126,843, with Level 1 and Level 2 inputs.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-29 Fair Value Table.html", + "keywords": [ + "fair value", + "cash equivalents", + "marketable securities" + ], + "tokens": [] + } + }, + { + "chunk_id": "d2ae6e55-f115-5c69-a201-f18c966e3416", + "type": "text", + "content": "Cash equivalents and marketable equity securities are measured at fair value and classified within Level 1 and Level 2 in the fair value hierarchy, because we use quoted prices for identical assets in active markets or inputs that are based upon quoted prices for similar instruments in active markets.\nDebt securities are measured at fair value and classified within Level 2 in the fair value hierarchy, because we use quoted market prices to the extent available or alternative pricing sources and models utilizing market observable inputs to determine fair value.\nThe following tables summarize our cash, cash equivalents, and marketable securities measured at fair value on a recurring basis (in millions):\n\n[tables/table-28 Fair Value Table.html]\n\n(1) The long-term portion of marketable equity securities (subject to long-term lock-up restrictions) of \\$266 million as of December 31, 2024 is included within other non-current assets.\n\n[tables/table-29 Fair Value Table.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Fair Value Measurements/Investments Measured at Fair Value on a Recurring Basis", + "metadata": { + "length": 995, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Cash", + "equivalents", + "marketable", + "equity", + "securities", + "measured", + "fair", + "classified", + "Level", + "hierarchy", + "quoted", + "prices", + "identical", + "assets", + "active", + "markets", + "inputs", + "based", + "similar", + "instruments", + "Debt", + "market", + "extent", + "alternative", + "pricing", + "sources", + "models", + "utilizing", + "observable", + "determine", + "The", + "tables", + "summarize", + "cash", + "recurring", + "basis", + "millions", + "long", + "term", + "portion", + "subject", + "lock", + "restrictions", + "266", + "million", + "December", + "31", + "2024", + "included", + "current" + ], + "keywords": [], + "connect_to": [ + { + "target": "a2c15371-b763-5e46-89a5-01a8e8108296", + "relation": "embeds", + "ref": "[tables/table-28 Fair Value Table.html]", + "position": { + "start": 712, + "end": 751 + } + }, + { + "target": "e5735ebc-e049-5b2e-92eb-ad1c7594d541", + "relation": "embeds", + "ref": "[tables/table-29 Fair Value Table.html]", + "position": { + "start": 942, + "end": 981 + } + } + ] + } + }, + { + "chunk_id": "3732a340-67a9-57d7-be74-f8992f5a1fe2", + "type": "text", + "content": "Non-marketable equity securities accounted for under the measurement alternative are investments in privately held companies without readily determinable market values. The carrying value of these non-marketable equity securities is adjusted upward or downward to fair value upon observable transactions for identical or similar investments of the same issuer or impairment. Non-marketable equity securities that have been remeasured during the period based on observable transactions are classified within Level 2 or Level 3 in the fair value hierarchy, and remeasurements due to impairment are classified within Level 3. Our valuation methods include option pricing models, market comparable approach, and common stock equivalent method, which may include a combination of the observable transaction price at the transaction date and other unobservable inputs including volatility, expected time to exit, risk free rate, and the rights and obligations of the securities we hold. These inputs vary significantly based on investment type.\nAs of December 31, 2025, the carrying value of our non-marketable equity securities accounted for under the measurement alternative was \\$64.1 billion, of which \\$45.6 billion were remeasured at fair value during the year ended December 31, 2025, and were primarily classified within Level 2 of the fair value hierarchy at the time of measurement.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Fair Value Measurements/Investments Measured at Fair Value on a Nonrecurring Basis", + "metadata": { + "length": 1386, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Non", + "marketable", + "equity", + "securities", + "accounted", + "measurement", + "alternative", + "investments", + "privately", + "held", + "companies", + "readily", + "determinable", + "market", + "values", + "The", + "carrying", + "adjusted", + "upward", + "downward", + "fair", + "observable", + "transactions", + "identical", + "similar", + "issuer", + "impairment", + "remeasured", + "period", + "based", + "classified", + "Level", + "hierarchy", + "remeasurements", + "due", + "Our", + "valuation", + "methods", + "include", + "option", + "pricing", + "models", + "comparable", + "approach", + "common", + "stock", + "equivalent", + "method", + "combination", + "transaction", + "price", + "date", + "unobservable", + "inputs", + "including", + "volatility", + "expected", + "time", + "exit", + "risk", + "free", + "rate", + "rights", + "obligations", + "hold", + "These", + "vary", + "significantly", + "investment", + "type", + "As", + "December", + "31", + "2025", + "64.1", + "billion", + "45.6", + "year", + "ended", + "primarily" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e419e364-6208-5c1f-b4ef-053c7ffad219", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ca65f157-5068-54ea-b1aa-3d384330270c", + "type": "table", + "content": "
As of December 31, 2025
Due in 1 year or less$ 26,735
Due in 1 year through 5 years37,001
Due in 5 years through 10 years12,769
Due after 10 years13,317
Total$ 89,822
", + "path": "tables/table-30 Debt Maturity.html", + "metadata": { + "length": 338, + "summary": "table-31\nThe table shows debt due by maturity: $26,735 within 1 year, $37,001 in 1-5 years, $12,769 in 5-10 years, $13,317 after 10 years, totaling $89,822 as of December 31, 2025.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-30 Debt Maturity.html", + "keywords": [ + "Debt", + "Maturity", + "Schedule" + ], + "tokens": [] + } + }, + { + "chunk_id": "1e2f785d-54c4-533d-8a15-387e130765bb", + "type": "table", + "content": "
As of December 31, 2024
Adjusted CostGross Unrealized GainsGross Unrealized LossesFair Value
Time deposits$ 2,217$ 0$ 0$ 2,217
Government bonds27,55183(214)27,420
Corporate debt securities18,30079(222)18,157
Mortgage-backed and asset-backed securities14,43763(385)14,115
Total investments with fair value change reflected in other comprehensive income$ 62,505$ 225$ (821)$ 61,909
", + "path": "tables/table-31 Investment Summary.html", + "metadata": { + "length": 754, + "summary": "table-32\nThe table shows investments as of Dec 31, 2024, with total adjusted cost of $62,505, fair value of $61,909, and net unrealized losses of $596 across time deposits, government bonds, corporate debt, and mortgage-backed securities.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-31 Investment Summary.html", + "keywords": [ + "Investments", + "Fair Value", + "Unrealized Gains" + ], + "tokens": [] + } + }, + { + "chunk_id": "bc1334b5-4a0e-5406-98fb-40c9a6dac277", + "type": "table", + "content": "
As of December 31, 2025
Adjusted CostGross Unrealized GainsGross Unrealized LossesFair Value
Time deposits$ 3,353$ 0$ 0$ 3,353
Government bonds49,087443(26)49,504
Corporate debt securities18,346242(32)18,556
Mortgage-backed and asset-backed securities14,337174(128)14,383
Total investments with fair value change reflected in other comprehensive income$ 85,123$ 859$ (186)$ 85,796
", + "path": "tables/table-32 Investment Summary.html", + "metadata": { + "length": 755, + "summary": "table-33\nThe table shows investments as of Dec 31, 2025, with total adjusted cost of $85,123, gross unrealized gains of $859, losses of $186, and fair value of $85,796 across time deposits, government bonds, corporate debt, and mortgage-backed securities.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-32 Investment Summary.html", + "keywords": [ + "Investments", + "Fair Value", + "Unrealized Gains" + ], + "tokens": [] + } + }, + { + "chunk_id": "8171126d-2e6e-596e-93c2-c3f671df6c8d", + "type": "table", + "content": "
As of December 31, 2024
Less than 12 Months12 Months or GreaterTotal
Fair ValueUnrealized LossFair ValueUnrealized LossFair ValueUnrealized Loss
Government bonds$ 11,119$ (126)$ 2,576$ (88)$ 13,695$ (214)
Corporate debt securities4,228(17)6,838(168)11,066(185)
Mortgage-backed and asset-backed securities5,222(106)3,813(279)9,035(385)
Total$ 20,569$ (249)$ 13,227$ (535)$ 33,796$ (784)
", + "path": "tables/table-33 Investment Securities.html", + "metadata": { + "length": 868, + "summary": "table-34\nThe table shows fair value and unrealized losses for government bonds, corporate debt, and mortgage-backed securities as of December 31, 2024, totaling $33,796 and $(784) respectively.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-33 Investment Securities.html", + "keywords": [ + "Fair Value", + "Unrealized Loss", + "Securities" + ], + "tokens": [] + } + }, + { + "chunk_id": "58628132-a0a6-534c-ac13-d31792ae5b6b", + "type": "table", + "content": "
As of December 31, 2025
Less than 12 Months12 Months or GreaterTotal
Fair ValueUnrealized LossFair ValueUnrealized LossFair ValueUnrealized Loss
Government bonds$ 4,230$ (9)$ 1,174$ (17)$ 5,404$ (26)
Corporate debt securities91502,429(24)3,344(24)
Mortgage-backed and asset-backed securities1,377(4)3,035(124)4,412(128)
Total$ 6,522$ (13)$ 6,638$ (165)$ 13,160$ (178)
", + "path": "tables/table-34 Investment Securities.html", + "metadata": { + "length": 850, + "summary": "table-35\nThe table shows fair value and unrealized losses for government bonds, corporate debt, and mortgage-backed securities as of Dec 31, 2025, totaling $13,160 with $178 unrealized loss.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-34 Investment Securities.html", + "keywords": [ + "Fair Value", + "Unrealized Loss", + "Debt Securities" + ], + "tokens": [] + } + }, + { + "chunk_id": "a423fc68-eb8e-5682-bb00-0a9a800def71", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Unrealized gain (loss) on fair value option debt securities$ 386$ 30$ 254
Gross realized gain on debt securities182482572
Gross realized loss on debt securities(1,833)(1,553)(316)
(Increase) decrease in allowance for credit losses50(2)30
Total gain (loss) on debt securities recognized in other income (expense), net$ (1,215)$ (1,043)$ 540
", + "path": "tables/table-35 Debt Securities P&L.html", + "metadata": { + "length": 697, + "summary": "table-36\nThe table shows unrealized gains/losses, realized gains/losses, and credit loss allowances for debt securities from 2023 to 2025, with total net gains/losses of -$1,215, -$1,043, and $540 respectively.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-35 Debt Securities P&L.html", + "keywords": [ + "debt securities", + "gains", + "losses" + ], + "tokens": [] + } + }, + { + "chunk_id": "9e11d899-922a-56e5-a3d9-3313bde9b1d9", + "type": "text", + "content": "The following table summarizes the estimated fair value of investments in available-for-sale marketable debt securities by effective contractual maturity dates (in millions):\n\n[tables/table-30 Debt Maturity.html]\n\nThe following tables present fair values and gross unrealized gains and losses recorded to AOCI, less any expected credit losses, aggregated by investment category (in millions):\n\n[tables/table-31 Investment Summary.html]\n\n\n[tables/table-32 Investment Summary.html]\n\nThe following tables present fair values and gross unrealized losses recorded to AOCI, aggregated by investment category and the length of time that individual securities have been in a continuous loss position (in millions):\n\n[tables/table-33 Investment Securities.html]\n\n\n[tables/table-34 Investment Securities.html]\n\nWe determine realized gains or losses on the sale or extinguishment of debt securities on a specific identification method. For certain marketable debt securities, we have elected the fair value option for which changes in fair value are recorded in OI&E. The fair value option was elected for these securities to align with the unrealized gains and losses from related derivative contracts.\nThe following table summarizes gains and losses for debt securities, reflected as a component of OI&E (in millions):\n\n[tables/table-35 Debt Securities P&L.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Debt Securities", + "metadata": { + "length": 1381, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "summarizes", + "estimated", + "fair", + "investments", + "sale", + "marketable", + "debt", + "securities", + "effective", + "contractual", + "maturity", + "dates", + "millions", + "tables", + "present", + "values", + "gross", + "unrealized", + "gains", + "losses", + "recorded", + "AOCI", + "expected", + "credit", + "aggregated", + "investment", + "category", + "length", + "time", + "individual", + "continuous", + "loss", + "position", + "We", + "determine", + "realized", + "extinguishment", + "specific", + "identification", + "method", + "For", + "elected", + "option", + "OI", + "align", + "related", + "derivative", + "contracts", + "reflected", + "component" + ], + "keywords": [], + "connect_to": [ + { + "target": "ca65f157-5068-54ea-b1aa-3d384330270c", + "relation": "embeds", + "ref": "[tables/table-30 Debt Maturity.html]", + "position": { + "start": 176, + "end": 212 + } + }, + { + "target": "1e2f785d-54c4-533d-8a15-387e130765bb", + "relation": "embeds", + "ref": "[tables/table-31 Investment Summary.html]", + "position": { + "start": 394, + "end": 435 + } + }, + { + "target": "bc1334b5-4a0e-5406-98fb-40c9a6dac277", + "relation": "embeds", + "ref": "[tables/table-32 Investment Summary.html]", + "position": { + "start": 438, + "end": 479 + } + }, + { + "target": "8171126d-2e6e-596e-93c2-c3f671df6c8d", + "relation": "embeds", + "ref": "[tables/table-33 Investment Securities.html]", + "position": { + "start": 708, + "end": 752 + } + }, + { + "target": "58628132-a0a6-534c-ac13-d31792ae5b6b", + "relation": "embeds", + "ref": "[tables/table-34 Investment Securities.html]", + "position": { + "start": 755, + "end": 799 + } + }, + { + "target": "a423fc68-eb8e-5682-bb00-0a9a800def71", + "relation": "embeds", + "ref": "[tables/table-35 Debt Securities P&L.html]", + "position": { + "start": 1311, + "end": 1353 + } + } + ] + } + }, + { + "chunk_id": "bf0c7272-91a1-50eb-b0f6-bde9399063f9", + "type": "table", + "content": "
As of December 31,
20242025
Non-marketable securities:
Total initial cost of non-marketable equity securities accounted for under the measurement alternative$ 20,940$ 28,429
Cumulative upward adjustments22,70944,485
Cumulative downward adjustments (including impairments)(8,431)(8,820)
Carrying value of non-marketable equity securities accounted for under the measurement alternative35,21864,094
Equity method investments and other2,7644,593
Total non-marketable securities$ 37,982$ 68,687
", + "path": "tables/table-36 Non-marketable securities.html", + "metadata": { + "length": 827, + "summary": "table-37\nThe table shows non-marketable securities data for 2024 and 2025, including initial cost, cumulative adjustments, carrying value, equity method investments, and total non-marketable securities.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-36 Non-marketable securities.html", + "keywords": [ + "non-marketable securities", + "equity securities", + "carrying value" + ], + "tokens": [] + } + }, + { + "chunk_id": "4ebf09fd-0bd8-5ce9-80bb-49ec83307335", + "type": "text", + "content": "Our non-marketable securities primarily consist of non-marketable equity securities accounted for under the measurement alternative. The carrying value is measured at the total initial cost plus the cumulative net upward and downward adjustments (including impairments). We account for non-marketable equity securities through which we exercise significant influence but do not have control over the investee under the equity method. Certain of our non-marketable securities include our investments in VIEs where we are not the primary beneficiary. See Note 5 for further details on VIEs.\nRealized net gain (loss) on equity securities sold during the period reflects the difference between the sale proceeds and the carrying value of the equity securities at the beginning of the period or the purchase date, if later.\nAll gains and losses, including impairments, are included as components of OI&E.\nThe carrying values for non-marketable securities are summarized below (in millions):\n\n[tables/table-36 Non-marketable securities.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Non-marketable Securities", + "metadata": { + "length": 1028, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "marketable", + "securities", + "primarily", + "consist", + "equity", + "accounted", + "measurement", + "alternative", + "The", + "carrying", + "measured", + "total", + "initial", + "cost", + "cumulative", + "net", + "upward", + "downward", + "adjustments", + "including", + "impairments", + "We", + "account", + "exercise", + "significant", + "influence", + "control", + "investee", + "method", + "Certain", + "include", + "investments", + "VIEs", + "primary", + "beneficiary", + "See", + "Note", + "details", + "Realized", + "gain", + "loss", + "sold", + "period", + "reflects", + "difference", + "sale", + "proceeds", + "beginning", + "purchase", + "date", + "All", + "gains", + "losses", + "included", + "components", + "OI", + "values", + "summarized", + "millions" + ], + "keywords": [], + "connect_to": [ + { + "target": "bf0c7272-91a1-50eb-b0f6-bde9399063f9", + "relation": "embeds", + "ref": "[tables/table-36 Non-marketable securities.html]", + "position": { + "start": 987, + "end": 1035 + } + } + ] + } + }, + { + "chunk_id": "955a097a-2011-5988-9405-f5336a7e2488", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Gross unrealized gain on non-marketable equity securities accounted for under the measurement alternative$ 1,806$ 5,582$ 22,666
Gross unrealized loss (including impairments) on non-marketable equity securities accounted for under the measurement alternative(2,894)(2,210)(1,271)
Unrealized net gain (loss) on non-marketable equity securities accounted for under the measurement alternative(1,088)3,37221,395
Unrealized net gain (loss) on marketable and other equity securities7901561,907
Realized net gain (loss) on marketable and non-marketable equity securities sold during the period690186778
Total gain (loss) on equity securities in other income (expense), net (1)$ 392$ 3,714$ 24,080
", + "path": "tables/table-37 Equity Securities.html", + "metadata": { + "length": 1091, + "summary": "table-38\nThe table shows unrealized gains and losses on non-marketable and marketable equity securities from 2023 to 2025, with total net gains rising from $392 to $24,080.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-37 Equity Securities.html", + "keywords": [ + "unrealized gain", + "equity securities", + "net gain" + ], + "tokens": [] + } + }, + { + "chunk_id": "011d1d79-02f8-5cd6-8fac-78a92d7db421", + "type": "text", + "content": "Gains and losses (including impairments), net, for equity securities included in OI&E are summarized below (in millions):\n\n[tables/table-37 Equity Securities.html]\n\n(1) Excludes income (loss) and impairment from equity method investments. Refer to Note 7 for further details.\nCumulative net gains (losses), calculated as the difference between the sales price and purchase price, represent the total net gains (losses) recognized after the initial purchase date. This represents the total economic impact of the investment, regardless of when the gains or losses were previously recognized. Cumulative net gains on equity securities sold were \\$748 million and \\$387 million for the years ended December 31, 2024 and 2025, respectively.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Gains and Losses on Equity Securities", + "metadata": { + "length": 742, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Gains", + "losses", + "including", + "impairments", + "net", + "equity", + "securities", + "included", + "OI", + "summarized", + "millions", + "Excludes", + "income", + "loss", + "impairment", + "method", + "investments", + "Refer", + "Note", + "details", + "Cumulative", + "gains", + "calculated", + "difference", + "sales", + "price", + "purchase", + "represent", + "total", + "recognized", + "initial", + "date", + "This", + "represents", + "economic", + "impact", + "investment", + "previously", + "sold", + "748", + "million", + "387", + "years", + "ended", + "December", + "31", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [ + { + "target": "955a097a-2011-5988-9405-f5336a7e2488", + "relation": "embeds", + "ref": "[tables/table-37 Equity Securities.html]", + "position": { + "start": 123, + "end": 163 + } + } + ] + } + }, + { + "chunk_id": "06e4d243-d405-5ce5-9fbb-0ba419013abe", + "type": "text", + "content": "We primarily use derivative instruments to manage risks relating to our ongoing business operations, including foreign currencies, interest rates, commodity prices, credit exposures, and market prices of certain marketable equity securities. Additionally, we enter into derivatives to enhance investment returns. We also enter into derivatives as a result of agreements with third parties to backstop certain obligations related to data center leases. These backstop agreements are accounted for as credit derivatives.\nWe recognize derivative instruments in the Consolidated Balance Sheets at fair value and classify them primarily within Level 2 in the fair value hierarchy. We present our foreign currency collars (an option strategy comprised of a combination of purchased and written options) at net fair values and present all other derivatives at gross fair values. The accounting treatment for derivatives is based on the intended use and hedge designation.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments", + "metadata": { + "length": 964, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "primarily", + "derivative", + "instruments", + "manage", + "risks", + "relating", + "ongoing", + "business", + "operations", + "including", + "foreign", + "currencies", + "interest", + "rates", + "commodity", + "prices", + "credit", + "exposures", + "market", + "marketable", + "equity", + "securities", + "Additionally", + "enter", + "derivatives", + "enhance", + "investment", + "returns", + "result", + "agreements", + "parties", + "backstop", + "obligations", + "related", + "data", + "center", + "leases", + "These", + "accounted", + "recognize", + "Consolidated", + "Balance", + "Sheets", + "fair", + "classify", + "Level", + "hierarchy", + "present", + "currency", + "collars", + "option", + "strategy", + "comprised", + "combination", + "purchased", + "written", + "options", + "net", + "values", + "gross", + "The", + "accounting", + "treatment", + "based", + "intended", + "hedge", + "designation" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "04559ea8-772f-5690-9d2e-decdf2312b8c", + "type": "text", + "content": "We designate foreign currency forwards and options (including collars) as cash flow hedges to hedge certain forecasted revenue transactions denominated in currencies other than the US dollar. These contracts have maturities of 24 months or less.\nCash flow hedge amounts included in the assessment of hedge effectiveness are deferred in AOCI and reclassified to revenue when the hedged item is recognized in earnings. Hedge components excluded from our assessment of hedge effectiveness are amortized on a straight-line basis over the life of the hedging instrument in revenues. The difference between fair value changes of the excluded component and the amount amortized to revenues is recorded in AOCI.\nAs of December 31, 2025, the net accumulated loss on our foreign currency cash flow hedges before tax effect was \\$60 million, which is expected to be reclassified from AOCI into revenues within the next 12 months.\nAdditionally, we may designate interest rate derivatives as cash flow hedges to manage our exposure to certain interest rate risks. Changes in the fair value of these derivatives are deferred in AOCI and reclassified to OI&E when the hedged item is recognized in earnings.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments/Cash Flow Hedges", + "metadata": { + "length": 1191, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "designate", + "foreign", + "currency", + "forwards", + "options", + "including", + "collars", + "cash", + "flow", + "hedges", + "hedge", + "forecasted", + "revenue", + "transactions", + "denominated", + "currencies", + "US", + "dollar", + "These", + "contracts", + "maturities", + "24", + "months", + "Cash", + "amounts", + "included", + "assessment", + "effectiveness", + "deferred", + "AOCI", + "reclassified", + "hedged", + "item", + "recognized", + "earnings", + "Hedge", + "components", + "excluded", + "amortized", + "straight", + "line", + "basis", + "life", + "hedging", + "instrument", + "revenues", + "The", + "difference", + "fair", + "component", + "amount", + "recorded", + "As", + "December", + "31", + "2025", + "net", + "accumulated", + "loss", + "tax", + "effect", + "60", + "million", + "expected", + "12", + "Additionally", + "interest", + "rate", + "derivatives", + "manage", + "exposure", + "risks", + "Changes", + "OI" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3e9ab57c-07c2-522f-be3c-53bf2a98b2d7", + "type": "text", + "content": "We designate foreign currency forwards as fair value hedges to hedge foreign currency risks for our marketable debt securities denominated in currencies other than the US dollar. Fair value hedge amounts included and excluded from the assessment of hedge effectiveness are recognized in OI&E.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments/Fair Value Hedges", + "metadata": { + "length": 292, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "designate", + "foreign", + "currency", + "forwards", + "fair", + "hedges", + "hedge", + "risks", + "marketable", + "debt", + "securities", + "denominated", + "currencies", + "US", + "dollar", + "Fair", + "amounts", + "included", + "excluded", + "assessment", + "effectiveness", + "recognized", + "OI" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ddc290d0-3ec6-5969-b1d1-aefc7443426a", + "type": "text", + "content": "We designate foreign currency forwards, options (including collars), cross-currency swaps, and foreign currency-denominated debt as net investment hedges to hedge the foreign currency risks related to our investments in foreign subsidiaries. Net investment hedge amounts included in the assessment of hedge effectiveness are recognized in AOCI.\nChanges in the fair value of hedge components of forward and option contracts that are excluded from the assessment of hedge effectiveness are recognized in OI&E. Hedge components of cross-currency swaps that are excluded from the assessment of hedge effectiveness are amortized over the life of the hedging instrument and recognized in OI&E. The difference between fair value changes of the excluded component and the amount amortized to OI&E is recorded in AOCI.\nWe had no foreign currency-denominated debt as of December 31, 2024 and \\$15.4 billion carrying value of foreign currency-denominated debt designated as net investment hedges as of December 31, 2025.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments/Net Investment Hedges", + "metadata": { + "length": 1009, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "designate", + "foreign", + "currency", + "forwards", + "options", + "including", + "collars", + "cross", + "swaps", + "denominated", + "debt", + "net", + "investment", + "hedges", + "hedge", + "risks", + "related", + "investments", + "subsidiaries", + "Net", + "amounts", + "included", + "assessment", + "effectiveness", + "recognized", + "AOCI", + "Changes", + "fair", + "components", + "forward", + "option", + "contracts", + "excluded", + "OI", + "Hedge", + "amortized", + "life", + "hedging", + "instrument", + "The", + "difference", + "component", + "amount", + "recorded", + "December", + "31", + "2024", + "15.4", + "billion", + "carrying", + "designated", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "49e1229d-c921-56da-b27b-180efba338f0", + "type": "table", + "content": "
As of December 31,
20242025
Derivatives designated as hedging instruments:
Foreign exchange contracts
Cash flow hedges$ 20,315$ 23,852
Fair value hedges$ 1,562$ 0
Net investment hedges$ 6,986$ 14,203
Derivatives not designated as hedging instruments:
Foreign exchange contracts$ 44,227$ 56,085
Credit derivatives(1)$ 0$ 16,940
Other contracts$ 15,082$ 15,900
", + "path": "tables/table-38 Derivatives Table.html", + "metadata": { + "length": 747, + "summary": "table-39\nTable shows derivatives designated and not designated as hedging instruments, including foreign exchange contracts, credit derivatives, and other contracts, with values for 2024 and 2025.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-38 Derivatives Table.html", + "keywords": [ + "derivatives", + "hedging", + "foreign exchange" + ], + "tokens": [] + } + }, + { + "chunk_id": "b933a60d-6dc2-519c-bdf9-8a3ab62bb266", + "type": "table", + "content": "
As of December 31, 2024As of December 31, 2025
Assets(1)Liabilities(2)Assets(1)Liabilities(2)
Derivatives designated as hedging instruments:
Foreign exchange contracts$ 1,054$ 0$ 316$ 197
Derivatives not designated as hedging instruments:
Foreign exchange contracts2005939284
Other contracts4741932498
Total derivatives not designated as hedging instruments674612416182
Total$ 1,728$ 612$ 732$ 379
", + "path": "tables/table-39 Derivatives Table.html", + "metadata": { + "length": 851, + "summary": "table-40\nThe table shows derivative assets and liabilities for 2024 and 2025, including hedging and non-hedging instruments, with totals decreasing from $1,728 to $732 in assets and $612 to $379 in liabilities.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-39 Derivatives Table.html", + "keywords": [ + "derivatives", + "hedging", + "financial" + ], + "tokens": [] + } + }, + { + "chunk_id": "005cb45e-8711-528d-a235-10a015c48732", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Cash flow hedging relationship:
Foreign exchange and other contracts
Amount included in the assessment of effectiveness$ 90$ 857$ (978)
Amount excluded from the assessment of effectiveness8477(45)
Net investment hedging relationship:
Amounts included in the assessment of effectiveness
Foreign exchange contracts(287)223(765)
Foreign currency-denominated debt00(393)
Amounts excluded from the assessment of effectiveness
Foreign exchange contracts0011
Total$ (113)$ 1,157$ (2,170)
", + "path": "tables/table-40 Hedge Gains_Losses.html", + "metadata": { + "length": 1027, + "summary": "table-41\nThe table shows gains and losses from cash flow and net investment hedging relationships for 2023-2025, with totals of -113, 1,157, and -2,170 respectively.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-40 Hedge Gains_Losses.html", + "keywords": [ + "hedging", + "foreign exchange", + "effectiveness" + ], + "tokens": [] + } + }, + { + "chunk_id": "68d1ddb7-2617-5c13-81b3-3f37764c3aad", + "type": "table", + "content": "
Year Ended December 31,
202320242025
RevenuesOther income (expense), netRevenuesOther income (expense), netRevenuesOther income (expense), net
Total amounts included on the Consolidated Statements of Income$ 307,394$ 1,424$ 350,018$ 7,425$ 402,836$ 29,787
Effect of cash flow hedges:
Foreign exchange contracts
Amount included in the assessment of effectiveness21301740(233)0
Amount excluded from the assessment of effectiveness2403701070
Effect of fair value hedges:
Foreign exchange contracts
Hedged items0590(59)0(9)
Amount included in the assessment of effectiveness0(59)05809
Amount excluded from the assessment of effectiveness01501301
Effect of net investment hedges:
Foreign exchange contracts
Amount excluded from the assessment of effectiveness018701370189
Effect of non-designated hedges:
Foreign exchange contracts0703350445
Other contracts05301740(148)
Total gains (losses)$ 237$ 262$ 211$ 658$ (126)$ 487
", + "path": "tables/table-41 Hedge Effects.html", + "metadata": { + "length": 2030, + "summary": "table-42\nThe table shows the effect of cash flow, fair value, net investment, and non-designated hedges on revenues and other income for 2023-2025, with total gains of $237, $211, and -$126 for revenues and $262, $658, and $487 for other income.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-41 Hedge Effects.html", + "keywords": [ + "hedges", + "foreign exchange", + "income" + ], + "tokens": [] + } + }, + { + "chunk_id": "f134386f-f7db-5a36-93f7-e03b6c11ddc2", + "type": "text", + "content": "We enter into derivatives not designated as hedging instruments to manage risks related to our ongoing business operations. The primary risk managed is foreign exchange risk related to the remeasurement of monetary assets or liabilities denominated in currencies other than the functional currency of a subsidiary. Gains and losses on these foreign exchange derivatives are recorded within the “foreign currency exchange gain (loss), net” component of OI&E.\nWe also enter into derivatives to manage other risks, to enhance investment returns, and as a result of agreements with certain third parties to backstop certain obligations relating to data center leases. Gains and losses arising from other derivatives are primarily reflected within the “other” component of OI&E. See Note 7 for further details.\nThe gross notional amounts of outstanding derivative instruments were as follows (in millions):\n\n[tables/table-38 Derivatives Table.html]\n\n(1) Notional amounts for credit derivatives are the backstop obligations related to certain third-party data center leases and represent the maximum potential amount of future payments that could be required in the event of certain default scenarios over remaining agreement periods of up to 15 years. In the event we are required to make payments under certain backstop obligations, we may receive equity in or cash payments from certain counterparties, the amounts for which are not reflected in the notional amounts for credit derivatives. See Note 5 for further details.\nThe fair values of outstanding derivative instruments were as follows (in millions):\n\n[tables/table-39 Derivatives Table.html]\n\n(1) Derivative assets are recorded as other current and non-current assets.\n(2) Derivative liabilities are recorded as accrued expenses and other liabilities, current and non-current.\nThe gains (losses) on derivatives and non-derivative financial instruments in cash flow hedging and net investment hedging relationships recognized in other comprehensive income are summarized below (in millions):\n\n[tables/table-40 Hedge Gains_Losses.html]\n\nThe table below presents the gains (losses) of derivatives included on the Consolidated Statements of Income: (in millions):\n\n[tables/table-41 Hedge Effects.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments/Derivatives Not Designated as Hedging Instruments", + "metadata": { + "length": 2274, + "summary": "The passage describes the company's use of derivatives not designated as hedging instruments to manage risks, primarily foreign exchange risk from remeasurement of monetary assets/liabilities in non-functional currencies. Gains/losses on these are recorded in foreign currency exchange gain/loss. Other derivatives manage risks, enhance returns, or backstop data center lease obligations, with gains/losses in other OI&E. Tables provide gross notional amounts and fair values of outstanding derivatives as of year-end, with notes on credit derivatives' maximum potential payments up to 15 years. Additional tables summarize gains/losses on derivatives in cash flow and net investment hedging relationships recognized in OCI, and gains/losses included in the Consolidated Statements of Income.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "enter", + "derivatives", + "designated", + "hedging", + "instruments", + "manage", + "risks", + "related", + "ongoing", + "business", + "operations", + "The", + "primary", + "risk", + "managed", + "foreign", + "exchange", + "remeasurement", + "monetary", + "assets", + "liabilities", + "denominated", + "currencies", + "functional", + "currency", + "subsidiary", + "Gains", + "losses", + "recorded", + "gain", + "loss", + "net", + "component", + "OI", + "enhance", + "investment", + "returns", + "result", + "agreements", + "parties", + "backstop", + "obligations", + "relating", + "data", + "center", + "leases", + "arising", + "primarily", + "reflected", + "See", + "Note", + "details", + "gross", + "notional", + "amounts", + "outstanding", + "derivative", + "millions", + "Notional", + "credit", + "party", + "represent", + "maximum", + "potential", + "amount", + "future", + "payments", + "required", + "event", + "default", + "scenarios", + "remaining", + "agreement", + "periods", + "15", + "years", + "In", + "make", + "receive", + "equity", + "cash", + "counterparties", + "fair", + "values", + "Derivative", + "current", + "accrued", + "expenses", + "gains", + "financial", + "flow", + "relationships", + "recognized", + "comprehensive", + "income", + "summarized", + "table", + "presents", + "included", + "Consolidated", + "Statements", + "Income" + ], + "keywords": [ + "derivatives", + "foreign exchange", + "risk management" + ], + "connect_to": [ + { + "target": "49e1229d-c921-56da-b27b-180efba338f0", + "relation": "embeds", + "ref": "[tables/table-38 Derivatives Table.html]", + "position": { + "start": 903, + "end": 943 + } + }, + { + "target": "b933a60d-6dc2-519c-bdf9-8a3ab62bb266", + "relation": "embeds", + "ref": "[tables/table-39 Derivatives Table.html]", + "position": { + "start": 1606, + "end": 1646 + } + }, + { + "target": "005cb45e-8711-528d-a235-10a015c48732", + "relation": "embeds", + "ref": "[tables/table-40 Hedge Gains_Losses.html]", + "position": { + "start": 2047, + "end": 2088 + } + }, + { + "target": "68d1ddb7-2617-5c13-81b3-3f37764c3aad", + "relation": "embeds", + "ref": "[tables/table-41 Hedge Effects.html]", + "position": { + "start": 2216, + "end": 2252 + } + } + ] + } + }, + { + "chunk_id": "df0054e0-5df8-5e78-a292-66b970b035b9", + "type": "table", + "content": "
As of December 31, 2024
Gross Amounts RecognizedGross Amounts Offset in the Consolidated Balance SheetsNet Amounts Presented in the Consolidated Balance SheetsFinancial Instruments(1)Cash and Non-Cash Collateral Received or PledgedNet Amounts
Derivatives assets$ 1,776$ (48)$ 1,728$ (516)$ (721)$ 491
Derivatives liabilities$ 660$ (48)$ 612$ (516)$ (9)$ 87
As of December 31, 2025
Gross Amounts RecognizedGross Amounts Offset in the Consolidated Balance SheetsNet Amounts Presented in the Consolidated Balance SheetsFinancial Instruments(1)Cash and Non-Cash Collateral Received or PledgedNet Amounts
Derivatives assets$ 842$ (110)$ 732$ (140)$ (231)$ 361
Derivatives liabilities$ 489$ (110)$ 379$ (140)$ (15)$ 224
", + "path": "tables/table-42 Derivatives Offsetting.html", + "metadata": { + "length": 1237, + "summary": "table-43\nThe table shows derivatives assets and liabilities with gross amounts, offsets, net balances, and collateral for 2024 and 2025. In 2024, net assets were $1,728M and net liabilities $612M; in 2025, net assets were $732M and net liabilities $379M.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-42 Derivatives Offsetting.html", + "keywords": [ + "Derivatives", + "Offsetting", + "Collateral" + ], + "tokens": [] + } + }, + { + "chunk_id": "b0ab3736-7c8b-57b6-aa2c-67afe672951e", + "type": "text", + "content": "We enter into master netting arrangements and collateral security arrangements to reduce credit risk. Cash collateral received related to derivative instruments under our collateral security arrangements are included in other current assets with a corresponding liability. Cash and non-cash collateral pledged related to derivative instruments under our collateral security arrangements are primarily included in other current assets.\nThe gross amounts of derivative instruments subject to master netting arrangements with various counterparties, and cash and non-cash collateral received and pledged under such agreements were as follows (in millions):\n\n[tables/table-42 Derivatives Offsetting.html]\n\n(1) The balances as of December 31, 2024 and 2025 were related to derivatives allowed to be net settled in accordance with our master netting agreements.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments/Offsetting of Derivatives", + "metadata": { + "length": 856, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "enter", + "master", + "netting", + "arrangements", + "collateral", + "security", + "reduce", + "credit", + "risk", + "Cash", + "received", + "related", + "derivative", + "instruments", + "included", + "current", + "assets", + "liability", + "cash", + "pledged", + "primarily", + "The", + "gross", + "amounts", + "subject", + "counterparties", + "agreements", + "millions", + "balances", + "December", + "31", + "2024", + "2025", + "derivatives", + "allowed", + "net", + "settled", + "accordance" + ], + "keywords": [], + "connect_to": [ + { + "target": "df0054e0-5df8-5e78-a292-66b970b035b9", + "relation": "embeds", + "ref": "[tables/table-42 Derivatives Offsetting.html]", + "position": { + "start": 655, + "end": 700 + } + } + ] + } + }, + { + "chunk_id": "92e430cc-3234-5cd2-9d06-f2fe8cff3715", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Operating lease cost$ 3,362$ 3,304$ 3,345
Finance lease cost:
Amortization of lease assets469413553
Interest on lease liabilities353165
Finance lease cost504444618
Variable lease cost1,1821,4251,739
Total lease cost$ 5,048$ 5,173$ 5,702
", + "path": "tables/table-43 Lease Cost Table.html", + "metadata": { + "length": 682, + "summary": "table-44\nThe table shows lease costs for 2023-2025: operating lease cost around $3.3K, finance lease cost increasing from $504 to $618, variable lease cost rising from $1,182 to $1,739, and total lease cost growing from $5,048 to $5,702.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-43 Lease Cost Table.html", + "keywords": [ + "lease cost", + "operating lease", + "finance lease" + ], + "tokens": [] + } + }, + { + "chunk_id": "dd49735a-4b7d-5ca1-9a3d-746234b026be", + "type": "table", + "content": "
December 31,
20242025
Weighted-average remaining lease term:
Operating leases7.8 years7.6 years
Finance leases10.4 years8.3 years
Weighted-average discount rate:
Operating leases3.4 %3.6 %
Finance leases2.8 %3.1 %
December 31,
20242025
Operating leases:
Operating lease assets$ 13,588$ 15,221
Accrued expenses and other liabilities$ 2,887$ 3,209
Operating lease liabilities11,69112,744
Total operating lease liabilities$ 14,578$ 15,954
Finance leases:
Property and equipment, at cost$ 4,622$ 6,822
Accumulated depreciation(2,037)(2,025)
Property and equipment, net$ 2,585$ 4,797
Accrued expenses and other liabilities$ 235$ 441
Other long-term liabilities1,4422,059
Total finance lease liabilities$ 1,677$ 2,500
", + "path": "tables/table-44 Lease Data.html", + "metadata": { + "length": 1581, + "summary": "table-45\nThe table presents weighted-average remaining lease terms and discount rates for operating and finance leases for 2024 and 2025, along with corresponding lease assets and liabilities.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-44 Lease Data.html", + "keywords": [ + "Lease terms", + "Discount rates", + "Lease liabilities" + ], + "tokens": [] + } + }, + { + "chunk_id": "f2892caf-e15e-59b2-b913-7bdd049cdfd9", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Cash payments for lease liabilities:
Operating cash flows used for operating leases$ 3,173$ 3,425$ 3,370
Operating cash flows used for finance leases$ 35$ 31$ 65
Financing cash flows used for finance leases(1)$ 705$ 405$ 1,988
Assets obtained in exchange for lease liabilities:
Operating leases$ 2,877$ 2,510$ 4,070
Finance leases$ 564$ 313$ 1,606
", + "path": "tables/table-45 Lease Cash Flows.html", + "metadata": { + "length": 763, + "summary": "table-46\nThe table shows cash payments for operating and finance leases from 2023 to 2025, including operating cash flows and financing cash flows, plus assets obtained in exchange for lease liabilities.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-45 Lease Cash Flows.html", + "keywords": [ + "lease liabilities", + "cash flows", + "finance leases" + ], + "tokens": [] + } + }, + { + "chunk_id": "0aaff577-19af-5e90-bcd6-e5688cb6d39d", + "type": "table", + "content": "
Operating LeasesFinance Leases
2026$ 3,275$ 491
20273,082345
20282,510335
20292,061314
20301,669241
Thereafter5,6541,143
Total undiscounted lease payments18,2512,869
Less: imputed interest(2,297)(369)
Total lease liability balance$ 15,954$ 2,500
", + "path": "tables/table-46 Lease Payments.html", + "metadata": { + "length": 611, + "summary": "table-47\nThe table shows future lease payments for operating and finance leases from 2026 to 2030 and thereafter, totaling $18,251 and $2,869 undiscounted, with imputed interest deducted to yield lease liabilities of $15,954 and $2,500.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-46 Lease Payments.html", + "keywords": [ + "Operating Leases", + "Finance Leases", + "Undiscounted Payments" + ], + "tokens": [] + } + }, + { + "chunk_id": "55d6d5b7-45e6-52ec-8fae-d4442d5edb7e", + "type": "text", + "content": "We have entered into operating and finance lease agreements primarily for data centers, land, and offices throughout the world with varying lease terms.\nComponents of lease costs were as follows (in millions):\n\n[tables/table-43 Lease Cost Table.html]\n\nSupplemental information related to leases was as follows (in millions):\n\n[tables/table-44 Lease Data.html]\n\n\n[tables/table-45 Lease Cash Flows.html]\n\n(1) Financing cash flows used for financing leases are included within financing activities as repayments of debt. The year ended December 31, 2025 includes \\$1.1 billion of prepayments for finance leases not yet commenced.\nFuture lease payments as of December 31, 2025 were as follows (in millions):\n\n[tables/table-46 Lease Payments.html]\n\nAs of December 31, 2025, we have entered into leases primarily related to data centers that have not yet commenced with short-term and long-term future lease payments of \\$5.8 billion and \\$52.7 billion, respectively, that are not yet recorded. These leases will commence between 2026 and 2031 with non-cancelable lease terms primarily between one and 25 years.\nIn January 2026, we executed a power purchase agreement which we expect to be accounted for as a lease resulting in future payments depending on certain agreement terms of \\$9.9 billion between 2027 and 2047. If certain contractual conditions for the project are not met, we would instead make a one-time payment of approximately \\$3.5 billion and assume ownership of the power generating assets.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 4. Leases", + "metadata": { + "length": 1535, + "summary": "The passage details operating and finance leases for data centers, land, and offices. Lease costs are summarized in tables. Supplemental information includes cash flows, with $1.1 billion in prepayments for finance leases not yet commenced in 2025. Future lease payments as of December 31, 2025, are presented in a table. Additionally, $5.8 billion in short-term and $52.7 billion in long-term future lease payments for not-yet-commenced data center leases are disclosed, commencing between 2026 and 2031. A January 2026 power purchase agreement, expected to be accounted for as a lease, involves future payments of $9.9 billion from 2027 to 2047, or a one-time payment of $3.5 billion if conditions are unmet.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "entered", + "operating", + "finance", + "lease", + "agreements", + "primarily", + "data", + "centers", + "land", + "offices", + "world", + "varying", + "terms", + "Components", + "costs", + "millions", + "Supplemental", + "information", + "related", + "leases", + "Financing", + "cash", + "flows", + "financing", + "included", + "activities", + "repayments", + "debt", + "The", + "year", + "ended", + "December", + "31", + "2025", + "includes", + "1.1", + "billion", + "prepayments", + "commenced", + "Future", + "payments", + "As", + "short", + "term", + "long", + "future", + "5.8", + "52.7", + "recorded", + "These", + "commence", + "2026", + "2031", + "cancelable", + "25", + "years", + "In", + "January", + "executed", + "power", + "purchase", + "agreement", + "expect", + "accounted", + "resulting", + "depending", + "9.9", + "2027", + "2047", + "If", + "contractual", + "conditions", + "project", + "met", + "make", + "time", + "payment", + "approximately", + "3.5", + "assume", + "ownership", + "generating", + "assets" + ], + "keywords": [ + "Leases", + "Future Payments", + "Finance Leases" + ], + "connect_to": [ + { + "target": "92e430cc-3234-5cd2-9d06-f2fe8cff3715", + "relation": "embeds", + "ref": "[tables/table-43 Lease Cost Table.html]", + "position": { + "start": 211, + "end": 250 + } + }, + { + "target": "dd49735a-4b7d-5ca1-9a3d-746234b026be", + "relation": "embeds", + "ref": "[tables/table-44 Lease Data.html]", + "position": { + "start": 326, + "end": 359 + } + }, + { + "target": "f2892caf-e15e-59b2-b913-7bdd049cdfd9", + "relation": "embeds", + "ref": "[tables/table-45 Lease Cash Flows.html]", + "position": { + "start": 362, + "end": 401 + } + }, + { + "target": "0aaff577-19af-5e90-bcd6-e5688cb6d39d", + "relation": "embeds", + "ref": "[tables/table-46 Lease Payments.html]", + "position": { + "start": 705, + "end": 742 + } + } + ] + } + }, + { + "chunk_id": "ad3f644d-c44a-5ff7-bf05-1a6fe7ec0939", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 5. Variable Interest Entities", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e89b53bb-1875-5779-be90-433810d27bb4", + "type": "text", + "content": "We consolidate VIEs in which we hold a variable interest and are the primary beneficiary. The results of operations and financial position of these VIEs are included in our consolidated financial statements.\nFor certain consolidated VIEs, their assets are not available to us, and their creditors do not have recourse to us. As of December 31, 2024 and 2025, assets that can only be used to settle obligations of these VIEs were \\$8.7 billion and \\$5.6 billion, respectively, and are primarily included in cash and cash equivalents. As of December 31, 2024 and 2025, liabilities for which creditors only have recourse to the VIEs were \\$2.3 billion and \\$2.0 billion, respectively. We may continue to fund ongoing operations, including the potential funding of employee compensation programs, of certain VIEs that are included within Other Bets.\nIn February 2026, Waymo, a consolidated VIE, announced an investment round of \\$16.0 billion, the significant majority of which was funded by Alphabet. Investments from external parties will be accounted for as equity transactions and will result in recognition of noncontrolling interests.\nTotal noncontrolling interests (NCI) in our consolidated subsidiaries were \\$4.2 billion and \\$3.4 billion as of December 31, 2024 and 2025, respectively, of which \\$1.1 billion and \\$841 million were redeemable noncontrolling interests (RNCI) as of December 31, 2024 and 2025, respectively. NCI and RNCI are included within additional paid-in capital. Net loss attributable to noncontrolling interests was not material for any period presented and is included within the \"other\" component of OI&E. See Note 7 for further details on OI&E.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 5. Variable Interest Entities/Consolidated VIEs", + "metadata": { + "length": 1675, + "summary": "The passage discusses the consolidation of Variable Interest Entities (VIEs) where the company is the primary beneficiary. It details restricted assets and liabilities of certain VIEs, with assets of $8.7 billion and $5.6 billion and liabilities of $2.3 billion and $2.0 billion as of December 31, 2024 and 2025, respectively. Waymo, a consolidated VIE, announced a $16.0 billion investment round in February 2026, mostly funded by Alphabet, with external investments treated as equity transactions leading to noncontrolling interests. Total noncontrolling interests were $4.2 billion and $3.4 billion as of December 31, 2024 and 2025, including redeemable noncontrolling interests of $1.1 billion and $841 million. Net loss attributable to noncontrolling interests was not material.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "consolidate", + "VIEs", + "hold", + "variable", + "interest", + "primary", + "beneficiary", + "The", + "results", + "operations", + "financial", + "position", + "included", + "consolidated", + "statements", + "For", + "assets", + "creditors", + "recourse", + "As", + "December", + "31", + "2024", + "2025", + "settle", + "obligations", + "8.7", + "billion", + "5.6", + "primarily", + "cash", + "equivalents", + "liabilities", + "2.3", + "2.0", + "continue", + "fund", + "ongoing", + "including", + "potential", + "funding", + "employee", + "compensation", + "programs", + "Other", + "Bets", + "In", + "February", + "2026", + "Waymo", + "VIE", + "announced", + "investment", + "round", + "16.0", + "significant", + "majority", + "funded", + "Alphabet", + "Investments", + "external", + "parties", + "accounted", + "equity", + "transactions", + "result", + "recognition", + "noncontrolling", + "interests", + "Total", + "NCI", + "subsidiaries", + "4.2", + "3.4", + "1.1", + "841", + "million", + "redeemable", + "RNCI", + "additional", + "paid", + "capital", + "Net", + "loss", + "attributable", + "material", + "period", + "presented", + "component", + "OI", + "See", + "Note", + "details" + ], + "keywords": [ + "VIE", + "noncontrolling interests", + "Waymo" + ], + "connect_to": [] + } + }, + { + "chunk_id": "69e9bca4-089a-5c9b-8c91-da6def74f219", + "type": "text", + "content": "We hold various forms of interests in Variable Interest Entities (VIEs), including certain of our investments in private companies and renewable energy entities, certain leases and credit backstops with data center entities, and certain backstops with energy infrastructure entities. Because we have determined that we do not direct the activities that most significantly impact the economic performance of these entities, we are not the primary beneficiary. Therefore, these VIEs are not consolidated within our financial statements.\nOur investments in private companies and renewable energy VIEs are primarily accounted for as non-marketable securities under the measurement alternative or the equity method. The carrying value of these investments are included within non-marketable securities on our Consolidated Balance Sheets. See Note 3 for further details on investments. The maximum exposure to these VIEs is generally limited to the current carrying value plus future funding commitments. As of December 31, 2024 and 2025, future funding commitments were \\$1.5 billion and \\$1.1 billion, respectively.\nLeases with data center leasing VIEs are accounted for as finance leases and are included within total lease obligations disclosed in Note 4. The maximum exposure arising from leases with VIEs is limited to the net carrying value of commenced finance lease assets, plus the undiscounted future obligations for leases that have not yet commenced. See Note 4 for further details on leases.\nCredit backstops we have provided to data center VIEs are accounted for as credit derivatives. The maximum exposure arising from credit backstops with VIEs is limited to the financial risk over the remaining period of the arrangements, as reflected by the credit derivative notional value. See Note 3 for further details on credit derivatives.\nBackstop agreements we have provided to energy infrastructure VIEs are accounted for as financial guarantees. The maximum exposure to these VIEs is limited to the potential amount of future payments under these arrangements. See Note 10 for further details on financial guarantees.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 5. Variable Interest Entities/Unconsolidated VIEs", + "metadata": { + "length": 2125, + "summary": "The passage describes various interests in Variable Interest Entities (VIEs), including investments in private companies and renewable energy entities, leases and credit backstops with data center entities, and backstops with energy infrastructure entities. The company is not the primary beneficiary and does not consolidate these VIEs. Investments are accounted for as non-marketable securities, with carrying values included in non-marketable securities. Maximum exposure is limited to carrying value plus future funding commitments, which were $1.5 billion and $1.1 billion as of December 31, 2024 and 2025, respectively. Leases with data center VIEs are finance leases, with exposure limited to net carrying value of commenced assets plus undiscounted future obligations. Credit backstops are accounted for as credit derivatives, with exposure limited to the notional value. Backstop agreements for energy infrastructure VIEs are financial guarantees, with exposure limited to potential future payments.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "hold", + "forms", + "interests", + "Variable", + "Interest", + "Entities", + "VIEs", + "including", + "investments", + "private", + "companies", + "renewable", + "energy", + "entities", + "leases", + "credit", + "backstops", + "data", + "center", + "infrastructure", + "Because", + "determined", + "direct", + "activities", + "significantly", + "impact", + "economic", + "performance", + "primary", + "beneficiary", + "Therefore", + "consolidated", + "financial", + "statements", + "Our", + "primarily", + "accounted", + "marketable", + "securities", + "measurement", + "alternative", + "equity", + "method", + "The", + "carrying", + "included", + "Consolidated", + "Balance", + "Sheets", + "See", + "Note", + "details", + "maximum", + "exposure", + "generally", + "limited", + "current", + "future", + "funding", + "commitments", + "As", + "December", + "31", + "2024", + "2025", + "1.5", + "billion", + "1.1", + "Leases", + "leasing", + "finance", + "total", + "lease", + "obligations", + "disclosed", + "arising", + "net", + "commenced", + "assets", + "undiscounted", + "Credit", + "provided", + "derivatives", + "risk", + "remaining", + "period", + "arrangements", + "reflected", + "derivative", + "notional", + "Backstop", + "agreements", + "guarantees", + "potential", + "amount", + "payments", + "10" + ], + "keywords": [ + "Variable Interest Entities", + "VIE", + "Exposure" + ], + "connect_to": [] + } + }, + { + "chunk_id": "c509dd08-2741-5972-81c9-b65c9a1731bd", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 6. Debt", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "822c3bab-18e8-5557-b541-c98c63bff65f", + "type": "text", + "content": "We have a commercial paper program of up to \\$25.0 billion, which is used for general corporate purposes. We had \\$2.3 billion of commercial paper outstanding with a weighted-average effective interest rate of 4.4% as of December 31, 2024 and no commercial paper outstanding as of December 31, 2025. The fair value of the commercial paper approximated its carrying value as of December 31, 2024.\nOur short-term debt balance also includes the current portion of certain long-term debt.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 6. Debt/Short-Term Debt", + "metadata": { + "length": 484, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "commercial", + "paper", + "program", + "25.0", + "billion", + "general", + "corporate", + "purposes", + "2.3", + "outstanding", + "weighted", + "average", + "effective", + "interest", + "rate", + "4.4%", + "December", + "31", + "2024", + "2025", + "The", + "fair", + "approximated", + "carrying", + "Our", + "short", + "term", + "debt", + "balance", + "includes", + "current", + "portion", + "long" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ca9c8f5a-9435-5c09-8dc9-45d0edd8b4e2", + "type": "table", + "content": "
MaturityCoupon RateEffective Interest RateAs of December 31,
20242025
Debt
2016 US dollar notes20262.00%2.23%$ 2,000$ 2,000
2020 US dollar notes2027 - 20600.80% - 2.25%0.93% - 2.33%10,0009,000
$2025 \\text{ US dollar notes}^{(1)}$ 2028 - 20753.88% - 5.70%4.00% - 5.79%022,500
$2025 \\text{ Euro notes}^{(2)}$ 2028 - 20642.38% - 4.38%2.57% - 4.51%015,585
Total face value of long-term debt12,00049,085
Unamortized discount and debt issuance costs $^{(2)}$ (118)(542)
Less: current portion of long-term notes $^{(3)}$ (999)(1,996)
Total long-term debt$ 10,883$ 46,547
", + "path": "tables/table-47 Debt Table.html", + "metadata": { + "length": 1185, + "summary": "table-48\nThe table shows long-term debt with maturities from 2026 to 2075, coupon rates from 0.80% to 5.70%, and effective interest rates from 0.93% to 5.79%. Total face value increased from $12,000 in 2024 to $49,085 in 2025, with net long-term debt of $10,883 and $46,547 respectively.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-47 Debt Table.html", + "keywords": [ + "Long-term debt", + "Maturity", + "Interest rates" + ], + "tokens": [] + } + }, + { + "chunk_id": "1b434ccf-107e-5ce5-a075-710e149efda3", + "type": "table", + "content": "
2026$ 2,000
20271,000
20282,676
20291,764
20305,500
Thereafter36,145
Total$ 49,085
", + "path": "tables/table-48 Debt Maturity.html", + "metadata": { + "length": 280, + "summary": "table-49\nThe table shows a debt maturity schedule from 2026 to 2030 and thereafter, with total obligations of $49,085.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-48 Debt Maturity.html", + "keywords": [ + "Debt", + "Maturity", + "Schedule" + ], + "tokens": [] + } + }, + { + "chunk_id": "855a388e-f3d6-5674-880a-02994519d0b7", + "type": "text", + "content": "During 2025, we issued \\$22.5 billion of US dollar-denominated senior unsecured notes and €13.25 billion of euro-denominated senior unsecured notes for general corporate purposes.\nIn May 2025, we issued \\$5.0 billion of US dollar-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 4.89%, and a weighted-average maturity of approximately 24 years. Additionally, in May 2025, we issued €6.75 billion of euro-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 3.31%, and a weighted-average maturity of approximately 14 years.\nIn November 2025, we issued \\$500 million of US dollar-denominated floating-rate senior unsecured notes and \\$17.0 billion of US dollar-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 4.92% and a weighted-average maturity of approximately 20 years. Additionally in November 2025, we issued €6.5 billion of euro-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 3.44% and a weighted-average maturity of approximately 16 years.\nTotal outstanding long-term debt is summarized below (in millions, except percentages):\n\n[tables/table-47 Debt Table.html]\n\n(1) Includes \\$500 million of floating-rate notes due in 2028. Interest is calculated using the compounded Secured Overnight Financing Rate (SOFR) plus 0.52%, reset quarterly.\n(2) Principal, unamortized discount, and debt issuance costs for the euro-denominated notes include the effect of foreign exchange rates.\n(3) Total current portion of long-term debt is included within accrued expenses and other current liabilities. See Note 7 for further details.\nThe notes in the table above are senior unsecured obligations and rank equally with each other. We may redeem the fixed-rate notes at any time in whole or in part at specified redemption prices. The floating-rate notes are not redeemable prior to maturity. Interest is payable quarterly for the floating-rate notes, semi-annually for the US dollar-denominated fixed-rate notes, and annually for the euro-denominated fixed-rate notes. The effective interest rates are based on proceeds received and contractual interest payments.\nThe total estimated fair value of the outstanding notes was approximately \\$9.0 billion and \\$45.6 billion as of December 31, 2024 and December 31, 2025, respectively. The fair value was determined based on observable market prices of identical instruments in less active markets and is categorized accordingly as Level 2 in the fair value hierarchy.\nAs of December 31, 2025, the future principal payments for long-term debt were as follows (in millions):\n\n[tables/table-48 Debt Maturity.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 6. Debt/Long-Term Debt", + "metadata": { + "length": 2744, + "summary": "In 2025, the company issued $22.5 billion in US dollar-denominated and €13.25 billion in euro-denominated senior unsecured notes. May 2025 issuances included $5.0 billion of US dollar notes (4.89% coupon, ~24-year maturity) and €6.75 billion of euro notes (3.31% coupon, ~14-year maturity). November 2025 issuances included $500 million floating-rate notes and $17.0 billion fixed-rate US dollar notes (4.92% coupon, ~20-year maturity), plus €6.5 billion fixed-rate euro notes (3.44% coupon, ~16-year maturity). Total outstanding long-term debt is summarized in a table with maturity dates, coupon rates, and effective interest rates. The estimated fair value of outstanding notes was $45.6 billion as of December 31, 2025. Future principal payments include $2,000 million due in 2026.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "During", + "2025", + "issued", + "22.5", + "billion", + "US", + "dollar", + "denominated", + "senior", + "unsecured", + "notes", + "13.25", + "euro", + "general", + "corporate", + "purposes", + "In", + "May", + "5.0", + "fixed", + "rate", + "weighted", + "average", + "coupon", + "4.89%", + "maturity", + "approximately", + "24", + "years", + "Additionally", + "6.75", + "3.31%", + "14", + "November", + "500", + "million", + "floating", + "17.0", + "4.92%", + "20", + "6.5", + "3.44%", + "16", + "Total", + "outstanding", + "long", + "term", + "debt", + "summarized", + "millions", + "percentages", + "Includes", + "due", + "2028", + "Interest", + "calculated", + "compounded", + "Secured", + "Overnight", + "Financing", + "Rate", + "SOFR", + "0.52%", + "reset", + "quarterly", + "Principal", + "unamortized", + "discount", + "issuance", + "costs", + "include", + "effect", + "foreign", + "exchange", + "rates", + "current", + "portion", + "included", + "accrued", + "expenses", + "liabilities", + "See", + "Note", + "details", + "The", + "table", + "obligations", + "rank", + "equally", + "We", + "redeem", + "time", + "part", + "redemption", + "prices", + "redeemable", + "prior", + "payable", + "semi", + "annually", + "effective", + "interest", + "based", + "proceeds", + "received", + "contractual", + "payments", + "total", + "estimated", + "fair", + "9.0", + "45.6", + "December", + "31", + "2024", + "determined", + "observable", + "market", + "identical", + "instruments", + "active", + "markets", + "categorized", + "Level", + "hierarchy", + "As", + "future", + "principal" + ], + "keywords": [ + "senior notes", + "fixed-rate", + "euro-denominated" + ], + "connect_to": [ + { + "target": "ca9c8f5a-9435-5c09-8dc9-45d0edd8b4e2", + "relation": "embeds", + "ref": "[tables/table-47 Debt Table.html]", + "position": { + "start": 1178, + "end": 1211 + } + }, + { + "target": "1b434ccf-107e-5ce5-a075-710e149efda3", + "relation": "embeds", + "ref": "[tables/table-48 Debt Maturity.html]", + "position": { + "start": 2656, + "end": 2692 + } + } + ] + } + }, + { + "chunk_id": "a018a591-8148-51d3-b582-646169002ca8", + "type": "text", + "content": "As of December 31, 2025, we had \\$10.0 billion of revolving credit facilities, of which \\$4.0 billion expires in April 2026 and \\$6.0 billion expires in April 2030. The interest rates for all credit facilities are determined based on a formula using certain market rates. No amounts were outstanding under the credit facilities as of December 31, 2024 and 2025.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 6. Debt/Credit Facility", + "metadata": { + "length": 361, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "As", + "December", + "31", + "2025", + "10.0", + "billion", + "revolving", + "credit", + "facilities", + "4.0", + "expires", + "April", + "2026", + "6.0", + "2030", + "The", + "interest", + "rates", + "determined", + "based", + "formula", + "market", + "No", + "amounts", + "outstanding", + "2024" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2b2d2ac0-0092-538c-b643-058516c6fe70", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "40ff1546-023c-57c8-a256-3f1da0e38c20", + "type": "text", + "content": "The allowance for credit losses on accounts receivable was \\$879 million and \\$924 million as of December 31, 2024 and 2025, respectively.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information/Accounts Receivable", + "metadata": { + "length": 138, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "allowance", + "credit", + "losses", + "accounts", + "receivable", + "879", + "million", + "924", + "December", + "31", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "21d337f8-fda0-588f-965f-bc5062201dc8", + "type": "table", + "content": "
As of December 31,
20242025
Technical infrastructure(1)$ 141,852$ 203,679
Office space45,40348,348
Corporate and other assets12,57414,463
Property and equipment, in service199,829266,490
Less: accumulated depreciation(79,390)(98,485)
Add: assets not yet in service50,59778,592
Property and equipment, net$ 171,036$ 246,597
", + "path": "tables/table-49 Property & Equipment.html", + "metadata": { + "length": 659, + "summary": "table-50\nThe table shows property and equipment net values of $171,036 in 2024 and $246,597 in 2025, with categories including technical infrastructure, office space, and accumulated depreciation.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-49 Property & Equipment.html", + "keywords": [ + "Property", + "Equipment", + "Depreciation" + ], + "tokens": [] + } + }, + { + "chunk_id": "fe1a858e-cec4-541a-b178-52de4bcdc8bf", + "type": "text", + "content": "Property and equipment, net, consisted of the following (in millions):\n\n[tables/table-49 Property & Equipment.html]\n\n(1) As of December 31, 2024 and 2025, approximately 60% of technical infrastructure assets were comprised of servers and network equipment. The remaining balance was comprised of data center land and buildings and related assets.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information/Property and Equipment, Net", + "metadata": { + "length": 344, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Property", + "equipment", + "net", + "consisted", + "millions", + "As", + "December", + "31", + "2024", + "2025", + "approximately", + "60%", + "technical", + "infrastructure", + "assets", + "comprised", + "servers", + "network", + "The", + "remaining", + "balance", + "data", + "center", + "land", + "buildings", + "related" + ], + "keywords": [], + "connect_to": [ + { + "target": "21d337f8-fda0-588f-965f-bc5062201dc8", + "relation": "embeds", + "ref": "[tables/table-49 Property & Equipment.html]", + "position": { + "start": 72, + "end": 115 + } + } + ] + } + }, + { + "chunk_id": "99d57e29-6ca5-594f-9be4-29ba68130cc0", + "type": "table", + "content": "
As of December 31,
20242025
Accrued fines and settlements(1)$ 9,830$ 15,594
Accrued purchases of property and equipment7,1048,877
Accrued customer liabilities4,3045,029
Payables to brokers for unsettled investment trades3,866950
Income taxes payable, net2,905523
Other accrued expenses and current liabilities23,21924,584
Accrued expenses and other current liabilities$ 51,228$ 55,557
", + "path": "tables/table-50 Accrued Liabilities.html", + "metadata": { + "length": 721, + "summary": "table-51\nThe table lists accrued expenses and other current liabilities as of December 31, 2024 and 2025, with totals of $51,228 and $55,557 respectively, including fines, property purchases, customer liabilities, and more.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-50 Accrued Liabilities.html", + "keywords": [ + "accrued expenses", + "current liabilities", + "financial table" + ], + "tokens": [] + } + }, + { + "chunk_id": "236081ca-188b-5992-af93-63eb678917fb", + "type": "text", + "content": "Accrued expenses and other current liabilities consisted of the following (in millions):\n\n[tables/table-50 Accrued Liabilities.html]\n\n(1) See Legal Matters in Note 10 for further details.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information/Accrued Expenses and Other Current Liabilities", + "metadata": { + "length": 186, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Accrued", + "expenses", + "current", + "liabilities", + "consisted", + "millions", + "See", + "Legal", + "Matters", + "Note", + "10", + "details" + ], + "keywords": [], + "connect_to": [ + { + "target": "99d57e29-6ca5-594f-9be4-29ba68130cc0", + "relation": "embeds", + "ref": "[tables/table-50 Accrued Liabilities.html]", + "position": { + "start": 90, + "end": 132 + } + } + ] + } + }, + { + "chunk_id": "d8172b90-2954-57be-9152-d03c3b8036e2", + "type": "table", + "content": "
Foreign Currency Translation AdjustmentsUnrealized Gains (Losses) on Available-for-Sale InvestmentsUnrealized Gains (Losses) on Cash Flow HedgesTotal
Balance as of December 31, 2022$ (4,142)$ (3,477)$ 16$ (7,603)
Other comprehensive income (loss) before reclassifications7351,344842,163
Amounts excluded from the assessment of hedge effectiveness recorded in AOCI008484
Amounts reclassified from AOCI01,168(214)954
Other comprehensive income (loss)7352,512(46)3,201
Balance as of December 31, 2023(3,407)(965)(30)(4,402)
Other comprehensive income (loss) before reclassifications(1,673)(116)698(1,091)
Amounts excluded from the assessment of hedge effectiveness recorded in AOCI007777
Amounts reclassified from AOCI0782(166)616
Other comprehensive income (loss)(1,673)666609(398)
Balance as of December 31, 2024(5,080)(299)579(4,800)
Other comprehensive income (loss) before reclassifications2,5111,146(734)2,923
Amounts excluded from the assessment of hedge effectiveness recorded in AOCI110(45)(34)
Amounts reclassified from AOCI0(169)164(5)
Other comprehensive income (loss)2,522977(615)2,884
Balance as of December 31, 2025$ (2,558)$ 678$ (36)$ (1,916)
", + "path": "tables/table-51 OCI Rollforward.html", + "metadata": { + "length": 2073, + "summary": "table-52\nThe table shows the rollforward of Accumulated Other Comprehensive Income (AOCI) from 2022 to 2025, detailing changes in foreign currency translation, available-for-sale investments, and cash flow hedges.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-51 OCI Rollforward.html", + "keywords": [ + "OCI", + "AOCI", + "Comprehensive Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "f1050b65-9d32-5ba8-862d-7bc26d4627d4", + "type": "table", + "content": "
AOCI ComponentsLocationYear Ended December 31,
202320242025
Unrealized gains (losses) on available-for-sale investments
Other income (expense), net$ (1,497)$ (1,008)$ 213
Benefit (provision) for income taxes329226(44)
Net of income tax(1,168)(782)169
Unrealized gains (losses) on cash flow hedges
Foreign exchange contractsRevenue213174(233)
Interest rate contractsOther income (expense), net611
Benefit (provision) for income taxes(5)(9)68
Net of income tax214166(164)
Total amount reclassified, net of income tax$ (954)$ (616)$ 5
", + "path": "tables/table-52 AOCI Reclass.html", + "metadata": { + "length": 1138, + "summary": "table-53\nThe table shows reclassifications of AOCI components for 2023-2025, including unrealized gains/losses on investments and cash flow hedges, with tax impacts and net totals.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-52 AOCI Reclass.html", + "keywords": [ + "AOCI", + "Unrealized gains", + "Tax effect" + ], + "tokens": [] + } + }, + { + "chunk_id": "dc5b25d7-c0d4-5db6-9dbb-d7a090977963", + "type": "text", + "content": "Components of AOCI, net of income tax, were as follows (in millions):\n\n[tables/table-51 OCI Rollforward.html]\n\nThe effects on net income of amounts reclassified from AOCI were as follows (in millions):\n\n[tables/table-52 AOCI Reclass.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information/Accumulated Other Comprehensive Income (Loss)", + "metadata": { + "length": 312, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Components", + "AOCI", + "net", + "income", + "tax", + "millions", + "The", + "effects", + "amounts", + "reclassified" + ], + "keywords": [], + "connect_to": [ + { + "target": "d8172b90-2954-57be-9152-d03c3b8036e2", + "relation": "embeds", + "ref": "[tables/table-51 OCI Rollforward.html]", + "position": { + "start": 71, + "end": 109 + } + }, + { + "target": "f1050b65-9d32-5ba8-862d-7bc26d4627d4", + "relation": "embeds", + "ref": "[tables/table-52 AOCI Reclass.html]", + "position": { + "start": 203, + "end": 238 + } + } + ] + } + }, + { + "chunk_id": "c0b18650-9348-568c-9a9b-cf50b9707b85", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Interest income$ 3,865$ 4,482$ 4,337
Interest expense(1)(308)(268)(736)
Foreign currency exchange gain (loss), net(1,238)(409)(382)
Gain (loss) on debt securities, net(1,215)(1,043)540
Gain (loss) on equity securities, net3923,71424,080
Income (loss) and impairment from equity method investments, net(628)(188)281
Other5561,1371,667
Other income (expense), net$ 1,424$ 7,425$ 29,787
", + "path": "tables/table-53 Income Data Table.html", + "metadata": { + "length": 873, + "summary": "table-54\nThe table shows financial data for 2023-2025: interest income rises then falls, interest expense increases sharply in 2025, and net other income grows significantly, driven by equity securities gains.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-53 Income Data Table.html", + "keywords": [ + "Interest", + "Gains", + "Expenses" + ], + "tokens": [] + } + }, + { + "chunk_id": "fb1ac6ca-b7d2-5950-88ed-e59dfc2ed124", + "type": "text", + "content": "Components of OI&E were as follows (in millions):\n\n[tables/table-53 Income Data Table.html]\n\n(1) Interest expense is net of interest capitalized of \\$181 million, \\$194 million, and \\$447 million for the years ended December 31, 2023, 2024, and 2025, respectively.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information/Other Income (Expense), Net", + "metadata": { + "length": 270, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Components", + "OI", + "millions", + "Interest", + "expense", + "net", + "interest", + "capitalized", + "181", + "million", + "194", + "447", + "years", + "ended", + "December", + "31", + "2023", + "2024", + "2025" + ], + "keywords": [], + "connect_to": [ + { + "target": "c0b18650-9348-568c-9a9b-cf50b9707b85", + "relation": "embeds", + "ref": "[tables/table-53 Income Data Table.html]", + "position": { + "start": 51, + "end": 91 + } + } + ] + } + }, + { + "chunk_id": "7f529d94-9f36-513f-b5fd-c9d68af8650b", + "type": "text", + "content": "Pending Acquisitions\nIn March 2025, we entered into a definitive agreement to acquire Wiz, a leading cloud security platform, for \\$32.0 billion, subject to closing adjustments, in an all-cash transaction. The acquisition of Wiz is expected to close in 2026, subject to customary closing conditions, including the receipt of regulatory approvals. Upon the close of the acquisition, Wiz will be part of the Google Cloud segment.\nIn December 2025, we entered into a definitive agreement to acquire Intersect, which provides data center and energy infrastructure solutions, for \\$4.8 billion in cash, plus the assumption of debt. The acquisition of Intersect is expected to close in the first half of 2026, subject to customary closing conditions.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 8. Acquisitions", + "metadata": { + "length": 744, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Pending", + "Acquisitions", + "In", + "March", + "2025", + "entered", + "definitive", + "agreement", + "acquire", + "Wiz", + "leading", + "cloud", + "security", + "platform", + "32.0", + "billion", + "subject", + "closing", + "adjustments", + "cash", + "transaction", + "The", + "acquisition", + "expected", + "close", + "2026", + "customary", + "conditions", + "including", + "receipt", + "regulatory", + "approvals", + "Upon", + "part", + "Google", + "Cloud", + "segment", + "December", + "Intersect", + "data", + "center", + "energy", + "infrastructure", + "solutions", + "4.8", + "assumption", + "debt", + "half" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f7fc6732-dccf-5a8d-841f-83c6acaac100", + "type": "table", + "content": "
Google ServicesGoogle CloudOther BetsTotal
Balance as of December 31, 2023$ 21,118$ 7,199$ 881$ 29,198
Additions2,44129502,736
Foreign currency translation and other adjustments(38)(4)(7)(49)
Balance as of December 31, 202423,5217,49087431,885
Additions1,26916301,432
Foreign currency translation and other adjustments807(24)63
Balance as of December 31, 2025$ 24,870$ 7,660$ 850$ 33,380
", + "path": "tables/table-54 Google Segment Balances.html", + "metadata": { + "length": 827, + "summary": "table-55\nThe table shows goodwill balances for Google Services, Google Cloud, and Other Bets from 2023 to 2025, with additions and currency adjustments, totaling $33,380 million by end of 2025.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-54 Google Segment Balances.html", + "keywords": [ + "Goodwill", + "Segment", + "Balance" + ], + "tokens": [] + } + }, + { + "chunk_id": "44935a41-bf1d-58b6-acbb-9cbee3baefb2", + "type": "text", + "content": "Changes in the carrying amount of goodwill for the years ended December 31, 2024 and 2025 were as follows (in millions):\n\n[tables/table-54 Google Segment Balances.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 9. Goodwill", + "metadata": { + "length": 190, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Changes", + "carrying", + "amount", + "goodwill", + "years", + "ended", + "December", + "31", + "2024", + "2025", + "millions" + ], + "keywords": [], + "connect_to": [ + { + "target": "f7fc6732-dccf-5a8d-841f-83c6acaac100", + "relation": "embeds", + "ref": "[tables/table-54 Google Segment Balances.html]", + "position": { + "start": 122, + "end": 168 + } + } + ] + } + }, + { + "chunk_id": "ae9e85e2-f37e-5f6c-acdf-3d119ca64aea", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7405d768-cf67-5f93-bcc8-d425a49081e1", + "type": "text", + "content": "We have certain content licensing agreements with future fixed or minimum guaranteed commitments of \\$7.7 billion as of December 31, 2025, of which the majority is paid quarterly through the first quarter of 2030.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Commitments", + "metadata": { + "length": 213, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "content", + "licensing", + "agreements", + "future", + "fixed", + "minimum", + "guaranteed", + "commitments", + "7.7", + "billion", + "December", + "31", + "2025", + "majority", + "paid", + "quarterly", + "quarter", + "2030" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "da31062e-dda9-5fff-a5b3-c00baa1074fe", + "type": "text", + "content": "We provide financial guarantees to certain counterparties, in the form of backstop agreements with varying terms through August 2026. These backstop agreements support counterparty procurement of long-lead time equipment for our future power purchase agreements. As of December 31, 2025, our maximum potential amount of future payments under these guarantees was \\$5.7 billion, upon which we may receive certain assets. The fair value of these obligations was not material.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Financial Guarantees", + "metadata": { + "length": 473, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "provide", + "financial", + "guarantees", + "counterparties", + "form", + "backstop", + "agreements", + "varying", + "terms", + "August", + "2026", + "These", + "support", + "counterparty", + "procurement", + "long", + "lead", + "time", + "equipment", + "future", + "power", + "purchase", + "As", + "December", + "31", + "2025", + "maximum", + "potential", + "amount", + "payments", + "5.7", + "billion", + "receive", + "assets", + "The", + "fair", + "obligations", + "material" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9849757b-607e-53da-9c9f-580004e39c56", + "type": "text", + "content": "In the normal course of business, including to facilitate transactions in our services and products and corporate activities, we indemnify certain parties, including advertisers, Google Network partners, distribution partners, customers of Google Cloud offerings, lessors, and service providers with respect to certain matters. We have agreed to defend and/or indemnify certain parties against losses arising from a breach of representations or covenants, or out of intellectual property infringement or other claims made against certain parties. Several of these agreements limit the time within which an indemnification claim can be made and the amount of the claim. In addition, we have entered into indemnification agreements with our officers and directors, and our bylaws contain similar indemnification obligations to our agents.\nIt is not possible to make a reasonable estimate of the maximum potential amount under these indemnification agreements due to the unique facts and circumstances involved in each particular agreement. Additionally, the payments we have made under such agreements have not had a material adverse effect on our results of operations, cash flows, or financial position. However, to the extent that valid indemnification claims arise in the future, future payments by us could be significant and could have a material adverse effect on our results of operations or cash flows in a particular period.\nAs of December 31, 2025, we did not have any material indemnification claims that were probable or reasonably possible.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Indemnifications", + "metadata": { + "length": 1552, + "summary": "The passage discusses the company's practice of indemnifying various parties like advertisers, partners, and officers against losses from breaches or claims. It notes that the maximum potential liability cannot be estimated due to unique circumstances, but past payments have not been material. However, future claims could be significant. As of December 31, 2025, no material indemnification claims were probable or reasonably possible.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "In", + "normal", + "business", + "including", + "facilitate", + "transactions", + "services", + "products", + "corporate", + "activities", + "indemnify", + "parties", + "advertisers", + "Google", + "Network", + "partners", + "distribution", + "customers", + "Cloud", + "offerings", + "lessors", + "service", + "providers", + "respect", + "matters", + "We", + "agreed", + "defend", + "losses", + "arising", + "breach", + "representations", + "covenants", + "intellectual", + "property", + "infringement", + "claims", + "made", + "Several", + "agreements", + "limit", + "time", + "indemnification", + "claim", + "amount", + "addition", + "entered", + "officers", + "directors", + "bylaws", + "similar", + "obligations", + "agents", + "It", + "make", + "reasonable", + "estimate", + "maximum", + "potential", + "due", + "unique", + "facts", + "circumstances", + "involved", + "agreement", + "Additionally", + "payments", + "material", + "adverse", + "effect", + "results", + "operations", + "cash", + "flows", + "financial", + "position", + "However", + "extent", + "valid", + "arise", + "future", + "significant", + "period", + "As", + "December", + "31", + "2025", + "probable" + ], + "keywords": [ + "indemnification", + "agreements", + "claims" + ], + "connect_to": [] + } + }, + { + "chunk_id": "7c72474b-7a29-5c21-a649-05b176151c33", + "type": "text", + "content": "We record a liability when we believe that it is probable that a loss has been incurred, and the amount can be reasonably estimated. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the reasonably possible loss. We evaluate developments in our legal matters that could affect the amount of liability that has been previously accrued, and the matters and related reasonably possible losses disclosed, and make adjustments as appropriate.\nCertain outstanding matters seek speculative, substantial, or indeterminate monetary amounts, substantial changes to our business practices and products, or structural remedies. Significant judgment is required to determine both the likelihood of there being a loss and the estimated amount of a loss related to such matters, and we may be unable to estimate the reasonably possible loss or range of losses. The outcomes of outstanding legal matters are inherently unpredictable and subject to significant uncertainties, and could, either individually or in aggregate, have a material adverse effect.\nWe expense legal fees in the period in which they are incurred.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Legal Matters", + "metadata": { + "length": 1163, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "record", + "liability", + "probable", + "loss", + "incurred", + "amount", + "estimated", + "If", + "determine", + "range", + "disclose", + "evaluate", + "developments", + "legal", + "matters", + "affect", + "previously", + "accrued", + "related", + "losses", + "disclosed", + "make", + "adjustments", + "Certain", + "outstanding", + "seek", + "speculative", + "substantial", + "indeterminate", + "monetary", + "amounts", + "business", + "practices", + "products", + "structural", + "remedies", + "Significant", + "judgment", + "required", + "likelihood", + "unable", + "estimate", + "The", + "outcomes", + "inherently", + "unpredictable", + "subject", + "significant", + "uncertainties", + "individually", + "aggregate", + "material", + "adverse", + "effect", + "expense", + "fees", + "period" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3ec53275-6590-5261-90dc-991db7865d0a", + "type": "text", + "content": "We are subject to formal and informal inquiries and investigations as well as litigation on various competition matters by regulatory authorities and private parties in the US, Europe, and other jurisdictions globally, including the following:\n- Shopping: In June 2017, the EC announced its decision that certain actions taken by Google relating to its display and ranking of shopping search results and ads infringed European antitrust laws and imposed a €2.4 billion fine. In 2024, we made a cash payment of \\$3.0 billion for the fine.\n- Android: In July 2018, the EC announced its decision that certain provisions in Google's Android-related distribution agreements infringed European antitrust laws, imposed a €4.3 billion fine, and directed the termination of the conduct at issue. We appealed the EC decision and implemented changes to certain of our Android distribution practices. In September 2022, the General Court affirmed the EC decision but reduced the fine from €4.3 billion to €4.1 billion. We subsequently appealed the General Court's affirmation of the EC decision with the European Court of Justice, which remains pending. In 2018, we recognized a charge of \\$5.1 billion for the fine, which we reduced by \\$217 million in 2022.\n- AdSense for Search: In March 2019, the EC announced its decision that certain provisions in Google's agreements with AdSense for Search partners infringed European antitrust laws, imposed a €1.5 billion fine, and directed actions related to AdSense for Search partners' agreements, which we implemented prior to the decision. In 2019, we recognized a charge of \\$1.7 billion for the fine and appealed the EC decision. In September 2024, the General Court overturned the EC decision and annulled the €1.5 billion fine. The EC has appealed the General Court's decision with the European Court of Justice.\n- Search: In October 2020, the DOJ and a number of state Attorneys General filed a lawsuit in the US District Court for the District of Columbia concerning Google's Search and Search advertising practices and its compliance with US antitrust laws. In August 2024, the US District Court for the District of Columbia ruled against Google. A final judgment was entered in December 2025, which, among other things, imposes restrictions on how Google distributes its services and requires Google to share certain search data with and offer syndication services to certain competitors. In January 2026, we appealed the final judgment and moved to pause implementation of certain remedies. In February 2026, the DOJ and state Attorneys General also appealed.\nFurther, in June 2022, the Australian Competition and Consumer Commission (ACCC) opened an investigation into Search distribution practices. In August 2025, we agreed to a settlement with the ACCC requiring, among other things, changes to our Android agreements. We recognized a charge in the second quarter of 2025, and the settlement was approved by the court in December 2025.\nIn October 2023, the Japanese Fair Trade Commission (JFTC) opened an investigation into Search distribution practices. In April 2025, the JFTC issued a cease-and-desist order requiring us to make changes to our Android agreements to ensure they are consistent with Japanese antitrust law. The JFTC did not impose monetary penalties.\n\\- Advertising Technology: In December 2020, a number of state Attorneys General filed a lawsuit in the US District Court for the Eastern District of Texas concerning Google's advertising technology and its compliance with US antitrust laws and state deceptive trade laws. In January 2023, the DOJ, along with a number of state Attorneys General, filed a lawsuit in the US District Court for the Eastern District of Virginia concerning Google's advertising technology and its compliance with US antitrust laws, and a number of additional state Attorneys General subsequently joined the lawsuit. In April 2025, the US District Court for the Eastern District of Virginia issued a mixed decision in the DOJ case against Google, ruling that neither Google's advertiser tools nor the DoubleClick and AdMeld acquisitions were anticompetitive, but that Google's publisher tools unfairly excluded rivals. A separate proceeding to determine remedies, the range of which vary widely, took place in September 2025, with the parties presenting differing remedy proposals. The DOJ's remedy proposal includes structural remedies that could have a material adverse effect on our business. Closing arguments were held in November 2025, and we are awaiting a final judgment. After that judgment, we plan to appeal the adverse portion of the April 2025 decision and potentially aspects of the remedies decision. A trial in the state Attorneys General case in the Eastern District of Texas will take place after a decision on remedies is issued in the DOJ case. Given the nature of these matters, we cannot estimate a possible loss.\nFurther, in September 2025, the EC announced its decision that Google had infringed European competition laws through \"self-preferencing\" practices on the buy-side and the sell-side relating to Google's advertising technology business. The EC decision imposed a €3.0 billion fine and directed Google to cease and desist the alleged \"self-preferencing\" practices. We appealed the ruling in November 2025. We recognized a charge of \\$3.5 billion in the third quarter of 2025, and we placed bank guarantees in the fourth quarter of 2025 in lieu of cash payment.\nIn September 2024, the UK also issued a Statement of Objections concerning Google's advertising technology and its compliance with UK antitrust laws, to which we responded.\n\\- Google Play: In July 2021, a number of state Attorneys General filed a lawsuit in the US District Court for the Northern District of California concerning Google's operation of Android and Google Play and its compliance with US antitrust laws and state antitrust and consumer protection laws. In September 2023, we reached a settlement in principle with 50 state Attorneys General and three territories and recognized a charge. The court preliminarily approved the settlement in November 2025, and final approval remains pending before the court. In May 2024, we funded the settlement amount to an escrow agent.\nIn December 2023, a California jury delivered a verdict against Google in Epic Games v. Google related to Google Play's business. Epic did not seek monetary damages. The presiding judge issued a remedies decision in October 2024, ordering a variety of alterations to our business models and operations and contractual agreements for Android and Google Play. We appealed the judgment, including the jury verdict and aspects of the remedies ordered, and in July 2025, the Court of Appeals denied our appeal. We are in the process of appealing that decision to the US Supreme Court, and we implemented the ordered remedies in October 2025 while the appeal is pending. In October 2025, we reached a settlement with Epic to modify the remedies in this case and resolve certain other lawsuits Epic has filed regarding Google Play's business. The settlement is contingent on the court approving a proposed modified injunction. Epic and Google filed a joint motion to modify the injunction in November 2025, which is currently pending before the court.\n\\- European Digital Markets Act: In March 2024, the EC opened two investigations regarding Google's compliance with certain provisions of the EU's Digital Markets Act relating to Google Play and Search. In March 2025, the EC issued preliminary findings of non-compliance in both investigations, to which we responded. Given the nature of this matter, we cannot estimate a possible loss.\nIn addition to these antitrust proceedings, private individual and collective actions that overlap with claims pursued by regulatory authorities are pending in the US and in several other jurisdictions, including across Europe. Given the nature of these matters, we cannot estimate a possible loss.\nWe believe we have strong arguments against these open claims and will defend ourselves vigorously. We continue to cooperate with federal and state regulators in the US, the EC, and other regulators around the world.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Legal Matters/Antitrust Matters", + "metadata": { + "length": 8226, + "summary": "The passage details numerous antitrust investigations, lawsuits, and fines against Google globally. Key cases include the EC's Shopping fine (€2.4B), Android fine (reduced to €4.1B), AdSense for Search fine (€1.5B, later overturned), and US DOJ cases on Search and advertising technology. Other matters involve the ACCC, JFTC, Google Play settlements, and the EU Digital Markets Act. Fines and charges are noted in euros and US dollars, with ongoing appeals and remedy proceedings.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "subject", + "formal", + "informal", + "inquiries", + "investigations", + "litigation", + "competition", + "matters", + "regulatory", + "authorities", + "private", + "parties", + "US", + "Europe", + "jurisdictions", + "globally", + "including", + "Shopping", + "In", + "June", + "2017", + "EC", + "announced", + "decision", + "actions", + "Google", + "relating", + "display", + "ranking", + "shopping", + "search", + "results", + "ads", + "infringed", + "European", + "antitrust", + "laws", + "imposed", + "2.4", + "billion", + "fine", + "2024", + "made", + "cash", + "payment", + "3.0", + "Android", + "July", + "2018", + "provisions", + "related", + "distribution", + "agreements", + "4.3", + "directed", + "termination", + "conduct", + "issue", + "appealed", + "implemented", + "practices", + "September", + "2022", + "General", + "Court", + "affirmed", + "reduced", + "4.1", + "subsequently", + "affirmation", + "Justice", + "remains", + "pending", + "recognized", + "charge", + "5.1", + "217", + "million", + "AdSense", + "Search", + "March", + "2019", + "partners", + "1.5", + "prior", + "1.7", + "overturned", + "annulled", + "The", + "October", + "2020", + "DOJ", + "number", + "state", + "Attorneys", + "filed", + "lawsuit", + "District", + "Columbia", + "advertising", + "compliance", + "August", + "ruled", + "final", + "judgment", + "entered", + "December", + "2025", + "things", + "imposes", + "restrictions", + "distributes", + "services", + "requires", + "share", + "data", + "offer", + "syndication", + "competitors", + "January", + "2026", + "moved", + "pause", + "implementation", + "remedies", + "February", + "Further", + "Australian", + "Competition", + "Consumer", + "Commission", + "ACCC", + "opened", + "investigation", + "agreed", + "settlement", + "requiring", + "quarter", + "approved", + "court", + "2023", + "Japanese", + "Fair", + "Trade", + "JFTC", + "April", + "issued", + "cease", + "desist", + "order", + "make", + "ensure", + "consistent", + "law", + "impose", + "monetary", + "penalties", + "Advertising", + "Technology", + "Eastern", + "Texas", + "technology", + "deceptive", + "trade", + "Virginia", + "additional", + "joined", + "mixed", + "case", + "ruling", + "advertiser", + "tools", + "DoubleClick", + "AdMeld", + "acquisitions", + "anticompetitive", + "publisher", + "unfairly", + "excluded", + "rivals", + "separate", + "proceeding", + "determine", + "range", + "vary", + "widely", + "place", + "presenting", + "differing", + "remedy", + "proposals", + "proposal", + "includes", + "structural", + "material", + "adverse", + "effect", + "business", + "Closing", + "arguments", + "held", + "November", + "awaiting", + "After", + "plan", + "appeal", + "portion", + "potentially", + "aspects", + "trial", + "Given", + "nature", + "estimate", + "loss", + "preferencing", + "buy", + "side", + "sell", + "alleged", + "3.5", + "bank", + "guarantees", + "fourth", + "lieu", + "UK", + "Statement", + "Objections", + "responded", + "Play", + "2021", + "Northern", + "California", + "operation", + "consumer", + "protection", + "reached", + "principle", + "50", + "territories", + "preliminarily", + "approval", + "May", + "funded", + "amount", + "escrow", + "agent", + "jury", + "delivered", + "verdict", + "Epic", + "Games", + "seek", + "damages", + "presiding", + "judge", + "ordering", + "variety", + "alterations", + "models", + "operations", + "contractual", + "ordered", + "Appeals", + "denied", + "process", + "appealing", + "Supreme", + "modify", + "resolve", + "lawsuits", + "contingent", + "approving", + "proposed", + "modified", + "injunction", + "joint", + "motion", + "Digital", + "Markets", + "Act", + "EU", + "preliminary", + "findings", + "matter", + "addition", + "proceedings", + "individual", + "collective", + "overlap", + "claims", + "pursued", + "strong", + "open", + "defend", + "vigorously", + "continue", + "cooperate", + "federal", + "regulators", + "world" + ], + "keywords": [ + "antitrust", + "Google", + "fines" + ], + "connect_to": [] + } + }, + { + "chunk_id": "73fa74f6-1bed-56f7-b46c-9fc2732899f6", + "type": "text", + "content": "We are subject to a number of privacy-related laws and regulations, and we currently are party to a number of privacy investigations and lawsuits ongoing in multiple jurisdictions. For example, there are ongoing investigations and litigation in the US and the EU, including those relating to our collection and use of location information, the choices we offer users, and advertising practices, which could result in significant fines, judgments, and product changes. In October 2025, we finalized a \\$1.4 billion settlement of certain privacy matters.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Legal Matters/Privacy Matters", + "metadata": { + "length": 552, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "subject", + "number", + "privacy", + "related", + "laws", + "regulations", + "party", + "investigations", + "lawsuits", + "ongoing", + "multiple", + "jurisdictions", + "For", + "litigation", + "US", + "EU", + "including", + "relating", + "collection", + "location", + "information", + "choices", + "offer", + "users", + "advertising", + "practices", + "result", + "significant", + "fines", + "judgments", + "product", + "In", + "October", + "2025", + "finalized", + "1.4", + "billion", + "settlement", + "matters" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c4449e1b-64b6-5c73-99b8-d77e690c680f", + "type": "text", + "content": "We have had patent, copyright, trade secret, and trademark infringement lawsuits filed against us claiming that certain of our products, services, and technologies infringe others' intellectual property rights. Adverse results in these lawsuits may include awards of substantial monetary damages, costly royalty or licensing agreements, or orders preventing us from offering certain features, functionalities, products, or services. As a result, we may have to change our business practices and develop non-infringing products or technologies, which could result in a loss of revenues for us and otherwise harm our business. In addition, the ITC has increasingly become an important forum to litigate intellectual property disputes because an ultimate loss in an ITC action can result in a prohibition on importing infringing products into the US. Because the US is an important market, a prohibition on importation could have an adverse effect on us, including preventing us from importing many important products into the US or necessitating workarounds that may limit certain features of our products. Further, our customers and partners may discontinue the use of our products, services, and technologies, as a result of injunctions or otherwise, which could result in loss of revenues and adversely affect our business.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Legal Matters/Patent and Intellectual Property Claims", + "metadata": { + "length": 1324, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "patent", + "copyright", + "trade", + "secret", + "trademark", + "infringement", + "lawsuits", + "filed", + "claiming", + "products", + "services", + "technologies", + "infringe", + "intellectual", + "property", + "rights", + "Adverse", + "results", + "include", + "awards", + "substantial", + "monetary", + "damages", + "costly", + "royalty", + "licensing", + "agreements", + "orders", + "preventing", + "offering", + "features", + "functionalities", + "As", + "result", + "change", + "business", + "practices", + "develop", + "infringing", + "loss", + "revenues", + "harm", + "In", + "addition", + "ITC", + "increasingly", + "important", + "forum", + "litigate", + "disputes", + "ultimate", + "action", + "prohibition", + "importing", + "US", + "Because", + "market", + "importation", + "adverse", + "effect", + "including", + "necessitating", + "workarounds", + "limit", + "Further", + "customers", + "partners", + "discontinue", + "injunctions", + "adversely", + "affect" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b6b65ac8-8783-5172-b1fe-06de896aa01c", + "type": "text", + "content": "We are subject to claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders involving competition, intellectual property, data privacy and security, tax and related compliance, labor and employment, commercial disputes, content generated by our users, goods and services offered by advertisers or publishers using our platforms, design of our products and services, personal injury and other tort and nuisance theories, consumer protection, including how we moderate content on our platforms, AI, and other matters.\nFor example, we periodically have data incidents that we report to relevant regulators as required by law. Such claims, consent orders, lawsuits, regulatory and government investigations, and other proceedings could result in substantial fines and penalties, injunctive relief, ongoing monitoring and auditing obligations, changes to our products and services, alterations to our business models and operations, and collateral related civil litigation or other adverse consequences, all of which could harm our business, reputation, financial condition, and operating results.\nWe have ongoing legal matters relating to Russia. For example, some matters concern civil judgments that include compounding penalties imposed upon us in connection with disputes regarding the termination of accounts, including those of sanctioned parties. We do not expect these ongoing legal matters will have a material adverse effect.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Legal Matters/Other", + "metadata": { + "length": 1485, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "subject", + "claims", + "lawsuits", + "regulatory", + "government", + "inquiries", + "investigations", + "proceedings", + "consent", + "orders", + "involving", + "competition", + "intellectual", + "property", + "data", + "privacy", + "security", + "tax", + "related", + "compliance", + "labor", + "employment", + "commercial", + "disputes", + "content", + "generated", + "users", + "goods", + "services", + "offered", + "advertisers", + "publishers", + "platforms", + "design", + "products", + "personal", + "injury", + "tort", + "nuisance", + "theories", + "consumer", + "protection", + "including", + "moderate", + "AI", + "matters", + "For", + "periodically", + "incidents", + "report", + "relevant", + "regulators", + "required", + "law", + "Such", + "result", + "substantial", + "fines", + "penalties", + "injunctive", + "relief", + "ongoing", + "monitoring", + "auditing", + "obligations", + "alterations", + "business", + "models", + "operations", + "collateral", + "civil", + "litigation", + "adverse", + "consequences", + "harm", + "reputation", + "financial", + "condition", + "operating", + "results", + "legal", + "relating", + "Russia", + "concern", + "judgments", + "include", + "compounding", + "imposed", + "connection", + "termination", + "accounts", + "sanctioned", + "parties", + "expect", + "material", + "effect" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "35c771ad-7edf-590f-bc06-3c0a70fea643", + "type": "text", + "content": "We are under audit by various domestic and foreign tax authorities with regards to non-income tax matters. The subject matter of non-income tax audits primarily arises from disputes on the tax treatment and tax rate applied to the sale of our products and services in these jurisdictions and the tax treatment of certain employee benefits. We accrue non-income taxes that may result from examinations by, or any negotiated agreements with, these tax authorities when a loss is probable and reasonably estimable. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the reasonably possible loss. Due to the inherent complexity and uncertainty of these matters and judicial process in certain jurisdictions, the final outcome may be materially different from our expectations.\nSee Note 14 for further details regarding income tax contingencies.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Non-Income Taxes", + "metadata": { + "length": 900, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "audit", + "domestic", + "foreign", + "tax", + "authorities", + "income", + "matters", + "The", + "subject", + "matter", + "audits", + "primarily", + "arises", + "disputes", + "treatment", + "rate", + "applied", + "sale", + "products", + "services", + "jurisdictions", + "employee", + "benefits", + "accrue", + "taxes", + "result", + "examinations", + "negotiated", + "agreements", + "loss", + "probable", + "estimable", + "If", + "determine", + "range", + "estimated", + "disclose", + "Due", + "inherent", + "complexity", + "uncertainty", + "judicial", + "process", + "final", + "outcome", + "materially", + "expectations", + "See", + "Note", + "14", + "details", + "contingencies" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c0245856-84f7-579e-a4a1-a6f695604a26", + "type": "text", + "content": "Class A and Class B Common Stock and Class C Capital Stock\nOur Board of Directors has authorized three classes of stock, Class A and Class B common stock, and Class C capital stock. The rights of the holders of each class of our common and capital stock are identical, except with respect to voting. Each share of Class A common stock is entitled to one vote per share. Each share of Class B common stock is entitled to 10 votes per share. Class C capital stock has no voting rights, except as required by applicable law. Shares of Class B common stock may be converted at any time at the option of the stockholder and automatically convert upon sale or transfer to Class A common stock.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 11. Stockholders' Equity", + "metadata": { + "length": 687, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Class", + "Common", + "Stock", + "Capital", + "Our", + "Board", + "Directors", + "authorized", + "classes", + "stock", + "common", + "capital", + "The", + "rights", + "holders", + "class", + "identical", + "respect", + "voting", + "Each", + "share", + "entitled", + "vote", + "10", + "votes", + "required", + "applicable", + "law", + "Shares", + "converted", + "time", + "option", + "stockholder", + "automatically", + "convert", + "sale", + "transfer" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c6d07343-e993-5a69-b576-8da141606332", + "type": "table", + "content": "
202320242025
SharesAmountSharesAmountSharesAmount
Class A share repurchases78$ 9,31673$ 11,85537$ 6,501
Class C share repurchases45052,86830650,19220338,897
Total share repurchases(1)528$ 62,184379$ 62,047240$ 45,398
", + "path": "tables/table-55 Share Repurchases.html", + "metadata": { + "length": 599, + "summary": "table-56\nThe table shows Class A and Class C share repurchases from 2023 to 2025, with total shares and amounts decreasing over the years.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-55 Share Repurchases.html", + "keywords": [ + "share repurchases", + "Class A", + "Class C" + ], + "tokens": [] + } + }, + { + "chunk_id": "4f83fe1b-75f1-5283-b817-098a4bdc725e", + "type": "text", + "content": "In the years ended December 31, 2023, 2024, and 2025, we continued to repurchase both Class A and Class C shares in a manner deemed in the best interest of the company and its stockholders, taking into account the economic cost and prevailing market conditions, including the relative trading prices and volumes of the Class A and Class C shares. In April 2024, the company's Board of Directors authorized a \\$70.0 billion share repurchase program for its Class A and Class C shares. In April 2025, the company's Board of Directors authorized an additional \\$70.0 billion share repurchase program for its Class A and Class C shares. As of December 31, 2025, \\$69.5 billion remained available for Class A and Class C share repurchases.\nThe following table presents Class A and Class C shares repurchased and subsequently retired (in millions):\nYear Ended December 31,\n\n[tables/table-55 Share Repurchases.html]\n\n(1) Shares repurchased include any unsettled repurchases.\nRepurchases are executed from time to time, subject to general business and market conditions and other investment opportunities, through open market purchases or privately negotiated transactions, including through Rule 10b5-1 plans. The repurchase programs do not have an expiration date.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 11. Stockholders' Equity/Share Repurchases", + "metadata": { + "length": 1255, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "In", + "years", + "ended", + "December", + "31", + "2023", + "2024", + "2025", + "continued", + "repurchase", + "Class", + "shares", + "manner", + "deemed", + "interest", + "company", + "stockholders", + "taking", + "account", + "economic", + "cost", + "prevailing", + "market", + "conditions", + "including", + "relative", + "trading", + "prices", + "volumes", + "April", + "Board", + "Directors", + "authorized", + "70.0", + "billion", + "share", + "program", + "additional", + "As", + "69.5", + "remained", + "repurchases", + "The", + "table", + "presents", + "repurchased", + "subsequently", + "retired", + "millions", + "Year", + "Ended", + "Shares", + "include", + "unsettled", + "Repurchases", + "executed", + "time", + "subject", + "general", + "business", + "investment", + "opportunities", + "open", + "purchases", + "privately", + "negotiated", + "transactions", + "Rule", + "10b5", + "plans", + "programs", + "expiration", + "date" + ], + "keywords": [], + "connect_to": [ + { + "target": "c6d07343-e993-5a69-b576-8da141606332", + "relation": "embeds", + "ref": "[tables/table-55 Share Repurchases.html]", + "position": { + "start": 868, + "end": 908 + } + } + ] + } + }, + { + "chunk_id": "0ad5d51d-0a0c-5bd9-a523-aad31f89a6ff", + "type": "text", + "content": "During the year ended December 31, 2025, total cash dividends were \\$4.8 billion for Class A, \\$703 million for Class B, and \\$4.5 billion for Class C shares, respectively.\nIn April 2025, the company's Board of Directors increased the quarterly cash dividend by 5% to \\$0.21 per share of outstanding Class A, Class B, and Class C shares.\nThe company has declared a quarterly cash dividend in the current quarter, and intends to pay quarterly cash dividends in the future, subject to review and approval by the company's Board of Directors in its sole discretion.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 11. Stockholders' Equity/Dividends", + "metadata": { + "length": 562, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "During", + "year", + "ended", + "December", + "31", + "2025", + "total", + "cash", + "dividends", + "4.8", + "billion", + "Class", + "703", + "million", + "4.5", + "shares", + "In", + "April", + "company", + "Board", + "Directors", + "increased", + "quarterly", + "dividend", + "5%", + "0.21", + "share", + "outstanding", + "The", + "declared", + "current", + "quarter", + "intends", + "pay", + "future", + "subject", + "review", + "approval", + "sole", + "discretion" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b0745017-4108-5452-b2ed-e590fa29864b", + "type": "table", + "content": "
Year Ended December 31, 2023
Class AClass BClass CConsolidated
Basic net income per share:
Numerator
Allocation of distributed earnings (cash dividends paid)$ 0$ 0$ 0$ 0
Allocation of undistributed earnings34,6015,12434,07073,795
Net income$ 34,601$ 5,124$ 34,070$ 73,795
Denominator
Number of shares used in per share computation5,9228775,83112,630
Basic net income per share$ 5.84$ 5.84$ 5.84$ 5.84
Diluted net income per share:
Numerator
Allocation of total earnings for basic computation$ 34,601$ 5,124$ 34,070$ 73,795
Reallocation of total earnings as a result of conversion of Class B to Class A shares5,12400(1)
Reallocation of undistributed earnings(287)(37)287(1)
Net income$ 39,438$ 5,087$ 34,357$ 73,795
Denominator
Number of shares used in basic computation5,9228775,83112,630
Weighted-average effect of dilutive securities
Add:
Conversion of Class B to Class A shares outstanding87700(1)
Restricted stock units and other contingently issuable shares009292
Number of shares used in per share computation6,7998775,92312,722
Diluted net income per share$ 5.80$ 5.80$ 5.80$ 5.80
", + "path": "tables/table-56 EPS Breakdown.html", + "metadata": { + "length": 2363, + "summary": "table-57\nThe table details basic and diluted net income per share for Class A, B, and C shares for the year ended December 31, 2023, showing allocations of earnings and share counts.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-56 EPS Breakdown.html", + "keywords": [ + "net income", + "per share", + "diluted" + ], + "tokens": [] + } + }, + { + "chunk_id": "91aacc88-6e89-518e-8f58-13b93a494f1c", + "type": "table", + "content": "
Year Ended December 31, 2024
Class AClass BClass CConsolidated
Basic net income per share:
Numerator
Allocation of distributed earnings (cash dividends paid)$ 3,509$ 519$ 3,335$ 7,363
Allocation of undistributed earnings44,0856,52042,15092,755
Net income$ 47,594$ 7,039$ 45,485$ 100,118
Denominator
Number of shares used in per share computation5,8558665,59812,319
Basic net income per share$ 8.13$ 8.13$ 8.13$ 8.13
Diluted net income per share:
Numerator
Allocation of total earnings for basic computation$ 47,594$ 7,039$ 45,485$ 100,118
Reallocation of total earnings as a result of conversion of Class B to Class A shares7,03900(1)
Reallocation of undistributed earnings(520)(67)520(1)
Net income$ 54,113$ 6,972$ 46,005$ 100,118
Denominator
Number of shares used in basic computation5,8558665,59812,319
Weighted-average effect of dilutive securities
Add:
Conversion of Class B to Class A shares outstanding86600(1)
Restricted stock units and other contingently issuable shares00128128
Number of shares used in per share computation6,7218665,72612,447
Diluted net income per share$ 8.05$ 8.05$ 8.03$ 8.04
", + "path": "tables/table-57 EPS Calculation.html", + "metadata": { + "length": 2382, + "summary": "table-58\nThe table details basic and diluted net income per share for Class A, B, and C shares for the year ended December 31, 2024, with consolidated net income of $100,118 and basic EPS of $8.13.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-57 EPS Calculation.html", + "keywords": [ + "net income", + "shares", + "EPS" + ], + "tokens": [] + } + }, + { + "chunk_id": "27725503-821e-5738-8db0-f52316729da0", + "type": "table", + "content": "
Year Ended December 31, 2025
Class AClass BClass CConsolidated
Basic net income per share:
Numerator
Allocation of distributed earnings (cash dividends paid)$ 4,832$ 703$ 4,514$ 10,049
Allocation of undistributed earnings58,6828,55754,882122,121
Net income$ 63,514$ 9,260$ 59,396$ 132,170
Denominator
Number of shares used in per share computation5,8228495,44512,116
Basic net income per share$ 10.91$ 10.91$ 10.91$ 10.91
Diluted net income per share:
Numerator
Allocation of total earnings for basic computation$ 63,514$ 9,260$ 59,396$ 132,170
Reallocation of total earnings as a result of conversion of Class B to Class A shares9,26000(1)
Reallocation of undistributed earnings(627)(79)627(1)
Net income$ 72,147$ 9,181$ 60,023$ 132,170
Denominator
Number of shares used in basic computation5,8228495,44512,116
Weighted-average effect of dilutive securities
Add:
Conversion of Class B to Class A shares outstanding84900(1)
Restricted stock units and other contingently issuable shares00114114
Number of shares used in per share computation6,6718495,55912,230
Diluted net income per share$ 10.82$ 10.81$ 10.80$ 10.81
", + "path": "tables/table-58 EPS Calculation.html", + "metadata": { + "length": 2392, + "summary": "table-59\nThe table shows basic and diluted net income per share for Class A, B, and C shares for the year ended December 31, 2025, with allocations of distributed and undistributed earnings, share counts, and resulting EPS values.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-58 EPS Calculation.html", + "keywords": [ + "net income", + "shares", + "EPS" + ], + "tokens": [] + } + }, + { + "chunk_id": "9ea7f890-c381-5f49-915e-17051ac64a4d", + "type": "text", + "content": "We compute net income per share of Class A, Class B, and Class C stock using the two-class method. Basic net income per share is computed using the weighted-average number of shares outstanding during the period. Diluted net income per share is computed using the weighted-average number of shares and the effect of potentially dilutive securities outstanding during the period. Potentially dilutive securities consist of RSUs and other contingently issuable shares. The dilutive effect of outstanding RSUs and other contingently issuable shares is reflected in diluted earnings per share by application of the treasury stock method. The computation of the diluted net income per share of Class A stock assumes the conversion of Class B stock, while the diluted net income per share of Class B stock does not assume the conversion of those shares.\nIn accordance with our certificate of incorporation, the rights, including the liquidation and dividend rights, of the holders of our Class A, Class B, and Class C stock are identical, except with respect to voting. Furthermore, there are a number of safeguards built into our certificate of incorporation, as well as Delaware law, which preclude our Board of Directors from declaring or paying unequal per share dividends on our Class A, Class B, and Class C stock. Specifically, Delaware law provides that amendments to our certificate of incorporation which would have the effect of adversely altering the rights, powers, or preferences of a given class of stock must be approved by the class of stock adversely affected by the proposed amendment. In addition, our certificate of incorporation provides that before any such amendment may be put to a stockholder vote, it must be approved by the unanimous consent of our Board of Directors.\nImmaterial differences in net income per share across our Class A, Class B, and Class C shares may arise due to the allocation of distributed earnings, which is based on the holders as of the record date, compared with the allocation of undistributed earnings and number of shares, which is based on the weighted average shares outstanding over the periods.\nThe following tables set forth the computation of basic and diluted net income per share of Class A, Class B, and Class C stock (in millions, except per share amounts):\n\n[tables/table-56 EPS Breakdown.html]\n\n(1) Not applicable for consolidated net income per share.\n\n[tables/table-57 EPS Calculation.html]\n\n(1) Not applicable for consolidated net income per share.\n\n[tables/table-58 EPS Calculation.html]\n\n(1) Not applicable for consolidated net income per share.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 12. Net Income Per Share", + "metadata": { + "length": 2638, + "summary": "The passage explains the computation of net income per share for Class A, Class B, and Class C stock using the two-class method. Basic EPS uses weighted-average shares outstanding, while diluted EPS includes potentially dilutive securities like RSUs via the treasury stock method. The rights of the three classes are identical except for voting, with safeguards preventing unequal dividends. Immaterial differences in EPS may arise from allocation of distributed versus undistributed earnings. The text references tables showing basic and diluted EPS calculations for multiple years.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "compute", + "net", + "income", + "share", + "Class", + "stock", + "class", + "method", + "Basic", + "computed", + "weighted", + "average", + "number", + "shares", + "outstanding", + "period", + "Diluted", + "effect", + "potentially", + "dilutive", + "securities", + "Potentially", + "consist", + "RSUs", + "contingently", + "issuable", + "The", + "reflected", + "diluted", + "earnings", + "application", + "treasury", + "computation", + "assumes", + "conversion", + "assume", + "In", + "accordance", + "certificate", + "incorporation", + "rights", + "including", + "liquidation", + "dividend", + "holders", + "identical", + "respect", + "voting", + "Furthermore", + "safeguards", + "built", + "Delaware", + "law", + "preclude", + "Board", + "Directors", + "declaring", + "paying", + "unequal", + "dividends", + "Specifically", + "amendments", + "adversely", + "altering", + "powers", + "preferences", + "approved", + "affected", + "proposed", + "amendment", + "addition", + "put", + "stockholder", + "vote", + "unanimous", + "consent", + "Immaterial", + "differences", + "arise", + "due", + "allocation", + "distributed", + "based", + "record", + "date", + "compared", + "undistributed", + "periods", + "tables", + "set", + "basic", + "millions", + "amounts", + "Not", + "applicable", + "consolidated" + ], + "keywords": [ + "net income per share", + "two-class method", + "diluted EPS" + ], + "connect_to": [ + { + "target": "b0745017-4108-5452-b2ed-e590fa29864b", + "relation": "embeds", + "ref": "[tables/table-56 EPS Breakdown.html]", + "position": { + "start": 2319, + "end": 2355 + } + }, + { + "target": "91aacc88-6e89-518e-8f58-13b93a494f1c", + "relation": "embeds", + "ref": "[tables/table-57 EPS Calculation.html]", + "position": { + "start": 2416, + "end": 2454 + } + }, + { + "target": "27725503-821e-5738-8db0-f52316729da0", + "relation": "embeds", + "ref": "[tables/table-58 EPS Calculation.html]", + "position": { + "start": 2515, + "end": 2553 + } + } + ] + } + }, + { + "chunk_id": "5b441737-ac75-5d50-ae6a-017dba5b4a90", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 13. Compensation Plans", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "bf282e69-0b25-511e-be42-debd4bc1d949", + "type": "text", + "content": "Our stock plans include the Alphabet Amended and Restated 2021 Stock Plan (\"Alphabet 2021 Stock Plan\") and Other Bets stock-based plans. Under our stock plans, RSUs and other types of awards may be granted. Under the Alphabet 2021 Stock Plan, an RSU award is an agreement to issue shares of our Class C stock at the time the award vests. RSUs generally vest over four years contingent upon employment on the vesting date. RSUs are awarded dividend equivalents, which are subject to the same vesting conditions as the underlying award, and settled in Class C shares.\nAs of December 31, 2025, there were 534 million shares of Class C stock reserved for future issuance under the Alphabet 2021 Stock Plan.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 13. Compensation Plans/Stock Plans", + "metadata": { + "length": 702, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "stock", + "plans", + "include", + "Alphabet", + "Amended", + "Restated", + "2021", + "Stock", + "Plan", + "Other", + "Bets", + "based", + "Under", + "RSUs", + "types", + "awards", + "granted", + "RSU", + "award", + "agreement", + "issue", + "shares", + "Class", + "time", + "vests", + "generally", + "vest", + "years", + "contingent", + "employment", + "vesting", + "date", + "awarded", + "dividend", + "equivalents", + "subject", + "conditions", + "underlying", + "settled", + "As", + "December", + "31", + "2025", + "534", + "million", + "reserved", + "future", + "issuance" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "90b8f517-413c-53ca-90aa-4d6e770eda80", + "type": "text", + "content": "For the years ended December 31, 2023, 2024, and 2025, total SBC expense was \\$22.1 billion, \\$22.8 billion, and \\$27.1 billion, including amounts associated with awards we expect to settle in Alphabet stock of \\$21.7 billion, \\$22.0 billion, and \\$24.1 billion, respectively.\nFor the years ended December 31, 2023, 2024, and 2025, we recognized tax benefits on total SBC expense, which are reflected in the provision for income taxes, of \\$4.5 billion, \\$4.6 billion, and \\$5.0 billion, respectively.\nFor the years ended December 31, 2023, 2024, and 2025, tax benefit realized related to awards vested or exercised during the period was \\$5.6 billion, \\$6.8 billion, and \\$8.1 billion, respectively. These amounts do not include the indirect effects of stock-based awards, which primarily relate to the research and development tax credit.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 13. Compensation Plans/Stock-Based Compensation", + "metadata": { + "length": 840, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "For", + "years", + "ended", + "December", + "31", + "2023", + "2024", + "2025", + "total", + "SBC", + "expense", + "22.1", + "billion", + "22.8", + "27.1", + "including", + "amounts", + "awards", + "expect", + "settle", + "Alphabet", + "stock", + "21.7", + "22.0", + "24.1", + "recognized", + "tax", + "benefits", + "reflected", + "provision", + "income", + "taxes", + "4.5", + "4.6", + "5.0", + "benefit", + "realized", + "related", + "vested", + "exercised", + "period", + "5.6", + "6.8", + "8.1", + "These", + "include", + "indirect", + "effects", + "based", + "primarily", + "relate", + "research", + "development", + "credit" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "85f94d24-2424-56d0-9ec0-fe3c4b199ab9", + "type": "table", + "content": "
Number of SharesWeighted-Average Grant-Date Fair Value
Unvested as of December 31, 2024299$ 122.77
Granted198$ 188.82
Vested(181)$ 133.90
Forfeited/canceled(34)$ 142.33
Unvested as of December 31, 2025282$ 159.75
", + "path": "tables/table-59 Stock Vesting.html", + "metadata": { + "length": 438, + "summary": "table-60\nThe table shows unvested shares changing from 299 to 282 during 2025, with 198 granted, 181 vested, and 34 forfeited. Weighted-average fair values are provided for each category.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-59 Stock Vesting.html", + "keywords": [ + "unvested shares", + "grant-date fair value", + "vesting" + ], + "tokens": [] + } + }, + { + "chunk_id": "af282188-be23-5510-802e-fc12116d1a2f", + "type": "text", + "content": "The following table summarizes the activities for unvested Alphabet RSUs, which include dividend equivalents awarded to holders of unvested stock, for the year ended December 31, 2025 (in millions, except per share amounts):\n\n[tables/table-59 Stock Vesting.html]\n\nThe weighted-average grant-date fair value of RSUs granted during the years ended December 31, 2023 and 2024 was \\$97.59 and \\$140.04, respectively. Total fair value of RSUs, as of their respective vesting dates, during the years ended December 31, 2023, 2024, and 2025, were \\$26.6 billion, \\$33.3 billion, and \\$39.7 billion, respectively.\nAs of December 31, 2025, there was \\$42.9 billion of unrecognized compensation cost related to unvested RSUs. This amount is expected to be recognized over a weighted-average period of 2.6 years.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 13. Compensation Plans/Stock-Based Award Activities", + "metadata": { + "length": 828, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "summarizes", + "activities", + "unvested", + "Alphabet", + "RSUs", + "include", + "dividend", + "equivalents", + "awarded", + "holders", + "stock", + "year", + "ended", + "December", + "31", + "2025", + "millions", + "share", + "amounts", + "weighted", + "average", + "grant", + "date", + "fair", + "granted", + "years", + "2023", + "2024", + "97.59", + "140.04", + "Total", + "respective", + "vesting", + "dates", + "26.6", + "billion", + "33.3", + "39.7", + "As", + "42.9", + "unrecognized", + "compensation", + "cost", + "related", + "This", + "amount", + "expected", + "recognized", + "period", + "2.6" + ], + "keywords": [], + "connect_to": [ + { + "target": "85f94d24-2424-56d0-9ec0-fe3c4b199ab9", + "relation": "embeds", + "ref": "[tables/table-59 Stock Vesting.html]", + "position": { + "start": 226, + "end": 262 + } + } + ] + } + }, + { + "chunk_id": "f66ea29c-44a6-5e42-8068-0295a35bfb25", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Domestic operations$ 73,600$ 108,076$ 143,591
Foreign operations12,11711,73915,235
Total$ 85,717$ 119,815$ 158,826
", + "path": "tables/table-60 Revenue Breakdown.html", + "metadata": { + "length": 384, + "summary": "table-61\nThe table shows total revenue from domestic and foreign operations for 2023-2025, increasing from $85,717 to $158,826.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-60 Revenue Breakdown.html", + "keywords": [ + "Revenue", + "Domestic", + "Foreign" + ], + "tokens": [] + } + }, + { + "chunk_id": "861ebe58-6ff6-5ecd-a9e9-7ef88de801fa", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Current:
Federal and state$ 15,716$ 21,101$ 13,378
Foreign3,9353,8525,028
Total19,65124,95318,406
Deferred:
Federal and state(7,482)(5,350)8,243
Foreign(247)947
Total(7,729)(5,256)8,250
Provision for income taxes$ 11,922$ 19,697$ 26,656
", + "path": "tables/table-61 Income Tax Provision.html", + "metadata": { + "length": 770, + "summary": "table-62\nThe table shows current and deferred income tax provisions for 2023-2025. Current taxes total $19,651, $24,953, $18,406; deferred taxes total ($7,729), ($5,256), $8,250; total provision is $11,922, $19,697, $26,656.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-61 Income Tax Provision.html", + "keywords": [ + "tax provision", + "current", + "deferred" + ], + "tokens": [] + } + }, + { + "chunk_id": "95d1b703-a13d-5e30-a147-c52e45235b9d", + "type": "table", + "content": "
Year Ended December 31,
202320242025
US federal statutory rate18,00121.0 %25,16121.0 %33,35321.0 %
State and local income taxes, net of federal income tax effect(1)8231.0 %1,1991.0 %1,6061.0 %
Foreign tax effects:
Brazil:
Withholding taxes1,0641.2 %1,0410.9 %1,3840.9 %
Other620.1 %120.0 %230.0 %
Other foreign jurisdictions(74)(0.1)%3530.3 %3960.2 %
Effect of change in tax laws or rates enacted in the current period(829)(1.0)%00.0 %00.0 %
Effect of cross-border tax laws:
Foreign-derived intangible income deduction(3,980)(4.6)%(4,568)(3.8)%(3,931)(2.5)%
Other2150.2 %3210.3 %2950.2 %
Tax credits:
Federal research credit(1,575)(1.8)%(1,792)(1.5)%(2,088)(1.3)%
Foreign tax credits(1,396)(1.6)%(1,373)(1.1)%(1,684)(1.1)%
Other(498)(0.6)%(198)(0.2)%(98)(0.1)%
Changes in valuation allowances5130.6 %6030.5 %1,1700.7 %
Nontaxable or nondeductible items:
Stock-based compensation expense(602)(0.7)%(1,743)(1.5)%(2,601)(1.6)%
Other1690.2 %2030.2 %9550.6 %
Changes in unrecognized tax benefits4320.5 %6890.6 %(1,123)(0.7)%
Other adjustments(403)(0.5)%(211)(0.2)%(1,002)(0.6)%
Total$ 11,92213.9 %$ 19,69716.4 %$ 26,65616.8 %
", + "path": "tables/table-62 Tax Rate Reconciliation.html", + "metadata": { + "length": 2819, + "summary": "table-63\nThe table shows a reconciliation of the US federal statutory income tax rate to the effective tax rate for 2023-2025, listing adjustments like state taxes, foreign effects, credits, and valuation allowances.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-62 Tax Rate Reconciliation.html", + "keywords": [ + "tax rate", + "reconciliation", + "income" + ], + "tokens": [] + } + }, + { + "chunk_id": "9b2b554f-311a-504b-a32c-81537c2d8eb1", + "type": "text", + "content": "Income from continuing operations before income taxes consisted of the following (in millions):\n\n[tables/table-60 Revenue Breakdown.html]\n\nProvision for income taxes consisted of the following (in millions):\n\n[tables/table-61 Income Tax Provision.html]\n\nThe reconciliation of federal statutory income tax rate to our effective income tax rate was as follows:\n\n[tables/table-62 Tax Rate Reconciliation.html]\n\n(1) The tax effect in this category primarily reflects state and local taxes in New York state, New York city, Pennsylvania, Minnesota, Illinois, New Jersey and Wisconsin.\nIn 2023, the IRS issued a rule change allowing taxpayers to temporarily apply the regulations in effect prior to 2022 related to US federal foreign tax credits as well as a separate rule change with guidance on the capitalization and amortization of research and development expenses. A cumulative one-time adjustment for these tax rule changes was recorded in 2023.\nChanges to US tax law enacted on July 4, 2025, allow for immediate expensing of domestic research and experimentation costs, accelerated depreciation on eligible capital expenditures, and other tax law changes impacting 2025 with certain changes effective in 2026. These changes are reflected in our results for the year ended December 31, 2025.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 14. Income Taxes", + "metadata": { + "length": 1301, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Income", + "continuing", + "operations", + "income", + "taxes", + "consisted", + "millions", + "Provision", + "The", + "reconciliation", + "federal", + "statutory", + "tax", + "rate", + "effective", + "effect", + "category", + "primarily", + "reflects", + "state", + "local", + "New", + "York", + "city", + "Pennsylvania", + "Minnesota", + "Illinois", + "Jersey", + "Wisconsin", + "In", + "2023", + "IRS", + "issued", + "rule", + "change", + "allowing", + "taxpayers", + "temporarily", + "apply", + "regulations", + "prior", + "2022", + "related", + "US", + "foreign", + "credits", + "separate", + "guidance", + "capitalization", + "amortization", + "research", + "development", + "expenses", + "cumulative", + "time", + "adjustment", + "recorded", + "Changes", + "law", + "enacted", + "July", + "2025", + "expensing", + "domestic", + "experimentation", + "costs", + "accelerated", + "depreciation", + "eligible", + "capital", + "expenditures", + "impacting", + "2026", + "These", + "reflected", + "results", + "year", + "ended", + "December", + "31" + ], + "keywords": [], + "connect_to": [ + { + "target": "f66ea29c-44a6-5e42-8068-0295a35bfb25", + "relation": "embeds", + "ref": "[tables/table-60 Revenue Breakdown.html]", + "position": { + "start": 97, + "end": 137 + } + }, + { + "target": "861ebe58-6ff6-5ecd-a9e9-7ef88de801fa", + "relation": "embeds", + "ref": "[tables/table-61 Income Tax Provision.html]", + "position": { + "start": 209, + "end": 252 + } + }, + { + "target": "95d1b703-a13d-5e30-a147-c52e45235b9d", + "relation": "embeds", + "ref": "[tables/table-62 Tax Rate Reconciliation.html]", + "position": { + "start": 360, + "end": 406 + } + } + ] + } + }, + { + "chunk_id": "a52ec118-6577-53c0-93b4-25e06b8f7439", + "type": "table", + "content": "
As of December 31,
20242025
Deferred tax assets:
Accrued employee benefits$ 1,834$ 1,951
Accruals and reserves not currently deductible2,5523,570
Tax credits6,3847,314
Net operating losses3,4724,953
Operating leases3,3363,337
Capitalized research and development25,90324,758
Other1,3762,143
Total deferred tax assets44,85748,026
Valuation allowance(11,493)(13,942)
Total deferred tax assets net of valuation allowance33,36434,084
Deferred tax liabilities:
Property and equipment, net(9,932)(13,256)
Net investment gains(2,978)(8,242)
Operating leases(2,986)(3,103)
Other(1,008)(1,289)
Total deferred tax liabilities(16,904)(25,890)
Net deferred tax assets (liabilities)$ 16,460$ 8,194
", + "path": "tables/table-63 Deferred Tax Assets.html", + "metadata": { + "length": 1406, + "summary": "table-64\nThe table shows deferred tax assets and liabilities for 2024 and 2025. Total deferred tax assets net of valuation allowance were $33,364 in 2024 and $34,084 in 2025. Net deferred tax assets decreased from $16,460 to $8,194.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-63 Deferred Tax Assets.html", + "keywords": [ + "deferred tax", + "valuation allowance", + "net DTA" + ], + "tokens": [] + } + }, + { + "chunk_id": "5d7e2613-c86c-5780-88e6-bce9040c95d8", + "type": "table", + "content": "
Year Ended December 31,
202320242025
US federal$ 13,689$ 19,921$ 13,658
US state and local1,2242,6972,919
Foreign:
Brazil1,2641,1011,368
Other2,9873,6343,581
Total foreign4,2514,7354,949
Total cash paid for income taxes, net of refunds$ 19,164$ 27,353$ 21,526
", + "path": "tables/table-64 Income Tax Payments.html", + "metadata": { + "length": 668, + "summary": "table-65\nThe table shows cash paid for income taxes (net of refunds) for 2023-2025: US federal, state/local, and foreign (Brazil, Other). Total: $19,164 (2023), $27,353 (2024), $21,526 (2025).", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-64 Income Tax Payments.html", + "keywords": [ + "income tax", + "cash paid", + "tax breakdown" + ], + "tokens": [] + } + }, + { + "chunk_id": "c0cec73d-62b3-5dc7-8868-a93d4d725875", + "type": "text", + "content": "Deferred income taxes reflect the net effects of temporary differences between the carrying amounts of assets and liabilities for financial reporting purposes and the amounts used for income tax purposes. Significant components of our deferred tax assets and liabilities were as follows (in millions):\n\n[tables/table-63 Deferred Tax Assets.html]\n\nAs of December 31, 2025, our federal, state, and foreign net operating loss carryforwards for income tax purposes were approximately \\$13.0 billion, \\$25.1 billion, and \\$2.9 billion respectively. If not utilized, the federal, foreign and state net operating loss carryforwards will all begin to expire in 2026. It is more likely than not that the majority of the net operating loss carryforwards will not be realized. The net operating loss carryforwards are subject to various annual limitations under the tax laws of the different jurisdictions.\nAs of December 31, 2025, our Federal and California research and development credit carryforwards for income tax purposes were approximately \\$771 million and \\$6.4 billion, respectively. If not utilized, the Federal research and development credit will begin to expire in 2037 and the California research and development credit can be carried over indefinitely. We believe the majority of the federal tax credit and state tax credit is not likely to be realized.\nAs of December 31, 2025, our investment tax credit carryforwards for state income tax purposes were approximately \\$1.3 billion and will begin to expire in 2033. We use the flow-through method of accounting for investment tax credits. We believe this tax credit is not likely to be realized.\nAs of December 31, 2025, we maintained a valuation allowance with respect to California deferred tax assets, certain federal net operating losses, certain state net operating losses and tax credits, net deferred tax assets relating to certain Other Bet companies, and certain foreign net operating losses that we believe are not likely to be realized. We continue to reassess the remaining valuation allowance quarterly, and if future evidence allows for a partial or full release of the valuation allowance, a tax benefit will be recorded accordingly.\nCash paid for income taxes, net of refunds, were as follows (in millions):\n\n[tables/table-64 Income Tax Payments.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 14. Income Taxes/Deferred Income Taxes", + "metadata": { + "length": 2326, + "summary": "The passage discusses deferred income taxes, focusing on temporary differences between financial reporting and tax purposes. As of December 31, 2025, the company has significant net operating loss carryforwards: federal $13.0 billion, state $25.1 billion, and foreign $2.9 billion, all beginning to expire in 2026, with most unlikely to be realized. Research and development credits include federal $771 million (expiring 2037) and California $6.4 billion (indefinite), also mostly unrealizable. Investment tax credits of $1.3 billion expire in 2033. A valuation allowance is maintained for California deferred tax assets, certain federal and state NOLs, tax credits, and foreign NOLs deemed unlikely to be realized. The allowance is reassessed quarterly. Cash paid for income taxes is presented in a referenced table.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Deferred", + "income", + "taxes", + "reflect", + "net", + "effects", + "temporary", + "differences", + "carrying", + "amounts", + "assets", + "liabilities", + "financial", + "reporting", + "purposes", + "tax", + "Significant", + "components", + "deferred", + "millions", + "As", + "December", + "31", + "2025", + "federal", + "state", + "foreign", + "operating", + "loss", + "carryforwards", + "approximately", + "13.0", + "billion", + "25.1", + "2.9", + "If", + "utilized", + "begin", + "expire", + "2026", + "It", + "majority", + "realized", + "The", + "subject", + "annual", + "limitations", + "laws", + "jurisdictions", + "Federal", + "California", + "research", + "development", + "credit", + "771", + "million", + "6.4", + "2037", + "carried", + "indefinitely", + "We", + "investment", + "1.3", + "2033", + "flow", + "method", + "accounting", + "credits", + "maintained", + "valuation", + "allowance", + "respect", + "losses", + "relating", + "Other", + "Bet", + "companies", + "continue", + "reassess", + "remaining", + "quarterly", + "future", + "evidence", + "partial", + "full", + "release", + "benefit", + "recorded", + "Cash", + "paid", + "refunds" + ], + "keywords": [ + "deferred taxes", + "NOL carryforwards", + "valuation allowance" + ], + "connect_to": [ + { + "target": "a52ec118-6577-53c0-93b4-25e06b8f7439", + "relation": "embeds", + "ref": "[tables/table-63 Deferred Tax Assets.html]", + "position": { + "start": 303, + "end": 345 + } + }, + { + "target": "5d7e2613-c86c-5780-88e6-bce9040c95d8", + "relation": "embeds", + "ref": "[tables/table-64 Income Tax Payments.html]", + "position": { + "start": 2281, + "end": 2323 + } + } + ] + } + }, + { + "chunk_id": "9feb0c3b-d76c-58fb-b466-48c594a35674", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Beginning gross unrecognized tax benefits$ 7,055$ 9,438$ 12,619
Increases related to prior year tax positions740896278
Decreases related to prior year tax positions(682)(83)(1,301)
Decreases related to settlement with tax authorities(21)(311)(2,183)
Increases related to current year tax positions2,3462,6792,099
Ending gross unrecognized tax benefits$ 9,438$ 12,619$ 11,512
", + "path": "tables/table-65 Tax Benefits.html", + "metadata": { + "length": 776, + "summary": "table-66\nThe table shows changes in gross unrecognized tax benefits from 2023 to 2025, including increases and decreases from prior and current year positions and settlements, ending at $11,512 in 2025.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-65 Tax Benefits.html", + "keywords": [ + "unrecognized tax benefits", + "tax positions", + "settlements" + ], + "tokens": [] + } + }, + { + "chunk_id": "83c3330c-0ed3-52ff-8b90-4ce949b45540", + "type": "text", + "content": "The following table summarizes the activity related to our gross unrecognized tax benefits (in millions):\n\n[tables/table-65 Tax Benefits.html]\n\nWe are subject to income taxes in the US and foreign jurisdictions. Significant judgment is required in evaluating our uncertain tax positions and determining our provision for income taxes. The total amount of gross unrecognized tax benefits was \\$9.4 billion, \\$12.6 billion, and \\$11.5 billion as of December 31, 2023, 2024, and 2025, respectively, of which \\$7.4 billion, \\$10.0 billion, and \\$9.7 billion, if recognized, would affect our effective tax rate, respectively.\nAs of December 31, 2024 and 2025, we accrued \\$1.1 billion and \\$1.2 billion in interest and penalties in provision for income taxes, respectively.\nWe are subject to the continuous examination of our income tax returns by the IRS and other tax authorities. The IRS is currently examining our 2019 through 2021 tax returns. We have also received tax assessments in multiple foreign jurisdictions asserting transfer pricing adjustments or permanent establishment. We continue to defend such claims as presented.\nWe regularly assess the likelihood of adverse outcomes resulting from these examinations to determine the adequacy of our provision for income taxes. We continue to monitor the progress of ongoing discussions with tax authorities and the effect, if any, of the expected expiration of the statute of limitations in various taxing jurisdictions.\nWe believe that an adequate provision has been made for any adjustments that may result from tax examinations. However, the outcome of tax audits cannot be predicted with certainty. If any issues addressed in our tax audits are resolved in a manner not consistent with management's expectations, we could be required to adjust our provision for income taxes in the period such resolutions occur.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 14. Income Taxes/Uncertain Tax Positions", + "metadata": { + "length": 1881, + "summary": "The passage discusses the company's gross unrecognized tax benefits, which were $9.4 billion, $12.6 billion, and $11.5 billion as of December 31, 2023, 2024, and 2025, respectively. A significant portion would affect the effective tax rate if recognized. Interest and penalties accrued were $1.1 billion (2024) and $1.2 billion (2025). The company is under IRS examination for 2019-2021 returns and faces transfer pricing or permanent establishment assessments in foreign jurisdictions. Management regularly assesses outcomes and believes adequate provisions have been made, though audit results remain uncertain.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "table", + "summarizes", + "activity", + "related", + "gross", + "unrecognized", + "tax", + "benefits", + "millions", + "We", + "subject", + "income", + "taxes", + "US", + "foreign", + "jurisdictions", + "Significant", + "judgment", + "required", + "evaluating", + "uncertain", + "positions", + "determining", + "provision", + "total", + "amount", + "9.4", + "billion", + "12.6", + "11.5", + "December", + "31", + "2023", + "2024", + "2025", + "7.4", + "10.0", + "9.7", + "recognized", + "affect", + "effective", + "rate", + "As", + "accrued", + "1.1", + "1.2", + "interest", + "penalties", + "continuous", + "examination", + "returns", + "IRS", + "authorities", + "examining", + "2019", + "2021", + "received", + "assessments", + "multiple", + "asserting", + "transfer", + "pricing", + "adjustments", + "permanent", + "establishment", + "continue", + "defend", + "claims", + "presented", + "regularly", + "assess", + "likelihood", + "adverse", + "outcomes", + "resulting", + "examinations", + "determine", + "adequacy", + "monitor", + "progress", + "ongoing", + "discussions", + "effect", + "expected", + "expiration", + "statute", + "limitations", + "taxing", + "adequate", + "made", + "result", + "However", + "outcome", + "audits", + "predicted", + "certainty", + "If", + "issues", + "addressed", + "resolved", + "manner", + "consistent", + "management", + "expectations", + "adjust", + "period", + "resolutions", + "occur" + ], + "keywords": [ + "unrecognized tax benefits", + "income tax audits", + "transfer pricing" + ], + "connect_to": [ + { + "target": "9feb0c3b-d76c-58fb-b466-48c594a35674", + "relation": "embeds", + "ref": "[tables/table-65 Tax Benefits.html]", + "position": { + "start": 107, + "end": 142 + } + } + ] + } + }, + { + "chunk_id": "16709af2-137b-5cc7-ac98-2bb8d6910538", + "type": "table", + "content": "
Year Ended December 31,
202320242025
Revenues:
Google Services$ 272,543$ 304,930$ 342,721
Google Cloud33,08843,22958,705
Other Bets1,5271,6481,537
Hedging gains (losses)236211(127)
Total revenues$ 307,394$ 350,018$ 402,836
Operating income (loss):
Google Services$ 95,858$ 121,263$ 139,404
Google Cloud1,7166,11213,910
Other Bets(4,095)(4,444)(7,515)
Alphabet-level activities(9,186)(10,541)(16,760)
Total income from operations$ 84,293$ 112,390$ 129,039
Supplemental information about segment expenses:
Google Services:
Employee compensation expenses$ 46,224$ 44,560$ 45,124
Other costs and expenses130,461139,107158,193
Total Google Services costs and expenses$ 176,685$ 183,667$ 203,317
Google Cloud:
Employee compensation expenses$ 19,054$ 20,519$ 22,078
Other costs and expenses12,31816,59822,717
Total Google Cloud costs and expenses$ 31,372$ 37,117$ 44,795
", + "path": "tables/table-66 Alphabet Revenue.html", + "metadata": { + "length": 1813, + "summary": "table-67\nAlphabet's total revenues grew from $307B in 2023 to $403B in 2025, driven by Google Services and Cloud. Operating income rose from $84B to $129B, with Google Cloud turning profitable.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-66 Alphabet Revenue.html", + "keywords": [ + "Revenue", + "Operating Income", + "Segment" + ], + "tokens": [] + } + }, + { + "chunk_id": "b2b6f826-08c4-594e-81f4-6ebdd2184358", + "type": "table", + "content": "
As of December 31,
20242025
Long-lived assets:
United States$ 138,993$ 195,337
International45,63166,481
Total long-lived assets$ 184,624$ 261,818
", + "path": "tables/table-67 Long-lived Assets.html", + "metadata": { + "length": 378, + "summary": "table-68\nThe table shows long-lived assets as of December 31, 2024 and 2025. United States assets grew from $138,993 to $195,337, International from $45,631 to $66,481, totaling $184,624 and $261,818 respectively.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-67 Long-lived Assets.html", + "keywords": [ + "Long-lived assets", + "United States", + "International" + ], + "tokens": [] + } + }, + { + "chunk_id": "221efafa-075f-5748-a22a-8ddec1063ee9", + "type": "text", + "content": "We report our segment results as Google Services, Google Cloud, and Other Bets:\n- Google Services includes products and services such as ads, Android, Chrome, devices, Google Maps, Google Play, Search, and YouTube. Google Services generates revenues primarily from advertising; fees received for consumer subscription-based products such as YouTube TV, YouTube Music and Premium, and NFL Sunday Ticket, as well as Google One; the sale of apps and in-app purchases; and devices.\n- Google Cloud includes infrastructure and platform services, applications, and other services for enterprise customers. Google Cloud generates revenues primarily from consumption-based fees and subscriptions received for Google Cloud Platform services, Google Workspace communication and collaboration tools, and other enterprise services.\n\\- Other Bets is a combination of multiple operating segments that are not individually material. Revenues from Other Bets are generated primarily from the sale of autonomous transportation services and internet services.\nRevenues, certain costs, such as costs associated with content and traffic acquisition, certain engineering activities, and devices, as well as certain operating expenses are directly attributable to our segments. Due to the integrated nature of Alphabet, other costs and expenses, such as technical infrastructure and office facilities, are managed centrally at a consolidated level. These costs, including the associated depreciation, are allocated to operating segments as a service cost generally based on usage, headcount, or revenue.\nCertain costs are not allocated to our segments because they represent Alphabet-level activities. These costs primarily include:\n- certain AI-focused shared research and development activities, including employee compensation expenses and technical infrastructure usage costs associated with the development of our general AI models;\n• corporate initiatives such as our philanthropic activities; and\n• corporate shared costs such as certain finance, human resource, and legal costs, including certain fines and settlements.\nCharges associated with employee severance and office space reductions are also not allocated to our segments. Additionally, hedging gains (losses) related to revenue are not allocated to our segments.\nOur Chief Operating Decision Maker (CODM) is our Chief Executive Officer, Sundar Pichai. Our CODM uses segment operating income (loss) to allocate resources to our segments in our annual planning process and to assess the performance of our segments, primarily by monitoring actual results versus the annual plan. Our operating segments are not evaluated using asset information.\nThe following table presents revenue, profitability, and expense information about our segments (in millions):\n\n[tables/table-66 Alphabet Revenue.html]\n\nGoogle Services and Google Cloud employee compensation expenses include the costs associated with direct and allocated employees. Google Services and Google Cloud other costs and expenses primarily include direct costs, such as advertising and promotional activities, legal and other matters, and third-party services fees as well as allocated costs, such as technical infrastructure and office facilities usage costs. Additionally, Google Services other costs and expenses include content and traffic acquisition costs and device costs.\nSee Note 2 for further details relating to revenues by geography.\nThe following table presents long-lived assets by geographic area, which includes property and equipment, net and operating lease assets (in millions):\n\n[tables/table-67 Long-lived Assets.html]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 15. Information about Segments and Geographic Areas", + "metadata": { + "length": 3645, + "summary": "The text describes Alphabet's three reportable segments: Google Services (ads, subscriptions, devices), Google Cloud (enterprise services), and Other Bets (autonomous transportation, internet services). It explains revenue sources, cost allocation methods (direct vs. centrally managed), and unallocated costs like AI R&D and corporate initiatives. The CODM uses segment operating income for resource allocation and performance assessment. The passage references tables for segment financials and long-lived assets by geography.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "report", + "segment", + "results", + "Google", + "Services", + "Cloud", + "Other", + "Bets", + "includes", + "products", + "services", + "ads", + "Android", + "Chrome", + "devices", + "Maps", + "Play", + "Search", + "YouTube", + "generates", + "revenues", + "primarily", + "advertising", + "fees", + "received", + "consumer", + "subscription", + "based", + "TV", + "Music", + "Premium", + "NFL", + "Sunday", + "Ticket", + "One", + "sale", + "apps", + "app", + "purchases", + "infrastructure", + "platform", + "applications", + "enterprise", + "customers", + "consumption", + "subscriptions", + "Platform", + "Workspace", + "communication", + "collaboration", + "tools", + "combination", + "multiple", + "operating", + "segments", + "individually", + "material", + "Revenues", + "generated", + "autonomous", + "transportation", + "internet", + "costs", + "content", + "traffic", + "acquisition", + "engineering", + "activities", + "expenses", + "directly", + "attributable", + "Due", + "integrated", + "nature", + "Alphabet", + "technical", + "office", + "facilities", + "managed", + "centrally", + "consolidated", + "level", + "These", + "including", + "depreciation", + "allocated", + "service", + "cost", + "generally", + "usage", + "headcount", + "revenue", + "Certain", + "represent", + "include", + "AI", + "focused", + "shared", + "research", + "development", + "employee", + "compensation", + "general", + "models", + "corporate", + "initiatives", + "philanthropic", + "finance", + "human", + "resource", + "legal", + "fines", + "settlements", + "Charges", + "severance", + "space", + "reductions", + "Additionally", + "hedging", + "gains", + "losses", + "related", + "Our", + "Chief", + "Operating", + "Decision", + "Maker", + "CODM", + "Executive", + "Officer", + "Sundar", + "Pichai", + "income", + "loss", + "allocate", + "resources", + "annual", + "planning", + "process", + "assess", + "performance", + "monitoring", + "actual", + "versus", + "plan", + "evaluated", + "asset", + "information", + "The", + "table", + "presents", + "profitability", + "expense", + "millions", + "direct", + "employees", + "promotional", + "matters", + "party", + "device", + "See", + "Note", + "details", + "relating", + "geography", + "long", + "lived", + "assets", + "geographic", + "area", + "property", + "equipment", + "net", + "lease" + ], + "keywords": [ + "Google Services", + "Google Cloud", + "Other Bets" + ], + "connect_to": [ + { + "target": "16709af2-137b-5cc7-ac98-2bb8d6910538", + "relation": "embeds", + "ref": "[tables/table-66 Alphabet Revenue.html]", + "position": { + "start": 2799, + "end": 2838 + } + }, + { + "target": "b2b6f826-08c4-594e-81f4-6ebdd2184358", + "relation": "embeds", + "ref": "[tables/table-67 Long-lived Assets.html]", + "position": { + "start": 3597, + "end": 3637 + } + } + ] + } + }, + { + "chunk_id": "b0321c4f-8f9c-528d-844e-e3c549a64bb7", + "type": "text", + "content": "In January 2026, we recognized approximately \\$32.0 billion of unrealized gains in our non-marketable investments. These unrealized gains reflect an estimated increase in the fair value measurement following observable transactions that occurred in January 2026, and are subject to change as we finalize related valuations. See Note 3 and Note 7 for further details on equity investments and OI&E.\nITEM 9. CHANGES IN AND DISAGREEMENTS WITH ACCOUNTANTS ON ACCOUNTING AND FINANCIAL DISCLOSURE\nNone.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 16. Subsequent Event", + "metadata": { + "length": 496, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "In", + "January", + "2026", + "recognized", + "approximately", + "32.0", + "billion", + "unrealized", + "gains", + "marketable", + "investments", + "These", + "reflect", + "estimated", + "increase", + "fair", + "measurement", + "observable", + "transactions", + "occurred", + "subject", + "change", + "finalize", + "related", + "valuations", + "See", + "Note", + "details", + "equity", + "OI", + "ITEM", + "CHANGES", + "IN", + "AND", + "DISAGREEMENTS", + "WITH", + "ACCOUNTANTS", + "ON", + "ACCOUNTING", + "FINANCIAL", + "DISCLOSURE", + "None" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f79e5f7f-6ebf-58cc-bc85-2994df22f5e0", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9A. CONTROLS AND PROCEDURES", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "72d13b7e-befe-57df-bf33-24ce62a0c074", + "type": "text", + "content": "Our management, with the participation of our chief executive officer and chief financial officer, evaluated the effectiveness of our disclosure controls and procedures pursuant to Rule 13a-15 under the Exchange Act, as of the end of the period covered by this Annual Report on Form 10-K.\nBased on this evaluation, our chief executive officer and chief financial officer concluded that, as of December 31, 2025, our disclosure controls and procedures are designed at a reasonable assurance level and are effective to provide reasonable assurance that information we are required to disclose in reports that we file or submit under the Exchange Act is recorded, processed, summarized, and reported within the time periods specified in the SEC's rules and forms, and that such information is accumulated and communicated to our management, including our chief executive officer and chief financial officer, as appropriate, to allow timely decisions regarding required disclosure.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9A. CONTROLS AND PROCEDURES/Evaluation of Disclosure Controls and Procedures", + "metadata": { + "length": 977, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "management", + "participation", + "chief", + "executive", + "officer", + "financial", + "evaluated", + "effectiveness", + "disclosure", + "controls", + "procedures", + "pursuant", + "Rule", + "13a", + "15", + "Exchange", + "Act", + "end", + "period", + "covered", + "Annual", + "Report", + "Form", + "10", + "Based", + "evaluation", + "concluded", + "December", + "31", + "2025", + "designed", + "reasonable", + "assurance", + "level", + "effective", + "provide", + "information", + "required", + "disclose", + "reports", + "file", + "submit", + "recorded", + "processed", + "summarized", + "reported", + "time", + "periods", + "SEC", + "rules", + "forms", + "accumulated", + "communicated", + "including", + "timely", + "decisions" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "770a25c9-21f8-5fee-8e3e-ea0bec58decc", + "type": "text", + "content": "There have been no changes in our internal control over financial reporting that occurred during the quarter ended December 31, 2025 that have materially affected, or are reasonably likely to materially affect, our internal control over financial reporting.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9A. CONTROLS AND PROCEDURES/Changes in Internal Control over Financial Reporting", + "metadata": { + "length": 257, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "There", + "internal", + "control", + "financial", + "reporting", + "occurred", + "quarter", + "ended", + "December", + "31", + "2025", + "materially", + "affected", + "affect" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cd7f642f-1acc-549e-bb6f-2651442e8a3e", + "type": "text", + "content": "Our management is responsible for establishing and maintaining adequate internal control over financial reporting, as defined in Rule 13a-15(f) of the Exchange Act. Our management conducted an evaluation of the effectiveness of our internal control over financial reporting based on the framework in Internal Control—Integrated Framework issued by the Committee of Sponsoring Organizations of the Treadway Commission (2013 framework). Based on this evaluation, management concluded that our internal control over financial reporting was effective as of December 31, 2025. Management reviewed the results of its assessment with our Audit Committee. The effectiveness of our internal control over financial reporting as of December 31, 2025 has been audited by Ernst & Young LLP, an independent registered public accounting firm, as stated in its report which is included in Item 8 of this Annual Report on Form 10-K.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9A. CONTROLS AND PROCEDURES/Management's Report on Internal Control over Financial Reporting", + "metadata": { + "length": 915, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Our", + "management", + "responsible", + "establishing", + "maintaining", + "adequate", + "internal", + "control", + "financial", + "reporting", + "defined", + "Rule", + "13a", + "15", + "Exchange", + "Act", + "conducted", + "evaluation", + "effectiveness", + "based", + "framework", + "Internal", + "Control", + "Integrated", + "Framework", + "issued", + "Committee", + "Sponsoring", + "Organizations", + "Treadway", + "Commission", + "2013", + "Based", + "concluded", + "effective", + "December", + "31", + "2025", + "Management", + "reviewed", + "results", + "assessment", + "Audit", + "The", + "audited", + "Ernst", + "Young", + "LLP", + "independent", + "registered", + "public", + "accounting", + "firm", + "stated", + "report", + "included", + "Item", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8b0d135a-ac28-5cfc-b9f3-ea1d14f71fc0", + "type": "text", + "content": "In designing and evaluating the disclosure controls and procedures, management recognizes that any controls and procedures, no matter how well designed and operated, can provide only reasonable assurance of achieving the desired control objectives. In addition, the design of disclosure controls and procedures must reflect the fact that there are resource constraints and that management is required to apply its judgment in evaluating the benefits of possible controls and procedures relative to their costs.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9A. CONTROLS AND PROCEDURES/Limitations on Effectiveness of Controls and Procedures", + "metadata": { + "length": 510, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "In", + "designing", + "evaluating", + "disclosure", + "controls", + "procedures", + "management", + "recognizes", + "matter", + "designed", + "operated", + "provide", + "reasonable", + "assurance", + "achieving", + "desired", + "control", + "objectives", + "addition", + "design", + "reflect", + "fact", + "resource", + "constraints", + "required", + "apply", + "judgment", + "benefits", + "relative", + "costs" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c0fd0d75-4689-5c53-a935-0f31a392520d", + "type": "text", + "content": "10b5-1 Trading Plans\nDuring the quarter ended December 31, 2025, the following Section 16 director and officer adopted, modified, or terminated a \"Rule 10b5-1 trading arrangement\" (as defined in Item 408 of Regulation S-K of the Exchange Act):\n- John Hennessy, Chair of the Board of Directors, through the John L. Hennessy & Andrea J. Hennessy Revocable Trust, adopted a new trading plan on November 10, 2025 (with the first trade under the new plan scheduled for March 15, 2026). The trading plan will be effective until March 15, 2027 to sell up to 8,400 shares of Class C Capital Stock and up to 4,200 shares of Class A Common Stock.\n- Ruth M. Porat, President and Chief Investment Officer, adopted a new trading plan on November 29, 2025 (with the first trade under the new plan scheduled for March 2, 2026). The trading plan is scheduled to be in effect until March 2, 2027 to sell up to 154,486 shares (gross, plus any dividend equivalent units) of Class C Capital Stock issued upon the vesting of Ruth's Alphabet 2021 Performance Stock Units, as adjusted based on performance (shares sold are net of tax withholding).\nThere were no \"non-Rule 10b5-1 trading arrangements\" (as defined in Item 408 of Regulation S-K of the Exchange Act) adopted, modified, or terminated during the quarter ended December 31, 2025 by our directors and Section 16 officers. Each of the Rule 10b5-1 trading arrangements are in accordance with our Policy Against Insider\nTrading and actual sale transactions made pursuant to such trading arrangements will be disclosed publicly in Section 16 filings with the SEC in accordance with applicable securities laws, rules, and regulations.\nRequired Disclosure Pursuant to Section 13(r) of the Exchange Act\nAs previously disclosed, Google LLC, a subsidiary of Alphabet, filed notifications with the Russian Federal Security Service (FSB) pursuant to Russian encryption control requirements, which must be complied with prior to the import of covered items. The information provided pursuant to Section 13(r) of the Exchange Act in Part II, Item 5 of our Quarterly Report on Form 10-Q for the quarter ended September 30, 2025 is incorporated herein by reference.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9B. OTHER INFORMATION", + "metadata": { + "length": 2187, + "summary": "During the quarter ended December 31, 2025, two Section 16 officers adopted new Rule 10b5-1 trading plans. John Hennessy, Chair of the Board, adopted a plan on November 10, 2025 to sell up to 8,400 shares of Class C Capital Stock and up to 4,200 shares of Class A Common Stock, effective until March 15, 2027. Ruth M. Porat, President and Chief Investment Officer, adopted a plan on November 29, 2025 to sell up to 154,486 shares of Class C Capital Stock upon vesting of performance stock units, effective until March 2, 2027. No non-Rule 10b5-1 trading arrangements were adopted, modified, or terminated. The disclosure also references required reporting under Section 13(r) of the Exchange Act regarding Russian encryption control notifications.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "10b5", + "Trading", + "Plans", + "During", + "quarter", + "ended", + "December", + "31", + "2025", + "Section", + "16", + "director", + "officer", + "adopted", + "modified", + "terminated", + "Rule", + "trading", + "arrangement", + "defined", + "Item", + "408", + "Regulation", + "Exchange", + "Act", + "John", + "Hennessy", + "Chair", + "Board", + "Directors", + "Andrea", + "Revocable", + "Trust", + "plan", + "November", + "10", + "trade", + "scheduled", + "March", + "15", + "2026", + "The", + "effective", + "2027", + "sell", + "400", + "shares", + "Class", + "Capital", + "Stock", + "200", + "Common", + "Ruth", + "Porat", + "President", + "Chief", + "Investment", + "Officer", + "29", + "effect", + "154", + "486", + "gross", + "dividend", + "equivalent", + "units", + "issued", + "vesting", + "Alphabet", + "2021", + "Performance", + "Units", + "adjusted", + "based", + "performance", + "sold", + "net", + "tax", + "withholding", + "There", + "arrangements", + "directors", + "officers", + "Each", + "accordance", + "Policy", + "Against", + "Insider", + "actual", + "sale", + "transactions", + "made", + "pursuant", + "disclosed", + "publicly", + "filings", + "SEC", + "applicable", + "securities", + "laws", + "rules", + "regulations", + "Required", + "Disclosure", + "Pursuant", + "13", + "As", + "previously", + "Google", + "LLC", + "subsidiary", + "filed", + "notifications", + "Russian", + "Federal", + "Security", + "Service", + "FSB", + "encryption", + "control", + "requirements", + "complied", + "prior", + "import", + "covered", + "items", + "information", + "provided", + "Part", + "II", + "Quarterly", + "Report", + "Form", + "September", + "30", + "incorporated", + "reference" + ], + "keywords": [ + "10b5-1", + "trading plans", + "insider trading" + ], + "connect_to": [] + } + }, + { + "chunk_id": "cc4c3124-4c66-5fd7-863e-f016f6ab1f29", + "type": "text", + "content": "Not applicable.", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9C. DISCLOSURE REGARDING FOREIGN JURISDICTIONS THAT PREVENT INSPECTIONS", + "metadata": { + "length": 15, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Not", + "applicable" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "defbd9dd-3fc7-583a-9c79-490402dd297d", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART III", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d4f9774b-4a3f-5f31-b460-dfe3ad0a1508", + "type": "text", + "content": "The information required by this item will be included under the caption \"Directors, Executive Officers, and Corporate Governance\" in our Proxy Statement for the 2026 Annual Meeting of Stockholders to be filed with the SEC within 120 days of the fiscal year ended December 31, 2025 (2026 Proxy Statement) and is incorporated herein by reference. The information required by this item regarding delinquent filers pursuant to Item 405 of Regulation S-K will be included under the caption \"Delinquent Section 16(a) Reports\" in the 2026 Proxy Statement and is incorporated herein by reference.\nWe have adopted insider trading policies and procedures governing the purchase, sale, and other dispositions of securities of Alphabet by directors, officers, and employees that we believe are reasonably designed to promote compliance with insider trading laws, rules and regulations, and applicable Nasdaq listing standards. Our insider trading policy states, among other things, that our directors, officers, and employees are prohibited from trading in such securities while in possession of material, nonpublic information. The foregoing summary of our insider trading policies and procedures does not purport to be complete and is qualified by reference to our Policy Against Insider Trading filed as Exhibit 19.1 to our 2024 Annual Report on Form 10-K and incorporated by reference herein.", + "path": "GOOG-10-K-2025.pdf/PART III/ITEM 10. DIRECTORS, EXECUTIVE OFFICERS, AND CORPORATE GOVERNANCE", + "metadata": { + "length": 1385, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "information", + "required", + "item", + "included", + "caption", + "Directors", + "Executive", + "Officers", + "Corporate", + "Governance", + "Proxy", + "Statement", + "2026", + "Annual", + "Meeting", + "Stockholders", + "filed", + "SEC", + "120", + "days", + "fiscal", + "year", + "ended", + "December", + "31", + "2025", + "incorporated", + "reference", + "delinquent", + "filers", + "pursuant", + "Item", + "405", + "Regulation", + "Delinquent", + "Section", + "16", + "Reports", + "We", + "adopted", + "insider", + "trading", + "policies", + "procedures", + "governing", + "purchase", + "sale", + "dispositions", + "securities", + "Alphabet", + "directors", + "officers", + "employees", + "designed", + "promote", + "compliance", + "laws", + "rules", + "regulations", + "applicable", + "Nasdaq", + "listing", + "standards", + "Our", + "policy", + "states", + "things", + "prohibited", + "possession", + "material", + "nonpublic", + "foregoing", + "summary", + "purport", + "complete", + "qualified", + "Policy", + "Against", + "Insider", + "Trading", + "Exhibit", + "19.1", + "2024", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "88dcb81e-5112-5eb3-9277-155678d391c0", + "type": "text", + "content": "The information required by this item will be included under the captions \"Director Compensation,\" \"Executive Compensation\" and \"Directors, Executive Officers, and Corporate Governance—Corporate Governance and Board Matters—Compensation Committee Interlocks and Insider Participation\" in the 2026 Proxy Statement and is incorporated herein by reference, except as to information disclosed therein pursuant to Item 402(v) of Regulation S-K relating to pay versus performance.\nITEM 12. SECURITY OWNERSHIP OF CERTAIN BENEFICIAL OWNERS AND MANAGEMENT AND RELATED STOCKHOLDER MATTERS\nThe information required by this item will be included under the captions \"Common Stock Ownership of Certain Beneficial Owners and Management\" and \"Equity Compensation Plan Information\" in the 2026 Proxy Statement and is incorporated herein by reference.", + "path": "GOOG-10-K-2025.pdf/PART III/ITEM 11. EXECUTIVE COMPENSATION", + "metadata": { + "length": 833, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "information", + "required", + "item", + "included", + "captions", + "Director", + "Compensation", + "Executive", + "Directors", + "Officers", + "Corporate", + "Governance", + "Board", + "Matters", + "Committee", + "Interlocks", + "Insider", + "Participation", + "2026", + "Proxy", + "Statement", + "incorporated", + "reference", + "disclosed", + "pursuant", + "Item", + "402", + "Regulation", + "relating", + "pay", + "versus", + "performance", + "ITEM", + "12", + "SECURITY", + "OWNERSHIP", + "OF", + "CERTAIN", + "BENEFICIAL", + "OWNERS", + "AND", + "MANAGEMENT", + "RELATED", + "STOCKHOLDER", + "MATTERS", + "Common", + "Stock", + "Ownership", + "Certain", + "Beneficial", + "Owners", + "Management", + "Equity", + "Plan", + "Information" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d1574abe-bbd4-53f1-b244-f7e8896a51d4", + "type": "text", + "content": "The information required by this item will be included under the captions \"Certain Relationships and Related Transactions\" and \"Directors, Executive Officers, and Corporate Governance—Corporate Governance and Board Matters—Director Independence\" in the 2026 Proxy Statement and is incorporated herein by reference.", + "path": "GOOG-10-K-2025.pdf/PART III/ITEM 13. CERTAIN RELATIONSHIPS AND RELATED TRANSACTIONS, AND DIRECTOR INDEPENDENCE", + "metadata": { + "length": 314, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "information", + "required", + "item", + "included", + "captions", + "Certain", + "Relationships", + "Related", + "Transactions", + "Directors", + "Executive", + "Officers", + "Corporate", + "Governance", + "Board", + "Matters", + "Director", + "Independence", + "2026", + "Proxy", + "Statement", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "553ce4dc-fa20-52d0-93fa-89793700eb42", + "type": "text", + "content": "The information required by this item will be included under the caption \"Independent Registered Public Accounting Firm\" in the 2026 Proxy Statement and is incorporated herein by reference.", + "path": "GOOG-10-K-2025.pdf/PART III/ITEM 14. PRINCIPAL ACCOUNTANT FEES AND SERVICES", + "metadata": { + "length": 189, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "The", + "information", + "required", + "item", + "included", + "caption", + "Independent", + "Registered", + "Public", + "Accounting", + "Firm", + "2026", + "Proxy", + "Statement", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0830dffe-7f86-5393-8278-d932c955e81b", + "type": "text", + "content": "", + "path": "GOOG-10-K-2025.pdf/PART IV", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a1500ada-20a5-5ebf-ba5e-80b9c1bd9731", + "type": "text", + "content": "We have filed the following documents as part of this Annual Report on Form 10-K:", + "path": "GOOG-10-K-2025.pdf/PART IV/ITEM 15. EXHIBITS, FINANCIAL STATEMENT SCHEDULES", + "metadata": { + "length": 81, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "We", + "filed", + "documents", + "part", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "dddecd11-4609-5a63-b797-ad5e0f42c39e", + "type": "text", + "content": "Reports of Independent Registered Public Accounting Firm 45\nFinancial Statements:\nConsolidated Balance Sheets 48\nConsolidated Statements of Income 49\nConsolidated Statements of Comprehensive Income 50\nConsolidated Statements of Stockholders' Equity 51\nConsolidated Statements of Cash Flows 52\nNotes to Consolidated Financial Statements 53", + "path": "GOOG-10-K-2025.pdf/PART IV/ITEM 15. EXHIBITS, FINANCIAL STATEMENT SCHEDULES/1. Consolidated Financial Statements", + "metadata": { + "length": 338, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Reports", + "Independent", + "Registered", + "Public", + "Accounting", + "Firm", + "45", + "Financial", + "Statements", + "Consolidated", + "Balance", + "Sheets", + "48", + "Income", + "49", + "Comprehensive", + "50", + "Stockholders", + "Equity", + "51", + "Cash", + "Flows", + "52", + "Notes", + "53" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2b63de4a-1d84-5ace-830d-73c017485072", + "type": "table", + "content": "
Balance atBeginning of YearAdditionsUsageBalance atEnd of Year
Year ended December 31, 2023$ 1,213$ 3,115$ (2,737)$ 1,591
Year ended December 31, 2024$ 1,591$ 2,895$ (2,850)$ 1,636
Year ended December 31, 2025$ 1,636$ 4,128$ (3,408)$ 2,356
", + "path": "tables/table-68 Warranty Reserves.html", + "metadata": { + "length": 467, + "summary": "table-69\nThe table shows warranty reserve balances for 2023-2025, with beginning balances, additions, usage, and ending balances. The ending balance increased from $1,213 to $2,356 over the three years.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-68 Warranty Reserves.html", + "keywords": [ + "Warranty", + "Reserves", + "Balance" + ], + "tokens": [] + } + }, + { + "chunk_id": "568517af-53a0-584b-bde1-eac1456f760a", + "type": "text", + "content": "Schedule II: Valuation and Qualifying Accounts\nThe table below details the activity of the allowance for credit losses and sales credits for the years ended December 31, 2023, 2024, and 2025 (in millions):\n\n[tables/table-68 Warranty Reserves.html]\n\nNote: Additions to the allowance for credit losses are charged to expense. Additions to the allowance for sales credits are charged against revenues.\nAll other schedules have been omitted because they are not required, not applicable, or the required information is otherwise included.", + "path": "GOOG-10-K-2025.pdf/PART IV/ITEM 15. EXHIBITS, FINANCIAL STATEMENT SCHEDULES/2. Financial Statement Schedules", + "metadata": { + "length": 580, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Schedule", + "II", + "Valuation", + "Qualifying", + "Accounts", + "The", + "table", + "details", + "activity", + "allowance", + "credit", + "losses", + "sales", + "credits", + "years", + "ended", + "December", + "31", + "2023", + "2024", + "2025", + "millions", + "Note", + "Additions", + "charged", + "expense", + "revenues", + "All", + "schedules", + "omitted", + "required", + "applicable", + "information", + "included" + ], + "keywords": [], + "connect_to": [ + { + "target": "2b63de4a-1d84-5ace-830d-73c017485072", + "relation": "embeds", + "ref": "[tables/table-68 Warranty Reserves.html]", + "position": { + "start": 207, + "end": 247 + } + } + ] + } + }, + { + "chunk_id": "5d443d19-0796-57e7-b32d-c7f7ef7189ec", + "type": "table", + "content": "
Exhibit NumberDescriptionIncorporated by reference herein
FormDate
2.01Agreement and Plan of Merger, dated October 2, 2015, by and among Google Inc., the Registrant and Maple Technologies Inc.Current Report on Form 8-K (File No. 001-37580)October 2, 2015
3.01Amended and Restated Certificate of Incorporation of the RegistrantCurrent Report on Form 8-K (File No. 001-37580)June 3, 2022
3.02Amended and Restated Bylaws of the Registrant, dated October 19, 2022Current Report on Form 8-K (File No. 001-37580)October 25, 2022
4.01Specimen Class A Common Stock certificateCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.02Specimen Class C Capital Stock certificateCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.03Alphabet Inc. Deferred Compensation PlanCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.04Transfer Restriction Agreement, dated October 2, 2015, between the Registrant and Larry Page and certain of his affiliatesCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.05Transfer Restriction Agreement, dated October 2, 2015, between the Registrant and Sergey Brin and certain of his affiliatesCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.06Joinder Agreement, dated December 31, 2021, among the Registrant, Sergey Brin and certain of his affiliatesAnnual Report on Form 10-K (File No. 001-37580)February 2, 2022
4.07Transfer Restriction Agreement, dated October 2, 2015, between the Registrant and Eric E. Schmidt and certain of its affiliatesCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.08Class C Undertaking, dated October 2, 2015, executed by the RegistrantCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.09Indenture, dated February 12, 2016, between the Registrant and The Bank of New York Mellon Trust Company, N.A., as TrusteeRegistration Statement on Form S-3 (File No. 333-209510)February 12, 2016
4.10Registrant Registration Rights Agreement dated December 14, 2015Registration Statement on Form S-3 (File No. 333-209518)February 12, 2016
4.11First Supplemental Indenture, dated April 27, 2016, between the Registrant and The Bank of New York Mellon Trust Company, N.A., as trusteeCurrent Report on Form 8-K (File No. 001-37580)April 27, 2016
4.12Form of the Registrant's 1.998% Note due 2026Current Report on Form 8-K (File No. 001-37580)August 9, 2016
4.13Form of Global Note representing the Registrant's 0.800% notes due 2027Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.14Form of Global Note representing the Registrant's 1.100% notes due 2030Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.15Form of Global Note representing the Registrant's 1.900% notes due 2040Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.16Form of Global Note representing the Registrant's 2.050% notes due 2050Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.17Form of Global Note representing the Registrant's 2.250% notes due 2060Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.18Form of Global Note representing the Registrant's 4.000% notes due 2030Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.19Form of Global Note representing the Registrant's 4.500% notes due 2035Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.20Form of Global Note representing the Registrant's 5.250% notes due 2055Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.21Form of Global Note representing the Registrant's 5.300% notes due 2065Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.22Form of Global Note representing the Registrant's 2.500% notes due 2029Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.23Form of Global Note representing the Registrant's 3.000% notes due 2033Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.24Form of Global Note representing the Registrant's 3.375% notes due 2037Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.25Form of Global Note representing the Registrant's 3.875% notes due 2045Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.26Form of Global Note representing the Registrant's 4.000% notes due 2054Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.27Form of Global Note representing the Registrant's 2.375% notes due 2028Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.28Form of Global Note representing the Registrant's 2.875% notes due 2031Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.29Form of Global Note representing the Registrant's 3.125% notes due 2034Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.30Form of Global Note representing the Registrant's 3.500% notes due 2038Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.31Form of Global Note representing the Registrant's 4.000% notes due 2044Current Report on Form 8-K (File No. 001-37580)November 6, 2025
Exhibit NumberDescriptionIncorporated by reference herein
FormDate
4.32Form of Global Note representing the Registrant's 4.375% notes due 2064Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.33Form of Global Note representing the Registrant's floating rate notes due 2028Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.34Form of Global Note representing the Registrant's 3.875% notes due 2028Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.35Form of Global Note representing the Registrant's 4.100% notes due 2030Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.36Form of Global Note representing the Registrant's 4.375% notes due 2032Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.37Form of Global Note representing the Registrant's 4.700% notes due 2035Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.38Form of Global Note representing the Registrant's 5.350% notes due 2045Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.39Form of Global Note representing the Registrant's 5.450% notes due 2055Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.40Form of Global Note representing the Registrant's 5.700% notes due 2075Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.41Description of Registrant's SecuritiesAnnual Report on Form 10-K (File No. 001-37580)February 3, 2023
10.01Form of Indemnification Agreement entered into between the Registrant, its affiliates and its directors and officersCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.02Form of Offer Letter for DirectorsAnnual Report on Form 10-K (File No. 001-37580)January 31, 2024
10.03Letter Agreement dated June 3, 2024, between Anat Ashkenazi and AlphabetCurrent Report on Form 8-K (File No. 001-37580)June 7, 2024
10.04Compensation Plan Agreement, dated October 2, 2015, between Google Inc. and the RegistrantCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.05Director Arrangements Agreement, dated October 2, 2015, between Google Inc. and the RegistrantCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.06Alphabet Inc. Deferred Compensation PlanCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.07Alphabet Inc. Amended and Restated 2012 Stock PlanCurrent Report on Form 8-K (File No. 001-37580)June 5, 2020
10.07.1Alphabet Inc. Amended and Restated 2012 Stock Plan - Form of Alphabet Restricted Stock Unit AgreementAnnual Report on Form 10-K (File No. 001-37580)February 4, 2020
10.08Alphabet Inc. Amended and Restated 2021 Stock PlanCurrent Report on Form 8-K (file No. 001-37580)June 8, 2023
10.08.1Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet Restricted Stock Unit AgreementQuarterly Report on Form 10-Q (file No. 001-37580)July 28, 2021
10.08.2Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet Restricted Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)July 26, 2023
10.08.3Alphabet Inc. Amended and Restated 2021 Stock Plan – Form of Alphabet Restricted Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)April 26, 2024
10.08.4Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet 2022 CEO Performance Stock Unit AgreementAnnual Report on Form 10-K (File No. 001-37580)February 3, 2023
10.08.5Alphabet Inc. Amended and Restated 2021 Stock Plan – Form of Alphabet CEO Performance Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)April 26, 2024
10.08.6Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet 2022 Non-CEO Performance Stock Unit AgreementAnnual Report on Form 10-K (File No. 001-37580)February 2, 2022
10.08.7Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet 2023 Non-CEO Performance Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)July 26, 2023
10.08.8Alphabet Inc. Amended and Restated 2021 Stock Plan – Form of Alphabet Non-CEO Performance Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)April 26, 2024
10.09Alphabet Inc. Company Bonus Plan, as amendedAnnual Report on Form 10-K (File No. 001-37350)February 2, 2023
19.01Alphabet Inc. Policy Against Insider TradingAnnual Report on Form 10-K (File No. 001-37350)February 5, 2025
21.01*Subsidiaries of the Registrant
23.01*Consent of Independent Registered Public Accounting Firm
24.01*Power of Attorney (incorporated by reference to the signature page of this Annual Report on Form 10-K)
31.01*Certification of Chief Executive Officer pursuant to Exchange Act Rules 13a-14(a) and 15d-14(a), as adopted pursuant to Section 302 of the Sarbanes-Oxley Act of 2002
31.02*Certification of Chief Financial Officer pursuant to Exchange Act Rules 13a-14(a) and 15d-14(a), as adopted pursuant to Section 302 of the Sarbanes-Oxley Act of 2002
32.01Certifications of Chief Executive Officer and Chief Financial Officer pursuant to 18 U.S.C. Section 1350, as adopted pursuant to Section 906 of the Sarbanes-Oxley Act of 2002
97.01Clawback PolicyAnnual Report on Form 10-K (File No. 001-37350)January 31, 2024
101.INS*Inline XBRL Instance Document - the instance document does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document
101.SCH*Inline XBRL Taxonomy Extension Schema Document
101.CAL*Inline XBRL Taxonomy Extension Calculation Linkbase Document
101.DEF*Inline XBRL Taxonomy Extension Definition Linkbase Document
101.LAB*Inline XBRL Taxonomy Extension Label Linkbase Document
101.PRE*Inline XBRL Taxonomy Extension Presentation Linkbase Document
104*Cover Page Interactive Data File (embedded within the Inline XBRL document and contained in Exhibit 101)
", + "path": "tables/table-69 Exhibit List.html", + "metadata": { + "length": 15187, + "summary": "table-70\nHTML table listing exhibits (2.01-104) for Alphabet Inc., including descriptions and incorporation by reference details for SEC filings.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-69 Exhibit List.html", + "keywords": [ + "Exhibits", + "SEC Filings", + "Alphabet" + ], + "tokens": [] + } + }, + { + "chunk_id": "981952d0-4f35-5d70-8278-d4baa74c6cf2", + "type": "text", + "content": "[tables/table-69 Exhibit List.html]\n\n◆ Indicates management compensatory plan, contract, or arrangement.\n\\* Filed herewith.\n‡ Furnished herewith.", + "path": "GOOG-10-K-2025.pdf/PART IV/ITEM 15. EXHIBITS, FINANCIAL STATEMENT SCHEDULES/3. Exhibits", + "metadata": { + "length": 183, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Indicates", + "management", + "compensatory", + "plan", + "contract", + "arrangement", + "Filed", + "herewith", + "Furnished" + ], + "keywords": [], + "connect_to": [ + { + "target": "5d443d19-0796-57e7-b32d-c7f7ef7189ec", + "relation": "embeds", + "ref": "[tables/table-69 Exhibit List.html]", + "position": { + "start": 0, + "end": 35 + } + } + ] + } + }, + { + "chunk_id": "67ddeed7-c2ee-51fb-ada3-a387f94472c0", + "type": "text", + "content": "None.", + "path": "GOOG-10-K-2025.pdf/PART IV/ITEM 16. FORM 10-K SUMMARY", + "metadata": { + "length": 5, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d8df950a-a463-5032-b530-e48c2902589a", + "type": "table", + "content": "
By:/S/ SUNDAR PICHAI
Sundar Pichai
Chief Executive Officer(Principal Executive Officer of the Registrant)
", + "path": "tables/table-70 CEO Signature.html", + "metadata": { + "length": 199, + "summary": "table-71\nThe document is signed by Sundar Pichai, Chief Executive Officer and Principal Executive Officer of the registrant.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-70 CEO Signature.html", + "keywords": [ + "CEO", + "Signature", + "Registrant" + ], + "tokens": [] + } + }, + { + "chunk_id": "276e3598-d5f9-53ef-8d3b-0d48aaf59185", + "type": "text", + "content": "Pursuant to the requirements of Section 13 or 15(d) of the Securities Exchange Act of 1934, the Registrant has duly caused this Annual Report on Form 10-K to be signed on its behalf by the undersigned, thereunto duly authorized.\nDate: February 4, 2026\nALPHABET INC.\n\n[tables/table-70 CEO Signature.html]", + "path": "GOOG-10-K-2025.pdf/PART IV/SIGNATURES", + "metadata": { + "length": 311, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "Pursuant", + "requirements", + "Section", + "13", + "15", + "Securities", + "Exchange", + "Act", + "1934", + "Registrant", + "duly", + "caused", + "Annual", + "Report", + "Form", + "10", + "signed", + "behalf", + "undersigned", + "thereunto", + "authorized", + "Date", + "February", + "2026", + "ALPHABET", + "INC" + ], + "keywords": [], + "connect_to": [ + { + "target": "d8df950a-a463-5032-b530-e48c2902589a", + "relation": "embeds", + "ref": "[tables/table-70 CEO Signature.html]", + "position": { + "start": 267, + "end": 303 + } + } + ] + } + }, + { + "chunk_id": "cf5c96e9-bdc1-5842-9cbc-c3e9d008b003", + "type": "table", + "content": "
SignatureTitleDate
/S/SUNDAR PICHAIChief Executive Officer and Director (Principal Executive Officer)February 4, 2026
Sundar Pichai
/S/ANAT ASHKENAZISenior Vice President and Chief Financial Officer (Principal Financial Officer)February 4, 2026
Anat Ashkenazi
/S/AMIE THUENER O'TOOLEVice President, Corporate Controller and Principal Accounting OfficerFebruary 4, 2026
Amie Thuener O'Toole
/S/FRANCES H. ARNOLDDirectorFebruary 4, 2026
Frances H. Arnold
/S/SERGEY BRINCo-Founder and DirectorFebruary 4, 2026
Sergey Brin
/S/R. MARTIN CHAVEZDirectorFebruary 4, 2026
R. Martin Chávez
/S/L. JOHN DOERRDirectorFebruary 4, 2026
L. John Doerr
/S/ROGER W. FERGUSON JR.DirectorFebruary 4, 2026
Roger W. Ferguson Jr.
/S/JOHN L. HENNESSYDirector, ChairFebruary 4, 2026
John L. Hennessy
/S/LARRY PAGECo-Founder and DirectorFebruary 4, 2026
Larry Page
/S/K. RAM SHRIRAMDirectorFebruary 4, 2026
K. Ram Shriram
/S/ROBIN L. WASHINGTONDirectorFebruary 4, 2026
Robin L. Washington
", + "path": "tables/table-71 Signatures.html", + "metadata": { + "length": 2126, + "summary": "table-72\nSignatures of Sundar Pichai, Anat Ashkenazi, Amie Thuener O'Toole, and directors including Frances Arnold, Sergey Brin, and others, all dated February 4, 2026.", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "file_path": "tables/table-71 Signatures.html", + "keywords": [ + "Signatures", + "Executives", + "Directors" + ], + "tokens": [] + } + }, + { + "chunk_id": "69de83b9-0974-5f20-acec-e0a38add0359", + "type": "text", + "content": "KNOW ALL PERSONS BY THESE PRESENTS, that each person whose signature appears below constitutes and appoints Sundar Pichai and Anat Ashkenazi, jointly and severally, his or her attorney-in-fact, with the power of substitution, for him or her in any and all capacities, to sign any amendments to this Annual Report on Form 10-K and to file the same, with exhibits thereto and other documents in connection therewith, with the Securities and Exchange Commission, hereby ratifying and confirming all that each of said attorneys-in-fact, or his or her substitute or substitutes, may do or cause to be done by virtue hereof.\nPursuant to the requirements of the Securities Exchange Act of 1934, this Annual Report on Form 10-K has been signed below by the following persons on behalf of the Registrant and in the capacities and on the dates indicated.\n\n[tables/table-71 Signatures.html]", + "path": "GOOG-10-K-2025.pdf/PART IV/POWER OF ATTORNEY", + "metadata": { + "length": 889, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: PART I, PART II, PART III, PART IV", + "tokens": [ + "KNOW", + "ALL", + "PERSONS", + "BY", + "THESE", + "PRESENTS", + "person", + "signature", + "appears", + "constitutes", + "appoints", + "Sundar", + "Pichai", + "Anat", + "Ashkenazi", + "jointly", + "severally", + "attorney", + "fact", + "power", + "substitution", + "capacities", + "sign", + "amendments", + "Annual", + "Report", + "Form", + "10", + "file", + "exhibits", + "thereto", + "documents", + "connection", + "therewith", + "Securities", + "Exchange", + "Commission", + "ratifying", + "confirming", + "attorneys", + "substitute", + "substitutes", + "virtue", + "hereof", + "Pursuant", + "requirements", + "Act", + "1934", + "signed", + "persons", + "behalf", + "Registrant", + "dates" + ], + "keywords": [], + "connect_to": [ + { + "target": "cf5c96e9-bdc1-5842-9cbc-c3e9d008b003", + "relation": "embeds", + "ref": "[tables/table-71 Signatures.html]", + "position": { + "start": 846, + "end": 879 + } + } + ] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/doc_nav.json b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/doc_nav.json new file mode 100644 index 000000000..eaaefde5c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/doc_nav.json @@ -0,0 +1,2169 @@ +{ + "version": "1.0", + "file_name": "GOOG 10-K 2025.pdf", + "stats": { + "total_chunks": 300, + "text_chunks": 226, + "image_chunks": 2, + "table_chunks": 72, + "max_depth": 7 + }, + "sections": [ + { + "title": "Root", + "path": "GOOG-10-K-2025.pdf", + "level": 1, + "summary": "This is the Form 10-K annual report filed by Alphabet Inc. with the U.S. Securities and Exchange Commission for the fiscal year ended December 31, 2025. It includes company identification details (Delaware incorporation, IRS employer ID, principal executive offices in Mountain View, CA), checkmarks indicating compliance with SEC filing requirements, and classification as a large accelerated filer. The report notes that as of June 30, 2025, the aggregate market value of shares held by non-affiliates was approximately $1.9 trillion, and as of January 28, 2026, there were 5,822 million Class A shares, 837 million Class B shares, and 5,438 million Class C shares outstanding. Portions of the proxy statement for the 2026 annual meeting are incorporated by reference in Part III.", + "chunk_count": 1, + "children": [] + }, + { + "title": "PART I", + "path": "GOOG-10-K-2025.pdf/PART I", + "level": 1, + "summary": "", + "chunk_count": 29, + "children": [ + { + "title": "ITEM 1. BUSINESS", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS", + "level": 2, + "summary": "", + "chunk_count": 17, + "children": [ + { + "title": "Overview", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview", + "level": 3, + "summary": "As our founders Larry and Sergey wrote in the original founders' letter, \"Google is not a conventional company. We do not intend to become one.\" That unconventional spirit has been a driving force throughout our history, inspiring us to tackle big problems and invest in moonshots. It led us to be a pioneer in the development of artificial intelligence (AI) and, since 2016, be an AI-first company. We continue this work under the leadership of Alphabet and Google CEO, Sundar Pichai. Alphabet is a collection of businesses — the largest of which is Google. We report Google in two segments, Google Services and Google Cloud, and all non-Google businesses collectively as Other Bets. Supporting these businesses, we have centralized certain AI-related research and development focused on advanced research in AI and developing the frontier models that serve our businesses, which is reported in Alphabet-level activities. Alphabet's structure is about helping each of our businesses prosper through strong leaders and independence.", + "chunk_count": 16, + "children": [ + { + "title": "Access and Technology for Everyone", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Access and Technology for Everyone", + "level": 4, + "summary": "The Internet is one of the world's most powerful equalizers; it propels ideas, people, and businesses large and small. Our mission to organize the world's information and make it universally accessible and useful is as relevant today as it was when we were founded in 1998. Since then, we have evolved from a company that helps people find answers to a company that also helps people get things done. We are focused on building an even more helpful Google for everyone, and we aspire to give everyone the tools they need to increase their knowledge, health, happiness, and success. Google Search helps people find information and make sense of the world in more natural and intuitive ways, with trillions of searches on Google every year. YouTube provides people with entertainment, information, and opportunities to learn something new and helps support the creator economy through the YouTube Partner Program. Google Cloud helps customers build for the future, improve productivity, reduce costs, and unlock new growth engines. We continually innovate and build new products and features to help our users, partners, customers, and communities and have invested more than \\$200 billion in research and development in the last five years in support of these efforts.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Making AI Helpful for Everyone", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Making AI Helpful for Everyone", + "level": 4, + "summary": "Google's AI strategy is a full-stack approach encompassing AI-optimized infrastructure, world-class research, and products. Key differentiators include custom TPUs like Ironwood and GPUs. Gemini 3 is their most advanced model, featuring enhanced reasoning and multimodal capabilities. AI is embedded across all major products, including Search and YouTube, and offered to Google Cloud customers. The approach is guided by principles of boldness and responsibility.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Moonshots", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Moonshots", + "level": 4, + "summary": "Many companies get comfortable doing what they have always done, making only incremental changes. This incrementalism leads to irrelevance over time, especially in technology, where change tends to be revolutionary, not evolutionary. Our early investments in AI started out as moonshots but are now incorporated into our core products and central to future developments. In Other Bets, our fully autonomous driving technology company, Waymo, is now providing fully autonomous, paid ride-hailing services to customers in multiple cities. Isomorphic Labs is reimagining the drug discovery process from first principles, applying AI to accelerate the development of new medicines. We continue to look toward the future and to invest for the long term, most notably for the application of AI to our products and services, as well as other frontier technologies such as quantum computing.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Privacy and Security", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Privacy and Security", + "level": 4, + "summary": "We make it a priority to protect the privacy and security of our products, users, and customers, even if there are near-term financial consequences. We do this by continuously investing in building products that are secure by default; strictly upholding responsible data practices that emphasize privacy by design; and building easy-to-use settings that put people in control. We are continually enhancing these efforts over time, whether by enabling users to auto-delete their data, applying privacy technologies like on-device processing, giving people tools to control their experience, or advancing anti-malware, anti-phishing, and password security features.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Google", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google", + "level": 4, + "summary": "For reporting purposes Google comprises two segments: Google Services and Google Cloud.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Google Services", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google Services", + "level": 4, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "Serving Our Users", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google Services/Serving Our Users", + "level": 5, + "summary": "Google is committed to building helpful products that improve lives worldwide, with core offerings including ads, Android, Chrome, devices, Gmail, Google Drive, Gemini, Maps, Photos, Play, Search, and YouTube. Over 25 years, Google Search evolved from simple web links to a multimodal experience using large language models, enabling natural language queries and features like AI Overviews and AI Mode. The company enhances digital content discovery and creation through platforms like Google Play and YouTube, supporting videos, podcasts, TV, games, music, books, and apps. Investments in Android, Chrome, and devices like the Pixel 10 series and Pixel Watch 4 combine AI, software, and hardware to make daily life easier.", + "chunk_count": 1, + "children": [] + }, + { + "title": "How We Make Money", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google Services/How We Make Money", + "level": 5, + "summary": "Google provides world-class advertising technologies for advertisers, agencies, and publishers, using AI for over a decade in products like Demand Gen and Performance Max. Revenue comes from performance and brand ads on Google Search, YouTube, and Google Network properties. Performance ads drive direct user engagement, while brand ads enhance awareness. Google invests in stopping bad advertising practices. Additional revenue sources include consumer subscriptions (YouTube services, Google One), platforms (Google Play), and devices (Pixel).", + "chunk_count": 1, + "children": [] + }, + { + "title": "Google Cloud", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google Services/Google Cloud", + "level": 5, + "summary": "Google Cloud generates revenue from consumption-based fees and subscriptions for infrastructure, platform, applications, and other cloud services. Customers use Google Cloud for AI-optimized infrastructure, a developer platform (Vertex AI) with over 200 foundation models, cybersecurity solutions, data and analytics, Gemini Enterprise for AI agents, and Gemini for Google Workspace.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Bets", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Google Services/Other Bets", + "level": 5, + "summary": "Across Alphabet, we are also using technology to try to solve big problems that affect a wide variety of industries, including transportation and health technology. Alphabet's investment in the portfolio of Other Bets includes businesses that are at various stages of development, ranging from those in the research and development phase, such as X, our moonshot factory focused on developing breakthrough technologies, to those that are scaling commercialization, such as Waymo, which is expanding to more cities domestically, entering international markets, and further scaling operations. Other Bets operate as independent companies and some of them have their own boards with independent members and outside investors. While these early-stage businesses naturally come with considerable uncertainty, some of them are already generating revenue and making important strides in their industries. Revenues from Other Bets are generated primarily from the sale of autonomous transportation and internet services.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Competition", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Competition", + "level": 4, + "summary": "The passage describes the competitive landscape of a business facing rapid change and disruptive technologies. It lists various competitors including general and vertical search engines, online advertising platforms, other media, cloud services, AI developers, hardware makers, digital video services, social networks, and workspace communication providers. Success depends on continuous innovation and attracting users, advertisers, and content providers based on relevance, efficiency, and quality.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Culture and Workforce", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Culture and Workforce", + "level": 4, + "summary": "Alphabet emphasizes its commitment to employees through industry-leading benefits, career growth opportunities, and competitive compensation to attract and retain talent. As of December 31, 2025, it had 190,820 employees. The company supports labor rights and maintains an open culture. It also partners with external companies for specialized services, using an extended workforce of vendors, temporary staff, and independent contractors, while ensuring compliance with its Supplier Code of Conduct.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Government Regulation", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Government Regulation", + "level": 4, + "summary": "The passage discusses the company's exposure to numerous US and foreign laws and regulations, particularly in areas like AI, competition, consumer protection, data privacy, and sustainability. It notes increasing scrutiny, evolving and sometimes conflicting requirements across jurisdictions, and potential negative impacts on business operations, costs, and competitive position.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Intellectual Property", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Intellectual Property", + "level": 4, + "summary": "We rely on intellectual property such as trademarks, copyrights, patents, and trade secrets, as well as confidentiality procedures and contractual provisions, to protect our proprietary technology and our brand. We have registered, and applied for the registration of, US and international trademarks, service marks, domain names, and copyrights. We have also filed patent applications in the US and foreign countries covering certain of our technology, and acquired patent assets to supplement our portfolio. We have licensed in the past, and expect that we may license in the future, certain of our rights to other parties. For additional information, see Item 1A Risk Factors of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Available Information", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1. BUSINESS/Overview/Available Information", + "level": 4, + "summary": "The passage describes the company's investor relations website (www.abc.xyz/investor) where annual and quarterly reports, proxy statements, and SEC filings are available free of charge. It also mentions webcasts of earnings calls, notifications of financial news, and corporate governance information. Additionally, it notes that executive officers may use social media like X and LinkedIn to share company updates, but clarifies that information on these platforms is not incorporated into SEC filings.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "ITEM 1A. RISK FACTORS", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS", + "level": 2, + "summary": "Our operations and financial results are subject to various risks and uncertainties, including but not limited to those described below, which could harm our business, reputation, financial condition, and operating results, and may affect the trading price and price volatility of our Class A and Class C stock.", + "chunk_count": 6, + "children": [ + { + "title": "Risks Specific to our Company", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS/Risks Specific to our Company", + "level": 3, + "summary": "The passage outlines significant risks to the company's business, primarily from its heavy reliance on advertising revenue (over 70% in 2025), which is threatened by reduced advertiser spending, loss of partners, and new ad-blocking technologies. Intense competition, especially in AI, requires substantial investment in infrastructure like custom TPUs, with no guarantee of success. The company also faces risks from supply chain disruptions, intellectual property protection, brand reputation, international operations, and the inherent uncertainty of investing in new businesses and technologies.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Risks Related to our Industry", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS/Risks Related to our Industry", + "level": 3, + "summary": "The passage outlines significant risks associated with developing and using AI, including reputational harm, legal liability, and regulatory scrutiny due to issues like harmful content, inaccuracies, discrimination, and intellectual property violations. It highlights challenges in implementing AI responsibly, the impact of problematic content on platforms, and the evolving landscape of device access. Data privacy and security concerns are emphasized, with risks from cyber attacks, software bugs, and unauthorized access potentially leading to litigation, fines, and loss of user trust. The text also discusses the dependence on internet access providers and the potential for restrictions or increased costs, which could harm business operations and financial results.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Risks Related to Laws, Regulations, and Policies", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS/Risks Related to Laws, Regulations, and Policies", + "level": 3, + "summary": "The passage details extensive legal and regulatory risks facing a large technology company, including antitrust lawsuits (e.g., DOJ cases on Search and ad tech), evolving AI regulations (e.g., EU AI Act), data privacy laws (e.g., GDPR, state laws), content moderation rules, intellectual property claims, and tax changes. It highlights potential costs, business model alterations, and liability from global compliance challenges.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Risks Related to Ownership of Our Stock", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS/Risks Related to Ownership of Our Stock", + "level": 3, + "summary": "The passage discusses Alphabet's share repurchase and dividend programs, noting they are not guaranteed to enhance long-term value and may increase stock price volatility. It highlights the concentrated voting power of Larry Page and Sergey Brin through Class B stock, which limits other stockholders' influence. Additionally, it outlines charter provisions and Delaware law that could discourage takeovers, such as the ability to issue preferred stock without stockholder approval and restrictions on business combinations with large stockholders.", + "chunk_count": 1, + "children": [] + }, + { + "title": "General Risks", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1A. RISK FACTORS/General Risks", + "level": 3, + "summary": "The passage discusses factors causing fluctuations in operating results, including cyclical nature, seasonality, and geopolitical events. It highlights risks from acquisitions, joint ventures, investments, and divestitures, such as management distraction, integration failures, regulatory issues, and potential dilution. It also emphasizes reliance on highly skilled personnel, competition for talent, and the need to maintain corporate culture for effective growth and competition.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ITEM 1B. UNRESOLVED STAFF COMMENTS", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1B. UNRESOLVED STAFF COMMENTS", + "level": 2, + "summary": "Not applicable.", + "chunk_count": 1, + "children": [] + }, + { + "title": "ITEM 1C. CYBERSECURITY", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 1C. CYBERSECURITY", + "level": 2, + "summary": "The passage describes a comprehensive process for identifying, assessing, and managing material cybersecurity risks, including software supply chain vulnerabilities, IP theft, and infrastructure compromise. It details oversight by dedicated teams led by a Vice President with over 20 years of experience, incident tracking and escalation procedures, and board-level oversight via the Risk and Compliance Committee. The company states that cybersecurity threats have not materially affected its business strategy or financial condition to date, but acknowledges future risks cannot be assured.", + "chunk_count": 1, + "children": [] + }, + { + "title": "ITEM 2. PROPERTIES", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 2. PROPERTIES", + "level": 2, + "summary": "Our headquarters are located in Mountain View, California. We own and lease office facilities and data centers around the world, primarily in Asia, Europe, and North America. We believe our existing facilities are in good condition and suitable for the conduct of our business.", + "chunk_count": 1, + "children": [] + }, + { + "title": "ITEM 3. LEGAL PROCEEDINGS", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 3. LEGAL PROCEEDINGS", + "level": 2, + "summary": "For a description of our material pending legal proceedings, see Legal Matters in Note 10 of the Notes to Consolidated Financial Statements included in Part II, Item 8 of this Annual Report on Form 10-K, which is incorporated herein by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "ITEM 4. MINE SAFETY DISCLOSURES", + "path": "GOOG-10-K-2025.pdf/PART I/ITEM 4. MINE SAFETY DISCLOSURES", + "level": 2, + "summary": "Not applicable.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "PART II", + "path": "GOOG-10-K-2025.pdf/PART II", + "level": 1, + "summary": "ITEM 5. MARKET FOR REGISTRANT'S COMMON EQUITY, RELATED STOCKHOLDER MATTERS, AND ISSUER PURCHASES OF EQUITY SECURITIES As of October 2, 2015, Alphabet Inc. became the successor issuer of Google Inc. pursuant to Rule 12g-3(a) under the Exchange Act. Our Class A stock has been listed on the Nasdaq Global Select Market under the symbol \"GOOG\" since August 19, 2004, and under the symbol \"GOOGL\" since April 3, 2014. Prior to August 19, 2004, there was no public market for our stock. Our Class B stock is neither listed nor traded. Our Class C stock has been listed on the Nasdaq Global Select Market under the symbol \"GOOG\" since April 3, 2014.", + "chunk_count": 183, + "children": [ + { + "title": "Holders of Record", + "path": "GOOG-10-K-2025.pdf/PART II/Holders of Record", + "level": 2, + "summary": "As of December 31, 2025, there were approximately 5,861 and 1,611 stockholders of record of our Class A stock and Class C stock, respectively. Because many of our shares of Class A stock and Class C stock are held by brokers and other institutions on behalf of stockholders, we are unable to estimate the total number of stockholders represented by these record holders. As of December 31, 2025, there were approximately 57 stockholders of record of our Class B stock.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Dividend Program", + "path": "GOOG-10-K-2025.pdf/PART II/Dividend Program", + "level": 2, + "summary": "Dividend payments to stockholders of Class A, Class B, and Class C shares were \\$4.8 billion, \\$703 million, and \\$4.5 billion, respectively, for the year ended December 31, 2025. The company intends to pay quarterly cash dividends in the future, subject to review and approval by the company's Board of Directors in its sole discretion. We regularly evaluate our cash and capital structure, including the size, pace, and form of capital return to stockholders.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Issuer Purchases of Equity Securities", + "path": "GOOG-10-K-2025.pdf/PART II/Issuer Purchases of Equity Securities", + "level": 2, + "summary": "The following table presents information with respect to Alphabet's repurchases of Class A and Class C stock during the quarter ended December 31, 2025: [tables/table-2 Share Repurchase.html] (1) In April 2024, the company's Board of Directors authorized a \\$70.0 billion share repurchase program for its Class A and Class C shares. In April 2025, the company's Board of Directors authorized an additional \\$70.0 billion share repurchase program for its Class A and Class C shares. Repurchases are being executed from time to time, subject to general business and market conditions and other investment opportunities, through open market purchases or privately negotiated transactions, including through Rule 10b5-1 plans. The repurchase programs do not have an expiration date. For additional information related to share repurchases, see Note 11 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. (2) Average price paid per share includes costs associated with the repurchases.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stock Performance Graphs", + "path": "GOOG-10-K-2025.pdf/PART II/Stock Performance Graphs", + "level": 2, + "summary": "The graph below matches Alphabet Inc. Class A's cumulative five-year total stockholder return on common stock with the cumulative total returns of the S&P 500 index, the NASDAQ Composite index, and the RDG Internet Composite index. The graph tracks the performance of a \\$100 investment in our common stock and in each index (with the reinvestment of all dividends) from December 31, 2020, to December 31, 2025. The returns shown are based on historical results and are not intended to suggest future performance.", + "chunk_count": 3, + "children": [ + { + "title": "COMPARISON OF CUMULATIVE 5-YEAR TOTAL RETURN\\* ALPHABET INC. CLASS A COMMON STOCK", + "path": "GOOG-10-K-2025.pdf/PART II/Stock Performance Graphs/COMPARISON OF CUMULATIVE 5-YEAR TOTAL RETURN\\* ALPHABET INC. CLASS A COMMON STOCK", + "level": 3, + "summary": "The document compares the cumulative five-year total stockholder return of Alphabet Inc. Class A and Class C common stock against the S&P 500 Index, NASDAQ Composite Index, and RDG Internet Composite Index. It tracks a $100 investment from December 31, 2020, to December 31, 2025, with dividends reinvested. The returns are historical and not indicative of future performance. A line chart and supporting table provide detailed data on the performance of each investment over the period.", + "chunk_count": 1, + "children": [] + }, + { + "title": "COMPARISON OF CUMULATIVE 5-YEAR TOTAL RETURN\\* ALPHABET INC. CLASS C CAPITAL STOCK", + "path": "GOOG-10-K-2025.pdf/PART II/Stock Performance Graphs/COMPARISON OF CUMULATIVE 5-YEAR TOTAL RETURN\\* ALPHABET INC. CLASS C CAPITAL STOCK", + "level": 3, + "summary": "ALPHABET INC. CLASS C CAPITAL STOCK Among Alphabet Inc., the S&P 500 Index, the NASDAQ Composite Index, and the RDG Internet Composite Index The graph below matches Alphabet Inc. Class C's cumulative five-year total stockholder return on capital stock with the cumulative total returns of the S&P 500 index, the NASDAQ Composite index, and the RDG Internet Composite index. The graph tracks the performance of a \\$100 investment in our Class C capital stock and in each index (with the reinvestment of all dividends) from December 31, 2020, to December 31, 2025. The returns shown are based on historical results and are not intended to suggest future performance.
0 [images/image-2-Stock Performance Comparison (2020-2025).jpg]
line chart [tables/table-4 Alphabet vs S&P 500.html]
\\*\\$100 invested on December 31, 2020, in stock or in index, including reinvestment of dividends. Copyright $^{©}$ 2026 S&P, a division of The McGraw-Hill Companies Inc. All rights reserved.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ITEM 6. [Reserved]", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]", + "level": 2, + "summary": "ITEM 7. MANAGEMENT'S DISCUSSION AND ANALYSIS OF FINANCIAL CONDITION AND RESULTS OF OPERATIONS Please read the following discussion and analysis of our financial condition and results of operations together with \"Note about Forward-Looking Statements,\" Part I, Item 1 \"Business,\" Part I, Item 1A \"Risk Factors,\" and our consolidated financial statements and related notes included under Item 8 of this Annual Report on Form 10-K. The following section generally discusses 2025 results compared to 2024 results. Discussion of 2024 results compared to 2023 results to the extent not included in this report can be found in Item 7 of our 2024 Annual Report on Form 10-K.", + "chunk_count": 58, + "children": [ + { + "title": "Understanding Alphabet's Financial Results", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Understanding Alphabet's Financial Results", + "level": 3, + "summary": "Alphabet is a collection of businesses — the largest of which is Google. We report Google in two segments, Google Services and Google Cloud, and all non-Google businesses collectively as Other Bets. Supporting these businesses, we have centralized certain AI-related research and development focused on advanced research in AI and developing the frontier models that serve our businesses, which is reported in Alphabet-level activities. For further details on our segments, see Part I, Item 1 Business and Note 15 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Trends in Our Business and Financial Effect", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Trends in Our Business and Financial Effect", + "level": 3, + "summary": "The passage outlines long-term trends affecting consolidated operations, including expanded AI offerings that may alter monetization and face competition; increasing non-advertising revenues from cloud and subscriptions with generally lower margins; significant capital expenditure growth for technical infrastructure, especially AI, driving higher operating costs; continued R&D and acquisition investments; rising traffic acquisition costs with a shifting rate; potential future debt financing; and an evolving regulatory environment leading to compliance costs and business practice changes.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Revenues and Monetization Metrics", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics", + "level": 3, + "summary": "We generate revenues by delivering relevant, cost-effective online advertising; cloud-based solutions that provide enterprise customers of all sizes with infrastructure, platform services, and applications; and sales of other products and services, such as fees received for subscription-based products, apps and in-app purchases, and devices. For additional information on how we recognize revenue, see Note 1 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. In addition to the long-term trends and their financial effect on our business discussed above, fluctuations in our revenues have been and may continue to be affected by a combination of factors, including: • changes in foreign currency exchange rates; • changes in pricing, such as those resulting from changes in fee structures, discounts, and customer incentives; - general economic conditions and various external dynamics, including geopolitical events, regulations, and other measures and their effect on advertiser, consumer, and enterprise spending; - new product, service, and market launches; and - seasonality. Additionally, fluctuations in our revenues generated from advertising (\"Google advertising\"), other sources (\"Google subscriptions, platforms, and devices\"), Google Cloud, and Other Bets have been, and may continue to be, affected by other factors unique to each set of revenues, as described below.", + "chunk_count": 6, + "children": [ + { + "title": "Google Services", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics/Google Services", + "level": 4, + "summary": "Google Services revenues consist of Google advertising as well as Google subscriptions, platforms, and devices revenues.", + "chunk_count": 5, + "children": [ + { + "title": "Google Advertising", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics/Google Services/Google Advertising", + "level": 5, + "summary": "Google advertising revenues come from Google Search & other properties, YouTube ads, and Google Network. Key metrics include paid clicks and cost-per-click for Search, and impressions and cost-per-impression for Network. Revenues are affected by factors like advertiser competition, ad quality, device mix, seasonality, and emerging market growth.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Google Subscriptions, Platforms, and Devices", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics/Google Services/Google Subscriptions, Platforms, and Devices", + "level": 5, + "summary": "Google subscriptions, platforms, and devices revenues are comprised of the following: - consumer subscriptions, which primarily include revenues from YouTube services, such as YouTube TV, YouTube Music and Premium, and NFL Sunday Ticket, as well as Google One, which offers access to our most capable Gemini models; - platforms, which primarily include revenues from Google Play sales of apps and in-app purchases; • devices, which primarily include sales of the Pixel family of devices; and • other products and services. Fluctuations in our Google subscriptions, platforms, and devices revenues have been, and may continue to be, affected by factors in addition to the general factors described above, such as changes in customer usage and demand, number of subscribers, and the timing of product launches.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Google Cloud", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics/Google Services/Google Cloud", + "level": 5, + "summary": "Google Cloud revenues are comprised of the following: - Google Cloud Platform primarily generates consumption-based fees and subscriptions for infrastructure, platform, and other services. These services provide access to solutions such as AI offerings including our enterprise AI infrastructure, Vertex AI platform, and Gemini Enterprise; cybersecurity offerings; and data and analytics solutions; - Google Workspace includes subscriptions for cloud-based communication and collaboration tools for enterprises, such as Gmail, Docs, Calendar, Drive, and Meet, with integrated features like Gemini for Google Workspace; and • other enterprise services. Fluctuations in our Google Cloud revenues have been, and may continue to be, affected by factors in addition to the general factors described above, such as changes in customer usage, demand, and supply availability.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Bets", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Revenues and Monetization Metrics/Google Services/Other Bets", + "level": 5, + "summary": "Revenues from Other Bets are generated primarily from the sale of autonomous transportation services and internet services.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Costs and Expenses", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses", + "level": 3, + "summary": "Our cost structure has two components: cost of revenues and operating expenses. Our operating expenses include costs related to research and development, sales and marketing, and general and administrative functions. Certain of our costs and expenses, including those associated with the operation of our technical infrastructure as well as components of our operating expenses, are generally less variable in nature and may not correlate to changes in revenue. Additionally, fluctuations in employee compensation expenses may not directly correlate with changes in headcount, due to factors such as annual SBC awards that vest over time.", + "chunk_count": 5, + "children": [ + { + "title": "Cost of Revenues", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses/Cost of Revenues", + "level": 4, + "summary": "Cost of revenues is comprised of TAC and other costs of revenues. • TAC includes: amounts paid to our distribution partners who make available our search access points and other ad-supported services. Our distribution partners include browser providers, mobile carriers, original equipment manufacturers, and software developers; and • amounts paid to Google Network partners primarily for ads displayed on their properties. • Other cost of revenues primarily includes: content acquisition costs, which are payments to content providers from whom we license video and other content for distribution, primarily related to YouTube (we pay fees to these content providers based on revenues generated, subscriber counts, or a flat fee); • depreciation expense, primarily related to our technical infrastructure; • employee compensation expenses related to our technical infrastructure and other operations such as content review and customer and product support; inventory and other costs related to the devices we sell; and other technical infrastructure operations costs, including energy, equipment, and network capacity costs. TAC as a percentage of revenues generated from ads placed on Google Network properties are significantly higher than TAC as a percentage of revenues generated from ads placed on Google Search & other properties, because most of the advertiser revenues from ads served on Google Network properties are paid as TAC to our Google Network partners.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Operating Expenses", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses/Operating Expenses", + "level": 4, + "summary": "Operating expenses are generally incurred during our normal course of business, which we categorize as either research and development, sales and marketing, or general and administrative. The main components of our research and development expenses are: • depreciation expense, primarily related to our technical infrastructure; - employee compensation expenses for engineering and technical employees responsible for research and development related to our existing and new products and services; • other technical infrastructure operations costs, including energy, equipment, and network capacity costs; and - third-party services fees primarily relating to consulting and outsourced services in support of our engineering and product development efforts. The main components of our sales and marketing expenses are: - employee compensation expenses for employees engaged in sales and marketing, sales support, and certain customer service functions; and - spend relating to our advertising and promotional activities in support of our products and services. The main components of our general and administrative expenses are: - employee compensation expenses for employees in finance, human resources, information technology, legal, and other administrative support functions; • expenses relating to legal and other matters, including certain fines and settlements; and - third-party services fees, including audit, consulting, outside legal, and other outsourced administrative services.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Income (Expense), Net", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses/Other Income (Expense), Net", + "level": 4, + "summary": "OI&E, net primarily consists of interest income (expense), the effect of foreign currency exchange gains (losses), net gains (losses) and impairment on our marketable and non-marketable securities and income (loss) and impairment from our equity method investments. For additional information, including how we account for our investments and factors that can drive fluctuations in the value of our investments, see Note 1 and Note 3 of the Notes to Consolidated Financial Statements included in Item 8 as well as Item 7A Quantitative and Qualitative Disclosures About Market Risk of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Provision for Income Taxes", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses/Provision for Income Taxes", + "level": 4, + "summary": "Provision for income taxes represents the estimated amount of federal, state, and foreign income taxes incurred in the US and the many jurisdictions in which we operate. The provision includes the effect of reserve provisions and changes to reserves that are considered appropriate as well as the related net interest and penalties. For additional information, including a reconciliation of the US federal statutory rate to our effective tax rate, see Note 14 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Executive Overview", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Executive Overview", + "level": 3, + "summary": "The following table summarizes consolidated financial results (in millions, except for per share information and percentages): [tables/table-5 Financial Results.html] (1) For additional information on the calculation of diluted net income per share, see Note 12 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. - Revenues were \\$402.8 billion, an increase of 15% year over year, primarily driven by an increase in Google Services revenues of \\$37.8 billion, or 12%, and an increase in Google Cloud revenues of \\$15.5 billion, or 36%. - Cost of revenues was \\$162.5 billion, an increase of 11% year over year, primarily driven by increases in TAC, content acquisition costs, and depreciation expense. - Operating expenses were \\$111.3 billion, an increase of 22% year over year, primarily driven by increases in employee compensation expenses, expenses related to legal and other matters, and depreciation expense.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Information:", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Other Information:", + "level": 3, + "summary": "In 2025, Alphabet entered agreements to acquire Wiz for $32.0 billion and Intersect for $4.8 billion, both expected to close in 2026. It issued $37.3 billion in senior notes. Other income and expense included $24.1 billion in net gains on equity securities. Other Bets had a $7.5 billion operating loss including a $2.1 billion Waymo compensation charge. U.S. tax law changes allowed immediate expensing of R&D costs. Share repurchases totaled $45.4 billion. Operating cash flow was $164.7 billion, capital expenditures $91.4 billion, and employee count 190,820. The company noted potential impacts from international trade volatility.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Financial Results", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results", + "level": 3, + "summary": "", + "chunk_count": 9, + "children": [ + { + "title": "Revenues", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues", + "level": 4, + "summary": "The following table presents revenues by type (in millions): [tables/table-6 Revenue Breakdown.html]", + "chunk_count": 8, + "children": [ + { + "title": "Google Services", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Google Services", + "level": 5, + "summary": "Google Advertising Google Search & other Google Search & other revenues increased \\$26.4 billion from 2024 to 2025. The overall growth was driven by interrelated factors including increases in search queries resulting from growth in user adoption and usage on mobile devices; growth in advertiser spending; and improvements we have made in ad formats and delivery.", + "chunk_count": 3, + "children": [ + { + "title": "YouTube ads", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Google Services/YouTube ads", + "level": 6, + "summary": "YouTube ads revenues increased \\$4.2 billion from 2024 to 2025. The growth was driven by our direct response advertising products followed by our brand advertising products, both of which benefited from increased spending by our advertisers.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Google Network", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Google Services/Google Network", + "level": 6, + "summary": "Google Network revenues decreased \\$567 million from 2024 to 2025, primarily due to a decrease in AdSense revenues, partially offset by an increase in AdMob revenues.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Monetization Metrics", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Monetization Metrics", + "level": 5, + "summary": "The following table presents changes in monetization metrics for Google Search & other revenues (paid clicks and cost-per-click) and Google Network revenues (impressions and cost-per-impression), expressed as a percentage, from 2024 to 2025: [tables/table-7 Google Ad Changes.html] Changes in paid clicks and impressions are driven by a number of interrelated factors, including changes in advertiser spending; ongoing product and policy changes; and, as it relates to paid clicks, fluctuations in search queries resulting from changes in user adoption and usage, primarily on mobile devices. Changes in cost-per-click and cost-per-impression are driven by a number of interrelated factors including changes in device mix, geographic mix, advertiser spending, ongoing product and policy changes, product mix, property mix, and changes in foreign currency exchange rates.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Google Subscriptions, Platforms, and Devices", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Google Subscriptions, Platforms, and Devices", + "level": 5, + "summary": "Google subscriptions, platforms, and devices revenues increased \\$7.7 billion from 2024 to 2025. The growth was primarily driven by an increase in subscriptions revenues. This increase was primarily due to the contribution from growth in paid subscriptions across both YouTube services and Google One.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Google Cloud", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Google Cloud", + "level": 5, + "summary": "Google Cloud revenues increased \\$15.5 billion from 2024 to 2025, primarily driven by growth in Google Cloud Platform largely from infrastructure and platform services.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Revenues by Geography", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Results/Revenues/Revenues by Geography", + "level": 5, + "summary": "The following table presents revenues by geography as a percentage of revenues, determined based on the addresses of our customers: [tables/table-8 Revenue by Region.html] (1) Regions represent Europe, the Middle East, and Africa (EMEA); Asia-Pacific (APAC); and Canada and Latin America (\"Other Americas\"). For additional information, see Note 2 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Costs and Expenses_2", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses_2", + "level": 3, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "Cost of Revenues", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses_2/Cost of Revenues", + "level": 4, + "summary": "The following table presents cost of revenues, including TAC (in millions, except percentages): [tables/table-9 Cost of Revenues.html] Cost of revenues increased \\$16.2 billion from 2024 to 2025 due to an increase in other cost of revenues and TAC of \\$11.2 billion and \\$5.0 billion, respectively. The increase in TAC from 2024 to 2025 was largely due to an increase in TAC paid to distribution partners, primarily driven by growth in revenues subject to TAC. The TAC rate decreased from 20.7% to 20.3% from 2024 to 2025, primarily due to a revenue mix shift from Google Network properties to Google Search & other properties. The TAC rates on Google Search & other and Google Network revenues were substantially consistent from 2024 to 2025. The increase in other cost of revenues from 2024 to 2025 was primarily due to increases in content acquisition costs, largely for YouTube, depreciation expense, and other technical infrastructure operations costs.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Research and Development", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses_2/Research and Development", + "level": 4, + "summary": "The following table presents research and development expenses (in millions, except percentages): [tables/table-10 R&D Expenses.html] Research and development expenses increased \\$11.8 billion from 2024 to 2025, primarily driven by increases in employee compensation expenses of \\$6.9 billion and depreciation expense of \\$2.4 billion. The increase in employee compensation expenses was primarily driven by an increase in SBC expenses of \\$4.2 billion, which included an increase in a valuation-based compensation charge related to Waymo.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Sales and Marketing", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses_2/Sales and Marketing", + "level": 4, + "summary": "The following table presents sales and marketing expenses (in millions, except percentages): [tables/table-11 Sales Expenses.html] Sales and marketing expenses increased \\$885 million from 2024 to 2025, primarily driven by an increase in advertising and promotional activities of \\$1.2 billion, partially offset by a decrease in employee compensation expenses of \\$214 million.", + "chunk_count": 1, + "children": [] + }, + { + "title": "General and Administrative", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Costs and Expenses_2/General and Administrative", + "level": 4, + "summary": "The following table presents general and administrative expenses (in millions, except percentages): [tables/table-12 Expense Analysis.html] General and administrative expenses increased \\$7.3 billion from 2024 to 2025, primarily driven by an increase in expenses related to legal and other matters of \\$6.2 billion, largely the result of the \\$3.5 billion EC fine accrued in the third quarter of 2025 and a \\$1.4 billion legal accrual made in the second quarter of 2025.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Segment Profitability", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability", + "level": 3, + "summary": "We report our segment results as Google Services, Google Cloud, and Other Bets. Additionally, certain costs are not allocated to our segments because they represent Alphabet-level activities. For further details on our segments, see Part I, Item 1 Business and Note 15 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. The following table presents segment operating income (loss) (in millions). [tables/table-13 Alphabet Income.html] (1) Alphabet-level activities primarily reflect expenses related to our shared AI research and development.", + "chunk_count": 6, + "children": [ + { + "title": "Google Services", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability/Google Services", + "level": 4, + "summary": "Google Services operating income increased \\$18.1 billion from 2024 to 2025. The increase in operating income was primarily driven by an increase in revenues, partially offset by an increase in expenses related to legal and other matters, TAC, and content acquisition costs.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Google Cloud", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability/Google Cloud", + "level": 4, + "summary": "Google Cloud operating income increased \\$7.8 billion from 2024 to 2025. The increase in operating income was primarily driven by an increase in revenues, partially offset by increases in usage costs for technical infrastructure and employee compensation expenses.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Bets", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability/Other Bets", + "level": 4, + "summary": "Other Bets operating loss increased \\$3.1 billion from 2024 to 2025. The increase in operating loss was primarily driven by an increase in employee compensation expenses largely due to an increase in a valuation-based compensation charge related to Waymo.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Income (Expense), Net", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability/Other Income (Expense), Net", + "level": 4, + "summary": "The following table presents OI&E, (in millions): [tables/table-14 Other Income_Expense.html] OI&E, net increased \\$22.4 billion from 2024 to 2025, primarily due to increases in net unrealized gains on equity securities resulting from fair value adjustments on non-marketable equity securities. For additional information, see Note 3 and Note 7 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Provision for Income Taxes", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Segment Profitability/Provision for Income Taxes", + "level": 4, + "summary": "The effective tax rate increased from 2024 to 2025 due to a decrease in the US Foreign Derived Intangible Income deduction and a non-deductible EC fine, partially offset by prior period tax adjustments. US tax law changes enacted July 4, 2025, allow immediate expensing of R&D and accelerated depreciation, impacting 2025 with some changes effective in 2026. OECD global minimum tax rules (15% rate) did not materially affect 2025 provision. New OECD guidance in January 2026 introduces a safe harbor exempting US domestic operations from global minimum tax, but foreign subsidiaries may still be subject to local minimum taxes, potentially increasing future effective tax rates and cash taxes.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Financial Condition", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition", + "level": 3, + "summary": "", + "chunk_count": 17, + "children": [ + { + "title": "Cash, Cash Equivalents, and Marketable Securities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Cash, Cash Equivalents, and Marketable Securities", + "level": 4, + "summary": "As of December 31, 2025, we had \\$126.8 billion in cash, cash equivalents, and short-term marketable securities. Cash equivalents and marketable securities are comprised of time deposits, money market funds, highly liquid government bonds, corporate debt securities, mortgage-backed and asset-backed securities, and marketable equity securities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Sources, Uses of Cash and Related Trends", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Sources, Uses of Cash and Related Trends", + "level": 4, + "summary": "Our principal sources of liquidity are cash, cash equivalents, and marketable securities, as well as the cash flow that we generate from operations. The primary use of capital continues to be to invest for the long-term growth of the business. We regularly evaluate our cash and capital structure, including the size, pace, and form of capital return to stockholders. The following table presents cash flows (in millions): [tables/table-16 Cash Flow Table.html]", + "chunk_count": 4, + "children": [ + { + "title": "Cash Provided by Operating Activities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Sources, Uses of Cash and Related Trends/Cash Provided by Operating Activities", + "level": 5, + "summary": "Our largest source of cash provided by operations are advertising revenues generated by Google Search & other properties, YouTube properties, and Google Network properties. In Google Services, we also generate cash through consumer subscriptions, the sale of apps and in-app purchases, and devices. In Google Cloud, we generate cash through consumption-based fees and subscriptions for infrastructure, platform, applications, and other cloud services. Our primary uses of cash from operating activities include payments to distribution and Google Network partners, to employees for compensation, and to content providers. Other uses of cash from operating activities include payments to suppliers for devices, to tax authorities for income taxes, and other general corporate expenditures. Net cash provided by operating activities increased from 2024 to 2025 due to an increase in cash received from customers, partially offset by an increase in cash payments for cost of revenues and operating expenses.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Cash Used in Investing Activities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Sources, Uses of Cash and Related Trends/Cash Used in Investing Activities", + "level": 5, + "summary": "Cash provided by investing activities consists primarily of maturities and sales of investments in marketable and non-marketable securities. Cash used in investing activities consists primarily of purchases of marketable and non-marketable securities, purchases of property and equipment, and payments for acquisitions. Net cash used in investing activities increased from 2024 to 2025, primarily due to an increase in purchases of property and equipment, driven by investments in technical infrastructure, and a decrease in maturities and sales of marketable securities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Cash Used in Financing Activities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Sources, Uses of Cash and Related Trends/Cash Used in Financing Activities", + "level": 5, + "summary": "Cash provided by financing activities consists primarily of proceeds from issuance of debt and proceeds from the sale of interests in consolidated entities. Cash used in financing activities consists primarily of repurchases of stock, repayments of debt, net payments related to stock-based award activities, and dividend payments. Net cash used in financing activities decreased from 2024 to 2025 due to an increase in proceeds from issuance of debt and a decrease in repurchases of stock, partially offset by repayments of debt.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Liquidity and Material Cash Requirements", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Liquidity and Material Cash Requirements", + "level": 4, + "summary": "We expect existing cash, cash equivalents, short-term marketable securities, and cash flows from operations and financing activities to continue to be sufficient to fund our operating activities and cash commitments for investing and financing activities for at least the next 12 months, and thereafter for the foreseeable future.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Capital Expenditures and Leases", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases", + "level": 4, + "summary": "We make investments in land, buildings, and servers and network equipment through purchases of property and equipment and lease arrangements to provide capacity for the growth of our services and products.", + "chunk_count": 10, + "children": [ + { + "title": "Capital Expenditures", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Capital Expenditures", + "level": 5, + "summary": "Our capital investments in property and equipment consist primarily of the following major categories: - technical infrastructure, which consists of our investments in servers and network equipment, data center land, and building construction and improvements; and • office facilities, ground-up development projects, and building improvements. Assets not yet in service are those that are not ready for their intended use, including assets in the process of construction or assembly, and consist primarily of technical infrastructure. The time frame from date of purchase to placement in service of these assets may extend from months to years. For example, our data center construction projects are generally multi-year projects with multiple phases, where we acquire land and buildings, construct buildings, and secure and install servers and network equipment. During the years ended December 31, 2024 and 2025, we spent \\$52.5 billion and \\$91.4 billion on capital expenditures, respectively. In 2026, we expect to significantly increase, relative to 2025, our investment in our technical infrastructure, including servers and network equipment, and data centers. Depreciation of our property and equipment commences when such assets are ready for their intended use. For the years ended December 31, 2024 and 2025, our depreciation on property and equipment was \\$15.3 billion and \\$21.1 billion, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Leases", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Leases", + "level": 5, + "summary": "As of December 31, 2025, the amount of total undiscounted future lease payments under operating leases was \\$18.3 billion, of which \\$3.3 billion is short-term, and total undiscounted future lease payments under finance leases was \\$2.9 billion, of which \\$491 million is short-term. As of December 31, 2025, we have entered into leases primarily related to data centers that have not yet commenced with short-term and long-term future lease payments of \\$5.8 billion and \\$52.7 billion, respectively. These leases will commence between 2026 and 2031 with non-cancelable lease terms primarily between one and 25 years. In January 2026, we executed a power purchase agreement which we expect to be accounted for as a lease resulting in future payments depending on certain agreement terms of \\$9.9 billion between 2027 and 2047. If certain contractual conditions for the project are not met, we would instead make a one-time payment of approximately \\$3.5 billion and assume ownership of the power generating assets. For additional information on leases, see Note 4 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Financing", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Financing", + "level": 5, + "summary": "As of December 31, 2025, the company had $48.5 billion in senior unsecured notes outstanding, with $2.0 billion short-term. Future interest payments were $1.8 billion short-term and $35.7 billion long-term. In 2025, it issued $22.5 billion in USD and €13.25 billion in EUR senior unsecured notes, including fixed and floating-rate notes with varying maturities and coupon rates. The company also had $10.0 billion in revolving credit facilities (expiring 2026 and 2030) with no borrowings, and a $25.0 billion commercial paper program with no outstanding balance. Additional details are in Note 6 of the financial statements.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Share Repurchase Program", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Share Repurchase Program", + "level": 5, + "summary": "During 2025, we repurchased and subsequently retired 240 million shares for \\$45.4 billion. In April 2024, the company's Board of Directors authorized a \\$70.0 billion share repurchase program for its Class A and Class C shares. In April 2025, the company's Board of Directors authorized an additional \\$70.0 billion share repurchase program for its Class A and Class C shares. As of December 31, 2025, \\$69.5 billion remained available for Class A and Class C share repurchases. For additional information, see Note 11 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Dividend Program", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Dividend Program", + "level": 5, + "summary": "During the year ended December 31, 2025, total cash dividends were \\$4.8 billion for Class A, \\$703 million for Class B, and \\$4.5 billion for Class C shares, respectively. In April 2025, the company's Board of Directors increased the quarterly cash dividend by 5% to \\$0.21 per share of outstanding Class A, Class B, and Class C shares. The company has declared a quarterly cash dividend in the current quarter, and intends to pay quarterly cash dividends in the future, subject to review and approval by the company's Board of Directors in its sole discretion.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Accrued Legal and Regulatory", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Accrued Legal and Regulatory", + "level": 5, + "summary": "As of December 31, 2025, we had short-term accrued legal and regulatory fines and settlements of \\$15.6 billion. This amount primarily included EC fines, in addition to accruals related to other legal matters and regulatory fines and settlements. For additional information, see Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Taxes", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Taxes", + "level": 5, + "summary": "As of December 31, 2025, we had long-term income taxes payable of \\$9.5 billion primarily related to unrecognized tax benefits. The timing and amount of any payment related to these unrecognized tax benefits are uncertain and cannot be estimated.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Purchase Commitments and Other Contractual Obligations", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Purchase Commitments and Other Contractual Obligations", + "level": 5, + "summary": "As of December 31, 2025, the company had material purchase commitments and contractual obligations totaling $149.1 billion, with $113.0 billion short-term, mainly for technical infrastructure and inventory orders. These include energy take-or-pay contracts, licenses, and inventory orders, reflecting non-cancelable portions or minimum cancellation fees. Additionally, the company provided backstops via financial guarantees ($5.7 billion) and credit derivatives ($16.9 billion). Multi-year power purchase agreements without minimum dollar commitments are also entered into, with payments based on actual volumes. Risks from international trade developments may increase costs.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Pending Acquisitions", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Financial Condition/Capital Expenditures and Leases/Pending Acquisitions", + "level": 5, + "summary": "In March 2025, we entered into a definitive agreement to acquire Wiz, Inc. (\"Wiz\"), a leading cloud security platform, for \\$32.0 billion, subject to closing adjustments, in an all-cash transaction. The acquisition of Wiz is expected to close in 2026, subject to customary closing conditions, including the receipt of regulatory approvals. In December 2025, we entered into a definitive agreement to acquire Intersect, which provides data center and energy infrastructure solutions, for \\$4.8 billion in cash, plus the assumption of debt. The acquisition of Intersect is expected to close in the first half of 2026, subject to customary closing conditions. For additional information, see Note 8 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Critical Accounting Estimates", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Critical Accounting Estimates", + "level": 3, + "summary": "We prepare our consolidated financial statements in accordance with GAAP. In doing so, we have to make estimates and assumptions. Our critical accounting estimates are those estimates that involve a significant level of uncertainty at the time the estimate was made, and changes in them have had or are reasonably likely to have a material effect on our financial condition or results of operations. Accordingly, actual results could differ materially from our estimates. We base our estimates on past experience and other assumptions that we believe are reasonable under the circumstances, and we evaluate these estimates on an ongoing basis. We have reviewed our critical accounting estimates with the Audit Committee of our Board of Directors. For a summary of significant accounting policies and the effect on our financial statements, see Note 1 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K.", + "chunk_count": 5, + "children": [ + { + "title": "Fair Value Measurements of Non-Marketable Equity Securities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Critical Accounting Estimates/Fair Value Measurements of Non-Marketable Equity Securities", + "level": 4, + "summary": "We measure certain financial instruments at fair value on a nonrecurring basis, consisting primarily of our non-marketable equity securities. These investments are accounted for under the measurement alternative method (\"the measurement alternative\") and are measured at cost, less impairment, subject to upward and downward adjustments resulting from observable price changes for identical or similar investments of the same issuer. These adjustments require quantitative assessments of the fair value of our securities, which may require the use of unobservable inputs. Adjustments are determined primarily based on a market approach as of the transaction date and involve the use of estimates using the best information available, which may include cash flow projections or other available market data. Non-marketable equity securities are also evaluated for impairment, based on qualitative factors including the companies' financial and liquidity position and access to capital resources, among others. When indicators of impairment exist, we prepare quantitative measurements of the fair value of our equity investments using a market approach or an income approach, which requires judgment and the use of unobservable inputs, including discount rates, investee revenues and costs, and comparable market data of private and public companies, among others. When the quantitative remeasurements of fair value indicate an impairment exists, we write down the investment to its current fair value.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Property and Equipment", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Critical Accounting Estimates/Property and Equipment", + "level": 4, + "summary": "We assess the reasonableness of the useful lives of our property and equipment periodically or when events indicate a change is necessary. To determine the useful lives of our technical infrastructure, we rely on multiple inputs, including historical asset performance, expected technology advancements, and our future infrastructure deployment plans. Any change in the estimated useful lives is recognized on a prospective basis.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Income Taxes", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Critical Accounting Estimates/Income Taxes", + "level": 4, + "summary": "The passage discusses the company's exposure to income taxes in the US and foreign jurisdictions, emphasizing the need for significant judgment in evaluating uncertain tax positions and determining the provision for income taxes. It describes qualitative considerations such as comparable tax exposures, tax laws, and settlement likelihood, and notes periodic evaluation of these positions based on new regulations, judicial opinions, and audit activities. The provision includes reserves, changes to reserves, and related net interest and penalties. The company is subject to continuous examination by tax authorities, and it regularly assesses the likelihood of adverse outcomes to ensure adequate provisioning.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Loss Contingencies", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 6. [Reserved]/Critical Accounting Estimates/Loss Contingencies", + "level": 4, + "summary": "We are subject to claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders involving competition, intellectual property, data privacy and security, tax and related compliance, labor and employment, commercial disputes, content generated by our users, goods and services offered by advertisers or publishers using our platforms, personal injury, consumer protection, and other matters. Certain of these matters include speculative claims for substantial or indeterminate amounts of damages. We record a liability when we believe that it is probable that a loss has been incurred and the amount can be reasonably estimated. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the possible loss in Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. We evaluate, on a regular basis, developments in our legal matters that could affect the amount of liability that has been previously accrued, and the matters and related reasonably possible losses disclosed, and make adjustments as necessary. Significant judgment is required to determine both the likelihood and the estimated amount of a loss related to such matters. Until the final resolution of such matters, there may be an exposure to loss in excess of the amount recorded, and such amounts could be material.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "ITEM 7A. QUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 7A. QUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK", + "level": 2, + "summary": "We are exposed to financial market risks, including changes in foreign currency exchange rates, interest rates, and equity investment risks.", + "chunk_count": 4, + "children": [ + { + "title": "Foreign Currency Exchange Risk", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 7A. QUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK/Foreign Currency Exchange Risk", + "level": 3, + "summary": "The company transacts globally in multiple currencies, exposing it to foreign exchange rate fluctuations against the US dollar, with principal exposures including the Australian dollar, British pound, Canadian dollar, Euro, and Japanese yen. It monitors and hedges these risks using derivatives and non-derivatives such as forwards, options, cross-currency swaps, and foreign currency-denominated debt. A sensitivity analysis assuming a 10% adverse change in exchange rates estimates the potential impact on financial position, with offsets from derivative contracts and changes in accumulated other comprehensive income (AOCI) for forecasted hedged revenues and foreign subsidiary investments.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Interest Rate Risk", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 7A. QUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK/Interest Rate Risk", + "level": 3, + "summary": "The passage discusses interest rate risk related to the company's investment portfolio and outstanding debt. The Corporate Treasury strategy aims to preserve capital and maintain liquidity. Fixed and floating rate securities carry interest rate risk, with fair value changes recorded in OI&E or AOCI. Value-at-risk (VaR) analysis with a 95% confidence interval estimates potential daily losses, with figures for 2024 and 2025 shown in a table. Actual results may differ due to limitations. Additionally, senior unsecured notes with carrying values of $11.9 billion (2024) and $48.5 billion (2025) bear fixed rates and are recorded at amortized cost, so interest rate fluctuations do not affect financial statements but impact fair value.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Equity Investment Risk", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 7A. QUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK/Equity Investment Risk", + "level": 3, + "summary": "The passage discusses marketable and non-marketable equity securities, their risks, and valuation methods. Marketable equity securities, primarily publicly traded stocks, are recorded at fair value with market price volatility, totaling $6.3 billion as of December 31, 2025, with a hypothetical 10% adverse price change decreasing fair value by $631 million. Non-marketable equity securities, mainly investments in private companies, are adjusted to fair value based on observable transactions or impairment, with carrying values of $64.1 billion as of December 31, 2025. These investments are inherently risky due to early-stage technologies and lack of market data. Equity method investments, totaling $2.5 billion as of December 31, 2025, generally do not fluctuate with market prices but may be impaired. The passage highlights the complexity and volatility of these investments.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA", + "level": 2, + "summary": "Alphabet Inc.", + "chunk_count": 107, + "children": [ + { + "title": "INDEX TO CONSOLIDATED FINANCIAL STATEMENTS", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS", + "level": 3, + "summary": "[tables/table-19 Financial Index.html]", + "chunk_count": 9, + "children": [ + { + "title": "REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM", + "level": 4, + "summary": "To the Stockholders and the Board of Directors of Alphabet Inc.", + "chunk_count": 4, + "children": [ + { + "title": "Opinion on the Financial Statements", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM/Opinion on the Financial Statements", + "level": 5, + "summary": "We have audited the accompanying consolidated balance sheets of Alphabet Inc. (the Company) as of December 31, 2024 and 2025, the related consolidated statements of income, comprehensive income, stockholders' equity and cash flows for each of the three years in the period ended December 31, 2025, and the related notes and financial statement schedule listed in the Index at Item 15 (collectively referred to as the “consolidated financial statements”). In our opinion, the consolidated financial statements present fairly, in all material respects, the financial position of the Company at December 31, 2024 and 2025, and the results of its operations and its cash flows for each of the three years in the period ended December 31, 2025, in conformity with U.S. generally accepted accounting principles. We also have audited, in accordance with the standards of the Public Company Accounting Oversight Board (United States) (PCAOB), the Company's internal control over financial reporting as of December 31, 2025, based on criteria established in Internal Control—Integrated Framework issued by the Committee of Sponsoring Organizations of the Treadway Commission (2013 framework), and our report dated February 4, 2026 expressed an unqualified opinion thereon.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Basis for Opinion", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM/Basis for Opinion", + "level": 5, + "summary": "These financial statements are the responsibility of the Company's management. Our responsibility is to express an opinion on the Company's financial statements based on our audits. We are a public accounting firm registered with the PCAOB and are required to be independent with respect to the Company in accordance with the U.S. federal securities laws and the applicable rules and regulations of the Securities and Exchange Commission and the PCAOB. We conducted our audits in accordance with the standards of the PCAOB. Those standards require that we plan and perform the audit to obtain reasonable assurance about whether the financial statements are free of material misstatement, whether due to error or fraud. Our audits included performing procedures to assess the risks of material misstatement of the financial statements, whether due to error or fraud, and performing procedures that respond to those risks. Such procedures included examining, on a test basis, evidence regarding the amounts and disclosures in the financial statements. Our audits also included evaluating the accounting principles used and significant estimates made by management, as well as evaluating the overall presentation of the financial statements. We believe that our audits provide a reasonable basis for our opinion.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Critical Audit Matter", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM/Critical Audit Matter", + "level": 5, + "summary": "The critical audit matter communicated below is a matter arising from the current period audit of the financial statements that was communicated or required to be communicated to the audit committee and that: (1) relates to accounts or disclosures that are material to the financial statements and (2) involved our especially challenging, subjective or complex judgments. The communication of the critical audit matter does not alter in any way our opinion on the consolidated financial statements, taken as a whole, and we are not, by communicating the critical audit matter below, providing a separate opinion on the critical audit matter or on the accounts or disclosures to which it relates. Loss Contingencies [tables/table-20 Loss Contingencies.html] /s/ Ernst & Young LLP We have served as the Company's auditor since 1999. San Jose, California February 4, 2026", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM_2", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM_2", + "level": 4, + "summary": "To the Stockholders and the Board of Directors of Alphabet Inc.", + "chunk_count": 4, + "children": [ + { + "title": "Opinion on Internal Control Over Financial Reporting", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM_2/Opinion on Internal Control Over Financial Reporting", + "level": 5, + "summary": "We have audited Alphabet Inc.'s internal control over financial reporting as of December 31, 2025, based on criteria established in Internal Control—Integrated Framework issued by the Committee of Sponsoring Organizations of the Treadway Commission (2013 framework) (the COSO criteria). In our opinion, Alphabet Inc. (the Company) maintained, in all material respects, effective internal control over financial reporting as of December 31, 2025, based on the COSO criteria. We also have audited, in accordance with the standards of the Public Company Accounting Oversight Board (United States) (PCAOB), the 2025 consolidated financial statements of the Company and our report dated February 4, 2026 expressed an unqualified opinion thereon.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Basis for Opinion", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM_2/Basis for Opinion", + "level": 5, + "summary": "The Company's management is responsible for maintaining effective internal control over financial reporting and for its assessment of the effectiveness of internal control over financial reporting included in the accompanying Management's Report on Internal Control over Financial Reporting. Our responsibility is to express an opinion on the Company's internal control over financial reporting based on our audit. We are a public accounting firm registered with the PCAOB and are required to be independent with respect to the Company in accordance with the U.S. federal securities laws and the applicable rules and regulations of the Securities and Exchange Commission and the PCAOB. We conducted our audit in accordance with the standards of the PCAOB. Those standards require that we plan and perform the audit to obtain reasonable assurance about whether effective internal control over financial reporting was maintained in all material respects. Our audit included obtaining an understanding of internal control over financial reporting, assessing the risk that a material weakness exists, testing and evaluating the design and operating effectiveness of internal control based on the assessed risk, and performing such other procedures as we considered necessary in the circumstances. We believe that our audit provides a reasonable basis for our opinion.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Definition and Limitations of Internal Control Over Financial Reporting", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/INDEX TO CONSOLIDATED FINANCIAL STATEMENTS/REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM_2/Definition and Limitations of Internal Control Over Financial Reporting", + "level": 5, + "summary": "The passage describes a company's internal control over financial reporting, designed to ensure reliable financial statements in accordance with GAAP. It includes policies for accurate record-keeping, authorized transactions, and asset protection. Inherent limitations mean controls may not prevent all misstatements, and future effectiveness is subject to change. The text also includes audit firm Ernst & Young's signature and references to Alphabet Inc.'s consolidated financial statements for December 2025, including balance sheets, income statements, and cash flows.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS", + "level": 3, + "summary": "", + "chunk_count": 97, + "children": [ + { + "title": "Note 1. Summary of Significant Accounting Policies", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies", + "level": 4, + "summary": "", + "chunk_count": 39, + "children": [ + { + "title": "Nature of Operations", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Nature of Operations", + "level": 5, + "summary": "Google was incorporated in California in September 1998 and re-incorporated in the State of Delaware in August 2003. In 2015, we implemented a holding company reorganization, and as a result, Alphabet Inc. (\"Alphabet\") became the successor issuer to Google. We generate revenues by delivering relevant, cost-effective online advertising; cloud-based solutions that provide enterprise customers of all sizes with infrastructure, platform services, and applications; and sales of other products and services, such as fees received for subscription-based products, apps and in-app purchases, and devices.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Basis of Consolidation", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Basis of Consolidation", + "level": 5, + "summary": "The consolidated financial statements of Alphabet include the accounts of Alphabet and entities consolidated under the variable interest and voting models. Intercompany balances and transactions have been eliminated.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Use of Estimates", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Use of Estimates", + "level": 5, + "summary": "Preparation of consolidated financial statements in conformity with GAAP requires us to make estimates and assumptions that affect the amounts reported and disclosed in the financial statements and the accompanying notes. Actual results could differ materially from these estimates due to uncertainties. On an ongoing basis, we evaluate our estimates, including those related to the allowance for credit losses; contingent liabilities; fair values of financial instruments and goodwill; income taxes; inventory; and useful lives of property and equipment, among others. We base our estimates on assumptions, both historical and forward looking, that are believed to be reasonable, and the results of which form the basis for making judgments about the carrying values of assets and liabilities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Revenue Recognition", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition", + "level": 5, + "summary": "Revenues are recognized when control of the promised goods or services is transferred to our customers, and the collectibility of an amount that we expect in exchange for those goods or services is probable. Sales and other similar taxes are excluded from revenues.", + "chunk_count": 7, + "children": [ + { + "title": "Google Advertising", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Google Advertising", + "level": 6, + "summary": "Google advertising revenues come from Google Search, YouTube, and Google Network properties. Customers purchase inventory through platforms like Google Ads and Google Ad Manager. Performance advertising revenue is recognized upon user engagement, while brand advertising revenue is recognized upon display or view. For Google Network ads, Google generally acts as principal, reporting revenue on a gross basis.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Google Subscriptions, Platforms, and Devices", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Google Subscriptions, Platforms, and Devices", + "level": 6, + "summary": "Google subscriptions, platforms, and devices revenues consist of revenues from: - consumer subscriptions, which primarily include revenues from YouTube services, such as YouTube TV, YouTube Music and Premium, and NFL Sunday Ticket, as well as Google One, which offers access to our most capable Gemini models; - platforms, which primarily include revenues from Google Play sales of apps and in-app purchases; • devices, which primarily include sales of the Pixel family of devices; and • other products and services. Subscription revenues are recognized ratably over the period of the subscription, primarily monthly. We report revenues from Google Play sales of apps and in-app purchases on a net basis because our performance obligation is to facilitate a transaction between app developers and end users for which we earn a service fee.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Google Cloud", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Google Cloud", + "level": 6, + "summary": "Google Cloud revenues consist of revenues from: - Google Cloud Platform primarily generates consumption-based fees and subscriptions for infrastructure, platform, and other services. These services provide access to solutions such as AI offerings including our enterprise AI infrastructure, Vertex AI platform, and Gemini Enterprise; cybersecurity offerings; and data and analytics solutions; - Google Workspace includes subscriptions for cloud-based communication and collaboration tools for enterprises, such as Gmail, Docs, Calendar, Drive, and Meet, with integrated features like Gemini for Google Workspace; and • other enterprise services. Our cloud services are generally provided on either a consumption or subscription basis and may have contract terms longer than a year. Revenues related to cloud services provided on a consumption basis are recognized when the customer utilizes the services, based on the quantity of services consumed using the relative standalone selling price allocation. Revenues related to cloud services provided on a subscription basis are recognized ratably over the contract term as the customer receives and consumes the benefits of the cloud services.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Arrangements with Multiple Performance Obligations", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Arrangements with Multiple Performance Obligations", + "level": 6, + "summary": "Our contracts with customers may include multiple performance obligations. For such arrangements, we allocate revenues to each performance obligation based on its relative standalone selling price. We generally determine standalone selling prices based on observable prices of our products and services sold or priced separately in comparable circumstances to similar customers.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Customer Incentives and Credits", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Customer Incentives and Credits", + "level": 6, + "summary": "Certain customers receive cash-based incentives or credits, which are accounted for as variable consideration. We estimate these amounts based on the expected amount to be provided to customers and reduce revenues. We believe that there will not be significant changes to our estimates of variable consideration related to customer incentives and credits.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Sales Commissions", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Revenue Recognition/Sales Commissions", + "level": 6, + "summary": "We expense sales commissions when incurred when the period of the expected benefit is one year or less. We recognize an asset for certain sales commissions and amortize if the expected benefit period is greater than one year. These costs are recorded within sales and marketing expenses.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Cost of Revenues", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Cost of Revenues", + "level": 5, + "summary": "Cost of revenues consists of TAC and other costs of revenues. • TAC includes: amounts paid to our distribution partners who make available our search access points and other ad-supported services. Our distribution partners include browser providers, mobile carriers, original equipment manufacturers, and software developers; and - amounts paid to Google Network partners primarily for ads displayed on their properties. • Other cost of revenues includes: content acquisition costs, which are payments to content providers from whom we license video and other content for distribution, primarily related to YouTube (we pay fees to these content providers based on revenues generated, subscriber counts, or a flat fee); • depreciation expense, primarily related to our technical infrastructure; • employee compensation expenses related to our technical infrastructure and other operations such as content review and customer and product support; inventory and other costs related to the devices we sell; and other technical infrastructure operations costs, including energy, equipment, and network capacity costs.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Software Development Costs", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Software Development Costs", + "level": 5, + "summary": "We expense software development costs, including costs to develop software products or the software component of products to be sold, leased, or marketed to external users, before technological feasibility is reached. Technological feasibility is typically reached shortly before the release of such products. As a result, development costs that meet the criteria for capitalization were not material for the periods presented. Software development costs also include costs to develop software to be used solely to meet internal needs and cloud-based applications used to deliver our services. We capitalize development costs related to these software applications once the preliminary project stage is complete and it is probable that the project will be completed and the software will be used to perform the function intended. Costs capitalized for developing such software applications were not material for the periods presented.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stock-Based Compensation", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Stock-Based Compensation", + "level": 5, + "summary": "Stock-based compensation (SBC) primarily consists of Alphabet restricted stock units (RSUs). RSUs are equity classified and measured at the fair market value of the underlying stock at the grant date. We recognize RSU expense using the straight-line attribution method over the requisite service period and account for forfeitures as they occur. RSUs are awarded dividend equivalents, which are subject to the same vesting conditions as the underlying award, and settled in Class C shares. For RSUs, shares are issued on the vesting dates net of the applicable statutory income tax withholding to be paid by us on behalf of our employees. As a result, fewer shares are issued than the number of RSUs vested, and the income tax withholding is recorded as a reduction to additional paid-in capital. Additionally, SBC includes other stock-based awards, such as performance stock units (PSUs) that include market conditions and awards that may be settled in cash or the stock of certain Other Bet companies. PSUs and certain awards granted by Other Bet companies are equity classified and expense is recognized over the requisite service period. Certain awards granted by Other Bet companies are liability classified and remeasured at fair value through settlement. The fair value of awards granted by Other Bet companies is based on the equity valuation of the respective Other Bet company.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Advertising and Promotional Expenses", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Advertising and Promotional Expenses", + "level": 5, + "summary": "We expense advertising and promotional costs in the period in which they are incurred. For the years ended December 31, 2023, 2024, and 2025, advertising and promotional expenses totaled approximately \\$8.7 billion, \\$8.7 billion, and \\$9.9 billion, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fair Value Measurements", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Fair Value Measurements", + "level": 5, + "summary": "Fair value is an exit price for selling assets or transferring liabilities in an orderly transaction. It uses a three-tier hierarchy: Level 1 (quoted prices in active markets), Level 2 (observable inputs from similar instruments or models), and Level 3 (unobservable inputs). The hierarchy prioritizes observable inputs. Financial assets and liabilities measured at fair value include cash equivalents, marketable securities, derivatives, and non-marketable equity securities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Financial Instruments", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments", + "level": 5, + "summary": "Our financial instruments include cash, cash equivalents, marketable and non-marketable securities, derivative financial instruments, financial guarantees, accounts receivable, and convertible notes.", + "chunk_count": 10, + "children": [ + { + "title": "Credit Risks", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Credit Risks", + "level": 6, + "summary": "We are subject to concentration of credit risk primarily from cash equivalents, marketable debt securities, derivative financial instruments, including foreign exchange contracts, accounts receivable, and convertible notes. We manage the concentration of our credit risk exposure through timely assessment of our counterparty creditworthiness, credit limits, and use of collateral management. Foreign exchange contracts are transacted with various financial institutions with high credit standing. Accounts receivable are typically unsecured and are derived from revenues earned from customers located around the world. We manage the concentration of our credit risk exposure by performing ongoing evaluations to determine customer credit and we limit the amount of credit we extend. We generally do not require collateral from our customers.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Cash Equivalents", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Cash Equivalents", + "level": 6, + "summary": "We invest excess cash primarily in asset-backed and mortgage-backed securities, corporate debt securities, government bonds, money market funds, and time deposits.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Marketable Securities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Marketable Securities", + "level": 6, + "summary": "The passage describes the classification and accounting treatment of marketable debt securities. Securities with maturities of three months or less are cash equivalents; longer maturities are marketable securities. They are classified as available-for-sale, carried at fair value, with unrealized gains/losses (net of taxes) in stockholders' equity, except credit loss allowances in OI&E. Some securities use the fair value option with changes in OI&E. Realized gains/losses use specific identification and are in OI&E. Marketable equity securities are also at fair value with gains/losses in OI&E.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-Marketable Securities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Non-Marketable Securities", + "level": 6, + "summary": "Non-marketable securities primarily consist of equity securities. We account for non-marketable equity securities through which we exercise significant influence but do not have control over the investee under the equity method. Other non-marketable equity securities that we hold are primarily accounted for under the measurement alternative. Under the measurement alternative, the carrying value is measured at cost, less any impairment, plus or minus changes resulting from observable price changes in orderly transactions for identical or similar investments of the same issuer. Adjustments are determined primarily based on a market approach as of the transaction date and are recorded as a component of OI&E. Non-marketable securities that do not have effective contractual maturity dates are classified as other non-current assets.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Derivative Financial Instruments", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Derivative Financial Instruments", + "level": 6, + "summary": "See Note 3 for the accounting policy pertaining to derivative financial instruments.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Financial Guarantees", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Financial Guarantees", + "level": 6, + "summary": "In certain arrangements, we provide reimbursements for costs incurred by third parties during power generation project development phases if specified trigger events occur. We recognize a noncontingent liability for the fair value of our obligation to stand ready to perform, reported in other long-term liabilities. We also recognize a contingent liability when it becomes probable that a payment will be required and the amount can be reasonably estimated.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Accounts Receivable", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Accounts Receivable", + "level": 6, + "summary": "Our payment terms for accounts receivable vary by the types and locations of our customers and the products or services offered. The term between invoicing and when payment is due is not significant. Additionally, accounts receivable includes amounts for services performed in advance of the right to invoice the customer. We maintain an allowance for credit losses for accounts receivable, which is recorded as an offset to accounts receivable, and changes in such are classified as general and administrative expense. We assess collectibility by reviewing accounts receivable on a collective basis where similar characteristics exist and on an individual basis when we identify specific customers with known disputes or collectibility issues. With respect to current accounts receivables, we elected to assume that current conditions as of the balance sheet date do not change for the remaining life of the asset. In determining the amount of the allowance for credit losses for those assets, we adjust historical loss information to reflect current market conditions and customer-specific information to the extent that historical loss information does not reflect current conditions.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Convertible Notes", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Convertible Notes", + "level": 6, + "summary": "Our investments in convertible notes are primarily recorded at amortized cost which includes unpaid principal balances, deferred origination costs, and any related discount or premium, net of allowances for credit losses, and are included within other non-current assets.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Financial Instruments/Other", + "level": 6, + "summary": "Our financial instruments also include debt and equity investments in companies with which we also entered into commercial arrangements at or near the same time. For these transactions, judgment is required in assessing the substance of the arrangements, including assessing whether the components of the arrangements should be accounted for as separate transactions under the applicable GAAP, and determining the value of the components of the arrangements, including the fair value of the investments. Additionally, if our investment in such companies becomes impaired, we may need to re-evaluate the accounting for the commercial arrangement, including reducing any remaining performance obligations.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Impairment of Investments", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Impairment of Investments", + "level": 5, + "summary": "We periodically review our debt securities with unrealized gains and losses recorded as a component of stockholders' equity and non-marketable equity securities for impairment. For debt securities in an unrealized loss position, we determine whether a credit loss exists. The credit loss is estimated by considering available information relevant to the collectibility of the security and information about past events, current conditions, and reasonable and supportable forecasts. Any credit loss is recorded as a charge to OI&E, not to exceed the amount of the unrealized loss. Unrealized losses other than the credit loss are recognized in AOCI. If we have an intent to sell, or if it is more likely than not that we will be required to sell a debt security in an unrealized loss position before recovery of its amortized cost basis, we will write down the security to its fair value and record the corresponding charge as a component of OI&E. For non-marketable equity securities, including equity method investments, we consider whether impairment indicators exist by evaluating the companies' financial and liquidity position and access to capital resources, among other indicators. If the assessment indicates that the investment is impaired, we write down the investment to its fair value by recording the corresponding charge as a component of OI&E. We prepare quantitative measurements of the fair value of our equity investments using a market approach or an income approach.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Inventory", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Inventory", + "level": 5, + "summary": "Inventory consists primarily of finished goods and is stated at the lower of cost and net realizable value. Cost is generally computed using the first-in, first-out method.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Variable Interest Entities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Variable Interest Entities", + "level": 5, + "summary": "We determine at the inception of each arrangement whether an entity in which we have made an investment or in which we have other variable interests is considered a variable interest entity (VIE). We consolidate VIEs when we are the primary beneficiary. We are the primary beneficiary of a VIE when we have the power to direct activities that most significantly affect the economic performance of the VIE and have the obligation to absorb the majority of their losses or benefits. If we are not the primary beneficiary in a VIE, we account for the investment or other variable interests in a VIE in accordance with applicable GAAP. Periodically, we assess whether any changes in our interest or relationship with the entity affect our determination of whether the entity is a VIE and, if so, whether we are the primary beneficiary.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Property and Equipment", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Property and Equipment", + "level": 5, + "summary": "Property and equipment is comprised of technical infrastructure, office space, corporate and other assets currently in service, and assets not yet in service. Technical infrastructure includes data center land, buildings and leasehold improvements, and servers and network equipment. Office space includes office land, buildings, and leasehold improvements. Assets not yet in service are those that are not ready for their intended use, including data center buildings and servers in the process of construction or assembly. Property and equipment are stated at cost less accumulated depreciation. Depreciation commences once assets are ready for their intended use and is recorded using the straight-line method over the estimated useful lives of the assets, which we regularly evaluate for factors such as technological obsolescence and our planned use and utilization. We depreciate data center and office buildings over periods of seven to 40 years. We depreciate servers and network equipment generally over a period of six years. We depreciate corporate and other assets over periods of two to 25 years. We depreciate leasehold improvements over the shorter of the remaining lease term or the estimated useful lives of the assets. Land is not depreciated.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Goodwill", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Goodwill", + "level": 5, + "summary": "We allocate goodwill to reporting units based on the expected benefit from the business combination. We evaluate our reporting units periodically, as well as when changes in our operating segments occur. For changes in reporting units, we reassign goodwill using a relative fair value allocation approach. We test our goodwill for impairment at least annually, or more frequently if events or changes in circumstances indicate that the asset may be impaired. Goodwill impairments were not material for the periods presented.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Leases", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Leases", + "level": 5, + "summary": "The passage explains how lease arrangements are classified and accounted for. Lease and non-lease components are combined into a single lease component for fixed payments, increasing lease assets and liabilities. Variable payments are expensed as incurred. Lease assets and liabilities are recognized at present value using the incremental borrowing rate. Lease terms include options reasonably certain to be exercised. Operating lease expense is recognized straight-line, while finance lease expense is also straight-line with separate interest expense.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Impairment of Long-Lived Assets", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Impairment of Long-Lived Assets", + "level": 5, + "summary": "We review leases, property and equipment, and intangible assets, excluding goodwill, for impairment when events or changes in circumstances indicate the carrying amount may not be recoverable. The evaluation is performed at the lowest level of identifiable cash flows independent of other assets. We measure recoverability of these assets by comparing the carrying amounts to the future undiscounted cash flows that the assets or the asset group are expected to generate. If the carrying value of the assets or asset group is not recoverable, the impairment recognized is measured as the amount by which the carrying value exceeds its fair value.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Income Taxes", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Income Taxes", + "level": 5, + "summary": "We account for income taxes using the asset and liability method, under which we recognize the amount of taxes payable or refundable for the current year and deferred tax assets and liabilities for the future tax consequences of events that have been recognized in our financial statements or tax returns. We measure current and deferred tax assets and liabilities based on provisions of enacted tax law. We evaluate the likelihood of future realization of our deferred tax assets based on all available evidence and establish a valuation allowance to reduce deferred tax assets when it is more likely than not that they will not be realized or release a valuation allowance to increase deferred tax assets when it is more likely than not that they will be realized. We have elected to account for the tax effects of the global intangible low tax income provision as a current period expense. We recognize the financial statement effects of a tax position when it is more likely than not that, based on technical merits, the position will be sustained upon examination. The tax benefits of the position recognized in the financial statements are then measured based on the largest amount of benefit that is greater than 50% likely to be realized upon settlement with a taxing authority. In addition, we recognize interest and penalties related to unrecognized tax benefits as a component of the income tax provision.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Business Combinations", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Business Combinations", + "level": 5, + "summary": "We include the results of operations of the businesses that we acquire as of the acquisition date. We allocate the purchase price of the acquisitions to the assets acquired and liabilities assumed based on their estimated fair values, except for revenue contracts acquired, which are recognized in accordance with our revenue recognition policy. The excess of the purchase price over the fair values of identifiable assets and liabilities is recorded as goodwill. Acquisition-related expenses are recognized separately from the business combination and are expensed as incurred.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Foreign Currency", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Foreign Currency", + "level": 5, + "summary": "We translate the financial statements of our international subsidiaries to US dollars using month-end exchange rates for assets and liabilities, and average rates for the period derived from month-end exchange rates for revenues, costs, and expenses. We record translation gains and losses in AOCI as a component of stockholders' equity. We reflect net foreign exchange transaction gains and losses resulting from the conversion of the transaction currency to functional currency as a component of foreign currency exchange gain (loss) in OI&E.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recently Issued Accounting Pronouncements Not Yet Adopted", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Recently Issued Accounting Pronouncements Not Yet Adopted", + "level": 5, + "summary": "The passage discusses two recent FASB Accounting Standards Updates. ASU 2024-03, issued in November 2024, requires disaggregation of certain expense categories in financial statement notes, effective for annual 2027 and interim 2028 periods. ASU 2025-06, issued in September 2025, modernizes accounting for internal-use software costs, effective for interim and annual 2028 periods. Both standards permit early adoption and various transition methods. The entity is currently assessing adoption timing, method, and impact on financial statements.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recently Adopted Accounting Pronouncements", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Recently Adopted Accounting Pronouncements", + "level": 5, + "summary": "In December 2023, the FASB issued ASU 2023-09 \"Income Taxes (Topics 740): Improvements to Income Tax Disclosures\" which expands the disclosure requirements for income taxes. We adopted this ASU for our 2025 annual period with the comparative periods updated to reflect additional disclosures. See Note 14 for the revised disclosures consistent with the new standard.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Prior Period Reclassifications", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 1. Summary of Significant Accounting Policies/Prior Period Reclassifications", + "level": 5, + "summary": "Certain amounts in prior periods have been reclassified to conform with current period presentation.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 2. Revenues", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 2. Revenues", + "level": 4, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "Disaggregated Revenues", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 2. Revenues/Disaggregated Revenues", + "level": 5, + "summary": "The following table presents revenues disaggregated by type (in millions): [tables/table-26 Google Revenue.html] No individual customer or groups of affiliated customers represented more than 10% of our revenues in 2023, 2024, or 2025. The following table presents revenues disaggregated by geography, based on the addresses of our customers (in millions): [tables/table-27 Revenue Breakdown.html] (1) Regions represent Europe, the Middle East, and Africa (EMEA); Asia-Pacific (APAC); and Canada and Latin America (\"Other Americas\").", + "chunk_count": 1, + "children": [] + }, + { + "title": "Revenue Backlog", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 2. Revenues/Revenue Backlog", + "level": 5, + "summary": "As of December 31, 2025, we had \\$242.8 billion of remaining performance obligations (\"revenue backlog\"), primarily related to Google Cloud. Revenue backlog represents commitments in customer contracts that have not yet been recognized as revenue. We expect to recognize just over 50% of the revenue backlog as revenues over the next 24 months with the remainder to be recognized thereafter. The estimated revenue backlog and timing of revenue recognition for these commitments is largely driven by contract duration, our ability to deliver in accordance with relevant contract terms, and when our customers utilize services. Revenue backlog includes related deferred revenue currently recorded as well as amounts that will be invoiced in future periods, and excludes contracts with an original expected term of one year or less and cancellable contracts.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Deferred Revenues", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 2. Revenues/Deferred Revenues", + "level": 5, + "summary": "We record deferred revenues when cash payments are received or due in advance of our performance, including amounts which are refundable. Deferred revenues primarily relate to Google Cloud and Google subscriptions, platforms, and devices. Total deferred revenue as of December 31, 2024 was \\$6.0 billion, of which \\$4.6 billion was recognized as revenues for the year ended December 31, 2025. Total deferred revenue as of December 31, 2025 was \\$8.6 billion.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 3. Financial Instruments", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments", + "level": 4, + "summary": "", + "chunk_count": 14, + "children": [ + { + "title": "Fair Value Measurements", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Fair Value Measurements", + "level": 5, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Investments Measured at Fair Value on a Recurring Basis", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Fair Value Measurements/Investments Measured at Fair Value on a Recurring Basis", + "level": 6, + "summary": "Cash equivalents and marketable equity securities are measured at fair value and classified within Level 1 and Level 2 in the fair value hierarchy, because we use quoted prices for identical assets in active markets or inputs that are based upon quoted prices for similar instruments in active markets. Debt securities are measured at fair value and classified within Level 2 in the fair value hierarchy, because we use quoted market prices to the extent available or alternative pricing sources and models utilizing market observable inputs to determine fair value. The following tables summarize our cash, cash equivalents, and marketable securities measured at fair value on a recurring basis (in millions): [tables/table-28 Fair Value Table.html] (1) The long-term portion of marketable equity securities (subject to long-term lock-up restrictions) of \\$266 million as of December 31, 2024 is included within other non-current assets. [tables/table-29 Fair Value Table.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Investments Measured at Fair Value on a Nonrecurring Basis", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Fair Value Measurements/Investments Measured at Fair Value on a Nonrecurring Basis", + "level": 6, + "summary": "Non-marketable equity securities accounted for under the measurement alternative are investments in privately held companies without readily determinable market values. The carrying value of these non-marketable equity securities is adjusted upward or downward to fair value upon observable transactions for identical or similar investments of the same issuer or impairment. Non-marketable equity securities that have been remeasured during the period based on observable transactions are classified within Level 2 or Level 3 in the fair value hierarchy, and remeasurements due to impairment are classified within Level 3. Our valuation methods include option pricing models, market comparable approach, and common stock equivalent method, which may include a combination of the observable transaction price at the transaction date and other unobservable inputs including volatility, expected time to exit, risk free rate, and the rights and obligations of the securities we hold. These inputs vary significantly based on investment type. As of December 31, 2025, the carrying value of our non-marketable equity securities accounted for under the measurement alternative was \\$64.1 billion, of which \\$45.6 billion were remeasured at fair value during the year ended December 31, 2025, and were primarily classified within Level 2 of the fair value hierarchy at the time of measurement.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Debt and Equity Securities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities", + "level": 5, + "summary": "", + "chunk_count": 10, + "children": [ + { + "title": "Debt Securities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Debt Securities", + "level": 6, + "summary": "The following table summarizes the estimated fair value of investments in available-for-sale marketable debt securities by effective contractual maturity dates (in millions): [tables/table-30 Debt Maturity.html] The following tables present fair values and gross unrealized gains and losses recorded to AOCI, less any expected credit losses, aggregated by investment category (in millions): [tables/table-31 Investment Summary.html] [tables/table-32 Investment Summary.html] The following tables present fair values and gross unrealized losses recorded to AOCI, aggregated by investment category and the length of time that individual securities have been in a continuous loss position (in millions): [tables/table-33 Investment Securities.html] [tables/table-34 Investment Securities.html] We determine realized gains or losses on the sale or extinguishment of debt securities on a specific identification method. For certain marketable debt securities, we have elected the fair value option for which changes in fair value are recorded in OI&E. The fair value option was elected for these securities to align with the unrealized gains and losses from related derivative contracts. The following table summarizes gains and losses for debt securities, reflected as a component of OI&E (in millions): [tables/table-35 Debt Securities P&L.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-marketable Securities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Non-marketable Securities", + "level": 6, + "summary": "Our non-marketable securities primarily consist of non-marketable equity securities accounted for under the measurement alternative. The carrying value is measured at the total initial cost plus the cumulative net upward and downward adjustments (including impairments). We account for non-marketable equity securities through which we exercise significant influence but do not have control over the investee under the equity method. Certain of our non-marketable securities include our investments in VIEs where we are not the primary beneficiary. See Note 5 for further details on VIEs. Realized net gain (loss) on equity securities sold during the period reflects the difference between the sale proceeds and the carrying value of the equity securities at the beginning of the period or the purchase date, if later. All gains and losses, including impairments, are included as components of OI&E. The carrying values for non-marketable securities are summarized below (in millions): [tables/table-36 Non-marketable securities.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Gains and Losses on Equity Securities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Gains and Losses on Equity Securities", + "level": 6, + "summary": "Gains and losses (including impairments), net, for equity securities included in OI&E are summarized below (in millions): [tables/table-37 Equity Securities.html] (1) Excludes income (loss) and impairment from equity method investments. Refer to Note 7 for further details. Cumulative net gains (losses), calculated as the difference between the sales price and purchase price, represent the total net gains (losses) recognized after the initial purchase date. This represents the total economic impact of the investment, regardless of when the gains or losses were previously recognized. Cumulative net gains on equity securities sold were \\$748 million and \\$387 million for the years ended December 31, 2024 and 2025, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Derivative Financial Instruments", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments", + "level": 6, + "summary": "We primarily use derivative instruments to manage risks relating to our ongoing business operations, including foreign currencies, interest rates, commodity prices, credit exposures, and market prices of certain marketable equity securities. Additionally, we enter into derivatives to enhance investment returns. We also enter into derivatives as a result of agreements with third parties to backstop certain obligations related to data center leases. These backstop agreements are accounted for as credit derivatives. We recognize derivative instruments in the Consolidated Balance Sheets at fair value and classify them primarily within Level 2 in the fair value hierarchy. We present our foreign currency collars (an option strategy comprised of a combination of purchased and written options) at net fair values and present all other derivatives at gross fair values. The accounting treatment for derivatives is based on the intended use and hedge designation.", + "chunk_count": 6, + "children": [ + { + "title": "Cash Flow Hedges", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments/Cash Flow Hedges", + "level": 7, + "summary": "We designate foreign currency forwards and options (including collars) as cash flow hedges to hedge certain forecasted revenue transactions denominated in currencies other than the US dollar. These contracts have maturities of 24 months or less. Cash flow hedge amounts included in the assessment of hedge effectiveness are deferred in AOCI and reclassified to revenue when the hedged item is recognized in earnings. Hedge components excluded from our assessment of hedge effectiveness are amortized on a straight-line basis over the life of the hedging instrument in revenues. The difference between fair value changes of the excluded component and the amount amortized to revenues is recorded in AOCI. As of December 31, 2025, the net accumulated loss on our foreign currency cash flow hedges before tax effect was \\$60 million, which is expected to be reclassified from AOCI into revenues within the next 12 months. Additionally, we may designate interest rate derivatives as cash flow hedges to manage our exposure to certain interest rate risks. Changes in the fair value of these derivatives are deferred in AOCI and reclassified to OI&E when the hedged item is recognized in earnings.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fair Value Hedges", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments/Fair Value Hedges", + "level": 7, + "summary": "We designate foreign currency forwards as fair value hedges to hedge foreign currency risks for our marketable debt securities denominated in currencies other than the US dollar. Fair value hedge amounts included and excluded from the assessment of hedge effectiveness are recognized in OI&E.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Net Investment Hedges", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments/Net Investment Hedges", + "level": 7, + "summary": "We designate foreign currency forwards, options (including collars), cross-currency swaps, and foreign currency-denominated debt as net investment hedges to hedge the foreign currency risks related to our investments in foreign subsidiaries. Net investment hedge amounts included in the assessment of hedge effectiveness are recognized in AOCI. Changes in the fair value of hedge components of forward and option contracts that are excluded from the assessment of hedge effectiveness are recognized in OI&E. Hedge components of cross-currency swaps that are excluded from the assessment of hedge effectiveness are amortized over the life of the hedging instrument and recognized in OI&E. The difference between fair value changes of the excluded component and the amount amortized to OI&E is recorded in AOCI. We had no foreign currency-denominated debt as of December 31, 2024 and \\$15.4 billion carrying value of foreign currency-denominated debt designated as net investment hedges as of December 31, 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Derivatives Not Designated as Hedging Instruments", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments/Derivatives Not Designated as Hedging Instruments", + "level": 7, + "summary": "The passage describes the company's use of derivatives not designated as hedging instruments to manage risks, primarily foreign exchange risk from remeasurement of monetary assets/liabilities in non-functional currencies. Gains/losses on these are recorded in foreign currency exchange gain/loss. Other derivatives manage risks, enhance returns, or backstop data center lease obligations, with gains/losses in other OI&E. Tables provide gross notional amounts and fair values of outstanding derivatives as of year-end, with notes on credit derivatives' maximum potential payments up to 15 years. Additional tables summarize gains/losses on derivatives in cash flow and net investment hedging relationships recognized in OCI, and gains/losses included in the Consolidated Statements of Income.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Offsetting of Derivatives", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 3. Financial Instruments/Debt and Equity Securities/Derivative Financial Instruments/Offsetting of Derivatives", + "level": 7, + "summary": "We enter into master netting arrangements and collateral security arrangements to reduce credit risk. Cash collateral received related to derivative instruments under our collateral security arrangements are included in other current assets with a corresponding liability. Cash and non-cash collateral pledged related to derivative instruments under our collateral security arrangements are primarily included in other current assets. The gross amounts of derivative instruments subject to master netting arrangements with various counterparties, and cash and non-cash collateral received and pledged under such agreements were as follows (in millions): [tables/table-42 Derivatives Offsetting.html] (1) The balances as of December 31, 2024 and 2025 were related to derivatives allowed to be net settled in accordance with our master netting agreements.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + } + ] + }, + { + "title": "Note 4. Leases", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 4. Leases", + "level": 4, + "summary": "The passage details operating and finance leases for data centers, land, and offices. Lease costs are summarized in tables. Supplemental information includes cash flows, with $1.1 billion in prepayments for finance leases not yet commenced in 2025. Future lease payments as of December 31, 2025, are presented in a table. Additionally, $5.8 billion in short-term and $52.7 billion in long-term future lease payments for not-yet-commenced data center leases are disclosed, commencing between 2026 and 2031. A January 2026 power purchase agreement, expected to be accounted for as a lease, involves future payments of $9.9 billion from 2027 to 2047, or a one-time payment of $3.5 billion if conditions are unmet.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 5. Variable Interest Entities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 5. Variable Interest Entities", + "level": 4, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Consolidated VIEs", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 5. Variable Interest Entities/Consolidated VIEs", + "level": 5, + "summary": "The passage discusses the consolidation of Variable Interest Entities (VIEs) where the company is the primary beneficiary. It details restricted assets and liabilities of certain VIEs, with assets of $8.7 billion and $5.6 billion and liabilities of $2.3 billion and $2.0 billion as of December 31, 2024 and 2025, respectively. Waymo, a consolidated VIE, announced a $16.0 billion investment round in February 2026, mostly funded by Alphabet, with external investments treated as equity transactions leading to noncontrolling interests. Total noncontrolling interests were $4.2 billion and $3.4 billion as of December 31, 2024 and 2025, including redeemable noncontrolling interests of $1.1 billion and $841 million. Net loss attributable to noncontrolling interests was not material.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Unconsolidated VIEs", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 5. Variable Interest Entities/Unconsolidated VIEs", + "level": 5, + "summary": "The passage describes various interests in Variable Interest Entities (VIEs), including investments in private companies and renewable energy entities, leases and credit backstops with data center entities, and backstops with energy infrastructure entities. The company is not the primary beneficiary and does not consolidate these VIEs. Investments are accounted for as non-marketable securities, with carrying values included in non-marketable securities. Maximum exposure is limited to carrying value plus future funding commitments, which were $1.5 billion and $1.1 billion as of December 31, 2024 and 2025, respectively. Leases with data center VIEs are finance leases, with exposure limited to net carrying value of commenced assets plus undiscounted future obligations. Credit backstops are accounted for as credit derivatives, with exposure limited to the notional value. Backstop agreements for energy infrastructure VIEs are financial guarantees, with exposure limited to potential future payments.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 6. Debt", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 6. Debt", + "level": 4, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "Short-Term Debt", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 6. Debt/Short-Term Debt", + "level": 5, + "summary": "We have a commercial paper program of up to \\$25.0 billion, which is used for general corporate purposes. We had \\$2.3 billion of commercial paper outstanding with a weighted-average effective interest rate of 4.4% as of December 31, 2024 and no commercial paper outstanding as of December 31, 2025. The fair value of the commercial paper approximated its carrying value as of December 31, 2024. Our short-term debt balance also includes the current portion of certain long-term debt.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Long-Term Debt", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 6. Debt/Long-Term Debt", + "level": 5, + "summary": "In 2025, the company issued $22.5 billion in US dollar-denominated and €13.25 billion in euro-denominated senior unsecured notes. May 2025 issuances included $5.0 billion of US dollar notes (4.89% coupon, ~24-year maturity) and €6.75 billion of euro notes (3.31% coupon, ~14-year maturity). November 2025 issuances included $500 million floating-rate notes and $17.0 billion fixed-rate US dollar notes (4.92% coupon, ~20-year maturity), plus €6.5 billion fixed-rate euro notes (3.44% coupon, ~16-year maturity). Total outstanding long-term debt is summarized in a table with maturity dates, coupon rates, and effective interest rates. The estimated fair value of outstanding notes was $45.6 billion as of December 31, 2025. Future principal payments include $2,000 million due in 2026.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Credit Facility", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 6. Debt/Credit Facility", + "level": 5, + "summary": "As of December 31, 2025, we had \\$10.0 billion of revolving credit facilities, of which \\$4.0 billion expires in April 2026 and \\$6.0 billion expires in April 2030. The interest rates for all credit facilities are determined based on a formula using certain market rates. No amounts were outstanding under the credit facilities as of December 31, 2024 and 2025.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 7. Supplemental Financial Statement Information", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information", + "level": 4, + "summary": "", + "chunk_count": 6, + "children": [ + { + "title": "Accounts Receivable", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information/Accounts Receivable", + "level": 5, + "summary": "The allowance for credit losses on accounts receivable was \\$879 million and \\$924 million as of December 31, 2024 and 2025, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Property and Equipment, Net", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information/Property and Equipment, Net", + "level": 5, + "summary": "Property and equipment, net, consisted of the following (in millions): [tables/table-49 Property & Equipment.html] (1) As of December 31, 2024 and 2025, approximately 60% of technical infrastructure assets were comprised of servers and network equipment. The remaining balance was comprised of data center land and buildings and related assets.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Accrued Expenses and Other Current Liabilities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information/Accrued Expenses and Other Current Liabilities", + "level": 5, + "summary": "Accrued expenses and other current liabilities consisted of the following (in millions): [tables/table-50 Accrued Liabilities.html] (1) See Legal Matters in Note 10 for further details.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Accumulated Other Comprehensive Income (Loss)", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information/Accumulated Other Comprehensive Income (Loss)", + "level": 5, + "summary": "Components of AOCI, net of income tax, were as follows (in millions): [tables/table-51 OCI Rollforward.html] The effects on net income of amounts reclassified from AOCI were as follows (in millions): [tables/table-52 AOCI Reclass.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Income (Expense), Net", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 7. Supplemental Financial Statement Information/Other Income (Expense), Net", + "level": 5, + "summary": "Components of OI&E were as follows (in millions): [tables/table-53 Income Data Table.html] (1) Interest expense is net of interest capitalized of \\$181 million, \\$194 million, and \\$447 million for the years ended December 31, 2023, 2024, and 2025, respectively.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 8. Acquisitions", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 8. Acquisitions", + "level": 4, + "summary": "Pending Acquisitions In March 2025, we entered into a definitive agreement to acquire Wiz, a leading cloud security platform, for \\$32.0 billion, subject to closing adjustments, in an all-cash transaction. The acquisition of Wiz is expected to close in 2026, subject to customary closing conditions, including the receipt of regulatory approvals. Upon the close of the acquisition, Wiz will be part of the Google Cloud segment. In December 2025, we entered into a definitive agreement to acquire Intersect, which provides data center and energy infrastructure solutions, for \\$4.8 billion in cash, plus the assumption of debt. The acquisition of Intersect is expected to close in the first half of 2026, subject to customary closing conditions.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 9. Goodwill", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 9. Goodwill", + "level": 4, + "summary": "Changes in the carrying amount of goodwill for the years ended December 31, 2024 and 2025 were as follows (in millions): [tables/table-54 Google Segment Balances.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 10. Commitments and Contingencies", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies", + "level": 4, + "summary": "", + "chunk_count": 10, + "children": [ + { + "title": "Commitments", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Commitments", + "level": 5, + "summary": "We have certain content licensing agreements with future fixed or minimum guaranteed commitments of \\$7.7 billion as of December 31, 2025, of which the majority is paid quarterly through the first quarter of 2030.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Financial Guarantees", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Financial Guarantees", + "level": 5, + "summary": "We provide financial guarantees to certain counterparties, in the form of backstop agreements with varying terms through August 2026. These backstop agreements support counterparty procurement of long-lead time equipment for our future power purchase agreements. As of December 31, 2025, our maximum potential amount of future payments under these guarantees was \\$5.7 billion, upon which we may receive certain assets. The fair value of these obligations was not material.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Indemnifications", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Indemnifications", + "level": 5, + "summary": "The passage discusses the company's practice of indemnifying various parties like advertisers, partners, and officers against losses from breaches or claims. It notes that the maximum potential liability cannot be estimated due to unique circumstances, but past payments have not been material. However, future claims could be significant. As of December 31, 2025, no material indemnification claims were probable or reasonably possible.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Legal Matters", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Legal Matters", + "level": 5, + "summary": "We record a liability when we believe that it is probable that a loss has been incurred, and the amount can be reasonably estimated. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the reasonably possible loss. We evaluate developments in our legal matters that could affect the amount of liability that has been previously accrued, and the matters and related reasonably possible losses disclosed, and make adjustments as appropriate. Certain outstanding matters seek speculative, substantial, or indeterminate monetary amounts, substantial changes to our business practices and products, or structural remedies. Significant judgment is required to determine both the likelihood of there being a loss and the estimated amount of a loss related to such matters, and we may be unable to estimate the reasonably possible loss or range of losses. The outcomes of outstanding legal matters are inherently unpredictable and subject to significant uncertainties, and could, either individually or in aggregate, have a material adverse effect. We expense legal fees in the period in which they are incurred.", + "chunk_count": 5, + "children": [ + { + "title": "Antitrust Matters", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Legal Matters/Antitrust Matters", + "level": 6, + "summary": "The passage details numerous antitrust investigations, lawsuits, and fines against Google globally. Key cases include the EC's Shopping fine (€2.4B), Android fine (reduced to €4.1B), AdSense for Search fine (€1.5B, later overturned), and US DOJ cases on Search and advertising technology. Other matters involve the ACCC, JFTC, Google Play settlements, and the EU Digital Markets Act. Fines and charges are noted in euros and US dollars, with ongoing appeals and remedy proceedings.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Privacy Matters", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Legal Matters/Privacy Matters", + "level": 6, + "summary": "We are subject to a number of privacy-related laws and regulations, and we currently are party to a number of privacy investigations and lawsuits ongoing in multiple jurisdictions. For example, there are ongoing investigations and litigation in the US and the EU, including those relating to our collection and use of location information, the choices we offer users, and advertising practices, which could result in significant fines, judgments, and product changes. In October 2025, we finalized a \\$1.4 billion settlement of certain privacy matters.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Patent and Intellectual Property Claims", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Legal Matters/Patent and Intellectual Property Claims", + "level": 6, + "summary": "We have had patent, copyright, trade secret, and trademark infringement lawsuits filed against us claiming that certain of our products, services, and technologies infringe others' intellectual property rights. Adverse results in these lawsuits may include awards of substantial monetary damages, costly royalty or licensing agreements, or orders preventing us from offering certain features, functionalities, products, or services. As a result, we may have to change our business practices and develop non-infringing products or technologies, which could result in a loss of revenues for us and otherwise harm our business. In addition, the ITC has increasingly become an important forum to litigate intellectual property disputes because an ultimate loss in an ITC action can result in a prohibition on importing infringing products into the US. Because the US is an important market, a prohibition on importation could have an adverse effect on us, including preventing us from importing many important products into the US or necessitating workarounds that may limit certain features of our products. Further, our customers and partners may discontinue the use of our products, services, and technologies, as a result of injunctions or otherwise, which could result in loss of revenues and adversely affect our business.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Legal Matters/Other", + "level": 6, + "summary": "We are subject to claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders involving competition, intellectual property, data privacy and security, tax and related compliance, labor and employment, commercial disputes, content generated by our users, goods and services offered by advertisers or publishers using our platforms, design of our products and services, personal injury and other tort and nuisance theories, consumer protection, including how we moderate content on our platforms, AI, and other matters. For example, we periodically have data incidents that we report to relevant regulators as required by law. Such claims, consent orders, lawsuits, regulatory and government investigations, and other proceedings could result in substantial fines and penalties, injunctive relief, ongoing monitoring and auditing obligations, changes to our products and services, alterations to our business models and operations, and collateral related civil litigation or other adverse consequences, all of which could harm our business, reputation, financial condition, and operating results. We have ongoing legal matters relating to Russia. For example, some matters concern civil judgments that include compounding penalties imposed upon us in connection with disputes regarding the termination of accounts, including those of sanctioned parties. We do not expect these ongoing legal matters will have a material adverse effect.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Non-Income Taxes", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 10. Commitments and Contingencies/Non-Income Taxes", + "level": 5, + "summary": "We are under audit by various domestic and foreign tax authorities with regards to non-income tax matters. The subject matter of non-income tax audits primarily arises from disputes on the tax treatment and tax rate applied to the sale of our products and services in these jurisdictions and the tax treatment of certain employee benefits. We accrue non-income taxes that may result from examinations by, or any negotiated agreements with, these tax authorities when a loss is probable and reasonably estimable. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the reasonably possible loss. Due to the inherent complexity and uncertainty of these matters and judicial process in certain jurisdictions, the final outcome may be materially different from our expectations. See Note 14 for further details regarding income tax contingencies.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 11. Stockholders' Equity", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 11. Stockholders' Equity", + "level": 4, + "summary": "Class A and Class B Common Stock and Class C Capital Stock Our Board of Directors has authorized three classes of stock, Class A and Class B common stock, and Class C capital stock. The rights of the holders of each class of our common and capital stock are identical, except with respect to voting. Each share of Class A common stock is entitled to one vote per share. Each share of Class B common stock is entitled to 10 votes per share. Class C capital stock has no voting rights, except as required by applicable law. Shares of Class B common stock may be converted at any time at the option of the stockholder and automatically convert upon sale or transfer to Class A common stock.", + "chunk_count": 3, + "children": [ + { + "title": "Share Repurchases", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 11. Stockholders' Equity/Share Repurchases", + "level": 5, + "summary": "In the years ended December 31, 2023, 2024, and 2025, we continued to repurchase both Class A and Class C shares in a manner deemed in the best interest of the company and its stockholders, taking into account the economic cost and prevailing market conditions, including the relative trading prices and volumes of the Class A and Class C shares. In April 2024, the company's Board of Directors authorized a \\$70.0 billion share repurchase program for its Class A and Class C shares. In April 2025, the company's Board of Directors authorized an additional \\$70.0 billion share repurchase program for its Class A and Class C shares. As of December 31, 2025, \\$69.5 billion remained available for Class A and Class C share repurchases. The following table presents Class A and Class C shares repurchased and subsequently retired (in millions): Year Ended December 31, [tables/table-55 Share Repurchases.html] (1) Shares repurchased include any unsettled repurchases. Repurchases are executed from time to time, subject to general business and market conditions and other investment opportunities, through open market purchases or privately negotiated transactions, including through Rule 10b5-1 plans. The repurchase programs do not have an expiration date.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Dividends", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 11. Stockholders' Equity/Dividends", + "level": 5, + "summary": "During the year ended December 31, 2025, total cash dividends were \\$4.8 billion for Class A, \\$703 million for Class B, and \\$4.5 billion for Class C shares, respectively. In April 2025, the company's Board of Directors increased the quarterly cash dividend by 5% to \\$0.21 per share of outstanding Class A, Class B, and Class C shares. The company has declared a quarterly cash dividend in the current quarter, and intends to pay quarterly cash dividends in the future, subject to review and approval by the company's Board of Directors in its sole discretion.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 12. Net Income Per Share", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 12. Net Income Per Share", + "level": 4, + "summary": "The passage explains the computation of net income per share for Class A, Class B, and Class C stock using the two-class method. Basic EPS uses weighted-average shares outstanding, while diluted EPS includes potentially dilutive securities like RSUs via the treasury stock method. The rights of the three classes are identical except for voting, with safeguards preventing unequal dividends. Immaterial differences in EPS may arise from allocation of distributed versus undistributed earnings. The text references tables showing basic and diluted EPS calculations for multiple years.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 13. Compensation Plans", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 13. Compensation Plans", + "level": 4, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "Stock Plans", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 13. Compensation Plans/Stock Plans", + "level": 5, + "summary": "Our stock plans include the Alphabet Amended and Restated 2021 Stock Plan (\"Alphabet 2021 Stock Plan\") and Other Bets stock-based plans. Under our stock plans, RSUs and other types of awards may be granted. Under the Alphabet 2021 Stock Plan, an RSU award is an agreement to issue shares of our Class C stock at the time the award vests. RSUs generally vest over four years contingent upon employment on the vesting date. RSUs are awarded dividend equivalents, which are subject to the same vesting conditions as the underlying award, and settled in Class C shares. As of December 31, 2025, there were 534 million shares of Class C stock reserved for future issuance under the Alphabet 2021 Stock Plan.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stock-Based Compensation", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 13. Compensation Plans/Stock-Based Compensation", + "level": 5, + "summary": "For the years ended December 31, 2023, 2024, and 2025, total SBC expense was \\$22.1 billion, \\$22.8 billion, and \\$27.1 billion, including amounts associated with awards we expect to settle in Alphabet stock of \\$21.7 billion, \\$22.0 billion, and \\$24.1 billion, respectively. For the years ended December 31, 2023, 2024, and 2025, we recognized tax benefits on total SBC expense, which are reflected in the provision for income taxes, of \\$4.5 billion, \\$4.6 billion, and \\$5.0 billion, respectively. For the years ended December 31, 2023, 2024, and 2025, tax benefit realized related to awards vested or exercised during the period was \\$5.6 billion, \\$6.8 billion, and \\$8.1 billion, respectively. These amounts do not include the indirect effects of stock-based awards, which primarily relate to the research and development tax credit.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stock-Based Award Activities", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 13. Compensation Plans/Stock-Based Award Activities", + "level": 5, + "summary": "The following table summarizes the activities for unvested Alphabet RSUs, which include dividend equivalents awarded to holders of unvested stock, for the year ended December 31, 2025 (in millions, except per share amounts): [tables/table-59 Stock Vesting.html] The weighted-average grant-date fair value of RSUs granted during the years ended December 31, 2023 and 2024 was \\$97.59 and \\$140.04, respectively. Total fair value of RSUs, as of their respective vesting dates, during the years ended December 31, 2023, 2024, and 2025, were \\$26.6 billion, \\$33.3 billion, and \\$39.7 billion, respectively. As of December 31, 2025, there was \\$42.9 billion of unrecognized compensation cost related to unvested RSUs. This amount is expected to be recognized over a weighted-average period of 2.6 years.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 14. Income Taxes", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 14. Income Taxes", + "level": 4, + "summary": "Income from continuing operations before income taxes consisted of the following (in millions): [tables/table-60 Revenue Breakdown.html] Provision for income taxes consisted of the following (in millions): [tables/table-61 Income Tax Provision.html] The reconciliation of federal statutory income tax rate to our effective income tax rate was as follows: [tables/table-62 Tax Rate Reconciliation.html] (1) The tax effect in this category primarily reflects state and local taxes in New York state, New York city, Pennsylvania, Minnesota, Illinois, New Jersey and Wisconsin. In 2023, the IRS issued a rule change allowing taxpayers to temporarily apply the regulations in effect prior to 2022 related to US federal foreign tax credits as well as a separate rule change with guidance on the capitalization and amortization of research and development expenses. A cumulative one-time adjustment for these tax rule changes was recorded in 2023. Changes to US tax law enacted on July 4, 2025, allow for immediate expensing of domestic research and experimentation costs, accelerated depreciation on eligible capital expenditures, and other tax law changes impacting 2025 with certain changes effective in 2026. These changes are reflected in our results for the year ended December 31, 2025.", + "chunk_count": 3, + "children": [ + { + "title": "Deferred Income Taxes", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 14. Income Taxes/Deferred Income Taxes", + "level": 5, + "summary": "The passage discusses deferred income taxes, focusing on temporary differences between financial reporting and tax purposes. As of December 31, 2025, the company has significant net operating loss carryforwards: federal $13.0 billion, state $25.1 billion, and foreign $2.9 billion, all beginning to expire in 2026, with most unlikely to be realized. Research and development credits include federal $771 million (expiring 2037) and California $6.4 billion (indefinite), also mostly unrealizable. Investment tax credits of $1.3 billion expire in 2033. A valuation allowance is maintained for California deferred tax assets, certain federal and state NOLs, tax credits, and foreign NOLs deemed unlikely to be realized. The allowance is reassessed quarterly. Cash paid for income taxes is presented in a referenced table.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Uncertain Tax Positions", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 14. Income Taxes/Uncertain Tax Positions", + "level": 5, + "summary": "The passage discusses the company's gross unrecognized tax benefits, which were $9.4 billion, $12.6 billion, and $11.5 billion as of December 31, 2023, 2024, and 2025, respectively. A significant portion would affect the effective tax rate if recognized. Interest and penalties accrued were $1.1 billion (2024) and $1.2 billion (2025). The company is under IRS examination for 2019-2021 returns and faces transfer pricing or permanent establishment assessments in foreign jurisdictions. Management regularly assesses outcomes and believes adequate provisions have been made, though audit results remain uncertain.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 15. Information about Segments and Geographic Areas", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 15. Information about Segments and Geographic Areas", + "level": 4, + "summary": "The text describes Alphabet's three reportable segments: Google Services (ads, subscriptions, devices), Google Cloud (enterprise services), and Other Bets (autonomous transportation, internet services). It explains revenue sources, cost allocation methods (direct vs. centrally managed), and unallocated costs like AI R&D and corporate initiatives. The CODM uses segment operating income for resource allocation and performance assessment. The passage references tables for segment financials and long-lived assets by geography.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 16. Subsequent Event", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA/Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS/Note 16. Subsequent Event", + "level": 4, + "summary": "In January 2026, we recognized approximately \\$32.0 billion of unrealized gains in our non-marketable investments. These unrealized gains reflect an estimated increase in the fair value measurement following observable transactions that occurred in January 2026, and are subject to change as we finalize related valuations. See Note 3 and Note 7 for further details on equity investments and OI&E. ITEM 9. CHANGES IN AND DISAGREEMENTS WITH ACCOUNTANTS ON ACCOUNTING AND FINANCIAL DISCLOSURE None.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "ITEM 9A. CONTROLS AND PROCEDURES", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9A. CONTROLS AND PROCEDURES", + "level": 2, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "Evaluation of Disclosure Controls and Procedures", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9A. CONTROLS AND PROCEDURES/Evaluation of Disclosure Controls and Procedures", + "level": 3, + "summary": "Our management, with the participation of our chief executive officer and chief financial officer, evaluated the effectiveness of our disclosure controls and procedures pursuant to Rule 13a-15 under the Exchange Act, as of the end of the period covered by this Annual Report on Form 10-K. Based on this evaluation, our chief executive officer and chief financial officer concluded that, as of December 31, 2025, our disclosure controls and procedures are designed at a reasonable assurance level and are effective to provide reasonable assurance that information we are required to disclose in reports that we file or submit under the Exchange Act is recorded, processed, summarized, and reported within the time periods specified in the SEC's rules and forms, and that such information is accumulated and communicated to our management, including our chief executive officer and chief financial officer, as appropriate, to allow timely decisions regarding required disclosure.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Changes in Internal Control over Financial Reporting", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9A. CONTROLS AND PROCEDURES/Changes in Internal Control over Financial Reporting", + "level": 3, + "summary": "There have been no changes in our internal control over financial reporting that occurred during the quarter ended December 31, 2025 that have materially affected, or are reasonably likely to materially affect, our internal control over financial reporting.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Management's Report on Internal Control over Financial Reporting", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9A. CONTROLS AND PROCEDURES/Management's Report on Internal Control over Financial Reporting", + "level": 3, + "summary": "Our management is responsible for establishing and maintaining adequate internal control over financial reporting, as defined in Rule 13a-15(f) of the Exchange Act. Our management conducted an evaluation of the effectiveness of our internal control over financial reporting based on the framework in Internal Control—Integrated Framework issued by the Committee of Sponsoring Organizations of the Treadway Commission (2013 framework). Based on this evaluation, management concluded that our internal control over financial reporting was effective as of December 31, 2025. Management reviewed the results of its assessment with our Audit Committee. The effectiveness of our internal control over financial reporting as of December 31, 2025 has been audited by Ernst & Young LLP, an independent registered public accounting firm, as stated in its report which is included in Item 8 of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Limitations on Effectiveness of Controls and Procedures", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9A. CONTROLS AND PROCEDURES/Limitations on Effectiveness of Controls and Procedures", + "level": 3, + "summary": "In designing and evaluating the disclosure controls and procedures, management recognizes that any controls and procedures, no matter how well designed and operated, can provide only reasonable assurance of achieving the desired control objectives. In addition, the design of disclosure controls and procedures must reflect the fact that there are resource constraints and that management is required to apply its judgment in evaluating the benefits of possible controls and procedures relative to their costs.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ITEM 9B. OTHER INFORMATION", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9B. OTHER INFORMATION", + "level": 2, + "summary": "During the quarter ended December 31, 2025, two Section 16 officers adopted new Rule 10b5-1 trading plans. John Hennessy, Chair of the Board, adopted a plan on November 10, 2025 to sell up to 8,400 shares of Class C Capital Stock and up to 4,200 shares of Class A Common Stock, effective until March 15, 2027. Ruth M. Porat, President and Chief Investment Officer, adopted a plan on November 29, 2025 to sell up to 154,486 shares of Class C Capital Stock upon vesting of performance stock units, effective until March 2, 2027. No non-Rule 10b5-1 trading arrangements were adopted, modified, or terminated. The disclosure also references required reporting under Section 13(r) of the Exchange Act regarding Russian encryption control notifications.", + "chunk_count": 1, + "children": [] + }, + { + "title": "ITEM 9C. DISCLOSURE REGARDING FOREIGN JURISDICTIONS THAT PREVENT INSPECTIONS", + "path": "GOOG-10-K-2025.pdf/PART II/ITEM 9C. DISCLOSURE REGARDING FOREIGN JURISDICTIONS THAT PREVENT INSPECTIONS", + "level": 2, + "summary": "Not applicable.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "PART III", + "path": "GOOG-10-K-2025.pdf/PART III", + "level": 1, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "ITEM 10. DIRECTORS, EXECUTIVE OFFICERS, AND CORPORATE GOVERNANCE", + "path": "GOOG-10-K-2025.pdf/PART III/ITEM 10. DIRECTORS, EXECUTIVE OFFICERS, AND CORPORATE GOVERNANCE", + "level": 2, + "summary": "The information required by this item will be included under the caption \"Directors, Executive Officers, and Corporate Governance\" in our Proxy Statement for the 2026 Annual Meeting of Stockholders to be filed with the SEC within 120 days of the fiscal year ended December 31, 2025 (2026 Proxy Statement) and is incorporated herein by reference. The information required by this item regarding delinquent filers pursuant to Item 405 of Regulation S-K will be included under the caption \"Delinquent Section 16(a) Reports\" in the 2026 Proxy Statement and is incorporated herein by reference. We have adopted insider trading policies and procedures governing the purchase, sale, and other dispositions of securities of Alphabet by directors, officers, and employees that we believe are reasonably designed to promote compliance with insider trading laws, rules and regulations, and applicable Nasdaq listing standards. Our insider trading policy states, among other things, that our directors, officers, and employees are prohibited from trading in such securities while in possession of material, nonpublic information. The foregoing summary of our insider trading policies and procedures does not purport to be complete and is qualified by reference to our Policy Against Insider Trading filed as Exhibit 19.1 to our 2024 Annual Report on Form 10-K and incorporated by reference herein.", + "chunk_count": 1, + "children": [] + }, + { + "title": "ITEM 11. EXECUTIVE COMPENSATION", + "path": "GOOG-10-K-2025.pdf/PART III/ITEM 11. EXECUTIVE COMPENSATION", + "level": 2, + "summary": "The information required by this item will be included under the captions \"Director Compensation,\" \"Executive Compensation\" and \"Directors, Executive Officers, and Corporate Governance—Corporate Governance and Board Matters—Compensation Committee Interlocks and Insider Participation\" in the 2026 Proxy Statement and is incorporated herein by reference, except as to information disclosed therein pursuant to Item 402(v) of Regulation S-K relating to pay versus performance. ITEM 12. SECURITY OWNERSHIP OF CERTAIN BENEFICIAL OWNERS AND MANAGEMENT AND RELATED STOCKHOLDER MATTERS The information required by this item will be included under the captions \"Common Stock Ownership of Certain Beneficial Owners and Management\" and \"Equity Compensation Plan Information\" in the 2026 Proxy Statement and is incorporated herein by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "ITEM 13. CERTAIN RELATIONSHIPS AND RELATED TRANSACTIONS, AND DIRECTOR INDEPENDENCE", + "path": "GOOG-10-K-2025.pdf/PART III/ITEM 13. CERTAIN RELATIONSHIPS AND RELATED TRANSACTIONS, AND DIRECTOR INDEPENDENCE", + "level": 2, + "summary": "The information required by this item will be included under the captions \"Certain Relationships and Related Transactions\" and \"Directors, Executive Officers, and Corporate Governance—Corporate Governance and Board Matters—Director Independence\" in the 2026 Proxy Statement and is incorporated herein by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "ITEM 14. PRINCIPAL ACCOUNTANT FEES AND SERVICES", + "path": "GOOG-10-K-2025.pdf/PART III/ITEM 14. PRINCIPAL ACCOUNTANT FEES AND SERVICES", + "level": 2, + "summary": "The information required by this item will be included under the caption \"Independent Registered Public Accounting Firm\" in the 2026 Proxy Statement and is incorporated herein by reference.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "PART IV", + "path": "GOOG-10-K-2025.pdf/PART IV", + "level": 1, + "summary": "", + "chunk_count": 8, + "children": [ + { + "title": "ITEM 15. EXHIBITS, FINANCIAL STATEMENT SCHEDULES", + "path": "GOOG-10-K-2025.pdf/PART IV/ITEM 15. EXHIBITS, FINANCIAL STATEMENT SCHEDULES", + "level": 2, + "summary": "We have filed the following documents as part of this Annual Report on Form 10-K:", + "chunk_count": 4, + "children": [ + { + "title": "1. Consolidated Financial Statements", + "path": "GOOG-10-K-2025.pdf/PART IV/ITEM 15. EXHIBITS, FINANCIAL STATEMENT SCHEDULES/1. Consolidated Financial Statements", + "level": 3, + "summary": "Reports of Independent Registered Public Accounting Firm 45 Financial Statements: Consolidated Balance Sheets 48 Consolidated Statements of Income 49 Consolidated Statements of Comprehensive Income 50 Consolidated Statements of Stockholders' Equity 51 Consolidated Statements of Cash Flows 52 Notes to Consolidated Financial Statements 53", + "chunk_count": 1, + "children": [] + }, + { + "title": "2. Financial Statement Schedules", + "path": "GOOG-10-K-2025.pdf/PART IV/ITEM 15. EXHIBITS, FINANCIAL STATEMENT SCHEDULES/2. Financial Statement Schedules", + "level": 3, + "summary": "Schedule II: Valuation and Qualifying Accounts The table below details the activity of the allowance for credit losses and sales credits for the years ended December 31, 2023, 2024, and 2025 (in millions): [tables/table-68 Warranty Reserves.html] Note: Additions to the allowance for credit losses are charged to expense. Additions to the allowance for sales credits are charged against revenues. All other schedules have been omitted because they are not required, not applicable, or the required information is otherwise included.", + "chunk_count": 1, + "children": [] + }, + { + "title": "3. Exhibits", + "path": "GOOG-10-K-2025.pdf/PART IV/ITEM 15. EXHIBITS, FINANCIAL STATEMENT SCHEDULES/3. Exhibits", + "level": 3, + "summary": "[tables/table-69 Exhibit List.html] ◆ Indicates management compensatory plan, contract, or arrangement. \\* Filed herewith. ‡ Furnished herewith.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "ITEM 16. FORM 10-K SUMMARY", + "path": "GOOG-10-K-2025.pdf/PART IV/ITEM 16. FORM 10-K SUMMARY", + "level": 2, + "summary": "None.", + "chunk_count": 1, + "children": [] + }, + { + "title": "SIGNATURES", + "path": "GOOG-10-K-2025.pdf/PART IV/SIGNATURES", + "level": 2, + "summary": "Pursuant to the requirements of Section 13 or 15(d) of the Securities Exchange Act of 1934, the Registrant has duly caused this Annual Report on Form 10-K to be signed on its behalf by the undersigned, thereunto duly authorized. Date: February 4, 2026 ALPHABET INC. [tables/table-70 CEO Signature.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "POWER OF ATTORNEY", + "path": "GOOG-10-K-2025.pdf/PART IV/POWER OF ATTORNEY", + "level": 2, + "summary": "KNOW ALL PERSONS BY THESE PRESENTS, that each person whose signature appears below constitutes and appoints Sundar Pichai and Anat Ashkenazi, jointly and severally, his or her attorney-in-fact, with the power of substitution, for him or her in any and all capacities, to sign any amendments to this Annual Report on Form 10-K and to file the same, with exhibits thereto and other documents in connection therewith, with the Securities and Exchange Commission, hereby ratifying and confirming all that each of said attorneys-in-fact, or his or her substitute or substitutes, may do or cause to be done by virtue hereof. Pursuant to the requirements of the Securities Exchange Act of 1934, this Annual Report on Form 10-K has been signed below by the following persons on behalf of the Registrant and in the capacities and on the dates indicated. [tables/table-71 Signatures.html]", + "chunk_count": 1, + "children": [] + } + ] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-Stock Performance Comparison.jpg", + "summary": "image-1 Chart Title: Stock Performance Comparison (implied) Y-Axis: Dollar Value ($) ranging from $0 to $400 X-Axis: Time range from 12/20 to 12/25 Data Series: - Alphabet Inc. Class A (Red Line): Starts at $100, peaks around $165 in late 2021, drops below $100 in late 2022, and shows a sharp upward trend ending above $350 by 12/25. - S&P 500 (Green Line): Starts at $100, fluctuates between roughly $100 and $170 until mid-2024, then rises to just under $200 by 12/25. - NASDAQ Composite (Yellow Line): Starts at $100, generally tracks below the S&P 500, dips significantly in 2022, and ends near $190 by 12/25. - RDG Internet Composite (Blue Line):" + }, + { + "path": "images/image-2-Stock Performance Comparison (2020-2025).jpg", + "summary": "image-2 Chart Title: Stock Performance Comparison (implied by legend and data) Time Range: December 2020 to December 2025 Data Series & Trends: - Alphabet Inc. Class C (Red Line): Starts at $100, peaks around $165 in late 2021, dips below $100 in late 2022, then rises sharply to over $350 by the end of the period. - S&P 500 (Green Line): Starts at $100, fluctuates between $100-$170 until mid-2024, then climbs steadily to approximately $200. - NASDAQ Composite (Yellow Line): Starts at $100, generally follows a similar but lower trajectory than the S&P 500, ending near $190. - RDG Internet Composite (Blue Line): Starts at $100, drops significantly to ~$60 in late 2022, recovers slowly to end around $" + } + ], + "tables": [ + { + "path": "tables/table-0 Stock & Note Listings.html", + "summary": "table-1 Table lists Alphabet's Class A and C stocks (GOOGL, GOOG) on Nasdaq, plus multiple senior notes (2.375%-4.375%) also listed on Nasdaq." + }, + { + "path": "tables/table-1 Filer Status.html", + "summary": "table-2 The table shows filer statuses: Large accelerated filer is checked, while Accelerated filer, Non-accelerated filer, Smaller reporting company, and Emerging growth company are unchecked." + }, + { + "path": "tables/table-2 Share Repurchase.html", + "summary": "table-3 The table shows monthly share repurchases from October to December, with 1,050 Class A and 19,190 Class C shares bought, totaling 20,240 shares under a publicly announced program, leaving $69,503 million available." + }, + { + "path": "tables/table-3 Alphabet Stock vs Indexes.html", + "summary": "table-4 The table tracks Alphabet Inc. Class A stock price alongside S&P 500, NASDAQ Composite, and RDG Internet Composite from Dec 2020 to Dec 2025, showing Alphabet's growth from $100 to $360." + }, + { + "path": "tables/table-4 Alphabet vs S&P 500.html", + "summary": "table-5 The table tracks Alphabet Inc. Class C stock price from $100 (12/20) to $360 (12/25), compared to S&P 500, NASDAQ Composite, and RDG Internet Composite indices over the same period." + }, + { + "path": "tables/table-5 Financial Results.html", + "summary": "table-6 The table shows financial data for years ended December 31, 2024 and 2025. Consolidated revenues increased 15% to $402,836, net income rose 32% to $132,170, and diluted net income per share grew 34% to $10.81." + }, + { + "path": "tables/table-6 Revenue Breakdown.html", + "summary": "table-7 The table shows Google's revenue breakdown for 2024 and 2025, with total revenues rising from $350,018 to $402,836, driven by growth in Search, YouTube, and Cloud." + }, + { + "path": "tables/table-7 Google Ad Changes.html", + "summary": "table-8 Google Search shows 6% paid clicks and 7% cost-per-click increases. Google Network shows 7% impressions decrease and 7% cost-per-impression increase." + }, + { + "path": "tables/table-8 Revenue by Region.html", + "summary": "table-9 The table shows revenue percentages by region for years 2024 and 2025: United States 49% to 48%, EMEA 29% both years, APAC 16% to 17%, Other Americas 6% both years, hedging gains 0%." + }, + { + "path": "tables/table-9 Cost of Revenues.html", + "summary": "table-10 The table shows TAC, other cost of revenues, and total cost of revenues for 2024 and 2025, with total cost as a percentage of revenues decreasing from 42% to 40%." + }, + { + "path": "tables/table-10 R&D Expenses.html", + "summary": "table-11 Research and development expenses increased from $49,326 in 2024 to $61,087 in 2025, rising from 14% to 15% of revenues." + }, + { + "path": "tables/table-11 Sales Expenses.html", + "summary": "table-12 Sales and marketing expenses were $27,808 in 2024 and $28,693 in 2025, with percentages of revenue at 8% and 7% respectively." + }, + { + "path": "tables/table-12 Expense Analysis.html", + "summary": "table-13 General and administrative expenses rose from $14,188 in 2024 to $21,482 in 2025, with the percentage of revenues increasing from 4% to 5%." + }, + { + "path": "tables/table-13 Alphabet Income.html", + "summary": "table-14 The table shows Alphabet's operating income for 2024 and 2025. Google Services earned $121,263M and $139,404M; Google Cloud $6,112M and $13,910M; total income from operations was $112,390M and $129,039M." + }, + { + "path": "tables/table-14 Other Income_Expense.html", + "summary": "table-15 The table shows other income/expense for years ended Dec 31, 2024 and 2025. Total net income increased from $7,425 to $29,787, driven by gains on equity securities." + }, + { + "path": "tables/table-15 Tax Data Table.html", + "summary": "table-16 The table shows income before taxes, provision for income taxes, and effective tax rates for years 2024 and 2025." + }, + { + "path": "tables/table-16 Cash Flow Table.html", + "summary": "table-17 The table shows net cash from operating activities increased from $125,299 in 2024 to $164,713 in 2025, while investing activities used more cash in 2025 ($120,291) than 2024 ($45,536), and financing activities used less cash in 2025 ($37,388) than 2024 ($79,733)." + }, + { + "path": "tables/table-17 Foreign Currency Risk.html", + "summary": "table-18 The table shows foreign currency risk impacts on OI&E and AOCI for 2024 and 2025, with monetary assets/liabilities, cash flow hedges, and net investment hedges." + }, + { + "path": "tables/table-18 Interest Rate Risk.html", + "summary": "table-19 The table shows interest rate risk values for 2024 and 2025, with point-in-time figures of $208 and $162, and 12-month averages of $230 and $184, respectively." + }, + { + "path": "tables/table-19 Financial Index.html", + "summary": "table-20 The table lists page numbers for financial statements including balance sheets, income statements, and auditor reports." + }, + { + "path": "tables/table-20 Loss Contingencies.html", + "summary": "table-21 The company faces claims and lawsuits requiring judgment on loss probability and amount. Audit tested controls, reviewed legal counsel letters, and evaluated disclosures." + }, + { + "path": "tables/table-21 Balance Sheet.html", + "summary": "table-22 Balance sheet as of Dec 31, 2024 and 2025: total assets $450,256 and $595,281; total liabilities $125,172 and $180,016; stockholders' equity $325,084 and $415,265." + }, + { + "path": "tables/table-22 Financial Data.html", + "summary": "table-23 The table shows revenues, costs, and net income for 2023-2025. Revenues grew from $307,394 to $402,836, and net income rose from $73,795 to $132,170." + }, + { + "path": "tables/table-23 Comprehensive Income.html", + "summary": "table-24 The table shows net income, other comprehensive income (OCI) components, and total comprehensive income for 2023-2025. Net income grew from $73,795 to $132,170, while OCI fluctuated, resulting in comprehensive income of $76,996, $99,720, and $135,054 respectively." + }, + { + "path": "tables/table-24 Stockholders' Equity.html", + "summary": "table-25 The table shows stockholders' equity from 2022 to 2025, detailing changes in shares, paid-in capital, comprehensive income, and retained earnings, with net income increasing each year." + }, + { + "path": "tables/table-25 Cash Flow Statement.html", + "summary": "table-26 The table shows cash flow from operating, investing, and financing activities for 2023-2025. Net cash from operations increased from $101,746 to $164,713. Investing activities used more cash over time, while financing activities fluctuated." + }, + { + "path": "tables/table-26 Google Revenue.html", + "summary": "table-27 Google's total revenues grew from $307.4B in 2023 to $402.8B in 2025, driven by Search, YouTube, and Cloud." + }, + { + "path": "tables/table-27 Revenue Breakdown.html", + "summary": "table-28 The table shows total revenues by region for 2023-2025, with US leading at 48-49%, followed by EMEA, APAC, and Other Americas, including hedging gains." + }, + { + "path": "tables/table-28 Fair Value Table.html", + "summary": "table-29 The table presents the fair value hierarchy of cash, cash equivalents, and marketable securities as of December 31, 2024, totaling $95,657, with Level 1 and Level 2 classifications." + }, + { + "path": "tables/table-29 Fair Value Table.html", + "summary": "table-30 The table shows fair value measurements of cash, cash equivalents, and marketable securities as of December 31, 2025, totaling $126,843, with Level 1 and Level 2 inputs." + }, + { + "path": "tables/table-30 Debt Maturity.html", + "summary": "table-31 The table shows debt due by maturity: $26,735 within 1 year, $37,001 in 1-5 years, $12,769 in 5-10 years, $13,317 after 10 years, totaling $89,822 as of December 31, 2025." + }, + { + "path": "tables/table-31 Investment Summary.html", + "summary": "table-32 The table shows investments as of Dec 31, 2024, with total adjusted cost of $62,505, fair value of $61,909, and net unrealized losses of $596 across time deposits, government bonds, corporate debt, and mortgage-backed securities." + }, + { + "path": "tables/table-32 Investment Summary.html", + "summary": "table-33 The table shows investments as of Dec 31, 2025, with total adjusted cost of $85,123, gross unrealized gains of $859, losses of $186, and fair value of $85,796 across time deposits, government bonds, corporate debt, and mortgage-backed securities." + }, + { + "path": "tables/table-33 Investment Securities.html", + "summary": "table-34 The table shows fair value and unrealized losses for government bonds, corporate debt, and mortgage-backed securities as of December 31, 2024, totaling $33,796 and $(784) respectively." + }, + { + "path": "tables/table-34 Investment Securities.html", + "summary": "table-35 The table shows fair value and unrealized losses for government bonds, corporate debt, and mortgage-backed securities as of Dec 31, 2025, totaling $13,160 with $178 unrealized loss." + }, + { + "path": "tables/table-35 Debt Securities P&L.html", + "summary": "table-36 The table shows unrealized gains/losses, realized gains/losses, and credit loss allowances for debt securities from 2023 to 2025, with total net gains/losses of -$1,215, -$1,043, and $540 respectively." + }, + { + "path": "tables/table-36 Non-marketable securities.html", + "summary": "table-37 The table shows non-marketable securities data for 2024 and 2025, including initial cost, cumulative adjustments, carrying value, equity method investments, and total non-marketable securities." + }, + { + "path": "tables/table-37 Equity Securities.html", + "summary": "table-38 The table shows unrealized gains and losses on non-marketable and marketable equity securities from 2023 to 2025, with total net gains rising from $392 to $24,080." + }, + { + "path": "tables/table-38 Derivatives Table.html", + "summary": "table-39 Table shows derivatives designated and not designated as hedging instruments, including foreign exchange contracts, credit derivatives, and other contracts, with values for 2024 and 2025." + }, + { + "path": "tables/table-39 Derivatives Table.html", + "summary": "table-40 The table shows derivative assets and liabilities for 2024 and 2025, including hedging and non-hedging instruments, with totals decreasing from $1,728 to $732 in assets and $612 to $379 in liabilities." + }, + { + "path": "tables/table-40 Hedge Gains_Losses.html", + "summary": "table-41 The table shows gains and losses from cash flow and net investment hedging relationships for 2023-2025, with totals of -113, 1,157, and -2,170 respectively." + }, + { + "path": "tables/table-41 Hedge Effects.html", + "summary": "table-42 The table shows the effect of cash flow, fair value, net investment, and non-designated hedges on revenues and other income for 2023-2025, with total gains of $237, $211, and -$126 for revenues and $262, $658, and $487 for other income." + }, + { + "path": "tables/table-42 Derivatives Offsetting.html", + "summary": "table-43 The table shows derivatives assets and liabilities with gross amounts, offsets, net balances, and collateral for 2024 and 2025. In 2024, net assets were $1,728M and net liabilities $612M; in 2025, net assets were $732M and net liabilities $379M." + }, + { + "path": "tables/table-43 Lease Cost Table.html", + "summary": "table-44 The table shows lease costs for 2023-2025: operating lease cost around $3.3K, finance lease cost increasing from $504 to $618, variable lease cost rising from $1,182 to $1,739, and total lease cost growing from $5,048 to $5,702." + }, + { + "path": "tables/table-44 Lease Data.html", + "summary": "table-45 The table presents weighted-average remaining lease terms and discount rates for operating and finance leases for 2024 and 2025, along with corresponding lease assets and liabilities." + }, + { + "path": "tables/table-45 Lease Cash Flows.html", + "summary": "table-46 The table shows cash payments for operating and finance leases from 2023 to 2025, including operating cash flows and financing cash flows, plus assets obtained in exchange for lease liabilities." + }, + { + "path": "tables/table-46 Lease Payments.html", + "summary": "table-47 The table shows future lease payments for operating and finance leases from 2026 to 2030 and thereafter, totaling $18,251 and $2,869 undiscounted, with imputed interest deducted to yield lease liabilities of $15,954 and $2,500." + }, + { + "path": "tables/table-47 Debt Table.html", + "summary": "table-48 The table shows long-term debt with maturities from 2026 to 2075, coupon rates from 0.80% to 5.70%, and effective interest rates from 0.93% to 5.79%. Total face value increased from $12,000 in 2024 to $49,085 in 2025, with net long-term debt of $10,883 and $46,547 respectively." + }, + { + "path": "tables/table-48 Debt Maturity.html", + "summary": "table-49 The table shows a debt maturity schedule from 2026 to 2030 and thereafter, with total obligations of $49,085." + }, + { + "path": "tables/table-49 Property & Equipment.html", + "summary": "table-50 The table shows property and equipment net values of $171,036 in 2024 and $246,597 in 2025, with categories including technical infrastructure, office space, and accumulated depreciation." + }, + { + "path": "tables/table-50 Accrued Liabilities.html", + "summary": "table-51 The table lists accrued expenses and other current liabilities as of December 31, 2024 and 2025, with totals of $51,228 and $55,557 respectively, including fines, property purchases, customer liabilities, and more." + }, + { + "path": "tables/table-51 OCI Rollforward.html", + "summary": "table-52 The table shows the rollforward of Accumulated Other Comprehensive Income (AOCI) from 2022 to 2025, detailing changes in foreign currency translation, available-for-sale investments, and cash flow hedges." + }, + { + "path": "tables/table-52 AOCI Reclass.html", + "summary": "table-53 The table shows reclassifications of AOCI components for 2023-2025, including unrealized gains/losses on investments and cash flow hedges, with tax impacts and net totals." + }, + { + "path": "tables/table-53 Income Data Table.html", + "summary": "table-54 The table shows financial data for 2023-2025: interest income rises then falls, interest expense increases sharply in 2025, and net other income grows significantly, driven by equity securities gains." + }, + { + "path": "tables/table-54 Google Segment Balances.html", + "summary": "table-55 The table shows goodwill balances for Google Services, Google Cloud, and Other Bets from 2023 to 2025, with additions and currency adjustments, totaling $33,380 million by end of 2025." + }, + { + "path": "tables/table-55 Share Repurchases.html", + "summary": "table-56 The table shows Class A and Class C share repurchases from 2023 to 2025, with total shares and amounts decreasing over the years." + }, + { + "path": "tables/table-56 EPS Breakdown.html", + "summary": "table-57 The table details basic and diluted net income per share for Class A, B, and C shares for the year ended December 31, 2023, showing allocations of earnings and share counts." + }, + { + "path": "tables/table-57 EPS Calculation.html", + "summary": "table-58 The table details basic and diluted net income per share for Class A, B, and C shares for the year ended December 31, 2024, with consolidated net income of $100,118 and basic EPS of $8.13." + }, + { + "path": "tables/table-58 EPS Calculation.html", + "summary": "table-59 The table shows basic and diluted net income per share for Class A, B, and C shares for the year ended December 31, 2025, with allocations of distributed and undistributed earnings, share counts, and resulting EPS values." + }, + { + "path": "tables/table-59 Stock Vesting.html", + "summary": "table-60 The table shows unvested shares changing from 299 to 282 during 2025, with 198 granted, 181 vested, and 34 forfeited. Weighted-average fair values are provided for each category." + }, + { + "path": "tables/table-60 Revenue Breakdown.html", + "summary": "table-61 The table shows total revenue from domestic and foreign operations for 2023-2025, increasing from $85,717 to $158,826." + }, + { + "path": "tables/table-61 Income Tax Provision.html", + "summary": "table-62 The table shows current and deferred income tax provisions for 2023-2025. Current taxes total $19,651, $24,953, $18,406; deferred taxes total ($7,729), ($5,256), $8,250; total provision is $11,922, $19,697, $26,656." + }, + { + "path": "tables/table-62 Tax Rate Reconciliation.html", + "summary": "table-63 The table shows a reconciliation of the US federal statutory income tax rate to the effective tax rate for 2023-2025, listing adjustments like state taxes, foreign effects, credits, and valuation allowances." + }, + { + "path": "tables/table-63 Deferred Tax Assets.html", + "summary": "table-64 The table shows deferred tax assets and liabilities for 2024 and 2025. Total deferred tax assets net of valuation allowance were $33,364 in 2024 and $34,084 in 2025. Net deferred tax assets decreased from $16,460 to $8,194." + }, + { + "path": "tables/table-64 Income Tax Payments.html", + "summary": "table-65 The table shows cash paid for income taxes (net of refunds) for 2023-2025: US federal, state/local, and foreign (Brazil, Other). Total: $19,164 (2023), $27,353 (2024), $21,526 (2025)." + }, + { + "path": "tables/table-65 Tax Benefits.html", + "summary": "table-66 The table shows changes in gross unrecognized tax benefits from 2023 to 2025, including increases and decreases from prior and current year positions and settlements, ending at $11,512 in 2025." + }, + { + "path": "tables/table-66 Alphabet Revenue.html", + "summary": "table-67 Alphabet's total revenues grew from $307B in 2023 to $403B in 2025, driven by Google Services and Cloud. Operating income rose from $84B to $129B, with Google Cloud turning profitable." + }, + { + "path": "tables/table-67 Long-lived Assets.html", + "summary": "table-68 The table shows long-lived assets as of December 31, 2024 and 2025. United States assets grew from $138,993 to $195,337, International from $45,631 to $66,481, totaling $184,624 and $261,818 respectively." + }, + { + "path": "tables/table-68 Warranty Reserves.html", + "summary": "table-69 The table shows warranty reserve balances for 2023-2025, with beginning balances, additions, usage, and ending balances. The ending balance increased from $1,213 to $2,356 over the three years." + }, + { + "path": "tables/table-69 Exhibit List.html", + "summary": "table-70 HTML table listing exhibits (2.01-104) for Alphabet Inc., including descriptions and incorporation by reference details for SEC filings." + }, + { + "path": "tables/table-70 CEO Signature.html", + "summary": "table-71 The document is signed by Sundar Pichai, Chief Executive Officer and Principal Executive Officer of the registrant." + }, + { + "path": "tables/table-71 Signatures.html", + "summary": "table-72 Signatures of Sundar Pichai, Anat Ashkenazi, Amie Thuener O'Toole, and directors including Frances Arnold, Sergey Brin, and others, all dated February 4, 2026." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/full.md b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/full.md new file mode 100644 index 000000000..07a0cb035 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/full.md @@ -0,0 +1,2198 @@ +# UNITED STATES + +# SECURITIES AND EXCHANGE COMMISSION + +Washington, D.C. 20549 + +# FORM 10-K + +(Mark One) + +ANNUAL REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 + +For the fiscal year ended December 31, 2025 + +OR + +☐ TRANSITION REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 + +For the transition period from \_\_\_\_ to \_\_\_\_. + +Commission file number: 001-37580 + +# Alphabet Inc. + +(Exact name of registrant as specified in its charter) + +Delaware + +(State or other jurisdiction of incorporation or organization) + +61-1767919 + +(I.R.S. Employer Identification No.) + +1600 Amphitheatre Parkway + +Mountain View, CA 94043 + +(Address of principal executive offices, including zip code) + +(650) 253-0000 + +(Registrant's telephone number, including area code) + +Securities registered pursuant to Section 12(b) of the Act: + +
Title of each classTrading Symbol(s)Name of each exchange on which registered
Class A Common Stock, $0.001 par valueGOOGLNasdaq Stock Market LLC(Nasdaq Global Select Market)
Class C Capital Stock, $0.001 par valueGOOGNasdaq Stock Market LLC(Nasdaq Global Select Market)
2.375% Senior Notes due 2028Nasdaq Stock Market LLC
2.500% Senior Notes due 2029Nasdaq Stock Market LLC
2.875% Senior Notes due 2031Nasdaq Stock Market LLC
3.000% Senior Notes due 2033Nasdaq Stock Market LLC
3.125% Senior Notes due 2034Nasdaq Stock Market LLC
3.375% Senior Notes due 2037Nasdaq Stock Market LLC
3.500% Senior Notes due 2038Nasdaq Stock Market LLC
4.000% Senior Notes due 2044Nasdaq Stock Market LLC
3.875% Senior Notes due 2045Nasdaq Stock Market LLC
4.000% Senior Notes due 2054Nasdaq Stock Market LLC
4.375% Senior Notes due 2064Nasdaq Stock Market LLC
+ +Securities registered pursuant to Section 12(g) of the Act: + +Title of each class + +None + +Indicate by check mark whether the registrant (1) has filed all reports required to be filed by Section 13 or 15(d) of the Securities Exchange Act of 1934 during the preceding 12 months (or for such shorter period that the registrant was required to file such reports), and (2) has been subject to such filing requirements for the past 90 days. Yes ☑ No □ + +Indicate by check mark whether the registrant has submitted electronically every Interactive Data File required to be submitted pursuant to Rule 405 of Regulation S-T ( $§232.405$ of this chapter) during the preceding 12 months (or for such shorter period that the registrant was required to submit such files). Yes ☒ No ☐ + +Indicate by check mark whether the registrant is a large accelerated filer, an accelerated filer, a non-accelerated filer, a smaller reporting company, or an emerging growth company. See the definitions of “large accelerated filer,” “accelerated filer,” “smaller reporting company,” and "emerging growth company" in Rule 12b-2 of the Exchange Act. + +
Large accelerated filerAccelerated filer
Non-accelerated filerSmaller reporting company
Emerging growth company
+ +If an emerging growth company, indicate by check mark if the registrant has elected not to use the extended transition period for complying with any new or revised financial accounting standards provided pursuant to Section 13(a) of the Exchange Act. □ + +Indicate by check mark whether the registrant has filed a report on and attestation to its management's assessment of the effectiveness of its internal control over financial reporting under Section 404(b) of the Sarbanes-Oxley Act (15 U.S.C.7262(b)) by the registered public accounting firm that prepared or issued its audit report. ☒ + +If securities are registered pursuant to Section 12(b) of the Act, indicate by check mark whether the financial statements of the registrant included in the filing reflect the correction of an error to previously issued financial statements. □ + +Indicate by check mark whether any of those error corrections are restatements that required a recovery analysis of incentive-based compensation received by any of the registrant's executive officers during the relevant recovery period pursuant to §240.10D-1(b). □ + +Indicate by check mark whether the registrant is a shell company (as defined in Rule 12b-2 of the Exchange Act). Yes □ No ☒ + +As of June 30, 2025, the aggregate market value of shares held by non-affiliates of the registrant (based upon the closing sale prices of such shares on the Nasdaq Global Select Market on June 30, 2025) was approximately \$1.9 trillion. For purposes of calculating the aggregate market value of shares held by non-affiliates, we have assumed that all outstanding shares are held by non-affiliates, except for shares held by each of our executive officers, directors, and 5% or greater stockholders. In the case of 5% or greater stockholders, we have not deemed such stockholders to be affiliates unless there are facts and circumstances which would indicate that such stockholders exercise any control over our company, or unless they hold 10% or more of our outstanding common stock. These assumptions should not be deemed to constitute an admission that all executive officers, directors, and 5% or greater stockholders are, in fact, affiliates of our company, or that there are not other persons who may be deemed to be affiliates of our company. Further information concerning shareholdings of our officers, directors, and principal stockholders is included or incorporated by reference in Part III, Item 12 of this Annual Report on Form 10-K. + +As of January 28, 2026, there were 5,822 million shares of Alphabet's Class A stock outstanding, 837 million shares of Alphabet's Class B stock outstanding, and 5,438 million shares of the Alphabet's Class C stock outstanding. + +## DOCUMENTS INCORPORATED BY REFERENCE + +Portions of the registrant's Proxy Statement for the 2026 Annual Meeting of Stockholders are incorporated herein by reference in Part III of this Annual Report on Form 10-K to the extent stated herein. Such proxy statement will be filed with the Securities and Exchange Commission within 120 days of the registrant's fiscal year ended December 31, 2025. + +## Alphabet Inc. + +## Form 10-K + +## For the Fiscal Year Ended December 31, 2025 + +TABLE OF CONTENTS + +
Note About Forward-Looking StatementsPage
3
PART I
Item 1.Business3
Item 1A.Risk Factors9
Item 1B.Unresolved Staff Comments23
Item 1C.Cybersecurity23
Item 2.Properties24
Item 3.Legal Proceedings24
Item 4.Mine Safety Disclosures24
PART II
Item 5.Market for Registrant's Common Equity, Related Stockholder Matters, and Issuer Purchases of Equity Securities25
Item 6.[Reserved]27
Item 7.Management's Discussion and Analysis of Financial Condition and Results of Operations28
Item 7A.Quantitative and Qualitative Disclosures About Market Risk41
Item 8.Financial Statements and Supplementary Data44
Item 9.Changes in and Disagreements With Accountants on Accounting and Financial Disclosure89
Item 9A.Controls and Procedures89
Item 9B.Other Information89
Item 9C.Disclosure Regarding Foreign Jurisdictions that Prevent Inspections90
PART III
Item 10.Directors, Executive Officers, and Corporate Governance91
Item 11.Executive Compensation91
Item 12.Security Ownership of Certain Beneficial Owners and Management and Related Stockholder Matters91
Item 13.Certain Relationships and Related Transactions, and Director Independence91
Item 14.Principal Accountant Fees and Services91
PART IV
Item 15.Exhibits, Financial Statement Schedules92
Item 16.Form 10-K Summary96
Signatures
+ +# Note About Forward-Looking Statements + +This Annual Report on Form 10-K contains forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. Forward-looking statements generally can be identified by words such as, but are not limited to, "anticipates," "believes," "could," "estimates," "expects," "intends," "may," "plans," "predicts," "projects," "will be," "will continue," "will likely result," and similar expressions. These include, among other things, expectations regarding the growth of our business and revenues, including factors that may impact such growth, and fluctuations in our revenues and margins; statements relating to plans, expectations, and trends about our core business metrics, costs and expenses, capital expenditures, sources of funding, products and services, strategic business transactions, and other aspects of our business operations and strategies; statements regarding the global macroeconomic and regulatory environment; as well as other statements regarding our future operations, financial condition and prospects, and actual or potential risk and liability exposures. Forward-looking statements may appear throughout this report and other documents we file with the Securities and Exchange Commission (SEC), including without limitation, the following sections: Part I, Item 1 "Business;" Part I, Item 1A "Risk Factors;" and Part II, Item 7 "Management's Discussion and Analysis of Financial Condition and Results of Operations." These forward-looking statements are based on current expectations and assumptions that are subject to risks and uncertainties, which could cause our actual results to differ materially from those reflected in the forward-looking statements. Factors that could cause or contribute to such differences include, but are not limited to, those discussed in this Annual Report on Form 10-K, including the risks discussed in Part I, Item 1A "Risk Factors" and the trends discussed in Part II, Item 7 "Management's Discussion and Analysis of Financial Condition and Results of Operations," and those discussed in other documents we file with the SEC. We undertake no obligation to revise or publicly release the results of any revision to these forward-looking statements, except as required by law. Given these risks and uncertainties, readers are cautioned not to place undue reliance on such forward-looking statements. + +As used herein, "Alphabet," "the company," "we," "us," "our," and similar terms include Alphabet Inc. and its subsidiaries, unless the context indicates otherwise. + +"Alphabet," "Google," and other trademarks of ours appearing in this report are our property. We do not intend our use or display of other companies' trade names or trademarks to imply an endorsement or sponsorship of us by such companies, or any relationship with any of these companies. + +## PART I + +## ITEM 1. BUSINESS + +## Overview + +As our founders Larry and Sergey wrote in the original founders' letter, "Google is not a conventional company. We do not intend to become one." That unconventional spirit has been a driving force throughout our history, inspiring us to tackle big problems and invest in moonshots. It led us to be a pioneer in the development of artificial intelligence (AI) and, since 2016, be an AI-first company. We continue this work under the leadership of Alphabet and Google CEO, Sundar Pichai. + +Alphabet is a collection of businesses — the largest of which is Google. We report Google in two segments, Google Services and Google Cloud, and all non-Google businesses collectively as Other Bets. Supporting these businesses, we have centralized certain AI-related research and development focused on advanced research in AI and developing the frontier models that serve our businesses, which is reported in Alphabet-level activities. Alphabet's structure is about helping each of our businesses prosper through strong leaders and independence. + +## Access and Technology for Everyone + +The Internet is one of the world's most powerful equalizers; it propels ideas, people, and businesses large and small. Our mission to organize the world's information and make it universally accessible and useful is as relevant today as it was when we were founded in 1998. Since then, we have evolved from a company that helps people find answers to a company that also helps people get things done. + +We are focused on building an even more helpful Google for everyone, and we aspire to give everyone the tools they need to increase their knowledge, health, happiness, and success. Google Search helps people find information and make sense of the world in more natural and intuitive ways, with trillions of searches on Google every year. YouTube provides people with entertainment, information, and opportunities to learn something new and helps support the creator economy through the YouTube Partner Program. Google Cloud helps customers build for the future, improve productivity, reduce costs, and unlock new growth engines. We continually innovate and build new products and features to help our users, partners, customers, and communities and have invested more than \$200 billion in research and development in the last five years in support of these efforts. + +## Making AI Helpful for Everyone + +We believe AI is a profound platform shift that can bring meaningful and positive change to people and societies across the world, and to our business. We aim to build the most advanced, safe, and responsible AI through our full-stack approach, which spans AI-optimized infrastructure; world-class research, including models and tooling; and our products and platforms that bring AI to billions of people, developers, and enterprises. + +At the foundation of our full-stack approach is our AI-optimized infrastructure — a key differentiator enabling us to power our own products, such as Search and YouTube, and support the services we provide to our Google Cloud customers. Our technical infrastructure allows us to use and offer our customers a range of AI accelerator options, including specialized Graphics Processing Units (GPUs) and our own custom-built Tensor Processing Units (TPUs), such as Ironwood, our seventh-generation TPU. We are focused on driving efficiencies in our data centers, allowing us to leverage our technical infrastructure to deliver our products and services at an increasing scale while simultaneously enabling world-class research and model development. + +Over the last decade, our research teams have pushed the boundaries of AI forward, which is displayed through Gemini 3, our most intelligent AI model yet. Designed to deliver advanced multimodal understanding, Gemini 3 represents our most capable iteration of agentic and generative coding technologies. Gemini 3 integrates enhanced reasoning capabilities to support visualizations and interactive user experiences across our product ecosystem, including Search and the Gemini app. + +As technology continues to improve rapidly, we are focused on bringing our latest AI advances to our products and platforms. We continue to help our users access information and knowledge, express themselves, and get things done by embedding the power of generative AI and Gemini into our products and platforms. Today, all 15 of our half-billion-user products — including seven with two billion users — use our Gemini models. For our Google Cloud customers, our offerings are helping organizations stay at the forefront of innovation with solutions such as Gemini Enterprise and Gemini for Google Workspace. + +Guided by our AI principles, we believe our approach to AI must be both bold and responsible. That means developing AI in a way that maximizes the positive benefits to society while addressing its potential challenges. + +## Moonshots + +Many companies get comfortable doing what they have always done, making only incremental changes. This incrementalism leads to irrelevance over time, especially in technology, where change tends to be revolutionary, not evolutionary. + +Our early investments in AI started out as moonshots but are now incorporated into our core products and central to future developments. In Other Bets, our fully autonomous driving technology company, Waymo, is now providing fully autonomous, paid ride-hailing services to customers in multiple cities. Isomorphic Labs is reimagining the drug discovery process from first principles, applying AI to accelerate the development of new medicines. We continue to look toward the future and to invest for the long term, most notably for the application of AI to our products and services, as well as other frontier technologies such as quantum computing. + +## Privacy and Security + +We make it a priority to protect the privacy and security of our products, users, and customers, even if there are near-term financial consequences. We do this by continuously investing in building products that are secure by default; strictly upholding responsible data practices that emphasize privacy by design; and building easy-to-use settings that put people in control. We are continually enhancing these efforts over time, whether by enabling users to auto-delete their data, applying privacy technologies like on-device processing, giving people tools to control their experience, or advancing anti-malware, anti-phishing, and password security features. + +## Google + +For reporting purposes Google comprises two segments: Google Services and Google Cloud. + +## Google Services + +## Serving Our Users + +We have always been committed to building helpful products that can improve the lives of millions of people worldwide. Our product innovations are what make our services widely used, and our brand one of the most recognized in the world. Google Services' core products and platforms include ads, Android, Chrome, devices, Gmail, Google Drive, Google Gemini, Google Maps, Google Photos, Google Play, Search, and YouTube, with broad and growing adoption by users around the world. + +Our products and services have come a long way since the company was founded more than 25 years ago. While Google Search started as a way to find web pages, organized into ten blue links, we have driven technical advancements and product innovations that have transformed Google Search into a dynamic, multimodal experience. Large language models have made it possible to express more natural language queries, vastly improving the types of questions users can ask, and the quality of results. For example, AI Overviews makes it easier to ask Google anything and get a helpful response. AI Mode allows users to ask more nuanced questions that might have previously taken multiple searches, using Gemini's advanced reasoning, thinking, and multimodal capabilities. + +This drive to make information more accessible and helpful has led us over the years to improve the discovery and creation of digital content both on the web and through platforms like Google Play and YouTube. People are consuming many forms of digital content, including watching long and short form videos and podcasts, streaming TV, playing games, listening to music, reading books, and using apps. Working with content creators and partners, we continue to build new ways for people around the world to create and find great digital content. + +Fueling all of these great digital experiences are extraordinary platforms and devices. That is why we continue to invest in platforms like our Android mobile operating system, Chrome browser, and Chrome operating system, as well as our family of devices. We see tremendous potential for devices to be helpful and make people's lives easier by combining the best of our AI, software, and hardware. This potential is reflected in our latest generation of devices, such as the new Pixel 10 series and the Pixel Watch 4. Creating products and services that people rely on every day is a journey that we are investing in for the long-term. + +## How We Make Money + +We have built world-class advertising technologies for advertisers, agencies, and publishers to power their digital marketing businesses. Our advertising solutions help millions of companies grow their businesses through our wide range of products across devices and formats, and we aim to ensure positive user experiences by serving the right ads at the right time and by building deep partnerships with brands and agencies. AI has been foundational to our advertising business for more than a decade. Products like Demand Gen, Performance Max, and Product Studio use the full power of our AI to help advertisers find untapped and incremental conversion opportunities. + +Google Services generates revenues primarily by delivering both performance and brand advertising that appears on Google Search & other properties, YouTube, and Google Network partners' properties ("Google Network properties"). We continue to invest in both performance and brand advertising and seek to improve the measurability of advertising so advertisers understand the effectiveness of their campaigns. + +- Performance advertising creates and delivers relevant ads that users will click on leading to direct engagement with advertisers. Performance advertising lets our advertisers connect with users while driving measurable results. Our ads tools allow performance advertisers to create simple text-based ads. +- Brand advertising helps enhance users' awareness of and affinity for advertisers' products and services, through videos, text, images, and other interactive ads that run across various devices. We help brand advertisers deliver digital videos and other types of ads to specific audiences for their brand-building marketing campaigns. + +We have allocated substantial resources to stopping bad advertising practices and protecting users on the web. We focus on creating the best advertising experiences for our users and advertisers in many ways, including filtering out invalid traffic, removing billions of bad ads from our systems every year, and closely monitoring the sites, apps, and videos where ads appear and blocklisting them when necessary to ensure that ads do not fund bad content. + +In addition, Google Services generates revenues from products and services beyond advertising, including: + +- consumer subscriptions, which primarily include revenues from YouTube services, such as YouTube TV, YouTube Music and Premium, and NFL Sunday Ticket, as well as Google One, which offers access to our most capable Gemini models; +- platforms, which primarily include revenues from Google Play sales of apps and in-app purchases; and +• devices, which primarily include sales of the Pixel family of devices. + +## Google Cloud + +Through our Google Cloud Platform and Google Workspace offerings, Google Cloud generates revenues primarily from consumption-based fees and subscriptions for infrastructure, platform, applications, and other cloud services. Customers use Google Cloud in multiple ways such as: + +- AI-optimized Infrastructure: runs on our Cloud, at the edge, or in customers' data centers. It can be used to migrate and modernize information technology (IT) systems and to train and serve various types of AI models. Our AI infrastructure delivers cost-performance for AI workloads. We offer a range of AI accelerators, including our custom TPUs and specialized GPUs, as well as AI-optimized storage offerings, and efficient AI software. +- Developer Platform: delivers a fully managed AI development platform, through Vertex AI, for accessing, tuning, augmenting, and deploying custom models and agents, helping customers build applications with more than 200 foundation models, including our Gemini family, third-party, and open models. +- Cybersecurity: provides AI-powered threat intelligence and cybersecurity solutions to help customers detect, analyze, protect against, and respond to a broad range of cybersecurity threats. +- Data and Analytics: enables customers to migrate, clean, prepare, and feed data into their models. Our data platform also unifies data lakes, data warehouses, data governance, and advanced machine learning into a single platform that helps users analyze data using AI models across any cloud. +Gemini Enterprise: empowers teams to discover, create, share, and run AI agents all in one secure platform, bringing the best of Google AI to employees through an intuitive chat interface, helping to automate workflows and drive smarter business outcomes. +Gemini for Google Workspace: brings our AI-powered agents into Gmail, Docs, Sheets, and more to help users write, organize, visualize, accelerate workflows, and have more productive meetings. + +## - Agents: + +## Other Bets + +Across Alphabet, we are also using technology to try to solve big problems that affect a wide variety of industries, including transportation and health technology. Alphabet's investment in the portfolio of Other Bets includes businesses that are at various stages of development, ranging from those in the research and development phase, such as X, our moonshot factory focused on developing breakthrough technologies, to those that are scaling commercialization, such as Waymo, which is expanding to more cities domestically, entering international markets, and further scaling operations. + +Other Bets operate as independent companies and some of them have their own boards with independent members and outside investors. While these early-stage businesses naturally come with considerable uncertainty, some of them are already generating revenue and making important strides in their industries. Revenues from Other Bets are generated primarily from the sale of autonomous transportation and internet services. + +## Competition + +Our business is characterized by rapid change as well as new and disruptive technologies. We face formidable competition in every aspect of our business, including but not limited to, from: + +- general purpose search engines and information services; +• vertical search engines and e-commerce providers for queries on topics such as those related to travel, jobs, and health, which users may navigate directly to rather than go through Google; +• online advertising platforms and networks, including online shopping and streaming services; +- other forms of advertising, such as billboards, magazines, newspapers, radio, and television, as our advertisers typically advertise in multiple media, both online and offline; +• digital content and application platform providers; +- providers of enterprise cloud services; +• AI model developers and providers of AI products and services; +- companies that design, manufacture, and market consumer hardware products, including businesses that have developed proprietary platforms; +• providers of digital video services; +- social networks, which users may rely on for product or service referrals, rather than seeking information through traditional search engines; and +- providers of workspace communication and connectivity products. + +Competing successfully depends heavily on our ability to continually develop and distribute innovative products and technologies to the marketplace across our businesses. For example, for advertising, competing successfully depends on attracting and retaining: + +- users, for whom other products and services are literally one click away, on the basis of the relevance of our advertising, as well as the general usefulness, security, and availability of our products and services; +- advertisers, primarily based on our ability to generate sales leads, and ultimately customers, and to deliver their advertisements in an efficient and effective manner across a variety of distribution channels even as trends in advertising mediums and user preferences change; and +- content providers, primarily based on the quality of our advertiser base, our ability to help these partners generate revenues from advertising, and the terms of our agreements with them. + +For additional information about competition, see Item 1A Risk Factors of this Annual Report on Form 10-K. + +## Culture and Workforce + +Our people are critical for our continued success, so we work hard to create an environment where employees can have fulfilling careers and perform at a high level. We offer industry-leading benefits and programs to take care of the diverse needs of our employees and their families, including opportunities for career growth and development, resources to support their financial health, and access to excellent healthcare choices. Our competitive compensation programs help us to attract and retain key talent, and we will continue to invest in recruiting talented people to technical and non-technical roles and rewarding them well. We provide a variety of high-quality training and support to managers to build and strengthen their capabilities — ranging from courses for new managers, to learning resources that help them provide feedback and manage performance, to coaching and individual support. + +As of December 31, 2025, Alphabet had 190,820 employees. We have work councils and statutory employee representation obligations in certain countries, and we are committed to supporting protected labor rights, maintaining an open culture, and listening to our employees. + +When appropriate we partner with outside companies on a contractual basis to provide a specialized service or to temporarily cover a short-term need. The employees of our suppliers and staffing partners — vendors and temporary staff, respectively — and independent contractors who are self-employed, make up our extended workforce. We choose our partners and staffing agencies carefully, and review their compliance with Google's Supplier Code of Conduct. + +## Government Regulation + +We are subject to numerous United States (US) federal, state, and local, as well as foreign laws, and regulations covering a wide variety of subjects, and the scope of this coverage continues to broaden with continuing new legal and regulatory developments in the US and internationally. Like other companies in the technology industry, we face increasingly heightened scrutiny from both US and foreign governments with respect to our compliance with laws and regulations. Many of these laws and regulations are evolving and their applicability and scope, as interpreted by the courts, remain uncertain. Particularly with regard to AI; competition; consumer protection; content moderation, including access restrictions for minors; data privacy and security; intellectual property; news publications; and sustainability and other social matters, we have seen an increase in new and evolving laws and regulations, as well as related enforcement actions and investigations, being proposed and implemented in recent years by legislative and regulatory bodies around the world. As we have seen in recent years, different laws and regulations on the same topic may not always have the same requirements (and sometimes may seem to have conflicting requirements), and even when requirements overlap, the rules are not always consistently implemented, interpreted, and enforced from jurisdiction to jurisdiction. + +Our compliance with these laws and regulations may be onerous and could, individually or in the aggregate, increase our cost of doing business, make our products and services less useful, limit our ability to pursue certain business practices or offer certain products and services (either in certain geographies or at all), cause us to change our business models and operations, affect our competitive position relative to our peers, or otherwise harm our business, reputation, financial condition, and operating results. + +For additional information about government regulation applicable to our business, see Item 1A Risk Factors; Trends in Our Business and Financial Effect in Part II, Item 7; and Legal Matters in Note 10 of the Notes to Consolidated Financial Statements included in Part II, Item 8 of this Annual Report on Form 10-K. + +## Intellectual Property + +We rely on intellectual property such as trademarks, copyrights, patents, and trade secrets, as well as confidentiality procedures and contractual provisions, to protect our proprietary technology and our brand. We have registered, and applied for the registration of, US and international trademarks, service marks, domain names, and copyrights. We have also filed patent applications in the US and foreign countries covering certain of our technology, and acquired patent assets to supplement our portfolio. We have licensed in the past, and expect that we may license in the future, certain of our rights to other parties. For additional information, see Item 1A Risk Factors of this Annual Report on Form 10-K. + +## Available Information + +Our website is located at www.abc.xyz, and our investor relations website is located at www.abc.xyz/investor. Access to our Annual Reports on Form 10-K, Quarterly Reports on Form 10-Q, Current Reports on Form 8-K, and our Proxy Statements, and any amendments to these reports, is available on our investor relations website, free of charge, after we file or furnish them with the SEC and they are available on the SEC's website at www.sec.gov. + +We webcast our earnings calls, as well as certain events we participate in or host with members of the investment community, via our investor relations YouTube channel and website. Our investor relations website also provides notifications of news or announcements regarding our financial performance and other items that may be material or of interest to our investors, including SEC filings, investor events, press and earnings releases, and blogs. We also share Google news and product updates on Google's Keyword blog at https://www.blog.google/ and News From Google page on X at x.com/NewsFromGoogle, and our executive officers may also use certain social media channels, such as X and LinkedIn, to communicate information about earnings results and company updates, which may be of interest or material to our investors. Further, corporate governance information, including our certificate of incorporation, bylaws, corporate governance guidelines, board committee charters, and code of conduct, is also available on our investor relations website under the heading "Governance." The information contained on, or that may be accessed through our websites or our executive officers' social media channels, is not incorporated by reference into this Annual Report on Form 10-K or in any other report or document we file with the SEC, and any references to our websites are intended to be inactive textual references only. + +## ITEM 1A. RISK FACTORS + +Our operations and financial results are subject to various risks and uncertainties, including but not limited to those described below, which could harm our business, reputation, financial condition, and operating results, and may affect the trading price and price volatility of our Class A and Class C stock. + +## Risks Specific to our Company + +We generate a significant portion of our revenues from advertising. Reduced spending by advertisers, a loss of partners, shifts in online advertising, new and evolving advertising formats, or new or existing technologies that block ads online or affect our ability to personalize ads could harm our business. + +We generated more than 70% of total revenues from online advertising in 2025. Many of our advertisers, companies that distribute our products and services, digital publishers, and content providers can terminate their contracts with us at any time. These partners may not continue to do business with us if we do not create more value (such as increased numbers of users or customers, new sales leads, increased brand awareness, or more effective monetization) than their available alternatives. + +We believe AI is quickly reshaping the advertising industry, including how ads are delivered online, and we and our competitors are constantly adjusting to meet this shift and provide new and evolving advertising formats. There is no assurance that we will adapt effectively and competitively to meet this shift, and that such advertising formats, strategies, and offerings will be successful. + +Changes to our advertising policies and data privacy practices, as well as changes to other companies' advertising or data privacy practices have in the past, and may in the future, affect the advertising services that we are able to provide. In addition, technologies have been developed that make personalized ads more difficult, or that block the display of ads altogether, and some providers of online services have integrated technologies that could impair the availability and functionality of third-party digital advertising. Failing to provide superior value or deliver advertisements effectively and competitively could harm our business, reputation, financial condition, and operating results. + +Expenditures by advertisers tend to correlate with overall economic conditions. Adverse macroeconomic conditions have affected, and may in the future affect, the demand for advertising, resulting in fluctuations in the amounts our advertisers spend on advertising, which could harm our financial condition and operating results. + +Our increasing investment in new businesses, products, services, and technologies is inherently risky, and could divert management attention and harm our business, financial condition, and operating results. + +We have invested and expect to expand our investment in new businesses, products, services, and technologies in a wide range of industries beyond online advertising. The investments that we are making across our businesses — such as building AI-optimized infrastructure, including our custom TPUs, and integrating AI capabilities into new and existing products and services — reflect our ongoing efforts to innovate and provide products and services that are helpful to users, advertisers, publishers, customers, content providers, and distribution partners. Our investments ultimately may not be commercially viable or may not result in an adequate return of capital and, in pursuing new strategies, we may incur unanticipated liabilities. + +To meet the compute capacity demands of AI training and inference, as well as traditional cloud computing services, we are entering into significant leasing arrangements with third party operators, which may increase costs and operational complexity. We also have a number of large, long-duration commercial agreements, which could increase our liabilities and obligations in the event of nonperformance by us, our counterparties, or vendors. In such nonperformance or an industry downturn, we may incur additional liabilities, have excess capacity that we cannot easily redeploy, and not receive payments from our counterparties or customers. + +We have invested and expect to significantly expand our investment in property and equipment, including our technical infrastructure, and we expect these assets to benefit our business over their estimated useful lives. Changes in facts and circumstances such as changes to historical asset performance, expected technology advancements, and future network deployment plans could change the period over which we expect to benefit from the asset and impact our financial condition and operating results. + +Innovations in our products and services could also result in changes to user and customer behavior and affect our revenue trends. These endeavors involve significant risks and uncertainties, including diversion of resources and management attention from current operations, different monetization models, and the use of alternative investment, governance, or compensation structures that may fail to adequately align incentives across the company or otherwise accomplish their objectives. + +Within Google Services, we continue to invest heavily in devices, including our smartphones, home devices, and wearables, which is a highly competitive market with frequent introduction of new products and services, rapid adoption of technological advancements by competitors, increased market saturation in developed countries, short product life cycles, evolving industry standards, continual improvement in performance characteristics, and price and feature sensitivity on the part of consumers and businesses. There can be no assurance we will be able to provide devices that compete effectively. + +Within Google Cloud, we devote significant resources to develop and deploy our enterprise-ready cloud services, including Google Cloud Platform and Google Workspace, and we are advancing our AI platforms and models to support these tools and technologies, including the development of our custom TPUs and how we deliver them to our customers. We are incurring significant and increasing costs and new liabilities, including contingent liabilities, to build and maintain infrastructure to support cloud computing services, invest in cybersecurity, and hire talent. Meanwhile, our competitors are rapidly developing and deploying cloud-based services and capacity. Pricing and delivery models, which are subject to increasing regulatory scrutiny and requirements, are competitive and constantly evolving, and we may therefore not achieve our business objectives. Further, our business with financial services, healthcare, and public sector customers may present additional risks, including regulatory compliance risks. For instance, we may be subject to government audits and cost reviews, and any failure to comply or any deficiencies found may expose us to legal, financial, and reputational risks. Evolving laws and regulations may require us to make new capital investments, build new products, and seek partners to deliver localized services in other countries, and we may not be able to meet sovereign operating requirements. + +Within Other Bets, we are investing significantly in areas such as life sciences and transportation, among others. These investment areas face intense competition from large, experienced, and well-funded competitors, and our offerings, many of which involve the development of new and emerging technologies, may not be successful, or be able to compete effectively or operate at sufficient levels of profitability. + +In addition, new and evolving products and services, including those that use AI, raise ethical, technological, legal, regulatory, and other challenges, which could harm our brands and demand for our products and services. Because all of these investment areas are inherently risky, no assurance can be given that such strategies and offerings will be successful or will not harm our reputation, financial condition, and operating results. + +We face intense competition. If we do not continue to innovate and provide products and services that are useful to users, customers, and other partners, we may not remain competitive, which could harm our business, financial condition, and operating results. + +Our business environment is rapidly evolving and intensely competitive. Our businesses face changing technologies, shifting user needs, and frequent introductions of rival products and services. To compete successfully, we must accurately anticipate technology developments and deliver innovative, relevant, and useful products, services, and technologies in a timely manner. For example, with the rise of AI in recent years, we have increasingly focused our investments in building powerful AI tools and AI enhancements to our existing products and services to better cater to our users, customers, and other partners. As our businesses evolve, the competitive pressure to innovate will encompass a wider range of products and services. We must continue to invest significant resources in technical infrastructure, including the development of our custom TPUs, and research and development, including through acquisitions, in order to enhance our technology, products, and services. + +We have many competitors in different industries. Our current and potential domestic and international competitors range from large and established companies to emerging start-ups. Some competitors have longer operating histories and well-established relationships in various sectors. They can use their experience and resources in ways that could affect our competitive position, including by making acquisitions and entering into other strategic arrangements; continuing to invest heavily in technical infrastructure, research and development, and in talent; initiating intellectual property and competition claims (whether or not meritorious); and continuing to compete for users, advertisers, customers, and content providers. Further, discrepancies in enforcement of existing laws may enable our lesser known competitors to aggressively interpret those laws without commensurate scrutiny, thereby affording them competitive advantages. Our competitors may also be able to innovate and provide products and services faster or more cost effectively than we can or may foresee the need for products and services before we do. + +We are developing frontier generative AI models and building AI capabilities into products and services across the company, tailoring them to the evolving preferences of our users and customers. AI technology and services are highly competitive, rapidly evolving, and require significant investment, including technical infrastructure, development, and operational costs. Our ability to deploy certain AI technologies critical for our products and services and for our business strategy may depend on the availability and pricing of third-party equipment and other technical infrastructure operations costs, including network capacity, energy, and equipment costs. Additionally, other companies may develop + +AI products and technologies that are similar or superior to our technologies or more cost-effective to develop or deploy. Other companies may also have (or in the future may obtain) patents or other proprietary rights that would prevent, limit, or interfere with our ability to make, use, or sell our own AI products and services. + +Our financial condition and operating results may also suffer if our products and services are not responsive in a timely manner to the evolving needs and desires of our users, advertisers, publishers, customers, and content providers, or if we miscalculate those needs and desires and invest significantly in areas that fail to gain sufficient market traction. As new and existing technologies continue to develop, competitors and new entrants may be able to offer experiences that are, or that are perceived to be, substantially similar to or better than ours, and the consumers may change how they obtain information online, potentially reducing the utility of our existing products and services. These technologies could reduce usage of our products and services, and force us to compete in different ways and expend significant resources to develop and operate equal or better products and services. Competitors' success in providing compelling products and services or in attracting and retaining users, advertisers, publishers, customers, and content providers could harm our financial condition and operating results. + +Our revenue growth rate could decline over time, and we may experience downward pressure on our operating margin in the future. + +Our revenue growth rate could decline over time as a result of a number of factors, including changes in customer usage and demand for our existing products and increasing demand for competing technologies; changes in the devices and modalities used to access our products and services; changes in geographic mix; deceleration or declines in advertiser spending; competition; decreases in the pricing of our products and services; ongoing product and policy changes; and shifts to lower priced products and services. + +We may experience downward pressure on our operating margin resulting from a variety of factors. These include increasing costs for many aspects of our business resulting from a higher level of investment in technical infrastructure, increasing regulations, and increasing competition. Certain of our costs and expenses are relatively fixed and may not correlate to changes in revenue, and we may also not be able to continue to drive efficiencies at the rate we have in the past. We may also face margin compression from an increase in the mix of lower-margin products and services, in particular from the continued expansion of our business into new fields, including products and services such as Google Cloud, our devices, and consumer subscription products, as well as significant investments in Other Bets. For instance, margins on our devices have had, and may continue to have, an adverse effect on our consolidated margins due to pricing pressures and higher cost of sales. Due to these factors and the evolving nature of our business, our historical revenue growth rate and historical operating margin may not be indicative of our future performance. For additional information, see Trends in Our Business and Financial Effect and Revenues and Monetization Metrics in Part II, Item 7 of this Annual Report on Form 10-K. + +Our intellectual property rights are valuable, and any inability to protect them could reduce the value of our products, services, and brands as well as affect our ability to compete. + +Our patents, trademarks, trade secrets, copyrights, and other intellectual property rights are important assets for us. Various events outside of our control pose a threat to our intellectual property rights, as well as to our products, services, and technologies. For example, effective intellectual property protection may not be available in every country in which our products and services are distributed or made available through the Internet. Also, the efforts we have taken and may take in the future to protect our proprietary rights, including obtaining copyright and patent protections for our important innovations, including AI innovations, may not be sufficient or effective. There is also the possibility that an issued patent may be deemed invalid or unenforceable. + +We also seek to maintain certain intellectual property as trade secrets. The confidentiality of such trade secrets and other sensitive information could be compromised, which could cause us to lose the competitive advantage resulting from these trade secrets. We also face risks associated with our trademarks. For example, there is a risk that the word "Google" could become so commonly used that it becomes synonymous with the word "search." Some courts have ruled that "Google" is a protectable trademark, but it is possible that other courts, particularly those outside of the US, may reach a different determination. If this happens, we could lose protection for this trademark. + +Any significant impairment of our intellectual property rights could harm our business and ability to compete. Protecting our intellectual property rights is costly and time consuming; any increase in unauthorized use could make it more expensive to do business and harm our financial condition and operating results. + +Our business depends on strong brands, and failing to maintain and enhance our brands would hurt our ability to expand our base of users, advertisers, customers, content providers, and other partners. + +Our strong brands have significantly contributed to the success of our business. Maintaining and enhancing the brands within Google Services, Google Cloud, and Other Bets increases our ability to enter new categories and launch new and innovative products and services that better serve the needs of our users, advertisers, customers, content providers, and other partners. Our brands have been, and may in the future be, harmed by a number of factors, including, among others, reputational issues, third-party content shared on our platforms, data privacy and security issues and developments, issues in delivering age-appropriate experiences to minors, and product or technical performance failures. For example, if we fail to respond appropriately to the sharing of misinformation or objectionable content on our services or products or objectionable practices by advertisers, or otherwise to adequately address user concerns, our users may lose confidence in our brands. Furthermore, failure to maintain and enhance our brands could harm our business, reputation, financial condition, and operating results. Our success will depend largely on our ability to remain a technology leader and continue to provide high-quality, trustworthy, innovative products and services that are truly useful and play a valuable role in a range of settings. + +We face a number of manufacturing and supply chain risks that could affect our ability to supply our products and services and harm our business, financial condition, and operating results. + +We rely on contract manufacturers to manufacture or assemble our devices as well as servers and networking equipment used in our technical infrastructure, certain components of which we may supply. We also rely on third parties to supply components and distribute our products and services. Our business could be harmed if we are not able to engage these companies with the necessary capabilities or capacity on reasonable terms, or if those we engage fail to meet their obligations (whether due to financial difficulties or other reasons), or make adverse changes in the pricing or other material terms of our arrangements with them. + +We have experienced and may in the future experience supply shortages, price increases, quality issues, or longer lead times that could harm our operations, driven by raw material or component availability, manufacturing capacity, labor shortages, industry allocations, logistics capacity, inflation, foreign currency exchange rates, tariffs, sanctions and export controls, trade disputes and barriers, forced labor concerns, sourcing requirements, geopolitical tensions, armed conflicts, natural disasters or pandemics, the effects of climate change, power and transmission availability, and significant changes in the financial or business condition of our suppliers. Some of the components we use in our technical infrastructure and our devices are available from only one or limited sources, and we may not be able to find replacement vendors on favorable terms in the event of a supply chain disruption. A significant supply interruption that affects us or our vendors could delay critical data center or network infrastructure upgrades or expansions and delay consumer product availability. + +Our ability to scale our technical infrastructure is increasingly constrained by the availability of power, water, and land. For example, energy supply is constrained globally due to the significant increase in demand for and limited availability of energy to power AI compute. Securing this capacity involves entering into complex, long-lead-time arrangements. Additionally, manufacturing and supply of servers and network equipment for our technical infrastructure, particularly for specialized AI chips, is limited to a small number of qualified suppliers. Extended or unforeseen disruptions at these suppliers could impact our ability to meet customer demand. Failure to secure sufficient capacity in a timely manner would limit our ability to train models and serve Cloud customers. + +We may enter into long-term contracts for materials and products that commit us to significant terms and conditions. We may face costs for materials and products that are not consumed due to market demand, technological change, excess or obsolete inventory, changed consumer preferences, quality, product recalls, and warranty issues. Certain of our competitors may negotiate more favorable contractual terms based on volume and other commitments that may provide them with competitive advantages and may affect our supply. For example, industry supply capacity for AI accelerators, including GPUs as well as our custom-built TPUs, is highly competitive and rapidly evolving. If we are unable to negotiate favorable contractual terms or our competitors claim the supply or capacity first, we may face increased costs and supply constraints, which could harm our business, financial condition, and operating results. + +Our devices have had, and in the future may have, quality issues resulting from design, manufacturing, or operations. Sometimes, these issues may be caused by components we purchase from other manufacturers or suppliers. In addition, quality issues with equipment used in our technical infrastructure could constrain our capacity to support the delivery and continued development of our products and services. If the quality of our products and services does not meet expectations, we lack the capacity to deliver them, or our products or services are defective or require a corrective action or recall, it could harm our business, reputation, financial condition, and operating results. + +We require our suppliers and business partners to comply with laws and, where applicable, our company policies and practices, such as the Google Supplier Code of Conduct, regarding workplace and employment practices, data security, environmental compliance, and intellectual property licensing, but we do not control them or their practices. Violations of law or unethical business practices could result in supply chain disruptions, canceled orders, harm to key relationships, and damage to our reputation. Their failure to procure necessary license rights to intellectual property could affect our ability to sell our products or services and expose us to litigation or financial claims. + +Interruption to, interference with, or failure of our complex information technology and communications systems could hurt our ability to effectively provide our products and services, which could harm our reputation, financial condition, and operating results. + +The availability of our products and services and fulfillment of our customer contracts depend on the continuing operation of our information technology and communications systems. Our systems are vulnerable to damage, interference, or interruption from modifications or upgrades, terrorist attacks, state-sponsored attacks, natural disasters or pandemics, geopolitical tensions or armed conflicts, export controls and sanctions, tariffs and non-tariff trade barriers, the effects of climate change, power and transmission availability challenges, utility outages, telecommunications failures, computer viruses, software bugs, cyber attacks, supply-chain attacks, computer denial of service attacks, phishing schemes, or other attempts to harm or access our systems. Some of our data centers are located in areas with a high risk of major earthquakes or other natural disasters. Our data centers are also subject to break-ins, sabotage, and intentional acts of vandalism, and, in some cases, to potential disruptions resulting from problems experienced by facility operators or disruptions as a result of geopolitical tensions and conflicts happening in the area. Some of our systems are not fully redundant, and disaster recovery planning cannot account for all eventualities. The occurrence of a natural disaster or pandemic, closure of a facility, or other unanticipated problems affecting our data centers could result in lengthy interruptions in our service. In addition, our products and services are highly technical and complex and have contained in the past, and may contain in the future, errors or vulnerabilities, which could result in interruptions in or failure of our services or systems. Any of these incidents could impede or prevent us from effectively offering products and providing services, which could harm our business, reputation, financial condition, and operating results. + +Our international operations expose us to additional risks that could harm our business, reputation, financial condition, and operating results. + +Our international operations are significant to our revenues and net income, and we plan to continue growing internationally. International revenues accounted for approximately 52% of consolidated revenues in 2025. In addition to risks described elsewhere in this section, our international operations expose us to other risks, including: + +- restrictions on foreign ownership and investments, and stringent foreign exchange controls that might prevent us from repatriating cash earned in countries outside the US; +- sanctions, tariffs, import and export controls, other market access barriers, political unrest, geopolitical tensions, changes in regimes, or armed conflict (such as ongoing conflicts in the Middle East and Ukraine), any of which may affect our business continuity, increase our operating costs, limit demand for our products and services, limit our ability to source components or final products, or prevent or impede us from operating in certain jurisdictions, complying with local laws, or offering products or services; +- longer payment cycles in some countries, increased credit risk, and higher levels of payment fraud; +- an evolving foreign policy landscape that could harm our revenues and could subject us to litigation, new regulatory costs and challenges (including new customer requirements), uncertainty regarding regulatory outcomes, and other liabilities under local laws that may not offer due process or clear legal precedent; +- anti-corruption laws, such as the US Foreign Corrupt Practices Act, and other local laws prohibiting certain payments to government officials, violations of which could result in civil and criminal penalties; and +- different employee/employer relationships; different labor laws, regulations, and labor practices; and other challenges caused by distance, language, local expertise, and cultural differences, increasing the complexity of doing business in multiple jurisdictions. + +Because we conduct business in currencies other than US dollars but report our financial results in US dollars, we have faced, and will continue to face, exposure to fluctuations in foreign currency exchange rates. Although we hedge a portion of our international currency exposure, significant fluctuations in exchange rates between the US dollar and foreign currencies have and may in the future adversely affect our revenues and earnings. Hedging programs are also inherently risky and could expose us to additional risks that could harm our financial condition and operating results. + +Disruptions in our ability to access future financing or manage our indebtedness could adversely affect our ability to execute our strategy and harm our financial condition. + +We may from time to time access capital markets for debt or seek to enter into other forms of financing, such as leases. Any difficulty in accessing capital markets, entering into other forms of financing on favorable terms, or managing our existing indebtedness could increase our costs of financing and restrict our ability to invest in our business. Furthermore, the combination of our current and any future indebtedness, including obligations arising under leases, backstops, guarantees, and potential liabilities from large commercial agreements, could harm our financial condition and reduce our financial and business flexibility. + +## We are exposed to fluctuations in the fair values of our investments and, in some instances, our financial statements incorporate inherently subjective valuation methodologies. + +The fair value of our debt and equity investments may in the future be, and certain investments have been in the past, negatively affected by liquidity, credit deterioration or losses, performance and financial results of the underlying entities, foreign exchange rates, changes in interest rates, the effect of new or changing regulations, the stock market in general, or other factors. As a result of these factors, the value of our investments could significantly decline, which could harm our financial condition and operating results. + +We measure certain of our non-marketable equity and debt securities, and certain other instruments including stock-based compensation (SBC) awards of Other Bet companies, at fair value on a nonrecurring basis, which is inherently subjective and requires management judgment and estimation. Gains and losses on non-marketable equity securities are recognized in other income (expense), net (OI&E), which increases the volatility of our OI&E. The unrealized gains and losses or impairments we record from fair value remeasurements in any particular period may differ significantly from the gains and losses we ultimately realize on such investments. Changes in fair value on SBC awards are recognized primarily through operating expenses. + +## Risks Related to our Industry + +Issues in the development and use of AI may result in reputational harm and increased liability exposure. + +Our evolving AI-enabled products and services may give rise to risks related to harmful content, inaccuracies, discrimination, intellectual property infringement or misappropriation, violation of rights of publicity, defamation, data privacy, cybersecurity, minor protection, and other issues. As a result of these and other challenges associated with innovative technologies, our implementation of AI systems could subject us to competitive harm, regulatory action, legal liability (including under new and proposed legislation and regulations), new applications of existing data protection, privacy, intellectual property, and other laws, and brand or reputational harm. + +Some uses of AI will present ethical issues and may have broad effects on society. In order to implement AI responsibly and minimize unintended harmful effects, we have already devoted and will continue to invest significant resources to develop, test, and maintain our products and services, but we may not be able to identify or resolve all AI-related issues, deficiencies, and failures before they arise. Unintended consequences, uses, or customization of our AI tools and systems may negatively affect human rights, privacy, employment, or other social concerns, which may result in claims, lawsuits, brand or reputational harm, and increased regulatory scrutiny, any of which could harm our business, financial condition, and operating results. + +People access our products and services through a variety of platforms and devices that continue to evolve with the advancement of technology and user preferences. If manufacturers and users do not widely adopt versions of our products and services developed for these interfaces, our business could be harmed. + +While the modalities used to access information is evolving, people access our products and services through a growing variety of devices such as phones, laptops and tablets, video game consoles, voice-activated speakers, wearables (including virtual reality and augmented reality devices), automobiles, and television-streaming devices. Our products and services may be less popular on some interfaces. Each manufacturer or distributor may establish unique technical standards for its devices, and our products and services may not be available or may only be available with limited functionality for our users or our advertisers on these devices as a result. Some manufacturers may also elect not to include our products on their devices. + +It is hard to predict the challenges we may encounter in adapting our products and services and developing competitive new products and services. We expect to continue to devote significant resources to creating and supporting products and services across multiple platforms and devices. Failing to attract and retain a substantial number of device manufacturers, suppliers, distributors, developers, and users, or failing to develop products and technologies that work well on new devices and platforms, could harm our business, financial condition, and operating results and ability to capture future business opportunities. + +Problematic content on our platforms, including low-quality user-generated content, web spam, content farms, and other violations of our guidelines could affect the quality of our services, which could harm our reputation and deter our current and potential users from using our products and services. + +We, like others in the industry, face violations of our content guidelines across our platforms, including sophisticated attempts by bad actors to manipulate our hosting and advertising systems to fraudulently generate revenues, or to otherwise generate traffic that does not represent genuine user interest or intent. While we invest significantly in efforts to promote high-quality and relevant results and to detect and prevent low-quality content and invalid traffic, we have been unable and may continue to be unable to detect and prevent all such abuses or promote uniformly high-quality content. Increased use of AI in our offerings and internal systems may create new instances of problematic content and increased potential for misuse and abuse. + +Many websites violate or attempt to violate our guidelines, including by seeking to inappropriately rank higher in search results than our search engine's assessment of their relevance and utility would rank them. Such efforts have affected, and may continue to affect, the quality of content on our platforms and lead them to display false, misleading, or undesirable content. Although English-language web spam in our search results has been reduced, and web spam in most other languages is limited, we expect web spammers will continue to seek inappropriate ways to improve their rankings. Although we continue to invest in and deploy proprietary technology to detect and prevent web spam on our platforms and to evolve our policies to address emerging threats, there is no guarantee that our technology and policy enforcement will always be successful, and our users may have negative experiences that make them less likely to use our platforms. We face legal and regulatory challenges to our efforts to address low-quality content, and our ability to address it may be constrained or made more costly through added compliance requirements. We also face other challenges to the quality of our search results from low-quality and irrelevant content websites, including content farms, which are websites that generate large quantities of low-quality content in an effort to improve their search rankings. We are continually launching algorithmic changes designed to detect and prevent these efforts, but we may not always be successful. We also face other challenges on our platforms, including attempted election interference; fraud, content inappropriate for minors, misleading or deceptive information, and other types of potentially harmful content. + +If we fail to either detect and prevent an increase in problematic content or effectively promote high-quality content, it could hurt our reputation for delivering relevant information or reduce use of our platforms, harming our financial condition and operating results. It may also subject us to litigation and regulatory actions, which could result in monetary penalties and damages and divert management's time and attention. + +Data privacy and security concerns relating to our technology and our practices could harm our reputation, cause us to incur significant liability, and deter current and potential users or customers from using our products and services. Computer viruses, software bugs or defects, security breaches, and attacks on our systems could result in the improper disclosure and use of user data and interference with our users' and customers' ability to use our products and services, harming our business and reputation. + +Concerns about, including the adequacy of, our practices with regard to the collection, use, governance, disclosure, or security of personal data or other data-privacy-related matters, even if unfounded, could harm our business, reputation, financial condition, and operating results. Our policies and practices may change over time as expectations and regulations regarding privacy and data change. + +Our products and services involve the storage, handling, and transmission of proprietary and other sensitive information. Malicious software such as viruses, software bugs, theft, misuse, defects, vulnerabilities in our products and services, as well as cyber attacks, phishing schemes, and other types of security attacks, which increasingly use AI, expose us to a risk of loss or improper use and disclosure of such information, which could result in litigation and other potential liabilities, including regulatory fines and penalties, as well as reputational harm. Additionally, our products incorporate highly technical and complex technologies, and thus our technologies and software have contained, and are likely in the future to contain, undetected errors, bugs, or vulnerabilities. We continue to add new features involving AI to our offerings and internal systems, and features that rely on AI may be susceptible to unanticipated security threats as our and the market's understanding of AI-centric security risks and protection methods continue to develop. We have in the past discovered, and may in the future discover, some errors in our software code only after we have released the code. Systems and control failures, security breaches, failure to comply with our privacy policies, and inadvertent disclosure of user data could result in regulatory and legal exposure, seriously harm our reputation, brand, and business, and impair our ability to attract and retain users or customers. Such incidents have occurred in the past and may continue to occur due to the scale and nature of our products and services. While there is no guarantee that such incidents will not cause significant damage, we expect to continue to expend significant resources to maintain security protections that limit the effect of bugs, theft, misuse, and security vulnerabilities or breaches. + +We experience cyber attacks and other attempts to gain unauthorized access to our systems on a regular basis. Cyber attacks continue to evolve in sophistication and volume, and inherently may be difficult to detect for long periods of time. The development and implementation of AI technologies may further increase our exposure to or exacerbate the risks of cyber attacks or other security incidents, particularly where such technologies are exploited by third parties to breach our or other parties' systems, including when such technologies are used to target our employees or impersonate members of senior management in order to gain unauthorized access to our systems. We have also seen, and will continue to see, industry-wide software supply chain vulnerabilities and attacks on telecommunications and other critical infrastructure, which could affect our or other parties' systems. We expect to continue to experience such incidents or vulnerabilities in the future. Our efforts to prevent security incidents and address undesirable activity on our platform may require us to spend additional resources to prepare and defend against such threats, and could also increase the risk of retaliatory attack. In addition, we face the risk of cyber attacks and data exfiltration or compromise by nation-states and state-sponsored actors. These attacks may target us or our customers, particularly our public sector customers (including federal, state, and local governments). Geopolitical tensions or armed conflicts, such as the ongoing conflict in the Middle East and Ukraine, may increase these risks. + +We may experience security and privacy issues, whether due to employee or insider error or malfeasance, system errors, or vulnerabilities in our or other parties' systems. While we may not determine some of these issues to be material when they occur and may remedy them quickly, there is no guarantee that these issues will not ultimately result in significant legal, financial, and reputational harm, including government inquiries, enforcement actions, litigation, and negative publicity. There is also no guarantee that a series of related issues may not be determined to be material at a later date in the aggregate, even if they may not be material individually at the time of their occurrence. Because the techniques used to obtain unauthorized access to, disable, or degrade service provided by or otherwise sabotage systems change frequently and often are recognized only after being launched against a target, even taking all reasonable precautions, including those required by law, we have been unable in the past and may continue to be unable to anticipate or detect certain attacks or vulnerabilities or implement adequate preventative measures. + +Further, if any partners with whom we share user or other customer information fail to implement adequate data-security practices, fail to comply with our terms and policies, or otherwise suffer a network or other security breach, our users' data may be improperly accessed, used, or disclosed. If an actual or perceived breach of our or our business partners' or service providers' security occurs, the market perception of the effectiveness of our security measures would be harmed, we could lose users and customers, our trade secrets or those of our business partners may be compromised, and we may be exposed to significant legal and financial risks, including legal claims (which may include class-action litigation) and regulatory actions, fines, and penalties. Any of the foregoing consequences could harm our business, reputation, financial condition, and operating results. + +While we have dedicated significant resources to privacy and security incident response capabilities, including dedicated worldwide incident response teams, our response process, particularly during times of a natural disaster or pandemic, may not be adequate, may fail to accurately assess the severity of an incident, may not be fast enough to prevent or limit harm, or may fail to sufficiently remediate an incident. As a result, we may suffer significant legal, reputational, or financial exposure, which could harm our business, financial condition, and operating results. + +For additional information, see also our risk factor on privacy and data protection regulations under 'Risks Related to Laws, Regulations, and Policies' below. + +## Our ongoing investments in safety, security, and content review will likely continue to identify abuse of our platforms and misuse of user data. + +In addition to our efforts to prevent and mitigate cyber attacks, we are making significant investments in safety, security, and review efforts to combat misuse of our services and unauthorized access to user data by third parties, including investigation and review of platform applications that could access the information of users of our services. As a result of these efforts, we have in the past discovered, and may in the future discover, incidents of unnecessary access to or misuse of user data or other undesirable activity by third parties. However, we may not have discovered, and may in the future not discover, all such incidents or activity, whether as a result of our data limitations, including our lack of visibility over our encrypted services, the scale of activity on our platform, or other factors, including factors outside of our control such as a natural disaster or pandemic, and we may learn of such incidents or activity via third parties. Such incidents and activities may include the use of user data or our systems in a manner inconsistent with our terms, contracts, or policies, the existence of false or undesirable user accounts, election interference, improper ad purchases, activities that threaten people's safety on- or off-line, or instances of spamming, scraping, or spreading disinformation. While we may not determine some of these incidents to be material at the time they occurred and we may remedy them quickly, there is no guarantee that these issues will not ultimately result in significant legal, financial, and reputational harm, including government inquiries and enforcement actions, litigation, and negative publicity. There is also no guarantee that a series of related issues may not be determined to be material at a later date in the aggregate, even if they may not be material individually at the time of their occurrence. + +We may also be unsuccessful in our efforts to enforce our policies or otherwise prevent or remediate any such incidents. Any of the foregoing developments may negatively affect user trust and engagement, harm our reputation and brands, require us to change our business practices in ways that harm our business operations, and adversely affect our business and financial results. Any such developments may also subject us to additional litigation and regulatory inquiries, which could result in monetary penalties and damages, divert management's time and attention, and lead to enhanced regulatory oversight. + +Our business depends on continued and unimpeded access to the Internet by us and our users. Internet access providers may be able to restrict, block, degrade, or charge for access to certain of our products and services, which could lead to additional expenses and the loss of users and advertisers. + +Our products and services depend on the ability of our users to access the Internet, and certain of our products require significant network capacity to work effectively. Currently, this access is provided by companies that have significant market power in the broadband and internet access marketplace, including incumbent telephone companies, cable companies, mobile communications companies, and government-owned service providers. Some of these providers have taken, or have stated that they may take, measures that could degrade, disrupt, or increase the cost of user access to certain of our products by restricting or prohibiting the use of their infrastructure to support or facilitate our offerings, by charging increased fees to us or our users to provide our offerings, or by providing our competitors preferential access. Some jurisdictions have adopted regulations prohibiting certain forms of discrimination by internet access providers; however, substantial uncertainty exists in the US and elsewhere regarding such protections. In addition, in some jurisdictions, our products and services have been subject to government-initiated restrictions or blockages. These could harm existing key relationships, including with our users, customers, advertisers, and content providers, and impair our ability to attract new ones; harm our reputation; and increase costs, thereby negatively affecting our business. + +## Risks Related to Laws, Regulations, and Policies + +We are subject to a variety of new, existing, and changing laws and regulations worldwide that could harm our business, and will likely be subject to an even broader scope of laws and regulations as we continue to expand our business. + +We are subject to numerous US federal and state as well as foreign laws and regulations covering a wide variety of subjects, and our introduction of new businesses, products, services, and technologies will likely continue to subject us to additional laws and regulations. In recent years, governments around the world have proposed and adopted a large number of new laws and regulations relevant to the digital economy, particularly in the areas of data privacy and security, competition, AI, and online content. The costs of compliance with these measures are high and are likely to increase in the future, including as a result of differing, and sometimes conflicting, laws and regulations. + +New or changing laws and regulations, or interpretations or applications of existing laws and regulations in a manner inconsistent with our interpretations of such laws and regulations or our practices, have resulted in, and may continue to result in, less useful products and services, altered business models and operations, limited ability to pursue certain business practices or offer certain products and services, substantial costs, and civil or criminal liability. Examples include laws and regulations regarding: + +\- Competition and technology platforms' business practices: Laws and regulations focused on large technology platforms, including the Digital Markets Act in the European Union (EU) and the Act on Promotion of Competition for Specified Smartphone Software in Japan; regulations and legal settlements in the US, South Korea, and elsewhere that affect Google Play's billing policies, fees, and business model; as well as litigation and new and expected regulations in a range of jurisdictions. + +\- AI: Laws and regulations focused on the development, use, and provision of AI technologies and other digital products and services, which could result in monetary penalties or other regulatory actions. For example, the EU AI Act came into force on August 1, 2024, and will generally become fully applicable after a two-year transitional period (although certain obligations have already taken effect). The EU AI Act introduces various requirements for AI systems and models placed on the market in the EU, including specific transparency, safety, and copyright requirements for general purpose AI systems and the models on which those systems are based. Various countries, including Brazil, India, Japan, South Korea, Singapore, and Vietnam, have also enacted or are considering enacting regulations focused on AI. In the US, an increasing amount of legislative and regulatory activity regarding AI is taking place at the state level. In 2025, state legislatures considered more than 1,000 AI-related bills, including on fundamental model research and development, synthetic media, algorithmic decision-making, and many others, and took a variety of approaches to AI regulation. For instance, in 2025, California and New York passed the Transparency in Frontier Artificial Intelligence Act and the Responsible AI Safety and Education Act, respectively, each of which imposes safety and reporting obligations on developers of frontier models. At the same time, the White House's Executive Order, Removing Barriers to American Leadership in Artificial Intelligence, prioritizes deregulation, while its AI Action Plan emphasizes accelerating American innovation leadership. + +- Data privacy, collection, processing, and portability: Laws and regulations further restricting the collection, processing, or sharing of user or advertising-related data, including privacy and data protection laws; laws affecting the processing of children's data (as discussed further below), data breach notification laws; laws limiting data transfers (including data localization laws); laws limiting use of data for AI training; and laws requiring data portability. +- Copyright and other intellectual property: Copyright and related laws, including the EU Directive on Copyright in the Digital Single Market and European Economic Area transpositions, which have introduced new licensing regimes, increase liability with respect to content uploaded by users or linked to from our platforms, or create property rights in news publications that could require payments to news agencies and publishers, which may result in other regulatory actions. The scope of the text and data mining exception is being challenged before courts in the EU, which could harm some aspects of our business. +- Content moderation: Various laws covering content moderation and removal, and related disclosure obligations, such as the EU's Digital Services Act, Florida's Senate Bill 7072 and Texas' House Bill 20, and laws and proposed legislation in Singapore, Australia, and the United Kingdom (UK) that impose penalties for failure to remove certain types of content or require disclosure of information about the operation of our services and algorithms, which may make it harder for services like Google Search and YouTube to detect and limit low-quality, deceptive, or harmful content, or, on the other hand, may impinge on the rights of free expression and access to content. Additionally, new regulations apply to online child safety, including access and content restrictions as well as other limitations for minors, which may also conflict with rights of free expression and access to information. These regulations could result in our having to modify our products and services and monitor minors' experiences on our products and services. +- Consumer protection: Consumer protection laws, including the EU's New Deal for Consumers, which could result in monetary penalties and create a range of new compliance obligations. + +In addition, the applicability and scope of these and other laws and regulations, as interpreted by courts, regulators, or administrative bodies, remain uncertain and could be interpreted in ways that harm our business. For example, we rely on statutory safe harbors, like those set forth in the Digital Millennium Copyright Act and Section 230 of the Communications Decency Act in the US and the Digital Services Act in Europe, to protect against liability for various linking, caching, ranking, recommending, and hosting activities. Legislation or court rulings affecting these safe harbors may harm us and may impose significant operational challenges. There are legislative proposals and pending litigation in the US, EU, and around the world that could diminish or eliminate safe harbor protection for websites and online platforms. Our development, use, and commercialization of AI products and services (including our implementation of AI in our offerings and internal systems) could subject us to regulatory action and legal liability, including under specific legislation regulating AI, as well as new applications of existing data protection, cybersecurity, privacy, intellectual property, and other laws. + +Further, we are subject to evolving laws, regulations, policies, and international accords relating to matters beyond our core products and services, including environmental sustainability, climate change, human capital, and employment matters. In response, we have implemented robust programs and initiatives and adopted reporting frameworks and principles that may require considerable investments. For instance, AI's energy and water demands have made efforts to reduce our emissions more complex and challenging across every level. We cannot guarantee that our initiatives will be fully realized on the timelines we expect or at all, and projects that are completed as planned may not achieve the results we anticipate. + +We are and may continue to be subject to claims, lawsuits, regulatory and government inquiries and investigations, enforcement actions, consent orders, and other forms of regulatory scrutiny and legal liability, including competition matters, that could harm our business, reputation, financial condition, and operating results. + +We are subject to claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and orders involving competition, intellectual property, data privacy and security, tax and related compliance, labor and employment, commercial disputes, content generated by our users, goods and services offered by advertisers or publishers using our platforms, design of our products and services, personal injury and other tort and nuisance theories, consumer protection claims, including how we moderate content on our platforms, AI, and other matters. + +The US Department of Justice (DOJ), various states, and other plaintiffs have filed, and may continue to file in the future, several antitrust lawsuits about aspects of our business, including our advertising technologies and practices, the operation and distribution of Google Search, and the operation and distribution of Android operating system and Play Store. If we are unsuccessful in these lawsuits, we could face significant expenses to implement the remedies, and such costs and alterations could harm our business, reputation, financial condition, and operating results. + +For example, the DOJ and a number of state Attorneys General filed a lawsuit concerning our Search and Search advertising practices and our compliance with US antitrust laws. In August 2024, the US District Court for the District of Columbia ruled against Google, and in December 2025, entered a final judgment requiring remedies, which, among other things, imposes restrictions on how we distribute our services and requires us to share certain search data with and offer syndication services to certain competitors. In January 2026, we appealed the final judgment and moved to pause implementation of certain remedies. In February 2026, the DOJ and state Attorneys General also appealed. + +Furthermore, in December 2020, a number of state Attorneys General, led by the Texas Attorney General, filed a lawsuit in the US District Court for the Eastern District of Texas concerning our advertising technology and our compliance with US antitrust laws and other laws. In January 2023, the DOJ and a number of state Attorneys General sued in the Eastern District of Virginia alleging similar antitrust violations relating to our advertising technology. In April 2025, the presiding judge issued a mixed decision in the DOJ case against us, ruling that neither our advertiser tools nor the DoubleClick and AdMeld acquisitions were anticompetitive, but that our publisher tools unfairly excluded rivals. A separate proceeding to determine remedies, the range of which varies widely, took place in September 2025 with the parties presenting differing remedy proposals. The DOJ's remedy proposal includes structural remedies that could harm our business. Closing arguments were held in November 2025, and we are awaiting a final judgment. After that judgment, we plan to appeal the adverse portion of the April 2025 decision and potentially aspects of the remedies decision. A trial in the state Attorneys General case in the Eastern District of Texas will take place after a decision on remedies is issued in the DOJ advertising technology case, and could result in remedies that could harm our business, reputation, financial condition, and operating results. + +In addition to these regulatory proceedings, private individual and collective actions that overlap with claims pursued by regulators are pending in the US and in several other jurisdictions. Adverse results in these or similar future lawsuits may include awards of monetary damages and remedies that could harm our business, reputation, financial condition, and operating results. + +Other regulatory agencies in the US and around the world, including competition enforcers, consumer protection agencies, and data protection authorities, have challenged and may continue to challenge our business practices and compliance with laws and regulations. We are cooperating with these investigations and defending litigation or appealing decisions where appropriate. + +We are also subject to a variety of claims including product warranty, product liability, and consumer protection claims related to product defects, among other litigation, and we may also be subject to claims involving health and safety, hazardous materials usage, other environmental effects, AI training, development, and commercialization, or service disruptions or failures. Claims have been brought, and we expect will continue to be brought, against us for defamation, negligence, breaches of contract, patent, copyright, and trademark infringement, unfair competition, unlawful activity, torts, privacy rights violations, fraud, or other legal theories based on the nature and content of information available on or via our services, the design and effect of our products and services, or due to our involvement in hosting, transmitting, marketing, branding, or providing access to content created by third parties. + +Various laws, regulations, investigations, enforcement lawsuits, and regulatory actions have involved in the past, and may in the future result in substantial fines and penalties, injunctive relief, ongoing monitoring and auditing obligations, changes to our products and services, alterations to our business models and operations, including divestiture, and collateral related civil litigation or other adverse consequences. Any of these legal proceedings could also result in legal costs, diversion of management resources, and negative publicity, all of which could harm our business, reputation, financial condition, and operating results. + +Estimating liabilities for our pending proceedings is a complex, fact-specific, and speculative process that requires significant judgment, and the amounts we are ultimately liable for may differ from our estimates. The resolution of one or more such proceedings has resulted in, and may in the future result in, additional substantial fines, penalties, injunctions, and other sanctions that could harm our business, reputation, financial condition, and operating results. + +For additional information about the ongoing material legal proceedings to which we are subject, see Legal Proceedings in Item 3 of this Annual Report on Form 10-K. + +Privacy, data protection, data usage, and portability regulations are complex and rapidly evolving areas. Any failure or alleged failure to comply with these laws could harm our business, reputation, financial condition, and operating results. + +Authorities around the world have adopted and are considering a number of legislative and regulatory proposals concerning data protection, data usage and portability, and encryption of user data. Additionally, the increasing adoption of AI technologies, which rely on the collection of large amounts of data and use of such data for training purposes, has led data protection authorities around the world to consider and adopt new and evolving interpretations of data protection laws, imposing specific obligations with respect to the processing of personal data, including required notices, consents, and opt-outs. Further, the increased risk of inadvertent disclosure of confidential information or personal data in connection with the utilization of AI technologies may result in stronger regulatory scrutiny, leading to legal and regulatory investigations and enforcement actions that could harm our business, even if unfounded. Adverse legal rulings, legislation, or regulation have resulted in, and may continue to result in, fines and orders requiring that we change our practices, which have had and could continue to have an adverse effect on how we provide services, harming our business, reputation, financial condition, and operating results. These laws and regulations are evolving and subject to interpretation, and compliance obligations could cause us to incur substantial costs or harm the quality and operations of our products and services in ways that harm our business. Examples of these laws include: + +- The EU General Data Protection Regulation and the UK General Data Protection Regulations, which apply to all of our activities conducted from an establishment in the EU or the UK, respectively, or related to products and services that we offer to EU or the UK users or customers, respectively, or the monitoring of their behavior in the EU or the UK, respectively. +- Various US federal, US state, and foreign privacy laws related to the processing and security of personal data, including (1) comprehensive privacy laws that provide data privacy rights (including, in California, a private right of action in the event of a data breach resulting from our failure to implement and maintain reasonable security procedures and practices) and impose significant obligations on controllers and processors of consumer data; (2) laws imposing obligations on businesses that collect or disclose biometric information (including, in Colorado, Illinois, Texas, and Washington); (3) laws governing the collection and processing of children and minor's data and how companies provide age-appropriate online experiences (including, in the US, the Children's Online Privacy Protection Act of 1998; the pending Children and Teens' Online Privacy Protection Act (COPPA 2.0); similar US state laws related to children's privacy, such as the New York Child Data Protection Act; and the UK's Age-Appropriate Design Code); and (4) laws regulating internet-connected devices (such as, in California, the Internet of Things Security Law). +- The EU's Digital Markets Act, which requires in-scope companies to obtain user consent for combining data across certain products and require search engines to share anonymized data with rival companies, among other changes; and the EU Data Act, which introduces new data portability requirements with respect to connected products (i.e., 'internet of things' products) and related services, as well as interoperability obligations on data processing services. + +Further, we are subject to evolving laws and regulations that dictate whether, how, and under what circumstances we can transfer, process, or receive personal data, as well as ongoing enforcement actions from supervisory authorities related to cross-border transfers of personal data. The validity of various data transfer mechanisms we currently rely upon remains subject to legal, regulatory, and political developments globally, which may require us to adapt our existing arrangements. + +We face, and may continue to face, intellectual property infringement or misappropriation, violation of rights of publicity, and other claims that could be costly to defend, result in significant damage awards or other costs (including indemnification awards), and limit our ability to use certain technologies. + +We, like other internet, technology, and media companies, are frequently subject to litigation based on allegations of infringement or other violations of intellectual property rights, including patent, copyright, trade secrets, and trademarks. Parties have also sought broad injunctive relief against us by filing claims in US and international courts and the US International Trade Commission (ITC) for exclusion and cease-and-desist orders. In addition, patent-holding companies may frequently seek to generate income from patents they have obtained by bringing claims against us. As we continue to expand our business, intellectual property claims against us have increased and may continue to increase as we develop and acquire new products, services, and technologies, including AI technologies. + +Adverse results in any of these lawsuits may include awards of monetary damages, costly royalty or licensing agreements (if licenses are available at all), or orders limiting our ability to sell our products and services in the US or elsewhere, including by preventing us from offering certain features, functionalities, products, or services in certain jurisdictions. They may also cause us to change our business practices in ways that could result in a loss of revenues for us and otherwise harm our business. + +Many of our agreements with our customers and partners, including certain suppliers, require us to defend against certain intellectual property infringement claims and in some cases indemnify them for certain intellectual property infringement claims against them, which could result in increased costs for defending such claims or significant damages if there were an adverse ruling in any such claims. Such customers and partners may also discontinue the use of our products, services, and technologies, as a result of injunctions or otherwise, which could result in loss of revenues and harm our business. Moreover, intellectual property indemnities provided to us by our suppliers, when obtainable, may not cover all damages and losses suffered by us and our customers arising from intellectual property infringement claims. Furthermore, in connection with our divestitures, we have agreed, and may in the future agree, to provide indemnification for certain potential liabilities, including those associated with intellectual property claims. Regardless of their merits, intellectual property claims are often time consuming and expensive to litigate or settle. To the extent such claims are successful, they could harm our business, including our product and service offerings, financial condition, and operating results. + +We could be subject to changes in tax rates, the adoption of new US or international tax legislation, or exposure to additional tax liabilities. + +We are subject to a variety of taxes and tax collection obligations in the US and numerous foreign jurisdictions. Our effective tax rates are affected by a variety of factors, including changes in the mix of earnings in jurisdictions with different statutory tax rates, net gains and losses on hedges and related transactions under our foreign exchange risk management program, changes in our stock price for shares issued as employee compensation, changes in the valuation of our deferred tax assets or liabilities, and the application of different provisions of tax laws or changes in tax laws, regulations, or accounting principles (including changes in the interpretation of existing laws). Further, if we are unable or fail to collect taxes on behalf of customers, employees, and partners as the withholding agent, we could become liable for taxes that are levied against third parties. + +We are subject to regular review and audit by both domestic and foreign tax authorities. As a result, we have received, and may in the future receive, assessments in multiple jurisdictions, on various tax-related assertions, such as transfer-pricing adjustments or permanent-establishment claims. Any adverse outcome of such a review or audit could harm our financial condition and operating results, require adverse changes to our business practices, or subject us to additional litigation and regulatory inquiries. In addition, the determination of our worldwide provision for income taxes and other tax liabilities requires significant judgment and often involves uncertainty. Although we believe our estimates are reasonable, the ultimate tax outcome may differ from the amounts recorded in our financial statements and may affect our financial results in the period or periods for which such determination is made. + +Further, due to shifting economic and political conditions, tax policies, laws, or rates in various jurisdictions may be subject to significant changes in ways that could harm our financial condition and operating results. For example, various jurisdictions around the world have enacted or are considering revenue-based taxes such as digital services taxes and other targeted taxes, which could lead to inconsistent and potentially overlapping international tax regimes. The US has imposed a 1% excise tax on the fair market value of shares repurchased, which could increase in the future. Over 140 countries are negotiating changes to international tax policies led by the Organization for Economic Cooperation and Development (OECD), including a 15% global minimum tax rate. In January 2026, the OECD announced a "Side-by-Side Safe Harbor" that exempts US operations of US-parented companies from global minimum tax rules. Adoption of minimum tax rules outside the US could increase our effective tax rate and cash tax payments. + +## Risks Related to Ownership of Our Stock + +We cannot guarantee that any share repurchase program or dividend program will be continuously active or fully consummated or will enhance long-term stockholder value, and share repurchases or dividends could increase the volatility of our stock prices and could diminish our cash reserves. + +We engage in share repurchases of our Class A and Class C stock from time to time in accordance with authorizations from the Board of Directors of Alphabet. Our repurchase program does not have an expiration date and does not obligate Alphabet to repurchase any specific dollar amount or to acquire any specific number of shares. Our cash dividend program pays regular cash dividends to our Class A, Class B, and Class C stockholders. Any and all future cash dividends are subject to declaration by our Board of Directors in its sole discretion, and in accordance with the requirements of any applicable laws, rules, and regulations, including the Delaware General Corporation Law. Our cash dividend program does not require, and our Board of Directors may decide not to declare, a cash dividend each quarter, and does not obligate our Board of Directors to declare a dividend at any specific dollar amount per share. Any such decision by our Board of Directors may depend on a variety of factors that it may deem relevant, including but not limited to our earnings, liquidity, financial condition, other capital deployment opportunities, level of indebtedness, and general market conditions. Our share repurchases and dividends could affect our share trading prices, increase their volatility, reduce our cash reserves and may be suspended or terminated at any time, which may result in a decrease in the trading prices of our stock. + +The concentration of our stock ownership limits our stockholders' ability to influence corporate matters. + +Our Class B stock has 10 votes per share, our Class A stock has one vote per share, and our Class C stock has no voting rights. As of December 31, 2025, Larry Page and Sergey Brin beneficially owned approximately 89.3% of our outstanding Class B stock, which represented approximately 52.7% of the voting power of our outstanding common stock. Through their stock ownership, Larry and Sergey have significant influence over all matters requiring stockholder approval, including the election of directors and significant corporate transactions, such as a merger or other sale of our company or our assets, for the foreseeable future. In addition, because our Class C stock carries no voting rights (except as required by applicable law), the issuance of the Class C stock, including in future stock-based acquisition transactions and to fund employee equity incentive programs, could continue Larry and Sergey's current relative voting power and their ability to elect all of our directors and to determine the outcome of most matters submitted to a vote of our stockholders. The share repurchases made pursuant to our repurchase program may also affect Larry and Sergey's relative voting power. This concentrated control limits or severely restricts other stockholders' ability to influence corporate matters and we may take actions that some of our stockholders do not view as beneficial, which could reduce the market price of our Class A stock and our Class C stock. + +## Provisions in our charter documents and under Delaware law could discourage a takeover that stockholders may consider favorable. + +Provisions in Alphabet's certificate of incorporation and bylaws may have the effect of delaying or preventing a change of control or changes in our management. These provisions include the following: + +- Our Board of Directors has the right to elect directors to fill a vacancy created by the expansion of the Board of Directors or the resignation, death, or removal of a director. +- Our stockholders may not act by written consent, which makes it difficult to take certain actions without holding a stockholders' meeting. +- Our certificate of incorporation prohibits cumulative voting in the election of directors. This limits the ability of minority stockholders to elect director candidates. +- Stockholders must provide advance notice to nominate individuals for election to the Board of Directors or to propose matters that can be acted upon at a stockholders' meeting. These provisions may discourage or deter a potential acquirer from conducting a solicitation of proxies to elect the acquirer's own slate of directors or otherwise attempting to obtain control of our company. +- Our Board of Directors may issue, without stockholder approval, shares of undesignated preferred stock, which makes it possible for our Board of Directors to issue preferred stock with voting or other rights or preferences that could impede the success of any attempt to acquire us. + +As a Delaware corporation, we are also subject to certain Delaware anti-takeover provisions. Under Delaware law, a corporation may not engage in a business combination with any holder of 15% or more of its outstanding voting stock unless the holder has held the stock for three years or, among other things, the Board of Directors has approved the transaction. Our Board of Directors could rely on Delaware law to prevent or delay an acquisition of us. + +## General Risks + +Our operating results may fluctuate, which makes our results difficult to predict and could cause our results to fall short of expectations. + +Our operating results have fluctuated, and may in the future fluctuate, as a result of a number of factors, many outside of our control, including the cyclical nature and seasonality in our business and geopolitical events. As a result, comparing our operating results (including our expenses as a percentage of our revenues) on a period-to-period basis may not be meaningful, and our past results should not be relied on as an indication of our future performance. Consequently, our operating results in future quarters may fall below expectations. + +Acquisitions, joint ventures, investments, and divestitures could result in operating difficulties, dilution, and other consequences that could harm our business, financial condition, and operating results. + +Acquisitions, joint ventures, investments, and divestitures are important elements of our overall corporate strategy and use of capital, and these transactions could be material to our financial condition and operating results. We expect to continue to evaluate and enter into discussions regarding a wide array of such potential strategic arrangements, which could create unforeseen operating difficulties and expenditures. Some of the areas where we face risks include: + +- diversion of management time and focus from operating our business to challenges related to acquisitions and other strategic arrangements; +- failure to obtain required approvals on a timely basis, if at all, from governmental authorities, or conditions placed upon approval, either of which could, among other things, delay or prevent us from completing a transaction, result in the payment of fees or penalties to a counterparty, or otherwise restrict our ability to + +realize the expected financial or strategic goals of a transaction; or investigations or litigation by governmental authorities related to our acquisitions, investments, and other strategic arrangements; + +- failure to successfully integrate the acquired operations, technologies, services, and personnel (including cultural integration and retention of employees) and further develop the acquired business or technology; +- implementation of controls (or remediation of deficiencies), procedures, and policies at the acquired company; +- integration of the acquired company's accounting and other administrative systems, and the coordination of product, engineering, and sales and marketing functions; +• transition of operations, users, and customers onto our existing platforms; +- in the case of foreign acquisitions, the need to integrate operations across different cultures and languages and to address the particular economic, currency, political, and regulatory risks associated with specific countries; +- failure to accomplish commercial, strategic, or financial objectives with respect to investments, joint ventures, and other strategic arrangements; +- failure to realize the value of investments and joint ventures due to a lack of liquidity or an inability to identify buyers or negotiate favorable terms for intended divestitures; +- liability for activities of the acquired company before the acquisition, including intellectual property infringement claims, data privacy and security issues, violations of laws, commercial disputes, tax liabilities, warranty claims, product liabilities, and other known and unknown liabilities; and +- litigation or other claims in connection with the acquired company, including claims from terminated employees, customers, former stockholders, or other third parties. + +Our failure to address these risks or other problems encountered in connection with our past or future acquisitions and other strategic arrangements could cause us to fail to realize their anticipated benefits, incur unanticipated liabilities, and harm our business generally. + +Our acquisitions and other strategic arrangements could also result in dilutive issuances of our equity securities, the incurrence of debt, contingent liabilities, or amortization expenses, or impairment of goodwill or purchased long-lived assets, and restructuring charges, any of which could harm our financial condition and operating results. Also, the anticipated benefits or value of our acquisitions, investments, and other strategic arrangements may not materialize. In connection with our divestitures and certain strategic arrangements, we have agreed, and may in the future agree, to provide indemnification for certain potential liabilities, which could harm our financial condition and operating results. + +We rely on highly skilled personnel and, if we are unable to retain or motivate key personnel, hire qualified personnel, or maintain and continue to adapt our corporate culture, we may not be able to grow or operate effectively. + +Our performance and future success depends in large part upon the continued service of key technical leads as well as members of our senior management team. For instance, Sundar Pichai is critical to the overall management of Alphabet and its subsidiaries and plays an important role in the development of our technology, maintaining our culture, and setting our strategic direction. + +Our ability to compete effectively and our future success depend on our continuing to identify, hire, develop, motivate, and retain highly skilled personnel for all areas of our organization. Competition in our industry for qualified employees, particularly AI talent, is intense, and certain of our competitors have directly targeted, and may continue to target, our employees. In addition, our compensation arrangements, such as our equity award programs, may not always be successful in attracting new employees and retaining and motivating our existing employees. Immigration policy and regulatory changes, and uncertainty regarding such policies and regulations, may also affect our ability to hire, mobilize, or retain some of our global talent. All of our executive officers and key employees are at-will employees, and we do not maintain any key-person life insurance policies. + +We believe that our corporate culture fosters innovation, creativity, and teamwork. As our organization grows and evolves, we may need to adapt our corporate culture and work environments to ever-changing circumstances, and these changes could affect our ability to compete effectively or have an adverse effect on our corporate culture. + +## ITEM 1B. UNRESOLVED STAFF COMMENTS + +Not applicable. + +## ITEM 1C. CYBERSECURITY + +We maintain a comprehensive process for identifying, assessing, and managing material risks from cybersecurity threats as part of our broader risk management system and processes. For example, some risks include our software supply chain and other third-party dependencies, vulnerabilities in our products and services, theft of our intellectual property, and attempts to compromise our infrastructure. We obtain input, as appropriate, for our cybersecurity risk management program on the security industry and threat trends from multiple external experts and internal threat intelligence teams. Teams of dedicated privacy, safety, and security professionals oversee cybersecurity risk management and mitigation, incident prevention, detection, and remediation. These teams comprise professionals with deep cybersecurity expertise across multiple industries and are led by our Vice President of Privacy, Safety, and Security, who has more than 20 years of experience, including roles in technology infrastructure for two other large public companies. Our executive leadership team, along with input from the above teams, are responsible for our overall enterprise risk management system and processes and regularly consider cybersecurity risks in the context of other material risks to the company. + +As part of our cybersecurity risk management process, our incident management teams track and log privacy and security incidents across Alphabet, our vendors, and other third-party service providers to remediate and resolve any such incidents. Significant incidents are reviewed regularly by a cross-functional working group to determine whether further escalation is appropriate. Any incident assessed as potentially being or potentially becoming material is promptly escalated for further assessment, and then reported to designated members of our senior management. We consult with outside counsel as appropriate, including on materiality analysis and disclosure matters, and our senior management makes the final materiality determinations and disclosure and other compliance decisions. Our management apprises Alphabet's independent public accounting firm of relevant matters and developments. + +The Risk and Compliance Committee has oversight responsibility for risks and incidents relating to cybersecurity threats, including compliance with disclosure requirements, cooperation with law enforcement, and related effects on financial and other risks, and it reports any findings and recommendations, as appropriate, to the full Board for consideration. Senior management regularly discusses cybersecurity risks and trends and, should they arise, any material incidents with the Risk and Compliance Committee. Internal Audit maintains a dedicated cybersecurity auditing team that independently tests our cybersecurity controls. + +Our business strategy, results of operations and financial condition have not been materially affected by risks from cybersecurity threats, including as a result of previously identified cybersecurity incidents, but we cannot provide assurance that they will not be materially affected in the future by such risks or any future material incidents. For more information on our cybersecurity related risks, see Item 1A Risk Factors of this Annual Report on Form 10-K. + +## ITEM 2. PROPERTIES + +Our headquarters are located in Mountain View, California. We own and lease office facilities and data centers around the world, primarily in Asia, Europe, and North America. We believe our existing facilities are in good condition and suitable for the conduct of our business. + +## ITEM 3. LEGAL PROCEEDINGS + +For a description of our material pending legal proceedings, see Legal Matters in Note 10 of the Notes to Consolidated Financial Statements included in Part II, Item 8 of this Annual Report on Form 10-K, which is incorporated herein by reference. + +## ITEM 4. MINE SAFETY DISCLOSURES + +Not applicable. + +## PART II + +## ITEM 5. MARKET FOR REGISTRANT'S COMMON EQUITY, RELATED STOCKHOLDER MATTERS, AND ISSUER PURCHASES OF EQUITY SECURITIES + +As of October 2, 2015, Alphabet Inc. became the successor issuer of Google Inc. pursuant to Rule 12g-3(a) under the Exchange Act. Our Class A stock has been listed on the Nasdaq Global Select Market under the symbol "GOOG" since August 19, 2004, and under the symbol "GOOGL" since April 3, 2014. Prior to August 19, 2004, there was no public market for our stock. Our Class B stock is neither listed nor traded. Our Class C stock has been listed on the Nasdaq Global Select Market under the symbol "GOOG" since April 3, 2014. + +## Holders of Record + +As of December 31, 2025, there were approximately 5,861 and 1,611 stockholders of record of our Class A stock and Class C stock, respectively. Because many of our shares of Class A stock and Class C stock are held by brokers and other institutions on behalf of stockholders, we are unable to estimate the total number of stockholders represented by these record holders. As of December 31, 2025, there were approximately 57 stockholders of record of our Class B stock. + +## Dividend Program + +Dividend payments to stockholders of Class A, Class B, and Class C shares were \$4.8 billion, \$703 million, and \$4.5 billion, respectively, for the year ended December 31, 2025. The company intends to pay quarterly cash dividends in the future, subject to review and approval by the company's Board of Directors in its sole discretion. We regularly evaluate our cash and capital structure, including the size, pace, and form of capital return to stockholders. + +## Issuer Purchases of Equity Securities + +The following table presents information with respect to Alphabet's repurchases of Class A and Class C stock during the quarter ended December 31, 2025: + +
PeriodTotal Number of Class A Shares Purchased (in thousands)(1)Total Number of Class C Shares Purchased (in thousands)(1)Average Price Paid per Class A Share(2)Average Price Paid per Class C Share(2)Total Number of Shares Purchased as Part of Publicly Announced Programs (in thousands)(1)Approximate Dollar Value of Shares that May Yet Be Purchased Under the Program (in millions)
October 1 - 3184113,305$ 253.20$ 255.0314,146$ 71,231
November 1 - 302095,885$ 286.00$ 281.466,094$ 69,503
December 1 - 3100$ 0.00$ 0.000$ 69,503
Total1,05019,19020,240
+ +(1) In April 2024, the company's Board of Directors authorized a \$70.0 billion share repurchase program for its Class A and Class C shares. In April 2025, the company's Board of Directors authorized an additional \$70.0 billion share repurchase program for its Class A and Class C shares. Repurchases are being executed from time to time, subject to general business and market conditions and other investment opportunities, through open market purchases or privately negotiated transactions, including through Rule 10b5-1 plans. The repurchase programs do not have an expiration date. For additional information related to share repurchases, see Note 11 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. +(2) Average price paid per share includes costs associated with the repurchases. + +## Stock Performance Graphs + +The graph below matches Alphabet Inc. Class A's cumulative five-year total stockholder return on common stock with the cumulative total returns of the S&P 500 index, the NASDAQ Composite index, and the RDG Internet Composite index. The graph tracks the performance of a \$100 investment in our common stock and in each index (with the reinvestment of all dividends) from December 31, 2020, to December 31, 2025. The returns shown are based on historical results and are not intended to suggest future performance. + +## COMPARISON OF CUMULATIVE 5-YEAR TOTAL RETURN\* + +## ALPHABET INC. CLASS A COMMON STOCK + +Among Alphabet Inc., the S&P 500 Index, the + +NASDAQ Composite Index, and the RDG Internet Composite Index + +![](images/35a1be409ff10928ffa22cc46162da7c3489970479062a937876e325e6f4969a.jpg) + +
+line chart + +| Date | Alphabet Inc. Class A | S&P 500 | NASDAQ Composite | RDG Internet Composite | +|--------|------------------------|---------|------------------|------------------------| +| 12/20 | $100 | $100 | $100 | $100 | +| 3/21 | $120 | $110 | $110 | $105 | +| 6/21 | $140 | $115 | $115 | $110 | +| 9/21 | $150 | $120 | $120 | $105 | +| 12/21 | $165 | $130 | $125 | $100 | +| 3/22 | $160 | $125 | $120 | $95 | +| 6/22 | $140 | $110 | $105 | $85 | +| 9/22 | $120 | $100 | $95 | $75 | +| 12/22 | $105 | $95 | $85 | $65 | +| 3/23 | $110 | $105 | $95 | $75 | +| 6/23 | $130 | $120 | $105 | $85 | +| 9/23 | $145 | $125 | $110 | $90 | +| 12/23 | $160 | $135 | $120 | $95 | +| 3/24 | $170 | $145 | $130 | $100 | +| 6/24 | $205 | $155 | $140 | $105 | +| 9/24 | $190 | $165 | $145 | $110 | +| 12/24 | $215 | $170 | $155 | $115 | +| 3/25 | $180 | $160 | $140 | $105 | +| 6/25 | $200 | $175 | $160 | $120 | +| 9/25 | $250 | $185 | $175 | $130 | +| 12/25 | $360 | $195 | $185 | $135 | +
+ +\*\$100 invested on December 31, 2020, in stock or index, including reinvestment of dividends. +Copyright $^{©}$ 2026 S&P, a division of The McGraw-Hill Companies Inc. All rights reserved. + +The graph below matches Alphabet Inc. Class C's cumulative five-year total stockholder return on capital stock with the cumulative total returns of the S&P 500 index, the NASDAQ Composite index, and the RDG Internet Composite index. The graph tracks the performance of a \$100 investment in our Class C capital stock and in each index (with the reinvestment of all dividends) from December 31, 2020, to December 31, 2025. The returns shown are based on historical results and are not intended to suggest future performance. + +## COMPARISON OF CUMULATIVE 5-YEAR TOTAL RETURN\* + +## ALPHABET INC. CLASS C CAPITAL STOCK + +Among Alphabet Inc., the S&P 500 Index, the + +NASDAQ Composite Index, and the RDG Internet Composite Index + +![](images/4c314e6ebadd1d2697f30ec183852cf94657538413c2e56652e5be51dd631c1a.jpg) + +
+line chart + +| Date | Alphabet Inc. Class C | S&P 500 | NASDAQ Composite | RDG Internet Composite | +|--------|------------------------|---------|------------------|------------------------| +| 12/20 | $100 | $100 | $100 | $100 | +| 3/21 | $120 | $110 | $110 | $105 | +| 6/21 | $140 | $115 | $115 | $110 | +| 9/21 | $150 | $120 | $120 | $105 | +| 12/21 | $160 | $130 | $125 | $100 | +| 3/22 | $165 | $135 | $125 | $95 | +| 6/22 | $150 | $130 | $120 | $85 | +| 9/22 | $130 | $110 | $105 | $75 | +| 12/22 | $105 | $105 | $95 | $60 | +| 3/23 | $110 | $115 | $105 | $70 | +| 6/23 | $130 | $125 | $115 | $80 | +| 9/23 | $145 | $130 | $120 | $85 | +| 12/23 | $160 | $140 | $130 | $90 | +| 3/24 | $170 | $150 | $140 | $95 | +| 6/24 | $210 | $160 | $150 | $105 | +| 9/24 | $190 | $165 | $155 | $110 | +| 12/24 | $220 | $170 | $160 | $115 | +| 3/25 | $180 | $165 | $155 | $105 | +| 6/25 | $200 | $175 | $165 | $115 | +| 9/25 | $250 | $185 | $175 | $125 | +| 12/25 | $360 | $195 | $185 | $135 | +
+ +\*\$100 invested on December 31, 2020, in stock or in index, including reinvestment of dividends. +Copyright $^{©}$ 2026 S&P, a division of The McGraw-Hill Companies Inc. All rights reserved. + +## ITEM 6. [Reserved] + +## ITEM 7. MANAGEMENT'S DISCUSSION AND ANALYSIS OF FINANCIAL CONDITION AND RESULTS OF OPERATIONS + +Please read the following discussion and analysis of our financial condition and results of operations together with "Note about Forward-Looking Statements," Part I, Item 1 "Business," Part I, Item 1A "Risk Factors," and our consolidated financial statements and related notes included under Item 8 of this Annual Report on Form 10-K. + +The following section generally discusses 2025 results compared to 2024 results. Discussion of 2024 results compared to 2023 results to the extent not included in this report can be found in Item 7 of our 2024 Annual Report on Form 10-K. + +## Understanding Alphabet's Financial Results + +Alphabet is a collection of businesses — the largest of which is Google. We report Google in two segments, Google Services and Google Cloud, and all non-Google businesses collectively as Other Bets. Supporting these businesses, we have centralized certain AI-related research and development focused on advanced research in AI and developing the frontier models that serve our businesses, which is reported in Alphabet-level activities. For further details on our segments, see Part I, Item 1 Business and Note 15 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +## Trends in Our Business and Financial Effect + +The following long-term trends have contributed to the results of our consolidated operations, and we anticipate that they will continue to affect our future results: + +\- As we continue to grow our business and meet the evolving behaviors and needs of our users and customers, our revenue growth and mix along with our cost and margin profiles are being influenced by a number of factors, including: + +Expanded AI Offerings in our Products and Services: The continuing evolution of the online world has contributed to the growth of our business. We expect that this evolution, including user engagement with AI products and services, will continue to benefit our business and our revenues. As we continue to incorporate AI into our products and services, such as with AI Overviews and AI Mode in Search, and with enterprise AI solutions on our Google Cloud Platform, we may monetize differently than our historical consumer and enterprise offerings which could affect revenue growth rates and margin trends. When developing new products and services we generally focus first on user experience and then on monetization. At the same time, we face increasing competition, including from other developers and providers of AI products and services, which may affect our revenues. + +Increasing Revenues Beyond Advertising: Revenues from cloud, consumer subscriptions, platforms, and devices, which may have differing characteristics than our advertising revenues, have grown over time. Certain of these revenues have been growing at a rate higher than our advertising revenues, becoming a larger percentage of our consolidated revenues, and we expect this trend to continue. The margins on these revenues vary significantly and are generally lower than the margins on our advertising revenues. + +Increased Investment in Technical Infrastructure: We continue to invest in capital expenditures as we scale our technical infrastructure, in particular for AI, to meet the demand of our users and enterprise customers and to support research internally. We invested heavily in capital expenditures in 2025 and in 2026, we expect to significantly increase, relative to 2025, our investment in our technical infrastructure, including servers and network equipment, and data centers. The costs associated with operating our technical infrastructure - depreciation, energy, equipment, and network capacity - are expected to significantly increase as developing and serving AI offerings require more compute power than our historical consumer and enterprise offerings. While our technical infrastructure costs increase, we expect to continue to drive efficiencies in our data centers, for example, through the design of our AI models and our TPU and GPU-based technical infrastructure. + +Continued Investment in Intellectual Property through R&D and Acquisitions: We continue to make significant research and development investments in areas of strategic focus as we seek to develop new, innovative offerings, and improve our existing offerings across our businesses. Acquisitions and strategic investments remain important elements in our use of capital and contribute to the breadth and depth of our offerings, expand our expertise in engineering and other functional areas, and build strong partnerships around strategic initiatives. + +Traffic Acquisition Costs Growth and Rate Changes: We expect traffic acquisition costs ("TAC") paid to our distribution partners and Google Network partners to increase as our advertising revenues grow. Our overall TAC as a percentage of our advertising revenues ("TAC rate") has been decreasing primarily due to a revenue mix shift from Google Network properties to Google Search & other properties. Our TAC rate will continue to be affected by changes in device mix; geographic mix; partner agreement terms; partner mix; the percentage of queries channeled through paid access points; product mix; the relative revenue growth rates of advertising revenues from different channels; and revenue share terms. + +• We have raised capital through external financing in the form of debt and we may continue to seek debt or other forms of financing in the future to support our capital and operating needs. +In 2025, we raised capital through the issuance of debt and we expect to continue to assess the use of debt and other forms of financing in the future. We expect to continue to enter into finance leases, primarily for data centers. Additionally, in 2025, we provided credit support, such as through backstops and guarantees, to certain infrastructure related counterparties and may continue to provide additional credit support in the future. +• We face an evolving regulatory environment, and we are subject to claims, lawsuits, investigations, and other forms of potential legal liability, which could affect our business practices and financial results. + +Changes in social, political, economic, tax, and regulatory conditions or in laws and policies governing a wide range of topics and related legal matters, including investigations, lawsuits, and regulatory actions, have resulted in fines and caused us to change our business practices. As the regulatory environment continues to evolve, we may continue to incur fines and we expect increased costs associated with compliance, modifications to our products and services, and limitations on our ability to pursue certain business practices. For additional information, see Part I, Item 1A Risk Factors and Legal Matters in Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +## Revenues and Monetization Metrics + +We generate revenues by delivering relevant, cost-effective online advertising; cloud-based solutions that provide enterprise customers of all sizes with infrastructure, platform services, and applications; and sales of other products and services, such as fees received for subscription-based products, apps and in-app purchases, and devices. For additional information on how we recognize revenue, see Note 1 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +In addition to the long-term trends and their financial effect on our business discussed above, fluctuations in our revenues have been and may continue to be affected by a combination of factors, including: + +• changes in foreign currency exchange rates; +• changes in pricing, such as those resulting from changes in fee structures, discounts, and customer incentives; +- general economic conditions and various external dynamics, including geopolitical events, regulations, and other measures and their effect on advertiser, consumer, and enterprise spending; +- new product, service, and market launches; and +- seasonality. + +Additionally, fluctuations in our revenues generated from advertising ("Google advertising"), other sources ("Google subscriptions, platforms, and devices"), Google Cloud, and Other Bets have been, and may continue to be, affected by other factors unique to each set of revenues, as described below. + +## Google Services + +Google Services revenues consist of Google advertising as well as Google subscriptions, platforms, and devices revenues. + +## Google Advertising + +Google advertising revenues are comprised of the following: + +- Google Search & other, which includes revenues generated on Google search properties (including revenues from traffic generated by search distribution partners who use Google.com as their default search in browsers, toolbars, etc.), and other Google owned and operated properties like Gmail, Google Maps, and Google Play; +• YouTube ads, which includes revenues generated on YouTube properties; and +- Google Network, which includes revenues generated on Google Network properties participating in AdMob, AdSense, and Google Ad Manager. + +We use certain metrics to track how well traffic across various properties is monetized as it relates to our advertising revenues: paid clicks and cost-per-click pertain to traffic on Google Search & other properties, while impressions and cost-per-impression pertain to traffic on our Google Network properties. + +Paid clicks represent engagement by users and include clicks on advertisements by end-users on Google search properties and other Google owned and operated properties including Gmail, Google Maps, and Google Play. Cost-per-click is defined as click-driven revenues divided by our total number of paid clicks and represents the average amount we charge advertisers for each engagement by users. + +Impressions include impressions displayed to users on Google Network properties participating primarily in AdMob, AdSense, and Google Ad Manager. Cost-per-impression is defined as impression-based and click-based revenues divided by our total number of impressions, and represents the average amount we charge advertisers for each impression displayed to users. + +As our business evolves, we periodically review, refine, and update our methodologies for monitoring, gathering, and counting the number of paid clicks and the number of impressions, and for identifying the revenues generated by the corresponding click and impression activity. + +Fluctuations in our advertising revenues, as well as the change in paid clicks and cost-per-click on Google Search & other properties and the change in impressions and cost-per-impression on Google Network properties and the correlation between these items have been, and may continue to be, affected by factors in addition to the general factors described above, such as: + +• advertiser competition for keywords; +• changes in advertising quality, formats, delivery, or policy; +• changes in device mix; +- seasonal fluctuations in internet usage, advertising expenditures, and underlying business trends, such as traditional retail seasonality; and +- traffic growth in emerging markets compared to more mature markets and across various verticals and channels. + +## Google Subscriptions, Platforms, and Devices + +Google subscriptions, platforms, and devices revenues are comprised of the following: + +- consumer subscriptions, which primarily include revenues from YouTube services, such as YouTube TV, YouTube Music and Premium, and NFL Sunday Ticket, as well as Google One, which offers access to our most capable Gemini models; +- platforms, which primarily include revenues from Google Play sales of apps and in-app purchases; +• devices, which primarily include sales of the Pixel family of devices; and +• other products and services. + +Fluctuations in our Google subscriptions, platforms, and devices revenues have been, and may continue to be, affected by factors in addition to the general factors described above, such as changes in customer usage and demand, number of subscribers, and the timing of product launches. + +## Google Cloud + +Google Cloud revenues are comprised of the following: + +- Google Cloud Platform primarily generates consumption-based fees and subscriptions for infrastructure, platform, and other services. These services provide access to solutions such as AI offerings including our enterprise AI infrastructure, Vertex AI platform, and Gemini Enterprise; cybersecurity offerings; and data and analytics solutions; +- Google Workspace includes subscriptions for cloud-based communication and collaboration tools for enterprises, such as Gmail, Docs, Calendar, Drive, and Meet, with integrated features like Gemini for Google Workspace; and +• other enterprise services. + +Fluctuations in our Google Cloud revenues have been, and may continue to be, affected by factors in addition to the general factors described above, such as changes in customer usage, demand, and supply availability. + +## Other Bets + +Revenues from Other Bets are generated primarily from the sale of autonomous transportation services and internet services. + +## Costs and Expenses + +Our cost structure has two components: cost of revenues and operating expenses. Our operating expenses include costs related to research and development, sales and marketing, and general and administrative functions. Certain of our costs and expenses, including those associated with the operation of our technical infrastructure as well as components of our operating expenses, are generally less variable in nature and may not correlate to changes in revenue. Additionally, fluctuations in employee compensation expenses may not directly correlate with changes in headcount, due to factors such as annual SBC awards that vest over time. + +## Cost of Revenues + +Cost of revenues is comprised of TAC and other costs of revenues. + +• TAC includes: + +amounts paid to our distribution partners who make available our search access points and other ad-supported services. Our distribution partners include browser providers, mobile carriers, original equipment manufacturers, and software developers; and +• amounts paid to Google Network partners primarily for ads displayed on their properties. + +• Other cost of revenues primarily includes: + +content acquisition costs, which are payments to content providers from whom we license video and other content for distribution, primarily related to YouTube (we pay fees to these content providers based on revenues generated, subscriber counts, or a flat fee); +• depreciation expense, primarily related to our technical infrastructure; +• employee compensation expenses related to our technical infrastructure and other operations such as content review and customer and product support; +inventory and other costs related to the devices we sell; and +other technical infrastructure operations costs, including energy, equipment, and network capacity costs. + +TAC as a percentage of revenues generated from ads placed on Google Network properties are significantly higher than TAC as a percentage of revenues generated from ads placed on Google Search & other properties, because most of the advertiser revenues from ads served on Google Network properties are paid as TAC to our Google Network partners. + +## Operating Expenses + +Operating expenses are generally incurred during our normal course of business, which we categorize as either research and development, sales and marketing, or general and administrative. + +The main components of our research and development expenses are: + +• depreciation expense, primarily related to our technical infrastructure; +- employee compensation expenses for engineering and technical employees responsible for research and development related to our existing and new products and services; +• other technical infrastructure operations costs, including energy, equipment, and network capacity costs; and +- third-party services fees primarily relating to consulting and outsourced services in support of our engineering and product development efforts. + +The main components of our sales and marketing expenses are: + +- employee compensation expenses for employees engaged in sales and marketing, sales support, and certain customer service functions; and +- spend relating to our advertising and promotional activities in support of our products and services. + +The main components of our general and administrative expenses are: + +- employee compensation expenses for employees in finance, human resources, information technology, legal, and other administrative support functions; +• expenses relating to legal and other matters, including certain fines and settlements; and +- third-party services fees, including audit, consulting, outside legal, and other outsourced administrative services. + +## Other Income (Expense), Net + +OI&E, net primarily consists of interest income (expense), the effect of foreign currency exchange gains (losses), net gains (losses) and impairment on our marketable and non-marketable securities and income (loss) and impairment from our equity method investments. + +For additional information, including how we account for our investments and factors that can drive fluctuations in the value of our investments, see Note 1 and Note 3 of the Notes to Consolidated Financial Statements included in Item 8 as well as Item 7A Quantitative and Qualitative Disclosures About Market Risk of this Annual Report on Form 10-K. + +## Provision for Income Taxes + +Provision for income taxes represents the estimated amount of federal, state, and foreign income taxes incurred in the US and the many jurisdictions in which we operate. The provision includes the effect of reserve provisions and changes to reserves that are considered appropriate as well as the related net interest and penalties. + +For additional information, including a reconciliation of the US federal statutory rate to our effective tax rate, see Note 14 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +## Executive Overview + +The following table summarizes consolidated financial results (in millions, except for per share information and percentages): + +
Year Ended December 31,$ Change% Change
20242025
Consolidated revenues$ 350,018$ 402,836$ 52,81815 %
Cost of revenues$ 146,306$ 162,535$ 16,22911 %
Operating expenses$ 91,322$ 111,262$ 19,94022 %
Operating income$ 112,390$ 129,039$ 16,64915 %
Operating margin32 %32 %0 %
Other income (expense), net$ 7,425$ 29,787$ 22,362301 %
Net income$ 100,118$ 132,170$ 32,05232 %
Diluted net income per share(1)$ 8.04$ 10.81$ 2.7734 %
+ +(1) For additional information on the calculation of diluted net income per share, see Note 12 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +- Revenues were \$402.8 billion, an increase of 15% year over year, primarily driven by an increase in Google Services revenues of \$37.8 billion, or 12%, and an increase in Google Cloud revenues of \$15.5 billion, or 36%. +- Cost of revenues was \$162.5 billion, an increase of 11% year over year, primarily driven by increases in TAC, content acquisition costs, and depreciation expense. +- Operating expenses were \$111.3 billion, an increase of 22% year over year, primarily driven by increases in employee compensation expenses, expenses related to legal and other matters, and depreciation expense. + +## Other Information: + +- In 2025, we entered into definitive agreements to acquire Wiz, a leading cloud security platform, for \$32.0 billion, and Intersect, a provider of data center and energy infrastructure solutions, for \$4.8 billion in cash plus the assumption of debt. Both acquisitions are expected to close in 2026, subject to customary closing conditions, including the receipt of regulatory approvals. +- In 2025, we issued senior unsecured notes for net proceeds of \$37.3 billion, to be used for general corporate purposes. +- OI&E of \$29.8 billion for the year ended December 31, 2025 included net gains on equity securities of \$24.1 billion, primarily related to unrealized gains on our non-marketable equity securities. +- Other Bets operating loss of \$7.5 billion for the year ended December 31, 2025 included a \$2.1 billion employee compensation charge recognized in the fourth quarter for Waymo, primarily reflected in research and development expenses, based on estimated stock valuation. In February 2026, Waymo announced an investment round of \$16.0 billion, the significant majority of which was funded by Alphabet. +- Changes to U.S. tax law enacted on July 4, 2025, allow, among other things, for immediate expensing of domestic research and experimentation costs and accelerated depreciation on eligible capital expenditures, the effects of which are included in operating cash flows for the year ended December 31, 2025. +- Repurchases of Class A and Class C shares were \$6.5 billion and \$38.9 billion, respectively, totaling \$45.4 billion for the year ended December 31, 2025. +- Operating cash flow was \$164.7 billion for the year ended December 31, 2025. +- Capital expenditures, which primarily reflected investments in technical infrastructure, were \$91.4 billion for the year ended December 31, 2025. +• As of December 31, 2025, we had 190,820 employees. + +We are monitoring ongoing developments surrounding international trade and the macroeconomic environment. As a result of volatility in international trade and financial markets, we may experience direct and indirect effects on our business, operations, and financial results. Our past results may not be indicative of our future performance, and our financial results may differ materially from historical trends. + +## Financial Results + +## Revenues + +The following table presents revenues by type (in millions): + +
Year Ended December 31,
20242025
Google Search & other$ 198,084$ 224,532
YouTube ads36,14740,367
Google Network30,35929,792
Google advertising264,590294,691
Google subscriptions, platforms, and devices40,34048,030
Google Services total304,930342,721
Google Cloud43,22958,705
Other Bets1,6481,537
Hedging gains (losses)211(127)
Total revenues$ 350,018$ 402,836
+ +## Google Services + +Google Advertising + +Google Search & other + +Google Search & other revenues increased \$26.4 billion from 2024 to 2025. The overall growth was driven by interrelated factors including increases in search queries resulting from growth in user adoption and usage on mobile devices; growth in advertiser spending; and improvements we have made in ad formats and delivery. + +## YouTube ads + +YouTube ads revenues increased \$4.2 billion from 2024 to 2025. The growth was driven by our direct response advertising products followed by our brand advertising products, both of which benefited from increased spending by our advertisers. + +## Google Network + +Google Network revenues decreased \$567 million from 2024 to 2025, primarily due to a decrease in AdSense revenues, partially offset by an increase in AdMob revenues. + +## Monetization Metrics + +The following table presents changes in monetization metrics for Google Search & other revenues (paid clicks and cost-per-click) and Google Network revenues (impressions and cost-per-impression), expressed as a percentage, from 2024 to 2025: + +
Google Search & other
Paid clicks change6 %
Cost-per-click change7 %
Google Network
Impressions change(7)%
Cost-per-impression change7 %
+ +Changes in paid clicks and impressions are driven by a number of interrelated factors, including changes in advertiser spending; ongoing product and policy changes; and, as it relates to paid clicks, fluctuations in search queries resulting from changes in user adoption and usage, primarily on mobile devices. + +Changes in cost-per-click and cost-per-impression are driven by a number of interrelated factors including changes in device mix, geographic mix, advertiser spending, ongoing product and policy changes, product mix, property mix, and changes in foreign currency exchange rates. + +## Google Subscriptions, Platforms, and Devices + +Google subscriptions, platforms, and devices revenues increased \$7.7 billion from 2024 to 2025. The growth was primarily driven by an increase in subscriptions revenues. This increase was primarily due to the contribution from growth in paid subscriptions across both YouTube services and Google One. + +## Google Cloud + +Google Cloud revenues increased \$15.5 billion from 2024 to 2025, primarily driven by growth in Google Cloud Platform largely from infrastructure and platform services. + +## Revenues by Geography + +The following table presents revenues by geography as a percentage of revenues, determined based on the addresses of our customers: + +
Year Ended December 31,
20242025
United States49 %48 %
$EMEA^{(1)}$ 29 %29 %
$APAC^{(1)}$ 16 %17 %
$Other Americas^{(1)}$ 6 %6 %
Hedging gains (losses)0 %0 %
+ +(1) Regions represent Europe, the Middle East, and Africa (EMEA); Asia-Pacific (APAC); and Canada and Latin America ("Other Americas"). + +For additional information, see Note 2 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +## Costs and Expenses + +## Cost of Revenues + +The following table presents cost of revenues, including TAC (in millions, except percentages): + +
Year Ended December 31,
20242025
TAC$ 54,900$ 59,926
Other cost of revenues91,406102,609
Total cost of revenues$ 146,306$ 162,535
Total cost of revenues as a percentage of revenues42 %40 %
+ +Cost of revenues increased \$16.2 billion from 2024 to 2025 due to an increase in other cost of revenues and TAC of \$11.2 billion and \$5.0 billion, respectively. + +The increase in TAC from 2024 to 2025 was largely due to an increase in TAC paid to distribution partners, primarily driven by growth in revenues subject to TAC. The TAC rate decreased from 20.7% to 20.3% from 2024 to 2025, primarily due to a revenue mix shift from Google Network properties to Google Search & other properties. The TAC rates on Google Search & other and Google Network revenues were substantially consistent from 2024 to 2025. + +The increase in other cost of revenues from 2024 to 2025 was primarily due to increases in content acquisition costs, largely for YouTube, depreciation expense, and other technical infrastructure operations costs. + +## Research and Development + +The following table presents research and development expenses (in millions, except percentages): + +
Year Ended December 31,
20242025
Research and development expenses$ 49,326$ 61,087
Research and development expenses as a percentage of revenues14 %15 %
+ +Research and development expenses increased \$11.8 billion from 2024 to 2025, primarily driven by increases in employee compensation expenses of \$6.9 billion and depreciation expense of \$2.4 billion. The increase in employee compensation expenses was primarily driven by an increase in SBC expenses of \$4.2 billion, which included an increase in a valuation-based compensation charge related to Waymo. + +## Sales and Marketing + +The following table presents sales and marketing expenses (in millions, except percentages): + +
Year Ended December 31,
20242025
Sales and marketing expenses$ 27,808$ 28,693
Sales and marketing expenses as a percentage of revenues8 %7 %
+ +Sales and marketing expenses increased \$885 million from 2024 to 2025, primarily driven by an increase in advertising and promotional activities of \$1.2 billion, partially offset by a decrease in employee compensation expenses of \$214 million. + +## General and Administrative + +The following table presents general and administrative expenses (in millions, except percentages): + +
Year Ended December 31,
20242025
General and administrative expenses$ 14,188$ 21,482
General and administrative expenses as a percentage of revenues4 %5 %
+ +General and administrative expenses increased \$7.3 billion from 2024 to 2025, primarily driven by an increase in expenses related to legal and other matters of \$6.2 billion, largely the result of the \$3.5 billion EC fine accrued in the third quarter of 2025 and a \$1.4 billion legal accrual made in the second quarter of 2025. + +## Segment Profitability + +We report our segment results as Google Services, Google Cloud, and Other Bets. Additionally, certain costs are not allocated to our segments because they represent Alphabet-level activities. For further details on our segments, see Part I, Item 1 Business and Note 15 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +The following table presents segment operating income (loss) (in millions). + +
Year Ended December 31,
20242025
Operating income (loss):
Google Services$ 121,263$ 139,404
Google Cloud6,11213,910
Other Bets(4,444)(7,515)
Alphabet-level activities(1)(10,541)(16,760)
Total income from operations$ 112,390$ 129,039
+ +(1) Alphabet-level activities primarily reflect expenses related to our shared AI research and development. + +## Google Services + +Google Services operating income increased \$18.1 billion from 2024 to 2025. The increase in operating income was primarily driven by an increase in revenues, partially offset by an increase in expenses related to legal and other matters, TAC, and content acquisition costs. + +## Google Cloud + +Google Cloud operating income increased \$7.8 billion from 2024 to 2025. The increase in operating income was primarily driven by an increase in revenues, partially offset by increases in usage costs for technical infrastructure and employee compensation expenses. + +## Other Bets + +Other Bets operating loss increased \$3.1 billion from 2024 to 2025. The increase in operating loss was primarily driven by an increase in employee compensation expenses largely due to an increase in a valuation-based compensation charge related to Waymo. + +## Other Income (Expense), Net + +The following table presents OI&E, (in millions): + +
Year Ended December 31,
20242025
Interest income$ 4,482$ 4,337
Interest expense(268)(736)
Foreign currency exchange gain (loss), net(409)(382)
Gain (loss) on debt securities, net(1,043)540
Gain (loss) on equity securities, net3,71424,080
Income (loss) and impairment from equity method investments, net(188)281
Other1,1371,667
Other income (expense), net$ 7,425$ 29,787
+ +OI&E, net increased \$22.4 billion from 2024 to 2025, primarily due to increases in net unrealized gains on equity securities resulting from fair value adjustments on non-marketable equity securities. + +For additional information, see Note 3 and Note 7 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +## Provision for Income Taxes + +The following table presents provision for income taxes (in millions, except effective tax rate): + +
Year Ended December 31,
20242025
Income before provision for income taxes$ 119,815$ 158,826
Provision for income taxes$ 19,697$ 26,656
Effective tax rate16.4 %16.8 %
+ +The effective tax rate increased from 2024 to 2025. This increase was primarily due to a decrease in the US Federal Foreign Derived Intangible Income tax deduction, a non-deductible EC fine and legal settlement in the US, partially offset by changes in prior period tax positions. + +Changes to US tax law enacted on July 4, 2025, allow for immediate expensing of domestic research and experimentation costs, accelerated depreciation on eligible capital expenditures, and other tax law changes impacting 2025 with certain changes effective in 2026. These changes are reflected in our results for the year ended December 31, 2025. + +The OECD is coordinating negotiations among more than 140 countries with the goal of achieving consensus around substantial changes to international tax policies, including the implementation of a minimum global effective tax rate of 15%. Some countries have already implemented the legislation effective January 1, 2024. This did not have a material effect on our income tax provision for the 2025 fiscal year. + +In January 2026, the OECD introduced new guidance including a "Side-by-Side Safe Harbor" which, if elected, exempts U.S. domestic operations from being taxed by global minimum tax rules. However, it does not exempt foreign subsidiaries from local minimum tax requirements if implemented. As more countries enact these global minimum tax rules, our effective tax rate and cash tax payments could increase. + +## Financial Condition + +## Cash, Cash Equivalents, and Marketable Securities + +As of December 31, 2025, we had \$126.8 billion in cash, cash equivalents, and short-term marketable securities. Cash equivalents and marketable securities are comprised of time deposits, money market funds, highly liquid government bonds, corporate debt securities, mortgage-backed and asset-backed securities, and marketable equity securities. + +## Sources, Uses of Cash and Related Trends + +Our principal sources of liquidity are cash, cash equivalents, and marketable securities, as well as the cash flow that we generate from operations. The primary use of capital continues to be to invest for the long-term growth of the business. We regularly evaluate our cash and capital structure, including the size, pace, and form of capital return to stockholders. + +The following table presents cash flows (in millions): + +
Year Ended December 31,
20242025
Net cash provided by operating activities$ 125,299$ 164,713
Net cash used in investing activities$ (45,536)$ (120,291)
Net cash used in financing activities$ (79,733)$ (37,388)
+ +## Cash Provided by Operating Activities + +Our largest source of cash provided by operations are advertising revenues generated by Google Search & other properties, YouTube properties, and Google Network properties. In Google Services, we also generate cash through consumer subscriptions, the sale of apps and in-app purchases, and devices. In Google Cloud, we generate cash through consumption-based fees and subscriptions for infrastructure, platform, applications, and other cloud services. + +Our primary uses of cash from operating activities include payments to distribution and Google Network partners, to employees for compensation, and to content providers. Other uses of cash from operating activities include payments to suppliers for devices, to tax authorities for income taxes, and other general corporate expenditures. + +Net cash provided by operating activities increased from 2024 to 2025 due to an increase in cash received from customers, partially offset by an increase in cash payments for cost of revenues and operating expenses. + +## Cash Used in Investing Activities + +Cash provided by investing activities consists primarily of maturities and sales of investments in marketable and non-marketable securities. Cash used in investing activities consists primarily of purchases of marketable and non-marketable securities, purchases of property and equipment, and payments for acquisitions. + +Net cash used in investing activities increased from 2024 to 2025, primarily due to an increase in purchases of property and equipment, driven by investments in technical infrastructure, and a decrease in maturities and sales of marketable securities. + +## Cash Used in Financing Activities + +Cash provided by financing activities consists primarily of proceeds from issuance of debt and proceeds from the sale of interests in consolidated entities. Cash used in financing activities consists primarily of repurchases of stock, repayments of debt, net payments related to stock-based award activities, and dividend payments. + +Net cash used in financing activities decreased from 2024 to 2025 due to an increase in proceeds from issuance of debt and a decrease in repurchases of stock, partially offset by repayments of debt. + +## Liquidity and Material Cash Requirements + +We expect existing cash, cash equivalents, short-term marketable securities, and cash flows from operations and financing activities to continue to be sufficient to fund our operating activities and cash commitments for investing and financing activities for at least the next 12 months, and thereafter for the foreseeable future. + +## Capital Expenditures and Leases + +We make investments in land, buildings, and servers and network equipment through purchases of property and equipment and lease arrangements to provide capacity for the growth of our services and products. + +## Capital Expenditures + +Our capital investments in property and equipment consist primarily of the following major categories: + +- technical infrastructure, which consists of our investments in servers and network equipment, data center land, and building construction and improvements; and +• office facilities, ground-up development projects, and building improvements. + +Assets not yet in service are those that are not ready for their intended use, including assets in the process of construction or assembly, and consist primarily of technical infrastructure. The time frame from date of purchase to placement in service of these assets may extend from months to years. For example, our data center construction projects are generally multi-year projects with multiple phases, where we acquire land and buildings, construct buildings, and secure and install servers and network equipment. + +During the years ended December 31, 2024 and 2025, we spent \$52.5 billion and \$91.4 billion on capital expenditures, respectively. In 2026, we expect to significantly increase, relative to 2025, our investment in our technical infrastructure, including servers and network equipment, and data centers. Depreciation of our property and equipment commences when such assets are ready for their intended use. For the years ended December 31, 2024 and 2025, our depreciation on property and equipment was \$15.3 billion and \$21.1 billion, respectively. + +## Leases + +As of December 31, 2025, the amount of total undiscounted future lease payments under operating leases was \$18.3 billion, of which \$3.3 billion is short-term, and total undiscounted future lease payments under finance leases was \$2.9 billion, of which \$491 million is short-term. + +As of December 31, 2025, we have entered into leases primarily related to data centers that have not yet commenced with short-term and long-term future lease payments of \$5.8 billion and \$52.7 billion, respectively. These leases will commence between 2026 and 2031 with non-cancelable lease terms primarily between one and 25 years. + +In January 2026, we executed a power purchase agreement which we expect to be accounted for as a lease resulting in future payments depending on certain agreement terms of \$9.9 billion between 2027 and 2047. If certain contractual conditions for the project are not met, we would instead make a one-time payment of approximately \$3.5 billion and assume ownership of the power generating assets. + +For additional information on leases, see Note 4 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +## Financing + +As of December 31, 2025, we had senior unsecured notes outstanding with a total carrying value of \$48.5 billion, of which \$2.0 billion was short-term. The associated short-term and long-term future interest payments were \$1.8 billion and \$35.7 billion, respectively. + +During 2025, we issued \$22.5 billion of US dollar-denominated senior unsecured notes and €13.25 billion of euro-denominated senior unsecured notes for general corporate purposes, comprised of the following: + +- May 2025: We issued \$5.0 billion of US dollar-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 4.89%, and a weighted-average maturity of approximately 24 years. We also issued €6.75 billion of euro-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 3.31%, and a weighted-average maturity of approximately 14 years. +- November 2025: We issued \$500 million of US dollar-denominated floating-rate senior unsecured notes and \$17.0 billion of US dollar-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 4.92% and a weighted-average maturity of approximately 20 years. We also issued €6.5 billion of euro-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 3.44% and a weighted-average maturity of approximately 16 years. + +As of December 31, 2025, we had \$10.0 billion of revolving credit facilities, \$4.0 billion expiring in April 2026 and \$6.0 billion expiring in April 2030. No amounts have been borrowed under the credit facilities. We also have a commercial paper program of up to \$25.0 billion, which is used for general corporate purposes. As of December 31, 2025, we had no commercial paper outstanding. + +For additional information, see Note 6 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +We use contract manufacturers for our technical infrastructure and device assembly and may supply them with components purchased directly from suppliers. Certain of these arrangements result in a portion of the cash received from and paid to contract manufacturers to be presented as financing activities on the Consolidated Statements of Cash Flows included in Item 8 of this Annual Report on Form 10-K. + +## Share Repurchase Program + +During 2025, we repurchased and subsequently retired 240 million shares for \$45.4 billion. + +In April 2024, the company's Board of Directors authorized a \$70.0 billion share repurchase program for its Class A and Class C shares. In April 2025, the company's Board of Directors authorized an additional \$70.0 billion share repurchase program for its Class A and Class C shares. As of December 31, 2025, \$69.5 billion remained available for Class A and Class C share repurchases. + +For additional information, see Note 11 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +## Dividend Program + +During the year ended December 31, 2025, total cash dividends were \$4.8 billion for Class A, \$703 million for Class B, and \$4.5 billion for Class C shares, respectively. + +In April 2025, the company's Board of Directors increased the quarterly cash dividend by 5% to \$0.21 per share of outstanding Class A, Class B, and Class C shares. + +The company has declared a quarterly cash dividend in the current quarter, and intends to pay quarterly cash dividends in the future, subject to review and approval by the company's Board of Directors in its sole discretion. + +## Accrued Legal and Regulatory + +As of December 31, 2025, we had short-term accrued legal and regulatory fines and settlements of \$15.6 billion. This amount primarily included EC fines, in addition to accruals related to other legal matters and regulatory fines and settlements. For additional information, see Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +## Taxes + +As of December 31, 2025, we had long-term income taxes payable of \$9.5 billion primarily related to unrecognized tax benefits. The timing and amount of any payment related to these unrecognized tax benefits are uncertain and cannot be estimated. + +## Purchase Commitments and Other Contractual Obligations + +We have material purchase commitments and other contractual obligations primarily related to energy take-or-pay contracts, licenses (including content licenses), and technical infrastructure and inventory orders. As of December 31, 2025, the total for these commitments was \$149.1 billion, of which \$113.0 billion was short-term, mostly related to technical infrastructure and inventory orders. These amounts reflect commitments and obligations through open purchase orders as well as the non-cancelable portion or the minimum cancellation fee in certain agreements. For those agreements with variable terms, we do not estimate the non-cancelable obligation beyond any minimum quantities and/or pricing as of December 31, 2025. In certain instances, the amount of our contractual obligations may change based on the expected timing of order fulfillment from our suppliers. For additional information related to our content licenses, see Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +As of December 31, 2025, we provided backstops in the form of financial guarantees and credit derivatives with maximum potential amount of future payments of \$5.7 billion and \$16.9 billion, respectively. For additional information on credit derivatives and financial guarantees, see Note 3 and Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +In addition, we regularly enter into multi-year, non-cancellable power purchase agreements with third-party suppliers that do not include a minimum dollar commitment. The amounts to be paid under these agreements are based on the actual volumes to be generated and are not readily determinable. + +We may experience increases in the costs associated with our purchase commitments and other contractual obligations as a result of ongoing developments surrounding international trade. For details on risks related to our manufacturing and supply chain and other risks, refer to Part 1, Item 1A, "Risk Factors" of this Annual Report on Form 10-K. + +## Pending Acquisitions + +In March 2025, we entered into a definitive agreement to acquire Wiz, Inc. ("Wiz"), a leading cloud security platform, for \$32.0 billion, subject to closing adjustments, in an all-cash transaction. The acquisition of Wiz is expected to close in 2026, subject to customary closing conditions, including the receipt of regulatory approvals. + +In December 2025, we entered into a definitive agreement to acquire Intersect, which provides data center and energy infrastructure solutions, for \$4.8 billion in cash, plus the assumption of debt. The acquisition of Intersect is expected to close in the first half of 2026, subject to customary closing conditions. + +For additional information, see Note 8 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +## Critical Accounting Estimates + +We prepare our consolidated financial statements in accordance with GAAP. In doing so, we have to make estimates and assumptions. Our critical accounting estimates are those estimates that involve a significant level of uncertainty at the time the estimate was made, and changes in them have had or are reasonably likely to have a material effect on our financial condition or results of operations. Accordingly, actual results could differ materially from our estimates. We base our estimates on past experience and other assumptions that we believe are reasonable under the circumstances, and we evaluate these estimates on an ongoing basis. We have reviewed our critical accounting estimates with the Audit Committee of our Board of Directors. + +For a summary of significant accounting policies and the effect on our financial statements, see Note 1 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +## Fair Value Measurements of Non-Marketable Equity Securities + +We measure certain financial instruments at fair value on a nonrecurring basis, consisting primarily of our non-marketable equity securities. These investments are accounted for under the measurement alternative method ("the measurement alternative") and are measured at cost, less impairment, subject to upward and downward adjustments resulting from observable price changes for identical or similar investments of the same issuer. These adjustments require quantitative assessments of the fair value of our securities, which may require the use of unobservable inputs. + +Adjustments are determined primarily based on a market approach as of the transaction date and involve the use of estimates using the best information available, which may include cash flow projections or other available market data. + +Non-marketable equity securities are also evaluated for impairment, based on qualitative factors including the companies' financial and liquidity position and access to capital resources, among others. When indicators of impairment exist, we prepare quantitative measurements of the fair value of our equity investments using a market approach or an income approach, which requires judgment and the use of unobservable inputs, including discount rates, investee revenues and costs, and comparable market data of private and public companies, among others. When the quantitative remeasurements of fair value indicate an impairment exists, we write down the investment to its current fair value. + +## Property and Equipment + +We assess the reasonableness of the useful lives of our property and equipment periodically or when events indicate a change is necessary. To determine the useful lives of our technical infrastructure, we rely on multiple inputs, including historical asset performance, expected technology advancements, and our future infrastructure deployment plans. Any change in the estimated useful lives is recognized on a prospective basis. + +## Income Taxes + +We are subject to income taxes in the US and foreign jurisdictions. Significant judgment is required in evaluating our uncertain tax positions and determining our provision for income taxes. + +Recording an uncertain tax position involves various qualitative considerations, including evaluation of comparable and resolved tax exposures, applicability of tax laws, and likelihood of settlement. We evaluate uncertain tax positions periodically, considering changes in facts and circumstances, such as new regulations or recent judicial opinions, as well as the status of audit activities by taxing authorities. Although we believe we have adequately reserved for our uncertain tax positions, no assurance can be given that the final tax outcome of these matters will not be different. To the extent that the final tax outcome of these matters is different than the amounts recorded, such differences will affect the provision for income taxes and the effective tax rate in the period in which such determination is made. + +The provision for income taxes includes the effect of reserve provisions and changes to reserves as well as the related net interest and penalties. In addition, we are subject to the continuous examination of our income tax returns by the Internal Revenue Service (IRS) and other tax authorities which may assert assessments against us. We regularly assess the likelihood of adverse outcomes resulting from these examinations and assessments to determine the adequacy of our provision for income taxes. + +## Loss Contingencies + +We are subject to claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders involving competition, intellectual property, data privacy and security, tax and related compliance, labor and employment, commercial disputes, content generated by our users, goods and services offered by advertisers or publishers using our platforms, personal injury, consumer protection, and other matters. Certain of these matters include speculative claims for substantial or indeterminate amounts of damages. We record a liability when we believe that it is probable that a loss has been incurred and the amount can be reasonably estimated. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the possible loss in Note 10 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +We evaluate, on a regular basis, developments in our legal matters that could affect the amount of liability that has been previously accrued, and the matters and related reasonably possible losses disclosed, and make adjustments as necessary. Significant judgment is required to determine both the likelihood and the estimated amount of a loss related to such matters. Until the final resolution of such matters, there may be an exposure to loss in excess of the amount recorded, and such amounts could be material. + +## ITEM 7A. QUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK + +We are exposed to financial market risks, including changes in foreign currency exchange rates, interest rates, and equity investment risks. + +## Foreign Currency Exchange Risk + +We transact business globally in multiple currencies. International revenues, foreign-denominated monetary assets and liabilities, and investments in foreign subsidiaries expose us to the risk of fluctuations in foreign exchange rates against the US dollar. Principal currency exposures include the Australian dollar, British pound, Canadian dollar, Euro, and Japanese yen. + +We monitor our foreign currency exposures and hedge foreign exchange risks with derivative and non-derivative instruments, including forwards, options (including collars), cross-currency swaps, and foreign currency-denominated debt. Gains or losses on these foreign currency exposures are generally offset by corresponding gains or losses on the derivative and non-derivative instruments. + +Considering historical trends in foreign exchange rates, we determined that it was reasonably possible that adverse changes in exchange rates of 10% could be experienced. We performed a sensitivity analysis on our foreign currency exposures to estimate the potential impact of this adverse 10% change. The estimated effects on our financial position would be as follows (in millions): + +
ImpactAs of December 31,
20242025
Foreign currency risk
Foreign denominated monetary assets and liabilities(1)OI&E$ 135$ 671
Cash flow hedges of foreign currency revenue(2)AOCI$ 1,627$ 2,096
Net investment hedges of investments in foreign subsidiaries(3)AOCI$ 660$ 2,942
+ +(1) After consideration of the effect of derivative contracts. +(2) The change in accumulated other comprehensive income (AOCI) would be expected to offset a corresponding foreign currency change in forecasted hedged revenues when recognized. +(3) The change in AOCI would be expected to offset a corresponding foreign currency translation gain or loss from our investments in foreign subsidiaries. + +## Interest Rate Risk + +We are exposed to interest rate risk related to our investment portfolio and outstanding debt. + +Our Corporate Treasury investment strategy is to achieve a return that will allow us to preserve capital and maintain liquidity. By policy, we limit the amount of credit exposure within our investment portfolio to any one issuer. Our investments in both fixed rate and floating rate interest earning securities carry a degree of interest rate risk. Fixed rate securities may have their fair market value adversely affected due to a rise in interest rates, while floating rate securities may produce less income than predicted if interest rates fall. Unrealized gains and losses on our marketable debt securities are primarily due to interest rate fluctuations as compared to interest rates at the time of purchase. For certain fixed and floating rate debt securities, we have elected the fair value option for which changes in fair value are recorded in OI&E. We measure securities for which we have not elected the fair value option at fair value with gains and losses recorded in AOCI until the securities are sold, less any expected credit losses. + +We use value-at-risk (VaR) analysis to determine the potential effect of fluctuations in interest rates on the value of our investment portfolio. The VaR is the expected loss in fair value, for a given confidence interval, for our investment portfolio due to adverse movements in interest rates. We use a variance/covariance VaR model with 95% confidence interval. The estimated one-day loss in fair value of our investment portfolio as of December 31, 2024 and 2025 are shown below (in millions): + +
As of December 31,12-Month Average As of December 31,
2024202520242025
Risk category - interest rate$ 208$ 162$ 230$ 184
+ +Actual future gains and losses associated with our investment portfolio may differ materially from the sensitivity analyses performed as of December 31, 2024 and 2025 due to the inherent limitations associated with predicting the timing and amount of changes in interest rates and our actual exposures and positions. VaR analysis is not intended to represent actual losses but is used as a risk estimation. + +Additionally, we had senior unsecured notes outstanding with a total carrying value of \$11.9 billion and \$48.5 billion as of December 31, 2024 and 2025, respectively. As our senior unsecured notes primarily bear interest at fixed rates and are recorded at amortized cost, interest rate fluctuations generally do not affect our consolidated financial statements. However, the fair value of the notes will fluctuate with movement in market interest rates. + +## Equity Investment Risk + +Our marketable and non-marketable equity securities are subject to a wide variety of market-related risks that could substantially reduce or increase the fair value of our holdings. + +Our marketable equity securities are primarily publicly traded stocks or funds and our non-marketable equity securities are primarily investments in privately held companies, some of which are in the startup or development stages. + +We record marketable equity securities at fair value subject to market price volatility. These securities represent \$5.1 billion and \$6.3 billion of our investments as of December 31, 2024 and 2025, respectively. A hypothetical adverse price change of 10% on our December 31, 2025 balance would decrease the fair value of marketable equity securities by \$631 million. From time to time, we may enter into derivatives to hedge the market price risk on certain of our marketable equity securities. + +Our non-marketable equity securities not accounted for under the equity method are primarily adjusted to fair value for observable transactions for identical or similar investments of the same issuer or impairment (referred to as the measurement alternative). The fair value measured at the time of the observable transaction is not necessarily an indication of the current fair value as of the balance sheet date. These investments, especially those that are in the early stages, are inherently risky because the technologies or products these companies have under development are typically in the early phases and may never materialize, and they may experience a decline in financial condition, which could result in a loss of a substantial part of our investment in these companies. Valuations of our equity investments in private companies are inherently more complex due to the lack of readily available market data and observable transactions at lower valuations could result in significant losses. In addition, global economic conditions could result in additional volatility. The success of our investment in any private company is also typically dependent on the likelihood of our ability to realize appreciation in the value of investments through liquidity events such as public offerings, acquisitions, private sales, or other market events. Changes in the valuation of non-marketable equity securities may not directly correlate with changes in valuation of marketable equity securities. As of December 31, 2024 and 2025, the carrying value of our non-marketable equity securities, which were accounted for under the measurement alternative, was \$35.2 billion and \$64.1 billion, respectively. + +The carrying values of our equity method investments, which totaled approximately \$2.0 billion and \$2.5 billion as of December 31, 2024 and 2025, respectively, generally do not fluctuate based on market price changes. However, these investments could be impaired if the carrying value exceeds the fair value and is not expected to recover. + +For additional information about our equity investments, see Note 1 and Note 3 of the Notes to Consolidated Financial Statements included in Item 8 of this Annual Report on Form 10-K. + +## ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA + +## Alphabet Inc. + +## INDEX TO CONSOLIDATED FINANCIAL STATEMENTS + +
Page
Reports of Independent Registered Public Accounting Firm (PCAOB ID: 42)45
Financial Statements:
Consolidated Balance Sheets48
Consolidated Statements of Income49
Consolidated Statements of Comprehensive Income50
Consolidated Statements of Stockholders’ Equity51
Consolidated Statements of Cash Flows52
Notes to Consolidated Financial Statements53
+ +# REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM + +To the Stockholders and the Board of Directors of Alphabet Inc. + +## Opinion on the Financial Statements + +We have audited the accompanying consolidated balance sheets of Alphabet Inc. (the Company) as of December 31, 2024 and 2025, the related consolidated statements of income, comprehensive income, stockholders' equity and cash flows for each of the three years in the period ended December 31, 2025, and the related notes and financial statement schedule listed in the Index at Item 15 (collectively referred to as the “consolidated financial statements”). In our opinion, the consolidated financial statements present fairly, in all material respects, the financial position of the Company at December 31, 2024 and 2025, and the results of its operations and its cash flows for each of the three years in the period ended December 31, 2025, in conformity with U.S. generally accepted accounting principles. + +We also have audited, in accordance with the standards of the Public Company Accounting Oversight Board (United States) (PCAOB), the Company's internal control over financial reporting as of December 31, 2025, based on criteria established in Internal Control—Integrated Framework issued by the Committee of Sponsoring Organizations of the Treadway Commission (2013 framework), and our report dated February 4, 2026 expressed an unqualified opinion thereon. + +## Basis for Opinion + +These financial statements are the responsibility of the Company's management. Our responsibility is to express an opinion on the Company's financial statements based on our audits. We are a public accounting firm registered with the PCAOB and are required to be independent with respect to the Company in accordance with the U.S. federal securities laws and the applicable rules and regulations of the Securities and Exchange Commission and the PCAOB. + +We conducted our audits in accordance with the standards of the PCAOB. Those standards require that we plan and perform the audit to obtain reasonable assurance about whether the financial statements are free of material misstatement, whether due to error or fraud. Our audits included performing procedures to assess the risks of material misstatement of the financial statements, whether due to error or fraud, and performing procedures that respond to those risks. Such procedures included examining, on a test basis, evidence regarding the amounts and disclosures in the financial statements. Our audits also included evaluating the accounting principles used and significant estimates made by management, as well as evaluating the overall presentation of the financial statements. We believe that our audits provide a reasonable basis for our opinion. + +## Critical Audit Matter + +The critical audit matter communicated below is a matter arising from the current period audit of the financial statements that was communicated or required to be communicated to the audit committee and that: (1) relates to accounts or disclosures that are material to the financial statements and (2) involved our especially challenging, subjective or complex judgments. The communication of the critical audit matter does not alter in any way our opinion on the consolidated financial statements, taken as a whole, and we are not, by communicating the critical audit matter below, providing a separate opinion on the critical audit matter or on the accounts or disclosures to which it relates. + +Loss Contingencies + +
Description of the MatterThe Company is subject to claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders. As described in Note 10 to the consolidated financial statements, such claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders could result in adverse consequences.Significant judgment is required to determine both the likelihood and the estimated amount of a loss related to such matters. Auditing management’s accounting for and disclosure of loss contingencies from these matters involved challenging and subjective auditor judgment in assessing the Company’s evaluation of the probability of a loss, and the estimated amount or range of loss.
How We Addressed the Matter in Our AuditWe tested relevant controls over the identified risks associated with management’s accounting for and disclosure of these matters. This included controls over management’s assessment of the probability of incurrence of a loss and whether the loss or range of loss was reasonably estimable and the development of related disclosures.Our audit procedures included, among others, gaining an understanding of previous rulings and the status of ongoing lawsuits, reviewing letters from internal and external legal counsel addressing the matters, meeting with internal legal counsel to discuss the allegations, and obtaining a representation letter from management on these matters. We also evaluated the Company’s disclosures in relation to these matters.
+ +/s/ Ernst & Young LLP + +We have served as the Company's auditor since 1999. + +San Jose, California + +February 4, 2026 + +# REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM + +To the Stockholders and the Board of Directors of Alphabet Inc. + +## Opinion on Internal Control Over Financial Reporting + +We have audited Alphabet Inc.'s internal control over financial reporting as of December 31, 2025, based on criteria established in Internal Control—Integrated Framework issued by the Committee of Sponsoring Organizations of the Treadway Commission (2013 framework) (the COSO criteria). In our opinion, Alphabet Inc. (the Company) maintained, in all material respects, effective internal control over financial reporting as of December 31, 2025, based on the COSO criteria. + +We also have audited, in accordance with the standards of the Public Company Accounting Oversight Board (United States) (PCAOB), the 2025 consolidated financial statements of the Company and our report dated February 4, 2026 expressed an unqualified opinion thereon. + +## Basis for Opinion + +The Company's management is responsible for maintaining effective internal control over financial reporting and for its assessment of the effectiveness of internal control over financial reporting included in the accompanying Management's Report on Internal Control over Financial Reporting. Our responsibility is to express an opinion on the Company's internal control over financial reporting based on our audit. We are a public accounting firm registered with the PCAOB and are required to be independent with respect to the Company in accordance with the U.S. federal securities laws and the applicable rules and regulations of the Securities and Exchange Commission and the PCAOB. + +We conducted our audit in accordance with the standards of the PCAOB. Those standards require that we plan and perform the audit to obtain reasonable assurance about whether effective internal control over financial reporting was maintained in all material respects. + +Our audit included obtaining an understanding of internal control over financial reporting, assessing the risk that a material weakness exists, testing and evaluating the design and operating effectiveness of internal control based on the assessed risk, and performing such other procedures as we considered necessary in the circumstances. We believe that our audit provides a reasonable basis for our opinion. + +## Definition and Limitations of Internal Control Over Financial Reporting + +A company's internal control over financial reporting is a process designed to provide reasonable assurance regarding the reliability of financial reporting and the preparation of financial statements for external purposes in accordance with generally accepted accounting principles. A company's internal control over financial reporting includes those policies and procedures that (1) pertain to the maintenance of records that, in reasonable detail, accurately and fairly reflect the transactions and dispositions of the assets of the company; (2) provide reasonable assurance that transactions are recorded as necessary to permit preparation of financial statements in accordance with generally accepted accounting principles, and that receipts and expenditures of the company are being made only in accordance with authorizations of management and directors of the company; and (3) provide reasonable assurance regarding prevention or timely detection of unauthorized acquisition, use, or disposition of the company's assets that could have a material effect on the financial statements. + +Because of its inherent limitations, internal control over financial reporting may not prevent or detect misstatements. Also, projections of any evaluation of effectiveness to future periods are subject to the risk that controls may become inadequate because of changes in conditions, or that the degree of compliance with the policies or procedures may deteriorate. + +/s/ Ernst & Young LLP + +San Jose, California + +February 4, 2026 + +Alphabet Inc. +CONSOLIDATED BALANCE SHEETS +(in millions, except par value per share amounts) + +
As of December 31,
20242025
Assets
Current assets:
Cash and cash equivalents$ 23,466$ 30,708
Marketable securities72,19196,135
Total cash, cash equivalents, and marketable securities95,657126,843
Accounts receivable, net52,34062,886
Other current assets15,71416,309
Total current assets163,711206,038
Non-marketable securities37,98268,687
Deferred income taxes17,1809,113
Property and equipment, net171,036246,597
Operating lease assets13,58815,221
Goodwill31,88533,380
Other non-current assets14,87416,245
Total assets$ 450,256$ 595,281
Liabilities and Stockholders’ Equity
Current liabilities:
Accounts payable$ 7,987$ 12,200
Accrued compensation and benefits15,06917,546
Accrued expenses and other current liabilities51,22855,557
Accrued revenue share9,80210,864
Deferred revenue5,0366,578
Total current liabilities89,122102,745
Long-term debt10,88346,547
Income taxes payable, non-current8,7829,531
Operating lease liabilities11,69112,744
Other long-term liabilities4,6948,449
Total liabilities125,172180,016
Commitments and Contingencies (Note 10)
Stockholders’ equity:
Preferred stock, $0.001 par value per share, 100 shares authorized; no shares issued and outstanding00
Class A, Class B, and Class C stock and additional paid-in capital, $0.001 par value per share: 300,000 shares authorized (Class A 180,000, Class B 60,000, Class C 60,000); 12,211 (Class A 5,835, Class B 861, Class C 5,515) and 12,088 (Class A 5,822, Class B 837, Class C 5,429) shares issued and outstanding84,80093,126
Accumulated other comprehensive income (loss)(4,800)(1,916)
Retained earnings245,084324,055
Total stockholders’ equity325,084415,265
Total liabilities and stockholders’ equity$ 450,256$ 595,281
+ +See accompanying notes. + +Alphabet Inc. +CONSOLIDATED STATEMENTS OF INCOME +(in millions, except per share amounts) + +
Year Ended December 31,
202320242025
Revenues$ 307,394$ 350,018$ 402,836
Costs and expenses:
Cost of revenues133,332146,306162,535
Research and development45,42749,32661,087
Sales and marketing27,91727,80828,693
General and administrative16,42514,18821,482
Total costs and expenses223,101237,628273,797
Income from operations84,293112,390129,039
Other income (expense), net1,4247,42529,787
Income before income taxes85,717119,815158,826
Provision for income taxes11,92219,69726,656
Net income$ 73,795$ 100,118$ 132,170
Basic net income per share (Note 12)$ 5.84$ 8.13$ 10.91
Diluted net income per share (Note 12)$ 5.80$ 8.04$ 10.81
+ +See accompanying notes. + +Alphabet Inc. +CONSOLIDATED STATEMENTS OF COMPREHENSIVE INCOME +(in millions) + +
Year Ended December 31,
202320242025
Net income$ 73,795$ 100,118$ 132,170
Other comprehensive income (loss):
Change in foreign currency translation adjustment, net of income tax benefit (expense) of $63, $(49) and $180735(1,673)2,522
Available-for-sale investments:
Change in net unrealized gains (losses)1,344(116)1,146
Less: reclassification adjustment for net (gains) losses included in net income1,168782(169)
Net change, net of income tax benefit (expense) of $(698), $(190), and $(276)2,512666977
Cash flow hedges:
Change in net unrealized gains (losses)168775(779)
Less: reclassification adjustment for net (gains) losses included in net income(214)(166)164
Net change, net of income tax benefit (expense) of $2, $(151), and $174(46)609(615)
Other comprehensive income (loss)3,201(398)2,884
Comprehensive income$ 76,996$ 99,720$ 135,054
+ +See accompanying notes. + +Alphabet Inc. +CONSOLIDATED STATEMENTS OF STOCKHOLDERS' EQUITY (in millions) + +
Class A, Class B, Class C Stock and Additional Paid-In CapitalAccumulated Other Comprehensive Income (Loss)Retained EarningsTotal Stockholders' Equity
SharesAmount
Balance as of December 31, 202212,849$ 68,184$ (7,603)$ 195,563$ 256,144
Stock issued1390000
Stock-based compensation022,5780022,578
Tax withholding related to vesting of restricted stock units and other0(10,164)09(10,155)
Repurchases of stock(528)(4,064)0(58,120)(62,184)
Net income00073,79573,795
Other comprehensive income (loss)003,20103,201
Balance as of December 31, 202312,46076,534(4,402)211,247283,379
Stock issued1300000
Stock-based compensation022,9370022,937
Tax withholding related to vesting of restricted stock units and other0(12,507)0(16)(12,523)
Repurchases of stock(379)(3,359)0(58,688)(62,047)
Dividends and dividend equivalents declared ($0.60 per share)0410(7,577)(7,536)
Sale of interest in consolidated entities01,154001,154
Net income000100,118100,118
Other comprehensive income (loss)00(398)0(398)
Balance as of December 31, 202412,21184,800(4,800)245,084325,084
Stock issued1170000
Stock-based compensation025,1300025,130
Tax withholding related to vesting of restricted stock units and other0(14,842)00(14,842)
Repurchases of stock(240)(2,514)0(42,884)(45,398)
Dividends and dividend equivalents declared ($0.83 per share)01520(10,315)(10,163)
Sale of interest in consolidated entities040000400
Net income000132,170132,170
Other comprehensive income (loss)002,88402,884
Balance as of December 31, 202512,088$ 93,126$ (1,916)$ 324,055$ 415,265
+ +See accompanying notes. + +Alphabet Inc. +CONSOLIDATED STATEMENTS OF CASH FLOWS +(in millions) + +
Year Ended December 31,
202320242025
Operating activities
Net income$ 73,795$ 100,118$ 132,170
Adjustments:
Depreciation of property and equipment11,94615,31121,136
Stock-based compensation expense22,46022,78524,953
Deferred income taxes(7,763)(5,257)8,348
Loss (gain) on debt and equity securities, net823(2,671)(24,620)
Other4,3303,4192,108
Changes in assets and liabilities, net of effects of acquisitions:
Accounts receivable, net(7,833)(5,891)(8,779)
Income taxes, net523(2,418)(3,226)
Other assets(2,143)(1,397)(4,542)
Accounts payable664359907
Accrued expenses and other liabilities3,937(1,161)12,939
Accrued revenue share4821,059899
Deferred revenue5251,0432,420
Net cash provided by operating activities101,746125,299164,713
Investing activities
Purchases of property and equipment(32,251)(52,535)(91,447)
Purchases of marketable securities(77,858)(86,679)(103,773)
Maturities and sales of marketable securities86,672103,42883,240
Purchases of non-marketable securities(3,027)(5,034)(5,716)
Maturities and sales of non-marketable securities9478821,367
Acquisitions, net of cash acquired, and purchases of intangible assets(495)(2,931)(1,592)
Other investing activities(1,051)(2,667)(2,370)
Net cash used in investing activities(27,063)(45,536)(120,291)
Financing activities
Net payments related to stock-based award activities(9,837)(12,190)(14,167)
Repurchases of stock(61,504)(62,222)(45,709)
Dividend payments0(7,363)(10,049)
Proceeds from issuance of debt, net of costs10,79013,58964,564
Repayments of debt(11,550)(12,701)(32,427)
Proceeds from sale of interest in consolidated entities, net81,154400
Net cash used in financing activities(72,093)(79,733)(37,388)
Effect of exchange rate changes on cash and cash equivalents(421)(612)208
Net increase (decrease) in cash and cash equivalents2,169(582)7,242
Cash and cash equivalents at beginning of period21,87924,04823,466
Cash and cash equivalents at end of period$ 24,048$ 23,466$ 30,708
Supplemental disclosures of non-cash investing activities:
Purchases of property and equipment included in accrued liabilities and accounts payable$ 7,435$ 10,326$ 15,090
+ +See accompanying notes. + +# Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS + +## Note 1. Summary of Significant Accounting Policies + +## Nature of Operations + +Google was incorporated in California in September 1998 and re-incorporated in the State of Delaware in August 2003. In 2015, we implemented a holding company reorganization, and as a result, Alphabet Inc. ("Alphabet") became the successor issuer to Google. + +We generate revenues by delivering relevant, cost-effective online advertising; cloud-based solutions that provide enterprise customers of all sizes with infrastructure, platform services, and applications; and sales of other products and services, such as fees received for subscription-based products, apps and in-app purchases, and devices. + +## Basis of Consolidation + +The consolidated financial statements of Alphabet include the accounts of Alphabet and entities consolidated under the variable interest and voting models. Intercompany balances and transactions have been eliminated. + +## Use of Estimates + +Preparation of consolidated financial statements in conformity with GAAP requires us to make estimates and assumptions that affect the amounts reported and disclosed in the financial statements and the accompanying notes. Actual results could differ materially from these estimates due to uncertainties. On an ongoing basis, we evaluate our estimates, including those related to the allowance for credit losses; contingent liabilities; fair values of financial instruments and goodwill; income taxes; inventory; and useful lives of property and equipment, among others. We base our estimates on assumptions, both historical and forward looking, that are believed to be reasonable, and the results of which form the basis for making judgments about the carrying values of assets and liabilities. + +## Revenue Recognition + +Revenues are recognized when control of the promised goods or services is transferred to our customers, and the collectibility of an amount that we expect in exchange for those goods or services is probable. Sales and other similar taxes are excluded from revenues. + +## Google Advertising + +Google advertising revenues consist of revenues from: + +- Google Search and other properties, including revenues from traffic generated by search distribution partners who use Google.com as their default search in browsers, toolbars, etc. and other Google owned and operated properties like Gmail, Google Maps, and Google Play; +• YouTube properties; and +- Google Network properties, including revenues from Google Network properties participating in AdMob, AdSense, and Google Ad Manager. + +Our customers generally purchase advertising inventory through Google Ads, Google Ad Manager, Google Display & Video 360, and Google Marketing Platform, among others. + +We offer advertising by delivering both performance and brand advertising. We recognize revenues for performance advertising when a user engages with the advertisement. For brand advertising, we recognize revenues when the ad is displayed, or a user views the ad. + +For ads placed on Google Network properties, we evaluate whether we are the principal (i.e., report revenues on a gross basis) or agent (i.e., report revenues on a net basis). Generally, we report advertising revenues for ads placed on Google Network properties on a gross basis, that is, the amounts billed to our customers are recorded as revenues, and amounts paid to Google Network partners are recorded as cost of revenues. Where we are the principal, we control the advertising inventory before it is transferred to our customers. Our control is evidenced by our sole ability to monetize the advertising inventory before it is transferred to our customers and is further supported by us being primarily responsible to our customers and having a level of discretion in establishing pricing. + +## Google Subscriptions, Platforms, and Devices + +Google subscriptions, platforms, and devices revenues consist of revenues from: + +- consumer subscriptions, which primarily include revenues from YouTube services, such as YouTube TV, YouTube Music and Premium, and NFL Sunday Ticket, as well as Google One, which offers access to our most capable Gemini models; +- platforms, which primarily include revenues from Google Play sales of apps and in-app purchases; +• devices, which primarily include sales of the Pixel family of devices; and +• other products and services. + +Subscription revenues are recognized ratably over the period of the subscription, primarily monthly. We report revenues from Google Play sales of apps and in-app purchases on a net basis because our performance obligation is to facilitate a transaction between app developers and end users for which we earn a service fee. + +## Google Cloud + +Google Cloud revenues consist of revenues from: + +- Google Cloud Platform primarily generates consumption-based fees and subscriptions for infrastructure, platform, and other services. These services provide access to solutions such as AI offerings including our enterprise AI infrastructure, Vertex AI platform, and Gemini Enterprise; cybersecurity offerings; and data and analytics solutions; +- Google Workspace includes subscriptions for cloud-based communication and collaboration tools for enterprises, such as Gmail, Docs, Calendar, Drive, and Meet, with integrated features like Gemini for Google Workspace; and +• other enterprise services. + +Our cloud services are generally provided on either a consumption or subscription basis and may have contract terms longer than a year. Revenues related to cloud services provided on a consumption basis are recognized when the customer utilizes the services, based on the quantity of services consumed using the relative standalone selling price allocation. Revenues related to cloud services provided on a subscription basis are recognized ratably over the contract term as the customer receives and consumes the benefits of the cloud services. + +## Arrangements with Multiple Performance Obligations + +Our contracts with customers may include multiple performance obligations. For such arrangements, we allocate revenues to each performance obligation based on its relative standalone selling price. We generally determine standalone selling prices based on observable prices of our products and services sold or priced separately in comparable circumstances to similar customers. + +## Customer Incentives and Credits + +Certain customers receive cash-based incentives or credits, which are accounted for as variable consideration. We estimate these amounts based on the expected amount to be provided to customers and reduce revenues. We believe that there will not be significant changes to our estimates of variable consideration related to customer incentives and credits. + +## Sales Commissions + +We expense sales commissions when incurred when the period of the expected benefit is one year or less. We recognize an asset for certain sales commissions and amortize if the expected benefit period is greater than one year. These costs are recorded within sales and marketing expenses. + +## Cost of Revenues + +Cost of revenues consists of TAC and other costs of revenues. + +• TAC includes: + +amounts paid to our distribution partners who make available our search access points and other ad-supported services. Our distribution partners include browser providers, mobile carriers, original equipment manufacturers, and software developers; and +- amounts paid to Google Network partners primarily for ads displayed on their properties. + +• Other cost of revenues includes: + +content acquisition costs, which are payments to content providers from whom we license video and other content for distribution, primarily related to YouTube (we pay fees to these content providers based on revenues generated, subscriber counts, or a flat fee); +• depreciation expense, primarily related to our technical infrastructure; +• employee compensation expenses related to our technical infrastructure and other operations such as content review and customer and product support; +inventory and other costs related to the devices we sell; and +other technical infrastructure operations costs, including energy, equipment, and network capacity costs. + +## Software Development Costs + +We expense software development costs, including costs to develop software products or the software component of products to be sold, leased, or marketed to external users, before technological feasibility is reached. Technological feasibility is typically reached shortly before the release of such products. As a result, development costs that meet the criteria for capitalization were not material for the periods presented. + +Software development costs also include costs to develop software to be used solely to meet internal needs and cloud-based applications used to deliver our services. We capitalize development costs related to these software applications once the preliminary project stage is complete and it is probable that the project will be completed and the software will be used to perform the function intended. Costs capitalized for developing such software applications were not material for the periods presented. + +## Stock-Based Compensation + +Stock-based compensation (SBC) primarily consists of Alphabet restricted stock units (RSUs). RSUs are equity classified and measured at the fair market value of the underlying stock at the grant date. We recognize RSU expense using the straight-line attribution method over the requisite service period and account for forfeitures as they occur. RSUs are awarded dividend equivalents, which are subject to the same vesting conditions as the underlying award, and settled in Class C shares. + +For RSUs, shares are issued on the vesting dates net of the applicable statutory income tax withholding to be paid by us on behalf of our employees. As a result, fewer shares are issued than the number of RSUs vested, and the income tax withholding is recorded as a reduction to additional paid-in capital. + +Additionally, SBC includes other stock-based awards, such as performance stock units (PSUs) that include market conditions and awards that may be settled in cash or the stock of certain Other Bet companies. PSUs and certain awards granted by Other Bet companies are equity classified and expense is recognized over the requisite service period. Certain awards granted by Other Bet companies are liability classified and remeasured at fair value through settlement. The fair value of awards granted by Other Bet companies is based on the equity valuation of the respective Other Bet company. + +## Advertising and Promotional Expenses + +We expense advertising and promotional costs in the period in which they are incurred. For the years ended December 31, 2023, 2024, and 2025, advertising and promotional expenses totaled approximately \$8.7 billion, \$8.7 billion, and \$9.9 billion, respectively. + +## Fair Value Measurements + +Fair value is an exit price, representing the amount that would be received to sell an asset or paid to transfer a liability in an orderly transaction between market participants. Assets and liabilities recorded at fair value are measured and classified in accordance with a three-tier fair value hierarchy based on the observability of the inputs available in the market used to measure fair value: + +Level 1 - Observable inputs that reflect quoted prices (unadjusted) for identical assets or liabilities in active markets. + +Level 2 - Inputs that are based upon quoted prices for similar instruments in active markets, quoted prices for identical or similar instruments in markets that are not active, and model-based valuation techniques for which all significant inputs are observable in the market or can be derived from observable market data. Where applicable, these models project future cash flows and discount the future amounts to a present value using market-based observable inputs including interest rate curves, foreign exchange rates, and credit ratings. + +Level 3 - Unobservable inputs that are supported by little or no market activities. + +The fair value hierarchy requires an entity to maximize the use of observable inputs and minimize the use of unobservable inputs when measuring fair value. The determination of fair value involves the use of appropriate valuation methods and relevant inputs into valuation models. + +Our financial assets and liabilities that are measured at fair value on a recurring basis include cash equivalents, marketable securities, and derivative financial instruments. Our financial assets measured at fair value on a nonrecurring basis include non-marketable equity securities. Other financial assets and liabilities are carried at cost with fair value disclosed, if required. + +We measure certain other instruments, and certain assets and liabilities acquired in a business combination, also at fair value on a nonrecurring basis. + +## Financial Instruments + +Our financial instruments include cash, cash equivalents, marketable and non-marketable securities, derivative financial instruments, financial guarantees, accounts receivable, and convertible notes. + +## Credit Risks + +We are subject to concentration of credit risk primarily from cash equivalents, marketable debt securities, derivative financial instruments, including foreign exchange contracts, accounts receivable, and convertible notes. We manage the concentration of our credit risk exposure through timely assessment of our counterparty creditworthiness, credit limits, and use of collateral management. Foreign exchange contracts are transacted with various financial institutions with high credit standing. Accounts receivable are typically unsecured and are derived from revenues earned from customers located around the world. We manage the concentration of our credit risk exposure by performing ongoing evaluations to determine customer credit and we limit the amount of credit we extend. We generally do not require collateral from our customers. + +## Cash Equivalents + +We invest excess cash primarily in asset-backed and mortgage-backed securities, corporate debt securities, government bonds, money market funds, and time deposits. + +## Marketable Securities + +We classify all marketable debt securities that have effective maturities of three months or less from the date of purchase as cash equivalents and those with effective maturities of greater than three months as marketable securities. We determine the appropriate classification of our investments in marketable debt securities at the time of purchase and reevaluate such designation at each balance sheet date. We have classified and accounted for our marketable debt securities as available-for-sale. After consideration of our risk versus reward objectives, as well as our liquidity requirements, we may sell these debt securities prior to their effective maturities. As we view these securities as available to support current operations, we classify highly liquid securities with maturities beyond 12 months as current assets under the caption marketable securities. We carry these securities at fair value, and report the unrealized gains and losses, net of taxes, as a component of stockholders' equity, except for the changes in allowance for expected credit losses, which are recorded in OI&E. For certain marketable debt securities we have elected the fair value option, for which changes in fair value are recorded in OI&E. We determine any realized gains and losses on the sale of marketable debt securities on a specific identification method, and we record such gains and losses as a component of OI&E. + +Our investments in marketable equity securities are measured at fair value with the related gains and losses, including unrealized, recognized in OI&E. + +## Non-Marketable Securities + +Non-marketable securities primarily consist of equity securities. We account for non-marketable equity securities through which we exercise significant influence but do not have control over the investee under the equity method. Other non-marketable equity securities that we hold are primarily accounted for under the measurement alternative. Under the measurement alternative, the carrying value is measured at cost, less any impairment, plus or minus changes resulting from observable price changes in orderly transactions for identical or similar investments of the same issuer. Adjustments are determined primarily based on a market approach as of the transaction date and are recorded as a component of OI&E. + +Non-marketable securities that do not have effective contractual maturity dates are classified as other non-current assets. + +## Derivative Financial Instruments + +See Note 3 for the accounting policy pertaining to derivative financial instruments. + +## Financial Guarantees + +In certain arrangements, we provide reimbursements for costs incurred by third parties during power generation project development phases if specified trigger events occur. We recognize a noncontingent liability for the fair value of our obligation to stand ready to perform, reported in other long-term liabilities. We also recognize a contingent liability when it becomes probable that a payment will be required and the amount can be reasonably estimated. + +## Accounts Receivable + +Our payment terms for accounts receivable vary by the types and locations of our customers and the products or services offered. The term between invoicing and when payment is due is not significant. Additionally, accounts receivable includes amounts for services performed in advance of the right to invoice the customer. + +We maintain an allowance for credit losses for accounts receivable, which is recorded as an offset to accounts receivable, and changes in such are classified as general and administrative expense. We assess collectibility by reviewing accounts receivable on a collective basis where similar characteristics exist and on an individual basis when we identify specific customers with known disputes or collectibility issues. With respect to current accounts receivables, we elected to assume that current conditions as of the balance sheet date do not change for the remaining life of the asset. In determining the amount of the allowance for credit losses for those assets, we adjust historical loss information to reflect current market conditions and customer-specific information to the extent that historical loss information does not reflect current conditions. + +## Convertible Notes + +Our investments in convertible notes are primarily recorded at amortized cost which includes unpaid principal balances, deferred origination costs, and any related discount or premium, net of allowances for credit losses, and are included within other non-current assets. + +## Other + +Our financial instruments also include debt and equity investments in companies with which we also entered into commercial arrangements at or near the same time. For these transactions, judgment is required in assessing the substance of the arrangements, including assessing whether the components of the arrangements should be accounted for as separate transactions under the applicable GAAP, and determining the value of the components of the arrangements, including the fair value of the investments. Additionally, if our investment in such companies becomes impaired, we may need to re-evaluate the accounting for the commercial arrangement, including reducing any remaining performance obligations. + +## Impairment of Investments + +We periodically review our debt securities with unrealized gains and losses recorded as a component of stockholders' equity and non-marketable equity securities for impairment. + +For debt securities in an unrealized loss position, we determine whether a credit loss exists. The credit loss is estimated by considering available information relevant to the collectibility of the security and information about past events, current conditions, and reasonable and supportable forecasts. Any credit loss is recorded as a charge to OI&E, not to exceed the amount of the unrealized loss. Unrealized losses other than the credit loss are recognized in AOCI. If we have an intent to sell, or if it is more likely than not that we will be required to sell a debt security in an unrealized loss position before recovery of its amortized cost basis, we will write down the security to its fair value and record the corresponding charge as a component of OI&E. + +For non-marketable equity securities, including equity method investments, we consider whether impairment indicators exist by evaluating the companies' financial and liquidity position and access to capital resources, among other indicators. If the assessment indicates that the investment is impaired, we write down the investment to its fair value by recording the corresponding charge as a component of OI&E. We prepare quantitative measurements of the fair value of our equity investments using a market approach or an income approach. + +## Inventory + +Inventory consists primarily of finished goods and is stated at the lower of cost and net realizable value. Cost is generally computed using the first-in, first-out method. + +## Variable Interest Entities + +We determine at the inception of each arrangement whether an entity in which we have made an investment or in which we have other variable interests is considered a variable interest entity (VIE). We consolidate VIEs when we are the primary beneficiary. We are the primary beneficiary of a VIE when we have the power to direct activities that most significantly affect the economic performance of the VIE and have the obligation to absorb the majority of their losses or benefits. If we are not the primary beneficiary in a VIE, we account for the investment or other variable interests in a VIE in accordance with applicable GAAP. + +Periodically, we assess whether any changes in our interest or relationship with the entity affect our determination of whether the entity is a VIE and, if so, whether we are the primary beneficiary. + +## Property and Equipment + +Property and equipment is comprised of technical infrastructure, office space, corporate and other assets currently in service, and assets not yet in service. Technical infrastructure includes data center land, buildings and leasehold improvements, and servers and network equipment. Office space includes office land, buildings, and leasehold improvements. Assets not yet in service are those that are not ready for their intended use, including data center buildings and servers in the process of construction or assembly. + +Property and equipment are stated at cost less accumulated depreciation. Depreciation commences once assets are ready for their intended use and is recorded using the straight-line method over the estimated useful lives of the assets, which we regularly evaluate for factors such as technological obsolescence and our planned use and utilization. We depreciate data center and office buildings over periods of seven to 40 years. We depreciate servers and network equipment generally over a period of six years. We depreciate corporate and other assets over periods of two to 25 years. We depreciate leasehold improvements over the shorter of the remaining lease term or the estimated useful lives of the assets. Land is not depreciated. + +## Goodwill + +We allocate goodwill to reporting units based on the expected benefit from the business combination. We evaluate our reporting units periodically, as well as when changes in our operating segments occur. For changes in reporting units, we reassign goodwill using a relative fair value allocation approach. We test our goodwill for impairment at least annually, or more frequently if events or changes in circumstances indicate that the asset may be impaired. Goodwill impairments were not material for the periods presented. + +## Leases + +We determine if an arrangement is a lease at inception. Our lease agreements generally contain lease and non-lease components. Payments under our lease arrangements are primarily fixed. Non-lease components primarily include payments for maintenance and utilities. We combine fixed payments for non-lease components with lease payments and account for them together as a single lease component which increases the amount of our lease assets and liabilities. + +Certain lease agreements contain variable payments, which are expensed as incurred and not included in the lease assets and liabilities. These amounts primarily include payments affected by the Consumer Price Index, and payments for maintenance and utilities. + +Lease assets and liabilities are recognized at the present value of the future lease payments at the lease commencement date. The interest rate used to determine the present value of the future lease payments is our incremental borrowing rate, because the interest rate implicit in our leases is not readily determinable. Our incremental borrowing rate is estimated to approximate the interest rate on a collateralized basis with similar terms and payments, and in economic environments where the leased asset is located. Our lease terms and payments include periods under options to purchase, extend, or terminate the lease when it is reasonably certain that we will exercise that option. We generally use the base, non-cancelable, lease term when determining the lease assets and liabilities. Lease assets also include any prepaid lease payments and lease incentives. + +The current portion of our operating lease liabilities is included in accrued expenses and other current liabilities, and the long-term portion is included in operating lease liabilities. Finance lease assets are included in property and equipment, net. Finance lease liabilities are included in accrued expenses and other current liabilities or other long-term liabilities. + +Operating lease expense (excluding variable lease costs) is recognized on a straight-line basis over the lease term. Finance lease expense is recognized on a straight-line basis over the shorter of the lease term or the useful life of the asset, and interest expense is recognized based on the incremental borrowing rate. + +## Impairment of Long-Lived Assets + +We review leases, property and equipment, and intangible assets, excluding goodwill, for impairment when events or changes in circumstances indicate the carrying amount may not be recoverable. The evaluation is performed at the lowest level of identifiable cash flows independent of other assets. We measure recoverability of these assets by comparing the carrying amounts to the future undiscounted cash flows that the assets or the asset group are expected to generate. If the carrying value of the assets or asset group is not recoverable, the impairment recognized is measured as the amount by which the carrying value exceeds its fair value. + +## Income Taxes + +We account for income taxes using the asset and liability method, under which we recognize the amount of taxes payable or refundable for the current year and deferred tax assets and liabilities for the future tax consequences of events that have been recognized in our financial statements or tax returns. We measure current and deferred tax assets and liabilities based on provisions of enacted tax law. We evaluate the likelihood of future realization of our deferred tax assets based on all available evidence and establish a valuation allowance to reduce deferred tax assets when it is more likely than not that they will not be realized or release a valuation allowance to increase deferred tax assets when it is more likely than not that they will be realized. We have elected to account for the tax effects of the global intangible low tax income provision as a current period expense. + +We recognize the financial statement effects of a tax position when it is more likely than not that, based on technical merits, the position will be sustained upon examination. The tax benefits of the position recognized in the financial statements are then measured based on the largest amount of benefit that is greater than 50% likely to be realized upon settlement with a taxing authority. In addition, we recognize interest and penalties related to unrecognized tax benefits as a component of the income tax provision. + +## Business Combinations + +We include the results of operations of the businesses that we acquire as of the acquisition date. We allocate the purchase price of the acquisitions to the assets acquired and liabilities assumed based on their estimated fair values, except for revenue contracts acquired, which are recognized in accordance with our revenue recognition policy. The excess of the purchase price over the fair values of identifiable assets and liabilities is recorded as goodwill. Acquisition-related expenses are recognized separately from the business combination and are expensed as incurred. + +## Foreign Currency + +We translate the financial statements of our international subsidiaries to US dollars using month-end exchange rates for assets and liabilities, and average rates for the period derived from month-end exchange rates for revenues, costs, and expenses. We record translation gains and losses in AOCI as a component of stockholders' equity. We reflect net foreign exchange transaction gains and losses resulting from the conversion of the transaction currency to functional currency as a component of foreign currency exchange gain (loss) in OI&E. + +## Recently Issued Accounting Pronouncements Not Yet Adopted + +In November 2024, the Financial Accounting Standards Board (FASB) issued Accounting Standards Update (ASU) 2024-03 "Income Statement: Reporting Comprehensive Income-Expense Disaggregation Disclosures (Subtopic 220-40)" to improve the disclosures about an entity's expenses. Upon adoption, we will be required to disclose in the notes to the financial statements a disaggregation of certain expense categories included within the relevant expense captions on the consolidated statements of income. The standard is effective for our 2027 annual period, and our interim periods beginning in 2028, with early adoption permitted. The standard can be applied either prospectively or retrospectively. We are currently assessing adoption timing, the method of adoption, and the effect that the updated standard will have on our financial statement disclosures. + +In September 2025, the FASB issued ASU 2025-06 "Intangibles: Goodwill and Other–Internal-Use Software (Subtopic 350-40): Targeted Improvements to the Accounting for Internal-Use Software" to modernize the accounting for software costs under Subtopic 350-40, Intangibles–Goodwill and Other–Internal-Use Software (referred to as "internal-use software"). Upon adoption, we will be required to account for internal-use software under the updated capitalization criteria. The standard is effective for our interim and annual 2028 periods, with early adoption permitted. The standard can be applied either prospectively, retrospectively, or under a modified transition approach. We are currently assessing adoption timing, the method of adoption, and the effect that the updated standard will have on our consolidated financial statements. + +## Recently Adopted Accounting Pronouncements + +In December 2023, the FASB issued ASU 2023-09 "Income Taxes (Topics 740): Improvements to Income Tax Disclosures" which expands the disclosure requirements for income taxes. We adopted this ASU for our 2025 annual period with the comparative periods updated to reflect additional disclosures. See Note 14 for the revised disclosures consistent with the new standard. + +## Prior Period Reclassifications + +Certain amounts in prior periods have been reclassified to conform with current period presentation. + +## Note 2. Revenues + +## Disaggregated Revenues + +The following table presents revenues disaggregated by type (in millions): + +
Year Ended December 31,
202320242025
Google Search & other$ 175,033$ 198,084$ 224,532
YouTube ads31,51036,14740,367
Google Network31,31230,35929,792
Google advertising237,855264,590294,691
Google subscriptions, platforms, and devices34,68840,34048,030
Google Services total272,543304,930342,721
Google Cloud33,08843,22958,705
Other Bets1,5271,6481,537
Hedging gains (losses)236211(127)
Total revenues$ 307,394$ 350,018$ 402,836
+ +No individual customer or groups of affiliated customers represented more than 10% of our revenues in 2023, 2024, or 2025. + +The following table presents revenues disaggregated by geography, based on the addresses of our customers (in millions): + +
Year Ended December 31,
202320242025
United States$ 146,28647 %$ 170,44749 %$ 194,22948 %
$EMEA^{(1)}$ 91,03830102,12729117,15229
$APAC^{(1)}$ 51,5141756,8151667,68017
Other $Americas^{(1)}$ 18,320620,418623,9026
Hedging gains (losses)23602110(127)0
Total revenues$ 307,394100 %$ 350,018100 %$ 402,836100 %
+ +(1) Regions represent Europe, the Middle East, and Africa (EMEA); Asia-Pacific (APAC); and Canada and Latin America ("Other Americas"). + +## Revenue Backlog + +As of December 31, 2025, we had \$242.8 billion of remaining performance obligations ("revenue backlog"), primarily related to Google Cloud. Revenue backlog represents commitments in customer contracts that have not yet been recognized as revenue. We expect to recognize just over 50% of the revenue backlog as revenues over the next 24 months with the remainder to be recognized thereafter. The estimated revenue backlog and timing of revenue recognition for these commitments is largely driven by contract duration, our ability to deliver in accordance with relevant contract terms, and when our customers utilize services. Revenue backlog includes related deferred revenue currently recorded as well as amounts that will be invoiced in future periods, and excludes contracts with an original expected term of one year or less and cancellable contracts. + +## Deferred Revenues + +We record deferred revenues when cash payments are received or due in advance of our performance, including amounts which are refundable. Deferred revenues primarily relate to Google Cloud and Google subscriptions, platforms, and devices. Total deferred revenue as of December 31, 2024 was \$6.0 billion, of which \$4.6 billion was recognized as revenues for the year ended December 31, 2025. Total deferred revenue as of December 31, 2025 was \$8.6 billion. + +## Note 3. Financial Instruments + +## Fair Value Measurements + +## Investments Measured at Fair Value on a Recurring Basis + +Cash equivalents and marketable equity securities are measured at fair value and classified within Level 1 and Level 2 in the fair value hierarchy, because we use quoted prices for identical assets in active markets or inputs that are based upon quoted prices for similar instruments in active markets. + +Debt securities are measured at fair value and classified within Level 2 in the fair value hierarchy, because we use quoted market prices to the extent available or alternative pricing sources and models utilizing market observable inputs to determine fair value. + +The following tables summarize our cash, cash equivalents, and marketable securities measured at fair value on a recurring basis (in millions): + +
As of December 31, 2024
Quoted Prices in Active Markets for Identical Assets (Level 1)Significant Other Observable Inputs (Level 2)Total
Cash$ 12,407
Cash equivalents:
Money market funds$ 8,154$ 0$ 8,154
Time deposits02,0812,081
Government bonds0746746
Corporate debt securities07878
Total cash and cash equivalents8,1542,90523,466
Marketable securities:
Marketable equity $securities^{(1)}$ 4,7081054,813
Time deposits0136136
Government bonds028,70928,709
Corporate debt securities021,11621,116
Mortgage-backed and asset-backed securities017,41717,417
Total marketable securities4,70867,48372,191
Total$ 12,862$ 70,388$ 95,657
+ +(1) The long-term portion of marketable equity securities (subject to long-term lock-up restrictions) of \$266 million as of December 31, 2024 is included within other non-current assets. + +
As of December 31, 2025
Quoted Prices in Active Markets for Identical Assets (Level 1)Significant Other Observable Inputs (Level 2)Total
Cash$ 15,305
Cash equivalents:
Money market funds$ 11,349$ 0$ 11,349
Time deposits03,3533,353
Government bonds0602602
Corporate debt securities09999
Total cash and cash equivalents11,3494,05430,708
Marketable securities:
Marketable equity securities4,4021,9116,313
Time deposits000
Government bonds050,54950,549
Corporate debt securities021,56521,565
Mortgage-backed and asset-backed securities017,70817,708
Total marketable securities4,40291,73396,135
Total$ 15,751$ 95,787$ 126,843
+ +## Investments Measured at Fair Value on a Nonrecurring Basis + +Non-marketable equity securities accounted for under the measurement alternative are investments in privately held companies without readily determinable market values. The carrying value of these non-marketable equity securities is adjusted upward or downward to fair value upon observable transactions for identical or similar investments of the same issuer or impairment. Non-marketable equity securities that have been remeasured during the period based on observable transactions are classified within Level 2 or Level 3 in the fair value hierarchy, and remeasurements due to impairment are classified within Level 3. Our valuation methods include option pricing models, market comparable approach, and common stock equivalent method, which may include a combination of the observable transaction price at the transaction date and other unobservable inputs including volatility, expected time to exit, risk free rate, and the rights and obligations of the securities we hold. These inputs vary significantly based on investment type. + +As of December 31, 2025, the carrying value of our non-marketable equity securities accounted for under the measurement alternative was \$64.1 billion, of which \$45.6 billion were remeasured at fair value during the year ended December 31, 2025, and were primarily classified within Level 2 of the fair value hierarchy at the time of measurement. + +## Debt and Equity Securities + +## Debt Securities + +The following table summarizes the estimated fair value of investments in available-for-sale marketable debt securities by effective contractual maturity dates (in millions): + +
As of December 31, 2025
Due in 1 year or less$ 26,735
Due in 1 year through 5 years37,001
Due in 5 years through 10 years12,769
Due after 10 years13,317
Total$ 89,822
+ +The following tables present fair values and gross unrealized gains and losses recorded to AOCI, less any expected credit losses, aggregated by investment category (in millions): + +
As of December 31, 2024
Adjusted CostGross Unrealized GainsGross Unrealized LossesFair Value
Time deposits$ 2,217$ 0$ 0$ 2,217
Government bonds27,55183(214)27,420
Corporate debt securities18,30079(222)18,157
Mortgage-backed and asset-backed securities14,43763(385)14,115
Total investments with fair value change reflected in other comprehensive income$ 62,505$ 225$ (821)$ 61,909
+ +
As of December 31, 2025
Adjusted CostGross Unrealized GainsGross Unrealized LossesFair Value
Time deposits$ 3,353$ 0$ 0$ 3,353
Government bonds49,087443(26)49,504
Corporate debt securities18,346242(32)18,556
Mortgage-backed and asset-backed securities14,337174(128)14,383
Total investments with fair value change reflected in other comprehensive income$ 85,123$ 859$ (186)$ 85,796
+ +The following tables present fair values and gross unrealized losses recorded to AOCI, aggregated by investment category and the length of time that individual securities have been in a continuous loss position (in millions): + +
As of December 31, 2024
Less than 12 Months12 Months or GreaterTotal
Fair ValueUnrealized LossFair ValueUnrealized LossFair ValueUnrealized Loss
Government bonds$ 11,119$ (126)$ 2,576$ (88)$ 13,695$ (214)
Corporate debt securities4,228(17)6,838(168)11,066(185)
Mortgage-backed and asset-backed securities5,222(106)3,813(279)9,035(385)
Total$ 20,569$ (249)$ 13,227$ (535)$ 33,796$ (784)
+ +
As of December 31, 2025
Less than 12 Months12 Months or GreaterTotal
Fair ValueUnrealized LossFair ValueUnrealized LossFair ValueUnrealized Loss
Government bonds$ 4,230$ (9)$ 1,174$ (17)$ 5,404$ (26)
Corporate debt securities91502,429(24)3,344(24)
Mortgage-backed and asset-backed securities1,377(4)3,035(124)4,412(128)
Total$ 6,522$ (13)$ 6,638$ (165)$ 13,160$ (178)
+ +We determine realized gains or losses on the sale or extinguishment of debt securities on a specific identification method. For certain marketable debt securities, we have elected the fair value option for which changes in fair value are recorded in OI&E. The fair value option was elected for these securities to align with the unrealized gains and losses from related derivative contracts. + +The following table summarizes gains and losses for debt securities, reflected as a component of OI&E (in millions): + +
Year Ended December 31,
202320242025
Unrealized gain (loss) on fair value option debt securities$ 386$ 30$ 254
Gross realized gain on debt securities182482572
Gross realized loss on debt securities(1,833)(1,553)(316)
(Increase) decrease in allowance for credit losses50(2)30
Total gain (loss) on debt securities recognized in other income (expense), net$ (1,215)$ (1,043)$ 540
+ +## Non-marketable Securities + +Our non-marketable securities primarily consist of non-marketable equity securities accounted for under the measurement alternative. The carrying value is measured at the total initial cost plus the cumulative net upward and downward adjustments (including impairments). We account for non-marketable equity securities through which we exercise significant influence but do not have control over the investee under the equity method. Certain of our non-marketable securities include our investments in VIEs where we are not the primary beneficiary. See Note 5 for further details on VIEs. + +Realized net gain (loss) on equity securities sold during the period reflects the difference between the sale proceeds and the carrying value of the equity securities at the beginning of the period or the purchase date, if later. + +All gains and losses, including impairments, are included as components of OI&E. + +The carrying values for non-marketable securities are summarized below (in millions): + +
As of December 31,
20242025
Non-marketable securities:
Total initial cost of non-marketable equity securities accounted for under the measurement alternative$ 20,940$ 28,429
Cumulative upward adjustments22,70944,485
Cumulative downward adjustments (including impairments)(8,431)(8,820)
Carrying value of non-marketable equity securities accounted for under the measurement alternative35,21864,094
Equity method investments and other2,7644,593
Total non-marketable securities$ 37,982$ 68,687
+ +## Gains and Losses on Equity Securities + +Gains and losses (including impairments), net, for equity securities included in OI&E are summarized below (in millions): + +
Year Ended December 31,
202320242025
Gross unrealized gain on non-marketable equity securities accounted for under the measurement alternative$ 1,806$ 5,582$ 22,666
Gross unrealized loss (including impairments) on non-marketable equity securities accounted for under the measurement alternative(2,894)(2,210)(1,271)
Unrealized net gain (loss) on non-marketable equity securities accounted for under the measurement alternative(1,088)3,37221,395
Unrealized net gain (loss) on marketable and other equity securities7901561,907
Realized net gain (loss) on marketable and non-marketable equity securities sold during the period690186778
Total gain (loss) on equity securities in other income (expense), net (1)$ 392$ 3,714$ 24,080
+ +(1) Excludes income (loss) and impairment from equity method investments. Refer to Note 7 for further details. + +Cumulative net gains (losses), calculated as the difference between the sales price and purchase price, represent the total net gains (losses) recognized after the initial purchase date. This represents the total economic impact of the investment, regardless of when the gains or losses were previously recognized. Cumulative net gains on equity securities sold were \$748 million and \$387 million for the years ended December 31, 2024 and 2025, respectively. + +## Derivative Financial Instruments + +We primarily use derivative instruments to manage risks relating to our ongoing business operations, including foreign currencies, interest rates, commodity prices, credit exposures, and market prices of certain marketable equity securities. Additionally, we enter into derivatives to enhance investment returns. We also enter into derivatives as a result of agreements with third parties to backstop certain obligations related to data center leases. These backstop agreements are accounted for as credit derivatives. + +We recognize derivative instruments in the Consolidated Balance Sheets at fair value and classify them primarily within Level 2 in the fair value hierarchy. We present our foreign currency collars (an option strategy comprised of a combination of purchased and written options) at net fair values and present all other derivatives at gross fair values. The accounting treatment for derivatives is based on the intended use and hedge designation. + +## Cash Flow Hedges + +We designate foreign currency forwards and options (including collars) as cash flow hedges to hedge certain forecasted revenue transactions denominated in currencies other than the US dollar. These contracts have maturities of 24 months or less. + +Cash flow hedge amounts included in the assessment of hedge effectiveness are deferred in AOCI and reclassified to revenue when the hedged item is recognized in earnings. Hedge components excluded from our assessment of hedge effectiveness are amortized on a straight-line basis over the life of the hedging instrument in revenues. The difference between fair value changes of the excluded component and the amount amortized to revenues is recorded in AOCI. + +As of December 31, 2025, the net accumulated loss on our foreign currency cash flow hedges before tax effect was \$60 million, which is expected to be reclassified from AOCI into revenues within the next 12 months. + +Additionally, we may designate interest rate derivatives as cash flow hedges to manage our exposure to certain interest rate risks. Changes in the fair value of these derivatives are deferred in AOCI and reclassified to OI&E when the hedged item is recognized in earnings. + +## Fair Value Hedges + +We designate foreign currency forwards as fair value hedges to hedge foreign currency risks for our marketable debt securities denominated in currencies other than the US dollar. Fair value hedge amounts included and excluded from the assessment of hedge effectiveness are recognized in OI&E. + +## Net Investment Hedges + +We designate foreign currency forwards, options (including collars), cross-currency swaps, and foreign currency-denominated debt as net investment hedges to hedge the foreign currency risks related to our investments in foreign subsidiaries. Net investment hedge amounts included in the assessment of hedge effectiveness are recognized in AOCI. + +Changes in the fair value of hedge components of forward and option contracts that are excluded from the assessment of hedge effectiveness are recognized in OI&E. Hedge components of cross-currency swaps that are excluded from the assessment of hedge effectiveness are amortized over the life of the hedging instrument and recognized in OI&E. The difference between fair value changes of the excluded component and the amount amortized to OI&E is recorded in AOCI. + +We had no foreign currency-denominated debt as of December 31, 2024 and \$15.4 billion carrying value of foreign currency-denominated debt designated as net investment hedges as of December 31, 2025. + +## Derivatives Not Designated as Hedging Instruments + +We enter into derivatives not designated as hedging instruments to manage risks related to our ongoing business operations. The primary risk managed is foreign exchange risk related to the remeasurement of monetary assets or liabilities denominated in currencies other than the functional currency of a subsidiary. Gains and losses on these foreign exchange derivatives are recorded within the “foreign currency exchange gain (loss), net” component of OI&E. + +We also enter into derivatives to manage other risks, to enhance investment returns, and as a result of agreements with certain third parties to backstop certain obligations relating to data center leases. Gains and losses arising from other derivatives are primarily reflected within the “other” component of OI&E. See Note 7 for further details. + +The gross notional amounts of outstanding derivative instruments were as follows (in millions): + +
As of December 31,
20242025
Derivatives designated as hedging instruments:
Foreign exchange contracts
Cash flow hedges$ 20,315$ 23,852
Fair value hedges$ 1,562$ 0
Net investment hedges$ 6,986$ 14,203
Derivatives not designated as hedging instruments:
Foreign exchange contracts$ 44,227$ 56,085
Credit derivatives(1)$ 0$ 16,940
Other contracts$ 15,082$ 15,900
+ +(1) Notional amounts for credit derivatives are the backstop obligations related to certain third-party data center leases and represent the maximum potential amount of future payments that could be required in the event of certain default scenarios over remaining agreement periods of up to 15 years. In the event we are required to make payments under certain backstop obligations, we may receive equity in or cash payments from certain counterparties, the amounts for which are not reflected in the notional amounts for credit derivatives. See Note 5 for further details. + +The fair values of outstanding derivative instruments were as follows (in millions): + +
As of December 31, 2024As of December 31, 2025
Assets(1)Liabilities(2)Assets(1)Liabilities(2)
Derivatives designated as hedging instruments:
Foreign exchange contracts$ 1,054$ 0$ 316$ 197
Derivatives not designated as hedging instruments:
Foreign exchange contracts2005939284
Other contracts4741932498
Total derivatives not designated as hedging instruments674612416182
Total$ 1,728$ 612$ 732$ 379
+ +(1) Derivative assets are recorded as other current and non-current assets. +(2) Derivative liabilities are recorded as accrued expenses and other liabilities, current and non-current. + +The gains (losses) on derivatives and non-derivative financial instruments in cash flow hedging and net investment hedging relationships recognized in other comprehensive income are summarized below (in millions): + +
Year Ended December 31,
202320242025
Cash flow hedging relationship:
Foreign exchange and other contracts
Amount included in the assessment of effectiveness$ 90$ 857$ (978)
Amount excluded from the assessment of effectiveness8477(45)
Net investment hedging relationship:
Amounts included in the assessment of effectiveness
Foreign exchange contracts(287)223(765)
Foreign currency-denominated debt00(393)
Amounts excluded from the assessment of effectiveness
Foreign exchange contracts0011
Total$ (113)$ 1,157$ (2,170)
+ +The table below presents the gains (losses) of derivatives included on the Consolidated Statements of Income: (in millions): + +
Year Ended December 31,
202320242025
RevenuesOther income (expense), netRevenuesOther income (expense), netRevenuesOther income (expense), net
Total amounts included on the Consolidated Statements of Income$ 307,394$ 1,424$ 350,018$ 7,425$ 402,836$ 29,787
Effect of cash flow hedges:
Foreign exchange contracts
Amount included in the assessment of effectiveness21301740(233)0
Amount excluded from the assessment of effectiveness2403701070
Effect of fair value hedges:
Foreign exchange contracts
Hedged items0590(59)0(9)
Amount included in the assessment of effectiveness0(59)05809
Amount excluded from the assessment of effectiveness01501301
Effect of net investment hedges:
Foreign exchange contracts
Amount excluded from the assessment of effectiveness018701370189
Effect of non-designated hedges:
Foreign exchange contracts0703350445
Other contracts05301740(148)
Total gains (losses)$ 237$ 262$ 211$ 658$ (126)$ 487
+ +## Offsetting of Derivatives + +We enter into master netting arrangements and collateral security arrangements to reduce credit risk. Cash collateral received related to derivative instruments under our collateral security arrangements are included in other current assets with a corresponding liability. Cash and non-cash collateral pledged related to derivative instruments under our collateral security arrangements are primarily included in other current assets. + +The gross amounts of derivative instruments subject to master netting arrangements with various counterparties, and cash and non-cash collateral received and pledged under such agreements were as follows (in millions): + +
As of December 31, 2024
Gross Amounts RecognizedGross Amounts Offset in the Consolidated Balance SheetsNet Amounts Presented in the Consolidated Balance SheetsFinancial Instruments(1)Cash and Non-Cash Collateral Received or PledgedNet Amounts
Derivatives assets$ 1,776$ (48)$ 1,728$ (516)$ (721)$ 491
Derivatives liabilities$ 660$ (48)$ 612$ (516)$ (9)$ 87
As of December 31, 2025
Gross Amounts RecognizedGross Amounts Offset in the Consolidated Balance SheetsNet Amounts Presented in the Consolidated Balance SheetsFinancial Instruments(1)Cash and Non-Cash Collateral Received or PledgedNet Amounts
Derivatives assets$ 842$ (110)$ 732$ (140)$ (231)$ 361
Derivatives liabilities$ 489$ (110)$ 379$ (140)$ (15)$ 224
+ +(1) The balances as of December 31, 2024 and 2025 were related to derivatives allowed to be net settled in accordance with our master netting agreements. + +## Note 4. Leases + +We have entered into operating and finance lease agreements primarily for data centers, land, and offices throughout the world with varying lease terms. + +Components of lease costs were as follows (in millions): + +
Year Ended December 31,
202320242025
Operating lease cost$ 3,362$ 3,304$ 3,345
Finance lease cost:
Amortization of lease assets469413553
Interest on lease liabilities353165
Finance lease cost504444618
Variable lease cost1,1821,4251,739
Total lease cost$ 5,048$ 5,173$ 5,702
+ +Supplemental information related to leases was as follows (in millions): + +
December 31,
20242025
Weighted-average remaining lease term:
Operating leases7.8 years7.6 years
Finance leases10.4 years8.3 years
Weighted-average discount rate:
Operating leases3.4 %3.6 %
Finance leases2.8 %3.1 %
December 31,
20242025
Operating leases:
Operating lease assets$ 13,588$ 15,221
Accrued expenses and other liabilities$ 2,887$ 3,209
Operating lease liabilities11,69112,744
Total operating lease liabilities$ 14,578$ 15,954
Finance leases:
Property and equipment, at cost$ 4,622$ 6,822
Accumulated depreciation(2,037)(2,025)
Property and equipment, net$ 2,585$ 4,797
Accrued expenses and other liabilities$ 235$ 441
Other long-term liabilities1,4422,059
Total finance lease liabilities$ 1,677$ 2,500
+ +
Year Ended December 31,
202320242025
Cash payments for lease liabilities:
Operating cash flows used for operating leases$ 3,173$ 3,425$ 3,370
Operating cash flows used for finance leases$ 35$ 31$ 65
Financing cash flows used for finance leases(1)$ 705$ 405$ 1,988
Assets obtained in exchange for lease liabilities:
Operating leases$ 2,877$ 2,510$ 4,070
Finance leases$ 564$ 313$ 1,606
+ +(1) Financing cash flows used for financing leases are included within financing activities as repayments of debt. The year ended December 31, 2025 includes \$1.1 billion of prepayments for finance leases not yet commenced. + +Future lease payments as of December 31, 2025 were as follows (in millions): + +
Operating LeasesFinance Leases
2026$ 3,275$ 491
20273,082345
20282,510335
20292,061314
20301,669241
Thereafter5,6541,143
Total undiscounted lease payments18,2512,869
Less: imputed interest(2,297)(369)
Total lease liability balance$ 15,954$ 2,500
+ +As of December 31, 2025, we have entered into leases primarily related to data centers that have not yet commenced with short-term and long-term future lease payments of \$5.8 billion and \$52.7 billion, respectively, that are not yet recorded. These leases will commence between 2026 and 2031 with non-cancelable lease terms primarily between one and 25 years. + +In January 2026, we executed a power purchase agreement which we expect to be accounted for as a lease resulting in future payments depending on certain agreement terms of \$9.9 billion between 2027 and 2047. If certain contractual conditions for the project are not met, we would instead make a one-time payment of approximately \$3.5 billion and assume ownership of the power generating assets. + +## Note 5. Variable Interest Entities + +## Consolidated VIEs + +We consolidate VIEs in which we hold a variable interest and are the primary beneficiary. The results of operations and financial position of these VIEs are included in our consolidated financial statements. + +For certain consolidated VIEs, their assets are not available to us, and their creditors do not have recourse to us. As of December 31, 2024 and 2025, assets that can only be used to settle obligations of these VIEs were \$8.7 billion and \$5.6 billion, respectively, and are primarily included in cash and cash equivalents. As of December 31, 2024 and 2025, liabilities for which creditors only have recourse to the VIEs were \$2.3 billion and \$2.0 billion, respectively. We may continue to fund ongoing operations, including the potential funding of employee compensation programs, of certain VIEs that are included within Other Bets. + +In February 2026, Waymo, a consolidated VIE, announced an investment round of \$16.0 billion, the significant majority of which was funded by Alphabet. Investments from external parties will be accounted for as equity transactions and will result in recognition of noncontrolling interests. + +Total noncontrolling interests (NCI) in our consolidated subsidiaries were \$4.2 billion and \$3.4 billion as of December 31, 2024 and 2025, respectively, of which \$1.1 billion and \$841 million were redeemable noncontrolling interests (RNCI) as of December 31, 2024 and 2025, respectively. NCI and RNCI are included within additional paid-in capital. Net loss attributable to noncontrolling interests was not material for any period presented and is included within the "other" component of OI&E. See Note 7 for further details on OI&E. + +## Unconsolidated VIEs + +We hold various forms of interests in Variable Interest Entities (VIEs), including certain of our investments in private companies and renewable energy entities, certain leases and credit backstops with data center entities, and certain backstops with energy infrastructure entities. Because we have determined that we do not direct the activities that most significantly impact the economic performance of these entities, we are not the primary beneficiary. Therefore, these VIEs are not consolidated within our financial statements. + +Our investments in private companies and renewable energy VIEs are primarily accounted for as non-marketable securities under the measurement alternative or the equity method. The carrying value of these investments are included within non-marketable securities on our Consolidated Balance Sheets. See Note 3 for further details on investments. The maximum exposure to these VIEs is generally limited to the current carrying value plus future funding commitments. As of December 31, 2024 and 2025, future funding commitments were \$1.5 billion and \$1.1 billion, respectively. + +Leases with data center leasing VIEs are accounted for as finance leases and are included within total lease obligations disclosed in Note 4. The maximum exposure arising from leases with VIEs is limited to the net carrying value of commenced finance lease assets, plus the undiscounted future obligations for leases that have not yet commenced. See Note 4 for further details on leases. + +Credit backstops we have provided to data center VIEs are accounted for as credit derivatives. The maximum exposure arising from credit backstops with VIEs is limited to the financial risk over the remaining period of the arrangements, as reflected by the credit derivative notional value. See Note 3 for further details on credit derivatives. + +Backstop agreements we have provided to energy infrastructure VIEs are accounted for as financial guarantees. The maximum exposure to these VIEs is limited to the potential amount of future payments under these arrangements. See Note 10 for further details on financial guarantees. + +## Note 6. Debt + +## Short-Term Debt + +We have a commercial paper program of up to \$25.0 billion, which is used for general corporate purposes. We had \$2.3 billion of commercial paper outstanding with a weighted-average effective interest rate of 4.4% as of December 31, 2024 and no commercial paper outstanding as of December 31, 2025. The fair value of the commercial paper approximated its carrying value as of December 31, 2024. + +Our short-term debt balance also includes the current portion of certain long-term debt. + +## Long-Term Debt + +During 2025, we issued \$22.5 billion of US dollar-denominated senior unsecured notes and €13.25 billion of euro-denominated senior unsecured notes for general corporate purposes. + +In May 2025, we issued \$5.0 billion of US dollar-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 4.89%, and a weighted-average maturity of approximately 24 years. Additionally, in May 2025, we issued €6.75 billion of euro-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 3.31%, and a weighted-average maturity of approximately 14 years. + +In November 2025, we issued \$500 million of US dollar-denominated floating-rate senior unsecured notes and \$17.0 billion of US dollar-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 4.92% and a weighted-average maturity of approximately 20 years. Additionally in November 2025, we issued €6.5 billion of euro-denominated fixed-rate senior unsecured notes with a weighted-average coupon rate of 3.44% and a weighted-average maturity of approximately 16 years. + +Total outstanding long-term debt is summarized below (in millions, except percentages): + +
MaturityCoupon RateEffective Interest RateAs of December 31,
20242025
Debt
2016 US dollar notes20262.00%2.23%$ 2,000$ 2,000
2020 US dollar notes2027 - 20600.80% - 2.25%0.93% - 2.33%10,0009,000
$2025 \text{ US dollar notes}^{(1)}$ 2028 - 20753.88% - 5.70%4.00% - 5.79%022,500
$2025 \text{ Euro notes}^{(2)}$ 2028 - 20642.38% - 4.38%2.57% - 4.51%015,585
Total face value of long-term debt12,00049,085
Unamortized discount and debt issuance costs $^{(2)}$ (118)(542)
Less: current portion of long-term notes $^{(3)}$ (999)(1,996)
Total long-term debt$ 10,883$ 46,547
+ +(1) Includes \$500 million of floating-rate notes due in 2028. Interest is calculated using the compounded Secured Overnight Financing Rate (SOFR) plus 0.52%, reset quarterly. +(2) Principal, unamortized discount, and debt issuance costs for the euro-denominated notes include the effect of foreign exchange rates. +(3) Total current portion of long-term debt is included within accrued expenses and other current liabilities. See Note 7 for further details. + +The notes in the table above are senior unsecured obligations and rank equally with each other. We may redeem the fixed-rate notes at any time in whole or in part at specified redemption prices. The floating-rate notes are not redeemable prior to maturity. Interest is payable quarterly for the floating-rate notes, semi-annually for the US dollar-denominated fixed-rate notes, and annually for the euro-denominated fixed-rate notes. The effective interest rates are based on proceeds received and contractual interest payments. + +The total estimated fair value of the outstanding notes was approximately \$9.0 billion and \$45.6 billion as of December 31, 2024 and December 31, 2025, respectively. The fair value was determined based on observable market prices of identical instruments in less active markets and is categorized accordingly as Level 2 in the fair value hierarchy. + +As of December 31, 2025, the future principal payments for long-term debt were as follows (in millions): + +
2026$ 2,000
20271,000
20282,676
20291,764
20305,500
Thereafter36,145
Total$ 49,085
+ +## Credit Facility + +As of December 31, 2025, we had \$10.0 billion of revolving credit facilities, of which \$4.0 billion expires in April 2026 and \$6.0 billion expires in April 2030. The interest rates for all credit facilities are determined based on a formula using certain market rates. No amounts were outstanding under the credit facilities as of December 31, 2024 and 2025. + +## Note 7. Supplemental Financial Statement Information + +## Accounts Receivable + +The allowance for credit losses on accounts receivable was \$879 million and \$924 million as of December 31, 2024 and 2025, respectively. + +## Property and Equipment, Net + +Property and equipment, net, consisted of the following (in millions): + +
As of December 31,
20242025
Technical infrastructure(1)$ 141,852$ 203,679
Office space45,40348,348
Corporate and other assets12,57414,463
Property and equipment, in service199,829266,490
Less: accumulated depreciation(79,390)(98,485)
Add: assets not yet in service50,59778,592
Property and equipment, net$ 171,036$ 246,597
+ +(1) As of December 31, 2024 and 2025, approximately 60% of technical infrastructure assets were comprised of servers and network equipment. The remaining balance was comprised of data center land and buildings and related assets. + +## Accrued Expenses and Other Current Liabilities + +Accrued expenses and other current liabilities consisted of the following (in millions): + +
As of December 31,
20242025
Accrued fines and settlements(1)$ 9,830$ 15,594
Accrued purchases of property and equipment7,1048,877
Accrued customer liabilities4,3045,029
Payables to brokers for unsettled investment trades3,866950
Income taxes payable, net2,905523
Other accrued expenses and current liabilities23,21924,584
Accrued expenses and other current liabilities$ 51,228$ 55,557
+ +(1) See Legal Matters in Note 10 for further details. + +## Accumulated Other Comprehensive Income (Loss) + +Components of AOCI, net of income tax, were as follows (in millions): + +
Foreign Currency Translation AdjustmentsUnrealized Gains (Losses) on Available-for-Sale InvestmentsUnrealized Gains (Losses) on Cash Flow HedgesTotal
Balance as of December 31, 2022$ (4,142)$ (3,477)$ 16$ (7,603)
Other comprehensive income (loss) before reclassifications7351,344842,163
Amounts excluded from the assessment of hedge effectiveness recorded in AOCI008484
Amounts reclassified from AOCI01,168(214)954
Other comprehensive income (loss)7352,512(46)3,201
Balance as of December 31, 2023(3,407)(965)(30)(4,402)
Other comprehensive income (loss) before reclassifications(1,673)(116)698(1,091)
Amounts excluded from the assessment of hedge effectiveness recorded in AOCI007777
Amounts reclassified from AOCI0782(166)616
Other comprehensive income (loss)(1,673)666609(398)
Balance as of December 31, 2024(5,080)(299)579(4,800)
Other comprehensive income (loss) before reclassifications2,5111,146(734)2,923
Amounts excluded from the assessment of hedge effectiveness recorded in AOCI110(45)(34)
Amounts reclassified from AOCI0(169)164(5)
Other comprehensive income (loss)2,522977(615)2,884
Balance as of December 31, 2025$ (2,558)$ 678$ (36)$ (1,916)
+ +The effects on net income of amounts reclassified from AOCI were as follows (in millions): + +
AOCI ComponentsLocationYear Ended December 31,
202320242025
Unrealized gains (losses) on available-for-sale investments
Other income (expense), net$ (1,497)$ (1,008)$ 213
Benefit (provision) for income taxes329226(44)
Net of income tax(1,168)(782)169
Unrealized gains (losses) on cash flow hedges
Foreign exchange contractsRevenue213174(233)
Interest rate contractsOther income (expense), net611
Benefit (provision) for income taxes(5)(9)68
Net of income tax214166(164)
Total amount reclassified, net of income tax$ (954)$ (616)$ 5
+ +## Other Income (Expense), Net + +Components of OI&E were as follows (in millions): + +
Year Ended December 31,
202320242025
Interest income$ 3,865$ 4,482$ 4,337
Interest expense(1)(308)(268)(736)
Foreign currency exchange gain (loss), net(1,238)(409)(382)
Gain (loss) on debt securities, net(1,215)(1,043)540
Gain (loss) on equity securities, net3923,71424,080
Income (loss) and impairment from equity method investments, net(628)(188)281
Other5561,1371,667
Other income (expense), net$ 1,424$ 7,425$ 29,787
+ +(1) Interest expense is net of interest capitalized of \$181 million, \$194 million, and \$447 million for the years ended December 31, 2023, 2024, and 2025, respectively. + +## Note 8. Acquisitions + +## Pending Acquisitions + +In March 2025, we entered into a definitive agreement to acquire Wiz, a leading cloud security platform, for \$32.0 billion, subject to closing adjustments, in an all-cash transaction. The acquisition of Wiz is expected to close in 2026, subject to customary closing conditions, including the receipt of regulatory approvals. Upon the close of the acquisition, Wiz will be part of the Google Cloud segment. + +In December 2025, we entered into a definitive agreement to acquire Intersect, which provides data center and energy infrastructure solutions, for \$4.8 billion in cash, plus the assumption of debt. The acquisition of Intersect is expected to close in the first half of 2026, subject to customary closing conditions. + +## Note 9. Goodwill + +Changes in the carrying amount of goodwill for the years ended December 31, 2024 and 2025 were as follows (in millions): + +
Google ServicesGoogle CloudOther BetsTotal
Balance as of December 31, 2023$ 21,118$ 7,199$ 881$ 29,198
Additions2,44129502,736
Foreign currency translation and other adjustments(38)(4)(7)(49)
Balance as of December 31, 202423,5217,49087431,885
Additions1,26916301,432
Foreign currency translation and other adjustments807(24)63
Balance as of December 31, 2025$ 24,870$ 7,660$ 850$ 33,380
+ +## Note 10. Commitments and Contingencies + +## Commitments + +We have certain content licensing agreements with future fixed or minimum guaranteed commitments of \$7.7 billion as of December 31, 2025, of which the majority is paid quarterly through the first quarter of 2030. + +## Financial Guarantees + +We provide financial guarantees to certain counterparties, in the form of backstop agreements with varying terms through August 2026. These backstop agreements support counterparty procurement of long-lead time equipment for our future power purchase agreements. As of December 31, 2025, our maximum potential amount of future payments under these guarantees was \$5.7 billion, upon which we may receive certain assets. The fair value of these obligations was not material. + +## Indemnifications + +In the normal course of business, including to facilitate transactions in our services and products and corporate activities, we indemnify certain parties, including advertisers, Google Network partners, distribution partners, customers of Google Cloud offerings, lessors, and service providers with respect to certain matters. We have agreed to defend and/or indemnify certain parties against losses arising from a breach of representations or covenants, or out of intellectual property infringement or other claims made against certain parties. Several of these agreements limit the time within which an indemnification claim can be made and the amount of the claim. In addition, we have entered into indemnification agreements with our officers and directors, and our bylaws contain similar indemnification obligations to our agents. + +It is not possible to make a reasonable estimate of the maximum potential amount under these indemnification agreements due to the unique facts and circumstances involved in each particular agreement. Additionally, the payments we have made under such agreements have not had a material adverse effect on our results of operations, cash flows, or financial position. However, to the extent that valid indemnification claims arise in the future, future payments by us could be significant and could have a material adverse effect on our results of operations or cash flows in a particular period. + +As of December 31, 2025, we did not have any material indemnification claims that were probable or reasonably possible. + +## Legal Matters + +We record a liability when we believe that it is probable that a loss has been incurred, and the amount can be reasonably estimated. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the reasonably possible loss. We evaluate developments in our legal matters that could affect the amount of liability that has been previously accrued, and the matters and related reasonably possible losses disclosed, and make adjustments as appropriate. + +Certain outstanding matters seek speculative, substantial, or indeterminate monetary amounts, substantial changes to our business practices and products, or structural remedies. Significant judgment is required to determine both the likelihood of there being a loss and the estimated amount of a loss related to such matters, and we may be unable to estimate the reasonably possible loss or range of losses. The outcomes of outstanding legal matters are inherently unpredictable and subject to significant uncertainties, and could, either individually or in aggregate, have a material adverse effect. + +We expense legal fees in the period in which they are incurred. + +## Antitrust Matters + +We are subject to formal and informal inquiries and investigations as well as litigation on various competition matters by regulatory authorities and private parties in the US, Europe, and other jurisdictions globally, including the following: + +- Shopping: In June 2017, the EC announced its decision that certain actions taken by Google relating to its display and ranking of shopping search results and ads infringed European antitrust laws and imposed a €2.4 billion fine. In 2024, we made a cash payment of \$3.0 billion for the fine. +- Android: In July 2018, the EC announced its decision that certain provisions in Google's Android-related distribution agreements infringed European antitrust laws, imposed a €4.3 billion fine, and directed the termination of the conduct at issue. We appealed the EC decision and implemented changes to certain of our Android distribution practices. In September 2022, the General Court affirmed the EC decision but reduced the fine from €4.3 billion to €4.1 billion. We subsequently appealed the General Court's affirmation of the EC decision with the European Court of Justice, which remains pending. In 2018, we recognized a charge of \$5.1 billion for the fine, which we reduced by \$217 million in 2022. +- AdSense for Search: In March 2019, the EC announced its decision that certain provisions in Google's agreements with AdSense for Search partners infringed European antitrust laws, imposed a €1.5 billion fine, and directed actions related to AdSense for Search partners' agreements, which we implemented prior to the decision. In 2019, we recognized a charge of \$1.7 billion for the fine and appealed the EC decision. In September 2024, the General Court overturned the EC decision and annulled the €1.5 billion fine. The EC has appealed the General Court's decision with the European Court of Justice. +- Search: In October 2020, the DOJ and a number of state Attorneys General filed a lawsuit in the US District Court for the District of Columbia concerning Google's Search and Search advertising practices and its compliance with US antitrust laws. In August 2024, the US District Court for the District of Columbia ruled against Google. A final judgment was entered in December 2025, which, among other things, imposes restrictions on how Google distributes its services and requires Google to share certain search data with and offer syndication services to certain competitors. In January 2026, we appealed the final judgment and moved to pause implementation of certain remedies. In February 2026, the DOJ and state Attorneys General also appealed. + +Further, in June 2022, the Australian Competition and Consumer Commission (ACCC) opened an investigation into Search distribution practices. In August 2025, we agreed to a settlement with the ACCC requiring, among other things, changes to our Android agreements. We recognized a charge in the second quarter of 2025, and the settlement was approved by the court in December 2025. + +In October 2023, the Japanese Fair Trade Commission (JFTC) opened an investigation into Search distribution practices. In April 2025, the JFTC issued a cease-and-desist order requiring us to make changes to our Android agreements to ensure they are consistent with Japanese antitrust law. The JFTC did not impose monetary penalties. + +\- Advertising Technology: In December 2020, a number of state Attorneys General filed a lawsuit in the US District Court for the Eastern District of Texas concerning Google's advertising technology and its compliance with US antitrust laws and state deceptive trade laws. In January 2023, the DOJ, along with a number of state Attorneys General, filed a lawsuit in the US District Court for the Eastern District of Virginia concerning Google's advertising technology and its compliance with US antitrust laws, and a number of additional state Attorneys General subsequently joined the lawsuit. In April 2025, the US District Court for the Eastern District of Virginia issued a mixed decision in the DOJ case against Google, ruling that neither Google's advertiser tools nor the DoubleClick and AdMeld acquisitions were anticompetitive, but that Google's publisher tools unfairly excluded rivals. A separate proceeding to determine remedies, the range of which vary widely, took place in September 2025, with the parties presenting differing remedy proposals. The DOJ's remedy proposal includes structural remedies that could have a material adverse effect on our business. Closing arguments were held in November 2025, and we are awaiting a final judgment. After that judgment, we plan to appeal the adverse portion of the April 2025 decision and potentially aspects of the remedies decision. A trial in the state Attorneys General case in the Eastern District of Texas will take place after a decision on remedies is issued in the DOJ case. Given the nature of these matters, we cannot estimate a possible loss. + +Further, in September 2025, the EC announced its decision that Google had infringed European competition laws through "self-preferencing" practices on the buy-side and the sell-side relating to Google's advertising technology business. The EC decision imposed a €3.0 billion fine and directed Google to cease and desist the alleged "self-preferencing" practices. We appealed the ruling in November 2025. We recognized a charge of \$3.5 billion in the third quarter of 2025, and we placed bank guarantees in the fourth quarter of 2025 in lieu of cash payment. + +In September 2024, the UK also issued a Statement of Objections concerning Google's advertising technology and its compliance with UK antitrust laws, to which we responded. + +\- Google Play: In July 2021, a number of state Attorneys General filed a lawsuit in the US District Court for the Northern District of California concerning Google's operation of Android and Google Play and its compliance with US antitrust laws and state antitrust and consumer protection laws. In September 2023, we reached a settlement in principle with 50 state Attorneys General and three territories and recognized a charge. The court preliminarily approved the settlement in November 2025, and final approval remains pending before the court. In May 2024, we funded the settlement amount to an escrow agent. + +In December 2023, a California jury delivered a verdict against Google in Epic Games v. Google related to Google Play's business. Epic did not seek monetary damages. The presiding judge issued a remedies decision in October 2024, ordering a variety of alterations to our business models and operations and contractual agreements for Android and Google Play. We appealed the judgment, including the jury verdict and aspects of the remedies ordered, and in July 2025, the Court of Appeals denied our appeal. We are in the process of appealing that decision to the US Supreme Court, and we implemented the ordered remedies in October 2025 while the appeal is pending. In October 2025, we reached a settlement with Epic to modify the remedies in this case and resolve certain other lawsuits Epic has filed regarding Google Play's business. The settlement is contingent on the court approving a proposed modified injunction. Epic and Google filed a joint motion to modify the injunction in November 2025, which is currently pending before the court. + +\- European Digital Markets Act: In March 2024, the EC opened two investigations regarding Google's compliance with certain provisions of the EU's Digital Markets Act relating to Google Play and Search. In March 2025, the EC issued preliminary findings of non-compliance in both investigations, to which we responded. Given the nature of this matter, we cannot estimate a possible loss. + +In addition to these antitrust proceedings, private individual and collective actions that overlap with claims pursued by regulatory authorities are pending in the US and in several other jurisdictions, including across Europe. Given the nature of these matters, we cannot estimate a possible loss. + +We believe we have strong arguments against these open claims and will defend ourselves vigorously. We continue to cooperate with federal and state regulators in the US, the EC, and other regulators around the world. + +## Privacy Matters + +We are subject to a number of privacy-related laws and regulations, and we currently are party to a number of privacy investigations and lawsuits ongoing in multiple jurisdictions. For example, there are ongoing investigations and litigation in the US and the EU, including those relating to our collection and use of location information, the choices we offer users, and advertising practices, which could result in significant fines, judgments, and product changes. In October 2025, we finalized a \$1.4 billion settlement of certain privacy matters. + +## Patent and Intellectual Property Claims + +We have had patent, copyright, trade secret, and trademark infringement lawsuits filed against us claiming that certain of our products, services, and technologies infringe others' intellectual property rights. Adverse results in these lawsuits may include awards of substantial monetary damages, costly royalty or licensing agreements, or orders preventing us from offering certain features, functionalities, products, or services. As a result, we may have to change our business practices and develop non-infringing products or technologies, which could result in a loss of revenues for us and otherwise harm our business. In addition, the ITC has increasingly become an important forum to litigate intellectual property disputes because an ultimate loss in an ITC action can result in a prohibition on importing infringing products into the US. Because the US is an important market, a prohibition on importation could have an adverse effect on us, including preventing us from importing many important products into the US or necessitating workarounds that may limit certain features of our products. Further, our customers and partners may discontinue the use of our products, services, and technologies, as a result of injunctions or otherwise, which could result in loss of revenues and adversely affect our business. + +## Other + +We are subject to claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders involving competition, intellectual property, data privacy and security, tax and related compliance, labor and employment, commercial disputes, content generated by our users, goods and services offered by advertisers or publishers using our platforms, design of our products and services, personal injury and other tort and nuisance theories, consumer protection, including how we moderate content on our platforms, AI, and other matters. + +For example, we periodically have data incidents that we report to relevant regulators as required by law. Such claims, consent orders, lawsuits, regulatory and government investigations, and other proceedings could result in substantial fines and penalties, injunctive relief, ongoing monitoring and auditing obligations, changes to our products and services, alterations to our business models and operations, and collateral related civil litigation or other adverse consequences, all of which could harm our business, reputation, financial condition, and operating results. + +We have ongoing legal matters relating to Russia. For example, some matters concern civil judgments that include compounding penalties imposed upon us in connection with disputes regarding the termination of accounts, including those of sanctioned parties. We do not expect these ongoing legal matters will have a material adverse effect. + +## Non-Income Taxes + +We are under audit by various domestic and foreign tax authorities with regards to non-income tax matters. The subject matter of non-income tax audits primarily arises from disputes on the tax treatment and tax rate applied to the sale of our products and services in these jurisdictions and the tax treatment of certain employee benefits. We accrue non-income taxes that may result from examinations by, or any negotiated agreements with, these tax authorities when a loss is probable and reasonably estimable. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the reasonably possible loss. Due to the inherent complexity and uncertainty of these matters and judicial process in certain jurisdictions, the final outcome may be materially different from our expectations. + +See Note 14 for further details regarding income tax contingencies. + +## Note 11. Stockholders' Equity + +## Class A and Class B Common Stock and Class C Capital Stock + +Our Board of Directors has authorized three classes of stock, Class A and Class B common stock, and Class C capital stock. The rights of the holders of each class of our common and capital stock are identical, except with respect to voting. Each share of Class A common stock is entitled to one vote per share. Each share of Class B common stock is entitled to 10 votes per share. Class C capital stock has no voting rights, except as required by applicable law. Shares of Class B common stock may be converted at any time at the option of the stockholder and automatically convert upon sale or transfer to Class A common stock. + +## Share Repurchases + +In the years ended December 31, 2023, 2024, and 2025, we continued to repurchase both Class A and Class C shares in a manner deemed in the best interest of the company and its stockholders, taking into account the economic cost and prevailing market conditions, including the relative trading prices and volumes of the Class A and Class C shares. In April 2024, the company's Board of Directors authorized a \$70.0 billion share repurchase program for its Class A and Class C shares. In April 2025, the company's Board of Directors authorized an additional \$70.0 billion share repurchase program for its Class A and Class C shares. As of December 31, 2025, \$69.5 billion remained available for Class A and Class C share repurchases. + +The following table presents Class A and Class C shares repurchased and subsequently retired (in millions): + +Year Ended December 31, + +
202320242025
SharesAmountSharesAmountSharesAmount
Class A share repurchases78$ 9,31673$ 11,85537$ 6,501
Class C share repurchases45052,86830650,19220338,897
Total share repurchases(1)528$ 62,184379$ 62,047240$ 45,398
+ +(1) Shares repurchased include any unsettled repurchases. + +Repurchases are executed from time to time, subject to general business and market conditions and other investment opportunities, through open market purchases or privately negotiated transactions, including through Rule 10b5-1 plans. The repurchase programs do not have an expiration date. + +## Dividends + +During the year ended December 31, 2025, total cash dividends were \$4.8 billion for Class A, \$703 million for Class B, and \$4.5 billion for Class C shares, respectively. + +In April 2025, the company's Board of Directors increased the quarterly cash dividend by 5% to \$0.21 per share of outstanding Class A, Class B, and Class C shares. + +The company has declared a quarterly cash dividend in the current quarter, and intends to pay quarterly cash dividends in the future, subject to review and approval by the company's Board of Directors in its sole discretion. + +## Note 12. Net Income Per Share + +We compute net income per share of Class A, Class B, and Class C stock using the two-class method. Basic net income per share is computed using the weighted-average number of shares outstanding during the period. Diluted net income per share is computed using the weighted-average number of shares and the effect of potentially dilutive securities outstanding during the period. Potentially dilutive securities consist of RSUs and other contingently issuable shares. The dilutive effect of outstanding RSUs and other contingently issuable shares is reflected in diluted earnings per share by application of the treasury stock method. The computation of the diluted net income per share of Class A stock assumes the conversion of Class B stock, while the diluted net income per share of Class B stock does not assume the conversion of those shares. + +In accordance with our certificate of incorporation, the rights, including the liquidation and dividend rights, of the holders of our Class A, Class B, and Class C stock are identical, except with respect to voting. Furthermore, there are a number of safeguards built into our certificate of incorporation, as well as Delaware law, which preclude our Board of Directors from declaring or paying unequal per share dividends on our Class A, Class B, and Class C stock. Specifically, Delaware law provides that amendments to our certificate of incorporation which would have the effect of adversely altering the rights, powers, or preferences of a given class of stock must be approved by the class of stock adversely affected by the proposed amendment. In addition, our certificate of incorporation provides that before any such amendment may be put to a stockholder vote, it must be approved by the unanimous consent of our Board of Directors. + +Immaterial differences in net income per share across our Class A, Class B, and Class C shares may arise due to the allocation of distributed earnings, which is based on the holders as of the record date, compared with the allocation of undistributed earnings and number of shares, which is based on the weighted average shares outstanding over the periods. + +The following tables set forth the computation of basic and diluted net income per share of Class A, Class B, and Class C stock (in millions, except per share amounts): + +
Year Ended December 31, 2023
Class AClass BClass CConsolidated
Basic net income per share:
Numerator
Allocation of distributed earnings (cash dividends paid)$ 0$ 0$ 0$ 0
Allocation of undistributed earnings34,6015,12434,07073,795
Net income$ 34,601$ 5,124$ 34,070$ 73,795
Denominator
Number of shares used in per share computation5,9228775,83112,630
Basic net income per share$ 5.84$ 5.84$ 5.84$ 5.84
Diluted net income per share:
Numerator
Allocation of total earnings for basic computation$ 34,601$ 5,124$ 34,070$ 73,795
Reallocation of total earnings as a result of conversion of Class B to Class A shares5,12400(1)
Reallocation of undistributed earnings(287)(37)287(1)
Net income$ 39,438$ 5,087$ 34,357$ 73,795
Denominator
Number of shares used in basic computation5,9228775,83112,630
Weighted-average effect of dilutive securities
Add:
Conversion of Class B to Class A shares outstanding87700(1)
Restricted stock units and other contingently issuable shares009292
Number of shares used in per share computation6,7998775,92312,722
Diluted net income per share$ 5.80$ 5.80$ 5.80$ 5.80
+ +(1) Not applicable for consolidated net income per share. + +
Year Ended December 31, 2024
Class AClass BClass CConsolidated
Basic net income per share:
Numerator
Allocation of distributed earnings (cash dividends paid)$ 3,509$ 519$ 3,335$ 7,363
Allocation of undistributed earnings44,0856,52042,15092,755
Net income$ 47,594$ 7,039$ 45,485$ 100,118
Denominator
Number of shares used in per share computation5,8558665,59812,319
Basic net income per share$ 8.13$ 8.13$ 8.13$ 8.13
Diluted net income per share:
Numerator
Allocation of total earnings for basic computation$ 47,594$ 7,039$ 45,485$ 100,118
Reallocation of total earnings as a result of conversion of Class B to Class A shares7,03900(1)
Reallocation of undistributed earnings(520)(67)520(1)
Net income$ 54,113$ 6,972$ 46,005$ 100,118
Denominator
Number of shares used in basic computation5,8558665,59812,319
Weighted-average effect of dilutive securities
Add:
Conversion of Class B to Class A shares outstanding86600(1)
Restricted stock units and other contingently issuable shares00128128
Number of shares used in per share computation6,7218665,72612,447
Diluted net income per share$ 8.05$ 8.05$ 8.03$ 8.04
+ +(1) Not applicable for consolidated net income per share. + +
Year Ended December 31, 2025
Class AClass BClass CConsolidated
Basic net income per share:
Numerator
Allocation of distributed earnings (cash dividends paid)$ 4,832$ 703$ 4,514$ 10,049
Allocation of undistributed earnings58,6828,55754,882122,121
Net income$ 63,514$ 9,260$ 59,396$ 132,170
Denominator
Number of shares used in per share computation5,8228495,44512,116
Basic net income per share$ 10.91$ 10.91$ 10.91$ 10.91
Diluted net income per share:
Numerator
Allocation of total earnings for basic computation$ 63,514$ 9,260$ 59,396$ 132,170
Reallocation of total earnings as a result of conversion of Class B to Class A shares9,26000(1)
Reallocation of undistributed earnings(627)(79)627(1)
Net income$ 72,147$ 9,181$ 60,023$ 132,170
Denominator
Number of shares used in basic computation5,8228495,44512,116
Weighted-average effect of dilutive securities
Add:
Conversion of Class B to Class A shares outstanding84900(1)
Restricted stock units and other contingently issuable shares00114114
Number of shares used in per share computation6,6718495,55912,230
Diluted net income per share$ 10.82$ 10.81$ 10.80$ 10.81
+ +(1) Not applicable for consolidated net income per share. + +## Note 13. Compensation Plans + +## Stock Plans + +Our stock plans include the Alphabet Amended and Restated 2021 Stock Plan ("Alphabet 2021 Stock Plan") and Other Bets stock-based plans. Under our stock plans, RSUs and other types of awards may be granted. Under the Alphabet 2021 Stock Plan, an RSU award is an agreement to issue shares of our Class C stock at the time the award vests. RSUs generally vest over four years contingent upon employment on the vesting date. RSUs are awarded dividend equivalents, which are subject to the same vesting conditions as the underlying award, and settled in Class C shares. + +As of December 31, 2025, there were 534 million shares of Class C stock reserved for future issuance under the Alphabet 2021 Stock Plan. + +## Stock-Based Compensation + +For the years ended December 31, 2023, 2024, and 2025, total SBC expense was \$22.1 billion, \$22.8 billion, and \$27.1 billion, including amounts associated with awards we expect to settle in Alphabet stock of \$21.7 billion, \$22.0 billion, and \$24.1 billion, respectively. + +For the years ended December 31, 2023, 2024, and 2025, we recognized tax benefits on total SBC expense, which are reflected in the provision for income taxes, of \$4.5 billion, \$4.6 billion, and \$5.0 billion, respectively. + +For the years ended December 31, 2023, 2024, and 2025, tax benefit realized related to awards vested or exercised during the period was \$5.6 billion, \$6.8 billion, and \$8.1 billion, respectively. These amounts do not include the indirect effects of stock-based awards, which primarily relate to the research and development tax credit. + +## Stock-Based Award Activities + +The following table summarizes the activities for unvested Alphabet RSUs, which include dividend equivalents awarded to holders of unvested stock, for the year ended December 31, 2025 (in millions, except per share amounts): + +
Number of SharesWeighted-Average Grant-Date Fair Value
Unvested as of December 31, 2024299$ 122.77
Granted198$ 188.82
Vested(181)$ 133.90
Forfeited/canceled(34)$ 142.33
Unvested as of December 31, 2025282$ 159.75
+ +The weighted-average grant-date fair value of RSUs granted during the years ended December 31, 2023 and 2024 was \$97.59 and \$140.04, respectively. Total fair value of RSUs, as of their respective vesting dates, during the years ended December 31, 2023, 2024, and 2025, were \$26.6 billion, \$33.3 billion, and \$39.7 billion, respectively. + +As of December 31, 2025, there was \$42.9 billion of unrecognized compensation cost related to unvested RSUs. This amount is expected to be recognized over a weighted-average period of 2.6 years. + +## Note 14. Income Taxes + +Income from continuing operations before income taxes consisted of the following (in millions): + +
Year Ended December 31,
202320242025
Domestic operations$ 73,600$ 108,076$ 143,591
Foreign operations12,11711,73915,235
Total$ 85,717$ 119,815$ 158,826
+ +Provision for income taxes consisted of the following (in millions): + +
Year Ended December 31,
202320242025
Current:
Federal and state$ 15,716$ 21,101$ 13,378
Foreign3,9353,8525,028
Total19,65124,95318,406
Deferred:
Federal and state(7,482)(5,350)8,243
Foreign(247)947
Total(7,729)(5,256)8,250
Provision for income taxes$ 11,922$ 19,697$ 26,656
+ +The reconciliation of federal statutory income tax rate to our effective income tax rate was as follows: + +
Year Ended December 31,
202320242025
US federal statutory rate18,00121.0 %25,16121.0 %33,35321.0 %
State and local income taxes, net of federal income tax effect(1)8231.0 %1,1991.0 %1,6061.0 %
Foreign tax effects:
Brazil:
Withholding taxes1,0641.2 %1,0410.9 %1,3840.9 %
Other620.1 %120.0 %230.0 %
Other foreign jurisdictions(74)(0.1)%3530.3 %3960.2 %
Effect of change in tax laws or rates enacted in the current period(829)(1.0)%00.0 %00.0 %
Effect of cross-border tax laws:
Foreign-derived intangible income deduction(3,980)(4.6)%(4,568)(3.8)%(3,931)(2.5)%
Other2150.2 %3210.3 %2950.2 %
Tax credits:
Federal research credit(1,575)(1.8)%(1,792)(1.5)%(2,088)(1.3)%
Foreign tax credits(1,396)(1.6)%(1,373)(1.1)%(1,684)(1.1)%
Other(498)(0.6)%(198)(0.2)%(98)(0.1)%
Changes in valuation allowances5130.6 %6030.5 %1,1700.7 %
Nontaxable or nondeductible items:
Stock-based compensation expense(602)(0.7)%(1,743)(1.5)%(2,601)(1.6)%
Other1690.2 %2030.2 %9550.6 %
Changes in unrecognized tax benefits4320.5 %6890.6 %(1,123)(0.7)%
Other adjustments(403)(0.5)%(211)(0.2)%(1,002)(0.6)%
Total$ 11,92213.9 %$ 19,69716.4 %$ 26,65616.8 %
+ +(1) The tax effect in this category primarily reflects state and local taxes in New York state, New York city, Pennsylvania, Minnesota, Illinois, New Jersey and Wisconsin. + +In 2023, the IRS issued a rule change allowing taxpayers to temporarily apply the regulations in effect prior to 2022 related to US federal foreign tax credits as well as a separate rule change with guidance on the capitalization and amortization of research and development expenses. A cumulative one-time adjustment for these tax rule changes was recorded in 2023. + +Changes to US tax law enacted on July 4, 2025, allow for immediate expensing of domestic research and experimentation costs, accelerated depreciation on eligible capital expenditures, and other tax law changes impacting 2025 with certain changes effective in 2026. These changes are reflected in our results for the year ended December 31, 2025. + +## Deferred Income Taxes + +Deferred income taxes reflect the net effects of temporary differences between the carrying amounts of assets and liabilities for financial reporting purposes and the amounts used for income tax purposes. Significant components of our deferred tax assets and liabilities were as follows (in millions): + +
As of December 31,
20242025
Deferred tax assets:
Accrued employee benefits$ 1,834$ 1,951
Accruals and reserves not currently deductible2,5523,570
Tax credits6,3847,314
Net operating losses3,4724,953
Operating leases3,3363,337
Capitalized research and development25,90324,758
Other1,3762,143
Total deferred tax assets44,85748,026
Valuation allowance(11,493)(13,942)
Total deferred tax assets net of valuation allowance33,36434,084
Deferred tax liabilities:
Property and equipment, net(9,932)(13,256)
Net investment gains(2,978)(8,242)
Operating leases(2,986)(3,103)
Other(1,008)(1,289)
Total deferred tax liabilities(16,904)(25,890)
Net deferred tax assets (liabilities)$ 16,460$ 8,194
+ +As of December 31, 2025, our federal, state, and foreign net operating loss carryforwards for income tax purposes were approximately \$13.0 billion, \$25.1 billion, and \$2.9 billion respectively. If not utilized, the federal, foreign and state net operating loss carryforwards will all begin to expire in 2026. It is more likely than not that the majority of the net operating loss carryforwards will not be realized. The net operating loss carryforwards are subject to various annual limitations under the tax laws of the different jurisdictions. + +As of December 31, 2025, our Federal and California research and development credit carryforwards for income tax purposes were approximately \$771 million and \$6.4 billion, respectively. If not utilized, the Federal research and development credit will begin to expire in 2037 and the California research and development credit can be carried over indefinitely. We believe the majority of the federal tax credit and state tax credit is not likely to be realized. + +As of December 31, 2025, our investment tax credit carryforwards for state income tax purposes were approximately \$1.3 billion and will begin to expire in 2033. We use the flow-through method of accounting for investment tax credits. We believe this tax credit is not likely to be realized. + +As of December 31, 2025, we maintained a valuation allowance with respect to California deferred tax assets, certain federal net operating losses, certain state net operating losses and tax credits, net deferred tax assets relating to certain Other Bet companies, and certain foreign net operating losses that we believe are not likely to be realized. We continue to reassess the remaining valuation allowance quarterly, and if future evidence allows for a partial or full release of the valuation allowance, a tax benefit will be recorded accordingly. + +Cash paid for income taxes, net of refunds, were as follows (in millions): + +
Year Ended December 31,
202320242025
US federal$ 13,689$ 19,921$ 13,658
US state and local1,2242,6972,919
Foreign:
Brazil1,2641,1011,368
Other2,9873,6343,581
Total foreign4,2514,7354,949
Total cash paid for income taxes, net of refunds$ 19,164$ 27,353$ 21,526
+ +## Uncertain Tax Positions + +The following table summarizes the activity related to our gross unrecognized tax benefits (in millions): + +
Year Ended December 31,
202320242025
Beginning gross unrecognized tax benefits$ 7,055$ 9,438$ 12,619
Increases related to prior year tax positions740896278
Decreases related to prior year tax positions(682)(83)(1,301)
Decreases related to settlement with tax authorities(21)(311)(2,183)
Increases related to current year tax positions2,3462,6792,099
Ending gross unrecognized tax benefits$ 9,438$ 12,619$ 11,512
+ +We are subject to income taxes in the US and foreign jurisdictions. Significant judgment is required in evaluating our uncertain tax positions and determining our provision for income taxes. The total amount of gross unrecognized tax benefits was \$9.4 billion, \$12.6 billion, and \$11.5 billion as of December 31, 2023, 2024, and 2025, respectively, of which \$7.4 billion, \$10.0 billion, and \$9.7 billion, if recognized, would affect our effective tax rate, respectively. + +As of December 31, 2024 and 2025, we accrued \$1.1 billion and \$1.2 billion in interest and penalties in provision for income taxes, respectively. + +We are subject to the continuous examination of our income tax returns by the IRS and other tax authorities. The IRS is currently examining our 2019 through 2021 tax returns. We have also received tax assessments in multiple foreign jurisdictions asserting transfer pricing adjustments or permanent establishment. We continue to defend such claims as presented. + +We regularly assess the likelihood of adverse outcomes resulting from these examinations to determine the adequacy of our provision for income taxes. We continue to monitor the progress of ongoing discussions with tax authorities and the effect, if any, of the expected expiration of the statute of limitations in various taxing jurisdictions. + +We believe that an adequate provision has been made for any adjustments that may result from tax examinations. However, the outcome of tax audits cannot be predicted with certainty. If any issues addressed in our tax audits are resolved in a manner not consistent with management's expectations, we could be required to adjust our provision for income taxes in the period such resolutions occur. + +## Note 15. Information about Segments and Geographic Areas + +We report our segment results as Google Services, Google Cloud, and Other Bets: + +- Google Services includes products and services such as ads, Android, Chrome, devices, Google Maps, Google Play, Search, and YouTube. Google Services generates revenues primarily from advertising; fees received for consumer subscription-based products such as YouTube TV, YouTube Music and Premium, and NFL Sunday Ticket, as well as Google One; the sale of apps and in-app purchases; and devices. +- Google Cloud includes infrastructure and platform services, applications, and other services for enterprise customers. Google Cloud generates revenues primarily from consumption-based fees and subscriptions received for Google Cloud Platform services, Google Workspace communication and collaboration tools, and other enterprise services. + +\- Other Bets is a combination of multiple operating segments that are not individually material. Revenues from Other Bets are generated primarily from the sale of autonomous transportation services and internet services. + +Revenues, certain costs, such as costs associated with content and traffic acquisition, certain engineering activities, and devices, as well as certain operating expenses are directly attributable to our segments. Due to the integrated nature of Alphabet, other costs and expenses, such as technical infrastructure and office facilities, are managed centrally at a consolidated level. These costs, including the associated depreciation, are allocated to operating segments as a service cost generally based on usage, headcount, or revenue. + +Certain costs are not allocated to our segments because they represent Alphabet-level activities. These costs primarily include: + +- certain AI-focused shared research and development activities, including employee compensation expenses and technical infrastructure usage costs associated with the development of our general AI models; +• corporate initiatives such as our philanthropic activities; and +• corporate shared costs such as certain finance, human resource, and legal costs, including certain fines and settlements. + +Charges associated with employee severance and office space reductions are also not allocated to our segments. Additionally, hedging gains (losses) related to revenue are not allocated to our segments. + +Our Chief Operating Decision Maker (CODM) is our Chief Executive Officer, Sundar Pichai. Our CODM uses segment operating income (loss) to allocate resources to our segments in our annual planning process and to assess the performance of our segments, primarily by monitoring actual results versus the annual plan. Our operating segments are not evaluated using asset information. + +The following table presents revenue, profitability, and expense information about our segments (in millions): + +
Year Ended December 31,
202320242025
Revenues:
Google Services$ 272,543$ 304,930$ 342,721
Google Cloud33,08843,22958,705
Other Bets1,5271,6481,537
Hedging gains (losses)236211(127)
Total revenues$ 307,394$ 350,018$ 402,836
Operating income (loss):
Google Services$ 95,858$ 121,263$ 139,404
Google Cloud1,7166,11213,910
Other Bets(4,095)(4,444)(7,515)
Alphabet-level activities(9,186)(10,541)(16,760)
Total income from operations$ 84,293$ 112,390$ 129,039
Supplemental information about segment expenses:
Google Services:
Employee compensation expenses$ 46,224$ 44,560$ 45,124
Other costs and expenses130,461139,107158,193
Total Google Services costs and expenses$ 176,685$ 183,667$ 203,317
Google Cloud:
Employee compensation expenses$ 19,054$ 20,519$ 22,078
Other costs and expenses12,31816,59822,717
Total Google Cloud costs and expenses$ 31,372$ 37,117$ 44,795
+ +Google Services and Google Cloud employee compensation expenses include the costs associated with direct and allocated employees. Google Services and Google Cloud other costs and expenses primarily include direct costs, such as advertising and promotional activities, legal and other matters, and third-party services fees as well as allocated costs, such as technical infrastructure and office facilities usage costs. Additionally, Google Services other costs and expenses include content and traffic acquisition costs and device costs. + +See Note 2 for further details relating to revenues by geography. + +The following table presents long-lived assets by geographic area, which includes property and equipment, net and operating lease assets (in millions): + +
As of December 31,
20242025
Long-lived assets:
United States$ 138,993$ 195,337
International45,63166,481
Total long-lived assets$ 184,624$ 261,818
+ +## Note 16. Subsequent Event + +In January 2026, we recognized approximately \$32.0 billion of unrealized gains in our non-marketable investments. These unrealized gains reflect an estimated increase in the fair value measurement following observable transactions that occurred in January 2026, and are subject to change as we finalize related valuations. See Note 3 and Note 7 for further details on equity investments and OI&E. + +## ITEM 9. CHANGES IN AND DISAGREEMENTS WITH ACCOUNTANTS ON ACCOUNTING AND FINANCIAL DISCLOSURE + +None. + +## ITEM 9A. CONTROLS AND PROCEDURES + +## Evaluation of Disclosure Controls and Procedures + +Our management, with the participation of our chief executive officer and chief financial officer, evaluated the effectiveness of our disclosure controls and procedures pursuant to Rule 13a-15 under the Exchange Act, as of the end of the period covered by this Annual Report on Form 10-K. + +Based on this evaluation, our chief executive officer and chief financial officer concluded that, as of December 31, 2025, our disclosure controls and procedures are designed at a reasonable assurance level and are effective to provide reasonable assurance that information we are required to disclose in reports that we file or submit under the Exchange Act is recorded, processed, summarized, and reported within the time periods specified in the SEC's rules and forms, and that such information is accumulated and communicated to our management, including our chief executive officer and chief financial officer, as appropriate, to allow timely decisions regarding required disclosure. + +## Changes in Internal Control over Financial Reporting + +There have been no changes in our internal control over financial reporting that occurred during the quarter ended December 31, 2025 that have materially affected, or are reasonably likely to materially affect, our internal control over financial reporting. + +## Management's Report on Internal Control over Financial Reporting + +Our management is responsible for establishing and maintaining adequate internal control over financial reporting, as defined in Rule 13a-15(f) of the Exchange Act. Our management conducted an evaluation of the effectiveness of our internal control over financial reporting based on the framework in Internal Control—Integrated Framework issued by the Committee of Sponsoring Organizations of the Treadway Commission (2013 framework). Based on this evaluation, management concluded that our internal control over financial reporting was effective as of December 31, 2025. Management reviewed the results of its assessment with our Audit Committee. The effectiveness of our internal control over financial reporting as of December 31, 2025 has been audited by Ernst & Young LLP, an independent registered public accounting firm, as stated in its report which is included in Item 8 of this Annual Report on Form 10-K. + +## Limitations on Effectiveness of Controls and Procedures + +In designing and evaluating the disclosure controls and procedures, management recognizes that any controls and procedures, no matter how well designed and operated, can provide only reasonable assurance of achieving the desired control objectives. In addition, the design of disclosure controls and procedures must reflect the fact that there are resource constraints and that management is required to apply its judgment in evaluating the benefits of possible controls and procedures relative to their costs. + +## ITEM 9B. OTHER INFORMATION + +## 10b5-1 Trading Plans + +During the quarter ended December 31, 2025, the following Section 16 director and officer adopted, modified, or terminated a "Rule 10b5-1 trading arrangement" (as defined in Item 408 of Regulation S-K of the Exchange Act): + +- John Hennessy, Chair of the Board of Directors, through the John L. Hennessy & Andrea J. Hennessy Revocable Trust, adopted a new trading plan on November 10, 2025 (with the first trade under the new plan scheduled for March 15, 2026). The trading plan will be effective until March 15, 2027 to sell up to 8,400 shares of Class C Capital Stock and up to 4,200 shares of Class A Common Stock. +- Ruth M. Porat, President and Chief Investment Officer, adopted a new trading plan on November 29, 2025 (with the first trade under the new plan scheduled for March 2, 2026). The trading plan is scheduled to be in effect until March 2, 2027 to sell up to 154,486 shares (gross, plus any dividend equivalent units) of Class C Capital Stock issued upon the vesting of Ruth's Alphabet 2021 Performance Stock Units, as adjusted based on performance (shares sold are net of tax withholding). + +There were no "non-Rule 10b5-1 trading arrangements" (as defined in Item 408 of Regulation S-K of the Exchange Act) adopted, modified, or terminated during the quarter ended December 31, 2025 by our directors and Section 16 officers. Each of the Rule 10b5-1 trading arrangements are in accordance with our Policy Against Insider + +Trading and actual sale transactions made pursuant to such trading arrangements will be disclosed publicly in Section 16 filings with the SEC in accordance with applicable securities laws, rules, and regulations. + +## Required Disclosure Pursuant to Section 13(r) of the Exchange Act + +As previously disclosed, Google LLC, a subsidiary of Alphabet, filed notifications with the Russian Federal Security Service (FSB) pursuant to Russian encryption control requirements, which must be complied with prior to the import of covered items. The information provided pursuant to Section 13(r) of the Exchange Act in Part II, Item 5 of our Quarterly Report on Form 10-Q for the quarter ended September 30, 2025 is incorporated herein by reference. + +## ITEM 9C. DISCLOSURE REGARDING FOREIGN JURISDICTIONS THAT PREVENT INSPECTIONS + +Not applicable. + +## PART III + +## ITEM 10. DIRECTORS, EXECUTIVE OFFICERS, AND CORPORATE GOVERNANCE + +The information required by this item will be included under the caption "Directors, Executive Officers, and Corporate Governance" in our Proxy Statement for the 2026 Annual Meeting of Stockholders to be filed with the SEC within 120 days of the fiscal year ended December 31, 2025 (2026 Proxy Statement) and is incorporated herein by reference. The information required by this item regarding delinquent filers pursuant to Item 405 of Regulation S-K will be included under the caption "Delinquent Section 16(a) Reports" in the 2026 Proxy Statement and is incorporated herein by reference. + +We have adopted insider trading policies and procedures governing the purchase, sale, and other dispositions of securities of Alphabet by directors, officers, and employees that we believe are reasonably designed to promote compliance with insider trading laws, rules and regulations, and applicable Nasdaq listing standards. Our insider trading policy states, among other things, that our directors, officers, and employees are prohibited from trading in such securities while in possession of material, nonpublic information. The foregoing summary of our insider trading policies and procedures does not purport to be complete and is qualified by reference to our Policy Against Insider Trading filed as Exhibit 19.1 to our 2024 Annual Report on Form 10-K and incorporated by reference herein. + +## ITEM 11. EXECUTIVE COMPENSATION + +The information required by this item will be included under the captions "Director Compensation," "Executive Compensation" and "Directors, Executive Officers, and Corporate Governance—Corporate Governance and Board Matters—Compensation Committee Interlocks and Insider Participation" in the 2026 Proxy Statement and is incorporated herein by reference, except as to information disclosed therein pursuant to Item 402(v) of Regulation S-K relating to pay versus performance. + +## ITEM 12. SECURITY OWNERSHIP OF CERTAIN BENEFICIAL OWNERS AND MANAGEMENT AND RELATED STOCKHOLDER MATTERS + +The information required by this item will be included under the captions "Common Stock Ownership of Certain Beneficial Owners and Management" and "Equity Compensation Plan Information" in the 2026 Proxy Statement and is incorporated herein by reference. + +## ITEM 13. CERTAIN RELATIONSHIPS AND RELATED TRANSACTIONS, AND DIRECTOR INDEPENDENCE + +The information required by this item will be included under the captions "Certain Relationships and Related Transactions" and "Directors, Executive Officers, and Corporate Governance—Corporate Governance and Board Matters—Director Independence" in the 2026 Proxy Statement and is incorporated herein by reference. + +## ITEM 14. PRINCIPAL ACCOUNTANT FEES AND SERVICES + +The information required by this item will be included under the caption "Independent Registered Public Accounting Firm" in the 2026 Proxy Statement and is incorporated herein by reference. + +## PART IV + +## ITEM 15. EXHIBITS, FINANCIAL STATEMENT SCHEDULES + +We have filed the following documents as part of this Annual Report on Form 10-K: + +## 1. Consolidated Financial Statements + +Reports of Independent Registered Public Accounting Firm 45 + +Financial Statements: + +Consolidated Balance Sheets 48 + +Consolidated Statements of Income 49 + +Consolidated Statements of Comprehensive Income 50 + +Consolidated Statements of Stockholders' Equity 51 + +Consolidated Statements of Cash Flows 52 + +Notes to Consolidated Financial Statements 53 + +## 2. Financial Statement Schedules + +## Schedule II: Valuation and Qualifying Accounts + +The table below details the activity of the allowance for credit losses and sales credits for the years ended December 31, 2023, 2024, and 2025 (in millions): + +
Balance atBeginning of YearAdditionsUsageBalance atEnd of Year
Year ended December 31, 2023$ 1,213$ 3,115$ (2,737)$ 1,591
Year ended December 31, 2024$ 1,591$ 2,895$ (2,850)$ 1,636
Year ended December 31, 2025$ 1,636$ 4,128$ (3,408)$ 2,356
+ +Note: Additions to the allowance for credit losses are charged to expense. Additions to the allowance for sales credits are charged against revenues. + +All other schedules have been omitted because they are not required, not applicable, or the required information is otherwise included. + +## 3. Exhibits + +
Exhibit NumberDescriptionIncorporated by reference herein
FormDate
2.01Agreement and Plan of Merger, dated October 2, 2015, by and among Google Inc., the Registrant and Maple Technologies Inc.Current Report on Form 8-K (File No. 001-37580)October 2, 2015
3.01Amended and Restated Certificate of Incorporation of the RegistrantCurrent Report on Form 8-K (File No. 001-37580)June 3, 2022
3.02Amended and Restated Bylaws of the Registrant, dated October 19, 2022Current Report on Form 8-K (File No. 001-37580)October 25, 2022
4.01Specimen Class A Common Stock certificateCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.02Specimen Class C Capital Stock certificateCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.03Alphabet Inc. Deferred Compensation PlanCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.04Transfer Restriction Agreement, dated October 2, 2015, between the Registrant and Larry Page and certain of his affiliatesCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.05Transfer Restriction Agreement, dated October 2, 2015, between the Registrant and Sergey Brin and certain of his affiliatesCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.06Joinder Agreement, dated December 31, 2021, among the Registrant, Sergey Brin and certain of his affiliatesAnnual Report on Form 10-K (File No. 001-37580)February 2, 2022
4.07Transfer Restriction Agreement, dated October 2, 2015, between the Registrant and Eric E. Schmidt and certain of its affiliatesCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.08Class C Undertaking, dated October 2, 2015, executed by the RegistrantCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.09Indenture, dated February 12, 2016, between the Registrant and The Bank of New York Mellon Trust Company, N.A., as TrusteeRegistration Statement on Form S-3 (File No. 333-209510)February 12, 2016
4.10Registrant Registration Rights Agreement dated December 14, 2015Registration Statement on Form S-3 (File No. 333-209518)February 12, 2016
4.11First Supplemental Indenture, dated April 27, 2016, between the Registrant and The Bank of New York Mellon Trust Company, N.A., as trusteeCurrent Report on Form 8-K (File No. 001-37580)April 27, 2016
4.12Form of the Registrant's 1.998% Note due 2026Current Report on Form 8-K (File No. 001-37580)August 9, 2016
4.13Form of Global Note representing the Registrant's 0.800% notes due 2027Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.14Form of Global Note representing the Registrant's 1.100% notes due 2030Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.15Form of Global Note representing the Registrant's 1.900% notes due 2040Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.16Form of Global Note representing the Registrant's 2.050% notes due 2050Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.17Form of Global Note representing the Registrant's 2.250% notes due 2060Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.18Form of Global Note representing the Registrant's 4.000% notes due 2030Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.19Form of Global Note representing the Registrant's 4.500% notes due 2035Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.20Form of Global Note representing the Registrant's 5.250% notes due 2055Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.21Form of Global Note representing the Registrant's 5.300% notes due 2065Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.22Form of Global Note representing the Registrant's 2.500% notes due 2029Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.23Form of Global Note representing the Registrant's 3.000% notes due 2033Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.24Form of Global Note representing the Registrant's 3.375% notes due 2037Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.25Form of Global Note representing the Registrant's 3.875% notes due 2045Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.26Form of Global Note representing the Registrant's 4.000% notes due 2054Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.27Form of Global Note representing the Registrant's 2.375% notes due 2028Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.28Form of Global Note representing the Registrant's 2.875% notes due 2031Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.29Form of Global Note representing the Registrant's 3.125% notes due 2034Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.30Form of Global Note representing the Registrant's 3.500% notes due 2038Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.31Form of Global Note representing the Registrant's 4.000% notes due 2044Current Report on Form 8-K (File No. 001-37580)November 6, 2025
Exhibit NumberDescriptionIncorporated by reference herein
FormDate
4.32Form of Global Note representing the Registrant's 4.375% notes due 2064Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.33Form of Global Note representing the Registrant's floating rate notes due 2028Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.34Form of Global Note representing the Registrant's 3.875% notes due 2028Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.35Form of Global Note representing the Registrant's 4.100% notes due 2030Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.36Form of Global Note representing the Registrant's 4.375% notes due 2032Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.37Form of Global Note representing the Registrant's 4.700% notes due 2035Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.38Form of Global Note representing the Registrant's 5.350% notes due 2045Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.39Form of Global Note representing the Registrant's 5.450% notes due 2055Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.40Form of Global Note representing the Registrant's 5.700% notes due 2075Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.41Description of Registrant's SecuritiesAnnual Report on Form 10-K (File No. 001-37580)February 3, 2023
10.01Form of Indemnification Agreement entered into between the Registrant, its affiliates and its directors and officersCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.02Form of Offer Letter for DirectorsAnnual Report on Form 10-K (File No. 001-37580)January 31, 2024
10.03Letter Agreement dated June 3, 2024, between Anat Ashkenazi and AlphabetCurrent Report on Form 8-K (File No. 001-37580)June 7, 2024
10.04Compensation Plan Agreement, dated October 2, 2015, between Google Inc. and the RegistrantCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.05Director Arrangements Agreement, dated October 2, 2015, between Google Inc. and the RegistrantCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.06Alphabet Inc. Deferred Compensation PlanCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.07Alphabet Inc. Amended and Restated 2012 Stock PlanCurrent Report on Form 8-K (File No. 001-37580)June 5, 2020
10.07.1Alphabet Inc. Amended and Restated 2012 Stock Plan - Form of Alphabet Restricted Stock Unit AgreementAnnual Report on Form 10-K (File No. 001-37580)February 4, 2020
10.08Alphabet Inc. Amended and Restated 2021 Stock PlanCurrent Report on Form 8-K (file No. 001-37580)June 8, 2023
10.08.1Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet Restricted Stock Unit AgreementQuarterly Report on Form 10-Q (file No. 001-37580)July 28, 2021
10.08.2Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet Restricted Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)July 26, 2023
10.08.3Alphabet Inc. Amended and Restated 2021 Stock Plan – Form of Alphabet Restricted Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)April 26, 2024
10.08.4Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet 2022 CEO Performance Stock Unit AgreementAnnual Report on Form 10-K (File No. 001-37580)February 3, 2023
10.08.5Alphabet Inc. Amended and Restated 2021 Stock Plan – Form of Alphabet CEO Performance Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)April 26, 2024
10.08.6Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet 2022 Non-CEO Performance Stock Unit AgreementAnnual Report on Form 10-K (File No. 001-37580)February 2, 2022
10.08.7Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet 2023 Non-CEO Performance Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)July 26, 2023
10.08.8Alphabet Inc. Amended and Restated 2021 Stock Plan – Form of Alphabet Non-CEO Performance Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)April 26, 2024
10.09Alphabet Inc. Company Bonus Plan, as amendedAnnual Report on Form 10-K (File No. 001-37350)February 2, 2023
19.01Alphabet Inc. Policy Against Insider TradingAnnual Report on Form 10-K (File No. 001-37350)February 5, 2025
21.01*Subsidiaries of the Registrant
23.01*Consent of Independent Registered Public Accounting Firm
24.01*Power of Attorney (incorporated by reference to the signature page of this Annual Report on Form 10-K)
31.01*Certification of Chief Executive Officer pursuant to Exchange Act Rules 13a-14(a) and 15d-14(a), as adopted pursuant to Section 302 of the Sarbanes-Oxley Act of 2002
31.02*Certification of Chief Financial Officer pursuant to Exchange Act Rules 13a-14(a) and 15d-14(a), as adopted pursuant to Section 302 of the Sarbanes-Oxley Act of 2002
32.01Certifications of Chief Executive Officer and Chief Financial Officer pursuant to 18 U.S.C. Section 1350, as adopted pursuant to Section 906 of the Sarbanes-Oxley Act of 2002
97.01Clawback PolicyAnnual Report on Form 10-K (File No. 001-37350)January 31, 2024
101.INS*Inline XBRL Instance Document - the instance document does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document
101.SCH*Inline XBRL Taxonomy Extension Schema Document
101.CAL*Inline XBRL Taxonomy Extension Calculation Linkbase Document
101.DEF*Inline XBRL Taxonomy Extension Definition Linkbase Document
101.LAB*Inline XBRL Taxonomy Extension Label Linkbase Document
101.PRE*Inline XBRL Taxonomy Extension Presentation Linkbase Document
104*Cover Page Interactive Data File (embedded within the Inline XBRL document and contained in Exhibit 101)
+ +◆ Indicates management compensatory plan, contract, or arrangement. +\* Filed herewith. +‡ Furnished herewith. + +## ITEM 16. FORM 10-K SUMMARY + +None. + +## SIGNATURES + +Pursuant to the requirements of Section 13 or 15(d) of the Securities Exchange Act of 1934, the Registrant has duly caused this Annual Report on Form 10-K to be signed on its behalf by the undersigned, thereunto duly authorized. + +Date: February 4, 2026 + +ALPHABET INC. + +
By:/S/ SUNDAR PICHAI
Sundar Pichai
Chief Executive Officer(Principal Executive Officer of the Registrant)
+ +## POWER OF ATTORNEY + +KNOW ALL PERSONS BY THESE PRESENTS, that each person whose signature appears below constitutes and appoints Sundar Pichai and Anat Ashkenazi, jointly and severally, his or her attorney-in-fact, with the power of substitution, for him or her in any and all capacities, to sign any amendments to this Annual Report on Form 10-K and to file the same, with exhibits thereto and other documents in connection therewith, with the Securities and Exchange Commission, hereby ratifying and confirming all that each of said attorneys-in-fact, or his or her substitute or substitutes, may do or cause to be done by virtue hereof. + +Pursuant to the requirements of the Securities Exchange Act of 1934, this Annual Report on Form 10-K has been signed below by the following persons on behalf of the Registrant and in the capacities and on the dates indicated. + +
SignatureTitleDate
/S/SUNDAR PICHAIChief Executive Officer and Director (Principal Executive Officer)February 4, 2026
Sundar Pichai
/S/ANAT ASHKENAZISenior Vice President and Chief Financial Officer (Principal Financial Officer)February 4, 2026
Anat Ashkenazi
/S/AMIE THUENER O'TOOLEVice President, Corporate Controller and Principal Accounting OfficerFebruary 4, 2026
Amie Thuener O'Toole
/S/FRANCES H. ARNOLDDirectorFebruary 4, 2026
Frances H. Arnold
/S/SERGEY BRINCo-Founder and DirectorFebruary 4, 2026
Sergey Brin
/S/R. MARTIN CHAVEZDirectorFebruary 4, 2026
R. Martin Chávez
/S/L. JOHN DOERRDirectorFebruary 4, 2026
L. John Doerr
/S/ROGER W. FERGUSON JR.DirectorFebruary 4, 2026
Roger W. Ferguson Jr.
/S/JOHN L. HENNESSYDirector, ChairFebruary 4, 2026
John L. Hennessy
/S/LARRY PAGECo-Founder and DirectorFebruary 4, 2026
Larry Page
/S/K. RAM SHRIRAMDirectorFebruary 4, 2026
K. Ram Shriram
/S/ROBIN L. WASHINGTONDirectorFebruary 4, 2026
Robin L. Washington
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/images/image-1-Stock Performance Comparison.jpg b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/images/image-1-Stock Performance Comparison.jpg new file mode 100644 index 0000000000000000000000000000000000000000..fa5c1a32c1dd89a9d01c80db76279e88c179c04a GIT binary patch literal 90564 zcmeEucU)85mUa*krHb?#5L6TarC4YIlrB;wbVNi11f+L@ARxU7CFx zN636gNiKuV(2uLF7&Bm-D>x#ormVhtPbY&zu4-P%tsGu(Dmb zdX4}34KeXMckkVoc<@j`QAt@vRY&)U9#r4J(Biq}3#*sbHqI`tZtfnQUV(3ef;|YTYE=mSNHJ9=-BwgkIAV8Bx-SK zd1ZBN9kaW)e{gt&J;wc-7YT^;U#11z|I5Vw&AjM$7uX&N2aR&w|-B~i8 z+vhIHX_1>d(q9&NbDlvy_Dk{i3%sJ*XvSwwLljJWVhdL=zoz!v%>HW=3;2g-_UDQH z>%68wRHP)p;*rvUAfO{0e}My|M`BO23}^MYLra;a|5@v!%xA9|yyIor#h?+8Hd=Q5 z^~$@ILGmgc$3!(u%4_B-VlVp*Bg8{|dps66hkd_F83VDL!MKqm&>Da;l1yd9Gl z8&MbKbt78m*9WIuYm?l#u(^m44&2!{5y3Po`0N$rW$y1WfOIfLrEPu!BXzuADsp{H z6>jndvnZ~798l1bWiNw%`i_w~?9nUv<2mv_%jJzb_)IHSf!SIKsk8FD7FBoG+1^Y&czB`ky#~|ZPAK~)vRAQDUTphY zR_$zeL+TikXFrD78qg`cy@NXA?obs*o{O(W2-jo3EaY!l88&L%Sva7+Byg{wE?8(U z(WhNe(w<{~zt3(Jc428uigY2x#7t+jtcM`kn@X1X#_*2Y%zfT~a}q9HpK07%{a9;; zY2TLnTE1=I9&=sIsgEamo`T%2>{Hw;szo;9pkZ}n*W@_Sjzx2(AhfZJipKc6nn3#oJ^F~mv-M#dwOYpr= z$LRaB6Nw%x9foGD)MvmHSVaR@N`kg>>l<@dCQL9@@8X-NgpF~gQ1cwa$uHeT&9TBm zvfyF=o4U@OgVQ~i6^BcBo`#RT<0Mm206m2{tb_N3O^MHn*4T>2$eL`Sl^$RV-f!0+ zLDqWYA*Y~4fud8;#QgqeYTQ5TYW!dRJ|gMpgSRq#aMOp7S7~H#s@P?iF)J;GBD7*W1vhv0h(QU%SYhw+2{82>%zpn#V46x0a?ycVKy z&FJ9G!YSyOVx#fkh`>+k_sbOhG5#OVfj~Sq7MY6&|D29|JS-NsmCEf8nWgCf!OemA zrz8v=GDPGy@JQ0EF2m?nCNDK4QBSuVTHYQ=dezLFeU66a{(#NCw0j$_A{xC-{c<&) z^M%VpCP_+9r%H0S`xN&f2_U!k(727e@Jn@^pW*a263eN%-_K3dj*um{(_H|)jd4;Q zfQ{Ec*s+<`o zoH9#x9y@K7M5ZCJXJ9H-dKstxLqsB2TQYFipA+zfB!6L9D>r_(`8CfG z7_?-`{uqDfIeWoUY zhp|3NR%r6}w7U5HYrEKvGg@6r6WIeR^5@hP4oDiM~ByjPy@T!w9n0N=TL+pNN2*qC9eV%ybJ9k%rhP)%e z=CD3P9iy}`5i$^B&pl#CRXk(&fqp6POsy>G+1vqBnj00ld3a?ESBP-P#P^}K3Fz%( zW{HT(uY>8a!IoNGodZSE>&vA;gaQ4ARtz*>uu~Y(8k|uL+hvGaUAb96(s3;|P11Yt zoJJU)7YRPM8jpo5Vu~u=J)fo9z7t;AN@+>BcvThDXtgY6Sm_s zEh&&-3X!NSn(>jvjK|G!$S|5vuz$L8uzwVK`Rzb29W)-^fg5iO$CEljg2nFD-CIqD zhos%V=`-@?>L7`>x78W{n?4rj{TMb)zo9Jh;ppfvNOK`aa)T$Cr5 zfG1y+s6ZZ*{`c?C$TiwC)BSHjD#guk%=WOI^-&qdWR)G)k5}G&ajQO?!I5xp%|!4I z_y7MI?jX3ZSp~kO#^LIbrHsjzm^sBo;sfwUg#M^Q`iRSP%4GER&xq=yrZwzDFxIWz zi~-vXDZ?WT)Z4yqm2t#k_wlyz%vXoAE3F2fi&?O0FzYM?;9(1mqGNdKj8l*W)^L*3 zQ3^GXG$Apx?DqkX$DANd^J_ zsA??JEN!W-DwK#kfP|ic25mzyocj&eG4C5+9JiK)dpl0Qya)c+g}0q?*}cILJlUd~H`pS=0wQ?&uUZ`LY4n?s_X97_<~Xe`_%Q4r2)T{lzr^au`xz0BQX( z{;TJJ1ghN_{%AzVAsdJ(&RTXJbU1WmdBB1TEHO$E=_eZfeD7zB)7s>^@crZ{lgc(b zi_Q_dvJcZMRStJFAyxI{NZ39bthpY%iF?wXe;I$IULpdks)vcfPnN6lh=(O14}FEm z3E_fiSMWMRFf;FCu0RC_sP*2TEc-9({vVD%g8Y9Hkm2MnvhDEdID`Zh8L-pO_#uUj zD|N8I-V}+8F#cFbAyTL(mQm{_mC?#vg3WK&_Jyy|KjNCPi2Tb74>XD-Qg@k*b93|H}HsF8H+ zU5zoZT1RPO(X(icfX6|LtYr26pK0d8cI_;*u%SQISbXFPHkc&FlZyg~{gMiypY`r_ z-&*>n0qd@TaA2oQXyr$Y9)?L4Y8WK&jErkkCxUiln&^%P$?X(8WeW2>ju!UalE>={ zCO48LyJJn;IMbUC7ozU+nAMRS;An)y)ULG#X&p&Qhq)Y0^-GtD;F}<>2Cu2tM(a!69 z(q>KF*CpL_AM%mgPrM^M-de|~bk&JXD=k2r~f}0=h{MAr^*xS-O&im;iE7O)du# z7!(Z{FX(@7M@60Ue(@WWt#C!)n3hgKO_`@4ijd}#9`WsxJ=v#j)nqZ>>~c6o|A4dq zs2xz`(Jj3QV&yyhOnV<&Sm3GWRwHG%_sC~WmbiM7fw>Psvz4Mtl}>~r@8n)as#Q#V z+zjtncRr^0DQ=rvXM(zR>p3CYk%4tHl>}c6iS#Z{$kb!HWhi@_;r6>xZ;~)%#>rHF zjzaN%b;t>mna25s>ox!lZX)w=UrWIaF80@g*oT64<7-8U=Yv{4E57HR({&<|cI-lC z{gtH9ALjobUG{%{T@e&XQ%s0ou#RP?pCv9x%R)Se2HW{H9bX9EVoPl=yo^umS!&fd`r!m4IR$-z z7LH$gyARuhP)`|@AfGinpWmsxw!4Pao0VWVzMg57{H&oo?EUz(umZst$ol=*`%m%g zpYZNqV46R&xBr;pN3?uaOU&fbj;4>_vWPC%riT_bPM|}IE~}v)xta+VQKDr3_~R5r zN6rh;<2@)!K{gb(Wk=J)Mm&p;`H5D$g|J`=$PX#0iD99u^>!`N@-)micq^0va4;rH z=`tzQ#Fj+W%BkG+ePR(2Rhb?pP<9;i_t0*CWX}EyefOs=@LzdM;R|K3W?JgklxrK~*`G#Kkyrl~)C>IVyMJ&BVriOSjn(QKT|D|e0Brfa zA}JmF(FvW^5;n)2t=sWal$cXcFn+)kogGUEU8t43W2{Q=)A)aFdLIMSSUaplimm*L zo_%Je{el96j_6+^+CQ18B68H#99Wjmy}A^QL}gPI-e7}_lcDwwD!HvS>>#-jSACXx zfl6ivjMS_>K+*qhF9{#)L5JZG;#?)xVxiDXziy=Lj;`fU;r1)U@C&}jKx*AXS!#t8 zk}c4Pxe^kovoM>TkU2wH=!&lraDrBJtLEhtVSu;$ zO9zKxA|)3FS~fP`e+y}cCb9X0q%Llp?=|9BknOX_KN=M-e2U(AD;8dD_hMX#X>wbu z(Ie(7$vc@GUU(tM>j@)@>lXS6qQ!(wn8`yMc{sP;tlu}|WGoiy&ia(y;~GM`3h}hK zMs0)AwzF|;;8J&^A}D%Ami`xc_ZRB2Lfr4MRy|KE~I}-q1 zZyUV#{LgfKt{>wX@!w@8;qy76WLvtg{;=xH#JE|bJb>_{AG5f1Ze9Ka`amj7)+Pk-*} zn2VsO3U8}+b0YI2!R|g*SEpG9f9hA`!eMo{&5@@d+L@QQcjgRLma(*b>`@WYZ6Hss zEc7;~K)c}}Iyl$!i<(tdhc8R;i`%tKKKajA9J-}Lz91f=ArZmF=(&;Bw&Yun zi(IfM4I472zqR_DdvBj$JM7=ov?fcYB%yMpE|j4`H{vZ{igOt%FbbQ?|Cx$*Gt!+H ztOo#Hu~QIRNOQ3|^>#`8jOVUkC)LTB?Ry(<{y@5apIa=@V78h6(ZEtyqc?qBLrBsdu3VQpNb<<76b_q`NrVR|rb%3doP}u4g{jIOfXWux7AjnpmL z2SyLwAH0xCAO)1h%o`o}bM??>oW|j`kei$DTJut}PC-7qurvt8t{KUidLg#rx>Zky zy?SrKe5)J^qSvu#JCh$z@##}uL(ZU8d?bYnT65~g&IGRZIEX>ZB&7cz64U)&gERjZ zSXY)E;K%-Eiudvg1qYx%poD=nlq-Upb>L`uuj(jc%v& z2j|}Pi(>Wj7i=}%@|o(D0Qt;||1hy?YJf2^qLr?Qq;~}GWOw8O#yuVzwqLvP!)4>b zju1PYY|xJr5>xu@X=9I$_8V9G?XyK5DWSOm4bU4b^y&@!Zb~=rV;U}*Vf{Qf6*KA- zMB*2cg%=-cEcLm2Sj|j7fA`~nW?&F82UFCa^2>^F*f3rsd%_9g%pDtgOhKom6dTs6 zMrZiMatye{^qhgazf-OV9RI)b3ey#rwpAZOU9w@0H3~d>0RBJ#Ee_iW8fd=FK7Llz zvkzqzZaoFjbnrZ}%>@y!gajVP*c?Ph``q|bxJH!owM(uVG2xzDEOsMse9c{efW!?iL!o;I>uugPTs|{Y3CA*hHvWpnbt9coBTtTwR~e+lvtGm}GpiD0 zNg-tiWu=e3x~`uL-^~RkIhL!VLD6s{xzxwhQwQsz;Y4XC4irpznpUFolm(~kxVBa- zS$D63M15JFTK#x)P#{c!tvP9)I0Zegy;T0rsnn?IIQ{&it6CqDajO1CS^;_ z&W*u9?TI+N7)GwFmW^|oh;&YIyS5EE@(Vd8*pBC?M?TS(>cs3(F?@5Ky>B;kBnOCH zaxVURt;&;8#LwaVjqr`cR!ttaqkYnuG=BT3;&vf*MafcsY!2f5)|Vk}Semh#j3oa_ zJ~jNKNG1^E z$GD>@$5>eRTK>)DQ2EPI{s83_7Is+92`%}7Gwv#&p<&_o(H=LOK>Q2^!jLVQ=YK00 z`SqLt#Av)gN;+d^z2*~=UCni7D+2zM+)0C_kOJ9o!^8d*)Fwk#T|B}dJ>$fOX?=XQ z_OTmLjHFk%eMeX~PHlk7W*o;Ia9^S2!AYy$v$IjvO%I|V2*$O$GCaet@DM`1^u9{A zr@Z-Vjc(oj4(Ct%8>WyPAK5`6A4^}_h@19embJrQij&82OcVwNl~iwk$DgI zM|CmFDqEKm8Vda3E3#}1TFT0rD%>PpSN)m(B0XW}n>SDNZUewYSSxx@z{YDq`}|$w zi@sk;KmUoX`=9lQ^6^IQY*Nx$y+EDFFL*%qx*+ujfQR1D?YE&@et~@@{Yd?ZYVue@ zoi%D{XmWq+W@geXU(D{e6Yt7hg22E&oW?dyxYj%2=DN{Qk~@{#%El<%9QS0N%tJ=w z(wzA0^1%qwA+@x(kT3}Qh_DGbU-(p+JJKTTcDdF$5Z726n^UWd=V~oj$&69ydi1vz zfzkE(1FVbGGWhN5^C9qj<>#lM?AQL}%eIo*Sgo(JLHj4G_Qm~O%sm$Si_rs&8!mb~ zlq&i9dpv2V3?VDYV*@U7^8#$(_sV73x$k95TUO*2w8k(k)6#`#mNnv)@aM3qI<2Hi z=YcT|M?=H@pq!2I;5i}*H`uX*CS***1_Qr{wbenpWpTD@-}je!S{oe!ey|EleLD6? zCSO18lhxb4iJM7R%m(nMAko>?B7yII65%FKiW>!((#gBMZqA_wPy9@03*F=$lb1vH zsy>Muiz_|OD+MrR!OCyI^7)9818JXVo&sBpm}<2g3nHWqIUtlsAWD##FX)rjCZhLbB*<2x1vd_FO%Kn4j$;^0dMh=&EY_TiXo zwbD{3_OpqUw3)n(Da6KK__qLk2Jn`&w3zvj>6R&MRXLfUy-MPxx>rv(^^rtPmw!?j`G%66Y=qJ83PkS&vh*uP} zzo6z_!t?$vyBXv<9*egP!7|om`YPcgcDE0%69tGAc&)8bEd_V5(7z3m8J8gRcx53W z<_8r%9T#X{%dHcV&>IIGS?*kd7zRI%A-?Lf>>p*+T%@NbRwYhBxQ*^Q)`epqEhMS< z+)8<9-A0lFif>VP6h0!}6`BGp?d1M_AvT85;xQtN-SR1@u&#e+0TB}QF(XxV;kL5} ziE)&k%skwzyEaVrSF;69P*zdcYt67QsaJYeJ(7p>rGEessnxxI@9+LnuL-@5p>0jh zd5agCYEQCYoR7cZ`{g;6TdUt572QbK`Wh$G!{@G41-En~0{BVWTz42fB%h4=$bB`P zGcd6}k~n*r?*&z#(AurM`Ue3@Rr$%_uxzX|T3EGyEZwTX^}4kk&v(1tHj;f3PaM?9 z4&BqDHz$x{q%NQ_g--9VJ5-qDcoQl@GkKph$7W4jkXd!yB1&54QAtCTXf@M1+577U zByZY40L30yE_~r?rW6!lFPW`tLf$R0Du@I!#vMN76p@Ar>mV?S%j|`Bo9DzwI861e zy#faKVtkrfmB~WB2}HNo!mx5uKxL?WHc&|&unNN_PgIT+6y9WNp6Z`t7uOD$yO&@= z;T8GWf+HEUXBOIUXGmNaq@KMxl~C~oR&chmys{!jvSOL_iV|Z&NcSGa+B$m|3 z#j0|UFN$(pF7;ja=gJXo;_-_CZSxGOys3SRe-Xk z$xvsOfyQga^EpGZEc58lRoFglG!f^3$xCCd2p&E$3_g2-%{b@@r$6KHU!(y3XBL=? z9@LvsQ!2wXtClRx5}bRAS2NaaTgy(QAG8zsWh~j_S*6L-O;#{1{AiU)>-q&`}hkBuSfk5}DD}4HsR7VX?cmdy4#Og;x+`wubFkI^X zW;$iomNut`8#|ulBnygcNU@+!EimT}lFy6Y8!rHC#}zx;5AQ#vIJf~O;B;Pmz^M`aFUc18eEpU-91Jx|`OEA2`2+ z#Cb)J`jxjxIAsWVdArG^&_;gwFrxqJMZA$34oJoMIZ0I}{C9 zeabRmGU3annE8mL2j#Yu*GWcSn0`J5No^Omu_d|N`_p7+BCccOkwdP$FWp35 z@J|!2jG}uXfvb`y1ObV1#(R6Bl_UXaA|*MTH4UB1_y+~19zCsLf+#mdha2NyxwM>_ z0|-zp2cdp>3+t{B+%6>s_p&C(RImG`hI}j1S*Q&1YqJs0@5DGtAj!;+E&@h)6!=z! zUW^vOfRa5_aB_9!UZ>>KRiBG^GdyiCfKk?^07w0cUY7vUWY5;QZVrjfgd$MZ40 z2Qk68>8hbRVz=)rLP_~#nnB>rdPlX);vV$I>agyDMUqH|l%=~)cF!Piwg1@rv(*3s zQyOS17s;^rxaXL9>a>&P+t(e0tX0^99b?X;<}J6$>CHk1x73P;c1WNjZlY96UxWIw zSDDub;}zx*3ON}@fa;L0UnSNjdv zlE0@wNKj^%0Jxu0NUeIdt$giD@zC>~;{*Ig16ufcFbhG2=LR3QC(qFb;{U4Ul0|>##pOUzr2S%$~IQUN8pmib%xWNvlz< z3TyvEVeI?EJ+tw&3hCXju<>Q~Y5gL*#Y+!v1S^@oWlOA${$QQ+;x~uLU&PzaB~vkk zO6*W~P0>8b>bOojF;?FQQ4Q_C1&tjFZXRnL)m;;4+<1-IiAYi`x4YA<>BV}Qx z?hr9s$2chGPvP^aZbv8b#}`I^+DD#(1mPpAlEfz(&2vn382ij4W?Jqt+8X3SwUF0O z`myLWJfWh$f3A60Kutf9|FX%~@E_1C_qyu{nTM;s=Wf(X6$Lgg%&tR14~aEy^G>gU zmTdmOns3VhGQ+RyhVAgo2g~y&eBArmPJR2N@v7Qd(D;&bo2Ge5xYWcX@>FkUjtkR{ ziv)5VEEH}FxnzHJmMNY#CRJ)pRv1eYi0ON(zv{IUCZyhEU`gd?n-a_e6P;KU=*|C_ zeatUpg=MLk<;@*ihtu<@53RWcKbSJWntfNgw#V3IY}3Gv0f49oas{&6&d8%QN&}%c z=}$nT)XS%icBK~s^9Zf@kZXH*fuf4C%F?$F?9RS!m$+%e;Fka&yR9E-+uu_{Z0t3T zPih?y9wD-?SKhPjGB0~DN?#B3v+39{QIBq^XBs7aeYMt)Xt8D1<%sWd6&YY473s`B z5SKTxxpFV;4$QeJ_iM=vGsdtn4Y3XIfesoJOX&RgDQh;@V3{rphYR4=SbvyIxl{GA zxOdJ;mL{wO(w;u+?L242HQDB*_71(aYra?-5r^yP8Gc1b=&5D%4hc)Vi1tGSe z1#4K2@f+6e4uI)BB1=>N_!Ftu;0LGYP-;7wVET#+;jaxxv$++gF(?(HXAb%5Q-V(%t3 zM+!1$E$rt-uOkhleVbMxV4E!&FxKpZR`tZB`x$Xf?y1u49L+}46#(}dH?|C%^DR+6 zC{%>(yWFn~4Z$@--ks3QH4cp1roH9J+Opb}_L*RO9CXFMv_MsQ`fHtN0f}2x$scU;3h(AgJ@`%( zhqq{_sAYNg_7s7aFiheOT*2qW-9LY_#!t9aBrSmQ8H0KMV={?9FV@Ojzq-?t9 z?5Kt(z>Y!?jeA9xU_o=-QS!wpD$vrKnXuc&v4^9+qeao5a|tOQh=)3d&Km?ff{oXw zxvPVP9XsOTbeq?-VcfRNn%IPe4ep1cY97RxcE{ri@}*=X57q#Afp*}^`v-(v6{4!G zX~zQ349%B3>c*MM==f968OXwut=pHT?uh8$%uXy9xbFpPQoLFZC^*NC8p?U%F3zl0h?@ub6X}g zzfTr>5j%zOyu^Q<3NlY0jEV|Clas z8||V}8=LtFHH|e@8t7K9tx(r>mA29tTG(?a(&qMCk56>?%#r*PNXNY!?(RDm(O~RL z-Gb7Wy1oOH)#dwE7L#7ounf(=--LEseYGJDE?2;;F7cjg)k34Y9GV zjy0TDg-bg|`eV+ltV+<17rV?;Z}?q-Z%Ge{)2pVs2m#uDhQR&^pbo?<;>Djd{V?~(a+5$8o2#1{M>8E-24m)?R)|KBJ5gCWn zWD}2YMCG+|Hp!FTw&3RV)twu%kk*yLmo4e7b={jQgRcc1hi$jKCelQe8t)9?8~c}? z+jJABF1A@a(Bi6Mz{bysuHp>P(#M~!7wPjQC5A}atpjE(|Mx+`->hZ*3$EawE*G>i z*}zGMGuDi5U3MH}vz$G)5>L3+N=w<^_W0`s@NDn{T+mln?nB*?Lfb@-_B1F-p^HA)k zgG18lL&op6a7Mv&YNl{|E7^%|%!O?6l6nPSMh;K8EALF@KTGhp&Y^ z`OFkeNRC@eTXRMHW;0~tLCAh-ifVfi?T5wM< zz#ufE3@qlh)uOiT%X}a{3TSDUg2S}e_Y0SHcQ1fzBY+o3!kuwYD;)5-W`T8@p6o zCOLSoHwoy2;H-5cCZu(k-rC+y%lH#eY2dJjQK}3h| zKySJF0?D*U1ukr$d@pr8D7$FTRtfP_ zT>6BIH`R{x@vf_%g24}tSf-x14{_{-G=Qh9PlUt;E1|qG9|%+69E=|%!R;vsL<(|Y z(AAkyaNs)DCz&3d$f0m~Jvv(u2RPr9yKes5`?wM1gm zZb`rk)vYo`7q&(6Cne#CEFXf;RD+0puyMU5!S47_>D5ccZD(pd$HjL!_~TTRd3f7_WF_c@Sk9Kp~3w1U!I_bwg+q%)?I z{r|}B{BMBM;)P~a}V@;)6#mInRx0f!55dih{$liddc zu#M4e*MYFailr|@Tf|{hw2X0{A89T@;9b=?UIXbl#UsAdzx{1Gmx8j zroPfw1OCMlrXPNpKG_vzjbRdfA1Qq~phHJ|uVB57IB1C;P5#eWO8uD$Cc4ZVqG8MK z9bb

~hBXNCm_YyAfsuhN?joRH8FCm z9AUeHwhNYoqBWO~EPtyrN`uhTu}6vXV=JW z@|Q1lp|M}9f7+M<&Ix)#+?#t@WYHpXcf{u*69*yrgyTEU$^3R2;GL#wyfz|GS3F9a z5b{|Us-EbXD=<7^1up2nU#at7eGMaHTQ5^AtG!heO9wQz`3*LCsa*IS5&W&YO)4pl zcA?TZ`egN|pTXDnC~#=XQ1f`vmZ~6*(^bSEJwTLin9mWR=LVu}ortx8Gm0A(cI6ng zy$H#G&p|di*7R@dd!)3*U+}5znMvb|i3wcQ=!Q_NBnsUvV;p#qL_4On(%L5nGZSt- zuv0?PhPZ4Zru~LdSi2=Zbc`3d0N$#mydL)xCw-wHm`nW_h%? zjz5v$CZ3oB)J=tlh@9Yb)Ly6Ku@IR$Ms!zfsq5QxCE&44!}p}-_Byz(j={Utx_D=s z!iL=7P0=X_={-%vtK(PpOox>N4-&U3h)?nJS>ASOsg^m}J(9|T6IEfis_}V!;%}G7 zGmEZUb;m6=4{H(Y3UO($4E)WJEd$FjAuddZj5c*FQF>s_WVXD3>Ge&f!44|5BW-aC z@}D;h1+p0yXoI-!Tw^!8v||XxH~dTwyV0+i{%@OOW+ZzQqC_-qJPgd2BqPO$_pftF z4Hh`%zMjygU3lXJ*~xf=8+R4Z$wu6X`4l&^_VpU+wJcA!42-a5ih#K-W67cW@iZRt zKwnXV4jbFqw}p)FWc>j-)H3ziGlVW>e}W$nUOnFsN=_^EMW)g}?u2W;tsVb#PT2u^gOhx#YBZH{fMlxa`7I)Hlq}4o7qfKlZITVjd-hr zrGXhn+eYW3+Z2Gt&%h?(G0m*dyWfKdm1v!00)%deXAGr&&r$mY*%Ll$`Ng`m4Ik3- z?avb~Lh}th%pR0<#Ah2Vx>%X4?7bS#%S#XXWzb1MS_f1L8pyGe#!Rk$4L6TE8CneI z$&M|noV2Ms1xzP-yb_?o@166m)vteFN{6F3kPuL|6k~346PK6PfCMhMkb+;1@xin> zipZC>yEXmVc-)*`0{UNUdoI=??C5Iw>; zGo*8qiOcA_pS%ka$EEVsM{?Yb#2a>Xi~QxZt)CK(-r#U*=bew_t+##7H1o&iuQMZj z*>e_5Cp2dbjZB-L)dbwnfV5l=1{XSM2t=5Pv)(J=(?lbUc#2QTi-0w6 zGP}ZKE{GxAI{HgyMt>t>!`w8}9V%`0wKKg=;{@D_bHLS_DBg4raU>-kXAws*s+hx6 zU0<~k;#g&e`sc&hp)Fql@+X+6$jkt)RgGJKDb&2Uf{`Xi+V)aMhlx8RD!WPsLG~v6 z=KHD&BkwLW|(PUdEu>D#-8Ej2z*`>yq_;2{)~ zc^OI7YCwo8x;5yS{ZP%b4_a-BjyVlOGJ}0I*O88wloz ztia6mCvIl&I0@^K_wzSZ5aoR3-B-EH9=B}huO>-EEbgwbzUL;q8KKG6LMS%qg^FiJJ~EZUi65zK+|iFi#SlL|B?Slvj) zGzRft;6d9io>sRc6XLG>4aid8d)L59*ufUNK{ymm$S$(*e!22iwF}a1fr@cn>+|T9xs1sPkh1i3ELz3bzL-}J z60#RNzLVfDxfK&|U)s5e!wq3l30J~Qywy|aYfs935Ui~lmlKe~5%?)kmcz@(pnOc` zs&xa);9F)rCsw9)bby?twroUTV1$z=2WGV)(ZxkMRisg*47!bs>RansRD`RFV>dN@qt^Sz4n)$cC#vT zM+8VtP+}6r;urKP>qyIaEpeetH4aqDyI2jyIC-`D2&&ejE9YL4>V@67zii~~`QmAt zh5RCa7dI7}|BUMIiZR~9nvDQ>B2#$^!qV7)DQ2=}w@XC?^6EREa=w18zlg59U(^QE9B5JCh@1r}qU8$5k-C@PLx^>b)@Bqgau-k!K8*YBmSTZrOA{zC# za^sj#XvV&qhV`F|2JEBcw3L7d;5-qsgiQlF)oKoJ1}bH@AC=OO5;{AOo)CY$?EvCn z;3BqR(Y%72{yi|HjV5hcW~P#K$m{kXsMINja@U*e!MigEH#O3vx?KNhu#_Bx`56C z68qr->?RVEb_H9g0LInWB60u*7HFA>t;r`JX(5pqY=kjF&(LiZKngD%|D0F-)0wui z!7-&48L@~{K7Vhqc3`Pe2sYU<|7C*#hitv&w5CJl!pU)qHOWa@oQpZ&x2=n~p8Duh)CA7Lgi@^x=JXD{45>Vz^^=P=?CEM?<$u5 zo5vUQX|x(DM$k_0EQE&OOZ$QE(*WS|^+u+r>Rnm-VmW1%OTM0T z6QIZ6lXRc--P}*^8}*UPwB{z%Y6dWYn~a2kN{J|#Dkgk@(Icf*)RWTr^>mHg%PhIW z=3N-aSiR&z<2%~<}O$emP;UWG7M>jU!1ozPZW44_3Z4*-hD`-0Aj?CVePefIc4?sbWP6o z2o639HOj%ll!uECIVI+p!0q&cfVz?kW@tJ<2MPTFGwOal|2=#8q!9~Q^eE%v#3;SX zq8Be;Ts#H!*k*HE{HQc*t^qvY&A#B#14g+yN!JXan_eNG~aNzD%kZxNaA&_w^Qu80j!6Omazg5Qs23(kd@W;opbtp z&M|!W+jhfM8>dAhLnB#fBO#?-aF*kMNXvsQv%wHO%G`AqXm}4wgKqZDkl#jHID<`N z=ZlP8z&N_9>$yVa=AZ}2>Nz`d{JpybpU%ULbm=IRW=Yw=l!M?x2Yx?Bcuzdu6=g;{ z6__f==dj(HwTFEZwwDhh!3G1FcBYHoJvX8UFWDMSF9hwy9u~^a-Jb8!UO8wJE<{~J zYgZ!n1Jg1l+3v#Wx9BhTXe8@QW<3quWi9eLK{8dlQM96`0L^)t%D^%}A{kks=amA_ zA=Sd;XPBlMV^$7|f_vnL%Qo57W#;a1f}aW=r@h2ANFX|7o&ddlNbxO+nU!u*QJn0p ze%vvtPiw0NRIHvcqXh{9=_5pz85wrX2IUTyH}2qG6gU>4nqHQZp!Imq{*g_D*Sa&# zJz!r@lB(jFCpl(@b7gl8r-uUq)h84g{-7p?GUyiy&Fn%l3*ERMz-B4_&3rKG`N~gG z1;>cOTs_58WJ(1V<v!(+*B;wod#qmZtoIX+ zH}P@n@FfeA--klBq+AgrJQVvDbObz!9TVFdCv>d(RkwF_ru#7q>RuTcH#kT zD&ehClTBN)Pv7hsR5mylITCkcjJN{Kn{UL6YIc{Q^&$iD*!G1TKd)@Owap8h(&B)9+Yx7A+S}P~pqAh79v{*+s&G%FH|L3;)i= zFvANrPfz$4^w~fNJ8p~RdO-CyDg6aC5V_)(B*<3ip!ig*4eIS10}#``+4qe zzko>VLq7~) zyP1NgCX8u6cOUlDG-~qub2jP z-A`-94@{W-vUNGT==dot(ls{!RPyD^q_O;VseQfsH)NAOTd3LmsJ&)3o_EiK#p4!- z_qu}PkE{>!q3yF0qw;PGCyi1W-iD?@$CM}1OqP=;;-*{>aUm^!GD&}q#hd5FaHIHw zBb>3_MT+sVtZdr@=?gRrv~``X7TL-)-fQ7f6x~JV6xz>~Q2MR=U=Kl#M_Vtd#rf(5 z&VKOw1{*%WO__RcZ_ihMdHul32uNo7NG@{tyOs9;GM@dfjwkpv)~KFqchwnHum3f& zMBq>^fEFijn7oBz^3YI96_yG_y9~cuRU!KYg1%enGz#qE6xccS!#cLk+2Q@5ttXuM zMO*c6B3M!iEL0h9UC=ac4wsQyBceJ&`LYMMKX9DI$irJzagfh$zJ@^njCw0@U|(MQ z=X;E_b<`~R0NO9LE-Q`JVAG`I+ub?=%G^z*BW;RWiOkk0K|f}bVk*=E<+K5AqL!S}!3(cGYG*`h-_E-afpo^BkH%)JDa#e@D z4&n`$TK|&ykbF)}pPkpmr=fy%f?{FuT=F^#?tskTqK3WmF~`Zhl7)%!#L?{#-jhKt z&H)q$V&|?KKDWAAtjx*hfOAw&%}ERV{|AWvQiFb8NMT;`5bQ>CecYs13_m630HfQ7 zhB(U$l<*#xEDp{ERJI&M%1mk%>};K2@(AD#wMC74Ow6VP8m8M+QB}^aBgp5sn?(KU zsyvNXo^{bmtsHvCuXe%cKQxQB^Dftk8fTy_Zvedi5rk(K9Hh3e_PhB;fWAPNbQ?+F z56ulWMd0Wn%SQaBV028;l%t8Y^l?=33J0UU!U!q6$^2JdDJkJ)a;QUIg!bnMKJFS_WC;dG5{PU#WwcI+b{LAi(K0#EuwC>-LhI$usbA3YILU!vd4{o|D4TJDZhJFN+Rpil;}Fxd>f0cH zfN)G*NfefLa9FET=XwXL?C4brgshx=A4BZxYE;i~ra18}9mfEtV!;=oTxKenfXv46 z$BBR|uLJpl&`4=O?FnqnJTVTN*Uj<6p~mI?%wp)WR;Gg1hI#Y%9DCw(3-iz4?r%+J z9R%hVlI{Iuaf;H`_PE~_w0WFx`37JR$&&YaPViWye;&pk_IAlxeYzT8E?QJ&h>HzQ z-~~^ZZPi|GB2j6CsT{%u99`ARTq0JvG_-buU%t~|P(8s1n!e&EN2j0&66_me^ZrBA zaP3_GiUANCdab%EMnFcc=I`><{~UJ=W;5DXNys`~BYi55gDYYc+EcR>en_!hypa#B z2VW7&l(%89hG0eJLf6c^BRyDI)n<-W-WT3XPb%1w^^D^rZPhL;ajEryvZ6IyO_1%Q zXS;Gu0H;!34F}{LXo+AblkW{NA6q`0y%PqCllsM@NZ53&8fxa)@HDl+r5|fgH0yyr zY!OnIj>ms%VVAlafbsu{n_l`PJ1-IQrBPYnlZJ;PbS3iAqorw49DQG$pk#L;1a7NV^WtYo;SHUp0=vBFPFV0EX%ZL|&edN9=C!rN&Z_}?#lbYw9Jr!Ni37LPL?1K2ae&)g8F3cKa zkRNqp9v3T4fMc^sT2e{qkQxNQ_|cYU)ud!i=B5HOtr_L|2qWU5;_`cG%;`*XdQ|v3 z#R%Oy5j3Qsr*qFQDU|vd+NnUiPRdr$B8Y_Lu@x7bD>ucb2%)gDFjcSmhsH(~lo;wMT=xMn(!j~5)J1E9cowCB6N0ZV*;uccxq?k@qF0l^Dy~jiI|&5n~s|F_a`r3 zNt+ryDRwu^{q*adx04lH^PVN){JnO)^lf0w!aaZVZyK8m`8Cmt3b7)pF4{)Yo)Qxe zJb3d04LO>t28SdwJ_t@rMTP=Agf#adInBw&>TF-3yEpBoM;E zKzKJ+SHd*Nu-*&wENECK6?0+9->Q65F=1ul(1dV{VkvGBJ4Xc_?ndRRNUR{#gAGK^dDL{rdg^GAGv}!R}Bu*74^`yqS6GB|!mau1{~&m;>`udm6^*PA9f} zq#RP|Q`Dn^^>f~OVxxz?#aF}4Dia6yC>{a69M`_tsV1@oakqcGLX+|G zlx9{qu`*sivFk4;zvW-9aQ}mK@PD6|wIZ**)C57dA{IMxX-418k74e#~3x64rS?79KL&_#V z;X`#PwivQ&@#pp=31A`PHnP|@#QjQ{;(42ozDy@@i z>)I-19m=!vBz}PYHy&wydYq1GJdvB;5VQQ!lB}U%aT=-MOCGrpFJl_hA`3 z39m^S02P&Zc=;=If`j%DUmAb;GDEz<2zDT(p`x#I?PjD_JzU676hkZnn1P z{)O&6C zN**BYa`^QuT4Ig{H;?|ZHi`X5EB$}T3^1$tBJY!9XS!fWGM``i^$D9#$y_F`Q$6Y`k+{#@AE4^9(r{o>H>%U}e;xY0cEGDjOH%w&LFKL9b6Wa`C%ghZpzA zE%}-U>gq4w1=g;hSZ4>CZxpkbj;r-6_$^Oj`SMu8g9Yc4BDo)6+vYyS{&VLp;J2ujDgEoSbOT2k z$a5Kq9)EpL$UAUnH1>4;rMsmC9m>W$=wg#xdpEr}m%nhA?EkWDJbHLGmWD$u<6Ksq zf@2r)O{xjl@@S$9)bQY8z!Yi$kUN2{J2rDAZ7)poV1*I!ktA2(2#26%E0wqvi^Btn z1f%Rh9PHLQQDxblI{4r(Jl?+@hrGvS!~V;{BVcxy3WV77D4$*g{+-0Im8{P zezT*V8LfJBrygYWu})a52cCr^?^}G#_I|83=-K$f2XlsMs&Oc$I1{>HI%4<+bV?<< z&>U~m`Rz6E-SJF4cU|AOip@^qa$4G;Pf4+1u8RDqb?l*__AqTaM$kAivVAEUpcYaC7LQkiT49=9Qwdx zL#a#9OATUHWes(5;nILyirH8AibtwWz*nfrSXO-*SuQvqq27A}NI@yiPs;THjMJ{Kxg8>2W2G z9uW-n09UMMpkJViWO!Dt;gfLdK|}T>b5?gqgU#1Km10WJzBd43NevBJFl*jw{WZ&= zeFR;N{v26QGqRRF3qMz?9TF(PjiIvUkEAvb2o#agdE=I{6I^-yU~ z;`$9t!OSj-kZ9$k&|gRPrO*t^Eh9XLQ~t@x4Weht6`Zvvgm{Ijzn%y^-^k_&yWh9C z9(`yA@Kyt!W4(Ze^n$5X4;B{&>~pZ+z>O8^Q}aC7VHYBf!362)?bOw2dFwaS%g=?D z;#LhH(V1uT;ETwp+^tf8wm%ibM^(3cX1dNQm8aO7GyWKMvi3HD{Mq8^>j;XO9elUo zdH|H6nH7xvLv#CBcZ+8QNznB7@W6WnXaE3;d?(~9%87=VWiw^m3!0C!%G$SVFBI{buzNoE zcyPD=9<*Uyfk*+1w#}Yy#!R?~s!th-&osi)E%8cJTU=>=8yv9KP52=Mwcg8^1{T2g z{^k%nx-BLqYao^>2unv!``(P~3t?bn{9y5eT0?HQUfFb~OQpipl+Hbb3&UYVj2lU9 zP-Y&FefIO@m!@u9{P!sUI8FKQU7DGFxWkFAS3ZpWsYKrz4I_en{GwnkHQy!|y=F1p zb=q^*=P!{RGkOaUkp(5y48C9*nEnZr+z&_`B;oxS>z%sAw8UC%-b`4po7@!?9GZcY zsJ5=FyDMTJcwV9ykzF`3M}A)ZIedJg=bh?)-2n#hGaLONYm6FEH~V^NNlM9oXx1Oq zNX~`S`(I*vQJc6zD)JAbLJAPIHo8&KOIM$S~v^uQPdkGJW=BsC zJfK#=U9oe@@qwG_1&mImHv_Yf6K}*O?ExS&s7imN!FCC8di&!s`V8kVmn9~D%XAEX zAMbjSINT*^OUQnQK4e~#hF=yjywmd{TuV9SQed`J|Fk!>-Q;rKD;KvLb0TrX{^%qOFBONxf{dER*05Qxcwjo7K8 zT_4Oxc~6lizADK3{2(C~hQL2GS>?DR1!_|OXQNkV`spM!dnpcKR`t_f@>YqVWW%`{ zs7ciVQWThaM{G)ouRTp3R;^j@509xzAM;bjlRgOnrYt- zX7+m>ZkII9hVg1jl!o4hS?K`BwnUk#06q2ee-HrwB-!oXm-GUWJ>e4J^u=B8h0ltI zKgUmh%$fSZUxILH_q7e(A@`6zl&a|Pt<0@Mkk1z6RrjA%lbp_p$+8u)zzLTC#lGOm z5tME#UAW#c%d*?^cT&>XuYupkDUfH&>HHQZ4%_&$I-VED;}h>y0YK~H$;l1!z-!+NNsZ3^5ugzjRlfG>Rzm170SfAFqD42;4K>OZ^J<^q zy;m2tfrjw`wm=QUHA(K9c3P+EMbQFsqQE!t?{>A)E9yKB3$#PaTT~MjfCgppHWOF| zo)5e{s3hq}#)N!5*uy!ahAOuhmO2$d)x3)NOjm(Q+X@|3D!~KZy~5iHXi|ygqN8dU z8m(#FY}+@2?B^Hg-uZ6B{0rG`elGfAouwNK@W3S3qN?K+v94nNd{j%fFGKe?;idjeDpTC6P3r^EwK={qA%ujadN5=EsRIXS_uw9_`YX`{2Ble@S6 z!T}y}2#$$7OJOXAety;eZAtNZq3S-xaz4D86P_lXYT7W?X&pE2UM#jg43Q{{^!1Rc zEh-YrppiSJZZf#gt!YLmtcMX#sN9Jn29?a=?=}(#=C@~Ok-cU#r#=pagJ-fsvmeP< zX2l-JW#x^bIPFK^M0d`uBEMn zH!#^d^b%Y4L6JwIm(VD^<|pxyxs6as0pdq&Od1jj|_O^xG;gdIe#T>1g_8>&9|P#TlpDH@w4o{}75<lgoO zWcg?J{lASd%sqsNE)&fUoVWkbI5e{RTm{ubC!K>&NP2(MJ5*g_{`>lws%8~(C(%$Z zO2s&N{?w+14<@XcU->-A{K7@+b?0yX#nkmZoP2;|rSXmT(Xw^jzYhjcDXR;F~%QhXhMuzI?ZJft6abv-)Q4NRb%JF;r6F zPrGLJF@LK-{N-J%Hx%6R3D3HMRv871Aww1dq|u2}!rU|WF008M>`~8*8#Ib_xLkUC z*`D1_tRR?%@s$>$@YRx*`NW#lp+hao)vs(DWx3g^eBJ$)RNcxs<^;b*a~7?iwIdmq zWsB|ua=ts|B8t3rG1gC#bUCN2XaaK#m^!pWJ9yQ<8EkLgEj>rQJH=R+_iS$O8SRnO z8Gjw>F6;!bs2ocL|9)4z6UayoJ1*Px6=q&tTa0h2jQ>M}0_goPz~J3BSi6%8I4Gc6 z1Rd%y8Y(@Ho_bujp=Su-g>8d%e`qqyEb%}l&u1<|rrXr>P%8Cr@FG2VE|{e{)NN=T zDFL}I>SsRyoqLl!$5h3Gf9phxaivzGs>A@zIL+~Z!lR&vxA;#trn-d5PDgx-?i2Nh zj3dSw-`fo2+P`RRU%IPDS&qCCyvYLV`;qRV?0>OZ^ld%QAP&i4Z=zHA?m=T6``It# z3{s6&kJOsiWOdcz)XJyO&lnH|I_EXYzj4mbRkJ6TQ=s#d&Ve=n)whd8u(f)!I!{G8DA(4qM;P_# zLuUcqC$2OBXMFm?PWDqP?awsspzer(J3eJYfja?2s($hL?@lx~ANEpKSHEA9Nff-F z(-a9|0Sr7_kzz;8i6!WZK3;u%Updzv6rPXs^v!}5Us^9Ht;P1zA{=7ls7+8Y?_UA& zG5s8OwxO5c1ZJ@fe9Yl;4f$*^zzAi1tF^j7y`|=lIXT@4q>xIAxH!Q?1xXN~r6JIW zrwn%@#c&q%Zxb$ywOn0y_%>-?IsNHiYV-jGEFHgk)9GE+9w1=ILvfLObqR}VE&(oX zZHjCpL~AiHquJj>C&KU%)PBqE*g4(W8aYdRxSX^$VN|EzVF3Q)w?J+4mdn;)nYRs} zuja~(I=Wu7gmH2^+ZJ*#sDdGPNp)vdR&s-z}Y}Bm|E~JI8a{B+Gn! z=oYS2^)XX0zEIKbn2LJLU_4ClP%Y55c)YO3W@%SaUGu8+Eie{?(jj;?s_GF^ceo0y zVKC@il4tzdMYAa2!nR`rePsR@i(OcGiJ-t+)4$?}%C;>07EvMCIhLluM@sK}{M_0g zB-9L`1#h|(9v;7>ZKA5la?6xX(N}XW4kyp@bf2MSehObBxEVgzRQc^n{GCG^@ReC9kp{R(SlCCbkisG&y4O30+9ErpAGY3J*Yjv>KT}RjTlhB zAEwC*Qu#x}j>)v=gba8%6*vd}k^xX8FNo;!48&wM{mwVy_@TIuBAC7yCb3%u1m$>K z*ExOj^o8387jhB(cmEZpEO0BR@Xc)+3AG(#R(zQI9K)1PQbpi?h}CBqFCwFsTZhk$ zhQ%4H4zEXJQriF|`t2&|X%!qkubj-@(Ol$szp>+lc{x)_Q`Gg%E0g@QH;Ql1 z&goLlYYo+`?9WDIn^0g%!!_JT4-SSEoAdWO-i;QeSFLn6c#vQ!-sJN}b?ju!(qtn$~qGKK(eG}}uCg^B^gG)wPL zi+$cX+e5y06ycrt&yWwxR2f77|3YV`2Y|WhR|0G(%?V&aQaYhB$GgVC>pq3{!ezpL zi{Gp})R;T4PAmPA4)J~3`O4%{vMvXX3b>~d&RB)0vI9$P;ROVW!6d4Hr<8Mu5N{^0*z zM_Lhf(Q;YCSL6DuX$aJI|J8Qp_yHIeYH}oVFg6P%u8Ba7ZiD4FX88s$f*CU3(i(iJ zp6m5SI%Sn5>Z*II<*8@X-|O7*_gbXWN+?Zy2??=lrKm7&nH&S*4SR+l`(Xr7bV{9d zPga|*t!b7w0r{)Naz22$H0KrSyY;j;7>$t;&gTXh%_JSsGDGw_AI}#K=SAfjCP0bhqVM{83_rGgp zk8%VjYskF;7eU_z9wA)!A5Ewd@27xCziwMD)4$w=T&XJlji`$rWF#M$)vQAEz{nbc zTjnD-u6p>S+B?Yp0N5Fgu2)HJMMx5?TnB$nqxWSbn~~AQIjgb2DXbr;tqQJXy$J#Jaw1@r3n% z?M#ooC1SMCLb3b+A(Bh~y?4u{wRLkB+UKZ4&wT-FeMbd9AE>`rVqu>QZBU&+H_Q=_ zEbryVMu|qvqzDVoY*~r&lfXlW$Jx&be@)R08cs!mA|o#{JkMOOs^a9xS?Lc?`-WmX zB&#I~{@WZ~c39zq3v?Jir9wmFd&vYmogI>`^nNHDmIX}G>WhFB!C?8n=_vo9r~I#% z%7bsXYFp>uhRrCDqun8nVE#c7!Y0$QYoP3q=+Edk!-H(7f_x zEc`_@^fJ`~X@)PibACOOo?7aCd&`t4 zx*4#={}^jfvuGSRDnbxmmP=h-0w_P)MB)*?RBh4u#2#ERml-SQdOW`IUbV(@YiPS1 zNq#h1Rqf~}KVeed@-*6}3pca}a~ z{(&Y{YVQUp=G8fVcT_IxhkHBcfDXC`oXOKg9RoUEUil73JZufjO^SEk8#bHd?^+a0+| zH@@Fma)-+2_|;KA=uH};`)?BRwB3$xjq)zfSjs#Mpt>Z(GNfiWL0yJsgw5`6e!`X~ z(2XZ<9r&bqRR4kW>OLqK*dk6`dVgrKD8ouP0=ohyTJe4iy<&JeIOmLZ$(di-_ao&&NsAsGlp9D0YIJ?n@ZE z!F$dKkJAMZELY4)jQJ;7{qOByAzA-x?Ms3i7hbBCpRVQbnpaqs%N?kCu**^qvi=s@ z8hV!Uu#|I_)CpqnxOYX`*ZtEn`u5EAX2Oc;CPemd6Kg?x?WD|O^1$~GO&q!RhU~kc zoU41FanG5f0u+`V`kDWj8A&+}{MK7sseqjo66Nchr-Ny~TKfIHH&f|zp#W0;Qp}!3 z?)Aa`OX`=yc%7xefKt>qNEQ=k`iW0@(u23z605Nk`Kd}4U9EKXC)b_w)Jd`Q^3Wdefstk%WHLl zfs6nw8Nyu(N^pCuXYC!=TJ5_jIe4z7x$D|TG)Zg^w`JkPUf&;F^~0} zv-D5^a(K>p@1o{+R3Cnwd1&INFPCUX095v~9L-IN35bQcFYjA_Oc~DznV)Nm{PBmT zQI6tEag>$$lnDq-4FZ7qmhegi@jcJihS{R;81tfrX;O=Q5VZIh{)a~22&pYOItUO; z*IU}g8bMiGx#_zDLTgl0^Z{hi7t|%y0;FVJs-g=RviB13{GJ`R4DHVZ5RAJe4lnW* z2Gwys>~N(4kj3i7x~L$skwEr;}hM`=8i%66)t*9e-fv}qoWu#;D^v7zgTuEo2=-wU}$VawxhpRKm)YXKymA(=5Ct8w@# zp`6VLW>|jLuh;1(Sr5QJO1IQ8?oBq*_V{v5*VqpWKm;#$J(md+!Fc)g`)y4&LfLTh z)}EwYR`|%`-?}4M5pOnBNseeBc_({3ny*=wDGfYaRE#>WTyJJp^SZDKdar-ONUwb` zfBzS^* zT#ob4<6jNJTDJA5igI&|*NYa&M$Sj*cULMm;j^RtsZbTL$NOoYk_*ug=rL>qI)#t1v6zic!U8xP1O&@l(dS_E*^gG54E^@%Qa9Lv7eZZHR21L)NHeQ)8n zco9grSO2;gs2`M!q*ea`gS3?_K3IsZyBjQO1LL zehM`*F48O4G}U}%i^YppQqAB)^$6*O-1r9pDy!;=yD{X56C5PhOV7|;sDhE+q~TD8 z`DPX(`nydYdd1jO`Lx%C+Wrw7B+4W!d{0f;nDsNW$s%=jifc}oGP*owJvs9GX&5tY z;W8#IqEF1*(Ek6D;=d1HZQPoQh5-urhHruK0YUpGVs*sSL&%swKS7d@kK*}!YO~N^ z=jn!BsUt~o-{-&-%MBADig@-lFG)g6n6t^xC6nf;PZf(BFxf+C2zh#3|%`J$SV%5kVOo z0l+$MX18xD5*A$|p_E1+dq2jtEA#!Oe}>qrVIIy?wj}PoKjE7&D8Ds3)=lk{9r-Owd2n5C82_kgj-3{U*7m zp+I<;H2m~R;G0j#GNwT@pc1#w(b~6%c})2jdDhU2K$*`5R`t!0J+SREH;z*iTDPA# z6hZF;KI)z(MpuS};z&S6_}O0t&z1gDB~RHs0wydn`2{ zkA#^I-BE1*mi#l07yi&p1J1l_jaZtUf$Bd>G~$nJYT%2rSa(MH=MXhMpR&0v>N%d7g72HSv(`bk%M#k zSrcEp4-EdIHx_bPd*|Ktzzc^E$kq{q$&Z=wli4{m_5OZIkL$71vp+Om+Ox}))i#$DtVaopjP@1~D_ev1oZd-jfR zVzwN9Tt^BLG6Xdlh0S*bYrh*yk^mXhy`?52A7<6rM71ld_A=iU2~9WiXU6(>KY2I% zFue&P!lYKVDsFK*F9K*0yV6#R7L#G8sSx=nsMS%~$$^VQo^C0&RO{n~< zdA4%zftd-X2P~fb$^GRvOIVlg2A9kJZuK8>QDMT$?E#9T&Zlef*6YyA0rY;^SwADA zGZR8CmHp&y-0FbPN^-abbf!XnAQ>TbX6;=wa?lpnBdPK75plXCCQvH1)OI_U{$*dk z@;*b{B3Q{IXYR2FX#Xrk!cQVyi5NbNOq7DsdN;Iq!JQ7)!mJ>|xVYG4JWm4gJ~($k zQO5DU;McADUpK|0FBkY1LfQTE4?rg2DONyWS>4xBKYJ;wbwTil;bsYMMZ(?GT|a&Z z+VzYFu~gY9A4cZm-sR*PZULB#myFAfAw~{^o~}^WqHmVEIa#$+d*SKy<&R8_)uw)! zcJ^s4o`;w$$SdtHe6ed($lKXDj_5-L2~cgZ2!Z+H5npBU(H;Gl_35Z$M_JF$uQB!C zmkr89_n-R0AIipUgInd@_WfR98has64u#y84k}aHhZ)~1t{<}#(?|PEjR-Z1nm=qn zcBpuE$svyNM+)}_Rm;f5lm1t+aCI!2Z|6kgu`YY@59%%twJklLl1+HsFv_k-H(PX) zKSiWGuglkcx(-voO1Ez9IQEQ4E$}6LL<@1;NrrA`Z}r#s;t|s7<%8Qi<--W8>gvc> zR*^3Qv&=wIxZj`lmf>;W+-O#GYO^&md=5I@O;|*lGiFuP&*NWTFN5)O1^Q zqm}xUG8hyTg(iWBN6ER%c8}%MmYg2aWJtG++_3t{Rm$&9ryS>gCSq&+lF})tI3R0W8jkz!RlQ2DhbNXDORX z+iIA&!Nz9&)~9bq=BNn3*{sw8SqL8RPrCt5d zvWBSk;Un4J`wcv`7$9lT%sKdRjh$}(B%fIwQB;YrhAq9Cxb~vqp2c*zmyFLFzA$Q4 zUw(+cuWH>%jG)_NOklxZmmY#+$)9aiuiX(b6^l_c#cH)ofTr6Z(# z!~!U4d5v@vU_hN$xqyzMPl5(nh<>B)zjXreRYWKB=!%`O!rfwdgLE~~(R2eI=J58O z^kSz}!?~+_{Iva;Z)YaY0KM)Zb1)?QEk`8GO=#Emdk)~Yt!ION@I_ng$o57CHk-)( zp%DXHOwFGw`s&=MBpF;AKF+^sG$IcN&*mBfbF_tJVgL4~!r+>Ycm0IcF2d$3w6Yb_ zS*WqN)p@ff=z`Be_?`=9g#w-yYQE^LG7|itxe>>tRS6=^3wyA-d?btKNK_?;T1TTi zRbYOw?E4x{m+xz<8fmWv`cj>!%7XV49V-UqO&3>{Tx_8}Hv$b_8ZZ-?J~Pj)Tb?Ri znG$I#%%K};wUe${?D_ybW3e$5zL97n#raJ9g@q+%-L61*rLdO(ZLwg(j$z7%t)%II zBx|O?gXhE`-@u!tvEktUZ!6M1aus^9f2B$3i2Wr4D5?)hvniTX$jfx3?`^s?lC z#N&(gbT#tbX~PLM^Y0XG=02~rZXJU|C|COrT>2VG&-XzQ&75;pSydf*!$}qOkE~n* z@-d^Xa1&x_!Djvav{Dl`bIoqH0xjhPH`qJh;9c-gc8YpZZ2hWb*r$TxSy@hx1!UCZ z{j6*B&NqggWqqLkh*r}1I{S>111%^ zt#;56-7sV?1wtr_QB-t%OnC!)Ik#H@+q9$hv>%fnPqpAD!RS|nKQ!+7cq-@TE)t6} zG&Oa!ia^S#133=5yyhSm-q2dB&NBAo3xIdf8-%)H>`E_nErEhvo*IkhS6WRjo(B~z z2Ud-4%yJD@`9p0Ts=c24p}DLJmR|W54p6d{)CZL&@fN>_fU%YPA5-gjwQJ8m^Sg=y$tYXESTY%ybEjSR3h(wD zpy%HjOK`&z{Y+|7IG;bEf(_kgRC`a~g6%;*!}g)Lt-hWL*~%h{I!De}KMFU9!$}n@ ztNA_iW3{$^c~mNGNr2iERG8UsS9{UMCR3(B`B*9_!Bfu=W%5~ea3!^_<8C4NUi+I$ zOe3G)>T5^CzOx1IVT2>dV{?(DeF^I+&O$q0j+NjLz=E$ud&sGfK=?ZPoRKxC ztu9Kdg;khpadWayca{ktFXs4RJW=DFd3oNe|h%G*XGw zT16p86iTx9d8P`Gyj28P=|h@iQ!_?G80{Y#4xST(9LRO+n@k(eW;gCb9??6o57P_J zG%E`ZC~$_)V-`Qd_w9#a*}*!Z?K}4fCC**L_SCAP-(vkcut{{?w4U_+M0{DYb4Iu~ z3@Go6KQX}J9&~cR_zIVs`voNwIg&MpAD##~|H4l~5R9M@s=Jhlz>L@^m1TpaA*U=> z{g#%^$?$75yct*iL=_I`x^rif|xhm!%ED`z?gSDriVP?dDyPVIHZ>slLNJT4p~8rc?N1=4w(DoC+8ye z#x7*pf0230WghcH@%wc@cwfT|ctJT{=!ZiR^1bSvyCEG!rV(nwjFEk{iFs{M=NQ-< z)xf%>xHdv@{BhG*-94*;$0NGu2aejsVDAOREXVxx)?FAa{i<`QH{szcqt|H_NmbCQ z1Aes_H)ZCTORhncT69d|#bNNuqFo?-7$^$GGHu~Vh3y+?e6%#w{k`LzHN!-+lE*!k z7PwU_H-8Z(weExMDkxU*&B(f@ACJlEF`f2dwds#&4J)(FxM_|;Ql&x?VB!?6oQlY9 z`tikQdPd{zlu@gLL_`wx&;!YNl4xJETZn3+BC!BJUzff5JI(1I8WMP0)$L)Y4A3UJ zrufic9cf4#yC7DyDL=G}1GJNo4zKwYtAX4Ib^vy2vZ4VR)v1#3rOTP2wY`RM8kKP+ zTv|o@1_dP!3k73CIZQZ1!F%zZZ}d`mZ7F8IQG1t1y#{STu7fCX(@Mh>!`m%-q#H*r z)SDBpi%Rn!xEBK}n9xU}7@+^LwSe~DAa}S|8nP|;^Q)=vSuR^*_$SymHw@qUM2gnE zGRPccaFav@L8IPkn+`7grkD2o)?|5Nl=6BL!hKWtb~UIz@(hrM0A$bBXvPbu#-^(I zu);+@I}bJgpX%K+SkZyX+Y4*U#p?`#Swi zeqoT$)7FD{(#KCGM*=JQCQT|?Kth%qi-m%eV$fJKFG2V5bPa2z?sL>ZCkDU_w30p& zVk~@0{Qch>OPb$$xFfu@21qF^se(TwEkCAGb}^$Oh|2YTJ=(snfoE0#X0mwT$R*g~ z`cPZb$UTR9jEi;tA^{2SAo8OJ^LL%6$B_o6cHTfAZ0~2?`f(z@soR129MY)}iUl7A zR`zIx21@UrMcD-adfqWd%t1d@(F@ewoLV+iNf8kRYn4}5YVEdO%Yb|>c2Bp<%w@Hw zEkm{UpDH8JAvA)aMH8lHCi(0E3B!QumN`d$i^eS)(wlOq_}+@33gaqklXLi5x=mQp zWVu5m%j5o+1kw5R4|7gl(v3+%01W28SjkP$olg#BWtDE|2Se|dc87N?7B{U@51PLD zv1gH`unz76!q977C#mU>Q0mn$EB*@kaC{~`c`GL?6H`PTCXJt1jdw+~1+oWI3_j-O zld4@|8zwi7?>3jwLr`{$Gd(5j^*9C(J2k;G}JuIlb zwJbFx)^MZ;7>Xyr>n7l8nLIuZ)X(05kMm@ z5qXyQEjy$3K_m!oh+Ne0m11VTclpWax8|Mmg-Y(+)zvi>A8%I$J>UZLj=YzO6h0K% zW!2V5{<9bHFK=6>a&J&?qp&%lXcjl6!l~{GZsj{@iReN&bi>@{`l`d1uGJW0#%Ujk}K2ee3uO+{YJ_J;R z&uU2YDVUgRzdC(vlt`Q8x^g}C{JrR$(R!LgIMYTA9(1b_6hNQwJJA!Xd8hVia@Q@v zceg&Y^f#FjvKM+)jfdPR>dIdOVd2{nQVY*i5yn=2|b%+!fC9_cCUnU7cZhy?Z`q{|*tKJCCh8&LfvXiPOR2{7y0zrOEr zJ>tl`tO(tRqx_HlG%v_Ta%*R1eB) zWf7u5_tZvS)-CZdmu=y6f0C6Q4SR@lXb6k9iA$BT3tzi7M~}zLj&Rbve-Exx0l)%7 zn-jR#qvI;5gnV)s zFe^J|czHqo|FHMoVNGr8`e+ao5do22gMfk}AOa#?M4CvIUPYva4uMdEA|N#a z0!l}E?}U!@-kbE^TY{7TA^zrCYwvURTKn9)&pmhTea>^w{ey?hFrQ>H$CzV$-}}Db z``v)__m)oSY=jTe%_HG2ghR&Ky1Em=30L8azAd>%G+H$suIt3l{`!`EbE zCAzyjmbGbNd3KcV0fpTSPkOz3zJ1lGV+71u;)`{!M{y|rX*lGeJs!-%xYisE)`)J* z0qO&FddQv|Q@%XCxlB_ddy#msxN`b|#^7pd9?A5r(6kg$! zp5x+kg$HIV-MlBY$*3YprZ0;qwTtD6(wXr@y3y|=)s62a0&N;kvb*2dG}wUB-5AV!kfQbYrXR?5z>F|r>sKBQ(=hp zhoHFZ)R5!H+aE9a_yL^}e8+u+VIDnBRxuMIp?+zm-Jv+{AMoXhiu>YvjRg94UWU3^ zWNRu(YZMO`^q%?e*5@0~$35}IL4QtcZx_h{Dl>GdHner_`3Ui^3mC~F|*o|AN37ym+cu|l*QXV z#|kOz>3rxIN_dO8w(DZ=iyYpK(i<~D8f9{6J`@^p{08`ExqdCni5s^b9S){XWv)pu z3Pa^Ji#!^qx2*dN?@?@7TgTy_z@P8D4hH;!mcmTKWT{kN;qBZ#)V}<@EE}j3Fr%yl zMyQPTKPm^bc3)tF|cv0dL-|MzN3`^FV zdcDtP{&tYIA>*y0JJsNq0Xc3oAG2YpK_EZZ6`f~gOG+ChiuF}=8Hsha4Q>;f`l9CB zXDMDzUMZp2j3q?y?>DQAIyHU(&{?zkRvHKP;}Jks{xDLuDLOEltpGL=gu8r8y9zbC zsI`5pnD)^d*gjIhVeYFjML?frkyKxXc$$8rDQK^{=W^;#5X=W^PZ-$BV{g`Hn@ukFay$SwOwU}LnCC=h2dg4|7eij!^}M^CY@_r) zxKwjw4o0;eg?_4#&Xd@G9_(Z!CdaV>c^bUby2!bd+NCFCJl3sGhh!`EhXnS!&`&Y#jUA- zf^->*M&gR&qSJt3cqNe%{7#Ab%6a(aS#aSQyWif6z%4$x$^lNSYUo7IA>@G6K&;YL zP^On;*m7XVUrdBKG;0Gwut{T@B7JCWGQ3ZSEgcvN#Ejs+_p+f)jJpAiP|J69&lejT zg^(SgpkW&DbFq+B91(y<4dIHljH&4rfetGyXZapVOx#p`uP;0SQ@$e7nwJ%OE`|$_Ez@(g@K+wPF%+xOP+KB zVTkdr;=t+dPf$1lm4zz@=2(oe(@zkew{n(?i;K4CMINBN6$dPaIdNlkK=r2|?v%=S zTbAt9?1c9RSk@}@LB)o9>szWA)rD1$RO5fmMY4?p++WUpXgRo-h)muKzif`4%oLq% zd3o#zB-snNAjITQR{rc{T*<68g+;9R_d}{8tBf}LfT)v5k0d`-UUmrmuO`u#5)G>u z!^4O?^#+o>fga3dBA>q~%5S_2<}7mft#R?c)k6LU0z}T-2@n71_n!Oq+`Cb~($AT3 zN*lgqGL9ZuZT5tlo}aez+*okL&mLOf2qcWhtGF$eXZEWFly0#xW!8Cj;4lWJKLp1V z@jv90`&?XK1^g5D1p6R9lBkZr6Yk>zye}RtL_fe1?|B?COAQ{iXt^*ih=Z-n*u~mj z<@krErn9O};(c}-$pm$Z*}GAyp56~BM&;+;dXwpjIyLe`AB*rPMLb1TIK zIRzl$47F3DkuyJxL8E~|1n?Cq1D;egM|%Tn{0@4HqNCCOUDG3Q3fB*C??j>H{ADlR ztjnH5jycaoP6tZ(ck&RUaeIfmY|6v84$8NQA@b+> z%`lD{(_<9dAJeuQ-LD5Z`3}F~G@k)b$loAWUAywP$_$V%sY*XgZ0M{c57)m7ZZEVl z^58IA(Xf=VGqG$SLG1?~C-N&)2FPsQH9UaND@C1+?^0?LWZkXpnvF}30cw^sLjUgy z2>v_%_9tf%|1hw~)hk3~L0oOEawc1#_;7-d1h6Uq`rq579I=|kTg2VRFtMBLVfln3 zJmR?mP8&X4kPS78ED1=`VzL%5&UJl`A1~X_=@{Le#^85f$QgT_C6!=-^ogPMIN4$> zKP>_8Kw%ij`e#i#ahjV$z3~R0{rkYJ%^J`;0SnPZ$>2FKnp6JHC##9wjtGYq2>94I z^^6g0y_dJ%NaaN_R^E3aLfJT_m6m+FxQz46#lD4bszgi}&C!Y@jo*36KcoN5wIXpt zSml*UZ=-V3nDe)n`PHb>Mqs`MX*v#Dp7{hX)Cr=UTl8?9c6UU~rX5!eZ`??>*gRc0 z*M{X6rhW=ge|S}xK>LgOr%h2MowGH;xfE>9BaRd8gDf@0XoQPv|3xBi@Pgb3lQQDFcg?#6zAcV^t+{E6^fIID1@)~noYGlCF zn7`$Pa+S`$DtnrTfIi{Jj0b!{xwX@OB;!thBtNrP^BMn?M`ewx==>4oQIQ(&_SuHT zZahY9>kQEjzpAq$CEoXgbR2ab)w^3v$Gt99rmCreCqR7NhTAEXT+c}p>PyF z&In)bg=e^oH(nM4+an%{PA9+z^g}Ihr^Iqtj+gtq+biG(A2~uOV8-or5&L{36?{12 znLOlgp@2rV4SSE6fX^!dj8@50yr{*VP zyc4@==Xa*DX<`AMDSzt|@m6jE)nok^V+@?>nj}V_SqP{X+=efR8DnL@s7f(=K)X8* zXjWm3nBX(N9l%3l0N`);BjhH5d?X2o4dz)9yVt-!R(-%L%dRTLV{yahI%} zK2Bv8LG0?sXX#&_CZ_mQ%v#P9dU0gfy6|X$?)93TQ z@|`sJFkPJU}4WfTe< z0|aRBw|d#px;b#I{`Z;?_f7@5V*8qT%&z0O$9iZTzAN{8WF@WexnMH|gEv(bR&L!` zOVe|jSp4~9K~aH8_1l_`hk)dr{xD}P3dwDm!$u=zPh;9yPKLOlP|174s(`{ph&uTp zj#F7~L1nmJR0b^@xWsv~sh3#a)&9czJM=842^{SJD2JPv@YQB{q6MbjbbL%ZiGa_u zulNw6iulGGs4%rHrQTnO&Fg~st1XIO_colvOfY2a<;Rcj%b`}gfnfTGMo`|a&|6f- zt0N7ogZ6pfOk~%#m!_%09e~JI|InZa0eGMS=a`JMbXtU#Yol2ECpRNSz2-Ca$M)iYn-g z2^>=gbLRX93ED6H$C!ZrV->_zwy%szpD)UdBxOy-Db=!YYq^#bZu)@Bgggp%9u*(+ zxSMXID@vb zZW*06@m8=UxDQvds=QLB^6@(;dV?xzswaCk>6p8$S;x2}(|XT4su4>pDP@FUw7L#q zw9Njk6Pbx#o$YnW+5rD~w+rI~n$m=$crkuC46!?l_Ba%l!t_Eb8;}pq%Yl4k0XwZl zq4xrb`7ZmF-IA*r%clSr=8I~~a6QzfGH_+hJUA~G2;<~=uXQ|r{4xL0pv-l}A=W3k zlD{qZ_Nl(*|Pgd6n@{N=)Y?f_}jZ-W<5tg zBsj{IYvqcRqn$&CEpN0$VCVrI<$_kATlHbyc08dtIbiipi$%@n6ls1%W&{-|E<`=+ zCXd-T9Z|!8mqK1tBvfkmd}q&~#a#zoK$30AI((>{hNd8oSeT8zHw?6tu>9; ztjLU#RetXQ=F#_3M^`4Dla{dUZpGe6OKy!G*w%M#Wb#rXz1)@=?;sk*z~vK$7;bo5 zU0PA66stAZ-w@eO>r)W~UEn2}v*<@1ClPI#zL$jywrNi|lwmf#=Z9%Zy-!b>sX=teVh2FiK-`I-`q7e7QR6tR%k+cNwE+ z^tyUbn_|X84 zvJq7+2AHZy{s8hifJ^ds3OkpSOO9>LC#;WwA7^l2`2o?vU@sg6$N(zUb`JK};S-(0 zd9ETY=-4J-c*ha2$s#R)#C4z^v%VPOAXCwh>M|SLn%(%t1{WP5X|XaOamOBu_Y<@m z?J=;x9K~RMqFbNfAx+)J z{vg;nwQ2MK6O`?~Xm@m_pQN?#L)tL`Hf=-l-bp=p(f+-Q#Z9@4g5I>WS!DbfIQFox zo}5iaeoT$>)xA*A-3uRoDjP9DAaMl*4xcj6W!+c~l#0^Gb3@oW%~2jXR)S7{l}0zL zgLvNPVgMIRot91ht5k&=<-P1fK+7^@1htWQ-haf@mv$MjRpHEk{#(5IALXn6!{1xL zabrSMQTu+Hhg}VQ&hAjAY=B8mx>Xi)wg4?WDDhG=Nwq%X-}GVegqGKu&yZlY8B@G#8oc5P?1>wJ|2?4IK{aq(>=Ca?%10&b3S zIUrVYG7GpgzI$T!TF6%NZ6P^UbP0YLhG4Ikc$|IQZJz{MyH~z?uz9(sf$rt802WOho3gBBPsEW$O4VHBH8EBqT0Hy?>5*7 zt=O{W3V1&x-5PTd7m^t)hxiI68bQm0Hz#g`KmwnSR(zm+n2AN#kVV9GLl~k(ea>pC zKQ7sl=IQ24pI@hU>(OzW_88O(+DPfeIbJw=t5(ZmN31#T!t&Dmhr|7wUi>9Uv>-D0 z$k2S8P7TVSB&)*5M7vN5cLtcGi6Z^VB+)#bLRR5$4bs3-@(2T!cDZMmmj(MBoW9;( zZH4j88$%dHSvdlc@84*_#@xU4m%;9rbsc*rxsI%!ewnUiv|uWGa`Sf}W*Y9^yb{0} zv1&+AovTbklPgY*+@=7v0hg$oX@UjvIcO88UitHyg#0i5EUp;ziCi7O2x#4OI8Xrur&dRL z!?>IM5V%kJ*5;YPvP#{sA^S*H`?D8*AnQa>2ZluLhCNcjK1_jmWCu4vhvD`SHMR$BSKCXPw|4t(smUle-5 zHO7^Q779AFrQUuqeDuNS+1To%p*0q?ML3v)cEw|mFu|&U6xA5MjuxYTQyGJ9^S0*> zTMg#f(0G>YO#5x?>d^^R5n5r-o&%NacX_lytH9;vfsCl0d=woabM?TYfqjq|L%LO3 zd(Cm=e!B5SteRl_RCSoF>C>0?$_nD*-j53g$gOE8w!zo1O4)B|qu&|u(Ox^a_hUCn z38aUKU8n~)&ru>;ljw^BUyPQ!o5l#+1~hM1-POjO7-hvBDLq741MUe2p~RnWz7iMW zlpL>04ij>m_Tm|1eJ$1##mL9&`g&o(2k0=fcy3XOnu~>Swtm!;VIYG(Av;JRq36tA ze{Z1QLu8@GN_9W*`a`~tp7fx|Vlhdeajl^nuHh#r1B?m=a^68j4FLQ~##OLlWQX`G z&(+jB?#r|jNnlk8rDR^goA3StRCB)s8VthEM*nb#*Z9XX#$r<~t+MrJ#cBdlGDUM( z8F;QPn9`l3koE_E1At`pdE7g_EF1yQrxc)8e|L!Q^QWA#EI3OdfZ=B5mR2?89(82haJvcros9KA9M69|;{W)JvDmx~D>LWJ zbdP`(nYQlM^FpF;5=n=6lS4Xxn1X*e1lj*n&M@mr{fbFqz{|E~emPC2=$*3HX#+j{ z*GE}rPXxV2|mUWkkW=fK29NR|5Kib|HK?U zn!O}0`K+cUM0He8e)I8|N5sACoW134MzFL#z4?TRit6o8(AA8_3xWz2 zEFk092FS?2Jp_XDKjn<}I{{%!K`uXl5mKKbk_JYoe+PP4pn{R%2C6D<=fW3M070U! z&_+Jl@83fF{_syaqmW;IaQI;VWw?^G>ZQ4b_ISIi?52(N|Ei~FiJL@VmR>&_N+Ol0 z3nFunU-kXhP1Zlmo;;ojX$3DC_?xMK6xMf-C*`X|gV++{K!Eo0UosSb{Fc%~aT4C; zdMn`GybyAFTFY~_leXRa^j~*>mIZf71kM%RSw{u0@H#~#o)Cmm3@qy<|9)@%C#__E zzg`-c?FD><`6-qeN~`kC1Sv*2xRK;V`2F`?k4502z^F@$7%*Q7)@n?6uF?`x<@|G2 zjX$hSno2KQBJ`}Z-N?5@B4DX*suKk7o(zbo5^BmgP1y%5)(q$Gfz?L+dM=4xu41{Q z!TdPS_+K(=W__t&VR2L6tmy-zZ`aBzREd_76=i^2U96EYNy?=UC z5;2aIWC{p@La9@;%i6!-?GZ6~qO}U@cnKWAFbmmaGzuOy6mr?JSiB| zz;{V;l0UDnqS|O+LA6IO*%BRkKc5jy#eXGT-Y)kvtAwgV2IN2wZw|Q%6&3B%HW-Ii zl)M^`IrDGFEU~4w3$q39Z~TE{=@~pcE#LP1l#u#xrt99W@ToGN-lu=RQs9cO7*%KW zWw%&;Joyg`j zo7f(AQgqJ0TT}jm7(LYFB~IDuev&If5?W&yWNRV(^8TxUDJS9A@B+)B-*L8dXS<(j z!b6;`@YL9IFYP^MlFIvdEBxwEO|GxwVb1R}?(5|vYsM5^*^x=odLNz)^wqk0cxYw( z9>%?56qVIy=9Z&)R|zR}MPtgjU3C0!Pto zoL>>9NN=(5v`%=;i2LT)YeuaYPzpGHdW3-dS&!NqqC;P6HY3Vxd9R0AP*0m*~ zhvEXF{O_Ohi+nP*V6rFMazg84^v&m=4Lk4)^eV%?Rs=k1Uy*r4660*F*W094P_wE0HYjv$ z{w9{`mNDi9R%HGSnlz@t(S??VFZ(n)&^h!%j`{OW9dDuBlJRw`sa~sM!)Ero7vmc( zD3-x1^}mAGTG}7P6w)#JFKp#>>J%XkuJ4?5r-z=bBxiuEC%#~ems;rq!vs}`w-reW zMztfQ%^tpCZ_>vjRUyLy6z6<^1P9EbEiS}roQ6W$>&hy!=0khGxOT}IH2*zDCTkXE ztTue4h>$qm!a+8QmJN%DOHSpnpw10T{9(=_K9qfE1vibZF#%@sowZe4c#6;Jmh1ZY z{6Ok@QPy8p8!Ep-*hGsFLzSYpscuDiV*4}YNgI!k51*U};E{!UX^GJ+ei6(p6H;I- zNoxmNP4&H#W2)y;l$t2tzqx_3jx%f?fIR4A%T3y9DPfN8hpo%Tb|g`iR1OsuSd_*c z5M_RnzgBmvbXtBP8C9rnFbL)KuW6m*MCFq>ro3>oj(dI>>K4fd7!Zu45EiNF){xd` z!D8k7G=}%6J#_=5m^4nbm_8g9-1}`Vr<-5P`0ShVCTgk>eXj_nG7Ly%$Rmz+t6hz)%} zvO_F&Ss`Xc`lB4?Tw62;rbl)YhJ1!*Fl=$o($nkCvoC+x{f72B+0(0%-Bo$pOmYir zC3o^N>Lsn+^2(YnLNPrNGO( zS{uKqc5|)uG6E2$xA!vKNHX%_9=Zwl5CioV8$LXPV#=jq6j~K^u)y9A^AFFX#g^xM z`NSB|97<@TyzEUf@b@FdyDNyDz4N+JEjFhl{*j2OT5I%o3DQ}i0Hv-PMP8aVPr((M zK_NQxPIb>55ZLBI#%1mjdM9qB%#t1peuWlC^f=nh5c%Q+j+jsqX~CfVzsjMx5NP|b z%%KV{TVm7-6O@6CqZ45Z4$ax8JJhs;_t_-M6!G z=GfGEc=K2-TJuvSQja+!i`9Lb^R}*|@8^E0c5mU7t71ZEd~G;YmOs`wH8VPAfG%cO z-fW(70SfvtKv=ywRZm*ivvl(c<_qfNRqMRXrxoZFm9t@G>QWek;`ojLq;^#4DK>%t zbW>S0!G)>R65r^rAZ^{bE_$s3T4=b@aBVP3yQ&0e&5&fmNG^1O$8Kle5j+F`AVH4M z2D&fZ_)a`KE!HUHI~U)u=Ut}TNQxEA#Ta6xY<*ZD`iU}ow&rp@odu)CyVFO_59A8=k;{dJMkzPnWPVU7{y->kPK^iTn| zU8OfI`{pBW4t~m69LPSb&d@o<+waBj{qYl{(U}el`~=qJebMB-;FUd~i|YWs^^{A) z18;uSp`vwtx45orf@MhMlsDeaBiRMCS}jW)AL_)H!PmFRIfoOMYJzq8FG7XY>s}qL zX9?4_2`z^+d~=Y&I&EofBpbwm4s*G_poPL*imrwbdk%;!9UsU96Dyj{bFKL5+ou$M zV)S*VI|7IrR5b|pM%pc)I9X88??ccEv(SVN2Ek_?#V<@CuouoKkf|Ocu zFX$h)!OjLaJUK}rGyhr*4w$=JjfVTZ8s$yX**A^Y-Fq5(X~RLD8B4#N{p*5~TN9IN zo4y&V7q+}pSs9cOQBQe2E<|U%e4lJHa!aIO#+#29KMd$pUu-Jzd@e?@>jo)f9RlzKt_=_(Nqd)=4yFV;>DeG|Pn|Wtx=3{0aS# zEZTTW$pMNB#@B^DU8m$oPd|J3(>*>ZbTzzPC$sz$M1~oV9m7)LU~DaK`Jx^TmIBcDQSJ;XV<9r7FUdwI6GL-0}n3S53Q~ zi;q>2iFRTifR^n*jORFwI*uGnREy+|=_VtVF$V3*no39>@#GbD5L63`Q1|d0B+U7Z zqPulVirtrxJr1PzBTRb9Jj>C|BZ&f)I9_k{YE3l0BH@{TpeBMPmTmf=Fo2gCBd0nV zFNZ{vgv<&563!rBaQza_q%KI3KqN>}n*FhN%d2a{tyU_owsKukP+ezI*gi@jffHhK z2ilF8L$pf6<^A}=b>5s0J>JSh=C$SH=wZ{cs1@Wf^Mkk%O`emYtQOjH;fMxvb>|<< zlPrjJBR|wh>j9>*$yKDjRgtM@FC1zVKe`aHH>Or$K*%s4L{C9Fyg$%6{j3Yq7gj<|!9>`M(B1=jptNYsP%F9V#0lcjX%c7)m z$qyp?fnth~h2-6Ok!_XAMx84q4*4zJ;DgSj_*x4wW=v74$KY?79#vo=rl%!|F7okF z;_LB#-78u`HYLopvdZEccwQ#?GZ@cik<20syQl#xyl6J3rwJa*#oHJy3tusK7HnVc z$3iO&%9p2r7CV%zwv|;<09h+VOTOQy?p|Za)1TI5Q+WNVds1B_7)@3pZGxJ9hIyM| z&1dCtoOq)$-fgD-dz)CL+SmtU^yWvrLvgBg$!m4a*_cT|{jlfyoNt27bPh4m?~8ea z@2jY)9Gz*W{Ur_~Sa@?-BsnUMS5^0;$QvFOgjvFEE)6%)c3+mcZ;GqDh-JZ+HqZOe?Bz@LL%v{g z1cVw%O~-f46w=D7k&OX5{nDI{os!Z9yy0sAOJmU*w-2Icn=w@Wag1tR6rtdA!mWgnu`OniamrlTI@1QaB4FJmHd zFe>`U@HW!~88hdqU1#4XP978q;j{JI!;sD+1hJLJew#R4dhGQ@dtOG1JEdPhuuhj! zwNrDlJn%b%v;8PE=ZAPFBE?o?l-ekbz$cX zUK$95m)G6(c-cxX@Xph4d$orl-1UKYJd?*8Z%h1CRDwrffPk=l9b}0!Z1=FiCB*Sz z`lA)ZktG0|=Z%aWjU!zr#vO-E48Y*`od+@`-m) z5g(fIWS+22D%YA`sW@34eTnb=0=d0g|3i3WPdX8!w)&)$db1;C%~WfRs=%IfIKB+Sc)+A>E5ncrKBBh^@pCj_&cexD2JRX+LUZn~8pvo$Gm* zmIJx@gn3s6mV{iKF7iXVO0b}gvpgRK-ENl3wmMPYr!@`h`L=#E;Kit2k#|M4C6$g6 zRQ@Xxqbsr=bf`k|Bnl#uSRX0e>|W604O+on^5uk$w8z2U9Y#1iT|aSViivUj!M^_; zZ*$8MC5L4{u(X9PRo^9= zx1g_#rR0;o>}Lnwc|tNni1sMAB@&2gBXUUn<>l~rTb^&VPiFR;y1G)xEWXyoH}~XX z@R#mOkW`OajJha2CyZC|447rVUKc%(ML3&zazqI{A1ti-#V3U^_{Z*@Iq+53 z-g@(dL;D$Z`P#_~UCM_tkA-2rV$harAPWTc^5T1ph1Zm8w1w^MFys`s9ZM+JyENfB zVd4F#zUN-IZPDPx{lNWw^T-K3MfnWm=<6lP^52C{Dp zaJ?Vfk`&fHw+CH}@iBMQBX7Uw%Gt_0;nsfDor42xB9oOz<0g zX4UEaQEBwzx*4KC4^WEStKYpyinl(|o>0}B&YFvHzjW_HVafN_IMh>#P`-MMcQ$f- zQmC`r{{vXc$UTp96ekF?URs+{oMk#(k7*ksZjG`K9WZI1Q^_to9#J3iQp@a*(XRA} zP7i+lN$Or_6&nE~^W#FfZ#4WcPpw82b1ECZ&QDOfRiyBmzLpr7Qx*d7)CXB$^(w-r z^-y+Ymk1gW+QE4DLiyRC>Ym|}FO9zT=Y{oNQNE(`(nuE_cSiDXy=h`1U#opyOV|8W z`U2;G6lTvoJce>|rD%zFT|7DILQ}iU{1IY0rz33HW`&3rk(k7MjO&2=(WQxt z`d~_DFnKTVaUpt`q1H8Q>`s8pz3wMwuV5@^!KeF89=MVs499bn(EDc}4QyP;1RtqW zkI{W!RB>ekIFlKK6KWzbAIqJhw{d(&_TA<7P1-U(u}Gh+;w4{jDh#fO8jrt7PS@iu zDEd5Xqps515jAkSRQ?MjJ}`ci2PU)Pw~h-9V@iKKx8+|F30m-s^Z5@G<*_fOg)oZ_ z5yRDtfK*8Kv;tl5l7QXQ;*NXnU;|Y`@&aPsn^>EX@-Tawvlk-vQ#;)Ap9^nu#IU8k zvOrjewAXM4`Zut``1OXkO-kbSPIlbx+PxEuW?<5A984Y=uJmDk;ZiSq(oCP5wx#!Q zse5(QKCUZVo3M!SlDNMGR;!`11nu_rPQw?L(g2)7Msz=wtG^f!-cJZEk>PLPSjP@$ z^;XRTeu7j!iMde^tk&XTughry#s_^)B2>6WG+WyZNX!Y1p33{fQ-ul{(Kn5bP3AvA zrqihHNb%>u(*)d+&asNmDV3wU`R2ifdgFbC)ZZT&IJicN6H(CDCPg}#R3B)@0u+0VzzfLz(VovbT)&!D3Rpl#6AgVR^O|jQySkZ}Osp^kPRxXu7 z;klZNmwlb^)>FSp9E~_}4pxcNQQXW=X`P@^|bxS9EVBxY2CRc@{DXupH@8J6jZTs_pI?om(-o`4_owo}` zm*4Ur1xtjq;Z3jogYK)GRAVKoQi!m!V%k>dD7_NYY9Y^2Jf|kFv8e4=+Ur0i6oI$* z_6U}4p--~OLvE6rUvMNw&~V5;us4`UU$0Qq}JykD&I(cxji zb6_)%_L(n8P0@!XSR z=gp})H3Ye3){c-m6~l^D_hWI_;`Qw(jPpS*p1jwvhA5x3uhTMbpm*8JYlCMmv4_0v zQfqg=xa`&-R2~N?-lkRyN@L>m?`l#)Ijokd;6WbINq`Ma6OBjc_T}i3>LdklN;6_S z6V*5Wp@9XHL`ETYqNfsYf3l&)ItwzFvN-#@>zZ55uP=^nXUWVHrEpP)FSx@!FET@Sgy)X z5N4jL(VbPpS6ktw20e5w9oTF6o_OCq5Y{tj*?5cP{VS(a+=mZmo;9y4*ZMaRJ#-Rg z5+@rkf^oFwEftYSd2Oy5-Fjg36vp?W?qS5yI)Kz8P)GS5+8qPrQlqLh`1$zT5fNko11LDMIRzJU9U^fYrg|mN|k7B|5m* zGN5Mm_E}Ty>8FKH&s?ZxFvLw~;a?n~OOiK@#0Nf9P+fY(cmR+}9Jh?M;#+%G8EyC(!uzFc<(+UT#OWefVpAhQhS zc$^8eIR`q7@cKXw#uQ>ER3|NYn=bm*TIAV0$?$u3wf)^pFk+!7{tfGi;07>E2}frn zm$j+oNbGs1s-h5*|MEmKZD_|T8@W}e?sUYkg(i1bPxs7+yquY{ z8nisSuiCv+y!CXMm9uMMr+?`iybPnwl?KaGAFE?afq|R26Q6wVxJb<=tA6v|dEn#} zD^V{VD>ITTKV&O#QkR6@T_TSU3FzsRbF#dI@Wb-iFOreyTiuW!p*=Y0ZMmXLyBhza zU<%+7bFJf|D-jun*4m*xwzhOJPjhwrnP7}@eufi=jU-m;)_V0*`g$Er<|`#TAV#1U z0$B^QB(R{G)sE~?Z*(!b$1ud3KT$;|95x+t2;lX>x<>TT6TXG?M1o{w2K=301$Y9~ zt7Tt_;`t3ITLEsM+_#LcZJy;@QlE9Gy`S#BLg(ZTI&+cqH$aJ%!d|w<;s|P;kD+~Q z5p3W_d~lp(ZUT-&lms1Wq7XkQ&w6@@9tZ@(lOy=~T9fW-c;B z_&P=;LHY;Y);TY6E{%wpE8mQL!5Cofx=7o`AsgL9;~3n&F|(D4aL0O#&vIPFJv$7~HvMovc1-9x}uz}#^A z%sU*2Xa@Kcr&rvNPqWC%=oo4&l0MC9w@y*}v&!XeQ$JG1<{Btxw|XojEYFpQ*{1P^ zh;cLRg*a1eJyPZsnEq_=;34m00tabQPW&j0Nv4(2J@^fX-bDh<*|PObDG|eV_Hz-k zn&5(Ov;(Uzv67{TmSie12R}*8zl%^!jJb+Tm<*7R^>9p)rb`}N?(vs=X#*5 zP!mLXSD}NHVV*7LxQH*RU5oh!CUrSc_D}%%B4$u`ii20wc7e?9ivS)#+|;8h!p71Y zr~w;G+3D*%tVwX5r=GR|(6{6hoxEs#9TJ5}vZuk6Pv6NN^o|?v{=OGJet!i!3lzX7 zTY2GSz&;DEe-L{K=beV$$Z#36pVebIPXo`bCr{j|_4qYW*|C98ILjCO5`I2bgMJbI&=7KqsiZD)c6TJHuQTPn-%qaL8TBpY+F5^*RmvufNXB`o-t?VQI@J>V z4X(aAEAiwHKI|B2+0pcyFxNernlS$J+MHt7D@U{)%($Zst+-78rFh~G`9(VwT;?w+ zdkA`1nrK-T&=Cf1zMX>=(~SgM)}<<$=F9R`(NY!_&O-`&^`w#+AX6{(L4kRazQuu? zmNAcxX~xeIR4E{|p$ZHZ_ns~=e?FryBx_Zn(kJTT4Fm9h@?Cm z>#|&=*N+(yaI~7Er>N<$gJ#ofoBC_g33CkRHRWv^Q;I1vSjK9LSiCQvY!TmilO8*K z_dE|ba4O-8U!W3bL~DGNH-JG|Zls-deLohUjpGAF>wkhsqlc&UWh@{hnenlGJ17KCd57*_RRi-cH-sl-0Sdk2R!)efGASP-%yUJ?l-{V%QVfUlO$ouCJJ|^jQW)Nh-_DTCE}4 z{@})gIkHnLx}#2@RYiQu$&_ne>a?|T}m+5Q#;(nsVi@n24j~mAI z9^;nnF-Oh$$@JTN8IAqDS&)!!dD+Z=hW@hIU^AE zU)vPX9B5&ZR529oDy=F}<)sUUX*Sqqg=_pkcz~9$>b7FiXn4!io#JOTY`TK8 z<@cSbda=uH=HHkk+w-u`1`h!TJ6LkF1k2F#5b{+tf3NXw4c@z7CH7MV-U>se6<8gU!C|WWm zWk)KmTDykZ+AoT$zga+Uc;E!rlLXO9jqfD{3Uk~E$|AnHYkh1_|6*qGSZ#A2;8-PT zbuhWdTvChT<~=IaK|Nheqxvyso9`Ut8IPG0g8NVd?~ZipfKY&lHw3Cv!EMI+U^=75 z=G`Qjp&uyHTRykp3aYS#r-&mItIMBgz4q{n=*dyP_XVJqB}%8?Z=Pkbj*xS;osAG%CBqeTN)I@`_1iU!$!{hjGZwf9Qq;Y6 zd9#`Eb)94r?qPl4;$F}q_$o^^nrW(YxC=`y|Hh^^*#NLzJ-UPkaGJ4goE3S~hWt`P znZ*%k2<+tXLG0M*N1c;z#{I}T@T}JbUclqPIn+H8sZN+>qV2_-W?;$8 zTs~4*!4)REbo*kpcQ?kIg9CLqVEr8cBJUD`Zpwq8cYL|#w;p+D#?Q{rwAtFdzY?i6 zmBxV&81(Q@m{y?4I*3UqjMR9(*k@g2wvQ{nPI~Rm?qUOiNknhNwa1z*iC^1Ssv+hs zn2~*S&-svTh8ccsRf5U5bJIBma+d*eMs&`l+%42&=PpNBcokn_$rpg$={G8wsWCwd zeUN7r0qv05`I4E8nycd)Hc+*Wr*M_$hfF^6>J`KL`(pg2ug+$g947EzMKOA{Iys2E zK@>OuCWA!<`Cf(#3PqXn_fs1qfsz$}!&DW4ntueAfXK@w8DYmu-Sch7)FKjLRu5+s z?d>1LC?bvgSU%=q*3UdX)6z2iN?7I_{$CKpDa15W$BCI+Q?6g!_`=uKEc9P#ECz>0 zB>A4Z6$zmz2sD z9!Cn|M3Frj0`{)X6_-Ovy@~}x*XhA(yD+oFJJfrsQBjL*?bR2Da+BA-m;C zi*zoNTOq>Nm(S;m#LhsMQ4;|b7t(FgZF9WTgGMcQ$kO5p+b+AT{r|(>dq*|(wcVlt zlp;-#-l77cA|TRRK#?vT=|ushNfVGzBOoBX2`F9ZLg>9kdhfkM=slqZ2=VS;`OaI; zckg-c8TY??27{3eo3Phj+3Q(Pna`X8A|D`&xS~I!zW)b^Ua>ePKYu=dNK)eZ%2X?S z;VNx?kAjh%p(kLf%2@?%(e630(%SFMl~b?~HdLs;A=o0P+1>WU*nos1vb1r9cB%)=x}GuyPb8FHq!M};=<@g)nV zJMpv2jVN9MHx9AtXjv0< zWHM<^=xM7_=>A!@9U=sgk70&q%4qOs)9d5j(#2e1k^%(t z`lMY4#T6BdS4fugP^o6IPIj@aA%d4Uc4&30=IjDAvzgQ)Lrmw$o{L(jERFEw8cyGA zOHWeBjq(%LVdnXh0;pG;%%Y9fgL7)FC+H~i$G={hI3gE6RsIGM%}-4Wo+ZU&$sOsw z@}%@wzWzwbu#{H2XA0z0Yl2h9rbqv)ptMAwf+J=zZFtI zgIn&MkybdSNJ!OYw7~po3OcpK?QSA4H(g zVlWV>7&LvH2)Mn(@rsWFpj{1?JRz*BWmozE&6b1C(Y|>d{p;kKUi?lv5%yNjQ?EiL zilQn_0?9hJFYTFeEK9eU5t>o20W0yblC+VjWl(lf#1tyhW-kQ)66@wyO-qpGS0k+* zGc=j)Esm%y6~WyY&*Fn+vX)2ytitWU_N9x%umCDdDpz)@m@i3`K@IFVbJ>CMb9(e5 zd=BQ3BccF(7_7qcm0lUC_4QE`VhegQe*xa%nmjQADd$*d#tkNLuxmQ`bYoG>8`i=n ziX$~@g>9>`>e5SpgLe;EVaPyW`^7yp(>KeU!?@iI=(k?!E`J0;-+oWBO}V@&^T@Q> z1}Pm`W@CCM!m6TVbi!VI=*$l^Z`7M*v#=)0F;^q&!pLlnghXef zjOQ^2EHN+LEplviXS8`|E+E}hVZ{;s1u+_N(T6>)w*$tQK&VS~JZoQ0j028;(oLoI`C;1I^vMq7FkoEdY9n|?PlR*dX5bD+_O2(kU)8hVdp?z!5)hes zA735cSUO1P@_@iZ(QqEGVR+3V{mFc!GQnK_6*o)~#TNJjldgp(uPhZ0yiBxmiJ|jF zslP~5m0LC=??_`o_06>Z>@KiL2+e2TaDCB5K-n9*{TI}=h#7gcwwKr{rEA3HYEH%J zdfChTm`q{G88hjkVLENJEi5o?(pEM*5G2O#5T5la6m&%XY@>C7p&GRp{HcNSO@UqF z&5&Ju`A|*aBknGFk3lUMl>+vjND_P&!;J(7%g0_+yyf;leo-RJItCpD5izVDMaK6v zTjXJx(S4plKN7`*0}Tl){P*a+4CM!FxIJItUZJ;)*!a=+V$i(y^iNS)QF$iW$L?D5 z9K889Oef}jgd{*yYWgqwBV(Jo;DE}}bz4Ju`)nZT%tyETIM!eD6y-xMH3)e3xG*5WRMC+Y;x-1Nx&qf_8N#u+Z!# zKBk`FUaZ(E&vO%_fPtw;x-8Fo}UBGoBu)2AVKs?GQY zxa-8?E{miuI=4naEPA2Gc7vainol#b#Wujkr`@ib1Rw!l=FBZjb&MG3|7y9vGD!a5 z+5>50Um?m?6=ojgosJ!Q>i*?|ox_BksQK>ryQ+xYLvalv6QmiTwW*lN!xb%AXY#Bx zZzJZlc*V=yadn_d5Gy*u5PLsHOAzm+&a3;ZAX?h^_@-2g>+xC7xS=-{SnP&%$ndoP zz-R@3-=G1pEG=11#(F;<6)dKe{wxId9U#4Ot|Ymg%DY1ClS(u%r@KUEb(jZX#Tup^ zY1z}a^OSDX#5@url!+M(Vt?Jq$cl#*KRLnJFI68&ezgeedw)>+Y~Nxm#SLhajJpP^`ZM94`iuG3o>@RWH$41N zz;q%4?qd3A!WHp(ffF?mv-~X(Tcy|S<>En|nlQ|#p&ywWDHZ5BWAv1^t(vNpsh?M} zn^wlE^Fy0-fr=za)1});Rt|Pp4Ff~W$WG9ZOZd^wE z;apb~V*gPad0@R8PCgY@3>|I5-5uL)psRZmk&w(ay@6=b>abDZqya%0kFj>uXue=U zAabzQW0KEi!`x$eeA)F!3zwXSbkKp&LBOO)oo;=$+4c-i+d5Q;n*lpPCPM@rWc;=5 z`z?deY=`akq*a{S_`ADh=)qvh+1)hC6ZcZerUD#lZ*expH2K{OR~zj5Y_E$$wYbkV zcf%vahuiOUR-$Fe^mvvx53q00Yn12;{IH-62aJLUIrx? zqn&A;#xvq(Z)H|mId)QekgZbg!4UJEU_XQWOkPSl`_-I@DV7Tpn4jZcHV-wwkE|Vg zM6iGPq)TzGcJnFY1y#D?e&t}Gid(Td9OKwV-^AKf`x_KhToy6h=cBypf?1iVof1;V zgNn4K-6<7EPgE8(qD&b>V!iT9bZ#?j#YH}m+9WU?I9+UMlB`BlM2iG@8&wWZ?#q3U ze|kzGZI=;ab!re z{K?QZ{ixCNCWTZ%@!#(8C9Rcz@K^BN%Kcb2b-9P-@eavvc~-x-&) zB{$alKr<)`TA8vS5R~9)QkEQVl~iOlk-U=3%;Jj^u;B!%I8d;dXtQ+7^M%A+Q&ai1 zHHY2K+WZQ8idiLrxla+5q$Lqv??cK5WAq&MYJXi&8jk-xJJsz~!_a^giogszP06~) zWnOvk0q@$}Ygbq|5Cy9~GX0Y)+gg%y%-rIOg(jcp2acbi9ZttGl4ft4XH@lP! zi`A3{k$S$3A$>a^p#kHJLP-t(T*q(~qvE!}Y+ZUsoe?Ri32zHCGXpiL{!I?&+rKm( zoSqID8f7;EtK$Y4P!4*jxl5#+@F(Vhs$1tR0SDR2M1#Q7kp>=Q)D1g`D0{ZrR&gUH3htE+1-BL7u=S;GOxGP*wnLCEn6`W6P2~;7-4#}`xP%W@7u$k1 zj}O8O-$vP4*n2GV*eRgSUj|K_A_>;^I_M!&$M=dcx=Q#!(pCL7$Ae5`)yjj9*u5861tYu)XTJ5r%)Ifp@ooo5#kmk)y45R{US>#n}W z9=2ZHol~aRRN}H)&2sW-yiZ_GiMOc%o!Dqkf6JFY5Smfj(;?IRhV^T>3r=YCPskw9 zDvDqblGx*3F3N_a9k7IJam7Fm*#X}um&I3z7SXCsu-!ybRS~RQe#lv+8+jp!|1N&u zH%K5~>RhyF;i6+!t2C|WkS1xymA{)gJV<*??~5(?MkWgP%_E9!`@$+PS}<W;>V z#`&hFg$PhzfCd;NPO?ojg!p)S`UF|p1WJp04<=2g0<6E)P(F~A&-z= zUm_(di`Loux;BbLAr*MSjkLv|A)@-5Nf#upv-Xm6(}q0)G)Gpk3i7{*0=6`ulX4}f z>Na&kJ44PwVF8nctH?E;A^EaQi=UC=w=O*Oa;H^uZA)giC*G0<#`4Y0WnB9rxg=uj zJUY{`fvh~Kj0$TY5jyU^{zc?a^-+;iLB)ghk3u+hn`&M2&JM#DZz@@7!+M!c(tuiB zq3khBR7ZsyoInPXybKrH59@t(z&gis%bqK$PW}6}0OP6=Dm?`VE1^Z-DHxvl}C_bQ7Ypgi(2l zD^CMo0=|>bUp=lwN!}tOZ$i)7tLxU|xlTXTefeHw6%28eZpl-B$rcb9--XfL=~w`l zPYRilV&oqPO4M)heh*q#=T*P&{CTB|JvVnquVFG}8_aMVMlOG*EG}PhkwPu#G)pR* z956(C3R%epxm=iq?%kbyBPydy&a8uudTgrvX5T0(HgsU{gO7^|Eo=< zNE#Wxb`sob;;^D-r4ksxJo+>L6+;w;L}*G!j}A3It77lW)J>MB_55{ok36v;-Wv7b zayl6l)_N1_(o&VwQSC+DpZlZ3ulok6@*!1>zbBv=s-25Y`0nB4&5mYnRmqku_#U*d z&G5$Nhj-0YMkPjFC}JuQZIo!RiQ;1W9Pkhg` z+@t)zBBSC5ldsKKyp0lDJS&cP!*ZwK-Q|(84D~Uky;J~Qg1YUA?K0MWiLvrZe$;g6 zggpxr)MX#K%Bw4f+lf+BFV=3(px|o9vD9K-dgbmkUH4e=UthD|7YvHYiZp-94vKYY zW^3f%(mUb1X|zq3q~6X-DAJ#P!CPtz0T>?D`xrdGi?Pj(UrD!mj$+c)$9!*NidEB$ zSV#4zjAE$iSQsgu#f4uu@4Udc+gs&9%R9Afw&Q?$=`xjbpH_FpdA%3aprfV?tJP;L zR|fX!sB0%;UaC-wXN>38E5$5VuFEV7q2cELDK3`HwV6m#OLfehYAKnS&4|1|2^5C@ zEaPg*5W58U#hEIhT3?Jb=-_ToR9=emopg+G_`BAJE1kN6-Dr(I@~8J?pv>Cy2SfN= z&1fLEV=t`V3ElN$H-|^@d|n9;2RkR7Yy!c+;;Jl*uSKFyDB-%4Qm6KT44WAW*CS2g zt%HUIzr%}CA)%eqAUJI6Ba}Jbxwz4o2Wuxmv+-#9s#3P8-L40>l22Vm?J|EqD>7AQ z)Q8Fto_uNv#CnJbVhEiBix1Y_L_d=swDuRfSdfi!2bs~0nqXP@5h96 zmCJS}F2=(`Dk0gFS$(5&H#<5X)ojeMP#K$rVejN)T#R7MP)>fR*qH11Wn#!!G!Yth z(JAF49|~HAOlD$VVcgv9ZJb_RKKm4}|7I`y+D!)cX7Fb?f(R<1g3|q*YVXx&RF59|>Xn$nTjpo>hZ;Ji-HK zR&$pxNbfD?EII2ULq+fEe+NX@2x)J=x_=DP-u%n8`MY8B271b%{rx2Wj7%csS+@Q! zixVl^_W${=|C|Nmd4#`f%}FXuvxe~C(ngc_P72}6m&W#N9d*k02|8&j!TvKfk+MPu ze#M;UMhQMr-CMYWql=UMhNsJ5N+^!?-r?od!Shz)$@LUjiO$dxMDXq0!v) z*w%+%pz}|4wa0FHhOWC72Q^U)3-RaMT##4j*}6aaRK5%O$q=2pdEHItUp`uVaqjO5 z8a=JosG{X$?XB!A19|Mf`;m0(XWk&eKJ@&TsU}E96MgjNK^AoFpTO^O7kvKpGR`aW zmFu;9MzG;rX=&x?yC3DH)u9VBw6iRl_W`Hsy8hqX4VI-Luc5t9g0-^pn;DODhhGX^KUk@pWRprY>C^VJ5gfH5mMa8W~45Lxkc4(i2~dQ_j_?`RN0C zY1Nxvx3=9(*INb-jXWIeUni-!AKkIJ$sbV~o-aiXdR6`Lmmu0SFvT;9L zbb;us)Gyt3>raIE-@^%7pIA7Z-`n3qrV%k%ugUZ*YCr*%W<0G{mC@t87vgS#@RaAP z*$v!GaXW$d@IS!Qf4fWn{z{vE6_5$D0lqA!d=O%nded=$f@%{weWKzIzv8cN3X8U; zTadex4N^XwJ@Bgd_oLko-RXkyEl0bsZ$(@tLzR4iGkpsWP!&KM`oL_ZDyO)-L^s1; z)acgeqzy4ywmv>_Gf^YRg`P_h;11;eT3FONZTpypIv0&v9`(9~;LmT(7TDmDC0+Qx+?3XsppSs= zb5#|!ibfqVO&9zwno9>F+PB1assiu$bA)m*=m1Valnjw_>^~S79{edQn$K|iSg=Ba zNQ~&CB&U6Tgff_7tNR19lQU?!QD&2}<}dX5e@);td!%)en5L-+1CV%D_)fEZA3rkn z$CjZ?v`x~R_{6V(QrlZUi}{loLfezS7o_p7Dfwt(+e}9#WG8VP$o}iU#f#CDCuZx6 zlq&cENb|n8b_4rg)Emz}nrtGfPkhqyWH7f@O5sv^aL3;(71qkehKb6yiJM2v6lTUd zWI)aEWfUH&pQ-y|^;Dc4$>%G8#$%2l~os1wM5KXfEB4&k+5O;Tz*h!V0iSA6B8K(I(VfH7;k$n|sU z@}^&T5T0)=93FdgL)%M1~6tT@DeF zLMflYn5)CyR6Vhk=agG`|G70{3FgJXcv_V74=eo77Ww}i|DSpJ|G_iRIwJWx!)^w? z|1HN>bIx>{n)mL@*X2Y+u^>q_P|tiHIxulEFVdQ1v29q{RaGH>V}#~yGZ#VNwUYvL zS6N%^ogVQZ?QL?9cqsRioi_28Y4uh=j7>g`IZ~w&f8bvzNJ;VY8#)oh%l*?Cqx}Z{ z?3(W_%lk3$)pm3X7gSh;*f}7@qUJz`$@Nd`qrtORu+SJg_k2<_BW9U`^k>?q$P6Y> zS|~SfYn!K9ZlT$hu)Dd@o}`}}T?d<~=X-g|B~=i>2=ov3B~4?~a8(HvGwkSKA2CrD|?bZw~vA~uo zFd||oX+teq!8Mt|(`pFte=Ng}rv1D79;oyFuf@RhB#>sTP`egRdZTgk zs`!qzPu!B9Johxzqy}m5hBb|ggCXn~^>%mZHzRJ)wYrYy)a>#bn+#+)#-iB${e_kh zqx#wv{{>n;jxu3dnSIqi$f3o#+75^?!7oh&)F$Xuss|IIZgp_}2Dz4iCROsb>FRV- zonc;l@pcYDN|rNf?r+tvhyKfYJiRX0_J zZckWmRL9FIJbK5d&$bj<*}E-$3rnB19g(wu@t={XurKm6X#qd$xe)+7a+8{x&5|M- z3Q`UEkDlDOj9~yaMPl~6{D1KV|36sbkw_@qKD#EhN{*WTLx)p_W!imwWp3_P9Op1r zJ2!cc?wofozVF?3VbjyB{w zrd0;bxwe)Qlds?=m296%tN%`ny@rvIC?*Ls`y*+;G*hBPM>FBAD9AHunHMJjB^*G_ zDiw2=r3kcNiv>`Keh;HzWJ5$=a+)JP+W6=Kih0-lctDy_9g>mI#`vJcG zf~t*6@eg&~*Hf9Lnn_f&X;D97&o#~a*V(cm_|8yL0Mgrx=TZPGU|Wsd9$v~d2?UUd z)TkVUr{RwtkQde+O60c+-t2xziJ{IXLiM{PL8^1?EkpPv|5_zQ59aafC&mw zl(#a9)Kq^u!z<*2z1A;+U@Z^GugL^hOBGrKd(r+I1e?U#rNIqabi*Ayr&uVM8b!p% zzp$Q;_L>a7!Hth1HgJ7JGXWw^_DIL`lQ7_1F=iNm$IiAb z9_`Ry0g4moLR_idImO~`3enZI;#Tbc_W#M;p@l_DmpOrtsS%V%Qi=b3YYL}DseoeC zDdOxL*3n=GP%?%fHYWHkaflTKEVtt=*kGaA*%hoKs_-RF2G{)yHl&P|?mIQxGm*f- z_FxBxzd;e;nL2lXm}>^mWm^OAZ)hNwyl@YgJblE$#UH0p@CMc2AV?`}5$IQEf&e!~ zaLI)S(0TtMOeru8t8_g${|y45DTrR2&>IdVz$e#}XC%wLJ^2}Ei2dgYp7fS@14&(; zC6$OL6@2{3iNeXZ6822wuemuO^q>X_DqA4tDQ@)P?Ko1FP&-h zd$J+G+Wu-+LLPtzm%e%nT@yDz;D*)tdgWX9Z`41#Qo^s1vS@=N)u-E&KHNL)f!gyv z-Kk{(sgGHl6a^q@uX4U2&4MNq2lDM{R_m+f&NVErNC-a)7Gb@Ri=@3a9>@cV>8F{- zJ*5?=dV~s{ z5*SApOB1#YjoABebM1tf>VU&Y?2S%DRqQ14ptaX7@@=q+t2cmymVbI>O`zlrL&AyL z3{9fKC8ezF-u!%V+kbsv36Zdo@$O_>N}6dJZ582o-0ur8i-1Q6vkT|t-?o)`ILk8N z?nzy^7D|UNy0=K5vIM0frEg-Q3@~3LX}{Z+Y{s-(^S*7crZeCefavJ1AdZeRNO-Uw z0KHYaxM{=+`Hh|h9?d@8$q%;@G;p8TyrR?^cTy&VeAJbJf>j=GL^i96X5A=z5JQsc z=SC8DhaOPjBHfn#5ASCv6Zk>s3Avr7{Zlr@eUoR;>c8OM^chj}ttt93t5IHdm3cz! zV5zuT2Q*0IZ!jkZRCq;Gt^t`T>By6MGdd%vV~4m@W8nC9LRnqK*9HNqagSB-QKZex zVC0Ec6wTl=P{%@Ti%f(v9j>pN@e&IRtBG-8j|QIL?heuc+R(p}R5MBV;0L(LC%N}T zgGQ8&yR^^s_?G7h_s@{-;jV)iWj9uetNs2Kxpz7~IZu?x2aWN?0U>}4jC&{3mjwrRTVUaaWl6^>bAvrw!0eTO~8h9SmB!a&}# zwj`d9ipH|Uc)xQ=2_WxqZ^luKcynbK85m5ra2|H1i^-Vkb#g{&4?2uLqYSQ^fwj?~ z)#6+QkD)KzKkz>jHok>bh^9G~5)U0g!NLRw>7c$dxQ=Zu7wU6V5Hr&IJ%by|1T*$& zET9x@k{ftT3@_e`6^_%YB~uJw1>IheYRd8&B6Mu+)}my^DkaI!#4JAV^8p!=q52t0B=weM_7xY%jO<9t!f9V+8tKpX(777Fj(ngMz()!1S9vB zN!Li&&>Rk&KC<-gdbyER%~!Z?-j}o7qJ|LQEVVunXUJE@JfV##w_O19p~6u z$p?F9E@{VjGj_j0H%FCGIdMN{Y~qYfxA&u`r zG=7!g8og)3L?|VaKcFyDS}b%$)bSfc?k&L=k(o4LqP)@@v_54|^&%_JsWb4J=c?QB zPv1Miz&?J%p2tnai~&-fBh(^$N-ig&xOA%9%@y^+=RU!#T_5C;u~Ok86xK{v=-qF) z^`^EH^(&Z5dIMuO%I<^83RK?LXk!~+zA#OMW90w=J!qypnq+vnB<>P2;%r^bJVu#} z_nBZ)dQXG!?Bf(v0jd7ko9;J=GqoExJepXdef9;GzAUs|xzr?vS`3wB_Q=8j2>!Oc zBBh+UEC=$?Ut0Gh4-4N@I1io`%768|2U$O-i~Hc6f% z!`xN(>7`Rui~-46sl1(H^LKakZU!&bwya1vPj1>Vop7opf1zD8x17;a(7eii4LAi= zi)Sam&cDY^d42_mRwD{)gH%@19xMIxY7g!4dO^Lmf1N^2$tTo^T!I0qd(?7f* zz(YFORmYygqRxpb(ht*$|M$is=<{DgdAi4i%>^2UZS-M-d(~H0X6VdI-w7J_P@9f90mk_ux5z+UaRy9(15Xn=V6RYbH$Qjp7i6jvIT)N}ANpT*2Tm zC9A>C#eVg26!py1^umlW1F|Z`znbdw6|=wB%24QuYG_j0w`Ix@cP6)k;FBE&AtRCr z*P2Q>0a|cNZ~Qyg)nm9>n7D=QBDtH9TYczsJe|prA{Db8;f&If6~hIxMm@}z8C4wo z&fZNGUsQ1L7HJ-?n>Os7jglm8*E!0GDA%(VeH~aIZQ$!RzV{sYWRs+ySXxv);B`JT zsuHO11@L82nYLwEgL)J^=UX=XlPMUw(NzB^yw;`ZS^>X}8NSybz0RJ6S>>7YVc`SuWf|nYF={#~Cvxc;5)8zzRuXi7(

-^*G^cV?&MHN z@jKr*DpoCijN!R8m~D44{j6lWuCTmCv1648g!8}_jQ&+2{-;{oe;V@R1?i71b!A-K zQb?u8o~U1L>dNh>x(Kbt-t{k(hrHg>fLQ`CR%h73cQXkiq0VPnnJfY}B34CBUaXp# zs#$aW2jYC^+v#URnFC=s&)lnHZlcS32bW=hYVi%}GKb%w8pPR9A)uuzKzVT;Fj%y^ z4`5=A8-P`l(#BoTcnre^b$}vF;45Iy{rf@+r>cu*^uS5jySLzDz*v9}DPOCc|3_EK zytH`y8`t2(eJ=`gZwG`^X<;7^7dFVE#Cw&;*r|<-v$|Glu2x<}IeE~oi$|;% zWZ41chqr=FsUqSnS4FK9XSF##&g{xzeCn-NdRr6~5D4-unJD9g+k3fU9Eh~pQT)fN zINEh#CIRfDNpIQ$h}NQJ^(A4R7m4u!nQtHhJu``kB43Pzuv_EEMjhpTOm(=CCP&>@ zW``~H#NzEW<4(B3A&k&oUDWuT`+}s8=4)BW5Af_xwlfVyyZ}qK#R!l)n7U$5MHk~y zXkc?E=}_NSN4;U`rh!5&D0D@V5>2_hR~-x&RLKfZs0-o@Ma~?~zOo`9Q?Vf8l7zrePREuZE>?Pu?!578j=kYKGs5~BQFY2`HtmCb}mT~q|q_Q1ji zlh-ZZ%yn;D9(U#WGS5zLNz-EGvM}arIj9+ z<$e9Bs$`XfgEUPY4l=;hF6})5-%v$L-5X(*)*JceaJQ!NDf3MROBT1C^p<*VN}6Yj zlyIy%^jYUAo*UofTSagbZ70?g?T~__PHs(lGbU2k^t0EIwz58i0l$^QZ!|~jq!96$ z?o8N}6qVZ=_Q2UEpTU0Z^}y5+c_&+P?FE4h%g-&t$Zt#Y6375 zP5NH3(`17ZUHX#BBGt^}Siv%p$7%iXJ-94ATIm*`pV8PlB2x*?n67T^9aWyncDtVc zBLXIryu?`x+@v}chVbNc+w)aIAE9)MOUac8KICf#IpUKqO<7ng9gY(b$a~VLCn8`n zP(Me=g{QRDnJt0b__Vw4&F_MgV2B+x8u?fkv`HLY)Ovq~3c*R#nv#Ymu}22BBpag6X$%vA55F=FHH2v3Bnj+|1xrN{fRI z1f$3|FW9~$Ii%;#w%g`cwe?L|QDkhiXa|5nDzQmE**C!VYpJ=QGxpU+ zB-WzQl}QbeZRe?F>*;fRYv#%R6Yi+EX+)iZlU>9p*2N8ZK)>V37IN*@=0rj`Jnw;+ zK(x$Dv1Ue`{73H26@a@4lXiiEa>-QGj=o!aTzP_BqqAath6kjC;_Wc%o$z9EtyaxZ zBh^XQ?0f6|i%#q;PpSbKbGJfAKX*W91m%~?{XY9DurJ>YZ+Se7ngpD?Y%pVqwgWSJ zzs$4Qe;f2b*j;_qO%aEbcPN~wQTGV4aRMR8@yloZ8TD{G{ z09_|d#huvpjo+YYYXa@o^qXswl2nDZM>Z{!S|gIHwH18vM^rA&$3hXAP@th4!aZy; zeoT2|D*3s3`kp(dp*tA*oapr^7u*u6j+09dO+Ptz2aL6l+amFaJp>l$som~O4Ntiz zky65Dgs(KewSyVD4rfQ|#^pho4%#L)yWduEW;Rqgm_E>p2*^?5u*fwX52TN1y0fqa zo1O3_FBezgOG-F2iE-mFamO2t8Q@Qcd#sITcNC)*J1^Qly_=vH@0pXWa6)Hd`=n~Y z@m?J<#Wvbd@kxWoecI84ag*#h{0!1A$S2o!(CdD*w31mcQO9H1N2SgTl^jI3Zbf`L z-9Y3D#kj4Qfj1MJZ7&8t?OKjZC-y~ltfe72Et=V*=e>O6yD`r|@m(ov#A( zw?_m`wOG5dJ{CzmZY{1foe-p?!b+m%(r%*`qwdv>*biH(@H{8c99p>%k>ZjDBG{2( z&He1HU}wKlOl-i)HnLX9$N!@~jemU>l!5)mc$$VI$mD%MJp;nB@0U#&C{3c>!ciUCRhUhl8GnokrFlv1KNq_G^JBai`{f9>!8SclEFE;5cSu2~Re9Gy{xOIna8n%5@K{rZr2s36 zN=c}TFSjwP8Cy5lU!kWCjA`azAw2g00^~24gEoQuFrNpq2cv=oJqbH zWq4wEpCD?u31BLFI(|#DYy>f4Ar_ePZH)?6#}Ex^w0pnPTK94v@utYw@L;SkCU(J* zKdAlPsyz`={x-R-rVOL!V6bcP&8xUet;s#GMXJN?tCf4*M6!o5>EDs%Sd}Gd3QQH$ zLCB8!Oe0*unFdZ1Dkbs&0PAcT>JN`IJ=b*z)yKCv#_KY|@}b-cA9Hv<3HEoZRchFh zT;%uU_nD0vO7VJ`YPWn4*zgqQ;`7r+$=ThRsqOHa@y&v>56dN4p5aVec^MRrt)>L1 zjZaw#_U!6Zb1(Qg+sXuEA0N-PMN0+pg?JM!57KqB%rvEx%{}L5ZF3mz${c+kt#~Rl z3;==7f*L574~Wm*ex39Z z)I-ku?KNd+1^b0Z6-IfEcZj`{RV=C6*^sk3u8bUUn51ezRHbDbV7jF&TxeXq5DoI6 zFrHOL6H4y~M0M%OE-&7&_N6JeGr6S}IwCCgzV@vPZj=Rh-;}%lSb51yRwYsVz&CDW zS+r8bQq}~#t*uFF-@Nh{3=KY|>o}c-n|xK4Vb7!{(*A{C`yJ)gjE~M`Qn_ZeFo9fu zO5%7XVu9M)Ga0AB3(LkR0PX){9HSLe8MvX&nM_>^mao?=UQ5U{OM!~)o5(PXMd6Fn zym|c%v+%JWcVYAM7p_GYp`e~INueLxbqvFfuCpGl;~E8y)wUU|!YjU(bKr>N8T<%q{N zGg29q6%=N#KES@7aFlZ3+;ZhW+T*STJdP?gqizh>PTfh%^t;)eF8qnH_eUxUq;9EH zkGf*QbrW{V?)QaXD&*+LU5xkq=OxM_$F>DL0Ch8MIZ;uwkxDq8YLFRCsoL|DeQNy= zTwet0tObB3nx8vXTyd(-uw;#PjeI#X=>%Q+B`r*1#j>F3YQjwY;qD(+_ivnESN|E~ zPdPM^O@r6D2_|j0vs|!PHD#kbDf9YGO-%NE>FGH|ux7tmIZ7mGS8!fkt#Xt-bX5DT zT}{k4+0|DcpY}w(4u4z6bRR!0K6`qz&ZH?hw{>;KD5ooP&dBz{GFBaFy9U6afC@b- zxA|f{oA_LZ^Z>f^&Q9!48EmOZBMlxofE5q?11qj@z2PCXXr}Q)01~DB z|FZt~&$Yn+?wFRgy+Q9o*{oXqay*$sE&D@x;;z$t>kP81ThB)uB(hqww{!s(nzCSs zvDY*@R%tCnO6lOGP&%_au!TzqkR6~=+_h-cH`!O*wL#w*;GjZ`yU-)9VXynRIxx#WE@vrw>UUQ=i*p`h)=r7fm{rUS`ThOO*i($Az_Y= z&fz3qDuUY3%dv!uUmTX9akUq#NWDALeBj_{NIT=eo9-I5 z-&tD_tH3fWsr945;YL|W*DFu|2V)vU=?mU8DAu)EV^)SX{k2%;5GqRSH5STN+9~T9 zy2Rx@sds{54Da@IRyi#4Pq^|S8%aYyK8lgMF9Th+a(PvmJ~^}vu#-5WJ~X5B!iXUJJL8{J`pqwZq3s7Tu852{;X2@^@<{tDuUCQP(iXa)V5PN4^#i z5;4hAuHr@!S=4^~a=ih(BH655R&{*XwR~@+*OMPwZv)~kJ=VJ!i4h-|MRp1E%e>1VJDSF3q$lR%{oBmsn=t4&LZ;&{^c8(9E z+bmw^alQ&#bHx_$-C*T$nh>CZ)TJfd`7KKrK_}RHOTH88>jSsLNnIIWQ_(d-fOv$ zU1PjrPiwnRYCXArHs{EqH_%VQ;slgYE)r%;hkdaUv zgj9|IyX#_zB+)X0PEu?8!YKiH=kq@Mhe$cLYo>n4v@u{NMux*w>)qPMhjb{r-8jGKF0*io`ZEX&A6}PkRsSiOaTs4 zlewEth1ghZH#=)@2T$AKd~cuP6vk=vdVs_bCIU#A;~oOJ8s7VtTtJ5O46S#CcK-W{ zQizy;AJ_$7a~L#@Y=mCpR#@}9ChHoaKJvv69)lCLrTWsZX?ZR8ZEmCtOi1n^C%)Ok zvCmQRiJq)*hpc!`o%dN$i8b&L2qE-Q$AXUMCu)ne1c}a`bj_5|4or0}s~M8Z?-=dw z^42O==Tfd}9U@$s<`ocCyH zAwS2C36k&QvfLXMSlw7!pV&vWzuCJrbcvp0+5%_;5_KXss3e_Wj6JohrWSA)B?5mHvimhdAL9K0;Qp1XNkaYxwc_)0 z0@Uw62XDDY6)En_vUp9ZAX7z*#sw<^WF|UdTsg=l39lz}2gEmH#ZHSML&aE&OmEXN z)IcrLrNFS?{EUNCAIZ2u;Frd~Bg0m2`f`bmYlS|lnd2k+l~pd_rXQw*euKI#Fn&0! z89>y=g=t+D&2)TdG@C!67d~rbB+w(wLdRaVQ7bvo6t!|iV2V0fsLWM=G-TlKaF+@! z{Dja+yT~`yE(*hJgKv{|w_bm8Fi|g_5#9Z=gV09|C=V@(j!=fPK|(o67@=V6B};!L zY z(7~TP^V)yixt6r*jpr)PPRha#52bo0FM;4?^ z8lgOU7ABV+Ld`>7=E%o)v6RzErmxA6r>{_cLPj*?J38bVm+EpstZQ*(!U7bnr8?;?Szhs|9k^>FhvoM30nCe({a_TWK0;x$i zqt?Px@J$?7W$d1--&;9NtQW(^5zr^AhbdTwbLk|jYJ24w&jP80maWtN$H7-f(Sn^7 zIK@#5J=@*U{j}|YJ3)u$Jaa*WE)-bnm~ci|lKYs)mSXnSZ=ocy%FNq{zsoExB@%v~9&arA>7z5w7X#X#Hi zvG^v0Lti!DTWk_;KUM*^?)xd8iV&7o1}#f+FF9fim~5d1o%DKo>Gqs^g2&%7gNV!h zK4!6`7f%Yoa5u^Iel2i~I$}vqLj?1)r}MVrj02c*&&B|BLdxYW6TX)IXXGl&Il&xc zxjj|-#+1vh=ZeyaJBE<^=%DyzhQf3$0lJ11qgED@#_1?;t6cRyfT~k-#h$W-gz^;L zOR^yvG(+>`|7!2bqMAy!XfQ0&DnrwZ0vdf_qYn{LXqkgdDp9eeBLXH-BZLqUkRb_y z#Aa54>41PT2nd3p1Ti4RfPsb^7aD#5Ay4wnN`V^;PxBOT5M%?KQPPVNbPG7fTU0&e%F?(0u^IY z(Us(S^3usk!6IC(?swMmdYE#-g8K`rtNFBIn~V^DO&(35U%)R+cj;i*Tm3qFj$wpx zyS2WRW3McEx1}v?r_lvQQ2|6J0-RN&A&S}+KWAN-tbGgfoq_O+J?ItZiJOy`+_bjF z4)~^B4Y*}kRm1AshsU7$OQwY3>)oJcYnOQL4{+~n=M*S0;a;JCU|ST~*S`(*xp ze7EE#Pm{{RlMd>RXTy_62=j|ThuqO+ATnxP*Rrg^l3WO7z34gAt$p0#gq#g)Y7=w5 z9Js|d)t6DpZOuwyN-pSCyhA4ZQ?I}2-yZ7k++=Zqu6q7x8*-Z?$o`5{^r=1*bM3s; zuPL-g+xW&0aq3Ty)awTewiRZ1^_erH?ToO|bDb>+A-##Jx$-2{XfBNw5yF=9qK{3IInm4pZ z|M4RF-^)$^_AdJW9WVEan)IAGCfF~QG|KX)n1p&1XKLZU%UxvBFw z#11C;>DAmYYi}_u`Q*A>w+DV5B%@ZSQpT`l;@+7=9w}`#m9KX6cbqFpje&$?<^w>+ z^LKnwCCRMOOXZ+~b=vI-;l;k&5_6t2jCO^oEDbB_*jJf*Lu>B^g+cEu^K>U%i>bl- zxw7W(q~;ak;gTP#Xb%NsovE9cdG0@^;Ddjk5-tF#))8W+bk{R2Nps*DadiH2J%4JIewgsFLzqn{(=m+MfK@aQ z%b^6(Iz>{QvBowGOsgc!U$y2fX$*p@qe_C^jt>yItuC&oI#DrGW2wLQr;OdD+4b!C zX`3j^($kTLfjQpg+JxtV0|AFm)Pxw0%v~7{-fSCgY7>RZiNLDMwaw%z8;1f_XdS=7 zA8QBAV}ofAU*DyK%M-6gD5iqjW9@`7d|j0Z6uu2d83U4I5}E+lBTdbzwV0jt{^c~n zWSofr0oJ|9BufAs8^x8N8wG9|yl}hoS1miN`u&=ebdG`$lxIy?3`w5^yta-kD?rkN z`MG)AA|SE^Sdm;RV^P$wZIbuk9JcE+T{p=&N1rg5Z^$k}S@Fe`Ujw8@8qxtlPlVodA+bp5FL1YEM|@X~fp?K)Elt&5gD3DAg)K{KrvU#Ir4g@SW3)YY~LE(|O#P z>qq1-?*I@s{Pi@bT$6OldNyFuHtu%u87805m&NP=SKWY0Rsp&(#NE1z# z;0<+yj_{i)wpN#t*a$$#RV}memWY~(JTm_2swF?=lxVGHcHm!g-c%0$7P;xoFm|s| z(qtKWHzZ6$N>+nv&RD!@t|N{J3{K7tG4?`}D~|69`vXN>j-$3yg)2)L^H!#y;=kgW z?{lA27bX)e=VS-*@dxO4JuGkEid_$tdqeeTN%0#UdY5RP<8_jNt#vcz;!Jg%a`{f% z5v#(?-X@4rG6F@fc#0{BV=gJOYX0cUE4&x_fU~-Z;f{-Qh6Uf#5O4C!OqTE-eK|Ah z#b2a_U{7j0Z0z(QP`P#tZt%bw$suj+CC`T-n@3q(M=bm>h&q=`uHL_k{TAfTv(-XYR^4ISwv z2-15e)DS{=oiofOpG9ACdS*hS=b+Nvas^7 z-M-Bwa*yXBzo4KXh*L~bR6ydvBSC@RUqW#8>Q!F) zu@4~uD{zULfRLKtq74Ay|4(9qKRtkde-KPRc#ADecCO{S!UPHSk7{gJs-z*W*4^b9u{Iqq_D z-McR+BrGB-CNBF_PF_J#Nm={3j;@}*fuY4K%hy)cHnz?#u5Rugo?d|;f`UUr!=Q2T z9}^OjK7CG3&&bTm&dJToFD)ytsI024scl8HwRd!Ob@vR9jE;>XCw@-OFDx!CudJ@E zZ=m)M4v&sc(5IN+{UQJe|E*j2=YQ+iU-gR`->*wVM1({nzxzdS$sOMasfmd1@L#?z ztx59Ik%m>k{|c>4Y+6anRklZ3dvxYbL!>v@1?M?Xzq|IQp8aDT3;6%kv;Whv|J$!g zfSiy3fAa{b0Wffa;mWSdbS&%Y3C-a7@eVi=1U9&1tI*%(&c3;phgS5;ifZVp3;E=j zD^9|{-dN13F0Oq~dcR^O&GN-d`lQ6;6)WoYRYe-_6ljeE?4|Bi(yUMKN!U|HfpT5f zq)MmZfZ>Xa7kLYbysYbQkFMI1S+r#@UC@X0B z?3MLcl)p|Qj0t3W!|S%{mc~6S<@$z2)9{$lUO)h{Avfj_zE+F!$XM2%O=k}{E zqEZ@c@32nb`|}c>3sl$pE`TeU^iqVrtlX0FY&O@Ll!$(0kTJAIHW{u<_=94PZDSEiS*d)^A|&IhKsmBA}4kXL9D;8K| zuTfL+@z*HbFLS<>Vx|ld3hYiZo>oL{bSiu!@uzy)D5qm?aM0w>*!PcQ?nuMxe>4N! zA>%x=(v4_UcQ=XZqJYUCea-T(jIBnRmY*vX@CPNE%f3&SQt5$;@X8pPej%Id*;+-==D2 z+{A+49Gl)6-nL|jBU8+HxsywPdrpv{_#+p;NK;?Md0a~T%%dd1N1o3pMq6-ihl)sp z$jqVa0)SP6cjxlVlkfTSE5{LxW)MPqFMtnbPcMK}7cd+>rcRDvYS<|w;>x3SzoYZq zg&~y^3gt?@^hf*ejX~K_1HC$mGS)F`!)?J^2}v0eokmhZ(qOImgjtfR#eRmhx}tXc zs76BaSTUBa7%4FS>6F+1TX@TD%8E^4Vc}ZK4>yE0GH(pBL~#wi%NHjtkpo_!<6Cu^ zxsp`kj_p46e{rdLcS@e&9OYk1sW3w4Y&Ph!OJhB6d@Vwjy@o(Q&3X1K-Cnvt!srbq)`Z!V#F5Zr9cR&31nEt{8f|IG;J@>r;lxGX6 zs4jpmJ^VU`H?2Rpb*L!^(Eic>SH^(OfBphECEaW~IyvX|`z`^#0PdK$;~Fo3EWlw% zjuz1Tqy4Xpfi6kN1+c_jd;uV54~sC(dlvvU$aNNThWkj+BuA@3@JIXK83TQ0&;<}v zTz3JOL58a~2@w~-J{=OWxr@=beEvUGUj0jdkFH-;k-mqwLd}?~SetO-X3f-1sb{*+xIZ;$ivuqR?PPZEkbw=B@qn0kRO!S6VYNUN83{39gvq28;Q= z%c}-WZf8uj!4?d^6xj*bky1FwI!Q|%pVkO=z*+C zufS_Mf4Ec95_EG8zZc3arcp+hq}m#06~iR6i?X^s9PRFA*erF5ZC8|foh{VW0$k}( zfcwwpj_&!NXE0zk9m)m#(f-*n=s=mm&%@^HdYdPbUUxOkz9m^k?q%IgFYddionVmf zI9+#J^`tFQX0|=)Mxxu^^v|#l{B0hu*#d)^b0nqVYQ_n-f+@v-m3qDT%GmR)1xHub zm0KJKN|L|!nno{xX1mEv7{$H*2U^}KmF~(xq~9~-46+J#&m9g(xzcOZY5#Orq3FDM zd~~2i-&pB z(KK_x;kn(}LO;X&Q^c-PP}K3?DcQP$UAqJhPvQR_FJHU&iqU7CXJ5hLxc3vbK=I-8 zC9zHabuFy^Yz~%d$i>>wXi;K(d~1(lMMpDEUVOq0D_iU&|qu|??GM~5Ccgn-B;*fEKpMxH3QL)iGIZ6n(9*qA;pGEbzE(O{pA zV?~(iaV&crD__#$O2W%)m6Goa(#wJP24m5qVCuJbkUo}sXR3Mz<(qXGZcmmiNeRqy zTfpB8(4QC0C{28%8cRw>i-c}9sNEqhswfQ5b2ESF77-imhdT;fu)n+8D$a>$aM~BK z9Q@4MbYFyFteNQi@)c|@VF7x?lBpBHBr}&wV^HthdOj5BAB(0ji3)T%nPD2?V;76_ zR?KoJVBtJPe)Z8oy^b$J^SdxnX;hB7GQHMM3%F%K6mrX-$8pecrs0ut8Loerr_dQz zCVgA%XfhR|3{>G!Qp8S+j=T~lKx+7^gndXti_3aixisYibNYK0+{kx4Oxkof!!#`# zFyD^l7_F#`92=c}MJGmNj?eL}A{mfD;aS{iKJ(20ZhKinD8m@_TbX?L6aL}%;^JKC z@9`7xkM_@u!9D~!+4&7(mnlEXL$=aGaxx%|ri9GtY`BY1Pa5_*KeC6iKiFOay5X10 zpbiWkeNR94Ew^}rf2!r|Yx<-0?mkdYkbXM=VMzno@jSyZaAPlk`C_kVL=otWuXRXy zGi;YCWu{Jj8k7W%#B*QajQ3fCbQl1aONq5F>8|8vbhaTAyoL*Sr|`EfxA}MIxW{zk z=_^j_sT>%uDL<)Wsx!b`;nufJj{bs|w4n-JhmQ_$!_ z5G!`E4S&N!vq=xO)^YLoO%;2`pD6|(CqA##c#eq@1uIXcTq%Cx!5PL8)`5(YBQKP^ zB5AA4MM$N#XBF%cK3lmyN%_src~ZY^x>E(Vcnmt{s2=egv4MeoSDVOKqe@kp_b3rZ`>Hn-H~qT91A<= z{wc`ID`<4@_WP|8>rZf~ zd4)Lvd~k~N`Wo}*P@tY~+)*?8cpAgE(V>U%W!doKGRmWO5ti*Y_imCqcY@3bZ6|bL z#fL6{QY_!?{_2#y)4t6rw8HYhqKT^-BaB06#r~UlV@Y^IWmNcb(*m9zabZ^cudw$`#vWCx8zq8dJ?H?I~60BxHBn=`d zDRn_dwFJ`W0a`?B#cS$W4(Hr2(}>!h*jHz=jV|`6)t~F_B=}R{UIpoL9>UEBUCM{b zAFE38f3hG5-RVpOe->JTPzna-d!)U!yp>kQ>`XI#0fZc_;s{Nrj6XeJ_KMEiU9DhK z(r(e7Br`MHH2b6}M>pZK_Ls}rYSU{hB&6K?o1XSpS(~?Y-%8@QfE}epT7on=Pe$z9 z{Uasz`n`@Khx99KAB_dU^>VLS6gR&Ti7sMLrtyvo6nBw`Rg+{4`-nuE$mG>iW#$j= zVgqXB1UYh^+f^N4N>xFSP1!gP&kh5-$Vs9c?-FM=i1Sp>s$kR5Ehtkpe?w1+_#<~s zro=gaROa!e_k2Ufl;2S92;^i?QBiw#QFhUyeML=EIlnBwq;RFWnU9`9@#dsM@L^bm zs_wG=HtR;)t2i%!gdL^~45wGCaxZlF=h-IK`Wz=$CUMNS7bZvM#NS@69Pi%Th3>5V z2}b_~BW!f^XQU+;q2C`e!#^%F)aNWTmaNPeu2Wa_UOC7jJnIYsz3;vN{Pl8w87tV? z^S1R>a>dk@Z2hjAE2AI`9zPuV9UhQ@�r+moh*>PUcIe&u?skGyc z3|wEz(dsb#MY-Z=MkgTxS%6*oHL(t5-VwX*vA}uBBMMfju!m2fEau%vn@aaTFG!_H_;|DNURUjihg(sN(+ za%Rzw5tAoD_AatiSf#~#Jk(RlGBIMN3YvYM9SsX`65CpMs1VsyFL#mD>XQ#oBU(OZ zP2Z){%YCPKLgy@l6q>!{b8X1=reH@M|HDU}ckZNBy=E7d^ap(d6j06`!q=^lrsRR@ z-#Ahfw{bt6K?ydONxz0jzLUJ?$6^>rkARU&HrJG=?9+X2>G{E?1DplM8~KT$1_E~* znQX~jRc;M=+jbL|J8^gIKRoqQOAQR{TBPf=om9(7#R~0RuJI+xHo3gaQTOcnQ0sPz z!uLtZT*4-FT~3SlUXwm$cBDxY#RCX*7*rGMb#4Qe>KPCXdUA?Sy*`pehqMKfJ*b6xBYO zTDz`5yvfqvW~pJc9s}Wj!bIQkg=_B9{?_t!5jz{3T|z)Fb=4v&FzRSMSqFCP7Heto z^*s5?{6O&-Ax49^)~>b#54APr4L;2gY}kzcWK&A_$kbSyE4X))2oJXJL*CfH4yns8 zfTm5HP||#j#TU)-nR|3Y#s1_uED<^v09xS?g1KyX0nDZu;1~nAePxZJ7{&#SV)FAJ z0-eghTJ9g|?;lS8^El2;(91oHzWhY;(Ras5p(UIZ;Xxrhh*9eLu6}|ixH?4Yn?7DP zLNhMc+vRmHP|^1!p2o7hw7(R+(9-#*Txc~U*fvydq$(w0&d_WQ4y!G}D&4J4JDc1w zp+1~|AKxyel^Rnu%Q><B{=i(0;Uc70*^3Sl zRH>{R5qwOh65J(#xKu!}MkxwXp8WTT_6Jq`&r-qv@%lnAy&Y)6kFl;!x9vxxF?hEM zz}<$mR&jUX?%RQE#$F6(I809#p8~iu7i=(iH$xwp*R5JM)bqW`pk1EIvf$8c3~ELp z2-yZ*_brwCA*{0EJUu{mu0C%%VqAqd2c2=Y4n8X8tUbqm-XJ*3*za7i%gYMR=~nej zwi1OGnlFtH4f!<9;bFE>ipF0SWq)5M{R4CT_c0f-_hwi5+0|W6QRtN)k%xpr^M*p1 zGrXv{tNXLopZdV1R_uPpRJp|`3n?YXCae?>gxiwr8=pZOA#IEMLiDZQK;7SK#+J-d zoR^=fpO9#yQ^0dV?RqB_2~y$C>!laKF;8RmJ*?Ic#LWAYITX;qTN0ZuB|^E&PhS>6oc^7Lh9_{A+`AEM*qg5OIq6vg{_<>J6TT&i>oqqna*MG$J{}`+P znNdBi}3PFCM(P zq*a2TbZnFt1z>6yKoH>Kq8HM)0D_9Yt?-3BUQgn+&yn+X%u4kQr+M`5>p3bg9yUNX zyKwk=;)+2%o1MkXq+-@2+)De@L(I2^@4T!%?-K4CJUTcKwb zMdbupmDZwCBD1oiUGKM@FuD{Qj(y4b#1I3al-wLi2;{P42@trl`-JSbBjqo+q(9sf z|Fj|bHn^R_Xv>sIh4zh#7aF|7$9Xh~7X)Be55Ik%vEAo;;1e2lc)J99jFYc{fi@xw zJ!-AZITkD1%U`&^@N>X=#G*-$&IwgaZb_GY2VQeEJ3X#hj$To9ls!rKRm-;S2;MR< z<~q+$Zc@Dfx*4Bq%DKS=|DJh{t{*+y8iO8aJrS2E`^41KP1|hM8>=SW>4*~=(&I$o z&4U+^`(}tIh=^i6bN=FkXXD}Tv?tURm zjH{J-XS7=$b5^qT#yc;6I#&M@80A~=8Z*@nqLQLmpL`JZ($l->hXK2{D5FE!TaHiI z)cr^AeLYht)I)JRafeEC4ipN<+wK~&v@K#PiqSJxl2Ks3ULowuIbO>~#x*J0J*=Z~ zMJDRzjLoHcWO4dSOOc(Tw->(G-(28B%L*ch;Rpx%K2MQ8cW;jHo1A`2UtI>I&pnHq z+Az`K6?|ki%&dE3-?od|9(f|YB-XAQr1z6vlXvcvo*29=5=znrC$@=P5JB;#l-b#b z)QvLRaprxfT5!jL4$ zwVKaiRMt`G93D?sN^@|%Z67jFVS;iG5TbaDV7^hahm}9j*h6F#vIV|-56pRgwW1{` zLL~~u&+Asz6IQTUV#03JI*=>wqt)f(WK#$ z8XNS6K_y|lLyGK#zn1)9NyU>>omN)F<+9WW#*T~$a;oJgG|!!TmyEi(B9|tDDp`7k zNuE4Tj)MZtc-Qae3t)}75Hcrn0hp%k6}8kfj9WhJy>S*Z#bNxgYfYiZrQ#2E?w?^K z%UsVjUFW>|FT4l=rDmhpt97&Hpzk<6|S|p_?##G~W zYg6=-*19iVED*xm2371W+6Vi%_?Rl+vUAoE?q2|lZJ=gkHIix%p?F3WVv3P#tAbZU zs_Utcj#xQ&ItG2Gkt6YgAL>=@<&{yeRrGJ+kRtY3*LoR;c+8J0uOOAJ}H51 ztEuI1JuW2sH^Gr!P0>e?Ef}fsn>VV?XQ3CsoS6rn=Mp>ThN5*ceKYi`$Jou{#)Re7 zPo}V!A-nktZQe(0N+6|=96_yP9lgFuN1GO7+xF>Y<-y~0Q?y)c(rZ9-o;6-s`f6Y0 zhQHUB{vmk;x@@z35z2swEA$-#T0uwmY7MR8#J^Z$3D_5>xnSVM( z9R=1ua}T}VD*QuNrF^ljigTX<^v#Pe;4pg2fo$^0W~B|uH@@PPRg$%bVA6fa)%8-3 zqz>v^uo=EjZ{I5I5OfcW#()`5po<7jG35fOaPc_AV3vP~ocp=&79P`VOTGEekj>xw z><`5AHxN%ovPV&xO-=%9*kyz1_kef2al>jGgkn;vy(Vrhs`2S7fAm*EJdkyaZrgHD zpfGIOOP@^_@8M1y!QDZ-;WR;$lRpR4%kQ?F?vpIo0dpNXo~{Me}qCcd3DKV}<*%!CR{ru}_1h zqhDNB^$_44h`?I(%g(6A&-O+@0-}d&9JOsF!ri=UZZHZIqUfalWQdx3>ji2(zqMsZ zn;cOP=5T9yz#8d)`W4nwR~t2R1s_JLf-cEjCz4}&&u}TTXuU<|KG#0#Q%D&6{y+g% z9Ix%@I{$^h^N;$X|1~`-&nuWL)P`BF+tSAx<*hf=IHYda2qzZiwrcy7=vhv3Mp+2GL#zY5X{;jPT5JWF(Sv5j^@!_A=~ z#i&vimlr+q_{lUp9>Va~lSfx16su%8Uq|TZQR*RW$WKMVSYQSm1~tt#z|h9!|Z8WRJ=U zlVK5oABdHhvokMJWNt7W*eknmz1^xTdgr@c99gd!8uWQ78QI!fr{V@vsXa3tiYF^Z zf@KHKMGI%Idh{P4-0duVUipIEkYSyDS9kQPI~~vMrtAj%SUn^i>@&v;S=&ctMXqF= zn@2BTqDOu2+YTQ4oj(Y!<{2|@`sp~C_vQ?|nKvsfp{I))APQ&K;(FyOMo#4wSD!JT z166J48Z4eY5s0(@1gkwqu6cW(+l_6eC+KKNbfHSn_vSXmwWcz50-#()YApYe0)hK_ zQo<}Gml5{Cv=3G%G^T^&<4b?d`*CfyPN_x92VDTZ&LLlzGD!TPQD<8Ai4$-3F!^$J3~ zL?`A$B^$tuz@`PVJg*{5@YC*`9ePl4@bc+<_UJW_O+%r%@e6?5N2p+vKG`j}C!$uU zp)UdRtSzj!Y4LSY*wba5NXrfC>pH@?%^Z@+fOfqOUyB zz25P5F>XFnrMefGYLWEZZucgR95&N8GL^#>dnD>$9uh&sft26n|G!`rT>Q8*V!Wgi8)%!NurS)DFGR`%3oOYKwBvvl9$y-CWZZl?A);y7~u|)K$A6iJobJ2>p63t zKl<>(?{!%PUl3zLE&ccU;9XGj=5h(fNnwKteW17f}Erg zX-6bjC<=DX!$Erw9}?S~#6C#b?<_9DWY>J_#wng7HY&o_<@KLUJ9Rth;H}342a*fW z1uXqyZpE0~ex}n|ZZeL^D`5wpz6u!&DX8U_;Y;+we{&qi_1?Z# zUs{`Ko%Q+C93URA4(CC~FAQ-D2U&1-<(%IeJp%v2KE|0=d^3c8_5A$Pzh&U2fmU?i zv4Yq_)3smdIuperoylpl-ii;D`tgwyM?xDZRD1QNjKcvBbsN4=NCYC!GA8sy_K>Zx z*Fmnm6Zjy-askN8gIv{7zUmGRMP5QLL$YUgt?DEtW^WNCl2K){HGRP~T}y{@7w%oc zF9j+})Ii7-B87!!G9a;vr0+eP*@&X%RTU*|f#Bp(eFi<`;ipJ*-i8kUod0=)ibQU$G-rVkNR zdN4)t*e6CF5;kzir0U7}q5K|JZ$aQhE``WJd_xv%Al5k+y`H?()>o$$g=d_T$gUh% z*kK<|;kT1FWBXUwn1@S!iWR#FBu(4cw4`BRRXh!E@ zj%de$s)vrQwcdNz%L+;sVPf#cZ`GGnng1oO7g3QmPoZ@nlQ^Ifh2bavVd2Jes@vT^ z`_Y?9)--Gma+AVhz9A<}D9^ldissg5qvdC5_l;1U>R$8O^PLgJ^-A7ZCg{?|GoeFt zYTWso_n!0Qvo+JO`o~}610I<;K>Zjxi_o0vd&LR)mGL=~4|yb_MSDn8X9Z5gqz(xN znvUTecB2G$M|`M=zJ!bPK1lEJ_ToHUIxIs8Ofo%cYoc6+3PF32TGxU!gPynrfnv_@ zTT9*cm>HiDWkJInY58-XJ{^uBCQK##Q5>zc_XLThq7OORRighg}N_+xijik3lM502Dsb zMUq)LQ0xm#cC-y^b$Qg#%SQ?8D-XrM@T1~_b{tg+%I`z1eOvUKz1m)TM=A=U@Sza2 z#%0b9n>bLAVV4s@tc4tNW>>{1+Lpd*uQHY>kc#OeSM%{5K&UM`7;{|aX)f_{0a0Kx zE^yVN3_A3Z=Hc2amV?7{yW#iE&FfEy6+}Nyb+K3)KtVj!bHbtnmA^gxgk30gEY7Cs zWIG>38XnXzE;rIKe4gpR^^nZstDeeV@bCZs^l}c2A9?#yDZ(_W!8~^R#hz|k85?$O zg}g$bTjnz??|Hs$d!?J-;P&~q)MuDG58x^zK1OM_xfGRIfW-yC zvuqy$CPeNPLtW=Jt&b!MU+aXm_Rd5iU8}H!wotgx`L}X;gPl1jKUSSZ^E7$hM-?S7 zJ}y!@bfvml;#PpvQw+od2Q$nYi=bQj@o?fk`m@t;Bo^bld;u__y9K%BOG@9l-b>28 z021#*LeDT%otp?WO{@I^)6*b)WJ`c0j>Z*p`uPz28GoWa*He}g2AM{WNMG~^Z z)6D#i`4!QzkZJf`i8!X zL)(jUXA?RP_xDHRgA$4?H5|M<>hTZX3Jd#b`QZ;_5ZV#^ek3+uyjH(Rd}OS$jJC|b z0Qm4)B3;e$|G~ZjHW^g`N-M648Cv~Y?`SErVWv7Lw)XV-DcSGw{A`dcaacV9Efv_* zG+b7osU7|3@jHfli`@4erwX>B9NKE9kr5J43*JNxz1=Um!}LzwEN=sX8Qa9s53=s9 zgwrLz%mz#BQ!P3p_n3&CW+{BQj~_c0%_y9mCf(RT?A*ce<>AZ>55*A_cva}SciB)s zf6NGp>|DqKokFW>1Y2-pb>TxNEApA#p+Jw(Apj9yPI(y|ynj~l$n`~SO01m6 z`9K@5ALPhc7C407xd6&u=Uo5}=;RAv$8@$4vKpm;3p5@O$tXAfDK#TW^(OajV|f*R zd$uc)J`U>-`R0KYY5nPN;+dQT?`OsHq)zXuBxQl&SF!mMxCr$!X!E>*FuM~~NS>>!@! zY0>B1dZaHro{Q(Z*0}y2>Gk{leY7^9^pOo6(A$inWE+P-{`E7 zS~}3Xzc&1Q-Xck$Q^L|+JG*xfe)*3{1 z*O=*~b=qxuY76e*_N}t9103jxL6&Lit5c|Xm3w_MUZoE)E=fRI!Lc;0i)i?U6;_Ea>p+$jX&oB2s{1mdO*&=dL~*#?G5v%d!h#3 zIA6b-;@M_XR>qxx*dM3$<(I^qJYQmY(b1!fQ{#uBLVc`@>eV2gg8fP{`7OknU0ZqY z<^@26PW^0gXWrp~N1bBIY#PbB@uTy0Q@;pIgk;$>s_&4toT_A8`B}7YdcnlZ)P}Ie zjA0jhfX=T7@PJ{7N%ybQFuC@hzg-V_y1$sJ9q&&lI&4csD7hw|?UZMH48a;5g0h}& z$dm4izpGZR&QIx&tHU-ZNq8>`4sF9!zkg(s(G0N!Z#+N2pSg5=xVdf&d0)@sIyJCO zwgmpUS0#_n1_SGdOc%iYym|GjGahgayDv$xZ^jNJww#7Vv2YY0{_Cb*l-$`D`R>nQ`|Uc!~pC=chNoVaK>yw<8XG%x54>r7hPq>FoVetfIq< z<2lMh1V&%pk3_}D9e@|t(F^q_25A3$TaI+Rt4roRYX5c@WX`0Cbg|7McW|h#v8!K) zjb_A^$x3Zl+x6&|1}Y|VIeJLpZgrpb!W+Af!&|+u(@jUb&e^DdhOQj$Uf!(jJsqHt ziSABbicxPAPu@Cg3K2I+K{mM8$GCx6{bb|O$&1GmzZCWsd+M4t580A4k99@uVALMr zCLdd?=&!%U4}@^flg=SeaHL_$-sJ}G;5e@B=Ler>VO_6quA6S-iN}stic+BroqHYr z3hU}|`U^K>#}sPX!CMnpkFZK1nO!~SfHJh%rMa060dRh+KSUe0Y~(gWxOd)owcO^D zd!jrzwYkVs$W9-(`3QEbfF9FBQen#iQLhU*BYNIp#O_`Idwd7tF9YKzIqK{|s%4!& z3whJoYM-X!JPIlXDob3ZN%@8#9qDJTTOnb_7eFnz^}CNsV04`t1vc(nb)3>qaoxJu z?p#X!6zy=f5nW>1J${esY_w^dF%HMftd2Rf^wr^Aii&y22|x`yMhf4)sJpJjJl1lv zt0?*S9YgY`WaA^7>c?9bfTWh#MvIKD+Wc=e>eFAtvp;|Ruj8DgTQQYF^=u`{4_kJN z8GRnyBXQBj4$K>j<}o_fiSOx)f2z&@CbV(J2EsdykOqy6t{=Y?+E1-mN~mzctC>g^ zw@q^J$C!{^058|g-qc3ZpD2@E5{ncM!kmE5ZI^M(9*?%;6pC!|=c8q$_0>NEk5gk6 zCfts2er3joEX#a$;MmQT>U@`bJeO}-qNH{)vpaZ(p^oa?KP3e&*1OK37RWR`Ujw5r zj0r~i%ywW9^@FGm=;szSiNaZm(7fRM{1LB=)_UHB%~?`BeN!Tl7;ynKRK2}QlmTwc9`uE-C}4KI`DAUjIE}>s`FX_3(*|vNi$r-_yQ-8x_@YOdN_Q=VPBhfo; z0pQV9jE?hyWoZd)Y~R+fCnG5dD{!u+ZiHH?g7a4eMHbb4ODFT$d2`G=ieq3N%bRxy zMvC-T84fp8d|#3tR+4I(2Ej*)yrg31zky^|T6;v7*e>h#iec5+_KIvB5yGeWb*0>^ znn7TTfDqPQB<5f=czm3U+j_ewT`?`!3QubxhAOhb6|nAw=*gCnPp<+bt{QbG#RWy(W@><6bw zggAY<1ChlEKcUs+&Zr}Moe1I1t#={>cfgN*jo`~rzsNZzZEbF$Sqv@7*DitWLo-yn z7jxviFZWXo`yzyz_yFGz{GkU92)cIM&v4ul1hl!sHVccB*i_hKF3P7C8a%sA(QPAd zd`=Hvg&s29YBf{~ub2<)s|M{)I(%-TMC=t!=D6m`+9&j%f@>f9s@iWo{KVj`9KCV@ zsG}K^htu4x3rd%@s>gWex+qpe(l3C>SE`=QUS|UDyjV%g95NKqHd8NfYfH!kiB+2L z_?)DXrW4qaL(|UVMCNOH`CU0<#g?KnY-Vzsr=yM&CcnP5--0bTL#G4!$sEtBJ0SI* zLXx6OLXK20CY|nMBZz4jVOxi|c4UM{x!QgonhRH9FZku@2A-Lpv8|Ky`snnA0u-F= zSBrZs(Dpn@i@m~MYKuP_eI@Tnb8|ZM@Q&Iw-2j!r?#h$j1fBGM$+Z7V_~~?1&b1IK zA$^T?!PAd#GF{Z_SV3vz9By>v26%#7Ku!BJwHTO`g~+<{1``$jU1 zyf=RJUjPR)FLX$TykMPlh(f_-ULYb6f~9cC<>YF%=D-H^?( zFYP{yBN^Bck7KtJSXcXP9CrE7(8Irdmge27-!yLA3fmQ>{i5>jZhEd%Rc7bOkqQT< zKLaxg*^6BZ#F>wSPtGA7X=4FrihAz`Rnw+6KFpBJXWNyQLQCq`-MTG_?UyZb3>Y<)PEJ?T46Nyl&Gid9{Rh{&^q` z=A!`xyV3)Foy4~V4rAKqyV~0$o2E}uIwelW9`^Kl=McoFnx40VV)j(6;yIyNXV0?I zDxDLnlh~UET^-=chYhz{Cv}6^X$CQg1c#X zt*N+o_kn(Bi_@@~02gn1rAY}3b50H36w?^>O;j8Dfi7;#>37imM;G;RYfn>0r5WlwS#2!&@St2I6&v85_-1qX8{P3aV1g)D!afXlu!aQM7B z&B7L7nwFgPU^G5*PMzI}WIB>WnuO8qWzR`M>kZt`+L}hcGNaxS9>T3l_Ku^oQ;k%G zCnX>^nvHN630y^v`o zMHWo`ePsIge=2%E`)u~Z>ZskQVUHwu*GVmf#Tl!?|ICjGth`*o_aP^7_~-QL+uN$L2t=RwP?Cxi`C93G$-`X63kfY z=1BML^%g3yXYtRf4Tr6gnH3Mx2WLs;*PbRWe&`h@RUeHlZuUOdhLXB=KeORsl{(MGdW) z+ZQ~__p;J#N1it6XWQhDM!QKB2EQ_s%Gm)hKek|UtGt&&O!%ik z%iG7Nnn43dCfjMD4^HEm#rLd+GHc#_{0M<_`{lsaW=Xer4J^kVm~MAUY5BzBBnQ?_ zrYnjXzLF6a@*}RFk4DP)qe&B?>p`Nj^9qspGk@CZ#uMNUfv$1J;?OFnn=oC2fQE&#^mg2*(Lg`^GH zJi0nb=7L|dKYu>hEXZZSuZbs3{q=5(?aXZB?E8rub35yFrP~+4pr;TAJitfR$Gzfn zs0hdEyBrh~M0wHkYxcJMm)Cwjn}V2wTo;?DCYQ(%k0=K_)Hg?t3RU3H8@;hM**&;m zgG;zTA+3>lU&M;VEwUoh|Lv z4CGO~Ary+=aehDQJvpgVS{MvSsyhZWY$bq zQuyugj4QQ`mZyU|_x2{!@DBE1VADkH3rcwA+1;$yL(Ka!?tDw0pdvS<+l?fYr@Pyb zT1p&?D8@^VGY=~=@}Trsp!@X&FnzXu$_z1#F0kOTQ}*dC!!>C-J&$hPajEsI$ll1O zNLbQlu9JNkGN#T7M@HE)Uq+UXAFu{)bJd4bzKKPcTTX zjziA%9uMw#QQTpfxBdiQq8J zS$F0<8)#&WJ7ci%wkjJBruWqG=uE;s5v)Dv2g>qd%KBCjR5~>&6Vuv2cbVAyJ37K> zrOmr!>w$r~!5d2tNb9i_DC9(CW%?nae=&YblK8ErhKY%S9rKAq3RFD#a4IV}%iX&D zRtM-cSDb-Gu)zL99i~2WQ`=65Hj5O z+N0joUPOg#N6VFB4=&&f1W(u1*&Jcl`7sH8L{AH(*Fau*ae*lp<)UaF@z{J3l& zOMFY)@W{bArO4*F7oN;A_5VA2kN)qz(&zN}wi@u@orYJ1Zb8)4sd z+jp^Z*);xkn4l8RTllT_Jv~lWB(Df9c=3h$&@Y0hXD8GhiL-)MA{{J`?IL8lS?iYR z-ERfU^CutiXZ~7iyoT%9+f1#ltlBS}M1|){RxxLi0%h*-bvN<8#8^j9LCoHgxF=Om(3@|Ygc{)e zu_K&%hk;mbD+B9qqR{Cp$14RQQ;2`jSP#7+I?zLj2Zw{!gm$xusQ83%$h=p=cFbnX5gBM}|h|k>(rmhDNqhq%|Y<98xziY$)xtdVdbv9JcoBX^$yF4^{ULbNd$xt<`M&m>FHl4FPba5BY zH@l63?%V8qatme!>$W>#qT9c8OPf4ahZ6i;F<;$hAUrCx49jB9$xz_z7@wr@T0hO;ohss$>B znWCD5*T0++zu&vkh{k2(HB%}dUP8-`x(DiJrER|;&0jE7{dMPj@*{^v4s7ZbirO`V zJTps2LSFd{I)1iMTL_vIm8739*}ADZaU`oNvUWBO7H<=ezbla6Yt6KKf8Um?xUBVW z7~3xqFtD*8?-l!ihg^d^Sw9$vllKe5*%6}vI zZq;0an%IfpE0LYIX<%+|NGLcLuC(V6m4+O91Fw|L_3schXh8nSuVcBKP?=`KQS^&j z@H)A`!#O0>qpH>2ly3C;%Gi?KT_1!fSkNA6!0#i0YHw%jA=HG`bgrSVmzhc_m=bA1B&=57x2E1wlkDs2kQ;oQPO zyxb^H@aMSWzx}&z!|0px37@H{T>Fu6Ju+Cf7m_L8M{8(XgyVy@X_vULL^*Z9ieU2+&tDBNjbOTL&-B^NRtMIxC(O<`RV#MuWbehr-Q&`@bS5n(H z=>WIw-3G@CKmuutvp**&R?tG{6i!>dNe}3zk|FuDh?4M2fM;VlB3@Rg%6&SC`uPsL zn|%S0ZXV8(STmYZG=3pd1TSYIe!Rh8e|FBRBvEU-n>Q3|2BBFuz>-b3_?l5m6(}%t ziK-NXra_2F>0@a2q3ePJA5KPI9&t|;A#0zCRIN6Hio6CH7?`;7qpsiT{?$`;&75&L zjgEL3j-_f|5o`iHASwvo_-(D|#hS->E~^fs_bfj*O;aY90U#$(k#p&wj*3-?G@{lm zTPe6Zi)X><A-(pn zHT+IBIt1Pad!Wd20TeqDUI1yYbD8woCOdnQ7=xM}=06Bi&DHBkl-U}|@WyX14jw;l zQ0{1PcK0L#9SuLBS~fm9Y2&3Fr6%ijSu9ubKds~rF5lw|<)d7wxJHQ8?jl(efGDr= zG}1@mSyoM{^-GYgPV9ZZib%X8-0kQB2z=>0>dhK?0kkzJ`xHf%3-|u~auRt?vaWB5 zm0Qs3aBv`yyzEAG{!uU221hwqzC~N2WDL1|0f=FvK`}*RI&!3`#J2&GOVPP{MOZHw z$GM;X)-qpMicyY_&}a_A%vjF0VGQCin*%Cxbz=qwx84i1o@h}me6Gd@ldL9ha24kp z_%WvpH5F&&YnoiK-{#$B+uu%w(&DzPY5fxL9ILsKbtE1xWfREJY5B@ek_VhkofY(1 z7*Q*K@XYV#Sb1)7?=ibkCJ(J;DBrqY5a|#~qzj=)??gaAYNRVl?=|!eQ92^M zhu(V$HIU-n&YZb(&OI}8XXX!{gbfKiJA1$HTEDhVr(LymR2AKUq+nFa$-@nICT6X2 zlde*-#&Gp=h|%Xoty9*Y-XL_8kN4$a@q(g)9?~t)g3LhXvU?WubgOE;Cx$>n@!fpe ztDvNwkyc^@3Dm#)eG=OE1YKy&S@7`&i*V8h?F!#t?+9c2Y95CMs=aD4xmY)68AnSS zclNHYZdtJH4&R^+byb5x?(|0wEp`z>MUU!~!(j#1ouqJeC`jZiYFGujm0Oe-(mMB6 zYPNCSopWrZkOZ3BaJl;6e3(Qhr`Xyzt*Hi}t9j1%mLpJzB$T3hsxhS56#fKtUX3+G z?*fAc5TQ5m!t~Sb6Og)!WBB8m?s1MNoAJM#9oY)*PLio}i%`d>;WPQ*PT2=&>2#-u z!tDWuKlxmG`uRC<^(Dma4Fu1W{&qv`!gIkL;JP~DteR?WC;sApuuzV+x>`c~<&jV# z%Xp}f`#g;CiB;z+6M|tfu5ZoJ4;aO^=T=;#^}ZDX!FtQ ziKT&egU|0(bDrbIGOm@#-lF4D$>Q{imTh*j`c1Y+i(@FJG{wgR`vClQLz{z<`#Ry zkK988ZrG^EniE-NfG=NBj{W|{XU9j<^Y6t$p`*2?sH+fFPp(%299OI|{0yHzGiYqA zcPZkASJmzwW;b=)X-)>eQJdbHq3GIaw(Y#4Lh)`NU_T^X9e}W4QU($fjl@6btLzoX zlVgqLR}wZLQUdMp!K}eUKAmlwdn|4lH>gRV2C8?-tdPf?1V?+5TcLMHa`4ZPtlz#U zzH?)1qh9B$?zf(R^*L>D6Gx8vTYU0>MCRoKSU}RXT}n24x>**zBlBrol0}RxW!3oU z>0!~&_m?BwwC1OEz!F@~!^>x9!`}UiwvIwaBSt(%9g_iV9j@Kdv!MBgajpY{t@XDv z2K;~TIr;Adxc7`erX-$M5am+emG~dOcIfe@+EFZz9e@2HPf_mjRstk34~a7jWhXyB z;!zR4_N~sF!2#Lj4AiQ%5h-vse&z6*1(dFiw};@|$%t+AEc%+8Z)+EjFm$K0y4LuL zr|-~jsw8BTPJScYY6UCL`8ue)t^1kF2HAT>H5zc=M{RiS_t9*bo~cY<^NJJiA$tPT!Bva5jDc~F=m(SA)9Wp#8A2w%so1$swUx6; z(z{E6H#g;$+VT1&sKISY?2xLSJV-@?v$yc837;h|dMSHRn8(=fZUoD~coQ6*sgCzsI>zq<8LbJ zU7zfR{?|BzXWo{K!{vSoX0KP9t055As*s$(L%{Y~a^i8H~P0is%hZV|s_rfPUMKMuY^5`eg zIVV}$BtS=$_4FUM0Qr~8&dImdShZFcqPGB&RJX~EdaBk(w_sf};BL2=DK}Tu6RVz~ zim68m`qR_$^e(TK(4_N@hKM}39r(~#>p61Ny;&}>8z2dC0!bYnD3i?oRDk3 zr{pICZpN$_T9x1qu&Bo+!)DJKc{Nw1&&z6npwlXkp8I@wND%j}?lN z<+YJwP7%W{6rGjBq6rm(eFkn0!xe{b|F^qYG>*xI?!+ao_ zAf5!;r+pbUl?1h!1JhaAEmwR?chwcpvn0D0<|6``d;MKe=Vb;U0w{g$Y*n}nZ%QLy zNuJQU{WUM)fx(NOUAZvgT@4f`{F57fbKje<^!M2oW6^8hwB==;<*9=r`egI$4K9nG z3dCEY%hpCpBBzY?0YS<4H*ZuG6jY<0vS(GPhG``*#T&iNE430WasEudT0g8kG6#0x zwkx_|uhsp~$th9xgd$gxuBOMS7Sdv4tA1?s?2v27roe#xD|J88&rAM&L^oEJ;v%B+ zGHfQTW-xCWd}2l2FZ*cg7~TGDWtES*l?7II0~+<4O1Lpwx{!^Z^GQnAL)C42{b{?* zop&>dylbOTw@&2OTP`CuF2EC1n}N7XLq+s@34$TA;$AuutQ(UoyPT_381X`getYFq zLv^)G?6i|kz6#>N&yJHuI0a44svXr#^C`5d8~moq$A0Gz3fx{+pPZT&o$48QHu_(l2y47&S6lJmUD29VSNA%daic#xu!OX1nu2pC{Y>lE778n#?4$%cSO0@ zK&(RerqiFtvg{3nD0M8yd`tWjC11hy2jJ~)7>IgpNJaOXs%{M663i#2U`9e}v}aBt z^E2{BT~w#CSd7}`*krwqpm*yjWqw|<%#**Vy7s(P@^r$*d98vb^jn?T3B5z?X<0>< z3pIn;)OTU#Y&2Flt4NizKVLX_9zfiEoLQreSTZ|RUHhSXlmgyrmv|9~+#fs&|BFFY za~oq+fk-T!igAV{!iQFW^rV-{gy|E7X6Vdz z`0Q)u@nCXOpC?|nN`#n3h%0tZ0sjL$>VLOD{(GSJ|31zBRh-Uj@KWr^<31(}llN5H zfEOf8`c8LFc^s=mxhTqc<(4ECCQF8`>1yV)T5WEBUpg#_Udk$8Kjlov+qEcpb2Y@p zsLda|+jCYxA4*SAuNiu0ubv)Rzfx?8e{_q()8h2<nfu(MjIhI`-jL=FEJdnIH5AHv~8K#+FG^vr^p8 z|Ke*C*rAqoN)Lc;VxFBQd#nSNe$Gp+aGnz!A*5oZS4-RIPu%K%xsU!|1@PbY)^}Js z$@}C&Q)i{e`<}MgCppVH?T6r%3Revc4Zm;hy8Qtyy)>Gd*|n4ccj{&gQ@HS=CG&vs zfD5^+WS(29G3bP>>_G-r!WQTyl-Xn@#)$nvmfdt(1c&1cdGHk6R0|DC5~K`O8gNiv zO*dg{T_k_{R}q^J!Ql>lUq3X&P~>Y8jEq7u4@siE^9J>ROKL{~2Pl(ZU;mNn{_l>? zDBwr6;e}X84hhFCG&g0Z7#zGSSM8Uc zyHS&~R-sput@gcL+Y&|rAp3#dT$N*mjm@4C@w&KOHq2_c&xd(tdh4rJ^ya%Hugo8S zeO8w26WfNT|GJdXXwj$MxoTvbXR>0qjoH{E{t8F5)gFIbB8u4JW$!di!2RQnxEyA> zaOEO)flnGJ<91WlcO>T;EkUI7P9z|mFg?S~Q>ia)JR8SlEypU;RW*!;4piK4zv9}h zITfuZrLnNN`}#H72RmBW<0!W)8j!Z-wk6xLK+)6pT9Ix}Azg5CQZ-NwGcIIs)65?# z6tIf2IRP#N4by+R!!)$TUh6gu>UZA~0{G$g;6nR^_LfjSn|{l8Uvr^KYyS zxc~)%*m$#oY7Y}`Tm>{l3tP}*6>ARdN-SNEBF8=IjrnW|Iso99dtWnjEU%qtiIy}; zd=lgwk+dX8=YUdW@MhC zHE=3kH-+SjrrKPTQMiGvtxxDlhDVm(dFns3hy2xQ!6GRdoj7BbPiat2eV%b7H~1ti z&gC+-gT2PR;jM+aM|Xhef9I%@sTt9Cd!9*tm}hARkARKl3=EIqBWgcvTl3EZr72$G}47}7Rk z*J(pRy!ST0M!EIs@qzAFvgyZU6=k?G)07Xf4qSA>Qq|zV;lTyL5|jaI*5{&H%aHfW z&@7yw!n4yTE6s{3!%uS1S&poD0Nk%l3^bOv@*$D!nRDK#fa%ohc(P@Y$16^|oQhDwX8X z%nx{8-k}BKD@cVxdLW!yOG+6g%XlD?`#gr>em<8&jkrWH;6Z$L;qO;9#@>xavE7xx zV}Z=P=h1?sVwpx5amhh`A!OgiWMTenXWHr#whpqe@SExug2G#C`DSBygsv}GBh1yJ ze9&M(L|`*7J=f$t-!dS!3}J_iWbUO8WqyE81^#>(w9C0nwyO7`!H<6YrgZO1p1#nc z)B3iN1~Z}Lf`{O~&=Lf*_E z_XK{h(VFJFU=8>9I=0e!qF49p}99 z1-kL=a`LZ)2Jm+~8uVd8MBdDHJEepVN~C5N6noA%49Y zr5^~Uq7TJZ@lE+~Aan^^*1Zoz_sCVJR7oPwKsuM!;94CIxH-E3oD20f~K=O%ew z)|s(6IQE9hfoA;SI`RTn!z}P|egi(!{~&zhZ^!;u0i5~M(-W2>4hB=dihrpHp2*Bo1HeC zRAdl4)Y|CoS*(ZMqm=WzRkgYL+^Zq7rFAj}ZgO$%y!2#hUj;0&h~}k|y+eNUhsOvH zKF1F9RD<3`)o2#l-|Nc7_o1F1*KHIcK$=10-awNk=<~BI^I6pv@3!;8zJhx8p+3_m zJzzHHs86HsAJfkM+&`ekf1k8!w!waTu>7&>bkDiS*kE^;Q_#8_AP|+ly`JFvwxGo9 z%&hhDbE%1{IsMOJHmt0QQ%5geW|-8hi+KbjhHTf>2A-Mt^+^Erq$9pK zASyOQ0zgKS`yBT45}1hR2IzAb>?Xnf)as>l$zuvqA4}pLN3A|cKA!zeh1X~&Ae^L^ zv;R76lRwPf+Q0CdN?`PGpVq1&q6f;jD*R0#-x@mYjqSlqtTSoLGabk6qejIwv5^=c z2`#Q5U1(|?TVf6-@>@*0U1<>|6{h{7cFR;hy-owxZ*8B?JVw2e${4{*MTOwO<%IZW zP?3ayyola9BNFIhr4z6$KPhZvI%2UMt!a;&`MhwD7l?d$l=BU7NVbOPI!kGVEI9!q z-uTxvzATo1I{M;+Ag%)46O5rBcD?IrFqch#q%Sp|aYm)$N#3q2L=9$ipek$i5V&T~ z@`bwDZe>(ehnoh<@~$@f(YVE%*E3qhh=Q5YSHhf+xt$@%X|Y~oC{sb?$-x)GaJ*xYydyv#?ID>d0M!) zJMY_``gXpTP~9faq3K%SJouhCORxen@zCf0qwU~7JI33vby$>Wbu1R0aV=Zh(HFQ< zrH6mH#m`?%Ox>xfY+c%2Fc#|wB4<(jlQZ2vi~FzAN(;i)|K(w+n4Ffc@%jA z>yg0g((Px!y3h?oQj(m2|C|9|bH8!V)n9U~?&2;aPmqzGhN?kiNJy!)N1^U;bfUb= zAIQM`pDHI+{iB8jwns8Fbf|Iz^oh}pRTNGNn_MCLvN)UFvhI#(RGrw=0GCB%ES;Q^ z(D$>|;hyvS6}fLkV)VYqUj!r_TGELVKry;dA%)V#E8`~S##u;l#plp1uBXWZwch9E zI&RuRvO3rK(`})%ub{xhn4fXNm3Q}yV&h%Os2@VkccpGSgd`k?V#)Y3GdhJ(D2&w?vwvoG<*1!;@C6_9zl|aGKvc86;?OA=mjjvtRIY#Ybp!8qg z?4GJvk>fDCkbuk4<2M-%krQsxr2OP(odx%5>g#Ok63a+Dj{$49Q)sZbcL~}@*9_ZjYo)DP%#a$p%+6_O%kQ1Wo4Zx_aP6`y^gK@&jk%_?Iqx#`i>X=ta0o zS^)cdP*@NMJ&I_O#NVT$*eNl26<>06SAP(Pj^S zgfwRHrwlqFR<;cJ=3&7;-1rY|KQDUfS*i?|mG!LjnsYAy=^eJ4mk#rhNP*3t&L;FY zNmV?oIf-=!gp;1}sP|493ItF_%UaM7Q9B%%e~}&Hr7Tog5nEuhD4{joB{4BpHO;-| zy9IA#BHYhk_Z4uiL6z~NrC*=PGCEkrqNRs7r*G{V`p>=fTXVS-Z9-$u-OJ~#b8vbU zWVNyKV@{f}SFrh%*irHF>ni#$z$)=$hkoO$MZN6cZCj-&i(k+ew^qoT+SzZDFL()M zhbM67fQmGf%OyXHjfe_| z5WaxQsqq}3w0!@ctr`E(@&B36GH`u+x@~-0AWOmMS)A+Rv%VPF{Sscb`mjdQQf7nYxQl%XaX5i`{*iSh!nt=o$#Ev*9p(U> zdkooVWZsQM2il-q&W}b2@i)`LiVqmGGggnD+>OHNJS~wE=gt>D2s&&Pb3u$@Mx!LW<;!h?rG)?N1}Wsp~tC36`RgnL$~0JjO{HX(${(}vN(<7Pa9d;zu(jdh5c zL{9V7Hj5m;xV3dl9!&X(RW}fboV$62rp)SJogG=xS2(pA^g_`9^svZ6g6_#sWm9=5vkTRA5&J^QK9UNANlXMXp3*9W}VQr>I z=>hj3$VPVOY?ud*p4g#qte#LuLoQ$Wi)U(elGRU{eZ0N122|3!pZDnPe)}&|1*vL? zK%?`Akqk}QCG4C9fm(pwQ{raJ*@tq^wsaDLK9nmf-9ug)0R9~RofFX751@+Ij@Pm& z@-|DH%7!DW8shB)?cq7)D0YhVa zYyjfi#kS;!^2a5N5JT;oCTUG!nB$<0LySggW5tk-)^g9h_?&G-5a(~I0`Gh@9@_ks zU#DRtL9J(ZPa9nKiRjx$ME|T`L|-Fts{^v{5eIOaxcr_GsUbg6BUeUoy)LPx__JiU z;~e6yN(pLozHwnD2P4r(xsTJg3DSK(p*Yy{kxAx|O@6~*DVCg+PL=R;neSsSUT&`B z@SrG1^8VAZ9D7LvXbxTh7$Je=)GWc z@#TF_jUOYc45^7Jho{Lu%W}0uh8LaZ9zyVr0J7yxH-%)Sd`+1+!3F$mkKAh$cRdB| z@%@xk#Di2vLg8c7Wo5PxvyY>^71jWd#=_E&ZPd%iZ4_GXY(ng+3zR^en^}dvJSgOD znr7d1;&XMEggIUG>3{bW*!7G&eX?q9t!rr$j6EB>&cq@$i7!L5Whn?LuL!Yb1P#Vn zDjuHeDWTqh1~=FrP{8fUR)+=>uZsJv0~%uJCYsW@*~p#S-rWwYRV{UItK>8FZcLPy zl5z>O*#|{W7Su;c>h_EANh^{kLIT%(fI@3jqHS)~vpik|i`kI*(Z0f1;&42j7&o&E zlOL9JPFu^2yeH;>o_lYKnlXSr7`^L_7gYZoX<>Jv)P;Y)YGv>s*9j9?rGDmKu0!Pp zu==E(B;%#1Hw+?@XUQOyb6x)?VatGgv-Y92pFO#$avAG3K&Q1qE-Fom6rEgxz zFU>uD@4|HQ`9{1r;B#nTnr|D}G@c*%D~K!-CbnOI#GZHa+2Iwh5oGXNfKXOwsp^e`c@7KJyA@n%#ZgD6_}y>eZcSbmFRDc zHeV6r!dX3P-QuOD=oEya1L0|=KU|SK2gW~9krXFS>!#$^^T6GV=o8N?>|aw|7kz}v ztFJ;Go2iqnP{NWH9)N^w!0lX%;;{{9G zeL`Ikzn9ctpSO18Pv$tq2_7%0l9A_B3BGB%XNx;C z-T$7coe%Yrb|e!77sB9iCIr1%-+rS=7{j}XowYncPpD;+9#1w0$t9jd#tEGy^aUwa zhQ?i#q4Y28PWFN6;#Pgf>h2;zWK-s*9Po45|S;d4gIJ)|=m~8-wer9L0yi#H6%a zu?$Y2%ViVpfyCE_G!=Cq0Q;|;VqWV!!Z72hTd)j-TDy{F&I!GrE`^zJlbzh+dD zhI0X|!@st%u5@2lFl($zLdWOy*=F`O8McU=AW6;9fSPtU2MRcYJDOk3h@cNmdglgl zb(6cFr8sI9&8%X@`q{bXP=2)|d1?!~_F3>G7rI(=@b&=$SObreX$XSPcMI+~)!o%q zXx3NU2hy&zwhjPGGFJB8yii5{8h5rS^0I2g4*){ClkyMb_y1u0@qgQ%|LpM%DZa$f zddu|HGwBH^=~Yx$y=1X8IUbMUACJ;2mSi;6v7-`d8vaFNR zXM%lf^&DA3yPmp|&`}l=rxrm=60E5iw=Wd03eBmKA-0_+g`Os7dblKJ=~{}W;hJ!dIoLiqA9LqQ zFBcy{!Df|D%$*G zSfqWgHhg8v?$zKKqS*l*9qDfA%yGwZt%l_Q+SHAuz)b(u$p&!xqYi)}nWC9^fwpVv zu*#@8qGU1$fHjf%q8sqAkIvn*D*aET_kIw|Rkx>;)%BgrUP(zM&cvU5v(0OWD>!6> zwRos}x>Vl~ZVKM(R9PVf!P;uB>;#8xEm^|t%_aGI9AKmisJ8ln4ZS|6CZB$n(S7uZ ziN(*AX_@w=gtj_DaPHRI@_bJlVa_&VgFU+KEy~%v2(VZ?QZAXxvkTUXsPXD_tZLa> z12h$!L*_$XM!Bh$wh2A#D;(Fb@!dLA563R{WY6cbE#y23zPZN@`948#ZENH+i?nuD z8n)F^>^e`n(SaDL+VZ(Xs9a#TEUrA0G#;5amk0T}01SxpgA4hYF-Gg}b zUZw>Fr;<1!wz%fz(J-c&mdc3swC&fhSrw$D*ZEQVp2BPk3N=6-j(8vc05joo-_iuE zq&15J1rs^|T==RQaqjM0On&hvCp2aEJ`I1x2-?*tJ$R6w)x1NHSV zYV&u;#fRyS$-UcUh~RSG+39G1*b2o$_BYifh2K;uzJP)TN{;0w-GLrn19l>pg4i1AY@?G4}${ zI7BFZ?j#O##2+q-6axt1DY3_88YdTQR2f7k2%sxDr0PNmZ-#G><^^&rAaNQ4=H)!( z!N}7zSGy@C)qk5wni>6-Yn+LErc2j+{!vUjK#1Aw^yn4_E{Id# z8b%o+?EAZ4R<)*!iTb zmy~V-(l%5~8@;g(MHG63+Z%H4-eQd)c0b@^mX8ZfQ>0KIP@EI38DCyMszxLL8lU!C zxz&wt*}|BwO+AhN!O=+{mY>?ek%3_tsOUyCx%OTh(WD_x==%NuRF?~fnAd$?>e)!V z@3|A#+h!?_`=IQcQ*9Myuu4_!LA>#ts*+096oaZko9}Cb@d<=pMZ+Xr?-gf7hKISU zDVF8iXvvXA#_`EVN*VfJwIxr4m!fy2&&w z(`vEdXmFd-J_Mm34p~8BrLM z1vrv=R@aYe zmnA#24`P#hgem=nT1xEgJQ0J9)V`6SdUl@u8QVtIqp;ZdUZzU+!oM7bLt($*JZU*C z{|ZBW;VNkgSaN>BrW-+(Lfx;Uog~y}T?$qWkpnl=no{(&q3(F;e&GiN_kh~gL+cXs zPQr3-SSoBjiqzrEQDxz&>#_1t1aPs8e+^xG2b&CbcL$|_M>E|kR$7!DRydgahEZ$Z z+@W91OE{~B>*d|{GF+Jl!!G*h*`;u&FgB}0{EsbVE1x*$?i;Gw2;h94Ol;G)B=HVQ z!i2P^Xg-N6IYR}9(RfyW1y-}r(9jL}#u)FhE7l)9OBSmJeGiM0+UMHJ?EM}3;Fx#a z4cI&F8wPjL%?1KjAxkaCub%ks-6v5)3mw~%M9XVxY&OU~LD}jOtuNzCnp@S;aR6d; z(DQP?Jz==T_@;T_&ugN2dhKDLSKk(PTc26Xp{^lIPqxmF`IJow-8f!}CYU#vx62G1 zOx~2Z#9>4-a+o>pUoPiMH7m>6F|;{BAr{@)JG*qtf(CqU>P&1!t0fjCQ{1zE`U8k| z9iN^AlI4Xc;M6f8n7qEAEbGDW?jR@sxu{5(hRbE1yfQ1%;;5^>pIr>C^0hPNB!^AY z>c=yTm(pbO`0gD3NS^u*tAjju&JV-idg|#T@7{hsFPNk zWJK(+XOnoZ7(DSTYbecepPYgXloyKRq&9a2gAW)}^~)W?0BM(61lu zrLt@6JX#23kU5g#QljU3{Wx!J7@^Z9UZRfRnSlkAX5#0Xg~=@*3^aZL>Tg1HPYtdWWf&q-qsEE_nAEpCN8%UAO zeR@8!=b01M#YH?cf~3y(KpWRE&V5k=@&+5yMul&7Vr;$#`1@~&mXV^+DHOr)3EA_|Kz0^9Ygt3o_!N6Q z)B8KU)vGffdmG$lf+^Ptq0QE#T2I%tKDz#86BA6&_R|@xn%snmzYs$3?(_jKom@V= z)0lG#4>`d$gIP$=Qmre8u=AF0>y6D@@}!cnq)Xr&011%hCTJrxG=`W4mat0!Guu7h zXQ1~OA7g}^7T*0$6$rFoR#vZDd2U?0 zKLZq2?;}w3tr|x6>tqdr%d5nY0eEZ+nY-!bQE_a;uzhQ-Sl5k1$BDJkwUPSKR-oX4 zNa_75?^7+`kojc6eY`E*K)>60oClm?^?bK9*~VeX2{uT?0w;R9o3bj6orVS?hce%X zKPKv{mu?#I7Qx~!K8OOiAiIxGTjJx{VtsT9hbSRX^=;&YBxo0-w3`!gv~PBlC%&)i zqktw-9sj0!yMItrC(|d=OKaI>l8L)p&lw$|+ndjxK}NnTZCO}S^?affclUV|yh{sc zklovW#_jc)|IXKI!)e@FrMlsC?=7ajT=%=5<}_%Rvx}}QFvaFP$X_#}2w5sC~Z(l`&vD zJk0uP#=msfqIJ|)w01YvFAwd2^dpqbb&EING=$k$($t@-JvefB;rr^Y&btgJmLkVi zTr{B)8h9$(d2d3;skUBa;DRO>*cSd?VvBiR{)41!PB>I)^mfuFc`T&f3^46aUfgMh z3J2kTKzE; z?`qnpq~2eid|rRbrh0gLZlDm&HTn4Z!y`2}5@1genUeWo$PiW-)1-?JjD6k7iTYmR zd%(iDkH;IPp!}|$Zb27bcg{;G_tNuv@`5&GCD9MjO)2#gc6;(n5>hSm2hru8f*b0q zdYIrN%Sg-YUcvez`V;dKfkrOmqa~a9$|~wqjR!Ni?Q4cYw;JyZJ-n^7RyNoXj6G7- zp5F}$ccx+{g0;=;PPT65K#xawljK9=@%BRK>;tqQPC2nMbw2UsVx)J$lXE2ghvBff z??pZ~o-h6HZj;)mV*{=v_Xa={pLEUkh81<}+wd2j*^z2WCZDq}N>(OZ5I3_?*AJI> zs`yxmnpF<-(d4_`HNW$*Xqkk~J_(JoOZ_qF*22>o8?p^Jed1~A@RuJ_&%^U4bP1R? zhmTt+4cOQ3&I*jZrhLR^dTyh4zc#VT@cpK;_&Tml*Z~71Rl6HvpiUtz;!+$>y3T@X zzzY^C&&%8C4=#ep10h&f>`XpMH81mC!E{g{0W~W-Ibl9}$XKP4H9fk5YDJ#|(kSI+ zc9{dcBpJpmU%Ph!9MLkn=@($=F7&FlqO679O~a~IehJpLZziPxeNdzHheNpWK(&`% zR>YFa`eDcVCC*9KeKdpDjXvkZV^&ql&pOF@q0c(RY&!z2bYKH-$VjY5Oycb3tF+9T zy{ya>;C|}Ksp9S}e`C+KR7QW3Xpv325Clv*nX`+qFU_n9E({TaRq6>mJ2u@q)F<0DesylLWo9`y=sba{O|^flla-Oy=FvJ|3F%D_uI-}^ZX*RLWDCulv)+7DsVNg{~#QJS~a?Uod z)v{)h$OdM+cwORQAyF@9JB>CdX=9Rgl&@Qh_W&sX4|SdyJbYd#6FnV~lU=aN>QiU8 zG%XW7zUt z#tJ<5N2mKi^MWHpD;d6uG3ft_)3M{`PCz6EY!o!&PA z7L8!Q0(BWV`Q2Yo(n?Cn_vK4rt;kH?L!PdLZwWr~F~HSyx#EiVfTgR@RQB7&6Y%~j z>gCFtA(GM@pYNrp^u?k-vmKla)U}N-6_*UKHFW!+wLyFl90>CC zII9gC?DCM@0zL~=xc#UhGyX!_Qn_dL0<_NHH`OUEk{xNIM*ro3El2xtsu8J`_jA4J zk$(Q^w!@smFGE2cIA1r0fcm~xM6#d1vW$y2RIm-UeeWY5eL6(9%FZ>SuP_oHO%JS& z0xam^ZWG)FZt4AmrNocqla07bgI!*@c<2Y$1TVp0Z#YtW8^ z_$!*zg&OgP(`~UcK{f}mJ(vQV9+tEl^hoKCaq|ozVR^e~UO2W)_q_KxCBOEWbG`>F z8wLXLyS)aDX2=q@F!|ibDp9Ezlp0^ZBy`q*^QqYv6EQK9j6o_FAnx}WiM)}PWSCiQ zUEi$*=K0RobzDY*cI`igT_H(hhkqKZx9RB#q&;eh@&PwV^#bOn2IID-(O2b4dOX)z z=+kbKy?YyU(;DC+cNF0~dnG#K^K1p_jHPfcpxCsw4`9%gNPKz4D|}*x2qbBRd&xeK z6kq>if;YM^Am#u2dvYh6Pc;OP z(*re{Hiy`#zg~^-aD03ScI{z~h9_-0qtaW3(?EKu#+suLbRl9t(GX()URl znHjeq{-8p8%mFWDgC2YAp%0f#5QQ`IA?8#g0zqm)>*c?xw(^s=WlzBPLVu*~)Ehvx z_FeYTrWSU;2})cBIJSW&^7MdT@n3Ww=WZN&R^AfE{5jRvlF-sVCC4pY6=`oL-w;WDrmS;=32?#uZ!fA#)>c7=ae z@fgHPsBC9+M|@rL8|5DGnYw)WCnk(R?a>9bUr*0P!K|p@e9o}>O4PB`YYRR$?&vSrm5HI~PbbXBtzNN6bhZMO6c5Z z1VsC~Xt>)iRLD{jnB?V;P)p-_dmW;w`dyRdY@!lTvz_ox2>|Ddu`Rt{-w^IE{PlE( zEk=4k-{KDo;v@Ken$w6lrKAKibH52_=NS*!7{7MggQ_OAi+yK9cR@N(Pu&vm3r8}q z8->a+n%M0#B~KLHoFP8MIHhGSe@9WG1ZeBuRC@DbcE=CC1IT~(TR7=L z->2mb&7pfoXk*kvI4FE_5A^)#xL>xiyC9%5_BGc(^dZ*6r z#G|T7G1;e^;BTm0M7lb%S*4gh^&_WEoUUp+gRC-m5^_o#vXL5rkymJtcqFbCRwb>$ zWGK0t+ay~!#&I0_i;iNlB!b5aw)OdJZp7b?=w+opEkdq`@L`B5!i2SZfGXgg#U?9G z{%3d*VZQ}YqttD+DYp9*OuhRoWVqi4|BKRxe;%zr}GG7rv-X7(= z+b$QcYGGMld8zj}%;@8D)KxA1svn@MGV{T9>^~5Sd+MEK$Y7Omnd>cn?)r4T(K8_j zV?Zwu#S>`~&cK@%1`JI)`(CoQHF?iK64<>AkHM#ZWswlz@ngz0JiY&sftUuOT>7!B zwS#vfe8-j~k%>T7KAGw+BFnYv+AQz0SksIyQBz~ySxu*A^cYZelNnIfiUR(gvGcWmf){4I%6NBq`pc&~ni7G`7H1H2SKf+EsrF_4ND4 zuyEFIbR_ZWk_g$T4Gp)Hig1-z8P$@s3+E0QC*M}uWUoGPjwWw-zMY3^zJ_KxdT=-O zUUEaJqg6`>|EdWkE5RQ91n5>Lq1)abqh`R|CCU4ac3js@Gd`*ozk4S~qbN)D2Gzut zf$A0iO|?Trh0Dz8=@jfZe>OVLp&xKVd=^A$_>ZQV*YG3fssg=a{v?-aIdnPlyk1oK zN!v#?3x3L1n3hK$!#2N3`Y_=6%@p+)iH=#BM(4DmCHtg8A6QNN=ZKBqQs_>u-LkUN zfoAQYS$*=ffKE~114Gd6FX%c+DpL?Uw$Qs0g0;(BJF*ZRVJfBrG)1YJl;?&0x;5gq zKP!M6G~>U7f%K;fXF-=YYz>cPD}~uhUtE0!%jJlBfK#qTjuu*2}t-O+K;1W8Bg(L>N73#*eu;$mGXws<~%YsDoR;PKZ@j z#eSDEMdgn?xJh2JC)0Q?G5n^g5CZtM^MXd_R)Q(#2}weqF(NhIABsThvC*!({%vt7 zv4fo7Xm=;?kpH^an3z^1UU|pPVvoN0`^oDG!f`YBp|VRWGa>$8786Kz7ReMJ;HwH5 zV($1q?7eqblk2uG8bm>*35s;0R0RQPg0!Fr2neA_hoB-|KzgVFk={W;X+e=DgoyO6 z^dh}UZ_-;r4I#-qpL4D`*IfJAXRWpOU3;H%@7;g!1RmZm-&@B!%5VGzBy^dcFhChOY8%*z?qm#EqTG9rFFYQDn(iLlS$n1@ZtNCTZkU>@TJ4 z*O%pIhSdsOpsH>xd!$tUzOMHG51Hv3lUyfp;F{+%oSz!HR;0L2IaWhz`N!MAnS#j; z=CI_7nq6VZZ0>&B+;%Pv)=svHF5+-(H*`+&n3giNuPV6>iWNW*ytN1GHnSy^)gD{^QStP{Uq2&PshA+X6nz z>Qi=Okm<+$hmTD?eO?+`f<3j?5jk4M=>j|7pU6)XY5*9b;H4wla#{-?*7rbU61Mj3 zj)#qdFH?$#^QAT7_=TH9hm3k0;+=V-Wu&Qr{Cog}JK@++PuWU0L*68B$Z6RLpX2%_(vQH~x=M6NC(CY?kr)WA3XAnHxy)337 zNdz~nIcLEjLTE1CPlW(gLoy>fmoCXRHWNZNx!{e2Dswi&D0u zM$>SNkSrfQigjxKQbQ9#?Woxtw@~Fcn+F}NVFpBRi;GG^)@uGu%6aJfTD!qKO_Oun z%9kpDOSFvnuQ;2BHNyat*Tov>cHG9A(HL6&CHs0;M}BMP1n=kPVm=1`y{OYaHEN%k zX^2dB=JqmPGktDfxzKY${1%{2$8Do6dK)~w8wA<-p(3G3zz^?gd*1&@NQGViC&5B$o3bciQx@GLWBW{W(NS1>?z}`uxKHzN#3Tg?#4J_gC1*{l7-r!R0OqYJGj+$x0hsF#&)eu#=LA|W8M1keqyp;w z{Zs$j@ZWzW6ZD_H`@gLK?oE$hKooF9O9!GH;6Qy9eXjpmJmO0jBxs=y*@8IbG)D&s zLx*(HDwp>%0{lGe;C9CCVMBdtD|-mlP~}}-EwO3v_3;e_QJk;;;q#;9)9h(>W1^6P z^KgjH@mL<*c`+xXw?j_JdeoJZrnO3h zBpw`FF3cIJSmSUmj_X_mJg%R(1Z1gA9~24mt7Dh^#|}u=W=Ua9+diIYAJ_1$Sv5oG zD>Z>YzOgVfo#m>3j2Zg1Bn>$bh@wM>4DpU6(_Y>l?~JLBqP@NE+*Lb!p5SUV?y%tN z*qk>hl!7eeKd|WfT$tQ7wQnKD?jSf}!a1$1J9ywPYg%erFSg5;&xTugHixGl23s0B z^w{NDJ1NW8dC|X0aFz9mY_1uyE|(n=4vufm$B!u~dwYlKDY2r3cn5&+?p&HKTEO?& zqsjHXd#AeOS`StoRD^U#wTiXzH(O==WhUkl^&G$pFoEb)?zjWp;LT>nrE50QLfrNx z76^c3THIlN-8oxw@Ls$XhmkM)P+_{i%r18Vi}ZGw#H_+0EgGP+tVeD@iH{339w?tv z#qVkC(S6LrsIO=!yJhRwfp`1CA0y#_6;0TK=%XxKzVp{49*gYp&aHd1?He$UuI#Ye zU?7WHtWz2D(B29&n+^lUN~UpjBFB+lhg0CaE+F=9JVOS!*;;XcW!-=_EXh*_Qu7>vtch5TF6#*(Aaf+JT6HZ28d<0O~JB zSs*{h>#(Qt=qVu#Hv~2f_wUGP_;CKMjr7^Vn!-KS=3##t^+<7I&`99eAN{RbKcQH$GQT-8}@|Bg9ySRFrch$j_j7s_(n8`W!*#ACwRA zRbBn@&DJ+Fw`|M3!|_wkNvHh%IsY@}1=)$Ctw8|fv?DR z^I80O@3h+fHbApv@i-`(ptc7Q!GV{dlE%-9o~O3ASun1rWtS(@$Z)p_Wp~q)_MQT` z^;!`M++^!sRPyHrB!W;kds`;Bk* zhU`7G38u2Gp8MgrW3ehR%^kj{VsaS*E5u=Uj10T&@3`q5W&e2Sz@I&knWL@__E z)2ILp?Tw3=OGCNi$m1svafAFK7}fcsJa9GR7XS3C-q+TMrDVvQD#0%Aei@{xnRLc5 zzi7rZ7_qw!aRpZt5We-vNEYl|0dS>=gM!BI6~KuGKqK=e4-K5JeQ7*)58GPz5{Hxk z4Kfo0ED6_0Eny^|PP{<|BjkG(-SK`SJ@Ejb($^-%N zO`*iFI)xa)L^wor=nbhDdF`*%J6)Ayt0)eS=YTRd?s{552yt@ zl%w^^+pd-ON72(CONEb>FB!b}b=O=!cm5m?=r+%oXd&SqCO4|usXWc=HDc*?^`5E1 z8#5rw%j|lu(}3YcJ0~DdA=ZADb8~SjRUuBIy(~rR0^(O`Wu}sijcSt#5@h^w>;7!B z<5_UwQ&RafRIml=;jG6GJBxq%TM-7y6psZ*igt+oXxZxhE#5BjW9tSxR%mRUk&=8v zX#i?qMP!L|rom*i94EBM7VTYem^=>5(->%xbUbZ}u+a&_+JAsnBmr4Zl4GslGZRl0 zYN?R-v%MkvjFK7rZ3v3?Bd6`Bzkoc)8;b=)%5jv13JPN=W95w{m0g5q{$#RY*UACyTDwG;z{sn}q$wGq4 z296H8OZP)m4+@gz_(By(sC;DB9pbg6?i52@$i_uP(=le2NM3Gnas>Vb#H30PLDB$m zPFa1xzykcdSoc|+b^`}sab*?&+BFcsY|#Py^GwZ+`xi$bJ5|72A|dm#eOS6#T7cDT z58#BLv~N7xhLD8g3Er4x=^+nY|3TodB+@03FvKeVN5t&Uuf3E#Clw~%oAWVY6Z`W| zM3WbC0o-sE5$AK|BvmCVAcyiGAG~;&sjwX4KD(opjb|N%1}R*IRi>8uTgTyfwUDR% zxazgC0RN5|-D;VeRd+V2;jw+Yh}((c16LtGc~;+Uchmjzkc`I`gy&fOey;Hw&)O~7 zd4VV`Cr`lZbx#2IEyw8OI$`?F->N$4_|zFEBOHG-s|$@o)^~2&*nY~#xC%|M@t5ZL zJcI?#ll?Vp^SJdOjvd$f&B3kE{`{20a|$4E&8T?EU^xb3*vc9^R(~1)OckAeGRxUj zO}vlJ9s!D0T+?R;_tfc=SGsh}aP9M>@0NnSbE|KQR!v`O3=^2u zfkn&%*$HPSuVQIaAMQ(C_KRyhM^eS0Eo)1Ga>Ne4-SFZohB=>eS4{jY zqM^;`5v{X?LRiC-(T2C zA!qpxr#t&?L`0v5`??cS0=;js{#<^e#ZMobW;z-*X$M;Kw2A3(6OZ#GfpK&S%QYCB zK>0!B=WQFAlGv3pcv5=+CnjI8zyduXn&5bL7636r+Zli00biTB?oz7I`8L68HRtk( zKf-3g?*6xnAk==>u;7R(^BIbvEgsqEBgMC1`4Jgp!IvZSBS70iP_4>7^@VOxB>+KP z=bLljrzhFF{4=#?H;pW!W>e;g{>xqk@@1+aEPjrQZ_iGDw6}*|_45AB;&-1={W41r z+wxam9@A19QQLb8enDu&9r&LQFUh*6{qa|OUn@V;L3cEVk&%cnrTH0z*pug<3k!Yb zSx-nYi4M;9+8x@30r)5PK3dxvv7kI44AX+R^gQ44z$v$EwfOv-$a00-x9s2jzu^P} zprC&!S^iI*^50`Qf=p(Z*x&^TZvCMv8Q37<>Cbf8(csaF{EyK&Oz>J!iOiLl4G5Xz z(`mk_LbY@2S8L4|NqDmh5OOox9L!i-CVScz{ddi4(MwJeN5ZOARC{zChQkjH{cjA` zNu<=(eapEk;z*agN#cc1-b?EAH<*t+h&&X%O3cDcCACs6WH<-d$MzCkqZ@`ao|8%g!<&h8+BxBjx%)a z;^w+YsH&_DcCmSI&U_6YQ!PKDGmKdLLV8{6fOY$7XhdmnvxLNwm4Gm>B08Mfo4XYx zteKmauL!ldeg&KbK|MD21J#2d%sc>BXIpuxD!o$$uJwFSh0c6}Vhx5k~l2M%?fhlLkjHE@krJW()1v zw)O=*hO*?m`@-JD0YT`|Bvl+^vAMLP0e-Yic>OKV`Kv+rv0S*@ykV%EOt;jyCUFq( zKQ?bEbjM`x_SNg+W5E_NBmnXpwqx@Hjto-(!;x)AD)>7*2D=q&B0J|~iEZ9)c;CoF zKuPlZiCqr4`yk%`Gtv7*|I6uMQrEu4%8ofwT`-0IjXr7{#q`wdx{AyJaM-Fe@BZIr z3I4PF`BZ1IPza%f`#72$aB3HEM>3Q4KXXP3dIH*E1ngoB77^0WnfEQhVp1!`pbc8n zTRlf~wIvUCj#Q8sEH}TqV~qJMoZ?8$FlDOrJlFJIlC-PK-o;qeXP#)KF+ZEf*>DcR z!^km%0Cf$9f{CZJisYFY4&6l!SGwQB+l>$$IM?O=sYz4QsW;kYifLZ+X~g6{^2{)l z`z+(u_Q5QfU2-PN5K^bIrZ%L>M5J?xI-z6!VY=Hy;AKdp!qsHS1O?WOZ_(s}v@baU zCA?;q0p8Hw)WD``p&O|$=4iB~ZLOLRc9s&2Gi3aTD_npCZ(TGR-FKv2yceAhl;=~E z+i+h?G!2e>q&oNxTTW~Ym-E0(b|S=~HE+=u<=M#8e`r&bfiLbnE2p~Vr#CM+eZKd& z#q%gp>vQg6zclC>!C{~<2-oM7L=YSq|MHFNHsFr)0H~qS%}MoRA^T|&T2GU^m@L=T zPuY@dMGD+rj-8lMTXHOE%b8k~20iUR{W=KHQjX7WG#(rguOsHg`yneQ0QF2GfU4An zo2;{SPg}#me$K8eT4y`BXV(-tg#IHuNdMDT5dabV`#&zTK+X+=?F>hh9%kV!Y;+#B z%@{7LMu`QTh2%*Ww?%?5LV63TVUn`5jB0kvGu=gY6^9kcPbh?U=fCQTwhcDpt!tpN ztq3{=X82=NQES(@<%tOyi#G$`lUWftM@ou%hOy?+%K=yFM;Go2?{LT77Nh9DC=EDD z-1;O76*=!LLL*X{BJ21Sg66%_^he5#2;o}`L0_Y#4EzEqD34pES~Hm&G3}vsjD;p& zjuLT?leS!%#CllBy*aD6cO^FxCht&{wo=>n9ld>cYSttOoK*dw1-xA=2ApJB#88y#}m`^Z9s{Y>GqS46anSD zzp9OV0|1kE^Dm&?OJk6Y9lQbYKU$694>arVsMi1a{42uu@6N@4iZt_|mG=MB?tkw? z|8Mu5{zEPOpV8$1ztHCYv^Da(Rr3G0&;J;&-~Wj8_y42_|F`Xm*%mA!B{eC!HHP+; zF8So5JuBk-1ePB68GN0~RQ8hW!3}+_2#^@$F0INmH4W#zXlkx>G#~HIJ+`LEY3g9P zxJ7O-wzKc=>hHANhG*X1%n5SwWR>I%JG$R1y(!r5=d~!1U3KXLUMQVCZ=0z(tmjCA*YKu6c@MS>0k(%RXL0NJAgMH zXM>=+Tv~0%7moDMg>b$i<3km~kLLG1Q%CEI#y3)ByAoin9PgVB=VPU=TGWvp5NgK? zRk<`7G}tIqwYyZG`xw$;qp#xhik;i19{BsUwz_9BP5E)Z-|5}?bSE8u?4ZaxtfSLA zm{W(Mexi1lAzVx9mNd(jBE#P#(`BD1wIKJ&!vM;s(UX~Bf^CMYYkL*+mR#ZYz{B{5IQ`O$UO9gQ^W>sElN0}Pvn32g8~_pEI;DC!hDw6L+ex9Hf-bNAUw=jlEjA{$Gi=AC zSS7Br##8xcN$JJD2$YHZmmm5+iwgTPpTk-q!l6UY%l2#jdl4*fTw4dhPF<2D*ZVi65w6h)Eax=cM=F zdFo>y+Myu7{i(a{y}_#<&@x)Zoy=Ec!~X>mC(r{-AE?$p4+P%xay}liZrA41aJ^2?YyD?D&@LK$FYySP4$r$h0nz5lOvbCq*AiD^`_%DrB%c`$#ZEGn){i8`*FpWO=j3?p+g=B5mS zGb31W4G2bCgBfxB#Ohy;)4Q>j-(e;0S{M_X)q8QrAKYK(572l~`0?D8GYa7O(1qhBf0NFT3ymlUB zq+|gt=Gv0k0?7iuo1%iJoR2wCjT^0v`vS8cCZwFc^r1k1_5jmQA==9tx4}!s3iLo$ zxk)>56hMH)IO>8fJpxf?0gUB;HCo8ofaD{=Swt0N|EfuHBX*6B=w`8`dx7~;pz6Dv zzpgMa71U=NfA}FuUl)G?DFK7a_S~f2@A+%Te}C{LCXGM8`@YVvPCc;2l9pK}o~?r+zYP{Ep2x zpbS9_uOkoX(*O_T747OT@G`QXIlu)^hQ#BNegWP6b)7&UU)}ycvU!a0 zJA^2F9MKoB6S7kKvjSDlc1VyZ-2=c86u>Tll^|BR(ET8C(4+HzwUUddkjI=i5E}!( zfHHvbu~j~@8U34WM$t(EU^&y0QFA-Tj5iwMfYRUPPQbc7RwVD&{_{kTB`hG1FFOPK z2zkKx&Jw`DT%JAXM8p?_Q{PnN?jPL$+od}LjQF^5lMTq^5e!JOovtO9Zu{F=`ST3@ z&F@7S8m+7pG49C|ndN!oy@Vn!rt{IN#pl*`Ok7hgR7YDiWn>Q@qOamC=4WZ)mCZFF zr20Hldy^kyiMd%vww(7T<*xNz6OFb6R9ByU-F7*R8@d2Cu&ahWlrw8+qsag6_v}gAGP00Am09sNTpsHUz+k3DbEtK$&SxT0+cbsKg1H^l|dT6FSxOf zYIqsS)AW1UMoXr~^;D{tpwO7F?N|2V0z%6LSGGi&G6h~N8e_Ku`GXD2br(I;#Nvff z&8{DORv<&^+LZkghEj8ltadY!1FcJMMa@(^%}pS)K_@Td<>`FZ{#dwE^-ILu3{d9b zI-70ON6ObtFIwiPGlBA7dx)6MLnIkrdZdZ(#tEDYWV0-UTr(yyVHFDI%xdT?#Gd)meI-xndp~8-@B8`ZR!hl;N2w2%o6?Bku~%vvZwMuPJ$1LLqfgPB z-3z6@8y7*koYn>u#hQsWSfIN1V%b}S8^-;Nrb=EszRq_9NrCIu9q`D(Kx5wUIU(u^cut={ACEsG9lE0DbK(@mY)?_ip2(~4C3mf}g^x_!vQ589op+kui=xP| z7si; z(3>{$I#Ir8dVD-RS0&&GQ2yz#w@#ah!tkUzC(W=La_pJ(!uKuxUi~PM>P3S@`V-8Td6W!lVqPYEf z_H{Z5g-_~|=9iJV@WK`afUZx;j^?nGHtTRkf9i2-9rw%gs4tY6{`urXWhe`+u^S90 z8ug*u<2Z)9^P&FoapmUd>llxh3e32H1$in>3;qXHpJNXk$F^0DL35t#4kP)0ynoGX z(C-c=I<*lX&88k;r0fQ!^`MH;o1Y2sc3)aa z98s$Ecp?pqE$@%>ex9|hJP(d|kfJlSsM{&Ad^D<*Z)lK8yB%RQAv` z!*Z4D><0AYu93^dI?#O~N!4L&!kkIn^D#EMm<5Y&-ui{OL_t4nbljWgYOC_Au$(cDg?6Rh6_r%gY|Pxl^xuP9ZT}*+8G{$~hrwT22iJT_Q_p@}*gP3;clNWn zM3&@_wfxHJS?t~&<3s-(RT9Zv!|lTa$|vA<%wM_kloTcWiA^iH;cf0364WA&^SQ(Q z{hEW4<6F&Uns?!Ccc$)it{JmBM-rSd8_gznabZnZ4vGnh7sd-!vvNvV)}Td9dxb6e z#A-gu>2ZBo==yYXvlt)?SG!d16xjZ{f%1!;95=US7S`c&FGN!k4bOj+SCc!RT-VtC zJhu|71hDXrB&zl=M;d?QmgRrz7KmG?1S9J~UUc@nPdu zu=}W#((>dw9m~2bZ5i44z+bLk16aWVP}e$Hy4?IpQN6Psxn+sM z=#=7tP9T?q_VpxSTEYZ2|8)!tSPP- zT6JaSoQo~PV?<8j0I=BCOxl)Arct!}+3g&-2dK(uPH1eeF}1(^e8#vpY4X9l#YFMf zk9;FLuoW4^!a0AZb{4Vmj!THLO9}WO*zLJ#(}y_6cXq7J5@RIl5iXT9u=w`}a7)tm z9D?Yts1Wr2li%${4yIPuncNuHvFq0BiR#>3w|H!poI)%A5H#fEEi+;RbzBqhQ9igX z-kI#v^XcQ)R<}?4{Jf@o98o(V`;$=}b=?l;TI6CY;9AR) z>aAe|#@xny0&5+qeXOVL-Mq09$E{-~wBr^7Q+@irESs9& z0v^5abJF&4KVTr(s=WbFp#tXB=_vJ+K#<&;A9$Q zT3Ctj08>vhOV%3AH@;v9(8&M(8_gXj%E^N=6Fpt=htc@Ub3nQ1r<#b~| zV2|8p_f~Dm`&WOzUn<_^o7fp#oV76HtuHzYQ!|^p#%NuKUerqS7K(jcsLos0|4f+i zKN=~(9{H*{Q^wzmT7~%D#O=QVw^8Z_`^XZ1 zO9L-9WEN`3`_RN#by1>4Rwpe#9z=UZRr{&HpVq@ z_dNC35IJp{LB;u`m8X3R%5?OJj@>OL4_gdCxB{+y!iI%y9mL3eihWQBm_$BO91l+- zL6g5L{KcN_Z8a)+3ma&fafwPMg9PuSZdRvA@$_dmWTb#3Rg^9+m2@TUi_gnU_9UwX z;S~XK{?`Owc{cDE`$AP59WG4+GxoB~T>OiShCRL)#qu^$6%@)Y`)&G@tLT@b1JI$p zVE6t`_zc$oLD?o7cXTIM)AVrM!Qb}eoyfRpQd}RVU2^>oZJvyywdGdwu9=y7f)YcD zH#@+6_#w3wgN}0FVK$nAt=YL8+UD6Bs6DyH+m5}u4e2bVTNIh3=?&q`kt@N;n1*O- zY)&a`5^DR7Ab29yVB624h}Zz7B}ieGGL`^tEcP++hHiK1j(cBO6pjJnNYFlB@FfP? zLKWj>P=h%Nmni*!3mx&QZ`k}$K>PtAx~-op-u}l>$)!5qg#tPL`GcymASX$->cP=o zK8T*~V%wBX51LJT^6txJIVb`|p1NX^fO@q&Y_@P*jSm=6^}D z_fZ7+me*A@L{5|@(|5%a1uJneS1NEe=;SBgOFX8g)wsc<)h7t%aX8bwO_<*g2b7BM zwZz^wg{K)^v7+Cf?%JNpEX6v1#UBNWY#u@R-g%Kv#$4Z;`eyYwuwHbm0b#O90nB2ZHG3m#bjDC6mr5I-hVm)d;DZ~1sBmMZjFQ?^~1vxLqgZY-l0-kh1 zqK@&wJw9iCWAi-i_AMEsCWoq!9cMd?p%bV|Xf*&hEixjemLB)l_BR z$g^G9RfX<))_F^gg?UWNw!}#ct|nd`Ixv0G@?K?S=@spF{?qs;m4u7)aJj4V!-HGS z%XJkDawQUPju~U`7Rw?INUz93Fq18SRc`mumc(=4dZf&>9)Uu@+|ueHO0OXOHiL-z*xU4YGqEe)xM!VEmd;9`mnR-xgmtAoUM;GBf$Bx1KmO} z<%EHI)#UjGsnh!8AzNPo3}yewvr#aMy#f>V^pKe?=XE{a?ApXfRoYT2ANE(z6VG_L zPJDe2*EKbLv@p}#lYDdAXzQVLrySr6Ov22z7;;|6@TGl(J(~Vp=5|4}diXUzwhh5r zx+p@i)*vxp(nM5O5A(0@^2dQo$qDT~dq@g+TlZHmg+xS-l3V^vU87}5TuR2bXiClu zF^6F}kz186^BZb)%4h0W(~8%g1ZET_eOYM`Mf!lkzK+Ulr_FcVASnCOZ0+;pAumA0 z$alB-3oQW){UBrgXPTZMxYb0nt$IFfOQCn$RHMN1oA|vY^{ORrj?Xs%Qz8XxU5i=| z5DNYZW2w=g%B?)nKg?RI`+E8Lsq5?Q_)($ymB;`vd{K4qAjU`2OL;WCnSG2tB=MZD zTT(lB<5FbfT-yCQ9n>Ve%k(X9FUoojNow$tJE-Rs+Elv4Il({RG9v2C=csg!(hT+> zXHEsm^InaV{$|G@z@VnDP1eUB=a;&tH&^fQwLv24A?ArHUtIn-CY`|_Nl%oK$F`Yo z_Q6`Jnm{fGxTQg=6jKrR1fMbfY_?A8lyUodd#;ZQQPcC`uWpftOqDK1fK1k!eEj0p z>%RB-og{tv=P-fc;{7^CsYn3&N>>%FYX6 z7-D*BY=)Kdg=bkj6t6#@>y>dM3L@rH!kq}Jc{4FiGh}Exz4DOZHD(G$+BgxD9+D?JF)cS&sj1SZwsl2~lYzN3W>>T+F4rLo%{{trU5mYp0<=K< zoZZBpf1`S>Tm8GY_eGdHx1xjA6sd>z!4U>Nmm01CbO|jP>dZ0SYj#cwy_cxpw9CcI zE+SzEvp5mMN-R^zv+@q1CzXlB)1^iB6i>l|Pmg|h$d~ADoFk^qrR{rvKyKpwBCS%` zkk4$AV#OSSFDC+LoRF ziWXbwKdda@t13B&>yo!32*F9K2zWeo5Cxh z8k1KYJ*-75zv%_nHav{B{LTvcQi^Qhynr3en2%$y&u)?7D=WH^x=r73+c2GdgcywmY11xTWE9}`+6K(VsG<-IPP`-*rf4ucX>ZS(YLkAl7^Bi=r^ z|08BdKU3bHDUrSpGP{Do?;n~o zY58+2mkrJA`nFewt2ha>iCpFA1l;TA;E?+mSggbrAHzqjH7h?BUS|Wy3S;A6=7upg z%?0cHa8yK62uA8t`c&&HtHH+N_xW4>)A+d`dq9YSKMxm0EXXza4o7Mv+Q`V|nieWs zR`_}{`^EmCy@MY`zTDRYQfulh@6J3PO^4myJ9pQ}W(gKtaFj)>F~8MTFu%j@0HbIe zwiy-@daO~%vd!im#nRpdU4U?F;Z_8i=cX0ZT!qH;gL0qQoENia(>T$9#=@=-qiyaB zJ#!DgmstHp+cUNL-MSt@k2Q5&=7!Q&bh#s|f45|Fp$)tpg9SX0%YkB!T*D1F9Yigr z&TaZJSqX30y6S2i={0lC-vR#bmMHHI$|CD`dGO+_qvPTWfd9%;J3Rd6CECFAiN9)F zx6vH%LJ~b4wMKqBD06N4QTfYs%Ga@%z)k?*{(0(56qfr={}R@az^Af~A6FP~$@NdC z@()f{Uiq{W2t~2RzLXrF;giry`L6h+jepaE&I7e8 z(E!U3Nbz%Wj*m|Fk|1+#mZ?*?ifKa@Dl@dDd^%?$+JEsVY$bJRJ3s(O+pbw#8KoQ4 z&|O1jb?F9GLff#}qu`-fL8-e~C*&M77{*yu_pMTQBEP&Wj`(zYDv&ado3B=g?ZM|^ zp`wExrDjh9nU<0W-s$raB_&&pELqghmp+W!!E@KHHil%SF4zex*Lp;l5kC6UQn>7x zkNngz7*&mjdQtCZ);3&(O3j)EBHlkKdKr6`D^PkioV6`aq5d6C>VCbWg1=d9pARzN zn)cI|Kgr2^TmyJ9OMOid?z0T5CZh!|_D+K|cSjEqqSgiw->emavbzF9SveN|2@o+6 zJe+)(oUbqLCSJPXvRmEGYXtJ|kS_?3s@OR*Zt7JvBD6?U1C)>1b*7EHHAALlmOP^; zV#nH7z5J->71td)4**`==YC3K*5fTqP(k)_z?u zwKxw~ncz?|q(*u3yxnzta@xk51AFWbFDBTKOZWL{pW(QuX%4>oX^8PCO2a+O>!>21 zg(yG-<^d>3C1FT~r8QSoN_v2`iZ1PR&b@7V0&IN^uP@eOOP2mDDXhpG&S%h%cAsD1uW|xp0|#uxwtd5 zx02%oXKKEsz&7NR_F$N|=F_kf&F1;<;|x_DRI@{=%X0E^*RklfqoMl=cN}e;67R2^ z&$i`B!f8_)7vq|nNS8;lqKq0|^zQCLl2xLY5}<5vXerd%{5N6-goOxp*m!u_Jp)I6 z%@PM=5l=?bx26z*m3+{_A>LpmjG&z7cfW2}hak&18e%&yghq22s?FjmvS~AcOk=_F z%8D1g(^o29NMF4Z63YYlfCA^VPnZ3?Xu9D?t-5e%^M;GvsVRFYw;WXZg8+lq_=}q| zV|uU$?vrs$Q}znyu)^6ACKV|@H8C;!fXzYntdMMoo7cZ;KkNL=mB9usxP}S>2xv7u znTTo4zX-GIbGm%$?eUAf$yK_+dL@-%p>ibc__}CYp|iJB>7^@=C|ItLtV>xnabbXxp6YJF#Np75ZGIqp)8&W2A}188 zB|cPH@5_OnQ0Sza_-4KMXwR=`YdY2?(K-Gv4w^fZ7kQ87Z4( zP@CUkxSh$yy!D=r=R}LDy5mb95HB)bt+Ojy=VTd^h%qsa!nx``J2>kZeb0=LxKb<* zXy_ZG$QJ}&%HXoMWhWZr#f~4Uycdi`!(YdG7R&k`9cEFBgmH3>97v54Bo_?T4m3G+ zsHV8)plsx!++%UWhGg?6&KGgPn#lS3bcQv&W60FNy2;C^&X?7KtmIR{$G#pLb=)4# z&REC-UyhrYlA}}qjl|nc`44aNkQfJXB5gKpy6c9wY9996iK;V<9xBC^TZUfsiMB)^ zWKlPs#(@nmQsKgzo06xIO-XsF@x5xPEH#hl+vK5ZnGXH7zvVgJ1l%-qIORI-0*AX- z1t(gJFYH_MQz4%vz`ViDLP{+P9C0m7B*Brs@a-#1H@j)tIvzif5)cS#+GnRX@xu5= zwN1o@daPMj_si+CNfzG!aV#E4e(&5`JefsS@p`*fIS>L4dq1>E=qok;`WYhHT>;wG z1bO#_%;aF%=H%D4LLJRiFd13oJS8`;5Va^5wQV`ijVN64HW_O<`yqDZ%5IT+{I^V_ zHtM)pmkpBQ$ADAVP~;ps7_%<*HhFz_*UiBWEC#*m2cpa)=c|V_Ny!|nggI{esW~`1 zsP{Yl$mV zs~VFN_s<)j?zZ!|;T>w5USP=iOEwxr@=$b8A331RnGtha-!42m$@<87k( zt26t|1{adFhEW%&<$_xCfIKmicNiR%on7@jcak1b+2YTnmSWE(Y1+FK#+m zR_(kp?0zhPTTpyHnMbe~{Lo|Zhbvk(^T(Q|82zwT4N zgt2l-#pk5%#ATRk@9Z@5+(Q_1MIF#{2PE0_?38`#<#$iUE*HWlCqxJ0kq?{>*LKYF zLdr5IObd}rwj^t!tMpR*aoFQ8q#Z@Ku?9D@hHN>F6UB)hxrj@*OfGh8uaMobOoZmr zfXtFtnh|IAuL-^XJp7}iw#LhwE4qt|$HVdqc_3(`J-|Uqt2$*twM8D>a_bh8LCxhT z_7LG>%iW1}L*^#B?KqQh3_8XG!UBK4t|`zaiOKm~cK7g+(v(JKN!)ha-kSN+uO4j1 zh=m>c=6ML!tb<y5-66lb?;+m9)* zG-%SUPTD__AKeLbZO*8}FM?xS8!nc9#yXYe5fm$HGAE8UW2}6iW;u#o|7y_3O<6gF z1+`b!VpRb=D%&dFpcJiEyjnB#hQT4csgt_Cx^ppMcDL2R1ZUWcp!RU|Y7>X0ISA%Z z{qXiSa{Urt)p^?~(ETHj`Bx$pRYzv8uxCGGGnZvhQMeeTMzD08r-3{meNR;a8nDwZ zO|+9n;rw~v;eJZ~#T|y#S^69MebhMEoV`y->kqxRTq=_etm&mC+_AKxu%>K#ffZ@y zK9t^2nWM;kLvwM>($9vsA3L6EW7{aB9+~(RxFOiR(tzb}fzaF61caa}Lf+en`=@H; zThd9kH0QI&d{k&eI-t{J< z_lo8{w|u~(&c7kQp0%fKC%VvicHN3w5adX&#=qHf#^>Z{_p z;PZlxKXPcmv>8k(AuCV!aI~p>YEb=iHw#-FOYhViSx}OGLNauMbE)%uMa_jNATHm4 zK-sQkUIS9z;E_N<>N{b2{Y5HlI#n?+zBGu3ewdsWxFZtOT2fz*?Q0XHal11u{5E?+ zjj}Y9AE7wU^u$f)2xqxJ`gVC}2uD2P@)mQDF?y5HG6;Iv=KNz87!WlSU znt^*g9FjbXS{V{ycpV0aN0dCVK)00E|P!YO4sq3!cpT&M* zF;EtNxSWH%guxlMjw(ohdd)jfE>z)Dc=u~kQuuPmV(?dm*XVqLXurqo#Tf^m-Ohr# zj*_pZzuh&>XAe2A;_B*sP%PD~z*aUCa0Qq5<+(%>&T0LNnPhFrW2dd4K&Vo$Kb-Rw z88H7C&9PVsl%&MuL1|4{jd7x1=7ng{(_U?7@digBeuBAh$i1isLtGP4duoFI$H-dI zZNzIM%ytrA$ovV1u<5vn$>>!kBv;@!N-gwM4H_PSkteGq$vyvNB@i9%eo6gPG1!GLtwR zDDD$(`+iiFVEQ--RE1nxw49`5A^6fO0vuNhNJ?C8CqaUA_4DvbCYVd7>U#R(X*B5K z?9#_ZS`W6gi~e)<26rbyBkz9Y_Q)ItLP0M3?BJ%4xRv>_5NnL))Toa~W}19RE!5G= zfjK5VtD~E)j#d`Gf>N887XUb98|bhxBi||}Q1t~Gq59x8op_beLmrC0dXQTL2KjK9 z;EO`oPm95IeFyvxZD-S}&vwwnDw@T+;e6tr{5x)e-xS-W0514@=d*BuBmIgP*s5bB zXmN?Q9fW*5hAB2DvEf3^c_lRv92-bX0&`1k>LQU5$uhpQNal;s8kvUYH794@z$`7t z_^7u&e?wcZ6V&wye?hRfXdd@9+itOW^sq*AQxu&1u#VjQ4zFlAK1|G?;7JyHzP9Rm_KHuUd_7&0%sQ^`Ys)nMYdGWP&itz7E1F_a zd>qRAC`wKcP@bJnr^&`iwOp=ARY{`dT>187Fxq1>34bz{vo>=%OatqgS8Os>=dv!E zmDzvoyOqVU9a2}Cuu7hYbHSwkFZSLutf{qI8x4Y>(gdY96%ddn(rW~y$-;n2ktzaG zA{`{u(0dUOPy|8|0clbry@T`;igZZmHK9fb@tf;i@4Mf%_jSI#uk+*lIoJNdMJAb< zb29VH=NaQ3_qcD-tB{Ml;yIC=qDRK^s!300gfbi{^4M3-Y~xqX9RC3m3~68-IU1qh zFy6%F@w_-|Wqy*^WVTn_dr6T%f(k}ELd~WcU-_{te_BNuQ@XkqZt;>fIxKq9LEIAu zZ8p=0d|2sDYVnSlwc4|VipO~}if(dV`!fwP1TJUVgjJlibaIcxN$?ER zhSd7QM(6Po$uC3E|B9bwU-ewX=QQ6kfh8)Lu^HP`kGGyz0M;5pt6=lcF~W0fRufu4 zf^wL8pu%F>OnP6NWR*LXX&uE;w}wEA(r*&)O|4?7zhs?#ex}}EWZK=z`>v{PMy}a(6=YLuaE|6asbz{9SMS8Hy5dT!i8$=2G5K?!S;D#K%_ggFEuGPO}QciZgefvUQT4^v7}z zLz5uy--!w}xVCNwp(+MJUKX?p%I?K>m7DeD{iU1q{z=CW1Ma=vBg>(HnvPAxRY}1` z7KaYA?_g{!wc37ml_7I|V%_D6pC0RroHZ!C^B;|qH9Dj{Z#3myYcxt89s7ZF;sX`~jlad(4w5mA16IJTXW)sXoa1^#c> z)%~h_w`Xs?u2xW0iox-9JVj7#olLRTW}x>5(eNw*!^C&qB=tV-rhE~ifXKKQKIRJy z-ot3=RBgf(`(nQR#5x0w=1Gt7yYq5WCuF9FQxDsZFA^xst|z z@w^bD5YX}%v~%Qd0OleneoQ-+W-=Hvp6m4KkQyub?8_OD^21LOZ{h#)&9Mcu*2$;Y zn+G1}QZGkmvia69!9C^3sRHjjyfMd(&h+kLd$(=!FHg0EPU2$tht?@30A&kBxodv~ zJYhK6nB$_xr+xy zpNNsRASI|v^|jh@-gTCm5Bhbxq371yj(ClxR~8O$TUd@WMW0lRiPA>we74_r7Rfv4 z;0ROUQ}n(NP!P|V+g@P7i3W=9#mw0mI!J7zPV=@#+0U&GWU{GSqB&`Cw2_{0S_{0> zlF7V@!nc0emkRGa+gy)8@vvYQy}XGD8!V*X zZF}P9ilu)QQF0m&>8=OImhM^TB!TJg)Fl7u1;2I*WL{Z#-*Tw5aaIG6@!izsSS6+!GTjC-UH`t6c+|755xfGw0{eR2 zdMXsqeO~V2Wi@%n4d?%y`0;BNXVy*p6vlV+vu1W}6;g}GJxs5fUW*_wi5ww3p9ot( zIJ#Te=9=@Qvi|6{ed&7VGPqqM!2g1uL_rOk;-`hDRRc|Xo9=+2jx2M1O(wm?`Jm4w=4?)xLlBGYl($gYW$M0x6-!L_=PP^XU z8LU8cZ#_eR%nErEn3VkE+wQnAKGkJ5EXW|OM}NJ+8T*UYJ~uMta4+&ifs)3=^AZ|yy?&OA@@ zSG?@;9vCp(_tk`YKvc=vntdFPzy07Q)1^oOR^bQucgXaz7&J2hoYy7a@jB{tRj(2% zc+Vtz*b;!^oWd~Y{TB3)C5uS9S;|eO{_q2_NT-Q#kmbk?Jk3w?BKuzSaOmk&y(1!` ztWkA&HY(j$y{b3$(!72k8S%q@*0?DitcR=phz8T)E88UM)@$@prTwq9)neQ!tbQhW zZ+83WQm8J;u40Gk;7jtG%=xuW)kW;TK_#_TrvZhq5)SC{+p02x_JmE+@ouA7VTrie zuP3BKx>kM1?6K{9(KYb=HEIN%1 zLWJM@WTH*jrW+$ELuwE#ZHg2aqXju^+;DW>7;iJ-KxObz@x z;R)8UB;d|ennu%@SNY6N>CFhFfy|FrvfZvgOlv%kn=h%{oUw?nW_Vy>?;vui+$6?; z#RDAlgWz!5KU5%Qa|wx8g4a*2eQutvs=T{<6%Hga1RK^aPMPu4aoP?>murA9rWVxyJsaGUtV( ze)n2j&}KN_rTVSE)}T-vJo30s0b&Krgj8CWYx6xxO^1EuoK5VMyHx-%bLux}Vt(@S zaA=8>pS>+QLY2NTDRyZ{eyY=F8 z<+_^Aj+S$R){^a;Z(u{pb3h?QzQDb+!!#%))Ios7)w60mcH)dd>ovK~U3|eK+ z`kE3vOOGf**!w%Ee2=+|4>_#blVKDPY(ymebK+uxu94^;W+IOU2732n-+?dmx6-}B zasWZ-qU}_qD7(*lJHF~`QY?l&PDI|WCF7XnAT@I8A+7o~2)9o;c__T^!+zofQbk(;!pFE%=AR9;XxL zn~%3dZ-8pG;@Pu41-11YOU~HU`_&sn*%S-{8e}Fa6_TAH)OX1rsDWPLx!V}csZbnr zGuwJ~N;bPk8IAknhnrc!_33|!jMMT|)J`-DY{?DI;*+0_9u-p}tEy{I)(6ucPi zO0n0_`6Ey?;0m*vC4jMS5zFV9ibwrgj+*YSia!kBCHKC5cbK?66L_C++m$PnVClN- z`O!jcg4AN8ve@RncM!2-2v!&8VK z{L*ER%iDU3=r^w_aTZZZw=crU1tljCNij}^-Us8!u_r^~#!9#AM~FOlpXb=2DyVCT z{gB8`I$uZ6y#pDArK6*@?;P|85IX$fLo7VrF}OE(_6NfD1959lFNh6;Z&t=*78yne zdf3U2%SIlHOfa7r4sreJ#_0R0h1@1#R>5H!WWPawj8M2Q4?2mm!}xZY558nE)z(KI zhnq5Za&Mo#`VNTzUvu7BscFWA1}|;q=b_$ElE>41N8HB*Ypn#Z>?SCWDu^=1p{Y#_ zYQHDmo{-Y!O&zV^C0%pzhp_wx$!z@9fKbNOxga``+&Kz+)}CsBl-go;P6psdK&m=X< zVl4roFGKi|O}!tTbz_R4b~ng-xz2#?obAiS_I=K$-`fKG#b&dhxmeo}pa^?;lgqk# zVCre+siB)Y&;DHU=7yl*qhR8cYGugtEcqDx^7AaCwre{SJg>|DQ6T_50r;LejYI)_ zrYGDBpbbN4H>8qlYATSC3P_PAvv_r;Sun;75w75vhp#j__hsP_DS2&Yi~In>kj4ER zzUMlb+|Egh#CouXHRZ1yn#|8P^{L%rvz#XgF5pLDL`oNcdZbUwi-s=w>_+1cx_Fdx`6j78l3m&7~4Geh${ zi9-zPBU*ge0^>zh+%p3vAOD)cxU48xk`a`fB{Qfq@+Lq_;`nt6FagZP@vh~t7_@Rj zMKe^GO}ElWk|pmY`&Xn`(9{}cw#=XQi`|T5M8r{{#NdN@$K6|Ig)wJIzR7bFXa87o z{-fjRU-a_7{U#kI^1zI@yUq&2CDy9M?FC^6i!yH8D8T7_BL2H>Uji5qCv4jUl@1 zqbr&=+~>XAE4Gi1KjVZfq(3l(_Nz$;JNa zaT%gb>1JBnj6(g3i89gM)Ynk`j}Pyw``l9e4H^PG)czCkTV{~~CT=(-3G?I->}E}k z-DG+{(wT#{%vg8QP+FISx0};cU{h*(QWQ?prkvNz7ENn6ZJphJ(zMsh{rm2elwXIr zKgK;tx%;VjhPSKXfMogIZ;&@c7itgq5g2AG`@I8 z%si2d&cvjLjIP9`JBRM(%X{&lOP~Ly;#*K`pL#Zw;`wEsI?kHVyf+ zX#_jqlt*0DTlPgfGvPVj^gF?3>X^a%qPbKc!J2<|Q3(E!EDtvfxQ`o-@|0hch9pTK z&F&mXZ*0}gw{0edEHnW+fQRLzu&omvEdn=sg)xh#hFL>_cZ()IB1Z=Bc)foIp993E z*YOdCSbz@{zIVB4qN<;&DwVWX!<32Z;T24lng^#eRyZL}0za_=)p{equ+C%_Tllk&HY}zU%8K%x15*v#dYaDH)!(8~x*>eiu^xvn8EU)`Vb5zY0 zf<5C$ZgS#>G3NPdAT_8Xb;@U{+R+jhm0+1spL;=p_l*Bq4le<#1C8OI#y&t9_EnfCO<86U%n%hHoWAVkl&tBnw zDh$!6c1bBIBA!KCP;Q;+G$B$dHJx9+QnlR=2d$jDDAhpQvTj&h+dRDlMK+J`zm{e~ z#^1MjcSh6tmSOCl4eAeGZBvCy&fj{f?;e}sB5AM(1q#HMte_Ri^01kq2ArsVs`gYZwmbLRpx3Glt*ml_ z;JBN3R!01nrq}K5Q`+CcySvGPC&{j|Yu>+kt>WM6;tUb!rc0R4bg_z>NoS9bp)q&n zjM=3$9{MKkTq*72Ck+RJiH^L2Z9S+lvB?`v{doTItd?62J;*2JA!Kkreea(z|2+51+wL<%p7#dJmZcSg_#{%aj_((UMRl`+1>~u9bIZ zRh$r1z7xQg`j_50!+JWR!_mZOTx9aJN3J?ucrJex70k~I!Coa<_I)Yr!*P@3a!N{b zrq$(?&)U@0SGeOPH@=dCiZ=dw<`nwTlR=`<`vk%4NWOF7uEUzHPH5-jLf<0_mTv!h z<>3mDB0Q5xOk0-thhBiRD3Vf)od4)U=Kd4tOOW$nJ@)Cwfx{(+ne3A7eLl-C2k7-I zL^V9f>eMB-=$ya1ZGv|%`8lmVW~v~MiBs23dHIubouMC3(K?fPF^GuvgPsqnFh5mP zY73_y1cB&V7Wa3LH>Y4s>QmL!)YbD|Ph)fscm?bqgtYW>d3h^B4vr;Je-*a>*(ptN z8$C?1RZq-w^v{PO^+s*N#aT3;D?GV6^mxU6Piy1eaA+fw?Xp@OV%Ly0ya?7-w!inu zBPnsjm;b0N{jX{aKQMU+B5+v7=!nm72;)vAV=luV;&7UH?(R{LU-tlXVtmm6hX(Q#qV|hqo@A) zF;u`)`0mya<@8Q|`hysEa_Me}Yu2$_G-%1gIehB=He)(~K8N(gUJ$Oca=Vk z4)&6+D?9r~w9gMJ=f!b#KoeN~lqdHk8B&K91o^p9qTEi8ve#YOc9h9{FQ|PcZj!7u zU&YntK#PF}?SU(jKh=fH66TtpbaL&(!DVV%xM%G?0LSt3h^T_}y$-mQ$70iK(mq3G z(cv%h?t>njQrIU5j2VxKY4YV2KL7j`tJ8Pj!joe_th~BdJBH)#o>}^EKj& zW8Ji;ADx65k0Nbexp`4Jfiry=fI*s>S&Oix&dj)ax%wWxOdhj7={E+&zks(t?<0j6 zV7f>wz6<15BZB>{4b33GB^FrcMn`!kMnZ_6$EwgR%k)9I}ir@firk4 z{TUkb<8>c|&iUAJHiMp46_LZ4xCG&tf%UaNf(>Ss+r0%O0XjN6f*&fR4C7h1p>b0I zTqX|Yf+&K{mxca(htvbwYVAuQCm$Msx`nT8mLjZ`;&!Q@$fW0Dlut>ec@r{Rk|7-0 zBGBsHN9ApvmzZ(1;wiZ;_5YL<{CAsgP_s0oyXG{esKEHlf<N6#Hs@^mq=!8YTWuq^ z&!JhNqCz-KrohJEMs>yY6)!Cn7)<)u8BIuDk;uHauUF^z&E(No-HBd0&H@AoaJ+36 z87m{G!retap+BrbjlR5lk9;RP^OY7v7u0ha=E22YQIcPXW=WJ=6^ zO6jQA%d-^V*$X=~|D=y~4EvI~l)zw*+8Aa(L5gp7J-yA`Hm9K+lyPp*lkOwhE&TZW zG2Y$VtH2y$?^gAQ(TIn?l?ikpTb5FDAr2TaT8$qwJFYcEG85K&KBxru@S%#zW$+2* zs%YoHq7oH#k$|BA3T!i$Ie?5c0j-|mSPFEs(|F9beX?H_ytB@h=0Fc-w*s>F-ykh~ z(6XYt9&s4uIV&I2xcrJr8GhO+|! znbSg|m>sK4dB?P*pth#4s6;`uwruvpIFy%e8Z8BqaxBRAxCij)coIM;G3|>8k0XHjf3m?4- zr?-688W`$K% z<&&$XMQSV{(#1*SqB5$cWk%%JF*EmfZN&$&dC_WA`FPMfAI#lAMJ8pOR|Me+53y^a zw47DRjP6^X;^Pc}zE}m@sdcyJtI3P2O6VVp;{G??;P-$E1&94fJw;~=8I<)o z-ORlCqJQ*?O3tH-+R;OjDG&lRRu?C#-&l{&L6;g^TrtJR`$7?w5A~BBzY5#}vDNih zl$U~h4WiIGOr38NJoB~;P7`l;eE7txlvhXq`s)Duv~3*-`fWx2V95-}eIY&jui3Ox zs;A$4lCuQ)G;619@D&Bnx8~OXQG$Y~rSXsl-fmR%B`T*we;Tq~**sCjz+ZCnI6RSynngi`PabX2BPM+rykv~RB8-yS%<@X8D3K13E- zUnl;_%Ts`q^J^eU5Gn-d`yDj|V=QiVEbU)2hl=xk7zI_RWq2oX-VA_eOO6<(5`3f8 zu?&q%H>g-cAN)aa>diF}4T^b;<%=R17IYQk3jnrYW>;U-(;n?lKjJ~&pyS>iB-&v= zTq9ON5Tmy|_|$jaOQkNd*RiOV1t6^w9)OqQAvqjfKD?hRhHc13-i>OpR?>|GKe(5& zAL)?^HmV$R+P8`(p59%cr?KuaeQ9-=O{3Z>mw9m-ui)(n)G`$qk)yngZkjJ%J>|Ns zydlFO2-U|YEM3qam(x14F)G0FO71|F(FTLOzGPazDdtS#J6qk_n$L~tW-=~9$$&eS4`H@D#h=53yK zt%(v(SYX@3ZwwUz`0zXy7m@hXj-h$ve&)yDAUGS`_z)33W42bY>>Md~lP*r=McGQU zyzAZ7+SgB|oH!n~vM{`mWKob@7F#Z7^OO1Lv5|Z5dE8@TQjMGFW6MmvC||YvkQ3wo z)yO{GcZ+H2oyl}=xc!PtQ3wj*(=|%Gg@;%)GaA2cuvZ*CudptbKF3 z6gdp5q>PB_3TW2lg~XUwzKP?s=c|R623oGwMDC`qO`(~hz%{KuQcaR9UmuB``C3n@ zVh>;7$V;;|v+kn4RFCi8wd2%?%$*|{Q^SyPIf3Q2i^U(~Xp&3k%-g;!^tcMYZ`SE6 z1UCez(`Zy|K4kUy8^-V#)?n%t@26I1C;Cdr?B*X_0V~ufC-BFTVuLtlxuoeE0{q2Q z6}gQeL`J4a@Om})ujQVylk0GX3s~+$&Mj!C0qHt%u=9VU3tc&GgUNW*v^Ku5)5!8&ujCufI){Gx#HQX*R<&<`yA+(>1+0k|< zPXaS+6%Yj2g`j)S0$)o@^OzsU@^PTq_BR%u@82J8Q`b@O*)JuLY;W#AWG~9v?~p4G zaju3iJEYH@jTa&fybP7xwArlOf$|SIjYh2;7CKJw_H?yb5_a2~w7>N|dPqIq=x6b5 z(%>r_(4OjDIj*1pG5Z>Ao0c)rY(dAOU*N{@hYy}LKk%2o?#sF_NlRynFJF3FTxz=c z8}t(T>FP(zL+MrZhi{fVyX{C3*o2^(eR>$=k4YPR6xztTAyjiTzyEr{h|Sm2h+}}d zp^Elz38{&5FmvkAt+g+8;xpA|0cG46zZjEM$`llBtp(rwj8%Vkrc}JVQH=t+G1BPj z5)JwGEw2n?%q+qbNWRYCiwK73!Pklqql3l|jLau$mri^BINg!Us>`NTDQ89r`)OfX z=wRswi{16%2?Z~fQcY!()MR{+DrLy+SSww%ef4ULJD2P7(d7537q0o?6H~D*%3?0t}vHPqgL{_2XIFC5cCp{Hvx=`S|ZCcQY%M|N5pMwG})h)qdeDT*UR-64$#_lupw3DPqv?W!T zLZ=MC^U|m&d)CoRtt5=6%acs@mKzMwcD0ntU~uK;s5_D&V1etyy~Bm4Ek%hi^-a7v z{;|{&7<=>GOIywe7DA81_N`N4FM|-}5$76{Br=0{)dh>BLHLYo2ejct9vnkLj0E0K zyU=vb1eQKnR?6s2DfEE+6`lJC!;YewJ;7~~$Cp~Lee#`|8Kme5PZp)-KvFfon@Zsp zkWGz1KjOpCCYcbGv2|1<1KM9?UG*a$vaSx*r^d>^BL0@0wl7f8x2O3wqP_X*_;HL; zXWDgjg>N0V#II)~EHkEV>-fAYrX^U!rU zyUo5`!NlUCw+UC#i*(>e6y}Z}No8=olUT^MKb-VVmQys|D5cFw7xBztkjPp5F z%W?KK)*U5r{0nxu16b;Ir`#RQB44PV_#FY6xzZ$fn#|Hv+03HJg8j&1$1N$;qq~OS zdG1?8{T~Fg@8OHfPW7*EQG}{zpT6nVbGMk zJ02YB_jGBwO+kt9+Hu9&Uy;n{wcAmgg#qsS9WI8 z=Jw%dkz7Ka?z&<#6*#y8{Y;b8F8y+^{V@Beb#HMyc$z@%$yk3x6uUr3vRtonxc;-( z(WRxUS!^iat2lmyjFlG=9U!Q2x6U6;c|rLwar#Y3_4>g|B1!ieyrVfFq#jzUB!Jha zht|U*MUUsnrS+RC#}NaG*E5_9cc*H>tUCb=rpoTvj)zdUYx&>eDyHARTP(8=;4?Sw z!gg4szM#v68pJ(uqr4HQRF_SIfHa3ZLzR~ATKF-E2~{A!e)~~ofdoE}$r+NIau6P3zh*0s36Lfc&T@LoRFj z>c;2yw-7&w&q-b_E{8W2nFuemVCMZC;-wkBeLt{_N_d&=p6~{Eaxa;|8lFtma8cnC zog^uJZRlTYlhZTmgWCs3kVV3eiTr{8VZfqH4@!99aE#m<+7x??!cqa`^c80Ld!T zsKkY8+W{0iA%F=Ccd8%u1^aYvWQ*piUGMi0r9br+0GPCBm1v?%nhp{vyb_k2VYu()s=! z$B^_Xl$MjZ_W2PXY0z<(_hK+w8*~q$r zZ^@mhmzJc)6b1^F!if9Z=F)eIyqAGmZL`9t5MQF>n;KVpr}KELX!YCgxw0>Gd^uB= zAk=D0v;NGRQc1>5^-(OjmouXeR_phMWwUA0O&aXp%aWlu*p5^!)cjQ zKiVQhl0*58qo11fDb=jLHUpJE>2(qUwRc3ghrdbJ;^pHO>KA6tsXcly;Xg`)cN5m zlVRaHvK)w+V_kzU+fFCMxZ526Vq>eCR`1$5t~{2ZKX5c zYzsu|OfNK$)A-_^AWvuPhr`RV7tUh^ayPY9QE zB3!MVKQ>3pf>X3|@CoWT-eN##ETw7Zu}k;e-e-Gq3Ehp)v?rWJn)zYH25Id~O59m9 z>pSlltY;mfH9_Qiu2b9FCaj9J4SZdh(xw$yv$YT#2_5p!lzeHo+U|9V`e6W~q2hnY z^8T|@^51Mzr8Ae^5rY7RI-VN3tweGlDz&nQA5m8UARQ%<;cc`IU!7h?UhZ1~_}~71 z^Y0tLPO^rtno1#SaBuB%Hf=z%+7WSqx4`E8eWSQqudjikrid`S|34h+c4V! z8Ym>R*HyYd!*b_4sfz~5eb+f_!^%miie-Ok(+Q8#$2c2lR?{yguMz^8F@^ghx)ij9 zcu0@Aq|}UajL4C%4kL?!b|}iPgLm3IaPK^ifJS;+P%JF?X=3*y9Y5D-Q$JYHw2t~1 z8sT!w5oE<4|Jl%94dmK#U4a*luE?QeDa|`4&*PYD_PyUHs{5S+Q6?zbsdOF3Ti!{6 z;kwpeFsLxuKC?DPs4OC7zSn=mnJK^4#2O7c@Hc5~md4#}$l0DL%m2*V{poE|ekkXw z_QYUHjurYL7kQE%V`LeOtIHOx(-W%zF5DdXR-a*&)d=zhzw8G%+e@5bw{k}OP-ufq zi1rwcW(wdQ*^@w`YQ*TxyF42;qrX$UCKnItQ9Luu1*i?KO3c`RoLu~Q7whrBWI6O} zB{RL$t8HZ2(sL604YL}6y-bk9rT_#pkHIIiy6H29^?*c+s?$KLpcQC^2KHrxleP)v z=Y`I6+S7g^RO8AVE%939QY=SnAjOi-+xNN@shbNoa8p5plZa*cIL=eGbc56@H8<|` zI+zq*hH(>c#_6=dH7zuFsB!Alb7WeVPw~gqrbMx8y+27p4q~(B&kTWrsP5>U@#WDs zm`U4kf2~)S){NBY7VFq)z$EH5dDGixl9MTg8f(UL-BzCa$DB(CfOZYZsvJBbfFGtb zXP?fPpSyKt3KvG6B-QuRpKyshpcp4I6fHwQTkr6hgcsrzmB#?x$3tM*XCT`8s&9$|V9SYo;x%$_Vg1Q9xNBomhxX$jicp*W@p zq1V`Lqqe%2EolP}?$JMaOA_1xhPay-sDZpA!kU>KjMYD)O9k~NQtdn=Xc=~B`eoj* zS5bjzG!<^>b>$KCkt4$qYCRFQwFf%&TMs_^Edsh^rY~z1Ta;?HJaj2O%@M>Js<29N zpEp?hO&ok4wb+F(A8@CuxV4SjuJUci>ML5bFDT|$S4iz?s7o?53am;8|+H0vv} zcN3oFWB66?A$1%h|D+_dKFPxqh99u~))s6kZ{}{?p?&>A#`(_vOeoq~F!UuiS!<)U;3D}8Fexcb5eemZy}xhh z_X)UZd@G5DPs=%9knY0w`v3@r>+c(;x$&AOly?v(Pd)(#Vdoa-2QAAQf8Q|V&r4Da z{_)?PgMTIAWu?Gxko9KE5p4}HjSCWQ7Uyqnw54{B-*V+ zeZgCWVu*QSn9W-igeNl8Vc?g{q0;pR$mQ9eC6a@Jt4MogF-^qW8#fB`o}#k7Kj#}r zo01F;A3fY{9R9IjlTPXcITj*5u;ilUhbaeUi6;9^b4r|KONYO0dwO`0{abUsTGZ>? zX>T4``gMf-0uplQs2!jXb9n=)#Ex}J$y-o>xY~o_)kBj)SB`i{&kS~N>XPzC@r9ui zRn@oj-=9*1n49FoAU(~OO)|yXk*t!$IH6MBdI#P~?N{>}fO6w{^szBNDWDz~#4~zl z!r8OGdfYSfWwQT+vKvlUe&(Gwo(H4^72plgOd+LC5)dl0jVWXK2FcIJ)y~?Sc+FO3 zncK;8@7j`-w0-f)(ETaRZz^f*g&!x z7#W)_INzA%=KMfP!GbA2DmMTQut-5@`ts3B+Fvyf=MW;eE|ZYyRCr zdO3aFU;Q6;8BREwpKCN#M;0c*IC+I&m){2=oE+vET%)5a4H)(Sh=5i>+Iv})?57>_ zK$&;2`C^xpK0?Ik-UXRTVI0S+a0l5{isVK8d>1n-)w*l-8&> z$F#ItwH0yi=FG1$>g%-*i49*Io0qDAT!`1OYQ(r=3w%tcagu}F#aZM^67TFyv4kb% zpbdGNlz#hwD|mqB4tbHLJnkn|I}?YQOLVh&&BDiVBOdUm0H*yo>|s}y;J7eXPx?j1 znP*9`KB{`7uZNrF(}zIUUqpr)G#9x8=Pew7=n3Rkn^9_#t0CtWtOkGX<-E7i+vE zfSxF(&_=pKyh%6bolM>;IITnZx`q==FV6;!a z`9D}w72Gub;FwskU6F%kSDL{ahmKg4D?|nHyYr(t4P*9_+opSXiC| zIJsQVpx3q^Ri_{7$T@bXOV2RTu3E)fyEq_X8ilPeT(pqS7JJW=so9H8b6rQnob>** zoGkFO79L;6mzoSSqf@yRwP}DWnV*R_5eRJmqP#Vx8CWBh5aRGB6sH*YfTQa#2 zSvQ(Ip2a)+DSyq0*NA%DbkfIV7I892inZ8csH1<=k#ruEx66BdXTrl5h&eu$T)$!`9D{L^!EOmeA zv?nFAr=r|i`;u7j+TCcK0tzSPq zIg7sp&>elkQ1g%?afmmJoaGDO?ZA^;Rsaz8BK1tV3rc+p7O<#0!c+U`;MMrbA=x(P zeI-*X7cPr(pJQM_9FAzmmjFu&an{cAR zdnMw9fzH+E7wyuPgvb3qz-iC6qbq>qk`k7?-_?(XGj2)s3(+x`dh9A9dSf|Drq6ql zfi8uH|Cu@dpD88Z8m-VwSY(*9Y9k9Ek#RU=T1v zMwO+@$W7!8>X^~pON^;f?2E=O;_WEQ&zN!mM^Ipr>>T|Kq9*YK{Cti814uYNefC_O zfzT4oH>+B=_uMcX_?8};!AF_^r${d1-Gu-0*^7_SM2jwp+6ec#!g6XLjSj2_+U9 z-PKS$D;k+aQHLrCSoCI3sy1kTnK0;_7S|J`-hi|1Hl}Y+PYAL?>GXTMn$BB9gUS~(Jg`f@r?5cNZ{hZAGyGQ zs{_2PKyTA^bE(sWaNQ!Y~Fjv-Nwu z+y|E$j1Lf?-=JPAd=QafF&U_69F0XpEZj)zF-Y&-Zb0k_LRTvPovs=kdO9ujM9acNS-TOInU8bgur{g$5GA-iTnanvIu)~k2zoPKpVOi@R;TR`%q#yTISOOb zPw1?GA==!bX~tzA?P4TTvnwkHUJboX+iq(moa<8L!T7rDF<@Hw8#RZ?(dvUoiB(QA z&tehon`Tr=JbDUA_a8lc@gW05@yuKc%cNWr(r#t^qAaenQioI>b>q73g^CA7e9cck zNEQ?x9&jt^(j$bmW#UhE(z6>6*~gXSaq!r^J@7Kwm=M$BU{$7*tgi!%1{`ni9Rj8a zMI{RUXhqV_T4SaJiGD%XfLUQlHgYS znulmTrRMv^$iW;kTT;5$n09#Q5N-H#ae;-M%Li1W=B-q7vd`uBv;mvvAZ=jGnEa-4 zp7(Q>$(-iTTZ|la7x%tY6zBjtcRma!afJ)Lxen4$TW?EGv6SaXjX#bQ;i9Vl_@X8& z3KsDRm`q~0WX!puSoP1}HIT0j!z$~YIXneA6B(APDL-nZ@Bw) zo}E>_R4%cEW^?ih(uF|9Q-Gg_9NAkP70#_6pAdJZ>bbJJ@kHa0Aq36EVhS_tm=T^V zD2Nz$UJIVoeC0wfVCpw_*ebi3adukf{wzj+EPxzJ>59HNksnVkH51ktn#{S3T&GBq zU8{4B^ihyrxP~RqSL~Ypj@XeJj+(VMNZ_rwARwB6DPk`g2awlikc8XyUJ;GAElcL? zYmsBpGoAuAwyFHe{g(ABd)vFFWynI9eEXTpqyimAyQ*aNL6x%bK$0+e#mUfsc}O;! zHn;Grj~af$!-QJR(5?X)@6!C~y%gBE?ms+VF8?>6<^LbwyXooA*<(JohVtZBIM=G( zPw*9SQKfkvGEQ^c52PV^Sm_HA@PU&vQKZ2^P}893vYl4Bx^}Gj1{%Jshi4-OB7R8& zE-OqLgnM98_{fdlpu20oLFnz;-12Rpwvu_VBg_5sf7~~r)R5?P`fVk_Dhj|RG?tvt zF2Psk771ot*g^sjmpp_JJ^|1~OxaAdQjNv0#f6R}#dtm<16NSX1`w+!yTfrF@c(?V z{AUyJb+|8$=7|gf|2%*kfQR{rI)lGKT~j6n0DsN^aXIY0xqkH(dWks`hxlopWCOdx zOowEHj8Df_dfL|B1@T8sDu?KZpt9W(r-ItrbXA2>m4QUkSf`iwTjIPpZ;zBTYUh}3 zvW8mj{-3=QSowIW?TqT?vzZqodp@0S=YM3Y?tAr*?7ElV&;Fg)x6S+*yP?0BMZk}P z%F4{PeAeNnA6_4=y5F=%ZI{z|9*bAKa#|h^h(yc=YMn-YaI?^#z+M6;nM6 z=e$iaxZhe8E5i8Po_)*hzs4@t7GKd>mn-KRl_{8~lyO?~m;@7>`(8Qz%ldEk>HN6$ z?a`@c+R^isPt5a}85E;$9noWEyZR!qyDRy@cJ14x=eul#S7%f$(OwlB>oNTiQ|>pb z2eN-)KRn;BUQ$ij;_g4OAkO|)g?8_xDb$sLt z9O}{Bb7f0J$@^`$PwlFk%ex`+s+k${TS=8C38jBtFZ_|5UYhvfw#(K%+h*N*X2DUB z=G3>m@Imv5b=!dB$@80U<^G$wC_C_#Sh!ig+FFy2?Sdw$Qy%j$GArL(@Zni&rT&ri z^3tcZXXKP}Eb6knrxr3rL;)BMH_m5$xuwB;q{gsR{qSzi8tX^-Z1=Oc^R;LDIvqQh zCsv}aAbjI9@L-8`#}C}U{cHZWZA(|axc19->iLt2X}Q-duiB<4&#QRw`*(OXFw*=F zuAb}carxPYdCM2PnpSSM>8$y!r41c2ViJtM=hd}~zMM60KHJ)*cQdbh@8X^@OOt&w zkJMJ+U9|Vxx1JaLqr25k_PUMb%2&HWcW=M3Rn|z~{Cmn1ZGDM99)GGXuTT9bel)k- zr%hKr@WLMBWQ!fMst+X1df4u;`e&b!rOmhXNx&ZE`j{`R%IAnSVaLJ1=pL8sL;N>m6W-sDmVzUf>>T2t|y1suozdo-P zwApgDtkGXP+gDZHkuht#UusI-nK8rV?9`vm$t=Iqy5E}p{b&2Ks_SOV)~-x_>6nfg zN7hdJ>7C5*I}J2Z@Oi%Q%kAIgC2Qnsd8eM4R`$V9ZHLE<%wrhE>o$Ap?zt5_@_P$i|I&J&Lu8(HE#%4uNt9gWfRbQ)XE?xOY zH#f_I2{8@b?V*$U;hj;VW{@DJ=WS`02%X%i= z($QvHFE8BVk}3=C~ofZ v&{@zG)aKF~+LOK2lbM})!?%?ykhba~ZCnL!XdLCEY_uK?6Z~n#{{Kw?`8;_= literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/manifest.json b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/manifest.json new file mode 100644 index 000000000..1274bf598 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/manifest.json @@ -0,0 +1,325 @@ +{ + "version": "2.0", + "job_id": "job_232bb2ef34f5", + "data_id": null, + "source_file_name": "GOOG-10-K-2025.pdf", + "processing_date": "2026-06-16T03:34:57.577101Z", + "processing": { + "page_count": 99, + "billing_status": "charged", + "cost": { + "micro_dollars": 148500, + "credits": 0.1485 + }, + "timing": { + "started_at": "2026-06-16T03:29:57.593829+00:00", + "completed_at": "2026-06-16T03:34:57.538564+00:00", + "duration_ms": 299944 + }, + "stages": { + "timing_ms": { + "document.profile": 7552, + "pdf.extract.standard": 239927, + "toc.detect_range_llm": 1437, + "heading.hierarchy_llm_call": 17771, + "toc.parse_hierarchy_llm": 1267, + "md.detect_toc": 2774, + "heading.merge_pre_pass_llm": 2196, + "heading.hierarchy_llm": 16541, + "md.predict_headings": 20573, + "md.deferred_summaries": 27874, + "md.build_dataframe": 3, + "pdf.parse_md": 52091, + "document.cleanup_unreferenced_images": 3, + "document.compress_images": 0, + "worker.parse.document": 299579 + }, + "token_usage": { + "prompt_tokens": 15915, + "completion_tokens": 3407, + "total_tokens": 19322 + } + } + }, + "statistics": { + "total_chunks": 300, + "text_chunks": 226, + "image_chunks": 2, + "table_chunks": 72, + "total_pages": null + }, + "HIERARCHY": { + "Root": {}, + "PART I": { + "ITEM 1. BUSINESS": { + "Overview": { + "Access and Technology for Everyone": {}, + "Making AI Helpful for Everyone": {}, + "Moonshots": {}, + "Privacy and Security": {}, + "Google": {}, + "Google Services": { + "Serving Our Users": {}, + "How We Make Money": {}, + "Google Cloud": {}, + "Other Bets": {} + }, + "Competition": {}, + "Culture and Workforce": {}, + "Government Regulation": {}, + "Intellectual Property": {}, + "Available Information": {} + } + }, + "ITEM 1A. RISK FACTORS": { + "Risks Specific to our Company": {}, + "Risks Related to our Industry": {}, + "Risks Related to Laws, Regulations, and Policies": {}, + "Risks Related to Ownership of Our Stock": {}, + "General Risks": {} + }, + "ITEM 1B. UNRESOLVED STAFF COMMENTS": {}, + "ITEM 1C. CYBERSECURITY": {}, + "ITEM 2. PROPERTIES": {}, + "ITEM 3. LEGAL PROCEEDINGS": {}, + "ITEM 4. MINE SAFETY DISCLOSURES": {} + }, + "PART II": { + "Holders of Record": {}, + "Dividend Program": {}, + "Issuer Purchases of Equity Securities": {}, + "Stock Performance Graphs": { + "COMPARISON OF CUMULATIVE 5-YEAR TOTAL RETURN\\* ALPHABET INC. CLASS A COMMON STOCK": {}, + "COMPARISON OF CUMULATIVE 5-YEAR TOTAL RETURN\\* ALPHABET INC. CLASS C CAPITAL STOCK": {} + }, + "ITEM 6. [Reserved]": { + "Understanding Alphabet's Financial Results": {}, + "Trends in Our Business and Financial Effect": {}, + "Revenues and Monetization Metrics": { + "Google Services": { + "Google Advertising": {}, + "Google Subscriptions, Platforms, and Devices": {}, + "Google Cloud": {}, + "Other Bets": {} + } + }, + "Costs and Expenses": { + "Cost of Revenues": {}, + "Operating Expenses": {}, + "Other Income (Expense), Net": {}, + "Provision for Income Taxes": {} + }, + "Executive Overview": {}, + "Other Information:": {}, + "Financial Results": { + "Revenues": { + "Google Services": { + "YouTube ads": {}, + "Google Network": {} + }, + "Monetization Metrics": {}, + "Google Subscriptions, Platforms, and Devices": {}, + "Google Cloud": {}, + "Revenues by Geography": {} + } + }, + "Costs and Expenses_2": { + "Cost of Revenues": {}, + "Research and Development": {}, + "Sales and Marketing": {}, + "General and Administrative": {} + }, + "Segment Profitability": { + "Google Services": {}, + "Google Cloud": {}, + "Other Bets": {}, + "Other Income (Expense), Net": {}, + "Provision for Income Taxes": {} + }, + "Financial Condition": { + "Cash, Cash Equivalents, and Marketable Securities": {}, + "Sources, Uses of Cash and Related Trends": { + "Cash Provided by Operating Activities": {}, + "Cash Used in Investing Activities": {}, + "Cash Used in Financing Activities": {} + }, + "Liquidity and Material Cash Requirements": {}, + "Capital Expenditures and Leases": { + "Capital Expenditures": {}, + "Leases": {}, + "Financing": {}, + "Share Repurchase Program": {}, + "Dividend Program": {}, + "Accrued Legal and Regulatory": {}, + "Taxes": {}, + "Purchase Commitments and Other Contractual Obligations": {}, + "Pending Acquisitions": {} + } + }, + "Critical Accounting Estimates": { + "Fair Value Measurements of Non-Marketable Equity Securities": {}, + "Property and Equipment": {}, + "Income Taxes": {}, + "Loss Contingencies": {} + } + }, + "ITEM 7A. QUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK": { + "Foreign Currency Exchange Risk": {}, + "Interest Rate Risk": {}, + "Equity Investment Risk": {} + }, + "ITEM 8. FINANCIAL STATEMENTS AND SUPPLEMENTARY DATA": { + "INDEX TO CONSOLIDATED FINANCIAL STATEMENTS": { + "REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM": { + "Opinion on the Financial Statements": {}, + "Basis for Opinion": {}, + "Critical Audit Matter": {} + }, + "REPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM_2": { + "Opinion on Internal Control Over Financial Reporting": {}, + "Basis for Opinion": {}, + "Definition and Limitations of Internal Control Over Financial Reporting": {} + } + }, + "Alphabet Inc. NOTES TO CONSOLIDATED FINANCIAL STATEMENTS": { + "Note 1. Summary of Significant Accounting Policies": { + "Nature of Operations": {}, + "Basis of Consolidation": {}, + "Use of Estimates": {}, + "Revenue Recognition": { + "Google Advertising": {}, + "Google Subscriptions, Platforms, and Devices": {}, + "Google Cloud": {}, + "Arrangements with Multiple Performance Obligations": {}, + "Customer Incentives and Credits": {}, + "Sales Commissions": {} + }, + "Cost of Revenues": {}, + "Software Development Costs": {}, + "Stock-Based Compensation": {}, + "Advertising and Promotional Expenses": {}, + "Fair Value Measurements": {}, + "Financial Instruments": { + "Credit Risks": {}, + "Cash Equivalents": {}, + "Marketable Securities": {}, + "Non-Marketable Securities": {}, + "Derivative Financial Instruments": {}, + "Financial Guarantees": {}, + "Accounts Receivable": {}, + "Convertible Notes": {}, + "Other": {} + }, + "Impairment of Investments": {}, + "Inventory": {}, + "Variable Interest Entities": {}, + "Property and Equipment": {}, + "Goodwill": {}, + "Leases": {}, + "Impairment of Long-Lived Assets": {}, + "Income Taxes": {}, + "Business Combinations": {}, + "Foreign Currency": {}, + "Recently Issued Accounting Pronouncements Not Yet Adopted": {}, + "Recently Adopted Accounting Pronouncements": {}, + "Prior Period Reclassifications": {} + }, + "Note 2. Revenues": { + "Disaggregated Revenues": {}, + "Revenue Backlog": {}, + "Deferred Revenues": {} + }, + "Note 3. Financial Instruments": { + "Fair Value Measurements": { + "Investments Measured at Fair Value on a Recurring Basis": {}, + "Investments Measured at Fair Value on a Nonrecurring Basis": {} + }, + "Debt and Equity Securities": { + "Debt Securities": {}, + "Non-marketable Securities": {}, + "Gains and Losses on Equity Securities": {}, + "Derivative Financial Instruments": { + "Cash Flow Hedges": {}, + "Fair Value Hedges": {}, + "Net Investment Hedges": {}, + "Derivatives Not Designated as Hedging Instruments": {}, + "Offsetting of Derivatives": {} + } + } + }, + "Note 4. Leases": {}, + "Note 5. Variable Interest Entities": { + "Consolidated VIEs": {}, + "Unconsolidated VIEs": {} + }, + "Note 6. Debt": { + "Short-Term Debt": {}, + "Long-Term Debt": {}, + "Credit Facility": {} + }, + "Note 7. Supplemental Financial Statement Information": { + "Accounts Receivable": {}, + "Property and Equipment, Net": {}, + "Accrued Expenses and Other Current Liabilities": {}, + "Accumulated Other Comprehensive Income (Loss)": {}, + "Other Income (Expense), Net": {} + }, + "Note 8. Acquisitions": {}, + "Note 9. Goodwill": {}, + "Note 10. Commitments and Contingencies": { + "Commitments": {}, + "Financial Guarantees": {}, + "Indemnifications": {}, + "Legal Matters": { + "Antitrust Matters": {}, + "Privacy Matters": {}, + "Patent and Intellectual Property Claims": {}, + "Other": {} + }, + "Non-Income Taxes": {} + }, + "Note 11. Stockholders' Equity": { + "Share Repurchases": {}, + "Dividends": {} + }, + "Note 12. Net Income Per Share": {}, + "Note 13. Compensation Plans": { + "Stock Plans": {}, + "Stock-Based Compensation": {}, + "Stock-Based Award Activities": {} + }, + "Note 14. Income Taxes": { + "Deferred Income Taxes": {}, + "Uncertain Tax Positions": {} + }, + "Note 15. Information about Segments and Geographic Areas": {}, + "Note 16. Subsequent Event": {} + } + }, + "ITEM 9A. CONTROLS AND PROCEDURES": { + "Evaluation of Disclosure Controls and Procedures": {}, + "Changes in Internal Control over Financial Reporting": {}, + "Management's Report on Internal Control over Financial Reporting": {}, + "Limitations on Effectiveness of Controls and Procedures": {} + }, + "ITEM 9B. OTHER INFORMATION": {}, + "ITEM 9C. DISCLOSURE REGARDING FOREIGN JURISDICTIONS THAT PREVENT INSPECTIONS": {} + }, + "PART III": { + "ITEM 10. DIRECTORS, EXECUTIVE OFFICERS, AND CORPORATE GOVERNANCE": {}, + "ITEM 11. EXECUTIVE COMPENSATION": {}, + "ITEM 13. CERTAIN RELATIONSHIPS AND RELATED TRANSACTIONS, AND DIRECTOR INDEPENDENCE": {}, + "ITEM 14. PRINCIPAL ACCOUNTANT FEES AND SERVICES": {} + }, + "PART IV": { + "ITEM 15. EXHIBITS, FINANCIAL STATEMENT SCHEDULES": { + "1. Consolidated Financial Statements": {}, + "2. Financial Statement Schedules": {}, + "3. Exhibits": {} + }, + "ITEM 16. FORM 10-K SUMMARY": {}, + "SIGNATURES": {}, + "POWER OF ATTORNEY": {} + } + } +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-0 Stock & Note Listings.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-0 Stock & Note Listings.html new file mode 100644 index 000000000..59ced0ad2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-0 Stock & Note Listings.html @@ -0,0 +1 @@ +
Title of each classTrading Symbol(s)Name of each exchange on which registered
Class A Common Stock, $0.001 par valueGOOGLNasdaq Stock Market LLC(Nasdaq Global Select Market)
Class C Capital Stock, $0.001 par valueGOOGNasdaq Stock Market LLC(Nasdaq Global Select Market)
2.375% Senior Notes due 2028Nasdaq Stock Market LLC
2.500% Senior Notes due 2029Nasdaq Stock Market LLC
2.875% Senior Notes due 2031Nasdaq Stock Market LLC
3.000% Senior Notes due 2033Nasdaq Stock Market LLC
3.125% Senior Notes due 2034Nasdaq Stock Market LLC
3.375% Senior Notes due 2037Nasdaq Stock Market LLC
3.500% Senior Notes due 2038Nasdaq Stock Market LLC
4.000% Senior Notes due 2044Nasdaq Stock Market LLC
3.875% Senior Notes due 2045Nasdaq Stock Market LLC
4.000% Senior Notes due 2054Nasdaq Stock Market LLC
4.375% Senior Notes due 2064Nasdaq Stock Market LLC
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-1 Filer Status.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-1 Filer Status.html new file mode 100644 index 000000000..e3c5a6745 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-1 Filer Status.html @@ -0,0 +1 @@ +
Large accelerated filerAccelerated filer
Non-accelerated filerSmaller reporting company
Emerging growth company
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-10 R&D Expenses.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-10 R&D Expenses.html new file mode 100644 index 000000000..cfa884632 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-10 R&D Expenses.html @@ -0,0 +1 @@ +
Year Ended December 31,
20242025
Research and development expenses$ 49,326$ 61,087
Research and development expenses as a percentage of revenues14 %15 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-11 Sales Expenses.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-11 Sales Expenses.html new file mode 100644 index 000000000..63ad1199c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-11 Sales Expenses.html @@ -0,0 +1 @@ +
Year Ended December 31,
20242025
Sales and marketing expenses$ 27,808$ 28,693
Sales and marketing expenses as a percentage of revenues8 %7 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-12 Expense Analysis.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-12 Expense Analysis.html new file mode 100644 index 000000000..fc4a86baa --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-12 Expense Analysis.html @@ -0,0 +1 @@ +
Year Ended December 31,
20242025
General and administrative expenses$ 14,188$ 21,482
General and administrative expenses as a percentage of revenues4 %5 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-13 Alphabet Income.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-13 Alphabet Income.html new file mode 100644 index 000000000..868a6d8c0 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-13 Alphabet Income.html @@ -0,0 +1 @@ +
Year Ended December 31,
20242025
Operating income (loss):
Google Services$ 121,263$ 139,404
Google Cloud6,11213,910
Other Bets(4,444)(7,515)
Alphabet-level activities(1)(10,541)(16,760)
Total income from operations$ 112,390$ 129,039
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-14 Other Income_Expense.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-14 Other Income_Expense.html new file mode 100644 index 000000000..bb31c5283 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-14 Other Income_Expense.html @@ -0,0 +1 @@ +
Year Ended December 31,
20242025
Interest income$ 4,482$ 4,337
Interest expense(268)(736)
Foreign currency exchange gain (loss), net(409)(382)
Gain (loss) on debt securities, net(1,043)540
Gain (loss) on equity securities, net3,71424,080
Income (loss) and impairment from equity method investments, net(188)281
Other1,1371,667
Other income (expense), net$ 7,425$ 29,787
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-15 Tax Data Table.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-15 Tax Data Table.html new file mode 100644 index 000000000..bcd3f80a7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-15 Tax Data Table.html @@ -0,0 +1 @@ +
Year Ended December 31,
20242025
Income before provision for income taxes$ 119,815$ 158,826
Provision for income taxes$ 19,697$ 26,656
Effective tax rate16.4 %16.8 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-16 Cash Flow Table.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-16 Cash Flow Table.html new file mode 100644 index 000000000..c287eedea --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-16 Cash Flow Table.html @@ -0,0 +1 @@ +
Year Ended December 31,
20242025
Net cash provided by operating activities$ 125,299$ 164,713
Net cash used in investing activities$ (45,536)$ (120,291)
Net cash used in financing activities$ (79,733)$ (37,388)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-17 Foreign Currency Risk.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-17 Foreign Currency Risk.html new file mode 100644 index 000000000..330a135df --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-17 Foreign Currency Risk.html @@ -0,0 +1 @@ +
ImpactAs of December 31,
20242025
Foreign currency risk
Foreign denominated monetary assets and liabilities(1)OI&E$ 135$ 671
Cash flow hedges of foreign currency revenue(2)AOCI$ 1,627$ 2,096
Net investment hedges of investments in foreign subsidiaries(3)AOCI$ 660$ 2,942
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-18 Interest Rate Risk.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-18 Interest Rate Risk.html new file mode 100644 index 000000000..2c826b330 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-18 Interest Rate Risk.html @@ -0,0 +1 @@ +
As of December 31,12-Month Average As of December 31,
2024202520242025
Risk category - interest rate$ 208$ 162$ 230$ 184
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-19 Financial Index.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-19 Financial Index.html new file mode 100644 index 000000000..3007ebe4e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-19 Financial Index.html @@ -0,0 +1 @@ +
Page
Reports of Independent Registered Public Accounting Firm (PCAOB ID: 42)45
Financial Statements:
Consolidated Balance Sheets48
Consolidated Statements of Income49
Consolidated Statements of Comprehensive Income50
Consolidated Statements of Stockholders’ Equity51
Consolidated Statements of Cash Flows52
Notes to Consolidated Financial Statements53
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-2 Share Repurchase.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-2 Share Repurchase.html new file mode 100644 index 000000000..cbc7b2a2e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-2 Share Repurchase.html @@ -0,0 +1 @@ +
PeriodTotal Number of Class A Shares Purchased (in thousands)(1)Total Number of Class C Shares Purchased (in thousands)(1)Average Price Paid per Class A Share(2)Average Price Paid per Class C Share(2)Total Number of Shares Purchased as Part of Publicly Announced Programs (in thousands)(1)Approximate Dollar Value of Shares that May Yet Be Purchased Under the Program (in millions)
October 1 - 3184113,305$ 253.20$ 255.0314,146$ 71,231
November 1 - 302095,885$ 286.00$ 281.466,094$ 69,503
December 1 - 3100$ 0.00$ 0.000$ 69,503
Total1,05019,19020,240
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-20 Loss Contingencies.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-20 Loss Contingencies.html new file mode 100644 index 000000000..188c3231c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-20 Loss Contingencies.html @@ -0,0 +1 @@ +
Description of the MatterThe Company is subject to claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders. As described in Note 10 to the consolidated financial statements, such claims, lawsuits, regulatory and government inquiries and investigations, other proceedings, and consent orders could result in adverse consequences.Significant judgment is required to determine both the likelihood and the estimated amount of a loss related to such matters. Auditing management’s accounting for and disclosure of loss contingencies from these matters involved challenging and subjective auditor judgment in assessing the Company’s evaluation of the probability of a loss, and the estimated amount or range of loss.
How We Addressed the Matter in Our AuditWe tested relevant controls over the identified risks associated with management’s accounting for and disclosure of these matters. This included controls over management’s assessment of the probability of incurrence of a loss and whether the loss or range of loss was reasonably estimable and the development of related disclosures.Our audit procedures included, among others, gaining an understanding of previous rulings and the status of ongoing lawsuits, reviewing letters from internal and external legal counsel addressing the matters, meeting with internal legal counsel to discuss the allegations, and obtaining a representation letter from management on these matters. We also evaluated the Company’s disclosures in relation to these matters.
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-21 Balance Sheet.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-21 Balance Sheet.html new file mode 100644 index 000000000..ddd4da740 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-21 Balance Sheet.html @@ -0,0 +1 @@ +
As of December 31,
20242025
Assets
Current assets:
Cash and cash equivalents$ 23,466$ 30,708
Marketable securities72,19196,135
Total cash, cash equivalents, and marketable securities95,657126,843
Accounts receivable, net52,34062,886
Other current assets15,71416,309
Total current assets163,711206,038
Non-marketable securities37,98268,687
Deferred income taxes17,1809,113
Property and equipment, net171,036246,597
Operating lease assets13,58815,221
Goodwill31,88533,380
Other non-current assets14,87416,245
Total assets$ 450,256$ 595,281
Liabilities and Stockholders’ Equity
Current liabilities:
Accounts payable$ 7,987$ 12,200
Accrued compensation and benefits15,06917,546
Accrued expenses and other current liabilities51,22855,557
Accrued revenue share9,80210,864
Deferred revenue5,0366,578
Total current liabilities89,122102,745
Long-term debt10,88346,547
Income taxes payable, non-current8,7829,531
Operating lease liabilities11,69112,744
Other long-term liabilities4,6948,449
Total liabilities125,172180,016
Commitments and Contingencies (Note 10)
Stockholders’ equity:
Preferred stock, $0.001 par value per share, 100 shares authorized; no shares issued and outstanding00
Class A, Class B, and Class C stock and additional paid-in capital, $0.001 par value per share: 300,000 shares authorized (Class A 180,000, Class B 60,000, Class C 60,000); 12,211 (Class A 5,835, Class B 861, Class C 5,515) and 12,088 (Class A 5,822, Class B 837, Class C 5,429) shares issued and outstanding84,80093,126
Accumulated other comprehensive income (loss)(4,800)(1,916)
Retained earnings245,084324,055
Total stockholders’ equity325,084415,265
Total liabilities and stockholders’ equity$ 450,256$ 595,281
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-22 Financial Data.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-22 Financial Data.html new file mode 100644 index 000000000..5d4bee350 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-22 Financial Data.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Revenues$ 307,394$ 350,018$ 402,836
Costs and expenses:
Cost of revenues133,332146,306162,535
Research and development45,42749,32661,087
Sales and marketing27,91727,80828,693
General and administrative16,42514,18821,482
Total costs and expenses223,101237,628273,797
Income from operations84,293112,390129,039
Other income (expense), net1,4247,42529,787
Income before income taxes85,717119,815158,826
Provision for income taxes11,92219,69726,656
Net income$ 73,795$ 100,118$ 132,170
Basic net income per share (Note 12)$ 5.84$ 8.13$ 10.91
Diluted net income per share (Note 12)$ 5.80$ 8.04$ 10.81
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-23 Comprehensive Income.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-23 Comprehensive Income.html new file mode 100644 index 000000000..7164d268e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-23 Comprehensive Income.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Net income$ 73,795$ 100,118$ 132,170
Other comprehensive income (loss):
Change in foreign currency translation adjustment, net of income tax benefit (expense) of $63, $(49) and $180735(1,673)2,522
Available-for-sale investments:
Change in net unrealized gains (losses)1,344(116)1,146
Less: reclassification adjustment for net (gains) losses included in net income1,168782(169)
Net change, net of income tax benefit (expense) of $(698), $(190), and $(276)2,512666977
Cash flow hedges:
Change in net unrealized gains (losses)168775(779)
Less: reclassification adjustment for net (gains) losses included in net income(214)(166)164
Net change, net of income tax benefit (expense) of $2, $(151), and $174(46)609(615)
Other comprehensive income (loss)3,201(398)2,884
Comprehensive income$ 76,996$ 99,720$ 135,054
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-24 Stockholders' Equity.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-24 Stockholders' Equity.html new file mode 100644 index 000000000..cf84b5ed6 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-24 Stockholders' Equity.html @@ -0,0 +1 @@ +
Class A, Class B, Class C Stock and Additional Paid-In CapitalAccumulated Other Comprehensive Income (Loss)Retained EarningsTotal Stockholders' Equity
SharesAmount
Balance as of December 31, 202212,849$ 68,184$ (7,603)$ 195,563$ 256,144
Stock issued1390000
Stock-based compensation022,5780022,578
Tax withholding related to vesting of restricted stock units and other0(10,164)09(10,155)
Repurchases of stock(528)(4,064)0(58,120)(62,184)
Net income00073,79573,795
Other comprehensive income (loss)003,20103,201
Balance as of December 31, 202312,46076,534(4,402)211,247283,379
Stock issued1300000
Stock-based compensation022,9370022,937
Tax withholding related to vesting of restricted stock units and other0(12,507)0(16)(12,523)
Repurchases of stock(379)(3,359)0(58,688)(62,047)
Dividends and dividend equivalents declared ($0.60 per share)0410(7,577)(7,536)
Sale of interest in consolidated entities01,154001,154
Net income000100,118100,118
Other comprehensive income (loss)00(398)0(398)
Balance as of December 31, 202412,21184,800(4,800)245,084325,084
Stock issued1170000
Stock-based compensation025,1300025,130
Tax withholding related to vesting of restricted stock units and other0(14,842)00(14,842)
Repurchases of stock(240)(2,514)0(42,884)(45,398)
Dividends and dividend equivalents declared ($0.83 per share)01520(10,315)(10,163)
Sale of interest in consolidated entities040000400
Net income000132,170132,170
Other comprehensive income (loss)002,88402,884
Balance as of December 31, 202512,088$ 93,126$ (1,916)$ 324,055$ 415,265
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-25 Cash Flow Statement.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-25 Cash Flow Statement.html new file mode 100644 index 000000000..138e6818a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-25 Cash Flow Statement.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Operating activities
Net income$ 73,795$ 100,118$ 132,170
Adjustments:
Depreciation of property and equipment11,94615,31121,136
Stock-based compensation expense22,46022,78524,953
Deferred income taxes(7,763)(5,257)8,348
Loss (gain) on debt and equity securities, net823(2,671)(24,620)
Other4,3303,4192,108
Changes in assets and liabilities, net of effects of acquisitions:
Accounts receivable, net(7,833)(5,891)(8,779)
Income taxes, net523(2,418)(3,226)
Other assets(2,143)(1,397)(4,542)
Accounts payable664359907
Accrued expenses and other liabilities3,937(1,161)12,939
Accrued revenue share4821,059899
Deferred revenue5251,0432,420
Net cash provided by operating activities101,746125,299164,713
Investing activities
Purchases of property and equipment(32,251)(52,535)(91,447)
Purchases of marketable securities(77,858)(86,679)(103,773)
Maturities and sales of marketable securities86,672103,42883,240
Purchases of non-marketable securities(3,027)(5,034)(5,716)
Maturities and sales of non-marketable securities9478821,367
Acquisitions, net of cash acquired, and purchases of intangible assets(495)(2,931)(1,592)
Other investing activities(1,051)(2,667)(2,370)
Net cash used in investing activities(27,063)(45,536)(120,291)
Financing activities
Net payments related to stock-based award activities(9,837)(12,190)(14,167)
Repurchases of stock(61,504)(62,222)(45,709)
Dividend payments0(7,363)(10,049)
Proceeds from issuance of debt, net of costs10,79013,58964,564
Repayments of debt(11,550)(12,701)(32,427)
Proceeds from sale of interest in consolidated entities, net81,154400
Net cash used in financing activities(72,093)(79,733)(37,388)
Effect of exchange rate changes on cash and cash equivalents(421)(612)208
Net increase (decrease) in cash and cash equivalents2,169(582)7,242
Cash and cash equivalents at beginning of period21,87924,04823,466
Cash and cash equivalents at end of period$ 24,048$ 23,466$ 30,708
Supplemental disclosures of non-cash investing activities:
Purchases of property and equipment included in accrued liabilities and accounts payable$ 7,435$ 10,326$ 15,090
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-26 Google Revenue.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-26 Google Revenue.html new file mode 100644 index 000000000..09a64affb --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-26 Google Revenue.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Google Search & other$ 175,033$ 198,084$ 224,532
YouTube ads31,51036,14740,367
Google Network31,31230,35929,792
Google advertising237,855264,590294,691
Google subscriptions, platforms, and devices34,68840,34048,030
Google Services total272,543304,930342,721
Google Cloud33,08843,22958,705
Other Bets1,5271,6481,537
Hedging gains (losses)236211(127)
Total revenues$ 307,394$ 350,018$ 402,836
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-27 Revenue Breakdown.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-27 Revenue Breakdown.html new file mode 100644 index 000000000..8a8a7aace --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-27 Revenue Breakdown.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
United States$ 146,28647 %$ 170,44749 %$ 194,22948 %
$EMEA^{(1)}$ 91,03830102,12729117,15229
$APAC^{(1)}$ 51,5141756,8151667,68017
Other $Americas^{(1)}$ 18,320620,418623,9026
Hedging gains (losses)23602110(127)0
Total revenues$ 307,394100 %$ 350,018100 %$ 402,836100 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-28 Fair Value Table.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-28 Fair Value Table.html new file mode 100644 index 000000000..56b10df24 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-28 Fair Value Table.html @@ -0,0 +1 @@ +
As of December 31, 2024
Quoted Prices in Active Markets for Identical Assets (Level 1)Significant Other Observable Inputs (Level 2)Total
Cash$ 12,407
Cash equivalents:
Money market funds$ 8,154$ 0$ 8,154
Time deposits02,0812,081
Government bonds0746746
Corporate debt securities07878
Total cash and cash equivalents8,1542,90523,466
Marketable securities:
Marketable equity $securities^{(1)}$ 4,7081054,813
Time deposits0136136
Government bonds028,70928,709
Corporate debt securities021,11621,116
Mortgage-backed and asset-backed securities017,41717,417
Total marketable securities4,70867,48372,191
Total$ 12,862$ 70,388$ 95,657
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-29 Fair Value Table.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-29 Fair Value Table.html new file mode 100644 index 000000000..a993d018e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-29 Fair Value Table.html @@ -0,0 +1 @@ +
As of December 31, 2025
Quoted Prices in Active Markets for Identical Assets (Level 1)Significant Other Observable Inputs (Level 2)Total
Cash$ 15,305
Cash equivalents:
Money market funds$ 11,349$ 0$ 11,349
Time deposits03,3533,353
Government bonds0602602
Corporate debt securities09999
Total cash and cash equivalents11,3494,05430,708
Marketable securities:
Marketable equity securities4,4021,9116,313
Time deposits000
Government bonds050,54950,549
Corporate debt securities021,56521,565
Mortgage-backed and asset-backed securities017,70817,708
Total marketable securities4,40291,73396,135
Total$ 15,751$ 95,787$ 126,843
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-3 Alphabet Stock vs Indexes.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-3 Alphabet Stock vs Indexes.html new file mode 100644 index 000000000..257ac772a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-3 Alphabet Stock vs Indexes.html @@ -0,0 +1,160 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
DateAlphabet Inc. Class AS&P 500NASDAQ CompositeRDG Internet Composite
12/20$100$100$100$100
3/21$120$110$110$105
6/21$140$115$115$110
9/21$150$120$120$105
12/21$165$130$125$100
3/22$160$125$120$95
6/22$140$110$105$85
9/22$120$100$95$75
12/22$105$95$85$65
3/23$110$105$95$75
6/23$130$120$105$85
9/23$145$125$110$90
12/23$160$135$120$95
3/24$170$145$130$100
6/24$205$155$140$105
9/24$190$165$145$110
12/24$215$170$155$115
3/25$180$160$140$105
6/25$200$175$160$120
9/25$250$185$175$130
12/25$360$195$185$135
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-30 Debt Maturity.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-30 Debt Maturity.html new file mode 100644 index 000000000..f1e3717b1 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-30 Debt Maturity.html @@ -0,0 +1 @@ +
As of December 31, 2025
Due in 1 year or less$ 26,735
Due in 1 year through 5 years37,001
Due in 5 years through 10 years12,769
Due after 10 years13,317
Total$ 89,822
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-31 Investment Summary.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-31 Investment Summary.html new file mode 100644 index 000000000..d7a0ede29 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-31 Investment Summary.html @@ -0,0 +1 @@ +
As of December 31, 2024
Adjusted CostGross Unrealized GainsGross Unrealized LossesFair Value
Time deposits$ 2,217$ 0$ 0$ 2,217
Government bonds27,55183(214)27,420
Corporate debt securities18,30079(222)18,157
Mortgage-backed and asset-backed securities14,43763(385)14,115
Total investments with fair value change reflected in other comprehensive income$ 62,505$ 225$ (821)$ 61,909
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-32 Investment Summary.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-32 Investment Summary.html new file mode 100644 index 000000000..2f180fd1e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-32 Investment Summary.html @@ -0,0 +1 @@ +
As of December 31, 2025
Adjusted CostGross Unrealized GainsGross Unrealized LossesFair Value
Time deposits$ 3,353$ 0$ 0$ 3,353
Government bonds49,087443(26)49,504
Corporate debt securities18,346242(32)18,556
Mortgage-backed and asset-backed securities14,337174(128)14,383
Total investments with fair value change reflected in other comprehensive income$ 85,123$ 859$ (186)$ 85,796
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-33 Investment Securities.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-33 Investment Securities.html new file mode 100644 index 000000000..7ffa9c96d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-33 Investment Securities.html @@ -0,0 +1 @@ +
As of December 31, 2024
Less than 12 Months12 Months or GreaterTotal
Fair ValueUnrealized LossFair ValueUnrealized LossFair ValueUnrealized Loss
Government bonds$ 11,119$ (126)$ 2,576$ (88)$ 13,695$ (214)
Corporate debt securities4,228(17)6,838(168)11,066(185)
Mortgage-backed and asset-backed securities5,222(106)3,813(279)9,035(385)
Total$ 20,569$ (249)$ 13,227$ (535)$ 33,796$ (784)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-34 Investment Securities.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-34 Investment Securities.html new file mode 100644 index 000000000..4e0f01f7c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-34 Investment Securities.html @@ -0,0 +1 @@ +
As of December 31, 2025
Less than 12 Months12 Months or GreaterTotal
Fair ValueUnrealized LossFair ValueUnrealized LossFair ValueUnrealized Loss
Government bonds$ 4,230$ (9)$ 1,174$ (17)$ 5,404$ (26)
Corporate debt securities91502,429(24)3,344(24)
Mortgage-backed and asset-backed securities1,377(4)3,035(124)4,412(128)
Total$ 6,522$ (13)$ 6,638$ (165)$ 13,160$ (178)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-35 Debt Securities P&L.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-35 Debt Securities P&L.html new file mode 100644 index 000000000..d6e86f01c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-35 Debt Securities P&L.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Unrealized gain (loss) on fair value option debt securities$ 386$ 30$ 254
Gross realized gain on debt securities182482572
Gross realized loss on debt securities(1,833)(1,553)(316)
(Increase) decrease in allowance for credit losses50(2)30
Total gain (loss) on debt securities recognized in other income (expense), net$ (1,215)$ (1,043)$ 540
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-36 Non-marketable securities.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-36 Non-marketable securities.html new file mode 100644 index 000000000..99457d79c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-36 Non-marketable securities.html @@ -0,0 +1 @@ +
As of December 31,
20242025
Non-marketable securities:
Total initial cost of non-marketable equity securities accounted for under the measurement alternative$ 20,940$ 28,429
Cumulative upward adjustments22,70944,485
Cumulative downward adjustments (including impairments)(8,431)(8,820)
Carrying value of non-marketable equity securities accounted for under the measurement alternative35,21864,094
Equity method investments and other2,7644,593
Total non-marketable securities$ 37,982$ 68,687
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-37 Equity Securities.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-37 Equity Securities.html new file mode 100644 index 000000000..16d8453a7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-37 Equity Securities.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Gross unrealized gain on non-marketable equity securities accounted for under the measurement alternative$ 1,806$ 5,582$ 22,666
Gross unrealized loss (including impairments) on non-marketable equity securities accounted for under the measurement alternative(2,894)(2,210)(1,271)
Unrealized net gain (loss) on non-marketable equity securities accounted for under the measurement alternative(1,088)3,37221,395
Unrealized net gain (loss) on marketable and other equity securities7901561,907
Realized net gain (loss) on marketable and non-marketable equity securities sold during the period690186778
Total gain (loss) on equity securities in other income (expense), net (1)$ 392$ 3,714$ 24,080
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-38 Derivatives Table.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-38 Derivatives Table.html new file mode 100644 index 000000000..6490cd063 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-38 Derivatives Table.html @@ -0,0 +1 @@ +
As of December 31,
20242025
Derivatives designated as hedging instruments:
Foreign exchange contracts
Cash flow hedges$ 20,315$ 23,852
Fair value hedges$ 1,562$ 0
Net investment hedges$ 6,986$ 14,203
Derivatives not designated as hedging instruments:
Foreign exchange contracts$ 44,227$ 56,085
Credit derivatives(1)$ 0$ 16,940
Other contracts$ 15,082$ 15,900
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-39 Derivatives Table.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-39 Derivatives Table.html new file mode 100644 index 000000000..288219e02 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-39 Derivatives Table.html @@ -0,0 +1 @@ +
As of December 31, 2024As of December 31, 2025
Assets(1)Liabilities(2)Assets(1)Liabilities(2)
Derivatives designated as hedging instruments:
Foreign exchange contracts$ 1,054$ 0$ 316$ 197
Derivatives not designated as hedging instruments:
Foreign exchange contracts2005939284
Other contracts4741932498
Total derivatives not designated as hedging instruments674612416182
Total$ 1,728$ 612$ 732$ 379
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-4 Alphabet vs S&P 500.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-4 Alphabet vs S&P 500.html new file mode 100644 index 000000000..4583cf851 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-4 Alphabet vs S&P 500.html @@ -0,0 +1,160 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
DateAlphabet Inc. Class CS&P 500NASDAQ CompositeRDG Internet Composite
12/20$100$100$100$100
3/21$120$110$110$105
6/21$140$115$115$110
9/21$150$120$120$105
12/21$160$130$125$100
3/22$165$135$125$95
6/22$150$130$120$85
9/22$130$110$105$75
12/22$105$105$95$60
3/23$110$115$105$70
6/23$130$125$115$80
9/23$145$130$120$85
12/23$160$140$130$90
3/24$170$150$140$95
6/24$210$160$150$105
9/24$190$165$155$110
12/24$220$170$160$115
3/25$180$165$155$105
6/25$200$175$165$115
9/25$250$185$175$125
12/25$360$195$185$135
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-40 Hedge Gains_Losses.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-40 Hedge Gains_Losses.html new file mode 100644 index 000000000..2c9068376 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-40 Hedge Gains_Losses.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Cash flow hedging relationship:
Foreign exchange and other contracts
Amount included in the assessment of effectiveness$ 90$ 857$ (978)
Amount excluded from the assessment of effectiveness8477(45)
Net investment hedging relationship:
Amounts included in the assessment of effectiveness
Foreign exchange contracts(287)223(765)
Foreign currency-denominated debt00(393)
Amounts excluded from the assessment of effectiveness
Foreign exchange contracts0011
Total$ (113)$ 1,157$ (2,170)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-41 Hedge Effects.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-41 Hedge Effects.html new file mode 100644 index 000000000..84242824e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-41 Hedge Effects.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
RevenuesOther income (expense), netRevenuesOther income (expense), netRevenuesOther income (expense), net
Total amounts included on the Consolidated Statements of Income$ 307,394$ 1,424$ 350,018$ 7,425$ 402,836$ 29,787
Effect of cash flow hedges:
Foreign exchange contracts
Amount included in the assessment of effectiveness21301740(233)0
Amount excluded from the assessment of effectiveness2403701070
Effect of fair value hedges:
Foreign exchange contracts
Hedged items0590(59)0(9)
Amount included in the assessment of effectiveness0(59)05809
Amount excluded from the assessment of effectiveness01501301
Effect of net investment hedges:
Foreign exchange contracts
Amount excluded from the assessment of effectiveness018701370189
Effect of non-designated hedges:
Foreign exchange contracts0703350445
Other contracts05301740(148)
Total gains (losses)$ 237$ 262$ 211$ 658$ (126)$ 487
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-42 Derivatives Offsetting.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-42 Derivatives Offsetting.html new file mode 100644 index 000000000..05bdf4199 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-42 Derivatives Offsetting.html @@ -0,0 +1 @@ +
As of December 31, 2024
Gross Amounts RecognizedGross Amounts Offset in the Consolidated Balance SheetsNet Amounts Presented in the Consolidated Balance SheetsFinancial Instruments(1)Cash and Non-Cash Collateral Received or PledgedNet Amounts
Derivatives assets$ 1,776$ (48)$ 1,728$ (516)$ (721)$ 491
Derivatives liabilities$ 660$ (48)$ 612$ (516)$ (9)$ 87
As of December 31, 2025
Gross Amounts RecognizedGross Amounts Offset in the Consolidated Balance SheetsNet Amounts Presented in the Consolidated Balance SheetsFinancial Instruments(1)Cash and Non-Cash Collateral Received or PledgedNet Amounts
Derivatives assets$ 842$ (110)$ 732$ (140)$ (231)$ 361
Derivatives liabilities$ 489$ (110)$ 379$ (140)$ (15)$ 224
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-43 Lease Cost Table.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-43 Lease Cost Table.html new file mode 100644 index 000000000..4d122eb67 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-43 Lease Cost Table.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Operating lease cost$ 3,362$ 3,304$ 3,345
Finance lease cost:
Amortization of lease assets469413553
Interest on lease liabilities353165
Finance lease cost504444618
Variable lease cost1,1821,4251,739
Total lease cost$ 5,048$ 5,173$ 5,702
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-44 Lease Data.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-44 Lease Data.html new file mode 100644 index 000000000..33e8e52ad --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-44 Lease Data.html @@ -0,0 +1 @@ +
December 31,
20242025
Weighted-average remaining lease term:
Operating leases7.8 years7.6 years
Finance leases10.4 years8.3 years
Weighted-average discount rate:
Operating leases3.4 %3.6 %
Finance leases2.8 %3.1 %
December 31,
20242025
Operating leases:
Operating lease assets$ 13,588$ 15,221
Accrued expenses and other liabilities$ 2,887$ 3,209
Operating lease liabilities11,69112,744
Total operating lease liabilities$ 14,578$ 15,954
Finance leases:
Property and equipment, at cost$ 4,622$ 6,822
Accumulated depreciation(2,037)(2,025)
Property and equipment, net$ 2,585$ 4,797
Accrued expenses and other liabilities$ 235$ 441
Other long-term liabilities1,4422,059
Total finance lease liabilities$ 1,677$ 2,500
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-45 Lease Cash Flows.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-45 Lease Cash Flows.html new file mode 100644 index 000000000..f60465bec --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-45 Lease Cash Flows.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Cash payments for lease liabilities:
Operating cash flows used for operating leases$ 3,173$ 3,425$ 3,370
Operating cash flows used for finance leases$ 35$ 31$ 65
Financing cash flows used for finance leases(1)$ 705$ 405$ 1,988
Assets obtained in exchange for lease liabilities:
Operating leases$ 2,877$ 2,510$ 4,070
Finance leases$ 564$ 313$ 1,606
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-46 Lease Payments.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-46 Lease Payments.html new file mode 100644 index 000000000..944a31bd6 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-46 Lease Payments.html @@ -0,0 +1 @@ +
Operating LeasesFinance Leases
2026$ 3,275$ 491
20273,082345
20282,510335
20292,061314
20301,669241
Thereafter5,6541,143
Total undiscounted lease payments18,2512,869
Less: imputed interest(2,297)(369)
Total lease liability balance$ 15,954$ 2,500
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-47 Debt Table.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-47 Debt Table.html new file mode 100644 index 000000000..00745dc67 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-47 Debt Table.html @@ -0,0 +1 @@ +
MaturityCoupon RateEffective Interest RateAs of December 31,
20242025
Debt
2016 US dollar notes20262.00%2.23%$ 2,000$ 2,000
2020 US dollar notes2027 - 20600.80% - 2.25%0.93% - 2.33%10,0009,000
$2025 \text{ US dollar notes}^{(1)}$ 2028 - 20753.88% - 5.70%4.00% - 5.79%022,500
$2025 \text{ Euro notes}^{(2)}$ 2028 - 20642.38% - 4.38%2.57% - 4.51%015,585
Total face value of long-term debt12,00049,085
Unamortized discount and debt issuance costs $^{(2)}$ (118)(542)
Less: current portion of long-term notes $^{(3)}$ (999)(1,996)
Total long-term debt$ 10,883$ 46,547
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-48 Debt Maturity.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-48 Debt Maturity.html new file mode 100644 index 000000000..38cc89a30 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-48 Debt Maturity.html @@ -0,0 +1 @@ +
2026$ 2,000
20271,000
20282,676
20291,764
20305,500
Thereafter36,145
Total$ 49,085
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-49 Property & Equipment.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-49 Property & Equipment.html new file mode 100644 index 000000000..78a461043 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-49 Property & Equipment.html @@ -0,0 +1 @@ +
As of December 31,
20242025
Technical infrastructure(1)$ 141,852$ 203,679
Office space45,40348,348
Corporate and other assets12,57414,463
Property and equipment, in service199,829266,490
Less: accumulated depreciation(79,390)(98,485)
Add: assets not yet in service50,59778,592
Property and equipment, net$ 171,036$ 246,597
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-5 Financial Results.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-5 Financial Results.html new file mode 100644 index 000000000..0b0319c46 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-5 Financial Results.html @@ -0,0 +1 @@ +
Year Ended December 31,$ Change% Change
20242025
Consolidated revenues$ 350,018$ 402,836$ 52,81815 %
Cost of revenues$ 146,306$ 162,535$ 16,22911 %
Operating expenses$ 91,322$ 111,262$ 19,94022 %
Operating income$ 112,390$ 129,039$ 16,64915 %
Operating margin32 %32 %0 %
Other income (expense), net$ 7,425$ 29,787$ 22,362301 %
Net income$ 100,118$ 132,170$ 32,05232 %
Diluted net income per share(1)$ 8.04$ 10.81$ 2.7734 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-50 Accrued Liabilities.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-50 Accrued Liabilities.html new file mode 100644 index 000000000..bf30657ed --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-50 Accrued Liabilities.html @@ -0,0 +1 @@ +
As of December 31,
20242025
Accrued fines and settlements(1)$ 9,830$ 15,594
Accrued purchases of property and equipment7,1048,877
Accrued customer liabilities4,3045,029
Payables to brokers for unsettled investment trades3,866950
Income taxes payable, net2,905523
Other accrued expenses and current liabilities23,21924,584
Accrued expenses and other current liabilities$ 51,228$ 55,557
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-51 OCI Rollforward.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-51 OCI Rollforward.html new file mode 100644 index 000000000..b4b6d2c7f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-51 OCI Rollforward.html @@ -0,0 +1 @@ +
Foreign Currency Translation AdjustmentsUnrealized Gains (Losses) on Available-for-Sale InvestmentsUnrealized Gains (Losses) on Cash Flow HedgesTotal
Balance as of December 31, 2022$ (4,142)$ (3,477)$ 16$ (7,603)
Other comprehensive income (loss) before reclassifications7351,344842,163
Amounts excluded from the assessment of hedge effectiveness recorded in AOCI008484
Amounts reclassified from AOCI01,168(214)954
Other comprehensive income (loss)7352,512(46)3,201
Balance as of December 31, 2023(3,407)(965)(30)(4,402)
Other comprehensive income (loss) before reclassifications(1,673)(116)698(1,091)
Amounts excluded from the assessment of hedge effectiveness recorded in AOCI007777
Amounts reclassified from AOCI0782(166)616
Other comprehensive income (loss)(1,673)666609(398)
Balance as of December 31, 2024(5,080)(299)579(4,800)
Other comprehensive income (loss) before reclassifications2,5111,146(734)2,923
Amounts excluded from the assessment of hedge effectiveness recorded in AOCI110(45)(34)
Amounts reclassified from AOCI0(169)164(5)
Other comprehensive income (loss)2,522977(615)2,884
Balance as of December 31, 2025$ (2,558)$ 678$ (36)$ (1,916)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-52 AOCI Reclass.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-52 AOCI Reclass.html new file mode 100644 index 000000000..54ec663fc --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-52 AOCI Reclass.html @@ -0,0 +1 @@ +
AOCI ComponentsLocationYear Ended December 31,
202320242025
Unrealized gains (losses) on available-for-sale investments
Other income (expense), net$ (1,497)$ (1,008)$ 213
Benefit (provision) for income taxes329226(44)
Net of income tax(1,168)(782)169
Unrealized gains (losses) on cash flow hedges
Foreign exchange contractsRevenue213174(233)
Interest rate contractsOther income (expense), net611
Benefit (provision) for income taxes(5)(9)68
Net of income tax214166(164)
Total amount reclassified, net of income tax$ (954)$ (616)$ 5
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-53 Income Data Table.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-53 Income Data Table.html new file mode 100644 index 000000000..837510fb8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-53 Income Data Table.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Interest income$ 3,865$ 4,482$ 4,337
Interest expense(1)(308)(268)(736)
Foreign currency exchange gain (loss), net(1,238)(409)(382)
Gain (loss) on debt securities, net(1,215)(1,043)540
Gain (loss) on equity securities, net3923,71424,080
Income (loss) and impairment from equity method investments, net(628)(188)281
Other5561,1371,667
Other income (expense), net$ 1,424$ 7,425$ 29,787
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-54 Google Segment Balances.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-54 Google Segment Balances.html new file mode 100644 index 000000000..547402460 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-54 Google Segment Balances.html @@ -0,0 +1 @@ +
Google ServicesGoogle CloudOther BetsTotal
Balance as of December 31, 2023$ 21,118$ 7,199$ 881$ 29,198
Additions2,44129502,736
Foreign currency translation and other adjustments(38)(4)(7)(49)
Balance as of December 31, 202423,5217,49087431,885
Additions1,26916301,432
Foreign currency translation and other adjustments807(24)63
Balance as of December 31, 2025$ 24,870$ 7,660$ 850$ 33,380
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-55 Share Repurchases.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-55 Share Repurchases.html new file mode 100644 index 000000000..c8880a76f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-55 Share Repurchases.html @@ -0,0 +1 @@ +
202320242025
SharesAmountSharesAmountSharesAmount
Class A share repurchases78$ 9,31673$ 11,85537$ 6,501
Class C share repurchases45052,86830650,19220338,897
Total share repurchases(1)528$ 62,184379$ 62,047240$ 45,398
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-56 EPS Breakdown.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-56 EPS Breakdown.html new file mode 100644 index 000000000..5f7c9687a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-56 EPS Breakdown.html @@ -0,0 +1 @@ +
Year Ended December 31, 2023
Class AClass BClass CConsolidated
Basic net income per share:
Numerator
Allocation of distributed earnings (cash dividends paid)$ 0$ 0$ 0$ 0
Allocation of undistributed earnings34,6015,12434,07073,795
Net income$ 34,601$ 5,124$ 34,070$ 73,795
Denominator
Number of shares used in per share computation5,9228775,83112,630
Basic net income per share$ 5.84$ 5.84$ 5.84$ 5.84
Diluted net income per share:
Numerator
Allocation of total earnings for basic computation$ 34,601$ 5,124$ 34,070$ 73,795
Reallocation of total earnings as a result of conversion of Class B to Class A shares5,12400(1)
Reallocation of undistributed earnings(287)(37)287(1)
Net income$ 39,438$ 5,087$ 34,357$ 73,795
Denominator
Number of shares used in basic computation5,9228775,83112,630
Weighted-average effect of dilutive securities
Add:
Conversion of Class B to Class A shares outstanding87700(1)
Restricted stock units and other contingently issuable shares009292
Number of shares used in per share computation6,7998775,92312,722
Diluted net income per share$ 5.80$ 5.80$ 5.80$ 5.80
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-57 EPS Calculation.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-57 EPS Calculation.html new file mode 100644 index 000000000..353fe2921 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-57 EPS Calculation.html @@ -0,0 +1 @@ +
Year Ended December 31, 2024
Class AClass BClass CConsolidated
Basic net income per share:
Numerator
Allocation of distributed earnings (cash dividends paid)$ 3,509$ 519$ 3,335$ 7,363
Allocation of undistributed earnings44,0856,52042,15092,755
Net income$ 47,594$ 7,039$ 45,485$ 100,118
Denominator
Number of shares used in per share computation5,8558665,59812,319
Basic net income per share$ 8.13$ 8.13$ 8.13$ 8.13
Diluted net income per share:
Numerator
Allocation of total earnings for basic computation$ 47,594$ 7,039$ 45,485$ 100,118
Reallocation of total earnings as a result of conversion of Class B to Class A shares7,03900(1)
Reallocation of undistributed earnings(520)(67)520(1)
Net income$ 54,113$ 6,972$ 46,005$ 100,118
Denominator
Number of shares used in basic computation5,8558665,59812,319
Weighted-average effect of dilutive securities
Add:
Conversion of Class B to Class A shares outstanding86600(1)
Restricted stock units and other contingently issuable shares00128128
Number of shares used in per share computation6,7218665,72612,447
Diluted net income per share$ 8.05$ 8.05$ 8.03$ 8.04
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-58 EPS Calculation.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-58 EPS Calculation.html new file mode 100644 index 000000000..44aa99563 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-58 EPS Calculation.html @@ -0,0 +1 @@ +
Year Ended December 31, 2025
Class AClass BClass CConsolidated
Basic net income per share:
Numerator
Allocation of distributed earnings (cash dividends paid)$ 4,832$ 703$ 4,514$ 10,049
Allocation of undistributed earnings58,6828,55754,882122,121
Net income$ 63,514$ 9,260$ 59,396$ 132,170
Denominator
Number of shares used in per share computation5,8228495,44512,116
Basic net income per share$ 10.91$ 10.91$ 10.91$ 10.91
Diluted net income per share:
Numerator
Allocation of total earnings for basic computation$ 63,514$ 9,260$ 59,396$ 132,170
Reallocation of total earnings as a result of conversion of Class B to Class A shares9,26000(1)
Reallocation of undistributed earnings(627)(79)627(1)
Net income$ 72,147$ 9,181$ 60,023$ 132,170
Denominator
Number of shares used in basic computation5,8228495,44512,116
Weighted-average effect of dilutive securities
Add:
Conversion of Class B to Class A shares outstanding84900(1)
Restricted stock units and other contingently issuable shares00114114
Number of shares used in per share computation6,6718495,55912,230
Diluted net income per share$ 10.82$ 10.81$ 10.80$ 10.81
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-59 Stock Vesting.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-59 Stock Vesting.html new file mode 100644 index 000000000..33ea45f66 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-59 Stock Vesting.html @@ -0,0 +1 @@ +
Number of SharesWeighted-Average Grant-Date Fair Value
Unvested as of December 31, 2024299$ 122.77
Granted198$ 188.82
Vested(181)$ 133.90
Forfeited/canceled(34)$ 142.33
Unvested as of December 31, 2025282$ 159.75
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-6 Revenue Breakdown.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-6 Revenue Breakdown.html new file mode 100644 index 000000000..61fa9f7ba --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-6 Revenue Breakdown.html @@ -0,0 +1 @@ +
Year Ended December 31,
20242025
Google Search & other$ 198,084$ 224,532
YouTube ads36,14740,367
Google Network30,35929,792
Google advertising264,590294,691
Google subscriptions, platforms, and devices40,34048,030
Google Services total304,930342,721
Google Cloud43,22958,705
Other Bets1,6481,537
Hedging gains (losses)211(127)
Total revenues$ 350,018$ 402,836
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-60 Revenue Breakdown.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-60 Revenue Breakdown.html new file mode 100644 index 000000000..1d777bd17 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-60 Revenue Breakdown.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Domestic operations$ 73,600$ 108,076$ 143,591
Foreign operations12,11711,73915,235
Total$ 85,717$ 119,815$ 158,826
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-61 Income Tax Provision.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-61 Income Tax Provision.html new file mode 100644 index 000000000..02d415cbd --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-61 Income Tax Provision.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Current:
Federal and state$ 15,716$ 21,101$ 13,378
Foreign3,9353,8525,028
Total19,65124,95318,406
Deferred:
Federal and state(7,482)(5,350)8,243
Foreign(247)947
Total(7,729)(5,256)8,250
Provision for income taxes$ 11,922$ 19,697$ 26,656
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-62 Tax Rate Reconciliation.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-62 Tax Rate Reconciliation.html new file mode 100644 index 000000000..b5e6f0d2c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-62 Tax Rate Reconciliation.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
US federal statutory rate18,00121.0 %25,16121.0 %33,35321.0 %
State and local income taxes, net of federal income tax effect(1)8231.0 %1,1991.0 %1,6061.0 %
Foreign tax effects:
Brazil:
Withholding taxes1,0641.2 %1,0410.9 %1,3840.9 %
Other620.1 %120.0 %230.0 %
Other foreign jurisdictions(74)(0.1)%3530.3 %3960.2 %
Effect of change in tax laws or rates enacted in the current period(829)(1.0)%00.0 %00.0 %
Effect of cross-border tax laws:
Foreign-derived intangible income deduction(3,980)(4.6)%(4,568)(3.8)%(3,931)(2.5)%
Other2150.2 %3210.3 %2950.2 %
Tax credits:
Federal research credit(1,575)(1.8)%(1,792)(1.5)%(2,088)(1.3)%
Foreign tax credits(1,396)(1.6)%(1,373)(1.1)%(1,684)(1.1)%
Other(498)(0.6)%(198)(0.2)%(98)(0.1)%
Changes in valuation allowances5130.6 %6030.5 %1,1700.7 %
Nontaxable or nondeductible items:
Stock-based compensation expense(602)(0.7)%(1,743)(1.5)%(2,601)(1.6)%
Other1690.2 %2030.2 %9550.6 %
Changes in unrecognized tax benefits4320.5 %6890.6 %(1,123)(0.7)%
Other adjustments(403)(0.5)%(211)(0.2)%(1,002)(0.6)%
Total$ 11,92213.9 %$ 19,69716.4 %$ 26,65616.8 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-63 Deferred Tax Assets.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-63 Deferred Tax Assets.html new file mode 100644 index 000000000..3e4a0078c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-63 Deferred Tax Assets.html @@ -0,0 +1 @@ +
As of December 31,
20242025
Deferred tax assets:
Accrued employee benefits$ 1,834$ 1,951
Accruals and reserves not currently deductible2,5523,570
Tax credits6,3847,314
Net operating losses3,4724,953
Operating leases3,3363,337
Capitalized research and development25,90324,758
Other1,3762,143
Total deferred tax assets44,85748,026
Valuation allowance(11,493)(13,942)
Total deferred tax assets net of valuation allowance33,36434,084
Deferred tax liabilities:
Property and equipment, net(9,932)(13,256)
Net investment gains(2,978)(8,242)
Operating leases(2,986)(3,103)
Other(1,008)(1,289)
Total deferred tax liabilities(16,904)(25,890)
Net deferred tax assets (liabilities)$ 16,460$ 8,194
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-64 Income Tax Payments.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-64 Income Tax Payments.html new file mode 100644 index 000000000..85735423f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-64 Income Tax Payments.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
US federal$ 13,689$ 19,921$ 13,658
US state and local1,2242,6972,919
Foreign:
Brazil1,2641,1011,368
Other2,9873,6343,581
Total foreign4,2514,7354,949
Total cash paid for income taxes, net of refunds$ 19,164$ 27,353$ 21,526
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-65 Tax Benefits.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-65 Tax Benefits.html new file mode 100644 index 000000000..b63fbac92 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-65 Tax Benefits.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Beginning gross unrecognized tax benefits$ 7,055$ 9,438$ 12,619
Increases related to prior year tax positions740896278
Decreases related to prior year tax positions(682)(83)(1,301)
Decreases related to settlement with tax authorities(21)(311)(2,183)
Increases related to current year tax positions2,3462,6792,099
Ending gross unrecognized tax benefits$ 9,438$ 12,619$ 11,512
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-66 Alphabet Revenue.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-66 Alphabet Revenue.html new file mode 100644 index 000000000..187a3ddca --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-66 Alphabet Revenue.html @@ -0,0 +1 @@ +
Year Ended December 31,
202320242025
Revenues:
Google Services$ 272,543$ 304,930$ 342,721
Google Cloud33,08843,22958,705
Other Bets1,5271,6481,537
Hedging gains (losses)236211(127)
Total revenues$ 307,394$ 350,018$ 402,836
Operating income (loss):
Google Services$ 95,858$ 121,263$ 139,404
Google Cloud1,7166,11213,910
Other Bets(4,095)(4,444)(7,515)
Alphabet-level activities(9,186)(10,541)(16,760)
Total income from operations$ 84,293$ 112,390$ 129,039
Supplemental information about segment expenses:
Google Services:
Employee compensation expenses$ 46,224$ 44,560$ 45,124
Other costs and expenses130,461139,107158,193
Total Google Services costs and expenses$ 176,685$ 183,667$ 203,317
Google Cloud:
Employee compensation expenses$ 19,054$ 20,519$ 22,078
Other costs and expenses12,31816,59822,717
Total Google Cloud costs and expenses$ 31,372$ 37,117$ 44,795
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-67 Long-lived Assets.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-67 Long-lived Assets.html new file mode 100644 index 000000000..a5ad18c5c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-67 Long-lived Assets.html @@ -0,0 +1 @@ +
As of December 31,
20242025
Long-lived assets:
United States$ 138,993$ 195,337
International45,63166,481
Total long-lived assets$ 184,624$ 261,818
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-68 Warranty Reserves.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-68 Warranty Reserves.html new file mode 100644 index 000000000..6382a5b90 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-68 Warranty Reserves.html @@ -0,0 +1 @@ +
Balance atBeginning of YearAdditionsUsageBalance atEnd of Year
Year ended December 31, 2023$ 1,213$ 3,115$ (2,737)$ 1,591
Year ended December 31, 2024$ 1,591$ 2,895$ (2,850)$ 1,636
Year ended December 31, 2025$ 1,636$ 4,128$ (3,408)$ 2,356
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-69 Exhibit List.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-69 Exhibit List.html new file mode 100644 index 000000000..f0c7d2291 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-69 Exhibit List.html @@ -0,0 +1 @@ +
Exhibit NumberDescriptionIncorporated by reference herein
FormDate
2.01Agreement and Plan of Merger, dated October 2, 2015, by and among Google Inc., the Registrant and Maple Technologies Inc.Current Report on Form 8-K (File No. 001-37580)October 2, 2015
3.01Amended and Restated Certificate of Incorporation of the RegistrantCurrent Report on Form 8-K (File No. 001-37580)June 3, 2022
3.02Amended and Restated Bylaws of the Registrant, dated October 19, 2022Current Report on Form 8-K (File No. 001-37580)October 25, 2022
4.01Specimen Class A Common Stock certificateCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.02Specimen Class C Capital Stock certificateCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.03Alphabet Inc. Deferred Compensation PlanCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.04Transfer Restriction Agreement, dated October 2, 2015, between the Registrant and Larry Page and certain of his affiliatesCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.05Transfer Restriction Agreement, dated October 2, 2015, between the Registrant and Sergey Brin and certain of his affiliatesCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.06Joinder Agreement, dated December 31, 2021, among the Registrant, Sergey Brin and certain of his affiliatesAnnual Report on Form 10-K (File No. 001-37580)February 2, 2022
4.07Transfer Restriction Agreement, dated October 2, 2015, between the Registrant and Eric E. Schmidt and certain of its affiliatesCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.08Class C Undertaking, dated October 2, 2015, executed by the RegistrantCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
4.09Indenture, dated February 12, 2016, between the Registrant and The Bank of New York Mellon Trust Company, N.A., as TrusteeRegistration Statement on Form S-3 (File No. 333-209510)February 12, 2016
4.10Registrant Registration Rights Agreement dated December 14, 2015Registration Statement on Form S-3 (File No. 333-209518)February 12, 2016
4.11First Supplemental Indenture, dated April 27, 2016, between the Registrant and The Bank of New York Mellon Trust Company, N.A., as trusteeCurrent Report on Form 8-K (File No. 001-37580)April 27, 2016
4.12Form of the Registrant's 1.998% Note due 2026Current Report on Form 8-K (File No. 001-37580)August 9, 2016
4.13Form of Global Note representing the Registrant's 0.800% notes due 2027Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.14Form of Global Note representing the Registrant's 1.100% notes due 2030Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.15Form of Global Note representing the Registrant's 1.900% notes due 2040Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.16Form of Global Note representing the Registrant's 2.050% notes due 2050Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.17Form of Global Note representing the Registrant's 2.250% notes due 2060Current Report on Form 8-K (File No. 001-37580)August 5, 2020
4.18Form of Global Note representing the Registrant's 4.000% notes due 2030Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.19Form of Global Note representing the Registrant's 4.500% notes due 2035Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.20Form of Global Note representing the Registrant's 5.250% notes due 2055Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.21Form of Global Note representing the Registrant's 5.300% notes due 2065Current Report on Form 8-K (File No. 001-37580)May 1, 2025
4.22Form of Global Note representing the Registrant's 2.500% notes due 2029Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.23Form of Global Note representing the Registrant's 3.000% notes due 2033Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.24Form of Global Note representing the Registrant's 3.375% notes due 2037Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.25Form of Global Note representing the Registrant's 3.875% notes due 2045Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.26Form of Global Note representing the Registrant's 4.000% notes due 2054Current Report on Form 8-K (File No. 001-37580)May 6, 2025
4.27Form of Global Note representing the Registrant's 2.375% notes due 2028Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.28Form of Global Note representing the Registrant's 2.875% notes due 2031Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.29Form of Global Note representing the Registrant's 3.125% notes due 2034Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.30Form of Global Note representing the Registrant's 3.500% notes due 2038Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.31Form of Global Note representing the Registrant's 4.000% notes due 2044Current Report on Form 8-K (File No. 001-37580)November 6, 2025
Exhibit NumberDescriptionIncorporated by reference herein
FormDate
4.32Form of Global Note representing the Registrant's 4.375% notes due 2064Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.33Form of Global Note representing the Registrant's floating rate notes due 2028Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.34Form of Global Note representing the Registrant's 3.875% notes due 2028Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.35Form of Global Note representing the Registrant's 4.100% notes due 2030Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.36Form of Global Note representing the Registrant's 4.375% notes due 2032Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.37Form of Global Note representing the Registrant's 4.700% notes due 2035Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.38Form of Global Note representing the Registrant's 5.350% notes due 2045Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.39Form of Global Note representing the Registrant's 5.450% notes due 2055Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.40Form of Global Note representing the Registrant's 5.700% notes due 2075Current Report on Form 8-K (File No. 001-37580)November 6, 2025
4.41Description of Registrant's SecuritiesAnnual Report on Form 10-K (File No. 001-37580)February 3, 2023
10.01Form of Indemnification Agreement entered into between the Registrant, its affiliates and its directors and officersCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.02Form of Offer Letter for DirectorsAnnual Report on Form 10-K (File No. 001-37580)January 31, 2024
10.03Letter Agreement dated June 3, 2024, between Anat Ashkenazi and AlphabetCurrent Report on Form 8-K (File No. 001-37580)June 7, 2024
10.04Compensation Plan Agreement, dated October 2, 2015, between Google Inc. and the RegistrantCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.05Director Arrangements Agreement, dated October 2, 2015, between Google Inc. and the RegistrantCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.06Alphabet Inc. Deferred Compensation PlanCurrent Report on Form 8-K (File No. 001-37580)October 2, 2015
10.07Alphabet Inc. Amended and Restated 2012 Stock PlanCurrent Report on Form 8-K (File No. 001-37580)June 5, 2020
10.07.1Alphabet Inc. Amended and Restated 2012 Stock Plan - Form of Alphabet Restricted Stock Unit AgreementAnnual Report on Form 10-K (File No. 001-37580)February 4, 2020
10.08Alphabet Inc. Amended and Restated 2021 Stock PlanCurrent Report on Form 8-K (file No. 001-37580)June 8, 2023
10.08.1Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet Restricted Stock Unit AgreementQuarterly Report on Form 10-Q (file No. 001-37580)July 28, 2021
10.08.2Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet Restricted Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)July 26, 2023
10.08.3Alphabet Inc. Amended and Restated 2021 Stock Plan – Form of Alphabet Restricted Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)April 26, 2024
10.08.4Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet 2022 CEO Performance Stock Unit AgreementAnnual Report on Form 10-K (File No. 001-37580)February 3, 2023
10.08.5Alphabet Inc. Amended and Restated 2021 Stock Plan – Form of Alphabet CEO Performance Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)April 26, 2024
10.08.6Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet 2022 Non-CEO Performance Stock Unit AgreementAnnual Report on Form 10-K (File No. 001-37580)February 2, 2022
10.08.7Alphabet Inc. Amended and Restated 2021 Stock Plan - Form of Alphabet 2023 Non-CEO Performance Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)July 26, 2023
10.08.8Alphabet Inc. Amended and Restated 2021 Stock Plan – Form of Alphabet Non-CEO Performance Stock Unit AgreementQuarterly Report on Form 10-Q (File No. 001-37580)April 26, 2024
10.09Alphabet Inc. Company Bonus Plan, as amendedAnnual Report on Form 10-K (File No. 001-37350)February 2, 2023
19.01Alphabet Inc. Policy Against Insider TradingAnnual Report on Form 10-K (File No. 001-37350)February 5, 2025
21.01*Subsidiaries of the Registrant
23.01*Consent of Independent Registered Public Accounting Firm
24.01*Power of Attorney (incorporated by reference to the signature page of this Annual Report on Form 10-K)
31.01*Certification of Chief Executive Officer pursuant to Exchange Act Rules 13a-14(a) and 15d-14(a), as adopted pursuant to Section 302 of the Sarbanes-Oxley Act of 2002
31.02*Certification of Chief Financial Officer pursuant to Exchange Act Rules 13a-14(a) and 15d-14(a), as adopted pursuant to Section 302 of the Sarbanes-Oxley Act of 2002
32.01Certifications of Chief Executive Officer and Chief Financial Officer pursuant to 18 U.S.C. Section 1350, as adopted pursuant to Section 906 of the Sarbanes-Oxley Act of 2002
97.01Clawback PolicyAnnual Report on Form 10-K (File No. 001-37350)January 31, 2024
101.INS*Inline XBRL Instance Document - the instance document does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document
101.SCH*Inline XBRL Taxonomy Extension Schema Document
101.CAL*Inline XBRL Taxonomy Extension Calculation Linkbase Document
101.DEF*Inline XBRL Taxonomy Extension Definition Linkbase Document
101.LAB*Inline XBRL Taxonomy Extension Label Linkbase Document
101.PRE*Inline XBRL Taxonomy Extension Presentation Linkbase Document
104*Cover Page Interactive Data File (embedded within the Inline XBRL document and contained in Exhibit 101)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-7 Google Ad Changes.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-7 Google Ad Changes.html new file mode 100644 index 000000000..415be9f7e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-7 Google Ad Changes.html @@ -0,0 +1 @@ +
Google Search & other
Paid clicks change6 %
Cost-per-click change7 %
Google Network
Impressions change(7)%
Cost-per-impression change7 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-70 CEO Signature.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-70 CEO Signature.html new file mode 100644 index 000000000..1cb3e63e1 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-70 CEO Signature.html @@ -0,0 +1 @@ +
By:/S/ SUNDAR PICHAI
Sundar Pichai
Chief Executive Officer(Principal Executive Officer of the Registrant)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-71 Signatures.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-71 Signatures.html new file mode 100644 index 000000000..dc989bc8d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-71 Signatures.html @@ -0,0 +1 @@ +
SignatureTitleDate
/S/SUNDAR PICHAIChief Executive Officer and Director (Principal Executive Officer)February 4, 2026
Sundar Pichai
/S/ANAT ASHKENAZISenior Vice President and Chief Financial Officer (Principal Financial Officer)February 4, 2026
Anat Ashkenazi
/S/AMIE THUENER O'TOOLEVice President, Corporate Controller and Principal Accounting OfficerFebruary 4, 2026
Amie Thuener O'Toole
/S/FRANCES H. ARNOLDDirectorFebruary 4, 2026
Frances H. Arnold
/S/SERGEY BRINCo-Founder and DirectorFebruary 4, 2026
Sergey Brin
/S/R. MARTIN CHAVEZDirectorFebruary 4, 2026
R. Martin Chávez
/S/L. JOHN DOERRDirectorFebruary 4, 2026
L. John Doerr
/S/ROGER W. FERGUSON JR.DirectorFebruary 4, 2026
Roger W. Ferguson Jr.
/S/JOHN L. HENNESSYDirector, ChairFebruary 4, 2026
John L. Hennessy
/S/LARRY PAGECo-Founder and DirectorFebruary 4, 2026
Larry Page
/S/K. RAM SHRIRAMDirectorFebruary 4, 2026
K. Ram Shriram
/S/ROBIN L. WASHINGTONDirectorFebruary 4, 2026
Robin L. Washington
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-8 Revenue by Region.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-8 Revenue by Region.html new file mode 100644 index 000000000..5611609c6 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-8 Revenue by Region.html @@ -0,0 +1 @@ +
Year Ended December 31,
20242025
United States49 %48 %
$EMEA^{(1)}$ 29 %29 %
$APAC^{(1)}$ 16 %17 %
$Other Americas^{(1)}$ 6 %6 %
Hedging gains (losses)0 %0 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-9 Cost of Revenues.html b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-9 Cost of Revenues.html new file mode 100644 index 000000000..b866ee4e0 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/tables/table-9 Cost of Revenues.html @@ -0,0 +1 @@ +
Year Ended December 31,
20242025
TAC$ 54,900$ 59,926
Other cost of revenues91,406102,609
Total cost of revenues$ 146,306$ 162,535
Total cost of revenues as a percentage of revenues42 %40 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-goog-10-k-2025/toc_hierarchies.json b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/toc_hierarchies.json new file mode 100644 index 000000000..88e968e06 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-goog-10-k-2025/toc_hierarchies.json @@ -0,0 +1,19 @@ +[ + { + "toc_range": [ + 43, + 48 + ], + "scan_range": [ + 43, + 193 + ], + "toc_with_level": "| id | heading | level |\n|----|-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------|\n| 1 | # Note About Forward-Looking Statements | 1 |\n| 2 | This Annual Report on Form 10-K contains forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. Forward-looking statements generally can be identified by words such as, but are not limited to, \"anticipates,\" \"believes,\" \"could,\" \"estimates,\" \"expects,\" \"intends,\" \"may,\" \"plans,\" \"predicts,\" \"projects,\" \"will be,\" \"will continue,\" \"will likely result,\" and similar expressions. These include, among other things, expectations regarding the growth of our business and revenues, including factors that may impact such growth, and fluctuations in our revenues and margins; statements relating to plans, expectations, and trends about our core business metrics, costs and expenses, capital expenditures, sources of funding, products and services, strategic business transactions, and other aspects of our business operations and strategies; statements regarding the global macroeconomic and regulatory environment; as well as other statements regarding our future operations, financial condition and prospects, and actual or potential risk and liability exposures. Forward-looking statements may appear throughout this report and other documents we file with the Securities and Exchange Commission (SEC), including without limitation, the following sections: Part I, Item 1 \"Business;\" Part I, Item 1A \"Risk Factors;\" and Part II, Item 7 \"Management's Discussion and Analysis of Financial Condition and Results of Operations.\" These forward-looking statements are based on current expectations and assumptions that are subject to risks and uncertainties, which could cause our actual results to differ materially from those reflected in the forward-looking statements. Factors that could cause or contribute to such differences include, but are not limited to, those discussed in this Annual Report on Form 10-K, including the risks discussed in Part I, Item 1A \"Risk Factors\" and the trends discussed in Part II, Item 7 \"Management's Discussion and Analysis of Financial Condition and Results of Operations,\" and those discussed in other documents we file with the SEC. We undertake no obligation to revise or publicly release the results of any revision to these forward-looking statements, except as required by law. Given these risks and uncertainties, readers are cautioned not to place undue reliance on such forward-looking statements. | 1 |\n| 3 | As used herein, \"Alphabet,\" \"the company,\" \"we,\" \"us,\" \"our,\" and similar terms include Alphabet Inc. and its subsidiaries, unless the context indicates otherwise. | 1 |\n| 4 | \"Alphabet,\" \"Google,\" and other trademarks of ours appearing in this report are our property. We do not intend our use or display of other companies' trade names or trademarks to imply an endorsement or sponsorship of us by such companies, or any relationship with any of these companies. | 1 |", + "toc_tree": { + "# Note About Forward-Looking Statements": {}, + "This Annual Report on Form 10-K contains forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. Forward-looking statements generally can be identified by words such as, but are not limited to, \"anticipates,\" \"believes,\" \"could,\" \"estimates,\" \"expects,\" \"intends,\" \"may,\" \"plans,\" \"predicts,\" \"projects,\" \"will be,\" \"will continue,\" \"will likely result,\" and similar expressions. These include, among other things, expectations regarding the growth of our business and revenues, including factors that may impact such growth, and fluctuations in our revenues and margins; statements relating to plans, expectations, and trends about our core business metrics, costs and expenses, capital expenditures, sources of funding, products and services, strategic business transactions, and other aspects of our business operations and strategies; statements regarding the global macroeconomic and regulatory environment; as well as other statements regarding our future operations, financial condition and prospects, and actual or potential risk and liability exposures. Forward-looking statements may appear throughout this report and other documents we file with the Securities and Exchange Commission (SEC), including without limitation, the following sections: Part I, Item 1 \"Business;\" Part I, Item 1A \"Risk Factors;\" and Part II, Item 7 \"Management's Discussion and Analysis of Financial Condition and Results of Operations.\" These forward-looking statements are based on current expectations and assumptions that are subject to risks and uncertainties, which could cause our actual results to differ materially from those reflected in the forward-looking statements. Factors that could cause or contribute to such differences include, but are not limited to, those discussed in this Annual Report on Form 10-K, including the risks discussed in Part I, Item 1A \"Risk Factors\" and the trends discussed in Part II, Item 7 \"Management's Discussion and Analysis of Financial Condition and Results of Operations,\" and those discussed in other documents we file with the SEC. We undertake no obligation to revise or publicly release the results of any revision to these forward-looking statements, except as required by law. Given these risks and uncertainties, readers are cautioned not to place undue reliance on such forward-looking statements.": {}, + "As used herein, \"Alphabet,\" \"the company,\" \"we,\" \"us,\" \"our,\" and similar terms include Alphabet Inc. and its subsidiaries, unless the context indicates otherwise.": {}, + "\"Alphabet,\" \"Google,\" and other trademarks of ours appearing in this report are our property. We do not intend our use or display of other companies' trade names or trademarks to imply an endorsement or sponsorship of us by such companies, or any relationship with any of these companies.": {} + } + } +] \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/chunks.json b/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/chunks.json new file mode 100644 index 000000000..968bff0a2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/chunks.json @@ -0,0 +1,2525 @@ +{ + "chunks": [ + { + "chunk_id": "f224df75-3d75-5a6f-9ffd-9ce11c4d6ed3", + "type": "text", + "content": "Meta Platforms, Inc. (META)\nFirst Quarter 2026 Results Conference Call\nApril 29 $^{th}$ , 2026", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf", + "metadata": { + "length": 94, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "Meta", + "Platforms", + "Inc", + "META", + "First", + "Quarter", + "2026", + "Results", + "Conference", + "Call", + "April", + "29" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "770126b8-2884-53c9-b340-0f2e3302dcd5", + "type": "text", + "content": "Thank you. Good afternoon and welcome to Meta Platforms' first quarter 2026 earnings conference call. Joining me today to discuss our results are Mark Zuckerberg, CEO and Susan Li, CFO.\nOur remarks today will include forward-looking statements, which are based on assumptions as of today. Actual results may differ materially as a result of various factors including those set forth in today's earnings press release, and in our annual report on Form 10-K filed with the SEC. We undertake no obligation to update any forward-looking statement.\nDuring this call we will present both GAAP and certain non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in today's earnings press release. The earnings press release and an accompanying investor presentation are available on our website at investor.atmeta.com.\nAnd now, I'd like to turn the call over to Mark.", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Kenneth Dorell, Director, Investor Relations", + "metadata": { + "length": 891, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "Thank", + "Good", + "afternoon", + "Meta", + "Platforms", + "quarter", + "2026", + "earnings", + "conference", + "call", + "Joining", + "today", + "discuss", + "results", + "Mark", + "Zuckerberg", + "CEO", + "Susan", + "Li", + "CFO", + "Our", + "remarks", + "include", + "forward", + "statements", + "based", + "assumptions", + "Actual", + "differ", + "materially", + "result", + "factors", + "including", + "set", + "press", + "release", + "annual", + "report", + "Form", + "10", + "filed", + "SEC", + "We", + "undertake", + "obligation", + "update", + "statement", + "During", + "present", + "GAAP", + "financial", + "measures", + "reconciliation", + "included", + "The", + "accompanying", + "investor", + "presentation", + "website", + "atmeta", + "And", + "turn" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "298b8cfe-f1dd-5464-b5b6-1e2548dad1f5", + "type": "text", + "content": "Hey everyone, thanks for joining today. We had a strong quarter for our community, our business, and our progress towards AI.\nMore than 3.5 billion people use at least one of our apps every day. We saw a small decrease in total family dailies due to internet outages in Iran and blocks in Russia, but otherwise trends across our apps are strong. Daily and monthly actives on Instagram and Facebook continue to grow, with video driving all-time high engagement across both apps. WhatsApp continues to see strong momentum too, including in the US. And Threads continues on its trajectory to be the leading app in its category.\nOur biggest milestone so far this year has been the release of our Muse family of models and our first model, Muse Spark, along with a significantly upgraded new version of Meta AI. This was the first release from Meta Superintelligence Labs, and it shows that our work is on track to build a leading lab.\nOver the past 10 months, we have built the strongest research team in the industry, and established the scientific and technical foundations to scale very advanced models. Spark is just one step on that scaling ladder and we are already training even more advanced models. But Spark has already made Meta AI a world-class assistant that leads in several areas related to our vision of personal superintelligence, including visual understanding, health, shopping, social content, local, creating games, and more. We're hearing very positive feedback on it so far. We've seen large increases in Meta AI use since releasing the updates, and the Meta AI app has consistently been near the top of the app stores as well.\nNow that we have a strong model we can develop more novel products as well. Since I first wrote about our vision for personal superintelligence last year, we've been focused on delivering\npersonal and business agents to billions of people around the world. Our goal is not just to deliver Meta AI as an assistant, but to deliver agents that can understand your goals and then work day and night to help you achieve them.\nMy view of AI is very different from many others in the industry. I hear a lot of people out there talk about how AI is going to replace people. Instead, I think that AI is going to amplify people's ability to do what you want -- whether that's to improve your health, your learning, your relationships, your ability to achieve your personal and career goals and more. My view is that human progress has always been driven by people pursuing their individual aspirations, and I believe that this will continue to be true in the future. People will be more important in the future, not less. Meta believes in empowering individuals. Those are the kinds of products we're going to build, and I believe they're going to be some of the most important and valuable products of all time.\nWe are building a personal agent focused on helping people achieve the diverse goals in their lives. We are also building a business agent focused on helping entrepreneurs and businesses across the world use our tools and others to grow their efforts, reach new customers, and serve existing customers better. These agents will work together to form an ecosystem, and whether you use our personal or business agents to achieve your goals, I believe that the future will see a massive increase in entrepreneurship from people creating new things that they've always wanted to exist but previously didn't have the tools to bring into the world. We're already testing an early version of business AIs, and weekly conversations have grown 10x since the start of this year.\nWe're also working on using Spark and our upcoming models to improve our recommendation systems and core business in Facebook, Instagram, and ads. Right now, our apps primarily help people accomplish three important goals: connecting with people, learning about the world, and entertainment. But we've always wanted our apps to understand more of people's goals so we can help improve their lives in all the ways that they want. These new AI models will let us understand this in more detail. So instead of just looking at statistical patterns of what types of people engage with what content, for the first time in Meta's history we're going to be able to develop a first-principles understanding of what you care about and what each piece of content in our system is about so that way we can show you more useful things for what you're trying to accomplish. We'll also be able to create personalized content specifically for people to help you achieve your goals as well.\nSince our recommendation systems are operating at such large scale, we'll phase in this new research and technology over time. But the trend over the last few years seems clear that we are seeing an increasing return on the amount that we can improve engagement for people and value for advertisers. This encourages us to continue investing heavily in what we expect will provide increasing value over the coming years as well.\nOn that note, we are increasing our infrastructure capex forecast for this year. Most of that is due to higher component costs, particularly memory pricing. But every sign that we're seeing in our own work and across the industry gives us confidence in this investment. That said, we are very focused on increasing the efficiency of our investments. And as part of that, we're rolling out more than 1GW of our own custom silicon that we're developing with Broadcom as well as significant amounts of AMD chips to compliment the new Nvidia systems that we're rolling out as well. One of the primary goals of our Meta Compute initiative is to lead the industry in efficiency of building compute, and we expect that will be a strategic advantage over time.\nTalking about building physical goods at scale, our AI glasses continue to perform well with the number of people using them daily tripling year-over-year. This continues to be one of the fastest-growing categories of consumer electronics ever. We released Ray-Ban Meta Optics this quarter, designed for all-day wear rather than primarily as sunglasses. And building on our release of Oakley last year, we have some exciting new partnerships and styles that I think are going to have the potential to reach even more people coming later this year.\nAll of our glasses are designed to easily update to use our newest AI models and features. I'm also really excited to see the glasses evolve from being able to answer questions to being able to be a personal agent that's with you all day long, helping you remember things and achieve your goals.\nBeyond glasses, I am excited for more of our metaverse efforts to be powered by the AI models we're training as well. We remain the biggest investor in the VR space across the industry, but we are focused on making our VR business sustainable as we invest more in other areas like AI and glasses.\nBefore wrapping, I want to talk for a moment about how AI is transforming our work. We're seeing more and more examples where one or two people are building something in a week that would have previously taken dozens of people months. I want to make sure that Meta is the best place in the world for these types of people to come and make an impact. We're building the next evolution of our company around these people. There's a lot that we can do to enable this: building the best infrastructure for creating and delivering products at scale, streamlining our teams so they aren't bigger than they need to be, recognizing and rewarding the people who are having outsized impacts, and setting ourselves up to try many more ideas and take on many new projects in the future. Of course we'll continue pushing to increase our efficiency as well, but overall I think the future is about building many more higher quality things than we've ever built before.\nAlright, that is what I wanted to cover today. We are living through a historic technological transformation. We are among the few companies positioned to shape the future, and we are on track to do that. I'm looking forward to delivering personal superintelligence to billions of people. And as always, I'm grateful for the hard work of our teams, and to all of you for being on this journey with us.", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Mark Zuckerberg, CEO", + "metadata": { + "length": 8271, + "summary": "The speaker discusses Meta's strong quarterly performance with over 3.5 billion daily users across apps, highlighting the release of the Muse family of models and Meta AI's advancements. They emphasize a vision of AI amplifying human potential rather than replacing people, focusing on personal and business agents. The talk covers improvements in recommendation systems, increased infrastructure investment, growth in AI glasses usage, and the transformative impact of AI on work efficiency at Meta.", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "Hey", + "joining", + "today", + "We", + "strong", + "quarter", + "community", + "business", + "progress", + "AI", + "More", + "3.5", + "billion", + "people", + "apps", + "day", + "small", + "decrease", + "total", + "family", + "dailies", + "due", + "internet", + "outages", + "Iran", + "blocks", + "Russia", + "trends", + "Daily", + "monthly", + "actives", + "Instagram", + "Facebook", + "continue", + "grow", + "video", + "driving", + "time", + "high", + "engagement", + "WhatsApp", + "continues", + "momentum", + "including", + "US", + "And", + "Threads", + "trajectory", + "leading", + "app", + "category", + "Our", + "biggest", + "milestone", + "year", + "release", + "Muse", + "models", + "model", + "Spark", + "significantly", + "upgraded", + "version", + "Meta", + "This", + "Superintelligence", + "Labs", + "shows", + "work", + "track", + "build", + "lab", + "Over", + "past", + "10", + "months", + "built", + "strongest", + "research", + "team", + "industry", + "established", + "scientific", + "technical", + "foundations", + "scale", + "advanced", + "step", + "scaling", + "ladder", + "training", + "But", + "made", + "world", + "class", + "assistant", + "leads", + "areas", + "related", + "vision", + "personal", + "superintelligence", + "visual", + "understanding", + "health", + "shopping", + "social", + "content", + "local", + "creating", + "games", + "hearing", + "positive", + "feedback", + "ve", + "large", + "increases", + "releasing", + "updates", + "consistently", + "top", + "stores", + "Now", + "develop", + "products", + "Since", + "wrote", + "focused", + "delivering", + "agents", + "billions", + "goal", + "deliver", + "understand", + "goals", + "night", + "achieve", + "My", + "view", + "hear", + "lot", + "talk", + "replace", + "Instead", + "amplify", + "ability", + "improve", + "learning", + "relationships", + "career", + "human", + "driven", + "pursuing", + "individual", + "aspirations", + "true", + "future", + "People", + "important", + "believes", + "empowering", + "individuals", + "Those", + "kinds", + "valuable", + "building", + "agent", + "helping", + "diverse", + "lives", + "entrepreneurs", + "businesses", + "tools", + "efforts", + "reach", + "customers", + "serve", + "existing", + "These", + "form", + "ecosystem", + "massive", + "increase", + "entrepreneurship", + "things", + "wanted", + "exist", + "previously", + "didn", + "bring", + "testing", + "early", + "AIs", + "weekly", + "conversations", + "grown", + "10x", + "start", + "working", + "upcoming", + "recommendation", + "systems", + "core", + "ads", + "Right", + "primarily", + "accomplish", + "connecting", + "entertainment", + "ways", + "detail", + "So", + "statistical", + "patterns", + "types", + "engage", + "history", + "principles", + "care", + "piece", + "system", + "show", + "ll", + "create", + "personalized", + "specifically", + "operating", + "phase", + "technology", + "trend", + "years", + "clear", + "increasing", + "return", + "amount", + "advertisers", + "encourages", + "investing", + "heavily", + "expect", + "provide", + "coming", + "On", + "note", + "infrastructure", + "capex", + "forecast", + "Most", + "higher", + "component", + "costs", + "memory", + "pricing", + "sign", + "confidence", + "investment", + "That", + "efficiency", + "investments", + "part", + "rolling", + "1GW", + "custom", + "silicon", + "developing", + "Broadcom", + "significant", + "amounts", + "AMD", + "chips", + "compliment", + "Nvidia", + "One", + "primary", + "Compute", + "initiative", + "lead", + "compute", + "strategic", + "advantage", + "Talking", + "physical", + "goods", + "glasses", + "perform", + "number", + "daily", + "tripling", + "fastest", + "growing", + "categories", + "consumer", + "electronics", + "released", + "Ray", + "Ban", + "Optics", + "designed", + "wear", + "sunglasses", + "Oakley", + "exciting", + "partnerships", + "styles", + "potential", + "All", + "easily", + "update", + "newest", + "features", + "excited", + "evolve", + "answer", + "questions", + "long", + "remember", + "Beyond", + "metaverse", + "powered", + "remain", + "investor", + "VR", + "space", + "making", + "sustainable", + "invest", + "Before", + "wrapping", + "moment", + "transforming", + "examples", + "week", + "dozens", + "make", + "place", + "impact", + "evolution", + "company", + "There", + "enable", + "streamlining", + "teams", + "aren", + "bigger", + "recognizing", + "rewarding", + "outsized", + "impacts", + "setting", + "ideas", + "projects", + "Of", + "pushing", + "quality", + "Alright", + "cover", + "living", + "historic", + "technological", + "transformation", + "companies", + "positioned", + "shape", + "forward", + "grateful", + "hard", + "journey" + ], + "keywords": [ + "Meta AI", + "personal superintelligence", + "business agents" + ], + "connect_to": [] + } + }, + { + "chunk_id": "731d62b5-a669-54fc-a7e9-0b04c8f6271e", + "type": "text", + "content": "Thanks Mark and good afternoon everyone.\nLet's begin with our segment results. All comparisons are on a year-over-year basis unless otherwise noted.\nWe estimate 3.56 billion people used at least one of our Family of Apps on a daily basis in March, which declined slightly from December due to internet disruptions in Iran and a restriction on access to WhatsApp in Russia. Absent these impacts, growth in Family Daily Active People would have been positive quarter-over-quarter.\nQ1 Total Family of Apps revenue was \\$55.9 billion, up 33% year-over-year.\nQ1 Family of Apps ad revenue was \\$55.0 billion, up 33% or 29% on a constant currency basis.\nIn Q1, the total number of ad impressions served across our services increased 19%. Impression growth was healthy across all regions, driven primarily by growth in engagement and users, as well as ad load optimizations. The global average price per ad increased 12% year-over-year in Q1, with broad-based growth as we benefited from ad performance improvements, better macro conditions versus Q1 of last year, and currency tailwinds in international regions. This was partially offset by strong impression growth, including from lower monetizing regions.\nFamily of Apps other revenue was \\$885 million, up 74%, driven primarily by WhatsApp paid messaging and subscriptions revenue.\nWithin our Reality Labs segment, Q1 revenue was \\$402 million, down 2% year-over-year due to lower Quest headset sales, which were partially offset by continued strong growth in AI glasses revenue.\nMoving now to our consolidated results.\nQ1 total revenue was \\$56.3 billion, up 33% or 29% on a constant currency basis.\nQ1 total expenses were \\$33.4 billion, up 35% compared to last year. Year-over-year growth was driven mainly by infrastructure costs and employee compensation.\nThe growth in infrastructure costs was due to higher depreciation, data center operating costs, and third party cloud spend.\nThe growth in employee compensation was driven by technical hires we’ve added over the past year, particularly AI talent.\nWe ended Q1 with over 77,900 employees, down 1% from Q4 as the impact of headcount optimization efforts in certain functions was partially offset by hiring in priority areas of monetization and infrastructure.\nFirst quarter operating income was \\$22.9 billion, representing a 41% operating margin.\nQ1 interest and other income was -\\$1.1 billion, driven by unrealized losses on our equity investments.\nOur tax rate for the quarter was -23%, which was favorably impacted by a tax benefit of \\$8.03 billion. This benefit partially relieves the \\$15.93 billion non-cash tax charge we recorded in the third quarter of 2025, which reflects updated guidance from the US Treasury issued in February 2026 regarding the tax treatment of previously capitalized R&D expenditures in the United States. Absent the tax benefit, our Q1 tax rate would have been 14%.\nNet income was \\$26.8 billion or \\$10.44 per share. Absent the tax benefit, our net income and EPS would have been \\$18.7 billion and \\$7.31, respectively.\nCapital expenditures, including principal payments on finance leases, were \\$19.8 billion, driven by investments in servers, data centers, and network infrastructure.\nFree cash flow was \\$12.4 billion. We ended the quarter with \\$81.2 billion in cash and marketable securities and \\$58.7 billion in debt.\nTurning now to the business performance. There are two primary factors that drive our revenue performance: our ability to deliver engaging experiences for our community, and our effectiveness at monetizing that engagement over time.\nOn the first, we're continuing to see significant gains from our content recommendation initiatives.\nOn Instagram, the ranking improvements that we made in Q1 drove a 10% lift in Reels time spent.\nOn Facebook, total video time increased more than 8% globally in Q1, the largest quarter-over-quarter gain in four years. Within the US & Canada, ranking improvements we made drove a 9% increase in video watch time on Facebook in Q1.\nThese gains are benefiting from advances we're making across the full stack.\nStarting with data, we doubled the length of user interaction sequences we use for training on Instagram in Q1 and increased the richness of how each user interaction is described, enabling our systems to develop a deeper understanding of user interests.\nWithin our models, we’ve significantly increased the speed with which our ranking models index new posts, which is enabling us to recommend them sooner after they are published. We’re also applying more advanced content understanding techniques, which is enabling us to quickly identify posts that may be interesting to someone, even if they haven’t engaged with a lot of similar content. These and other improvements have enabled us to increase the diversity and recency of recommended content, with same-day posts now representing more than 30% of recommended Reels on both Instagram and Facebook, more than double the levels one year ago.\nWe're also using AI to unlock more inventory by auto-translating and dubbing videos into a viewer's local language, enabling us to recommend a more diverse set of content. Over half a billion users on each of Facebook and Instagram are now watching AI-translated videos weekly.\nLooking forward, we're making several investments we expect will deliver more valuable recommendations.\nThis year, we will continue scaling up our models in several dimensions, including their size and complexity, while incorporating LLMs to deepen content understanding across our platform. This will enable us to better match people to a wider variety of content aligned to their interests.\nAt the same time, we are executing on our longer-term efforts to develop the next generation of our recommendation systems. This includes building foundation models that power organic content and ads recommendations, as well as developing LLM-based recommender systems. Our focus this year is validating the model architectures and techniques in these domains before we scale them out in future years.\nAside from our recommendations work, we are focused on deploying the models from Meta Superintelligence Labs to enable a new set of product experiences.\nWe're seeing encouraging results within Meta AI since we began powering responses with the first model from MSL, Muse Spark. In tests we ran leading up to the launch, we saw meaningful\nengagement gains that accelerated week-over-week with each new iteration of the model. We're seeing similar gains within Meta AI following the broad roll out of our new model, with double-digit percent increases in Meta AI sessions per user. Muse Spark is now powering Meta AI in direct chat threads across our Family of Apps, as well as the standalone Meta AI app and website, giving billions of people globally access to our latest model.\nOverall, we're very encouraged by the momentum within our research and product roadmap and look forward to sharing more detail on what we're building over the course of this year.\nTurning to the second driver of our revenue performance: increasing monetization efficiency.\nThe first part of this work is optimizing the level of ads within organic engagement.\nHere, we continue to enhance our systems to show ads at the optimal time and location. In Q1, we also expanded availability of ads on our newer surfaces, including bringing ads on Threads to people in more markets. On WhatsApp, we are making good progress with the roll out of ads in Status, with hundreds of millions of people now viewing them daily.\nMoving to the second part of increasing monetization efficiency: improving performance for the businesses who use our services. To do so, we're deploying AI more deeply across each layer of our systems and tools.\nWithin our ads systems, we're delivering performance gains as we deploy more complex and predictive models.\nIn Q1, enhancements we made to Lattice's modeling and learning techniques, along with advances in our GEM model architecture, drove a more than $6\\%$ increase in conversion rate for landing page view ads.\nIn addition, we’ve been investing in more performant inference models for when we’re serving ads. In the second half of last year, we began rolling out our new Adaptive Ranking Model, which is an LLM-scale ads recommender model that we use for inference. This model improves our inference ROI by routing requests to more compute intensive inference models when it determines there is a higher probability of conversion. In Q1, we expanded coverage of our Adaptive Ranking model to support offsite conversions, which drove a 1.6% increase in conversion rates across the major surfaces on Facebook and Instagram.\nWe're also leveraging AI to make it easier for businesses to manage their campaigns, develop ad creative, and engage with customers.\nThe Meta AI business assistant has now been fully rolled out to all eligible advertisers on supported Meta buying surfaces, providing personalized recommendations to advertisers, resolving account issues, and surfacing campaign insights to help optimize results. Performance has been strong since we began testing the assistant in Q4, with common account issues being resolved at a 20% higher rate.\nThis week, we're also introducing Meta ads AI connectors in open beta, providing advertisers the ability to connect their Meta ad account directly to an AI agent. We've always supported advertisers both on our platform and through tools like the Marketing API — now we're extending that to AI so businesses and agencies can analyze and optimize campaigns with the tools they're already using.\nUsage of our ad creative tools is also scaling, with more than 8 million advertisers using at least one of our gen AI ad creative tools and particularly strong adoption among small and medium-sized advertisers. These tools are benefiting performance as well, with advertisers using our video generation feature seeing more than 3% higher conversion rates in tests.\nWe're also seeing good traction in using AI to facilitate customer engagement. In Q1, we expanded Business AIs on WhatsApp to SMBs across Latin America and Indonesia, as well as on Messenger in Asia-Pacific. We now have more than 10 million conversations each week being facilitated through Business AIs, up from 1 million at the start of the year. We'll further expand access to more countries this quarter while adding more capabilities to the AIs.\nWe also continue to invest in the Value Optimization suite, which helps advertisers maximize their Return on Ad Spend by prioritizing the highest-value conversions rather than optimizing solely for the most conversions at the lowest cost. Adoption by businesses has been strong following performance improvements we’ve made over the past year, with the annual revenue run-rate of our value optimization suite now over \\$20B, more than doubling year-over-year.\nLast, I want to touch on our commerce efforts.\nPeople discover products on our platforms through ads and organic posts, with brands increasingly turning to creators to promote their products. This is contributing to rapid growth in our partnership ads product, with its revenue run-rate more than doubling year-over-year in Q1 to \\$10 billion.\nTo support the product discovery and purchasing happening through creators, we're expanding our solutions beyond ads. Last month, we rolled out our affiliate partnerships offering on Facebook to more test partners, so creators can tag products from participating retailers on their posts and earn a commission when someone makes a purchase. We have also started testing similar experiences on Instagram.\nWe see a real opportunity to help people more easily discover and buy products within our services, particularly as we incorporate AI deeply across our platforms.\nNext, I would like to discuss our approach to capital allocation.\nCompute is becoming increasingly important as it determines the quality of the services we can provide, including powering more capable models and delivering innovative new products. It is also becoming more critical to how we work at Meta, as we are entering a world where employees are managing agents to help them generate new ideas, run experiments, execute tasks, and build products.\nWe are investing aggressively to meet our infrastructure needs and ensure we maximize our strategic flexibility over the coming years. This includes substantially expanding our own data center footprint and striking deals throughout the supply chain to secure necessary components for future capacity. We are also signing cloud deals that will come online over the course of this year and 2027, allowing us to scale more quickly. These multi-year cloud deals and our\ninfrastructure purchase agreements drove a \\$107 billion step up in our contractual commitments this quarter.\nOur investments will support our training needs for future models, and most importantly, provide us the inference capacity necessary to deliver personal and business agents to billions of people around the world, along with several other AI product experiences we're developing.\nAs we grow our infrastructure spend, we remain committed to operating efficiently, and we recently shared internally that we plan to reduce the size of our employee base in May. We believe a leaner operating model will allow us to move more quickly while also helping to offset the substantial investments we're making.\nMoving to our financial outlook.\nWe expect second quarter 2026 total revenue to be in the range of \\$58-61 billion. Our guidance assumes foreign currency is an approximately 2% tailwind to year-over-year total revenue growth, based on current exchange rates.\nTurning to the expense and capex outlooks.\nWe expect full year 2026 total expenses to be in the range of \\$162-169 billion, unchanged from our prior outlook.\nWe continue to expect to deliver operating income this year that is above 2025 operating income.\nWe anticipate 2026 capital expenditures, including principal payments on finance leases, to be in the range of \\$125-145 billion, increased from our prior range of \\$115-135 billion. This reflects our expectations for higher component pricing this year and, to a lesser extent, additional data center costs to support future year capacity.\nAbsent any changes to our tax landscape, we expect our tax rate for the remaining quarters of 2026 to be between 13-16%.\nLastly, we continue to monitor active legal and regulatory matters, including headwinds in the EU and the US that could significantly impact our business and financial results. For example, we continue to see scrutiny on youth-related issues and have additional trials scheduled for this year in the US, which may ultimately result in a material loss.\nIn closing, Q1 was a solid start to the year, with strong execution across our core ads and engagement initiatives. We’re also making exciting progress on our AI research and product efforts and expect to build on that momentum over the course of this year.\nWith that, Krista, let's open up the call for questions.\nOperator:\nThank you. We will now open the lines for a question and answer session. To ask a question, please press star one on your touch tone phone. To withdraw your question again, press star one.\nPlease limit yourself to one question. Please pick up your handset before asking your question to ensure clarity. If you are streaming today's call, please mute your computer speakers. And your first question comes from Brian Nowak with Morgan Stanley. Please go ahead.", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO", + "metadata": { + "length": 15558, + "summary": "Meta reported Q1 2026 results with total revenue of $56.3 billion, up 33% year-over-year. Family of Apps revenue was $55.9 billion, with ad revenue of $55.0 billion. Daily active people across Family of Apps was 3.56 billion in March. Operating income was $22.9 billion with a 41% margin. Net income was $26.8 billion, or $10.44 per share, including a tax benefit. Capital expenditures were $19.8 billion. The company highlighted AI-driven improvements in content recommendations, ad performance, and product experiences, including gains from Meta AI and Muse Spark. Revenue guidance for Q2 2026 is $58-61 billion, with full-year expense outlook of $162-169 billion and capex of $125-145 billion.", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "Thanks", + "Mark", + "good", + "afternoon", + "Let", + "begin", + "segment", + "results", + "All", + "comparisons", + "year", + "basis", + "noted", + "We", + "estimate", + "3.56", + "billion", + "people", + "Family", + "Apps", + "daily", + "March", + "declined", + "slightly", + "December", + "due", + "internet", + "disruptions", + "Iran", + "restriction", + "access", + "WhatsApp", + "Russia", + "Absent", + "impacts", + "growth", + "Daily", + "Active", + "People", + "positive", + "quarter", + "Q1", + "Total", + "revenue", + "55.9", + "33%", + "ad", + "55.0", + "29%", + "constant", + "currency", + "In", + "total", + "number", + "impressions", + "served", + "services", + "increased", + "19%", + "Impression", + "healthy", + "regions", + "driven", + "primarily", + "engagement", + "users", + "load", + "optimizations", + "The", + "global", + "average", + "price", + "12%", + "broad", + "based", + "benefited", + "performance", + "improvements", + "macro", + "conditions", + "versus", + "tailwinds", + "international", + "This", + "partially", + "offset", + "strong", + "impression", + "including", + "lower", + "monetizing", + "885", + "million", + "74%", + "paid", + "messaging", + "subscriptions", + "Within", + "Reality", + "Labs", + "402", + "2%", + "Quest", + "headset", + "sales", + "continued", + "AI", + "glasses", + "Moving", + "consolidated", + "56.3", + "expenses", + "33.4", + "35%", + "compared", + "Year", + "infrastructure", + "costs", + "employee", + "compensation", + "higher", + "depreciation", + "data", + "center", + "operating", + "party", + "cloud", + "spend", + "technical", + "hires", + "ve", + "added", + "past", + "talent", + "ended", + "77", + "900", + "employees", + "1%", + "Q4", + "impact", + "headcount", + "optimization", + "efforts", + "functions", + "hiring", + "priority", + "areas", + "monetization", + "First", + "income", + "22.9", + "representing", + "41%", + "margin", + "interest", + "1.1", + "unrealized", + "losses", + "equity", + "investments", + "Our", + "tax", + "rate", + "23%", + "favorably", + "impacted", + "benefit", + "8.03", + "relieves", + "15.93", + "cash", + "charge", + "recorded", + "2025", + "reflects", + "updated", + "guidance", + "US", + "Treasury", + "issued", + "February", + "2026", + "treatment", + "previously", + "capitalized", + "expenditures", + "United", + "States", + "14%", + "Net", + "26.8", + "10.44", + "share", + "net", + "EPS", + "18.7", + "7.31", + "Capital", + "principal", + "payments", + "finance", + "leases", + "19.8", + "servers", + "centers", + "network", + "Free", + "flow", + "12.4", + "81.2", + "marketable", + "securities", + "58.7", + "debt", + "Turning", + "business", + "There", + "primary", + "factors", + "drive", + "ability", + "deliver", + "engaging", + "experiences", + "community", + "effectiveness", + "time", + "On", + "continuing", + "significant", + "gains", + "content", + "recommendation", + "initiatives", + "Instagram", + "ranking", + "made", + "drove", + "10%", + "lift", + "Reels", + "spent", + "Facebook", + "video", + "8%", + "globally", + "largest", + "gain", + "years", + "Canada", + "9%", + "increase", + "watch", + "These", + "benefiting", + "advances", + "making", + "full", + "stack", + "Starting", + "doubled", + "length", + "user", + "interaction", + "sequences", + "training", + "richness", + "enabling", + "systems", + "develop", + "deeper", + "understanding", + "interests", + "models", + "significantly", + "speed", + "index", + "posts", + "recommend", + "sooner", + "published", + "applying", + "advanced", + "techniques", + "quickly", + "identify", + "interesting", + "haven", + "engaged", + "lot", + "similar", + "enabled", + "diversity", + "recency", + "recommended", + "day", + "30%", + "double", + "levels", + "ago", + "unlock", + "inventory", + "auto", + "translating", + "dubbing", + "videos", + "viewer", + "local", + "language", + "diverse", + "set", + "Over", + "half", + "watching", + "translated", + "weekly", + "Looking", + "forward", + "expect", + "valuable", + "recommendations", + "continue", + "scaling", + "dimensions", + "size", + "complexity", + "incorporating", + "LLMs", + "deepen", + "platform", + "enable", + "match", + "wider", + "variety", + "aligned", + "At", + "executing", + "longer", + "term", + "generation", + "includes", + "building", + "foundation", + "power", + "organic", + "ads", + "developing", + "LLM", + "recommender", + "focus", + "validating", + "model", + "architectures", + "domains", + "scale", + "future", + "Aside", + "work", + "focused", + "deploying", + "Meta", + "Superintelligence", + "product", + "encouraging", + "began", + "powering", + "responses", + "MSL", + "Muse", + "Spark", + "tests", + "ran", + "leading", + "launch", + "meaningful", + "accelerated", + "week", + "iteration", + "roll", + "digit", + "percent", + "increases", + "sessions", + "direct", + "chat", + "threads", + "standalone", + "app", + "website", + "giving", + "billions", + "latest", + "Overall", + "encouraged", + "momentum", + "research", + "roadmap", + "sharing", + "detail", + "driver", + "increasing", + "efficiency", + "part", + "optimizing", + "level", + "Here", + "enhance", + "show", + "optimal", + "location", + "expanded", + "availability", + "newer", + "surfaces", + "bringing", + "Threads", + "markets", + "progress", + "Status", + "hundreds", + "millions", + "viewing", + "improving", + "businesses", + "To", + "deeply", + "layer", + "tools", + "delivering", + "deploy", + "complex", + "predictive", + "enhancements", + "Lattice", + "modeling", + "learning", + "GEM", + "architecture", + "conversion", + "landing", + "page", + "view", + "addition", + "investing", + "performant", + "inference", + "serving", + "rolling", + "Adaptive", + "Ranking", + "Model", + "improves", + "ROI", + "routing", + "requests", + "compute", + "intensive", + "determines", + "probability", + "coverage", + "support", + "offsite", + "conversions", + "1.6%", + "rates", + "major", + "leveraging", + "make", + "easier", + "manage", + "campaigns", + "creative", + "engage", + "customers", + "assistant", + "fully", + "rolled", + "eligible", + "advertisers", + "supported", + "buying", + "providing", + "personalized", + "resolving", + "account", + "issues", + "surfacing", + "campaign", + "insights", + "optimize", + "Performance", + "testing", + "common", + "resolved", + "20%", + "introducing", + "connectors", + "open", + "beta", + "connect", + "directly", + "agent", + "Marketing", + "API", + "extending", + "agencies", + "analyze", + "Usage", + "gen", + "adoption", + "small", + "medium", + "sized", + "feature", + "3%", + "traction", + "facilitate", + "customer", + "Business", + "AIs", + "SMBs", + "Latin", + "America", + "Indonesia", + "Messenger", + "Asia", + "Pacific", + "10", + "conversations", + "facilitated", + "start", + "ll", + "expand", + "countries", + "adding", + "capabilities", + "invest", + "Value", + "Optimization", + "suite", + "helps", + "maximize", + "Return", + "Ad", + "Spend", + "prioritizing", + "highest", + "solely", + "lowest", + "cost", + "Adoption", + "annual", + "run", + "20B", + "doubling", + "Last", + "touch", + "commerce", + "discover", + "products", + "platforms", + "brands", + "increasingly", + "turning", + "creators", + "promote", + "contributing", + "rapid", + "partnership", + "discovery", + "purchasing", + "happening", + "expanding", + "solutions", + "month", + "affiliate", + "partnerships", + "offering", + "test", + "partners", + "tag", + "participating", + "retailers", + "earn", + "commission", + "makes", + "purchase", + "started", + "real", + "opportunity", + "easily", + "buy", + "incorporate", + "Next", + "discuss", + "approach", + "capital", + "allocation", + "Compute", + "important", + "quality", + "provide", + "capable", + "innovative", + "It", + "critical", + "entering", + "world", + "managing", + "agents", + "generate", + "ideas", + "experiments", + "execute", + "tasks", + "build", + "aggressively", + "meet", + "ensure", + "strategic", + "flexibility", + "coming", + "substantially", + "footprint", + "striking", + "deals", + "supply", + "chain", + "secure", + "components", + "capacity", + "signing", + "online", + "2027", + "allowing", + "multi", + "agreements", + "107", + "step", + "contractual", + "commitments", + "importantly", + "personal", + "As", + "grow", + "remain", + "committed", + "efficiently", + "recently", + "shared", + "internally", + "plan", + "reduce", + "base", + "May", + "leaner", + "move", + "helping", + "substantial", + "financial", + "outlook", + "range", + "58", + "61", + "assumes", + "foreign", + "approximately", + "tailwind", + "current", + "exchange", + "expense", + "capex", + "outlooks", + "162", + "169", + "unchanged", + "prior", + "anticipate", + "125", + "145", + "115", + "135", + "expectations", + "component", + "pricing", + "lesser", + "extent", + "additional", + "landscape", + "remaining", + "quarters", + "13", + "16%", + "Lastly", + "monitor", + "active", + "legal", + "regulatory", + "matters", + "headwinds", + "EU", + "For", + "scrutiny", + "youth", + "related", + "trials", + "scheduled", + "ultimately", + "result", + "material", + "loss", + "closing", + "solid", + "execution", + "core", + "exciting", + "With", + "Krista", + "call", + "questions", + "Operator", + "Thank", + "lines", + "question", + "answer", + "session", + "press", + "star", + "tone", + "phone", + "withdraw", + "Please", + "limit", + "pick", + "handset", + "clarity", + "If", + "streaming", + "today", + "mute", + "computer", + "speakers", + "And", + "Brian", + "Nowak", + "Morgan", + "Stanley", + "ahead" + ], + "keywords": [ + "Meta", + "Q1 2026", + "AI" + ], + "connect_to": [] + } + }, + { + "chunk_id": "86135d4d-0cb6-5f78-afbe-a78c154cb2f7", + "type": "text", + "content": "Thanks for taking my question. Mark, I wanted to ask you just about the level of investment you're making and sort of the signposts you're watching to ensure you're going to generate ROIC on all these investments behind Muse and the other products.\nSo if you could just sort of let us know some of the key factors you're watching over the next 12 to 24 months, whether it's Meta AI, Muse advances, core algorithm, what are you sort of watching foremost just to make sure that you're on the right path to generating healthy ROIC on all this CapEx and infrastructure spend?", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Brian Nowak:", + "metadata": { + "length": 571, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "Thanks", + "taking", + "question", + "Mark", + "wanted", + "level", + "investment", + "making", + "sort", + "signposts", + "watching", + "ensure", + "generate", + "ROIC", + "investments", + "Muse", + "products", + "So", + "key", + "factors", + "12", + "24", + "months", + "Meta", + "AI", + "advances", + "core", + "algorithm", + "foremost", + "make", + "path", + "generating", + "healthy", + "CapEx", + "infrastructure", + "spend" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "aed8925e-845a-5524-a9ea-722d49e70076", + "type": "table", + "content": "
building the businesses around them and decreasing the costs. And this is kind of how we’ve done everything over the last 20 years of running the company, and that is basically the plan.
Susan Li:Mark, on your second question, we aren’t providing a specific outlook for 2027 CapEx. And we are, frankly, undergoing a very dynamic planning process ourselves as we’re working through what our capacity needs will be over the coming years.Our experience so far has been that we have continued to underestimate our compute needs even as we have been ramping capacity significantly as the advances in AI have continued and our teams continue to identify compelling new projects and initiatives. And now too, there are very compelling internal use cases.So our expectation is that compute will become even more central to the business going forward. And it will be critical to determining the quality of the models we develop, the types of products we can introduce, how productive we can be as an organization.So we’re going to continue building out our infrastructure with flexibility in mind. And if we end up not needing as much as we anticipate, we can choose to bring it online more slowly or reduce our spending in future years as we grow into the capacity that we’re building now.
Operator:Your next question comes from the line of Eric Sheridan with Goldman Sachs. Please go ahead.
Eric Sheridan:Thanks so much for taking the question. Maybe if I can build out on one of the topics that was discussed in the prepared remarks. But just the opportunity set that sits in front of the company with respect to putting agentic compute in front of both consumers and enterprises.You’ve long been associated with sort of the consumer landscape. And I am curious about how you’re thinking about extensions of the media engagement parts of your business model and the commerce parts of the business model to become more agentic over time.But what do you see also as the opportunity set that sits in front of you across SMBs and enterprises where historically, you maybe haven’t had as much product velocity? Thanks so much.
Susan Li:Thanks, Eric. So I would say in the near term, obviously the sort of biggest focuses are some of the areas that you mentioned around deepening sort of engagement, obviously with our existing community and user base, making ad experiences meaningfully more personalized, more engaging, more valuable, helping SMBs find and engage with customers across our platform.Those are some of the, I think, most intuitive and adjacent opportunities to the business that we have today. And then, of course, as we are able to build out
", + "path": "tables/table-0 AI Compute Plans.html", + "metadata": { + "length": 2793, + "summary": "table-1\nMeta discusses underestimating AI compute needs, plans flexible infrastructure buildout, and explores agentic opportunities for consumers and enterprises.", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "file_path": "tables/table-0 AI Compute Plans.html", + "keywords": [ + "AI", + "compute", + "infrastructure" + ], + "tokens": [] + } + }, + { + "chunk_id": "c49da176-49fa-5a81-ae97-013949560b26", + "type": "text", + "content": "That's a very technical question for basically where -- the things that we're watching are to make sure that we're on track building leading models and leading products. The formula for our company has always been build experiences that can get to billions of people and focus on monetizing them once you get to scale.\nThat's -- we're seeing a little bit of that here where basically we invest in advance to build leading models, and we convert that into leading products. And then we think that these are going to be some of the most important products that get built over the next decade.\nSo I think just like anything else that we’ve done over time the basic milestones that I look at are around, first, technically, are we delivering the quality to enable a great product; then second, when you have the product, how is it scaling; and then third, you look at the monetization and then you drive up the efficiency of it towards increasing profitability.\nI don't -- I mean like I don't think we have a very precise plan for exactly how each product is going to scale month-over-month or anything like that.\nBut I think we have a sense of the shape of where these things need to be. And I think if you look at the usage of these and the quality of the products and the quality of the models that are out there and the use that other frontier models are getting and the trajectory of that, I'm quite comfortable that, A, the lab that we're building is on track to be a leading lab in the world.\nI think Muse Spark was a very high-quality model. It powers Meta AI which I think is now a world-class assistant.\nWe have an ability to be able to grow that and have a large amount of engagement. And over the coming quarters, we're just going to be tracking how do our next set of training runs go. How do our products scale? How excited are we about the products in the pipeline? Where right now we're very excited.\nAnd then we'll also ramp up monetization over that period of time as well. So I think that those are the set of things that I look at. I think for the kind of specific financial questions, I think Susan can jump in if there's anything more to add.\nOperator: Your next question comes from the line of Mark Shmulik with Bernstein. Please go ahead.\nMark Shmulik: Yes, thanks for taking the questions. Mark, I guess now that we’ve got Muse Spark kind of out there launched, how are you thinking about the team’s focus here kind of divided on to further model training runs and kind of further specialization in that personal intelligence goal versus product launches and kind of shipping more product out the door? And Susan, I guess, kind of as a follow-up to Brian’s question, I know it’s too early to discuss 2027 CapEx.\nBut we’ve had peers mention tonight a potential significant step-up. Any way to think about dimensionalizing kind of how we think about some of the returns or traction this year and how it might affect the 2027 spend? Thanks.\nMark Zuckerberg: I mean I think the roadmap from the team is -- has been pretty consistent. So we have the research team which is focused on scaling increasingly intelligent models with capabilities for the specific things that we're focused on which are business and personal agents.\nSo we're -- we just released our first model, and I talked about in my comments how we're climbing the scaling ladder towards greater capabilities and scale for the models. That work continues.\nWe have our next set of more advanced models in training now. And that is -- that work will, I think, just continue. I mean that's a loop. I don't think we're going to be done with that anytime soon.\nWe're going to have teams that are just consistently focused on training more intelligent and more capable models in the ways that we want. Then we have our product team, and that team is now really unlocked to be able to build things on top of our models because we now have a very strong model.\nSo before this, we had been prototyping a bunch of things using other different models, whether it was our previous older models or kind of using the APIs from other companies. And now we're unlocked to be able to go build things and get them to scale on top of our own models. So I think you will see that over some period of time.\nI tried in my opening remarks to give a bit of a sense of where we're going, but I think that more of the details of that will become clear over the coming months. And I think that these are just both loops that we'll iterate on. We'll keep on iterating on the intelligence.\nWe'll keep on working on building new products and scaling the products. And then as we get to product market fit, we're also going to increasingly focus on more agentic capabilities, enabling agents to help people be more productive, but also agents for businesses and enabling, frankly, those agents to interact with each other and build what we hope will be a thriving commerce ecosystem on our platform.\n\n[tables/table-0 AI Compute Plans.html]\n\nSo I would say some of these are a little bit further out, especially in that latter category of things. Again, the focus is on building personal superintelligence, building a consumer agent that can work for you and help you get things done. That right now is a consumer experience that we're focused on, but we think there will be clear monetization opportunities over time. You can imagine commission structures or a premium offering. And on the business side, we're seeing a large opportunity, of course, around agents and scaling our business AI initiatives.\nI think I mentioned earlier in my remarks that there are over 10 million weekly conversations between people and business AIs on our messaging platforms. That’s up from 1 million at the start of the year, and we’re going to continue expanding globally in Q2. And business AIs today are currently free for most businesses on our messaging apps.\nBut as we make more progress, we expect that we will also work towards establishing a longer-term monetization model. And we'll also consider other services that we can offer to businesses in the future, but we don't have anything more to share today.\nOperator: Your next question comes from the line of Youssef Squali with Truist Securities. Please go ahead.\nYoussef Squali: Great, thank you very much for taking the questions. Maybe one for Mark and one for Susan. Mark, Ray-Ban, Oakley AI glasses continue to perform really well for you guys, but EssilorLuxottica owns and manages a lot more brands.\nWhat are the gating factors to see the launch of additional glasses under these other brands this year? And what would be a successful year for you as you look back at 2026, maybe in terms of units sold?\nAnd then Susan, on that 10% RIF, how much of that is due to efficiencies for maybe AI implementation versus just the need to stay fit? And as you look at your employee needs over time how do you see that growing maybe relative to your overall top line growth? Thank you very much.\nSusan Li: I can go ahead and take both of those. I might answer your second question first. And I'm just trying to make sure I got all of the parts of the question.\nSo in terms of what the sort of kind of the optimal size of the company, I think, over time we don't really know what the optimal size of the company will be in the future. I think there's a lot of change right now with AI capabilities advancing rapidly.\nWe're very focused on leveraging AI tools to substantially increase our productivity, and we're seeing that reflected in the accelerating output from our engineers. And we're generally approaching -- we're approaching this with a bias toward wanting to use these tools to build even more products and services than we would have before.\nAt the same time we're making very significant investments in infrastructure, and we are very focused on continuing to operate efficiently.\nSo I think we will be continuously evaluating how we're structured just to make sure we're best set up to deliver against our priorities over the coming years. So that is, I think, your second question. The first question was about the AI glasses.\nWe have -- we're continuing to see strong growth in, obviously the AI glasses sales over the course of Q1. Demand for the expanded portfolio lineup has generally been quite strong, and we're seeing sales shift now from the prior generation of Ray-Ban Metas to the latest generation which I think speaks to the value of the improved features like extended battery life and features like higher resolution video capture.\nSo we're pretty excited about the progress we've made with glasses. We see strong interest now in the Meta Ray-Ban Displays with the Meta Neural Band.\nSo that's an encouraging sign that there is consumer appetite for display glasses which is kind of the next generation of how this product evolves. And yes, so I think this is an area that we will continue to be -- that we continue to be excited about and are investing in.\nOperator: Your next question comes from the line of Justin Post with Bank of America. Please go ahead.\nJustin Post: Great, thanks for taking my question. Mark, it took about 10 months to get Muse Spark out. I think it’s a pretty good pace. Just help us understand what kind of unlock that is for some of the new products you’re developing? And how is the product cadence going to be over the next nine months on either consumer or business enterprise products built on top of that model?\nMark Zuckerberg: I mean the field is moving pretty quickly. So I mean I'm very happy that we're -- I think the lab that has gone the fastest from standing up the lab to having a very kind of widely accepted as strong model. So I think that's good.\nI take that as a very significant validation of the effort that the team is working well together, that the infrastructure is working, that the effort is on track. And I think that that's basically the main thing that we've learned over the last quarter that I would take away is like where and we started what is this pretty big bet, and it's on track for our plan.\nIn terms of what exactly the cadence is going to be. It's tough for me to say both because I don't really want to share competitively sensitive information and because I think some of the stuff we are more focused on quality than hitting a specific date.\nI mean on the research side, this is research, right? We are trying novel things. You don't exactly know when they're going to land. And on the product side, I think we care a lot about just having -- let me put it this way.\nThere's a lot of agents out there, right, that people are building for different things. And there aren't that many that I would want to give to my mother. And I think getting to like that quality bar is something that I care about more than hitting a specific week for launching or something like that.\nSo -- but with that said, I mean we're in a zone here where the teams don't check in with me like once a quarter, like we make meaningful progress day over day. I think that's part of the fun of developing in this world is that people can make very rapid progress. Small groups of people and teams can make very rapid progress.\nSo I think we're going to see a lot of innovation. The timing of this call is it's good in some ways because the Muse Spark release, I think, was positive. The Meta AI first release, I think, is positive.\nI think that that shows that we're on track. I'm trying to kind of paint a picture of the very high-level direction that we're going in, but I think that the picture is going to come into focus a lot more over the subsequent quarters.\nOperator: Your next question comes from the line of Ross Sandler with Barclays. Please go ahead.\nRoss Sandler: Yes. Mark, just sort of related to that last answer, but there's a lot of new consumer applications kind of cropping up everything from like an OpenClaw to something a little bit more consumer friendly that you would build for your mom, like you said, with like Poke or Dreamer which you recently acquired.\nSo how are these new ideas, I guess, changing your view around the direction that core Meta AI or Dreamer or kind of your overall agentic strategy needs to go? And then the second part of it would be, do you think the lab will stay in this consumer lane? Or do you think you need -- or you want to go down the route that others are going down with code writing and like the recursive self-improvement loop and in that direction kind of in parallel, just thoughts on that? Thank you.\nMark Zuckerberg: Yes. So look, on the OpenClaw and other agents, I think that they give you a very exciting glimpse of what types of things should be possible.\nNow they're pretty rough systems today. And to set up OpenClaw you need to like install a computer locally and then get into a terminal and configure a bunch of things that, again, like there's -- maybe there's hundreds of thousands of people or small numbers of millions of people who can do that. But what we're talking about is delivering personal superintelligence for billions of people around the world.\nSo how do you make a version of that experience that is a lot more polished and dialed and easy and that has all the infrastructure basically done for people already and that just works. And that's kind of what we're focused on, on the consumer side. And I'm really excited about that.\nI think if you had something like that, that worked quite a bit better than those systems and was easy enough that people could just get then I think you go from having something that hundreds of thousands or millions of people are going to use to something that is going to be addressable to billions of people. And that has been our primary focus from day one of the lab is being able to deliver something like that as a product, and I think it’s just going to be very exciting.\nBy the way, the same thing is true for businesses, right? I mean there's the personal version of this but there's also a lot of people's goals are they want to create things, right? They want to create websites. They want to create products. They want to grow their products.\nThese are all things that good agents are going to be able to help people do which I think is partially why this is so exciting. And in my opening comments, I talked about how today we can handle a few goals for people, they're big goals, right? We can help people stay connected with people they care about, learn about the world. These are big things that people care about. But they're not the only things that people care about.\nAnd one of the things that I would love for our products to be able to do is just understand people's goals specifically and then be able to just go work on them for them, and check back in and whenever you have questions that you need answered. So whether those are personal goals or you're trying to create a business or do work.\nI think that this is like -- this is stuff that I think literally every person in the world is going to want some version of it. And also, I think it is something that scales where the more you want to get out of it, I think people are going to also be willing to pay a lot of money to have premium or high compute versions of it. So I think that this is like, it's a very exciting area.\nBut I think what you all should be waiting to see is like whether we can build the version that really like just works and how effective we are at converting people who are using our products into being hundreds of millions and then billions of people using this stuff. And then over time how can we effectively convert that into something that's increasingly profitable by monetizing it and getting the costs down.\nSo I think that that's the roadmap of what we need to do. You asked about whether we're primarily focused on consumers or also recursive self-improvement. I think that we've talked about two main goals for the team.\nI mean one is this kind of agents version vision of what we're doing. The other is that self-improvement is really important because you can't build a leading AI product if you don't have leading models. So -- and you're not going to have leading models in the future if your models can't improve themselves, right?\nSo you're getting to a point where today, the models are still able to learn from people -- and then I think at some point, the models will have to improve themselves. And that's how the growth is going to -- and improvement in the models is going to happen. And if you don't -- if we don't have an ability to do that, then we or anyone else, I think the companies that don't do that are not going to be leading labs, they're not going to produce leading products.\nSo I think that is a table stakes thing that we are focused on. Now does that make us a developer tools company? Not necessarily.\nI mean I'm not against having an API or coding tools or anything like that. But it's not our primary focus. But I actually think people conflate coding with self-improvement more than they should. Coding is one ingredient for the model self improving.\nIt's not the only thing. And we are focused on all of the parts that are going to be necessary for self-improvement in service of the personal superintelligence vision that we have for people and businesses.\nOperator: Your next question comes from the line of Ron Josey with Citigroup. Please go ahead.\nRonald Josey: Great, thanks for taking my question. Mark, maybe a quick follow-up to a prior question around personal agents and business agents. And with Spark Muse now live and more models in development, do you look at the personal agent opportunity which we talked about earlier on the call more of a short-term, medium-term, long-term goal, I'm sure it's a never-ending goal, but when we see a product, is the question short or medium term?\nAnd then Susan, I think the ranking recommendation model improvements are very impressive to see, given the size and scale of both Instagram and Facebook. Could you help us understand just how doubling the length of these interaction sequences can drive greater usage? There’s a thesis out there that maybe some of the rating recommendation improvements are long in the tooth. So it seems if there’s a lot more room to go. So any help there would be helpful. Thank you.\nMark Zuckerberg: I mean I think that the agents work, there's going to be short-term versions of it, but then I think that there's going to be massive upside for delivering more intelligence and more capabilities in the models. And you're kind of seeing this across the industry.\nEach month, each generation of models, they just have more capabilities and can do more things and people absorb it and are able to get more superpowers and it’s awesome. It’s like the most exciting time in the industry.\nSo I think of the agents as the product vehicle for delivering that capability to people. And we certainly -- I think this year is going to be a key period for establishing that as the vehicle for how people are going to use this, but then the model improvement, I think, is going to be something that's going to go on for a very long time. So there's a lot to do here in both the short, medium, and long term.", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Mark Zuckerberg:", + "metadata": { + "length": 19231, + "summary": "Mark Zuckerberg discusses Meta's AI strategy, focusing on building leading models and converting them into products for billions of users. Key milestones include technical quality, product scaling, and monetization efficiency. The lab is on track with Muse Spark model powering Meta AI. Future plans involve personal and business agents, with monetization through commissions or premium offerings. Susan Li addresses workforce efficiency and AI glasses growth. The roadmap includes continuous model improvement and product innovation, with emphasis on delivering polished, easy-to-use AI experiences for consumers and businesses.", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "That", + "technical", + "question", + "basically", + "things", + "watching", + "make", + "track", + "building", + "leading", + "models", + "products", + "The", + "formula", + "company", + "build", + "experiences", + "billions", + "people", + "focus", + "monetizing", + "scale", + "bit", + "invest", + "advance", + "convert", + "And", + "important", + "built", + "decade", + "So", + "ve", + "time", + "basic", + "milestones", + "technically", + "delivering", + "quality", + "enable", + "great", + "product", + "scaling", + "monetization", + "drive", + "efficiency", + "increasing", + "profitability", + "don", + "precise", + "plan", + "month", + "But", + "sense", + "shape", + "usage", + "frontier", + "trajectory", + "comfortable", + "lab", + "world", + "Muse", + "Spark", + "high", + "model", + "It", + "powers", + "Meta", + "AI", + "class", + "assistant", + "We", + "ability", + "grow", + "large", + "amount", + "engagement", + "coming", + "quarters", + "tracking", + "set", + "training", + "runs", + "How", + "excited", + "pipeline", + "Where", + "ll", + "ramp", + "period", + "kind", + "specific", + "financial", + "questions", + "Susan", + "jump", + "add", + "Operator", + "Your", + "line", + "Mark", + "Shmulik", + "Bernstein", + "Please", + "ahead", + "Yes", + "taking", + "guess", + "launched", + "thinking", + "team", + "divided", + "specialization", + "personal", + "intelligence", + "goal", + "versus", + "launches", + "shipping", + "door", + "follow", + "Brian", + "early", + "discuss", + "2027", + "CapEx", + "peers", + "mention", + "tonight", + "potential", + "significant", + "step", + "Any", + "dimensionalizing", + "returns", + "traction", + "year", + "affect", + "spend", + "Thanks", + "Zuckerberg", + "roadmap", + "pretty", + "consistent", + "research", + "focused", + "increasingly", + "intelligent", + "capabilities", + "business", + "agents", + "released", + "talked", + "comments", + "climbing", + "ladder", + "greater", + "work", + "continues", + "advanced", + "continue", + "loop", + "anytime", + "teams", + "consistently", + "capable", + "ways", + "Then", + "unlocked", + "top", + "strong", + "prototyping", + "bunch", + "previous", + "older", + "APIs", + "companies", + "opening", + "remarks", + "give", + "details", + "clear", + "months", + "loops", + "iterate", + "iterating", + "working", + "market", + "fit", + "agentic", + "enabling", + "productive", + "businesses", + "frankly", + "interact", + "hope", + "thriving", + "commerce", + "ecosystem", + "platform", + "category", + "Again", + "superintelligence", + "consumer", + "agent", + "experience", + "opportunities", + "You", + "imagine", + "commission", + "structures", + "premium", + "offering", + "side", + "opportunity", + "initiatives", + "mentioned", + "earlier", + "10", + "million", + "weekly", + "conversations", + "AIs", + "messaging", + "platforms", + "start", + "expanding", + "globally", + "Q2", + "today", + "free", + "apps", + "progress", + "expect", + "establishing", + "longer", + "term", + "services", + "offer", + "future", + "share", + "Youssef", + "Squali", + "Truist", + "Securities", + "Great", + "Maybe", + "Ray", + "Ban", + "Oakley", + "glasses", + "perform", + "guys", + "EssilorLuxottica", + "owns", + "manages", + "lot", + "brands", + "What", + "gating", + "factors", + "launch", + "additional", + "successful", + "back", + "2026", + "terms", + "units", + "sold", + "10%", + "RIF", + "due", + "efficiencies", + "implementation", + "stay", + "employee", + "growing", + "relative", + "growth", + "Thank", + "Li", + "answer", + "parts", + "sort", + "optimal", + "size", + "change", + "advancing", + "rapidly", + "leveraging", + "tools", + "substantially", + "increase", + "productivity", + "reflected", + "accelerating", + "output", + "engineers", + "generally", + "approaching", + "bias", + "wanting", + "At", + "making", + "investments", + "infrastructure", + "continuing", + "operate", + "efficiently", + "continuously", + "evaluating", + "structured", + "deliver", + "priorities", + "years", + "sales", + "Q1", + "Demand", + "expanded", + "portfolio", + "lineup", + "shift", + "prior", + "generation", + "Metas", + "latest", + "speaks", + "improved", + "features", + "extended", + "battery", + "life", + "higher", + "resolution", + "video", + "capture", + "made", + "interest", + "Displays", + "Neural", + "Band", + "encouraging", + "sign", + "appetite", + "display", + "evolves", + "area", + "investing", + "Justin", + "Post", + "Bank", + "America", + "good", + "pace", + "Just", + "understand", + "unlock", + "developing", + "cadence", + "enterprise", + "field", + "moving", + "quickly", + "happy", + "fastest", + "standing", + "widely", + "accepted", + "validation", + "effort", + "main", + "thing", + "learned", + "quarter", + "started", + "big", + "bet", + "In", + "tough", + "competitively", + "sensitive", + "information", + "stuff", + "hitting", + "date", + "land", + "care", + "put", + "There", + "aren", + "mother", + "bar", + "week", + "launching", + "zone", + "check", + "meaningful", + "day", + "part", + "fun", + "rapid", + "Small", + "groups", + "innovation", + "timing", + "call", + "release", + "positive", + "shows", + "paint", + "picture", + "level", + "direction", + "subsequent", + "Ross", + "Sandler", + "Barclays", + "related", + "applications", + "cropping", + "OpenClaw", + "friendly", + "mom", + "Poke", + "Dreamer", + "recently", + "acquired", + "ideas", + "changing", + "view", + "core", + "strategy", + "lane", + "Or", + "route", + "code", + "writing", + "recursive", + "improvement", + "parallel", + "thoughts", + "exciting", + "glimpse", + "types", + "Now", + "rough", + "systems", + "install", + "computer", + "locally", + "terminal", + "configure", + "hundreds", + "thousands", + "small", + "numbers", + "millions", + "talking", + "version", + "polished", + "dialed", + "easy", + "works", + "worked", + "addressable", + "primary", + "By", + "true", + "goals", + "create", + "They", + "websites", + "These", + "partially", + "handle", + "connected", + "learn", + "love", + "specifically", + "answered", + "literally", + "person", + "scales", + "pay", + "money", + "compute", + "versions", + "waiting", + "effective", + "converting", + "effectively", + "profitable", + "costs", + "asked", + "primarily", + "consumers", + "vision", + "improve", + "point", + "happen", + "labs", + "produce", + "table", + "stakes", + "developer", + "Not", + "necessarily", + "API", + "coding", + "conflate", + "Coding", + "ingredient", + "improving", + "service", + "Ron", + "Josey", + "Citigroup", + "Ronald", + "quick", + "live", + "development", + "short", + "medium", + "long", + "ending", + "ranking", + "recommendation", + "improvements", + "impressive", + "Instagram", + "Facebook", + "Could", + "doubling", + "length", + "interaction", + "sequences", + "thesis", + "rating", + "tooth", + "room", + "helpful", + "massive", + "upside", + "industry", + "Each", + "absorb", + "superpowers", + "awesome", + "vehicle", + "capability", + "key" + ], + "keywords": [ + "AI agents", + "model scaling", + "monetization" + ], + "connect_to": [ + { + "target": "aed8925e-845a-5524-a9ea-722d49e70076", + "relation": "embeds", + "ref": "[tables/table-0 AI Compute Plans.html]", + "position": { + "start": 4952, + "end": 4990 + } + } + ] + } + }, + { + "chunk_id": "3ac15566-f61d-5c0b-ac9e-eec9850ddf78", + "type": "text", + "content": "And then on your second question which I think is about the ranking and recommendations improvements that we talked about in our -- that I talked about in my earlier remarks, I think first of all, there is still a lot of room to continue improving recommendations over the rest of the year, and we expect we'll be able to do that to drive additional engagement on both Facebook and Instagram.\nA couple of the things. First, we’re going to continue to improve our data infrastructure that’s going to allow our models to train on more data. And we’re adding more detail to how we describe the content that users have engaged with in the past and scaling up the complexity of our model architecture to take advantage of those larger data sets like using even longer histories of content interactions, and that should all be in service of improving the overall quality of recommendations.\nWe also are focused on making the recommendations even more personalized and more relevant to any given users interest. There’s work we’re doing to redesign our content retrieval system to show more content that matches the full range of a user’s interests and to tailor the diversity of the topics we recommend to the broadness of someone’s interests.\nSo someone with particularly concentrated interests might see relatively more of that content while people with a broader set of interests might see kind of a greater range in the topics that we show them. And then finally, we're continuing to make improvements to our sort of LLM based tune-your-algorithm features that allow users to provide more granular natural language feedback on what they want to see more of or less of in their feed.\nSo the sort of the kind of the sequence length which is the thing that you called out is one of really many improvements we made in Q1, and there is a big roadmap of further improvements going forward.", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Susan Li:", + "metadata": { + "length": 1882, + "summary": "The passage discusses ongoing improvements to recommendation systems on Facebook and Instagram. Key areas include enhancing data infrastructure for better model training, adding more detail to user engagement descriptions, and scaling model architecture. Efforts focus on personalization through redesigned content retrieval to match user interests and tailor topic diversity. Additionally, LLM-based features allow users to give natural language feedback. These improvements aim to drive engagement and are part of a broader roadmap beyond Q1.", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "And", + "question", + "ranking", + "recommendations", + "improvements", + "talked", + "earlier", + "remarks", + "lot", + "room", + "continue", + "improving", + "rest", + "year", + "expect", + "ll", + "drive", + "additional", + "engagement", + "Facebook", + "Instagram", + "couple", + "things", + "First", + "improve", + "data", + "infrastructure", + "models", + "train", + "adding", + "detail", + "describe", + "content", + "users", + "engaged", + "past", + "scaling", + "complexity", + "model", + "architecture", + "advantage", + "larger", + "sets", + "longer", + "histories", + "interactions", + "service", + "quality", + "We", + "focused", + "making", + "personalized", + "relevant", + "interest", + "There", + "work", + "redesign", + "retrieval", + "system", + "show", + "matches", + "full", + "range", + "user", + "interests", + "tailor", + "diversity", + "topics", + "recommend", + "broadness", + "So", + "concentrated", + "people", + "broader", + "set", + "kind", + "greater", + "finally", + "continuing", + "make", + "sort", + "LLM", + "based", + "tune", + "algorithm", + "features", + "provide", + "granular", + "natural", + "language", + "feedback", + "feed", + "sequence", + "length", + "thing", + "called", + "made", + "Q1", + "big", + "roadmap", + "forward" + ], + "keywords": [ + "Recommendations", + "Personalization", + "Data Infrastructure" + ], + "connect_to": [] + } + }, + { + "chunk_id": "0c897e87-efc8-593c-9088-1315a13c74a2", + "type": "text", + "content": "Your next question comes from the line of Doug Anmuth with JPMorgan. Please go ahead,", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Operator:", + "metadata": { + "length": 85, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "Your", + "question", + "line", + "Doug", + "Anmuth", + "JPMorgan", + "Please", + "ahead" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7fd2203c-dce7-58e5-a6ee-2c54187fc018", + "type": "text", + "content": "Thanks so much for taking the questions. Mark, how do you think about the step up as you go from leveraging smaller models in the ad business to Muse Spark and future large language models going forward, what are some of the key unlocks across engagement and monetization? And then on Manus, can you just talk at all about the strategic importance and the role in developing agentic products for Meta and then just current status around the tech and the deal.", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Douglas Anmuth:", + "metadata": { + "length": 459, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "Thanks", + "taking", + "questions", + "Mark", + "step", + "leveraging", + "smaller", + "models", + "ad", + "business", + "Muse", + "Spark", + "future", + "large", + "language", + "forward", + "key", + "unlocks", + "engagement", + "monetization", + "And", + "Manus", + "talk", + "strategic", + "importance", + "role", + "developing", + "agentic", + "products", + "Meta", + "current", + "status", + "tech", + "deal" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b14c8a21-e918-5e9d-95d1-b7e50f4681c7", + "type": "text", + "content": "I'll take that question. On Manus, we're still working through the details. So we don't have an update right now.\nOn your first question which is about going from leveraging smaller models in the ads business to kind of the ads models growing. There’s already some work underway, and I think I alluded to some of this in my earlier remarks, even kind of in the current landscape of the ads roadmap, where we’re basically trying to advance the architecture here to allow sort of -- to allow us to leverage the abilities of larger models.\nHistorically, we haven't used larger model architectures like GEM for inference because their size and complexity would make them too cost prohibitive. And the way we drive performance from those models is by using them to transfer knowledge to smaller, more lightweight models that are used at run time.\nThe inference models are bound by strict latency requirements since they need to find the right ad within milliseconds, and that has, again, historically prevented us from meaningfully sizing up -- scaling up their size and complexity.\nBut in the second half of last year, we introduced a new adaptive ranking model which enables us to leverage LLM scale model complexity of a trillion parameters, and we made advances in the model architecture and codesign the system with the underlying silicon, so it maintains the sub-second speed that is required to serve ads at scale.\nWe also developed an approach that intelligently routes request more compute-intensive inference models if it determines that there is a higher probability of conversion and that lets us drive both better performance and increased inference ROI.\nSo there's a lot of work being done there before we even sort of incorporate more of the LLM work into our underlying ads ranking models.", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Susan Li:_2", + "metadata": { + "length": 1800, + "summary": "The speaker discusses advancements in ads ranking models, moving from smaller to larger architectures like LLMs with trillion parameters. Historically, cost and latency constraints prevented using large models for inference, but a new adaptive ranking model and system co-design with silicon now enable sub-second ad serving. Intelligent request routing improves performance and inference ROI.", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "ll", + "question", + "On", + "Manus", + "working", + "details", + "So", + "don", + "update", + "leveraging", + "smaller", + "models", + "ads", + "business", + "kind", + "growing", + "There", + "work", + "underway", + "alluded", + "earlier", + "remarks", + "current", + "landscape", + "roadmap", + "basically", + "advance", + "architecture", + "sort", + "leverage", + "abilities", + "larger", + "Historically", + "haven", + "model", + "architectures", + "GEM", + "inference", + "size", + "complexity", + "make", + "cost", + "prohibitive", + "And", + "drive", + "performance", + "transfer", + "knowledge", + "lightweight", + "run", + "time", + "The", + "bound", + "strict", + "latency", + "requirements", + "find", + "ad", + "milliseconds", + "historically", + "prevented", + "meaningfully", + "sizing", + "scaling", + "But", + "half", + "year", + "introduced", + "adaptive", + "ranking", + "enables", + "LLM", + "scale", + "trillion", + "parameters", + "made", + "advances", + "codesign", + "system", + "underlying", + "silicon", + "maintains", + "speed", + "required", + "serve", + "We", + "developed", + "approach", + "intelligently", + "routes", + "request", + "compute", + "intensive", + "determines", + "higher", + "probability", + "conversion", + "lets", + "increased", + "ROI", + "lot", + "incorporate" + ], + "keywords": [ + "ads models", + "LLM", + "inference" + ], + "connect_to": [] + } + }, + { + "chunk_id": "2c0baf12-cf4c-5762-97bb-a9347aee6765", + "type": "text", + "content": "We have time for one more question, Ken Gawrelski with Wells Fargo. Your line is open.", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Operator:_2", + "metadata": { + "length": 86, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "We", + "time", + "question", + "Ken", + "Gawrelski", + "Wells", + "Fargo", + "Your", + "line", + "open" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "34220948-ddf8-545e-8872-7ea394f1bd12", + "type": "text", + "content": "i: Thank you very much. Two, if I may. First, if I -- you talked on the Muse Spark launch. You've talked about two categories or two verticals. You talked about health and wellness and shopping. Can I dive a little bit -- ask you to dive a little deeper into the latter on the shopping and commerce side?\nAnd maybe if you could -- were there any learnings in the 2021, '22 phase where you pushed deeper into commerce on Instagram and on Facebook? Any learnings from that period that you might apply? Is there an opportunity for a next-gen marketplace-type business in commerce?\nAnd then the second, please, maybe, Susan, can you talk a little bit about -- based on your model improvements and the content recommendations, where -- how much visibility do you think you have to kind of the growth trajectory on the core business?\nYou continue to grow at basically double the pace of the industry despite being a very large share of the industry. Could you just talk about a little bit about your visibility into that continued performance? Thank you.\nMark Zuckerberg: Yes. So I might give you a somewhat loftier answer to the question. You're asking about shopping.\nI think it's sort of an interesting example of the way in which the work that we're doing is different than what I think others are doing out there. These products, they -- AI agents get better when you fully optimize the stack.\nThat's why we believe that we need to be a company that builds frontier models in addition to building the agents. And then in order to do that, you, of course, need to build your infrastructure in order to be able to do that well.\nSo we're undertaking this large investment to be able to do that top to bottom. And I think a lot of the way to think about the investment that we're making is a bet that the individual things that people care about and that people are going to be more important in the future. And that's sort of like -- and I think it should be a pretty obvious thing to say.\nBut I think so much of the rhetoric around AI in the industry is around like a company trying to build some kind of centralized thing that like does all the productive work in society in some way or something like that. And that just is very different from how we see the world.\nLike our vision for the future is one where society makes progress by individuals pursuing their own aspirations. And some people care about big grand things like curing diseases. And a lot of people care about personal things like finding the right shirt for my daughter. And I just think that we want to -- we're going to build things that help deliver this vision for personal agents for people.\nAnd I think that part of the lane and what is interesting and differentiated about what we're doing is that that's just so different from how I hear everyone else talking about the work that we're doing. So even though I think some of these ideas, they seem like they should be so obvious.\nI actually think that our approach of trying to empower individuals and building consumer things is just in the details extremely different from what others are doing. And shopping might be one kind of specific example that I think is going to have interest in commercial implications. And I think people -- consumers are going to like it.\nBut I don’t hear any other labs out there talking about how they’re building an AI that’s really good at shopping. And I think that the reason for that is like not because shopping is the most important thing by itself, but because like empowering people to do the things that matter in their lives, whether that’s local or understanding social context, or shopping or personal health things or understanding what’s going on around them visually which is going to be really important on the glasses. These are all elements of the personal super intelligence vision.\nI think like a lot of this, and when you're thinking about kind of the investment in Meta over time I think you should think about it as coming down to these set of values around what do we want AI to do in society. And if what you want it to do is empower individuals and build a world where the AI is in service to individual's goals, then that is what we are going to build, and I think it's going to be incredibly valuable.\nSusan Li: Gosh, I almost wish we could end on that answer, but I will answer the second question which I think kind of has two versions.\nOne is a version of like what's the revenue outlook? And obviously we gave the Q2 guide which embeds, I think, both a range of kind of macro outcomes, but also the work that we've -- the ongoing work that we have to continue improving both the sort of usage and engagement on our family of apps and then our ability to continue making the ads better and more performant.\nI think the second question is maybe more of a -- the second version of that question is more of a higher-level question about kind of the overall trajectory of the roadmap here. And one of the things I will say, having been working on this for a very long time I'm always really impressed by the team's ability to continue to advance the state of the art here. And our planning process now is, I think, really fine-tuned around this.\nSo I’ve mentioned on a couple of calls, the budgeting process in which we run a very sort of ROI-based process to make sure that we are funding all of the ads initiatives that we think will drive growth in future years. And that’s something that is both quite dialed in.\nAnd I think that our ability to measure the impact of that has been pretty robust, and it's been a very important driver of our ads revenue growth, and that continues to be a process that again, we ran in this past budget and -- as far as we can -- as we have line of sight, we feel good about the investment opportunities ahead of us.\nKenneth Dorell: Great. Thank you everyone, for joining us today. We look forward to speaking with you again soon.\nOperator: This concludes today's conference call. Thank you for joining. And you may now disconnect.", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Kenneth Gawrelski:", + "metadata": { + "length": 6052, + "summary": "Mark Zuckerberg discusses Meta's AI strategy, emphasizing a vision of empowering individuals with personal AI agents rather than building a centralized AI. He highlights shopping and commerce as key verticals, noting Meta's unique focus on consumer AI for tasks like finding the right shirt. Susan Li adds that Meta's ad revenue growth is driven by ROI-based budgeting and continuous model improvements, with strong visibility into future investment opportunities.", + "page_nums": [], + "document_top_summary": "This document includes: Kenneth Dorell, Director, Investor Relations, Mark Zuckerberg, CEO, Susan Li, CFO", + "tokens": [ + "Thank", + "Two", + "First", + "talked", + "Muse", + "Spark", + "launch", + "You", + "ve", + "categories", + "verticals", + "health", + "wellness", + "shopping", + "Can", + "dive", + "bit", + "deeper", + "commerce", + "side", + "And", + "learnings", + "2021", + "22", + "phase", + "pushed", + "Instagram", + "Facebook", + "Any", + "period", + "apply", + "Is", + "opportunity", + "gen", + "marketplace", + "type", + "business", + "Susan", + "talk", + "based", + "model", + "improvements", + "content", + "recommendations", + "visibility", + "kind", + "growth", + "trajectory", + "core", + "continue", + "grow", + "basically", + "double", + "pace", + "industry", + "large", + "share", + "Could", + "continued", + "performance", + "Mark", + "Zuckerberg", + "Yes", + "So", + "give", + "loftier", + "answer", + "question", + "sort", + "interesting", + "work", + "These", + "products", + "AI", + "agents", + "fully", + "optimize", + "stack", + "That", + "company", + "builds", + "frontier", + "models", + "addition", + "building", + "order", + "build", + "infrastructure", + "undertaking", + "investment", + "top", + "bottom", + "lot", + "making", + "bet", + "individual", + "things", + "people", + "care", + "important", + "future", + "pretty", + "obvious", + "thing", + "But", + "rhetoric", + "centralized", + "productive", + "society", + "world", + "Like", + "vision", + "makes", + "progress", + "individuals", + "pursuing", + "aspirations", + "big", + "grand", + "curing", + "diseases", + "personal", + "finding", + "shirt", + "daughter", + "deliver", + "part", + "lane", + "differentiated", + "hear", + "talking", + "ideas", + "approach", + "empower", + "consumer", + "details", + "extremely", + "specific", + "interest", + "commercial", + "implications", + "consumers", + "don", + "labs", + "good", + "reason", + "empowering", + "matter", + "lives", + "local", + "understanding", + "social", + "context", + "visually", + "glasses", + "elements", + "super", + "intelligence", + "thinking", + "Meta", + "time", + "coming", + "set", + "values", + "service", + "goals", + "incredibly", + "valuable", + "Li", + "Gosh", + "end", + "versions", + "One", + "version", + "revenue", + "outlook", + "gave", + "Q2", + "guide", + "embeds", + "range", + "macro", + "outcomes", + "ongoing", + "improving", + "usage", + "engagement", + "family", + "apps", + "ability", + "ads", + "performant", + "higher", + "level", + "roadmap", + "working", + "long", + "impressed", + "team", + "advance", + "state", + "art", + "planning", + "process", + "fine", + "tuned", + "mentioned", + "couple", + "calls", + "budgeting", + "run", + "ROI", + "make", + "funding", + "initiatives", + "drive", + "years", + "dialed", + "measure", + "impact", + "robust", + "driver", + "continues", + "ran", + "past", + "budget", + "line", + "sight", + "feel", + "opportunities", + "ahead", + "Kenneth", + "Dorell", + "Great", + "joining", + "today", + "We", + "forward", + "speaking", + "Operator", + "This", + "concludes", + "conference", + "call", + "disconnect" + ], + "keywords": [ + "AI agents", + "shopping", + "personal AI" + ], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/doc_nav.json b/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/doc_nav.json new file mode 100644 index 000000000..df7223dea --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/doc_nav.json @@ -0,0 +1,119 @@ +{ + "version": "1.0", + "file_name": "Meta Q1 2026 Earnings Call Transcript.pdf", + "stats": { + "total_chunks": 13, + "text_chunks": 12, + "image_chunks": 0, + "table_chunks": 1, + "max_depth": 2 + }, + "sections": [ + { + "title": "Root", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf", + "level": 1, + "summary": "Meta Platforms, Inc. (META) First Quarter 2026 Results Conference Call April 29 $^{th}$ , 2026", + "chunk_count": 1, + "children": [] + }, + { + "title": "Kenneth Dorell, Director, Investor Relations", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Kenneth Dorell, Director, Investor Relations", + "level": 1, + "summary": "Thank you. Good afternoon and welcome to Meta Platforms' first quarter 2026 earnings conference call. Joining me today to discuss our results are Mark Zuckerberg, CEO and Susan Li, CFO. Our remarks today will include forward-looking statements, which are based on assumptions as of today. Actual results may differ materially as a result of various factors including those set forth in today's earnings press release, and in our annual report on Form 10-K filed with the SEC. We undertake no obligation to update any forward-looking statement. During this call we will present both GAAP and certain non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in today's earnings press release. The earnings press release and an accompanying investor presentation are available on our website at investor.atmeta.com. And now, I'd like to turn the call over to Mark.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Mark Zuckerberg, CEO", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Mark Zuckerberg, CEO", + "level": 1, + "summary": "The speaker discusses Meta's strong quarterly performance with over 3.5 billion daily users across apps, highlighting the release of the Muse family of models and Meta AI's advancements. They emphasize a vision of AI amplifying human potential rather than replacing people, focusing on personal and business agents. The talk covers improvements in recommendation systems, increased infrastructure investment, growth in AI glasses usage, and the transformative impact of AI on work efficiency at Meta.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Susan Li, CFO", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO", + "level": 1, + "summary": "Meta reported Q1 2026 results with total revenue of $56.3 billion, up 33% year-over-year. Family of Apps revenue was $55.9 billion, with ad revenue of $55.0 billion. Daily active people across Family of Apps was 3.56 billion in March. Operating income was $22.9 billion with a 41% margin. Net income was $26.8 billion, or $10.44 per share, including a tax benefit. Capital expenditures were $19.8 billion. The company highlighted AI-driven improvements in content recommendations, ad performance, and product experiences, including gains from Meta AI and Muse Spark. Revenue guidance for Q2 2026 is $58-61 billion, with full-year expense outlook of $162-169 billion and capex of $125-145 billion.", + "chunk_count": 9, + "children": [ + { + "title": "Brian Nowak:", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Brian Nowak:", + "level": 2, + "summary": "Thanks for taking my question. Mark, I wanted to ask you just about the level of investment you're making and sort of the signposts you're watching to ensure you're going to generate ROIC on all these investments behind Muse and the other products. So if you could just sort of let us know some of the key factors you're watching over the next 12 to 24 months, whether it's Meta AI, Muse advances, core algorithm, what are you sort of watching foremost just to make sure that you're on the right path to generating healthy ROIC on all this CapEx and infrastructure spend?", + "chunk_count": 1, + "children": [] + }, + { + "title": "Mark Zuckerberg:", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Mark Zuckerberg:", + "level": 2, + "summary": "Mark Zuckerberg discusses Meta's AI strategy, focusing on building leading models and converting them into products for billions of users. Key milestones include technical quality, product scaling, and monetization efficiency. The lab is on track with Muse Spark model powering Meta AI. Future plans involve personal and business agents, with monetization through commissions or premium offerings. Susan Li addresses workforce efficiency and AI glasses growth. The roadmap includes continuous model improvement and product innovation, with emphasis on delivering polished, easy-to-use AI experiences for consumers and businesses.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Susan Li:", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Susan Li:", + "level": 2, + "summary": "The passage discusses ongoing improvements to recommendation systems on Facebook and Instagram. Key areas include enhancing data infrastructure for better model training, adding more detail to user engagement descriptions, and scaling model architecture. Efforts focus on personalization through redesigned content retrieval to match user interests and tailor topic diversity. Additionally, LLM-based features allow users to give natural language feedback. These improvements aim to drive engagement and are part of a broader roadmap beyond Q1.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Operator:", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Operator:", + "level": 2, + "summary": "Your next question comes from the line of Doug Anmuth with JPMorgan. Please go ahead,", + "chunk_count": 1, + "children": [] + }, + { + "title": "Douglas Anmuth:", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Douglas Anmuth:", + "level": 2, + "summary": "Thanks so much for taking the questions. Mark, how do you think about the step up as you go from leveraging smaller models in the ad business to Muse Spark and future large language models going forward, what are some of the key unlocks across engagement and monetization? And then on Manus, can you just talk at all about the strategic importance and the role in developing agentic products for Meta and then just current status around the tech and the deal.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Susan Li:_2", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Susan Li:_2", + "level": 2, + "summary": "The speaker discusses advancements in ads ranking models, moving from smaller to larger architectures like LLMs with trillion parameters. Historically, cost and latency constraints prevented using large models for inference, but a new adaptive ranking model and system co-design with silicon now enable sub-second ad serving. Intelligent request routing improves performance and inference ROI.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Operator:_2", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Operator:_2", + "level": 2, + "summary": "We have time for one more question, Ken Gawrelski with Wells Fargo. Your line is open.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Kenneth Gawrelski:", + "path": "META-Q1-2026-Earnings-Call-Transcript.pdf/Susan Li, CFO/Kenneth Gawrelski:", + "level": 2, + "summary": "Mark Zuckerberg discusses Meta's AI strategy, emphasizing a vision of empowering individuals with personal AI agents rather than building a centralized AI. He highlights shopping and commerce as key verticals, noting Meta's unique focus on consumer AI for tasks like finding the right shirt. Susan Li adds that Meta's ad revenue growth is driven by ROI-based budgeting and continuous model improvements, with strong visibility into future investment opportunities.", + "chunk_count": 1, + "children": [] + } + ] + } + ], + "resources": { + "images": [], + "tables": [ + { + "path": "tables/table-0 AI Compute Plans.html", + "summary": "table-1 Meta discusses underestimating AI compute needs, plans flexible infrastructure buildout, and explores agentic opportunities for consumers and enterprises." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/full.md b/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/full.md new file mode 100644 index 000000000..373e7fc1f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/full.md @@ -0,0 +1,419 @@ +Meta Platforms, Inc. (META) +First Quarter 2026 Results Conference Call +April 29 $^{th}$ , 2026 + +## Kenneth Dorell, Director, Investor Relations + +Thank you. Good afternoon and welcome to Meta Platforms' first quarter 2026 earnings conference call. Joining me today to discuss our results are Mark Zuckerberg, CEO and Susan Li, CFO. + +Our remarks today will include forward-looking statements, which are based on assumptions as of today. Actual results may differ materially as a result of various factors including those set forth in today's earnings press release, and in our annual report on Form 10-K filed with the SEC. We undertake no obligation to update any forward-looking statement. + +During this call we will present both GAAP and certain non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in today's earnings press release. The earnings press release and an accompanying investor presentation are available on our website at investor.atmeta.com. + +And now, I'd like to turn the call over to Mark. + +## Mark Zuckerberg, CEO + +Hey everyone, thanks for joining today. We had a strong quarter for our community, our business, and our progress towards AI. + +More than 3.5 billion people use at least one of our apps every day. We saw a small decrease in total family dailies due to internet outages in Iran and blocks in Russia, but otherwise trends across our apps are strong. Daily and monthly actives on Instagram and Facebook continue to grow, with video driving all-time high engagement across both apps. WhatsApp continues to see strong momentum too, including in the US. And Threads continues on its trajectory to be the leading app in its category. + +Our biggest milestone so far this year has been the release of our Muse family of models and our first model, Muse Spark, along with a significantly upgraded new version of Meta AI. This was the first release from Meta Superintelligence Labs, and it shows that our work is on track to build a leading lab. + +Over the past 10 months, we have built the strongest research team in the industry, and established the scientific and technical foundations to scale very advanced models. Spark is just one step on that scaling ladder and we are already training even more advanced models. But Spark has already made Meta AI a world-class assistant that leads in several areas related to our vision of personal superintelligence, including visual understanding, health, shopping, social content, local, creating games, and more. We're hearing very positive feedback on it so far. We've seen large increases in Meta AI use since releasing the updates, and the Meta AI app has consistently been near the top of the app stores as well. + +Now that we have a strong model we can develop more novel products as well. Since I first wrote about our vision for personal superintelligence last year, we've been focused on delivering + +personal and business agents to billions of people around the world. Our goal is not just to deliver Meta AI as an assistant, but to deliver agents that can understand your goals and then work day and night to help you achieve them. + +My view of AI is very different from many others in the industry. I hear a lot of people out there talk about how AI is going to replace people. Instead, I think that AI is going to amplify people's ability to do what you want -- whether that's to improve your health, your learning, your relationships, your ability to achieve your personal and career goals and more. My view is that human progress has always been driven by people pursuing their individual aspirations, and I believe that this will continue to be true in the future. People will be more important in the future, not less. Meta believes in empowering individuals. Those are the kinds of products we're going to build, and I believe they're going to be some of the most important and valuable products of all time. + +We are building a personal agent focused on helping people achieve the diverse goals in their lives. We are also building a business agent focused on helping entrepreneurs and businesses across the world use our tools and others to grow their efforts, reach new customers, and serve existing customers better. These agents will work together to form an ecosystem, and whether you use our personal or business agents to achieve your goals, I believe that the future will see a massive increase in entrepreneurship from people creating new things that they've always wanted to exist but previously didn't have the tools to bring into the world. We're already testing an early version of business AIs, and weekly conversations have grown 10x since the start of this year. + +We're also working on using Spark and our upcoming models to improve our recommendation systems and core business in Facebook, Instagram, and ads. Right now, our apps primarily help people accomplish three important goals: connecting with people, learning about the world, and entertainment. But we've always wanted our apps to understand more of people's goals so we can help improve their lives in all the ways that they want. These new AI models will let us understand this in more detail. So instead of just looking at statistical patterns of what types of people engage with what content, for the first time in Meta's history we're going to be able to develop a first-principles understanding of what you care about and what each piece of content in our system is about so that way we can show you more useful things for what you're trying to accomplish. We'll also be able to create personalized content specifically for people to help you achieve your goals as well. + +Since our recommendation systems are operating at such large scale, we'll phase in this new research and technology over time. But the trend over the last few years seems clear that we are seeing an increasing return on the amount that we can improve engagement for people and value for advertisers. This encourages us to continue investing heavily in what we expect will provide increasing value over the coming years as well. + +On that note, we are increasing our infrastructure capex forecast for this year. Most of that is due to higher component costs, particularly memory pricing. But every sign that we're seeing in our own work and across the industry gives us confidence in this investment. That said, we are very focused on increasing the efficiency of our investments. And as part of that, we're rolling out more than 1GW of our own custom silicon that we're developing with Broadcom as well as significant amounts of AMD chips to compliment the new Nvidia systems that we're rolling out as well. One of the primary goals of our Meta Compute initiative is to lead the industry in efficiency of building compute, and we expect that will be a strategic advantage over time. + +Talking about building physical goods at scale, our AI glasses continue to perform well with the number of people using them daily tripling year-over-year. This continues to be one of the fastest-growing categories of consumer electronics ever. We released Ray-Ban Meta Optics this quarter, designed for all-day wear rather than primarily as sunglasses. And building on our release of Oakley last year, we have some exciting new partnerships and styles that I think are going to have the potential to reach even more people coming later this year. + +All of our glasses are designed to easily update to use our newest AI models and features. I'm also really excited to see the glasses evolve from being able to answer questions to being able to be a personal agent that's with you all day long, helping you remember things and achieve your goals. + +Beyond glasses, I am excited for more of our metaverse efforts to be powered by the AI models we're training as well. We remain the biggest investor in the VR space across the industry, but we are focused on making our VR business sustainable as we invest more in other areas like AI and glasses. + +Before wrapping, I want to talk for a moment about how AI is transforming our work. We're seeing more and more examples where one or two people are building something in a week that would have previously taken dozens of people months. I want to make sure that Meta is the best place in the world for these types of people to come and make an impact. We're building the next evolution of our company around these people. There's a lot that we can do to enable this: building the best infrastructure for creating and delivering products at scale, streamlining our teams so they aren't bigger than they need to be, recognizing and rewarding the people who are having outsized impacts, and setting ourselves up to try many more ideas and take on many new projects in the future. Of course we'll continue pushing to increase our efficiency as well, but overall I think the future is about building many more higher quality things than we've ever built before. + +Alright, that is what I wanted to cover today. We are living through a historic technological transformation. We are among the few companies positioned to shape the future, and we are on track to do that. I'm looking forward to delivering personal superintelligence to billions of people. And as always, I'm grateful for the hard work of our teams, and to all of you for being on this journey with us. + +## Susan Li, CFO + +Thanks Mark and good afternoon everyone. + +Let's begin with our segment results. All comparisons are on a year-over-year basis unless otherwise noted. + +We estimate 3.56 billion people used at least one of our Family of Apps on a daily basis in March, which declined slightly from December due to internet disruptions in Iran and a restriction on access to WhatsApp in Russia. Absent these impacts, growth in Family Daily Active People would have been positive quarter-over-quarter. + +Q1 Total Family of Apps revenue was \$55.9 billion, up 33% year-over-year. + +Q1 Family of Apps ad revenue was \$55.0 billion, up 33% or 29% on a constant currency basis. + +In Q1, the total number of ad impressions served across our services increased 19%. Impression growth was healthy across all regions, driven primarily by growth in engagement and users, as well as ad load optimizations. The global average price per ad increased 12% year-over-year in Q1, with broad-based growth as we benefited from ad performance improvements, better macro conditions versus Q1 of last year, and currency tailwinds in international regions. This was partially offset by strong impression growth, including from lower monetizing regions. + +Family of Apps other revenue was \$885 million, up 74%, driven primarily by WhatsApp paid messaging and subscriptions revenue. + +Within our Reality Labs segment, Q1 revenue was \$402 million, down 2% year-over-year due to lower Quest headset sales, which were partially offset by continued strong growth in AI glasses revenue. + +Moving now to our consolidated results. + +Q1 total revenue was \$56.3 billion, up 33% or 29% on a constant currency basis. + +Q1 total expenses were \$33.4 billion, up 35% compared to last year. Year-over-year growth was driven mainly by infrastructure costs and employee compensation. + +The growth in infrastructure costs was due to higher depreciation, data center operating costs, and third party cloud spend. + +The growth in employee compensation was driven by technical hires we’ve added over the past year, particularly AI talent. + +We ended Q1 with over 77,900 employees, down 1% from Q4 as the impact of headcount optimization efforts in certain functions was partially offset by hiring in priority areas of monetization and infrastructure. + +First quarter operating income was \$22.9 billion, representing a 41% operating margin. + +Q1 interest and other income was -\$1.1 billion, driven by unrealized losses on our equity investments. + +Our tax rate for the quarter was -23%, which was favorably impacted by a tax benefit of \$8.03 billion. This benefit partially relieves the \$15.93 billion non-cash tax charge we recorded in the third quarter of 2025, which reflects updated guidance from the US Treasury issued in February 2026 regarding the tax treatment of previously capitalized R&D expenditures in the United States. Absent the tax benefit, our Q1 tax rate would have been 14%. + +Net income was \$26.8 billion or \$10.44 per share. Absent the tax benefit, our net income and EPS would have been \$18.7 billion and \$7.31, respectively. + +Capital expenditures, including principal payments on finance leases, were \$19.8 billion, driven by investments in servers, data centers, and network infrastructure. + +Free cash flow was \$12.4 billion. We ended the quarter with \$81.2 billion in cash and marketable securities and \$58.7 billion in debt. + +Turning now to the business performance. There are two primary factors that drive our revenue performance: our ability to deliver engaging experiences for our community, and our effectiveness at monetizing that engagement over time. + +On the first, we're continuing to see significant gains from our content recommendation initiatives. + +On Instagram, the ranking improvements that we made in Q1 drove a 10% lift in Reels time spent. + +On Facebook, total video time increased more than 8% globally in Q1, the largest quarter-over-quarter gain in four years. Within the US & Canada, ranking improvements we made drove a 9% increase in video watch time on Facebook in Q1. + +These gains are benefiting from advances we're making across the full stack. + +Starting with data, we doubled the length of user interaction sequences we use for training on Instagram in Q1 and increased the richness of how each user interaction is described, enabling our systems to develop a deeper understanding of user interests. + +Within our models, we’ve significantly increased the speed with which our ranking models index new posts, which is enabling us to recommend them sooner after they are published. We’re also applying more advanced content understanding techniques, which is enabling us to quickly identify posts that may be interesting to someone, even if they haven’t engaged with a lot of similar content. These and other improvements have enabled us to increase the diversity and recency of recommended content, with same-day posts now representing more than 30% of recommended Reels on both Instagram and Facebook, more than double the levels one year ago. + +We're also using AI to unlock more inventory by auto-translating and dubbing videos into a viewer's local language, enabling us to recommend a more diverse set of content. Over half a billion users on each of Facebook and Instagram are now watching AI-translated videos weekly. + +Looking forward, we're making several investments we expect will deliver more valuable recommendations. + +This year, we will continue scaling up our models in several dimensions, including their size and complexity, while incorporating LLMs to deepen content understanding across our platform. This will enable us to better match people to a wider variety of content aligned to their interests. + +At the same time, we are executing on our longer-term efforts to develop the next generation of our recommendation systems. This includes building foundation models that power organic content and ads recommendations, as well as developing LLM-based recommender systems. Our focus this year is validating the model architectures and techniques in these domains before we scale them out in future years. + +Aside from our recommendations work, we are focused on deploying the models from Meta Superintelligence Labs to enable a new set of product experiences. + +We're seeing encouraging results within Meta AI since we began powering responses with the first model from MSL, Muse Spark. In tests we ran leading up to the launch, we saw meaningful + +engagement gains that accelerated week-over-week with each new iteration of the model. We're seeing similar gains within Meta AI following the broad roll out of our new model, with double-digit percent increases in Meta AI sessions per user. Muse Spark is now powering Meta AI in direct chat threads across our Family of Apps, as well as the standalone Meta AI app and website, giving billions of people globally access to our latest model. + +Overall, we're very encouraged by the momentum within our research and product roadmap and look forward to sharing more detail on what we're building over the course of this year. + +Turning to the second driver of our revenue performance: increasing monetization efficiency. + +The first part of this work is optimizing the level of ads within organic engagement. + +Here, we continue to enhance our systems to show ads at the optimal time and location. In Q1, we also expanded availability of ads on our newer surfaces, including bringing ads on Threads to people in more markets. On WhatsApp, we are making good progress with the roll out of ads in Status, with hundreds of millions of people now viewing them daily. + +Moving to the second part of increasing monetization efficiency: improving performance for the businesses who use our services. To do so, we're deploying AI more deeply across each layer of our systems and tools. + +Within our ads systems, we're delivering performance gains as we deploy more complex and predictive models. + +In Q1, enhancements we made to Lattice's modeling and learning techniques, along with advances in our GEM model architecture, drove a more than $6\%$ increase in conversion rate for landing page view ads. + +In addition, we’ve been investing in more performant inference models for when we’re serving ads. In the second half of last year, we began rolling out our new Adaptive Ranking Model, which is an LLM-scale ads recommender model that we use for inference. This model improves our inference ROI by routing requests to more compute intensive inference models when it determines there is a higher probability of conversion. In Q1, we expanded coverage of our Adaptive Ranking model to support offsite conversions, which drove a 1.6% increase in conversion rates across the major surfaces on Facebook and Instagram. + +We're also leveraging AI to make it easier for businesses to manage their campaigns, develop ad creative, and engage with customers. + +The Meta AI business assistant has now been fully rolled out to all eligible advertisers on supported Meta buying surfaces, providing personalized recommendations to advertisers, resolving account issues, and surfacing campaign insights to help optimize results. Performance has been strong since we began testing the assistant in Q4, with common account issues being resolved at a 20% higher rate. + +This week, we're also introducing Meta ads AI connectors in open beta, providing advertisers the ability to connect their Meta ad account directly to an AI agent. We've always supported advertisers both on our platform and through tools like the Marketing API — now we're extending that to AI so businesses and agencies can analyze and optimize campaigns with the tools they're already using. + +Usage of our ad creative tools is also scaling, with more than 8 million advertisers using at least one of our gen AI ad creative tools and particularly strong adoption among small and medium-sized advertisers. These tools are benefiting performance as well, with advertisers using our video generation feature seeing more than 3% higher conversion rates in tests. + +We're also seeing good traction in using AI to facilitate customer engagement. In Q1, we expanded Business AIs on WhatsApp to SMBs across Latin America and Indonesia, as well as on Messenger in Asia-Pacific. We now have more than 10 million conversations each week being facilitated through Business AIs, up from 1 million at the start of the year. We'll further expand access to more countries this quarter while adding more capabilities to the AIs. + +We also continue to invest in the Value Optimization suite, which helps advertisers maximize their Return on Ad Spend by prioritizing the highest-value conversions rather than optimizing solely for the most conversions at the lowest cost. Adoption by businesses has been strong following performance improvements we’ve made over the past year, with the annual revenue run-rate of our value optimization suite now over \$20B, more than doubling year-over-year. + +Last, I want to touch on our commerce efforts. + +People discover products on our platforms through ads and organic posts, with brands increasingly turning to creators to promote their products. This is contributing to rapid growth in our partnership ads product, with its revenue run-rate more than doubling year-over-year in Q1 to \$10 billion. + +To support the product discovery and purchasing happening through creators, we're expanding our solutions beyond ads. Last month, we rolled out our affiliate partnerships offering on Facebook to more test partners, so creators can tag products from participating retailers on their posts and earn a commission when someone makes a purchase. We have also started testing similar experiences on Instagram. + +We see a real opportunity to help people more easily discover and buy products within our services, particularly as we incorporate AI deeply across our platforms. + +Next, I would like to discuss our approach to capital allocation. + +Compute is becoming increasingly important as it determines the quality of the services we can provide, including powering more capable models and delivering innovative new products. It is also becoming more critical to how we work at Meta, as we are entering a world where employees are managing agents to help them generate new ideas, run experiments, execute tasks, and build products. + +We are investing aggressively to meet our infrastructure needs and ensure we maximize our strategic flexibility over the coming years. This includes substantially expanding our own data center footprint and striking deals throughout the supply chain to secure necessary components for future capacity. We are also signing cloud deals that will come online over the course of this year and 2027, allowing us to scale more quickly. These multi-year cloud deals and our + +infrastructure purchase agreements drove a \$107 billion step up in our contractual commitments this quarter. + +Our investments will support our training needs for future models, and most importantly, provide us the inference capacity necessary to deliver personal and business agents to billions of people around the world, along with several other AI product experiences we're developing. + +As we grow our infrastructure spend, we remain committed to operating efficiently, and we recently shared internally that we plan to reduce the size of our employee base in May. We believe a leaner operating model will allow us to move more quickly while also helping to offset the substantial investments we're making. + +Moving to our financial outlook. + +We expect second quarter 2026 total revenue to be in the range of \$58-61 billion. Our guidance assumes foreign currency is an approximately 2% tailwind to year-over-year total revenue growth, based on current exchange rates. + +Turning to the expense and capex outlooks. + +We expect full year 2026 total expenses to be in the range of \$162-169 billion, unchanged from our prior outlook. + +We continue to expect to deliver operating income this year that is above 2025 operating income. + +We anticipate 2026 capital expenditures, including principal payments on finance leases, to be in the range of \$125-145 billion, increased from our prior range of \$115-135 billion. This reflects our expectations for higher component pricing this year and, to a lesser extent, additional data center costs to support future year capacity. + +Absent any changes to our tax landscape, we expect our tax rate for the remaining quarters of 2026 to be between 13-16%. + +Lastly, we continue to monitor active legal and regulatory matters, including headwinds in the EU and the US that could significantly impact our business and financial results. For example, we continue to see scrutiny on youth-related issues and have additional trials scheduled for this year in the US, which may ultimately result in a material loss. + +In closing, Q1 was a solid start to the year, with strong execution across our core ads and engagement initiatives. We’re also making exciting progress on our AI research and product efforts and expect to build on that momentum over the course of this year. + +With that, Krista, let's open up the call for questions. + +Operator: + +Thank you. We will now open the lines for a question and answer session. To ask a question, please press star one on your touch tone phone. To withdraw your question again, press star one. + +Please limit yourself to one question. Please pick up your handset before asking your question to ensure clarity. If you are streaming today's call, please mute your computer speakers. And your first question comes from Brian Nowak with Morgan Stanley. Please go ahead. + +## Brian Nowak: + +Thanks for taking my question. Mark, I wanted to ask you just about the level of investment you're making and sort of the signposts you're watching to ensure you're going to generate ROIC on all these investments behind Muse and the other products. + +So if you could just sort of let us know some of the key factors you're watching over the next 12 to 24 months, whether it's Meta AI, Muse advances, core algorithm, what are you sort of watching foremost just to make sure that you're on the right path to generating healthy ROIC on all this CapEx and infrastructure spend? + +## Mark Zuckerberg: + +That's a very technical question for basically where -- the things that we're watching are to make sure that we're on track building leading models and leading products. The formula for our company has always been build experiences that can get to billions of people and focus on monetizing them once you get to scale. + +That's -- we're seeing a little bit of that here where basically we invest in advance to build leading models, and we convert that into leading products. And then we think that these are going to be some of the most important products that get built over the next decade. + +So I think just like anything else that we’ve done over time the basic milestones that I look at are around, first, technically, are we delivering the quality to enable a great product; then second, when you have the product, how is it scaling; and then third, you look at the monetization and then you drive up the efficiency of it towards increasing profitability. + +I don't -- I mean like I don't think we have a very precise plan for exactly how each product is going to scale month-over-month or anything like that. + +But I think we have a sense of the shape of where these things need to be. And I think if you look at the usage of these and the quality of the products and the quality of the models that are out there and the use that other frontier models are getting and the trajectory of that, I'm quite comfortable that, A, the lab that we're building is on track to be a leading lab in the world. + +I think Muse Spark was a very high-quality model. It powers Meta AI which I think is now a world-class assistant. + +We have an ability to be able to grow that and have a large amount of engagement. And over the coming quarters, we're just going to be tracking how do our next set of training runs go. How do our products scale? How excited are we about the products in the pipeline? Where right now we're very excited. + +And then we'll also ramp up monetization over that period of time as well. So I think that those are the set of things that I look at. I think for the kind of specific financial questions, I think Susan can jump in if there's anything more to add. + +Operator: Your next question comes from the line of Mark Shmulik with Bernstein. Please go ahead. + +Mark Shmulik: Yes, thanks for taking the questions. Mark, I guess now that we’ve got Muse Spark kind of out there launched, how are you thinking about the team’s focus here kind of divided on to further model training runs and kind of further specialization in that personal intelligence goal versus product launches and kind of shipping more product out the door? And Susan, I guess, kind of as a follow-up to Brian’s question, I know it’s too early to discuss 2027 CapEx. + +But we’ve had peers mention tonight a potential significant step-up. Any way to think about dimensionalizing kind of how we think about some of the returns or traction this year and how it might affect the 2027 spend? Thanks. + +Mark Zuckerberg: I mean I think the roadmap from the team is -- has been pretty consistent. So we have the research team which is focused on scaling increasingly intelligent models with capabilities for the specific things that we're focused on which are business and personal agents. + +So we're -- we just released our first model, and I talked about in my comments how we're climbing the scaling ladder towards greater capabilities and scale for the models. That work continues. + +We have our next set of more advanced models in training now. And that is -- that work will, I think, just continue. I mean that's a loop. I don't think we're going to be done with that anytime soon. + +We're going to have teams that are just consistently focused on training more intelligent and more capable models in the ways that we want. Then we have our product team, and that team is now really unlocked to be able to build things on top of our models because we now have a very strong model. + +So before this, we had been prototyping a bunch of things using other different models, whether it was our previous older models or kind of using the APIs from other companies. And now we're unlocked to be able to go build things and get them to scale on top of our own models. So I think you will see that over some period of time. + +I tried in my opening remarks to give a bit of a sense of where we're going, but I think that more of the details of that will become clear over the coming months. And I think that these are just both loops that we'll iterate on. We'll keep on iterating on the intelligence. + +We'll keep on working on building new products and scaling the products. And then as we get to product market fit, we're also going to increasingly focus on more agentic capabilities, enabling agents to help people be more productive, but also agents for businesses and enabling, frankly, those agents to interact with each other and build what we hope will be a thriving commerce ecosystem on our platform. + +
building the businesses around them and decreasing the costs. And this is kind of how we’ve done everything over the last 20 years of running the company, and that is basically the plan.
Susan Li:Mark, on your second question, we aren’t providing a specific outlook for 2027 CapEx. And we are, frankly, undergoing a very dynamic planning process ourselves as we’re working through what our capacity needs will be over the coming years.Our experience so far has been that we have continued to underestimate our compute needs even as we have been ramping capacity significantly as the advances in AI have continued and our teams continue to identify compelling new projects and initiatives. And now too, there are very compelling internal use cases.So our expectation is that compute will become even more central to the business going forward. And it will be critical to determining the quality of the models we develop, the types of products we can introduce, how productive we can be as an organization.So we’re going to continue building out our infrastructure with flexibility in mind. And if we end up not needing as much as we anticipate, we can choose to bring it online more slowly or reduce our spending in future years as we grow into the capacity that we’re building now.
Operator:Your next question comes from the line of Eric Sheridan with Goldman Sachs. Please go ahead.
Eric Sheridan:Thanks so much for taking the question. Maybe if I can build out on one of the topics that was discussed in the prepared remarks. But just the opportunity set that sits in front of the company with respect to putting agentic compute in front of both consumers and enterprises.You’ve long been associated with sort of the consumer landscape. And I am curious about how you’re thinking about extensions of the media engagement parts of your business model and the commerce parts of the business model to become more agentic over time.But what do you see also as the opportunity set that sits in front of you across SMBs and enterprises where historically, you maybe haven’t had as much product velocity? Thanks so much.
Susan Li:Thanks, Eric. So I would say in the near term, obviously the sort of biggest focuses are some of the areas that you mentioned around deepening sort of engagement, obviously with our existing community and user base, making ad experiences meaningfully more personalized, more engaging, more valuable, helping SMBs find and engage with customers across our platform.Those are some of the, I think, most intuitive and adjacent opportunities to the business that we have today. And then, of course, as we are able to build out
+ +So I would say some of these are a little bit further out, especially in that latter category of things. Again, the focus is on building personal superintelligence, building a consumer agent that can work for you and help you get things done. That right now is a consumer experience that we're focused on, but we think there will be clear monetization opportunities over time. You can imagine commission structures or a premium offering. And on the business side, we're seeing a large opportunity, of course, around agents and scaling our business AI initiatives. + +I think I mentioned earlier in my remarks that there are over 10 million weekly conversations between people and business AIs on our messaging platforms. That’s up from 1 million at the start of the year, and we’re going to continue expanding globally in Q2. And business AIs today are currently free for most businesses on our messaging apps. + +But as we make more progress, we expect that we will also work towards establishing a longer-term monetization model. And we'll also consider other services that we can offer to businesses in the future, but we don't have anything more to share today. + +Operator: Your next question comes from the line of Youssef Squali with Truist Securities. Please go ahead. + +Youssef Squali: Great, thank you very much for taking the questions. Maybe one for Mark and one for Susan. Mark, Ray-Ban, Oakley AI glasses continue to perform really well for you guys, but EssilorLuxottica owns and manages a lot more brands. + +What are the gating factors to see the launch of additional glasses under these other brands this year? And what would be a successful year for you as you look back at 2026, maybe in terms of units sold? + +And then Susan, on that 10% RIF, how much of that is due to efficiencies for maybe AI implementation versus just the need to stay fit? And as you look at your employee needs over time how do you see that growing maybe relative to your overall top line growth? Thank you very much. + +Susan Li: I can go ahead and take both of those. I might answer your second question first. And I'm just trying to make sure I got all of the parts of the question. + +So in terms of what the sort of kind of the optimal size of the company, I think, over time we don't really know what the optimal size of the company will be in the future. I think there's a lot of change right now with AI capabilities advancing rapidly. + +We're very focused on leveraging AI tools to substantially increase our productivity, and we're seeing that reflected in the accelerating output from our engineers. And we're generally approaching -- we're approaching this with a bias toward wanting to use these tools to build even more products and services than we would have before. + +At the same time we're making very significant investments in infrastructure, and we are very focused on continuing to operate efficiently. + +So I think we will be continuously evaluating how we're structured just to make sure we're best set up to deliver against our priorities over the coming years. So that is, I think, your second question. The first question was about the AI glasses. + +We have -- we're continuing to see strong growth in, obviously the AI glasses sales over the course of Q1. Demand for the expanded portfolio lineup has generally been quite strong, and we're seeing sales shift now from the prior generation of Ray-Ban Metas to the latest generation which I think speaks to the value of the improved features like extended battery life and features like higher resolution video capture. + +So we're pretty excited about the progress we've made with glasses. We see strong interest now in the Meta Ray-Ban Displays with the Meta Neural Band. + +So that's an encouraging sign that there is consumer appetite for display glasses which is kind of the next generation of how this product evolves. And yes, so I think this is an area that we will continue to be -- that we continue to be excited about and are investing in. + +Operator: Your next question comes from the line of Justin Post with Bank of America. Please go ahead. + +Justin Post: Great, thanks for taking my question. Mark, it took about 10 months to get Muse Spark out. I think it’s a pretty good pace. Just help us understand what kind of unlock that is for some of the new products you’re developing? And how is the product cadence going to be over the next nine months on either consumer or business enterprise products built on top of that model? + +Mark Zuckerberg: I mean the field is moving pretty quickly. So I mean I'm very happy that we're -- I think the lab that has gone the fastest from standing up the lab to having a very kind of widely accepted as strong model. So I think that's good. + +I take that as a very significant validation of the effort that the team is working well together, that the infrastructure is working, that the effort is on track. And I think that that's basically the main thing that we've learned over the last quarter that I would take away is like where and we started what is this pretty big bet, and it's on track for our plan. + +In terms of what exactly the cadence is going to be. It's tough for me to say both because I don't really want to share competitively sensitive information and because I think some of the stuff we are more focused on quality than hitting a specific date. + +I mean on the research side, this is research, right? We are trying novel things. You don't exactly know when they're going to land. And on the product side, I think we care a lot about just having -- let me put it this way. + +There's a lot of agents out there, right, that people are building for different things. And there aren't that many that I would want to give to my mother. And I think getting to like that quality bar is something that I care about more than hitting a specific week for launching or something like that. + +So -- but with that said, I mean we're in a zone here where the teams don't check in with me like once a quarter, like we make meaningful progress day over day. I think that's part of the fun of developing in this world is that people can make very rapid progress. Small groups of people and teams can make very rapid progress. + +So I think we're going to see a lot of innovation. The timing of this call is it's good in some ways because the Muse Spark release, I think, was positive. The Meta AI first release, I think, is positive. + +I think that that shows that we're on track. I'm trying to kind of paint a picture of the very high-level direction that we're going in, but I think that the picture is going to come into focus a lot more over the subsequent quarters. + +Operator: Your next question comes from the line of Ross Sandler with Barclays. Please go ahead. + +Ross Sandler: Yes. Mark, just sort of related to that last answer, but there's a lot of new consumer applications kind of cropping up everything from like an OpenClaw to something a little bit more consumer friendly that you would build for your mom, like you said, with like Poke or Dreamer which you recently acquired. + +So how are these new ideas, I guess, changing your view around the direction that core Meta AI or Dreamer or kind of your overall agentic strategy needs to go? And then the second part of it would be, do you think the lab will stay in this consumer lane? Or do you think you need -- or you want to go down the route that others are going down with code writing and like the recursive self-improvement loop and in that direction kind of in parallel, just thoughts on that? Thank you. + +Mark Zuckerberg: Yes. So look, on the OpenClaw and other agents, I think that they give you a very exciting glimpse of what types of things should be possible. + +Now they're pretty rough systems today. And to set up OpenClaw you need to like install a computer locally and then get into a terminal and configure a bunch of things that, again, like there's -- maybe there's hundreds of thousands of people or small numbers of millions of people who can do that. But what we're talking about is delivering personal superintelligence for billions of people around the world. + +So how do you make a version of that experience that is a lot more polished and dialed and easy and that has all the infrastructure basically done for people already and that just works. And that's kind of what we're focused on, on the consumer side. And I'm really excited about that. + +I think if you had something like that, that worked quite a bit better than those systems and was easy enough that people could just get then I think you go from having something that hundreds of thousands or millions of people are going to use to something that is going to be addressable to billions of people. And that has been our primary focus from day one of the lab is being able to deliver something like that as a product, and I think it’s just going to be very exciting. + +By the way, the same thing is true for businesses, right? I mean there's the personal version of this but there's also a lot of people's goals are they want to create things, right? They want to create websites. They want to create products. They want to grow their products. + +These are all things that good agents are going to be able to help people do which I think is partially why this is so exciting. And in my opening comments, I talked about how today we can handle a few goals for people, they're big goals, right? We can help people stay connected with people they care about, learn about the world. These are big things that people care about. But they're not the only things that people care about. + +And one of the things that I would love for our products to be able to do is just understand people's goals specifically and then be able to just go work on them for them, and check back in and whenever you have questions that you need answered. So whether those are personal goals or you're trying to create a business or do work. + +I think that this is like -- this is stuff that I think literally every person in the world is going to want some version of it. And also, I think it is something that scales where the more you want to get out of it, I think people are going to also be willing to pay a lot of money to have premium or high compute versions of it. So I think that this is like, it's a very exciting area. + +But I think what you all should be waiting to see is like whether we can build the version that really like just works and how effective we are at converting people who are using our products into being hundreds of millions and then billions of people using this stuff. And then over time how can we effectively convert that into something that's increasingly profitable by monetizing it and getting the costs down. + +So I think that that's the roadmap of what we need to do. You asked about whether we're primarily focused on consumers or also recursive self-improvement. I think that we've talked about two main goals for the team. + +I mean one is this kind of agents version vision of what we're doing. The other is that self-improvement is really important because you can't build a leading AI product if you don't have leading models. So -- and you're not going to have leading models in the future if your models can't improve themselves, right? + +So you're getting to a point where today, the models are still able to learn from people -- and then I think at some point, the models will have to improve themselves. And that's how the growth is going to -- and improvement in the models is going to happen. And if you don't -- if we don't have an ability to do that, then we or anyone else, I think the companies that don't do that are not going to be leading labs, they're not going to produce leading products. + +So I think that is a table stakes thing that we are focused on. Now does that make us a developer tools company? Not necessarily. + +I mean I'm not against having an API or coding tools or anything like that. But it's not our primary focus. But I actually think people conflate coding with self-improvement more than they should. Coding is one ingredient for the model self improving. + +It's not the only thing. And we are focused on all of the parts that are going to be necessary for self-improvement in service of the personal superintelligence vision that we have for people and businesses. + +Operator: Your next question comes from the line of Ron Josey with Citigroup. Please go ahead. + +Ronald Josey: Great, thanks for taking my question. Mark, maybe a quick follow-up to a prior question around personal agents and business agents. And with Spark Muse now live and more models in development, do you look at the personal agent opportunity which we talked about earlier on the call more of a short-term, medium-term, long-term goal, I'm sure it's a never-ending goal, but when we see a product, is the question short or medium term? + +And then Susan, I think the ranking recommendation model improvements are very impressive to see, given the size and scale of both Instagram and Facebook. Could you help us understand just how doubling the length of these interaction sequences can drive greater usage? There’s a thesis out there that maybe some of the rating recommendation improvements are long in the tooth. So it seems if there’s a lot more room to go. So any help there would be helpful. Thank you. + +Mark Zuckerberg: I mean I think that the agents work, there's going to be short-term versions of it, but then I think that there's going to be massive upside for delivering more intelligence and more capabilities in the models. And you're kind of seeing this across the industry. + +Each month, each generation of models, they just have more capabilities and can do more things and people absorb it and are able to get more superpowers and it’s awesome. It’s like the most exciting time in the industry. + +So I think of the agents as the product vehicle for delivering that capability to people. And we certainly -- I think this year is going to be a key period for establishing that as the vehicle for how people are going to use this, but then the model improvement, I think, is going to be something that's going to go on for a very long time. So there's a lot to do here in both the short, medium, and long term. + +## Susan Li: + +And then on your second question which I think is about the ranking and recommendations improvements that we talked about in our -- that I talked about in my earlier remarks, I think first of all, there is still a lot of room to continue improving recommendations over the rest of the year, and we expect we'll be able to do that to drive additional engagement on both Facebook and Instagram. + +A couple of the things. First, we’re going to continue to improve our data infrastructure that’s going to allow our models to train on more data. And we’re adding more detail to how we describe the content that users have engaged with in the past and scaling up the complexity of our model architecture to take advantage of those larger data sets like using even longer histories of content interactions, and that should all be in service of improving the overall quality of recommendations. + +We also are focused on making the recommendations even more personalized and more relevant to any given users interest. There’s work we’re doing to redesign our content retrieval system to show more content that matches the full range of a user’s interests and to tailor the diversity of the topics we recommend to the broadness of someone’s interests. + +So someone with particularly concentrated interests might see relatively more of that content while people with a broader set of interests might see kind of a greater range in the topics that we show them. And then finally, we're continuing to make improvements to our sort of LLM based tune-your-algorithm features that allow users to provide more granular natural language feedback on what they want to see more of or less of in their feed. + +So the sort of the kind of the sequence length which is the thing that you called out is one of really many improvements we made in Q1, and there is a big roadmap of further improvements going forward. + +## Operator: + +Your next question comes from the line of Doug Anmuth with JPMorgan. Please go ahead, + +## Douglas Anmuth: + +Thanks so much for taking the questions. Mark, how do you think about the step up as you go from leveraging smaller models in the ad business to Muse Spark and future large language models going forward, what are some of the key unlocks across engagement and monetization? And then on Manus, can you just talk at all about the strategic importance and the role in developing agentic products for Meta and then just current status around the tech and the deal. + +## Susan Li: + +I'll take that question. On Manus, we're still working through the details. So we don't have an update right now. + +On your first question which is about going from leveraging smaller models in the ads business to kind of the ads models growing. There’s already some work underway, and I think I alluded to some of this in my earlier remarks, even kind of in the current landscape of the ads roadmap, where we’re basically trying to advance the architecture here to allow sort of -- to allow us to leverage the abilities of larger models. + +Historically, we haven't used larger model architectures like GEM for inference because their size and complexity would make them too cost prohibitive. And the way we drive performance from those models is by using them to transfer knowledge to smaller, more lightweight models that are used at run time. + +The inference models are bound by strict latency requirements since they need to find the right ad within milliseconds, and that has, again, historically prevented us from meaningfully sizing up -- scaling up their size and complexity. + +But in the second half of last year, we introduced a new adaptive ranking model which enables us to leverage LLM scale model complexity of a trillion parameters, and we made advances in the model architecture and codesign the system with the underlying silicon, so it maintains the sub-second speed that is required to serve ads at scale. + +We also developed an approach that intelligently routes request more compute-intensive inference models if it determines that there is a higher probability of conversion and that lets us drive both better performance and increased inference ROI. + +So there's a lot of work being done there before we even sort of incorporate more of the LLM work into our underlying ads ranking models. + +## Operator: + +We have time for one more question, Ken Gawrelski with Wells Fargo. Your line is open. + +## Kenneth Gawrelski: + +i: Thank you very much. Two, if I may. First, if I -- you talked on the Muse Spark launch. You've talked about two categories or two verticals. You talked about health and wellness and shopping. Can I dive a little bit -- ask you to dive a little deeper into the latter on the shopping and commerce side? + +And maybe if you could -- were there any learnings in the 2021, '22 phase where you pushed deeper into commerce on Instagram and on Facebook? Any learnings from that period that you might apply? Is there an opportunity for a next-gen marketplace-type business in commerce? + +And then the second, please, maybe, Susan, can you talk a little bit about -- based on your model improvements and the content recommendations, where -- how much visibility do you think you have to kind of the growth trajectory on the core business? + +You continue to grow at basically double the pace of the industry despite being a very large share of the industry. Could you just talk about a little bit about your visibility into that continued performance? Thank you. + +Mark Zuckerberg: Yes. So I might give you a somewhat loftier answer to the question. You're asking about shopping. + +I think it's sort of an interesting example of the way in which the work that we're doing is different than what I think others are doing out there. These products, they -- AI agents get better when you fully optimize the stack. + +That's why we believe that we need to be a company that builds frontier models in addition to building the agents. And then in order to do that, you, of course, need to build your infrastructure in order to be able to do that well. + +So we're undertaking this large investment to be able to do that top to bottom. And I think a lot of the way to think about the investment that we're making is a bet that the individual things that people care about and that people are going to be more important in the future. And that's sort of like -- and I think it should be a pretty obvious thing to say. + +But I think so much of the rhetoric around AI in the industry is around like a company trying to build some kind of centralized thing that like does all the productive work in society in some way or something like that. And that just is very different from how we see the world. + +Like our vision for the future is one where society makes progress by individuals pursuing their own aspirations. And some people care about big grand things like curing diseases. And a lot of people care about personal things like finding the right shirt for my daughter. And I just think that we want to -- we're going to build things that help deliver this vision for personal agents for people. + +And I think that part of the lane and what is interesting and differentiated about what we're doing is that that's just so different from how I hear everyone else talking about the work that we're doing. So even though I think some of these ideas, they seem like they should be so obvious. + +I actually think that our approach of trying to empower individuals and building consumer things is just in the details extremely different from what others are doing. And shopping might be one kind of specific example that I think is going to have interest in commercial implications. And I think people -- consumers are going to like it. + +But I don’t hear any other labs out there talking about how they’re building an AI that’s really good at shopping. And I think that the reason for that is like not because shopping is the most important thing by itself, but because like empowering people to do the things that matter in their lives, whether that’s local or understanding social context, or shopping or personal health things or understanding what’s going on around them visually which is going to be really important on the glasses. These are all elements of the personal super intelligence vision. + +I think like a lot of this, and when you're thinking about kind of the investment in Meta over time I think you should think about it as coming down to these set of values around what do we want AI to do in society. And if what you want it to do is empower individuals and build a world where the AI is in service to individual's goals, then that is what we are going to build, and I think it's going to be incredibly valuable. + +Susan Li: Gosh, I almost wish we could end on that answer, but I will answer the second question which I think kind of has two versions. + +One is a version of like what's the revenue outlook? And obviously we gave the Q2 guide which embeds, I think, both a range of kind of macro outcomes, but also the work that we've -- the ongoing work that we have to continue improving both the sort of usage and engagement on our family of apps and then our ability to continue making the ads better and more performant. + +I think the second question is maybe more of a -- the second version of that question is more of a higher-level question about kind of the overall trajectory of the roadmap here. And one of the things I will say, having been working on this for a very long time I'm always really impressed by the team's ability to continue to advance the state of the art here. And our planning process now is, I think, really fine-tuned around this. + +So I’ve mentioned on a couple of calls, the budgeting process in which we run a very sort of ROI-based process to make sure that we are funding all of the ads initiatives that we think will drive growth in future years. And that’s something that is both quite dialed in. + +And I think that our ability to measure the impact of that has been pretty robust, and it's been a very important driver of our ads revenue growth, and that continues to be a process that again, we ran in this past budget and -- as far as we can -- as we have line of sight, we feel good about the investment opportunities ahead of us. + +Kenneth Dorell: Great. Thank you everyone, for joining us today. We look forward to speaking with you again soon. + +Operator: This concludes today's conference call. Thank you for joining. And you may now disconnect. \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/manifest.json b/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/manifest.json new file mode 100644 index 000000000..439647643 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/manifest.json @@ -0,0 +1,64 @@ +{ + "version": "2.0", + "job_id": "job_12a9a381bb04", + "data_id": null, + "source_file_name": "META-Q1-2026-Earnings-Call-Transcript.pdf", + "processing_date": "2026-06-16T03:35:02.680217Z", + "processing": { + "page_count": 20, + "billing_status": "charged", + "cost": { + "micro_dollars": 30000, + "credits": 0.03 + }, + "timing": { + "started_at": "2026-06-16T03:34:17.225197+00:00", + "completed_at": "2026-06-16T03:35:02.672609+00:00", + "duration_ms": 45447 + }, + "stages": { + "timing_ms": { + "document.profile": 4249, + "pdf.extract.standard": 34338, + "toc.detect_range_llm": 1055, + "md.detect_toc": 1090, + "heading.hierarchy_llm_call": 2513, + "heading.hierarchy_llm": 2522, + "md.predict_headings": 2839, + "md.deferred_summaries": 2622, + "md.build_dataframe": 1, + "pdf.parse_md": 6799, + "document.cleanup_unreferenced_images": 0, + "document.compress_images": 0, + "worker.parse.document": 45392 + }, + "token_usage": { + "prompt_tokens": 6675, + "completion_tokens": 170, + "total_tokens": 6845 + } + } + }, + "statistics": { + "total_chunks": 13, + "text_chunks": 12, + "image_chunks": 0, + "table_chunks": 1, + "total_pages": null + }, + "HIERARCHY": { + "Root": {}, + "Kenneth Dorell, Director, Investor Relations": {}, + "Mark Zuckerberg, CEO": {}, + "Susan Li, CFO": { + "Brian Nowak:": {}, + "Mark Zuckerberg:": {}, + "Susan Li:": {}, + "Operator:": {}, + "Douglas Anmuth:": {}, + "Susan Li:_2": {}, + "Operator:_2": {}, + "Kenneth Gawrelski:": {} + } + } +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/tables/table-0 AI Compute Plans.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/tables/table-0 AI Compute Plans.html new file mode 100644 index 000000000..86563c184 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-earnings-call/tables/table-0 AI Compute Plans.html @@ -0,0 +1 @@ +
building the businesses around them and decreasing the costs. And this is kind of how we’ve done everything over the last 20 years of running the company, and that is basically the plan.
Susan Li:Mark, on your second question, we aren’t providing a specific outlook for 2027 CapEx. And we are, frankly, undergoing a very dynamic planning process ourselves as we’re working through what our capacity needs will be over the coming years.Our experience so far has been that we have continued to underestimate our compute needs even as we have been ramping capacity significantly as the advances in AI have continued and our teams continue to identify compelling new projects and initiatives. And now too, there are very compelling internal use cases.So our expectation is that compute will become even more central to the business going forward. And it will be critical to determining the quality of the models we develop, the types of products we can introduce, how productive we can be as an organization.So we’re going to continue building out our infrastructure with flexibility in mind. And if we end up not needing as much as we anticipate, we can choose to bring it online more slowly or reduce our spending in future years as we grow into the capacity that we’re building now.
Operator:Your next question comes from the line of Eric Sheridan with Goldman Sachs. Please go ahead.
Eric Sheridan:Thanks so much for taking the question. Maybe if I can build out on one of the topics that was discussed in the prepared remarks. But just the opportunity set that sits in front of the company with respect to putting agentic compute in front of both consumers and enterprises.You’ve long been associated with sort of the consumer landscape. And I am curious about how you’re thinking about extensions of the media engagement parts of your business model and the commerce parts of the business model to become more agentic over time.But what do you see also as the opportunity set that sits in front of you across SMBs and enterprises where historically, you maybe haven’t had as much product velocity? Thanks so much.
Susan Li:Thanks, Eric. So I would say in the near term, obviously the sort of biggest focuses are some of the areas that you mentioned around deepening sort of engagement, obviously with our existing community and user base, making ad experiences meaningfully more personalized, more engaging, more valuable, helping SMBs find and engage with customers across our platform.Those are some of the, I think, most intuitive and adjacent opportunities to the business that we have today. And then, of course, as we are able to build out
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/chunks.json b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/chunks.json new file mode 100644 index 000000000..c68b83064 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/chunks.json @@ -0,0 +1,2115 @@ +{ + "chunks": [ + { + "chunk_id": "1e8af615-ab47-50f6-999a-08bdb1d69c7b", + "type": "text", + "content": "", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9d9790c1-65a7-5b12-be61-31b87992d270", + "type": "image", + "content": "\n# Meta Earnings Presentation Q1 2026

0\n[images/image-1-Revenue Forecast by Region.jpg]\n", + "path": "images/image-1-Revenue Forecast by Region.jpg", + "metadata": { + "length": 81, + "summary": "image-1\nQ1'24: $35,635 (US & Canada: $15,451; Europe: $8,327; Asia-Pacific: $7,338; Rest of World: $4,519)\nQ2'24: $38,329 (US & Canada: $16,593; Europe: $9,135; Asia-Pacific: $7,721; Rest of World: $4,880)\nQ3'24: $39,885 (US & Canada: $17,389; Europe: $9,358; Asia-Pacific: $8,050; Rest of World: $5,088)\nQ4'24: $46,783 (US & Canada: $20,982; Europe: $11,154; Asia-Pacific: $9,012; Rest of World: $5,635)\nQ1'25:", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-1-Revenue Forecast by Region.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "bcea04e7-bfc3-5817-b400-5f7991495f4b", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterUS & CanadaEuropeAsia-PacificRest of World
Q1'24$15,451$8,327$7,338$4,519
Q2'24$16,593$9,135$7,721$4,880
Q3'24$17,389$9,358$8,050$5,088
Q4'24$20,982$11,154$9,012$5,635
Q1'25$18,259$9,527$8,224$5,382
Q2'25$20,045$11,366$9,148$6,004
Q3'25$21,331$12,072$10,020$6,659
Q4'25$25,643$14,198$10,893$7,403
Q1'26$23,667$13,296$10,631$7,430
Total$35,635
Q1'24
Q2'24
Q3'24
Q4'24
Q1'25
Q2'25
Q3'25
Q4'25
Q1'26
Total
Total
Total
Total
Total
Total
Total— (additional values)(continued)(continued)(continued)
Total58,13714,19810,8937,403
Total55,02413,29610,6317,430
", + "path": "tables/table-0 Revenue by Region.html", + "metadata": { + "length": 3581, + "summary": "table-1\nThe table shows quarterly revenue data for US & Canada, Europe, Asia-Pacific, and Rest of World from Q1'24 to Q1'26, with totals.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-0 Revenue by Region.html", + "keywords": [ + "Revenue", + "Quarterly", + "Region" + ], + "tokens": [] + } + }, + { + "chunk_id": "72a039a1-a82b-5da5-a685-4891b6d0fd14", + "type": "text", + "content": "In Millions\n\n# Meta Earnings Presentation Q1 2026
0\n[images/image-1-Revenue Forecast by Region.jpg]\n\n
\nstacked bar chart\n\n[tables/table-0 Revenue by Region.html]\n\n
\nOur revenue by user geography is geographically apportioned based on our estimation of the geographic location of our users when they perform a revenue-generating activity. This allocation differs from our revenue disaggregated by geography disclosure in our condensed consolidated financial statements where revenue is geographically apportioned based on the addresses of our customers.", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Advertising Revenue by User Geography", + "metadata": { + "length": 617, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "In", + "Millions", + "Meta", + "Earnings", + "Presentation", + "Q1", + "2026", + "details", + "summary", + "stacked", + "bar", + "chart", + "Our", + "revenue", + "user", + "geography", + "geographically", + "apportioned", + "based", + "estimation", + "geographic", + "location", + "users", + "perform", + "generating", + "activity", + "This", + "allocation", + "differs", + "disaggregated", + "disclosure", + "condensed", + "consolidated", + "financial", + "statements", + "addresses", + "customers" + ], + "keywords": [], + "connect_to": [ + { + "target": "9d9790c1-65a7-5b12-be61-31b87992d270", + "relation": "embeds", + "ref": "[images/image-1-Revenue Forecast by Region.jpg]", + "position": { + "start": 61, + "end": 108 + } + }, + { + "target": "bcea04e7-bfc3-5817-b400-5f7991495f4b", + "relation": "embeds", + "ref": "[tables/table-0 Revenue by Region.html]", + "position": { + "start": 158, + "end": 197 + } + } + ] + } + }, + { + "chunk_id": "2d71f6fe-2d26-5c37-b52d-2ec472072997", + "type": "image", + "content": "\n
0\n[images/image-2-Revenue Breakdown by Region and Quarter.jpg]\n", + "path": "images/image-2-Revenue Breakdown by Region and Quarter.jpg", + "metadata": { + "length": 55, + "summary": "image-2\nChart Title: Not visible (Stacked Bar Chart)\nTime Range: Q1'24 to Q1'26\n\nLegend Categories:\n- US & Canada (Darkest Blue)\n- Europe (Medium Blue-Grey)\n- Asia-Pacific (Lighter Blue-Grey)\n- Rest of World (Lightest Grey)\n\nData Points (Total Revenue per Quarter):\n- Q1'24: $36,455\n- Q2'24: $39,071\n- Q3'24: $40,589\n- Q4'24: $48,385\n- Q1'25: $42,314\n- Q2'25: $47,516\n- Q3'25: $51,242\n- Q4'25: $59,893\n- Q1'26: $56,311\n\nDetailed Breakdown by Category (Bottom to Top):\n\nQ1'24:", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-2-Revenue Breakdown by Region and Quarter.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "df8a9368-8ea9-5791-b0ca-a11cb1d2b614", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterUS & CanadaEuropeAsia-PacificRest of World
Q1'24$15,824$8,483$7,481$4,667
Q2'24$16,847$9,300$7,888$5,036
Q3'24$17,609$9,492$8,220$5,268
Q4'24$21,783$11,503$9,245$5,854
Q1'25$18,605$9,680$8,439$5,590
Q2'25$20,371$11,532$9,366$6,247
Q3'25$21,751$12,268$10,272$6,951
Q4'25$26,472$14,480$11,182$7,759
Q1'26$24,079$13,500$10,914$7,818
", + "path": "tables/table-1 Quarterly Revenue.html", + "metadata": { + "length": 1435, + "summary": "table-2\nThe table shows quarterly revenue from Q1'24 to Q1'26 across US & Canada, Europe, Asia-Pacific, and Rest of World, with totals increasing over time.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-1 Quarterly Revenue.html", + "keywords": [ + "Revenue", + "Regions", + "Quarters" + ], + "tokens": [] + } + }, + { + "chunk_id": "72a039a1-a82b-5da5-a685-4891b6d0fd14", + "type": "text", + "content": "In Millions\n\n
0\n[images/image-2-Revenue Breakdown by Region and Quarter.jpg]\n\n
\nstacked bar chart\n\n[tables/table-1 Quarterly Revenue.html]\n\n
\nOur revenue by user geography is geographically apportioned based on our estimation of the geographic location of our users when they perform a revenue-generating activity. This allocation differs from our revenue disaggregated by geography disclosure in our condensed consolidated financial statements where revenue is geographically apportioned based on the addresses of our customers.", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Revenue by User Geography", + "metadata": { + "length": 591, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "In", + "Millions", + "details", + "summary", + "stacked", + "bar", + "chart", + "Our", + "revenue", + "user", + "geography", + "geographically", + "apportioned", + "based", + "estimation", + "geographic", + "location", + "users", + "perform", + "generating", + "activity", + "This", + "allocation", + "differs", + "disaggregated", + "disclosure", + "condensed", + "consolidated", + "financial", + "statements", + "addresses", + "customers" + ], + "keywords": [], + "connect_to": [ + { + "target": "2d71f6fe-2d26-5c37-b52d-2ec472072997", + "relation": "embeds", + "ref": "[images/image-2-Revenue Breakdown by Region and Quarter.jpg]", + "position": { + "start": 35, + "end": 95 + } + }, + { + "target": "df8a9368-8ea9-5791-b0ca-a11cb1d2b614", + "relation": "embeds", + "ref": "[tables/table-1 Quarterly Revenue.html]", + "position": { + "start": 145, + "end": 184 + } + } + ] + } + }, + { + "chunk_id": "3601bb8f-ed16-5aa0-ace8-c84bbc142527", + "type": "table", + "content": "
Q1'24Q2'24Q3'24Q4'24Q1'25Q2'25Q3'25Q4'25Q1'26
Advertising$ 35,635$ 38,329$ 39,885$ 46,783$ 41,392$ 46,563$ 50,082$ 58,137$ 55,024
Other380389434519510583690801885
Family of Apps Revenue36,01538,71840,31947,30241,90247,14650,77258,93855,909
Reality Labs Revenue4403532701,083412370470955402
Total Revenue$ 36,455$ 39,071$ 40,589$ 48,385$ 42,314$ 47,516$ 51,242$ 59,893$ 56,311
Family of Apps Operating Income$ 17,664$ 19,335$ 21,778$ 28,332$ 21,765$ 24,971$ 24,967$ 30,766$ 26,900
Reality Labs Operating (Loss)(3,846)(4,488)(4,428)(4,967)(4,210)(4,530)(4,432)(6,021)(4,028)
Total Income from Operations$ 13,818$ 14,847$ 17,350$ 23,365$ 17,555$ 20,441$ 20,535$ 24,745$ 22,872
Operating Margin38 %38 %43 %48 %41 %43 %40 %41 %41 %
", + "path": "tables/table-2 Meta Revenue Data.html", + "metadata": { + "length": 1798, + "summary": "table-3\nThe table shows Meta's quarterly revenue from Q1'24 to Q1'26, including Advertising, Other, Family of Apps, Reality Labs, and total revenue, along with operating income and margins.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-2 Meta Revenue Data.html", + "keywords": [ + "Revenue", + "Operating Income", + "Quarterly" + ], + "tokens": [] + } + }, + { + "chunk_id": "ca5ccbff-1066-5fde-aca3-b6ba138b6c99", + "type": "text", + "content": "In Millions, Except Percentages\n\n[tables/table-2 Meta Revenue Data.html]\n\nWe report our financial results based on two reportable segments: Family of Apps (FoA) and Reality Labs (RL). FoA includes Facebook, Instagram, Messenger, WhatsApp, and other services. RL includes our virtual and augmented reality related consumer hardware, software, and content.", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results", + "metadata": { + "length": 390, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "In", + "Millions", + "Except", + "Percentages", + "We", + "report", + "financial", + "results", + "based", + "reportable", + "segments", + "Family", + "Apps", + "FoA", + "Reality", + "Labs", + "RL", + "includes", + "Facebook", + "Instagram", + "Messenger", + "WhatsApp", + "services", + "virtual", + "augmented", + "reality", + "related", + "consumer", + "hardware", + "software", + "content" + ], + "keywords": [], + "connect_to": [ + { + "target": "3601bb8f-ed16-5aa0-ace8-c84bbc142527", + "relation": "embeds", + "ref": "[tables/table-2 Meta Revenue Data.html]", + "position": { + "start": 33, + "end": 72 + } + } + ] + } + }, + { + "chunk_id": "dca8cc6c-9b0d-5666-968a-73413e158858", + "type": "image", + "content": "\n## Segment Results
0\n[images/image-3-Quarterly Percentage Breakdown.jpg]\n", + "path": "images/image-3-Quarterly Percentage Breakdown.jpg", + "metadata": { + "length": 63, + "summary": "image-3\nQ1'24: 18%, 27%, 7%, 9%\nQ2'24: 19%, 27%, 7%, 9%\nQ3'24: 18%, 28%, 7%, 5%\nQ4'24: 18%, 25%, 7%, 2%\nQ1'25: 18%, 29%, 7%, 5%\nQ2'25: 18%, 27%, 6%, 6%\nQ3'25: 18%, 30%, 6%, 7%\nQ4'25: 18%, 29%, 6%, 6%\nQ1'26: 18%, 31%, 5%, 5%", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-3-Quarterly Percentage Breakdown.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d3ae4b0d-b92a-5856-8b16-0d428c30db36", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterBottom Segment (%)Top Segment (%)
Q1'241827
Q2'241927
Q3'241828
Q4'241825
Q1'251829
Q2'251827
Q3'251830
Q4'251829
Q1'261831
", + "path": "tables/table-3 Segment Margins.html", + "metadata": { + "length": 905, + "summary": "table-4\nThe table shows quarterly margins for bottom and top segments from Q1'24 to Q1'26. Bottom segment stays at 18-19%, while top segment fluctuates between 25-31%.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-3 Segment Margins.html", + "keywords": [ + "Bottom Segment", + "Top Segment", + "Quarterly" + ], + "tokens": [] + } + }, + { + "chunk_id": "d3424da5-a6d6-5f35-ab5c-8119d0b3d511", + "type": "text", + "content": "## Segment Results
0\n[images/image-3-Quarterly Percentage Breakdown.jpg]\n\n
\nstacked bar chart\n\n[tables/table-3 Segment Margins.html]\n\n
\nGeneral & Administrative\nMarketing & Sales\nResearch & Development\nCost of Revenue", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Expenses as a Percentage of Revenue", + "metadata": { + "length": 269, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "Segment", + "Results", + "details", + "summary", + "stacked", + "bar", + "chart", + "General", + "Administrative", + "Marketing", + "Sales", + "Research", + "Development", + "Cost", + "Revenue" + ], + "keywords": [], + "connect_to": [ + { + "target": "dca8cc6c-9b0d-5666-968a-73413e158858", + "relation": "embeds", + "ref": "[images/image-3-Quarterly Percentage Breakdown.jpg]", + "position": { + "start": 30, + "end": 81 + } + }, + { + "target": "d3ae4b0d-b92a-5856-8b16-0d428c30db36", + "relation": "embeds", + "ref": "[tables/table-3 Segment Margins.html]", + "position": { + "start": 131, + "end": 168 + } + } + ] + } + }, + { + "chunk_id": "f5aa0d1a-8149-53b3-b7fa-a7454585d964", + "type": "table", + "content": "
Q1'24Q2'24Q3'24Q4'24Q1'25Q2'25 $Q3'25^{(1)}$ Q4'25 $Q1'26^{(2)}$
Income before income taxes$ 14,183$ 15,106$ 17,822$ 23,553$ 18,382$ 20,534$ 21,663$ 25,354$ 21,752
Provision (benefit) for income taxes$ 1,814$ 1,641$ 2,134$ 2,715$ 1,738$ 2,197$ 18,954$ 2,586$ (5,021)
Effective Tax Rate13 %11 %12 %12 %9 %11 %87 %10 %(23)%
", + "path": "tables/table-4 Quarterly Tax Data.html", + "metadata": { + "length": 765, + "summary": "table-5\nThe table shows quarterly income before taxes, provision for income taxes, and effective tax rates from Q1'24 to Q1'26, with notable spikes in Q3'25 and negative rate in Q1'26.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-4 Quarterly Tax Data.html", + "keywords": [ + "Income", + "Tax", + "Effective Rate" + ], + "tokens": [] + } + }, + { + "chunk_id": "2b6e6995-508b-5a12-8b97-b1d3560d8aa9", + "type": "image", + "content": "\n### Effective Tax Rate
0\n[images/image-4-Quarterly Financial Performance.jpg]\n", + "path": "images/image-4-Quarterly Financial Performance.jpg", + "metadata": { + "length": 67, + "summary": "image-4\n- Q1'24: $12,369\n- Q2'24: $13,465\n- Q3'24: $15,688\n- Q4'24: $20,838\n- Q1'25: $16,644\n- Q2'25: $18,337\n- Q3'25 (1): $2,709\n- Q4'25: $22,768\n- Q1'26 (2): $26,773", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-4-Quarterly Financial Performance.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "21d81faf-1ff7-5aba-b902-161d97ffd732", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue
Q1'24$12,369
Q2'24$13,465
Q3'24$15,688
Q4'24$20,838
Q1'25$16,644
Q2'25$18,337
Q3'25 (1)$2,709
Q4'25$22,768
Q1'26 (2)$26,773
", + "path": "tables/table-5 Quarterly Values.html", + "metadata": { + "length": 751, + "summary": "table-6\nThe table shows quarterly values from Q1'24 to Q1'26, with figures ranging from $2,709 to $26,773, peaking in Q1'26.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-5 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "ef8c8327-d540-5df5-804e-d5323c71e730", + "type": "text", + "content": "In Millions, Except Percentages\n\n[tables/table-4 Quarterly Tax Data.html]\n\n(1) Provision (benefit) for income taxes includes a one-time, non-cash income tax charge of \\$15.93 billion accrued in the third quarter of 2025, related to the implementation of the One Big Beautiful Bill Act. Excluding this one-time tax charge, our third quarter 2025 effective tax rate would have decreased by 73 percentage points to 14%, compared to the reported effective tax rate of 87%.\n(2) Provision (benefit) for income taxes includes an income tax benefit of \\$8.03 billion in the first quarter of 2026 related to the U.S. Corporate Alternative Minimum Tax transitional relief under Treasury Notice 2026-7. Excluding this tax benefit, the effective tax rate would have been 14%.\n\n### Effective Tax Rate
0\n[images/image-4-Quarterly Financial Performance.jpg]\n\n
\nbar chart\n\n[tables/table-5 Quarterly Values.html]\n\n
\n(1) Includes a one-time, non-cash income tax charge to provision for income taxes of \\$15.93 billion accrued in the third quarter of 2025, related to the implementation of the One Big Beautiful Bill Act. Excluding this one-time tax charge, our third quarter 2025 net income would have increased by \\$15.93 billion to \\$18.64 billion, compared to the reported net income of \\$2.71 billion.\n(2) Includes an income tax benefit of \\$8.03 billion in the first quarter of 2026 related to the U.S. Corporate Alternative Minimum Tax transitional relief under Treasury Notice 2026-7. Excluding this tax benefit, net income would have been \\$18.74 billion, lower by \\$8.03 billion, compared to the reported net income of \\$26.77 billion.", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Effective Tax Rate", + "metadata": { + "length": 1698, + "summary": "The passage discusses the impact of one-time tax items on reported net income and effective tax rates. In Q3 2025, a $15.93 billion non-cash income tax charge related to the One Big Beautiful Bill Act increased the effective tax rate to 87%; excluding it, the rate would have been 14% and net income would have been $18.64 billion. In Q1 2026, an $8.03 billion tax benefit from U.S. Corporate Alternative Minimum Tax transitional relief reduced the effective tax rate; excluding it, the rate would have been 14% and net income would have been $18.74 billion. A bar chart showing effective tax rates by quarter is referenced.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "In", + "Millions", + "Except", + "Percentages", + "Provision", + "benefit", + "income", + "taxes", + "includes", + "time", + "cash", + "tax", + "charge", + "15.93", + "billion", + "accrued", + "quarter", + "2025", + "related", + "implementation", + "One", + "Big", + "Beautiful", + "Bill", + "Act", + "Excluding", + "effective", + "rate", + "decreased", + "73", + "percentage", + "points", + "14%", + "compared", + "reported", + "87%", + "8.03", + "2026", + "Corporate", + "Alternative", + "Minimum", + "Tax", + "transitional", + "relief", + "Treasury", + "Notice", + "Effective", + "Rate", + "details", + "summary", + "bar", + "chart", + "Includes", + "provision", + "net", + "increased", + "18.64", + "2.71", + "18.74", + "lower", + "26.77" + ], + "keywords": [ + "tax charge", + "tax benefit", + "effective tax rate" + ], + "connect_to": [ + { + "target": "f5aa0d1a-8149-53b3-b7fa-a7454585d964", + "relation": "embeds", + "ref": "[tables/table-4 Quarterly Tax Data.html]", + "position": { + "start": 33, + "end": 73 + } + }, + { + "target": "2b6e6995-508b-5a12-8b97-b1d3560d8aa9", + "relation": "embeds", + "ref": "[images/image-4-Quarterly Financial Performance.jpg]", + "position": { + "start": 799, + "end": 851 + } + }, + { + "target": "21d81faf-1ff7-5aba-b902-161d97ffd732", + "relation": "embeds", + "ref": "[tables/table-5 Quarterly Values.html]", + "position": { + "start": 893, + "end": 931 + } + } + ] + } + }, + { + "chunk_id": "10241b1c-edf7-56d0-95de-3f8c79b67396", + "type": "image", + "content": "\n
0\n[images/image-5-Quarterly Financial Growth.jpg]\n", + "path": "images/image-5-Quarterly Financial Growth.jpg", + "metadata": { + "length": 55, + "summary": "image-5\nChart Title: Quarterly Performance (Q1'24 - Q1'26)\n\nData Points by Quarter:\n- Q1'24: $4.71\n- Q2'24: $5.16\n- Q3'24: $6.03\n- Q4'24: $8.02\n- Q1'25: $6.43\n- Q2'25: $7.14\n- Q3'25: $1.05 (marked with footnote 1)\n- Q4'25: $8.88\n- Q1'26: $10.44 (marked with footnote 2)\n\nKey Trends:\n- Steady growth from Q1'24 to Q4'24 ($4.71 to $8.02).\n- A dip in Q1'25 followed by recovery in Q2'25.\n- Significant drop in Q3'25 ($1.05), the lowest value on the chart.\n- Strong", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-5-Quarterly Financial Growth.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "6522b598-11ed-5413-957c-9e3d99d36cc8", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue
Q1'24$4.71
Q2'24$5.16
Q3'24$6.03
Q4'24$8.02
Q1'25$6.43
Q2'25$7.14
Q3'25 (1)$1.05
Q4'25$8.88
Q1'26 (2)$10.44
", + "path": "tables/table-6 Quarterly Values.html", + "metadata": { + "length": 735, + "summary": "table-7\nThe table shows quarterly values from Q1'24 to Q1'26, with figures ranging from $1.05 to $10.44, indicating an overall upward trend.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-6 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Trend" + ], + "tokens": [] + } + }, + { + "chunk_id": "88ea2b15-d8ac-559d-8374-55f30e01732c", + "type": "text", + "content": "
0\n[images/image-5-Quarterly Financial Growth.jpg]\n\n
\nbar chart\n\n[tables/table-6 Quarterly Values.html]\n\n
\n(1) Net income includes a one-time, non-cash income tax charge to provision for income taxes of \\$15.93 billion accrued in the third quarter of 2025, related to the implementation of the One Big Beautiful Bill Act. Excluding this one-time tax charge, our third quarter 2025 diluted earnings per share would have increased by \\$6.20 to \\$7.25 compared to the reported diluted EPS of \\$1.05.\n(2) Net income includes an income tax benefit of \\$8.03 billion in the first quarter of 2026 related to the U.S. Corporate Alternative Minimum Tax transitional relief under Treasury Notice 2026-7. Excluding this tax benefit, diluted EPS would have been \\$7.31, lower by \\$3.13, compared to the reported diluted EPS of \\$10.44.", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Diluted Earnings Per Share", + "metadata": { + "length": 859, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "details", + "summary", + "bar", + "chart", + "Net", + "income", + "includes", + "time", + "cash", + "tax", + "charge", + "provision", + "taxes", + "15.93", + "billion", + "accrued", + "quarter", + "2025", + "related", + "implementation", + "One", + "Big", + "Beautiful", + "Bill", + "Act", + "Excluding", + "diluted", + "earnings", + "share", + "increased", + "6.20", + "7.25", + "compared", + "reported", + "EPS", + "1.05", + "benefit", + "8.03", + "2026", + "Corporate", + "Alternative", + "Minimum", + "Tax", + "transitional", + "relief", + "Treasury", + "Notice", + "7.31", + "lower", + "3.13", + "10.44" + ], + "keywords": [], + "connect_to": [ + { + "target": "10241b1c-edf7-56d0-95de-3f8c79b67396", + "relation": "embeds", + "ref": "[images/image-5-Quarterly Financial Growth.jpg]", + "position": { + "start": 22, + "end": 69 + } + }, + { + "target": "6522b598-11ed-5413-957c-9e3d99d36cc8", + "relation": "embeds", + "ref": "[tables/table-6 Quarterly Values.html]", + "position": { + "start": 111, + "end": 149 + } + } + ] + } + }, + { + "chunk_id": "af36b58e-2cee-5e6c-8ff4-c9fb50e97b81", + "type": "image", + "content": "\n
0\n[images/image-6-Financial Growth Projection.jpg]\n", + "path": "images/image-6-Financial Growth Projection.jpg", + "metadata": { + "length": 55, + "summary": "image-6\nQuarterly Data:\n- Q1'25: $13,692\n- Q1'26: $19,840\n\nAnnual Data:\n- 2024: $39,225\n- 2025: $72,215", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-6-Financial Growth Projection.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "b88d2c70-ee9e-5b87-ab52-b0dca01f0c15", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PeriodValue
Quarterly$13,692
Q1'26$19,840
2024$39,225
Annual$72,215
", + "path": "tables/table-7 Financial Data.html", + "metadata": { + "length": 432, + "summary": "table-8\nThe table shows financial values across different periods: Quarterly $13,692, Q1'26 $19,840, 2024 $39,225, and Annual $72,215.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-7 Financial Data.html", + "keywords": [ + "Quarterly", + "Annual", + "Values" + ], + "tokens": [] + } + }, + { + "chunk_id": "c75bd5ec-99c3-5021-8908-08e583d7f320", + "type": "text", + "content": "In Millions\n\n
0\n[images/image-6-Financial Growth Projection.jpg]\n\n
\nbar chart\n\n[tables/table-7 Financial Data.html]\n\n
\nCapital expenditures for periods presented were related to purchases of property and equipment and principal payments on finance leases.", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Capital Expenditures", + "metadata": { + "length": 291, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "In", + "Millions", + "details", + "summary", + "bar", + "chart", + "Capital", + "expenditures", + "periods", + "presented", + "related", + "purchases", + "property", + "equipment", + "principal", + "payments", + "finance", + "leases" + ], + "keywords": [], + "connect_to": [ + { + "target": "af36b58e-2cee-5e6c-8ff4-c9fb50e97b81", + "relation": "embeds", + "ref": "[images/image-6-Financial Growth Projection.jpg]", + "position": { + "start": 35, + "end": 83 + } + }, + { + "target": "b88d2c70-ee9e-5b87-ab52-b0dca01f0c15", + "relation": "embeds", + "ref": "[tables/table-7 Financial Data.html]", + "position": { + "start": 125, + "end": 161 + } + } + ] + } + }, + { + "chunk_id": "b715df54-7381-5136-b33f-9890f209b7c0", + "type": "image", + "content": "\n
0\n[images/image-7-Quarterly Data Trend.jpg]\n", + "path": "images/image-7-Quarterly Data Trend.jpg", + "metadata": { + "length": 55, + "summary": "image-7\nQ1'24: 3.24\nQ2'24: 3.27\nQ3'24: 3.29\nQ4'24: 3.35\nQ1'25: 3.43\nQ2'25: 3.48\nQ3'25: 3.54\nQ4'25: 3.58\nQ1'26: 3.56", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-7-Quarterly Data Trend.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "77c244bd-8c3d-5f95-a585-d0d453340034", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue
Q1'243.24
Q2'243.27
Q3'243.29
Q4'243.35
Q1'253.43
Q2'253.48
Q3'253.54
Q4'253.58
Q1'26 (1)3.56
", + "path": "tables/table-8 Quarterly Values.html", + "metadata": { + "length": 721, + "summary": "table-9\nThe table shows quarterly values from Q1'24 to Q1'26, increasing from 3.24 to 3.58, with a slight dip to 3.56 in Q1'26.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-8 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Trend" + ], + "tokens": [] + } + }, + { + "chunk_id": "f73965b0-c0e9-5908-b23e-bd02d9259fd1", + "type": "text", + "content": "In Billions\n\n
0\n[images/image-7-Quarterly Data Trend.jpg]\n\n
\nbar chart\n\n[tables/table-8 Quarterly Values.html]\n\n
\n(1) The slight quarter-over-quarter decline in DAP in the first quarter of 2026 was driven by internet disruptions in Iran, as well as a restriction on access to WhatsApp in Russia.\nWe define a daily active person (DAP) as a registered and logged-in user of Facebook, Instagram, Messenger, and/or WhatsApp (collectively, our \"Family\" of products) who visited at least one of these Family products through a mobile device application or using a web or mobile browser on a given day. The numbers for DAP do not include users on our other products unless they would otherwise qualify as DAP based on their other activities on our Family products.\nWe do not require people to use a common identifier or link their accounts to use multiple products in our Family, and therefore must seek to attribute multiple user accounts within and across products to individual people. Our calculations of DAP rely upon complex techniques, algorithms, and machine learning models that seek to estimate the underlying number of unique people using one or more of these products, including by matching user accounts within an individual product and across multiple products when we believe they are attributable to a single person, and counting such group of accounts as one person. As these techniques and models require significant judgment, are developed based on internal reviews of limited samples of user accounts, and are calibrated against user survey data, there is necessarily some margin of error in our estimates. For additional information, see \"Limitations of Key Metrics and Other Data\" located in the Appendix of this presentation.", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Family Daily Active People (DAP)", + "metadata": { + "length": 1783, + "summary": "The passage discusses a slight quarter-over-quarter decline in Daily Active People (DAP) in Q1 2026, attributed to internet disruptions in Iran and a restriction on WhatsApp access in Russia. DAP is defined as registered and logged-in users of Facebook, Instagram, Messenger, and/or WhatsApp who visit at least one of these products daily. The calculation involves complex techniques and machine learning models to estimate unique individuals, with inherent margin of error.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "In", + "Billions", + "details", + "summary", + "bar", + "chart", + "The", + "slight", + "quarter", + "decline", + "DAP", + "2026", + "driven", + "internet", + "disruptions", + "Iran", + "restriction", + "access", + "WhatsApp", + "Russia", + "We", + "define", + "daily", + "active", + "person", + "registered", + "logged", + "user", + "Facebook", + "Instagram", + "Messenger", + "collectively", + "Family", + "products", + "visited", + "mobile", + "device", + "application", + "web", + "browser", + "day", + "numbers", + "include", + "users", + "qualify", + "based", + "activities", + "require", + "people", + "common", + "identifier", + "link", + "accounts", + "multiple", + "seek", + "attribute", + "individual", + "Our", + "calculations", + "rely", + "complex", + "techniques", + "algorithms", + "machine", + "learning", + "models", + "estimate", + "underlying", + "number", + "unique", + "including", + "matching", + "product", + "attributable", + "single", + "counting", + "group", + "As", + "significant", + "judgment", + "developed", + "internal", + "reviews", + "limited", + "samples", + "calibrated", + "survey", + "data", + "necessarily", + "margin", + "error", + "estimates", + "For", + "additional", + "information", + "Limitations", + "Key", + "Metrics", + "Other", + "Data", + "located", + "Appendix", + "presentation" + ], + "keywords": [ + "DAP", + "quarter-over-quarter", + "internet disruptions" + ], + "connect_to": [ + { + "target": "b715df54-7381-5136-b33f-9890f209b7c0", + "relation": "embeds", + "ref": "[images/image-7-Quarterly Data Trend.jpg]", + "position": { + "start": 35, + "end": 76 + } + }, + { + "target": "77c244bd-8c3d-5f95-a585-d0d453340034", + "relation": "embeds", + "ref": "[tables/table-8 Quarterly Values.html]", + "position": { + "start": 118, + "end": 156 + } + } + ] + } + }, + { + "chunk_id": "aa5cb3b1-5f79-5129-8780-1c287ea081f9", + "type": "image", + "content": "\n
0\n[images/image-8-Quarterly Financial Data.jpg]\n", + "path": "images/image-8-Quarterly Financial Data.jpg", + "metadata": { + "length": 55, + "summary": "image-8\nQ1'24: $11.20\nQ2'24: $11.89\nQ3'24: $12.29\nQ4'24: $14.25\nQ1'25: $12.36\nQ2'25: $13.65\nQ3'25: $14.46\nQ4'25: $16.56\nQ1'26: $15.66", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-8-Quarterly Financial Data.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "0310c851-f6a3-5e12-ae44-5ad8b06980d3", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue ($)
Q1'2411.20
Q2'2411.89
Q3'2412.29
Q4'2414.25
Q1'2512.36
Q2'2513.65
Q3'2514.46
Q4'2516.56
Q1'2615.66
", + "path": "tables/table-9 Quarterly Values.html", + "metadata": { + "length": 730, + "summary": "table-10\nThe table shows quarterly values from Q1'24 to Q1'26, starting at $11.20 and ending at $15.66, with a peak of $16.56 in Q4'25.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-9 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Trend" + ], + "tokens": [] + } + }, + { + "chunk_id": "822fa03e-bd63-53fa-8c26-85d829b25c2f", + "type": "text", + "content": "
0\n[images/image-8-Quarterly Financial Data.jpg]\n\n
\nbar chart\n\n[tables/table-9 Quarterly Values.html]\n\n
\nWe define average revenue per person (ARPP) as our FoA revenue during a given quarter, divided by the average of the number of DAP at the beginning and end of the quarter.", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Family Average Revenue per Person (ARPP)", + "metadata": { + "length": 319, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "details", + "summary", + "bar", + "chart", + "We", + "define", + "average", + "revenue", + "person", + "ARPP", + "FoA", + "quarter", + "divided", + "number", + "DAP", + "beginning", + "end" + ], + "keywords": [], + "connect_to": [ + { + "target": "aa5cb3b1-5f79-5129-8780-1c287ea081f9", + "relation": "embeds", + "ref": "[images/image-8-Quarterly Financial Data.jpg]", + "position": { + "start": 22, + "end": 67 + } + }, + { + "target": "0310c851-f6a3-5e12-ae44-5ad8b06980d3", + "relation": "embeds", + "ref": "[tables/table-9 Quarterly Values.html]", + "position": { + "start": 109, + "end": 147 + } + } + ] + } + }, + { + "chunk_id": "b5f986d3-1c0d-52f0-841e-e7075c1fefa8", + "type": "image", + "content": "\n
0\n[images/image-9-Quarterly Growth Trend.jpg]\n", + "path": "images/image-9-Quarterly Growth Trend.jpg", + "metadata": { + "length": 55, + "summary": "image-9\nQ1'25: 5%\nQ2'25: 11%\nQ3'25: 14%\nQ4'25: 18%\nQ1'26: 19%", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-9-Quarterly Growth Trend.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "4fa3c11a-07e2-59c2-a659-8a32de28b309", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue (%)
Q1'255
Q2'2511
Q3'2514
Q4'2518
Q1'2619
", + "path": "tables/table-10 Quarterly Values.html", + "metadata": { + "length": 470, + "summary": "table-11\nThe table shows quarterly percentage values from Q1'25 (5%) to Q1'26 (19%), with a steady increase over five quarters.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-10 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "54e64331-f2e6-56b1-8811-82888a7d05d6", + "type": "image", + "content": "\n
0\n[images/image-10-Quarterly Percentage Growth.jpg]\n", + "path": "images/image-10-Quarterly Percentage Growth.jpg", + "metadata": { + "length": 56, + "summary": "image-10\nQ1'25: 4%\nQ2'25: 9%\nQ3'25: 8%\nQ4'25: 13%\nQ1'26: 13%", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-10-Quarterly Percentage Growth.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "9f128ac9-4c37-56c7-a2f7-a7eff8bfbbfe", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue (%)
Q1'254
Q2'259
Q3'258
Q4'2513
Q1'2613
", + "path": "tables/table-11 Quarterly Values.html", + "metadata": { + "length": 468, + "summary": "table-12\nThe table shows quarterly percentage values from Q1'25 to Q1'26, starting at 4%, rising to 9%, 8%, 13%, and ending at 13%.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-11 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "bf0afc23-5239-5463-bbf2-75433fc9bf91", + "type": "image", + "content": "\n
0\n[images/image-11-Quarterly Growth Trend.jpg]\n", + "path": "images/image-11-Quarterly Growth Trend.jpg", + "metadata": { + "length": 56, + "summary": "image-11\nChart Title: Quarterly Growth Trend (implied)\nX-Axis Labels: Q1'25, Q2'25, Q3'25, Q4'25, Q1'26\nData Points:\n- Q1'25: 5%\n- Q2'25: 6%\n- Q3'25: 9%\n- Q4'25: 13%\n- Q1'26: 17%\nTrend: Consistent upward growth over the five quarters.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-11-Quarterly Growth Trend.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "a5d52424-7c11-543f-a11e-2b381a149fac", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue (%)
Q1'255
Q2'256
Q3'259
Q4'2513
Q1'2617
", + "path": "tables/table-12 Quarterly Values.html", + "metadata": { + "length": 468, + "summary": "table-13\nThe table shows quarterly percentage values from Q1'25 (5%) to Q1'26 (17%), with a steady increase over five quarters.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-12 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "165cb9bd-677d-57b6-945d-25f2774911e5", + "type": "image", + "content": "\n
0\n[images/image-12-Quarterly Growth Trend.jpg]\n", + "path": "images/image-12-Quarterly Growth Trend.jpg", + "metadata": { + "length": 56, + "summary": "image-12\n- Q1'25: 9%\n- Q2'25: 16%\n- Q3'25: 23%\n- Q4'25: 24%\n- Q1'26: 23%", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-12-Quarterly Growth Trend.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "cb760784-f96e-572b-bb7b-137a772b9672", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue (%)
Q1'259
Q2'2516
Q3'2523
Q4'2524
Q1'2623
", + "path": "tables/table-13 Quarterly Values.html", + "metadata": { + "length": 470, + "summary": "table-14\nThe table shows quarterly percentage values from Q1'25 to Q1'26, starting at 9% and rising to a peak of 24% in Q4'25, then slightly decreasing to 23% in Q1'26.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-13 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "c31a8a40-eeed-52f0-b6c6-79ca910aac06", + "type": "image", + "content": "\n
0\n[images/image-13-Growth Trend 2025-2026.jpg]\n", + "path": "images/image-13-Growth Trend 2025-2026.jpg", + "metadata": { + "length": 56, + "summary": "image-13\nChart Type: Vertical Bar Chart\nX-Axis Categories (Time Periods): Q1'25, Q2'25, Q3'25, Q4'25, Q1'26\nData Values:\n- Q1'25: 1%\n- Q2'25: 7%\n- Q3'25: 9%\n- Q4'25: 14%\n- Q1'26: 17%\nTrend: Consistent upward growth over the five quarters.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-13-Growth Trend 2025-2026.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "9ea8a6bd-70c5-5cd1-8a98-2d41db83c89d", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue (%)
Q1'251
Q2'257
Q3'259
Q4'2514
Q1'2617
", + "path": "tables/table-14 Quarterly Values.html", + "metadata": { + "length": 468, + "summary": "table-15\nThe table shows quarterly percentage values from Q1'25 (1%) to Q1'26 (17%), with a steady increase over five quarters.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-14 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "946b1845-9da8-502b-8c22-bc890e9bd33a", + "type": "text", + "content": "
0\n[images/image-9-Quarterly Growth Trend.jpg]\n\n
\nbar chart\n\n[tables/table-10 Quarterly Values.html]\n\n
\nWorldwide\n\n
0\n[images/image-10-Quarterly Percentage Growth.jpg]\n\n
\nbar chart\n\n[tables/table-11 Quarterly Values.html]\n\n
\nUS & Canada\n\n
0\n[images/image-11-Quarterly Growth Trend.jpg]\n\n
\nbar chart\n\n[tables/table-12 Quarterly Values.html]\n\n
\nEurope\n\n
0\n[images/image-12-Quarterly Growth Trend.jpg]\n\n
\nbar chart\n\n[tables/table-13 Quarterly Values.html]\n\n
\nAsia-Pacific\n\n
0\n[images/image-13-Growth Trend 2025-2026.jpg]\n\n
\nbar chart\n\n[tables/table-14 Quarterly Values.html]\n\n
\nRest of World", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Ad Impressions Delivered YoY Percentage Change", + "metadata": { + "length": 803, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "details", + "summary", + "bar", + "chart", + "Worldwide", + "US", + "Canada", + "Europe", + "Asia", + "Pacific", + "Rest", + "World" + ], + "keywords": [], + "connect_to": [ + { + "target": "b5f986d3-1c0d-52f0-841e-e7075c1fefa8", + "relation": "embeds", + "ref": "[images/image-9-Quarterly Growth Trend.jpg]", + "position": { + "start": 22, + "end": 65 + } + }, + { + "target": "4fa3c11a-07e2-59c2-a659-8a32de28b309", + "relation": "embeds", + "ref": "[tables/table-10 Quarterly Values.html]", + "position": { + "start": 107, + "end": 146 + } + }, + { + "target": "54e64331-f2e6-56b1-8811-82888a7d05d6", + "relation": "embeds", + "ref": "[images/image-10-Quarterly Percentage Growth.jpg]", + "position": { + "start": 192, + "end": 241 + } + }, + { + "target": "9f128ac9-4c37-56c7-a2f7-a7eff8bfbbfe", + "relation": "embeds", + "ref": "[tables/table-11 Quarterly Values.html]", + "position": { + "start": 283, + "end": 322 + } + }, + { + "target": "bf0afc23-5239-5463-bbf2-75433fc9bf91", + "relation": "embeds", + "ref": "[images/image-11-Quarterly Growth Trend.jpg]", + "position": { + "start": 370, + "end": 414 + } + }, + { + "target": "a5d52424-7c11-543f-a11e-2b381a149fac", + "relation": "embeds", + "ref": "[tables/table-12 Quarterly Values.html]", + "position": { + "start": 456, + "end": 495 + } + }, + { + "target": "165cb9bd-677d-57b6-945d-25f2774911e5", + "relation": "embeds", + "ref": "[images/image-12-Quarterly Growth Trend.jpg]", + "position": { + "start": 538, + "end": 582 + } + }, + { + "target": "cb760784-f96e-572b-bb7b-137a772b9672", + "relation": "embeds", + "ref": "[tables/table-13 Quarterly Values.html]", + "position": { + "start": 624, + "end": 663 + } + }, + { + "target": "c31a8a40-eeed-52f0-b6c6-79ca910aac06", + "relation": "embeds", + "ref": "[images/image-13-Growth Trend 2025-2026.jpg]", + "position": { + "start": 712, + "end": 756 + } + }, + { + "target": "9ea8a6bd-70c5-5cd1-8a98-2d41db83c89d", + "relation": "embeds", + "ref": "[tables/table-14 Quarterly Values.html]", + "position": { + "start": 798, + "end": 837 + } + } + ] + } + }, + { + "chunk_id": "89b992d8-3719-5047-bd85-bf2049cb3b31", + "type": "image", + "content": "\n
0\n[images/image-14-Quarterly Percentage Data (2025-2026).jpg]\n", + "path": "images/image-14-Quarterly Percentage Data (2025-2026).jpg", + "metadata": { + "length": 56, + "summary": "image-14\nChart Title: Not visible\nData Points by Quarter:\n- Q1'25: 10%\n- Q2'25: 9%\n- Q3'25: 10%\n- Q4'25: 6%\n- Q1'26: 12%", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-14-Quarterly Percentage Data (2025-2026).jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "5a9a3ff0-3766-5811-887b-a4c69a905b66", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue (%)
Q1'2510
Q2'259
Q3'2510
Q4'256
Q1'2612
", + "path": "tables/table-15 Quarterly Values.html", + "metadata": { + "length": 469, + "summary": "table-16\nThe table shows quarterly percentage values from Q1'25 to Q1'26, ranging from 6% to 12%, with Q4'25 having the lowest value.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-15 Quarterly Values.html", + "keywords": [ + "Quarterly", + "Values", + "Percent" + ], + "tokens": [] + } + }, + { + "chunk_id": "d30a3592-47ea-529b-940e-06e06192447d", + "type": "image", + "content": "\n
0\n[images/image-15-Quarterly Percentage Data (2025-2026).jpg]\n", + "path": "images/image-15-Quarterly Percentage Data (2025-2026).jpg", + "metadata": { + "length": 56, + "summary": "image-15\nChart Title: Not visible\nData Points by Quarter:\n- Q1'25: 14%\n- Q2'25: 11%\n- Q3'25: 13%\n- Q4'25: 9%\n- Q1'26: 14%\n\nKey Trends:\n- Lowest value recorded in Q4'25 at 9%.\n- Highest values observed in Q1'25 and Q1'26 at 14%.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-15-Quarterly Percentage Data (2025-2026).jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "8683a3c7-6919-5b6d-9c2c-9a0eb6f0b6b6", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue (%)
Q1'2514
Q2'2511
Q3'2513
Q4'259
Q1'2614
", + "path": "tables/table-16 Quarterly Values.html", + "metadata": { + "length": 470, + "summary": "table-17\nThe table shows quarterly percentage values from Q1'25 to Q1'26, with values ranging from 9% to 14%.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-16 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "762954c1-ba35-524d-80c0-f06075229c1d", + "type": "image", + "content": "\n
0\n[images/image-16-Quarterly Percentage Trends.jpg]\n", + "path": "images/image-16-Quarterly Percentage Trends.jpg", + "metadata": { + "length": 56, + "summary": "image-16\nChart Title: Not visible\nX-Axis Labels (Time Periods): Q1'25, Q2'25, Q3'25, Q4'25, Q1'26\nData Points:\n- Q1'25: 9%\n- Q2'25: 17%\n- Q3'25: 19%\n- Q4'25: 12%\n- Q1'26: 19%", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-16-Quarterly Percentage Trends.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "5afee24d-42d3-5007-b886-b5d5ad63061e", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue (%)
Q1'259
Q2'2517
Q3'2519
Q4'2512
Q1'2619
", + "path": "tables/table-17 Quarterly Values.html", + "metadata": { + "length": 470, + "summary": "table-18\nThe table shows quarterly percentage values from Q1'25 to Q1'26, ranging from 9% to 19%, with the highest in Q3'25 and Q1'26.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-17 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "00c61283-6184-5f21-b017-8a2071ef7eac", + "type": "image", + "content": "\n
0\n[images/image-17-Quarterly Growth Trend.jpg]\n", + "path": "images/image-17-Quarterly Growth Trend.jpg", + "metadata": { + "length": 56, + "summary": "image-17\nQ1'25: 3%\nQ2'25: 2%\nQ3'25: 1%\nQ4'25: (2)%\nQ1'26: 5%", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-17-Quarterly Growth Trend.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "b23a116d-5aea-54a3-bcce-5d8436f03562", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue (%)
Q1'253
Q2'252
Q3'251
Q4'25-2
Q1'265
", + "path": "tables/table-18 Quarterly Values.html", + "metadata": { + "length": 467, + "summary": "table-19\nThe table shows quarterly percentage values from Q1'25 to Q1'26, starting at 3%, decreasing to -2% in Q4'25, then rising to 5%.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-18 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "a7037604-0c50-57eb-94a4-42dfea6d1a37", + "type": "image", + "content": "\n
0\n[images/image-18-Quarterly Percentage Data (2025-2026).jpg]\n", + "path": "images/image-18-Quarterly Percentage Data (2025-2026).jpg", + "metadata": { + "length": 56, + "summary": "image-18\nChart Title: Not visible\nX-Axis Labels: Q1'25, Q2'25, Q3'25, Q4'25, Q1'26\nData Points:\n- Q1'25: 17%\n- Q2'25: 15%\n- Q3'25: 20%\n- Q4'25: 15%\n- Q1'26: 18%", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "images/image-18-Quarterly Percentage Data (2025-2026).jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "1283ccb7-06c4-547a-a21a-59fa0b3930c3", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
QuarterValue (%)
Q1'2517
Q2'2515
Q3'2520
Q4'2515
Q1'2618
", + "path": "tables/table-19 Quarterly Values.html", + "metadata": { + "length": 471, + "summary": "table-20\nThe table shows quarterly percentage values from Q1'25 to Q1'26, ranging from 15% to 20%, with Q3'25 having the highest value at 20%.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-19 Quarterly Values.html", + "keywords": [ + "Quarter", + "Value", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "946b1845-9da8-502b-8c22-bc890e9bd33a", + "type": "text", + "content": "
0\n[images/image-14-Quarterly Percentage Data (2025-2026).jpg]\n\n
\nbar chart\n\n[tables/table-15 Quarterly Values.html]\n\n
\nWorldwide\n\n
0\n[images/image-15-Quarterly Percentage Data (2025-2026).jpg]\n\n
\nbar chart\n\n[tables/table-16 Quarterly Values.html]\n\n
\nUS & Canada\n\n
0\n[images/image-16-Quarterly Percentage Trends.jpg]\n\n
\nbar chart\n\n[tables/table-17 Quarterly Values.html]\n\n
\nEurope\n\n
0\n[images/image-17-Quarterly Growth Trend.jpg]\n\n
\nbar chart\n\n[tables/table-18 Quarterly Values.html]\n\n
\nAsia-Pacific\n\n
0\n[images/image-18-Quarterly Percentage Data (2025-2026).jpg]\n\n
\nbar chart\n\n[tables/table-19 Quarterly Values.html]\n\n
\nRest of World", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Average Price Per Ad YoY Percentage Change", + "metadata": { + "length": 804, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "details", + "summary", + "bar", + "chart", + "Worldwide", + "US", + "Canada", + "Europe", + "Asia", + "Pacific", + "Rest", + "World" + ], + "keywords": [], + "connect_to": [ + { + "target": "89b992d8-3719-5047-bd85-bf2049cb3b31", + "relation": "embeds", + "ref": "[images/image-14-Quarterly Percentage Data (2025-2026).jpg]", + "position": { + "start": 22, + "end": 81 + } + }, + { + "target": "5a9a3ff0-3766-5811-887b-a4c69a905b66", + "relation": "embeds", + "ref": "[tables/table-15 Quarterly Values.html]", + "position": { + "start": 123, + "end": 162 + } + }, + { + "target": "d30a3592-47ea-529b-940e-06e06192447d", + "relation": "embeds", + "ref": "[images/image-15-Quarterly Percentage Data (2025-2026).jpg]", + "position": { + "start": 208, + "end": 267 + } + }, + { + "target": "8683a3c7-6919-5b6d-9c2c-9a0eb6f0b6b6", + "relation": "embeds", + "ref": "[tables/table-16 Quarterly Values.html]", + "position": { + "start": 309, + "end": 348 + } + }, + { + "target": "762954c1-ba35-524d-80c0-f06075229c1d", + "relation": "embeds", + "ref": "[images/image-16-Quarterly Percentage Trends.jpg]", + "position": { + "start": 396, + "end": 445 + } + }, + { + "target": "5afee24d-42d3-5007-b886-b5d5ad63061e", + "relation": "embeds", + "ref": "[tables/table-17 Quarterly Values.html]", + "position": { + "start": 487, + "end": 526 + } + }, + { + "target": "00c61283-6184-5f21-b017-8a2071ef7eac", + "relation": "embeds", + "ref": "[images/image-17-Quarterly Growth Trend.jpg]", + "position": { + "start": 569, + "end": 613 + } + }, + { + "target": "b23a116d-5aea-54a3-bcce-5d8436f03562", + "relation": "embeds", + "ref": "[tables/table-18 Quarterly Values.html]", + "position": { + "start": 655, + "end": 694 + } + }, + { + "target": "a7037604-0c50-57eb-94a4-42dfea6d1a37", + "relation": "embeds", + "ref": "[images/image-18-Quarterly Percentage Data (2025-2026).jpg]", + "position": { + "start": 743, + "end": 802 + } + }, + { + "target": "1283ccb7-06c4-547a-a21a-59fa0b3930c3", + "relation": "embeds", + "ref": "[tables/table-19 Quarterly Values.html]", + "position": { + "start": 844, + "end": 883 + } + } + ] + } + }, + { + "chunk_id": "84e9ab1c-4b57-5907-8a88-7b1464921a7d", + "type": "text", + "content": "", + "path": "Earnings-Presentation-Q1-2026.pdf/Appendix", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0de1bf51-3db0-563c-9a63-c49b71d6b225", + "type": "table", + "content": "
Q1'24Q2'24Q3'24Q4'24Q1'25Q2'25Q3'25Q4'25Q1'26
Net cash provided by operating activities$ 19,246$ 19,370$ 24,724$ 27,988$ 24,026$ 25,561$ 29,999$ 36,214$ 32,226
Less: Purchases of property and equipment6,4008,1738,25814,42512,94116,53818,82921,38318,997
Less: Principal payments on finance leases315299944411751474545754843
Free Cash Flow$ 12,531$ 10,898$ 15,522$ 13,152$ 10,334$ 8,549$ 10,625$ 14,077$ 12,386
", + "path": "tables/table-20 Free Cash Flow.html", + "metadata": { + "length": 959, + "summary": "table-21\nThe table shows net cash from operations, property purchases, finance lease payments, and free cash flow from Q1 2024 to Q1 2026, with free cash flow ranging from $8.5B to $15.5B.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "file_path": "tables/table-20 Free Cash Flow.html", + "keywords": [ + "Free Cash Flow", + "Operating Activities", + "Capital Expenditures" + ], + "tokens": [] + } + }, + { + "chunk_id": "d7b5e91a-79c7-56d2-8e9e-d1f5e6f4df71", + "type": "text", + "content": "In Millions\n\n[tables/table-20 Free Cash Flow.html]\n\nFree cash flow (FCF) is a non-GAAP financial measure that has limitations as an analytical tool, and you should not consider it in isolation or as a substitute for analysis of other GAAP financial measures, such as net cash provided by operating activities. FCF is not intended to represent our residual cash flow available for discretionary expenses. Some of the limitations of FCF are: (i) FCF does not reflect our future contractual commitments, and (ii) other companies in our industry present similarly titled measures differently than we do, limiting their usefulness as comparative measures.", + "path": "Earnings-Presentation-Q1-2026.pdf/Appendix/Free Cash Flow Reconciliation", + "metadata": { + "length": 689, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "In", + "Millions", + "Free", + "cash", + "flow", + "FCF", + "GAAP", + "financial", + "measure", + "limitations", + "analytical", + "tool", + "isolation", + "substitute", + "analysis", + "measures", + "net", + "provided", + "operating", + "activities", + "intended", + "represent", + "residual", + "discretionary", + "expenses", + "Some", + "reflect", + "future", + "contractual", + "commitments", + "ii", + "companies", + "industry", + "present", + "similarly", + "titled", + "differently", + "limiting", + "usefulness", + "comparative" + ], + "keywords": [], + "connect_to": [ + { + "target": "0de1bf51-3db0-563c-9a63-c49b71d6b225", + "relation": "embeds", + "ref": "[tables/table-20 Free Cash Flow.html]", + "position": { + "start": 13, + "end": 50 + } + } + ] + } + }, + { + "chunk_id": "8073d0e9-c154-5d30-a77c-64d8dca092e1", + "type": "text", + "content": "Family Metrics\nThe numbers for our key metrics are calculated using internal company data based on the activity of user accounts. We report our estimates of the numbers of our daily active people (DAP) and average revenue per person (ARPP) (collectively, our \"Family metrics\") based on the activity of users who visited at least one of Facebook, Instagram, Messenger, and WhatsApp (collectively, our \"Family\" of products) during the applicable period of measurement.\nWhile these numbers are based on what we believe to be reasonable estimates of our user base for the applicable period of measurement, there are inherent challenges in measuring usage of our products across large online and mobile populations around the world. The methodologies used to measure these metrics require significant judgment and are also susceptible to algorithm or other technical errors. In addition, we are continually seeking to improve our estimates of our user base, and such estimates may change due to improvements or changes in our methodology. We regularly review our processes for calculating these metrics, and from time to time we discover inaccuracies in our metrics or make adjustments to improve their accuracy, which can result in adjustments to our historical metrics. Our ability to recalculate our historical metrics may be impacted by data limitations or other factors that require us to apply different methodologies for such adjustments. We generally do not intend to update previously disclosed Family metrics for any such inaccuracies or adjustments that are within the error margins disclosed below.\nIn addition, our Family metrics estimates will differ from estimates published by third parties due to differences in methodology or other factors such as data limitations or other challenges in measuring large online and mobile populations. For example, our methodologies include measurements of our user base that have in some instances exceeded estimates of addressable online and mobile populations that are based on data published by third parties.\nMany people in our community have user accounts on more than one of our products, and some people have multiple user accounts within an individual product. Accordingly, for our Family metrics, we do not seek to count the total number of user accounts across our products because we believe that would not reflect the actual size of our community. Rather, our Family metrics represent our estimates of the number of unique people using at least one of Facebook, Instagram, Messenger, and WhatsApp. We do not require people to use a common identifier or link their accounts to use multiple products in our Family, and therefore must seek to attribute multiple user accounts within and across products to individual people. To calculate these metrics, we rely upon complex techniques, algorithms and machine learning models that seek to count the individual people behind user accounts, including by matching multiple user accounts within an individual product and across multiple products when we believe they are attributable to a single person, and counting such group of accounts as one person. These techniques and models require significant judgment, are subject to data and other limitations discussed below, and inherently are subject to statistical variances and uncertainties. We estimate the potential error in our Family metrics primarily based on user survey data as described further below, which itself is subject to error as well. While we expect the error margin for our Family metrics to vary from period to period, we estimate that such margin generally will be approximately 3% of our worldwide DAP. At our scale, it is very difficult to attribute multiple user accounts within and across products to individual people, and it is possible that the actual numbers of unique people using our products may vary significantly from our estimates, potentially beyond our estimated error margins. As a result, it is also possible that our Family metrics may indicate changes or trends in user numbers that do not match actual changes or trends.", + "path": "Earnings-Presentation-Q1-2026.pdf/Appendix/Limitations of Key Metrics and Other Data", + "metadata": { + "length": 4114, + "summary": "The text describes how Family Metrics, including daily active people (DAP) and average revenue per person (ARPP), are calculated using internal data from Facebook, Instagram, Messenger, and WhatsApp. It highlights challenges in measuring unique users across multiple products, reliance on complex algorithms and machine learning to attribute accounts to individuals, and inherent statistical uncertainties. The estimated error margin is about 3% of worldwide DAP, but actual numbers may vary significantly.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "Family", + "Metrics", + "The", + "numbers", + "key", + "metrics", + "calculated", + "internal", + "company", + "data", + "based", + "activity", + "user", + "accounts", + "We", + "report", + "estimates", + "daily", + "active", + "people", + "DAP", + "average", + "revenue", + "person", + "ARPP", + "collectively", + "users", + "visited", + "Facebook", + "Instagram", + "Messenger", + "WhatsApp", + "products", + "applicable", + "period", + "measurement", + "While", + "reasonable", + "base", + "inherent", + "challenges", + "measuring", + "usage", + "large", + "online", + "mobile", + "populations", + "world", + "methodologies", + "measure", + "require", + "significant", + "judgment", + "susceptible", + "algorithm", + "technical", + "errors", + "In", + "addition", + "continually", + "seeking", + "improve", + "change", + "due", + "improvements", + "methodology", + "regularly", + "review", + "processes", + "calculating", + "time", + "discover", + "inaccuracies", + "make", + "adjustments", + "accuracy", + "result", + "historical", + "Our", + "ability", + "recalculate", + "impacted", + "limitations", + "factors", + "apply", + "generally", + "intend", + "update", + "previously", + "disclosed", + "error", + "margins", + "differ", + "published", + "parties", + "differences", + "For", + "include", + "measurements", + "instances", + "exceeded", + "addressable", + "Many", + "community", + "multiple", + "individual", + "product", + "Accordingly", + "seek", + "count", + "total", + "number", + "reflect", + "actual", + "size", + "Rather", + "represent", + "unique", + "common", + "identifier", + "link", + "attribute", + "To", + "calculate", + "rely", + "complex", + "techniques", + "algorithms", + "machine", + "learning", + "models", + "including", + "matching", + "attributable", + "single", + "counting", + "group", + "These", + "subject", + "discussed", + "inherently", + "statistical", + "variances", + "uncertainties", + "estimate", + "potential", + "primarily", + "survey", + "expect", + "margin", + "vary", + "approximately", + "3%", + "worldwide", + "At", + "scale", + "difficult", + "significantly", + "potentially", + "estimated", + "As", + "trends", + "match" + ], + "keywords": [ + "DAP", + "ARPP", + "user estimation" + ], + "connect_to": [] + } + }, + { + "chunk_id": "dfe08e92-9dfe-52f0-8032-ef0cdcd5408f", + "type": "text", + "content": "To calculate our estimates of DAP, we currently use a series of machine learning models that are developed based on internal reviews of limited samples of user accounts and calibrated against user survey data. We apply significant judgment in designing these models and calculating these estimates. For example, to match user accounts within individual products and across multiple products, we use data signals such as similar device information, IP addresses, and user names. We also calibrate our models against data from periodic user surveys of varying sizes and frequency across our products, which survey questions are based on monthly usage, and which are inherently subject to error. The timing and results of such user surveys have in the past contributed, and may in the future contribute, to changes in our reported Family metrics from period to period. In addition, our data limitations may affect our understanding of certain details of our business and increase the risk of error for our Family metrics estimates. Our techniques and models rely on a variety of data signals from different products, and we rely on more limited data signals for some products compared to others. For example, as a result of limited visibility into encrypted products, we have fewer data signals from WhatsApp user accounts and primarily rely on phone numbers and device information to match WhatsApp user accounts with accounts on our other products. Any loss of access to data signals we use in our process for calculating Family metrics, whether as a result of our own product decisions, actions by third-party browser or mobile platforms, regulatory or legislative requirements, or other factors, also may impact the stability or accuracy of our reported Family metrics, as well as our ability to report these metrics at all. Our estimates of Family metrics also may change as our methodologies evolve, including through the application of new data signals or technologies, product changes, or other improvements in our user surveys, algorithms, or machine learning that may improve our ability to match accounts within and across our products or otherwise evaluate the broad population of our users. In addition, such evolution may allow us to identify previously undetected violating accounts (as defined below).\nWe regularly evaluate our Family metrics to estimate the percentage of our DAP consisting solely of \"violating\" accounts. We define \"violating\" accounts as accounts which we believe are intended to be used for purposes that violate our terms of service, including bots and spam. Our violating accounts estimation is based on an internal review of a limited sample of accounts, and we apply significant judgment in making this determination. For example, we look for account information and behaviors associated with Facebook and Instagram accounts that appear to be inauthentic to the reviewers, but we have less visibility into WhatsApp user activity due to encryption. In addition, if we believe an individual person has one or more violating accounts, we do not include such person in our violating accounts estimation as long as we believe they have one active account that does not constitute a violating account. In the fourth quarter of 2025, we made certain updates to the methodology we use for this estimation, including to incorporate updated data signals as a result of improvements in our ability to identify activity we believe to be violating our policies, as well as to focus on the most recent account activity when determining whether to include a person in our violating accounts estimation. Accordingly, in the fourth quarter of 2025, we estimated that less than $5\\%$ of our worldwide DAP consisted solely of violating accounts. We believe the increase compared to our prior estimation was a result of the methodology update described above. From time to time, we disable certain user accounts, make product changes, or take other actions to reduce the number of violating accounts among our users, which may also reduce our DAP estimates in a particular period. We intend to continue to disclose our estimates of the percentage of our DAP consisting solely of violating accounts on an annual basis. Violating accounts are very difficult to measure at our scale, and it is possible that the actual number of violating accounts may vary significantly from our estimates.\nUser Geography\nOur estimates for revenue by user location, as well as year-over-year percentage changes in ad impressions delivered and the average price per ad by user location, are also affected by data limitations and other challenges in measuring user geography. Our data regarding the geographic location of our users is estimated based on a number of factors, such as the user's IP address and self-disclosed location. These factors may not always accurately reflect the user's actual location. For example, a user may appear to be accessing our products from the location of the proxy server that the user connects to rather than from the user's actual location. The methodologies used to measure our metrics are also susceptible to algorithm or other technical errors.", + "path": "Earnings-Presentation-Q1-2026.pdf/Appendix/Limitations of Key Metrics and Other Data_2", + "metadata": { + "length": 5182, + "summary": "The passage describes the methodology for estimating Daily Active People (DAP) using machine learning models, data signals, and user surveys. It highlights challenges like limited data for encrypted products (e.g., WhatsApp) and reliance on IP addresses and device info. It also covers the estimation of violating accounts (bots/spam), which is difficult at scale, and notes a methodology update in Q4 2025 that increased the estimated percentage of violating accounts to less than 5%. Additionally, it discusses limitations in measuring user geography due to factors like proxy servers.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Earnings Presentation Q1 2026, Appendix", + "tokens": [ + "To", + "calculate", + "estimates", + "DAP", + "series", + "machine", + "learning", + "models", + "developed", + "based", + "internal", + "reviews", + "limited", + "samples", + "user", + "accounts", + "calibrated", + "survey", + "data", + "We", + "apply", + "significant", + "judgment", + "designing", + "calculating", + "For", + "match", + "individual", + "products", + "multiple", + "signals", + "similar", + "device", + "information", + "IP", + "addresses", + "names", + "calibrate", + "periodic", + "surveys", + "varying", + "sizes", + "frequency", + "questions", + "monthly", + "usage", + "inherently", + "subject", + "error", + "The", + "timing", + "results", + "past", + "contributed", + "future", + "contribute", + "reported", + "Family", + "metrics", + "period", + "In", + "addition", + "limitations", + "affect", + "understanding", + "details", + "business", + "increase", + "risk", + "Our", + "techniques", + "rely", + "variety", + "compared", + "result", + "visibility", + "encrypted", + "fewer", + "WhatsApp", + "primarily", + "phone", + "numbers", + "Any", + "loss", + "access", + "process", + "product", + "decisions", + "actions", + "party", + "browser", + "mobile", + "platforms", + "regulatory", + "legislative", + "requirements", + "factors", + "impact", + "stability", + "accuracy", + "ability", + "report", + "change", + "methodologies", + "evolve", + "including", + "application", + "technologies", + "improvements", + "algorithms", + "improve", + "evaluate", + "broad", + "population", + "users", + "evolution", + "identify", + "previously", + "undetected", + "violating", + "defined", + "regularly", + "estimate", + "percentage", + "consisting", + "solely", + "define", + "intended", + "purposes", + "violate", + "terms", + "service", + "bots", + "spam", + "estimation", + "review", + "sample", + "making", + "determination", + "account", + "behaviors", + "Facebook", + "Instagram", + "inauthentic", + "reviewers", + "activity", + "due", + "encryption", + "person", + "include", + "long", + "active", + "constitute", + "fourth", + "quarter", + "2025", + "made", + "updates", + "methodology", + "incorporate", + "updated", + "policies", + "focus", + "recent", + "determining", + "Accordingly", + "estimated", + "worldwide", + "consisted", + "prior", + "update", + "From", + "time", + "disable", + "make", + "reduce", + "number", + "intend", + "continue", + "disclose", + "annual", + "basis", + "Violating", + "difficult", + "measure", + "scale", + "actual", + "vary", + "significantly", + "User", + "Geography", + "revenue", + "location", + "year", + "ad", + "impressions", + "delivered", + "average", + "price", + "affected", + "challenges", + "measuring", + "geography", + "geographic", + "address", + "disclosed", + "These", + "accurately", + "reflect", + "accessing", + "proxy", + "server", + "connects", + "susceptible", + "algorithm", + "technical", + "errors" + ], + "keywords": [ + "DAP", + "violating accounts", + "methodology" + ], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/doc_nav.json b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/doc_nav.json new file mode 100644 index 000000000..2fd8986eb --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/doc_nav.json @@ -0,0 +1,306 @@ +{ + "version": "1.0", + "file_name": "Meta Q1 2026 Earnings Presentation.pdf", + "stats": { + "total_chunks": 55, + "text_chunks": 16, + "image_chunks": 18, + "table_chunks": 21, + "max_depth": 3 + }, + "sections": [ + { + "title": "Meta Earnings Presentation Q1 2026", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026", + "level": 1, + "summary": "", + "chunk_count": 12, + "children": [ + { + "title": "Advertising Revenue by User Geography", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Advertising Revenue by User Geography", + "level": 2, + "summary": "In Millions # Meta Earnings Presentation Q1 2026
0 [images/image-1-Revenue Forecast by Region.jpg]
stacked bar chart [tables/table-0 Revenue by Region.html]
Our revenue by user geography is geographically apportioned based on our estimation of the geographic location of our users when they perform a revenue-generating activity. This allocation differs from our revenue disaggregated by geography disclosure in our condensed consolidated financial statements where revenue is geographically apportioned based on the addresses of our customers.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Revenue by User Geography", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Revenue by User Geography", + "level": 2, + "summary": "In Millions
0 [images/image-2-Revenue Breakdown by Region and Quarter.jpg]
stacked bar chart [tables/table-1 Quarterly Revenue.html]
Our revenue by user geography is geographically apportioned based on our estimation of the geographic location of our users when they perform a revenue-generating activity. This allocation differs from our revenue disaggregated by geography disclosure in our condensed consolidated financial statements where revenue is geographically apportioned based on the addresses of our customers.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Segment Results", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results", + "level": 2, + "summary": "In Millions, Except Percentages [tables/table-2 Meta Revenue Data.html] We report our financial results based on two reportable segments: Family of Apps (FoA) and Reality Labs (RL). FoA includes Facebook, Instagram, Messenger, WhatsApp, and other services. RL includes our virtual and augmented reality related consumer hardware, software, and content.", + "chunk_count": 9, + "children": [ + { + "title": "Expenses as a Percentage of Revenue", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Expenses as a Percentage of Revenue", + "level": 3, + "summary": "## Segment Results
0 [images/image-3-Quarterly Percentage Breakdown.jpg]
stacked bar chart [tables/table-3 Segment Margins.html]
General & Administrative Marketing & Sales Research & Development Cost of Revenue", + "chunk_count": 1, + "children": [] + }, + { + "title": "Effective Tax Rate", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Effective Tax Rate", + "level": 3, + "summary": "The passage discusses the impact of one-time tax items on reported net income and effective tax rates. In Q3 2025, a $15.93 billion non-cash income tax charge related to the One Big Beautiful Bill Act increased the effective tax rate to 87%; excluding it, the rate would have been 14% and net income would have been $18.64 billion. In Q1 2026, an $8.03 billion tax benefit from U.S. Corporate Alternative Minimum Tax transitional relief reduced the effective tax rate; excluding it, the rate would have been 14% and net income would have been $18.74 billion. A bar chart showing effective tax rates by quarter is referenced.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Diluted Earnings Per Share", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Diluted Earnings Per Share", + "level": 3, + "summary": "
0 [images/image-5-Quarterly Financial Growth.jpg]
bar chart [tables/table-6 Quarterly Values.html]
(1) Net income includes a one-time, non-cash income tax charge to provision for income taxes of \\$15.93 billion accrued in the third quarter of 2025, related to the implementation of the One Big Beautiful Bill Act. Excluding this one-time tax charge, our third quarter 2025 diluted earnings per share would have increased by \\$6.20 to \\$7.25 compared to the reported diluted EPS of \\$1.05. (2) Net income includes an income tax benefit of \\$8.03 billion in the first quarter of 2026 related to the U.S. Corporate Alternative Minimum Tax transitional relief under Treasury Notice 2026-7. Excluding this tax benefit, diluted EPS would have been \\$7.31, lower by \\$3.13, compared to the reported diluted EPS of \\$10.44.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Capital Expenditures", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Capital Expenditures", + "level": 3, + "summary": "In Millions
0 [images/image-6-Financial Growth Projection.jpg]
bar chart [tables/table-7 Financial Data.html]
Capital expenditures for periods presented were related to purchases of property and equipment and principal payments on finance leases.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Family Daily Active People (DAP)", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Family Daily Active People (DAP)", + "level": 3, + "summary": "The passage discusses a slight quarter-over-quarter decline in Daily Active People (DAP) in Q1 2026, attributed to internet disruptions in Iran and a restriction on WhatsApp access in Russia. DAP is defined as registered and logged-in users of Facebook, Instagram, Messenger, and/or WhatsApp who visit at least one of these products daily. The calculation involves complex techniques and machine learning models to estimate unique individuals, with inherent margin of error.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Family Average Revenue per Person (ARPP)", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Family Average Revenue per Person (ARPP)", + "level": 3, + "summary": "
0 [images/image-8-Quarterly Financial Data.jpg]
bar chart [tables/table-9 Quarterly Values.html]
We define average revenue per person (ARPP) as our FoA revenue during a given quarter, divided by the average of the number of DAP at the beginning and end of the quarter.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Ad Impressions Delivered YoY Percentage Change", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Ad Impressions Delivered YoY Percentage Change", + "level": 3, + "summary": "
0 [images/image-9-Quarterly Growth Trend.jpg]
bar chart [tables/table-10 Quarterly Values.html]
Worldwide
0 [images/image-10-Quarterly Percentage Growth.jpg]
bar chart [tables/table-11 Quarterly Values.html]
US & Canada
0 [images/image-11-Quarterly Growth Trend.jpg]
bar chart [tables/table-12 Quarterly Values.html]
Europe
0 [images/image-12-Quarterly Growth Trend.jpg]
bar chart [tables/table-13 Quarterly Values.html]
Asia-Pacific
0 [images/image-13-Growth Trend 2025-2026.jpg]
bar chart [tables/table-14 Quarterly Values.html]
Rest of World", + "chunk_count": 1, + "children": [] + }, + { + "title": "Average Price Per Ad YoY Percentage Change", + "path": "Earnings-Presentation-Q1-2026.pdf/Meta Earnings Presentation Q1 2026/Segment Results/Average Price Per Ad YoY Percentage Change", + "level": 3, + "summary": "
0 [images/image-14-Quarterly Percentage Data (2025-2026).jpg]
bar chart [tables/table-15 Quarterly Values.html]
Worldwide
0 [images/image-15-Quarterly Percentage Data (2025-2026).jpg]
bar chart [tables/table-16 Quarterly Values.html]
US & Canada
0 [images/image-16-Quarterly Percentage Trends.jpg]
bar chart [tables/table-17 Quarterly Values.html]
Europe
0 [images/image-17-Quarterly Growth Trend.jpg]
bar chart [tables/table-18 Quarterly Values.html]
Asia-Pacific
0 [images/image-18-Quarterly Percentage Data (2025-2026).jpg]
bar chart [tables/table-19 Quarterly Values.html]
Rest of World", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Appendix", + "path": "Earnings-Presentation-Q1-2026.pdf/Appendix", + "level": 1, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "Free Cash Flow Reconciliation", + "path": "Earnings-Presentation-Q1-2026.pdf/Appendix/Free Cash Flow Reconciliation", + "level": 2, + "summary": "In Millions [tables/table-20 Free Cash Flow.html] Free cash flow (FCF) is a non-GAAP financial measure that has limitations as an analytical tool, and you should not consider it in isolation or as a substitute for analysis of other GAAP financial measures, such as net cash provided by operating activities. FCF is not intended to represent our residual cash flow available for discretionary expenses. Some of the limitations of FCF are: (i) FCF does not reflect our future contractual commitments, and (ii) other companies in our industry present similarly titled measures differently than we do, limiting their usefulness as comparative measures.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Limitations of Key Metrics and Other Data", + "path": "Earnings-Presentation-Q1-2026.pdf/Appendix/Limitations of Key Metrics and Other Data", + "level": 2, + "summary": "The text describes how Family Metrics, including daily active people (DAP) and average revenue per person (ARPP), are calculated using internal data from Facebook, Instagram, Messenger, and WhatsApp. It highlights challenges in measuring unique users across multiple products, reliance on complex algorithms and machine learning to attribute accounts to individuals, and inherent statistical uncertainties. The estimated error margin is about 3% of worldwide DAP, but actual numbers may vary significantly.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Limitations of Key Metrics and Other Data_2", + "path": "Earnings-Presentation-Q1-2026.pdf/Appendix/Limitations of Key Metrics and Other Data_2", + "level": 2, + "summary": "The passage describes the methodology for estimating Daily Active People (DAP) using machine learning models, data signals, and user surveys. It highlights challenges like limited data for encrypted products (e.g., WhatsApp) and reliance on IP addresses and device info. It also covers the estimation of violating accounts (bots/spam), which is difficult at scale, and notes a methodology update in Q4 2025 that increased the estimated percentage of violating accounts to less than 5%. Additionally, it discusses limitations in measuring user geography due to factors like proxy servers.", + "chunk_count": 1, + "children": [] + } + ] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-Revenue Forecast by Region.jpg", + "summary": "image-1 Q1'24: $35,635 (US & Canada: $15,451; Europe: $8,327; Asia-Pacific: $7,338; Rest of World: $4,519) Q2'24: $38,329 (US & Canada: $16,593; Europe: $9,135; Asia-Pacific: $7,721; Rest of World: $4,880) Q3'24: $39,885 (US & Canada: $17,389; Europe: $9,358; Asia-Pacific: $8,050; Rest of World: $5,088) Q4'24: $46,783 (US & Canada: $20,982; Europe: $11,154; Asia-Pacific: $9,012; Rest of World: $5,635) Q1'25:" + }, + { + "path": "images/image-2-Revenue Breakdown by Region and Quarter.jpg", + "summary": "image-2 Chart Title: Not visible (Stacked Bar Chart) Time Range: Q1'24 to Q1'26 Legend Categories: - US & Canada (Darkest Blue) - Europe (Medium Blue-Grey) - Asia-Pacific (Lighter Blue-Grey) - Rest of World (Lightest Grey) Data Points (Total Revenue per Quarter): - Q1'24: $36,455 - Q2'24: $39,071 - Q3'24: $40,589 - Q4'24: $48,385 - Q1'25: $42,314 - Q2'25: $47,516 - Q3'25: $51,242 - Q4'25: $59,893 - Q1'26: $56,311 Detailed Breakdown by Category (Bottom to Top): Q1'24:" + }, + { + "path": "images/image-3-Quarterly Percentage Breakdown.jpg", + "summary": "image-3 Q1'24: 18%, 27%, 7%, 9% Q2'24: 19%, 27%, 7%, 9% Q3'24: 18%, 28%, 7%, 5% Q4'24: 18%, 25%, 7%, 2% Q1'25: 18%, 29%, 7%, 5% Q2'25: 18%, 27%, 6%, 6% Q3'25: 18%, 30%, 6%, 7% Q4'25: 18%, 29%, 6%, 6% Q1'26: 18%, 31%, 5%, 5%" + }, + { + "path": "images/image-4-Quarterly Financial Performance.jpg", + "summary": "image-4 - Q1'24: $12,369 - Q2'24: $13,465 - Q3'24: $15,688 - Q4'24: $20,838 - Q1'25: $16,644 - Q2'25: $18,337 - Q3'25 (1): $2,709 - Q4'25: $22,768 - Q1'26 (2): $26,773" + }, + { + "path": "images/image-5-Quarterly Financial Growth.jpg", + "summary": "image-5 Chart Title: Quarterly Performance (Q1'24 - Q1'26) Data Points by Quarter: - Q1'24: $4.71 - Q2'24: $5.16 - Q3'24: $6.03 - Q4'24: $8.02 - Q1'25: $6.43 - Q2'25: $7.14 - Q3'25: $1.05 (marked with footnote 1) - Q4'25: $8.88 - Q1'26: $10.44 (marked with footnote 2) Key Trends: - Steady growth from Q1'24 to Q4'24 ($4.71 to $8.02). - A dip in Q1'25 followed by recovery in Q2'25. - Significant drop in Q3'25 ($1.05), the lowest value on the chart. - Strong" + }, + { + "path": "images/image-6-Financial Growth Projection.jpg", + "summary": "image-6 Quarterly Data: - Q1'25: $13,692 - Q1'26: $19,840 Annual Data: - 2024: $39,225 - 2025: $72,215" + }, + { + "path": "images/image-7-Quarterly Data Trend.jpg", + "summary": "image-7 Q1'24: 3.24 Q2'24: 3.27 Q3'24: 3.29 Q4'24: 3.35 Q1'25: 3.43 Q2'25: 3.48 Q3'25: 3.54 Q4'25: 3.58 Q1'26: 3.56" + }, + { + "path": "images/image-8-Quarterly Financial Data.jpg", + "summary": "image-8 Q1'24: $11.20 Q2'24: $11.89 Q3'24: $12.29 Q4'24: $14.25 Q1'25: $12.36 Q2'25: $13.65 Q3'25: $14.46 Q4'25: $16.56 Q1'26: $15.66" + }, + { + "path": "images/image-9-Quarterly Growth Trend.jpg", + "summary": "image-9 Q1'25: 5% Q2'25: 11% Q3'25: 14% Q4'25: 18% Q1'26: 19%" + }, + { + "path": "images/image-10-Quarterly Percentage Growth.jpg", + "summary": "image-10 Q1'25: 4% Q2'25: 9% Q3'25: 8% Q4'25: 13% Q1'26: 13%" + }, + { + "path": "images/image-11-Quarterly Growth Trend.jpg", + "summary": "image-11 Chart Title: Quarterly Growth Trend (implied) X-Axis Labels: Q1'25, Q2'25, Q3'25, Q4'25, Q1'26 Data Points: - Q1'25: 5% - Q2'25: 6% - Q3'25: 9% - Q4'25: 13% - Q1'26: 17% Trend: Consistent upward growth over the five quarters." + }, + { + "path": "images/image-12-Quarterly Growth Trend.jpg", + "summary": "image-12 - Q1'25: 9% - Q2'25: 16% - Q3'25: 23% - Q4'25: 24% - Q1'26: 23%" + }, + { + "path": "images/image-13-Growth Trend 2025-2026.jpg", + "summary": "image-13 Chart Type: Vertical Bar Chart X-Axis Categories (Time Periods): Q1'25, Q2'25, Q3'25, Q4'25, Q1'26 Data Values: - Q1'25: 1% - Q2'25: 7% - Q3'25: 9% - Q4'25: 14% - Q1'26: 17% Trend: Consistent upward growth over the five quarters." + }, + { + "path": "images/image-14-Quarterly Percentage Data (2025-2026).jpg", + "summary": "image-14 Chart Title: Not visible Data Points by Quarter: - Q1'25: 10% - Q2'25: 9% - Q3'25: 10% - Q4'25: 6% - Q1'26: 12%" + }, + { + "path": "images/image-15-Quarterly Percentage Data (2025-2026).jpg", + "summary": "image-15 Chart Title: Not visible Data Points by Quarter: - Q1'25: 14% - Q2'25: 11% - Q3'25: 13% - Q4'25: 9% - Q1'26: 14% Key Trends: - Lowest value recorded in Q4'25 at 9%. - Highest values observed in Q1'25 and Q1'26 at 14%." + }, + { + "path": "images/image-16-Quarterly Percentage Trends.jpg", + "summary": "image-16 Chart Title: Not visible X-Axis Labels (Time Periods): Q1'25, Q2'25, Q3'25, Q4'25, Q1'26 Data Points: - Q1'25: 9% - Q2'25: 17% - Q3'25: 19% - Q4'25: 12% - Q1'26: 19%" + }, + { + "path": "images/image-17-Quarterly Growth Trend.jpg", + "summary": "image-17 Q1'25: 3% Q2'25: 2% Q3'25: 1% Q4'25: (2)% Q1'26: 5%" + }, + { + "path": "images/image-18-Quarterly Percentage Data (2025-2026).jpg", + "summary": "image-18 Chart Title: Not visible X-Axis Labels: Q1'25, Q2'25, Q3'25, Q4'25, Q1'26 Data Points: - Q1'25: 17% - Q2'25: 15% - Q3'25: 20% - Q4'25: 15% - Q1'26: 18%" + } + ], + "tables": [ + { + "path": "tables/table-0 Revenue by Region.html", + "summary": "table-1 The table shows quarterly revenue data for US & Canada, Europe, Asia-Pacific, and Rest of World from Q1'24 to Q1'26, with totals." + }, + { + "path": "tables/table-1 Quarterly Revenue.html", + "summary": "table-2 The table shows quarterly revenue from Q1'24 to Q1'26 across US & Canada, Europe, Asia-Pacific, and Rest of World, with totals increasing over time." + }, + { + "path": "tables/table-2 Meta Revenue Data.html", + "summary": "table-3 The table shows Meta's quarterly revenue from Q1'24 to Q1'26, including Advertising, Other, Family of Apps, Reality Labs, and total revenue, along with operating income and margins." + }, + { + "path": "tables/table-3 Segment Margins.html", + "summary": "table-4 The table shows quarterly margins for bottom and top segments from Q1'24 to Q1'26. Bottom segment stays at 18-19%, while top segment fluctuates between 25-31%." + }, + { + "path": "tables/table-4 Quarterly Tax Data.html", + "summary": "table-5 The table shows quarterly income before taxes, provision for income taxes, and effective tax rates from Q1'24 to Q1'26, with notable spikes in Q3'25 and negative rate in Q1'26." + }, + { + "path": "tables/table-5 Quarterly Values.html", + "summary": "table-6 The table shows quarterly values from Q1'24 to Q1'26, with figures ranging from $2,709 to $26,773, peaking in Q1'26." + }, + { + "path": "tables/table-6 Quarterly Values.html", + "summary": "table-7 The table shows quarterly values from Q1'24 to Q1'26, with figures ranging from $1.05 to $10.44, indicating an overall upward trend." + }, + { + "path": "tables/table-7 Financial Data.html", + "summary": "table-8 The table shows financial values across different periods: Quarterly $13,692, Q1'26 $19,840, 2024 $39,225, and Annual $72,215." + }, + { + "path": "tables/table-8 Quarterly Values.html", + "summary": "table-9 The table shows quarterly values from Q1'24 to Q1'26, increasing from 3.24 to 3.58, with a slight dip to 3.56 in Q1'26." + }, + { + "path": "tables/table-9 Quarterly Values.html", + "summary": "table-10 The table shows quarterly values from Q1'24 to Q1'26, starting at $11.20 and ending at $15.66, with a peak of $16.56 in Q4'25." + }, + { + "path": "tables/table-10 Quarterly Values.html", + "summary": "table-11 The table shows quarterly percentage values from Q1'25 (5%) to Q1'26 (19%), with a steady increase over five quarters." + }, + { + "path": "tables/table-11 Quarterly Values.html", + "summary": "table-12 The table shows quarterly percentage values from Q1'25 to Q1'26, starting at 4%, rising to 9%, 8%, 13%, and ending at 13%." + }, + { + "path": "tables/table-12 Quarterly Values.html", + "summary": "table-13 The table shows quarterly percentage values from Q1'25 (5%) to Q1'26 (17%), with a steady increase over five quarters." + }, + { + "path": "tables/table-13 Quarterly Values.html", + "summary": "table-14 The table shows quarterly percentage values from Q1'25 to Q1'26, starting at 9% and rising to a peak of 24% in Q4'25, then slightly decreasing to 23% in Q1'26." + }, + { + "path": "tables/table-14 Quarterly Values.html", + "summary": "table-15 The table shows quarterly percentage values from Q1'25 (1%) to Q1'26 (17%), with a steady increase over five quarters." + }, + { + "path": "tables/table-15 Quarterly Values.html", + "summary": "table-16 The table shows quarterly percentage values from Q1'25 to Q1'26, ranging from 6% to 12%, with Q4'25 having the lowest value." + }, + { + "path": "tables/table-16 Quarterly Values.html", + "summary": "table-17 The table shows quarterly percentage values from Q1'25 to Q1'26, with values ranging from 9% to 14%." + }, + { + "path": "tables/table-17 Quarterly Values.html", + "summary": "table-18 The table shows quarterly percentage values from Q1'25 to Q1'26, ranging from 9% to 19%, with the highest in Q3'25 and Q1'26." + }, + { + "path": "tables/table-18 Quarterly Values.html", + "summary": "table-19 The table shows quarterly percentage values from Q1'25 to Q1'26, starting at 3%, decreasing to -2% in Q4'25, then rising to 5%." + }, + { + "path": "tables/table-19 Quarterly Values.html", + "summary": "table-20 The table shows quarterly percentage values from Q1'25 to Q1'26, ranging from 15% to 20%, with Q3'25 having the highest value at 20%." + }, + { + "path": "tables/table-20 Free Cash Flow.html", + "summary": "table-21 The table shows net cash from operations, property purchases, finance lease payments, and free cash flow from Q1 2024 to Q1 2026, with free cash flow ranging from $8.5B to $15.5B." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/full.md b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/full.md new file mode 100644 index 000000000..014fa9202 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/full.md @@ -0,0 +1,420 @@ +## Meta Earnings Presentation Q1 2026 + +## Advertising Revenue by User Geography + +In Millions + +![](images/04c3b09236cd08fe8aba1f3603a61bc37f033000642793becf32c559e0f4a1e7.jpg) + +
+stacked bar chart + +| Quarter | US & Canada | Europe | Asia-Pacific | Rest of World | +| --- | --- | --- | --- | --- | +| Q1'24 | $15,451 | $8,327 | $7,338 | $4,519 | +| Q2'24 | $16,593 | $9,135 | $7,721 | $4,880 | +| Q3'24 | $17,389 | $9,358 | $8,050 | $5,088 | +| Q4'24 | $20,982 | $11,154 | $9,012 | $5,635 | +| Q1'25 | $18,259 | $9,527 | $8,224 | $5,382 | +| Q2'25 | $20,045 | $11,366 | $9,148 | $6,004 | +| Q3'25 | $21,331 | $12,072 | $10,020 | $6,659 | +| Q4'25 | $25,643 | $14,198 | $10,893 | $7,403 | +| Q1'26 | $23,667 | $13,296 | $10,631 | $7,430 | +| Total | $35,635 | — | — | — | +| Q1'24 | — | — | — | — | +| Q2'24 | — | — | — | — | +| Q3'24 | — | — | — | — | +| Q4'24 | — | — | — | — | +| Q1'25 | — | — | — | — | +| Q2'25 | — | — | — | — | +| Q3'25 | — | — | — | — | +| Q4'25 | — | — | — | — | +| Q1'26 | — | — | — | — | +| Total | — | — | — | — | +| Total | — | — | — | — | +| Total | — | — | — | — | +| Total | — | — | — | — | +| Total | — | — | — | — | +| Total | — | — | — | — | +| Total | — (additional values) | (continued) | (continued) | (continued) | +| Total | 58,137 | 14,198 | 10,893 | 7,403 | +| Total | 55,024 | 13,296 | 10,631 | 7,430 | +
+ +Our revenue by user geography is geographically apportioned based on our estimation of the geographic location of our users when they perform a revenue-generating activity. This allocation differs from our revenue disaggregated by geography disclosure in our condensed consolidated financial statements where revenue is geographically apportioned based on the addresses of our customers. + +## Revenue by User Geography + +In Millions + +![](images/cb8eea1da323c9d810e87825f4e0b32b48b4b6c55b74851f35a95fe436e4dcc5.jpg) + +
+stacked bar chart + +| Quarter | US & Canada | Europe | Asia-Pacific | Rest of World | +|---------|-------------|--------|--------------|---------------| +| Q1'24 | $15,824 | $8,483 | $7,481 | $4,667 | +| Q2'24 | $16,847 | $9,300 | $7,888 | $5,036 | +| Q3'24 | $17,609 | $9,492 | $8,220 | $5,268 | +| Q4'24 | $21,783 | $11,503| $9,245 | $5,854 | +| Q1'25 | $18,605 | $9,680 | $8,439 | $5,590 | +| Q2'25 | $20,371 | $11,532| $9,366 | $6,247 | +| Q3'25 | $21,751 | $12,268| $10,272 | $6,951 | +| Q4'25 | $26,472 | $14,480| $11,182 | $7,759 | +| Q1'26 | $24,079 | $13,500| $10,914 | $7,818 | +
+ +Our revenue by user geography is geographically apportioned based on our estimation of the geographic location of our users when they perform a revenue-generating activity. This allocation differs from our revenue disaggregated by geography disclosure in our condensed consolidated financial statements where revenue is geographically apportioned based on the addresses of our customers. + +## Segment Results + +In Millions, Except Percentages + +
Q1'24Q2'24Q3'24Q4'24Q1'25Q2'25Q3'25Q4'25Q1'26
Advertising$ 35,635$ 38,329$ 39,885$ 46,783$ 41,392$ 46,563$ 50,082$ 58,137$ 55,024
Other380389434519510583690801885
Family of Apps Revenue36,01538,71840,31947,30241,90247,14650,77258,93855,909
Reality Labs Revenue4403532701,083412370470955402
Total Revenue$ 36,455$ 39,071$ 40,589$ 48,385$ 42,314$ 47,516$ 51,242$ 59,893$ 56,311
Family of Apps Operating Income$ 17,664$ 19,335$ 21,778$ 28,332$ 21,765$ 24,971$ 24,967$ 30,766$ 26,900
Reality Labs Operating (Loss)(3,846)(4,488)(4,428)(4,967)(4,210)(4,530)(4,432)(6,021)(4,028)
Total Income from Operations$ 13,818$ 14,847$ 17,350$ 23,365$ 17,555$ 20,441$ 20,535$ 24,745$ 22,872
Operating Margin38 %38 %43 %48 %41 %43 %40 %41 %41 %
+ +We report our financial results based on two reportable segments: Family of Apps (FoA) and Reality Labs (RL). FoA includes Facebook, Instagram, Messenger, WhatsApp, and other services. RL includes our virtual and augmented reality related consumer hardware, software, and content. + +## Expenses as a Percentage of Revenue + +![](images/66dd2af450faf8e82755c0df14e5ec835c56c7357551e5f3f8333135e237a2ee.jpg) + +
+stacked bar chart + +| Quarter | Bottom Segment (%) | Top Segment (%) | +| :--- | :--- | :--- | +| Q1'24 | 18 | 27 | +| Q2'24 | 19 | 27 | +| Q3'24 | 18 | 28 | +| Q4'24 | 18 | 25 | +| Q1'25 | 18 | 29 | +| Q2'25 | 18 | 27 | +| Q3'25 | 18 | 30 | +| Q4'25 | 18 | 29 | +| Q1'26 | 18 | 31 | +
+ +General & Administrative +Marketing & Sales +Research & Development +Cost of Revenue + +## Effective Tax Rate + +In Millions, Except Percentages + +
Q1'24Q2'24Q3'24Q4'24Q1'25Q2'25 $Q3'25^{(1)}$ Q4'25 $Q1'26^{(2)}$
Income before income taxes$ 14,183$ 15,106$ 17,822$ 23,553$ 18,382$ 20,534$ 21,663$ 25,354$ 21,752
Provision (benefit) for income taxes$ 1,814$ 1,641$ 2,134$ 2,715$ 1,738$ 2,197$ 18,954$ 2,586$ (5,021)
Effective Tax Rate13 %11 %12 %12 %9 %11 %87 %10 %(23)%
+ +(1) Provision (benefit) for income taxes includes a one-time, non-cash income tax charge of \$15.93 billion accrued in the third quarter of 2025, related to the implementation of the One Big Beautiful Bill Act. Excluding this one-time tax charge, our third quarter 2025 effective tax rate would have decreased by 73 percentage points to 14%, compared to the reported effective tax rate of 87%. +(2) Provision (benefit) for income taxes includes an income tax benefit of \$8.03 billion in the first quarter of 2026 related to the U.S. Corporate Alternative Minimum Tax transitional relief under Treasury Notice 2026-7. Excluding this tax benefit, the effective tax rate would have been 14%. + +![](images/30b2b78bad460d4d556a833ef9335a1a542f201db01e14c35d8a4dfee7dc30f3.jpg) + +
+bar chart + +| Quarter | Value | +| ----------- | -------- | +| Q1'24 | $12,369 | +| Q2'24 | $13,465 | +| Q3'24 | $15,688 | +| Q4'24 | $20,838 | +| Q1'25 | $16,644 | +| Q2'25 | $18,337 | +| Q3'25 (1) | $2,709 | +| Q4'25 | $22,768 | +| Q1'26 (2) | $26,773 | +
+ +(1) Includes a one-time, non-cash income tax charge to provision for income taxes of \$15.93 billion accrued in the third quarter of 2025, related to the implementation of the One Big Beautiful Bill Act. Excluding this one-time tax charge, our third quarter 2025 net income would have increased by \$15.93 billion to \$18.64 billion, compared to the reported net income of \$2.71 billion. +(2) Includes an income tax benefit of \$8.03 billion in the first quarter of 2026 related to the U.S. Corporate Alternative Minimum Tax transitional relief under Treasury Notice 2026-7. Excluding this tax benefit, net income would have been \$18.74 billion, lower by \$8.03 billion, compared to the reported net income of \$26.77 billion. + +## Diluted Earnings Per Share + +![](images/18b2087e7634065c98a94b1db07c5ffc81637e79c2778e767a443fe647caab09.jpg) + +
+bar chart + +| Quarter | Value | +| ----------- | ------- | +| Q1'24 | $4.71 | +| Q2'24 | $5.16 | +| Q3'24 | $6.03 | +| Q4'24 | $8.02 | +| Q1'25 | $6.43 | +| Q2'25 | $7.14 | +| Q3'25 (1) | $1.05 | +| Q4'25 | $8.88 | +| Q1'26 (2) | $10.44 | +
+ +(1) Net income includes a one-time, non-cash income tax charge to provision for income taxes of \$15.93 billion accrued in the third quarter of 2025, related to the implementation of the One Big Beautiful Bill Act. Excluding this one-time tax charge, our third quarter 2025 diluted earnings per share would have increased by \$6.20 to \$7.25 compared to the reported diluted EPS of \$1.05. +(2) Net income includes an income tax benefit of \$8.03 billion in the first quarter of 2026 related to the U.S. Corporate Alternative Minimum Tax transitional relief under Treasury Notice 2026-7. Excluding this tax benefit, diluted EPS would have been \$7.31, lower by \$3.13, compared to the reported diluted EPS of \$10.44. + +## Capital Expenditures + +In Millions + +![](images/53d26c7b71ef63ad991c159a42e21b61d95abeeca3a75490684c36173f3f3020.jpg) + +
+bar chart + +| Period | Value | +| -------- | -------- | +| Quarterly | $13,692 | +| Q1'26 | $19,840 | +| 2024 | $39,225 | +| Annual | $72,215 | +
+ +Capital expenditures for periods presented were related to purchases of property and equipment and principal payments on finance leases. + +## Family Daily Active People (DAP) + +In Billions +![](images/6b5fc1dea6e1fddf8c44f6f5eb426a2eaeb92c83102c3be20118b88d97b92ea2.jpg) + +
+bar chart + +| Quarter | Value | +|---|---| +| Q1'24 | 3.24 | +| Q2'24 | 3.27 | +| Q3'24 | 3.29 | +| Q4'24 | 3.35 | +| Q1'25 | 3.43 | +| Q2'25 | 3.48 | +| Q3'25 | 3.54 | +| Q4'25 | 3.58 | +| Q1'26 (1) | 3.56 | +
+ +(1) The slight quarter-over-quarter decline in DAP in the first quarter of 2026 was driven by internet disruptions in Iran, as well as a restriction on access to WhatsApp in Russia. + +We define a daily active person (DAP) as a registered and logged-in user of Facebook, Instagram, Messenger, and/or WhatsApp (collectively, our "Family" of products) who visited at least one of these Family products through a mobile device application or using a web or mobile browser on a given day. The numbers for DAP do not include users on our other products unless they would otherwise qualify as DAP based on their other activities on our Family products. + +We do not require people to use a common identifier or link their accounts to use multiple products in our Family, and therefore must seek to attribute multiple user accounts within and across products to individual people. Our calculations of DAP rely upon complex techniques, algorithms, and machine learning models that seek to estimate the underlying number of unique people using one or more of these products, including by matching user accounts within an individual product and across multiple products when we believe they are attributable to a single person, and counting such group of accounts as one person. As these techniques and models require significant judgment, are developed based on internal reviews of limited samples of user accounts, and are calibrated against user survey data, there is necessarily some margin of error in our estimates. For additional information, see "Limitations of Key Metrics and Other Data" located in the Appendix of this presentation. + +## Family Average Revenue per Person (ARPP) + +![](images/69ee3e417cc0aafa526249ef7516c0534d071580683739d371d661a469174c58.jpg) + +
+bar chart + +| Quarter | Value ($) | +|---|---| +| Q1'24 | 11.20 | +| Q2'24 | 11.89 | +| Q3'24 | 12.29 | +| Q4'24 | 14.25 | +| Q1'25 | 12.36 | +| Q2'25 | 13.65 | +| Q3'25 | 14.46 | +| Q4'25 | 16.56 | +| Q1'26 | 15.66 | +
+ +We define average revenue per person (ARPP) as our FoA revenue during a given quarter, divided by the average of the number of DAP at the beginning and end of the quarter. + +## Ad Impressions Delivered YoY Percentage Change + +![](images/1ceed1b0c01baaee83a4726a6b4eefef5bf667a54d447b89fc2e3b86d09dd0f3.jpg) + +
+bar chart + +| Quarter | Value (%) | +|---|---| +| Q1'25 | 5 | +| Q2'25 | 11 | +| Q3'25 | 14 | +| Q4'25 | 18 | +| Q1'26 | 19 | +
+ +Worldwide + +![](images/3d7c8b8150273b0424346c0236701a2d3ca46a8bd6c280f7aa0fe23263f2c4d8.jpg) + +
+bar chart + +| Quarter | Value (%) | +| :--- | :--- | +| Q1'25 | 4 | +| Q2'25 | 9 | +| Q3'25 | 8 | +| Q4'25 | 13 | +| Q1'26 | 13 | +
+ +US & Canada + +![](images/7730ba4dddc9d16def970b311b185cdeeb820ee8b74a07990578374671844dde.jpg) + +
+bar chart + +| Quarter | Value (%) | +|---|---| +| Q1'25 | 5 | +| Q2'25 | 6 | +| Q3'25 | 9 | +| Q4'25 | 13 | +| Q1'26 | 17 | +
+ +Europe + +![](images/4ae2d5c21673df9bbb7738fd43738928621492d56e05971e00d937f91392e274.jpg) + +
+bar chart + +| Quarter | Value (%) | +| :--- | :--- | +| Q1'25 | 9 | +| Q2'25 | 16 | +| Q3'25 | 23 | +| Q4'25 | 24 | +| Q1'26 | 23 | +
+ +Asia-Pacific + +![](images/54c657a8a01a5bcd888ddc327efa65c732ffb37249d479f006499d74d4c16b12.jpg) + +
+bar chart + +| Quarter | Value (%) | +|---|---| +| Q1'25 | 1 | +| Q2'25 | 7 | +| Q3'25 | 9 | +| Q4'25 | 14 | +| Q1'26 | 17 | +
+ +Rest of World + +## Average Price Per Ad YoY Percentage Change + +![](images/8aeed655b1ccdc28575f62ec6fd1c87a88a48e532b69a9ef58af983a8a26c7b1.jpg) + +
+bar chart + +| Quarter | Value (%) | +|---|---| +| Q1'25 | 10 | +| Q2'25 | 9 | +| Q3'25 | 10 | +| Q4'25 | 6 | +| Q1'26 | 12 | +
+ +Worldwide + +![](images/6fff140b362a6072e614ad803fe465b32a6041d0cdab0b97524a7f1941afae7d.jpg) + +
+bar chart + +| Quarter | Value (%) | +| :--- | :--- | +| Q1'25 | 14 | +| Q2'25 | 11 | +| Q3'25 | 13 | +| Q4'25 | 9 | +| Q1'26 | 14 | +
+ +US & Canada + +![](images/9a575f96f4f858dfda4ce54927edb636f3de8b48b9be76685326bcc5dcf6a2ae.jpg) + +
+bar chart + +| Quarter | Value (%) | +|---|---| +| Q1'25 | 9 | +| Q2'25 | 17 | +| Q3'25 | 19 | +| Q4'25 | 12 | +| Q1'26 | 19 | +
+ +Europe + +![](images/017fb79beb891fa5395d300f1da2e699cac968cd7f10eee696b99b2ba07a9653.jpg) + +
+bar chart + +| Quarter | Value (%) | +| :--- | :--- | +| Q1'25 | 3 | +| Q2'25 | 2 | +| Q3'25 | 1 | +| Q4'25 | -2 | +| Q1'26 | 5 | +
+ +Asia-Pacific + +![](images/3a107545a760a2591da8338f448f9ee878e6c20ab4c220fa0199bd1b8131c9b4.jpg) + +
+bar chart + +| Quarter | Value (%) | +| :--- | :--- | +| Q1'25 | 17 | +| Q2'25 | 15 | +| Q3'25 | 20 | +| Q4'25 | 15 | +| Q1'26 | 18 | +
+ +Rest of World + +## Appendix + +## Free Cash Flow Reconciliation + +In Millions + +
Q1'24Q2'24Q3'24Q4'24Q1'25Q2'25Q3'25Q4'25Q1'26
Net cash provided by operating activities$ 19,246$ 19,370$ 24,724$ 27,988$ 24,026$ 25,561$ 29,999$ 36,214$ 32,226
Less: Purchases of property and equipment6,4008,1738,25814,42512,94116,53818,82921,38318,997
Less: Principal payments on finance leases315299944411751474545754843
Free Cash Flow$ 12,531$ 10,898$ 15,522$ 13,152$ 10,334$ 8,549$ 10,625$ 14,077$ 12,386
+ +Free cash flow (FCF) is a non-GAAP financial measure that has limitations as an analytical tool, and you should not consider it in isolation or as a substitute for analysis of other GAAP financial measures, such as net cash provided by operating activities. FCF is not intended to represent our residual cash flow available for discretionary expenses. Some of the limitations of FCF are: (i) FCF does not reflect our future contractual commitments, and (ii) other companies in our industry present similarly titled measures differently than we do, limiting their usefulness as comparative measures. + +## Limitations of Key Metrics and Other Data + +## Family Metrics + +The numbers for our key metrics are calculated using internal company data based on the activity of user accounts. We report our estimates of the numbers of our daily active people (DAP) and average revenue per person (ARPP) (collectively, our "Family metrics") based on the activity of users who visited at least one of Facebook, Instagram, Messenger, and WhatsApp (collectively, our "Family" of products) during the applicable period of measurement. + +While these numbers are based on what we believe to be reasonable estimates of our user base for the applicable period of measurement, there are inherent challenges in measuring usage of our products across large online and mobile populations around the world. The methodologies used to measure these metrics require significant judgment and are also susceptible to algorithm or other technical errors. In addition, we are continually seeking to improve our estimates of our user base, and such estimates may change due to improvements or changes in our methodology. We regularly review our processes for calculating these metrics, and from time to time we discover inaccuracies in our metrics or make adjustments to improve their accuracy, which can result in adjustments to our historical metrics. Our ability to recalculate our historical metrics may be impacted by data limitations or other factors that require us to apply different methodologies for such adjustments. We generally do not intend to update previously disclosed Family metrics for any such inaccuracies or adjustments that are within the error margins disclosed below. + +In addition, our Family metrics estimates will differ from estimates published by third parties due to differences in methodology or other factors such as data limitations or other challenges in measuring large online and mobile populations. For example, our methodologies include measurements of our user base that have in some instances exceeded estimates of addressable online and mobile populations that are based on data published by third parties. + +Many people in our community have user accounts on more than one of our products, and some people have multiple user accounts within an individual product. Accordingly, for our Family metrics, we do not seek to count the total number of user accounts across our products because we believe that would not reflect the actual size of our community. Rather, our Family metrics represent our estimates of the number of unique people using at least one of Facebook, Instagram, Messenger, and WhatsApp. We do not require people to use a common identifier or link their accounts to use multiple products in our Family, and therefore must seek to attribute multiple user accounts within and across products to individual people. To calculate these metrics, we rely upon complex techniques, algorithms and machine learning models that seek to count the individual people behind user accounts, including by matching multiple user accounts within an individual product and across multiple products when we believe they are attributable to a single person, and counting such group of accounts as one person. These techniques and models require significant judgment, are subject to data and other limitations discussed below, and inherently are subject to statistical variances and uncertainties. We estimate the potential error in our Family metrics primarily based on user survey data as described further below, which itself is subject to error as well. While we expect the error margin for our Family metrics to vary from period to period, we estimate that such margin generally will be approximately 3% of our worldwide DAP. At our scale, it is very difficult to attribute multiple user accounts within and across products to individual people, and it is possible that the actual numbers of unique people using our products may vary significantly from our estimates, potentially beyond our estimated error margins. As a result, it is also possible that our Family metrics may indicate changes or trends in user numbers that do not match actual changes or trends. + +## Limitations of Key Metrics and Other Data + +To calculate our estimates of DAP, we currently use a series of machine learning models that are developed based on internal reviews of limited samples of user accounts and calibrated against user survey data. We apply significant judgment in designing these models and calculating these estimates. For example, to match user accounts within individual products and across multiple products, we use data signals such as similar device information, IP addresses, and user names. We also calibrate our models against data from periodic user surveys of varying sizes and frequency across our products, which survey questions are based on monthly usage, and which are inherently subject to error. The timing and results of such user surveys have in the past contributed, and may in the future contribute, to changes in our reported Family metrics from period to period. In addition, our data limitations may affect our understanding of certain details of our business and increase the risk of error for our Family metrics estimates. Our techniques and models rely on a variety of data signals from different products, and we rely on more limited data signals for some products compared to others. For example, as a result of limited visibility into encrypted products, we have fewer data signals from WhatsApp user accounts and primarily rely on phone numbers and device information to match WhatsApp user accounts with accounts on our other products. Any loss of access to data signals we use in our process for calculating Family metrics, whether as a result of our own product decisions, actions by third-party browser or mobile platforms, regulatory or legislative requirements, or other factors, also may impact the stability or accuracy of our reported Family metrics, as well as our ability to report these metrics at all. Our estimates of Family metrics also may change as our methodologies evolve, including through the application of new data signals or technologies, product changes, or other improvements in our user surveys, algorithms, or machine learning that may improve our ability to match accounts within and across our products or otherwise evaluate the broad population of our users. In addition, such evolution may allow us to identify previously undetected violating accounts (as defined below). + +We regularly evaluate our Family metrics to estimate the percentage of our DAP consisting solely of "violating" accounts. We define "violating" accounts as accounts which we believe are intended to be used for purposes that violate our terms of service, including bots and spam. Our violating accounts estimation is based on an internal review of a limited sample of accounts, and we apply significant judgment in making this determination. For example, we look for account information and behaviors associated with Facebook and Instagram accounts that appear to be inauthentic to the reviewers, but we have less visibility into WhatsApp user activity due to encryption. In addition, if we believe an individual person has one or more violating accounts, we do not include such person in our violating accounts estimation as long as we believe they have one active account that does not constitute a violating account. In the fourth quarter of 2025, we made certain updates to the methodology we use for this estimation, including to incorporate updated data signals as a result of improvements in our ability to identify activity we believe to be violating our policies, as well as to focus on the most recent account activity when determining whether to include a person in our violating accounts estimation. Accordingly, in the fourth quarter of 2025, we estimated that less than $5\%$ of our worldwide DAP consisted solely of violating accounts. We believe the increase compared to our prior estimation was a result of the methodology update described above. From time to time, we disable certain user accounts, make product changes, or take other actions to reduce the number of violating accounts among our users, which may also reduce our DAP estimates in a particular period. We intend to continue to disclose our estimates of the percentage of our DAP consisting solely of violating accounts on an annual basis. Violating accounts are very difficult to measure at our scale, and it is possible that the actual number of violating accounts may vary significantly from our estimates. + +## User Geography + +Our estimates for revenue by user location, as well as year-over-year percentage changes in ad impressions delivered and the average price per ad by user location, are also affected by data limitations and other challenges in measuring user geography. Our data regarding the geographic location of our users is estimated based on a number of factors, such as the user's IP address and self-disclosed location. These factors may not always accurately reflect the user's actual location. For example, a user may appear to be accessing our products from the location of the proxy server that the user connects to rather than from the user's actual location. The methodologies used to measure our metrics are also susceptible to algorithm or other technical errors. + +## Meta Earnings Presentation Q1 2026 \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-1-Revenue Forecast by Region.jpg b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-1-Revenue Forecast by Region.jpg new file mode 100644 index 0000000000000000000000000000000000000000..20fc74adf40353f1189a0626d171d7b96763f749 GIT binary patch literal 157397 zcmd?QXIN8Pw=jymq9P*F_YO@$+aLjHqC)7!kc3c!NDl-g5USvotw<4&5|FL|0wg4n zl0bmK76k+Z1QL2cO6Wz9A}D%!-@V^+&Ue4>$1Tr!?m73Hc{10UYmGVPSaYp0#vEn- z9Qyf*=YpxBi6PISLp(f(4t_j8M|t#ljvYCA^yrae2i37-$Bv&kd-B9VIe+HNsk40N zFJ9z3&&S7qSyYh!lCS_D-<4ml2#bkHNJ#Jt0;B-qQljD#;{QN$=%DL~<0sCYJb6z1 z65l29|J&u~pFDggkNBLrd*l#+=P=))BYcN`zT*))AoB2$LkETDU*-6TL&r`YJ$&TU z!EN^oJco`PIdt^M@$p@mJMbkfZznmw-o? zuHDvu^>c*h>^~s*j_~p5@hrdokGTGupC1vj?4Jt5KT1mVq^qlFl+{U-W&*gfJ}<@1oI0WxjKtjOm8Uzr z`4zHH@hyj>Kp#ytEhcK0aQOXeIwd#pEw6?rKC+JzbH|H=)2cQdm1zC%%h(IgoV~ROQ#L0{IP#tol#!6{ zhRW5cci8<1YcRq{2YcxYNJ6xNAv#D40|EnYtC=ZVjqzOkZ~m?Qr;$H*ZpCcBBO|-L zWMTG!8YnR%JdX`Cfyc)VP*E~)xK7qD?0U8SOQW|xYK7<1@pbXLhN%NEBh`>KN3LyY ze4@eR+qk}Ou;KkQFLe(GO1qlsZEhsU@w!nnckbC|7_u6QqxpTns>hC4B)xHA?g1C< z2QB>aWfipx9f__$QXT2ZOs%$y`?IkZ6N*bdJ+3)2jlMs5934@2SxK3e?c#u9o1K0} z_>9#tivOIfU})`grR9Q6rhhSc*ikEMjKAO^P-9ml@+sQEhR|SfglfNl8k^I<)K?** zV@ zTb5e3-IBQ3Z@eAyo>BM)g?xjJArQ;Zxslq0?z*EUdH)*3aCP01)r`kZLIP%#K?^+-^<)Z?E7H=vR$0naO*KhPqL`w>#7UV3v1|i4G87K` z`r5%7NY~KR#0xwOL=x#Ji+r%vE#DGayHDE3{=R-I)@4D1)ii=4O(wUZ(h$7{V))pUlmmyV73IkBukpeG%GKq+VUmh$QEb}aJ z#wG2VHF;kx2wc&B`;%uHc!O-JU;A*~rer7lp-P*@+UdFJ@~Q3-!}ryMh6eqRMiY$N z=l&wzP2Jln^`*oH9H}kt!?&A#T85D>#}kL2D2{vyG7+E&oG&v0pi0zhcr8!uTEUxL zMJ>CSkZ^Yoa?NJ<(_Xa={XUHxjqI}Yb%LcoJFh#re1Y$KMEB2fZ-Zc*QyyoBKT9tjToU^~Udqx`zJK#M6Io#D^?*47@poj&8K^ zYTa(EnQI;5&dnLu0rSr|)!MYDye(fDWnh%md!?}*LAbeIswcw{z@WqLmrOE-hgSj=H|62ysjY@ zZA-tRW#C5&>A3Q%627kPSbfQi8*%-{WE+}r%hUi(F#C(tw~(3@B!{@NS7SH;8&ll$ zZ7cqGOdbLNNXSLVuGih892^#5KMpac&zsS#_Gqd)q9S3;H#`kO4Zb=aWqKX zDfj4%5GG%!(vpDoxI|v=W#z!j-F{k zy_C}?Ei<=dePxnV?^s9A?7T!^6jspg8+Ha zJnsqF8#-sY8Mt%!y!2GM97iNomYMT@a5|cZ#VcJGXnhwP=ukLQ=ZqD2lc@*}NsA(e zeyJT;1Fqt!Iz%qw6f5q1n3+~lFX zz@oew2e;@RwGyB9`ZYQ1SMcZFURfc&v7NN|2L{@Zl?3k*LKmb&aZfy4iws_5YL zG@bi2gVOwMi~yDIl!cK(zra)z+w3#4x`N2fKiDS%WxADwMJxwwoP~#0Ycy1x4{Jxf zOe`CxA;=v>ViVt$)v7YKAOk(3XEo$6kX~8_L4b5clj1g%k6()vNfL#JZk4!v>_P!3 zS|5P!VP)dTZAXpdXNKx_#V)PnEC|H21O_KaZz)yki6gO;O{(A z|Gk;}KZ|-jAmcL7TiRb80*?i#GkS|8j+{@?3H0~@xf#+WHF|2Wc@q$8!rH( zFY7E+qNj`=Bnl_{w0+537B5BVk_Ee+U-vzlJTJ4v<~5>C?{;pVhk*{d&M6Oh-~6nT zL;Y7v_YNxpsh8X2QhemyhF|7E%U&;+pnJz^#j6pMeOKpi4ETRyxc1FnIhT-r{$yAU z@?aFI#K9=xuS&NY9%@FtH9hfYhV!r1n}4J5s`!NsKjs~l=jQ{2IU>>>raPvo1_OQo zpbH|&IC}#cSrqjR@=b2D%(0?xh*R{-;e_A+n%FBL(J!O0a6ahmg}LnO_fYOVDenhL zaEKI#RAON1=4OS%!YjH(Zy!v(LX_s+Fw+5;)C4G>{UR4h9&idyu4$bYqEKJb zZds>ST+>9kl+H07+PmSOvbcZtd+jInFQQ$`H@O1i_IAtsK|$7?!t!oTI~wQP{`gdK zd3_4|!bG=rY^8}i%y1gZ7ouLSbq}&0E1ix&ob#Ex2UNa?rx+ufefOr?3mx<+@>$S3 z?Ql7}raqIiCU1o2ejrf6GS|0cwe$<+)YCZpW#wB~jI|*R3)S+%yK?q(Qc7QhoW0a0 zK(Wu$TKA_v>|;r+D}1vGlrl0*HY&`X>ChC6(Z&SU*W)&~uP^50t0ez9`R&Z`bJd>W zQR6n!C>Y#Zm>U~k`hFc}Fv$X^236O5ksyw*oFpb-UxKZE6c>1*7yEW3g@RrM=GGdR zW&nxbON4!u-Rb)THnPAWAb)&B!*G_Y!`Bq)h5*imX}xyHK+gK`R<46F$xuh+qG{}$ zVO{DaBYDRp19n0`#+dJc^b>nE8A|uDubeKiRWwOCa@(uGc zq5eI>95LAHa&dpqeG6HWbH~BU`)IvpfiCY>GHbA&>%l; z4RlzUTEdKQE&D~g@i?4sAgD8G43J(Dbc0T<9b4FTD0Ez)>}aTK4Zd4r8m{fi8!e!j;}S7IzNDI$yjM?BTx%Vw6W=zG*PPj#WaBF87-z&b(0 z1W^S`i+SK+OB0#**a)QjuL=S z6CF#p>(6}OOG@vpc`_KGTvXSZSd_7FbGtsqT24S9sri0~WM!aVKRvmW-cc>=YUz!7 z@yc!C+Ka&;YDBL9pJQq4Gn9KMbyG{(#GBNJ81(@8PeXG!g(C@ETf+Jbu z@E=ER5WX3Tae@kib^xwv zaoxzpetR3}dg{+aR1SeSx=pWvjw)C~92OjU3%@d+gF51OrK0$*JvIjf{~Csw*jsQ= zN&8-Zco-puZOo@%LMGv5I(j5i1a74Ab55sHTCqm@c{BKq=`WC#p3m*mqcUWMZ;xj> z)<)h{ekR_ybfgN@DQa%$8a7&sEVA^9aFO;_~-9HzGT!wsj4%~Ji>r;`#@zIW|l`wPD%TQwzesU`DYc1I^X zOam=Appd{Hm9>42zI|hg0-+W&a9=F1Ux>8Z3z!iAh^IX)+oFQ<%SVd`CQ5=<^Yb%e zgu=mo8%zzHN}uU>0O&wTSP6g8a(pnpYInLh8@ah={}Lro(vy)X6!rW^V_euaN{iU3Vf*KNod4iPgP%QM$T;^!Vv zdO89!2<#BFbGc9{2rcdY0Qi&V&&fc%kK}BMN(ch0jSzC2%i@<%pf6x*U5m4-cy&wF z+h`}2y7W)CcoyeTXX=y3oR>c()A(~TYg%iS6&Wjs#R|N3YzzlF1IrQm{&CQ&m99YU z0k<~{)K;cBjR5#z#2MI8X40`!IP^CU-=)a-=RPySS?a zoFVhpdcFp}V$OAVR-zR7hy2=zO2OwNiI`6wiPm z#YHmXzrKEK#rOubRnWfI=tLDn*#m{}WyW*3A)gw8k%p*NfATvg=RIxT9?ebOp8SVb z+PBjUThw@yVJoIWRazAKGt%O2D}M(0@g7aK`k`bfXh+uc!U|H7Q=e5CKb_kUqY|DR0S z`R3`bM}N$&y|NnnuB7+7)vDj#rIT71otlWe#$JQP(L6V`>Xq4Dr^h|_g7b@JU?MQA zB9dKxdF7U}k?VX)M{@8L3fz%M#Rpsb;yF4D!v8(G~^FtWqw zn2r$L&A=XWSB#Xx24VSNw*c;y1}n4F_DunFcBfnUsKDN+bX~l|!{TzA=4aqUopW+Y zjXM2YnL1fx$bz1ckv;+;C1Aao7y}3Uny_N?8bPx!9ZmV2+hnVMp7snuNie@p5orW(N^muf}L}j!i)E zzVwzCa4l034DEYGdfIm8GWgzC_H~JfJIt=Ty@F?+T&qC|lwDG?LAT7;YMja{z+_3K%N+)`sAK2F095IB`v zd%#`qi;E#v`oXVUfeUwu(FVS;?Jtu274VY}3OAzEpAq^uWtWaVtj+wP?^!}i%4P|4 zw~g=Ocg=oWh4K-kG=_Y`(ID-6?a6*_3Z9MkzONYCZ%0YRMBQow&_14K9Cm zEDr|z7Xuw8KzWpZx#5_rgSw?A(_L0gr?IlL<>Og5c9cy%lC^Eq_ctiQGw_}p@1(GZ z)>}}iGnP9+ew+?ne%aaVcs*46mygvr2Xaj*YkBHH`z1_kMW={(%j_UXKyXSezqD|> zyg>H^@#;QFw&ykoS~~_kC@OYVR+ki!UagYQuM!mg*S+ERcSI^r}6*lJBxdH0hzxBcPBn=!kH&r*g9}yRdTeVCa8G#k7 zETp*P;NoA;@;4>2%ngQ-8m7qs!@9@S-HcD`9G3pI64*hD$* zD})YX8y$u+iCeAk`C8$!F+ZN8{~P;H^v)!I;{TS@_(U)8bM22qdW`z{JHITvXD#M| zJXkC_vydX+E)RIJ4=yK}NnYcw#OY3Ane|X$8hrF8j z5AWKH$0jjxZ*^)*be<=8ykhsRs4h4iPYcS-S`fOF9rN%(co%%h)P$j7r)>k-gVMXZK!jid`9u}@DoHn%Z3owBX0 z55ZFFeR9&^D0yyB(CFBxFwh-3W4L=^7#;ry1WB&8IU{{^;|*L@i~fqp^7REb1GrF~ z(4Rb&kDELFOR7iY(A8ZAG1=;^?=0UfeI1Mg@hz*_f4L)*)wdNA73ZF+J7CaV9FfV} z6T5*{Ux=#m4)a;Q0G9g9yzHqPyP0%DyStexdSBOGi?Nq%IO?ZiYAYsfesx#Uw%3*d zhII?-W?isx&=}Ct>JqRR*us{UH8~=w#LSPW+J%0BwQK|KzV>;b4_bT2HJD(}NFA!q zlQ8+H9T5@0|7Pw(o$l4@X>r`gpFD4`E&qbbcM_8?AiVUx20 z3(=Ma=Z9vG*iG5u%KiA5FBZYzsr=hj6`j3ZFTQ;LUZlD%5C-HUZ_S)?HE3|@I%9wmRg zbYO&-v_(=}sUDdZV_F$dI`fzD>sCM+c*JjJ+kbd6k-pE|qgSLG>8GrN0ovG9BE}~( zJMyUD=9h7LNYR=A4YajA*9nIaccpw+!QCq929B~y&C;)&9E!GILGY0xASa?%1}Z28 zQxTH0@SP75p1$2y9{kkow6)E=5YUJ3w<@rVBxakaA=h@Wbo-{;SHot*?R#vb0vt>3 zRSJ6$l?~YDN&%gx-tcn;O>%5W|1S0mk*5Eq0yKl3MHv5nLh#OCIzags!RHr9HPfaA zLC_E3eocu$X0MgXc+m7N&(V3FBR6@3d3ZkZ9RHn1jpxuy-3;f}>WP}~>FfFR%G37# zqd!c+^WO<~kNolfr?ij%#p==p&S##&EJk?+1t&02n)n6giOXLkyEVp%u>;>x4G;7k z6bEa@8`v3nf+fXGNGC1GyKNQD(TLT~%u{c52_wBuyse{V&gK+bl#5%%nv5X!eSDT> z@l=4#>pHZo!@NUMWg{nVF&s$0ot_w8DQir6BP1MX$i2=87{yn>5oWGRDGgbsP5#8rfoz3yDSwCc3@LQG6ZaS5qk0OT0~D zX18fQ2x~2dXncd9p3qjh94FBmT|Nxb*$SV7s?VXZ@>!xL$b|}-4VO{kyTu;6%GKx* zC_R$!YLk&-7D`2^rc@H_ghAR10$S_olLLiqA&?vPIV62K7C|*@_9@VI(}T0+o}mj_ zY$arE@ojFtwm^O@ia^9GEkUtai&nA#$k;y3Vc)Mb=jKvz;Hq!nuoXlSQn^;8y>60p zx;VUXlMZb|@5m}N;6rxV$ZGiM@TUy#9BK1s69$gJLP>KK2H0TlWE*c6 zYitWENxGI)r6=?uM`}d8wm%1Ek>?ZN=9p@?e>(O3OvF?QR3Xs0>qE%7GcBr?+J)-4 zh~8?}WwZuZm7`>5yqF<2Ynasx$}-mi!$0LYwlptib*k6mAjbm-FwEe`qin@^3y3}LtT6Aifp$h1-5eqPqi$A ze;5wLWSj6qI7%|$r(kFNQ> zbHqK+2JG63tZUzSZJ$!d86TfAQLZCET3-Mjp_w18!OOFOa9pWAZ-{V|lgn+$QRqTZ z-x0%q$&i~exr<9}ZOCtvOo>txi7D`;bbcMk+5X9sny)|T{@uso+rxJ7eBS>C(%O)o zQ#?IjFT*WyR=v4cTOmr!8jFC7vfDiFq^4Mq7DWgdy`UYEF5axz?d;mUd+t{fe)3df zWCeV}h_V;&sV=-9X>G4&o-g#NezGwk88ow{i1YBdXdG_-*~U2k+>X8SXT`kKom<0W zogO(Xddj{d6F#iK)iT`c;9ya;Eq(^#)guTQ9JoFa0Bgp2 zW^eSw^m0yA)pcG&EiIwYTiRKvsivN~^Q+Y;yvw+A<}v9``;aK@{Gr z4@-tw7Qtc8z{{*GxiuvXeQSMDwF-R^h@`9V06rMtqW7r19t@`MI9hia)Keon;#N5; z5%x(xBuw1rSLbyFWRi=szu7^+LD9!6!`(^(*=r~OumbX1Y10QX-NNktNOI6ILt1vK zF}WIhBOwKmrUAXsmS1xp3|p6TbkCt_4rdf$E58B*dxBXTRTkbI!NekrO~;O`HOX~y zvrKzLeK@G~DXV%fYUx;W7}r$_x4BNUJ0jX3Fbsop^M&3}R$>AfNcDM1_R zb(d9CD_hBLQ_4o`N}Ko`^}4OuUv?5xt`1?=CJY`K>2kV1_%rK6t^pa z(tq0wH28`vlT3lbS8VWwRiizC2U(G%{U=sN#O>LoZCahynGN#=Z*Q?9FNRADCq;0d zL-?+%gi_AcWM`l1jnjdANMTBO8nc3z($dm&aN3sIj4{^{sN7Knqt)#R04^UMgv7ZElsrWlk>NzsDg*4SD*iIKSUX4xAJe7T%#E3x&~ zmEDk=XeACls0&;t*aq5|JW@y~F^XKFfWf%^*sRHsB1G*^9z#ck0pJP*^W?ME;(%&~ z20^p}1cxscECD~S|7z(wK3AwDOpk=qw7KqRD~mqWw2!H#X}WqqFv=CH^`hW*z@pP3 z8m~~fpI_BN2Qkhw{v6JBPZQu{8z0xzo>cuwwXM~wb=RE>rEloVh`euaANxxM|C`?Z zhp`-WlLNTQ9x3mdO6@9bx zGTm1uIq1eLR}2-Q_5eu_4ifTG%D#uxD*G89uk8!jmxtaX6gQEZ^S0U-9za3RS2&d} z0k}$w=6#tFPb6{yq+{=DKQfIk&x8C@`t;9~&d#g30)n*DU{V-SJ%Rv|9v;*t>l)xp zTU&|kHBer2*|Qh+u~l8C2%9wY@Fv#u>u~)MC+M3#<>=f-J%F6wee=j;5UTi!YjwR} z(vyzGGnhTVL1Fsl_dQPEsf-G z79b{Y4hSp-b<1*Q@+>cmv*D@tA%omEN~KY)Fbr$!g--pSJg;NIV+?7SAHkw!=|3E` zKCE;v_>lebHU(PaUjZoN4{vFUKJyc4fe1$PK`+m&(`}357Bb!sf*!=b_~Gs!Hk{E- zRv)HlTp`{iJkiMCfl(#4pic$hGqy0$CVV+!-bap@&E^EY#8C~Y_avzIS%K#kImk3IrG~-7)mC!}ENI^g&R}?D1ZjhlW`Qo_Y;gy=5MECQB=1-L^IF6H=X>4 z#hdV7$l^V;%d)t{r;uj+ekgZrM-p-abTaDq-MHmcwFiDa|4cNw+9b63|1d-T$jKS# zFMmme|G7S&i6`FO(>pyH3tgtWDe0gOHB| z7;9gWwh-a~L=`&MJ5N&I2x3w)?@ehpZ)LGcwv+8NpbyU@uf{XD(U1jB)3?C~Y?TF0 z{lT$C5z}c8Z4+BrDUAWfI-gxvwJk!?mLnkcGst4_?2|}`3X3coxT1NTO@;eorl6TCDPx$b8@2~zwf1^uu%X_C(buZ@{socISs8}93 zC!af1^uQOspB{YUwzo^=NV)pZ%79O#F0a%Bhf92g4*3(pUs-=s@i!J&!B&JpY7jmG zzuAJMji?v&2S;j1qQ>535I@fXRpfSFmycUs@fVnGyKbPor{!VYt8aEu+2|}fk%25S zFxWe6K`YPj8vucMszR8}MBRhj*YApcP`}#hU({<9}UwetMgdLLw zMa^P43A2)NR>u4|uT3L3}1Uu^ss12unvE9?$PHi z5vjy^a)dZ5&@q8k+VGJ|(H0mk#zKj1ebv3pMdD1f(L1*DFzQp*|7n`yRP!Wew?BL9 ziV=c}&L>#e$tlkN{0fXj} zYo33xTo|1z*SjiGak+AH^d-e41v}_Yj_S8fK2p_xtz4lQ^)i1F~kP zH3e*m@ntQD3f{Jo;GaAPZy&bZ*CS@mn=Rcb)_z}akbBYfYgAchEYFkQ|M^89X^Fz! zt+MHkWA^sHU-rxwCPr8Umu%Gp?i4lNOVb7~Scg}J%+KuaY>JoX?P=;o?w`bCRDE9@ zOj%CE71E44Pd6 zaKL`|hOfql)ozJ3@wIKd!qS;BJL(0SSW4xD9hz!F0t5;`Pt||I&OMd9znIM|Mj!7D zcu9v_+mWp%u;wcqyWdupWf1}fpR(37j#u`5Tr?%Y4cPvV&}@@CalMjzj;_puZAG%@ zg?=?SD4$|q7+4+)qLDqy8;o4~*)?|1h?n;m&kVZ1?Sh(wr@ua5d1{CHrHYL*i`VoM z^u7!Z>_2DRi;w_vqofpIj!Mv{t(xoB5DA>yf`tjlv~~C5RgTQGCWZgH)$_RJC3}7H z0)?*~9eHMfEh6QY77<1PFEs;|kAZJH+^M4H8$qRfv{Y==(zJ=o_q(+!4TA#-L|-*H zEkv_0e*sf!A75gW*Gh#tjKtw~4aDKYelf`4gAFfD4@YipGw_@H34K;3E0jNaL3yH! zgN)G+lW&s&pM_vE2MfseIj8JJ{BrUA!Ke|mqRBC}egHEuYcjLpT#E=mx{+sptL3Ac381-Z!~3HRHM#-XGzZj zbc6r)D{qMZ%dc#|ncU&Qb6xv?);a%QUD2trUvzM%6DA$Be<2X!`Z^^CnXV93V#!r> zCc1wqPXpV(9eI6w*oo0^)77xRSGs9Kr14|3U^Jbt>FbkwWi#2-CAsJco5&RWJ( zWHCOWetlX(sBN3+%3J7epXSLP$ofPj6WzLULa_F;42D}GQZ8{^9BG8qRIU+s@GPcQ zCN>(m7gaALI=MNZrPS7M@diZpDX2J0R-GjcuO9x-p`V1=!?b zyFBNOFl%{z3lNO!;W#8ydYP1!!+&^=z<2Bk>fsE%s>0%Nj$P3Y3rkdy+1+o>47BhL zreKeM`EbhT?gb5;(=!*yNu@aYAiBg08@~9H=g4%w4S9ZLGx;-;Y))&`3AIGh zV3tM+we2cYnyQlC5i5%IxT@J9cH8hVfXmn%SgjyELwSgzUk`ScbPZtadk8fM20AUO1B7TWR6GQBX<0?X@)E zWBHpaf0!>%>yP@{_y~ZfxKeUZOtz-=Whv_Pli)=Xj{j8WsYglTHRKW<=2EzAj^t=I zifowed}pY*8Yxap*^KB;?6yQ>fc`2EeBQP?@xqbrdX-`%FLGenjE%jBC#y=ocNXSa zpo4Z?rNa*P+vW@KJt+Qj411{?wb>{bNQ2k#5>ULS0_lCj)pg@`2>zLr`P73g7^Sra zYr+lR-GiCET3mklDy+c@tn4SxdY)>-!!N+bg+!l4BhM;mA52RdH*AxUgY9f+B|h}p z+|YQ6yOXO>V{aId7U4q|pvx0_v0_*X<*msrF-2v|)RsFt+>qW|wf+9}20X~^-g|PAKu9#ZSw5F*S!7wQGDaBph|HyXZniuEAzhGGkAyo z+TZxKqsr$JE=POyNy)&V@G(j?+a_FzcMw3TAJZ*OXz!ZMbBs$9Nv%;Js5BM$=YT8f zgvnM_1N|ROXACfvUINex(HU4z+%M53#sbh=(-oz!r=%IdA#h=m7(8e;2{E2Oh@go4CTP%8H|O<98ji55nS-}Q#hS> z2GiOG;$y%jaRwcq+>>|5t6ekqSoOC^2MYkor}EM#IqE6P@rVjai`6CY+)e`qgp2Ig zA|zo6#E+ZLt={r=LavE3K1Z-ULNf+~&V4KyUQJ1=jLDLzMVP*ZYObyUf)|MlDD`A6 zKRN94Rf|UxL24r-X0uIoFg6Mp_mijBv;3Y`Rx;8O!M*>Zs6!G;D$R8pST?ljEK4IV z;{p+O^CaLen>zjdZ3-z$5p9~K#BtrMH!Hu50Re=wxp{y7Uyy1?%!pgM$)=-u0W3kS zh)AH)k~45Ud>d5O(mACcgwY6EG4oMhxP@j3T)X8OqvZw%JF3c3x2aYVYVh^TZzX_) zu+XmJfyhP&4y`kK5OqqVQ0%aJ;^jw})X4?}??h_|Ba<7mj&2;q*N(_9kV?7cZEEL@ zKv6d)_&x|D4CZz9YdXjCitxq6elkrey)mXVFonYLj2bMgYKSNG=&VQ24$(jSI^67J zEzfZZl=tYAd>H=xy&K1rVzTR`9dqn+_(Mz@T%oFnNwcQC~yf;GIWi+XKnTW=#cZ$3&YiN^! z+;5ZyW!J-p)bi^4dG&)zUY5W)sP{xlM7=O;in|Wry-@HYzT$zH)fQUQJU0#dx?lgzVtfOiE|Ss}h)BhksI8}q^;qs`qfR>;D`xtL9Sh%i z4(nqp8xLYbx}65UKj=)pb>KsO-hV0hj2R&kq#b$#v8LJ40FU68XXRdSr#9cO*_dG5 zzlAj%R6F;A+2hU}=U6x&2D!z_$)tsZspi_ZfG4MAl|Ng8*_Y_@b6EG*9O#}zUvT#f zanpb!qG4zCR;dFR&?bGwRkiWM#(I8c2xrZ`pgi^AQ`W=xR?8tMc}{-niZ9NikQ65! z1HM-iMryxb3;WuE>>@=j?THq91bH{yP$)gv`+_7J|+$dvTCLmL~FgHIP zF;41j$>kn+fp<#n1hgdY?rl)ApISJrRKv?th~yfW-;H3vy2sL2rEuK!t<2hKUARn` z!$QcS*?yI!Em1i$K3V@G6rg1XNs3wo;$@HHiR-YLTTyEf3j1zp-)|KDxb5joSgzq! zSwz((_o^4iJ+C_@g;`Un<$!}jGOt)xJ{e7Tuv}QN2}^ z+|%hjkEeaVx@!;^R@S?IatKVS|2k%e$L=)}K! zXZ|g)Wz6KS{=joT_@5a6OH?Qt@+HkXhu3293v2|X6}+a|*@Y-8%|nsiTqsf>4F0$` zcP~WO!81BK;<4`e?d~`-sRW)^ zaiK<-Jl>f`>r7(nPrEUDGQ#UbED5YkRL&nB(ONMdrJO|gqk*7BA_XHWZ7KBQgkhz+ zN@*+TV4Dj7aR2tMC$U6AY_LXQNJ;FYR)p0XciQ}mGA}vTG1rUn(#@{%#LZlID3J<+ zTQ?0cd%V<_wz{2;q0o^*&++gI7vG#`6Y%dPvRu!Uv8s2&E)O2mWS8Y)Is{_c)b%bO zFpqp8_Gl)E@&sTzVVWZO+x{|Fl)N77FLfjX65y0>3jwP3RtTQYY`}M=BStAtKI;)p$ zSkcAbe|4Lx_LHX#@Owjus=D^$DsQGM{FVif7z8yS&eFnbhRPc9VY{Z&>9vb{4hp`V zfg|V1fS1X28S#n|y(+q@#U0ku0hOQaRFXuiaznc-DUEh$_QyL2PN^JfQSjRpnNn8? z*<~66h^UtwA}8D&Adj0e{01Z0K0xWjEALanR#mP&LDJy6r%URptgPAS8=ca`X+`+n zOv+9AHo-;;HJJ*)?=axqI;KUV*D9k@q?`S>7jYPXKXCG#F{{dio-}W(sJ5ZGZ7NuP zbq0CLN_ZBri)`R~;1#s>qt_{@Pe2{zQ&qOX)(XJaF2*WY+F&NY$Y^v(eD1>QkTdreU{ZnWk^WvD?^ueip`s9zt79RQU* zHn2PyH=~~{nUv#o1OPmoPRZ{5n4Vif|)?gRsC3bSI9VKb>Jg7LKQ+$YkUOfaD=wG$)-yE8tM zYMJ$`A#L@Ma4s_L2c1tK0+zaTV|G**J7SiR(Fn86-;v)x-@BPo@sI_B1wvceru1}w z9UDX!)wcz(?V}_5RNnvO(XW(FOokyvR^bR>XlvZDy2ZD`{nZ7FqVNuDsrr23z zPpbhpHBDlgJ*VC%S$cj@F*<7=F9S+BFy(1wk5K>@KK`=v4WfB>|Hqin|5{w0H0-ZF zo;p|_AM?GdV}583YIQWWBX+bMuBX`my;@po(IJS=Q5^)?4`;9GBZ*Sf#8_5wB+0H&gz`~zQ%=Okc9!QZ**Y-B49vFkPp|8bQa0nc-?>t-3pLD6WQj|Gs5h%*ADEr zwK7bBS3EaNatlTIPa*6CiXIB2xnpf!TKf)q`*3cAt1 z>|hkrC6Y_T(h1sxU{?&Y#mz6K z57rYTuYqTt5WN6&7pc!B1-3_+PUfqVGla~2<1*fCZT{;7XPt|smq#X+=bva1>;yn_??S5CSW-T9if%_!Ax9Lp& z{xIAIzs~Vjo#`C`Ouq~^LEP&P1ncg8G`r@BajArBURM?`Y6f_|c8^5$3zgT%7C)na zXNp0j?GwdV^}14)fJVaOsJIjla$({JryZb+hSReOL9v9XH-uGt_DJ{ELsl>v9#BTz zJY&Dil81lnY$OM2q>-Y(*4_v0jow5%oA@m5JHK4TYQ!$(Q?fPPe7J^NhTFRLw@YM* z0k@?pmwenkY}YLOCbm@wYJ{h0)(h?Oa&ASq#0(+gWkV(fW{$6q2P%sn?eVOs&o5uS z%cv{cfZ4pgg)mG7a{HvvPHZHb1G5r8{gcPqetV~S_Lc|JE>fw6Zpi$$PR|x4+@D)V zXbXLcjen}|J8cY^=kOU3USVLWm$ge$x+dj0Evv9)IDB%*LB1Rtfe@-+49BzuM)fFa z-y|}O9@N6rqbGXX=dViJ1W^wh55z5iAMyTlTj_X+p)bDWm7+RFpWt3jK(LF>N1HxL zfh@9Zm`TX13`@EEx1JA&qk4k_$%~-NA5z%W&^6_oU4y#MFkIIN(Xz(a2Hf@C^8(YR zD^wprEiE*{Fe3u3%?>F}tT6oW1*0P1xHFv*8g^#aUs-W4aBz3MYGiL{_GKAMTd`uo zFLf4g@4UjrOp21>C?=xqS$0=I#flJ_+Be}ZcRv!=tNMR1_nu)*W?#H0Gj<(x5D=y4 zNEHxJs&o}XFNP$97LX2s&_gd{fe|oaC;{mbAe96u0TPNMQX~inB=n*J(m{|S)teb- z=KlZp+~=Mz=RD`~;mry>qxiwbdqBpRR#^Q=mMshV|hyD@rv z&;=J1FY!bTt3}j+Ink6g2fi7c;~eGO-AqAyy_Jvv$r-I=UYVSMb@M@NaV_xXyDl=!_h?Kezxl3{~dNR4bo!!3@3bz2#r4Zs>gW5C5a;RUSf;1A_-e zhSZug@)>M4jr*Y?%{_bhGmjnHx&qZ4uqJXX@efa6yTV~M7vJC25ZpRt`#8aBToeBeMp9nMuj+}cyNUK)Xl)4$SfvL#c%}gR=&ipd zC3gVLGAz1~t$vyPz{;1FtS*lj&^NArtiy^nM3HUg91E5og~UkoO&*Imx=g_36TVJN zRPTYx-;_)N9e;<`Qm|~}=$ww?B}FIFa|d%#-B3vxPDaF6cgWAJgnlKr4ax1^)s~PU=e*i+P zi(fjD;aBVDO_Hwz0Zkp^k@YddZ7cjlc1v5+0Lu>g9A{$dC!kUIYW}LD{2FOfG8S0p zEqlxzSiXtjtRhWwhOZES!#*QiXE9(AJhdNSS*FHGi^rYGXE1}*EJuSE~UP(QQ`~VP|mXDqTlSQadfL>=u<{8Gf2@{ z{gRr2&=8lzf3AT)xGd>?t}mpwZ`4tf1zfJgj~#UDmbU9SVZI(_CUO1RE_D05_6P(0kN@LKx&W*UDG;s5UB^SzVht#oJ2bKJY=sQ9#+gtQuy zQ?n1&T7g&Mc`-N8mV(mwO+DKKmTpVr^RmN*D<8F_X9Sm1mewD64xGAHM2JBkLA!P% z9lU3=4>@6LT^Uk3QeRAm@R#xwF#n%c7}w1J72876o%v)z!uOu;epjUZ-FUHPe7E?J znd|Wz* zl9a8ag}!Y`=L@d8xBnXVKZ7rSgwf;Li_Y`FHtX?VF) zUr1G8cj#kZotkzi_KFLcgw6CONluht=Fptw9K9Ja{cKTss(<(NtlZI?{iSkZxo5=u z7H&e+K&RF)k~zIK_DxJ5|FmKv6`a<+1y)&jLU`*;wdlU}-t;<6K-=;~QVS{g6KO-M zg0}K;{=5nR;K=@v@&vVO(MS_gXIsOhOQeZdY#C$12LqRJE>4M9HY2G;OjtN{gyxJ$ zJUj?b7`49x^r4GYS&I1ZCTD$t?a0pP^}LWA>A$OJcX4Q>G<&%$NFjFCGa=fhjoUNC z4E(m%PSys6=#trXj*3?CQzTC5hw1>A?3(UM&=#N8bQO$CJzx&IxjMtKA3!lcfv;{E zeH&C8SYVldk((_p0vWSfIH`nQsor#D4k0 zA_6(_zCr@vdaby7YAky-za4$UBm~{?0&Y+C8aGvGS{8zcB~YHEw5u3s?gW4~hDH&n z;Z53B-IPmX9dX=orw>crTHi0^;UA3-yhB!|Ax?|=l@s(Q{2zZR&Sj8CTOlVid%Rcs z!)k{kwe!HD0k$%t@M!v>0b`OT1mMP|r6u%+(G5*4g{>@)>JAsD5|ZN=n=_AN%c^?c z(G}9I3{91GwMQ@jpFHwoGwH?RhdF7f)vQ3pf_%zr*F?eqHazv#ecHqps+yq&9j3v= zW@^kf(yoWRFc}{fTboM5qfUx2ZM^o1=J5=AS%Qf{4Q4(3+KdY}(9t0%L0UB7H-^M# zyOB3jIoZ~f)XzQ+6*I$MKfk=BF~D$XC0%`(w)|v4k!&g}ZeWt!1#BdaJPuo0?B{UH zimQ(NHkS8>&?hC{oh3wQ1@tIVSHdVhxv@2WY-^xp61pr4aCC@n$`O~YY@6kG6y(A| zXx+O$wNe`ycZ0Sk6Jx=aJfBwLN<%WXM`r=OuVK*o)2DU)pdU zz5na~SM)LH&*-B7UJEZKwR*=GVz22vH$PJV`S=t?ciK;cECQ^p<`!&N$v3zdCdL zQ?1E@%q?qpGQ`Ay#Yt*we7yF)KX7SU(rP2$c8EKk)EdU1Ir($Q)4b}S#ZhBB;2nl7 zHM11%aLp5MC@(mx*64(~bxyj`mp!oY`Oy}(lt>}EtB9YsRQkeblY|8LHk!h9-(e=} zFLVq`G%C*|SDb5JiJiUD_fY`m((taIVveAa3NDaD8<(0Tp%0Y*jobF5b*7Yloym@} zyY}M|ao_I?%a7~#1b*i`TeCSCy4{abuN#=CUzym8TUY+HukCsy~`!4zahikaP%Ow1IoZ>I0 ztA96;ukl)c@bE(f^H{y}U8cs-^&CuLgU!?7gAwIG?}i`FQQD_8 zQ+7@fq919d?6*&Cf&a+h5{^vQQkuT->#0`7&sKAaIlqDMs~4v;pWQIKYgqD)>sZ#fZfgCa za*pL^+nD|xp{6P{jN$r@>+1dg`3NR0#I?mOsqLg;WjXlrB~C)=`q)67ZQE{ zzjhBlJUje4DH9{Uss8h|e_Fy-%l$ud0ssD@{)L`D#J2T!`NjleY`-3joh!VX&J$?> zYlZMpRW5glxyJw&O(&!(z}o1LiZ-@Q{wqkp69OT5xK#^)ZeStv{Vjm{OH%-tiUr06 z0yjyrSHL=^6eSi2rG`_365HF7A1+pn5|&l#hv#pKGg1;~{=UIqed6S51$vY`~io1QLOp6DqImymj!zR*tHt=iPy)7O}(g-XMhEVy!nrj4VV=s+2 z>_V$zo;1mOr^tA=MO9OK@&U1aMJ%<%@-SEvHgrAsNw~{yCBWfrh0> znBkevtgMft$d|SvNaZJ)=B89Ba;?f6YQnV_^EXEtbkU|~d;N;5NAK5;elYTW?sD^~ zMK{}#*26n+;;BW8#7NYdneLT7{d-y7tF1TokN}3j%l*$@Yq}d3SwFu^(exm`wClZJ=7Y(< za!ueA<^%df^ry-1IIi_PzU zYk9{VrnQj2(LLUC=w=CN1LdbAUM7rq;JjX7}iJ9or5}Tx8OAHLb_J%fX)qq2^Fy%|3NorP>$CucZr| z?o5qv<1}?}1qH`fUWr-RjP4NVOsPZx$(*tFjsP55xiepuhnE)i`!fV)IkXh-q7cml z8%A@G&|s78_>rG&h!E{u_Yh>L``ip6{XoaY8u#W%4m!7&tpcLHt_Db1q6`X73(T)oIzS49{GV#pH=oSDL&iLWoCZOyB;{f^yHx<=4{fwr&$HvWd`+ZdTX)Ee!fO7w<_1&55{qM* z7bm5CFqk%GD1Q(Y#YHy=JlOp!{;*`YQCB*cP>g@6P2s?$Im*BOHLL!DOP?case83e zll%X@{>Hidkn*Pf3MWkM>0tTY2YhVaenda;@D&Lx*t-QsS9@Ezl)S>G1jZ z`sYr>k{2or(#dSccKVY~>iljlK~br9}7C6iI5Blqw$={C{Nf;`Sjsv(a^Vpw zi%2_zJVqBcU>Pbw_~|n}NfE$sWRcgf-_I}#U#$kjF##vZY;p8TW;H2KTc`^7TYcR@ zVuaip5V}@_pjROU!#K+7+!u?wkd@^g_?n{@&lydi(bwvOHk*A_E0a$;*VbjF5}bL#WXJ(R+YiOUB(04~_oaQAXtmjhpCZ05;T+ z{}MbiQ&4)BT{EB;P}>g$sawlD%as)nF}VqV*cIyC^B&c&)y!;Q@icdzOjINntaicu z13_Cu1qCl9tQIs&V_sFK-F#buBj)!6k2DlaLz?4sqEkGY-r1#C!NH8zQAOy2Lgcl1 ze4^ukGq1XtYtk0BApcyMf-!NXcbAcKlIrWRq12pPmd&HYew`(uZ}DnJ_ngkeED1gA zR$59Tl*T|_U8HidVWv*zo8;qkW~>4s10#=4b!k92k*4kCF!M?Ox7FaGDSVE7jLF-I zPEnqV^`6e_~{LIFB`MSbM8mU35m^)`wQUJfg2} zA5E+;?qTOj;hXWz9CN9!*PG_|QIocv^5Zw>xPJaSuiF{=AgyAwfHhwe=!ZiK8bD#> zpci+y{VFJc=EjA*!1j~*ZXE~~pwgVRVn0(L%d$51s}3CdrRWBmj1@s1$MjqY2V*yF;*E_H`6G_>ZQVOSH{F2;_onBV}uG zMjXB)Bn0c#Ce%4b+e|xkgtevxHMeU;NZ1!e?x~VOuCDKusW_2&+wOZL(^nP z*nnkO6dl1VG-gwNGgZu-itwobi9ab%$UT4#>e;KgxQd82l~uMxetML(tPD|vkuxArKlyp5iv!fzRgw-{Xi38cwhU_0H>&&9u=N86t9kYlD(h* zN%J~nq1$xPW4CRX8d9pgHH7xLGGTfqKkKzZt5ko2`T>f2tUK@DA` z$&xK^Z})*3wK9K_%>Zm0e?r_Bpl^*ArbCa$!m1GV`s4GiBWLsDS;1?~{Tm9wAR%Tu z4MHLo9s^BF?XILoNRAtv<9;}67*W|fPFWAhpas-;GV^ z>^`xZhbsEZg=N&rh1K5=hv|!V+;NM3nmW+NF2|-4U#Ge1URe8T4H+VD6%}h=O+(uq z?>8B!a9R|$u!frl6E6_te6=xUB{*E^x}(~;#_KFG9j ziRO7Q6V8brOOK5;KFNpsAaXQXeONTL1wl^`r0eq3vvwENE$XUx&zn0cokgr%`=b8s zQ??@5;S*T6-Rw#QG%RAYll>a0qax+i5AS+0AD{cOq~6eiI&Kn=TaB*k(5xOKP+2_C zQIabWhf!>=2r2+9IbAqO-u%=(gF>96_i7Y|729N<2)UM5kv|q4*OKa;YhKXfjm~L5 zdb5{qZNO0T7h{L7dX{BZ^?SgPI@qfD`nY0$es4>mte@bLUnUw+@L4mx>Pc=-Y8>0&malkZD>JR= zDGAR2>T|w6r?A24$}y&3lrN7Dcg@&z-AyTMWmV8cr&iK5Sf3fxwzXcKax6Q#`rQNq zWbZRSET>gw?QEC7>QY+)YJC`WnGt~{KGaBxu)VIR7V)Rq7A(+)tjL2waAm?O9`44W zjUw6$I0}*ogo5id=#-@3td*6w&rh|dF^0^Q^#Z=nrKu63;psVvNFI;H4)W$oejhs( zI4+sbKCe_V_dYqsZi2ER{WQ(QHXCHguMc1DaP>toodXC-**hj;4&F?LTtOF6p|j7) z4^zLeo>jhQ4u~zZW&4d-LQ(KSjIuV66$fYrGKj3-F(oblTTnev{!HqN==>!8r&DH~ADT|LvG-<(Auzpk2WCS+UVJgwn-Qh*kyg#W5iLC%_i%bT zZ{LAm>+w%#5;n|epgrxh*SLkzP{t%tR>Y$I-7Cvhhn3Q{;KKL%{@ZF3<>}DwMnSU; zgIV;{Ez#)2bXaHFB^RtV{~M}b)O5FE{$n$h)%8N{-F*p-N%PUA4L<%%=CF)231B`e zBcpJ%8H){4%|O<`fTeTGYC@0X*c&#!ma%HW&)S`iF{8Z*1~lzu`4wEk>A1VSR))Q= zPHW2KlW@;cggmRqN}AF;{V-K_K4RtdaJx&Q@g1Uu=hq^!GrT$FEl&)Ql$qMmeOl_x zu6V)JHuQT^;y_~_t4~xkNMBC^rwU{HS#{|TXdcgMymv%x*R!axf>F zn4$eU%eERqD!oSKaqw4>8-oH`Q}L}GyoLNs%ytIPhnLLg&98`%Mo4AmdPli(GQ^BP z*kDo@%4b~H6NPtmI-zPt-Et5Kd)c+(utdom+t!3>@xH#ErY!Wis-{j zBajh~F38+$_dW|O^{9(Ke2aSX@%wyy)t^Y|H!c_5(pKc~%yrn4Sb^Z}TZdV~U)pWh zpP6IW-rGN4a<%z>7Yh7-=FRC7_K~agQOlgVr~X_NVzFs?we}mAn1z4)ryK5@w{M)f z{ayC^)nEBsxh%B)8yCj>Qo}o)28!eAAw*1`6u6?0nSG;Q^7!Zfl>Xn}Zl?2u7owUg za~>wyF_}+@^F{C`TlXN9GI%fqG98>bw6&(E;*RA$6H9URUJRq22bewS_`nQ+H)^gj zXC8W}ZcuH_qC&9ppkET^L`uCye20Zx7^)^uMjIuR-`aeFhps3nui-4)y)=P+ITxnP z?+`r6N$_WxUVVa4W)dq@y*6z|of|fFpc8j0CTT5pj- zY2kxMqs*4~JhSRELN238wWhME6^xp*Xf>I2^9_jp*EehUJi9-3TvIzT9EvvX{voP) zdqfiRu2+BgPLPp%Hopr#FZZY9H(tzPhSU%w+sH6|xJ^Ljncrmb{y1KWl zQbhyvCQq;cB3R0_=zjsuygUcX8~1T7jX}86-GWpqtwAg8nB^cDY_vN?n39T$_V*c@ z2KeOA0f5CfmCf!7kllx+gq6p`4gBZ-*d0x4lA@zq82~fvhsXv)wnE5-ZaAq3mr(hs z(p=2d1JNhy4n4)5l4@7z(^@;A{OU|1xsm?(`}R6_Z2P;zdNa)_b|0dfzLRZ=6IPoVJjj5@o2_&DU#xRA%*a&m5G6R0_YPoT-gm?xjgFSlNe1bMQo8}EeIL9Of-?i2IHv8byP8=4@k z8`I4Sd^4$`C|d3hwUGXim0IulaZ}3xqTwsv^solA?IB0%tsdLOq7gG?wS`5Ua;&|0 z1+(e5NNkgu5?LlcCc;A4^cVs4XJMs z#8%lBnR-17rE#E%UpT*L9}smh@d(6@H366%XyARwd&cbdIZB>Y)6y(ij)k(_=9O(| zG+v`}Q^=M>%U0KGM$P%JC%+H~oXY?vc^nn1)@m=?GPK353zU04q6%gOI4#fL0vi&)XD z)kHscNAdIA`kFY=xv|9XG(RH69eDh(-=)gVVb)$QBqSxs8OCRXc1p>v0JL8vA%3b^)-`~t@uFy#$^F!?DqDDx}@n72>U}1i44! zzi=ja2Ewd=w+w(RI|P62DwR-C5z+LXm;Q*4&lBR1f75it1@R?LUK-SwCSCp{L_Rwd z-)L`puid>jl(S|mGe0lrz^cd3k6Z!lM)1BE+c?=CLjE}3uIR8;W1ya-?g;9r_zrnfn9QM(qh%?~&=xl}j_4 zpp$u~pqA(gl&c^o-+KVZ!*k=wZ@s*H;1uK7GYX72v9JMe<-+%sd-!#=7*DNwWIorA zf71^LPmYHb`_neJ8e#C_%$gi@{(YjGFO+dz+cCu&&+pF<(6~+6il>b$VO7kVUWFm7 zno1kRzqAXkfDGsc^f*nKXJ88(PCZlGa2Q{=sW$!7ByGfGxnc;}xH~L&Jw(~@G0|P7 zBx%-zwI#VLFwkE#{a>Fa@9rAXS^Mg~L*F#Vm1x zoJCQq1@c>Dn|IYsVX{np7@A)F^P|3YmKN%L3z`h^z#4glhiU;6mYaDELTf~8gpVg{ z_lR>1DFRqIA+wU&8IXO8x^1v;klcf$JITygLPp(3J9Td5){4D2dYXd`O5bMB;-4co51qdpVsFEtZvH97*Vm?-`#U%W6D+*H|L!fuB{ zIY+bU;4Kfag>M6U+iz|EOuar*6eT0Q-!7m%>o_l~Q}oDkY;^-%tAS#n3k>DgSHsmE zoR!#^yAj&6pw`8cX-3(gq?(sx+J!Q92ZUHw>lM=4M0|@*J2}a##slFW7t2O%yEZ!- zcb+gq$ijDh?m8ke>lqLoINMO&C0Xv_^{JAeO&lnlfxFd$TOk@1squAty?NQ@2;305ymusv!!%t&Y*a{kfI^#`pb36e8hrgI6^dRD zUR3AszeC!wQ6!JXg=>!0j?W^cuhcglWITQCnAjB-4g0N%epy{=g#C#+D*UuYLU{BI zq0R>;Yb{xxa_Pg2+yQmKH*-0vs4wSZB)HCqZBqg*^TPo?=Jw%#MK2fzSWT+i%x*s| zzrdNdRq*^m7{NZev|gho^>eGKY4+t=bpa$aLGVePL%H**6|u5h4$IDSoC zstR;kf9IS5qd!+{7?)0E*oho#-UC{EMReuRWgOcZN#k3%0G7Vf8DR}nN?>L_p?2N} zS42pBi)iy`HRFKT#VP8cT7Hn7W31-6TzczE_ud?jj%Z4g-O@%ilHFy%GUmO3HY2T1 zXG5l*dX_!kd!YhhOlg-J+Q;?FE7N3E3<0gKk+sN*fPGqz!d0=A2cxdOr*ty=K$Z5* zpEPpX)(}!Y_M=k0Gs?L?^zQ1kYem-DEReYYyMX!oxU|_b1KF@03)toaCVWR}prQDZ z=PudP!|9X!ca8lpU7OVkln-PQH)#@>B~?4i1WqiL2gr7ny2$(Ew3VDN(2=@ z#P*Hqc|56i$PXzdBsv+3u_15GWN=EY>hky46H6I`uqI1CdWV=a{G)F@3O12XXaT#F zyO5#N5;?l?4wX5i}UmcdLV((QBKAn7& z>Yy^0oB=O~ItsKIplaB=YNiABnG<^=KLx7S`4lG#ZiUJ^Fxo`M`1_s?V^Xx)Zjc@EPP>71dT& z=;-7dnN&eDEA=!QA7{J4sjEN^K>#vB!FhS)@>42fY+mJZF4Uq2BH8?)IbHbda_E}D9%#9y-rgO6 zwrewOKl2K5t6cX2U$&R|P2zJEjwvT&$!jfh`UXRIxk7xr7 z=QQAgyPmzEmK3DE*TBmJO*X(`T}PFU_p{3~kzn-lDG*IEjT6NUBo{ojpv=d(!87yA z85!$X@zmRGn!@45K!Fq|KC7#a|1XIEJUIbe#{|A_giliYsEP2Z{c6Zwi#!b@ zl@J~*lLt1Xn)=@c`bw@R%_5Ve7CK%QkMFVePLNd((P68gCDC67KU|VT%5Qp1_I4jE z#5`i>#u;jE6SsJN*5c5YtXd-ugjXlEm`; zs&M-w*BEY9s|MM>X=YFCmxWh6{rm||w-tAL)(MxTD96McLqZ@;lN@e4(V2YSS~jIQ z*dlseKeCEFVxh2`IqxzEmB3}Y$2Yqk_s8kt!3t3_YIZ0QH@UN3T5M-otwVsyX;sAW+ zLB&xMr%8!j!37?Rp8P>ha@!(mfvNp1bCA#1lQEvcamT<`T9xU{4Y)m_{K4jRAHMR$ zv7s6GzUiaE`$ZefO~czHLi*bQ8T_ZsgVQ0kAh>Zop7FZWKa9YLlmQzFWK_}x+3U#i z76pd&j{9%;vXZx#nmdJ3mo4RQ6-f;SIm)h{!Ivqs&lu_n+0D&=6{UmLvmQG03UeQ2 zN0wWT;%eE7)PkI%`e)hoZ?b0@4L}`G9^EM_q!MY@$(=bBO^CiQCnTO)74v4cINVdj zV~Ta0)c&3)uBB>OEaXd<#SBub?=InMg&PJ|Ilz z8hf$lvs9Q7YxGWZ)!;BZ+cYq-!eq-ItEwON6A2k{&hNqI417CooyZuKq6%7sZwQ5Z z9=IzrP7GqmoZNP1i7Te_oO@2UQInn9YqMvgf6uY+GwpxP)!E@m9S9)*qQAoD)`P7o#h;IqX5dQ1u;RjEB?uoQ z@!G(!qIj@`eQE{u&*ZWW4r59~Ue)-KN^|uR8Q&o9okdai5xO^wTW{A|FVESouKU({ zPL>=12L4Q>&F%!6^5hV}_pBTFqt>Uq$DgDk$r< zGX$D34~$)(m!}m-__z~2!Y*}z)Mo~QBIkoaIt3%rluXrnX zvzT{OvVtdivdjkLaHs1LzBhD?@Ei8i0BdY{^dJd$%Q^6&_Lw`-4;iu@a}%ohsiZhb zBMCO^iJgELs_3%2`FWZ43Y+NJ4-TH_*l%3d3^-7z(B09QuH8oMOIctrMT_cSvZcxG zvQ-R>zQW@gX0p^{Y_+sPzBjy_VNDv1tcdF!Nv#Q$P32F@!-UUhe7#c8za~ z9yvGL5E7vkH)h__^U1TY%|`p3Y9j`;#Kdu`T~AUG1`;&HDoP@1JN==b!i6by$oso0 zo3KIOs33DeCcz8Af}DSKWF5U;caRJyY#Bv}(MLw{)rgq&o)~ftHsOS& zIOOI{AzAwm9JFR(Dod%{ZhQH)^|IE=ZA3?|Owa(i@$}+jvSH)vn;+_#lbd1NFB8qZ znD*Dld!g4<(_R$E2w0IY-7S8(5s`BHpWD=t0e9zC?hhgv_`3rj*yt7Pb+oVPbzcMj z7EWCnN$O}2r(ACKm5E;uAb50zO)d%P9Q5^@@fSM90pgLiVTM?Nmj(aDi{mYva=%F_il3vNFKFcZjpO#|<8 z7=P|L$iJwd8C3$=A;yeBGU)WQ{!9k=eO{Kh`xoV!D`9^ALwcbRa_#rVR1@c2`YJ9~ zca?Qw9B1h;#=sh+SxVf|eG*%C(6gKNaj1Iyr6m}uzml3tOF=u(f-`AfA??XhQ!Xo! z!{s>M!HvyYJBgqg8F%tww}i!6^DXCYuw$vCNQ6x69EBS`T)Nv4(Z>r0O5AN+O7wu1 zoj0I>gZaZh1a|~?7Tn_HobU{So~jpl%%Kxc*EAjZ^r-B<40meYQ_U6N`0=bb_h4OYvhvizQHaGm@cFA(KSjJw{A2!Hp>&ue$Z zu*}!WKJ@D{4ciWRAniP$F>ohj&2zP4e_9eI+C?{5e4(mG%i&fx6xlUb16ow}1lont zuZ9ZOCvEtB3YXI9f3AwCLfw4sEs@l2Dn^pjH(pq{@=}?QmkK@kq(EH%u2a9>!BU+1 z3G;K?9Hmm%mp5=)!9ul@(5JsGbAsJ9kKFLp6EbAt`-dR}4nzAuBP`x5tSiX%EGNHO zccT5v?;)JyPNJ@K-z85X98oWE``9NV#e+J50&)ROXx*3m-+9 z$~v|)X=qmri3K+Eqcz+j8gAe5UUc8~3r+UON1Ba1q`jjsl^t5ofWnabb@Cr&%B^|M zmfHTM?YG2g>uQF^T*ijC!`CzcaIt-#EKd+>V)iNNvm81UMh?jICJaW3SuFbK$>ltZ zhhGx7=;2!Ho)~4fLN(5qZu9~#j=d~B5a^REYADs&vetf6dl-6i|ID)FZ+kQM1TH($9tpZ72uu0%tVf1eWY#v$hLxRkp@bqKNMK1@+BLHd#Ccwg zK*hwzEzE;BH8ENQ3&wvA!6W9X_gYRLsv=CuYe8^}H6|ZspWz$24yq55tZ$ityMx{k zU@X227T=9tV1KK*CGnFsCxYseXgnHwr+7XOy9U?VZIQObe&3U6_-lf2E+?t@eb2+!aoIBe1S;Gs zmK_-I9)5G&FenH|w*>rzsU{GQt^lA0$z8l0Tx{fQMR#6w;YBX5AucIB)6#18ON&aP z_XTfHa<)66AZ*oGtg;T^*wVq^O62e(KF7yW4+>|9AH9KkP9|)$aWTpw#qJZ#oz^CjR*0+F!DNU(WnnZP^;~ zpX`+T;=~LYH)~61B+@RrAkZEtlR0bVHg4# zeWr`sgMx7$*p^Ox8rg0)>({6_=qk*wc?#L|p=6VPB!LmyiKth!4nx7o5(?Y7e_mD>Gz98i-Q^!8gdyaup<(bwcLb7>GZkFW6fK_qrW|Zhws}@= z^-wC)JdV{4h!R=wAO6O5^~10-CB?P{-5m1!o0c9wp19xpwxf`jQBC$FxS9++{pYH{ z&c=zU97*0OG3#AA!Ka9V4_$gezudZpx8mdzApJ88rq?2AS_+S>##|!9I;=KlrXY9C zE9%%t5wpeZx_M|slp-N^TY4jb6zoGHckr1bY&LDlU}%b^xq5{qhY)ks$mSDg}buLCnDEVEZEQB2&VXP>IJyKivTsXmhEj$NHj z2tW>3dk|I}OGCj){<`bHz;@xD9#B8Poefdm+NKMct8SQ(F>tKtly5htyEnTcTgH<7T~p}r zHJnC|3gWft!bmu-p55V1Z)3o=wxx!IU^aGL!~^*64^hI@!zW7QfKI84UQ)Eq0IhbZ~1mVSCuzd2y( z?#xC)1F1H|LikV%q+F-Zi=o$^~Ei^Kep)%gI z%P-WPWm0bteoOvG!d2yAP<{7zL|pZc|D0R@E84HKPNhqVKD%;XM9*7 z?GGulJ#TxdY3Nd(XmOl0vmrG+;M)BC(WAL`#nBiAb8m?g*sGmI6GmYU_0QTq%CBrk zP02YfnCC4BEe|}h@lFaGU;+O~wY0pT;h@6&)R?$q1NoS@UkWs&X|~q;usCO2*$UHG z8L^LYvQbFQiKIc5%Y=>hjoa7nuRh+K&rRFg?%Nu=+>10IlIh;d9L~jpSzj}A>o*rv z{XPFkZOA#SZa-%68deRDojb88o4gU@c5?{RwIK$R5S39wDt)MpDu_s@WFZ7hedeSx z+r>VVr64!KRTY<%6rTYc0)hhC)@5?F@z#^eBZ9NgQm=g52w$&^7d6_OP}x+)#DZG@ z$XB%W-u!^&m(gd)f_I@aCGtRfe@5mtvb*1x`wueG8+o9oGF$~NIl_E~LJB(E*_Oix z+y*sK2e%@1jM>Ys4g+M2eFF!E1^X+%Vs4#oG_8Wc$6@8l+U{*wtobLl13$1UHG)%p zMsq@+M9-ivYJ4$V5HhIfV{vwl19mbsGJf%&iw@vMmtIU?JNP_fc>DEwaP} zI?L?dYhY7(zIf;&=4R7oTNqHOL(?Mta$}^`?Bq+oSAA+Em?4g`Jt8-ar#`5YBqVl6@AYBF8Qh4$D2W#24c7+DE#l34IpvX7%(wXXN!qh! zWV!gO&~2lx^3r}*W(>D~8DS&1U}WA~=%6GJD=CLTc~`k&0Bid3-&g?7^SYxI8F&P~|9 zorbeLBh+mu<>Vq|m$B0EybhtjxE$jxA~sK9VH}@gpy@o6@x<(Ds)J~GvgBg@-Pca@ zsB{p3x1Y^5BChZQ(;0J7q!JfX64W{H3FZ)j93=w)i}dq z@$;9L1p_8NOw?`H&2hVdP!sc5GK6=LJK6qHKr(iF%M z?yOEWU{`OdB=nN-NSK_q|H03jIqJRc=3^=%AEvTSxQhISAQrPVEG^+ktoERDZ9F_uUAJ<%D6*(9L7N^2cytYCa(0Tdp_{S2*Sn| zk-;82pCmG$3k3F;@RsC~P@#|;4dGH}Puzid#|ZcnN)Y<2PAyB6Wc(=SjQFJ7F2_;} zuM$(Ydd(W&tuAc|34I!yW)*_LR7qY@O3yl5VpiTp!a6~H+);E>egoE)cGiCi&&~hTInNxi#G^H}e}v z#F50F7Yo{~rPBkO*i#*G*BXL0%4V|K>7CCd3y00uHsCkJNH3z89E zn3!YRZkqbr*kt5~yQ5QqIxfr6xWe0H85^TpnaGQi4KF}O@o zWV>uhdUF(X9a{CqJRtvx#tZpSoU>G|e7~1Sjc<1oN$#A(D23{+_BmN_HB`OVh;Fz= zjws|Cl^H4MQ!$t8@m4tn_`jHY@35w_w`~+>?1~N|B7LMvXws#tbZJ5oLPrGzLX!@m z%~()CK)Q6Lg(M_M36MZw0BJ#5Na#gMD1m@+2*E!#Hy?>bfWbKu; zS2o$p^*qmg-_Pw!Q(eBp4Sn-~_8I4s$Sqt%6#3=AOd>~90PA52dmLn5*NB-%P_A-W zNAnyPw}(jwRG39wOXJ?zzcJf#%G-HRTLF93i{lBepjhv$z;<9gLQ0(hJFE2?vaZ$J zO7c03+-*0FcAd&e$6-o52xfL+xUWnBlkZ^#b^E?Y^o9qBQy#iJ1EUn7`O(}RkNs0f zgPv-QmCx)jNpO*TBpYYyIiW3tQSb-QwV{?&ip z2|6$zkatgV?IPAWe%Acg3C@4NIrIP6>uPg^y!!3PBWP&kt=?hR**i|FcwsRa} z9&BCinM2+?R{#EwA0i*$y8pB~{Gue|;`YIo=s)Jojf1xE3pC@b-;8e&N#8nh{;fNt zAY2RF8^!ID?4|Xy^Ch#^vtGS1)-mdf+-CYuwCg|KuUOtcI;uTs)NqZ$yx)G6ZteXx z^q*B4zH{+MzcMAD3UjQWZ)o?!XUmj$+a9o=%}O3|^RnIX3TxQq^bOaLekz4LTwzOL z+e$PCcaVAj)RiV-22A0L2=SOK>BP*d`QMmMGU9T6Ta<0LXo$ASc@j1{VY8SpleV7J z8ia&r6y@EGnkY?$h~-e{kOb$s)8vW%S~QQvtz*|uwKC@P^WPjpm#(=Vc^nRZV7zx| zWk+QC4aMKSP5l4uYirOt*TkxI-WAk}W$dcP~__pA{2thg&Lfuuc)(MG3V6^u%xd1I5>W1y6*}fk{$sS#>t` zVEK_HXHE?(lzp+p`J4dZC4hrZn=G&*j8XGs+9oEsdN@wWAug|k;h=d&;+6-hnxVC3 zwkVXjdQ|1EkLHW~RP8{)CSjGJoQb(k?YYeRgO_#m*|Zxezx;;U)&i`qCHbVaR_FEOGjN+*Ql-;*Q>BENBN)78ce-ei1=-QE9Sa z>sWPC{C+ByH=P8lU&tueLvNMe8FHmTEt^F0(+VDz&f( zysZOL-&$TKfA4fPB$#cnIsO92%#)FUOy1s%D!jpS5~*yUqxEivrR}*{yFvcP3}!s% zq00NPEK9d^i%u;{zs{_$8rQu=3yLwJo7q10a;|vbY)P}I+`9XdRo`AT4i+mHTlNf&dt8Xz1DVmD@e`u@+(GaZH8;SxGou`f|M`H z9;i0psWA9BT0rjPR}Zp6#zR)}9iYg-F(9h4ZY!WgAq=rJTLZBiH-Z3G4dl5|XWI!F zXdtp)m9IW}X!{unyj(LXAtvt^5mJmwc@(1R(Q~pimQ|>%lY?#@2cd+xtW@Gzoy_;T z3{>(x%E!+E*jR&!n-7I)B38)Hnkm@Snp`)|k<(iz)|utu4`3yRX~A93U)I@1=B*MD zqm8OjZSS|j(<`C1=LZa^vdHU?%uz$bwww%aD|-XXo_?Gmg1_%ncgT-q1fz->Jje&B zvg5`<)G%FY1=_1@T*Wi5pVRqXFP3+^ELKOa+(ayms~haa4k$g+^QvmW_)SE=Zxb#k zNUT3Qzj?mE{rYAadU#^>aa7y6p5j64D)tJuu-1(F#7;9}w5zq8vGubh{2a#gF{Wuv zgfqI)g{;uwQC%C#Mo#=W)Tnl5`g9G zc&7+CsMgA99NkdFm374dYa{PDfT_KO`rXp#+*|@r+?I3N6KVERewKNcbeefZURYUzv{7 z7T(Al>l*sLP#r@w43bU7)lRNPWKnUm&_lISh89X_4-K{zc(i&nxGs0f(6OvXEMNo{ z!a$o4cXP=MW0d{lQBld2wdh^@QP)L!#8)PYk{Q)(?0}jW2WjZKLKh*tU7+DBdz=IWXat60sHZ&L;_67#mVnYLg zI{xK@XBDIQE7N+t^A_(9L;gOL3;p*nh7#Ki+|!Kt4jh|d#b=r7ExPij9I?KMuF(Lg zZcny+4#^CEN_xRHTJmGf)>TMY5~l7mZ@kXGT(|v?-QP%gG{3_1bM0?Gdj56%i82JA zhOUOfGoC;B_byD8u?w@m!NlhGt@mlcuT0@d+i8m;j+Y`D^!MF1Q(FutRI{0!%|p+FzN!Mj=3q=-JqOFccnqyz*i2IgmUubGdbCILErn?)r;N$PyQCiodMaB~TgEw3Acwr;aS&ZdG^}O2>k(oRjyJ z>9&cG+=gz=LVm!qe(ssT5J<($9X{--+md9zPd7_JF#SrjltYdB6RRj1E#F3;N8DC;dK>WB@JoEZ0iE3t zdUkGf5h+qs0-JhG(d|Q}m7zL9pPAsc29PkWSMtX2xV@3W#Y88a&f4ay;F(zZiY=U4 z4Xy^+xIZ4*6P4;l7)|_c>NrN50IjX}uuzRmNE1GJ@-|eRJYg{San#JOX|^S`SlBkzwc^JXUt4Z)#{8%`$||d3&#&#D zy?#*@VTlmN5DSzv%NW_gZ0*!Vd|YbYq;{^qU}ARv&qwRxK+@-$FiO1-5AWm&)fYF- z!r3q4tGhMxM?b+|clC_9d^)KA?bONdANnUpqf}4YL6WiuM%oiQaL<%9G5@1U9>Mo&hbL^9I|t9j70gcqb+>IRCX|FJ z!>@|GlY5*kf9>18y2|Ej00gyP7Y*5+U||2PhoUISOEjVCMfa^y6$}bHI`!nLu->%@ zotN_hHc^Bu=Qx82BTKo0W?yIz>e@|Yo;PSdPIyvpcs#73`Bt2-0~qpg*7*8|`R!20 zoHdlFxW5>Kb_WA~>p@4{**LE02n-6s<_ zul+^Kw1{OPYTH|T?H9JzAaCQmeknm4zYBE_e1}5BQ+Fk+p>CmKFM9YDH8Qf@*7yNQ zA=K^h0_deq2W(%L`1PEZ#o4vQd54x3$c&f|g>2=p-A|%PFZbd4O#4e$7~t&anB0z46{%m+T-#^n=0VohnbiFQb{_v-=pIG6Kg0j6 zv1qc0JZw7S$vanZbAu3MY%O6qiv=}_7 z7Wzr@l89`*`7ZFqXa4y>v`CYii>|$8;BxbT(W(7#*lP)Imuh{NeDZ)xBBxcMPS?3$ z9qQ7qsj2=-=b__0|DbSK)~?TlWERj;aIyXP?0NdPMgACKz)alg?i#>B{2hZ6ZCW$R z>EXWw*!}FUOQ}g8!TC`O$SF5XHiLZQR6#w+ySZ-7>xeuVn4R>K?Fm=`V!KWQ5D!XG zAACUt>YQHF45lT*M?N1Q)yzQqc6=#DS9t~gk2j^lM|uyC);!&TN}gP8v@Mnocd!*izWG!es|_;Y{0ta>>CqI4KT9+UK?uZ62|)|Fs6~4DdMT`@ zF>a>CxSnU=4DE~vQ;&g_T0Mk=L+IA^0T8jQb=QijqciVxgHb3>CaOKb-M|SS;==j! zJqaL>kxwL5(9gdN2X@f%I*AHa%Docm?8WW4Jw;8a&o=PJn_AQ5*`bKOO3-HstwgM1 zpMVw-r|6r*^E@;&WcXoGm|4T0yu!?sU;C?GMYCKcYh^`Wv zR?l$&y>yWfy}Z)V4|e5L&P=9*^7A|l+ZWk4vL|;+jIm#mqIx)D@%G~7Nv?z|V*yT< z)+cxr@tE~^i#la9%i|)H(SF5-T+cphnN+}EKXEP6W|NolcvAN6Kx072n%J| zX!R8fiuh@egIXMwYNrU;HzK%Kh<#A87#Nr3P6Nn?6u6>I}}Z6v7HtxXPzx-NqS@~&{O+Cie@fW z%BSwW^3w+zSPIq_&rU~2Z?#<@cW}(Sy3r|4C~>?&ZE_^~05FSmYjEMVvzW(*GUsSk zhd$RxjW1_0a)*C=rAVDoMM4a2So&&o9H}r|(WT~{I1YhXMZd{HCnP_{J3fTVK}61f zDfjeK@AFH=>Q)<%Pa|>^R@JlVf6-uZEMQfad$oL|5b?4Fn1psEtk!DK*3_}+NtM%w z$5d0~D`Noaxh<9c$wKAkaCawsAPQaSL`jhbD((aB-X$!&O9;jVtv&#*Lw*k}Yc*1u zNi~OO%(z&YX!RIeD8%8Twmf{_X%!{f&U(szlK1XXN@%DZ=+v-eY$Y!;Jek)?OA==a zww@)g)2!tn>?Ca0b9;f&nwXwOp8)@?YOF<7m^Zn1VLW58)7@V(TOem)uewUk$JdQ} zXC)tq?##W!5_yw6pK7u1f8oW*h2()!A1JfDlqb_=_YX3?k2Z6&IrK4O`iqzKn6l+{m1mUS8 zn&6!(2PFuLOCHdQ-s}_NzWI~*@wU6jWlWH;><5Q9MF~yu@;9E_)kmw=Vn50mx#jz- zr9mvMgWrSEwyYBiqx>Jftj}TCZp>;QolxO3Gu_i*r}RtH276d- zvhU7JK^G`KSn1GB{vEQ35&W=V@8Sj4_jS!soBjwLQph8|tfzgIIm0jJubplRI4|-* zV~}MY_+Op5e^e7pp8VGL*G&(Rst3F6cWz`qJu{?%;q{kG%zI+4lxXtv$ZT`QJbxbi z_1>l}`*~1v^P%6ho5qdLmKhq%MMtp(+aEQQPUX9*Rb1@8!}4tTs&V7t;|)Z^x0c%r zBPQ<68?T=|P&vu^KEg(o=JmY=rjRsx<6!n0qd#M)!Hde7d*51QMu-q4&XJcX#2PJK z-(SBNH}|_~Z4xtI98ntdYV-_g?m0Ie)V+Gw^*w&a5Xz)X?l|q>-Na$%~VqO{Bm=b?!#|L*(V~x3)^}=ik z1`(Z+Bls~xxZhw}M;cU?ma zcdR`KfJX(ftDzfWY&1mHn$K~e?O5#xb;f%k;JHZrBgNp4PaJ*nd^#%Ss+gtW?)~`k zE_ZlgXxWN*0sylio5Kxelktqvx)(R*U!Pb_!KX65s3v!p?VvOt0beg~eA=aIG7N=v zt&!N?y`o%O(DAb!r8%RA*cR*aGQ@A7+mC?hVCDhmrB)SMWUKbK)z7N|;~2?l=_eYZ zTl`=zOkNGC3i$-dS9BpO}ec*8XkX|wGd$mHqtiS%fL;; zf6uZNEY-o*gU!^g^vc;J6-bc%RJv{WQ4~%Twe_^;5Z7XAeg74%S3S{v_vnj{YtgK`zr-#3IMtJ1uYW97dHvD#x((1++ik7zi!-T{S!vZ08ZNgTJ*h|Mq z%upCxgfG)7y&O{n9MLVWv0 zgg~OIUjiYWT1PQFfcMxPzs_XG{m0KVy@J@{MS&2`x|A-Ewz^(g82cbY>GT*2*p!70 z9PCiCm#D-auv`)fB^X2DBt>ROJJUJH!`vtoVLbMa7{k29@urK@y2?Xf_La2 zOyKX&@2^Q=dnTjy1^>?Y1WU>#r)#R`|^1+Q_i#jmL=7nv}X%iDa;pFw)R@)c?GiJNMs#MVR1m-8>-vOCm z&91hd*fhtN2^;UIQe==(&1!e`t2_$wN#+{2)g6mb*@r@)jt)j4S>?~q*sj}iQ1PP+ zm}Hq=8_}8Kq9m?0jhg{C5V@ajRcddT8&uX}pL}JKGi$zVoc_we`ejnE;td8dug=s* z%ADE5fs73t7qRGkC6AdbepixIit@LqOGbaau zLldjx*kt4RH##bN@fmsfPVQuO?jCt@S-)sht9JbLXw*o}9kpQiNaXt>Yi+0U$Y!r} zQ-Gj54x(6*fl<}(Dajd;qt=fX3>OVi!uW43*yngn81F)e$V(-1#;@XCe(6MEwH+~u zNJ!On*MzcsQ&gPT^R+zxDH&ww60^jPTxy!RYX;6{;A4W3iCHWR$6hwH)T#ok9JUf4 z)#^p5F-IrH?Yo9h))e#=?I?C+B0$RXrf5uR?jU|0nUMBWcs(h$6Ap8|%& z%YS;Ng$A5W9k>Uiv{Kweq8Nf2pJ%o2a8%m5@7pXh>ZtSSLeHj*JG!qXo`)+sl$P?? z4F3Kr<(Khk66NyVgM`AyG6iK2EZ;P$kdW7^KOeM!K{)*7RmTRM2=G_v*K1k?Ak80HgmOCcpIR10zcL2P1|Q8GJ!`LzJxu#-gxJI#k+?} zD=Ea6FYEBQr01D94g&<2N0K~d;IS9K)BBEiNoL++h>Gc$&B3Y!D1Db0z879@tAw0p zu;Ne#fHoE1j+^=QBwe&0+TWvZFq2z7uAt;Y!!Q7Vh2)-JyCab#k!T&48`%c#K;}ec zDW@!~rrgl>I`7~~iDe`~Kx(WNTtpK&0%1?4&oZn|-~lpGJuc&f4GxlX(;xZixdl2U zaH;e9TVI$ZgM^`x1XO|q9X#c#_4-s{X3{y=?20CgacFCG%}cQpLa&bQb_;Q?&uD8p z7m@;ol`R&*Jf>0?B#lcY!M;O!n1B#M>c`YDQBU7o0bf(Ac`P?|&3pcU!AEru&;2fJf<-9Tx4cx zf2hzBavi8X=oT4e`wIc zWmcPhShap6*v4S57Dq7HtC35VuFF<;lkPUEI3b7zAhgX)>6g;acbqCNTFR*rju;V{ zirlHuiv`(AOj*CTd{1o1jVuj-H$KtgQQo+J*?)q)<6n0?lY#lqGqB#L|NS)n@^_F! z!+_{|E)5s)d4CnBxmh>gxn>&_o|fnTTlW1w{sw&iwEolY|0vDR{7X4%ZTA&N*uDR( zJ^Ftn3w<`U_UcUgo}KOo^dzGYZWob3r+IGjE1PkDfPCIoR}*rMA@7BUhXCcdMv8X? ze4+%{BfaBV8oM}w#?U=Ak$^#~hZ{nI3eYx9(2$6q*_8BTUXUas7$?Cs&>c`+acJ_G z;9{LTn4#Q;*H2iNC#kWE9aj1P1SmIm+Hvb_b3OD-6Y`$je32jaY<-WV6{bdiLWhzR zX*G}Yt!mo#=)ruQ3>V7FItZv1xni&@L zjlZ=-GDJWPJdD%up!&o-xnjM70Px4j@lZ2=wHqM2=9-$ z{dnayDOvwvT&EE6x;5_j%JURJtg!`5NhUk9Ba6!;#y!_u!w%Cha;{p;QVbq_icp%W z555r}z6v8lrH-4hA8&MZPxI$Y?b6j+W3peY z*M#u6u0Vo{4D!t~STjeq4eVZ;T!YMxa|i0vY)^TRGos7!OJLid#JG42@|{#z33~n9 zK4Ve|!M{K|Lq~sId7k&~{h*12VAAagwPC;4hV(%l1}6?Q7fpCcDB5mx8y)zVpVQ)4 zx##TR7w6~3?NJ;f*0oH`&&;U=HP#Hr5&a8yT-*mU^^aOD;O@)tkc@Hz0pR&g>&91J z7kCAV_N#D92$5=C<1b+4GlFjBHf6;aEy&}UI7*{FRDJI7q8g@90?V#^m+FvvnFQK= zHPH$TMrKtEKkcukXVK`o0oAlvf#70~rm@`NK74pA+P)_$SRgGiZ{yd&bCZ7A-DCTg z@!<^7SW5KWH;WG)D9AUA3P~aUY$5lGRI(w4*pPT7`2nJ&7j?hcq!z_*{pbl15<*KyUe`9Rl$`~5ck263k<-3MQeW8>D>g(bC#P` za)GcpDQgIKi2m66mZ{L^9bPr~@+n`zbL4feEB_$qU6Pt!W+_2CM|jDXGAK`3Vt-J`T>k#MsW=H+X*4aoy& zQEd2&f}C@CKd5+_cCIHY-m~UGWWz66s@-tseQ7T4R*M@Yw2vgK!UDmE{e{)l=6Y$V zy+U^eAB4=#<3?-Ot{8V?tT2UQ9(d$fS&!DHgB$0|0g&^DtS{a~mbu)^b$|F^(GaC} zGSmqi#4qBiOFw4|y*m#cm+SJ^K9Tt35vgG9!IdHi|Jrm+vZ2%mlSH|?+jCg8^^rxP z1mb-CNhJo!6KzDJrc0Ihjz@9KGgM?p&-&OT)pk`swmbd0%?s>kZY@>I}<4N~gDnLjd9Q$hqY4nez!xvW18^-}&UUHf^ zNN(faz0vS%x`*PG<`*t@CyWgID4E)I1LZCLVR4OmN3kNY$Z|shZVoj_W)z<6q8YhX zr2mQ|PYRK*=;|eqWdf$w=G)o=8A{Akzt6gQHJrwwEIkY-+n3awAI7O!#oW!fmD$eG0rcu z(6Wy{PTVnFmtX&NrgTnCIpDJ_9xR zOPW4*+afGk0$rlf@jc{ZyBChiUcIpf z1Z{>IFj}W%CEMbSR+A@INm#l^@TujfRN|u7t;LWhec;BLSFGErv$@B!DG7Ot(I0$6p zYj|Z|4sgikL}sV^9r_j*n~!pPuk?{>n?7Y0BeV5hqFTIDXbLa}u*G(>wHVvm!cayY zflHTz)Ob>SYR?WJA8Q+*YvZ?)s}0tUTPp-O?so!GcyjDmFITbG`=+p8M`~}U750M{ z5gyA4WdnCMGcvS4z1*bP_yZ(0qAp?G4s{)?TcBhH2};dh#Y~~gc&yy!v3B>(U7$et zwtfE$17gD9b?fxEv^u767iaY;gZ3|v(W^F?g`&@idGw9}Fy))~bOBDUnr1?Hiji;D zb9mSt8tk`F*s9`uYnawJ@vKOY;iuy0;yNIQT`0vUa&WVyd&j^FZ^^>vgY6`>T|!2< zWU)uW`mADR1U}1|ccsx>#L)!-xRQMEBQ@p=drd9&9o_4asxMy*vA`ITIgawOxPX%ZUi??F}Vc5gj^)n zI_>4S$s$;72YM!Ede>JsjlVKMRF;~FxhyLq+D{I3Q48|*Sp}P+n|%fTmpi7E5XnRl zD}dxp@AX`%k>}ldshmdVLi}OylWN&>I~`d;DZhjR)!N_>AubGJCqvEn{XO#$r?Kn= zVJ`kPqa!67>W#w4TMD{Go>jSqtAX1Efm{WdE6>?qc@S~F!zp&QT3?x}1wViWM!%fU zz*R4(iDdk3iH&T$W%@3^7v~``VAQ$IHGfSD{%oS}KB4E((PB@?!K zxr8!+6ZbNJ3M@Sxvz}4g-S>2`+R zO0xW`(yI32*c7D-)rz>YIkLVEh$@-~--Rp`Iy$l{VHoZgGf~UYmiA=k__)m@yzoxo z=thNGMl@}1bnAo!j}e`?8ndAYwz9;Y?}|2uJ^U&{BA1vLe0?T^OqFnkFYzb zJ@<&U^y!5Au^qL{Dk+ZXzKw{Xm_H7ydhpfwx8&*kKsr48cMj@vH=3v%w zX#-7oWj@x(Ukm`9gOw-5&raMlc?VH@c0Sk5@hj8zhEcxi2{&IVBRcl6T8Jk0Y^C8W z&)YG<61|m8vj1fn6(2p4xF;Oytn{VJvv>LNaOwFl-b-=lY(dc+X1d{a084@04)NJn zCfDFCi6X`*^?ssraxtpzq(^oZ;{+=o6`eOZlg4pRrh;nJVtq4kj?y?7BcIH%&eB!P zFN`hG(lV5P)3*E$L*m7w?e^lQr`@aamqx0$9X5W^FX{9A9D?5IRL^{WYf0tTl3Gx; ztdW@d*2tUrS*`hx!;2DHhY*NR^j9W*BUA1^+{X}0JQ_4VyPksQ#|4@qku_Sz&<67R z%0q5tN%H>IxW?j{8FRCK=XxzWvE2C~ouH0Ew+AU-nMlk51)I#>MKK{EZ&Uj6w(|gM zs%?uvRjj;Bk8E{80&PbsHVCns?QAR$UH&`%Dn^eE7sl)mgNrSMB~6ODrQR@|#ee%x z==fiFFN4jb@-rHf1&4)u1`mt8o;E_9v6OLF&lEXR18cqLJ+ViMWH>0*ZVnbI;!n=v_!0JSVeQOSx;!yZ3)5GW#;c#8=@c@ysRGz zPA2y2`cgh(N05SH+!X*Bh%ir07C5v)-$r(9yDCUOOpXw=Wc*0x`HL{JALk?NK!{Lc zREJ4qhv4!Aq_(teZT1TiUbV!m2=>i;7ZCt^QQMmmZQ;vXJ z;d1$;XKa3)`JLA4a7*s`bIKGIMZ3G$8p^1B6YYzYE3Xroy}q;Sv-|GW_C(UmS#%BS=*Z8vA)ntk)o zrW05p`Sabr4{ZqhsOKQ4^BJmOb+$co&WuEPpF>r7PWlq>W&+g)(UT@MEeXRvYa5_3 zF)BFU>riU|TFak|Nis9*ZPTt5?a@%$8wW8`>L#etKdMA&)+NSg>H4D z$+hmwxzC$d&>OE8w!etZ-t^t!snL*nd#v%>?}h2tOxG=f{_sNO2DfoL?+C%mEQGGw zcBPpnn%f20(>YX2T{=!OF{ddYOJ3_7UHnw@3H7w6Sc)wWx=2k#A~shNx}%i@yJx^b z!b;cBCnq@wi}A$aw}+V!&@rt1yC)a&Z)udnX@n)gsO-?iH{x(NTpg^ec>>*;>g*sgaqpnfa_Fra42KCOBm z>ZY@@y=gu`86YN^Q%p?xef;7ffMD77FNMX$g(*TsITnIiAl(mJ(1YUQ@LPX%AA$wL z4rRR=dBcd2aDPSXLM;PkZDN&6_LTGnq02rBwD$W9qaJned{o7wxAii-&(mP!_^Vp_ zX)oG74jgkVsFImiT31^2!FAgpkyLPKU1#kN3ATcN)Yzy!8$2~32xT9J7@OGl?W!ww zumm;VNZ~V|@%hJ|c6RB6^Ic0i?YOzrL-_5dU%o9X-(Iiz>_{E)h3#AqPfI4TOZj|d zn%bH4d9fJWlQxb2d?=;0bWFMN_sVab&$VBT_^tldb&0<;%^8|90>)50@EF8r^E=o&OABfj<2bVYs3`o;!c>+kqAN?0;=yHobYAuU{p; z+IhG6`vGKp?9^uW41M_M@%ICT^|#xHvNihw5m{>A4)R+6t$~>TCTxKBoV~Y~zcSrS zaz{?tO7@kQc0k02XeDCcF?NDChY9%R4v%raFxSV9`PR?>zEA&M`#vA5k~Qnt8=^dz zsdH$Mk)c0fD4N%op^@X8u z=Pi*o0iw{K2mt5EB3qS@iT}1k)t8?KT*+RD)$+K1-jWiG^!HsRY; z8Bl6Q!1Uu$eiNE{DVu!tj>P{mC*I5JQFnz;(gIk#zR&;1m;CdLG3$Gx@afX;^H)$1 zi#77ldX?*KiY9ti^Ek9NIdQ#aeSPizw_AVx?AaIg8eE`oQrHPsEVF=rt&PM;jB$7S z%D~Zi_n2oeXVK?&rx9&WKHGdMU(Jn0YPhOD`5CoEQFuf41F;)~Y5B2<69~+O=M5|&Z8MK1~}Kh2#AL52Z>l>M&nSHVgqRZBQVuqZ@R@ zY{!X&u=E~a=c-eG>7aH|9gG77hP8CgiUHZdX>0Nq3HclMwhk=$Rh0U74CMKH?2A#7 zWn^G$nWjSEeM4sg@k<|bb_{69D}@45K*az6qP`^TRrG8P*bsiX=lnZY659kBX$jws z@@lX^nuUy&75I#jMARG~-H5KCCQrqMInS41C$It88KRpW+8G=}WgQR!yUkeS8yjI( zj{P`>q=oR^xrQC7 z6VG&pRJBJ%_KWpF5T>U0UbqJw8o_7Kebu`&Z&3w^ki5l7{4gg$k2*z3UeD<&hvLup zZERADkbwP^7J5uR+(*M7N2CHHAy@B2oF1&&cSutA$Qr7!CuuieBR$rSCF z9}0$OkMKMsRV^8;k$`{1Ch_IqAQ5b4nr;V9Ml0w&?TMEeiV4XKD*C;dU4N$Izy0-S zGM{AJLWX?d$dS%(li$N4E*EP2=F>dv5M_N(74GfmosUG<6y(NUBIdq^yLWvF7(Iojs=;w$BIuQvr@uM5B^|s0(-y0sL0&E#HjLN z@|w8y+wuRxe-sWGEm`>SB?L16;rQBx{A0U*qE-0rT$12$FSjMI>OMw?Nw~+hh^3fv zw~B;;Q6xb0;?$g{!9(e90VfT?YUOwN^DGWtYQLW`cu>lF+)PBkfDI% zk^vQarv66f0LF5lGAZR1dy}P2sfV#WfG7ncF<>pPP#*=oi8-~ptU&-E)J6D;(VcmT z@u=3!-mOZ=$uX}*W@a`qyzZ(V1%mW8wqem&dwRQMl{bebik#NgyQzC2$O)07E{>cX z0Gkhh9{>O}VXfIjJ&eh8_?a%58sk1`0@(&`rnr}TcfjI00*VA&OHabkR``o(GdE{? ze*d*U5lD-@f`{IHIaOm;8=e`lo*)|!Kg*~EIG4yM4P7x>kZNTm>_lp3$P?E~ewfwi zvl-;a5-HlBifce?vq59<$`yw&VTJKryV+8$nie(*E}nG9MN=R0d~JN;Cn6>jWy%RReXcPV7rU9Fi_=+@ z`1NMBj)cS5fb{gQOt95Tc%1db;Ch=?|0cedY%R%pk0OuGVE4pv;*9m`j$}VNb^B&v zd{esZpr{n!DfrFCVb6n3Z8iT-Ju%A2czZa}Ujsn}iQ=9hsPtHZs*5oZ z;y6(^wZozQOfp|XQn@2PM(R4#1<&99x<|*1JVFLx`x6>-Hih!oRokUiMxC?t!5532 zY19iM0R#hwTB`-RRcfkTVy6FlkQO5Y6w@5-AMjdhzAZs6S6f1xji=L?BGf9RgMWlG z<&Gl)Fd7Myi9MY!xDmWnB=XLFQ-?udBw+gw@}~HkX>vC^R{hQyr;IRZ)Pu0l3l57; zW|C2jAQf^f&et{-*JUS~Y$L_{sNcv! ztaQt(=&i*Dm*Uu>?K3SZIp)J>h*_7mwY!KGb?}5g}rrs4i#9P|L8h@aJ*Z5g^m=H8YU`YJ#l-9(4k4H(BcL%lE!>j>P9QMQ8yT5GOG5C5T=kBbEOmT0_1b@Nw zNnQX$tUQA4#11nOts{#OPjBCwFpG>O8A(omKp%D0*>SL$(a#nD^*dOr5LE7w3K3YRWSj~X3B~xMTb83 zdCrKq0oKF$&>~Zh_fes<%4+pzgD*V(5C6O~w`Q?%H;n1R_geJFejbIOX(nOtH`774 zwkb=nq)FGgn^kiZ1N=u6!7Va0Ori^&M8x6KjMdqtU0 z*OO0W{OihAWBQTH`S*wacea$kI^uxWWXS&QAtYfJbaYW?kAGFCCVjT-&2Hdu$LMQ5$7a>zqu42LWfcfM2?o_Qu8?~%rEha;6{U-`2 zu*AeppW?DcuWDrirg{|@NKHa1+1aV`v(lw8E8)fY{Z{pzi?&3sI{iz4Ho#XV-XOc{ z$ecUnMJ>M_yD(y0VJm(7^1xouk|)LBjq4yF_Mx}$_;c-6rGNwNYLytrc>$eD^Q)NzuwCNRp_<|j;4{%8sv`F4KR-(PMY9TsT;Z(hvb6LrEQBYX&3S%_ znLg&3M-5B~F;T}2vMF?qae5h#nk|N;vjlS2IO-7?LVPpva=fO>h@KB;@wFhji1Dp7 z9!)r%Vm=F6OpQG}&>umjw>e)1t)=qBE$YH5`^T>fEsD}O>M5WmI27Oy@aB5iKj!I| zHs!mYyCK&)uOH5?!O&`vkzKsGo%r_j1-mysq4GU*o`%;E?uNID6Sml14a)!mHBVXTdQdxCN%>S3hz>d*PTQ7{wSjd(9YR% zxzNe4N&OYc-qmNQ>tD!QPeybW+*SvyZMR?aD#8@N{a*Qdil_v|q{Po41ttsiYq77J z+8v-WH%DUKfYI_SH*)N*-oX!NM_Hq#@TpveRT&&?u88|7Y?#2%=$!|_!oV-7w*>S@ zEhFAru9E5S+Fnib)-PO(ld|oz)6c@=o%VQB{nAr)^yb#x%W?Ew8N)x;bS8gkZGY%! zDqRvAJMPB$M$7nv)@+V|2|_L>hkQ4_c4VqJ0TJ2RQqCXX#h-Ujw9G8xlAY^pr7&nV zlHn#OnSx5iLzMRGwZLse1J)O24)2c@codNF4B;qwx6&)^NI|G)l?}=+YPc?2cM2|N z{HCA68-=xc7oC4q-wwG0Pk~oQQBVBl&?EI(r-ZS z2kx-h3xxO4>cjY{3f4@wK9ctEVxu9%VOHq zIGv84ZC|O)f`iyOH@@OoD&?f{$^J=$2%rEh~24d1bcoj}0_A z1tl0J#dQP7K9G5%*8fG^ zdw?~OhV8<*mbIaah$wXh1PKC4@2K<^gb+druu2b2I)r8e=|~9(LQq;DK#&r8U`0C8 zlK`Pf2`v=qojv2a?)sgx|M#ErfB%{5VrI;|@4RDXCht@3`+kbW^UtMMhcDwl!cJJr zJvqZOS3Wj;mX$ggaycZL3O#TC(X&K|vGrvfE++@)iN^ZcqI~PrT!h;((BQ_-OB-2uXzP0Wg}F--UT z`wvjLJMFJyV_ORy`maugeLA4KacsY?1gELmrva7C8?18F;>u_6*IAgsd_LW0-Qzg# zNxzkNIs6>Kl3Te!G7ETl9Oz2S+Z^Zj0S}AZ*$*XlJ^;{1u?tkiU(P7~XOVYN_iYxb z{}6fiLjaNY6(I1gd-}f%yc@hd9`Nu9-HRtb+@F)IzO<=}3l2fS9XWMEC0NHAxEgn1bYBfTGBp%e!@npGI{S>jQjn7KcbWUQ<@ zm|~S#X}Zy*ts8K826yIMO+bl5xNWn8fyxLQF3*YK z;mg1r`l68DEHIGgdQllWJsCzh4V# zYTy#SVX({CE|H$I89+09K&Ir)VV5>;IZTA)D65m=d4ov|QMxstnOCJl$R7U}ke+01v3k}bbT?vJcZOjwoM?sx2fydyqN z>dOxzGz)6bj4X(yn7(r}*Qd&%kio5OuZ=O)-;++G%1YHM5i6FSqNG$1$MP(YBi|Su zm|$9ekbs~m#1m0K2HXL>H+#q8I4elR!^H&eC#R`O;0qihn0l2yn5R@ZlYZG_QHOU$ zu|<2l>4b~trjpdv+)-N7{md*XP@{R8C<|!z^vo>h?;8Vl0ZbW7c$p9%+uX!NaU=O_ z5_>$aSM>5;Xk(|Q70sRw+h+8-!3@Mr_M@}_!1Qn%^AtR6If-Z#^4&fX)&6cX38t&J zl%eSEx_W+lX?}WA`y|*RC%6Ln!lO4MvofQKk?kUL#;;qA7V@Xfdt>EI;68jIV?~<} zVjjvePK!^y%z;;l8y0J@EN?zDNGg_qrEN>O!9nGx%?CeA|EuzJC67 z)T7e>^8q&~m|$m*+;~;0vFQhnv}~oXp+(BvhAz|)cV-73U_rZn?cu!CnxDrx-)ncD zN?&7n^5?gA{FL@t0r?6aulnuu*1`xrr!n@T!`U?fOIzj+PT?Lp8R>8BzJ1knbZU3) zFMglwrx6}63mt3;mV5LiZnw>cSLf({dG%+*6YlmLCC(P_$EDBS(dzzZUq_DJ&^hok zo$Pm->R+e*pA0X}o{O1cN%sos!>FZBxE@$Smb>>1dOz&i&K@YzS%((6i5J2^I0gn$ z)Z&e;LS~p=PCgBE)4{SRO3(h0IXjmpl@MU-kdRiX?KFS!XBtdPPfezu^x@W}if1LN z*qd270Pm@6`6+5wdbtTEZG7UB5dD8k*vJv9y4?jCY}xFg#&F*ar4=)dLSNjB#+jEZQ$n%jEYs zV%l8EEl|gzzwub2csKG+D9#PXz!;8=FS~uwx`E_YaVMteZEm(>cy1HwO-;K`BS2B; z>-^i`3I8D`B`nzwyVy!6RrBrJKlx7>$}Qo$9b{|Hi4n3$kfL+2`hM%bZ3D}GOub7t z{HS3P;v6`o_A_;ZPD<=s;}aL1G{Z3VBDv}F_MJgQS2+-RK0&3Id;nx;Tlq+}^_)aV z{7nAR^udh(G;7=Lr`=ZAYh6II@nL;R`t^Ekfs4M1(&^Ht)&kuX+?Zy*YBrviT!&1v zNW9;UUr%0t<-#{BI(yyxYx7_K)Md?Jz8M1L_vz!k|rgtEZKWi*bQxr9)Fk<;>*-W zY#h5KfqL&}@e%o?`wyyHA1^;3P;L;5AgA;VRHUaBodQo?A{Ze0sj<(!n?YvovP>ORrv8c49m}EC-%%?uL&E#GB@tS#qeQptuXT-LRAU6#&c?Z8L_9AbSG|m znL5R~$ljc{cJq!EErb)ymB1Q~)qP90eR!{P?!RK348rF@uSaxhq6NESDpJ@*#t70= zh5ODDv88)XeAbzj{pO>tn#qYgjXDl`dQ4C~J(wKi)SJ6=`yoYDTJ+pgIoVHZHkm|h zy~NBlrYx9$dRTV3xzWRgZB_S6>;p#NZ--Hp-PjJ+zAW{( za7u;?o-19z7HQ6vKZ&lyE|(&Cj7F9l%`Lb?6^iKfb+@iyP9%cCiPB0roZ`@Va4Toe>es6zPr@|s5LyZEWZ z-*OsIw|UwVk1)}ElkLasZ(28|hIZxrRg1J1OO2308aF&yZ}PuQ zF42<3Jepux+|DIIp%c&3+LxCLqeUG0uH_Y=_K=(MIaF3tIZ&zzfkvrKP0P(SG5b5- zH?8L6y})`@nV2!Mk%&g~_Gjg_EvrDGy3A~smZ?BZ_mw*8#lE%OVV}aBOWK>ve{!Q5 z$=_brtKQFYv}Hd$kmj==YfX7Rz81tKqh6l27?ul|lzeO4n9P@-e0Kqcl<{Bh_qT5U zxCIe$QSp8^i~_cgU`*XJg)y$ArI`Ru!I=OK$N4$BpYDEZ|G%(4=F5rF{k#^~aQO2= zZA?4ni?8Ctz$Y)RNnf9;9lq&uWvFl)WMj~;`P$!w)^ej;4A3{o987)1zCi^=QBHLk zw#_4qXt?!$B1}B()Iw{*&_S6Y=DZj7%Bk3)QS3(#Q@o0q8D%EC1}_fonp|t=bBWT3 z1M-^&-M0-j$|#({0kViW+idsL=1#^;m}Cl9j!59=x>vn+PU5;*iyHRX*c2>JY@f>~ zGJeQQ%DYx3^a=t0(YRdp_!26v%?);EQ!j(cKlzBti7r^3Ae*X9CJ>jf0j9r&fjEek z*{ZzjDL|p1_t`00AM*;X5Y!w!>uB0>4f%;!Ax@Sp{8zB$*?UJ+PuJ$MgzCv{dV+6% zW~F86v#FgczX|3>mC1-|P&~4a&leW>r3+p0rNY}lm|I(XOrJwmm(=cb!ZSc1p^II;O^z1Pmz0I`K;WZjz{EvyYtMi&b2;OUSb`{ZrK{1dB@4a zs=7JvdRV@jtJb;W-^_ND`?bE)56hpqMHrAa!Em z51r_a{R(tA_MG=56MDE-7MPKy4V&W2S1t@3t?x3`u*?#!Hdl2)8@M5$e4^vzt|fKv zw<{N-dPrd78N%@7+&GwUoF~4{VGpIp{FJHQzJ?H67|?(+WlTCFz+T&A0dm`EOP+F$ zeO%&{`}4{@HK>JQyXH+!v2M6{q3*#$IjeVXWQhVCYuT)3F@%D{EkEhs-Xw}vhegA3 zLyr%5l@FNYvujv@=d4}y0U{dFyi6{b5XiK>du$JKPh3Zde}2lpq_w8N5(1*GUI)Fj zO$k@mR})E;1leYKjBfWntKg4u&~`TC`gL=(5MPe1&>!fpJ7d~%u8B!j57~2Aa78P4 zEqh~YZg7{|WMds*`8lqD5W723wo z7C9AtdcAostkPPZT*lF%Qc}HSlu*9st;@CbSQ$(*&$y@1jc0pY5E-3l%Vlk8)mAIy z#>lF7I{+YLeA=;&>TPWrb%(z;uwdalk-0WUf1PGBv|@Dh+%Yu@{A#n%al0VsUO8Bm zb=XV(@51e0k>k*B$Z;$)?+|*Y+`%_avMNS>=%x7Myzw&XW)W^{Nq5C;E2WZmwP571 zapm|4#@@}b(Y!-iUGLo9iZk(ObKE|&>4WU=A5Lrpcvm2TfDABFp=KDbNbsh}DgSE+ zyH&>8^S|h-4fuIswsHGvnKVKTWPczavl0B|_C0w{vn5LUSj`eg%wMC&F#LaXdXMNP z#-osLSmTpABP6l@`HqavTX7&mUZ!QJ#W7he>+@{so~1F9bU#Hu0iI*vN$z6iY-Umb zfOFa480iLT^Mrm{DUfdlOwsi8FRvHFai*A8^9m>O!)W^(Hh=(pGER3!ab}tVTgZTAU7kxQK&nkpWwA(BNq^0 z&K;0pPIK;F-gfYJ^865ExH1JIYFkdX=S>hP$Zm-=9W@M3&^w0Kq2G{2eyD~iIXwvf z?^}+$QN{Cng*s*s$#-Z1t9^2$W^MdUf^R!xVr>BTHGe|pRwpz!v#N<(2oI_1odXkuqaF`PW`(T@`-i5-PftFx*p#=I2tsd#d%8|)7VKrP3MI!S4rtHeO68G= zBs`NO5~a|UamzZ=eT;kuJ1hN|jF-X|-SS!U?6~IV)*>LIEvA}NrwGx9bX&+9h+pz> zQxm9&KbzVwR~(&binF)36Gy=rT8^Un#Fs6Z+1NsH%8N%?A-zgk8G>VJX1yh&HjlCe zLBdzKHj85Lfn4;D^U-zXi1wrv9Z}MX1*h$+97bPfk2;vq!dCo(pez;PiT%cCZ1m}h%tWvd0G77)(}VD^e6%Yhz{;b;Z@3S9 z1ehU`cz?67P7U*%st5BO+#7^MaGb9qeI9tM921k@_`;g+(~{J15NZAdnEzwUtK@kW zke_(rUda;2nrQI(3!h8s*NZoNMI1cXpZD)~t8h&wp3vY`HU}ktCIXdJ$%cgxWF?o9 zRc>@hnv9HHtVw7T6&>JnFA*}`!n$#|l>`yALA{4|*7WHl)tq_JURD$#X*7u2Ownx8 z&{et>-!IB|r?P|Do6QFmmjU>oylfG{;07Ojp|YEIb)w^&fmDSlzMf#Behu&D?8VBq zjq9FlzIADpfp?F6XN~h079m$G{8|n=68@H}l=*Tf?RN0)_v5C&{`pFMUmgD+qrXaj zc7&FO+j^3fkjDL9+VA`>I*d(psfd2jGw+(bZeLwft?NynG zhvv-^_1kE_XtRIzBmbal+x7muBt>_P<@Z0^AJ@SVtSuBS!PK4TigDv5V*H?wzevKY))-j6#p-Yjt_ ze^oXw``n=@4^d05>aF&8RINUx?Y#WO01OBSw}v4)AI>kMN{r%0=YlkoYx*{Z4~6|| zIy~>J#lnjau`%p`g*~#F&@Sj8lsc5qzFKqqYk$DGi8C|7ML*A0bgzDVNZWy`Iw1edyR8RB{fFFagOas$8UaQf8^3h7oFrE_Vp zY0gt(^9Sep(R*HeF8Xiil?vn~dapQ%&wJl5?>?;8A@n^q`11JmECn<179adkedI<; z+)af20!vgb1|>VhAniI6aUlDkp!t*Mqiud_Wqdb&n0h~>chIp1)>>1T7k$G&oI3G= zxzJ!}DJKI#e=*0AslVFc872b(?dm`c*Sb?-!mU1UDMB$7M8S4a>7I`ee3+Z6L$1L~ zJ>H2ueJYzUB2GW)P}pS^mto%c(TF%r#ynzbp{Ek;rL7D6C_S0j+&;gvIW>dck8qW` z;VC)AOuZTT@srIA3UaR;cq>U8Fgwuox)Bh(MEuINS&yiAfzMcJ0t{shb{VJ+0`Rsq!exO zLc~L$=#hhe@VmcZ5~k9&ObHDgZI$1!5xnpJd?)Gt#zy>mnj|wbqoB}zNJT?We@4=G z{yaF4<_A0T;i3}f>-Dw0qQrdQ%(m4C4g5aXV#Iv^GQpMxg& zcgO_ptmNe`nxz{~_I6;quXB0?h|mNSI1Qr+YgUPyWu9rqByB4ak@;m_#G~20R}uiX zM2UCkg;f=Cg=Cglj`{K6T{)n(3?>k}#Ew_2I?DWh{MoUIi;H>V*cvq_-Q}R%O_<$M z-eLFVuTOlG_CekYxZYvNe)@tulp>1=az9l6mE)JA_O!jfMvQ@aMN`rNZkhh+pI`j% zpIyT%$Ty=3j(C)9?45j8+Iy(xqYtKSxhaSPwoj7V;BSk3T@>OqU57Ukq)sy|(xoA7pOF+Vdoazt^>w%niUMgZQTx*8NhX5pbTJX`k)n0sLwAd#gh2WrA z6p99fI_)=5^d$n>gwhgQkO4QtC7A09pxS?c4DFDP>T_)v!W+>??R7jZF%-Pvq22=W zjzLhP0o%TwPei%%#6U4_QwyOpt5sECX+&Z|o^4?UR=5Bq7^iiXBKsNi$2W`(w&>q3& z0Eh`7M-g+4S^7*!&wDK#g$2PXsMA{!F7rBX7^R+43wwEfmfa@v>;G%dyG1?HN^inC zMP-G1SGp)PUAhp>j6@uha1O@gjfD*7Igy}F;M*9JEo&y|=>X3gD)d%mC_NeJ3@;jm zpvmi&#L5|Cc#o%VPlg(RG0D+jzoFpD7KevrS#0V_V=p85BPq4Dt*=T?zemOTia6= z+rQj3hS8)OA~cYN^70o4vY7Ym)XQe|^3e&(A9Q0uT$N$nR#6^^X{3-zVb>bS^IcK2 z3%!V&nJqpK;Y6)&INE<>t4>Y?KZVJ2%=kEOy0t_;?O%u}^O9b=n2Zq&398j)91auIC z!ogB3Alfei%eO~t^h;cUGv#}XeoFpQT+^b+p%`WnpXLY9^xx|tC^j<8M39;oVll@D zb1~|)$nH7UgUtAO*`b6xdZt2q(T16$8kfp>%(I(e)X(YH=<_QFNCPtE8911MKt;uN z68|K)!Qovm!Yd#*H**SP-AbIvEu7fkKX!G2s$Fp>;-@991;N)U9=Xb%PIWp18szCN zW~z!(cL}^ii*(=iR>nSTmlK~Sf3BDopRIjEyFGTc2t?e|Q!h^?GP9mo>u~Y67#6>z z+S)k{w`I2@SX+%rs1%a*36F>HFE-}&*adqn4cde3rlGoAHN1&+Tj~qS4&#Sh@w5vG#99UPl zRu`LTV_d3z%`g&=+S|27%Bfs@>Z!fVa!IW(xSq|L-GjGNkjZak&1!Mu(6JUaVP}7< zZhcIK7t|W9AoxtRcxM4FU9a1B7k2LBO;tv>WHX~QCouZV5TgEWBZAxzx&g*F>jlMp z`L%K1u6~Z92=1wW(HX^-M}GZ?uP^IF1!ueHcKlnhf3is8GD>88=~(gMsUhjBD^6ca zQGa8P{i@_OuxN?gmhfGYeh9`4TNFppMuuj~sKF9JoHHc=*{hO{{`}`3rxUt!+g0&o z(!hAw%ysAMdfOsMO7;G#T%%y4+>tII)4ED`OzrNs4zsK;r$EsbfnPIAZ2w(mNzz0D z`+*T_FHC&XYf#KWiKITMrC)ZMthzfk0Eb7iBUMbmgkq5#0A?@Mo zPz~$Cvc^_Jbg`Z_OdQ%4-PHZl?u65b`$uWMvUEMT)0F_8YMXu&)Y_-Ry|9!j7cpqV z+JdsS+nsD7uQ$<&Nm`=|aZ%o+9`G`Kms@&EMe@%l0A;^I?tS?s$(KqGT0mN6c}scf z5h!=IH|^;x6N+II6v3_NC1V{OXBAKO%ZS#Iycft#tmo7-1~}g%-^W&|uG=+U)=7ox zsIiUCx_Gs|u{qp1{q3qKl*pT8H&^AB8!D2l;+Xe(y*DT&)cYL~6>$;_;ffti>6Iuit_h^_oXMCUGfWsTBc*T_b zH!r|h%445j=JVEIU& zFQ%*r2pL0tu~h%J4ABbw9m1f0cc8BA)}=`O;6O>!T#`&E^8WAp>%?r|ad`)@2s4c> z$y}bPn@hJ+^XuXSxQ_}WzI6bI#(-II<@#Yq;D?20vxn1)RW5^wr{!DCl)bkARKswpI-x}lw)N}j0Bgm*Z*}qB$uAP2zxGl`_&QkYk{%9j*H*5a|(C>|AVgf<+dPD6Yj4D53$FU zyp;m0WUABhd8CYAC(pgvZS@=$?X8ge;u>84WzWSsrc0^p+Goku5SQsdmOsh2$#1{C zOaC{VKvgrTT{-ZU`mEEjmV2B}qD!U_Z6`cmJi9&7yMLVv_+1yo^ zc~5tjj!ton?x!C)r8+;bO;dLR1&S{AZGhi17F=Zo8UxGm3ttS{UiyH37F@^|w|PrB zr1ak<8qZd#hhnjz#LrmjD%TaSC;dn*?J*&*IbkT%w#|boffRCv$ju!W%QGK*AGhb7 z%dYR8ClE%3#A!9!_);bAUXcPdF7LtFXf}R}ztDI~7vj=+D(qM(o~d6Cs~+<5~_Frwst1fs-qapu~F4`JxRv2%q(qUZ&x}6 zz%gM?z^9^_(adK90Htg9O&eOZlGp~5*lHF;mE^9ccRLrGS_7%z?XM#D`h>5WK1Qc~ z8MeX9vG?Ik`xT{ih~_?1EzJ-wOKRByyw)Q(Zz3_r=k3!7p17()OViHw{1?i%b~nHg zeI5q$IANxkDsW@ayPJ^TbvYOnIUcMnmh^)UOY4%2NZ}#Kq+%v%0jOR*ulp*ij%gOZ z5DR+3hs=VDYnrPXDJxt4NS#Zm1bAFuw68m^C(^!H#uS#Bf!eIE}J7B-IY^j)b|ISP0c;OvOt60#~uOd*suhH zDN-wW@h?=KRPt)TlBsyY#=_RRLQLYiLL_5O&;Je>pTnVIvc1SPs0Wj^ZdQK<>KAbo z%h{TcrjOAl*vo z!tDn^dXApgl-N}tG*bQ&Yp#7_u6&lyZj2zLSUxU(diPg~WRX_PQbp%|1(%OAw`VsE zIbI+}%9#4PUD9IV@J%>q0v-|Ry4YaZ7-$kc@Hle_UlNT}EB8U97Fn7SD$}-@U&11T zAtH6B^=HA9P)p10OYf4M*@tR_!xqp>E(@M>ZYpE*}kzT87>DltK znB3^omTGO>u>}aplp-F5Gw6=OUO{V}ql#d`YHwsf zgi2zCQ+=|f4u&Z*i68-v7CP8Nz}}a>!ro0;c4ofx^W7l(naZdU)1UzXW@zJi zYD#jO-tm;&NW{81O#1slY+Xm<&l76JOIRISn#U-t=8<%1c|4r-0FLr_L>~*(`(y5uWP)Vj^KD^!}#=#Jn9yL0#qhrqj^V88ED&0uxy--i8K0o``P-sh5G%^JcDtlMkrlXH`FxLAm9_a99MUsd*__%MaB5#3fF_Xbb3ACI{cUL0oVpc85cuWd_NvpW?yhv z4%<@puATUU?#}S@rIt5TGdlW#^4ni+$bXi@B>?N!?^he>#{RbZSe>#|-S|#`WKGNL zb;|iKtkX)uYeDpI&ZHi?Pk&uE6U?KffELtnX46GbAG`c|GtFh^owqK+4NMY-o%7UAAUzo2F;s7wOs(|kMfUn8Ds*|h-Y2P8nS`Kw&-ds_cVb~Y z;5RW(YC59jLvw0tYb|+2IY9?r4qjD=wDbiTc^z&VfCMrR&+($R)?l#!0Yh8deTtuV z(;u#ZE1-x^?jlU(BH9iHq3b#bH_(9_NktQ{2Tg<&d%&)V4 ziqRRvZY*xk39e~5Ki!X0e+~D-Nl(ZO( zU(pQDv=t4a{ML|VfLyJ0cCskzkx=&P+h*%%nGi$kfU?VBq6-t6Vt!cf{&#ab^+EYp zY26PUyj8fw#rmKE@^>`VkT=|t572>vODmH;L-<5U70K|B*)kjFEfXRc?dK!Tyz`1( z1zBaWTrz?~38iGfxo}cI0ayoQd?QagRfHVx*lb)8ZGsH7jMm7T35#2`9A`}#@S1>1 z9VjQ+R*Tda!fGy%;YRjCC4`dlQ?_>+ zn}M1nq}tcJIu{ZtQQf-pmA4)nJGu8a4v3q9>V?~>-m%Q*x{r6r6>rj8Mr|E#JuP(V zH91tBYjhG%VJr6?T;9t=?yQeOX9$7!t5j5`a~CJ>97G6u5saf*w7^I6jl$4zKxnbT zj8jGT{`Q?h^}JY<#!;|6y7)yYDFjI{$aoHS7-iQwiOctWRn6niFv3r2k2KdLwFBxH z(Q?Y&)=7u={3-`%>{^C-j^%xXVlK(dV2-*g3!YdmOF7?lE(rp5pk#=t{(*5}#ZHFu zNL*fW(1nqVj9ne)QVP?!p6M{cT3)b(JQ9@Am%}hfMbUIGJr;Oeh3xC+mtR^MVp@~( zDx&v<^hpe~i6L<2;8>_hx)9@(c#T;`rJmENg(N=BZ5>eC`RM7nqzKnZ<mS>M_I_l-BR8g$Ls#*($GrRMfLkK5WUX5S*6>2+WX zT^8QEoJb=9(1~vrh8m+Nl(7bsBlS{GR4( zkF7Aq;7SHYAE9}|BiOm&5w(k3ELqd3?o)yhhPmMZqJFr{vc3 zLKm>3;vV=W3l*KYa~btsCO*Ss(y~xrE-w&Mi>ym!9o#KY&ur$%Jy5T9yvwCsvw-p; zk!E!n{N5CccqgIA$?QR*PWirm4+a)`RQHT2wvj&Cq@(Ahhrf3_gG=`I3tRML-XV;u!VkT&>Dj8W) z6;{LQ2YV`Q?Rvn?W@CYJ1~k(2{6mnXMD494?=n&+dpSu9Ow2(R#5b{;gfCip-cHrK zNI2tC-ao&0sOy^TWcvw*>d?u~|hc|-02#M;cwGEY$G{cfQIt(~Gr z_S8UJ&8;;v8DC~2;sjoS3JR$l03@ZTweQ=1H8xE-5ek$6tPW z*u9WNF@?4vvOG(wWAR%6^ZXo2WMs=WM9$*6Fb>xj24FvP%7`)}_zOn%u|AnQvvR)f z6LDCDQL(PD_to}ScyT68$b3D2jn;@3kd2q5Nxi;KP|B6^w6d~<`Osexv+Qu#qM{Qs zz%-)4AUAIHlQlS|FCZ#B0pXI_^e^|N&U{Jflf!N8rKoiZ%%Iw53GK7Oh$1E)S!qvB ztk1)Ox0;P{q9`WauEMbs`@=ggWS&>Fe6b5?QUQC+GTIOft~e=Djyf}3^4 zbAimz*Oa%+(1S00S_%>`1mYKAlU+K&=9xLpkuP~SbAh>eBgF8D($B?iBcn0%&P^h2ZnqnTP01wqu=x|94xl(%3azM7LTfu>dc*=N2 zMyG3RScl&|ygMknvXd>YvYYMXn*bjl?o9-4Kx+UC%*^9f;Bb?51G;ms(3>0CPHL%(*t zX$t@+&lAQJr7!f@J{jk_78JjO_*yOYl+BZGBVcQCfF%>{R$_1Eb)(k8hom_5*SbC|H(-@GHnm#W4n!EMH3ku(>aROQF>5 z+)gQibJhM?OWw8ZoCQ0)W3}5eRWm*fq*UUY7!5Rz4JETLl~|WBlo%Tvz*nzH3*O~V z#QIF6eX_dM&Lv4OpqV>-s)NU0&5?rR5+246+*6VdAhQOrGOX)Js<228YGWrv2r5Y) z<&INIx4b$?f`;#Eo$;j8Et$qFMTR1wXqEN|bXW*U^ULh-S3H5cYVWrdkw-!bg6zm) zuwD;H70ml;mq>mj*DU&Sf$w8Jwj49=Th8)NrPk!Hw1eJ6HK3nGi^pYTC`U_VtV5}* zo(C%LyvrO{?(;~{me|eT)=G!&4l`~WbNCx8>&AdojXwqOv3ccvup8OPNhFyVVh`WN zWhCYVGm#{AQ*rH<0+HAW*HlYD7gtodGyAtn?L(Lv45~g?U*aT6V&Y7g`Tz+L;p2J# zrDWgko(=M0aZr!s__)R&bkp*(YW)PESTVp_zBN^~P?oOv+{J0PogW3*4Co_b_RCxUUq3rheew>jzsQVKl%>w?NPH7`A zH#g7E0wPx`9jQ(!k%#9JwYjX3wQAE=^vX}@o_sy^e<4Dj?(Ya8AQZi{Q<6VovBCef zT1%?w=Ze;lb=8Gy9-D3bS@Lp>qHcDjouVJ~{2*bOE>F<+l`?OUhl!-iDA>vDe8WHJ zG9}qKmPa;se;G=|FU7F0^tn6e2)Or4zNl!sQJ#{?5pE1Q!qzV@ z3dxZf_8T}Os>ml-k_zDvS)iEfb3{@R?b>XKT6ECDqia>{AYA0F<#lh@%F5(qx+(Y0&Ae9HzC$5p$Y zIveeGZ?0Ug^Knr~NLp^mrws=N*qL;eO^$t1NJg7?$g^27y%KF{n;0Qwo5#2og_8$a z8wE)fMM`6Zf4%p zJh?DJAEEv|6<5vM7V9^%FGs0~r~{OKKL|hdvx|oyIzX7->1Zb^>yc6w8`#Z7+C3G^ zXR>*{MVPa|r2_7~o?wU#k2b{V=|S%d5y6WlGE3#QitQ3w8^H3yCfD(U(-KuuVJb8%ga` zD6;01NW2e$r32jCV><8RWj`ukf;k3JVt#@_XQS@tzP z_VI~$nlX=r4*k8HtmHUyK^RXcscV@DUIzw++G;fpZ*^f8#oX_pdRvR;NUn9s?f$jl z&O3eN&9MKE^a#A-@fF4C?#0y*6}#H<$X9OX9psc3ac z?!8^1jyUXYc@m=hp zmf|a?ziAJW6YE#q_ukT#y#3nnujSakS|>O&C;%5b`IFme#@OI8x)V=;)~C+DzHIn+ zl=2GrJL{pjOE^)%;w_zrJJ9HdCHhGp2fh%!88au*)=IQHk}(;IR-6M+38T_ig`IZ( z9hDF$obZE%V*0exYLzfwOC~V2Q9^5@z6QE#kJcgtY=Uf5_s~|=Z z1Ku?dujq@h1+YNl$P%9o3g)-jPQ&CEZeA+&w5kw2A15a+#Vcj)+V>szFVGY?AN~!tsE1iLGEu$6PWI+=C(1qGC)AD z-F!lm;U$|8t7ea)#4~+)_K9aMc?78zFXPq}rAAxiW^>=$42e%BX?FV+Rrd9DoRsO> z6OV##W$>MVsx$zaKNpm})fO;=o@^Bq>{T4z*>cqOq9uU2V%O~!{TZ*KIy?CQ z_bQ?cGC8xo%9qQ-;RU!6+;EU=Er7l{Ww3FlO{^eXYGj+xN-oYrxfSt%Nw;JQ{@;8NBfP7oSr* z^d;0v+1cjNk!Hp|!QQMN4C92xh1VyD9@8D>ndqx4WkGg+irC&V4^MGsO(dD5S)8H9 zm|XLs9ZNL`PAn(G=Os)t(z)_eXRN!R?Nx>f)(=u&vX}dT-L&R|y#K-5fD`~L-20mq z9wKB>P5KdgGtG)$o*6DQ)qi&zb%`s;X1vRSVW5S1Op(oSe^OUkhqL9N6oB`9+1T^* z*lGw|s{d_cYkuj);c(zb6MB7733K$Q{QLA8mqB^?YQKkyUgPh{&piI^)2HuBzW4kv zztlf`X21L}q>P?E2~(l#$hSPaby3$DJ*Yccu_*r5&XMr_Y{i|^eR?|=EGPh>}XS=PcgZ#8+ z(ryz}g{K4XUk;uekNC@I`oS4UkS?v}8)6t+e|ga2e6fPV1?6_J+Q%k0`Bnc%o<8}< zWFT*wHmM8-X?q$z$Cl3f$y2U6L$qL*>TrCJqw~38FW|h8DG7Lv_>KCu7&rvO3Z{gSm#P#(2n3!YNemOqOgkHJj6epn4tST)s zAQGnnFt-fAeB)W(^*t53JJLZ^ZdaK?ACWBu$!f{sPkyiq4)JFr2J?u>rSg1K=W_I@ z3%0=BUnt~;o47yX#iGxL`O4w^rG6gCvn(kZ%<(p1M)}j-ilOU#9PqoxEuD(ZaSsZJ zS5hds0N2<4@&)(68_kzjNsw+W*|&MmjBdCDvqc(K$>nA~?qB$ik@$#jk@_0BY1w<| z z7Ic(xZO!nZX=S%)xjAkXcXhPWc+cQ66eh3eqcNW@elkFSbSG-y&IpQ$d3XI5ZWfyo zK5FTgk4;-Q0JK)yxRZq3@`syrEIhb+4|~QEYSOqMYI9NJ-BXb}tIx+rhH2=fFK)X_ zGS=Q9q!uy~x-Pt@95GObjaUt{O7uh%zn45AyZ@1=>@gmr& z=`gZsilug+-8nI2rv8IAQ{Py{9;d}e^%>rBHK+5?r0KDOJipZDAd;MaxH{~vMvaeW|mApY(D-lr3L;Fid8O)o1t@AYHbf?|U#`OQ(&00RnqC=}Pd$0=VWR3Rp7n*rWLA(J zfB<%QLtLv7HW^}U4XxDu9O_qY>}iumVXDh1z1$X6-B|XhyX!_#JFH^R zRL8#vW5WmWUpT0ulVOEWUX_-rQ1}4k>SC71!zCe8}*Cs#|96{+gu3+<1 zvgcoRa%Z{uj8q+2O{~a7tV)b6&gv>c78Yb5iMO5F*3h3`h^k}m=$MiiwkMt^IGo|n zgOA@eh4j1^M_+(680_o>m1uZ4u)f>Uv%p&G45y9npXD)kLQq-q4&ZglAU;r4I}XU8 zuR>=sIX8icK^4++jnm;){@ME_@{b%)oi=OB8AamCRMai@(D~g3&lAWD0Oedp+0Kr8 zcfio4ierfwOm06&?Y`0}c6krATr~pU(_&XJ6B}&kh<37Fx0vzsipitldPKqV`#RPa z(sEX~!xFj?-O*t03&xXSoxXY3sMMoJu4u)DjJ>0Ej<;VGz)$C$U;Bg38eN{jtrsY! zePDZk@w=<}iLvRH$SwCK9^KE_0lY~)SDNDAnh6TN1w0lZDSS#h6H(6uqm9CTd2|KT%42Zf=kq(Glyn5{@H>qPS@4qqk9Z*ee z>$STnppO3yHvSIBS1hg*p-g+=-VI{dY_lwc>uEd_ld^F(`0aSb!xfjK_7HU z6<$lm7M&R63dCznr!&rrKRRf4BjB%w%Z$(du1To7D|@wH3LB>VfxulXFdKnK6by1Z zIzRASt{#ouNdsGd*?2s*C0K^vlmp(mehdKs3>G3x-1V2A9r_=%XqoNv2-=a-F0$kA z{`FqMa{KvzrKch|0<@u(6{>!L}2D>3(PM z#hpZZB@3RoRR&u!gP%IXR{h{^tX;=gHR5vZVVQcBb-Q&)I^#7dY-TSVCZ;xPJ`yW0 z@NOuICV5G#K%FUQBQNB2oMmix1=1ZLLm+A77I${^;pyn8VZ3O%3%p@sZhaN^F2pn9 zrg+8WALy8-WW*@r=yK{AV}eZGvKMlTXQm}gf|f`LVo<(!)wtZj$5yuU%2X5_BGV{8 zy`^_8sDqw+a+))8vf9$FW_(_LU?dp9vvgL)Dgix`YiMS3+Ex<^_EFC*&iK-$wTds; zGQCL9)?ohR#58|=R^W)tcsXw(zALE+W_5Dv2>$xJKZ8*`Y%+2M+nHrA)R4-q-r4N6 zI}7Ge`-MNTe7haLC^(-a4NA=?eFOI=5-K(S<=W4$P}}%Mmo6>RBCvD2pQ(OojvNe;~I&7 zQD#0PoLM*-e?62ex4X%I>Bw*%S4>38;qoSq+DvKA}^f!S8MIA*x}-??p-yYCsM4$QLl2=^uK z6qiK@y3f~&gv-;w`4Q(=a?Z@$_PLVqDMX@ z^j07NQ1GB$OX z32ZB4E(c$ykUn>^r@E7AY-`8h!^MU)06ltUC!`40i!umtq9xEG=aP(4P?6=2% zX^$x%QBJqI9YSzu*y`=Y@Fc8bO&v>NXJjh9N7w~iQSN@>+6&k~gcW$qu^k<$dS6!6 zAyNi2L1^qHG;nb?QB@u*^y0Ro0zTzW97NgGM(9J8m(WK@UT3S7DHJ(5 zAK9R*H+B(nn@)4CW)1d=ZX^&xN2l4fE-aJ~H^|6_Lq=Gs09;(uX6a|kOEpdAEK)j9 z0Pd5}AutfHEQ`T1S{1aFJ%0JVk~}Df7c7arU1I27FD?$L4_tQ;({IJJ0*HxC1xZM~07ajF7!uY+bP$i^kK~M_tE@SM`tWFnYCD97f(O`}BD^ zo(nCZ*NHx_c(_0l0t6DoE(D2z!2&u{-9je%M^m!0l!$H?$cvQqV=9@|erS{kL5%0! zC|qTQ?Mp~-@)Kuwf>#ROg!BMueKSY50UOPUa!eL$sJP2TiiMJ$bdNr$RX<`PX%-Yz zU~7Y;&2zt=%0d63aXQ(=q*pOZs3w`rYbhdLn0D)uYyRRbJ4q^-icQy27+hLTI%srr za$ws=NElZ-4MvGuM7(#nXqN4B{gFcNr=hs;#KZUHd{6sv#JnyXH1|#DRm0}!{!yTZ zz22jG5PXx62FSz05g~?j!lw$EZ3Er-U+tZ#FgNt6=z6)sCnlH}*#aB177w1DiEmqy zO8aa*PGF2@p@773el4pGl#3uQ9K&L`Ar+7j`fJ)N3JD zUqRi@-H->pkfN`kM}v=8)X!uH8Dv(l(9sQU6A1*&fok3t3 zHEDJeu{Kv7Ifj+{-ga%2dX7%%r zJO|$z~NFZrbgF4&mPH z|Ae3u!Nol-S<0-XY%34m(Co%-eahLc6c9RbV0?^f;QN!U>A%J~3)YH-$mcY471#x% zzNmXBcX+Du<|*W=@&X|?Yjn{4eSyl1y| z73cQE^}vgfx)v=((8IbrAT{ncpv%WFcb^@-b9fSTLze!M{Y0+5bXnSp+FOH^=Rh%T2Iw7=q&$V-(hyo*Y`;$cC6#(mu(_ruL#O z`~}x}*%l41D+ry<$OMctp%|1xvVRuF5p8FS94 z5|iW>am}{gFWS536HX1j0CWgn2aUTpvGP8h1NZZ`=L)lPHYSXsxLnFgU#q{A(X31c zyO21?Sx2f31*nDVW@jy^)pK*2T0|mT7H}kO7|VPcnIsgGB6Y4XGBOq0_XWu;T*_7u ze>zG>13URTvpX})TB24V+JdhrrpwOC-nl;#QqO;ZO);+MzaL;y8^ZVexJHsV0kmeWYskJ3d5eY2i;eY@q0&WbdeSPJ7TP2|70{w zf&uM-zF=KUkeTQW=qMHcz^!%^TR>@o%g)JGhCkh zZ^n3v$N1I908#kJUIQDKli?yTb0T(t8CkZK_BFP$|Endh!wL9T%yN6&-K@4p*;?9e;8 zcJmw1Zl%v|YTu6O9<;IInE79zy1zLpY5FnSOeh;4@Lt-d*THil-6Jp$8U$6~TYZ{0 zgb^s;H67hN7wK|hRM2`ya+@N}RVM{~^VikWJBfd6O5F$jw||aKmJhSm^rrN2mrlfu zmqtuB1G`(m?#@KS=q7;OaXblnR{tOCXgTI2dT1ZG??Ff973oQ3p+d(hM)zbKR-bN$ zNDE*=r~coO!?5U;gccGD?7mz$hQ(5yv2Gz})|dFa%TiY~)rJMsP*&}f4}+^)UW%K_ zYgbm=DRW!iaTSk_J|@Tw_CS-iAk31xn7tT1&kE>D)GW(C8dpJPy*5*6MHIjy7W4Y} zvsHQiXwd(1_sYs{D}6rIt)X7G*JGM#vYM#nh1PcgM#WKGs$BzoTc5<1TUPW%%TtoX z1A~(QgvR&nVy!v)W-kpAR|qw3>?#rYY)->TF|}f z^31C+>|}j$b)~fxcusHaz2CWdF5!gd?(fw76)s3qFY4Ds)>7d!1rF+V%N>h-rC2IE*eDQt!s}!B?#B?D}{deu< zPNWLQJaNb;@59Js6L)vD1ehDFX2wdCbz=^4aaE|`G=kLD@wsM0%~;KjtLC-H_Wfo+ z*aNRsOve=02D^Mi31pIO-*ws|uABq@k=mx*b+MvEJ%76VK$Os&N^uCvr8nbpSH(x8 z{%t-uqwm{egR)%f8hdllD4`~#7#>KTR)zwc+!-St1cl+CWr#Ot5Wpc72W}zt zthm1ufisdzNZS8|sUo>^@b#&dEP8glll{zv+J)n#9UOpy%qzS?%i>X_C&n< zci^y~*5-YS4>T*>qajxMo*M~|>4)i9?YXjXgbLcv_xkR)jBaQ5d7lSeboepgtU#%k zX!iQ21iSboThEcDYMXM7m58<-??&lm=`+3oB(iRbI!u?FDU_&EyeuXdnj^_+I5wg& z-A3Pk!Z@R%cP_jOp$U6XT*QqmM&;a!EDAFbUEWkmG2RM`Rh~I+P1j_2ubwdpT^Fsz zp7V;$1%1k1aY|VaT)<3HGvv0u-Q^~qD;ipd_nU_!c|u)H+}bV{@s+?rbP>Ag3R7;` zeEgYhMb6@pM`SWma9$iw+(Hq-9?Ef47QUECtf%Y|li?NJX-3XtSJG5$Ahu{3_~>h( zfJF-1wiTjWkX_QMc#&UIu8DDd*q&H2IutPW@uB57tEL2Fg?WHQchR-aQ+o8vY;|NU zoEK+*jk2NkRz|||Itv5z!&pORWZh8dgREdVc|8vQAh^MCT=moFIT<#IyX|P*ds~H@ zqfgI<)|zU)#Qsg*cpnum($A5@IN>y(DlLdc|aJz|VYDILc9G zsY8=QxHA=9&AOY&w2QJ;A+UMo>I*!s_~@QpvBlHcImyWX5UrgNzg!|Z z@$oo~#q2$Cw|E}x9L`W#8IHIhR~;i`&G;b$U;Eg&f?u~2j;15SEjM=x~4!6IOwVcL0(KG6r-hLisi&qHE5$jj*&wRQ0NdjqZ z;Tu-QM=aB0NA_<#dE#thYU5{59_H~aj9Y)K^SI5;tp)Q(N6b7vS#+9oJL2|?9|u5R zQp=)^d@3VMpDmVGX3Mmy`^cK!J>NDfR9<6lyo|OQEjwJM909J!VoiN6U(1Nm9+Fjf zyN(NeT?%IiXspw5KapQAYp9;hui0Yl4>eUZ>7bqqr&;z0^~;0oLq%lti_YtDW#S#@y6${ z(M|Ho^*6|B+-R`u#Oh93VqwFyg+6ooJw0TCZ{eq|jz=D1bfRn{AIEJ)ee)Ql6D${L z>x`=;4=-O4@nZQ0DaSu%I^fFP7XL{)h${a9w%H=q^xH0X{UV^jH+}KIfiU4`-$xu} z9?A?qw|J@U7>5<4X*(Hpc);3|mCNu=u#Lz{>~iQ_q3L`XPV_5i2dE@{HQMiM_zm=n zZTm;z&iGE6erVW$Qa1Ci+97z)6_$Iy_Xr*3`@hsd1D1;rnMXgW3STTAq#_>_b1@!* zZ|6=KdV7bC**ETx7w0AF0sX%-f{$$eu_ph{&vSrr!uj1eS!ZEf7zP*>-4(|(;*+UJ z6V?IBHuoy4{AX`>o7wbMPnOu0--}2OSUL-CO%b=NE3{v#=`nvUCt8;;*>qlQi5e!_ zddlCS9_Tnt^EYbpq)~pu+qvU-IE5?MWTroM{Pydxxm@>aKOA?zNdS{6xb=?8&dM+> z2MK67i%R|qy5OBA>fyw)M9f>^DtF#17Q)9)&3!W9m?Y>D`d@DRaM&Y7zDdfkYAJ-BJ+Gc!dNwpB;>fx z(_U?t21p>A4I3dlpTbe#3CMlkocOmRHcAXNJ}Fo^8%%FjN(uqfbu2=)`rT`)KPnhhB?cw->n6LpF|^t_H&im>fh&?k)7*b5j^H%r*R z!6i#%we-^ldRZY*rq7Q5X4NM(vCl+b?LPO6&GMc>!pynK^P7`qs%y==-r8R0U$xw` zn@;&pY%b|^gwJ?PzpA<9XDRj5F(>3^ojeax2UhS@LD(TSaj=G8IW+Ri+ zHP&{I&a(nozM?yQ{B4s~is=qo2}Pu;Gq7Y`VpN_0nywoNWiVlPk+7H7EdzY=i)~$ezxNCaA`D*sEP_pXEJ4y zlxx&63%ZGOlxJ(_UOg<#7RuA4m&+_4Vfw)d>*e8bz~ z_t*gK4;1zb*hn40<;~d=3iLnL4Jvb6b?j30^`@enZqLdNRF)*33sSm!ioZI4`jppO z3<<8Qlz{y+l*91^$p~ZEV28PhHjq>pt@XSgJVV@BBjYm*AV@8o1vhxe-TKFxy(Faj92ibe}Mt&K|vaxtMn;7cw|_nlFNL&XRqC+s9#z6QHO0?|8?tglgKVu0k&Eb-b`~Nvns2{9C}yB z34X!qpf;9|&RU)_87AaVA4fRy4a_(exFWUY>e!Q*u75T@C8JP(+$beAp)5EwV@&OM z{Bp8DeF7;FCy+L}PUqE@eX)?w!oHyGDaKb*J2q~s+0F7&g!Z`^6?7vkP z_dj+7KL2Y+AR(}Qof%;l>C?Nx$YCpv;3f2{aA`PVCq}s#|41U_lTp*CY4c2Qa()S) z9S_nUs2WRejNp-Qw9+RQf_qQ;abq7lt=iShis8og^i8UhoHAiLP`f7>TtO`-gn<~N z$k;}7#<*9Hyxa&qTbiKdUCS%!y=J*rIN{K9Hii0NGrg$#;1<1G94_j#ftmsM^<1Ao zU{?N1`b-D(B`c9LM#f1`R!K1IwP~Y|ON6bmS4D`Kx%*gdJ$ZfJ+pDUTTMY5Bf$-@T z7SYW!*65C|_?w24jS$_+b&(4<>r3ltbDCzBSC+zkg_TSq3OUp*vXfDYdXLhJTX+E& zvT5<*GeH|O16mk$3DuhZ>4(WFq%FARIQMlpQht z$ud>bY=6S{40uAKINkN&>Us!1ePp|;Bpy+>Uczq&&@~4$Ht5|(Qib>jvmzTKv?k{@ z9>4|t<*jqm+Q2env+{Prh5D@SrBhptV}g3;*oPgF zGwq}_Rihr}!&Zx1N5No7_smcQKy+=T>`D?E2*~T?J?lJpIWtlrKG?=ED%#wxfn9q9 zlbaBeoSc~-(OS=^+!ttUX>9=1s=bnqDA-OyhUz*Ll&7h8T{cHQ8L6gq^s8;iz!W>x zdt2yp2*>im)zogKo+?{5-F(AC?)>6btZYk!OC@8FHju`SxKJ8^p5QUfH64@EaJw2+ znF><|8iDh~ub>PLU1n+3!Ls=V-i>MKCDe}Mr=7KstZAW>wdVP4jrSj>YFb{L3({*h zT26Qie7Xj&&Y-O60DVkk5A;Z8$7);Q^(`e+p7-&%@;EQ{_7W3vK**{d*qh7@L|N!f z?fU(qt^^n2nf%F`c3`l*tO~ce<-LAY%M{{FNvt}zRgPH1CLjY`uDP^~a#dP*ppUX~ zj)~1SdTb@DT$B(h;`sW5%#c5>LJ%StvL5jlH9+A4F4&DxbYt& zi02j&fO`SZkR13c)@sdwoOrTv=e6xkicRiX)`phh=>4D%JKytXhq3z&Yucbcw*>(+ zJpJwRSvBEfvyEw)IO_3iTX8*ANx_!)ddn2r3pxqVg{w;F& z!#G!KbHfXUELiPEY@JIPmo6(>gH1|>s82i#H{WF!8dMk%n9KC=ikzRXYPE6LoJC@! zWZ8L|U}0-cF8-g>rbHM)4AGiuJm|_tw?6HTz03B$7^9>VDN~v-i459=-%!nEQ#O3Y ztaAV)7zCHmv_M}98+oQBy~4cA&FDvd2!!R&%F50fm`)w0Rokt*8irWuH==9YInl>a zu=#vid(;;mY?jCv9&6AC_E7Rj1~1DRvK>MsU6F%$Umaw7t7&JCO0WL&1L|c?U*imB zW`Tb9F`d{i@8K07mR@nG!{OWm<~tvO$gaY-Rl-Q9WCpkTkLmyP5`305Ik$g?2roi+ zLcV4AqR0l9^z?+olX(Trx3-+Lbl(QX1k2B3`Db*?WfxooIAw(Zk;y>{?vL^Zty-}bV#iKsHRFXb?vFB;ZhXAnzRe`LOV5D=n~bjc(V(ujZD3 z30fs5v8CIytmPE5nJK4N_Ax{;aq*~>q0qV#!jtJoG#<$JUL`h#t-Xb>V9Z3}Pxb7< zrh*)cEb}e^j#|bra2+_70{2Lc5m{`dtf0k z8|dp38Tu&IUwHmgKy@QA-oqItNk%ZsUcgS&;SKoR#?)yg6-J60dJ|Mtp2TJ|X3X3! z7$GTL=f=b)JD7#vx*k2FRt=rhj^)AQil-p%hK3SYaC-Vcn7qB+s?8?tr#5>XNeNkBd+z%u(}^+PE;Vu(u857P{nh zDamJPs8n50!#hi;fRMGG+8%D4SH`Fc!(~0~gVR7xhS?M|Zx%P(kKVh2yWLMw(Py?U zxzni#XcAcD)-yJ`;yM#x4}uqns$=q|5_P5Sb^(;GF}}In)v(zvMNR!%Qz@Y$R)XZ6>J6F2M)DuYPQio9LD5^+`@3BJ*} zSyrJs?T=gv5FxkPvKHD=sD{?X(kqMhSJ=`<m*3wPl&;%=Ln_}i-qFZyLH0}(uAb+{P#k)(%$vjN;?0$gTX^`zIoD=oRW zKTRo`zwy}J`=|6?mn@uU)|`Q+gzE8&Zp{gPChT~*OXp3M#P{ZsGx-HuRmOQ8!q`LE zrqpc*0c^_QA%AO=?t%A81HOlo65KGX@q+lIAJQEG6SYKya@@q<|# zc>PlaT)TgyA9FOy^L()CK!J>!WXV&Hxz4X3tKz=7H4Bx`8EFM~a(L^CE#fkQhB8}@ z$0?jHpte`F*6V3(gfG)0c)U~+9zPN8A5*srv$#yeOchNbno*hiH&^c{W zBhpq%2dhqyQ3Qz_3Ec(|XEX7?02Zvja2G62hVTlOX9rIFcpm>o2p%{8C(iL~od8uO zC|6wcZ3!MOV>rjyZHt{@hy$3oHW$8v-ja9RGj+T}`;`Brp&G%q#G@uE_8v{wq^~*( zw0Q5-gj{cqI>`QWzlt0Lot!pI05B!5A}$Qccxrd-1^+mZTTbRKv+~N|SI~`RALQ{6 z97AGTM+G4Jw{lKPNIE6c%N({iu&VTF}d>^s-^}SK$)F@RhSM*}_h1|02Dkf~eZj2K)%IC6})SUIm z#xqb#(vRi)mExH@^b`Go2^rbb0mvLST%R5N&;S5XKnHc~0h9wzUql=`*(~sHqbLh3d5?YtpW|J6r3j9&# zKCX>4E#sdZouOO?{qdvok5{j?-$_oG41Xs|L zTw}=f6iiy1lU_Lg_uE(X4vfAGpF$riGXDy4c}@bKvnK5=o$Nk$)p+Zl4sfVWM=JV( zyHn!&YslY+$$qv>55;J*^r!C*pHsnFX%Cd3H)cgMt} zm|eW$?>+lNP_FgI@_#eOuiOCCeLS8$y+-kjDln}C>L%EjRGcWYWjHCisLk^u*3vg8 z#()el_}Py!-(L@pJoNfZL<~-iOFj?(WuXgh;=c4br?_6ZF(9X+I<#-Vf46zyOZgrL zVLd+ql`vA}xua;fU@l3f%{RaWrfgHZO{9Ekx}PKz=`T1ItSd0W&J-SwelOt<@NJuj zx7{d}d}+@*&;tb1FYq11&a4JaSzoj?&JK-5pNBl{VQgXHR3o)D+c3*9d78{wbvp$( z6`KVCBwaCKQSDu~TAvn5767bfbB1pS)cH&j5&Kb9yYOX7haX!Z;;#;cgpP5%LB%Bl z5806Pc_OQZS++cE%D7YM3cNgqS0Ng%IW#^$A$?VUra$qnSx#-nDwSMh!R!B;FoX}0 zq(Lyz-TLUl_AW65CScvOP)oAS^+%oJe_m!0o$(HzooZLBR!_^#BI2^~p_mRX`sLi2 zv)xn_Qg;h@ZmO5ng~8mI8T`ox*8^Ha=j!viBjC5i-4+~6;jsH@1_8_}@Y4fRhJoOs zyco8%MYWXqT8HjZk&P$*6vI;#{>NcyERzhgi-3yI>9u!aIwi%qHtH#qGXH*4@4V{$7Q1SFIW&3WdVFRAq`3Zo zgzZi((_XkLcYfJ@OjcX3)zX7(No)IOI1)XE+=}Ysg{6iwjbwSQEslKmv+jc=+Bq!s=)YLbIc15ZuH_!+P5P zEkbS{koJx7ki)lGXO|!!6;Uz=)#0~_bQob9gvX9p>ZMLG)jg)2EaFb~wrX{(YO8ah zq1jZ}ltnm3|x4E&ShB><=mp@Lfea}ycWia-T0bIk~cEc3XaI0 zZOWIGi%iZ4PJe*xiRrmqdKHDaxokq)fGutCqdoCr^ySmA#w|UGY~GE+rYubKn<_;M zk1#QYn$)ukakv-IQ#DrGNj7$A{!I_Fv<>a;gt0kBDnbFojMsi-G<@^;S{1J?d~l-U z($V==oe`b52|Jw|*hDrHuY2rS1iNX7vYrdt14dw;Xyi!K>m#u_3FK z;xAY6w?vbdp+Dijb{qZpf90GKf?01q7r?&H?yAHm5#r)xM6x2dAKNiSo(wLiWBGT1g4P==ysvWYmtoy;l^#JZCdY zwB)}Oi*}q_C&m;J%i2@>!S)v6krFrSY2PQ{NmkCfF5y~})2FgERVy`x)UgX-u(u@O zmD7JvzU>zuWgc1FMePmq<$()bq>q&m?Mr39191%sq zB9L)41%4%`Z;oA|wXbEyjlk|5uRRfx-!~WYaBy~^i~8!ze*o@U5;JN)%tVZOvU1mL zN!Gpt9r)whZU6k?iyF)wtR6Y8*S>#MvB+jUw+W5j;qok65GaNEm9#GOs%IdERRZ`Us*pe|oDX?nY@7 z6GZTv#>`)!+5dpAnI0Gp(Q$5h5kB2Id06Z#=tD}U{>m4Qt$gtVr=xd8_A?cWmM?7# z7Y$x;ZE*UtdOf6;u6!fB872?FmU#h9za7cH4k1mxeaK&%He?@wAlP(&B=PZ{JjL`} z9Z0m~NB3j5oaH*&tMi?6b+In}lh+YWiZSFJd>Ra+c5A+D&A|C%t0%wsV!tt9Zjs%D zU#-5pv~ypoZL*iO{i5O0>20OYjFM$JSuJv{dvAa|?rYOt2;=>&djWe_6U}x09OeE0 z@(n%-PpgxC+~K5?oO=V(yI&IH5^fdrtQT&c-?P#6W@y_U7Z5u)bs9*9ho0_Rcl*YX z_t#OUXFpaf**$+Zc+T)-(R8Z7W50RQnnZyI7%&nGS=H5`F{9x}RT!mt|M<-)JrKE} zI40p;?B1Y}V+ zM``FjH0CHkg9e?5{f3XY(Q-i2S9S2!Ke3@1C^?wgo*VI0pTD?g-)#TeZiAf%px7z? z5_HEnDCzr52VQOP0P)qmlbfI+)xvE7RLe_~JsHurb45MwS{^(C%U5gqyo_2Jr=J~! ze;@Go%ZZx5cB35#6{OFu#{ihZnn^pLxd**DYO;c;`mi9kEcmE2@NMj0KZhKMI3g?6 z>By-+y`-EjYz1>kDd^)&a?~T7hpmQv$_S}qik4~Ed8n6^u{Y%o)DjH=*vGG+oc7)4 z^HKTGy>rm7plxP>ZZ)h3{Sa81XwW9T0g(o3G#n}(jh4>2>*`nuA>8rxsHCJ6KT=0S zQXW2@mYxnz^Z;CEPS{->_l1vfieVV$4+LtAJ~Py&)-G_Wbi$x+$idpbwyV2509 zem?#Dcn0Rq&r>cvhjxy7!+K}4!AKM)QNJU2^zI8y*dL%9!LrB$z|G>|BIq64k$N)hAQN8R&pG6JUzp)tl z+S3r;vwg^or4KuC^l zwJ3L5YWI9zWZt_>pRkxCLtC#jQ(}VVWYl=7KD~4!Yn8ZSvkf|s+_RB04v{RrdT5K0 zom#aqXgJ5l7~fKPM6vQU(k^kpssPq&N#Ed07A766<;W;j9h1Wd#*t}CZ96f*isVgi zOTA7ZUqFyl&&bQ+{eqA=tV8R4Q8o8n?F@IE!d35Q@@2QumvKJ{!)T?nV~IK>=WWAz z%CPM0m%kDj1H%}jtvjorW_oJ4_8Y3{{Pk0-;PQa9OgACs1WdA>TJ>;1t*9`Aw25;A z;~1nxy*!L!or2C&7>mav=JD#HPh}do=nP-u*gQfZtV|(Br0^Il8HiL&&&~m#%Q*GaOYh^Ho^z}X>|Jtjcmr`^U;Hqz3@%o-ZhSZICaqk~ zcN40J$NYmyhmG z&bWdc0ih<{Nt${X8`#Dh|1)gj`Me6?iJ>fdEKEX#3%0$^#qXX`Av~3oiq+k%IM%%P zoMd5GG_*O`dhUPecc-NH;}aXpnu|_eSxba6(Wxs?(_)Wznfi#-IbNV%24Xxoj@ z-j%EkjcPJHJ5t72SMR$l_;?>|@Z>EMR?p6g-7j`4t)&h7b_yZ5FqF~JyS-vgHOGB> zm)5V62M=|)`l}rG>fL=U?po@5j)}ZYKS3#&YH`1jO-ZOANhMH(sO2#i<69jZ zFWTDJ2+_b@WJ*QfedwTL^lI}+2vUP~)#PgjH=GH=6M(8jE1p}t6{BbXM=OJbbR z$qtyNPI89Ldvz^7hul=C5!x9USh)zj5ax-os*-`GS_}6>^f;%-4Ar*G*wWLzXbfVD z;hNJe9K-1r6@2DZUBT;fPiDK|Z2&y?eny7&WV8lZk+NOm?XEM~1bEyF%7D9cQ^} z?dQx|euDyl{21?or@hQQQi50}2EUdEW3QdY4o%c5X9^1uvNYkk{$D)W5%p!~N|1A# z=BW+W)ZL1c7l$$vn4eS;k7*1b^ry1VC}$XAvZI8v&{3EBKP;TkBZf??F=tjk2)VdP z;Pes?a!MykwP5hhPws^g{lj#6tC8&JKm?m#3H6JvhYLgXQGwcz=#qKB9uK zbXIzoE`7Lq%U$M+;q~nFmq!sP&d5xF=k(DH+4|3v!#z!n4`!zaT`;{zGxB7nMs1QZ zu+D#|_FRI@yDfd1a)~9r9WE6k2rUPuMzudJY_Um0UPcH!Qny;rMV*vr3coYi6ASey zmng>?I4qgKxVuZ~>^uCb{9}sOu6m~@IeHeM7|=FN*4rjwj+4e-w_d7@(IB$n9{$&) z^O1xvE?B~z?2&vio8E0}Z~GVvs#Q$mT^Y2HvfP-(xLpArNig*7P+3%yJaM+Bz=UF^ zOHWN-K-=q4>5BF5!{Mp}3jP_{mQB<0Y{?T$g3&WzoSF|ImdG%bM!`FBz;CauqOCtC z(%D^`wfHdMZJ1?yqVj0x1J^~3Te!lHk4f&b@7f$490fHsv-muf0W}`F(5JCIBi*>M+st)c zCYe2{Yi_T=nmRM7e>OQEvCtpnT0(Gr(j@w~@aN}PT{2fs&gyX1j{I|qd{M0mEyekG zN6Eklp!8Y@B)ta+ao~2zY(O?Je%`xNJW&@Fq_1yR%)5qC!ut*$yH}GUI3e&NNERtX z8$%Gh>)sja`O&mf)k~i*G1Z1xR`Pav0e{BOuFS8~DzvpjsSj$8rrN+~;r%v?)?R|d zj>7CQt1v!CqyX<*d43nnS$C5sIEqwBDP=AuxClDMc)R~lqRdIVoia z%Qo&@Th^GCty1DbNlIiCdM-G-wumsd65DrCk@YFJ1LKb@?+VEEzL+@eUe{OPhDH!F zX%s&!s4F^_Xe|?7r#T&RA|mE)nMII`7>F6%xVHE5ybWugVi$%w3=}_}R-ZM6=}u`( zS@ilQ>EHPzztFt9Wp21hcsvHjmSHim#TvSs)Xwsl8MOwt^^Tf?n?>O!gI>#A0b&;v z`2%bsDDR#fVZH-ASPuP+d_!t~qExud$`DuN*9-R=Jdd@l`~x(OS^4WX=nrTgJ{%TV zYcGJ0C6ww>U)zsu;NuL{TOoHG4;{7w9ry#7=HS~oL5zfgtR4QB#kjw&TlqWc19XH8 z^!>S=`H4-wf6u1Sd1k9>@r9HdT_d&-LImo<V65^4BRSQb?XYdL3kS@Whz1MmE;z-c+X1uvtd4^D)K!*Hl_OK2|zb(r`P)}K(F<8 z!<v;(>}cC2?qpHg%igGdLchsk}tZ?C&CfZP3fBf zYau@ySO2X;8*#AzoE~VL8X)adZ_9@nl?pq~oUL8B4Ag=hmlKBZr1#+CqBjP@D2Jj> zbY|UCbKP`Wy*UXw5BRRYcEqutdzLE`@-b1H!Bd`K*$ON=2%f!@(=JwtmYLrmwtGcLR_I z%Z3o>?#GHQI z7p+~*m$T}SUawjw!b?4>27T4^w;J29hxN}*$6kPbQ;rZEJRzBGvo7F%Lfa$U(<1l} zn->YNqV1@%7gI!T8lw13R8uLLHHZC}{un{>|3%J&@Gl_9P-^(Bid z3lBjr_J5>P^=g6hS9ff%kjlhn*Jg{~cGaM-0P=_jNykrz=KmP12Utu3g$Y>+)-W zdhGS>d7#XA;Z#(@%W?ZB!JS*uDg#1kc@L)bOt1xx`jy5$c!!|vU^^cDm(!Y~X*mc^ z>~X_>qbG(0qWj*>l=+-Ifv$caFMgSX?+d@jT3GS;jQ%1-{%!iix^_?0%sh}>4>Z5r zThs33?a&jMR!Av`55wnTdu~6tOQ#{W3~8zq%7`N)G7Q2c9QZ@2IP>)gRxOe^p|iSc z0Osm%q?i(ov9c`-Hj0%!2^PDNQO4*3W|{6ao1RL;M2jRx2Qc;q1Qx(f)sBctz)$rh zV)ChjCU`7So6nJa;wd_lPnMm-Ynq*omR}ZWD-HQz?p;jXA{8(*TcxK5`^r_>rTSjW z6_G2Yp=7*OnQ(-QIb)GTRt4=<{it>;2$U0GMV=~Ry)~vkVZ;*!CLql7N-5VD=&?Bc znxid>aN6V;q0w>dBkvV~36H{wFQpD-^}g$lkuP-{H7OnIN|#chGuGR5H_8F!?~T^+ z>=$N1JYN1|=>?nvE2s|Pt)iM~1?y94Ax~)iMY4NjTLO+LPX=8z`9JKvcT`hp`!iy41cu%OMg{3%APJ!*C`b#05;_RXFoM#hOP7{Pf|O7~8&N>OfIw)0fJhGn zq$5R79B1Bf-ru*@_pS5KIqR&GKX$TLp6A)wN%p<(a^2U}TkUChDaH#&=*$vE|75MeosYTI&-ka}S3 z0a$HuPB;gQ9Fcj_OU_HIqz4e{#30~U#rPi00XXMQ5+}Ouowc2HpM1U0~j`U$<7t%cmGv+;V-q}N89-M-zB(%v(rm`?n4X*Ob##{nxPl`i(y+X6z-??1>p zf%^bITKwkOn`~~&!Ez$wZsC0{jhhiOl5*uH+v4NW;V3=?bf{lXQ(C9vre_-8tmB9- z7_N^kc{`vdZib3<@E=yD%0TtToXVRdMS~kxUxGxv*>1o~-4^eqmrl31@5h&y|1_w? zM_ba7@?h>zb&uAZ(#jZ<88jdNWV&Ysncv-lpevC)btYcDGZ5gstlXSCA z0mD|EX>ERZJ|BA0e-ueer=Pekz45WH?~m6HetAtDl*7@Ut`aUQG`)ov5m*yCBNoVp`TsmHWAwUjk4DGB=tFq>tiOE|> zhQb=vB#f6W*Q6%aE|OX|;uQCe6}%W|e++YZkX|DAslTx7O0;!8SJP04TrZtIX=;aM zCE~G|#KQO`T=ESCwU%XobE2zs2|6Y#OFhs!E=FAB5}To1N5}NNZV~S^bJua^eu&e+f>E|i`krkXPcLc1 zXV7>dU5SN|4UU~^yF0s;C_pa0lodk*E}zUe~5l@ydbAi|LW`mNwKw3p?mfZZ1k zL{;<4r$JAUWT$L0GyAA+!scn_b%08+<)~B<^u8IW2~mi-plP825zFc|a;+(MC@y9* zhP4LQS+{-E>0Gswytu+cb~1z_)Zk3%d>P)4j_Y|rdj$_*YUIY_L}pPieW7?m^G>fq zEp{fLi!uW~NRwV{itXP<+8_rYv}6OnYO;+H0Jv4e4QM(g!<)5!X98Pl(RjxtuRpjj zsyfD3=b?Fxc$MDdO{*acg@fJ5f)aNtxx{TS77#@!)3f59ZAjWsHa9xf2Wr?Eg z2J`W1CCfD$Y3Y%)0bac#W(GQh4LO6D4B{to6S7DC7)yTRCgq5``WMV8(6c#O>VPGY zg{l2G5q=)JM>kc{Ve#UpuCx7wugncv4u2v+exKhM5Bzs4?CF2F5CEuulBj1NHuLny zGS!b`&pyA$uDytUy9} z#70{lv+c_EG5;D+eD&Nll}(8IerG-d!}du3djILKHy=&>J(un6{`LNCjy3N8)!K*O z{NG&KRd^(UEhUTO2maAU^Xx$VM(pCg48_FZqT%faX;oz{w9mmjox3(d6Ynmvi$`jb zDW@OKa~%1V{Vg5hxTU8M;Bl~S^o3(E{Ed%QU+!r0!#=jb%3l*0bo=lrj?He(QUA@W zWsm>*P|tXC-2BrI4!cJy&e^l%KEkx0Jt~cH|9!aL_#v<9>@}&M#z$A@IgbAN=V+cG zU(AcV;_YLh4!}!rXP<4ukdsVPFXutU$;erk$%Twhf^8hfoxJ`W{P*U##rB2ZuBL~g z`4*3VzSeW_onQQ*XQ4jv1IG^u-t8|tI8$$QB9mc z%qa3aK*r2GH%I%oWsdt?KeFKo|Jw(r2RgVN{Ab<=Mp)A^~NdnEHA17W=-pL+K6`)?j&2lmHIMuR#e@XThHzWq6NpzCK> zy*EDdYP0mx?(dMw&2Qt*xukjdoG(WD2vZ7hs_Ea|694}KYPI>!dAR& zcR2m-AH&K2+0leT_U>lPX|qJ`xf}dvM~HwwC^A~l-%5I+ESmTCPU3$)<}m+x)4z9& zD(lW}EA?B!3)Ay1C8f{S@x2Y3l9>df{SXmNJ$qeB?^6v}RGp>? zuzd7O+EVz4jp4qu>E@HWFzH0k`BNy+b+!ddl~d{di{+XPKk-rVUzt#BK5*WDtduLs zsNbf_e&JwaX0K75nq;4~Qa3nG*M0xn9>zWV)lc@Rxb}0Kt6MVtSC007dSdtn{YEpk zs{N(!XZ!P&oGdn~nd4g+c?cHpz0C^H{@$QDB?x7*@yqu({GGo)0tbhm=Xc>dTa0bY zY*FUZ$K+6JL5?Wn&)Kz~uX0?U|29Q(f072Tm4WYP^niA6eizyU_qEf*r^AFe4lz>y zwfZ6Cw&<=bXASe~G4xOWTKhQuzYJEm3hTdY+F6e&=<%5?4pqf(GmmmJFiQc59Y{0eTFCc- z{fphN!iJ*H7KDI)S=0Z6 zH|yre*DyZnZ_=z`j%(ka(hqB(B8?~E{LDlz%zBFa+?jIvWDWOukeq|;N*Ah&SiGfLr)uapvOR*!GirF`z+YaadX zLjAF6FDCTiMA#Y6!2+OZ22d8eC{%Ra0o!?L^w&dIzs_q&%s_z_g09F+Klj3Xcb!@a z_J&c_GbYaLqIo`5p3Msj2XkEfSAk3{sB;{;%03du&p(>IWqjvp_+tLKg0ps-9C%gUamY-69r^d;Y{}Xo zTGP+ZKL7Tery-pE&-MFOrFuV@0Hj^%EI@ij1bHMLM^ZuGJ*Sz5=4>OgV86BO@N$$!iWxm3f0sXwV(jmH15h;>y z;hGiSCPGakSbn5I1(ydRyNyvc*NxAenAZsrpo)fSCdfoxik=)@E+jNM?#|Sk zfQqGZ{ghimvbMNSK6ARr)tX;#YzryWR~_6Lqkd68Y4DvfemNR4ipQm~=_kRR&ZlBx zqO$AC3@tZ3>haHMz9(UKIGYU8#gdj@^-^_g%gurSn~B;sH)S?T5BYcDLh|g>R!QBS zGGnM3NvJsQspkC;oK=hH9+LSlq~H6=)h@${8`fasvIQRR`nbli+UWEz9LFq~{*k}U z>~xH`ZiUM<(JDZu=2iaqOS{3lsqKN?=)$ryB;J@c9rM9Ox+wkBy#_Zw;7e)0E@#Zh zCaf<)X?_A_HB_7;z(dtV0WfW2o7axeH!(V+vKJ~{Y(_~Jc%J%pW5OUbd;=E=wZ8bi z0VH7fmPqOIkO#hddK2FSQP+0h7w*H|L7F`8m4G&sP=xYOOx@e+(vIlO~?vI3)R1S@j!6=N|~cQgNwv z*y3vHUpUx^*TqD3Z-}&Mecqe!oydD%IHKhlb?$6Y?HU&g&N+Z^criu^K#DlGe2uO1 z)}mH5f6XKp!J{v%Yvem-uDCyOjrVtuYt9~}pHvX6PYV$kl6w;~ByoW)?VcyHH1H=} zA@JoUQu*<5FD@8n@s!}{Vm$e%uPIi4Y)}{2UIbGZt7Qv#dk;2ql@4$9UH6Z^1yfYy zwn1Q>^f4t_IzIH6Og%>~RoM$`1zH+gv8fpb^OtX>vZMf}aaRDi$QDc|l&;O@&u3ul z6yxYl6)}1AAt|Z+24a%go1DP=DwU}%Wk;lmg^U|NIi#cHbBo2E%5(g1xEfh>r!p3H z)VDfw86JW zgi8Bmtw**B`lvv5jPXh#-tDKJlw>yKVs8w&w!Ni1ID>KN{PF46tsme}!cAke;c`zE zU@>Zlq%LHlU}b7#aK5pI7qF~xA|^KOK3`pnOmJck(ndG`j@c-db&*y#A=B@xsJIMa zFimFKc|H(zOfRdGZJgb%@+|jAZ51xK*^yYvPh3Y$`#`(!{WlfPu{6Kc{5Ow=~ z+W~^jYQ_d6C6s>?qmQrsRyjQV=2L)}Rz}j=vf=Sl0lItNvMR zQ}+&4UEjyOKZJ|>>l1%_I>h|1(%u#7-hWm9US-z~y!;gU>sijfrhJ#xfVX}ZEVyfZ zZT;&IKl4Zzez&KK^WL9rj6eMRXUSXQXAV9i>Hh+ceK_`Sw71vn!szdm%Ab$ucc2m7$GLk$80;P-5C>XfzR$HSjVH0!oz3(KP7S`byhrK&WA z8$HEEc-iq7_5MlSRVXZp8Gi*^wTY0ERmR~Zi>>L$2vZJeEQni@dS%C0n#O7MrhJPs zztR-TOxrHh?v;LCGS5|C$@3yf2ZjyhgZ`hI2#+&T2NK{R;>B3Y=?70uPk$t$q~Pts zTWrD<*&|EZhK2}2{6lb6uL%fhCajcHDHAe2!(4>GZ=OEx0qz~&S_~}g2QNaEbd9~0 z)qQ&ml&8#gbQZ^TJ9knmZ_rVk(XRfC^Hi+@HYm&f7^uI=(j}{bHx?8jybkvOPd-w& z-4yVCHJ-)red^v-_gOlB&{TjD^+}=-aNoYLm=mhhwJN@7ol8j89>WDx(8-Y#{UJoL z@&W&Wh|EpdF}FNYCmQ$4{S*)EE}Ez` z$z_n&Wr&pnzf=dG@=cSRO*s#2G(q_Ln9RJ7eI#Nx%kq?(QOtRHZ#ldeXM^mrv~weH zjaEQ!ac3J3`AwR|WaCELJMUKaRH`Y8ZYXW-MZnxu#8=D)*+xE|HRKTS@k2}wkH!gK zdcZHY{5D87f*l#02d~lA&KubyZCr9`Yim2P?Z{v1O}5otSQ0@N)$4uLx{pI-vY9-MI z_^e?d&g2TC8$gQ=%R%%9fDBL2+=&K-xNBb9YP46_Eo7;`JG}~BYw?;HbF-d*^zB{J z&6vTt*La)cuEb*R-Hy_t%I%x|z;ryr0E=MqHS#aZmG}0~sdnD4Wvs3m`z0YO-Ra1q z-t12BIvIbX6qR{IFSpEeJ*oZ7q?s8gqJWpsx0{^4QEgSKa%_x`eYR{yo}9Y&GZGh; z!0&At@1$*xwN@!7)aehw3+#=gR4V4_aTaB(D3yaWsbwpcRz=2OQvBnE(+Kkcf92uba*vgs?_m58aix#P%KEgiDC zF<=dhVoOSK){QP+b#IETDJqGM!Q~EsO|L|D1SceSiIV*7qS4OYSE0oYxH2b6D()5QRk!fAs)xspw zjHW_(SZ@JMN)KY>T7=1zz$PUmsH`ZNU5aVgS+zyIwBJ@6yH2Wy(IIVPvy}A0lEi4% z^u~gy$@Hp>!fH;%=8c5fy-r6cljAJs)j!SJp=eb(pWz3eQvL`y^YQTS%5}O>uTeOr z@Oeod9_`jZOt@rO7Bcksh>m#b-q}Wt2gwXU3@Ms$o_t;%l8KEqeI-?w16ZG}y8%0A zGOr1(<}^VRI>HN!%F@+U2kN^Kz9Da(Wd!R$`Gri)`|ui`{BVtp56ng61g^F4 z>(Ki~s_!HFg2iJ(cX&SWDM^#u0_Ay|{1Lr&N>qncDo2chQ$?E?ED zPtD~NM5&CVIN*>J#Oo1!LJ#6q77-DZyU7DNuYCQ!eLzo0ozhJNjPW*j0y5?wEtFfw zaPt`K+zp}xBBrg9YCQsyU}-vRz(n1&1>f3>Q)q3}va`&YIXi=3#9(XQT&gO?EAIQ?aDY5+0+(ZDc6O z06V+B36DYOGgsjig47k+E#IX@=7$+3aDtZ`S)++BPA9+Pv!8_6+&LZ0AKzE7*Ea|V zG|{`SF}5B6|0#z%&bTZ+ef3p)hwizl4<~0j6o%1hv6d|H0Y^N75^uYJH&dT-emJ}T zkb6qq23I4gVwaFQ(o{R3=xXzzpX{d?`@VZgkDr~BvBfp6dMQ!L445WT=AEH;oCq~q zrj4bgD|UM_#)bPOL^8vhxI!PS2hRJ$-d9=W&s#pP(%I;&i6EpacFRAzyQXtn47Zg+ zWGaGp`N6lDk-pRZViE44aXxp8pe}s_{VZB-O?_kZ_6w!`fh8ClN)Z2U^iIu)C(TQWJiWiO_L}B z?++-2V~k<0o3n$5}#7T*WrlA&y3dk0S%5lv<2J z=@IBClJ;5lA5u+AjJdid>d$z_`eEkQ(a>W_bu@F@-Q+t`eAFQ_=sIo!1=fa@Adj*8 zgC@!S%u~TSxN4p9QtIz0SKEexMXFk$q4wVI?2MMVWA#29)kc3yN&j}1c~m)RD8DeVwiKej1^8OSj3-Au^rqshCPS&0nbv9YgkSVE$N>xq%A=hy*QUv>7u zodB4PztyFQx6XGr+`GI-Gwov7wV)g`>vcIF(4K54N!^^u1cho%<(RNA?!-{ae zi9{RIhWoqK*}C96(Be=8J#^~aK-C6p@qUdB(~W6YcT`a(3+k_gOWQpoJ)n<-b(Q(E zFp8&n!x(H%4EHrW)_mAFv0w zhUoiud}eMNwXVBZ-|lrG*iemzKM_iFEDAL=h{kB{7~#(y3!|_~_Zs4j3~4Ke@|Olq z!2`t`AdJx!^GW$SpxrgPX|Pva80?M>gh%kqW}1xtBqqMEb2)^FaD-0lN=n`Bt;UfY zu#|aCY3G;@0aC){Wyq;yAg(9MjQf-GW%KqC#%#W=dpE-itCL3FFygWCq3&4d=L2Lr zEmAd<>0!USZ|tcL3^KDdN~|27*Xg$68}*j}4_$D1fs#rpfb9TllP$YaKV9aOS2=7Yqpr3S=TGS#i60Nva_$&-tBZ)McdN(g9CYf7h#h&S{>k`c|(^w*@ z@5Xy{DJMRpxT!Nv6HA}kC><3wQn&)#Z&l#F4AzCTOq!u$BZ05(RA?VKRwwSAoKtkG z3i3^5al_ccibUj^F@9hHK<}Kd19)^ z9Q7S)ZDz;aJt3wi<+0ezRoT=nxN;=XPv#vdOT#(!B6idTMqi9?g_sHE+$RD{XVeG0 z@X*HIr%0K6-M(OQgDW1d0&RcSdh7IvtbRO1x3kcZOCkBezDr%0qo2AOAr5dq=rUBo z74Bwqk4x{8&uxft=@>e z5;f~vde_tMGs-cz!a4*xm`1E0|5rBE1D8m}c1XdbGm^XyC=` zamkGPAgPzoJTC4J!h1B+I!fHQ-ynU1e7bBmMR!EJN(ak)wg7-QM#c`-qW1AIxg|Ad zvNVZ(5zTK(i;hJ46eDLO)Wr=_0HO-61^H>a83>eg#@?nnZ zv=n;Tq9K8JuRGiQHabJ;jdq^>8n^q-M0yw!i_6IPE+J6fu6Uq?^~ZLlJ9=YYFM@Mv z@fjyk8~sWeMvM_KHSUiT8EjBfG<=xVtwvTc^ACH=lAG(HJ95p| zkte>(%3vhAFuXLsM6*;mn&TPd9~@7=4*wG(v+=d`^NT8oB0GXH-;kQ)BNQBj??W1YYFV;tox)Z_xY;M2mc z*4lgKa(p|~)f3leJy;V>uHm~qpU%3egX#lTvg|%re!i1MeP|~an?29Li~9al*f5i? znHb~npQ~>z!O&W#G3!63C7C=8;dM=nSlE9p=EOf5rpdwa?e51W3rP8!ETE(Q>O1qr z1JJ@9Jl;ie_p+!TNRo1wOl{+DHBlANLxw(HqJXSH59~Xs$lZQbbXM$<#~kJXKu(nbN$RAi4BClvrOBnqj9T z2v)cw>gnTs1JMfBZS%sU1^T>)4H3)eWQ)uO%j#(Rgh{KjyVi>?!%1%v^#Ce>rekU| z@-aDbgF4kB-tVeysK^zo3d2k+ZqNoPai{S@zxYF$(zO4ZzJX7q(hG$zV0bb)X2 zY7kHY_sO??!}SGZno7D+XjaUeThcJD$oj6Fy3!m4o7t`K67?v^?zQ!OZ6#pD@N!CT zdZPxG$c7|y+gDED4!ABLjg{3@0fiFg05x7gD|3u={1`pc!JATs9sVTh?oI9qX}Cz{ zfjoTjm_q3adX6!S8PyF{l$jB&>Tw8z81p7{hsuuRwJ@rlv1pr=hz0FicR!6b40G8M zJ+o$Pz2PTQ73rpw!%+1D9xY;e4C?_dNtl2M6?Gr8C-K-kno4MYP?ol@+wfo+=d+2% zk)?u3r|m(mDlkt5;wL|W|)?V=Qf=eZri%oWcwOOxAfcT z@qM=4SYe-l&KS`A2N1MJ5wEw{+s2=PF?C!1%>l3S_~)Mi!tWR$fzbeX~*%46~%y4C6k zFc|YD=M1#h;uq6*L*_K|u1_gCr2%hBMsQm%Zxu_^tQ{W6_4oavciPU(D?Lq6u@B&F ztf^Xtz{|!)oR`jLUfvpb?k62<4KnYsb4PFJinXXFB|4-DU>j703(0C;;O<`%HHZtI zCq1f1zt3v2VO*Ctc^~#@8{6bF*c}eeJ=J;t4`n$)%Q3yd`If9gM5vl;R-Ho)s5sg2 z@`bRADtR+Hx~t9-K&jH4I`pc~M_Fp=5`d-WUO}5bP+oxX{1A+IgQw^b_0ABiw;h9b z6sE5SeMmolsQu`A?vfddtWKif zD=aB?_VssZ%Ja)$e?<&B2|J$)Rj`K^u7MHT!>f31_uM??4bpESgS%J2E|l7ZO&+wh zR?3RGaM4tXz{f3oVOeaZz8cHl(PXPr~4bQxU_5T`IUJxrhMWj zYGBDWeP`~m>z_G`g6G?eus2pPn4|zJGyXpPw23T|r`W2@i6r^1KKIHOSIo-S^oOV1 z^0om=YV1s4&5zPe7FC2+%=yXo(1C|1RaLm?!!BV{9 zR~&ce^p#^+Sa+PB?)9ksSWDoQ z$fA~Fbc)yX&8}ri+V--!hsQWxVX+398(`^{TtRj#4N;SUM zB)`f7xyz62I0M*CPOfav*wLGEBZJ?bJTm($ZFjjMI8;-Qe-y}Az~mkX+<;6E`18hW zNeETi5v!;@`Uv&dqZ98w$6?$c2S>75ud)(=dnaj)k3xfi4J0Lhp!N_beJ3Q+89BO&uo8 zTR&7HH@ItMx2wjwqWXp7dF985FC06a%uTe6f4qZuYUzeHT8>tc86fsHWEtEZngH9r z>9r)gw<*?RZ+-G@#XH*-zT~DGgX6gul0jxF>(+!+V?Y1(0CRyM2*?a#CFZEBQ$g$8 zOj_6MoYU-VI$!2t=siZXOA)Xi`NGlL@A9VBPQUpz?-!1?AWxgaLtELpj(x-%O;N>> z>n3%!_{xEw|i{;KHfH|VcGcB zn73B$iRI1R>j&fN8zKB7VlT}C>ui?+Z&&BiW*haPUDAtgfm6ic5?o29x zRNnR);*(R`BJ58J;-=bEaJrWGW-)CTnw**PX*KHdF8*+Msw^I_$9n@}w3@p!K>r+Spv8^^t0>aa#Pk;H+=aE_+ zU@=fX_tA_yekws3+4H|E5$UNIdHJ&|L{9&y5UC7Ld=p(175$DMU7-q|<6J3-M;wS~ z@*I_3U8*d8Yj6*1ZzSM7qfAx`G$uxqnk&9+LM$eO!M@kWlQ z>?t9DIdVJh&%Uze#r6>U&u@$S9Wk;QnwnwgCbnQ2VIi+9Mz37-RT>Wugja30cyD=+ zEUI&NV!3C@;)o=l5qBBify+-)KL2OLE=SCxBs$^4B_xHL& zmQetumnt~WQ(XBA*?Pe;rkP$h{2@9AFT6H68u5&5D}_|KB%@64z%t*X)FnTJEe=RDHi!ro;YCqzmPPW`-H z#;Yl(F>no!FveG&&FXLW6gCt@_G!nu#6{VsXT=8jdVX=qt!Iv+^N9Llp`8{+{Qbdd z+2bF5g1>M?U|hZWk$@hozn~U&aMwJ!I(|r}1Xol7++3cFHWO&rS?dn;xkgrL%oXq5 z{G2gVjn)-xVF|YfIzS#of-WKO-uA+)y1>()^o{J_shjw7LooAOzj^p27*#-mk>ayc zfi-(dErp(5pKm6N1TN$Sg;<0*k1S^t_YSh2&#O)Q#WXJPFBC&iEpyKN_=ccDvUNkP zACF1is9j)P|ERQY@}LD+8+FOSxUSo7-&yL9>aC7a-xB2>mW_O`l4v6)UAG;CLohpzFpf?Fh3;SW^>Z!WcV zPsUTr#w(YO8+7qIM$A;5F1-XstjNbGeb#RXv8`p^$zwXD7ZzGEMSW8Yx3%=@P$U!y z0-ciSLiZ?#_|}4LN&J3f2!y8jAj{Q4%xA2T&;*^Q$`$+P^4p1NmA&W%wyQ0nl+rjE zx)5U-UO#9ui3fD*C$>t6OV5?|$vqcoEP}!1md?Onila_q7N!>@&7pwh<#pSSEf)tW zDq=k}6lmPzEHIc-*nZB4CsZ+n0Pv`z_P}|)so+kGHJ?pe_m*R+lweSA{hey7`#5;A zk)NB_3&1KEA(b_4yy>mO1Pyu7`1l=dRO=UVMM4wJn5Vle)U#@vHqT2z8X)pdp2XB5 z03-;+kO8l^x$M2X>$9B^vE|Pz;k}^Ks?%kjjBPdbcp2U0hY>AE{krBRCyD` zOJkGgZI$Lrj)jWjq#k1;6o`QZp>&#w7^5!zR5Y?|fL2 zN>`#DC|eV}!vWW;j?dUv>qcCxIHZGnv;J`tm()w-B14Cfg|sMR!5 z6Q%Raz^5TLZ%i#V60T}{LDc4L zfE<2K+)8I1K}vKQ<*IaF%bXu2PHOV{S(IJyJj(thS%a8pJoFzjxVc2PERu<987ZI@ z>U%Dws2K6t_{RI5jFz1qCpLRr)mQeoz@Z(o@`~zTjB%O#|D;w9r@F&y$!tiIAN13_ zpEr0y`CQR*Kvz5?L(s%CebStX`%qBdbSWJg2)tY91(_}R*we-`UR}$AzRwPy-!5*p zQfl5=%4cK5dKV4CD%`?XN3>qmu+d{-IpKw$-zBTD37j2rbwB!!ySjhjnBV++!u?fk zA5ttf@*@Pk84o^ZGIA~blSrmWFnzlHg1M|>r9qA|tn@%+(V~ClQWz19LD0h%@V-i2 zrgXP^g;P=W%knf69 zFEPtCYJu|{qMw?qmalhKEG@o8e-i>#<1UAow5~;*CgxR){6h0p^3pn9hyi6`n+;fp z*DK*wrOFzW_0><`ZEH#eH0nW3K~fA;>5KcOk!2o&k#M8bPKT<*(WfXUfXR=IoxpG8uR(CUdG)RmQ}re7L%7@ZuvJjozKtv_E3?PUGF8 zpqQdv>Y!Z1*z)CoC(VbnAVOR-)h$c}U$C>9zm2;hYu2NU*Ha^_q#}hSN?PiBBQ_eL* zmzh){_{}|DD#k(OSlwp40-#d_8m<);`TS&hk5`2q#9(k~)MdAtp(NBeNb~6}2F|-6 zHy9UCMhUC^YFs?lvF#_{X1QnFW#^r9L9TAOx=JsXtb(EKe8rx6br8_mRhbJpKX~|4 zh3|{e;~qQmE#4g?h>nDnVG76tjqpXkHld{QvRyoO8vb^Y_mXah$VgknfIkbQRXUi} z7DDc}R6o`aa#FXD3taW!;yvAU;z{K5qn$KkQRqd^Lv|jcjw^5juaZ!)mZWQr3YODJ zgeY+cG)klAgM-VcRxKj*xvR91kyi;Os$JVa(Oj-`0Tjf`34WhL>AK2A$-lH;tn2HE z@Ug)2Y`v=DVH5Y)XoQA_D!Jho(@V-CRFwE)o;-ssHwo&-onQ>6l!bt-vmb2lg_Yw6 zMipt^mIMNdoSOFIN`EK54x;lR2b#d6w#F_^DpfKHY?U036%J8moHrUrH#U^3OXcu* z?{Tn>v8>%u2(aS)(FT0wN%e=7rMn#X_kUgE|F6RF;S0qfW;fAq)q_4A{KZxGA5-*Q zgOBdX?^}Ny`-ZnZ|B4ZlkK#_wyuhawoS?N(Hb6@@dca&c+ ziqHLZn7$N1C^a{7tF$Fd#RohHl@?7FfwmEK+mpP4TqOaEfKfd?BTZemm`Ou`8Z2M!!&0YAXGje3xPgguLi4LF`Jig|{ zlN|lA0}Q0(a9fD9KY>Vxwj!!6hC28I^61t9okZg)K)f#awwd_&PoAK=i##*M7%1`jF;=goS}A`Qj9p(-HtLpvWAECT|Q|O9)V}Ie@t6Y6m`adswi=}N47h$cM#qE z;{`IO@uO0unTfs8LOykUDv2TOm)pGCF9R6K?&y3VOPW$+Jc0F~BZ93mPgQ`egeQQT zcw8z9eH>Yf>Re5=k7hW_d-a+N@Xudmy#k^%|>xEdRoSfwz-+Mld9re^7s;)7omqt zs!Z;i*BGKfqu5nVu)D1mY)~bS?c32N3OUoL zLK0LOD4-T6pT3My^Ug}rruPWVP=A=`3KmLN48}+=Cz#G9{{}b8YuO$V8*C?4gFqFr zo1Ged4^cM;W1=(8pjJFZ0ym2)p3M^B$CIgJEWA}9f@wVpJDSeB0*1)r1W55}%Y}va zAquVD5#E)T@s9C5WA4>oI8ui#sM%&adO8DBJ%T?tf4XoI5uoQ5>w%+HSrD~0LX$g= z7c5MB1ULp9@ku!V-z{V_QJG3_KN#?*CY!Mi)ljUlbMB_;NlAbk$RsoF<{ln)V+G)9 z3o`_&y%`#IK0Y0Rl)(Gj$aE>CQKip4=#s9k8@?;N;wkJHD3Syj`&9UTUBfX;#0Z!H zgI2jgIwN?-k(^|%^#OplelJ&dTaawuqYy`at@6b#>CU9ALI9`Xki>H=2B9oznFH~m zj?9d&0ueDrzWy&s%0{nbvv!+$5#=`12D1#D-i`TzPL*U~CGYlx;1_q{#B6I1>caB! zo4m_cibe(CkHeprgzM!ul7nxmeE1O6-&#V2^$nP^TMF7m$`rjj?tKYE*53+(M9@Lxl zGNrYz=t^_m*;eW0i-!-iEEg8Nz)>&6Dv9^YtWkoMP1x`nN?KKxbt?q3Le_urZqQc5 zDr=}tgJ#MSh__48ED<3Sba$y$8f2jo{05Bbv|wjWYXJ^r?PcS<~SR#ndS-A>}-HQ14?64JNKGW?b=x`8vUG%4pwZ8AW_n(&f`%{?ZM@n9>cd(CCj z?kriGe+sE!lwG9XHPUm57bS+E9WE-9ys$BK0bK$X035%GS_Rt!NI&KH)zA&C^49p;emU=4r}r94#f0PYts-&-;n=@~XknGpdfz->&uZ zWjN63pN?e^M;iPUOOvh^6Vcce`sPOz_6ZR(+1>;cwcQMc->;)TTFcad%)CxT8W|WB zDpA97lO1!PO4%K0NLy8tAFvZr+v)PR@#s=?>a2=f2hD>Upm52W;$%tNA*mM$Pz`nn z3e>SB6NT@Ok8M&KUeqyvf}JLAakp|m7*K=TEnNA+@vhW2Gy-FR*yZOe=jUIX8?&#E z(dh<|;A2WWVZu&!6;r$QuE9yr2-zE%@Y4xw@s@!A&ZkQ)^AvkdE+%>g_=L`+bd*m_ z@?HkBrFL0?&n$?~`$nzOB2 zfvu$iCpLJ*BL?Nn>9Y^hDLb0nhRHJPMaGksY!cGx#u)#pN}?*tXncg8KCq%ru@06( z^f;9FwY-_E={F{Nn5h!dYRB6}f%QoYP5NMUxhqS0TV`ZE z3HjC*Ce=IQ2VyHHU~!XAR2k@lvS(cm;zPjX=DeSz7flj-Hal5z~ z^Q}kW7J3cy<{_)nMTKfYBAF!2L5@}=7i?_!e(ao;BRZtj59=D&L3mKb*|p>7>(vEN z6Op7VW^PI4I|D-PvL8D@2#PCqQ0N|L`kS~&xrwcJl98ma_CNUzE8w+9ab=hN$`)Ob&UsgPpgnH=m zY^S5y7hP9x;g&^|MG~{3sLl*_Sf9DbZ@Yo;1Wa9S^U8guB={C$X3oODslW>xrraK| z>-Kd^eBRG$Nk^EF5P@h!Q7-%ww)nsO4o*MbSz^* zMMZ|{3{^rK!O)ANbO;!d5P<|#1R?ZLLX{CKp-Gnxp`{RduOmtqkP=EDAVMgi_s)EA z+%q!!-TQgoKi~U4-|_AI$gvh#cUIO4_sY7j`?}6E9Zw0Vn(K!`ThOI_A8XcT?$CYG zq4f%E@KAYN_JcHy5xii@h?qkJF()ee{FBxiGCF%j!eBzzB%8YD+DMt6!Hc+I6C-`T zPkF!R_Bb+wsyUi>BT~Q5%*+B}W)#?QH$hU4d$5`mCak%KCiKT#pBq>Ypk>_Xt-4G3 z==ARB!rkeVx2Wv#!0Q7+ZOyriTRNqBw0B_4jhO^`H#8=JP{Kzu(Tg=9nwp7n(QNVx&dKWa++=tWE7E1)L_$|_Z*geV zN9bDgS$7EdRa_)iDX;!ktY{6jx^Wh)g;9OiGoPYg7B5S-SiMUZs?zf4buu&PZOODC z60iN3=WD;l6Okc7tH00A&fS?$w#Vpo1oRQiw2PM!^X?Zl@oPE)Bl-m`;|8`V!LcY% zr32SIays6u)nv>-A}ckJz~u#N&P5Y9@=W^19nH{9qM?ckvgWoDTv??%p$n9vfoeiw}W5bZDws>m~8b)C?)56SM{X@aE#tqw^IZ(YKFkm5e*D_H3bT# zCyVGbeG&;j$5Y{ujuvwg6~y`Ty%TqNfkEsmS?k_ zQJEVsGV#Wh9IFK`F7Ig?*E?ltoK;ue4ZGO4n6Ox*B-TRHnn?3p@6g#yfpSw6gY2}{ z8nR5(I&0U^`M|{yU85Jqmd+gRogJ202R8~w4?XT!jD6S>)h7Y5eq!y`+x)7NQgqsTCvJf} zoKp!aF{L|Rf6H;26wE}aT~e_ws_2K>uWI!fn*%KYR~StKHaxBH58*|aTC`NU2_*2rcPUEpgOPxVa{3t*k!Zt69efzn>swK6R4v z)49R@j)?YLdyMy#N^!A!7f9w!gKIe11I}t7HX)~k?v1H2D^7^sX%j32L!|{ny6X$7 zBpv|KK@IY;h1(7$k3wH3TL>FE#Lu`C4yRR`qOvkbFTEZltDbWNo0ePfkB>{#WX#mC z(4G$f&12KzBT7**kXjv{jarNVJfjsb<~-4=P<=i`g1q65FNU29rwS#lT{JVz!Bq_1 ze-|Ir>BSYNVP1PW=mWX);L>?--#I5cuEENUe%mT@)Z&%d5Is@Uk2&Vi;H&Svc&fon zx-{U?6W)cFJ$?H!B|e$8luP$B!UgaTK?ZC&fzc=tYq0aPSrAQ@qlnsDM|I=~F~mG! zw;7KRmz5*BE}8aPaIG)^yU7)m2#<)$$-?-EEK(OJNp#YB?^@rIy{CDGf!=a-lh1IF zMUvTkUJtbhYFjRnnLp}yrF=5-v2jmZc`jO*4W1X^&e7PIx5-MH4l&GFC+$n?DngTE zN;O6FxQI&bm~dr5sD{QQAqt}5RcS9{b18FSRb^CXL|EP3@<;16bQRf`0|;+=>tmRZ z+vUdhstdMbY1IXX_93V_#7atuX*skJat%v-MS51x|1h}3JIGSmhbTdc!EL1ZK>QiD zu%#v@Ps&z3i|yyc*TrLfeJ3yqooRijFaryV>1Sc77_F*qfofGmWO$Fbx9fPJgYkbu z12J%CCWCyDxWxbV$5r6}s>DU;K(kimzL)jWSPlMea+YZ^2`#QhT}xDxQZ5$(=N~p+ z?=~v$)tGvqrHR$yaNZY307cRB5FU8S&qp=X5p*_J>I-<_ugBV~cBm?>z>aZ{O=#}2 z3gGQ8Fu|ZI?cAlLs+Ly51br~#C-De9F?#fZZy1^>6rsalXkvEDlRO_e_cSm^Jb!jr zU^EQjGI8%nEbu z<>bQ7DR;qnJ%6l(8`4TNHj>+{GW-8^jOIIG)b>9zAInelQe z_>iWvKrp(34`Qb5xP}dw^Y{&eAbWJLr6Yo#YcQz}+Y=(T)Z;g9-2tF{G#0y1YtS^9 zc@Tz=!?$|%H~madeW&}W6Z#h?k6ISiq?mg=_lXD?5_HRlGty^MN_Gc3^UjQwg_m=93xsYRCpcyA_;jVjzvU)`*TAD=9$1B(?R~8tV7$k#gA5z`iy_CCtfthMio2Ur-#6m!BWkKWFVU5> zLPny0XE9fmK9pyyR}7K@C(8zY~zR#dG~rY1Log8FRgA#Oe#OGwfmg4Rh?(R9)*#r-{4cznciSzaSG%

B$0O2rUqj;PSl5Xw|?vxUjN+&#H;DhB@5&sn{We0g|zg zy#t1Rv=AR^o$4K;)qE_0CnO9~nMrN%9q&$bF9+5~0~euNNqykHwL>6E{nvC@;YJk(!2A3A#DTztT`Yn2%)PU5!h4wpxrS1uIxnfM0b z_?lZNw|3o~89pIcz|D;Io9~B6=UPm>VFpl;RkxD8bLp%i5QxlVvRTYtgi33!PjojO z25+(Gh_j;Qd+6_OzXcdKYp4Je4!?_d)773t}bxSc_;i>|6_ezWMkMYyoSk&PCV zGRw*(gv#j%NN@*hX2B4-9&Ad?fQhr8o({d`k9VaE&SmO{w z2pU(vxNjmIMHJZ66ZTriu^Tik$ORd9+rwlv^>?x!^VD8kULN~*Qukbl&Np_B+GjH#d#|{6^g4m+2^;31&T2g8&6pMxi zJa@_BQGH1{dSE@S>#@Jq_4_zl(hb<^r*m>GNd@FBS&V4MD32*})i%^O=ErwM^B`8rs% zpojo~Ydy2v%O+x209g3QIC(&pM7SYa&_fRFG9Chu+4PVaGM6rr67h5`%WHGKNW2ky zx2=R_iFNIjBAvTpL@>|hX3Kfh(x-MhvkJ4i<2kkDT&ufedcnZ@qjU#2cGk_#kr2oj zVB!KWmYr9o6efo~I~=LGuW~*r*88v{+VKy7LfJ`V+ROutP?J#>JG;W8aCtjZxXQ*1 zGRzL7WQj`$k84Cf~&VL6KG4NYb+SW?FORg=1&^D&G@?3{H~@-ltpWH zswZ+Y!Nqv80qAo-KhAmVxAEXU^YL7`=fFU=uAw>ItAm}^rmG&Xa@J^bQj~P$R(F#vV zhNrT3&+ENWGyT!+%?!gzVuOA0JlKeG*;%4KMlta71Abs(MTescrREt-gxjpwJ>kc0 zQ;en!xy^Uotu~OHtKDT4?#YzfWdjk*h_Rnv8Si&d(W{nJo)U~%(L#Gg!*sl96F;!^ z{ioo1XVAzGO=qP-bgHQ@9}OY9Hi*3U@!Xl~K%#133A#j4!R#$Nx`u&U$O6OIIpmwy z35A%LDqnjo+gOm&YPjy7CD$plzJ zaUEX65qw=Lyebl@q=VKp%v^3mk%K@nx4JA@GrsK(4Acc7E89e^*+Y@r>@86+b&aO0 z#kooubgV-$gMLRbE`7;pRJ3O%gk~)*RS4iZn3hSsw~)r zQ}oQtfQVLl1iXT_N@wio!Yd&w743$B z@RNz(Jvzqtou&SY(5A`Yr$P3jTebtU>^=I?WnK(6?l{-C!RmX=^V1o4dXCM7YHfPy z1-nA_Wed*TUJ8W*Bom1EB<@ccg%#7u`gz|i9ZOYxvstLw+*FZ=qVrJdv%?KvpTrSM z3khTgm=)-@S;IvW#F}BWp*&MbqB%?Q!Kzq|s$MiC-A=tQb#O?bD*b54>FRryE@ni_=TpRfD;`~WC3UJ!2 zot+gXGZpi_aL4-XO0D~BYH}~xC*vmE$0ORx_?%1&yr%trN?nLoGlkgiLZneX6RZz| zQ=-qr`g~uZ?!lmYN<9KzW6iWLf=vrrfvCw`*i$^=g_Yq$%VL$`gX+e_^@jx>RvGI; zPpy3ni%0CEiz#TZ=D_E>(KK1MgQHpCnQ{FDC3|?nRdmu$0wQBsLYE`~UB{BdT&v1o zg?>(2QD{!mPSQ3t3t5M@>SvcvRaVt><2ns(@#WCAU$m!clasmuO&6=;ah}V2dU=&y zY6TyebqZUFBF*(L{R0;HW1HNwf4HIQJr78E`w~%O#k0JXjb0^5c{pg{NDh}^wv}6( z43Y`|by2>YCU$c#b&atGc9}Mjtd;_m6cEue{V~O1ea?Ph?Xpy0QQbaheto>R+%WKb zzg1L;>C=jrMt5)2`T6Q*S9)4{%+gUL7@IO#LwyBsIrB<51&Glcy0$hyXouM}<4B!A zl?;wWL*F@9yC(CrKHi`4A`qw!825N9!>;>Sp~S<9bX=gU-#^f zO0bs;G4sO(Y1_kkR|JauWlnNVL}^-&Z1Ajp8d#}Mr_6Bq87 zlWm`eFDOSkQ|UG;h6O1EQT2w^@;t?vZ2(R;~LvcT5%CpY4g5)C%22>&a<&o z$qhSZRB0MG?gsKEv9l3sf>&EN_%EoWb@Gv@w9k5gLSsa3s1;x_-E@x#(6qh?ISf3JR9d-dU!^zwCNncU*H|nv2pMg_i;dG4RbIe+#PI%P zSNDz}f<5!R`=wW3vA4?dtzNWSge5cPQ?5Qv2b`f)*(=H8^)3UcFXH0{rej9MBOrBZ z@%W%n-~8t<(@i`EYg^V;i~EeA%t?O#5Yg#0QaJL?!aBHW5P_q&&bjlmR~4C=@j7|m z+KmKrmlCbZ-s&nw8S(>v_VB;oKx^ULEiz)(6Mckn!e$r0FsxGKda}>A{rFwNKeG72 zKAw4l?z2X@QT;|xs$D9xN;&p!Tog)8g+b(B8f$4LjNe-Oio5-t<>W61mf1Wb28gn| zWC`NkjR*a<+@bjtCo7@SfyQo#78i*$4x(sb@TBMU&ibyC;33O?x82+Qui+W!J5TqBST%rEa_kkaRQ6VtD-Y&E88spN;QifYrruy@o;lS%7oA|hiL^#0q z+}8cv!PSYF9e0ps`amz&kxI$8#q9Za6t07TLN|g`Zdsli@osAO?k{@DUybC>cSGSE zOsnMU{nM(LRU8l#^RDQt{o@oV!M1Xm&HgVY86^(v2fb@<#Sqe)&`*OVEvB_XPLavp zL1BU*vdc1exCMB}#n1`gBq< zY}z4#((X#Au#%K}B!{_1c&?5RhVwA>^cv=X4841b{@N%9$01pAj9aKD=2}|dLWppJ z=m;sHL7tz8HsmzF7?Lo{>83G>Th>9PlXFcUY3N88z&{pB9dr@;%Ev-$yOrHlBxGm_ zKm;GoBgr|S0VZs}DYPcofvAwa-C zTq)P$cb5daJY!r_tPEr@vWSH6?sl0z9X@(%MQa7jf?0{w=i*C!USH_jH&MtVYxKzD0O%*(psebjsAMjyFf8ooy&h4i0^ z3W3f@Z;-;J1EUfA&ieeq7iRC4zaQnl+CKXa+*;cHPeIR^vZa(*PGD9jPj%@I3s2H- zfBZ)OA?{Qz-uvRt{#W8oP}0*o<~ipG?lDu4n3&fD>OdYo#mFY=SRPd zWciWr$5Sk+7vhZk9+7i*vI2S>bF6?8$B9WlWoBlEp#>55%?T-|T$^LS|Gku(<*bvimsInKo1L)}oGW|J>{+<8SFS-3y+Ilzw+7Z8 zgPwkbgd1g=Ow4q>0z_M(@z^*=NC45*0F||Rb6%RHVbhkG)nuC)uxi2oU?C{ZGUx|m z9J=+f?vjC6RO?SpHh#CS46QxJ6)<*T99hP#jxot&$9%99Vtn4$L03m`0r@eaO&}wDU8%z-DH=`Gu7H`_jVmLY3P`0PXuKUejg_dO zZUCt8xwOqk&SLT-3R17{@AzG0mkp_41gSLIB}e1SAr#1Og`GkIFw6r3)LK2RK&?-o zEqFu`RL6iGoST}*A-(OM#q3&fPld0PauK8Kb2VV*AdRNROfOO~c~u)pL)8uO-o6u9 z-J#o5CO={q`SkUk*kT!RmXsR}_ueKY@BhxCu9()kKeq%$ERDDER0j3f0cB7Ile`iH zrxYr3(5SBvv*a5B^MS$3yvQ1S%SBcZ?M*8F=SA~-Jj_CIwcRETHx`(cvg)C-B7){6 z78ga>%`#~xDRfFVH%u3G@JJ!dSGko^oJdzF56Y{B8sPLE{c`Q`Mh9j zd#-M7OSQx2ZMOLh#(L|OxWSg|GVEuRD2zp}sZj6L&YaquGFYeQpzPk_71Hy7d0^qZ zlOL~@Qi1rzG7KNWRb&3vXn})b6?8V~RJOy$R*nY%Kz(S+g^`?y;9qu)CCjwPueQ=b zvF;gQ7dij~A6|&kc(vnH!aOb=&cJHga zhT{tu)$b39!a1HH3M;bBk@^DrA_i3%N4z^Q_8hxe;Vq0-pQ3`4qSlSGW08879FL{p-Tf~da`J!vOL>9o{>-*vJg^hk<)#>vA;|rJm`|#n~6D#KuDaO zy{p-@ic{Xh3Zog*1XZHnSrY7qZs-eROx0d2Ht!r@c&oRI8SWQAV(y)kZ6 zH$%3t!a1~#o^AuVJW0Tg074MCFd~+Thk>?yIIm%kr67n(r)%oAJ|oL&h%>C$vmN98 ztrZ=KFTD4+7z?mm4e#vB$lg}s60FfVaKPx=8+|J{uU)&|m=ZivbXKn);m-(5Ke%A1 z2yq%tLMQVcClkm8O~8(%K$zi~o@?vVi6LYCqjJaH@$36p^!(W?&t627l$~+c1&Mdq zd2 zcTi_S1~+wgYwAs>>il}#7PJJ)SL!#b%RTN@%cj9(vllFS6ytXY5+i_8E#BKb=c0*I z1#nf_5{}k5U@WU){u#PYUKvQtFFiM$Cb$;W@m{*dGcRDmKy7YM-O%1l&!c>eojB;= zkYX?dxg2GKCz65%l0x~;A5-s;u7)No?7UhRqQQ}gU>{}c-&r&myznd{D%aG!v^2cy zeUR~%z{mrwmMiK|Xd{yfz9PXZpqDP`*Ul9{%rtMTd>Mh&# zrZ|(Ts9y7(#)X;$DQf%U$U;!oLLb)Qc6Jt;*aelk!mN+uB4Xc|i&Zki>Z0O3W|}3# zKwxb{QAo0x8{Mxv^EC*89Ta(}OQ@@Ok;QPF!sZ><|Uu8bc12s_zU(^>{T{ZfRA^2jqNw}hlz`>nmJhF1fBL$B@Sj0$flfWKk|8i z?uRU)^<(7iN(-j7dBWC?pS@FeriGFFSZ!X84roT@$=Xh3V3}DY0?~vUpzx0sE)8rg zZ(W;26J1p^YG^{ecKC-x_%M_8rKc1h6em|!^!yXy^PJ>qTE^TvbE9{%U27%oDwDDT zf_bG%rvM_3>>PYT8cjw?D&vBXK4V(xejC?|RU&H9R0<63#1ytTnaH6M3*B^cHHrLr zmo+pds-r7z5e74r-4amwM(^jh#_F`Vk<`BS%Ambd^N%TO)=}N-F`#H$Mx|o>q+>5) z7QU-*$*B&5xrTGRa1QJ|4LAD{h9ub{h{PQy!)VylKJId4Wx^x;w~`AP#NQlIPJr~a zS`Hq7XP#8EKupq4MZCD_SwGc;%AOgdp3ZqKFvv`D1)68dtkv-mYv4(QM5GBU&~_)8 zM1h>&Szb?;s6*>R57InXhwZnQ+DkYZCjG!~jB7=2MEMs5)k zH&$6w4Sx@&eXOJgA?8nQo-s!{Wn?=@C}ZE4fO+WfZFcyRbgF7;Za=V(Q5C{2A%?d} z(BZ!4niy&agL`3O1ID~x708c0LOn{}*XSG$G1I!8Ct27>>FI8qg~W#?8#iLB=Eq_I zm)E`1NepI4H%NxUuGr&cs1)Dt39>6_6O9y_pp<|Xv}-*K%^U_h8uQ}2DA3Se4r6b= z(WCm5=#bVDB3Yx&XRZp1MqR!=vMuO3;a$<;v6@I~hsP6^LQs+`KsHa7Qi>E+$zS;|;); zcPpL=(0fG~O=jjm&yAn+FnbRD^q5gw#hy^URx+l04K zQ<2|Kcd0vP6w|XwD0*Hv9v`AHi>QFX(`k@GP%_j!j7Oj5RhB z*6~2`&w3H@Rs>{M)J?wO zijx##3ZTJWBrqExxu`#r>AG<*4NgtaV@zKw5w{^Yr&p@{5QaWLiCk81I zOx9@tU63G?BtmlE^upV(yXr*;%z&MUjf`3mXS9a`DGKcweL1hDh}jUOO_g6=Y6KVswti^;9q zdLEUCouO@4W5!07g=!OoOm1O{j2m!+s@yECP0da)~eu@i^W z;RR?-+gOY$aUyE4)h<+rgM|dxzPm8Vx8K@@xy{*#An71 z9;Y40-`KqL9tPFF&Ya<2ZZawCQqt~=R}{+5H=oSTgmtE6pmZ(*R^K5Leed%Bh%(9W zLe48whRmra@MWqr*V>9GrsF=pM}5FZeGZ+?Zg|D{W`8S7gh7x;!HorKX}D!sVL;45 zQBNanpgZ@e0*L%VBBQ!}X3J@A>+`xn8df?vj% zb`Kr4K=B|C+n!}y{eaGUg)9nHUk9BhIXBIBrIbERtH$!aaY=DsEL^QU_jJakCupWA zg>J~gs(}B%R+6Swz4_-4B<=YnyoT&a>UAvP~3g>5z`TICA(==&~x#!#F zSe$LXwmdCTWjL@vSt30q-rBg9Jg;V11+ORX5YF6RXDR`5b$)*MwrHiE<>=x4aQ>6e z`GVZv9Z_L&27A0YACDyEd~GFA1=^7B9X$ry*K1+pr8*TfW^M83G0gSnIo%GtiheWv zDe~;nrRImUEw9admRpB+b$lXQ6LaA}j>0!)IHf!V{M%}_$U7!06@Ta032o}vMnCum z)ZP~UV*EIF>|1_UVCaxvY{{c~W?DW!k3-X>@?I^TMd;4q{rsEUD<)`FDRj;04|r=} z^|uD{=};zJWXOWCEz-w@JJB;rB&1{pyxFF;wyD^8`t+h&F12UR!@sy6 z%5ciQj4*}VrBUG8*GJw+7H~YjZ1c5t~eOu_9Lqb2y*^%tgcAK8pXNP zCUKGkGo5I);t1E+*a!gXk#m5GlO)aCV8s#9do%RFduS%62$8;AS$omB(}Z3(e@S=1 zUEkDk9_=pV7S%!2@Rgj75Yz%xScLR`qI@#QUGq|}zTlVaa9hu=D%1T$a(wP1@O8oP zmEQ^q+F|u#t_@xud=6eEH1>2!F^5V-MtqGjz*Ig4UR316!=2oK2!XiV3-~V8qXf53 z#-vJc$Tc8d5u>7A#Eld##g{2_G-I9IM{RW!fus-I{@WHJ5)HQPBlr z(p&9pzX$B=lp#|FNEflDJH){*{0z5_L7||D{undg&U4woMz@%+tS6pIWz4Iftsnx( zK}%zjP@oSF<#Wo-!g?!!{`VVYeavZLvx{T+TyWQWx^d&MSc+IAA=p6Dh%>u%wP?42 z%8g04lUOmBA4oCNEofXfzv}_Jq}s*K2#RcO-s`4y{VsslCS9y|xp3xhAeyL~ggP8VnBF!#b zm^o#?6Vf-x%hV!igxajo7OpXMN=@u2sdFf_neP`3+DRO?un?=Ir(puo6Nt}oO^xfS zFs<%_5fha)tDd#~JeJhpjN;9)C8@^b^>9n~NDrpxwE<=s=0Rdz=%8n<#9Zk(*w_F> zqTWUQKv;40!=+F~O%p_nWDfOs<#GW{4tFm#C$z8!plULbGoA5i@#JaJHqz!rl@<4` zin)VTAWZy1-$z)9069J3Z5*L3%qR=K{W@P*FSOwXppL=JDcpE> zEZ-zjN5VWYfNS>wz}rd?1UFPl#I`qWA*5dsn$y`YkuT8t7^HRcNL+UjyvE_Ok>Xuj zYA?!(I3-z+8E73>F3frz2-Gi}HS|cz!4CSDA(%Cpp%FyU;XZyR1+FzT?or+oz>sxe zYJWTf^ovtEG2?k}SABTz!Xmw*b(2WwOZn;K@pZXkFQl!deFEx|4OOv?mXaFaY zL1*7E6<41piCOaNxTU%Jhn~#K7gZ{om_j_Jt48+YzYBNsX~ID?fSx;%NBwM9|lgtMm7COaGBGwfYeoafrFvmRT^rf8>knMub@c9bO`)cWwqt0+^c?D2Lc)9<7TzS;Hom5N6#LqJO!7QNMRROy+$)ju#~ zoL{_{RZFXj_gJ1CKBO}(2Ek0ml@pV3mGZwZt_%!cnqSSnlluMH@ZztHe*T^1xbF*n zrWo@_o>QfxOhnmIPF?fMOT4AKQmD@7RtM_xLt!77|LJ#FbW;!Q=g2n?=8qHO_a1+! zgpbZl{vkTU{43F!3ojm24gxzUU`ki?DoX$rnniG^qTUS(v*r7YgBZpWsSuJ>daUru;MW~u?sRGK^7{i zNu#a{vDVL*DDGySy|~z}xdr2vajs$Hu0$s{x(nA0PDc8LEg0e#06Vczx7wEx%1%H+ zu%py{x!wHv@6PwVShVBYWx!G};xOP*|q z2sJ&SdUv&6{gL)Og_$^aslU+dM0wW2lPZf6tK77(oHDTtLa2y0XBb+OaOP^u5I&)r zV+A_Q2ojofWFVD%u{}tk+%rs&dgE{O7jlkFq+AOJ);C+n9rHWO<-X*Q?wWad(+m${%As|STD;zEf}4v3M5+vl%53PEX+Hr0poS6*DglVLH-T3Sw{Q^yPTFS>TIdDOr+R<%6;88_~SCpwMk^4DHgM1M$SeS$!~@F2)qW%qZ9nT@=OLqs>5u@-2aRPbA49(| zo1IP#+Z{34%#US3J-Fg}s$*LvbfP={ODD88&3>sU}J+Pv%|bBkGa=m`SiWsPeqBj?q)KqdpFubMrf8#3c9MY>x`bFU-f|)7>hsgRs-YZ| z#%3Zfmw1llW_-UTmVZn-ZdMH)-RH*gz4oCi&L4~4%yonDf75esH+LSXD@^%?IV|Mf z6Ltg8-cZY=6i53RRzq0^5Sx}bW-uTOsH~F8O)~k>InJ+xK0CawA zaqqjQH@`OhJ2`mMc9)l0`-L2A_LUs0^Hi8MCFa5(2D`5fn9ps#f-6aN!!>h4_g6sc zx$;kHFM1h_E|_mz%-F4pP#C)2!{`Ktqy+4{9ITXaueznl!7YdYn%kFQq^2nww zd$m3JsMpyyfBdd5zU!>{_2`Aa_QPC|5>k{Yf4N+X-#qQFFr|=pc$Kc8%Io@AlYNIRE%v|J7OeP#8nsV`#31q{<$(DcdK9_3V;|x~|p{m{SOh$1K(K zwvFs{%N=yn;TKkDr6-xymW5eED(^IJvqKF67ax{Lx`H96@z@T4`zwxEv277Pwd{ya z-D25FU3PDsX;028JHm2hrhzG+yqDP66rZ$r1^-yj$ zNj#tQzBP|eRc89gYWeU~sBZ?h!wz4z4VQ%Y<=F3S1lJ%UGgg5Q9l1isoUJI#W_qST`EM+99f!^`bCEG}D8E_Zcb4+> zKIH0{;7+mPk+X68axQx}S-}73DUj`V_!?(K&9+o>_L`UB)H&SxTlD5ct#{IvzxeMg z&3no$-&Or%9QBW#9Q`k1vC^x4vj2(29{x7|hsI(Lf5iWK`QOI1_9h%PdEVG4lTlH*Db{7^3@0MI%VwbQV)0JI3J3Wu#v>Eq< zOwL>_FCRgU@c62KaG1H$GDLdPWU+H+UPiTa?d2li2?YgUaZN9Nn-~qj5jJvs}*rreb@4cXgZ4& zLKSdMzv7f!a1EKnjc+9v2K~y+E~O$xntBsjIz7F(V?f;2)&xW9f_J3WNCH( zmF0)lBd;w1Y|S+`C;mEm^ONt<4FZXM+D&4f3T(~gD^E7Q9p`@rIA6}ie+D>uq5m1+ z{4w77zd69!o26bYOlH!r7r5#a&QGT?*E#-LKUZyGfEbndnT2xq-8;NL8{Fq>URGLf#csdlRDln}=AJErf%%WdOQ0QOV zK}tp+?1Pq&T5p&{zsE>{q;?MO=R*%;7_($Qm^(D$Ls)zkTzivW8y|;Wo-lM5jL^u= zm43e|t!bn?SwVWPp$Wj@qMOjM`Q$v>)J}sq2r_Fuv#3zrAR?9NvSo~H82RWn1^VVy+9Dlr+#}iB(`2U1az2Ga>NsQOilacZuWH%Ga`vZ9DrFc zO-zS*=3bJokKh{*Rl0`nM0Ps?vv6gP%(Mm<2o>g~mS8OlhjQH_Lwf>~MALx*vx7M} z5deVXnRZ7)6@DEltCe@7T`afqsx4E!Hvq|4rFA}F)?%9}spR#_Nr>Jy zAxeNsXb;2o`IGWA)Xg)SjJ+1t*+pngFx!^&FdHBFaBqTA}4sDI+igD81*u!Fzbc5bP2)gtdA2Zg++%N>sey$^xWTRw59Bhca1GSyi zsN{<%L|GZPNiX2?E<<4k+iaKQr;>3SRq*a12@1#Q!30*|4TBhXRitTeE1@kKqk{D8 zb5F4QSZTGkQK>Qg=wYGN=J!J*H8d_^muHSHEnp5TF7lazbRW*r`D2yNAjOJzR2^kk zO2{-BXsv>|b`8zL8o@B*wdo+A~+byDE;nLX&Xkyb8d*gmh=i>|%SjlOql@Snu&W z%h56n&5T>^RDAB~Gf+ELt72qo-?T&0@Wmd7vI--*2TJk)4CAdNG+erde>DssN?pFS zR0k3Q##iFR*TQ?WH!PFESi;)nSeIZn$)KA4+$&kQnLuP8z7Ggd<4#(^H<4hSwD8@| z{CSX$K}n1si6KnmN@(7mz^HF}F3sV*T$dyKURkWy_`;PW%Zqw7EbO9x(hJk7VmI>g z0<17AzQNv6MI1k<@&t}=Yekdvftai8K)1F0oRUjgXKmji1-i(FA)cJsgn%r;m}HyD z5HD3Zf%rNu_7mhmzlil5u{~>M(<|vwGF>m2Pb`< zG^qn@;YhYFigCp_^R_8*vP|r`MTdrl`89mCvwjn5IA^|xxMsln7_t;!8+vN+80C(Z z*xm_M&6184{#7+W2TjS$rT&v&= zg><-%YYchVx)B|`gpmpTVM5uu)Qesvh%_0CTp%ivhEh+sPMNO zljF)RpX;SexHdDZYaCpE0kf)0#Gf(|H_$IxuHXrpfU0gra@HqyIbw&3RLX1_tn80X zX=?udx8ds1ggXPtq~Imkcr*%^$7lOKYEyG$5mx!qz`n<@ul9xA>MHzB0~D=M6VeO) z9Qd}ird;GL-{V+iU@z=q?e~Dx7-LWShas*Bbah0UHU?M$xADVdP{ zt8KYxW!-A^$VS{y`LrnG#{dyM!=sxbd8l4K~&sKtbfOPxrlai5HHE@kXb%7 zXdcaKOvJg=6PLA|JB17$Sc-$%V7UZWr-%qnrg|%Fl`N?H;SgZ5t9*UWj+G` zKAZ3FEazXdPTw&9YsEVwXiZ~vPxYo;)?alzhEhcBI&}-RcN5C+St@K`cIJN3D6e>L zx8{4(lUWF6x{K(-ik+Xy6de}lVy86X&A`@lvGMoHV3x-4IWZ67$+j95* zQrp_oqpsincoi*?d5|@HVE4G&w_rb+^YQC@29GDq^!7m&dwu)q3mV0By+bMOPE_Up z$*(>*G$-%#?cgMmCy^3FH-qddI7&qZlNo?}p8tLl!h}TWu(`Y-jz2Nb&ul8$T5h%K z$*K%a!>wy^Ob1PE`97rxTHvHo##T7ZxIfy|7-{VAZy#L!ab4%@%fCAPg8Os%uT78C zoCx}}*?DOA<5#u*FPl%Se;IveMN)XazTXdj?%`LLmu~PPgUro$s)Xl{k=zdl7bl$&+~T7WAt8l4E%emNXu>rW_CiytlS4}~D}=`YM`|J@g5)^?Su zEWQ8ZZ;AWOrLj8+uG{2Yt#9qUEO-CszWV-Kl;Jk*+xIB9t2mhTQJlK-)}!23EW<4d z^Zlcw87>9W#u|3`_Z8;H(XDpH<1fr+sDaFY1P=7Se2v`c-mDydcBKEd_oa6I*%N6U&EU@&BeT_GHJXm8qSr)BagR?XM2n_3aJ#?jCgh&wZ6)@sq-C zxP#QX?9{5+Mp(tU?%!D;j{d+p+pnM7pyg1s=NDNom-x;!abx1H(1K#6?)UfO0&i^! znD^lTR`OEso&M~7KZP>k|Db<)xr|xklKQj9sr=bl|MxpC$HL)Q@cevnyBVR>(z}2v z4kwU^ILrH%_{8_x&*MO&jB5u`?%* zO?|tPs(!n?wp|_k|LCjUn9JU6wZ4KW#xMV$_TD?H$#iQU#y*1RAOcFUP$e`G0s?}K zUJW4x2t{c^LO=qc3XYB_#ee|;2~7i}k{~6N&;}F`7IaXNYFkfUQ6d$#45@3p4!3 z$B@@|2D*}CO-*1~5i^iNpz25L${El-x;Ql#q0h>|;oH!?|Im2&Z)3Q$=s&Fw#~{DRa}2flee-(; zG(CgVmtFdc_M-qFTZfidAQ_OpV%=Svo;69G!gAldhmOPd{wgGnFaIgh{g-b3>fV2m zycP&^3l>A+Ke(3tcK)8zUZ*UlSBx^s6Dx+xmUC+CSa0r{jSloOHI0J<{85W&is+kp z#jH`_yI=m6wq*ZP+A=Ve_{HO2C$BwINsIrfSBZ~uz!hK8YabOFnAI^_l;wHYef|gE zIfm0=2)-{)gqMAs5rJ)b3 z|KHlo{~t^IJO9w;4I3K0#6{CEH>O-2ci+`&@NwPAxrAzR3NTW#y>Dx1RshAgb`+vlFNP zyDHAl^BmS4z}@%Nl3@3lxk%&lD0A|W9wtgK=47W3ss_S8^?bFizvW1UzMe|WQ``ww zpu!(~Z{PEZc#VG)5l1lpui^4Ws6ekNZrwyXT;qwn$Ys)@zqJv(;ClS819R5+T@dRJz7L;x z6UOuvS#A{F+^?taZ(A9JeH-#W6+nFUasJF($moZamm@o1QoSAe$`-F5YApv*@T;Bu zFx*#h{V^f9)VR=Hh2UufS*3=Fv@4^dh>{ma*1sjpU2)nh3cs;|G++Qr^!l?aIdU$8 z3%>aeIu}Rbg!^$Dgz@BDn_=>_dx-;C+JjTWnRJ&Xw)y@1Zg?x`aRZurwCfgqsIxk~ zH=>Q7$jtG@kxyVkMJ5st=WL55=f&9%Xm#YzmK7kYZCrd;pb=XxHdcb->#_0c+)=2) z(NiB{t02lzpnT1KBBro@aMoWhXCEO>WN}RocaO+hkeV@^_;Q{xY!rNp`fV3q4||BI`vNHg@F)F8$YYBGYTgHTAtqXav51Eqbb~Y>8n|= zGh^Xd#bJMJ6+!ICMg#z+B$rQ3#8`)`6||mURqOURB^M*qBD}edwl&kiu%gh^j>e*9 z^-?EtC3jgT-49|f;9XRh=_S_EB{!bzv&_<(R(|FW093sry=YwJac4Gxz?anM)67~u z)z&3DW;9KHT{Dt_LCJV}E<=JoV|BcsPizT?{WE$vcdV{&EozAip!6?%zHwaU{Wvtp z4P!YN9G&rcpu@%M^)QXCF+)hk?f1xSxg6M)p>wCIAh2;Fg!}F-mhO?Ff&u_SUj4pf zu()H*JuQxv+nG!}zZ z^bt?OoMvx5r)yhGY4<7wayC1Cmguw)db#;qcs1%&gcrc{(LjMMhnSh3Fl_=AX{W2j?v<&)N{LsI>f|XCTf&#l`l)N3 z%<~LXC-oP{#`Y84rBU)*o!HAo{*FV2x&pRv6)P7;e4AO(@AM!Ek^TY?43>c6&GWP7 zf6?N8j^4lAtyH#jPPEI9)qV2@I$k$3MoID+%c563LaXYY3Txu|*NJ?FXlJnTtI8dZ z@ZL>MThm9vJ%^xrU+}4|E@zyEg_#IbY_8p@1sT07tbb$#Tz$U!_eE!X;Hv0aHu-M5 z6GXpHF*~T#_0%7H1?daYY9?DslVHb)zBD~HL<+*`F>%p9k}ap4@-``0wN2HD+KP%> zlhFSiKc$hJwrpai&ptMSo68_De*NJ8wKXKw1HMs>zMHFfWRR`Zi54zO*C!9$WWleP z=*W7Fk{$Ysb#c5oKLXDentz^)xuu`LMq`4oVt@W;=0)9XjWW!cvgd^4Nc8?V8yJt>*lNnoO%&9*)gutKbH&n3 zj%hNt;Lhx?dg*ops4qA$8Xa)L+PI$?pTkUcGUxcfeW&2%rUmZ$MMm>>uKW+Dd zaoyzaI)!AX`$#}K()FT~8E+h#Mm=^^mYSB}YWugIjSdAbC3iq!j_|UcZWi=PJ|+1l zY+}X`WE4dNj_0+XQcamio1!DSu|6FpWf&K&{`tA=;D^OMy?rC|qURmOl}f&Y=cK>z zVJ7w6OE?FrBB!@)e87e5Knk}vCni=AIUwz@bjLiza>AXqj5X`TNnolp?t3#&Sko5h+BeOlIO!MIN!%74m- z3gP4P2u(UIb9m5m&r^k4jUlJ_JPr?D$bNkFWMjw?h1<^#UdaCS@Sqp;CnI28w`O^w zarrjdQ=|4G$A?p5KDorTHk-I9=V0CNXb~_RJsKgT7}d=N_f0R<5QN@bTZ9we9HAL zkSPosWE23cj2?$iCE^SdA!&z9-UGaZG`fB{?)?2}f1am>TiKLxT=*BzHz@F1bHlcY z0mc|EWU6b;Y7<}YRJK8}D41r7>zaO5ly$AS^HUc4p6S7^?>{jHdUSOCNUE7Ip~~nu zT{pC+I;j1PQqn1}VZ@xO-Gzahc1>U_Jy4H@a0PC1$!K5Sqs5cfZvd5+&uTr&mShUl z)H(heYXu&hM2E2ly;LFS;o9nUgA4Lu--V$rfn79Fkh6TX{x<>sTj~Dw#nj1bj5Kzc zcPbq%*f=@2ATb8+H7=>GbG~bXxteub@1XE&TA>88J2avnv~yd(?hajIc2-mi@NGbL zt6eV9L|t+~2Fp7;TV=H8`;#}S(b1b3Jx`PR+Mk$cl3&b?ySFO!RiJegFk#|S8s(T{ zQA{IbVmbaX{*H{h`Cx=rEi_rC^l8yk0g3P8dh+-WJ`@t);)1?slkq%wCA+8eC%Q4v znUt|6UYk8%3Cu}*xNBtwlm+WAnogmVss}uDtock|e}~UC2Av%oom{EHoN$xfpjk3= z(&SPXcMI>@NUs{K8?ZDJi|9@QE|C+oW?LMA_v{L;%VB!sdyzjxe!j{OyjnydzQW^5 z=-xN>D?7?!+aDo77b!2JE58d{T35n{K>%P<#9O8H0$JyRl+=72;}m6IwA1w zkss1}x9{ZlyD#!J3%|atcjOoKAD@2q-(8eXfX;KTmDcTtZqi66yl+!W$y)_{Yb zejo@W`o2tf{+eClS&*6ik$pbtpq(;jjkbcP}k} z7_JGrj?u@|!la5VBq;W~)f)CwCsQOpJ7q!JjQk?+cBjLh1}j0KSGQf5E&X8X$*Ens z+3ExDAM#RvH=Tzgk;u>O7B0U0;T0Y!4&UrmEX>Fs;?WLiTR-1Q%EC|3_NHb`-fl^UzR^00?XJS2RK~b$qrF4r{k-?=>wm!@8;S!Curlv4(YNiME9`4Gw4Ah6LoGPeAP(%?BB#2 zwAyrm5pA&3 z)g`_<$PHne&@Us4L7=DOP>M^-%Nzn625@#}Qs?kD4e3^G*v+Ltt_K)Z}S>`D_YG1CtgPUc3A7Xg!qT8k)_v1>6R~% zURFnX>5B-*z0ECZwgFxsjfn=&fIG<#vc6lG2(86meDA|C&<(JgQwq3kHs84W9l3_x zI2PCv_sOV2CAZ(cucyJqxrAjtm}m3~D;(Yyo5dqFx~cs8zfNQRjQYQ}m(xm9ocEsP zt$|GY8zyFAPn79gL^OL<;dps&0Oz1?3?|t{!Ch6?VRit0FT)P9&Byo=;!MiCOZzx= zsccb+EQrgQ_cD#;raYDoYSZPKB;!)nCQZ&0J^o3#*jCH9nVu+zMJ+oE$4dO|w+3(* z?l|vW;LQ{NC1p{YspPu@hH|;^5<9kq?O>Xh-}54jHi*|!M9^ruvXV|D2E8p@+E#Z~f9r=s zB`kfqN^`qj>YORMxq=tPw4NFMDgfvTepv&%T?GS-^Pp7S(|&wPGIwi!-@p$;5TmUe)=IUrP8Ts4)KKlOIQm*JSV|{(|NhI8*+Zf7Bs|=O~R{%m>jl6<}>mJ~# zXiYSlG2iqiKfeH~F%D^tKC2@69STBHLS;yWLb+g31Wv^QtYoDjstCH`75>$*=5L{2 z|8B;)5F~LrDt)2xo1B*xl zI}?G1D14PfM;JP{Jh`@o6wIyLK5`M<0u%7|v$2B$#gYL;KCQv;0$1t)ms<%Dmr*cp zdQLe-j+^@YW+7R817DqX&v@ec(lLt#&D3&>$Lp)w;p?B&JyK~)jh}49_bC0)HHED` zFR1N7s&zRd8yq~Fi2WVoI%#BQ*r*^fo_}n4qcnf!F8v2~_V)aE!_n?9=DnQWG5rIN z2?WOv2?bbLLa+~}QzKpBv}~)+zf1K0y@m9_X*Qm{o4i>&ZvGc~8mW(0R)Yp4ajfyF zwLO(TzS>Uy{rWKr<_s5h=5xDYb)9SYw*yazzyo<881F#uz#Ew5RP$FvYov%+yjt8< z&W}9fD}4TCenB#Y;xqPGp88&9K=8X>r6aU`abT!xL7BsX0*M^2ZAzKHQykwR`uU7Guszhz%wq|^>UXwpZ(<72t;yAR`9onp(ayHEpHD{ ze?+V|w`yPk<(2*+xOWIedCL-q?B{O0eDGhH0zYf@-eb@OUSHCWKLb#uD`GA8>cPwP z=UdXJ_;|01Ps5|ZXOBqVw*1FUf4}!%y&NpE=T5wy%fw9EuV^J^fAar^!t=_AyLM7t zJLZBn4%l}y3@xih7Us`%N1P*VHuuv--m%BBri6=y1QZZu8qVdu_w$HwCbUQ2DXt?l za`FCNosk|Cp>POTcXDJjInnG~2DyVi3YCvmkba<5@7pf7#v`38pr;YCJDL~rEGwGX z)9lq(Hy7>UwGs@0>5e<3d+X;D$*-XF&@h=gWkwhg5!COR`3@AKUH+NtaZ zKM_hiVIu3(WRtac&wM}bb7?!FkaLj>^lH@2*1JlJDt~_(B)MFNy<9U;^h8I<2>Lk@)$)N;G3GY;(o9v?eo6Jc3X`I1$r&rxJthC%`grKO zxI{#U;NzO1DfwbnNF=6r?B={c1U1*cWfQDNhJ@qSy5Pczg}GUuOpe&3LCY@<#?Mr> zKl%>S`Ky}Dd%=%4nZtUNO;M-$9A_k1S_WFsT|JDw+m!Q#-;$zr83P4sLpw(}jp} zFDUR9rggf)XVhS+VGl^0w99I*_~?9ScV&V zU$#MhjY~3o=uD>;6=Ey97UXv3XkL#$MkNN$RFQr_1&*S-t zENZKozp>z7ZPmx=Z1>AAYY0j-cbd#GA`?w}U`%pDXnNA=tHD#+A6CE9@yr^Q2ha;S z{qY0~L*QyT#!)gS{?|MR#Gf|^@p2se&uH9pXgHdbYqq#tH)Xur)~fA0$K6hfn1~#= zBo_?YcP)jusF;Kotja_TfCT8gy}hAc8ZH4f+~DoNuE zZ91GQlvb+_cXlpo4p<%Yq8CIMDD)U+>1OrkIp-e7zdY>{^=eVKt^+6-b>y(di&{Yo z55r`I4S-&bMu!gI=fvsE-Fusngp_LYIlv-BN?3BLZ5>X@tkX0y+Da7&@jWFWCBtR* z6P$U2>YMEHsx#UCPVZs!=Rr}2z>q2lW-4U)emAN9q;RA`hf46y<_vY2ajcpR(GksH z1mX(-9?57mTj`01VO!Ycd5fs*VV$T({7FYQK|>@VLp)BJQgn~y3V&hb99GJ56fe5T z31%lp){YmHt-mIUxcY~JowEI1^A&gX?$jcGwY^5!P|=tzqe*%9(wg(d9IC?2U(R#2j(c9Ra`p0 zd3WQWqHmt*D~t}1^i^7{1f>&t9jRe6q($aQ+C{&I6>3wX1whV8yq~G_54$f+^p*EA zQicev_?y3tM7?kZc=7FC{S(6PnQu0?)Gj#jRURZnziFmv*SdJTsWiih7zI{*cQ9QS2Wu(hbT zr{54MQk5aP37^9+Byd15$H)Nv?g-T$VuH}n>YJq?8_5z zd4Xb5%2v@U{ug8R(Kke>FnDk`#}=kSd0$)Y7Lr_Y_DkNJ(~N4UZN6Q)E};%PC$0JgH3ex5w2tzOtfDLhB_eO|_*=P9ku)L- z2UUd>js`)h7FSSl>{QjqYkLna;fCSmNuWadhOLW852K8gs*Z25BoA@2W62F+tC#hA zLWY%xm8)3&eSP`sOYXg9m9f*UHL_~etXDoKAQSuaJLl}$;HDBapX;}zLnuYE z3yS9Au;yX(38=&)@bJ3P{U3YB55>otlb%=!cj?an49fQe($E+;AYw>R5Wv2 z7YjngvBC7Lr&{iVP9hko4q5z`SQK1~OK6(>E7>ylQ^9E0~ZAhrp!>b=-do)89JUo$JO`_>NDn^h8 z7TjsUl+K$)rIceR9 zsVe%TZV(Oqtw2Uk zNIyDiqi_^I%hHYz5XAAQva?^?+BO$q=j){v;6<{l*pBD3PWs1gqxC%3ATXIx;t4@>y+mhC_sI3ipsdR1uuqG7 zC;#M!`7RIWl1M&qB?XA-%edb9cPt3F7tV;1P2I6yx^XnZah8C2F_W$i;wvbZxs~K2 z)=uaFI;+rwC~1K6a#piLPgU>MTwvK5T06Iw&v}%N&^lihF?v|NRn@h^WAKX8s*(oM zwB?&ckF9c8!w?Wd6WulE*w2cP#`BhDY+@_3CJh(1et8uVJrshj5T^;BSTkr1O(N!d zaI;&KARTUMhqPPXyncbm-WS%78&4gmP7yP+RWB)dEXi1F?@wB?IpaQdY08v9h;Xf@ zCT&XgG2`^Qp`k2~ z3ztjJbm4w_pn@?7Fq?@F*#l0eBv_l6gVRn!K z0Lh0$ob%3im!C2Z0emuAbfsTDymv^su!PyW;3bS4jvu(z&5$vkDBaF9(VX2*r8eo* zr>9m^qL6#m!J%fp&CY4rWx?ewW*1pX5t^2!RUw~`c6#N$45n3>_%b6^7}TmWQL^-p zgosZ;HTSRI7PMvzr9Ed?a3Y(gad`6=PES!Lx{Q)yjmK4bmx`3npN8u@mc%nxu3ba* z`DeR4`rXnHvReRGSXHol*4NipBJO~XmnTNN<6z+EkA@$hJJp^$;$wt)@~zi5yDi^2 zzh;4rv2{Sh4^<1+Be=S$g(>v#RSOF@rD=3InYC*AJwuEoWh^v9_{x9z6XuQq`{}v@ zNUMAABMC8U_3!#AMcolr)}{H$^=2-!w~ZX(=LwkDY=3*e*%Qrg%DzTyt_E{dDKU&S zGuvX&HI2|bcLmAiq63CThK5re7W5g9Qm)D}4|ctFwzbt%i7_KX8?mFO?UasA0k-6Z zMm8ty%yGJN$L-OCHEBY$5f615Z%UsyoKWq8X?1RnIlgkwUUT#>USEG2F_RZEUsZqR z5HK=K^l0S}_cWayl0f6s;o$-is>sfOP1|am4m8(05_I7zV?U-g6f!i1clm?QoSf3b z|Gw1@F+!@>C$FYI|72vdZ#u``L|B|2Ia+DVTk?iX{^Em<%4srqR)nwYvAy_SN{J2Y zVbJ2l*>5tpE9V>dmT%(dR@rtJy&}o@B5YY(eI_%*CS9Hgc3mF6U2j$pocG?MU~baJ zl5U<5_pweO5?ov}@Q^MfkZw8Tqq&is)`5tT&M$CR$1J$AAqx;!zo9T%|KDcE)fk%*(RkhIPz${gA3d(B*!!0)olt!q-=0;-c9VNWJK#~ zA!}gRJG-|f{{V#w+~{Ib7H+~Bd1695P? zqM+1(OV;eIl(Je0iBtSW!XPX|h-OR>5+?tA^!{K}yLmSVPOT%67>%vV%cj{57^oW1 z4h{em8Q|Hy1IV0t{utzAgoyL%=V9#ix7Z6cJt9JZ6F{w6j1RvA&FW3>ePdm(QB!z< zqkBRHnmYWg{GP+v%=&?fQ+$cp8WimHzEl!4~Hm*8iY?Ex*$ z%Flsa{tFbvyi~AcYKPIcJAWqx{2Q$IJP8AMs>%I!LbNo|r2sjEyN-9%mo72sWxZ|j z$Vc6FS*1{6P9&HdR|ic7qvt2o`S{`&Dr-)S^0z?Z|CZhXQ|M@f%DEg z=+LPZB~L?}IH@%;I=GB~pWmiwc@RZK;(G6@R$L-w7D2|Zctog)tb|MR4#ks)W3GSl z@L4V&%uoJ_qDb)YLU_6BQ{rPum2Qf*pB?R+cn1!B{G}kUy8(t>ZFUH#(i%mf*@+pE zW3F<=KQBY%%mHPP-sLbxY{J)@{JDO(&UR0V|MR|J(cpSJC68ph58hMr{7PF;){D`` z8G!~JdYE*()|FR&JJ;W(o?!euLWf{RDV0D_tMn(XJ5ipvI!&zmRGKXuQA94Pn&X(` znCafW*)>V1F=3qA(Vw#7yRN;F@7nyN`nVioqmnpw$VT$GUL-zbyh=n*ANGF1DqD~Jf)$hpTP^q(P*_RWJgE$^I;Xu{eL;+fn`%{Qe!# zl&epzt7S@bVeWC${$l4TV}=j}5+?0YCg_4pYc4;9L~G5<4SzM-v}TEAjpe(+_kf9rGa=`aQ* z+ci|x(U#rg1ngQg0K8|2lF- zhyM9kn|yf(xP+P=>ElCPpG-Z50v66XaHm{Nk`Z%b|IIGGw#4l8RJGNgB@VC8;1O5Yz#uOC|CrJXO6~N*;gw z)2Smu-zUZcpTdEp?AW+>yZpmdZQHP@p%_WVCtcq;uP6Y|Ls!w7m!UL}*cljt23l{>_8-y1i#&K3oE(e2f=D0HEFrkJ3bN);Eif<@hs0jat^KjK4Cpg4WD z$4vp#IZd4c=)bI0uI+PjUveC^W%rfK*N$MhpA6Eqr_}7L&mU0Z9?Dbn@tdQUO)8hW zs-Q}st=8FROxbpp6a~Rah_$}&bKCZA?MW6>Yv5k_gU zCSqw z&Txe9W@!HV+IviSfE@f|{)2l8IuWV7O{}$70BA4J@u;?N7d=K(hLLr&RjmdmB!aIT55vCv`9$?dN zRI($befiTcC7HwVm3shk6G*K@F~aY}uWF0iZ<~Dhd`A1*B&%oes&d!^Xee>3%Hj8TW@a_N#<)ev?aJ=>v@eMSNhV@a3InWs8 zTPWv5v$$&nm8mRHf?Pha7;@?Ehq&L44KrOm6FkL?iHfo@xk`PDS0qIK;PWn5(eQdz zGvFXiLmtsrIwO2~9aSt6`kHO$rg*7e*Pl&Uv>hrEGB7uD9#EZ(Z;`=<@p{`7B}*1F zufmWWc^i6Wu;*{4sR0(sIY87Oe6ognEeN?pz?|&_v42SUqWp@LasTam=Cqv?BPR}~ zj&o7*N;V!^F5}9OQ-@n6Qt7CB@$8nkrf;LBf66;}%t1j7UAWW3qU>^NBul$L z0?>5_dlp}C|0Ej(k$Apf33RTdkW*fI`hWF@(=W!yNQ_h{AqNCqh=4!%O6f$JnPk(4 zMQN(oq@7LHfp!)AkQ{%H!LaF$m(^&@EqJL^W@Upht4G^Q4^?jB3pFBc9F%{r&+S+tXp--fsM6x+)+RW2<)m|naPKcHVZ1ZIxEsWuUh@kS} zNf++7R$w|%`d>XH{2KR4TE4oaM zEP7a=P;g~~%&I>`%>4Pd3&R&yUO??lV)V)`y<}=dq(T|h*t_40FD7QRZ#|Mxo@1dz zXJ~ENhqS60vg~*((}~ty(-z?5lTd}E$$d6nef==~jI;;=q;jy8_jpSWKCiDRQt>@t zgYv2Mrl0kBlGX$bS?(v75j(|o|BQ*yS5K06%`w!5*$jLJEHGzPavAWIplMwSK`q?) zecK8%E3PA=CcnZfLj1`M-#_@SU0*S@X93J$cI0O0Rz2G~e{1KrqL9uF_OMeQ^Pm}5 zimtVDu}Wyr+S#+HT>F`~ zql2JK=jT`UG!HPf{h1Cu3AN&C_M6KFwu7Xh#}9IP%cpubXUBp(IF1=*bc(>Ey^`8;<|LwnH8RCiVcs!3NN6pcEr0i%ZZ)JqCU8+N#!55VwzBT6v!}jebG#W%;y=F4OosmMVJfep5qmp_*AC^cz3D|w;Yq6P{aI<9FkmE%w zn)|@bNDe3!XsLy?U@Z)9^Pxpl-!E2QV?8e`M=Dwlsi2JMbVHJ-b!6!%BQz_Xw$S2; zJhI1d&L^ac;P;xN%n`N$_ZjAJFwURRRqefjsLeF_PAYl{hKPbW(l*)tGp}sZ?CYa0 zt=u^pqW)XPk+Wx#UMbcV9sB)y>F~Ftn)yg6LZb%Oi zNBchJKfLscx6feGAOxZr?i?HWC$(HoPl%H09RT|m#gy8znv2a_P(jcP3s_ls-zEa0 zZ;)tiWpW$4AfpI4jY*bsy~n;#Uz)WkhF?`@j=tTOxmipJDKf}t+H5(US;8_M1QZZl zlTBop>2+=L8Vx#Yt-*ecO`d%_fQQKqx^_Kg;x@B6r?NXOsRNRwL+p24H=5QLHi0mJ zBKobTsu!Y)!+1@{tnlE+ITgN<${`8hJ6@93sN&?d68|aQOv#$SoR^CFq&Yzn92x!P z*TUcmCE$sLS$;=^ho`kWx~h7%%C%-&=lDtba9F&2lW1IDp@%pI0fUIkfP^1P2WS0U zbwNbvDrA>3~a&0EU9X0TT~) z6!;ElmYO0T?2PS{%n(!tBGw_Z5>VLjrbF(G*!5=J6l(ih+)il4TQ^6}{ckE+_g4gB zgoK|p?wNV0ZDq#KZaQO88NWFui3zzhv6bQEVlFCgII%+)4$kmc8F8jvu8j#s#9I36v|Uu#rhK7R*x ztH*v`JY9Z#AmSXQU-h!AQHVkAxuCJV(h9PB4-{keE(v>**U@13^1z{dMv8Vrmn2Z5 zDU6z~56|C{mPTDYaHu?6v8=Ole zzj6X%tvI1=5JT9kaZ^7wxCgRy$avI$7}l#j9K}>rNn3d%J9m;hx(jV^RUI3Q$=fI1 z>|acShou}E71sH)diX^6+Yx#?)OxNW+Bm%kS!#Q8A`c4{%pkXlM5MJei)iHNNg|<`5zP2^ zT+KSUP_y=?;U?=iJ0s^0??gJ|nMn7%n?|D4z7U**ZSk%wYeO)Ivlv<Qo~uYyW0u_TNXaZUbr}Ax8Qthdq^s|q~WECXnTS3 zU1!DdAaA%1y^%UncRc6=$pRu2{_3A-ocVaAROwm%0K3Icc5}OKph7KQh;=I^)xKm% z7jFWfta7`K09G3(1hUj5PwP$7hJy+>OgoMl1vt1vTkTqPO@KldDb3VMy52?ClQ;O| z1`lYDw8G2GZd2JRMNy60E=n1$+xIR>Wi1CyLGMV)+o zJ?YM!Z&TaF=@1mg^>7n3-B;fFD&&h?Z|H((Z9f!NAaN_diXG6aFYl3NyiyhPm_oVh zTGFHbsMGQR3^~oE7ij?y#!q9Uq|9qokD3h$GcU`FC!&Z4+GeC4DH#rzH{~rpWjF?_ zJUpNBGzskt-*(yVbj4n*|E>Nt->-gaV`UN5F&02Z{IGw4pJ~|&A)_cLH%Ggiy=mr^U3iW2n-U5|3xK8_6~Hw?LAh&|%w5}K&>Pt<;R2KGX->EfNkgS=h!Nx4jN)`!lY z|7;(7Vk|c@_`tfvI}udt1^#T8Qgd8LJ2aJA-bosqggDbR9yRjl*a>3EIk0K^aAe_z zzGy7ILK&pfYgcZF?cjB%As~a0ZHqStMaV6`oLC8xBXsWSK`8D;$rF)9*p8w!yEo$~ z)zC|uQ9j%856HnR8YmxBA3j zG3c}$Cn&T!BCOZTfj3Jyo~N~^MXJ44kM=(8Qywph=vJjM> zKThR1WL}~TpabnyE6Q%dJQ1AdWD9`)=@Hd6y5u7I`0As5$DGe~eR1xE*|43);ErCz z%0Zt3qgMsVC6~srm6p`unt0xP-YAo+6i=S3+lwPIk)t_3efZUIZV-x~7?6t;HT{x>72nlgG(*F~0W67^+XgjDzft!L`Mw*bFMlnYmdUA#KZrzNJ(( zgO#327wz3_Jju0wyKfwKa&B(op{R@uCbk)8$&$Y4hsV)v30T#HRW?L2q7nVgf?jd+#3l?(}Lu<1np)>svN;dPY=O=4=4a|IU>uqFh z3j^Fgy85=9GIWc@Hgn%`Myl?{W}v2^ZHFf3o&4?+pG+$Z+H0#wMPvq)FM$P_awXS{ zdMva}->!Ui&N&1eC$pRg7cx5ewlT5&(m)c+I^X(G63Ade>T%U(Y=%n1Ykgmb4XXjO z#0@b(+|v)Af6*aY&Ply?WawKHGR=9L|M7_i@pO9JGA80P=yYsu>m~WaQ+1yzw4vaV zQq#c+f}=x<4#N_H5Q&_!S<=DZA(g-WmCrEbRRM(&l7VJETmkt}5{*0dWA#Dw$S}OV zaD2tAtE?z)w?fJ4B8*ZNRlcerjAqbCh9(mab>1a3-K$i+_q`W!4|ws-HMl%NoEAT*H4Z!6aAZv9)}NKf2Z+jB9ADpc;0C^=j=!t}UgH1018v|JK4C$(u|@tIUj0UdUe=?j0l{~$rj16^WW51#Vt?o04%u6tJI?O z!w9scZK1{uy-@oMvBzr)NfLF+x9UX5IfmZ98v1?-c}-z!YW9Lh;2c`^?-}n}g#xom z(@+^}t9vc$4nya56`~7;+}!b{vM~GWijT6HU6=7{X_5sD|i4 zh-b^m$;V-+fjM@bMRSf48KPq0PBQ>$HCg#ze>$Y+_XnTNQmA2fyUm>*Q4@=c(-9f- z6T!W!AL5r&l0!u0pHwY5<=wof??OasiqbJ{+vFKUPGU>M3zMR5hgaaUP(AlX(Wn7$&xdYC8G#P5)jErMI;GGl1w8ANNfQCMG1m{NRpgsBqNfO z09*nR+S=JWxO;d$@$&ZZ4S5zC7XCaUGCtvD zVp8&}*D0A<@3V7qKjeKZEi136tg5c5ZEk68Ywzgn`Z72)JTf}=ZG2*WVR31BWp!U!L}lF#Sa#cuCX-*+1zqTvvq=S2Lb+MhK0k0}=X|46gHD)u+MCIAvV z9B}gRC;&y^7{!wrMD)MMf2%PNJ1lf=X+gQqI*2`gX9;Zd22goPgwM}D_mzq`)cO8{ zu9(ItOsktdvhtM(jiv=~89_?Yr$_)6qWq(wN=0NSvjgCttTP`7?Ut5ipb zRj|EQ4Zy=O`cm+4Y1Z<0E^a;xVDV8Uf8Aj7lwdey5SLe zU4=_pvO*@6en^xf^N<6Nj|jIl55Cc_^#05C1V0v>lgxyz6|PC zwD>Ml_Dd1Rs?ruu@2TdJs~+J9MjDJkI>HRnZplmwMJzt8^P0v2#!`W=a&7!r>-6em zoX8h)LJBuTzP+p|PtmUwlPMaN54(0%(R7=?Gf*VTAYEDV^SK!_m*nX0Cr=Kx(h}i| zvy4*V9v{kos7)7`w?i0TJ!SF!fg&<&C;*QhZRG{Bn0v)B6xMZh4FZ%l1^0ri(b635 zo9h<3(!dB?KO!(nTOUqZCv4-Y^V~^Jq>(%s) z6wYITPIw-4tRD;XtC(p_T*^7A5Jx*LV1Wxrs6p7I`oQQOQbw$|d z>x|&lue>7HRWF6`F1}@c$!`%;89%(~e`ztl$I7^-zoMI?e1sK$K`!)>CmqNBY!Rua z=}MLD{3(}{>d8?X-MpHS^B4IZpIf0H%r>w9^&$Di>nrknqmuFJQ%(%moi3pcJ52Mv zJUtQ0t=}#6ZRa;BYnI}IKXT`B$Mm3xq&+Vtg}ctVP;X)cQXKQ67&M)PL=DXOU7j4^ zzjd8js7q<@uV5jUs5fHftM}Jr`NqIkG?112vlxIy0tLIMSl3+nY4CvU+EGz zb-M!=An_n|yj?Ral(K7XF1fX19qV_d>Kn1Ps^kgT+H&-K{}1}T9U`Qn6@oj8wpHPB*iAN*8$GLB?Q!AL zx(!Nmfz;pWV|I=L(-0& z7g^G}4pmk%rGf^2?Ygpo9So~u)arD4+860?lJAchVb{_iRfFXoRwRxJQiun&W03;e zf{*6!l}^}=OObWrtlDk+jb(b*U3c1o2CaWKr$gQJ(bu0buyF>mK z(0N{eaJbLQPkecoZT9#(j>=VeGKg}!h-Jc@@P(cQE8=T@?y&MKx@^|FVq{q0Ia~c8 zse5ZBGkNmot@->GkI(N+f~5#MxTD>N_um@5N8+{YBVC$bm`npg7SV$rq$hM-ns7AC z@V#Urc``Hlc`_q`UmYufm!X}Z`B-3#<@A-2;I0lv)_%uR`2PRH&MNb1FH$ zs9wZA5p((SwNAM?%E`f!MrQ9@SfJU|V!8rdIVd-Do~r4q5*^;#rv^}uq=wBm$EGU@ zP9LwmpPIQ|mCXA6DNI2ARBTe-p)q?t6?f-W3UR7v*S&9U-9&k}WP%}dMK0~Lr1ubl z1H%PEG(&omP(BK>dO$g7KPTjBUC*M)qirSJWry!Ya`NEhZB9f#D7-{zXVpW93v zaSxF?Qb5+@$UzS8-(t7N0tyx1Ytv$~9S|RezdIEfNb8H@8m;*mk_XF?I?WT?Gxi)s ztv6lNj0O~IvxZKFF%pA^r;h5F-T`Mp!T3i}5nOSJkyya6j^mM|DFCX%fZ_&=I)kJ zbH$KJq^8DRY5c5ai0f4*oeV7SzyQkZEr6{2LQt7{7hb{9yp#H#Ib$)d-(Bamxp`5? zIRtSFJv10zawtgrweqco0qoUR;~+w39j5(Q{_%npDH_B5PoLu@+~h^YEG^|A3$qj( zl4@mcq;M@v)5`57Ld2Xpu~Olp?yYJ zpfN^G#o?~{ubP4XZ(9ST=kA+oio!nmrSjj+jN~Qx z-2w40_fcu{Ae>9_V}ZtWBn*AvfQoH2L8zf)CdN*oeOMrizWiybcJ!=*%Ly2~j|Z~) z-e7#;rw0x){K^+~{=St5_y5(#LKgBKUIbSXNs-#b0*3t*VPcRqj|EOHc@(42;aEVK zij>7*K!v*3Ba}){>D+(CVK7`gKX}mn=0~u9zW3Ld*g3L9hmv}eobG(2{_Rg8ud{=m zF~4_%EJj`s>)i4bruSDHaVALjA2lp-GvKnzie?TcQ?UdwSzV>wd;CmN3V(p+guY z(yxU!kjjO0y6f?e#Yh`fR>)=$4(ae*$V|N9x11fvw2D5`~-Hc6-AHExeMX-M6VjM!J(P}E`QX`bgC3n3{19u}ZHRY7+d zkZG5G0>~7jVcineN{#&(`WRHHw$}pjGg5)aAFQ2oQ{Tr+%WKn&&IbtiCWm{}6>_UZ z&KzO%-c{J~M7o8oZNDG6w9T+k*68dWm!@XyAwjTlUky0f8UONuDE3_Da&Bu|n1``?|? zfA-8<+WDe&@5?;9lF92>054B`<2nvL#S7mMC1%f~*>GmjVQ12*Cz&gflPg*)Q(qP9 z3E90Cd)yQ+6r-i>b3cRbgrv>+&KJ&VKll2JI7D?d0zUG{qDd2^+WhIQ^=+paKZtwM z&6~XGHtevbg#x7Cd|*h1MfUuDv9s~E0nbL*=$3jAXsnm*E^X73wRq0v83z6Rv@&u165t6hTM zWDNP3p;o=uu>j5~|2HK7AON<^y&@VJ?U^~?n0|ZPvU*lXJnj|D5;JBYUE-Kluu__o zP&J)NI9X>x5DWOM5Mp+wFJXZWxakqm3Yvc|mA_e?-Q*;G>dTe=uh;I&6UY#)^Qup| z8iMm3cTIw1=pxlWr|UA_*UIfJ3izRk^LoA2Spt4(Of$|9o$vO3#$NKH9vb->l{I%_7Yhjan4XdHb2|Zq9C8A+Af{zpTuTL&tO#-_3X5yf5mY+vU7B zityk(6OoTw!M$#}sr0(E#7_CxN1QYDZQ>gK8scL&eb6B8`?u)}>uHYt^~xE>-W^J2_S4VYa0Dp1?_8aOH60Wy z!jT>xlbp2(;YhCPf$s#jfx>uMzUAka*4wm9ZtLjhyU?m+_{$AdXsc9Bt2+c=jX22; zfG;JOCqJzh^IGV!L7OdHco5yV#_rk@*T489>(h=2&69oPzCrcq;nj)RmI-s)NBm8M zS!@!=tZq=$$h(|P1vvy~%uMdPDKq2sQ6J+%1F8o+y(3@Vz7QFHdTW$=r{?>x5&Q0s zwo>on8}XX&?=)0Gg`7fgN<|r%%KDZqJe3W_Z=t||dGp@S(|8vAY$wh`>ETOGr68OD zam4`4h_o}WWaU5)1V4XIxBQuc36PqzI-w|cK^>W{Ii~Fjqn}Smo%#uSNQ6wSn}GrP zZ_l(6KxSc7w;VC>Rj}_ED}AGG1s)K+)R}^hI{{N+Yr~gcHy%9z!~47EzslwfVITrn zvKfSL?y;Pb7<$?vp%_&QehmCF$i%easf~E(skosw zIBPfYv493prsAE97V5FWJbj}01`EWn?AA#@QRlv4Trd*m(XrlXlmGYvakne05nJU( z5zyxEBUoUK*7^jF8~|CsG7wy5Sin}Wz@i#E*?*@#d|)98TpJ)QB_Lzd7E>V78QZof z822=7r9mdakXQWZmqE9>IHHlE4nZ5%Z~QKPWmHGvK9jjMKDqCfT~6w=kr5c_ZgJ~) zJ%6@P%!l>c>PK$ZFHACBlWxJ4*S2*StA)1-Uu+U6KZ@iT>o&jGeQ}jDwIuR*yil;{ zrr2Z~FB5Tw%j#`4MpxFG1A1yIdO>_ONk`mq-^Wey!!-u)$Esid>_}(`MrZ@NE_8lJ zh3q62h)mA2AS}Xk|(>Ila@H z8v$LLmIm`K^lEd3#JBq^G>H=%O_(xuvlyPYGjKnR8%sH+kJPl|^7OeIGaa}oimNu5 z67adBWObHM+^j)XxvsG;B2%iAUp4xQMvzZOqqeUE)othPMU5|?f5lT7g`z=HC$6#^ zQ%{QpROS7$y{+1dPZ%D-H$OuUtVdeh_#XK*Q0bYz`bSWgMH@+rx#7fsX-KN2U#6!yZgT~cj~iHoI2H1fd!O4 zSp&4+PUwM%(!ikwm7e+P&0iVsztd(w>i#EoqydCL{iIeO=~vNFsN_@MZqQihffEf7 zPK3PtM=)ZmkrG57SO8=w%(H*HKY-Lq4$Tb;iW$PxX#Um(ioP^$<>0p^obiWTeICQ_ zAXRJo%sQ=k+tDOug4f$T%i7U3?jR?zbvm(kA@alvLrH!S>=LSv1)6fP0KMY;6@>69 zKt;vB4y2pHBcN-Fd01c$ycBd5ue_5nA*giBIYs0^XUhcp1|R?O zEK*Aj6zAHh4i>;oN84e6XDzMp1D0Vdu*-`D+GACj2Uzs9XR&%hiVO7Ko9VYHo|JaYy$&YnK5bq&~;(Iwu!1e{(({c%aQ?>PqSe)>EOb>HU@S`mN1R=QGO*z61Pl zTJeDXP-d?$SaFYV1iD!ZU6a`GjINToybGeF+q%E|s{iZUGb9`E!2+$tbP5S}wOp^- zDl`YGmX@>hJ5CRLiI&i6^HjT4&CB~QCEvMvQcfwFKmMloT`^S`N{U1YQ)(Bs<#Zm0 z8G0n?@FLcfxS3>p?*t0hAP#C-tuEf^zPFOsbu+xLS7Aq?;-f|Mn}$hw;dvwUb4lpo zpw21b7!%Z;J1CHM+TnP#Og-0fho8eSjRAjA?g7P+JZ~Uv-@@F4zg|0Zo@wUCn@54C znB$q^yVVz7F-{SF<&%3`={Y8Nf7+HyL?A_8(Xt1^M(}-s{0E%5%>E&^p@?8dXgSj--Z~m zpAaQ41im2%&+&#Tm+4an+1g?Or$)V+1|9DGkL>Se>;kXuCQf}?H?={#CgNVbZ0wbt*MbJN0APe9qoDA}pA2K7?q_6*y z{NTwTo19lf!;aDqzks)n^^ffLHA7EWwiGvCf;Shx%jey8(E&L_KH|R`>;6u3`W~kW z=;q=;TNJhou?#>YU~YT>VYlMxx%Cb=o07SgdMeKDpsoCpf@2=qRl{7PFFy6sIVEVg zRbMs-LkCM(p|;aeq`s-vg1yiEL6Cmuc0ElXX#{GR(pG2Us356nGBw zNWb%-H9NQ<1c=Fkj+Krn8_V}5egZA}uSH|(vzX^l!oO?*cEV6mjKTuxAPSSP*Xqq_ z*9K7+-7hsfn|9cq%}R8|1B_ItjHCyDSMFD5>ON#gi{fsQU5NG%+i|FGsZW|E=Jnzw z=EjPCJtYZu+U7J6SUs~=3fKH$G~r+QaNob-byuHO`l_CHRm*WB=O7j++d8!Rbba-> z_1elMNRAf?`KsB0eq$ANc}d8;BGI_Fjt0$>j-JHW^<4fC_5fM(Yw8QLL?*~84Vq-3 z0d)qKFq=y}v$qx}s@5{MkBRw>_Zh50rIKlraxD=gV@FTw)d(1JbJS(k@Und!K4ymM z;(lnt`;?s-Hi;kq4Rp!Me_bxW?`!$vqRc-$8zGcfctOY9@Ak{_1XmGL)!9Y4ntnDw z80t^Wq=8uW4$bewNxeFINpVRd3Ik^c0hwh3LH&k##6z)5-PWH$F{f7l zPR_r`E1!F2Rr=VT2@UOLsTYu-N#*hUHT?YFL(1*PcrA=HMifr3!`^H1o2h^5Yn&9vLxh9&3E{2-4=8WB89;x$8f=r z7Dw;{JTPCzg@Bx}rG$#Y4aPnBWDIRU@hJd;$EK%Xy!kB?ZXCk|xkKlau)u<2`-iMJ zh#bu3>;uK2%{w5N%V3;EUQk4)AU^UeUF%~Xht7LVfJ=bP1x?6;)&{fkzdgO)AyWE+ zd*_kWSt-=eLO-lfIAD$EDM(aD6w_zM{>O!~g)KVecjlc&T$xhpoA{Gb`Bv;;s@d_? zY6rGzWqwDSv*dAtAGp+Y?f$x2&zi&k0;d1*`7F-96b!!PH>$WfhXpR1FM9-@HHO-6 zE59uPt!uxCSold;0I^T|WAS8cjCxNVj`nZEATgeYZMM3j>AE0t76X6%xq5mw^@ZP& z=sTf36h<8;gJvj>vzJJvX93~)9j^X##+mfz1`YHNS#P5c63jK(PjgS*e(ONuK*C18 zXZ<_4O0V7fUCuMXezV{4Eb6|xtyL*^5iC9QKK@_t@=b55HIRkIDaml!Vj_R)SIBZzZCi3ZRWS#XqNtg}kqazjc8#p?u*d_5M`X zKU?4^*!ZacLIC>+O1dg@{deX5a3+6U9{^6lObH0DASXnt{8@-&G|ac-)SqRE|83!D zhu9RT*!^EcEsiQI0DA?WW$A6KE{@v*nSd4Hx)oA8{*c5182=*Z z&JZ+|WgJ|M`^)ZFAn*;EcgoPllyB_F39JL2fYiWj4i+F%EOWsEt6x?zr&370$7Os2 zhVVbO{P77U zq&V_`fC=EF`Jap$^?-ZI*L)RIOb;AXf(?!kASHsOzz{bUUMyxUGX*8kEVf_o$XE6i#w|7@59SN_>a=Z`_Y zDPBh%hsFme36!?Z2Ji4LO_$?YeendYU9FTXaysf;THj`S_TWZTh8U>X$w^4 z#^cY(s-3o7JDHw;3|rE%?)&gh2Exp|{eQSUuCx4sS25>)R&SDJKSMbru)WO{xj;wF zq3p;z5gdfJ3H_15qhcE~H<2eIs;q>k6U zd|TG}s$z~qdW$xfpknijX?*5@c~M#Gx3)wIuP`1s@jVEIX}SF)lcuF+{2Q28r=*VF zIvw_f)0MCD%mSkHwPM0Ao5yD0bt<&i@$Jn<(kN>UAM^zF`*<-w9J5&zsLw-}^56~S zrt#Y``z#qR#AwQvG~w14n~)InXz0?oS2Q#DqR!G$Qsq}wQ$UU4hC?^#Fb*btNlkcj2V9V-(cI-Pywy<7gYTJVKIFT*v2`fLFOm7`CL(u35 zHyvYvtw&a9Z(WE;t|Oen(~)lHQNb2xRe~(bvEy4t4*^n(u$_f_X(B@DcI;D9F9tmN z6gvz&SXkBqnRkm)ZFtibmf5=B6Os+8U*&iqdsGBdMuxoGtm1d1>~?yL=I|n-^*O-5 z$bjOpr(E!_c~UorCWSWPPpIr0oxyo*C7hgi}l300+fq zhGi<5v*xvNR6{t-esgY6tHD~zc`pT7D~f<@NkV@(w~O>-^h%JhcT_7)8pB}P(WZcV z<<;SRc3*Nt5#OXfGNQOuxPYy)v?9>j$lxR#P0VrXU9@8`Aim|yvpDQ@+cTz$Ynv$j zJ|F6X0a3;|Jb-K%r)iB@VuWpMObH?J@!FHsNA9dm_kC?k`tI{r!QW1-k4X{74BY4y z`+mMy1QAtKh!DP6R&ugh#Zw9vQO}cBudea;`)hJ1&d2DT3O)cSPHn*FC~m>!IsI0a zh5cw{!u2BEN(vto4Rgi2BX}gs0T~!~8^L<$!lu~gG_e7+$h=C})^hxc`ZW5Lz5oQE zj0e^?Eff+FVoJl?Nz>6s@zTw2LUGhl!LG=KAenA`k+Nq~L1nMTzT#y(az_mh|6JZN zgv5AW5{DDcPPMk#@Gie&iej`q%pmhJMVu~8l-XfuOKGb#G|F3@b=AztuWj7o!wuxQ zk>(>=YMnqxVY@zgG6FKX6IzCWmuP=ntFCv`Jf0u}PS2wqYcj~Vn`h)OBXF5p#g0cz z+P84$LfvHV?vYn`);_JZkXicnv7$ewKJn4r#;cn`8VS%*Wg7jBCz~gjBWID)n{ZRa|A&e0!In=p?oK>rj>^DmCzY*tdi~g{Lte}SGkFFvOt&d@by)$I47K@}` z)IPMiH(uCnLEazSrma!`y6+S8ajU6aM2QM*v8exUN+rj>a$(2d4TJUP%VI;UZt4dG zMhOcZx^GGajd7|c&qY_tUtK-y@nZRAYJ1cw0?ju{%oBsByweKoy)uM%t z0}Jpi2Qq~OEIPT|sk$vN=?$TSxG#}+(!&wHT6QV^K0n4dFdESIAhvKlo{fJuhy)4im~P8X6Y=A#m==AiRDi?;r^WiOC)|G@~*mF2*noEBi)JiDkD|#JHkYy z^(3D6#mr_{5V5d@>(-^O2R$!c(W(HJ1G97jNb2T>f!wA?SiR#6He76@X*2!2MN zqov=0p(WuCBah}9%B&8fBjG%#T%A0+Vzu&$urGuBW*+BBG$^Q^;J)`TVSnIB5`E>1 z>NP5y#l=d+TpE+=N*K-xJIr+i8vOfrM4? z7B1PX#7>Vh-6DD+7H2$5=29%dLL}I8_%)eQd|Rq>{3_i-t-Qw27RqAe)!iKL6!})$ zOLRT9NTcgeGCL9bKGCn8U)x3=lil*wC4fZuyLb{5i3^jPMm5}gD@&26ybjZwN;~&( zYQ|rE&}jm*?Bske+Bu%{4yPKlF64yLVk9Rt$9ThtKBnTmtA#nq&CS`j;2XclO?`*t zBvfFsyQjzey;xD>G_(iW4o|8%jVq*$UT0!ahwp(bSCsU`%uIU;I$*&`r)4WwCCA6o zsarIT&AC?uwrXNUSeQdon>v2`lP23ozqq$pKtvC%l_G(V3Qw{M5=~$j(TV2-` zHTs~yH?LR*YqcS*6IRdX?8Q%xMFWs;ojg#m4IVtpKd`g7zx_M2raEkJLNh9Y=BoAW z-mQL|+hgheCc_V^`U@Xf-&YtB?u*^1U)*Qk%OEQCo^7?{XpEn1PI-Jm^|3Q2ke=5y zq3|}vq*#H|UgS;Qd{slJw8xF>N+==nWsCGMk5?X6bnEGBLYoF&5!3=(ylnp;@yM!+ygQ7_L@Ze^4yuvAt?+z*qqxuF+CVA!A@IET`O;O&m-4_EM3K&xi7 zA773$z>i6V`&}&r5#aEe7X=uitLsJ{wtw5bsddoF7$3kl8!9J+VFWLtB22jlbVqnJ=FAM z>!}KI4f0Y@LO;zQzcK3RXF;x+f^a)Z66HYSA8nj;!Rg+rDA zzWHa5I-r+|@EGiS>`JxErG9+k%|1rg=1c0Xz=EiZSkQb*T_f1AYryG5Yn1yHpR=ICe`UeR>WkO6pV!Dd)8R8$cOjZsiR6|%q2hex`ol2U=N!Qg z$QdIIULLR>QZM&pA`_ZS-;c>Kl76svXDnx>VfDQr-Fe6o1Ndh9q_=fv$1{m(!&Xw4 zM4gyikJVPx(Rq9R8lxP~v*DpaD(x~QPw;jL$v5xF5#OB=7+Xb<6?-R4nN?3ZL-?T2 z7E5DWhIJvi>!T}UTIE&FCD)&W17&X1R(wKT2yU5OhaS)rp{mm(j-*n;!&xK0jQYJk z_od{fwUE-p)SNCv`wWCHB+ANcN7$MAyRowu);oM@!M~6d%y$Uj&pkLOmPANJy2vSi zU?D|Xi_)Gr3*AA)&M|jyK>4Z=Y&)%_Br?u%HeQ_dYg&7g&Aq;h&Poe4J7f;Evtf?{ z`|Ri5jK>cDga-W(0UQa`IswaE&IJl`yBRUmtCe%2g`308llV+vJel2oyiq%sL3q+_>a zGZHAt_V&6ta9`w`aoYuM&gWAd8yzY07~Zu?!-;^(V#kZuo|HX%80|De$Uzx97h7)X zJ=BWAXX`fIJ!~6lmn>M1t0upVLwZQ9aU!RVpgEejBtACeQJwB;kAuE-lOa~-t}f)g z9j%)c5~@>|eEQ%MyU5|ym2wAuGrR9wmn)y$zxsj=cJ~UO>6urAcoWoCK65~4Zo^;Y z4k$63?S8yxlkZ5cH7hU|M&z6NJ_Oz7o9J^K<$4x z&*|mXaQ3=nR}X(#~6CO(Ji6 znS&`j@kEi3uQ|eM;3INA3sDx_hOXI7WXn}HHr&z28?7F_=lrNs-%HXwo-oU}c~9wi z)+Z{UM_IRG@ViZY{pN_XF4!`0Lsl7sB#@S6t8CHvnqdrBpP{9YTBu zB+jegN>ls9eU-VnW+%dw^cm(-_3#af4zEWBNd~&axJs<3!-6RC3&`Tn2vS*tfrrC_ z3n%x}M17x}1hK-bLktd`bXsTRZwx&;FnRRIL-epkzi0_%kQYH;!B6-`wd zEpips^dX6%p$9jTBx8m}ukd~Gy2-D7=Nt0Z7?~z{jJ&nEB`+=1#FC+fDF=3jG`Hx(jbsOl>Y1o8QUgoS1+1{_@{ zaeDy)lc({28Qt#$c6jBxTv-Zv^Thm7NrHSK3?dtow{Xz;Sp}z*8c{a}3d`4SEr(tK z_sURg=&YJj0bMq+bZI6Qx5dxBel>@);rLY|##Rr3`QRZ>?BQ$5au5|yNG7jSm9cXR z87K98X9r&T-D7iHeA(ckqc_vD0-w*U9?Xd}F$+xY*M{Q1bTo}c(C>Y{pX{D~`-z(! z-hjCL#XhCM#gZ~Dw!N>ubzWSl`F6EWY90w5GTceK;#M=4cZl!Cp9=2(jjWh9GKvK^ zl(iU1N`5cU+G7tQyo8sCmy+B*!Cdn3nv-XiWfxOh)=3oL^l^1eA!{}gm?;@J*peJJ zGZ|N=QnJQ@DC*B5ibWCA(aa18=fscpc9nY?ld0@<_`#yC+hnBnDhNu8(rB}dK$;rA z?732BrH(NMoi&Gwq?jV)Wa}uZvg+ZeX+X`n%a8B8AbX6nnlPLK9A0qCKt-DV^j2qm ze#iJAhIMVCAmo^(juSrZ486`j;z-;UNFUOQQZ~KWEd639%-q%y`gBK$A@cLE=!-ki z!Loy6N#7i2gmyjHb%x7LJc*4Pjk95LWG&wK*59H>tS~#NuF0wt$rcK z?1xtlK*f|O*R>>I(N$8q{{55N-DK%}-!UFLkuvCD%ZSGGMqNJ-D)fxB-*#=&)g)9j(oYl#+g6sPO(_kld^eGJU>BsF{IZTk`aB=#vjb z>90>}$Ha1x?d?My$7!s7ej!dWpNxlv5HDo*V`nP18gD4>cPZ|~S^B(Rmx@E13_HIc zS~cS`8W+2M!!~ryH4@nNaJtRpqFyz8SM6pl{>QGE7W`P(gZ$yA*Q0p=vgV{9)&`J1 z+2W|H`^^#8q{49Po9!zUBPQT+fS(2KG{#mQEZG_!Z5jA`U^quT*9D=ap6A88?M3LX z;cX?bv{mE9Lp26JAzYq|ck>i{M-zKeQAG0<+y*zX$b(Dmk$_-%H{GKxHj2*|w+vWT z3-64eW#4apRxB?VdC;0h2f=GPsi&U{_?ER=St`|W( zY!e??zUHCtq1g}Npn%;HGgV%=RYDc0g`uqdVLTUPjvM$!3^nL&+cMExwvdigM_)T% z_064xFMjJo7pECekwoJI=8NfGbI=DKt zoh+hJ-MNul!aUcHyLKG6K`);=rFDrj)Yy8xGrwXP{Bj65`BMDIBZfPJRW!mKR+qWqPGNQ?u56WL0JE=#Sr=Wz5T9#4^Fg#EYl=x-7??O zYj;dJ(ryVM*fTLbYff{iS4&N9^)w5eKfr_cwNU`tF&Yl;q+vNJ^u2#Na8N9Ve3LV4 zHMQ2>`(VJQp7JW}&-lcgnMov{Gx1F}10c6iCiLZ`Y` z8RgEEtM6gmLW<#}UFaC!TX9+6>_$Zh60}$;WhF-50QN>#0vooovL?gpYJ@K z%<&8P?il|GksKi>>yYxKcz*Lu^0P2<1jaw!}ZZx)UT!aQ}X+BGMwjtvS=*Jo-#HsvVQHF zw&6#yO=^)DBt1)re6rIbdC^4sLsT`f^K&0Rt-!3jg<8IUqwm+^dVRz#e2|#oeBQNd z8@va)5bk*>L}xxgV{2}>e)4=5+j!Pz{WLQ7VKNza3!8uN8uITu*T6FbJN|zFQRO-( literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-11-Quarterly Growth Trend.jpg b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-11-Quarterly Growth Trend.jpg new file mode 100644 index 0000000000000000000000000000000000000000..fde4339b15a0c8e50f087035cb3fe5a618eea5a9 GIT binary patch literal 23122 zcmeHv2UHZznsy_?fMm&WRHC4$NER3nNg_FC6p$tWsHkQ{^&1SCn$ieykg zGLm6N(g4HE-{0N6yXg1buzT+LclVwevt*D|15D)+W z0r(HVPXO`&2{AE*n1}=dfsm4toFS*9ASWXuXFf|yMaRwp`zs$b0U>B~CUr zKCz3Ju3i%s7KZXlNQ({Mc741 zZ~-_)LqJGFfNuk!U_Xfn{`3L<{X=ky@H7!IgoN}A8F)h_HE@c6knq%LLL#Emr@^}e z!M_8iX^76Aza~dat92K0!Ie%ZB=I!~=gqQadhG!e*Y$gDp`>RR7@3$^xOsT_F7gYD zh>D3zNXp+*P*hS@xvitCr*B|rWNdZc`hktDoxQtnuV2`s@QBE$Xhc%-lcy=q zp1(-V%FfBn%P%M_Dt}i|S@piUrnaTEt-YhOtGj1#Xn16F?912j*}3_J#qUeYE34=q z+dI2^`Af!1>bp9IgSvf7p zU02!*LLnq{Hxpl%HIs5)*GAFba~n9rz$HA(js9(De;V1}H?YwE*2w;2VE;L;ae$nV z0GvER8UP0DAM#}dpZRAx_|v9J^1di@)(>C4{3<3q)%vDVDQPjyr2OUT>Nwje2D26g zOWkvJE-G!sM=HXz=HSWb>26S@pkTNyx zO0p~iUqMM}$<{eq<32jHm{N4>MXZ9VED@d7Kkwl_C5$HOjHtgM`r6^am{2Gk&AwG1 zS}edoUd=;=CP_hGk%_T;(Ln8;OYd2GJNhKUsFEB|Olt;MD-Z1Bp;YJb=_KPOiOTgJU9R z_|{WAaAr2ZclPr2mS_4Y`UcM!4csX<;~39;fT%Al{Xo9HBKW%OUFzm+i~JczFX>=r zUGIIPW5Wa9;_NQ!7kSPqS=#%*@4WV#?-*E~VdRVSmvf%{X+tcTA*HZC7vLYdtefSx3Rq7WBh?@FrESfdBnh zRDNTFSdBNgek=!_9>`4XX@UybMHuUR>z>R`5@$>GdW#HXUQ4NZYdEgp;CQzE^lcg% zMc`CkDXghMR7N0lj;zkJYS>YoJJ!7-?gjILdo%qPT@kBWn-!p-NN7RuWuE#cJ*fCG z-|)b>US3=s3>PX;iGpFH4#kfBa_y(>{DSfd-tUQy*+=pVo;Ny`@i30x=Ul|->Ch>@ z*>K|FIFU0&(g6F#Y1OkS@1^?YsiJV*dGfDK8n>JpeMN65QW2tIJ+;sX90#rm^@_vZ zv9|6Q`*rpDzS0crMafYKnNmZ|k=oQ#D_0owL+Us7>XmNh65#4pNu7oiG6xqV4<>Bx zV~LHH1qsIAqr3-a*4hz-#?8ot)!>1(u{Q}M&CJ9F#jS&V_M_F2~5N zbus6pMGW~d%h002--quoyo)L}`&3h%!S>E7bC13Fvkb@D9tZhwPNFG;k}pl8V|tsE%px@JK|5Y!|_S_fAJk3EM`hZ+CQ!Z z1-6xMP+NU+`+^6I)>K-0=5^}I)^xQLl?0S=A8-(F6mb=$DMX_^&!|RB=BVbR8({@3?kr-x(Y|7f*Z#f^textNrjqpiOi=WH|yy-i9L>| z@)nq>!+n`xzhs44qLEV5tQsl^W)#lYdqW{xy7HA2 zYh;FmQxEY}Lp|Cokiw9*8l0dAIxPj@!ySX#**xwBJ%oEOs%|hi68n(AQNi+IsIh*s%$j zep6yrPI!0YoVcctLvy|ATy=m|Nqc^$u%vzrl*6j?BfMflX+hGU-n851I=keY;p0ig z>FO&H9?LG0^Qns>&JpR1)q_jQ9y^s!HJ_x4JuD~4k{YuPzT@a%~O2;?juDn^nU z3n*QF7u(@s*`aehr)05W?l9vpVMaA4)5w67(jV-)Ps+7UXA|U_yed(EBe(aywk9^N z|6{a-=82$>{%P{~0$3(ec`Sf7(Z1D)@DRO;kEj8S(g!4aZ_p|cx^>8M(ORl)NQ#_?!&z$zsw(p2Doj+>X zpExo9%ZlWEQ5L_fi5Ah|dgD{9g;;r-cc*F<&CIZG;Nk4Syu#`RTP6&rD70 zDPJzi(04M-wvri<^*5wT3mx}-(-&%y&yi2ZiQRJII_iu;a&%^VavREuAvCNTgU_Br z9nZh~Od6d#-lMe7gpRjDwA4cBJx%n7Y$9Cg!mQpJW=~2_TE%yKC`AOcxG2$}_M7rs zyoOB8m6Pll0m#>$jt^d8(~~2O?4Tj1>Y5uE)|zsebMh2VHSL9m7IHNq-Mgjb2!ql2 zsua*Z05JSl&e)~jpS93=MR+~Vi)xc_Wo^|yzRiNfo@KI1vLUu&l|d=0o6rZsc^aJ% z80gTD^|)?QLhp?)MMu~?lzmMUV)6E|Mp*?+!d%SldPtIr_~Dt|5zQhxa^UJk+t+oW z`hUcbDFx;lvB^i%hbxS@2!SQUA@t|+C@a(OBwrTd-PJ4vAMopeG{dm(?@s4HkF@I% zAG~M0r^kX4;2XDJteeVkjxA>)+J37G{BJ%KEsG3y2t}4GP_ z8_b*2zciKoYemmlVXdyP@$26jXFYtJaRIv;N4rowz^H^3jHkc@ed&dGASfFT%&oQ9 zW+y@_Z0de~Qar|tgB1^CNbBH^mGD6O*rNfZ3skm!UU(pUV7BpKdjz&M49A*5f9VL4 zG%M~F$3Xf99w5tZg>Nqv!KVvAm0t_{xgX}V3T#o**}osL=j*Hht^VDOd((F+tA@|| zcApvS2vH=En(f|Lv5*?mLrvgrQzqk%@W66a&deJTG4~aIzN~W5lTI9QBmxio(Aa{n zr8y?*%26n0Ivmu+NiWejW|gZ9r4S&Y^~BnKrT$FOX(D&3R}YUwZve z=k{E3xwJ8i0pn$aeKIDzm+%S@WK7=ZeHcn|nk}q^;oW%b8ew z=VwVCf*G-R6vSP}_zvTAsOZY_EX=q2WfnOs zbcR~ZXNM%r|VjGAQOvkKhTo|?=@!rqY6Zx78H#OnJciZ zM%-ijMTa7cBOc&}sNjJ>Jg{S;?N50lLq$u0K4kEuCjDpEqUIuKftK--_-MwqLtX9L z;VDYjO6Tbw$0_fKvyGLHkY{_5qO#=41a((U?0HoCrz@~1=j{8}&&N%QH(RP3;nE`K^~(Kd>WovT z_)((8uxKkZZv^r})iA(uPqm4Q{Jx74XDEfcftNu8Rbz5{`o#~S+r|m>@2X-J7(2wo z03xj~21RSNiy4s>FDhBbCfKYNwMHh&=u*b6l5`b1h_#Not>D;(9C`782}L<) z{7s;%g#XjRQTHh>hJoa3Ie~Z00RLtgI4DQ zK1q$EEj*Alp8Umi^6)l-?+sWA7CRZWwvt1XIZKXQYbS1%3Nvzt#PpaquI)k526ohpv+3hd1oFWm>~R z3T-pV`J|XyD)(l)4&cy6BGX zyXu%s$k{k0n8>qHXSQ^nCDP+)a^W}TBS>mzLknwnlq=4Y#|e?_Tqs|!7sy;0gWJCw zyfpiYEt=u{(?h=?dgRZh>`8ef^k-SO_5wcKc$C+mSaR`&?lO3b?fsMY{rQ}JBtR4o zB!z*2f(ATrc)9p#Y4PlP>=Sn0<9$3Z;Ha`Lz8T03TCx!36K)QI_5cI07Tg~Xm|Xy! z*X;pl3wi9XmyBDjSYCpbv;KoWKkei<{RxBFAkm-{OL=s-HAr<_s3ZM2vd!<9PRsDO z%Kw;l^Uc3yiUhS$zyId`h%;kvf;`9mE#5`uWnzgA|1q^J`vwTwCd#U@D)d@$2W5M@ zbjF#+&?9h{Mvf5>wlhTD?zV-Z#E{ zhfoDi+>E=iR7rZy%`d&%^m1q0^BlBhzenKZ-HYB2h2AvNLFd+e*n6$9a_Cok`e+{q zbwkCfvo}?=V%rypDgPjlaBM?2V1jplWrkqbeUL_q(9;oI8Mna$=|AAp58^BO@PPCJ2=HD71wi!=8<#!%UukwG=EpPA&{KgLe1W(NCR`vBHS!QWT0~;%J(jM7vnCZYMQ}jWM$jD6fca`gR_$ip#i$6nSini4M$En=9}n^I5mz4u2c#rpb1e zKqo=^vky`h?HP@Rg_O<>J-=fUZ5>=Vi;@Rqs*zXA&9;Xee@pXbC)X+Pse6@Fm;Yr z$49X!yniHR?Z^{z2gsVdAV9nL)@o=1F`&^SFISt>!$0%ydDo*K&wod3?h}x`F zC;i>p0D)fYwHY-FO|;b1IrJm>&j#P+7W0Z^SOx!I#?pU zkJ?!5IA2eu+>Tv>wYg{5=R3@3!DpE34Fq#NJf)E?jaMhac?;LlsU(n=XhB5!dDQOr zl)ks0GC)*#hIGN6BQM9RVZv=@zXtGfB-4)9@CnLaAyHSRsAD>e+15?#b}joBAUe9~ zISpwQnSUX8-YY!E2^nLV5kO<+Y3XylwGs6g<3u6x&HUjD9_wY(-Ojf&LG+E9`@$yn zpt3xFa-9?XrVIlX(Mq`6g2i~?A$&*MV4!mQSv&NAuNHY+2C8cn1By?p58c3;_V-Te z+LPyjc6Y&W;F|N{LPnP!0(r8qBpG@l2uiGq%f$oVM1F__x?eebh6g;xx<+!Z^QWCc z@Rg9Cyc4|eZ{gEgWW@9jO^=NIX*{4%=~np*xhRy5Wbex<$U-!&oMdqSla_r!so?pF zs-(F&1q9QRKT(64Lc4OsWW|9l#PUqB0C^~tl`Pv*M7_dnBaPWN(-qY-Q-tL=2Ogy; zu(J-AxItNR-r_#KRlJrad*?R6{v#iWzA9IZ(IAR(Q?ts?u?3YW^BLSjexPIKekchv zO+FXP80*XsXr<@y1j56YP>KhNXUtDeh8 z$4BjevHz6_oNZRp!H=pg*~#ABwUDV=rt^q#l9_M@@odck#8o=V(CR?d*;a=JJ~IoW zQHqgTv5ya`j_CN7Yo>g5=TnWvymV^y6{%Xmw#*z-lYQ#iB%Ut7g{CARQ?O~yelL2} zrYeS@^Oz81@0*Es&r8-lostpE({YjgmImcnmuw3V^mKQ1$pL}ju^?78SES0DWy!Ww zv*E8o;gHkmIn;8@JNdn?;^U)NrRZ7gEEp>TMOzok+k&&w!KA1I4uLNMNhs$KOH{Fj z=j9>uRYo?I8|0QYT$CycySycy6*-%o^uIAwmETJ<`sg_*{farGiPs2s(lr-DrE;OK z6EQo72tq3-#4H||eTdP(1#QAm?4VOB{M{_<;HVnReeufgY5!E0)=m!&x#cS=1A9CP zpSc0j4KkY+`1Zal9thr05}>r6+~g~T;u!iKfuv(|9uE+V$74)nXhC&V7mtP5OBlKf zF1aeW-~m)s0BC2yyrCG%Vfcv_ds<(3-#`dN{`BzdpIVI5XZu()sbvboUd|u|Db2S| z9APE&WP@&RznguY`)k9=ei|vX=}o8a6WL4eJzi-{GAQ|}sHfy+gG#>v^5IVLzd$ny zb{Jb=2h`v~M@GZay!BeJpz=+mgdSp&55_zVRo(M$oC5<~(YA0Fsd7=drZ;}Ey2oosZhK$nGp4mH@z-^C`tYlLg6R0j|f_!t-N<{Lf# z|2#L47(S^=|H+MQH#|TCUyvOElW~$j&%A?~!mn4Bt8d1w!`t$V)*W*bcDr)nCkwLFzh&xVf@H>7?ceuYe8&csF7KNYk66ARdHCSPyk=Ra zOHA)JWiZQXV_60AkoYSem<|U`aoA5K>h$~`VS|zBFXINsgb(q+Rtso6Kz}OPZIB;l zy!a7Izw3Rn4?lHCse;PpdqB2ApxJS3F1%l~4~M;i(K>m)LHP9`gKxmX0B&Vxu|x3= z%uFG}(fw5Lm)#4p`CGl==$HP-bSC{8%3rg5Xt!5x%V}dHu_RbORW>-SH6+=*^_PjU z2SaHBxR5IN{w%yD{>1s~y1G;aBg07CSu+SO%+DD|#ETjDdn+e&!DqSq;mcxqc;L5v zGEh{~GHi!+2#&f4>&U}(3^p#af&qmfMl9b?g;P+BiU;(NVx+=2Y#CWpV1(^zIjog z82VRdl0K-v?|<2B|HGiyg1tOiQ4A$gpqR?Happ_EsckgE%Dlr)%3+d4K%>H6ikMz{*4xy05{_;pkM7kK;^n1NC1 z@2LUoCq1KfQY)W0*E|h7se$7EK5<4zKy^|BWuH`q!+(uc0Io3F$`AbeO=B*`t6n;v z?#D#URqInRZX4y zy3|ux$GRPp?uXcI{0kjOP8*42GMfBJ!T;SnB(wjgdV|QQ9Wl$|61(>3rc*0I8cW(> zQs)WAR-C}rP|B4OUq9SHhs;jwvOo$dzl{t@y*97@Qa$e59Y>y- zg>+&itu0yh4f;OJm~Q<*{T27Xd#lME-Jv~9mTw_pyTvjL4K>5q>v#QqnoD`&-4iki zyVv;$%@MPejZv2=3&z#M@~d+Fav}wiij*{gh54EpJqQ{MC|CGe!Kr<`yDQz}h0cA9 z|CFFnr2zPWTb5f6r*+CAehkWS(F!iyHO+V&;NY9-pg-=IlKMWWkI5x7tu09rAWVNk zFs^0p4leJ$$M9vfT95>TuNN)S?_**yD)OS#KzD$9C`E*bd7$HmU;OhkzzP_;{3R(p z)J4b^$f~|C~wNLMba!+o412RdMkrFU! z&tCI6x!Fdl%JzC2j)zUJaJL9{x{D>1+9bdf=AqkYhUep+pB|kKy1B5^B~7r~ok_We z$S>}@TpI81N^r_H)BC2Pi!7Hur1jWYb8Ls%X_uKn<<=ry5z!36%5ElQDPo~sYrZ&; zc*WyL^+CgM22(f{S^A3(^}z7Ez4LI!m4RcVHmaxO`641hJ9~0r9!*EEBCKy7f;7TB zzN5M(A;iB(oVyNtaRX(HyhXndAGY4f&=dl|Y zOpY$Gpz_IXW63K`sf~=}EqiZ$?}5)GAUorBO_n|OcX?mBjPI)sJzu@|c6+0FOYFvb zS-w+#0-w1s@kRE}t}sUIkP2qYhh0K+Zj&`()r7G($9AHrF=4H-6m@LAiNvOX*Z&P1 zYV@kNM;`8$$lRD;_v8h~;g^4a&F({Lp3#1-bx;SSW0{G7?!9p^r6tiB?6-udGQs)rU>}h=sd0onA zj{csT@004tsy@)Z-H{m_{+O+RH&}X^@IzMp5{@>50W?ZqdV+Qf=zpJcC`2`s6|OP3 zs}`_+Sb?m_jxX_yHEsfLkLT#a}Yf zGEt74Ptd}$!O|%AjHa>I?fe>NRcRw*hlSoQ%Prdy>=;~|Yh*FP+~>N$Q`_*w`kZ_W zm9{_Jc*&$A{rdaI8{RUEr}`;6NoIrh7v_ItQmJH>NnM}wzd~my(q3S@5EGulxEi@O z=|B4Ir!AW&LvOwoaSPAdI3$Jw#7p&w_)V$w7c!b1h90hi((^)PW%RuW?(wMXV1ZAb zH-+&)tY~s?n>*dOMy(^xbF~88yggJ>=_unMQ2i*Or}|u6`uwCpT7XDxhANu4c((gh zx#Fjl^iTN^s8)Q>hnw}qEvGH8dLvtRzteB4&km&knh=e^nedT%ljGvTDUp#>z8{W; zF8Td?6VuZl+{eYq(D|AGa7rBTI?(?hQ{SlqrZ|p(!7eZ>;qXj9#Afarf1q=#m={Bx z)XL5EDHkCoj>d3}%$0JxDc16#(re!pfH<|-4*!|aTsng2Odk_9)buz7Lc(s^T`JO@ z;H%tD5$=Y+2g*EmFRkv5&z=1MZ1sMxqs8?2ikJ43`JFO3rjOU*SKQ6-6KA7u7?bO( znwhmA>CEq+$_dbSecMP-0PF&!`z;5hM4a6h<15WFQQbOa=6F;GG?cyd^M=S%85inyJq1*<&`Otowa8*{O4${ zC3~w?f6q(06~U8Ypc_l@;DI>cm-^$_#;6e!Qp8Snb29awsFR>_)7|v^iGJ@)(je+f z>FRH0Xq@iLzF0|jKfITL)_Xr?d(~#^PUo(c+34f!@nvIEQ3i*odA73EjmZqZ+&K+W zy9U0v^ZrX*i8l8WWk?_e&0QNSvh!U?(*_cy3iJ1xFCn<@&ZaB32(p@Ia zo4w5{RP#rprY|)1feD7Q)oV11Xz=nYE22fS&w=|K4xu&gEC{hGWoi;%A8v_8L{+)G znNZ$uUeJV;4*Zg-YVaJZUYCvs3^P7aUhzcnB4c62}ck^FJxK!hB<2pz#< z&>1@^2{fT5YQ@bl_wrlXNs1o$R_gBUAH_FCsA7{wY?C|6E1+ z5A1N-{o{E@re?^&===)WbQuqP-9PtyiiY;lJ<4lgq5O+<1Wz-4f;wAM;a}GEs!}d} z##~yO#6A6t)%Ev8MSS%-D3SC%x7!yG#9aE)cyzsjLSoA~XZ&$X!O`cbV85|+@I}2w zlNp;i7|K&7lODG>+AuFmf6xfZkfj`Jwi6Db7))`IKOjWrsl)CVFxvS$IHLXa>g&_R zEkk)ec6}jS)H2U1qPbZ>AlpRvw$vERq}dEFbWqZ%j=Q%xkrSotr5vc?ieUlV0&bz+ z^(|6p625Y}T5Zwi=<&9fzU`v05LcpjPc2)hbgZR0hPCxsm80|VGiOpxGA@C#;ll+_ ztV1%!%GcGV*uDXpauaQma=#CiFrh9}&jReM{<3?7S@8Q3+(5F4+kn=sNbF?U#&XKed)^IRVud@LxXJx z{H(z=7g}MaD_CisoG7Ai=}h8mPx`>02awzpNJqDS@|q$a9=}P9sju9>tM3RjP=Jh-z{FLK#(`#`>v7Im1SZ$S+bPPI;#l@25Cr zw7`xPYsjEyS0wu$XPBVr2M}cLk8fC9enoZDg=8LClS#oWB_+NYpvtR@U1s*wv0BpWohmR3V9O_$LMBIN$KP$ zWZ4}GOulV=U`=!27?Z-~9^;Wo;mm|_+BO>&w%Z6ONlD&%=5MvwlP>M2zxOW3#39wR zHgNjILU+vb3A@uCcz_2vAWK(~#y_SobocE$wX$2+(%1cyHyuJS4gIXrLNXLzpCUrW zoFtB2%_5`gHLPdTr25<@z6V5ELcY5nLkdfkTBg_=(Hx~xI&z3z6tph6mRO*&AcKNTo*L_=Ekb)vvuUc2W)o>+Sks|*B*3>Fh>_Q+`q zk!+dfeVCCxg9mP>Ij*0ptk&g`urzyq&NV?2@e!j_)aRY`&G7RguJasCl6&|=G=>^3 z3GAFqZT-{vk3dBj=8gI&im&id2 zeph^6Ioohv;g(jo9uwfFZxAhfBM`2T{@FwONc9$G5n4u%ypyUM!OWcy9F_Y#y-5?2 zf#nuB3iPJ~fq4d2&wSs#7`1*}yND1H$qv!Y*aK6f{!9hJ;S=xOX8sT8m>~P7P ze6a0+@*E^X-SzNZbqTNPu*`V})e;AiyMuRO)$iBmr6KPNw}w_$DT8IZf(hOOl#`l} zWQ_6Lm@YcKY);zPc92Z|`M6HFwhGw+LAw_j8rsqadwQIuBtCW5r7>NgFMx2iOJu`% z@A6GW2M2x1(dMXaRxa`*O1R|CQ6CY;>ql?NOO?tfnqX8L&DgZd_h8|QFCLf@!j@pneglt{$qFEB5hIAy1GsN|r`c4aJu#_<8gtMR* z7C!9g8~@7Lyw3g0Qleq1bKBAoCxZ;vkkfo!l%tSh*yT;VsLomQXIj0^6fNPYObdkX z3SggurQWwe>DqU*jcii}+qMHDe0vH_(M_jAc6kwNQWUcm7{!W|>*XIdK65VQh*iNHIp$|eAuTfkIQDpXZ6EcI216do>z>{bD(7!lCGlQ7Y zn0aQjOOt6-#Ep(A*CI$dh7~4)VUjyNe1v2zIKC>hTFyR0zv`KGo))eSl@5Oci8(Wa zHXI9@dANmjgZ|0Ih-)EC`#Z1=50Pa$5Z5YbcLj#_tnvmsblbxs;-I?tA4OBd1G|@1 z+n$*VjGVjbTUY<&Zm7!qIgTdRE%S&=B@7lKj(5*`PI5A1SN(o*)gS==6l9*!T?ux__je9Z?r311YDtW8peXYI=LzSu*1uM!IIxhu6YSTg6-(#*yut9Kz)JZKj zcHx~>cmls*U>EO!!E2!i?KcFF0ujTNQEqmbPQZw+n1G2HQjkd{jxta7o=fwrH!4

#F^YDVyi2aRM@Ed zg3yv5_~3b_u9eX`KLB-^x6O+J%(liLSkrDUd08n;lxaATZhY}5oyRfY{uGk#>xFlG zU_by!E3fy5ft7O?beU7}u|Yq-0WEuFWrqlM@wg_`FGG9@(cPUTe@>iDm zagWiAt+m9jQuHO&=CXc#y6xn(%PW<-YtKjc4AmHS39I8PQd#*1RiSn8s7sNgi#`5> zF8a+B@6cw6h&0>5sHVF}k)Wzy3cGS=zIx?X@J$9%wfC}3e4pz$lxRB0*3)O+lYsV= zd%O%(>I@PNGq0|2a121^|W3EDQ=Ozs(c6qShKPs}LYS}{t8 zz8l6`&nsh9{q09bMHiUl;SAHY5TsPH`SVfzc;4ruZ!6NgnVMl~8nQ(C!&ewJhDx2; zRo~DEhb!rXywXeMBWw~3&|s(o{Tos@lY8x;hjq#%XCV2!`GulInjP=G9 zgZYcr0V=s(KLt91XR{k_P>@a*G}9$2uJPFI^%y1Bf%H#W@by}aHwU`k#q;eKQi=l?z5aar(^Rw4ZjuK1{|+an=9iCIm%kG}QP5{TOo8^)O!VFZy;%<`iJ#gR+uUW| zW;ScS2ho?wPX~~u{sJoZx4A0Ar;mD>c(+S$v`tYo)S!v-YBE)bJ}lh#%~IrNWv2r6 zX8Iv!cKIeoc8PXqMVVfiUIK99ibIb10uMPPTgG9rpuX$~bT@v~>DzoAJ!pgqT=y{JOXmu15{I+`jjGC`u`46qH2 zg#Hl6#2rrpfq=%RWr-z9Dd1H3JV;;AnhsmzPm(W$n3UIku;YN}GTkwlTXuT1kZpfbRpObpeL4y?kwt zp~N-MH0gl?j_p-Kpu_@8e+}k+?o0VS0oHE8FLyaTHS0cffeGdxC|`p<(Th@D8?TjM zZ3P5a5M&td*|jy5DnGBxm*_kN%BkNq>B!|`ALAI!cz AZ2$lO literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-12-Quarterly Growth Trend.jpg b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-12-Quarterly Growth Trend.jpg new file mode 100644 index 0000000000000000000000000000000000000000..bd6ccbb7f38e5ac80d03ac12caa700e7c4307385 GIT binary patch literal 27098 zcmeIb2UrwM);8QoQldx@kf4BqB$W&@45(xgi9<#OML<9#$3c)FQ9w{ca?YS2IcHIF z27v)2ry&k7%>3=X>m#twv%BwJ-+tfje?54$Rn=5=b$8V{bRXm5R=f7lG5^BV7S2dzy87202C*2 zhH;y5aae#86gapPIM@b&8N5$?oWJ}4fBnHZfs2PvKzQ;L5iwYy=rnKw2N(AQ9xgsU z9v)cR4g5WTM}dFl!Zj%ZN|pPBEY?(f&qEVVvfe1DrB>})W8;7D%<~ix%~@JH`iqy? zIWBVw2nq>{h>G33B`qT>C$FHUuAvFj($+CDe)!1bv8kDjt)0Dtqm#4O3vVA^znA`D z;jbei-@J`_mzb2ClA4yDk@?|cVbQ1JlG3ue`i91)=9bpa-95d1{R4wT!-&bL>F+bM zbMp(x^^MJ~?H$za-jC@W7?O!@qEi z;Ea?C;eBgL7QW{vscwWO6x5z#-#7Y7_XTnYdR?CfzQdJz42{CnTQ<=jR36?>wjQW2`mg#|iVzStKG{(#QU*M2!s zZs;>t#{E92X|;49(VycFk8JoERAh7Ao|>USf#`Lkm+qTEbI_(wTxT2Hsk_l?X0Oat zu3ZZ=iWC~2k1c^tDLK6y8)kYWaS7F-dPd(=he+@pWmsgl08l9E4Cfx8E2^s3K zS`Hh=0;b06twWc@Jjl=Q+vq1>9hw_{&`29R+&&CVOnlxu;OhvZ?`T~8WZD(+=`5e# z)2n!{8>ne_I~{?WW#($*T@L%ZB+7TlEB%049Tu{zW+pSHU~W6nQh zpfh|cw8og=bj|E}2loDAwACaE&6=D%Ndf6y{&4bT4ST6XUB$q}I67zt3v4Zo@6q?~ zG`Uy%B?kBdm8y$4-Q z@(O?p4Vefq{5ZZa>}<~ed^>0({qmew zEO4OYO!T45N0m;l<@_{-{4VvmEgFB03=WP&f8cMA4_o6Drm^Qf>}#eyV^OO+PiG?G z;Sf)%!1c9fO2(sS$(%>ps9kvJ=ALElszTqAT{&J=PcDrka}wD}i5GSR(MqQTwt2hs zLi{9HA6L=$>E5aGN2Y0LUgGG@CBLIc^cf3yw;}R&5$RZ9JOT?$rhmWxCFEveDHb4c zY*s>=jedInhKT30@5ZYudQ-IQ-5=05k-|6Ew&VNBiqAS+L$ZvdUm#Cl3wD34y>wQy zbXY~pogVq<&1AW3+?~*R8Obm@8Akpd)>ufRnW3Ss=_h^z{m%uubty3;`6>5D3#4tp z>Zx&)rH}N*`Zw8FXCt-x)F3_%OyD$>CJtAkouvplw0zO1aM$R%Sku<$86 z4C1Yq=plYZVJ*)O!}?7*!)#~O+%j0@w%;9EJN4UvI6Y`HpNxmuXdWbWX;Cjp{%7$< zPNM8gGNKRD16BIpMohQEzUkk7ReNT|!N>#OQKrbn0SRsD95F@m4a*40Xt}j)Y+QFX z>6zoYHL}U~T6GJOxW|l=LvwN8m7hK+8<>8^|MA)E-oP7uSC-O(qOgU`>zbeychT~>hX!& zA^eTj&de{jI(&Az=7!s&&kPle)~eimr94JfAv_g#hpVf?3Q;=+B~H{Xa5LtXZW@{}%0-S_wdm1srh2nN@8J@)8%3qbQ*s z-RjT_H!azsIsl4V=-WKO%5Kbg^F2tKuhr%p&y!{U-8Z*C^JQKoW^5K=@=RZVIWTdS z6|7nLPD6);IA|2T>Z-;oS>Yc=SMaX66nDx z_c%856xR&Ba$H$cogUvR{@`H}$EX~?d7@&b0|5=R-G20;|4B|%6K%a6ALUk^=BJ1W zWkwydZHKc4Z1NV9Q3hI=Z`UpFK9eDuKrU)7n_s$7xzc4gxnb@w9F!VuYHBCwDRa^4 z63xAKd4^ZlarXL%$r&7r6}l69mc=>03l!|KY6(;G9@S1!h+3uyp{uL^}cOpJ>Qd98a{8eooW*=TxK?JEX zUpNoh>LhfTVRpD^li3$G+sL;#cDejhWt=@K$vf(e!vhD7347|fv)gwUble&v>SpMs zu9a-FQW*@Nfb}9fPNG;923!f@S%#Apo`MohA2o1%Ff;66 zqN}~#Am2_EIJMC+t9@$X%DH)(VrPla%;6Z$@4=JJrnM8Z-`N=KYIApLtIh`0IiDHK zj(yVJB%h3n^LEdLM1?&*u}BQG+TuI^R{#40Tpmc}`V3d4$t(*QH~ApHx;D#td@B1J z5M{l+om@;(Nr_MYnEVQxRzi^C_0pQV*Ta>rKe-e>I_;`B0c!Dj#5u&l8|_be4%nXfB8wDG z48fG3I`{Y-r_>ye@rR4tC<)!Ny5SnD3hql^%u295&D_eD+z@n(O+7*Eo zGAgp^ZR*K)OMA-x;y1}c5ap!Tl?%zeY!OC$F4LP#9?#6j@|GrhpYU>7iz%YDuBq-mF#RA9MZ zW^0?tvtk2r_hrTXyy%+L{r6V;?x8w#i zKuK&ZTvXj86|GHdE`0GhAym3`KDb&z$l~g^OTlV27F32*TT{lL%bc?tB19w;nji)J z@yJUBZbWn!mDDl_PGj@cg8!oxDHC)$@3O?c0%Q2e*(+%xz&Yg3<80WkB)^?FGQn)>-w!W;zb49}$Gy)ZQBu}H2CY+wsg zr*45ZNKNC62M+KfIs$YA5B!53soZz25N>G_-cjWbNyuGqY8p|>_i5_B!aGbgL~s0w zFje_dX{90}&%5ZlPps1Q)`(!Oky1%GPV>CHZiaGn!gnli##QOz`U@r#Iq?$O(0`t?oDaVxm!4#c89Q@}(E zT5zI#@%mG?-dvuzuW#*4IG-jB+UHoX)(#VRJMX-+k7X%+s@~FF;Y*k|#!g*+)9p0J zqP-JgU5u4ncLgL%DqUa7g733_ZqM(iK%=%KW}?H+1n_kI z?t%vkQ$&`?ay4UtliT(Mbo*9%@DVXn;r#buqg!QMEpUWm$lh7_x&sy{>cRrV4e-u# za6qdMa^qOfiYpr=R`_Dh`eA_tumU@LU!QrcLiT{TdiOdO7@nKH-$6N$o_7H>SSO)# z`dFZK95Wz`zCwWoZmeSNV1ey2QCgdZt%bkWPp}LF65{N?&Wti(#dwdSW3a%oExZoN zyvd6NrdP3mc~YqFzjP-$V&skE%a!&Ww=wnW?mg#2zU&(;SD^;^ce_5<$Fiq_FYMyq zSO-pL8`UP=6xH;3>y@jc73KO`KNZ_+GhhM7=j{-#fyb#aZ{L*_Oi0jIAm4c33xiy* zQ+@Lay`h+PyS=_gl);5)lNsOfP2+G=!NscbB*B=>I&EX2FzLvanl=X8OXb6tOZ60l z7M!4gg;sQ?2AmQsUAeqh9|p2Z9&hBZzNjBTm6tQ5+1-MuVnkhV6*k)5|4y1Y%BvUangXqZ94I|> zmFu)qAJ`nHG8M|R-HyIhaGtx~+?at(ZraO#bK&yF^-;ade(?vJ=7wj`%BeJ6COu0$ zCcgT_3ydb|GwWLL;R}vkz@f=r?V7Rq+3Yw)m`^4jX2nz86^wK`Ys92OrAGb26W z&k_IIQ6*P~1wu-z7n9&(dt--7-BA&DUZ|@`KQ{h7?~uC7ksKqNa}HA_ZBz6s&z|2u z+XFVbXM4m{r*3!LP$}Ey>PS=^U&T-_s8!~*U~pcDT1Z%$0# z|Bsp?m4paof5E)dh6~pg-gG<*y%qR8ROS>r1W2CfF6r&Dhsc142KVn-v;WR>T(fxr z&~>OD78szd-lQ1q?#rb0-S9-Y*=nxeVF#;mxDtTV2&LZdQj1rnf(*-QeU%EadxL zFm|qj5Ojk2L0iX=rF8gkDSXh}WVa|K%0nd zRdY`o3nZylJN$oOv23RxEHqNf%QBuG9(i zQBTw*Scm2k{Tt?fxA{+8^WTx~E>3o!Lk6H*O;--^V4X!91C4RJ%(?KzK*PP0Sw7S1 zgOlKl>8<&rX|wlH3JZif+{OZVa@CtSI@EFTSfHgm6+Vz`xCi7A&5FHJk){HE9YS&RkdWKwmh$^EvSIT8!Re!Ih;ZU5bAl{4xiICZx=4y*$} z^+&zQ-RtxyRRGon%*XJ&~@JV zxGivMdvM(-ZBIU|(5cs@2heU`v4H-$-GTnCL#6)LUovf3KvBMV{?MEMO{{YRYT)aT z$KzY;Trt1$mmxmI67Rr9xXFj-({$C*(U}9+tm^eUTUOu{66BgC>OV)@I|?@V_`3$i zacNr5JR)uxk&QPeA+g{bx#RJBclxiK%f!{|ChZt%%6MSH^YQc9Q{@h~jYK<&o>qfq zQjU%?!SkfyJ7HtF_;qB4;ug0Zwm@feLy?hbRtn{nGQZ=kRHs+dQ8;~~fx)C&r|4WF z!_t!5WsNIailn?3E*>mbW-|Gho-mn@E&OJr8X()XvPD;_=ok>X80>`Kn3*5Zb0l;H7!>W>=ee&CGBZbTM8Mfv3F7p~Q*HV4{x@#hs)LoIJ zh7vj-I|e6T2*1hFB7Dy0JK;{mYckn1(8)bxbwvQ~o61%eCZGIxdHVg@my^38jQk<} z0ld#0Ebc^!E62AwZQtsDLO*rgwcu$fdrzLBSzjJy<(g0kVV&Prp$xsBH#@(RZ%l)0 zN&7Xg`B<*N)GNX5SlGrMMd16$Iz)OC3%FHF8FZMN;|C>c ze(K`BdnK<@e@ZO#2UKME;&*Refr?ecO-v9Td1t*7|KkLI~Zei2-5Zf`>Ux!Wd)4=TUhI}r9?#{&3|C0LP6;gH~H zXKsPUH@553UaS+Y70;etx1FghZg9S_SjAM?bBf3$(A#LJ%xm|#f#5_3P21YON@cfQ zmS_X_uAQWC(^Q~`^($$L5h;0>;ZSoO=*uelF;Fk z!v3q=73UBG_mc)g#_l-FN$sXsJqdrls!!O!Orviu=8uxCA4!RXt*@_I!`c5d(pUTe0*|_xf zVG3k4=Cdg4^VZz9nkpd)F3(Bga=TBx7I!WN*%Ssx(M{XbQrDs(154s`8RpLJm8?37 z756?|Yt@^63R&lpMch^$N2TfacMB=XD{`I@i*y@A6K*6VNG~~mqf*g+cf^$cfxaJE zkGSS^j#wBP_0Kq;Sc6N9*}cQ-%;CJ7_E8I>EN0U$V2T%W6Fd#@^Th` z6>=^|sn1DEQ)&7fF&BSAb^2#LK)AuVt9o%7cKuL6)rHj3O%QBGDEHH10aKh-S0!+k zI%K(lMrl`up$H80uj z-4Oh)u~3Z%pT3+0AE3noo2+U){=N+^5Goa|nEzu~qWZ-yfv8pQr8523Z|?a}R+iD> zFp@2q_*Y8<)NvV)oi+2R%*5P^`p3&22A+}dP0u0Xz~xt3TM0@kb%Wk6Dt?8mWW}kL zC`y%N1p~Jn@kBK5n8NQrwdCF`=Q8y>-%47>dfIuN!0dqPE`dP`;_F%1Yo&D4$l^24 zJ=SH2jd~zvTRAC3L1ntbQdFnYh#3^9WhXezem@YO7hA;*+E)iQSU~RqX!T3G5gS5W zwL!CTZx&P12)gb5C6>}k!7QM8c>Bb075~>Vd6oe4?pfQ8y-m<8lB7=1-Xli`l$jhj zgBE%`QD?4A#u4zxHPj9ZFx$WfWH5GHm4{0y{G2f-(-ZwWTMo~&pLwptHD+{(j_`r2 z|E&41yz>2=j#RKw&HQ1jOZ3VSGiDEZ#EB^KX%Z9$t7m`tdGW(3*Kwn>BJ+q@YWA?; zE-J(Xw52HH54ZbQ@?2jrR~8)h9)+h?dT)JAs#v0nk4WMc+7uBVNQOnCLGSpL5Ez;^ zBn4k(pX@V}g74L9k&Yo4A5s_=z-7Y%^iTvhlKFsyJ`Y16j}Z)!wlE_8XoNY_K>-eD zZ6+3&Fpis&MPG~s-JS$5FwkGq3T-NwORTkj0={;CUE4i1}FG&4~f{Lj_pil!+f`oZst%ukk4(la6(+ zwW)HiTnI{+jb0Pk7w*L;z;{WPwZ*vEcw3c%U zpb_E&&QJ5+tYc@)l1lke%G9828NL4_dD_YIGF-n1E(?Yq;B%J^ybWB|%rUnzsu#DF z5AeAkS?48~AH_;wq%9c*4d4=z;c;l?`OR8h&^R71v(B ziF-uAw<+pxA&7aS?}m6 zZZ{9SZn`vS&7Lh{3)Rop+eAqs*8`JYbCozxc$j_tUD_1&_v&--Y0?z4#@RCzb@ASgi)A5|aU+GodjM$BovC>CSQ z2bG2YZ+6HL^ZQSL@5B##@rWh$F9Lj}aenmKd%~WpD>+)p(_%k&eftR`?hzMzh7BEaw=ad?2UooYk_tP!3b)mfiAfEAOkD3CYgb-7Rl=`izi{u&2NLvrqQpm`+eIe2$CM zpCbAH8_Ydx9MWq~S@~LCush+J%yN--`3bW|Dx4`zIE`TC8K)(*Y;u$dnH&~~mA70n|2Ff#Hw#Vs7 ziRm@eu5F2|5tT!vG09?XFKAu`<=4_mGkK7}LQ(Y%XRE{prq+smTg?UG;|k<$7Z|nT zR>*rX)Jw0`(;mF5@)3HDD8EAfSTSXoj3`b>Rl<}uSZm<5qpqQ$Pgzr5>Y}}aNT`=g z7svHavv|hACn4seoT51sC4dsiGn|&O_^FN5IBSt@@p8nSI46hf1q@p(fe-PAy7Tp` zmKM!S@lW{;vLwNc6*Y|BRMF-1no`nKsSc~K*sSiKYYBkSS}C-X%;!&`mbf2 zRYQX6X@zJ^wJ~U;sa6vGt=}^jTnl#p8>{-y*x>*9*kHC5&9{#G2p-jyel;qj?nl-< z4T(aWkJ7F^Ak%=jF>do*3HlA10}N(+Pj>8)j=C#is9gg)K+=I3*GwEr%K*e(;=eUd zS0g~G!xb8Mr{&bT;Tp)m0fXngu&>ay{0b0;ktz>qh1Z1~ss4isk&>SG4vx6*_z~mf zz#Im4!hE>?Uswx@Dux0eYoQKlazL;Gy@&;fpobg$CH@|nSY>+n4yoXW>;HvI!SnZ? zi0Ut_2Oaxj30*kMvM9Za5+`DQ^H{ncaP_Eh+aAyH@LVn4^@mc?; z930)Wb2uvgpB*EgMT5{Giv$=KHONC}#plACq3f)R(5(e5@cG@SveAqfM`E$4=N}=5 zvM3%fqUt?+;0VP823Pi1a7`4WgU`loO<;jNKCvC0CYR`*K@bQCtNVG0isC6N^z*{J zJ(Se}L8ZcxV&?U5=nEvc8R?VgZ}oq+`)f+>e9Ua0tv;$v^CM?(*@GthFM{cDM0Ahl zxNG@z)IXX*8g%R5{*jx^toitGtoP&Q5yVt)R`oZh;Rn-Tmt_$nF$2cDgV2kM&WB41 zk2J(6K&T?&NNE4(N&%CefbKFCTJ7ypw-5Z5r;4rDjZ*>}J?}njfju)kf}?N38^(9p zor}THG=0Q|w>fE`nI%|V`|Zz*A5Q-UDR9y_f+h5}JxD+@b%4tlyX4_^Kk_i>RNOAb zR@~ke{JTEB4Mr4%JeWaUmj}bBgSI;JHP8(!{c`wD59ghswN|?fSD=fvAbxLbw^D>K z#IK%yS^>5Pc_m$;17-dEqhqn)`llZGv)!L|=6}7NS&9WfYG_MHHpp)R!D!Aot~g`| z+HbGpz`1%owU}~nHy8|2w1cqU?*=+>yD-G?7K9V1rpQ6s5Nn=Y9?>z?QxH3|xeZET za{kY2QNLS`(kMq zqbEq?f6YT}N{6)Sw z%r5-5aSa*X9MLas%txV1oq8ksXb}9Iky=C(sG}6pde8r2k}I(h&=$jOd7clabS~ zEnx)4CT!mRQEXxp?m50LVgMfi;{|gqi{H%as*Wk1Y?IPT2q=EPD56})VjzD;MUL(z zJRAr4PmhWmF=sS758FCQmh1%^j8rPR??L0_cegGPCZTCSW2kNeQvq_|Wn>K&4XBT* zA0x=u`5#dZ`TIXFVW5jF#z+ps4qSyXUIr;giW-cqadnP4V3!PtpY@NWgq5!JV3(X$ zdk5j*8mF+yxiNk;x~8RQllW&C6=WX&`$z+aZJAupXa!(^qa|C>tLq`VY?jR{v0U zKh8lGkHp$xvFKyds7^PA;)mlV{oBg=Cpqbu?BN_UV*V*|5_L%Wg3i~X5!r|F=UOE3 zl(q{)*h^|>JcKxMZT3wX;x;cAV@Oz%er4myf{)KAlkJX|&^a|LG_ADu@FwO`zwNnY zLD#Df%-;sy|BAlF>oWPnb)R+SDg?nHHE?qVA;<)J4u*z70!XGSxb;I4RlUXnMdp-- z`f`bs{iAA0QH1~HwFLggdth7>eu0J|NX^W51*A*Q=)Se*88r|5@P3wkzi1o_JVAwH zyunaU?F)_9F+N`Y)={A#)t_VLDZSlzmYs+>^XE zSNbR>WvsX>eVU_5bTc>nm*iwWEUmB?6XSZ18A7rj5LsXWuaoJA@);y^bAK#!{uo?r-`OuAj<-{A!x@`lNSAW;fG0W->F@*Ab$hu$~^dCg&!3sQEp zJj?I9pJ~zul&%|F(~fX4CtOYyGi;>wd(NXJa&5qTGKPdsNeyjA-Y zQRYrPV|a8mJl`#&rc_DxcNM4EZYV?c5v)V=2F~Akevw< zDVu_g{{!nR9%sdtuA(#~e(daZHAGP`qXzURlu1!pwF3^}iU~ZOCDM{K>w+Vm${882 z!=+_-`1{lUKv!l51qUBhmMYn(z|c6Nt4!6y6n6Wq$79 zyQ^m~JZ!PA^yTWxewFcao6WDv?*o7Dt66_^F0rJnxG+%OM*MDq2q~*K9s2PJoy+O- z+~#bSxq*DyY3Ds?wrcR^Y^^-UUyn#MbTG2!&0(Z5sz)q_WVeMy z7qrgO}L&sK)|{>A0E5b*nX*R0u2K`u*D| z_bS8Zg;QRiKU!muIms;5NUI$olvub^g8P2HNdW z7$gID6J*W<7mqrunCNU33Znx?jK!44|kGa8}(SPM0e%&>Hi%McL zEa1md3iW%mu-dpM^v1$PFhAx*WRPFeOw-eo9SU<*Opia#RnY|TsweM|Ju-%U=oZK zu})hVs|}$`r&vy(eLov1=zZm-`}x4=O$ScGeVM4|0gi|GnI02BK3Yf;6VO?_HKD`JT=4`O0{A@PWZMnbhh1Z{EqS_ z6;6i~^rZyr_?`Y8tJFE?!__ZaQ)r;AOraGlL^Q}KPdhpekj*>DO; zriZC0SsmnST#bPub3fMT86P~xvjxr4%-$NkIa?tAzJ2(BcBHNLR^TDB@y&wyR)Dh?v&*OE7~Q#kWlcjo$&A=5G|go-w6-{q9?oySKSN z>Q=NX#Cfdvmb(I1*{32!HwmuW-&y^nw}1zkb7hRd&mV7>xjqTK9(D2&_cQlu1z-{_ zoxX@+L0(5$$(cT}K+wfbczXNhIxjC>>rW~{=|ghL^-PDC2Sz$ARmVAu=3hO7>TJFz znZ@vx4Ar;8p_LtSwe^M;c&gUFco;+U?>O&!auzq5 zmo69TQWwX{#VQ8hexm-0YERveSRneXg_T`E#O3XNCM`(s8Tmb9Oayad(d%JT%L}21 zU`;;^#!fUE-x3yqmcO*$M1Se+^e#^^>%na80X{(Q z;3F*okL-bsLX?Ie~iq zaR5E%c^vh7ID=~GxFKu80pkr<6-`_dAqpc0_i)cO-<~SyP=UFsAp$}u4x(H7QuEgx zOxjc(W*MK>cbD|5q_~=Eczst4-*}E0HpWy8WOfq1F*Cpmm&=db-f4naOVPGo5^CLM zu9)tF`j*`(NMAgaS-Cv0ZBZVRYGO6cBy%mUJ!-PWRV8QfD^F8^gG7}1WeVN+3;I%v z&Fm7@5%x?J$kdpu{O+nI+r<)?*$8v_^C3rukQ*E)iW_aeA@HqMMiUN5ZFO#YHrHvY zzILF{PI?mb?CT3cz(LSRt1EkJ1oxqP;2I67_k%RnG?tuNBHfW@Jfl%I* zj+w_%28R-)K$0I(?(oQPLWSUJ^EZ{xK)s9-28~0&An*Wxhlyn5v^WY zr7tu&BdZjNFvjadM5i_U)ft6W;z*76G;Vpxjt#IJR{|t2eO!LH$ zF#j#ZWm`w{dfBt0BLa)POcP8s4uW%P&@s<5jSR(&!9yreB`Z|RlBAwl10T`Y+}-S2ni}5 zF@?*rGb8GHiycvlJU8yF#()#<+-#Q6x+&W`TZri`>`HUVfp`OSE8 zxsPRvi?qCn`{6Plu5<d z&x>f`#J2c(L!);X#eogt-4B-uV@Jc$iveXwc(8!ozCi|J-}(`GM?Ujwx0?*^Eag_@ zi`%iR-Q9aV=S|F;#+}Y}b`pS4{?X8gb(JXcMH$h}`)M5EYNh+{P1^7#7#!F(sfS<@ zIvjoO0_`Tdk%I-2E>HBx2i^e%7Hd-)eUoQ(Fy+?UF9~idI4o_qilBi;Sg)(-BoyzK z9fVmka;2O1p?B?sa|B2}4x}*1QVJ7z#~v(A5}cyNG|@=_^+w+$d};f5^?`mE?uW-4 zF~QI@=NVUx$+q%?n;42}k{QK;i~Wci3wRGRYl0$F*tVk^h8?T9i?g3dd0Xl@?0jvuf}p zmw-iRu6Dl11ldfZP^d^73Yx`|Dgy5jF9td1ouR&44+jLel%!#%QGB1TU!4f~ z0%JuA>%X+^ir01D+gpBPYuc*OE<1TgBsZ?EO#L0Z&y(y&SU{Klk@64KREX1xDsK== zO_`)c^B{If)R=hL)jUO7>6U;6=k4^4x(rQbp)05L0w(xpEY6#3bWFXQsSusK*yAdn zZFZ?&@#_u57yHih-=p4!qn>i5mVoA>qSn{xGScE9k5MC)lINRbx|{K20ks*Ea@=Xf zNP>a&xF>vX+DaJZx0qgR=nL9Zok3dU`9F#2=Ev_9p}cX&dWGCtpZa$06$9{bmXEIbsiGN9 zZ0HOuf$Sx0%J`OnP#?@WDIh;eV=m5ltzn4Yq#9lq;0oarvmhCf|!TyWSDt@0C#z}o4GiIR*AcHCRlZt^blhW4C)|zgD@BkK zjyOK*{74~`z;UoS3kEd}t!DKs->mI7iSf7W*=MEcR(O$o{vyw;#t4#M>mbN;7@zCV z{Uz|LUS+_h4rgox=`_7Yq0Wmo!ATjs4_9O?r=LP2^x{e4PCNQE*T1vTAA_>{O%l$8 zRuZBHP9@Iigaj8!4^&7JnJ&P)mYr~Xi}xkcDwQVla2&l1kt74iacx>0rW;DE->Jyn z4$A8FW+6KK(i30}no^QAIKcq;NikcTIX}>fWbu}!)xZUavYavowUzOUzV)iKJsV*O z>fkjetfx2hU$Q%8JW!-{g(py2h4q1)Yq0yoa$I{V($Fg#t#X22)N}p9enzheQ-cE) zPB-i|`sT?h@hFYwGul41M!V4E#vxf$x;R zCj1f&v1UaXX&Zhl?FheCiW!3COPq>Q0AspC@@$Q*#C)r$L81+KC^8 z4@xEnD$6?S#;L-hR@jo{IoQUcL6k(Ak+~E8`yCZo8|yeuA*-l@dA*(b#%?%*#(T4n zQws|uTZzK?tCZKbGsJZ#%B{}4D3MpnO*q|PlfH1B01Ghn;`G8^JGRJ@Hb+_WhrN@IZ&GAp3!o`5;iicAgUDm1dS=&JdRuL*Pt5+bW7(3$A z(dQl80yy7buTf{s&@6Pxa`Yq@071|Tg%9Qvq#zMu& zK(+<`(7~&i%5>a}T(wD>k7N-D-M3NA!@RxqXH3Fl)8_E+H7BDL3Z$$Eax4TC>&mH( zLb72=c5QY`q`} z;6_rZUAQ16Q9_X6D15VYZO8XP&&t@@DE-00TM@OQEBJs_m$b=ZWN~rX#c!YD#A+Ql zErMkpoQO-u3+9((2nnb#5iJJ?9A59yk)dEO9=h>;@A>M23!F!p&{2q-bR_!apmN3+ zsLQokOSm^YV^z;5YHg!&PrO4YF4MGAVTuGX7$%V_*86*trdch&H`b~^ukpGhMlYJo zK2G6Lh?N^bkQG6en4SS=s6fEyg4S)ug5i=%Sp+5_Z6PGXQCWpX70QDfc6N^JBi$$%P*3wzV%p&es#*LW5L+xQ$kHfquuy7;0lqm_2Z`nTY_R!GDmJc9+l63 zUgAdm%>Bh=xiXofb5FgF>X1&@bWFEG{TnMoOc7w-^%M2{YYKa9)(x-YLOy^fJk@O9 z2s`k&UsTQd)w$BxF`zzD;$lxw%H%8c8By4(jGHX#F2RQnWrUg}*?cEO@iZX1YQi#qVQKCOauUZ4DZ`JZ^bM3c{YW74#(4;nF+&vYFmfJ9Q&QkNN zOK(U%tpLgf0MRMB+D5<@O7tZ4oxYlGl-ybJ0`iRZnFDKsD0SP1`ASTkruXsRM3U$_ z8==j%0H+X7=yZys?;<-Ukb1ta+0mxXVNW)oNMC1C_fc{A`1kB5jXt{Kp1^^5LHvo5 zaXQC7TRqbGhLf)>=*Mg z99ekh{H=iPi?RKa*tARhUjf@H(ly=r?idqoBkQcM8?Lhs5pgqP2jwse13U1Wv^xz} zd$iyhw+d4An-+8bNHG8#<0ozsyg^c`7nBD{ceLm8+wBicEfg&Zwm25o;_8=$sP?S5 zH)l!*Dz=aZN3p(=&3RB2g1odq`A{h5$*a(!6q5MRE#fi{PDhdRsI=^Z0Op4$^p@v$ ze7Ew{a~*!F3ptarXaZBeCO3~IuEf6!I-kt7NT?`^9G<+!#+Mk1rmua3K0hs#28 z-CDz3E4xJk_laUv+OHPub z8+Dxf|L5H2xo2kXv`<-G#a_E>@4e!?)-LQQb_TehB&R3`;NSoN z4)_nijsv%WbNKiK_-D@%5D=U{e~yrtiiDVmh?tIol8lP+5`>BI5(5J>+ckD(79Lgx z29BGYJlAgs3JOBlMI}Y}C9Vkw@_+9H=luEe#6-ljBqX%_ml-be|8IY=UjXuRIL~pL zaB*0GGvqk980FV4E#pN5f@F~@d2w3c?_@754pJSCNtfyA*+h!AZ;_%`;AeMeSKUO`dmfyP5kE$v4-#!pR5&CD$<9i5zAT;1F~{9gJ81ilIij(qzr zDmvzUY+OoeT6#uiR`$owMa3nhW#tu>4UJ9BEv;?s9sL7?L&GDZ-^OO=<`)*1mRAs~ z$erE2{ewf)(ed|j;Q+Y*H7xM|{~FjY<01#gbp{U)7mwikxNy$6f(I@+-r37H@F{Mo z5g6H1vhY7YM=oqqq5WlK|FeO;_&*xip9A~bxW)it zTpV!maLEBEaCpp-;!F6S<6n0U+K?At?O_38M5rh1s<=GGDT2P2a3s&MS&C>t;?aR- z?Jk9~OPgCa5udYE>t06g;Jt(19dURthO@km|CT*tNp<+n9cE3!3PdEo)!2T8Gxhf3 zV5K&1mb>W7vjlv55?Y~X&7#r)v`U7Kj?9~S>J%;(V&Gd68w>b9U9A;C`6qWVh`pN|Q@rTFJ>DEC=|;WT-^p;Hy1zi7Pg#J; zmo%eVTwYP?GWxzJ;$1ww%dT+Z3n3^AhW&Wp-sxGk;jf!sYV-74ByQ~YDxEZk@>yk% zM*OWSOqo4AJV*H4^U%`Ux#?kLAE$1)nm9Ci*A8_oJjL~=zbAP<7Yn?uyfT%ZG&c39 zsXv~RNZ}Q~3we8M=~)LUcH~F5fyQGSzYJ}?l3g)6jVxA&5FASgekDR^C>p8!24p$`qf8 zBD-xYG}aV-sD8(zR8$%kiv`w(TLVgL3Ea*MB8pZc+OtI=R&gQ;Uu-uQ&Rplb^4R8> zDe8fOh6S~>#W|pG|I3U<)Y+(w(o_q^$qxt#3gWa^4kW7g&8gI6fq&=2&9#N^ zh~XJ@?Hg=QINv7l^bTw@d{Iv;#D@_JH0XY+&w^Q=#&o=rUp$5T*eQ0qlId(jxUl3o z`ksB@!J=+a)FZ`<3O^B$*5@Ku3_gn1y%ep?S9t5BU7mmo%D(4z7BI$-vH` zE6|oCuCu9bubUxbhhy#LuGe^F%M@C-H>al)?Wd|1AKoEMI;-YOV07DZSVnPbU1xXt zmCyCS#scB^v^3cGMdP!nYL!ZR&}((ECgzMr5ytPb&#hOzNNfrT!8kXWLrGEw zyyUM|e6+OwI2nD-e|D6*2kAWzw^dS_l;>U#fsvu43<*&J8pC-S2{yWlbczdK2oniT z^xHo|NzkGxtF?rKo3~ZSQx>daM_+ZC9&jnNKQxPJGFfQr!Z3 zc{n_v*PQAWW+f@Lg4vk}K&;qF+qP0GX-(V{9#wr2Ln$2dQJ3cGxeU#_dw72E1@6Hd z%7!VrqV2plr8FV^6yu9g0mPwe8+*BxvErXoK79fi?Xo|Wn)KO!w}blZ&_;U56XmKb z>%Oqw0fh^ch_?@+vzPmzD>pK+z>E?s6tGZxRJZ+R1-?f+fCaX%KwH953QnrW#1>cp zMFF3e^>-2&0aJ&Y_0iH`E~Sf+XcHs2T0FA?gkS$c>JsQ0X@SdIgGd}kyzmV zdwD`9#)$*KRVZ0oigP;=KG#XRv89}P=Oi^6T3wrpq#Ah-k65^r*_=-?VW=qdGRVwS zmt1}!XNF9Z^{v5Ty6TnTa1(pZNwxykiWYvQ7gjB`ya*Q$esdS$3p~QeW@bY19Z~j* z!~Dm+bqj_Au(e^nQ@n>t!o&S7%>Ld}fNp5MwgOk$IkT>;W6qe6uQq4a9

5bW|!k84dejgNYNRsg!VaG8dPV~$S@Oa z)r*$y9@ejRQqLHkCZz3n;OH;=iEe9#K@mgHD0i2j99TmU(k0*ala8KHP0tNQ3cqsv zD+=wv3$lFtmu2*xjIO>CM%k-zQ3rl`FU3MLRNsHZpCLWRY#-6eQd(^_w`O}Dfp1)@ z{Oy*mo8w`4_5t;>6xCpJMp2&ImRitE;aiU{fuFj&-sG%(Ss2n6l%KOUP`!w1SmfP% zzc5Z^8_C9+W=X(Cids7wSR7QkH!Mjsd)gDUuZ;WAYUFYKbu1uXb|HlVTB|^JOt`_F z*o(HBo%EQ|u_#*^z}abP>x(2vVm8~(3rOD78RbX`zWuAeDlsyzg$^!-?jJ+<2=Ynt z9Z*A9fF(El)F=EDXRXo7{FxE-6-bzJCYU4X?~ak*jVR;ATXygtpD<2ts}3wH`PTW&ta`QSsX zvJ#{?S=8W^PwI86&zaMio|pfYJICdWw;d4`qENTQ66KZwREIe)L9j6=t?3SbM7{{! z2|Q5e_*#W3J&F%Q$-y&+={|7ID`SP9svjA*O8O56q*X0ydMQ zTOl7bFNUC5-RD+}O7zXvHO*ZGxoxZ4rqZ?;C2uIMb||K~&+%e`W*7}A)3i*H2w-0V zLzSMM)kbzj6{4QcPM%LCN2sMKM37!Q+XhVp@-=#$#G#~!>xf~+SMv}7ro73~lektS z`~`2dE)@k)Kyhsu7I-L@=eG8OU;G5A8n5oIA4W7M7jHbfMY(cQN3vMcvFK?U;RT9M z*X^XzJV%O1rP*`cm%_KZXNB5?3@@yyuSSl~G=-d}W8_zqAT7_olYUi0a~)*S$QQrx z6PeI2%sf#m1Z&GN#sY?^SU}oidu)uJR&@>(i7elgU=HBd0~CfIxtY7sYND193$*3J zz#c%0C#@8_t}of6loHHMY4M;l-l z)MTme{MN_wjHx#~qRs~c7%;Y6Isct}TMHv=g$0gJp~q)5dY&LQJswi2X~+MwyNR?I z9xM8KyPNOl`_kn0W{c!^S_)a~IW&*pI zzyc#_@GdLgJkU=(M29V7YS=JVO1jjHJ>H=7;3k3M3;c2H8ipF~aE*c_D_EeHsam+E z^CugXn_T!3!<-V#_3Bq+_Lu8Jz87^n+=d`I0wkySPX42j%1&$^nArQsky3c`x8(ny z_;Fd+wyd8q3ry)X+Fn@QDVK!=C7oaS6b}V8;O;A4q1SX7>^Kz8JU)IB!zBQfrcZLd z#U+i`Ov%|gy5bHLj~*&Al?&NSEfKZRcv_^!U8H-&`nbdA<$b(jbpb9;)5{vfk@o;< z^#RR+`~~y%eInf$wa9|6dD8n|1aTaGQIco=dQA+Yfwd*%!A7NEr!+%5lirGRJx7!! zJtqV`SRiGrRyWtEFzUM*&*A^=ylr8~U~QYQwSLv5k}q1YZBLB6BNi}-gbAhd28S)z z_Q_H|@BM=^79%x`^_Y{V4{OAl}HNw8+sQ`bzn%3tG;6>Vkq3v11F zEHgC>ER7MqKAAfmv@ea;*{#shb-h*EC0umVZcTqahVR7<{%L6}CNs{%vzp%VnX zuFzAD4KMmxZrk%KYc}x)n|FnpMo+p{Bm*sX>1pC;m%kakXy)^Jn9U*nw3Eom*teYE zO3+%D)3IuJ$vA{!d?k^0;BtxQhjZzHg%ZT*Ch|lP33V&w%bB^~5E^Q-x02LkseoS& zobX@xJ3m|E5$L5@Zpg6ryZQ^i@Bc(v>5Bd6j{ugyfa^15=5wj{Ai!}v4R9##dTNx_Aqm<8BV3w`7 zZbc{)Hlq#;RV1xRow)m^K1aKg`vcdN27w!U0F!xvUT@0B=~b6YdMs8aZM??pI2ms^ z*vBO;c!ZQmG%2saJ#C$7MR7QpW;!3x=3?Y}aQh z!#{V?K9-6_sOVMcgW$AC;>2G%HwldxD zYw%Ds;l6_Y^4cvt|1iAZ)G8X~ID=TOkm4TkQ+5~S*G;RDgN2+NduSz3u6W!_w!6mX zRK*?Z3n&UziUVQ}a+WACZ}NpP+n}S3*VV-h3xvI2fl}=yxNOjlzIj^Rd@H$1kLQYON0#)-c(~dJfP2)o_(byF==iiwqvQF7v@BY0_m9V>&(l_# z!xA!ebp^J&7#>q@Bo-gNoHGa2^OfJZBmZ#${%7fprE0^OPD5H*E5@$DULqIM2NwuL zTlG%4(^Foqlj!YylOkH#ZrV}&WWdGD^Y!6%V5`4;D=?006?(xqvN=syn4hk?l_83N z2<~8S;EyzGf_*b!y!4Fhbv!z|e$g1>hLMDd6>XO_PLI}7E62XQsMU1$h`}F2e&mQ6 zj&Z3HSkxPqXNOrpUf1$>W?1@9b&OJL-@b6}ApwI^yQoN7Sy6cXG##(gDMz)7>#c{$ z6WYKRWRD(*&Zi=^o|imAD?@@Lgxsl1lYm_c0~}^!))%4-7p&r{KG$q}qZ4_MD(e^H z27PlAip`j7Z&Z{6Twx6_n})*5=J>)00-x0ttkR4}Wnw=>`I+3IRrn_6QJj&-TTp+3l~M83Y- zR@fW~Xf&%n3l~f0Jl<{nYG>a~>nL0NnQ^~L(lj`DY>vCiJnyn~pVITbnpdhwl<69=9cF2*xhzzvroEnq;Ugq%7|)z_sMggde;oi1G@<8wX{7sfTL9) z+l6RQt>4P7+Zux*h4ez;=HY28`SH+MV^kH!?*VMO7z@mqgDINuxvgUo6)fNb0-Zsr z5rM_>{Z3k4@So^GWV03NdWxTc1?oTqw5hMHaf}rq#=X>o(#Bi^L#n~)!wo1Z?w8&d zj5f&k68@hbP0-=`!K>qY!AC_H8t7|K3*S6FQN>&h|G8sPTF*P)?6O|pz-SH9JiT`* z9&-u?u9TX^)2|G;;SY;~Ri0>uF3TDfik|hA%=*g@t5B)^oG~#*6=CjdIAZE|!EmND zKYDFd^=DVniX`Nt4d~K6dh?5o{fo{0$rsE!Q-uZmroklI@!s2`T+U}2u962JK1BLs z0FK+`^Glq5$oES~P#VxeiFy&t@sasQEAYp6?qB2d?Xh2?f(*}#s*e3xDZxz>e-Qn@ zJ#*%*F+AI>SG&cdDt%TgmPf6JvRTlI&4%!^NZVtN7i2oXwh6zNJSy@0#51*HmGEI(FW`E^e!m1L4WH4+PUGW=rm0TOKEv|t+(9k`hw<@_E zMcMHAtkgFKjp2jdN8@QUy{&j0*9)7TKa?+EAf)_^pVuRZ>dd@*d@bG;?@qlH=c6-W zR2Ocb?1Nj7TOm$xa+IpdKyj-@KGPl3S8}&P{8<@AMMdUn;^*KI$a`dpR4b)IJ!AEZ~l=)O( zDDHBI!Kh(@^CQ8q6>(220ORQ^so_e)nfn-8_Z$nus zEz$lT!{`5K;duU^S4sr<#EIb9x^%{7XS!UVX!BBq9 zN(mnQ^1BYc-#gZw-9o?8fu7o9fp4KsLtoL7Aeco2u1y~FeCYfXUGf*V|mhJbdnkPGA6skMM#AAWmLi^Oe+_0PI z&S>GKQqy_=yHM+_J_xlh+ys$z!yBNdC6oPd5`wwy0@>e%U|w?2?R!(H=`Ab#;9~v% z8^Wj{yNH$<@VbhyF}@j-+y3I_c%irU%sk?O8+VJvJiTXydULK+cSkw@*5n)2q`gct ztKiiGu{iE9q-*%h_RzYRbIH57fb{Elx9D1M5FgGcO8Gv75u;^Qj30>!XJ@FSuR`k? zUcPCiYh}J*FPo6Z&~YFh$N5$ntiT!&OEXW8eLn+;z17q7M^a)aRLK=H5FU3T$K@0~ zTjMQXVh$#2PKf&+TeFt`s2~2&6W`DO(NO=Uz_PqOeWxv(ea90QbsAY9N@UGXT}_%v zz<#^mT3mjlB*A4xf09jIkoJiyI2ZCJXHRl7^b>v&VlB{d)EWi`+1G zcp;c_EU?~EX}RPBrmWih!4&Zb?f#OaE0z8*mreixa&}$Sz`rJ z?tDdPTz_=4yyLWaO7NGJAG13D3{@e-aH>aomQoi7KW(t?Ub|ZMG?X7-k&)jm@4>8D zKnaZW(dv!WVYrz^np}wct0REnwBM@NJky(p+sYBrRCej>B?&*7!jp$+IG5=iHOPs) zJ>9Vu`}WBftxGCRmBZCl`xdktnsyiik3!}*+L)-7cWT}|G{eiB^H_keeXfq|wWsOB zs^{oYR|b5MdUa3Ja^)_*If5XkDk$l+MTD--_~*ciC-^g|z}xa=rpCAfOU`WVdp9`| zVon}%5f+(Eipp-$*HyhpOB{l^lzyDtKfnLNs-E7z_$(jkBcLyLxj0fvum4H&OL|f& z8U;pDEHq%Gs|NyX`a+ISGT4F)=t!)EW4>Btd}3Nw`nm4s$h8HsBW4BLKfQ?wuv+E% z>DqF(B0(iYha;ur#&3VI3KWVqn?_8nF&N_btLIcg(E;nwwI%53x%&dEU_E?A_nVZz zRZ`$Wen}P^s=NJyC22o1c>vZ zx%!a`FH2ByYPMqPWxr;7275*V%Y zOpF49xT~PN{hSecRJl}TJUb~ey?PA+*u~27IC$7`E0y!T2GIwWGBA=mwMcilx}K*D zVsTatu-$!ou-JP;9?hG79t-pgwPJ8fybfh8+CpklrCx#nW63P^=onNnJj$w65WBKf;%UFO( zr{~n^^^uTAv)L6hW9_OO&^1;t`eSkpE+g|XawaI`s;Y9YUJT&lI2ZDs zmF}Kzr#KvlaxwGpTJZd#GBGU zHWx&R{?`hLYb$^@Am6dFcWr2Cs%@xRp=~=ly1tB%XSM)1t;G~jZ`6f#3$M#92 zj(-Rbm01gXvKJ~^*H9_Of%~_mcHhk24mTWf)~%7zJr!cBh)p+na23b+16j;eO?0qD zNkN>d#`M`}XgKSN*9BQsM~R<0xstC#G} z)3kt?dgkSmk_{)duZ@xHKXCbF3V&GmXUPlst6aTj#*=qhzE*`#hxbo#so}$V>LR#qVGA_ey_XK)&9Y>|GxPTM(5{6g}W6s)N&zR z5=f#3#{@a4QttQBIovzf$~sj+_A`$k5^`n>DhGQeLO`|nKo`tuM8l8jz>KRrniEV7 zywHJx`Vy_;=}&d^uOuS>vsKEgYVe^(YQ1zqeA+PW%TP4gRaP=-6TxL5v)tq26cr}q zeD?c^S4jPvo>H;&(mKeh9ThYz+GH||G?^}|<-ZX(`J0`#*^VDA4pd54kH_%$qD>wg z$IW|r&&7``$9v<-u~}Z!o-n8!wE@Xj2he$GMmlEVx4v8)dAZ9^3VNtsI>?AXJF0h^JX*5n5-bggRxH@A zx42B{iABxV^}*o%MHu>Tc?t8OPudwO>*?ia_eJakq?ux}SBEbN6_gKN(>$mAjQC5U zx5)9sc+XdP8>!{b8=JWT7EFcN3#0NeO)VLGmS1$B-02zOypyigx?t5U7*1fxbge306R8n% zgJkhZ3FI;W@p+9t9yGYNEn62*Jey~(b(3aiTrB#*RX27zf?U8O_x6>^V!qG1uTpKw z&46KB{}Qud^i2~O$C*cz6vi1dcCaY-gSxc(Ip}M7l?r!F%5xJ>n6EdOh4@J`C7&jn zfI&ty7C798wWJ5Tnc+3^IN*#JRsVGRSKN+id19JSB0j^_z*WqiGSUqnOsa4qpnsA|meXev;aYqf;5! z5PDVF!D;+;yl&JY9o%7gIs_xNGheIuaL?T(U}Y(e+wJRHrHB@>q5{C@YR`?f;#+|=q2aVr1Q z>kM!^bKL%qoHmbrp9>;eRHZlZVPZ>YoIKS$l*!df>%$8b{uBGp9>4J325aF0E#vaT zoFs+ij2J#r2PaCKiMB3EGrj$ZQ__N$rX21FG+$FfKudQZD4C{lkoX2JYe5ju9$?IU!on!^}|+HfMvM7rwA;H3%M8yDLji*2w%BD&3S2Q2AqHj=#~CuHfJ8%6Ld zrg7bpQ1SA-QhlB_GIt4ZYUr8vtT1Q#tL3x3D|f7lV;WQx0mn70U^^xKl7VRf>RSHbD`$o^j>n?BU2rW%hChrQtNs)({ z+*lx-*QNGyRb654rr7|m&eDd~p?UTEG$V3^3dd&c{^vqqa&mlpIWwO5F2)pY?P~1~ ziM6t)LY%IAxQ=dipYq>cu>@)VI4lIBXRR3lJq>zixww?+6VBeRVk)Awh>$euAK(K`cP=A$S({6N5N= zFuk^(J>?ZVk&l{{;~`F?Sy(Qh!b$&AgZvf3QBb(Ad;wT^tH}CEtvC!U;W1?|pk8Rv ze=uiSBUycu&niVX(dCfwvv}bl0{ul&;KWg3SE6o1m+@5iG+E%tG@l#)e#fl`ZOr?5 ztQcz5K(EmF{G0QqW^n@O!ecFwu8(N`5$4c5gRMK|Hld@g^F^f9&PimtDX~=9f}A|h z=ooKvlg{jA=ug%gTksBjeIfEhy}Px53P9Lg_Qe(KYavrv2Sv$2tnGQ9ALqga2NE|} zP(~f{VZw5d=_E~pEla%x`PSQZ(9KwDwcIzvhqcA0f#`K*@7c~!Pvw#!mI3yvJ2f%% zEX<2V1Et(@)#FX~H&Io|49>F92ulRo%robk`86lqM^`H%N-u^yZOTOXuBz@Nk$5IN zQ2lHdqcvP`14C!}5SpDmn@IouWr0#;iwFPp$1rj|{I}bc%^P^f3!w7qNB3L5#ZlZy zOgkCAviND5)XT`^fGF1RUNx)NJ;xTiTl=h{_`Z@JIgB!G4D_AgmXeWM-~x7iJ`d?r zJmqewCZCyvHs3d#G*os95G~D*T*YZ?@74fWPmK~<rlKK=sgY)@7H zOH`V(HYSzj@eqM6LV4ZWWs?Cx{U63QhdQD>GBGkNEr5tbTOvK(`$li8$;m@VDq6{MGld2;>fWAS6S z%^=RInkwV?Va(G^z#M0BpBOU_BL zf&Eq(YAtxqi7)W>!!pbuQsPj+K+g-Kd76F;QNG_(Mw>)-r?jfHI{a%%d9V{^5A>Lx zWaM8)3I|{TvJxU*!`l_HMiB=R_r4*z$+B1_t%iB?)R#KwlN%Ii-zI(u>BWmwh>+`7 zSbddx<^2p<%g6d0i!${i&13(xewc%3RQLjkd-QlukPk=s%ZYY;khdvO6T`fAuUD&XIkwM6W!+_!BbLtZ-WgRn;2B92;*~T^M0m-pHNH=RuiaLgyx3x}p2rRk)`kUrtF}r&va;y(H6A-0fKc z2=AyZ$I=&2x1zcFmvyZ+EEg55TF2=mzp%YMg7^DMk~|4zbEgr~^0Ik+U6+yHb0p4< z6{Do}%I7m|rQRUHz<4J7h|}QmF(}okX8{l11Gq)DF1^*b+w#jdN(%h^TyFCUsoL}R zvwXR-R)9>fe#^bCNfLG<`?jG{{WN9Mol2!Bp<{u0=#gto&m{f-d&>l&*GtU?kWK*w3g`3QUUc~}B<^9&1e)_it z$$eYPA54Vp;k4?Fqg6^mrij)oN<&+wgkbF)1~Bg{5j=wd3_ z7V+M(md4kt^!L6=Pb1yVROfsNDT9&k$9Jlt_E>J0J298HqIwB(eyD`_uR_xpFI|_& zHMtVDSEChBy{vq-Jc7B52;R*={Nyvyp6pkUBSx#RGiL$U`*H>2%@hFn1+M3#IO>{Y zFRO3c&M}gnPmS>DeMRB6+fl%j7?^9_CFrtUJ)lt7;^!#N07xn(kzGTY)4-_rJYTuF z-@Q^ey~K^q8RBl=J0c5};sABjm$XssnLyST7dJ81T^>=BlQp5AVzksh-IuFM7@fh9 zCM_=y7@ksGjxY`u;@NBk2zT0u>n=!uJ>K|?&~eujC$<3=eAqDW*5Y$P+w<|+dfC{@ zb?6ow>(&YU!HT$RzCwR-jwHQmSVzlc1`a5VWV>V+pj}b=VrNX*^L@1TQ(7mZHw}@) z!$jQX8@Vak$stDM_yDyK6`)0iBx^#BB6?EBd`{G=4f&opsHrlXB>HMh@gqO3G+D$w zIZf4yU(`^m>g;a!Cl3X2PIjA+mfmo#AmcUN zwF+t(ze2~(G7~DC<-{|Pr>LO3N6k&p>KUim7HERU%cWa*0T+mq?|t#04CC(hp+Ou{ z_GG)gSX8*lqs|X|&CVlVYR6rqN{QgN>PgA%KP|ymO2mOq5>y<3TY*4nG;0}-tFqdC zq-$t^TlXw1KRji@X(PO$%9a4t52c{9zV`U!S}M@l{4G#M&Yf2BR$hL(`@Pp}u61;I zc|6yL*Oo3mUGY%?P5rb64#G9gBrKVYLVv7T;LCW*wTe-1`9;l@G*b=ruQFM5Y##s) z97TZQ9+1z0e0YM|YKpETjX3GlHRBbW-&{_%O=9qTtLY5JU(={Ybz3qwE2t^L-bv)F z=k7OHf=ey^_E1cK)2y{+RoXK;Q=?r4SYz*f{M_pE*49+yr)_UoPJw3kt8dNbNw^J$ zj0T+QuFY-Aw9MJ>uxzq3pAL*nBMd4x3tPI^si`Ow*SWdxAg{&D;Xn_(TXu=4O^p1) zw{TXx&2a^GyzZFkM?_cBMlf9zos!UgQi+N-}PvoWn-AX z!B8IWhG2ZfW~?S&(CcVKh*KW!ES@?MqxzF`{)^Ie_eo*zlgN~RG5@$kEshdQq(l1W zNfyW>wp#Z;sruh%^SQ9AL9#O;YP|P_dT8|kSEFx>N3j)7$t?9U3<|!(CcvWs@m3#Y zt0pDe@l}TzZBog9>ejZOM5da@H}~#5pWo96-$|;r4xdcOaFx<7G-#~fklvPL6e6{% z=%x}hefF6Cac*lVVO$nBAe<2j~k42 zNr%T;I+a3Q`7i1|^Q515Dos>Z`~aX9me=BD_ww;k%eVdK2At`MoFdAk&Rq$w;7|c< zylxlysiHird5y^%DVwrV3%YFF1Tc3d*)cW4)oaaTuu$v0^K8TW;#RMjS1*YmBOP0} z$(D67I7LV$*^CLj=`Y8ZU#|Na+tG@{0!OBN#e9*!-qleimUp^64LU%}p&sk1V zQ#UsWNAKwADHPtmwCQT>dlnFTWc=b{w3bFFj;W~#K)O`@(Si*XG@o^8h}J^Koj#p) z3vhtZumT<$F~?z?STG zF8T5pq=g&VMO;H%x_V9^_N#L7Xp!mbFsbiQ`@O6!_N5fW?Oj zkXJx>zS17=Ma{Ai!S_N$&yF1OToONe`t1oN%zMUr+7ejGExsdonSzVH>w$wBR&6Lx z53iq1!{dn|ZHM0@*v0o1xBMk9PaZbN-fLQ2G^_JE)K#}8CbpbT(#MnytV=h@Hc(8R z;Cqf?nff*j;p*5_ApO;IoKDz-OMiG8W8ECk?(in-K}JLnX+~!N{A%#n;V-#A$v?~66YT&&J?Rkx} zjIY!}B1u&o;iOLPg5iWsW-@Q}3KQP#12Z^BWW$wK)2MAQ#hj;t4h7snUz+R)Z`E*r zyY868HRdgVpRWG76+6Pi zWpJmU7?3-cbBLS{;&ozVJ&Z2x*0E@v?2(z47CF3HL+}Xq-gd%j;~VB8XVqs{&k8j9 zT>8A-{ed@E%PG{4=c|+j8OnpOE_=+oO1TP5?IreB!``#_O2FzbC4R1aI%XJXO`dQ1 zu3hMB>O#|k_$cNSZf^hP+WNU9?W2eHDfG=ASCHJJ>d2!Z+^_B3_%$&=o z(Vx|RTq?r4-MVHFl%-d1k*TMr_xaO#`ir~+v`NHugl>@#%9S}JWoImDdrG%USy^H3 z*ski!>+J{Dn)4PHB(ky*5s|R>`O1G^y71R>212pY2$S4V$tX07wk}*IbC9V8M7%UB zGZI)~Pu!xH#%M%$+D*_MA0QnyOu+!sklIcDL9c&aezrkr;e~hSoBYe^W+)UzB*nFI zSISB~0z$1EGyQe7LY~ReehU+xFBN;36K~GPJ2kaL6?c)=eBr*v>8*@M+Bz1!?lhWx zy3GmWEca8T%BgUkRD_Z>Kt`vo3_6FGxlccvJjrLb`C7VJ6$ax%tR$UDsr1j|z54v= zjpB&U1dJLHA%n&D3EtOlN6(GjzET%t64X|2z7hNu6~xtp=2p?d1k(4M>`kK`YG{Ii zw(Y4wn+kNoDWv6deUPu%mZ#i)J2vj$n~A6 zaF%wwB$3o}-ias8Z`&>1OsmpU$Do;PSu~b}&svmW+1+D%LTCfw)D^-%V78U+DDdyS7(m z>Ov}Sqe`%yuaV*_V(5Py0Jl#PN~o#Sozg8kmI?dp0hKuM)a-d(eSL~^Ai3ybOP}%E zuWa(wH)&h=xz)$z2cG36t`39m09)ujVh!OIb{A#I>m`W4wOx8Vc7uV5yCV2T_vnlX zosW@2ftJ(8tsS-#KnmW_%JdI*?=-er-_F`+v%lYIuVrgMXaHqce+=`Vg%w4|E!_%Y zw{K_ib)oL<`}-G_X$%&3N*c5%i?3*ELp~<9f;^ksC(UFxO(H}hWoAv(JXmOZv){1bBe{=$pbm@nHlb_*FH0%Ef(qs87#Q$m?lmoY@Z9 z4uu^4_5zGQDlWKVkT;(@O;acy1eli8;4*pL7O%NU3T_C=1IveZVJK;dyOitm&4&;( zL}e}xGZIYt`Wyq8i=XaALdFDN$y2YNVovjuwV532PQe$plAwJ9G ztMLw}3k7X}ZwKq>)gMfkxbF~5Ew3qDs=GPw?h`=fQ~@0;*SaxbS9Ocy zzz00Wm&3E^@E~?p%xKYJAsU5!_;+UI?1?54hHa(#f>3?mN)n-mX{<#m0;&D&4(p2> ZUp7@8R2{W`@pyh)FZgfX28#XmzW{XT1JM8g literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-14-Quarterly Percentage Data (2025-2026).jpg b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-14-Quarterly Percentage Data (2025-2026).jpg new file mode 100644 index 0000000000000000000000000000000000000000..f2ed9902c3cf470249e18c43d21f46e0b68e7c15 GIT binary patch literal 22947 zcmeIa2V7K3wl}&FrI8>aNwN|IMNmnS8zhU+f|8R85+zBFjUYjC5Ks^r36eo_&N)d2 z$ux*$=q5JMG~YgVX3nT{?wy(M-S^)8-aE~&=-#z=DSXa;NpEbML7Tu z4*>AMKj3r{kOl||2p|OK2q6&2`SXNCB$T8i#Ka`Dmng_68R?jr80i=ouCVcNTw%G% z%D}*R_xjD-cli1FnK^{Tg?Pny`1pB$%LMQI`ST>iB$r7^FY{hyxXSx)e@+_!azZ>A z{3(1q7T^py9zHqVX$!y%*7F?Re|&(ye(=uVpFKwaAv{k+3_ehL5jcZ~kALPY{<(8! z&w@{Tg7<;5%GE8DzyO=8{e~+KIe(3X)e>!v0vlh zyncgUK=7`Ru!yvbtem`p;{8XD)ipG=v~`S(pPQJPnOoR9I667IxVrhi_VW)2d=nHE z{Wc~x?%n(Nl+?8JPZ^n6*+s=ArDf$6l~qm6Ev;?s9i3f+L&GDZV_(N7kn;=Q7QZho zudJfCcXs#o4={&EzqJbw!2e6L!2f?~*e~rO2itY#>{g5od}w(VyYZKKMUW287|XE)(rd8c>OA^$}j*rJ%FFd%YG#`a z$!k=zT#8kSs%v1yQ+hXQo=ro#-0?%kQrDR91|0h?POi~J7DnB9!va}MV0LyM+AH9z zQHO=uv2d#Ee)R^P;FYMijq=58y7QFg>X?Z$yVTl6@RQ6 z#8@>-^D^57%-jfn{u|}-551%1XyWE0eMDfvR^H~#e4^NL&dP^V-VK~>o}q4S(OI%N zX8Tnk(l4CJ&jaE`B3@JrA_rkcXsGYHn5faymlMI42W2M+yTXx;ot^ts&W2-~y9lzC zI)XuBrL4=bT%Q8JyzSI}p_#AsoM|=PZt&Wqr%0P+cjZ~zdw$b>PqFS><7el38_2?tQbJhtO&r|C$V8Y7rO#GjG!Ai#tXB%`MyYl_oU%I-1+1{>_)scJwuWLKI%+meW;r; z%dGP72$0||xemE`Clc|AYzZc}PM7YU6q&!KF>&FIeYW276hB@n7!M2}%PQbI1^faQ zOomkIs@OC3IXz4vLSIBS`^xNssxyh}$vwz?6q=R9>{xFz#FCehE%Rv&U}ohEUkAwX zA~u-WCJ*V!oQ}ua?u5AwD$t7?HXlvb2RbNOseGkUjC(7t8gSOjvD;;Go*cEePG=B( z);T`m^Gs;eMJ8Y1Tj6fL%>y*63k5PJR|Pt6>1LqoqMOad#2AnPcZ+K{688_jfr)N) z^Q#ew88t2sTYZ?Dd`5|Zs}P^3E@p7O+`m*X46~d zxPPT~*Pc(JJ}#0$zC9t#$|n9Vb=!?nT&ijMguuvNj|S?NX5?oHgwEpFeE4fRIM=ai zfY8j7h+8;cl^*3ZC;rERaZU%Lo%ZHp_KjD%I_U`@=-g@hR=T?Zo zfP-l_!2VfKTYrsu3MSIQ&RYew z*AZNhuXU&s?RSJ;#JJ|}R`;>!H`Ev-O$;}qhuy}aTy`2q>&OQk0wb}Ll_CD~);)4l zfGA$HCS}q$<0!vyy~P-Cj2rNW9RlJzz3q&Ao8Dh)Sx& zYkk3|?p*;3MhY$v9<;?DYqZ!)|J>b1dThH_^jZmV$@-VE76f@)OkN_5;^3zdo3WE; zaw?PyLckftcEkIXNHV5LOv-on^SR$X4?j6ua!K$d$ew|3&Sy2Lkip$n6!YK91k6** z%^#!)Uwe|%O~8=!p0bL&hc3LWyH~DB#*#B;M)55+;hq(T#QQDTh|0u1onZlDE1o9H z$xay!RQ{3?&TWhv$;_xczcxNNDHla7WMB1NvAFLE!A!=Gm?l}lYWXaMhbmJ*8sD`T zd}~*`#&&J`$$rh;WjA&IH|eq&D~&R@LoN}rCsay&kJiw6n_#0HO82k8*?$FfKi(@K zu0I8$))9Lr%s3yeX)ZqmvJbIPioUy_Kc!OanXrDi476>`%*3cVT~jQsh?4QY7A5~U z6wky&2ylxw;9h_+8n|R*T-=*qO-#sr9poSt#=XJfVpegi1AT$~#I-l|JP{j2G`-{1 zHdl5#`d~Uch8WfQy*u!4ai_qG6v5T&)5~Ew4Ru6?sKg8ef1z?#S4GrAxr=Z|&)VZL z^A90G+sr>CD0o~_;x2xJSZMJa^3^kKS|K(bzK~;w-b;H7Hl<;A=gERMu(vEmC){LD ziXAG{viax`^Ab3pq8iq@6)Y4vE!}c=wKuohiY*ed`0-G&@3{z|PXXie&EDxW$EKJ6D_v|Kf=9xTwkLxvMHB};HfhM1etL7DM zPmA*sP6==6!(Z22H$`%Dj!%jzv4z%|zk5>cDN1Hx^-+#zc{U~J2I-j~&g6d<$cRyQ zv9a`sqc!_epjcBI9p9!DPCPNq62+ZQfOR4v>{|*F6Gk0t3}D>_tL^NJNAq`XD@=D| z58Ka4ZmsF#K0dX|`U-pQd^u{ym~+)VsBEv2NMcQ%168c*f6%c1In|dthhL4yl_3f$ ztvB1&E5}^4Qe{!WXJh_e>Ko5(zP(OtcG@XGTm$!Wy*v6g#>Ufd>3Bc96X9KJyaiWc#CpH$}4bHz{ zFRcmdp%{w6=cvrmU41m5eR+lmH9Z@=_g+kQp6V9CNxw|<8^V=WC#@$I89uwZP)3(j z-P*C3O|jzb_V2t4Y_Ix#y$OH22;>dFvPTTAXDgw38e@apCnSf5tM_qWMcnjFIK4>{yMFLsu-@vBOo37$VLAf2kb z-SMTGuc<4seHi~pJ=%((9NjfPsUB+Le=~6Y?Q@04XGxe+v+|uAsiWG_DiZ_=eNnWF zJM3#U-s8|jH+ZlFGf!=Hd#?FJNWAqYZ~Pl)q>Vn#`Hb9|CUQaB3O@f~KpUY}(R6=c+X#Y@KbGI6dX+scjat$Yv~drdRj42O@}ig-K;dmltkRg@M!3*yY;FBk9F))d06!uuV8XaHD`L|2 z%b8h=l_AiXKIX34)aLzSEsV2DX{B;>^NTku-xzK`B3zJqlb!q_)I!6>6kqY84QNpLf_0bxfqDEw*4TTVcNxf* zG&VQtu3MHN#x-rfa8E8qd=9(aeDaA=ey2bR#o^0}3VXN47A&yWMzVcNDlACckN&EM zGEM0gftdkYYR< zu+krAxEr+FjtYxQ!x*Z(OPhE#>uy^^t8&k;??4VlIZNqcg`?9m*C`9loGPw<`&L2j za>%9Iz#&YGVax{2+;~vOQeILw*p+y%BPvsiXHQ6sqD_i%?OpAFnAJk{8nLH6EErZ) zDjX&Cs```4-Aju2<&;D0sbkzGG=dRd2+n5aLxxS#KgrTHUy|`93Zf)?3d`-i_b=e_ zKf~+)r|(~&m}jb=x58hK6Om034`LkZE_dxa(1H^k<5DVhPk~SxO*c#3uPaQl>%k@W zO=c*Q=B#KX5TmuR$x2*Wi#uu)RME$4mmxdZ%1-PV14&k>g2r+;%ijFeNPbNUo(|TC zr#?wa6{VcJD1J-+tmj;gpuf?sa@hJrH-ir`!FXI|m&gji+?M@HBfQE&# zqVoejE$is}bsO>r4rZx_HW(56c66?_Q?2VT`kK2@`2~NdtvOFmz6uer5H-8%j7WnT8>ZQcX4$~6z`^d89vnKKq4G+)xKgre(1RSqL zvMFjhSm%pB{$4{p!401;59FA2vRZ$77$|(FDI=M9I!6cYhhsx6F4oQl2s{#vBWjh8 z(wqv3=w8ESA8Ni7M##7e-0cbJTnx}(NUU7i&KGU3`jVRLLeh0jHB+gGApU9z0qghY zyUTV4?sjvs#TYnE(892_|0gPItxbB{76)y<^+=M&Js(>%qMP?Qr*%Mnc49s307(5 z{a+`peSNB*6CLH*xv8xC$Joc!g%i+&*;uAtHi;6esr=ev>bG0)xZoNFc?vMa-#Xrb zvt@^U>qCmg9p693!L6ZNYw%F!3C-D^W4luTu8XxOr7HZQjT@{x1rU?r$49dZ^7}Cd zSlkqRcJvgW%aaBE$KCXVpyJIO>WPUi&;fy^qw2f$%ACUOo`IEShx+%;_N( z49d|3IN4ybBP{a~zGi^N=dVpRk92XD!Y{)6ogbY74VBQgaLj!%aN#924Lb$Sd;>eJ zGbO)eI(`=Q*2Nd+TZO=#>%{THn^fn?jxMaiF*F>QQDsUko?e?@(?OA(Xvhh~5sD>_ z##&wrx=CYFj%8=Matb7GW}E`AUV=mSW2OSpved-I|GYNA=f?AXX_x#j`5F(%`mBQk z{yMdVd6#kc6hQI7&;`#i(wPRU5|?XYk2dF5Wow?3iTQQVQ-nkhMH`$qSVC#yWJPC> za5g8@F%(}b#9@jlL0qg)Qhqf2)MU()h41x|ViMNr+;s!RUPm89S_=#J3Ge4?;;0XT z>g!EU0Wav$-hI-D@38^>l$w2D;pH{zC&$oZ^eHfw`r}i(hnL-DJn>{bP63eoy0;q6 z7M}w3;LL#{ZCrAW_u&T%r$B)`E16k%GG_te6!<9`%e_Sf_v*u8;d{&9o^Fuuo}=11 z1%8YLO)$0d1zY2%z-&0KmLWtuMZ!C!4*qjA^H0@_3@DNRXSIHLdq42#Owi#e@MB~4 zHbtnWssaLbC&$rCdOEFK8<*+d;|XDvIjH^9l3KxifS(vD{&#U$%fz- z5zjY(w0$P~6!=lJM)Y-Y(syU~&>Xne7Kt`S@&7QYsXt3^ky*x*YJvJ6MbGRPImnIY z{H*^!OPCRq;YS_? zS=xz*Q=^nH`CAVvhdl0F_U5m9-iy3y*GA>cl-S3xqgmhxGD)(Bul95+d7ivQu}AlY zNs5H*Jvw}hAoVaZM{3`42(|GnvAH(4cz%1zR^5B)Gd9~WNP>5;M%vu;(+Y1wV$@+c z=RH8EK{YaZv}CO?_CtB)>>Bk$k9#P(1WodUx7r_XVK0a>4{H)08McTbF0>`L0W+LWx{a~!{V~()S<_R~lV7n=)VCxz8K-HL4>yZUx%&?A`WAJnG=sT4 z*ow>y==zsN7R$T!r?0a1JZ-*sCc4Do?3nHx{A`Bbt~S%dyzHs=<8hJoh1ZaN=RJXT z56|q6p8k3loq!}b&i zSEe~RtdG*n9sBW8UH+vhDoI$DQlnky$233u#@mH7F&~mY!qpk$VrDcUr~sVIV-QN- zL(`pOUYCoo{uXiyV9mi(c_-)pZ~iKPNFP&vTWC!7MZ=`PRR5sik2lw0t%!}?vF$9>OMHrRgwOD!{%u z;y7^ul~&4VolZfP2N$S2z22Mx1{dMWrMatgLsMfQl=D6O0m|{eo&1O_dL#AU_&iSm z!%OgGxvVj%9nryb!olM=B`)njQ$h7~R_Na*gq#6!Q;N$Gd@ zL~`!LZ44$$^Etj!T4BIpgP=uwr7X*aQ26RgC1y1k+sF}T^3z`g;6DVyKpEYP6X!QP z_>NOFDE{$Eov{^`8i*4gCuHPnpzI)_)nRP z`A_UbvjY3GD3%D`Mznwfqx!4R{-w@8DR5-WCj1=y+oM`gZoi)b1Llv!gEJ~u_*G?} z{U%Gl^LH1rHnKw|=*B4Y`$)r>=Z{iY<0QOT0g8C%9}OeXQCw5~Hpenz4++5q>l|AA zB8SD9mN0vRAkLo`d}PskHism$aL}xLM4s(vuW<@Q^d6jeD1x|7m!S{24HifEE2jT; zJ^70U_#>4t{Ig2ng6fT=^q1)7skqE2FqR7M=lzYWFo3)e ziga9_-6a9-pY71LntMUQ7wtRtPl5k3tJFxy|ID0Hu^bD>WHg2_OLKF3Vrk}g52|hD zB-cIH@(Rc?9N?6{kP6}n;#jD_I`>;{@3YPPrXLMBT%C17@b*u^IP7dE*Qn(WS2t2C z`7kgB5Pyt5{L#q%7I%b?B*O-R9Nmx?et(PWAIbkX2)7DWR02jbU2K~@GE{JSNY3=acp1J?GFs&?!|1>IKNZFmPrZ>>S063izk z4{H}I2WQRutA7L$hH4m1oXiOYuI36jF)aj)El3fvP=$+wu@rUe&z1#h6ctdG}{zk?;d0z2v-_P6N0NWtvK ze#`!A{q9x6hhyC{zr>6_nwm&zl%;%3_j?BR(2JnrlNyDt1%e&v=P=TP1=nPe5G~N& z&k9z6a&+FEnIEiW}pFy89!=o0?(wUKx)66Hs!s;2d~ zcK)gN|39Vozk6K;Czo$7e2n}z=Dw2ikJu|=4nl+dyFZ$nzs2ot`M|etuR`}$z@3?( z{Z%yc&)0443C&1Q@Y6uc)L$GFezKMnX#aHi7vssxAkMrn+rK^k%QyZq*=y)FOU8*u z=LuI$KG{#mbl%!Q-WZf{7M~xmfRJ<}la2nJ6r)sx>i!z+@W>a3BTB>3_oqcB?&=FC zzPZ_J@jA*q={gj7xgWX=`2^Y`ZovHjgJZ`G>6k( zHFrU@CIuBP9*8AG!p-Bwi&b2!uwQTdT#+TRf4O=E))+qfvl3Lb+9sjrYmNM$Kwoi{ z{nC(hLt+T3>%jYlnGaCv%s-BUJeHLX zasuqmV$Z=)nf$l@y#&d`v%e2^>c%KyXAg3MrIN>yhvMVB-E44YGBB#(7`D9v`$rt? zB)msh&NBy-dC;X*vSSSA-(?1T>a~$1xFlgpI0d}3aFp<-JIU~VGG%Zu!L<$%fQAJ} zwciRYms?%i*J^wG%~7CnY&Z`IyHMC|Ch}BbKN2J~99R|ukiI7qfSbwQzeSj37#pvY zZF-c;dj5kb$^ZJWgq^$KzxE^pa(#SznriPK%40^oak0U9(;4E<$m1yO1By+%9W;j# zKgGhE*pc-|fH@p<9^A_Q!qki49xznmSu?J@taKKb7d1pfl; z{d+jIWC$tB2-!7xMkV`ACdwAl!?jZXH=xk4MIi$4I{NuP%b$pV@(iv9egJUQ>w1HC z7XQtm$>RLKeei!kqAB=wa63hsr{eNE&kgfB^IM( zZ$tBHz&MG`rbMw>2vX{7_&Z6+CPvzjaW*aMJYgDrgQKkfQJoZ9ua9tX(fnP!&cV1XdBr_QMgRiC?H@uX{I9^(=?^CqVe!S8qaw6Vfj-eb>{*akzX7-C zh@po|t{}yZ0xgXEcv-T~K~R(ew@Ssfpr}iRxNG4Ug`20qQsNTs47eC;=_g310#lwp zhrsUCfP7vNj*0)XCV%$HpK|i&-1!qv{>0cn_0E5zE`C@xvXP&mWuz3*`+=t1k&$mW zME&Sb>ibvMY_3GKY4AFh+v1Zf-*h)F`? zrSeO?fWFW1yh`EasxtqXE}BL10Rs7fBDjbwy8MlYfI9=zjiH#lZ$dQW0NLLvm0YrpsQkv?}M>7f>3efMcJU;2d! zNq4NfJ$Y%_7uTA^g;t+xZl<=I9G98&IP_*aQL@7Tmi?PI^R{lr53aJsL-g~*cnLqP zOjv!2He!r<%fiWeoGpCf-VaHF2772CmA=xnj27x?_t-7T0IbWJA;;w4QNMwy$#$`u zOv02RpEt=^DsBbeEn7hzyI+#l>lJd48U`-F)`Hf=_^bpF{``}!jhEW3U%gtl8}t^c zebj1jBcV;RipP(trcVAwe)a``_FIHVJbVJj3GWx_!yN^m0>Qb7P`#mD_!|-2$4uqL zw(0>%S`$(424+78au1z;CeLsiPdbY5JE>}rVS--3!GJc}AWp@8{<2ZfR*$F#Z`zB7 z_GykfON*RwLtP8O$m#K(XsIh*RY{~;zwe%&`RncpQG8N{W$McItbyWUtQ=1Db=t=FYhWLl)p=wj6Ux{p5x!&6d=(9?P1}bWalBxBUc;3p#h1buC2G99e&9+$BNfSUq2L zcq5h?%5QoU?)LRLNxIegn-;s#o_z;AcJ zj`y&*!>O52hzFib)!FEb=1_laFqxj0#&ehLLFReJiWi7Js%Z-mruR9@HUfKT_KMbx zsV1vrdhX#yI+S?LE}CuSZf38SrT-f(gl7T3A15OED zW&{^Up0eO!n-Nk|wUW=6hVr;#!n}=-^zTAWfo{GwNM)Wc+D4LMoqBAMFPyJi&r7#h zlHPziq4m;92JVum+0(XJj_w=wV^9Z##qw%cOwQLeCne{Msha~#=NYIZUON&L{Dbos z0VK#n(-dh*6>IyvBAVqgJz9tg*=ss^Ct^nNF-lM4(IR z{BGei>Heq;l8p8IB<@JKKr5MyPl6Y%>Ccl2qgd(Y&JAGlki^GO9m4wO#!|FqCRYVj z4RjRPeAK5xM1umfhpvqB2PY7}CW-opFCP9(yfL2w$-P@|I$V;a6=`$Df$4n*>kcJe zPX3e&THwu?Tya@cgHgBX_2`B^4M@Hv#ltZP7IO_X14CV{<};p4CgI!oPKx^l3boZi zO>OR#o#glYvv}X4B;%O3z=m@Rx!$ye zT^qL?HI2glrZwpAF#>CiPfE8$<4YHp)h2P7@7uhgCz=(8u&(Hrz2^>@a3K9u!)-!I zh>a5T7A0yM_24-w2;$Gvw^kDNPRp4u}s~CToPn@a=XRB59(t zVFEj!V z46Yyg?CG~W@p@kXvtQIHq>dKxoUf8kGw!1e(&n^0uItn4uGg(=3vWltfYBaN{z$K&^=!A(bZZe{9Yi;56? zX6~kC@`h4cUc&uWHNtcB#Pqhr#L0u5Vfy=J%7B~6d(mlgjkrq-J+YKd4Go<@{5<`m z;egh~p}I-v*D~wvyB>(uouM`T*f*!Z6;e@LEocMpmwnDmT#25*1aDmbEWLOr>iT8M zlpoKsWrnQ;)fQ{LiRXHGBj}>7etk0NuR1x4l^;dYj7Cn3PuBM`SZ5Pon04paX?525 z6!2+H0=Dm$X8%r4?0HqzDR4nV!y6#Ea1OBbqC@8=A0hk2BzQ>*)K1 z6M5y4M@s1Dfu)gQ)JQqVaF`U+eC@@Hl2?4l?r}w8!qa3kF)H(O$vf-q3G_x0&_FN4 zF2o$y=%@$T8I`R{eBENMi(eToc-qPW57DM!i$yiKu6|bCoW0H%bfpns9B#k1qKUG`3bdf_g|PXmn?vVMEz zNMtTXf6tK8T&HR(&GMO-S=YCE;ko_1P$i|JQIN6WrcGZ}PE3bqiiZNOC*^@+_VZB( zI_xFX8doC1uRN+ozgR+D&bo;;FEA4s7BLOxP9^V=RC?ENxsP_i2g<_LzmfN`xGZR5 z<_^Oh&c~|#;F(yofyiJHi2fUM11Twa*AKzvlpCfjh$CIzE5!V%9a z6Gy^}$7q)XFny}>blX&BZ9ItfKiL(mAjOT|L42uTKD!ow_cfO1U~HRDf2OO|DAZY@ ztA;H3i(5QU?U(~Po$tb}hM$zxmKYcl=4^*!E)X8g1#LO#s|qBt%sy0j?q{Mu$)-n+ zW1Xqwf(BKQPR+>n&}ejEZ6fh+E%?Az^m}+^`LDC_6i{$-cDy9q2m(C+z|-t^=P_59 z-(_=VDgoW=xJS0e=aUXa+rM*-5okHPzyaXWjiL?Sr&uOi}T*G$-$V+YmbHB!N0Zzuad%&KDDf48}-o4)G<4VB&GYZ z*hc;QX?4 z9Glc05#!5T)uf%u;jRhFcDE}pe=5o7?B{XIGk4Xf1;^ffzeT*Z`U0gBn-l(LtmQz? zmwKcq8g_Et8F8=*nuV*CFC49@4diWZ-JI!~mzEf_@{hI9uceT9Gpyq$_t|&$I{oa!5Bl`sEg~+B#Zoy!MI8+nu?DLGy zb3goA>T_mtMJ5$U@)I(q787rAwVEs~OJ+Kc4hgdjSWk0C$2(|NE1w=FJiqT#=e{zv zm4x6S-FL~}a)|&l0OvS}6-8~7#ifQST3=dS-IM0e>A*iu728|el4MMDJqY$NvbP5< z^MqUeuab@)6=|p9z-3}V)&{l!d#`Tn$cSTO-8l< zbq{7|1{;qSrYBeP z78J=g;vp%ZY}gzhI)C+$JG3S!NR+I4=&&Cmo3dOL5VrSxY;W<-N(X#kqYA?0dxCg0 zQ!DlOhTp?8ue>1^JL%}YAo4RDKKUb$pWsm@v!VJ44?b6v3KnZOpvc2-w%rL& zw>>%u2r9XCNhyDYj3?FV_8BFU!=Qo#r?8pV_9xGx=jDp|<>+BV`ou-I@7g>eJN_sI zO$1a?I*AnXDjBJ|)%0=~1fxG#Ey8a`8szCQdt@c&!hR3Q$!^rch^l#>5X}e7Cjc@2$h05LousQAqCSThhHz zyCnCRSuYmaqj5K|Ml;VP%F^n__t|5*u9b$KMt^itdLS+BF4bxkL9kV_`Ngj~+nKyy z+^U(<`UyfJ(*{ajrqsN+-PL?0{2rIcR&$lWUa@Z9C!LZPcUq(rj>tRm`83PWHvToT zb~!6;`WkI>-xXA}GS>ye-mS|oJV=-qEO#XLg`rBj^glP4Od(-IW;h2tP2_IW3X{M{ z$=R5+617DsaiY%vPOes^0*>dRvZENaq%jhXeZaG`ZOV=(EG&^WW+!>h4#$hmHhW@z z^fEY9$0mkVwYt0Sp1l1s7p!&c=o2DYi6`8Wy>;QXEP0z=rnoNtU`=&%U%r|1_9}%7 z>|uRfL=;CnEpO#ocwE+A@`Z&)uM5%2k)?@GLncZNjVhM9`V28N+bjB6cyDt6sX5=7 znk^hfJ?Q$<9|z7WuXK1;HNpH@1l;32_`o0x+?zG z+wLfS(~_*ofZ33A`Hof`*_~^O9~^FaiU*u2|5?cL;cffj?sL;BoQ*szGr(4f?+I6O z+V92!z|N4^#;6kIGnqriJE5V$^j;XUcfwUxaH|$IQ#&_pt^cU`PD0Lg72{HTJN)Bx zx5#CAfhXAeBjaLesKl`eHzRxSB~m>&N<5?ak###4rhPJq&a+rC7N*p`)uWvvND){k>NqE&4 zLK8ehWjR}Fh5_}hZpk2PfOKg^tgiocwg{Oo1Nov)><~J0*gKASE>GE9YUn-foH=GI zqG<0L^?TG10b5x5IZTa!*^o+;)qT77>1*Y6132-u^d!RkCoUYJ#%~%m_E`zz{c`smS2oWIIa}X91`PBG22#T4@)Gm zUSbcNlK;#h#MD%|^+t#EHIBAoScbga`GtC{`a|L~l2?xQa)Ze(VhS43q@wBrCd1tG zhtHCD-JA}+uY?=hl`LI@%t?40toPg$iH=qG;%A0VQ#J+$AW<2fZbS&m_;oU*gUAPCYh9 zv+BGOOXxd4b&_W56ekJm{|L&WJtf8?Mh-NPNQj{^toUd2` zXWu7VXp9}yeAoj%|CN8*L@Gm?T1zM9hNl9s_e?Ztn(OPvT+rx@sLMq*)l3t(yaH*{m4m`XTnRRN?c%8}hJz9*)60gYP?jp_h2S87~Mi+DX!S-9e{G zFa_%9yt2!_e}_Rj@%^&y(S<$v#w4<&iN(?r$B>EKS>BwTMbk=9)DAT%J~Fy1jIfXJ z)Np04(tL5gS=`Lg{?VsI@z6nQf8!g?L$j+~6irOD)gWT+7g?}u zj!$}+FIsWBABbC~|9dY)y)CAVbP;>>iMzPtYLehI<)pLK-sJ^f&0Qt>yt@iMUpCkE zyNT8*UQAZXKL|;()A`3Uoqu>-?yr3VxoSTbXGoI~$ltwJ@=Pk0f*N@h$Fq=m;+kLg zr4AqzS>Ecv7M#r-RoeY^IH6`Bvo5hVu^kt6~VB{hNsksu%-AVIQ{L}+r( zL2?F34U$7Qv4N)hFXx<@0l%5^oqL~o?*E_vPV*GKcUA3OJFNAtckQ*S@gw+YfJQ|@ zSpgs*0008;55SKDasU|#2`LFN87V0#IXM{x72RnnN=hoGGiM=mY|K!0HfB~<4z8=* z92fXFSy_2RF7sWxE+ixb@FZjRw!Pf$` zWCXVedk6_G0H2V#?Be>;&G)sRe?Z`y zpoqx#QPD9UV&hU$)6z3Cvp#?MR#=26E-5W5uWx8 zlbn&&AT_Z+dqMCO8Qrb$PX%@4oHy>H=pVo6r8vhWG;s(y~=^vRt!LT zrB^@&)C9ZIw0(QPrqVOLB}+2p=Xy$n%&Nl{&5y(BevMKw-I0U4aZiGkIImW8%Xh<~ z9%06{zFqKK(i(Ei3PwBnmgbaXuZZnnoc_Yb2p zmO7WUIA(3{(70(MTnqJVA(*Fo_vKIXDF@aqm42kcUHaVI_}S24ET!_+lq&jZ`ixkm z2_7Jq8oTo;UDa-tKwffNlnuhzKcTw)F1@?MHO}A2@1x^H5FpRi>2}pGqdVR5+Sr7#y1N~TS)w_gZpEw= zX>!`2f5K?GJl%ap4S71?pB)c%g~l&Ck>$@J-sBx^|XeW1R6o{vICaQFC(%!SZsyEpp8K)PZeEVwVR~rTgSR1V?7?`WFAmK-nyz!a5-wMR;ho^$%W%ogN=9g zZL!fs0d@L=jK(w^>^>xT0GScl{EWXsUm_1aemqw9_c0W59+zp`FVH_vHI@6GSWU5gQjk1*mAwkc9n^ri_d3H9QAt6d;BZGLRPo% z;v^Mnp!!@dbG>z)WAoFtNj{7dvwcwkbhM%=F| z{KeSn%noHm0#vC4buNH=Okn=qQMryA!}EgA+k~HY4~C*6^Tu7zDE2rSvKyM~thfu^ zKAIVPK+NTb<;}!6Ov5pXYp|jn`iU<;^6swRx#I4r4sS1m1CwKmNi^BBx8(eR6M z#4_u;#|vln_Umo91W*66*vjjvi$mfQYawy+_D)2>s}dfWe%nk7r@ru?<(h0^Lj~v$ z3R|yfII-QDUow3|L7py96yf%?JdAaT{nFt5R#%sMH~j@Cow+Uc%o&%4tL&jf@Hx(G zJTT0I2V(P|rsyeY3~~EgyP1c1NM#D|rXrn#&exv8-RknBGmX@%p?r$Io1dL&uPwC{Yyz_#9H z`luf9uVyCiWT%7ld}UICcz}PwLCC&Pe;_qqL0u>nA+^ox=g@O-Mk0JTcs4H;J; zveL;Q=@Cyz&Y7xqx8>KEDT*ut`H51%nMJG0wyB&KVEIS zC^AjNh`Vro8=KwZ;XbQWc{zRcb@sA05XiYw1ot?E3=AYkPb9gm5N@!sUvO|ABdSot>qO^h7zItTGk5}O#f zipZ1?lOTls*JQ*?3*H08pdorXE#wzw0x`e2W6lpeXTfSZY^WgG}=@Z|BnAZA~S*NL@h?1<( z-EZ9t6NBT}LS9pzuC!M=6^ad2UiVxX;^VW~5g^kYHEibcvF7O=*dr!>ko)Aet#j;r zq5!7A*|sy^-%yK8dF$#O6Kr=fCxPeTuraE?kvS$1D0+cCEo;UvhYkNd&;rz3QejmZRg1drdFPv3G zwfJ8O8mJ5BTPJ8EJ{W&1OB8n8Pn^#;Vevt&(v8BgEzjQKGS$f=8LQBU>5q9m6)!bsI zTezvsfx8h#qnd5CbrjGLO-fYp=*q6f^4b)1#wg|!X0~4YMMJt}47&LN+02_1i#oz& zBJaz3t!e9Bc|kl`eE*E6UI|6u0S-1fTvggH&SI36m)p1AXKhJI6w7>*ci`0ZCct&C z4!Tdaj0bkwVe{5yX!Ecf4{o@*rM-=0t{By7D#2ONR#>B_pZbg~p=Ba%+`K;d`kb{N z1sy{TJormCaD|R zS0O;{s;Z1aR48WD%GDgSrAZ{&(3Q88yf$qZ=Cv5PlX$9(1L` z(TYvVX_n7=7k4|$(prdI9I)PaU}ZSlH9+=lqnoMa9)pWBO0453HcB<@mP+Ck*lCQZ z8mc()+4h{rjpr}TI4Y7}0JsN{N;V>&7Ha5)vK(Q=OXAK+Orc%8!PlUT<9Eq-xZ`44 z&`YK>M(-cd1iyqm3avF%yg}QTdlY%$G2ed=r_0wp@BQf}t zFHU#`MgyC>H3VA;*rDH&w!J0XpB-`G8g4U7RhNZv99 z4pH#tu;ZChW}ct2-wzZ1c|+)4B+q+x9pV^BgJ4S?I4l&MdMuqs5+rudL3#o7&M1AG zpxZz3fENTaY>t|05pM}{imxzKT*Vn^;U4d6tcm>~&98EGax(tv%-M$;Y~o-c-PDbCU&W> z2^1V9=-SV{jQ^&1D96V81K8O+5_}h5kby&lmxb~QI_k!Qidgq~?>{57QB0{~+iw?Xd zrySj|R%5+&NpNGcvhoQ*@RI`#xO1I(r-Rq@tw#=aoY%5`Og>_fVe7JP+No05LaM)c z+~cV6(_*(X+OG6IsW#_6#rD)WjMrXcx$&-Nb@ncq0S*%=*}$>gHS-F)os{>7sm`3ryxb6Grwwi6&W#b?( zXz#WSp-Ndl|9M1$8Et^ytr)|yHqOtjtcSy&FeUZTUi^K2=^aJ)6~<>`6;wp}OLy~b z4bQ0kZ2I|mxKd8|C?TJ=&VOks;%;y4lA4g}TPae7IQ@h(^)CzO+iY2+Jyn7S$K)T% zQzR2>c$1o_-n009s{WGKj>L>3jY?*w=nXBS?R1K0uh0uNH}Yvc81^ds47@up78D<2Jn$A2G1wc}!tfnDP$Pf`n)>HV)@V2A zw0-eF{Q{aBd$5$W2Z|b|95&;CiNtw2b{}ry0jwcwuvCf^G6)677~r_@K$77N9JU4z zFo3#svi|UZBE?D#Bgk(JTEqi>-AMKQoqpIx2OdB>zPPp8jOFt$b$>5u35@QtBjR-F&ei)qX=DpN&1i#XWG0)X?HEm+Bw?4gVD|iF) zVK(rGh}=?N1@w*dRn*T7KlA?F4U?+!51#{N>gs76O2cbsrLjZNg^1vMC-;QrnEX+z zNp`iTBo{^+pFDEQkVZ+*l2rC}7^9~`%xbE8@}|FiLp*XI906(uIRQ=1JVms^&m z9@=hx#wqqtI=&y;SKx6G*^Am`(^MSdyDHxuTg?$9_RREScwMa+Q(9Es%?;~>jfQ+i z1J~5Mr9npzZPyt2{1ga%Ke9SYvi5s%V19ZCJmG&9Yy9w|*P4A2KfF?LlfD5Xr)f)! zanVFmf1=ghIY(*s{0Ea#8-LjP+)f6@iiIa%F}Tv&?Mr`|yV8^Uo#&w~MW>uYeG$?R z-2-kQ!tS6Rp{e4;Con-P-^-Gy-UUWUrmuY@XglIf%Q1vB#c>CA|Uk2+0HFua49m>PPW_YVqIyw$-D_pC7JojRnUE05#dDiPP9%z#s zfKR@~0}A}YrRfFp@Do7(JFIK{$ZGh10oQv^qCBBIYnE7&Lp3POb;z)T7of6|Vl7VO ztY4*sK49%Gn>uC-g^s0#YkL*1P@DhQ%lA(d9Q4qZPK%2Pf9n*S`99tY{cIu0#K_M` z3 zF+%EU(3U#J&<%P-h=8`iFkfIZPK)qu>H#>2Rmjhq1rDiwi}^yo6$P=-C;$)0tSN4) z~?lVIm16KvAL5@L&-#*?jFH^9;97U^su>4R?zh(JdiR}+P;dr z0&lxD0Gj~SK|a?$s(fT2dKF}pPyV1(zm&|=G#%c`-hr!{1&3SSBf(f_$#|3W8@L|o ziU+o7bq9nq^izU(Z`?aEbN{iZLg+IjDK?$Nfbl^N=H0UDC0Ou`mdTki@L>$a%(>+0 zHz%^8x_vyvgsT$w5&AzALL^^JoE3Wv#>$#+gzl6fp|5Ycg=|eOiDc9(+H}&YwIw~Q zBZljJPQx<25zv~|9b{f3Xba-vtyykdfG!AI=FQuP^bBBR)7wmExXYtd;@g6QAbT>k zu&(k6xJJLD%~eBHO`Y#{!FfN4-#Sw!&KH^ZOSpmy$M=QqeF?foNA0c6g4{S|fT_^% zQBAPd9&5_;Ka<9@E$6MHXcpKbLe*F-fJ9WBEvBxB*u&m2QMtJG9q zTH0n~ZB<;JSBh|7OlhLiX4-qms-KeM@JWQQgJ@{rU6$FSDwD?`yVXgfAU7R?plx&91PX`2-vT3XLDUR zku{a-oI!ySbYJJUC3u2P4;efVPN{#1sx##RL*_Etr{QjTL)M1np+ zPS32IC#6vM1RnExgOJoP16SQR-24mB*%iH}{?}_`{8~#Pjcpi=xkr_hhdy4lM?r#+jOwkaEc5_P)j@}#e zk8=lS?eA!3VA}-S=Z*_I?1e8j<+Yy_d9~-4pLHi#qy+t#KQWfa;=oeo({QY(weh+f zcCsq85{zd2Jy(!8kHXvv;`#iCRhRi{4tT3y4tyB0%sU@+V2mWi2I{pO+2K;EEpkl? zLe#*y&%W^wpLtT6UnczKyy??J0{6gz1>&cUsWXHJ1Fm$lGsBLYV)pr62i*mJ`EV$Z zJ*?!^v{^T2TRby?1%e=ua=g^v(p87kAeRqpz*VQ=qNas6ra-nlrvTmA;oiof@1$am+*|NKa!}W&T)_!C4c$1uW0VehelI0-TU)mMxRxDHJK_Oq zu-S0XaRr;!qvpwLEbJix!VDfz0&ynuS1@lo*G#JO^D0B|a15mx;!*_TP+AdDEA;Z%);jXF-C;us5~IudMJ)=AFw zpW%>vFANMMc-KHX{~4CFfr2DdpClHe*azTZfL|88f&uqd zZmi3(yc{;79pN^B8R?o%M zD*B!m$ED|QNDfLA8?2#bAG#fOungZmYpgQdY5beq=&Czd{X&tyJKXfmSWU3qFG1Q@ z#efUeTg?u^fC9y}1!AC^z!7875HkAv{!kE4Zr7ufe;hp`9P$FSs$~d)n*&$%M^;*- zPh^?aZ6U9JS?T)UKU^rwzk>nY9L}pCD>i~3iVw#Q zXLx`XK0F50+I-kif|XgpJxmx#(Rt4%A6L`Ou#?77`y1944}_hD;>tU)(Un%k{>^7G z;52{XI9}?nY2M@3(TLFK2nEO8Ho*hN_u(pURxQ;Xs=)_x9v>#c1>2yc&3g#Q-eZu6 zSR0r*!%x!FiDwQS&Y;o?-ui8DPW{UNCKT0o^K%^Y_X$Vg!_2+57&5r2pY}^A09~ROL1q`{IQXh64W|!yt(`1C-{?LQj_{Vk6yrG0elgWCs?A- zx;X)=`<-{c4Gr)se?lDcBePE!A94Sp4o407r&h;JiATU>kVLN`DbF#HIlhgApdb4O z!{RY`Qexik-OT(75=xR+f5oQ%!&D?EdxC;@B_m-2!SfclDf^Pk;`(AT65Lf#&kt&| zteQ0*AQ1!?uz}~jI>xBC`k>9!)WZXH!8PE*_=O?RlcO7|L756*dWcWRQ*~?w$}D6j zEqa*Ypz>Yowbs4OYn(sM$mSSh9sh~ZOiX6WJo4Mzz z-?xXu1E2tXDeIN)OeBzZ2MNd(2&J159+<)dI0O`xCCR@8o`vIqg8_%+-DfBYUhvBC zvcH^g*%@2asl#7Y4W%ZK7Mgo&fHmL&3g7z4m>TfOcpz8tM8&L&Qt^M*sH_Fq)zN7> zZLlNt$*7X_jiX-`t$<*EqF!;x({A8*kTpMfleS~G|JbGQ8=a=o=|rIwx~UaUz;S6u zM#fp#``%qJkvD<=HXqQFbDT(3~Fr!NHmZ?qxj}P?n%d5ZNeSj}{runW1sq!s`x}S<_qqT7$i)@Eu6!FyG5_0TeAgV< zDoU}nO<)9ooHz-)1xDj)mEo8;U?2r_?PfR%@j#+W5{6s=51duRa-D;3b#37Rf{S1{ z&4TcLPzOj~e7bQjC*jjvHCQz$>YXkg=$7cl62;(wH_4|}cm;50C}H~xVARh$0`zvl z(4;I#nI$)1s58d6OWwjfIW-t1K0L6Huz)+I3hrP9+3vqv^6xSEx19W2cm9nh|HjyV z+ns-%Exzuwa`RD&#P`Th8!qS}zh%$yCGh_%)4F4mCFaSN5ynPkwD}aN8xqchlSS6; zShwz6%!YE0DlITt3el2GYQ0Cai{nCfzv(crh`u&DYEI7SCx+F}pXMn=GlXJqyiIu8 z&eFk$U4P|@)y_on_C509wczxjdk`c{5tIrLs0wLxQ#M>74H?MhrSgGT0o^@_4j9eS z8_y?=pJsZiH%08MO5R{K%w|F)yX{=M_GxM%-xm%bZZnm%6|$IsFBU2-s~Z&-u+`1X z%}W)$NhFrT)jWnTmeOJeWpud2WE#Mr&x&Iw@B+tx&I}Y5I9HcpW#2&ef>SJ59^pM+ z=6=MRgGO|ACcPF|x4GnM9`YdjQyuhL-YtRst7qV)@a=~#akvjo2^1T5b%|`mZKxGw z=Dm4b*_Y?{>d%RYL{|1IF&2fsQ2%b7I9O@>Wm&I(5qGvo^^GZ)@{8Fk*N2&=Ar3?3 zM}#WkH(yz1RKC6JM-fd(P{y4(tKKgpC8A*Mn>ohC?!FOe>u^Mh0PM8mS{fmz8wC)e zYx#TwVb`wYx5K7IIagZqHf{~s>>OQSyET5Ei}8t+M!-F*m=J;79FL1xhz_1Jw|Xxm zA)hh=@WjC|RNjy^oSf+*5#Lr}Gi1+>1b)rp6pRIvq}wN4C85^a$j{&BIZDS`@QhlS z@k^351{@@r71-gj0PgA+*RJ)+DWr1<|3EhDIdp18i45WOTwO1I3)=1i{*5#l`*}Vq zf`e0|$){nI-N+?dZuScZA4Vd0#r5}G#qd$nvGTaiUZ*vQ#&iA)cT3Okh&pjn1@r{< z{bB1HNky7QQN*w4Tq_UtfwTft zr2JWW>(94uhplg_Z#G%DGAO-OMx}qaeMr|Cos9Go9q@QEu<(pybwF1!pt);8(H15oaV`D557P+~zv<+wc={&9Z0nKNzA1=iC`BiE=tr9FEGus&eFh^`u7F72t5 zvbwr+!EROIAtXae@r5gW$qW<%^LJGLsx00$X{}_aP52S+XIRuz)WN1Qt5IAO%A5V_ z4V(K`VnX2*n+2g?StTYiZ5(mbKK7%5ae_-FOC=N!xR7pMD&L@Rjf2ox!mL9iKfhaz z<6E$HcW3S!6yV=Jx;nqSRNuh*q(BO@gc&A41&x_lf1ZfqCHi(Hvy7>Gi;kcWP1sl- z5Wvr;fA{*!!uEELriohUky{%>(%Dub%b^djHPG{F8Zf3$GcI1#s32Cq9{XmIL_4mp z9Fse9Xc?@b`0~XU`*}r13a|J;SURAK+Gz>p8tirIkZ{d$E}p*?+a0igX?#OrG5~9> z(fcGMd_eG8!(}ES?I?_)dVn;~ltu3ue)Avno}!aA?^=_o8=7vo=3UNh3E)qT6U^)q zq-T_>M$>n-!O?;|W6FHbH0aD?6FGn#n3S8m>qc^g$9O z;RfmQ7RE9lH_l*ujn8eRLCk2RJ;GbOD_I?q^ae2TSyM}K^f|i>FUG}6DLPpmg%_a@ zw7Op>$ubMdC9APUqmQtjo0B;ZF`+3&S9`2_+Qs)Sl!ScsYTqjYfMBm8f=qL}^F3?p z&(mx5onAfxe_wzrZaTi1{SsOV+k?^SyigQ4{o^}{`mhV(l8<$d+IzvJ`gWvH+*!tu zL4P?a)ASp6`c{kFFBX3FmOB-3%QsQIY(<9M9TwzaG9&Rpzfr4aLRV${P608IC&je+ zNT_um|J6@bMgGeZdkuo53OgjFH9;lCD2$Gt1fk^S=MeYl7wX$sYxla{LS0wIJ3iC& zQ6@yb1Lp5woHDF=WyqwnV2I1o{?nqyc)Nz&74hJs9(Ar)lVJ!fxJMw_Dd}j30o=^5 zYnd~$WK6%~DR(cv!=KBHlqkY36`Tf zGPP#Yq-a1xUbNQNqx}2tP8U$WO^j`kJfLU@HGVUkb(Pj8_hA{WU^8Uyg(A`_rlG?= z-?ikya6V})m)eIt2gHYuUNVoX^guamX&EaH`<#)0Tp=LAokL#tmg*8D+hgPHjf(0u zWEA|`%7K!1V2AZx81&$F{_frREW;w6)khp>Dr`KHk$&mZ3|WL0-)>jW+yNEs0jn1s z!H72TtS#%0B@y4dy-^n{Lhi&+b(AV3n?sFu3TFk*W(El5Pnc}1ORKsbe8dCp>^@~# z*lPzO-oo*R*tRc|`i0-u+U~s^$4TP>6<`T_HfLgGk~s*s{V5^MNrHh8KsP*}#6kKp^ z6@^9tHdf+(#Zj4^?k<;F&l`S+C-3D;`8FmBp&vx3C+jtsKgxOdL$F67(p_DrrMq}lbsFt%f6`Dikzs!c3C7CLROwL)(2pw&ZL#u$ z@a>MSBP^{IUb<5EUhthP6Ol{45C-8ugO%1LvRf8K+kt=8G zt2z4}`QeQs2fpNpI)B3x9i zwI2K_i0V-5As)i--NBU)XWnM@Hy*s`us+|CL<8W6y)?H)g%_hmiVs5GYez?rb40mt zQX@YNIkpOiT$rz)A>_)E={6k+v$KAUQS}rU-QmBY#iifMMw;=04BG_Qy4EynERq## zSzL9p{CeRv)mrzhS})%!b5}@{cyq<)<#TaafhFb@J9d*;jyJR!U%?fXwMZ)&t!IvJGf#H7)ciP$}*VUNEtlx;8TkKO>IC`gi z=^NH$rUfniE?VOUA5|f_h~PTwsp|(rtwnEgh|^lXHT?`6)p>HQS^OSplG-ec2Atfu z{(7xI{!gOTpJf)N5=r|b@BP5UG*{C_Uq-CB3h)^g7dF$Gj?gzl2DFox+~h;wTG%)q zI{BYzYWGp+{xN(5Eq*XIUVmo1y<@>ZJV*7%J6Uef&?Hmm)$g8Gsiz!&IziRS1}VtX zZkoL5$-F7Wty0Y3mQz~)PU+6}v=etom^6zh0w!kwd5y3p!^-En z#_N3KQQ$qNp!}5EkFAS@htOmmlVAfG8|BZqC&FS^5E*ud>MVwJSYrlqR^(05aGbsv zf2QBA5}5=!9(V#0sILsj`9SlrA(swz`*|DVpXC*50vZK_eD^o*MqMoNcz4e<;_fnTz`vGeVw!pJ)$>CFmhs#_maLYh5lA{cwP<{{!=6dmk)JkTrqkpL##2l zQ52=u?@es$#sFMeHrUpaxg5HWu$>O1*r4A)Du}3j=(daXRRL zxih4up{XN--K*o!Z%6T~S~!L4KVQl{m;R7+Tk4SMAsV) z;yLBplb443JIW)aA8`AS&vH4}PE%})|AA%ANb`C7W!? za@ACeyGcrO<)(ySFzta>)Xsb6kl-Pg%hgQpQu7P)1$|nYj3N1#^eMFlg<@Hp2k$q6 zMm=+i=pN~p;VA6AcUJ7@dy76uTNXx;K3)ZI4DcL|hU9YNhfs&cathKG#P8?a8bW40U{EO-rLsxB$0fCR}y5|1s z@Ao)tRyJhVN_-`uQiu-ggB2!(Znc^ALVLFZZg?|qxpD<}_ScIBJuzSmz$+a@dh4o! zMEwNmv6r8&b?~WdoF#n!nBa=TQlB^91BQ~~1JAS=7YDw33L-ZOrXTFg;>6)+R>ZfX zYevOi$C(v<2;q3O+ZKlWLC^-$VZAsTb)<9toDBQ0ObI-;TN^f5RiT|RqN8sKy6I!L zIejjUeSvlBZDm8wVSMj=3+n2wEzyu=tpC#VE21$}ZcB*?(;2e86mY-CygyZGk1vAL zyM@Da{*>$QUx|8@nXs)miN8Ic`9$Xmv`gM(p#Swj}1 z((+OnyMmy^~DU`JxX>iB<2v z*iR?N)^-ccO(Ak~2yla{?38awk;)8V6_G_~_L}E5k6-&zQQeFTD+eUK#kYp41pg3$ zlFV)y?Za^_j^+67K`SoSOgx9*;vJ;-%`)To3|i?N7(K=ca(3yoB`4XuTs4z@LN1Q9 z%3h0k?`)iVA%f+HKlLkW}8Fk#P!)HiLyB|4wL*}jXQIG(BM??DOsR>t~3(zTTW!`btSUa*H z%>5#eYWaebcd7VUZbEHsB`&H1)1rcA99TnCQ7kL#P1rqan`LW-HYLvHr<=p;+lDy9 z+T!b7jICK|C|@OIe8({*Q;pi3tv9BB_NbpcbPSbjAv;Yf%DEICZd7PwGm}rfqSqZn z`leEjwuHw;j*yBq^rI8tz7@9}2k=4~aXjBp&57ov5({>>uMuvE;e02HcF5-u{dwjk zuRDe8Z*V<2W9U6?Z^YxzCQjUwitfXmU-D!a7Jy*s)K*f>J|yQefBNXr zI~o~M-&PXu3uUN8KbJ43RXlGpl#^ZzoC*7E^!iBNpsIKKNTD)Ph&!~*7?T_B=2U} zkT)+@FhUgq>WN^h(WKz_vK_7Qq(*38l}c$1;Cg>S)(3fbpY!n6#>Q2L0*8-hm(?#- zsGNYT%aMxnbzJbt`!ydjdu7=5QD02VvaOl#Oz}Ins^t04s+vKt$?j@sagP4YF@|?EL@@)t(=MO=Tza=gZ`G#L}hX!~K2U66&O`3QPH6f!JGj%7ZgK>}#-v zd(t@HlFg0mu9_JMwE7XbOVZvl7&@)+6i3e9>rGBXG8C4^@l`%CN#$M^53`q{L!v@o zQD=<#SlUtgFRT2N-J$G(dT1h!NY#ggVyDZB1C#jPM_Nz?u+mruD570n<*&nf(}x^K zKeknx#dB9Hs3aMYr4NAX0sLNlYA|Kg&(6xo%L-B7#$~6c*8op7Ko_IvE>v!K^gdWm z;-#Eo;&@^BfjK%~Ak>2R`^E1<@L=tp zyUuS@X2>@xs9^O~dZkk=n44;@@>!OBWC&GwsYBQ`YMJiG5ILP1C>N==tKQK$uBuSH5qdDKi%oKC8s zQh4DN>KvKrdnD$8SY^|hT+s=`VgocC2ewZnd3P_5y8#CskTlH}2*lndLy+Da%Jgn1 zfllC@zPCy$!jZJuRydXc+xCxXr)@7ZSNpQ46ZY4*tptQt-cKf&&aNc|$bMN*@R@8b zw_8Qq-|@Ccvs!40;Lg`DzkaOCx39!%ruxK3X(MdKC}F1R{-Jgst{)ss%dLGsk68Lw zwapL?=GjWwnF&)F^bOa@hIjib=P$XNF^NgJF3Ya8aP*MET{3m2MiJ?QMfYnPSeO(@ z`}-=#XX-pfnhu6SSfUYXji=nRA3<9OW4$g|m)ONyLnsZ>S58;NJAeW_2Gi7=JcUZ{ z3ZU=$^!c3=OGAvo!lN!b_0N)V!0O^ZcrWO$JHB}(40_jgCN6kNPkVgJZy#w>SZ!F+ z6S_wNdA-e2Lv8uA9*H$<%>57+QSMB#ry<#ZQ>4 z>JUum!o7llQzJjlmVa;Rabgr(b+hj)Up`UhzpC%owNugg>EkQnhE0C$^I|bf>zM`Y z(?i;?JtSOa;>r0k2IPMbm)w@D%7AGMpbhE&`7aXwgYD)7?XHD&)=1uL)p7Tl0;BQ7 zOXL-tS?gjX<~pf=UJf0m-2Wg3=@r6#>an2`x&_NHmeN zN|q?lO>CN`dF7co=b-1zy))l8ckX@fPWvmXyQsZu@3q(Zuej@Q^l%EGyQHqE4iFFk z00H<9IQ#^h2gpfD$w*1a$;ikkD99-VAfrEhwKQBbn5vaxdr2%Zo+c}hn1 zw4A(x;`s||>KYd{FBupb8Jn1znOob~+SxleI(gpl^7ird^AEop5qa-^6yjmxqsLE@ zo+hVc=j7()7Zg5!@#=MXMP*fWP3@c3w)T$BuD9>{2L^|RM@GlSr)Os8zRoW!E-j-s zHn+BSb}@U{@AD!6i2gDy@c+L|?5BCrgYzOJCMF^#`#vuMLLcxTq9-OfCP~Vmsz-Li zo$^dRk{l^ac|{&%Lyz&ZYCGb0BcY zMgT(p^dp(StUX@p!sV^c?q?y{A*$?T$HB+sNAcMUmFC?uB1^cENr({oZ$BvgOAAQj z^FBMV9{V71%$qzf3QLKCFgjy*W7N8ftsioq`WW^?_y(hUGdo^Td1bLcErFf;yf8JL zx+-%^;Ki5a1HqhE$Hy!>i7EI4tYdGJb)U-q)|FDrI!z>vIU&*>Kb|}HY=`+>;8`?J zNvUtSuf?V0?7J6VzRGdgv{Ck#0yRIL#nzPq^UXAGz;II=66-q?p%6{i}1*xOu2!E z&`NZiyc=@o5Kzc5u3+wWel$9GJ+<=?U=l00cwdt}aI1=d%O2_jB&17euEaeSk5uWK z>OnxJpStFyUa5wjy672G`c)-)wbQ%_iLq|;hsNfjVUZ;{^^9}xiRd3(Y$cSEf>_en zoO1GKDRZBVXKCk>yTJVVl`&`TgVMg{6O?4nKXMbpCbGfW;N!5&Qeho023TCZDe^Xd(gEd2LHcag zLbmwc?l%#dLnbZ2}O+j%Kt=b*~46a*e689)*#2@imC5 zX6mUi0sn-9U>;7dXDzQ!@c_ADSkTh4$pvGUucKz3 z6(g+V&*(6(Z#Y)!FVkDSb!NQQmdE!L`5|)~7(sl3zx8p$9s=+K;^72;({3$`^v7tT zwtkFttGK62e8qf9TA z0u#Joi9%vI*u_L`7)6|iI>>ngrWR18tSxZHDb&RM+A9S&l)^~eM5jA$U%9Qy*XSr( zPGOO%Uy+BeK-IM;ERFbxsyv%xIeVou=9#N*@Csz<+DN+&-PI*arn{z>`x?H5DhZ})@8D`|Po6urNu`i&;6nChey!Mfw{* zIdGg-9}^KD;>5=L>(}U2efXS>I7RhO1^>Bvm)Rm1#kikyWzZ$HgF*KFrd7F2yw~uE z=yPA93qyh0ELC2NG6j&mBL~;Fp(4g&qt62=|(a4hAe&j6~V#>n@{v zR$YV&#GMO1j2x>9%6cqnGmP>>R6kNU z2!83gG32GC&8eM&-_hfuD>dC!3Gq*~44Q_}V~RQ~dDGVv+S4PbHo}be^y7 znJAPxjldWJ)o-ol?~8@nCOz?yH_{tc-&F<n9jzD9qSXyDOUgyR!hcu)3 z=QkxE2Ba9jI<;%>rUryb(W+)=zRX4l0e^Os3eQyG?#scq`(X#Pa~VdXrjT13cz28-}g`4qpXEihH!NuiHhbd#9mIgw1*@>px5qg8I1UGhH zd3@ZOf^^pwxkaQiR#r0{glXRPU@&`%{F)Bu_;ws2fBa%rtaU8u&;>S1Wgx4<`Iu1M zm&aZ%#sg{tf`j6D^evQg>2o_ogDD%_$KM$F>3V-hlo& z_aK@t94D{$emj<-;$+c*%+Sq|_5WvL0IJSyPi*xNuv$L^GE;Uhy|E%)q(H)5N+M;U z5^op$`h#XWyw^IH4AY_>*0@;?Uq|6%-Jf6W6iIj^Wi0#lMAg8u^Zhb(Psny5$y$l6 ztxKSSS9Lic$@!Vw7 z=c;sl7X2b?d=CNodUt2Xyt8lZ+34CgW^Y~p`Yf*X5Gb+SBR=Wk#zY^(??A8dl8ren z+t=+P>YFG_k}Vf!Z`yI2y?Z9K!(C)?5#=44W!W&F zR&c-cDzlUf@>t8tZdAK#@8kta#V_fqr~-?A86?E1*NIiv;_|B`+vKpMYt_wUNM7{1 z0ge@M&a3FLer3<`Q~vhzur9=km(D#NwjpH7#!Pc=w~qos{p`YEXy?~!q^bF4jx3U6 zrWSb(wLybA^FcZ_`Ez%(-`~+wQ@x?aq)YHmJjf!B7y%=-CRD7g8RDJxZIk0QALiX& z2W1Reb5_AGt$wBLQ|y@LdQRD9Lo_}I#1 z!wsJ2`)~AaX=$GZPOGx%!|MYw+U9GZG=b9EkhXgk^+Q*(U?1LbOxkhRR{C&l!_s^P zHMdlm*1h>o)jsy3E-2*V9f`vXv@v2bOmxUFwya&^Q4_}^yt)1151p2D;aK8Qysy(BDZF9ihuTR#we8 zxUCdml8*^LSB~b1Kb`8&`96_Yaa*069I2vnw>GATCU!k4KUFiLN}yVOc9DusrG7xC zMoc-bwONkwnlGp3lc^_(Z$BjthFpYZGWQ#@ct0>TJMVZalsAI#cwDpLHg&~I&?uz- zgM);m9l;CvCh4GA3T)kEUq()>ndXqvT^aHqgQ=QLqNAtvmL%UF@ncY!lqq)k@$e|LtrvggAd{mpIL~^fFHex6YXI>1bPCj_EAD}H&*Gl zn3*@%k6t_kB5G!u_ST&bfnXA@80Bo0uzzfacq~}8TRAoHkNT@q zU)7BFS*aWXM}`!Su}Bv_1k!h!cBhOyq=NosMEDQJq1V+q&bN3pZ@vq+^s_Uxa7%N;)7AgRNr@0g&B~bd&Qx{+6w45W{P3 z7s?0w`4~QT*7I>a#yQFi&+6+~pF;@cN*v*2qOVOX@UE|=@Dbc0fR!xbsZ=sM1o)oR zfa-bk5LoOe_WyFrQ~3=-D4+L8vtP|ULi|1OZqiinZaw(o&BhYsqjK0Do6IB@8+{0z zXWn+P*ByVVr^cN8C+?|U&G4PNlQDAskDoDwK782yVie1GmbEZHr}%zp3_F{x*%eiu z?97J2TjisHD=S^1Io}2@(m}gJMEcC0@Eq@}&aWtH+e?M!5i}r^f z1jE0@8gakPc;g$vq-Vi3{Wr8IS#2N)iwWwx_@+;XfLfe`uKY_G+89h34B(5E5807K5~1$V#$i5Y0mNGA%Hw~2xK+*mAG)n z`QgCF7%}{r05Wg&gfz`v=xbW}441Z&HW^SM1iNDdolm4Y?&F0WnSz?WJ~ifq?;7UU z&;!B7WHfBfV+6K>haa#F#ar-*fNkDi1>5W$0$bL`%C40P=^%9s=#I4dtu_2Scz3uq zcz0pbmZX*O$UAc=E{4PHAV>`yF#=opy3pk^sLVd}{;bjF@Vvh1o2?w==^cG}DRvue zqwmfI=`+(T5WM0%QE-BxCLiJC~)C7iU02c8jnQ*J!1hs?m(Fr^!C#`@t0=lqep;e9t( z3K?x+R5xoHzDHDHv^N;s7WY>@>9$3a7y}P zKM%b;@(aasMb^Awz*Z@~wXLcOMqx9Y9Murl*8(@zH}Tmf}ihNnA^Y^~Fpn_`TW3GOAk@=A31v-C{Bk_Ah?m0tGoR;p`hr0DlL1c;V7 zN-z(#7972`=Os4Y#VoM5yj^)VavtSlm;$Gsu=xD$vax6I>NC^}^@}~33!BAlC`jy` zfPjV*>O-sWSEWza#JnpZWU!OGv!ygQoFIaa?b3-*-}2YiW~{dM?^n||2iM*HRpG? zWT_@8tRR%W20CH;rhHzjA95lCZC-2uM%#k5NOIq`LqLDT4ux;B=FTr*69}x2xaFZP z<=3e|Mw)2uAYT#Wb7ov9JM#~mApD_G_7Lc*Dgc*b0pB&WR55SGaR|&=41jJe2x>M? zg=?4bjBOwXjV}DH$cFO36cUG!Yud(7V#O4<5S#F@rVXJ9u*-e;V)aVDFuUs=jq6}+ zKe%BQaai7vTgXiOzswO+*zEz z?#OlhH5Wb0leRp0XoL0=FUcp97s}8Tw9!SF`}1%LB!qKokAsSxZ26w5i1GLjegBTSWhD=eTv=rz3fClCSr1Ve!z ztuAAz@E5c>@eQ4uvW%ns=?fZaOvdUhN6?e(eelmPbj$ZOUPIIq-?VTD^u?NP*N)By zfP=jZqTwH0Ad&+JUY$k`-aNhYnRdU zi{%*$p-G587vwmJLil>(BMv~L6qp4qpCh%=9$dMGsXB!MWDXMe_DTZy2zu(!*Ao!Awa&nEjx4w z+*pFT?)&T*#gFg0ZQ{Gr?bff2b~_!g_IHVvjLou)^!mbPN%P!`St7?5{4@~{CFbaS zS36E=vrH@Lq0_cqWTeN=)}>s_B_wnq%P_DkdhSqbSH_W_W9@dQXg6?>7}>kcw0h>! z%l_*Ujyt!_JFZ-63-t)nDN(uMkp#+n7--1+n6vpoe_N(n+M>l1Fr<+^i@b=G!+^v)9=m=3ufj*iH5@N{vT(s0 zM2XzYztsr>lQF%Y7VxHN(ZQy=;=$!IqzM~u=z4g@n`$c(oKASF|M4Zow7VPR?DbXm z*jp4@h~xH<*G4LyD#3EDI+;km=f55EIMzdRbqkv^BNFI2rR+oyP-JFiVMTk*Rg}0S zL$~D$Jk9*mO@W3rzmdNk-#Ek3G9KG+!=<*5Xl;TDLRUvA#c!Ce zpMNsYcw9AmOg26eC++EzHyS^2b&XG@bkrbb`>pI^BRg6k32`T)>GkKFGyJjmEtf@j zNBzolX;p z!(r#f|D-OcFseSDu$Vc`Qo)4ajs@De`1;0*wZ}AONvvKVRGeh`)tqMj+4R^N-xQC9 zZ-_2|QDD#@ekE(@BM7IM0q?)%`DeNkRD)-EmVWXO{F+*<_L$t@D+f(mSG|^)W2zMS zKqH)4;QQTwcF^3>q<^Sptj=b8AB5mpS$Z~)RSYY6xI)4zI#`IT7*%4bI4CU z=cxRlgt66MzhX|HsC;N8X5nl}Ac!q7~mV7AhS ztvcoX-*R5R3gxYexnZr^TqI*q$NUh)$OmtJ259g%ga!ku4nFOB2Y!$Pnw5{v{f)Bl zu7MkefFqd_)53V0fq2HQA;olsF_y85B%POu9HB(kgf8^rSE>_ z?@`_#v>j3#DTNsU2{VxaDzl7zJ|hhedsO-{(3a|y|1(1jyhgB=K?0xCYXJSV9ng$& zAIUXI#W8~}+!6oQ^u1^H8E}C5x<7S7@GkHinq)V!UP^pxlP#kMEFti%izHW=MIldt zVZeqN@r{kIea? z?oB+)E~2T)9t?QGuv}%xkMH-;B)3d~K`&hy*7;X%>5px{^ zr}lOlGN$5j;a+g;Ds)3u>QehT?bkyN*@&*uqXK}MjJ4X38HJBL09k0Fez-w+)&v)O z8oE6P+mb`dmgJQ6!0v+$vd{gA)Dbo&qgZkGQ{)Z;;^&>`ov`-JRvX71Cb!pM$%W}1 zk9HLmg|c6rKI;CcSfK53apnsL@SZ~;hofx8x;gB*@}mF^SVVmbtOJ>MLEo32Nj`MPx=_c(6BR=VJeZkEoLkALvphWkbrGTOlV^+25XgWp8(gKM@a z8zem?zDF-I?G+|D4{VJC~oOj@1 zunihe8?ThZyGCM6UgPKwfd{V0L*QBtUOhj66>@*uTqwIt?r19|xgTH$dFdy0?(ogC zX9jFg_#pdt1q}T9{U>$3mvR)BoKBSGDR;+#D;j%rrjRE%v!BA)qd!5DSJ{VM@R?7K zHY5Biv8b>=1|1vqdqJSMJ^)`?7C)di8LCJ-?{ou%7Q4`2nclQf0u(863s4r>o3@B7 zjG|{Ddkou6STqJ45_wl;Xqg}az?=~Ju#IH8qPi*tAjYX3Y{D?g!jW|$_5{|)F)?jl(Q%1$7C7y%X z9cG(fMe~Pq0IUVNN4gB!z~itv=MLd9JUJ+*a$wG}v=5xMfpYY{HvbYPJ=4_RX8YX^ z{(nAhRBXit?@B5Gg)wFlOHlHn9eocMtBR5c`|z6FDZ~ z!~duTw8jX#Q!Ba_3#7;hI-qMpdPx6*VJW8|Rp#)XLFg_@Z)Zhk4;zlhQ?`9hM})5q zXKZeh9pIS$jxkd#7OH%41dShE_+*R5rxMu0GGrekMAWYGlmD^>j~URAi~44MlD_#B z@c24BcBNbRABDw4l)#r)zML8+)lsGvF=12mZ~tPZ{pnn~M*m53nHT>tfmnevZmGXf zBK5#Hbdm_}~6wcX!M1 zcV@}?ZIK9+naO^m`|o@O===oO=J4M*t=Njy(!43WBG2NfS?r@`U@CFr7s*QorRE=y zywE`C6Zt@UKF#CV2eBQdQ$b&m-oFpv7hg43_k=OCe|ukkmYvvfwM&rsAW6x`2`?W%iA*$n1{{^Hi1{0H5-9JOva_`6?5{s}XOvtd3e(p9B}Bpg+Slhob@F_w z#>ED~NT6PHdY;XN>N4L%AM6dNY{`=%!Mrs=-XSo>51+PR(q)l61g05KlLvrI z6OK3@Cj^#ZYqi3+fc`@O#Q;NxK{Eu)-5+ z;F-(RSgH7+czSWrr|u=vIs}?QZ}m+)=0ZKFrO(oc9|8oRDNQ+=9UYMn|MMWgqTnC< zzNM%3rVDg8e;7y3&%IX)Wl!5!gO3rIZo&eNa0~?$qxW;)7EF4#U;Q-0lO}H(d35Ts z9K)9!p%~)^rqVO)R!7Mif0U%=<+`LMN688fB$fyt;_rqhEWj9-FcCZkAKDQL&Qlqj zEP=r<6tq4I5{n0ysuU3)O81c2k7=<#=MP1Jz69f37-uec|dAX(y`sZykS3YkYCF&rGL`2+{XX-}Lui zv61lKeS&uLf58UL{~djHCs~w7+DGNoiQIku$@v&Gl!!F>5TGE*z-@wAiDpeWCS?%R z>Y# z$a>E7;D|wsH>%{45Ro{ZfeNdCPFIXXzN!^6G2$s<- zq%GhHK@+&G@82W&H%M)UnVjQ~`;j0llOL%=1sMNo_$Ml0NosZq}~2%sYBu z$c6AN&Jhw*Z!{fUIM9amc<+3yGlLGmlLQ-V$;%KdJu(X5{1l}APWZhh@qKO+CdandoecTwe$-L!ts zg9Vr6uyzs7$`PL{yIl!itB>-#CeXCmh8~razo*mH=q)7;xk$v|n|J8rrTgigZUUOh~h3_PP&EaY6=gg$~@Z_s`%%#WXTD+oDr?|`Tn zEb%C8!qfs=PQp}gHZYME3i00I=iagRfkr{c2i(OH8Jv5T;d z>LM4N8NnvH#+QikcSC!z*G%T2D3QeNnGj_$#FPw#V>!WjCDS)$h_}wOQsKdMNoDEL zg)-(|-|_kRo;R$!rxTT=J6;&|PQ^df1$m*zmuGo?_YI4jTw?veMYi%d51kLr=|lBy zFP6;*7V(VbT2a>gnjW*KBuCh%AfCe^%9Myr;g8lcWRWwhK;4jPk-q#Y&s^uq-W-g{{<~taR$r5;tf6p^C*h}zhHubl&H735z z+md5I`^%Wf2k&--0e=_fc>~{~ zj$qaVel%$Y^H3reyES0FXc7{1A9=RF!b02m3Ih918`UMhO;*iU#zf$amE`C+~oQVnQ?pW%Lbi^YMHO z=d!vUE5Uzp=j(=ix8JPH@VL# zy3UOe3)jSN@oX==^iAwYKlM;^i^}xYH)cff!w7*UmTsM1_g{FaSVbDm8oBKtQg7nI zY4PE9m1KzIdM7dMEHIK=M5_BvmDq-UB7_C8NrtwH@q#iqnFmMFl)rTrrj{x^cK*7@ z2WjKrpQ)cTy)ZP_&*$R7kJ$HgGvs6I;iljg*8k({eT46#PsHa{48h9EsZ@gs?VV3k zovO+tdd|I13C7l0&laSm>P^0FGsD6`#!mq8fE$h5bjndj(>yI}>K+PYHdYQL%Mak1 z8u0_;{U%A6q8a6b$~+lGhPIu`T>gPtSCXC^t1Z~gvpiX9x_x3_n|i;wE_9~KQ1Gp? zvGWO5uLc+lW@0jOM_U6F%`el{pBgp(`Z@ulOO^@N0PWN`POqNQ+JEJZBo8mDx-J}9 zF$8V^ii1@yJ+LqxtHL%cH-Q=XP@9ZWAQN&qyM9+4U5d0uc4lGnhmNP{6MJR?4F2|9 zbU|wVRF#8vTkPP*u=)Zmv`O6d>}{-3LoWK>=1^x$78w0rf$cAgiJ-kpO`o?$rz zXtzU|Kqt+Mm{l^bB3n+F@f1;fj!k?=h_WDJ)7gs0*`GFtJ3jgjngjNwCe|H;e$cwq z-@v25;PBA&nUX7!@X0RT-AppQs~XXPx3OORONw01UN0o~P;w&IP(eBy&l{n>iMnXP z48PNvH^rHZ^_~@O^Il|heC}JM=>8-2Pv4l?d#ejH!U^x3v>O;O?L!O|A zFa=sZIg`~m-u3L*+>JBvNe$T*=C6vab5 z^9PjJ3mj{=C(3HIqLZJ!Vvk!FR=ypO0ly-gh|wnc=vC|_a8Z+8H^{^E5RkOYILl*= zn#F5C>A_UyBvxYN+AM??7PWO3d`-o5+ozf}Nw=ry&jYds;5yEzq3Yh7H+`e*3y&{5PZB+Z7tC5rKq^w+zfrVd1&WKv`FV1bLiaf4 z0%fP`KwfmpX`A3DetxW^d%;asmogy&+IF3nFvy~1j=K|ACHaKp`4c19 z0H2aUOhu;5KF9chf&3=5b9~on=iC}pdW6plnoB{aS!>Az?oY=*x4e6xcD12z>p=Zu ziCFRULl(AiUH7f6OtK|xY<`hllD9!_FHFN~E(lU#juge%@;~7b7`|isprPJ2*>kz1 z3U(is=l}Q*dxQNt+l%(}-?Uziy3Z=bdizX72iZLM4uNt}F>X$^in7pi#SU_GDw{qh z(4|x}U+W5Xw6QU{u3ZzIVXy?GdRc88_RYjH3k_Fa(Q}@=x+I8=fWjp3RUNFNv4tC} z>xebg=2n!(!IrdTnWfQ;AV*XC$J4`*)NdUUbKc`x8T}#3tVVq@tFA(`gD(1Bu~mX= zlo>ZqVGE5YGf9X5+QEue-_8!Cz1)sqWd5x2({BQ#>%y>oxs6O}akkp!NrtWHGtl$^ zk7!|W)PqlP!C?)IRs*!Jv#u?@=Z!7U&g$yo8a@$*<7$LPAjM6!Uwo#1e@tO4{ZcsY zqgOdWa;Bq(datO=NP?|ImVWbfIsPr2Bi8O-{^G~mclIxY_XKUYtidyPN+@%RmUql~ zG6Yt)vgH#q*E(Mk)nrnyb68oJVzl&}pXl;OWUIm2K1cxIKK~-<^tIw)JcHCFaATy`h>|fEq#-!-YM&^=OB|iJOx8d1eST!lAqpyrP>ME5k`6;z$(b( zvuEl8;8Ei;RrrJ}oOGTcon9bunq>I9;XWxy(ZU zQN*`(gMz56Q}3J8!j4P)Sn$UkVK^V#MtxQl^s+8V5Ub@ty{d!jot4nz;rH|DHxGu*&piOdGuz8!I)RlOhxHqNl|B)5Q8$n&8;WT z|J93Z5+lkN5&##0rIB~52Bwc|6IXUMp(0)<`Sc;6xa|al?R3+ed*`h62twVZ_Xx|e z=@u#dzF-C05w>Q=ck`VxWo18{KK@;U!WV0GpX)C&AL`t&v_LI#YjliHya?Gl5e(Vy zNx)J9+w3-`H~p?o(`!lYFwfMEM7!WDO4~`ZyhP(77#EyQ+6DUX`oQu6tc7e(H8b-O z)3-2ZsD$TUJ$pUr@@t2#j!6NHuboVQgZ`4fqeLwGv@<`a!8;`gGVs8!xQ7vQ4W}BA z_gUcGIg$KLHmL>QE_vu<_X(ER`ZH+t_45@8v5&GE&w0j(p)qlfm&*xEbO%agE#%4@ zLj^;zIdGV@&g<9VI&v2ZqYo@ba$=? z6V17^AAAQxw_6#pCg)l2-XZroJ;>%6v29uFq?&yvebP?KxGCWc*g^6rlNVjxDDj zBGSUrZE~x57=Qe>3Qpp+`2OhZc8OvK+BRp>*TA4EhGCV6nWzS=yS@a>;Tf0QW2IJ1 zAGy_}8?br4rBuY*;2$qmv;>}Lh;yUXavhPYw$|jMZz~+oIh{p?VsTbqZwO7G zTx7qJ%twz{rKZV1=sLP;PNZiM`98LTPHv#;0y?f8>g16wpXC|uv2k)(>{&pxU|yfM~3Cv`NwNSN1A5M zRLIbIk$KE<_m~^reLjDS=uE_@cjz9Q{G(HigOX`mDJ-! zPXK1NDMZN07D=54+^pUWtXb`sN#a6FO#oM^Y9I4w zLpFIoVIEu&_HKRS&ywsTFqEH4cxgd_&wFxLb-yXX4zsR0unnP}_^j#1@%dAmZQS)t z3BG6cy}e|H#idz~!%wmkbA_FBj>T99$u5M?nmsM6dYr8fua!89lfL$o9%BsATolT^ zQFC#dEfaDQWysf-t}?itS~ee)7JY&90foF(7v;A91~Rf#W|GBO@iHhyAJZ~-g*stb zAN@-J&k5Q7<3l*n0M1OQ^Xw^tou}SOW4DUok*w>57`0V9Z^3c(`*gR@B_#CBIA^2P zzWKu=WjUkR0A5@6wTsQr_M>JfsTJO1}w+pP1C|RvK|7e$g|*vk9ewh?A&D6NCu7e9O72Lb|oiZjT3)t8X* z_i^Vg*TxhCx;Noo@2cR#t2Th;6it~7s$_nJhr>Oh_>BS#VpE6EnZq(P--%{8>gh*Bkm=NH07cKQM3os`X{2t~0pKbqkzPmUf>>QH)qcKR6m z+dKs$;~}R-@gj=`SNKxKcgWv%KJChd6q3Q7I9I){uBQ)LW zT6Z9q-s#C9aC?F0I#+0V8&722DIuZ{t2n=zYsV#>i!rV&%UjjjNm;`yMc*r z_aBK8tn4p!IRwgzOP^7HU~^+%HVsY4P_^tclK zQb?m+G9Bqg%{L0Ay8Y{Ta);CG7oW(PyC#sNeHpEL!aOQCR^<_w%&s(3E=1H_Pn177 zhCy1UEWev4&$IqoUnW(An?qfkZ(u1t#E=&%uD&w?O5>38&g~*Q`c0 zt2UN<3#}pM$qn=D5uCQ2K_g7`|j!-eDE3z1VGbFSB{hx{Qb|w;B^wI`L+B z+$~})r`A1WwdM`Hx zA_cu;-3I{}^+>v}U65O=j>yW?{(jM0iZAXWQZ^9P4fOhoQ{VQ3SP9`k3yDoUYB}SyGZhxArLAqv#D2+ooFi$3!BB8|B zG#G#v(bSm)iz=6v14T=g{Rid5pF3<`8@vu8l~70|JI7)3svJ&fv%qYI%$kaE?Cv=H`PPE{Ay*uH`Pi< zZ29t@ZaVpin1oM6A;P_je|+z-f)xiwHBiSUo~YPlXHQaLv30)y=bT!OPD0ef=w1&8 z!j#8iT+BvzU-A?a&{(Fa0lt$__3kChJ8x}!f3R(3qp9dnDuGI{aLrj#B63|>)u8!ot$cf`7y zsjr2=*1i$ry_T4C$q1(6shsqjc6sL-g)SV)ww6;GM9WwrsT;ugpACAXxVvSF_YH5g@?_JdNH4U z%zVvzQ&6~<=bg&qfuALaJ|P_3B{;8G5V74ux$?@{l2u1#LjUrUOq&otJ~#KJI5n_f zkV@EO>?@SwOHe_+oQm|;82F}D@MX*BFO;0|P~V365myFS+2)Q2vpm^ah(GoAx@Bv{ zrPcA%TOIPH8u}=$k2BR?2d58i73fnx|LV%XbI_^RQrZ~8w{Z;2h-)4KfrweWfaB{4 zU$dKKuO9j}_TIChXSsEKGeo!?Or6)u-EzU5$GKVh z2w)S5xq>qQ3+}Hi9ALkv*GuQ;*(+!SX3TQSQyj|XPfWxltS^~>J6lCx4dN`Of`;X4 zuh!-v62E?AKS6XbMXPexK-P^8lhol|aB1KU`zr6#P?1;m35%v7z=87+Sf>Z4c5s~j zLQ11-dZyG|1KPHL+}kPnX{*oo-7K_J9V)bq8HVv~JW#l}eoPoTjCA8n4Eot^)fK}T zOmyj3X1G48r1dJ2B+-KuN13r;nFvZuV@{FNgdCbCJ@$D*uEiNr(`%-Ti+)^*pE8kF z1K&iq-jj=z7}!W;8bKEtuKU)tlwqEJ(qZly6BqsT^{L=D3J?7rAPNmbjot_dY zcg`Y@m-a4l@C8V?Tl^34Av`*Eg_zN7H!!zv2*VOM)>ROUA?r2TJPfISOW1a5YU3w?T zr@8i^BQI-Qb|?9^2a1=!a3zdP$s3mD9}DuWT%#PCrN(K8A^H+r=Ce?J p=R%Z5BqZ{V0n>*PpEG`m%!?3P=%uB?Ki#hNeI@?49FW7Y{{yEqUVi`p literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-17-Quarterly Growth Trend.jpg b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-17-Quarterly Growth Trend.jpg new file mode 100644 index 0000000000000000000000000000000000000000..2842a5e82a828a66008430a26ebdf3f203201b65 GIT binary patch literal 20182 zcmeHv2|QKp+V^4`B4fxrl(B@&88^z5WGM4gNl3^%+ahFV2qDBaMwya%F0&-_oM|I7 zdmA^ly}#9S-t)aro#&kAyx;VD&-XjqZ(%KK-Rr*Q|8?Eh4L^pT1!%6TX{rGP1OPw) zegXJN;3{y6goKoY_!KEADH+)*ateA%iqoeln9tBr)3dX1aT#lhEniB|Yy#FZLuh^AxvA=_dyLeiV=RJ=b6|^0SOg%q+Zo`~nv)N=Qmc zUy+f$dQDYLT|@Kwt=k5MM#d(l*7t25*xK1YbaVIc^z!!c4S5!TR%x%QKv<&l`zQF#E&1mxXCsult}q`K6pl!yvXg4Z;Q`%Os)R1D@0YFvn|5WV z-&keU<64k-AiP8*Zd#*6NN@6CkC^aG2T`mA=3Ee0Ff|vP#o@8GLTFY^3=QT$4K?8ZL&^F8n9`Hpb_v)kBgy1lWmrYgPelxxb&hxV@c`Y;FXo-A z%kR+BlbUy`y2VZsvd=`AEwZ}2-Wh~{k4IWP&8a`WJp58r?&_=8e6ZQzzMragxL~F`P=K1GWo~>4Ei!gv@^|f};@yOsv%ISi$v&TMMQ6u!j#jP8T7<(Tcq@m_%@8er0H~O$#7-$Z>tU5> ztF_zNis*hr9iE~--YPT2IbLbK`ADXt>iS3n8jjJK`<}qls`*aaE|@8`*fS=%A=G*m zL5&rhJvJQPlk4uTz6-6XhEnr@UM!im_Of8Pp|L*-UF&gSk z>d~FN+|YLq@IVWK)ZF|Vf?;hkw=S)|+dV127hA4~OGhO5CU~buStEN{?MElJF>qvW z*8>`})-U-Q%pWI|^!jW`>Y!m90`=%Rtvhe@7Jv^m^K`d9d8LauAQ)pM|}$%0=z z{ipsY+FSqov<5G8^fiHxY>(9_HB|wh_NrMZ0S0#Chs_4tA#a9kP}lU@=%s5v4rD}W zqSH%xhZ_p)k60ARE}1;Tt;P>G=qOv3&O?b*=jm*htTZ20PF8EQI8DXe>bSm*2WXaA zI$BsK%{IghV>iCvBY-$_rxn+5!3RXN_r&7LCa>0EGN_VJ%O>NS8bq;5N;1t<<5=Q5 zRv5QyF|(~kEA~Jg4Nvoez_j`M&Z$whSol^^8zQ=%j;!w4Gv`d=1Ra5od!%FVxlP5h z7N+A8jE$EP+y%%IvMws=%E&uSP&ch;Duy_`t(l#nY0LE&su-#;(TFB?H$Pb-cxJ$n zSK;bQk^nm;(mi0L88G(^2Ll*^_^O*qO zTLmE)uX)yufD-qHySE=osekv-t!sbmO3F++I8ygHtk5p4PfwLTj9yO__~XO4@lxrq zhX~b_solW^XD_by?>shsLFy_2c7%y<(Wje_RJ+a0SM3HbwD1TFI4UlxK~0;oybG?O zspfK5X}_+D^}jIqxHqA6<)B!w89`P)=02R-YPjLz$InN5;WW+e=h*nI*GAMxN(Fg3 zGzU@tqQ^DwMOC4=%-T$aj$KLrx66@-ITf}|mX?$DBXJhfH(w|-b$2cSJ@~7W?8=Fuj zT1LO?W(gV~%z2GKuB6-Zge@F#G#MXtz{|A>?U2Kgr#2Q|INiSZMPsiy1m5P%sG8v~ zKUkryS~aElF!)NuVQxUvLcDp>lX`J4WRDHj3`ui8vT>Q)r6sm+{y6*nmWqY(w_5Iz z*>t&7@19|m6f+b{o^&3|RTcHFfvzBr@XQ3MYjTmh&S`XKowmxplT zDrdPJ(@po;Yw;QRW*gyNez9SZN1?l#lwGvi@5q`J<-8d!Z{d&~A2`Zp%B z-uoLN>qN*_U=Fn$_lrGS2rF5xo&bBfsNx1Uq_)6Aar-Fi(}-{3Nj@)Neit{_-%N$}>a$ z^^WE2dSYU3u#850Xq1)|una7uB*L}7(LeHDm{4p&Gqw|I8@3@SMerH#-$;Jec0UH-LVR?S$d%aSmsc;ce2V!4^5t%2uRpmp#dCNsFu?c`YqW43hIyH(x4JNYSCvkcatoP6&V5oiGl> z*D`rkq_!2(GTZ)co_%EOwbsIiUa!-o#qR)uGkn3UPsb%R&mYMKLru>I3%ZX}XoBF% zgP!2XU`J?OU2tNNn%e%V3TiJgwO)E0N!~M-o`btTn&jQm%GfA7I*+Of?{biY&0MGM zKF4m1*rXo9F|5D5o#&`FJRIo=Y4Nn==u6TnXJb%&l;b}j%2rw;fU0b4qV$Ne(jHmg zbm1n6%F^|6{CcX3jmCGh`nH?dr^i4F(*h*6LAhf|TXA-NY_sPc|-y#soLjX^DeW!JM6Oe%#cmFO}VF*W_xav@O-`lzs#+zsybgIz1 zi;FNYIfFc144s?wv`}n8JAg}3wOSsrpW5&6wVrG)?P^=&c}0PdDprrA9{m;$x9A}1 z&6pMVEFP%T{@!r%w&>bXgzXH^1B=uXlE;q~LVbClEMI7OE@NB;=9L)qoZiz@eSB0T zV`|Fm#dO8w^CaSzM>TbkkevCUkSqTl4ePJKt((!n{S654vxo?M8}f5Ny_U2)90JT!X;3 zx5fpw&d?Z1vnc0VQAJpV+;-3AdH^oQZRJV|!p9<07N@fL-ZREitG?`Q713$73GiV( zHI>K>^S50Pb`6FVBM$b;=bsK(s1QtHe3jnWN3lc4@gwyG&C?PwpG?o@WyByX0D&R?h8)N|>-?2*C$2R`#;qMGFs~-e$_P zTQ~bQw~CQ=QS0#jY>G1(94=e&mQi{_yTa0ynX2^q0s>iZ;ExW$cr=L7J#^04IoQ9k zJiNq1qVa0Z#>#Cpx_ZmPhMxbfq}x)-GY-0PR6=8{1BBT&PTKxEw0)eN^p4=zn=l7N zI3Mfd8zs4+O0w~bqX>UzJP z!ZyT1%kQ|FEBU^c_u?HE)r-JK1z*}Hx>(_%3=|#++v~Y^80o2U`;=TA5NNO`4`1Ia z0lZawY=@oKM}Ukj@th(Sj)fOg(2G0yMY)&=ojKE)%-%1YkyC29erwq$2j=eM*@{y$V9#A&JC+GSQVX9 z9TJHt!mmTmGDTl8=#obn#}9okveR^#b%tMW$qa=xPg>5(riZQq3$1R4R973@MoT%!0os(|r<82t`oqwcdKgXe}Hz!tDgbok5 zw&1V`+{KvR-_I7QkoAuTkzw+?=C{Z&3;X@Z&^Z@x#Bn0$!2%vY-*&yc|Lowi;72_0 z(`8sQ5)D1L1lFUfH&`LoeSiVwCO$)uT5D}oI0FA3H(9CH|7u@EtZ#_dmg1fA+VUp zr*zhQL+%w-$8?Y{>#k8aIFYpxJh04~htv7-YLLSvW_TdO32_L;0}KY8ogh`@f-N9b z9&4~&%0WB;G9H-^PHnUZ5A;dvU=g4ovSY)`)vNU^Ie&Rx9V^`6jC)cG16hPM_>Ira za8SK57QQnl5O+KPas~4a`t#U7n6yd%r*l*1rnZZDoK+9SFgG#C0z@uG0(k+K1RdPI`onTD zRjs3X_Oc!^@i}26W?|`~-#PWAohB4S<1JueSOwd;G!7xf9&gM0FQV@|tvgu4MRJ2D zzU@$-!C1Gl=tRNhE~2JtEs~)lA6xIU-9g8`2&n1dU`BO3lXPZ5FnGsJ39*PC=hH{V zM^cGBCT)@_(aAzwt}$qa3%vciF7>Zu{WCvzwjvSRJRNwzDi%~jAxkpL19Uv@+ABP; zwhcWYF8Cy9{OIm2kntBg|1j6vM+8H+$%bKTwXkKrdeery*%)lFht<(0;)v{RdRq%p z;U-+5P#c_Y2I23I>L&QiwoXaY%yT@TCghpWhBd?kaRC81tksdIeTR$I1OF{mdc&p5 ze+MlQtRWyxBWsAwPk5kg^u19GmI@D;6~pRCalMwGjC&d`l1=`i(r z%3lfJpI9}lHnh1+%ys9ve`qh<|IE^RR>NQV%t4 zJp-+-7PPwGpQ0uN7L+%bH?2^$<(x;Pt5yfCFsudVPiNka^sZd{kq06e3sdS7$ezeC z5{eTBP40JR^-^#d8b39^B&=`|DelQp(g_;$3XGs9`2DFRt18sc?o4!^fiutgn=E)5XZTGp*=I>g=ppjM-wFqsnT1o&&a|Diq+@BvGGNM zD>W;S@Rk8gMZ&qquAqzRtJkvOs$M@P3-bZqcqS}{{Av_&du8a7f>N`yp6zGuocSu ztK8)3e;+kjn_o5bp#ySo0d)HPpIeeSLjMs)`tNASAB0Oh#x`@bELbMI2@ZK2HbTRX z>!_fzqgg?ho66-i@qo%n9*&~v*n)F{g9i^}27TJT46*%~W1lv*@AcdbrdJL%sDk!2^n5sKn5e!#@NDDLtY}Y;9!>>N zh5qx&emp>))P~sE01@Mp(^e{a>T=W%V$#7#tB)ze1%S~NbBjZZ}@S0|*Bg_LM2~cSAZXUl&-qyWYPugS}{V7S>zQgd-B-{5^dYyS(&n z@AvJtX8??C8wa)De!Nyi^|3}I#q`t3DzB`{z&+2VUwQi0>F8xbLgdekRo@SffG%5= z)aG7qPIUDLk* z!(#ax!3#i6A75=n?21glX1nn~#b~1U^SNGlMG1}Ns+P4QQ+2$zvO6nhH0gH9;~#Q^ z`yu|sS`5B+?c4pUiRBl|#o*{LrM&9LLNyc0VA>d3l41A5z||OkwwvsqSU0l4K>W(; zXZwr!$)A~(>yrG*5cCA0Cz(i{BgPFV$_5WiEdOMd9(AgK+bB!ZGe^hV>LjId+Sbsq zD0O)P5ISF40|8U$FzB?SV$FB)OenelM9W-i)IcOa|V~ z;GTdn=7}~QX#Wkt{Pf^Fy@4w?)UMa|*CBsE7|@SqT!+oxf}v~C@ZYe>`VSNj`o-ll z2n_uU)(He2AZXhq)4`HEOvD{%De41q_YqE=@lS|DS#aj<>82>@c?TH({I=${Z~5Iu3p1r ziae8)7+jaIOP?EhPw9R84n?|uEegKTNQVbH@cq#@b5zSIwFBIb)9^1y6)A`adF|8L_CaCiBwP(cF z{D}!^{%Xf>xf(~Ok&=jO718HE$`#>|+43*sOA&7z9{h7@;y(k!{%_oqThP8^SA|x~ zP1{&7&>M=&BW#3$7Hc&xas!Lq5{YL^r&0lb3!NJfk>-MAn;a41dD6IFl( zot+X4n;Z`28YS5b2}5EpZbt@FlO+BI(EfDtf6)?=pNaYbH(w)Ru9np@#7!@7PCneWPNYP!mVN4Cf>(}kP!4!UMbl9HT2IJzM>VBvl zjapPk$YTAFZ8CQ6Sg-1IT{lxN>rR?Y92QZCs(u%ZQ-L$OdGQY&_xB9R_;N7AyymUKWz$OQhtgNcd~^pT zN;OFJc3i0!EzA6v(J{f#%4+23p0^y?_hX@V^TeGTybYM5m)LOSiy!WHc-C^5phjE0CPbdyN45{`}<* zB1YHcM6eu_3$7U$z*`Ik=Il2g8E;0A7#AxOQDM0A8WeO6Pi)5Uv98eARq2PQRSb6c|n^=J~(>IlUxm5Zk_A;?Q9f$8oSb}>S! z0j)bEr}CICX+i)BbXWe~iRpZYY&{2C|Kj$p0|UqXjphX95NK?0h)nS;6VxyE3Tb56z6 zx#BaDtx;1oT(3_A2g*0p1`Xeb9ktt}OdEW@f(PVTk{MXf=4Zj8_Q)0~#K8xH1Sb$kpuQ(A2kg=?E5p*jR(-dfFlr|TO0v{DUX z18tA5)6(EpcMel_3$(f|RK}eiZJw`wvQw$?O6;}ECx(x(=0wyI7Ps1Tun)zZuUZ-q z#){0EtjkMTnmzX(Ygw9mwXiS`5c*u1ABZnIHN{~uFU7%ysbENJ>a8>`+E%D>?&jYv z@h%a;*v~q2vsR%=DkH1nAPa5$Gv<4qyZnP|mhpa%oG-npWTZ1Ygd6od4O&B}C<8cp z&D~SQd5KsRw9VJD{F0~s3biISFyXm*s5c&X%&81k&-T`?>Y{Q2&tQz+nN461KB%yK zw_$7I=7N17amYgT2JOPgzxB91aN@6id8ga;^Ox*uCJ1wJoSk)b=v*?xeC+kw_nlmO zB+qx1y8}!X(-~s4pu~F+f$+^65BQm9urde7)kT~kYSwdP2TR?Ij*Ckt!(X8+nmEd3 z*Qv5LzW~ZQsd>T=FBG0|AAYYZ61h>{U%ve4(mh(U*30jp$qGGE!@(D3^cU&%bdCQ` zh4CM*)Bfi(w!PVA<~O(NtM0y9cv|DZ8AM7LBq31Pw0>IWa04t#Jc)yWQipnJLw@4>Z(-~v6Uv6NuR0iONe`}wPxX42EuO;EEgkQiX_-@fx-D|RfGYZ=>U zo;7Rk{1oKy_SF0kT@PUYpYiuU`l+HiYIR8R-}5s@0{9 zp@^;R(#8AhmHkH&&wSo~$9OLL;$`TGJ3*hGF^+F^u*Fwqh+`^GcQuruGTrZc(TZ%) zuzRZ6USz8<_ZcF7UbihJM!P~vuXbON@SJUDCoOYp#T!~RK?mzlI+G)>))K0R(t@OE zl4YXJsZS?zPL#Oi?rYGxPn5dB@W#uReKrHO_j3 zF-*6tjRo?3*l4_faq>}-v-n7Myxr1)lhw?i8i9w6tWNKb3kyBa%yda`e-xr$cJZ9o zop;$h@j=9Z5JUTn|4r0lLW|0WUf0{R<+O`1F*O+ZNMF5KHs;yO_cbTmk+1Gs6q6Rq z;pho!RCPxxM6>M61LMNLCSdl4tYCpnNFEX^vKl zpIf+u)ET#2bG#)b~8u zX5sslvC;JAZ~9;~8k`XsR3G33MEBX?OMAV{b~uiDEejPBL&M)>DW<-gK@|+2ht1h} z+v5Db!Dd2CSwy1BEduety5X76gO5bQBJ)2UWKhB3-?2u>G3=#B*ly|SV=b|$Vn!#? zjV%k7v`$v0&K^(_-4DSQXn$z-w~F$rGdf~yvSx2K^7>vsIT}h4iIHIUTy!kj`BN68tSi0f4imoGaPF{hA)>AwH87L5o6%(7~VY}?SwIPTK^iLOP5hU%AB z2wcZj;45V0WITWjPak2wd0-}5q$QD&@>Gh$n_FR^xO4+^f?(tWef8E^n9whVMm#r3 z{F99w<|j+B0uW!g;EPnGRoHC0hNaIzdz%?edImB`_UISo@nCVvbbhzeH)C2RK_|R$ z>9Wgf_l(oG|3-CmEvC(k%@lgRpRv+*=EGyK-XW$F zB2n%_zIv3o(~}_s%t_XqdDa*fU|AsR?ZeuN#OL=8l6FpIS}w)uWU%)D7Fnh4ob)S! zJofErJkfW7#$CbDD>wtAwm_kn=Jwt%WqvW`Wdaj(p7R{jy95Jz8hNfH?PB8QU(4n< zAlGUf>^C{=hvw$1EZPF|o**9ek?N?r9zk0vA*MBEc1}FOURSPFYuCkO$Z$iWJV3{4 zk}~>CqKELCcK)|Q(H)UzrU}zUB_Hg`6Hs*8P#C8+J4J~di3VV05brZ2h3_Ns5Iop8j zrDqkKb#xcJ$HX+L=!L1;#`y$NRfKgQ6N?-inWCApV`ug0 zDX`qgJ*nk2+x==TV)2D0E3xV&ot|YuBc))wdfvQ`d0nr;UfZ>_sH*Ct{UTohNCc`Q zD0^_z$$78rY$lB7X~Fw%8NRaz&|HyL1=1Ty%~eLhBTU7PL3WGR$bFxLL}#aLFelKL zT{eLHaJD|q&-PMF*jKyfrcU#5BQ+QfI1cbmG?uq`>#kyR6xt%~^o0{`m&S7_ zsiJ&L;p&m+TlV(}Dsp$iqdaFfI;_5UWKb30Xr~~aGwY7xUU}ozz1szzzK&strgcNZ zkx6u2SHvx^QAo<{5=PUmFDcxx?$}n3Cx}647p%Q)g&*N^YKvyVDtuT>Hp&(|JNOr` z2&T_hXHZ|u`eox(v02A7S9}YO&8y0>%7}9D+RRkvViSRdwognu<@C)souh_r+qLaJ zlJF;3T5FPIs=Rk*xx;&%J1&>Mag)Y=+M1CZn!z}z4Hc!FPLmpY%S2HcYI8)}S+H-T z{gPO;tlOe`w@q7#LvQZTNuG9oq>nclp;LEvadQvq(by2>i4*OwVE4P~z^=F_tw#*^e;SIvzL zbJD2wBCO23_IW&TJ<4$*Aw=GVg77LqpKl0!x53u*L!bLnL*q*xhvxj_uK~cOLoZe? z&&qqUwWW0;=YilU))N-){`cOHF%Hd5J4hAUT*PHgL0_OhZDQ*EJ2A~RJxqY3P)YE5 zc8YZLjhc(A%ABGb77FwPt@hIp5v19al~sz6a`Fygf}9I(0C6+ORnVx7gJYniUcvi= z$;T-AOX$vg`b-Enk&&vU^SlUMZ;=J(X>V`l@2}RY6J!adly~=_pj*Z0_(*nyY2{ecOLaUg7T?jLr1S59`NbC zwYy=gD2npTEp|BTo1Pp%PiK~|#6!fekoGZhr}sLrtP}1|%*D=ngJ_pzL=kqq%s$7y z@=Wrh-D54EQxm3424oz`gw z<|^Nm)-@+wVk7sttE`nc5Wn_>TRNh-4eRQL9@UHF)fm-WeDG;4KJ?bcx6z!@glQLa zEwx;xU>S6;H)bSAPMi8utyIK=Sl-v-y)o%Ny5gc+qZ#`Q)X%WAIctex7fr7^deXmI ziXeQw|5X;glAl4XfUzky|D5DYd`zbhdopg_sr%Rn1^rs*P{gm@^isgbNpKLp*Rc+|(b7!a4sUChKcbd1* zwySsUpeezfg92sK6z1r^;@O@5xv4(sL(FDLRg>>0DEW;W2WM+fzIfU4XqGWUb=~y6 zueSPDAf7Fv8k_jjWEZZA%J4jSP4|tqlRpJWLV85CA??~v@gzBRQI?**!{*U3P*j%KP5Jwpv!i0_6};SXu$Hxna)Au0Tg~aGO8XwsPbpJ=&A@hx7q4@w_9|o7USsmI zjmM2A9#+2SaP;IqWqPja@h6%2>JHQ0=XG^{r%i;5bB!}1oxH3;cl9W22kG^!@4UP$ zL#k^>)XYij2{GV%^SK-E_wc)J(nnp6{gbP=AJp%(NJ*uXw3g6REEY0~X8K?14+pj& znMX$X>MU2!sIr7!(|t#JN7g8{3pH=(=y_zkmq{EopOlbAr1A0fMn|4vQz+n>-BL@y zL&!y>P&RWB_+8sRJz5~Lhq zHP^^!OB0>B!_n&SV9%z7 z>V`bdy1kir&7(AH4ZV2m-c@pAv;{%6sv^I6Eb*%2_)T^1Q(owMk)U=wP>!9^&gKMl zO`}|akBLZhCDDO~X>vgrn&7h(?Jw5I5=VW;j%%V^pWu?q$uNH=7=A#JHJn zwb*X^lM%&VHXdGZ{PLJ^Qyve%_=#Gbvn!@y6&c;uc^0|# zA84zquXt?Mu#20$)FmMe$DYfzGxdKgR4A@fbx_s`y3@c0(u&6j+Dk&^)HL?8=}9b; z<{FF1_gIxLs$b!#`Bb7k?BSrms1vt;q{k)zJ*yDJ-sAH%%Gz%p+(290kbmvoKv62q zHP~}fE^133epCjv?#yx8g7spL53EH&tm9U|1HVZ@2gHnkl`OBnt;?0p65Ry(&NPDd zMQ+7}@1C3#l2@kvw9+1Rys6uobbK*@*nct`SWYk{67Ej&b&B=Im9^)_VDCctonTgD zLssu>I1kh9D3y07K#A}OjzfE{`+{TuZ5!TF>8vBRUN(>5jG58Px-iqBOh^jLj{LD4Q1xX z1(Yo9==lbcf`_~$$)5MydwCCR2T6i1*PkxwLQ+$9NGy2_@qY$8O1z`P(M*Mqxis6h zK)%GKR=Yccj%=X2`3Zr!I|bB}hx52%gxq(H{OEn%~Rs7EE-99UO}-V)n-Oe&J7yxb`KYPo-X zjLUc|A-`e&{YtlA}Gg2V&w0_3RdrQ+dlfJdOBdRYa(xq zAb!=M&uLfa^sP$23y%uNU83^g%!;oTj9z-PY7BcE3#KVfqYKMF@f{~yA8)hW0ZV5J zYq>*;^k!xgR(f&Xn}GIH#Up?Be*nmjA~|Ax?iPUi1+`Uros4<`Tc+iX zNj8=hcV~7x*@i5QQIu&A%kKo3P!%deD#7M5PVsA%e>*BrI*s`i;V*{@7Tg2 zl`dyX-ii`~zYd!8`oKDGj7}YV`*5hx;k2`XVovmx-TXinjOV=8VBSzkY~to_Fhk>Z zo@GFn47NbJjgxA_{e;)KUV*;b?;-LAsjV$X%a8}S*O202m26`K#X)76nMFyLm=^Pm z_nQRkc8~}49W9*OMjlN$O-E-&^9(IY!pjR3ECURuM{U**cX)T0HhV3}KCY}NvAm4@ zL|;ZjPZ8&*KWK4QW1P*#m`1$<@ZpEc-}l$ak7CqmfOp>ziYU%?%ukbW(q>gO52a$iaUIuH3+{5qVswe^u|^qY6RF`Nmm=@j zE+m`8^g-gyIiDn2cy-hwp6T;xDL2`3zjXqRW*^jE!&uKfP4cNo4hx*WIRP)r3L(e}kwUIutozwBDW?T)>pd^ds(}=xj8@K)(`)*K7&9Nk?8s zj)Ut-^!AXC^7li7pjkpc+QF2SEv(JliQ@R1c!P>~$T_YT8a=y&X@%ZeS4XYBG`?3$ z#rb9m;DHEs7)vW+&N)0~;C}FhI8PPs*j#>V&nsI>OyK%*9U2jZXsyyM_mtW=elbPH z>7iaMi}2kQ@-i<&?I#|LMtV{w==s)p64qFSCz)2fKqw7AD%LflTe#7GS%qNH%bIk^ z$f97mHn8qt65?j^fra8?P=+tT%?xT0bV;=7_my&Xj=6K)K3~WmlmfwB&lWt7$L|UY7oQaHcg~5}-#+3PH*e*HPWE1hD_WW(I zB1+Ei47=h!Yb=NKp>|r$SMfuZB}uTEtt#AK-PzXf&RU*Evb2S3K+w8ebknFCSXoM# zLth@)@P8sTBYqG$-tKY`u_pML3py8&c$}-j9EqKX~zV?s7oHEIho`ibf*fe7nvEixwYmHSXxMR0MnXu0ZzQ z?oo$NvkpD5PjC4g8?E`|GXz5gy}snHg_>v|Rdq-sXHGnts{bfm#tf{Jmbk!y3e;fB zfz)I3HPhe^Q&s^VTfURMjtk5?G}y4}(8~1AmDo^g*LY+iCLAN9m_a?av1XcnX1GMs zkxF_<$hNBf`GIT1y$jVbNM`mmb^U=tNk41neA5mWMqM^{#f&}9P~f&R{W6*3en{ci zp?5FcTWF#i;^EE)xgpxB?Sf>g@tBF3TYcnx;)$k+HTVkXyC}%KP7)n!j??{qW906M zRQL4?GY3|?jXBQd5N5@^q2m;w!eGnUad+)PiEs|>C-O~Cg;3*n2m4YJX_1v)v4@G< zQN^jxI*+35Af1xbG$PRzrNo=rXOpv(CM!E<#gFIfI6Glbl|k&)IrfX~f^tIo?E>6P z24|?h9HqqkiJl8OREr3mT^&(PXJl5p`eNp)d0+zi?1Y8s*3FSw7W4j2L#PBgMol56 z>H1nB2)nwO3Jd4Ai;zs$zL`N}R$U)Pm(E3TT7DXO6lVf_d~v22O4|b%3}bCFS5o}T zz-CG_9lpU20yb1%{V1rH%zK*yWaorDDE03PB;7m!e=Mu>QRkwpBN%Aa7;fJoSSAK!e^EG++Dp$mTQ0T``NFrg6~8-&H44>K zHbkopIDOSuWuBM6yvKdze1qq^nUxXW+|tU%%4j4VR!rDs^M^$|Z?oh~ArM=Wk`qOw zc-%UqOPQ6oM&3dmzmk`m+jDs81JM#7BH;$@fL*gV0mp>7n=>u+K<|f1?0v6Ui#mVs zF13dZ1Ws{9Xi$N=UVW4!fEiq`tiGP6euV`Rz?>hC<`J z{*|iy1^R@_Oy)fW>U_yDWeQJ}I*}ImZ*k5Gy9JRDTRU)2^;LA;7_Gc$t6&oa7d@IU znO9;Cyo`>HvCXjfM7+x=E#BJS!6rt!DjFXhoQ=8rMkfzr)Kjez<#8e5_TFR=0oEm@ z<}PZd#ngXI%s4scR>ID=YR+(-FGl(R1oGr>cg6bXJmacGL8D2ZInLQjti4gRvutbJ zK4#k%9DXOWF<1L!wmO10qMeytnaD3hiD_T)GHUOnwuhtWVNm4D6g^gvAt}L^=&EQT z;YU`geGA7n!=szYbB77as|1@mHw7Qe&*g;)^NO98Ip51`og9RXc1T5YKD9v=?S)m$ z`Psi}p0o+|($Hyg3l^gUJUPfv$KYmxShE|X6ZaJ#jjeZ7&0ZGF+Awfm>#bTrS@?*l z*eJEj+%7n`c}ntOkbK9S6*klZ4;+kw<;aJ-t6v7{6>{z3zG>x|9*R76yRgGt<$*(A zS@ktc1Fm{_P$m>sNuN127ok-97jQ(l;=A=!!r|0ZzEX7v$aLx zsm>R59K{l-!q0_+z}FO9mikFm%#J+M9CF4~t!)?+9M5|4A;-FSwlx(r!HA10&T1q= zOJ!!QF0?p`lSJQBv=pysd>Uw6ymcF$82V;q+dS?1h$s1Vb&7ZEOYZ^Bk&=f;RLRB7 z=Z6Xh9yfVJUlr+#R=qpvFNta^M_QH2=dgt)8VVkIVeKG{b2kx>&kuR{yvVYsrd`4W r@xZ67Ed9|} literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-18-Quarterly Percentage Data (2025-2026).jpg b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-18-Quarterly Percentage Data (2025-2026).jpg new file mode 100644 index 0000000000000000000000000000000000000000..6510f2492d29712a1e22fd3c06ca3bdfe3e0cd81 GIT binary patch literal 25843 zcmeIb2Ut|Wk~TbKkR(b@BZ3kJL?p)nlq?R2fMk&%K|pe52m*ox0TD%E$Ve8+IY$FI z=b$9P8R7uL@Ez~HdWGxW-G86m{r3BxJ3eRn^w8&=?&_+ys=BI=$D_xyptGuqDvBT+ z91sWx_y;ox7lA47P%*w*Z%zT043g?B3 zyzI=(T%z2(R|SNGgutBQ(qh-7t_TWU`*9N-5)u+JQZjmSa{6m*%xu^GKYxxJK~%&z z?s%PeI2S>QKVd1fH z&*NVtynL0ImY$KBmHjp+_kBreS$RceRdsVqYg>CqXIFRsz~IpE$mrPk-2B4gm#<69 zE32r@t?ixNJ@o#;k8$CE@cue1;QOxwJ2fsUU|hKP_;~n4KgNZF>kb@vRQLpJ0)*$} zG>J?d&tJS2Kuj$klUm$F!Y+6lNps(+?+h)6(A*`|kD>i#WdE^&1^%Z-_G@5&AJ;gD z3=apGJUl886ts81mF9ouU&ntd4a8=cd}K?Ju?3}$qy>r!lg0MUDLZCIkSFl4WYMx< zSJEU3j^c-=)BW1=l>9{hp*d^4wxYCEHDN)_;1}{lOrbuv(G!~Q2YlGy)~y)D zHuqfwFSyZzC`XXF>SqV07o=F`4^8M+A|1&rnVvk+OXZY=-KTtHEoHjE1SPg!6Y8|A zO1MZ*JzFjj>y2XwZi1+^CKk-$8^kLId^3nUpj^5;4RssFaw}vjcXTU-j47pvaUO%@ zh4yuYgFo)uTk*H{Fipyi2iJ#E6&t%3WaRsMm)n*jp{=A2w4voMsBGUV)!opIX7sm->3Giin8atYm2KmjfFz4?$^Hhm%E6!=ZMmHHN$#4$s*%WbSpmB?6tX zhTz#RmGnsI^r}d0@;WMJCXLQ=v8WVD_tnDlJ<{85-(4=JZBYw*zbBYSOS0C4$N4TF z%7~8LYS$n4&}vz)pYh|2hjlNyc46nS4e@TtG%~cvFVDUBqKZr}4a2=|tkmmC?j1?q z8Siwj>0Nu979QIi&Xt1uf9-$}wHZSOFx*+MUzg7aE?&3K{`%|~B)4is_{~AnSbYv2 zUR_=@@69LT&BQfV`7FnrvXbolRPjWA4U}o^01;aH#J)6-rsmaG;?BO3xktAiWlQI5 z){Mq;41G=vNA;lfS+5;~QtkOMgs(AH#psfGhMgKCnI+Ydu*_HN76P5jgcn@+zQbak zCb4!XII@bNTCuD(W&Ndj3e`hb-f5d$x^q6Dv0X)GCm|DkXfjZ%=8%T2Q4bs|e0T8$ zYwH`bxhL#Wj{OM`@U^;VwY2nPkAjp~&&9j*k!q5~jT#bh#@^{3$Dq3?rSOd6zK47v z3Fj>~9_a9%9|C1y5y2rg6S<=$=u-GvtvB+}2&GeB@gkwz)+UCB!?}Al=es|3l~RhzqF}6f zRtOsQK@P1iLeD0HRta&ALA~nk$!Cz+;V$a;)sh_a4JIrfr?p?nj3bZvOph}<+z^3b zu%17XW{<%Z7#cxl+z|gmKaP=Tz4dOjk-eyWjEGZu+xS0U2-5ptHA>K6xXOG zRqW^~aIhdHacHKOTX|we;+?+bPxc9gA{hx}^4Sfb#gR+14xvSJYgJ|H##V(pD)j3v zm*R0&of^+|j8MNwjKAbd_0d*_#Q_H+{5&1r+K;^v{?%eN*GZLodx2;;ZpU;ox zdO>TUFcr0P>9lE!J^Pn-$C&{s-CPeu3)17Y9O7ouM{i`;1%}a<)FLq%Jk(J8D{yF` zuz^5&dkemZpN11O)bZeact)4wret(zb!25C?MBt*{j!`$nt>s5I+dX7 zeFZKP4__7xSeXq)JNab``=xH^k8-7jU;nRAuHw=3=>Cpo9m$N=k$%d_3B0a72Zcg| zCX;hJCg;$JEZ0^)f96VJd3uA%9Z!06xL~OT!lKmfWUE_o%}~FddAw8Q!@+~Y`k7GG zIUnL&1YPzGjokN!{-8#@#M1bP33usqogq6Hqbm!VwNCG3YhJU!Qdk#9o*#pndr`&} zt~_YCyi%|vv9Tpy`8+Y+X5n7Pf$(t6?BNH=tNOEcF=yed18;PbFp{Xx@pE=pmQ;u4 z6Q@ljWl&V&v$*uG`1hX_ToIRkXuaG(%bz+lqK=YDEYH8o7DJV8)wQf=Af;=)z)yR9 zPWr6z%hs*#i3`zUWTT%|wrLiVc^be}wbpjD!L|^B2!*-?{D|1Q2RO+)!x)qP4d@(W z|I!no7(JOTWfs7qfm0m}ystiI|1p2&EXFjQFFC!)>cv{-R7`2RNI9jHPX`h6T&VDB zd4+(n{&iYc!|46|%bPG-v@kGM8b z8CGQdt2uSCk~K;@X04a|R_}?rq2tohIl_-_7v~jV6zd#llv2?5KTjXnO>o2-wYj^u z6VJK}FKc)cL;}q739a2F=^$-c=}=#QB;Z_tq+4`AxEP zkJt7cmw1{53m4OP3vqn!+iuHu90pyfHE(NZDo}&|twWc*$Va3UMfCLI(8l*mwsz^C z$~lr+UCzLRVUAeg$v$;Sl8ikmFf0(_5JielmG7wJ)%_$LM^?+9z`M=Er~;GVb_^=o z`++FlnRgD(Q)IH3-3e0Yzf#5&^rT?nic-!u==lI+732qXL~AmC9ZDEI&F0h%yAE;b z=!L6Rph8>8g?8(iOYeVFe(|WT>+tCjoM)So$O0)z#BUbs-31!S}6Mh>jeiEE4Oi`vgT&ReR&$ZuVh~Sk3dCc zb?-_C?GmSV$C;iOB(Zo5a)g_7*TBvoNIjgorD;2K-^H#$>n^&?@FpE% zlk&obeI%yw(6-!)$DpuGOc>3_4?}krpKGnOQx@nBN|9&iV|YRh!VGOLaxGccwuDS7 zeTZ0S$nwa}%!ib2cw5kX^pPI47lz?mkI>yRzL#}veW*MZ2kKv#T;j#Z`3O)$^naem z2Q~*AuTm!X+H*WvN*_+#DOWN3MyQfP&IsXGMA630(|_^hLD}nE#CsM|$(gOeeb8Q$ zWkuWNm}|sWduxIwyu^G#Y!WG!(J{Gil$qZ*qP8@NerC4 z`PMKca)fk1Dq!u_M_t7!npXuLp{mH8_dU7Bm#PJtaY+Ti_r!&DucM-Z=Ee2-M2^f~ zDc-!9$;6s+RuHXuuUgb9?cGE>Bkme_3{O+1Vdl#|H&gM{$}<5hb*Qt`A59(Z!WCva zj)>fF5=%3G(1tolBD@DX-FFOH<}*761(0Bc;0{*Dpl#aXV^HJcV-O?m*I*jNQC&K~ zlKLITAl&m9tCQ^sKAnL;551L+K`CX?SbR=&-^tdbiPSpq-Q8%w*s%p;h+(Mo-y8SR z3325ggVxu<*uY^Z`Yd8W>G!rNSFAMkjzJUS&?AC;IFb$89{pQWS;OA6l4DRsDg1!+ z?lEXq9v%t)t*L)is=}R$A)&)6pFG<`y^MA@rEA3V92L?s;gIe^KjG^R z?m~Fi-KlCWe=oy%!~9^QH_jW++TN>xX0ZWT%LdQ}59pr;fHaW)hO+hfUb?&0GU^Ru zt>5F4QX&?nLlOx%c7?c#_#B>&@N(2vAlp7B-f<`{REvqu8NUO5IFhZZbeGg8p!{k3&#dft1?W}qnVtu^E5dGa#@tzJ^&cC#{uZ}N)n znDzyGJ~W|_vP)6FxUApoNdf=zW}WrB+id|v>i|sc{H21A4Nt1c5kbO-q)BH~_>2{< z5f97XG-pjoe&l-u8FzUay$+`sGc>x9Z*Lc~sAw1z&p^LjBrt%bww1`rt(fK>9M=C(6lFvrKu4OK%y>5W12+jikl4>pln{*4}*Yw|vius9X`wm=5;G zSf>qSCZNgb z?^B>53{arf!1*T!(d+BTL!Ib(O%y}r`$L8*>vAHvAU@TAy&!rdLu5VMilDfW6HlmA z%YI(9<0SwcDg^P0+}DO;y`BdtY^ujTRxEbQQiWmmyisVV4^0Z%@7v4@_vL-|vD%M@G#so@p7}I1CzGD#l78HJ zuqXFfF3N094nOS}AXrphu0+BdeGjyEkeuXisu%+7f1|zWYdo@M2BTql}~fl)K?_7-_N%Tzan38(?1<_ z$lPbg8+b64{7K=#a$>&zbeMIHe9i6otA{XT@&Z!4;LOL~)9?r3Z*AD*Sv>~rd=bU^ z+unYUQ9A}*EGz=1P7A(F_?T?pIGn36==tgEe%3dzRPfF{@R4^HCfFCx1nP4U;EoOG z!Cu_{nFGT2$zQj30Z%05-6=O1F9@|4xMM7UpSudEMoR&$=TuzOYm%%DmU z!UwN#`QS=bj)^S?{rE+87{t?(mlAIovaF=-Vs0958Yr#cNDC#eWxBLIQjR2jPNvba z^gUdJMUhB+m!towijDZ?@nY5l6kd~l!T1EPwH^B#wyJV%9jN9rD6>vztF5Ock}#9y z{n4NfPdfTKE?mx#zs-c$>OkbRZI1PeHCuM|FH%;whVjq8absu6t3&&GUdf(SmryJD zD$*YIbgPn^M{un%YrPtOVo$Ga8GE766fNjz!8cVs`NZPM#0#de(Kl!7U>eRaq;Uh? zV;OpxRNvSELuETQPC_kzvE5E_;kp73y2gV$f@`9?GIywazC9JhH(REY?lCpA49=MH zQ9*$x3g-?*f-IiQ2Z(8F_6zUrK#y#^gl^>|)KblTKv3I!pFakjyK41xBT!t7q9wHn zOEv0VWJ=E-6+h2t87p%U-E+I5zB-XyqLDg~m>7pzWatd>1(&Y*vU=7jV=H6v!CKBs zA9^0JZqaV-sm?BXnX@WzL!z#9+p<>A>t{Tpb&Xiy zbl2aSLBo8s0sHDh!{KUyT%n`igO1&^+fu^)s@{l%?IlcX74C-TtvPwR z6fdBB?BaQMVQF}7grvNO8VEzX&I8L_AIaV_b?<@5!WqV9) zc9q_j#+rzz-PMPEFTIEOoSpv4g$}a%vEi)kupe{(pVK_@mp~tzN7PoR;zMQL3bWAuVabR~MjyDv5y>@OwHjd#1@vzoVGYGbzuG+ZZ z%-M8cIL`lgdyO<8EbZxDP+u5Fbc=9>0?nlhr@#o5-*65fR$ksQXyp0K0ZxJH2NXXh zyt?7YQyrKw#4O@gP zGj@Bw9{8~)6WDy(OQ}t7n<}~u%uWyuIRx_xnf+TAoT5Wp`52p zMY+~^P3O~V)O<`drZEwL5{#%B^bIN3*_stZOG(*Kekint$KA_b@%x6L@qhx!D6_qX z50(bC7wM3So%~c+`yk}?{agyP!XjcTaaWrCW<{UNYdeWi@Z~97ZfRJK%;J`=RccL` zVn+hHqGcL_LOhu*ebGC*tkG~zsIQp4CZ0@%^WFOXNSd{mMOI{1i>bxgC}LfkfSJjI zI9;yPnn;L^r*KcGkJjvW+UeDL=J!$9{w+s=;=?NOCI+IQ-ta*r-Qyi^JGaBu`U~3O zyGxIhmb zV|h(jL3rm1(UC_sHe^R*GVGhvF$nnz{TdrI4)|uGXu#1UL4-9dO9K))dmSEW+NVE8 zKlK{%&)XXUTdweT}wh=Ax zea>;EBWL&#dEVFV1K3;xCT^V_Th)LK89G1OpS+}{K&_+Lbi(lV@(*EHb9VXfl0K8g6r*5zC^F7gQfHPO*jBX`Q~$JPIshpRu;MG+DD6QCb5z&rDZ z1E!H$-5`u5Fq!?72PoJP`RZI1jfm7nI9I_<+|&BuZ*|JIQVf7qt^-K1u_k=^Vg-b< z39)&}s$u5~9X1f+^v*j8a_>(I94J_1!zNQdpozvBws|&Jcfcp>wNvvn>0xCa<7FMq${=eS8F;F!nj@U{1F0TaEEuEgJJ!>;7k zJHT}%!oMt^lS+EvDlTB%0mCtw+&SoMz(lbNcu;?9) zty{#(*9YWhSJg=aeSFLNlOmAMenNnikYs%fGQ~dz^*QOL>%PGNwHS#3Y%|F*C~YyD zrRznY2A<^*OIn+eeBIv`3kD?H7Tt#%);3^EcZI7*zj_0dxheODff6J^vGQ&WJ7dt7 zMfq!spMI}Ngp?it*XbqxLN^O^0#?Du1Y4U0EcKGx=$M4T7qy1~Iq7yd0`&$UAiyN{ zI{j+?3l#sllFBmE%)igN0QAul`m89$cIo>GRvOltoK9|1puR)`x0{=AQk*J(2Z0; z?VZ_PZ8u0di)X)+N(VjFRwk&)(kuzA8-I!yL;SA7ko3L-zFy|bg-|PVF{fe>UeFguH zG~Vqqo04lLJA0g4`HWK!ou!+@xw3#~{%Zeo%($Ns#~?OIlY?)FkGjLIo0cc4N!Rs# z0h;E20>1u=ai`Wu%1*AKSjqOpFQ^SZy46yHzk-mSfp+{C@%00V}_L;8y} zTZl;LCZ|4pR1Ugr7{5GplAE|YoGgmxgftcGo&PO{Wt~F8unI?N*!obcKV%^9qaT1* zoeeqgkzRoR3A|j*_&z`JVFOt9`7f3|#VUK@NO;7|ri2^-nA||}MV+DO#jK7bKu&V} z{~TDN#@QhXf-b7|!PmO^kMMQA0E->i8wGUOT7a8+73_5=oiiT-k7GOCYgTSl1@^j^ z$&t4xHrUw6IQ>M`K*=l_eipD*Zk)c3;D?x=W|jT-^>Jv2V-UxoGT;jXC#hvOc7?L5 z27#W(`~bb5;Z2+^%HOf3Zoy%nWvA%iFm8XC+`e2qc}{^^3wZ3W_P?%s06PXfODZ~Y zgMEdpN$d-rsD4f@kVyr=*%J8r>(cZ8xFzS=(~j6bUjY4f0da6eb=&3yH$7jdx;BbT zm4i8D1QAvBmjg2L>~zN|snHXFK-ckO>j2|e zzUWGk=BXbF(u;ebPk`s8{ZUB&fn$8eOjIzU$Dml>4EP98_Kv4{Gipt)TBU_|{y@K9 zQKkLR-$7>@Re^~*7&-Fo!mJTi7%=xd0*tgDkNwsD=XnWWGI!n@!PanY(d;DoygE^R zpfQX9)*c{&z}H{F>y)kY-@jl{yR-nqgbgEh&%%!AE_n%`nN`c7W-g zs+*wu7IW$soxaf}L@V$}O}0M``ZUATKP<|?G981uWcs0N09!8GX-Y?OZ@3C+DwzL3 z!C%1{pO)W|$H8L)tK@5otzX2t1{oAYJ*WgA=*Lrkwf}jp2e`S5oVLdxqqzO>gU|Ih zPu5G`+QQxf0ACv75PhRr%^O(P<%pUqE|I8&GtOa1kJCl|piq(7~E z1NtUs`ja}w&ZuncjWyUIF(>wEY#rlCo)X)*SuS-M@Pa-ALh=tP`iH8rZ?Xy1(_%yee zi0Jmi$$=?GSc7)}p4=9Bo810eK>k`-qk^MKn{*oA5HePD9g`b1rCP&j)Kxc1|ZufHJJNyF#AKX_Cmjs}?eqHFfiV*pJQ zx{M=F^3)sm%Lc6H0d2b^c8XhnDeLu_lbWlx9Mq6}ew*`#aARd7O0S zhgzI?9(qnuYwExJ1zV?_1+a$4NwWuh!YI18C!y1%Kz;jkb`H)wf(gDT3OI%%CGchK zFGH))ll4+J@g)@S&^yY18tZ9hDSCa{o4*4yQ}s>)P+O1b_pXDlqJKx6>67l*n%S!j zfI0%dU-9#t*5w$qPcja4*c^MmzN9D)aF>6p=dUsjtOD0s+5lgFK`)vUYgocLrz|6R zLZ2c9YF)YRpY2bV?(p9|_98_6@1aONn6uZ9B{Yq@Kgo02TQ5>P%}=)^expSErMrQG%Sx)g z)HuP%AmYqu%+YECS^!Wy;ptc^HSqr75FBX+m@+g5r5c2En2Rrb*TPd$9 zEon4AiU~`@38YdjjV}TDQ;gB*JIR2kB5V;DVw(-C^jD!g-yY2%bEJS(51A1l1!@r9 zoPk7O4h?@S(d#-ZFn<)+H3L?if>q|A`@2;@URFr@i96OLa(y-Ju3;9%@W23d&{Nif z#TN$ucE_Q~J|#g4?a{f0zxNFY0Q`2t!mr`tFP#cI05m-l*9_lUcLtKd$d!J-p}Uag z$Ch|EF^Av#KU|8`Nief{F+8xU*&K8x49*kq*^98v#YNMti6eslB+iY-s zF+Dt|W=48a1e8v)e}z?s;J?=*h3r2unO(KuY#7Ra806bne+CM3S6OlU%9LqI$An4Jb*5?6#+dN}1p{F@g~{Qd ze=0jZ9}WM<82>*)@c#n9ivh3j9jcaiPI($Qb^&VEjra^F$G>JRC~r)=P#CgQ)! zf6?myJs!|9qE#NFx43&&(XkEKLhu1lEbf43pGj$C*X;pJ-!yz0Q>Ys~zVpmNAKUC4LKC7%t3hs@5hEUq-nTWxlD zJyK7ZHFGT8#6T2ZE7gj^-Lh8Z`b%He$!X;kbyHT-bAk;uh~NrNGA!tzEZd5}cIC|X zb)2NslOZzx2Zv`Ng+L0&))6&+3 z?9_#mE>37fi9~W5HG$tQf%}X85ZRRT^t3azy)-o0MMf1-Qq@S)Y>3e*5hbU6m1#6X zfaU$x^S3UgY8tX8pwYB73&D4v@r{yy~bMFN9-7brl(Ow4N!wO(D|M z)8BjI@|{A@Hw$~arc$}rV)6jT&>h2PTJf2cwf2w`vpX|^-S1eFWsOuYUuI<=s8uny zf5^dU-5OKblDbC|v*188g}yiH`(m6S_`|h|SIGiVFJ-oIhWpgX58r*0UEu81rOfex z4Xl6HAGlduIhdgp*vHI$zKwxH>#uLcMLTs|!wPJBaef9>ew7w$eisIOq^+(9lk5`CQ7IbryekyK=$?xmP6pQ`p7q0G;Fs8U zQYaPEs-Tn@`M#aC;4&GD%n4>*Ei*D-JPK?F%`ng_&!d?(117bu^$E2aGwO)Pft6xn zZ((igQe;UyNOy`pL z;)+N$5=D1p%E3Y%$aPR#T}`W`pVz#${*`HfMDd0%Xu~!H9-hNqb9Y%ZL1fi4!3`=) zxs^hBBduzN+~dPdl$d!rKY2G<>A3|rX{EOXHEWs0?<#9tdEd47pOtty{uw^&tHyCofJ+B zRvzk~mN)_mSy8uw>R_e8nmqp+aP-lEi99g|tj zevw$wfHnJFp)HoSiu7wmsC{q)W_wE#J|lKyh1a*7&rj)5ql)68vViceUPG#0o%K45 z&a?;E(h1o26q@Pbht{KCtbPN}zpLEp*P4tS>Rn+Bw>eNRPi3}aycN(~p(#seRjBx9 zhH+bflU$y0ZZo}j2GfkF@>lbg zR&106;}I_}wJYxs)qz|}H4SyjP@L)m4vFpE<93Q)ZVjJ(!zJ@-iF26?uPm)n9XcO^ z5S|}jeId$dTij?ZV*xhC0mr+mcCRuhTd-dUYN7^}{vk8qN8C(!$}BP|ibjs(P&dxQ zveV;xyY6^>mehpgeUTe8E}L#ey*P2k={f8wnNw-LUxOPD)`gd5h!$b*Y=9T{e30W2 zU>#*YU<7SHSBRvcl23~u(DWzp+-pUU%epKS-EF^6SAMTw$j8BOM{I*ayy$v9n6W#? zLr~kAxXJ%*PtL4gDX53sW={eLhRoH5^*QK^$bY_;CfBZf1ZVt`iZmSdx>8(V+fQ1c zujZ+`xIu-Z4o4?>9RYPdxR*wY-r-_h4Dru zA{Gj0HB{@9*PSOEkro8`!U32PD&jHtrpRN|j7nXo$jPm{wj?-`G2VzP59I z7@|>cdau#RYm;(8mNibn2|-ylwlhPPJ|*D&)ctBra)CcF=L}(W^*`huZO*V58hSi! zcVndslz#@{P5@7{b0E@2(yhI;UqqX~cyX0O^TIcAvsIY0u^q#qX#4I~Ip5ByjzR(78O{@qkC_|h#uo=%R(axJb`+9(EHN=BV(R?}s^L|CPW^FIe1?KI^ zk_qgyLg4ePVZQlUal1^GZ(1JRe675HXCc`y-nohOJS)?{v_UM#p#39DpCyMwATdpF z%}5)hF)TE*ZpXR+qdRA?-&OK48d%*~(}t0uZ3bE-?vFRpe3kknGs!@S_Bu13D^XtX zAkBKyct({y3@Re+&+-ZMXok*i9sZuTF#K)J`ENR4k3tgmmHryvGb$EHY!b)*mj3v) z{k5dD+V8iD8_OBxQA+&BVXCEel;M{dmvvbK+6@7VEVEe+-b_&I2_-sABZH0c+Yv=6g8Wp}c zP++YN{=$vGqn%fh=15wam&Lk9n(^^^$i~reX}Q@Bolw8 z@e9RI+<7}6ML#SSqe*S5tq5$uukDy4&6O#}bL2ATaPRmC8RWtXL`Tm(^)9$iHK0|A1dlFCVK43nG2{7zw zwwm&RNLEaFXF>LHWLe6-tk)x~`KBiGsHv_jxk=#|TPzu4)Q2M=Y-PQ6uP?cqE@C7R zy}GGuUVK>%S!Ka1ab1>VKIKiFdxiYEXd_edxXI-TQIWl8IWAL|uTxD(%k8C5>~^ zw$)W7{6nPe`xFZ2HkotLZRF;c>t1l#El#4ix;rJ0DC_LU$D1tdSjtLOs0AW(pIimy zKCY9`!I; z_i-5zCC+{Te%a=nOo%KPq`U3$)PQ$9{<9v{K|AGbUy|)_kfNc|(Va-1CD^Me_9PL( z%h#Y1phZ8HX+3K{!a#$GO1@Z?47RLi7o-*%`l}E53~mjU>@pqN#!5WN)L~!6DHOel z*3N9C+)WzX8|6@}638v;WnMz%1KZ)Rfk`Kn3YP`d?R%9|<2)sudxJ)bG@OV1rSFG- zKz@DeT12U5Lk99pCum%$d<56jfBrKG=P#wM@IsxvI%V@f51w7+^C$OBF&sNk|4%Mu zK;8fyyBl&UJz`xgrUs=Yd)`MJSyvVnNm1Z{Y$v`yj51D zd1MF!vpLlUM6$&xW13-rRmx7JZvng!KA1wuJ_v+?=Wx(L){f??)+3cutR81>(fbkM z>lV5e@u5qb)z)dJ#9t;{ps0<<+bQg6#j-`Wu!2QziGG%iqQ-75G`w4rrJoi@k$1eJ z3)%Lm&-C1^(!v%{ZgTj^S5cnx@SADzLWJ<9w#Xm{RHBUb3SnrSrGT$}`8URgiwf?R z7kH?N90q^1@phd>V9Oq$k3qTvMHsbIb!%S-Db1@}tm1h1`%Q~c&vDjVThVGa>IZ6# z=1D)iwu~YTBMI)Je{?q)l2<&$D=h5xt}bwq4d>%E!inTpZ3Zb6W`;%$5|qi@bA?qC;&jXo4f?3XeLyOVVz=5G(8vs==;PNdyYG!nz=B%U zUv)SoiHxc?eAOR%#K}EvO_W^OQs-cMDdqy2je14&Hic}4Y*pVaW_F6rp{StXsX-o& zP;SOj@-^F(lEH^%^dDd7J~B*MHGL)qj&~e)Z=PbLSK573GSd@5cSTc{1q52!DYR{q zAoU>9d_U|W8Frg4&g{ne8`jDmS9WPDypiR%yoEZu?b9oSv{u8iGU>w7CxD9Py1}5X za-ReL=!4EDl|)tFpi@N-rl{jO2ZW5iKqvRreDAuU7$ybgS zFEnqg4fI9?LaGNZ4+=Y+@lC%Tf3dnU*!UXuF|d~r?A3%W;^BN8Lm%1W^jfKxc|>lG zc$%NsT6v$0@7dOdpe!>*-7MNie3mq4$y6p=DCSj`)B>-J=KA;Pnj!*t=c^1Ln+iv- z;e5lQGLQ>X7y~$2+L^OO`{#ua2eVvwtbfEK{dx#87oA0V*Szip@m!quMStxMpTh+) znVsG~S(vn8+QzA??6+uN;Q5=qAKKlLBP4eoH8~wSCu_{DR?ZSb!`YWZ# zwigEHraPg|ASs0NGK92AhQ2mI1Acz3%64V5x4q>#5&Fq`2gr>Yr()rgu{u+;yO#Wf zWy0C^mbZL^SiSscJUwe;Mr18xJZ5)yPc*MC^HplkB2=4)Z;q^a$4HI0KU%b4B#6@1%wkrnQK0#pPTvQL@4{ z)S4~l6wdJ|`t#&QfQU?ujTWE;`e?T{*C~Q8rLkj!?@34+IT`S`My6}Nli3s$0EUXj zGaQa$3>u3RgT756w>g#W`7;sn%%-m*kjFMR^OR#~op0Rt;xa-&H!aY)^x&+QZoZb$ zxf|xLI*Rl>Ef>N;UoN2I=R8x+NgpWk+mI($R-M8eKrJEYuG%J#!z$J{M!;goTncI2&k zpS#j2s&<-kp~#43@HXB69C;3kOd%6Mi8gnq>U_NiYQ!a6*e*=k;6SK}ikx|`!XaZ0 zVdJ5Xi9x950EtVV1Atv`(m);>4m*$*G;l*X{qBPH4GkK=Zoo^y^_SYafsWqQ%N)Sy zo*y?>T-pm{uV8&lp)~kFsGbLS4T?(Iq@J~w@YGN@eZbvOo}ebF07P?*-F2k_S8ZHu z`}7hESxlf(?+%S|M~nXD*H;sJ)XKs&`&#e4J2>B&LIw}@Vm{yu(YAk9L5MfOKX zyc~OuxQ}yWTa}+TLvEEAk!4VKWmu0=SHQnxRq3)n;3QLFZoc9R%^OTa#sI3U18@VAARM;rNKN731je6k!uyEUO`Nh?V8QF=2O@A%U z+&D~$5mCtHA&WjXBb%|5y68nGy<12LQNT44Mv=DnFPk0m!w4KK)ISL@D`pSO?CVel z?GzEC-Pm1SJ%RVVaea)Vr~lTYQ3ZbW-8jiT-CB3cb0pC&skO3wKSlSUmw9*)mMRGk z>+j`&#Z|&m<6Kuo4F?z42;9{(RNXsg{?_W15-6@~m!@Bv)Q(d5ebR%n{``BdoXeF9 z`tCjm3M8~S0%;Aoa=AoaSxS>}b_-{6Lb7D6Ymk3KjQ2+swLJ z*{fG=svqhOKQv}|UTP6wxE0-!2_!H+ejp3%Q)3149{mCE47Qq>XtKtYb0B-v97M`& zL!-M)qzc+PGIV2wZI8RaOO3Aw^gh5*zX`H+%V9_8Hq3rzYbsiuH?V zCbz~C(F{$mlvexGez&?$S- znR6I^>B?BOr2y1NC;2a-i^NB^1s1xu^HZNkoK=*|*Yg3-jeWT$6ZxQW)2*G_Pu^b) ze9j|lq2qkV+%bq!4kG`JQP0qBl4w(NByro0NyUg%E(lh3W}VE}!2kPT+)A9G!xIdf zZbBRedgE84LrKjP&4=efnvhTw3|}peJZ@_WDzN)59N5gqqy+=2mCIoXWQsX zxIg7f?6+{bt-Nqa{@|=QdvCWTEuC&~V zxODV}jz_|?IHY5sKWE&-K>poJb-Rplj`U~PyMLwIOb8>(aD3XN(Nwz9rM0}=t4(Kc znxKY>8&+xl;&3L9QV#?5>(s^;U)pufDs@+`!Um)vJwA{;wAb{}P6L&Rnrp74z4p}- z-UlPuoCEKQ?$r+P3bP>Wi_5FWlInd_51m(H+JOAkQ~UukYeKQNxQkLXv3zV-p}!bJ zN}?>1w8{vCENBoV5HMm zV?7h9?MnOMzUfR2GNnh*>;0u^>LCaR`;NfAc06!L0Z3u4fUnen;_j%jTr3yX2kx;6 z*U~RNl}}>n5vusqxOc;4gG$u)>$>&A zR_it)Io0R{z!P)k&@Fm|o)?>Y1UuRW(ezKR4C_Z2Yp-X}k&Ic$jviRPj}ELAEgrPm zP4|lDe84V~9O5q)O))xyo@oneO?-TrPi89gS!g=Va{35StgmfvSpnWH6hZAH8;w_H zkri(#bcyJ{**-_w#}s*(8Pp6|lzWz1GnZ<8H{|~2_-0c~%C+#oUz{JUlS~VwzY+m? zzoEcO1m=9~<{~*2QG}7!X7+t8PK~Nrigq(sSyUE+jC-{0T_#A!%wDW^k9z`JQFJz2 z^R*iAqquOh!sb2+wsc*yekUYZdwn45`_(9R z<5#R%Rd6C(8S37EL1EG{dsk~G2lBT1xY|e!_xU8wY0*$j;(R8xD zx(MX-HVdY}dnwg{ekubMe?gb;U@OMLRPxBo)&i*x8$41F%(x61!CzJ&sURxBLBjjC z@k2~iyyUaGHFxF zsv(EK`WLuOEePHx*WV`S8Qe=D;kDE0Zk`^xr9(e_)%THr&S}FZWOW?!L8Tzes82PR z*(*>$OH-CrAd*l0=}YqSB)mx97)6lx&sEWgh}{kC*+?s z0Y)Mjk7BB}(mNI|T-7Bhi}~ls^5-SQwG*$xl+c-*X(wr$R8avXrFry$lj%N-F3Kb2JiT`xFU};wx1H ze}kSSvEwYgL^Vc2zb&>%wST`(qGX4hQf=so=q~LZnHc#++!hoN^zmIVVHjK6B~7)u zsMb|zbp2YpY@>K%XO}nzQiJ^be4I1%Vd>t5E^4?qcv=cEI5_SXnrVIGHMz|6R5U^C bReG$nt^9xM|2y!@j6Tg}|BLF-kHvTf)_3@AmnR0&Ox-fVnKf&zcjle-&MK21 zLq9(7{bFWhYQ(p1A0OYoy&vC?QN9~|2lpR1aA5zzz2xA*gNJ@Te)#9T>*TRxM~(}e zJatOoq=0~+keIOGX%T>cz?lnYM8w6z8~-S#P^u|Y5%^xi|?Q7(9io09zO8X z{v&(MUcd0|+rNL`f&GV0oILc?&xd|G#<%aM{Ra*T9Qygx1r@<-;KQdi-2y`ZLK3PP z!J!XR(<&;fdd{e+U$=okVYWf(882S6;Xmx^de*2LJYx7Ch`x;YVph&s7;& zkdepZ%!{#mgwF0Uu*b$fJp7UG`)O~?_L?pV?A8CY$HN~?{+aAG38?HpbwPEHfddk5 zm#+stJe~SugzxzNJwSo|0(@8amR|lls{iWc=wOSX!h4v`Gwh(L{9iD>U!3ei*(FF^ zDljcyjyyf_V z|LTMHpTz%}ph;DauP9Fy_Q20Tgko^DJ_)%Km)ldT>EH)V0Fqe1cbJ4EMy1J`<;=K? zrM2+Wc;C(}ga$?0o*ZJ?=~>$Y)24-hb)%dphA`9{;*(l2Efn6K6c$#X|G}qFJG!e==mjS;Hw{OZH+Qa&AZL@J9YVtbho-H(6Z7si+?-*agwr@L zCYIMm!kq@+j$`b_CxqoBZ>w}c5t^aoZ*+0Ut$raMyeJpd$36l%2vwdB=EBa$S`JEV zvq2x|+U{ODj($1*yog$#B(Z@uVox%WMD+0PN5WwBX;&1TW+L!%T}Rr!eofV^^_>e? zb&~TW4)W3tS+GndjAWc=CDY|FDHyQulczNrY--f6-u5dhZ(0|nP>`N?KlmILoLk}V z;uFc!7w5wYX;Hac5;D5y8J8>fan}^~N@mLv;pEfDBg!JZt<^fAHlNI~T1r4{v1ouJ!NwC>~Vn4C>P`%X&As zOa<7rY^l>hQy(?ak|Va2eU6|Xd{5AZ#3;w2tmldN@PO9SNli(wHdpKQ1`|C`9%5;1 zy7w|9(k;27p7+iDk%ft7wiPWKE5OMXDhQcw@PjYDJS>+i9~#!L$jI$ZP_)SI=X_-S z;ER;YfB=}5w>mK9jvfeRux>BJv-)Q-8#{k z`JF?%!gb`_<7t|aFDn}9TZ=9i<73@?hG)mr4`LLpcKsj+VTL|>r+WumhGC@a0Lr=| ztP{+1N)QfHN~1gRY?+n;gfzT-kSW03Zmi5V2w%85l$7D}z92JwIioKm#|)Fn^P*FF z^Kmc9K56QSFeF$2+%^pbv2K4Tn(&JK{3R=Ab5qC9YA5U!bl&;SOd9W}E6+<)1uFIJ z>h1-w4h986U8+G+xn8|!%YMxfkz@fKzDAsidC~7@++Sp6=g@%Na_-+U>H0Pd>HC7o zi>Fk;$IF9#~q?HuzFw7i_dMzl6(SZ2)36Y(qiLowAs$(%}*3n`vJP35L<;LPESdTtnY+Z2#A}?d zW>^;5%j7a`5~qwxM;IYtj$9^kh}`5*=W-9L@~$TQ+m_Ke)z1}IuLg$yCyCAf;%HBH z7I%6m1)o_EONJTAsLrsI^iFoaP5(8RNnr5%1PtaXF)FORIB7L&pTJ-sI?zpPbvksmx<|(qcD*rqs1!Me z6fVp(>5)QA#huGDo4BnDs2x(Q>?&Il${DRfu={%h%6J}jmi=$=GQs)YVJro&tm7}+ z1}B0Q0TWhBHuy<>MhV`2E~hdrxv}_jRj4NTa;sBmN&U5O&@k5-W=y>};Zc?tUMMc6 zoTlav=SQJtnc7N_6oZ4Ghk8om$N8!+jKsPUrtw|$X9Son) zUUr;S9`;BVnWfG%Lw!d)H&~v%r`Odx3HE1Ng#gHsGz$rJ%G_g?I&W26Jdh#NxdA`u z7piE9rl<)RONH>pgYyOo;VF>d4pe%=LN8jk2W1;de7Uf#qZ4ZpGS9x3{{@G&j>ynX zMguU0w)9J-#Md|TDs$4J&P?9n&6XZB7}&Us#{12IeYTIvHwyGNNmHtruK9kcqTJnY za{GI`*}SZPr;Sb5r1djn?DE%wBR~gSIn7LyIcXSzr)n&Y%PKswc&ipdc$@X>BPK!BK zEFJ2xUf<~v(lTN54hjk)V+C_QOYHW@X;ckaz>~Go`g_ZM8>A;A3JXpM1Bo9y;VsN3 z1&k85eW{$ULs(z&G~F4AfXOZPWc$fOj?R)9m3eTzOSobR-q)g!ah>gEER_H1Pxv)-|LJKU>!;~3}m2zOyRbve~7INM;SHnnAT4x{CU{^g-x z%nv?O8vjd2K_i=v%bOC#82@Cuh5QuQ!lecjYVTNl5fuq=+PuLO z`dqFKMc}>0HAT#C_~^?uGQygS1JHdZ(=_fF%ig93)or#rt&V?ub-U)1(pGqpSx?w)w2=jrW;RILkJuD)TpJm$tu z2P3l)`^^>@pUUe(c%^N1J8hcAroSm-??iW>=nc(}>u19Ux-7On7O%in!t>aM6L1*; ztyKw5ldZ^>xJJnKncy^o>hJRDBo*w7jrv%Y6f0E`PYmgY)a zp7mxa zUQ;**7C(Lr_-?Qewzbm_TihS|u2p<^mXKuUy%=2>hh2MGcKg+fd-0pbPk?SPy~j`F z@;r_|I%|FYcdv_lXWGIFGS;E}-0^(;#agESP7nRF$H}2H#xyWMEDs{b%Fck^h{go6r&{NV~HQ60}nRT=gLR0Dw*%b8yHB!~m z@=IiI$ir81rr!{LCrB^yFpE(I-7Zv-pf3|TVP;;^w^jWyphurPk|VCPB^6v4EH0l= z(zn0=Em}v`YZOFV&=qPJ-W~CorLfFFBFpmn4z+A`c$mqS1QiTWCEW0Hp-!4t$n!v~ z7l|J-v!r&HXU0qc0>YsYN~#SfP{Sp|G?$nv2gPvYG2q(g3N*X@7(TUUMd@?Jw}M#M zY|?Osj-)e0*L(CF<&g#ZDmeo0zTlA!qd?)3FY7wBfsw{GoUd^HFpl?_c*E`BDFlP#+ z_dd2=tsTgE^+wxnech2nZW3ku|IpwFD!q%|umTBk6J9j=)s_PT->MN!M_cdTL} z9rf?o^oLhD>*3@lQy~h$E)7nsyC3MYd3Cs}d5K;Mjqrux@}nz$XpLO&YM~MD@|9}E z<6WsvLqb;Rk)_W*%uZ!Joztf-wM|sjqs8Inb5wfQ2JLtV6*-pO{3WZ{d0=a`DN;QD zyRi~?_lX~n41sgGXkKOIrM~L?lYX>@_u0Ig+0n30fcWR^PMzYF!XaNAjKF=!EAKLaog@Q)D72_7mmU!jv>8}M@b6Ln%~CBbBy2t4ETDyyE^)UjH6ool zk1L+=gKMeDq&}yZcf3xv+V(}_20wA0e8P>aFn{H+T5$yK6;;#^zV+X@rlsC9l(;qs{i`^$sqBp2E^%{|Cn{jg zjFi3jhB&;?GQ%(bgRchW#ZY%OB)|z0I(d}dj}P|emj1gi`9G_C`ruE;ARF&rQCBj) zb@dW>+4r3%sBY|Nr3>R@QlDZ3qQXa>)29>1UQ9rj1%Y+%bU83`v?>?I~ zS#s!t`V1?Eh%|Px92d`uh0v;#I|Uqa>4u%C%DmIzvgS|$oS#8;!Wy~AN0H4Rr0wNU zcGpF;o*v|0hLQFkrf&lX-&OD1ji^;#b6FzTa`nwa7!c0nb2F5bkyJ(n8SF9ca(3LQ zL{RV85z~oBSAT^wn54=T=qRYBW&1!Xn$t@PAnQd_w7ziFYI&wKv^W)sYkOk-*7Ahr zSROoaQb6E*lgY9g?X_4>Z4!=k{WXg5_Bbr9uhn4c4CRg%86FCY#7lhnbp#3HS%XKT zPF0VV+4O~AlEpTbl^_KN?>`L}aeNTtvrQ>&T(@ZFFxmn1-uv%$0G4b1+61zT_Und0 zRDQbYP(cv!c85fLVay?_kTJ>YEZ0hm6%;ZN^RlFDWVabxtAj!+N(ETXFGyM2Qd~?~ zDd$D-xw|-|dOvR6j}Atik9^?-y~aA4@K!Eg?~`%l$TLxK?M>S&?C=(3GA3__e4blB zEvqDuB8T96@Nwy%*Lwg#R9f}f9vu3Wkk9>%?juU%yb>tQ>mSU`@)Xpanzexeq~vo z=8D}ppGY1X-90wleZ~3!sngxH1FT~L$<~otC%$Su@nmGfoH^<*km@DziH;U0^5=Z9 zUU8>gE~~UYsnClsgnkCpyhlIxUh7OJ^yQTF+)b+-7QQAr$pHmBww@aV$~z^zt$G6P zk21=iDJc^MJ|@=L@j`>=^rhCy(2W3#RMk~dZtVC>(D#muE4uK}H%NkGfw@mH$rG9u zjI>dE2QH**w?Brx_ToAE8>H4e6v}ZnpoQ=-azKQ*PDM+~w*$?>M+V4U4+DGlG_w7- zN@6Zb1W+&TmR?fM}K;F@bB%pPiQQ0vqL$snk8W@ z4L#v5)^KULJc9GLx~PtdTQR;})$fcP+vvMG_9`%O#^s$M@B7biY5pjtGeuW7_r`<* zOs&drqYNOoDApgn0f@-G})CXM_G<7!6{}UbBrn>VA$*nVyi%4 ztm{MTa~!ESf{jn2h%|CJnK=Gdi%EfNp55Hayi-W2zUknVxvQ=hqHO5Fydw1K3L4XZ zO@hz(_h-t-(z6aLRh#b5Z6tloVW+&ChPB-y@v=ywPeKl=`xVGrLPUQ58W{J!v}rsj ztX9!VAJOZQtJHm&uB1>)?*@h>&~2Av9iM*gR9Q*wxWF&{#W#3N;nTZzP)-4J(k?g7 zQ@|m90!Yppx9Y|+@_rlk`Ae6);p7`K$Fcjt2b;LyTL(F3)4y+D)6n0GT=`c$^;$E8 zHsfrr%*OI2(I4{j?umVZn6$G*>(0!*eP+* zDs#meGjnF}qCbF_fI-0O6=R324o?UtoAVt>Nsm5{vyj(Uds&1@Dss|Cp7diop+PmgDMC#8)TNF-iNZ9VOKTf1#e5-b=ArwY2lg2^(sUB4z&?E35=_qA!lj%#!@}`KTqwv8EUv^u+Y@w@k!CVfOiu^pU==IfY1)!GbJ}@I~F&S1YLfH$MA+&HY|s zU*`7g*PE~ZOcSjj&7W1>8lL8fKlpc;|Lw~AkJh;gq<^9&_t~@`bFEvD3G16>wfO@hf&#%Rhu?81F*`o2q9p`e9p|9diG^8y?e*%p%(!B`Ifjz zW>c)dG#*T57rLr8uP!FBoede9*Tb6{4}RQiQs`ps`n}^Iw$g#Fs6KOh z^;nF#`gVGwkJB{Dt!?i+)XIZS6VpY4+CBC>z@mQ4GJH+<8`2yTtR>;uh5FRCPReL* z)ni$d*lslf#+x*TEvZW7FP$kwM8h(@hn502)>xq$j(;4Ky}coK7eAvKbfW8Aha1Oi zXEy(#qd$h>z|t3eY?4ai%ve!oppzp6m#TqJXFH%F%Tx0Q4m$rW{+?g_#i5|ab2t1R zVKy3j#kYNX`K$GNp4ioWxBgu~YUl(nm)$3={5JS6GX`8qP}}_GfuDtM{sp)vp0h8l zI6Mcf8*`*Ks=wbU9pR=hOnE-vI>4hR#44w+Xc1!I%-_0BW0LP~R2l}+&;Wr5<&4hGm95P!<=x4Y%GBbrubc`7P^IA2|yys6Wy1D8+J zazm`9F1B7~zXcXXYalywKiEPycgcf&HH|6eBOgo9;Zt!wfs(oG*G@^Sam`V^0ZM-t z={e;^6oY5<>dWa)m!sLuj`?z|969mUUKh_sdLvY&|IbW_civ|z^6aH?BP*rf|NT|? z->$rG3l6n6Zo{9cHxwVuLbZNX1P6)UvE5 zxSFGhCL~-$k}lpVz{k}H6(z>AIhkrMx@64JJoL>uwO)t7-L8toBM4{fd*+XbLrqzV zME}ncwSlM}Rk}scN2eW&Ge%P@xW(Z7B3HQyQK^|+y`b9IikIRRBYrV`&g;%d zN&E|YyhCDan}TAPNWKA@IQm$s9#}riGiWbVg+1vLnaX~tJzF9>VwkLw^_gsDuZU?>xc~<#LuR;3r zrprD#8}<0G-c@bJjy0Jtakk>nATYQmEs|aQI*VP%>~z=sit9 z6T|gFFdY5Y6e%=#?;zyLqdNp_r3$<)Pxa>Ly;0o-gj5gX6sVM8_B{q)c@oy69T5^3 z*2*qme$MZlmInoEc!p+%-cuU{)>2eSK?s#I<%X-}DH#6in+iQ-Y?Q98C;UqA@UF*{ zN>(ejUE!Y5CjAGWNA-y1St3oL8;Uprb{4b3ie-2=Y|XJvm`HZ3uBm2-rx2yycV?hN z+*IEgsia#YK^86zf41%m)}8! zkOg3EsB2xne;_+ANMdHrhb3;QjlR*t+XM{@_1)@u8dFddk^=AheY0g|F*wtr{z?C} zrO=)0cZNx1O3lVAyRt83{q2GP0q)q9esXPuW^=d$4WNAiHmsq2unNf*y}lOh}j2Qr!YVqb+=M7?D>6#N}lsW;plYw!SHkMxWiy^F~zt}Dwe^EFGPcLKm# z?qzfOMM&O_8gUB_0RL>OjP!%gcC$X1VCcFc4O_1us0FBho)YJ$>7U^Q7Xj54mu(z} z^-`I4JWSTtPjmtR5N)}x3YU~}N7&TZ?0x$)PF^{+HlFVsaxZj;iRTZUib;$>$j`kc z1KF+2%R7P<#QyvreCo3I`W{|g$e(4H%M{kSh?2Qp4kFLEr}iwxU@p5;HRXeA$MtL^ zmy!~9lkfV6za-wx(}xwUMY038{J&av9UC%M5~PrN^#h!fI_w@d6 zS2l?2A~Nc4W=@qU-+Hql;A$oC{pt_C5kkMdmhsp9Kll&@V11f+5Ae;#ab-8a+SK{! zle?s-8-0)e>U&}&+>m^1=o0fW{GY0mRw{p^(DmW`$7pZz2q9)KnU9VQfujB)b#ra< zZ~Qg5f06vB>K}IhZ%O{c^*H|z z=j*D?^VybV)%&#f>t6k9*xbfx7SF;dHzl?Um($P=_b)^JLjTzy@KR2|vDZjajD5SUxzz`JQZ-m8BelaU1)_lN%wPZ(n~`&nR}}Kp4|GEU_ufhAB*8GCU z;=c+XOR>p|Sn;Y55YSE^9Z0QmUgrmJCy%FF8nI1227eLUnGEUX39%aw~M`ASiCSol<|FYTb9wi3 zXOhPoJWYA$a4&7TD>0cp0cKI})rD}>ajP+sXYh#aw%Hkl+>ko^G8g=)x-L%dxKpL( zGX697%_k9!T8r92&9=D9`T7limL3=K^a3PhL|a;0|2A6uwF3Uvrg~YqxW+EOoSBfa zm_*Eh@yy8lWOu<#7(6v;rY)_@i_92o@3HBK2x!7bp?l-cu-8L2jvTbRILQF!r43Rh(_y$Zw`>ebG&sZJ2 z7iW_Vnl4m8TLfocxphsWqxz_2fsdTh(Xz}L@Uv36u?DGh+I=oTeM_fpl)i#=ch`}E`bJTE}W`z7IDQd=E(>ZTqi#Yb^yu6#C?@ij> z%9fe8c1djPBKM-XO!vkn^>)r;v7cJ1MBDQe`kqij0AO>6Ttfm zYAkCgHy{*8yv9R6Oy-Of`9LsBH40Xs| zV@|M$IaZNdQd{qnKnms5cJJiC^q(d%w-w&SKXa1Ai^AJ*_Xll z55bPhr8nIR&PCulU?8I}pw;K+-bY%vViJ zkQ4`j{|+sG^q zpZ-?l{~Fv=6f9c3WFtG;2O#Yzw}G_esebFFZ?j)+BvG{gPzx;-fBzG5-xCyDb?+;G zHR)gKKJ8yt+8Qx!&$bI?8&RJ9*s{8{s!^&X>l;TEuN~MPsq>8vlBZmFPF~)m+z6n& zawhuRA{kjsCxelZ+r=;^W##j#YpT?sDJ2+7q5OiIJiMgLmZOWnj3S>D5^j^RPW0fh zhOwDN2qd~EW0UC^@CxoEdoSd05&dA6 z{HfxPRpahvnWh#kCC19aKy_6*4`lIMe2rUBt^8yKGKtoh+MfReBiO`+e-*j~%%0@$ zL0_ry8BJQQWcPEFFkRm|tW?Loyb1isq_?VaxvP8@v8C)+rq_(AN56+U@gr49cd{6EwwOn*<8t$n zQYJ~SnR#NSAsPwQ3tF>LlC#7zoQgWMojZn$PiQ@&B`@K3{Lin|pJ^<~pzr^GHgN$? z;ZR9Grbs9(PVme$4wVQCBYiepgd~OGZs30%o=qlJwwiHn zs7s(^0cyG9gz%V_><{1Y5o`hUFLaIALiJ_(&}?F@qpL`lxI(a#JCQhF%juCYBEJsD z3KLPy5gjWFXFNmqjsoJcN5~tjYI;p9@m8>c&@f8ZPjnZV$8OWC05^!uEQSe|_J*sF zWX9Kju4b?P;G4VfvTEL7K1sbMe;C;a?&*PO2Ay%I-d0Z--kmr{o9dGGtU4CfO%7de zOE(^~H@)v0tO)JIl^IKzB%?ZhA8;mzU)}a4rX`MwtM#ypm922{qoUw50|d=-A}BaLt1)y+>qWE9 zXY}=OP`71O=M&zDht;a}QsxV%+@0J806-=0?rf;h%Rz?up6}pL5fd^xLuz}&D$JuI zlmJ5rCBf!~!(7V+ePvNo`is`6-YgEXR2}4(T;o_3!mbQKY%&!R+hQ$%Eeo13^3zXN z7h^YR;w8gsfW*WF_(9P9`%F=@Hjg%wJb?4wX*I#nARMwEt*CpWN7GTqSuPYrk;CRL zn<5-V3n!bz29?0~%wq4Zh8NA1mDD`m)L6uw46R@lukmWkr9Bm*Bui(5^!q|}4Da)= zcj@7QVYj-&J$%B(6l{4GZQFL|1B`#)^1_N#(zuDa37efDE9bf&g z1^bKDpb;C1I+qOYWI>!78cN#hfo4sEH5bW3t~X94h@X;GTu!wCS#J+i+3@J-JH8t+ zl^8TDJ(g}-Wy`6_l^@K+J@=`u9~Ie>O5PQ>R7*_Ul3IdU`3x~V=qG%ZC5p>@bubk? zl#Twa1geXzxnoL@MgIg=6OVfOY@je&t^cZZLs}>|Y(D8DqYq(*1yq zS_7TTK3amL@X=;^%b>PiJo#O=CfL8Ehr;j+8!f5cb(YC~<>=X8Vb)xe8%*F>Y%`bc zf(*HxH#3*oy>)`&Bc;Q}_g;|d!>hz$R!w86~6!Xp-*>sOQgAkpwX}+Ah=LQKjY39n0vNc44y8~Hs@)Ffg6&WsC(h; z+=%ku7VkD2iv_n08sbZ|&dKw}N8M>qXC?M;dlIJ>Wnb*1p3t&ZfzsM^nhBz5(K*rT z+66wfxtO8##WnxD(qJ_XP8{rN`=0C(ED*FJB2ls^NP)pi=P;jo_}f5e=P$dBw+lo9X9O#(yKH&;jTdx=ylMQ~nH;cMlvH3Frad>$D@NJ=+NPjq z03$bXm-2I0U^tSKwZI=3M5sdmbrKU3BT#2J;?Puy%aP7c)ggMYi1ffTZlM^}ZdvdC zD#3HBiy$BV`wzYaO~wruA8}`hz6P!l;^axVZD4DNX%`_shPBB$oSq%hu&-qDzmQ)^ zLY6s{)Zz=>sSp$$_qt&(K%2CcYHb|YoRKq8vbYYj(Jp6bUnoI*mrJCFH!gkqX)4A4 zql|qxqWrUVnR;h>uOQ@Y=}Jh=FwkM{`fzyVn!f<-Qd0XxlP`}l(JwWT8d=4>F4vn} zsIkn3{=K2^puk*)qh9@;LFI2g!}qUP*)AK%?wxg$>&?l3OQKNeK8typB1SD03Kd?5 zr)TrPx@J+Ms>{)ysBQzEA=aD6jtxCl`(-%!yC?a@8a5qdS5)J}h?Fe3hrx*|-Y9M# z+Y=@*C#7HVN-jSV&HTp9A=nIs&}w_cuZ0k5ZOA61Vpxq6@w=K6zASGXs030FT-Mf8tigdFtqR0JqsVORl3-#d0lTH6}Z0Y@UgKD>Wp zvg(j+cqNw+3Uvx7QkB2FJ^Jb$5lyCvw{21=Xl+{C#yhfLNiwN4twTQbOcJID)T!qY zr~k}Xqc7l20dqpJJPJ7;Im+KE zE6Tns*3A?fj0Bv#tpvFU-w_^WAL?47LDw`Ba|@!|d&^S7Bn)>r19*~`7-zsvEd8N5 z@omxg-X{8Q73Q4i=2&71`@1(fg@0!mWX%1+rzR7W^yVC&?rz=hd`7)|NB*_;`F=7k ziXJE!y^<;>g)-*(>{fj}-Sv2?JuL5m)z#_0Pr&U(w9LuB;vgopv*w7i`kMdR#|NRGxK3TJ=Hn7TqJJsWT${q3r6vva^jM0|XxV>@J`6PKLAsP8X6J6UpG5c z@AoyWc6UWVq79TWF{3`1%V#GymZl!MdPk)%P|lN)>iQ(og0!Ebhi`KJokl(VmYFw_ z0+F9TTGw-PS6AJLqL)s0o}>WjgGxoy~axff3c7=Z8z85P{JS<+jK zRr24MKX>ty5JU1FiOK8G`PsLZQ%PGr<8+aRKIw|Kl~CcccM8qT3n|!o-P#33kTL(8 ze36T{uqiVQ&3I3M79LMHc)fzM?A{X$Q>d;*v(?P#Cbfmp+=x6%*W0wuewI2iLJ^HR zXa6RakO_G_7TADCYYq=&c0+rc-0&!&( z(iRF~->gBlt%CJ-(;o{Y)^j_*mVCSCDt>(PuPKFxXKW2*+}L@h7Q+rnZgfdg*lwD~ zPKbto6^icMh$>!_KDJPH+A8^rJE0W2^TizF*tul&?#-#_{~nFQ4Azi!Cdu!lR}S3S zh6UJu=&Ytak9ZL_-7`lNAIO((0~E^!3ne9%M>{>Mi%S?*i_Rsn!*p?GHY|CO42XojNJ^N#Mc$`; zy|0!aA^b>Ki%ccVcI>>PM;)XxLaqpRC(f=G3`<}(YvO?y<_38q1G(MmBD%0{bH#@# z$&zIBDO(n!@kJR{c%a-pUGAd>6lLtQL=H3hX}JI%Q2wPiBF7vnA{ zCIHh40zn*uhT|Tr=@k2{StgatyY?)1SLu$DpWdBI&ubU=8JNo7n2nyLJhxL5A6Ze~ zY~0-Y-xDp=;qa|OEL;yF*1gm+mg4S`^coB};j!#99A&OFn1!VLGW#yousSKI7=hOo zqmOmot$W=q@_90-YJMlyGjR|cSjReAla_NRK)YA@(5EHP$3`hxqG$UXlyl#CrJ&`Q zdNxngSBP0K>;doV>HJN<<0I~JUE9p6N`rwQ!B>c3E}e<#xh7(4X?n~pfjHF9+S!>A z7^7E_XGoi&?2(#Ln0X+oOs#|DzjvsdZK|YqPjFI4p#3q-#n3DI#hz0qD-<9zQln{I ztve&msU4=UZ@wE6W5-Il7a9Xed&^dwoAvHS^`Y*tjQV@dk|bQcB}A&FI%>bjkXasV(-eodQ~a#LWq%+x%y1CEyRaoi zmpT1hm^Z(s%})KZT4d75*I|(r4I$d40WV4fi9|S_S^Mn=9~IVozx*@SFZ@UqzuBwr zPYY({z?%X)!u`=`;hwB>eWhHgs9yhRDD*QF+7*zag-_m#26L!0wCb%~HtKVDWVx`j zq`9N+M>Qo)-LoX7gt{c013)I^mA{gQ(KFEcfEVS&<;)yRW^Dt4#PzYawpk*88k%Y1 z8B@PkQk~y#JF9)`C~0tJ3YoAwNZ2Jij)d&QfHN&?3EJ@EacBi4kRTb5= zl4gh)cpz_Cqkj-YcU}v1TFFd1tUHS%Ev+dbdoT#E75K%*)xNVfIm^glzqdDT4mhv0 zylZe0d1~jtn~srkS|)vl16Q4bSoXmZceTC3of*M)4}Ky)_)`x39JjkEQMi6T=iJoW z=$%yX_N$H~dH7FH9vIbNiLS*VMVnd@zF6RN;(>6=L|8<5f&jp(Q-#-`lgPOhx*ys5 z!j|zuZgSq;7*)(NkbC^mA8@Hodj~$Ou8@HWn<-TXfDAXynl&yuF8ZVnL4zTgo9BwG zi*ej|wg#@u_yHI=pP!pw}!D2fgspcG4J5}K6IK^+Ob2q8p5Q|W<# z5{gt61t|j3J18}UBvL{NZ4?0s(i2K(QUZhyQk3SGXMCLJdC&X4zw`Zm-#OPjb7<^(lDudBw`Pp_$xvb)rkzcuoO|hLPr?oKqAA6bRs#po$`5EwT)T(J( zhKPnT?F4qKQBa$fv8U!i9{(|-sU1(#}C-&gclK3skxLR?h%-rl=S|6l<3T@=I6W9I<$qR~3 zOE-CQ1?hqHOQrY8VheC(71q{@cZy;pp$VCG`BvDRuPDZ_-OlqS8Kp)t5-IU7zRkOzZ!?D@ZrmwzFOSjIR_NhM{#DC5V z3(9jG2-wHl%qtvhYqlLc{4d+<;f&;P$*`x{G|Y^-@3Hhbye#%0a?{q!U3!L#41fwOHq<-hW*-}<25Ewpsb z?Cp(b2WJ&d-272u`^RBTgZ1;r+@F6-ZLIqLC_R>ERc6(^dFd4E&%avv z+))WEG|E~SU1sQ zrTs=TXvqLh%FE4a3HDqo%b&!-$uw1flhDp(tim8fzx|Bq)ojrpK6^^?55^c4K(tBR-O zEhIZeo^ljQ{5|gP{suT5rtf^=UlLAPjRqEcYFm5D5DCjbd}HJJ?a$`he_y<_#)g(x3-t@Sd);d*Bk; z;!U-D10A?E2y!clFt<}Ex%#p6LbrR;$G&$g$5wt zEJm}+=9=2KWLwdlH z&07^pRyE~V24Z9P=$DOAu&^`vv(q5;KH0^_S(i$}r0RM*^LMn~O6gzf5-~>_J@wvt zpo^cd($+CJPtVR7x*&Qn2(!07B2=d+{S>;%HC?*{w+yPZh zmW%M*e6plsxzLE!D=RKaV+k4W(^j@0(6?Rnf?l&O78jI#J$q*aOb6X30g#n6|F$CO-H`LvT74R59{`IyKXx zJ&NLNj`_+Ygs$43NJZQ?ZPEr$yscOkiW6nGO>124rf&m>XewQ zfR)ixQ?==^`m-b`-IrY&nn+f2P5fLalwn00vq{-}u=`%hdi*D$`^$5x z07sFy)CZs!9`^?eMBYT@8ca*1nblG+ATNIWgLC}k@=uzZa*6Nf+IRV$b`^e&f}6nO z9fyC+qax3KOs$Ha+S69l*m9oyh$H#$fZLK?Fq0fy1yT{ry@^?=ioyYFp61kcYi4}C zH99V?ad&V@XOUxYH2Z;{bhQ5bs%V1>bUY985ngFj#IJLs+2nRwD$WYzQ@?Pejd?gP z^uemU+??)UW{xB}+X>qQOn3I#Q#WR^9#Uqgs}8dH2yOM~>ePOv+%;bRqL7pg%SMsn zQavfrL|76*b-yjycGO))J*^-zj_4^*ied3MI$P1W)`#2anVo4VwIj;?KYJHfMMwnS zw832Xa^7;M)yO#BN^8~Lu<_B}v>S;;P8+}=&~90CZv|BRgjN{SROob*g1FDP_>|MA z;ZMN+ta}16D)!#}>Pdq=@R)i+CM-?Uq1DJaB^fcJ;v19jFnNWq^M+IXz`$t6j_#gs zaLoh(YDyu&0& za5sw{*vx!mLv#JhT^CsBtI#CLbH*{1VA^V2@Qv-PnBWiaePoaJfCaw)N2`d9t<=AL zx#F6VYxjeb;@OTn))%uB=T#%J$zh4lF>BX!H>h&wMAL zFbXp&eDIC!@i(@Z5{@o^J(E3awBktr_JLS}5eKBd#WVrce}(=YRY&0Bk(&8rT#}Er zKh9!3su3-YwTnU9?Kuz~U*bh^0hm%AqDf(XRx9F&6k%@PYQ3$+2QFZR>^jSn< zav(0HTOsFGhlvML9#Tm^N0QBG91|EW@sKFRq=ypqP+qld;){$? zWjm7_J|%b)v*1j3hG(NtQc(KcK)DMygF(fSKNEavFWZ<RL5_`Zt*J29PjQk#M zetOnfdwYy|t;Tt$dCow8FnQnK!Q;p&WhT(~J=^d9F9fdLK#zdlQq76YN&?NPDyfqn zasqX2mDql%k7D`USz6o3OKg`|w>D=H8|U-8N>!mBOLwZi9(VY=cskUSuO%JLhVuRX z0G#>Tfv?s6W2MWZ2A(>zbaj>`J`1^mi7=%qoSyWOxAR!ZF(1dS(D-bi)wwp72}Ij1 zJ84^m44rpGZJ`szNv*CGItxfXdXTurkgw;vdSP6bNfe+*O54fN$uUy|zVi>B0RRY; zQ?gtV;a>g>Hn1eg4W!sC)>)O9*Bx5ga=$cvd4Q38Rb1)}k~V*P$~{s~&>Zib`1@me z8P6axsPfj{>3wJ3vmQM5bE|mUZtwH*gBkV(Ek7rS4UT$~`(?zsV>mwW=b+?;@|WH5 za6tnbU_T+I#jqo*&T%Z}khbA`|JlnEZ0i$cgE*Q3WZcbt2>|Z2&~^0-G3q*VUsQJ` z>x{k)A{%OOx5)FOEorTjhMT6Wi$4Box&e-_i80+G_rA3;Ye-m_^Rj8kN&k{pTj*r2 zaE1QDkp-_pk^tgdXI`--GB1)!cY0>^PFzUbhd=T>0xM3bZjP5;;(4j2{`jw9XGw#p zQPy_hHQml%7Y(b#Ca7tMLyf>pcg*5uS+=SU)EEfd_5mBO18ROiAUa!^P#KbPI)mkW8|}6QzOhZ;^dQN(kYZV&__d7Zb_P^b`oXo6!mMU+Wt%XXc7UFer^~LW@j)Xt ztln$~X*ZiN_nOdmM8Svo_M{9WwID3ftuL9MZ^ap8g^Y0(PRjI8Sqt+Dn^-162GqFE zjWzJZpjPZZZeNT8H<9yL%lU=*m^|;**&-DG!uNm=zQ zlSOy;Y}5DAFH)~=FHJab&-RUql=;0AAI+8s^On#}1gsUzKbIv+%8&Ac7b0pN`2o|i z_x;S|PVIkv>lrYqj309T@CvCMc=gIZ``yK5UB^k6``qFG!xPUbV-+Mniw_H}9 zifhS_MLVZ`WBVwd#gZ6!U27==4Fwz4__?I^1O1(S43r_ugV}WFuSL)gc&JCBx%quG z<^?zYhc_dnD1&K?tn(sN|HVDdaJ>KhWy1Dz?&!se4+nkuYVW19kO?DBVQxJ+(|lgc zJ>3hXBsTWcf7bi!>s0&U%p+@U$ghrn&D&QRE_x-qif&%i>r|6?xu?1F_0NBY5&!+n z|MTS#v%^l-NQM1z+;skk@r@1JeEFEgFed32qA$p=d;fIt08FXI{_Z^xW%X1d*VP<| z>Qqa-x&E)&&5MT`IFPW@pi(`xkY1zx1j-FEp6asN`O14@wR9*zSPqwK7FH!C%}wIY zA{>KZHSnF(W?WBMjJph~1P>punm~9!sk9_{D*6M>PFe9S&yske%R(H~la^LYtnn_t zz0&bIc}Qp3ZyP9QvHdcAq09SpO~VkJ;A^#LQ@ERdJCC5NqNXbXRfwHo%v`g3I*Q|x zgNIaM{@fR4FJ^?H-a<4)+};Lq?>8Sh}J&+Q!89K}vzrklI4nb)Ad zdKS(DamfSPt5pdQ$-qtof-PqoC^Ies@y~CXpa#@27roJj#i^A|AggPUsxuua_LA={ z+Y=nj%AB*FC_goPb6Zed8sj-`g(KIbOSM<3+P`BK`6=x@Y&;XJK{pJ!K1bxFdHr7I zUrJc!Z&XpvePg=Rh(u+y70K6`H{sMMrMw#eo^h!g4$i&QvnRPNVG#2IQNu&3Zw7R% z2gchLqSNXLqr?7K+T(!7LO|@Gtk@=DV5f2RrZGcjhKe^8MoX>2J6!`1l5();4xXtN zzFtYXd)0DYFkrk6t9kIWm+u4>KU&d?m<#&N_sObfT!`1g_(=SD(~olLW${S4TNWOR zIIpz+L$M&do{`c*t1d{selRDI7Ycj-`14MspMTNofvg~#Mx(@)cVe2zc}1F*5rnbH zb(qfFYwJ=bQnd-YKHDZ9%?DLRtpd53#(`WsZOCnxxN1bOGbcL*o>q}CsLrd)a?Gm4 zn+G4CJP;g^Vuj^-Uu=*#^9gKrM!XZskAM$wWH40&RXR=7&hka|4hDDEQ2m3lvIbj_ zVE|tgK&zuU!LoRlhxOM{TKB#OXFh{5OB8xwfBoPNYvG&ZSl!+#$8QzD%XTjYVpgo#8x6`Atq`TY+SX= zJR^*#l@wpil}SQ6|NLy#o$gN<63AP+fyQ28L}tZ_#a%AePZc8SOrskf$^n9_LyLMn z=l-PB-7~C@u)tlxsMG|EMYnd_mz1755O!k3waZc+b82@b>c{Ro^GnDKES+Ac@50ct z#*qsvSX!Fuh#t2{W{-?xdT2*NsL`3ad5`3!9U9zQX?=hNwlf1>gLTbV*%x#iSaAjJ zoiRN3E0EUI)ca45*GXyBj8(tBz7O}Dqf^?Ni`KdpZXZtC)Gg&?>`zx9gMBM+Q0U2S zcD-3%!~3Z_qk7&&9*K?8)caGm4T{eGzBVXlVoKl>vdaxn+ zytgV7JMo2V_4b0Qicn2@{ssL!1^IL-bhWjy%i;z*Y-|Ki+?z3gjggK~H>o$_Z_4jJ z5iuEWBZ}J7WNx?+q63uKj_ZG)d;V)P{I@)EMOsX*D-IKx8=Vi8j(YzEYNpqmaNZb> z>|OcIZMT(b(J3Z_r%-~jDk#X8cs>RIUma@}W4#p+y;$Yg_Z zZ`w*3y08?51q#qiW)ZGJs5~LXK`ZHDTd4(P?-p%4WvhH!-I@#CeCFE2FEw^DHGDYy z_7+Kh(yq^N$nleHs#V0mGF{Bsg-G)d11o$iH}99Ci8w-V&n-vYLc&f_eonkP4%8{ zrT&LPgmZUrA|Bz7B_7jGR^3STuPj3k1%=y5AX2+3Zvi@evG>-4vtpzAf(9hl2*D@l zN|lm8CUC}D68H*pPnNfrGDPC9yi&@5_*D0I_CamY?z1|S3Kr$%LvAg!!#+SAAf8wY zYT#)e%l~pK8Da8E@iu4It`z=MvSnjbE6NoGInfz@qG~*}Mmx^>k>=>Nkc7Pi78zT{ zhR@It4Llbc;cEd}h@E+dv06Gj`Kqcr4M!#qNf-{UUT?Ror|CDYIuXB`W3}i;z$4?{s%O49fprrJh-X2w+?dJ6zSs+^nq;t3n24N zY7_LcGA8eN8zw@P;lGWig$~`UZ7OaEwNgCzAUwC^@DR6_9e1-N`J8DH!XF@UO!NAS z`x%+8K4DalLf>GE_31UElYJ`_#wF }5$*!nT2TC1&4&8Vo=xO)7jjM;}zLIdgb z8>MUB*tAHpzz3C}+x7`Ao6ZWk=&*ELstJ8Y>|MDd(;;r9ZAw*K_44d_5lFD<+KD-A zEmhIAdepCVom}ngF+Au<)unC&YnR%Ni`e3Q)YJ+Xr{pdK+5!O1LCG&gg#46pD#krY z?5pIjP+=3F)@DH!!M@6XNYvcyC79pXxIhzu=ls!8GJKe(H}~Re_hpyW{Gz*y z!ME6~{{vy>WM<@TgzS%!z zDh0JzF;^~o$26PU+T1je?l^g&yzy&L0ajcfrqS@%^^07Tj&lhJ2CqV{?U^@h@34Lu zdMYuOE?#$yG?q$a`*4*8Hh%q*?VTJCknJ2)=M&ehxED_uw$wSaGcT_y&#coqiiDzI zm;IGjq5t|7Uw#0CxjjD?1T3FjnBVU$<~*;_k!l421g95zfUo8ce_v61<%TDhZ1`hJ zt1Sm;jfA4-jh`nLVq33b5*}wgEsMDt_17oo<>xWvmdd#xfj#vgbe~UL7E1^3OXY{H zaNo$LQO8OfQEQn&EYN^Iyy*&jC`7ZTf0LQ%jtd3xNV(|S)tN5egSO*s*Wcy+shbq- zNi?A{yVQ4vlvS5-uJFv6guymkM|8Z^KUps**FEmemC#Gp_%;F!0c@QMv|`1AdCKAhR$Tc8h;8X|Id70 zI{u9%2#|j_k&i=#lnS~!$L=%5r;AnNf?;?e=NCIV8tU-2e4nQj!{W)MDskyMn%bOib<6a}rb)C>MA7 z@rd}~7W>ZRoq?KP$&vHmv$qChd}YS0;qSd1lVbqHv&K^o6XV_>uNf_kM_L6GTUze= zCb5S$K~1wjlO_z8P}HL(i6*OT=jUKOWPfbYxY5Gwuc_EVCpX6>ei z2RYDn61l#NbjKjGj7$a{2m>r(9~BLB6IL&DNgx-2`?Fw2v2@i&m$sb^qxAv)YzM+6 zb29I-0yywyZ(U&?6^*)+o}(H=LAypwphqYe7*Ncu-UP&^CMZe?WWEFWU|Ll-MN;ZD zS0i54wCWk)Sc0XO;?8|2ti{0x3w(4I>h8R7N^W(KLVmt*dIqdRKV4+Iz9x-g=~3S- zB57xrBM_p6O?0N$@1PGm)o!~rwy`+{pH21nmhbfuSNFmhzeY4DjN1tb(k#@A?=jCe zzi{;%v0J-5lr;~BSN)n8&qM=NiuL89`KbM{T9 z`KHM`4ZbPaTwN3ak=O41D{4ji}i%xi(Y9Z_p>ia=2}L@i3<#mU~_an zN8)i=v(aB0Y~t%^%1i8dxf3Pl5!nLenMbSQbr0)F-`JM@E<`kY?r{WZDHaja3~Wvz zURs#b8T24w56RFsZowtw$%C58iHOJVd#KA`8F@{OCI?Q?vO=BtX@_uN_D>} zdJQF_BlX&b2#%#eq|voF^&G}b9--j%s@M(~v)9bH_jOhM-tMkIo$4V`t_za6A6<*k zGHKPOJ)f_AfvlxfBsueI({~~aDi1~Xs8$9j%tzm|s}8mYES|xqnydq1XEnE8s8v_= z;))-ZkClgJ?#U*gt9AYTHkh)bHl{cn6;1crVN`^^@(?b-?{q=D*3JQt*Z+EN|8fKO6%uH75cl;J?Euq2IMSJ+C``S&Bv9}AyRA33Ls>dJR;FvrDZ~xi zPDD(Zo5P4g9jr(r-0U2b6jXYfd6MQ7lr;{T7|@LO(}D^;8n0Q0=iJZ5+uN1e=4|Nr z@($R!AF9Fpss^?cK~yOht8?ndsvSVk$V1OEY!%s+m9xj>!X}76zop9@ zdbP9-(*Q$i2Wjwz)2!LcldBO}N}5*z;Qr?QnHH(ki@FB{4z&@e5vNYzc$#t}Yc<{T z`IZ zVy6UCW>{U+tl<5?HHLIy8{0#sP|WZpM9BeRS!a30ULUONQM#HgI7?r%P%Ok_T@P}# z<$K#O&usnfe~MlsDl}IjR2Th>vl4F8Wp@ormotm#K! zKHQL{!5b%CO|jE+Pjl6sHtsVkzF+r;`ST_Nt-wGKn-JL3L%dZXYR| z9wGiV1fm*QU}K~>a_5QA@EGQKa~5Ex%+iP}Y<{1FO?wNy33*T)$x@P7cpZZGm`u?z z8N@P-++l5&f~Wcivv5vBMyVL9rnF|IO~ZtHBrYz&*~ zDZ4~gEKg8J4q%h$mqZ}ZQ-4j14!~CzGndLcx|htJa_&+|eOx@pGju*hMFf|hJiDcQ zel49(qff93Co;&IT+hYSb_Y5`E#Ewgt`bj6s7pB-%)fkc`R}6b7gM3e>%U~b`fdOr z{6cNqd{>Z`iFM2Inpg>KxBQQzTq{MvC&#8?@odq6dc5qPI(XBWLth7`oF9LrN2AK_MBeg z9gHOC-%C=8w?q_fq;9J%N?HsIr;lZ)=cvpH?bUX~kE>hTq@@Y?u!Lcx{%JLzW&b6I z@GlDb*7(tq)z^>BT4wUi7tYCDVoUixqTjd2aL(FY?u{i5SzrSzN@&(VuyL~zA&lht zpRIQGfAg@bXNl#NoD-PU)8D!W+?+-m`?qD%5=#_t`tQwNd`8K=rppT2o#I;nu7YBQ zjLa275zbt&FPk@QJ8O=2JXoc?3W^+%Nm>yj2Rf$df`kYz=BlOlvZiS?*#5N=>*e%) zTW<5!8vkO!^m3@CZP?G8t`%>Cs21k8pdwd>v0G^&pm70|tsgOVNsdx};pSW{=t93JJK#%9ut5Pe`)q7IKJ|v&z&x>4qL5QqOw;ionda zKw?9baZfO~1s({AnNlU*`?;lQ@`}<6rgQnIq*I!x=@)cvJ%wOTpgDx>75O#W(Sb9e z$IimC-1c-0xbI%b2|e%w${~2a%GMO^m@Gnle=ZE*xI-vLP)djSWeu z2ddaEyZWhN503lRKoAD?rB74|JTW_Gi0l!Ms43)oq(K{AY&3U1XPSFv6Tx5g%TPWD z!mW1$@4MpcboOfB#FgxZo%cCn#;%`i}{1vS&l0Z<0gP9Spy5*d) zd{k@%0tlF)To?HcqqNH|`#ZC^LB84lfKjxqJ2ap7MO5Es`}z9!yY2tTmdg(3A%Hwt z-~nRS(bAZoY~7U&MnlxyEki zHGYV0$nN}e(WHKXf=5+SvbCCY73z{I3Nl4)L2F*2hD0lKsN$@e9CRE+$_JY-$e1{1 z#m9%&G)YcsX=H3UK+{AHJrbdmx|ydnrX~*=`8cnG^;#wIKIPKlZBmWCN2lhH>MFn+mvG)$mM)!De@!cwH}(umZXXmSTn za=K6|kCwHVh8vyzLeJK-bbg#zty#u`oU`ZBEmse)(5boXFE3K`zZS18I8rh6g4wS1Qz2zA*-i8=ha^1OP5JKi~_M zTQt*nxz{3$Nt(~g_Ame$`7VqQ)X$o=FiowL#dnGhD3&-BV=6j$%-)e zC}t(y;BsNQTuJoBQzxWQEe~Q?=~nTK>A?Dm(xg(iY7NxHY(&Zxbmmk#1&vz=g!~49 zGw&9W1}v_(Ra|-Pw5u$h_w!FI9NIWLFTV9bnw&~N35q1e zfI49(-*$mzWtj-6)J`Xup^*tE7opX6IgO7FEUSU)o)YUvO|JT@kF{-$Iy!cfIpT?D+XCHgZ2Zub2{GO`$v!cdb_l{a7<+ggq~dhL0|#;%iK>8=~& zzq}G3s8pEfjG$Cw#bSV`w}PRoK8}8rrc$AYQ?S-Vya75pZDqV4@}{w{rOM>$9_DU* z7E^aNzWP@U8A^%y3JbPfz4mk3`HtK+o-V##ZzVL%$-<^$M?xIuWE^||Jzc5h3bIoV z47kn)?fvUD`P0VtY#+xjtKt0_LOpE(jy^LJlC6X0^Q*ysiQNs9#xyG9F*vN3V=c8646iFALy?0k<|o~A`Ia?CDF+Od z`m1K$=euCGvU7j_1pV``q55S5QxcBNGYrbeR8k<>i28t7!9JhC;JN3^A<{1pQ~e}haZ$X`_5nPXIMpC6jP8BRPoj3hBf4& zzGG?^K3OK*7C<-`6`f)~mbTcz{W&pq?R9*T=v2Sr41XQNC1>93y#Kz=?nnLiiSdqg zz&q_m7iZ3Iy!0(UlFvWSp?+ypZNw_GHvrL@n`DzbUzSjq-h>F!oe2&OXXS^|$~GRM zJ#|w`@vRw=%AlwTiyE0I=+;v*X5RW31+j84oa=eF zbXI+z25>Vy=G9El@+Ejs+|&t(CRYPC(NO9pRe_UzqExo@k_D1 z#wsk8s_*~>HzyoZi@R}4?P#sU1*)M^;{&Iw7z|fMU zjUJ$u&JE8RW<5*f62M}fjT=9(@a}K_$7KjhKYHQX_J>%xvf(}}8Dc9K99#$E6raa_t{NhJI++-G~10|cXHfyR`C@a9$0H0%+3uQQ*`wfYtD5;B> zK$te3TS_#PWi{ft>HG&OqJU_{6Un)IRQp)T(^e_~s*oG|;hi`a>%&5q<(?+Kl;m1@ z1@{{2(%d&85IC%8S|Zpa9jMRE-4A1a*}+<;_;7YI`j@1_e`qr7zi@0&%HO=!*RtHiKaknTsgNL%jAM z{C^j3=0hUhFc#T_#=o0vKWkMb+*L7iH_LhUECDGpY%)NH%wJah%$3dyW;^r8=I42~ zA9RK9C$N{6%hjpKsd|3|zWcWo%S3#x+`~5mFHd|pH+t!N&Hqiym7fCH8Y0fF{s0%W zs`2J74J|jom@CC_ascTvCLw{PlL8Zh%xVVTdAIn*yxWT8UI?}{o>Iwe7Y;M6-0_b! z5Pd~vl8b81*W~SKJ|cAVEY!Z6VDw?&$V<_)dC}Q-Y)MAKAhe0v34R!zF6tP-w2$s! z>A6g2N&)FXSsWeih;s|2(a;3rDf6JOe5oyFWApSs?9DOk?AkNqiaF5@Q8cj5hX3s2 zCCV3BIJO#_YTk=2gwM{dP}?9BGs6ir^(0h?sG!!4Am*_2ARaV&#e?6ve+{KSYuV$q z#&Ty4FYgY!lY1OE z`+DEo|N6W*QO8@XG)UviBgGMX70J2nQ%&b7cL#&p(SR*4<3jS@7gmh)F9XF$VqY4) zb9>20AX8)k^9ly2gXP2E1XaIFbw-iVahWUZO@%hW1P?8Qr9r}r(xnnvN%v+8TPOi_ zH(5?+4sU;QGo#W5yS5wI4RVVZfMekod-VY+^il*-Iy@Pe=fl5 zbg~tqZoIQL3t=L@0>1CJWHw}Bc>}sQ(VEOL6Y>Ol5uHc&y(7slC8vIDHlSL5_wFpPDsWF zST3J1l?sK#eFjNYZOpZX0;dhQ_`|UUGm(%tIk!W&?Q1`yk326uTYZ;*w6JDBc`)qT z(mOu(YFXW>@F*QHY6G z1{FjObDu4YIU%fQPGG`u&%iyW1be>31(-_RKptitIyoz>P9(wEojq= z|780oX^-quB>PQ_O{2QbcC>~ax%Nm^HP+dzA2P8mP2bedeGqzftpGlGn`G47n6s`o zf373sYT9a@pTf$NkEaxJFmTy&FxMOHy5;`4`3OM}^)y8g(Iz_uAgym~&+0*-o$)&Z zqyYNYixZZKzWwUrQbPeGF162U?T*>N0l^!?4GNZpZbTDorl11{fD#pN%*v;sR@-y! z*@B>MQq?UtN~nmaERob|nz9;?@08G|9*Eh81l9TRO89)0wzfYS13d?Oh8U=&9h|$ncltZ(To`IA-1=MZcqf7j4ZUnR?1;!atxyF9Hk4%`ncO7qFrTptYjN*jGF>)Gup+y>^NW9lq- zqs;spyX%{kwqFxbSM1J`WOcR)$$=~cveqo6s;Zh&?N1hudD_3-L{QJtRX5BO(K7

?i+Neyri*eQ^OYFy2IerplWRkUql z%qtN2_692LS1CZKPB5hn%P`m4NK**ZRdaSJy~42zULNO>G7o64tKdfFu9z8m)aRNs zy#+ZXLv)h$lP$=iy7NN3fg6^WHQf7yulCY^Ca-t#I~vZH2KSQgOjl}z%V`&NHqN{q zdr~-G%@i1iNtvgnTP|Q!2AX&y7I_jMyHfs`Ek&3ymyH#pxkdQf9Nv%R== z8_E0>gYy@SFb_x-*cE}qtA+e#tPwUw&Yd)D_V~;U-WibISe`qVZ!UE=-ql>EEFPY_ z%0bmJy^=%YQr%~?Rv*SM;)8MS-hBHEa+W}l{*;rSl=x?R8Wff6$IAmmo9;ZSXcR2W z7eaD%DFoD+lWJw4AHpU2inFXEGth>_w|=#X$hF_af1nq>B~ z6_sct2*=J%b;L`yL<_7w=!Z9bV}pscn=R^)dV67ilBzm2%hsN4k7d3MHFS&~R*4Oj zq0^wAgg6Hy=*icaXa=G#-Wcetc3(;Tw;S$wP9vyE3A(sA2GbRS5538tkgn#{!H#v~ zX;F=mZ5TpayY(Q8`6JKpXt;VkgIr^MSy+GP?}PDsLIK4-Sp#&d^-1P5Ot#cSj*7BW! ztG2}C#G2+2WWbW;V13sqUMfmlG{%L2+m?G#6S7LLJUM!=p;yp&Bt?LGQiy*ES}hQy zt0yKz(4Q1c#5BZwbcU%-XLjg$65PDC20DEDeK&6);3M-b*5ttf98?IKtIwFu&1^SE zS?;QP(%})QRW+(`LXQq*-m5fPF}1zPbW8X7fPg~c7TteB915$fXDIja-}M)X|N%7e!N1Jqc$Zi!Yxpa^8G%$bT2x;I^kB z2Hv?XMCa=#^=l>g+`YKicu&HceOWfny-;bc{FR~1P0yGLWrCv*qe8b4A9T0N0q`=| z0X5j=hO@EQM>@MK)v!Rr8g_5NuF)PN7cF#E@X@ zU5;zmEseH&W(LURdp=`e!Y-!iyVcu%N}eR}VFm%h>&&urTF(pvOFaRxF z$6j+jn)$oHN>=cH-Tr@pelS^V6>2h{^;4{~!u&0*ugoWCB)K%>1}9WaX2nt`aX_-ZGUoKHd;05>=yjY z@uq4=Y*(By&-wMac1>Za(S)?p?9V>GDu4K1{eO^zgR#uKjqmz!3>B91c9xC%@xQgp zmg-qLf>){P{H%`wj=(An|Dg|;+B1lMUS)@zb$WOHYR!+@|B8D9ukdqZ``Mx(R5NZ7 zdWEQuR6tftrrE_NL*>6&r^;IgpIYtuR$0d%vd+;eX{8~%d z&;14<-s(~E;T4z0YvaGg zk}!bWcjSE$ROkqvi!%newm*foqd5adngxR0ymsN|_JA63CYp~Pckl|X`Q0I)-{nA;s)53Rt+Bbnhl;m4w zjQiU!?jl3teN;ZgG@PM)5BbYbyto^8==2r*Y5(7mrTLm}Hnx8TMgJJ#|IgpQFcu&h z<|9@;+s@<-3_7>hR)?906fy9R=XM!7HcN*G2{-`Hr1 zGZ6u<{d;LWl|iAU!$hy5xKyKKWf7}R)dh|r`0Vmiya-AT#!Htb=%u=(Nw=w z&vF?%QGda!N0I>Y-B7+(J!7l3IUi`dWqKMxRNmoVwBBL#k4>%6vK4x>7NcWE5>Zf?s1GH|1*tnFH(scO3eK`LT{GSZ_7&;XPV zD3VQ+MqTsZSSe?xZstTCtXf%(cM(HhOFi9V3O0EolPunY0m`Zv6yTbyk_-$FUsr<)H)ejulJFnlC!sNxr*!XhfiRl7>Ay?23 z{WA-%Z12RTbv*ESt61Y4?q4Hy>_d)F^k7aOq&(PAptHFCwGyXNwTxOl#0#e8=&Nv3 z6bap*^UcF#&PSM)MVFNiO$Ef;49Pt5{oMb##si9tW~tgmre&HpT4;m)wPr29m{uda z+M-+UXI#;W@kH!<(_-=4+eSdy58G)1Vadflh7IZlLOHzVY5*Tuq`S?(wzs0Fh=|fus)V*digXo169^#*p|hnE zr1#KlfCK?S0SQe)OMxIr=*31C5J>2u2c#DX9lY6=y}$3A^WEQhp8MSU+<#beEyfya zt|W7U@3%qt%Hkf2SWQzCt!2_jW*M-2WxYy<_KUQ@Vj`GmIm?S5IX)wK zc*Ff_`cBk)Yq_jlC%t>wklj@IcEG`eyA8)TCZ9{|+UC`IWpJxfmW?jAs1jpXm(F=H z?|AGC;4Yi!yW_%M2T^3SumKfI7d#JrPVSX6LAMM@m_6075{EUhN^XhOuDai6R{Zsd zQB#;k{pQ}zZ|mlnEOGX0VdDs)k$5>akghdce_*R!URGc(L!11;XZDx6yYKpYo=qD@MT6eA-n6GGQb-+QPa*l=*tuG2-E`^ociJ- z!6wF%N|dAH{Qzs7Z*R(;e(*37Hk)ese%2hW($%3mkj7Y=K)s|_(w4tUx+e;()V;>0 zLAa&uvBW9VGN0W8Cs#Fs&&aE!QSlZDMNiw2Slvj*dRr6YCpa`jiMxay3I-L`&)&HW z)jx^A^Y&E<<0a-QahWgDw`6DTz<;q5uydyk;%IszW)e+>$MG32C2KfVoAqSe((`*R zB$oBSU@+P4f>iqHwvvkRsj>Rw9}BxLF^bz*i5A}PFM?Phi^#;jm{-m&oaD}!pkyFd zCSQwE3B8m3;JC(`HA{6OC&jx$Rsp~EN0aQLRxx;S-WU%EJDmFZbFZrr-K&C1^!4dA z)2ZS_n)xryp!1`X*{y;>wdT3=e!d3npj;rEih{K0?>t65f(fYINm z#8&`UF>4cPC#^cIkM&aSc;$AgL)h$X0IPcWl0nHef@1d;N8NjNZP=_z*nb7sX=3oA;Bzwz2wAV^KY6XNY(!Kr>jIu|2W!NLAp{ z>XoAb$|O_BKtB{~zBtmtdoG}?=qSdq5UjM>8MvSsZp zp(qU2B^zl%;t>qda88Uqj;h;{T)Rp?+Xhm82I9(2k{7k=6JikgWnH968<_O(y%jKX z(V&`89o&oYNh8+;H_RQk_|JM&jPty6P6$JtQQs>`pbhh0wrvVZ zTlw>;*%u??BKt9o1hj4ryA6FwGHJ4ZQfaj;eBMDG-`evNpe@DolhHeaS=0x%L_2o4-{ zQrO1Y%gxGT^d+3&|BxzD(EA;b$8YiDi#~k6=KhomeQkzkW1`XEql=4>hCx(Kf|l^z ziS^L=ZFM!9PV63uqvT{}M&~+FSvWVA#2Zp)uBNqA|Kj?F*iK~jrgiJ3-3uX_ui3CM zWubA@uxpMBrt~wv3}XJ|j2QZ#%vsc3Gk=||-nCmqJGCFJnr-r1R$1q-y{s?$eQOb+ z4VmDABZfnczxE9O?pg04tm~V<_mFD7ds&&qj${`mvfRMVuRQ}fE?**HP?npWO?I__ z4S>-kPmP{q&r<0>^LS8FK6Co~9Ox_jJ%f{Hwp98oZ;;22Pv-v@Mf(uuSM1Gc-czw7GukDF z!C8(I@S9$K0%^2NSdPXbNH=|E5X%%)y^vZ~3Z{;rjX=3F>3=9fw0k`Gs zW@m->fC^v5%E^Z;`?fsk#V|7K#uCTFh1ATgXhjk)N8AyovKp(npstZd^H1){rQfyo z{^AJ4p)Z+~E=KV*%XdLW=NrtS&RfW4l zK6levUF)>?h^O4kOKB&MHMyo*;R#g(%WiMX#-Do@#=1ScP$XmPfttY!N#pCmG6VLS zLt;udh@W;||@MoCDH0vUC zEZ3ZRtkkzF>G)!j|2jH$WR ze@@{~tnqk)?1J8k#zYEK@{cmtJ88)VDKHH^T+GJ7xnP0IZky;uwUTIB_Yho=$L^2f zt7uvMX$32CtK&z!A(Btx3ybaD*ljz?T3xIT50^2bS95{nQJ0@>RU)L6D%}Y;C>?c2 zuuL@~Q4qVTFju)nM|Oj<`m(Z)emkxAiG|N4p>)xzrv^3jjJmlP5VCvl+{3&N-q>|S1%&Tq2qQXA*bDdg#SrmLJ=~-C- zZ1XqS);=MKNY{$E%9DYzY(zn2y%RiQ_-ew9?s%oyk|hOp|KvyJu@@JGCX3#1@75)~ zWN1X`O8gu^9M6ud6y~Bzl8|OkQU$IT+r|^8rtx>d`mSpI`VCnxSNE^}+GqT}c&RXnle4WX|CGFV`4=P|u~E7BKH( z43}L=lnQUrH>Q%U?7!qT^&X1;yORM%e>}Ohi+=VRsbcy4CR~*;&@)dcmRuG!6jo3v z;8*N=1ddl_Fll_{1O_QTCO|_N{2J3TsExprlOHmIAAlifK%ilii++ehkL=U(IzVc8 z%NNw3t}&+adRt(iP~c>W)xxGT%bSwQ#VKCR(6Yu1ajZl_n$lzK8dT7fSl)pg`6F3df zNo4|`UL+omiN_75>t+~eEOP3+!}OXF89;#rLS-}E)S@urVvn?ToxrLuj@6F>;ES#Y zlo^~-T7>gQeklNF17klmELtUa`UBSbJ-s`VbU>qWmQZ1WAgIqa{DC3mfmzl)+ z=0VQd>zh()mSORw*GDWJ&oUjo|NSTbf1kKLGNoc%9b(^6{Njboat7R!>TcwVFEa9) z2HAd=<$aN9u!EG;ybvNi_Xr!H``f9$Kg|Te^U5$gO~=p72bocy9O3gK z$s1WzvJb~Y4%37M8b>&4G-=-ubk)C4(LAcjk}*G% z-0v3+>p<}*ZCfk(I5+PYA%=tlpt^H`w*ESx>kv_{8t2~V7DFtZowurIr(FM?DdgB1 zF{NW&3X8|D>+z9>lamp1CE zc0auH#hzaZqc(ne+rDOY8#^Rh+4eZSU&D*&zkL_1bev8k^mO!#NcJL~Q(VeEMhV_| zSvU)01+%r{E1GQ1)rJs=gJ#R>V8Q%4u`@tdeLX9@rb@6yYTV?6kVkYI$UKYQbJeWX zHOsi&!^T~}4N)R!_3`~U+rezbfJ7mlT!wln;aR4Xel-33F79!tccd-O3xx`WrJ8nm zd{)F~OwFZ$rJFCEET>STU%U?%s=w=(`dFsid_VYfzVDzPqkO;(q+N+61C? z8eXm5jYiK97X9~CU|QTS;@3^n;ZguRZELD#J0PYkqc16Sgiz2^N~VL!5O)TsMP6Xo zLg{$)i>9Zm^RD*Z-BqzElnvixkPh!<7zaMSXq7Ic#F6QK*!=~F1=s|fZ`C!GIInn6 zmRBW5*x?|Vb5Ym;kRe0@^yvezHp zF(Vm~5O}i&my2ZU%MujW0$-@`K^Zd8!QFd4Ek%nV0YzDLSK8)M3D4$wgWzc#H#UWB z8PtAbV&HLvFeTkavjGdr3_*B<4Oo+m@e+Pd0kHo&*7QwPYr8OhN7u|J#XcK943}!I zGst}}0l8sybS1`BY4D9nwPNiu#zUrv5!~RWJ4C3LK#OqEK1Sh82Ncq_%LRJ-Z66@k}SoOaeOiTbt&NLK4#1 zs-#Sgu&FQ}0&2aD$*bt;k)Ap4%X+KZP0o*otL;aHKLF#18?|R_E1DGbWr~6vW}gRz z;4+wNdXlee2Fxv8pG=y0aky;8(o#Tzhz&8iNw(zz(B#2^L0fAHLvp!3`{W*%&`qP$%ygQHq^x^J?2Yf$}J)rh za-6gdPSVHN#kRB;J=P(P$Ac{SD*=4;4Z!Y4e%v47QmNGh+602dVK+X;M!}*&qe)h{ z>W#PlmsW9wa~K%x8_^GN5R%> zMh0!<+E}(rufHzECSPu~wdiz?8ka9S?0t!%?hQ+^ZV68Ewm13lv~%Ik=?7#ofd_Wr z7v?M>-On0Ef2UAn($OI-bZ%{BF*LxQyxPEpE>Sn=ZHuMApaNXKX8T06jzh{gOPT~V5J^T$Avyu$H}+ZkK?%wlgNEZ1{fWiWbx zfgo|*f%CF;r)i}JzWOf(c$iRlk z#(s`bX8(JkZTw#~gJ;f`IG3r}$8Tq=dOtC(&Y!)A%pfyefc~`mzWKRkrgtfBxACAP z7_$;mrT-Tf?Pam44D~NxhZu-mfq%9C-?go@TJt5-wFnk!mB=$7Ccjt^#kisC$ zs`NA2YuITfNe6s4I?q@RI1=oo)oz}ln;~&cOHPEewSKDH%)BsGx0wiw245e4?z>w3 zJpSU;dyMCNi}O%*mdG`FLw8F-L86?llkk_>sbP4_g&P1}^5QgVIG?x9{?=Xcpkk&2 za-z>aJTaLoCMcIwj%g*qZzOW#=2#Wzj<3677rtKdDlx^+?{nb9220-QBa5Lq>)w1z zM0tw_1Li(qqx`$@O}#<680=WqYp0zPcyWm4tOU>bf*7kQFSlZg8DZQ!)xhCVWrjI1 zv7b?bU{u#*YRxuxAul1}{D*0mi!pFnj%8RVaN2Fdm&%f})Z0#lDc$qGMgh-OY`$Ob zJmyk?59p#gst!pH`z|H(SRenbN$!DM8?m5Ev0kUa@m@DyyIC*6u`QV;K@XojL`hoF z^SJFuO*iM4VaZ667Y0H+wbbHpe)H^*kqKiSMu;#(VhM97clwfBXp(JaGOe9RP>%Bp z(FHjA*5cCCwl?>^@zXG%G+hoLZ6q<$@`ahGa)$jvH|}CfUiuZC z-lHgOJO4s0{*XK9N&}I>w9cW*-Fs-wzE@7`)43J7CHTSESou0(QK2{h7F!5>p9f=C z*1k#dZRH_s-Wtm3{@l9b*8;k%S73~9Vk=JW?&BZwr(#rm7y){5ovihv6z)Oe-u1Hk)kcHMI@aPqy%0dI%(ZotEKI|@ zWNQ2cgB5)&uiDjI89P{rY23A3V`dIur?B`utI|j;k(GzBa=#p3OUv2!s|?Y&{%4ih ze4(t}u6eJNLT3`t9C`+^RDMY#2pmeUNu6wskEP{UZ_QvUJav%;lZ0Kr)PSIL*{7^> z*ZktUG>7>w_UCIL0^$X3;|JU#s^_Slp0CYGCkuULigg(@Vk$@tmo9upZgUOH`_uAd z2O)fPd)Qar$3AZvIUw6Xwa-FV5w$)pMMy{*lfDoh&(oT-H+=SMyPc|n9O6o%nr!nHg0l_(Z(nVZ4A0WpSijcL752)nCXQ%>F zAfP~IX$61_X_!88AA1=$zo&M4yt+))t+9PgJL*+MTACk+en7^!} zxRLm5cw`fJYWB?Sbw*Jk_tPmlp^vTGbAiTn- zD!m<0T{4qhRXF@5+Sf{F`KRVo!(lHa_uO%LNm za_(OaceRvd7a%1*Z#t%|Kbm05!Qv7G3u3vPUwCN;jM%RX$=10jdm%U60=w~|jEr|uB6l$1PoZQA{JDdGs#ssTE7U;6=#WZ~pi z8bc^IRyRG8d_n-EB^0E5X9BWXGqSK@GLf+Cm78VPisS6qCnNJ^ZbrOi9R-9wYwS=i`?Zs~;LXNs|>fCw^_2Nfku{B^wLW+RFtAp`OeCE|1Vg8BC-H1?p zaO9xZ_3B@58d+ra5+m6pl>?7F4EsAF^F0THiw1T?>bj!0!MLu^;%4}6mWWxTpzc`*>wZUiR&cTc@AdgSySDuk?Pb!y=-`)GPd&f?$y_WuDzl@#`E?{ zf96HOnm@u|uz474Iv4nc$I_<9l??Mo(js(IQR|+Kl%SEQ+7>o~Xy4nEdyJJ`Ypf5l z#VfuCYV_%~qP;bY-vfoJB`{cUt4p2RMUL6^ncEYoJ$f{)HZJg&Sm0!JJ7?w0!@dpv z17)P6QsX+7*v0vt+)Pa*_CRI-enT`I%u}PyV{UUoa9Hi3tZZr3D}~s z5zq)wKo#j3iC|nXhY(FZZOsN9@V{vx6sc!qCNffJo^GZk%zMw^#+u`A zla8EHRbH3eI8FquZJH*-pbFmA8GzoRWHf#Ic&AUu^>o*gUStkRF<)9se%T+*DES>Z z`C$F!0I*0#E;c(F)I+`_kCgX<u&CE)Svszcq?Fu=eye-@}PE0F7gSoT8^T79IAZwmPDDyetnd- zuyXyw#UbkUT9sdL^Wd7TjkAlGHW;=n(WUE50XnIFYe@FC#)253!E>44BWkjxF4Z2% zXD0hl2{P_genJUwxm(zV24}9@IG#EfEUVJF&hE3}Qs`5<2I=mlEO$3fLFw0`8(5JQ z7`r#{kH$ggD~HX^gSzR71_cp45{t@23%q=wdG*woX$UZng0uSXSuzDHOX2@hmMo6! zbP!e+T7U9)`@*0C|Hh%=UMl;Oc)-`jxyh}W(RbTW>2aMm%!!4%F98W-C|V+nH!W}` zNH4H0`oMAfpkd-d>@h$>@>4|AW;oI>5EsY07XSA@74EN{wad;p+EiQcpHiR zME$%mUs!r;-xrs3`*Ltsu3~d>J3vcJ9|mPST$Lc(yfL|-wcIHrmT=aB^g@P`&_lr2 zZMCTKH`MiY23?;_Lz1yERJVV30ykjN1&^JxN0tQmRT{RCf?hUssvqI^Pew5+Q5i3h zVCM^A>5FHrbTdDWQhkB0u1T?!^Y<`$4^oMC0Bk~0T*HKnaNRrS!di<i+AAxs+Ca&Jn!Ju(OR(2~5gew5G$kaq^=B0JYpi*x zm^JBHN&0@$Mif3Xgp_-sc~c;akn+h00-4uMSsIw~lf`8(M}WG;()rdAsjU^jCRJfk z`)c2WPUf##THSNi=1Pc^y&DOZcM)W`in&L^kiMzFsUR~{AK}tGBnc+1L2(gFCf$=R z2H=F=tExe%Qw!rwV{#lS*4TNPRvk@7sfP7MTUUN^q;>r=&%IV!1x@1mWl5G1)auRk zhqF8uO2Y|o0BHm`l~*Yhdo13eAQiH3u*gE)mnqdQLix*6JlxiTX5+#HJ;ICIf=0V5 zZ41N5%Qhp;fW5xGW^-+5^@R;}_;mgB?Ll=0?{t3lT=U;`3}<&YsmM4$@Dsi7pxxTP zK)d2`0NRK5>Lk_3-0C@k3N7YU1ha4@<&UaVMkR)x@afAZq<+`U3H4d|06J-tkGf&& zUU<({6BEdK*G}%XZxJFNtPX- zj*5N}9{m;HrU*U#b@Q?9uqgyaeuX|nDU8-!|Ol#(gW%J(L zt<213n$>**QvC8Y71PMK{ObP$LYbB+I@ZjDAN&3qKV5$j%g*8Je=9bHX z`BPeI(w?L9pT^Qp_{&+`t2$JH>E3x*k`A~+f>sIYky@Wnr&jIy;9$yUvbc)9OzB!a z`@NX=F|CNEu|r?VKlfS)n_P3w#U-)@dC9a+mcCFP8%7$F!(PtUI|LHtTu4+^0UnwWw08i--2@|G&cVkUj6!ghaPA;KUII2%el6D*qSZ(-kFhM z#5g=c0atX4E}JXwUtSL$=JKtB4fDrz$@wqyb5}ww_`Ek?3_Hg_ljePrqI1t#h$jHP zF*)EntVh@e!juEE=u&_-rtDJjQnt?Rfdb z*FSsx2yPiDieQp|zGh)cS99+=s)mAW4)2~c_r~hX<1fp^Af(u-+G+7;L`-{<0ws`j zY?4~*R$gd@=DPk&Q_fm47-o^1F{(8J?a&wwhbvWpIG(N^?%@f+(CTwyla17Mvh9nz zx?5XO;h(-%&PFUo__^9TbKX^iPqQJ4x(b-drVHMKoFK! z#X8ZJg8WlKjJ51sAY2l$di7@RY@1)roT)QMg8jqi{JZKm*_ytW5 zH~9y4QmaEmBmDy#BK}e?Z|(aTLD$^3y6KsR*|_EXb>JYe>G38)(vCw-C3Dsy0Ho{5 z>pOk_u;H&@`>!n<_5PMV=4;@=-NVAwhgMR7(_i=Ar-O9sc)NO{7($b0RORpgwEqXG ze!C@Uz%Hh$OvdLaVl$gTv9z3VyUHnyhhvd#UfyS){|scX?%isS`qr0vr%aRCtUVv= zQcyuj(d%HJ&j*Tx_t;p+vZLJ5>F6tXcuclP4Td#$Jev-Bq$cqLx3F`S>m>$Z@ihy#BYKII7xfx{{HKsR;C2hZ-2jo5WEZb6h7R^g`h^sGp%)C@ z>JjV}RK{<&j)EJ`yJ{9c{Z++sTrRKwpwsu5%RFIyfdwD$pDC_HWv)b0z=UvXUTHlQ zj`9O&=Y!C!FSXTcc?=CjR(UqNno>?lM@1#APqPab6p@HXp?N$vF#%@q(K`{w_VJ}) z0$(NSNrhEU!>^Oh02yQ@iPy)hv4Nn+^56w`NBHie`FWoODRT4T>}7raEUh902_FMn z9UFC`vgWP_pHt6g)N_hRZpOJ;^6wG7@c4jAF3@*3@(edfNw+;*=HdO@!+6;^fN&K< z-HAZnaW^nspQ3YArPJ1R_$`}X0YAJX8gMhP4$4)vo-KTVI(2{um-cv8K#GmAow6ICZ){CwF~2DZqg z3)a~*uHZ#j#=WUBcOt&HZaD0Gg^-&fZ-;W=M^CL{GI^ti>H&^#>kZeRtYzf$ zCguazM+@(DNtxZ#6M$tgeOw|xY|TCFaA5vD;~15D_9ZQD`LD7a3sD!A@)MDiV3&{k zjPE`4ukW3)!bJlrKbc0Y-7QS5gO&;3D{RX zaNR)-81`Y}n%0NqctZGQ#oo~(SV%PoBky@#Z{6i(p_jWX7z~D7tHt`HDbU(7UH$IJ z8Z=3ACfK{$&o03doG;gGCFdZ7*l;lR0Qz$?o-}g_Kl5WFJxUAZLh7dyIt@}xZ}1V# z)A(PI7wdP}}oJUXjkm$czL!u~Z!0<5h%xxST%|E0@jA zhP*(4oNC1YB(BFg-Fz)q4*2lK|gT^d%}^mUjB(o##GzjHCrS zyk$#=9}IiZJ`b~)yp|TkR*=51tkpHkrH;eII~Js{QXs?-vwQ)KIfO5#*?xjEK2l#< zmIHvOw3qMih>^#G*H7*1WF>bb>pgd?Y-xB>iPg?;~5cEa_;Ji zP=x_6AxMy_wBTTXUMVR_Ma?F!^aBKJ)XWMNNFVT;nF6tvUI{u^-#aRkI$5bbi*OP$ z^Vg+KY>czB2f?{dj7zaA<^M=wZKTDa!j>-W;MOiXH&4y|dRK5$plP%We5(i9Jnk+)(R5wzHkso0x6IGx;R#gik`S?cDTJzGO4Yh zjvAO#{`rpC2E5vTS4(Ges6ub4sKz=#jZ;@wSF^VC#cN2J)^aUhOo5gX$s;OpHEB_E zuvRdyaL3y~x=p`U%nD~8v?WIL^7NWac08>iUUk`c7CAus+}G;&`O_jTtI+MMS)Ig1 zYYA|Wm+ACLzieLpq* zt+?C}+f?5Sw8q1{VEhc=X;6LS)5v}~|57dd8A`KXk>{rQ7}I*ahwlz z-1Mka*^O+3>{G;DPV!XLglc09p-M@$_Y(Z%oUeZRikxOa%spLHVXHmcuIhVx+`Ur8 zhizczBAcoVyXleBHZzZ&H%2tG2qWqu4Hnfvkey3*bwhx~MT?trdB}>?3=`=16cFP3 zs#IemDA(Xsy+sIHErlbUYtZ7Oi-pF)ZS zMa6$AEVRDvzoBpGtovnQ${|p*rLqm!&>UP^)FgG$`Z6NyykKE|x8&#UGR2)dv!tSs zPO9a4*w|VNEF-7FIJeTZ1tN?T-Rza%s;1_y%`SFwuRlZV1JzCMS^5cs}UG1*nY1Ws`0u@-TNZ?-cH;wV5Grr2NDx4Po zh$|r70sB_C;WbLG`xd}9e)d7G^b1z>@?_qk9{x+>Ym3+OE`D~f7OK-gI*3WWe z?c4il4fgj+Hoqd|DflK^uGYHhWYhCJ7SG2xF6ku3`J$%gVS^cC{kB+>^hCR;`Pzfr z1uKlH>FjS`0lh7za0D6z6S&q1sqKqp&OCJNI6mb)-to1C|BQ2!ctE!^l5C7*Q4x&P zD}ie%PGc~}c+pD7Oj(Pwua7p`JT0jd4*EN1aXYuayo9-4mXUwD(0vJ6*RFEe_~!5O zKV$mH(Y|j?4Ks5SnS5-AQ~UZ-(X1xlTQ?g3;OxJ$B<@aiQFg$P;wYu$c00>dG#uQ(av8N<9%6lIjMdkpRCG16@4ZW zaheAU&-isrMaoJ|JgfnP4c2P8aQQTuTgkBK?_)@tnou)_;Af`cR$PLZDi8LCUzXKB z2iF?{_81WhyK2_zWdT!T2Cg$RAGE76FErbnEwi|>DEF%TgP?6%6SrL23PzSi;{;$({|AuL>+jU#qoI!6Vj-Z?)+#g-j=v zxfDL^a%q<(b;NK_%d!g4wNoqCJXfdiY#;SY{Yopfz(526#X0D&+uT3PMk&4?8xpXJ zR(v_w9^MyOZ#Xv}o@_cSlKYJ*wH%*qX|Q*f)g*(2tUz04g1f2oY-@2D&-d-yLx(*P z0LkB0=YZDFlB{M+3pXJY)45kk3G**Bp^R6i#GGV|3hg$FmWtIgN(W3MalOc7N@KiR zDpW7;c+)2`?OwUW@)_fNN^)y74FvSaNV=Ho}jvAuKDD6GL{=8_s2^sr@LzV znlWyHo+A||ikkUG!(E$P={xwO9#EsH1vXtsm3||P?GBLyP2S1-a9GUetkSMf(eEmO z?u$d8o?YBPZe-h(GgCx5l zI}{h&KKYKUk_((TkwFF-6?0WDQ~Ad}LEHV00>=~A@OYu1vt-6sV4|AcB}BIH_pV7T zuDrWu%T*z$h`=Yr*}Is6brW6l(>uN-C)5X#mjj+(%X`R^g}64C0B^z%P5v7=h{OPyBX!wDaL(nDjWnrn#zQl2R6jshvFW4 zqZdw~#FsOOZ{w9thnmuy?dM> zcfhQ0JuXskbv3B~E>hz!3MA+5V``%Y_XaYr-lU6!IThLX)yp|HLyF%27gX5aHTc@v zKTu)0s$OhyQkOiThD_^{lBynmD_H%Ap8m;h6o{)tg|t)#eh9tWS@>nPdo!BlFO}|? zk{^bDQMuip)CVp+oSAvEFB*4ZiEPy@5gMe{U9pn!VnTVoo$2VSAErP4{*&A|3i3NL zE*ITCv(2F9PYhPc&SztA&+!Jk|1dnfb|~;r2T%9$%+9q4{D z=fw|U$5apfW&7LB&y9Z>oRVLGj7k3%@!jBV{lm;d&t#0bt_7>(KNv5eYU ztLwj$)kd$=duSYwJ1p@XT3u@$I#a4~a+VH?Zfwv@{MooYn3U{)Q)Uc z?AL=AW4A#cKKiHK7?_$ket;pbZ@8*Emg?&W6jf;#*hTS!`QCqpLgxieekfN7oYwLj zWMXB9vjF{{M=5FEXJ_YNet8tlWOd>>&FW+5L?xCsm7*8_2K4H%r^Qt+gZsh73qgqm z-d4DFPAoq2(7Hj_tIOtbd%y3&{wP2S!D5@-=3>x)Zknjh-=bN^gU(o;@r_(6<%MmZ zw_;Pcv-7Ow@s!=z!jHRRP%%86D=N2PNET}rm6Bk+$SPexpCP&JT#3ja*{hgZv2dXB zIQiGC)xftgz`J!Dj4W9!EJOn8^$|fbBtrsgJoO@edMUfYU8!hG(s@T^C*R7MVEppc zYkhap2W}m(Ib`0tYS^A5NpU5`BIzLeO}BLcNe(rmJBWzLwklTg9sc?*I6rSlhy7R; zI^r0zL7&Yg#;Ve?mmz$ldep^X&=TZWt)zuy&m)Se!;jvxGm>Dn>J^fQ1(x5K*?W>G z^qC|}Jc=PJ^395Sv-#)dxnmQX_HEwuK_xcwQVl3f0&UdmAJ7goZ|>} z_Mc<9<$^CDHV_|BgQZN zdEb!NPr5x!tjNNIF(lQYGancQ??je^uDiWbE$TkqiQ*|D+ys@wB(?I@)tRaB!N+ge za$HT-oRdl6e@5S=?z1+s#=0b_EJvsDwm)}vVaOHHuq0-Fw{m%V8$DzMekN9t5_;;w zY52v~$=0Lc_)+WGd0IUj;u0fz7)Ft;PV`ClD!oHhpyBgN7=Rf^}?60}Uc}72}{w;nV9cH*b;IhYDG>}g}J@_C0k@02p5##cG z|K;*oLtrY`ij#^QBd$73*l^xg@3Ajh0n|HWl9doa*`X}I|J@y-0#Yc;_Z~8;X+jTPtHSG3zNNct((8$+y4D=$i*YRfF76u?@cM5pDFyO-QG7Xvwnl= z_uoB!*#Co%{%=re=aVm6UT*X2nDjAnrTEmA-V}rG226Ih|9lKf=>XwryIp3YMh_5_-j|MnjKY^!BEx1k_(2F4#yrkJFOb?Ut6MYmXd{TCu&M3{QQ$Y5t;*`9+LS z3fSVQ8`nG58$$5cYpk)CacycJr$Yp7h^cjjoJlBvmZ!yVIW?Z^LH6;9-awshxDM|W8=*oYdOn#a zFNFb`fq4-cGz&|I6NEatj4N|Yoqa7Uo%&+GrlHnBk(X3`({E{#w#cS-5tE2Zhd}yO33@M(D+Q#qiXuKL6(E0MTY#P8*rKPf|~~Z<){^-1YgWU z-QJxp>6Gs?LnWb7f9j%Y>Fh3VW?}5CA6dqi zCEqIH+MgDcb7>Wh0uU%QJ*W<9!H#ciuCRrpWp4-t2TUNkSQYrLCO9ZoV=LVd{|QGg zQGdGje_G#TuonG)88PuPBBq5(I|gyi;eSTT;h@)-Pn24@t9`nn@i!nDf}4eZ7*zE9 zD;Aq+T=`-B9p5}jy*hsQviHPxr|9jrrDrP-&Qz`j&m?Bm=gxmeBJ*&+{l_=ck-5vu zDLb=knZiHA8org3Hqv)-i2rAS4HMICiOcTfAe+71kTO>->-!f}+epZa^OBKDt=VL=(*Pc= z+%Ik%W^=94rKm)2pBl3YApKIOyX-V*T{E46bOg7!`*uknT}_BF^fL$D@(fA{a&0Xz zgh4CWkx8yXFwbEny>E#o^$*jmpYo9g-sc-j*KON4#k|U$UYS2OM-k6hiF(g5;|J8} zU{eMM5|4))zyeoA(sQWu^dMsK0RA$2M8nD4j67MWZqznEqY_d0a&p0QS2aTz+4vTw zbRG94_Z=%h5cnQ)y9NeBCPGA~)-s8hlA-mbX165tF^2^o&`2l*sZ~Vu+Unb8)~dJd z3{F)jt<$PpHdd2hEc%3)SL-;R>F+^T-zb&kMpeg+MK|?TV_#kk4lrJO;OcHP>>6*_ zF9yK*SG9CCZhVbFVN*by_(>HL<+Jxf`**~5<|%{b**y}3d&8{zh?sd$aMz-A(2F3) z(w$4Xt>{Lh;?m;!UQr59L$Ed>Ghh_@Mc)z3+1pmCI8^~~catLCJHD`{p~Yw0^e|eW zT8)G_DT_A~-S1UO*5b<*UBPcOrj@7F?4@gf%!l1}iY(;S2V8QpavnqrHe!^Qy;}26 z=KwXD%y6UuReU9t>iK%^s5_>&+^vO=Pgr}H0E2T&NB5?!-I8fVm@Veap_PtI=Hjr} zAV*?+HFfcvb{$FWM1XCx$T{?Wd>6#XX9xM!C%Z~BwM$=MTdTU*;|W$D=SrcQ>4?Lt zrjp7K_7NCNy1X?0B@D#r_e`lbp|_X$lDl!br$-Ss5792hhFPj*Lyfbyrd1U$KNn9& zXiyZ5(O-(oL`sSc8h6M+fs-F=_SZB_{nVWIuhgx8ny#LyaXx-+BgVPM_ztN9<7?b* zygfb6#_C`Eq3Fb(Vf8Yc(*>A^pD|wS7s2?%Zaz;=3_Qz%GM^K*^zQAGnRBaVF$v2)&Xqe}?L$Ojr|fdCzB+_VlHJrori-qtWng;2}<1a*MGDybkA#I%EDzsIb?b$%|H-bl69wuO=HP*bBP7%USPn+`Bhg44%bw zZVlAiC1&Rw(qj@RPywf|JjXndUGJ>-j7}H`f=8P!S)XIkcp z;)w^*YKWm8I@2+ZAIAThzy0s&Tc+P4enuNVT_61~SfyJ(u}YWW14aLeRT2R%3J(;v z(TBcrTA{!}IvXxt`+j?k=~cWPD(+ApCPcyOf9vd_;_i`_!VKTb9qL&I$bZ^;)KFWH z`h6x3{~zYeLvFk!Jv-gsn4XpB{?o1p4T_NI#y5|+oc(vV{~;IPF~73>nO1T0>?QQg z?Nxv3yhDkhx0!%Alk{ez_M)vEa5|Ha#Thc&gWd!upLP%H}(P?}1W zu0TLa&?UVKB!tdVKp^xII;f~9U_j7-v>+`Z2?^3WEKzC%1QL)gB|s92l+e+erK|1p z?fu;Ep6A??KW4^E#vF4dPsYqU-uG9AkS(5ed#q5_W@y{fv5gL5C-r24Uay47ck4uP9sn)RagDaaTs!@~rIK|6h461@*phX0(2?=?Z55j}yoH z+y;V~mhl-_swcur(+pBGNk0|oqid2{Z+(n~qd4L9{KxlpWqH>5U2T1Cg03wW%W4C^ z6lfWDUJ`B4K^dv%CDZI%I_o*uf7&heh<3{`x0Sg<*Z}$u;=n6me$8U5OB?=g%lj7y z@ppMqk+N-JR~Q5XgAy?pEgCd6F*p@z%uT2sm&uUmW%>;am=9Kl29KAjl=`@Y$cD0{ zo1s$3WCbIU3|#ZHHq^T)Ik_?22KIFDbb=+`E~dxO-EadP*x~Zp=Fx-e<>9EL3Z5^) zS&w?2?lWTucnW+Ss2n2kseC2hG*XSDI7xw?DmC~i^La^mV?{2OvJ-CTE@^#P+KiOO z)e$QxP>rY8J4$BxlO(SBw;D#^q*`i;y@7?5gRP(F?h`e_HZ%T0+qYKFAd0s18n^rQ z$QnwlB``&~{K1ly%sQ=afb4pVKhhmVnaMupo>DTV-PWDQEprmP9Q4GVFfiMQP%aU9 zp#+-o0c!c@vI>icOCNW{!4?z5^ymq9#Kn>knOLr~#zw*o{!;sUzAJtV6>8*Hm$1-L z!tl`K)fYF8d;{E|#StJ;2{Aw`?5iwMFr)KwfBIYsG+96<*}JC0=jp!U_!IJQ+=OUz zzmI462miAC2m*r30;Qaqpsoe)vaGSerRhpV9n4ujQjlLm-x)cKK*Y7cs-y!KefEWu z+T+DZiBF2=aQW}ogzuf4kj#A3P4ymAXw_GIP6M69!quhBJYILntHE87^R1H1T*dg zdI1sIXX|$Hj6x7*U6d{>J}gI6V4qO5|VSxllz1l}t#S z8V1doSrOJZ*#$}^=Y^31Hm@K_!Y1!ch!nb*@*kbH1f0Ljg_`qbIJ)l+6!isLp;Ooj z#R3@TrFgEo@~+mMnp>yseDq`wE}CbUkdtD*jC@O3w&x@&sK^w*?I- z-O}Vbn-FPag4VWq>7Z9V`K~S5+SK&0y2MHKtD0y0ojr9bKr>pui0;Z{>jmWxp)#As zVX~ASq5+E;w$I-l`_f2^^N^#>LYa|`VpccJQ1qUuu?g*bp9za#C;dfcde&+>MOjzc zGg(=TzpaAB9g2%Hg{dJE-mfEar9f4dI`qfI$r2pxrQ*6Ro#PHS^~QqT?1L6BeFPU+ z*w^$+it7-^#>eJR_Ri>YxzS=RC;Zy@La!opDy}hwe7_#L8c6M&uFxWLtv|Q8HF9U3 z#!u0XG^BvSOFqRFDtcju_leWGA>|(7+G?1TPldy;M!K9jadM}KlJQd(PTT{57KDb4 z-AB(-1EdVz6zgaDnOJSEn$%l=_Q?9RLWPR41(v$7jf`#EhXH%|5OQB^DaKJwJ}#T;ha-!d*PfI zXZYpJGE#zZWt%5Ul2_}ku^lA=9V%cZ;?b4HENoQNnx=|?Ee(NZ_dx>a=wks{TeRW9 za8kmNx2$Q)Tc_@B>6JSPoLYo^v>f8Z5Rj$hrDEJkcMLl<%d0f)v{KAEC@b{>rHS9S$l-i_+TvCAB*9(*yHG z%+TQr=`5m?-g)&AXzB6mzZV;yE4Ui=v4r{N0jetZe@wt1QwnfHSJ$@jXUjcCc ze~id#+KpV=xHxg`1@-2ynj3$HSiil?uJC$C{!_5+z+BCKH+BQFM^R~Ge;2)XRC~N} z#xh(h8)yQg&Fw`rm0OsubuDJ^XuQ~yyR&zY{ZrEcz;Z0hI$5sGY=ZNvhGPO6HNcYb zWxWAt-jMt4hT7DCVZ< z_I?vdOD(4-RG`cK<@5Zn!p0XPFWmY|i~4Nt#*2jk>dxXeLQeZ~L;fzXCaNyV$z<%b zt6&xGI^P8mM1m4Nlgbi4XQgarUP%kE>Rb0_MH_GRdv};%PxxP#b=pCisYFP+ZQLI3 zC~~*V3m$078>coUT=^(=Vlt}15*xpGO{v;NVV)VI47XqGc&LeRD6f;xV>ReT#K0-E!oDs z{~S`8f*n9WX5xp)dK;P|kt^^xWuTP3={JBJs@^rV=>f}}l6uC;|H~@6ySOg40eH~G zaYr;Gu_GE7TroH-#ho%~GVjuAp>EbO9GA&;g4}Q4a3;VNL$Fz-_F#i{>ePF~8gIIk zS2fgnw;XH0^TZKtdi96Jrzi^mDYbzSJnfiQ`%gw2&5| z(j#Y|&~G0~TsB}!)(?h|>K6wp)F9eB{njqTW!@Qt*r@K#lVvfUB!!6Qv}1gzL24W=rwxgzd3dTCtc2w(cOJ2jwx#zhijCP} z>M>1U1?Udd^QkOF@#~+S^~u;Jw;Ahg59kZjw(3xgsSZW1X->arL-VA)i>)NRmEAL* z<*KY$U=qIp?DTr1F~h8z?=|RVojrslM$bz|hxSnP)>q+#7SGRWNL7N4Q)(nE~`N#!H&vh#`9U5IwiX6h{Yz*^gR&W32~~<0Yb1z zqTTd0Mzi>eOp3L!-ZQ_p^ghS+$Mhbf3i!`$K>0A*uR*KIq<0cj;h7O<<{DwjP!UHf z#N7^B_wKBSi|t|W;7Q_c?5&&#E6VX9M7v(aPW7DrY{I=jOo#5Nmw4$=tw) zWM#e&FLrfr^HV+-l!OchE(|SY*TZZDM0-)T$(KKjHbQm1-wTJ)Gz4TT)L0nn90I9my~qf#`L=-W=#Io zack52{i%a7p22gS>=pYGZe=Z=P44S~K_hv}8OP^#fz4_$A(LjIa-y1$l5sV|((tgo zHo+{M$8i~5(ZiKpM)Yt{OcZ=^9sC&o0#|TUjB>Fy?|#z(}fRog^Z!6?-{F2>gMX< zr~yf7OoOx)_z5u!DR*S+v`7?N9PSn4^W%N9T)C92>=w*A-aUM_fn!s3NPK>Drme{5 z#5BG=e)z^l9Y(P$H?G6C-z4c`wbthCo`9<1SKwgWROa^2T17<+Ja)gVvDCAsG&AeAcnlo1Nnhk_LgpaUwTlEANIE60M4AX`jl>q8C~**1nC?Eh*(3FZYKceT6I2*mdaW+< zbtX{3p@RO5C=j znG(AcL8f_Tuw%=c=WH4l&-!`iO6H=&6(~rDG8DR^SZZlG{7Zt>;)lKwx7GCz)UZWD}`zpvp?*p?fM1i0jNe$uXy+e0c2F3V#7%!eo>Lr`1lIT%u zC7kU~pIUQ4$GFPbwRvIbw4&5n%RmqFPn&6y5yxLSgxK9SsR|iFntMq2=_pBPxZ%Pf8z`OdzZvEqjn@}>c_<4kUaz^xA$4=0#hd+24kFk#mLdQIS1v+d$4S{Cu5 zD7@r-R(;8{m1bA57q`!JC{ZNtBI(;AF*!^AOk9;@(=nr_fmNe#0M&#vR)U20w;y+z z->Dk`iGfcZu|@X8pGkL~HgdyTFPOlL&) zEfyKEXVCOze6y=8$J_plg&U z&7-X>@xiUMpZ?vTy0z~%ZB9$>#X-ORxIT&6Ue%PaeN%0{9jiQj{$=o>YzPn>k4REb zfl*9!jKqb-?%x0L?Or~9*s!wL`7F!h!{ws$nwLZSWfq69^F~*eoL(Dc^EF_8_It-e zKe_|%Ulh(GOt}5BXtZ4zb{bT-$Ogh@15f`lw?Du#o;dcSmu6+t0%+Xt!kR=b_3vz> zapVO{mZBcTyK2zR#j4O<3>y}pH$P9D>?R+RclHHSps6k(X2;mdez zY?(b&QEDsZWRtH82A8CPAx3j8Z84^U)^tNcGQ&MN9UkFtc6IIMSDB=MmWW3ADs`d* z_baEXt3|7xUY2SNy>UD25o@iP%EP0|({cxuhw0%JOlDG_e1Zfj&4~w37+PHAv`5X24!?5h2~Ul`gJZt&O50jK6Hh> zaq#wAf2l;9P5jbZ?Hz#&aT{ZooPryfrW|SD)&YKkDyy5FeE~Cev4yJrT1JwooZvJo zkry5H-=bKgKCh_p<9Z?J@k?OEsF#0!@NpIF=(26&sBnG!K*$y@S4fhjM*VZi0vDY3 z$&K+}>pJ>_bA2=w`YA$7)N30UA<;b9lwEC#2V*|lya-`=$S5pz11Co!o4a~{159lm zW37Vp0?~Qt*%+5xjGu+XOD%}0I1mWDVr;DN8g*U>iGOl+aSZJ6L86*K8IH-Vp%@0H z^GGg>_^F+M1$nPJMkIM%2&yw2lCyaKiQ($K5SOW(WShMostZps2qG!}tS6KbYJK)< zt-fHjhZP0lzm;_I-FiyCvw%Vo^YW-G+f$-xxQa;nFGBr0Vr3RZ#*xm%d3RPJfM0y2 zJK~XlHWYT*!@R^Bd8UR(if`=IXyz9Jwe7tcp-R^?vpm5go#~NI0NY7Q-ZnGhj6|FF zL(FX9GScq6P~F%h&SMSFa<@Y<^*7m_QUjsVaHvvy^Ws`!gJyy;H30tz{Nji z$L~)EJc8wSWvjR2BGy%por+v?*|7IpJG(s}c;;N#*Tsh4p{##cxg0Q%bJytdb|g-` z>RUIDA2uzsO=VQgDlUH9@MDDKtw_g8S;~A(;*lBcUX_HS*@f>8r!3owddYYhfZI&< zXy+Trbl#WBl(8;iuZ`NO?o0L}8;(8|(%Q_5}wjOuc z4*)}P@Q9-h#rp~N*S7{*M@%Y9pk$>nfx~^tp#Aw8wdmr;Ox3QtU5o3nhoka{PE?l~ z5@bw-49xO|)In81x|8JaiUkWC)R$^liQ7nte(y_Mvr71;3>}Ou{24aGlF* zEDbTm(zt`KNr_f}%5d(qY}qwvBAB;-x}3osqEmwVvM}Cr{rrwcC(IC?l}q`kGw)L9 zuY0S-A1AreyxaL?mJ!vHEoVfc1o{Qc2lGfwbykVLuz{&c?(o_IQAahs3Yc|!{7jPz zM>QhA`m;e@JaRm>x%Khvce!HXo9MRocFk);SlJ z523FJ5^(N2opH3~XX!|ZsRxC#D)eO8YraMy?I$Y(qo<7(6aD2?UTp?An#aE%T|B<1 z8%mb4{zZ7=r{)3b2TP)b%I5+#|1|C0g4S}yvMysYE~C?yhI~m2;bBO%W1A0tb^_iAJH)PV)&UxIgJ~(ZMiZ=eO-P2y|iidLx zkTl<-8hzenZSJ{~OGLfcFDWFMHI8S4T$95EmFXkcUU#b`rCdd%%{W^^XYRJbZ%l2x z7muOm*)ZIoDEK9BNjP>SJ$kW#H^z05k7ytG;bJ&MQ4}^>l znVxJuFU|B-@N;DMAKV&@o|9@zJfjw2oUs--Hg8)%^twR`%B-kOPlfk^dnA|j)=l)g zeACR2;l*7n(KRM5BuDc-x%g{ewbBijvik?H^XomP7S^}Mh9nD&yfN|Dg)SXoI7s4H z7QF)M5sT-kHRv|OaIwnJ(R(VV9;-0Ex5?kz-vGK-`?t3MM~~e9G4=j|W5Mj^%L2K# zisFxIr`87AZ_FiMq4i~S>uO+dXV|jQYEM4*N0Ij1{%myB-Wj&~H?OxM3hu}5I8{fT zR9ztV<-L)sjvrHc`*V{^MzOC@^iNn2Dy&tI; zS@D+5jqhK}KUQ*gG%I^QcW%$p)`q{=q5P`>>!=U_+mEmB$CiDrCjQg@T+jJ+w_7c1 zqb&orI!qG#j zc(ZtmH7-BJcZK4zdH+=muBGJRMV6P;L99hp`7S1@#FRfVwzPt%D<#k(@VJ}jeZ;wg z;YCcKH>QS#|GSKxiV=xHq2#RDIk6gLItR{pGvpSV(3c!AD^%N@WIUReD#Z(OA(GL)VsLYE;DtUZH9N5o!l-o zzhdMP4JP_J@mQv?O+u6%p8Ql#F(Ib%lc#FJh$Hs0OJQMw4n^gekr|qYA`1mYhA6S% z_2=4Bc%*TU4v}mZcBtt2Gk@ZJnKRwbdN=UQ z0uK*3ys<~pU{-{)f~zz3nOb~*G*xdoeP|>&U!6Ov&Uv5?VPO8z>!%@zYHp2i+@XLs znS;kJu?;F6anI5PWmU1^2e>O7r9H>A{4w0qQZ4EY&eJHBkl1P$F^0|o_1-!)mf}Z2 z>MaP}z*bw;Wd2^&6N#72uAvUGEn%@FiNwpVMw*YXD||g~b$#=N%NF7r;8yW+pqXtk zSK!Z*0T_UrGBku z@|o%|bhp>5-cheLirc=0Zbb4S60Q0wW5#X|yPOpEij+6qVe?S_>Pq|k46NdVe?>b> zG7n>pF;nSNQ%jk1Hr!uYHg+m6E=ma`<(rUM4%Q(%yX7LD~I1COP#(6gSXO~J^gg=vdYXrS$_)*3FNJ_i@qmBK>oTZ0d?7R}+FjrN7# zk;f^&`wE9H_WzaR{pV!)e{0JLv$uDyUa61f-^N8IkM7Y;Uk+r*SS+N>&6!7_%O?^! zp2*%faQUCZ9OLea%ou56p#ZMx(S(MI%ff>h*JmZ{_F-G$+OsTBgQghw#~%lLkK42V zh}+#;d^5pWf9JGQvQyaPUMfJGa$Gd|_n1O*3XUcWwhnSeH`+!{YoN_HuX?O6`5FXb zp5m+v+Pfyk6LT+iijpqdc+5iqHP!V zWF0VDimXqCL}|3h9Ot!UkEokx@gU!)wOgvDX`- zxcnGq*Km7B?MgaBM>0$ImC{r72CTe#p{BO_X8egd$~ogxxVW0>cl>NL*+UV&rL)4#i4NZ(OX{E*yU?ZMevzdxf)dmW}L zeRd~q5M+MuW8T(W8Oqe*q5pteMi(kZr<5v$|1`FsvIsVD{3+-B>;Ck|o9A-l2B9a~ z*O}f38wA)PbjEVTYpr=zUp&EqucEe;VlT=SD8>I_HnW$dNGeg*Z(}(f5zD-)!E2)< zdqJHPvpZp3HN0rAzFBK^2boMVHLX(^YNJTbonK^Z_IL=VuayqN_*4UtXZu)am6Aw{ zB(8vW#p#!8#>aA<)QUm`Ub~=FK*t+xJ*ap5Vw3$=U8NEuo`T;3fts{AM6V$qve=_q z3ZjxD*2^t50u&WB4I%fn9Lne;NvSI8h$~L_fSx3}!2?;BS{bGbTnS5A78jQk;}O8V zbP$^f36xY~}_!VN&Rc3Vq9k&*I#J$zjF}Yp|W;CPy&;*V})Y<2|*n4*s126xgkU;&8Lp!Pl$m@p8M=?Kx??ERu;Y1 zrjD$!_i=yKMf+*9WQr|pnN!;=s0elFy}8L|$<>78_=oLJ+~zjs8Glz=ymxP&hPifW zehg*t(QADHKi&8~4}M-O-aQ?hsk_(dejUaB4>-#ra6Qsmb-Y6`eSk+zR5>*74ld8( zR@Y|D94lXY`x-@o7S=fe=URfFsPwG^5ntb z;7n95OQy*oYKMAE5MOF@x7t>USd|_Y1~a&;oCh(cRYBERdUGmsge}LD^u>b|ReukS zd>5jG>fLU_yxk9{*%#iYPyV!cR@e%C9x5#%L^L&Qb@Q1cZ&Y9zTq1$KJ$7G(^XsBh0(vR~lA@~%5>9S3N&9FK{V0hjloMPiqn>REh*_BWp36q)lPRR*LF zTudx)1!e0o?$w>D-q~ z#~sgKP1Uy=;>_%F@eWAB6_PDiR-8g8^KtCP@Hrk_I~IHA?pKkdbIEGg#ya<;qJ2x9 z+%qfNH{8bLJLd;hLc%O;E;3C5s--{CO?eWt%>})@3c2?k-p6Rf#AGQ;OjF-|L8_*f zy-9<0dO%j39@I7u>1!Sr(l%);V{5?EtGDXdqF%qTl^LKG=8<0pA5-paIBSqG4*ZOR zo~R?zho(n>e4XC_i$Tg-B`bP#jthx{{`V9wYB*>$X*w+Sj^Ngri=CakMCqhZt>p_W z^lYzSd|3;No{K9KxG}C#(9t-%5wNfz&R-WSDGS;u*`>qhgCI_ErVewtxpL~aJ8$SK zH8%B+OVVgw?Sv{bWq-kQdZ`gf)c5G`PBZ3c5-!+&Q@xs{s}}blq9wJ>T4! zh(em=CEG7U3)%nfhzyIt?c-!|Kn!>zrb(bFXpG0OSxa? zOushhcASU5#K3trZSIXoYUp)0T-ZzM!PE0^99g__|G;a&NdUkNaPS6z7XbJOIP&0B zPW8(<=;Lx_L5z>=)Ge!{g)APKSR&#~+*fS+v&r7(>G-|6k z6`{+iDPQV_2S&z0Y`yc@&*i#qTiaenSC2{*O7zA9jR>(Bq(_~^YsKc=-s4|q=Dz^~ zz5zO$OP3|*3x#VoD>NnS#iqr_z_5mElLEu4QHAtyxyOCbES@|^pi-^QN^k|8S2f-Ra?_-&M45w* zm9>j~Y&6-PPpXwqL7ntF^s#lo&`f8r8^H)uv59M36~FNFyAT)rMx_l&0DqHOM%&HW z{B>a0g_u=pYUXQg-#Fu8^FF+k3ZZ{~EI$YwHEXY)YfGx}aww>xZRj9*RBg3Akmttc zlq4YaoA=Vixr$rO3zlK5sF$w@Oa6*O09vOr}4unxA2 zd%(f-GuQyJr|59~xKKFNg;ZCWV(eAA^b~7jKvGIaJ0!#_J=aB|Jr2jrLHF@c-$#;Zq)URIRrwd##uCTc^4B2Jl3 z-K>QBB-4n}>(47#+Ng>h{Me3jlyPY2l}Ao2u&nYofRJ{@h=}y?^n8x@w596kgc;`(%0aRuNcG5qM+gTaI!S97j2DCapq?oj8K8-OIw z!J&paX7<|Igk+iYUFE8%vafeoPR}f>v-&2?&v6KcbB;FHM>G11n;~$}&LnzSX-m{eKuA-{rf@b7DiD#{ntfo)JxklE z33glp=Ou@!7s4sI?K`Cg7w>;u8D+}Ey>s-l6>W4K6HQu7LAx+zfHgWl7){5-mbB%`y}Qxp6x}WTLu143Q#)reu2JN+wyBCBTcN2lyMk$p}1mCXitkq5s2gxU)gYWI%#htu|%`Mb-DN492MD11O&V#636Ouq&!du z3#OLBNv|@8@jb@|vfIRKRkJetJCV|PH6Lb}`Uh7Drg9f%F=~k!)in7Ppre{;m02?* zav=E-W_fFqim3n2#WtVxMZaYVzF0SIL=PSE@T~!I8>Ci++Pv!Ky#AK0@(o#m&bwhS zpL{&Tw&sLh&6o+rp<*@YQbh?Y!(+jygl@n?>C@dOmlWe%U2WtVA69O3yE#SnDb&Mq z6J$A9`FJf_cHKBF1Q?duLTyg+`Xvb4V2`dQPnLor4Q)4{W+tg;h5J)6z%Mm{RGnn^q6obIs79XFR6QkBF`aRhv_CE6jRxrF_#&?E0Uy zN1dQ>^L14oH-=~`U>?*JVKEMp*{94R>`ZHof5{KrQ4Qy6B<9w1SytL0Ka<-M@$G3+ zBUl%d<@iGBmNs05U8ZVh_*^D2;XI@W;&5#?`>nD61%itf1@2D)gB+!cp$*}X(ClOZ z%&F_%3009fM=KO&bLQ1Imdh`0pd-kjBd2S}{*q^73mV*iQr9i~)kV>+&T&I<`C8)k z%U%{jzGGnHBw}X`8kUSQcpocZdPb`o!BSvX^T)p*Jc!1~hGT9ysBdjJ|8z@?cERRaUD&env85{gB`(P<>F&;4f0KNzRiR5W zfTO400{~xH3N632a+bBCMCA{>0UZAKL8r>D<>-Ga=sa-od*$SRtYY%|AjIXzt?2vx zh_yMH@HdX;pO~7vj5Xe{e_(3Nv&fEP?%x2Oh1OY@}nou^W&HEDywkBT2okw<%k2tW3gA{qdwA^Fmdnj!fJ-sCG2{`>RT*P zzhEHA+9N<9WQqeGtms_R^)3m+dq z3RVp*CirMc^Gf{uM!8qQpF1THGhEt2PWtjvhU;!bE4i(c8YUFT1(q-OEwE ze7bjxRj^(hjxr}F`5X$9wPML}^}VXK{pC+5e0bETK2c+M5q!4axrDG)n7@|R=O_1{ zg$3APYsQvjBtfA$fxBbYXhr8|)sm9(HfjDz+bkc-Bgbc%{)Bpam@IO29SJ-=5nfxb z7fX(PC#c!V?)0b??_cAe?QZ0Z-f3oqF}SfT-uYh(mI7Cnc`4U&0oiI+!5FH zRo$|r%OPcDa2VZrJb<$W3322(d$Z(FgcW9{W7ABrg6j+z4C>RDqO7Wnj^b({j*qh4 zvvSOpP1?}R$TT_2PbD9l!q4br>7~sj6iG@r{xo?0B}J>V1y!frKX9D)Rk=o*ZKS;> z>ufz~WXJe^Y^A^mYHkn0M6Lf5an@Lz@!7DsXfv)t88x>dBa7G|S{(-|07GF-`!S$l5yDcLI^ zM1{*X8(iQ2zCPkf%t&G2*IZjqXuAuh#-afc6S7uI?Q}}Cgz}$dddLVex@JpDM*<&E zLVWZMs9Kg^+SI@FAWrc+t}n}Ne|b2}C}ag)t_D`b@~#PIudG6{zwDI-Nx9pN_8#tY zG^K&;P1}7DCG>fBOay1^JQdM7SQ1)ckNk^=(~9piYItUH@3b({YML-8 zfgI1hjYn7R+>9Y?sPgi5QcYq@w8mzuRJwh$93^a-m&As(hf+aWTd%>(28#1SI;)cm zB#1?kdbX`cGuFgLWbQ^z`Q>_@RaEqIlzXEq40G0V!;X{fnf}wEW}#aM)_83J*yPI3 zSS}nwnJ(f#No*77VaZO|$#pj}jvupqN)swXyzCXiVDGN_wZW!m&O55cw!*v{D|3uZ z_3!o) zLeODsTj8CaN)E%Zg|gs^=kFOJPaD2$U#P2HfQOH^Pi(~-N2L0+3IrI6(75hqRLbzB zp7OA!r)Sk$9DV3kX~1F^@Z72jN+)V0B@rWUGBKQf-lYrs-p_OL+7ee^x9e-I9#_^N8Lw&7*hi0vR{j}e{sP1 z!;-lwLLR9shEP1^zRd zr?@s0jY%u}MLNk~TqeTX5+4y4BN9$c>eJfIQ_z`3v#v25E|zKHVD;W_kY6Q|K3eK^&d!#1;(FwuFhTA@;(P9g*fIw zhAWI0C&hQ{62`|IWlw_Gu^!hQyxW-zJ=(@NVq7Q7Z?W(UPeX;7CLKcSV9^-PJ{u!c z4AfRxKzqS`O;WW(xdB`qsEE*1W(LF_w^8MYMxNL8e2q@Jgz)xby{vc`>|KHyjtCw1qjh|7jng#chjA>d4eE2z;Nk0 zZ)|+4r1d?<16lYqZ3>H=PQHMJ53n-6%bMPsM|w<>^C+?Gwok-mFQepNQIUh})C2{$ z-fB$Wa7u1!`=Tz8BO$WZ@?i{WJjt~RSE+_Fe#1XXLx_WG3;VE72z)^7T+W_>79O8UG+j=*h zrO2jjdgi&k0s$zPo}ZG^7x&;Wo?(MKltp0?90&QdUuluaTFC`Go~Z7NO@AP43`=Ce5z#pc2XHxe2YcWp)k&?04y#1-v&}NnonU zxn@^^*ukD?F2Oo7Gpy>CJ{;#Hn8pVLuxoXJ4T#o7k1op67HeUqueZO5zDh$Cui7~7Ia;mvK(}S&zRDHBFz(lc|Isheh#4&Vf%ZpALKYw-n7+Ng7XqT;j$vpfN_(nx|i;tUGVJJ@67qRpA!72fz4pZF&|5vjhlHE{(0vtj;ou zah(5G(UbzU!lv5(uR~_5I`%FSW7_)z*&CgS$qL1~-vA;3^d0Qa2WlSu%RLRCNvL%B zywfs!HR6d#&fg1%>+4}XU(46eC>}VFl>g(1OCxSyTq4#2w-?1k)Vu^w_@KPAK?Nln zadQl?Wlx|&g(wg*JncYoNk??P0M~Fz)>AmuInSF&4O>I&G#;}6jnN}Yxma{%V2{D^ zD)G%kruBHI6aOyTQWs3dCh;z1=ZwcQOUA?5z~`#kTKfJo{-GggE|t$i9$6fP8D;Dz zQxcM-=x$<3{hMWFbf{!p#wW;J`!cd&+#^OlXv?yeg%Px*v1bM{6CuQ}C z&H-`K*%`pwcW}Nzu`mF^6g5aclSe-td$uk~o7cPb{iwfWZcwq$>SuUQtpS?@jIlt9ft+TUi;TsydA+`$}XPe%w#_J=`epvhirQ7 zXDYhFyss9Rw8z1C32F1b`UMLez3XM%jLsp9d6qWJkF%VWgxQ1d1iY|ixoLD+m^LRQ60j|L$4&FQ$^htts7xcIBNVv!c9{OTdvTLwl& zDQg8p1u1a`y-dgR8^EY2(5B3MKtCYB%q0$iLSL5@lEgGNs$a?*v{4#zFYBkXAKu~1 z!l8{@mnD@fiyL~{-g2GnvV0JOZiFq~E?TU#ee4{s9L^&CBE}u~hWwqFC zI>8kW;cBHQG?n6?4CEHW{F;lsp~8H1o&M>W8cKe-vy=*mygs;-56%~i!$CkWKRfZn zW2okvDfZ6|Rc|Snk6BA)RU>6xHw+pf-b<$&Vh7DFZI;92b5?(0DYGtzn#IlNdgCNz zRp4@hN7Du3>6_5a`mAXIrC>`8rU$z?)*3k_-?MK|o_@BN{&eZ%`s2IL>r0bEGb~JR zbtj{8u#b{zrco^>0`(+|zkpSm8;c=%*y<7KvmPR`t`BafKuc_hQ5;v>vRQQPGq?W& zT}hEE*Q+cZy}(X>uNVDq7;QS-kh`eT_BSB^aq8cCH=6eh2d*$dh5v;b`KMx^hUp(D zeeFqwR9M;XsqG)jHLozv|0(p$U@O9!H}RfM3s_ezSPeUpP zE08&iYeXk?1fUpA9c8h`NoE!{c>cgqzk^ZR&@vIKCyFD1u)0^iQGzWU2Ayy@EDE-z zGp!*|l@&f)%bJKyRVxsnqBYrQzz3LVA@vON)6qTCLey5iXOfE5hOEtuounL$W2FyLHEfXKVE$yaEm> zD`Fz$$p(c_i<3Hu%^@Gm$)@^hFbPcZFdWI;B3W{J}d89rn60z4<}o8S?He<^OuEO6B3e3Jzg!m3Tx2Y8c=qCeZeN2(sF#% z59$u5EC-sAOsS;v&P6Z9gBkJ_)nK2VER2QqVX?{Ij=XSE5i}Ra? z-u5=(R~tkwTqg2#Z7nXnV+`b?Rl%voR-81_4K_XRJ;AJSVX;)AMB?u@b}dd6eFMD8 z5B$60;mWZ1H&qV%itry*j)USEu#I1ISyoHtM?7g^IvT|dfTOxs{`uek|9OQ}5%djU z0Tk;Q1FCIanDArStq(L{)WcK)_b}Ht`Ix*rf-FL@bNkN|y0Y0XzX3k`kU{*lQ5N3- z{XXQN{)GOEnvs9)|NV}lm;52+UJ{0SeP3x`^RI@^|8jTprZOM0qE|oPU+ib-$N?;> z*mWT#XLorSYjLg3CW!m^IfV_vr?B7tbo4%&ytXM+HY~D@^XsauM331XhVzq7YMs_? z6O!Y`2_V9DldKodL{L{eBAeS`DiG~ui4$(hHabQn41>T)8DOX$v`MWX#cw|YH z_tS|xFH_Y7q-3S@!|OI=yO$UK|=#`$8@~4sIvsaE)rA z?puhRrtQT-xTLm|-%0xoH0m@o4*P7-eAd+|#t{y@{u6_g1 z46lPB7W!{wKn9&p?bTpO_Gk3s){Q#4RG66N;2zpLzRqG(`{fiOi{Lo4Yq;HfZ!uc!urg=$5Ta2q>cy}I?fL*t!VERRg+I6{r-Fhi1ia)CPqng{i! za^d3Js*d|?IOfpc86JHeZK1{{2)!u}n>i4K%fGNe|1HmvY(1Hncu1RIVT?h1xZpf> zezaOw2q%nG!3VVQ@72eT7L4MIeBxdWE=HB|(nR|=_*@Q|*F`cEBSn&HLh+dt(}%Go zuf(O-9Nct zk`K|c`vy2Tgq2*V6?WxP)tCs3nS)o?g<-$oq=dCwH)ZWO*)j*?ueJQ?{qXB2KRjJn$Jv+B?%PPL(@4P4=9o6W;^&oDr*^8T7k8x z#={;m-9gjo1Cq5$#V61kVYSu%is5(q%57NAcSgG`qb>iB)-D*Xuf_vH$Lm*GI-&tDeB2Dm)eS#mV24b06+MTTijV{ zWMzm+eYo>Iop-o@?li6KvS8!4J|8HfC9nQdat{FDBOF~A?wI??lBV37Ir(?HV_&1U zIOjKjLf!rEgCG8TWhKv$aG+mbVksFg$Pr0dsv0Z5qk2wT$W$F25eRTOsaa^X-v-X( za`LDSLpoSb=~a5JDxFU=YPvZ=5rgff5d*zL!39yryaxiQO@{R1 zn~U9r@ay5F+O+Y6AgIdpqtD(!_sTlMAG6EZ5YdM4(qJV?tKr2CNkQs6zqh`qO^#KQ z@u}P4b&z&ZKJz!XXj!+-+^VD*ru?4%(mmrrcB94omw4o4XDMY40MklnfhP~cWrmfW*^5A26iNSxlZpLnV>b1E?jL(II;+Ej?ENI(bh^}4o*Dn0sjwk?*Z1-)@=<( zJyt{yB68>+Kv1f5DFKz4Ap%>{S^b(Nj z7wqT#-+S-Z?)Ut8^6a&;S!I*GclKO!jyc9b#@+eOnH>A0$GFtKUZy{DAGFkh2j%`4 zlN*r_P;YvPHpjHh=N9F|>GiMFEMYw(+T(Aac*jv*QgUKLD+z2pdgcWIm%ABh%<%TS zp4}4d?K^5u(eF&3rAP^6TW-#rnC-8Ef_4P1}g&I{P|$Q<$y-7!+h>o6f%_wXWzX>{?|tO> z+*Ptrv+s2zC!=24@Ic};ROF?kOC&ycK+ZWv>w(_mrfOGh=7RooE;VK;z~FN(YP~~V z&6p>%5m2LL=J4J5H*3z1mNiE5Yb#+lpt3FZNX)e*>P!vIb#H-_Hbz+EMT zje@BhTA{CXom<(5#C_BgriX@5q$Tm>gz@78*2XjK2l3Y6)MfqX`F4v)eC zmw*u(ghA$*mW@zKFk1ilElz$u#9rbIP``Gn-2s!H@$R*|s{ zx^nlsOg@^vQ$1M#&AJb^s69~jN-$2=S#9!hz{ zQU-5D;0135!HW5_YldYf1N^QmC!$cBlX8J!bxMvj*yzF-M~@_K`g=U^B+(Xnq_4XL zlb{PhpsO8I>~!z}e-I?cv0+UnZKq>fhssa&=Btq$+Zr!5_~t$H4*tU<){Pan z#GG@o47iQ}9io{$ zC)|%nAM5xBOFr-?AHF~>>HMkRcQY8<{ft?5%!|X2i#LySefatl^K~nqd9a;<&bYP~ zq`1GXU?#6Gd^)uXd>FXo5ICgZT6gSB1pvOq=YQ?^U(}EI{b8;GiGDKm`)u2f6q11- zkxclW?{jH)!0DR+z>Nn#hWfdR`7dx=he88+*Z~18g1G&*mE{)t{%lsz+U5(P^o+Ha z3ed5Fp)D=MX8@*m>^gO+6t6;GOj~p9fQu0_T5#@r6VHv$fVj82hWj=lj~8w|UQmSR zr}v)J)!o+Oz_|9x_=n*C+RqOzsjK(COnnn2Uw_|K+|zRaI=N^FdG7 zIwOm5+28NU{tivK`+e74b<$_R)xS3jckrJ9r~Cgt-Jb#I;xBVXJ-elY{B#J=;f?9o z;)xn$TE9a{_Zj+mGLO5BFnlD9`kh~wE z`pfEj5j+Dw8!ZyEfE~!J%>`Fr@fr8pAvs=i zV2MKZJ>st)#3#S zrja*anjei{%08I<1{=RvoQb6!<-5%lrz!B$sv^^t?4xmuen-cU#4+U#H1IEXyd|Jpe0lGUk~0lsMI zDps+6UCM|uxbGge3Z7+oWvJHp1#odBAlej}>Scp`j}qSmvSTL`gBsQc&27fyI{1tR zY-7{ss6T=4byyjzC+aud+~8QKN$)G-po?swX(J9sG0q@yX=LMqMoRqvV&5JjQ1} z_|A!+3q!{X5{48J2@^|ap(>-1?F(A#TXCU7W$~NiaB|Uxr=tX$*pAeatg!j81l5E) zyREIv4Q7zg>(Q53Gr-OD4~_~!yzZ|}HiOKakJThdm1yH}jtv7b%UxRQy|* z2va*z=Ix6Kvgz4Q{-Ki$w3R#pCCha_x%F8toD|Ma;Z2^mv;2*7;+uCLd!k7d)4of* z9Q@H5p8*CACNn9{VrYU(sLa!D6<~h#d%Qo<;3G_@jK^l=c|0SVjgxVeQgI)27i5+r z?2R8%K1T1}r%pC~IX{y(y`ZeLYNMf_4o#MUYyTG0L%6+wS#+l+JqAYHW$Wc5vkk$ z&Kl!1*u|bhq?3g2ic#~sboei97Fn+0DehWGQ>f!e$DL80O0z3QJh6{BIhPtgOf!$Z zSK=HI0qh)P(~9{uXn*~pt;ZonhwhgAP& znN2}-I9&f3_Tpu?6rI3X6g1B3xkqZPbr`+y*S&GrUcp!fmCP4P=;{!Z`#>pCnRblL zPv&^>+e&LH+8u{SQ87tiJo3Spp#gyZ6{OttrJ`%G=cdx{+&Bq_I_CJpR~ifD=Uo`C zCOt!5@%K(1)0T1>bR7u6%@Rjjh2`f6`(kZIb-`Tit;wVQ4prV?6UtM{oexvRyB?NF zr6;?Ll#P{q2Dp*ni$mJG!dv2hIk^JGN2(m^Yz`)Sw1==Mn-)vLY^2y6zwtrp?IyM`W5jyt#?=Ra2|2IEg_yLEHuGtJ` z2^PHoOva_BE%tjrhzc>E0R%T{?rwj8u#)J#voB4)U74@0vTl!vEz}OpD9q|31zkuP zHk)H2hlZ7UT{h_t|Jn_}zX%D?G`BpN;uK!qA!B2wk|3(%R9mtT~V}g{CiR2Gw-2(P z0Scv27Q}N$3O!OI9jQNH^_)CkR*>|XPTk{f-97fm+a^ti}l~RH{2s3+28b zVLBt$D&FK%Uv)BT@|KA+%1f{w=5I9AE}d;C`%%v&vk(WCOc+xK8Z1+a-#2(9j^6f> zz)s$e3Cym^~2>JbB^3fO1a+=&X9t4S9TsJ>*>d}K@aERnwCjBidd z1kKz$i1tVVh`P87lGKe zvYT&9i~EHkO*rf`GqCM4`fgG}IllTuTFW%E<1K70DP=#rFle+3qRvvm9c7mXUKrIm zk)tVmJi&d$Uct6kS3Yd^&%W8qKkGBPq|HdPbATgh!(Xq3YdSw2Y>7r1K6)Wui2ai6 zsps=up8+kbJGH0`O|ooba?x@t(BxIL9`h#3RaA+r%DJLH1Rq3bm*T5(&4AH;31BW0 zrD_?Km3LdBsQiwPS3f$oB4AoCDMEsquZiU~5W^N146O$zSnaRys^U^$_+FnpNLy8w z1RY04G8cuaG!uNT7^3*>8s6+Z^y6M-7^yBd?$GapR`q3-|ACxH$#ODzjnS*nfC$TsOu z3!mZ~uMrgO@yd~~S=Ino!g~y!JG~^`i{izP(2XFCoGPiWM8um zBei0bVY`|>xiBW)+~8dnKK|tV=Wg{`$tzk-OyVZaBSuj}I=Rt$-NsJBRI(egALF#& zMD?QawigC$f)c!>q6(tP^2kQG1M`s2v+@(e0mtohPlMKMY+!?#O>qBvA=Xe=P46e|b(% z_HCd?ID9J(#p?b@x%_qLd&9AK>vuTbxbw5}=uIQPEfoYQ9oh!}rQQ>k=J)DLD#6z- z)UBZTv@orAhXiUz6!l9T=SG!wku|G+{?e6;M{?zB^M9l55=GT^;rOPD7na$*@r7uc z2KV5{_*6*#H&vH^8~Xk~V5Orbb|u!@P}lqBpA=tsb)y_L#~60~EmGkr@)Ae5)`sbwsKJLjR&YJKLBoFLC`usK=yU2J_2fiqflj z9g$i`GE+Yh>>A;K&h|j47M8E>V4VumDB|s^2&ET{@YN|jsLoKCz;&!cNV}@~2<|lw z75W&zv*`iffd3bdAIp%12PF;#HtmtYR|0kYvv+aRlW_}{=@Jc(VlIlk5>@6*ql`oq7TyxrsgF za5sGHtd6d-GTyE^x5!X4H&{0JxAR`)F1ON=X$+)Cp06S`($HhHqtIYcnb&hjG^^a? z&w^f=^n~W^9+94<>>Q;-6}Q{lcATX&;Mcsy<#z5SPUbHKApQZr>-Vx)(c6x1A^n?)`>)AZEbpZG9;VJ3dA=n%zeABp5vu}wpj@|QYwcd5LvA% z-sS0>ko3a8Nm@^&riPeae;WEKztZ$od1pO|Z09d3$$PeFZH+=Fn&84SX&3hLQs9te z4>lf_s6UUAVfnj@;V+oDflB}wLQTnN3@CUka!@aaujW{*66UpZ$C^R#H zb!_lV^eFlp?l(NvT-;NDmQ4#}6KkJ2P==-1We~Q$M0EA;e>vkPV<_Jvx>5RUY2O35 zH=2;N7Hykt`v5?9`#-}*m&$)}6HtRv7C(YgVMR+rhe0X#e+f!i$%ls-T*L1=O_t5$ zL;CExUN7$*v;#v7x2R+qst{!Uyxo)7HbspXkjjET#nja=Fjy=EL2-ds*pIlm6%{6* zA&a}-d=C|?!L2A6ta-HPky=Mubw{jlgJc<|fjHAFeskhXALojJ{!(}HM6CW?jiMXj z60=Q2^m7d?NSHYRB`i1Y@vcylVv}yVQqsn+4JkqpDs|HPe6er!JVH&eb-g$y7(C0v zT$4D_BG5w=>Pn&CXmYex_o~$G+Nv-2yBjgB)D+E<>W)utv&j)BdcTM@vPas$5>8s8 z5>$mZ1N9i0CL)rQn^+NCtDVf)`u%HqS(Jb)&c(~4p;_YKyI_c%TPsb&#gzQ0$9_qE zj-HL(UfLxTS~HtbbY)V9QaaOEb)VD?Q@h?Gi?i%Cm3NT>J1FMeKxbIY?D;^6NG*w? z-34AlkBmTyKTF9xYm1x9t9*jKhzBhVS3PjVWd8AZ#-F82(HS`zy2@O}E#2{JYK{9# zYZ#F!{53JN^|H@^?kzrc(`T-?BxxB-ncMEqF$uO8?=LHr@D^$W`f~kb$N`Wbo*g-ZnOS7rMD)mb~>9Mz$ zw%ono3{rY>FU6Vbw|TN^31-vTh-}Z6#v3f5D>{yY`GZ;oyMGZyhWpE<^m^~$ng{+` z_OxZLzG0PUk$5u0#ozx?6Y?VxQ{V64a@~_e4?5LkNAsAh{4pye6=NH1yE;teoWk{z z1QQ(cJgZyPX`SrJ5qZ&c_-RY&Ac#JSDjWXpM7iZ)DaaJ_+z`66lafoTr0=;(z}XBd zxuWr_N*!n2O;Z|&uS&YUx++*cqaKw0u{}OvTgvbclgD5sybv8jpr7f-#=rZF{nrim zWmPR*ULq?*EF(})jPq+?r_%`an;wb@p8;*F)aWO=dab~VEsJsSLP?9Gfm(*!-9J@U zm6W}&r)w3x(cLV}Ly3FXmzh(u)mK3vtVXz{k~-^j$Cwx2Bp^p42JFQF`h(88Lonr3}3;WnP z0Cn|u(`Xx|L~7QmmK|5}BbT7h_O3(xv4nZ{+zMk;)KqTL2WXLrU->6K7UyVH?iBUn z9DAHb+H{jXmnkZE$Ua6|GRgNYk%4XjI*IvXn2LC_+a^Q#1IMw5Tf34aej5tX=jx=7 z{jA^k4A3p}nY#}-yY%tP566*TOw&p{sPfrvp~7N32e+Mh)PdS8k)nV3n8&Ga50!7l zPilvA>OXO~p&@gWjD3eW?*AZ8_|u|kdilHp4FRyncKp-Nc;qM1^_f3v176ldjRvk} zsD!cC<$BNUHoQ4ZnR`F*^&pN>y$c{7widZpA<`6|lGct&2?ecnLuZMfb(hpbN#B2X zGLJyFuK7A*{R$yn@p`}1TO(@Oy9O7htl#-DiIhM1GR2Xb$KnF=mob+{5N~x^>C}jr z`UM&o_HVI<7F{Ljr21$5ty-scHy$!zqJmgvMv$0cuBgO z@6G$i;!pHQho}!Uh)u{xz1)ZT{l0S@_`fF+%ozP-dO7Kf#r69XtDRVB){^dM+;%U{kuS|-kI z)*6-|#K87$jk+(WI1)1R-&*tlpa| z(zAq$eL)z#Fy^-i>LZbwh5`Gd{Z6bi=e{dJCZ3p373e9h75Qg43$uv%(|j_Kjtbf2 z?$PZWDC2pQOH6!b4dHKV37LA8MrXpzXTx|V^@baGw&8b7hraKc)|GNR+`wS=jn_x4 z=Rx;XFE%`j5W1%~@eZ*r%ldUgWwEp`=PLLU=r?``W`{utm|z^e6eOCZsk7yAw=W?8 zd;=b+WR+HK=)lOP-CE{&ZP13Qa%hD&@lMv1><|Z@%dQ;!h_i66gCElpN z2M)B-aXvAILc87d^3K5c)L6ybn0+K+`JVD|6q>)BwfjCWk<4>Xo}6>NI{ZYZ_~Zr_ zm0#wHTPfCC0S=F_RFun$PdpCtW0MjX1{s<`>;~nh^>!7h(j}!W_Pg<_i)X1M&K**V zb6wR=I7F8Uwj6LrTjk7NwJ;}5W?r1R_&VNOb)4g=a`B#T0M<9$N8}Vh^-QuI$()w?meKj2)LZVLEi{9B z7IQ1|`3uw5+4tqzz)?YY6*KAPn$vGR9kqP|dXk@M_b7J}^?Ez&$uF&;e9g16XQq7Q zJ_C;P+cXJ&22eylP{WWDsBS4~mlU-!Q&)^TSejB?5Tms~S<$^4dNW%NWS;_{UjoCGw>u)eF}LZ^(kGKtoyO zWEKY6MNX)IYk}=TN{vP}qnt>v3ksIYj+?QH?##obc=8q4PD z>uIRM_MizQ4wQSP5gse8W*ewZr6|LP@6mMF_`_#|DokA&JU2F58PM^~cz;`$_fe+O zU`U}}UQO!uMX=}ox(N{UdNdiL-Vk!yW3Q0~XZ*a{iaVIkm{~L;@ixh;x9_R(ZZ61B zR+)br4%%(l4ouhQDD5xsfRechpe9-jigd-JkJg)PLl>x#8;icWVVGYdSYKqna=TH4 z>rvW^tddK-`fl^{`u<5SbEFOyKi&dl#TIawKmSS~k$$*Woyt*`&jtA?KGGqCaOt`m zo3A&o&Zn4BKb{&e!29~Hc%_rSqdpWXmuL z8TtgnEb^A_)YUM@Xx=0#kl_a=={fF!@U&f{{Nzibl$?UCGtEJ|x?)nM5vglLC{>dI zekaz%i>()7!yKh=p4zOB$=AL`!-QjDiF~iZYin9-dUhAX6D?6?W8lrE?N{8{*F6N? zIhCJ1oJNO?C*y_1P4DBcp)+%kh3HUN3~#HR_|yJuS1DOJ`Ra%>8p-75yt$7t&~Pvd ztCJQEZz8@j=&kFZk(~2rJRh3<-px)RS9x*OyGUXx^&DWDX<-oj@TYz)tnml97eF`ga+YAr!pxhhcm>clCz_v5m*@%SyX zO3h#Vt`N4L0pZf~pi%7qL=5;WTskv&{IV9qeA4|70fH9kZH^ncf;M2Dg7OUl#m!R-rIn9d zZ*wE=v^3DDYTFMf$q0k)7-6&-%oy4tNAf&UB0TF?f*FhbSMFS`FwMl3LN(^Wk4QG~ zh$KRvB&M|BIoO@kt`~hjb@$?O`lUAfI80xgw=#=_@JwhYi#x-}tu?1Ac1HoO2fvfV~4)0Hmm7Y!YefpOmj5#%1V9T(L zdQ;X&n6b_4ljqIM1o?OFzCw7LiNz*>q|I-9OzLnsCEw*Go?HYnFf(Jx4r61J)0t`R zi4l&3gtOOs^S%*l0$Q;ORAHEG1bSl^9Woa$1)X*#B;>!3A6eN{mGtqEN+p6v>hIUX z85reP7Y2H>iu*`1g{P^?C*3eRSADSX)zth{KFU4cX`Fo+T)M1Zq#=G?VVf>gJb25{5W+pG&!y+MEo@Gkv38n6%Bjv~EjY_+FuMyEIyC zqJ?F!9rd>X**g6R+U7X7!*R*=`)DQ8{9oob!G zoltr4)zl1g&VC!e-qtf)_VNXle@>)!`%s)GN`%SH6n*k8P`ErDaAE=gp!)Cs@jm)5 zQanL#{xd0_7CE?eN{0}t}DO=5;8G&pIdU=aGVK5~N!rTV8Q<|uJVf2Gf9-MA>5 zFY9NWKan92n2i}~ZV6ozwfY940s!2`ycM3HLfr4|sB7G<3@F|c1ntJ2@7B<7`~W;$ zfIK+b{@2%k_{;Xl4rsra5i2#VxUQfPg5}(7+{@#$6DG+8W8|I&afs{?*8beMS6%bp zWbo^HcX&Fx6jijI6%&n#Ny;N9C&Klf)~Ka9F0TMj$2Bow*+=I+VchhEG6ZgVfvZ(D zyed^m3!_tEQx>m{Jmms}P#AxkJg+pDQ!#y{w0)qx)hJI;WLQ}0^kR!a-I|_KUWTw=1^jv4XLnS)+nj@OY*1@Wa-Gjdy{wmzPYa(qL>-(8De9Y0#s+F8DHK4{2x z=XRoM2l6tJp(9r2=+sSEDaMeP-Q;Q;AyAauH9$fB3s;G7C4U#tI_4ZkRhWJN9ac1fA$|jk5dB@McP- z|EIm;*}52#m(#(17DYZ6iR&^!Irp_Vu*a)SQO?=X&f47I+rD4RSx00MRVyeND=S=L znf^@;9beF-M$m!2G!?qB@-gR~or!#ngPDVg9+wrWTvM6blOlox>Z7kmjM~?|Szp-f zuqfS^&J*^}G*PA1GKtpD@9LmRvcgVPjGB(qM;(i+Y~AbAb_vOYEkG#_L8$fYBCtU2 zBXY-rt$UV~pGjpNZmK$7G|<4w=MKn#f40(?vmE1&iAvI|s;OvF2z)*#C{T%=23E?Z z39>Fcu{i6#N;3N-c;MLYjFhxMLtWHel1BvST<`U?hze-yV3d_jKgFKZoLEQ5*_1}Y z19T&ZC*C`)7$bs|StYb!>2~X}EOWDd4DPkJV+Cr;Mg>vxK6LldO;@x&t?BCC?NG1-?OSlDjqqnW8}Q$0 zi1{r#pNq~HhttXmXm}@I)oGlSkfLKIu*Kq415>&AX=cRQWr}Es2@E7MmOY(YJbePy z!@zJUzuKktEiW_UII?TopBUiB`=`x-OGzQFNn6h$CX{xAWS6u!;urv;iMP0Ps-u*D zX64>U6diidHabzwA|brAV579DyGJW8$-~cWHH*qn=s0qsDq)fLH^-+eYSWgy!p85n zX!&xDSLWMb~L4RK={MkTo zR{HL5<@l3Bg@g`uyBHsgRPyKzExy?svu}UzU5wIuhs4B{b^$q5wc0lyk{VdU+gc{@ zXc`mIc2jVl3%kDg9LBH>@u%g)rOqlhdOTtKI=Jkd2R~`Ff#Mt=G3=ck%h7ba?tZoz z!ihW}fl|u|Dih02rh1Pd!+uUJ?@<)29WcY=@_PRUnjo{F*X|6zgN_>y5nAL%g~ z=kvF9R*X8`g%e)9>}BRH7pi;f-7JqX((b!hk)nfeRhK%U_Q>r9Vuo zE#;Q3rt0J)lP@uvOj0V)h_K5ZsD1C@F{#*iYB7AN%2+~X{+1c}!emG%YBy1uRY<)A{Iu2?K&gRk!99O>nyf$I+r=Lg~_G zg!bH6J-*)ZMKRwTL_|}riA{_&Lw8LHSEUNd$|tN|`Rt=^wcM&TlF+QhNv(BUU9=U# z$n+Q3%Lg`H|B+viF1}u8PY(D*OuSM097* zFw2}B>)Pnwk@njoL3Y9HFmnYN@BvZ#zN|(&b+7Ng?B%t03SzAe_RQZ$MX3sNr15|7`?ga$@N@|MT5R0#l;)$tJ~d0n<(J79?2giCFQ?Op!%%?$n->G zQ&(%#tVZ2n-uDiU5BmSj+$?OzhF$NY3;O!J!G_whCQiyq0c%fdIN9z-xI^459d zuW0+eRoTDw?|$%qJd})G#l$$st%v!MMxux*{?lMeVbi(rb6PTpio||+_SS8MEX53y zuA%H@wo`{uhvxrhQ3qiv_ z{Y2%aW5D`$uzPG|_=uppv2dl%q%r=qOTj^-o$M)8;fR52SwOQ-@ z>dv95^oZ!@n3`&c0tJ*jS;QLI>w5Z++){qaA#|tDu1jUx*al`V9=|Y=6MYeh# zVeaOvB_N2QnFDL9a~tOV7FsWk!qU>kiH6Wx2#dD#B&MVl-D8lV399$y`?E6JD!xeB zuv`%DHXg;uHB_eYgkZv22Gwkkb2E#v#l&z)8LFUALiELYVvk=u(z`W1DilwvO0?l5Fcr} z6Q?6R-Wx#Xa}=L>>+EIhw}eQ*8W<10j3vT-N}G`GPsbQJ>$-PR^t;CFr(ab<+GJ-O z{TX#g7qoDqL?~*OQ!^ixVqN-f1WU6ol-0)oj#FFO3XgmU^8lmvho|pXgCdHKa~_W$ zp;J}WEW9@RFuF;?j8pwi?*TI(d~I7#`cuTfYV&(&tyD2C zygmSnkJvVmz%ML4YI1SagzzVxT5ucI6}EVW*TMRr;6xKkXGN3FA<`T?BH}kKZa4Tc zt9KD0O?7;3v<Vrj)#iWSYS&u-9poTxgpgg|Na7gY$#H@|G zsG5zkBS&byuWaJQ`cIXep^LXu)Ihy0LTMF-gx*RxxWIbuX@6e^l5(wAYctKn1djE9 zJMRQTx;hQ&pXrI8DTGwHF`k*4jt83)8)@u(M7m$LnUqhKwxx`fZD##_hkX1I3^rE7 zA0sUTQ6Nmlf*K+IOUn$qkYV(S$cNA@o4Z}ZTYA@%7R*m9+ayQnH54w!5&F&yz9ic@ z@;p*c<1%fqFsSf$dpw!W8dXzhl=K;Jab7S%1017~s8aZdYF~9r;|!F@i=WhEK70lc zl}pAa6r5qCDz=GQ6WFVY?suvxo0zh7jpw9PWL!HKs9_c1wKnC9?Sm3AF3vNV4txdI z@541tOJzT-jy6rJu7L=lk30HeA;v;tCTN59NAN4}CjCP$k`2teiAy2I;RTfmXNVpJ z`^$;0p8?)cBi@^0p8*i2t6Alj@uUzS2^8?T0 zH<7>`f`MdYEWzm7k|=Rlh>H4kIEhlxYm7*uhW@MAOxJG4@xCZ*a!9(n??C3w&2v^3 zUwbPCbczNVBc}`u-eQr&A-^5h`==FRvs2^I1-`s#{0LXG-=O6jHB- zsA4#VX~L3Q*DoI2&m4Qzw$E`Bu*~%4A5sfPKP~aE{Oz|TC?9>UOfhCRV(;>3P^)f^ zV&U(t#7*ZfnM9Qjz8uW2$wU9wGl@c%Dk^)7KlVHWu`a7~=ts^)-?+!s?w)7CE9rB8 zzc8l;TQv2wPIpk60)(b)3^?tjmrs=Lguz?wrYLm`8HC%n%AG_c|J@&M_NVweVS{x| zR*m)*du&SA{Kq0X$bVjT(yoy-QQ$57RB(_sj z__U*7kv6s+pO)+}__V8eKyX!>uD@L>P<4H7xh=w%yL)s-ab1Tkm`hhNt^sh<*5T{* zaAvDg_6;1duC$|aiOzgW)%POdz&=B*-tKB^#5lx%dhMm?vS}^o#BAEeKXpltp#s8# zBA^PMTsz#-cgg7!7M96TBc~3+&jq<25+8eV9%gO|IDI|UN5b57zvgo&r0!G=*vyVF zXTbgPC6j;FrF7W6W*z{(w*Pu?08Bt!wZxEIZ~h!U@0a>?gT;CKMejAydsNuy(3@|K zS$#&L#-SO*FN0(B6NJ(iLTC&QrrvGD#1)sGzx$53`3g|_4dhbT9nqp z4ow;Zm9_vw0W)GjGU`?;%X}htD_l`pLR!@IrgmP)5^C%-U|TMPM30JWs6G>R!0D_= zjYt}P22g?AlT`M$kbktI1-ZE%jjLU>u=m!y;x#i}GRSkM6Vw9IpQCWQdRSS3+YMY* zl1w>Q=7ng0)({r3VBpo6CJ+serSy#~>AWfB)p50MZ^{EJKiNBH|DFvJdp_0fYX`9Y z_B{d4ehR&LY{JNhnLo$C5fp&~Q&nt4&UR@~bd(bNac zMW$k3axN*7~wgS$y~99~=%(l~`*XJ7o104*awE zD>Qay8y~zQ3>hs;vqE4#5F>P#k<;39Q)>wQPn~-bW2Giz8Jg>HJ?zh8#!+vblU#(K zdQHQ#Z841aNNK3i1ze$54Y$ie6y8$HYTJv!rLwo&*+fqLZodQRa>p&rqkJN1Mw4xB z?-Y(u`4?|!ZBB;ARCvxvIoc3#+C0o0_KZL=Wr<<}VN2E&g(4frnQ|7%2{E3kvCo6= z(B>HxE*Yl?{f*HxROpDH7#We=4+2nXfTiBIeflWK-)))`HeL+~a^yVC=rr_t@ zdQRt-;2_2?)ax0bpNUoGB&9JpVhCz4NfmsE>ZeItgd#X@n5U|6fD$VhkV`W3t~xi? z?jH7>?t0^W)5U4)2F#85xrs2agQoM+-DtW}939%!3&nR2xcp!R+Li+uDQLCEnji(R z?qlFefb%XrjdiG`hjHTIf$ZakMDN=i}v_gVJ;rV7f4vg&vcuZ zt}VWA`M-8UU5CRSz94MBASM34Zl@#w+wjdV4Zm;Lt_;tAB&GXEtqFL}^TU2|PRvq( z^TF(|0T)fbv_0~X5OQCL-BI`hXuCdaBj6Vw1l!J`4+3y*HI>Q}|I14lHc@v= z^cjS7wQ^<02;fSxI6nid+jAAmZ+`|L*z58QPW8ceLfWY0mv!Ahsh$IujLo)#Sla>N zYgf6fX~Qo6fTVj8uLG2QLTkRhoZmCl$)5~SLUbt$(x8z3~~4g4iz&$#GJ@ z#r&h0>W@(+xnD*4yCU{R6c$tVxq5;WW~ie&jgeR`+)=6dK4bLH5hKI@qnY2w(G~ri z?&p}6A+PpRjO2DWQo}4#6-o=izwI2nJ!a1XZE<^&m(M38y0OakCRBH>j!f0VRsaA# zE`9y7SO-SlZDM0K@pUm=IqgnMM!SkT0I%;1bWMp zuqDcoe4lA7<>HEZZq0|VnNWI?)wFk0Pm|NESEgNKxh~UZ z+ouogou5~@?-=}WIyVa#IuO)>*x`w{{ka#Bx&Li%>e_Umd72{E`0HVw$R(@E-1s&R z$QhyvUsh&Y%H`rkF?(ZU&4mC+jY(I$Fk&vgJ@%O%qcq+-jSE>y@}7QDzm-!b(j?|1 z=j^I7=X_d%8WsQMtak*XriGeuXF@xrwpDzNg;fW|6J z2WTF6OHcjYqujr+LzfHGK&PAheAnvBVWOz&F8Y(p?hmUj^?CEvNQ58@dl>!Wa?!7n z3wh>JQQ@`Eh|i}PPgOH8O4Fau+1jsE$0M8fvhpIp1`%pgO~5A3$<{}qOE0So(52m# z<|JQK3CVbWwvTvw+IA=Vbs@x?(<*G=Y4YJ*n3b_UJY5#nE|KO{OFDyDB`UftCkS6` z-_`SY4;y1aF0JQLmFDUL%#(4$$d>~=xj6~;gHvEAt^mR*BvU~b88*+7P%%Ie%B;Ml zO{>+T(cmo0`uv^Q<3_MSuXEHbDl--*dAGj_TH$`tniQ1eVlZq;nMUlsTX|*zb4OVsY-u$cs%cE z`l`0iEvWM?o~+kSlbR6uN0pxz2;k!TbD4NJ9Vo)?9Jq;9g1(?2CEv!2@q#XMYiCYz zL0*y}SL}^74|(FSUtGMZ2R6l~-2c&MKuG9X@3N{n8Dp_16K?`DA}cF6he2r)%DO9~ z6v;c`Q#j*cGIK+rXbnETS*BPF|M zM>;7xe-iuZ6#pvES%~L{)G`b`H|1ZA*O7xo4lF&#+NUTbw3=x*`r4$~LJSF>9jbg* zOieSb$Ls?2+P^TU3T&6Vw^XJg~csgVo*zFN|$-G!F)>@Ke>$0%LD31l3(}W4JWb76yU-qU6YH}IGQ!M}mG>w$HQ;dw{H6wnX>;U%BUNq(QOgaswsWk}jT2IuEcbtBNbXx?cs2y}Dh zx*tL1Elovv@yK-G6LFhOF0Wt#hPi}`iJyjQ7+T^T$u?xLt;lnV-55V_ICNyxi#>RD z8y@Spt7ToPu~~wnEA$}h)eWk*Fh={Qt`3(aau4!!viRNYor*2yOAvo&s1u_-t$U$} zKC>qpsLNzeTI#qA$odF4N)>RfS?53WD%Z0zk*V&fW$tTWUg9BNE ztWKCdIEsJp?Rj=s{MPytbPyW*OJ>~;aVNNo&~j*(*83CInB@9L0jIxo{O?T*r%kQv z;G_mcIpNZYTVA=^$~-L`65{OULwuIEAe$)$kk$a{`1pv)eV)G|NOylOzDys)OA!J_RY_Xii=m z{QKJAlc2f9$@S--0q@A{uu+8@9eeVb!vd@b=vf4mVJo{ls-5Z-JfG(;dpPULPT2j| z?rC|VX?c**@2!9IXRCjU_kKmCMzQ$O`bU4C|HbLP{F8^%E$Ym={)h+a`S1`3;DrQf zvaM%{VBj3uCqB0_&Gsynu!w4iGyK0W#H(S9OK~ehdHW^8nfFF$n#Y4ly0n$ghB$XF zvH8nWlh(kV0Pzy@AIhj-uQhc^vvFSS0Hp4E=fUI5z0Coi$7vEvxe6VMI~PJ-$^|Vt z-T-cteQkU4nWLzWrN~wU_ja2PK-QEj5DV128MF<6x!>+<)9>3OzgVm%d}bQt zE;zV){JBkC2hbt5x3%v4cEbC`*S19g+0|#DA>z6@_wF$)d-^=o)}0C8c+TTa&75KI zL0eq=Tj)z%&2Q}HKYcF08m#d!H3oT^s()Z{F(&wpf2Q&U2>_zFLlrHLtU#=-~WZfk_Lj)d}mhL2CaRLdiJx9HU+P2 z!uqI*z%0!zk=5UN9BB&r8N#w8Ej6uRbeFUg)URo>T&Fw-T0j9b(ht71JuFUsPf9ue z;d^B36*9FhXY~tEE%2)m8@jf^3;4PSJ?wLxin;5}yJ*{5OPXIk5UU7`rnfv+>c4Zw zk*c|pRU5;&<6dMoz-t9L6HVcZHF`?=;(|hI*5QYodE#} zP3avAp%(*$(2)+IN$A}MN(l@gAT3BBogyWa&;}5YARv%X0s-j|nsfzWZpQh|jC0od z-*eVo>;CUuXC2mpY!>_7NjBNx!18fC3RokaUuKuOa(TA?ebah{MUgcv z6M)~j@ZouaP0rO(pR9?47^G(pU^#IB*OWZbIOy*enN| zWt+~BKKtAm`&d`RNkf^pBm$C#M00AKv_b{aQw(e}>Gx~Q#&H6IZ!7L3z#-D|GmUS0 z%yg@Q!9Y4YT3kY2VGX;ilxq@DrGBv@Rq>46*}9%oWu!anj03ydGJ$%U^!}!8X+IrP zZ>pR?UF68XkKBy(vBfm{tt^YJFxx#6;?vD~`HC_X(bju-N03?7=~{hHuk#tNqD~`f z+I>+G2Rgv>ApEMc64<_#Uaw#n+jXJeMQIpLKm= zUM3N7y-}ItQ(#QMwtQ4`tKNHaI@Gr8(Tm+mLrOsCv@O*fXbS@~B#j(j$;FigT*Z2u zXAD-{7_+5%KhMwYHMb2j$-i2)CQy9kbKu(Jv^aOPfUI>{s##Uom~9xV?f(4bX>^Kb zc#(Um-IAPdP23MPB>7B53p4ys6+efJbSZ3eCXYlto8qma=*DYfWPqy%A7v&|E;ujE zys8h(7s}{v3O<*ZwEj`p5zXS1SE&Kwy4GdF!Ye&{}!YdvtNo0-2w z!)1th*$*!k&8&+vQ<6O?e37pEmwXZwIy;N3DOylb(_cQ>dU#671%Mz!%X;f>ArkhNqN?w^UWmh+QQ%ok&DfdP~Jz%`p1H9Dmc0_zw&Vd4M)j(n5GEbWg zy?I;c3@U-cwTalgRCeWGe7fb?3Eo6)Yt21!!(noP4o@VnbBa3j7UZ>#MWlDO|s($M)iSYJbfrfaVGhyaJ3fCjQ!BixAcAzk$`AdJ6UB|kK^ z$)>OxK|mm)fR4y|J!7hiadrZ}6F(2Sfi+f!7O6$lY3~W61Z%yuM_llUtPWbA0jXv= z?-B?Fhxqsn{<-mt&LXX$Ro5ob7&InGrIz)GEPWyDYQ}!@V4TR^lfT5nUA&?4rCJ}J z$3MNJ$n|KDoq?jB8=^=ymJk5LLfpku=4}lK8N>cwY?JzOeeL_(cResQx9DVODBnqT zWg%-i<@4E-_d{rG`Oti$sv|2fXJrUP(ujZe{6w-_`32iXj_Yt3pxpj^l2(`1m0@d9 zea5HR6wsKMl)%+}ztp-2wPZs;n2c5nu5Qx_zn(FoV3P^Ac=M%R4%g`d>JBCeI`7yxe@oJ+{})v40#x;Cq`IgY`%n-AL^L4)scnJug@`kS*wNl^kd>y!wm<}=l8 zK?i&ndd2fgm@%kwRm&eS=xWtCc1?jC;X}%QvGd}c?sPnY2shS!_IMFh!Yz6CE>^rW zaC>5j;}TICI2Et7H9PC@w1iZX#pjX`Kmuj(M%XXq7w$Kv-?MRFg6^BC#fOv)4(sh2h$>t;ERSf;7I$-5 z<~RyqAJoMr&dvnhfyreWNn%!O^$+v$R3kkr|KKB+0U~Hf3JhCkA|pIr8nS25jb*y$ zSbloV!tym+PW~gT{rK!>ufL;MMU}r)5S}JFS1nt!B%ZL;P+cycbpH7_KAzJoqi=7V z?!5g_LFU|d6>0w@TkoDhbTH9H93@}OM!I)=#q)n>k&iB@4wCsJw!M+ZbU@ZVJaLP8 z%VUxLnPq-`rDxu#+|If%jyH_=(08lk`Mmk@06igAe7Sl0voLo@%q~}LaKTeP&H9+n zW!D4TjhQt5cIrNk8-4g}?u7#Vdel-mg$t0=Y+nIXs zB`tBL<>2`e9}@WY5ic0+U+BKx%$#qq$&X0a{+kClbbJ2ixgSg7QkITl8Hd8GBRoNJ zUhTU-?enfF@1K}jsEB-{Ai`b9@~2mn>1R_iSmRaR zg+*9=QNPausQWtTWGnXggRR=I!%NI|?1iPhqQgfIwymO_3i>Xo?(nn}EwKp9D*Zc* zauDc<{>e{c9y9pM@N<~Mx5}~b|3Ntx#xG45o2~^#xH#NtCvm5?9StJ*g1BpdI}$28 zQA}4?rSaeOx9`(kCocw$keG!s^$Sb-SNnakLiL918D?L0Gz+i0ufV8ZzObB3 zy!Gds4sZ{-K%*Iq1CDi*)0L|lo~`)^uVuts10F>8MLAe6-21LH?!iCj=Z`Ic&lM)$ z=h#fVjrRP_{h!3F|H>wBOu-*zO85U$ru=_Yro_Eyd-(mG&sA*sQ&Q=6(!8Q3BO}K3 znMmYg#bzK=`C<$RZTT=v)?_X`TiOK=;y;}D;dhqPHK5;F8cHRPdAKQ>|NblMm9Ahg zKRh*8ZE!6B%uhA&w&hfxfpy9!Y2U;SBOTc}etho2a+&4yU;p1fHoklK*Kc1R&;9wE zw9q`=-fTJSatP$bFTX_?k?2A+siI_6xXtD9yrtpdJ}s>&Q`M!8scn?s8htm{2m48n z(#$ZA+!b;gYd>wWwvc!D$Zy-S*@=(C4x{nG=F713%eXh~_D@;9`}ym>XG2JbzkNY0 znSCi!qK4&X(4#&?n*@p-L-?VMy>?eEG+L@yP8L1?y6=B}xGKMt$@^`!Z0_pGfrvuT z|C!AGx0%d>ZQrx#|BG?)4T#eGn{C+4H%v;O5C8dKoLBv3JGt=Y zZBIcpE(S@HGOnuYj$NarJk2*)N9pvND){8xaDB@neDB+%F)zl(|6GiaQz;YF=_->5 zmzFzQ^#@dDMq>Uy8HwhnlfahM;G}^f=v9SD~xp3pT=lC)2JoX>P)JzE3@Q1%2c@Z=L`D-o!9*==wBAk#bspm zkujcX&KsT_`TOQJ`m71^v%#P!?+oCiZRVZ`n>uR^eTnK@J$911g5 z%StOeF-q^bYl&Zn(Cv&54rtdE*E09-HeK2(x5>&hi<527+19$_N}@^0hTo!}V`c*a zWoqOPcb$r&`R{+}X)y8GFt?SL6h5^u?uxIpFjwHY&Zql)xPlDd1vF(FhY2>lr&&DF z^UiXE2bB;bnHcapcE5Wk6XNyG`e4 zlf0)vMy{=AL)2iCGq85}H_BuEkiGf{j*g@)b0N~?FXN)}W-Q!UiSts6YL)RBZT{f- zhgfW#xw!)`cih{OO!^Pbg(D?d{L8tGb5K}R|4vSIE);h@`2k6T^Q5OLfRI3?5Dl{cZxzT`od z{2eKaC{cd@7GFzE<2c_$)ROBGGpFw#5`9H$+YRFL6Wu+4k~5R->f)1Gmi*|~4WJxj z9(I%dJo`gpdu|g^J4gKu`7Ic`&l^kqzxpdVt}H8DBJeMWFtt*(9teMazeePw2_L}PTauZ zPK*_~Idf-s)^>&5KujKgfHg9+;?*-V9^;1_y%PZ|i$ET&?7)#NTHX$IUAqDmOXCW& zS=I&^lLOBohhl}4siK1}biVw-U5}#!&60%+x-{_}?T*7FrttIcQ5+x5BR45*lmq`h z?b>_RU~-+7{^*GK*t>CgShiBFhl-{iXi4%eSPv~h+5Kr8aP?BfKt-8znj%1%ljAXR z6UZedn>dWarXQ5ObeDzKoldubdfERvBp?2G^IB3dN(*w$ zXEK5cX6L1lUBTwawNXJ;8QSTBkv$rIg#({MvtqVa#=&`nAUwb}z(;#jziHYBgV*OU zp3kINWPZGoLkr40toEvTjxH{94I=WFXKIUTA593(vc6ax&I^yH0bnA%f_9as&8mO; z^~5jLy)iChyOo)a8g-q<_C?KqvS2-PJaNp+9PmC_M8WJ3=7v=rrg7 zqdS~eMG0~~gC1s&Mw_L;FL3qkM!#;N42$62XQ%{c^7?*?3o$HTRyydNa1}____#)* zsiJo(zB@gpG3Y(66*nu1%&aVU|2d84Vj(tnjEqEcKNFtrJwq;2PDPcd^c0GwMNJQJplu2~*Zmw1vp>kcAXXC53*~G%5#z!K+!?LT`|_dQ zLz83D?;Fehvp)EbkN@F+p6a~*w<`h%0b2A=636quvn2ywJTKt4W@PL$V1)jINQr z`z(R}-yZiIM)tJXOM#4uCm+J40|6T5QrA|*Ccj-Qq)0=*(yJkV4hxn4MoE5j07;jk?7=IpIS}67qrKZ4;_D)=~47szV zx(!OTj}H ze~&+(;rusJvqAT7%6Kg%WBr5`TQZCM>%MPK5DUx6Zz7L#-zc)xnTqV~t(Qy$|Ju)A zM|}5<3cF2wa%tm4PC+ln@@2z$CT0TgRg7)7l108hvEBUeUuZg(Zvzi;0ka8%RWs0G z;njg(`b!yW1N+HOjc3(>oVJgT@;3)Lq7ggn_<;B4}BXun;v$#fX^wa&s`J)ll zPTH~!p9~plxs-*(xPKtZek`}Tw|2G{Bax*N8rV#5L!u*9x6F0Ub`7)ua_hE40%k9+ zXB&^<+J0wQj~|h<7PV=Y#J2*iu+Zx-2{sN_^#s(eFACXkg!+TY_{G;l9$xcY6O!C; zS-eCh&5JH6gvUzBJqUy*Bqob>r4eA+%qWE&hi&%oDP1-IE8r=HL%jHr3%6+X$jQgJ z_>E6hNl4^kj@W@!STS~;3sp4sO6px+*WC*PLDubz&64J9$QbJZZE$kH5T3CiBBL$2<$b-mS@-q4w|dT zr3Zo8Awi*^8ZkDD0c%}6!t`k`#+oY8$gvu)x*MHD8hqahVHNZI$SjWpH*s=2Y2W~0 zrh+nGQZ9NO(2aYH3`$ChG0p@KN0E-dFJ}M^8yn+H6RC1##O~&h_b0o}F1{--OSMj# zGMcwntH?|9U80As0ZXBULkF4RbahENvG?t&2S2FGHZdS)@@BD%!r3|onF?UA$vGk? z`%o)iagL6VkW)ZF28f-b6OW?TEDOHRl^r^iLWYM-^y(kB=kE7|$?`E9F{cZE8vUMR z@x-npk~#vl=C$$S@({{ZM)(?Oqm^b8K0nY7+AGFUX`PE$Xl*Ti0isQHu;P8-tiJGp zxSftrVd|oHky~2}jjS;);in(R(db1h`@-k*i(m!@CW0}Zc7py;p)dk0kE2}RIbNqm zz@SisnoJlO(G|Ncb~&y6+?cj`cT;OwoAmdK$(4PykS}T@wQgX@T4q8{-05}D`@33A zY^pcARUut-b5C8f29MN4YA_0@62|UiCBB6!i)1=yhVJI=+!QVK%PNHs%`-#(0$`>Z z6^>SyriE)SM>$?8PPR~Xi*U3&;1)y*hHj@pCQ#F5V+%wFZKGZMtaMD4S8ZxK+nmeW ziT})tm%gSBs{SnH8=1YSqTD@cJ4~|T;{*3RC71nF&jiy*{MK?D|7`i%H7-iUYL@$f z{-O~E>&FC*DW_PIJN*<+s|JC2^M-z$YZH;F2g^HWRwRX7SzV`~n+Hb(xN(a^BUYZS zghxpcnfZ(b$K@BDAns4RSAU_5+mD3nOCl%a+!}{21z7~5;RCk!6OW-k)MxjFZO=FK zlnZlL<*2hm%J+kqErIM*If_&BWx-eP!NC)7#v~hA^r{G}b4C?({%MxYUB%oyKhZ>h z3khGdQytWYuJnK_pU#7a*ysq$_;qtde>QupY?dNyj9QaF8sxR;a;vl45FV6CkrjGVmy=8jjX zjoav&i?u9pc?ArPl9Pxc&)ae93C+P=+0|F2ltn-sX$4QRUB@;0{UEv}atd%~riKjEjS9I5-9@G&?sqd>Qd zCk)TYE%w6Pd@DAOu+NZGi6QoXHgx3{qdog8 zQN>zoJ-%~1s9<7Ntty|H*&ki4pF|oO5?O^+JtJss0hGB5P65TPYy4rR-u^W^lUm2< z`nQYv=A;eg8`TNx+f#Eth*_&=txO$DlzzVK(+eybi*8~970FYnaTDW1yH1PZ#KQxf40%NdF<)-o(THy=*Bf~_#Ml^T~Xtq{&}#ruTowG4;~16MjjAR0o2 zJ3h1mo7IJc89Yv)?^$pZb!J37j+{>Vs+0FKO*G-;ncDW zA(w=L$1nFkFMf}4Q`LWFIU!y`tPzLDefEQ@pc8Njb-AyzP9AigLuIOYqU4 zBXc*E$=K(K?Zi+#U{%Y{wq_;J}IB8!fWKXC?q86%Un{$P4ef&bA=6+M+wXQK|>2-m9;G- zz5!mB(nS4`a>KrY06zMaEkEZJH5cl>`-ar7L*2Yc%>f5f!uUYL$*Vj(Q`?Y0xfe|v zSHDzWw!j>5M4XRWe_Cs^GfCNDx;IQ(IR$8VBA@Dgs3J-$f+Gy13L;bOD{r%#ql-mYqX=(Lexi*W7 zq&V!VX^nf*j)z|P+@RY2j3eL)8IDCo7+;tCqj+Xl51D2SeaLQv2X)vZH;~dG)@b96 z-z#F6`YdWCG)_=WrrXK-HTBhJB}HvKtIbATDBLSg8z`3y+Z?;wW5Pfop=BIr5Rvzu zuh>kEip#mqh?n)1JQ$xZ%mDD$Y6Xn;cJx#r0C_{5PxUm0$wzDcmVOq8^U0aJKKWH3 zuvS`!h<|3epPRC@+a>&Zgb%K63c41dIS?WMp7fnS;6>N9m|+`hEID-K$~E&Dk6?Ez zs~=zjoluV4>iYvbb{aE*x!HZ|`)cmVXPK${wHUm5kKene z@q}u6?=KNhYVAtg&`#bi>EeoZO~aGJVT32U#cVuZYzpOLA4l!TFQKa&Rty0>I|$p# zExYP>5ALMNK)ZS8GQgq48ko>G{RFW^@vx`sASRgQxL`8<4X5?er3^r^`8LgkB9HR%>uAqttYg*A@;hX zW9I{Q3Dau=WM+FX#ng1Z$iIgC?^rN2#ET|vymvcJ!K@gWm_cOk7h^J!?gwvYoa+1R z-)PMk1s7E%9G&kmBp;Az;f&MfJ>uA5k)furH-rpBOv;rzZwKWARuAi&)!k88$IQr5SH0@oKsatDcwXOGqepqp)$c$@9IZR4PO13?1 zk87i+0i+kF-cV+z5Zt60L%K!e>JjK8U0M`Y7n*}ctO4&rvm}Ke)h|e3E9muVN=Cl; z+Rm0x-e^?^wowVNX%3vjKQA7#TH3nd_{@6hJVc5^YH> zt<_rQ9nogqv_6NGpZj@ZQZ-fa!frWA7;I#wC({Zxkypb$$9G$n!UVTIG@866MiA8G z?KpmCSzbPdpT*|+E>OwPi)o7`lMwEgg$#*j=K%k!9>K-CC zF1hXCw~t$zEv>VbB!K)d;eO%7e!SM&hv$jZbR(l$BTTpoZA$w9BEf8zDOkO_#Ed2; z*JTl95tE1n?Q)cIu=m!w!ev4I%MkxkL)UeSGT!HT zV7>K;2_5fsDt_2YVEr;(V%RIm+sU;rNPK`dACJ#t^B>H0qm{Osc!mXX_$9S4#AYnZ zxxbH3?J^#zJr9RNnUF9h$Wb6NR8;=UmCegGkwt25if#WjZDE?a2mO||@NWhC`@iJv zUrTi(MXR$TyO3Ka4TR_ZUD~>qEA=rQ-9*8BUW&PM?>ORLNit2F%<{z#Ed17gPCqBV zk+275_vTM%dU;BF6Fv({5A6jo4n|}i`c-SZAg%Dd$t0)@GKGWvIGDM?K&}7FK5o{o z_|j)QHBWg~+Wc*2A2BiUW&(vj-x!f6W0~Lix^rNS#yqVp&xekEUS?7OAe9Wrk7y9c z>^c@lefW?CAKwd7d-WZB;g}=(k}Pwb2QsVundK`}IjpX4EAm?L>97FX%Wt*Y|BE!- z-(ZCQWyR4`L%axQK+ktfCn#!cqQdi4AT8Id+s~Ce>sqXLT`VtmEY*5`QfqBj6^njX zh913Reo_x0dq%w6@U12Yvftew>B6Fe{bDh~Pejq$KD;-Lr zE1ai0)&7_ry3UI}@D6rImfEsqc>s2#6@^;FW>>(hC$>2?VGBfX6X9G~B&EEwB zHDm`Ucb4cWLaD6wOiwV^Sm05u(Q3^Z35%usUUG^YaptlM5p*j%T>c%KMX-qcdl zn$lTwu=f-flBi+r5%aLzS5GCSiE(oFWX#Ngr9*bgR1UR|q^DYL(lrCq8EU43V4y{H znwcT_MKGgniFy^$I~k^O_kJ#rCRI)$f~}w2>spQ;(r4uChrEygerPC})J9^gKl55x zNADcD(m8#a>T}^`%{mf3t4>n91)Nwj{&L!9N9E@xs6%E#UKO*R7Nvuz8TT*rva|6c zpuDIcf}E6-a@a?*zqAj|$Io+k4dqaB0G$TB2bu4Hx&&&JGhRA_L3VY7UPv~0g5xWL?)3@sa=)yY&kSMmZIK9sKFm20OHT^`hyYhy;aXDfl87py>3WU7* z#a51l0&pNzE}(=naNCUhJfS2bp-Uv}fxU2lm84!#wQ#}!$T31%rE_2Q7ett|xlfIv zI`0Bf+9AC_{=-Z8s9^-JcaX|qibt(ioEuw1PWxjv$@De3@NoPkFR`E^k`+O~) z*(`2JYwiT>wC>-)GseFRB29x#gGd4V@!Eq5dokZ(CM~BX(>RSs5j0*llil{w-hJ&G z>qy8@UvocrxEa43fUg_w9yF)ETSiCUY}}Uu`%}VF!ODQbho@mpc$KaD-(UHe#ZYg@ zxVn5$4iapQ6N}iJH-6D=yNB;F?C9H)#VMwLXbb!CH^< zgC71w#+S=W8vn01sJs3xMA<~DZ?DQoAa1Hi9-WP$1Y_UVMvQx9l(3O#I z)#j*@%Fg-(Y98)vF=%SJ7GXKar7o}4S%Vlqw?+qR?e?3;;Agv4>bGZ1ZGZ!N>NgE^ zB#UgVc7-qzV-kUYgxAxJ{Hw}gi&r+4=IJ3IgnE^MiIJzz?W(Apa51-`-F*R9tl&hl zKGa=5lN@BY4Fw2|Xr9;nA!Ye>sWe{l;DX0%I|*)PN9z^eHC2Gv)w^Esth*7YSZp07r-&ZyYbkB0>yqys;UZ|UZ`P>!Smrs_?obio=0rOUO2{Tu(d_~wpo(^|B&*{Jj^L`L;R-+R8@h#qsvy= zPG@0Br58ZpUcokJAT{3fX#RK?JO%^ox=U@jsbAFq9Q<#(1(W<(~O&WQDx2YmeMU;A-v1EYDLttqxP zqxVG!OL>8N&-VCS_caQZa_)_kCi&QJxRq8@p;4Ry>OQK=^MMw$(!Gjh5`LmPX*$$p z$xGm@jl@pyTslhMQYiLHX`rOQ>l zhys;-y$Q`j_ud48Uku*aaX9ET7l%LbSaU?larPTJh(3ld%O&16GXNNtc#U_AnzsQ| z69o4%f8+b;)klGoB%3{z=w5L-Jtfyi3GXe;DG~wG5T%$){VfIT>tmTusg*vwwH~xB z{U3eP&Gwt**)9^YS^bmhBS1<4_L1gwEbBUc8Q{V|qvL^^H5-Ds5Zb45`JzJhyWDV~ zxkdQG?6Xq)O3m;ARjoljlLp4ocSha~rjt!zBLSD(M7|7IlV>|Kq3v)n3oA-0s`Ts| z*BhH04vxhPmK}jS;%azuCxa8LsK07a`@W>4g24Tq?98Oj#iOcLVR@!w+*~W|`sFS` zf5RfsrQJ;#V2|g^?%+=43`n5c^OluFHbio(fb4v6-LGlJ7}I zxjE71YfQA1Z61i}-oJ|dUKq4ib&KKEjH_6Q;+B^#vPil2V%OmLye}*-%sm0kU7KCf zXt~!@?!9YKs@tQGTSNL`qxCkR8CokC{*p?@Cpk6n*lsaQH*O}k4GF7q4tL18M(2;2 zTickKnUNqW!Tpy^caj3Gfv59C)o4bai5&~ZbWN7{xdLy|dFVfe7A(e zSxug5fs;=y?iI5ui@r-()y}g@zCsLLd`&!Yex-V^7f@Qdlz3c3lZtPm6`_L!5I9$Y zmD>bNF{*4=ECt+psrs6ZI6rgY#{C21RD*~LII7=Tl>mghpg8+{T2bG_WUwFX|%^s&3Vo z>b3K$j#udn~oM$rDF&+6|npKbl|+p6MB2E=bW zf635uNE#*iF>WkB=={XQ7oYiKzMU!Plqx;mJQlrVF5diA(Yw?9U&a_FFnI+VdkyniPsH z7S*a`){Mvy7G9OOeKEE-NUoVzl9pS$b3_C7xO4YRW>$FlD1Bm-bG>ybt}}gGx&cmo zF#5?6)X>xZ81oz#=vF$`J9TGhi5}&e`N)IXSKMIZbUeTrMi`L6H#Kr{su z9-^danTvrX2s-qUPxX!VOiD%K%Q@KWvn$Mk+w_|};zW{E#eDy`)!>Eqr1cgP$vYGv z*XpnOI7R_qx+HTU>kn)tjdAt5PzmbmpYLUsiu(1ec%gzB>U_OP_7edi7U3h0W+}X7 z<7G9K^t%KbNnr%C5ROm+*?z)}xANxVGVJ-mss>(mUXpNVt|VJ*5cpuowGevE{1Vp~ z&qX{ept1wp1!->VaraIW;`gtGcKKKz7g$(;#!BSq^Js~+C9K$Ha{8I%)b3ehJxbmu ze5(3|Roa_^<21~Hb{aguZ%k65Ad;kcZ_29-C&d0Mb&#Y)&EB;FJ2`^tyDt8MVtjZ- z8O9%A#P6d^N;+4bI6mvsslplasJK&H>@k7TgTTdhW6INn2B^bl}r}O~d6H?#tWGk?ZZ_Cgx^{Oqt!$LfdDNv#z_WzfAzMPu(%M7!goh{!ZJcT8i%K)7M_iH3lPj`3OXEJd&B zn0Kj*>)EgOn~}9pnitBo?HR~SSnE8&wpl_@Y7v6df;p;GV(;ulm6adx*5GSR3=F#& zNL|Eop`N0el)cj`5UY#2bRSh!TzQjyS%rk`1jA`*jN?XQ z`WeJo{PXw@ung+z%<2bY4u>zYGtuS$C1CBmfP-7|w}7>OCBFXRI^?8$iJ+z+^KpWFaEu?_B?vJ&Wx5G zp8xz;(hM%?9o0{54-fEPcxHo;VN~@YEPR!VrrxU8+SqkcLvn`vix3FX85Obn5mC(O zvI|k-E}Uxidd=wWD*(}TuqGkabJ+s5b5STWxQd|kVMN-|>qWvlAMmaHQAuvT-G$74 zPgInVkJ-Lpd`o2G=PfNQAQ(KCZI7fDIPL^5TLGhJZZB)>1mFxEsjUhK4s1ISu_R?G z;^=qQ`(iK3gKzyweTQC%w@mvzYLuj%YArIq3EN~a<1nMuofY6wH}>|7RKFbX(9Zsd zwc;)>fv-hvTiNE4#2O6qHXHoO;5&s&v=IZKBTXcmM4thIkM z*>RBP=tnD!<6b5E`nj%Tgg%sP#id*b3r8y}W872JJ+M|PG0xm)Rxge8f^m{6#+jw; zp4qnbnOgSO8|n=o%qB6VWnVuzV0rb zH8RE%Io6z7GJH@|U_1c6Ya}<&v@FpJhWg5b&3;srud3P59hK5ql@d{Z7d;f{W!qs6 zlQb1XGSv-gKXMHOrUx>!8;@!a?UqHJpq8=1Q_kw9XhvDs$G6$~wlXR3Z_SP|>NQNT zhvDIWz#i?TJv!STS#c-i-{VK6Ks|Nffm3#S6c{d9B@O`kRA+2xV z{xa!OIo?GoRhdZWAS2WSKQJUYLWg)GJog3imwm+U8gQ14U(1l+Fr=7>TT|Q0WN?f+ zeEF54^paXYkNRZVieH8!p*Xwh&W_=9uWccP?IO3B>LCWN{`+^Wj`0|{#URA>Koc5cZNUO&<&QLcBaYP+ zrET~j!9lLVcbq&pa>}pSq61=$Z4;toyOn~u+6?59D(mQ^lec|NcU8lUPA5@wxLO4y zyTVmGvB;&lM~8S^7eXn(kW&gd2$oYdAPl_lU1sq3>tDapNY4_PDR0E3^^N%=ipFB$ z3i+j@uZL9>g)^yn2I3Ql`fh4#)EpQ*`&g(lqIzU{v936Rr zVQ2ffmK8Q^=2eG7j&4@u4T9(C@yELwyTT67X2}qMR zoe%++0Dbv4KE$dM^C9;syq~iq6`hI;i33GK2_kP&QN}C#8U0Yv??WVlIlQ9Bb~4=Z zUJV>Pb6?$QV72V3K~#XOv3V|ZEu;B92qF@yRkQXAC#Q`p4nm@K6>LzNQTLjGJumZQ zsP@es*rvo8X>=z6$r=~IfKdTksOm^!;9Cm zSER~T{9M8vCX|jJR9RizI3@B8LFc>tzrDyFE1TR)usELD zgSLT_wT%C)cKJ1N$#2OkTceCsu91w(q#fFbXLa6Td#+xL>kU?McdA%3e<81clV2 zWea%B*1B$w?yRw8nJ)vfQA12n?0X#QfCbYC{>gOQ-I=&HyIP-L6w!>H{ zV0ubJttdie0MZnMwSrdNX}~e?0QP9hB}O`)P;w<9FSUX*rV1Lx*zgA$yC}L{TqgPW`IJzDq#q9% zh?+>vozHNw*z{7gPN)NSLMqgp8}h}++ujeC80D;22jy6T4%19>oRf@R-@aYFGJ8he?}Um`2aoeEPNFUG$WfeIxjC1L5fw0Q2jxkB!4JIiJUOZ0|`i zE}<0&Fp@O7eY{P(?<^Q`+6U#7AYD}mU%6km_-Ax$i9Mw z3^W3pkapxtz`|-SLy>7Zf&jdHIS{O645$NsFs%8V1ut>X?kTU>b7{*2D6DpqIHbl}SQi6#Z*^mG1wKXsYP@sUQ3KH}QUNb|Pf5wV1s& zH>G_42l6;&&zFcmlNf8a<@a_zIoQbZc9E_Rz}wBvw5n-mD8X2mIL}MXzGLLQ*xCLR z$A{;`T-qMoI>8sk6M}lp2xskZ#q*WOr3L1%h*L`=Qn`a*`N?#L(9hJYl{hpdsCK1@ zc0lU3&{-GZoh*iirYiBil2@!=?=Cb``;={bMXHSd+;>>nFG)djEZ%3IH*&I@Rk5%f z>-?o^{DzWM{Qs5__sU;J+y!o)R$1zvP1!5j(E-)1FsG)*t5kaqgyC!-0Wfsh?jG{=h->)aK>)G$tdtR9eq-Xvc0-WBp5VM6FPKQ+Y7=8R{&f z+jqH?Fdpjh-8I_zgk}$~Mg_u_X${)LG&rP@E))VL=Y7nJ(tABN`HpDZSVX|TmWg#L zPOTsl<<+_I7V|e55VaoyhX=+(iY2)AM^~x~I>t2LXda&3uwKqZ|4K0%^%FDBcrBkC zep_1iow7{cgZ#VvD5K6k>w9M``fFnFK~%#E&x6z5m(Iq^YxOTpr>KEA>2>O!tv1C+_&)sfI_noQ}g%3y+A7yVRiEhmh%bP z)9+vbKx3qERzfa3uj|8^^@n2$6~ns!Wq_`LT2$K+*G3Sv&2o3p65#$xAe0RVsCssqnZppC^aHCK?DgVulPDHVL!c*E??i|&MAK-EnG zq(FZ04YsDdOpd^b zd$ox}*|F*D%Dr&Jf#`X?dGPfn%xy`FVR<%rRq$4CXr_XE^yj`*kY++s0A1S>YQR1o zbt4sW*%dnb%| zjV(NUl+@T`NXyNS5F50bi}=6TJMXxrvUTs{*t& zgpQ~qND+`OT|!BdASHo>Ho^b`0s;vgln#LekSYkiQRj?$@0mIGJ@+m5-p|P&VFPRL zWbd_}wVt)s^L&5h&)JT5__Xk{9m%?`h8*fFte$J5OBYKlTLSDL=E`ko&ZC&L%NCK5 z=@y9PmDxw>@M(pk9G+X+r;u+LkaBZ;5Uq|lX~1XE*B*>T&rcal`I4GFYp85#HW*iY z5vAGRQry(3`1-ch&aIqkaxTxPxTC!(z;D+7P+@?Bsztk2xrd7Wnzt^cMux8ywx?bD zxD?15Rt;(1jLF$tR^X$`xBDu%sf6B(R%ULejU@H|`dh>KsA^h@N&pTTwDMb&`^;Wn z)g%trkOq0a)1(Ops?dJS$0Tw(1Zio^0;^_IG+Oril+#7M`!b4Bo0u#v=1x+tlBx!) zXi0EQZc_lC@U7Yb`@z1t9%WMz;WWb`zWo5UeyIM8PR|Qme^bqicBK;+Nd_dX)Ep$% zDmwM3LjqVO;gUtw+aDL!&aS`9ncR7r3!3jH=d2b;yVl-tHZ!nzG0I~#mnkS@`LKvL zI$L8e-o4r!kBeZ&g&Ik$z8FB-y1kn2^d4tRdzk4xBXlly`UVxwV1KAheFg0$wBPrA zMvBA@Hk}p=+cc$_|3;jrvw1$XM|7r&^o`7F0WrPeo)is@1|RQt7mdHOoGcf+L=MxI z`y-<-<13Pp^61|s87UW_GVpi+(>;|~y*@B$3kJ-JKcg6h8a91}JJRd!zcqlcFIw9o z-+I-jt&(r(u2l_)nr!xTnhz>luN~zob$dJ{r$u@K-$SFAG(b(cD930yh@HC!)+f6q zTGaZ)x?M4$vkeV4>{~c_9P@_ahjk!P6BzlTsL2+wzNnlVnv1Q7`QutSlpSmL-iqd@ z$fwaoQ7e$)u_gLMP;;!|xi(u~5cmPFk5ZQ-HlSO{ zVCNRcG=zBE;X~fJvnJ}2sf&BFcz2TmZ(>BZ)>vZQBp}RTKdWp!PbqadhhFKr7vs?H zVkvxZ2wniBxIhwhxMd?YSlM{HjbBy6ou#ikA=7URJe`*<^z|y`o|oOkd;7S96(#Ts zMr87}5s~*((zCT)H%3C_h>|>I<~T_XgDz*>HMoRCsHO}^oGgm9hcg{i{86ctfxQld zfdwTa%xQH*6Pm; zCF8X`)Eq<)YKhiATxol&`TPgepy2~*GU4Llq-CpDHCU@m_+SHjY_Fi`7=Gcl=1LA@ zP8hD%N<^TJhj;k ziI<0$o0!SHYIG~fSRPgpJ8!;85mpGEr21wKS65lpq)xlH%ZC*Vp5%N5d|%<8ZmZp@ zebuPP_d@-BY#em&P>XpnPT$|Mx1UOS)U8~0EXrcK@y5JR11W4tD)g=T=uzalg^Ee_ zNMf%wT-z&KqzHYF>v})XJTkUT^q{^izosa8)b;TK0y7lxa2ZW`CRK<0+Ij?VNGD=|Q3F(tEUMG+mvr+Pkc*piYmm>5ev0 zH3tPzImy@|Qb*ul5kf3#aaR8uR9%1|0R-Yw)peDI;Ry2-2AQecZkloa-of<28uPusU$ha0)0+*BpbXe2c8fHU&JOKHUhEJp{j?(Ex_RpNB# zuSwTYVP$1=0B$47Kgjn?O1#epB;t2@yi50^2=26 zKElmJbGL@krCkLv0;M;Z;f&+E{1`?TdSd9XPtDMwCPZN#g`E(WVQyA1vXr_cp45y+ zqthqDJ=hk=`wd)pT$M#;T?iwbaN+C(Bd`WF2;A1B326_QytmSP5#l`;Yo?8yN7O*h zZgZ$)B|yOSfh$N)ZLC5JH!eqfkYdf9WZ`n08O44p*_0dOJm_5k-lVcIce@{w(0Hgd zB(CMP9T7}qh{vXDo+|>5cJ`NKK4o}bLrWdOdlv`cl0v~prXcIsvF{rY(g!3}_o1bR za_urrYprksP_xVWFr@$?u+J~%xj|8_Yzsf`QsmpAB`$;&^ZaPejsxH>UbjZLc!GbG zsQEZYNYSdz-gol3bcuJ-a7i=3<6)U9WOi3j++g8Gp=dF`i z-fZP34czVq5DFWx_n3}ID!F*I>Bdm1kX1hks<4^D!ND5LmGI7A#2D`4S1>Q0&R*X0 zN=KE~TLfgU>Fi-kV1xnvHXW(d-Pag<()nYiA<2dt@n0@lHKLA`jJ6{DI#fY2edbtb%Cn#9ht#sIN+9Dc$Ktw9 z1!EzZBKF-odd~`DLS5jN0^gJJL9O+*aXQttS?dV)>|xxf4nIBT1`Qu1dLz{a88@;mhohA z#@m+CXC_dS&4tWOys3~CZc?i{ikp%13~DZq@g%7Mmw1SD0|};Bu^-D7ND)mm2ZHdtrx|KRdLpqC+qUS%{4sl zJL-MpPWJ|R!03nGQ16W%taZXn2i+9D?VrLk+=94_g$A8Apk?bJ&}%pYkgyxO6gJ^i zM|W*x?mbNS81zkK7WHsUtw81tuq*ebrBAjvLMpj(6mTmfwvrjhxC>4Jad2mCZMF5A z&x?=6)V1EK>lT*02CwVCiPIGUq_JtHBuR^78F`8PfT^t zzUqO;#fcE;8jzH^>2}ic`2E@63SiOLam5uYzFWMJHN9$9N^An$@cPcRxs$2wrs)oY)dlMQ~1VG4Nl@858Kls<|WGq14^A+Z zsCbiC4WKA_)$-w{oME(M37Py(;HpHjly;L2KZnv*Lg#D~H1OuJb0Dq^!FO70m@{z= zGo|uEvvN%{myc9XegQ5gkuMY&h5c~)*hH27WTKc+AEnF83}^|7OARdWO{^UmnYO@| z)S)KtFB;wXQI^jV=RMLZIg?}Znt>z_F)HZ_!uV-zix*E{#%VK|G^}xB>o_tld_~MC zPUpVHRj<)O8Op1_zoPBbUhmNvPUk)`zr_n>{S_!S6C7?7FiC+{LK91M}M}O$5czE_Ww;S`G1-?(&tD1A9&yQ zi5kO%0W(T2T+8+o>XzyuqWIXx-QKP0C+mr*y6?OTsN9rYVr?^*1UoZA`s*@;NZ`@b z=@apZn9g}#QoNM5L$uFC%)QdS{^Qvw7w+hJM=YiQTSPWoL*Iq+ar;K6Lk+LjR4tpl zeZ>Q#^tx&rJ7=ysE&=7qRc~xXqrOkP)JCzSQyNOJiz?HxIc6(mOU}|JZaffYNxo}^ zjP+7#52(GTEbv^d3OTc8eFDwg8=UP=gk2>Z6PN#m;3UcFlG+7^Z0(+7E>_y748Mqu z^}h^2EH_An)R}@cxa_~l`8`|@^?x1w& zE6n*7FjdMwV2V^-79tFXDguUVx3zP!6MFlleN3wmL_m}?jW%p_TxQQ!$*mP$C>`2vuKESj2Xgq?Raqod8+31POBpVhx{~aacZAEP|73 zloLRw0veVz^#ctfy4Tr*e;CL&s`p<3(f*vQOkEGC>jo#ZsC33kN{3%9y0uxIlc zw?K*~Vq`)730@su6CWzK7yB z6XNS0l6jqvczexp%>H0QHefue6DFaHMb9wFA^IzmLn1R)reykxIUPw?UVkb{Y}dZV zQPh9ca3C={mf3{qlE zWe>Zj1dERmt|ibPyI3MH)$3_=X#O19c-i*n)51hWoOVL^w{~2)eWmeC=1gsG>`Bj*YWk zKUCQR0YaX^nhx5rJ5|5t8Lmq3C-;i=)sRvm2IWLD@_PDWgY#yyT@|#~+zZiA7`XoU zbB=(?3urW{0R_~Q$kasHvn{%STAZm|%9aWs&j@pjFbZa%C+N;-jx26Nj$g#abY-MO z3;<_x$xfMC;oWkCOHK=5vPMSUCE;sZ3WT2Xo+KMft*{#KT;cYEabX&jN6AGh@z+87gsMiNTN|LK&Z7Lh{jV=)uo+X_P?|IY^+-orCXEhb&uH*X(o=1N>q7hCtc zGdL4N9+VHYh6mS-3EgmZPim}skq?0|cbh2F4I%+}XaIYE|8Q}2{fZ8XT@pXz!bT$) z_+e7#p>9ul*V@6x+xCsqnr#eb4hjb1j9s35D8R$g6C@6640}RxF`LU`$EgZ7&jJ&g zV=ZT~s1wM^{(vp+j!2hnRo1|gq8bw&OPgx&Cc#BW1BUc!>}E1CAdn9%Pp!pCc0SF{ zFR@P0%1~Jm+?r*zWW@{0?ciyKLnz=j*VEClh1sPH{Ctcq#B`9b&e2m=H;fkRUM6C| zLAgg_szyAu#$gXkW*;Z%E^_Op6vsPO-f-mbuWqm9D$v?H^tL?s^^6m7w1gg1+LtMk ziA+s0G83G`oP{(pw;(jVQ6@LhkKY|@o&60i9vntBPYza~Zo}+~D;#m=(AOm^`S^bJ z>;zv{fc=m0fuKN@b4J_kU5ETs%K4m76;jm%%VG%>w%WKrPG2{2BvHj{{vmQZN6Iij zskSn+-8^>9Y9cZ+wgIXxAZEXq>Buf`$yMyOIZO=;vq1N#_yw6k?M8w;y&Zg_E8~n~ z;{^c*nCuQb#@39~x0l@G z^BE0$6ZDxe?b7GOOiM-J%Y8YUiNP2-wb7fX>Q^te?ea0M94pPZ6Btql-zHO@3>uJE zI_SC`H7wchg0EiL_Db#%=0d6rj;&jqtEZJ_j|8kdeAE%`mby~3bP+_0*$vgoi91XX z^wp6aSPY7#HgB*7mEYqkk&nU=GE;d*ndwfkW#W@4E#igVXa;@(#0q7r1ZcZeFj{y) z(lwJ^ykYChY5s_kQOh40Y`4AH(_+vgv|?x&RTk79R#it>^DMl4tn^Z5!mLV_vq+#1 zxHyfhq)&)-mj@OI6cg{dmn&f_+`^2^r*;P5;gHQOh1c@6ueOP*&EP z&}(>vG|pSphNO1jC-DM;UtUiXZ2tBX?O<^{lX+%io3-KfCVB>oQm=iUzkO*ZTVaSPr8&m{-7>}{P4uU1^HCvD#?g$^ z&d8fD3iB(kPnWu?U!7NJOzN&Gmz{ulsD0m!H=u5Hc9U%&XIrS+gk~7YaB@}DrLkA2mKkg6Y2tE?=Tuyc3U4@Ii9g#*3wlB*AFmFG+ zyA<68iEU5Hz$*qqBIb7C=(TVkobPmB&{3<%LDt7U=M4*dpd5Vh=YF8Vt%gn&5?ftK zug8+Tixx}4S_90Pi#I_AUF(!E>=0S$Cr@7`>h&Uu`rT7fpBnLL>WMjVa9jca&&pYp zoB*J#orLkV^cDv_{)|xjo(rJZ>k2>X65X*PtcmW`X%eD>Pb^9vKJ(oh+(@t`mc7+jxMB?%Ul)= zDxfuj+T(O;d>VY-xB<2D>zsr0@%}q0nxXI{p0_23P^wHIKX=agYo=MPE&LRRG=!h^ zM?7@RURcWUaOF1tuN5iMq@9_dZkB`pFS(5Bf68TiIylp_ekI3iTpxGPP^ocv&9V`x+}!#O7@~SwGWQ*QB zJlq8w4yn+qZ!8WdVbX`CWQy$VD7x~$;MYaSRb@=_ooNe^6ZbPJJo8aQc5v|~35!it z^~=vlBAI8j2BqHZ83CBLYg__IwC=aAndgthg^#r%q}j#xD+!KEN2X_|A%1w3(PiwI zNgkEk^|&h3%_HaG{@%MhXJNQ6N6m?XHgewB#Rd;S2shT~NW34;?rCAc9@VD_D%+@L zk#g2^>gSxnK#jJR1sF)xUY`mD>s(${wLYJyA{Cs%3`rdRt#mDC2w{-e!VyP~qlQ&J zzkg3WnNJ9*F-7q=4+MD)nf`2HFJkJM<;Lc$oD1){=23zW=D;3&>B|mU^PE;Y+&oU(SLys6ml$MI}1~ zQjFlv-%MH}t!}<5Mx>%@X$vE1nY+w{^RZEig%F%aIinACz)9`{Q;>-%Embl1}&Nnl$x}#*aJa zjy%4dXp$;_bK(1-ItmM?&sW!GIo*>>k7VX40tC{x!MmkgM1|?AV5Kmf$xmsD?>=Mv zPfpJn{%ymVc`|R6rt!^@J2R>Jf#4wPi}VRH_%XZPg^Ql#07mg{@UPbLU3Wzc^6e1j(`;m?T$JH|hfU$eMn2Z`NsEBQloYhz9{xE5+;LLlM9xt&2@@c1OX-Ul z5?3*YB0%65ce&G5$|KCJP18%&ohTjiGC@C_PnMStP3v|r3i!_-~`vLv>SL|G*oyq4m+!*eE(!LepayS-KN6jk^@ab6_l zse7;tHqvE@JJ)dNth%N&aY!nK37&FJy<|n5Vy^DyTy(2Cobwv4W3AtY1)u>>g6zv0 z1y$qT0zARAvZ9>C!xRWb^T4Ibz_-`eT+FkvxaTapcMB?L#R`5nbG=(n2z?MBRQUqn z@n%U)OzxdgZIV4*@+kjPZY#3VbQ3NIxm+)^qz$R_B;6#lg>S=WC$`k^~K%bWf-i~a|_si-c= z*a{;7^y79(UNK+bPlT8d!1)QS4*|=a0ixC*T)AsG2Ip5MkE*a#P>U=a+_>`uQ}nD+ zP)AksCxDLel*L;swd!c>>iDl%G&5s~^7ciQi?D4Ln#sM#-e##uG*Ggr^f&eQ^(%X} z5bB`q+q>EYl!KhOyC88Y3N*I~DB|n7hhpn4qx{uj8vmwuVp~J3JPycfq<+s=xkh^ z`o42)0{mbf>T=EET6R~4^Ek)!P~tjTV1Lhz3#lgvR$1_5Ed{Y0>T@f%Fbl}St}T$E zdwU&a=XCHq&)|&-eXM=hQ79DFe(wh?_VF0_JKK^A?J6Om%D1Q$lCj3FD;a7L^w5f4 z9d3V0vaboz$pS-L6`^52aw4hE70-;lWq9a|<||k7)|1lGeiOT6`ZIwBR}5n8PPNzC znA@dO#8h?$djtuLcr{dT%R`^b=*>NzJ(PE(BG=wfQiwF59a z7r?FSOeV~T{)~Km25{6-yY~UhA*9=|l2z~T8KD1M?&ntuL?2T>Lw`^G{K|0uao*?K zKh66*CeTgS-u!NSrtE!_de#O1IoUAnS1c;cp5G3Ybt=*Ce;5mGH!8SRd}gX{qW*|N zu6OVJnFS4*_0Y^9wWoJZxj*^zmA?CiBJTSQVjMBIqqx2ceZ|t#8u%@={HVy+S|=V%9=-3q z)A>c_WU_xk?gNXuSTxI>&YsVOzmfo!5UgdyEsJ5^cdI4b?*yfBG~hof0oyY{MnxyP zROY^YM2G(j_QJb8hW~^1a^nx$%Tx|ngs=7TlQ5-q|FZTjGEiNA)zl66<=8#z@8+oS9FyMNbvKb{ct;BO+pdGB%y)rVPwZEr(3WYfL{vtAsty8njdz#V4skFP^~FIpdnvMA;( zjGtx_kybPvUi>OmcV>Nw8L4d=_K5$1Ue85@Dw;(q->7&uoarZ^w-xxxk*dtlg)Y># zNS49h`)j&7YKRHIRL6CrcjHGwm_mP(slmDK!7yYumDSNr)v0PYmC$}Z1uSHd=w1Y3 z3skf{40e{T;t~L+&4|6M?btExHW$@^nlEjNJ_if?%($3+SIZIIhm-=pKuLJK!#UvR z3AqJJDqd(x@)??>m`!0t9YmLum(8(wiz`}WdltRnZ!9SFoAe8G^y8R&s9l`V*-Z`C zN&!FAZpTc!^Gc<)tJJi7jy6@E97TXcRX?~M>#cqyrVkzStI;{#TxJ}9XxRw`d$L2K zXr5=tZ{jLv6&73SR1hTiPhKqoW~|vsZY4j#H{4vG)Qz2e?NtdDmZS7&tOrx0?vFOy z^l5x>O5}mvCb{_Ps5YXy&sfJDVOiXB)jdcon~(9&S{=~ zbyOH7v6fQuBoRL{kP8+#TqbO$Y(S-(B2jg4bMTcpdSJ3L=Uh znRaUnrJj3?#S6|5A; z3AZ#ekefGhuzd8pGRw2(M2M+rnZkL{rP@G<;n5VR6>kEYl4SL3W;!i2viRPEW8@4a2Mi`awmR?|7;!BU=Icfp)*2VXgiiqENe)vL`*p5-DZRj1yaGDY;Z#37-219k+k(DHs`2Lnkh)-2o|ZgZlycf= zrddgf`xrKZueJaCv4rSi@13o#I#Y%0nUCQ)VI`(j&aKmHn~2Z$RA}pHbL?*S(ym;uH$MB9Cpq6^&sEpp5n{%UXgh0epJ0>%x$i zOy&XBwo<|3u{#I*^0s?#ZZ^-(IrV}i(dWlU@V@#!X&E>Wq$T+I*5f@vzLf=wTNRT* zGNbh8W0@7Rvs!!2M)UlC9I3R_zIm@7#(Ma3aVGn8&HA?xw)OY#nZe1eW`%f)FJTv` zv1Zc5?siWTMtTX+Famc#jh7_P@g_KlYk)6xzi9iOdq(MEF(ej^uDhJcj=IYwb|Q@R zG-W!+U(NDj!_CICmr86OZ{#!kU1Sl+m=Xl*4a?ui9YO9Tc^;E2u9z=edp);{+PqL- zF!@kzNdCxc7E?cJjD3t*f=B^gm*Q?E*DQ9t+0 zzV0r8QR2p@Y^qd^&ldE!jv2r=Lls5MX!~#Fw+j>R>CP}0S@pNky@({T$kSNE-~41A zU|3l%vgla`mYFS`A&L{q;_Bps`Y1mQUYct2{s`cpF_1=nS=7EL&y78uebLW*U5Tj5@8n`t4I< z`AVksUy`l=l5Ay>{!<*ziCXJhWB(9`^DndfyOOQ{1I$uTs9!a;bw@qOiH|6v_+j3bVgNt~5jN)TKZByck28i*Ver78+ zWY59%h(#ts=8oNoyx1xF{vYfQUD?m`!a;0^<-i0B3pPR_bLh~OKL+i65A%{I(*OAN zU!sct5>@=~AgcI3V#oF)T~;jV)YReCMyDJu#Ou=^SRP~R+Z=1BwoWp7FH0?S?FnH& zu!tdnrW zo>4{-fp_j^NF;R%+lQUpIuje=f*s9soG&4)$*sw>m~Z>W9VXEsS_F{yQs0z5f@|dX zgN*i7c64xTlF9`vQ;zYyxkvDh;%~V#6!>025YTT-JPl8FK3L-w0@|HH` zkBrH=(RH@ZWnL$BuP2lX!3=u{PyM$ILb44!ewC?tpsPY1bTCY6L@)q@xV=>Rq@QwVG%(q&z;6%7o^R`QbK!t>Z@gR6M{O77(SdWZm zl-*r#inp!g-UU||lpBzJU20+=-=}{}KWkLJ*bhf&Qx!wR7!11kyk0*)mzit~n7I7* zHKc7i>>!)s0jt6ISTTjP?7U#HA98$r--SA&HQ=n!kCR9;AT+ia7+CDrb5D4P0!Bt%DK?|;yO~A-3OLOBv4CL^deRe z;Hjx!q$F5m?6)OlKJ72InzbyRT}kJ5JMNrcNa!Zk?fRzW<)N%DWF=Tl=`(LQGTXV} zR-)9LOQ@g~ZJ}bCY94i-XczCrFm!IN<&KQo?0?NyohjvQsldbASGjA1C)6V8m(QA* ziz{<$m9g6ps^5xoEWDveT8t}q1~DL2!u=L*-^~r2N9mGN4Z3w7`zYiJGO&I{UQ1`` zPKcq=;TzB!atLxnlUso;v_Ck;zb6E>%PlDV0ylimyQMRwMS7J?==QeDd(y9>HMmEp zFx!YyEn9J_(R=mWTAP33xA#b|H|Y&ndk!TjPykmWOxU$&U|PnG$@o3T^mVur9v}ks zON6G|+uP3Ug}7-q7#41ts5#V_~%adlGIJNyM7Yoor4>!EWj zQZrwF0*f7_K0nBZaj?F_=3S(#fnWB{U60ipbv8U~Cxgzg9JuqTu-Mqb(*C!>c}8!m zCZRKaurJ78l-XzT=ue2>Ph~&9{nUJ=*T?5q>}J^lJW>-04jz zN@CK~lRvPio}SY^#~kIt)LB?2UVXl4-oeqv4PxR(TYaKXL8mL@%??~=s{s3zR?bOR z**jl7CFbE*s?qzctX5uUvX7TpO62Z6d@z$0^j<4>cKXtlGv|h`d^+p9FPpsK_hH%H zZMn-}@X804w^3^kMZ6zY;W_>B)?&X%ecbwW4}5)v<<6(met%O&RizCS5o75uc3Bqq+LDQ z9?bl)!u$f@AU%KiHsNw)sK6zxIQT{A_a4$ed9r+WMdtT2zxNkn?q6To3viQdVAL?B zIF(D@W(WmWbR8$Jeqf3G=mnKUbmrntn}lKaB_B)YT7EF?Rpj(o-1)wUkQ^u}x_2fRZ#dt!kvWp3O@1 zWlL0#7!uFcU{vqxOVO2$@8qw znd+a46SkgxK(-^1+4lx7bqkZ+kiFdbIty5C;*g7zRK>FHvo5yBF-^);{0x=mcd=GO z#W_=CaJ@_4VjFvZrQK-&Kp?Bv1hCxo8rZ7R3E@szO4ADhh%@uAoK@MFkXy`q_H@h)#J>8aYjxt1 z0=%wW!QDqs8PunilVipKHI;LLvWa)FLX9)r{IGQ~OxhK8V3vp%55dR2T<;if0>4NH zdlgD>*2>xxl?~3r4K(#f2L@tNtfa@-A>Kf9rYu;T=1g>UxEnJ# z-rQ)lP2iQ-|Ewf!q5nMty&Sil>X&8f^8pHYtjlGKFgF z)`sO-ufeS38 zTn_Uvtw(@Btb3KwVN?x?=ofIXHc(~PwUoq+ulxDlV!lHP`4D$zh%yp8>kynmSwpbX z4L7|@#%g(S4}J8#tS%qlkDm3&Cffq#X04`i$w6=vXUnWH{$#>RpG3ve1h1ZP1&gR5 zY$je~;ac9(H_6SHdA&Ona z-AVMIcOVzYsPBcQ_RN7g+=3(`gl<+7BC)GfT+xZFwkY;)&PdW7pi}h>&dMia--dxGw#4Lryr`jmQnP zQ626z5~?grAH`cIu-!wPIW`(K;x-gVp3g&2+w~hg6YS81(dUC%VSwtIKqehgALZEm z!jkGwsJ?|Pqwl{4a#r=F>liO23CD4n5V^nvdmE3Sl*tr*QCgB34dOyDWOBK9WBehI z=G)t{PHNAKSNn#fe4ODa(4jbjxVSj&aboKYbTPzc8uUG7^i5QbU5QhsN(|i6cQ>Tc z?1E`K?6surk2+_qBrU0JMXDZhbWw7Ewp))UI;)L~BuT&5{&Xq`AfwWlg^p@v#pUp? zgd#nEs1_Tfm>)}BGUDsei*bj>Kyc@ebi6E52mnkUn+&-@G+{vfkxQBSL%*eo*XBZc z=2iHBCcV8)974C!2Q(~5u-`C)9iEzmb@Hymu-089C5VcUguB=#aMNxzRJ=5UcOe3n zY)h7PXp36&4mG9}$IvI1a-%1GW6o0&G~%_ed&g$uWE(GsIAA6358Kq=D2b&g9A%hp zcfNm$#oFje`gEJ4R$ksqpa84fq!<3m36m{N;?Ce>S%(Zr!uCV%khEZLBLPvR>mu^EEd2QT`hMNh^{ ztTeoBdjpp=)h~!{_wWnVvwhmLHk4#t4QiEqUUp5Nrqxg?2oP+24)v_7jNgyTFxgbS z3mUP>$COZE%EJ^yECx%0Y9+QRDK4j!jC`58 zLeF(lXlTxrgA3T?{0aicybV*h39qsiVe?5)o*F*6&>(aBhu&p3wq^3%j$zBkX zJNckwJ7RfX{+Dv6Rw?y4bf&^l`(9oRi_h){77{*VH3VNxS>P@*f*Ze^a=8E}$9YeK zWEk2H7R@U6();!%x@gLZT<>IdeYQ&9`@|2Fd+f>6O9)S5EG=CeH*5_|dicbNZI zlThcK+>l_7jrt2=UJmQ0OK-j2c7X51sH5i?EFtN)b2$a#ymmO5{b~`SzN&ot!Ix|K zdJ2eGqnaK}X5Ieds(D{si;wqkYufzE&8=C1FINqXk^jKbtNePzY*$-*-&_0bkQ=v^ zTE2X30h^Bcwa&^-&M;8-umAg>pYNYnooiMAG~mpuq|Fy$3Si>rfcv2_HB=8?C9bD6 zzt-ww9gNPcb#LYrzTadnf1I*jKO9K2?lS$rvb_F*#dz$yn2OL@OqeSC150E7?$xoN zv{h>Td(|+C!kv=>UzhY}USXYih11sSC*SPX{!)b8*bQFwYIM`-i)M@RLr!ZPmkVl=Zm%d@|6*KuyCt=G`G(MpL~6_zc5y7 zH=!#d>th-4lf9s;2bEv8Rz9iRep;z$C=F>jTeia~`Cfgu;*<6HdP>Tchf94w?QijC zM)$_3HI4TF*D6o#pK4$Gizb_pi|*5(c;OIKk0RaYDr_YW%S!9Oth zf1-c=yZt+A!@4xkFEju{8bX**+QEmfY*1bS&7*4K+T2Ns%MylP4`B!0Dm$!nMDqWv z(vJL}tc?qVQ$}X{<-=7GhOaiJ^Pth>>{$nv3oc@fYMQl^E- z_Vq62$nvKh=Klwe^qbc9f4C|?`|O;D+U2b+%i9rd9Qk0Gyp$w)hhl?T4iKoiN?%p+ z=5a+==mJ&`EtMQ5H{#i>D$ zZ4>3of)3$0_#a!0zbG6!+?t|4KdJW7@^b&4lM3?tk54K8uIYW#DE_@O%=qQ=MCh1< zQ|9U0brk*DQ39MP52WAb>!!LNR9y^{3LwGMa9gZlUuJxBZ>$`WP1aT_{r+m6m;YC0 z`{w`sMDiQ6{YScn$`7_dB^($aAI&F?GckT!td zor4F9RnjFgVql}|?B;bl?uZ@?)%!WSXKif4)>s3H)PVH{3fR~)7tI5dV>W}r=9P^( z8EAT(%MP1l*v@l`b#?Ctmg`^tqSJH7A56Thgf*PXl;|dN#|EPdA~Z%5vJWbI00RIH z{tr%-=et#S4%64xg%By=$T-I8_KHled8vonz!Er7I$78ChEDr zesk9ae!F+w#8!1CLC~Qon_@KKK{Ticr*aTXk5AXX%GT6~>Ft#Ql?i%6-2;{vjQGKK zZBc?zLz6rj;+2c2RL z)m4M;dIZONRqvbLS!ag(x_hY2eN=FxiuqbFT%7eqzu+`)`V{)0x`9z;d-hqYb3UcR z78^?~ZXsedQ61bQO$}@Dvg;ebiet4F?cvx>jW!;7@lFazwy#zOQxipP;@~)%lUFlb zv73XWBhYTD!Yckbv(f^BqUI$g911>6gnGXTEi{kPtDKm+B%7(*vg+f%nGzE9{4(jg z1@9u-N{g93J=bY7*q|X;D%RD6GMs0gS*$mqV`kWSPx zUlJC`iN*Bn$W($UYAH|v9wGMlF{xpkD}oh#&SDmG{InBd6)tyhgPaX5zjhT++SE1(@DlM-Su(UirKQGoBQ{a;T&+-!qK=Zg|Ib$vD3!J1W-A~2k;SvDS z+?4ZpN0-A}6Ru~$9dGG@eV0ZV2w->j2$|dO<>H!7+OiQR>&bc_Mkc;jMZTc0s?-LH zN%&vwy=PRD`PVM&j5=faqk|9_L39SB1p%cBh&8kT!4Q%_2%ypfp@$Mc$A%IF1|%S% zr~yI>2~q-t5LA>dJw*b+LJv)lq9A%Q&-0wK*1X@|wa&-$DY?JhxpUq7xA(sGzAk}F z>gf2pHkIZx>VaMMwZ&QNSrV7+%o$=VvK`^jLEe%3Ykj|ba?<|o56gOQDMMKE*mnM7 zc9FouCmfA|ooZ=%cE9Pg;lX22`g5roGj#S3u1rPATQ+II^tVfFZ}_spbG(CIk8qP3 zP=jV^QY{*xSj+jWrN{5D?Hn>ZyL*yX$`-TvC;1+K_Qz6%bzgBL6L9>OicmX^EaVX) zT;b>ms!}VdnKTz%%7H$=UQAc&R^_55x@3e&Qho3Uz-wyS>RMUZHQt<%mkyG#S$pn1 zwh?a|%xrm)gH@)Xu0|He9>a3N>FD)9O^?f8V_Pe^LS+jRFM?kT%cceCrtKQf*&4;a zSf2kJj1q3KRk8FrIcqigv?B2&?D(GS<5n{M^0~#Q?RSl7Z zpdNuh8M2PqyulRf@_+vN7oq`eg7vIfSD7P+f4#m4iu8R#aawF6G}u`kiKj=n+_{~#@ZLpgk)pXW@sA?i}>cud)q>K;``96bqkvmMJ8_jrGq-fHQm>yJr%A( z`{r`JuP!g{p&N@Js_R&Y8@;yI_XEK0086hmqQUKbv%0W-m(F)-R{#_`D{+Jrso?-o zww<0Y-?pu2H!;5hcjwjqzLu10tCg0UqM8YjOjbD6l(bCW3yVOA2uu!2zh*QH$n7EP$07n9J#5so#4uNPg+S zfwS9iqshJ#>QO5bdm%5_XPX5X6&#Er1(&ZeT1L$2Jk+VR0f1N_$ha?kgVS-%#*MZk zzW%+m+5SU&ti02bx8&^ZX3Ym~`6;eaK#J#=f!N zmko_^XS3OT%-OZHqI2Yvu1sV3W80*vXrfy?;3MyfZCy?6zH~#=-5IUnt1eKqNU?Tq z4wc{bxq%YD;`G(NC&?<2JyOc+jIBacIC=Qj+wBX)ou+H=a{6Fdh<1ezb;9JFaz34n zj~=j}89=|GDB}_+u$xE42xi@kL=P6l%Hrugsv3N`F=S7BDQ~5~#Q%M;&u??^F*?{* zlK*R>o1OYA_<^VgS*JR0?1n-w=Q6V7(`W&KLu%616K_GJE!{>Q+@=*vcut-g6!u%_~1GkTb(kw~7p$8$3?PcN8qU zIw+yv<=j>A~>LY2^=}|fynX)1?iul|Humr4)s8fE_2m^^`BG3gu5?C7R!2R%pW1myE0^_bh_6x!5xBD)je{{z`8kCR??-NK zEz5J~O>?`ynPDG6F{*1$nW1JKbfABfe;@oB@GBf1fy_;8d&^!mmfBg{_n_lhOy7P)H`2NC_y1KeS!lz4XsX6=@YMU7lAMbGo z^vb=BoV4fuUvT_RZw%fmw(+3PkZW0&ZTKJprghlJSBH@LilF`hcH|k=Gl+V5z44CN zFa}`#y}$izcX`Pus@4Ujm!+02nCs#sPw4X^X-u&r#BGe_^1WB-LsmJ9_+tmYtsGPN zvh{ohHWA%Nq0TbRbPA9(gYX$y6pO2eo$-r`jcyM=`l@|&i%}_VeicoB-s5s&mCf6U z_}$e6B(d?{+ei3(Ch_=?xymM~rv%d$uoj)3IH;_>4~%nhO^1}ZwabPCL%0u#3Qq67 z#UYbk|1AAU`H%mN(Es}UMXhH?MO|4fouRoXr(LE`i{<0&AEn1v@AB4Rfu)hxZ!TKG4#u0zqfq}s}7l3ai_jH2S$%#n-)}2Qhcq0o+`Y*jrFm@&U&~#@E7up zpEsiLe*?E4VS#lEHwx^eTfSIJ!Y=fw({`5HXCl!Cr$51Mq?(E6g^2XSzRA7Q2Iyu= zaS>y?bV(Wln{;I@W~a)|?-HEzv2`4i;a)faF$h5moW2EWH+$yw=wEFf4KgtI3#=EJohBP9dhfq$qnaoX-*)Wx zI9#qp$zXjz)y@rF%vPr;?7chL3QJXaJAqTF55eCKsv7FtP3w@N?5xcZ@K>bgQ5%~1&z+JKCFet z#SW?TbZHsw=Byh0M4^{LRnE9DjsL{)#lO(FMJ)|9 z&rmlDjMJpvZd2{3*z@PkuC8J>i|(qjDyT4!O+e1nq{3J#wGgtSCuJxV77%sD(Fb-CIGMGNj)dI5D%ci1}5dDGei(gOyJ;2 z{JE$V5Rf&khihqGqM1&5Sq@#oQ{xhCCt5h%#VBrPtHN2$omaYhmgaM^JL)aVYRhWM z!8UR0i6e!}8ADgu*eA%~t@HzrhLl!07)L_D?}dblRN0x%39Nkm2%UXr(5zHsAUynd zK-o2pPFaabEOrV!`<0WdX*{RF$Szjq+X_N75$Zf@(||JR4@H5%Z0TxU=k?+RUrzf@ zC65{3n??^nUE}=nPw*4g`It>Cwv@K?H|Uu)+_xBjUX3y*1=bLN@j0Kd;qIO>8CmpV zW`~r1Um~Yz87V6Zff10`y>RDq(ZvPKp~Tavc{&m~=g?lddr?n%l)9;M`<=e#Uv9f& zcx3oO*mJuXF~j$;Gq|$~xzE5Ftj9wVJ6~)cdfd4Gh+ggUgIgw>>2ht8)R@J0YjSPx zN*B%XbZNlLODd)wst@HSrE5-M6lwW$ewcwOM0_+a8^{ZM^7IyeL&MAZ_(=EB#AMjR zvuMBl*E>dTV->Fx8nuz7Czgn**5J-ec&3s|@W*N|oh2pNjCivfDVMI*sW8L@SXt)1 zPCWlFOe<&K{9O#Gt_w=Hoa4PRvhW&s326l-4f^Wv=OJwG%D-4xoK>`dKO0+8+(((A zgl(TOw##GSiBW{-2hQBf3w3gwlB@<4GS*^F*uI$&JtU$h(O;qW@6Nvoqnyg9_{qUI z$a3zY_pFXt-UUY(mT`R|4y;AiK(Zqrj5ANBWt?%uPilJE5GbA|&l*|aILoZ1aTT7Y z6p-ZcQfY)pC(TvQ&Ew?kPAJYn57F_Wc4-#uOL=nwx0t0c8tWX8`?PDiMI?J&w&#vA z!(TjdiQn__rQ~O!zevuY(1>1DeqgM`p{560yS3gw;wv+Q8xJqZw?0HGD}&svfCLR& z$)vR(V!w`@Z-8AYU%gAAV=V8(*VremnRrU%0GUoPDnuSZW_^6y%&Zb?-##x0rZe?p zAo~8`B~8a=ViX6eoFK~_)$D)eHY$BitYPJ@9-W3?dWSSg3A?++(?G`UdHyF>!AMyos=8Y%P8i3+bf zSGkDFsGFp$neNwN@21_T3r#vKvk{}G@~KUmM8HXxaAg%eQx?m z_JxpmHD(R}FlaL`sM~2X5qGU|A1Ka3LBDIO<8@!Uo7}pD9Z)*tIxDp{HZ{S;p~}zm z&giUHGHpLkwzgGWS-jadBKBR!Zf2f~xh8^1iw@%m0xvJS3vRRi;2`2~>}PY2oC!w8 zuVk**0-p4%msgGb;c(Eq7)Q(Im~af)0D5EbruFcUXU$C197@)(JfkHa zDJsmqGht)q&s-o)yWq9Yx4%3dHX{ul-j6;UQ$?2lWtMD33)s%8+MIlDa*MFAORE=c#Hh4ur+cNkWVeqqx@D2X{T#EbCn=5bDtIbT& z(-slwpeZ`f&Kl$VOiD`ka5O9O`z@Hml6#sGOp|$xN9o0l5*V*;KkNXLs)DA8M3J_uzv0`7`omUT5hka8>X7Yw%D9mVT{AkmFees z23E7-w3$LC9<&znkYu2|53${=oYuN@!A<172nP(6C1sTo4|bJ6nI8(Z|3lA>9+1gG zG(n+I=cD%qb9;{Y;mCtEOv^W4i2>I?*B3U$leG`*KnO_V=(%(U&!FKZlbl@Xud6s^ zRNdr}Zz~vJgv>SpfakA>Li9^Rr)D0H-POdCr=e zrphv#oc`89p5H@qx@z;})MUHq`$DiwtvWSVP-$dWFIY{(^6{ zpyiW(QHTTdXsHjQJOs~}!{Z*7f(i8L z7$*;alrCh$o2pa$9w{(Y54;_M4szY#ZyyJ(d40-x1IkJUkN*xCRSHSa0#{yIaBZxF z*W1r|ot=}L@>v8NVJI@%Zw2N?G@$7Q`=L#L{m_fK0vw1x6M3*T_NCqYVq)$SFsj*uz z<=_p;JmPLtxLNvuIs;Y}A-VxLm;7gK(It$aDrmcO(yA zjR~nYyb4Cema$l#=ZD^qvWthWHoK8sUmo8x>-9?{k5{<7&}OyYwYVopWNGvgN)6Ir zzWEK#KHf)rhy%p08=`Q3%dx7e7KQ>NtC%6ERofESp48^19gnb!AEIjK+oq#xMb0+J zpI8_tDz{0uD>dto0t0R+TgS;G-@(zU$v)*XrjQ~=6iwhER=68&s)2hiBXPclqjc;m)-j1#e*#K`bYd0^)@9NwDP?tm1MyQ+i&)=5mE)9(|`lp~vVauzunJT;oT}&raWs~}nev`ZcEA*B@(KS;+A|n+|n0()C zZg}vJ-~MPX3|>y7OuAIBHf^&%8H8OMi)w%7EYD<>otGy6C!?f74)m$=({+2!7M=?2OQ7veAVHunNZCv zvt!Vu^YvD~^BdA5tE-u@(2lgbh-t#r1o9V7?(hFfIsSiM98)l@Onrk_XLnGWA=NjB z)q2$mu=}>2vK5)7FiS zX<{(3%B}uP)T$FurmJQRrtx<-=eHlq{Vzu1os5yN8Jw7Wu=V;{A9+EE7aT3fVO6Z! zPV{sRtW2GWSYx&yL={~D$FEH1%6y`Dwej8DlNwfR;5KfN6F1be-dzAwJiwZf4U15VWvuWV!}S>lwYSoY*NW zo4#0k3f5|OvGz@dmWD3@;ABdBc`~*0w+a~N$Bv6+UkdNLY;IN@UTe^`Dop2DE(B#y zNbjZ&TqBn!bLmvZE8OW%KWoy>B5U^*Q4k^f)xHny%2-njfgb~Z(U5>zG)UBY3 z!DPg_27A5WqE`T4R*-mv_}Q~hzH$=s7RJKl0lz;Q&!~AaAaC&}Ly&8!?^sTsj|1fE z0*5({U&2)G&#H*3eFfJ@>Dg0YF zn`nCU@iXXfkR`mVB(o%~EcSC%A7SmY+4+CNECgr6vKlLF5Fhsv3SkZh)~zud`u*mW zW!>-7$u58t6-hBhyYAN91-I`vA6ns#W0n!*%H8N?2tVpdQ(o2q5!Z51+~cYL@up<8 zZLymE>uMBP`*@#BW&C`WgvD1PDT`!vU};%8w>KAA0^8pHNE`_1zwz zFy__`@6ci8&$&&E;ih}lOQSz_TuY|YYU>N$u!~eSDw7}?>6y|l-`ogUbMy*e;~exp8EJcjuUY0llL=otIOium6wKvs0z2y04G=Isbn|Sf_EGF zv(I05L*LoCNtRM8WAAzF+6V?5v$;5{%CUdPld*HaXxp`D?=xkjC4ZE!$CyUr*s(}( z%Eh;y{cd6AH~{?g)%Zzg<+nd}JbivFX?yF(j@tc*!(<}hYOaO6GJ)MuE@j;5R384G zS~EiclJtz{)Bqrwo{$oj6<3w9|ISP8cG?R-v2ZRxpG|5-qB3gIDgp&e*OFnC+rl$h^c_A)s@109a&`_A2Pxx9Meb}HA z?xEgoXoGAVt-KcY>C&t9b2C2e-d}EP$Nkt5q+m@Jvn(Hi(E14AmtZ>PZB{i7zqq)h zwc_6tPmvj8P2C<*DyX%=R7*$XG>=#yY;)4RfgCv8U6$RriU(2G?>V{ahQ^8@wK-I3 zYyQ>}niXH^V;s7O-Z9yoULX9Q5#nw&TXMY>tTFQ>qEv1s|LEK{xIX$wmc-6JSjW&r z9Gf9`$pe5LsPO(;tX1N^;V;`g);jG=6}ZhXY;1YAQ9R7xtjocRuN`1y$*YHHk$nfe zt1pk8_xm?MpJ+0iA7R%Uo26-TX;9NH@dNz*Q96PCoIc^^qmbqJO@K+a4&c|7dm6gQ zBVZO$={I(Va8a~Y#`t;boVD6DMH+n9r|zFAE8uoA8&-ejc`68dYxck(ZY}L%@T@0t z_lY^vWRN3I)k1xFf-0QO49VKNfBmP@Z0EOOd0&F_zTadAb2EBv>-~ixxxKM@kwXV- z-QXT!?#XIBwbIEzYoYxr1U--lkjKBM#!+gX&D_`TjL&v|%E;IK7_E*pedL0koVcI5 zw60#zp%)~b$*$Y?$e$7&mmOdYAdHQ!oKxM-9lpVK4CM~kcsKyHWc4=F0%&OBwxFjQ z`%`JI-EnihRmCoueaeZhrkalY^CqZ$6ioP;=$2-1jU4i*DbO*hF;? zF$TdwLLC`L^jfxsd`7tCm*juHwlYBU?~y zu**JUvXcQcg5PKIMJ}azKv)!vNg0mEp)M%S8sqoIsWsiFwMI0Zru|Wt`s_tVKCiK@ z#FNi2lnaPY-ZKKLn6OAKOR_<4- zfypz0>l;C6mi;KZb7JR8jkmt2APKile+!%~#aL9^uDXqn3{QC5*hmi3pEP>XYIonV z6-gH1r%J0{OvKsp(Iv(v=3A{vN{%LoVP+H#@5Yg?mS=@{}em|0#ospp`aKa(AKtg_xYMiIW zax3(ttlPkECs)ji`h*?q>x8ohZoA!6A_V=8_=V2 zwkCl-#e(T_TlZVY_=CuwA73vXsRmIOBK;&1G}+ZSZ@vQ=&Hkm4>hEkr_LzpSP zKh(V+{WWh7O9IExBfO7Xp7l`4On>7ICfo@>AJlF8fJ`r`)pu=&TlQ%uIW}{<9=&gH z(!Re}&mv;N@9MwYA9kGN810x|F!WODt{cX)$0TFI#JNors{B(roz)FT<{Ggu>p*R2 zQkj=AEpd|lTjPJiW2md6?O_`lNX9BN4dIEnwH2PhIu1uz#Gf_wJ{XMpUd*PJLlpmEXUB-)pm^8 z=v7%Ik6fJ7YrfB_b=urp=0RP2Kv^>cIhA;?{@B6bHxZTD`)N&omFYcSuluJZpFa84 zR&xN6VEPR;+>M+mjfjzZ{d_p=157wx48V7uKB_l4n9?0uJRLkDM!UX03O{!zh(3v! zOh{f~Cjx$Jc9jtr+-XYM6Uo@D`B2~Y;3JrZ{-mXN8rIsxVKQ3u-{mbL6#Oo>U)-`pW+~;DKFzfuWz2Minn4cR@@9&H zyU!2-mfbspP)EC2I=a*)ul2XIG7sOS1ia}OVmxzq-wt56{_eg|&q)Ypsl6Ccam}Wu7E_@k1KX0hz0u;kL)S&xGh*^4#19 zt3)YAwO{2cP1{Sl0zPA8vvR&9^LMiSTas=`KH!=xbv$h_mle^x46ls-p5iucx_smN zKX>;ibnB8wt&ORdb^aPFgJs*PozydE06q2U)|G?3owahN(E_BJshsDN2I#p+kIQ4} zn%=xOpWC-6mzS|!lbHJX-Zix@dy_2Zw8*39eKxgzb55xKK*?jrq+>T>+KWckxyjWcyr&rV}7 zjn3H~`R*dq!I*4YkrqHBTHyKZtm~_@zUf(QLff_~Cdigey90ta+@ARtM7_`dWFH5* zEb_n^6o~czbAV*!$0RVB5a+ROc#>nDxiy&Tsp1JmnQBs0iN$q5swT!tE-Kvk9DZVo zP?(;MH37s7y7FHh`2-gDW$~y}9L7s1G~p>?bp2on2KWwo!BotzwlW}Om&nyM zbzE|gsWY?6v?bQR>EiZ?6c@^Gj`aEh;_2hrmB%3=kLA?|FAvR&{9>BZ7z4RdZ&GDR zy7$6{0h$k*ZhgDK-6vdnB8~N8ogfuBkAzeTC*g!@?be|ll}Z*Ks0EGe#6P16eR}Na zv#{iXPj7lpWHkZkg+%4oO*wJ_S^PRIJTPT8fZ-2vH3yDvHi(!kNDviPkVG%#V$E&E z^VLR)e1LuR3B9HnQe_B9;I?A%cE0bqZd0hCtq|Fb49}hmh)K({vw?C|zRJVyy}F(J zq%N?m*8dWHA*aR}J;gwc=0kw}*PKT1B@mAbso{K{A_2sEvJFcw->XT%Oi$VY=B4c4zxn_HQllh*0$@^T31Y ziKvN>1SIX5oaT_aL-cb8a9U0|4iz?Db6R$gnX0{EqIixKi>pwEzOjG|`_y8{NX$~` z5V+jo$qM6cP1ehx6j2bFke#jgO5<7*kS_BPWjp`(IO+cZEx<0Ot&rIi6mJkaBhmCK zE@_TKdlCW(DCe$nW}c$V2l6hBFUeKu3MFS?>%LTdM^WILk_cjl;)Tk|#5Ngjnhty>pv4 zHOiMvKcwyHZ5Ti#`&d^yRtv}@flGwzx4d4z>oTOj=eW53l@z*u&waoXb6EC2#DQB2 zPIfvUI;DD4;TGA{h|{SPw`)(~PR>uz1;?PZzDCNueuf%+6E`zWvr{+9vfOt4GzMXB zz0>6lLkLwf1-@g_FM5Um2aRA;L$~Jx6VPwyHod@3W}TlG=rS~KQR2SFUr|DM>)h1tC1dq z8hZNoTi)t382=EYJO`>WFCzB19OMRqL*lzQ ziSoZm_@VplS9?*8ch@tNxq7?c-X)!_#w#Q*4a!`dC%d(zlnqCltc__^TMCCNvS9|- zw`U6Fl}R`Q`clKU9RsjR+_sHwY4IAAs z=P@R7PR)8y09E8igEDMF9{%UU9Ndy<;rlv~5ga@htX0am`EcmXKZhA7BrAbjeib2_ zSGISAAFsv56(k)p@Nsd0OReWA*DsqO9Nss_!T9JCVGh(8Dv=o7#_=u8TC7x@mJ(^> x62sUP4Y8_S@V06=N)>SEyy$~Jv5S=rhk&OI&;6bCZ!G=4tN+_i;HMwM{|j6UHBkTn literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-3-Quarterly Percentage Breakdown.jpg b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-3-Quarterly Percentage Breakdown.jpg new file mode 100644 index 0000000000000000000000000000000000000000..b98fe6544b8d0d94d2526e2ca27a3abe48829c49 GIT binary patch literal 117606 zcmdqJ2UJwevM{;gww5+H*O5xd_};lvR)gP*4B>1^ELm z=Kv|-I@+~s*U+vbAJ?y6N56rCaRV9fv9U36K=@!V2pNS+>=wOT=Ex3Li6PZT4b{&L%15Cqx zA45>X%@a;=i;#$iUqI8`!z(@^v8Hx(jF^^BOIz3diDXhvcB}uE)_(>+VzOPcrsa=+zuxr5vsUME z*wO^-;Ro>%*Ul7qlJ^zk?MD~S!gRMj_+A3{dzLlQte(VdFWmiPwkhQ0oc*=xkb;}h zJON#>%=1%Q;ESb+=7;ZJjkxxthyI6`yTLZ3w=I1eRnMkQ@hFW-9`Bsrz)|=gpciep zApXf@8`0n8VcO&8N6au1+#d9wb;18%%ecoUA|5HEO8C*TW<>ZgJ^{tYqf=L6sJoM~$*$$(5+CL!PK1$>oLc{;{*6p3AS|69c#reKT zvq1^_o$%^p{_5u6FKRW!3v?g-N{B2boxNI8gKAtz@!8hQ=>EJs5 zmB4$+Aoh_n{YH6u=mU~~--+KmO)CGzU~t1LmW_HS9{Yq7^arx(CMtU;=3m3-;8VL7 z7by2t3nL@?J=>WT_2@aGi!r}{4Au~bVs-m3#9z>K@d4ejka(VGXl#wjvfQp~prjwx z$GdrFw=u95I{p#+e>JTR^vYEX-j4+nq!a2q*R4o|Q2*{of(Y6FkB-*A6enVKqm;BX zDoyXKe_ZYB>sY*Kq%wdO?{M!WP3)W)Bca z{$mSpBVEAamd{QJrtQvdy{8X&edCa+x^ktN8V&?i%>N_y|7QByT@CuX>SQKb4O(LS z@5hIHA=&WsgUaI^i_MEWzzv0e;0*=egt`I#p7)HoiLq9t;CRxgFVI?$ep1VHa1(H$LGozK0eQts{c6@CefSasm1 z#X1lOe{>*+eKcY^!gXiW=&u+yQF@HUBld*KQ1H(cgJjWlUpgc1}~K zD1W_mg7@p~6MO~HyhanB>r+7p-x`hYa%WnLBEMr%`s2_AuSH^0G>z?ExdZAwwV3^H=0ERi4DWn-?-zW{C62}Z4s9iAhZA?Lc~?yK?)8p-R|oJC z0E~Vh9M&MMV~TA*mcjSvUHR*;v4?t#JYx5c(k_9P!2DNeYZ_XHES#uCp%T=XV6vV~ zdl2jhfiP+cwEX0hOnhIM3&q+*zRbxNHNWU%WXhB~&jSGbgR8L0*f|3i((|PGM1$hd zGY9q+ad>8e19O#w-I&A;j-cP1qjH`l6t@_azE#qu#=>!&q|?yg)$mFLB`GgeFZ|Qd z^V+)e|Eb~o_eu)cUy^qXUZt*APh_zPiUcJ;sixwC51mrp&Cx>2Fc@x;504k^g@X>1(G7d$`Rz9UQIb1rw36%i^=SC|B_o zQ=Pz070|p-7OTr59)1Zx{BQgL3X#10^9hNRTq)<~ee5cf(j$w4li1^u4zJ&_@Bcz& ziezCwm^PCULm#LADxLHUwDV9n#{B?BF=Wmy9MFOKQ&fM6V#rJ%L_ezGz$bH8VB#@!5Pqnz&)0F>1F>3vRwwDzZXWzTRfwUtK_ZXG z&EnhAN^p8Xx{TF*76)-{6_Jy~5QPCSj(-ItMo<(it6RoEp z9$f_>6tHE)<{Lz-zo=_Jmn5wP9UWeuZ>JZ4y{N z&%W;DC6Ek-vPGuq%oepoTA68edy)Bbeqm(D=?f^cyC^b>+tpCcmw#&>Io~7Vk6;e* zAQDH&negSqwvCFPMK+|3Z`@kleu&X8e$yOl=u|pU`J~!Hn9z8kL3f5|cDGdDK)7eO zSKky@WwTFCW>X7SB~}HJbN!zF6#3HWV=_Jxf`lXCt#H*4R8^C`@RP_@?S`1pf~HEk z6vZZvopl~Iez@y8%qTi|X`?%-(OMid2}z!5sw)|g*>oSP&bP_8H!|xi2>NP{m)@DK z8S!jRlnRn4jF#W&XuH8G27_^)>EKg>^LP{H74#Fz#+fX~N@x!s*{6Hcz4Ua2=(Hyu z=BrAVuRzQ5CK#5-?`Us%Ikps&+sd565OAl-c29c!=_#;+xgmvN;r2p*Q+1x1vV6)& zL{}DLYQnClMw#}9=iI!@oZeB}f z!U*+sh)eq3r_2q~9}m8}RaoD8r$YQ!Pw5ugZ5!KNvVJ~BKDM%M5StQyiIw_)D|?g! zStadGSG(J$U*pj6*WoYGKOPyBKe$`)(R(Y3@BKCB{kL|voVVZFJ*1lVtFn@GkWC`E zzxzQ$LrZd=3P-#+@w%QOjI-`**4?WP`qwc&yIH-5Df9bo#3RUgS;m3tF-p;QZT!fm zM>hdXdB8+kc@bq5roDxDi5shxlM)~N)kc8cC%g}~183P8v1~r*GD|Q0@K)+1x--<3 z1d5{&1|d*RimM#ADW;y5e8%&-=>wd7C`rHiMW?6FC@CI+mhd~jT0_Ks>6@@=E_b08 z95vO$r0jrPGHcyo3t0$D)zj0VjqH2a`=k>1_r9-|2g8R|hb5AOCg8m>Ad(&(M5+mD zt|*gwlH+_Zf)r}&4Kf#{m@<+LQ6cNoJOo}O{ZKMD-C4I112ZPr_kYwsn!i+Nc0g;?%D0uYf0Xf5e5W4O-T zL{S-NzQkM2o7WLpH224Ss7Dd?K?+9snYZ4yer75bF zde@o>U1=#ifrp1bee2t|shMiyd4z)(Ev%rbNQwN84(50AL`-2)yW<=`ngKS?$%)2E z?fnyfMr@!vA_kr(8pWtzMt8dj-VZIft3@#E^@TiCzxwb!Z5Q5?+dclPb=t9*QQy5M zKVVI-LTVqeB|izy)(+7wJ@SeZu`M|4dl`;!Zw%}7aDBtz?QnAuYX4@6O(9KHA4CyR zOxvh=0AnsuwiaP~OYDV5h+3ZfwPVXVxLU(7wgJ+7`f0K8V*s;e=9`b#61qM+ZlL)RRz76-r6fkk(`G*|$8o6pp`O*hjtIyh%xR;bL~FrQK8 zl@3u(^(zpDW)tx)WNpT3E46g)OWl7V*WJWb(pVOsKfH{2T&NH_SYy!xVOb#ef*YRc zIK!Ov#VORemGu)4n>BgEQpy`jLsIfC(n~6=?&i2@fOimb>~0?YWybN%iaS}i+8%4U zt61~zIg!Z&Ya_n}0G0SGBZq#kPHpUic?0KH{(A~@cDY6EUG{SsX^SYn)j!c>l>t-< z;MWtGkrueli(I{)nqLBG?+`g-Owvy6g7)-+_6*oT$l{;j#hv&{iw|`j=h%B(p)N${ z#Wyz1?1MQ96k*eVhy+q^g=h7gIICKJ9``ymZtdZ(UFylOG@e-0)?{?$S?oMo3D&v< zh7MgeXud0Fi{#O0rfy|;?$)j1thq=(0Z|PWw=uhmzUUsqae}nT=H&<_pf3NT4dnHL^`9@8+k5yDy^D|A&3Xl=Lv*$&#~WV&uDd@Q z`Qw3tT^o(4A$D3m5S(@@`wM<3J3mo;4=^kTfbm*`3TuuQa=gWv*(_MT@_L~VRtwcy z{UzXFJfq^leCy71)D6{f>ro=>{^>Z7?v$VIX6pX-d4B>!fmI_@Qbs*G(d{1B1z+hIUQa9amPNM0Z!aKM7h=*0AE*&?^M9 zv>Teo^J+BCw$BlKmpeLLjiJ@3`AF9lXCeBTi*ocfw~h|UPRX&^rnP%jG~IY6KJJLd z=#Ix`OdWl5r$Zg>MG+zk(~BRfKR%6~tJO1c5&B)hP{*+~gZ-2$6V{}8N+7ehD9soi z=gO9A@-qLHVJJhHsIRGkXHV$!rSUv36U|dTP*P)g#u5gJW_%_{lV8zNdEN>3@qoBZ zm9ls;r6ESf360qclXOeprvy`F)SA&aCXB1rhm@DEuC(oBCk{1|R zVw&DywY4pyq0# zfT<~&l=V>+c50VjssTbc4)8=Ol0Rp7qsPpx^5@U-H`UBdUVDmLp`aMOywW5%jO>>H z?TsBX^GvR^#_!?O6_11aMs+i0A@sor1xHpSsUQF#D`ptLBk8W!(_)EVUmE6*@ok_e z%SmxJ;meCLu6p1pqm!9CdI=nIo@AX#y0IUUC^JObofPBr_};TSAmuo;l9PE_u5Dx{ z^Zmw3)Is(QYyomeBFE=xWT{ zv^u*SPsN?AR^|Cm-}2;XI>g;c?-lg48>PDf2QA_K0`U$y&R#rNt1CWd`V{O~o_%8n}?dhbr7ryi(_ zv0x)40EXMZy1b5GLGC56u&sS<3haHl_npyDi$RG0bF~Rr*8ZKolU|iF%w9J?H`Hi< z&AEj^IZ7@;p;4jNQ$Fs?C{?3$eIg`Kt7Dfl8LN&yGpn~6>PC@jRZg+qKxR~P8sV@x zUbu!3h$K>TZ3CB~m#8MDSs@5@ERX^~aI9rC*fZp_bH95xx_WI}OE)&7;jObfKS7kj za@R>C!BVRUHkPG)8`qlm$cbzPeRukC(6yK6x{7r4buV(I567M%8VT67iT(-MU?LAt_{GvdluYn0n5A|?ZFbi3_TB`7J2#tuVXL124E`~Y;`itFvfZrID`?z{_?kl@?Wx0QLq9qg_%n(SnxlA@Fg=)DYaCM zCv?eQt11R?+$dKI-+H6g+i?_ez!&7T8SDLNJRvvg$qvD?h`ZK-tEh#)%5D6EoV^bv z4C2ISlT1CHt0l{ezC9-y64WcAbFcm*5%E61#(=D(@5cE0xu7dz_CR3tSLbtK=98CH zHDYvE#KwS2fb`?S>F=uKep8DmM7iKyXgT~{)$`xf2;|Ye+qOKs;+OI_HO9CbC-45^ z`>|x&6T+G5l`g*uKmJXaU7r??_xL3+{_vRAu9L9g1xoiXlMFd)!XWfv5Huy2_ke9* z%j~0Am8(N|6Y5?7*t!j%m{`xX5IMzE=ptfsAcn?!t zcil|9O7HgJ0hW#M4@RbzxgSY!UsT^DKXiFJ5D z-(Z!?5xLj?`%ceIVI`9RbhhL!PbQUKod}-8;iRC-anL;LCBWjKW$;FqAj6<>g+eik zs6wb|Kb_#BGb(>OlW6uOc%G>!jVMp+HEZ~RjHG4&4&fybob+_6fMzD3;!8z=bA`<% z0NV+ik7m$XT_#{5HRCfE-Jma__hW@W@ z6bknFuo~EVmTAkSdhTy5YQDW;Wmlk0A8978LsdHBN`};wD_#tCdCEsUmR|_A>SEEs z#ByvN(zaz-d+_xjpALM1y>-uXUMXzbgA!e`hK`revrs@*#{Gb?p2d+)GJU|y-2=2N zbl{tf&S=%zLY)!Sfmn2^ubdEYR7kid&P!j}y+xFIZyIT0RMYmK;dz`2%XbGivmrH z!X0B|@y~D_MLNdgYFonV)40baE_tR*<3*;U+krBMq;Y4P5H&sWb4$dF33HhkRD$>G z1+=Vs!W1zoGRqkQhDO+H;M^EiZu^^y3Oq%G9}S;nO>sx-mDL3fmq5(I*GrWJj^2o2 zHAIk>?Op;=ZnfqaxPtgv&Dl!9C*VQQJ88c!<&|?@tcnd}1CPN5M+VBK+lh<$JrOsA z2O3mZZJO9x<(-Au`zXCfK2Y5gkjhE%FdyR1AcHk$y~mz$S)#pd??aT%6vO?VzFco@ zP6R&MjDx{lSKq8dATzE>b6OtVfIsLZrW^e>B>~oT2@sh)KZSa^`)O{#M9B#khs-s~ zmSEz#Tgilay(tb~Xrh+|V^R`>1a}p8*7+TS(g>s; zQtjeN*-N@-!Wz68#dj!QTN6-Nj=f;|@$9f@{FDSIX^HuzhFMe%Su;Px_mgt~^3VeD zs;sjH*XDV)RxEeYkYnVm|Mg2?W~?oNrNkmUs&X(ul*N`Z8Ec4{uWV@X(X_T6hr5cW ziGWFi?&fHTNh=*A(bI}4+bk+YQCZarrBtZA%qDzvCg-)`hx!J4QCqhXo*&i*+SL{< zG{G71eroWf`0OUEO{&@Z5Wu4h`aS3X@UJfgZ>$OPax$?JK_I-hJAxL2wkcy#O)GLu`v+hXE-J+X6*#)=TtF^>Xg21 zHZ>fWK`#JfRfRZtwSPbn?+G+4R^8-1scilJFw(H8^5Mlhq*+-3fq1!{!8ZuC|s<=-p6@fW3 zLf2&^Mz*JIwku1J($tLdpjgfHO^9mrxA4c;i4B`~`~qfiTQCZ3mJ}70&GHYk*v@u# z)yE5Yj8JSBm8Tqz$aFIy>ZVLcCyA7BFfVnV$ zhAshwcl;2n|Jq$VIswro5RZ0LRXv=hY;+-wKiS@8X;-G=#1EE)&^HkFBM&`@)lp?c zi8Kw%nyD-d-3=**FkIQHmurQ~GI0esLzGBgdSI5Th9MvVUWfI!YxIWf^)s0WXLeoq zjZ5^*-$>-1*39=c(bm8syahw}7!l%Rp|aLs1B>yR>k2Fwxw`b$GIAU}EsoH_O3FZ8 z6Ol~S;+nm<_>?4vDdD|mn50=PoEEj+H_Z{BKbgg7gico_ghDxD$6}<1rQxnKYN|3d zGMerdf!I?P&3wBtUua*!b@SJoDoq-U?0HDKuTSJ#crUjys6V6A8Xk`B-@2gCp6F;k zbfixqV0AUus<{9|`)jI8XkW`p>)S<)l4nH}&wh+7bwXOCpHEaHYqSK|CCv%p)dVNm zA2`25JugCY#pR)6CFA64L={Y2aLi0*W|kC-S0@^L40T&(!i#y5F?Yw)wfbD=Pq;M9 zXzr|zK>Y+IcgG7XwoFmchZC}>xoK;-#b|Y6jB;8o+LNF5n&Lh;X~ld|J+8RP5+QP- z!{QlcAxtVFE#jNsV^uavL^qo;!Dr&>3L~#X>DjL_rLb0sYrbX!>yAbLI^=8M`RI+5 z1V!5;y7$N#Zgqq=a`eJ! zY3F2f?mF?`0|nXnXtE3q92!737*cFRlZRy{W28rNEFqCoBG~Czv|aA?8nF;Fes7L! z;uX|ZE=(G+^LYhjOM?t0GsmJ=WUJxZW2i0rm;a%>quVU*) zH@l-F#-T#lgoI-4&Y72hCLN2or#4=a5lqq0Rxm8D^3`#At6btE`)!0_FIAPGbk_6S z70*}ZARm7$t7=({;w8I80`+wpCKv%3oyu#YHZ2FOj>4D{)699`A>NNsloAdO9F9FT z+Cv^|&C^{t29AAzWP((Tj8ubA24DOhT@)1*+;bHaZ1%E>+h{M{f9`GtkUE@(kPy$^ z{5D5$dj~9{LPOjw5kUJYy`qr_P>)9w5XAksTSrG4@T2#PUe!eLq**pZEH9ithm4Xt zNuVZoTv5kEDBvN`Q?E|esWPrS0Gh=k0hBQbr1eNZ3ILsP0GAa&5d8BB@!xm8i-30% zj<~z%dJq~`hEB4Spf;7ZPY8_Ph=sb8FK%axfKA!#u+y?QRPN+>N(I9lv-D>G%zcz=7jjKc;MQT5?W?@IQaKaaGYzvbEzE?ZEY3L4_Q$&r>7Wz!*|o8+Far>Q z3;wSAX4g@~J^N<~_SL7igiBqXP!TV*X}(p%vX^PfGcQeBVcNH`Q+KWFbyvzSTeBQ& z(gE)jQCb;&kd0}awElYU&5%ULD4h(Xz{V!)M4@YrpYsgH+`Ozf6oML01=a8sz`2Ep zI?CCLaR~`5ZVT=aPCqSU9w&H7rCSm0#>dFuUFDe^FK40w^4=O*xD(ZQ{}y9c7x?XK zp*M7@Q`@no$tlvx$N6)r*%%FQ9|GOlxuDY8ms21Pfu8CLme=nkbvBzCAe%a>*DlHm zPX|nG$3B7__nI|8pwd0P_S_qO=ja5f?RqSnPej>ofR-R3U{iYy8f6*xrm{}YVWmJ* zp4Sc;>jfFR>oD5<{o?NXgI^vrGSwv5^3xvnhy_0%X_|GhnwtJ_N-m4D*=DclD|*VQ z(EYBxWxeFMW`ZXo)cEJg(ZA!={z$go7`Yohx8S{gDs2}C9|z9=~jNggVQ_l4@74(rc?Rmkx`J` z%xfCVDNUc{XE=f-J8ua4yo)e}+OS$9RknZ`B){pyDA8*;o*YAdFJz*IUC|d?cMI41 z;-gOEuKuB~RSZ=wdag>UV{7@XTK$~4tfDj3p}{u;G8vhX{pvK_reB*p`3Y18B6V-A z>l{FahYICMOmoh-QyXO;_I?<}#wTC$wt)GI`1k61%NkH-gW&z|4t#hkLscV`5(+dI zwe0te&<)WvDj-h#Raoy&z30^sWy2$JDR{WWa8qijq!6JVKf@FUqm0rBXZ+Jyr=zd( zk12=S7Vrz3(@NhT8Kv7W<@71OQi!kAv>azq_vNeetewpmf1=@|a-BzUs-C`>+r*Te znK~;Wau0J6=^ru@P3q@01qMoM%w*uD4^IPh{oWB|T2m{o(=(sHHL|4gI&5^>2vlE1CsvS@|vWOfIOI(stK5 zN^!(FfN$EA=jv=nPStiimywa)H=eYHD?NSiT0TlO7DRAD&9gh6yHZ>cPx>gMfxpsX z37>@4>Uc~@W24xkrPLMK`B4Obz!zrXr7kS2q}c*E?9$rNmHwa1zf!}C|^KZZ&6POo7_?| zJyoXKcnLBJgLu5Pfa4%O`jZzo`>Oku$<=nkc`lP3Q{I=!(SIBKQ zq%X{^f$0%|=JwSH|BD#q>g3lqmciUfIsL~KeX+kb7+U1h58o%YRt4BvQ3pFzl>q~v z>#0G}s{ni-n-h2wgnYpF0b~(d>oxjc(HhC$6%r%@M5ZX_i8$+^OF%$;%cWi=e9ATU zTP=mkF|>wIr#@*DplH70@7Bo84~UVQ@Y&`u;BMG$Anx7~=8U0dwEfof?h{Uw$dh<# zXlYoU7oqNEWOTQEQZq$MGLf%?xy7~;PVb~P6IbxbXyPYK6T}j8)+K{^ z>No2Q_Ig6xOgC!r*F>!AUur9x$wbI!My`h8(c0RSrE0Qp;$s=V>Fhb-L>*@+1;N-t zzuBoc%4_bYu*UCHyd9oxr({%&GVLO1aK6=(^vGbQ>25Q*1$;Dbx0$YPdV{07-=g_@ z3Pqk=+1s*gEHI-3c_m${mexfB8F*#a3o9-81)fOSs!Ds=B}S2L1( z7ql=|7Rj6sd5iZZ=*x!JCSt_f&~iAo;_R@=%$+e#)Re47LOH|SmA5N2>C)Namz&9q zcEoiu5VJ7o-o0s}n7hvpI89$hf)5j^A^yb;GsKb?NhV^1%aS}7aoW)QQRr6R}tV9bD3sgvDa?In(ZgPKaaeBL24#aW$roY1LuoAwgL0! zvrJJBkA=&mo$3KBy{mCW@j|66#pQs`G3SzfvR@bFS2|L(mAnKpkO`!K$tHd8N6M9f z59V)?!b$r(A^M#}Aw{}ZE+WP0r%YKdGBN&=OP&WmMGuXUs#)BL;nzJp#J(T~?iwK4(8BP@PA-2CyfB02D_Q5}pya z0A*$!SQHFS;BhPBD~Cre3difap|hj@cZvnn=+z5UrDDU~7-TE($np!BnjBxXByM=-C1f77U%TlAl5>~2 zXg95m23EV+dUorS6(Tlvc*bDQLpaWHjT|*!L+A`A`9Wj|X(WX#%td;PXj0%BvQb>f z8-H8mt?8fhJJs+xf9ElI=P^a%KO-ffL1*0i%I?W2@vJ${d-gMF{E_x?OzTDV0wFzJ zj3-fb)uCoIB(K`M(Q-9*dboV2dMJ_=FZ|=L+L@k@VdADW8taI?>F&_ zukcT8a6|um-(rmjZ@2V`KcgUj7=iRUga7b46Z*^REV%d8s>Hd&F0Ty%(bf~xDV=v_ z=E%ctLkIiPcx#oX^8ps~%B=m@Uw-dGf*yB<03>HfPU*JYQIMP=In6EolXnR$3IArE z`w|G1_}!wwrMO@0_ngaKv50IEV-1o3eD@e1Zry19dsBXo#IK&qCw10ufXSgi0-*o< zZvQ{0xP9P!3jX|>v6vdPQYn5Ju+*_|YaB8DkBSxj%O2Wkvn*qo#%czyW@%1C{V{fO zYEQxgigM9#ni1p{G?6w2IZZp_;bGgP`;;D&fgpSY@1pq=q2Pa-AJjvdVq59^ZutkP z8^%YuJyJ(FCnOKJ(XG>e5zfm+nG&@z%=OX>s*; zPeRdd{@aOjFqO~k{i-yCs@hKx@rFUyyGT$&;bnB7lqIi+d42PQ4?%~%wMjg6gHE6{ zr$dws&OplehZZ0UI#c$$!&f8ksjJ;!5F7FXs^{LZUfQWL*5Ho=@) zPsBGg&ies0SUH8r!IcS2VH0NvX4>+}#j7o(9-WA%k<^i;i|4J3fNw@dhPB|VnIzd0 zX>van^TjzV@CEl;@tHg)Nn&!q_>nuCdy5S#UQy0uUS_Hh(Z;yf8-I~=)l&$>L< zR*Ds+2UKZ7P)*U+xE@mtg7CHx!`4PPRbIb=4YK2qYSW%Jp-5VY8Fg@N1ClYa&f&(#Qm5)8tFDX&DmNVz|*$Z zN`gLmUZhDN!qy}ID7?&n=h0w@j*gBVBwOoF8gK)-jN%lN?B)A323itXbG>#V=xsTp zn8+8H4Z6-6^jiv$i)dVP61a$4mIgMsteAwMF&(!wELtJ+bV6I&l$_-K)qD}ZXSrZ) zlp?)BQsBf*63SBS!S?QIPT*Z3adqt6+~65H`YruISJMOy2mSU?t?1R16SgiP^BH*Oc?bsOaWQ%G zRjTLoayIOoM>iG_925_&g8`O5*HdATD^I*`NACiyo`>5%J@h6(>27BImO=0n{~wC= z1MKr}b^4eWK&}EveHk{+^*hT(x#O4qH9x#kwWNvvc_Fp>?-lrqvp6@`r&03LMB88& z8>3|>`e6DzpR2h}=!oI{f;Fo`#}-=8Or$8|Kz72R(*l~!%3@5s>Sy%Pl=rjaty16m z_4-=esLCI*Wc{-tLmiTe{$ClKe~(!4yv(V!z65@*0UWz`L8GO&Hs2XFeXgD_bFkN? z&HC7!8op`FK0m=3-?>gBvuS4I{szF=yXr{f{qo=Jz@4;Idd7K1cDU2Q_ZWpijdH>u zFG`I@{;ky;OXKjs$R}GRJ)XO{%P)NG57(__KFH>@3W4EYoA{Xp`H!%&eS4iGy&t5dW%U2K{fUTukz$!n$ z!tQcfw8}{>eH@kl?CqhR|6THoSJ@qv2m9u5DC9@6BU#L`jA z#cw1vIaR*(&Sh1GnOJ0#%4d;r=Nuc%aGKDHhVjB645YK&uYk79CM)UeDMx421q%=r zWj;wWO~9e!lplL^>Z^6Y)kaVGGcT?uJ*?nQCe8R*ZbZly@;?TAQP#J%)OxxO9EI?uHR-+C6|PM6W-Td;cdT?s9+EkiPm!{TtI#Yd zO6OzkgyY5$;;NW?Qr(5sB~@iuPqSj)JP6H~(xYO@ypz44uNpul9tgDF!eS7b5?H zVRMWxCm<*3yE3P|=V8gRxvv!1hAQZH_s|_ZBcia`kPM$HF3X>1IKJvuw@r>`T=_MK}X<0051=GTyY{j zmetV;>v-BJ zT>@h{%^@5`xBJArW$)Em9Aguvyj>pOV-k|(IS9&q^$}A7$rc*_le;xV?!m{*j~mT) z)zb)C$n@^0S)|T+UGy29g=QkZ3Ml%s#6#V41|G}$nvp0=J`p4$2j9BOUsj+t3k`?)-O81|c(xm=J1&lPbzW&m(alh0sT^G`VYg_pHcb|mw zaz6Z}W$}J#@O!tBmfONFl`Zv44aZ{qJBZ26x0x2#a=D07YUK+`wseIGou!;9QiijIsm}e ze0*b>n^rO|(~lz6AsuC4LiH|{1tP0L)qFlzgYM0tSQMxNa;O9|Foe08XpudX;l8W8 zpbQNxQN0A%E`cW}a!<&vNr#c)=&RL#vEVmsn8{OVa*q!`o1tF`ennu?!yhq#^e9qQ zZ6K2wmNkfh(=t{-m@{4#_VSiwC@MO8eM?9fEvCp$l?ttFB>lqA&dp5vU|1pwmB|kj zIbkMjZbAAjnkfVd)()e`80R{rx0{r4-&ES!hkhGU7ALTpb@$Cj`0t7C*!71;t||pu zYvq1^w^DWRdb@SbO6Jw|*vVUUL-MSo38TxBN(;N8Ymk7=6doVL5PmaQVIjS)B;-Ax z^jI{+Psc|L-H&q0x@b)cSA!z1u;`-PwkB}Mq+dmGli)T8x^w@9C?s}Oc8MF%8ulsS?eEN1WhvhXu>q5Q*U*f4wT zSlN8N5fkd8ki;kVdY&24vSnee6RBSN0tTY4a)W5!bmW*@Evj+f#(=vrr5Okj49TPZ zw-CjBeQDJO4G6xk`>B6do0Z!8e8HcCp!H0fE0|TlrSCj#@?+!t0r57sDf9-)RHWlFA_Qu#Ft1`1{uh`PK&!j$ymr{uLt&M8 zTo=a2z48I69!WDdJ9Xzu$0V+aTp?R>3qrC_=l1xeCrW1z1R)DQV95bO1b|I}Y}liu z@1hYkDmK{)seZTB^>$Qb9HrmNF}5C`-K_32z4vUN2lI(xnO1Iz#l3s$Ub!_*dw9*-ooBd6j8B>P)+Y?Ll`U&;vKKhD= zOdFO5rRc`I6mZYQsq>56@xw3*>>@WoOhOvTXhZ z79`qO%5qBP5KX~?A;#cv0Zj?qTLN=ck>H0Kr=7Y)aj}~3zX4Tr~hynNbTs4%Q65i;+nMWw3xjtfm)q!#w z05V2qTyr8YQqH;!rEI!#%_COuSRbm+1Ye_J;Hv)h;Xl$rftu_qgYK%EFwExfkG`Im zKu~wookK~d0S4smzk(WHm-S{YQ6^(8PU!d7EYg;a&6hxghH{wx%c^sR$NrpYJ>#tW zhE`17aXB$;uPZ$=dR17vi|`FCh7Z zpB!zuD<>ywDz%$W@U~zc5gvSfy!Zn#vd~FtP|}P{n!ZBTB~Uq;?*@ae#Nv)viec|% z;|@&J#(S28;C>}U6uAyla@8Zz!2{DiNM8=x(QI+QV!=9S};(z1svt z-2dB_(xl5fBZesreJ|hW$kKWeJr(RDlYO>Eteq?#iN2$NsEdLdqYpd0egN1=;3KWE zpA!)6Q9cx=YInm zlG5X%AAjO?Xm&(m8r6>W2Oy+Ez<*c%qs~8-TNKRx_~ z28@YXOS$J=-!AH39yfrABI>rfQf^JGSF*DKL6${yCHF^o#UQOmv_^Bb6s{5STbos+ zMc9>8RH3(4NVJ9f_H87zf+Rz4(dlG|@O9B}5%_8K>Y@@q+1DDr_p|ZdM)s@y2!R&R zP}`e_0tQZJFyrG%+v@jJWkZMd$bU##?$wXaIAwKvWajWlzbIwN)YNakpN8#Q?miT5L*{4rfkteKMo z_H6vMIqaRDC%s>T0l4miKXF74ec3guKtm-8J;-l)JTmj3ePxXuejEGVv5VP!*dQJPgkpJclJHCuQ^ynVOqE~e z(Mz_}rdXubMe_ZX{WwjGyLD9zzsaSp^B`v9b~fmxXVee#)W6n+ELSe)@b*vck&mBW zYX?VGvA5Ni_=J6ltBShu2aHK+3T4d(P+ucy+74^GhDc$J&1vT9!Br|CgOWOOZJUEE zMusIJe1uV&+u_*>>AuGlu+gGLx!k7_cz3mScz5XZ+8zd1zKrM^U^K7qSGM|CMwxV2 z)Ds^;ybWg^=h>;oi*4{=>Hkoug(AuAm#MbFxely|hti9!^7e-&W0jsvJCnYMpcQQHb z&KLuWDDDrarDR~g?~K>|g;_HzJId$4{1_KfG2sFQ_v`JO_rws!hsr{uGtEv{c2 zU6&%5#IBt*B)VPF{+Qe=<$~T6`@ zeU!vLDuiS^Kd%2Eu80v4@p{?D!jjE#x7?n&G@0OL$@Qs}%nd0v5r zE)M_WmL1NE&|)Lybsku0&*IO`qvWsQu!A3KH=lYN42V{y$oy4RhJ&z+W*y&;_~L(2kDgJ$_Ug3!8fJYQNT ze23){z;c+V8Vb-)C?8lCva*YnQ-GVt>1XnS`XgU8Z?P8Ju=KQ~6Z4%}D>Jp_={tGeGctY7WxzJyV`4bD zojY>0Y^}Fi`FyL9<+Xfiw9ZtfEyMV64aoSr@+h_8Ek2L=$r{*3lVgl@+f>k7z3QO} za4ssu*WIAT(i^;$J<7Qbr>S0?HLgHU+SM_3O0H-vysjeRB{FCHYLqV{B%ngr04klG zWmdVfUQfK;?Sks>ojq(iAsp$Z)R7T%J@Uu42r%vSc^=likRQLgyPo$Bd%6lVa^hE~ zc`rK)ai_h#@BKbXdf}BYunskP!?RZwh9$fIfKtsOVT7My;~Mmg`6}fS0B~LjkY|Oe zv1qxg+hl6SL1ZjM$#z$@+lMhNk5BnvfiSI#KI9&>uuvcA{XD8U8&CC$fd46UoG?p@ zovntB4ZS5_*7&P8`%IN{J~VQfH#we9ZAe^MCwRkkx0N&k3e|r@`Yq=FVedVlqG-N- z(FPP11tkjzNR}aINrL2zFbqhRC>e(Yzzjh^K!SjPWRRSLlB1GDM8Ml1 z1AN~(_pN*H`k(XedGE|xGgURctGc=?>|J~RYL|@J8Y!CxW1ZhO$q$hZZ=3obGng#( z7XAiUjH*_Ic^QA1h!u?N-|u|_6VQ?rpa_xsbzri?ko3dEL}$c}_OkhJ;A-!3oxARha`LV9f1^!xA)Z#X0zuN4;6=d1RvzGxkgf#rsUUoq^*p~Q;udX{~1veZaXtVOSbmjUA5#O##g17i$l}$)R094&4jf6oGP^JED zaLOD(mE3miHvoMyu36yIlea@}>Qw0TrAX`hbb!}EV)?h)pYr*AT$BXk2x`x=ZVO#k zhCSOuK5!BeK8;P~D+%XXd-qpvDsNBi+a9V6MR>tMA=7I!az$lL-ny@dwfYu)#j6bu z!7ETAQ|k#LQ);aXRT?W2dXyMY^LYj1)TP?psXbCp-fWc(YdM1NVuWJXWFA!FYqPO; zjQ1ZXKvU%+zsawEh;x2S4C^DFF@J|+6tfv&8`v*Tk1X3G`8<~I&O6H(ufb$cALotC z&9joqufFVGiT@mngZ&N2 zuN(>)aoI!?22KWH3cAtYIw4^XbG?Qe9+hSq$Z(7)XG@2Y8vN{m)6f$iAfwY)n#}D2A#8_TihPP3rBA$YB5Dv{_8cveTt_7=Ldf7Sb#6x2mmT&6{R;hRhZ*}$ zl(Si}b#mdwBY`xnEOWi2rOOj?zu<+xhFC105EfvE$ks>wXoH&DQS2g9Nrg0DJ`5-S zqGt1wYvQz3IV5Egmd>@Rs@qBPs(M&Rm$JMejn$Jq1cC4&zkxOZRLdW?_o&q{bd^KM zF#53`-+!R_Z!SSt^RwO%-~ch#ZY0e8bB(r+6h4e#0nH)^`-Zy4&UE_F}w z=N-0HOU9xPOEa(rmYIlUSh^6lW^_ylnK*UPP!=o(AM9I_NaTIj$CGv-{|#Jcae0uB zls2xdOa=F)b4D#`R>IUT$n^ir)R{FcV?@#o$>k4OO>*3&e+ygO#zfuPe6p;p% zOf=&4-3-|XsBCvyo+*R|`vHr z*m@KmVq9D%uUX5chARuIL>$!N{nJWDSL8yZSf5RCHw(AtR;soH6x}GGVV!?63}2rl z%i89hw&<8rGldEtT*P0V@>|Y-w-a~E$WK>1?gicj&-loA8~4kd@{nvdxHP1vq|rQ<718~q04(&_2afU8m7 zzy(IyJ4Cz^o+Ge>8}AEP2q3He2gS%es%JV{cZ{m~ObSHNwCYZp2+iqMFSy`#$ck0z zWGei@?eZ(ei{=4W0>{NQ#9miGZrSu9i9D|^XFlR`9 z-(Gl>7IPN?3f9=f^iR z>&|je=;QYZ*HJLkcV);{lxz%6m9eqPtI{d-e;-((qJn$ZX}r;>A_qEL>?w3n(zl39 zWt>++QeS{XU{V;V+{T>w_9os9of!N3tI>DAt@6;;;-q4&s3>xA#X}I(9M_s~Un^)V zo3r7p#+t@$JnP?GC&_cwk9~;U@K#$qTOMvD&7~%(81PwgE5AbxekUa&?Or$?x|Y^0 ziIN#{%SxtL3&i9kw$v3NB1OKkq7^Rw$WiKhe8YDZE69RJOseA9k>vQgC2xNN+LZhh zpO*HL9<%KW$R9YtdY&bGC24f%CzdcSF=&bnIyAogK5$pTN1wDyv2R*Nrlqa#k;HF+ z?{&GnKO(+~ihF=$Z;*ybooN7kZtIT@v!IU$8+PQ8{Qxk1a}FI4h>!l%NN(=Rs_HMB z+TnJuCw)e;06`(D{ji{rgVB!N9G%`3Mg`;#ORniO2YD+;_z#u`fMpE0@%XwVrz-7% zl$H{n?#48ae>GNR|5hT7(#@e6IX6L7;kMKgjcEssHmI^<_APZ?UphMWIc!x;?dt-x zQdvm>pVtdYp%2rHEGF*R6&Ze#On$NdB;#XOE#8XxJ0k%fEI-{4K`z>8fhb2CD+Yw= z2hj0`UY73Tl+b`2mk5)nz)~DN-JV!p{4o=Sx&DN36?Zz0uGml7=GMWZlPoWYm@liY z#U~i@hmDzzs0|mVY2?To^0vx$Hbp&Wxrk^IvoB@Mb<}1^++lx+h{VU9mX@z`LmD8Y zycf*0^~D(9d>UZ)q?f^W^6!hl4>9DW7BfWL8{=V|c-Wy}{*k|Nwpn@E_e$EpJT8&7 zXGUu2Vy?+yu0y-G8Q#Q2J3`a$rQY0BY)~$Ft2SDFCsBx(v(vbSJs~3s028K(x!X6elP#3Y_;=d5HR_g_!61cDYaJ#2@w00`A zW_1IDOvIMyx#2m%d~3Po%T$okHGI;U*`0W z&k0xWlK5QH&%^Ej;}CRGgp%4Gc(urJbY&UjYMK?uGaVwK2f3NO%il>V^R{J2hTr2c zZ9`V2)){bTFG%8&(pSVh#u|Loy)YrLXURaFu7%ean7?RVA0na4;9I1Ufao{;q$DCh zE7>2EmRodV6W!z9?2n&*#X-=7jQ%v`HhwZ9iXcH~xKn!KW4-MbyFpJJ(pVE&)dONd zsi#*m{sKV(#r_DRSz$twk-(SR<|Ek)O@XTmPZ*wGzk6Zka@W!G-}sxY;FJ7$LjuXS z!Z$~7i`%I81YEY0or6|s*7jUa(0#=ygm`Mnw@0-lZ?%p1!SWEWGzN!@BIH7Kr<~q} zamVMZyZ+(SW_!5}+4tws9r?bNvfCf%Rl#7ywSMa&nnEN#@R_m;aE1UR087Bya+iq71qh0S@5Pby`RNS(f_z~Y&Td-IAQ zsW#e|!-!}HYW2rt$$0eDjbElgysm(BjiHFLj=Xx4T7n6ME3NTDMyIre*Q)}lk<`uf zbiMg2Hb|#I`SBu(g?sdF?`;rBC)k&o-m8Y5On*$nlyCjU=JqXxb37T=sCIsOk$$qk z6;QD5S)=uhcyw4U<$b!D1U;wyu}x+L=T=lVHn@05u8f#%?}Cc*j?LdJaJmhw)d2u= zWyH{kC%N|sr>oR*4YhALnX9IyWMELP3$E8QT(xP=xZo}_=4S@ClD2s#O0izOeJ2@4m;ywJzdLOMvqar>*7zgGQ`BNaUd3#aqGKzpv&4~Bn&CQ@*LVH|Ba!80NL*-MW62jIfv5qw z%)%Fva31wrMETKHv6Z{l>msBI%v1qz!``erUD?g*{M|-82g%ix!@I2|@xu4>d$hJ? z<9(UZK#gBberRyYm*vbDI;++FOj%7=srR4UC5l;tG%q)CCN1ptJ)TRK{UjX~a%g-r z2z%Vj6H^+mQ#`2=f9pxi`gF5HLN%j@>D`)S6Gd&{!t4KlwI_f+q6)e;$cpqCj?>3b z)G#dPb+!}vAWfeito+s7MLle9z+O)h*Wp%0kH+(We6OM+6GOyEU=*oPakjiH!=Tji zC$+=Y*NEDZ<+9N5m`L1q+!frf8WrtNAuBLl3v0&?UABG~N5mR=w5$sx;EK))KfXzS z|Jy^*g-Va9faAls*6z7@`cYHzaGl+J%PJm; zRhT=O`-MtoW^?^zp^`yttoyO{T+*OZ=Yw}bq{^85`SLO{d|2JFoLYF)3KAAeUVr^^ zXL`Ha%G^OTz+jwSn^tLqX~fo02VJ?_Hf7dDk$d+YH4?VW3}JfhD4ll~Ds#XJRc>@_ z?pfg?_)%fV%x7l5X0as6%gpm~h274oTtQaS2UAZ#cmu{f#$%i2Jl<@A<9zLqB{>AG zWzo9xXK-E>0)%y4~1+00xlv@R42sX9v*F1qObOkquF?{ z!g*pQC99I+=j1!ZkmJlHW6uB*G87@k^^j)~Nc+1<16*e|B?W6jbJ2oQG^ZRs@oTF9 zo;Pr*pCRu~p1*%!o;^RVwxhNSv1UY}#AirNaFyn$>>d?cR;iJ^pAFq1!XQJ2CTy8yaCjlo7A8i z?wA?cH*mANT-P~oVXe+KcB~`>=K0R)jIC2!34OoC=7&&o84%)u>aT0Ngk{`YYAo*! z)F7r-@2v1QOK-0;Vb3RJaW27;!r#M(RrF96a^&VK=0h6eUSKbCQz9S+Vuke5w3oeN^1BxD4~&3YOnv>S_D_r9a`CTO@vz0_cwm z)Ghx{S07*hPgftPcl;;(mn!|y!Wtab;2Or+NvWy?#tzk&D~b1Ru494@lOf!Vjk5y# zFBv-8rT$9B?M^7@EXy_#efr}20gaAFZw81gs??05ikYA~8VEDrNy5V>9-RddVQ{ijUvV0&p}^;mZY+f; z@^T0J$$zl2T==-C;g6`@(l_nAq*p>PeQSZoRCC`}GaBAPPngC}f{A6?(P*(`zox`9 zn7#IVbSXG>`iiVI7t>Q}`sx8iijUmH2*eb>Mu^j{+K+A|e!uKwLRe73oIHPld}-?! zG5XNBmmj_ChdJdj1&bW0OAD>;JV`0yhESXJ*H#ox-wm4{T9_6qF6z6kKBiv6_n_Iy zJpOLP`fWMcOtrVt_f3*!XL$6N8Q~sqV?@q(e{bMn9_RymuBPu&6JL_w2b6&WKb;RJ z{8+S&Ig)a>CSRf&{`iy$6+B3F+0J0{G-a1q&6X?3Mp-obmnRqln1@F;dx)0z$2^Rg za4Mz_Q?C!P)lFE43 zoeMl&sK+{c5DcR&*XG17%VbnBr}e?aY^;oxbL~0<)D#T~?#PIL@7g7oeqrcnMJFAc z6Oc-UHu_zUOyLLUooJP_4ucCs_@8{xZ>hSfeI2yj7N3tb3RH$9k9HpVyZ4R>L*BQm z1EeD$8M-FX@+EjQ9chNja~d>4Luoe-mJ%N?VkeP z-Q(0?Kv%ssNu8q36LvGt$EhzvJ{o_I)Y`(6P8G@ZAQ({?f~ynYp?FBJMWx53amXCt zpZy!qld9gJ|u$Vw3^tdRNY24x2c7gb6cS+hO*Wx_hK22l3O)zV$lbeYn?00R z8aQ1>7I~P;%2GjQjD)O%1x^v;q+V&+(^~xl_PL<_+itIGBV%Gyav09bb9{ypGn94slTY{OdZVpqjl(?kE|@(sni)|JMezCeC5CTtdZQij z^@GfNQ*1MuYvLOs=GV*HcL$$62prB2aBwm+@b`eW(! z>=V$+rr19^%Dnfst)VoU>uupzLjYGnDw0RxSho4sieVzT#7YEP!2uU^tM9NTZ=&6{mKd$N z>+;v4wwK~-peI`|6woE*T^0S){#WwKHd)4JKSA^?2&pYus)L^LbG5}L_-MS-d|b8z zaPs~k`(Fn9pvaFxG%Q)Hiy>zlz_6;$#G;#Dw?{0)&TWj;mF6y}9bYo(caU2-=%lze zm)g<~V2IxZfQS5K(J#XhFzNCYQQu&Tg7&NduXMr9kLH%xK@kSkZ&UUFRYWY)@adcoAk_Ao@#FMog-7LKdl`~ufNU;;~bBcq6>4dLMu zrc+?OqstAUz?w76g64b-<)h%zP01PLSq?Fz9^=8q9-|{`%AS26&W`y^ZhAD)P zhtkQZgS-BoR4z$z4`IP`l2+`+O4T;_M~0vxkNfMn#U=3@)Ci8iVMXae-(PoI3+Nfx z1L^5Gm{YjIg8VmqdhtYdp2Q6o;j0Pi!d$jTo4dFWb=fWDR}6MUI7v-AYVdTso??br z*1N4lzIkGTpR3B@#$w^bHE)V*lV4Nr!aLR_AzSnIiQ+v|v1|8o`aMfBge%%6i1kPA zVog7usqojZqpBer2*0^#^G#>Kx_Y72(D9SrO;S&rG#otTz9IgixPI#%yub`Zc`j_Mf~_^k$L z{j$1oyqX}%&M;6b`YoKliGHa*#o>AsR=ymLV4;|gFiEi0zM51zJT&joUWxi|KqYiR z+=Jur%?GuEm?g(A%6BWDWGD5aF5MqUDBlbA5+FBJ`G2^2P4@qVizGI&;BXAe#&Yds zL-N>uzj^3Lu8H)@Y2JF>82^8IO>-9)T1cvtdhq24w=&Xb-=zR4y)z<~CKe<}vg$ZW zlYofPN6h3V=CsaQ!rjo0`(`1qsPe9Z8Y;#aNLz!38r z6d5W+=@G?0pMsrx$oyoEAtc9OXV-6HC|aP>TUSEK=Tha<@@*S={1hJv?W*i&UQ)9~ z{EKSRX^(x$Kp@G-N2la}B4|Uq`0o;kBTWwaltcx71L#lj#Wt;KE6o#I`?+0KyMea; zvmAZ|DKNp(#Dg0@vxcW5+p?WYywWMT%Rb2!+7EbRmWcwulM(CmIy55r$(Eo-T;`YF zl<)l|lrVq3bq&D4J>U|lqekkyvKVR2g^GDgalKt2_4pWsM?Fb1{{4&F!ZT&>3ldVn z9e-x!e<2e+dw`?Z#Z?XeMR(y*Mhc}ai(BjO12@R8`&CKOw2pL&1+A=Fe*;Joz$ZEW zpjWp)wPc(W9`9%7JdFxwc87JQV|jM>Olsc%S|Ch78Ebfq zeV#F^kCEMY{>gUr>cgki8a%Q30y_OorM*=yN*38;Hn|A%J`V*aiCTNklwgslmN}CC z-J4<;m+XrjN)Q8S9&>ZQfe$aMh}yr${!oK__eV~DGXK8U*&9#e`$P?E+x#+DWFM$}ze2-9cymDa zvgey*hTu1)Vq8nTz0r(B3y(#u!&@yuj1KK>ICCB6+X|WPM3r!7e_N+Q+k@-9@pSl@ z27wX%`j-&+OWj}O8=M`Yt#SemWI!*Kznp&V7G`G6Gng94%QZXiH9TkGdl+xyFfHL5Dn&maKRJofBNY|B?oHlBGwBFNmzs!IadE$PYZ6lY^4=&M z*zUOycW|$>#dQl98hxAIvp5OZgUb(uuFjK!SIx*dIDTw+@SFrUIj6Qkdh^fv9IqW4 zo2Nc*_n!oh+P(VfY7CBU_Ph_D(u=rHTX=Nh6ckG6Gq}?xbh<2{7j?@g? zHY{t`@kTH8K6%CH%luhRhYRwo83`KtU(GgtFds_221rTj01W=WR=QrRYxCwiWK9O) za6xqo5>$jRud&1`z76NVzZuoop=4m#|EBcP4F!Q*$%3}_Op~sFE6af)x|M8>(47|? z_v`v|{p}-Ha|Y^mI9dd-G5R($jS!QYoqV?j$&oZihAXevY7eL``}^nk$LN|lsXd+vd`zid zz1m^k`mO|Or%S{wjc9y|^&1!n+0-;Ha>k8-3#73zsfHEP80LpBS(?o#p}&8pE3h4y z6VQi%G!0%D%X|}aJvocl?ZTXxht=%M%9}A|sl$bRJF0}Z>U%;hGVoNLbVP~w5daq8 zyB7Ata8el`1fFCk`pP^jvSk2ASJZrs$pCr%ob3*+FE~6ESn5vREZVUnyGzt_WbW7Q z&)poKW2b+1XY8}#^*c=qjL(HhUHwuuM1iH26LM65VcM#`yN&ZOX^7l-LdM=kuDtoR zedGMQ75Zgy%E&%l(9!aPQ=9^5fO^9lzNwj>j+vWBHdq zRoXcWX*G1{&IuL!bd3+R9tPH-$WKCy!gU5aZZ`V%U+^Svr^5?h>3wELyd3-ZB6G|( zMqZ;u=bUCh!#T6G6`eyYTBk(o%^?A7EALa%KM}NFG9EShO(mm#)1JMQwy_&|J8Aft zO%oJHu3!`>2N%1KJun^d%FSD3;9}7zECO_TF+&1hkd}qfg8&S83n)x*|5EGFFHXMAOIBmRv2{lk*b% z1tRKOljwGL&-QvwUwsCqb~B7|D){-*>*XxLjWA(uZVw>{_j=BUn2iCsr!%+R%mn^c z61z#YlWOvA3q=CXpF4?5gQX~A1(Vnu{JFu)?|-h7-b*1Qby0iN?s8Hx&B>j6!Q#)~ zIqjrozTw$EQYnNtIVVqUdFfBh(lVC5sDPKQl{@A34G`ivV;1Z!$QqRMY+t!lYBLby zo4c6Sgv?d?_~D>U)%6yuA<4dfqqt;9@$1`M+;ylXP#AORhM<^6XnI*?y;`yzh>9g1KC{G%;UcYb-&R;1v2{aagL=k`Un{szb} zw7f{b*ZUwj!Y-J^>6P4cF)wCzqwn@JSV1wkIidsJZO(sVvMqd%ybJ0q{s{@+``?kw zC=V{JIgCgI&|&4WkiH%+h&Jl^W=ZrFmSeK z-X?{!R2PCQSgvU>ZXS-Hpvn;icNR57y};=n!(}elg_KF^T_qj)_%6A!f+MZKk?%3b za(e*9X$;bNeWB^pN-$2UEDvGyJoYU<(uxauEqwx z=!;Fg4ZK7R7*rVDUg5R&eTwkX=;oH93CX#HDH*7Iwk~ZGy(o_&zIdz~>5BQID(RW| zs<|H`3UG6IqX$voebOO{z4gKN$?AwOinp5LxwW@*7D+?9m=R2$D!2-WQyW>JAT4UUScke;(1uU$KrP6)}dj0Z_#My3sguYQ0UxnFi45OQWD;rFB z7VBHbZgebA{iMu`9W&W?;BS%Ca#EUrPh^HoWTselqcWf`qdDOU5Kt^lPW4nUE4dNk z+I&r3J0Oc_Tu4))@JYF8;|N?O^hS}NgqQR`D@~i}DaQU#jHU1|2Y1@}?EAfFDg6>?C^ z-8!(RKdImrz5A-L_i)pwzp%0IL&-H)8*=@ugW2X>egEYHwr}*pa{G!x*GT>g{2TDA z&+=`vT9(#OGwfl9L^a_T- z*nRIo270w(Sm z=<7rw!B?E=>E{qN@u0Vu>U}iLwqO!`*)LgNb^CAb^hjT+SA~Sm7&rv#RC1yg$@8_8 z3gtV%RWuaE+{$369%*)i~I%y%UY21#?UmFNUMIp0G6G=3ioc$j%Ia; z_!Cr@l-5}Ca^+aom^T01bK_IYmw%@#-{l+^sFnZK8md~0i@KcPa`XSLfAO7L^3|mr z$jYu&O(XZznQGn&^_LGYMtmG!s}Xc1u(iZtmEM!D@(|DL(~1yx%B9EreA0j?C*y?M zhty!+-VC;}kHZ&^vR>!76lLTPRhJSX)bI)U`gnnZ%5WT_l^7%4-7YaJS-vW^$>isw z)1K+(H+0Z(igcn1ZCxq=m2Y=Wm2a#W@tg-W##8jqJE8C)m5zANDNfr+(fM4Dqb+5x zy|IHI{jG3NhrZY9QKsL_mE<>jIwumE4+uIZI!AzBbX^NYQ5ml0T$zY+nk%pV1o!MY={&wVSzP*6C%Aw2w0s}{^mm^< z+VWn~dONbi4RnQyq7$`A{+b@$qWeiTljZ&C^b}AS)jASJS@BMz%%w+iX^x=`?KYh^ zaO0Tp^ao9lcIN1dk+XR^OZY~IW3T2?rj5_>sR=Fb#ioJ#H257+S^iyJS~2|XIdvL} z^iOzBs>tPGw4T+j7)ep9?uWduZx=3qgBJb=MvyL{D92bJ`7E@ z?u7aRW8+G`(l9h&NGqY*x}EO`jS}}5JQfyJ+Mke!b-{XLtR>^wLk-EGiZOTR%{!BW zw{)h}wU^21CoE*K2QyO~&EKK;!LcF!!L6XP(GN8SjX&XY*UDANv<&9=|+E%D+XJSS6?Q^57b9R|j$k_}qu|N?57i3s7 z;_SC-ti?ap&r|h*XrNow-x42eJgM;4pVwjB+MVhi+_jF4B>l~+p>}b84 z0`fnj9^iuaN%>e9j^CUJ3dFv;MN?USJso9yTGKaeelP+=$P>ELKx|o%_|gwX+1Upb z{&v0R5~iC@*q~u9TJbe~w~4tos(A@^mcI7({8UpicoKCQH7Q8@*kvM)cb_4R830gq zIl0G`5r&&bUvlZQ;l!QiZ)+lRfV03@nYN(O{~il$-uBlg88Cxl`Ew#%MdIUL25ayN z(UwFkWi<~!XjH>vFx=Nh5~kN9hIE8Trn z!EpE$I#AfyfFaR%2lO5HZ!e&;gFZDQyuM}$Ywb(xfD2V8#^s@D1Md%@2Ts}FKvmRt z%MEgS<%9lgq<(-e1}5moli_grs7$9;ksqol`JVJQuaTI#&8GgUP*$(SJ-&q;c<;DJ zCne9JN`Ct9mpp#HJ>PB?es!i;Z&T$s1C|c>iss1Tb7HWovkD1?7 zxt{b=DU!_Dl@`x6R01`^0pR}&0zKi*xhbM6ZUWzDH(seMuML({4C+ax2tQEg#b9jAEElh8D~}uJ{xP1WRhC;&?5#Nd!!6l zYX1}loS95d^2*JP*Y50n3w*KW>IyFoK9?u~LF{|Hqo53w<$XeY`G`L!zd0IZJT7A~ zwnzK9*gxF{Zl7E~37B=e#%9XHrFPyF1Q9!w?XJzLywAH^(U@1>FQ8c2A3>HcE40}6 z1m!aE*jLqSe&RFTJJ*QR6vho#DL#IMD~o*{LVK0KB_+|n-vl7zJ+FC>w1?FQW5h7< zG;8vy-rP?~SOvW)_fbsHvPrnMs4$w;-S`UVD#VA|8%)b zVbmwH_BQp7U)Paxu1YB4icoXqcOH8zWZq zBqP=@N1UA!&wH1IaUSbFRc7R{fArGp|34CgFM3p7dZt-RiIL*RRZs%w6a8k%H}lz{ zd){f#vZ$kYav4B_YkvOcYVAtiqGBg`Q8vsL|PAJKE8m6!o>$rG@TjvFT>NH4EJE{zp6DRXk-osHvKM{Yy{JJTHUqT%y72 zFr=(P2FQJMk_Vk&+G#M`>S(d4sK+~(c!cy*;A#Ob|Gmtsr(BCE{ha>Bj)~kN zRm5K)&{`QbJAO`F^c7)%SAlzA&YW1ETUjb4>PmB({Bm!tQN&2d+7W<0i}?%SB^T`%`+>*TgFLY4^|6*F;~>wAzcBUwjupXORz2 z$xkBZtf*l)wROpj-PMU~UlWekYJ%JVXX$AMbXwwbENf}*oHmDTV-M;T;}7r_d7qO1 ziCjX7g?@3{y@J*ZZB$!fM^NHVAkW}f^hOu`Lk3(#J5%$P{ZRU$A~(yrkgH0TZ+DIF zS(NuxnD+@GDxh*A4|?${Ds%IWPUGFBa|te20pCzc`!;wtz8UA+I#laoJ#+kFZnlqL zV1Flf8Z91rFulN;7H~VVI|lft}3*hlvIX=&odI{%`BzzWasyr!)a0Q51h|Ff4~7<8nce>L-%-(u^*!rOBR zaN&M6&kyKVxPMNxT9 z!Yevev4>N-8xK4R*|-N)63-+3D8j4e*9qD+XWk7M!U| zRezB)zHh$|kln01uFKgE21F9e& zWmfzdM0yc)4LX3$bS5pj4t!@5``d}<`sanhrI2pH(azo7Tig7}P6?|H%;lBdfO(9| zKf5Wqr2xKzv|UM3scVkI=(z;c%V$tP>oo=eFRDtI6;W&7>@G!SY!M81o((qe+_!8x z!D2nmD@UV@BB}Z(aN!@%0$eE2ZT}6}7VPHS8;VV^ywiC;#SZ%N%CIg_1t6yru+QCO zr`)mQfu<2sVY71<=-6~l6HEDz8E&-Ux;v^oEH_oB7uXP}ko*=Pe#J?e`{9$_Q7 zPz4UN>q>g=-sm^}bm`MybPZMz+*>~6u10Cu-pLbXdOj6r`y1#f8)5_qnNb-vT3};l@y#Ed|m&!PDMNATt1FxD#edobS5=vz1D*uUcf?jK`PN3?e?E?kKJCN zwN#wxHd4+cMrElKgN1?aW523@$Wgx|#pg5|w-V3GjLoy#Lr3ma*if{foONp1>$5zE zeyC}dv$gLQ#cX0nk)lEr>k{8Q)hGtOT~4l2tda#!wh$V6jUv3*kG@Rpct`lu0zo}- zW|$DZD#Sj%2oN!$s8QiL$ z{+sHb%7$B~3XdcA3?s8)vh%_>P%2yic+Gbz8WK&OYCOc93aEU)frcmxlmLUO_XXCB zq9XWVLmMbF~a&bb7^)EB2$F*(3TZ13YI08PGU7$RP-v(a|6C%#n) zgcALw3l6hO*=+Yj(f+Yu0??RZ?SBJfD>aborE~KMZCvM59iS_?#WBk$Z!Az&{_~+a z($PEDAC0qt85nVofsq>807tY`rXjA`U@DIe0o!n*-b^Qx=r|MBgG8Uq?bB zd(eTs53Srt2LsxmdIMq3!=wuTsM8-G15=Y6@#K0?Nv-QhSGci-Pbkr3fR9fm!`tG5 z1MeUok`!}RO@`#UJjlq{1Kpfb)AvCS@~90T-%wDH5=FZM0RLrX8w{QoPdeoZ2DO#4 zIm*i%9~RUY)34{h?2s^p;&Q8= zUn04UpbwGL7`}R0!LV%~+yle%f{8T#RX|`|FOxoAWMDo0iKUq z4lu6kb!0J$;5oeNKjSMZ{905^w2-9cxWp^;NrO{ji@nGamAoM;=<9)u`b-7sxCr0< zVY>{Te20BVNOwS;0u2AIL>~u(!JJ}{%bLCfz9W} zJYqVz5*~n*izCX{peIO(#^N!;7}`UG6%=^?g5;M`XGIg&<-n9AEu#@ZNrAZuxmnC* zeGS=<;O~$!i1_SBED%-&#$Ik2Sg@^*oZ6jSXHi4AQ#4yvd;TOQ{6%j^;umH~M#5n> zg1p9ldc^jX(CL{dvYm}x#T~!r zX7&#awQtN=bZ8lqR54$1QLKt42G6lyPwrH~jpb12K)$wpIEB#)eY1T_(RcY3LX{fA z)3z<)PN%CmIMVR8C$72inL)kfPJomBs)X_tTr~w_1U7bRyRtjkra- zWpm*aamPNVx{}{nixKUZPCsxE0>mL0B%1I5x)$KSj=H3c(7NQsUQ8ArZ85&%PxEXe zH+Q;_D@8=4b&~3kbe7tVk!OSkD`~w9OgwkFM;+6p<#% zR36rP!pXtNfuAPYw5ei<&?RkEgOy`PFlLw#W5I9a7so{tf$+6V4dFK>T&xKHz*t=Y zkz7;7_WP~8#J_=&sKo)&?HOw#C%TDlgWtfUL@@(~t<$J#NuLA$RuKaOO^r~a22(2T ziW9im{qUnR+|;wZDjb*8gx!=FX;_Jnfl`$(AD9)Tlo|7`BgVAWi7+-{zFFy+FL#6@ zeIBU&1|S`ZB3Zg#2H_1o#Fu7?SS(;%49le2_|R~@)&4EIRn<`Y_g&8>k`^kz+Anp% z29fpUlxsR&v_B=JwADrGqOB+iL*t*(XQmfeG3HpSSrjy?9TFb&H+^2^d+?D5uDaAR zSfQei$Moy6qM}8$fU`A=PEwOIdn=u(kE&Z!5bY}&Hv0ws+{C1I$T|daGrw!+sm_xx z`S1zaNgtb5LBX*fYBZyU@c|=C@$H|s^(B*nh!UlndZAAKGKeLsrX{t(fXBKl_2o=j zFEbloZo4wRlgARely^5+msd@jHM!cfHKI(ZK=sb5Z#bTyY3)aenwP~36TH?#^`@II zB;p(6O>>}(%$DgA{*8ev9_m_T4oS!&T%JO#hY1#q(Jf$;w}O*-i%6x!GDoBCD0MFR)-PH=E6Y%U*`yFK=cNs0 zvfEJFVAOAL88JZ-_-1AH$kMF7-B??6GYEHgxxC(jqp8=}?tG4a@P#m0@DQ1Qy|^`D z?-cp78ivCnW@!C(S>d8u0Oz(OOszL(Y_X?QL(GCVz#Tv0=g%m(jM3e8p}9_r1?tlF z&s7c@JE|2`3d}(;xpd5rp>IN@FT;A3>^ed)sUP2uGt{IA_&i)0G0u?9@Q^w!n1MR; zPJVPmY`Jv2n=V;hzhc1`LD1pQr@RI12Sj6UY~<29@;P<~AzkTX!qoJJY(5SSBDuM9 zWN7{DFU{=sxE`Aq4R)JqjZr`08630b>#xEwoa4HyFcxP43~xSrIdA7=-uqyNHm{n< zQPcmc{5J0|R`?v#b9NScLmmbk!wUasGX?un>V~~=!pa?L?fO7P*hj9v5=@OCIh60^ zlnhfvIHby?dhFCR19FdNVs%Zt%dOJdM+0}7w3Z^-dkp!QWZOgObplGY~bTZfDo|Tr>nz%4c??)Yk|jR zhzZ6$LDX)48N=X1zrZ|W9%rgz0HrUphJ>>*(U11*JxkSwrgt&YfQs9v`VDG{6niau zH);fe-zgoEA3fhi34-9pwY+eF+dJ5$9!RS#|@d|*zf%ktSJ-i zzq>xsO(QO^3zaKt&ng&8vwB>Sv_MIECoaJ*72*3FFQIhFth7Hqgu1AZEACst>>vVB z*%u^hqiQ#;!5J=={vf{Bk;J8KP~6CYh$n6Z0&iMpNJ4Uy5=OM??JqFR*3=;|IY&7*nVWW zTtQ@#LZ1TMlsY$b@EIDUsYMTBJdMoy!oP$DbEFB+*QNZCXHcyC;?aO%V z%~(Q_WI2iHL&^Q@m@OLP+})5T<=S?*?}@WiGs9z6@P3vLIm-VA(SHd<51!1tDPxS! z+?EWuSmsWQ&{QWY)vr&8svK`h(!lT22Z$j5dXXNj5eOkJ?VSsV$Rkrg$aET7$&*m^ zWZJs8?mP@k7#QpgS-0k5Yg5O1!5+amj0M|XnMWoKn|-%lb8k1Vz4%oF)}h?e{w`w& zcC$Fw%tZTg_n7bO2As%NEW4A&jhR3h(eIp1&1P6G521R|dPr06DCgc-Lsj36#GNA1 z$Dv6RcQ&$0U_GGyHP3Y;RY8d^Ys7^>)8(oO$GjqIOlwCagr)D00xh*b{DaD-CQpFo z%UN@!&?_BLk}} zf_oe#CATV^>0cla{yz^K2s#ebUjK>@EXt_iEGFQ2RucXuA;i*4jz06=5kT%=X488; z`~=RLZxHX=LeLTx*V<&-sKNPF5zBwp-ds}Aw+8>i0a3jnJqKezT(8c;J({HN8t>Z* zG+3kU3rh9JE7w=_87tZ{m*EZuQdWA=G6`EcIl*c(P6I-i`4RiTXx_bdc31aQeOQxm z)fxI9E+*{8e;IY%%zqV$*CC$O8&|w7D}2Ks-EUq!dcDm$a>)lj0## zWCA5<(&^iIGG=$N*eiU=P5xFeVo>Chy_r2t#5etmiKZL5pIac?R)Rr{z00|RL@*+J zs21F^Px@=_0+0I}L)LdidHpiV8gg=84ejd?>Lq)-{Qai)en=SPqB@~mG^Urq^4EH&F~~o)uj%UVPUiI$2URN7sXlBavCrOO*D9iD{fy1&@R zh0m;^%e!1$kT}9-%n`Sv6dSPV#P%xFm`X;eAuuePO3XUpTfVITOU!JKM&U-MUFvc! zK%Dy@utY?=@E7J(@4r5=YCJ!S5%?sE<%QJ8hIP?#_~&tXy@|yFCykDv$mHiGqj6a0&?~;d+i;f z_{}z>EZnuo^==96@Nk=b=|&9Etvxdrxt48wnD3A5Y8*D{hNwEM8f{BS@pEe}CFc{H zlu{(ceqQRp<>3h2uyVn>LHC+J-#46VXvGqyzW6`bdk?UtmVIqJprW9lbft)dCZS0Q zy(qmSCG?1NNGQ@3kg}EDL3)s05}KfZ2_Q<7-g^^}-UI~c`h|T&Is5GYeeU_bz3>0t z=enMU%q%i9YcgwQer0Cf_o#eq+F==SyAd-z)G~>R z*0-WLG$O>9@V&$N$YqPE#+CJKg?CYfN5{J*|0nw<+9JHJ7pQi@)TKG`_t7hI zOs^^Z4_UME>&uWUcXT5lNa=;BY0kn>dw0YI)Aj(|DXKVjETdEd$^g$L$m3nb)56fsa}wpuc2F4_(Jg|gj5=~nvrJ%7Zz^~%yDxT!wM0j zwiv?_^HSd@8!9B{BYQJNO2re({HrS5Rfx*OC`&Z6Lrp5>OF7j%|>LnEV z5y7nKtmJ2^5cqj+dy3}J^R*vhvSko_l;q`9wK|zTZ zFr|Kbs9$I|==pZL8KXPf{g2uf8z_-s)?Nu^C7hL5>|b-!!l`S}LO z1ufw>gwLXFS}`;i3fmpQI+E7Xurloff|};Ni{9@KKxTRm$wo@k{Yder1Df6iFO-%f zL}gEG_a^1TL91H)APT$w_|aLPKcHCnp#Rvqzj$Jd_(vc;*LqgJLbs zC)Ld-IhwA1U?C;FGI^)@)%k|N88}sR7bgl6QSe4{eVjQ_`B`>%7HjXu^O6k43xx1w zLXrW;irSj4IZjoFTgyG%39*tFz1S~F#Dvl*B9hXTwAmp1TWY&cW#kpiASyS@r7X?7&y&q8?)HQz_$j8u1QsO6KQS|nvCAr5&@zbms)mtxzX!_GQ**(M5;M$yo2_Z= zte5bq^a0RT=OL+{@hp|W#rrD~=Px!jDQL0T!`R2RbO}A>7a!yv3H!4cKOMegu8d*S^RruchZ2s|5)7|{dm(D{fPdGIE=eL-%lLpH) zmL@LFGlvD#5X^X0#me&v`k7~C-4~-HClMCwlr|WZL;*ilT4a`V(~bDpaw~#3XfbTD zG#zHr9Bijr>b%0lz~7wIwhvMkrj61MUqK*CIMTVYvUU62%ZO3)A~I2&`SmKq`1wFQ=Y`I3y_c~^S~(~F&XX3 zB8qjGMaOj@s4xaL?t0ymMJ2j9hr$qY&u1tWo#QOTDrcn)1sl|rxibHjSJ2OrPc9T3 zD=CuMt}Fy4UPc{I1(gu<8JLH&=765HwaHQx9MVX{jDb~B^1IRo^ZR8BtnvkCv`fF} zz-9V9tYjeEyB!e$8$@abVK0vOX*u~WR1?_gQw=A5c42;6isIP4)FGCsI(r^}YJZN} zlm)1CXF+|kY+p{ z0J7DLNr`U(QvVJl(SK&68}%hmbAym4>Bg3Fre^e&yVsKD`FGkNDg>#6wfApGg;0nJ zKA_r>ar&AeaL=oim@p!Un7*XQX$(vFTfQ_4mG^u9+j=_}K!cR>BenBp_AKKzx@b5P zzL)Kyta<2mE#L9TO-dwv%0_^VZ$|%#*UK1#S5mT-`>DE66Qwohyv7_TGi|p!u*9qM zv9pYk?1W=sGkzwJSB(a7nzw&sx9oR%;ruRlCT^n za*dWknJ#2)0I77tm6sD@*_l9QI}?_Etx}fDwX+|drni|W|JFCNbZg>ktc_}H;MPE7 zuMqp2bCt@R3LujeCJ3F9*WC^bphq0_V6(^o)f`DbsMuL-DB#?cM_oc1j9n>?B?G z6mzA_&X;XQ87@9;#~!aFCa3kfh(b}#(G_UrfpuM|4C~UA>oAr;C{40HvV7K;;aQb^ zkVU$>HIH-iqvo2tsM*Qu+)#JVcH5FGkx=?B=nCY<(OW9l>IoP7=*MDPq1udIqCDD^@#yU zPr3A)EWyyW_w^k>am6?l;lcxsW)7Q2tcnfS_54BM zR`=xZfv^?ZuUTajTR{v>ZsYX&{KeZ)E|p)7&Fq*KQqW%+onSuu9~T zJoXs)NOR*g#-iqqH~e(I{rb+3jwfBy0vQZCCK)KBw|O~EjKdPBE;!V2TDC4%saevo zBnk*D><1UdH3h~GU=#;JLL9)?FW%RLJ4N0Q&|-8)haIJ)9WE!NUeTV~5-u~bgHev7 zmm-g7Wt_;{Oxb`a6Q_XOoBavkL%0S?@f#qD1js|OUXm9kGpz+{WNQgwznNtzYg08Iep;IwfjQJO=wThHD5w* zcMgQ8*DcXrbNq5^w5(&D{$pOJs+L8nz}@5sFzuqHOz7bbn$*ue>j()6uBP9hRYJBa z2!kHG#0MAZJe1gCASIh7VkE%K;Fb&(g82b z3F*%mlr^D(gOi~brb?^}zl37>Bj8rgH$Z&X=g*(KXyv^gc_NiM zXYtsW3F1_o|T!GIQKx3n9E9L$0<&q*a#IVQ_h1Oom zFVuM&Y@L8;ffX@x0{~^%-8O&qUI1Lc(KT)=vI{Tts2J#FDx~3td4t1t!831?=^Qm) z^1h5To23_LueZ#kF9~vEi*+ScjY-N=yAvJlOCs84X;J52k3nGphk9mmdKb{l>HJ77 zdESI*3#zTF9Q{NGq`t_qQ5PDBmS$l=PGY`Qi^;p%J^F0p3*E+V(y0Z60xR039ns4T zE~R7{57rkonF8~{R7d(?#q4X{Il#&ws`0>}$w}@Hmltm4Sp9~n&W?ZQEciNtRH28S zA(lb-7BQMA^&%kiiZtcP*fTX~-ngH?wa;mMs7;1@BURB4G%5kq!vO89Kqn za!ut<7{xs{hPOc^WZJG=SnzavlALUI!=$Oj6}@Zyq3Y_%L3eVG+%Pbav7vA~#D@Y@ zoC+#wo8Tj{M6|m25Y~O9#PW)j$}DR@eZ!E?k9`m$`mC~7X>8U&xV0FXl`|>NW=gkC-ozCy>2qx%5ST4#@JAaHq>UTjlY~Ns=TAa!ZmfAvO}xk6=e0A zrk2SA0MS^$Uv=QY&o|1DUafvR@H1F9uryk}L6@L>p#uWFIH|Uz^G=;&uPNklTullV zKy|pEH$Z-t?t%`tnp9*=p(b)GkONByr(dI|2`?xwwNdf+yW7+tt061eyFn8X zJnuc!CVmQ3(C8NSnO>+srfVSgsE;jY9J|m8ZEvyMjqMrL(>@>#QE;tw;&~JmAXIWS zHmTH`AOl_GZgrH@xs=UVu05X!n-g55m5*r}QdSP$eZrfG!4~=_~ zgYbjUiHYw!+GIWT5+W9g1Gg&?t8Ng{eSW6mcbQ)!X?j(~JBQuCmRF&#JNyrGy7RVI z1bEpHAJ&~e#u5pvWwhlLwmM&);I0C4+n1O5)YkPH1z zQF+0ra%t&-3XT>0C)Tg^qXu&X+X50TTNQ;W3_ndl>;^eQnA>6S;cJPtWc#lglt|m} z_)-{w&3br*^ot-69X(>_?8xyF#GJB&`JI?pL-;M^W$BxDKaVtd@+-IJc*H^w#x7*C zys@Sc8;|8E2i##6>)Gj`tXHKRJtVdeCkavBQ>TDH@oU?$nd~i~f zC#9y4gjDkxM$~{{37v7a@8KB(}s1l2DkQ^Yg`uSy}UqFzIpfC%la0!>Ee&oDh)bWm}jL^VK;x>d5FZg%=1YYx)L-%JuA59^F-U zPSuj%<85Bf)XD~Jz+I$GlL@cJYrUy2)f41$5!_o|j>u;aAMR{$d$uH6cAmP=xBX13pBK`G9AzrKy_ z>on6#v2{mdeyU!7&oLl12vpi0|^`IQwa>Td&FJ;Oxt+>`IY*T6b^cE zbk`g)xdGumAJm18267us?bN!5(Z<_C3P?-g?kI9yBs{}Pg#t=L zpcSj<9D`D60WSDnTi@+2>N2?M&UvZ1UTCB$2Wakyq_`{@_kVu{KAc!@vID(r* z6jCK0p;Hs)k}?|cWp}R1eQF)Rq4hT%jv}JNdtviPpw6a2Fo}?#uTALTg^nQdmMu9p zO;h_qtvTfOhzZX?;Cx<_yMS5$Ijb0$q%p=MsFT`9&&$TH0%*Zf1 z_}L+ohy=aj6UD-2;Jh<)8Rs%6yPudh$_AsONRA~IH+@iX=x`WIUp}ganA>UpoLWF3 z)SX_Z^)e0!_9VzR=Q>Q&xz#}sqWUc42_s7?H&?1_YptE_E9*+FCCnT%kVakAlxM!` zIBRrpaJ-C6x8vx(dTq!z!0wt6oFIl{njiSK0>F4!9d!AzFn>mRn_*j{H!?|FRU){WFRFY{X<`6tSe>c; z{JOM~AQcqcq+qEmH+&8u3eOwfEPpWe8 zA1npta>qW9AI#RCT*bO$PVG4U3%4Jeue;E`enQgJL(AQ>n(#P0bv2GWvh z)o#q*O1|}lC?z^eQcElBI?F)d!LpXtSEi9w>ld@L(pu7580KUf2JV!Po=p%ax>%z$ zcsixo15)d0IP!_k7O64ol*05kO|-w}`xqm3HS2Xb%F=LA^EV2pOS9~BHW zdMzjdfd*>65IYsVe$`oW2x)eC8^U-@Q|ejDJD1ls2{w#2^{f>jv{4zA{AD$;OT!5=P1_c?Vh|7SI0Dy!Ma`abLYH-gEJ&T8%686OT=#`iN(DKC4QYs~cAcAZ|-lz4HmO$g`uTHqSD-3V8%f5ZO z)%T^s;Pq=l<6d4OpGey4yi?3MZ2kJFJNBlVLODS}Z8@;cjo+cS{u&IagD`F?c0rpE zDu&^6asGoshGD9B-#@xG&NDs+A5Jq7j`gcD8@|$xiR};0abH-NtWlp zo*5?RO$-f0G%*F|+@#`Vpm%#EA8jGcO$G&6_UPm4)@j0!{lxCV6_Sfe)t&<|vYb&C zqE~DgVsjSILGhA{xmPuqH$#5ANp2_@IhQjtS-=|P6*zz@3XNTveN!EXa)9uvHM^;(Yo}HrE?i@;ZF7{eS>wzX;34DB zajbijB|RxKkXm0Q8=(GHaG#d}l53?|0wvdG)X5)O2+q%j=!bWF_5Tpn+_al{v!7i` zN4qg8K3?aYU3nd5E(6RDWR6&Z?^@DK%MVM(C534w1`${|yqH$P01HIliwJy0daVwn~3WtJmh*f=ZxwaahfAm;2GXVtB44V!Zgy zY;?C~)yDY@8+{liZFwR~nCYKz^{g+)GQTT(Nn+i4^?5AW zq)-bA^@bc`Q2f*%Rn{~Ek&(d&firxFC5*5;Y}*>`m6hef3;VlG&|i-Fb@u7P>s4`Y z?(nAlj!(5BG+#FhNAzmsAP<9;loy3h%LTa`ouE^eK zTLvx~_~73L{9Ewo{~2E-<#Wjwww&ZZJl14qp+xNZbDaOS@(=|?X``qHN3fN7D4DE3 zwRk zyfbygi6r_2EGjZ4;}j!K_GJ*~;uT0yQj#;-Z;J`bgnU`xz?HtEV$)5Ao`U)|OwtA& z-2-Fkh-+9zGH3ZHS~D0Nau4@ooLRWy19#1=|3O=OTf3^bqY2w|n{-gO;XG!ksCr<@@O5JlCF?Uqum@qgqI7b+ zZtw%A*#QqmnMY;zJjXv9+RaP&a*EA3T%e>Q2p&iD2{eks`p?1Tb)X_nn z*C_1Hx)+Sy3!{xpaW=WP{5UU(_zLpOSg|9zrO4?O8lto8;434Z?Gy=co&1Y;4Zg)5 z=(-BPp0~$-003C+0suILaQD~XNM4j+5X?Mp{73*aQpsd_rfL;rrmh)v#B7N(}PcYY~JZl3@2xYfXK8(UV5gefbMQg>; z9ZE>snAy4X>K8p_8`gdo1jc`<4kfyX2qIp!bHr|PQ3n@2`F!5FH?S@k)zaKj#Gl!p zM^+`|pK^`0r(HF;gq|Jj3e!WH50)v89xUFwbs-}mhx;5v#lU@`3%P-MK;LhXq1en; z^FD}(q~1{g#-OP(7+_pE%R=4M$IA2Cl%Qm@Pw)pc?Axnr!b)I&=Wll0ls3#Rd z_~07;WeyIXz4_!}@CByVGjzLFO*4D@O)z^>OY#rWa_+I#@GU%mml z0KR-cJunP#d5-i5iyrbf9Y@DEO(9$jBJ8k{U6cb3+czfN^)1$Z&RDZGS}*-^e~AP` z39S-6Hy&fHbDtg2??7u^jmx*ir}xOs%IOU%lVl8(e?TAlR+lr`AF0x)J4}uLVkA5^ z#vKaLQG#f~QlRLF&=;HeUznasPLxvMve2043@az7lIc(LaIgF3`t zI9CnDBJW|>h@m_UP6!M!?;0w`f}r$5E!M)LOT3ICWitb-HAa# zPAa-VL4LUxnIDQ=QIuAu4birMkvc`w=ztf!38PR!Q_^hL(AF*I9(S=!36qT>oOMHW zjFhw&bA4^o+RTdBXGGJxQ*X|1St@5j;e{i4=qBVlTTeW1XQdkgH*YS00(TVZoDJLt z9htrX%%l<$gXa)ACW#rK_XEw(M0vr_bTl0}$wN56sqB$+H!wWh#%PMg;@bSow_aCk z5sRXfcbc<>8HM)i<_?fjYo3?DoGlcBvaIpMaw z;Xnnq^2O|lb5MSnz9bc*!WedI<;Wm0UIkdKiDgY%qkM$)2r8+)f`OEKAS#qwT0VqZ zD*)dx>Uq4u9J;0aquQeBxhT|;@t#zS-*qnw**5lT@`3YWn#)Cw%wWJM)aV)thB{;VHD+VYmZ8uz*4!> zNPU0$5aQD4n(0jAx4OZ*K|>iCHYyevq9ZTk_Z4I11;}aF=_)-cFQ|$0Gv=&oX6_(P zn6-|92?~{zF-|H;f*)9KLd=`kR_=+AsqKj^@8eds8ppfW`Gm9COKUQwmUyl zeXmPGn=X{LDb#$ez=SEPV*1uwT4|G8)0VuE0LA~oC=&Ov^Y@`p#LHVeKz{#z3L3A3AR_5-!$ zvhbVi1@$Z+BIrpzw$)SMEGgWlfybcX=5ae%%t@Y3oKmxqLYWMcKnQ57KritjyHk28 z4PzjZ)MOaKmtGjL-=8*T!eZUBTL0)ya~v>)fXO0OH%a$qH6aB7gTZ6CnO6;d;L?PQ z&No0SSYiSTb@_ElE1XY8^&)vgLc~(%=KStX>0>i^t9mAe5ha=EtYF^WQ(xV)nA@he z_CA4{LjGMq*n)#`r=*>|aw}7}=gk?z88y7YpXb(XDp7Y`(GgK(A&kst+twSIwXi@+kP3^|^|zYsBPZ2xVWPrYC~IoI zjC%ubdQ3hP_n=;0wL^0iP1%5XDT;ODQQmV6aBB@ zx%*`iV~J<}-z{RmAF#vv;ZdlZ%?}1PtJ(Hmq+p! zKg%zEmj8x?i@*3;e(|&1KbZ@^ho`VgRd1h0;{Kf%bfpW>()T0BFMgI^{475@!*S#% ztJp7^!e2Cn{{=LKzxY{>`E`Hsvz(rVe(|&X;%E8A&vMMJ`@e~wg*p#di`|?H-bm-x zmT3u`&(wK&j~v9A7IE_y6qx3bxcjeZW#yJGx8|7FkaI03Z_u8U)O2)b&$+E)I{V1Ui{a$kYjk-M zKc4>MTZZlPv%NeYg5$T_cf(KWX6BW4h_Nu((D;h3AwKywG@_3#{-iR(C*MZxA0zSy ze23?!DrhT<{|YkFz#`BKJPBPvT%U0z&0>R)3oBAQGx!F;y;c=ba!eo`cT6B`@n1?H z{7W+blFX;2@^Sp%l+4#o$UYah`y4G6g13d0K>e)N&yp6x0`T@CN08Y;WrluO9ysr~ z5(g*zK^3~{=i3~60Z=}#FCKjrbNuFE$*<8*mvbT}I^E+A>xz;jJj9j^K3@tQZ)jsV zNk!}2eK7n<<3oa(#Tuuv@M7l9%ZcooTR(C!2>e;u*REBDB&cGnj#hS%_H93MHMNf! z*1Gh{eX^`p@-jZDk-x0n^c?BQ^mluCD{X^HRZ~C*N;( zJBTrD2QhQMOzcSX4|>Q^VV3%tn0JG_!Z})!ub zsS`e-#+6DgFT|-G_@^zcI$w4T_>3M8c^yCGE$#St41B)1OKgPfLm%JnL7WTD1%;HH zlWv6jd|YfqUjP);#z_)AZm#((rz_7sS-xekyeiX zawnx(g~|d~j2<0P?ylznUPr!p?gAAf-~(K6K559~#bMij@K~B_VN&QZs*Y#+?mdXSb20Vya5BH`Y{z6vjmo-iPIPwY_-?AZCO@)6yjxaGHGxx z9Ek3X{&^cp{rSKD^iACQYbKw6uW7IRnCI5uqGIdM83kHu@p zAhk-)n3D)Q3*P0$Fc4Q@YJ9a8_-;)g2IEEzeQ?WszBlLv*QbIhBjgV|h8&@kfs^k| zlEen3*UGS*9rkV3zf;?%C;eFssdzfs|8!PIr2lu90qnLXtl_zs@6w0ca?QL*bLrY$ zv#L_%Et$yhPaE`QBuxdAjtIlltz)uCjT|f+>!OA-;oI*GcT5#|dFA$uVWz!rAm2B@ zVWWWHTf;4LDdL{&&8VOoG723*M7h_B+Mo6-WtKlSH9Qe;@6(mP0Sd5G<8gl~q+H_v zyj}nMNXtL#4wsFM&GJ2i&$Fw-!%OY51VoWYDp|GY zO$Wme!obyQt5aI6_fGU7`!d07|9Ry<9mX#~!+)c~^=Dn^FVYgzhpv*oPNIx4Kk8Lw zT+RtiKAs{?%XvJ=OV%l_hqlhuXSWJfC~?G*>L?^PA*M3lTi(XANc+){II4`}YBeeb z^>25qTt^TJm{hN%ieTThuJTSSHC+B}ytsd%hTEB%_|S_s*^@zl9&0+5zWT)`r$5{AOl%if%IY8C8~xRHQ!=db}Kr52oh$jO7uI1UMJ` zsLDSgI7BPZ`Zsm^3^8j{dE%>6~eoJ~MSWn(K-~O2FR4VOzu5${Q-bcta2*)QP4i!0TyH^u1k60y- zx)4_?o5i@D2Y`#lh(A6+vq*sQlZ)nN{$7|6t?KJBqw0ROmf|@+)YdluYMGYnE>7YP ziaapQt4|6cqEWnSeWPJ=93nj?os7x{Wz77+uO`U{vPdv}PwY45JKkc?Y5vota5g$# z)(*H@baR8}a6aXB0N2{j8-I~%{*f5~%?}brsWQ5DnSmof>;mLn@lgz3Q*oP=`M6#r%lhbfhjB(SqeWElqe#$1WcFVC#Kgd-%1L;z00a%V(NxuoV1M+?M<6i_gnTw{bZC@EZWsMUrkGYoDvzbI*HJsK2|0 z{~?0lkxbS5l(lwQ_^AIeYd7-Y$#XE_tmCFN5UHq^&t$9SeGaP2g!@Z0;W`P~DyLqVB#M)skFZ zSx{H?n?RLyXYL)lk+qjZn-?ZFa>bKYv7J{Ka31#dq(a}rbA8^DrO@ge%1yphwDI9K zRBnrnX>`Qco`Ju6fHOJ{|Dm)Cx}X!Knd9 zx%Mx%*$jY|?GHZyAXQ1}!1dy5;LO2m*yWfsk5>{Y9e*5<8DGGr zdC|3zHG87X1w>=YC8v(nqPg`?oI!2Y{j^dch}Jm%8`M{*nS4%CpM;FR-Cm<@t8D4= zmR5V1{mc)Jj(?RtOK-*7E zJ6zu_tKqqP;ttHVF*7b@k=rtw`SC~$wT%7)f@KYSDC|zK6{ICraXBw{XO`5yR7{LrXCRNies_4C_* z9NC5aXzMmEO_1*2S-sT~X5qh(a z!rhb7%VvMZF<2(l!9E|r%t?;#Ar5AS#Y>cRmg4huw@NS#ab(KwdhRoWwnq~yeU{bb z$0INMpGKZ>4fOwNtCDu3HAIopyq~!g~ zbZW5Cmgkb#<$lO1oW^!{j+Z^jho~-`?9iWS{}B`0b*tD~)4@(4K0hSjF(v8|jd(=N zLIam1;0G(}nV20LH+#uV$AdvK5*XL`DD7xh72?fy!`Y;;xIHYpcN0KZ`;X-ZxN=S# z86Och?>y+EW4XO!sjR2lR*U`vi=yzZ`N%9-(WnIkE+2(;yu=qu6w2`2>*JsK|? zo1mTT777!-I-fEMZJ(Nek|y+`gt4xQcy^oH{5SZOP7?nJ70&vOi6pwkrhN^3zM;fQ zFzCd&`HHTabb{5KKDe-&fG-k<$me@%a^_`~xm>hXMU`NM@rXg8K+^MBjfwT?asOmt z^Xf4G?ohAjkAlBP2wWPckJeCL5O3~W2IL@h$PNa%2RUqvz1Q>ai#@lE`6dAUKR*J8 zkwlrPJ2|it9xSR{BQ$@PzV+bLP4p*~ax~x^{3O$rRk_dVF2z&_YKsOZ5|FE;+@9gC z4~~cEzc8e}cDwPZzLUAlL-(J>UO`2l2x~z1X5B*LDXI7FKbCqfqII3cNB~aLNoBr= zXIz?84)+GjxX@@~>bx)k46$>p zBdvMUDFf8oeL^(F+)6fEIXj>YuO6G6Llqhzv@x+@CT}8k1|kNDVH&w&nR! ze^j%i2}V$C>YK?w=94dy*OeI5pTrY`<%t4cP8( zV2M4Y2}wJnboFQ2f5d(Wy3079>Z)sU9bX%6VZmlc_XUp?&ft4b|GR3OJ~p>)eo*uN z_*K#PXI{JUd}6zEL@YKPr$JuyQ~i`8H*hDjzJW`7k{G0NO3b~g(67StL>;d6>X=ixvpQMBdSx>}JJOWydBP*zW%D-b` z;( zKF?M2a^`G6ci8J6MIZeZEp>42%-&HP#}^07*g2sknE&9Z_|l_*R3i<*naGnI-$T6R z(^Bq%>`&eErTX(PBP-Z<5IgxjE$)(Fu8aSX3y~aKn`T+OdV@%Tz;d3W&^?nlteUn4 zx&3EqTClxMwvaTZhnmnr&X*R=XOi=;f?QNT^zg&)@h zDz&j&IR|p#LvQh1(@q`&I5G8(2iyOHsVBY2U9t6GUtBGdtoi}~=f?N)iPXkR#;Y$j zw|+9btB_aYc1i&98(@X@&JDn#?)S7O0WR$y1>E79o5>o}7rJhmR5!vqn^^C?ALTZo zxcDqWQeIM#>i$Vip?TLqr-yFw(sO(ZrqA7M-2?(2Ijfqg<{cA*8m9#F5u5>JyYeAsre)efc2-C zLxkrV+Hp-R<<9DSrh<4V6*yO8?e4cHN&6gotC`t1r8_G%?M|6}2_v9hA07Qj#bph` z5ARcN0I*Zr=?ly9ck`yAWb?`hASvg~rj z#N>g&TTbe(Xm($hq0kJk8-#iBJ(w6+7%GYUA^+%s*OVX*ODkevRA?1e^d7|W4Pe9# zynXft0Pt7ak0f^@>-u0RUL{-#Sqk-t~Vw0dnw|2RO^=O9s+(;y`2SrhPzPTo%w2+BawM7;fncG^b zt)ZVzL38zPvPB(?^C`@x&q<}N6|=A%+Nxi=TzVicBhwRZ``SJ;By@r1jPEV&OU+fH z8Rk^#AYzNC;^+#o4tP*b+05RdxnWKp`?@GWmv8T8`NHEk7)QJ$3I$rMjv}i5YC*gxR2w<$tOaGH zy(|~^4sH2vdqCb;NK_z*^_Dx|`4;_djh4I+Go_4?mXO*cRZm|$zK83xlkc46yCKv> zq$F78c1?9bX&pp{)pp%<>)ZT_JLir>{2qiWl$s2;LFIF`Hy*2_H+!7!_OO}1a?3BN z=&96^B%{&6f`PUSltEfh1KCl0OVUG6>t|4@&^NcwVs+QQ-^V7T--kixAwU>$y1q!9 zB6t*ezVp#DP=+;Bp;-RynPu)_^ul01Kbksi4>rz~83w)ruz~ai6+XmWDEnI2E7MPc zW$l`L4Y=Tx)K`fZ$R#MMmHv%4N@D>d*gJ~!oQcGu!$r!|)}8X!6`nF_uA)Y}G8F(< z7QTxc9F0@Ha@y>lUi2aBmpKOGZvh-(yl=6nLa|gKh`EZ|{QQU~JM=W~jkufEn5u(F z&R1AIjRV#w5qGM{=k*&D^>t_Y0)QEXF^aqRr5h_MRD(WI08shxhq_u++&!j|D}nn6 z0F*u7xx46qG$7C)CB)eekMW~eNXI{}&?%GXpI!1_;&m6@eJ-{89H+uwIC-+#3<}hPKGSlnP0FpkZ zSoOT3duNqUj{-LGU4-~;u_$+;xg@t)>6+FZ;4d4fc zrS5yH9VtYM`}n`1epR;WI?52Gv(L(!_jD3)_OI3fNe*eeaEVtSrDCY63l&ED(|xLE zeH5)7!F;I2^0Erb^G}%o@xU_m>%s3iv$RT!5l?%{dk|K4xP8vbvAtC18(9h}l(7Hm zOKJC3?JePRI=l}!*VaG3bZ4;7l<8RrnZ&)Vayt_1HvOACDvUS;bHL$FxQ;J`>l;9a z$@AR}{OqQ_0ZdFl$kTWz=6Y~Hq!hRSxlx~XuDjrwOh>y4=cJ0)eQ9%m(Y@phc&X7r zm~a(8#+nI0Z`kJwL+4?-D|$^Aomp{@`+yz#)Ix;ZS+i4~(rce=iyET=toM%79bcyv zR1v+?{u->FH!-UP_+8ICE7T4{u?>aTDz1v#-j}C4(yaB>9>8SUmG^!F+S%FLitgEvnVWL+k#86dAGxn>|Gz4ya7~x6yK;~PNmgt zEDJ@UmXnSmh|>@I698C?_^)>7GJzujqnmm@`B%RIu9PRM*h|I*SgNR?#S17v%j;<8 zVWO<&b`fSQ$m}f(x~!%?BdK^;dq+cj0A+@(K}JS(6KbBRsFLK1irkI6+~;6Ed-2_< z7!9JN-(D~joS)m0<$Prn{%Ro{3h>+*1tUZ*GAqh8CLP&-xVkV&6#Vs?oHJ-76JEnf5#Oj)Sb!X;~=n1xLY5 z-0l>m(8Kr6c=N*45L-mvR_W+F^GR@f4_YE?a%ojOJ7BoW8e$>T<`ayH%wUdfhxD4u z$dtFs_6NA}X!VrfZ@0+qS2)j7F_lA%`~u^!Vf9nV#;6sS^1&(6X7ygVnG&v>A3K+t2o=xvU zmN880FpFK`c5v!?36>*TJG*CbQoE=+I7v;0D)=yLCiTkOdQzUE`iH(jTx<+nlo=w5 zY}BR8!yweiD5(Oe1l0Sh(5A0Ko(`PkRg&S$97~IdP(S=C@#|+t9|!s^K09w*6?tFq zp>PTty=Fs9*FNLQjJ2!GqM56oR)$b%< z^~I}UD2>Z98nTXwcO%sDlw{(P^fH@w6Tem9>?s4N!YAB>&#glPJQW}w!S+<=R@jz_ZY#qk1tOPbuwmdxhNK-u7J zOm$0;3zke&gNj*iXRtxL#^xjJNYRcun(`aqA=9Hdb7BFqAv+ zgzAjYyO|eBT3%gyT)T}B!MB*5p|A6Nc#Bj>9eJ61Km}c-w&Y~XpDL!68bGi5ECLH7 z-qiFU9c(!3n$H}DHy?e7GC`zf$@4Wu<{(Rq+?w0Qbwg_VRX7}1%Oh~m{cf_;3W-)P zkXK<*3W(6eYTk>&e9|?zY?^B4K4r<3N#G~J6A%@M*Iky30=ydLaJ2xEY$ZJ{kK}Bs zukSS5%E48nH&-&(>_VR0+(_eajR1EivylM7-NZDg2}Ui)3@(Ce%9U+xcPm7uRwT5- z7s{*`A~TdNF`slY3gm-pcH-U($g6f}&W(cAQW9bkE&OHULl+#VA5+ov%c#ilX(mAp zQLOpj07aqAcvQ%l7=r0g<1pj8Yz$+Pl45IpXhYz(5(=tcB#l7oaK zFNi}udvgHKRw7Gzr;bR8OXa-HqzX#_gRX@WY1!VL78wh7ajL9)e4Kg8;y_ss+uUW) zk#u=G+-yqR+yP!%7LPfLzY8bwIbwXgM>7{2axYo1LkoTx&QKX~M1l1!YsZ zh<;$+$*S8T$L4aAuVrTCX(5P5GCuAJxdFbDI=NF)5^c&9`qQ{63kUCIe3Q6$W~`X3 zfJ0T7fslHk%V3Cb!?U(Vo@o@&;z7pkbEw-mf4d9tvMovM-9kwAA|Ki+Ml*(M;em1K zpi}YGux;*Lvx5thp$IjNTifoJV&Rm78D%9`7i@!vUG;P%g(%vUKt=}fZyo!E9bu_r zxbCth>{V^&GpA(ymxWMCu@}!Ukn4xRxwNQKG+g;+X6o0Rb-ST1j&qe?q}Q3BghcUP zv{jbQb*|4Uh|`5dZ#s6FPx6yp8hT)yNtLQ;0wyNRbb5Zi?xhw%kwilzwE(?Rhdj@k zQVAqXPHZW)9_cJC)15U~p;1`ct6MRCyCTijBR)S8#)leELGkSk2Z56BeR&`6Uu1LE z$qD+rrZ&V(%Z)&H?%~s3x~~dP<;E$RoE{S;$Luq(>g-2mSOw16tQaH&BWpc)iv1tN z@intKru(smX0^T3T6}#ainvI51|Fp@Uf8X@Bi-Xg>X861%*$uokm2qL{szd~eF1bf z9H*$%lq}xjd1M>L6=N9DJkt}8>n@>M_)!a3t6l3h(CNrf#unQYl>Fl3e49vwb%%cV z{3ULw#F+Z4A4n9XQlisZx8yZG*^Vp(_B*0A?Vfgp)Y?80Y2n~v9ZK9~BcEJiXT7#- zzo?^B`$}FF1||8@RYicnt43#3$b~7@@8Z1_A&@B>y4cfg0%CCbOddQg%o#7EwA^QN zjjfqOlSLQp;Ln$*F=Ng5+Ci91TiMdt5~N2?tWM5n(}Ti@vbdP?#`DQF)dn@b(lKr` z&`y5?tjS_d#i1)S48%}kZ3j+7st2G`9>akC+?Hg)H7~f)%Sncd+(?wwzttgVmX3}(j_2DhtNxafPi8s z0s$$Z2c$zNN+>EtL}{T)4ISwu5UPMgMWjiW5;`a#9aK<2K*V1xD96{Y?|tt*=dSfz z>*fzKLuU5Oo@YO0cKJLXK=B`f=$FKYdIjE$1L&I)$K~$ThXTdU)LEv}EhhC>8OLsN z;_xU)kzO_ZTz;jlDX9;7oc~y08R9~G!{&v6L&$2m)c|%iSFKOvrjePf*KlC!X@j_hPS!@JRBy`g~L{9RSuv&YGmnV_Lc-5%5oPpTs(N1vATc?Z;YTqXNU_F=bbNbzTEh> z{=qOBxHR4sC6tYrl4LWc9VxPDdBd(j#GSd-!f=%QeZeokY1roRwoV<1kuO#MH$jN zGaqz3tf4oPGoXQFhdcRz;9iL?aQP$F$3^j{?s6<>w{cU8^>+lW>FO0=ECl^)8nraV z{QFM7rV;4D#Dfr;r|Rq8O9=)hn4TR_DVH$bRstgu28u!ruS?KHAcSL!C|b{cVpGHo zF@R&Jm*6#)r}0(eFdjpb`3UgSZ-tbVb39LTd}$;e*5pZF?M>>tS4C!qxza@zy*_*I*O& z?CuCf5J#hT%Hgi=7U9>1t$j<;>q>bmpbd%vo`h5v8?Hp0sFGePKdulHhq>L)Az`NE zf#mGdF^cb(UVJ7%^GUNd_E9^x3Q8f)Ytrqoh$(sqS5#5y4^=D!308oXykZ_9{I+OT zj)5=NhEx|ulFw696gB8`a4b9-`OutgvFi9+ zFC^?Qu#V`P#D?jc&e?uS=ok|LgB-0K-WGY$DBt>cD@k?)IUI;1Dr7ZRQU*nka%52= z`6gOz2xoiB%M6^#3bZ^BaO#hO#>QDoiIlZ^#+(^1H&HpwMSP)mP-u+;32kwrq^YU+ zVOXcv;HT4Y37jBu#0h_YGG)u#P%_Ag2K7TS_@+7oz5_j`(MP(nlp;ph(;bGy zZkFTga9$-&`iHDvV6-y&mK#!89t!QxtV@W4cB;)OS2PC`=(JxOkA~$XE$_IJB}n$F05%B$@0ZIS$Y@r3QGuw z`{zpuqUl4`JkLd@AEf6{gsLwJ;C=Xt5g~b851!svFI2G@j;_j(c3QC+B@&y1r(B=B zne|nRFS=C5XV4_IF2P+72bpvnDxfPro;dIS8E|MW_#+$QRFwQ$2&y z;Xy9PL`#G2RhlLm$Il0zvk5Gn%P&YurqXJze06r{V9P@nCnbTKkL|rhRN-hakZq-pb#RD1NhjVEaI%*Jl<)hVS!20c^N!K+Wn)JJw%K`L) zY>m3Zw1-nEI9A)vuzwsWihFM|7v>czd=omUX6`8BZ-u0T-fE7TvBCRDR(d_;L=EXH z-AOHpL&~ucOs!cZL2;nYXf8azW7==Gva z13z`ihiF9JIiqAy(0g}W4eC+hsLW-F@zasnVtAp365dYjv#%x7BhHzEIIGiOltcD+$K8i zS**AI#pcltnL`tf0?Nr~O2eo;8(UUeY4)_ridFoWwQ-R@0XSpI;|8qJTwd3)qiuGp z!QvEB<8Bj`*!!5P%Ip*s9ir!p&zD=<0y?HWvecV4C_Gak6gJw-j0rB5Zjw*+#+(TZ zs_cb$Qih+SSv?4m0O1H|$y7Ti&f-Xq__m5#N{ctqgp?pa4+9!8^`BsqNF?rn93!W8 ze0ii&@6R!IVVC6_(vm7h$Wu<^J-C5IS+;F&K=sa{#}8a4IQml{z}mDo9$aEf7oj51 zQLT6CRUu!hCcoehrJ>NeJmt#yZY8gvS6v@&bB~REI3(Q=KAceCQ?RDpf>(V)ZKcd} zAE>UHJqA2rZ^y6Sc=2Lv*gBYcp}G=AUO_#3b13COL!&&aiEsa1jZ{LjKx+yqVZy)D zO=+-6-{Sq|Y`WXhlwwqUPsC?HSPC54S%xs^Lsa$@5d*GTbBG*tRtT3-B4vX-m#1l* zYfZ|*i=`eee)`eqqce2IEn!nIKiy=%Ve=b)wEvv$daJheWJ_J*yq+n0QiI7lVW_Rx zXc5Audf;MpK6S~H_OSJOTc$WFSyEb)e!gj_45SISP2vuk; zURQUW**MIGe)J>S?G1Or)9SzKiNkKWzdT?~DB9;-JDuazV+G zf^Y87(YG!dt0nvJOf6@Gw^Ys!70RPIUFccyY8p@Tp5(HssnmHEDpv>w<7!T!3yVcQ z4L>iJ3b=qVp*phPnoO4MeD`T~#_)t#hxj*2kKt4y1x2kP!H*8+pmE}bLF0w;ux#@M zC#5Gke%&*3Z^6O)!M^_A@J(!jVmrUwPtWfcv)1L3$%?)kG9d38ugF4rt~e5 zvRpd_6h|yinN(dKg&^u^P+Fj9>M;RXPN1krd)G*APFYrz-g-}nJEXz7Lux%mI8h}g z({d?9q17aSUOPrVDnUi3Ynm~c62x9^lAk1)q@|?~3&C(FDB;6J#49N7wlxG-cUwFx zxz$m6%E<_P6@T}k2pq(9ta&t7mDA=-p9mb~a++EXJ~V_uL9W!IlY&DkYhp$q#}d!f zp*Hx)6&7;ki^0HAwiUP5vf)rd8T(LVp`|Qh7J7O7K++k`hkV}47uF&D14;u(J`2V*W3lH=nYuQBc6)0bV2i?aE7kw;mwx0p5@6xl}-YPFOClW z67FsXQ$Zcue?o7Lii=GP8KZqX;dY57|6en+rS6&A?zxRZvQ0o+munmWuW!IdpZE16 z!zP};Z?SKlBbIUKXRq%}rz~BgwD3yv z@QVIfDP8LK<>d8r+plt$V^f%`tn^d^q*{g@KQS6K)3ohFkntsr5_P|H#|Ajx=bFh> zl@#!dAyBJkT?*nxmz-iGJ}7EO)jE4kRJfUTstG)xez6}hCUlGUq*AFy!9+g?5N00X zbs6kDs0mdx>B#R>W*Zz9UY{|_gpbQr1y7x3rC{S%QdRI(^fW6~^ay+z;NA3)kWPI9l;)Qd!HSWa6WCL2X4$>kz*G$Or1sh%jUy$dDSu!@cbapPI_eo!el*ze zf^kSVX1>x^;)vT2&F$t80(f+_^;wQ}7H7rgP|`s;4B8Gh*OkjLlo08Rtfql4<@YHn z%1a+DJc_qsN%o)D>A73Vc?dkh2LypFN(HMl$ue_ONqw=_6lfau8~hFG4LaJz4d8W- zM|e3SA*e@ufsQukz#<5jX{^B2Bk!xT9lQgX@l#7Lg6G)lCfdR@U=SuU8V}s^GB?E% zRQtoT$4*nefciRZYT{O|KzdL&H&<3$8Mj(ibSpL`G$|RFUj3#{7t0(pt~3d}U#4Mc zR4+~IA1w*#D!zh4f6T1*n6uB(MTx~dQl_?~BF>j&c63x*DyqD9Qt5>EI*wf|YsSSB z)!X6nZm($6y@#`n$F6w2uKo*$)lADFX6fLlqpb7J$cyry#3s!Y2T0Mr+xZ|E=1cn- zuu2-s-yK-_Cnp(+Zpc%@S4_`fGPSG;XGgEuDH?;rVa})z*vjEc4J-?Erg;ju^Ej>9 ztcw~?;RY@l(ybUZAwM@qtq&Us?r9pr{hB3CC{6A$K0cEMSJAi0`Ra|gn$gFA=nU5= zSFI~nNc-VDuyw>yMsJ*jf#`C^(*Vj0x=;P^sAE)!KxS&YGJ|HJ5j5na9Q*Blh8;X66g+yFa zRck;(@K6s+$IuLTClQx)T;`hM3#ZFXEuX0DZ}c0Ga(IhLTG*@inGLo~QFt7ZY4OqI z?1NlfR|t&oCZD@ho>7N4pU9w^h8Bca*DsssyFXH|M+X@{{x*AqDW?2+tQLo8!3ox%O~0mZ}s;d$?Vi8o%| zLj;-tID^6|#^{ve;HbehZ-jBtmj<`tFlUD7%smGMqUi2Vwx~sPT0nnZ`sgf zYfOipQfNm5*Mk}{xjuYj+NLyEX3?Gipxpdt~4T6q&hth-^0VS-xPU}4x| z=1ZiR=3oAE$NEkfUTUd4(a^hWOjbd6a`1^>YyO zsbCH)23^R@8e%=W5(f0|kVa_D3-FQ`8x*VAR4mL&v=2X$u1pNOZw+)fJ&;UIbD9?# zyA@3xYNeEePn&`6BxsW2%TX9r^$4--8l*BQlW^ zYwP#r@yy~xQuc7sF*pobme631@I|%*$V`4t%3S`jB2VG#B3RPe5TBDdbLRkG=bC(Z zTSnd7v+_>y83RQW5N_$>t5a;8NGq z+Y-(W+B*ygDUIt~i}m^F45EJEL|mbdb}J>1(fuhoOuYY#AWr@W!ucSdf{IoG{CbRe z&{NdLOHe ziug|y#?#%k)TWUJ*3Z2`+UZLgfyIXyjOO`9Sxpk_4U*DLqmTsgSdT_ahK+7(R75ix!jXAX)_9%g@g|RD6_FhRWrXt%# zkOtB4>^3MSv-dLpqZkWGaA)5&kT0b_6uL||*56-q`*3cG8n#&E)|V;1Cs==_#GLM& z{dG;2mhYwBS8og(IQ%&>a=&+l0QT>hfF8JKX8J-zMq{US8`Ftn!N;O&0e~3K(T`6A z8K-?WP3!{OqKCKB=${coJ}EI@kM0rK;?S#w>JRsKh2&#Cb%or26OmEzCc8Ht*LYt@ z!OvMP(2!9ioPpwvL7|1?*`^#*Nkwm!*G{92%2}joFYW{_3EED>QkV?Tn)=!a?THSB z*ur4ACs_!Sat_%{OgT1L z^F9+5%zes&ak}m!t_ZZ=Xs@L3{D-cEE|%%%hI*BJ22}%VM%!&y=Y^{c+L!}KNih#z z4iU@0e*h^VFx)T4N3%f{UgH3F!{Kn)=VrohL^NzcE(9R)aU4y^%gPSxU>;q zfs?*=f2)M)<>HP4fg82`?s=?= z!N>^}^CZq3;p{NGtJWE%V#Z9WU<>2NWkdOdVxQa=!X#E+4S(rsqnyQAJV-CCIqc3M zw$dV952RqY2UOvXGM`C=vZ0o9#PZyGq>!ZbUtAB~Q}$Yy#c}J!74)*Ma%o)T?BTmE zn54Z<#5x~jBaLxir0dWUUbo~;*oZ;-u$3AG&xiP@yf;^fq)~f8^sG`XitAp2Y~L?? z8jF;@cDz^E`Dkuqo;krX>b1(c!z)=K5@lzuhX8k;*2YU$(fFIBKLL=YeE;+7j7STy z+-P5rmOuu)e5n7B;&@E@a~_SwdV;|MfHEu+h?q`Qzmdtp% zRi#(KS`bB3JL>?ePANX4>p$r;$cqsdjQvTGNN1H8n z7M61z)vE;O10&{3-3yQQ#4%T@+#Hmj;?KG}gL(yceYrJ&1yK@+K_oel#VORM@ zJ&(6ka1kmg>nad&;d&nBP_Yx zL)J^($MUDESZ=>-ykS!>?#<0~WU9=MI_{7Ryf`_y!7_T@T2~EM5GRk;KM$4YSHop; z^_9Kc${DDGkzOB6MiUyKlFV`AtuHQ#aDez@Zv>+p%V*gOMhpFp>*4hgD(PFvGo~Va zqRC1@F-MQi)MlzwWjt5%e=_?RTsDsv%4w@`L!PydJ1QIS&LlyfyQczzOqmG@*KlTH zm2Go?7~x&xKhpS#zvi7beyCcM{4z+Rl=h>K`oQ8zHx@%~uB$K3UF8IZvhWucZgVvx zc{A7)!VPXLgeKQb!*%6Q_7E`o8ragv$60Jkl}eC zT^ciFaxAu*clO*}_o=+8j}mNVMyB%)ONxpx`YvzFl*|c=n;MhjN0XbMpU>-P(AVsQ z6*)|1;tB~!llQV zwc9r@{9H$xqko^2VSxy+aon}M$9_^0BwlH?X{;ucx6sJFz8Ic~mpbTD*pjAmNv*{RVsCTj7p=!Q6D zZi`t6aF#o8#G8wc&rs#wWP8d)iXEAE|cF-b6Xfmg9f|+#FH!K=w5S^Gdg=UNP;Dbgi>|FP(t= z!O%}(=0r=mA>A9sc^9=slY_Cu^@_4C?w4J|V=%?uD?{*0%a<1fTxs;JOFoghQk0W! zI+Mb%LI)ky>VCvA81e}Xqr9YutUQaSGl-PSk*=Gq0paW#zc4*MVA0posdcqdbD3k& z&TYZJNhl>KVZJ)}sN-SYQoODwp`}+gMq8SgMB z)c=LwjeHCzv$l3kEHlmNXqJPfQHO6T8ReDB}`N+GyDxWzUbir@N2!tRQejo%qX+gG93}|iHtYb z8U{l!u?qsyN*>M}+1tIBJJk4rs)FCDnODi#)19V-5#V7ZG%m@ zKGgj^=BXE3a;hXRBSS4L9O75fu>IPwy(5Ulx=4*}$smfk2-;mL9_uXJ0Or?0bMvjK zgnN(nzKZfQ|T7%15kJJP8uN_Cn_n=o!xHuMShxo};+>HZ^yxykvynaIYsK2Oavt^ubG{lmb&QBUut(&KdlL{4Sk8-#y za^}1u5x3^k-HD*6##(jX?{-1l%tJ4}hZ7p=o~4l1#$q;%NRv4RiLk@|?Oqq@JbNB~ zkcYLaOg-aCsAJZL7Kt)_227q>OAGDl+iZOVX)X8&+?ifdHZdtuYiY`dskadm>1-yN zLT0nZl2g<23x>RPYp2wBx=O$bp4tVR)ENr63v)VUOJ&vKAX_kfbAfePg}y41=uC8U ziE7_ZjjX!OTXAD+)IcMEf2b`R;^y&K5Me~UPADcfdA34?%d}!!vt#jMyiySgiuuvY zNUCIVEEbzWpb74eglTwKAgBT2Uc8)mNjj*xNDZxtR*y=ZGQpFV0 zlN}9WE=|->$aXrqxkk}NI4xBv3M878jPu(EO=EB9)Qhlps8vvvN0+=$nkE*(1YCt` zF>>=#BWDssCE-^)S+LKE6JwX<#7veNm!LRr; zfz3K%Ws&o-&Q*NO!4YPZ$!1QE4$|?@2Pz^}YsA=anUWb0E%=oz{Z9y3DDU-7Y{Uu_ zISpev=<3OQ@@UUp*4*685i^Qn@`UCN--G7J_XQg`r962tfwk;2YUeE{D%Q=4rd@kA z3Jk^V;lyj|7U{SdSN+Cx|M~k%!Q=Xt%oh)q8`R_-GjCRf`izm|k}!7y-V9-nbIp*? z<7Olq9}w^id3^kL-SnVdS8O4A>oBwxw9K(f*HEPiY;q-py&rd_@){9ytIwqQM2wE{ zrvc>X2P@xG$}KzmPKdkGHhTPEKS{v|r9u_eDgtPZ0AkBo6~yPsm4`k2@G;PgWAGWZ zjDRk|Up0Hmt`B%yA<1&G+L)ec%&||=*sdmmeb%keA1}jSk&kk;WMs;C=4zZl^#*Bc zx0Cx;vvG*LzG5(;4y{_ORA4L3|0qmprtS{ih!*n_u{X1qZ^BN?X8ijE|jTyS79j?yx?zB?HI~9SMYQ#0y8F{g}uC; zHW4NhXbHWW(=KA(gyzTfd4hxUN2dlCamD2A<(iHcgG-rG`m0;>Ndsily6c#mhuZ{C zCzCp=ZtB7HV$Q*yfwb|vS~qIo{H-trv#&QTV%4c3(Qg-7QHY^ag_yjrKfEy zCtYcKMdAaW)i6bMy_kW}C{(oez0dHOa;4BIiD15|=kh)iQevEXF)nPeh7VJUw%+0P zr6A4e(;b2&)*(irU2JZq~* zllrJMf>?PLIb)un*l-BPqzbNZ|31bq-BkUeqbOyw6BxG(`##JZxhnsG!?0r3eFnL zI~Nho%^)v=$YiIVTNm=v(${EXMLoAz^4AwRB9f|yLEM4sONif~7hTR&RP(Ven<>Iy zgCg?8PGa2pRwl>LoPzDvP^?I}Rr?zFl!#Tu=l~nr%8+k{Qr@jm(dr1U(jB2tqfnuq zo_34#akq*HWk7^mF~y>(*d>8X5mnqQ*qYQy6|{@I`bWV#WS5OcqCULBOwq73B?TJ2 zUM&x~Qh?KXoL9`6P>1m=T?KuzI528xquS_?j|+@IUo#}lfxCKAJ)=KI`TVSYFrb@HRfMd{;cunPRBJAO!cpmMa!< zn25kY%IK7`TN;vrE~PlO%FD7$-!T15!v%|SF*Ihhs+*d7s@x&}AWpF6P5Vv||jZMDac0vlrPTe^oUw{!|V#_KNWGDV2 z+k8`>J(0X$xk0y!sFxd<=?Hz|UnKlNo4TC8Vz!f0x@HVQzvd23 zPMN3=Ne|Uwg~;DAo_p{T*yPPGa`6$=FbG#^{dy>GgU0DTZb(YA^GSGm9Di9Cr%lx5 zOXfs#g}7Lc!bVRYTdu-cL#deaQ*B(jIWDr&%16y1QmbJi?8hoGUPjNFT;18D=t!ZD zz40QSwD)ME+)X17J?;w$@gajHqsc6N1fi4=`{x5X`hx-Wggh#}7(Nl+W!sr)Yx}M< zg@sZg@ZbSH=VCP>YxjY1(UE&86@_8rM>Q`&jqcmoQ}C<#*k;yu8W)WTJoz*&FWn^F zymLufwKM`-kq8_;3=0sB4qj@WOQTGQRIDwZ00K9+`lnYK#D*#&H)s@F}BFjfPMeQ8vET?=gaqj zU$KQ3Hb>tyj;bxmXF}xO0PjrAvAm`bS^#?cv`P8j^^X;=-938`KCSE79AteRvo!WWNU5y+IwGl1}^ z4^8y{A!feeME@UR=KmpPzM@0_|ALs=qu2je-5=??|M()9y>MU6z`!>RA2`Qn2EcnY~YV`3;~y;Ke>~9L zdi)I?DxX%ccPV;Z@@p0uO912WZY@X=#U7s12-1aX!DkEk0z`sTkI3>qX~hG82DQ5- zfB*ic7xZ5KrToLcGSq&T0mbZV1{4=^R&pLv;->GPJ@n9wh&@=*tb7;OL{wA5evAE9 zyIs}0=lmNakRN4#-Pl7?LrL!$8-l+UEbUnXFedLFzW?VC5bVKE#G1CHU4h0v62OD2^AkXRNOQl!J zL(dkAJlD~cc=U5yQm6gw*4AAB^R2&M1R!a$7o>Jul;#Nw?Yqz$kuqaRVIL|7EH46R zi+4|9*n==kE&RgL-X;2efxe?~fZsshe=xIeJl}mcA~yOM3vjV@I);%n-kbWm_RVsAM3oeZ}``iwC``|P+wX7y=2h<$>FmeO#rcq9~S{o{8Y9J zmBdLH@#AXWBV7_QkOTqujWhZk1pEg}?H`)9^YdoDZk_n`{qFdz>#smH z3vbC8_?#TMwOjP*L0%5|l1uXGLl!S>>2H`?rsd{4-T1lfoR@H8ZS4#IviAyKw*PrT z%-^oC?*~8hecCQmR0X5^`V;#Xh-hT|7oKPXX-e|<58)Q#94dE@o9ihdBy<|~}EBCLz4g=UCf1X79Q`v%?usVCvHzVs_|l*XGwRqxWQnh==0r?Vb72nJEeC5j;YdZPg zJ$fb!yEOdn=k}PU@Ryts0MzaoyWeA|J#H@I@Lrg<*o%>0Q||l>Fg?olAbqt-aoXe7 z{NKfC{|-rM_wK)a@1~6S#t!{8WyH(9DIoLuKIpT z$Yq`r{|krw#Boz1mwDgTH)GB6K$)L*6H$bxE4R7`C2Wyy@e34R=-CXztM>5&26!B z?JOIX(d65e$zpu+~)%JGc4!nB* z)u|UsSDIBphdu+?PLcX@AAae}CDWqQ(Oy2hLXK!BeMx|0_x-lD2YaRenX2qXnX8bh-{W3G zN-BUiy{sIC9KQu;P105%1gIl$zJ;hcsCK zw-kQYCY=9i$@AXp`>6$?XiO5#EruLphr-iN+I~Z{tC0}x1aP|ktM{bf>YMTdzSTzf zyDiw=B_oAglBBy9>ggMp*-x4^wTIh26wg5#qBG9$x+2mERwoI=>1zswUBeYU{eM)E z-D3O*dwG7$G#|$5z1^$0%kDmRQlkrozk0u~_2JLhG4K4FAw?7YAjkBN{$G+qklGe3e$%|VXIInb zjdr*G&vfONd=f)OmBhSNcFrwPhk!Ss!#7~*+w`(5d{ghR*WWk3sJ#foIW~4C zYNmum5<8B*&2oRg51rn^(AblYt4>cogx}g%yp|4q%k7&U{+_aYE_Led`;{8Y%K(Q@ zw%sxx?pg3^v4}}C5j5Da2{NaF*q)W$<|G65Pu_&JP z{I$FlQy=oFs(0R)Jerx)Kb}S??XTxr&L`5f++ma8ILpcPtipMK{iT~JOLIx>*D{K^ zj%Ut4BM;E^S`X5f8BT{K8S19AsGEyV_QycVqt+l8*NiacMRrr3>k?=sX16`csWk*la zT%D*#g2Sv1ZyM2gb zTwE4UMBw~5jK>`pGVTLS9z`*1Hv~uSOg)y|-m7r8eR7_Ygt7QrVPu-nU;r z{Ws>mu(FQ&;%9)*XMjk2=G3dhS~+7v7jxwlj-;lPOwUY~;wmeaA_ne_)V4og6QKeI z{?=Zfi2v=s|7T|Wr(XXj4{=j$kcqNWRWUXRlN2e$7)r^{6H^*jjLeCn6RThnzQmW^ z9Iu9VfBH@m;cLp>uGtl@_T-z7Zz$m&&XbFMOYwV&{n2u!q>%0G`-3QW5acd@FbHfQ zP=N0FKb+4d(Acy7C(CThx#mLst+D(zfN9WYGpiN6^9S#wXzal8tu_C24hNgl)bLq< zwU{+Qzv}!GO(S0*f`)UDC3vGatD8D)t{F{bJ!4U7iuN+?T*k)kb=xFsq4RsKCI1hB z+J7(I{{{cwc4sbi=|j+toMgL$uY%zh!(K$#c}T)AJt$Qv6%jj}fFAz6(FR>Vw~qZ1 zfEE8X0OO0I?`LfB`R2oa@8O}wOaJ8=?y=j!Mau{OC}x&xevYcuG?)LNmNXFY(8#Yr z+{@p_qEyXTuOrKUIzrHY>`#^nv;1G$QKU6H$B=xs2FEntafPTvr2=XG!z(A3`uZ`- z$gVWY8}ux_Tcq%Jp!ctCr@_u&X*p|tVry_so9@R|59@B%_?LZ;)9HD*Knbp2f|}oYYMyWQ_4hw=p$?!x ze~h;k1lGT~;or-O?$V#52rd+Iq?WFK@DEO)UYRY%$McV#U>o5x&{X~tA@%1`i6aiU}q3Lvq88kLjc{Wd!4z*v+N#nJEAu>UW$|qvN9$P zTd;}$UXcH$8+f#IN^x-LwQpqj`;*8bm3KbyRJYeUly#|zu3AN*EaJ-5q|g*$a^;+j zZ}3+$K)oB}EB{cYFYkKrt^fR%B=p~eek5CN+VD;Pe*d<3?hh`Ria`Yq{gW)%MnUr> ze>4w5`F{#M*{Z4Zl``D^bO})Ae{}@Zi8L*D;LPppTmx*Gnb|7A`Piy5{&TXpPoyB! zVa&EW4i1kYxZ(3&Ijo}hI)Mwh|5v=e!uXcJeF=94jM8Sf|Gpf(GGo62t&GOPOCxd{ zcKE<@8Y)w~qA`wh?B27V!4Cf=b-1>y<;s7Ej*cUv{-a&sTpfPO9O~for{`!^tMW%l zU)JaBt^0#&H^4RaeG3|X@2)s|)(=;~yZ%r@BP;w*oesefifYk+h#4C9e-9TpqhBW| zsG0fF%!s{9U_aj4rVxkmlRR{QUV>T5F)b|I(kM7MIB_`Y#L#&fF&T9os^1I!K*fI- z*|+v)U+$eY=Hri?gVkJt2G16yQ{bQPnc1^{qMHh@56Wb-H^vug4`%WAF)n;uexqS$ zZhRY#E!*H|pNG zolRBg$QGY(P1d$K3&VvJB%~dHR^9^JZ3GFeHuK3t0_V$7bBUd|o1MI(XiNj;A@bM< zQ|TbxQRh+)wl?Wx!&`lBhAI&?D6R^N#qJ6^DxLbmQaWt1Wy*(hEnG)i756h~Q*pi& zqhIMxqI!qn0J(C8Hf?9c7fNq0#zR|g#`P$wG zuBR-#&x=6#=ntB5xzS%x?6k1utEpIZ8vH8@w5gak4mqqcN12B0c$s4jT;zK)gIaRx zqY3Z;(w@DDna7$Is7>Ikcx}17I8z!_5+5CIyooCQWGA3nLs1{+4P!@xXxAID7wrP5 zOe%HNW@aU|hZnFDVx3AAbV3D2`E)N-MAJrXsV(Ak^6(N>JRzl*%@!V0IpRj<6&3rF zoazjD^Jg6o&AEs7t|3zyU9-<*TZhx3*WhAQm?r_lPgGNn&IR`uq~sG}f^ft+%M}6C zh2z~^xHTRug`Qy*>ZKayF?1M5prSAXOefP!@@Mqg#pvQ3bEI#w$E8m|=UR?~P5L_Z zN34sTY&(!P4*b;$3v_ZEN|(39q28l5a~CmrWE<|>WQaEWaald11Mv$uj_g9Nr+$Y_ zX;@1MRCpD^vVC3m&U_m--1FDGG?0_DxOVL8f)w!fYkL{_RC7L}7ZS2pLxoHRT& zI5ix21XK;p9XHZpQ@UO5PM(}yn5Eacra9?Byb?ICphsEKNuIO}a-XL#E4?*<*B~7*x&d=uFP4VHW%eaI57G zi30^Oa?JOY*e&=x*eafPnhiJgMIX&~pO6%T7W5J@Cb5JX%2-DdO&Z5GkIeIYc!aeb z$+^0@wvw_IFezG2$o0%nDxc=;;Z_bkuBO-39bx?X;ip?pAG22k&(ymC&vqQjqF3Uu zueuOY2sN2%heLZ(cnn2*3*+uRO-+yJ7m4l&1@bGFaVCTm!NL6Z#}h|gF5JBvm4t~$ z3rG#VakDbE$pnGyMzTlpu-b5G4pl`}M2u*Zxf@T(7yE=eE{=^;2uh;Mwj}T1&GoFBf-{g zo}Vu0>ICFlpJ(YQ9Pifi^Vym)kEkBWTX{41hZJ0`m~7X&`Mgo%x} zXb7cSv~tificN9Ezw$POhx5jS2@dvj#cB2G2z0_<^@Lxc>(hMP4@SqgXbHEM_hIEv zSu|DVDB;+(uIoVHPU8wgwHqk4BNI1({ITMJ%_pK*u-jLv`cca`d~LpMX{FI}pQ22b*PKwes@0@zNb6k*m-iuQb#Yr#xL)WxQbrfzs?;9r@Z4J-|_GDX4j^p?% z^vqY_T#Sq%Q5C-8SrW4njtBr*D@`=X<7aT{Nx}gKq^}3 zxr*5PbriqMq9qmeRt3&}eBT|>s#wsBOkBNIzMJ)DKyZ7q;dK!@a}gPCTQ@}*%h*9Q zM)k3FK+KITJd4Y_n5$+-B-pj+wCLtSi^Eo|xDs#kK5#LgP))&Wm^Y(kL>oFnX0P(m zQMDr-)tC)%F38W?#Q#*}<+A8S34SlVyP2U)n#*35G00Xz)~k3fpqIUbSYc zbemZb#@z6a4*CSYs|a>Th?ih#iKC>`;3wb|QzH8GIOALMF;M8^rD=o&(eQ!Xt=JSg z)}Xw2+E%qzj{7Dr>evjm+?R~QPTk2rsj6`{he9)FLOG<=g00#F_lN-@0YjHign02J z>OIxE*YV05!YQVXraC!S*8lKIDNCPT)_o+(enPMvVh~lV#R*y+ETGp$0-=G^U6tde z2Nk>QjWcYmINs?S1W3)yc93vkBR+!Yl#nzhP|f~Vu3?45qKorb@n@1*P&ANQ=FRBG zs`cT6kvC5rG>Z?*Q+&F@IWNv@K(qFsW3DKGs>tn_vmRCLt>K_{>sLM5 zH&sBi=GC2Bqwv3udRMlj-x%`25Qr{Zi`82MKI&2FE@{tJwAE$jH>eyACoQxjX~ICs zX{xg7@KPu3v$-0=WlC}WOMS^n)&t>gS;md-I+NF!tc#C`;Z8;#Y3^(mTU0dQCzQ(T zFjqez@Lsv>2w@Ktz8?-$q87PpE-CBrczvrAJ^`;zc9Ve6(dFl3trcPRwTFshyxmC( z<4#hTiLVM{mzu(GRfs~cQ6k5f^R;V={W1;rNRGMAJP8u616l~1gDq*5I^)%@vm~7b zQicYvIxx?Y>fyuZ1h_83iB#}|CKnK2aWER{@l8V9i^;etaC1;z6Y_XR&ER(du#EN| z-k#Xaje>*%3T`Ep8%26J8M#&hcz9UR6Y0R8sNh7FO1D&}YY@CSxYS2t2uZE6UG=eS zGm6JVk7)*FPn$~|eMD|2oN)nZ2qgM{WOLc)+hLo1A09aE&>d>6!DI*W>3EP#WpPn; zNp($w|6QAU87=dArVVGR&&9^#TU9KF0QN4UJlHEgHSJsSk zo2y2$kIU(yGKBoO&S=?3BXo;lY;F+U8b@(>neB)wZV72W2|l?IBNQF2rk7anme*P; z_L4sg`4~Uq!TvtxJd%sN9vy~Z;JEFMFfX7NL=CyI+1f00mKkjyLYgmIXAs0~EAD21 zO=qj!z*OS6WxM?vE%aJerLVd8Mud9Y8R1GDYJBg@;qH)vikil*E6Qh`U67sAGW7kt zul0IaBHNAaOy-6==Dfn@y&dz;SqbV=YgDm=2argS?SLam_ppBuD%MFfD-LgEZH8l`K;^Y=H zc}kfzxkK;ri*{C};KJD3pU8z*d-DRf2K@Mgn#hon?s9AnT2y!|2R5$af}=;ae{Jai zKtC!d+-`#{qyktlIxFjHJv}jxW*j}7Vg!f9={#7^(8qnE4|X{ z=bB^vip%?AqMjMrG-7)<%qW^$ddJ)YH|B@(7Cm#XEd~2<>50so$glH@HQ{ekRn&zh z#W^GMsK*T62pn9bI4X5Dk&+461kUfTz*<3JD(G5E5%)vNR1-t3;v^!|9g$;37>&l4 zbyMTI#{1?o;n@!2SkD{@87`|C!*mLEh6^z}_d!>aa^Yp68nvsj52>Ath5P0q^bZ$h z-P!Q>v#!h%1AMAA^2Erv&eZ8>>Q|}?DFygkT#GBNB12I5x?azY2@(_$0@9_1u7ZFx6%`aw^yPc^{qB3?e(#Ko zu}}WVS|@wYv-jC^ojJ>uCeh1$hMf?>$3!m0tZ>)w` zdtg;t07Gffa-gMg>JNxY*h+7F6{J$%;-&N~=pk~v+V{HUtGF13NZ~-o3s@+%?Lpzo z++2-4ABE^jajZ$QJda9DqpLft-AS#7MDp14B2^-EvTLhL9xyV58(xZd%Pa1B!Sh0mo!efIIac+mclK&@qWvfzYg>3T%| zaL@V_%KCK4S&oz-B~kmA4w@4;=;C?l)*1VLCIyaTwrmk@q~{c%qUcS#Z8Qu6A;HT; zWi6}?&6=K+AaHl6HqqtEIUR>uVaPsAgv>C=$>IfAXp2_pTFm|WYcBehJhtUVW6+yd zeSw!%>Rw%jQFZ(sS)GSMp*QlS&+qLO2v$1;@?tG1vf@(wO9yByVat@Pin?T;uDO@I z#H}@bS)%(IRve+Hh zgSY>1&c?#nk|HNhjj?w?9tzv%{tFO>gtzvJG(U>C#1GZ%I>HM|mbMCUU9a##rBOlf zscGPs#WFSfUz{P4SXj^&2K{MUy|FtF+Z)rzg52y%CH4EIGX$KE%nL@j6kn3VEqZma@K; zWCX1X2PZVn>|q;uqjhAFhNiuax7`YZFzv=R0?JEWUbpoA*aV(&w%(#f>RMaUCR%m`2}Y_RthMK{_Jp(wr+ z)4#I4J1;wh%9kTfL}XwRIROXrmRtd#?FYfk_Pmg$e8Kh4iGD zAAq6JYcI**q%^ptl?~u(7j6_OOWJ+G-tKroWq-_N?MAOJZc$NPY25SVK*#nDP`(g8 zuETXRB_KxsOkX`r3eWulE};{B!-HD%HlwFDtG^?tFT9{OA$-)prf41ZXyTi_M>+nJ z)hBI;;e5Zd!K2i5pLOEooeT5+cp<*38N#cO$r9l}(WiD^mM>AKFDwk$Eu?#kVZ^4# zlafxF<~}3N_}*|0zpGc(k`Irqy55IWcHvzVt0PdTfsU?9fpi6TdGnGoq^{nH)7kct z1coJcIorbr!wxHS^tS}1>9{#E`^{^?y_L9cFo56uJ2#{gBp~ELbu4*NRE326|&u zniUF-u4zaX11=yCV#hIKe@3l^9RgwQI?ZDaP5kHeaqk_rl_mo6@*{*ez60l6Rl6JM zY*15ipLPMGtzG^2YE-p!4c{xD{K4MR9{^yC{n=j+@}C7g-p-Qyni*fhhS1xfgu)6Y z$WcVkkZ5D$=2)YI{^mM*l|9~^pMk91h0i=Bf9j@9v zIJvDEg$U+|H_%s*_m!zOvv+0)0Y@nz-aoU<$3!a@=k*-gm{eIHwJKhvau0m~k_Bh~ zGVyl87bs$X))t-(<7Y}i+CS9IOwF>gos1Ccj%dbC1xOeM4~3Ul71!BZ(eyy@WBbs! zT{>=>!+?~hLV<2gYzs<=o8;%;A34nSWRu4{>9iGU^&Nm;>T_N{5KS3Xv+yuv=!605wg>AGr){_#O&^_6R#H&BHtjw7j^w6(CFjF9f%^ry=_?2!ZE~RV_q> zEBPZWW^~Wk`vwU*e|(^%G83`%VuDwPunhsJRJ;pVGR@fQk|``Hh-b+ZbQhC=xNRnv zrXGY;d9|&M=$iF8`b`F*m!(|BOy?WmL^n)n3Z~FZV86y*la#hXGew$p9Gf4 zo7N0Fpw8)>sBhCoDd;*N$`9(XKb%Yp%Lpq3xqK|FKr3HtR4vF86?k{M@c8~Fr@M}e z1&0(2T@ItcVnaoGwJ~YHxZX-a_34zgz{`k2kEA4Gl~?SbjvPm%kg+ROLa1sc#$LPh zeYS*mLR0m|1>&mMBOf^O2A_a-V3(Y2!@&{BTHoR23ajtVl)7|Qz>`Xb1KT0_g2a04 zJp~(la^7AeD7U3$USWsxbgJiQdVb|G6<_5E^LiMls91pfdc03||VW zXq8{>&1BEM4heP+gG6}Dbh@RI1^hc`I}2BzK~RH{V#5*M2zo~(@>XLGr&v8{7PKC`%E^gA#_ z21Yy?1>U1w;>?D-Ey=66j_@ZdEn5b3FJ6L>3EA_L|G2_(0BZqagq$@WjtH7OoIihd zV<)0v6Af)JMgu=Gl*EKVpcu!wFB!76D{=ZD7XocF{G8@@|w;lu1~Xuit+QBzSN1P$1=HtUS66aD?jhowDZ-Wf5{me#q3Gzt<|<^H&v9hxx~ z(Y9RM87UapocJ|D3J)nuR@CJ|iba|LfExq?Fu-_EOlK|QwP1{b%;MPBv7rs&;{J7 zgw|wPuO*CKB@8f3sCJnfbuy%?<~vvbbpA5Yh_y(e46!O5BX+a#ns2gl&vNCl#M{Eh zZnJlXtAB@wo~>A62PY%A3N4L_(wedEL+egz!wsd=b0ti5A6k<0ONIGPywUE3I%p3x zny8rj1a=5U`)Ld%Jv)v6Rp=pJKQOzA&QC$1Qk8VM0Y=fdRZkAJN7l>imu+(gt=(H2 zkyM}bM{0fUYOx)vp~}FwKAEo~BCD5b%<<24X330)5oiCkmJRS*Sjh2JnvdW4x%M;f zf2yd*;~lpSzd=7H(M=!!xa_eW{n+ugJ>RIt^H`ttE59FE9h~{g^rK&2VjBNdzb@e0 zjQLY&3mTdpXV&bNL}yIl2exl*s;A!RyjuUJn2y<0&9qNGqpw>kVz^?uxL3{qrky{mCCw8NYxKyln`&GM5+2F8J=JMA4@ zow{rOA710XaK!xegKq=QAl(y~UsrUKi4NS8E|%h=R?AarJGZfane4PnE7#)#X4ZQ} z462Yu>swtJc)H9ZIr+fwHR!k281jW0-rX{O!y=Yem*I zp+?5D5!U3MVdTgV<#pPhq-Evq5?&BMJoM^KY(-3B#58wcZgR7n`^2}oP}ms%i+^J)w8?`YAPX-t=lHQYFpmP;Ja{A`%w3M-IUWaMV}| z`wa71;5-r&wsInW9Q&)Jr3XEE`|FdjhrayiBl%)j2?=?+$WjREYSz$eJQ`$=7)ZfP zfx>(iFz`)88F$vaFD_eKo}$S7TBJozuYcP@|NOvQwvY6w8myKDpjo2v1)o~TA&hA= zrn!VYBk91TIf?KAIM2(Ez?;ozwx(*BoGYqE4LxE*b5R%M_IqLWkqcWmF?z%Bn?I}p z%fe3SW`k}FH?-#?sTaRn0NRbR8{HW&r=|oJU+J4uR&dd+wshQp1xAeLpf5eEH%JzQ z9YMG7ykdA1%$9}ur^%29w8?Sp{t#p%RV`;23JC$6qj~JsC2EyCMm*420(@mr4}VzJyQq-X_KBh zR%uq{iPb2QUMT*&95^!4?-36MbJI=a|Fj}H`MYlJCe}t$6#6ui)f1Y2r+u~@?cU6F zfHnajv6SsHZo#^ac1jsB!fNn|t{bO`xj9!c+}oOFIcj4d0f~dX`^%(m zB1|jHXANJ@0X2(kR*F-LDO6&e6aEAUnZql>c+;K?<_0b zF%*Ua_#NkJxgLvsCqtV|Zjs%Mor$Ow9o8?odHK1I#!>4P!@bl7l!B~l)FSH2zsBk* zeINO)Ys+8=aSW&l_c;faqGqc!&D`^W>}Mn^|0bjUNai1=y=>?uwl$b%GdfU~8Y5EZ8omR~8_WQ?aay|ElEBb?6}=UI zTfJX_nD$XPps$(GmYJe*>LW{GK-305JWy2qS2WCE_fo&tP5jg);) zGd2%^lgdTk&hrQE0)Ll!anP^OMuvp+k=0nRv~3nzHJ>+hX4KuvI)FTkvv_mTmkjM6 zPSMC<)E|`gJri}zPZ>~INXC_=goiAm$E}{idFyC{I4{Qhqp6yk`ul=f?6da)ZN(iw zn%xGS)zV0zWE0BwKZ&X@$UUHj_nUuTE=D=Hsh>m1t$Hr4GbM= z-obiK8Ex$Of8C&YZZCT>eRK=ixy>+8Y@{WAPHAB*lu}JIr@2%w zEAi_L>s~BGDgR~S^Tjpx*W8S#;YL-bSw+}CNwmYyAP945(Xoa|>L==tA?hKB86DjG zu(U@65q*h9dm~k2Wm4qYpY#(B{_M#?@Yui=P(AdDLJXQ8ZfnHKUq0rOzf1`;`O8$S zzfPV4#l%>9IH*TxwrsI_JMT4i0btZ z1rzd^bvRZQGYjIq7jHjUg{+)xnrJ`*72>TWO!X$~4qKIrujSC9wyi9T*{&g)Drte( z4zZ4T(ip@ENXmx~BpSCmeP`X`>pvs58Vl#EG$z6oT-iAJx5?UFvYm{I63waXf`7Jb z2wqeOE7>3y( z5E5-&R=}o;okTsw1*>G5K2Tz`_8r+PkS8)fe5J~bXw;7xI}sKJUW%JcOV(=Z2bvY+3m|3 zhaid8mj?eT?Zf1aCMEmMJAl>eJO10Z;J*>@VU&IrVByq6jv9R?q8)KUyQA7M+FJnETr5gNzV;G73-^Sv}#yy>F7G1zo9)?9Bk@Z)O%2=GQi%=yCZ< z;9E^jh7HY3i~p6IgqjKiUs=r)ihU{H89QvmCqJkI(tZp=~_>)x_r(ihZ6qym4fLm<0ZLchT!mjM6Ft$egQqQQjVi~AU& z>+ZMm;A_TjX~RVG>b0&Gy3VkW^1Ps@=<&Bg#!k!r+{w51!}FC-dCh6Zx(b6{AXJvQSQfN4LY%FjcTd z@4h+0SQFaHfpAHpW6kAftG>(e{@ZU5CBD&mQYf0-6cp3SF7{~0wg-gM_PXV_Kxw8C6Vdh0le?6YfMR9T?y zHfGP2be1;V1&bS7fY@d2Mf-v);D8vuaaT43*ToCPmGueJL9VpqisS!=%70E06<%;i zd^Wo6FHWp2)_cxPK1DOLL4Qo5MTEQxy88i=@8Q;R1C&yt>Rtf}<~TMOW{+r%)v^yw z?|KE;_MGFi2T7+b(HOyUd7q@kGpgu1qGavW2=4hLTbvr?Bh0C)83A zN2B*7kwZ3TDj`I;Y?Vw+QuvhR`zzjv^0!b?Ox{gPGb^hWek@f$@v_KrsH5fl#W4tB zHfXLyy!9+{R~dmHx)!zKuwrSF$elhnsxs5tOJn=>xM}^S@?2R_uZUq>m-Kx?4pe7C zP#v%S`;yupRZPaC$Neq*&ixF_$WfD;$B;b%6c=UY`Zd&PuqIJcK-)wkEm2Txn!#y~ z?>o0@2Bp;82w^?|6I4beKNdww6=r0Zs%8u8TiOS=|G9tbhMHBA?>U-E15AlK9USSL zIw6skiE5Eu#E1M6)k|RSaoGPcFtW0M?7pofNXqvXOABbYvp#|7Rfnr@bOt6KTdPQ`6pzC=E>9737V2@LE6rJ*6x~!$1-PLrq?mNOun! zUn1QsG5qtvNCBHQG{!`8{C$0?;`)i_*NEjH;G+8{2n7w;L<9XJBEu_x4y}Lng`c@f zx*pi3w6TdcG&Ui89N#y#U3ljr>WsgJLv6g!l^jUlI-fOvZ}s*R2>**PH>rDmfR;bF zbgOnVc>3GhGd<~$Fv|K(587O!%Ef+DcGJz6=qHVvcy+OljH*<`GC!E^S{11%IvEq2 zt7wQFf-3E9qQ5rfi1S9x>omqYSv)wR6I!%mQGF&j5Dy&A3Y!324ov61{F`6`?g~8N6iMYceY`6&} z4of-mpLs!grZZMf%hgGs8-?Dp{P0g9Y?yoUuf{~BLN{yo-WcnQ-WK}}#hTT9fn*L& ztV(Avp;;w~8hv+NI9KAD^l4kudYgA*>mIF=+s}2yxE!7RKi;aVcULCIXfxkfb5)qQ z@D@UgNdSzlf)}Z>OC8J1lBD*9^@SJBC`Y+<#Wldd*vsI6nDLox8L&3E(+;E28%Sr0UCQS; zVa9Md2^^_4*$k-KZZ-IcugdIU*t1+vkW9V*A!sskD?sTeRtEqnOh|hj8Veu~V1zxU zFKBaBd%Kle;!a6;#in1&`O5@9u_Rsp0i>-syX)~p9J}{K#V`GaFAP_laDgK9BsV9M zm#8J>r40luYn3Kr(qFI;Hd&?yK)!(;IY*95yQYw$gpB)#DeX=hdIQ8B3MDrVwDe?U z&e`GgRNYyuicLY{!79K->cF)lrh({<{0~C@)3QiLlr`xeJ)&l`D_-1vbQNqaXIPzFfv$k|B+`lIa?CbjKR42%4za zXmXtnP39eLCRyUmrqFLatFD?$cs~rhbD3SsQ;1cIce?~8Sv3IzxmBzizQ}m19FM|p zC8B`#h`wZ<51@%`_<~zq0n&dr?2gk_t;M`^#>FIL%7T^-mlH+Y2}vw%{1bl+V&Ph< zH+QMCN@df6Ws9s)!HgF{W2|p096~MGyj)`CstRO_Zn-v>>V_BfgelGO zOIoj9RIv@PK}!s*!LpV;t892}WTf44BK)$4yz!QQ zaN1(CByHMBtt{31ZI%SLTkQAZsL=JJ8j_*!jeti&wyrr{AE6-bGz2sw8XBO3WE4H*4 z(zVWy0)L-Oen+oq&mJjrEFk>8m0tQYWo>fb>@0H2`Sc6gGakxg+Z6~f?hK+1dIQpb zo_0~QJt^>ubQy=XOAHC6dy@`VozV#oHXL^ApxYBmauK7sp3|ux{xYd>KL2RUD0kef zt7~6;r;UlWXp)GJRYh{MarED#o85qUvZvlAVqW(h>3~q@5XjdT@Zpzl{U)$$Z(d4^ENU)2bZBZ=pQHw_8Mmm%?f`=ARIF-I3}+eo zUfM6cd?i&a7KEI6%-(g1c+m#g$<B#U9fnlzteD@-}Do)qWP=LL+tf5k$KTqi6^T2wFS>Ma3pPY-{6V!$A`rB%VOVu zPEd@nu66j9t=wB9#!^oFD@Ayg7qTV3AAI+HvC9*j&0X44SZ1$)Yk7Cg{Yg%BH-? zO~B;CKHqsZ^cMVJh|N*?@RMAFP4 zsIwNw@Pv=_gv8rMsmt}vNVRt6YwLbuQ_97TA&Pump*&($R>}a&;oi;6qNGJ2VGGjo zW_B*A=J&|+wXa-6p9)L5j1lTmKd8klJrLlc?H6n~m2Sx3{SKS!_7AhAI$lSt7ZKK1 z+{GM!`(8L;t$>huisNSfGA)a~q@B`UlRkfV^8_i3$bN}o-c|QAW^fsv(=PoU#fRyMlx#WnWv5Z`8exm#&l(eT-Ia1ALHQNNqcl#@M!#Ob*H- zf?}a2rGf322YE-$2Bup;TX=^NRL!0TKc<#yp(BB-?<_8=4T^F`4~Y;?=i9ndYn)eO*yMBPDW)>>OR%oxv|{!k z3v138Bo<3tdb`4?C5fm14`47cO|*QL6|2m_?4Ra#f&^2IA=Y)TWDAb~C4)a-CGw~! zzD%LT>5>r}4SA049QWbkMFcxPw3DhTk1SlJl;&*Dchs=E4|{Jalp7^8Hgy;zhDsE) zrZ=VK5=HI?-0Ld{RL)PnUFDX1Q@*isAQd~SrMAo*!^t7VR$qTW8ee(=|2|u(Mf&A- z#%4z^Pvjz>4d#QQDW=S{x$Z>|&9(8sT0S`!e7bF^+z#c->nbFWCKnqHq->W@W53>0 zx>BHmHEvu@E{;LB)oKh{2a+b#qgz3d8_Nc*g&w3+mK#`|!Af537^cQ{BdmfMQO{uK zjrK8`xmM2c%-K(|<>0%Ir2OOsW1xG05#xJ?qlxS(AJh-iyB<@wJAR=6TmixPcDM~M z<42-dr5TC5DYJkNgn9emiZGq2J96OQ2g9fi=g>!eA~_Ed*Q=YROGnHx51(fiJz)Og z&Rkc*TStLCZ>aEQ7>aHsyNtD7@D{pvcBHeSV=w1eg;17q?{(Nbe~<51_1q$7O+%Gq z6Ioy4R9z}a>W7I!SHjFfA-&TP_thzR^(gZR^o9FF-Nj@HCk zR|Z6+%G{HcSgP#G?`I>etTX^B+S6GJ8C`)o^oj1Ux&CIX>Iw2 zPwJEAnq;c978*@9I`{^tpGp{tgTl_mq_v)G@Wl@X`ZD#x&xfO;ksH{b40p_7;nioP z;ohD@KQ5aAhgF()O^rg%Aj%cnF?fKOrck%j9 zq-Om9sx!Pg{?WL1KSxlVgTK*NEAz?Y#ZAc~* zI2s$znXyW{Ru9Z<@jG_RaVINlak~gotj~AGe8=dRG^yTZQ8_ROY*xK&mbUvd#yTi_ zulyc|s|Df@-(U;pIbPH|15XZd`sajMbq?I?CYAvvrhl?H`Skbd027N!>52P7SN}iZ z&yoVMSS(PjP2+$7Nk?SrT9n;?H8g4zc`J%r5_;PE;x1$1dFsXJYX0|DxdkcLE0P+?fzl zjRwd}WCzYCjK$)8sb=2qP4rY2Aky6VlQoySMQTV&!Y8I#&D&Na`@Ls4wdYsZDnTk` z+zK*)<7h?i+cj<0!a5iJ`F+Np&BS@*=Riw1A0zsi$MM_^W5GIAPwlnKub&&|LbK$F)yAstG$Jc?H6IwW5i5W+qUkW4&@SX$sLAA2> zGmw&PgYC8QvcfPU=e^)$H9pj*&t%!&X7!>s%gfY6)^fUa*j+0Hx0 z3}FNfzCnD{mx3lnkK8Xg$8@@d?>|^PnS1{yT*DPZV)Tafsd3izdT&=oZM`m|M-CWw Mfd6}(`8)r=0OvF@F8}}l literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-4-Quarterly Financial Performance.jpg b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-4-Quarterly Financial Performance.jpg new file mode 100644 index 0000000000000000000000000000000000000000..97fbb612f38054d9e5a1498a7dbd477ee656ed11 GIT binary patch literal 84470 zcmeFZ2UuIzwlFBiEiOP5(QLrPV1ubPh+Y$$659fTL5L;@5JU%q5WOUE0|6qov5`be zge{N+!dQUl*i;(?L>5A(3WRCGG=u16lzZPB^JebMeE;{~f4;eQ-T}^8oU_*6Ywxr7 zT4%4a_uuc20lvCnb=?YZ=nw#KNcaQT9|Bwh96fyG$dSWGg{PxOj~+XI`pe_OL+sS4 z6Q_Ye;Gz9ifSi!YFTObZ7r*~`9RKpr(Jzjj zIDAC-*y}4{E50~#=;(2=FOMHPdg#b0;lGEE96bg+{^i%_we+>Wv-gU8Ndt*XD(E?d zBT)~t%9-z$q;#%ABYwH_SVCG6tb6MZ2ggSQ*7${M-oL)SsHAM^bN5N1!B4&x_nKRl zg;ksp*8N}V7xvuYBVQa9KDq$>8*_la`1=nIg|%w~zt%c$e+2ZMxPn*s{vhD=VIdaq zFc4r4sQK|f8}aXOwfpw4?939a9w74y064bvpG`Vc{h!dk&DHK&29nvxU>jSl^idLi zM@^A?H!?Sk-fd9c8|!fC_sHtQ0;6&%%AjAKXB?L62Q3Rg2BGf0zRQEw4f{5=6nCnk zOPvLt9$QmN%7e%%fmztX-1m-bCYv>8-({lNC$a1X{3CeUIi|bR%MU1>k9tG!!%5hf z%-4+U()l(C`C7w{uL6UU9;61 zkNh_B5ycsWl%%3BTi@r<5SmPV#Z^OQ&*l?umIuARI>YPU_(E1QLRzJDB_L;Q7O&~e zC+kl!VR0#oj2LASo91EX$~|1P=qD*qk^y3!>H6drd-!|Ypgxw^xa&DPVR6e~%C~Pc zmUlkHySPBA8D{c+H15gVqX)p1^sPPcO~3b)>N%{tIrR!QjU9%>fq*(t^XFB|?-mjW z)eob;3y636#C6D>xPmWkURlFd+Gx#FA<+SnWmDyR(5TU;$D!WpgG204rnNC8 z&(UfhU_9HKnu1Nm7JTvhO#WAM)C<1(o!~qh56N67!2m@H+C{%?y}6c(2?3kBg|Pb; z5)|IadN5_L6)M-H6-pH;W!=2%EEsGL1NemgES5`{|o(5*Pn_zAbC{*&Ez;B$uF zc~kn?Kk(M@3_xhyyPb1Hytis9b`+CeeETK5DUVlxosJ=IKJnSA z3zy+i4~MC?C8lsv&-&f+&$ITi^RCp0eQGAC4HTH1Q`!eq;EMtS@}ow;W0|Mb(H8pv zXwUUbuY?-rSOG2-yJW{msazLD=@pNA;8_Dn;X#M6ldFRiP}KZTc@MTyhY@BC=~>}g z6|jd=hHSSuhnli@jjgCmI=pB$(2zMsnexk%LD@0>z}?U9!-^_7Go2H^i`HhSw7He* zj+g5<*lI%Bm*=Yrg19q_H#p053qPZw>t5Qy*?RP+JazAY8tBlIVWKKg1s^4B;od%= zx-w>uK+j43TG_}%B7qUcE5O+ctdHt?;NOOyX-Bo190MKo&f`<-t<;F|eXY?`@l4T% zSEXPIDypj4`-=3(=+#g-ny{uO5e-E4@l>$b!ZIQ$levT{60~+`UD|yN>Nr&il9}iF zw&$%!I92@H$cn!=-8)Jk5wk)Mvq_LRD^8iHltD))3zC=L@gHlM zi(Uq5eJYEU2d&f>r0E9befPvZm4WCeU3`yM7HO(L#|}YkSxmM-L+oUADKX}bWtU;_ zNHao`B)ExHw;J9H+azIU$Yw{(lg~V(c{t*p>g%?B*r-P@z#F2=h6(kip6}a72mM0U z`HLHJ*Y5PcsEqinx&5)=#!E`f_6i~{N>IhB@7XmD0aoo@q&Z_(?&((a$4ozAl!-j3 zgV)j3nb0mIIFw9E#Vf^4Dc55RR$7DWt_8&Rt5DNw`t{gOOU^NQ{EUf!8rZMceSpi% zQhEER+GM?H=fo+WzGd8+!steV(yqAm=0c3hlC`@)jU%G1F7_1)y6eyy;`wy^MI{_p z6C{CktBePkCcGyGrV@rjcuUsz4W1{ze7T;IFE{s)`df!<>x3X*d&FspqrSyC4wfJ7 zS=Qy8I~%8<5WlNg-X(PgzMwp|L=P-*XqoI^OUbJvlW{Kdh6OWoja8l6vMY*pI&M?% zn}1w>O2HDbd?Ma2k#HWq>GzX#iP`f`lxZE(NperVpRP6(L!1QZTG*kz<4lZik_Qnx zU1?5gnJGvbd1IpZCBt<7_;0+3Zu9$IgN1P|Ne7DhA9h zE1Az>u_iokm@8)L40fa+^|X>@bXn{{4GElM@MV)pmm}K-%0`CPpaW9#>Jx2x>~TNR zf?u}#J>`}1Uc{gk>hwc;RylU}#lxHTJVoqs$949;!}K6e^3pX=cEhNU4F#jdGkJyU zWG+Cv#P4eZCZ-?P-I$mwG0--Qu5EBit}WaL1V_&A1BfP?U0qVHHU{MlMgR`IZz2T*&;Qh5J#eRpe)Bl2sv90>`r$Me~Dgw?&v)= z8Iz!^UxM;%fUGj?WTTBY~_yQHj!TQ_7PN+BNOPj;pacH zc8dvO{F@2xBX{p)e7~Jqf8m|$!bk5F8L8X244pQeiQ6>sVJxPYtfWT?J9{-1@jd;V zf=_{Ci^%Ib{c29~x3&QpMMs0g^aDEL`t=JwZVrVEx2}GCnR|IW2{*n=dE%?Lz-=Eb z%88dr!IJ00ne#rOXrDr|+*}ja2tK&tC($9XX^~(v7HUHkz`L(y^de2<=AIVQ^-A36 zIu;Hx7Tf+fQ~{q}WV-5y+)gt!-lEqG z6&=E*2dZGnd9|}CSvb;56N>!@`ls~iqd{kWZMEWPaO4}6l&MO}WG~t*kg@$PbXN#*-io1o z6XOfEcAHk<3*(a4e(e~drmY|-Y9`iq3qODZZ*-v1(z(?2FnMo7)Gt6J$NVcp2Q1;!py1-v{V)a4AyKud8ix!uZwMgfQ%xLXsC| zMn+@cuWdB`pq)hCMDdW29_$n|warpXQYaiPflPTZ$i}+|*WWFK1z-%3(sM_I4L<#X z0kLy<@Z-PNygMv`ig8gcmZ(clh+uD-rwkDDx~{~UGL%4lyS9|IApCS*D848b%`=*Q z#h35*@z|rjG0Q|Q)-I}y4$<}jV{s>eg-{iclko?7fDQL=2&{d;o4WPhIjyo}SinBu z1AKeN0`-oX?E5%cfaNq%D)s@NG~q# zdh|;s?c9skiO&wPem>CVBlq5RRb?11v$u@0$;MA4K1>c-AqBsoC+FACcIVE4nN12c?pbJBPyDtSljyQ@0E9VJ&h0|1d@2ikV%d_=^2p=|;x z3G^=U=tV0Zx9ebwT zs!fVHNb)Za9}yf8e1Z9tH(zJIcl`H1&R%GAXrTM;1759)D_Sw)T8^&$cHH)0E5DXf z)*RHC?tsnkJ&G3`VUM#z&LSr$tEkJ4(dq+LL9a=t%TJR%4wC%qz&A7C_AJvMpvLJ~d*o{{l)XFvcOQn=mcT6RO%}o|*gz0V?!(14GyTAHD_UJ)HR?XGZf((4 zo}-1(9BGI$E+*hc6|l3j*hRY=YRsA@@d#6BG0oK64~W#IMFjScibO2Ath~+D)m0>Y z!3R~@29_@sr07!kR9rHX$9GgG@&%KFN92rhX=(+*k}iGq6Uf*O~3;qPL{{bDAb{4AlU-&aqF zq#hp-4filg7Z1@l^%O(YUgGZbv1FL%%msJ3uktT2TWt*E_7u9MgIP!2-P#5Ti6x6N zZIP|3t#)^c0&GZHvQOTEOfLNLW%xu!bR@>>qaVbDmew<+$X3Q#j)g>2w|Ehvi23|U2d0g9M-n3G z9UH3_gT+N1y0x@(eZmunguq_RNnNSFsT;zNxfzKJiJ{cFN3xgnjlk#xXI32N@N+?< zby~Nwl%s=Xb&G?oVh)&Yh9<35wnndW#~*l7?Y9-5eEAR7v-4w z7u8ZpGiaXQ@`>1;P#$E%obS!XP;gayr#qxJ!N*RUm>Au;zV zB=oqFF_kTO)_#Y|ZnM%`VDod8actSG(#1-sYmGHZYmC2a?fUK~m#0z&owsqslZ#}p z@~D7@z%mhCt0f!6$x-#)SB6o4pQ`^YIuD90P~Qa1F>-zQd3$Lg(J$ne8)2h8G5YO8 z15~dk>{LR|I+0^O=#WZ_@Efvs8tG@Pw0Ex`u72|TlgGut^c9n)Xw4QsXm^xh|5PJl zU6t%jDx!o;5cQqi&Hnf^Zd#iW*#?VXOs#!E4f$v~XCHyf-(mJFRPk{|WN4Sg($F{T z!9~vx8|a0}{+D&xs;{Bm4Wx2lMD<`merl0F=02JM30Ln=E!ok~%*fD1s#0X07i#88 zYR|*?HNQ@+Xu9X~{X(}tS>1bA5u~acQlBjnW`GVPhKpMb&r+zQB68q9fX$hJowJUm z`rPG-`WDS|sZ+<+zOKu@#EtW=RZAgOFWJR+)V>P$OmVp^_DRfRO2YD^?+pGn))+az z2;B#?X>p%8fF#F*T{S#>DU(3b%TnVO6P6f;I^vw(A?73{;?o32Sq;sPVH$S6A9`qe zGrjWHURqZc)K&(lx73gDA=DP*@I~v8kF$8=oBRc%AT|7hA3PiUlq)^;>iSyS#O<5b z^?X{!^VcCgpQ`Z5DqSyS3r}VFI^Vcbt*%i`4;;~A0_Asl@AV0}iK{8w9ilnl{B2aQ zf%QtHwh!;)Xu%h{otA;NEN~T5H=2xHgb2(pOA5?j&sH^kv507mUa~W_Ee$;4Spx+F zA=b%dMX$v;tsx-~XF|+9LUMU}+_+$`8Ej59db^J)w-1mNTrHh^FA~&w*qxo;#&!Z- zYGHSiKZNPse!35MYmt=q7PAjXF8ia~+iwc)!ITHiy@*hMyNu$yPnZe}iw^spkajL> zfOnG7Rhgnp$l_8nyTjpn;pzhPxz3)<#kWhNf81?hu}yVhjslM3IzNAhE&;E!bY#Li zRmh}4EN;+m%yND()<&`9_N1fLh`nZIYpLB`=GJAc-er^SrL{HMUsg7FK3ZjSdMY&v zz7Dl@ileS2QeK<|sgT3$?@)0KHu6QOWdwZkTI z=9Z%7n_!mgw&Z$~qjhvtTu?zl;rjV=ZZ5uC#jMzRd&l>M>=zu>lu;FOY^6g}w?j2H zhf;}9X}S5Vnw_<@54h@kHpuoTV2;nFa;fEYN}b;PR%Oy)aRv!bo?5lqwzjxddz+Ib zYDq#e%-o72;J0~-RJ9o0U=u8AY)Op&L35|qGSb~@4uTXbR@Q8=+V-q;G zf$n8eI_$KZ$I&7>9C$4+KbYxgA&88GK%R}7^H0yIj)}TJDV$&*t+l1<1ZbyUj=QDP zCcAw%9IQp}nKQu}gcRx?ORFgcm(=!zWKbmFSc1So$Ay?ztAHIeid)ZPI;WI%hM0Wd znMO8h2gOcEQr>DOYCQ3vO9oQYKq+LjS_2#@2WC|`0Bv{UOOt+it8DB4u}C53VMK}O z*CS{$hD25*CZU2R?SHSRdQ|y%sun30rrzou#R#q-gDDH*&ER@*EsQ<-Mem?R48 zw{z*uiu$`+DvwWt)-JF)NSq@}!mb5)RDd}ht`$4U1vu9KU3MxLD;JALA6UIz{QQxr z=hH8Jx#DM|+Xuz$Z#4&3C4~Be-{OdFnuo~uzA3)9{&#owLDK)v!}7*!y!qNvHV~?7 z(Ppp3)H1!j1t=AI4IWdq`%#7*FN8+FFdR4zKxQo7k!aaF>2WB z^ib|<-j222w2_i>P6Re(2Gdy4qZFo%=Ak#rWKm;NH5}K4h480SXFOMvi`^={`_&{O zaP*4!upmi31~)s>uibMjJX^iR;a(TY>D18&L@G+}_GZ882oTJf4GCg+LYD~rHQc5KltU%Bn~KFW$zL^}jDI+_A%gm~uj!pV zJRL4R@Y;p&q3xq-q&1|?bOtv&ZL!#Nbx^VUdUJZ=56k+eN>&o0)Fi@A1*TMH!;A;g zvU-+buh}M}JMDW48Y_&)c9`>n5w$c=uMJm1PBhZf+1}WL96+Ru7Hv5I!N8umnJkrN z>WRSgvecw50a0JAjcpmNr4u(Oh*m8FI<5+>m6oQ!dPd~>A6|O1z6@6{@b?AIL_M6p zVCrd??o)j02bankiddXJu{KVv(mh&NXeCVzvoOL=LRH7IB`cnKC1pqz7gJ6|MNhwP zXFs48?*llN9-L%ZeRtyz+j$;+S%I48r=FIhm=T&=gK2CJsvz9BgY1sY43eMBSYMj| z^Cej3p^R0}^o6dJHfPjs^mFYIujk9I$vX6^vpZI7w;3|qag_x%Gk6k7e{H5Qd%Ar$ z?(z?V+( zh#e&S7l2PJtCay`uTHyqx+4bT+Vi@WC12=IvGs!*+@X#kp!!y;5dr$isfop4d_k9A z_w&$8sRd6AOLa5vX7Bp1>;vLz{N8CMJ`fc4?(!#_zOPwRd*3AB%vww~4K^pcn(PBc zscx%j1z!Jw9416YugO!JzTc&9JKN1!7i_b;k{%*InwOqSUW93ES@xEDv$x)KcG%jBg8oeNb;{FSh3WSPIPq>s-9>GghwSx(J`;(3W(4-) zpZ!>_?yhW9x1h1%^(@2b+ApsMn*s%I>q6&IP}xqU;*94Gw+b$L3~4uAY>QJt20AHr zIKB<2uvVji>eLo2PSldw?eEvR%(Of8f8aEe6s$?55HPZPap8OQ*`uh&HZ*)VjcEvD+3xzLQm&( z4}~!?HJFMVP$R3!^|z^ni|u0ug;_JQwkRJ?EavQGb}Et4U58@8Oh)LVK7;d3SDnV9 zY})eV$`%5G+tJH$QS8RVA$n-HlL{evij5%6qMKH&Gh;I7IT|-Q&~vf4O=lb*lg%sB z=?dW`h9R|@?c?H1Z%|g-lJ&NKA9bjuxO**cCOu?H%eII-U_=p0LeL0hAl#wCg;2c^ zHJ6Vq*o!M|gv7EL=n^NzTV92F5$>>w|P?u@u>COHpKyyFw{zts2nq= zI1X*R;W}YLqb&GCh9r_f_oyQSEv%@<$ngc&f)EH~bp@e6bEEzb!S2Y3LX|R2a!NPt z>TyqE z0x1PSO}5L9F$toGdb_Xr8r`vXIXy)?%oSQsDj!=gTDapnejoH8=4N+Abh+43U+VP*)dK0Kzgh8tzV6qBB%rzXd3XKeLV5nDevk6M=Fl2%vMq+DiZX# zM6bSg3Wk&25s)E_BzV0^R9Ua<3TO8k5clIed>`N!v8%&gfh_!50}opLu>v%fHU7%0ekH}dKahWqsBRnMJD zMpuy@-9x62gbzRm42D91qlE6ZSf%b`D4%lDa#6@F8DAY=1oL{&HYMfXT%nt zmO7|E$VP%B5@cug0gi1oO`S@)4k-|mD=jX&kWrn-!{*RC9H>s>C(7 z48n^o=Qmb2geC2;x`|PnBbh(TLRr!chq8Uo^n~B-qCu({UR1Z%;k0nL_^JaGYzmSc zp)cJ~o&=XJxvo3;#!6<_?4I{I3Jy)HstssJ`p_C;cV91MR?Q_%X)U$r)Zw@NyF)wD zEa(WCOuRyaKm_v=P~cw9g_E^^up-WMZUi)!95o!(OxF}sd2Mg2@UgXr9nDiPLd3GF zdWU`&^y4`Or#4FUGjnYWf*x9KqXx+%T}V@{23-qVmfS5K6KqmavJ5=g)rlCr`7qpR zG?qvs6`JQ{d22bxJJHj1-)E3&dikTJwE?rcJ!cEj-Cf!P+^V!{tV^!E&jDL9P-hZ5 zmHe^7*lbhrMY#>kw%yO|UJ;`9>*2LMnID6v^+UZ?>+Fm!;ptQE!eB)8Fgr%Y;2n>d zQ)ff>!15&RT-#$5u!LI?&}N_xB*(C?Uj(zOx)2E)_7b@1N%2p*U6Je(QD2-F*xY@Z zu0kAD6oJaVZ5d2ms~>HQ6l&zcG3V3oN^H&`R7)gGKHBI7KGk*p<1=DH( z+sL8p$0<`IJ>TPK^-)3t&GDW{DXEt`qov{b0pt}~o4Wk!2IyubdKOa<5~zxs>WI^u zvE4a5akL>3I$Qv59(Xlv5R^m`ALRCxF=Z^+5IYDYOop{$>=D#sV0VZzK!)C!bS zI#J5f$wj)MNY7O+rAJZ3d$KXuJ8T0%B_oXV3Y%U|u!fs|n@dPKek7`$giO+S27=jZbWYPx84qoK+-=_Y#hR@wo@hD`JwHT@&1u=rP_6rZra z)98zRW-qp^DST=l@Z8_A>^X^w-yLa`kFG1zPd3h79^GKyxj86;q-qqe} zm3yz3v(?`@tJ3OlO4@zy4VgKzvvjb;wo8q8n|z_22kYI7dfjovqd!fy2h7gSe?)c0 z(zUSU&2@wckO_pu5YCTGmr-3t9fG=ZVX{|^cwxs^IbVU(&4O|#Xz@}@Yo=FIn&EUy zlaZ-7Z?bCMR)Qu;7P~5Iz0vqQb02VB7>0D+AT;b!+MtQDtR16Si8*|jx#f9l47qd6 zH-Dt<$I!Uv%RUDbNT4~_8%91Tz2wrcu;3CR1-L9!Ve_+5S5hc zcje3^jjrPg;1(9KUz-0A;m3^m=o|fMxJA2jo&yGZrxNCCqZb`|GI(ZZ3M6}&Eup9V zcSX{Rz|G8523XZ+P#AD(=@8VyASx&nJRl|nTo|=ovz~i!W5Nk!<6RwwA7sbmV1r*? zLi1}vODV)!zc|AN&^)oUFr+rxGB-`z#7SPQ7de;ZPQTkWO!pD#*)=H4)3vj+QEBycsS@o6A5AU7#dW`#A)Cu>Bt{HNf;3*cyc)aG|Kfd+PMc(@Yf+BS zzZs_h8rq975*mm!ajjI)tVn!Ru2b0kh6f!?h=VQol)HUI0H&A{gC(0uTR%0_42TN{ z!so4XI6+Bxe|jjsY6)3YKr)ZayWqk6KnMWspc;<;u{xF7Ien;TDd`Uzq=ZKU+HC^#VidMPL{b373Bm;c~gD)imw}4pmQpz55h+1}v}FH)?zlo7CdYCvL#u}Q z{Cr$q6S90Y@4}!j<5u|xqrGY-KHk;WoLT$eu&pW^3=XWu5hro^yAswAI}X>vnR>I< zwcdi`_ruhxD_0mdE;&awUk{PLmBby>f245*_w0t*IQ`4W`}bOV%r(kW+NQvE`W$R0 zuP&=q&}M95sbzL)clH=|P*4An^72b1M3S$TY=So6w*%tfp922(2eq&p#aasBH9krm z9yE7m#OT>cN48wJmIcFZ>(cb(gVe?yn*2t{ZrfPj==93v3w2YDD;ZJJse%F%MtDoZ zo-oN_u~%y!aGyJDu5MzxJ}vybXbD>|`lhKgDJ!~w9>4Qq*_fN}vkwqys!lq99r^8? zXJP$r9amV#=+4Stb+mkHXo z-e8p$n78^?iG%DLvKluu*xKVy(kcBncUIsI%hh4z%2+j^=YyD1aF~;I{Tr~}KV`5V zEadAy5)$VhJ8TATR+K$Epod?4?A<(DdqqV39pLT({H#sYwSgy_nrh4&>NgUe#0Pio zWJ}CE?-pdotlb;L5WeJ}&-j<~RRvY**X(q>)!ckyG7)D;YUrIZyB^ITO`qK{yQz2k z=*L$FI^c`$Pv7it9Ct!x_@yqAVGoZ_)fP?M=0XEM@#&IX7Y<+0`}x8F^nVD7jQzvg z|JOGUtnd>rUftLpQY!j1uo3`&fJlDBX8s(2`0dCypQ(=&y@gdV%HD>1pzXaKK2U}b zun*N`mYQKDbsZI&eWp8(Kn&xh~sW=KdB1mJO}0a{1=W`HDl|9K`L2Elr=oI$0eg?o`B}3YeaUS_zz+0P>bm8-Ni3% zO>{_nT-sKl-+BE$*D5x>Qg@t3MufN&kwc-TyG3N|98!0aC)d_!m__NC9STsOVoLMl z0_Wn0=`CV-eh~J3+f>zVbSOo@fWbjdOql+vPg3!cDWFp*ub=<8hTG$1>YZYdRg#9A zY=$a^h=FAU>?|(7IN{0w<~YcGsAb~HIwfT^fF6Y>nn&I!2>dwVxSW>uBp_JH$3sdp z9Xe#-26AK}*mVgDs7?}wUrNHw?hfe*ow7fl$ zrXysv1F~AXn;eljBsdiy(eX>Cpa0d|EWv09MQ6%fy;ncZ*nw!^cF z3>&Zvv4#)RYF&Cs8#MG*UB2Fgw)Z?|;DcS3jL3BM(~ID`*t{uYYrs6s^XM==pOC7> z-}BEeA*SxX3F@Ym#{QC$$7@4hSvQ3NVVnM#=qyn=1xn#sNrK13J<~E-0O6cc|mwYeh848!i_EX8Z}WQwmf|8&6P7zhxIC-a_+Tm zHNeyeWM1(UBV-<7VoYH3Vm=6iAo9P~+fy+dRdEyvcvZJ`?VE-?KVe;m9<;d45KTj( z0&6_u&^~$fYz;f*5MZ&cUd5~Fr>ep*p3db+ctdKPE;7^x+(cr@ZPn73jCR92Olo$Y zrcT^rPse2J+)Aw$G=JozWPiec8lDq-+I~{gGCXPoh1?vnv)1#Wgl=7-JU&;HGKzW% zUybL+0)gfnr^>>ON|}x4*t|ZUd-!B2^hA=|qsf|dgyhnhOtO2M8=gdm);TmlEg*|@ zs}a-q<`sv&G#EJ`iShH~`+qQ(4&M52g#!TmzpQYMSa#mLA2T2S*h>qi;l%HLIZNsL zI84d0*+!w5O>uGh8Cxc*T_rTR`0U9|&1)rBoV)ZQ;_E*iHDKs4VOB$|-k~9f3JZ&? z@4}1HVm#!gToA|ViF-y%K%&>r%xcC1y%yd~ zp*%c#p0N_@k~)c#uiF+v_9bMUPIYk{XHVB;TcUN z+2_59mnIC3_HSeVLR4VKq>^LnM{1Qwslrs(xrariT-tAwYvpcRymSq6=%T5~3_&4! zd0~n@*VRUTi z8lSE6$fMbyi8-`8DTHJJ(&*HjXh1{OA{v`6dVd1`k^9}WeX@+LLLDcV<&$TVR%u7C zY}|%{>Q}8DR+r_|cC33=+lCAEJ2gIRsKZUVIR4G;4_tL*3emrtQ*A zC9&ARVnL7xB{eh1PXf$w>~&nX@5wLiznQkxbdA|EJUHr?>s#FwlYTMMrDdqeW^e(f zHE&qa%2;e<)pjL7tVPvUT=L=Zk31ZTB6h`6+JD+CJ%cfqh$@(s4`7hnBHR0hrh09e zpvK+ZEs*C8a>S)_tW{y}0?Q`+m!hMg=(PosdJLn(d3Vi@V05V3!)c zmK_PkPa^HvIS%FYEQeAq*aeJ78SVqH(brQjd9^z??zDxRY8$y(f|T6{$P|^w?8v2C zhTknL2n|RRXPdky4#OIULh3ETRGQKfyt)h$^Cxjum(|*R)LI$vrU-Ow#UQPb1vOqr zg^fP&Y7STaj`u00J0y@nxI(itj$77YLA54`ao!1T67!zpDEO3eDu%w zE^9m@*H?=#ZP+{{P;-E4?e;NE+dRHp>`ct4sbjr+9uo|0@ay#MF}E3_-rh*(rL9fj z5#}^P0Di&lS0x2spX#nQcQH|wS8jDry~{UpL)=)Fz-~fLxY=Or3R5>m1(Uxsf8woY zw!wUxUja{MqA``%{(|FjU%UT0y5pz!#*f1;6LX%D2pSvSsbnQY{7Th)@8RmV$V2$C zRZ-HLC7d;hE$t?CWsS&iscB?db|Kkd^XDK_FpE?BB$%q)cg%_0kGZI;eiQ7mRBjH>AyzMolZjw{`XbCCxXB0z+6JQ+v*m z3Ri+=0+WoHywBQ%cT!$CH_e?kwPyh#(A`q*ofsVV^;hJFj8AnJmgRS^yJzYI`)>J2 zP|AfNn#*T$x1LG{VkqLhkoH|Zc^~lo-EFb9N51df3sssYV^kGJk{j8m zdQOMbGq*m6mY$j}qSrSbk{Yr3| z`|g+D)qcT7{j?Fp3C~uU8fo_qoE{;Y04ppU&Rt3gLnUF~>LTo-ReE1;$1pS!wQL-; zpb{Y?4vvv5GlS3%hB%1RdC%oxtj@?s$Kls}S&j(8Z@cd4{4n;f1ZiP-MdMP}U89@d zc?#Cxe!vU{B0sE53uyyoojd7`O(|NxYB8*|maX;ugNdFc^LH**moKL~DV~^1({&XD zfq)TQBfU`^7Ry~XI)2(fpBQ;8G2XCCEB2|rLgQORf-9_6Mz040 zO_YK_CU5PMj9m1+5sK6wX8#{8fG2)vG+TR1+NnwyguOf9E`J~8x03;vKK3*KaP`1o zKVZy|Jm6rM+Uly0zdCPie+2ba`{>{C|Ai7cbO&>W_k2ENCv@hAH^}~FxRd|8;Vw!3 zVvfNVYd;?7fkTMRuabhuN%|rJr;@aTCMc9!n(w1Sk7U06`fupR#thiP{BgFi1&Z+m zS^?>Gj9jd3SXRrj$rTMxzVPmh>6y|fK9wCqVY9`dj*)iS__)BK7E@7*OGQm3??v9H zT&`;OJO`=DSc=)wV0TkXbhk9yz+*U~UqI6CR@5ysk#N5&<2~|H57N>7=7-YTEQbU( zC^ssepf`vc4B#)7(#23!)}^1$B7-pkr~|0G$-9BDaIWbSM{ZGEU>8%VL=AYP{*h&@ zRj**C7hb0RCtaoUWtSwCrMeZ)3lF z#SA4Ywejbsk;GzQgeodHGT#t2TECQS^d6V8*d&^qGp4Ox2*OXrps&0)(XFY8nuOZ} zg)Vi)rrH|6+-Jz04&AzYO#H31{(!t0@~)KTDAfdP39QsI)3HR1N(wjb$Q{-5^)N{6 z)C|YXm*lv7c=*)4AtFbIcV)#>4?Pu?LLKoKG(pOF9KD}9dZMJ`d5GllQzdV`%u`@v ztC{HZo$xLlTceW4YOk$=a_w!`NdnuDT=QSmu(SB4)sE}H-d3d$puEa`gkmXn_8zRV z^RoL`=s=sF(X|uOr=AF3h#+r!+W~H0vA1D2tj#%NRnKx-DIsbOVWPEnbIL2Df)<5w zuNz+A>lFN`hPh~8D?|2!d^^U-o=|!5jOg`jFC+w7Jnf@2&o3tS^+JK&7~BvsL55JS zw+!|#BXuO@J+L05mbcAy1A~5;xb1W$I;bh4Z;)vLm z4}?3nB)M%AE!F{7A?CX}k(#qkNo(8RH$B}`*<_cb?n+o6*~7M8J8HFZ<8-f?owfK23Sur)9@g)}fp$ zt@WB53HqbL{ep<`{8xFKl2P~a5+0l3hjRKAnnafe5nS!6igooC;llF#i$mZ1$D8}d zmqQ2q!&^7*Ew5ME*iF)=-qndR_5lwmg@AY6-+cM)XWA2=qn!w?pADn#T$kufe|Rzf z8w^2u=+)owzli%a?icIW&-u(>dmh+t0b%O(>-P7Di#`_(0Gv2=Fi}@k?O@jY)8HZ9 zvA+Ax_o0^$lKgYvV|5Pk1&oA8i#UfK{X+a_Az#+=f#wJB_%9pPsB;y>x>0uRZ&p&FU zz3)Q70;j--l5;%GFsa0vA_6IO@kXbsA`kl0v!N+q*u}urD&J?7=W=_?eCdU98EDdL zUZ{fvya-1(!UxWBJ5F@nn8|&7V>W&;*>5>}j4?Ve@U(JoATv4c94JeN!7<2JSY_mqrn| z{NVYxsolSJUde8lm>F7$IKL10s*hy~))P*1`1KXEPV4On`w)4dCyJ4ELg>oby}2x% zOnrs~Ux)Oo#f;ZXk_tRpesa;wr$vxMCN9HLn}DWlJ)Qy-*PWXwusHycu@O#whV zxC6+zRrB~)x9M7gQl`x=qd|4qTw%i2RP3W-cYI`0e!tP^+9~@sSjqV9#xg6hJe{^J z`V<37x3j*}I45iEGTGx1ivcNtv&2rVbipFrxtiVlu4SO!0z0+Y9!es{3oKx|R@`Gt zPgiXkOv4~QlHM`z7G25-HwbgHU7IR6ZRg(mASnypz9;SnRsSu^#sWB**naay#TZD&O(<}_LBsie-1j41eEBnFwGv5=jI+cGWld%Gzs ze_Z4ztZ1@oL@yh5FS7iO-=p`|I0|}yRcrwpTU@P-@aHwkU#fh(v`~VlWS3+4JS~q} zw0C~8?M_eVO4SgmNlYy-D6M_YH8KQddQH8HH!@dY?`{dBkg@rZ9r6Bog~GIyn;4qD zFpC}?u^XqX!q{L^X&SM`AE9SN;g*r(FIZ;pNGUi1+e{jRCg<-o$t!+8lxp}adFLfH z^Y~hpIU;`&R1_hi=40n;q**8Jrl-^mI`af1OzdI9DrU)u=q%*xPzC4l_QJMlgY>HK zRc>?KUAJ@#*QrHEXyXc~3ARg#fOj&wzK>u3Av4CIJU-0t&CL&A*w zr>(ZbxWNNEL4OHH;c-lL(PXCo+#E9Va?`A8++|qowAG7sx7^r7VJyt8k7ixwA7Vv* z5IOz(-?0CF`Xc&+%b+Cc6J#sM|Ge=h_4!jqn@Wkg^fS6HnZA{~lFq{4@#;h093c5W zga1F%dGa~nHv3<@mZfX?1cmNe${HK#9hI>*EdR_pL5!!{qwt}6xOormqgi$;RUlm% z2Goyb(ornn%WJsI6Xxd#$^~bnM@@{&>tifxPKVJGhOO1pLm9`v(Gto{Mmv-H^_!h) z_m)HG8oj0W!@66ubHtIsKy<9uuViMK@E3Tto+Uqyese}1Q^L|H0?x&iZ+zbXqLaG$u|per+~~P!i|Ae;sOy^( z|M5EimuPIj@v+|9K)kOsqqf9KQN&h>?L?eMZ)RVb78>nMXYWDley?t8W^g3pzlXOX zTtWsYc^+r+q4rUEwF@p3vM};FX|m>aO=!+jK7mRmc*C7WeA{LREOm~W+~{5%4IoxT z9hOSI+`%1^*VACh^b8?YYfQV16Xp!xPT!|*B>v_d$S8BI%Rxx6JQVw;<1|>1cvsDdL;Ny?r9qyHBO;Xe?{zsUWFsWL`K~8iuhfBkIqbp1-gd0u_N}0U= zu>_V3Nw#>pS9jq-nE{!p66Nj&2?8;SS_1_ zHhgW+F3UzD{v^+#;^loVb=okb3_7;B~c>4*AyX!ad?ScKUAxJqc$C{1WO+F zQTYxFg>vW#%9^Q?C{VB<6@fauAvIYhS}`&N{tY>HWtqZO8C9R@srO`TM=5V|*rB)X0nWP7eN#^=dEyZ2A>=Q~!P}+zk`P*J;?%P$p1` zy5r9@k1>UuPZ$~h{<&>8Ui@f=hq8m_k4C84FG6QCn1CCe}D3f8DJ&yF+8$9cuNWFx3)t_j&9ufAs-v$^?c zJRjf+%JO$V9)NF39bA5-$x~3Xlw%%R)S*%P9`rOnP{poDS6Yd@7arqzjk&yZI=kPHWcnVI7AC-IZ@*zyDv9Gt9kzFB z6eS0Gd1`7<^QjT#T5V-UoGQ8~2iYokqX<`OKMtDTaIFwtRFL9K2u!SU(g-)H?Zydx z=exA9+%-QPb+W+NH9oMgJLC?v$q|9Xoq$VV2BubWGdr^LGSPvV_|l3UF65Bt6Ee37 z%ujPzTChgZ+)^A5X~xM%j72BLVy&X-;@h-N>ly48$H>w7=|j;i3%ox&JP*#$FQE7i z%sNt?(>c1pHHL!;DWvmLg-L+P*f0LUmi^gOWt{IoOCUBjC))+97KlPcp7izGUN}*S zI%T5r1j~YPm6&hhZBOn9LB_aq!O%RwwX4tsZ7mO)%3^8GuHOB5@l@JWGfROC^?MJB zDH$+a>hGEBzf3}kBHyJHzP=vfB_vbo#}>%)x5#zDRON=CHiI$`-{)D=%Uw(t^wgJC z7~?m%o@iBTPMIpLGvLCwrMVH1w8R)1CM~xnc3Ehj{D9R{xouqs(`#o_N-7v*N?vdA zHl{jI1y+MqP|Xp$M|6FZGhALWK2@@QWa9ORcy&x1=MpA6>w3N6fFECH2(PO;{}WlY z^D-sq83(zUoWu5~Uw3((U6H^WZqAi=dcy2({fN%^)te|cyCwZjI^=0N!y>OlE$2{t z>1`|qQ|f)(Hw{CN&7VMG9BQBYbdf4rT3>Z2y2UG%0?`b2xnMT5uBfR!-P&(2iYB*qtl)_tK`R4-x{jTZm`J2TD5w&%fDNBkju~28K4D0 zj?AY?oMG)eZjIf?5YL zv?)clilQMcJm-zQ)jNIO_Prwe3g_dMWZuSPVccRxsB>hZj#$T$h5b$6FDIyG^yMB; zs)J3bv~qUpDl%)JMv>{J0p9aGOxeZw@KB%cRvR?WVN{R}8_Rlj2+G~oiAM76y0_zG zNM4#~2t|U12HJXZEXKr{A3kc9uya0{bO15ROP8KiW9%Q`c@BX8D_W;R z4g0B>ZK+((+*i_smL^;0lmMv&j)(TeSLaFgAA+~AE(ILWW<;#0LvFgj+T84nb&@YZ z`5sASt5v#PgfD&AJv+lQKL@E|=@5cWsNS3-s$<|9$MCIuQlv5lGQyh|m*@6fkuLaF zHWEp26AQ+=CB~`})D2kpMD_XUaaQHEkh4z8zG^-vGGV&BMP}>t9B$D9W78%-MQ`F< z-@2;2yOeJ7!Vkfb8N+k-&r^XzC+R^FY1`g#`Vs-WW5FPU1y$S^ca#_G6J&H19Jzr5 zsy!yWsAT}k!n0z$gWd}O53v@1dpW9_QrR$$O!VPV35D=)cj+TL);M%mei7ikY#&qYq9Ho3e9NomXLPXHzW9nDj$ox|I!>|4|b($^H8 zoD(#EqPot^A zT$XB`Z24YYtv`MNENieC&pcT-;N>&6x_!z>cWU?U*CwjZc`NBB%)TcNcdc+hhFfv! zzjoAKKtVhYibd#VK5@4Dxf5M6B3d9jQ$fsb70PbqR|fqb0%UhZxY}LV<;AAr8jK6KxiMJTgWV$m9w{m}etc=Dz6Y3T(fTe*M^N6&e_&+wA}--Gz1xM>U? zzdvf;Kgi{_O*^5kXoJ^{^Ro%i)%4*Q)Hx}i1h+C7LneaNF!Xat44XL4LHzckDN!VY zX}4ZM3(k5Ey5fZy-~#Y(z+eh7{P+Uf%`_2>0m(a$-b*rv+gPY~O_!`J z-+RzxA31oZlIT5r?9aCYVCS3EoFjRdmlPx=7VAs%Guja$i$*ir1^61&S*sNT!oH{K zMqFFVMa3)eC)865yS(zJFr&KGlSu;#jkCosZND3?=vNgp3L$H-Mc*-&O(1UUjA7nj0v(CI%!h2Tmz1Oj@{$Yv_8f(}I*YZ)N_?q4pCe zWgLSGBgA-ZmreA`KE?BFn*h`d+uQ6}`t|LHGR72~PIwxp(8?9&Y)+&v)9~I6P(}ZG z9K6)vdRp=6>POCR*#NJ3^I$qmhh#Ru^~q~~-hAiMhV6C%fE0|1Uef^rUgcg-cU%tio{VJX)jQl+@vzC(u1WXf>z`lE&3(n@iB-(X z?6Pp=P0VRBrl7@y?J?Uw)wCCVZ zjCuIG>b*<#(n2HR&9W&A!IG_Pfek!L2x>ms!3}^fw-4b@e4)t-Lk^x^mG!s6D5ZIl%S^iHbdm5H z&s1=w_ESZM2<7c2Y+f~0=}d<04|(XSVYTc^1OlT?X^c*5QhY!zTW|^JxNQAPS6AIp ziUgoZj@I!Z`r?AaG~@0EI>WO=mT*wnAs1C1$KckJL0aut6R47`d3vA+#0CoC>{N^N zCs)kTl6@VzRXjXw+x!YCLT=*Ykmy)Rk{uHwOKd-z?i*jc@1bLb+)OE(`;prvI>v78 zm_T%z1L0u1ea$(pG8*#A>`8`dVCKl^FsrPgO46Z(BZ9exjDzK7a`qQQtSkv=A7uU7 z<<9hcA~~@YZjlx8X!tqqIhn-mLWJ}YFF3v*!E!M^;leYM9QN{fI|Z?P)}$n&_6WWI)A zm)IbH;Q?9`CSmu)-wV|hqQ7jL^-7Psty62vL}O}}yyFI(=0X+JY+36K(NihGe1Y=) z0)Fr|qda>LavQ-tVi`)79nO5WeV_Kgr_+}IdD`@O*aMTql6>Fb)=3Z~of3#_k*jbKB%*J|I+b(Uxg_;sW~S z|G<6~u#DQ2-j)~F?h?AR#KEDyJlk?>$FwqGj-s|?K~~PEpTVl-HVI1S8x#Lw_9E<$KcrSLfp9h`m~htX0X8y@qix<4)@hlsOVRXg zzv8FwL2tj-NhsTBQ0Nk?j4w=_?Q1Eu#jW*Lr!&LFcA7qw*Tds6@sYGhD66A3`8~*C z@malh1fJ3zITI119TqOr_1{$Hb7M@2Tss%}HOsFB>C^uB_|^U3(MS36%nQ#-YxE-1 z?Uy8a{l4?07IWCCx#srdW#%;0w0O*mm%}^R3cR}B>=*d-e*a6hG$TQksNb6#=X|vH zJM?731GiBnKiCdfC1&YRGnANpg^qa?i*)__z#Q?hrBX7}KJJZL7?6>Qt z&)0(%^_$cFL++pCPyF3zI6>IG-oEsFQxXVh7*+g5LGI?R?01d+$9Lx5l$J?Ok0O_Q z`>&IoG*c+c77oDQy>|z9C)t(y^UwcZ9=JzF+l!-vqBB>fzxf5k{_;z$KZdq`yxI47 zSf)~Pb98#=I*xgmBm9-~pqSm#DcKoey2gCYW>ut<7TWyt)yg*P<|z=c&Di(je$c_2 zAQ0$V?$Z0pmCSRC&$RB{{T()|Lmw$W-tutcB=@0Y(e^bK=*A2n1th@k3bUJFpjWjK z>!$|gax>iL*CqR2@|1p*`uw%~q~~>uS{Q`0Oej8n zhlm&`wbQD-uB+h9X&IBn5ObMV7dx)v_bh;ihgF-}_zo~TJ~y_$8m`AOCEMFGx}chx znqYy@re8ODON1ns9`vH_K7Au!SfJS|#R$>TYSca)R7tR}4UivRFbF=iK!&HD>Rwmu zAJlBJdUciQ4+)qfn9$LP#58o+_esG5uM3@zm4pyB@!M?yh0rkC33w2CP)>~6j}-$p zG;p*-&X6uV_o9zjY=&XuyMR+^iy67%*dVH?hXsyel+S`c3(z-w4}x8(EHroz8g8|; z2cx5gYh!zn4Hkx|HrW&LHcC#*UAD)Vd5rpcBZx5v(a6NFs60?V$Jcu@>^+Dh==h!V zGHUEZDYSVUO6D^GixZlEi1l2EITn+JBbm8WXzvco^dS?@i&@2^cnzWPyrZG%mz1$v200U3JoKY*Cd@!v|tDE zkOH4=fe2wYBvmBB`<~5#_$r5685%84?j+;b4w){k@>e9pV9a*aIEv|-#G1O7l)QRq z&RL~jtC*~!HgDk%!j)Kiz< zA}6Q(!+jdUIdc3geG4}g@y@(T!=tqFdw94j8jkOOlSvGnY}gLC^{1ubihE>w5U<~+ zB)3pRI?hQ=d!@;`GsK96&g^vWMi~C&CJCXIAa}0W<&$hQ6|EQs8@WaXIxG{xLl2)C zVP0IHUSMlL%lB6skD41yivuXr+|r5RY123bgT^HR1u{v%W_7B~XHBE}P!L~Yy=C3C z!mdtyAk!G4pnuQFMku+>^HsR~J67R3>nE)F$zr~Q9)uM8?|4&loUU?-f{7-N7e%WZ zxu9>ikuA|)8K5*=^E4y!bmGY&p8#`QLYu5VNgXnM;v%|#ZP?j!r0Lk)6Ba$C7vVNH z+22i<=~-b}nSga^p)9|=0@tLwysq>gZWJ3{R|)&37@smMA6k)Jqub_nZ3^ngP_vH> zS!j=*qN~v&c<0?AuAUB&X4ltiCBXI=(G0FMulg=GWRfQr3`}@4@2x4Q&~`;+s0}hU z7*JdMX4wnETN@dJk9yTk_4NDtK1nBSUsJK!80UQtnxef21w0;0hkMRXOOYEL*WvKW^v2P*HR zKdWqg9H10__swE_Cs(LIEHcg%`a40#px&SpVp^}SV`O<=Z$tU6wPN#wQGQ$HxS2 zew5x@QvIhasSe!uMtKl(L0?2fh=8L1@eRkDH~O!C-WRiWo%h7{;v10v&7=Oef&cq@ zK=DSN=gkjCsHUj9j1_6CnkFE3#W*u1u#X<0JRhiph&jhho;>$zTO&UKpN z+kZG(pC`fVv|hVPtWir1a)O&EH;Xenv(qZ*uBf>)g9O<#Vk*Npuu4#U*z15qyAw6l zTz;Dxis~dMsd^^FOxhD?)}_u`p6i~n)Q$_5Qmyy2D41yRwii{w&wDC_Gx@{PLXh>UByJ_OnB(U)G~qcp+L-8h4Y#rLpujcbhb+m96R z2(4WrX-4G2$l~~Bu}h3Kre_@GM4?pR4J9T(IO=CWf=uJY(5L6?^|rMSN(stDTG#?1 z>Qfuf%zhh=;JFexK6XX*=atr^MM-_sMf#l9sdJyi>`MbvX^5o&tiwk%g>(=r<)7D+z#lQFPQkdcfW@Tj*8%R8f%ajyh1Xi0*~!Uvc$_JAhuw5L7CZmd?kJ zqcNI>C==k&h`pL{;BTCZ?n4wkiu_=1(}1o<&YH!u%FqXn=54Vxk(d!#s0Dty(7-x3 zofBfiQRY!qYEi10-F0D1X3?kD(wXCN*-A=z#t_z7>3R9m{KF;uf-De?$*d9%JI$B0 zLrA==np7kbW`})FuKb=t6UDdB@v z!jS&$eAT#w=HWD0+O_goh&^k=lZK|NNs=!1r_gPfM{sT~|E@KHJsfch=0EWe~op;5TPQ4T?sfw(xUp{%df7OFxgDdk7Fl z);n7U@fltNI3-v`El%VB{RmYWL{oD%GXsVbVv`}^wnd6Wl6o|FM+DSDITF(=zbjEH z94_jTpMuTwBN+X$&X~NI!G^Ne6Zu{jywygG_|A0Xzl24-f)5Y00mHWaggf^N6tA|iwnLxmI)AP9CrPoQb zliP?5r2XfsXUHPx$B%0D)7L(CRd8rxkG8nUgo`;>#3*D|{qk16(UT`vE#y1~lP(`V z_VT0PN#<9a6MCYXx8Hiw1DhrBkVfM12kjaALeK8X6SNEG#7sAR^xQf&f307cY*|xP z)bOGi)Xe_)lY!8uy5H#K=tt{@?37qjy7^!B&6TEz@R+sfEdN@m=dZkfIsD_U?AgF? z;}_~TRIxdGmMlZCMIe4{aoTBxRI@0~2rY1vF z(ZDNMPDj;9u*$2L1lL474C@4Nu8OY873%uHbn~*7b%JtoT@Z8X&q5fEHcZ+2U-jOl z^4;i48xIU%V2D#e3$ojMX2MNnI6am(SiyuDIzZ_tK}N{1C%&?5yt?qgaon?1G*E60 z)%5{p;~3LKwi`qv_o{$~ul7mi{w2OsUN|B=zOh zNz576#bGzPGe%e#fO`AZZP|#UEU>J}?v-l|-3K=h(0x$F*aIm_nd{Y?aEDr~MQ{4B z47XyIXM9hujN^%;DbhIqg18p$niiRjQ-sx9jb3-OeH4`Ti# zAX3<(-sy6;fEY-Q#a2Y42C9=zZ%Y*^+CC^vb+zghFXd`CZ~K|tR?J~ELE`eR=mJo}sYT%CCB3T( ztaQudBYsNl(pG~U@B*?4gv-dFfdkydF_e{cM@OH(wE%(+zLW5T)sfO0C~ZT+6XLxA zj_(k6bX>)IkXvj8B1y2L^*fW4A?X{_5AVk3iZobb8-?U?_t4@PC%00+bOKjf0Fx{7 zsq`~G1OE$fpQR&Bu88%p5?s;HbBd300mHIfhhoi?FP8<&cMVPq@HYtt!^$F8vVg~I zMu_dC=`{{90r%C7)ew>w6Rhh4E@R4;v3?>zUe&Ts7N&A!IGaoA@f?ZD&9s3sl;L3< z937L$dDUY6E%a7>m~vNB7FP$&4Zab%V;2J)lbcg>*JxD~yl@vlJu)4BexfR*S9%GE zvE*X{hen$Wf`$y1vu-7dPP;EpY-P{tqG!k)eqcBg<}{$i3*aI>ihQ6}JstP#g#qru zWW4l+o8wk5-+8nvwEGdP{DG6 zS5*7_L%?eXCd(7l*QQYW9otnTUIQ`WX3!Mt(pzn=x$dwT28Yd*`yhFgSGjnt$}~LD&*nL9sjVYx@YssDukvm01Pyia3YlA zZH?z7mz(zCh)M%sI-q(nY8#_P_D+{<-RmPAufC|&xNQ~B0P=d2Pc8>WJ^IkYBg7Mjw~QXhIt?Nk~Ke#31X>*&|o^jI0PbzKiRT9C2fL>~jc$C~3%fa;h40G+;n5Kmk34SHQzZqXtw-hDz?8ATc7Fard#ReWSE)&s4$ znSJGu0)RHKN(6|G1VCPbg+(y@z@{C6h>4j~BaX414`nuMA@!*dm&=rE`;#XhhGT*KWlDnqxup#!s1>CY-< zh58P$v)$`@!YmU4O#C&e13BN;NUK;^GJfQE!p_psA`pDQVLMB(U1<_Ez|l?m`uM_+ zKW9Fa1Q@J2X;tHA_-(Z)>~rs445-`sMV_rT1Ho7G*Gqm#Kk+jPmFYW)2M~}-YV@uxLi`X zwuVm~YMn_-bxEj*kfZ?HICp=mcDa5*wYUJkyj-w-dWpfBy?6}O@@t~9k2&E#+4h>J89-5VmE;n)|@FKn9Y|>F%{YTxF zk0`sZ7#I`X*DKGyFWYA=zkYZb4S_FrELjhojzIz^th$^2QM%mibM20Vmc@Re#A|>8 z>4u6mSFCezxjk9I1U#>!qwP?z^Vh2ZK=5QZTJ&-kni(U8e0tAdUD8veb^RNRDSS;~ zStekirrCX;{jn#H32)x&tTiMuE=Bz5!_&_lz8C-)O-%$}OO14Xr(K=9;xOeH`}tZv z{swgO%P<12hYEKs4xyVLToC6!xgf4Lmquv*!ms{G_+J4jweN%P>p*e>eK_J^Qyxmt z(}Ac}zZKA_X2(PO%ROFEcZdMu-(6%$BVn0^cH_~?NVVZE%=oUG7o&Z+V}G4iEST7*uj8NinBeo43I>MH@`u7JB5tj-T$K)ghYp|E1W5Ue_IJ{W-gXq|z4qAN> zn!mv#-L|wzOKqj*PdMdP>D?Y)!TjpbL%z#a=_%O-VYFlthYf*|mTfPpzRB_zD-tTi zPG@~P0yNf@IF-KE1*PIHeW-FCPF7=(bMA**mCQy-?e7=cq|-=!W^(aB^}Q&{_5l9G z+20!UJ_tsnc>1tBRQ{#J3A8zy>?kdt%9Z|*Vr{o z4HszxWjelAa~r~tP0^EKtI!5_Akxk*m>M+c01J;YPcYxiT))mJGnRDHfh*59Tz6X} z=zy0uz1-}iBt=0V0yQ>;YRbOLtlm~Hdqm2d=ZRyDxWBTlcPwxBw?eLz4L>IvZoMri z$TTY7UU(N|cxo}Kxio<%q4~wp;=PybI>Shc}G}ynWbu{zgTBi7tt?G462B{9&VSHW*YcY=lirH0?Y>T?q z3=Ab?Sn#bf=bcrrQ_?sh#Si0j{0hZa$#LP!+s@m@*)-`?1-&a!=G*;tfcq zwA_$1h8SAd39fN@taxoK)1^anphX_QaiCvtL!x71sI0Z=xE`ohuj`y&MEJALHp}3E zHi$!6Re6W7PqLjOZqbALob%6hkTgOFLXVr7hjULSm?&CLCE%tK#?XM%6F;0NT@{pe zx@ES$Gqf6qBTLP3t^az}k025ArmVs)sXS6Zr;iX_&os*KGWCQi)Ems1HqOslKP)?@ zd}(saFg$&zpl)R8w`{3CtP++S3wNcflho%*$ z^x@W1PzC{_YS_t^1=n6Z;Ia_b*fqRCw4IwWIER z_vFVe)q0A*n5zC}vgSuek58rDTM~R(AM7m&WT%hvP=66n`SfOd`u-=E1p8#Gp?L~B z`dLGhC7FO7q@_Q=B&Gz?;T`2UlC!Fm-gNH>qtp6T=eLNlZ*SP$R+yY=RcmnTPyM+` zw_jIrY=9%OCyx$S!XRQp2QE~!a~gZ4d6FQUq%IBsF)TQPP#m@WTUyvzBog&op?*^4 zCU-FvL!LT=_o_r-V)J8Sv*E+`q3vD}w8HEq_}EOkaH|{*cZLD81s~A`&QXQ0AkiX$ zwBY$yHNn@83}HKb`P0>KUcG(~-^oh^V#cjk$3#ryouTumho?YVw98C~zFqF~#0K%$D4}ZZs^@RflXnTNTLgi*73> z-78g~#@&y?!V99O#|^wWnS&3Dfd@C{Wn*y2$u$%3Qd>5eq2jN#-=*I$cBtdHM4M84 zICZLiI6v1jA4?z<&qhZB+(E{Z>@s<1@lLs;)o1Y>W!8i!w0qJmz<^G}gp@UQX-{=5 z4BatU0+8mShbUz$_qtC4M^dnIHmWICiveC|oW151t?zQtT0vJCYkhQ$v)+OybULLg zBP0h;x5|g{zgx2YhDYFDK~;HlIiy8w)!@>(lpCaTj+HGw!BVFlS`iJF@~}4+th=;M zhR~RsnWdlFPoqWUo^WQB^2EFcWgK}7l6B32Ez>brqp{v_K~J<@Sr|FrxX`72z@aAi zD)7)dovj6xRR(C4067{V>4phg|XH?Quocm&@JbOIy!P3 z-Cv?e4`2NI+$k~FB>hIwxmkIu?{aK){+i|4wRTi5$og+sGZzPa6=&*r7oQC({e1#l z{4fF5?*GQ@<;OStpN+-Zb>IU8vh3Iud}W9KYDwodNLL=|*;n`c<1_vxNACGQLwmpklk1We2f&^y4G}dr}gu`UM+;v3P2ap@*V`%Tz9ft3@0rT>pLSXQnye6T?Y!oMF_P3 z8^o7?^4}_6kRMEOd=C;E0GKR<^h3h|CW|jwPOjSxI};^acZ{kge(%*v{NH9Qzyx?C z9_@&RF18W>UarID8|t(V6c`d)2CHMp$mUg}uf?P294lE4!T}T(ur=gds~o_8J&XD) z8*H?I!^E67K)>;|oZ5e7WAnXQQx|;@AkX-#oDpAhsS=Ux4M=*rilfdgx^c8IP z%?d%-HjRik44+P?1dc-#$YuIE`d1+5x9(T-zg`a%>ooYjInv6t5TyDUfo;;-cjmTP zSaM|hqi+kUPPuBmu$!{efgnFk66v>5fBbqrqzt@ zJULu{Lg?AXu>7F*XSO!5s{_g}e`$I9spl2`a=fK%e+}SH%HoU+GMZ`XY5j6_ z@BCxmr&OQ5*(`PizZ9$rd;3tM#x91lv+ZwAcG`m9R0;efN7?0fTlV`3yrsAnmg4b_ zt?%ZpOb_kyyS>;AyfsGll1lmQo8LS^zkLut@WH5ZSnNYs=aD31Ee+~lo;R6_zh)HA zdu(u!>aN^D=2<#+BWXF!H+BLVa-bc%ZXBCSy##(jeEY=)Yros;mYdDm4ObvtH&eQ! z{|6cLXUp!)cOBTfwd=rvPcjs-VrAFmnAGK%$anJ-2ITH^fBXG!i`2{Ub&mMWW$9t0 zeN}Sv#PUP5W|Y!`ZZ6i`2|jjjZsg#`-{?@<0-m=8#8`LyHjpqBRt+9UfBRt|@O>T# zP2taK(_S7({pDi|9eeb5{RtaPwE4rW^MUvJizi$`Ci7+27Sgo~L3qUO_T0UPxGfR> zhEjUexg|A>3LPT{vXjCN9xb~0QT)KUdb=AItId?HW^d|W_Qs12Z*^-zhV)ncciO@Y zNurBcxPuK##LviNDsGq+sYq0EJzHwW<<;BsenkLn@8*4{Tap@}B5@AZw z$f?O+Lp=0<+bX(28?4Y@&a)53)p*H1IlKf7?o8@G8x^v_){ZLt zBZqZOEvjwgm1d1K<#l}o)V&82i%*tJ5mf8h*)T6OxE*}Q`Wx#T+OpG?34RcDXczY+ zVq#@5=f#Gl%36wbiN@Azr*3{Q`J_y2rAMow+Zo(_&kkZB0nKa)qV-7RVDjgoLKE~w zRP)+hlw+`D>7{E^#D~@e9s$)nO4T+d>Ly$|??Gct*38Wz@|tnW0VQHyq3b}<@1GW4W_h6T>{d z8zVxNxi+U;9XK^_Y|vz&pA_=Y;pLXC zT}KkF&0e}3#T1JEP!OE4vkjj1P}*#$d=GN9yCp&pShi|*KF0{bq6?zenaY|M)3C7H zn9)!V2z9f7c+lH4Sk(DRzX1zbij0oafx5U-9fE9z1@E_(SELdCaKbH^oyB<>`(ow+ zWWP|7dLzlyATp=KxPKKn(Wes@TxBoTBfyXJmmW~9eYycD-uO)IEb*_5l9WA-+v|+w2j~Ua(d!$| z4d0495KSefPZ_k@GOFpA_5jP0z}o86=}ElZYz%D(1M4L<$z`(=?Vo2TIypN;g^3g6 zvz=A}7Ej%D{`Vjf_Dt;Q!BwU>>6Mnya4;?p0AIwEN;;Zb!GvYGx@;*jwismqg&bTi zD>QjFZQTl9>D`{;Bq*__N_r3KXtxv5zLLo2uwGIg*&U?1BRHH%?BdEgLj}u% zBe1udI@lbk@w!Q%*X3m=>V^)`dZN3sSd-9*_HgEhr>&k zN`r(RlFfv`wmZ}@1yTnZztFA;9*Ob9pWApi?`*Px?H{x-%H$Ky(lvzzNdq*#q>fUX zHVvqtD!LM}j#Fi1v;W;`KNQ4S`^V}?jPNx*pSr@&na?wRq%CQxbF;V9_8B_e@18^L zUiwX=8#X2}9`=$@QXAF*kJ0XS2%ZQmBqf|2iRKkmKSere-)pC94j<=BKr>}uMvJoLIe@LmaB!;#Q**x|Wf z<@CqXlyw%<*S>-Fg0Qro&Dm0P)9 zH{+$f@4?tx4>}o!wW@oIz99?YG-xw12(sZIAA7aVH563SU;Q3L4Idwr89K~9e_~5Z z{qWNln=0{|Z>rtKG`1%&V!?OqM(wG8YJ0F zeC+LHLD%MJYb`c*a%1t*Mf1=OSWLDOV@tm4h0ePU1o3IG_rvSKmC+5B<;t=gT;AY3_1IZ07D zmTb9lMVgmR1XS1c zH09;Z29=skl9l2cT#pGDt0u%2GI%rl8U}Huq3-_7a?Z{~VE{w3HR#zIDI)@w zd$w$NViE^uL2WQG^8grWG)jAD z0K8(I^T}~mWs9@r)BK>HKArpqf%eqno_hRGQjZ5a{*=9}6@f41GJv!w$56w|WT)e_ znE=gdlaSkO9|x}3+*nI&nvZ#`Yy3DD2M0VaItg0vO1fma&2Ez}w$9ov$T==aLdA?k zJ=MaO?(`Y|6I%v}uv^$HL09E=2@ec+2ABpRrCV=wi zqyls47*~I~B2;Hir;Q|Uz^qt!H{lnz;_w~-?*qyi@@l%x`BN=&i)B%B!KiV{3MxCI zb2)OP`?xhgtLm=5<+GAi3Dj$F>&8;#KPua&hdmYiB=ncG&q~c8iy^-~2j1G)A$?8Q zrR}_-y)bzg3s5zd;{eWA`RVje?R{LC5EB0$6k7itq(vjiCK2@6#vb`KLW6B zz5v|ymmg?D{p}*yV!#~sQS)7;(MT#eQp|Q^(i-8|Ug{+9+tjyHT%Qqo{(ad=_G^p@ zy~RFVglnEU!@dg|rs+9@e!7g+dK7?78`!#v`sMqZpI;l`!kbvZn^+=&{0af{ZR7)G zhhRJD`t`|MC+5myH+kg0 zJ^)zW?ZO4S?C6a@xpFOj{mNYW#hNPYt19oCzdw5ZNqiSap48@7pd^FPZ*fT)r*~#i zeM$K#U#=7s!1%iB^NIHLGv@jp?JLM|D^C5_j@k<-i08pwp12Xw0@;}gVs@)gb}PR! z=;x0unDsd8aR{pcOICDlRUuw7DX7-xt~Mp3B!{=946!u3pT=pfpi8@foenm?0aPhw z$B@A4#-By^Cv935k4_KTNF*P{S9sRt3!D@9wfu9pUc9@oS^+T{D`j-oK6nxq&P+C0 z_gF8()|G0Hx=p-PDE)2!_g_15zsYC;%?UPIS*&}y`0FHE?I)uOX*X_=xe~iQVNtAlq9f+k47SRU}aP?&6G57UHMli+VHOY-4XxnrF`o ztfXwXgO1sJDwGgT~F$6v)~bBF9AR0GcyXkGXKjSL+ydfgS(=&|#BKjZ4| zI+$(VFLpZ7`AB-b40E(-Rm8SU7A0=8{qs#D%CW3Y8!Pz8=q`anYUC_d<@V}Mc zN$stZT0PeJXRpTjTAP?9(!rm0z5@9<)OHU-uI_7n{ILdmV7SWcEVE+i0w0eTzsQ|; z+5gLjX8HFA!`%g#95YzFpp`R~ z*|ZV>CAR;-9{uO$SrW0(d7VdVa;DC3rvx=>z4>^M{dg(6`bJ~qeZSePHifxq-W}b} zD)vS7&uhl;Q_}4-#Lv>$5es%tg#RX?QNUqrUJ>Wp-{6KgC8fU>-dFzk6^Xe zTnPcxUtiZoX{4WO)1T9uCsu?CG-t+z&KJ*kFpp|FPv-D-UfU?8cla!;?bN-EC|oVg z2tP)A&NM09R&WTJ(KZY1>(Gj_dmDP3oVDVY^n|h^fC)gMU2IYpO z%UucM&$lfA2Ja1l$ZKCS$RewClJu0b>R8T%jFE|e?Wn(`?A1-DwIL_yU!NiPRmv^% zzbj;k=6_$3`oR3&h%_aKqUx^|{#s%%HP>RpO7O!3Rr3*P^w*5E(xXc0LxDn)o#A#e zjVAH+FeUG6v%1C*=p49WU^yLV_lS?ZR2WFL$}d!|JO5X|<^I)ac>n*V3+Mlt{{F|p zJwAV5#&eAhGr}J^^~?4ky{w|~v5oe=wBDJ%?1pB{n4giZEI%D@0xydw95;Foa{AbD zGIg?bO2J6f|Kjb<jrasjUi0AfbhtKmtJo3?U3v0;mKih7d>?1Y{-*GLL>Yb~run)As$&`90tF46*h=Rkv00xJn)YQKRN%Lu?X*!l1FHrY4ZtL1mBoz~D=Bdp`XQfhE6T zcpBoJXlnl(ViSJF_QW5X$@Ty2;@_Qf<+T#PO!4e^Wt7yN>pkU@qgJ?>r=zIqJH2{e zxfD3cCM{R&-U$Y2NH~|W+#-%@{c>a~W}ZeILNjQ9U#0nYy&Y8(H;%HAmOkNM-TTH^ZWEqOz7r&E8oS9O`~9piJ^ z&wIO--MpHDbIMRh`>LF_%wpPNHtvFJW3Um5`}2E~Y1=^#Pk5`q>9VpckFCo|~14kAFH9Jn(d;+S40^PVVv!?34ceo^=Ot;wrCW!NJlOspF}^8`v_*CfBzDuZrViMc2#F#niFD_}T;a zJA@VEL~^G3-~ZU>Z?ii7!XEzD3I6A>>^~=kg#Ri){F%8z2fqoq>?L^$nJg`E{Vh}# zJbDV>pYeP{V%j`)qCqn|V0N?uot87t)g!T%FX&#Lmke?=${%p95`T@8kc>| zESV4K^{U z)+q+t)Vx;sE#wXgMp<603ZZ2GaUd^PVf}fu{@*F1{C|Jk@K-ePe~UnV_~!S?(V`RN zS6i*H1+Cf^`&P(7{(eifhuaO*~llR)xaTW2# z|M2G63*CRB%l}OCe{%K4?~+Mk+8dAb{;o%2;hTL@?b_A~iy*dT9}5>^PoFaitB4=? zfXApFYk|77-d~W7I=^&mO9mDF;YcfO^Q^_626TVHx~cz{2r~4N{X2 z<|tQUb7H6M3WuT#i*D@_LDv5a1pb5w{&xleuJGRtIDhj%*J!V3a=ydPzVE(2ev+dx4TPlF%^Vg|>XY%rtut&9e>++>SOd(F2n;r5!#z=wQ6u4~r# zrG4tmmc@$7os-!*IC}A((1oor(OORXj#gV*a&fYp@9sQ_bcbu&k`rO+um<|?wvE?_6{^MZG*ZsXZQ5M-z%)@LvH-oG{t+k?Kk7r zk^eU4&@Iv0?a;0BMGbzi+4GBd6H*5}m5-Ckgv9SS~jtW6p^{dH01YVlA`!}%7m(9D62ioyv^I%3B zbK&%x-wkRH_NXOW06l+lFnp~~w!r@o%pHl2X#V-KH~)4WhV5QHvfCpZ(4!}Zci#{K zJZlu;1OY)rc77J)4Ox#o`NztJ=T|tc6A~OXWQ?+JLYnfyZm$U(9SZ^_)oA4f(1z?? zp^W%$ddS;kRU!;2eu`5>NIhY1aD-5ppIV1mb{ONnk>GAW-$>Ah2dmNN@JOKy3kqd| zFyFBEb9%69KRYKXz2#6Q7tIOBdh(QJ(*i#~VlHV@Xro)F#h5 zUv;thtP@=GpySe>r8iPZ`5@w8$3MWozNN!vx|d9QlmTeG9QYglLzUguul0YeB3s~b zM{u9~k+uBMznKddAcbbp$a?RtISi$? z1b~%bmWi<33iU{*C#;}V$NH@aS@P;HZ5Z~VGga6u^Xa{~pRu5YzofEzry9__sJAkJ zPX3?B`6F5U3s?fd&dDbLur%9&9_=FR%?F*bqpgcR!mH+GPHELbJAQ?Ylaen(KfmAP zhzNk)t}-gx7tVuie*8^OFPnGtenDck)xyIa4Mu};#{-{66G3qGY0-W{a{zZa?AUA* zLMFx0M(Um)$^<$M*!#x-{KG}Y9V%NVhLgfo*cxR!t>Vz}MA__-`d`Ij^DiS8Ak_r1 zUW3etchz?7zhx1?$uB_I1=7C($`H?6+#nIk3q-! zg4CmpZ)s{?2XdkFYiwn9W=(iMqd2SogCm`lJ0Amrw^E8wceb|O1vKqfu?&$%?o)8@ z&044@4`?1j(Xp*uePA(8EGps0&E4yEU)ErRpuB+iD5>;ul|qznaZq}EGzSZ*l-oHR zSn}aH)#%iW0WILMaD;t^!1cMV!~O@&Ee;Tg>?4G#c=9Zx{|QTbsCPpPpYaA9yY^|- zZxeoc_TX=AQHrSeaM!ll{#{N0&-}mP7eIXZg5QgGWW8tC)F8-noA#c09-4=sNMZxJ z4I;KMAO&rIy8NphUabIHNZ_GHl~2)GMJyfN)y7H1>RV|-BZj4)7g+Ly?KhBv?*6wGaauI|#@ z*jz`n`gLwX7bfsZCw}X$4}lG^+u$!uW7WBMV($H4-de#|Ye_r=sC`#9|7P!k-(1XJ z1LY0n>{8F~U=mEE6<+p zeza>nyUI>Z=-E=wEl02DH}P+G9z{fHg{)#5oGo|hz-pI0;|8>Jn?C%7sGtvirK~2r zUIFn}+Lmf*<{JsGdSxH(wlHq;Y#K6#cT^=e^#EtF&WjK>f|hhmX6V-E+!6e!Fxqyn z_su-Np!CKU^9u;m_|nZvSs~3`09O-()r3OF*pTlx3h)^#d;Y+E|CMb2-KnqbM4>bd z)^aW5?cg9>KUJH%bT&C{N&OLXf#0^r3h^l7jj_;HG>^E}Mkm3DYWkVz^1|d>{+X!}64S)RA@KL$~ zs9gP%U-8ESi2Y6I@IQU`pHtJX$ov_4Ka&v<1N=%aW>>bBM&`}@mXAl0TIFJW z_@j+CybWAKLKEkpiAHdnPr$qUl=4A~wFP8(3iI4T+0n0Q5Z3tIKM`xewzoJN< ze4ZX`CEj)0jc?p`@Us#h(lBM-rK8_MOFOxGVb8CZ&fQGMD1v-s<)m_4^)w^x8~2u3 zV$$j!z(N3ni$Y5+r2cg9nw7vs1MGoXauG%3pP1P0zqk#c&~w);m0}{=$RV-2e&S8s zwJt|_9^hrnJnV)X@iT8a+6NqpD~Ck&n?Wi$A#swlasBtA4_>=cnUMb4!3ZWObV zSAP6I<0#SEw`BU}hi~RqfgcljXi4Pvl?|K{xKm$?lDqVgjLoo;MVNJvv?B)3%%L}* zy+>|Y$Z;Aa)ALTk!)VZsDmr$J_*&ubJ(|9TkG48I1FlSIXpOJmo$=5o^UfXXZ)1p+ z&MhcHYT@>$GE?o&KT5=q8~Zw9@%bnZ@ESU9hs(#OAQ;vtts)t-3H+R{MV7Hqapf8s zUpn`JBm?cH{n((QG_QwytE7-p4n$}lK9Kuhnjcxus9Bv$h!W^n@SWsisCgS_Pa+!$XT zNZEiCs9%O548)ILp1=0i%cg626@F7&72@dzl`lf8l*AIG`KViczkH>?YSceS9JOt1 z6Y((E6dN0uU%t+BXaVYytv8$EOB-kxP#q?x^X&?nRO|hNR9!U2S<n}cHjTBgvK{>BUOJlJyW z!rUqT?J;tm%cqn+DrP7`+~9}+xx4f$f|VkauqEk;T1UA=^xhmTFe36cspo`-8!Wjc z>U`ex8ZVQyT?rK?cXxu+6C(6pE1YQ#i0Z4~u~1D;w^uuY!T132-6_V0$L&8xZ|zNg z-jhC}W-9hcfZAnjMEKBI$?r~6<{S;eg$@7c9rpXW1AB5a|GCwy(SQ*jUK@W`qqvQc zpNq+Sb@sU>)K%GRW@<~?=02IodfAyBfN^%(7)~h8e68RL-IKiUE-%!og%h2MFe zS(f?sGN|HXmlwYp>+bJ^(3#eZNI5EqV}0@ch@)o7KZq~9qMx{&2lI5fvKZO(OrILU zjxVICn~I=Vxd`&pc3g^XhOwpH;^{%lrXc-{={lzv88N??Oj#et<$k+;bOd|rep@7{ zy-DKZ*yh{igx9AQZflQuo|0-`ekZc+%2fTn5KpL|8blCs9Ih1IHS5KfI4&>ZwvRs? z2n;H~O}chR+r$cQrAv$oBC}_Qw*8upx`=Q(v`NA|bm=Vf)y>Dff~#4!xt-9ByER6g zwaItQ1dXY>iB5ga@4gDHt8HsQR6#q|14rEyCaezM?5cGWmRqg99~qbF{S`y*G0zKc zn3-beQHW&XfF6_044Zd_k;p6?+KN3!KSsNdq zik!{CJt{m?0T$-&vY^=z6ZF)mrD*c>Y*B&Upa~VII57H#u9#(QF<92u>sbG^6pd4l0n zgPT0^MB(b$^UZC(CP2ZCy>)II+T^_v`T@%(KXq6C37pneqX{%plI`Pq8-9z(Xw^(wSr)CWQnoo z+*N1mr}xzmrlS&c13n@={$;1gecMfaf{>hizT&k)zr7kuClbH);){#F2f-eth=ETN zUBorF5z|FO&4ec84U$ffCa^?Z$2?9E*ezeET|!zpoq>gOmQYr#yDoKWgXwSdMe#4&-L939v`%Q^Qd28$|Bq+ z{Ju?w(ol5Yb@3U=Sc#Ft0lATz#1!f3*phV*gnT*;>%_38yS$oAMT!u2G2oq(@x8J* z+d&>-NaMT(qc}G*svqJo${2OioLG-+`LH`a=i|oF<4DE8rrc$*$~rHhYkf(6LW(C) zWJ#+%@9SuO@?zP%hV43GiwVg1F@Sjm6HFv8>&N5hm_gIn5N2^GCOG@mwxS>0>Dg>& zYAdU;3A-#lDkBdBHJ-BgsKYFO@6|PEgWF`uJ8_3(wbJs~<3Y6AWuaJ61QvC=J3IgL z$I+e=orvrU%>mPl9YrOSK|#?Ji}W4H^<$a+y~aeCceokK?6tzZqi6k#AHc!{rAZl2 z7WK>5_hz>;Xp;%vBzyEhv z*S4Cg+3JC#o>!a_!aG&+{{m@bh6ss_fDXr?jqYmO$A4aTbs1M2qo8NZ3&~s2G880h~m% zNlC3|JH$`hyM58;!6?qO9;n5&-Y8T`M(GWLKw<|!dJH(cg+3=~Dw z;n4iB)GCRoa7!JsS;+C$IC)7Gm-Aw8*terGPAUm1=%A#!s%MNER2aJj^88SPZvXg= zxT8X$P*xU48Az!u9;EHNJ=^*CG|6F$Pyd#g{Ch+weDkVVVm^??5WGsAVKvY$|NCQ`)>dNeT?f?UXp|3A*`m@t$4iLu2#yaurq~ zW6771F}{MMXWhEM`TSV1ps#DabS|6(#0^RayKYfCC#4%c`y(edt%m*8M%EtmPH=+d zb5H4do-B7Rq7kH7=;Q3_>?k;G{@!+;$KiH^BT>285pK$c5?6%#?zTbJmZRXC6@(#s z`_Z12ZX+qraZP_*-nq%}_Ir1D)@MlL(-b8iXUGB5-G~(qLpZwZyi52JXNbrj-H6a? z8ds(DaBnucwNvKAc2j}25F6(?_Z@9SQAMlgp(Gjtk!JM}(sfnEkGyPaX9IP8)Zm<< zGh#F_;_!Z55UOUfGwx*n9I$#=rIqGCUAc91X<2xJE%z$_HZ<&*p$OAo8|?1p1geBX zoopOqwF};|9MaCcUd7$CPA>Xv>%+$Q98I%jH}>iqMpsS;Mds%X7SLGyP59v3Va*q+x|50?$W@uA1EaRM%FYhTRJTsYmSlQN zjKC3JGln|8*+-A*Sh^$%vTU0a7SKkNEP`^)RvDLN1rEuRjG~NHtcy-qVwq#;#^S-? zNI#uw>ecGYEUVS9h1f#QC3Usi3}*X+W$m8|qf(Fs++Ok|Q>k8}{JIyoKy$Rv6DeuU5t&rS4_oxnZ-1Q?qNyuJqP9h5jT88y$ z<8hN?j%(TtLpjiw-*=`UR~r^8lkjiZ{F{yM@hyfV>NtNr5}lruC^Y&&*v|t?DxKY9 z$KthzuDJQ@8?0Wr%2+mM6s7gFnorBN_-+2aXAY+^qr#EReJE|Kj_4~c2&Q3wZfv5F zM!Y^S7R)Rr4~DYvlT$d`*4tw==7A}S1~&IaWLJ!=&i5Gx-adzQgvP9nFw_^Xd!AQu zQJ$Rk&?6LFHAULs$2gSN3SK!FYG4rahxeKLsyN^Mh~9EOTva#8Cwmgpb@X4(S>Oot zE8-0|ut?F)4%6D~n?!t$JHzuh|0O@8qrz_Tkt{9RR3rkm=O0|t9-oNuvlgl1Pb{H= zZUPRp0%bs+ph#EE9iA6weP!>{=*h#0W!d49s4k(5#R^0b$x;+$E+!`CZ>7OhytODV z^0~+2R&y;TG?RbMSk+w&If`@_%kgu|iTkJ19~P{>x2#?VtTFr%X=iVgzXNLcU|9Q3 zbB5I6GYRcNWXr|*o*AVX4U;zI`PCw-kvGSJ!3yDgtiJEISflXK3_X@*J)_CAm2l4=SyN!=FFC$h93`Q=((sm~8Xc*oj^Iv*~_uG|>C6HGqHG^)r6uJ(@g zIZ2lGvm>ian>UxvE{Q>)wXczxC$H9IYQDUS-d3WaVtvm2jfJjrlJ&_aDTDzXEt^F( z!O|C(hO)mebD!`jv*TlO@sZ{!RSfHb8dNAdVo!Zs&5hlZSCqPVy)*q;Y&?7;H4e%(cVhU1NvmJxK}AbnTxm z#P)d{YbvotnDx+DL<|bM8duIJCtKoW;~EdQ#8y z%$yT@Rz$?3!YG56^TW92@rkazbnlU7@?+oj*$^*GM4%vgZm86ov|fo7E}}>wmnWwv zUw^S>|GYX}jZu0j7V4+1gzkSfot85mKTflmnK8nKL5p~ zadT}mNzHZBW>hR;ypM~XaW0+{Rf^rA!j|hS%~UVAYL?m&28t8U|n6<$x(&G09q`$0ZqRc(E^zpa+e`mzf!@HQeNq(xM@b-3!z zPyyY;>fRG3+jxAAdrSPjJWrvsr&RO%$qI6dg=QviI(I3hVn9oeIV81C#7stf^i)1B zfk;g@L6ko36k?dyAVs)dPucg0wX#~IRN64nZ&x_)nG2`%E!(VTR8LfCj*dG=cGfP| zqtXsNora2YoXZx|2SJoJ8!?gb(h2OdI!qx_GPWSK3?4)G@lVs?Zg)aY+**mq^(TdQ zDhj!Z@>%f!gFOF4tNxvFuQ6CzG)8Sac4J{>-d*f*Y&qHt;c6aAAy6izp)B3dsW}Uu z;yD=o7y@P{MBvp>bfPSFMdSc?`|(7MS$maVxL{_i$unhA18Zr6je&~Xn}I$3w8#Yd zj*MIT!>bc>ZkN+keA0}0nY`GAg*M4!>IrHgNOIhzu zgj3%+jecoyCaTj6t_W7!GLn23Yjr17-8IPmafns6j+}~S81|1X>gY2uG+M&GFGn6m zX7@Ctwv1lDG#x80Xg*w5_fIV?Eh|wQn~|6Rjnz!nTu^f#tr^ZZB7mB)z}%WMJ^V?1hy-BlLvFU~Zx$u&v;?EawNr?-GO znXx)i*mzWcb?_*Bku;Nf#3Q0{>$QUE<1x$F@lF)5qRL%Xr{w7DN(u~w->cCpAQX_| zDn&r}IFV}jaO&;wA4^jbE?nZy_m-N<+M_lK(xFbagYKEj33z5KnLN*nRJ+mZQPnZg zw!q^A-Q*t&tq_)j+oTqglWV@px0agypIwJvyjpk67@o`;rk9RW{3Ryz$OXbO;AOoQ zlOm_E@#zS0hcu4Av;CE4b!lUrTESJ18Cm;cUNV+6@%Um96@|`UM1a%;DJ32M{Kw11 z?;wVncLF2V+oY2%DEH2KwWQ&uqjO+IhcV!tpyiAMO~N!c-)_lOFBd$eeOaQ6ahe*o z+3KNRacoJ*h1kDc_EhJsb1?0+6LEwA^q%uYrT|CGP-!P|>_qnv+=f5g{ z#(~+zCHf4$C;&E58V$K-Uu zY(khFFQ=>~e-2{2*N^&S zLA5@D#ArSyjDTz)T;r)h4eo32kMA{^5qU)#fQN4d-C`fOKEU@>!^X5FooFY{HJQ$d zhA6Tk_0~Js^@_(QWN3y?O{w>QZp79z-+kw<8tAky(UG~s6+&s-4C+{#xYJS8#z1GR zWIR_sbvnzk>x88sn5T2N1f>%p;~`imt>Oa80B(?z=!?0_e6?da7N4>wt}-HAn)r6&jXjpEV52Rcba0Wgg0^0;!w1bP0vwN^nSLgQMVC)ZE{Wjrx*+S}QYM$(~Ot_xfJrc{sfF zC6P;P5<5aW!}8US%IGyrP6)2yZW5p3r+GH4J{U1%+J3KbxLgG43z(({BRWXd_Aym? zfv1O;>LH3Z`LJo_`$!=4AP(p=*wJ9H{r)e{CelnFx|Q%geo2k6e=KXdSY92rgoL@Z zQ_;)l6^>3kixD4~AUgC~;bZxyB{4!6A-Sf_ zJG#R!D>55j`j8(SY_Z@?+u9#EK&VURcM3_T81iU{&03~2`5^3sar4wc7!g zDS2%7eCtEu@~z|D-w*4gPzIU;i_-1{g!f6qJMF-V5XGZA!n32#-qY4>_EozZ7=N`Y zKCjIT4>9wbTnxAvsD1GhY5sK;IXIPBZ73rXGDDTIlPpYY0%nHFEPO`h%Z)os@Uw+h(3?I!dwYUt!xgvNf^5M|+f#RA{g zU$mZ-e*L4J!GEF#djrA&k6Of*ZZI2bmy*}#9FRv$T{N98Ls1%|ZV9GMo}5_+Q2k=m zxOu!*{7S}n#Gu_gR^#urQ~TBwF3TdV#%4E(r+fEymE+YKIFx?o007| zI(CF{|5Gi6_l(9V8g6v2iV%y|nV|4qo^rSdM;oS&u!5S8JSwR;(Q&e~x)uY=?h}@G zp8?7g112)Y(6aEAIRREa*!w}K{re+kw~m?Vf)=klAG1;2zNd1*Xi*24!c??z8LUwM zd@9h2?|g1xU;DDmoh}Q?8#D-bidE3BVey%!MMwH zHyiwI!Wr}!mwq(?6bg{YV?hw7Nfo2DvruJ@=GY&;@~mOX2Qgr>aNf8z3a!6l_bt+F zuo*~y$P44LQZEyTlhb!r>7I2dTFpcEE2BAmkNdb73XRLUO#ad%`OtjBTV)T@8!d#< zS6U7rYQsH>8A3smTP31)Zg3LMinDonVm8AmZ%$)>;rHEZA>Tya)_+=lJX)vqKJT5H zNHgDtj21_z1mP|Scg6xW{LR}*;h7OE%c@2om#fOg1VCo8%lQITnf zpRD=${>6Q$Jo69QH7FE@xaKmE&Ld>xBZr?x+ZsjO8G_gz>Tquil*4gizvJkP!ODNZ zKtW;O0mwJ}a^2-X z5+~Z;yz%n%+e&Opr8}cuhqcX~_dX1}*>=W*)!>9TZt^Pz251o=MPS?L^W3NuaZj87 z_kla1{;Z&@_@s#V6)Ac}r-Fuf;9@tbgz>bl~;}$g= zfJDJ?^3K~6n;SNtNgs_&GV+p5fb8<+On}xwM`KqAUt%@ROaVERe@v$vA6%jv)vC94 zDv|@kg#wYHTAS^}B!Ltk5^=cWo)5jUHn>mb)gFE>7}UV2^RTL&>yZ^$EX<9>oXGhH zs>jHDD!11NuF6QrXh=Bl6*mgU#h_kBbC9nU7<$Ypd^*m#MQBD{^uEA05)3dC=;c*M z7tz2zrY=j4K#FK?UmTEfEipMkTJdK?ysc7;eaD6jNDS6+UkbH&lD4HN(ohcW;$_t4 zz{cX5dIr+>elV+3wBY6_eyz}WZyrv#JAI`=`MPkvDRDC=_Kt zH+JqkHw?-uS#%z^GK5rlZoRW*26;i9paxC9?fjGt8!pfLuxVPaC4LY&ZEYMBsaoZ+ z?J-)+L8qAt5F1C&iHy(b{u&8~q-QRmba%4^$wC?AS49s8A5&ob*BPmRG0k zC*57>yw8}UShEIo@!y8QZ|#5e{S)wVKol+Ht{VdRD&d={J}I7h zc^udSm3vJ;Gk@3z;nynkMy3zwDrAK&~J-AIIez;Wi-G|BXA6=6h&=7 zO<7P`_(srG++iriD1g*7eqqt9%6xW=i!KmC!`q2vG+(#j!OO(CnLOXKH#Tzju_{sf zG;cO`-sJS#7p5F?Iw8FxfyM}&*Ao`z`B-oHPw(p923F<+gWeI=l9(BbJ*4`b3C*ym zB5`p~T8p0mRRu#%)FBj|AZ9Q>2|6%-yun1iptffyvngQS)4RIXcN%vFpeT+Hr?i+s zo)dm|2v)S1c?NAj4RHxfDU7MtC|aHcW`$kZ&D(1e02gl z+@Fz;LyqoEzz7M2~k4;H-M&B<$l2qe&p<+tnC;&OAH6B?gCM4X$S+~9r^d?Cku z)H+q2)Z!%=g)0W8DT$6E!2Z@&GGI$C#*Z8GqLW?g$KZrUbg7fom_C1N6F;V7RT&*< zRRa0aw<<)I|jI2^en zWfbMVecI@V>aC`sV!=@4v4|cvBR6}BR7{?q`=C4Qkl{B|wGX0ZR^K{j)V$f*;N&p{ z$C?L}&3uCwSnGx5lpJyG3QV|{3HF&DevRd<2^I{tIAgweVX_ytU{XOk^rF=p=tS% zpKhe0Nmrcs!jaU~Vjh79iBKauB}xrN>RoV;ZII?Y7RAT)M# zF;+#yEOI)oS;y;T0+A*5i#VIh8zIthYf|RYVuUnf1OHkfWF%^*%-VT*tZ>7rY~v#N zip1$zTo(cs(}8erUvPljS;9-hm6-RbNSA=9(}%WTTlU9wPL+Y&5k7M^OF~BrrREnI z)s2Ea`){jMUEE-n+h>WH;dyrUQp4yPR@rL>!rbb{aSmwWJ(E7{g$B=&nUQ59oh10? z`AV@$X#SwcuV3TcsZ_bx_I=UuG`A5*A1ajAD;S>LxEIsQ$LKL|GAXdr95(jC+|=R^ z4Ss0XA1(lvvp>FV+M*!o3pqd6)z!R+R0#us3fX|cpi1dEQQ zAiSF(5XDY)O%ul){8D=NMHokg=7g0MD?~KN^AMkKd`=~0N6A_&2$|bO6&d8J@ zS8AGmIsKAnU%q10C|^aU#3oh#*1a>wAugTYjB*!{^Gm**86=Sk zY*9@c=Ml1&_U_PDCNKTo(%&N0+-?21xn+N>j4~AMht12f9$weui9YZ*WK@&qzzWX?AWQsy`9| z(a^hb@|lBIK*fo7X#1EPIC;!-dg+{2u44NbTv2{`z-G}V;st>=$Qsub7W7A-+lL6L zzG5nrqYRkS8HL;G-uu4oe3WT)vCO~G#X(ybzCo)Em$@NiijXR}BB&E=6EVd2)Ff1< z;-WF%av($dM@Z5fLt z%>9ie=MV%w8W^77`aCNX6@9|p5oYoM0)af(JhP?vDDc6K@+B3YD+t)e)!Io0r= z=g;8W`2H#VWrK&)vIdwNP)&BpY1=O*XS&UKDb0VWUy^$Kh`~?QJ}5PVW|RsGjMU@s zn)(>UM22|=ud?M6?SmE9^FDnReF9^nEAdP`G!u+P#Pee$?Pc-YNeVG+?QWrEZuaH$ z4t3hX-=a^~h^uqgd|5lsfhjdSc(6B!ni2eUfZ4g+=1ZQ_+|q!Bm?kp`DwH&=6&P

9 z3Q4~KY}URA5mh?kw+=n|y{#YgApo-mL9t|Mkl-mWUc8{G>Ab(XA@7n`KRos6;sxH? z6_~H7V>BaN40!>}+047N9!M2P8&cDqa#v=1$pg7}5Q@;2Mn0_6^wwGSgF(-6@_Ug- zn|GK|cBQTGAx6>ExvE!*&q7(OGc{%Q<+oTVJksPu*1mI{!&7_tcRWffijsmGb39H@Jvyx`SLuu7ui#n8rcTBYl?6H+uq#$+<+m@DBAfBfo!(FY3_ zT0g>bUiO`>`06p-T?}KJB49QhKJqqt&gUZ-YIm>B5xp|^>GvvH@#RYJ)WXo6XZ0CV zQ&|TMuNOtCeKLAsta)#HXj+RWb5DGP_i~lH07yaEEEdxc}G0ZW^ao0aNk51Iz^r=j26ndX6vPZq6^y{_w`#D{ zEmN5oQYc6F^7eB=?9In@@chWbbzZlp63zX6O*>uGXfU{MX)$@F0>~!V@Q-~tWPY|_ z5lF4{wOR-Ua?QM`VB}?11|uj=O4Vk5uk|-Yc4}WBfkM|qh+nX03+wk!wuR^Uw->_D zlTQl|5evXciuTp)_?md;BV>Ek^0_6Ub3H4}(9)U97O8&k1zGdg~eUscQ8ZrM`!Q|PQ1w|57YgR2~<&LX0 zd3!T_v%ikS|N5*?-g5Kal|1eb6=4h4+a5e+wH=G-Fvqwlc7hrv;Xo<($&%JHkAr)@ zScskR%`?{^P*|hVVQQeQjj+a}Zb;{Lx5~4M6R_O1gUy_2r5j`hYrM|PQ6T8ZD)6CI$wnCa@7 zceRQ6JB!v9D~9W0CLv~P^^;($`r20PEnTHdrh0Z07$vSLTGj`w%c|3KZgB2=^&Tac zpHeNk0-q?0M}AO7XNUn3A)_5k(0I>5drLKHB0Q;qoMYRkuzXuZRbEL>K9C zeMb#Cob{?M@-Pb*#i7U`LeQB$W67bf}{vG25t>>x1pwX7q_A@@G= z=3z!{rx7AwJGVC4$TsL;RNQm0XxpM_{I$Yv#QXOPdp^HtiuzR5+?Uuie5FasSzJKC z5eedl3w2-+*ZIri8D@e~2z@HO#q~mipR(F@p7|tG3F@7@8G|qttmzhp0=@xv2$?&| z)A=+_2=Q^^t0dsi6zVu8Ms5chqD)Q9*R(ebd1QB67PkZ$`aeRcYp&Iv1a^c87q=6M zL|LO`1bf9ock5gBk|cNP^faX08S2H0($XEG3~rn68+n;_Lm>KQoR@){;@)fuoJa$M zJ0Vb4>!79+6Ir)TS~VPoWnYX+tZ(+s9x@EL|4?tty9#4Y3R;lZtV`<^-M+ zyrGI#Ge+Ccu}xsg73o0X7Wq+#=W(6>TBo|PzGe}oXo5w|I@&yV#yUsEGWSMjPsGH` zLn>H$T~U4|pa7o+w?iO7J9k#MV7mJ&4_;i{-yJVg3(+iEfLQ1i$fc+=gVS6GOx8Q# z&O?mS7j$Y@&282EpbF?zyRMQC5@Bv6Bk4)CtgX7BtF?yy%W=>B%YGhGyeU0r;DKSY> z{kpFW61Ev$>s9W$ngawBhQNaTj<_R}h7OcS*MGQ`REalZuTaJQYX>XDfm${Ai2>{3 z{DaG4`;YGmILBYstSVls%N>b6ZTn)9lN))HLm-Za*LGMF=dFK;ihT|i)6tf|KF7lV z4&JV~S$CZQ6mpmeZQ;=9%Umb-B}aULRy-}55Wjfxr=^P+R97H7J$QoaDCM@3v}{-` zdXdx`H%(vsw6Ge&jCok_c$|DtV$(oqgu$Iu(z>S#@QhH}@Tsf=b!gixgDxxBOqt~) zS<{fwsykBMZ$A<6KJVK1`Z??&zW_!$Ds0a@yJvWYoHfis=`uME>z^cEXdbH4Rj*`t zedEJaBK|1B7)wxil8#aI7(Fv93<~YIjt>+>r93U>BbPADj;ktYW?Z?`6Y`oxMnm_> zK1*yP%&b74wd0Li-oieMs~TP~i=GYN@st&i*WT)X@o(UGU!t#WmEQ7FiOdiO;rlT#G^!eI0zN&4s=%UPlAm<_I$W;UGtpS^276%)x0aP6)b1G9)h{wZ@l!bM! z)jA2yO-NdS@{PD`7AGF?iwiMxT`lh{+aGbioOp6m#oNgj4V$~mJ@m(X3!Ca?&{$^dzDv!4^VLwpSfQUzTe=;n zowa+UAK5=;afQZOriAiYBS5&Tx8$fg6BMZl&|Sxm+yHO}JaC0KL#y(pXB zhmoW-^UFk1XeMdE>jq7Sn{p#pT zFI;#t%-N+tjGt^>sz1hCrtdWpzy=Q{RV9M=PaEg)58bNx>PPXdDzq5v#P*nCq64i9 zv_Q~Z+<-s;a0nAX3Y!D0-$wOJ*1sECa^2xkh$VEyj-Rp*XYA%Qy)s5(IJAKrp<6HH ziB8=V55lw4V^;S{U{l8ajJBh_-1bR2k1|CdyjG})_y*OB!DjagO-g)A{Eo@N2qhVt zrskUZ$X+kR>e+qOX5R!O;XH1E?7bw958%l@FgV}=14r3~hIKun%+ zBsOi9=f4q7&O0_Q?M&{AmBcsdcCH(4wU{=^7&Vvk-5M7gD(n3Ej|yV9PPN`P`aDZ8 z99@T{dQ%PBIBk~6SexicDucz&Iw}-6r&6c;zIbux$Iwr{j4wY9oO2bvFt66TieEw}eHO?w`6!pDA(h;c&~vuLc9&dND; z)3uMR>@8bOhtI5@Mj+fpQN#k2^{|fl*NC0BLiehgSxiXS;7s#+&7 zLh9^dmJCg?Ktyb6g#6xQ#$%6uQ&8lw1&bt)Vu2VcbdkkGhi0Z;VJ2UR=st4GLMdq9 z4L+&Kzo042y$xyP!_otmmGCI^qE9X`|2b~!*JWYK$q}SDhq9HN%gQDCB-B#G3(6FI zS$&5mRYctvt@*k#7AVKe<(rm?z91)TFH5T?P)D~gmOO{=q}kGzZi1_<1`3%CoylD( zJNNP1O28R>+TUDwW=a14Z zoHJ*d$W_lfDp=Dy>XWo^C?YRFp9!o; zht(oTyqrK=MLUEusVL`crYprqZ0akfY{zNu6g3?S_pU%|<>Dqup;>P`hQE8{nsV6CYlZh0 zVejN!l)L%q)Z>&jVWo^2#|Riu#!rne1lGhpzJGK};o7&nxGyLY~t zdO7LsuFPdsN|w8C&wO9u_+y9b0*%v!SF_$G_ivt&72|XE{9C=tA&J}n{xvP%BPc5{ z4LDg+X1{57-m|-)HsH^R{J<6>@Cd!fv-)3(?SHwupLa*~)C{Y_DXA5}U2_sA7iE8a zdMfML=kiy;p-2JO3hi$@s!!P)?3|t3etPpR;0j{bz_*!SJ-Zf_zt4>+PXDgH$IYC* z?ZDOkBT2eZhlBm?R$m1!t_Xg3e|6gYm=|a5*L=Eh;Q0*8ogzMGS8sbhyX?x3%+((` zp50eJGb^uZ?%%+9cMrO}_mj(BrlfrE!QYR93RjEj{xiImiM}Q^ojGq+(do^xX0JMK z&04m4_hrHV4BO2AzWkjr<91%#YR{K<=KG|4`?O{{vzst){$5$rQ2S@AR=!+SQ*@ho z-IHJTeov)Vn_jt@7M;&MEA7ffKdZMZXLR2=`E9GOZOnU~uctTo8g7q`x1M%^dEKjA z+2y5rxs#toK8wzmtykY4rp)|mYpBWXorc@3XYae}lPm3W-0j?9pPX|SkW)JY1D~QuFQ^LPJoa;MSLyk0Uo>mmtLMQ&&wYTkjN8Qr zCycoMnyPq{s=oiCL!>xvL9Jm!?peb3ov zpFQ0DzWouvKfX8q&KSVM0|4-Fe}Mf#z*WFu-p@Y!jQ242@9^QnNBE9?&c{6j`1y|> z6BIakQcyrpP)I~vROr-MVL`#uU!6WHAt5CtB_t~SwY22d;*wI5?;GLaR^>awcjEKU zPe`5;JSF-6`PhFA5d8eJ&yU{V<&g#)66E0(K;TbL>pDfG zZ$C~denUO2apgzwuVq~_Pm;e@h14&pT{ZssoUEL>mD|1L^IAqZqlTcn_j9>@oZFM<4r=!Jb$xf!{YI$zg9@X>ALu|v)3H=GG#}mUsu0A?F zN1eQU=JcFN>vlpJuZ)lru*t2o!Y(>GehC0LvZ>j`8GQYDQZTO&7}&s&$UcARZ_l3p zO2Fr}eyAVIOACks-bWu0agSYXiw`)k5r-`Z|_%og-4xF$0* z)MHa$irk_;XaHOI*x`$MuwIG6sx_!!gwC@7l}o0L5swo}x^=F1hfHO(2efMDH_{1; z@1!%7HO4B_a|1gq5;PaaQr6uoSQZa@L;@dl-o`wIEDp!24d))$ShOb;#rEdtC}6|N znjxw*G4nTGar=O)Zp$IzwnyLMh~^47S;ZU@jF3+1HbgAtMNB^HI{WY&@3^jE)DuX_ zZoOLF0FEw9GXzB!8-WJJ@>X;7T~=X{FUB9(?oDU8)N~{6KAsW(xu9OHvbVvJVJ>>q zwE14G%M;zllUZN1Js4c*@*hY#Zl*OaBE}!xIEYmO{gea21UQ;lE8;XeZZ0q;Eie{+ zQ&)@B^&PU@mTe2vZ9~H(Rp^zOeHhEZQ~tZ{3~~SH?ceyWiP!Qa`*vJtUD;Ky(hw5x zTV&4S)GDSMhEqt@dFZH!G;GBzPp*$cd(dIb-bi|G(L$9niDgJx z!ogU1k;ivN^?i_@!N-4{t)yIB4{wdD2a6P=oh$YMhaHzjA zqHs(K(AJAZGi?YSLx%h7;fgzXrBFQO;s{W0T50SX$DCJb<6si8f#Z_7JvndcT(uP5 z(YxS+DT}J*q3uswyrYOvlr)S{qa2a!K%}i8wDokn=n1LLM`))8iGJFZW@;d@ z95mB>CThhH9_1@sU?Mblv}~0U_$tt_0kM}in&$Sh5^-ctQ!F(0beLXcC1NswlrA5| zP`$t?B$rNDvueGMH9L6?>9+*>w9OFKEYZ$6k(Y4e6`pB{^PLjqVdSbfym;$%hbXVJ z?1T(8PFlD!NJGa;if#O`D za{3>x2bNVCTFZ*Nt*oF^QR6pG35J>no?Z)kl9o2f*{9jS2y04iI}Ayb%X94YOhuB_43VurW6wkonkv5Nchc%J`x9bm-J19w{rUz1HYZm5JF&1W+m;bRG`f9S0gAZb?811+xeriFytRXlCGMAFy4a2V01 zpZ)@!xSDJ`^AKbIt|B?%-7?iIwr^h9TAAv&kX8+^UNyl+>adk*zSyDi-slc#0j4XM#nvO*dq?YAmVBvVSm!Uuf% z>Rs^!FC5ZuKC^Q+jV_iP`RlMlq*z&1{pIw?(Hl!(_H?805KdZ#N-o&;N1Ypt$XmLM zc-|pL zU1i{Uqf#HY6$)J(giZq0sNldxuphH&uYc`J$<%UY6_^Xd$>9*RiaZ^xpX(Ksdj}}|s z2^N{j!C`SZ=os8+PqsJ*9+AJk5K^U}AKkkH>Ks4`Jf0xy5{5;Z;oBvSHqMjI+PJK? zRekIGKXsE*+ItM-S?(a`Uow1=oJK8&qVa-7``1{-WcXUH+@OD>)9(p0o8fZnx(KYQN6FO`W9$+U5-*m4ar| z;#c+6>TpY%uM z4IJ@E8oTtx3KRCGj|^WV6$?9LvcaZJtZv%Gb$g4$k;>G|)=n|$vU{ROz1rU86t1;d zbqea_HA^HEdZsVe$Nbm{4?#S7dkOw`muY{&pXbYC7M}?dtBaX>S#wGnW=|(8trZ&Q z51&rQH-Z8Rh0TJf`e$3LceVwx{yCbTgVIdW{`=84RGK+48W$E-nN z?d!6|U1II<9L+_z?;!NWh4@ud0Kk|f@g2`HT4`ak0x#z%GGC!S z_$Ti~-t%rB@H_JYKQJ$J9+kgMJe0c+_{f|M*fVXTxnpzI?q`hUaJ|TCWFN{Q3jQBjEal1yn@OoT`QBDRTRE$2oK!2kV=|2v@m#XIElTz$kd^VbLeJEs215deU9v_n-%)-oAV zs#~{fTj%NvMzG(FnyEEyJ)Y>GW_hDWRtm_1=&93~U?17D(UqaBcXYE;&1ce93?vLP z-9z^5yj9;7_W=78OMyl^>VDY#gm-lPWa|ORp>mDV*3vkP#Z5K)cZ&72)3M2R1lCOH zIDy2qu=v+E>ZdxDv=uW8c^|pII_7L-tZGgJJ0Bty(%_RWnKd=f+c@@E*FaaEy_@zHP0&_(-z zif7G3$?rt6zAC-9 zWP*YX!1+Bct7B-c(-P>hDOlYoiDU19Jj=`innq%xc7jwJZa+Q@e&g%lcV=CrfN_pm zHq^QUO)qdy0t-ofO|#yvDO!Q+hz{1sfS6*TF5t=aiWR`JxnVD0{Oy-4{?laGN!5E!F%%zTMlCsa)C}b7TRt&GUbos73d~LW?ex-D#{gHOkh3@qgiJXaUR|wpQoLbn2 zYa0}@`2wEi|m_3vMq&1_Sutz_*u6asx(HKM*beC^JGad@P;^u*0s!dmfM zy`vB6;*@#%;&sU?-K_Ug!`1V3eE2@-IgQ=#U*mCpO(kpS{d8h`bO8;z|s?#7*AqJL%9#>U;Gmltto4uxi zB9zeeO52>Cl^(0vg#@6$Mws=P=ybS|PEO2t2}Czi&daS&()Cw$k+(0kMfIB}iOXPc zK+#VPj!Z{}Iy;}~H?CLPpSdvY6YAnIkO%E=BDSu~*NVpE@jA;0pj|bo>NLS7LP4{A z#P_+8zXHbA1B|MpM=D|;P?$?uGfTl4%KjvSMeSm+u<xXu`=gC@JNRzh zO;FN^vcnr!&-%7lD|@Uwmabv=R;IS=h{wgVyuG2(2+;YYRav+yKgK@ zY8VUAM1;6s@=dU-4b-CTciQ;@g2Ft<0ZFHFAQw}k+j7jvIVHo+*UV~b-m}75c zmX6)j=>jXe($leJAOr+o3U zTTGK3l(Zxjbdo}rgcqqwrd1wg8W_>ft~R)z3#^Bf z9&Q0wIHKYU}MPNzj-v9R>89 z`RsW0^WdG)k^81Jk|2F<^5`=#-4mPmQ3MrBlY+1WQ4n^~^HGa53s$Kh747u3)P1;$R%#ZlydI-_U;CKJXOq_JXhA08R(ng&m$9eLDxn}Qca@?S38o~cIA zo$RY5)E3hrVGbMnfTUmgppAiu@Lyh@P5hxV@?yi3Aeg~twnc;w8G%fVnJ6eqlR~L1 zFa%+A(y~4Nk+@!o>8w!3-FUgJsoVgOzPL7;Gl|$Pw;5E`RM=ykYIJn>zn!5rQc@** z61TkyX%oQ~DN#+0_W>(b669GiF}a>JWB1%rap(!42(#KN@puoREzs1DwcH6D%T5s~ zz2YGnz&VpNZNq95^?!98NdO6wr1SHL$amFK-)$!x3#VAXH;?$f{#}O2-LU{bK2dL( z!yTWiZF zApGB{e4Jhw;2C=TaA6=p)nZDs$gVspDjjZ*gWwbr20V~Qw;-r<#doj+MNkJ&|K4{Q zc5mD`y;y=h!9Us6U+@&z8|l;QMd-^GZv(PP9D!ffe=RD*I;7rM*wj;oPl9bTjF_Iq zW=`+{(&HLjUc_Eo>*MU3o8qgojgx0GNrX{g3YNx{C&1FM!40P7o7>qB9#id%pT3=+ zsFai}&_6<c23G-dvTUJbNpz^tny*M@9g&hWuwfTSB*A~ z@Flj5RJK)fY~<&OUti&At57Y=Jc@&Rh-o*1W9t{Qd2Y8c&sUpO?lrb3(uJFIBRHLJ zypa~k(Rq=HyBUpB5(AGf=2tFc%<>mqT)Vl&itSla2TPvBln>PRGsI=Ez6yQ8X6{f-`b|@?I4eerC8*h*9svxVEmp zJChI+-PCSpEkckB?5dpzJ%>N=1oQBcTi%NdhaAVy@Ej?UGJ7C}@D7o!QV-9D@wUMKh68~5$U zs)LhIG<`g^=JpD(nVrMJc=Y-H3UfiJKxFviid`PmL1Z->g%{|g7t2)LbG$e0-JVHy z%n!A!@Y`ItL0VO7OlLuqC>>{W(#bH!%>7a{Ln+9uc~C$xLgwebtt{}>3w;aSLF;+M zbYEU!>l4j49V;W)!gQwLu0tTG-=g=uc^BLA@^lzcsos54$KPAvLAe>;%5afMHH9ax zgj<@1`y|0q!OyICT7UU{h8-OddVM-eXvRt}?7vzFHuMtKw2t+?l4GaEu4g`-H#!&j@bu7OPbfhV7A`HGA} zYM<+SPeJ+jesx*tn6*gXO8a>q@WN)dJZOI9+y04%5xB|?J1&q0?7s)PO9Ro}WA>hI z6{VpsY~PP;HTc*jt+(_YpT`pcutiXc;x=fi2705GDa^IL9{{&;?)X|F>zW=|ugA9N z01lglviICoz;B>5qM1jBM&njpa%kH*hye`HO2T$a_I`XWHr z*foOYWD`6eLj+Yv$rU)rP5zl%jTxdmMo6a`fydz#!uXG~7zQ?V_X}BjgxDL{Z z@$BDwLGf&TY4<$X)cqZEur#hz14^oTN;ANyuJ~PB2KKMy_kJHQ@w^NX`E-bC{i1aN zQfPB~?<~Ea?w1dwRhn~FEhNQtL?Yccs}MRK11Qn-pz1-?g>vw48G^!gtBA0Rr?RWc zA^IcFu^!PrgLLb>?#n8emO08QrKyRZ-*6#)0Zb0tlGdW^K4K|E59JLa+5#kgVb!%K za@Xk$6;WYhmxp-~+b{xD_uRWg@nw#MqKkYGD!9&d?=ebi!{+Ba!E~BmB!{i_#db|5 zLk!vdq>S+VEn6RK?C}_?;ooP!qrE;o*74-UxX{Y=Z;3DUFw{p)cGebPn_bC8q}lea zroGI#AA`)KSMtrHE>aL4#8{H`o~-v9Mw7Wqbo4xSitAIoAApQ`F~5x)nrl)Ps!xK&b(~RsAuZ3eqivu7%7l2J5VK}BQOaf%M zz8sxN_gwSypOyKwJb~DE{Tia%aa8Qu0vco<=qlTCu&&s6k`3;)Q^Lz9o zzBT4kZQ-z+z;GGCIPz_h7p{7oh9~ycd-%Vb3Uw%0 zogVq@cD#4qg_|KX9KO`CF|ZeowDEd13g!k;HR=i)d?)Vw+pSRRb{WH<*2I$sSY`57xr;vmS~?=>TWEX6fjf(C}$OLgI|OX!-ZhGVP8 zQs|hP6Iq(uPutWR4K7TZFRDzqzND9pYD7MwcWpmB-(=&}BTY_Ou*J%g4QV<|EJ=Z0 z=N7AN!kx>?7dC=dywY2*d6{9DLqzqS;A8rc4=RyGW;Q_KUT;C#xEUB>YKkzQop-Dj z$hx7flH6KKDM`yc8e43JV71R+BIL0UU4I~%X+(0%3M)cc97=AR(S0h9M_^i?e1fU1yt)9OsXP-3*cQUQ)tur^Tj_m+<1{Q zVe_C8Y(0m~M@Q{0^ksE)$A+4DQl`zl$tc6Og^qas&`EMHd@wJvviOuKf=0ft4v%Kc zK75)W7af$eAnoc!+=Ff=ifn?Mu={|4&>wA#g#vpoH=WE}=FBXi_5mFmnuFozU+OP4 z3{is`9gyIRqY-zqax1`R?IjbloR}d?4&9f)c~#y z1WjTyclH4`k693JN^5IJM#Gg}i!%)U%U3ZGgCYXPt|}9D5-q^94_HD%I%95pri;7V zwaX#$+mma7PX~;3zMb8b6_qC%ok@U=o-dCkyr4tmqvpVZ<9K4}TYMM|xRubzZqhwQj@9z;%>h0P|jo^ z&H50nZyylt*eg`SaoF@@=;WYKP-}BEI!j^e@mgAos`xl>ibL0{RxqUalr6Rh;eh5Y z+42n1S2yzUVU!SYMp`z}b{C?bapyKyL;O!4N00Kwz`9m30rYkK6$w+P3zvy{47bIv zXlV95JB>$trmqk5c#e2b2|UH?CO*(rJ?GYYe;<(Wk8Dk^r9eRAtM~AKGaa1w7fMjd z)7E9;EnAjJvTL!17jvi_DQu>0sY6^qr_Il!BcxIX@}<1Q7w6`m6||E{LrWXxagBDG?*m0riBVhUR7X|kng~$SyEUEqh?@+b++~9c zWfqxfnuF(eox|%1t{XQ)JuGg5)NgLdy8MK53+#0hFjF@TD>vju?P+>M)%P{@5gK``R2h|_3iE#1E*16S;dDVkGc_tnK5E&}i&=I* zH*M;1ECO9;Drt;nq`h3g=x#z+gwI6w+KvG$jPdc+e#B@!T%T-jaec7TtToh6E{CKL zlaI?d{sN621?oAgFwsLjwSThIqLn2y4K-VDF%YedObmHIQP2MKj?l`E;Z}cAcu8X8 z^vRH`Pzgy8Zmq{RE>6TQ+x->A$Wm3L00JZAM~&;`mJgfoEb87lFya0q#s5`>06a0T zE;TyxT-DeI2rhNvYBdbjXO<(k6n(|;78@{**__|*G|k04(cL$vut@w0KV|DXTHa3C zUhap--)ziwT!J6C4i|rS@bQ1WBxNs9w-X8t=y`0OwE#Kme|IU%j3iqywafMai0p2@ zPD-Zz`)f@QXCU$i@oy>56QYX+E8>pKMC}8(el?fPuY#=bIY{kWQictok4X6?^F^Pn zesjd^KRs+7JqpHV{;?W#H~zUq*d^K;CBXoJlmlb)kOSfqmH8VDU|XT$aD`s5%jW3O zRYVcrH^&@4AZP)|0yKOQibbpb)Feq2p_F>w16V~JXdm#=39H`p(Kq{UmaY>jB-NNA zvsn@yd=UOWQ$7ZMlcF&4g&1)!fLWAhQc}_B(8Jj3z|~|*whS6MIRCo3SawJ6_QB3l zwB-Wwc-Bl!^7(PkorEf>ceZ6^y6@Ajr@BWdPF z{=E9#a^&0b-&l@R|AgcgQs~7AzA$HY@m$?Lp!xG3jvurmh0*bEe&ZDxd0*jg6(__i ze9JxN)u9KheL%)GeztSIu*{nMacEfgz6jf4eJS90=RIY=ALkOHhh|@g3kfo>F!OB; z7XnB0qodeg1^qbL@6!Am59Y(B@(@kg4pP;><{m?aJ+d}`j548}E({XW>$1i7 znHLO}hmwWOM1i6NI6AkmDcvm$(cBV`>nFU3$6MygTrORlAFb|O$QD52g}Ue3BqHN$ zkSH-H!#36NY|&rMg18$oKjOA23eZ;ZJe9c%kENl-omF}6 z=7v+CvW2S;(-JAnp?$zG$RUuy=y{yp(IW|c3?;66C+ln@#4Q0V z_k)icEJWmbT-&Nd$ldaLZ~AfUf$Fs;Ke}L@fo6qA<4VoK!{Tgi7LO`r*><;HD0TE! z#EBw-n9AGPN1ZHbQ89T4PNz+5Y4Pw{V^4rL@4eqP`I9vLL8s$uUsU@1*oZqsLC=?0FX|G3O5tt*W=PApXRHF>RthQa+v4>vEH=*))S* zh9_R(Pz441@~UNJ5R9eHeZa($OO4GT9I3d*BhWuJL~Bg;%-D2VbkG1sr*&=pyV&*s zMV`QY zKH&ZROhKYT@yIRowFXPW;OUZhwnYjQ1$FvSS}r=^3Dj*?&*%2!ka24nZQiJ5fnKD- z&0XoWqJcr=3aCegLx6c-k%C$?qMz-M82@yAK>H3~!kvFLVGiE=-@Oe0(CByuLIC_q zj-1xVZ3m(heTA!iKMF8#Ia5l1=NfPCmk0QE)a|2sfdhByWNHcIK=iITSyduzme&B# zp#!l0y7)eFz%Z?8Xe|p==0E~Tnac=38-jKE@B0Jc^u>iX-B5CE`L2#-J2LXRN4NrX z-7Tuczw+FCh4#h;C!(`t(Way}7U{pd$T*o=@oR~mVv$#zwJT&n;1V{g!0lG)sUh^T zk)>X;Ngf*?L{D^xFK_CQxm>&8RbVvR)P7cTAzgcrMBeaQ_V@k{&mUs?YJRxqi-aey z6hpO13zJWwsd^n?)!MxdK^q5RvpH*da79HUJpIOsW9k`<+XB_^=)F?#T(U(>x?AKv z;2m}ze{^rHmHdUW(hRA0Pz+yatFqYf7 zdv=ZnLf86b+i%k|CthXL@xY_*H5R^5ZlDaskOp(AcgE7PlR`bwS-?h0gNZp_l>=Hx0__9lcpkF$n)o+bI($`C z+E(KabAv07zb#LLyi6Xb@72_PENtP?w2bhBp$z*7e0JWLP`psK+sjnXSm21V9T6zWm9dQQw0aj2LQW~WMNv=siVm;7KW24#=! zFFF4FX39#(J@J7l{n&EH4RLTnhKaQszL(q8E^XQbbF&{I0Ml3*@DHiEpt5n;MmC!V z9z)`U4~~TntGo?zEbSPKbxo91NZeMNaQLnca%p~0yG^n7d67G2DLACL&EF7A;?9D4 z^gLZk;msyp|IynSD#=axZ2^P8IW5LD!>|Gyz>Xj>BIW8U5AOhFZ(T!4 z?x8E+j-~(URRf7$Z~t9kg6g@7m_6g6Ui=6PIO;oida5UR(Lu86!tSm6fP*Gzxwe5S z2>|dz;?UnKLKtXs#1?<&Jp7}c?C0=N_V|WA?5ODiu-_iw!y((5%=QevPN3v76t}lJcAarTzrM5@rGmp=lg1oe1MDEtw^6DPTt2V@3W5vA_~Rf z50L*|OQD6EDA;xxrj1GbcTLEyh8FGvzIW_x6PFrP+y}r80CTgGDzfq4RSn}K51<%% z0GOMa#LD}3^!&%LFsad8SS~5rxirgz_Tu=b%nlCt@!!uCbL`ZFHPEa&Non|d&3>Ql z11$K2?25Wcg*ly+OT5kBeERbKa`oB0yNznIA#IflJ5nJhuqvq;gISBk8;_Wj&^jf| zHD2tE0~JmVRvg%6e!=s}p2GrD9;^+3d|0Xu1%I}5MDCM(mcsc@_Mm^H13mlZK>NHO zgVzWjW7;Gw%x4FGL+>dSOEo^d{`vs?f2ar{94oc~{6Mc=%?5{S>m^sj40ChO+QY$X zGwap^3Il9`?C@tibB_*m_gV4BMH;MdU^Y3D6*sp5svq+YKU)e6oz~w?@4AddjT89Q;W-fYZ6syqr=B_ZrXPhB6F4oex5qneqwT94vQCA47Sb1TQ!(- zI`yi8HNWtPgm07zNjwJXUD9$SoR3b=!wU|o>Pk!Livr(3;9$SwarG{QEVV(h8AuWx z9ejKH_2U^yxie|eQ7d;Jjo0Ke##RtOb@Q#L@fki~{OSY*hm}v``>H98p^L2iNZhzf z<90@d_h#)6%h$GlT&>mFs?vP<>*=reo>nck@A(~{KX1(MUGU{!g^C{PD1UCMw%T+@ z?>y5P7ol{%$V4R&){$M*-(79BYkM8z?jmLV&CUM@UFLs{Ab|f#BgonHE_Klf?1MJ8 zz|XPeBQM!D@aR!6YD_+QOPeazy~_W@i=OAMo$BCW#{!S8pukQ9KMJh}kMD^^%tX@- z;mb>-D+F^AwJ#j}EI`>HhTuEcY>ROU9yeS!wV(k{=Jhva4q#iC?0P^8x`q?|_tOi; z7W>LaU-o*^Ns7^Xjh2#5)IBqQivK>q);)T95BO|7vXJGetI=5&;TGx#a%!&;I|{af zoR3kU*XV-x5*BgRR1aH@N#xZV{v_&g58m`m`63}RA(Awj6@yEhr;#gQ%iovzIKWc^ z&9C04hGZF8%kV%agPP=yBlnbyg8Ue27B`pa{lHmU3E%vem$8kd)!#jKJKx~CFqw)g zMWE?}WIPZ!hE7AHFHGv8r)nERXalJ>zGP1ttl+Mwvy>bbH`azGz{<^Vauu95##xU> zi7JS+Ol)Nb>9SW?;)RS}GcT=Xs8;aYrdS{tG6|9z!WTAAWDf2Y@b-KAf$_Rb-v_MqXqoIOq&(1(VP0Uni5vE^EZNqti)O0 zbzQK=3x5;=&0hwY25rz(Cm!U#VdrFa(}YenAQo}R zFegryT6~v~k*|&2T$t6!?E-U6={E~;$JK_*jRb4ZbKp~L{;Jmd0Lkg&@hqh1vrpc4~3OFD;6%OUsUIG{<%R zgSi<_epsAGm}xLpuKC1`BkR)0yI%r?bkM`c?jc?uUZ;s=e`v9G%PY$ zA+My!1i5YIZlQ4-!?Y&OJ_%SY2yJMsru zoNNwF5JQc^vHYzMy~3PYz!rA)Fa%wE41>o@L*GqSMu;lEwtIa;$)Vo0z*2-|Pc&zG zVzmsf;)kWR{J%HtPPYXB`u9dFWcAGtbG>R<)_xEHAt0UHzu7^vn z|MX17r&+b-Vr3KGrDo_U5ceY(0tglJy)p}DZ&+d=;yA>Q~uF0g|5J09Hb4509iR;Lur01ewl-NhN;gHdB3u)BavUeQ+-jJ zh*e-x!Ra&D2~cxB*pgH@n0FUecXKO3KsHaV&Q+=JwI$TY5<=Irhvm2-X0{*KbZDHx zu0e};pS_tny$fy)53cL0R%0u%#{-w*+2#vxvy&-pzLeG%(FKNAza4O{V-G(Y(17Ka z42{bddkaRWxK^pivNU}g3=K`(g#^jSm@s(a5A#UxzGDxi;+ws_MPF82>{Ve9&k$q@ z)j-oWOA?H(_qYxvQm47NB_$b>YR^pqfZ9Y6NJ`@u*?pB@F;b^lV4EsAGBL1~ag+0{ z);P5P>3sFm>0mLfA>ootZ`f+23Ib%DIXP; zM{S-&i4;bRQhL@C&V4y@o-ll}jhuZpHjrwy1(nvUc_1x&F2(E@Kg*QG@lzH2~ zXpD2UQD*jf`|je@(s13q8%YN=ab?2GpZ{t>WY2%C@`>orsTQy7%+rsk_4ttuT6R_O zO%4mU<4OAW?vMYQtSLNS>GE`Yv5b;ueeMcnPM?k{zFXh4jg<+bsn(^tHCuJuyz<?NT*?6VDq^;wKz)0E%YB0b@SBf&hg{VZ>4ZLvIZ#fwK2J`5o0ihRpVQnOkmS& z6f2VxpKm*!az8XZImz+|$_XX-ep27x!7OSzeHm*Fx6%g`RX;~6f8WunJ%5SI` z;V)9upTka{sq`X2a4I0{ZLernboBDN^pNDk-%y5K>E6N##Jf$5*RHoXs8>4OY#g&w zjOCb)f@bci0LO;rHI_5XANMVIqm-iRdNt(hTywe&jcTgck(%#FnL8Qw(jK!kSx8mu z_X9T9s%mB1RVKC_6ml=uLMwBoWGDNv9ril1=VKsVMaa~LkGpP$%LW|NcQ=*-&uZW5 zk*gpr(R$5V1Ghqxn|;E4U2bn|d*{%c-iKRZuq`XY>%Br%s$^EDhvte@#X}C#W7TgVR;C==Vq;eH= zBY=|2pFT+(+Y=CmtJqfxF6U>{FI42y#TW@ExV%#ysCOX3+fEs2%n$`nQBAqxex5!N zQQoxIF1G9AUZ!SN$hs=iKjj3k3KAYyJ!WW8`O?}eits8y-L#{Tn|Q_cyHeWaVP)hz z7cX&V<1>qm^$^P@ zww%rwRG*sMj6~&h4{WG8B;;}9k_d~kQ?*mY8^`b(Rh;Xfev$4Hy__2=?O)J?b?c7{ zYQ;$mG?%M8l+6ycxu0y!^UE^ z+~F|lT6;Um)H3x39v7!vMIB1Aa7d6`{5;!ulXE>2Oj5vVcXK}$R$M)a7tV6-qI#nD z@TUTwQ+D6gb>sqzwtVk;NPGp{COWu>BC6#d)^)~{HJH`L+V~`6CKS1_t>@V-qf@ItW(Pb!T~NWPR0ED4gK# zqs{m_)9;Cu-)M)IbCBcCxx{s;6R00oant=uk;M!wg+p^TC%#iMU0ivwSfABh#jISv zE;aA=5nK`@$t;qZ^9Tz$XQD5@{(KtTQDVCfaQoQI%eB4@aLB*w=mSLX#^Oq8ZC7sO z$NsY)>g9*f0_IZ;`a*qBxGL1%ce3FMmzSo?FK83OpEU2B-V7K>&9u>f=f-)6Z6@KH zs~EGE9Up6jU1nWx$7kOx-m9m2)G&+0E#4_DZk;+nt?I|d1kAGD_rZ~Km`E1;*l^tL zfkxcu;PJVzCk?d9_r&J?K`g3ImZm<19jLH;;*rL-tvVu|VMs64LRmFBIjGYteAOD( zO~2JD6m0ElwaXI0X?Xx#9v;-3iG zscu5Rktw;-z{Po1+E4-KoCfHbTFTSi=hXEJBZS_#9GbQCqBA$b&OjuMl^`j99Tm~P z?q0N3ULXlh@4esX>b>;Tl>pY~71ViULvEIoAN``EOlQ!xr7Zj`$Zn)~Z_RbFQ{w`} z*9e8r3j>*HjzJaTsr9JuO(u%p?E_rn&7P|pw}g+-&R8RX#vahvO;lg_Iul!&qkysD z#_zsht(h8P!z-p<#k+JS^1aCEwI&ZwHo9P=<`dh~z0h0QU2#gV@B&{*6^0KzqB45P zM>xjUwOF(uSWJWkkD6<+@NL0L>$t^q^>>{y64TE$aw^p<>3d^*@+%{&7N|hC{ zYQy|L%nlu{DyeKVmRvte2X&LVA)bwvLaHPPYb&u~$*-`=NcA_Pkt4mQ40~@+&}t_AEso+)(60!AZWN8A1R^HF0W}%GYXKS} zs`7_BBsp)*8RR_E5>3JG0n;Wp4 z+`8AkD3ug?c=oH3V;w(Esd>@m;fifAY`VX#Cq5s%P}E3>PK%mrz8R_s_gpSo-!ZmX z$|50=JKq@bV3O<$iz zr)}G9yH%jv-!(;z57wfRg#VNs|8rK0BMRKS#M86q+6%9)rf@P9J8X`wop1h5CCE!E zgE&9y{SbPYZSswR@t>%03tun9W~{t5xGKDtH3s~SE_!+LQN#m9j$X59kJI?DXvCq< zykCES=R3Q1p?l)JgW_R~Nmpv-7Nc#n`u^u{-L7P+@bsgjaUtLoff`-~Ua%vd@rNtUQlO*?`!V0P(TO0_PL+{ndu9 za8r<#fBdpU5kxjLL{)IWP!L(aKrv{BbDm6(gz5V6r=QGITARw9embNFQ;kKs}FrR9`JGu%r81L187qFzO zyt(3MxA)8k!|Scoaw{z;9Z%>HX%q3pRnxiR0wNQ!DCIV1#1Oy$_W=@IEi3z{rL9t_ z<-zQmm79_xyVC4LrlqFz;=?vHSMCh$oa=bK8P=b;$?7O^+!f%DkZjp|E-3mDVZQn) zN@O!UoJ&Qp!_vA9`#k0`^5eY&fcI+!+#`RCimmwz+Qi?{Mk#JTMSf2HK+1=E`X2%7 z7!ONo^9}!dwDpoX-%G1-F@5+?OeNam2M95={_dMwA~0vUGa5{CH%7Q~t86)QRZ3nR zy{8x36InRIr*h}Rm_4E}(kguG@bLp$a^WX&)<0}*416Pbz}E)=T$lJ5GW#RpW9;Pf z){@%9UwAb}KD2xc%>8Nsuq z0RR4XU?Mj_l1M56tzUAE*=(Z{{0hj;RVurmecpTL1Lz6$U6=fK?@izWrxKrE&{?g; z)T6!-{Mq7w{iA|Z$N!g5Pu|{_k4T?PgW2l@;J+_RS+^uYX7A zU!VSO3R(LF<&EBSq&Z>s0o&hiEV?uv8g$WAw+R{MuwHH`sj`oMbK{5P$9IC?zCIQQ z_~h^TLR1wam$(n0ze*%Uo!kc)I{S6LFOSO>0$KbShyD+H?*Z0S)~*fXIO0cfzSg&BtRe_gb+G9HVADTloFc47$AWlB?J;6Dor4O7(xpH0qG?4 zUi`N@Bl=$Re*ZcD|Gwv(`3_tc$-eepSy@?ot!Lf)Sd#B~Wg@SXbscKj}1v*-mtZTB@)k#f#(jI-k|8!H7z59(Sc6jtWq`Fnl zl-9_v(W~xvK0Ec|uev4x57tn+2YGXZl>nXgUqhmIMHXPl{7h^v8&?`JU)|zXT|O-F z)8%M0YOR)!nDgYCbWwPUR!`p!&v{@|k5XVUke&X0?FP3&P4}kMNX@{LWlb%sC<{q; zq>>-gXUJPmsjd;!btrRZu2P8xKi5t?v%y<Rju#D5Hqud??w(i!UcbrAy z;_c=glCR&F{C7G>1h1w480Guhh{dCo#?a;mBZOP40?^+R)JeR}wu`=`s-mN-nGXdr zKWQ<~H)y@)fU6RZfLLr3Z;{`(;M`#Hg#66Ze%g*Nls`0l+rRy1zx}H?YIhDdDGI!v zUDuZ{O~TdSiG~KT1pd>HSDZMuG$*RC+k|bIZYw~H3H6PYNgsGH7M!oaBEBqGLPiZx zRXn-~{9390Mecc&WWUk0mhZ!or@P|o3j(LUYw=#ucQKYY(=cPdyD9m8D5AH1v&VQ9 zJ~>E5x-o6P?UugLMlQ`8*<}zLayEjG#MGDMrA>}|mrOeQ z+xZC}4v$R4dg+;43RW9>;<6`WC9XbCFN>W`7?;_Q?J!uEuYB%`^>D`yd$uC|zKJBE z3IuFonmXWp8f*UEsXf7$)<`Ka(a=}O!@n{EWrOk$yX;rZXtPuqe98a;o0Gdzy6j(` zK|FlJpETs{Q^Nx5iKHm^#lrFlg-*dg=LhM&v^A;XkPl}oaZSSU+|#1qp55y7R*2Am z+xQgTr7$9Vb3RbDFdx}?xlTDo^vij(X8|KVI&XQv25YZ+YZimtUDT0$aozM0g~-k% znyL)15ZSO=pVMl-qKSXYgP-U2DTO+q&7de1=g?Yl^`()6$>5-8X0#F*+mugfRnknx z!054)^5zc_q+~OA%Ah>fwHoZHYUYO}FSv)#c(GaHKljYAx{W0L2O}=Mx}|;*aark= zQWwJ}Y;QOTJIL{h@?cGp4VyXbvw6sU*Q_{ zC^z2MWx^|1WCrLtS}mggF@b(u9-RdWIh%WCUFA_x8@)i#s@77F*->UwAle$ zne__Xt8AuMhR?yc;Z*4@U)?@8rXyN1cRTi2ibUS2%lY{|&qwXn_+c=4n?|n=!@77m zW*N?;6cE@cd&`3%5ayx{N;r7^l!h5_SA!gz7^Tc_zI;;74Afp>XHc6I{g#LLUFk@L zaGp*5c*Q;VH~wXB4WC=w6f!zN}Hfit=cQa(ub1EdUf^-ov5bk zu3`IZP7?DL3@lW+Stb6e5Qkq8edmW3;2wTC^!)xwIyjViI1fG0W;laq2W3EmS5yPV zSmXY81&a-l3QvOgU!aUrS^?%R3s^yHt=Y_pzK#?vQL0>13x_+qT>lgkA(D?U7 zBG%Z+yv!8)zKoe2>eE;BIAHv5G|FM0!)4?N&^QifgN)qExu6F&aH;fK?iUM z=y(LSKtz3*iIzbx0>WIYgQTxLG@L|a+d-Q8it z2wOx}dYaxW_~n^(7QwR#t6t1Fh_s(< zyx{JJ19t3dtSC@|zI+W5zM%L9;A@?q@*c8IbMA|();(|7UD@R;;HufNCw{~~xJ5Mj zdB@!%ry6C$DPk&Xyie?&=sUVlqfyhiK?E~>q(jb9|C=G9>G*M=sU3!NCbhLo!WP-M zs^&4bAA{rcg%p^ZX$&ywb~T`}@oK}(;{3QFwFBvL1@S}vm(?`IFp`lL9<5AV_{!c2 zr}FbJ@^Aq~Ka`cQ59(9a?A0_PXfX@*ZmBY}sga+sV6ysC?M8)TrOTd)OWnhz z>Y$?p;p-LDX9EDmWxk#{JxnM$ckyy2$uld{)I}}W*?;u}*%0h5456S+H2`X%Nw0z~ zLjz$8vZBOHlY4c`#?KqSxsZUckHkrunGMzbpi=XGL32}(5`&hf$L>{2ZPiLUolu~_ z$@DlP%2$-jkgG3>JY1rqSOs>fN#+CbwNFlfL416Cv%en|CiGlM?cPMV3~5m)~>B?a~)f_s>dk)Wmg@f(QzWGtb4Ds-_k=>A(C zscg5`$1i;>2^<{TeD|D7^_D<=y>(b?WCQIMR{JvY`un2WZkGRTZWcvzZg5h(^Nq$Q zbK!M^3Z1s58`LGYjPwo=FYB;hvPe(@p^r({12<27ufHH)uaLCJE_oC|N*Z41zt+Yo z3&*(3ja7AFG9)8?{dAd?Zk+iU^raQQ=A{`E>Nnjr8;Ho}CtcNVmer^Ivq`sbP4zg! zydhI`$U9dif#BzqIl<#dta`Vn z28ij~zluf55ZO_3s{UsEGAPh0%Ck@V!X)z2%Hh)kRFMOo!V=J`*E1H1DrD|NSD7G# zRwp`GF#lc5UEdlFsmbv0B_P}9EA~cP^Ds1`pXrh~G&?wVK(E0wL82vnhX$B5UIGl==4)|Qvi6O8v!chfoEpfKI(XkTeD9W-o;nfiL=-;0R-gLc{$y^+AUdo+M_h& zyI-NuGbE^l(0FYV#x)`m$?ucb3hENg@|~KMb-BRKfzBK6YrcH%^;FPAA}f7YB)(@^ zLurb=lHOZJyX8-=ZtTk9=i`-})6(87o(4NDx?d;xC9%PR1Fo2iP^%#-7ShF!2crdC z{06;xCcUS296i;?8k+pc;r7AIfQhsayvfG{y;zjxRG!3Yd?1kRnDdj)y9=B@OT0gL zUm`ec+8JQ|74&63ku8DW= zzQaT#H`45?ovyV)P+f$*5XI0YAjK;Ke{vF^-a&goI$R=o&0lnxa;>94xr+pTL4v7v zI9RAIxJ9Iao_*=+c`wOcRi*9zVo_0hrmW%W$y;Hbm^wemRSHboIgD?6`Z? zfWm^FNUj{49R!F5`_(y{w_w0jSRmP8Nd3I4h|zgSiuo`apCWX`qY34ZG_e4WX=yw^&DB@t0!CpuI> zEVW?youH($U+_#MIOJmVMuC<+dsBC;7?D4^u{Wi&ehxcB8sOSSk{EZ2`ZWdlpTkpU zRx^W3E5aCdegiZ&{_2>;w>-IndXsa9BMXp(L?^)iQ+hmtgBg$MrmaB93#ut((L3}r zYa;8%x;^`6gvO}v7$=qWmme6gh{&jJ&#q~TiFBDN0A70gs}@{BKt3*;a?LFQo&bQj z-E?YkadAMj?tH2QII2i2Pq7{6IWW@Gc@_xf69cU*Y@qzYW*b82KILS^vb*^b*Q=cH z&wHlZr58%`pLI~6dB&UkteWMql+69Vz7aW^ywkw!;LReVWbdKyXx3(>d_%C8oMvNp z2j+Q(FLij>VO~`Gl?6loEl;PyTOMTt!oGx8Q^`qTjW-H7#}H?xHg59HnFl`Mc2`9O zb#yVLuFohI9`?F8FWc_ zhso-Zxyzz#RjtSSugA$ZUK{6Wde2?W!?Wd&4>SMxs8CYXleHpt-ekv*AL7nFG+@W+ zdA;EcF;@6m_KlB+#4~V)_{*}V-`jmC^l!_@cKUqNwS3_-M}2&$A2i+E_lx^iwmOUF z3**tU7TKHgiVkC3DS=NMI>->`Uafxg>u>eTKNNo#a&Bx4?8IjK{~ofHb@WZxC42Np znChKH0jXW2_OBh>cmm&i@&)1*w{EvKrXE3#Oe~^KgLQ-iEI&&EJ=Rp1A4~Z5|4oqd z!Uat;x-|e!#?@S9dgvqSP&nIev;IGYlUig^Wko}r|&-SUm>Tl z8wl0he0cR81yl|o6#4y4LH+qK;-cQ8g*zw9{ahX}T_h}PYaHc`2oLv&1w>&@aD-65ycjtLnI*Ax|h^E-JuZSaWk#E)LsfnVc2?D+3%IrR^`!GZ8To zyG4P>cb}jDizAp^M?~kefF6ZxB%r5}>AMnYk(A-5lQDniVbt95=YuEf%=F9|Nd8Ft zBA0wPoN#NMcJjxslGGwxs_R600Jp$pUh!f%R3SntcaCM<9B{X`Ba-BqY|eBM22%oy zo0lhJA5=JUY6!lptk7{c4B+cM+Ml`^mp+&-^5)=XTv)x-!}oMtK-}qp>+c4^yT7@? zEkjX)ARsl&(Iqk+@r9HM?GP#;L;1eY`W?!fOYUuzEVL21KIsk z>E1JwJ&hd{O2i?ty+Z79X05uZ&Cq4-oLDQ58pCUhX`mM{rP3pTi@Zk zqdN5H|A22b3i%qfJ0y7`_f@TX%CN|UxidS^wc6<<5$p2YOmsjR%+&NfIlSA8jpVqw z{H8`QM1ow05a%ux|1liJgAuk#-I)~y+U2D9CX%NMJZ7A*eM-UfX?5dTN9$8L}Ge+Tu@_DJ6z>D%aKBXv8p<-daweA}jP+w}h( zv~SxZeS4&DkM!*)(*Lg%q1!lp8>esM^lhB}-=o~Kjb66V%Qkx1MlajwE#+oo?v1AHRSvh90$KP$GQ0k$K&w$uOr%T<%Nar!n+ z-^S_NIDH$ZZ{zgss9nhaK}K;Kr*Gr*?LE@J&mL(SZxuhL=nhHMTlhH1Q>_DHYNWW( zxRkcfC4CSj)V)}RHZe+jlv*FnWBdKqj(rhI+b-lR?j1j50BE%d2wd4 zo1J2D-p8!~!V*gumIoVpC8pE+X~AA5?4y$$sgw1{qvB;5Q>E7)G?fbsoW#zEjxDpY z$a$X766qk9n$E@f@Nqk6MOJY|AV(+%%e6F8s`g64qjhzTYOC%KC*(l+?@}!|Yk57G zEObJF*yHLK*a`bXe+4nD&*xj=top?7mgky@CvG4f;gi7eML+AxEE z${A813J~d5XOc$p!-p?4HkDfp%LIGL6iQ!sy$*w?Z87$9^hU=9rT6 zaNr01GY=eRDbje3EfMC?pKGQl7MK$oZQjgbts@7}K?b86(XWnpavw9gHt49HU5b7i zv>mUBW`^EGOb#K~bC6np)J9J#CtkZgy)^c55iW1jU?{w%e|kX=;Bi72)-GeXdce)F z`F506YzHe~B<2#%R;bQ-6cH*n%qj_s_+^qwe0D!*a8jSGG#T$VuhIOCUTE&Dt(jirV0zG`fBn7tkc z^vXtL=Dhrok0|4tZ+Hb5A4LumDSs*-5)>!CDnxvHDCzpgq5wa(=cg*+$9LDBB|Q#m zugl*LP`rMp?a6#Sx+C)1yYhQq^FI|y;~zDbPwIKqxR%yOqL7OK*6iC14Xkw-lC?;- zaq5ake0_c2y-%v=IT#B&TF-!tK|Ai0ZYg>MXlZ15{RmwytPT@0J^it~%+2Ns)CRhR z+Q`p=2cGyWMaV(easKK97Q436w77r6}tYR>YXjoTS>89C3$p? zX>Vt`y-maT{L11akI%;j@bK_nF{(Hi@Hm>Aoc8MWQ?t!G<*#a&i~vTV*WWi5_NZS6 z!9Nr}{*e#Sf%%dA7GRk^v1#xVE1e;);CEZ=0|6Ia4B0EOCS>#`9R@(*}RSW*%N6<7+n(|HGK5KNWKVUXDj_YG4-2vy!vbx0gv8VRjra?ep zbI-}d3PV2zNRd`_H<~RC0x0BJsd@sMLU5L**AK;%6!t+Pkv}A%cs3O`87x z_+D&DSqVhy#>>vNvqGpl^f^OkOm2T(KOVapXXiP=)FA8XZ>(jJgDzLEfa*L?RA+eJ zSvo~7UPnYCwfB`5z5y@T;h^g*-b;Dk5$3ee*IRso|d9vIi4sB1SU-Vsu|9YIT z$Prgefqz3zB3?3k63^6O;WUw}rCLepXZzn+XiD98PrfSlebz3dIu$#tvRdqK;^SDf z?*uV%#^9H%qM{c)tEvbzh?%(7?O?_<>S<=@@4o{G?CJL6(xZ(`#Zp~k*qf;vd4pml zVk!MT*rMZ1tNrtOVu3(9Z}Nj??hA~nSKpXrb5y5?F^YV-tXF~;oGTZLa`FoE3oolY zk;SN_&1!S3hYtlK#7~7cibWM{5q|uASt-(Z)fYMz`n0EHO-MPTzfT8r5NX7fr2wap zJfkOz>2FYn5_j!A_owjy-oKoANk}|QL$#0|@63w1ZTGZkjTn+Ae`dTx=w->n?`{4l z%B!vNadBvbZ85Q&A}|2@l^riWsZnd2e&v50{mKDDu6pa}Zhh6d_=7V;*I~qOUMg?a z``VMLgUG6l(#_gICu{oU{ZBnG0OTPpJ49KF%d=JdRJ8L?i_*}nI6f_WD1A{35RY$v z)JPI{NWU_hva37ubOo3nrl_=hDMKb-iA$JsQVXzf&>stqyXdSm6tElOvh!BNgrO-5 zjPf7H=Wy_$)@Vg)=*G-ySD(05kSdAlZDUcPBkPRuR&y^6Tw0KCRRUAc#$(a1TK$lP zh|>?}QjFbyGlr@~F2ZU;$7=j91;&}1T=k4*7aV-c6C7{_s0PWec=UW(czQ`3NelbN zRu@ps9T9ZO?R~~nwxZo7N>ZHi*sh#DetwFESabNR75i9*zsFq}G9XV7p48`4NkTC- z;rL*r{L+W|g6K0<%~EML+OJfse@C(Xrbg6`jMlPZ^3=+eLk~@wy-8v#ng=@KhQvgD z#d!}KO9=-l(U0StoN6X!jLKF{zT=23HL@LS{E{BvIBEZK*3CM&47TTB!WG1oYgX=7 z!E#uDF*eUkfT5@OTvF<0Mh8XM$lXIw4e~>LYeaI`l4?%cW0u(vE}IiOcc-ON^NCli ztmeq-Qux^J*46}F{Zr44UDAw*C2x85^p8d)&Kf}=W0!SZIXam>u|Z0k4*Ej$7Mmg4 zr4GORDND=4@y{yLWkrGJbYI1^tGisQroZnt|FL^9#Q|3)^fy?~evNtgVfy}k6;b&A zB+M8B2s5+k(3r$>BWlpJP^WJHmc1SH*R zRhQ#q%*l_l&d#L&SQSW8syW(tl%eR5Wj5Va{FX;Of3vplb+8ITrsd-|mIZ3E7N?)}l2o$8@oA6EsKEyHU;R)%$(us4 z9DB>Nf+{CDNrt9cEGC5y1&>9)tUvl?{J)&u>*RmJ>#{L& zQ2X|s1vuoy>&F#xXIr9Hg+!S-MDfmH<~P?rcP#Y!?P-Rbn19Q&PU!{(bV{7{&0+w= zOA~Url)B>6zhC?6^>-m3rpw<}s{PAPA4W{{K67ay?gFM(D%)~`En>HQVJ|zdR=s0o zoa1(1Ci&Uk{b@JiL*Z@K-v3J0-gh3>ZgNINcT%id9n7wn`%D>vkA{$?i70ct|0$Y0E}zOyv&Ax${r^yY0=zK zUM{y*1^467N`;#vWR7QXO_6N(cRYfU?~4EVCn`=q8;7I5s>?sjIX1`X z-$Kbm~+u4gZl-l;y}Z1D;m1u7Cct zny@N47aGH<4rx30@v!5tvRkaH-Rp1UKh!5exG4*MDxm7MKEiY`j&(+4kn?DB>EIA` ztp_ETl8W;qegDAWyF=F3c*J<;XkzTo!r)638D!Ruc|(Jo{rCj=hE4732UyqN*R1%t zlCynasmUXK^@~WtPwPW8m&sp`fM%h3A@w1(geWyNS zBmQQ%%d(F7>tWUR%}Lv#`GwV|6C~l%r})hMZpm4a?!(U|L_e7!B~hQivqpFM0?z)i zqhF|f<#6u9PySmz!R#CXq{#19Sq8Qr8J4C%gKfx)g5X*GmGHi?1hc7Omth=Y?Xm=Y zPq4Z~7)VDMG&TwAM$Z?(m&n(=e3}fq!mJP=&QOpK@Bqx1Glo6O!}PYk-+dG zObwgnl@~?gXX4B7s6nW^*6tCqVJJ|0?2bR-Tkt~)-Sx6Dy|YnRN9TNgOv?q zD}0*PC9Y@oEGH{7+b0;79!92Jl*4bzjUa=MGa9A;`c@eh^c}j}M{j8|WG%+)B)+>N zIq&btj(_)gJzz(9m-aQ0ZVOayQSRX0tP@$h(Y0(E{8Vi>CIgF4iry%Ps6Lf% zEAJ&OvK%m!-mkmkheZB*6re_U|L(W?pATX5C(DI}eJoX@Baqi-02wu51j#>$-USI? z(mvUQLgDbz8q(R5>s=deb6a;d;rlsft4s{F|K-}pd(8WB@TM)Qi5sxMwUZ7F8D5W>;&2p%iH>8KSTQX~*sa z(yOoCx|KY=pLAL^FQlYnpymB=AOpi8(!X40q3fTFhM#23{U`*^57&jt3e5TP<3 zKNs=$xKyLy`%YVc9lk~+f3$ivG!H2lc1_p48`(2rY-u5+kl?|s8pvf@+7OA{dc0=L zX@4xW?qa0eSEC{8^!2tTnj3WGE-fh?n?&ZwhAEz``_CHg|8)5pJ2BypU#O5W2b!}~ z;Q&qxz{czYlQ)s$`B?^qC9P&@Z6$ie5TjzvDV@gRH;h|#PQh~7)JT$%E#*FgCFa@O z?VK0<1?O_pL>Pj!oSQcot_Qb)nTE$zZdv#1keq-DFHQ+fxboUa$5p5Ar#_zH|JZ5O zb2tBrPKPY7^5fB&yR?%x;)2d9suT#76nS@vzeH zer1DuPFlc?XTHDj5rr)(;5%xBN-^H8JF*1a5i^yI$Faj8_|)C=TWj~b&wUs$*Z*bb z;h*V>o{Br$>#VbWBKg`|o;6@R;1GCP6(Jl!58UiVeV21U&n@U%FT!e0J#9%{!+yv}!}& zkNSK7)7Zw;lvo#Qi52G-nZmF>_$U50UV1NzB8$Mt4iZ?ABAq=$KU*>2E}wI8_BQdx z%i4e>?k4^pt@}Oqe=;g|{>#q4*AR)_T2QJ-0`9-%nI^tRkh{)}U2a~V8lLWbe(TB; zQFs^ZB`W3>NbdXtw}!E218zk^1B>!a%9{v05GNyO5_1f<{?Ql^@&+XNY~qvt3Fw0a zRHZ}i{k5UsZ|41bUrLuSkoR?KNzea~D`Z~rn@6%M*3#A5I4sGn0ss6p7YwL6MUlW} z{DyT`jZBBE;mGo4tW$cuT{k*x1;Fkm@c|FiPANK&L;o5WUhVpq0YLs|e~CG+XmtJQ z);k2az7Z!|Kxg0b^nU=N`J%pE@w9g8`D&Lg{bV8_EvuYtUNZ^Cy7%cCnp_2ITO_(f ztT|()-RJYa_)E}lzVh7R?|qa1PlGq|y^Ayd{XT#n?LENh0053$f;08IZWU~|(8YAv z5p*d3&WLGpM(>$6rqaU6m=_cJn_WovxHj=opRu8Ok3YBljE}Hr8 z>g#GxYi?iC@q(@D)2 zaopBBMb|SAnmZ|4xDoP%pt{LSiQly2(Qc-g{~ zY1ab(=9g^68(02tg64n3yEUxQib&3B8Rm*M0>dbHt>DkHuS_=zHrMh2nXJtHaInA?(9d8HuH!&9jD)Sl2@$(k2^dX3;3bEdde;vTM| zvS;+#WBj_5Ykq?WVh9o;q z+w*U7F*q1y(+Z~n)3!epXM%9~!KHLnK}~j*DS{b62{>dBAYr;p(aOBFXXP)Air~QLR@z2*9v(m#y^dEND-gq&C>Q}`6QiC2%WwttDhzl|t!-Mv=#aXmr+I{YPXtB?~2JUjz{V>pWJZE|6KOdA(#KmNLVK z?JZ~q0aao%g1x`p*qwQ&V?|%eHKABf#KWsnD6gr3XvFuLtfy!x_Lk?AYYiaeYp6^& zKDj^@k~3>B^BFQVQmTuG1)tlv6%dBdI9n9GBs;$ERO%XK-kONlvFA#hLe_(Fi!>TJ z!-tJkYC-#v0}zt^a(Y}N|Lg?~o=}k4QDk5tj8BLK&Mre6UjT{`1A#;Ye$34&l z;sya@Xm=PZH7Ox10TkEF6FW#S-e3XYJBc$Te|TyC>{BOvMq*NGD<7*O(}iW*wP}RO z-;Zfik$1>IgY_`*GD_GbDZ=KJ-a9#u{US3_TFX^eyM~W!N<)LK2qDzCcOAQ7%Q81K z;koKoR@aaAe0fay%jvdgJIIOR+ zLlEpo84V{Z8dm!m1G1H(*H(&-hVsil>#2p#2t6rvH(wx7{Yr6I=una?vZNxpRvvil zx-UT2Ym$L|*R{`zqs+8i`>AB)w}%qI@RXaEuW$xIlMT(go4VJshpY;sPv?Jzi(AyZ=(`K&i7677a_lJVuHDe~9u=!`Znb6?N;w*9b0VoiTR zf?`q%LkLdNHR7uz=6g&b(DJ697y{7=>lTp&h+yj)h|L~?o@-L6E27hcErf<2k^+Z7 zOu|!J{E(u~V?Ih~9w*Ci+%Wd%*CFhCFstB@@U&beJU>$VVgx5*{$|NnoI8N%loKom zknoxVY3{g3cq95lYu6%bO2stYu^{m+Ps>}Lo1UloR%oInZtSix#OM9TV{%n}a!s@0 z%+Xy|lOqaMudlT&`d?<&D1zR^9&M@Tnbbv2cNN9@F7It2`*A1Mup^dX<9(Y(EXn4= z+@&4*Ix=Qr7n0s#$Q#DN+QZ zbWIL!Sfw(wFLHLb>8Mrcc1E^M;UhBh`Nh@P6$FnZ`yuu2sOP!)yip{P{rBLL1IICF zrpB`0`gN9pbvA0ZN!Ano*yg$YA_tVhrcQ&Y!7Gw#^!0lc*{}VYigICR6pG>dhe*&roSY?P|SX(P(1m@M1S zsc61B-_3{U*Pr$40Tfk3%KwbAk!3MN?C})yPBVzEm15{!pud(N#Sp0GuJ-NzfmE?0 z)hYXVKWbMbdLp&0H58lYwK}X$8$LMdgvrS66Dt@Lo8ah@cB#)9B>Z9Im==CYD+l9R%2c4ZX%C!*UXu;*X z34o68?9PcJG5C%Z;}m9qMQSaOt~XG07AFga_xw3LA6W8^gK;cDI9sSewfqLhsj6tn;R_F%-XsYBI_w{eqJ*mc3_F=MpOeavu-i$I zP=UrgA6f~Ol8CAhy4D#eJ;aLRr(m5_>5G2F*(~uUoa&}|KMjiBxZZc?-i_|&7Vl1$ zh^GAGDhMrfNF~oyh&YAMNeqwa4lZOr{j6tsl6pFn?Mq)#Y@AY-)T^#=!J=a%kg{Fk z#u2Bhq9U=gd4(QR3e{$J$=L|&GV=wu>BVtijeo_aXje4N)=XIDnfWmrJWcITIe2V_ z6SR?E{M*o*51HL{Mq_GK%@3R8u4pmZBeeog5%Q@(x=tj272u-0BbOY&yxmXR9G_a- zQ5kJ!gy7d7H*7xB6Q@-WRnt^fCX}5~nwIOG-i(tfsIs3}&(t40LstPz7KEUfp9AwS z6$BhkF};0&8LhJ|3LU#>lg?SVD%`wW@!YuUW-+QpqqM-Cuj zVT>yC{xG0(V;Q_8PRhV(R)LGTqkEZ8b_T|Y&7K2*kM`;Nq6~hDS`n=%=tq|POWJpZ zQM+K#?Cs60{LiI)BZ!m3|q(XOP=b4IBXMc@=7>98B z!ohVN%E|iK@d0W)hat*^;B%tey0d)iY|vgCK{PEWNz>D`a6_j!_)v1b*lz=z&YIzp zK|XPbJ-4ZbCbw{ss+LMkrj=@*U*?G3ZF&HGQkG_Whki|UjCQ-m2DgxWf0~HNIEjNY z_<4Wd8U;@9o&(Vi7Koh}+}ta7Y5=y}eT}><(g$gIX9=|{nh7$j-7QE%u$udkJ52&vli;mmS3A1;Ms#%0lj3P6XvPBaNfT3N6yETx6v_ID%0IKO9rSF49(0Nza=J%Q zN{pj0hgiTs2)17>_4D#m4^u8kd6THg;?--A9hhVUVm$T7Gh`ZhCU+AEez!B{^Ombd0+>#F-BDI`5h0;Jo?`BJCV0H5$GmpAz|KgJ9^fPs+jg+ndm=cMiiJ6o z5a~l*F;b^%;fBh&#z)?;Dm)QANg3Ix46*y%-Q!BfS%BdhdvCZz<4u#U6nT(&lbn~v zXt%0cg=N_|X-SI9t;NNg6I-47?g&_t5eZj2PZo6H&NTBLG`ZSSVP__~o?7nm=6XoR z*|L(Rnu^1UAHp_?J6uSv?OHLe|1z9GvwQYY+92WAIE7w*--Vovk$t$-zSK9nJ0&NyzAe#1 zt4HBeoo`K(Ak&J09_%R1o<1s}AD5QZK|?omkzg;Ch7(!}k7k~w1@T%0`4H4(4ZD-7 zj9##r8HDl^_`J6~$^3P`H{FlA67pz;P~@+l-fDFtRJsjA-Au9Nv{`%9u_Qb;tio>5 zOoHgR-vjvrtAng~WDSCvJ5peYUy>IeGHb&^3t>(=Mf_cSRRvVv6x{q|oP-LQ3uMU9 zS|>a=<_?WGGGG;PV^cf%1lE3&bkeD42h@p&dHa;Qh_~s5t549ywYHv^9JZiUff0Ue z#R-qTbgb19dGJ2RFO@%{rib<`jHG8}Lo}(zl4}GFenVF?G^j$~IZ!ScDI-97t$@5b ze1cIGGge_d`j+R!?xm7Q>{3d`2F6n*&u3-xLluI-j}+Do-02YcfRG>c&H>BT$1&cXZ6J~ z$YV!*4FI2iavUYAo-9k~#khs{fwV5N`NZRkT1WPVKc7sHB?s16c!9y)!;$xmK%x_A ziYFxeW~F?+)7!tgYvh4l)UY0=B)h6US!tR`z$y4;hW#?`k8?|xIh4vdm$kaXwZ3JokT8pipR`)XQ_lY{X zppbY9Vgo1obNfK%Z+TK4*|jK;Nlh7MH}XZjcr$%$Q{73He#e&}ebQrT?2Q8POQF`g zvvLb_S>~g;BXITKlH&;#KFr*#{UAjlg#U$@;^1VeRHK&L%=3^=+R6D5kCu&JVsvc5 zV+SJ}wZBp_M?ytz+6}-baInWoCiu*!Bm2w6@9gh&XDLmu*x!C3Z|NnLZ0H|2nuIAA1EbtBgZ*k( z`e{`T_ePR3< zvatF>i&$IBRsBt|m~=SxTgFPfQJ4ZFaesgX@;1)PDn%A`f~r)znadEnr zBs;;-3vT#C0)hWc@A_h6en9|H#SfKnV#yB?om3QA0neN7%vv+Av6Bp|(CLbOP&t`p z&|hpER}j}tJnfLaTuNX=*g(4uA_@o-LfpVqxvmz!Cg#&_sGt~KLqBalT{ zvKO9mauANm%2Tx5Jdvh-fz1;SpN#m?*05=2O}AaiqtD0zduB}#HJ9rwR?HA+K~!vN zvG2{cKxwS_@`|i8EbBv;%`MZ1Y12<;1a|i+b@U6huBjwRV)$&yl@K7Q4}<|)7PcJSip~!>00WUGmT6uu{i*q`84pH z=RJ+Va#88m=@+6eCOaX6vs|#D3w{LwkIkEvVEI}A*_r7L z9ILXpevquW*;^iAzDSa=G^IQ!BS=z>8-6P&4TYF5ecCLq+6tzI?t{nFClzTydNJhO zU9?DAV>RFI_Qmt}oMvrTY7?HB-XBprDZP>Jv+1lFSO8LF#_+#MWOs8ZIun}u`K^2w zj>zXQwS9xU;Y9zQ$vz|GEZuv>@w;BEQxd+6ZKzA9uV}jRX80Oe<8sSjs*bpYN*q^o zmD$4W*=woxQvH--oddl{P9veO8bWw(PxXoE&O#qfT`Rw!Vfto8oXSl9jjGBh3l-gS zw@iH5MX4(Zml(@!-e8s1azWa1Y{HJb!V_&22eEs#>fHToUzGGl_WC#!Fuy(FNU`V< ztce^A>l)&c)EZf2a+|xs%Ca{FI;J?&~ikd_=$)M5UZ6RL^v)#j8F!D8WM3&N+qbnnjjwb;z?Lf zrn&#JC!t>4SbkH*>(R#OQqe3_Hj>}O0zR$SAEM1a=(U0_;j~KBh$7hY&>CJtw$NE45Q!`ABrIv;XzmeRT z(9tSj1vTo+NErdK^(E(fFj5EM`Lgsy!`hVF*u`S zJX+lo(a%2^y>V|SJgL8|Dg&ZqO8~RTU3zL%Uxqlai2SJE#Wlmt!x9VPu7D04Xj1H; z1y!|HF4V_#kC4Y>f8$HnOR*O|hy*jkgq=EU4;xrMhv201qpN~{2|enXIztOxo-O-y zz1S(}XQr)PE%u7{C)f({_l8A8)|4e|(Xowi7!0aCnc8Z|-K6UpnTg*T(iB{+%oCFz zWP4Mjd;I0e@z-W3&~+dX_SY_b%!@OP{%-Zx%5E>?B)KxeaCxv#zWmr^EdB_wWAS2K z$8UXY>Lz;+TX-ih)UeRKl^D+&ts;ulZPTODssYE&n3k=;_;s)3H#KBnotO}}bOFz_ zm#(a@+ASAi?vNvDnSwA8^%33(ybf8BPpM;pp{Sk8VW`lbBw5&o(~iJSIXc@$TU7-vg)tRbyWtoR?=#Nk=hq7 zu|+fA@)RD(M7xaGFFp|cnQ-ix2l{NoEZz7}q{=n%eN6;(PHHQ^XojRriGSntXkXNG z{RW>XrGZ?%G5pl}>AfblgG^1;sn>@x;a#3Gf@HZ}7}9M3G6;rMml;*e=#MWgHgjt` zN)TsrP{v85`)=zYQO8vC86v&qO(36+7|HC_5p--pg=h5Unb31?>QLv^Q{UI;4Yj?o zI*RbtHAC}Z3XB>PWa>0V?q;ZZ>LrDKQG0t#?sk3QAyO{&49O5urDr>kNK8u0&yY@? znZ0D!ZBbf3uVP@Kp+G!UZOJdYW8wC)dloV8V5cD7nV0#L-9V`ho z>a}Po5ko=Gk6{Z0P+ly}z9P+~4EN1uW)sYF zfNL3%fzPXM2GXx-VI4ZBUBZvZH~AcpjvxDON@yQ`qUiQhripz{pEC2c6CB8RfAH(M z!}g&4Vy9*jlWH8-5A?gBTK0-THg`+KY%oS|EjcX^=Jgs?UCZ=LD^F%*C#Dm3tbRHz zcz9a`{n$~;tR*}~j+Nm~%t_41!DyQ_ zv6VeJeF`Zh(MH|#yTkONH-&sAW{kNS>qj%2BOX{2T&zmT)cmRF`h2u_w;Ox&9DBs& z|LN|#|I$3vx05$b)11^;>ei@jY&4oU5LcYdiKA^mj|Xv&s0k{#2e@rBjN*zJ3bf2EW1QxyN;1_jTQmr}~8TwB$u! zz=#`VEIosQT}+PHx@BH_y)zUZd?Ph3->22Jl)bzZV8BNVX9&T=q-0ik;?SVlweOkV z4BjTx9i0A^)H;H86+CEAoKcLH+y`Sx)_yBVLZQ+P3@Cm zU~KI2!_@w+A7b@=0g^GGH)r|}q{l;kPCKHPtSu34*gEB6H_G^)fBygFmfkWrnwSd ziHv%ln(nRweEURX6U3)6IJrVDyl50%RKCNcuaXHYK|5A;-ys~Z9u<>&}|hZp|qpbR2%?S5u%L-(;@Z?NB{2GHi5 z{f+gw=<}%;2eS!O@dl3J1Cep7n6U+#8Y5KNWq#P$8ch>A<7(xy`paK+h!|F=IqecY z9AZUNA?ak6eWUsMKjPW`C~eqnJ3U)e$TvcIJQ|FdTFwxcn?m_6$#p7ijE7HZkFS-7 z48%U_uY){ZAGut!F}6`NxH>8I;4bU?_pcLlvqkWaRh*dOd!W=kW&pZ;#H&dg7nHqP zItHABdEmM7%x>utbtlWiTH$&Z$7Vgs?6z}wFr^Cu8Ut@aP5b4X5wd)TFo@IKp`~M7 z9b8%e~x`lyL^HVv!G<=Y_aMXVQD`M z5wm{s?Q0mek$%vcSe3QoO%5T7;}>xi>AUm%yywI_%4#`d$S1r5-Dk_eU9XSC!58EZiQ z7eJ64@jsu!l=XCGH{B%BYTT+RY8}(Y}DbAO_uFw!2|9=DSNY}-|T&)S+&pFd4yXnTmYWZ$;dg9 z{Nrid$t~Y1?pVRi2nEGo>M+cFOz`qfP_k2KgdIWc-PR?0M9?|%POABAu_dfUPez{_ zz+om~o#q{sMArJ%-)f!i{n0Cyh%Tc8e;hVKAMg# zL^AB-aO~)>5tA=h^e$u9#Y7}tB0Tn$vtJ!`Uy;8&see9D4oPjK3{TI>%#s3=&+Pq2 zUMQyzH@Yl{rE&&z?I5_E;m9F&L?*-SZX!17Fn^${$^X{@LssITr+{L@j?v#A7Efy; ziB}0Fe1!TqVMgt0&LHI_-?Z}Y_xR4kOZI^WJT?9OxbU*!=dsYy+Mhl(PVL#1MS2uP zIenG;=b1l;iuC?s4Ew}fNOHOp`2ykOnl*ac896I6jVHz*Cvm&GVxT-bv4$$`(u|y| z#s!G+oonPIZG2^oO4~25HWC8stVTAHQO6Cc{Ovz^MpG3jahl{J6iNai{mT7D9Z0vek<_f`aL~uS!MhQexR(4*RZ^#k z587esc%MTmVys@y;qL@(84L|7Yi|w~q&j#2BE!y^m>L;hb*aQ-qq!2VXW3MDGtk4^G__T_F+NPi;k;s71QhRsQitX&zv zyr>NxH>s#I^QN2!Zc^C$BU4RX(@e1i_!;B)`Luk4WoT*Zt(Uaxx!ErMsP0iR+m~oA z{JBhRl_RnIJu}f}Aw|1rrROXmC_(LlLM9_|P2$D1*t7F0H)|s(Pq=61-ZKkXEj3*u*Q&&xWdYFfIR<Z_kGHWPe69EDraRCyK^n2$VF(}y6;+&%7 z^QqebsM-mu)Ef5*V-6R?2MA-=)zNbzdT5qr9nC7HY`ky1V9VRSto3ngHfh36P_S5z zyh?0l+n-36IrzWcAIQ4ltD?jwcL~+)6oA`h_&t4e=x~CuFpyTw`<^_TaBo7SNABc| zUuM4Eip~+0K!Ho6qIa5RPH5ZlFK<;xmt@>MGUo2vq(Ht#UYz_n2_pPh{c)wM?IqhT zmIHTjaMj7kAc)KC*N+Z=m5~ zRvCS|EpjaU+eu|yWsZ-#(AcRT(w0ic{7?Njb4yTA$Et`mmLMmao>vmJ}TUgv33K`1*L)XZ_BD@0(c1f^+Kj$+| zDJO=P+X~~;qA?u#SL-yOX2}eP9mS3)W*134{y;bz<@*PR?nP%-<(&?v4zwNMGYA_g z(H9>FD;1cA98Q;%-Bnns(0uEiMd&un9Y-wFAU7DrT#J#bs<_QO;wg|JjIw|~oO5V(&ceBqD zptrN%jgE&i7)f%E>2}!pR|q$4(zVa0a=h1awmuzy6ej-sac#!L3GdnA&PZMHZA2J3 zL-`P9^M>v1yB|iOaW%x*W~fxG|3=*lK{?;83Q#CmUJ55v=G|@b)YJj?X#oL&$xz4r z0&IBh8(WsanEVA@(n&;3=oGDbPYy#~g{|!mSBYmpR?i?k$vr_|p{=eMr8`t^LNC60 zA6Z~n{chM0XR|WY#YBl@z+foGwtLl}BnFC+zW&CyXJ$iweH1MVh-D^9xG?an*|4%a!$V5LKb_CBz*g2)^!;n!Q zCub0zE7gf_d4k3cXTt>SinoePm=3mI*J^xkh?V|2Yb-01-vkR2tqOV;@k zaKM%e+WBtc%TCBRqDvAg{@wrN z4e;SiPF$BP2LZJMtIP#;Qg4LI$Te{)1J{u8SwkV&xH9AG>c>vKslNq|buS zu`tnzJMlNednz~nO7Mj{!d-6`_+|iX_?)KDc#<(oMgioUjpjnsUn-`(+pY$RT0}mLxGCaL4lG ze$&lTl&XG1wJ&=|)4U~x8A5#RMj@lfHl{Tyq$Dg1ppG8i|3&a+RGQODt;@VSiljef zSPiHs9hA{oG5j(z@=E23zTMrCL*u*3L8<}aceawti$+JwXI&5&%J$AOflLs!B~JWP zt+7`Z1ye6tJmT9KAa~g_F-`)=rN!RV+M5rG6CQQ>&uFjc|C!3 z@sLL@bx1{p^3SD-WIl{Nu=?t)fYkM`Wh=6)O?1wF)rvLWCpf%`@}4&sv%g?#?|?mD zWBjt}tykXyOEKde52B{Vq;d~vFxmYR_b>jaXAiF?(Q1((o;kA^R!E2OaWeE_HJ;&0jFNQU6 zs=EK8g{sDTmz@PGs>Jp55V2jdJ%zDszRS93^l5EqSrC7e6<(r$zv3O-2tc{WoQ;iF zB8&3Cz7dlv`DiPZVx)PH(c6j6kbNq7-Ar^I+9lJTE~eDZ@(5kYBCVJ>-GH8_@I=d4g~6WA!VTPJ<)WeCpP>Sk4d+}pc^UiUyq!5+~a=@ zPh1hv<+uPI70Hb0^K{=%8A?=JovrCqYi8$FLUIYjx~t0m}5x~ z_ZiX%&hPYWw{woDy@VvvBMBQ$X!3aKSUqa9h@qF=8e_tVs{5!r5-`D$KNJ?t(6-3J zG0^!7mVcR>ns@SX?CoK{j84P`b>F%sE)i83+iyGG)N3HOc%&#i|H7l6R@}Dj+vupi zrPHeRGdU>RaWS*(I}7psg`z`7!jFgQc#`4Z5o_vd2rtl!DzvD27s@>avM88^3#AW-2q~eQx z4I@X)`GaaTHbrBOpHESfL;M`jFjES;llKG(&Mr2x6$-=2iLGHr&7BEXwj`>-HjwD6 zpHFS6q&=wTjia4D$q6Yx^hPR$xlLbQ0gmcN)Uwi#C$MLl9a!jb-I8<9JxN_5= zRJN|1{c`N8P{t?bQ_jZN4B_F!#hr@;-obId8>HpgX#xVNsq3gE4eKq}$`wmY@3*d0 zn*D|M{?O`AHP*e6Uf%BZuso~a8oAi$q=x6ZrghKy2X19kVL44Q*$saY1kUIdi9U+= zl^2P0TRuRvHQ!pyDPGVOv}+hD{hwdF*thMN-?uTmaeyQM`AJEVs4gpdUZs{uuuy;t zN~q^RYGRTSNHMM^`dsAoPLEbu9-wzP77(|$6UKnjW^`nm)~FA5yEyP2xyWR!sbnbW zr=i0baJE}>{XnZ(+PUb<)cz9RG2y2d<$VUOh7&ZY!YStDIVn_;OI&60PDl>7eaG6H z=JY8cJTQ|TEvY#BuseVirLZzaBChVPNfv=6aBC~|uYZs6FLhR4KKZ8oCm)R>pUzq4 zGR)a0p}-kzTEkUEk;@6ocFvcO0ocwS)equrjKByJR>1X*zhj|FEvOq$ryQrH(q{YX9b$2!?A1kXg$2u%*givAQ<#fzjB2bY;8RxUwhrUUR9LXb+i83yuQHfFv-^uAcCrCqmiidG!T&g*Rm}K6*Gkbw}N1HtmcoOuKaY8$Nd}XahR1GDxF;J{6p|TLF8tB637u8xtmQ%vujH zI>hVo*Ycf^2c?q(lPU#(5*XT};E^99d8R>@d^&KvaM}#V+^dvE>hAy20=H;>Z*f5u z9xgP3TTBxJ@P}o;|9fnodSrex#Cvi$qCk8O>PemRFNjOw+JV}TI3$v|7e8&vm)ZOU z9T1?^`Bl%)VRT#ML6`#V(!ius7A|eZz00-}?$+xn=1E1MSiWzN2|gU~AE;L1TtaqI zS%*iXn=1}A!xyQSKm=+(i6@^zN=k!^MRKlM2D}?7f}LG{bWzGLfz^xJ zd-xd{v5K7Ef}y^etD4($&wKW!N1eQPVK%pFfR9gj;a~ zkXjUPjN>8@?pAf1w;R4V-4K?!&}#L^R`Rm`UWd@#(sH74O5c{(rm4|n-rC~V?vPTv zxI89*o2m!LWseK{8{y1P@z%ZMyew9)VNuK_c^SjEK5qlpo9fKa4LjkUEC)|wMK9RW z+c24?4^*RdCEs(>_uqLDmH?9NYB^B!Y|n^ub#J4XymNnS^F2tK9Jbx^3mBj;w6MIw z5lm|)JFC#yP4gaAhL?)0I|Lo`9~LAV%tbUMvkE6`EhiSp|CpmQmTdYukV5yCOZ8XFqTXF-)3rD1{QMF zJTPEd$ypnTQhrLh=;N^g^PKGz7*l8FbLHu1JFkQwG>|GR#ALBGC5nwhWTC92u3YVCQb zon|T^7kU6K0UQ|aJYUL@6#z$owQ<2DQ#k0|WwXD77UN_NAo+U_4>Q}!a7$Yg z4>SjByc$dHxnJ+*lMi$(#oUPqOs!ye901vdfqR-|obu5sv_IoO^a66D4hz}TG zE}PwYktX{MMr86Tvu^`ylv>6z3ypJ@or!9+?Q|zqiTIVX__Cbo?Y5L>OH$8*L+i09 z9dJU^Wz*@6Mt^DqIjAY5aw|z&v@@u5Td6eTgA&Pn&ccwVHg)idO%3yEUds_(V-0(S zaT$&tn3hT~y?BO1$dZo~%wa9HvsU@h;;QAI&!$-DCt+j_}479^a z$>LK&L(C0p{0k@Rvfr=)=K|~b$W%?ZWQ>qGbIAsNXRv`p9E@@ah$()c$BytK1((kO z0JE>W><8$Z?^2%3tC+vn(8CNGhT-L;J<9m^TyeVn81@hT zkYf4;y#I7FRW~5aF1E}09o7lC?h4Z{TdYA)(0rEDpty6a7GN_;oEST!o-m8)%UHOY zBQWrNg5j;Gp{qMbW>D({+X)83M9g(mYZgJkAa#Y=$LRl;8Zk7QSUO;bH5{A%bNDQ? zoT*cJ))tBY)IJX1M30+qUR58gPtc@`Vv zu6G^;5Cd!;Oz{%^oEG~~6&k&-<1e;%lQ~s5@;1NQ=RXkhsWXW2^g?>I5z<0)xr;S& z`C7+BsjRTFiBR^e%jUAqmz zpmIn#;`<>Zc!WjoYmY}p5z8-g#@>N7Gr=3TceHdMq@>qSds#)j>H zc<4A;IlVLb(mZ@Q88c*jQ7zQ&q~VS(LWRk-%13}LwMnJ#%hV1E#6_W5e-Efn_Pubv za=qDNM9jBS#9cw`Lb>7%+c`CrCP^s!gaaNuNdv0@w}^19{dq z3UcT-OQCswXzjtK{8IS+euKt{x3jTc)G9Jz!*O9HcVKJ=^663efMGSu?9Iq;oJg}| zj65Tg%sGnk0cMcy{n7!w#~U`#T&AuBbP|mmK4j z%}z*zsEqOqKm-Sv#cXXOf3+iOgeAHe^@+JvQB9lB4?}TLCUwq6&cnG@9vg{TO=da{)&0eg%(E-WJsk7s;d1VnX2|}*~(xq zxXrpb{Nr~uFT&7gpUxI+m$s}QIQka=t@(I4hEFCYUDny)fmMs<>*G6{Qs7M-(2IM8 z;U6KtK$6{DljVwyAuiB;CO=#VHPv-^X*Bl-F37ys!0ZrkT=6kOu&4z}wlVB>omei9 zho89Uk$-1hvA+R2*6A!WT;43s9ouje^EROmJb?YuOIkKEg%G9=_oNO$3`MJ6aXda* zo05~%;tXbL=V|VWs7+2U8%q;zv_-kv2-UtUvZ$V1;_)j}S2?`dRErCZ)zQZ{Tpg+n zmz$_}u<%En{Ai*?w^k0q#I6CY<-qiZA8jGr}6v6nocF1_URFYYJp|8R5;yDaD>QY_Zw!*OM|jX_Etw}vIs0aZ5|p0 z*wp=h?eKhiC)EFX#Ja04FFZ$**wAU4_8g242)naxx`ACL>Uz)-eM=c4CScOGC~JvK zjW3d~GfToJ?JlrTvk+J;LS?$D6b$}!)_vIg=zK$to4rrbyzyQ0+tS|rAtH*rsu9~0 z+GR6d3U05*-!@MRVQIgQrYQL$ipD$qufv%*?GEqbn1fcDwJS%(>@Q~(fcwyY&E1y) z0CvYFFx{I^R_)^45W{F@Ln}&Y)vby&4gZ@a9b+fN^i}mD(;=;xnZ2_paX&V`L z`0Vzg*bZhiPtd!dKD6KH+bDyu#x=ju(5(>%{mieZJ8-oljB{vTck|toVu5Mh4&)L} zB-IF-<13o()orKnnmNX_WKDGgI-LU zI|lo>q=2;t%Lu1(l!(#LiTUy0V*OuhEWp>;2@vG@-m8xOb-FKSqDBm`b>?tWDs}`_ zaw46)ma!SPC;~I%JF8>nd;ASXC#>QD{qpeTt|%r28;pBJwswCK;)-dfv6`J?9V_1n za;bpG(Z?7bZS6=EA8~za2EeSaVKY^m%Gdi@=1FIw^`@h+SC@c8EynD4mT8yKVCdRp zpC4V9p24DAW{xlfSv1-d=xc>mCzw&ecZc z)zzN)Q9QBEi~)34>d6=>lZagvb;q+w(0Ths=)9^KAex``NY@F$A$K({*u}9OvXVPK zl>nTaXlNp91=tRriuqu^b{o}lZ$<4Eu&h2NT3+c(mUCIHOcevB9V0(_HqDUdxvVuF z?v`F`2l$#N&9%i#oe>#?O~8u;JXzUyzVBn3ir&|mf^V4E72X@6!evK%)umxfa^4T@ zTTSgKp&$l8IM>Rnh$s>{ARr*Vv*!uuO1;wqVCz$)gsI_Dk1MY*wz{9(9cR9d{%eWZ!GoK*34o%j#dGux3tm6 zD+UI!(FI$)Y@FS<)I4)O;{5%0W5DKDrHYOMit)DXnZ1l0?&u7x$VEsMn=Ig#g>nel z?n8Q{$l7Y)4%n!3g^_=rRUS2!?s6fyU2F<;Gsi@4jyLxbNKLKjw2^NIl}tkJbwJ>2`=e4hhM9(=;hqw*ER!S5lyi(lyt@6Jl7UewjktfwqUo>SCwbh>>ftBzH|8CJ=x@_#M- z)&J}VSYWXpo#kQ`e(2z<$E-N7W3V6K`V4T8TjBGsOb&@T{N3x;ULV_ugREEJI>^PQ z!^U{BKcD|Rw;Yy_<+RYI}JD44yc-B zLZOb|9{k5uw&%~;l>YNx^v@(_-!4>{9bxnYt@?I)46qP^sOtfQ@IdWP!dFcRHFuiF z)|`t|G;t_wlWp6aUWH4KRk>1^CY}qhdcWAPPp(UFY={X)JE@hnF_^~wXB@eqL|(_a zfnjj=J?ymzWXck$nViL6wzKI(6vxvLty4&4u?rTVJm9!OpKwz>h~W&RJk~chgnM}M zAO21yf-3hJamLxBrJ+%(L2Tg4ZgTu?!jImDXz5n(7-f^>EBOu!7H}^;v?4`F z0ID6vnYc-olxb7g*4N>Ob2^_4FM-fY$Yot9msnU}YL;lTWfZ|*=L+$>M)ny;PNN2I zJ0hqI3k3Og`}SGYF46d|d(9?j>SB4?6;bJ<7yg9SI6 z+|4vNMLPoo)2MwbtgN#iG}g#0>2{E1m#YSPCj^^oM`XtqQ!0hGg~7@mv|KC4yMVgL z%s>g4ct$!O?tEl~StQm4>8j~`mFeiijhHhpG^O4uF-x%R2Z7DffQppb?8ymt;dam9 zF9yG`tFx*bXv148BHB_eCpLOP)Tiq*AMZTg7(MIRvSb(yV|vHnkKCynQ6MR$Nnx;# zw`(INP?LchN=T%4+?Y#A6ytu3eup(SwP;Y1-#h(;3rICEs(2FvQ=ERj$EKv)$GK|Q zI6(ubR|_pGKfQ`HlU5=Flo!fmWXsDDvp$c&pKhHWkI%OpG;4A~`{T{MW!jk%>RzD+8Bu>n8=WfUBduY8mAt>qr2cZ zm5BEZ-e_S`R0;PdE&T$uU4ktNx&2G6ssOvE)5M?!Uo>}D&e0u~-0#wI&*)paM$KT- zJV#wKHDF>jCm~^7nx||AWv+PbJYX^yX^L>QB!RT+i$4KFB{ZB&vdI9T@MkkzAS-Vq z7H!*Geoy0g@)b|V!k$@&C*9{za|7izU_BldrE~zlbgp`6=^?1x;II^%Sd}uIPpu_H zmdDNl5#rl6r+U;4^zj7`_|x{--1veC zwQT$R_MPwPoDbbj0iBCNuu|m^i(_+8FW%jOtyHy0Y>>K`CfAS6ARzuhPOOFjdsxL2 zWq^$cr$-5>{Zz$tp6>MSEEmD{{&PlCFAS|CD1N%^DO9SEzQ^{6XzxY=Ip6M^GG0wu zcvdymwj1j?SO2}Ty{m%Z?IL;2KG`3LkV@V>BlSR8atk&m{Um^=c?U8UfkbCN$ceV} zi)(Gw@jkcfDY_%EKG5CS)Z+%TEWUqxPJ13t4>oY0O+959t7@fRIq+#W&EP5r!)tx> zLUNwt(n}9>9!ePhtxVbGBp3<<)vQqAad2S3G;L!@#rSgFWvN2|c>pIrN@Xr2cC z^RUEuRb_7LDOpV-u~eHv^$^ijN2cf%m5j=h4_dvqRHNJ&`Pi4aoMvLH z09VPono?yvyCr_`x~Y3nq2g{OAsr1N6$XX#E{j{1727pHvInT&y<*M*H;^7&d8{}p zyA6P;wl(!=+bP+Nrlo2>8^%pXk#+JKrZtOuF@ZIU5P=*2pbN4Wt!u^6PMqp+TDENq zxFjJ}ZgwBiH`fQ)H=8bq8&CDhkBdp5-(EdF(bJz12SYQC4 zphM1-WH(ph#6v3z7G!EfS(nxsN%S1KKw4=W^^3C#Fxh-)Og74}X}b#bw5oTQv#&u! zo8?oK`f(M-7!;I6a|;(YO$5LgR0#PS+IhT&S4ptO%~oao0dqI=pbGr)8&th?VF1_D zNEAjX)^9v6&eUKtcfwO~TPbvXEgL=0sg~`a%t77im8LSWb5^m^O(dU&5J3}6MTd53 zWo%ZxKS1&#eiU{WO`vFBsMI=dc6 zTb*rppLkvNZ}Od7d;KFfLea$q_r(V})twqZdY9A6KC2rN(6A!9=-WSK=C zt*7V)oC-CyBlLcG2>*XY?B9T?!PPat72tUt+hlO~Hs+iB5rJlARo(}d3 zwxd0b6O|7)vR}Jq*D^qNzejhM1IN%ROS`4oMUOyoOUxq)p0`bVfMnCA$|$dPCEL)i ztCT%u!Ur{vbms5$!lMY6EAqzNyvp-WA6Lz@jm8g@NdbV66r%YI+`HfSEErb-=h8YE zpFEPAB~d@(35F=?i=h1)mAomNh1`#tx9icR;Xdf5-u%Z6&ctGJ0o2VNi^XPDK+%ib zZM8V*9qaoA;Ync=TtzwS@%P|G<_Mx-;@lK>@fmI7+pU0JjnYwMT_@jk6$qjS1erE} zu{*mKGRe`KNgU@uzoTJn|oT& zuYqh4HA^qi-PS(Loe$o{#SWxD8!i>SdL6BG3!D3p7KxpGne`3PJEt3)J+sGl?JgD# zhUe%SQUl`erW{eZ{vSuP9~Aicl~Mv+Ir!>=av@z0No!l^!7DAfppw1{5Ugp`+r_EQxXrqctu6;n1dnqdk;xiD`VZO{V; z@uRouXZrH>{2;a6Fz2KkWV+?FY1Ic_#Ys6=-*om#iGKc#x4>wz{ z%B-THVUO*eZ?{C?%4Fx0JZl29kx$kH#5YUf=>!2-{b&o_k9%C1pM;1=Y=GEFaLgm1&1t6 zip~t}DdbXb{q{e3cHrbmT22d*lTvgcPg5 zvV$u#+81BSJM3;tM~g5t#{b!txTL{^A1^%oyG8oV zKg;5O@$*;O{?GRE0bPG0c>2#|w^(x+5{Fezhi@(5WP+c}-^bhe*6*zA=ChQm&g3XG z=+83v-~3$hZF?cKy=_P@p9g^x+8#W0yYeT(vpWuihs2_mF{1NzWrcI~)%xT>3g_C$ zXC)dz+(&iKMtKD`@B9Xkcf23%op<2)jprlp z{05MBy&rk!T|D0LedL{Y;P{Q_Bk%kMkaxTvdFLHCe(U+jJHJI_pZ_E8>;vOH=SSXo z50ZTWA9-gV81Fm((|G3~fAK~)r?dBoQ2!Lr+{9BFc6E+wd<=KOXIkV_Ck}3CX|jCZ zKx&=yhK+rUi1gh)qMy*Q8nYaXU#xTeiQWUMrrYQC*uZy<>*7}r-FnQP$Hw{a!9UN)X);ie{&bRwq4wddu-AG(8cw+n zHx$QNsmCHRcFLAWdu)pjd$-4eHI5gS!=5`OdqA01t6-kq+|3!D-O7eW(@?4a6_+5a zxmn?<&^W71nM<0JKU+Jhpt&V{E9wc;qd!W{qFg;J8l|SGZI;+?w#hX9oF+rOS{i;d zstyc=XA=Vy2k@HG31%g_p(W~Yk)sp7)tubnO%Cp(3(5I|g+VE}fi|CfC5k46`;&3B zQQWyot2t+4k?jM%{2uSQ?jDZ;#f6tBe6$x&LjVjG_jO#NxDz5bMDSt>vou(eeu4sM zpafq|99dMP!p=u_VJE9krEaP+hN$Y9LW<^bK<9VmcGfStC4cQB!j7Hz6_N-PZR@42QIy=asw<(F>dIyo_( zkw!TUdeS;SH#bNJsf1-s9UMpi@qqjqT-pqEecCJP(=^U2cN{OZ{w9fmAnRFd)Y_-u zCQ?T9NP70UPqVUbQ^>fgQ?Ua{U-`gF51xGn;MFsKl5)v?FlBP2L)K(jmsCe>aO9RU z3Bd@B-RZlr;2E#h1S#sxMQWQs^Hr4#m}s-<3>t=bU!cb>6`w6WUQ?>kJHSC5xeyFN z+LEuEL*)?YHKHG}yQC{8vr5&JD!S^2uupbE$_dO?a&R8MRS0RF?*U=O287hhU9nKb zSm$)7ma6YHK>Evsge1mX3q94n@(b!+Jfa^?$Mpc}usU<#Y#rtMKvD@Xxv`oHqPjga z(M=0=2yY%=59v4wgv97YdD4+nDA(+}qKvVD`yB)_!IJ6t9W6GG2r4v2<=k&Kwfuf^)VzTcrIc?O(K+kgLbMVR;My~k1f@jX$wp+i6kUC$d0*= zjOG;T+U45%72Fm&-&5a~oAShZbx^t0x%)c0miAJtF)ZJf`%bFTi>w5z#}g^JAX}9O zFjYhP`MSw~gKASF7dm!%SA5r>WZH40C5rW`1w`j7c zX6KDWer(9(1NVOHJh=(f$ zNJOSz&M=oZRKZD~3~#h7XZ-d^W)3CVP!t*V_mw^Gr+kxRp!!yY{; zOVx^+w|;85TRPyUyGvFiflS6D5~rJI+lf0$h^`T3F_k0uGNRv=$9;VT=7x>fF+vZ# z)I4t>BF#5qcv?wCVv1SX-GERX#_#-)#qHflOuklUl^Ho=ZrGnfFod@l8h_{f;B(7< z+{n_28E1uoM}1NFO9gf)Y(m!#3xUADZDEk(H-UyKv3pAK5; z%VQ$zL0o-XcBt#VFYGF^E}0lMZ$kWURLOo%7eWmxG{NuGLNhaF76NitI=!iOBHyHp z80dXF720OF_P0LavbH=1Q=ButVdM$knLCWr)EopMTXh;7({Xi7-tl#d{ef|nX6T_n z+_3XOI+bG6g7r)(p;4hOxyi=qP8kF3BZ!MveJ-n&%#`I#i9&I_s6{VB>k@O2B3J-E zk><fV<0Rnme?XN?$5>&NP8|Dc7 z8ehc{@5E?5tNIQNF*sGJtJ|okYph+-UL~}9h}38n{|IaZP^ei3PN1N#2F9j( z9D_NQb^^J$g%$w-Zf+5C4bMHc8_9q*PexqhXn-;{w|I0}(63PoZ`tKe#|Gw17wqtB zH}prJxjk@pE9E7;MBhEK96<&O7+RymB65>4Q0E=dBQI)-fO~9})?Y2d9rxHU;27gp zNvWJ@ZBQ)(^VAIz1?*pjE5o^qRnCpaPfG*!Ka-ZAG&EHob1jKz{$j z-qab0UHQ}53R@(qJZE@jlal?p8Bpc!<5sb*p9Psi64LNRe377NL^fxlO+oQcV~@9N zlN`*Ka@XuWs30R_(tb@4PcQF1afG$Eve_9BxU;FMQn1JNjU~#BGD7gQDz9*^&%QX3 zRp@^eDyG_841DCh)H|`X#|BGb*!J$R4GQhC?Xj(dtx$lV{9}4bhJ)70q8Sqwpwya- zcDd*cCp33x?t0ja3pQebQrmNP^}HZq=v0$?xbFf$Ss})q*G1Wj;F6hJI>v-Ym1kQl z!1LrZKF61VP&Ca`?GAG`4I>v28hT3hQ84Uux8H~hmr%uuQXGw`#y|w7^fynG<-xx3 zo&ejL<$?7O0pJcriYeTf%$<4Pv*cyl^tX)qKHF51cWTDDw#9Wr4@XCC$jiw_FP~!i zuwiI1g`7t+N(Wi{6fPf+uATFAmKk%wMdRIY#qik5YOC z`S~ApHXQdD&*%LUHU_oyx;L(Qkd4*?b<)P3+gin6@(%R2I2rRtiC;b2w?V*s-?&@M zEZg(%4(IL^YW-OT|CZ<7M<>1cbc!x|<~c5^1_v0%T`LKl;319VBD+u+ar>UBEF=wiVFGoMzd$1*$#9uZM4nJE!bP_<0)S`mX4uc^1vu|$6QPmv)8 zH%`@$8%W-7u3Q#m9lT`GV3e$S^9>>34`olLd$qW<<{9TEkG`|Ay zS83jg_!>aKYxv$1>NR{P{^*5%Ti`#lcm>gap}-H<*9ot2=B4=qVa(UfcHn3k2uD^P z@@L{1ql0{da+RPN@aZ54lp-cy`~l2~*F>>#A}CNSr&xaYT-o~?_}AEe7FAwNK$q84 zuh0KvxB6fu3g?+WIp;ZFtubw0b(OfVgWqG*|8ePNr_2wZybsa-kR#3rF@JR7%C_;4 zS2Ui;rO|G&87etn?&C6a+r9cA&|-h&U)&!NzDm6QE5ejNMfhvRT!Ybal6FSq)*Jxg z50wmT6E$wwS(9_uxb9AP3WL9o@LyxDj>}x7d)Z%QZAf?}cd*FrK7_nlIY8c7YrY}1 zmY3`yS{nA^x%RgLP7mKXCH+2HM}^aRBL}^kDT_Kcn&jFUy7Ov|)9V@46Xa}kVnsE$eP?gG!)|BRS?t?r1wsfwk->#?XhKjyYZz< z(zl2HSi&#%F~e!1Mj?0QM11yW=vK0FTA4C^7^rmfz4U*L&9a~5qGtk$=&kvXA+$f>fWcd4_>DbG|7Z{Y{)pqFJ-oja zy>YV(Z1A}MFsRIZZZOpN2|BzTlwUVgDP>`g?Z<IDp8BT6;aaSa|z>;Ilw)P?1DK@ zx}F$9Y?4rn)J0q~GeOtO6r=ZB(B)=Uy%Q$!I>>@`$%l01V0?`<+ z*vM!;4lf?8BijhzOn9&7U?VdY4dlOAUQ)d0P)hWC;K~0*2eU>FH|d^W8I72AbfZFv zNm+MU*9Z9QKHnvT8<|B(0QT54oW>(-HYvUy(p*(};T~Rs!o}-0`VJ3A@~`JM*_YQx z1JoC!2#B!aD2$GCy4r?X2JYF(h0&$T8^K(K&uxk6wkh~9F1>OjUdhMAL#NSZrh_c7 zlDGB4kbcK3Q%(W$RuL4)F)VS?V{ zdL|xzk&#%aZ?fJz#RJFSs~sRMAg0SxpmLOZwG0?NgNHO_9!j2}PAO>?M)zn$f(S~VoG9qr~uUSlokOsJGf1_8`Zji9Ox&=66%Q{lm|pFRVyV; z42E*HdWocD)Q^X`56`W6s2JR;cUOj#2E?%LQFu--hoEK%kcug^l?3bh66YkH7f)^4 zr3QA_3X=&=Deaz~rRE-J@LeqHhNqkzxRp;256NK0;Nq62PXIBnB)~?(xCH6BB65%b z_AjB_#>EMnkWIhfjfGkjTsA5|{$&^P{e_YQH{l?%GAjzrXT$5psBlpwT^VkXQ zM^uq-#xwFA`|_6MXryTVG+0?g;hj2bbh1=>AO)2&jZ700Z_S&#%#NoSF^X~p)lw)a zP81Ihe{%-8Z_*D3)2UZqmqhEM21PuP^Y}oE8{sr!&v_cDVxE&fI#Ve#2jhz!Uzu(L zag896*CVRz7YAL7!@6;U%JkW!uE}I&>>N;`&AZ=kCh3WC9(2w+6XibL!e!WC*Z>je zxCaXo{)sq~EXNS}3NhP?vni0Rr`q~afl9{kY)fQ%;8J;8(XwRQk`NlY6oWr`I)sm} zU5ttnQ|uBiSzaM6cO=>J6Vn|1w=}r9MK^#aJ4E-`gr%3PPp5$67EK2-gT>a+#j3Hm zEPvvepq=P`DOsq;$jh!-I(7OdtSf8pjiY7!iH*~S&DZITZ3>r`D?+Ol&EK`1lATkp;DsxWUQ)N*J&UTGV&xfspipQ zsc)R|^RAluowP`!`vQMYz3*Zl=XsRf;LATE(~sloKSJkso}JXHy0P3;J3Dz$ykT55 zy6o7C*_2e%rUVFZ1TRFr9-_Xy9B4J>ju>;qw?8}ipzpFJPJnqiytYv_mv9YxzC@>c z2INZ>Xjp2Nl-I~oCO9bw=e0S@_X*)S;|p=cdaYqr{VNnxZ;B2uT{-Le`elv2mLgKt zxprxqjb2%I`ovn1^U>(uR9Q3cg6VS& zR>RgPmrN>P$H_)?fda(cpNLBn&fpd~Y<)%LMiG${Iw7c)1cpxZMi4_Me7skm#UJW1 zmrFNt!nhz8Ui{?e1ee!Srr$>#*<-s*GEM_zAhM0!t0@!c&+)32TPoVeD;Q5XK|h!Zvxt5Oil~RsgN+j5 zN@SB3Ayl)DT<%=|oTCGRc%cB$JVCfnMo-9ZTi3w5dKiODM00@?6ft{CaG*z!|e$T?B1r{JJG85lvHk~~meH%#F* z;UR~lR)DOQeWfi#nr17Wg21=vn#AKu4SKfOgw8q&+U^vpK|T8LJ^bx@dc8v)y%0(T z=n)aq!xRvJ%q4PU&0H;3zK5G|!ZY_uX|h9Z>vZ4A9|@^Oa&pY{lWq9Ebxq+g&uojy z7oO=&J2(44XcXeMV&CzAZ4SpIt2sACa|Eot!dxgZG|xpSHF|PM#T~B^W_Ah@)h;*X z)#wE;hPswKf?z5Z0yloV)_R>!fmCojY2-ZW`nt8EWeK-B32dmVj~^NT4)+bpEn+MJ zJEuFSm1;K=EOxH%n>us5W4j7o-2s3WZ)!;LwRjo0`Ibtld-cxX6{dG3{bcd!5U0T- z$GSgzZtQ`pQagw89Zw5fwLvsrEs`y04a-u#EngB--)JMH3I^DgPh%cWL~fu4I~)=7 z?xUlte$M>v4NI1}XMR%mg>&IDth_PqKF4x0mbz`DkRm$Q2pREoHzzDnpS!e2MM6!g z!M4ZBG5c#&4Yk{J8L*{wf5KYlA+5E@?PThyV{-E*Ytbda$77$lGbR<9L}HwBSM8!xf;`}gWP#8gmD$Un z(C@8(UaUT{af5Iv0Q0y~4blbzI>Y16?fi}K<5&pi5^MGSa22QEa9A?k81(i0Pkm!i zI$z7a2pBrJT}|8k%JX8=y66P69Tm4}i&M>7`ED>gc0s3YHscZBil*nSU3<6b$Nz#v z+Y64iAv(aQn%@zn;M4!Nve0Pv4jzps=v8{KET;ipiZ@D*9yKbB@x$ax)^n5PGb4tZ zP%lR_x{e2ls65?clP`UPD<1pE|JzxwM$tEjyoDC@&A$Mf|F>`c@2u}+f?&09WnbFV zge9L}%82I`FP=g4LzCN?{T&MK8TWEZ5=p?eYc8Dmp zD`b$9)6cW48di9_2BEjD8V?4Sou7nF((R8tY`$P{VSN7mjQ-kG&INoV9ca|=Hs?}m zdGFcqqQ5kG>199w8VId`ri@Qw9lzOmJeia+6t|9uKU}J**DDQBi-CZ0dq$tzmaDkN zLI*0G6D;Ba(_ue7a~YiE3=EuLWKN1z<8lTv5kcHSE7q);glwar;*?lTsYY&~(_~phWd(n@@Mq~f zibtUBgR=-4Y-1qyr*;n*-$Ge1gcg-Y&$L4d; zyt*17+PB9xIw^q$Mc^dlXnhH$1{x1blT{es(<;-Mo@tX+jo%e=smS#guy=HFapiUB z>gTOF!f5jvZlx6yj2VRn$D$KT=!ZkNN-eZ6f)-E3NT9@EA@_CEVUZCP;z<(j&IT{j zOoT`tT~ih(GoBTew>u$Z9WEeXvwhi_np_je@pzdWQMqLN*}6M=Ucg6c+JlQKja5Qh z+>Q>Tl>~OEUzE)CbJ4y_3$Tg4tfZV5@heV|JlI$O96y-B7u_D%g;@;aYRR|`qrJ?6ngj2(YsEf}oXU^&|xg^_i*zAdZa-dCOp z)8y&^ihPo+Gj7>@_0*!wK>L#B4U*ge=@dd=F5zBDo)1O8f^6jjSFbyt={%ui{^`s) ztmIGXj7X^JJ6jG%@+n}lBrc%gr{yI z&qjfy8{CM;SXl=K#Vi3aYqj9gQ!A8O1X4(0fnRz(OIObeko(X(buDbG8N>AZ9x8Xm zyB7=Xx5N$#^hmC|ptA1TcPO;Z6@&;nJ)GUCiX4<6zzewg@j+<~bH?POzb+d&U*xD{-iE~s*8H;QEpuC+OZJ!`x%W8@7V zU21Hb`;y=LeL{IBgoDJmGCy!X?3DQP=X1vbQ+3`UDxi;h!R=PZ6!8>q|C{fmd*%5n zFDP-7*v4a-jNRu9zI8dzy-0e$*|Sl=KnEn_<}l-!nYIY6ggPvnNjrcn1eq;pjv8hffNn8{> zuE%s;X=0Df{~xnqYnN4fjQ7~S3Zp)sGIfOf*d-|ioAYFdQ8s<96;t-ux-2Gl&-=-m zJe$DPsr6{}srBr!T`gSuse-^dZI;*@GbG=-9HTck-ALuORep@hY|P1+m0p5(wwiM* zg^0&N&u#c9Zz)H+AqresbJH_oSWtL{zx_W)5%0Ee>obCYPP-^nNeh!6c38jmIKwV0bkq)jW`$BFq8Fy^k*-?ZoFN1hhAh1-^QMpIN?T&p)egRv7_l`B!`bM!9xdT-RrhE5<)PmolaOL?14 zzJPb2BD|dx5_&ns{T?K|O|G{ArhK8KGa5^31aDFe4n(FbEVIsN#Fa*2OH!dQl$b}a zaLR_A^o4;rYvk}Wy4Q_EB|Ku-yTQbk+Ys_7Kq+Nnspj6We_`PnkJ%zdyc&67!%RgU zlj~;^Dj_xtG^O%V078`sZNe95DJP8V?|T%;7r+Y~@EU9>)7LCL1kRa-& z7}gb)-_TyDEn?}dyt=T#_w&3YtJk<$XOT)g$}{ZNF1ngSL{zMn+aKSy3UJLh3>-)B zZW~$RHB|V$*4r!$PubtYpyqAT#8B5*o6wn)#*!&Ts%!EC%0)Mjw9RV-BT>25r7fZR zI=LR?3WpY+A!3zBi=buPW$5^ooF+>mT`>i9ovFB6gp1=pl8|k&R2@4$%M7aqWfO|% zYJl&atp`zz#28l5_l4F93_BPN-i?5Q+uNYE3gN`woxyRf=I_X?4?n0g6CDAGP}zlf zNT*Qc!{+aFUg|pVZ^_ei1nxCwqi4PmaCxSt@jkup=XtYh{Ao1LO9bm~~pMt8kUJI(^1NS{rkod&@e21g#8M|SDJv!BkM zpSS}S@acG7rJ2CxtX?*a^}x&#{RMT1jG{R~Ndkf?v6~RtIFy1;#;9c_pi;VhBP^Fg zgvOTg><++Zy(00pI3az!@iAnq@8-~zpC;GrI=Ns4{Y}#{9(2JJsM}OZOb5~3nxNQU zPl4a6O=B%frdjSxdQ3Sh@3F~colJ9>!R(|tb4KFTV}BGnPP)Ssb{iC#6iSrXm=wt@3b>w--U100|32HtCSo6~_`K!c-O7uN7@!Z<2) z$1AT>V1}EETo-365$I@1(GOKSie0>_r;IM&a@{)P9b)HX2XqGZ8-YErn8)=t1b&w+ z``c!PrUtr3H5#(SxaPDBTHav0S1ODjUuhe)<%IULibk9(7y)VGP1gfq+AzlR=k1Rg zdFoHHX0a39jr1Z~fR(bUdA#d8L6Q%kj;PnCACu1hxGdL5*Xr^1PQ%qW9=BQQtGXM7 zjd+2|Nr|d1^S$@zAr+!7m5D0tw$AP!Y-b$q!yh-7n+z;5UC_Yo0E`bHr=rfs(|WQk zlY)bY+@79@zgMMIrP#E+!0iyLDxyui9dD_#HcEU-_=z7qpxT%>b&SZnv@xDm)C3t% zt!90=V2@1?*BDYdFykLE5S4q&6q3E-hqINmzw=^5E77=5JT7HZuWY!J>*pN~ERF;7 z!#lr^kzMR!V$grQukvhc6T9U3kW%yQ81up6DM@{%XYfQ(%{a7rHCs(-F%UYj_P68Z zA4<0`>>vzr!GG>qO*q#}SO3GB?V|6PyU6m=p4tfjiHoGB@pNCRUz~rrt-{+rM?{{N-cVlu0zH#2+W1_zd z5G|WoL0ivIS+;m`b+^vpvzuAW;U=9EFa0MceNK45E68XPUIs!k19B(R%Y#;1y0I@) zCXm+8-Ru03G#zEDJ+`}(YAI9h&$pi*`0PFH{>Uz`PlEpY4j!=Irq9nQ2T%4J>e_K(Q#?@iQI-Z5_rqTBf8%h_z7y_f9W072k8^SIad z*c^|9GNRYLL@?Rc!6eqCVKwLZxGQlEErahSeC7M=9ql8A5tFGvNaq#LKy0I zTO*gno0|yUfn=v8)!?(%Ki>>>4|6tHxVX~4Jay$yHhLd<8BY(Nctr}r6Ji00pZ(bYbkcRsCEy*uVy?cNYn5z#M}ugNIcNfWK* zq%6a?)OEfS#Y|;ru>HwV3LTc}>&R0@32gJS`x*b?yZn!(|D65hCm&b>`#5|0>JuUV zzVA=4E`s@0<-w}2jY+LvjL8vRmno-Tcs~PePwU+NbVo&VqxVk^E3vaQe`F){@lM&1 zKo9Q9_LhyJo^5#mDQmx{(QNRpI@{l$zNi1E?Dx->`+CCGs;#AW^`+;nd=l*WljQ8Z z*@)l2m+IXBW#@P3eq^`l)DDe*VN?rbG@Ok@quoP9z(Tb`^$C^lrhDc4>m7I5r}4-p zPS2V$+C17z1edAuNv^Sm>@Tdp-M8BwKO8r{mR~CA2eSf(B=KJaP zhp^){_D^6rb29gjG)et*`0!H7|cg*5CIjeEWsq8#_%MA@qnq@O+cixCLw&_=NR?LExi33#RVq4roE5B5RpJM+MzP&d{WxsH>3&e8Y#01(2D zMmIlw@=nrMo>!~-Z)^PHs{X4hFaO0yHaXtvJjUzoP~dTFofM_L?HAhDQU1jW^axyj z@+T+S53*Fhe(BX1;M7vTWMEptvQxpjog-3zzb7ic>X#zfjFP`#4?S^8rxq?5@NT+SzQ5k)?R`oQjCnf} z(DmRlOGjA!!uuQlv;uuP(r*6Ar5vIx-4{(}4FPWYCyi*37#V2Uu)1ZTR^@#j62$!l z_=iI>@5BFbX7SZ6Q2*+&Kl?H)U|Y`mZmN;OylV+J7i*C=(0MZQNY@2TPul!x1DXxf zNQJQ$t3r!6LU;zb@4+*jybQW!(!#>2(nYnm6)K_`ct zDi&%LNrks18)tu5JmvmE;>uLxGRyovH)!BOT) z%bPr|zKs$9NV;%!y{cH`b8;2Oe393$#ayjC-F8|b#IBF?)5*cx{xsD$QS44Q#hlTW zcRv#06=+C8yKWqHoz zzRX#4k{lSqtLxp+eAx4iE;O6bb0{Bb`sXauPmBDjT#~;w^6#DR9r;zcTHa{ZnUPy> zGW7aYvp52-1pKO8sjt-am1ceQpOb0F{xxc&NJd4uS?3tjqcWN^n&eXG%*F4_d6#H^ zz!f`)+ARqGhIJWK!}ynjhQ`tQqh_MQ7yf{tt0<0&mW;;j2Yd3|WtVJ_c0QzuheUp&KG=r*7DsPCt^_7{l?*f_ouPm#-k#|_e*UR=M z8T#Sx=Zf0xQz5|zqherUAfPeWobNTCXTjFVWB9(`UF2tpZxWja)uG&&& zWtD2CA2NiUiWcfWTB*N`J^tHP>i;+6hJBx2X?`=Ta!MbVG-jrW9*cg&_u@cZ043bzqB6Gt1*OBHMV;JiBKVvwf5JYDTRQy7$bHeN4@DFdN zG6LARvgOsXlm78Zk>XMd>;J>|vy(uYMc>QBY8JUnJiK+)Wg@{?=FxF+V;NMS)88xw z0)tz9jo_PG_3rHbN{Krx*TI zo*QoJ($XT7y4c=9QKxWI*d1Ssf3qDyaqTs>p(^TCURdbwiQb=uM zM-5LmPLwf}RXEL+`n7s-$j#BecNDx%y6v@IseSyUoiu+W{$4#jj<<~fkGh7@#A|-v zpx@^fHV?bF-PBXe=j{9Hg)L%;E=+$wBjwIhcMg;221o%1ImW$WCM7Nk>Ng4d4SZkz z*Re;1ys8>SL)KQ5_ZODfHzJm@zn=bD@RwTo>Yugqe~~?_YD3}aC*Z(G0e^3Kjjq`u zF}ScyMt5Y<+T@`%k@q?*eHsIkYR5NnAMZR!9s;#is#yxVfJeo;YONQ?h0x79Aq1wlZSq%6IiOQueLwV&rWlRW$!KBD6x) z*CsQ-lxo;i2mo}Pkb5?J@2e88*wlF80oz8OlA&Fk%P^#^rA_3UkS+vz z@KxG5#MF;mSP7-ld&Hs~XOhIF>Fx<+FRwdorSUQiPzm2?-_i3Uh8tqe$uEz~sJ}}4 zh3l^Bww>U?y9K<_4YYa-fcn+B^`F;AR$;dwy}x(c<}~^&eBF*s7oBrhSJpLcWolC6*`1- zB&yM(&S9njI^S0ai|EcxhDW(%Pa(%|;SdF2V;OxpPZ?ohNTX+@WM zVe^7&p0dhOPu@EXmJPk-0(7El87^uJBcb+%%x=Yp`09_5drNjgqB8ddzBtZy^iAcw zajDdohw<1NY>u9o$hdID5zh*XpnG^cKC1*JFY)n#}RZs9j z^ou)#b^xZ$3(M0J$AJQ`0F?1C$hf*Ofhur1gE=njkL++b9=TBHKU3{>FE zO-)#=O$9#E^?hdfTEAotYrqRx2Ry2fr_#OfvpPS;7SJxQy(n~aS+XX07=rcq7MWOV z4zf#3Osf0uq`}t>59$~DJ2?GmPn4mS31Yi>=;oU$`W?pwjj&=fW$@L)4W^h_SzRA8 zGcrxpuj8!ZjyJ`+cwlv46l^!(d5|lzsY#(cRpK*g)U{#V=>kb5?N1^owo2)K_@dm> ztwi&_Noe{UzMhh0N!eMkAy_MGwIhPdS7yUHnZA_l+X8!R)jwd}`Nowwfo(4d$xa!a zr_*$DYr=6k!yUJ+U@!^I|I^-=M>TbBePgX~TZ>$ZfHJmNp@M)R%re-w3{?ZXTnO_Z zi6)YefDB=Vwg@t#WlF--8#R#t!7wLbkRc!>0U?GECIKOcOaWz{zWBEHTkGDo>#p~o z?|a|1ntx8tKF`_DInUb9e)ieV-oKrxH)C98-i|s2XJnPP`7U9+J1S1bkn8bpNTnt4 z0_auZlhQOa4!dw01V1eFK+l%?3A10dFD%PZ+T6Y$SamCD7+o$W-B^#)Jg3u3Zgd#O z1iB_vjfD+30M!k9q)Ooh(PoR@KlELqo!T{S(`*-!WPa@AWQ6%F0%b!>-LDJ)?N^G9 zB)*Gp*{RsbOIlni>)@q6ya`L!t^kc*DE$ny?j$Ah-@+jSSC=qAnHp6x@cM&i1oBFV zCf=H9I3R}b0RJ?1T3oh?fUhfoCM&f$12~7zUhb*VbaS80zQ2=m?Cfx0jA>NG)(RLP zKVU6&H5-kJo_7GNv>jJ&So)E*Zjm7yW03haJr=D%7Cw1+uOnxZTmpd?Ud5ub=%eh1 zHw~0&4XA{#iCAns!WzAi@act|&yFPO!dUL;+%NkZdQg;L`@kwkwecP)oep-GGx34D z{sFC~8Dz+J-Y- zESd)`GK8DP{_+jVvKh}quoY3>d@_NIWH{ChNJb2W&t=6xan+$54oVZ6H;r1>n-D5g zP!YNw9dqvkwbRh`pjZmzj*9X` zb@Bk|RB3R<6`va4-j%K6g@Y+=KW#-k&lzq@W%!>eUzo+F?)?3m$y0q$lW%bWn=A2Qz#-1|avfTk8dR(IH!E!lRm!R2S*nJQvY z(Xq=exl{T=9}5*o&~(+nBEJ&ZzDtYX7Cz|TdoDmd>I&r@Ql} zD|r@l?o^3|`eNTiOHO>_c!tJuE%3m%Uq-aqS0@g^(61?K(L1+LU?tKkY97K!eetHa zp{F8+wiM9eey?*mXq@cK!LELHkkbn?=TLXXE-V!)XPPm3qtWt!-fE!zYA}*o1jqL;W0WpamUo-Xvh{GoJi11r zS4oTpbc6Kvd!MaO%AgGym}m?TKm(79LEbvuF_f5XxksR#@FAmhBVjQ5aI3K;>v{0Z zs<$96W6v8y<~D-m9vwu0D{ZzlaA?b90o%BDiFi_@1H`ven9TRV&e43TIM(53LBey969L)Ij?%Tj=@MF4Jv?dH**xN5nD_gM-00IyXd-yOgR@z>&(spDB-mCVVFe^G7VfpuQA?p5I zVJ#PpH%pWPI%p_ZYid$k}21Lt;!c!J>Mh zKDKb=`W`T&vGC^mTiYyS1f^e-*Vh!M9XPC61&a`h8Y~C<%vs&5;ShrnK)A$HqYI>v}5Cw(Ue9G zhETc!<;N?^oc5?3b)XlB@-AGf zPv%l{baqiNp|Qdy2U0b+V{4Y6&B0gHrq_p-Q?!XC5X3M{)66Uab(M_G_8fd-!f9+u zUz2-!S=GVg;Agf`Z44@O%Lu4Kt^TpfiVQ7Qe??8S?wMDXiu?rY`(X?D+DargqIVUr;#@9DtghPupdRgFR_0_1$cX( zWiPFm8^>I93^Yb$nxIQAr0*sqWJMcs-KZndtT|_KwZW^)Lbb|8OD21~%XW_kwkxY#f`@_GuG0qWRuG}-7 zW+SU8QfrvL@FR&Q7kRL#@z(*bjRem-d6~>9!saP`u*F{}XlPA6ssLvNBQl&fXWXhg}uS+rb%_0alcsLDYMxL+45eAHN9o99rV=# z2+)EnA1s~R$?c%f_f=(8ffj>mf4wa|KkH#RjGF96oq8@38B+^|6Z_|l*n@^$+>UI| z13ZBe$iJGhJ-T;5x(STkU1)zcNjCq9t8AIV-s)-AQFhAi^yye4Iz=A#AX=|>f#s_A z*fu5V3spr&dg$YQQ6uVAr&k*yq7joy!7Zd^1swd5*#80jjBcx7 zQ5jt0nPQs&!7{zD1hJ)b;tork7FPMK!%9z7#LC2+BvS}FEvIw><=B)@UH=uNVc03#0aM|+y_;U<#0$(|eFmkLjh(-2)YF6f+5nooZoS5=8+E&Fhe zU9%*4E2Y>Nn$-FF(gw)EjjB*ivtc_@dtE`H#|vs4Dg-6fW1luWjK)nEdUm=WNd)wv z1QbSiH@&-7?JuGpOBa@Tn2DcztL`LZttqHYs)*wrJWk}`&q@0NfWSxgCK?=NMe17h zA>yV*+mVE0?NS>?Tnch&lCV*N?bY>dbOqan-_W&nWru-4F1=rL2fJJzwaB=+BJ>92 z4dNhGV7br>aIXEhmul)$rY+UIu1jWyHjiPfLkbb(w2 z&FU8`1jNpu1bBL}+hy1~W9w}Rf@T^6A(L2ufsLeV^SH;#jHp=GL8o0CtN)&Pt!b2{BsQeI`LSzY~F@-;gSEUr1vKg>nW^?iTwTZN&OEwkra8HLN}6;af6q+>V1@MabL)enyvUV@WJCuReaFkTxxb=Y4qr7O=YT zFcj-v=^zWJ9JI05!-GKVu*FAKEu+hnH^$26GZp+xz_D_n8kKH(&FAB54iXo7oouZn z$Gd}wY$(7T;K3>BT{Bq$chE2^k`oa;OCZp(=FsQ*$m)c}y`6~ao6^~%?A#!Oy{7$p z@}|X>6%%ASF^zPMpb*3S+2~_c3#b%XXqx_(? z>-uqzWy&r_N4hqu>`l;6CSN^)<0jgURd~x6JEG-W*>(?(OXI?@g&csjQ6+$NE+u$oWcnQ!%emrAl7_g8*+PYr%#Rk8dT)(M;bXD| zywI?nZn!CyuP9bSp)gqV0?C7)B)2h;%y_4Y7mTyyO)FVn_EDSCQqvfz?PD~bu4qJB zA*ZV)=iue@#6qvlZ}yFK(8ggwflSAN*9c~}LSV)#9?xDXnNoZtVOX!p=VeS_c6G#M zkNsO}ci0YR$W<=YNOF}?+PX-LpFTb8-J{(b8b@kKw7un_A{E4Kh%{FF;4QA7>OxM( zpJ8h%pG~Qj&Y&V?Ax%S!$lK`)WOGeT!~4j})p*9Sq}~ zCr_J}Q2T!8b{*0FH1OyQ06_Y*d^oX_tDBK`Q(U=w%$iVBDS5mhaO`rCj4eg2cvJ*2 zx_NBi^2qpP_Y$QYW+wDJv(20iB%n4~gJ6sLuF&oWLp2y{^gH;|vualxMs70`uR%h> z#*ApH)f+Ue@3_y$%(A7Vu+zwn*P;f1N6h=_(H680p-Rio3z2qj_4wV6sgME@_M&C6 zvD(WJ)M&kpvn^myR^`6T((g*=9})p$SBq|}U(7gL3|%B=>Yd`bqd9=e1ywcoOHku6 ztw{wz_vK)qR}H7Eod-;2p`5{CVcW#A&$i)j?}4EqO>t-ip`s#WScCa|ijQ8F^_91` z=~}-ott?L+lv8ni9{%PqyLRcvvL9aQErmpRn_B3h6oN*Hx}${2&*_0gGlGWUY;s{V z!%R8gX>^Jcx`{Zq=Nu*j_IV)B-F>+@_+rnrkh;GQeqO)ufl+N8*TK;Cbe9Xr?t!e5 z%@sqpHzp7BgATuSe7@W=y6(f$9_AQ9bb0M~jc5&+(GawV0ChN@Gubcjnndd#e;Rwz z1o&w%L}Nuq-a$@bAw8M~$7APlW#x>p<$M8qlZ@;)07;tDq1TWtFm1}YhsgtntE&%Y z&+6I3f@XL8GAhotBC>Lv z7)Y7ugOCwvB9PNpoiZPmev)|Uw`Ntve(`vYF*5=4{-O5AXRy5Kr^8>T>y9?wrHPe) zl!l?7anU6#fyiMc=J#8bdSvyBfGBEHW(FqhFTx_P@`YhUP>0RK_{e<%`^i`jB@Z$kM(lK z+$dY^Jy7K#WYZu6mvcaBc-6&4L(-!)OQ0UYR2&YsZjv2ca~W7NQ6y-vEqtx-WXUSf zK$hkV&&Qjo4b0+H$yKvEZZgLl{OdgJ1akgWa)V_@B)F9J4<~_zUFo2lDaxTYc5*^M?Vq|MsL#5CaoQ-2MJHno}l89UmZyoZ-DiJNT$6zU9h2 z-{ZxXZ}pX}mmf*s901X_F1K$&t)J~HA9(D$bblt(-p0#ex~pEgXYu&5*cqGwwxSysk@zIqJtdC!P zljke($#>$vIzP zoep0L{p;o7ZTUr87to?(E$QObjc|V<`{_ql#H{|^?=0%SmMM~WnM(G*6~A7wFJ5=B zUSIaL&|Q8$@jD0v^6})Ye(OHc;jP=?&5Bp1M-tj)JNZvW^B2;N39kIR?mk@q%Y1gc z*3VxkaRVaM~V97YLKs3YL z#!m6ot8AUkL5`(r5J-RzVe{5A1qr=h`+T;1m>70z^RmIf^7L&cIEjFt;YDyx^Q*$S z0HACy>7@UYoIIya3!}~X#g9cHiRWp%RcR~^;2)sZvL4;DI)B#tk`3x6kdTH@Fm-0%HN;t!jx7Z)ds zokX&Aj(Jl`Dpx-K{x_`uneYk0)cr!v#sB!~w%T*c8E83sWhucvmBM6r*_m&U>nQa3 z0$rP)i#DN1lEwabTPI{oCQ&_qv`4&88>>&*0Qdw1d8bzpw5B;e+}WY>Ll__7bnLZc@2=GMl{3Va6)F7~ZJq0*`~ z%Z$7b{ZXPdQ_gAGCMrBNzmvKeM=StnJw>wnbOe?StD~aS+pwRyVNS z398^M@%DnW>tHtflE<~TS&*WUbX|52D1>NqZxAuZXeehO<`i}LvS6b@R|g}TcOXz$ zXKAd?0ufX6RT>e;)vqo!I+7TEIT71DJEB^ax`xW3eG2l7Dq?q@Pvz89TAia+3$Z6n zcFxt0ca0HS{epQo-o^m1Ad=?*Y~8ov7_v9VwJ!Gjbk7de>Rlup)6aNzGn(|?&wNAH z;Y{|&16nDijvo9XkGZ>SPg(HFfso-`8x3Bd;@~^z&zBr?ZP~T}*SNI-8OB9+&3=E^ zjKLB{SrO^0S)r)$B<)R!R_k&b=R?fAg3U!$wL zMX8P=jzhK_Yd+5=cAICOy8dH}X+}NYh69Xwk~&K?6vuA>SW`L0sWfny9Na0u_}-Vy z8y9*;`fGMY`R>LGZKmmg1cZoNM43j$HTF%@_z4>B8ipfo(`>$J!23QahIHW!&F4O$oND$|?385znWqjwk7+ zr$yW10Kt<<95x-+bNb4^BD z+DMR42G8#Ee$C?Z3WTc55j#WL`1ww8O^5J6?Jiv-kUcBLxgl1|PoD-248^svRMg%+ zT;m=H`e1l*sYJ0S80tYbBc6Y_#`vDK@Ek$_( z90-~Xhw9rKm+g{0bq8k~F!!iO61$)8yMWKom}9AAqtxmKo}m-ZExD~%4pC%%9D4)6 z^|^6p8{|4DYqXj1etc??Q!NR%-)f8LSK~ykDOL$S)botgP7frS!{9%u42-|Is6df^lF>&-UlBcZo zk;Giq^M)%Yz~$136=hZOJd3=6f(z-;1Fsw)TjULBVc_})pciMJDgX|+h&kLauHLna zXTJ+}4V?9RM>3MC>1lTF{v&Kp{Q0gD=qV?ErE}E_CRJ*Up{{w7+LOT@1{{{+3W{TC@Q=#pRPd{ zhOaNEq;nMuy~tQ@1Wb8b@R^lyh4Ce`1_KVz{8b}5J&sAZ&{9?SKKyVx4q+rKazNKBvFG>aNF=38ddge7W;`y9v0sNkW}3a zE;PDtxJ3{k!eTYwxJu0MYT?bA&$@-aHc+62ZTZy1%C*2LXR9;QU{{*~PQ`z2+B02R zGC|IyPeZUMqlGCG1RdcLD~+=)@IHF!yA^2m|cu#YW#C4_A*7qMRSSKa#k2pm>U<26F80rXN-vN$fnr zql`7m>)4I30a%mIyRSN6+iHR{LB3BI$Of0kK4%w_y@R&OYGbuZz`4S@*CaUyvRI11 z@MOn)J1VRC!RAV0OG5vFWX^7#{msU>y|!#c5NVShR@Zt7y zzPtZhrr|;*d)w^G%G#2hCCqVoaHei?kHHk1YZBy5#A`yQIAKaX?ACdloo`tXZ;50? zAd9fYmWC7QXa|)CvgV5_qad62;q`9U^UzK(lurljY=e8z%r-b=b;g>4&5FJ-l!^}= zvTDPoYTgPN$y!*V3`1|TM1{dqcec*$Hd`5BiRkDH;*>0S@4$n))4}BElNxM=}4dUGo~S=3R6zj;Jqnt3&7Eq;bD^|Wjp&sr(Z&Ej zRE$#111jX%x)`#*Zb^&VbFj8Ok`OjBY%u4KBvwtt)pR~^o{*~>oIQ5(``_xIf2D}O zk)~DF-lgXEoZBMF?90@z-As5teI)V2W-aS?(B32@e)X~M`71=MI%;kIWmM~0i1ev1 zKDzR+od4xqBb+;*_VG29l4c7v{eZ@nNM-bJo^v+oQb)Kun!N1{3SR&iE^yz`!){?R zbVh?phXeCwp0pt507lMoDmr1)#3v$rx z8>P*bC^?;NTCQPHUBbTNh)fnr5-SM@QL*@FXwiYDdo#$4xqvQtjY{#R?pY?Pr?;W+ zL^uPIIz>h9RFwm1m1%CS>rw$C~2&`e!zeOP`fgDmEOg0HJ8r z$)WlSyFAc4Hh4g8uy@Kx{lt1%LBmpYqyL%E%e5ep)``%`wk-k;m-8Nr*5%JsLLza9LQr-dq zM8!IXmI=sYQ(3nXK&6xZ1A605!`bYtXbz+y*-MSPHv)wuF&~z4?tb^HpmNf zU7fO zs4O-{3%*86fTwo6qNOr7cM+XDSFY?cVyWM0`xKm;7}f2ynQyGf%q`fvebdH=kwz(U zb*_`^z>~hq5+m##guW#bCR^uM#L<+Ime1EVLz-A?s;&L!lAqCBLqiBp(!>$@Mmf0C z18>dFtu@x+;hy}m(sQL2C6;>MjDhGymtV0(XA6vlAMwVADQepIC>QiZUw!({>vq>D zZ)C(?(^yav@{(t1bjgR>AA_N#!iQ%uvuqwnY!R^oAAu^BM<%jQqN2_$>1h;81|zch zio`U$9(Rh+l(UHydf|GejS`>Gsdm$glytOnBQD>Sr^E%DMq{xHY0J%(DJv!p-wT$@ zJ5ic42$Er@Ze$e?ooNY>2OIJnoQ;@ak9tl%^boo%qbp(>f*jr9w{=IK4ZIC>gohTz zsbzm=M~^l{NxxV+F_FSrGhQ=ej^En%b>)Rl$hK2$dNk|0*4YD2xmet`?d8D!UU6LB znn9Xq>qJiV205JQzUn9UX7g~#$Zl)Py>vNym|J{nWQoqic8t|NYXisW$phV*UoK3e znpMlGNmNLAS`nt405cOuD$4q^HL-Pd{98HA%9iiPCMgQ&9>@2B$RU7L!=u$U2oTlC zGN{yZPJJe*mP(~zW@nIOES8*_%1v_c{jB?~ahQ5dtn}Spab@^eY>^3ioXn7SaCJI0 z%Toe%MVSo(Z)<332kLIm@i(nwMtmPX;~F|=Fzn1YEum&yHIt#=Ik)W< zWqfJ4ycS`-!j%>8E6e=6LrYeD_Afg7x2MMn1paYkv5O~BsE`WYbbqDY)>pQZOi-y9 zdu%S@eB(EF9CIJ7j(fdNP87N&`y*Q3<^0z+hF4a(ANIa#-fx=!4nf0cHWbbCr8P^x z^X>FtSdtWTNT0yy(NPMwOEjMB63DHM?ynlhf_&;b$=Rwvr%y|-l@FV}?jr;xyfud2 zdbejb`H|3SMVo14tJ?Y=403Q4TXum1j|O&$xYWNjwdCWL5(EcuafGMQ{I-JFVwH(w z)O0M_c@_#i6HMeYYZEry(44n{*(ZF|c}4F#4~Bv^C4JkC zJjIs-Eh=4ul-s)6EO;-)^&d$PBf|=vLz{|r;`?9ISgglDIr}4t&${9rbm$-C_novH zKz+wa{@uIr@KdXsS~??m0A@YY&X^7WbA7{K(K=Ger49k<&m-Myqm6lkby)A6yRW}) z3w`h6`vz|cUaZzE+j=Z(jMFB-`AZ#)L`W(ZFiZVKVwcQcA1-GvV!{xZA1hY@EjbV6zzmGA$OM8n zBp7BpD3UcU5b;nvHZ?0cI;|}AF4u|8rooT(3Y4An#a>RHxV{U0(|u}%m-IYL2?FC( zGzRglC1H33=}=o-g=2Df(KvK+WQ|xPxQE&Dnc23aU-my0rxeKPt6x^QUa&0boT$xt zx+Ye&++^@3kx0%c8*z5~HBw=Rv*St2uHuvnXTMwHiLcoX6RPq1)u zsnc*;mKiNVWZ1ohDHPWvxs%R$YW()$Cx(MCCfF`F6c7Eb+DWpuBGhYKp^ux9(FpXJ zT*j8ff;dgX=B6pOhj@l|(RTk!SL`uxQOC07J(Qh`8gSU8}9fcVKxm z`;2CU+b=N<7nRoI)?eX-4u$j4MoSNkI=gkGVp0io?NUES1ObYFGCPMf9ard$3`SP~ z`?aann=0mJ^be0$XT!_|m~02?F8B#z8CMfZ?swQ`*6>RC@?v*MOw$4RbuvqKDjN;| zHW*oW$Mc?b%Qiakbsc=KWMpZy_(9G8xo!(oL|ApT#uIm*w(SpZxkLssmKXun zk1i(xx}ldD^`boj{#??t=|Ga?xw2%7gOfjLA-M|Q{~osCzbb=cVng+ASWmx(yGqVv zv9-vG&wQ}Qy>}=zI-vJ^Sr5l&`db#dKDb}wJYRiq>i=Y%HdK84+thz${5$6g4kHJD z?_J+NlF;nBb@p-gtGit;r}>Vxa!QWkyl@V{Nk zHmJR_D*UhylzOgKhSO<}?5IOX)eE1W^H?y-zW-eo^wF^SHyTFSxgHB2rw*GtKp*`N zzGVOGm)9u!$OfsZNgaCd%Y*-e{OEt#>wNrRK;7IZJFlyo%2JtXn*H@Ful3?DQ)xIF z^ZoBk1N8wpSf*|I(l|fKC(zvSCyDO;|Fid#i--1# zB1;px7!2_2)!U|R|J=&prx$k;k8@+IcYbYr<<|M1XAY%^TmI|6w*9{E`Zvsd8Hbg& z*1M+>qyZ`_I literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-7-Quarterly Data Trend.jpg b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-7-Quarterly Data Trend.jpg new file mode 100644 index 0000000000000000000000000000000000000000..eacc3f2d0cb4a4b008a754586ac0c8cee8adc42a GIT binary patch literal 82491 zcmdSB30%`h^Ee!9z3M?zL_pz>$RCsb9{)YRma)IV1Te+~kxf!`6?BSSr~|G=?>2akb| z1CN9MPnX?i0N}yBy@#&u-J=fp0Jvu_aL;ZNKvl-%2YdI(3gDmRz`;HH5AFN#gT1n5 zw~uAL_+ane{Ra*oJh*S~p1nukef?0@iG#pTK7$yCm49VR2)*~XQ(i&Y& zTwNdfU|oHW z$p2;Jxu`E5C0M1|Chg99_NT)pr;f(2Y0U5kx4`w zm!d5(77i*0-oR;7dbCaPnU%Jex>gRn?p(4E z^yPufi0{V2OELI)LZ)qSvH)X#t9cqZ&5^5PL}Y>H35^ z{3@gVB0H?vj`s^LOH$2~Z~JuI53Am6>0kE) zYj$XkfI(~r^%~+vb<^Ug$@t0-rvEmw;;{X(cYSs~#EENphci7itMiA-l+;s9(48Rf zhMrwO2{Ky(dtqjp?mN|QliX%FoV$M4s0lmkA?Z&YEH<>`yPqH~k_;_5iA_JP-FFvVoEyB-H{^0KC z1?97u-@VmRyCWhZ+72nSv+5Gq3Z1PavJea=ULdf+skB`H_)UG~xa?_n*T^R?cxWGe z`lWuc8&n}Ndv#=d6i;W2GO7vb?MWoiD}=SMn!72j8=Np|)Av&#&JL(0JQG|`?_ejf zj-%BouiQF$t?#yjq{WcrtUwU7;2EbSJFqiPm$DMP3)h4;YF(_l#`alg z&gRm`l*c=9hat}gTCPIu368!`mP>}qT^y`zR}to9>bb)6w23ULoV(DVU1^$~i9Oe& z=MbQE`?yOLWOHSsdP%TaQeq~jf?Lfj*5AJHDQxXuL~Bn|9A4O%{rfRHVjq*i{ch4^m*~LYl%M_IqEM6+pnEdv)XDBI#8eKG0G?u@*6ZL z@LKRFD_4Jwvr+wqi>OEI9`G=E%HW0WS2E~Y=~78hA8#$B6kAa-Ovno=!ZJt*ii#Xq zIFKwgNES%FKCgoHG?0Nyv4zW1y%pl=6Lb<3dv#5ziP9?nLI0~HLo@yBU zFok7^Zc!&oA zeG8sj#)rcV<2Yp%b-&D^Ff%zlX6qn5=%Gz3+aL4hji`1`DfRl4>tq3F2qIXlLz4XO zTC}a&oc!9Ct577bggbzr^%!j?M&D^iD)3PsdC{2aun^;%$2k$>1~@f;q7$ANB+|-X zyT|#6XknaJSngumnDLSd@@{-?nq}MmS>D{-8Q;xrbcm`-tinh$)=L753P6bl1&uek zp-br>eYc@~b#r6JbQjS2a2K$8TkW$xY40+UK)&)Km}~ERyBQsPhJ$tmdg)@^RmO%U zl7tXd&@KQy(*>4KQ4H~3ZrmbBtRM)DzywPPCmxfrJTB=K%fHZhIg=;`KBl&VS8=V<*GZnmp zd1Mk)*Bw{W-(A-Ui&@s;CpM#9W&55_QJOOwg_2_sn>ZOq{o632=x@4vT{9=QsWS9GkB!)o^9vL1G^=x6PqFrNn2JIj8KRt7Lj= zoV!Q@pO{V(nx7IJ+69DujX3XBO^o!PQ;%skT(u! zRw|3o$Ot?PHZ5FSu2GNDXjlxbT)Lv6c5i@9b!ql=+@PCv^QtkySw+&6!l`6QIm zE%iOM3T~%Bk056%#V28mXAQovKqSQW!?(=%u7Jh?1@2%pcB%i?O>Z^#5 zi^FAD*@CbzjzLos$MjUg!oJ(B;kPKBS4KW%-5+0@kItp&+P2o+B8!CX>r+JqX`w_| zV~2%b=sbaBE3d0`Wj532gwkm?qZl7mR(3NzwBAilDJr{iYO$-K^figNM8Dm?mQ3m~ z#VI6qoT1nD=rFS*0-7wND-;N0JrJRYs^3+~t};7GoP^ccw#$IO2u(TT68U< zbpkTEhFv*e)XiqsH}@kI%OXlJ4THOY@R5TP**Qx*5X|yv3{Q7hyW|VkG*741$k1Lg zD?raVdxxYd-w^5UaQS7AIOWy(=3ySxMImC=mJ%(%Ynm|oinhyzBs1`_L-W$blZI6p z9s0xmajWwv?HnBs-_%+nA&D9Q+#bpzCoPJB^7l5a&*$P+9;ES9GrJO+uDt1);~lOl zxUJAvK)s%yeH*IQpFAOQ+JY@kh|vCnj~wT|t@`GTqt55O4ZZB77Iz9x?2itntl6N6kd-C)_~?NE%Je*M#fm#{ z_*R}&Z*{WKL-LwvYZYe6(=G*BT8g!wat-x4K1K2ByyW}Y-H zkqe$9toJp&2m@zyt7SO1mi2qrtO(x|mr6o9XnH3seHO;Auypd{n_Z%ZUk^>M^=!V_mxEZkye?);&Y6N@?nen}c-97WHxIRG|v) z(FP(ha@!FJeOZ-QL3N^cDHe#Y3mY>nV{Ka*pIqEl@!!s}&uuT^+x3^l%~3=eD_#$i z3(_T}wrLlG#L#;J^D8a41{5!p=h8eQ!Iyur38W0s369$tZtG_V;jFBbNLzVB1iCT( z*!86x;-FW_?H_HgDFg0vQ$(ltsZ_7j|;;TY4kh+d|vV>pa5+ z=O~_Bo~dOJ*70<07N(bhyM+l7HfSUsv~(UFchYk$emrR7jO{q7l-l6nrk`!s3}-3? z(OF50E7EzrjpSC3msuR;3|1AeF*|=cjw0eRM@wgS0iV{9#^W$+ioZYNS#`Cg^O08DG1HknRbnh*&(R!Qb= z!EMv!Km9hbjUTx9RrpZx`q_r@jG<)45UiN)-%e14f*34*$FmABy({Wi+Kk@QSy2Yh&Th0ouS#Up{{j_T}p{hqYC*Oi4Q!7nDZK7ql~Xg~hg(KNWjVa-Cn~$cZoGlTJRMTL zq!XPHeun9V#Z%$fhv{>l4l!mcbE%#yF5Fe7yh3^%a9;TgEc#Ax^}H8!_+i%z%OeX?xbPVS9Aiplw+C?_%^uP}MtMy-zyL&>Ru z&uAxxAE-QeVjBNKtfP>#pMdSrOY!6FoLNEh4Xlw+(BR8tYP64wC2mK1(t04cAAIsl zAMGW(9xHB-JT!0*)_bLBu?HgQkMicgKdZtQi}mJ&eZb26hhj0fkXsKaHZ z_s1mz#xKjwD*VsOm#pfhOuf+GzGI_y)CJLL*R+(!C61c}{>#cmpV@PR{BMzO7hbZQhmB`=b0URhVs))$vFfSDcndg4xm z+C}Z6o)WYdzTjvS6V8r_-+qymB?N*7z0_=8tu;SMo4M@T8sb7t%iot)Fu35Be<(PG z$+zfDVTec$)=)?B&SD!EMLX>qDv8XV&dG+n6xU+7URpZqrX^TC)QOo2cjY8T3{)N= z=3OFuwVgL^8gd{X;X3XeEHSJG9ho3Te~4OC@%0nX~Yd?5LT!C^gi(-8TK~uF~+B8N04xn zoia{J6&@7h4<%lPl^A0esz;Z))w|FrO{UdTP*H{T4T@&u4oE%#8eXz5S{^e1%bAf2 zkol>0qq61kc^Alen72lFV4$m{^;sJ>&8jzvm0aguqI(|xNX!z;L7OqdAj-_O^irZlciOY*H2BnJDu45f2op@F z0r4~o4&%E7$+YQeZTnbdettQUI>PVM^iOOjRJYw1dCSQgT*;&46*BOG#^(*{pX{&g zZAt9f*b;-hTqE|8SnvSsIR~@W02Jw=prLSU%T4nuHj}qJ=&GE9sckf%Xr(zfjqCyv zE$sLR>oaS0mO}aQQJjg?{3u^vC+?)D^58{vJ;yY@yel!XC|j(al&76OAiqt)oP3P; zjhJD3JC0XQGrYxxjjqX4SyL(Z(+^|2!Oe;~SMs{DLz@k3ErU|9rs|(bT-H!oK%5HH#^7U3>4E_ z4Qwu`Xj&_$1l{1qxbgE9ZYza zVKAO$iAuqfK9!_%*jIJ|is{?GuKT)=cFG*J1DSJ4Q*p8VJsVHrhYtL+6{=>Iiub{dup048ZKMxNz5VWE}$y?pd{xL_kWW3 z=L^oupA$nq=?=C3@wKV>8EN`8qMvGn#;pj}4BL)Kuc3PrjP7{ zadmnWn&E|XABL%RGKrX3G2dO^oG+ck4HfSKlC34LA~Rzyv~)%}sJ}MRHM#iJj|GQr zrO;Uu>TCxgVEK2Np)fd~(d)M6~ z(G|3%Q@el~IsJ*Gi!8T>O79<<`6;(8V3a8;L^;PQ;`Ee|hXTuQe{XZ5Ima-S7-PoJ zb*e(x1`Hi(xa%>q2DEV&yR1brT>G+O)P=zVh$FmOV!)i_Jx!gCi~8D*{V^=qJ3z2 zEHd$xc}zKQ)80L?IY?k!VjGNWZXwg8?q5#+L$>zHLpobvKfKtvL3J-^%X<5$@BRVc z4=qmgwUfYrq6q)4P7D;JX%RBHx3zRy^+#W%A+?J#J91tS;N_JXhV_<<8?qRp3Ws5vD(04H-ojYwyso%S5fEBc2dl{b6*xQ zO^$aNVztwGtnh~dW9>y^ZO+FX93$itUBimN_>i2i>Svwvw0rIaO2*nPR(4!Vq`8|s zuk$%BG5v|Ot^=~Om>Mp44B!81wx8p%1o@$3@ePvgFikRW$uGF$Vtpe`j0kD?qN+Ui z?K1alol0C}a?xJki~PazxZ`M)EanmhS^c_O&h}d4-J6&8_a-zFPT=7A=xMHNn!Ts* zZO0hHLds-?%SsO>WD65_DB9VlxZHj!$L&J4^SGc)HgReXtlaw;G$y!LY=Yor5?a*Z z+P}}@d@08I^`;tTspQKQaa1)WBW)%e#16C8N+WKxb#`Wd(52Es&o}SC=#A{hjybrJ zn~>YRB-P_xE@By5hYie-zldj=Hyb;xitGlekOsC$2-1KLdeanQX_3DSH&~X)y#Txa ziq(XX9>1p|{%@^1c;DnG7u*zY(9Sj- z_8bCBMmKmjh6k;;Rky~a@YtnY!1oWkR}`!GHNw3~EmK@qE+)-vB$$OEM$H8kC4xa` zlCu1M7QR9Kj5poB-jBY3J$K&_Y`S(rq(@)E$2}|@n35tB0?+DZz{IM#WYK7 z7Xmb7X9ussAz5$a?_GTTF#QR=_sOw`yr4}zBxN*ENC3sQraW?ZDAUAIc)GgDZx#4< zt9w&^H-J=chv~8(8<%5W*Z6n2a8!~=$NBP5p7{q@+~&E9widlnquLu^`LP>wN4Mhq z>!qBoyZ#(+l}i6xt&1v8wMA!?!n5tK=RB1_3=xZ@$by@3{F%&+i4nEcFVfI(k4*3n z|2p*h?^gN;Ui_icSiJBuUb}AoUT;|dKB-_xTlkVvy`ho#V-1oZ-`Lc%$fDot8o7Vf z&%G@q>6v>9M3H`)jV!v;*`k%b_WWe9-Nc1yGswg6WVif&I*!z0iy^ogbovhIaF;|W z8>kchoOPdMpHhKRY=vHOc9`v3LF2O|Rh1@q3ybWFGZv{2=uksSeHwvS|Nq}Q16Nn1d&_4e^z3h=-22R{_g^IRfT(Jc{GoMmrQm0u*t~() ztb_w9$W+@#mD?0enMDioq4-QOX{xrnfTl`r_-g~Ol4j29c}DA*hGi4NJdxr{5hhxq zL1O6W+Z}%`FN(XnW0lJACk%pp9=2P**(LRjb!xgLrwi_(gsvIBu!+(uaiFwQ&4AdE zvx)ta!tE2*J)G4X$u0ocGH>A8zWzhLVcVmw{^w}7GfJ*-m=zn-OhBOK7-UH*q_{FS z&mh6yTbn0loU3L(X?B}X^ew%E;2orh$YeUaw7{Ek*k%&JtUY$&#=1oqnIYM}+scl@ z!U&A`4HEF^WuNVjUsQ++s%WEPMEL%l0eZK3HW1O=?=v|;>!gAP5sx`KIgOyQ=Iw#C zjV{x4LHMN@23I~a^6baH{b>kGQ?*s{ea2Iyd{reu7~v!s{#OLsq7z5UTvsYwB} zd4#2OuHvHJ_(uLMc@V4z2;o6q`Rb-l&Dk0nlWI5 zYy+A1Ih!2Wz)H^g9(l{lBx?E;l74Rkn zH&noJRj}kkS2>5a`n?W^)twInDqt(jy`#9HodNAN+BAtQkb9%v&kO647Mw&*C}3e~ z<5${D>rgdu*2*89uCwUNzUzWia{z-%hbtELT26sdFOI$OG}-@AjuBWZ(`8BM+PV=$ zVHD8|qZd?AS8W-%Z0UA@qD1vBU9v0fO|H+`aPV9SpA8oswahN z?(fmtevi4?KbT8Luy+BLrp7Ji^@D`;(*LTfCR2Cc4PAzenwO>~tl`FNxk*Rh92C)C zmm`Ys@1Y2$7%59tc>2TadrQg5MK>F^`LC9!YWCIMDvh7Q&HmVpKfEDJ3uu2`io*=Ijv49vw8@Z)gtwZF|*i|?NI-UXbJTjqN2h_rSA zhHm`7t34gsTMcKS(x>Td5So}w+v$o%{Z(^^#a&P5Xrc`f~HH8G|we=9m${h4Km^;l(f z{D2&5acUQkv@9FFs_Djg%cll_KMtMum;ny91=jE(ubjXRwDC3FQSG!CC$`tvB7IK5 z);t9p|Cce9{vV?Izrs=eKl*}=>GPtqibKJe!T`bP28Y{Km@p^$Q#;#+nuk-1(sU(r zL1cvY^aW&W^RQHSJ}3>h(VsJ+`+Jb`slOEFe^c3$Qt*TRyD-@{$)U@lTFM%~voRjB z5)yZIaOKzyaVt4pEpg4<*NJp&UOi|9yV&KHESL`+qT?ll#2Yhe$&8pI&F}P8qS|jH z3p?qA2v1XzJ7sb8*dL4Z?-w6X63wBJHH@TWGU8WexzJI}WVK8P(niyqtxSE!N%~EekoB?_NrcE8Xy6_2l2{;)B9I&}{aHCGRNEYZ0aB)ob%=dTqiB#kTw?&9g7s-?A72+=c zRAE62Nlb`mBZ%Hh!5SS@dQ7}$?UI6*NF0@iFH#D$zKYwc5R{x<<(=JZxJ)ESlJDXM zzJIxvdZT_*H)YX;q^Pl+_C;$o*K!wt&syYjvY$f)iQb2jB2yvvZXzmF3uboietC^R z7Kww$+bWRCe!5ioV2Lds$jQ@aguePoZvk4k;8H68Q)jQqb6bax=mCSQm+8#vq$FD3 z*ytDzD$#-^FFtv?N69b5FVg50qx9?!wYVHsWvv%f(bGmBhww2qES3u#0+B!TH8l$R zZE~`-&a#O^zINc>j%E%`{(SLDCT?4=2^(@`7hu||euf=fYojOO#!jD>C6-{&n>*^T zC6WE$!ly3{yKD1p{0fjgxI_5(QJmhWUXMa2=jLEhrd_j<@OdNEo`;XJDe=!Miwe(a z2vWO;v235r&umbCNCm+-VC{ucBYE0&odoGa{*_W-a_L@F&2Vb?#L=kJBuXJgn@m-e zZP3mxTAPylinv`s^N)St8*Sn7^wX{9jG5T8gABdVC54qCR%4m7ol2qq9eY*4zJSi)SoLrRADiV%Hm!6fE zWM)I8@t;l2kCzqI;OA~T7fv#Bgm52H&6L^e zE-o#Vrpv8Zu|Mv)Qflq5pG(E3h0WT3Qfv2?p4y{1F}5CGW`Ke3WG#h;aJ~*62uN<4 zkto<(-{cHD8z}$=NEBT1WJz^}!p-tVss`G?YoY~wUF|(880$o7vR`m zVRJ@4sSe@F&T>ekU#ByHAoUtt)+wJ{(k!#({Gr^j$cUPu1~>X@c6a&ZDj#}{cN%Jn zi#oJ;N$-U*jtg=8$lG=rf3Bx6E7dVZH>n;8T%15k}YgLPpg+i;lwyM zD2b-slNTOVk+y%JutN;;Ux>~>={k5(n^2BXPKCnXSb<_74X_{*NYN&RXl^f>FBpVoz`$x!>KUL-t8YbdhEle2e<5w1;cNz$g^$ij3f{P4Fo8~Ymsa~{RENW~J{4ll$(tBN`tc06zn$B{4Nw`Dq9aya@%jxcZN|JR&>TiI-UuIW{L!*ES8& zcJW?Jb4Qs43Ef6r`Vl{O`Qf4|`+URI@>$-|#Hy^(yrCjT$`qYyTtvN|ESkuoof%e+ zr9yd=^|f(@-sxR>5%CDaz}IOrd}cIiDxqY{Ha9Aa=Q~1=?~dCDy^#Kfu$8kQCF4gP zsPq)LtkL*KomcWrS7xYuzQxo+q9si7Jz?uX$1dP6Tp!7~uKG*RQXAHs8B|JrN93Mx z^>Xem;BVY5kD|DH-fp=evQt8d%s@|6V|*iFM(hIq6JC6#Xe;ZDd7!vL*rv<>z)IZu z`CS10MAg*Au&Exif|W*l$*GPy(}^eV?DluO^3)hE^Z$8ww4SO0Wqius6H@;-fA5Ws zY|$4yRsC4F`fUqwCB$Up?|xol7vn*@0R8^1G)-Kt$dLi2-$J1ug^u4R)$2|7s zezSMr#Xn~L-DB_dzOz?MohufH#!L?%i}lMaQPu6=2v}C#ce5w7NVp7Wu6_sh;msf4 zV*J;hqNnO1^~2v)nD4V6@f|nM7N3-iAJv}9s4eRK=*~O9zls{E#-c!~(YDh)QuJ@; z=!LVto%lyy!Ury2Wlw|!{xEmR>{5n8Kvl=|$e#XtQ5o9W4A!?1UmjKd;V*&!0Q7Jh z+v58ZzVmwP>&Nj*nc;E28}tmiF3jt*zZ((Ge-As9@jkePA);x8GiTzyYQ12&9nohF zy}n6uN^Y14*YADq%uFd{VHuk-g~`HpyU?JooRPglxA*>HgUg;tQ;lGTcsvSU4BIZ* z+Od$Nc4B z+{db_WD)BkCf<+LXM+VL6NM@+WoMmUBI=r)9GWn-SSF_&3&oWE>^`>^$Y#XGX?97+x*Us z-?uoQc13^G?FhQ)l-u5Vw$s>esH%l>7L8bBq~70Hq>`6bIQyDkJ5{L6liQ%#q06UR z1pXr_@$6Q*R^qIFV^buA$K6<5RqOEG1u$S64@PI|Q1_>fLeqHV&gC|0awtQfMQ(8Y zY$vJ*%FRE2vE2|%Z?P%ooo1$*JQ#T50pm79tg4eJr_$&7EgIZWOllbr$cMt2*N`XN)yOk}la=?m z_&3HPCK8rrW;jsgAosdRTtG>Zt`tkfPG0rA;GF%|`9Z`S8+nDc^#&aW!x*o#mD~E*EdXS@KOo*q`tl2CMIyXBf!vwvI}TP61J;V zH7%~%%=h2;>~(M2!r_OrLJ0Z1p*tHpR5Q1fg1haKQoYg?B3@Z^Hk})wOtCP>k#>Z! zE(i}4MMVAh@xo_Mxcf8su$&wF!c+T(aaezTV7l7Nxh{}4%RPmNSWG|Ch6v7*T0+~D z)1mr{>xiDs%6wULH4xtH(lY$)O5WN@`Cz7Ddm$C2iJvOExhSx3c^XUiQ~GYYJ%oY> zf$7f#I=F!5P$y=4A0sIR3kyNX9Z>5J2+1d~yd2WI$lHc}1?Tdzl}&v`j3PUpN@-$> zcm7)O=hg?`dS)g|6*vRR%MYWJ@vCRo^}AV~$}S2NrtBKX{uQK~HcnjQMYN1` zql;ABJmS&bDywl*n+me*!bl94r#+6ba?yAR_L}=Im|moD^8LyGv)FPRP%$P3EGE7TQ#Ov=1cgd-ZIgA z8#1U{+;%vzzHJ@Gh5%ixkHsN4^Wkur+v3v3OCNnWb#gPjJ3MpY{;O5Z1A7uu3h5nt zGmIqNn(;8JniKSI$@NR=K4ueuzPQcdYs`^L&$E^fyzI zqw(KQ4eC1i&wddaT|x=KxU!#g*`eWG2rddvbf!*4C7$b8^Nl!69%f3My!487iz;|0 z=p&libgh@18?qg173^_4o?C0f*VgMdYI(9&+10CRMQRzHWDeP(QyP~y?Uxh_CsJ_f zlOZdMqNz*AdB>TnN#n{y_A}+f7#*J!U(N^#)mXXruur!OKb3#oKiCaB+Ez=2w|h z2z#ZIHyfN@7JJ}RyXh&|A)`DhQ80!<^8JTcmA;ul20fdqwR#$v=GvOLyYkZXa^LMs z9czWOfzBwcqe-j`nR}LUJ(*s6t!Z(*M>0ToD;&Q!RdzJ{>RNO`U5jsleD>(5cFQQ% zgpSKgiLv8s@y-Vjq|(R9+v7W5oga5^>Uxs%v_+kfe>hN&UgV}1HIWHfUC|j%em2${ zt8!oPIL7GY@MeSUeCR-47X2Xwm&PM`H*{C_Jl@&gxtwj#p|uR+A8yUj)4J25u0>A~ zHp)wraN51hzQ@yMlh3Bvup^JXu(6GXb;xRZjQuECGB_9$5UINB7(*GmIPoK>axRcCFOG5ibSbGr1}8H z7025<5X;y?0VoC(S1`HHxyb*-_62`A{e~GXy=dcd*HZrB0QEtqJWVX67$ZB3R)xoy zYlauq8adG>XTvWPET{PwRd2fq!=>6j^iz3Ee6^))`y&OwJ3;oB;VaPz81$HPNluZpkl-?hm6$AZGP^onL-+Re}{Rfd(9+Fr6dGuua{#6 z3HFMl6E`L}uf=Q0H2u}pC7|GwK&9ffJNe?lQvH`fYxyW=GNWcPKSL=X6v-YwfsNb< zlI%p%z-m!P&zm^ZxbpjQaCxL?gC5I)@!0wgBXKFu+^T$mUTotz+wd-eL6=qNQc8X^ z#7y^lX}iI`S-a7I>$CR*2*Q?gPnrnJ*2Y`F`xZMbvL&Z(v z72Z!4G2f7nD9W8}u)XEf&~ z6RRRq+7~SsJ@8ayL{JzSqeRyzxNFzMhYF@h=Er>)K;EsZ%R1?rFCOs| zUOui>bjL?d=O4vp6YMr7ZietOdOIM}My}w#`^#7pFYvJZ9^7HP*Y)y%?9int0m9Py z?bbIcm48d|o{4F{hS~xfugC{%yP?tpH`z(_6xpU|U~6cJ*ia{$9k}%b_4A3;$4jnV zHjQ^5gxsDg?@QVRtXfzx3u_dL#El* zXZNH0d9qNZ%lc%3(^Qp3Z|PSv&;COz;yepmf4N8RaGj_Az*Tq~9kL6sUytzjN$#m>v%?>8`=EROm>|gBtfi}u zh5ATyVuCq@s{|0 zB6Y36vQ{sAran^^viBK{?$1WEP|JpL$M+rj^6%*de8{dbcBziz$t>Fs)GxF1!`c#O zcTkIPr?#QkhDpF3;-7SNh${Etd$g_U)LgTAddr5sPx~ZpB{tmK9m@&3bj~U9+)o|9 z@0oi09%0q+lNVn`b-o@dO<4Hsm1D@vwYuw50X>ZoKTqmtvL9He{k&iM5Ae@Fm;0b7 zuwt2p+`KJazE!xWUAgUkdS!$>vT(}B0Z}sU_RKEnJ;DIMPFSbJVhQXns+2g!@)A)y z)|s0n_4aDPDzOJ&0shST`DFCS$oZm}19;nMyN5+#Pb2I%j=jl$Qz*NH+j_YMaJ}h0 zihIazu6`)w0W1Kchb#U5F=S^e4!Yj)U(!LlnOczrnCDu?bHsmy~(; z7o03CH#1J+`s7K_UMWmx7I++qHb5)pF`=>xW`nTSkbZjKa>>4;MU`V`&$M9GXRQWZ zw|pqruEI$QOq#Pam9#`0Fw1=~d6;GAWj8&TzwBnMjV}TM?W}tc;-#%>LNuqtqBkK@ z5-dgJBY&LB-m=QR^4c^}uhElr6K&w(+h{u<&P0Ov0oSF?whcr}nZ5@Uls)7Lk|ZfKz_*=0bRERg4q##D1{4(mM|ZXOMEY7i86V%n@ zC8(cAUO$;mb;?a{)h^q##L|hj^KmANoYX|rkKeDotZ<3z6tkUd!kAe z;3G8}EtUPzdv5Wg9QNLC?k$kZygFH79Dq7F)`>DhZ&&9bNtnk->ex>96(I^Mj$Yx6 z6Z@JwTfDdL5g(?nPTk)wePlQz$rLFgPiMVlc6fW8mfbB_*}iS7z&Rr>Os0amtfAe} zNw?vGZRiuFLP+2J;mQ(U^;X|H-%Flh%go^<5d}vny1NU&xaXr%OxhQh2f?;KyjBTt zzf~Z+)aJ3!Uv33yM?z$tmV4<;=3?u5bU)nz*%~~A%(z;}f*7`_sU=mlk|(nS))xCy zD_xa}Lu0v>xvseb?b=QJ;2{1*y}7BXc5@am-ZObo1<=>1e&k(si=Q!R_{6j{zLbFi` zqDdT_M~$kz3pb~i!7Mj%LX-4Q*4BN?LcZv|?;LMPZgdvJD?mdir^*f6ZK10{6Mi*k zuWvyeMaIjL^SRE=qkgvY8P3_5fvo)WV;J~Dk{y9cj^3oBe&Pn}*nFE*e*A424F9NW zpbAbsjzMW+VWA6$;Mo%P%5O8tB@`*?lMl6SDQq#m|TL& zi#&giOb5!{4RpTx{JNhZH(QG_e8$!_KQG=i?Q;YnWhVW|29`Dee{mVtACJHzmPIA~ z_K#Yx=LJR=j1g-alIoLX_nvaRG*1iPE=p{l!+hs{nws-uEKyR`a=n#fE}ps5R)zHs zK0U?fc%5w3sMHvr zM%h0Zn4w&!tHYWNcm~$cnJ(7wVdT_oTqN1yxZsgJp6yN|hFZpkK!Mv=z}(6+cdn~` zYae~zYqL^sM=(=1PoxEp1u?KhsXq`NS?iyUKcOa9`b1&^K5xKzdyYDX9H9LYO*>hu zL7}ihdBzrsA){XuWRdOXe!RBPowmd_@L(D#O#u4v!W3vL`b<+)Rf`}4tid{OAA4#KJp4cW3=T5NbY@UoIuwzHsOVy-4n z8ShOIjYs%970F|FHhh`1*Ln47YOBs!6N#dSLI|Hb>nU6k!egX`qJnlKUPV^fT~~f@ z5$qI80m*i}^k^;8R;C&=RyZi*A3fH-{cYZF+dFQ>kvXgdWm1-uW+wC4Vo;~r3g=*Y z?+B47OP|F?*1#^8d((tufpB}=#pny_T)J9?>0%Up*w@~nVPX6Yc5%t72nvK#Et?@c zPU692Dx}3}xK!?m&H}9Anu+4&99*z+BR^{P>S9HYRa-5ix;YFD+ELfI85&aTEjKh8 z$4akX=Ouc}O{8mM%D1CEUUeoFp(f$6?4Akf7u%bv5dqxzhzzsi65H+Xsob#p10l$; zjJJk1xy>HI(LLH}nEuE3;%3Nk3}jPmkD824Ephqvl#9Mg9!`qrB=reSl@{iS>D4nftvUBS$*aAtND z91ZfmiLwC40iWZpla4nLlSHH=OCdT1r^R}LG#nT(N+8T0)B5-?k%+vK4{r-ybGGr! z#?D<`w2exQI6Fi=Zre%)Ld_2f8>w!(=P&oYiQgm?nAc15efzuD8Pm>2+ij|2rwzDE z)cC_LXSS{H8#%#v>4zl}FTaJUmYiJxq+-sZxII?_+oeA5BYX#j3Sm4F>Z=9XQhfd2*afRu<@E(8LUvkuD!wH0(qx&_W2e0buMZe4qX=Fqeao+bEyAT%a}dlmgZ@O?LJV{GMrV4P$lW9ff|67Sy$l}3Ep zv7ENO4>o%j+TaDVap|1z^(_k~*@beHMZ#mj1R98GJ~3o#?SrkuyFp8*z-}{@A+$sH z_9LJDG1Kq4B&aZfzx>$#yxCfZFmzd!mQ4h@6a3AN3YMC*Z2h*ir{gctcvm^}`end{ zEeUEZNESKqH7d^jSz3>l1h6*= zv^A<$U17!fd$QY_-z|Ol-M#l||7}{UfBv`0y|*K#u4hp1hl4#o`~mz4=f^DEvrm3^ z?p*E5jZIbUI`v#@!MevQY)vP^E`J|6w;$8lBT%hDIerM`h( ziGv%VvQ0iRo-(lNvHn@+tJ{D-^BeZ)ecJwa9!r(;-)9aOm*3gH?oa#jpY&u;m0ROO zw8Eo?>q^nvDTePII;gll3|v}24f?S4*+0M!i0QN6Sp2%PO#05^g1FWD-cg+UB+`U= z4*1g>W##(gA$@gLXsZFJkp6HYooTAol>C3N_a0zPW!=JX98|`!Fc`4`!GRG35saZ5 zz)_G<2@xR_X#zus&@pso6sdxWQUV4A4U#|*si7(@0)dDDk!}dR_g?-0N;3ELeZPC3 z+rIyMqR*3@9oAX?fqP&YFOATloABu=H~NUP*H()WZ1`uv`UX0IL#YbX|Yhm zFRiKC#?9mMDvmDI1f)_Zz}XiXniV05N=8eh$o-dPydO1WiMFibKNsp-=s8yRjY^XW z-KMYv2#_xo<9FA6;#LFS^mtz~T@f<&Nq!x`P^Mk!v)sR0oYj!4+5hyjn>t0wJ5_mF zkKq8k;;3Rn20i-tng4u1X>jSicbZZLL_aem0;t_Yk*Z*9aDXA5>IH7rY}EeNN~k>t zOGf;OmnZe<*}T&-l3}boTqtTP-erL@+Cpbv%38w?`ZVuoKuhFcx;L%x>WsC zDRZXD#rz2)-(D#4$6)$0=nh}<<1gzmF!#uhMblLL941bAAae3U2_xBu9@(T!q`;X7mh95}p7dj$nqn( z9!8*Ywcd+~f#rR%@Fk%YH%SJb{Ehc?r7|%v-&XOQe}e#hbHFO#|F%a5W{=~H$Ew#x zrOZQsjO|{WTvq2_&SZ@DCIZIav% z%B_73XAKmq7$KC8mCLdF+}l1mA}mo&DpVDO`jmeTrLH{O?Zqy)F7IwdaDU$NE3Yyr zWwu{FmPD30)d{!C6nIs?dQiFD=cJ7H+V0x}57&C36EiB#{ibU8LnR}nM}RcZtgZ>y zQkBSgfjuxrvgnzpn$FCmy$qFUWT#sVzSqjrx~= z9#NFTj5Plf)Eea*_c|;66V(6X_d@Fuh6JI?j~K5t8CoEc`0iK6^Hn!=$i~xLM}l@G zd`apt&CInawv|x(i72ghln#=8)|e~XWE?lrri6F6CPu%&@oaOpe$4P9yv%Ye%Iv%t zMNwbWHUC>vo62x<+}fS?2f%3)dh^50@IJ-sdBn#`ug27Z9;A2EK9A_rpI+W?b+}AD zwWldy6NqcCb>a!>BNp9-ILvTatH}HPsx2tOP>MNHBVjBh3Hos1T%{s+PvpBN83|0( zj4R;FN9{qgONk@2b-?jY4*Igctom*Fsp79#UBh*6Yvy2@Yo5_cK4lpW>}(u}39Rq9 zJw;a@g~GkHY;TLMU4Gsj@ceIidoi~!565t|EF^wWxpLbLNpxCG3|Zq$SoF@wHv({H zY(2%vNRD%)fRn%HB@bsQEFV6P)BFqG8A%`ykLZo1Ufz4Q?`QqS07WRvBXbbS|5b!O zWg|j=VFUjEZxQ;Jg=Wn+YkzIUrqsT<4@2K_@|fO(3zKo?(SkvwG)Z?ob+25UTWprV zO^^3}`1ER89GQ5u#BunNbt%lX^y;g%Y?-i`=JPW+G%9&mpj(e3Ih*f)lUH@g?&%Uj zZT5M#36|LPQkz;nEI^x&rmb2#92q3s9z+Oyy4T=`?)Q751?Nih59%}s+J@b}9R^j( zTodp_Zo_JB-!3(=L|X9e{a@({R%>>A{-%|I9+il6NS#(k9vBnOaLe*qX@VJ^X$Dt8 zGX;nT@~$c13lMa;LFt06 zL~Z45gkm#GfS{`VseA`q#Zzd)BGgN$THwxU6FDYo#+AsmlSHoJS>=}UHR-h0H>}$D zqxu*`1pq3w1#ix(b2mt`xV^EBV_GZsFIDcic%23H4vDe*%@P7xlge*{AB_!kxX_lB zfh%0{Q}ZTH1pp!AEzoPA-4>C6s}CKwb?%*sOqcgzQ}T(dRDjo(PkieRPNRDbZ;Z3u z^6Y4Z@7)gbEwLp1>>&)oNj*D412#bS?BQU|#<%tPDZQk*##gpacDd#_f_mCh_cf2; z7S6Aq%KUEr+rDjA8jC7^&*b9Qv~$4D{5!JD)SeE#AYsW9R&V^Mm{RB5d%{lKI0czwqzJ$Px}@TC{8 z(WAKJ8()*B002V#xX%?VbZ#BszV|RogG#(a>QMtJI1s1#j9yj=-i-eG=Yo!sS z;b*+tWRY{`{;V$6`y2(=cD=rfF}(C=o42v$EL+ZgKND_;*pa+?qPuRQzeMZO)xgnd zf)S#>t2Vb46Khg?dB?@yS(hE}+}NtSO{9EALHaWmK@#UOYjUmg2ayk7VRtFBh>iH$ z)x(!|wdAgPJ$QV&{I1L#V2;d!lix#T=^RsR)TQOo#YMW>W}ko5gE6*rO9hyhZsL1_v>Ve(ImO`{K=V8IIbsX(Oq7hRRblU*e9*$?}-sp4jI)e@en z7-!Yw@XE@Ok!JWcXKbf#vF?qQyiUFGU#|3HeejuoYcf&;PIjE-&^ZTS;qwX-m-~ya zlQ!Hp+8r*kibMC#U*j3RQ8g-)Z1Z^M(A&+&VLNgDA8_KZNDkkV+e&Af16*CQHyQB! zSJs8MfBCjwK{hF&xkKlJ0h3(FM6ZmO<2!&`VO@qFE0INVzzTb2pZ_(|=2!gtRUTRy zuQ$Nkk_G31LSElvOv%yr$zU=XI6k`cVUQ7mwtdwY07h;>pT;hC9khv5EbX{hiQm4y z^ojpcWkOBH+1kB7vtR}6VrW% z{ln=_e5!rhxcKb{FQ%zVh|hm-N=-c)i`{`g2H<0M!hb}_weyvEcBXNT7?P=RnR~=* zX4m}SerpqwgT+28n8*#8;?VRkK3$Odxh*Q?JY&C|U;XxFZYf)K{Y=!1tDS>?U1#H6aLm7@X`%6I9s1@@$3MW|&x7MdM{tsWFo}IeN ziuq(mLqTu(oS(=7_d%l$UV@>rRk#12{Yu8eV@EFRw0pY=yT|-3IrGb&8)umL(^2zG zvyCi&_ptj3E;uXR;n%G)OuuG%qQS{wwQC+FyhTC3nM!wvF02HAHGkr1 z+JVafoLkT_KU*$#7<@T@Bl5;eRfBY<#}h}mVf2CH@(C84`^u)qD2?~f%VEW*@flD5kFYm`U=h>H%cBo_x2f%L3qwjjpn9ck4!ZLTz|xZ6|j4?u+-?LdA;2j z-ueuaSo#c$bwADrgnV<+?SRiS<%V#_l-GxM9uB3(FHZh6LEB#(pqRyilhf#aNjOt( zxZ0grsNFL^w`%Pc4K&Kpw!LsPZws1->N_4HMZ$EAJvVD&UJoC5-d%5@8*335%z_p0 z14F_39nqk4<@V&~RX2`ps`&IWvjjIovDBY6uszns7+gNL&D(bHj3XLz9pl__rq7^U7^?KvQ_Y0cIx$!X6UGwZW}=88@HEG94S4g|Jtx7CRJPZ{cNe{&%1Z zR2FxnH9rao<)QG2AkTW`y!kux2@#LZwbZf^{jzJI)Mk8(yMqYRpB;Zv54 z_@KMt1&;CyoAA4#tSL5653<8&D zo3J}GzgX%`>`Xn(%(1z$%!~Dg=a z*y9zy3tu;2^Pcvj>%@V8K&=FiGMry=Do#?&^L$F zwP1+PSo#Eis#@xer1VcrVnOJNN(}5sX62x}O_}wY-S;mtsijKSmUqRsMs3HrtqBAG zSfqx@mFvYN?zOn0$j(_{Ov*gI=Tqy;C|_OMD}4OvA9YY;|IOpsw{nj zO<8724Ipw+(fU@QOjpK@gMlcgd!27VEdk#E8A%a@s;ey<%~9dD$;p;o2TT zD^B{vK7%djp9V|jE70Z?$NB4-w#@6(>oqy>MZ~uVBYE zRoLD5+=cM6$`W5n&)GN17~BJ!#l-+XC^HTTouNGOklhYTZZu)g7ll}WZPf)%=96=( zZX4%U;|bdemtF_9(cxUcQ@t%n7O62-bSB9spwTEMbPVSq(=%8Zw1V5F1}^r zX5?Wj=$~Oqjix9Uf#xCqte_P;{p@JB0>X(Mv}c>(9H1lT-MebT*8$m+ilZ7KtO7+Q z#U-)F7If;Tg8?k4_iVfPcB(Kvf1Z?>g5|&uJC09mg8uu;aj0)u@G6YR+-D(){!c8) zX|UT@{ghAa)7}LC{dW*c3C>44TaylbJGuua)W3y4&o4IR0SwnJwte zr~zynCa6x&+FpW@4o2k<%DX*` zSNE**ZF1Pfq?Y#D7~GQa1h-PaS6I98CT0h_l777rCf*Og3>Jb!{KLVAtv|E;M77nK+cIG*`n z$J=im0CqGmO>^wvP{!Bz?3lPc_8r~Vq*wNGbLWvGH=xa=&k@E*@?C9iE34AaBWxWMYVvP_b8MDe0SZaI9;_;x+AMQTeg5CW5foXYg zGYt}uYwuVI%?)t5XbUtNx~~=lcp}IQ675lD7I%u5$Npw@128keg7-k-{$!RUX8Bnb zzc0|}V`Aic4xUZ$g9kZT?T(9$3oJH?6dtQHm2pc-*bOp$ zT5Yyorv7J+?%H#L^+B2L-Q2&J#DcV_ZSs{w{g`;+D=a$}!j6ToV{tjp7P6WC&5ng^dwtolkQH_;WMl72 zb}WP)3klfNB)2};u@C?|7Q&8&{D(KMV8=qfzj*~a7P84#*|89IEM#-97j`U!T^GX4 z!D82iZ0z{Tt_xw;g|O={{}VeTvtuEfN)|g7!j6ToVq6LdA?&)4Z?z!ox)63<$p7F*-t1TiI~KB0sLrkn+1mG>9SdQ{LfEko z0D!$m7JH8@_Fk0CE#>}Gr$w@3A?#Sl=Gj8*SO_~7!j6S(G6QxjWMdfEdt|Zq$YPwG z!QLZlb3-|HEQB2kVaGz)u@H7FWNYJh_8wW=8^8Zg9BRsrg=}i(?79$kT?o4_gneGf z1~Xuv7s40>b}WP)3t`7X*s+j}vqRYDg|H8O{ck_?mAyyShVH@MBa6L97JH8@#wj}N z^Fjas_IV)y0QSg9HYZNF9g6oFNDD=*yn|? z&rZk8vCmFtpPdeHXP=!8WD%;zlYnPvQ2e9|ZV(*c~K0AFig^|STcCD-* zS-9w&IEN7_PR%+0cpV^gee%Nf$KPy_jc=yuA^0oPBw^Quushz#v(pbWCO4*Pvd>OupPkM=J6%VBT^GVWJ3T9C#ZI4n zcKZL$+38&UA%xx{>9*)}*zn^92rNG~gkUD096qfkklyV*ndp0{(#t5(T5*H*eD=eM?a((saVN^>&>e7d-j1@0Z!0P?dD%4?Y4OPp5S_r~`8X zC`C`fho@P7;Ot28X$qKGQVigZQIN3AER`wuD}p%tn|iP*syps8&M5 z8B81ACwy)ym*ed`U1>=)bf!>=+7-S$fIayV;czW2N_R_&A65| z%2@*C+HNQ=e4W2+(82_>vE9pue}Y;!uX^~>u9nWk|1OmiHgiQM0U! z*Sji?&=BHBiqUxxQ1c*xYv=lRRk(}`1_NR?d8OEXN+26`_{`BR6fp`@#~F~j?5pJxR>Ak z_jk6Ky$;xVljZV~qK6R%lhMHO(WMWA;-Vg0w|HS$VOd^j_#IMhg{oTIk;#!iVxa9^ zH(I>MF$+=zG4uM7WjzBhWW`Zk-E->uMEM5!{on6&fjGA*7qep3r3yBVv1;Y=L4md# z+obDdB_68 zIT*bC5aO|lzm`FGy-CG&1|jx+qs;B`$xP`y_La6B{(&cO70GW=>Go)Thb4UPwy!K9 z{;P!HVs`wGEMXh>e*;T6ur#H5`XO;#8{Ddwaq}@hllx;lfDea8IiDvBl4p8#Ck3L7 zOKURk=SWT2#l{ZC_V-!*Q9+t~sYiy-N_>}_-%9;&f^J8H*v@897Y z+r9rQCN*ud4%pIZ=Kr#*EQCm8K8$s!BIYtH7JfZZ?5|tL7dMFzDy^-xbeQ>i`t&KP82k z^!*Pa{~tNc>+jI_Ht+aH{bC#UKk66TxW6%z?H~S)e(|!Sp`f>XV=G^Qat6oystsfo z+sj&@*8!L8+Vt0~9)2zjO4;j05v)>ZZ>lPR@9Cd#y;kq3LJGo7BoKelm5CY2{pwjg-#7L#)(RABz12%G zf8=}X*~4U#(2nNmm73K_qL@e_4JNBbFm%5%=tNwN=J)k9QbrN<|1$kS<3CRJ zzmgoxl{HQkPJUyY?UrXpD}3*En1}Py$R(R&!f&dx5;Wj<2vwZ?cHAy#@Qe!obURo?0-7k3skN=M3*DYet*LMX!-u{|3PKHr-;Qg_-5K=gVq6mBJx@n;n)=h zt$JRpWN4;aOTqSNC+`Kj`$<;EE<;4$TbDQdRo;=~0zTU243S2tlMe9t+?Y39sDAno z5}LN>yBfXL{9Vldt$O>m0rDM=G5SqO`A3dX_)XjR#xcGvqGL=mNCfJ0ex~-?!<03l z+LcdTcrId^1@WGLVC}?b+iZfcZH{G_D>yu>JY0*L?#tBG@7vpaf1dgOiJkU+=zrI$ z__p>wsCxIiI|htiRlNFy$3P4>v=Vyk1YMLc(S4h+JrPWq8Xs=DL1 zZGbU!zROAn+wL-Sz76B+ADQEKb-r(d^&NX|8^bpe+~)TGhTr}#8jnnRXK=FXMTJ?j zzhFx?(zp$kxZ>ekj*i2Uf9RdjeBt39HQ4TfF1!uAlVc=%Iww5meH^YA;p&{obG>FH z$$X(T+*0Pzcbzb=Z~Qy#y^@t=-pEQ-O6?nb>>jLnRBf@UPpOFTZv%Wn6#%D5?lR~F z&L1bY8bH3-$x9~~*G`c??k zuaynZgW#+@Sb|~2a%8D&YJ@JATf!Ni#PY}`+4eB`nrU)LxSwXM>14j`l7xOnxXMCG zIdB#(rwHQimQxD}!EjUeU%EajqWXtCsrVSd-?B(48f4enOb2H$mtA=Wgx`BGys5x`jv zB#rY&&*LeYJzjQ_2_BwtpWl8Sz}n%bafv)qBt@+rn--5Y5BK|3mk%dQ;du~Xm~>~R zk(Bx(%GIwE@JQ~rGJ$Z#VcijfV$0DZ4u)8J~Wq#mPg zfmyVQyI*wpNX#KscP)gLG+aehs)9W2t3IfB<(glC6ZEk| zD(%E|fCCt+Y*$v3+muyA0v`2p?W)G%%tRAZ(XX(mbwE@3UB*j3Tu=+9fkdXtI&=Ig zWk)*myqcVAwqBDSG7t@p4LhnXP1X>bJuH3`YfE$~AWBiWX>pu$o6q}@0C8AWC^s7_ z*YHNPCZS-pdR{|G%8m#8KXeFOjr~N5+Q4!qTLeCyBMv(>*ejeF+IqdC_z`&f4V$QJ_9au`fmTj-kX-p_%k)h3e z-n7{a4t8@jGbV}ysE!H_4yJMc{C=BU{zf~0YXE`V-GA&B6ol@Z2;jKOcfzh9Pk->~ zX(W2Oy`Sq2E~6n2YD?5cb%&xTqqMiYM}CyMDD`f%+)6G%U4MS8kk6J@TX>IvhGs$K zi{3nVGlB%3EiO4gr{;`+$?sqA5 zO(NzQ`Pfc6VTEr}W*f4UUmo&0xel27;kV6MLu;t!C~`jSn>&}5aa}8<%FM2(CFiI? z<5Wrzh|e-OB9gD9_{Fe&>BXL0SHs%!a{j{2je?uau~ZaJ{kHqXW9A(*nv>#GQ$Z1N;&zH#d%Ns&g-q~)<$ynMJW z3}`0ao)L&}q5-9-QWcb=gAW|qzws+|r^rL+w)hr+m5KeU#t9mL^GCT$A`ZaO z7Ko~@t=Z^I~F^=c9 z43o7K&L51>6^UAV*0#VvBqIyd;N|e<(MBL-(wkh&Fma*gkr0CxLVj(|CqxMOT)TuM zmmQaPyKUGcHv}4+xRVmj>&STaiYA3bLF!StDl6j4g#EabKyJfH9C1=h-OpXC=U(a) zhYTyvdopLUt)k2rf8;UAL>wY7`UvC&{EGb9DX6=yM$!N~iW?+!Xjf`CsZWL``J7$- zb90iQMh2-F>H^bN|v~TYGFmvryG#Nlqho zP1yS+oU(#ERt~|fsC-=!xyjHX0**3>4W;%o$0qp!Hi_@VziRNE05HZ5z+l10b2D0Z zKl~VDS%3TK=geRlsr^(4(RGfh4(Don_obKDrlrq3BwSH`zNe`Py2ai6=iStzEl(U& zW{l0E>G;hG{iNBj?2a5mRqbY&p$$$cHE3+qtt}5xGaPz~!6CFMKL$s{J@CtC^mgxB z_WdJIts4K(H}IGNBfGP-8 z4A5{27i!MnF29XW+_?))7kWyY1;xMZk2TPB(5t!`_E^(3NK9z>%al9zgImM{;^8cE zXGgdSwI=M47NRx^qy%FyYe#^C&k5Ro-YcV8Dz^6Sa(sp7lM)cFWnr=d@mhxK04k&o zGOIG_2x)*STwn~3jeKWgHb^zz>|9#&b*-Q()8QVY969f%DNFOn15#8lZL&lmmy&gb zR<5l{WpGyi)qn?#cLaGP)W4~_^jN>{jgPk3Yu#?krq4+@PeYNr(;Jhf)v&>^c0{N{ zST%#caW~e+J%$i}z$&+539qnn5y3|wqv?eT(O>>vZCn|jvUvE`Ep@1ks}yFMK%*xN4c)<7D@@HNbmFZ_yO ztni7lZhNzdPm^*6jmDWSvAD3uh$?iF5BSWDz7OXLy{{%;tOp*GTNO znWTT%U&~;vE6#dhM-3Ipmj!1FWgMu|x-PwjXq!kr@{CX#vMtv&unG~FFjd$ta?2-#d z3KLZ0O`?oao+#Z+4PFOaM;0+UtbffPFFeKSm-l#qiabi;$ zgz4(qjh=joR81Q`UrrmRDCcVdUHV6{7>GdL9fmg1oK;lL0}I{MCQzz+@jPwGYh!#u zbAPFADQT4bUQNqV#+e>rxx*0%=>~0e1h4@NrgCxW*}IlJ85+ zOgE{kE!}b4l4h30tq1gzr|nZDNT%SJF-!NUGYCsTv$)VCj30zl{sjrt{c&*h%PNB| zg;rH_RyiKh@$byChLGK7;KBt?UzcQL9evrjFY;5Xl};<6C`dLdln_jJY2K66%RSqm zt+MvPK7ldDw2jH>c(X8D{Riz+uxxKBL1PX#lyj+ zz@Z`a6&jBd#VY?=N`XLymiNzvCcVPqtnNc76#|5xodP)f%tYRjRsPRn3&^bq-lr-> z+5kQ9^Ooj*$)KsAP2EO8&~>G+>Ts`SRhZ#yJHA>Kd~Sc#Cab-i&!0Ckd?X}HX-zjM z4ksv9%+M%Pa@ z#sy=A!pW2a-1Nj(0@O{kwl}@4TOrr2jHh}QJB}9J?uQc_EJsu-m0yCjsfQ@z-Zx`p zvmOqvAYDW9JbDxOLo=FH^CJ+N+zouc4Q}aiw=dYHfikJl(?z=lxXmhWpU#8H=Bpao zU_g7Kr9HB!UK&kDTvu8@9Ynge<&@bH4N;b&Dd7`K3r@R9o3cr0R54d^$5V5gcc#+? z_FY8Xy>8uS5J<7aBggvay|Iq%o|;v0Q}A$sl1)S7)sWE(>tbm%BHAjh=hmdNs;qE3 zZLGO0ZU7@iJv*&TxF5HAD!D;&9q{@f*QV|r}?XF>Lg`l0S-q(fpwA{(NGfxmWF9T1%;}(2DfGNLL{zL?4#qgN>NI$ueXZ__w^}_3hY`oN>!)mvKyuC7C1=D0kD$ z-$zYg_F`P$PivJj@iFu%e2f*$vdINW8^&2haf5lZC{(G*UX_`Nh`mDmKSI0*b5@U5 zTVp_@azofifdgvFTy1Z8*P5IQ^0;0Krvn4Rjn8wtobi5v03y(~ee+k%lf1lqxa$LQ z)1RGcxakcysWLlVWu<>l0hjqX6cF^h?>}@&xEak}E(jZ*XCxW_Ro>ZN7 zr6P3oZ1hCeD!Tb|%7p(-GxK1 z@B*#nw5&aUBqz5EKH5nxKbZ^i1^cSk z#e9gqfaUX$+<)g?XQf$2gJm6BxG+Wx!wm!TSz4rzl&YD|XhfCw%-PM`wWk(_6qg4T z7{&I-E){VL+<>%xN+PDux~IqYv;-yyn;=V!UBihkr&0nTK$&ia0?Kvd9ttJ+TTkC( zx<^iU{j9M>d~SB`mB8X-?x`zXw+j!EE49|5AU!En?XvKpvVLo$3198peci>NI-upWjRqrzOPL}_)9 z3gB6l$!}|U^BZ&TVXlLLtD3c;j@3Q1`7sZ3uDf);2A7OczxN@DpF4Uv9sQ4ps9Er6 z)r4tA;A6nb>i`Xcz{;KZOjwLXemJH+ttsEcH%q-n`BV~EOIizwCU#qj%;0b$;gj@o zG;iO_mfSGPxuuEN`^tZnlN-w7KO2B;dx;Y;2>bxb{%-b5=oOa;Q{XB4>n1Ri{UFGGx^usP-Ir zx1AN|6RqITti4*mS|a=}GjlY`bsDv}kj?K{);H_9OYUZVJ2oD8JBqq=M6O~TFscVS zyI)JJWh}hb1?^IM{I&$qOthy-HciW=*Wa)SOo|*;DU>hQdLjFOJW~QUa(>e7pexl~^KiGh!X(D}Sj@SRU`(x3M(37n?EMR?-7MPtiM1yN+mJdQ>=L|V!j4Yc;5a)edc z`68-{rlslH1{v5o9#@J0k5>1DiWBElr=57Er5wNHqVVB0K^)}w(4;0J&IXHej-xg3 zRdX6yw0yF8;Y{?ykPP1T9Z8!yV@wbU1k)nG6d<^c^saK848IdguO6KEOQD5OBefP+ zeMT$J;xn{XhRBzuA;gW8s6Z%6QF&Zd!98@vQ*`8_j7gGoRcvE{ab-~a{GOoZSgSO7&@idfko%a?<@sQ}Oum`ojv#W=QNUh9#C z&!KT;W@%=kSQJ8rpy~{qop|8;H%j3!Y2E!S?`8HqbBrPO^a{GMu0Ukjojie6_IchB zX1&*|EUaMOfNE^#&+C<1u3reXjtze?=iO#g6g+SM!c~3`w zvnQ>1iQ<>eflW=n^#AL4?v*XfT z!^gl62)v~DVc_e46Y2K^GGSuV*yY_uD%W|_`ZJX*}1`bNA>n{+9|APGAD3WZ7v@LBjgk&(JPU2-98q*wsc65eZbVr}A9p-D8F^ zs(k{#Q#ngsC_3@y{9|gIMlAkhlY;ktM{%;yvG=iM)I+L1LQi0t+;uXyKf#J~@=rj#;X4yi=!M0e} z-$tQS;Lp64GO~O9;!$XZMqdom zR9m^`N-6XF)AWSTXK}y1ygV9QM}jI`L;{U_JH7(%6^!%)3PRIEra9s>1eT* z!a;npLD{W|h4?(FdUtI#Vmm%kP#6Ul;1ok+ zmCu(0>oXIq!lzHb#%z;tZpDXF)&U5)u!IPHsIqL((!`bJ-*1si>ejK?S5QgmFnM__6uf9NIvR-4l=a?7Bm1iRzOy00wf+Z#cT!!6^=FfWtp70!XU3<3UW%EXaW1Etd|991E_X z5c`heM5a$AiA=wucdyNwyixdV7@3YXcioRM%nTzG48AiP%CfY@gbK&v86rY`;%*9> z?iA-?;x73L$#*x|WGzg)qj#^=??;WgB&gD}ne9+q|m6c^_FMWe~9rsPMfW znj4ddU`T#Ps}@TOJI+V9##G5&$7F4iPWMn0xglUukQ$eOasWu!)m$Kl_jr&|pSE#k z%2WdX(7A9$lhs~lK2O(W^YB$4p|=n6jQUz6cf#GYdu4h`59HV;m+Dkg( z$~Ar7iYFtVru}XoU z%OyahaIO9JG;AbGqT-KScS}OKh6B3KWGU3T5-x!jjiUp>4btkalb8r7@7#-)+;Ga& z((C-E8_HM3l%%OWbwl}5hk*r#=4Wy|?5OfaD9?}_odb;x<;g*Ua*7gQ`c95a`&y|6 z?<=H{TDgkm;Ic_v(JptmN`*`ET@)@QF(`spF~!p`&=y_Aisc>^Vc zrk3lyQ|m`SZ3(HFhE3Y5uA*9wvDZF#MD^>OokJQIGhF6p8q}K*T0B%}!+s!m9k5vN z&NJaQkBCi3Y~SwWEItXz{cqBu_NZtG9MLd>NiDwwt^*DbBamQfozm`@26;CwYafXx zeq>pQoS7wU=J;g*H;PbG)Z%N;Xe$rpZ#=p}acC_I4>~z-C=*qi;W8;#kV+fNS{(z) z_lKI#)%eyr>6~jX<@)no>%6!Us6%FyN@9@U-3uoic0KAnp6qzDt1EoBfIv_dLJM}} zRj(G0OKM8wlKM$88Q-ko92WOHJYg>DrAjI>I=>pVzb767Gxc9%a4_g83G3X{%?{W1{W02|I+2OJdbj}1)?wG^W@ zkiwk@cXmDM&EJ35GC?C;`er%NA>kV4(6n++52uCD z1*drpYcwMg4Fp5i0ftvEvab9cp+wpv#;}h6k~*z`?DYcvQiwR(-uY_Y_7#T5MVJ7S zT5F+0W-r-%h!^>I(@`wPa!xCd>|7K(Z?7TNZivQBOoa4>qGr@~dS_TYES7ni!J|B3 zizN~3jJWHx)z$$oUQ5A`PB7TovSteiLwRF=SxC6{^`fPtWk?2>>wXnYSZi`(6ari( z2S0$Z37YO8>6^Pt#FR$&&7UU5Sx8snh6$)sUq$1Yxl_hzZT1xc#GU0cg3h6#$4eCbk#ujdf&i!%r6aI^E5=j+KpP~4UIkz z;Ce_Z{#13RU?iVNu+VCjHF??K)>M1enL7I~Qv(&M`hf*WX86a`PO-6l7{2LhjLmAK z?x~i(h+3D>bF~Vt5`;$0Y8-w}l+f!+N2er)FB({Z_|Nd#oxfBwqogToi*}ZOIV5}* z^*hEZCGr5mSPZqc7QEBq`Pppi)t<>#J7HXbOA!Kx#X|{F4M3!XieELod?!$T23%TTParTIJUX{X4YDwphA zJ0BFL&jhwdR4F9zR2Qob*%7)V;?Sryd(vO>OSeF~DKnr@;_k+$$Ue~}Kk#+O(#rmk03F{4^jnggCy zQD3VXxO=-8J2H@f!wH*p$Sw9R`qmQwEB3mU$NV57S$KE9`O_w?0M*GPQZLRYVpb^2tJq4+bwK{|^b=F(H9Dz1X|+UGC=$N&bjE^iv2a52UR~Zss(FVTTvKc`#oG&TM$}ElH9gf$FwNpqU5-gDR1*mVYLhEmop2~M z*}+Tnmgv-`<5o62BVBst#ZzGq*8xeDO;9^}b?=53jjR@g5#BNHxa&!G)_#NMNZRYX z@dif$um|1^rl|tc=?^EnxxwKf#iQp>QL;Mt@7K)qncD5TU!f+9)P?X_Pf&T(1|}nr zXE0_~L-#zBG5(Fm5B$~Q5lcPBpCG-0-xX_)4NlNF&21SUW#o4hD;7e7hYE7vqrOQy ze78*^hMZlAbCNd4CrQ#q;XlcbRRZf)Ta!}zCd-^YE#Xgn)aP$+ldVuKbn|m4bfy@! zt{l>080a%9-jMgM=}C589#2x!{NORPA=(pDk)&5Ck;3L{!5Buc2l^#eI(@u)#@O9Y zdAX=$)uf=~MzE|$L#u&IOPZyA=e2;!F8nc%ME42AyT-^)fdlgGBFDD0OkG;!``ZK)r#?XpG`)&{}WzXHW2q}a9el)IE?Es`Zc=g%C ziDjh^(o@J5s=lfrI>`)oMr9&)soTH-3(1k-p_X57^Fe0hD-TsnuLf$nxTlb4x{K2>~Z9dP>zAIfjS>jEpua2T}e zyvTboV{C!Zg-JJ?b?udctLloavgz==fuM>*BvF3LL#^kVTalXi3G8fyOS#M%Ehtvg?6*n>c(RDMPVzj&Otkj ztt9~(nxr6zs-QyUv7Vofr_%Tk^Qb-Zu&e>R-j&1}qE7zA;HnQJ>0RTx_|66MAlXC8 zxxTR2e$V3jaW=;n`t;W9%ehq;`6_ls5TMclq~{W`z#_{b#;B!10g2S7xyn?jx4vze zsMD&ejh)n@_R?mHU}1lFYQ)w^^btR~$8cHmM_YVF*|(8n>EtE`IKQP{xtxkap<;F+wkS*8mo69^M3D1pI#) z$rAW8#fxkylV*AREcG5vaK3>0epTJyF7klrG?!PsaFguod54ZXT}4K;6QRB+yhC7Y z^iIXCTv+tyiqAx|d5Kvq)mCA2@_pCxaQOLZYN2pp$fp)UN^^d=%x#a}u2udLe-~`z z&L;I4rwHp5jC}sfS0U?wwM>uRLk2k9aHuF&u5Kg9LaN}_>2OnuDak zC3<9fG3*nH6seDLfRXlgJH2qsF1*e|H`)x`)a)@@ViZgu(h+VAx<6U^Tgi@CpI2!l zMl?z9vp#bKqN9uKm9~^)xU~`7O3Gs@)1-T;kq2!)5F_cu73SOxEPu!LnM~Y|Juq>(^@jxlVLqsCISs% z&ZK!~n}X6m^!uE%p_WOt?%KWi_EoXTJY(AGCP@+8Kp3O}3{^mtch}GRzioMISTVS& z8sA#Sh_$y8R$DF6LzC)MZf#F9&p#!5p! zxqs5n;(N5xlC%yGOmlaJHmO%Ml@a(>&^8g;q0*Ce);E5M`k1b%!#OQL0anBBq#|jgI)3?q?%@3 zl2=iAhC_u?JqxYl>fTS}yrWD&!4c|{^yQzi15)kt@zQB(k8Ls} zto!b?k2Af;4w#$oIwGmHXlO_W4cnEgBl!7NR-jNp-eQL7s97Z%ad5xvHT$mLdyXll z6?{_rt4jBfig9U3Cah+xsT`BaP#&EVcRHywpFQf!Rg|my7xkf>J;(~i?Y~!rA{s5=IHhUHM!>v}0JciP8#aQ_)v9{Du-Sl-MrN0wj2@>;hiOiLJ2?-z9?rinTNV2Kw z&aXGQhm-Su3gHtZfPM`1Or=w(Ct7n?aT$ZKL^H`-NN>#fV}_>o2!1J-BuiX!H8*ta zfTX}5yt33dP*j;^Tqh@Ic%AcH(rR0Ns}1f{FG)-kE5?W&fUCSLy^IgfHyO@H!L_)n z+PlY`v0lQS-0x<4&kI%9uJ9xboAi4Iwp$9e>)P$QM})~ZJy!Q~09T1%gsPW+%kIwA zdI3hKNHn&%^H?;$j#2!GbK+4gygd{V0mVdS1YNzD{M^9(|10Cmo0={Y^md%rXDzJGBz=X~Js`CR<2@AX}NU_#cM9d4Y#!eosFGvkg%Jh-kH%cp_h zA_!95n>F_tR%F11yp@{qzRLa9l{DV6;Lfc=u#O=j1xki}!XTG` zrUTdaq&8h_mR#$#SQxB9<1wK^03EaFnIUtB4x4n&U0ny+6j-PzQL>x^1~~dTBg!Bh zdg4@Kn%qx?-?MPf>)wXM#PUibzNuq)=gECZIf0{D%djP5dzGFJuD4aa|H#O-^U^X_ ziy274pUiWmftj*?UKna$PK_ATjTYP{g6ZtSs#@_tPpFHlMK+)&ut@=C7I5pRL3NJ!mCW18L`_6e+#W%w! zmO5`W0x;WFw7SkFVqkBF z6X&%okFKdOetfB6$5HdQu5OC$xPA4+A&?}!j%lJ4dXD&7y+1zDE z&ZkAv63?#$RQkO2_2Y`FXs2Zf4b^?YtZ}5zRWBBEmK=Js070i zASNaf3MPce9xd~o)Me@`urEg+FrO-a6F&WpAbob0aafTjYt z%br;kUVo|t(?ew+Ly(w1uSx7SZC(KAkKM3sz`LV9#YrYIi8^$evKH7I-gf*(ec**1 zm1EZ({5C~*nuAj!=WfJ|j;ny8%wyXpS*%!Z3)lR$U!FWq3u&5sI|enB2@a?5_E9qc zH(Iz=Resu@w8qTkV`@*SMq$Em5+@|a!(dLvU!6Oq*5|Z7;cfNaZuR;T+XOG$Ti5!P z-reuh^d}IKa*qqTO@d0$qhI9`@CX(Gg}(7`G4*l$RLl^pdBUZGy1$56gpZy)2++X! z43=^e(`x7)D$fi=%Oc zRpGkDC;|*ViZ4Xz3gvYrV-S zV#!}F+$ZIfL?Q6@wOH7&aYrsQrEFQWsh1N|SWK6KcI^salPy9s_nUjXEKTZ!SVe3Z z9_CfTP*t)C4$y9N8wf~&f;2yT;P>al)#F=Q{znSeIXAjn)syow2)DuBI{NXm?CFAp z|GLCU{dB6qTNyjmwP`Lk-_YY|Rzy5EK9JHLOb2QGe5c`kJ()PFlN{McF8baP3Sc=b z&Fa=4bA}ftt=5VJ^LGDs&9a2JJ$p681;d+{Lvr+T=o~A^=B0`Oi#w#By+Yk96L=~N zl9iLE@#0oLofgdn1}mBmyEt3DtnGQKLzdWOzf|cCFMiMAg%USB>d4k83!%$_udhe` zR6EIx(CMk`pcb%C3+p=5s$y{ooJn9pI^;zCu4+zz=v80lF?%kP;ewe1?&)4@yVI?Mtsg>PV&aoxcI8t+sxxcTdP>81ROPMkOX=z4%Dd|x#p(Zt9qK?Fy?~W zx(VhM*qP;j25%huHTwH{((<;uuicY@gOT+s>Kpa1wTGPjYO~z^crFK9d8;2zV`}YH zysL5mxa>D63u48movE8!ikf(L6)N50*bQbHU92Q_sR#68)aOO2z27ErV5N<1z+04* zKDOk9b<^9QS3}GtQ0uX5RP|^QBP}#JDlfs(yrW_6^VyFVkIE-CcsX6>yaRu(9{zlB zEPp&m@*#2YtXR0R^9w^7v_34*Zy-TjT&2X-S=B!G{OPR$pkzOtV@6`N5HVNU{<5^n z$;nnT&yxNyjX(twcu7SK;fBcA8G!+<@P#@oPu?8QB##>DI zC3{Cr*wrC>>@=pN2xtAt$tW#HmS^0_ZnI&qZI2ejK}QU0#?s&`v9J6X6$2F!cU}<^ zE7?P>VpUlRoD86H+?b>h9{V<~6kP)!bbhPn%V_+S2|PRQR(pL)upU545-?d!KFc(?js_Xyb&sjK15gbvR+2TRZc15n9F^;PX5Fm zfIz3`#JE{U;z3rmD%vhKBgHe@XLExzjCu^(PYF$xqwP-Dz|dX1I42ZJ`x6!_dpMuv z^J2I<8$Z0$z5gXvb8$VH%o&FQPZ_z>d<(BS%<<#v_yO4lNb5;1$-?LN5_(d7;+v{K zf&mMLvDkpvx^RfQ6eq0i&R+|3wI5DXj2n$%`EnP))nJx+OH36*+S?k5M4f9#ot1Q$ zrIzY7Iw%>u@(gnEohm23b;K?t3@fD|gvC|1jwxLbG5*5Q17uz$SYL{u(-5zDjETz5 ziu8h0JZB#@FXrU~1e5L`M__pOPp9{?yV(bOkfFxTcKhxLJ}i8Y0g=sAn;GwvOg>Hx z=MELAsE8D=f2n=VKKooryJcN%CvhiRg8V7s`5{W{7XY_bqg?_`etkb1mE zT4)s@_&cG@o2ju63vr#@=4+_z%T4dQo@XDjUC`tM<~jRh3MPMoWiAw0B2fQ8JZAeE zC7g9_b#2RU#yOZ(``0Y=Sy5YA2sjl*yW|M^-#xniQ{L-TVWnc$I9ynI7)cG{h!`w(XILgU z6*=)DAh4sb%aZh9BfYB?t|z1%R=VFf(l|DEsXpNw_7Havx&BFuXe`<;9{X^-%2px! z5dLPuEbahr_5bMJm@Xie^03qInlXYy6~72vK&zSEUJ zJP1e~U>sGp>4}MHgOBx|Xw2z;U$A7v-N_ci_fI!@0&?4HBDV(PhzEIF9YaJ_lV*Mq zfv)`EKmwdvhh0n%@4vto3@*Iw(TFPdI%4c9wig`^R z<8|#@IdlB0-08}>>rod7kH+DerdA$GO1B$_q~@+Alv6o?)1gr}>#4E@uEKD@p-^aM z^JF~zH#}xK`{Yn4-o2i>7{alRiFqblyfO6ikV$IeboSL+HmsZ}m8o;fiKALzE|Hr< z1m{mACQ8Tu6}84{hi)p9tUHRM?p4P*oE}W(uurVpkvcjmB67?M@bFp%MrYyWoJdhf zV8{$7mQT==By^b%0YgQk;I_}*o`rKlBr??OuZ6h_ z(kiEXNM>zl)@{m&%@?iB6j`*-jE5KtG#4%5;N&4-j zXM)@7DXXyS+2UhRjGI+$dZVY2Fxa25aJP43LMPa&;&Eq*OLG0Wm4Re`UZ+zKV>h*| zuKZ)!ctl9u(dY@kEaG&Q1?6HTA_YgVGyq4FCnpjY7Xyy&dfor{*wfh}C`Z2wx_@j9 z%wRz&I(8N`Tlgg7%lOB6te30%wiRV3v7>l{5_YTRaPe{io67xki0NIVjck@p$8VO- zImacwp<|~lF2jx9h_xo88C*t6I@Uh8NdU6D_JhNjxIva4xY_-Ub!>G3X@)~@k&>m# zL}o5ipZzIlQu-n0sNUP_r(O?)_E{$^3EwMYK^|rWC%d7NkD{(ted(qT2Y1LXH}xQF33HSQj}n_x^+^m4*_i4CJdg6=`HQrcTDrlPibG}Oe^Q)9cBxmykM79+q3iw z1{0ce{ydb-wKDH9$095u9z5{7Z}aO&Rm91*rMc?y{$vXNUK5VupE#8y0ey1QG#P;H zX#f1t;!~?uuvNSnx$3x*BSezPK zI9M*>>eJXeTKqj{r{)wax-I4ZSPC!!ce&L>1YViqRO<;;lv{y43UpJ^XxaA7`zs5B zq3S-Fmzu|xQ(QfCNXv`sFPPrUT8N5rnHeR3+k)PwdsTTT zoi3H+Dv?hmEdkW_?<>a5NWjUy5;XbO|A8F22KVOS>@l_b?SNcW?_@88wbRHHQ4dxd z&|YH-pY4`^em`%hxBp+g)?d1HjhnB&Uz~f-9~Rnc?I$lTrhV_&3J;<%(*mRAO*(n? zce|#$t}SkLYMdi;fq4ckMB=oZf#I+=!LAZh`)xQ>6@AhAshx2vbANy}rO$&MDlnTKB+%9-a%)A4E|fS>LAQBl3Me=Ur9 zdW#=+GFdOZHK5PlG2E7qJ@n8o*JR8hz4FC74WpB@Rwe}17E41Kh>wA)q6Qf(1mcHG zen7p?XNcA(kMvG39}#K)5>QWu1PNa(Vrot|jQ5{2r5`j-HPK3YT}~dxVQl|(^M=h2 zi{+z#yz(=KCG|k}QwZdqy_}I3TQjo^3C50aWU=S$NK8~KK66q-v)GC+uSjd`gTP#y zfgm5cSw@2(Uh!Nn{Lu!hDB(HIL(imW7{T03qwqCheMr&vfJ8f2%3{8Jh5>d(nHv?W zx|Tm~F>kgv7hO<)kw{Afh>6$zcD#}QyWj)I2S6)RTk(uV^c7=3FqGSb(ayjcrc)`F z$6w*ngubOP&p84CY{1f z+WnZzT|%VG->h;9;*RqmYT|Jo$K(B>*k8T=R{D__C$WGql_3m403uniGY7Cud{GnF zz+&4ar?6KC*VeiyBd-TPDDg5LKc2^F-2Vy7DJ@Hph5Z4?_H|Yor)4W?jOehqis&VKp=XJp} z4sbp}M6H?&O=j*sPlFlp%Rfay5N^b3x^^+pGhWa3^Yakw8# z4QY?qV)L2|jouYaDJsWqM||=F(Z+`|F!(-ag<&8Lixw{KzVZ9&Xj>Tt<$!-U3DSXM zX2qTqOvJ3D2ajv|j6$9$)a9DF6CM%hUiA8o;uW{7Fnk*`60ddW#|=9G%rqqcaAxH| ztf1&OFNVV}_p(henOBx()$pBJkpPFrA!@%-gxcNid2h=xnnm$?v^SZ{2YN~~5~T%NHlBpHKd+qx zSi1MS;sugP_VA~{WNBwl?`e7YQk(B@x8RbRrrn;KhzJG@Zl8%h8?f%9<`iyK@7Ob3 z8Yiub%u4V{T?@_q96kjP!Q=4+rNajNAF@bK)&bjtNLAE@MgjR?kaqE3OIm|2HEr8_ z_FD8MtIl$Z%VmWSVoiU+9=)YbCma+=eyAB%Xru`^h4tzY7+Na{IcZ?G<`-sZ|!jAu^)BIq+i63k|{9hjo z-dOL?cJQYOBm4Q=JgRDo$`_|z0ms|gCr39K)xY-Z&b_H@6MQ(=zm))uJB+tInh_)g z0kZE;Pv_5H&2af#WIIrC$!AyAcwf$q4#9!oa26Ed?ZYAZu+z}k-G!WB*LbLI!x+@< z;D=!GU{bT^#4|?D%cMyfMdi)GnJ5Wo*$!GWbMDm0(X4+=D1Vjvq*-?kXQCsnSu-q& zPWG=1VQ{L+G>8)xjkZOko$N3}z+3{dj2gz*40(_6wJ;{2>P4=j?rLAh4bn;*rUGFe zfk~xQ<>>0VPDSs;?W^rc^d$CfD!S)E1!8Miw|#j?)pI$!Xb*iLhmTj6S@=}}u2{NM z*;Y&CcmY$__6=hqzp9`ji!DlQ5nbL%KmB?COgR!MOQI)A+G+GOZ3L%Qp|2wDeydA^ zg!y1<)`*Zbk`GfQq3wS_Ny)b97bT=brY=)A&qR?|xj8ivr(O*K ziAX&FPylqpE7<(Y>4O)FUPlC6(f^DNKlGxxKgXI9l$4yG#s~5)MRz;~;`>)dF8V7z zB4;<6W%tP=UFtS4P>iJ&Vwe~SSd-{16+yBWz^Z9Z+<0Oheo<(A;J^Pb%TtzQ+aAc7RrKhOQ`NM0F@l-*SL$KOu=`G!b5k(f9o M_y0V?p05M{4Jwz-i#AGr&_PRDcv;$kS&|@e|p!9+mQ& zxbjQm{PcVC{42lhQ|I_aia@?_{Pc@_zeve1I(6o>1`s6t)!ENJw~-Z*^8|l=>;B0w z;1a*^G>|VH1VFz15vPBjtmK5`htB|jV1VFh!elxG1QJ!(Tx>{E0fK&6$%-IvOMnmyhEW%wr@Uyz1QqnGzW7gJ!oU6R>dK$ z=AbLHE+mtl`OHj+`RL(xKuvnL%iX2!zuj>vaHg*xx(<^P@p?@I5g-TCw}qQ*R$J2h z0ot{MZXtQwC(~-1eTN}vy7uLPi1tZU&*c?PFC|nFhbrusMPBw#n#X6|x()X)K=gZeQNZ@YAWci5PE{zR_Vn z+zH@rbt^iH*Ubsd9=@z_i+eG$1~ZvH4#t=3#%4<%WPoITS6YYra9$CJo&)0{#1QZsibNG+_d=tYa_J1fUGW<*)8CM<(V?>NM2xm#TBOJephwyb=@7;#+i1@b~UQgCFC zX>|e+I?$F;C{(J{#1>O;*TMtDmv^j>vM==Vh+gPImJzGnK|0uK<=N7d^v^VzKQRCxp`(hP`t$M(8`+$-~s`pJ{%W%`f>voK{Ll8Rs@e<%W~8>P znk`teEV+0km}?^pmU)@4m|VP}Vo5X#&R*m|tb%gbp_xAMy+E%e1lg-;WUn@URH3zP z8Ku~@o{{cfU<`XQ@lxCW4TlU9Q*8%o!9ZYx(l%A!xB2R{k?z{_zM*+IftZHHk_c>LS*$$0Gt?19OP$5?g?Px31+4; zVJ2qei&lp1=<-ahgXEzwtTUx}LS|XW;%(M!r8QT~Xb9y>d__CiR$P8AF9cu6 zF=&V}&g~vN9@<36-I^J+{3?HJn?>Ys3Ioz)dm*c#FE^r)(R=)0M${xMV@ohA+&%p(-4>@|5dW=OsKR( znI3uN0;9L3Nd3D-{8=x#6K)7^9B7Dz@5y`EqfkIsPBPJ}UwXiJrV}ey2oeS{cWSPp zoM!ZsOkwz%Z>UrE;=3q0!=!+CE=CTUi8ui~-VFBMDjGQ|L5X@m0}ld+_N4CNP=D4OC!m9Hz10@&@db> zJgVQ~CvBVb?nnpS0w4Txhi66=5RNGgXGoS|l{uRX@D9Syg6*Yk)OESh-QZ$Oqo73E zV-^D!$7vlKzqSx33~KBBOA-+LCnx$;I&uGrMQCEz`oVsdcp3O2rz5W^rbk!{kFR7& zfk6maC3G?uo%La`5u;t6)!!eYC;gO=mREOqJhJ3SvmNrY^Ef4r8g6RSti$89=k*=< z4~eR*?oDf2dgg+M@H#86Fnt`x;NrTS270jW z+;k%~C=`MkMI`N0PKUA5#!1E|k$a+B#V=(&Du&Y}$=fN-ZPgO$iHk$h7?BedxOW1P)we zMKPv>*j_I;XY~RH3>PXVs%5sP7TRqL$0xU94Tci3-qo^E*Si-Ad~7t?X;cS}8n5^) zbkkP)Zfs`u^Q9T^HN*6fxgu!;zkI{D+V}6bd4|=u+t`bd3GGWpJwD;kwN)t8La9Wv z&bIW`tjy<~L9V1nL@GUouZ?ZD&OGTdDfxo<8 zTsFM7v6yAJdbs3V3ulN0xs-4klv3}Opp#R*bWO~^myUIxFVv~5*?r@Gqo!XCBM)aY zHcY#1Arl0B?;&-I_u=G}$>=1Hf_v}L2+M}k=?;-i$vwd%<2^jBNiW&B4PaAU`0QVD zxJ&pRho3$Kes~$#Y)T2|ao2hieYv?7KXYUCpxf)RaxcEg{BdgR+dqA|`HyjTCN7xM z@-tvu`sP1=`uEKM9WE#0MrPSyseDe?-8VkEfhaCB(w{e=jLWBU+K^eXfxYMM%#aT{ z(EI4Dpvx~aLqiJI(mw1m=Z^pCQ=D9NER-RQn#MZqrt()_k8`%}N`~YUfW6R^dGVf7 zZh~yfuCZqH3E-zZ-$|7dz@s?nsI~ud1tDpLCy<3Vhj%wl0M|8osSYOqfsQQ5y6ufa zQ`BnJ{VD4D+>s;Xk5)gDMRe6Ls>t;OP%f}m+IyJO4FP%_%3L)(0dzu%YoaHB)@Fe$ z;qKzC+s7Zt^xzLs9bT_ZMu#0&$o$b@+(&YJx+`sHt%hPh_)mI&hL$qZyZ;jPQHyV` zy&&$PQ{I#Bp!VDScQH}4!s&suL6TZ*c>aaZV9y{oOf=C=gxz2$IeGoy)@-+Na9jLi z=_TWw@-KfIRj0qqOOCrTt6k8d*Z~)YI!FX)ab_8=gGE7L18kh+4gz&GUz7OL9f`v! z;I&1^8(PT4tV``6FlA*F=cunWHoOrxr_t^!>iEW|-=8{K3+pOj;E_<4f2v$M-PQ;^ z_r%eAd$}X}$UfLxe_uNdtL}?B$I4|iS~^A(t>h@yrkd=YyxESG$CRY|yuEj&v5?^T(+cF=tK(^8Fd2qUiL-t*%QEpy9B4f?)rGa>>*;zGJXZ_21bxekfx$h-Z&&w z3!6EJsO{VLR_|SHAPFWCvxwp^Y9nW514W)*M?E%qY~gn%BQ zZLm-V7kvWYh4pakj)Xx?o{5EzFNB>e3HBig=a6E1sHVWgYIL?mQ+ zXf(oBR!rC(inumPCU^Wpf=;x~!jiGWe&j`%Cjp*2RXnQ5%Q0a~8I+M*vRFnk7F9Ap zdUq#j@1l#S$*P6Z()o^e_x8=o_1Ek3b6jU9l$sn2ge`&0hUwL!B55fGy2m^5N5j{E zKUDF*m)z~T#%X3R#{G2u;k%uN`v<-NXlex+K$#Os{ z*HcnBJx$d;LY7XkDOoZ;kW%wFHYW@rRU@a-hzVdotN<7h5=Zk(OPh!Y`?$* z+5VWA^j;;Z3(aR24^N$JFwS7xxPe>aLX;n4m6@PUmJZJ7=sb9Zg@}`E+a#d0$9Ra9 z{@IZ6meMreP^H>*9CU5~VH*Bnfo3Cy>@^t>u{QSJ?bwO*mLgUuW%Z1e2VV7%eNQ>v zM@D>=JM?-2-sgr-m5XD;9Q^X~^0;~1=C>H3ip3{@FCqjg6qhDe$fjFLW${-t-zR-H zZPRif_d+LR_v_5AYbL!JiW3r2P6>8k)*Yw$lt$UO7xL& zgd!jX7xsorM=84O+p3jX&g_ngSwC#E4Hgo+*RP0gP|7qQ1mi1D04CwPk}UxvPF>m& zsokq+`epSiDIG9TOYO2*7Gl640=9uGfy*6K@9=y>n@}Z0Ljw4Q+KVnMxk8ZU@9bpMR2hZ`KT4~!^Z3?06~D@eqI_n( zWE967lIX4%%vlJ8rDF4Qe0F5>Qn;Dn zVW=&;!W4faiytPzhL=-1eQ);ZWl$n39mMvSNwrWq8w91BLm5V3!$Hn_(iL_z;$5T< zdZZInSB-WZI1@Q0y-c3RU=@(kXpa5zn(|0z7;L6N#buVPjU@-G9cd(2bETYUsyAa8I*#Buy+9-ESiHO|;V92EHg zbvM3%cAy-&&+7cjLiVJaDTOm+x9Gay$o#(BftFxbWI*;hRt}3aq4VZ9>WoYJTuV2% z#VSq!+68c#GiD0}s^?4g2uWtI3@iU4>)Oh}7}9r_;-WIb8M&G|5Z30{6e}3IY!^K1 zLO^}H45lRV0{5`?7I$L~Oh3F5^G>7s=E2o^rJCO~O~59VOJ`mGXN{ zr`--FiBHK`0~vaGw=ze*nLYuO^0i51NcL#QjQH{}8CCCzRUx|BQF~2=C8tow_M6gE zIW()#4vj3f>7aJIMw8zU4n_W#-2EP;MKx6PTuuw|N7#z63}*4Ah1cNvh8EV?b!e6K zo!z)%g*PLuw<~NLICw)TgIRbS_(st@r#?0}EN6Y=lAOe(genqubC~!S@6By65GCX- zX=3Bn4H9+(iJEK%r(_8vN?885w5U4N?~w7@hn%P*Wu~_2PN=Z6>A`vxA?9aI%;xHU z>>XhdAkopBh@V3ReRJZ3U)%(xR5$yR?5wpZuzPd1cHz{$=52KQx=W`Dl|B*S?(N@k zw8=yYAsYDR?X(i(qwdd)(r5Lfb*I4l1No{$@{iw%4y1K+TQcYY*t_LdQqh>nyj5Eu zP=uCe;60S-@+wa3a=*&WmhY&y%kkynl{TW`FiC74Hg06(lj!Xt>k7kW#%89#EiP2M zsJm*6>OzFU*pWP5Lb*c;tP*|f|U|BG4YjtQ(11sa8m zL*R{50u05kkN(uJfmw?w1+~ECexoOF!-e97nLdO%04r)C1WGd$Nk zK;iAn@QHi5`_a_~3;41rn605tfxdtteVYgQ^Z);6-Py3Rv&rKZ_=1k{eT4iZIJ^3X z)Tck*AvuvhR1}aPQ7jo-;PNwetS%82bSm&EAnK3$fL{(*7jWs(ov~64=DBg>i3iF? zgYoM5H}+Ztx9wtI?*Hu|{!f3ee@gV;{5L}9Y8gMzP;}qS4AN;gI;Q++ZUnZ-1U4B1 z$3(NhJ?Qy>-VAB56*47E`I73wp?#O~hgNjZ7 z1lpXjW{h`d`p!`#M8E9vp=15Vo!XYGMxg1A*kg)J!7h`VB@gc18a!P1S^9>xwuQ@D zDN7pa+p<`__iIc1lPbR@{8y9I{~hJ{KT#I{gYs7y{{v<8A8r0K<-gkIzbtKbR~FP1 z3>k{sDA75j$P_l!6!_>Kk}hVJaO~%)PS|Blwh>>OKuvh_t7Njr(&9dGDcsw9=Xv(N zU~%Xbr-$ALg7})yqU*We-Os-L(Odl=%U%E^AXkQ}&5gIZR9P*N`wPigJ19E`QFHyg zloW1C)+dms`d?o70l)03Fy>$a3CYL-3=KAc;tp1eJg#DPfU?(kv9mPec4;@!*>CZTq?_Pc-h3GZY|F;0C-tF2WJ=C@iK7+fH{ABiE^NIIPeY<4j>gxi)o; z4&(8xdp%Keim}isd%uEwVTr)0zs=}H90wbaC#!RpVH8?JS)zqn#{~G&^+vdWu&;0; z#af>;0&5Rjdx@hbwDl^I?gG4ZRTj^c`})P2t1>_0ACXBBRgZ?>MXcnB<`M zFSvG92PRK<{?z zgP!?liH+SwrG1(@EWICNE|6JuDxG~rUQ!%E@QGBLTDA?vG}=Ph%?k6rFYqc@WvrK1 z41dosSNHtwMnSQDx7yr>-4sz-b!s6kyN3)6o!!OE#(^mbD`%m%o<`|$9ul{dwr^Fe z?yt3_ZWK_=z~V`72Fc0o9<^SK`ti`2u1U<@tZVemTV2drsX}(jk+fE>1xJT>w1CE; z`iru59NftMu5*3{_M~ST7A%Qt#wwdG_LG%;t`r%y6JUY8d#gd#}c~!ktLuVX> zVf@+b%8^38dDtA^Uvz#YduL_eEnZe`YuYJiOYk76$^WPUF+|AEUC7uRuyo<#4MU8@ z=h=yda{+PzUtPbz^%S6GG>XDzLeHpvKHKm-S_j@(H z`K#kT<`kjlLL}yI)z3xDkT=ffZ7yejFA(CD~{JM8U6$yG>(In(dwy>{vlBT13V!W-^UZ5I9AuThLO7aNhwJ;#mmR1OEi(I*Q%RkdL&)B8V z>t?ieA*H}m6~@4c@85C4F6~CY+E5<1mM%O}cOY?qWf8_{Z>ERMrOZBSPs?G(p3*X< zAeT2yK|QOc*Pm+IZKS83+ag#r8&dzQ#q7%Lkx(b1V9*{u97@BgFTT z{jm1^M?=RyQP2)B9q=Wy1IJ5-z1?KW3>1}E1m3xo35iN9M#$ye)_zmBMD*QjFDUAF z(CaY98$3S<_wwp<#2r+F8RhY|lbm;3E}stoUkS=xIrJeuG3SckKwfF#B$)AvklMvE zDfSypf?2pbS2Ev6XgMI@Lv`BpiOK0+eO0!@qAOmZ&B1Jb)mFwLdi~3V?~5HhP5?}- z4e(l$`o|XCo4#Ui;GZKS~6;kr)#5GI^c_%P^uS52b)5W zjcOO6_N-uidGKw&E(Q5MG9&$oVb|RCb}yB*NOe;l7v3=U5C#Tn2CoK>%UFyAmcq&M z4_U8nR}59B(*~MDbQg2IizBN_6OMx*U)QbU+Ge!1$(D_YjB>{+=CYN18VCqX#x&^5 zwSgGpn{^$auBgN?J@0D-*y4`}Hd5h{CVlhhPXq}9Ozeg`*<`+f{Av`RjIZcdnnUDe zoB(R}B}arqk0nhwl+@y*l71I0x4aX8i1RpPrRiI;_JXAxmYBYw#U4yI5({W$nqm-x zyjlmlf#&E}ydvP`do(nc7Y(%zjxa1=9Ys~=@G>C>lf}p1zTsY*)Hzi~<)!t#uOS$0;2H`riHqvVB(`UrD&i6A(-$J7{xs>2)_u!whBR@yOs}gMf); zCW5W$Qi4&$>eyOpQccX6f0|RAVvWlCj%=Rl{_?+}Qoe|`EW1m;-1y$H;K7y1a$%TF zX!|}N6G65*Vo} z44p|S@vrsJMl$RMK4g`6$NKMWw)QBT0Ib`5^W*!R?Z-T_{R4~oRc;XBsj6(tGGcN` zpV>UV_*@7DDUork^}Ksc&;9w^KyOvkf=HK4qUA{3zARqN#t78R=K?bxKCWXmWZk-Z z_&~AN?U1J0#D`&91|oH)-5Zm7i-sR2wi6z)_Ezt-K6ui-ad>c^R1-CZ6ahd7Rq1oUbB|V%E>{8 z-Pk5R4vs8X{4zOjke8RtY1#ZF`2=8>=XV}*_!H~!X+#~^)qm8O9!`+&^PD`i&7xWk zVmU}zBm?ZQt2kw>mz?x$7;BCTf1+?Hp`bG)%7qwE&kn^Sm@p*4vI~yq!zbn^?rLFLOy)=L|jtuxV?LQyjV;T%WNo5X0~wgYuTRMg)bu;2slIOrzEU(hP|`1c5fMUpzEDI=uMfss8 zruXc}t5E?%5p<(PhsutNLXyj)DH%uaV{|&UYPv(}1I;dHF{{Ob8Dd-vrSNSza$y&F z??Y*<%R7;IRqrGg6Vu}>->|&H>tc9r zYdc|%ky1FW96hcwkGx~vFAdX+$31p@x681$Vj(#1>!YZ?gK2|lj?o0!*vAg?UgbrK z9Las(5=n8}r3ZmmB0ZA(v=$mRF1cagTO*EfT_=DD4zL+sxMm3k0b7k{#sB6^3k(h} zzBurZz&jlxI!ViWEvRgTS!yy|8Z- zKG1_M99>>(kmHVR!=qD@fI#Xx-&ix8>4#rv_bW6_F0l7}MC7dQVoJAoBdH`P>;!<( zVX-F0#+9GH9kBLZ({BN>#Z_U!=aSlBN!Uu3$UtO6X`FdhzDUmrpz@+~w{9j%5(d^i z{!aJqZ))o7s=bvWS@QNfV%QreJw(D9))*@{atps50>=-Z9{LjSkyYT7=317% zu(ia-Pum9JS^qfNduTgtBuFm3*zw$aw!Q!7-2Pc45Z}QL8&HICjhKBl11L>z2M(t<9zXN0r9Hv zp0Dx-{I@@wizpCu(92zOSKW$I#=ZHl{lj{U|K)EDuD{<`uo~{WnlP(P9qV9uZ(2O@ zqJ;B7VaFBpuu`N>;@;%?l^%=AZJn%6Vq=R=YT&GPSyK|(i=3(!Zffg4BO0=2Gj~v2 z#3UYJJANeotko>QKb;9CW6O2Z?Zv#N8JcucqxTT|dJcH1_<3%*@^I#0NVJ)da1N){ zCrZ+8s5++6$o}5;-Q;A#-@18Cy-IBNMvE6d9yinNjGjqs3cJ!c_JLs%# zDKn9mi9S|7=#JfK0&hoU1zi+=t`VpG6}ILu`$7*-;)ao!BBq{hD@p2Q89K}4s@*E` z)h#s8)3(Ti+xT;ixw=qQBsqVNLfY^`-phrp?F8R}ei(I8CyUj&;1F!d$#iVHEoGkc zY0?Yr`%e~@@gsdP-GXVtjcLJpYaka4drAd!+d)KmOsa>hP&P!%DI44>)>=u5kK8~GLMmE8y-vU z@Tba|q4o5kV{0phx>XJ{N`ig@upnw4O6A(>loTNs5wo@}$c&?PPo|@x^aEl>@8$>V@Up z*oFvtOWkjIK(*~9CUEQtw!)NnEi-P$8mrYg?D{m9xTGB%dnKVU()0U5W`IB#@C1Mu zrAc7QI579iDI5wnc2SJT_zTfOdu4LvPF}x(8m_^;HXzSldf9IC__@nY_^!fpSB(n9 z(6(MG)u;Ng2Yr2*#|Y7KtP)Nllof8cX|eYEpu1kE2Zmlzhq>hV=%d72IUQY^9Y3)K zEpK5fTwk;7%16Z7~O`39Z?jc-PObJ4tpqvUIP zH=SbbLKY8b8ySyTGUIVW%f?|Iw4-PJF~{3;pPT?ZvaHU?zy4QrR|=kQDFFdDf3$;k za8t%#e{j0op)EDOO3s(Wc2Hyd}(%`t$CVa&r@||3WBx%!X^_zb%|BN3Y zk|SQ_`^RSSx~cN26KV_* zHnexqS_4lmFticeMPxnW=~)@+bxU;DRSwy|-OMfOTCeB@;uLpd3XDYE$*BYc+n+Z- z7v=@`!{JcSTX#($PYcLPG>Hz~B+S578EAlou8aN?l_UKzu`&W38JoLD1U2}^i<)$+ z;MUVyKJR=n6;d%X=9cyRQUlKoyDI#BD~F;}UfRG%WyhB2 ztPtQ)CZL!T0NmN!b@Jt4)2%12P04Mxnl_>Ag-V#_$~dDu-{|J{gE5qQ$X?rY*F6bn zm0H`(;6;(a)`{hIiamG264WYV=>`egm$qIUJdEPF5RHV(!D-RWsGpAW(rZn%=gIq) zBe7*zBo$HeB89uwHZc6lX(81++*Y#0IBDBxE;~`&-Y-4Lj(={P8?utQ*|w6Gim+ut zce!Y8%7(@3i9O?`t{xRik4;(=)M~djOfvND_#yHHKnWD`R!YiFx5l;of_p4I<*VMc zYJbzw0>mJ}89=_bgZZa}`*Z4FNlpq2H^%%n1peP5|J#EImML_&i!Dnbpo^jqoQ}$T^!yy8a?x)M9X`ky; z5k+Lppmkr*47!SnGQua8nEn|26vz@JImeF{NLv069L>3WOy)uIya~(5NPVt>d8`lV zLpnof=KDn^MyS}I(eo&&{;W0Tap|vRJ4wY^i}{eR^@wxk5v7OWs299fsLZS_!-S`1-uUE2 z%H5=cL_A;C$H-Gi0fkIUP=|+$O(+p3V^6F8@d1`hk|>8f4?`W~@`F;CJ+mvZK>F(3 zjrvSyBElf7vlfxx!3|7PYG->bhr=Wm+3mt0QKh3nF?62X$+ODs zU3E-vPM5;mT+71NQ~b-aR$xF$ZWiu&#k%2u>B!*u*qoI%C$>&VxPwD@i9zLvO=Mo? zoE`0xgEE|C8Pbj zBvaHxDpsM;*ogC|&5^&pk7m25lCgEDlyt&O8VJG0_~z@Vnwt!#&S6uE*HB&LWtBU9 z^G1(vhq?C8Mf8TC7ZfQX?EPFtjC}Nulnj?CY#icdhdj#Jx5$xZVM`Qr32?bPDql*x zUv(x}Sc^w=drR<}WjDll$x#PlZAQi}-dZ$tFKEJnB)VSq*|?73u9()B7g`+)L%X#) z!^#5lQM*-*vI*kl=+(;%yJ&*7-V|FFf+ zxkV5cdR4jz{IGnML0@(M=nbuA>Pa3)I>aILQlZZKnFO}P}^8%cQyYBb|*ZMbrk zu){9)#cEWE{^$KR>E>293$U@V0&6idVwJGa89SyaKVk)D+aN2~2A)KDSXhS=`_So6 z;2r`Wc#QI)X4#vyFlz}Cu%`-HBrC_zn3T(#z}iwnTLTc^k}o^uxxVom#U6)=En<}0 zVMwww-YX`$iEnJ#+=NsLby-`_k{>r-e%e)ahZy>z1O$py8^Xp3dze@PH~Eu~JX0qa z=IbnDGRStOi9RgVW+>xt*72P3ckU8RT?c1vfM6+*kwJ43<<0nDWS^G&r(0#O#~C+k z8R3p(c!etRIJFEf*I*LW;0*Pa&hyr@UQs1t;9{#z+A5kE1C@?#1E?+Ky>J$k&4gI- zl#dMYp@Wycy-XfDhcWnWs30QLT)sPw1fCAd2l!%n5_>h&UTM| z`iciD_S&i)ZBQmGr{3$YmT+tdtzwrK(sB8Ayot7Fv%0iv-4!)m8k5qrDci@r=G(V~ zSto!!Nsq@~8gy%J{WW7XrP=laJ5DNc1>^N1h0|0So?AKCQAT4unclnu^H$jtQ=iEh zB(#)^G9`~^-w{Pv{MDyjlO@S_?2J6?ag9lG+~9dUz3legZ;B%Ir_{z%sxEc$RYOCB zfF~c+rp1YR9TN*pqRiAZ*Pb97D{cxFxfUgpVJ7^xz+>YoMDAj@Z!<%0;h^cc@_y;3 ze4p>PLvWiM>W#UM(hJj8>U$Y7KkW8-8m92gE;mcpRU}s0)3UTJ^wwUqMhmAMdzw!h zYe`tkhM?$o4^un-?(+S;%=^DzruaX_^CLO`iRVXU|LBA1_cLETY@fQf>#+n^PHBos zb6t6;SU9>+Qg{N8<2vGFKJU4dG&f3N-H6O-fLyztQQu*tgVXDN_3+*m?O$KU|8v~@ zHYm@@&lse&5050oBl^{a5y?2g28=okYJS{{h%(vTckuZ}&n9B7_z0@AxjhUcN^S*I zvDQuiyN7>=(`RoUqK*y3f5G0*Y|}x9^{XD9yd3V_2MD8sT083V9PIHdiLYn1yfHMqqnC%~;_8m7GnBBe5+U}}66uJ`fup1oK(*yt zvhUVn@{v<lNvyDzdBL&k;~)BW}$HeMt$lg+cL5QQ{tsW5oX}^b+qN8sf|M88O!{dWp%6ccYE-HwCb1&)v5!Mtd1#C zIS<2!JZ`8)cO9tF!BVH540XShXbc(-yXucmH%jHhYc`Zq3j3%%yj)#+{K$;(<`JB% z7NL?IAHb=!MitHMx-KerOb-r~*L&Z9!Ix|tv$1nor{bMWQ0p%re(yORpsv|pNeTE1*M=G`pD#DLwlS}RC#$qu6wZRE{_$pCb=DUUH^r$W; zs>z&iTAa$bYMO&T>uSuQEf|p=)aAdyXwJIRY=m7F3UIO?Qtk{56dixSZ9x>sU%d$m zO^1;=&1GZ?%Gzo#B@;0_IozDtw#X4pd#gU0Ccel)3PyUPLMc|qfqV7Tqt7NGbSkAQ zVVADi{5n$axt4Kq>q3x47!c#u9#&SHlx`ZiJi*1n^@t)pdqu`l97CPPU`0le4@I|E zBb^5Ipn5KO?QOZ#!iMmTM%{5k{JlgB9OA5iLnXaF3djCDoY|p1S;LP$XtX*#Gej2S zIT0wX+$mUjPLu*tXgk3v@uHtg7#r=fOUKR%obuAk;pB@Bw@tjrJMv4Z;s-tG$LOr= z>uMd%(yoQ<*Th)iwT|D`?Db5NaXQ_8IPuX1jDH@QV@KW9;f<{hqp~9tTfJ7s z>7cyKU@v~~(a_RBEZMYv-@i1R4Y4-Q<86uMbo9_w-p4rTkaIk0AK-)@QBA_Bw%6q2 zBoDz-H>>oKEeOG2z?2W!c`HckkzO*OYc|371DjH4FoLH<)tQLk;hT<3ZY?&YOHdJ#T zy}h^Z@mHS))J4~U0Df@rfIh0p5T3A?_oU8xFEcjU6;(Ec4Loo8`ZpD3>VCGQ>Xa;g zeQ1Sf71O-p_t)kMan+&&&AD>kc24<1!{z~!xj`6=exu37SXxQjI#!BwT1~h+k6Qn$pv0t0~>5t!i^oRxj zSb;EGFUfilEbQ^{oUd)c)|059i)`t$C59Kw{{D58kBI&`Yc{|3^_S5>&$$NrqA)|ZaR68Yk+r6wJL@kagE| z<`a&zqC5t!12v(getTLL@?x#&*GsJOt!B+g( z`^WK)YA#0Ge!UKyWhDo8#6l`(`qapuKkA#n$Pb4HPhxyB-#;;*dPsir@g1(onzoJp z0DSF*{gvXv#}H8X+(+1-B-qj<1NQ1%&3C6h{K5S+KXxYkmmr$!OOagZfB1)*1BtVT zIzu+RHtwzE%`bngf66Ty5%)X1^Htm*ZX=lc%hcuXu4??j7633s5vvw@tC{gqW4YGz z+VOA0O8&pjBM{k~#&e4LL>T`u;eRbL7p2}qV|gg)vst1hf`}iC=JLf{Rf990Num*ae9aOC!%Yg91m%r zhNZqbMinF&ugw03av=yLX++-o;MNOB??x7lb4c4x`WvcvP9}@8hkku{EEAXqS-$-A z1OVxQ$RIal{$S%R&{EvCoUL&5tWRcck2b|S_Wpwl{r1MjZhXz16TnBppzcNp>>Y>f z9<=183-wNHcV-{ITM*j#yVRgR*uZc8pJaNoGPingr<4CJo8OtC|KI|aF#IQx!59HZ zFCrj4l{31bO*VDPNXDmD5AaV`1i5f3n}W1>3#J&YZ2WoaapwCAQ%Uu7`f-DdwNUf7 zsZsIkIo3~qeKoRGr)?a*`wKUW#@>njD{RXxri2s?hdjST=exc}jsRh3cg zMQirr+?Sp?h&oirQf{A5|5b#)%fUY%{E8j^hgbI6eARZuk<+%mdMV|7=h69I@+E}L z+Ci@S>9Y$T6aGs6um2Zdpr>9sMP3b&ICPh}2$z>dxKk`eR*KK>V?ST>U;ZlBFCLQG z3r>f6u0}aW4=8h*ecLTT6Zp74mBj{*z!*=t2g4{q9QVykbwXOx@bXyv5C%yNPRbj)0;b_ z8I48~BRcZ+SfylKp_?!ebWPtoVJ4Hr5B2*VCuJtjaCEM0f7>KV)40oNDeRe`ARl&L z>41i%Kim3rji6^Sp+P_9ow1zG4NJdX+*d9hCJcK)_DISDLRWC@#xiF@md3wzXJcek z`W#-q@>KJ#V02ZmTn9uFUeQeGCcQ!;!s*n|ZH<(o&$;KzM#P3|g3h5nG#blHxX$H^ z^1c1dL1JD}VGA67OfnGIR1p}8W=F+tSHYbP1gt)r?%)OSeW!f?l&Y;HZWkp`{g58I z9+q4r_Cs_$W-$=w?uQ3Amf*{9h4dz`rqKf}+EsrJ6JkO$Y2ur-g?ASe(9`^&fB(aL z(`lWsf(_tkg$>A3#8#5$Mu@U8iSqGSE5K@Ls{kWsElOXq9*~T*T=6qqC591}AfQD> zNI{WLf!8ZJ>!kpMUL=KH`UUNSoHAnp>tAc)Iyc+`>`=-}Mi(+$!fDo4%Lnu3_lx zjT8h^gw8NB{B~mS9|QEGeZ~LyCZp)H4a!%2#Y09{F#a-3IZ^vhsiRnigF$gb@#+T2 zYGp#5<{yI%k9@$;Iw$7G4=xKhW;E2pLis^s#DQxnKc8F-|xE^YEdV4FQENYnm|&G93v%OPWko@^Oeu;VAR!y)qbE?vHh3moazS{N@XRR0d5}eBtz-pytN8O%*YjN zBrK7Al`nzBjn;e+XSE*HkEdkw?fiUnCO=$Khlj)Ao_0$stkR6Vx24zpxf2ki#eOR3 z1(q4%P8`@sVb^!54|d*r!fkUZVRvHNQ>q>TA>(*29r}UTt)C*0h^5#Iuo03Pd*5w$ z+|3ruv_nk#CILG$iUXS~YkLmHA9~*-qUvpviLm)8gqpd&@$vzCB*(a7AiA!;ZD|F1 zd*jl%v<*v8WO9Clw|`L{y6T}zmJrn1q;WR7bHL6gCUQH(`&-X7r^KuD@hYL<23v50 z4&BDJS4Qk;gnsYvC|gBd|4|M0MfrpMM{(2HudaBMf{-Dko4HEV+iT+ZIagln zBwu0pVv*2}3zkhaclewSeys&Wdiab~pDUjG?8I>LsF^c8e+g}uJ2pjD5@KU2TH+gPN<^{Jm zXY)IJB^D+Nv;~B80yq=NnaNpL-)B6>kX`R$2=UV6?mm~20J5EpiP;{$j02o|vX(?0 zq?+*kE|fl+2}88qM&^xG=!pJKFs4Nm=9KeceN=m47K1O}47@Jfl5eTWS}!jgUj{c7 zGU~3s3KC9lgEuG`Q_xqW5pLYhYu+PVaC! z{MEFha@g=Hdo~1)ZigkMq=khYjQ%h7zB{g|Z2LFp*hK~rQK}b55D_GVUY(ILGX@w2 zLI(u|q$L7@(4%9cqar0#2Q)xJk!}b@q$vb2hR_4jYv>&U?}(%1{_f20z59FjSKj-C zKMwoAF6W$G*81+f)_2ji*r%f8o_(2w2a_+{Zy zG$J9W&(Gfb=GGmH=Z_9VeklU{cJT7+oBn=MT;Rn;F6DyWZIC;MsR>=oM^&l*6;e$> zFG(t>Uk&-@+}7&@J1wzf_kUD&v#k!~8a%A-skBaVbZCwsPMiU%&dh z)FK-fSn%5)@7)&isgOtXy$2z^Uo&-XGKgKY@yLP58kdUNDOn!z>i)dtX`9rPeacv2 zhq2&fHFi2T3g#HHrri9bnzF7Ipk=GG(Ym6Rr~{VOQxgvwr*UyoRwQ#gtt_=n13k=M z=G}=GN!SJfriyUKuvMM-G{M_G05fNFZL!#_Y-U2ziFzffhaQxob}D7r z&WoDrp`%w&Ikgp2Pl)TC?H_N zK2UjXt;}xF+noql@py5pgQvdoCWZ3;m9k_j_e6QhP}gsa(wpUB)tV@qWd2o#SxEpF zSj259*~im@krsU4s@sd?PD`3v_*-hi)n7N^i*aONYr?Oq@?* zG#&!D=SnS7W{klfH`3lL9CKKY`4tU@WL{2yK@0XvaA@C4(Aa(CM$JbpZ7IsE-1Sf*ZU_2Fqf$m-<5 z!BfHB0xw-!#Jc2;a3J`?Yp>5K{G9&EvdC!NHi?+Tt*^>ep(H=&>v}0YKj6}1lXv<4 z0$Fm$YdN7B8VGzE7T{un)Y(N7U^0jt9Ke>;2Z0b%74W)!UQK?GN)OGOS7Q-6T^kQX zPd^Nf&!MWgn$otG0^Uywx^Mc#o?5tc=x+Lf&|yv64RLtT8-HbBQ<)T)EkjXXxsZ<1 zmiISaxoSV(Q63U@SEnN_Y5g3@vXihS=&#S;v<*TjCIQC5dQ(;>W%v8R;|r!V0gdog z-Lc?^cpHCNy$AlB?>qc{Bx9@yA6Diswx{UGl3?~roa=3H?ZavHBeQj@o{HL?HP-5; z+o0~odAsVNIAQk7(UTN$v&FY!o%-Z24$fE;<^*{&u$b{J zHPoGM`>>MV=ho~^Lz^a>7J76-@)Yl{<1n;}Tq;(6CYjS1+PtM;-2|KkX37(N%7GKbVLv5GXTU1Wx6aZ&;qd?-!v)^7dVFs= zP4Kp$>K|(e8`FtSdjQkT=i{Akxz92!HXyzuQL&av2zri}r|je|j89jbl$`D2!Kob5 z@ID^?)d_Yli7tMXoK>jhkPf$;hJ+TC(mKQEG?642l;myekHj7Gh;dc6Uz7C2<`Dvg zd#_>)WNe(L1-d!h6*Q4O5g`s<)5>LyK85sa_uhA1;Yk^oEyEb zY~)pjt@6gg=zdV=q#&mwoS{Xr4)RjmnxqPbA1=P6ZtKQuj!K`O*+{?@J{x6>HzU-w z`_cz9^ra7LZ9n-fs$8T(ZDv;#sg!1%Lx3TSh=#TtG7a^-VOD=tQ~xAHB}F=~KYY0MvPTZ~^g7*I%G>`V65_3zT#j8vo5a(=SZwGM!A&RoHa-16hs3@@?_ zh4rKkdyV5tUlOi1E?Qw~>`eTgTD3rq$eQKZVcxxf5grrO z19OY|Xr5~ilc3G(67}r8r`81rAF5ASIV5VYJ*XaXP0Q<0_IZ$3Z-#41w9Kf!o|-|E z7YpvQ3{>qN5LO8g7hLf5gN^kBOq%L`pCJu}O-M`^J*XPU23TSZ1P?nVb8_4@qC+$t zEvB3Gl1JNSy&DT_RuoGt-87(!Iv!A{L1k5ib)Chqgj;0JB|CK}!f5)}E(gc1YNNnA zQFfQr{3CE3a*AgK8_A-goxk=b}wJ+C(l| zFrY?~wWT<~296?&*b?~mgG#3C$)E6Y_wN>^OLo;$AH1pK9o`Vi)@h73hMWLQRZFzB zdwYpi9|#d{FU8)Cr@}`RE}j%zB&#^*P-U77*H$zDPezRP)K!JiNih4NymHOLBWBLr z*kVMKv6U^FTOrX)?L0PTR8I~d|BDAG(2{}8d;CaS(@S#cqP^zV0X3Fk^}?TJ4aG81Ylgj;FN?N0_iPi2T2kB`Vp1+J|#Z= z>#|mHuSeHhsOxHy?P_B9VSVGnnJHLNe~7E)JCBRC!wG_oVs-kk(m5W>TEVHTQ%*wIZr^9K-gZ*s$XN<+Z zV;g+EyNaErC9iBe@@)Xl7`|zqTB+>0+h(Nvvz4B1-n0Nm1lK53_>)Q;jDE>B_NmZ8 z{OhN%kKe(pn6{38W7<-}D^RQajX)+9)6*bZmMtAN_s7^cB9R{(-}>x9=!DV=DL;Ng*}c z;DWnVy>&KDBPMxpd@SI5#4b}qYMFibzJ3C_Pci5FIJ;QbX?ch0DLbq}^rj)aH^=q?zX(@yYs=hDqt#2*PA^pp#) zu}K%J8j!bjHIIiIw1f!zP`i>b=?P#8zfeU|(~^CSr143tM5qL8fNs^%Wo#Q4aSYHpS-dUp*x{^Nhgb2;iA`1p?PxSsjb@XIX!jbSYajk z0Kr2}4ATNj*=KU{S+?h@+l|h0-6hBlAX=1Wq@998-P8`-*7ncl*V(4Dt!WtuJy|21 zn@G~cNtbOcr#8F!M4R|)#uX`q4EQDLKp>o@XKNO_;a$zJXVMcdi$&VxRm-)N>iCk# zPU`g@wL;pby(KjsSDtH~Bcit`wmq2zOYjDn2yB~zgkJo0Y?+bJa`6Dlyk>e0;;y1x z*_V25`kG10(uhj);l>wU358ikwM*rnY~m;Y=mLVO>7)C-s31c;36a}oJ1DV8m{qi5K83s*l1jjINaT^*Ev4nNY>>0 zt_yT;7fo(`r!8SBr+ki8uGgg3hz$xJ9Ff#xcSAJf`TV1|b4n0r1xvTsfmsy#=NiC4 zuG)ei@FBDiK8i>SBP&-J-b$LCe-#{C46V`Tz)O)Wes@pAjrbY8sOurv_t1zoZ4NC& zPRrRynR9S62|<4uaRR(yhRE3fSZAav7AkA zl+KA3-t@@#bKH(<`ZB|n9Kg^*!5usqeEsLAyGsyjd?f%JXxt*npAS8iq+Ig01?=rR z)1T8@J?NrQui15eWFa$BZrsFK*}gN|$IxCM_fj89X`bdAB?8o;0Iz0c(nr~#frBZ9 zEz)y|iN#fsRPV<4Ylcl+3EipA)dRvk`tk!Q(K?2!;t_z%MfmzU-sk^7Smrz3Ai6i@ z#;RYOh2$lb5V|?e*CsR?9+%X6XcW;?_PDC)aSbt7Ecx>3qOVKDp^u<;KHvAJAH0`| zRybNBgFg^d-IGvgN-(Sq)0n0;<(z#;R#h(hxiJ^lRrgbx=ueaUf=bfL#$yS_2r`ct z7Kd7JOx<*gP!&Aa9~;+Y6w)X=5})``t}-<50d7eiTPABrv06t>N45bTN_xS)igRlh z7phaJ#&5>6v+2J!|8%={cm=Sl-&1>Xwq(XM^rQs2RMTF~JRcy(~UoplK``U~r;ttZ%V` zqEbWt=D$$=zOUEUN@@WUv<^>o zH>NfV+P6cHfrZ{94uNYaTtnoza-!qWakL7BMd+eR zOa*DV+u1eh!Mp5{yy&_MK~sI?@o9PsCWfBA|mG8#MFZ{PnDwp(#^nj%< zf4%GX4e%tlpbsV&@!HoRM|Bjsqsfts!#z|B^AKM#z^4l9kcvX$mqRFiW*<(Myupav zC)u7p;c=4~)Q~#W9mxSnB10J}#t)J3@w*9nA?kN~JO&gzqAglfa9ieCYg+Ab{jLLr z({hQb#u?bD%_C5Fs#64VrjJZ6FY_vRGU2jvWzd}_7GGx`*Aw1KRTcOU$zFlxm9Jk( zyW76tIHx@eElVUUPJjm#hJST$vMwnXxD&4KccyM2wMn(PmSnUiOA|1P9_hiNt=vH!o2iibNuR_^{w`5BYOa2hd~kUB2;ak!pz7Y#n*^s}Fif^)E;#aV z89nlPysZR9vG@gL1)?G`C6QUKwWoKocrs~IEI0z`)_**2Xm-r+G!!{{(N>ZgYU6j1 z(zzVd=qZ;<#I(C_>cH$4pr)z9@mS#tW1~>y#T?7K)z_trZkZ^boLzO60WcXE;2@9d z=j7;$N5RS&Tun$Uv5p-6#Q8US70b05-Nb?@2Y7hQ617MEX9QA;PS6CF;u zXl@|&Ts%0Zx&--k|HgfR$ILj+Xa3jTFU4L7dZj=ZP9=)4cYQ8 ziN^j)$kA`H_CBd_L?7eWLFs&O4Z#z$!s~+K%d1=h%+e@>;XP)~ibIuv7i;PCFk#E? zw<5zuQvy7c4E0WRKmulk^evSz?4kP8i&Q?P4x=YzLU>u*g&Omhot z@aed%L1cI%-*Mb2K^pYSu^#E~N z^fkdJC^}%kX#C!l#RHPBzl=~$71)s&gqZ?|VHX{qJvC=2X(QR#>5xIX-k z+6NEz?;-qjhC}d65#aZ(9&I9mawy*#63ox*`gPvdGxFA$|E&}yYtA?RL2QfW|gj&?m?0vP2Pao_bf7U!);9L@oGxn!8b+Y7mI5c{QUPIG> zQ&qL}dWxFuZ`JmvlwYa1CZJ~9GvXz?2vC>~k_3;7O95SdJx=#Y^?Lkh#7m01qlRh| zT1dTL+nrvj1Li6nGEk=_D^sg~7;6G2xj+n^E_R_cN%=jR`uA%Lylg~lXpoawOcI3) zLPTjWB<|*{$&sD8(2x(O7e8hj=y=xgaC5^*V|}HePn#J%frhd<^~G1?=|UGP!wQl+ zUfXr}L(OtR2Jq23=Bb(Sj=@3w;MximEmdqICRv0}{jrw3kjKSVZ@y?_qr**;@)8u4 z^8`cV=AeaInMP;VzD$PR825(t~&y2mB2H~VI@e& z^(?3Hk0o`FdN|N|^m?fAqOz1_x9ICRwk|rQ2o30hs6P}z`#46E%#TYJwC0MGemCq> z2TRbmB;`t14Q8a~xWcSwfT$cPCz>G2Py1vO<#(=X_E5$w z!y?1dWr>`$wYCwW{D_h;E>+1Un@sdSMY4?|rMlNnL&Enxo4FzXQlSYFW%{}{a};Ww zkw%~phc)#D_+HJ-ieEZM@;bY>n9w+1rrAgclOdRq?9a|}mT#*3Ywl;Jw zwyX|Z-7G2=9CSmj`9q*$))p8D7 zHVLV>kDWc`>zCaCHYxRslvYBUIi`watjm`l>nc`{c0HD0G6D7+?ko=tXr^tsv}+>W zLY&gk>|3YCovwu@SJY^pI%<{sT6a(I7#41o6Bo&E-oI(3ZCTzRnw;!x=WQmNejaZc zVSX>v)HTf0d*Wtlva4K{`B({GnGmgqu!xAaaR{w;iL%E?wn&G*@~KPps> z>04{IUBvVUP-6BJ?Kkek=|0i-O zsxj4de$_*H8>C1A;M@L(hAEw+%{FkBC65vu}g0B%*h}PSihS3U4wmJHGX_^2wHCyiGCrX=lxdN#pWJ zvuKiL;}Q%yU0^gF9R1!dM%}bO+lMju=!JTw=P}IlIleY#8>X zzd!wEGK_1_-nH3G9L*wD-?Yk@J;(X+XW5URO9*@4HFgzsjraRu!gu~?<3-XO&sc$G zXHKn*hditji6rUW83}}RUD=ETIzXqAMCH%%9ja8jxL>GK;)lBo*Z<&#;pE1Y*Q*e@6 z)alB$oJAy2bRdh+^8GXq|3J-rV2eh&Q5@GR*~^yzA6B_J3Omts4GC?)yNqfLot|#; z#J(UQ+R>24i!du54jy@^?lj61cNUwIBBXr<3J#92i}4}pecMYs*fsrA!g$zz)669p zuIL<$s?+izbJ~)!RB0rfaTV`H3(So$oDZeXy<( zm%b3^xFDQ5k=*}avZj-2uGUdGwq^&Lu$a8`W)n{#BHb`SgNFM2VuRrQZjtH_xMGFI zQ`(d@$X)83wj3D?-KY3g2AoQq3GMELNPdbY1S|xdzESufl>u4;x^Z~kMmg-%hqw9xZ3~+-f0>^IW^vT2-((FjLujn zQ@i?PLfo4acoywmaLT!Lp+tkUNY&6jm~sS4k7O^*PNS*iT*0i^KD;r6t?Q2@B8_#T*PS4vz#Vv1So=>{cFT^aDSQHQ2}sdfFcH zdesN71EkD@5bY2(+W*!!lPx6VgE)HP&h&Ls}=CAvf%S`TQ*wRmGA+3(te_ir*Z05^@NoMiNQ zXt`bzildvu+ZKEIxa4NN+&q4J*;{wOd9&-rxRGbqEpF}*}Pjaqp z3Mn8ikM|LYu-}NJWnj0_{_VR0Xr)49w*6@XXgS8sap z7)uo!d$Bk_P6*{)TtxR#XZ7`&{64Yq&2uwv(d_{-YkGS9mjbU$Mx~U=!H_^u2OVUs zA%KaJ3*m`EhP$l_B-XibntR^x4?Tzb!@0uwzQun}9K9xokeHs(!D0jf*MG@w{wf zpws*rjkEgqZ1iq9<}6t$-Y(@WqZqy5?8>h>g#T?LK>U|yfA#4^d-=dZfA^wV*R);L zH72G8KPIUsZehYhVU`I74+$Czj&Z?BO9q_%cPB3e0xp_X?+--W-jRRbz~=4Fn-e`o>A$uHU*r%~DQIZ(|vu#Z1qm_2?Wm&p;U z%6->E3!iIfG;`)y=WYWakXjG3|McH`Q>#t{su$LSBKC4ieJ-NM6rbyeyb5G)KOtpU^Px~EUQB%BRTk_-lSLWLVyDt%DeilNc{*nLiB-0UQ z8WSNdZt0$}3y93@O19@84lk4bffhHsK~3Gg@%81TUv?izjep}}m4l+L_yv*+#qUsP z5e(lpy3yK@>8gGmQ~b2CKxN}hGu=iyOcj@7S>lDu?!&gCQT~D1Zlk6GxB*Ws4vmph zNkVr^LYl@)N@?6~Ar8X;Mb0M>EecREX2V=tr0S-+709Kzk~P;zIsKB>*@HJ}I-aoV zEjl+mX>Kq=O{8dHJiPMSR*Q0~Z10D~{a0dbXJvZvx>h+E)#Tv3mTF_L;X)=hV>}f# z(?>*R=QPU39I?8&Df#S;DV_u7Ave>Rj45(Yi3Uvfjv^O*s6wX;p#n`0KNkBuXatTJ z10T~a4LH$AksDEl1{&HrFuh}fvX6#pu>e;?0iT@5RoQ+3pXzuM3M}9_nye-m3iV#m zfUeO@g%Sxkw70)X&bpkcA7yrHG406#!xO%5wn5t<-(+KtZP0rLy>E3>z5LSXlW%IK z9)wdWw?W>0fKieKdaZODv!Vz16{kz8*Vn0oaJ<_+9Jg!qn>1Mb({^?s7v zOP5?CJ1ovCsUWzERXq?6T|Zd3+)4@B^5LFy$(&yPMVXa@vagQBEF7l3mh{ZQ<9*7~wE9jrideVfN4NfqRX zB}4TN+cLTaAGRqj*I(SCEG_0GD13OOb2{yolHUOfT9|D4$|We0Mo^W7t+Z^wVG(*8 z+4JG4iQ0{cT1v?&?B+QaiyP$b$I>~(>g*Na*TSpx?^T`@Pva(yLKkeZ=bA9#H09_% z*HW}%XL7s5q<=bx;m1n@gQsnVC=g9~d$d)P2fwhr1Bz^rDB=?rTdX|*;W?=aVQmk)irM+5X2^fhsg zpeX1`H1t+GQb`7EY+SJkQIBWLslGYepmNvRcw+8Ni$}*LKTl>(<98|2`oKl%JiO0$ zxOEGoM!$4;#XMj#)?od{Mq$|nlI+8l7Do`0$)OCyR!F$8)crKzfoNW)SN~6nOgFOK z7t?;(uXB5+`u{XA&-S-7SOA_6)cAfv=b944hS8jh+Rv;RPg;L1VYl9waOObH=NJ5Z z6|DSwk@C#^>uoK?#>**>1#h%_tPBTyN+O4r@ku%IOz!($JhO9~6K=p|D9L#Zq=veOdz%3J+YKo&|VElopnKyJe0B7oXO{a1rAPlV5=S0&z9 zGxE64;F52=g?Yf|59@K0a|8mP^(zAe$+|y1weyiYCgr1Ley?$sY)I|7Rh$Xz)6ONk zXlUPAo=G{5Y6%*i={H0?;NW^qV!ouC*-g`r>Q-%GvZksi1gD+ig}j-hSV@gRi&GdM zk`@NkJ}gWgge`M@qvouEL2j+^cN@=jSA+AoH1wz`RDd)8 zWq}PXL8i07Dby$<2pe)fCfFL*F8VYCL3!qwT}Q1X~l;0N2=;R^%ujJ?l;f_uN{ zAbXgWwCOtlZg!}b+i(1Jx^r1iKoZtXY3Vu zy_0*-BlpzF7WMTw#5g@?o$00!Z!bU5cIJW>bER@TqLQV^zHW=D3Fo#qL&NR52^-Ey zNxB|wmi>j7484yB&7R|({aoVFBZFhcs%|!_dpV|JQe$Yt9_=U3zL(m1=OkwS;N@>k zkM_1RSIu!P4_+dZlI^l#hjv`!9Qj2e)RXX>;q1^YzM=OvQreklto;*P)UBN|-{>)E z%6LEK1I(T0{=oTWkHY1gUw_-hOc*uuraI37kij_h$qSuO>xDyi+=&BFx9T3 zV%vO=Ql~Oh9zi?%HBf78)7MZ#=L0H&{>U$$yX(bzGIG}@Fa+PmY58)m)@7!85B^*K z)nmwiYZ>@J{`@MV@e}2UyJAW8>}6TRiE_tzbX62-y*}dzEpH2@^hTs?76vTZECthLUk)6YBId_ z?|Sa8_HH&E+X<7~wz~S>SxD5KW6*Cwh?m`VX!`q2M4u$SADr1mNx4ZOwM~<=*%2L=Hymz6N+M<;%n0ytPwr_VR1F z>J(QF!$J2fig@F#n3t>SAH;ImWf!nH0W#Ccz=7^KeCfv_(>6iZiA?%U^MH3*fRQZB zkTgj8N@=afyz;zP{5U%DN$mOUEf_SC4|sX5Yp#x%(3Uoxm=^((&Vmt(jD+?y_B7 zOZ-;bx2E|Z-;ZDF1%9(lE=#nHL@i~fxa88+g;g5%{r$?MgK`=hKO+tQTqJlP1K?&R zzxKtpJ9{D=n9P&gcAonOh06fLK*LWU??2YXPsR4{T1-^!u1c{{lq|o8R|fB`Q4a1{ zC10BY?T}9HxiEW<+xrl=F_Sy}b;7QnJ**ukj(T*!<@V=C+PMnBuPjVkj>jnfY$+6R zd~`qaUGkE;P?zm(ZMKlO&n5ofW<&c1Ry}hsHAl$*8v;2?dAhmW8eNzfV(0lVptC*e z(sDHTs%?h<`1EB#W{#5sX@2wKlN@qOhRcN#`Olfs%5vtq&yMka{+L`MtQ_k`Th?>_;$Gd@~Pyy7v1;2W4gK5?vfvV`3tq(iM!D} zR>1LYYc*QW71hdo_re@pX_UwIs}xhUY)p(WR>ivl-Xl4G<`Y-h>!$!cNBixa65rhW zEzX@V8Ok|!3VoyKMDdJx@~b?-I}ewy>i^s$d+(&nZNWcY@pA*djK3G+cU=B%Z!F5P zt5oU7qY0lY7iM;DHSc7eS3A%BDB@erH~ZY5Toz=KiEO)q{3UdEUuL(}H zv!A9@>;_11rxg}O%x-*vf1ro~q7rWOpT|G!W0OSgN@|Ryb)T2#9YlPg%hCujGeq}D*T&WD`us+ZS9Pe%cFNw609bTVe4Jnpo*?!%4S5e;l4wc znQGlXH00N|WQDJsF?TH9O}p7=i+AyS_f6>SoxH#yjRWcFr^{i!U7fDHZFeKBfZcjw zq3g1MB2(QUSIM8ty@rLXgQuJiSOZ|2#A3AAb~kxGRRP!smD8 zQ+Bn<_x5kuCT}x6@PXWp67zVyw&H;N%EHZ{h4UwCOM5;978TH=8R@s?H;y&umEiaV%0cYqo&xVt}B;cOnL9J@u+TCNQhwf@r>^0{WTDXb)v*NQ36au zSSLzMGwDARbbtkvSU`!zLi{g@Aix4jETF_9F#Hz~7_xv83n=Z76|sO4lX}MjN}ml{ zS*CX^)4TtM=xi*Y!~#kzpu_@7EYmxd=^YR?)ZMBc&N97YR!?{Rl2|@UEFYx=<`4=3 z#s#6WM(@TU>PSwM*elvrdApE&SXtamKdJC=_UK$Xa1y<@T7 z`Bz~BlveBxzbf=EQM*o2OY+OVZvnG<8A89!EB#fk1`?X+g4~IgJTi;SfkkxBbFsJZ z^Fi%Sw7utN?|pOYj>Ypw2O^n9%x?UYi(JYDz1tvn4pS4ln2)Mb{VSxJf?kqT&?gef zb6c+uJo{8kICqb}@$JVjpI!DV|V)-bs;wX847K<#P!~#kzpu_@7ETHu7 z0;L0k5hX#ydju^WPF=yp`UwkMpPJZ4*%`)zJ<>FtVL_oaWT5TG&nZv!xYKjDSLF?S zv>$F+>W!$exAyAyZ-ZhD*6yTe%x?7GVfx>()(ZXp72nbn(qZ=9D7I!35hOQ- z$@NMh9xxBS{F?kBxbOv>y#q=w;+lFbLUg_!ym_ujAFgsXK7ZI)Gj%n4Xv<_o{a&%Z z(LDO1htL~ym6r>(PjW&v=gi-2=8@2cT)t-UxjuITpGy9kQ%&4QJw`FIUn1E=#q`2h z;Wh5Elt%vg*_f7)UcKQ6+tsTMv0VQRb?JS1&w|DSp`uTA17M?%=uI`+L_PEH*K&+< zBCcE-kYc>asJVU;PySpg!SnS?4U|N!*~V=h?;NiA zr#m@@WB7Y7)py%us!PnMuM`+s7_})C>F$YM-Blf(OfX?t z-Pl&FW)P0L`q&xh@n9M!xgzj^_?Is1N^}W|IPPF%G5@ENwS2xh<*)kde`w$T^#B7N zn7SFQL~l^5=IUrA28uj+Z@#x6O;(NX7dzhqZV&E<*wUNWO&hE5bLr7ktTmPTWV@!gP|8=HyIP&XUzy231!v23Yzn||T_teQo_4PQ!I6a4> zsw<%Z`e_tFj3={io3!z!@njU}n1-bBgoFqWnO2ccXiEpB%4@WQ#G$P8xkv>CryCUR ze}6LcOW${XS+!j&W4OHYVdR??Ul!?1S^i07F7pHb z{dKuW>RKYlTS^F>tvy0t-la2u@7)ah3r{%?^+r!G$ugtknYbQHV2<0{izuP=Lagj= zKvmzArJcth6#C0A9{4&N!d%&Z_CIX?F@WIC;y^KzG914Kr6ZQwyw0lb_^k>);KITTx2+m_L``e#5mbf7Ovy z1MmJq>i_dS9Q-e=>n}$h@W5$u5^W<pijWU28FU#|6+-6<}LWkUi+6V@$(rf z1-|6hzSwqWPlqk`){0-PcT~Ih1xIDht=`e%g0N9oUrAt_V4^DcVU@m)N(T;CiXtt} z;4Ir?+k71C)o=FeoV|BMd(ON;<&#?G`gIPp`dG{Vd3U>x{wwvbzg~s^g<^iEk>;tD zO48jnBjulsc)qM`-IEQ9GQ*u#&i6QxfZ!%tK~lMld)`AKy0Uj@)1sw_PK; z{?G#U)WJs2OLzbH#Jj8ZjZo9S(6E!k{%Ul*ZVCI-MdRpS7f8h(*wq#jOALH|+EM8D zc?=iy*(WXq^nQpAuw=(hIgPxB92JyTrpCHjyfMR3Gpaw{#EQgYMNl)CIhDvSt8&pP zQ%?SK`DIw2`+v4JySB&h{|}wI|1qKA^foY{v0S>|`D)r+mDS6{IQJ|$$!3w>mn0FQ zychwhd%Sl%-YGb8B?#`_=qWk9Cbh5)S_ecXpLJ`^EiPWtbaqLN=hT=H4RjkD#2ytCX6zPRDcDor=UqCfkInNfIO zrGHC^fVGOABqYN2{`o(Vzpi>Q$5;PcnD_n%K-phUJ698|i=kH&UW!z9n zRL#s*$B6p|%|m5q8LUJMlcMme%FfFlYWl)yG?%y|^4| znLt-gSEAbf5PSO!8l}SF_3T$Ath+r?!%ZwJ$35BlPvtRc_(9-5t)=}B7VZ|g%bpBV zq^&mUMbtWbiaYNGABh_aTwGZm0u;VV&*C18mv+W2g`;YRn_uL~%`RsaX;5PgDs19t zLcQy!%@ge?PPY=#cH*?855%nc=Zr=Ggn zZPkvkd{yT_1lM)Eu5Tddz-nr|R~GEq%Gg^Z3g~W+on7+X2x`JP0p(;ghr42l3u$$* zn(JfSSPWX0B;{j&%}*$@rn}nOMZbmUH8hP4@ptFQ==rcFLYwVuE~>hr7??yf;t&W8 zJqa#xv*^TOl6tmANpq|6#ab0jV2nse(?Zw*k3?OLLza{J?(!4}6fqvY$u@Q%yrcIy zsmSW?8wb-ulXPz1;Rc6SWEpa4JoYRS5u4IZw0Ls^vpL?R8a3c~(u1p%Xe442>ale? zE*O)-uYug_$bJstdU{JcEDvEZ2KCarU#^_Sr9r9%kHpc`)Rea#3^nGxa;e^D%Xhb2 zb4qnxc(`;tIT-7IW*v19|4`_N5#7K=L4q^Eyx|tXV)WG8w3MLKL~Zv_PIVCxpCSZi zqXg+)Rakdcb5y-cKqA&5C2cEw-8BS(l0++GDyu4yh^=oMZ^LizMSebR0IopK^nmyt z)+X7T@TW%XJ}_e{98%lFb@vb^e)~hwIYNL z)llAhy<9qEL9z*9j-|lTr zeo%ng7iV4R^TsM^NuN?DS}j;#0_v%D*3pg~cU2B)F5r8tlW<}t#0G6Pe2tcxf<fqex956HSs5FBuZS314V-C7H1>?IG&_r=kkU~?&-pZZLcG6Y>Lw6hS`7$PKXJ5A z^6h-W8rCtiO7m^d z17Nn-qu(>~>94N6w)e`Xdpyru4}|MZAhwE-8Vs2xuyPsY3~le+>w%Z6Jpx%qCmFYW+B4bl?=sjBs-jdv{*%O51RAIRTy_0iB)FQr;~7?KBrN47!Z zo8INiB(J+qok_(bXW2BpU5o^65_OMHZIu&}xw!fx(>pkBa8->ijbt_!1}4;u`h;6d zdZl?1^^GQS+CO4b6jVWN34&eGX4@bLNeVWkz8`kCNT&&SE|itmdI_9HO82n44gWf~ zK-c_p z^LV+t)A%;1A1iXUskxWE7w+gnbX~cn_#&&)(orQa88@`xSh-Hs=USwIxrpQ88wrtn zgY&|wx@xU+&CBhVI4y}!VHNS2$#HQh2i0xeDpU}Ai;`a0t#ciE82knnXRmTHOY0K_Xf6QR>NPXU@S5@``T>>Z5 zZ26RY4_D{tN9au`9>Jw8jiske&EC_|5mv2gOHLy$bYKg-LlC{g5jixY*=W)Sq&t+FYa|ah($gByV>-t2L#{a4N}qlP#l@QC?^== zaX4qdtvz$pq^e__VCX&KZ6E|KlPJ%kb8tF=J5*x^onQNBU=k8WRQ%TRrXC^Ka^w(2 zS&_!J>0*@+%A-crj*5ungp4TE%EH4J1jT|4?NL_&>HIFQ$ih41~m0nwC-obfG z?36~;TTA;4hB}i@DZCjr772ye_fQPCK|S%d{xeC8Z(Cd;H+|&pS9K-cu&f=-17Mq0 zl_}#z%YMeDSo(Of8p!Yav}wAZWxo_TesI|~tSA~js?$4e2|2O4(xlcAsce5!!1}2( zG$gG3JmH!=TAGf3TuPK(#YJv|1U@d_t#U=*GP+WxWK`2qs2@|*|` zo2H6!-(gQUHi%Z?fWDI6PVPN0#kORLcl_cc-bHft&-R?whh`kLwhhf zYEz0Yf|q8*i{NbRF|*M617mEjKtVwtzw{j2s~!*=L{MsfZrA<}_ax;6UG3=W;M2

T|bN90erW*tRREoF_NpAFw^(X$0|KTwslN?EAMsTzarN1BtE?8_pg5Yvfp?w(R`hmb2VRJFS5v;zE$THqw+@b2&IMEH3{GJPR(@C4Q;D55jgTgK zOiYL3D|{(+YlGxXGqpC_g1?bCxAVo$ zX^2!ByD#3e-IfQYPloV~dS=opy=D>M=arLII2t)WvCY37m~Q5w7ixOy>|R+|c5|B$ z0-Ja)6afn$gcG7iD-$vSM9`NX`O84R$?r9qWaJI=`pM&k8{DUK0}{x_()d%pCHk7z z`r}hY>~mA&l6p_ycKuu0j4SGz-SRdF+O&Kvyx`6ttig%E#W~p5l)~WSa^PL~Tl?t% zlf>jDB&~>(Hrt-&*=?{`|3EFSm%{`uUmrei-}!(`ym9W-rssr9|E=Yb;jW1!GW4=q zd9~h{Y7rSPTN|$YkkT%->JRNnFu~24PsBS0$0a*Js)?JH><@DmRE+e}3M{(!ozN9M zbU%llkt0v3I9o0FtpkdKGcP~~GeJA>pn1W`#UO;}H4HiF81G1fPCwwg$pvZuiR@Cxp)E#D_r~O+;bN;LeJ-h(0kW-<>R`aftGM@zICOS?&rs-MYzF6 zZ2{#)WZHy@2F*I6p!Fj}PL(D4g1vDYhDKO24s!hi2NHb|8D1>7I~zplt~8|+IuSvz2P5ZlHcosUc9w=tQ(aY*j0nH%da|ggJHv4O_&dsjZV9gATeVI zFWBXl1mRq_GU!CrWN7_6|=mL<`~zQbU%;%>v(sz7QBZy)utzq(gR9Qiuz} zk7FE@%5V{e}Pl}KXnZh|?u$E>Tr8B{T18cSk`mik>A6zW3Y$1x#za&%}} z0P8ZwOYzU>5Cl*UsBfB`GvE;#YBA{J8c3Qz_87bl$k~x$psgL%a%}rY-@spPmm}+c zds*)R2Zo>OVOJ-$IjXz(`ktYOIN{l~?zt@MXTO2$W&jTd*p#v6Fu&CLIpp{F%ozQH zxw|S_w`k?GOij+Ps4Ia|BiXutz+rRcL>G)k-4 z>1*0(m1J5m`}`+t_Wm*V*dw&2y>~%dKWwDwJWFHMW9qUUWf=h$I43`;*?;yah+POd zS)20Q&1v3=TlaaUABu~}EujpLzCs%10qM&}5Y%}8^~N|nP4y0|J$%a7_s{SRo;dL9 z6cBrw^)g@xG}4n8=W3IIby*FHg(&%veJ@2WKTid7YtW4m8=&DNohw~RmM30(mI zWu|;P{hgX=Ek7Y!V266{$_d$tQ2xh#?e&&8lHBtkVL=E4HT8b?A*ISC#B!yTmA~pp zY(+T?1CVm_<<2q0Yx*b8GbV2x9i3a=G+(uM6|$M{tB(o&d(N<1(B&uJ`n6#FlOm*I zD@+G8ZbJzEn3WmjANc-K)!C3Q4vcZrwjef&ZvRS$=rJ3aOIeOt{28*po%38+awGaP zDf`N%@`8^%?@2~ZOYalMU)VDfo}=qF?lxi(O~2QbVcXaKuRZ_yx(E32{Ex@99#HLd zD5wU=Zx&coU5s{8UfpVD3W`DY*j3SWNuXhi5)!2K_fHGg%?(Vu7CrG&~^@0pX0?2?@60I}}^zRYJoX(D3AT_t!qi_*R9^h<81 z!d_j>%E-+j7AVk_VvP=bNE+;1C)TQcyefR-q+UFIN43F98Hw-eO3Pj=VsZ$I?&Lr& zTCCw4h!Mg-U~k@%i%U%D%C#dNU|XuN_E8xN+Vi~(;cA7O=CZsqH>>M)Ja;1Kn z7gnkQdG^I{7$kD$Qdvd8$VNm7ko!ae<`Pa`bh4ff9hp1`!z&X;%(_EQl%j5J!@vU! z%x)fMPJkfSSV%EAuu-_D> zt$Jiyn|@p#a`&9#TsCaE%0`q--DweWAY1g=1(Ej}GW_?)xua?(l{H6P8LU%vdSpQk zv%ax0QB%qVp{E>>Bp=TwR2&99 zGru{!s{%+8F|^FZY?!8mief6kHXIs9ECqK&k+>wz&)(^b=U!|( zn&xO<*K=r@K~F)^tNNrwisG~?OI9xB<)Jua?@*eaZAO^JP64mFr;0jd*)6TC$D&a|EP83lBs$j9v;rq?M4iJ@xnNxdnT4xKMM(|57H6Wgw z4OOsnrh8pP0rD}>xESg&Z}-M?U)oUM<*(u&3zOTyiAMumts`c5JI~V?30uY$zs3aQ zop^JV9avmv*5Z}bJU}(AL+6MA`nHR`g?{2xYV0ubjXASKF#w=59WK8ffj+<9XI{^MydRjIW!5tQjlFhQ zIrSNV?jZ2?kZkzp+4B}`o#1#rSZ6W(U{)Q(yx-}Yto%=P=xO`^phh-GNQ+fwo4V&W zL^_)>$X%-|iM;U^Nlc$>vS<5<|P;6>zK5q8a z@Ov}bG$Id~p;0Th^_;YclER$HhR`q1jn(!(9%v5hi@w-=smzq+`noU*{!uw5PHR7V zb?yl(pStsX-`QJlf=I?@Ar=p?b`^}b_T^<&R$iw~J1;x%;*x_BXmM-!YV|O!XL%YT zZ(f^KLRm!aLi>gZHV@}$>{X@+Fdk+Fh6N0)*=ya(vvl#@RY@r25Qv1u3oMnlsY5>F-k?#&slm<5JSbJXR}R5w2jJw_7esS1u5I9B2P#Z9B5)@Y{0G zNfWVc06dj~Xu8`NNj1uL@4`2O zYV*kp>%b>pwpD4S-LZ+TbBNA7LOrxy%1ua`=McufQ-kNV4*b`lXkDvCiGYD!N?;5F z=}d=}f4SgSP%ZlwJQBu{zSsUqjy}To;(#DVG>Vh6?piqp7p90O^}*djZwY`ID{7~k zU|Nb(uAF3%2#u1I5yo`ZlvVVUf7}!LUmJW>+G)Xd^ACsA-f8ht7XWTZi3=j<}|}QXvLS5xGEPis}u1{>v!Y&=bHmTifivkUax9iYVvZ-gF5C>Y;=;IcH0>m#lA@giI-Ym}lx_?<<4 zwU=+w!`})4O_E6c^XsGLD=>0IT4E=&p1!kbzXo@OgDg7)8ZPpUpEw{wo=t!i53zaI zmEjFb6oCZ*Kx6neN#|Z0gCIW3V9j15w2U@rN=m=xi4f@qm|d0ny>XH07amC+ZSg&d z6RVf14H#&XWu^!ua7jNSOx|D2Q&gUQwnAo@wO!1zH)~o*#Fv+=p{^+#31^VaV5(i3R zQ=gTm12-C-@AZ!qC*{#z=I8Bw3d*GRgy(hiYpy@C?b}re3MDsIZtH3uGyYLOGf?VK z=l2P6ChC1XOcTX4cg3aNcg2WwP1>AOjeZUD)4gxnGIcQ#9~Jy!$yMD;e~cQQ-#t&S zpym0?vLkTmKcR|grgcZRUGZWM167v)b3Q%p*Lbw zX3%szJt9xAVTX^GErlkWm~9zu0CP^YOJ)!T>ztH5GuFAJ&{q#Dj}P`ynykM(H%k}V zwGC4i9fmm9atjLtjs~#HE*CU%yJIgxOvR25r}~W4?zCKy4cDRxlAJumu8Q_zZ^3$0 zRYa=wyA9;%S9n(Bz%WZ3_qymE7B*axf$c0%0GfK+{SHrF8^UVUMKxWzR2=gbmue{p z+SFmj;7`!XDBF%;{qHW3aR2mk1U!2w4*c~}&zF+<4YMwKS&1x$=(7=yS&DfS`M_CP9vK(703z%%#e_KezP`s>{a#ZZfRtiJf{tQ zr-6G&S;?~$dG~_LQXp_wM+4}*-!)oCQ_s7td)1a#q#xM(c2zcU-A74i7^l%kGUl?Y zaxiDdY_XTtuBy{@%j4qPk1y&|AKZ3QPnYSI2 z{)mOVlo6#{Kg}lBnZEV%7AmVJ%1U|WYb)acj5YkpD}$rD`}Bms1U=+)pA!|^ks>*x z2w$oWaq8u^{n2nabR_%sl)oroBfTojGnqBI`0}*x!0HeBmyx8?mO|2`Z@A6qxJ}eB zP|KpBXU49AM)B^ogJ9yaLJe5k&V~Jt5(~{mp9eOrC-Bkzm+gAiBzQKfIruXb>uld& zIaG%jt`)j+43Ts?MF7y(F8zA1qFt0-elVrpTW}Ot=Pf49EGLsW*lit>$vuije=%yr z?R(O*?h{SWi$(JJ8DG$@id=5Rr1vCMmHL*|IEZ3?D$88W3I%~&yRb`3yDHgjx~S48 z%EuTIx|XsiEmw(w>^z_H^hD7YX1_xuL;5Q!1CG5`^&d(8PhC=KkAJUCP@ zi;hi@A(}92R1>O6PuIy;xy`#OSCZX>-mk-kCQW(K^K!7yT8>g=U+N$Pftp&YE&gh_ zDk^;ePis}TZXx?6Z%62(MiyKTO_GvYfB}J4JWmp_s9|O9^uInXwP(K>p|$~vIxf-x z_y+zqidr$IKED8mV$hqT;SXPZyJ8eR;1I|`Ml?@K$>e$fLIV!mLh2ZV4;C6&3T(~o zRlN~t`L|mVlPATGwG;Q&MDIfQvA|^C+S8t466knAPazTQmAC z{aF_{B~vy;s%vwQkSfH>%CA+NZ$W?@0Vj4K!)?jc#qsxL`0fqkYzSg&oMI0mnTsKi+2mI1-Rs>v;6qxVOu~j7uEjw)^|TM+(R` zB?#Nft@~Jsk1TN8IJ#pNsxUlJQuKBI_~Lx(c8#Q7=y&mth;HLn#C4u&YaqE+kqIx) zn5R8ii7NW}N{zX@HMZSE$;DRb;Kg5NcU4rCAr9E=w94MEn2UZN~)TfTCAEUnKwbbsnmS)Z1gyR}h^_Zo`0z*`%o|{ zsLYfKQH0X4?%Ipi8TW^h@;(d<#o8hhT_zY~JXQw25)mYK7{wZZr-i9T?~*l*cZ}5T z4z>9vF)?muQfsYu@DYqMhi&juJ>+=9LE0CY$#z9&tmc{LEeA5j?4FweP~^K-=rK!W zfN_x%2n8QU7ePk6xfYwI(rdbpA#F<&Ae&1_+2skkvrqUuftYgTuVX_W9;q!8Td}q# zAuT*NLs4OL?qUElx!M_{q%+yGL%S*u^uJxPzdFcQ7BA^6&3SntSlt)sEK^Yw6Kbfy z15YOr6VaB4;hx`Ong zrazkta5zDIHaU}<8)f^eZ(#cEI#x9MDtpD#T{56MRFUtRm#*%%QI>a?nec$4w`5(!k9(b;6ihoa=W`NZFCXP1jXAU=Iap|ra2_?n?CVPE{B;7u?`dCrYhA!kxP`Xyzu7p72A?J;f_qLbKXbFKS>3TWMq;C--_@&8rz?uh;R5XFQM9H zHk?$;5Nn+psZAm<@qY^T7Fye-xj1Ov)nf;a^QtDFe|ivdl5H=o!!x-4wfg$eRv7oO zdV@yd5Hst;(G3e5-7k2MuSp`;4}YT8YMm|C8J$Us9TnPV4z7^O|2dg3$2!qGMc8UI zZO|AuYpY;;tmRTGXc*61w4z@I=G^VtW;Vf{io=Or%#c)y4>O>|imw4vR>p?*$gcm@ z^bb53>Pt_BMQuX(3(GDcW`omy#0918*Yt{;GDjA5iI!BYeNQ7qN6B%iVBTkXMbqWa zzOV{tGG%wd{x+p{v!QU%JkFJbqB9(Owp!x|)YnS(I$Y8D`AoI)oeP?gczO%m zNBjqMeZ>qYFdb7P~5k#H}Jnp?1#p0qXMLr1Qz_n?D$g>g>7Q zwGN+RqxZL2TfC0irz!KN+Lto)nopdVcpTs_~w9GcRQ@RowBl;DQNHq z+5b!U*m^zG-(|A#+j2F{ne?K~Vo9B4TEULhT>;BEH?w4u_^Dp~GFJ2B*mXDz{IZtHZM$m!dNqQkv8`koK;MmT+4 zU_!^LElENxc0lONaQI$)@zUaET~ndQt3{tA_Xq7@3eXor|Dc56Bh&2Z?a|Xus$Ldz z?T69SI?9YS^P^?b{!cpvS!0|y;JAXPEc`QdIR9vhU5OgE=v!Dz#SF2~gfhHo!JnCQ znW2N%J5^P;=tooaG{uMz(mwe1;{wi<%IHT;w`LBPLEN*#CE1*287&;Um|5U;1_}b6#*}>7nR1rOTxXiSSgoVRE9AzVLZ~YuqR?%tehI-D> z-gq2G8H!GLE!tQbp~9coC~+*S)u<}SoSUChFX+~d3t{IkP}SOGYNHCDk0UOOG*!e$(MOA*3rD$ z{l?32>OJ3bF&%4krfp4Y@^p~zy`FF7z_32~NN!GBh+JIP)B=UVfjjRHI|Z|%jPnMq z86A_S-NH*I7O=>jGp~ta;Pw!9seK0w7J`@1DH*VIuc@ZUOj^gTN)k0ZY8x@@e6}*{ zlSp*LcjYnn{Eh8lLGthRNez|A*j<$ai0gUk9nL#5&Sdpo_q->M`$z|jL~Q28!XC2; zbC1iH?ee@5cEo>7@|jA;K1aG)!q%0Hai1Dc0O-6pWdWZQQShPd&nDEVhwWvROy*44 zmyUIYnCx-Y;?@sK$}?s|aG10PWdgv9_Pmx{7(@2B@n=DmbwOdh=*NXOk^$QiYK0`X zNHxIdHm2cK=aS~3e_jP-)mct}54*JCMm%J~_=XYFMY#Sq2kuOH5+@jMKOOTu7*7wk zY)fFadUoZIBB)NKeTvp!|Ru)s!bwyui3}WaCl%(P%F{aDY(ILmUI| zs(i#{AO<2uvjwI}r+f7AQYJ5`Xs72H+w*hsWQG!O)Gjf5mNL#g5Zr3{^jg=O@4Om! zlx!(xr-T6aSVgGR)Vb820B590qso!c8h?Iqt% z*32;uvMMX3&*v`;)AqRq#~ZF_j=2Ii7GYih12KV-UHFessife|Uju@2+bON%KsL_; zvS7pY7PpMmA-Jz*3-QkgV$ja~Z~Ky4K9SE3?vKl_@(8O-)F-zF%&Z}8c+45xDh*>4=aWX=?Jt7j!uhu$Qr!KUgQS)7;RxBVu0)QR>T2~^EAy&Og zXJyX$(0K{E3|nS=J%%TGjEF27nID6;`LDIF7wZ1>;_dHAl>6aF`CQ3X_rV8~i$`@W z{pD+UekrrHk%^te83cf1r$Z&WpMiMZO5Mm2>)hBvG*O(t&^Nv`q|y@`Y_{fqnOWN^ z-=DAdB7E)v>K#Py2 zeo!(Vx`u{@S)GG@sbo~3vk$~U*#8CUnMD6E9;c4bfPnyB%YN%xj@W+n zmY(gAb6|y2AXjirnU>xf?jzRa4l?a}+L7+4hAeP4^tlcvh#YT}sVXam#vCXZ z)Q82KerKyCwGBM#&^&sCgkFU@c%_i)K|X6UF6*ORHUORLf9>OgD!2ybwE%Oj5dOG7 z{#0lbLLB>e8q%h*Aa=3~zrCxXS^LLw&-)sE&Czpo4BFm?l~p(uu9Zcilb#giaq&u+}!sli3;Jtyns68*^ulz+frBr1JiyX{CMYzRk2^orNYupEh>Fh zEJVP0QsUzC{SwbnwY@CI&}WI(p?NJcL#OqAy~IM2Wa3tTHZ+hPjuGn4hGy$$H&C%} zxl-s}M0^%Mu|&tDCCu)ZlRS_{sFYdkV{#yyDNCv-5bz75Jgre<9fcA}WF>KWgZ4{U zqF?ulE9G+AsI+y`Jb1pXn>Dd+!0;5T_MYd-3pRzTw-DE4WsweJ|M~cGL>R7Jg1Frj zAWYCo5)Dv4pozIeQiy|2<4R}S6fwj6b(PkQ=7|$I(`~*{!hlvBCwN&A{6l+>>peIp z)TDbNBj3zmLCCdu|2oe;QnXgw?J-jKC@`P5zm%$|K8#7TiH#GG@7l+&IX z@JYx2n=a#zQw>d(qG=ag9nN6zuLh(76sjG31=|xIDBXzi=eikyn|UK|#~8Pj)$_7S zvBThE2h>BFT7v*MoPBbp;7PQOYj8y8GCHSYx;i^Y$y0_PZI?P!mtA01$!lw)*a&-= zh(uP)`h(snk8r30vo#s5p&_ry59Y=8v(rmndOEA_(DOa60G(A#jC4Yw0~HWd6Q%CI8<|G8SvY9MDE3fHq#gth=cPMW>%2F<~q5zoy%3-~8^v&}+Y zT?Qkqi+#6Q$$d84X-lalEk6jfU-d0h({HjLN()Mn15xP`a{A~J%nKD>o|f_4#Zy_H zOVzu5&A}u050JsZei%<^5N|$VOdL~pahtt1HeNy`mgr1lts&M<_e^sriz<)GrnHHX zs92k*qH|#vA8z^$CwhQ2ZnpHtbBr=1<>?>}Wk-9GUFQYTS-n|1H|ZccW?-^?=+xl; zguu^S8Zfs?EwwZyqx6F$XTg1c>$}aL`?RO2EmO&jz2H;j2a`_BOw9Xp)x9{xl9EJ_ zo`a72yt_37eV`NH^<@sB1gdX>Pxpg3D0T>mhRst(fysp<=k!4LrxtfrW@|EAe%$L8 z<$GabVe_15HKD*ZrDwmatI3t2ag$K%jrEknyyYcDK-_Q{8RM!Bhgk}h_)bawVbp0UufqiMZ|z(L<{UpApldHL>K+*LtmU+=Qcu8G#ZQ*Nq+au$3O zHP18oNd541%zsTr1A~@MPR4W=NW_!Gogbe2pC3XpM4A?Cjbul@=_4f8{-N(;@fx{i zzF^TxUi3k^P6(xC-)T&W``zN0{rU0usC_*$eqs7&1pt`xX|bw&yS#O4giamA?8%6&=6Ky<4{idrF_{dlm3zz>I_L3FQDpV;I~eHuq0!aQc{D zcgBGv>tY6U#cK3>E9ms+RzR?mErqAg77j-9wJ82?uM*BLBCf>1kz~K~NH&9LUq<;b zKh~5bfBd}gJ7~=QicjM-=G4YlBHNyGK0r`iKG!?r@$Km#r8?g_^Acp8e90&du=G?P zj#_U_zbE1OfMH>E35VaKsA6)`oG2+`#uJt{Af=2TR>N1V3W+_{7k7eL4!bIjJyrcy z8N}S?(cJP(?VRbj(nXxjKl`9D6zTp)tXkM_h$K8ZifmJsOE*N3AsnMj=vEOhX2QY` zn#jnc2A>KsbiQQDMV*-Db=)+zGk^+#9lwL#938tJZzIhwJuO)bR)(sql3q-sIn!$$ zY28QR+FIG97|LcXE^k^QcvTr2bqP3mjAcRw@xbKGmhpvGAw9?17}n)!ZJH8%gj71l zF<=A!uc+$3wz5aVZS9<+h3Zc=f%~4lgoXPw4t|3BoM`H6zhbnjQbSyat1B~$6O*7x zX{LHK4SuoKCJM=872P&kMjoFLy4#o8z;wpB(*%Op(F9VSw63(1p-aWxk(>W!Y=-96 zU1vS$a5_^UbWH$wLxli`&@^{DoLqbej5Z`05_0}aT7MJS0B6}HsQz}Pw#1X%|6p{&bG_-4GyJs8A4(u`;rAIbv_Zg_eSCO=Mrq| zP^EoL&hsFWyq`qv=x3>Mz z;R)>>87NIL-osdFxgR;JIK_out8ZT(CMJS=yNxUG=$J`w_iv{|mG?UG#%(Ckk>UJg zZ17Q>c8P3cD0wl4GAZWIpwaR$aL4ZsW7M<1C%Kjgs) zZF-)tGw154bmG0*=$eGkjHT;;a||F7+4ybv`Vp`R@7m0GoDVhcu5M@>$jOEx)JNub znaND99wiT^=RaDZ3LlP;=J*B6Fr6(BmzulS+5u^O@dqA@yLxOdT-JmL=#0~fFHkM& z=2C!S5P^eY<~gZutG8xqhqG-X9GnbJa4V~c59@_i{IX~Cr5IwZ&Z-DHFLqcjak`c@ z$a~=zknz>d0d3Uj;CX?y+@C=`K@G*Uv_(~g_t$9fTGX<7%|qiyVCnBik4;sIyQ*{% zO|){1Cm9FgX7Xzv9r~dqjO2W>V*)+NUJ3ynwbuw_wkc_A-GS|kl&#EL7o)l@MVGY9 znSF<@68lc-rLN{jk2vn6_mluODC*n#`e#^iN(s;5rph-THr%fb2_YW{r?{ZOR}qh{ zENUe`2Khz@QpsMN@H9n5PW0`B;2#v5E<8nAsr%&Ox@KU41+P+IP(Dy5M(`0(`52!@ z#m0L zwEs0PZboMUp-63;EZ9{UwSE+Lr(@>Qe|EGE0N2wX^_-s2(|G z#`2ijkD|?UgQDC{E2oDYsKY-M<~u=Dy+vTp{u}~fWBB)l+5Fzi(YKb;iuO=J7d6jm zJ*xubv_adqM~qQ~9Lh*?&0^`dZRm|K-9N^fo*vz{dJz2ORNkKZ1xpO)`=JnLCZJ=e>+V4D>oM%AJD4zbJ7I7X(Ls8p0Q z{{8A^=pTv5xO?MO5r;?RYI--Dw3Nz!>g%#%Xa^K-l@1hgQcRlHmxB1F?=KA-lw4hJ z^LaUk5sA+XvULB*r7UtPw{>V4@}fh*@)?(0v`rT--d76@R>-<4(mu3 z&{gZiqMQ_7q@O#oR?|oBvv(cRJaiGY#e7aX**!~*9k9P^De|>Oh7hPJDJaUa(k?#6 zNT^0F&aI>n>{8s<9sBL!p`eQyr5%$mY|bwd7@8;8$TkodaxzCk!(f(rbMG1(hGLd( z{vK93Oy=}e`qPt=OJ@%wcM|h_V|}mHB12LPO_d58oKg0jmzlG#-c8s&?ZM6~lW(N} z)F`nPU3dl>Ln!^u0(Xd=m0m8p%y!*+Y*5fVPne!(XE7@lp>ajO(_R*wtq~Iwz7%hI zhSs|Iy25=7v_t3&M&Yi?Lx6ob?T**Gv1*;WX3mXJb_>$ZDBl)|HjP^(Gi5afhRiz%^_ghC2q|9fOWzRA2sSm zAHK}akRB2eif=9_FWqvT)`|Mi9}B7IeicLVZyhc}a;kMVDu=~-V32yNywLG<$Ne7n z?TVf_^1pgn{uc`)5JQziwFcXF{2RSHZY`&nQtb?#9<4L+sImw;q;(1E3B3>`sU)@E z?Heq_J_y$s?p{=Y6U@oZ44s)YG$ZpSRxVk%C7!f6dI|$S$}=_qs`qT9R?;J<*XvrQ zinEQBvjj%gvExdqbIVVjd9A-*;{`oxOAC9k)@3OhyS}U9nnRzvLR&&(<4ST%VjvkS zvgYyUeQdf>bZG)~u_w28anSq#AkZdlU%uIF zbiWG%)QG}|r-`^}y?JGtRGguyczje<@AeRq^Jd&^JeI7-kMgT^@j{8~tt3=J(bwqm zCoioNN@tGUH{urd)a;lTwhA&BnDKcb!U_R(K{FD)fE;Y^LJF~9D+M+Ny>$+im+C1v zHmp|>&xx~R{MZ3SB&dSH2<5flo-q4*Wk$Jn**c<@A69aO88#Fd&VHq1(V*93K8j~h zM|TP~%LqA2a+boESt_{+Lx?hDOa>oQ1C(K=HdyDS3`?H5E6-!LjCs|)vph2^B|Jl3 z@MZ9u>FqOr_cr^f!)`q+2ZeR>tn8;0i=kV4xfmBkYJ=s1*un7ixKIFk*Zgti?ei~O z(5(+-8@_pHMGh4mEMbZvCLI9F-pva*5%fH=q(ER_F4-5<3+A7zO;30v)!i|=a0NwZ zPD-kUuPP_1TsN17bJ$6y(`|_^n4bCl z3(4Zj@oF0}d1k;tQ(tZZOx+h~{d}kl#biq)_)uzC?x*$Z9aChp?ia~Q^})?nLHwL3 z9DrgZ3LDvkF=bw@r+Nd_lnZ#hpFVYv^f|!VE2|4r=t~I6^uSA7CoFI&a$gidtUEB> zs=&@qc08%|0)78&FUw)fTN1#GT~oa~gUb&aE%rDwt-tCt@2PfI;3=4tM#1!I93zTR z0h<2As7Y|wi^mKCJCBQ!St`{Jpe72`{iwL3d8jGH7}D__uAo|ix9Ypk*n{Sng}zu` zjryresR|gbgDD-Xk_6^lnl~H@;=Z>zlx$3YmrM=Dx4<H1&J{>$c`AHJDx4l%U(%b~w*7DnOj?E!GF|Lx%a?0iGWg}VGd KcXeQQ@P7fvOa-O@ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-9-Quarterly Growth Trend.jpg b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/images/image-9-Quarterly Growth Trend.jpg new file mode 100644 index 0000000000000000000000000000000000000000..d39a9ccbc959509ff22241f8debfee1d35c48d38 GIT binary patch literal 24591 zcmeIa2Ut^Ex<9;WQltyg2?7dIl_E$BsDQ{pn)Ifk(gdV;LXj#(KtMqVy%&|rLj5;%S?)~ofeE&K3KjGP|?3KM&vew&w>wP!&D|QB;yr-a~0N~&N z01o&EV8?;m0MWT~gy#r|2nh*^iHS%^X~;;=pC@IYzCc04!U$nyVPs}z;}qaxW9Q>w zX66>V%6IL$h=>S;>!!50kd%P1h|sTt;1ClNlb$D~CnKX5y2N}*=)e8KHUd;cIB|GA zcsT3;E)@ICyxt_;>^a`1s(} zUf|~dJ{1A=rR#F%E~q~uWVfRc@{LL*;<)wkGp)t|l2iDxy&o~jMLK#0#>-c@xUcev zh~5wrzbSG1j=X~6U8Q@P545y(boKO2pFDkLW^Q5W;OOM+;_Bw^|1uyjC^#fEI_6Dm zT>RU232EsWnOWI6xp|*TO3TVCDyyoSnp;}i+B-VC28V`6MiF1Xjm^%@FDx!CudJ@^ z?C$L!93G*LPkx;j4uJRDw7~zrP3)(6QGxTq#mC3PC;W9@IJhq0fk%Z;aOwIvYB_bn zM|Kz3g?x!Jl|Cid5PjRB;1 zI3V%xr~oK%bi$qHL-N1r;Pu^swyFp>pM-oYFPC`P+0uHMB*#nGKL(wjfh-QzX=NnU zPa9`?vDLj8KN-gC)R=8nSrxlst!@Z^RSU~G?Rosq>}?-UT{Wd@NwtdI~ELF?A>{!{k1L+KS{;~CoB+R@(=-0 z7$^-XGwAqqnJOlLBL~2}FFEk+#^CUFPLb;8DT%Tj1J&oFeNSP8LIyQvVe@P=E^F6Y z9T|&qEz&JYOs2Xx{Ajybm-{I+%qwK8%evp_T$;fGjD2&*+2*BYFZP!CzsZgx z5Xrp7srk4rARS*fg>t9h{yB~nA#)(DHl_^=P;J?4^#;v8i^`ZRSI+U?qZ;O_$cLS* z!ihyE4YUw~bHk#%`5DSDx=QXD;#JxUYjthfyNztPAJBGr%I-XQ@JSipGgV(ep>=uzzCk)H0ZB8R^bSIac@nzNXMb3n%Z(Ic_;p*vnW7;0!>by1>&K#fech5gsKo64qPHuRi_QmnkA+Xn{H!uMh{YThVX zGurZbN#bl-h!9WI$h%Pl9Z*oA!Jl98aeB()gqalQ?qzNw7P^Q6kL2lC_Z51;p*=S> zpG@+VJdFmux;zc=^Pz3iVCs@MgjB`I6vD8&UMEMTrXk(qf6$n3m{XCZE9JKZy!M)MjyV^X@ZBN*(m8kh6L{Bg2k zhx&WZMuIc*fv?n-C$#nF*Mx2tOd&0Y}v!d2f zLoqrunij%+Qv6P~E;}B&Bi2vUvlXH{Ilnbo&zQ-)+|xQRSjPgP-n=Tps%Ebg%B*{? zh_}1QET=%u-2IKB$F|j2NswAO$}TZz*bi*$}%a@X4C zGA8ZqDK$Hb>Z57xKP2N&{W$2@@m%CUzT%}d&V752XfGAxVu~HsKrBG&l*FYrP`qVy zh4Wr?)Sf8=A%IIHEvZwXFn2eqT>%K~Das1jX90=N&7Bqs{8^4hlk+Plh9}9{7Ehuk zC)!#koad9$%iZ7WyhYPufv89gEKqEF_FkvN7z?xq^qtw;E?BjBlVE|4_FNdE0@KSg z`Xy+GqY4Gj)LS#AcB!cg9VmRSL=$aB@G-pBp|8xXBB1`tTEe-A>ctfRbBX?PZ?#9OnGI_)fB+t>ypMchW7>LK@yDL#cx?er2m7niK{v0upSq(gA# z!=c;T;I{cpSA?M+Dn9C0Y?|n%sB|D{H$sK-J!FI$QP=jEWCFzkKoPY zgAq+-Teq`3?;|7p@g-COGLa=gix>E;0;Ru(f4jLpBr~#lLu_;`_KjGaeL?x2z=G3? zyjw&pYn=Bv!+1(8koUUdzVNb*Bn(m$`VDpKnGB(cup~iCvN5ct{h;Sc+@gG4#9=6R z4&OiG#077TmFmE{+1pRE433Xj-DP^#YDt&l*d8=cIC`8>G+onOQ*)!^<@l(w_7#7@ zJ|lF^Jll(p*UF&>?KEpkGr9g^-ThoDXC%oiKe{nc(5Pz0e;Qiopfd(**`{f!c21>< z8mAW9rfLbA{MdOydd53$o|VDH^nTmllZ)_WsqH6@5F1X(N!c#b z%fC5}pJM{|O63Li=O%JhzMTg}pF(v;=gZ^7u>gSL5hSrlellyK`5sR=!kJl$dYlQT%ta_bnT*Ia3`KCAqW1>YcDUW`m|Ry6F0RA+{oJe=OHZXHwaBWyi)yG z3cYD8W-sg;tt5UzaxkKnM?(r+yK0_VYKdW zHmm+nSsSgCzWeq5nk+*fy}FSKjiEM=rMpqw$48fA&#L=uxuz5_4-aJymJtvFN4+cj zwmDNxK`8Rk_IY39ck&`*Ul~;$UL&WQ!_POn@VZcUJ2*Eu8~Bm;9mxU|%SHm}PtVe` zdn^v$ZNAJe?Z&}r-1d`eU)0V?n~7Ewn}y-6vR-h@__(*2V8dt?NZRO5isNf=bfbUD zK-t=kv6D1UeISUGLS19v-b`$yc+AXbAVPldI+ro8fTH+EWU=pW z@w)0*vTgA6(2PH)R(_ZkdX@BNrl!EHJ;Km&1;Umi(y@TI4rm&R<)DsdybJd#TZOo; zluNJw^l|!6=fwnR`B4m~3LXfwfGQ~#=wngDpo_5pt>RDT#^>dkgQe5|D+5N{VI4?5 zzyjxEnxT8`>R6zG7Ynq4?T;Ngeglf^6y|;ISRkNt7JfXRg9WBzvA`VI{@9WK**rap zu+*#$*0g}^L+=d3w!5&v&VwIMwPs4yJN(Clry6GMUCl7AYtW-b7%HzW9j)5$)T2MG zQWmp#=8r;8HHFNMUVc**ya7F$Nx8xhKUpx>n6A82?7KjeYh1JkuUg_uL*WFFDFn5L{ zYc@b{iWqu^1t7B^g-@X;xT;tH){fE%8_>7ska1(}0m)R~hw+I4iLryBfi-QHfGfj^ z1&+|L0M{NGd$#4Mjp_e%AVqfaf4OlW2IId7@IQJ_mUf3C@|UpTGM!(rz-uk|*$dS( ze675eiSI@0_L1sf{hi{XzDel4W9rhZrmDB4*w-pWe0QeUEE3tgRW~y>y#koGbNL?5T8UPXPeIHh6h}J^8^7!&JF9L)lbU)Lm zlEUs{pu$$=LYR*H%rs?drYC>daGBoS2m(jLvt^uME!ONZG6A2AlO*kuITo(w9YukB z_qgs$V~M_(BR>EMW!Y;N$1>lz#Mv|OFES`{>%2Jl3?L#`7`3oK%k+iLqIHt)OxbY_ zKBV2C=;49r#!3i{6|*HR&7%jTfb><-M_FIjxYMHKe&&pyr1*rV-U*cr>MIJ+J9QR3 zyIA&OqBpCZNuEYK>mP9z;i$eD+OGf$toy-M*bB9cgfVj3u!GOgeJbtz_iNu{DwJnK z`^8bb>m`3-{=%#89<0!)%bPy|{{cTAw8Ne^LH**J0YhkGfqup+_ca-2EYK-C2wl&I zt&k~gh;Mma1%DXf-M=R@|9|x91fZBO%4B_KPE}aoTmB?(6PgTMyTD~k^#m5!7*YQg z#XqnrBeU}mS>*R5_}SsY+s)6+coDj1yuK)xd{>j(5!ME}BLig~(GJx&>Dw~)I*hY( zi`5Wl!A#c#@e2=p7j=oRmfx!6Dpt*W1bT5f$VfB_)SkW+7?CT54=tva-|AUaKR3Au z*Yijb*>mCavn~wJU}iRjW_mp6%y1oV)=-E(OKV!8rRL_5Bh2YoBPymUK^`WGwJP=} zUzBjIc75s5>+ECKXFRekp%C*lWi{%nZfe9BUo2_s$C#MTYyrQ31F6yZ*{36gF>)Sr zojSLC-s!Vv5a7kMZKj-JVtvVKrzu7Ay|`3#%8*lUEAg=a33-8+5x5{X3svb|{f^r8 z(cTx4$NbNVPgmuRpK#<~IQx78Z>kfNMNY;Cm0lj)BqUHxyb2KsQ{=Ol%*@dfOI^~x zpFanuGBUKlkjIVxGFUoWF8UO5ZJFV+65k8eOeYg^w5eBd*;af3Pq;z{x;Kx`wU=CW zp6$dPK{spd<;}RxSf5zXmPfK!iE3|uf?g7d{ne6BfUU3kv~OU zRRy!VZ(|RcId`4KL-7}D zP+#+#e>M0n_P#I3;emf6w!ez8kzzCan43}Y%*pl)CqKR@eYOw%hKvJKU4ii#PHv7^ z9(}FPKEvN#I41}>Y$3%0axl`cf3C>5Ix+~lt0oe8 zvNnB!r{O^^3BLzga^Jz<>Hj}%MJL(AGot*`!b+|;;teArT^r_Fzct*bXK>QJ#Py}> zdeEMXPZ+_`wsD1*sco)q&r=urXyVuDxm#pz?ZIe6u8zNj@p-I$_m; z%yg0#>N*x+k42n$i(&lAYABk|$RRY{#cr3E9O@p~sB>=3fPOX)OU&@SEeU3q=3*w4 zzsZjJhvBW!yZGdbQg3e2n^pn{r5(C9F=K|3va%tq`>C!i9M^gqulGv6k20idrZ~PD z58DFW#Z;128G%)?$g_7v`b~qGF;v{@N$G5UUX^(k4%CT@+{Ug!?W{z#AF6f;Cg?`R z6e$=3x=z9Z=pRy47VDip!0p?#W-nkhd218_7@Bb*i)nUjLkU~V21~;)wk`Ki)T>@s z?7G|j*p+$LP8)4hJn0&^h4^gDYRY##Aaw4j)ZGUU_s%|t#VF;6-)jqR2t&)yVFCKG z$iS?N)rKbPT;^IBH8+P!R@FzwIX-#Ze7e4V?YT#Dwd>9G97sj~=Lp4DT8i%>`>z5* zz8Re+HJY*}$F8Uu-gdHOKHSat(2*L|zi{o~+?&vl%QWg5YCm%@2oaeM7w@mDaDJVd z7>kcqYa`YZdnh36-{bD&1_g0q)b2pTdPYO1WqOq`O7$8RcnLQFUFapybo<)F!Xd~Q z5W?GM9>xMl0o9#?JYFk$qlAVZd($!0UNDS*0vtooi4l#Jzj(d`1aSY_B?4jCMC zy8t5hU1_ffzUBW-H*}T@0fr7YbFsjTA$(r%f%?N5QP_{2ifGOgEO16B3PqofL0j>J z^3j>~Pm^rh6r<@&j17D21jjg;mR{GGMhIDz7oDFaL!1a|zkY4L+n(GmPX*XK$HvGiqvKcaE%kj z7mmQ-o5b^G1@f@)M!uW(SnOW3h93L356Riv%rWamUYqY31Ky0faI^lVI_Z?}2xuRe z#kiU^yI-DeQtn=Vb(eKD{asHpw?J6c7|xmmx%9fr%wo%stBmdfXNDNx$K?-f)KDRq zL@(@Rb(pefcznpxE2Vcw(p3t}Yj}WntxgIGcB>S*6iJ@=gsD4{v&6|+S}AMngS=#d zqq@8tXbOJ}+>BLs8Zc6&wkNBCyZQ&AyRdK&k628HA(UXzBZ~HB$Ye3lH)t3CS$X`& zInFi~Ac8K)o17fSoalHM(lt>WQ|VyppI~CAMyV%OB04}0lK;cNd9LrMV(zhKVm3@p z_^np*juCTk^c!jAv#$;N`XkK_5{b;#phbKe@DE%Gh@J&}mjsOCi-?Kr}kR*^?&)EfL~3yKkl&}RPEA1)SI5n0-IaazT~)Y|PrGvsZXlNwP6^P|W@ zCoO(mi(gTXbHDpQw4nND!UFT%3Lv9IgRxSvZZQ{X8Vj)Mi=8>fAM-zOxdJ1e2j$_p z!$0CQR?w*VWA1=~Z}B^5rc2ppPd0Eng%-BHq`GRf`G~(acmq_a8GjhnpOyOmjlLx9 z7xXP~k{e1+T7_==Gh~#I&R4b$7>K@m^=V53! z^XY>@lZ7(%iFatOaL|4K*!w$g{M>qguEt+g7J}dhqQ9(=;P?buyuTtE|3)%?MGC*C zs*Z|bUyI-|Wm)9O9CZJZiVfGtJrx>tgZH0*HZ}ioPMC!Tqpqk(bx^ca&yuXh*IQ&r zU>)rsSdPfX^p3GKkhzypyaGgZX%`fphiuU>hziUsg?tlh*L{d53l zadnWZxeR!eG=u=!t$zd7{VKcv)-iv2 zv=nF7Td*`p57o&M(0GM`Gms0I*&Gw@aau+0HBuXTB?(Yv^-7i1cbl!gAF-wNUrr};6yP>{ad2v$?hp-mfz-R=fSBYujEKA6eu0; z>M^-Jyl+j?-K?+qZ8fU8i6MhzOLvQ7_>8x^{^F{>cSG)aynbD&SUgiuw*+C^+;{|q zw*;8L5YrOx-WKj)>cOkbAH%IgzwqW}BUk-(r997T@Y84W>d3>Y8*w%cOKqt>%Q1c# zcALspS=xu@?I>gK=6N`!#o-t4rbMK0;}}Zw_=37?dy81TI3gWc9Mowgiv{Xqy*FMu zabkfN-r2~4IxSOVhCNENd8pH)z~qDScqYBfK+ZCY+cFj?d>RLa#6xn&(M1An%K|CX&kzicgdU)iAutL2zmbz*vR zsttO-eC8>^&d!jHhN?4pGgtq);ghY%8}W}aN~w(mGcBezC& zn>Mz9qMpgz)B&wTPq?4AWE5=Y)yp)i6R@>U#yYo#xe>aw^Cmbk@D4+p7jfrYf#;o_ zYLrd;#tl9utx7{S=Hq*?;L{V5E$&P&n|5oyNGX)y*<>d!oY*DV-t>q5NEJwJuk-Fn`rVgCK!8DNT*x zhS?{mJta(MS*3xpN|xSbwHx8%Z{JfB{z7ut6~TR}!stUwRIwqu5ASMwxepf7i-xzoyTHvJCD z{M1swP;ww1$^&{wU*Ll^V>9)Xu#V7%gFUYOqQehy%XgeX_~)-VYyUgZCU&uFc>Yok zLG{Ufs?~r#dWAWq-7Nxf!%HCUc7lolgALJpr3*rb$%>;`z_J1hB=5kcA4isfJK24% z4ci}0IMo$PC@dDdY!CBc$J=yghu;k(Pvddz|InQX+_`gXd;`($KvPC>d@Yzn#gP>P zLW%t#MESTrHTTVP!>3<BtWz>xxGh3UEapw z56I;wB!xBc2Zk-CL|p~HlNC(5F@~Q?ASy4jeB7b zZV=PnWds50;t=RQ$Ai4BpP09r90ZNIDSr=ZziV+SX10%6L{v{OL0I56?E#r*-A^4) zrF(KWz~z$zlRN$+PWX3(_9w*=3A-2@(19Rxtk0>Q5smQFfC{byKhn}i~r3a`O)Xaas?cW~XpVs5U+5MGE zzPEoxYd@%%{FGmv7~j8V2$;OdK?G*#4ycV*VEfli)DYcMKlO}iPK+ggmZy2Z_{RW0 zni|KEpU^&<(K!!@2ne=S{c5oOZt(7I`&spv{s3~a4`#BA*45YE{=u2j?Q?MTyaaw{ z-0uVY8_E4Ct>S->72uirlPbxt{z22M-1Gxo&p-T(?`-(FY5#@=#{ua23FM4aXIOOi z=*ROnM>n!RfG)#1-`{Wgx0CXxN!T%4i*q@8MgL*MxG`69dbfL72eK2IM&ivM4kZScZM{ybbyzPo`0L6e#txS96dPCUAGRksj7m z&BKk?orMy#u#<{WDx6^8e7xZq%fPvClm%;w`T%a+_4J(`o9q-nNP`P=#j=9Q-Xsdm z7=N)})12EPQ5rWO`wU=eR|79bbnT7i(xa|OUNDiG1I$g6ON5{9Hq0xcuQE}bkb>!N zzouC;9^B*pSG`E+6H3xHds@!Ty2y|ddH0h-8-e&kdvfBU;#0a7=>82AEZ_qsRs=Xc z(6IiP5!WvH!$L-WPXgKj^}A;O`RB6C+lPX|u$AYDV5tSTTqMx%iKHVh4>&Y@t-l0D zx}R*(I6-KhqRd$p6YZg_54AgY1`WQ{7|Fj;jzZ)xfF&EiZo~JK0P{3{BwhZym_IRH z{!0u6?XPF(o~?N&AkskDNBZeX+m&w+j+@)?dsCR!9f~uu7}yLcbk>NLmVd%R_n!y^ zLoXKiPK45bCchftf0MKwh@4!k{(#t`QEogBYM%d_%?|=By@==lveTUqZx`O z_z#M(nuq7!|Ii>MeNREfAJybaP71lpj}mFufXsLfT@iNpwHgU+HNu^hXh58CBvt%_ zAVGe*DEmw6{~o2Nz2EoJZj~`i>`;HR#pT-j^hC$!%%-XHb?z!ORd4On|ML~iQ2%El zLg2w|R#6hco0#q-R_q@Z3IPK$~ik0#h<0`VuSD4i5{idbyD!zF8=jM6yBI29t zA;~CU$gAVQzJNO{a@J7ir6)~2ThJ-2#g>0Kp3BirI4Ww*uCJeb7zZIYUrcJ)#Sy9F zb6xznnGO9F#|T%V!*{@+mo!lGwY-BEoWfJs#b$u)=EKJ_dY2v-)Stb8F)pA*J-lZ6 z=pm%jU!HzU^6`z7Rh{G*k|sJxp?FY`-$1=mP8T@1u2B&>V|u-0S9q|*)2d)C!5B^| zA^)bt$f)5>bl!r3YDICj<(?ol%2<`+N)-RZ=8aSrUg_qG9fZ8?4VpcG$2+#GHjU3d zI#STAVmLfvKVHDIk%e8qanN_)EB^^;3hT%+>^%GKA|f)g!=&2@a3^d1a`Cmuk+34^ zU70=ykjKPk6a2eCJzt}bQ_JfC4wk3XVSW|PcXFx(Y>OL7 z+Nwbq1`pxqA|3svxbmj89?K#)1I;2jlm-1A(Rb!~RvP`JgL7=l&Z`K0otT>7r1Zma z5oeI*?e%7wFtnKB7^#&SER93PJ?VQ~=%+Z$IEpH~(nkcM)831OD=L6BXm%`S8%{{a zVAD?g6`}h~-ETKH4-71>M3txU+P4pW*pCdtFv1l+CCvXqA@Z6-(ru75IwhuAI|=Sv zE0#Sa?_!k|yd2`saXiQjNqw$HDF1s*@z?enr*4`XbHE*fx5;^jxA=?|^8*+qLz@=xZZ1Xb<*`R5j$1LSkp^=%^QW7O*cC&VVnCo1}#MDMk? z8M6)-;a{vSAme4`DOZbG5yMGXERDpSTpR0|%AcEER$8JBC5YixEQU0%Kqcbak_iV) zk&7dn;c5DYW7%&D8Q89|$kE`e(1uvd6-pen2(Q=?V-V*yPh-xv3HN%f1vU3*%xfXT zs12*!9y!0PT5ELdn~{HHuC|G4E@XBLu2}x$BhBzmU0ISqO-M-`(2*94sJ)Kg#x*j- zM=m4#1$vDrkM-uD=%mYrzISpbD;&(=$rSgzb`r5BP3T@`fjn*2F1NB7v%hZ_7^k6p z{h-$s&5h^Xg&9(>pu4DZ-SC#X)N}d=F@3a!wK09{?KGBplhuGEB6ej!>}CB`7S>*! z8)CZWZrp6-t@LI)+ArE}vv)Hl8!^Bnef9QK!-f69H;(CT-b|c01!G_uSA}7`pA=on zT0L!9-7p_Mf|{=TxTp#znE4!Dnd9x*=>RQXMzxNckcN1$Vl;(3g;JIyHSIN!g41uO z{8K@KwoP4rd+JKOepvd80HOS|B`KWae*Ce(!tB-S#J1a)DsB&bc|m+a4sOc^*Luvl zJ)tTpCgKYF^eOU7m=4{n$$6#ut50_*f-|Gd7wp-Cud6kBNpPE)yiW9!3g<72acV49sK{5;oVbU?pBo|KyYh=aT} zv;BEA>;j4eI2cdtD13go&R(B>UMM=9GlyX=*PNB@(+*3_FLqR9i1G80JU@A>Po+YPuHw2u3p7Fj@Zx2qU~zpm z_lVTJ7N@guih*!RmbPyS1%0q181@Wq{M_Oiti9GwsX3x~m~{c8RhWyGL2iyF@?Bfq z8NWw)Y~ObGsrbCq9;%=4K|xXK8^4Svl&+$x>LdGS>>@F0HKCKto0^x}RO}$c2a>PC zZyeoE>ds6rj#KKz>4W?dIhu+p^=ug^1fxgcC$MRblYT-lJ(M^*5Zq4@w!i{jBZ&~| zmFlGa@=wDl$YP~nl&<$uUqai7llodq2+oOkHF$^$!uv63RjolcqgS#l=m1sy36e$9yDSZPTMXRNgi_*H`r-0eZBP1Ais{883$CY3FJ>t-iJD*8g$Dfc|B zYD)uI>1sjp`kml!(|5$u(vyl5#g|j}*?lC}c*9UGa~5TjJO&1%{2tG4)vy+2r^=uo zBFto|`lPf!C`s+|*(qcueVySx$a8nD$z0oJw>avbD1Asr^?rtNPD%-8H!nyj7_eVO z^eLdixVKJR8cH#I#shEqsxjW8zfZON#8tj8#hx3l{~+lx;?@vnLoGGEnvZ8$=fZrZ zQs$bP<;fGe!*TSxczauV@m>jU+JGHeF%(4Y@GER*RvS$p2qjpR#etXj_ce&%q~}_m zct;V%tJMZhQmAbDP0I~qXp^4KtCE}avQ>3VW%gIo2A0&m*N`%>*NTJN+~ zXPlJyih9sVuNL)$v zA}hHol%~)o4|;B7P6pwsO`!sYHRE?5k+*Z^0Fld~%V;;-i!f1#%pA`|bB8rD zy}akPx*y^ChjIhVG4qH4lo4fr@`#49l;G^6bmz%vc`kCk8F5(fTL1jbon2*ODTv*BDp?j6ecm2Vf`Vy_gvs`o2Nvf`IbS)O zzBf%uIicz(gfXqTFDF{3na4$k*vg$-aa5nc8=rtT`G_^lULLrg!gns)#^%+rx0+|t z6kNZ-&50%UJwjt^clI`iHG?=F7wvpB${9-Ly4QTDd^Jw{dM95{`b~%T9(KkR1=;UX zr?Za=7a*wB6e1A{N~{la^)Zy^+Cs7L<3> z=S!8Jl&cOgyB12FGR;yXz)I=a{dO(?z7I8_b20(x&*#@ywbx${N)g(oqGHSVWp>T2 zNgj^*zgA7<%EQzjAo_9s>eBp>luww5e(1ue@ZNum;VV4H7+UD!ev8wcKU6XNl`XxY zyESdd{3eNtOEY!5O-<3mM=Z9!!w@?YHxeX!{VcSXcu4K#mJyu(N1tre~Tly zzr><-UvaLzA`@`$Up0QZV~UX}g`TWnfgYEGT(#vlDs5oXv0WKxQ>cJwTS1ky26^~2PG}Pk9(&s4_B-IGEgN&K0KZMTf5Bt(pOa$2E zWh`?|B-HJkmm8LnOO-$OK4({6?5%pD#UnrQK8EEwCt?Tu!t$H*&igv}qkMr_RMGtRSTe6qb_xNh^6qN{OYhY7KHhTkR>r38c+$YEy+was(Z8;orwnX4Dc zOC~v9WAu6lTYv^*ZoHAt9UW=!Na$z7i95CMpaC*JKK`v^N3+^hQ|rlT)3=?-Osior z=T0{>(bDwmU&6xoIcRWNJTKodIe&jWeA98f;Zo?fRi`IQ9_VA!7EhZqcDr5cIF5KW zE4M=KMbs`cvbHXwN*T+k5Ed3hO4MFtju(e|9gHEY6Aa6})nL?=fpXr$5DpzJ`^zYPs-`VwathWYJ=;#@!n?1u6YIbsOn;9O5@s_s;x_BLNo{_C?31DuN(aMv@)IZ zJ4Tb9oS_X)MQUOTR@DVix|PCWSIe!gYf>A8ub@-LvS@d25n!8$U2&G|s%hJF$<_Jc zF!*SBes4XCYu=P3PRg3bd8_VcSLxquP6!{lZBbb66K_2%^UbPE5KmK&-uf(yRsq}^ z+6dHbGdiH7#LRfp^t3CZzx0^e^c{uWQ$^4&>X!|m0_I;SO9Fja-_DNhJ{vA3?lj`o zQM7ooRUq494m(rgv?i!Du1`RSh;VGH*hI|H{X0I#J`p2oaR@*s{`%K%H<*tOyucllV->Do_pTR z*6)kr(_Sf}FpY@KmL;%DzPaOWepr~L3m4O>VxqkeLof9NOl?zTbe$AYKTFJ-^U^kX zgVXU!n)qm+*5vaPUHK5FMEyq>vbN&uQ_5XTU40GCN9Y_p#YP#1nC65_VzIdXiPLeT4TO->1`DA$10XvA4GkNU3;6G z$I6nz7c>bf87vObQIRIz?(>;|L-EWs-5T#<;xuh1L9@d`lP)`4^^Fe{M-AoCk}wrC z7l51&UfgTR;1mAX*?F4n^1|Jr$6Z8>(onXfS;YQU^Q8e_4Vkh z=bh!p4oJH(@+Qf6x&-cCu# z28GJdscpkrjmvdV@aa&z@8i&_ySbvpK`sgL!CRsp)^>%QZ~SUFBww4-PtdCo%FRN^ zGRM_O@jixLdN<0@=T53|mme2vk8xTpiR$ z%91q3h76Lsl$~3mvo^Z-sCwz$&FTP;S7~_n`0((G8;^BO z5nmRnwaXI;%71-LbzV~q$JlMY??TyU8Tx@Y6DEdm0aJ)1;gNTJyXozfF2$;GuCdN6y^WOKd<~qs=fdR-QqRJv%BILh zdl1>h`{HQ?7K!P|oydbbV+s3nIQUert$ylU1D@NpQ*VaT(hVpONQBOzE#@5u6c23@ zLk{%3)d<=7_PM6})o@YEt*hx{ z--A}IGz1+KY!kW~kN@&|?WA==jIp-eO_1A`SLSSddbem!GWZ#%BA<<-S^*j0M)7L9 zh+pPvyVOPSF^|cY!{q3W-WEf03zv|pp-y*BwbSQvQ>tteJ73;tIh^%@_~aSOgc?z_ zx$A^n6zYGfeM?#=2mJsOyLwR~AS?4(ymL&(2l0-lH_>O{*h>L(WDd>VS0kJ2FTJ0X zzmCiBb&CV)xFz*km3hDTmtyW6=`Y3{sMw4#i{2}z*VQ?L?awheIBf8atN>ub+TT5j zbc|45sp}r15?>02rPHO1oW!8*po#jMS7I+voGDTLZ|!Jb!5=k)klcMab5e7r0b zr~ZDvNN&Gyj2pni`!!?4-u}25F&*!_cOvvRG-I?-gIn?zJam=k^zCV(c*pnRfqUL=hzmQ6H&9sKOmsZQj=dhHN!L{ApqlI{UV5jn11`Ko6`K zrBVWw*(oIcn1 z&($uWPb;ENta84PXn6*zzdu<79|r?#O6!L%?`eQ|%15wrly+KOmiay$*>GpbiOFik zrtDnYlvXTW$4lyL4)Oc3-@0iWns!teJDI`@!i0$S+2QiQi}-1hQma?~qDieIard1$ z&Z+PBaqQ*mb2VGjDOM|22piuI;xXf}gd3$eoh;Qb{#};BOGKc_(olh#(k-fFo$FHg zykf%-{g0njO`{%2*x}#Eb!{dzZwh0b%pb@}p}22lFR15QTw=P(}m2XQU}pE+ey z;J>8iH6U|JDUQ*@4ci+qw7ko(8ll*11O93~T}{GLmq!(=F1q}J#OreoZ8YxMMa5~W zC!6oehWSU$PjXkfyyP}@R#GUWU@zStIV@VbhR+0=)t*)*tZgHXJ=IeW1vJ1)A2n%h0OJHK=oF?AC zniZ{GGwd&UQ9SB;G;9LQ2KgK2qBmwlW=002RYtTKgy3&5)E;VvnPmY7xnmYJ^E(G?X^vGJmRZ)Y*hd*b3E^!g$ zV6K5X=$HNwV*e%dmSdoo25pmgkI8IuK%Sv!%Pp&`x(mnn)?5e+W^eieX~kJSDvNmP zF_Vct*H_h-U!)qa)L9L!q>yvWmN#5XO@lp?vLKrb9_z@Xv&RA+to2~{s=v9KCZZWD zkAH-lBulR!`Eo-PN~v!ptZZL#qsEy>=%qMkm>==>xIOYC z`wi+lB*gn&jGNpk7KsId3yaB6Tq?-A3B(P}$&TIj#Sia>eT0** zr4Lr$R{VbP+Vkos0%2Xm*>} l!kAIvl=PP&@#em(6ASJh6qz>vAY + + + Quarter + US & Canada + Europe + Asia-Pacific + Rest of World + + + + + Q1'24 + $15,451 + $8,327 + $7,338 + $4,519 + + + Q2'24 + $16,593 + $9,135 + $7,721 + $4,880 + + + Q3'24 + $17,389 + $9,358 + $8,050 + $5,088 + + + Q4'24 + $20,982 + $11,154 + $9,012 + $5,635 + + + Q1'25 + $18,259 + $9,527 + $8,224 + $5,382 + + + Q2'25 + $20,045 + $11,366 + $9,148 + $6,004 + + + Q3'25 + $21,331 + $12,072 + $10,020 + $6,659 + + + Q4'25 + $25,643 + $14,198 + $10,893 + $7,403 + + + Q1'26 + $23,667 + $13,296 + $10,631 + $7,430 + + + Total + $35,635 + — + — + — + + + Q1'24 + — + — + — + — + + + Q2'24 + — + — + — + — + + + Q3'24 + — + — + — + — + + + Q4'24 + — + — + — + — + + + Q1'25 + — + — + — + — + + + Q2'25 + — + — + — + — + + + Q3'25 + — + — + — + — + + + Q4'25 + — + — + — + — + + + Q1'26 + — + — + — + — + + + Total + — + — + — + — + + + Total + — + — + — + — + + + Total + — + — + — + — + + + Total + — + — + — + — + + + Total + — + — + — + — + + + Total + — + — + — + — + + + Total + — (additional values) + (continued) + (continued) + (continued) + + + Total + 58,137 + 14,198 + 10,893 + 7,403 + + + Total + 55,024 + 13,296 + 10,631 + 7,430 + + + \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-1 Quarterly Revenue.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-1 Quarterly Revenue.html new file mode 100644 index 000000000..a325f1611 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-1 Quarterly Revenue.html @@ -0,0 +1,76 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterUS & CanadaEuropeAsia-PacificRest of World
Q1'24$15,824$8,483$7,481$4,667
Q2'24$16,847$9,300$7,888$5,036
Q3'24$17,609$9,492$8,220$5,268
Q4'24$21,783$11,503$9,245$5,854
Q1'25$18,605$9,680$8,439$5,590
Q2'25$20,371$11,532$9,366$6,247
Q3'25$21,751$12,268$10,272$6,951
Q4'25$26,472$14,480$11,182$7,759
Q1'26$24,079$13,500$10,914$7,818
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-10 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-10 Quarterly Values.html new file mode 100644 index 000000000..13308a8a4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-10 Quarterly Values.html @@ -0,0 +1,30 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue (%)
Q1'255
Q2'2511
Q3'2514
Q4'2518
Q1'2619
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-11 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-11 Quarterly Values.html new file mode 100644 index 000000000..f0dc25902 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-11 Quarterly Values.html @@ -0,0 +1,30 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue (%)
Q1'254
Q2'259
Q3'258
Q4'2513
Q1'2613
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-12 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-12 Quarterly Values.html new file mode 100644 index 000000000..623c4cda4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-12 Quarterly Values.html @@ -0,0 +1,30 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue (%)
Q1'255
Q2'256
Q3'259
Q4'2513
Q1'2617
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-13 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-13 Quarterly Values.html new file mode 100644 index 000000000..83cc5a749 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-13 Quarterly Values.html @@ -0,0 +1,30 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue (%)
Q1'259
Q2'2516
Q3'2523
Q4'2524
Q1'2623
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-14 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-14 Quarterly Values.html new file mode 100644 index 000000000..1a00e97fd --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-14 Quarterly Values.html @@ -0,0 +1,30 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue (%)
Q1'251
Q2'257
Q3'259
Q4'2514
Q1'2617
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-15 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-15 Quarterly Values.html new file mode 100644 index 000000000..400090484 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-15 Quarterly Values.html @@ -0,0 +1,30 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue (%)
Q1'2510
Q2'259
Q3'2510
Q4'256
Q1'2612
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-16 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-16 Quarterly Values.html new file mode 100644 index 000000000..13db07176 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-16 Quarterly Values.html @@ -0,0 +1,30 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue (%)
Q1'2514
Q2'2511
Q3'2513
Q4'259
Q1'2614
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-17 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-17 Quarterly Values.html new file mode 100644 index 000000000..7a448d9d3 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-17 Quarterly Values.html @@ -0,0 +1,30 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue (%)
Q1'259
Q2'2517
Q3'2519
Q4'2512
Q1'2619
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-18 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-18 Quarterly Values.html new file mode 100644 index 000000000..ab99c7604 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-18 Quarterly Values.html @@ -0,0 +1,30 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue (%)
Q1'253
Q2'252
Q3'251
Q4'25-2
Q1'265
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-19 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-19 Quarterly Values.html new file mode 100644 index 000000000..25803e5e7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-19 Quarterly Values.html @@ -0,0 +1,30 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue (%)
Q1'2517
Q2'2515
Q3'2520
Q4'2515
Q1'2618
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-2 Meta Revenue Data.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-2 Meta Revenue Data.html new file mode 100644 index 000000000..dfbb20a18 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-2 Meta Revenue Data.html @@ -0,0 +1 @@ +
Q1'24Q2'24Q3'24Q4'24Q1'25Q2'25Q3'25Q4'25Q1'26
Advertising$ 35,635$ 38,329$ 39,885$ 46,783$ 41,392$ 46,563$ 50,082$ 58,137$ 55,024
Other380389434519510583690801885
Family of Apps Revenue36,01538,71840,31947,30241,90247,14650,77258,93855,909
Reality Labs Revenue4403532701,083412370470955402
Total Revenue$ 36,455$ 39,071$ 40,589$ 48,385$ 42,314$ 47,516$ 51,242$ 59,893$ 56,311
Family of Apps Operating Income$ 17,664$ 19,335$ 21,778$ 28,332$ 21,765$ 24,971$ 24,967$ 30,766$ 26,900
Reality Labs Operating (Loss)(3,846)(4,488)(4,428)(4,967)(4,210)(4,530)(4,432)(6,021)(4,028)
Total Income from Operations$ 13,818$ 14,847$ 17,350$ 23,365$ 17,555$ 20,441$ 20,535$ 24,745$ 22,872
Operating Margin38 %38 %43 %48 %41 %43 %40 %41 %41 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-20 Free Cash Flow.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-20 Free Cash Flow.html new file mode 100644 index 000000000..890dc1444 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-20 Free Cash Flow.html @@ -0,0 +1 @@ +
Q1'24Q2'24Q3'24Q4'24Q1'25Q2'25Q3'25Q4'25Q1'26
Net cash provided by operating activities$ 19,246$ 19,370$ 24,724$ 27,988$ 24,026$ 25,561$ 29,999$ 36,214$ 32,226
Less: Purchases of property and equipment6,4008,1738,25814,42512,94116,53818,82921,38318,997
Less: Principal payments on finance leases315299944411751474545754843
Free Cash Flow$ 12,531$ 10,898$ 15,522$ 13,152$ 10,334$ 8,549$ 10,625$ 14,077$ 12,386
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-3 Segment Margins.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-3 Segment Margins.html new file mode 100644 index 000000000..d45892344 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-3 Segment Margins.html @@ -0,0 +1,56 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterBottom Segment (%)Top Segment (%)
Q1'241827
Q2'241927
Q3'241828
Q4'241825
Q1'251829
Q2'251827
Q3'251830
Q4'251829
Q1'261831
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-4 Quarterly Tax Data.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-4 Quarterly Tax Data.html new file mode 100644 index 000000000..6554d881a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-4 Quarterly Tax Data.html @@ -0,0 +1 @@ +
Q1'24Q2'24Q3'24Q4'24Q1'25Q2'25 $Q3'25^{(1)}$ Q4'25 $Q1'26^{(2)}$
Income before income taxes$ 14,183$ 15,106$ 17,822$ 23,553$ 18,382$ 20,534$ 21,663$ 25,354$ 21,752
Provision (benefit) for income taxes$ 1,814$ 1,641$ 2,134$ 2,715$ 1,738$ 2,197$ 18,954$ 2,586$ (5,021)
Effective Tax Rate13 %11 %12 %12 %9 %11 %87 %10 %(23)%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-5 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-5 Quarterly Values.html new file mode 100644 index 000000000..bd7718aab --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-5 Quarterly Values.html @@ -0,0 +1,46 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue
Q1'24$12,369
Q2'24$13,465
Q3'24$15,688
Q4'24$20,838
Q1'25$16,644
Q2'25$18,337
Q3'25 (1)$2,709
Q4'25$22,768
Q1'26 (2)$26,773
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-6 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-6 Quarterly Values.html new file mode 100644 index 000000000..d0781aaef --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-6 Quarterly Values.html @@ -0,0 +1,46 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue
Q1'24$4.71
Q2'24$5.16
Q3'24$6.03
Q4'24$8.02
Q1'25$6.43
Q2'25$7.14
Q3'25 (1)$1.05
Q4'25$8.88
Q1'26 (2)$10.44
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-7 Financial Data.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-7 Financial Data.html new file mode 100644 index 000000000..eab216c5d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-7 Financial Data.html @@ -0,0 +1,26 @@ + + + + + + + + + + + + + + + + + + + + + + + + + +
PeriodValue
Quarterly$13,692
Q1'26$19,840
2024$39,225
Annual$72,215
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-8 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-8 Quarterly Values.html new file mode 100644 index 000000000..5213d9594 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-8 Quarterly Values.html @@ -0,0 +1,46 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue
Q1'243.24
Q2'243.27
Q3'243.29
Q4'243.35
Q1'253.43
Q2'253.48
Q3'253.54
Q4'253.58
Q1'26 (1)3.56
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-9 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-9 Quarterly Values.html new file mode 100644 index 000000000..7c517e6b8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-presentation/tables/table-9 Quarterly Values.html @@ -0,0 +1,46 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
QuarterValue ($)
Q1'2411.20
Q2'2411.89
Q3'2412.29
Q4'2414.25
Q1'2512.36
Q2'2513.65
Q3'2514.46
Q4'2516.56
Q1'2615.66
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/chunks.json b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/chunks.json new file mode 100644 index 000000000..22a3c4e4c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/chunks.json @@ -0,0 +1,1083 @@ +{ + "chunks": [ + { + "chunk_id": "043b2754-18d4-56c4-912c-727b0fabb127", + "type": "table", + "content": "
Three Months Ended March 31,
In millions, except percentages and per share amounts20262025% Change
Revenue$ 56,311$ 42,31433 %
Costs and expenses33,43924,75935 %
Income from operations$ 22,872$ 17,55530 %
Operating margin41 %41 %
Provision (benefit) for income taxes(1)$ (5,021)$ 1,738NM
Effective tax rate(1)(23)%9 %
Net income$ 26,773$ 16,64461 %
Diluted earnings per share (EPS)(1)$ 10.44$ 6.4362 %
", + "path": "tables/table-0 Q1 Financial Results.html", + "metadata": { + "length": 860, + "summary": "table-1\nMicrosoft's Q1 2026 revenue rose 33% to $56.3B, net income up 61% to $26.8B, and diluted EPS up 62% to $10.44, with operating margin steady at 41%.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "file_path": "tables/table-0 Q1 Financial Results.html", + "keywords": [ + "Revenue", + "Net Income", + "EPS" + ], + "tokens": [] + } + }, + { + "chunk_id": "9d6c51fd-67d4-55e5-8dc6-2ce359bb981a", + "type": "text", + "content": "MENLO PARK, Calif. – April 29, 2026 – Meta Platforms, Inc. (Nasdaq: META) today reported financial results for the quarter ended March 31, 2026.\n\"We had a milestone quarter with strong momentum across our apps and the release of our first model from Meta Superintelligence Labs,\" said Mark Zuckerberg, Meta founder and CEO. \"We're on track to deliver personal superintelligence to billions of people.\"\nFirst Quarter 2026 Financial Highlights\n\n[tables/table-0 Q1 Financial Results.html]\n\nNM — not meaningful\n(1) Includes an \\$8.03 billion income tax benefit recognized in the first quarter of 2026, which partially offsets the \\$15.93 billion non-cash tax charge recorded in the third quarter of 2025 upon enactment of the One Big Beautiful Bill Act. This benefit is the result of U.S. Treasury Notice 2026-7, which addressed the Corporate Alternative Minimum Tax treatment of previously capitalized U.S. research and development costs. Excluding this tax benefit, our effective tax rate would have been 37 percentage points higher and our diluted earnings per share (EPS) would have been \\$3.13 lower.", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results", + "metadata": { + "length": 1104, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "tokens": [ + "MENLO", + "PARK", + "Calif", + "April", + "29", + "2026", + "Meta", + "Platforms", + "Inc", + "Nasdaq", + "META", + "today", + "reported", + "financial", + "results", + "quarter", + "ended", + "March", + "31", + "We", + "milestone", + "strong", + "momentum", + "apps", + "release", + "model", + "Superintelligence", + "Labs", + "Mark", + "Zuckerberg", + "founder", + "CEO", + "track", + "deliver", + "personal", + "superintelligence", + "billions", + "people", + "First", + "Quarter", + "Financial", + "Highlights", + "NM", + "meaningful", + "Includes", + "8.03", + "billion", + "income", + "tax", + "benefit", + "recognized", + "partially", + "offsets", + "15.93", + "cash", + "charge", + "recorded", + "2025", + "enactment", + "One", + "Big", + "Beautiful", + "Bill", + "Act", + "This", + "result", + "Treasury", + "Notice", + "addressed", + "Corporate", + "Alternative", + "Minimum", + "Tax", + "treatment", + "previously", + "capitalized", + "research", + "development", + "costs", + "Excluding", + "effective", + "rate", + "37", + "percentage", + "points", + "higher", + "diluted", + "earnings", + "share", + "EPS", + "3.13", + "lower" + ], + "keywords": [], + "connect_to": [ + { + "target": "043b2754-18d4-56c4-912c-727b0fabb127", + "relation": "embeds", + "ref": "[tables/table-0 Q1 Financial Results.html]", + "position": { + "start": 443, + "end": 485 + } + } + ] + } + }, + { + "chunk_id": "5526d4d1-7db1-58f1-b89d-dba24d94e5fb", + "type": "text", + "content": "- Family daily active people (DAP) – DAP was 3.56 billion on average for March 2026, an increase of 4% year-over-year. The slight decline in DAP on a quarter-over-quarter basis was driven by internet disruptions in Iran, as well as a restriction on access to WhatsApp in Russia.\n- Ad impressions – Ad impressions delivered across our Family of Apps increased by 19% year-over-year.\n• Average price per ad – Average price per ad increased by 12% year-over-year.\n• Revenue – Revenue was \\$56.31 billion, an increase of 33% year-over-year. Revenue on a constant currency basis would have increased by 29% year-over-year.\n- Costs and expenses – Total costs and expenses were \\$33.44 billion, an increase of 35% year-over-year.\n- Capital expenditures – Capital expenditures, including principal payments on finance leases, were \\$19.84 billion.\n- Capital return program – Dividend and dividend equivalent payments were \\$1.35 billion.\n- Cash, cash equivalents, and marketable securities – Cash, cash equivalents, and marketable securities were \\$81.18 billion as of March 31, 2026.\n- Cash flow – Cash flow from operating activities was \\$32.23 billion, and free cash flow was \\$12.39 billion. $^{(1)}$\n- Headcount – Headcount was 77,986 as of March 31, 2026, an increase of $1\\%$ year-over-year.", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/First Quarter 2026 Operational and Other Financial Highlights", + "metadata": { + "length": 1290, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "tokens": [ + "Family", + "daily", + "active", + "people", + "DAP", + "3.56", + "billion", + "average", + "March", + "2026", + "increase", + "4%", + "year", + "The", + "slight", + "decline", + "quarter", + "basis", + "driven", + "internet", + "disruptions", + "Iran", + "restriction", + "access", + "WhatsApp", + "Russia", + "Ad", + "impressions", + "delivered", + "Apps", + "increased", + "19%", + "Average", + "price", + "ad", + "12%", + "Revenue", + "56.31", + "33%", + "constant", + "currency", + "29%", + "Costs", + "expenses", + "Total", + "costs", + "33.44", + "35%", + "Capital", + "expenditures", + "including", + "principal", + "payments", + "finance", + "leases", + "19.84", + "return", + "program", + "Dividend", + "dividend", + "equivalent", + "1.35", + "Cash", + "cash", + "equivalents", + "marketable", + "securities", + "81.18", + "31", + "flow", + "operating", + "activities", + "32.23", + "free", + "12.39", + "Headcount", + "77", + "986" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a409e3d7-75ca-59b5-9891-1ecd90985f3d", + "type": "text", + "content": "We expect second quarter 2026 total revenue to be in the range of \\$58-61 billion. Our guidance assumes foreign currency is an approximately 2% tailwind to year-over-year total revenue growth, based on current exchange rates.\nWe expect full year 2026 total expenses to be in the range of \\$162-169 billion, unchanged from our prior outlook.\nWe continue to expect to deliver operating income this year that is above 2025 operating income.\nWe anticipate 2026 capital expenditures, including principal payments on finance leases, to be in the range of \\$125-145 billion, increased from our prior range of \\$115-135 billion. This reflects our expectations for higher component pricing this year and, to a lesser extent, additional data center costs to support future year capacity.\nAbsent any changes to our tax landscape, we expect our tax rate for the remaining quarters of 2026 to be between 13-16%.\nLastly, we continue to monitor active legal and regulatory matters, including headwinds in the EU and the U.S. that could significantly impact our business and financial results. For example, we continue to see scrutiny on youth-related issues and have additional trials scheduled for this year in the U.S., which may ultimately result in a material loss.", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/CFO Outlook Commentary", + "metadata": { + "length": 1254, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "tokens": [ + "We", + "expect", + "quarter", + "2026", + "total", + "revenue", + "range", + "58", + "61", + "billion", + "Our", + "guidance", + "assumes", + "foreign", + "currency", + "approximately", + "2%", + "tailwind", + "year", + "growth", + "based", + "current", + "exchange", + "rates", + "full", + "expenses", + "162", + "169", + "unchanged", + "prior", + "outlook", + "continue", + "deliver", + "operating", + "income", + "2025", + "anticipate", + "capital", + "expenditures", + "including", + "principal", + "payments", + "finance", + "leases", + "125", + "145", + "increased", + "115", + "135", + "This", + "reflects", + "expectations", + "higher", + "component", + "pricing", + "lesser", + "extent", + "additional", + "data", + "center", + "costs", + "support", + "future", + "capacity", + "Absent", + "tax", + "landscape", + "rate", + "remaining", + "quarters", + "13", + "16%", + "Lastly", + "monitor", + "active", + "legal", + "regulatory", + "matters", + "headwinds", + "EU", + "significantly", + "impact", + "business", + "financial", + "results", + "For", + "scrutiny", + "youth", + "related", + "issues", + "trials", + "scheduled", + "ultimately", + "result", + "material", + "loss" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9be2402e-24b4-53f6-acd1-eada59ca225e", + "type": "text", + "content": "Meta will host a conference call to discuss its results at 2:30 p.m. PT / 5:30 p.m. ET today. The live webcast of the call can be accessed at the Meta Investor Relations website at investor.atmeta.com, along with the company's earnings press release, financial tables, and slide presentation.\nFollowing the call, a replay will be available at the same website. Transcripts of conference calls with publishing equity research analysts held today will also be posted to the investor.atmeta.com website.", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Webcast and Conference Call Information", + "metadata": { + "length": 500, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "tokens": [ + "Meta", + "host", + "conference", + "call", + "discuss", + "results", + "30", + "PT", + "ET", + "today", + "The", + "live", + "webcast", + "accessed", + "Investor", + "Relations", + "website", + "investor", + "atmeta", + "company", + "earnings", + "press", + "release", + "financial", + "tables", + "slide", + "presentation", + "Following", + "replay", + "Transcripts", + "calls", + "publishing", + "equity", + "research", + "analysts", + "held", + "posted" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9a15bdc9-4035-528c-88f8-d17a863d225f", + "type": "text", + "content": "Meta uses the investor.atmeta.com and meta.com/news websites as well as Mark Zuckerberg's Facebook profile (facebook.com/zuck), Instagram account (instagram.com/zuck) and Threads profile (threads.net/zuck) as means of disclosing material non-public information and for complying with its disclosure obligations under Regulation FD.", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Disclosure Information", + "metadata": { + "length": 331, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "tokens": [ + "Meta", + "investor", + "atmeta", + "meta", + "news", + "websites", + "Mark", + "Zuckerberg", + "Facebook", + "profile", + "facebook", + "zuck", + "Instagram", + "account", + "instagram", + "Threads", + "threads", + "net", + "means", + "disclosing", + "material", + "public", + "information", + "complying", + "disclosure", + "obligations", + "Regulation", + "FD" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "24582b3d-7fc5-52ba-9e33-b5eaf3804ef0", + "type": "text", + "content": "Meta is building the future of human connection, powered by artificial intelligence and immersive technologies. When Facebook launched in 2004, it changed the way people connect. Apps like Messenger, Instagram, and WhatsApp further empowered billions around the world. Now, Meta is moving beyond 2D screens toward experiences that foster deeper connections and unlock new possibilities.", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/About Meta", + "metadata": { + "length": 386, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "tokens": [ + "Meta", + "building", + "future", + "human", + "connection", + "powered", + "artificial", + "intelligence", + "immersive", + "technologies", + "When", + "Facebook", + "launched", + "2004", + "changed", + "people", + "connect", + "Apps", + "Messenger", + "Instagram", + "WhatsApp", + "empowered", + "billions", + "world", + "Now", + "moving", + "2D", + "screens", + "experiences", + "foster", + "deeper", + "connections", + "unlock", + "possibilities" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0698d606-e596-57da-9927-75546cf45353", + "type": "text", + "content": "Investors:\nKenneth Dorell\ninvestor@meta.com / investor.atmeta.com\nPress:\nMatt Tye\npress@meta.com / meta.com/news", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Contacts", + "metadata": { + "length": 112, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "tokens": [ + "Investors", + "Kenneth", + "Dorell", + "investor", + "meta", + "atmeta", + "Press", + "Matt", + "Tye", + "press", + "news" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d43eba36-752b-5159-97a4-858a5e555d0a", + "type": "text", + "content": "This press release contains forward-looking statements regarding our future business plans and expectations. These forward-looking statements are only predictions and may differ materially from actual results due to a variety of factors including: the impact of macroeconomic conditions on our business and financial results, including as a result of geopolitical events; our ability to retain or increase users and engagement levels; our reliance on advertising revenue; our dependency on data signals and mobile operating systems, networks, and standards that we do not control; changes to the content or application of third-party policies that impact our advertising practices; risks associated with new products and changes to existing products as well as other new business initiatives, including our artificial intelligence initiatives and Reality Labs efforts; our emphasis on community growth and engagement and the user experience over short-term financial results; maintaining and enhancing our brand and reputation; our ongoing privacy, safety, security, and content and advertising review and enforcement efforts; competition; risks associated with government actions that could restrict access to our products or impair our ability to sell advertising in certain countries; litigation and government inquiries; privacy, legislative, and regulatory concerns or developments; risks associated with acquisitions; security breaches; our ability to manage our scale and geographically-dispersed operations; and market conditions or other factors affecting capital return to stockholders. These and other potential risks and uncertainties that could cause actual results to differ from the results predicted are more fully detailed under the caption \"Risk Factors\" in our Annual Report on Form 10-K filed with the SEC on January 29, 2026, which is available on our Investor Relations website at investor.atmeta.com and on the SEC website at www.sec.gov. Additional information will also be set forth in our Quarterly Report on Form 10-Q for the quarter ended March 31, 2026. In addition, please note that the date of this press release is April 29, 2026, and any forward-looking statements contained herein are based on assumptions that we believe to be reasonable as of this date. We undertake no obligation to update these statements as a result of new information or future events.\nFor a discussion of limitations in the measurement of certain of our community metrics, see the section entitled \"Limitations of Key Metrics and Other Data\" in our most recent quarterly or annual report filed with the SEC.", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Forward-Looking Statements", + "metadata": { + "length": 2615, + "summary": "This press release contains forward-looking statements about future business plans and expectations, which may differ materially from actual results due to various factors including macroeconomic conditions, user engagement, advertising revenue, data dependency, third-party policies, new products like AI and Reality Labs, competition, government actions, litigation, privacy regulations, acquisitions, security breaches, and market conditions. These risks are detailed in the company's Annual Report on Form 10-K filed with the SEC on January 29, 2026, and will be updated in the Quarterly Report for March 31, 2026. The release is dated April 29, 2026, and the company undertakes no obligation to update these statements. Limitations of community metrics are discussed in the most recent SEC filings.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "tokens": [ + "This", + "press", + "release", + "forward", + "statements", + "future", + "business", + "plans", + "expectations", + "These", + "predictions", + "differ", + "materially", + "actual", + "results", + "due", + "variety", + "factors", + "including", + "impact", + "macroeconomic", + "conditions", + "financial", + "result", + "geopolitical", + "events", + "ability", + "retain", + "increase", + "users", + "engagement", + "levels", + "reliance", + "advertising", + "revenue", + "dependency", + "data", + "signals", + "mobile", + "operating", + "systems", + "networks", + "standards", + "control", + "content", + "application", + "party", + "policies", + "practices", + "risks", + "products", + "existing", + "initiatives", + "artificial", + "intelligence", + "Reality", + "Labs", + "efforts", + "emphasis", + "community", + "growth", + "user", + "experience", + "short", + "term", + "maintaining", + "enhancing", + "brand", + "reputation", + "ongoing", + "privacy", + "safety", + "security", + "review", + "enforcement", + "competition", + "government", + "actions", + "restrict", + "access", + "impair", + "sell", + "countries", + "litigation", + "inquiries", + "legislative", + "regulatory", + "concerns", + "developments", + "acquisitions", + "breaches", + "manage", + "scale", + "geographically", + "dispersed", + "operations", + "market", + "affecting", + "capital", + "return", + "stockholders", + "potential", + "uncertainties", + "predicted", + "fully", + "detailed", + "caption", + "Risk", + "Factors", + "Annual", + "Report", + "Form", + "10", + "filed", + "SEC", + "January", + "29", + "2026", + "Investor", + "Relations", + "website", + "investor", + "atmeta", + "www", + "sec", + "gov", + "Additional", + "information", + "set", + "Quarterly", + "quarter", + "ended", + "March", + "31", + "In", + "addition", + "note", + "date", + "April", + "contained", + "based", + "assumptions", + "reasonable", + "We", + "undertake", + "obligation", + "update", + "For", + "discussion", + "limitations", + "measurement", + "metrics", + "section", + "entitled", + "Limitations", + "Key", + "Metrics", + "Other", + "Data", + "recent", + "quarterly", + "annual", + "report" + ], + "keywords": [ + "forward-looking statements", + "risks", + "SEC filings" + ], + "connect_to": [] + } + }, + { + "chunk_id": "4c634a6a-e022-5e8e-bc63-d804248bb980", + "type": "table", + "content": "
Three Months Ended March 31,
20262025
Revenue$ 56,311$ 42,314
Costs and expenses:
Cost of revenue10,2187,572
Research and development17,69912,150
Marketing and sales2,9082,757
General and administrative2,6142,280
Total costs and expenses33,43924,759
Income from operations22,87217,555
Interest and other income (expense), net(1,120)827
Income before income taxes21,75218,382
Provision (benefit) for income taxes(5,021)1,738
Net income$ 26,773$ 16,644
Earnings per share:
Basic$ 10.57$ 6.59
Diluted$ 10.44$ 6.43
Weighted-average shares used to compute earnings per share:
Basic2,5342,527
Diluted2,5642,590
", + "path": "tables/table-1 Q1 Financial Results.html", + "metadata": { + "length": 1326, + "summary": "table-2\nThe table shows Q1 financial data for 2026 vs 2025. Revenue rose from $42,314 to $56,311. Net income increased from $16,644 to $26,773. Basic EPS grew from $6.59 to $10.57.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "file_path": "tables/table-1 Q1 Financial Results.html", + "keywords": [ + "Revenue", + "Net Income", + "Earnings Per Share" + ], + "tokens": [] + } + }, + { + "chunk_id": "baafc521-c953-5d9e-aedf-6a867cbe6024", + "type": "table", + "content": "
March 31, 2026December 31, 2025
Assets
Current assets:
Cash and cash equivalents$ 23,426$ 35,873
Marketable securities57,75445,719
Accounts receivable, net17,47019,769
Prepaid expenses and other current assets11,1157,361
Total current assets109,765108,722
Non-marketable equity investments28,41027,524
Property and equipment, net194,776176,400
Operating lease right-of-use assets23,26820,404
Goodwill24,74824,534
Other assets14,2838,437
Total assets$ 395,250$ 366,021
Liabilities and stockholders' equity
Current liabilities:
Accounts payable$ 13,326$ 8,894
Operating lease liabilities, current2,4142,213
Accrued expenses and other current liabilities31,01330,729
Total current liabilities46,75341,836
Operating lease liabilities, non-current25,60722,940
Long-term debt58,74858,744
Long-term income taxes16,84921,005
Other liabilities3,6124,253
Total liabilities151,569148,778
Commitments and contingencies
Stockholders' equity:
Common stock and additional paid-in capital99,33795,793
Accumulated other comprehensive income (loss)(303)271
Retained earnings144,647121,179
Total stockholders' equity243,681217,243
Total liabilities and stockholders' equity$ 395,250$ 366,021
", + "path": "tables/table-2 Balance Sheet.html", + "metadata": { + "length": 2292, + "summary": "table-3\nCondensed balance sheet as of March 31, 2026 and December 31, 2025, showing total assets of $395,250 and $366,021, with liabilities and stockholders' equity.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "file_path": "tables/table-2 Balance Sheet.html", + "keywords": [ + "Assets", + "Liabilities", + "Equity" + ], + "tokens": [] + } + }, + { + "chunk_id": "6d7be585-ee73-50f0-b3e4-e2fc7d00fe43", + "type": "table", + "content": "
Three Months Ended March 31,
20262025
Cash flows from operating activities
Net income$26,773$16,644
Adjustments to reconcile net income to net cash provided by operating activities:
Depreciation and amortization5,9993,900
Share-based compensation6,0324,147
Deferred income taxes123(993)
Unrealized (gain) loss on equity investments1,075(135)
Other(17)(96)
Changes in assets and liabilities:
Accounts receivable2,1282,804
Prepaid expenses and other current assets(2,424)360
Other assets(1,082)(52)
Accounts payable(937)(1,034)
Accrued expenses and other current liabilities(271)(2,231)
Other liabilities(5,173)712
Net cash provided by operating activities32,22624,026
Cash flows from investing activities
Purchases of property and equipment(18,997)(12,941)
Purchases of marketable securities(32,978)(11,763)
Sales and maturities of marketable securities19,1764,784
Purchases of non-marketable equity investments(544)(100)
Payments for held-for-sale assets(118)
Acquisitions of businesses and intangible assets(372)(1)
Other investing activities15511
Net cash used in investing activities(33,678)(20,010)
Cash flows from financing activities
Taxes paid related to net share settlement of equity awards(4,423)(4,883)
Repurchases of Class A common stock(12,754)
Payments for dividends and dividend equivalents(1,346)(1,329)
Principal payments on finance leases(843)(751)
Other financing activities59222
Net cash used in financing activities(6,553)(19,495)
Effect of exchange rate changes on cash, cash equivalents, restricted cash, and restricted cash equivalents7112
Net decrease in cash, cash equivalents, restricted cash, and restricted cash equivalents(7,998)(15,367)
Cash, cash equivalents, restricted cash, and restricted cash equivalents at beginning of the period39,10045,438
Cash, cash equivalents, restricted cash, and restricted cash equivalents at end of the period$31,102$30,071
Reconciliation of cash, cash equivalents, restricted cash, and restricted cash equivalents to the condensed consolidated balance sheets
Cash and cash equivalents$23,426$28,750
Restricted cash and restricted cash equivalents, included in prepaid expenses and other current assets34071
Restricted cash and restricted cash equivalents, included in other assets7,3361,250
Total cash, cash equivalents, restricted cash, and restricted cash equivalents$31,102$30,071
Supplemental cash flow data
Cash paid for income taxes, net$541$448
", + "path": "tables/table-3 Cash Flow Statement.html", + "metadata": { + "length": 4015, + "summary": "table-4\nThe table shows cash flows for three months ended March 31, 2026 and 2025, with net cash provided by operating activities of $32,226 and $24,026, respectively.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "file_path": "tables/table-3 Cash Flow Statement.html", + "keywords": [ + "cash flow", + "operating", + "investing" + ], + "tokens": [] + } + }, + { + "chunk_id": "3589d1a9-74de-5e33-ad3a-63432b61a89d", + "type": "text", + "content": "To supplement our condensed consolidated financial statements, which are prepared and presented in accordance with generally accepted accounting principles in the United States (GAAP), we use the following non-GAAP financial measures: revenue excluding foreign exchange effect, advertising revenue excluding foreign exchange effect, and free cash flow. The presentation of these financial measures is not intended to be considered in isolation or as a substitute for, or superior to, financial information prepared and presented in accordance with GAAP. Investors are cautioned that there are material limitations associated with the use of non-GAAP financial measures as an analytical tool. In addition, these measures may be different from non-GAAP financial measures used by other companies, limiting their usefulness for comparison purposes. We compensate for these limitations by providing specific information regarding the GAAP amounts excluded from these non-GAAP financial measures.\nWe believe these non-GAAP financial measures provide investors with useful supplemental information about the financial performance of our business, enable comparison of financial results between periods where certain items may vary independent of business performance, and allow for greater transparency with respect to key metrics used by management in operating our business.\nOur non-GAAP financial measures are adjusted for the following items:\nForeign exchange effect on revenue. To calculate revenue on a constant currency basis, we translate current period revenue using the prior year's monthly exchange rates for our settlement or billing currencies other than the U.S. dollar, which we believe is a useful metric that facilitates comparison to our historical performance.\nPurchases of property and equipment; Principal payments on finance leases. We subtract both purchases of property and equipment, and principal payments on finance leases in our calculation of free cash flow because we believe that these two items collectively represent the amount of property and equipment we need to procure to support our business, regardless of whether we procure such property or equipment with a finance lease. We believe that this methodology can provide useful supplemental information to help investors better understand underlying trends in our business. Free cash flow is not intended to represent our residual cash flow available for discretionary expenditures.\nFor more information on our non-GAAP financial measures and a reconciliation of GAAP to non-GAAP measures, see the \"Reconciliation of GAAP to Non-GAAP Results\" table in this press release.\nMETA PLATFORMS, INC.\nCONDENSED CONSOLIDATED STATEMENTS OF INCOME\n(In millions, except per share amounts)\n(Unaudited)\n\n[tables/table-1 Q1 Financial Results.html]\n\nMETA PLATFORMS, INC.\nCONDENSED CONSOLIDATED BALANCE SHEETS\n(In millions)\n(Unaudited)\n\n[tables/table-2 Balance Sheet.html]\n\nMETA PLATFORMS, INC.\nCONDENSED CONSOLIDATED STATEMENTS OF CASH FLOWS\n(In millions)\n(Unaudited)\n\n[tables/table-3 Cash Flow Statement.html]", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Non-GAAP Financial Measures", + "metadata": { + "length": 3101, + "summary": "The passage discusses Meta Platforms' use of non-GAAP financial measures, including revenue excluding foreign exchange effect, advertising revenue excluding foreign exchange effect, and free cash flow. It explains that these measures supplement GAAP financials, provide useful comparisons, and are adjusted for foreign exchange effects and capital expenditures. The text also references condensed consolidated financial statements (income, balance sheet, cash flows) for the three months ended March 31, 2026, but the actual table data is not provided in the input.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "tokens": [ + "To", + "supplement", + "condensed", + "consolidated", + "financial", + "statements", + "prepared", + "presented", + "accordance", + "generally", + "accepted", + "accounting", + "principles", + "United", + "States", + "GAAP", + "measures", + "revenue", + "excluding", + "foreign", + "exchange", + "effect", + "advertising", + "free", + "cash", + "flow", + "The", + "presentation", + "intended", + "considered", + "isolation", + "substitute", + "superior", + "information", + "Investors", + "cautioned", + "material", + "limitations", + "analytical", + "tool", + "In", + "addition", + "companies", + "limiting", + "usefulness", + "comparison", + "purposes", + "We", + "compensate", + "providing", + "specific", + "amounts", + "excluded", + "provide", + "investors", + "supplemental", + "performance", + "business", + "enable", + "results", + "periods", + "items", + "vary", + "independent", + "greater", + "transparency", + "respect", + "key", + "metrics", + "management", + "operating", + "Our", + "adjusted", + "Foreign", + "calculate", + "constant", + "currency", + "basis", + "translate", + "current", + "period", + "prior", + "year", + "monthly", + "rates", + "settlement", + "billing", + "currencies", + "dollar", + "metric", + "facilitates", + "historical", + "Purchases", + "property", + "equipment", + "Principal", + "payments", + "finance", + "leases", + "subtract", + "purchases", + "principal", + "calculation", + "collectively", + "represent", + "amount", + "procure", + "support", + "lease", + "methodology", + "understand", + "underlying", + "trends", + "Free", + "residual", + "discretionary", + "expenditures", + "For", + "reconciliation", + "Reconciliation", + "Non", + "Results", + "table", + "press", + "release", + "META", + "PLATFORMS", + "INC", + "CONDENSED", + "CONSOLIDATED", + "STATEMENTS", + "OF", + "INCOME", + "millions", + "share", + "Unaudited", + "BALANCE", + "SHEETS", + "CASH", + "FLOWS" + ], + "keywords": [ + "Non-GAAP", + "Revenue", + "Free Cash Flow" + ], + "connect_to": [ + { + "target": "4c634a6a-e022-5e8e-bc63-d804248bb980", + "relation": "embeds", + "ref": "[tables/table-1 Q1 Financial Results.html]", + "position": { + "start": 2771, + "end": 2813 + } + }, + { + "target": "baafc521-c953-5d9e-aedf-6a867cbe6024", + "relation": "embeds", + "ref": "[tables/table-2 Balance Sheet.html]", + "position": { + "start": 2901, + "end": 2936 + } + }, + { + "target": "6d7be585-ee73-50f0-b3e4-e2fc7d00fe43", + "relation": "embeds", + "ref": "[tables/table-3 Cash Flow Statement.html]", + "position": { + "start": 3034, + "end": 3075 + } + } + ] + } + }, + { + "chunk_id": "4a694a69-fe63-509c-90be-f6e7f02e4334", + "type": "table", + "content": "
Three Months Ended March 31,
20262025
Revenue:
Advertising$ 55,024$ 41,392
Other revenue885510
Family of Apps55,90941,902
Reality Labs402412
Total revenue$ 56,311$ 42,314
Income (loss) from operations:
Family of Apps$ 26,900$ 21,765
Reality Labs(4,028)(4,210)
Total income from operations$ 22,872$ 17,555
", + "path": "tables/table-4 Meta Q1 Revenue.html", + "metadata": { + "length": 746, + "summary": "table-5\nMeta's Q1 2026 total revenue was $56,311M, up from $42,314M in 2025. Family of Apps income rose to $26,900M, while Reality Labs loss narrowed to $4,028M.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "file_path": "tables/table-4 Meta Q1 Revenue.html", + "keywords": [ + "Revenue", + "Income", + "Meta" + ], + "tokens": [] + } + }, + { + "chunk_id": "f404c94a-de9a-5cf7-b2e2-a029f3a9826d", + "type": "text", + "content": "We report our financial results for our two reportable segments: Family of Apps (FoA) and Reality Labs (RL). FoA includes Facebook, Instagram, Messenger, WhatsApp, and other services. RL includes our virtual and augmented reality related consumer hardware, software, and content.\nThe following table sets forth our segment information of revenue and income (loss) from operations:\nSegment Information\n(In millions)\n(Unaudited)\n\n[tables/table-4 Meta Q1 Revenue.html]", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Segment Results", + "metadata": { + "length": 473, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "tokens": [ + "We", + "report", + "financial", + "results", + "reportable", + "segments", + "Family", + "Apps", + "FoA", + "Reality", + "Labs", + "RL", + "includes", + "Facebook", + "Instagram", + "Messenger", + "WhatsApp", + "services", + "virtual", + "augmented", + "reality", + "related", + "consumer", + "hardware", + "software", + "content", + "The", + "table", + "sets", + "segment", + "information", + "revenue", + "income", + "loss", + "operations", + "Segment", + "Information", + "In", + "millions", + "Unaudited" + ], + "keywords": [], + "connect_to": [ + { + "target": "4a694a69-fe63-509c-90be-f6e7f02e4334", + "relation": "embeds", + "ref": "[tables/table-4 Meta Q1 Revenue.html]", + "position": { + "start": 428, + "end": 465 + } + } + ] + } + }, + { + "chunk_id": "af92081c-54eb-5b97-9d21-733bfd6fbe1d", + "type": "table", + "content": "
Three Months Ended March 31,
20262025
GAAP revenue$ 56,311$ 42,314
Foreign exchange effect on 2026 revenue using 2025 rates(1,749)
Revenue excluding foreign exchange effect$ 54,562
GAAP revenue year-over-year change %33 %
Revenue excluding foreign exchange effect year-over-year change %29 %
GAAP advertising revenue$ 55,024$ 41,392
Foreign exchange effect on 2026 advertising revenue using 2025 rates(1,734)
Advertising revenue excluding foreign exchange effect$ 53,290
GAAP advertising revenue year-over-year change %33 %
Advertising revenue excluding foreign exchange effect year-over-year change %29 %
Net cash provided by operating activities$ 32,226$ 24,026
Purchases of property and equipment(18,997)(12,941)
Principal payments on finance leases(843)(751)
Free cash flow$ 12,386$ 10,334
", + "path": "tables/table-5 Revenue & Cash Flow.html", + "metadata": { + "length": 1375, + "summary": "table-6\nGAAP revenue rose 33% to $56,311M in Q1 2026; excluding FX, growth was 29%. Advertising revenue similarly grew 33% (29% ex-FX). Free cash flow was $12,386M, up from $10,334M.", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "file_path": "tables/table-5 Revenue & Cash Flow.html", + "keywords": [ + "Revenue", + "Foreign Exchange", + "Free Cash Flow" + ], + "tokens": [] + } + }, + { + "chunk_id": "868184d1-5c3b-594f-8571-01637edc1de0", + "type": "text", + "content": "(In millions, except percentages)\n(Unaudited)\n\n[tables/table-5 Revenue & Cash Flow.html]", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Reconciliation of GAAP to Non-GAAP Results", + "metadata": { + "length": 92, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Meta Reports First Quarter 2026 Results", + "tokens": [ + "In", + "millions", + "percentages", + "Unaudited" + ], + "keywords": [], + "connect_to": [ + { + "target": "af92081c-54eb-5b97-9d21-733bfd6fbe1d", + "relation": "embeds", + "ref": "[tables/table-5 Revenue & Cash Flow.html]", + "position": { + "start": 47, + "end": 88 + } + } + ] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/doc_nav.json b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/doc_nav.json new file mode 100644 index 000000000..b9961879e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/doc_nav.json @@ -0,0 +1,131 @@ +{ + "version": "1.0", + "file_name": "Meta Q1 2026 Exhibit 99.1.pdf", + "stats": { + "total_chunks": 17, + "text_chunks": 11, + "image_chunks": 0, + "table_chunks": 6, + "max_depth": 2 + }, + "sections": [ + { + "title": "Meta Reports First Quarter 2026 Results", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results", + "level": 1, + "summary": "MENLO PARK, Calif. – April 29, 2026 – Meta Platforms, Inc. (Nasdaq: META) today reported financial results for the quarter ended March 31, 2026. \"We had a milestone quarter with strong momentum across our apps and the release of our first model from Meta Superintelligence Labs,\" said Mark Zuckerberg, Meta founder and CEO. \"We're on track to deliver personal superintelligence to billions of people.\" First Quarter 2026 Financial Highlights [tables/table-0 Q1 Financial Results.html] NM — not meaningful (1) Includes an \\$8.03 billion income tax benefit recognized in the first quarter of 2026, which partially offsets the \\$15.93 billion non-cash tax charge recorded in the third quarter of 2025 upon enactment of the One Big Beautiful Bill Act. This benefit is the result of U.S. Treasury Notice 2026-7, which addressed the Corporate Alternative Minimum Tax treatment of previously capitalized U.S. research and development costs. Excluding this tax benefit, our effective tax rate would have been 37 percentage points higher and our diluted earnings per share (EPS) would have been \\$3.13 lower.", + "chunk_count": 11, + "children": [ + { + "title": "First Quarter 2026 Operational and Other Financial Highlights", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/First Quarter 2026 Operational and Other Financial Highlights", + "level": 2, + "summary": "- Family daily active people (DAP) – DAP was 3.56 billion on average for March 2026, an increase of 4% year-over-year. The slight decline in DAP on a quarter-over-quarter basis was driven by internet disruptions in Iran, as well as a restriction on access to WhatsApp in Russia. - Ad impressions – Ad impressions delivered across our Family of Apps increased by 19% year-over-year. • Average price per ad – Average price per ad increased by 12% year-over-year. • Revenue – Revenue was \\$56.31 billion, an increase of 33% year-over-year. Revenue on a constant currency basis would have increased by 29% year-over-year. - Costs and expenses – Total costs and expenses were \\$33.44 billion, an increase of 35% year-over-year. - Capital expenditures – Capital expenditures, including principal payments on finance leases, were \\$19.84 billion. - Capital return program – Dividend and dividend equivalent payments were \\$1.35 billion. - Cash, cash equivalents, and marketable securities – Cash, cash equivalents, and marketable securities were \\$81.18 billion as of March 31, 2026. - Cash flow – Cash flow from operating activities was \\$32.23 billion, and free cash flow was \\$12.39 billion. $^{(1)}$ - Headcount – Headcount was 77,986 as of March 31, 2026, an increase of $1\\%$ year-over-year.", + "chunk_count": 1, + "children": [] + }, + { + "title": "CFO Outlook Commentary", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/CFO Outlook Commentary", + "level": 2, + "summary": "We expect second quarter 2026 total revenue to be in the range of \\$58-61 billion. Our guidance assumes foreign currency is an approximately 2% tailwind to year-over-year total revenue growth, based on current exchange rates. We expect full year 2026 total expenses to be in the range of \\$162-169 billion, unchanged from our prior outlook. We continue to expect to deliver operating income this year that is above 2025 operating income. We anticipate 2026 capital expenditures, including principal payments on finance leases, to be in the range of \\$125-145 billion, increased from our prior range of \\$115-135 billion. This reflects our expectations for higher component pricing this year and, to a lesser extent, additional data center costs to support future year capacity. Absent any changes to our tax landscape, we expect our tax rate for the remaining quarters of 2026 to be between 13-16%. Lastly, we continue to monitor active legal and regulatory matters, including headwinds in the EU and the U.S. that could significantly impact our business and financial results. For example, we continue to see scrutiny on youth-related issues and have additional trials scheduled for this year in the U.S., which may ultimately result in a material loss.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Webcast and Conference Call Information", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Webcast and Conference Call Information", + "level": 2, + "summary": "Meta will host a conference call to discuss its results at 2:30 p.m. PT / 5:30 p.m. ET today. The live webcast of the call can be accessed at the Meta Investor Relations website at investor.atmeta.com, along with the company's earnings press release, financial tables, and slide presentation. Following the call, a replay will be available at the same website. Transcripts of conference calls with publishing equity research analysts held today will also be posted to the investor.atmeta.com website.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Disclosure Information", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Disclosure Information", + "level": 2, + "summary": "Meta uses the investor.atmeta.com and meta.com/news websites as well as Mark Zuckerberg's Facebook profile (facebook.com/zuck), Instagram account (instagram.com/zuck) and Threads profile (threads.net/zuck) as means of disclosing material non-public information and for complying with its disclosure obligations under Regulation FD.", + "chunk_count": 1, + "children": [] + }, + { + "title": "About Meta", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/About Meta", + "level": 2, + "summary": "Meta is building the future of human connection, powered by artificial intelligence and immersive technologies. When Facebook launched in 2004, it changed the way people connect. Apps like Messenger, Instagram, and WhatsApp further empowered billions around the world. Now, Meta is moving beyond 2D screens toward experiences that foster deeper connections and unlock new possibilities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Contacts", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Contacts", + "level": 2, + "summary": "Investors: Kenneth Dorell investor@meta.com / investor.atmeta.com Press: Matt Tye press@meta.com / meta.com/news", + "chunk_count": 1, + "children": [] + }, + { + "title": "Forward-Looking Statements", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Forward-Looking Statements", + "level": 2, + "summary": "This press release contains forward-looking statements about future business plans and expectations, which may differ materially from actual results due to various factors including macroeconomic conditions, user engagement, advertising revenue, data dependency, third-party policies, new products like AI and Reality Labs, competition, government actions, litigation, privacy regulations, acquisitions, security breaches, and market conditions. These risks are detailed in the company's Annual Report on Form 10-K filed with the SEC on January 29, 2026, and will be updated in the Quarterly Report for March 31, 2026. The release is dated April 29, 2026, and the company undertakes no obligation to update these statements. Limitations of community metrics are discussed in the most recent SEC filings.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-GAAP Financial Measures", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Non-GAAP Financial Measures", + "level": 2, + "summary": "The passage discusses Meta Platforms' use of non-GAAP financial measures, including revenue excluding foreign exchange effect, advertising revenue excluding foreign exchange effect, and free cash flow. It explains that these measures supplement GAAP financials, provide useful comparisons, and are adjusted for foreign exchange effects and capital expenditures. The text also references condensed consolidated financial statements (income, balance sheet, cash flows) for the three months ended March 31, 2026, but the actual table data is not provided in the input.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Segment Results", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Segment Results", + "level": 2, + "summary": "We report our financial results for our two reportable segments: Family of Apps (FoA) and Reality Labs (RL). FoA includes Facebook, Instagram, Messenger, WhatsApp, and other services. RL includes our virtual and augmented reality related consumer hardware, software, and content. The following table sets forth our segment information of revenue and income (loss) from operations: Segment Information (In millions) (Unaudited) [tables/table-4 Meta Q1 Revenue.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Reconciliation of GAAP to Non-GAAP Results", + "path": "Meta-03-31-2026-Exhibit-99-1_final.pdf/Meta Reports First Quarter 2026 Results/Reconciliation of GAAP to Non-GAAP Results", + "level": 2, + "summary": "(In millions, except percentages) (Unaudited) [tables/table-5 Revenue & Cash Flow.html]", + "chunk_count": 1, + "children": [] + } + ] + } + ], + "resources": { + "images": [], + "tables": [ + { + "path": "tables/table-0 Q1 Financial Results.html", + "summary": "table-1 Microsoft's Q1 2026 revenue rose 33% to $56.3B, net income up 61% to $26.8B, and diluted EPS up 62% to $10.44, with operating margin steady at 41%." + }, + { + "path": "tables/table-1 Q1 Financial Results.html", + "summary": "table-2 The table shows Q1 financial data for 2026 vs 2025. Revenue rose from $42,314 to $56,311. Net income increased from $16,644 to $26,773. Basic EPS grew from $6.59 to $10.57." + }, + { + "path": "tables/table-2 Balance Sheet.html", + "summary": "table-3 Condensed balance sheet as of March 31, 2026 and December 31, 2025, showing total assets of $395,250 and $366,021, with liabilities and stockholders' equity." + }, + { + "path": "tables/table-3 Cash Flow Statement.html", + "summary": "table-4 The table shows cash flows for three months ended March 31, 2026 and 2025, with net cash provided by operating activities of $32,226 and $24,026, respectively." + }, + { + "path": "tables/table-4 Meta Q1 Revenue.html", + "summary": "table-5 Meta's Q1 2026 total revenue was $56,311M, up from $42,314M in 2025. Family of Apps income rose to $26,900M, while Reality Labs loss narrowed to $4,028M." + }, + { + "path": "tables/table-5 Revenue & Cash Flow.html", + "summary": "table-6 GAAP revenue rose 33% to $56,311M in Q1 2026; excluding FX, growth was 29%. Advertising revenue similarly grew 33% (29% ex-FX). Free cash flow was $12,386M, up from $10,334M." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/full.md b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/full.md new file mode 100644 index 000000000..85de9c016 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/full.md @@ -0,0 +1,128 @@ +## Meta Reports First Quarter 2026 Results + +MENLO PARK, Calif. – April 29, 2026 – Meta Platforms, Inc. (Nasdaq: META) today reported financial results for the quarter ended March 31, 2026. + +"We had a milestone quarter with strong momentum across our apps and the release of our first model from Meta Superintelligence Labs," said Mark Zuckerberg, Meta founder and CEO. "We're on track to deliver personal superintelligence to billions of people." + +First Quarter 2026 Financial Highlights + +
Three Months Ended March 31,
In millions, except percentages and per share amounts20262025% Change
Revenue$ 56,311$ 42,31433 %
Costs and expenses33,43924,75935 %
Income from operations$ 22,872$ 17,55530 %
Operating margin41 %41 %
Provision (benefit) for income taxes(1)$ (5,021)$ 1,738NM
Effective tax rate(1)(23)%9 %
Net income$ 26,773$ 16,64461 %
Diluted earnings per share (EPS)(1)$ 10.44$ 6.4362 %
+ +NM — not meaningful +(1) Includes an \$8.03 billion income tax benefit recognized in the first quarter of 2026, which partially offsets the \$15.93 billion non-cash tax charge recorded in the third quarter of 2025 upon enactment of the One Big Beautiful Bill Act. This benefit is the result of U.S. Treasury Notice 2026-7, which addressed the Corporate Alternative Minimum Tax treatment of previously capitalized U.S. research and development costs. Excluding this tax benefit, our effective tax rate would have been 37 percentage points higher and our diluted earnings per share (EPS) would have been \$3.13 lower. + +## First Quarter 2026 Operational and Other Financial Highlights + +- Family daily active people (DAP) – DAP was 3.56 billion on average for March 2026, an increase of 4% year-over-year. The slight decline in DAP on a quarter-over-quarter basis was driven by internet disruptions in Iran, as well as a restriction on access to WhatsApp in Russia. +- Ad impressions – Ad impressions delivered across our Family of Apps increased by 19% year-over-year. +• Average price per ad – Average price per ad increased by 12% year-over-year. +• Revenue – Revenue was \$56.31 billion, an increase of 33% year-over-year. Revenue on a constant currency basis would have increased by 29% year-over-year. +- Costs and expenses – Total costs and expenses were \$33.44 billion, an increase of 35% year-over-year. +- Capital expenditures – Capital expenditures, including principal payments on finance leases, were \$19.84 billion. +- Capital return program – Dividend and dividend equivalent payments were \$1.35 billion. +- Cash, cash equivalents, and marketable securities – Cash, cash equivalents, and marketable securities were \$81.18 billion as of March 31, 2026. +- Cash flow – Cash flow from operating activities was \$32.23 billion, and free cash flow was \$12.39 billion. $^{(1)}$ +- Headcount – Headcount was 77,986 as of March 31, 2026, an increase of $1\%$ year-over-year. + +## CFO Outlook Commentary + +We expect second quarter 2026 total revenue to be in the range of \$58-61 billion. Our guidance assumes foreign currency is an approximately 2% tailwind to year-over-year total revenue growth, based on current exchange rates. + +We expect full year 2026 total expenses to be in the range of \$162-169 billion, unchanged from our prior outlook. + +We continue to expect to deliver operating income this year that is above 2025 operating income. + +We anticipate 2026 capital expenditures, including principal payments on finance leases, to be in the range of \$125-145 billion, increased from our prior range of \$115-135 billion. This reflects our expectations for higher component pricing this year and, to a lesser extent, additional data center costs to support future year capacity. + +Absent any changes to our tax landscape, we expect our tax rate for the remaining quarters of 2026 to be between 13-16%. + +Lastly, we continue to monitor active legal and regulatory matters, including headwinds in the EU and the U.S. that could significantly impact our business and financial results. For example, we continue to see scrutiny on youth-related issues and have additional trials scheduled for this year in the U.S., which may ultimately result in a material loss. + +## Webcast and Conference Call Information + +Meta will host a conference call to discuss its results at 2:30 p.m. PT / 5:30 p.m. ET today. The live webcast of the call can be accessed at the Meta Investor Relations website at investor.atmeta.com, along with the company's earnings press release, financial tables, and slide presentation. + +Following the call, a replay will be available at the same website. Transcripts of conference calls with publishing equity research analysts held today will also be posted to the investor.atmeta.com website. + +## Disclosure Information + +Meta uses the investor.atmeta.com and meta.com/news websites as well as Mark Zuckerberg's Facebook profile (facebook.com/zuck), Instagram account (instagram.com/zuck) and Threads profile (threads.net/zuck) as means of disclosing material non-public information and for complying with its disclosure obligations under Regulation FD. + +## About Meta + +Meta is building the future of human connection, powered by artificial intelligence and immersive technologies. When Facebook launched in 2004, it changed the way people connect. Apps like Messenger, Instagram, and WhatsApp further empowered billions around the world. Now, Meta is moving beyond 2D screens toward experiences that foster deeper connections and unlock new possibilities. + +## Contacts + +Investors: + +Kenneth Dorell + +investor@meta.com / investor.atmeta.com + +Press: + +Matt Tye + +press@meta.com / meta.com/news + +## Forward-Looking Statements + +This press release contains forward-looking statements regarding our future business plans and expectations. These forward-looking statements are only predictions and may differ materially from actual results due to a variety of factors including: the impact of macroeconomic conditions on our business and financial results, including as a result of geopolitical events; our ability to retain or increase users and engagement levels; our reliance on advertising revenue; our dependency on data signals and mobile operating systems, networks, and standards that we do not control; changes to the content or application of third-party policies that impact our advertising practices; risks associated with new products and changes to existing products as well as other new business initiatives, including our artificial intelligence initiatives and Reality Labs efforts; our emphasis on community growth and engagement and the user experience over short-term financial results; maintaining and enhancing our brand and reputation; our ongoing privacy, safety, security, and content and advertising review and enforcement efforts; competition; risks associated with government actions that could restrict access to our products or impair our ability to sell advertising in certain countries; litigation and government inquiries; privacy, legislative, and regulatory concerns or developments; risks associated with acquisitions; security breaches; our ability to manage our scale and geographically-dispersed operations; and market conditions or other factors affecting capital return to stockholders. These and other potential risks and uncertainties that could cause actual results to differ from the results predicted are more fully detailed under the caption "Risk Factors" in our Annual Report on Form 10-K filed with the SEC on January 29, 2026, which is available on our Investor Relations website at investor.atmeta.com and on the SEC website at www.sec.gov. Additional information will also be set forth in our Quarterly Report on Form 10-Q for the quarter ended March 31, 2026. In addition, please note that the date of this press release is April 29, 2026, and any forward-looking statements contained herein are based on assumptions that we believe to be reasonable as of this date. We undertake no obligation to update these statements as a result of new information or future events. + +For a discussion of limitations in the measurement of certain of our community metrics, see the section entitled "Limitations of Key Metrics and Other Data" in our most recent quarterly or annual report filed with the SEC. + +## Non-GAAP Financial Measures + +To supplement our condensed consolidated financial statements, which are prepared and presented in accordance with generally accepted accounting principles in the United States (GAAP), we use the following non-GAAP financial measures: revenue excluding foreign exchange effect, advertising revenue excluding foreign exchange effect, and free cash flow. The presentation of these financial measures is not intended to be considered in isolation or as a substitute for, or superior to, financial information prepared and presented in accordance with GAAP. Investors are cautioned that there are material limitations associated with the use of non-GAAP financial measures as an analytical tool. In addition, these measures may be different from non-GAAP financial measures used by other companies, limiting their usefulness for comparison purposes. We compensate for these limitations by providing specific information regarding the GAAP amounts excluded from these non-GAAP financial measures. + +We believe these non-GAAP financial measures provide investors with useful supplemental information about the financial performance of our business, enable comparison of financial results between periods where certain items may vary independent of business performance, and allow for greater transparency with respect to key metrics used by management in operating our business. + +Our non-GAAP financial measures are adjusted for the following items: + +Foreign exchange effect on revenue. To calculate revenue on a constant currency basis, we translate current period revenue using the prior year's monthly exchange rates for our settlement or billing currencies other than the U.S. dollar, which we believe is a useful metric that facilitates comparison to our historical performance. + +Purchases of property and equipment; Principal payments on finance leases. We subtract both purchases of property and equipment, and principal payments on finance leases in our calculation of free cash flow because we believe that these two items collectively represent the amount of property and equipment we need to procure to support our business, regardless of whether we procure such property or equipment with a finance lease. We believe that this methodology can provide useful supplemental information to help investors better understand underlying trends in our business. Free cash flow is not intended to represent our residual cash flow available for discretionary expenditures. + +For more information on our non-GAAP financial measures and a reconciliation of GAAP to non-GAAP measures, see the "Reconciliation of GAAP to Non-GAAP Results" table in this press release. + +META PLATFORMS, INC. +CONDENSED CONSOLIDATED STATEMENTS OF INCOME +(In millions, except per share amounts) +(Unaudited) + +
Three Months Ended March 31,
20262025
Revenue$ 56,311$ 42,314
Costs and expenses:
Cost of revenue10,2187,572
Research and development17,69912,150
Marketing and sales2,9082,757
General and administrative2,6142,280
Total costs and expenses33,43924,759
Income from operations22,87217,555
Interest and other income (expense), net(1,120)827
Income before income taxes21,75218,382
Provision (benefit) for income taxes(5,021)1,738
Net income$ 26,773$ 16,644
Earnings per share:
Basic$ 10.57$ 6.59
Diluted$ 10.44$ 6.43
Weighted-average shares used to compute earnings per share:
Basic2,5342,527
Diluted2,5642,590
+ +META PLATFORMS, INC. +CONDENSED CONSOLIDATED BALANCE SHEETS +(In millions) +(Unaudited) + +
March 31, 2026December 31, 2025
Assets
Current assets:
Cash and cash equivalents$ 23,426$ 35,873
Marketable securities57,75445,719
Accounts receivable, net17,47019,769
Prepaid expenses and other current assets11,1157,361
Total current assets109,765108,722
Non-marketable equity investments28,41027,524
Property and equipment, net194,776176,400
Operating lease right-of-use assets23,26820,404
Goodwill24,74824,534
Other assets14,2838,437
Total assets$ 395,250$ 366,021
Liabilities and stockholders' equity
Current liabilities:
Accounts payable$ 13,326$ 8,894
Operating lease liabilities, current2,4142,213
Accrued expenses and other current liabilities31,01330,729
Total current liabilities46,75341,836
Operating lease liabilities, non-current25,60722,940
Long-term debt58,74858,744
Long-term income taxes16,84921,005
Other liabilities3,6124,253
Total liabilities151,569148,778
Commitments and contingencies
Stockholders' equity:
Common stock and additional paid-in capital99,33795,793
Accumulated other comprehensive income (loss)(303)271
Retained earnings144,647121,179
Total stockholders' equity243,681217,243
Total liabilities and stockholders' equity$ 395,250$ 366,021
+ +META PLATFORMS, INC. +CONDENSED CONSOLIDATED STATEMENTS OF CASH FLOWS +(In millions) +(Unaudited) + +
Three Months Ended March 31,
20262025
Cash flows from operating activities
Net income$26,773$16,644
Adjustments to reconcile net income to net cash provided by operating activities:
Depreciation and amortization5,9993,900
Share-based compensation6,0324,147
Deferred income taxes123(993)
Unrealized (gain) loss on equity investments1,075(135)
Other(17)(96)
Changes in assets and liabilities:
Accounts receivable2,1282,804
Prepaid expenses and other current assets(2,424)360
Other assets(1,082)(52)
Accounts payable(937)(1,034)
Accrued expenses and other current liabilities(271)(2,231)
Other liabilities(5,173)712
Net cash provided by operating activities32,22624,026
Cash flows from investing activities
Purchases of property and equipment(18,997)(12,941)
Purchases of marketable securities(32,978)(11,763)
Sales and maturities of marketable securities19,1764,784
Purchases of non-marketable equity investments(544)(100)
Payments for held-for-sale assets(118)
Acquisitions of businesses and intangible assets(372)(1)
Other investing activities15511
Net cash used in investing activities(33,678)(20,010)
Cash flows from financing activities
Taxes paid related to net share settlement of equity awards(4,423)(4,883)
Repurchases of Class A common stock(12,754)
Payments for dividends and dividend equivalents(1,346)(1,329)
Principal payments on finance leases(843)(751)
Other financing activities59222
Net cash used in financing activities(6,553)(19,495)
Effect of exchange rate changes on cash, cash equivalents, restricted cash, and restricted cash equivalents7112
Net decrease in cash, cash equivalents, restricted cash, and restricted cash equivalents(7,998)(15,367)
Cash, cash equivalents, restricted cash, and restricted cash equivalents at beginning of the period39,10045,438
Cash, cash equivalents, restricted cash, and restricted cash equivalents at end of the period$31,102$30,071
Reconciliation of cash, cash equivalents, restricted cash, and restricted cash equivalents to the condensed consolidated balance sheets
Cash and cash equivalents$23,426$28,750
Restricted cash and restricted cash equivalents, included in prepaid expenses and other current assets34071
Restricted cash and restricted cash equivalents, included in other assets7,3361,250
Total cash, cash equivalents, restricted cash, and restricted cash equivalents$31,102$30,071
Supplemental cash flow data
Cash paid for income taxes, net$541$448
+ +## Segment Results + +We report our financial results for our two reportable segments: Family of Apps (FoA) and Reality Labs (RL). FoA includes Facebook, Instagram, Messenger, WhatsApp, and other services. RL includes our virtual and augmented reality related consumer hardware, software, and content. + +The following table sets forth our segment information of revenue and income (loss) from operations: + +Segment Information +(In millions) +(Unaudited) + +
Three Months Ended March 31,
20262025
Revenue:
Advertising$ 55,024$ 41,392
Other revenue885510
Family of Apps55,90941,902
Reality Labs402412
Total revenue$ 56,311$ 42,314
Income (loss) from operations:
Family of Apps$ 26,900$ 21,765
Reality Labs(4,028)(4,210)
Total income from operations$ 22,872$ 17,555
+ +## Reconciliation of GAAP to Non-GAAP Results + +(In millions, except percentages) + +(Unaudited) + +
Three Months Ended March 31,
20262025
GAAP revenue$ 56,311$ 42,314
Foreign exchange effect on 2026 revenue using 2025 rates(1,749)
Revenue excluding foreign exchange effect$ 54,562
GAAP revenue year-over-year change %33 %
Revenue excluding foreign exchange effect year-over-year change %29 %
GAAP advertising revenue$ 55,024$ 41,392
Foreign exchange effect on 2026 advertising revenue using 2025 rates(1,734)
Advertising revenue excluding foreign exchange effect$ 53,290
GAAP advertising revenue year-over-year change %33 %
Advertising revenue excluding foreign exchange effect year-over-year change %29 %
Net cash provided by operating activities$ 32,226$ 24,026
Purchases of property and equipment(18,997)(12,941)
Principal payments on finance leases(843)(751)
Free cash flow$ 12,386$ 10,334
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/manifest.json b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/manifest.json new file mode 100644 index 000000000..7803c129b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/manifest.json @@ -0,0 +1,63 @@ +{ + "version": "2.0", + "job_id": "job_d669d759ca88", + "data_id": null, + "source_file_name": "Meta-03-31-2026-Exhibit-99-1_final.pdf", + "processing_date": "2026-06-16T03:32:14.805330Z", + "processing": { + "page_count": 9, + "billing_status": "charged", + "cost": { + "micro_dollars": 13500, + "credits": 0.0135 + }, + "timing": { + "started_at": "2026-06-16T03:31:40.786174+00:00", + "completed_at": "2026-06-16T03:32:14.798647+00:00", + "duration_ms": 34012 + }, + "stages": { + "timing_ms": { + "document.profile": 3561, + "pdf.extract.standard": 24962, + "toc.detect_range_llm": 1075, + "md.detect_toc": 1089, + "heading.hierarchy_llm_call": 1847, + "heading.hierarchy_llm": 1854, + "md.predict_headings": 1958, + "md.deferred_summaries": 2335, + "md.build_dataframe": 1, + "pdf.parse_md": 5430, + "document.cleanup_unreferenced_images": 0, + "document.compress_images": 0, + "worker.parse.document": 33959 + }, + "token_usage": { + "prompt_tokens": 3363, + "completion_tokens": 170, + "total_tokens": 3533 + } + } + }, + "statistics": { + "total_chunks": 17, + "text_chunks": 11, + "image_chunks": 0, + "table_chunks": 6, + "total_pages": null + }, + "HIERARCHY": { + "Meta Reports First Quarter 2026 Results": { + "First Quarter 2026 Operational and Other Financial Highlights": {}, + "CFO Outlook Commentary": {}, + "Webcast and Conference Call Information": {}, + "Disclosure Information": {}, + "About Meta": {}, + "Contacts": {}, + "Forward-Looking Statements": {}, + "Non-GAAP Financial Measures": {}, + "Segment Results": {}, + "Reconciliation of GAAP to Non-GAAP Results": {} + } + } +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-0 Q1 Financial Results.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-0 Q1 Financial Results.html new file mode 100644 index 000000000..ce0803ba4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-0 Q1 Financial Results.html @@ -0,0 +1 @@ +
Three Months Ended March 31,
In millions, except percentages and per share amounts20262025% Change
Revenue$ 56,311$ 42,31433 %
Costs and expenses33,43924,75935 %
Income from operations$ 22,872$ 17,55530 %
Operating margin41 %41 %
Provision (benefit) for income taxes(1)$ (5,021)$ 1,738NM
Effective tax rate(1)(23)%9 %
Net income$ 26,773$ 16,64461 %
Diluted earnings per share (EPS)(1)$ 10.44$ 6.4362 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-1 Q1 Financial Results.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-1 Q1 Financial Results.html new file mode 100644 index 000000000..71871c423 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-1 Q1 Financial Results.html @@ -0,0 +1 @@ +
Three Months Ended March 31,
20262025
Revenue$ 56,311$ 42,314
Costs and expenses:
Cost of revenue10,2187,572
Research and development17,69912,150
Marketing and sales2,9082,757
General and administrative2,6142,280
Total costs and expenses33,43924,759
Income from operations22,87217,555
Interest and other income (expense), net(1,120)827
Income before income taxes21,75218,382
Provision (benefit) for income taxes(5,021)1,738
Net income$ 26,773$ 16,644
Earnings per share:
Basic$ 10.57$ 6.59
Diluted$ 10.44$ 6.43
Weighted-average shares used to compute earnings per share:
Basic2,5342,527
Diluted2,5642,590
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-2 Balance Sheet.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-2 Balance Sheet.html new file mode 100644 index 000000000..e88951c6a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-2 Balance Sheet.html @@ -0,0 +1 @@ +
March 31, 2026December 31, 2025
Assets
Current assets:
Cash and cash equivalents$ 23,426$ 35,873
Marketable securities57,75445,719
Accounts receivable, net17,47019,769
Prepaid expenses and other current assets11,1157,361
Total current assets109,765108,722
Non-marketable equity investments28,41027,524
Property and equipment, net194,776176,400
Operating lease right-of-use assets23,26820,404
Goodwill24,74824,534
Other assets14,2838,437
Total assets$ 395,250$ 366,021
Liabilities and stockholders' equity
Current liabilities:
Accounts payable$ 13,326$ 8,894
Operating lease liabilities, current2,4142,213
Accrued expenses and other current liabilities31,01330,729
Total current liabilities46,75341,836
Operating lease liabilities, non-current25,60722,940
Long-term debt58,74858,744
Long-term income taxes16,84921,005
Other liabilities3,6124,253
Total liabilities151,569148,778
Commitments and contingencies
Stockholders' equity:
Common stock and additional paid-in capital99,33795,793
Accumulated other comprehensive income (loss)(303)271
Retained earnings144,647121,179
Total stockholders' equity243,681217,243
Total liabilities and stockholders' equity$ 395,250$ 366,021
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-3 Cash Flow Statement.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-3 Cash Flow Statement.html new file mode 100644 index 000000000..793f27a3f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-3 Cash Flow Statement.html @@ -0,0 +1 @@ +
Three Months Ended March 31,
20262025
Cash flows from operating activities
Net income$26,773$16,644
Adjustments to reconcile net income to net cash provided by operating activities:
Depreciation and amortization5,9993,900
Share-based compensation6,0324,147
Deferred income taxes123(993)
Unrealized (gain) loss on equity investments1,075(135)
Other(17)(96)
Changes in assets and liabilities:
Accounts receivable2,1282,804
Prepaid expenses and other current assets(2,424)360
Other assets(1,082)(52)
Accounts payable(937)(1,034)
Accrued expenses and other current liabilities(271)(2,231)
Other liabilities(5,173)712
Net cash provided by operating activities32,22624,026
Cash flows from investing activities
Purchases of property and equipment(18,997)(12,941)
Purchases of marketable securities(32,978)(11,763)
Sales and maturities of marketable securities19,1764,784
Purchases of non-marketable equity investments(544)(100)
Payments for held-for-sale assets(118)
Acquisitions of businesses and intangible assets(372)(1)
Other investing activities15511
Net cash used in investing activities(33,678)(20,010)
Cash flows from financing activities
Taxes paid related to net share settlement of equity awards(4,423)(4,883)
Repurchases of Class A common stock(12,754)
Payments for dividends and dividend equivalents(1,346)(1,329)
Principal payments on finance leases(843)(751)
Other financing activities59222
Net cash used in financing activities(6,553)(19,495)
Effect of exchange rate changes on cash, cash equivalents, restricted cash, and restricted cash equivalents7112
Net decrease in cash, cash equivalents, restricted cash, and restricted cash equivalents(7,998)(15,367)
Cash, cash equivalents, restricted cash, and restricted cash equivalents at beginning of the period39,10045,438
Cash, cash equivalents, restricted cash, and restricted cash equivalents at end of the period$31,102$30,071
Reconciliation of cash, cash equivalents, restricted cash, and restricted cash equivalents to the condensed consolidated balance sheets
Cash and cash equivalents$23,426$28,750
Restricted cash and restricted cash equivalents, included in prepaid expenses and other current assets34071
Restricted cash and restricted cash equivalents, included in other assets7,3361,250
Total cash, cash equivalents, restricted cash, and restricted cash equivalents$31,102$30,071
Supplemental cash flow data
Cash paid for income taxes, net$541$448
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-4 Meta Q1 Revenue.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-4 Meta Q1 Revenue.html new file mode 100644 index 000000000..ee22435c1 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-4 Meta Q1 Revenue.html @@ -0,0 +1 @@ +
Three Months Ended March 31,
20262025
Revenue:
Advertising$ 55,024$ 41,392
Other revenue885510
Family of Apps55,90941,902
Reality Labs402412
Total revenue$ 56,311$ 42,314
Income (loss) from operations:
Family of Apps$ 26,900$ 21,765
Reality Labs(4,028)(4,210)
Total income from operations$ 22,872$ 17,555
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-5 Revenue & Cash Flow.html b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-5 Revenue & Cash Flow.html new file mode 100644 index 000000000..08e59ef80 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-meta-q1-2026-results/tables/table-5 Revenue & Cash Flow.html @@ -0,0 +1 @@ +
Three Months Ended March 31,
20262025
GAAP revenue$ 56,311$ 42,314
Foreign exchange effect on 2026 revenue using 2025 rates(1,749)
Revenue excluding foreign exchange effect$ 54,562
GAAP revenue year-over-year change %33 %
Revenue excluding foreign exchange effect year-over-year change %29 %
GAAP advertising revenue$ 55,024$ 41,392
Foreign exchange effect on 2026 advertising revenue using 2025 rates(1,734)
Advertising revenue excluding foreign exchange effect$ 53,290
GAAP advertising revenue year-over-year change %33 %
Advertising revenue excluding foreign exchange effect year-over-year change %29 %
Net cash provided by operating activities$ 32,226$ 24,026
Purchases of property and equipment(18,997)(12,941)
Principal payments on finance leases(843)(751)
Free cash flow$ 12,386$ 10,334
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/chunks.json b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/chunks.json new file mode 100644 index 000000000..1cad054b6 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/chunks.json @@ -0,0 +1,14047 @@ +{ + "chunks": [ + { + "chunk_id": "d55ac241-0983-5356-98e6-5388e5b473d0", + "type": "image", + "content": "\nMicrosoft Annual Report 2025 Cover\nPrimary Subject: Two colleagues collaborating at a desk with a laptop.\nVisible Text: \"Annual Report 2025\", Microsoft Logo.\n[images/image-1 Annual Report 2025.png]\n", + "path": "images/image-1 Annual Report 2025.png", + "metadata": { + "length": 199, + "summary": "image-1\nMicrosoft Annual Report 2025 Cover\nPrimary Subject: Two colleagues collaborating at a desk with a laptop.\nVisible Text: \"Annual Report 2025\", Microsoft Logo.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "images/image-1 Annual Report 2025.png", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "9e0d738a-c578-5816-afc1-93e714a7ab5d", + "type": "image", + "content": "\nSarga N.\n[images/image-2 Signature.jpeg]\n", + "path": "images/image-2 Signature.jpeg", + "metadata": { + "length": 42, + "summary": "image-2\nSarga N.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "images/image-2 Signature.jpeg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f0fa5122-814d-55dd-aac9-1a40c56c0a8c", + "type": "table", + "content": "
(In millions)SharesAmountSharesAmountSharesAmount
Year Ended June 30,202520242023
First Quarter7$2,80011$3,56017$4,600
Second Quarter83,50072,800204,600
Third Quarter83,50072,800184,600
Fourth Quarter83,20072,800144,600
Total31$13,00032$11,96069$18,400
", + "path": "tables/table-1 Share Repurchases.html", + "metadata": { + "length": 2520, + "summary": "table-1\nThe table shows quarterly share repurchases and amounts for fiscal years 2023-2025, with totals of 31 shares ($13,000M) in 2025, 32 shares ($11,960M) in 2024, and 69 shares ($18,400M) in 2023.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-1 Share Repurchases.html", + "keywords": [ + "Shares", + "Amount", + "Quarterly" + ], + "tokens": [] + } + }, + { + "chunk_id": "0e66fea7-c5b7-53f8-8471-b407613a66b3", + "type": "table", + "content": "
Declaration DateRecord DatePayment DateDividend
Per Share
Amount
Fiscal Year 2025(In millions)
September 16, 2024November 21, 2024December 12, 2024$0.83$6,170
December 3, 2024February 20, 2025March 13, 20250.836,169
March 11, 2025May 15, 2025June 12, 20250.836,169
June 10, 2025August 21, 2025September 11, 20250.836,170
Total$3.32$24,678
Fiscal Year 2024
September 19, 2023November 16, 2023December 14, 2023$0.75$5,574
November 28, 2023February 15, 2024March 14, 20240.755,573
March 12, 2024May 16, 2024June 13, 20240.755,574
June 12, 2024August 15, 2024September 12, 20240.755,574
Total$3.00$22,295
", + "path": "tables/table-2 Dividend History.html", + "metadata": { + "length": 3219, + "summary": "table-2\nThe table shows quarterly dividends for fiscal years 2025 and 2024, with per-share amounts of $0.83 and $0.75 respectively, totaling $24,678 million and $22,295 million.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-2 Dividend History.html", + "keywords": [ + "Dividend", + "Fiscal Year", + "Per Share" + ], + "tokens": [] + } + }, + { + "chunk_id": "77c3bd2d-5b47-5b5a-929b-e8988bc51db5", + "type": "image", + "content": "\nChart Title: Microsoft Corporation, S&P 500, and NASDAQ Computer Index Performance\nY-Axis: Dollar Value ($0 to $300)\nX-Axis: Dates (6/20 to 6/25)\n\nData Points & Trends:\n- All three indices start at a baseline of $100 on 6/20.\n- Microsoft Corporation (Square marker): Starts at $100, dips slightly on 6/21 (~$135), drops to ~$128 on 6/22, then rises sharply to finish at approximately $250 on 6/25. It shows the strongest overall growth among the three.\n- NASDAQ Computer (Circle marker): Starts at $100, peaks at ~$150 on 6/21, drops to its lowest point (~$118) on 6/22, then recovers strongly to end near $250 on 6/25. Its trajectory closely mirrors Microsoft's in the latter half.\n[images/image-3 Microsoft vs S&P 500 v.jpeg]\n", + "path": "images/image-3 Microsoft vs S&P 500 v.jpeg", + "metadata": { + "length": 729, + "summary": "image-3\nChart Title: Microsoft Corporation, S&P 500, and NASDAQ Computer Index Performance\nY-Axis: Dollar Value ($0 to $300)\nX-Axis: Dates (6/20 to 6/25)\n\nData Points & Trends:\n- All three indices start at a baseline of $100 on 6/20.\n- Microsoft Corporation (Square marker): Starts at $100, dips slightly on 6/21 (~$135), drops to ~$128 on 6/22, then rises sharply to finish at approximately $250 on 6/25. It shows the strongest overall growth among the three.\n- NASDAQ Computer (Circle marker): Starts at $100, peaks at ~$150 on 6/21, drops to its lowest point (~$118) on 6/22, then recovers strongly to end near $250 on 6/25. Its trajectory closely mirrors Microsoft's in the latter half.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "images/image-3 Microsoft vs S&P 500 v.jpeg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "b1e3c716-1628-5ef8-9628-f29f51a19862", + "type": "table", + "content": "
6/206/216/226/236/246/25
Microsoft Corporation100.00134.41128.48172.01227.51255.13
S&P 500100.00140.79125.85150.51187.47215.89
NASDAQ Computer100.00150.44117.59154.73223.97254.97
", + "path": "tables/table-3 Stock Index Data.html", + "metadata": { + "length": 1195, + "summary": "table-3\nThe table shows daily values for Microsoft Corporation, S&P 500, and NASDAQ Computer from June 20 to June 25, with each starting at 100.00 and ending at 255.13, 215.89, and 254.97 respectively.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-3 Stock Index Data.html", + "keywords": [ + "stock prices", + "index values", + "daily data" + ], + "tokens": [] + } + }, + { + "chunk_id": "421aaa5d-a19e-55f8-b19f-9f0d269aa779", + "type": "table", + "content": "
Commercial remaining performance obligationCommercial portion of revenue allocated to remaining performance obligations, which includes unearned revenue and amounts that will be invoiced and recognized as revenue in future periods
Microsoft Cloud revenue and revenue growthRevenue from Microsoft 365 Commercial cloud, Azure and other cloud services, the commercial portion of LinkedIn, and Dynamics 365
Microsoft Cloud gross margin percentageGross margin percentage for our Microsoft Cloud business
", + "path": "tables/table-4 Microsoft Cloud KPIs.html", + "metadata": { + "length": 708, + "summary": "table-4\nTable defines three Microsoft Cloud metrics: Commercial remaining performance obligation, Cloud revenue and growth, and Cloud gross margin percentage.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-4 Microsoft Cloud KPIs.html", + "keywords": [ + "Cloud Revenue", + "Gross Margin", + "Performance Obligation" + ], + "tokens": [] + } + }, + { + "chunk_id": "ca7bdf17-9611-509f-9921-ed11c2b257a0", + "type": "table", + "content": "
Microsoft 365 Commercial cloud revenue growthRevenue from Microsoft 365 Commercial subscriptions, comprising Microsoft 365 Commercial, Enterprise Mobility + Security, the cloud portion of Windows Commercial, the per-user portion of Power BI, Exchange, SharePoint, Microsoft Teams, Microsoft 365 Security and Compliance, and Microsoft 365 Copilot
Microsoft 365 Commercial seat growthThe number of Microsoft 365 Commercial seats at end of period where seats are paid users covered by a Microsoft 365 Commercial subscription
Microsoft 365 Consumer cloud revenue growthRevenue from Microsoft 365 Consumer subscriptions and other consumer services
Microsoft 365 Consumer subscribersThe number of Microsoft 365 Consumer subscribers at end of period
LinkedIn revenue growthRevenue from LinkedIn, including Talent Solutions, Marketing Solutions, Premium Subscriptions, and Sales Solutions
Dynamics 365 revenue growthRevenue from Dynamics 365, including a set of intelligent, cloud-based applications across ERP, CRM, Power Apps, and Power Automate
Azure and other cloud services revenue growthRevenue from Azure and other cloud services, including cloud and AI consumption-based services, GitHub cloud services, Nuance Healthcare cloud services, virtual desktop offerings, and other cloud services
", + "path": "tables/table-5 Microsoft Cloud KPIs.html", + "metadata": { + "length": 1795, + "summary": "table-5\nThe table defines key performance indicators for Microsoft's commercial and consumer cloud services, including revenue growth, seat growth, and subscriber counts for Microsoft 365, LinkedIn, Dynamics 365, and Azure.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-5 Microsoft Cloud KPIs.html", + "keywords": [ + "Microsoft 365", + "Azure", + "Dynamics 365" + ], + "tokens": [] + } + }, + { + "chunk_id": "c70fc2de-5ee0-5069-8ae4-4f9152c8737b", + "type": "table", + "content": "
Windows OEM and Devices revenue growthRevenue from sales of Windows Pro and non-Pro licenses sold through the OEM channel and sales of first-party Devices, including Surface and PC accessories
Xbox content and services revenue growthRevenue from Xbox content and services, comprising first- and third-party content (including games and in-game content), Xbox Game Pass and other subscriptions, Xbox Cloud Gaming, advertising, and other cloud services
Search and news advertising revenue (ex TAC) growthRevenue from search and news advertising excluding traffic acquisition costs (“TAC”) paid to Bing Ads network publishers and news partners
", + "path": "tables/table-6 Microsoft Revenue.html", + "metadata": { + "length": 850, + "summary": "table-6\nRevenue growth from Windows OEM and Devices, Xbox content and services, and Search and news advertising excluding traffic acquisition costs.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-6 Microsoft Revenue.html", + "keywords": [ + "Windows OEM", + "Xbox content", + "Search advertising" + ], + "tokens": [] + } + }, + { + "chunk_id": "21d686c9-1f19-5e9b-99f4-7b4fdbc5cd0b", + "type": "table", + "content": "
(In millions, except percentages and per share amounts)20252024Percentage
Change
Revenue$281,724$245,12215%
Gross margin193,893171,00813%
Operating income128,528109,43317%
Net income101,83288,13616%
Diluted earnings per share13.6411.8016%
", + "path": "tables/table-7 Revenue Growth.html", + "metadata": { + "length": 1189, + "summary": "table-7\nThe table shows financial results for 2025 vs 2024: Revenue $281,724M (up 15%), Gross margin $193,893M (up 13%), Operating income $128,528M (up 17%), Net income $101,832M (up 16%), Diluted EPS $13.64 (up 16%).", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-7 Revenue Growth.html", + "keywords": [ + "Revenue", + "Net Income", + "EPS" + ], + "tokens": [] + } + }, + { + "chunk_id": "e4a917a1-1b21-5bf4-8860-fdc09eb4b30e", + "type": "table", + "content": "
(In millions, except percentages)20252024Percentage
Change
Productivity and Business Processes
Revenue$120,810$106,82013%
Cost of revenue22,42219,61114%
Operating expenses28,61527,5484%
Operating Income$69,773$59,66117%
Intelligent Cloud
Revenue$106,265$87,46421%
Cost of revenue40,17129,61136%
Operating expenses21,50520,0407%
Operating Income$44,589$37,81318%
More Personal Computing
Revenue$54,649$50,8387%
Cost of revenue25,23824,8921%
Operating expenses15,24513,9879%
Operating Income$14,166$11,95918%
Total
Revenue$281,724$245,12215%
Cost of revenue87,83174,11419%
Operating expenses65,36561,5756%
Operating Income$128,528$109,43317%
", + "path": "tables/table-8 Microsoft Segment Results.html", + "metadata": { + "length": 4940, + "summary": "table-8\nMicrosoft's 2025 vs 2024 segment data: Productivity & Business Processes revenue $120,810M (+13%), Intelligent Cloud $106,265M (+21%), More Personal Computing $54,649M (+7%). Total revenue $281,724M (+15%), operating income $128,528M (+17%).", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-8 Microsoft Segment Results.html", + "keywords": [ + "Revenue", + "Operating Income", + "Segments" + ], + "tokens": [] + } + }, + { + "chunk_id": "65e313e3-db40-55ee-a0b9-a74657e07419", + "type": "table", + "content": "
(In millions, except percentages)20252024Percentage
Change
Research and development$32,488$29,51010%
As a percent of revenue12%12%0ppt
", + "path": "tables/table-9 R&D Spending 2025.html", + "metadata": { + "length": 710, + "summary": "table-9\nResearch and development spending increased from $29,510 million in 2024 to $32,488 million in 2025, a 10% rise, while remaining at 12% of revenue.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-9 R&D Spending 2025.html", + "keywords": [ + "R&D", + "Revenue", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "301de650-4cd3-562e-a6a4-72b0cfb244d4", + "type": "table", + "content": "
(In millions, except percentages)20252024Percentage
Change
Sales and marketing$25,654$24,4565%
As a percent of revenue9%10%(1)ppt
", + "path": "tables/table-10 Sales & Marketing.html", + "metadata": { + "length": 705, + "summary": "table-10\nSales and marketing costs were $25,654 million in 2025 (9% of revenue) and $24,456 million in 2024 (10% of revenue), a 5% increase and a 1 percentage point decrease as a percent of revenue.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-10 Sales & Marketing.html", + "keywords": [ + "Sales", + "Marketing", + "Revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "4058dfb8-d7a1-5283-af19-829f40330f3a", + "type": "table", + "content": "
(In millions, except percentages)20252024Percentage
Change
General and administrative$7,223$7,609(5)%
As a percent of revenue3%3%0ppt
", + "path": "tables/table-11 G&A Expenses.html", + "metadata": { + "length": 709, + "summary": "table-11\nGeneral and administrative expenses decreased 5% from $7,609M in 2024 to $7,223M in 2025, remaining at 3% of revenue with 0ppt change.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-11 G&A Expenses.html", + "keywords": [ + "General and administrative", + "Revenue percentage", + "Cost change" + ], + "tokens": [] + } + }, + { + "chunk_id": "c0a4e50c-e185-55ca-9c4a-a71bbaf7ce12", + "type": "table", + "content": "
(In millions)
Year Ended June 30,20252024
Interest and dividends income$2,647$3,157
Interest expense(2,385)(2,935)
Net recognized losses on investments(349)(118)
Net losses on derivatives(260)(187)
Net gains (losses) on foreign currency remeasurements171(244)
Other, net(4,725)(1,319)
Total$(4,901)$(1,646)
", + "path": "tables/table-12 Financial Summary.html", + "metadata": { + "length": 1476, + "summary": "table-12\nThe table shows financial data for years ended June 30, 2025 and 2024, including interest income, expenses, investment losses, derivatives losses, foreign currency gains/losses, and other net losses, resulting in total net losses of $4,901 million and $1,646 million respectively.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-12 Financial Summary.html", + "keywords": [ + "Interest", + "Net Losses", + "Derivatives" + ], + "tokens": [] + } + }, + { + "chunk_id": "82944171-cfed-507c-a902-82c89f467531", + "type": "table", + "content": "
(In millions)
Three Months Ending
September 30, 2025$25,191
December 31, 202519,733
March 31, 202613,742
June 30, 20265,889
Thereafter2,710
Total$67,265
", + "path": "tables/table-13 Revenue Schedule.html", + "metadata": { + "length": 807, + "summary": "table-13\nThe table shows quarterly revenue figures from September 2025 to June 2026, with a total of $67,265 million.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-13 Revenue Schedule.html", + "keywords": [ + "Revenue", + "Quarterly", + "Total" + ], + "tokens": [] + } + }, + { + "chunk_id": "8d1714c5-e877-501e-83fa-d4750007bb69", + "type": "table", + "content": "
(In millions)2026ThereafterTotal
Long-term debt: (a)
Principal payments$3,000$46,206$49,206
Interest payments1,50925,52727,036
Construction commitments (b)26,8595,29032,149
Operating and finance leases, including imputed interest (c)12,798165,903178,701
Purchase commitments (d)103,9406,013109,953
Total$148,106$248,939$397,045
", + "path": "tables/table-14 Debt Commitments.html", + "metadata": { + "length": 1721, + "summary": "table-14\nThe table summarizes contractual obligations totaling $397,045 million, including principal and interest payments, construction commitments, leases, and purchase commitments, with $148,106 million due in 2026 and $248,939 million thereafter.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-14 Debt Commitments.html", + "keywords": [ + "Long-term debt", + "Leases", + "Purchase commitments" + ], + "tokens": [] + } + }, + { + "chunk_id": "7d1314f0-065b-54b3-84f1-55e25d235575", + "type": "table", + "content": "
(In millions)
Risk CategoriesHypothetical ChangeJune 30,
2025
Impact
Foreign currency – Revenue10% decrease in foreign exchange rates$(11,596)Earnings
Foreign currency – Investments10% decrease in foreign exchange rates(17)Fair Value
Interest rate100 basis point increase in U.S. treasury interest rates(1,415)Fair Value
Credit100 basis point increase in credit spreads(436)Fair Value
Equity10% decrease in equity market prices(1,213)Earnings
", + "path": "tables/table-15 Risk Sensitivity.html", + "metadata": { + "length": 1323, + "summary": "table-15\nThe table shows risk categories (foreign currency, interest rate, credit, equity) with hypothetical changes and their impacts on earnings or fair value as of June 30, 2025.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-15 Risk Sensitivity.html", + "keywords": [ + "Risk Categories", + "Hypothetical Change", + "Impact" + ], + "tokens": [] + } + }, + { + "chunk_id": "172af7d7-ea9b-5403-8809-b2545b8ea94b", + "type": "table", + "content": "
(In millions, except per share amounts)
Year Ended June 30,202520242023
Revenue:
Product$63,946$64,773$64,699
Service and other217,778180,349147,216
Total revenue281,724245,122211,915
Cost of revenue:
Product13,50115,27217,804
Service and other74,33058,84248,059
Total cost of revenue87,83174,11465,863
Gross margin193,893171,008146,052
Research and development32,48829,51027,195
Sales and marketing25,65424,45622,759
General and administrative7,2237,6097,575
Operating income128,528109,43388,523
Other income (expense), net(4,901)(1,646)788
Income before income taxes123,627107,78789,311
Provision for income taxes21,79519,65116,950
Net income$101,832$88,136$72,361
Earnings per share:
Basic$13.70$11.86$9.72
Diluted$13.64$11.80$9.68
Weighted average shares outstanding:
Basic7,4337,4317,446
Diluted7,4657,4697,472
", + "path": "tables/table-16 Revenue & Income.html", + "metadata": { + "length": 5292, + "summary": "table-16\nThe table shows financial data for years ending June 30, 2025, 2024, and 2023. Total revenue grew from $211,915M to $281,724M, net income from $72,361M to $101,832M, and diluted EPS from $9.68 to $13.64.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-16 Revenue & Income.html", + "keywords": [ + "Revenue", + "Net Income", + "Earnings per Share" + ], + "tokens": [] + } + }, + { + "chunk_id": "f996e041-a3b3-5bc1-b947-010fa5e45999", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Net income$101,832$88,136$72,361
Other comprehensive income (loss), net of tax:
Net change related to derivatives(5)24(14)
Net change related to investments1,574957(1,444)
Translation adjustments and other674(228)(207)
Other comprehensive income (loss)2,243753(1,665)
Comprehensive income$104,075$88,889$70,696
", + "path": "tables/table-17 Comprehensive Income.html", + "metadata": { + "length": 2283, + "summary": "table-17\nThe table shows net income, other comprehensive income (OCI) components, and total comprehensive income for fiscal years 2025, 2024, and 2023, with net income rising from $72,361M to $101,832M.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-17 Comprehensive Income.html", + "keywords": [ + "Net income", + "OCI", + "Comprehensive income" + ], + "tokens": [] + } + }, + { + "chunk_id": "8fec4155-9cc3-5780-a857-96e83308701e", + "type": "table", + "content": "
(In millions)
June 30,20252024
Assets
Current assets:
Cash and cash equivalents$30,242$18,315
Short-term investments64,32357,228
Total cash, cash equivalents, and short-term investments94,56575,543
Accounts receivable, net of allowance for doubtful accounts of $944 and $83069,90556,924
Inventories9381,246
Other current assets25,72326,021
Total current assets191,131159,734
Property and equipment, net of accumulated depreciation of $93,653 and $76,421204,966135,591
Operating lease right-of-use assets24,82318,961
Equity and other investments15,40514,600
Goodwill119,509119,220
Intangible assets, net22,60427,597
Other long-term assets40,56536,460
Total assets$619,003$512,163
Liabilities and stockholders’ equity
Current liabilities:
Accounts payable$27,724$21,996
Short-term debt06,693
Current portion of long-term debt2,9992,249
Accrued compensation13,70912,564
Short-term income taxes7,2115,017
Short-term unearned revenue64,55557,582
Other current liabilities25,02019,185
Total current liabilities141,218125,286
Long-term debt40,15242,688
Long-term income taxes25,98627,931
Long-term unearned revenue2,7102,602
Deferred income taxes2,8352,618
Operating lease liabilities17,43715,497
Other long-term liabilities45,18627,064
Total liabilities275,524243,686
Commitments and contingencies
Stockholders’ equity:
Common stock and paid-in capital – shares authorized 24,000; outstanding 7,434 and 7,434109,095100,923
Retained earnings237,731173,144
Accumulated other comprehensive loss(3,347)(5,590)
Total stockholders’ equity343,479268,477
Total liabilities and stockholders’ equity$619,003$512,163
", + "path": "tables/table-18 Balance Sheet.html", + "metadata": { + "length": 6315, + "summary": "table-18\nConsolidated balance sheet as of June 30, 2025 and 2024, showing total assets of $619,003 million and $512,163 million, with corresponding liabilities and stockholders' equity.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-18 Balance Sheet.html", + "keywords": [ + "Assets", + "Liabilities", + "Equity" + ], + "tokens": [] + } + }, + { + "chunk_id": "aaa89319-cae0-5959-a6fa-3b1a400b0a60", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Operations
Net income$101,832$88,136$72,361
Adjustments to reconcile net income to net cash from operations:
Depreciation, amortization, and other34,15322,28713,861
Stock-based compensation expense11,97410,7349,611
Net recognized losses on investments and derivatives609305196
Deferred income taxes(7,056)(4,738)(6,059)
Changes in operating assets and liabilities:
Accounts receivable(10,581)(7,191)(4,087)
Inventories3091,2841,242
Other current assets(3,044)(1,648)(1,991)
Other long-term assets(2,950)(6,817)(2,833)
Accounts payable5693,545(2,721)
Unearned revenue5,4385,3485,535
Income taxes(38)1,687(358)
Other current liabilities5,9224,8672,272
Other long-term liabilities(975)749553
Net cash from operations136,162118,54887,582
Financing
Proceeds from issuance (repayments) of debt, maturities of 90 days or less, net(5,746)5,2500
Proceeds from issuance of debt024,3950
Repayments of debt(3,216)(29,070)(2,750)
Common stock issued2,0562,0021,866
Common stock repurchased(18,420)(17,254)(22,245)
Common stock cash dividends paid(24,082)(21,771)(19,800)
Other, net(2,291)(1,309)(1,006)
Net cash used in financing(51,699)(37,757)(43,935)
Investing
Additions to property and equipment(64,551)(44,477)(28,107)
Acquisition of companies, net of cash acquired and divestitures, and purchases of intangible and other assets(5,978)(69,132)(1,670)
Purchases of investments(29,775)(17,732)(37,651)
Maturities of investments16,07924,77533,510
Sales of investments9,30910,89414,354
Other, net2,317(1,298)(3,116)
Net cash used in investing(72,599)(96,970)(22,680)
Effect of foreign exchange rates on cash and cash equivalents63(210)(194)
Net change in cash and cash equivalents11,927(16,389)20,773
Cash and cash equivalents, beginning of period18,31534,70413,931
Cash and cash equivalents, end of period$30,242$18,315$34,704
", + "path": "tables/table-19 Cash Flow Statement.html", + "metadata": { + "length": 8247, + "summary": "table-19\nThe table shows cash flow from operations, financing, and investing for fiscal years 2023-2025, with net cash from operations increasing from $87,582M to $136,162M.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-19 Cash Flow Statement.html", + "keywords": [ + "cash flow", + "operations", + "investing" + ], + "tokens": [] + } + }, + { + "chunk_id": "75b59a24-798a-519f-bcef-06088c97f96d", + "type": "table", + "content": "
(In millions, except per share amounts)
Year Ended June 30,202520242023
Common stock and paid-in capital
Balance, beginning of period$100,923$93,718$86,939
Common stock issued2,0562,0021,866
Common stock repurchased(5,856)(5,712)(4,696)
Stock-based compensation expense11,97410,7349,611
Other, net(2)181(2)
Balance, end of period109,095100,92393,718
Retained earnings
Balance, beginning of period173,144118,84884,281
Net income101,83288,13672,361
Common stock cash dividends(24,677)(22,293)(20,226)
Common stock repurchased(12,568)(11,547)(17,568)
Balance, end of period237,731173,144118,848
Accumulated other comprehensive loss
Balance, beginning of period(5,590)(6,343)(4,678)
Other comprehensive income (loss)2,243753(1,665)
Balance, end of period(3,347)(5,590)(6,343)
Total stockholders’ equity$343,479$268,477$206,223
Cash dividends declared per common share$3.32$3.00$2.72
", + "path": "tables/table-20 Stockholders' Equity.html", + "metadata": { + "length": 4663, + "summary": "table-20\nThe table shows stockholders' equity components for fiscal years 2023-2025, including common stock, retained earnings, and accumulated other comprehensive loss, with ending total equity of $343,479 million in 2025.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-20 Stockholders' Equity.html", + "keywords": [ + "equity", + "dividends", + "repurchases" + ], + "tokens": [] + } + }, + { + "chunk_id": "d7c618ea-146e-534a-ba94-d4a1ee5df5b6", + "type": "table", + "content": "
(In millions, except per share amounts)
Year Ended June 30,202520242023
Net income available for common shareholders (A)$101,832$88,136$72,361
Weighted average outstanding shares of common stock (B)7,4337,4317,446
Dilutive effect of stock-based awards323826
Common stock and common stock equivalents (C)7,4657,4697,472
Earnings Per Share
Basic (A/B)$13.70$11.86$9.72
Diluted (A/C)$13.64$11.80$9.68
", + "path": "tables/table-21 EPS Data Table.html", + "metadata": { + "length": 2415, + "summary": "table-21\nThe table shows net income, weighted average shares, and basic/diluted EPS for fiscal years 2023-2025. Net income rose from $72,361M to $101,832M, while diluted EPS increased from $9.68 to $13.64.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-21 EPS Data Table.html", + "keywords": [ + "Earnings Per Share", + "Net Income", + "Shares Outstanding" + ], + "tokens": [] + } + }, + { + "chunk_id": "8f0ab224-deaf-565e-89b5-1642ad95e14e", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Interest and dividends income$2,647$3,157$2,994
Interest expense(2,385)(2,935)(1,968)
Net recognized gains (losses) on investments(349)(118)260
Net losses on derivatives(260)(187)(456)
Net gains (losses) on foreign currency remeasurements171(244)181
Other, net(4,725)(1,319)(223)
Total$(4,901)$(1,646)$788
", + "path": "tables/table-22 Investment Income.html", + "metadata": { + "length": 1960, + "summary": "table-22\nThe table shows interest and dividends income, interest expense, net gains/losses on investments, derivatives, foreign currency, and other items for fiscal years 2025, 2024, and 2023, with totals of -$4,901M, -$1,646M, and $788M respectively.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-22 Investment Income.html", + "keywords": [ + "Interest", + "Investment", + "Net Losses" + ], + "tokens": [] + } + }, + { + "chunk_id": "577e18e7-d331-50fc-9972-61a95fd1d6bc", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Realized gains from sales of available-for-sale securities$40$22$36
Realized losses from sales of available-for-sale securities(65)(98)(124)
Impairments and allowance for credit losses823(10)
Total$(17)$(53)$(98)
", + "path": "tables/table-23 Investment Gains.html", + "metadata": { + "length": 1492, + "summary": "table-23\nThe table shows realized gains and losses from sales of available-for-sale securities and impairments for fiscal years 2025, 2024, and 2023, with total net losses of $17M, $53M, and $98M respectively.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-23 Investment Gains.html", + "keywords": [ + "realized gains", + "impairments", + "securities" + ], + "tokens": [] + } + }, + { + "chunk_id": "56ca9fcb-4e77-5b94-af9d-0e6f3a01dde4", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Net realized gains on investments sold$83$18$75
Net unrealized gains on investments still held536146303
Impairments of investments(951)(229)(20)
Total$(332)$(65)$358
", + "path": "tables/table-24 Investment Gains.html", + "metadata": { + "length": 1445, + "summary": "table-24\nThe table shows net realized gains, unrealized gains, and impairments for fiscal years 2025, 2024, and 2023, with totals of -332, -65, and 358 million respectively.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-24 Investment Gains.html", + "keywords": [ + "gains", + "impairments", + "total" + ], + "tokens": [] + } + }, + { + "chunk_id": "09f4dfcd-8f5a-502c-a2ce-c81f1c7502b6", + "type": "table", + "content": "
(In millions)Fair Value
Level
Adjusted
Cost Basis
Unrealized
Gains
Unrealized
Losses
Recorded
Basis
Cash and
Cash
Equivalents
Short-term
Investments
Equity and
Other
Investments
June 30, 2025
Changes in Fair Value Recorded in
Other Comprehensive Income
Commercial paperLevel 2$10,880$0$0$10,880$9,939$941$0
Certificates of depositLevel 22,653002,6532,3093440
U.S. government securitiesLevel 152,87871(1,462)51,4874,74246,7450
U.S. agency securitiesLevel 22,686002,6864962,1900
Foreign government bondsLevel 234924(9)36403640
Mortgage- and asset-backed securitiesLevel 22,55810(27)2,54102,5410
Corporate notes and bondsLevel 210,763124(101)10,786010,7860
Corporate notes and bondsLevel 32,51165(5)2,57101112,460
Municipal securitiesLevel 22071(7)20102010
Municipal securitiesLevel 31040(14)900900
Total debt investments$85,589$295$(1,625)$84,259$17,486$64,313$2,460
Changes in Fair Value Recorded in Net Income
Equity investmentsLevel 1$4,577$1,045$0$3,532
Equity investmentsOther9,141009,141
Total equity investments$13,718$1,045$0$12,673
Cash$11,711$11,711$0$0
Derivatives, net (a)282010272
Total$109,970$30,242$64,323$15,405
", + "path": "tables/table-25 Investment Table.html", + "metadata": { + "length": 10019, + "summary": "table-25\nThe table details investments as of June 30, 2025, showing adjusted cost basis, unrealized gains/losses, recorded basis, and classification into cash, short-term, equity, and other investments, totaling $109,970 million.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-25 Investment Table.html", + "keywords": [ + "Fair Value", + "Debt Investments", + "Equity Investments" + ], + "tokens": [] + } + }, + { + "chunk_id": "4a3c2966-3788-501d-a4bb-63808be3ee18", + "type": "table", + "content": "
(In millions)Fair Value
Level
Adjusted
Cost Basis
Unrealized
Gains
Unrealized
Losses
Recorded
Basis
Cash and
Cash
Equivalents
Short-term
Investments
Equity and
Other
Investments
June 30, 2024
Changes in Fair Value Recorded in Other Comprehensive Income
Commercial paperLevel 2$4,666$0$0$4,666$4,666$0$0
Certificates of depositLevel 21,547001,5471,503440
U.S. government securitiesLevel 149,6034(2,948)46,6591446,6450
U.S. agency securitiesLevel 21700170170
Foreign government bondsLevel 23193(16)30603060
Mortgage- and asset-backed securitiesLevel 29443(35)91209120
Corporate notes and bondsLevel 29,10628(318)8,81608,8160
Corporate notes and bondsLevel 31,6410(1)1,64001401,500
Municipal securitiesLevel 22620(13)24902490
Municipal securitiesLevel 31040(17)870870
Total debt investments$68,209$38$(3,348)$64,899$6,183$57,216$1,500
Changes in Fair Value Recorded in Net Income
Equity investmentsLevel 1$3,547$561$0$2,986
Equity investmentsOther10,1140010,114
Total equity investments$13,661$561$0$13,100
Cash$11,571$11,571$0$0
Derivatives, net (a)120120
Total$90,143$18,315$57,228$14,600
", + "path": "tables/table-26 Investment Summary.html", + "metadata": { + "length": 9976, + "summary": "table-26\nTable details investments as of June 30, 2024, with adjusted cost basis, unrealized gains/losses, and recorded basis across debt and equity categories, totaling $90,143 million.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-26 Investment Summary.html", + "keywords": [ + "Debt Investments", + "Equity Investments", + "Fair Value" + ], + "tokens": [] + } + }, + { + "chunk_id": "1539d95f-5366-5c5a-98e1-23db91787a71", + "type": "table", + "content": "
Less than 12 Months12 Months or GreaterTotal
Unrealized
Losses
Total
Unrealized
Losses
(In millions)Fair ValueUnrealized
Losses
Fair ValueUnrealized
Losses
Total
Fair Value
Total
Unrealized
Losses
June 30, 2025
U.S. government and agency securities$2,569$(51)$34,608$(1,411)$37,177$(1,462)
Foreign government bonds43(2)106(7)149(9)
Mortgage- and asset-backed securities841(4)189(23)1,030(27)
Corporate notes and bonds1,107(8)3,105(98)4,212(106)
Municipal securities00168(21)168(21)
Total$4,560$(65)$38,176$(1,560)$42,736$(1,625)
Less than 12 Months12 Months or GreaterTotal
Unrealized
Losses
Total
Unrealized
Losses
(In millions)Fair ValueUnrealized
Losses
Fair ValueUnrealized
Losses
Total
Fair Value
Total
Unrealized
Losses
June 30, 2024
U.S. government and agency securities$529$(12)$45,821$(2,936)$46,350$(2,948)
Foreign government bonds79(2)180(14)259(16)
Mortgage- and asset-backed securities201(1)409(34)610(35)
Corporate notes and bonds1,310(9)5,779(310)7,089(319)
Municipal securities38(1)243(29)281(30)
Total$2,157$(25)$52,432$(3,323)$54,589$(3,348)
", + "path": "tables/table-27 Unrealized Losses.html", + "metadata": { + "length": 7158, + "summary": "table-27\nThe table shows fair values and unrealized losses for securities held less than 12 months and 12 months or greater as of June 30, 2025 and 2024. Total fair value decreased from $54,589M to $42,736M, while total unrealized losses decreased from $(3,348)M to $(1,625)M.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-27 Unrealized Losses.html", + "keywords": [ + "securities", + "unrealized losses", + "fair value" + ], + "tokens": [] + } + }, + { + "chunk_id": "eba625c6-80bb-5830-8fb7-4881edfb68a6", + "type": "table", + "content": "
(In millions)Adjusted
Cost Basis
Estimated
Fair Value
June 30, 2025
Due in one year or less$35,108$34,952
Due after one year through five years42,46041,481
Due after five years through 10 years6,5306,424
Due after 10 years1,4911,402
Total$85,589$84,259
", + "path": "tables/table-28 Debt Securities.html", + "metadata": { + "length": 1239, + "summary": "table-28\nThe table shows adjusted cost basis and estimated fair value of debt securities by maturity as of June 30, 2025. Total cost basis is $85,589 million, with fair value at $84,259 million.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-28 Debt Securities.html", + "keywords": [ + "Cost Basis", + "Fair Value", + "Maturity" + ], + "tokens": [] + } + }, + { + "chunk_id": "8474c325-a28e-5fe3-9244-2d9d66a680c8", + "type": "table", + "content": "
(In millions)June 30,
2025
June 30,
2024
Designated as Hedging Instruments
Foreign exchange contracts purchased$1,492$1,492
Interest rate contracts purchased1,1501,100
Not Designated as Hedging Instruments
Foreign exchange contracts purchased15,2147,167
Foreign exchange contracts sold43,30731,793
Equity contracts purchased5,4344,016
Equity contracts sold2,1892,165
Other contracts purchased2,7692,113
Other contracts sold1,242811
", + "path": "tables/table-29 Derivative Notional.html", + "metadata": { + "length": 1768, + "summary": "table-29\nThe table shows notional amounts of derivative instruments as of June 30, 2025 and 2024, categorized by hedging designation and contract type, with foreign exchange contracts being the largest.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-29 Derivative Notional.html", + "keywords": [ + "Derivatives", + "Notional", + "Hedging" + ], + "tokens": [] + } + }, + { + "chunk_id": "d7f38500-41ab-5488-951e-0a6105cf64e1", + "type": "table", + "content": "
(In millions)Derivative
Assets
Derivative
Liabilities
Derivative
Assets
Derivative
Liabilities
June 30,
2025
June 30,
2024
Designated as Hedging Instruments
Foreign exchange contracts$89$(44)$24$(76)
Interest rate contracts150190
Not Designated as Hedging Instruments
Foreign exchange contracts248(809)213(230)
Equity contracts385(983)63(491)
Other contracts21(1)12(3)
Gross amounts of derivatives758(1,837)331(800)
Gross amounts of derivatives offset in the balance sheets(258)260(151)152
Cash collateral received0(99)0(104)
Net amounts of derivatives$500$(1,676)$180$(752)
Reported as
Short-term investments$10$0$12$0
Other current assets20101490
Equity and other investments272000
Other long-term assets170190
Other current liabilities0(1,639)0(401)
Other long-term liabilities0(37)0(351)
Total$500$(1,676)$180$(752)
", + "path": "tables/table-30 Derivative Assets.html", + "metadata": { + "length": 5670, + "summary": "table-30\nThe table shows derivative assets and liabilities as of June 30, 2025 and 2024, with gross amounts of $758M and $331M in assets, and $(1,837M) and $(800M) in liabilities, offset by collateral.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-30 Derivative Assets.html", + "keywords": [ + "Derivatives", + "Hedging", + "Financial Instruments" + ], + "tokens": [] + } + }, + { + "chunk_id": "9b401ce0-62b5-57f2-81a9-f1ae471ae44e", + "type": "table", + "content": "
(In millions)Level 1Level 2Level 3Total
June 30, 2025
Derivative assets$1$474$283$758
Derivative liabilities0(1,832)(5)(1,837)
June 30, 2024
Derivative assets03274331
Derivative liabilities(1)(799)0(800)
", + "path": "tables/table-31 Derivatives by Level.html", + "metadata": { + "length": 1869, + "summary": "table-31\nThe table shows derivative assets and liabilities by fair value hierarchy (Level 1, 2, 3) as of June 30, 2025 and 2024. Total derivative assets were $758M (2025) and $331M (2024); liabilities were $(1,837M) and $(800M) respectively.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-31 Derivatives by Level.html", + "keywords": [ + "Derivatives", + "Fair Value", + "Levels" + ], + "tokens": [] + } + }, + { + "chunk_id": "774362b2-4d34-5748-b999-a72bfbf1e1ee", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Designated as Fair Value Hedging Instruments
Interest rate contracts
Derivatives$5$(23)$(65)
Hedged items(45)(25)38
Designated as Cash Flow Hedging Instruments
Foreign exchange contracts
Amount reclassified from accumulated other comprehensive loss103(48)61
Not Designated as Hedging Instruments
Foreign exchange contracts(938)367(73)
Equity contracts(266)(177)(420)
Other contracts21(15)(41)
", + "path": "tables/table-32 Hedging Gains_Losses.html", + "metadata": { + "length": 2669, + "summary": "table-32\nThe table shows gains and losses from interest rate, foreign exchange, equity, and other contracts designated as fair value, cash flow, or non-designated hedging instruments for fiscal years 2023-2025.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-32 Hedging Gains_Losses.html", + "keywords": [ + "hedging", + "derivatives", + "fair value" + ], + "tokens": [] + } + }, + { + "chunk_id": "4911001d-4060-58e9-815b-96f22f51545b", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Designated as Cash Flow Hedging Instruments
Foreign exchange contracts
Included in effectiveness assessment$77$   (14)$34
", + "path": "tables/table-33 Cash Flow Hedges.html", + "metadata": { + "length": 1160, + "summary": "table-33\nThe table shows gains (losses) on foreign exchange contracts designated as cash flow hedging instruments: $77M in 2025, $(14)M in 2024, and $34M in 2023.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-33 Cash Flow Hedges.html", + "keywords": [ + "hedging", + "foreign exchange", + "cash flow" + ], + "tokens": [] + } + }, + { + "chunk_id": "67202e30-5eeb-5d05-a101-99d342e69724", + "type": "table", + "content": "
(In millions)
June 30,20252024
Land$9,338$8,163
Buildings and improvements137,92193,943
Leasehold improvements12,1179,594
Computer equipment and software132,83693,780
Furniture and equipment6,4076,532
Total, at cost298,619212,012
Accumulated depreciation(93,653)(76,421)
Total, net$204,966$135,591
", + "path": "tables/table-34 Property Assets.html", + "metadata": { + "length": 1568, + "summary": "table-34\nThe table shows property assets at cost of $298,619M (2025) vs $212,012M (2024), with accumulated depreciation of $93,653M and $76,421M, resulting in net values of $204,966M and $135,591M.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-34 Property Assets.html", + "keywords": [ + "Property", + "Depreciation", + "Net Value" + ], + "tokens": [] + } + }, + { + "chunk_id": "9a5bfc83-30b1-5018-8ff7-41a738c3dc90", + "type": "table", + "content": "
(In millions)
Cash and cash equivalents$12,976
Goodwill51,001
Intangible assets21,969
Other assets2,503
Long-term debt(2,799)
Long-term income taxes(1,946)
Deferred income taxes(4,676)
Other liabilities(3,620)
Total purchase price$75,408
", + "path": "tables/table-35 Purchase Price.html", + "metadata": { + "length": 929, + "summary": "table-35\nThe table shows a total purchase price of $75,408 million, comprising cash, goodwill, intangible assets, other assets, and various liabilities.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-35 Purchase Price.html", + "keywords": [ + "purchase price", + "assets", + "liabilities" + ], + "tokens": [] + } + }, + { + "chunk_id": "48034024-9c74-53e3-9b34-c6cf7c84825d", + "type": "table", + "content": "
(In millions, except average life)AmountWeighted
Average Life
Marketing-related$11,61924 years
Technology-based9,6894 years
Customer-related6614 years
Fair value of intangible assets acquired$21,96915 years
", + "path": "tables/table-36 Intangible Assets.html", + "metadata": { + "length": 814, + "summary": "table-36\nThe table shows fair value of acquired intangible assets totaling $21,969 million with a weighted average life of 15 years, broken down into marketing-related ($11,619M, 24 years), technology-based ($9,689M, 4 years), and customer-related ($661M, 4 years).", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-36 Intangible Assets.html", + "keywords": [ + "fair value", + "intangible assets", + "weighted average life" + ], + "tokens": [] + } + }, + { + "chunk_id": "6c9eaeba-45ae-5065-9440-b0d0155eab20", + "type": "table", + "content": "
(In millions, except per share amounts)
Year Ended June 30,20242023
Revenue$247,442$219,790
Net income88,30871,383
Diluted earnings per share11.829.55
", + "path": "tables/table-37 Microsoft FY2024.html", + "metadata": { + "length": 832, + "summary": "table-37\nMicrosoft's fiscal year ended June 30, 2024, shows revenue of $247,442 million, net income of $88,308 million, and diluted EPS of $11.82, compared to $219,790 million, $71,383 million, and $9.55 in 2023.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-37 Microsoft FY2024.html", + "keywords": [ + "Revenue", + "Net Income", + "EPS" + ], + "tokens": [] + } + }, + { + "chunk_id": "7bd19c3e-a5c9-5270-a340-a4edc82d6206", + "type": "table", + "content": "
(In millions)June 30,
2023
AcquisitionsOtherJune 30,
2024
AcquisitionsOtherJune 30,
2025
Productivity and Business Processes$31,359$0$2$31,361$0$96$31,457
Intelligent Cloud25,6760(28)25,64804125,689
More Personal Computing10,85151,23512562,211015262,363
Total$67,886$51,235$99$119,220$0$289$119,509
", + "path": "tables/table-38 Goodwill by Segment.html", + "metadata": { + "length": 2266, + "summary": "table-38\nMicrosoft's goodwill by segment from June 2023 to June 2025, showing totals increasing from $67,886M to $119,509M, driven by acquisitions in More Personal Computing.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-38 Goodwill by Segment.html", + "keywords": [ + "Goodwill", + "Segments", + "Microsoft" + ], + "tokens": [] + } + }, + { + "chunk_id": "232454cc-54aa-5ec2-a0c2-982930a29730", + "type": "table", + "content": "
(In millions)Gross
Carrying
Amount
Accumulated
Amortization
Net
Carrying
Amount
Gross
Carrying
Amount
Accumulated
Amortization
Net
Carrying
Amount
June 30,20252024
Marketing-related$16,502$(3,901)$12,601$16,500$(3,101)$13,399
Technology-based22,560(14,959)7,60121,913(10,741)11,172
Customer-related4,278(2,050)2,2286,038(3,051)2,987
Contract-based217(43)17458(19)39
Total$43,557$(20,953)$22,604$44,509$(16,912)$27,597
", + "path": "tables/table-39 Intangible Assets.html", + "metadata": { + "length": 2691, + "summary": "table-39\nThe table shows gross carrying amounts, accumulated amortization, and net carrying amounts of intangible assets (marketing, technology, customer, contract-based) as of June 30, 2025 and 2024. Total net carrying amount decreased from $27,597M to $22,604M.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-39 Intangible Assets.html", + "keywords": [ + "amortization", + "carrying amount", + "intangible assets" + ], + "tokens": [] + } + }, + { + "chunk_id": "94c04c68-141f-5982-90ac-5f46acfb78bc", + "type": "table", + "content": "
(In millions)AmountWeighted
Average Life
AmountWeighted
Average Life
Year Ended June 30,20252024
Marketing-related$1310 years$11,61924 years
Technology-based9129 years10,9474 years
Customer-related00 years6604 years
Contract-based1715 years384 years
Total$1,0969 years$23,26414 years
", + "path": "tables/table-40 Intangible Assets.html", + "metadata": { + "length": 2057, + "summary": "table-40\nThe table shows intangible assets by type (marketing, technology, customer, contract) with amounts and weighted average lives for fiscal years 2025 and 2024. Total assets decreased from $23,264M to $1,096M.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-40 Intangible Assets.html", + "keywords": [ + "Intangible assets", + "Weighted average life", + "Amortization" + ], + "tokens": [] + } + }, + { + "chunk_id": "9ff99e1e-0c88-5acf-bc6c-4cdc1c1ab453", + "type": "table", + "content": "
(In millions)
Year Ending June 30,
2026$4,594
20272,901
20282,034
20291,851
20301,382
Thereafter9,842
Total$22,604
", + "path": "tables/table-41 Revenue Forecast.html", + "metadata": { + "length": 834, + "summary": "table-41\nThe table shows projected revenue from 2026 to 2030 and thereafter, totaling $22,604 million, with the largest amount in the 'Thereafter' period.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-41 Revenue Forecast.html", + "keywords": [ + "Revenue", + "Forecast", + "Millions" + ], + "tokens": [] + } + }, + { + "chunk_id": "203d64ab-9b51-5ee5-830b-f7042569057c", + "type": "table", + "content": "
(In millions, issuance by calendar year)Maturities
(calendar year)
Stated Interest
Rate
Effective
Interest
Rate
June 30,
2025
June 30,
2024
2009 issuance of $3.8 billion20395.20%5.24%$520$520
2010 issuance of $4.8 billion20404.50%4.57%486486
2011 issuance of $2.3 billion20415.30%5.36%718718
2012 issuance of $2.3 billion20423.50%3.57%454454
2013 issuance of $5.2 billion20433.75%–4.88%3.83%–4.92%314314
2013 issuance of €4.1 billion2028–20332.63%–3.13%2.69%–3.22%2,7002,465
2015 issuance of $23.8 billion2025–20553.13%–4.75%3.18%–4.78%7,5559,805
2016 issuance of $19.8 billion2026–20562.40%–3.95%2.46%–4.03%7,9307,930
2017 issuance of $17.1 billion2026–20573.30%–4.50%3.38%–5.49%6,8336,833
2020 issuance of $10.1 billion2030–20601.35%–2.68%2.53%–5.43%10,11110,111
2021 issuance of $8.2 billion2052–20622.92%–3.04%2.92%–3.04%8,1858,185
2023 issuance of $0.1 billion2026–20501.35%–4.50%5.16%–5.49%5656
2024 issuance of $3.3 billion2026–20501.35%–4.50%5.16%–5.49%3,3443,344
Total face value49,20651,221
Unamortized discount and
issuance costs
(1,155)(1,227)
Hedge fair value adjustments (a)(36)(81)
Premium on debt exchange(4,864)(4,976)
Total debt43,15144,937
Current portion of long-term debt(2,999)(2,249)
Long-term debt$40,152$42,688
", + "path": "tables/table-42 Debt Issuance Table.html", + "metadata": { + "length": 5873, + "summary": "table-42\nThe table details long-term debt issuances from 2009 to 2024, including face values, maturities, stated and effective interest rates, and carrying amounts as of June 30, 2025 and 2024, totaling $43,151 million and $44,937 million respectively.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-42 Debt Issuance Table.html", + "keywords": [ + "Debt", + "Maturities", + "Interest Rates" + ], + "tokens": [] + } + }, + { + "chunk_id": "2a31a2e6-10ff-5f17-b76c-be44e548759b", + "type": "table", + "content": "
(In millions)
Year Ending June 30,
2026$3,000
20279,250
20280
20292,054
20300
Thereafter34,902
Total$49,206
", + "path": "tables/table-43 Debt Maturity.html", + "metadata": { + "length": 824, + "summary": "table-43\nThe table shows debt maturities from 2026 onward, with $49,206 million total. Key payments: $3,000M in 2026, $9,250M in 2027, $2,054M in 2029, and $34,902M thereafter.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-43 Debt Maturity.html", + "keywords": [ + "Debt", + "Maturity", + "Schedule" + ], + "tokens": [] + } + }, + { + "chunk_id": "a38735ed-f1e5-59f3-b5b1-bb9c3195605b", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Current Taxes
U.S. federal$14,086$12,165$14,009
U.S. state and local3,3422,3662,322
Foreign11,4239,8586,678
Current taxes$28,851$24,389$23,009
Deferred Taxes
U.S. federal$(6,250)$(4,791)$(6,146)
U.S. state and local(1,087)(379)(477)
Foreign281432564
Deferred taxes$(7,056)$(4,738)$(6,059)
Provision for income taxes$21,795$19,651$16,950
", + "path": "tables/table-44 Income Tax Provision.html", + "metadata": { + "length": 2861, + "summary": "table-44\nThe table shows current and deferred tax provisions for fiscal years 2023-2025. Current taxes total $28,851M (2025), $24,389M (2024), $23,009M (2023). Deferred taxes are negative, resulting in total provision of $21,795M, $19,651M, $16,950M respectively.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-44 Income Tax Provision.html", + "keywords": [ + "tax", + "provision", + "deferred" + ], + "tokens": [] + } + }, + { + "chunk_id": "35db7d9d-f52f-5807-9862-a55597bd303b", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
U.S.$69,212$62,886$52,917
Foreign54,41544,90136,394
Income before income taxes$123,627$107,787$89,311
", + "path": "tables/table-45 Income Before Taxes.html", + "metadata": { + "length": 1161, + "summary": "table-45\nThe table shows income before income taxes for years 2025, 2024, and 2023, with U.S. and foreign breakdowns, totaling $123,627 million, $107,787 million, and $89,311 million respectively.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-45 Income Before Taxes.html", + "keywords": [ + "Income", + "Taxes", + "Revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "efea2194-29ab-5e8e-9e8a-b1ade3dc4796", + "type": "table", + "content": "
Year Ended June 30,202520242023
Federal statutory rate21.0%21.0%21.0%
Effect of:
Foreign earnings taxed at lower rates(1.5)%(1.4)%(1.8)%
Foreign-derived intangible income deduction(1.0)%(1.1)%(1.3)%
State income taxes, net of federal benefit1.5%1.5%1.6%
Research and development credit(1.1)%(1.1)%(1.1)%
Excess tax benefits relating to stock-based compensation(0.9)%(1.1)%(0.7)%
Interest, net1.0%1.1%0.8%
Other reconciling items, net(1.4)%(0.7)%0.5%
Effective rate17.6%18.2%19.0%
", + "path": "tables/table-46 Tax Rate Analysis.html", + "metadata": { + "length": 2289, + "summary": "table-46\nThe table shows a reconciliation of the federal statutory tax rate (21%) to the effective tax rate for fiscal years 2023-2025, with adjustments for foreign earnings, credits, and other items.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-46 Tax Rate Analysis.html", + "keywords": [ + "tax rate", + "effective rate", + "reconciliation" + ], + "tokens": [] + } + }, + { + "chunk_id": "37e9fc8e-3678-5a29-9e1e-7562a9ca8023", + "type": "table", + "content": "
(In millions)
June 30,20252024
Deferred Income Tax Assets
Stock-based compensation expense$909$765
Accruals, reserves, and other expenses5,0504,381
Loss and credit carryforwards2,1141,741
Amortization4,1184,159
Leasing liabilities12,8746,504
Unearned revenue4,3243,717
Book/tax basis differences in investments and debt3039
Capitalized research and development16,89111,442
Other529426
Deferred income tax assets47,11233,144
Less valuation allowance(1,169)(1,045)
Deferred income tax assets, net of valuation allowance$45,943$32,099
Deferred Income Tax Liabilities
Leasing assets$(12,696)$(6,503)
Depreciation(5,699)(3,940)
Deferred tax on foreign earnings(1,148)(1,837)
Other(127)(167)
Deferred income tax liabilities$(19,670)$(12,447)
Net deferred income tax assets$26,273$19,652
Reported As
Other long-term assets$29,108$22,270
Long-term deferred income tax liabilities(2,835)(2,618)
Net deferred income tax assets$26,273$19,652
", + "path": "tables/table-47 Deferred Tax Assets.html", + "metadata": { + "length": 4372, + "summary": "table-47\nThe table details deferred income tax assets and liabilities as of June 30, 2025 and 2024, with net deferred tax assets of $26,273 million and $19,652 million respectively.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-47 Deferred Tax Assets.html", + "keywords": [ + "Deferred tax", + "Assets", + "Liabilities" + ], + "tokens": [] + } + }, + { + "chunk_id": "84b6f224-48d2-548e-bb74-41134065a57a", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Beginning unrecognized tax benefits$22,760$17,120$15,593
Decreases related to settlements(240)(76)(329)
Increases for tax positions related to the current year2,0661,9031,051
Increases for tax positions related to prior years4684,289870
Decreases for tax positions related to prior years(300)(464)(60)
Decreases due to lapsed statutes of limitations(25)(12)(5)
Ending unrecognized tax benefits$24,729$22,760$17,120
", + "path": "tables/table-48 Tax Benefits.html", + "metadata": { + "length": 2069, + "summary": "table-48\nThe table shows unrecognized tax benefits from 2023 to 2025, starting at $15,593 million and ending at $24,729 million, with changes from settlements, current and prior year positions, and statute lapses.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-48 Tax Benefits.html", + "keywords": [ + "unrecognized tax benefits", + "tax positions", + "settlements" + ], + "tokens": [] + } + }, + { + "chunk_id": "db4a750b-c288-58dc-99c2-ff1f5efc436f", + "type": "table", + "content": "
(In millions)
June 30,20252024
Productivity and Business Processes$50,567$43,599
Intelligent Cloud14,02213,683
More Personal Computing2,6762,902
Total$67,265$60,184
", + "path": "tables/table-49 Microsoft Revenue.html", + "metadata": { + "length": 1061, + "summary": "table-49\nMicrosoft's total revenue increased from $60,184 million in 2024 to $67,265 million in 2025, driven by growth in Productivity and Business Processes and Intelligent Cloud.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-49 Microsoft Revenue.html", + "keywords": [ + "Revenue", + "Segments", + "Comparison" + ], + "tokens": [] + } + }, + { + "chunk_id": "087287ba-e4d1-5ad9-b7be-1fdffdd0a05f", + "type": "table", + "content": "
(In millions)
Year Ended June 30, 2025
Balance, beginning of period$60,184
Deferral of revenue186,957
Recognition of unearned revenue(179,876)
Balance, end of period$67,265
", + "path": "tables/table-50 Unearned Revenue.html", + "metadata": { + "length": 733, + "summary": "table-50\nThe table shows unearned revenue balance increased from $60,184M to $67,265M, with $186,957M deferred and $179,876M recognized in fiscal year 2025.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-50 Unearned Revenue.html", + "keywords": [ + "unearned revenue", + "deferral", + "recognition" + ], + "tokens": [] + } + }, + { + "chunk_id": "4452390e-74dd-5620-b691-6e58c8dc77de", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Operating lease cost$5,524$3,555$2,875
Finance lease cost:
Amortization of right-of-use assets$3,408$1,800$1,352
Interest on lease liabilities1,417734501
Total finance lease cost$4,825$2,534$1,853
", + "path": "tables/table-51 Lease Costs.html", + "metadata": { + "length": 1670, + "summary": "table-51\nThe table shows operating lease costs of $5,524M (2025), $3,555M (2024), $2,875M (2023) and total finance lease costs of $4,825M, $2,534M, $1,853M for the same years.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-51 Lease Costs.html", + "keywords": [ + "Operating lease", + "Finance lease", + "Lease cost" + ], + "tokens": [] + } + }, + { + "chunk_id": "28d25b09-4699-514f-8b98-01fc0bd50366", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Cash paid for amounts included in the measurement of lease liabilities:
Operating cash flows from operating leases$4,931$3,550$2,706
Operating cash flows from finance leases1,372734501
Financing cash flows from finance leases2,2831,2861,056
Right-of-use assets obtained in exchange for lease obligations:
Operating leases7,8266,7033,514
Finance leases20,51111,6333,128
", + "path": "tables/table-52 Lease Cash Flows.html", + "metadata": { + "length": 1850, + "summary": "table-52\nThe table shows cash paid for operating and finance leases from 2023 to 2025, and right-of-use assets obtained in exchange for lease obligations.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-52 Lease Cash Flows.html", + "keywords": [ + "lease liabilities", + "cash flows", + "right-of-use assets" + ], + "tokens": [] + } + }, + { + "chunk_id": "da108065-e4c5-5301-95cb-bfa49a12d84f", + "type": "table", + "content": "
(In millions, except lease term and discount rate)
June 30,20252024
Operating Leases
Operating lease right-of-use assets$24,823$18,961
Other current liabilities$5,424$3,580
Operating lease liabilities17,43715,497
Total operating lease liabilities$22,861$19,077
Finance Leases
Property and equipment, at cost$53,876$32,248
Accumulated depreciation(9,861)(6,386)
Property and equipment, net$44,015$25,862
Other current liabilities$3,172$2,349
Other long-term liabilities43,00024,796
Total finance lease liabilities$46,172$27,145
Weighted Average Remaining Lease Term
Operating leases6 years7 years
Finance leases13 years12 years
Weighted Average Discount Rate
Operating leases3.5%3.3%
Finance leases4.2%3.9%
", + "path": "tables/table-53 Lease Data.html", + "metadata": { + "length": 3727, + "summary": "table-53\nThe table presents operating and finance lease data as of June 30, 2025 and 2024, including right-of-use assets, liabilities, weighted average lease terms, and discount rates.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-53 Lease Data.html", + "keywords": [ + "Operating leases", + "Finance leases", + "Lease liabilities" + ], + "tokens": [] + } + }, + { + "chunk_id": "e267be8c-3565-5cd5-9a8a-d0d1596d29f2", + "type": "table", + "content": "
(In millions)
Year Ending June 30,Operating
Leases
Finance
Leases
2026$6,111$5,008
20275,2375,157
20283,4955,187
20292,4194,521
20302,0174,382
Thereafter6,20236,251
Total lease payments25,48160,506
Less imputed interest(2,620)(14,334)
Total$22,861$46,172
", + "path": "tables/table-54 Lease Payments.html", + "metadata": { + "length": 1678, + "summary": "table-54\nThe table shows future lease payments for operating and finance leases from 2026 to 2030 and thereafter, totaling $25,481 million and $60,506 million respectively, with imputed interest deducted to present net lease liabilities.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-54 Lease Payments.html", + "keywords": [ + "Operating Leases", + "Finance Leases", + "Lease Liabilities" + ], + "tokens": [] + } + }, + { + "chunk_id": "36c1f271-f9e2-5fb3-b351-1f862f2b95fe", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Balance, beginning of year7,4347,4327,464
Issued313437
Repurchased(31)(32)(69)
Balance, end of year7,4347,4347,432
", + "path": "tables/table-55 Share Repurchases.html", + "metadata": { + "length": 1394, + "summary": "table-55\nThe table shows share balances for fiscal years 2023-2025, with beginning balances around 7,432-7,464 million, issuances of 31-37 million, repurchases of 31-69 million, and ending balances of 7,432-7,434 million.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-55 Share Repurchases.html", + "keywords": [ + "shares", + "repurchases", + "balance" + ], + "tokens": [] + } + }, + { + "chunk_id": "657a22eb-9366-5d37-b1fa-074b1683ae81", + "type": "table", + "content": "
(In millions)SharesAmountSharesAmountSharesAmount
Year Ended June 30,202520242023
First Quarter7$2,80011$3,56017$4,600
Second Quarter83,50072,800204,600
Third Quarter83,50072,800184,600
Fourth Quarter83,20072,800144,600
Total31$13,00032$11,96069$18,400
", + "path": "tables/table-56 Share Repurchases.html", + "metadata": { + "length": 2547, + "summary": "table-56\nThe table shows quarterly share repurchases and amounts for fiscal years 2023-2025, with totals of 69 shares ($18,400M) in 2023, 32 shares ($11,960M) in 2024, and 31 shares ($13,000M) in 2025.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-56 Share Repurchases.html", + "keywords": [ + "Shares", + "Amount", + "Quarterly" + ], + "tokens": [] + } + }, + { + "chunk_id": "0a3c08c8-285f-5234-ba6e-dcd0afd088ac", + "type": "table", + "content": "
Declaration DateRecord DatePayment DateDividend
Per Share
Amount
Fiscal Year 2025(In millions)
September 16, 2024November 21, 2024December 12, 2024$0.83$6,170
December 3, 2024February 20, 2025March 13, 20250.836,169
March 11, 2025May 15, 2025June 12, 20250.836,169
June 10, 2025August 21, 2025September 11, 20250.836,170
Total$3.32$24,678
Fiscal Year 2024
September 19, 2023November 16, 2023December 14, 2023$0.75$5,574
November 28, 2023February 15, 2024March 14, 20240.755,573
March 12, 2024May 16, 2024June 13, 20240.755,574
June 12, 2024August 15, 2024September 12, 20240.755,574
Total$3.00$22,295
", + "path": "tables/table-57 Dividend History.html", + "metadata": { + "length": 3219, + "summary": "table-57\nThe table shows quarterly dividends for fiscal years 2025 and 2024, with per-share amounts of $0.83 and $0.75 respectively, totaling $24,678 million and $22,295 million.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-57 Dividend History.html", + "keywords": [ + "Dividend", + "Fiscal Year", + "Per Share" + ], + "tokens": [] + } + }, + { + "chunk_id": "61be9b16-9678-54c0-89a8-bc758632e3a0", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Derivatives
Balance, beginning of period$(3)$(27)$(13)
Unrealized gains (losses), net of tax of $20, $(4), and $977(14)34
Reclassification adjustments for (gains) losses included in other income
(expense), net
(103)48(61)
Tax expense (benefit) included in provision for income taxes21(10)13
Amounts reclassified from accumulated other comprehensive loss(82)38(48)
Net change related to derivatives, net of tax of $(1), $6, and $(4)(5)24(14)
Balance, end of period$(8)$(3)$(27)
Investments
Balance, beginning of period$(2,625)$(3,582)$(2,138)
Unrealized gains (losses), net of tax of $411, $247, and $(393)1,560915(1,523)
Reclassification adjustments for losses included in other income
(expense), net
175399
Tax benefit included in provision for income taxes(3)(11)(20)
Amounts reclassified from accumulated other comprehensive loss144279
Net change related to investments, net of tax of $414, $258, and $(373)1,574957(1,444)
Balance, end of period$(1,051)$(2,625)$(3,582)
Translation Adjustments and Other
Balance, beginning of period$(2,962)$(2,734)$(2,527)
Translation adjustments and other, net of tax of $8, $0, and $0674(228)(207)
Balance, end of period$(2,288)$(2,962)$(2,734)
Accumulated other comprehensive loss, end of period$(3,347)$(5,590)$(6,343)
", + "path": "tables/table-58 AOCI Rollforward.html", + "metadata": { + "length": 6017, + "summary": "table-58\nThe table shows changes in accumulated other comprehensive loss for derivatives, investments, and translation adjustments for fiscal years 2023-2025, ending with a total loss of $(3,347) million in 2025.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-58 AOCI Rollforward.html", + "keywords": [ + "Derivatives", + "Investments", + "Translation" + ], + "tokens": [] + } + }, + { + "chunk_id": "94b78184-d619-5a1d-b89d-8bb7afb772ab", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Stock-based compensation expense$11,974$10,734$9,611
Income tax benefits related to stock-based compensation2,0271,8261,651
", + "path": "tables/table-59 Stock Compensation.html", + "metadata": { + "length": 956, + "summary": "table-59\nThe table shows stock-based compensation expenses and related income tax benefits for fiscal years 2025, 2024, and 2023, with amounts in millions of dollars.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-59 Stock Compensation.html", + "keywords": [ + "stock-based compensation", + "income tax benefits", + "financial data" + ], + "tokens": [] + } + }, + { + "chunk_id": "309e74c6-6705-520b-a28b-959fb9661644", + "type": "table", + "content": "
Year Ended June 30,202520242023
Dividends per share (quarterly amounts)$0.75 –  0.83$0.68 –  0.75$0.62 –  0.68
Interest rates3.4% – 5.5%3.8% – 5.6%2.0% – 5.4%
", + "path": "tables/table-60 Dividends & Rates.html", + "metadata": { + "length": 771, + "summary": "table-60\nThe table shows quarterly dividends per share and interest rates for fiscal years 2023-2025. Dividends range from $0.62 to $0.83, and interest rates range from 2.0% to 5.6%.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-60 Dividends & Rates.html", + "keywords": [ + "Dividends", + "Interest Rates", + "Fiscal Years" + ], + "tokens": [] + } + }, + { + "chunk_id": "792d3955-3e4a-56a4-8154-2e6d38d6218b", + "type": "table", + "content": "
SharesWeighted Average
Grant-Date Fair Value
(In millions)
Stock Awards
Nonvested balance, beginning of year88$292.28
Granted (a)39413.90
Vested(38)293.25
Forfeited(7)317.23
Nonvested balance, end of year82$347.44
", + "path": "tables/table-61 Stock Awards.html", + "metadata": { + "length": 1244, + "summary": "table-61\nThe table shows nonvested stock award activity: beginning balance 88 shares at $292.28, 39 granted at $413.90, 38 vested at $293.25, 7 forfeited at $317.23, ending balance 82 shares at $347.44.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-61 Stock Awards.html", + "keywords": [ + "Nonvested", + "Shares", + "Fair Value" + ], + "tokens": [] + } + }, + { + "chunk_id": "4e74811f-7ba5-517d-8687-fc6523746a50", + "type": "table", + "content": "
(Shares in millions)
Year Ended June 30,202520242023
Shares purchased667
Average price per share$385.10$339.46$245.59
", + "path": "tables/table-62 Share Repurchase.html", + "metadata": { + "length": 904, + "summary": "table-62\nThe table shows shares purchased and average price per share for fiscal years 2025, 2024, and 2023. Shares purchased were 6 million in 2025 and 2024, and 7 million in 2023. Average prices were $385.10, $339.46, and $245.59 respectively.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-62 Share Repurchase.html", + "keywords": [ + "share repurchase", + "average price", + "shares purchased" + ], + "tokens": [] + } + }, + { + "chunk_id": "0579405f-2c41-52d7-aca9-c24a01e2a411", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Productivity and Business Processes
Revenue$120,810$106,820$94,151
Cost of revenue22,42219,61117,202
Operating expenses28,61527,54826,875
Operating Income$69,773$59,661$50,074
Intelligent Cloud
Revenue$106,265$87,464$72,944
Cost of revenue40,17129,61124,109
Operating expenses21,50520,04020,424
Operating Income$44,589$37,813$28,411
More Personal Computing
Revenue$54,649$50,838$44,820
Cost of revenue25,23824,89224,552
Operating expenses15,24513,98710,230
Operating Income$14,166$11,959$10,038
Total
Revenue$281,724$245,122$211,915
Cost of revenue87,83174,11465,863
Operating expenses65,36561,57557,529
Operating Income$128,528$109,433$88,523
", + "path": "tables/table-63 Microsoft Segment Results.html", + "metadata": { + "length": 4793, + "summary": "table-63\nThe table shows revenue, cost of revenue, operating expenses, and operating income for three Microsoft segments (Productivity, Intelligent Cloud, More Personal Computing) and totals for fiscal years 2023-2025.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-63 Microsoft Segment Results.html", + "keywords": [ + "Revenue", + "Operating Income", + "Segments" + ], + "tokens": [] + } + }, + { + "chunk_id": "e3d92df6-6a6e-53d0-abb3-1136d50f5ae9", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
United States (a)$144,546$124,704$106,744
Other countries137,178120,418105,171
Total$281,724$245,122$211,915
", + "path": "tables/table-64 Revenue by Region.html", + "metadata": { + "length": 1263, + "summary": "table-64\nThe table shows total revenue of $281,724 million in 2025, $245,122 million in 2024, and $211,915 million in 2023, split between United States and other countries.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-64 Revenue by Region.html", + "keywords": [ + "Revenue", + "Geographic", + "Annual" + ], + "tokens": [] + } + }, + { + "chunk_id": "89bdd525-3b5c-53e1-947e-df3aefb5bdb9", + "type": "table", + "content": "
(In millions)
Year Ended June 30,202520242023
Server products and cloud services$98,435$79,828$65,007
Microsoft 365 Commercial products and cloud services87,76776,96966,949
Gaming23,45521,50315,466
LinkedIn17,81216,37214,989
Windows and Devices17,31417,02617,147
Search and news advertising13,87812,30612,125
Dynamics products and cloud services7,8276,8315,796
Enterprise and partner services7,7607,5947,900
Microsoft 365 Consumer products and cloud services7,4046,6486,417
Other7245119
Total$281,724$245,122$211,915
", + "path": "tables/table-65 Microsoft Revenue.html", + "metadata": { + "length": 2625, + "summary": "table-65\nMicrosoft's revenue for fiscal years 2025, 2024, and 2023 is shown, with totals of $281,724M, $245,122M, and $211,915M respectively, led by Server products and cloud services.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-65 Microsoft Revenue.html", + "keywords": [ + "Revenue", + "Cloud", + "Products" + ], + "tokens": [] + } + }, + { + "chunk_id": "f84caf3c-647d-582d-a0b5-ddbab86fdb05", + "type": "table", + "content": "
(In millions)
June 30,202520242023
United States$230,069$186,106$114,380
Other countries141,833115,26372,859
Total$371,902$301,369$187,239
", + "path": "tables/table-66 Revenue by Region.html", + "metadata": { + "length": 1247, + "summary": "table-66\nThe table shows total revenue of $371,902 million in June 2025, up from $301,369 million in 2024 and $187,239 million in 2023, with United States and Other countries contributions.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-66 Revenue by Region.html", + "keywords": [ + "Revenue", + "United States", + "Other Countries" + ], + "tokens": [] + } + }, + { + "chunk_id": "6656b43d-55a4-5cfd-9025-359ff3b95182", + "type": "table", + "content": "
/S/ DELOITTE & TOUCHE LLP
Seattle, Washington
July 30, 2025
", + "path": "tables/table-67 Deloitte Signature.html", + "metadata": { + "length": 230, + "summary": "table-67\nA signature block for Deloitte & Touche LLP, located in Seattle, Washington, dated July 30, 2025.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-67 Deloitte Signature.html", + "keywords": [ + "Deloitte", + "Seattle", + "Date" + ], + "tokens": [] + } + }, + { + "chunk_id": "6b48ae77-7813-5b62-87f3-ff41c622c790", + "type": "table", + "content": "
Satya Nadella
Chairman and Chief Executive Officer,
Microsoft Corporation
Catherine MacGregor 2,4
Group Chief Executive Officer and
Director, Engie S.A.
Carlos Rodriguez 1, 2
Director, Automatic Data
Processing, Inc.
Reid G. Hoffman 4
Partner, Greylock Partners
Mark A. L. Mason 3
Chief Financial Officer, Citigroup Inc.
Charles W. Scharf 2,3
Chief Executive Officer, President,
and Director, Wells Fargo & Company
Hugh F. Johnston 1
Senior Executive Vice President and
Chief Financial Officer,
The Walt Disney Company
Sandra E. Peterson 2,3
Lead Independent Director,
Microsoft Corporation
Operating Partner, Clayton, Dubilier &
Rice, LLC
John W. Stanton 1,4
Founder and Chairman, Trilogy
Partnerships
Teri L. List 1,3
Former Executive Vice President and
Chief Financial Officer, The Gap, Inc.
Penny S. Pritzker 4
Founder and Chairman, PSP
Partners, LLC
Emma N. Walmsley 2,4
Chief Executive Officer and Director, GSK plc
", + "path": "tables/table-68 Microsoft Board.html", + "metadata": { + "length": 1428, + "summary": "table-68\nThe table lists Microsoft Corporation's board members including Satya Nadella, Reid Hoffman, Hugh Johnston, and others with their roles and affiliations.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-68 Microsoft Board.html", + "keywords": [ + "Board of Directors", + "Microsoft", + "Corporate Governance" + ], + "tokens": [] + } + }, + { + "chunk_id": "fc7f99a2-f90b-53db-9944-c6216c05ea6c", + "type": "table", + "content": "
Satya Nadella
Chairman and Chief Executive Officer
Amy E. Hood
Executive Vice President and Chief Financial Officer
Judson B. Althoff
Executive Vice President and CEO Microsoft Commercial
Takeshi Numoto
Executive Vice President and Chief Marketing Officer
Amy L. Coleman
Executive Vice President and Chief Human Resources Officer
Bradford L. Smith
Vice Chair and President
Kathleen T. Hogan
Executive Vice President, Office of Strategy and Transformation
", + "path": "tables/table-69 Microsoft Executives.html", + "metadata": { + "length": 763, + "summary": "table-69\nThe table lists Microsoft's top executives including Satya Nadella (CEO), Amy Hood (CFO), Judson Althoff, Takeshi Numoto, Amy Coleman, Bradford Smith, and Kathleen Hogan.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "tables/table-69 Microsoft Executives.html", + "keywords": [ + "CEO", + "CFO", + "Executive" + ], + "tokens": [] + } + }, + { + "chunk_id": "edf99fdb-63b6-532d-a46a-4829e552912c", + "type": "image", + "content": "\nCertification Type: MIX (Mixed Sources)\nDescription: Paper | Supporting responsible forestry\nLicense Number: FSC® C132107\nIssuing Authority: FSC (Forest Stewardship Council)\nWebsite: www.fsc.org\n[images/image-4 FSC Certification Labe.jpeg]\n", + "path": "images/image-4 FSC Certification Labe.jpeg", + "metadata": { + "length": 241, + "summary": "image-4\nCertification Type: MIX (Mixed Sources)\nDescription: Paper | Supporting responsible forestry\nLicense Number: FSC® C132107\nIssuing Authority: FSC (Forest Stewardship Council)\nWebsite: www.fsc.org", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "file_path": "images/image-4 FSC Certification Labe.jpeg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "3c018846-994f-5ad4-83e7-57e493222eaa", + "type": "text", + "content": "Microsoft Annual Report 2025 Cover\nPrimary Subject: Two colleagues collaborating at a desk with a laptop.\nVisible Text: \"Annual Report 2025\", Microsoft Logo.\n[images/image-1 Annual Report 2025.png]\n\nDear shareholders, colleagues, customers, and partners:\nFifty years after our founding, Microsoft is once again at the heart of a generational moment in technology as we find ourselves in the midst of the AI platform shift. More than any transformation before it, this generation of AI is radically changing every layer of the tech stack, and we are changing with it.\nAcross the company, we are accelerating our pace of innovation and adapting to both a new tech stack and a new way of working. We are delivering our current platforms at scale while building the next generation, always striving to create more value for our customers, our partners, and the world.\nStriking this balance is hard work, and few companies over the years have been able to do it. To succeed, we must continue to think in decades but execute in quarters, approaching each day with the humility and curiosity required to continuously improve, while being guided by our bold vision for the future.\nFinancially, it was a year of record performance. Revenue was $281.7 billion, up 15 percent. Operating income grew 17 percent to $128.5 billion. And Azure surpassed $75 billion in revenue for the first time, up 34 percent. These results reflect the growing demand for our platform and the trust customers are placing in us. We take neither for granted.\nWe must earn our permission to operate every day, in every country, every community, and every customer interaction. That’s why we remain grounded in our mission: to empower every person and every organization on the planet to achieve more.\nImagine a world where every person can get help from a researcher, a coder, or an analyst on demand. Not just information, but deep, contextual expertise paired with action. Or where every organization, no matter its size or sector, can reinvent employee experiences, reimagine customer engagement, reshape business processes, and bend the curve on innovation for their people, businesses, and industries. This is the new frontier and how we will unlock the next level of productivity and growth for the world.\nBut it is not some far off vision—we are already seeing what’s possible when AI reaches the frontlines of human potential, helping small businesses become more productive, multinationals more competitive, nonprofits more effective, governments more efficient, and improving healthcare and education outcomes.\nTo share just a few examples across industries: Mercy, one of the largest health systems in the US, has saved caregivers over 100,000 hours by automatically documenting physician-patient encounters. As one physician put it: “the best thing to happen to my practice in 10 years.” A grandmother in Japan, who lost her hearing at age two, can now communicate with her voice, thanks to an AI app. A judge in Colombia is using Copilot to expedite due process and help tackle a backlog of court cases. Barclays Bank is putting AI in the hands of 100,000 employees, transforming the employee experience by simplifying how they access information and get things done. Ralph Lauren is helping customers find the perfect look for any occasion, thanks to a new conversational shopping experience. Carvana has reduced inbound calls per sale by 45 percent, freeing its staff to focus on complex, high-value support.\nThese examples, and so many others like them, are made possible by our clear focus on our priorities, our responsibility, and our culture.\nOUR PRIORITIES\nTo deliver on our mission, we remain focused on three core business priorities as our North Star: security, quality, and AI innovation.\nSecurity and quality are non-negotiable. Our infrastructure and services are mission critical for the world. This year, we made significant progress across both our Secure Future Initiative (SFI) and Quality Excellence Initiative (QEI), but we recognize our work here is never done. We must continuously raise the bar for ourselves and our customers.\nSecurity\nThrough SFI, we have dedicated the equivalent of 34,000 full-time engineers to our highest-priority security work. We strengthened identity protections, secured our networks and systems, enhanced threat detection and response, and embedded secure-by-design practices across everything we build.\nQuality", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 1", + "metadata": { + "length": 4442, + "summary": "Microsoft's 2025 annual report highlights the AI platform shift as a generational moment, with record revenue of $281.7 billion and Azure surpassing $75 billion. The company focuses on security, quality, and AI innovation, citing examples like Mercy Health saving 100,000 hours and Barclays deploying AI to 100,000 employees.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Microsoft", + "Annual", + "Report", + "2025", + "Cover", + "Primary", + "Subject", + "Two", + "colleagues", + "collaborating", + "desk", + "laptop", + "Visible", + "Text", + "Logo", + "Dear", + "shareholders", + "customers", + "partners", + "Fifty", + "years", + "founding", + "heart", + "generational", + "moment", + "technology", + "find", + "midst", + "AI", + "platform", + "shift", + "More", + "transformation", + "generation", + "radically", + "changing", + "layer", + "tech", + "stack", + "Across", + "company", + "accelerating", + "pace", + "innovation", + "adapting", + "working", + "We", + "delivering", + "current", + "platforms", + "scale", + "building", + "striving", + "create", + "world", + "Striking", + "balance", + "hard", + "work", + "companies", + "To", + "succeed", + "continue", + "decades", + "execute", + "quarters", + "approaching", + "day", + "humility", + "curiosity", + "required", + "continuously", + "improve", + "guided", + "bold", + "vision", + "future", + "Financially", + "year", + "record", + "performance", + "Revenue", + "281.7", + "billion", + "15", + "percent", + "Operating", + "income", + "grew", + "17", + "128.5", + "And", + "Azure", + "surpassed", + "75", + "revenue", + "time", + "34", + "These", + "results", + "reflect", + "growing", + "demand", + "trust", + "placing", + "granted", + "earn", + "permission", + "operate", + "country", + "community", + "customer", + "interaction", + "That", + "remain", + "grounded", + "mission", + "empower", + "person", + "organization", + "planet", + "achieve", + "Imagine", + "researcher", + "coder", + "analyst", + "Not", + "information", + "deep", + "contextual", + "expertise", + "paired", + "action", + "Or", + "matter", + "size", + "sector", + "reinvent", + "employee", + "experiences", + "reimagine", + "engagement", + "reshape", + "business", + "processes", + "bend", + "curve", + "people", + "businesses", + "industries", + "This", + "frontier", + "unlock", + "level", + "productivity", + "growth", + "But", + "reaches", + "frontlines", + "human", + "potential", + "helping", + "small", + "productive", + "multinationals", + "competitive", + "nonprofits", + "effective", + "governments", + "efficient", + "improving", + "healthcare", + "education", + "outcomes", + "share", + "examples", + "Mercy", + "largest", + "health", + "systems", + "US", + "saved", + "caregivers", + "100", + "000", + "hours", + "automatically", + "documenting", + "physician", + "patient", + "encounters", + "As", + "put", + "thing", + "happen", + "practice", + "10", + "grandmother", + "Japan", + "lost", + "hearing", + "age", + "communicate", + "voice", + "app", + "judge", + "Colombia", + "Copilot", + "expedite", + "due", + "process", + "tackle", + "backlog", + "court", + "cases", + "Barclays", + "Bank", + "putting", + "hands", + "employees", + "transforming", + "experience", + "simplifying", + "access", + "things", + "Ralph", + "Lauren", + "perfect", + "occasion", + "conversational", + "shopping", + "Carvana", + "reduced", + "inbound", + "calls", + "sale", + "45", + "freeing", + "staff", + "focus", + "complex", + "high", + "support", + "made", + "clear", + "priorities", + "responsibility", + "culture", + "OUR", + "PRIORITIES", + "deliver", + "focused", + "core", + "North", + "Star", + "security", + "quality", + "Security", + "negotiable", + "Our", + "infrastructure", + "services", + "critical", + "significant", + "progress", + "Secure", + "Future", + "Initiative", + "SFI", + "Quality", + "Excellence", + "QEI", + "recognize", + "raise", + "bar", + "Through", + "dedicated", + "equivalent", + "full", + "engineers", + "highest", + "priority", + "strengthened", + "identity", + "protections", + "secured", + "networks", + "enhanced", + "threat", + "detection", + "response", + "embedded", + "secure", + "design", + "practices", + "build" + ], + "keywords": [ + "AI platform shift", + "record revenue", + "security" + ], + "connect_to": [ + { + "target": "d55ac241-0983-5356-98e6-5388e5b473d0", + "relation": "embeds", + "ref": "[images/image-1 Annual Report 2025.png]", + "position": { + "start": 158, + "end": 197 + } + } + ] + } + }, + { + "chunk_id": "2cccd030-d6e6-5d15-872d-be4bd5bf89b4", + "type": "text", + "content": "With QEI, we created frameworks that increase accountability and accelerate progress against our engineering objectives to ensure we deliver durable, high quality-experiences at global scale. This includes improvements to change management, incident management, platform resiliency, and service health.\nTogether, these initiatives are laying the foundation for a renaissance of our engineering culture, where we build planet-scale systems that power the world, with the security and quality they require.\nAI innovation\nAt the same time, we have made major advances in AI innovation, including across two foundational areas: our Cloud and AI infrastructure, and our family of Copilots and agents.\nOur Cloud and AI infrastructure\nWe continue to lead the AI infrastructure wave. We opened new datacenters across six continents and now operate more than 400 datacenters in 70 regions, more than any other cloud provider. This year alone, we added over two gigawatts of new capacity. Every Azure region is now AI-first and can support liquid cooling, increasing the fungibility and the flexibility of our fleet. And just last month, we announced the world’s most powerful AI datacenter, Fairwater in southeastern Wisconsin, which will deliver 10x the performance of the world’s fastest supercomputer today.\nWe are also driving and benefiting from compounding improvements in silicon, systems, and models to improve performance and efficiency. And we continue to invest in sovereign cloud offerings to meet the unique data residency needs of governments and industries worldwide.\nWe have made meaningful progress on the next frontier in cloud systems: quantum. We announced Majorana-1, the first quantum chip with a topological core, and deployed the world’s first operational Level 2 quantum computer in partnership with Atom Computing.\nIn data and analytics, Microsoft Fabric is becoming the unified platform for the AI era. It is now our fastest-growing analytics product ever, with 25,000 paid customers. OneLake spans all databases and clouds, including Power BI semantic models, making it the best foundation for building enterprise AI applications.\nWe also introduced Azure AI Foundry, a platform to design, customize, and run powerful AI apps and agents. Foundry includes access to more than 11,000 models from partners like OpenAI, Cohere, DeepSeek, Meta, Mistral, xAI, and others, ensuring our customers can choose from the best frontier and open models in one place. Already, 80 percent of the Fortune 500 use Foundry for their AI workloads.\nAnd this fall, we introduced our first in-house models: MAI-1 preview, our first foundation model trained end-to-end in-house, as well as MAI-Voice-1 for natural voice generation and MAI-Image-1 for image generation.\nCopilots and agents\nOur Copilot family of products is helping people thrive at home, at school, and at work. This year, we surpassed 100 million monthly active users across both commercial and consumer.\nWe rolled out a major update to Microsoft 365 Copilot this spring, bringing together chat, search, create, notebooks, and role-specific agents like Analyst and Researcher into a single experience. And earlier this month, we announced Agent Mode, which allows you to start with a simple prompt and then work iteratively with Copilot—steering it as it orchestrates multistep tasks to deliver high-quality Office documents, spreadsheets, and presentations.\nGive Copilot a prompt like, “Run a full analysis on this sales data set. I want to understand some important insights to help me make decisions about my business. Make it visual.” Agent Mode gets to work deciding which formulas to use, producing new sheets, and creating data visualizations. It’s pretty remarkable.\nCopilot Studio continues to grow as well, with more than 230,000 organizations using it to extend Microsoft 365 Copilot or to build their own agents using no-code/low-code tools.\nIn software development, GitHub Copilot now has more than 20 million users and has evolved into a peer programmer, capable of executing tasks on our behalf asynchronously. In healthcare, Dragon Copilot is being used to document millions of clinical encounters, saving precious time that healthcare providers can spend on patients, not paperwork. And in security, we were the first in the industry to introduce AI agents that help defenders autonomously manage high-volume security and IT tasks.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 2", + "metadata": { + "length": 4410, + "summary": "Microsoft's QEI frameworks improve engineering accountability and quality. Major AI advances include leading Cloud and AI infrastructure with over 400 datacenters globally, the Majorana-1 quantum chip, and Azure AI Foundry with 11,000+ models. Copilot products surpass 100 million monthly active users, with updates like Agent Mode and Copilot Studio used by 230,000 organizations. GitHub Copilot has 20 million users, and Dragon Copilot aids healthcare.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "With", + "QEI", + "created", + "frameworks", + "increase", + "accountability", + "accelerate", + "progress", + "engineering", + "objectives", + "ensure", + "deliver", + "durable", + "high", + "quality", + "experiences", + "global", + "scale", + "This", + "includes", + "improvements", + "change", + "management", + "incident", + "platform", + "resiliency", + "service", + "health", + "Together", + "initiatives", + "laying", + "foundation", + "renaissance", + "culture", + "build", + "planet", + "systems", + "power", + "world", + "security", + "require", + "AI", + "innovation", + "At", + "time", + "made", + "major", + "advances", + "including", + "foundational", + "areas", + "Cloud", + "infrastructure", + "family", + "Copilots", + "agents", + "Our", + "We", + "continue", + "lead", + "wave", + "opened", + "datacenters", + "continents", + "operate", + "400", + "70", + "regions", + "cloud", + "provider", + "year", + "added", + "gigawatts", + "capacity", + "Every", + "Azure", + "region", + "support", + "liquid", + "cooling", + "increasing", + "fungibility", + "flexibility", + "fleet", + "And", + "month", + "announced", + "powerful", + "datacenter", + "Fairwater", + "southeastern", + "Wisconsin", + "10x", + "performance", + "fastest", + "supercomputer", + "today", + "driving", + "benefiting", + "compounding", + "silicon", + "models", + "improve", + "efficiency", + "invest", + "sovereign", + "offerings", + "meet", + "unique", + "data", + "residency", + "governments", + "industries", + "worldwide", + "meaningful", + "frontier", + "quantum", + "Majorana", + "chip", + "topological", + "core", + "deployed", + "operational", + "Level", + "computer", + "partnership", + "Atom", + "Computing", + "In", + "analytics", + "Microsoft", + "Fabric", + "unified", + "era", + "It", + "growing", + "product", + "25", + "000", + "paid", + "customers", + "OneLake", + "spans", + "databases", + "clouds", + "Power", + "BI", + "semantic", + "making", + "building", + "enterprise", + "applications", + "introduced", + "Foundry", + "design", + "customize", + "run", + "apps", + "access", + "11", + "partners", + "OpenAI", + "Cohere", + "DeepSeek", + "Meta", + "Mistral", + "xAI", + "ensuring", + "choose", + "open", + "place", + "Already", + "80", + "percent", + "Fortune", + "500", + "workloads", + "fall", + "house", + "MAI", + "preview", + "model", + "trained", + "end", + "Voice", + "natural", + "voice", + "generation", + "image", + "Copilot", + "products", + "helping", + "people", + "thrive", + "home", + "school", + "work", + "surpassed", + "100", + "million", + "monthly", + "active", + "users", + "commercial", + "consumer", + "rolled", + "update", + "365", + "spring", + "bringing", + "chat", + "search", + "create", + "notebooks", + "role", + "specific", + "Analyst", + "Researcher", + "single", + "experience", + "earlier", + "Agent", + "Mode", + "start", + "simple", + "prompt", + "iteratively", + "steering", + "orchestrates", + "multistep", + "tasks", + "Office", + "documents", + "spreadsheets", + "presentations", + "Give", + "Run", + "full", + "analysis", + "sales", + "set", + "understand", + "important", + "insights", + "make", + "decisions", + "business", + "Make", + "visual", + "deciding", + "formulas", + "producing", + "sheets", + "creating", + "visualizations", + "pretty", + "remarkable", + "Studio", + "continues", + "grow", + "230", + "organizations", + "extend", + "code", + "low", + "tools", + "software", + "development", + "GitHub", + "20", + "evolved", + "peer", + "programmer", + "capable", + "executing", + "behalf", + "asynchronously", + "healthcare", + "Dragon", + "document", + "millions", + "clinical", + "encounters", + "saving", + "precious", + "providers", + "spend", + "patients", + "paperwork", + "industry", + "introduce", + "defenders", + "autonomously", + "manage", + "volume", + "IT" + ], + "keywords": [ + "AI", + "Cloud", + "Copilot" + ], + "connect_to": [] + } + }, + { + "chunk_id": "298da42f-7e1c-5ebb-99eb-aac2b1e18aad", + "type": "text", + "content": "On the consumer front, Copilot is now integrated across Bing, Edge, GroupMe, MSN, Windows, and Xbox. We also refreshed our Copilot consumer app this year as a more natural, conversational, and personal AI companion. One highlight is Copilot Mode in Edge, which lets you chat directly with your open tabs.\nMore broadly, we continue to expand our reach with consumers. LinkedIn is now home to 1.2 billion members, and we are bringing AI agents into the core workflows of sales, hiring, and learning. And across gaming, we have 500 million monthly active users across platforms and devices.\nOUR RESPONSIBILITY\nAt Microsoft, we have always believed that we do well when the world around us does well. We demonstrate this, year after year, by making progress on our commitments to create technology that benefits everyone on the planet, along with the planet itself.\nThis means that as we drive the AI economy, we are also providing the skills and opportunities for everyone to participate in it. That’s why we launched Microsoft Elevate this year, an initiative that unites our technology, skills, research, and philanthropic investments, so AI can benefit every classroom, community, and cause. Over the next five years, we will invest $4 billion in cash and AI cloud technology to schools, community and technical colleges, and nonprofit organizations. And we will partner with organizations like UNICEF, Code.org, and more to extend AI skilling opportunities to people around the globe.\nOur skilling initiatives will help 20 million people earn AI credentials over the next two years, from foundational fluency to advanced technical training. And to support our work with deeper research and policy insights, our new AI Economy Institute will explore how AI is reshaping education and work, helping us bridge the gap between technological innovation and societal impact.\nWe are also focused on empowering teachers and students with the latest AI tools, making Microsoft 365 Copilot Personal free for 12 months for every college student in the United States. And we’re introducing new LinkedIn courses for teachers and students, enabling them to earn LinkedIn certifications that will help boost their resumes and open doors.\nAcross all of this, one thing is clear: People want technology they can trust. That’s why we are committed to driving responsible AI innovation and building safe and secure technology. As we do, we are guided by our values of respect, integrity, and accountability. They are built into the design of our software and services, the security of our data, our privacy protections, and our engagement with employees, customers, suppliers, communities, and governments around the world.\nAs a multinational company, we remain dedicated to creating jobs, promoting economic opportunities, and strengthening cybersecurity, digital stability, sovereignty, and resilience globally. In Europe, we announced five commitments to strengthen digital stability this year. They start with an expansion of our cloud and AI infrastructure, so every country can compete in the global economy. And they include a promise to uphold Europe’s digital resilience regardless of geopolitical and trade volatility.\nResponsible business practices are embedded across our operations and supply chain. Following the UN Guiding Principles on Business and Human Rights, we’ve strengthened our due diligence processes, expanded stakeholder engagement, and launched new governance mechanisms to ensure accountability. Our second annual Responsible AI Transparency Report highlighted our work to build and deploy AI responsibly. This includes developing new AI tools and resources to help our customers innovate within evolving regulatory requirements.\nAnother key component of earning trust is contributing to a safer online ecosystem, including protecting those who use our services from illegal and harmful content and conduct. We continue to take new steps to advance safety, especially for children, while balancing our commitments to free expression and privacy. Over the past year, we’ve focused on addressing risks related to abusive AI-generated content and partnered with StopNCII.org to detect victim-reported imagery in Bing.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 3", + "metadata": { + "length": 4225, + "summary": "Microsoft integrates Copilot across Bing, Edge, GroupMe, MSN, Windows, and Xbox, and refreshes its Copilot consumer app. LinkedIn reaches 1.2 billion members, and gaming has 500 million monthly active users. The company emphasizes responsibility through initiatives like Microsoft Elevate, investing $4 billion over five years in AI cloud technology for schools and nonprofits. It aims to help 20 million people earn AI credentials in two years and launches the AI Economy Institute. Microsoft offers free Microsoft 365 Copilot Personal for U.S. college students and new LinkedIn courses. It commits to responsible AI innovation, safety, and security, guided by values of respect, integrity, and accountability. The company strengthens digital stability in Europe, follows UN Guiding Principles on Business and Human Rights, and publishes a Responsible AI Transparency Report. It also partners with StopNCII.org to address abusive AI-generated content.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "On", + "consumer", + "front", + "Copilot", + "integrated", + "Bing", + "Edge", + "GroupMe", + "MSN", + "Windows", + "Xbox", + "We", + "refreshed", + "app", + "year", + "natural", + "conversational", + "personal", + "AI", + "companion", + "One", + "highlight", + "Mode", + "lets", + "chat", + "directly", + "open", + "tabs", + "More", + "broadly", + "continue", + "expand", + "reach", + "consumers", + "LinkedIn", + "home", + "1.2", + "billion", + "members", + "bringing", + "agents", + "core", + "workflows", + "sales", + "hiring", + "learning", + "And", + "gaming", + "500", + "million", + "monthly", + "active", + "users", + "platforms", + "devices", + "OUR", + "RESPONSIBILITY", + "At", + "Microsoft", + "believed", + "world", + "demonstrate", + "making", + "progress", + "commitments", + "create", + "technology", + "benefits", + "planet", + "This", + "means", + "drive", + "economy", + "providing", + "skills", + "opportunities", + "participate", + "That", + "launched", + "Elevate", + "initiative", + "unites", + "research", + "philanthropic", + "investments", + "benefit", + "classroom", + "community", + "Over", + "years", + "invest", + "cash", + "cloud", + "schools", + "technical", + "colleges", + "nonprofit", + "organizations", + "partner", + "UNICEF", + "Code", + "org", + "extend", + "skilling", + "people", + "globe", + "Our", + "initiatives", + "20", + "earn", + "credentials", + "foundational", + "fluency", + "advanced", + "training", + "support", + "work", + "deeper", + "policy", + "insights", + "Economy", + "Institute", + "explore", + "reshaping", + "education", + "helping", + "bridge", + "gap", + "technological", + "innovation", + "societal", + "impact", + "focused", + "empowering", + "teachers", + "students", + "latest", + "tools", + "365", + "Personal", + "free", + "12", + "months", + "college", + "student", + "United", + "States", + "introducing", + "courses", + "enabling", + "certifications", + "boost", + "resumes", + "doors", + "Across", + "thing", + "clear", + "People", + "trust", + "committed", + "driving", + "responsible", + "building", + "safe", + "secure", + "As", + "guided", + "values", + "respect", + "integrity", + "accountability", + "They", + "built", + "design", + "software", + "services", + "security", + "data", + "privacy", + "protections", + "engagement", + "employees", + "customers", + "suppliers", + "communities", + "governments", + "multinational", + "company", + "remain", + "dedicated", + "creating", + "jobs", + "promoting", + "economic", + "strengthening", + "cybersecurity", + "digital", + "stability", + "sovereignty", + "resilience", + "globally", + "In", + "Europe", + "announced", + "strengthen", + "start", + "expansion", + "infrastructure", + "country", + "compete", + "global", + "include", + "promise", + "uphold", + "geopolitical", + "trade", + "volatility", + "Responsible", + "business", + "practices", + "embedded", + "operations", + "supply", + "chain", + "Following", + "UN", + "Guiding", + "Principles", + "Business", + "Human", + "Rights", + "ve", + "strengthened", + "due", + "diligence", + "processes", + "expanded", + "stakeholder", + "governance", + "mechanisms", + "ensure", + "annual", + "Transparency", + "Report", + "highlighted", + "build", + "deploy", + "responsibly", + "includes", + "developing", + "resources", + "innovate", + "evolving", + "regulatory", + "requirements", + "Another", + "key", + "component", + "earning", + "contributing", + "safer", + "online", + "ecosystem", + "including", + "protecting", + "illegal", + "harmful", + "content", + "conduct", + "steps", + "advance", + "safety", + "children", + "balancing", + "expression", + "past", + "addressing", + "risks", + "related", + "abusive", + "generated", + "partnered", + "StopNCII", + "detect", + "victim", + "reported", + "imagery" + ], + "keywords": [ + "AI", + "Responsibility", + "Skilling" + ], + "connect_to": [] + } + }, + { + "chunk_id": "587a23e0-5289-5616-bba1-6eef994d8978", + "type": "text", + "content": "Our solutions, partnerships, and programs reach people and organizations of all abilities to help them thrive. More than 5 million people have participated in our AI Skilling programs focused on accessibility. We launched new technology to help people with disabilities play, work, and live—through an Adaptive Joystick for Xbox, sign language detection in Teams, low-vision keyboards for Surface, and AI-powered visual descriptions in Windows.\n2025 also marked the midpoint in our journey to become a carbon negative, water positive, zero waste company, and to protect more land than we use. We are on track to meet many of our targets and continue to accelerate progress for others.\nOur renewable energy procurement increased from 1.8 gigawatts in 2020 to 34 gigawatts in 2024, and we contracted nearly 30 million metric tons of carbon removal—playing a pivotal role in scaling the carbon removal market. We provided more than 1.5 million people with clean water and sanitation and plan to replenish more than 100 million cubic meters of water around the world. We are getting closer to zero waste through new Circular Centers, which contribute to the reuse and recycling of nearly 91 percent of servers and components decommissioned from our datacenters. And we’ve reached nearly 95 percent recyclability in our product packaging.\nWe are learning how to make AI more sustainable by design and improve AI-powered solutions. Platforms like our Planetary Computer and our AI for Good Lab are helping us find new ways to address the world’s most pressing challenges.\nMaking progress on these commitments takes time. But here, too—we are guided by a bold vision, thinking in decades and executing in quarters.\nOUR CULTURE\nAmid this rapid progress, our culture is more important than ever. The AI platform shift is reshaping not just our products and business models, but how we work.\nOur growth mindset is essential to our ability to continue leading this AI era. It enables us to innovate both within Microsoft and with those we serve. We must be learn-it-alls, willing to experiment, guided by evaluations, and committed to continuous improvement. I am continually impressed by how our people do just that.\nWe are focused on being Customer Zero, applying AI to reduce toil and improve flow in our own work while creating a playbook we can share with the world.\nWe’re also embracing a new way of working—one that expands job scopes, reduces handoffs, and gives teams tools to scale productivity in nonlinear ways. This isn’t just about driving efficiency. It’s about empowering our people to dream bigger and get to “job complete” faster, with less friction and greater impact than ever before.\nOur employees also continue to find ways to bring their purpose and passion to the causes and communities they care deeply about. This year, they volunteered over 1.2 million hours and gave $263 million (including company match) to 37,000 nonprofit organizations in 110 countries.\n**\nIn July, after we reported our earnings results—including surpassing $75 billion in annual Azure revenue for the first time—I shared a reflection with all employees. Fifteen years ago, when we set out on our cloud journey, we had a bold vision, and we persisted through all the ups and downs.\nI asked our team: What are you working on today that, 15 years from now, you will look back on and say, “we got it right”? Hundreds of answers poured in spanning near-term priorities and long-term ambitions alike:\n• “Helping the world be safe from cybercrime”\n• “Speech is now a standard human-machine interface”\n• “Using AI to improve the lived experience for people with disabilities”\n• “We changed the way people learn”\n• “Access to millions of agents powered by orders of magnitude larger models”\n• “Medical knowledge has become ambient, embedded in clinical workflows”\n• “We helped billions become skilled in AI”\nThese responses and so many others reflected a deep sense of purpose and belief in what’s possible. And that is precisely what is needed for us to succeed.\nMicrosoft has an immense opportunity and responsibility—not just in building these futures, but in helping shape the world’s future. And I’m confident that together, we can continue to think in decades and execute in quarters on this journey to empower us all.\n\nSarga N.\n[images/image-2 Signature.jpeg]\n\nSatya Nadella\nChairman and Chief Executive Officer\nOctober 15, 2025\nISSUER PURCHASES OF EQUITY SECURITIES, DIVIDENDS, AND STOCK PERFORMANCE\nMARKET AND STOCKHOLDERS", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 4", + "metadata": { + "length": 4512, + "summary": "Microsoft's 2025 report highlights progress in AI skilling for accessibility, renewable energy procurement reaching 34 gigawatts, carbon removal contracts, clean water access, and zero waste initiatives. The company emphasizes a growth mindset culture, employee volunteering, and long-term vision for AI-driven transformation across cybersecurity, speech interfaces, disability support, education, and medical knowledge.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Our", + "solutions", + "partnerships", + "programs", + "reach", + "people", + "organizations", + "abilities", + "thrive", + "More", + "million", + "participated", + "AI", + "Skilling", + "focused", + "accessibility", + "We", + "launched", + "technology", + "disabilities", + "play", + "work", + "live", + "Adaptive", + "Joystick", + "Xbox", + "sign", + "language", + "detection", + "Teams", + "low", + "vision", + "keyboards", + "Surface", + "powered", + "visual", + "descriptions", + "Windows", + "2025", + "marked", + "midpoint", + "journey", + "carbon", + "negative", + "water", + "positive", + "waste", + "company", + "protect", + "land", + "track", + "meet", + "targets", + "continue", + "accelerate", + "progress", + "renewable", + "energy", + "procurement", + "increased", + "1.8", + "gigawatts", + "2020", + "34", + "2024", + "contracted", + "30", + "metric", + "tons", + "removal", + "playing", + "pivotal", + "role", + "scaling", + "market", + "provided", + "1.5", + "clean", + "sanitation", + "plan", + "replenish", + "100", + "cubic", + "meters", + "world", + "closer", + "Circular", + "Centers", + "contribute", + "reuse", + "recycling", + "91", + "percent", + "servers", + "components", + "decommissioned", + "datacenters", + "And", + "ve", + "reached", + "95", + "recyclability", + "product", + "packaging", + "learning", + "make", + "sustainable", + "design", + "improve", + "Platforms", + "Planetary", + "Computer", + "Good", + "Lab", + "helping", + "find", + "ways", + "address", + "pressing", + "challenges", + "Making", + "commitments", + "takes", + "time", + "But", + "guided", + "bold", + "thinking", + "decades", + "executing", + "quarters", + "OUR", + "CULTURE", + "Amid", + "rapid", + "culture", + "important", + "The", + "platform", + "shift", + "reshaping", + "products", + "business", + "models", + "growth", + "mindset", + "essential", + "ability", + "leading", + "era", + "It", + "enables", + "innovate", + "Microsoft", + "serve", + "learn", + "alls", + "experiment", + "evaluations", + "committed", + "continuous", + "improvement", + "continually", + "impressed", + "Customer", + "Zero", + "applying", + "reduce", + "toil", + "flow", + "creating", + "playbook", + "share", + "embracing", + "working", + "expands", + "job", + "scopes", + "reduces", + "handoffs", + "teams", + "tools", + "scale", + "productivity", + "nonlinear", + "This", + "isn", + "driving", + "efficiency", + "empowering", + "dream", + "bigger", + "complete", + "faster", + "friction", + "greater", + "impact", + "employees", + "bring", + "purpose", + "passion", + "communities", + "care", + "deeply", + "year", + "volunteered", + "1.2", + "hours", + "gave", + "263", + "including", + "match", + "37", + "000", + "nonprofit", + "110", + "countries", + "In", + "July", + "reported", + "earnings", + "results", + "surpassing", + "75", + "billion", + "annual", + "Azure", + "revenue", + "shared", + "reflection", + "Fifteen", + "years", + "ago", + "set", + "cloud", + "persisted", + "ups", + "downs", + "asked", + "team", + "What", + "today", + "15", + "back", + "Hundreds", + "answers", + "poured", + "spanning", + "term", + "priorities", + "long", + "ambitions", + "alike", + "Helping", + "safe", + "cybercrime", + "Speech", + "standard", + "human", + "machine", + "interface", + "Using", + "lived", + "experience", + "changed", + "Access", + "millions", + "agents", + "orders", + "magnitude", + "larger", + "Medical", + "knowledge", + "ambient", + "embedded", + "clinical", + "workflows", + "helped", + "billions", + "skilled", + "These", + "responses", + "reflected", + "deep", + "sense", + "belief", + "precisely", + "needed", + "succeed", + "immense", + "opportunity", + "responsibility", + "building", + "futures", + "shape", + "future", + "confident", + "execute", + "empower", + "Sarga", + "Satya", + "Nadella", + "Chairman", + "Chief", + "Executive", + "Officer", + "October", + "ISSUER", + "PURCHASES", + "OF", + "EQUITY", + "SECURITIES", + "DIVIDENDS", + "AND", + "STOCK", + "PERFORMANCE", + "MARKET", + "STOCKHOLDERS" + ], + "keywords": [ + "AI", + "Sustainability", + "Culture" + ], + "connect_to": [ + { + "target": "9e0d738a-c578-5816-afc1-93e714a7ab5d", + "relation": "embeds", + "ref": "[images/image-2 Signature.jpeg]", + "position": { + "start": 4316, + "end": 4347 + } + } + ] + } + }, + { + "chunk_id": "9e4e4300-3f24-59d3-ad92-a2cc2c5d6b6b", + "type": "text", + "content": "Our common stock is traded on the NASDAQ Stock Market under the symbol MSFT. On July 24, 2025, there were 77,014 registered holders of record of our common stock.\nSHARE REPURCHASES AND DIVIDENDS\nShare Repurchases\nOn September 14, 2021, our Board of Directors approved a share repurchase program authorizing up to $60.0 billion in share repurchases. This share repurchase program commenced in November 2021 and was completed in April 2025.\nOn September 16, 2024, our Board of Directors approved a share repurchase program authorizing up to $60.0 billion in share repurchases. This share repurchase program commenced in April 2025, following completion of the program approved on September 14, 2021, has no expiration date, and may be terminated at any time. As of June 30, 2025, $57.3 billion remained of this $60.0 billion share repurchase program.\nWe repurchased the following shares of common stock under the share repurchase programs:\n\nThe table shows quarterly share repurchases and amounts for fiscal years 2023-2025, with totals of 31 shares ($13,000M) in 2025, 32 shares ($11,960M) in 2024, and 69 shares ($18,400M) in 2023.\n[tables/table-1 Share Repurchases.html]\n\nAll repurchases were made using cash resources. Shares repurchased during the fourth quarter of fiscal year 2025 were under the share repurchase programs approved on September 14, 2021 and September 16, 2024. All other shares repurchased were under the share repurchase program approved on September 14, 2021. The above table excludes shares repurchased to settle employee tax withholding related to the vesting of stock awards of $5.4 billion, $5.3 billion, and $3.8 billion for fiscal years 2025, 2024, and 2023, respectively.\nDividends\nOur Board of Directors declared the following dividends:\n\nThe table shows quarterly dividends for fiscal years 2025 and 2024, with per-share amounts of $0.83 and $0.75 respectively, totaling $24,678 million and $22,295 million.\n[tables/table-2 Dividend History.html]\n\nThe dividend declared on June 10, 2025 was included in other current liabilities as of June 30, 2025.\nSTOCK PERFORMANCE\nCOMPARISON OF 5 YEAR CUMULATIVE TOTAL RETURN*\nAmong Microsoft Corporation, the S&P 500 Index\nand the NASDAQ Computer Index\n\nChart Title: Microsoft Corporation, S&P 500, and NASDAQ Computer Index Performance\nY-Axis: Dollar Value ($0 to $300)\nX-Axis: Dates (6/20 to 6/25)\n\nData Points & Trends:\n- All three indices start at a baseline of $100 on 6/20.\n- Microsoft Corporation (Square marker): Starts at $100, dips slightly on 6/21 (~$135), drops to ~$128 on 6/22, then rises sharply to finish at approximately $250 on 6/25. It shows the strongest overall growth among the three.\n- NASDAQ Computer (Circle marker): Starts at $100, peaks at ~$150 on 6/21, drops to its lowest point (~$118) on 6/22, then recovers strongly to end near $250 on 6/25. Its trajectory closely mirrors Microsoft's in the latter half.\n[images/image-3 Microsoft vs S&P 500 v.jpeg]\n\n\nThe table shows daily values for Microsoft Corporation, S&P 500, and NASDAQ Computer from June 20 to June 25, with each starting at 100.00 and ending at 255.13, 215.89, and 254.97 respectively.\n[tables/table-3 Stock Index Data.html]\n\n* $100 invested on 6/30/20 in stock or index, including reinvestment of dividends. Fiscal year ending June 30.\nNote About Forward-Looking Statements", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 5", + "metadata": { + "length": 3336, + "summary": "The passage covers Microsoft's common stock trading on NASDAQ under MSFT, with 77,014 registered holders as of July 24, 2025. It details share repurchase programs approved in 2021 and 2024, each authorizing up to $60 billion, with $57.3 billion remaining as of June 30, 2025. Quarterly repurchase data for fiscal years 2023-2025 shows totals of 31 shares ($13,000M) in 2025, 32 shares ($11,960M) in 2024, and 69 shares ($18,400M) in 2023. Dividends declared quarterly were $0.83 per share in 2025 and $0.75 in 2024, totaling $24,678 million and $22,295 million respectively. A stock performance chart compares 5-year cumulative total returns of Microsoft, S&P 500, and NASDAQ Computer Index from June 2020 to June 2025, with Microsoft ending at $255.13, S&P 500 at $215.89, and NASDAQ Computer at $254.97, all starting at $100.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Our", + "common", + "stock", + "traded", + "NASDAQ", + "Stock", + "Market", + "symbol", + "MSFT", + "On", + "July", + "24", + "2025", + "77", + "014", + "registered", + "holders", + "record", + "SHARE", + "REPURCHASES", + "AND", + "DIVIDENDS", + "Share", + "Repurchases", + "September", + "14", + "2021", + "Board", + "Directors", + "approved", + "share", + "repurchase", + "program", + "authorizing", + "60.0", + "billion", + "repurchases", + "This", + "commenced", + "November", + "completed", + "April", + "16", + "2024", + "completion", + "expiration", + "date", + "terminated", + "time", + "As", + "June", + "30", + "57.3", + "remained", + "We", + "repurchased", + "shares", + "programs", + "The", + "table", + "shows", + "quarterly", + "amounts", + "fiscal", + "years", + "2023", + "totals", + "31", + "13", + "000M", + "32", + "11", + "960M", + "69", + "18", + "400M", + "All", + "made", + "cash", + "resources", + "Shares", + "fourth", + "quarter", + "year", + "excludes", + "settle", + "employee", + "tax", + "withholding", + "related", + "vesting", + "awards", + "5.4", + "5.3", + "3.8", + "Dividends", + "declared", + "dividends", + "0.83", + "0.75", + "totaling", + "678", + "million", + "22", + "295", + "dividend", + "10", + "included", + "current", + "liabilities", + "STOCK", + "PERFORMANCE", + "COMPARISON", + "OF", + "YEAR", + "CUMULATIVE", + "TOTAL", + "RETURN", + "Among", + "Microsoft", + "Corporation", + "500", + "Index", + "Computer", + "Chart", + "Title", + "Performance", + "Axis", + "Dollar", + "Value", + "300", + "Dates", + "20", + "25", + "Data", + "Points", + "Trends", + "indices", + "start", + "baseline", + "100", + "Square", + "marker", + "Starts", + "dips", + "slightly", + "21", + "135", + "drops", + "128", + "rises", + "sharply", + "finish", + "approximately", + "250", + "It", + "strongest", + "growth", + "Circle", + "peaks", + "150", + "lowest", + "point", + "118", + "recovers", + "strongly", + "end", + "Its", + "trajectory", + "closely", + "mirrors", + "half", + "daily", + "values", + "starting", + "100.00", + "ending", + "255.13", + "215.89", + "254.97", + "invested", + "index", + "including", + "reinvestment", + "Fiscal", + "Note", + "About", + "Forward", + "Looking", + "Statements" + ], + "keywords": [ + "Share Repurchases", + "Dividends", + "Stock Performance" + ], + "connect_to": [ + { + "target": "f0fa5122-814d-55dd-aac9-1a40c56c0a8c", + "relation": "embeds", + "ref": "[tables/table-1 Share Repurchases.html]", + "position": { + "start": 1132, + "end": 1171 + } + }, + { + "target": "0e66fea7-c5b7-53f8-8471-b407613a66b3", + "relation": "embeds", + "ref": "[tables/table-2 Dividend History.html]", + "position": { + "start": 1940, + "end": 1978 + } + }, + { + "target": "77c3bd2d-5b47-5b5a-929b-e8988bc51db5", + "relation": "embeds", + "ref": "[images/image-3 Microsoft vs S&P 500 v.jpeg]", + "position": { + "start": 2907, + "end": 2951 + } + }, + { + "target": "b1e3c716-1628-5ef8-9628-f29f51a19862", + "relation": "embeds", + "ref": "[tables/table-3 Stock Index Data.html]", + "position": { + "start": 3148, + "end": 3186 + } + } + ] + } + }, + { + "chunk_id": "68222691-bb13-5233-9d7e-ebe6f0987fd0", + "type": "text", + "content": "This report includes estimates, projections, statements relating to our business plans, objectives, and expected operating results that are “forward-looking statements” within the meaning of the Private Securities Litigation Reform Act of 1995, Section 27A of the Securities Act of 1933, and Section 21E of the Securities Exchange Act of 1934. Forward-looking statements may appear throughout this report, including the following sections: “Business”, and “Management’s Discussion and Analysis of Financial Condition and Results of Operations”. These forward-looking statements generally are identified by the words “believe,” “project,” “expect,” “anticipate,” “estimate,” “intend,” “strategy,” “future,” “opportunity,” “plan,” “may,” “should,” “will,” “would,” “will be,” “will continue,” “will likely result,” and similar expressions. Forward-looking statements are based on current expectations and assumptions that are subject to risks and uncertainties that may cause actual results to differ materially. We describe risks and uncertainties that could cause actual results and events to differ materially in “Risk Factors,” “Management’s Discussion and Analysis of Financial Condition and Results of Operations,” and “Quantitative and Qualitative Disclosures About Market Risk” in our fiscal year 2025 Form 10-K. Readers are cautioned not to place undue reliance on forward-looking statements, which speak only as of the date they are made. We undertake no obligation to update or revise publicly any forward-looking statements, whether because of new information, future events, or otherwise.\nBUSINESS\nGENERAL\nMicrosoft is a technology company committed to making digital technology and artificial intelligence (“AI”) available broadly and doing so responsibly. Our mission is to empower every person and every organization on the planet to achieve more.\nWe develop and support a broad portfolio of technology solutions for individuals and businesses, focusing on secure, trusted, and innovative platforms and tools that meet evolving customer needs across cloud computing, productivity and collaboration, and personal computing. We strive to create opportunity, growth, and impact in every country around the world.\nAI is fundamentally transforming productivity for every individual, organization, and industry. Microsoft’s AI offerings span every layer of the technology stack, enabling transformative outcomes across sectors and unlocking opportunity for every country, community, and individual.\nWe believe AI should be as empowering as it is powerful, and we’re committed to designing and deploying AI responsibly with safety and security from the outset.\nWhat We Offer\nFounded in 1975, we develop and support software, services, devices, and solutions that deliver new value for customers and help people and businesses realize their full potential.\nWe offer an array of services, including cloud-based solutions that provide customers with software, services, platforms, and content, and we provide solution support and consulting services. We also deliver relevant online advertising to a global audience.\nOur products include operating systems, cross-device productivity and collaboration applications, server applications, business solution applications, desktop and server management tools, software development tools, and video games. We also design and sell devices, including PCs, tablets, gaming and entertainment consoles, other intelligent devices, and related accessories.\nDigital transformation and adoption of AI continues to revolutionize more business workstreams for organizations in every sector across the globe. For enterprises, digital technology empowers employees, optimizes operations, engages customers, and in some cases, changes the very core of products and services.\nThe Microsoft Cloud provides integration across the technology stack while offering openness, improving time to value, reducing costs, and increasing agility. Our cloud business benefits from three economies of scale: datacenters that deploy computational resources at significantly lower cost per unit than smaller ones; datacenters that coordinate and aggregate diverse customer, geographic, and application demand patterns, improving the utilization of computing, storage, and network resources; and multi-tenancy locations that lower application maintenance labor costs.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 6", + "metadata": { + "length": 4383, + "summary": "Microsoft is a technology company focused on AI and digital transformation, offering a broad portfolio of software, services, and devices. Its mission is to empower people and organizations globally. The Microsoft Cloud provides integrated solutions across the technology stack, benefiting from economies of scale in datacenters and multi-tenancy. The report includes forward-looking statements about business plans and objectives, subject to risks and uncertainties.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "This", + "report", + "includes", + "estimates", + "projections", + "statements", + "relating", + "business", + "plans", + "objectives", + "expected", + "operating", + "results", + "forward", + "meaning", + "Private", + "Securities", + "Litigation", + "Reform", + "Act", + "1995", + "Section", + "27A", + "1933", + "21E", + "Exchange", + "1934", + "Forward", + "including", + "sections", + "Business", + "Management", + "Discussion", + "Analysis", + "Financial", + "Condition", + "Results", + "Operations", + "These", + "generally", + "identified", + "words", + "project", + "expect", + "anticipate", + "estimate", + "intend", + "strategy", + "future", + "opportunity", + "plan", + "continue", + "result", + "similar", + "expressions", + "based", + "current", + "expectations", + "assumptions", + "subject", + "risks", + "uncertainties", + "actual", + "differ", + "materially", + "We", + "describe", + "events", + "Risk", + "Factors", + "Quantitative", + "Qualitative", + "Disclosures", + "About", + "Market", + "fiscal", + "year", + "2025", + "Form", + "10", + "Readers", + "cautioned", + "place", + "undue", + "reliance", + "speak", + "date", + "made", + "undertake", + "obligation", + "update", + "revise", + "publicly", + "information", + "BUSINESS", + "GENERAL", + "Microsoft", + "technology", + "company", + "committed", + "making", + "digital", + "artificial", + "intelligence", + "AI", + "broadly", + "responsibly", + "Our", + "mission", + "empower", + "person", + "organization", + "planet", + "achieve", + "develop", + "support", + "broad", + "portfolio", + "solutions", + "individuals", + "businesses", + "focusing", + "secure", + "trusted", + "innovative", + "platforms", + "tools", + "meet", + "evolving", + "customer", + "cloud", + "computing", + "productivity", + "collaboration", + "personal", + "strive", + "create", + "growth", + "impact", + "country", + "world", + "fundamentally", + "transforming", + "individual", + "industry", + "offerings", + "span", + "layer", + "stack", + "enabling", + "transformative", + "outcomes", + "sectors", + "unlocking", + "community", + "empowering", + "powerful", + "designing", + "deploying", + "safety", + "security", + "outset", + "What", + "Offer", + "Founded", + "1975", + "software", + "services", + "devices", + "deliver", + "customers", + "people", + "realize", + "full", + "potential", + "offer", + "array", + "provide", + "content", + "solution", + "consulting", + "relevant", + "online", + "advertising", + "global", + "audience", + "products", + "include", + "systems", + "cross", + "device", + "applications", + "server", + "desktop", + "management", + "development", + "video", + "games", + "design", + "sell", + "PCs", + "tablets", + "gaming", + "entertainment", + "consoles", + "intelligent", + "related", + "accessories", + "Digital", + "transformation", + "adoption", + "continues", + "revolutionize", + "workstreams", + "organizations", + "sector", + "globe", + "For", + "enterprises", + "empowers", + "employees", + "optimizes", + "operations", + "engages", + "cases", + "core", + "The", + "Cloud", + "integration", + "offering", + "openness", + "improving", + "time", + "reducing", + "costs", + "increasing", + "agility", + "benefits", + "economies", + "scale", + "datacenters", + "deploy", + "computational", + "resources", + "significantly", + "lower", + "cost", + "unit", + "smaller", + "coordinate", + "aggregate", + "diverse", + "geographic", + "application", + "demand", + "patterns", + "utilization", + "storage", + "network", + "multi", + "tenancy", + "locations", + "maintenance", + "labor" + ], + "keywords": [ + "AI", + "Cloud", + "Digital Transformation" + ], + "connect_to": [] + } + }, + { + "chunk_id": "7658b3a7-805f-5e8b-b1c7-ac40942034cf", + "type": "text", + "content": "We prioritize security above all else and we offer our customers integrated AI-driven products addressing security, compliance, identity, management, and privacy across customers’ multi-cloud, application, and device assets.\nThe Ambitions That Drive Us\nTo achieve our vision, our research and development efforts focus on three interconnected ambitions:\n• Reinvent productivity and business processes to help organizations and individuals work and collaborate more securely and efficiently.\n• Build the intelligent cloud and intelligent edge platform to provide a foundation for our customers’ digital workloads including hybrid consistency, developer productivity, data and AI capabilities, and trusted security and compliance.\n• Create more personal computing to enable users to interact with technology in more intuitive, engaging, and dynamic ways.\nOur Future Opportunity\nWe are focused on helping customers use the breadth and depth of the Microsoft Cloud to get the most value out of their digital spend while leading the AI platform wave across our solution areas. We continue to develop complete, intelligent solutions for our customers that empower people to be productive and collaborate, while safeguarding businesses and simplifying IT management. Our goal is to lead the industry in several distinct areas of technology over the long term, which we expect will translate to sustained growth. We are investing significant resources in:\n• Transforming the workplace to deliver new, modern, modular business applications, drive deeper insights, and improve how people communicate, collaborate, learn, work, and interact with one another.\n• Building and running cloud-based services in ways that utilize ubiquitous computing to unleash new experiences and opportunities for businesses and individuals.\n• Applying AI and ambient intelligence to drive insights, revolutionize many types of work and business processes, and provide substantive productivity gains using Microsoft 365 Copilot and agents.\n• Providing training on generative AI and greater access to digital learning and resources through skilling programs and initiatives, grants, and LinkedIn learning pathways.\n• Inventing new gaming experiences that bring people together around their shared love for games on any device and pushing the boundaries of innovation with console and PC gaming.\n• Leveraging Windows to fuel our cloud business, grow our share of the PC market, and drive increased engagement with our services like Microsoft Edge, Bing, Copilot, Microsoft Teams, Microsoft 365 Consumer, Xbox Game Pass, and more.\n• Tackling security from all angles with our integrated, end-to-end solutions spanning security, compliance, identity, and management, across all clouds and platforms.\nOur future growth depends on our ability to transcend current product category definitions, business models, and sales motions.\nCommitment to Sustainability\nMicrosoft is committed to sustainability and our approach to addressing climate change starts with the sustainability of our own business. In 2020, we announced goals to become a carbon negative, water positive, and zero waste company by 2030. Since announcing these goals, we have made meaningful progress while having seen major changes in both the technology sector and in our understanding of what it will take to meet our goals. Progress toward these goals can be found in our annual Environmental Sustainability Report.\nOPERATING SEGMENTS\nWe operate our business and report our financial performance using three segments: Productivity and Business Processes, Intelligent Cloud, and More Personal Computing. Our segments provide management with a comprehensive financial view of our key businesses. The segments enable the alignment of strategies and objectives across the development, sales, marketing, and services organizations, and they provide a framework for timely and rational allocation of resources within businesses.\nIn August 2024, we announced changes to the composition of our segments. These changes align our segments with how we currently manage our business, most notably bringing the commercial components of Microsoft 365 together in the Productivity and Business Processes segment. Beginning in fiscal year 2025, the information that our chief operating decision maker is regularly provided and reviews for purposes of allocating resources and assessing performance reflects these segment changes.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 7", + "metadata": { + "length": 4445, + "summary": "Microsoft prioritizes security and offers AI-driven products for security, compliance, identity, and management across multi-cloud, application, and device assets. Their R&D focuses on reinventing productivity, building the intelligent cloud and edge platform, and creating more personal computing. They aim to lead in AI, cloud services, workplace transformation, gaming, and Windows. Microsoft is committed to sustainability with goals to be carbon negative, water positive, and zero waste by 2030. The company operates in three segments: Productivity and Business Processes, Intelligent Cloud, and More Personal Computing, with changes announced in August 2024 to align with current business management.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "We", + "prioritize", + "security", + "offer", + "customers", + "integrated", + "AI", + "driven", + "products", + "addressing", + "compliance", + "identity", + "management", + "privacy", + "multi", + "cloud", + "application", + "device", + "assets", + "The", + "Ambitions", + "That", + "Drive", + "Us", + "To", + "achieve", + "vision", + "research", + "development", + "efforts", + "focus", + "interconnected", + "ambitions", + "Reinvent", + "productivity", + "business", + "processes", + "organizations", + "individuals", + "work", + "collaborate", + "securely", + "efficiently", + "Build", + "intelligent", + "edge", + "platform", + "provide", + "foundation", + "digital", + "workloads", + "including", + "hybrid", + "consistency", + "developer", + "data", + "capabilities", + "trusted", + "Create", + "personal", + "computing", + "enable", + "users", + "interact", + "technology", + "intuitive", + "engaging", + "dynamic", + "ways", + "Our", + "Future", + "Opportunity", + "focused", + "helping", + "breadth", + "depth", + "Microsoft", + "Cloud", + "spend", + "leading", + "wave", + "solution", + "areas", + "continue", + "develop", + "complete", + "solutions", + "empower", + "people", + "productive", + "safeguarding", + "businesses", + "simplifying", + "IT", + "goal", + "lead", + "industry", + "distinct", + "long", + "term", + "expect", + "translate", + "sustained", + "growth", + "investing", + "significant", + "resources", + "Transforming", + "workplace", + "deliver", + "modern", + "modular", + "applications", + "drive", + "deeper", + "insights", + "improve", + "communicate", + "learn", + "Building", + "running", + "based", + "services", + "utilize", + "ubiquitous", + "unleash", + "experiences", + "opportunities", + "Applying", + "ambient", + "intelligence", + "revolutionize", + "types", + "substantive", + "gains", + "365", + "Copilot", + "agents", + "Providing", + "training", + "generative", + "greater", + "access", + "learning", + "skilling", + "programs", + "initiatives", + "grants", + "LinkedIn", + "pathways", + "Inventing", + "gaming", + "bring", + "shared", + "love", + "games", + "pushing", + "boundaries", + "innovation", + "console", + "PC", + "Leveraging", + "Windows", + "fuel", + "grow", + "share", + "market", + "increased", + "engagement", + "Edge", + "Bing", + "Teams", + "Consumer", + "Xbox", + "Game", + "Pass", + "Tackling", + "angles", + "end", + "spanning", + "clouds", + "platforms", + "future", + "depends", + "ability", + "transcend", + "current", + "product", + "category", + "definitions", + "models", + "sales", + "motions", + "Commitment", + "Sustainability", + "committed", + "sustainability", + "approach", + "climate", + "change", + "starts", + "In", + "2020", + "announced", + "goals", + "carbon", + "negative", + "water", + "positive", + "waste", + "company", + "2030", + "Since", + "announcing", + "made", + "meaningful", + "progress", + "major", + "sector", + "understanding", + "meet", + "Progress", + "found", + "annual", + "Environmental", + "Report", + "OPERATING", + "SEGMENTS", + "operate", + "report", + "financial", + "performance", + "segments", + "Productivity", + "Business", + "Processes", + "Intelligent", + "More", + "Personal", + "Computing", + "comprehensive", + "view", + "key", + "alignment", + "strategies", + "objectives", + "marketing", + "framework", + "timely", + "rational", + "allocation", + "August", + "2024", + "composition", + "These", + "align", + "manage", + "notably", + "bringing", + "commercial", + "components", + "segment", + "Beginning", + "fiscal", + "year", + "2025", + "information", + "chief", + "operating", + "decision", + "maker", + "regularly", + "provided", + "reviews", + "purposes", + "allocating", + "assessing", + "reflects" + ], + "keywords": [ + "AI", + "Cloud", + "Security" + ], + "connect_to": [] + } + }, + { + "chunk_id": "fd207155-8d31-50ce-acfd-e686c29275d2", + "type": "text", + "content": "Additional information on our operating segments and geographic and product information is contained in Note 18 – Segment Information and Geographic Data of the Notes to Financial Statements.\nOur reportable segments are described below.\nProductivity and Business Processes\nOur Productivity and Business Processes segment consists of products and services in our portfolio of productivity, communication, and information services, spanning a variety of devices and platforms. This segment primarily comprises:\n• Microsoft 365 Commercial products and cloud services, including Microsoft 365 Commercial cloud, comprising Microsoft 365 Commercial, Enterprise Mobility + Security, the cloud portion of Windows Commercial, the per-user portion of Power BI, Exchange, SharePoint, Microsoft Teams, Microsoft 365 Security and Compliance, and Microsoft 365 Copilot; and Microsoft 365 Commercial products, comprising Windows Commercial on-premises and Office licensed on-premises.\n• Microsoft 365 Consumer products and cloud services, including Microsoft 365 Consumer subscriptions, Office licensed on-premises, and other consumer services.\n• LinkedIn, including Talent Solutions, Marketing Solutions, Premium Subscriptions, and Sales Solutions.\n• Dynamics products and cloud services, including Dynamics 365, comprising a set of intelligent, cloud-based applications across ERP, CRM, Power Apps, and Power Automate; and on-premises ERP and CRM applications.\nMicrosoft 365 Commercial Products and Cloud Services\nMicrosoft 365 Commercial is an AI-powered business and productivity solutions platform that brings together Office, Windows, Microsoft 365 Copilot, and Enterprise Mobility + Security to help organizations empower their employees. Growth depends on our ability to reach new users in new markets such as frontline workers, small and medium businesses, and growth markets, as well as add AI-enabled tools, features, and agentic scenarios to our core product and service offerings across communication, collaboration, analytics, security, compliance, and other AI business productivity categories. Microsoft 365 Commercial revenue is mainly affected by a combination of continued installed base growth and average revenue per user expansion, as well as the continued shift from Office licensed on-premises to Microsoft 365.\nMicrosoft 365 Consumer Products and Cloud Services\nMicrosoft 365 Consumer is designed to increase personal productivity and creativity through a range of products and services. Growth depends on our ability to reach new users, add value to our core product set with new features including AI tools, and continue to expand our product and service offerings into new markets. Microsoft 365 Consumer cloud revenue and Office Consumer products revenue is mainly affected by the percentage of customers that buy Office with their new devices and the continued shift from Office licensed on-premises to Microsoft 365 Consumer subscriptions. Microsoft 365 Consumer cloud revenue is also affected by the demand for communication and storage through Outlook.com and OneDrive, which is largely driven by subscriptions and advertising.\nLinkedIn\nLinkedIn connects the world’s professionals to make them more productive and successful and transforms the way companies hire, market, sell, and learn. In addition to LinkedIn’s free services, LinkedIn offers monetized solutions designed to offer AI-enabled insights and productivity: Talent Solutions, Marketing Solutions, Premium Subscriptions, and Sales Solutions. Growth will depend on our ability to increase LinkedIn member engagement on the platform and our ability to continue offering insight and AI-enabled services that provide value for our members and customers. LinkedIn revenue is mainly affected by demand from enterprises and professionals for subscriptions to Talent Solutions, Sales Solutions, and Premium Subscriptions offerings, as well as member engagement and the quality of the sponsored content delivered to those members to drive Marketing Solutions.\nDynamics Products and Cloud Services", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 8", + "metadata": { + "length": 4068, + "summary": "The text describes Microsoft's reportable segments, focusing on Productivity and Business Processes. This segment includes Microsoft 365 Commercial and Consumer products and cloud services, LinkedIn, and Dynamics products. Key offerings include AI-powered productivity tools, cloud subscriptions, and professional networking solutions. Revenue growth depends on user adoption, AI features, and shifts from on-premises to cloud services.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Additional", + "information", + "operating", + "segments", + "geographic", + "product", + "contained", + "Note", + "18", + "Segment", + "Information", + "Geographic", + "Data", + "Notes", + "Financial", + "Statements", + "Our", + "reportable", + "Productivity", + "Business", + "Processes", + "segment", + "consists", + "products", + "services", + "portfolio", + "productivity", + "communication", + "spanning", + "variety", + "devices", + "platforms", + "This", + "primarily", + "comprises", + "Microsoft", + "365", + "Commercial", + "cloud", + "including", + "comprising", + "Enterprise", + "Mobility", + "Security", + "portion", + "Windows", + "user", + "Power", + "BI", + "Exchange", + "SharePoint", + "Teams", + "Compliance", + "Copilot", + "premises", + "Office", + "licensed", + "Consumer", + "subscriptions", + "consumer", + "LinkedIn", + "Talent", + "Solutions", + "Marketing", + "Premium", + "Subscriptions", + "Sales", + "Dynamics", + "set", + "intelligent", + "based", + "applications", + "ERP", + "CRM", + "Apps", + "Automate", + "Products", + "Cloud", + "Services", + "AI", + "powered", + "business", + "solutions", + "platform", + "brings", + "organizations", + "empower", + "employees", + "Growth", + "depends", + "ability", + "reach", + "users", + "markets", + "frontline", + "workers", + "small", + "medium", + "businesses", + "growth", + "add", + "enabled", + "tools", + "features", + "agentic", + "scenarios", + "core", + "service", + "offerings", + "collaboration", + "analytics", + "security", + "compliance", + "categories", + "revenue", + "affected", + "combination", + "continued", + "installed", + "base", + "average", + "expansion", + "shift", + "designed", + "increase", + "personal", + "creativity", + "range", + "continue", + "expand", + "percentage", + "customers", + "buy", + "demand", + "storage", + "Outlook", + "OneDrive", + "largely", + "driven", + "advertising", + "connects", + "world", + "professionals", + "make", + "productive", + "successful", + "transforms", + "companies", + "hire", + "market", + "sell", + "learn", + "In", + "addition", + "free", + "offers", + "monetized", + "offer", + "insights", + "depend", + "member", + "engagement", + "offering", + "insight", + "provide", + "members", + "enterprises", + "quality", + "sponsored", + "content", + "delivered", + "drive" + ], + "keywords": [ + "Productivity", + "Business Processes", + "Cloud Services" + ], + "connect_to": [] + } + }, + { + "chunk_id": "fec4368e-7a2b-558f-b2a2-a6a51a7e65e2", + "type": "text", + "content": "Dynamics provides cloud-based and on-premises business solutions for financial management, enterprise resource planning (“ERP”), customer relationship management (“CRM”), and supply chain management, as well as agentic AI and other low code application development platforms, for small and medium businesses, large organizations, and divisions of global enterprises. Dynamics revenue is driven by the number of users licensed and applications consumed, expansion of average revenue per user, and the continued shift to Dynamics 365, a unified set of cloud-based intelligent business applications, including our low code development platforms, such as Power Apps and Power Automate.\nCompetition\nCompetitors to Office include software and global application vendors, web-based and mobile application companies, AI-first application companies, as well as local application developers. We compete by providing secure, integrated industry-specific, and easy-to-use productivity and collaboration tools and services that create comprehensive solutions and work well with technologies our customers already have both on-premises or in the cloud.\nWindows faces competition from various software products and from alternative platforms and devices. Microsoft Defender for Endpoint competes with endpoint security solution providers.\nOur Enterprise Mobility + Security offerings compete with products from a range of competitors including identity vendors, security solution vendors, and numerous other security point solution vendors.\nLinkedIn faces competition from online professional networks; recruiting, talent management, and human resource services companies; job boards; companies that provide learning and development products and services; online and offline outlets that generate revenue from advertisers and marketers; and online and offline outlets for companies with lead generation and customer intelligence and insights.\nDynamics competes with cloud-based and on-premises business solution providers.\nIntelligent Cloud\nOur Intelligent Cloud segment consists of our public, private, and hybrid server products and cloud services that power modern business and developers. This segment primarily comprises:\n• Server products and cloud services, including Azure and other cloud services, comprising cloud and AI consumption-based services, GitHub cloud services, Nuance Healthcare cloud services, virtual desktop offerings, and other cloud services; and Server products, comprising SQL Server, Windows Server, Visual Studio, System Center, related Client Access Licenses (“CALs”), and other on-premises offerings.\n• Enterprise and partner services, including Enterprise Support Services, Industry Solutions, Nuance professional services, Microsoft Partner Network, and Learning Experience.\nServer Products and Cloud Services\nAzure is a comprehensive set of cloud services that offer developers, IT professionals, and enterprises freedom to build, deploy, and manage applications on any platform or device. Customers can use Azure through our global network of datacenters for computing, networking, storage, mobile and web application services, AI, Internet of Things, cognitive services, and machine learning. Azure enables customers to devote more resources to development and use of applications that benefit their organizations, rather than managing on-premises hardware and software. Azure revenue is mainly affected by infrastructure-as-a-service and platform-as-a-service consumption-based services.\nAzure AI offerings provide a competitive advantage as companies seek ways to optimize and scale their business with AI. We offer supercomputing power for AI at scale to run large workloads, complemented by our rapidly expanding portfolio of AI cloud services (including the latest models) and hardware, which includes custom-built silicon and strong partnerships with chip manufacturers. Azure AI Foundry is a unified platform for developers to design, customize, and manage AI applications and agents.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 9", + "metadata": { + "length": 4013, + "summary": "The passage describes Microsoft's Dynamics business solutions for financial management, ERP, CRM, and supply chain, with revenue driven by user licenses and cloud shift. It outlines competition for Office, Windows, Enterprise Mobility, LinkedIn, and Dynamics. The Intelligent Cloud segment includes Azure and server products, with Azure offering AI services and competing via infrastructure and platform services.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Dynamics", + "cloud", + "based", + "premises", + "business", + "solutions", + "financial", + "management", + "enterprise", + "resource", + "planning", + "ERP", + "customer", + "relationship", + "CRM", + "supply", + "chain", + "agentic", + "AI", + "low", + "code", + "application", + "development", + "platforms", + "small", + "medium", + "businesses", + "large", + "organizations", + "divisions", + "global", + "enterprises", + "revenue", + "driven", + "number", + "users", + "licensed", + "applications", + "consumed", + "expansion", + "average", + "user", + "continued", + "shift", + "365", + "unified", + "set", + "intelligent", + "including", + "Power", + "Apps", + "Automate", + "Competition", + "Competitors", + "Office", + "include", + "software", + "vendors", + "web", + "mobile", + "companies", + "local", + "developers", + "We", + "compete", + "providing", + "secure", + "integrated", + "industry", + "specific", + "easy", + "productivity", + "collaboration", + "tools", + "services", + "create", + "comprehensive", + "work", + "technologies", + "customers", + "Windows", + "faces", + "competition", + "products", + "alternative", + "devices", + "Microsoft", + "Defender", + "Endpoint", + "competes", + "endpoint", + "security", + "solution", + "providers", + "Our", + "Enterprise", + "Mobility", + "Security", + "offerings", + "range", + "competitors", + "identity", + "numerous", + "point", + "LinkedIn", + "online", + "professional", + "networks", + "recruiting", + "talent", + "human", + "job", + "boards", + "provide", + "learning", + "offline", + "outlets", + "generate", + "advertisers", + "marketers", + "lead", + "generation", + "intelligence", + "insights", + "Intelligent", + "Cloud", + "segment", + "consists", + "public", + "private", + "hybrid", + "server", + "power", + "modern", + "This", + "primarily", + "comprises", + "Server", + "Azure", + "comprising", + "consumption", + "GitHub", + "Nuance", + "Healthcare", + "virtual", + "desktop", + "SQL", + "Visual", + "Studio", + "System", + "Center", + "related", + "Client", + "Access", + "Licenses", + "CALs", + "partner", + "Support", + "Services", + "Industry", + "Solutions", + "Partner", + "Network", + "Learning", + "Experience", + "Products", + "offer", + "IT", + "professionals", + "freedom", + "build", + "deploy", + "manage", + "platform", + "device", + "Customers", + "network", + "datacenters", + "computing", + "networking", + "storage", + "Internet", + "Things", + "cognitive", + "machine", + "enables", + "devote", + "resources", + "benefit", + "managing", + "hardware", + "affected", + "infrastructure", + "service", + "competitive", + "advantage", + "seek", + "ways", + "optimize", + "scale", + "supercomputing", + "run", + "workloads", + "complemented", + "rapidly", + "expanding", + "portfolio", + "latest", + "models", + "includes", + "custom", + "built", + "silicon", + "strong", + "partnerships", + "chip", + "manufacturers", + "Foundry", + "design", + "customize", + "agents" + ], + "keywords": [ + "Dynamics", + "Azure", + "Competition" + ], + "connect_to": [] + } + }, + { + "chunk_id": "c908aa9c-3f43-53c0-8d35-2364ffea171d", + "type": "text", + "content": "Our server products are designed to make IT professionals, developers, and their systems more productive and efficient. Server software is integrated server infrastructure and middleware designed to support software applications built on the Windows Server operating system. This includes the server platform, database, business intelligence, storage, management and operations, virtualization, service-oriented architecture platform, security, and identity software. We also license standalone and software development lifecycle tools for software architects, developers, testers, and project managers. Server products revenue is mainly affected by purchases through volume licensing programs, licenses sold to OEMs, and retail packaged products. CALs provide access rights to certain server products, including SQL Server and Windows Server, and revenue is reported along with the associated server product.\nGitHub and Nuance Healthcare include both cloud and on-premises offerings. GitHub provides a collaboration platform for developers to manage code and incorporate AI and agent-based tools across the software development lifecycle. Nuance Healthcare provides AI solutions to the healthcare industry.\nEnterprise and Partner Services\nEnterprise and partner services, including Enterprise Support Services, Industry Solutions, Nuance professional services, Microsoft Partner Network, and Learning Experience, assist customers in developing, deploying, and managing Microsoft server solutions, Microsoft desktop solutions, and Nuance conversational AI and ambient intelligent solutions, along with providing training and certification to developers and IT professionals on various Microsoft products.\nCompetition\nAzure faces diverse competition from cloud service providers and open source offerings. Azure’s competitive advantage includes enabling a hybrid cloud, allowing deployment of existing datacenters with our public cloud into a single, cohesive infrastructure, and the ability to run at a scale that meets the needs of businesses of all sizes and complexities. Our AI offerings compete with AI products from hyperscalers, as well as products from other emerging competitors and other open source offerings, many of which are also current or potential partners. Our Azure Security offerings include our cloud security solution and security information and event management solution, which compete with providers in the cybersecurity and cloud security space. We believe our cloud’s global scale, coupled with our broad portfolio of identity and security solutions, allows us to effectively solve complex cybersecurity challenges for our customers and differentiates us from the competition.\nOur server products face competition from a wide variety of server operating systems and applications offered by companies with a range of market approaches. Vertically integrated computer manufacturers offer their own versions of the Unix operating system preinstalled on server hardware and nearly all computer manufacturers offer server hardware for the Linux operating system.\nWe compete to provide enterprise-wide computing and point solutions with numerous commercial software vendors that offer solutions and middleware technology platforms, software applications for connectivity, security, hosting, database, and e-business servers.\nOur database, business intelligence, and data warehousing solutions offerings compete with products from providers in the data and analytics industry. Our system management solutions compete with server management and server virtualization platform providers. Our products for software developers compete against offerings from major technology providers, as well as open source alternatives.\nWe believe our server products provide customers with advantages in performance, total costs of ownership, and productivity by delivering superior applications, development tools, compatibility with a broad base of hardware and software applications, security, and manageability.\nOur Enterprise and partner services business competes with a wide range of companies that provide strategy and business planning, application development, and infrastructure services, including multinational consulting firms and small niche businesses focused on specific technologies.\nMore Personal Computing\nOur More Personal Computing segment consists of products and services that put customers at the center of the experience with our technology. This segment primarily comprises:", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 10", + "metadata": { + "length": 4504, + "summary": "The passage discusses Microsoft's server products, including server software, databases, and development tools, which are sold through volume licensing, OEMs, and retail. It also covers GitHub and Nuance Healthcare offerings, enterprise services, and competition faced by Azure and server products from various cloud providers, open source alternatives, and commercial vendors. The summary highlights competitive advantages in hybrid cloud, AI, security, and performance.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Our", + "server", + "products", + "designed", + "make", + "IT", + "professionals", + "developers", + "systems", + "productive", + "efficient", + "Server", + "software", + "integrated", + "infrastructure", + "middleware", + "support", + "applications", + "built", + "Windows", + "operating", + "system", + "This", + "includes", + "platform", + "database", + "business", + "intelligence", + "storage", + "management", + "operations", + "virtualization", + "service", + "oriented", + "architecture", + "security", + "identity", + "We", + "license", + "standalone", + "development", + "lifecycle", + "tools", + "architects", + "testers", + "project", + "managers", + "revenue", + "affected", + "purchases", + "volume", + "licensing", + "programs", + "licenses", + "sold", + "OEMs", + "retail", + "packaged", + "CALs", + "provide", + "access", + "rights", + "including", + "SQL", + "reported", + "product", + "GitHub", + "Nuance", + "Healthcare", + "include", + "cloud", + "premises", + "offerings", + "collaboration", + "manage", + "code", + "incorporate", + "AI", + "agent", + "based", + "solutions", + "healthcare", + "industry", + "Enterprise", + "Partner", + "Services", + "partner", + "services", + "Support", + "Industry", + "Solutions", + "professional", + "Microsoft", + "Network", + "Learning", + "Experience", + "assist", + "customers", + "developing", + "deploying", + "managing", + "desktop", + "conversational", + "ambient", + "intelligent", + "providing", + "training", + "certification", + "Competition", + "Azure", + "faces", + "diverse", + "competition", + "providers", + "open", + "source", + "competitive", + "advantage", + "enabling", + "hybrid", + "allowing", + "deployment", + "existing", + "datacenters", + "public", + "single", + "cohesive", + "ability", + "run", + "scale", + "meets", + "businesses", + "sizes", + "complexities", + "compete", + "hyperscalers", + "emerging", + "competitors", + "current", + "potential", + "partners", + "Security", + "solution", + "information", + "event", + "cybersecurity", + "space", + "global", + "coupled", + "broad", + "portfolio", + "effectively", + "solve", + "complex", + "challenges", + "differentiates", + "face", + "wide", + "variety", + "offered", + "companies", + "range", + "market", + "approaches", + "Vertically", + "computer", + "manufacturers", + "offer", + "versions", + "Unix", + "preinstalled", + "hardware", + "Linux", + "enterprise", + "computing", + "point", + "numerous", + "commercial", + "vendors", + "technology", + "platforms", + "connectivity", + "hosting", + "servers", + "data", + "warehousing", + "analytics", + "major", + "alternatives", + "advantages", + "performance", + "total", + "costs", + "ownership", + "productivity", + "delivering", + "superior", + "compatibility", + "base", + "manageability", + "competes", + "strategy", + "planning", + "application", + "multinational", + "consulting", + "firms", + "small", + "niche", + "focused", + "specific", + "technologies", + "More", + "Personal", + "Computing", + "segment", + "consists", + "put", + "center", + "experience", + "primarily", + "comprises" + ], + "keywords": [ + "Server Products", + "Azure", + "Competition" + ], + "connect_to": [] + } + }, + { + "chunk_id": "f1b23c94-6032-5471-9374-749c14bc6491", + "type": "text", + "content": "• Windows and Devices, including Windows OEM licensing (Windows Pro and non-Pro licenses sold through the OEM channel) and Devices, comprising Surface and PC accessories.\n• Gaming, including Xbox hardware and Xbox content and services, comprising first- and third-party content (including games and in-game content), Xbox Game Pass and other subscriptions, Xbox Cloud Gaming, advertising, and other cloud services.\n• Search and news advertising, comprising Bing and Copilot, Microsoft News, Microsoft Edge, and third-party affiliates.\nWindows and Devices\nThe Windows operating system is designed to deliver a more personal computing experience for users by enabling consistency of experience, applications, and information across their devices. Windows OEM revenue is impacted significantly by the number of Windows operating system licenses purchased by OEMs, which they pre-install on the devices they sell. In addition to computing device market volume, Windows OEM revenue is impacted by:\n• The mix of computing devices based on form factor and screen size.\n• Differences in device market demand between developed markets and growth markets.\n• Growth of the AI PC category.\n• Attachment of Windows to devices shipped.\n• Customer mix between consumer, small and medium businesses, and large enterprises.\n• Changes in inventory levels in the OEM channel.\n• Pricing changes and promotions, pricing variation that occurs when the mix of devices manufactured shifts from local and regional system builders to large multinational OEMs, and different pricing of Windows versions licensed.\n• Constraints in the supply chain of device components.\n• Piracy.\nWe design and sell devices, such as Surface (including Copilot+ PCs) and PC accessories. Our devices are designed to enable people and organizations to connect to the people and content that matter most using Windows and integrated Microsoft products and services. Surface is designed to help organizations, students, and consumers be more productive. Growth in Devices is dependent on total PC shipments, the ability to attract new customers, our product roadmap, and expanding into new categories.\nGaming\nMicrosoft is expanding how billions of people globally access and play video games on PC, console, mobile, and cloud. Our game content is developed through a collection of first-party studios creating iconic and differentiated gaming experiences. We continue to invest in gaming studios and content to expand our intellectual property roadmap and leverage new content creators. These unique gaming experiences are the cornerstone of Xbox Game Pass, a subscription service and gaming community with access to a curated library of first- and third-party titles.\nThe gamer remains at the heart of the Xbox ecosystem. We are identifying new opportunities to attract gamers across a variety of different end points through our first- and third-party content and business diversification across subscriptions, ads, and digital stores. We’ve seen new devices from third-party manufacturers along with key PC and mobile end points that help us empower gamers to play in a way that is most convenient to them. We are focused on growing the platform and expanding to new ecosystems to engage as many gamers as possible.\nXbox enables people to connect and share online gaming experiences that are accessible on Xbox consoles, Windows-enabled devices, and other devices. Xbox is designed to benefit users by providing access to a network of certified applications and services and to benefit our developer and partner ecosystems by providing access to a large customer base. Xbox revenue is mainly affected by subscriptions and sales of first- and third-party content, as well as advertising. Growth of our Gaming business is determined by the overall active user base through Xbox enabled content, availability of games, providing exclusive game content that gamers seek, the computational power and reliability of the devices used to access our content and services, and the ability to create new experiences.\nSearch and News Advertising", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 11", + "metadata": { + "length": 4086, + "summary": "The text describes three major business segments of Microsoft: Windows and Devices, Gaming, and Search and News Advertising. Windows and Devices covers OEM licensing, Surface devices, and PC accessories, with revenue impacted by market mix, AI PCs, and supply chain. Gaming includes Xbox hardware, content, Game Pass subscriptions, and cloud gaming, focusing on expanding the player base across platforms. Search and News Advertising involves Bing, Copilot, Microsoft News, and Edge, generating revenue through ads.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Windows", + "Devices", + "including", + "OEM", + "licensing", + "Pro", + "licenses", + "sold", + "channel", + "comprising", + "Surface", + "PC", + "accessories", + "Gaming", + "Xbox", + "hardware", + "content", + "services", + "party", + "games", + "game", + "Game", + "Pass", + "subscriptions", + "Cloud", + "advertising", + "cloud", + "Search", + "news", + "Bing", + "Copilot", + "Microsoft", + "News", + "Edge", + "affiliates", + "The", + "operating", + "system", + "designed", + "deliver", + "personal", + "computing", + "experience", + "users", + "enabling", + "consistency", + "applications", + "information", + "devices", + "revenue", + "impacted", + "significantly", + "number", + "purchased", + "OEMs", + "pre", + "install", + "sell", + "In", + "addition", + "device", + "market", + "volume", + "mix", + "based", + "form", + "factor", + "screen", + "size", + "Differences", + "demand", + "developed", + "markets", + "growth", + "Growth", + "AI", + "category", + "Attachment", + "shipped", + "Customer", + "consumer", + "small", + "medium", + "businesses", + "large", + "enterprises", + "Changes", + "inventory", + "levels", + "Pricing", + "promotions", + "pricing", + "variation", + "occurs", + "manufactured", + "shifts", + "local", + "regional", + "builders", + "multinational", + "versions", + "licensed", + "Constraints", + "supply", + "chain", + "components", + "Piracy", + "We", + "design", + "PCs", + "Our", + "enable", + "people", + "organizations", + "connect", + "matter", + "integrated", + "products", + "students", + "consumers", + "productive", + "dependent", + "total", + "shipments", + "ability", + "attract", + "customers", + "product", + "roadmap", + "expanding", + "categories", + "billions", + "globally", + "access", + "play", + "video", + "console", + "mobile", + "collection", + "studios", + "creating", + "iconic", + "differentiated", + "gaming", + "experiences", + "continue", + "invest", + "expand", + "intellectual", + "property", + "leverage", + "creators", + "These", + "unique", + "cornerstone", + "subscription", + "service", + "community", + "curated", + "library", + "titles", + "gamer", + "remains", + "heart", + "ecosystem", + "identifying", + "opportunities", + "gamers", + "variety", + "end", + "points", + "business", + "diversification", + "ads", + "digital", + "stores", + "ve", + "manufacturers", + "key", + "empower", + "convenient", + "focused", + "growing", + "platform", + "ecosystems", + "engage", + "enables", + "share", + "online", + "accessible", + "consoles", + "enabled", + "benefit", + "providing", + "network", + "certified", + "developer", + "partner", + "customer", + "base", + "affected", + "sales", + "determined", + "active", + "user", + "availability", + "exclusive", + "seek", + "computational", + "power", + "reliability", + "create", + "Advertising" + ], + "keywords": [ + "Windows", + "Gaming", + "Search" + ], + "connect_to": [] + } + }, + { + "chunk_id": "e5a8027a-aa21-5098-92c3-ef48c9a513ae", + "type": "text", + "content": "Our Search and news advertising business is designed to deliver relevant search, native, and display advertising to a global audience. Microsoft Copilot is a digital companion designed to inform, entertain, and inspire. Our Microsoft Edge browser and Bing search engine with Copilot are key tools to enable user acquisition and engagement, while our technology platform enables accelerated delivery of digital advertising solutions. In addition to first-party tools, we have several partnerships with companies through which we provide and monetize search offerings. Growth depends on our ability to attract new users, understand intent, and match intent with relevant content on advertising offerings.\nCompetition\nWindows faces competition from various software products and from alternative platforms and devices. We believe Windows competes effectively by giving customers choice, value, flexibility, security, an easy-to-use interface, and compatibility with a broad range of hardware and software applications, including those that enable productivity.\nDevices face competition from various computer, tablet, and hardware manufacturers who offer a unique combination of high-quality industrial design and innovative technologies across various price points. Many of these manufacturers are also current or potential partners and customers, including our Windows OEMs.\nXbox and our cloud gaming services face competition from various online gaming ecosystems and game streaming services. We also compete with other providers of entertainment services such as video streaming platforms. Our gaming platform competes with other console platforms. We believe our gaming platform is effectively positioned against, and uniquely differentiated from, competitive products and services based on significant innovation in hardware architecture, user interface, developer tools, online gaming and entertainment services, and continued strong content from our own first-party game franchises as well as other digital content offerings.\nOur Search and news advertising business competes with search engines, and a wide array of websites, social platforms, and portals that provide content and online offerings to end users.\nHUMAN CAPITAL RESOURCES\nAs of June 30, 2025, we employed approximately 228,000 people on a full-time basis, 125,000 in the U.S. and 103,000 internationally. Of the total employees, 89,000 were in operations, including product support and consulting services, datacenter operations, and manufacturing and distribution; 80,000 were in product research and development; 44,000 were in sales and marketing; and 15,000 were in general and administration. Certain employees are subject to collective bargaining agreements.\nWe design our programs to attract, reward, and retain top talent while fostering continuous employee development and reinforcing our organizational culture and values. Our total compensation offering is both highly differentiated and competitive within the market, and we also monitor pay equity across multiple dimensions. We have invested significantly in employee wellbeing and offer a differentiated benefits package which includes many physical, emotional, and financial wellness programs. We also provide access to continuous learning through a wide range of internal and external content, supporting professional growth across roles and disciplines. Through our employee listening systems, we gather direct feedback from our workforce, enabling us to adapt our programs and address employee needs globally with real-time insights. Additionally, our culture prioritizes the security of both our customers and Microsoft, embedding this responsibility across all teams and functions.\nOPERATIONS\nWe have regional operations service centers in the Americas, Asia Pacific, Europe, and the Middle East that support our business operations, including customer contract and order processing, billing, credit and collections, customer lifecycle AI and cloud operations, and vendor management and logistics.\nIn addition to our operations centers, we also operate datacenters throughout each of these regions. We continue to align our datacenter locations and server capacity to meet the evolving needs of our customers, particularly given the growing demand for AI services. Our datacenters depend on the availability of permitted and buildable land, predictable energy, networking supplies, and servers, including graphics processing units (“GPUs”) and other components.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 12", + "metadata": { + "length": 4501, + "summary": "The passage covers Microsoft's Search and news advertising business, competition in Windows, Devices, Xbox, and Search advertising. It details human capital resources with 228,000 employees as of June 2025, including distribution by function and location. Operations include regional service centers and datacenters focused on AI service demand.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Our", + "Search", + "news", + "advertising", + "business", + "designed", + "deliver", + "relevant", + "search", + "native", + "display", + "global", + "audience", + "Microsoft", + "Copilot", + "digital", + "companion", + "inform", + "entertain", + "inspire", + "Edge", + "browser", + "Bing", + "engine", + "key", + "tools", + "enable", + "user", + "acquisition", + "engagement", + "technology", + "platform", + "enables", + "accelerated", + "delivery", + "solutions", + "In", + "addition", + "party", + "partnerships", + "companies", + "provide", + "monetize", + "offerings", + "Growth", + "depends", + "ability", + "attract", + "users", + "understand", + "intent", + "match", + "content", + "Competition", + "Windows", + "faces", + "competition", + "software", + "products", + "alternative", + "platforms", + "devices", + "We", + "competes", + "effectively", + "giving", + "customers", + "choice", + "flexibility", + "security", + "easy", + "interface", + "compatibility", + "broad", + "range", + "hardware", + "applications", + "including", + "productivity", + "Devices", + "face", + "computer", + "tablet", + "manufacturers", + "offer", + "unique", + "combination", + "high", + "quality", + "industrial", + "design", + "innovative", + "technologies", + "price", + "points", + "Many", + "current", + "potential", + "partners", + "OEMs", + "Xbox", + "cloud", + "gaming", + "services", + "online", + "ecosystems", + "game", + "streaming", + "compete", + "providers", + "entertainment", + "video", + "console", + "positioned", + "uniquely", + "differentiated", + "competitive", + "based", + "significant", + "innovation", + "architecture", + "developer", + "continued", + "strong", + "franchises", + "engines", + "wide", + "array", + "websites", + "social", + "portals", + "end", + "HUMAN", + "CAPITAL", + "RESOURCES", + "As", + "June", + "30", + "2025", + "employed", + "approximately", + "228", + "000", + "people", + "full", + "time", + "basis", + "125", + "103", + "internationally", + "Of", + "total", + "employees", + "89", + "operations", + "product", + "support", + "consulting", + "datacenter", + "manufacturing", + "distribution", + "80", + "research", + "development", + "44", + "sales", + "marketing", + "15", + "general", + "administration", + "Certain", + "subject", + "collective", + "bargaining", + "agreements", + "programs", + "reward", + "retain", + "top", + "talent", + "fostering", + "continuous", + "employee", + "reinforcing", + "organizational", + "culture", + "values", + "compensation", + "offering", + "highly", + "market", + "monitor", + "pay", + "equity", + "multiple", + "dimensions", + "invested", + "significantly", + "wellbeing", + "benefits", + "package", + "includes", + "physical", + "emotional", + "financial", + "wellness", + "access", + "learning", + "internal", + "external", + "supporting", + "professional", + "growth", + "roles", + "disciplines", + "Through", + "listening", + "systems", + "gather", + "direct", + "feedback", + "workforce", + "enabling", + "adapt", + "address", + "globally", + "real", + "insights", + "Additionally", + "prioritizes", + "embedding", + "responsibility", + "teams", + "functions", + "OPERATIONS", + "regional", + "service", + "centers", + "Americas", + "Asia", + "Pacific", + "Europe", + "Middle", + "East", + "customer", + "contract", + "order", + "processing", + "billing", + "credit", + "collections", + "lifecycle", + "AI", + "vendor", + "management", + "logistics", + "operate", + "datacenters", + "regions", + "continue", + "align", + "locations", + "server", + "capacity", + "meet", + "evolving", + "growing", + "demand", + "depend", + "availability", + "permitted", + "buildable", + "land", + "predictable", + "energy", + "networking", + "supplies", + "servers", + "graphics", + "units", + "GPUs", + "components" + ], + "keywords": [ + "Competition", + "Human Capital", + "Operations" + ], + "connect_to": [] + } + }, + { + "chunk_id": "5d60860e-a9e5-5004-bd6a-e7490c2d5ff5", + "type": "text", + "content": "We engage third-party manufacturers to produce our devices and have implemented measures to enhance supply chain efficiency and resilience, including the ability to relocate production geographically.\nThere are few qualified suppliers for certain components of our servers and devices. Extended or unforeseen disruptions at these suppliers could impact our ability to operate our datacenters and manufacture devices on time to meet consumer demand.\nRESEARCH AND DEVELOPMENT\nProduct and Service Development\nOur success is based on our ability to create new and compelling products, services, and experiences for our users, initiate and embrace disruptive technology trends, enter new geographic and product markets, and drive broad adoption of our products and services. We make significant investments in research and development for new and existing products, services, and technologies, including tools and platforms spanning digital work and life experiences, cloud computing, AI, devices, security, and operating systems.\nWe develop most of our products and services internally which allows us to maintain competitive advantages that come from product differentiation and closer technical control over our products and services. It also gives us the freedom to decide which modifications and enhancements are most important and when they should be implemented. We strive to obtain information as early as possible about changing usage patterns and hardware advances that may affect software and hardware design. Before releasing new software platforms, and as we make significant modifications to existing platforms, we provide application vendors with a range of resources and guidelines for development, training, and testing.\nWe plan to continue to make significant investments in a broad range of product research and development activities, and as appropriate, we will coordinate our research and development across operating segments and leverage the results across the company. This includes continuing to support fundamental research, which provides us with a unique perspective on future trends and contributes to our innovation.\nIntellectual Property\nWe protect our intellectual property investments in a variety of ways. We work actively in the U.S. and internationally to ensure the enforcement of copyright, patent, trademark, trade secret, and other protections that apply to our software and hardware products, services, business plans, and branding. While we employ much of our internally-developed intellectual property in our products and services, we also engage in outbound licensing of specific patented technologies that are incorporated into licensees’ products. From time to time, we enter into broader cross-license agreements with other technology companies covering entire groups of patents. We may also purchase or license technology that we incorporate into our products and services. At times, we make select intellectual property broadly available at no or low cost to achieve a strategic objective, such as promoting industry standards, advancing interoperability, supporting societal and/or environmental efforts, or attracting and enabling our external development community. Our engagement with open source software also causes us to license our intellectual property rights broadly in certain situations.\nWhile it may be necessary in the future to seek or renew licenses relating to various aspects of our products and services, we believe, based upon past experience and industry practice, such licenses generally can be obtained on commercially reasonable terms. We believe our continuing research and product development are not materially dependent on any single license or other agreement with a third-party relating to the development of our products.\nDISTRIBUTION, SALES, AND MARKETING", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 13", + "metadata": { + "length": 3832, + "summary": "The passage discusses the company's reliance on third-party manufacturers and potential risks from limited suppliers for server components. It emphasizes significant investments in research and development for new products and services, including cloud computing, AI, and devices. The company develops most products internally to maintain competitive advantages and protects its intellectual property through patents, trademarks, and licensing. It also engages in open source and cross-license agreements. The summary covers supply chain resilience, R&D strategy, and IP protection.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "We", + "engage", + "party", + "manufacturers", + "produce", + "devices", + "implemented", + "measures", + "enhance", + "supply", + "chain", + "efficiency", + "resilience", + "including", + "ability", + "relocate", + "production", + "geographically", + "There", + "qualified", + "suppliers", + "components", + "servers", + "Extended", + "unforeseen", + "disruptions", + "impact", + "operate", + "datacenters", + "manufacture", + "time", + "meet", + "consumer", + "demand", + "RESEARCH", + "AND", + "DEVELOPMENT", + "Product", + "Service", + "Development", + "Our", + "success", + "based", + "create", + "compelling", + "products", + "services", + "experiences", + "users", + "initiate", + "embrace", + "disruptive", + "technology", + "trends", + "enter", + "geographic", + "product", + "markets", + "drive", + "broad", + "adoption", + "make", + "significant", + "investments", + "research", + "development", + "existing", + "technologies", + "tools", + "platforms", + "spanning", + "digital", + "work", + "life", + "cloud", + "computing", + "AI", + "security", + "operating", + "systems", + "develop", + "internally", + "maintain", + "competitive", + "advantages", + "differentiation", + "closer", + "technical", + "control", + "It", + "freedom", + "decide", + "modifications", + "enhancements", + "important", + "strive", + "obtain", + "information", + "early", + "changing", + "usage", + "patterns", + "hardware", + "advances", + "affect", + "software", + "design", + "Before", + "releasing", + "provide", + "application", + "vendors", + "range", + "resources", + "guidelines", + "training", + "testing", + "plan", + "continue", + "activities", + "coordinate", + "segments", + "leverage", + "results", + "company", + "This", + "includes", + "continuing", + "support", + "fundamental", + "unique", + "perspective", + "future", + "contributes", + "innovation", + "Intellectual", + "Property", + "protect", + "intellectual", + "property", + "variety", + "ways", + "actively", + "internationally", + "ensure", + "enforcement", + "copyright", + "patent", + "trademark", + "trade", + "secret", + "protections", + "apply", + "business", + "plans", + "branding", + "While", + "employ", + "developed", + "outbound", + "licensing", + "specific", + "patented", + "incorporated", + "licensees", + "From", + "broader", + "cross", + "license", + "agreements", + "companies", + "covering", + "entire", + "groups", + "patents", + "purchase", + "incorporate", + "At", + "times", + "select", + "broadly", + "low", + "cost", + "achieve", + "strategic", + "objective", + "promoting", + "industry", + "standards", + "advancing", + "interoperability", + "supporting", + "societal", + "environmental", + "efforts", + "attracting", + "enabling", + "external", + "community", + "engagement", + "open", + "source", + "rights", + "situations", + "seek", + "renew", + "licenses", + "relating", + "aspects", + "past", + "experience", + "practice", + "generally", + "obtained", + "commercially", + "reasonable", + "terms", + "materially", + "dependent", + "single", + "agreement", + "DISTRIBUTION", + "SALES", + "MARKETING" + ], + "keywords": [ + "Supply Chain", + "R&D", + "Intellectual Property" + ], + "connect_to": [] + } + }, + { + "chunk_id": "03ae86ea-c9ce-5c8f-b69c-0d85585d7a8b", + "type": "text", + "content": "Our customers include individual consumers, small and medium organizations, large global enterprises, public-sector institutions, service providers, application developers, and OEMs. We market and distribute our products and services through the following channels: direct, distributors and resellers, and OEMs. Our sales organization performs a variety of functions, including working directly with commercial enterprises and public-sector organizations worldwide to identify and meet their technology and digital transformation requirements; supporting system integrators, independent software vendors, and other partners who engage directly with our customers to perform sales, consulting, and fulfillment functions for our products and services; and managing OEM relationships.\nDirect\nMany organizations that license our products and services transact directly with us through Enterprise Agreements and Enterprise Services contracts, with sales support from system integrators, independent software vendors, web agencies, and partners that advise organizations on licensing our products and services (“Enterprise Agreement Software Advisors” or “ESA”). Microsoft offers direct sales programs targeted to reach small, medium, and corporate customers, in addition to those offered through the reseller channel. A large network of partner advisors support many of these sales.\nWe also sell commercial and consumer products and services directly to customers, such as cloud services, search, and gaming, through our digital marketplaces and online stores. Additionally, our Microsoft Experience Centers are designed to facilitate deeper engagement with our partners and customers across industries.\nDistributors and Resellers\nOrganizations also license our products and services indirectly, primarily through licensing solution partners (“LSP”), distributors, value-added resellers (“VAR”), and retailers. Although each type of reselling partner may reach organizations of all sizes, LSPs are primarily engaged with large organizations, distributors resell primarily to VARs, and VARs typically reach small and medium organizations. ESAs are also typically authorized as LSPs and operate as resellers for our other volume licensing programs. Microsoft Cloud Solution Provider is our main partner program for reselling cloud services.\nWe distribute our retail packaged products primarily through independent non-exclusive distributors, authorized replicators, resellers, and retail outlets. Individual consumers obtain these products primarily through retail outlets. We distribute our devices through third-party retailers. We have a network of field sales representatives and field support personnel that solicit orders from distributors and resellers and provide product training and sales support.\nOur Dynamics business solutions are also licensed to enterprises through a global network of channel partners providing vertical solutions and specialized services.\nOEMs\nWe distribute our products and services through OEMs that pre-install our software on new devices and servers they sell. The largest component of the OEM business is the Windows operating system pre-installed on devices. OEMs also sell devices pre-installed with other Microsoft products and services, including applications such as Office and the capability to subscribe to Microsoft 365 Consumer.\nThere are two broad categories of OEMs. The largest category of OEMs are direct OEMs as our relationship with them is managed through a direct agreement between Microsoft and the OEM. We have distribution agreements covering one or more of our products with virtually all the multinational OEMs, including Dell, Hewlett-Packard, Lenovo, and with many regional and local OEMs. The second broad category of OEMs are system builders consisting of lower-volume PC manufacturers, which source Microsoft software for pre-installation and local redistribution primarily through the Microsoft distributor channel rather than through a direct agreement or relationship with Microsoft.\nLICENSING OPTIONS", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 14", + "metadata": { + "length": 4063, + "summary": "Microsoft sells products and services through direct sales, distributors and resellers, and OEMs. Direct sales include Enterprise Agreements and online stores. Resellers include LSPs, VARs, and retailers. OEMs pre-install software on devices, with direct OEMs and system builders as categories.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Our", + "customers", + "include", + "individual", + "consumers", + "small", + "medium", + "organizations", + "large", + "global", + "enterprises", + "public", + "sector", + "institutions", + "service", + "providers", + "application", + "developers", + "OEMs", + "We", + "market", + "distribute", + "products", + "services", + "channels", + "direct", + "distributors", + "resellers", + "sales", + "organization", + "performs", + "variety", + "functions", + "including", + "working", + "directly", + "commercial", + "worldwide", + "identify", + "meet", + "technology", + "digital", + "transformation", + "requirements", + "supporting", + "system", + "integrators", + "independent", + "software", + "vendors", + "partners", + "engage", + "perform", + "consulting", + "fulfillment", + "managing", + "OEM", + "relationships", + "Direct", + "Many", + "license", + "transact", + "Enterprise", + "Agreements", + "Services", + "contracts", + "support", + "web", + "agencies", + "advise", + "licensing", + "Agreement", + "Software", + "Advisors", + "ESA", + "Microsoft", + "offers", + "programs", + "targeted", + "reach", + "corporate", + "addition", + "offered", + "reseller", + "channel", + "network", + "partner", + "advisors", + "sell", + "consumer", + "cloud", + "search", + "gaming", + "marketplaces", + "online", + "stores", + "Additionally", + "Experience", + "Centers", + "designed", + "facilitate", + "deeper", + "engagement", + "industries", + "Distributors", + "Resellers", + "Organizations", + "indirectly", + "primarily", + "solution", + "LSP", + "added", + "VAR", + "retailers", + "Although", + "type", + "reselling", + "sizes", + "LSPs", + "engaged", + "resell", + "VARs", + "typically", + "ESAs", + "authorized", + "operate", + "volume", + "Cloud", + "Solution", + "Provider", + "main", + "program", + "retail", + "packaged", + "exclusive", + "replicators", + "outlets", + "Individual", + "obtain", + "devices", + "party", + "field", + "representatives", + "personnel", + "solicit", + "orders", + "provide", + "product", + "training", + "Dynamics", + "business", + "solutions", + "licensed", + "providing", + "vertical", + "specialized", + "pre", + "install", + "servers", + "The", + "largest", + "component", + "Windows", + "operating", + "installed", + "applications", + "Office", + "capability", + "subscribe", + "365", + "Consumer", + "There", + "broad", + "categories", + "category", + "relationship", + "managed", + "agreement", + "distribution", + "agreements", + "covering", + "virtually", + "multinational", + "Dell", + "Hewlett", + "Packard", + "Lenovo", + "regional", + "local", + "builders", + "consisting", + "lower", + "PC", + "manufacturers", + "source", + "installation", + "redistribution", + "distributor", + "LICENSING", + "OPTIONS" + ], + "keywords": [ + "Direct", + "Resellers", + "OEMs" + ], + "connect_to": [] + } + }, + { + "chunk_id": "52450c36-b7d6-5729-999a-0a37603c3ae1", + "type": "text", + "content": "We offer options for organizations of varying sizes that want to purchase our cloud services and on-premises software. We license these organizations under volume licensing agreements to allow the customer to acquire multiple licenses of products and services instead of having to acquire separate licenses through retail channels. These volume licensing programs have varying programmatic requirements and benefits to best meet the needs of our customers.\nSoftware Assurance (“SA”) conveys rights to new software and upgrades for perpetual licenses released over the contract period. It also provides support, tools, training, and other licensing benefits to help customers deploy and use software efficiently. SA is required to be purchased with certain volume licensing agreements and is an optional purchase with others.\nVolume Licensing Programs\nEnterprise Agreement\nEnterprise Agreements offer large organizations a manageable volume licensing program that gives them the flexibility to buy cloud services and software licenses under one agreement. Enterprise Agreements are designed for medium or large organizations that want to license Microsoft products and services organization-wide over a three-year period. Organizations can elect to purchase perpetual licenses (covered with SA) and/or subscribe to cloud services.\nMicrosoft Customer Agreement\nMicrosoft Customer Agreements are simplified purchase agreements presented, accepted, and stored through a digital experience. Microsoft Customer Agreements are non-expiring agreements that are designed to support all customers over time, whether purchasing through a partner or directly from Microsoft.\nMicrosoft Online Subscription Agreement\nMicrosoft Online Subscription Agreements are designed for small and medium organizations that want to subscribe to, activate, provision, and maintain cloud services seamlessly and directly via the web. These agreements allow customers to acquire monthly or annual subscriptions for cloud-based services.\nMicrosoft Products and Services Agreement\nMicrosoft Products and Services Agreements are designed for medium and large organizations that want to license cloud services and on-premises software as needed, with no organization-wide commitment, under a single, non-expiring agreement. Organizations purchase perpetual licenses or subscribe to licenses. SA is optional for customers that purchase perpetual licenses.\nOpen Value\nOpen Value agreements are a simple, cost-effective way to acquire the latest Microsoft technology. These agreements are designed for small and medium organizations that want to license cloud services and on-premises software over a three-year period. Under Open Value agreements, organizations can elect to purchase perpetual licenses or subscribe to licenses and SA is included.\nSelect Plus\nA Select Plus agreement is designed for government and academic organizations to acquire on-premises licenses at any affiliate or department level, while realizing advantages as one organization. Organizations purchase perpetual licenses and SA is optional.\nPartner Programs\nThe Microsoft Cloud Solution Provider Program offers customers an easy way to license the cloud services they need in combination with the value-added services offered by their systems integrator, managed services provider, or cloud reseller partner. Partners in this program can easily package their own products and services to directly provision, manage, and support their customer subscriptions.\nThe Microsoft Services Provider License Agreement allows hosting service providers and independent software vendors who want to license eligible Microsoft software products to provide hosted applications and software services to their end customers. Partners license software over a three-year period and are billed monthly based on units licensed.\nThe Independent Software Vendor Royalty Program enables partners to integrate Microsoft products into other applications and then license the unified business solution to their end users.\nGOVERNMENT REGULATION", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 15", + "metadata": { + "length": 4057, + "summary": "The text describes Microsoft's volume licensing programs for organizations of varying sizes, covering cloud services and on-premises software. It explains Software Assurance (SA) benefits and outlines specific programs: Enterprise Agreement for large organizations with three-year terms, Microsoft Customer Agreement as a digital non-expiring option, Microsoft Online Subscription Agreement for small/medium businesses, Microsoft Products and Services Agreement for medium/large organizations without organization-wide commitment, Open Value for small/medium organizations with three-year terms including SA, and Select Plus for government/academic entities. Partner programs include Cloud Solution Provider Program, Services Provider License Agreement, and Independent Software Vendor Royalty Program.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "We", + "offer", + "options", + "organizations", + "varying", + "sizes", + "purchase", + "cloud", + "services", + "premises", + "software", + "license", + "volume", + "licensing", + "agreements", + "customer", + "acquire", + "multiple", + "licenses", + "products", + "separate", + "retail", + "channels", + "These", + "programs", + "programmatic", + "requirements", + "benefits", + "meet", + "customers", + "Software", + "Assurance", + "SA", + "conveys", + "rights", + "upgrades", + "perpetual", + "released", + "contract", + "period", + "It", + "support", + "tools", + "training", + "deploy", + "efficiently", + "required", + "purchased", + "optional", + "Volume", + "Licensing", + "Programs", + "Enterprise", + "Agreement", + "Agreements", + "large", + "manageable", + "program", + "flexibility", + "buy", + "agreement", + "designed", + "medium", + "Microsoft", + "organization", + "wide", + "year", + "Organizations", + "elect", + "covered", + "subscribe", + "Customer", + "simplified", + "presented", + "accepted", + "stored", + "digital", + "experience", + "expiring", + "time", + "purchasing", + "partner", + "directly", + "Online", + "Subscription", + "small", + "activate", + "provision", + "maintain", + "seamlessly", + "web", + "monthly", + "annual", + "subscriptions", + "based", + "Products", + "Services", + "needed", + "commitment", + "single", + "Open", + "Value", + "simple", + "cost", + "effective", + "latest", + "technology", + "Under", + "included", + "Select", + "Plus", + "government", + "academic", + "affiliate", + "department", + "level", + "realizing", + "advantages", + "Partner", + "The", + "Cloud", + "Solution", + "Provider", + "Program", + "offers", + "easy", + "combination", + "added", + "offered", + "systems", + "integrator", + "managed", + "provider", + "reseller", + "Partners", + "easily", + "package", + "manage", + "License", + "hosting", + "service", + "providers", + "independent", + "vendors", + "eligible", + "provide", + "hosted", + "applications", + "end", + "billed", + "units", + "licensed", + "Independent", + "Vendor", + "Royalty", + "enables", + "partners", + "integrate", + "unified", + "business", + "solution", + "users", + "GOVERNMENT", + "REGULATION" + ], + "keywords": [ + "Volume Licensing", + "Software Assurance", + "Cloud Services" + ], + "connect_to": [] + } + }, + { + "chunk_id": "4420cfbe-f143-54d1-ae6e-e8af15304d41", + "type": "text", + "content": "We are subject to a wide range of laws, regulations, and legal requirements in the U.S. and globally, including those that may apply to our products and online services offerings, and those that impose requirements related to user privacy, telecommunications, data storage and protection, advertising, and online content. These requirements are continually evolving, and they can be unclear and vary significantly across jurisdictions. We have implemented comprehensive compliance programs across our operations to adapt to these changes and to maintain customer and regulator confidence. We monitor regulatory developments around the world and implement policies, controls, and technical safeguards so that our operations, products, and services meet applicable legal standards. Our business teams, with legal support, manage the compliance programs and prepare external regulatory and commercial reporting, and our internal audit teams conduct reviews of the programs and processes. While we have a unified approach to regulatory compliance, some of the programs and processes are tailored to meet specific regulatory obligations, such as with the creation of independent compliance functions required by the European Union (“EU”) Digital Markets Act and the EU Digital Services Act, which oversee, monitor, and assess the company’s compliance with these acts.\nFor a description of the risks we face related to regulatory matters, refer to Risk Factors in our fiscal year 2025 Form 10K.\nAVAILABLE INFORMATION\nOur Internet address is www.microsoft.com. At our Investor Relations website, www.microsoft.com/investor, we make available free of charge a variety of information for investors. Our goal is to maintain the Investor Relations website as a portal through which investors can easily find or navigate to pertinent information about us, including:\n• Our annual report on Form 10-K, quarterly reports on Form 10-Q, current reports on Form 8-K, and any amendments to those reports, as soon as reasonably practicable after we electronically file that material with or furnish it to the Securities and Exchange Commission (“SEC”) at www.sec.gov.\n• Information on our business strategies, financial results, and metrics for investors.\n• Announcements of investor conferences, speeches, and events at which our executives talk about our product, service, and competitive strategies. Archives of these events are also available.\n• Press releases on quarterly earnings, product and service announcements, legal developments, and international news.\n• Corporate governance information including our articles of incorporation, bylaws, governance guidelines, committee charters, codes of conduct and ethics, global corporate social responsibility initiatives, and other governance-related policies.\n• Other news and announcements that we may post from time to time that investors might find useful or interesting.\n• Opportunities to sign up for email alerts to have information pushed in real time.\nWe publish a variety of reports and resources related to our Corporate Social Responsibility programs and progress on our Reports Hub website, www.microsoft.com/corporate-responsibility/reports-hub, including reports on responsible AI, sustainability, responsible sourcing, accessibility, digital trust, and public policy engagement.\nThe information found on these websites is not part of, or incorporated by reference into, this or any other report we file with, or furnish to, the SEC. In addition to these channels, we use social media to communicate to the public. It is possible that the information we post on social media could be deemed to be material to investors. We encourage investors, the media, and others interested in Microsoft to review the information we post on the social media channels listed on our Investor Relations website.\nMANAGEMENT’S DISCUSSION AND ANALYSIS OF FINANCIAL CONDITION AND RESULTS OF OPERATIONS", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 16", + "metadata": { + "length": 3929, + "summary": "The passage discusses Microsoft's compliance with a wide range of U.S. and global laws, including those related to privacy, telecommunications, data protection, advertising, and online content. The company has implemented comprehensive compliance programs, monitors regulatory developments, and uses policies, controls, and technical safeguards to meet legal standards. It highlights tailored programs for specific obligations like the EU Digital Markets Act and Digital Services Act. The text also describes available investor information on Microsoft's website, including SEC filings, business strategies, press releases, and corporate governance materials, as well as social media channels for public communication.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "We", + "subject", + "wide", + "range", + "laws", + "regulations", + "legal", + "requirements", + "globally", + "including", + "apply", + "products", + "online", + "services", + "offerings", + "impose", + "related", + "user", + "privacy", + "telecommunications", + "data", + "storage", + "protection", + "advertising", + "content", + "These", + "continually", + "evolving", + "unclear", + "vary", + "significantly", + "jurisdictions", + "implemented", + "comprehensive", + "compliance", + "programs", + "operations", + "adapt", + "maintain", + "customer", + "regulator", + "confidence", + "monitor", + "regulatory", + "developments", + "world", + "implement", + "policies", + "controls", + "technical", + "safeguards", + "meet", + "applicable", + "standards", + "Our", + "business", + "teams", + "support", + "manage", + "prepare", + "external", + "commercial", + "reporting", + "internal", + "audit", + "conduct", + "reviews", + "processes", + "While", + "unified", + "approach", + "tailored", + "specific", + "obligations", + "creation", + "independent", + "functions", + "required", + "European", + "Union", + "EU", + "Digital", + "Markets", + "Act", + "Services", + "oversee", + "assess", + "company", + "acts", + "For", + "description", + "risks", + "face", + "matters", + "refer", + "Risk", + "Factors", + "fiscal", + "year", + "2025", + "Form", + "10K", + "AVAILABLE", + "INFORMATION", + "Internet", + "address", + "www", + "microsoft", + "At", + "Investor", + "Relations", + "website", + "investor", + "make", + "free", + "charge", + "variety", + "information", + "investors", + "goal", + "portal", + "easily", + "find", + "navigate", + "pertinent", + "annual", + "report", + "10", + "quarterly", + "reports", + "current", + "amendments", + "practicable", + "electronically", + "file", + "material", + "furnish", + "Securities", + "Exchange", + "Commission", + "SEC", + "sec", + "gov", + "Information", + "strategies", + "financial", + "results", + "metrics", + "Announcements", + "conferences", + "speeches", + "events", + "executives", + "talk", + "product", + "service", + "competitive", + "Archives", + "Press", + "releases", + "earnings", + "announcements", + "international", + "news", + "Corporate", + "governance", + "articles", + "incorporation", + "bylaws", + "guidelines", + "committee", + "charters", + "codes", + "ethics", + "global", + "corporate", + "social", + "responsibility", + "initiatives", + "Other", + "post", + "time", + "interesting", + "Opportunities", + "sign", + "email", + "alerts", + "pushed", + "real", + "publish", + "resources", + "Social", + "Responsibility", + "progress", + "Reports", + "Hub", + "hub", + "responsible", + "AI", + "sustainability", + "sourcing", + "accessibility", + "digital", + "trust", + "public", + "policy", + "engagement", + "The", + "found", + "websites", + "part", + "incorporated", + "reference", + "In", + "addition", + "channels", + "media", + "communicate", + "It", + "deemed", + "encourage", + "interested", + "Microsoft", + "review", + "listed", + "MANAGEMENT", + "DISCUSSION", + "AND", + "ANALYSIS", + "OF", + "FINANCIAL", + "CONDITION", + "RESULTS", + "OPERATIONS" + ], + "keywords": [ + "regulatory compliance", + "legal requirements", + "corporate governance" + ], + "connect_to": [] + } + }, + { + "chunk_id": "a41e7c83-8a77-5764-bef8-892dc0f445d0", + "type": "text", + "content": "The following Management’s Discussion and Analysis of Financial Condition and Results of Operations (“MD&A”) is intended to help the reader understand the results of operations and financial condition of Microsoft Corporation. MD&A is provided as a supplement to, and should be read in conjunction with, our consolidated financial statements and the accompanying Notes to Financial Statements. This section generally discusses the results of our operations for the year ended June 30, 2025 compared to the year ended June 30, 2024. For a discussion of the year ended June 30, 2024 compared to the year ended June 30, 2023, please refer to “Management’s Discussion and Analysis of Financial Condition and Results of Operations” in our Annual Report on Form 10-K for the year ended June 30, 2024 and our Form 8-K filed on December 3, 2024.\nOVERVIEW\nMicrosoft is a technology company committed to making digital technology and artificial intelligence (“AI”) available broadly and doing so responsibly, with a mission to empower every person and every organization on the planet to achieve more. We create platforms and tools, powered by AI, that deliver innovative solutions that meet the evolving needs of our customers.\nWe generate revenue by offering a wide range of cloud-based solutions, content, and other services to people and businesses; licensing and supporting an array of software products; delivering relevant online advertising to a global audience; and designing and selling devices. Our most significant expenses are related to compensating employees; supporting and investing in our cloud-based services, including datacenter operations; designing, manufacturing, marketing, and selling our other products and services; and income taxes.\nHighlights from fiscal year 2025 compared with fiscal year 2024 included:\n• Microsoft Cloud revenue increased 23% to $168.9 billion.\n• Microsoft 365 Commercial products and cloud services revenue increased 14% driven by Microsoft 365 Commercial cloud revenue growth of 15%.\n• Microsoft 365 Consumer products and cloud services revenue increased 11% driven by Microsoft 365 Consumer cloud revenue growth of 11%.\n• LinkedIn revenue increased 9%.\n• Dynamics products and cloud services revenue increased 15% driven by Dynamics 365 revenue growth of 19%.\n• Server products and cloud services revenue increased 23% driven by Azure and other cloud services revenue growth of 34%.\n• Windows OEM and Devices revenue increased 3%.\n• Xbox content and services revenue increased 16%.\n• Search and news advertising revenue excluding traffic acquisition costs increased 20%.\nIndustry Trends and Opportunities\nOur industry is dynamic and highly competitive, with frequent changes in both technologies and business models. Each industry shift is an opportunity to conceive new products, new technologies, or new ideas that can further transform the industry and our business. At Microsoft, we push the boundaries of what is possible through a broad range of research and development activities that seek to identify and address the changing demands of customers and users, industry trends, and competitive forces.\nMicrosoft and OpenAI maintain a long-term strategic partnership originally established in 2019. Microsoft is a major investor in OpenAI, and the companies have reciprocal revenue-sharing arrangements. We hold rights to OpenAI’s intellectual property, including models and infrastructure, for integration into our products. The OpenAI API is exclusive to Azure, runs on Azure, and is available through the Azure OpenAI Service. We also have a right of first refusal on OpenAI’s new capacity needs.\nEconomic Conditions, Challenges, and Risks\nThe markets for software, devices, and cloud-based services are dynamic and highly competitive. Our competitors are developing new software and devices, while also deploying competing cloud-based services for consumers and businesses. The devices and form factors customers prefer evolve rapidly, influencing how users access services in the cloud and, in some cases, the user’s choice of which suite of cloud-based services to use. Aggregate demand for our software, services, and devices is also correlated to global macroeconomic and geopolitical factors, which remain dynamic. We must continue to evolve and adapt over an extended time in pace with this changing environment.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 17", + "metadata": { + "length": 4370, + "summary": "Microsoft's MD&A for fiscal year 2025 highlights strong growth driven by cloud and AI. Microsoft Cloud revenue rose 23% to $168.9 billion, with Azure and other cloud services growing 34%. The company discusses its strategic partnership with OpenAI, industry trends, and competitive risks. Key revenue drivers include Microsoft 365, Dynamics 365, LinkedIn, Xbox, and search advertising.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "The", + "Management", + "Discussion", + "Analysis", + "Financial", + "Condition", + "Results", + "Operations", + "MD", + "intended", + "reader", + "understand", + "results", + "operations", + "financial", + "condition", + "Microsoft", + "Corporation", + "provided", + "supplement", + "read", + "conjunction", + "consolidated", + "statements", + "accompanying", + "Notes", + "Statements", + "This", + "section", + "generally", + "discusses", + "year", + "ended", + "June", + "30", + "2025", + "compared", + "2024", + "For", + "discussion", + "2023", + "refer", + "Annual", + "Report", + "Form", + "10", + "filed", + "December", + "OVERVIEW", + "technology", + "company", + "committed", + "making", + "digital", + "artificial", + "intelligence", + "AI", + "broadly", + "responsibly", + "mission", + "empower", + "person", + "organization", + "planet", + "achieve", + "We", + "create", + "platforms", + "tools", + "powered", + "deliver", + "innovative", + "solutions", + "meet", + "evolving", + "customers", + "generate", + "revenue", + "offering", + "wide", + "range", + "cloud", + "based", + "content", + "services", + "people", + "businesses", + "licensing", + "supporting", + "array", + "software", + "products", + "delivering", + "relevant", + "online", + "advertising", + "global", + "audience", + "designing", + "selling", + "devices", + "Our", + "significant", + "expenses", + "related", + "compensating", + "employees", + "investing", + "including", + "datacenter", + "manufacturing", + "marketing", + "income", + "taxes", + "Highlights", + "fiscal", + "included", + "Cloud", + "increased", + "23%", + "168.9", + "billion", + "365", + "Commercial", + "14%", + "driven", + "growth", + "15%", + "Consumer", + "11%", + "LinkedIn", + "9%", + "Dynamics", + "19%", + "Server", + "Azure", + "34%", + "Windows", + "OEM", + "Devices", + "3%", + "Xbox", + "16%", + "Search", + "news", + "excluding", + "traffic", + "acquisition", + "costs", + "20%", + "Industry", + "Trends", + "Opportunities", + "industry", + "dynamic", + "highly", + "competitive", + "frequent", + "technologies", + "business", + "models", + "Each", + "shift", + "opportunity", + "conceive", + "ideas", + "transform", + "At", + "push", + "boundaries", + "broad", + "research", + "development", + "activities", + "seek", + "identify", + "address", + "changing", + "demands", + "users", + "trends", + "forces", + "OpenAI", + "maintain", + "long", + "term", + "strategic", + "partnership", + "originally", + "established", + "2019", + "major", + "investor", + "companies", + "reciprocal", + "sharing", + "arrangements", + "hold", + "rights", + "intellectual", + "property", + "infrastructure", + "integration", + "API", + "exclusive", + "runs", + "Service", + "refusal", + "capacity", + "Economic", + "Conditions", + "Challenges", + "Risks", + "markets", + "competitors", + "developing", + "deploying", + "competing", + "consumers", + "form", + "factors", + "prefer", + "evolve", + "rapidly", + "influencing", + "access", + "cases", + "user", + "choice", + "suite", + "Aggregate", + "demand", + "correlated", + "macroeconomic", + "geopolitical", + "remain", + "continue", + "adapt", + "extended", + "time", + "pace", + "environment" + ], + "keywords": [ + "Microsoft Cloud", + "Azure", + "AI" + ], + "connect_to": [] + } + }, + { + "chunk_id": "0ffa9cf3-0de7-57f5-9489-d1707086edb7", + "type": "text", + "content": "The investments we are making in cloud and AI infrastructure and devices will continue to increase our operating costs and may decrease our operating margins. We continue to identify and evaluate opportunities to expand our datacenter locations and increase our server capacity to meet the evolving needs of our customers, particularly given the growing demand for AI services. Our datacenters depend on the availability of permitted and buildable land, predictable energy, networking supplies, and servers, including graphics processing units (“GPUs”) and other components. Our devices are primarily manufactured by third-party contract manufacturers. For the majority of our products, we have the ability to use other manufacturers if a current vendor becomes unavailable or unable to meet our requirements. However, some of our products contain certain components for which there are very few qualified suppliers. Extended disruptions at these suppliers could impact our ability to manufacture devices on time to meet consumer demand.\nOur success is highly dependent on our ability to attract and retain qualified employees. We hire a mix of university and industry talent worldwide. We compete for talented individuals globally by offering an exceptional working environment, broad customer reach, scale in resources, the ability to grow one’s career across many different products and businesses, and competitive compensation and benefits.\nOur international operations provide a significant portion of our total revenue and expenses. Many of these revenue and expenses are denominated in currencies other than the U.S. dollar. As a result, changes in foreign exchange rates may significantly affect revenue and expenses. Fluctuations in the U.S. dollar relative to certain foreign currencies did not have a material impact on reported revenue and expenses from our international operations in fiscal year 2025.\nFurther, global, regional, and local economic developments and changes in global trade policies such as restrictions on international trade, including tariffs and other controls on imports or exports, could result in increased supply chain challenges, cost volatility, and consumer and economic uncertainty which may adversely affect our results of operations.\nRefer to Risk Factors in our fiscal year 2025 Form 10-K for a discussion of these factors and other risks.\nSeasonality\nOur revenue fluctuates quarterly and is generally higher in the fourth quarter of our fiscal year. Fourth quarter revenue is driven by a higher volume of multi-year contracts executed during the period.\nReportable Segments\nWe report our financial performance based on the following segments: Productivity and Business Processes, Intelligent Cloud, and More Personal Computing. The segment amounts included in MD&A are presented on a basis consistent with our internal management reporting.\nIn August 2024, we announced changes to the composition of our segments. These changes align our segments with how we currently manage our business, most notably bringing the commercial components of Microsoft 365 together in the Productivity and Business Processes segment. Beginning in fiscal year 2025, the information that our chief operating decision maker is regularly provided and reviews for purposes of allocating resources and assessing performance reflects these segment changes. Prior period segment information has been recast to conform to the way we internally manage and monitor our business during fiscal year 2025.\nAdditional information on our reportable segments is contained in Note 18 – Segment Information and Geographic Data of the Notes to Financial Statements.\nMetrics\nWe use metrics in assessing the performance of our business and to make informed decisions regarding the allocation of resources. We disclose metrics to enable investors to evaluate progress against our ambitions, provide transparency into performance trends, and reflect the continued evolution of our products and services. Our commercial and other business metrics are fundamentally connected based on how customers use our products and services. The metrics are disclosed in the MD&A or the Notes to Financial Statements. Financial metrics are calculated based on financial results prepared in accordance with accounting principles generally accepted in the United States of America (“GAAP”), and growth comparisons relate to the corresponding period of last fiscal year.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 18", + "metadata": { + "length": 4456, + "summary": "The passage discusses Microsoft's investments in cloud and AI infrastructure, which may increase operating costs and decrease margins. It highlights dependencies on datacenter resources, third-party manufacturers, and qualified employees. International operations face foreign exchange risks, and global trade policies could impact supply chains. Revenue is seasonal, peaking in Q4. Segment reporting was restructured in August 2024, aligning with management practices. Key metrics are disclosed for performance evaluation.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "The", + "investments", + "making", + "cloud", + "AI", + "infrastructure", + "devices", + "continue", + "increase", + "operating", + "costs", + "decrease", + "margins", + "We", + "identify", + "evaluate", + "opportunities", + "expand", + "datacenter", + "locations", + "server", + "capacity", + "meet", + "evolving", + "customers", + "growing", + "demand", + "services", + "Our", + "datacenters", + "depend", + "availability", + "permitted", + "buildable", + "land", + "predictable", + "energy", + "networking", + "supplies", + "servers", + "including", + "graphics", + "processing", + "units", + "GPUs", + "components", + "primarily", + "manufactured", + "party", + "contract", + "manufacturers", + "For", + "majority", + "products", + "ability", + "current", + "vendor", + "unavailable", + "unable", + "requirements", + "However", + "qualified", + "suppliers", + "Extended", + "disruptions", + "impact", + "manufacture", + "time", + "consumer", + "success", + "highly", + "dependent", + "attract", + "retain", + "employees", + "hire", + "mix", + "university", + "industry", + "talent", + "worldwide", + "compete", + "talented", + "individuals", + "globally", + "offering", + "exceptional", + "working", + "environment", + "broad", + "customer", + "reach", + "scale", + "resources", + "grow", + "career", + "businesses", + "competitive", + "compensation", + "benefits", + "international", + "operations", + "provide", + "significant", + "portion", + "total", + "revenue", + "expenses", + "Many", + "denominated", + "currencies", + "dollar", + "As", + "result", + "foreign", + "exchange", + "rates", + "significantly", + "affect", + "Fluctuations", + "relative", + "material", + "reported", + "fiscal", + "year", + "2025", + "Further", + "global", + "regional", + "local", + "economic", + "developments", + "trade", + "policies", + "restrictions", + "tariffs", + "controls", + "imports", + "exports", + "increased", + "supply", + "chain", + "challenges", + "cost", + "volatility", + "uncertainty", + "adversely", + "results", + "Refer", + "Risk", + "Factors", + "Form", + "10", + "discussion", + "factors", + "risks", + "Seasonality", + "fluctuates", + "quarterly", + "generally", + "higher", + "fourth", + "quarter", + "Fourth", + "driven", + "volume", + "multi", + "contracts", + "executed", + "period", + "Reportable", + "Segments", + "report", + "financial", + "performance", + "based", + "segments", + "Productivity", + "Business", + "Processes", + "Intelligent", + "Cloud", + "More", + "Personal", + "Computing", + "segment", + "amounts", + "included", + "MD", + "presented", + "basis", + "consistent", + "internal", + "management", + "reporting", + "In", + "August", + "2024", + "announced", + "composition", + "These", + "align", + "manage", + "business", + "notably", + "bringing", + "commercial", + "Microsoft", + "365", + "Beginning", + "information", + "chief", + "decision", + "maker", + "regularly", + "provided", + "reviews", + "purposes", + "allocating", + "assessing", + "reflects", + "Prior", + "recast", + "conform", + "internally", + "monitor", + "Additional", + "reportable", + "contained", + "Note", + "18", + "Segment", + "Information", + "Geographic", + "Data", + "Notes", + "Financial", + "Statements", + "Metrics", + "metrics", + "make", + "informed", + "decisions", + "allocation", + "disclose", + "enable", + "investors", + "progress", + "ambitions", + "transparency", + "trends", + "reflect", + "continued", + "evolution", + "fundamentally", + "connected", + "disclosed", + "calculated", + "prepared", + "accordance", + "accounting", + "principles", + "accepted", + "United", + "States", + "America", + "GAAP", + "growth", + "comparisons", + "relate" + ], + "keywords": [ + "Cloud AI", + "Revenue Risks", + "Segment Changes" + ], + "connect_to": [] + } + }, + { + "chunk_id": "bebaff53-91ff-5baf-a265-141db4dd72ca", + "type": "text", + "content": "In the first quarter of fiscal year 2025, we made updates to our metrics in connection with the segment changes described above. These changes align our metrics with how we manage and monitor certain businesses. The key change was bringing the commercial components of Microsoft 365 together and creating a new Microsoft 365 Commercial cloud revenue growth metric. Other changes include combining Windows OEM and Devices into a single revenue growth metric that brings revenue from PC market-driven businesses together, as well as elevating our cloud revenue growth metrics to align to our strategic focus on cloud growth.\nCommercial\nOur commercial business primarily consists of Server products and cloud services, Microsoft 365 Commercial products and cloud services, the commercial portion of LinkedIn, Dynamics products and cloud services, and Enterprise and partner services. Our commercial metrics allow management and investors to assess the overall health of our commercial business and include leading indicators of future performance.\n\nTable defines three Microsoft Cloud metrics: Commercial remaining performance obligation, Cloud revenue and growth, and Cloud gross margin percentage.\n[tables/table-4 Microsoft Cloud KPIs.html]\n\nProductivity and Business Processes and Intelligent Cloud\nMetrics related to our Productivity and Business Processes and Intelligent Cloud segments assess the health of our core businesses within these segments. The metrics primarily reflect growth across our cloud services.\n\nThe table defines key performance indicators for Microsoft's commercial and consumer cloud services, including revenue growth, seat growth, and subscriber counts for Microsoft 365, LinkedIn, Dynamics 365, and Azure.\n[tables/table-5 Microsoft Cloud KPIs.html]\n\nMore Personal Computing\nMetrics related to our More Personal Computing segment assess the performance of our key consumer businesses.\n\nRevenue growth from Windows OEM and Devices, Xbox content and services, and Search and news advertising excluding traffic acquisition costs.\n[tables/table-6 Microsoft Revenue.html]\n\nSUMMARY RESULTS OF OPERATIONS\n\nThe table shows financial results for 2025 vs 2024: Revenue $281,724M (up 15%), Gross margin $193,893M (up 13%), Operating income $128,528M (up 17%), Net income $101,832M (up 16%), Diluted EPS $13.64 (up 16%).\n[tables/table-7 Revenue Growth.html]\n\nFiscal Year 2025 Compared with Fiscal Year 2024\nRevenue increased $36.6 billion or 15% with growth across each of our segments. Intelligent Cloud revenue increased driven by Azure. Productivity and Business Processes revenue increased driven by Microsoft 365 Commercial cloud. More Personal Computing revenue increased driven by Gaming and Search and news advertising.\nCost of revenue increased $13.7 billion or 19% driven by growth in Microsoft Cloud.\nGross margin increased $22.9 billion or 13% with growth across each of our segments.\n• Gross margin percentage decreased slightly driven by Intelligent Cloud, offset in part by More Personal Computing.\n• Microsoft Cloud gross margin percentage decreased to 69% driven by the impact of scaling our AI infrastructure, offset in part by efficiency gains in Azure.\nOperating expenses increased $3.8 billion or 6% driven by investments in cloud and AI engineering and Gaming, including the impact of the Activision Blizzard acquisition.\nOperating income increased $19.1 billion or 17% with growth across each of our segments.\nSEGMENT RESULTS OF OPERATIONS\n\nMicrosoft's 2025 vs 2024 segment data: Productivity & Business Processes revenue $120,810M (+13%), Intelligent Cloud $106,265M (+21%), More Personal Computing $54,649M (+7%). Total revenue $281,724M (+15%), operating income $128,528M (+17%).\n[tables/table-8 Microsoft Segment Results.html]\n\nReportable Segments\nFiscal Year 2025 Compared with Fiscal Year 2024\nProductivity and Business Processes\nRevenue increased $14.0 billion or 13%.\n• Microsoft 365 Commercial products and cloud services revenue increased $10.8 billion or 14%. Microsoft 365 Commercial cloud revenue grew 15% with Microsoft 365 Commercial seat growth of 6% driven by small and medium businesses and frontline worker offerings, as well as growth in revenue per user. Microsoft 365 Commercial products revenue grew 7% driven by the Windows Commercial on-premises components of Microsoft 365 suite sales and an increase in Office transactional purchasing with the launch of Office 2024.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 19", + "metadata": { + "length": 4431, + "summary": "Microsoft's fiscal year 2025 results show total revenue of $281.7 billion (up 15%), gross margin of $193.9 billion, operating income of $128.5 billion, and net income of $101.8 billion. Segment performance includes Productivity & Business Processes revenue up 13% to $120.8 billion, Intelligent Cloud up 21% to $106.3 billion, and More Personal Computing up 7% to $54.6 billion. Key metric changes include a new Microsoft 365 Commercial cloud revenue growth metric and combining Windows OEM with Devices. Microsoft Cloud gross margin decreased to 69% due to AI infrastructure scaling. Revenue growth was driven by Azure, Microsoft 365 Commercial cloud, Gaming, and Search advertising.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "In", + "quarter", + "fiscal", + "year", + "2025", + "made", + "updates", + "metrics", + "connection", + "segment", + "These", + "align", + "manage", + "monitor", + "businesses", + "The", + "key", + "change", + "bringing", + "commercial", + "components", + "Microsoft", + "365", + "creating", + "Commercial", + "cloud", + "revenue", + "growth", + "metric", + "Other", + "include", + "combining", + "Windows", + "OEM", + "Devices", + "single", + "brings", + "PC", + "market", + "driven", + "elevating", + "strategic", + "focus", + "Our", + "business", + "primarily", + "consists", + "Server", + "products", + "services", + "portion", + "LinkedIn", + "Dynamics", + "Enterprise", + "partner", + "management", + "investors", + "assess", + "health", + "leading", + "indicators", + "future", + "performance", + "Table", + "defines", + "Cloud", + "remaining", + "obligation", + "gross", + "margin", + "percentage", + "Productivity", + "Business", + "Processes", + "Intelligent", + "Metrics", + "related", + "segments", + "core", + "reflect", + "table", + "consumer", + "including", + "seat", + "subscriber", + "counts", + "Azure", + "More", + "Personal", + "Computing", + "Revenue", + "Xbox", + "content", + "Search", + "news", + "advertising", + "excluding", + "traffic", + "acquisition", + "costs", + "SUMMARY", + "RESULTS", + "OF", + "OPERATIONS", + "shows", + "financial", + "results", + "2024", + "281", + "724M", + "15%", + "Gross", + "193", + "893M", + "13%", + "Operating", + "income", + "128", + "528M", + "17%", + "Net", + "101", + "832M", + "16%", + "Diluted", + "EPS", + "13.64", + "Fiscal", + "Year", + "Compared", + "increased", + "36.6", + "billion", + "Gaming", + "Cost", + "13.7", + "19%", + "22.9", + "decreased", + "slightly", + "offset", + "part", + "69%", + "impact", + "scaling", + "AI", + "infrastructure", + "efficiency", + "gains", + "expenses", + "3.8", + "6%", + "investments", + "engineering", + "Activision", + "Blizzard", + "19.1", + "SEGMENT", + "data", + "120", + "810M", + "106", + "265M", + "21%", + "54", + "649M", + "7%", + "Total", + "operating", + "Reportable", + "Segments", + "14.0", + "10.8", + "14%", + "grew", + "small", + "medium", + "frontline", + "worker", + "offerings", + "user", + "premises", + "suite", + "sales", + "increase", + "Office", + "transactional", + "purchasing", + "launch" + ], + "keywords": [ + "Microsoft", + "Cloud Revenue", + "Financial Results" + ], + "connect_to": [ + { + "target": "421aaa5d-a19e-55f8-b19f-9f0d269aa779", + "relation": "embeds", + "ref": "[tables/table-4 Microsoft Cloud KPIs.html]", + "position": { + "start": 1197, + "end": 1239 + } + }, + { + "target": "ca7bdf17-9611-509f-9921-ed11c2b257a0", + "relation": "embeds", + "ref": "[tables/table-5 Microsoft Cloud KPIs.html]", + "position": { + "start": 1734, + "end": 1776 + } + }, + { + "target": "c70fc2de-5ee0-5069-8ae4-4f9152c8737b", + "relation": "embeds", + "ref": "[tables/table-6 Microsoft Revenue.html]", + "position": { + "start": 2054, + "end": 2093 + } + }, + { + "target": "21d686c9-1f19-5e9b-99f4-7b4fdbc5cd0b", + "relation": "embeds", + "ref": "[tables/table-7 Revenue Growth.html]", + "position": { + "start": 2336, + "end": 2372 + } + }, + { + "target": "e4a917a1-1b21-5bf4-8860-fdc09eb4b30e", + "relation": "embeds", + "ref": "[tables/table-8 Microsoft Segment Results.html]", + "position": { + "start": 3721, + "end": 3768 + } + } + ] + } + }, + { + "chunk_id": "a7e07b2e-0d3c-51b6-af40-5ecca315ae0b", + "type": "text", + "content": "• Microsoft 365 Consumer products and cloud services revenue increased $756 million or 11%. Microsoft 365 Consumer cloud revenue grew 11% driven by Microsoft 365 Consumer subscriber growth of 8% to 89.0 million, as well as growth in revenue per user from the price increase announced in January 2025.\n• LinkedIn revenue increased $1.4 billion or 9% with growth across all lines of business.\n• Dynamics products and cloud services revenue increased $996 million or 15% driven by growth in Dynamics 365, offset in part by a decline in Dynamics on-premises products. Dynamics 365 revenue grew 19% with growth across all workloads.\nOperating income increased $10.1 billion or 17%.\n• Cost of revenue increased $2.8 billion or 14% driven by growth in Microsoft 365 Commercial cloud.\n• Gross margin increased $11.2 billion or 13% driven by growth in Microsoft 365 Commercial cloud. Gross margin percentage decreased slightly primarily driven by the impact of scaling our AI infrastructure, offset in part by efficiency gains in Microsoft 365 Commercial cloud.\n• Operating expenses increased $1.1 billion or 4% driven by investments in cloud and AI engineering and commercial sales.\nIntelligent Cloud\nRevenue increased $18.8 billion or 21%.\n• Server products and cloud services revenue increased $18.6 billion or 23% driven by Azure and other cloud services. Azure and other cloud services revenue grew 34% driven by demand for our portfolio of services. Server products revenue decreased 3% driven by a decrease in transactional purchasing with continued customer shift to cloud offerings.\n• Enterprise and partner services revenue increased $166 million or 2% driven by growth in Enterprise Support Services, offset in part by a decline in Industry Solutions.\nOperating income increased $6.8 billion or 18%.\n• Cost of revenue increased $10.6 billion or 36% driven by growth in Azure.\n• Gross margin increased $8.2 billion or 14% driven by growth in Azure. Gross margin percentage decreased driven by the impact of scaling our AI infrastructure, offset in part by efficiency gains in Azure.\n• Operating expenses increased $1.5 billion or 7% driven by investments in cloud and AI engineering.\nMore Personal Computing\nRevenue increased $3.8 billion or 7%.\n• Windows and Devices revenue increased $288 million or 2%. Windows OEM and Devices revenue increased 3% driven by growth in Windows OEM, offset in part by a decline in Devices.\n• Gaming revenue increased $2.0 billion or 9% driven by growth in Xbox content and services, offset in part by a decline in Xbox hardware. Xbox content and services revenue increased 16% driven by the impact of the Activision Blizzard acquisition and Xbox Game Pass. Xbox hardware revenue decreased 25% driven by lower volume of consoles sold.\n• Search and news advertising revenue increased $1.6 billion or 13%. Search and news advertising revenue excluding traffic acquisition costs increased 20% driven by higher search volume and higher revenue per search.\nOperating income increased $2.2 billion or 18%.\n• Cost of revenue increased $346 million or 1% driven by growth in Search and news advertising.\n• Gross margin increased $3.5 billion or 13% with growth across all businesses. Gross margin percentage increased with improvement across all businesses.\n• Operating expenses increased $1.3 billion or 9% driven by Gaming, including the impact of the Activision Blizzard acquisition.\nOPERATING EXPENSES\nResearch and Development\n\nResearch and development spending increased from $29,510 million in 2024 to $32,488 million in 2025, a 10% rise, while remaining at 12% of revenue.\n[tables/table-9 R&D Spending 2025.html]\n\nResearch and development expenses include payroll, employee benefits, stock-based compensation expense, and other headcount-related expenses associated with product development. Research and development expenses also include technology development costs, including AI training and other infrastructure costs, third-party development and programming costs, and the amortization of purchased software code and services content.\nFiscal Year 2025 Compared with Fiscal Year 2024\nResearch and development expenses increased $3.0 billion or 10% driven by investments in cloud and AI engineering and Gaming, including the impact of the Activision Blizzard acquisition.\nSales and Marketing", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 20", + "metadata": { + "length": 4327, + "summary": "Microsoft's fiscal year 2025 results show strong growth across segments. Microsoft 365 Consumer revenue rose 11% with subscriber growth to 89 million. LinkedIn revenue increased 9%. Dynamics 365 revenue grew 19%. Intelligent Cloud revenue rose 21%, driven by Azure growth of 34%. More Personal Computing revenue increased 7%, with Gaming up 9% due to Xbox content and services growth of 16%, offset by a 25% decline in Xbox hardware. Operating income increased 17% overall. Research and development expenses rose 10% due to investments in cloud, AI, and Gaming.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Microsoft", + "365", + "Consumer", + "products", + "cloud", + "services", + "revenue", + "increased", + "756", + "million", + "11%", + "grew", + "driven", + "subscriber", + "growth", + "8%", + "89.0", + "user", + "price", + "increase", + "announced", + "January", + "2025", + "LinkedIn", + "1.4", + "billion", + "9%", + "lines", + "business", + "Dynamics", + "996", + "15%", + "offset", + "part", + "decline", + "premises", + "19%", + "workloads", + "Operating", + "income", + "10.1", + "17%", + "Cost", + "2.8", + "14%", + "Commercial", + "Gross", + "margin", + "11.2", + "13%", + "percentage", + "decreased", + "slightly", + "primarily", + "impact", + "scaling", + "AI", + "infrastructure", + "efficiency", + "gains", + "expenses", + "1.1", + "4%", + "investments", + "engineering", + "commercial", + "sales", + "Intelligent", + "Cloud", + "Revenue", + "18.8", + "21%", + "Server", + "18.6", + "23%", + "Azure", + "34%", + "demand", + "portfolio", + "3%", + "decrease", + "transactional", + "purchasing", + "continued", + "customer", + "shift", + "offerings", + "Enterprise", + "partner", + "166", + "2%", + "Support", + "Services", + "Industry", + "Solutions", + "6.8", + "18%", + "10.6", + "36%", + "8.2", + "1.5", + "7%", + "More", + "Personal", + "Computing", + "3.8", + "Windows", + "Devices", + "288", + "OEM", + "Gaming", + "2.0", + "Xbox", + "content", + "hardware", + "16%", + "Activision", + "Blizzard", + "acquisition", + "Game", + "Pass", + "25%", + "lower", + "volume", + "consoles", + "sold", + "Search", + "news", + "advertising", + "1.6", + "excluding", + "traffic", + "costs", + "20%", + "higher", + "search", + "2.2", + "346", + "1%", + "3.5", + "businesses", + "improvement", + "1.3", + "including", + "OPERATING", + "EXPENSES", + "Research", + "Development", + "development", + "spending", + "29", + "510", + "2024", + "32", + "488", + "10%", + "rise", + "remaining", + "12%", + "include", + "payroll", + "employee", + "benefits", + "stock", + "based", + "compensation", + "expense", + "headcount", + "related", + "product", + "technology", + "training", + "party", + "programming", + "amortization", + "purchased", + "software", + "code", + "Fiscal", + "Year", + "Compared", + "3.0", + "Sales", + "Marketing" + ], + "keywords": [ + "Revenue", + "Cloud", + "Gaming" + ], + "connect_to": [ + { + "target": "65e313e3-db40-55ee-a0b9-a74657e07419", + "relation": "embeds", + "ref": "[tables/table-9 R&D Spending 2025.html]", + "position": { + "start": 3606, + "end": 3645 + } + } + ] + } + }, + { + "chunk_id": "e1506605-4527-5a30-895a-3dbdabac2f1b", + "type": "text", + "content": "Sales and marketing costs were $25,654 million in 2025 (9% of revenue) and $24,456 million in 2024 (10% of revenue), a 5% increase and a 1 percentage point decrease as a percent of revenue.\n[tables/table-10 Sales & Marketing.html]\n\nSales and marketing expenses include payroll, employee benefits, stock-based compensation expense, and other headcount-related expenses associated with sales and marketing personnel, and the costs of advertising, promotions, trade shows, seminars, and other programs.\nFiscal Year 2025 Compared with Fiscal Year 2024\nSales and marketing expenses increased $1.2 billion or 5% driven by investments in commercial sales and Gaming, including the impact of the Activision Blizzard acquisition.\nGeneral and Administrative\n\nGeneral and administrative expenses decreased 5% from $7,609M in 2024 to $7,223M in 2025, remaining at 3% of revenue with 0ppt change.\n[tables/table-11 G&A Expenses.html]\n\nGeneral and administrative expenses include payroll, employee benefits, stock-based compensation expense, employee severance expense incurred as part of a corporate program, and other headcount-related expenses associated with finance, legal, facilities, certain human resources and other administrative personnel, certain taxes, and legal and other administrative fees.\nFiscal Year 2025 Compared with Fiscal Year 2024\nGeneral and administrative expenses decreased $386 million or 5% driven by Gaming, including the impact of the Activision Blizzard acquisition.\nOTHER INCOME (EXPENSE), NET\nThe components of other income (expense), net were as follows:\n\nThe table shows financial data for years ended June 30, 2025 and 2024, including interest income, expenses, investment losses, derivatives losses, foreign currency gains/losses, and other net losses, resulting in total net losses of $4,901 million and $1,646 million respectively.\n[tables/table-12 Financial Summary.html]\n\nWe use derivative instruments to manage risks related to foreign currencies, interest rates, equity prices, and credit; to enhance investment returns; and to facilitate portfolio diversification. Gains and losses from changes in fair values of derivatives that are not designated as hedging instruments are primarily recognized in other income (expense), net.\nFiscal Year 2025 Compared with Fiscal Year 2024\nInterest and dividends income decreased primarily due to lower portfolio balances. Interest expense decreased primarily due to maturities of commercial paper and higher capitalization of debt interest expense, offset in part by higher finance lease interest expense. Net recognized losses on investments increased primarily due to higher impairments, offset in part by higher gains on equity investments in the current period. Net losses on derivatives increased primarily due to higher losses on equity derivatives in the current period. Other, net primarily reflects net recognized losses on equity method investments, including OpenAI.\nINCOME TAXES\nEffective Tax Rate\nOur effective tax rate for both fiscal years 2025 and 2024 was 18%. Our effective tax rate for the fiscal year ended June 30, 2025 was primarily impacted by changes in the mix of our earnings and tax expenses between the U.S. and foreign countries.\nOur effective tax rate was lower than the U.S. federal statutory rate, primarily due to earnings taxed at lower rates in foreign jurisdictions resulting from producing and distributing our products and services through our foreign regional operations center in Ireland.\nThe mix of income before income taxes between the U.S. and foreign countries impacted our effective tax rate as a result of the geographic distribution of, and customer demand for, our products and services. In fiscal year 2025, our U.S. income before income taxes was $69.2 billion and our foreign income before income taxes was $54.4 billion. In fiscal year 2024, our U.S. income before income taxes was $62.9 billion and our foreign income before income taxes was $44.9 billion.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 21", + "metadata": { + "length": 3978, + "summary": "Sales and marketing costs rose 5% to $25,654 million in 2025 (9% of revenue) from $24,456 million in 2024 (10% of revenue). General and administrative expenses decreased 5% to $7,223 million. Other income (expense), net showed a net loss of $4,901 million in 2025 versus $1,646 million in 2024. The effective tax rate remained at 18% for both years, with U.S. income before taxes at $69.2 billion and foreign income at $54.4 billion in 2025.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Sales", + "marketing", + "costs", + "25", + "654", + "million", + "2025", + "9%", + "revenue", + "24", + "456", + "2024", + "10%", + "5%", + "increase", + "percentage", + "point", + "decrease", + "percent", + "expenses", + "include", + "payroll", + "employee", + "benefits", + "stock", + "based", + "compensation", + "expense", + "headcount", + "related", + "sales", + "personnel", + "advertising", + "promotions", + "trade", + "shows", + "seminars", + "programs", + "Fiscal", + "Year", + "Compared", + "increased", + "1.2", + "billion", + "driven", + "investments", + "commercial", + "Gaming", + "including", + "impact", + "Activision", + "Blizzard", + "acquisition", + "General", + "Administrative", + "administrative", + "decreased", + "609M", + "223M", + "remaining", + "3%", + "0ppt", + "change", + "severance", + "incurred", + "part", + "corporate", + "program", + "finance", + "legal", + "facilities", + "human", + "resources", + "taxes", + "fees", + "386", + "OTHER", + "INCOME", + "EXPENSE", + "NET", + "The", + "components", + "income", + "net", + "table", + "financial", + "data", + "years", + "ended", + "June", + "30", + "interest", + "investment", + "losses", + "derivatives", + "foreign", + "currency", + "gains", + "resulting", + "total", + "901", + "646", + "We", + "derivative", + "instruments", + "manage", + "risks", + "currencies", + "rates", + "equity", + "prices", + "credit", + "enhance", + "returns", + "facilitate", + "portfolio", + "diversification", + "Gains", + "fair", + "values", + "designated", + "hedging", + "primarily", + "recognized", + "Interest", + "dividends", + "due", + "lower", + "balances", + "maturities", + "paper", + "higher", + "capitalization", + "debt", + "offset", + "lease", + "Net", + "impairments", + "current", + "period", + "Other", + "reflects", + "method", + "OpenAI", + "TAXES", + "Effective", + "Tax", + "Rate", + "Our", + "effective", + "tax", + "rate", + "fiscal", + "18%", + "year", + "impacted", + "mix", + "earnings", + "countries", + "federal", + "statutory", + "taxed", + "jurisdictions", + "producing", + "distributing", + "products", + "services", + "regional", + "operations", + "center", + "Ireland", + "result", + "geographic", + "distribution", + "customer", + "demand", + "In", + "69.2", + "54.4", + "62.9", + "44.9" + ], + "keywords": [ + "Sales", + "Marketing", + "Expenses" + ], + "connect_to": [ + { + "target": "301de650-4cd3-562e-a6a4-72b0cfb244d4", + "relation": "embeds", + "ref": "[tables/table-10 Sales & Marketing.html]", + "position": { + "start": 190, + "end": 230 + } + }, + { + "target": "4058dfb8-d7a1-5283-af19-829f40330f3a", + "relation": "embeds", + "ref": "[tables/table-11 G&A Expenses.html]", + "position": { + "start": 884, + "end": 919 + } + }, + { + "target": "c0a4e50c-e185-55ca-9c4a-a71bbaf7ce12", + "relation": "embeds", + "ref": "[tables/table-12 Financial Summary.html]", + "position": { + "start": 1857, + "end": 1897 + } + } + ] + } + }, + { + "chunk_id": "43e1e4f2-1de0-51de-8e5f-b0213249ef81", + "type": "text", + "content": "The Organisation for Economic Co-operation and Development (“OECD”) published its model rules “Tax Challenges Arising From the Digitalisation of the Economy – Global Anti-Base Erosion Model Rules (Pillar Two)” which established a global minimum corporate tax rate of 15% for certain multinational enterprises. Many countries have implemented or are in the process of implementing the Pillar Two legislation, which applies to Microsoft beginning in fiscal year 2025. While we do not currently estimate a material impact to our consolidated financial statements, we continue to monitor the impact as countries implement legislation and the OECD provides additional guidance.\nWe are currently assessing the One Big Beautiful Bill Act (“OBBBA”) which was enacted on July 4, 2025. The OBBBA provides a U.S. global intangible low-taxed income effective tax rate of 14% effective fiscal year 2027 for Microsoft. It also provides bonus depreciation for certain assets placed into service after January 19, 2025 and an election to expense U.S. incurred research or experimental expenditures.\nUncertain Tax Positions\nWe remain under audit by the IRS for tax years 2014 to 2017. With respect to the audit for tax years 2004 to 2013, on September 26, 2023, we received Notices of Proposed Adjustment (“NOPAs”) from the IRS. The primary issues in the NOPAs relate to intercompany transfer pricing. In the NOPAs, the IRS is seeking an additional tax payment of $28.9 billion plus penalties and interest. As of June 30, 2025, we believe our allowances for income tax contingencies are adequate. We disagree with the proposed adjustments and will vigorously contest the NOPAs through the IRS’s administrative appeals office and, if necessary, judicial proceedings. We do not expect a final resolution of these issues in the next 12 months. Based on the information currently available, we do not anticipate a significant increase or decrease to our income tax contingencies for these issues within the next 12 months.\nWe are subject to income tax in many jurisdictions outside the U.S., some of which are currently under audit by local tax authorities. The resolution of these audits is not expected to be material to our consolidated financial statements. Our operations in Ireland remain subject to examination for tax years 2020 and thereafter.\nLIQUIDITY AND CAPITAL RESOURCES\nWe expect existing cash, cash equivalents, short-term investments, cash flows from operations, and access to capital markets to continue to be sufficient to fund our operating activities and cash commitments for investing and financing activities, such as dividends, share repurchases, debt maturities, material capital expenditures, and the transition tax related to the Tax Cuts and Jobs Act (“TCJA”), for at least the next 12 months and thereafter for the foreseeable future.\nCash, Cash Equivalents, and Investments\nCash, cash equivalents, and short-term investments totaled $94.6 billion and $75.5 billion as of June 30, 2025 and 2024, respectively. Equity and other investments were $15.4 billion and $14.6 billion as of June 30, 2025 and 2024, respectively. Our short-term investments are primarily intended to facilitate liquidity and capital preservation. They consist predominantly of highly liquid investment-grade fixed-income securities, diversified among industries and individual issuers. The investments are predominantly U.S. dollar-denominated securities, but also include foreign currency-denominated securities to diversify risk. Our fixed-income investments are exposed to interest rate risk and credit risk. The credit risk and average maturity of our fixed-income portfolio are managed to achieve economic returns that correlate to certain fixed-income indices. The settlement risk related to these investments is insignificant given that the short-term investments held are primarily highly liquid investment-grade fixed-income securities.\nValuation", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 22", + "metadata": { + "length": 3936, + "summary": "The passage discusses Microsoft's tax and liquidity positions. It covers the OECD's Pillar Two global minimum tax (15%) effective FY2025, the U.S. OBBBA Act (14% GILTI rate from FY2027), and an ongoing IRS audit for tax years 2004-2013 involving a $28.9 billion proposed adjustment for intercompany transfer pricing, which Microsoft contests. The company expects no material impact from Pillar Two and does not anticipate a resolution of the IRS audit within 12 months. On liquidity, cash, cash equivalents, and short-term investments totaled $94.6 billion as of June 30, 2025, with sufficient resources to fund operations and commitments for at least the next 12 months.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "The", + "Organisation", + "Economic", + "Co", + "operation", + "Development", + "OECD", + "published", + "model", + "rules", + "Tax", + "Challenges", + "Arising", + "From", + "Digitalisation", + "Economy", + "Global", + "Anti", + "Base", + "Erosion", + "Model", + "Rules", + "Pillar", + "Two", + "established", + "global", + "minimum", + "corporate", + "tax", + "rate", + "15%", + "multinational", + "enterprises", + "Many", + "countries", + "implemented", + "process", + "implementing", + "legislation", + "applies", + "Microsoft", + "beginning", + "fiscal", + "year", + "2025", + "While", + "estimate", + "material", + "impact", + "consolidated", + "financial", + "statements", + "continue", + "monitor", + "implement", + "additional", + "guidance", + "We", + "assessing", + "One", + "Big", + "Beautiful", + "Bill", + "Act", + "OBBBA", + "enacted", + "July", + "intangible", + "low", + "taxed", + "income", + "effective", + "14%", + "2027", + "It", + "bonus", + "depreciation", + "assets", + "service", + "January", + "19", + "election", + "expense", + "incurred", + "research", + "experimental", + "expenditures", + "Uncertain", + "Positions", + "remain", + "audit", + "IRS", + "years", + "2014", + "2017", + "With", + "respect", + "2004", + "2013", + "September", + "26", + "2023", + "received", + "Notices", + "Proposed", + "Adjustment", + "NOPAs", + "primary", + "issues", + "relate", + "intercompany", + "transfer", + "pricing", + "In", + "seeking", + "payment", + "28.9", + "billion", + "penalties", + "interest", + "As", + "June", + "30", + "allowances", + "contingencies", + "adequate", + "disagree", + "proposed", + "adjustments", + "vigorously", + "contest", + "administrative", + "appeals", + "office", + "judicial", + "proceedings", + "expect", + "final", + "resolution", + "12", + "months", + "Based", + "information", + "anticipate", + "significant", + "increase", + "decrease", + "subject", + "jurisdictions", + "local", + "authorities", + "audits", + "expected", + "Our", + "operations", + "Ireland", + "examination", + "2020", + "LIQUIDITY", + "AND", + "CAPITAL", + "RESOURCES", + "existing", + "cash", + "equivalents", + "short", + "term", + "investments", + "flows", + "access", + "capital", + "markets", + "sufficient", + "fund", + "operating", + "activities", + "commitments", + "investing", + "financing", + "dividends", + "share", + "repurchases", + "debt", + "maturities", + "transition", + "related", + "Cuts", + "Jobs", + "TCJA", + "foreseeable", + "future", + "Cash", + "Equivalents", + "Investments", + "totaled", + "94.6", + "75.5", + "2024", + "Equity", + "15.4", + "14.6", + "primarily", + "intended", + "facilitate", + "liquidity", + "preservation", + "They", + "consist", + "predominantly", + "highly", + "liquid", + "investment", + "grade", + "fixed", + "securities", + "diversified", + "industries", + "individual", + "issuers", + "dollar", + "denominated", + "include", + "foreign", + "currency", + "diversify", + "risk", + "exposed", + "credit", + "average", + "maturity", + "portfolio", + "managed", + "achieve", + "economic", + "returns", + "correlate", + "indices", + "settlement", + "insignificant", + "held", + "Valuation" + ], + "keywords": [ + "Pillar Two", + "IRS Audit", + "Cash Position" + ], + "connect_to": [] + } + }, + { + "chunk_id": "33116602-bb3d-5eb4-9a05-5ce844a3c5c6", + "type": "text", + "content": "In general, and where applicable, we use quoted prices in active markets for identical assets or liabilities to determine the fair value of our financial instruments. This pricing methodology applies to our Level 1 investments, such as U.S. government securities, common and preferred stock, and mutual funds. If quoted prices in active markets for identical assets or liabilities are not available to determine fair value, then we use quoted prices for similar assets and liabilities or inputs other than the quoted prices that are observable either directly or indirectly. This pricing methodology applies to our Level 2 investments, such as commercial paper, certificates of deposit, U.S. agency securities, foreign government bonds, mortgage- and asset-backed securities, corporate notes and bonds, and municipal securities. Level 3 investments are valued using internally-developed models with unobservable inputs. Assets and liabilities measured at fair value on a recurring basis using unobservable inputs are an immaterial portion of our portfolio.\nA majority of our investments are priced by pricing vendors and are generally Level 1 or Level 2 investments as these vendors either provide a quoted market price in an active market or use observable inputs for their pricing without applying significant adjustments. Broker pricing is used mainly when a quoted price is not available, the investment is not priced by our pricing vendors, or when a broker price is more reflective of fair values in the market in which the investment trades. Our broker-priced investments are generally classified as Level 2 investments because the broker prices these investments based on similar assets without applying significant adjustments. In addition, all our broker-priced investments have a sufficient level of trading volume to demonstrate that the fair values used are appropriate for these investments. Our fair value processes include controls that are designed to ensure appropriate fair values are recorded. These controls include model validation, review of key model inputs, analysis of period-over-period fluctuations, and independent recalculation of prices where appropriate.\nCash Flows\nCash from operations increased $17.6 billion to $136.2 billion for fiscal year 2025, primarily due to an increase in cash received from customers, offset in part by an increase in cash paid to suppliers and employees and cash used to pay income taxes. Cash used in financing increased $13.9 billion to $51.7 billion for fiscal year 2025, primarily due to a $9.5 billion increase in cash used for repayments of debt, net of proceeds. Cash used in investing decreased $24.4 billion to $72.6 billion for fiscal year 2025, primarily due to a $63.2 billion decrease in cash used for acquisitions of companies, net of cash acquired and divestitures, and purchases of intangible and other assets, offset in part by a $22.3 billion increase in cash used in net investment purchases, sales, and maturities, and a $20.1 billion increase in additions to property and equipment.\nDebt Proceeds\nWe issue debt to take advantage of favorable pricing and liquidity in the debt markets, reflecting our credit rating. The proceeds of these issuances were or will be used for general corporate purposes, which may include, among other things, funding for working capital, capital expenditures, repurchases of capital stock, acquisitions, and repayment of existing debt. Refer to Note 10 – Debt of the Notes to Financial Statements.\nUnearned Revenue\nUnearned revenue comprises mainly unearned revenue related to volume licensing programs, which may include cloud services and Software Assurance (“SA”). Unearned revenue is generally invoiced annually at the beginning of each contract period for multi-year agreements and recognized ratably over the coverage period. Unearned revenue also includes payments for other offerings for which we have been paid in advance and earn the revenue when we transfer control of the product or service. Refer to Note 1 – Accounting Policies of the Notes to Financial Statements.\nThe following table outlines the expected future recognition of unearned revenue as of June 30, 2025:\n\nThe table shows quarterly revenue figures from September 2025 to June 2026, with a total of $67,265 million.\n[tables/table-13 Revenue Schedule.html]", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 23", + "metadata": { + "length": 4343, + "summary": "The passage discusses fair value measurement of financial instruments, classifying them into Level 1 (quoted market prices), Level 2 (observable inputs), and Level 3 (unobservable inputs). Most investments are Level 1 or 2, priced by vendors or brokers. Cash flows show operations increased $17.6B to $136.2B, financing used $13.9B more, and investing decreased $24.4B. Debt proceeds fund corporate purposes. Unearned revenue, mainly from volume licensing and cloud services, totals $67,265 million expected to be recognized through June 2026.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "In", + "general", + "applicable", + "quoted", + "prices", + "active", + "markets", + "identical", + "assets", + "liabilities", + "determine", + "fair", + "financial", + "instruments", + "This", + "pricing", + "methodology", + "applies", + "Level", + "investments", + "government", + "securities", + "common", + "preferred", + "stock", + "mutual", + "funds", + "If", + "similar", + "inputs", + "observable", + "directly", + "indirectly", + "commercial", + "paper", + "certificates", + "deposit", + "agency", + "foreign", + "bonds", + "mortgage", + "asset", + "backed", + "corporate", + "notes", + "municipal", + "valued", + "internally", + "developed", + "models", + "unobservable", + "Assets", + "measured", + "recurring", + "basis", + "immaterial", + "portion", + "portfolio", + "majority", + "priced", + "vendors", + "generally", + "provide", + "market", + "price", + "applying", + "significant", + "adjustments", + "Broker", + "investment", + "broker", + "reflective", + "values", + "trades", + "Our", + "classified", + "based", + "addition", + "sufficient", + "level", + "trading", + "volume", + "demonstrate", + "processes", + "include", + "controls", + "designed", + "ensure", + "recorded", + "These", + "model", + "validation", + "review", + "key", + "analysis", + "period", + "fluctuations", + "independent", + "recalculation", + "Cash", + "Flows", + "operations", + "increased", + "17.6", + "billion", + "136.2", + "fiscal", + "year", + "2025", + "primarily", + "due", + "increase", + "cash", + "received", + "customers", + "offset", + "part", + "paid", + "suppliers", + "employees", + "pay", + "income", + "taxes", + "financing", + "13.9", + "51.7", + "9.5", + "repayments", + "debt", + "net", + "proceeds", + "investing", + "decreased", + "24.4", + "72.6", + "63.2", + "decrease", + "acquisitions", + "companies", + "acquired", + "divestitures", + "purchases", + "intangible", + "22.3", + "sales", + "maturities", + "20.1", + "additions", + "property", + "equipment", + "Debt", + "Proceeds", + "We", + "issue", + "advantage", + "favorable", + "liquidity", + "reflecting", + "credit", + "rating", + "The", + "issuances", + "purposes", + "things", + "funding", + "working", + "capital", + "expenditures", + "repurchases", + "repayment", + "existing", + "Refer", + "Note", + "10", + "Notes", + "Financial", + "Statements", + "Unearned", + "Revenue", + "revenue", + "comprises", + "unearned", + "related", + "licensing", + "programs", + "cloud", + "services", + "Software", + "Assurance", + "SA", + "invoiced", + "annually", + "beginning", + "contract", + "multi", + "agreements", + "recognized", + "ratably", + "coverage", + "includes", + "payments", + "offerings", + "advance", + "earn", + "transfer", + "control", + "product", + "service", + "Accounting", + "Policies", + "table", + "outlines", + "expected", + "future", + "recognition", + "June", + "30", + "shows", + "quarterly", + "figures", + "September", + "2026", + "total", + "67", + "265", + "million" + ], + "keywords": [ + "Fair value", + "Cash flows", + "Unearned revenue" + ], + "connect_to": [ + { + "target": "82944171-cfed-507c-a902-82c89f467531", + "relation": "embeds", + "ref": "[tables/table-13 Revenue Schedule.html]", + "position": { + "start": 4304, + "end": 4343 + } + } + ] + } + }, + { + "chunk_id": "ecfe0d62-fe3d-567a-aaec-4c707a41b54a", + "type": "text", + "content": "If our customers choose to license cloud-based versions of our products and services rather than licensing transaction-based products and services, the associated revenue will shift from being recognized at the time of the transaction to being recognized over the subscription period or upon consumption, as applicable. Refer to Note 12 – Unearned Revenue of the Notes to Financial Statements.\nMaterial Cash Requirements and Other Obligations\nContractual Obligations\nThe following table summarizes the payments due by fiscal year for our outstanding contractual obligations as of June 30, 2025:\n\nThe table summarizes contractual obligations totaling $397,045 million, including principal and interest payments, construction commitments, leases, and purchase commitments, with $148,106 million due in 2026 and $248,939 million thereafter.\n[tables/table-14 Debt Commitments.html]\n\n(a) Refer to Note 10 – Debt of the Notes to Financial Statements.\n(b) Refer to Note 6 – Property and Equipment of the Notes to Financial Statements.\n(c) Refer to Note 13 – Leases of the Notes to Financial Statements.\n(d) Purchase commitments primarily relate to datacenters and include open purchase orders and take-or-pay contracts that are not presented as construction commitments above.\nIncome Taxes\nAs a result of the TCJA, we are required to pay a one-time transition tax on deferred foreign income not previously subject to U.S. income tax. Under the TCJA, the transition tax is payable in interest-free installments over eight years, with 8% due in each of the first five years, 15% in year six, 20% in year seven, and 25% in year eight. As of June 30, 2025, our eighth transition tax installment of $4.4 billion is short-term and payable in the first quarter of fiscal year 2026.\nShare Repurchases\nDuring fiscal years 2025 and 2024, we repurchased 31 million shares and 32 million shares of our common stock for $13.0 billion and $12.0 billion, respectively, through our share repurchase program. All repurchases were made using cash resources. As of June 30, 2025, $57.3 billion remained of our $60 billion share repurchase program. Refer to Note 15 – Stockholders’ Equity of the Notes to Financial Statements.\nDividends\nDuring fiscal years 2025 and 2024, our Board of Directors declared dividends totaling $24.7 billion and $22.3 billion, respectively. We intend to continue returning capital to shareholders in the form of dividends, subject to declaration by our Board of Directors. Refer to Note 15 – Stockholders’ Equity of the Notes to Financial Statements.\nOther Planned Uses of Capital\nWe will continue to invest in sales, marketing, product support infrastructure, and existing and advanced areas of technology, as well as acquisitions that align with our business strategy. Additions to property and equipment will continue, including new facilities, datacenters, and computer systems for research and development, sales and marketing, support, and administrative staff. We will continue to invest in capital expenditures to support growth in our cloud offerings and our investments in AI infrastructure and training. We have operating and finance leases for datacenters, corporate offices, research and development facilities, Microsoft Experience Centers, and certain equipment. We have not engaged in any related party transactions or arrangements with unconsolidated entities or other persons that are reasonably likely to materially affect liquidity or the availability of capital resources.\nRECENT ACCOUNTING GUIDANCE\nRefer to Note 1 – Accounting Policies of the Notes to Financial Statements.\nCRITICAL ACCOUNTING ESTIMATES\nOur consolidated financial statements and accompanying notes are prepared in accordance with GAAP. Preparing consolidated financial statements requires management to make estimates and assumptions that affect the reported amounts of assets, liabilities, revenue, and expenses. Critical accounting estimates are those estimates that involve a significant level of estimation uncertainty and could have a material impact on our financial condition or results of operations. We have critical accounting estimates in the areas of revenue recognition, impairment of investment securities, goodwill, research and development costs, legal and other contingencies, and income taxes.\nRevenue Recognition", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 24", + "metadata": { + "length": 4323, + "summary": "The passage details Microsoft's material cash requirements and obligations as of June 30, 2025, including $397,045 million in contractual obligations (debt, leases, purchase commitments) with $148,106 million due in 2026. It covers a $4.4 billion transition tax installment due in Q1 FY2026, $13.0 billion in share repurchases during FY2025 with $57.3 billion remaining authorized, $24.7 billion in dividends declared, and planned capital investments in cloud, AI infrastructure, and facilities.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "If", + "customers", + "choose", + "license", + "cloud", + "based", + "versions", + "products", + "services", + "licensing", + "transaction", + "revenue", + "shift", + "recognized", + "time", + "subscription", + "period", + "consumption", + "applicable", + "Refer", + "Note", + "12", + "Unearned", + "Revenue", + "Notes", + "Financial", + "Statements", + "Material", + "Cash", + "Requirements", + "Other", + "Obligations", + "Contractual", + "The", + "table", + "summarizes", + "payments", + "due", + "fiscal", + "year", + "outstanding", + "contractual", + "obligations", + "June", + "30", + "2025", + "totaling", + "397", + "045", + "million", + "including", + "principal", + "interest", + "construction", + "commitments", + "leases", + "purchase", + "148", + "106", + "2026", + "248", + "939", + "10", + "Debt", + "Property", + "Equipment", + "13", + "Leases", + "Purchase", + "primarily", + "relate", + "datacenters", + "include", + "open", + "orders", + "pay", + "contracts", + "presented", + "Income", + "Taxes", + "As", + "result", + "TCJA", + "required", + "transition", + "tax", + "deferred", + "foreign", + "income", + "previously", + "subject", + "Under", + "payable", + "free", + "installments", + "years", + "8%", + "15%", + "20%", + "25%", + "eighth", + "installment", + "4.4", + "billion", + "short", + "term", + "quarter", + "Share", + "Repurchases", + "During", + "2024", + "repurchased", + "31", + "shares", + "32", + "common", + "stock", + "13.0", + "12.0", + "share", + "repurchase", + "program", + "All", + "repurchases", + "made", + "cash", + "resources", + "57.3", + "remained", + "60", + "15", + "Stockholders", + "Equity", + "Dividends", + "Board", + "Directors", + "declared", + "dividends", + "24.7", + "22.3", + "We", + "intend", + "continue", + "returning", + "capital", + "shareholders", + "form", + "declaration", + "Planned", + "Uses", + "Capital", + "invest", + "sales", + "marketing", + "product", + "support", + "infrastructure", + "existing", + "advanced", + "areas", + "technology", + "acquisitions", + "align", + "business", + "strategy", + "Additions", + "property", + "equipment", + "facilities", + "computer", + "systems", + "research", + "development", + "administrative", + "staff", + "expenditures", + "growth", + "offerings", + "investments", + "AI", + "training", + "operating", + "finance", + "corporate", + "offices", + "Microsoft", + "Experience", + "Centers", + "engaged", + "related", + "party", + "transactions", + "arrangements", + "unconsolidated", + "entities", + "persons", + "materially", + "affect", + "liquidity", + "availability", + "RECENT", + "ACCOUNTING", + "GUIDANCE", + "Accounting", + "Policies", + "CRITICAL", + "ESTIMATES", + "Our", + "consolidated", + "financial", + "statements", + "accompanying", + "notes", + "prepared", + "accordance", + "GAAP", + "Preparing", + "requires", + "management", + "make", + "estimates", + "assumptions", + "reported", + "amounts", + "assets", + "liabilities", + "expenses", + "Critical", + "accounting", + "involve", + "significant", + "level", + "estimation", + "uncertainty", + "material", + "impact", + "condition", + "results", + "operations", + "critical", + "recognition", + "impairment", + "investment", + "securities", + "goodwill", + "costs", + "legal", + "contingencies", + "taxes", + "Recognition" + ], + "keywords": [ + "contractual obligations", + "share repurchases", + "dividends" + ], + "connect_to": [ + { + "target": "8d1714c5-e877-501e-83fa-d4750007bb69", + "relation": "embeds", + "ref": "[tables/table-14 Debt Commitments.html]", + "position": { + "start": 838, + "end": 877 + } + } + ] + } + }, + { + "chunk_id": "2943c163-4f7e-5f78-9400-0ed277dc7a92", + "type": "text", + "content": "Our contracts with customers often include promises to transfer multiple products and services to a customer. Determining whether products and services are considered distinct performance obligations that should be accounted for separately versus together may require significant judgment. When a cloud-based service includes both on-premises software licenses and cloud services, judgment is required to determine whether the software license is considered distinct and accounted for separately, or not distinct and accounted for together with the cloud service and recognized over time. Certain cloud services, primarily Office 365, depend on a significant level of integration, interdependency, and interrelation between the desktop applications and cloud services, and are accounted for together as one performance obligation. Revenue from Office 365 is recognized ratably over the period in which the cloud services are provided.\nJudgment is required to determine the standalone selling price (“SSP”) for each distinct performance obligation. We use a single amount to estimate SSP for items that are not sold separately, including on-premises licenses sold with SA or software updates provided at no additional charge. We use a range of amounts to estimate SSP when we sell each of the products and services separately and need to determine whether there is a discount to be allocated based on the relative SSP of the various products and services.\nIn instances where SSP is not directly observable, such as when we do not sell the product or service separately, we determine the SSP using information that may include market conditions and other observable inputs. We typically have more than one SSP for individual products and services due to the stratification of those products and services by customers and circumstances. In these instances, we may use information such as the size of the customer and geographic region in determining the SSP.\nDue to the various benefits from and the nature of our SA program, judgment is required to assess the pattern of delivery, including the exercise pattern of certain benefits across our portfolio of customers.\nOur products are generally sold with a right of return, we may provide other credits or incentives, and in certain instances we estimate customer usage of our products and services, which are accounted for as variable consideration when determining the amount of revenue to recognize. Returns and credits are estimated at contract inception and updated at the end of each reporting period if additional information becomes available. Changes to our estimated variable consideration were not material for the periods presented.\nImpairment of Investment Securities\nWe review debt investments quarterly for credit losses and impairment. If the cost of an investment exceeds its fair value, we evaluate, among other factors, general market conditions, credit quality of debt instrument issuers, and the extent to which the fair value is less than cost. This determination requires significant judgment. In making this judgment, we employ a systematic methodology that considers available quantitative and qualitative evidence in evaluating potential impairment of our investments. In addition, we consider specific adverse conditions related to the financial health of, and business outlook for, the investee. If we have plans to sell the security or it is more likely than not that we will be required to sell the security before recovery, then a decline in fair value below cost is recorded as an impairment charge in other income (expense), net and a new cost basis in the investment is established. If market, industry, and/or investee conditions deteriorate, we may incur future impairments.\nEquity investments without readily determinable fair values are written down to fair value if a qualitative assessment indicates that the investment is impaired and the fair value of the investment is less than carrying value. We perform a qualitative assessment on a periodic basis. We are required to estimate the fair value of the investment to determine the amount of the impairment loss. Once an investment is determined to be impaired, an impairment charge is recorded in other income (expense), net.\nGoodwill", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 25", + "metadata": { + "length": 4273, + "summary": "The passage discusses revenue recognition judgments for contracts with multiple products and services, including determining distinct performance obligations, standalone selling price estimation, variable consideration for returns and credits, and impairment assessments for investment securities and goodwill.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Our", + "contracts", + "customers", + "include", + "promises", + "transfer", + "multiple", + "products", + "services", + "customer", + "Determining", + "considered", + "distinct", + "performance", + "obligations", + "accounted", + "separately", + "versus", + "require", + "significant", + "judgment", + "When", + "cloud", + "based", + "service", + "includes", + "premises", + "software", + "licenses", + "required", + "determine", + "license", + "recognized", + "time", + "Certain", + "primarily", + "Office", + "365", + "depend", + "level", + "integration", + "interdependency", + "interrelation", + "desktop", + "applications", + "obligation", + "Revenue", + "ratably", + "period", + "provided", + "Judgment", + "standalone", + "selling", + "price", + "SSP", + "We", + "single", + "amount", + "estimate", + "items", + "sold", + "including", + "SA", + "updates", + "additional", + "charge", + "range", + "amounts", + "sell", + "discount", + "allocated", + "relative", + "In", + "instances", + "directly", + "observable", + "product", + "information", + "market", + "conditions", + "inputs", + "typically", + "individual", + "due", + "stratification", + "circumstances", + "size", + "geographic", + "region", + "determining", + "Due", + "benefits", + "nature", + "program", + "assess", + "pattern", + "delivery", + "exercise", + "portfolio", + "generally", + "return", + "provide", + "credits", + "incentives", + "usage", + "variable", + "consideration", + "revenue", + "recognize", + "Returns", + "estimated", + "contract", + "inception", + "updated", + "end", + "reporting", + "Changes", + "material", + "periods", + "presented", + "Impairment", + "Investment", + "Securities", + "review", + "debt", + "investments", + "quarterly", + "credit", + "losses", + "impairment", + "If", + "cost", + "investment", + "exceeds", + "fair", + "evaluate", + "factors", + "general", + "quality", + "instrument", + "issuers", + "extent", + "This", + "determination", + "requires", + "making", + "employ", + "systematic", + "methodology", + "considers", + "quantitative", + "qualitative", + "evidence", + "evaluating", + "potential", + "addition", + "specific", + "adverse", + "related", + "financial", + "health", + "business", + "outlook", + "investee", + "plans", + "security", + "recovery", + "decline", + "recorded", + "income", + "expense", + "net", + "basis", + "established", + "industry", + "deteriorate", + "incur", + "future", + "impairments", + "Equity", + "readily", + "determinable", + "values", + "written", + "assessment", + "impaired", + "carrying", + "perform", + "periodic", + "loss", + "Once", + "determined", + "Goodwill" + ], + "keywords": [ + "Revenue", + "Judgment", + "Contracts" + ], + "connect_to": [] + } + }, + { + "chunk_id": "54fbdc6d-1456-59be-9b31-606bebefed7a", + "type": "text", + "content": "We allocate goodwill to reporting units based on the reporting unit expected to benefit from the business combination. We evaluate our reporting units on an annual basis and, if necessary, reassign goodwill using a relative fair value allocation approach. Goodwill is tested for impairment at the reporting unit level (operating segment or one level below an operating segment) on an annual basis (May 1) and between annual tests if an event occurs or circumstances change that would more likely than not reduce the fair value of a reporting unit below its carrying value. These events or circumstances could include a significant change in the business climate, legal factors, operating performance indicators, competition, or sale or disposition of a significant portion of a reporting unit.\nApplication of the goodwill impairment test requires judgment, including the identification of reporting units, assignment of assets and liabilities to reporting units, assignment of goodwill to reporting units, and determination of the fair value of each reporting unit. The fair value of each reporting unit is estimated primarily through the use of a discounted cash flow methodology. This analysis requires significant judgments, including estimation of future cash flows, which is dependent on internal forecasts, estimation of the long-term rate of growth for our business, estimation of the useful life over which cash flows will occur, and determination of our weighted average cost of capital.\nThe estimates used to calculate the fair value of a reporting unit change from year to year based on operating results, market conditions, and other factors. Changes in these estimates and assumptions could materially affect the determination of fair value and goodwill impairment for each reporting unit.\nResearch and Development Costs\nCosts incurred internally in researching and developing a software product to be marketed or sold to external users are charged to expense until technological feasibility has been established for the product. Once technological feasibility is established, software costs are capitalized until the product is available for general release to customers. Judgment is required in determining when technological feasibility of a product is established. We have determined that technological feasibility for our software products is reached after all high-risk development issues have been resolved through coding and testing. Generally, this occurs shortly before the products are released to production. The amortization of these costs is included in cost of revenue over the estimated life of the products.\nLegal and Other Contingencies\nThe outcomes of legal proceedings and claims brought against us are subject to significant uncertainty. An estimated loss from a loss contingency such as a legal proceeding or claim is accrued by a charge to income if it is probable that an asset has been impaired or a liability has been incurred and the amount of the loss can be reasonably estimated. In determining whether a loss should be accrued we evaluate, among other factors, the degree of probability of an unfavorable outcome and the ability to make a reasonable estimate of the amount of loss. Changes in these factors could materially impact our consolidated financial statements.\nIncome Taxes", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 26", + "metadata": { + "length": 3325, + "summary": "The passage discusses accounting policies for goodwill, research and development costs, legal contingencies, and income taxes. Goodwill is allocated to reporting units and tested for impairment annually or upon triggering events, using discounted cash flow analysis requiring significant judgment. R&D costs are expensed until technological feasibility is established, then capitalized. Legal loss contingencies are accrued when probable and reasonably estimable. Income taxes are also mentioned.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "We", + "allocate", + "goodwill", + "reporting", + "units", + "based", + "unit", + "expected", + "benefit", + "business", + "combination", + "evaluate", + "annual", + "basis", + "reassign", + "relative", + "fair", + "allocation", + "approach", + "Goodwill", + "tested", + "impairment", + "level", + "operating", + "segment", + "May", + "tests", + "event", + "occurs", + "circumstances", + "change", + "reduce", + "carrying", + "These", + "events", + "include", + "significant", + "climate", + "legal", + "factors", + "performance", + "indicators", + "competition", + "sale", + "disposition", + "portion", + "Application", + "test", + "requires", + "judgment", + "including", + "identification", + "assignment", + "assets", + "liabilities", + "determination", + "The", + "estimated", + "primarily", + "discounted", + "cash", + "flow", + "methodology", + "This", + "analysis", + "judgments", + "estimation", + "future", + "flows", + "dependent", + "internal", + "forecasts", + "long", + "term", + "rate", + "growth", + "life", + "occur", + "weighted", + "average", + "cost", + "capital", + "estimates", + "calculate", + "year", + "results", + "market", + "conditions", + "Changes", + "assumptions", + "materially", + "affect", + "Research", + "Development", + "Costs", + "incurred", + "internally", + "researching", + "developing", + "software", + "product", + "marketed", + "sold", + "external", + "users", + "charged", + "expense", + "technological", + "feasibility", + "established", + "Once", + "costs", + "capitalized", + "general", + "release", + "customers", + "Judgment", + "required", + "determining", + "determined", + "products", + "reached", + "high", + "risk", + "development", + "issues", + "resolved", + "coding", + "testing", + "Generally", + "shortly", + "released", + "production", + "amortization", + "included", + "revenue", + "Legal", + "Other", + "Contingencies", + "outcomes", + "proceedings", + "claims", + "brought", + "subject", + "uncertainty", + "An", + "loss", + "contingency", + "proceeding", + "claim", + "accrued", + "charge", + "income", + "probable", + "asset", + "impaired", + "liability", + "amount", + "In", + "degree", + "probability", + "unfavorable", + "outcome", + "ability", + "make", + "reasonable", + "estimate", + "impact", + "consolidated", + "financial", + "statements", + "Income", + "Taxes" + ], + "keywords": [ + "Goodwill impairment", + "R&D costs", + "Contingencies" + ], + "connect_to": [] + } + }, + { + "chunk_id": "12bbfcd8-5633-5ddd-b496-47d8bd0d731f", + "type": "text", + "content": "The objectives of accounting for income taxes are to recognize the amount of taxes payable or refundable for the current year, and deferred tax liabilities and assets for the future tax consequences of events that have been recognized in an entity’s financial statements or tax returns. We recognize the tax benefit from an uncertain tax position only if it is more likely than not that the tax position will be sustained on examination by the taxing authorities, based on the technical merits of the position. The tax benefits recognized in the financial statements from such a position are measured based on the largest benefit that has a greater than 50% likelihood of being realized upon ultimate settlement. Accounting literature also provides guidance on derecognition of income tax assets and liabilities, classification of deferred income tax assets and liabilities, accounting for interest and penalties associated with tax positions, and income tax disclosures. Judgment is required in assessing the future tax consequences of events that have been recognized in our consolidated financial statements or tax returns. Variations in the actual outcome of these future tax consequences could materially impact our consolidated financial statements.\nSTATEMENT OF MANAGEMENT’S RESPONSIBILITY FOR FINANCIAL STATEMENTS\nManagement is responsible for the preparation of the consolidated financial statements and related information that are presented in this report. The consolidated financial statements, which include amounts based on management’s estimates and judgments, have been prepared in conformity with accounting principles generally accepted in the United States of America.\nThe Company designs and maintains accounting and internal control systems to provide reasonable assurance at reasonable cost that assets are safeguarded against loss from unauthorized use or disposition, and that the financial records are reliable for preparing consolidated financial statements and maintaining accountability for assets. These systems are augmented by written policies, an organizational structure providing division of responsibilities, careful selection and training of qualified personnel, and a program of internal audits.\nThe Company engaged Deloitte & Touche LLP, an independent registered public accounting firm, to audit and render an opinion on the consolidated financial statements and internal control over financial reporting in accordance with the standards of the Public Company Accounting Oversight Board (United States).\nThe Board of Directors, through its Audit Committee, consisting solely of independent directors of the Company, meets periodically with management, internal auditors, and our independent registered public accounting firm to ensure that each is meeting its responsibilities and to discuss matters concerning internal controls and financial reporting. Deloitte & Touche LLP and the internal auditors each have full and free access to the Audit Committee.\nSatya Nadella\nChief Executive Officer\nAmy E. Hood\nExecutive Vice President and Chief Financial Officer\nAlice L. Jolla\nCorporate Vice President and Chief Accounting Officer\nQUANTITATIVE AND QUALITATIVE DISCLOSURES ABOUT MARKET RISK\nRISKS\nWe are exposed to economic risk from foreign exchange rates, interest rates, credit risk, and equity prices. We use derivatives instruments to manage these risks, however, they may still impact our consolidated financial statements.\nForeign Currencies\nCertain forecasted transactions, assets, and liabilities are exposed to foreign currency risk. We monitor our foreign currency exposures daily to maximize the economic effectiveness of our foreign currency positions, including hedges. Principal currency exposures include the Euro, Japanese yen, British pound, Canadian dollar, and Australian dollar.\nInterest Rate\nSecurities held in our fixed-income portfolio are subject to different interest rate risks based on their maturities. We manage the average maturity of the fixed-income portfolio to achieve economic returns that correlate to certain global fixed-income indices.\nCredit\nOur fixed-income portfolio is diversified and consists primarily of investment-grade securities. We manage credit exposures relative to broad-based indices to facilitate portfolio diversification.\nEquity\nSecurities held in our equity investments portfolio are subject to price risk.\nSENSITIVITY ANALYSIS", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 27", + "metadata": { + "length": 4422, + "summary": "The passage covers accounting for income taxes, including recognizing current and deferred tax liabilities/assets, uncertain tax positions, and management's responsibility for financial statements. It also discusses market risk exposures (foreign currency, interest rate, credit, equity) and the use of derivatives for risk management.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "The", + "objectives", + "accounting", + "income", + "taxes", + "recognize", + "amount", + "payable", + "refundable", + "current", + "year", + "deferred", + "tax", + "liabilities", + "assets", + "future", + "consequences", + "events", + "recognized", + "entity", + "financial", + "statements", + "returns", + "We", + "benefit", + "uncertain", + "position", + "sustained", + "examination", + "taxing", + "authorities", + "based", + "technical", + "merits", + "benefits", + "measured", + "largest", + "greater", + "50%", + "likelihood", + "realized", + "ultimate", + "settlement", + "Accounting", + "literature", + "guidance", + "derecognition", + "classification", + "interest", + "penalties", + "positions", + "disclosures", + "Judgment", + "required", + "assessing", + "consolidated", + "Variations", + "actual", + "outcome", + "materially", + "impact", + "STATEMENT", + "OF", + "MANAGEMENT", + "RESPONSIBILITY", + "FOR", + "FINANCIAL", + "STATEMENTS", + "Management", + "responsible", + "preparation", + "related", + "information", + "presented", + "report", + "include", + "amounts", + "management", + "estimates", + "judgments", + "prepared", + "conformity", + "principles", + "generally", + "accepted", + "United", + "States", + "America", + "Company", + "designs", + "maintains", + "internal", + "control", + "systems", + "provide", + "reasonable", + "assurance", + "cost", + "safeguarded", + "loss", + "unauthorized", + "disposition", + "records", + "reliable", + "preparing", + "maintaining", + "accountability", + "These", + "augmented", + "written", + "policies", + "organizational", + "structure", + "providing", + "division", + "responsibilities", + "careful", + "selection", + "training", + "qualified", + "personnel", + "program", + "audits", + "engaged", + "Deloitte", + "Touche", + "LLP", + "independent", + "registered", + "public", + "firm", + "audit", + "render", + "opinion", + "reporting", + "accordance", + "standards", + "Public", + "Oversight", + "Board", + "Directors", + "Audit", + "Committee", + "consisting", + "solely", + "directors", + "meets", + "periodically", + "auditors", + "ensure", + "meeting", + "discuss", + "matters", + "controls", + "full", + "free", + "access", + "Satya", + "Nadella", + "Chief", + "Executive", + "Officer", + "Amy", + "Hood", + "Vice", + "President", + "Financial", + "Alice", + "Jolla", + "Corporate", + "QUANTITATIVE", + "AND", + "QUALITATIVE", + "DISCLOSURES", + "ABOUT", + "MARKET", + "RISK", + "RISKS", + "exposed", + "economic", + "risk", + "foreign", + "exchange", + "rates", + "credit", + "equity", + "prices", + "derivatives", + "instruments", + "manage", + "risks", + "Foreign", + "Currencies", + "Certain", + "forecasted", + "transactions", + "currency", + "monitor", + "exposures", + "daily", + "maximize", + "effectiveness", + "including", + "hedges", + "Principal", + "Euro", + "Japanese", + "yen", + "British", + "pound", + "Canadian", + "dollar", + "Australian", + "Interest", + "Rate", + "Securities", + "held", + "fixed", + "portfolio", + "subject", + "rate", + "maturities", + "average", + "maturity", + "achieve", + "correlate", + "global", + "indices", + "Credit", + "Our", + "diversified", + "consists", + "primarily", + "investment", + "grade", + "securities", + "relative", + "broad", + "facilitate", + "diversification", + "Equity", + "investments", + "price", + "SENSITIVITY", + "ANALYSIS" + ], + "keywords": [ + "Income Tax", + "Financial Statements", + "Market Risk" + ], + "connect_to": [] + } + }, + { + "chunk_id": "77f172f1-9b2f-5c47-8d3c-6dab6de4f915", + "type": "text", + "content": "The following table sets forth the potential loss in future earnings or fair values, including associated derivatives, resulting from hypothetical changes in relevant market rates or prices:\n\nThe table shows risk categories (foreign currency, interest rate, credit, equity) with hypothetical changes and their impacts on earnings or fair value as of June 30, 2025.\n[tables/table-15 Risk Sensitivity.html]\n\nFINANCIAL STATEMENTS AND SUPPLEMENTARY DATA\nINCOME STATEMENTS\n\nThe table shows financial data for years ending June 30, 2025, 2024, and 2023. Total revenue grew from $211,915M to $281,724M, net income from $72,361M to $101,832M, and diluted EPS from $9.68 to $13.64.\n[tables/table-16 Revenue & Income.html]\n\nRefer to accompanying notes.\nCOMPREHENSIVE INCOME STATEMENTS\n\nThe table shows net income, other comprehensive income (OCI) components, and total comprehensive income for fiscal years 2025, 2024, and 2023, with net income rising from $72,361M to $101,832M.\n[tables/table-17 Comprehensive Income.html]\n\nRefer to accompanying notes.\nBALANCE SHEETS\n\nConsolidated balance sheet as of June 30, 2025 and 2024, showing total assets of $619,003 million and $512,163 million, with corresponding liabilities and stockholders' equity.\n[tables/table-18 Balance Sheet.html]\n\nRefer to accompanying notes.\nCASH FLOWS STATEMENTS\n\nThe table shows cash flow from operations, financing, and investing for fiscal years 2023-2025, with net cash from operations increasing from $87,582M to $136,162M.\n[tables/table-19 Cash Flow Statement.html]\n\nRefer to accompanying notes.\nSTOCKHOLDERS’ EQUITY STATEMENTS\n\nThe table shows stockholders' equity components for fiscal years 2023-2025, including common stock, retained earnings, and accumulated other comprehensive loss, with ending total equity of $343,479 million in 2025.\n[tables/table-20 Stockholders' Equity.html]\n\nRefer to accompanying notes.\nNOTES TO FINANCIAL STATEMENTS\nNOTE 1 — ACCOUNTING POLICIES\nAccounting Principles\nOur consolidated financial statements and accompanying notes are prepared in accordance with accounting principles generally accepted in the United States of America (“GAAP”).\nPrinciples of Consolidation\nThe consolidated financial statements include the accounts of Microsoft Corporation and its subsidiaries. Intercompany transactions and balances have been eliminated.\nRecast of Certain Prior Period Information\nIn August 2024, we announced changes to the composition of our segments. These changes align our segments with how we currently manage our business, most notably bringing the commercial components of Microsoft 365 together in the Productivity and Business Processes segment. Beginning in fiscal year 2025, the information that our chief operating decision maker is regularly provided and reviews for purposes of allocating resources and assessing performance reflects these segment changes. Prior period segment information has been recast to conform to the way we internally manage and monitor our business during fiscal year 2025. These changes primarily impacted Note 8 – Goodwill, Note 12 – Unearned Revenue, and Note 18 – Segment Information and Geographic Data.\nThe recast of prior period information had no impact on our consolidated balance sheets, consolidated income statements, or consolidated cash flows statements.\nEstimates and Assumptions\nPreparing financial statements requires management to make estimates and assumptions that affect the reported amounts of assets, liabilities, revenue, and expenses. Examples of estimates and assumptions include: for revenue recognition, determining the nature and timing of satisfaction of performance obligations, and determining the standalone selling price (“SSP”) of performance obligations, variable consideration, and other obligations such as product returns and refunds; loss contingencies; the fair value of and/or potential impairment of goodwill and intangible assets for our reporting units; product life cycles; useful lives of our tangible and intangible assets; allowances for doubtful accounts; stock-based compensation forfeiture rates; when technological feasibility is achieved for our products; the potential outcome of uncertain tax positions that have been recognized in our consolidated financial statements or tax returns; and determining the timing and amount of impairments for investments. Actual results and outcomes may differ from management’s estimates and assumptions due to risks and uncertainties.\nForeign Currencies", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 28", + "metadata": { + "length": 4486, + "summary": "The passage presents financial statements and supplementary data for a company (likely Microsoft) as of June 30, 2025. It includes tables on risk categories (foreign currency, interest rate, credit, equity) with hypothetical changes impacting earnings or fair value. Income statements show total revenue growing from $211,915M to $281,724M, net income from $72,361M to $101,832M, and diluted EPS from $9.68 to $13.64 over fiscal years 2023-2025. Comprehensive income statements, balance sheets (total assets $619,003M in 2025), cash flow statements (net cash from operations $136,162M in 2025), and stockholders' equity statements (ending total equity $343,479M in 2025) are included. Notes to financial statements cover accounting policies, principles of consolidation, recast of prior period segment information (effective August 2024), and estimates and assumptions.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "The", + "table", + "sets", + "potential", + "loss", + "future", + "earnings", + "fair", + "values", + "including", + "derivatives", + "resulting", + "hypothetical", + "relevant", + "market", + "rates", + "prices", + "shows", + "risk", + "categories", + "foreign", + "currency", + "interest", + "rate", + "credit", + "equity", + "impacts", + "June", + "30", + "2025", + "FINANCIAL", + "STATEMENTS", + "AND", + "SUPPLEMENTARY", + "DATA", + "INCOME", + "financial", + "data", + "years", + "ending", + "2024", + "2023", + "Total", + "revenue", + "grew", + "211", + "915M", + "281", + "724M", + "net", + "income", + "72", + "361M", + "101", + "832M", + "diluted", + "EPS", + "9.68", + "13.64", + "Refer", + "accompanying", + "notes", + "COMPREHENSIVE", + "comprehensive", + "OCI", + "components", + "total", + "fiscal", + "rising", + "BALANCE", + "SHEETS", + "Consolidated", + "balance", + "sheet", + "showing", + "assets", + "619", + "003", + "million", + "512", + "163", + "liabilities", + "stockholders", + "CASH", + "FLOWS", + "cash", + "flow", + "operations", + "financing", + "investing", + "increasing", + "87", + "582M", + "136", + "162M", + "STOCKHOLDERS", + "EQUITY", + "common", + "stock", + "retained", + "accumulated", + "343", + "479", + "NOTES", + "TO", + "NOTE", + "ACCOUNTING", + "POLICIES", + "Accounting", + "Principles", + "Our", + "consolidated", + "statements", + "prepared", + "accordance", + "accounting", + "principles", + "generally", + "accepted", + "United", + "States", + "America", + "GAAP", + "Consolidation", + "include", + "accounts", + "Microsoft", + "Corporation", + "subsidiaries", + "Intercompany", + "transactions", + "balances", + "eliminated", + "Recast", + "Certain", + "Prior", + "Period", + "Information", + "In", + "August", + "announced", + "composition", + "segments", + "These", + "align", + "manage", + "business", + "notably", + "bringing", + "commercial", + "365", + "Productivity", + "Business", + "Processes", + "segment", + "Beginning", + "year", + "information", + "chief", + "operating", + "decision", + "maker", + "regularly", + "provided", + "reviews", + "purposes", + "allocating", + "resources", + "assessing", + "performance", + "reflects", + "period", + "recast", + "conform", + "internally", + "monitor", + "primarily", + "impacted", + "Note", + "Goodwill", + "12", + "Unearned", + "Revenue", + "18", + "Segment", + "Geographic", + "Data", + "prior", + "impact", + "sheets", + "flows", + "Estimates", + "Assumptions", + "Preparing", + "requires", + "management", + "make", + "estimates", + "assumptions", + "affect", + "reported", + "amounts", + "expenses", + "Examples", + "recognition", + "determining", + "nature", + "timing", + "satisfaction", + "obligations", + "standalone", + "selling", + "price", + "SSP", + "variable", + "consideration", + "product", + "returns", + "refunds", + "contingencies", + "impairment", + "goodwill", + "intangible", + "reporting", + "units", + "life", + "cycles", + "lives", + "tangible", + "allowances", + "doubtful", + "based", + "compensation", + "forfeiture", + "technological", + "feasibility", + "achieved", + "products", + "outcome", + "uncertain", + "tax", + "positions", + "recognized", + "amount", + "impairments", + "investments", + "Actual", + "results", + "outcomes", + "differ", + "due", + "risks", + "uncertainties", + "Foreign", + "Currencies" + ], + "keywords": [ + "Financial Statements", + "Risk Analysis", + "Revenue Growth" + ], + "connect_to": [ + { + "target": "7d1314f0-065b-54b3-84f1-55e25d235575", + "relation": "embeds", + "ref": "[tables/table-15 Risk Sensitivity.html]", + "position": { + "start": 365, + "end": 404 + } + }, + { + "target": "172af7d7-ea9b-5403-8809-b2545b8ea94b", + "relation": "embeds", + "ref": "[tables/table-16 Revenue & Income.html]", + "position": { + "start": 673, + "end": 712 + } + }, + { + "target": "f996e041-a3b3-5bc1-b947-010fa5e45999", + "relation": "embeds", + "ref": "[tables/table-17 Comprehensive Income.html]", + "position": { + "start": 970, + "end": 1013 + } + }, + { + "target": "8fec4155-9cc3-5780-a857-96e83308701e", + "relation": "embeds", + "ref": "[tables/table-18 Balance Sheet.html]", + "position": { + "start": 1237, + "end": 1273 + } + }, + { + "target": "aaa89319-cae0-5959-a6fa-3b1a400b0a60", + "relation": "embeds", + "ref": "[tables/table-19 Cash Flow Statement.html]", + "position": { + "start": 1492, + "end": 1534 + } + }, + { + "target": "75b59a24-798a-519f-bcef-06088c97f96d", + "relation": "embeds", + "ref": "[tables/table-20 Stockholders' Equity.html]", + "position": { + "start": 1813, + "end": 1856 + } + } + ] + } + }, + { + "chunk_id": "a95275a9-cfe7-5300-9350-46502dd472fa", + "type": "text", + "content": "Assets and liabilities recorded in foreign currencies are translated at the exchange rate on the balance sheet date. Revenue and expenses are translated at average rates of exchange prevailing during the year. Translation adjustments resulting from this process are recorded to other comprehensive income.\nRevenue\nProduct Revenue and Service and Other Revenue\nProduct revenue includes sales from operating systems, cross-device productivity and collaboration applications, server applications, business solution applications, desktop and server management tools, software development tools, video games, and hardware such as PCs, tablets, gaming and entertainment consoles, other intelligent devices, and related accessories.\nService and other revenue includes sales from cloud-based solutions that provide customers with software, services, platforms, and content such as Office 365, Azure, Dynamics 365, and gaming; solution support; and consulting services. Service and other revenue also includes sales from online advertising and LinkedIn.\nRevenue Recognition\nRevenue is recognized upon transfer of control of promised products or services to customers in an amount that reflects the consideration we expect to receive in exchange for those products or services. We enter into contracts that can include various combinations of products and services, which are generally capable of being distinct and accounted for as separate performance obligations. Revenue is recognized net of allowances for returns and any taxes collected from customers, which are subsequently remitted to governmental authorities.\nNature of Products and Services\nLicenses for on-premises software provide the customer with a right to use the software as it exists when made available to the customer. Customers may purchase perpetual licenses or subscribe to licenses, which provide customers with the same functionality and differ mainly in the duration over which the customer benefits from the software. Revenue from distinct on-premises licenses is recognized upfront at the point in time when the software is made available to the customer. In cases where we allocate revenue to software updates, primarily because the updates are provided at no additional charge, revenue is recognized as the updates are provided, which is generally ratably over the estimated life of the related device or license.\nCloud services, which allow customers to use hosted software over the contract period without taking possession of the software, are provided on either a subscription or consumption basis. Revenue related to cloud services provided on a subscription basis is recognized ratably over the contract period. Revenue related to cloud services provided on a consumption basis, such as the amount of storage used in a period, is recognized based on the customer utilization of such resources. When cloud services require a significant level of integration and interdependency with software and the individual components are not considered distinct, all revenue is recognized over the period in which the cloud services are provided.\nCertain volume licensing programs, including Enterprise Agreements, include on-premises licenses combined with Software Assurance (“SA”). SA conveys rights to new software and upgrades released over the contract period and provides support, tools, and training to help customers deploy and use products more efficiently. On-premises licenses are considered distinct performance obligations when sold with SA. Revenue allocated to SA is generally recognized ratably over the contract period as customers simultaneously consume and receive benefits, given that SA comprises distinct performance obligations that are satisfied over time.\nRevenue from search advertising is recognized when the advertisement appears in the search results or when the action necessary to earn the revenue has been completed. Revenue from consulting services is recognized as services are provided.\nOur hardware is generally highly dependent on, and interrelated with, the underlying operating system and cannot function without the operating system. In these cases, the hardware and software license are accounted for as a single performance obligation and revenue is recognized at the point in time when ownership is transferred to resellers or directly to end customers through retail stores and online marketplaces.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 29", + "metadata": { + "length": 4407, + "summary": "Assets and liabilities in foreign currencies are translated at balance sheet date exchange rates, with revenue and expenses at average rates, and adjustments recorded to other comprehensive income. Revenue includes product revenue from software, hardware, and services like cloud solutions, advertising, and consulting. Revenue is recognized when control transfers to customers, net of returns and taxes. On-premises software licenses are recognized upfront, while cloud services are recognized ratably or based on consumption. Software Assurance revenue is recognized over the contract period. Hardware and software are often a single performance obligation recognized at transfer of ownership.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Assets", + "liabilities", + "recorded", + "foreign", + "currencies", + "translated", + "exchange", + "rate", + "balance", + "sheet", + "date", + "Revenue", + "expenses", + "average", + "rates", + "prevailing", + "year", + "Translation", + "adjustments", + "resulting", + "process", + "comprehensive", + "income", + "Product", + "Service", + "Other", + "revenue", + "includes", + "sales", + "operating", + "systems", + "cross", + "device", + "productivity", + "collaboration", + "applications", + "server", + "business", + "solution", + "desktop", + "management", + "tools", + "software", + "development", + "video", + "games", + "hardware", + "PCs", + "tablets", + "gaming", + "entertainment", + "consoles", + "intelligent", + "devices", + "related", + "accessories", + "cloud", + "based", + "solutions", + "provide", + "customers", + "services", + "platforms", + "content", + "Office", + "365", + "Azure", + "Dynamics", + "support", + "consulting", + "online", + "advertising", + "LinkedIn", + "Recognition", + "recognized", + "transfer", + "control", + "promised", + "products", + "amount", + "reflects", + "consideration", + "expect", + "receive", + "We", + "enter", + "contracts", + "include", + "combinations", + "generally", + "capable", + "distinct", + "accounted", + "separate", + "performance", + "obligations", + "net", + "allowances", + "returns", + "taxes", + "collected", + "subsequently", + "remitted", + "governmental", + "authorities", + "Nature", + "Products", + "Services", + "Licenses", + "premises", + "customer", + "exists", + "made", + "Customers", + "purchase", + "perpetual", + "licenses", + "subscribe", + "functionality", + "differ", + "duration", + "benefits", + "upfront", + "point", + "time", + "In", + "cases", + "allocate", + "updates", + "primarily", + "provided", + "additional", + "charge", + "ratably", + "estimated", + "life", + "license", + "Cloud", + "hosted", + "contract", + "period", + "taking", + "possession", + "subscription", + "consumption", + "basis", + "storage", + "utilization", + "resources", + "When", + "require", + "significant", + "level", + "integration", + "interdependency", + "individual", + "components", + "considered", + "Certain", + "volume", + "licensing", + "programs", + "including", + "Enterprise", + "Agreements", + "combined", + "Software", + "Assurance", + "SA", + "conveys", + "rights", + "upgrades", + "released", + "training", + "deploy", + "efficiently", + "On", + "sold", + "allocated", + "simultaneously", + "consume", + "comprises", + "satisfied", + "search", + "advertisement", + "appears", + "results", + "action", + "earn", + "completed", + "Our", + "highly", + "dependent", + "interrelated", + "underlying", + "system", + "function", + "single", + "obligation", + "ownership", + "transferred", + "resellers", + "directly", + "end", + "retail", + "stores", + "marketplaces" + ], + "keywords": [ + "Revenue", + "Foreign Currency", + "Performance Obligations" + ], + "connect_to": [] + } + }, + { + "chunk_id": "f4204034-85e7-5e6b-bf0e-14edcd8e72cb", + "type": "text", + "content": "Refer to Note 18 – Segment Information and Geographic Data for further information, including revenue by significant product and service offering.\nSignificant Judgments\nOur contracts with customers often include promises to transfer multiple products and services to a customer. Determining whether products and services are considered distinct performance obligations that should be accounted for separately versus together may require significant judgment. When a cloud-based service includes both on-premises software licenses and cloud services, judgment is required to determine whether the software license is considered distinct and accounted for separately, or not distinct and accounted for together with the cloud service and recognized over time. Certain cloud services, primarily Office 365, depend on a significant level of integration, interdependency, and interrelation between the desktop applications and cloud services, and are accounted for together as one performance obligation. Revenue from Office 365 is recognized ratably over the period in which the cloud services are provided.\nJudgment is required to determine the SSP for each distinct performance obligation. We use a single amount to estimate SSP for items that are not sold separately, including on-premises licenses sold with SA or software updates provided at no additional charge. We use a range of amounts to estimate SSP when we sell each of the products and services separately and need to determine whether there is a discount to be allocated based on the relative SSP of the various products and services.\nIn instances where SSP is not directly observable, such as when we do not sell the product or service separately, we determine the SSP using information that may include market conditions and other observable inputs. We typically have more than one SSP for individual products and services due to the stratification of those products and services by customers and circumstances. In these instances, we may use information such as the size of the customer and geographic region in determining the SSP.\nDue to the various benefits from and the nature of our SA program, judgment is required to assess the pattern of delivery, including the exercise pattern of certain benefits across our portfolio of customers.\nOur products are generally sold with a right of return, we may provide other credits or incentives, and in certain instances we estimate customer usage of our products and services, which are accounted for as variable consideration when determining the amount of revenue to recognize. Returns and credits are estimated at contract inception and updated at the end of each reporting period if additional information becomes available. Changes to our estimated variable consideration were not material for the periods presented.\nContract Balances and Other Receivables\nTiming of revenue recognition may differ from the timing of invoicing to customers. We record a receivable when revenue is recognized prior to invoicing, or unearned revenue when revenue is recognized subsequent to invoicing. For multi-year agreements, we generally invoice customers annually at the beginning of each annual coverage period. We record a receivable related to revenue recognized for multi-year on-premises licenses as we have an unconditional right to invoice and receive payment in the future related to those licenses.\nUnearned revenue comprises mainly unearned revenue related to volume licensing programs, which may include cloud services and SA. Unearned revenue is generally invoiced annually at the beginning of each contract period for multi-year agreements and recognized ratably over the coverage period. Unearned revenue also includes payments for consulting services to be performed in the future, LinkedIn subscriptions, Office 365 subscriptions, Xbox subscriptions, Windows post-delivery support, Dynamics business solutions, and other offerings for which we have been paid in advance and earn the revenue when we transfer control of the product or service.\nRefer to Note 12 – Unearned Revenue for further information, including unearned revenue by segment and changes in unearned revenue during the period.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 30", + "metadata": { + "length": 4209, + "summary": "The passage discusses significant judgments in revenue recognition for contracts with multiple products and services, including determining distinct performance obligations, estimating standalone selling prices (SSP), and accounting for variable consideration such as returns and credits. It also covers contract balances, unearned revenue, and timing differences between revenue recognition and invoicing, with examples like Office 365 and multi-year agreements.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Refer", + "Note", + "18", + "Segment", + "Information", + "Geographic", + "Data", + "information", + "including", + "revenue", + "significant", + "product", + "service", + "offering", + "Significant", + "Judgments", + "Our", + "contracts", + "customers", + "include", + "promises", + "transfer", + "multiple", + "products", + "services", + "customer", + "Determining", + "considered", + "distinct", + "performance", + "obligations", + "accounted", + "separately", + "versus", + "require", + "judgment", + "When", + "cloud", + "based", + "includes", + "premises", + "software", + "licenses", + "required", + "determine", + "license", + "recognized", + "time", + "Certain", + "primarily", + "Office", + "365", + "depend", + "level", + "integration", + "interdependency", + "interrelation", + "desktop", + "applications", + "obligation", + "Revenue", + "ratably", + "period", + "provided", + "Judgment", + "SSP", + "We", + "single", + "amount", + "estimate", + "items", + "sold", + "SA", + "updates", + "additional", + "charge", + "range", + "amounts", + "sell", + "discount", + "allocated", + "relative", + "In", + "instances", + "directly", + "observable", + "market", + "conditions", + "inputs", + "typically", + "individual", + "due", + "stratification", + "circumstances", + "size", + "geographic", + "region", + "determining", + "Due", + "benefits", + "nature", + "program", + "assess", + "pattern", + "delivery", + "exercise", + "portfolio", + "generally", + "return", + "provide", + "credits", + "incentives", + "usage", + "variable", + "consideration", + "recognize", + "Returns", + "estimated", + "contract", + "inception", + "updated", + "end", + "reporting", + "Changes", + "material", + "periods", + "presented", + "Contract", + "Balances", + "Other", + "Receivables", + "Timing", + "recognition", + "differ", + "timing", + "invoicing", + "record", + "receivable", + "prior", + "unearned", + "subsequent", + "For", + "multi", + "year", + "agreements", + "invoice", + "annually", + "beginning", + "annual", + "coverage", + "related", + "unconditional", + "receive", + "payment", + "future", + "Unearned", + "comprises", + "volume", + "licensing", + "programs", + "invoiced", + "payments", + "consulting", + "performed", + "LinkedIn", + "subscriptions", + "Xbox", + "Windows", + "post", + "support", + "Dynamics", + "business", + "solutions", + "offerings", + "paid", + "advance", + "earn", + "control", + "12", + "segment" + ], + "keywords": [ + "Revenue", + "Judgment", + "Contracts" + ], + "connect_to": [] + } + }, + { + "chunk_id": "83c2d546-19aa-519b-af22-416dc2c5c7db", + "type": "text", + "content": "Payment terms and conditions vary by contract type, although terms generally include a requirement of payment within 30 to 60 days. In instances where the timing of revenue recognition differs from the timing of invoicing, we have determined our contracts generally do not include a significant financing component. The primary purpose of our invoicing terms is to provide customers with simplified and predictable ways of purchasing our products and services, not to receive financing from our customers or to provide customers with financing. Examples include invoicing at the beginning of a subscription term with revenue recognized ratably over the contract period, and multi-year on-premises licenses that are invoiced annually with revenue recognized upfront.\nAs of June 30, 2025 and 2024, long-term accounts receivable, net of allowance for doubtful accounts, was $5.2 billion and $4.9 billion, respectively, and is included in other long-term assets in our consolidated balance sheets.\nAs of June 30, 2025 and 2024, other receivables related to activities to facilitate the purchase of server components were $8.2 billion and $10.5 billion, respectively, and are included in other current assets in our consolidated balance sheets.\nWe record financing receivables when we offer certain customers the option to acquire our software products and services offerings through a financing program in a limited number of countries. As of June 30, 2025 and 2024, our financing receivables, net were $4.3 billion and $4.5 billion, respectively, for short-term and long-term financing receivables, which are included in other current assets and other long-term assets in our consolidated balance sheets.\nWe record an allowance for doubtful accounts which reflects our best estimate of credit losses inherent in the accounts receivable and financing receivable balances. We determine the allowance based on known troubled accounts, historical experience, and other currently available evidence.\nAssets Recognized from Costs to Obtain a Contract with a Customer\nWe recognize an asset for the incremental costs of obtaining a contract with a customer if we expect the benefit of those costs to be longer than one year. We have determined that certain sales incentive programs meet the requirements to be capitalized. Total capitalized costs to obtain a contract were immaterial during the periods presented and are included in other current and long-term assets in our consolidated balance sheets.\nWe apply a practical expedient to expense costs as incurred for costs to obtain a contract with a customer when the amortization period would have been one year or less. These costs include our internal sales organization compensation program and certain partner sales incentive programs as we have determined annual compensation is commensurate with annual sales activities.\nCost of Revenue\nCost of revenue includes: costs incurred to support and maintain cloud-based and other online products and services, including datacenter costs and royalties; manufacturing and distribution costs for products sold and programs licensed; operating costs related to product support service centers and product distribution centers; traffic acquisition costs to drive traffic to our websites and to acquire online advertising space; and costs associated with the delivery of consulting services.\nResearch and Development\nResearch and development expenses include payroll, employee benefits, stock-based compensation expense, and other headcount-related expenses associated with product development. Research and development expenses also include third-party development and programming costs and the depreciation and amortization of assets used to conduct research and development. Such costs related to software development are included in research and development expense until the point that technological feasibility is reached, which for our software products is generally shortly before the products are released to production. Once technological feasibility is reached, such costs are capitalized and amortized to cost of revenue over the estimated lives of the products.\nSales and Marketing", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 31", + "metadata": { + "length": 4180, + "summary": "The passage discusses payment terms (30-60 days), revenue recognition timing, and financing receivables. It details long-term accounts receivable ($5.2B in 2025, $4.9B in 2024), other receivables for server components ($8.2B in 2025, $10.5B in 2024), and financing receivables ($4.3B in 2025, $4.5B in 2024). It also covers allowance for doubtful accounts, capitalized costs to obtain contracts, and cost categories including cost of revenue, research and development, and sales and marketing expenses.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Payment", + "terms", + "conditions", + "vary", + "contract", + "type", + "generally", + "include", + "requirement", + "payment", + "30", + "60", + "days", + "In", + "instances", + "timing", + "revenue", + "recognition", + "differs", + "invoicing", + "determined", + "contracts", + "significant", + "financing", + "component", + "The", + "primary", + "purpose", + "provide", + "customers", + "simplified", + "predictable", + "ways", + "purchasing", + "products", + "services", + "receive", + "Examples", + "beginning", + "subscription", + "term", + "recognized", + "ratably", + "period", + "multi", + "year", + "premises", + "licenses", + "invoiced", + "annually", + "upfront", + "As", + "June", + "2025", + "2024", + "long", + "accounts", + "receivable", + "net", + "allowance", + "doubtful", + "5.2", + "billion", + "4.9", + "included", + "assets", + "consolidated", + "balance", + "sheets", + "receivables", + "related", + "activities", + "facilitate", + "purchase", + "server", + "components", + "8.2", + "10.5", + "current", + "We", + "record", + "offer", + "option", + "acquire", + "software", + "offerings", + "program", + "limited", + "number", + "countries", + "4.3", + "4.5", + "short", + "reflects", + "estimate", + "credit", + "losses", + "inherent", + "balances", + "determine", + "based", + "troubled", + "historical", + "experience", + "evidence", + "Assets", + "Recognized", + "Costs", + "Obtain", + "Contract", + "Customer", + "recognize", + "asset", + "incremental", + "costs", + "obtaining", + "customer", + "expect", + "benefit", + "longer", + "sales", + "incentive", + "programs", + "meet", + "requirements", + "capitalized", + "Total", + "obtain", + "immaterial", + "periods", + "presented", + "apply", + "practical", + "expedient", + "expense", + "incurred", + "amortization", + "These", + "internal", + "organization", + "compensation", + "partner", + "annual", + "commensurate", + "Cost", + "Revenue", + "includes", + "support", + "maintain", + "cloud", + "online", + "including", + "datacenter", + "royalties", + "manufacturing", + "distribution", + "sold", + "licensed", + "operating", + "product", + "service", + "centers", + "traffic", + "acquisition", + "drive", + "websites", + "advertising", + "space", + "delivery", + "consulting", + "Research", + "Development", + "development", + "expenses", + "payroll", + "employee", + "benefits", + "stock", + "headcount", + "party", + "programming", + "depreciation", + "conduct", + "research", + "Such", + "point", + "technological", + "feasibility", + "reached", + "shortly", + "released", + "production", + "Once", + "amortized", + "cost", + "estimated", + "lives", + "Sales", + "Marketing" + ], + "keywords": [ + "Payment terms", + "Financing receivables", + "Revenue recognition" + ], + "connect_to": [] + } + }, + { + "chunk_id": "5cf2501d-86f3-5861-bf10-11f422a6942e", + "type": "text", + "content": "Sales and marketing expenses include payroll, employee benefits, stock-based compensation expense, and other headcount-related expenses associated with sales and marketing personnel, and the costs of advertising, promotions, trade shows, seminars, and other programs. Advertising costs are expensed as incurred. Advertising expense was $2.1 billion, $1.7 billion, and $904 million in fiscal years 2025, 2024, and 2023, respectively.\nStock-Based Compensation\nCompensation cost for stock awards, which include restricted stock units (“RSUs”) and performance stock units (“PSUs”), is measured at the fair value on the grant date and recognized as expense, net of estimated forfeitures, over the related service or performance period. The fair value of stock awards is based on the quoted price of our common stock on the grant date less the present value of expected dividends not received during the vesting period. We measure the fair value of PSUs using a Monte Carlo valuation model. Compensation cost for RSUs is recognized using the straight-line method and for PSUs is recognized using the accelerated method.\nCompensation expense for the employee stock purchase plan (“ESPP”) is measured as the discount the employee is entitled to upon purchase and is recognized in the period of purchase.\nIncome Taxes\nIncome tax expense includes U.S. and international income taxes, and interest and penalties on uncertain tax positions. Certain income and expenses are not reported in tax returns and financial statements in the same year. The tax effect of such temporary differences is reported as deferred income taxes. Deferred tax assets are reported net of a valuation allowance when it is more likely than not that a tax benefit will not be realized. All deferred income taxes are classified as long-term in our consolidated balance sheets.\nFinancial Instruments\nInvestments\nWe consider all highly liquid interest-earning investments with a maturity of three months or less at the date of purchase to be cash equivalents. The fair values of these investments approximate their carrying values. In general, investments with original maturities of greater than three months and remaining maturities of less than one year are classified as short-term investments. Investments with maturities beyond one year may be classified as short-term based on their highly liquid nature and because such marketable securities represent the investment of cash that is available for current operations.\nDebt investments are classified as available-for-sale and realized gains and losses are recorded using the specific identification method. Changes in fair value, excluding credit losses and impairments, are recorded in other comprehensive income. Fair value is calculated based on publicly available market information or other estimates determined by management. If the cost of an investment exceeds its fair value, we evaluate, among other factors, general market conditions, credit quality of debt instrument issuers, and the extent to which the fair value is less than cost. To determine credit losses, we employ a systematic methodology that considers available quantitative and qualitative evidence. In addition, we consider specific adverse conditions related to the financial health of, and business outlook for, the investee. If we have plans to sell the security or it is more likely than not that we will be required to sell the security before recovery, then a decline in fair value below cost is recorded as an impairment charge in other income (expense), net and a new cost basis in the investment is established. If market, industry, and/or investee conditions deteriorate, we may incur future impairments.\nEquity investments with readily determinable fair values are measured at fair value. Equity investments without readily determinable fair values are measured using the equity method or measured at cost with adjustments for observable changes in price or impairments (referred to as the measurement alternative). We perform a qualitative assessment on a periodic basis and recognize an impairment if there are sufficient indicators that the fair value of the investment is less than carrying value. Changes in value are recorded in other income (expense), net.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 32", + "metadata": { + "length": 4267, + "summary": "The passage details sales and marketing expenses including payroll, advertising, and promotions, with advertising costs of $2.1B, $1.7B, and $904M for fiscal years 2025, 2024, and 2023. It explains stock-based compensation for RSUs and PSUs, measured at fair value and recognized over service periods. Income taxes cover U.S. and international taxes, deferred taxes, and valuation allowances. Financial instruments section describes investments: cash equivalents, short-term investments, debt investments classified as available-for-sale with fair value changes in other comprehensive income, and equity investments measured at fair value or using the measurement alternative.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Sales", + "marketing", + "expenses", + "include", + "payroll", + "employee", + "benefits", + "stock", + "based", + "compensation", + "expense", + "headcount", + "related", + "sales", + "personnel", + "costs", + "advertising", + "promotions", + "trade", + "shows", + "seminars", + "programs", + "Advertising", + "expensed", + "incurred", + "2.1", + "billion", + "1.7", + "904", + "million", + "fiscal", + "years", + "2025", + "2024", + "2023", + "Stock", + "Based", + "Compensation", + "cost", + "awards", + "restricted", + "units", + "RSUs", + "performance", + "PSUs", + "measured", + "fair", + "grant", + "date", + "recognized", + "net", + "estimated", + "forfeitures", + "service", + "period", + "The", + "quoted", + "price", + "common", + "present", + "expected", + "dividends", + "received", + "vesting", + "We", + "measure", + "Monte", + "Carlo", + "valuation", + "model", + "straight", + "line", + "method", + "accelerated", + "purchase", + "plan", + "ESPP", + "discount", + "entitled", + "Income", + "Taxes", + "tax", + "includes", + "international", + "income", + "taxes", + "interest", + "penalties", + "uncertain", + "positions", + "Certain", + "reported", + "returns", + "financial", + "statements", + "year", + "effect", + "temporary", + "differences", + "deferred", + "Deferred", + "assets", + "allowance", + "benefit", + "realized", + "All", + "classified", + "long", + "term", + "consolidated", + "balance", + "sheets", + "Financial", + "Instruments", + "Investments", + "highly", + "liquid", + "earning", + "investments", + "maturity", + "months", + "cash", + "equivalents", + "values", + "approximate", + "carrying", + "In", + "general", + "original", + "maturities", + "greater", + "remaining", + "short", + "nature", + "marketable", + "securities", + "represent", + "investment", + "current", + "operations", + "Debt", + "sale", + "gains", + "losses", + "recorded", + "specific", + "identification", + "Changes", + "excluding", + "credit", + "impairments", + "comprehensive", + "Fair", + "calculated", + "publicly", + "market", + "information", + "estimates", + "determined", + "management", + "If", + "exceeds", + "evaluate", + "factors", + "conditions", + "quality", + "debt", + "instrument", + "issuers", + "extent", + "To", + "determine", + "employ", + "systematic", + "methodology", + "considers", + "quantitative", + "qualitative", + "evidence", + "addition", + "adverse", + "health", + "business", + "outlook", + "investee", + "plans", + "sell", + "security", + "required", + "recovery", + "decline", + "impairment", + "charge", + "basis", + "established", + "industry", + "deteriorate", + "incur", + "future", + "Equity", + "readily", + "determinable", + "equity", + "adjustments", + "observable", + "referred", + "measurement", + "alternative", + "perform", + "assessment", + "periodic", + "recognize", + "sufficient", + "indicators" + ], + "keywords": [ + "Sales and Marketing", + "Stock-Based Compensation", + "Income Taxes" + ], + "connect_to": [] + } + }, + { + "chunk_id": "55142c68-fbfd-5e5b-beb3-45fb775c66a7", + "type": "text", + "content": "Investments that are considered variable interest entities (“VIEs”) are evaluated to determine whether we are the primary beneficiary of the VIE, in which case we would be required to consolidate the entity. We evaluate whether we have (1) the power to direct the activities that most significantly impact the VIE’s economic performance, and (2) the obligation to absorb losses or the right to receive benefits from the VIE that could potentially be significant to the VIE. We have determined we are not the primary beneficiary of any of our VIE investments. Therefore, our VIE investments are not consolidated and the majority are accounted for under the equity method of accounting. We have an investment in OpenAI Global, LLC (“OpenAI”) and have made total funding commitments of $13 billion. The investment is accounted for under the equity method of accounting.\nDerivatives\nDerivative instruments are recognized as either assets or liabilities and measured at fair value. The accounting for changes in the fair value of a derivative depends on the intended use of the derivative and the resulting designation.\nFor derivative instruments designated as fair value hedges, gains and losses are recognized in other income (expense), net with offsetting gains and losses on the hedged items. Gains and losses representing hedge components excluded from the assessment of effectiveness are recognized in other income (expense), net.\nFor derivative instruments designated as cash flow hedges, gains and losses are initially reported as a component of other comprehensive income and subsequently recognized in other income (expense), net with the corresponding hedged item. Gains and losses representing hedge components excluded from the assessment of effectiveness are recognized in other income (expense), net.\nFor derivative instruments that are not designated as hedges, gains and losses from changes in fair values are primarily recognized in other income (expense), net.\nFair Value Measurements\nWe account for certain assets and liabilities at fair value. The hierarchy below lists three levels of fair value based on the extent to which inputs used in measuring fair value are observable in the market. We categorize each of our fair value measurements in one of these three levels based on the lowest level input that is significant to the fair value measurement in its entirety. These levels are:\n• Level 1 – inputs are based upon unadjusted quoted prices for identical instruments in active markets. Our Level 1 investments include U.S. government securities, common and preferred stock, and mutual funds. Our Level 1 derivative assets and liabilities include those actively traded on exchanges.\n• Level 2 – inputs are based upon quoted prices for similar instruments in active markets, quoted prices for identical or similar instruments in markets that are not active, and model-based valuation techniques (e.g. the Black-Scholes model) for which all significant inputs are observable in the market or can be corroborated by observable market data for substantially the full term of the assets or liabilities. Where applicable, these models project future cash flows and discount the future amounts to a present value using market-based observable inputs including interest rate curves, credit spreads, foreign exchange rates, and forward and spot prices for currencies. Our Level 2 investments include commercial paper, certificates of deposit, U.S. agency securities, foreign government bonds, mortgage- and asset-backed securities, corporate notes and bonds, and municipal securities. Our Level 2 derivative assets and liabilities include certain cleared swap contracts and over-the-counter forward, option, and swap contracts.\n• Level 3 – inputs are generally unobservable and typically reflect management’s estimates of assumptions that market participants would use in pricing the asset or liability. The fair values are therefore determined using model-based techniques, including option pricing models and discounted cash flow models. Our Level 3 assets and liabilities include investments in corporate notes and bonds, municipal securities, and goodwill and intangible assets, when they are recorded at fair value due to an impairment charge. Unobservable inputs used in the models are significant to the fair values of the assets and liabilities.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 33", + "metadata": { + "length": 4365, + "summary": "The passage discusses accounting treatments for variable interest entities (VIEs), derivatives, and fair value measurements. VIEs are evaluated for consolidation but none are consolidated; they are accounted for under the equity method, including a $13 billion investment in OpenAI. Derivatives are measured at fair value, with gains/losses recognized based on designation as fair value hedges, cash flow hedges, or non-designated instruments. Fair value measurements are categorized into three levels: Level 1 (quoted prices in active markets), Level 2 (observable inputs via models), and Level 3 (unobservable inputs using management estimates).", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Investments", + "considered", + "variable", + "interest", + "entities", + "VIEs", + "evaluated", + "determine", + "primary", + "beneficiary", + "VIE", + "case", + "required", + "consolidate", + "entity", + "We", + "evaluate", + "power", + "direct", + "activities", + "significantly", + "impact", + "economic", + "performance", + "obligation", + "absorb", + "losses", + "receive", + "benefits", + "potentially", + "significant", + "determined", + "investments", + "Therefore", + "consolidated", + "majority", + "accounted", + "equity", + "method", + "accounting", + "investment", + "OpenAI", + "Global", + "LLC", + "made", + "total", + "funding", + "commitments", + "13", + "billion", + "The", + "Derivatives", + "Derivative", + "instruments", + "recognized", + "assets", + "liabilities", + "measured", + "fair", + "derivative", + "depends", + "intended", + "resulting", + "designation", + "For", + "designated", + "hedges", + "gains", + "income", + "expense", + "net", + "offsetting", + "hedged", + "items", + "Gains", + "representing", + "hedge", + "components", + "excluded", + "assessment", + "effectiveness", + "cash", + "flow", + "initially", + "reported", + "component", + "comprehensive", + "subsequently", + "item", + "values", + "primarily", + "Fair", + "Value", + "Measurements", + "account", + "hierarchy", + "lists", + "levels", + "based", + "extent", + "inputs", + "measuring", + "observable", + "market", + "categorize", + "measurements", + "lowest", + "level", + "input", + "measurement", + "entirety", + "These", + "Level", + "unadjusted", + "quoted", + "prices", + "identical", + "active", + "markets", + "Our", + "include", + "government", + "securities", + "common", + "preferred", + "stock", + "mutual", + "funds", + "actively", + "traded", + "exchanges", + "similar", + "model", + "valuation", + "techniques", + "Black", + "Scholes", + "corroborated", + "data", + "substantially", + "full", + "term", + "Where", + "applicable", + "models", + "project", + "future", + "flows", + "discount", + "amounts", + "present", + "including", + "rate", + "curves", + "credit", + "spreads", + "foreign", + "exchange", + "rates", + "forward", + "spot", + "currencies", + "commercial", + "paper", + "certificates", + "deposit", + "agency", + "bonds", + "mortgage", + "asset", + "backed", + "corporate", + "notes", + "municipal", + "cleared", + "swap", + "contracts", + "counter", + "option", + "generally", + "unobservable", + "typically", + "reflect", + "management", + "estimates", + "assumptions", + "participants", + "pricing", + "liability", + "discounted", + "goodwill", + "intangible", + "recorded", + "due", + "impairment", + "charge", + "Unobservable" + ], + "keywords": [ + "VIE", + "Derivatives", + "Fair Value" + ], + "connect_to": [] + } + }, + { + "chunk_id": "9d006dbf-ff5d-5a86-a9b9-17b7730eddfd", + "type": "text", + "content": "We measure equity investments without readily determinable fair values on a nonrecurring basis. The fair values of these investments are determined based on valuation techniques using the best information available, and may include quoted market prices, market comparables, and discounted cash flow projections.\nOur other current financial assets and current financial liabilities have fair values that approximate their carrying values.\nProperty and Equipment\nProperty and equipment is stated at cost less accumulated depreciation and depreciated using the straight-line method over the shorter of the estimated useful life of the asset or the lease term. The estimated useful lives of our property and equipment are generally as follows: software developed or acquired for internal use, three years; computer equipment, two to six years; buildings and improvements, five to 15 years; leasehold improvements, three to 15 years; and furniture and equipment, one to 10 years. Land is not depreciated.\nLeases\nWe determine if an arrangement is a lease at inception. Operating leases are included in operating lease right-of-use (“ROU”) assets, other current liabilities, and operating lease liabilities in our consolidated balance sheets. Finance leases are included in property and equipment, other current liabilities, and other long-term liabilities in our consolidated balance sheets.\nROU assets represent our right to use an underlying asset for the lease term and lease liabilities represent our obligation to make lease payments arising from the lease. Operating lease ROU assets and liabilities are recognized at commencement date based on the present value of lease payments over the lease term. As most of our leases do not provide an implicit rate, we generally use our incremental borrowing rate based on the estimated rate of interest for collateralized borrowing over a similar term of the lease payments at commencement date. The operating lease ROU asset also includes any lease payments made and excludes lease incentives. Our lease terms may include options to extend or terminate the lease when it is reasonably certain that we will exercise that option. Lease expense for lease payments is recognized on a straight-line basis over the lease term.\nWe have lease agreements with lease and non-lease components, which are generally accounted for separately. For certain equipment leases, such as vehicles, we account for the lease and non-lease components as a single lease component. Additionally, for certain equipment leases, we apply a portfolio approach to effectively account for the operating lease ROU assets and liabilities.\nGoodwill\nGoodwill is tested for impairment at the reporting unit level (operating segment or one level below an operating segment) on an annual basis (May 1) and between annual tests if an event occurs or circumstances change that would more likely than not reduce the fair value of a reporting unit below its carrying value.\nIntangible Assets\nOur intangible assets are subject to amortization and are amortized over the estimated useful life in proportion to the economic benefits received. We evaluate the recoverability of intangible assets periodically by taking into account events or circumstances that may warrant revised estimates of useful lives or that indicate the asset may be impaired.\nRelated Party Transactions\nIn March 2024, we entered into an agreement with Inflection AI, Inc. (“Inflection”), pursuant to which we obtained a non-exclusive license to Inflection’s intellectual property. Reid Hoffman, a member of our Board of Directors, is a co-founder of and serves on the board of directors of Inflection. As of the date of the agreement with Inflection, Reprogrammed Interchange LLC (“Reprogrammed”) and entities affiliated with Greylock Ventures (“Greylock”) each held less than a 10% equity interest in Inflection. Mr. Hoffman may be deemed to beneficially own the shares held by Reprogrammed and Greylock by virtue of his relationship with such entities. Mr. Hoffman did not participate in any portions of the meetings of our Board of Directors or any committee thereof to review and approve the transaction with Inflection.\nRecent Accounting Guidance\nRecently Adopted Accounting Guidance\nSegment Reporting – Improvements to Reportable Segment Disclosures", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 34", + "metadata": { + "length": 4326, + "summary": "The passage details accounting policies for equity investments, property and equipment, leases, goodwill, intangible assets, and related party transactions. It explains valuation methods, depreciation, lease recognition, impairment testing, and a recent agreement with Inflection AI.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "We", + "measure", + "equity", + "investments", + "readily", + "determinable", + "fair", + "values", + "nonrecurring", + "basis", + "The", + "determined", + "based", + "valuation", + "techniques", + "information", + "include", + "quoted", + "market", + "prices", + "comparables", + "discounted", + "cash", + "flow", + "projections", + "Our", + "current", + "financial", + "assets", + "liabilities", + "approximate", + "carrying", + "Property", + "Equipment", + "equipment", + "stated", + "cost", + "accumulated", + "depreciation", + "depreciated", + "straight", + "line", + "method", + "shorter", + "estimated", + "life", + "asset", + "lease", + "term", + "lives", + "property", + "generally", + "software", + "developed", + "acquired", + "internal", + "years", + "computer", + "buildings", + "improvements", + "15", + "leasehold", + "furniture", + "10", + "Land", + "Leases", + "determine", + "arrangement", + "inception", + "Operating", + "leases", + "included", + "operating", + "ROU", + "consolidated", + "balance", + "sheets", + "Finance", + "long", + "represent", + "underlying", + "obligation", + "make", + "payments", + "arising", + "recognized", + "commencement", + "date", + "present", + "As", + "provide", + "implicit", + "rate", + "incremental", + "borrowing", + "interest", + "collateralized", + "similar", + "includes", + "made", + "excludes", + "incentives", + "terms", + "options", + "extend", + "terminate", + "exercise", + "option", + "Lease", + "expense", + "agreements", + "components", + "accounted", + "separately", + "For", + "vehicles", + "account", + "single", + "component", + "Additionally", + "apply", + "portfolio", + "approach", + "effectively", + "Goodwill", + "tested", + "impairment", + "reporting", + "unit", + "level", + "segment", + "annual", + "May", + "tests", + "event", + "occurs", + "circumstances", + "change", + "reduce", + "Intangible", + "Assets", + "intangible", + "subject", + "amortization", + "amortized", + "proportion", + "economic", + "benefits", + "received", + "evaluate", + "recoverability", + "periodically", + "taking", + "events", + "warrant", + "revised", + "estimates", + "impaired", + "Related", + "Party", + "Transactions", + "In", + "March", + "2024", + "entered", + "agreement", + "Inflection", + "AI", + "Inc", + "pursuant", + "obtained", + "exclusive", + "license", + "intellectual", + "Reid", + "Hoffman", + "member", + "Board", + "Directors", + "founder", + "serves", + "board", + "directors", + "Reprogrammed", + "Interchange", + "LLC", + "entities", + "affiliated", + "Greylock", + "Ventures", + "held", + "10%", + "Mr", + "deemed", + "beneficially", + "shares", + "virtue", + "relationship", + "participate", + "portions", + "meetings", + "committee", + "thereof", + "review", + "approve", + "transaction", + "Recent", + "Accounting", + "Guidance", + "Recently", + "Adopted", + "Segment", + "Reporting", + "Improvements", + "Reportable", + "Disclosures" + ], + "keywords": [ + "Fair Value", + "Leases", + "Goodwill" + ], + "connect_to": [] + } + }, + { + "chunk_id": "869a398d-20d1-54d6-be56-575490545590", + "type": "text", + "content": "In November 2023, the Financial Accounting Standards Board (“FASB”) issued a new standard to improve reportable segment disclosures. The guidance expands the disclosures required for reportable segments in our annual and interim consolidated financial statements, primarily through enhanced disclosures about significant segment expenses. We adopted the standard beginning with our annual reporting for fiscal year 2025. The adoption resulted in incremental segment reporting disclosures, most notably disclosure of cost of revenue and operating expenses for each reportable segment. Refer to Note 18 – Segment Information and Geographic Data.\nRecent Accounting Guidance Not Yet Adopted\nIncome Taxes – Improvements to Income Tax Disclosures\nIn December 2023, the FASB issued a new standard to improve income tax disclosures. The guidance requires disclosure of disaggregated income taxes paid, prescribes standardized categories for the components of the effective tax rate reconciliation, and modifies other income tax-related disclosures. The standard will be effective for us beginning with our annual reporting for fiscal year 2026, with early adoption permitted. We are currently evaluating the impact of this standard on our income tax disclosures.\nIncome Statement – Disaggregation of Income Statement Expenses\nIn November 2024, the FASB issued a new standard to expand disclosures about income statement expenses. The guidance requires disaggregation of certain costs and expenses included in each relevant expense caption on our consolidated income statements in a separate note to the financial statements at each interim and annual reporting period, including amounts of purchases of inventory, employee compensation, depreciation, and intangible asset amortization. The standard will be effective for us beginning with our annual reporting for fiscal year 2028 and interim periods thereafter, with early adoption permitted. We are currently evaluating the impact of this standard on our disclosures.\nNOTE 2 — EARNINGS PER SHARE\nBasic earnings per share (“EPS”) is computed based on the weighted average number of shares of common stock outstanding during the period. Diluted EPS is computed based on the weighted average number of shares of common stock plus the effect of dilutive potential common shares outstanding during the period using the treasury stock method. Dilutive potential common shares include outstanding stock options and stock awards.\nThe components of basic and diluted EPS were as follows:\n\nThe table shows net income, weighted average shares, and basic/diluted EPS for fiscal years 2023-2025. Net income rose from $72,361M to $101,832M, while diluted EPS increased from $9.68 to $13.64.\n[tables/table-21 EPS Data Table.html]\n\nAnti-dilutive stock-based awards excluded from the calculations of diluted EPS were immaterial during the periods presented.\nNOTE 3 — OTHER INCOME (EXPENSE), NET\nThe components of other income (expense), net were as follows:\n\nThe table shows interest and dividends income, interest expense, net gains/losses on investments, derivatives, foreign currency, and other items for fiscal years 2025, 2024, and 2023, with totals of -$4,901M, -$1,646M, and $788M respectively.\n[tables/table-22 Investment Income.html]\n\nOther, net primarily reflects net recognized losses on equity method investments, including OpenAI.\nNet Recognized Gains (Losses) on Investments\nNet recognized gains (losses) on debt investments were as follows:\n\nThe table shows realized gains and losses from sales of available-for-sale securities and impairments for fiscal years 2025, 2024, and 2023, with total net losses of $17M, $53M, and $98M respectively.\n[tables/table-23 Investment Gains.html]\n\nNet recognized gains (losses) on equity investments were as follows:\n\nThe table shows net realized gains, unrealized gains, and impairments for fiscal years 2025, 2024, and 2023, with totals of -332, -65, and 358 million respectively.\n[tables/table-24 Investment Gains.html]\n\nNOTE 4 — INVESTMENTS\nInvestment Components\nThe components of investments were as follows:\n\nThe table details investments as of June 30, 2025, showing adjusted cost basis, unrealized gains/losses, recorded basis, and classification into cash, short-term, equity, and other investments, totaling $109,970 million.\n[tables/table-25 Investment Table.html]", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 35", + "metadata": { + "length": 4353, + "summary": "The passage discusses recent FASB accounting standards adopted or pending adoption, including improved segment disclosures (effective fiscal 2025), income tax disclosures (effective fiscal 2026), and disaggregation of income statement expenses (effective fiscal 2028). It also covers Notes 2-4: Earnings Per Share (basic and diluted EPS calculations with net income rising from $72,361M to $101,832M and diluted EPS from $9.68 to $13.64), Other Income/Expense (net totals of -$4,901M, -$1,646M, and $788M for fiscal years 2025, 2024, 2023, including losses on equity method investments like OpenAI), and Investments (total $109,970 million as of June 30, 2025, with components including adjusted cost basis, unrealized gains/losses, and classifications).", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "In", + "November", + "2023", + "Financial", + "Accounting", + "Standards", + "Board", + "FASB", + "issued", + "standard", + "improve", + "reportable", + "segment", + "disclosures", + "The", + "guidance", + "expands", + "required", + "segments", + "annual", + "interim", + "consolidated", + "financial", + "statements", + "primarily", + "enhanced", + "significant", + "expenses", + "We", + "adopted", + "beginning", + "reporting", + "fiscal", + "year", + "2025", + "adoption", + "resulted", + "incremental", + "notably", + "disclosure", + "cost", + "revenue", + "operating", + "Refer", + "Note", + "18", + "Segment", + "Information", + "Geographic", + "Data", + "Recent", + "Guidance", + "Not", + "Yet", + "Adopted", + "Income", + "Taxes", + "Improvements", + "Tax", + "Disclosures", + "December", + "income", + "tax", + "requires", + "disaggregated", + "taxes", + "paid", + "prescribes", + "standardized", + "categories", + "components", + "effective", + "rate", + "reconciliation", + "modifies", + "related", + "2026", + "early", + "permitted", + "evaluating", + "impact", + "Statement", + "Disaggregation", + "Expenses", + "2024", + "expand", + "statement", + "disaggregation", + "costs", + "included", + "relevant", + "expense", + "caption", + "separate", + "note", + "period", + "including", + "amounts", + "purchases", + "inventory", + "employee", + "compensation", + "depreciation", + "intangible", + "asset", + "amortization", + "2028", + "periods", + "NOTE", + "EARNINGS", + "PER", + "SHARE", + "Basic", + "earnings", + "share", + "EPS", + "computed", + "based", + "weighted", + "average", + "number", + "shares", + "common", + "stock", + "outstanding", + "Diluted", + "effect", + "dilutive", + "potential", + "treasury", + "method", + "Dilutive", + "include", + "options", + "awards", + "basic", + "diluted", + "table", + "shows", + "net", + "years", + "Net", + "rose", + "72", + "361M", + "101", + "832M", + "increased", + "9.68", + "13.64", + "Anti", + "excluded", + "calculations", + "immaterial", + "presented", + "OTHER", + "INCOME", + "EXPENSE", + "NET", + "interest", + "dividends", + "gains", + "losses", + "investments", + "derivatives", + "foreign", + "currency", + "items", + "totals", + "901M", + "646M", + "788M", + "Other", + "reflects", + "recognized", + "equity", + "OpenAI", + "Recognized", + "Gains", + "Losses", + "Investments", + "debt", + "realized", + "sales", + "sale", + "securities", + "impairments", + "total", + "17M", + "53M", + "98M", + "unrealized", + "332", + "65", + "358", + "million", + "INVESTMENTS", + "Investment", + "Components", + "details", + "June", + "30", + "showing", + "adjusted", + "basis", + "recorded", + "classification", + "cash", + "short", + "term", + "totaling", + "109", + "970" + ], + "keywords": [ + "FASB", + "Segment Disclosures", + "EPS" + ], + "connect_to": [ + { + "target": "d7c618ea-146e-534a-ba94-d4a1ee5df5b6", + "relation": "embeds", + "ref": "[tables/table-21 EPS Data Table.html]", + "position": { + "start": 2721, + "end": 2758 + } + }, + { + "target": "8f0ab224-deaf-565e-89b5-1642ad95e14e", + "relation": "embeds", + "ref": "[tables/table-22 Investment Income.html]", + "position": { + "start": 3229, + "end": 3269 + } + }, + { + "target": "577e18e7-d331-50fc-9972-61a95fd1d6bc", + "relation": "embeds", + "ref": "[tables/table-23 Investment Gains.html]", + "position": { + "start": 3685, + "end": 3724 + } + }, + { + "target": "56ca9fcb-4e77-5b94-af9d-0e6f3a01dde4", + "relation": "embeds", + "ref": "[tables/table-24 Investment Gains.html]", + "position": { + "start": 3961, + "end": 4000 + } + }, + { + "target": "09f4dfcd-8f5a-502c-a2ce-c81f1c7502b6", + "relation": "embeds", + "ref": "[tables/table-25 Investment Table.html]", + "position": { + "start": 4314, + "end": 4353 + } + } + ] + } + }, + { + "chunk_id": "48e3d269-d22d-59fe-bfb9-93721c15cb09", + "type": "text", + "content": "Table details investments as of June 30, 2024, with adjusted cost basis, unrealized gains/losses, and recorded basis across debt and equity categories, totaling $90,143 million.\n[tables/table-26 Investment Summary.html]\n\n(a) Refer to Note 5 – Derivatives for further information on the fair value of our derivative instruments.\nEquity investments presented as “Other” in the tables above include investments without readily determinable fair values measured at cost with adjustments for observable changes in price or impairments, measured using the equity method, or measured at fair value using net asset value as a practical expedient which are not categorized in the fair value hierarchy. As of June 30, 2025 and 2024, equity investments without readily determinable fair values measured at cost with adjustments for observable changes in price or impairments were $2.9 billion and $3.9 billion, respectively. Equity investments measured using the equity method were $6.0 billion as of both June 30, 2025 and 2024.\nUnrealized Losses on Debt Investments\nDebt investments with continuous unrealized losses for less than 12 months and 12 months or greater and their related fair values were as follows:\n\nThe table shows fair values and unrealized losses for securities held less than 12 months and 12 months or greater as of June 30, 2025 and 2024. Total fair value decreased from $54,589M to $42,736M, while total unrealized losses decreased from $(3,348)M to $(1,625)M.\n[tables/table-27 Unrealized Losses.html]\n\nUnrealized losses from fixed-income securities are primarily attributable to changes in interest rates. Management does not believe any remaining unrealized losses represent impairments based on our evaluation of available evidence.\nDebt Investment Maturities\nThe following table outlines maturities of our debt investments as of June 30, 2025:\n\nThe table shows adjusted cost basis and estimated fair value of debt securities by maturity as of June 30, 2025. Total cost basis is $85,589 million, with fair value at $84,259 million.\n[tables/table-28 Debt Securities.html]\n\nNOTE 5 — DERIVATIVES\nWe use derivative instruments to manage risks related to foreign currencies, interest rates, equity prices, and credit; to enhance investment returns; and to facilitate portfolio diversification. Our objectives for holding derivatives include reducing, eliminating, and efficiently managing the economic impact of these exposures as effectively as possible. Our derivative programs include strategies that both qualify and do not qualify for hedge accounting treatment.\nForeign Currencies\nCertain forecasted transactions, assets, and liabilities are exposed to foreign currency risk. We monitor our foreign currency exposures daily to maximize the economic effectiveness of our foreign currency hedge positions.\nForeign currency risks related to certain Euro-denominated debt are hedged using foreign exchange forward contracts that are designated as cash flow hedging instruments.\nCertain options and forwards not designated as hedging instruments are also used to manage the variability in foreign exchange rates on certain balance sheet amounts and to manage other foreign currency exposures.\nInterest Rate\nInterest rate risks related to certain fixed-rate debt are hedged using interest rate swaps that are designated as fair value hedging instruments to effectively convert the fixed interest rates to floating interest rates.\nSecurities held in our fixed-income portfolio are subject to different interest rate risks based on their maturities. We manage the average maturity of our fixed-income portfolio to achieve economic returns that correlate to certain broad-based fixed-income indices using option, futures, and swap contracts. These contracts are not designated as hedging instruments and are included in “Other contracts” in the tables below.\nEquity\nSecurities held in our equity investments portfolio are subject to market price risk. At times, we may hold options, futures, and swap contracts. These contracts are not designated as hedging instruments.\nCredit\nOur fixed-income portfolio is diversified and consists primarily of investment-grade securities. We use credit default swap contracts to manage credit exposures relative to broad-based indices and to facilitate portfolio diversification. These contracts are not designated as hedging instruments and are included in “Other contracts” in the tables below.\nCredit-Risk-Related Contingent Features", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 36", + "metadata": { + "length": 4479, + "summary": "The passage details investments as of June 30, 2024, with adjusted cost basis, unrealized gains/losses, and recorded basis totaling $90,143 million. It discusses equity investments without readily determinable fair values, equity method investments, and unrealized losses on debt investments, noting a decrease in total fair value from $54,589M to $42,736M and unrealized losses from $(3,348)M to $(1,625)M. Debt investment maturities as of June 30, 2025 show total cost basis of $85,589 million and fair value of $84,259 million. Note 5 covers derivatives used to manage foreign currency, interest rate, equity, and credit risks, including hedging instruments and non-designated contracts.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Table", + "details", + "investments", + "June", + "30", + "2024", + "adjusted", + "cost", + "basis", + "unrealized", + "gains", + "losses", + "recorded", + "debt", + "equity", + "categories", + "totaling", + "90", + "143", + "million", + "Refer", + "Note", + "Derivatives", + "information", + "fair", + "derivative", + "instruments", + "Equity", + "presented", + "Other", + "tables", + "include", + "readily", + "determinable", + "values", + "measured", + "adjustments", + "observable", + "price", + "impairments", + "method", + "net", + "asset", + "practical", + "expedient", + "categorized", + "hierarchy", + "As", + "2025", + "2.9", + "billion", + "3.9", + "6.0", + "Unrealized", + "Losses", + "Debt", + "Investments", + "continuous", + "12", + "months", + "greater", + "related", + "The", + "table", + "shows", + "securities", + "held", + "Total", + "decreased", + "54", + "589M", + "42", + "736M", + "total", + "348", + "625", + "fixed", + "income", + "primarily", + "attributable", + "interest", + "rates", + "Management", + "remaining", + "represent", + "based", + "evaluation", + "evidence", + "Investment", + "Maturities", + "outlines", + "maturities", + "estimated", + "maturity", + "85", + "589", + "84", + "259", + "NOTE", + "DERIVATIVES", + "We", + "manage", + "risks", + "foreign", + "currencies", + "prices", + "credit", + "enhance", + "investment", + "returns", + "facilitate", + "portfolio", + "diversification", + "Our", + "objectives", + "holding", + "derivatives", + "reducing", + "eliminating", + "efficiently", + "managing", + "economic", + "impact", + "exposures", + "effectively", + "programs", + "strategies", + "qualify", + "hedge", + "accounting", + "treatment", + "Foreign", + "Currencies", + "Certain", + "forecasted", + "transactions", + "assets", + "liabilities", + "exposed", + "currency", + "risk", + "monitor", + "daily", + "maximize", + "effectiveness", + "positions", + "Euro", + "denominated", + "hedged", + "exchange", + "forward", + "contracts", + "designated", + "cash", + "flow", + "hedging", + "options", + "forwards", + "variability", + "balance", + "sheet", + "amounts", + "Interest", + "Rate", + "rate", + "swaps", + "convert", + "floating", + "Securities", + "subject", + "average", + "achieve", + "correlate", + "broad", + "indices", + "option", + "futures", + "swap", + "These", + "included", + "market", + "At", + "times", + "hold", + "Credit", + "diversified", + "consists", + "grade", + "default", + "relative", + "Risk", + "Related", + "Contingent", + "Features" + ], + "keywords": [ + "Investments", + "Derivatives", + "Unrealized Losses" + ], + "connect_to": [ + { + "target": "4a3c2966-3788-501d-a4bb-63808be3ee18", + "relation": "embeds", + "ref": "[tables/table-26 Investment Summary.html]", + "position": { + "start": 178, + "end": 219 + } + }, + { + "target": "1539d95f-5366-5c5a-98e1-23db91787a71", + "relation": "embeds", + "ref": "[tables/table-27 Unrealized Losses.html]", + "position": { + "start": 1473, + "end": 1513 + } + }, + { + "target": "eba625c6-80bb-5830-8fb7-4881edfb68a6", + "relation": "embeds", + "ref": "[tables/table-28 Debt Securities.html]", + "position": { + "start": 2047, + "end": 2085 + } + } + ] + } + }, + { + "chunk_id": "9a4c2eaf-8bf6-5f7d-a02f-a3f8e7a7649f", + "type": "text", + "content": "Certain counterparty agreements for derivative instruments contain provisions that require our issued and outstanding long-term unsecured debt to maintain an investment grade credit rating and require us to maintain minimum liquidity of $1.0 billion. To the extent we fail to meet these requirements, we will be required to post collateral, similar to the standard convention related to over-the-counter derivatives. As of June 30, 2025, our long-term unsecured debt rating was AAA, and cash investments were in excess of $1.0 billion. As a result, no collateral was required to be posted.\nThe following table presents the notional amounts of our outstanding derivative instruments measured in U.S. dollar equivalents:\n\nThe table shows notional amounts of derivative instruments as of June 30, 2025 and 2024, categorized by hedging designation and contract type, with foreign exchange contracts being the largest.\n[tables/table-29 Derivative Notional.html]\n\nFair Values of Derivative Instruments\nThe following table presents our derivative instruments:\n\nThe table shows derivative assets and liabilities as of June 30, 2025 and 2024, with gross amounts of $758M and $331M in assets, and $(1,837M) and $(800M) in liabilities, offset by collateral.\n[tables/table-30 Derivative Assets.html]\n\nGross derivative assets and liabilities subject to legally enforceable master netting agreements for which we have elected to offset were $452 million and $1.8 billion, respectively, as of June 30, 2025, and $304 million and $800 million, respectively, as of June 30, 2024.\nThe following table presents the fair value of our derivatives instruments on a gross basis:\n\nThe table shows derivative assets and liabilities by fair value hierarchy (Level 1, 2, 3) as of June 30, 2025 and 2024. Total derivative assets were $758M (2025) and $331M (2024); liabilities were $(1,837M) and $(800M) respectively.\n[tables/table-31 Derivatives by Level.html]\n\nGains (losses) on derivative instruments recognized in other income (expense), net were as follows:\n\nThe table shows gains and losses from interest rate, foreign exchange, equity, and other contracts designated as fair value, cash flow, or non-designated hedging instruments for fiscal years 2023-2025.\n[tables/table-32 Hedging Gains_Losses.html]\n\nGains (losses), net of tax, on derivative instruments recognized in our consolidated comprehensive income statements were as follows:\n\nThe table shows gains (losses) on foreign exchange contracts designated as cash flow hedging instruments: $77M in 2025, $(14)M in 2024, and $34M in 2023.\n[tables/table-33 Cash Flow Hedges.html]\n\nNOTE 6 — PROPERTY AND EQUIPMENT\nThe components of property and equipment were as follows:\n\nThe table shows property assets at cost of $298,619M (2025) vs $212,012M (2024), with accumulated depreciation of $93,653M and $76,421M, resulting in net values of $204,966M and $135,591M.\n[tables/table-34 Property Assets.html]\n\nDuring fiscal years 2025, 2024, and 2023, depreciation expense was $22.0 billion, $15.2 billion, and $11.0 billion, respectively.\nAs of June 30, 2025, 2024, and 2023, purchases of property and equipment remaining in accounts payable were $6.9 billion, $4.3 billion, and $3.8 billion, respectively. As of June 30, 2025, we have committed $32.1 billion for the construction of new buildings, building improvements, and leasehold improvements, primarily related to datacenters.\nNOTE 7 — BUSINESS COMBINATIONS\nActivision Blizzard, Inc.\nOn October 13, 2023, we completed our acquisition of Activision Blizzard, Inc. (“Activision Blizzard”) for a total purchase price of $75.4 billion, consisting primarily of cash. Activision Blizzard is a leader in game development and an interactive entertainment content publisher. The acquisition will accelerate the growth in our gaming business across mobile, PC, console, and cloud gaming. The financial results of Activision Blizzard have been included in our consolidated financial statements since the date of the acquisition. Activision Blizzard is reported as part of our More Personal Computing segment.\nThe allocation of the purchase price to the assets acquired and liabilities assumed was completed as of September 30, 2024. The major classes of assets and liabilities to which we have allocated the purchase price were as follows:\n\nThe table shows a total purchase price of $75,408 million, comprising cash, goodwill, intangible assets, other assets, and various liabilities.\n[tables/table-35 Purchase Price.html]", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 37", + "metadata": { + "length": 4492, + "summary": "The passage covers derivative instruments, including counterparty agreements requiring investment-grade credit rating and $1.0B liquidity, with no collateral posted as of June 30, 2025. Notional amounts and fair values of derivatives are detailed, with gross assets of $758M and liabilities of $(1,837M) in 2025, offset by collateral. Gains and losses from hedging instruments are reported for fiscal years 2023-2025. Property and equipment are summarized with net values of $204,966M (2025) and $135,591M (2024), depreciation expenses, and $32.1B committed for construction. The acquisition of Activision Blizzard for $75.4B in cash on October 13, 2023, is noted, with purchase price allocation completed.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Certain", + "counterparty", + "agreements", + "derivative", + "instruments", + "provisions", + "require", + "issued", + "outstanding", + "long", + "term", + "unsecured", + "debt", + "maintain", + "investment", + "grade", + "credit", + "rating", + "minimum", + "liquidity", + "1.0", + "billion", + "To", + "extent", + "fail", + "meet", + "requirements", + "required", + "post", + "collateral", + "similar", + "standard", + "convention", + "related", + "counter", + "derivatives", + "As", + "June", + "30", + "2025", + "AAA", + "cash", + "investments", + "excess", + "result", + "posted", + "The", + "table", + "presents", + "notional", + "amounts", + "measured", + "dollar", + "equivalents", + "shows", + "2024", + "categorized", + "hedging", + "designation", + "contract", + "type", + "foreign", + "exchange", + "contracts", + "largest", + "Fair", + "Values", + "Derivative", + "Instruments", + "assets", + "liabilities", + "gross", + "758M", + "331M", + "837M", + "800M", + "offset", + "Gross", + "subject", + "legally", + "enforceable", + "master", + "netting", + "elected", + "452", + "million", + "1.8", + "304", + "800", + "fair", + "basis", + "hierarchy", + "Level", + "Total", + "Gains", + "losses", + "recognized", + "income", + "expense", + "net", + "gains", + "interest", + "rate", + "equity", + "designated", + "flow", + "fiscal", + "years", + "2023", + "tax", + "consolidated", + "comprehensive", + "statements", + "77M", + "14", + "34M", + "NOTE", + "PROPERTY", + "AND", + "EQUIPMENT", + "components", + "property", + "equipment", + "cost", + "298", + "619M", + "212", + "012M", + "accumulated", + "depreciation", + "93", + "653M", + "76", + "421M", + "resulting", + "values", + "204", + "966M", + "135", + "591M", + "During", + "22.0", + "15.2", + "11.0", + "purchases", + "remaining", + "accounts", + "payable", + "6.9", + "4.3", + "3.8", + "committed", + "32.1", + "construction", + "buildings", + "building", + "improvements", + "leasehold", + "primarily", + "datacenters", + "BUSINESS", + "COMBINATIONS", + "Activision", + "Blizzard", + "Inc", + "On", + "October", + "13", + "completed", + "acquisition", + "total", + "purchase", + "price", + "75.4", + "consisting", + "leader", + "game", + "development", + "interactive", + "entertainment", + "content", + "publisher", + "accelerate", + "growth", + "gaming", + "business", + "mobile", + "PC", + "console", + "cloud", + "financial", + "results", + "included", + "date", + "reported", + "part", + "More", + "Personal", + "Computing", + "segment", + "allocation", + "acquired", + "assumed", + "September", + "major", + "classes", + "allocated", + "75", + "408", + "comprising", + "goodwill", + "intangible" + ], + "keywords": [ + "Derivatives", + "Property", + "Business Combination" + ], + "connect_to": [ + { + "target": "8474c325-a28e-5fe3-9244-2d9d66a680c8", + "relation": "embeds", + "ref": "[tables/table-29 Derivative Notional.html]", + "position": { + "start": 914, + "end": 956 + } + }, + { + "target": "d7f38500-41ab-5488-951e-0a6105cf64e1", + "relation": "embeds", + "ref": "[tables/table-30 Derivative Assets.html]", + "position": { + "start": 1247, + "end": 1287 + } + }, + { + "target": "9b401ce0-62b5-57f2-81a9-f1ae471ae44e", + "relation": "embeds", + "ref": "[tables/table-31 Derivatives by Level.html]", + "position": { + "start": 1890, + "end": 1933 + } + }, + { + "target": "774362b2-4d34-5748-b999-a72bfbf1e1ee", + "relation": "embeds", + "ref": "[tables/table-32 Hedging Gains_Losses.html]", + "position": { + "start": 2238, + "end": 2281 + } + }, + { + "target": "4911001d-4060-58e9-815b-96f22f51545b", + "relation": "embeds", + "ref": "[tables/table-33 Cash Flow Hedges.html]", + "position": { + "start": 2572, + "end": 2611 + } + }, + { + "target": "67202e30-5eeb-5d05-a101-99d342e69724", + "relation": "embeds", + "ref": "[tables/table-34 Property Assets.html]", + "position": { + "start": 2893, + "end": 2931 + } + }, + { + "target": "9a5bfc83-30b1-5018-8ff7-41a738c3dc90", + "relation": "embeds", + "ref": "[tables/table-35 Purchase Price.html]", + "position": { + "start": 4455, + "end": 4492 + } + } + ] + } + }, + { + "chunk_id": "075946eb-7d2d-5acd-b756-f4f9935a0e6e", + "type": "text", + "content": "Goodwill was assigned to our More Personal Computing segment. The goodwill was primarily attributed to increased synergies that are expected to be achieved from the integration of Activision Blizzard. Substantially all of the goodwill is expected to be non-deductible for income tax purposes.\nFollowing are the details of the purchase price allocated to the intangible assets acquired:\n\nThe table shows fair value of acquired intangible assets totaling $21,969 million with a weighted average life of 15 years, broken down into marketing-related ($11,619M, 24 years), technology-based ($9,689M, 4 years), and customer-related ($661M, 4 years).\n[tables/table-36 Intangible Assets.html]\n\nFollowing are the supplemental consolidated financial results of Microsoft Corporation on an unaudited pro forma basis, as if the acquisition had been consummated on July 1, 2022:\n\nMicrosoft's fiscal year ended June 30, 2024, shows revenue of $247,442 million, net income of $88,308 million, and diluted EPS of $11.82, compared to $219,790 million, $71,383 million, and $9.55 in 2023.\n[tables/table-37 Microsoft FY2024.html]\n\nThese pro forma results were based on estimates and assumptions, which we believe are reasonable. They are not the results that would have been realized had we been a combined company during the periods presented and are not necessarily indicative of our consolidated results of operations in future periods. The pro forma results include adjustments related to purchase accounting, primarily amortization of intangible assets. Acquisition costs and other nonrecurring charges were immaterial and are included in the earliest period presented.\nNOTE 8 — GOODWILL\nChanges in the carrying amount of goodwill were as follows:\n\nMicrosoft's goodwill by segment from June 2023 to June 2025, showing totals increasing from $67,886M to $119,509M, driven by acquisitions in More Personal Computing.\n[tables/table-38 Goodwill by Segment.html]\n\nWe have recast certain prior period amounts to conform to the way we internally manage and monitor our business. Refer to Note 1 – Accounting Policies for further information.\nThe measurement periods for the valuation of assets acquired and liabilities assumed end as soon as information on the facts and circumstances that existed as of the acquisition dates becomes available, but do not exceed 12 months. Adjustments in purchase price allocations may require a change in the amounts allocated to goodwill during the periods in which the adjustments are determined.\nAny change in the goodwill amounts resulting from foreign currency translations and purchase accounting adjustments are presented as “Other” in the table above. Also included in “Other” are business dispositions and transfers between segments due to reorganizations, as applicable.\nAs discussed in Note 1 – Accounting Policies, during the first quarter of fiscal year 2025 we made changes to our segments. These segment changes also resulted in changes to our reporting units. We reallocated goodwill across impacted reporting units using a relative fair value approach. In addition, we completed an assessment of any potential goodwill impairment for all reporting units immediately prior to the reallocation and determined that no impairment existed.\nGoodwill Impairment\nWe test goodwill for impairment annually on May 1 at the reporting unit level, primarily using a discounted cash flow methodology with a peer-based, risk-adjusted weighted average cost of capital. We believe use of a discounted cash flow approach is the most reliable indicator of the fair values of the businesses.\nNo instances of impairment were identified in our May 1, 2025, May 1, 2024, or May 1, 2023 tests. As of June 30, 2025 and 2024, accumulated goodwill impairment was $11.3 billion.\nNOTE 9 — INTANGIBLE ASSETS\nThe components of intangible assets, all of which are finite-lived, were as follows:\n\nThe table shows gross carrying amounts, accumulated amortization, and net carrying amounts of intangible assets (marketing, technology, customer, contract-based) as of June 30, 2025 and 2024. Total net carrying amount decreased from $27,597M to $22,604M.\n[tables/table-39 Intangible Assets.html]\n\nNo material impairments of intangible assets were identified during fiscal years 2025, 2024, or 2023. We estimate that we have no significant residual value related to our intangible assets.\nThe components of intangible assets acquired during the periods presented were as follows:", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 38", + "metadata": { + "length": 4472, + "summary": "The passage discusses Microsoft's goodwill and intangible assets, primarily from the Activision Blizzard acquisition. Goodwill was assigned to the More Personal Computing segment, totaling $119.5 billion by June 2025, up from $67.9 billion in 2023. Intangible assets acquired had a fair value of $21.97 billion with a weighted average life of 15 years, including marketing-related ($11.6B, 24 years), technology-based ($9.7B, 4 years), and customer-related ($0.7B, 4 years). Pro forma financial results for fiscal year 2024 show revenue of $247.4 billion, net income of $88.3 billion, and diluted EPS of $11.82. No goodwill impairment was identified in annual tests for 2023, 2024, or 2025. Net intangible assets decreased from $27.6 billion to $22.6 billion between 2024 and 2025.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Goodwill", + "assigned", + "More", + "Personal", + "Computing", + "segment", + "The", + "goodwill", + "primarily", + "attributed", + "increased", + "synergies", + "expected", + "achieved", + "integration", + "Activision", + "Blizzard", + "Substantially", + "deductible", + "income", + "tax", + "purposes", + "Following", + "details", + "purchase", + "price", + "allocated", + "intangible", + "assets", + "acquired", + "table", + "shows", + "fair", + "totaling", + "21", + "969", + "million", + "weighted", + "average", + "life", + "15", + "years", + "broken", + "marketing", + "related", + "11", + "619M", + "24", + "technology", + "based", + "689M", + "customer", + "661M", + "supplemental", + "consolidated", + "financial", + "results", + "Microsoft", + "Corporation", + "unaudited", + "pro", + "forma", + "basis", + "acquisition", + "consummated", + "July", + "2022", + "fiscal", + "year", + "ended", + "June", + "30", + "2024", + "revenue", + "247", + "442", + "net", + "88", + "308", + "diluted", + "EPS", + "11.82", + "compared", + "219", + "790", + "71", + "383", + "9.55", + "2023", + "These", + "estimates", + "assumptions", + "reasonable", + "They", + "realized", + "combined", + "company", + "periods", + "presented", + "necessarily", + "indicative", + "operations", + "future", + "include", + "adjustments", + "accounting", + "amortization", + "Acquisition", + "costs", + "nonrecurring", + "charges", + "immaterial", + "included", + "earliest", + "period", + "NOTE", + "GOODWILL", + "Changes", + "carrying", + "amount", + "2025", + "showing", + "totals", + "increasing", + "67", + "886M", + "119", + "509M", + "driven", + "acquisitions", + "We", + "recast", + "prior", + "amounts", + "conform", + "internally", + "manage", + "monitor", + "business", + "Refer", + "Note", + "Accounting", + "Policies", + "information", + "measurement", + "valuation", + "liabilities", + "assumed", + "end", + "facts", + "circumstances", + "existed", + "dates", + "exceed", + "12", + "months", + "Adjustments", + "allocations", + "require", + "change", + "determined", + "Any", + "resulting", + "foreign", + "currency", + "translations", + "Other", + "Also", + "dispositions", + "transfers", + "segments", + "due", + "reorganizations", + "applicable", + "As", + "discussed", + "quarter", + "made", + "resulted", + "reporting", + "units", + "reallocated", + "impacted", + "relative", + "approach", + "In", + "addition", + "completed", + "assessment", + "potential", + "impairment", + "immediately", + "reallocation", + "Impairment", + "test", + "annually", + "May", + "unit", + "level", + "discounted", + "cash", + "flow", + "methodology", + "peer", + "risk", + "adjusted", + "cost", + "capital", + "reliable", + "indicator", + "values", + "businesses", + "No", + "instances", + "identified", + "tests", + "accumulated", + "11.3", + "billion", + "INTANGIBLE", + "ASSETS", + "components", + "finite", + "lived", + "gross", + "contract", + "Total", + "decreased", + "27", + "597M", + "22", + "604M", + "material", + "impairments", + "estimate", + "significant", + "residual" + ], + "keywords": [ + "Goodwill", + "Intangible Assets", + "Microsoft" + ], + "connect_to": [ + { + "target": "48034024-9c74-53e3-9b34-c6cf7c84825d", + "relation": "embeds", + "ref": "[tables/table-36 Intangible Assets.html]", + "position": { + "start": 644, + "end": 684 + } + }, + { + "target": "6c9eaeba-45ae-5065-9440-b0d0155eab20", + "relation": "embeds", + "ref": "[tables/table-37 Microsoft FY2024.html]", + "position": { + "start": 1071, + "end": 1110 + } + }, + { + "target": "7bd19c3e-a5c9-5270-a340-a4edc82d6206", + "relation": "embeds", + "ref": "[tables/table-38 Goodwill by Segment.html]", + "position": { + "start": 1901, + "end": 1943 + } + }, + { + "target": "232454cc-54aa-5ec2-a0c2-982930a29730", + "relation": "embeds", + "ref": "[tables/table-39 Intangible Assets.html]", + "position": { + "start": 4149, + "end": 4189 + } + } + ] + } + }, + { + "chunk_id": "68666be5-1713-5c0d-b949-58f2d2493d46", + "type": "text", + "content": "The table shows intangible assets by type (marketing, technology, customer, contract) with amounts and weighted average lives for fiscal years 2025 and 2024. Total assets decreased from $23,264M to $1,096M.\n[tables/table-40 Intangible Assets.html]\n\nIntangible assets amortization expense was $6.0 billion, $4.8 billion, and $2.5 billion for fiscal years 2025, 2024, and 2023, respectively.\nThe following table outlines the estimated future amortization expense related to intangible assets held as of June 30, 2025:\n\nThe table shows projected revenue from 2026 to 2030 and thereafter, totaling $22,604 million, with the largest amount in the 'Thereafter' period.\n[tables/table-41 Revenue Forecast.html]\n\nNOTE 10 — DEBT\nShort-term Debt\nAs of June 30, 2025, we had no commercial paper issued or outstanding. As of June 30, 2024, we had $6.7 billion of commercial paper issued and outstanding, with a weighted average interest rate of 5.4% and maturities ranging from 28 days to 152 days. The estimated fair value of this commercial paper approximates its carrying value.\nLong-term Debt\nThe components of long-term debt were as follows:\n\nThe table details long-term debt issuances from 2009 to 2024, including face values, maturities, stated and effective interest rates, and carrying amounts as of June 30, 2025 and 2024, totaling $43,151 million and $44,937 million respectively.\n[tables/table-42 Debt Issuance Table.html]\n\n(a) Refer to Note 5 – Derivatives for further information on the interest rate swaps related to fixed-rate debt.\nAs of June 30, 2025 and 2024, the estimated fair value of long-term debt, including the current portion, was $40.4 billion and $42.3 billion, respectively. The estimated fair values are based on Level 2 inputs.\nDebt in the table above is comprised of senior unsecured obligations and ranks equally with our other outstanding obligations. Interest is paid semi-annually, except for the Euro-denominated debt, which is paid annually. Cash paid for interest on our debt for fiscal years 2025, 2024, and 2023 was $1.6 billion, $1.7 billion, and $1.7 billion, respectively.\nThe following table outlines maturities of our long-term debt, including the current portion, as of June 30, 2025:\n\nThe table shows debt maturities from 2026 onward, with $49,206 million total. Key payments: $3,000M in 2026, $9,250M in 2027, $2,054M in 2029, and $34,902M thereafter.\n[tables/table-43 Debt Maturity.html]\n\nNOTE 11 — INCOME TAXES\nProvision for Income Taxes\nThe components of the provision for income taxes were as follows:\n\nThe table shows current and deferred tax provisions for fiscal years 2023-2025. Current taxes total $28,851M (2025), $24,389M (2024), $23,009M (2023). Deferred taxes are negative, resulting in total provision of $21,795M, $19,651M, $16,950M respectively.\n[tables/table-44 Income Tax Provision.html]\n\nU.S. and foreign components of income before income taxes were as follows:\n\nThe table shows income before income taxes for years 2025, 2024, and 2023, with U.S. and foreign breakdowns, totaling $123,627 million, $107,787 million, and $89,311 million respectively.\n[tables/table-45 Income Before Taxes.html]\n\nEffective Tax Rate\nThe items accounting for the difference between income taxes computed at the U.S. federal statutory rate and our effective rate were as follows:\n\nThe table shows a reconciliation of the federal statutory tax rate (21%) to the effective tax rate for fiscal years 2023-2025, with adjustments for foreign earnings, credits, and other items.\n[tables/table-46 Tax Rate Analysis.html]\n\nThe decrease from the federal statutory rate in fiscal years 2025, 2024, and 2023 is primarily due to earnings taxed at lower rates in foreign jurisdictions resulting from producing and distributing our products and services through our foreign regional operations center in Ireland. In fiscal years 2025, 2024, and 2023, our foreign regional operating center in Ireland, which is taxed at a rate lower than the U.S. rate, generated 81%, 83%, and 81% of our foreign income before tax. Other reconciling items, net consists primarily of tax credits and the U.S. global intangible low-taxed income tax, and in fiscal year 2024, includes tax benefits from tax law changes. In fiscal year 2024, tax benefits from tax law changes primarily relate to the delay of the effective date of final foreign tax credit regulations. In fiscal years 2025, 2024, and 2023, there were no individually significant other reconciling items.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 39", + "metadata": { + "length": 4470, + "summary": "The passage covers Notes 10 and 11 from a financial report. Note 10 details debt: no commercial paper in 2025 (vs. $6.7B in 2024), long-term debt totaling $43.151B (2025) and $44.937B (2024) with fair values of $40.4B and $42.3B, and future maturities of $49.206B. Note 11 covers income taxes: total provisions of $21.795B (2025), $19.651B (2024), $16.950B (2023); income before taxes of $123.627B, $107.787B, $89.311B; effective tax rate reconciliation from 21% federal rate, with lower foreign rates (81-83% from Ireland) as primary driver. Intangible assets amortization was $6.0B (2025), $4.8B (2024), $2.5B (2023), with future amortization totaling $22.604M.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "The", + "table", + "shows", + "intangible", + "assets", + "type", + "marketing", + "technology", + "customer", + "contract", + "amounts", + "weighted", + "average", + "lives", + "fiscal", + "years", + "2025", + "2024", + "Total", + "decreased", + "23", + "264M", + "096M", + "Intangible", + "amortization", + "expense", + "6.0", + "billion", + "4.8", + "2.5", + "2023", + "outlines", + "estimated", + "future", + "related", + "held", + "June", + "30", + "projected", + "revenue", + "2026", + "2030", + "totaling", + "22", + "604", + "million", + "largest", + "amount", + "Thereafter", + "period", + "NOTE", + "10", + "DEBT", + "Short", + "term", + "Debt", + "As", + "commercial", + "paper", + "issued", + "outstanding", + "6.7", + "interest", + "rate", + "5.4%", + "maturities", + "ranging", + "28", + "days", + "152", + "fair", + "approximates", + "carrying", + "Long", + "components", + "long", + "debt", + "details", + "issuances", + "2009", + "including", + "face", + "values", + "stated", + "effective", + "rates", + "43", + "151", + "44", + "937", + "Refer", + "Note", + "Derivatives", + "information", + "swaps", + "fixed", + "current", + "portion", + "40.4", + "42.3", + "based", + "Level", + "inputs", + "comprised", + "senior", + "unsecured", + "obligations", + "ranks", + "equally", + "Interest", + "paid", + "semi", + "annually", + "Euro", + "denominated", + "Cash", + "1.6", + "1.7", + "onward", + "49", + "206", + "total", + "Key", + "payments", + "000M", + "250M", + "2027", + "054M", + "2029", + "34", + "902M", + "11", + "INCOME", + "TAXES", + "Provision", + "Income", + "Taxes", + "provision", + "income", + "taxes", + "deferred", + "tax", + "provisions", + "Current", + "851M", + "24", + "389M", + "009M", + "Deferred", + "negative", + "resulting", + "21", + "795M", + "19", + "651M", + "16", + "950M", + "foreign", + "breakdowns", + "123", + "627", + "107", + "787", + "89", + "311", + "Effective", + "Tax", + "Rate", + "items", + "accounting", + "difference", + "computed", + "federal", + "statutory", + "reconciliation", + "21%", + "adjustments", + "earnings", + "credits", + "decrease", + "primarily", + "due", + "taxed", + "lower", + "jurisdictions", + "producing", + "distributing", + "products", + "services", + "regional", + "operations", + "center", + "Ireland", + "In", + "operating", + "generated", + "81%", + "83%", + "Other", + "reconciling", + "net", + "consists", + "global", + "low", + "year", + "includes", + "benefits", + "law", + "relate", + "delay", + "date", + "final", + "credit", + "regulations", + "individually", + "significant" + ], + "keywords": [ + "Intangible Assets", + "Debt", + "Income Taxes" + ], + "connect_to": [ + { + "target": "94c04c68-141f-5982-90ac-5f46acfb78bc", + "relation": "embeds", + "ref": "[tables/table-40 Intangible Assets.html]", + "position": { + "start": 207, + "end": 247 + } + }, + { + "target": "9ff99e1e-0c88-5acf-bc6c-4cdc1c1ab453", + "relation": "embeds", + "ref": "[tables/table-41 Revenue Forecast.html]", + "position": { + "start": 663, + "end": 702 + } + }, + { + "target": "203d64ab-9b51-5ee5-830b-f7042569057c", + "relation": "embeds", + "ref": "[tables/table-42 Debt Issuance Table.html]", + "position": { + "start": 1379, + "end": 1421 + } + }, + { + "target": "2a31a2e6-10ff-5f17-b76c-be44e548759b", + "relation": "embeds", + "ref": "[tables/table-43 Debt Maturity.html]", + "position": { + "start": 2389, + "end": 2425 + } + }, + { + "target": "a38735ed-f1e5-59f3-b5b1-bb9c3195605b", + "relation": "embeds", + "ref": "[tables/table-44 Income Tax Provision.html]", + "position": { + "start": 2799, + "end": 2842 + } + }, + { + "target": "35db7d9d-f52f-5807-9862-a55597bd303b", + "relation": "embeds", + "ref": "[tables/table-45 Income Before Taxes.html]", + "position": { + "start": 3108, + "end": 3150 + } + }, + { + "target": "efea2194-29ab-5e8e-9e8a-b1ade3dc4796", + "relation": "embeds", + "ref": "[tables/table-46 Tax Rate Analysis.html]", + "position": { + "start": 3509, + "end": 3549 + } + } + ] + } + }, + { + "chunk_id": "5cec060e-348a-5733-8ceb-d7f52255220b", + "type": "text", + "content": "The decrease in our effective tax rate for fiscal year 2025 compared to fiscal year 2024 was due to changes in the mix of our earnings and tax expenses between the U.S. and foreign countries. The decrease in our effective tax rate for fiscal year 2024 compared to fiscal year 2023 was primarily due to tax benefits from tax law changes, including the delay of the effective date of final foreign tax credit regulations.\nThe components of the deferred income tax assets and liabilities were as follows:\n\nThe table details deferred income tax assets and liabilities as of June 30, 2025 and 2024, with net deferred tax assets of $26,273 million and $19,652 million respectively.\n[tables/table-47 Deferred Tax Assets.html]\n\nDeferred income tax balances reflect the effects of temporary differences between the carrying amounts of assets and liabilities and their tax bases and are stated at enacted tax rates expected to be in effect when the taxes are paid or recovered.\nAs of June 30, 2025, we had federal, state, and foreign net operating loss carryforwards of $390 million, $836 million, and $2.6 billion, respectively. The federal and state net operating loss carryforwards have varying expiration dates ranging from fiscal year 2026 to 2045 or indefinite carryforward periods, if not utilized. The majority of our foreign net operating loss carryforwards do not expire. Certain acquired net operating loss carryforwards are subject to an annual limitation but are expected to be realized with the exception of those which have a valuation allowance. As of June 30, 2025, we had $816 million federal capital loss carryforwards for U.S. tax purposes. The federal capital loss carryforwards will expire in fiscal year 2030 if not utilized.\nThe valuation allowance disclosed in the table above relates to the foreign net operating loss carryforwards, federal capital loss carryforwards, and other net deferred tax assets that may not be realized.\nIncome taxes paid, net of refunds, were $28.7 billion, $23.4 billion, and $23.1 billion in fiscal years 2025, 2024, and 2023, respectively.\nUncertain Tax Positions\nGross unrecognized tax benefits related to uncertain tax positions as of June 30, 2025, 2024, and 2023, were $24.7 billion, $22.8 billion, and $17.1 billion, respectively, which were primarily included in long-term income taxes in our consolidated balance sheets. If recognized, the resulting tax benefit would affect our effective tax rates for fiscal years 2025, 2024, and 2023 by $21.2 billion, $19.6 billion, and $14.4 billion, respectively.\nAs of June 30, 2025, 2024, and 2023, we had accrued interest expense related to uncertain tax positions of $8.2 billion, $6.8 billion, and $5.2 billion, respectively, net of income tax benefits. The provision for income taxes for fiscal years 2025, 2024, and 2023 included interest expense related to uncertain tax positions of $1.3 billion, $1.5 billion, and $918 million, respectively, net of income tax benefits.\nThe aggregate changes in the gross unrecognized tax benefits related to uncertain tax positions were as follows:\n\nThe table shows unrecognized tax benefits from 2023 to 2025, starting at $15,593 million and ending at $24,729 million, with changes from settlements, current and prior year positions, and statute lapses.\n[tables/table-48 Tax Benefits.html]\n\nWe remain under audit by the IRS for tax years 2014 to 2017. With respect to the audit for tax years 2004 to 2013, on September 26, 2023, we received Notices of Proposed Adjustment (“NOPAs”) from the IRS. The primary issues in the NOPAs relate to intercompany transfer pricing. In the NOPAs, the IRS is seeking an additional tax payment of $28.9 billion plus penalties and interest. As of June 30, 2025, we believe our allowances for income tax contingencies are adequate. We disagree with the proposed adjustments and will vigorously contest the NOPAs through the IRS’s administrative appeals office and, if necessary, judicial proceedings. We do not expect a final resolution of these issues in the next 12 months. Based on the information currently available, we do not anticipate a significant increase or decrease to our income tax contingencies for these issues within the next 12 months.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 40", + "metadata": { + "length": 4221, + "summary": "The passage discusses changes in the effective tax rate for fiscal years 2025, 2024, and 2023, attributed to earnings mix and tax law changes. It details deferred income tax assets and liabilities, with net deferred tax assets of $26,273 million and $19,652 million as of June 30, 2025 and 2024. Net operating loss carryforwards are reported for federal ($390 million), state ($836 million), and foreign ($2.6 billion) jurisdictions, along with federal capital loss carryforwards ($816 million). A valuation allowance is noted for certain deferred tax assets. Income taxes paid were $28.7 billion, $23.4 billion, and $23.1 billion for fiscal years 2025, 2024, and 2023. Gross unrecognized tax benefits related to uncertain tax positions were $24.7 billion, $22.8 billion, and $17.1 billion as of June 30, 2025, 2024, and 2023, with accrued interest of $8.2 billion, $6.8 billion, and $5.2 billion. The IRS is auditing tax years 2014 to 2017 and issued NOPAs for 2004 to 2013 seeking $28.9 billion plus penalties, which the company contests.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "The", + "decrease", + "effective", + "tax", + "rate", + "fiscal", + "year", + "2025", + "compared", + "2024", + "due", + "mix", + "earnings", + "expenses", + "foreign", + "countries", + "2023", + "primarily", + "benefits", + "law", + "including", + "delay", + "date", + "final", + "credit", + "regulations", + "components", + "deferred", + "income", + "assets", + "liabilities", + "table", + "details", + "June", + "30", + "net", + "26", + "273", + "million", + "19", + "652", + "Deferred", + "balances", + "reflect", + "effects", + "temporary", + "differences", + "carrying", + "amounts", + "bases", + "stated", + "enacted", + "rates", + "expected", + "effect", + "taxes", + "paid", + "recovered", + "As", + "federal", + "state", + "operating", + "loss", + "carryforwards", + "390", + "836", + "2.6", + "billion", + "varying", + "expiration", + "dates", + "ranging", + "2026", + "2045", + "indefinite", + "carryforward", + "periods", + "utilized", + "majority", + "expire", + "Certain", + "acquired", + "subject", + "annual", + "limitation", + "realized", + "exception", + "valuation", + "allowance", + "816", + "capital", + "purposes", + "2030", + "disclosed", + "relates", + "Income", + "refunds", + "28.7", + "23.4", + "23.1", + "years", + "Uncertain", + "Tax", + "Positions", + "Gross", + "unrecognized", + "related", + "uncertain", + "positions", + "24.7", + "22.8", + "17.1", + "included", + "long", + "term", + "consolidated", + "balance", + "sheets", + "If", + "recognized", + "resulting", + "benefit", + "affect", + "21.2", + "19.6", + "14.4", + "accrued", + "interest", + "expense", + "8.2", + "6.8", + "5.2", + "provision", + "1.3", + "1.5", + "918", + "aggregate", + "gross", + "shows", + "starting", + "15", + "593", + "ending", + "24", + "729", + "settlements", + "current", + "prior", + "statute", + "lapses", + "We", + "remain", + "audit", + "IRS", + "2014", + "2017", + "With", + "respect", + "2004", + "2013", + "September", + "received", + "Notices", + "Proposed", + "Adjustment", + "NOPAs", + "primary", + "issues", + "relate", + "intercompany", + "transfer", + "pricing", + "In", + "seeking", + "additional", + "payment", + "28.9", + "penalties", + "allowances", + "contingencies", + "adequate", + "disagree", + "proposed", + "adjustments", + "vigorously", + "contest", + "administrative", + "appeals", + "office", + "judicial", + "proceedings", + "expect", + "resolution", + "12", + "months", + "Based", + "information", + "anticipate", + "significant", + "increase" + ], + "keywords": [ + "effective tax rate", + "deferred tax assets", + "uncertain tax positions" + ], + "connect_to": [ + { + "target": "37e9fc8e-3678-5a29-9e1e-7562a9ca8023", + "relation": "embeds", + "ref": "[tables/table-47 Deferred Tax Assets.html]", + "position": { + "start": 676, + "end": 718 + } + }, + { + "target": "84b6f224-48d2-548e-bb74-41134065a57a", + "relation": "embeds", + "ref": "[tables/table-48 Tax Benefits.html]", + "position": { + "start": 3290, + "end": 3325 + } + } + ] + } + }, + { + "chunk_id": "64b89ba2-ec71-5757-94b7-ce60a3061efc", + "type": "text", + "content": "We are subject to income tax in many jurisdictions outside the U.S., some of which are currently under audit by local tax authorities. The resolution of these audits is not expected to be material to our consolidated financial statements. Our operations in Ireland remain subject to examination for tax years 2020 and thereafter.\nNOTE 12 — UNEARNED REVENUE\nUnearned revenue by segment was as follows:\n\nMicrosoft's total revenue increased from $60,184 million in 2024 to $67,265 million in 2025, driven by growth in Productivity and Business Processes and Intelligent Cloud.\n[tables/table-49 Microsoft Revenue.html]\n\nWe have recast certain prior period amounts to conform to the way we internally manage and monitor our business. Refer to Note 1 – Accounting Policies for further information.\nChanges in unearned revenue were as follows:\n\nThe table shows unearned revenue balance increased from $60,184M to $67,265M, with $186,957M deferred and $179,876M recognized in fiscal year 2025.\n[tables/table-50 Unearned Revenue.html]\n\nRevenue allocated to remaining performance obligations, which includes unearned revenue and amounts that will be invoiced and recognized as revenue in future periods, was $375 billion as of June 30, 2025, of which $368 billion is related to the commercial portion of revenue. We expect to recognize approximately 40% of our total company remaining performance obligation revenue over the next 12 months and the remainder thereafter.\nNOTE 13 — LEASES\nWe have operating and finance leases for datacenters, corporate offices, research and development facilities, Microsoft Experience Centers, and certain equipment. Our leases have remaining lease terms of less than 1 year to 20 years, some of which include options to extend the leases for up to 5 years, and some of which include options to terminate the leases within 1 year.\nThe components of lease expense were as follows:\n\nThe table shows operating lease costs of $5,524M (2025), $3,555M (2024), $2,875M (2023) and total finance lease costs of $4,825M, $2,534M, $1,853M for the same years.\n[tables/table-51 Lease Costs.html]\n\nSupplemental cash flow information related to leases was as follows:\n\nThe table shows cash paid for operating and finance leases from 2023 to 2025, and right-of-use assets obtained in exchange for lease obligations.\n[tables/table-52 Lease Cash Flows.html]\n\nSupplemental balance sheet information related to leases was as follows:\n\nThe table presents operating and finance lease data as of June 30, 2025 and 2024, including right-of-use assets, liabilities, weighted average lease terms, and discount rates.\n[tables/table-53 Lease Data.html]\n\nThe following table outlines maturities of our lease liabilities as of June 30, 2025:\n\nThe table shows future lease payments for operating and finance leases from 2026 to 2030 and thereafter, totaling $25,481 million and $60,506 million respectively, with imputed interest deducted to present net lease liabilities.\n[tables/table-54 Lease Payments.html]\n\nAs of June 30, 2025, we had additional leases, primarily for datacenters, that had not yet commenced of $92.7 billion. These leases will commence between fiscal year 2026 and fiscal year 2031 with lease terms of 1 year to 20 years.\nNOTE 14 — CONTINGENCIES\nIrish Data Protection Commission Matter\nIn 2018, the Irish Data Protection Commission (“IDPC”) began investigating a complaint against LinkedIn as to whether LinkedIn’s targeted advertising practices violated the recently implemented European Union General Data Protection Regulation (“GDPR”). Microsoft cooperated throughout the period of inquiry. In October 2024, the IDPC provided LinkedIn with a final decision alleging GDPR violations and assessing a fine. In November 2024, LinkedIn appealed the final decision to the Irish courts, and the next hearing is scheduled for December 2025.\nOther Contingencies\nWe also are subject to a variety of other claims and suits that arise from time to time in the ordinary course of our business. Although management currently believes that resolving claims against us, individually or in aggregate, will not have a material adverse impact in our consolidated financial statements, these matters are subject to inherent uncertainties and management’s view of these matters may change in the future.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 41", + "metadata": { + "length": 4300, + "summary": "The passage covers financial notes including unearned revenue, leases, and contingencies. Unearned revenue increased from $60,184M to $67,265M, with $186,957M deferred and $179,876M recognized in fiscal year 2025. Remaining performance obligations total $375 billion. Lease disclosures detail operating and finance lease costs, cash flows, balance sheet data, and maturities totaling $25,481M for operating and $60,506M for finance leases, with $92.7 billion in additional leases not yet commenced. Contingencies include an Irish Data Protection Commission matter against LinkedIn with a fine appealed to Irish courts, and other ordinary course claims not expected to be material.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "We", + "subject", + "income", + "tax", + "jurisdictions", + "audit", + "local", + "authorities", + "The", + "resolution", + "audits", + "expected", + "material", + "consolidated", + "financial", + "statements", + "Our", + "operations", + "Ireland", + "remain", + "examination", + "years", + "2020", + "NOTE", + "12", + "UNEARNED", + "REVENUE", + "Unearned", + "revenue", + "segment", + "Microsoft", + "total", + "increased", + "60", + "184", + "million", + "2024", + "67", + "265", + "2025", + "driven", + "growth", + "Productivity", + "Business", + "Processes", + "Intelligent", + "Cloud", + "recast", + "prior", + "period", + "amounts", + "conform", + "internally", + "manage", + "monitor", + "business", + "Refer", + "Note", + "Accounting", + "Policies", + "information", + "Changes", + "unearned", + "table", + "shows", + "balance", + "184M", + "265M", + "186", + "957M", + "deferred", + "179", + "876M", + "recognized", + "fiscal", + "year", + "Revenue", + "allocated", + "remaining", + "performance", + "obligations", + "includes", + "invoiced", + "future", + "periods", + "375", + "billion", + "June", + "30", + "368", + "related", + "commercial", + "portion", + "expect", + "recognize", + "approximately", + "40%", + "company", + "obligation", + "months", + "remainder", + "13", + "LEASES", + "operating", + "finance", + "leases", + "datacenters", + "corporate", + "offices", + "research", + "development", + "facilities", + "Experience", + "Centers", + "equipment", + "lease", + "terms", + "20", + "include", + "options", + "extend", + "terminate", + "components", + "expense", + "costs", + "524M", + "555M", + "875M", + "2023", + "825M", + "534M", + "853M", + "Supplemental", + "cash", + "flow", + "paid", + "assets", + "obtained", + "exchange", + "sheet", + "presents", + "data", + "including", + "liabilities", + "weighted", + "average", + "discount", + "rates", + "outlines", + "maturities", + "payments", + "2026", + "2030", + "totaling", + "25", + "481", + "506", + "imputed", + "interest", + "deducted", + "present", + "net", + "As", + "additional", + "primarily", + "commenced", + "92.7", + "These", + "commence", + "2031", + "14", + "CONTINGENCIES", + "Irish", + "Data", + "Protection", + "Commission", + "Matter", + "In", + "2018", + "IDPC", + "began", + "investigating", + "complaint", + "LinkedIn", + "targeted", + "advertising", + "practices", + "violated", + "recently", + "implemented", + "European", + "Union", + "General", + "Regulation", + "GDPR", + "cooperated", + "inquiry", + "October", + "provided", + "final", + "decision", + "alleging", + "violations", + "assessing", + "fine", + "November", + "appealed", + "courts", + "hearing", + "scheduled", + "December", + "Other", + "Contingencies", + "variety", + "claims", + "suits", + "arise", + "time", + "ordinary", + "Although", + "management", + "believes", + "resolving", + "individually", + "aggregate", + "adverse", + "impact", + "matters", + "inherent", + "uncertainties", + "view", + "change" + ], + "keywords": [ + "Unearned Revenue", + "Leases", + "Contingencies" + ], + "connect_to": [ + { + "target": "db4a750b-c288-58dc-99c2-ff1f5efc436f", + "relation": "embeds", + "ref": "[tables/table-49 Microsoft Revenue.html]", + "position": { + "start": 574, + "end": 614 + } + }, + { + "target": "087287ba-e4d1-5ad9-b7be-1fdffdd0a05f", + "relation": "embeds", + "ref": "[tables/table-50 Unearned Revenue.html]", + "position": { + "start": 986, + "end": 1025 + } + }, + { + "target": "4452390e-74dd-5620-b691-6e58c8dc77de", + "relation": "embeds", + "ref": "[tables/table-51 Lease Costs.html]", + "position": { + "start": 2071, + "end": 2105 + } + }, + { + "target": "28d25b09-4699-514f-8b98-01fc0bd50366", + "relation": "embeds", + "ref": "[tables/table-52 Lease Cash Flows.html]", + "position": { + "start": 2323, + "end": 2362 + } + }, + { + "target": "da108065-e4c5-5301-95cb-bfa49a12d84f", + "relation": "embeds", + "ref": "[tables/table-53 Lease Data.html]", + "position": { + "start": 2614, + "end": 2647 + } + }, + { + "target": "e267be8c-3565-5cd5-9a8a-d0d1596d29f2", + "relation": "embeds", + "ref": "[tables/table-54 Lease Payments.html]", + "position": { + "start": 2965, + "end": 3002 + } + } + ] + } + }, + { + "chunk_id": "aba2c310-8f8d-5daa-95ce-b3df70bd910a", + "type": "text", + "content": "As of June 30, 2025, we accrued aggregate legal liabilities of $541 million. While we intend to defend these matters vigorously, adverse outcomes that we estimate could reach approximately $600 million in aggregate beyond recorded amounts are reasonably possible. Were unfavorable final outcomes to occur, there exists the possibility of a material adverse impact in our consolidated financial statements for the period in which the effects become reasonably estimable.\nNOTE 15 — STOCKHOLDERS’ EQUITY\nShares Outstanding\nShares of common stock outstanding were as follows:\n\nThe table shows share balances for fiscal years 2023-2025, with beginning balances around 7,432-7,464 million, issuances of 31-37 million, repurchases of 31-69 million, and ending balances of 7,432-7,434 million.\n[tables/table-55 Share Repurchases.html]\n\nShare Repurchases\nOn September 14, 2021, our Board of Directors approved a share repurchase program authorizing up to $60.0 billion in share repurchases. This share repurchase program commenced in November 2021 and was completed in April 2025.\nOn September 16, 2024, our Board of Directors approved a share repurchase program authorizing up to $60.0 billion in share repurchases. This share repurchase program commenced in April 2025, following completion of the program approved on September 14, 2021, has no expiration date, and may be terminated at any time. As of June 30, 2025, $57.3 billion remained of this $60.0 billion share repurchase program.\nWe repurchased the following shares of common stock under the share repurchase programs:\n\nThe table shows quarterly share repurchases and amounts for fiscal years 2023-2025, with totals of 69 shares ($18,400M) in 2023, 32 shares ($11,960M) in 2024, and 31 shares ($13,000M) in 2025.\n[tables/table-56 Share Repurchases.html]\n\nAll repurchases were made using cash resources. Shares repurchased during the fourth quarter of fiscal year 2025 were under the share repurchase programs approved on September 14, 2021 and September 16, 2024. All other shares repurchased were under the share repurchase program approved on September 14, 2021. The above table excludes shares repurchased to settle employee tax withholding related to the vesting of stock awards of $5.4 billion, $5.3 billion, and $3.8 billion for fiscal years 2025, 2024, and 2023, respectively.\nDividends\nOur Board of Directors declared the following dividends:\n\nThe table shows quarterly dividends for fiscal years 2025 and 2024, with per-share amounts of $0.83 and $0.75 respectively, totaling $24,678 million and $22,295 million.\n[tables/table-57 Dividend History.html]\n\nThe dividend declared on June 10, 2025 was included in other current liabilities as of June 30, 2025.\nNOTE 16 — ACCUMULATED OTHER COMPREHENSIVE INCOME (LOSS)\nThe following table summarizes the changes in accumulated other comprehensive income (loss) by component:\n\nThe table shows changes in accumulated other comprehensive loss for derivatives, investments, and translation adjustments for fiscal years 2023-2025, ending with a total loss of $(3,347) million in 2025.\n[tables/table-58 AOCI Rollforward.html]\n\nNOTE 17 — EMPLOYEE STOCK AND SAVINGS PLANS\nWe grant stock-based compensation to employees and directors. Awards that expire or are canceled without delivery of shares generally become available for issuance under the plans. We issue new shares of Microsoft common stock to satisfy vesting of awards granted under our stock plans. We also have an ESPP for all eligible employees.\nStock-based compensation expense and related income tax benefits were as follows:\n\nThe table shows stock-based compensation expenses and related income tax benefits for fiscal years 2025, 2024, and 2023, with amounts in millions of dollars.\n[tables/table-59 Stock Compensation.html]\n\nStock Plans\nStock awards entitle the holder to receive shares of Microsoft common stock as the award vests. Stock awards generally vest over a service period of four years or five years.\nExecutive Incentive Plan\nUnder the Executive Incentive Plan, the Compensation Committee approves stock awards to executive officers and certain senior executives. RSUs generally vest ratably over a service period of four years. PSUs generally vest over a performance period of three years. The number of shares the PSU holder receives is based on the extent to which the corresponding performance goals have been achieved.\nActivity for All Stock Plans\nThe fair value of stock awards was estimated on the date of grant using the following assumptions:", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 42", + "metadata": { + "length": 4525, + "summary": "The passage details financial notes including $541 million in accrued legal liabilities with potential additional $600 million in adverse outcomes. It covers share repurchases under two $60 billion programs, with $57.3 billion remaining as of June 2025, and quarterly repurchases totaling 31 shares ($13,000M) in 2025. Dividends declared were $0.83 per share in 2025 and $0.75 in 2024. Accumulated other comprehensive loss was $(3,347) million in 2025. Stock-based compensation expenses and related tax benefits are noted, along with stock award vesting periods and activity assumptions.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "As", + "June", + "30", + "2025", + "accrued", + "aggregate", + "legal", + "liabilities", + "541", + "million", + "While", + "intend", + "defend", + "matters", + "vigorously", + "adverse", + "outcomes", + "estimate", + "reach", + "approximately", + "600", + "recorded", + "amounts", + "Were", + "unfavorable", + "final", + "occur", + "exists", + "possibility", + "material", + "impact", + "consolidated", + "financial", + "statements", + "period", + "effects", + "estimable", + "NOTE", + "15", + "STOCKHOLDERS", + "EQUITY", + "Shares", + "Outstanding", + "common", + "stock", + "outstanding", + "The", + "table", + "shows", + "share", + "balances", + "fiscal", + "years", + "2023", + "beginning", + "432", + "464", + "issuances", + "31", + "37", + "repurchases", + "69", + "ending", + "434", + "Share", + "Repurchases", + "On", + "September", + "14", + "2021", + "Board", + "Directors", + "approved", + "repurchase", + "program", + "authorizing", + "60.0", + "billion", + "This", + "commenced", + "November", + "completed", + "April", + "16", + "2024", + "completion", + "expiration", + "date", + "terminated", + "time", + "57.3", + "remained", + "We", + "repurchased", + "shares", + "programs", + "quarterly", + "totals", + "18", + "400M", + "32", + "11", + "960M", + "13", + "000M", + "All", + "made", + "cash", + "resources", + "fourth", + "quarter", + "year", + "excludes", + "settle", + "employee", + "tax", + "withholding", + "related", + "vesting", + "awards", + "5.4", + "5.3", + "3.8", + "Dividends", + "Our", + "declared", + "dividends", + "0.83", + "0.75", + "totaling", + "24", + "678", + "22", + "295", + "dividend", + "10", + "included", + "current", + "ACCUMULATED", + "OTHER", + "COMPREHENSIVE", + "INCOME", + "LOSS", + "summarizes", + "accumulated", + "comprehensive", + "income", + "loss", + "component", + "derivatives", + "investments", + "translation", + "adjustments", + "total", + "347", + "17", + "EMPLOYEE", + "STOCK", + "AND", + "SAVINGS", + "PLANS", + "grant", + "based", + "compensation", + "employees", + "directors", + "Awards", + "expire", + "canceled", + "delivery", + "generally", + "issuance", + "plans", + "issue", + "Microsoft", + "satisfy", + "granted", + "ESPP", + "eligible", + "Stock", + "expense", + "benefits", + "expenses", + "millions", + "dollars", + "Plans", + "entitle", + "holder", + "receive", + "award", + "vests", + "vest", + "service", + "Executive", + "Incentive", + "Plan", + "Under", + "Compensation", + "Committee", + "approves", + "executive", + "officers", + "senior", + "executives", + "RSUs", + "ratably", + "PSUs", + "performance", + "number", + "PSU", + "receives", + "extent", + "goals", + "achieved", + "Activity", + "fair", + "estimated", + "assumptions" + ], + "keywords": [ + "Legal liabilities", + "Share repurchases", + "Stock plans" + ], + "connect_to": [ + { + "target": "36c1f271-f9e2-5fb3-b351-1f862f2b95fe", + "relation": "embeds", + "ref": "[tables/table-55 Share Repurchases.html]", + "position": { + "start": 786, + "end": 826 + } + }, + { + "target": "657a22eb-9366-5d37-b1fa-074b1683ae81", + "relation": "embeds", + "ref": "[tables/table-56 Share Repurchases.html]", + "position": { + "start": 1765, + "end": 1805 + } + }, + { + "target": "0a3c08c8-285f-5234-ba6e-dcd0afd088ac", + "relation": "embeds", + "ref": "[tables/table-57 Dividend History.html]", + "position": { + "start": 2574, + "end": 2613 + } + }, + { + "target": "61be9b16-9678-54c0-89a8-bc758632e3a0", + "relation": "embeds", + "ref": "[tables/table-58 AOCI Rollforward.html]", + "position": { + "start": 3084, + "end": 3123 + } + }, + { + "target": "94b78184-d619-5a1d-b89d-8bb7afb772ab", + "relation": "embeds", + "ref": "[tables/table-59 Stock Compensation.html]", + "position": { + "start": 3745, + "end": 3786 + } + } + ] + } + }, + { + "chunk_id": "4859b1f9-256b-5961-aaed-0b88ac0439cc", + "type": "text", + "content": "The table shows quarterly dividends per share and interest rates for fiscal years 2023-2025. Dividends range from $0.62 to $0.83, and interest rates range from 2.0% to 5.6%.\n[tables/table-60 Dividends & Rates.html]\n\nDuring fiscal year 2025, the following activity occurred under our stock plans:\n\nThe table shows nonvested stock award activity: beginning balance 88 shares at $292.28, 39 granted at $413.90, 38 vested at $293.25, 7 forfeited at $317.23, ending balance 82 shares at $347.44.\n[tables/table-61 Stock Awards.html]\n\n(a) Includes 1 million of PSUs granted at target and performance adjustments above target levels for each of the fiscal years 2025, 2024, and 2023.\nAs of June 30, 2025, total unrecognized compensation costs related to stock awards were $21.6 billion. These costs are expected to be recognized over a weighted average period of three years. The weighted average grant-date fair value of stock awards granted was $413.90, $339.46, and $252.59 for fiscal years 2025, 2024, and 2023, respectively. The fair value of stock awards vested was $16.2 billion, $16.0 billion, and $11.9 billion, for fiscal years 2025, 2024, and 2023, respectively. As of June 30, 2025, an aggregate of 98 million shares were authorized for future grant under our stock plans.\nEmployee Stock Purchase Plan\nWe have an ESPP for all eligible employees. Shares of our common stock may be purchased by employees at three-month intervals at 90% of the fair market value on the last trading day of each three-month period. Employees may purchase shares having a value not exceeding 15% of their gross compensation during an offering period.\nEmployees purchased the following shares during the periods presented:\n\nThe table shows shares purchased and average price per share for fiscal years 2025, 2024, and 2023. Shares purchased were 6 million in 2025 and 2024, and 7 million in 2023. Average prices were $385.10, $339.46, and $245.59 respectively.\n[tables/table-62 Share Repurchase.html]\n\nAs of June 30, 2025, 62 million shares of our common stock were reserved for future issuance through the ESPP.\nSavings Plans\nWe have savings plans in the U.S. that qualify under Section 401(k) of the Internal Revenue Code, and a number of savings plans in international locations. Eligible U.S. employees may contribute a portion of their salary into the savings plans, subject to certain limitations. We match a portion of each dollar a participant contributes into the plans. Employer-funded retirement benefits for all plans were $1.8 billion, $1.7 billion, and $1.6 billion in fiscal years 2025, 2024, and 2023, respectively, and were expensed as contributed.\nNOTE 18 — SEGMENT INFORMATION AND GEOGRAPHIC DATA\nIn its operation of the business, management, including our chief operating decision maker (“CODM”), who is also our Chief Executive Officer, reviews certain financial information, including segmented internal profit and loss statements. The primary profitability measure used by the CODM to review segment operating results is operating income. The CODM uses operating income to allocate resources during our annual planning process and throughout the year, as well as to assess the performance of our segments, primarily by monitoring actual results compared to prior periods and expected results. During the periods presented, we reported our financial performance based on the following segments: Productivity and Business Processes, Intelligent Cloud, and More Personal Computing.\nWe have recast certain prior period amounts to conform to the way we internally manage and monitor our business. Refer to Note 1 – Accounting Policies for further information.\nOur reportable segments are described below.\nProductivity and Business Processes\nOur Productivity and Business Processes segment consists of products and services in our portfolio of productivity, communication, and information services, spanning a variety of devices and platforms. This segment primarily comprises:\n• Microsoft 365 Commercial products and cloud services, including Microsoft 365 Commercial cloud, comprising Microsoft 365 Commercial, Enterprise Mobility + Security, the cloud portion of Windows Commercial, the per-user portion of Power BI, Exchange, SharePoint, Microsoft Teams, Microsoft 365 Security and Compliance, and Microsoft 365 Copilot; and Microsoft 365 Commercial products, comprising Windows Commercial on-premises and Office licensed on-premises.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 43", + "metadata": { + "length": 4437, + "summary": "The passage covers stock plan activity for fiscal year 2025, including nonvested stock awards, unrecognized compensation costs, and the Employee Stock Purchase Plan (ESPP) with share purchases and average prices. It also details savings plans and employer-funded retirement benefits. Additionally, it introduces segment information for Productivity and Business Processes, Intelligent Cloud, and More Personal Computing, with operating income as the key performance measure.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "The", + "table", + "shows", + "quarterly", + "dividends", + "share", + "interest", + "rates", + "fiscal", + "years", + "2023", + "2025", + "Dividends", + "range", + "0.62", + "0.83", + "2.0%", + "5.6%", + "During", + "year", + "activity", + "occurred", + "stock", + "plans", + "nonvested", + "award", + "beginning", + "balance", + "88", + "shares", + "292.28", + "39", + "granted", + "413.90", + "38", + "vested", + "293.25", + "forfeited", + "317.23", + "ending", + "82", + "347.44", + "Includes", + "million", + "PSUs", + "target", + "performance", + "adjustments", + "levels", + "2024", + "As", + "June", + "30", + "total", + "unrecognized", + "compensation", + "costs", + "related", + "awards", + "21.6", + "billion", + "These", + "expected", + "recognized", + "weighted", + "average", + "period", + "grant", + "date", + "fair", + "339.46", + "252.59", + "16.2", + "16.0", + "11.9", + "aggregate", + "98", + "authorized", + "future", + "Employee", + "Stock", + "Purchase", + "Plan", + "We", + "ESPP", + "eligible", + "employees", + "Shares", + "common", + "purchased", + "month", + "intervals", + "90%", + "market", + "trading", + "day", + "Employees", + "purchase", + "exceeding", + "15%", + "gross", + "offering", + "periods", + "presented", + "price", + "Average", + "prices", + "385.10", + "245.59", + "62", + "reserved", + "issuance", + "Savings", + "Plans", + "savings", + "qualify", + "Section", + "401", + "Internal", + "Revenue", + "Code", + "number", + "international", + "locations", + "Eligible", + "contribute", + "portion", + "salary", + "subject", + "limitations", + "match", + "dollar", + "participant", + "contributes", + "Employer", + "funded", + "retirement", + "benefits", + "1.8", + "1.7", + "1.6", + "expensed", + "contributed", + "NOTE", + "18", + "SEGMENT", + "INFORMATION", + "AND", + "GEOGRAPHIC", + "DATA", + "In", + "operation", + "business", + "management", + "including", + "chief", + "operating", + "decision", + "maker", + "CODM", + "Chief", + "Executive", + "Officer", + "reviews", + "financial", + "information", + "segmented", + "internal", + "profit", + "loss", + "statements", + "primary", + "profitability", + "measure", + "review", + "segment", + "results", + "income", + "allocate", + "resources", + "annual", + "planning", + "process", + "assess", + "segments", + "primarily", + "monitoring", + "actual", + "compared", + "prior", + "reported", + "based", + "Productivity", + "Business", + "Processes", + "Intelligent", + "Cloud", + "More", + "Personal", + "Computing", + "recast", + "amounts", + "conform", + "internally", + "manage", + "monitor", + "Refer", + "Note", + "Accounting", + "Policies", + "Our", + "reportable", + "consists", + "products", + "services", + "portfolio", + "productivity", + "communication", + "spanning", + "variety", + "devices", + "platforms", + "This", + "comprises", + "Microsoft", + "365", + "Commercial", + "cloud", + "comprising", + "Enterprise", + "Mobility", + "Security", + "Windows", + "user", + "Power", + "BI", + "Exchange", + "SharePoint", + "Teams", + "Compliance", + "Copilot", + "premises", + "Office", + "licensed" + ], + "keywords": [ + "Stock awards", + "ESPP", + "Segments" + ], + "connect_to": [ + { + "target": "309e74c6-6705-520b-a28b-959fb9661644", + "relation": "embeds", + "ref": "[tables/table-60 Dividends & Rates.html]", + "position": { + "start": 174, + "end": 214 + } + }, + { + "target": "792d3955-3e4a-56a4-8154-2e6d38d6218b", + "relation": "embeds", + "ref": "[tables/table-61 Stock Awards.html]", + "position": { + "start": 491, + "end": 526 + } + }, + { + "target": "4e74811f-7ba5-517d-8687-fc6523746a50", + "relation": "embeds", + "ref": "[tables/table-62 Share Repurchase.html]", + "position": { + "start": 1943, + "end": 1982 + } + } + ] + } + }, + { + "chunk_id": "18761fc9-9905-5fb7-ac0d-39240265b989", + "type": "text", + "content": "• Microsoft 365 Consumer products and cloud services, including Microsoft 365 Consumer subscriptions, Office licensed on-premises, and other consumer services.\n• LinkedIn, including Talent Solutions, Marketing Solutions, Premium Subscriptions, and Sales Solutions.\n• Dynamics products and cloud services, including Dynamics 365, comprising a set of intelligent, cloud-based applications across ERP, CRM, Power Apps, and Power Automate; and on-premises ERP and CRM applications.\nIntelligent Cloud\nOur Intelligent Cloud segment consists of our public, private, and hybrid server products and cloud services that power modern business and developers. This segment primarily comprises:\n• Server products and cloud services, including Azure and other cloud services, comprising cloud and AI consumption-based services, GitHub cloud services, Nuance Healthcare cloud services, virtual desktop offerings, and other cloud services; and Server products, comprising SQL Server, Windows Server, Visual Studio, System Center, related Client Access Licenses (“CALs”), and other on-premises offerings.\n• Enterprise and partner services, including Enterprise Support Services, Industry Solutions, Nuance professional services, Microsoft Partner Network, and Learning Experience.\nMore Personal Computing\nOur More Personal Computing segment consists of products and services that put customers at the center of the experience with our technology. This segment primarily comprises:\n• Windows and Devices, including Windows OEM licensing (Windows Pro and non-Pro licenses sold through the OEM channel) and Devices, comprising Surface and PC accessories.\n• Gaming, including Xbox hardware and Xbox content and services, comprising first- and third-party content (including games and in-game content), Xbox Game Pass and other subscriptions, Xbox Cloud Gaming, advertising, and other cloud services.\n• Search and news advertising, comprising Bing and Copilot, Microsoft News, Microsoft Edge, and third-party affiliates.\nRevenue and costs are generally directly attributed to our segments. However, due to the integrated structure of our business, certain revenue recognized and costs incurred by one segment may benefit other segments. Revenue from certain contracts is allocated among the segments based on the relative value of the underlying products and services, which can include allocation based on actual prices charged, prices when sold separately, or estimated costs plus a profit margin. Cost of revenue is allocated in certain cases based on a relative revenue methodology. Operating expenses that are allocated primarily include those relating to marketing of products and services from which multiple segments benefit and are generally allocated based on relative gross margin.\nIn addition, certain costs are incurred at a corporate level and allocated to our segments. These allocated costs generally include legal, including settlements and fines, information technology, human resources, finance, excise taxes, field selling, shared facilities services, customer service and support, and severance incurred as part of a corporate program. Each allocation is measured differently based on the specific facts and circumstances of the costs being allocated and is generally based on relative gross margin or relative headcount.\nSegment revenue, cost of revenue, operating expenses, and operating income were as follows during the periods presented:\n\nThe table shows revenue, cost of revenue, operating expenses, and operating income for three Microsoft segments (Productivity, Intelligent Cloud, More Personal Computing) and totals for fiscal years 2023-2025.\n[tables/table-63 Microsoft Segment Results.html]\n\nNo sales to an individual customer or country other than the United States accounted for more than 10% of revenue for fiscal years 2025, 2024, or 2023. Revenue, classified by the major geographic areas in which our customers were located, was as follows:\n\nThe table shows total revenue of $281,724 million in 2025, $245,122 million in 2024, and $211,915 million in 2023, split between United States and other countries.\n[tables/table-64 Revenue by Region.html]\n\n(a) Includes billings to OEMs and certain multinational organizations because of the nature of these businesses and the impracticability of determining the geographic source of the revenue.\nRevenue, classified by significant product and service offerings, was as follows:", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 44", + "metadata": { + "length": 4436, + "summary": "The text describes Microsoft's three business segments: Productivity and Business Processes (including Microsoft 365, LinkedIn, Dynamics), Intelligent Cloud (including Azure, server products, enterprise services), and More Personal Computing (including Windows, Gaming, Search). It explains revenue and cost allocation methods among segments, including direct attribution and allocation based on relative gross margin or headcount. Financial data for fiscal years 2023-2025 is presented in tables, showing total revenue of $281,724 million in 2025, $245,122 million in 2024, and $211,915 million in 2023, split between the United States and other countries. No single customer or country (except the US) accounted for more than 10% of revenue.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Microsoft", + "365", + "Consumer", + "products", + "cloud", + "services", + "including", + "subscriptions", + "Office", + "licensed", + "premises", + "consumer", + "LinkedIn", + "Talent", + "Solutions", + "Marketing", + "Premium", + "Subscriptions", + "Sales", + "Dynamics", + "comprising", + "set", + "intelligent", + "based", + "applications", + "ERP", + "CRM", + "Power", + "Apps", + "Automate", + "Intelligent", + "Cloud", + "Our", + "segment", + "consists", + "public", + "private", + "hybrid", + "server", + "power", + "modern", + "business", + "developers", + "This", + "primarily", + "comprises", + "Server", + "Azure", + "AI", + "consumption", + "GitHub", + "Nuance", + "Healthcare", + "virtual", + "desktop", + "offerings", + "SQL", + "Windows", + "Visual", + "Studio", + "System", + "Center", + "related", + "Client", + "Access", + "Licenses", + "CALs", + "Enterprise", + "partner", + "Support", + "Services", + "Industry", + "professional", + "Partner", + "Network", + "Learning", + "Experience", + "More", + "Personal", + "Computing", + "put", + "customers", + "center", + "experience", + "technology", + "Devices", + "OEM", + "licensing", + "Pro", + "licenses", + "sold", + "channel", + "Surface", + "PC", + "accessories", + "Gaming", + "Xbox", + "hardware", + "content", + "party", + "games", + "game", + "Game", + "Pass", + "advertising", + "Search", + "news", + "Bing", + "Copilot", + "News", + "Edge", + "affiliates", + "Revenue", + "costs", + "generally", + "directly", + "attributed", + "segments", + "However", + "due", + "integrated", + "structure", + "revenue", + "recognized", + "incurred", + "benefit", + "contracts", + "allocated", + "relative", + "underlying", + "include", + "allocation", + "actual", + "prices", + "charged", + "separately", + "estimated", + "profit", + "margin", + "Cost", + "cases", + "methodology", + "Operating", + "expenses", + "relating", + "marketing", + "multiple", + "gross", + "In", + "addition", + "corporate", + "level", + "These", + "legal", + "settlements", + "fines", + "information", + "human", + "resources", + "finance", + "excise", + "taxes", + "field", + "selling", + "shared", + "facilities", + "customer", + "service", + "support", + "severance", + "part", + "program", + "Each", + "measured", + "differently", + "specific", + "facts", + "circumstances", + "headcount", + "Segment", + "cost", + "operating", + "income", + "periods", + "presented", + "The", + "table", + "shows", + "Productivity", + "totals", + "fiscal", + "years", + "2023", + "2025", + "No", + "sales", + "individual", + "country", + "United", + "States", + "accounted", + "10%", + "2024", + "classified", + "major", + "geographic", + "areas", + "located", + "total", + "281", + "724", + "million", + "245", + "122", + "211", + "915", + "split", + "countries", + "Includes", + "billings", + "OEMs", + "multinational", + "organizations", + "nature", + "businesses", + "impracticability", + "determining", + "source", + "significant", + "product" + ], + "keywords": [ + "Segments", + "Revenue", + "Cloud" + ], + "connect_to": [ + { + "target": "0579405f-2c41-52d7-aca9-c24a01e2a411", + "relation": "embeds", + "ref": "[tables/table-63 Microsoft Segment Results.html]", + "position": { + "start": 3653, + "end": 3701 + } + }, + { + "target": "e3d92df6-6a6e-53d0-abb3-1136d50f5ae9", + "relation": "embeds", + "ref": "[tables/table-64 Revenue by Region.html]", + "position": { + "start": 4123, + "end": 4163 + } + } + ] + } + }, + { + "chunk_id": "17d0e47d-6058-5b0c-81a0-e0660583c3d8", + "type": "text", + "content": "Microsoft's revenue for fiscal years 2025, 2024, and 2023 is shown, with totals of $281,724M, $245,122M, and $211,915M respectively, led by Server products and cloud services.\n[tables/table-65 Microsoft Revenue.html]\n\nOur Microsoft Cloud revenue, which includes Microsoft 365 Commercial cloud, Azure and other cloud services, the commercial portion of LinkedIn, and Dynamics 365, was $168.9 billion, $137.7 billion, and $111.6 billion in fiscal years 2025, 2024, and 2023, respectively. These amounts are included in Microsoft 365 Commercial products and cloud services, Server products and cloud services, LinkedIn, and Dynamics products and cloud services in the table above.\nAssets are not allocated to segments for internal reporting presentations. A portion of amortization and depreciation is included with various other costs in an overhead allocation to each segment. It is impracticable for us to separately identify the amount of amortization and depreciation by segment that is included in the measure of segment profit or loss.\nLong-lived assets, excluding financial instruments and tax assets, classified by the location of the controlling statutory company and with countries over 10% of the total shown separately, were as follows:\n\nThe table shows total revenue of $371,902 million in June 2025, up from $301,369 million in 2024 and $187,239 million in 2023, with United States and Other countries contributions.\n[tables/table-66 Revenue by Region.html]\n\nREPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM\nTo the Stockholders and the Board of Directors of Microsoft Corporation\nOpinion on the Financial Statements\nWe have audited the accompanying consolidated balance sheets of Microsoft Corporation and subsidiaries (the “Company”) as of June 30, 2025 and 2024, the related consolidated statements of income, comprehensive income, cash flows, and stockholders’ equity, for each of the three years in the period ended June 30, 2025, and the related notes (collectively referred to as the “financial statements”). In our opinion, the financial statements present fairly, in all material respects, the financial position of the Company as of June 30, 2025 and 2024, and the results of its operations and its cash flows for each of the three years in the period ended June 30, 2025, in conformity with accounting principles generally accepted in the United States of America.\nWe have also audited, in accordance with the standards of the Public Company Accounting Oversight Board (United States) (PCAOB), the Company’s internal control over financial reporting as of June 30, 2025, based on criteria established in Internal Control — Integrated Framework (2013) issued by the Committee of Sponsoring Organizations of the Treadway Commission and our report dated July 30, 2025, expressed an unqualified opinion on the Company’s internal control over financial reporting.\nBasis for Opinion\nThese financial statements are the responsibility of the Company’s management. Our responsibility is to express an opinion on the Company’s financial statements based on our audits. We are a public accounting firm registered with the PCAOB and are required to be independent with respect to the Company in accordance with the U.S. federal securities laws and the applicable rules and regulations of the Securities and Exchange Commission and the PCAOB.\nWe conducted our audits in accordance with the standards of the PCAOB. Those standards require that we plan and perform the audit to obtain reasonable assurance about whether the financial statements are free of material misstatement, whether due to error or fraud. Our audits included performing procedures to assess the risks of material misstatement of the financial statements, whether due to error or fraud, and performing procedures that respond to those risks. Such procedures included examining, on a test basis, evidence regarding the amounts and disclosures in the financial statements. Our audits also included evaluating the accounting principles used and significant estimates made by management, as well as evaluating the overall presentation of the financial statements. We believe that our audits provide a reasonable basis for our opinion.\nCritical Audit Matters", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 45", + "metadata": { + "length": 4238, + "summary": "Microsoft's fiscal 2025 revenue reached $281,724M, up from $245,122M in 2024 and $211,915M in 2023, led by Server products and cloud services. Microsoft Cloud revenue was $168.9 billion in 2025, compared to $137.7 billion in 2024 and $111.6 billion in 2023. Total revenue including other items was $371,902 million in June 2025. The report also includes an independent auditor's opinion confirming the financial statements are fairly presented in accordance with U.S. GAAP, with an unqualified opinion on internal control over financial reporting as of June 30, 2025.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Microsoft", + "revenue", + "fiscal", + "years", + "2025", + "2024", + "2023", + "shown", + "totals", + "281", + "724M", + "245", + "122M", + "211", + "915M", + "led", + "Server", + "products", + "cloud", + "services", + "Our", + "Cloud", + "includes", + "365", + "Commercial", + "Azure", + "commercial", + "portion", + "LinkedIn", + "Dynamics", + "168.9", + "billion", + "137.7", + "111.6", + "These", + "amounts", + "included", + "table", + "Assets", + "allocated", + "segments", + "internal", + "reporting", + "presentations", + "amortization", + "depreciation", + "costs", + "overhead", + "allocation", + "segment", + "It", + "impracticable", + "separately", + "identify", + "amount", + "measure", + "profit", + "loss", + "Long", + "lived", + "assets", + "excluding", + "financial", + "instruments", + "tax", + "classified", + "location", + "controlling", + "statutory", + "company", + "countries", + "10%", + "total", + "The", + "shows", + "371", + "902", + "million", + "June", + "301", + "369", + "187", + "239", + "United", + "States", + "Other", + "contributions", + "REPORT", + "OF", + "INDEPENDENT", + "REGISTERED", + "PUBLIC", + "ACCOUNTING", + "FIRM", + "To", + "Stockholders", + "Board", + "Directors", + "Corporation", + "Opinion", + "Financial", + "Statements", + "We", + "audited", + "accompanying", + "consolidated", + "balance", + "sheets", + "subsidiaries", + "Company", + "30", + "related", + "statements", + "income", + "comprehensive", + "cash", + "flows", + "stockholders", + "equity", + "period", + "ended", + "notes", + "collectively", + "referred", + "In", + "opinion", + "present", + "fairly", + "material", + "respects", + "position", + "results", + "operations", + "conformity", + "accounting", + "principles", + "generally", + "accepted", + "America", + "accordance", + "standards", + "Public", + "Accounting", + "Oversight", + "PCAOB", + "control", + "based", + "criteria", + "established", + "Internal", + "Control", + "Integrated", + "Framework", + "2013", + "issued", + "Committee", + "Sponsoring", + "Organizations", + "Treadway", + "Commission", + "report", + "dated", + "July", + "expressed", + "unqualified", + "Basis", + "responsibility", + "management", + "express", + "audits", + "public", + "firm", + "registered", + "required", + "independent", + "respect", + "federal", + "securities", + "laws", + "applicable", + "rules", + "regulations", + "Securities", + "Exchange", + "conducted", + "Those", + "require", + "plan", + "perform", + "audit", + "obtain", + "reasonable", + "assurance", + "free", + "misstatement", + "due", + "error", + "fraud", + "performing", + "procedures", + "assess", + "risks", + "respond", + "Such", + "examining", + "test", + "basis", + "evidence", + "disclosures", + "evaluating", + "significant", + "estimates", + "made", + "presentation", + "provide", + "Critical", + "Audit", + "Matters" + ], + "keywords": [ + "Revenue", + "Cloud", + "Audit" + ], + "connect_to": [ + { + "target": "89bdd525-3b5c-53e1-947e-df3aefb5bdb9", + "relation": "embeds", + "ref": "[tables/table-65 Microsoft Revenue.html]", + "position": { + "start": 176, + "end": 216 + } + }, + { + "target": "f84caf3c-647d-582d-a0b5-ddbab86fdb05", + "relation": "embeds", + "ref": "[tables/table-66 Revenue by Region.html]", + "position": { + "start": 1429, + "end": 1469 + } + } + ] + } + }, + { + "chunk_id": "4627b947-6452-598b-bc09-a09b07c70bf3", + "type": "text", + "content": "The critical audit matters communicated below are matters arising from the current-period audit of the financial statements that were communicated or required to be communicated to the audit committee and that (1) relate to accounts or disclosures that are material to the financial statements and (2) involved our especially challenging, subjective, or complex judgments. The communication of critical audit matters does not alter in any way our opinion on the financial statements, taken as a whole, and we are not, by communicating the critical audit matters below, providing separate opinions on the critical audit matters or on the accounts or disclosures to which they relate.\nRevenue Recognition – Refer to Note 1 to the financial statements\nCritical Audit Matter Description\nThe Company recognizes revenue upon transfer of control of promised products or services to customers in an amount that reflects the consideration the Company expects to receive in exchange for those products or services. The Company offers customers the ability to acquire multiple licenses of software products and services, including cloud-based services, in its customer agreements through its volume licensing programs.\nSignificant judgment is exercised by the Company in determining revenue recognition for certain customer agreements, and includes the following:\n• Determination of whether products and services are considered distinct performance obligations that should be accounted for separately versus together, such as software licenses and related services that are sold with cloud-based services.\n• The pattern of delivery (i.e., timing of when revenue is recognized) for each distinct performance obligation.\n• Identification and treatment of contract terms that may impact the timing and amount of revenue recognized (e.g., variable consideration, optional purchases, and free services).\n• Determination of stand-alone selling prices for each distinct performance obligation and for products and services that are not sold separately.\nGiven these factors and due to the volume of transactions, the related audit effort in evaluating management’s judgments in determining revenue recognition for certain customer agreements was extensive and required a high degree of auditor judgment.\nHow the Critical Audit Matter Was Addressed in the Audit\nOur principal audit procedures related to the Company’s revenue recognition for certain customer agreements included the following:\n• We tested the effectiveness of controls related to the identification of distinct performance obligations, the determination of the timing of revenue recognition, and the estimation of variable consideration.\n• We evaluated management’s significant accounting policies related to certain customer agreements for reasonableness.\n• We selected a sample of customer agreements and performed the following procedures:\n Obtained and read contract source documents for each selection, including master agreements, and other documents that were part of the agreement.\n Tested management’s identification and treatment of contract terms.\n Assessed the terms in the customer agreement and evaluated the appropriateness of management’s application of their accounting policies, along with their use of estimates, in the determination of revenue recognition conclusions.\n• We evaluated the reasonableness of management’s estimate of stand-alone selling prices for products and services that are not sold separately.\n• We tested the mathematical accuracy of management’s calculations of revenue and the associated timing of revenue recognized in the financial statements.\nIncome Taxes – Uncertain Tax Positions – Refer to Note 11 to the financial statements\nCritical Audit Matter Description\nThe Company’s long-term income taxes liability includes uncertain tax positions related to transfer pricing issues that remain unresolved with the Internal Revenue Service (“IRS”). The Company remains under IRS audit, or subject to IRS audit, for tax years subsequent to 2003. During fiscal year 2024, the Company received Notices of Proposed Adjustments (“NOPAs”) for the tax years 2004 to 2013, primarily related to intercompany transfer pricing. While the Company has settled a portion of the IRS audits, resolution of the remaining matters could have a material impact on the Company’s financial statements.", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 46", + "metadata": { + "length": 4371, + "summary": "The passage describes two critical audit matters (CAMs) from a financial statement audit: Revenue Recognition and Income Taxes (Uncertain Tax Positions). For Revenue Recognition, the company recognizes revenue upon transfer of control, involving significant judgment on distinct performance obligations, delivery patterns, contract terms, and stand-alone selling prices. Audit procedures included testing controls, evaluating policies, and sampling contracts. For Income Taxes, the company faces uncertain tax positions related to transfer pricing issues unresolved with the IRS, with potential material impact from remaining audits.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "The", + "critical", + "audit", + "matters", + "communicated", + "arising", + "current", + "period", + "financial", + "statements", + "required", + "committee", + "relate", + "accounts", + "disclosures", + "material", + "involved", + "challenging", + "subjective", + "complex", + "judgments", + "communication", + "alter", + "opinion", + "communicating", + "providing", + "separate", + "opinions", + "Revenue", + "Recognition", + "Refer", + "Note", + "Critical", + "Audit", + "Matter", + "Description", + "Company", + "recognizes", + "revenue", + "transfer", + "control", + "promised", + "products", + "services", + "customers", + "amount", + "reflects", + "consideration", + "expects", + "receive", + "exchange", + "offers", + "ability", + "acquire", + "multiple", + "licenses", + "software", + "including", + "cloud", + "based", + "customer", + "agreements", + "volume", + "licensing", + "programs", + "Significant", + "judgment", + "exercised", + "determining", + "recognition", + "includes", + "Determination", + "considered", + "distinct", + "performance", + "obligations", + "accounted", + "separately", + "versus", + "related", + "sold", + "pattern", + "delivery", + "timing", + "recognized", + "obligation", + "Identification", + "treatment", + "contract", + "terms", + "impact", + "variable", + "optional", + "purchases", + "free", + "stand", + "selling", + "prices", + "Given", + "factors", + "due", + "transactions", + "effort", + "evaluating", + "management", + "extensive", + "high", + "degree", + "auditor", + "How", + "Was", + "Addressed", + "Our", + "principal", + "procedures", + "included", + "We", + "tested", + "effectiveness", + "controls", + "identification", + "determination", + "estimation", + "evaluated", + "significant", + "accounting", + "policies", + "reasonableness", + "selected", + "sample", + "performed", + "Obtained", + "read", + "source", + "documents", + "selection", + "master", + "part", + "agreement", + "Tested", + "Assessed", + "appropriateness", + "application", + "estimates", + "conclusions", + "estimate", + "mathematical", + "accuracy", + "calculations", + "Income", + "Taxes", + "Uncertain", + "Tax", + "Positions", + "11", + "long", + "term", + "income", + "taxes", + "liability", + "uncertain", + "tax", + "positions", + "pricing", + "issues", + "remain", + "unresolved", + "Internal", + "Service", + "IRS", + "remains", + "subject", + "years", + "subsequent", + "2003", + "During", + "fiscal", + "year", + "2024", + "received", + "Notices", + "Proposed", + "Adjustments", + "NOPAs", + "2004", + "2013", + "primarily", + "intercompany", + "While", + "settled", + "portion", + "audits", + "resolution", + "remaining" + ], + "keywords": [ + "Revenue Recognition", + "Income Taxes", + "Audit" + ], + "connect_to": [] + } + }, + { + "chunk_id": "3ffc9122-2c81-530a-9248-858f70d3adbd", + "type": "text", + "content": "Conclusions on recognizing and measuring uncertain tax positions involve significant estimates and management judgment and include complex considerations of the Internal Revenue Code, related regulations, tax case laws, and prior-year audit settlements. Given the complexity and the subjective nature of certain transfer pricing issues that remain unresolved with the IRS, evaluating management’s estimates relating to their determination of uncertain tax positions required a high degree of auditor judgment, including involvement of our tax specialists.\nHow the Critical Audit Matter Was Addressed in the Audit\nOur principal audit procedures to evaluate management’s estimates of uncertain tax positions related to unresolved transfer pricing issues included the following:\n• We evaluated management’s methods and assumptions used in the measurement and disclosure of uncertain tax positions, which included testing the effectiveness of the related internal controls.\n• We tested the reasonableness of management’s judgments regarding the future resolution of uncertain tax positions, as follows:\n We evaluated whether management had appropriately considered new information that could significantly change the measurement of the uncertain tax positions.\n We evaluated the reasonableness of management’s estimates by considering how changes in tax law, including statutes, regulations, and recent case law, impacted management’s judgments.\n• We evaluated the appropriateness of the disclosures in relation to the underlying facts, judgments, and conclusions.\n/S/ DELOITTE & TOUCHE LLP\nSeattle, Washington\nJuly 30, 2025\nWe have served as the Company’s auditor since 1983.\nCHANGES IN AND DISAGREEMENTS WITH ACCOUNTANTS ON ACCOUNTING AND FINANCIAL DISCLOSURE\nNot applicable.\nCONTROLS AND PROCEDURES\nUnder the supervision and with the participation of our management, including the Chief Executive Officer and Chief Financial Officer, we have evaluated the effectiveness of our disclosure controls and procedures as required by Exchange Act Rule 13a-15(b) as of the end of the period covered by this report. Based on that evaluation, the Chief Executive Officer and Chief Financial Officer have concluded that these disclosure controls and procedures are effective.\nREPORT OF MANAGEMENT ON INTERNAL CONTROL OVER FINANCIAL REPORTING\nOur management is responsible for establishing and maintaining adequate internal control over financial reporting for the Company. Internal control over financial reporting is a process to provide reasonable assurance regarding the reliability of our financial reporting for external purposes in accordance with accounting principles generally accepted in the United States of America. Internal control over financial reporting includes maintaining records that in reasonable detail accurately and fairly reflect our transactions; providing reasonable assurance that transactions are recorded as necessary for preparation of our consolidated financial statements; providing reasonable assurance that receipts and expenditures of company assets are made in accordance with management authorization; and providing reasonable assurance that unauthorized acquisition, use, or disposition of company assets that could have a material effect on our consolidated financial statements would be prevented or detected on a timely basis. Because of its inherent limitations, internal control over financial reporting is not intended to provide absolute assurance that a misstatement of our consolidated financial statements would be prevented or detected.\nManagement conducted an evaluation of the effectiveness of our internal control over financial reporting based on the framework in Internal Control – Integrated Framework (2013) issued by the Committee of Sponsoring Organizations of the Treadway Commission. Based on this evaluation, management concluded that the Company’s internal control over financial reporting was effective as of June 30, 2025. There were no changes in our internal control over financial reporting during the quarter ended June 30, 2025 that have materially affected, or are reasonably likely to materially affect, our internal control over financial reporting. Deloitte & Touche LLP has audited our internal control over financial reporting as of June 30, 2025; their report follows.\nREPORT OF INDEPENDENT REGISTERED PUBLIC ACCOUNTING FIRM\nTo the Stockholders and the Board of Directors of Microsoft Corporation\nOpinion on Internal Control over Financial Reporting", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 47", + "metadata": { + "length": 4516, + "summary": "The passage discusses the critical audit matter of recognizing and measuring uncertain tax positions, focusing on unresolved transfer pricing issues. It highlights the complexity and subjective nature of these estimates, requiring significant auditor judgment and involvement of tax specialists. Audit procedures included evaluating management's methods, testing internal controls, assessing reasonableness of judgments regarding future resolution, and considering changes in tax law. The report also covers management's evaluation of disclosure controls and internal control over financial reporting, both deemed effective as of June 30, 2025, with Deloitte & Touche LLP providing the audit opinion.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "Conclusions", + "recognizing", + "measuring", + "uncertain", + "tax", + "positions", + "involve", + "significant", + "estimates", + "management", + "judgment", + "include", + "complex", + "considerations", + "Internal", + "Revenue", + "Code", + "related", + "regulations", + "case", + "laws", + "prior", + "year", + "audit", + "settlements", + "Given", + "complexity", + "subjective", + "nature", + "transfer", + "pricing", + "issues", + "remain", + "unresolved", + "IRS", + "evaluating", + "relating", + "determination", + "required", + "high", + "degree", + "auditor", + "including", + "involvement", + "specialists", + "How", + "Critical", + "Audit", + "Matter", + "Was", + "Addressed", + "Our", + "principal", + "procedures", + "evaluate", + "included", + "We", + "evaluated", + "methods", + "assumptions", + "measurement", + "disclosure", + "testing", + "effectiveness", + "internal", + "controls", + "tested", + "reasonableness", + "judgments", + "future", + "resolution", + "appropriately", + "considered", + "information", + "significantly", + "change", + "law", + "statutes", + "recent", + "impacted", + "appropriateness", + "disclosures", + "relation", + "underlying", + "facts", + "conclusions", + "DELOITTE", + "TOUCHE", + "LLP", + "Seattle", + "Washington", + "July", + "30", + "2025", + "served", + "Company", + "1983", + "CHANGES", + "IN", + "AND", + "DISAGREEMENTS", + "WITH", + "ACCOUNTANTS", + "ON", + "ACCOUNTING", + "FINANCIAL", + "DISCLOSURE", + "Not", + "applicable", + "CONTROLS", + "PROCEDURES", + "Under", + "supervision", + "participation", + "Chief", + "Executive", + "Officer", + "Financial", + "Exchange", + "Act", + "Rule", + "13a", + "15", + "end", + "period", + "covered", + "report", + "Based", + "evaluation", + "concluded", + "effective", + "REPORT", + "OF", + "MANAGEMENT", + "INTERNAL", + "CONTROL", + "OVER", + "REPORTING", + "responsible", + "establishing", + "maintaining", + "adequate", + "control", + "financial", + "reporting", + "process", + "provide", + "reasonable", + "assurance", + "reliability", + "external", + "purposes", + "accordance", + "accounting", + "principles", + "generally", + "accepted", + "United", + "States", + "America", + "includes", + "records", + "detail", + "accurately", + "fairly", + "reflect", + "transactions", + "providing", + "recorded", + "preparation", + "consolidated", + "statements", + "receipts", + "expenditures", + "company", + "assets", + "made", + "authorization", + "unauthorized", + "acquisition", + "disposition", + "material", + "effect", + "prevented", + "detected", + "timely", + "basis", + "Because", + "inherent", + "limitations", + "intended", + "absolute", + "misstatement", + "Management", + "conducted", + "based", + "framework", + "Control", + "Integrated", + "Framework", + "2013", + "issued", + "Committee", + "Sponsoring", + "Organizations", + "Treadway", + "Commission", + "June", + "There", + "quarter", + "ended", + "materially", + "affected", + "affect", + "Deloitte", + "Touche", + "audited", + "INDEPENDENT", + "REGISTERED", + "PUBLIC", + "FIRM", + "To", + "Stockholders", + "Board", + "Directors", + "Microsoft", + "Corporation", + "Opinion", + "Reporting" + ], + "keywords": [ + "uncertain tax positions", + "auditor judgment", + "transfer pricing" + ], + "connect_to": [] + } + }, + { + "chunk_id": "ef420677-48a9-593f-86bc-2f1112daebcd", + "type": "text", + "content": "We have audited the internal control over financial reporting of Microsoft Corporation and subsidiaries (the “Company”) as of June 30, 2025, based on criteria established in Internal Control — Integrated Framework (2013) issued by the Committee of Sponsoring Organizations of the Treadway Commission (COSO). In our opinion, the Company maintained, in all material respects, effective internal control over financial reporting as of June 30, 2025, based on criteria established in Internal Control — Integrated Framework (2013) issued by COSO.\nWe have also audited, in accordance with the standards of the Public Company Accounting Oversight Board (United States) (PCAOB), the consolidated financial statements as of and for the year ended June 30, 2025, of the Company and our report dated July 30, 2025, expressed an unqualified opinion on those financial statements.\nBasis for Opinion\nThe Company’s management is responsible for maintaining effective internal control over financial reporting and for its assessment of the effectiveness of internal control over financial reporting, included in the accompanying Report of Management on Internal Control over Financial Reporting. Our responsibility is to express an opinion on the Company’s internal control over financial reporting based on our audit. We are a public accounting firm registered with the PCAOB and are required to be independent with respect to the Company in accordance with the U.S. federal securities laws and the applicable rules and regulations of the Securities and Exchange Commission and the PCAOB.\nWe conducted our audit in accordance with the standards of the PCAOB. Those standards require that we plan and perform the audit to obtain reasonable assurance about whether effective internal control over financial reporting was maintained in all material respects. Our audit included obtaining an understanding of internal control over financial reporting, assessing the risk that a material weakness exists, testing and evaluating the design and operating effectiveness of internal control based on the assessed risk, and performing such other procedures as we considered necessary in the circumstances. We believe that our audit provides a reasonable basis for our opinion.\nDefinition and Limitations of Internal Control over Financial Reporting\nA company’s internal control over financial reporting is a process designed to provide reasonable assurance regarding the reliability of financial reporting and the preparation of financial statements for external purposes in accordance with generally accepted accounting principles. A company’s internal control over financial reporting includes those policies and procedures that (1) pertain to the maintenance of records that, in reasonable detail, accurately and fairly reflect the transactions and dispositions of the assets of the company; (2) provide reasonable assurance that transactions are recorded as necessary to permit preparation of financial statements in accordance with generally accepted accounting principles, and that receipts and expenditures of the company are being made only in accordance with authorizations of management and directors of the company; and (3) provide reasonable assurance regarding prevention or timely detection of unauthorized acquisition, use, or disposition of the company’s assets that could have a material effect on the financial statements.\nBecause of its inherent limitations, internal control over financial reporting may not prevent or detect misstatements. Also, projections of any evaluation of effectiveness to future periods are subject to the risk that controls may become inadequate because of changes in conditions, or that the degree of compliance with the policies or procedures may deteriorate.\n\nA signature block for Deloitte & Touche LLP, located in Seattle, Washington, dated July 30, 2025.\n[tables/table-67 Deloitte Signature.html]\n\nDIRECTORS AND EXECUTIVE OFFICERS OF MICROSOFT CORPORATION\nDIRECTORS\n\nThe table lists Microsoft Corporation's board members including Satya Nadella, Reid Hoffman, Hugh Johnston, and others with their roles and affiliations.\n[tables/table-68 Microsoft Board.html]\n\nBoard Committees", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 48", + "metadata": { + "length": 4205, + "summary": "The text is an audit report on Microsoft Corporation's internal control over financial reporting as of June 30, 2025. Deloitte & Touche LLP expresses an unqualified opinion that the company maintained effective internal control in all material respects, based on COSO criteria. The report also notes an unqualified opinion on the consolidated financial statements. It describes the basis for the opinion, the responsibilities of management and auditors, and the definition and limitations of internal control over financial reporting. The document includes a signature block dated July 30, 2025, and lists board members and executive officers.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "We", + "audited", + "internal", + "control", + "financial", + "reporting", + "Microsoft", + "Corporation", + "subsidiaries", + "Company", + "June", + "30", + "2025", + "based", + "criteria", + "established", + "Internal", + "Control", + "Integrated", + "Framework", + "2013", + "issued", + "Committee", + "Sponsoring", + "Organizations", + "Treadway", + "Commission", + "COSO", + "In", + "opinion", + "maintained", + "material", + "respects", + "effective", + "accordance", + "standards", + "Public", + "Accounting", + "Oversight", + "Board", + "United", + "States", + "PCAOB", + "consolidated", + "statements", + "year", + "ended", + "report", + "dated", + "July", + "expressed", + "unqualified", + "Basis", + "Opinion", + "The", + "management", + "responsible", + "maintaining", + "assessment", + "effectiveness", + "included", + "accompanying", + "Report", + "Management", + "Financial", + "Reporting", + "Our", + "responsibility", + "express", + "audit", + "public", + "accounting", + "firm", + "registered", + "required", + "independent", + "respect", + "federal", + "securities", + "laws", + "applicable", + "rules", + "regulations", + "Securities", + "Exchange", + "conducted", + "Those", + "require", + "plan", + "perform", + "obtain", + "reasonable", + "assurance", + "obtaining", + "understanding", + "assessing", + "risk", + "weakness", + "exists", + "testing", + "evaluating", + "design", + "operating", + "assessed", + "performing", + "procedures", + "considered", + "circumstances", + "basis", + "Definition", + "Limitations", + "company", + "process", + "designed", + "provide", + "reliability", + "preparation", + "external", + "purposes", + "generally", + "accepted", + "principles", + "includes", + "policies", + "pertain", + "maintenance", + "records", + "detail", + "accurately", + "fairly", + "reflect", + "transactions", + "dispositions", + "assets", + "recorded", + "permit", + "receipts", + "expenditures", + "made", + "authorizations", + "directors", + "prevention", + "timely", + "detection", + "unauthorized", + "acquisition", + "disposition", + "effect", + "Because", + "inherent", + "limitations", + "prevent", + "detect", + "misstatements", + "Also", + "projections", + "evaluation", + "future", + "periods", + "subject", + "controls", + "inadequate", + "conditions", + "degree", + "compliance", + "deteriorate", + "signature", + "block", + "Deloitte", + "Touche", + "LLP", + "located", + "Seattle", + "Washington", + "DIRECTORS", + "AND", + "EXECUTIVE", + "OFFICERS", + "OF", + "MICROSOFT", + "CORPORATION", + "table", + "lists", + "board", + "members", + "including", + "Satya", + "Nadella", + "Reid", + "Hoffman", + "Hugh", + "Johnston", + "roles", + "affiliations", + "Committees" + ], + "keywords": [ + "Internal Control", + "Audit", + "Microsoft" + ], + "connect_to": [ + { + "target": "6656b43d-55a4-5cfd-9025-359ff3b95182", + "relation": "embeds", + "ref": "[tables/table-67 Deloitte Signature.html]", + "position": { + "start": 3883, + "end": 3924 + } + }, + { + "target": "6b48ae77-7813-5b62-87f3-ff41c622c790", + "relation": "embeds", + "ref": "[tables/table-68 Microsoft Board.html]", + "position": { + "start": 4149, + "end": 4187 + } + } + ] + } + }, + { + "chunk_id": "c37e894b-93a6-5f38-ad28-59d1539a2e4f", + "type": "text", + "content": "EXECUTIVE OFFICERS\n\nThe table lists Microsoft's top executives including Satya Nadella (CEO), Amy Hood (CFO), Judson Althoff, Takeshi Numoto, Amy Coleman, Bradford Smith, and Kathleen Hogan.\n[tables/table-69 Microsoft Executives.html]\n\nINVESTOR RELATIONS\nInvestor Relations\nYou can contact Microsoft Investor Relations by calling toll-free at (800) 285-7772 or outside the United States, call (425) 706-4400. We can be contacted between the hours of 9:00 a.m. to 5:00 p.m. Pacific Time to answer investment-oriented questions about Microsoft.\nFor access to additional financial information, visit the Investor Relations website online at:\nwww.microsoft.com/investor\nOur e-mail is msft@microsoft.com\nOur mailing address is:\nInvestor Relations\nMicrosoft Corporation\nOne Microsoft Way\nRedmond, Washington 98052-6399\nAttending the Annual Meeting\nThe 2025 Annual Shareholders Meeting will be held as a virtual-only meeting. Any shareholder can join the Annual Meeting, while shareholders of record as of September 30 2025, will be able to vote and submit questions during the meeting.\nDate: Friday, December 5, 2025\nTime: 8:30 a.m. Pacific Time\nVirtual Shareholder Meeting:\nwww.virtualshareholdermeeting.com/MSFT25\nSubmit Your Question\nWe invite you to submit any questions via the proxy voting site at www.proxyvote.com. We will include as many of your questions as possible during the Q&A session of the meeting and will provide answers to questions on the Microsoft Investor Relations website under the Annual Meeting page.\nRegistered Shareholder Services\nComputershare, our transfer agent, can help you with a variety of shareholder related services including:\n• Change of address\n• Lost stock certificates\n• Transfer of stock to another person\n• Additional administrative services\nComputershare also administers a direct stock purchase plan and a dividend reinvestment program for the company.\nContact Computershare directly to find out more about these services and programs at 800-285-7772, option 1, or visit online at:\nhttps://www.computershare.com/Microsoft\nYou can e-mail the transfer agent at:\nweb.queries@computershare.com\nYou can also send mail to the transfer agent at:\nComputershare\nP.O. Box 505000\nLouisville, KY 40233-5000\nShareholders can sign up for electronic alerts to access the annual report and proxy statement online. The service gets you the information you need faster and also gives you the power and convenience of online proxy voting. To sign up for this free service, visit the Annual Report site on the Investor Relations website at:\nhttp://www.microsoft.com/investor/AnnualReports/default.aspx\nEnvironmental, Social, and Governance (ESG)/Corporate Social Responsibility\nMany of our shareholders are focused on the importance of the effective engagement and action on environmental, social, and governance topics. To meet the expectations of our stakeholders and to and maintain their trust, we are committed to conducting our business in ways that are principled, transparent, and accountable. From our CEO and Senior Leadership Team and throughout our organization, people at Microsoft are working to conduct our business in principled ways that make a significant positive impact on important global issues. Microsoft’s Board of Directors provides insight, feedback, and oversight across a broad range of environmental and social matters. In particular, among the responsibilities of the Board’s Environmental, Social, and Public Policy Committee is to review and provide guidance to the Board and management about the Company’s policies and programs that relate to corporate social responsibility.\nFor more about Microsoft’s CSR commitments and performance, please visit:\nwww.microsoft.com/transparency.\n\nCertification Type: MIX (Mixed Sources)\nDescription: Paper | Supporting responsible forestry\nLicense Number: FSC® C132107\nIssuing Authority: FSC (Forest Stewardship Council)\nWebsite: www.fsc.org\n[images/image-4 FSC Certification Labe.jpeg]", + "path": "2025_AnnualReport.docx/4.\tEnvironmental, Social, and Public Policy Committee", + "metadata": { + "length": 3976, + "summary": "The document provides information about Microsoft's executive officers including CEO Satya Nadella and CFO Amy Hood. It details investor relations contact information, the 2025 virtual annual shareholders meeting scheduled for December 5, 2025, and shareholder services provided by Computershare. The document also covers environmental, social, and governance (ESG) commitments, with the Board's Environmental, Social, and Public Policy Committee overseeing corporate social responsibility. Additionally, it includes a certification for mixed sources paper from the Forest Stewardship Council.", + "page_nums": [], + "document_top_summary": "This document includes: **Preface**, 4.\tEnvironmental, Social, and Public Policy Committee", + "tokens": [ + "EXECUTIVE", + "OFFICERS", + "The", + "table", + "lists", + "Microsoft", + "top", + "executives", + "including", + "Satya", + "Nadella", + "CEO", + "Amy", + "Hood", + "CFO", + "Judson", + "Althoff", + "Takeshi", + "Numoto", + "Coleman", + "Bradford", + "Smith", + "Kathleen", + "Hogan", + "INVESTOR", + "RELATIONS", + "Investor", + "Relations", + "You", + "contact", + "calling", + "toll", + "free", + "800", + "285", + "7772", + "United", + "States", + "call", + "425", + "706", + "4400", + "We", + "contacted", + "hours", + "00", + "Pacific", + "Time", + "answer", + "investment", + "oriented", + "questions", + "For", + "access", + "additional", + "financial", + "information", + "visit", + "website", + "online", + "www", + "microsoft", + "investor", + "Our", + "mail", + "msft", + "mailing", + "address", + "Corporation", + "One", + "Way", + "Redmond", + "Washington", + "98052", + "6399", + "Attending", + "Annual", + "Meeting", + "2025", + "Shareholders", + "held", + "virtual", + "meeting", + "Any", + "shareholder", + "join", + "shareholders", + "record", + "September", + "30", + "vote", + "submit", + "Date", + "Friday", + "December", + "Virtual", + "Shareholder", + "virtualshareholdermeeting", + "MSFT25", + "Submit", + "Your", + "Question", + "invite", + "proxy", + "voting", + "site", + "proxyvote", + "include", + "session", + "provide", + "answers", + "page", + "Registered", + "Services", + "Computershare", + "transfer", + "agent", + "variety", + "related", + "services", + "Change", + "Lost", + "stock", + "certificates", + "Transfer", + "person", + "Additional", + "administrative", + "administers", + "direct", + "purchase", + "plan", + "dividend", + "reinvestment", + "program", + "company", + "Contact", + "directly", + "find", + "programs", + "option", + "https", + "computershare", + "web", + "queries", + "send", + "Box", + "505000", + "Louisville", + "KY", + "40233", + "5000", + "sign", + "electronic", + "alerts", + "annual", + "report", + "statement", + "service", + "faster", + "power", + "convenience", + "To", + "Report", + "http", + "AnnualReports", + "default", + "aspx", + "Environmental", + "Social", + "Governance", + "ESG", + "Corporate", + "Responsibility", + "Many", + "focused", + "importance", + "effective", + "engagement", + "action", + "environmental", + "social", + "governance", + "topics", + "meet", + "expectations", + "stakeholders", + "maintain", + "trust", + "committed", + "conducting", + "business", + "ways", + "principled", + "transparent", + "accountable", + "From", + "Senior", + "Leadership", + "Team", + "organization", + "people", + "working", + "conduct", + "make", + "significant", + "positive", + "impact", + "important", + "global", + "issues", + "Board", + "Directors", + "insight", + "feedback", + "oversight", + "broad", + "range", + "matters", + "In", + "responsibilities", + "Public", + "Policy", + "Committee", + "review", + "guidance", + "management", + "Company", + "policies", + "relate", + "corporate", + "responsibility", + "CSR", + "commitments", + "performance", + "transparency", + "Certification", + "Type", + "MIX", + "Mixed", + "Sources", + "Description", + "Paper", + "Supporting", + "responsible", + "forestry", + "License", + "Number", + "FSC", + "C132107", + "Issuing", + "Authority", + "Forest", + "Stewardship", + "Council", + "Website", + "fsc", + "org" + ], + "keywords": [ + "Investor Relations", + "Shareholders", + "Executive Officers" + ], + "connect_to": [ + { + "target": "fc7f99a2-f90b-53db-9944-c6216c05ea6c", + "relation": "embeds", + "ref": "[tables/table-69 Microsoft Executives.html]", + "position": { + "start": 191, + "end": 234 + } + }, + { + "target": "edf99fdb-63b6-532d-a46a-4829e552912c", + "relation": "embeds", + "ref": "[images/image-4 FSC Certification Labe.jpeg]", + "position": { + "start": 3932, + "end": 3976 + } + } + ] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/doc_nav.json b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/doc_nav.json new file mode 100644 index 000000000..a3756d41c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/doc_nav.json @@ -0,0 +1,712 @@ +{ + "version": "1.0", + "file_name": "Microsoft 2025 Annual Report.docx", + "stats": { + "total_chunks": 122, + "text_chunks": 49, + "image_chunks": 4, + "table_chunks": 69, + "max_depth": 2 + }, + "sections": [ + { + "title": "**Preface**", + "path": "2025_AnnualReport.docx/**Preface**", + "level": 1, + "summary": "", + "chunk_count": 48, + "children": [ + { + "title": "**Preface** part 1", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 1", + "level": 2, + "summary": "Microsoft's 2025 annual report highlights the AI platform shift as a generational moment, with record revenue of $281.7 billion and Azure surpassing $75 billion. The company focuses on security, quality, and AI innovation, citing examples like Mercy Health saving 100,000 hours and Barclays deploying AI to 100,000 employees.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 2", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 2", + "level": 2, + "summary": "Microsoft's QEI frameworks improve engineering accountability and quality. Major AI advances include leading Cloud and AI infrastructure with over 400 datacenters globally, the Majorana-1 quantum chip, and Azure AI Foundry with 11,000+ models. Copilot products surpass 100 million monthly active users, with updates like Agent Mode and Copilot Studio used by 230,000 organizations. GitHub Copilot has 20 million users, and Dragon Copilot aids healthcare.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 3", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 3", + "level": 2, + "summary": "Microsoft integrates Copilot across Bing, Edge, GroupMe, MSN, Windows, and Xbox, and refreshes its Copilot consumer app. LinkedIn reaches 1.2 billion members, and gaming has 500 million monthly active users. The company emphasizes responsibility through initiatives like Microsoft Elevate, investing $4 billion over five years in AI cloud technology for schools and nonprofits. It aims to help 20 million people earn AI credentials in two years and launches the AI Economy Institute. Microsoft offers free Microsoft 365 Copilot Personal for U.S. college students and new LinkedIn courses. It commits to responsible AI innovation, safety, and security, guided by values of respect, integrity, and accountability. The company strengthens digital stability in Europe, follows UN Guiding Principles on Business and Human Rights, and publishes a Responsible AI Transparency Report. It also partners with StopNCII.org to address abusive AI-generated content.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 4", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 4", + "level": 2, + "summary": "Microsoft's 2025 report highlights progress in AI skilling for accessibility, renewable energy procurement reaching 34 gigawatts, carbon removal contracts, clean water access, and zero waste initiatives. The company emphasizes a growth mindset culture, employee volunteering, and long-term vision for AI-driven transformation across cybersecurity, speech interfaces, disability support, education, and medical knowledge.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 5", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 5", + "level": 2, + "summary": "The passage covers Microsoft's common stock trading on NASDAQ under MSFT, with 77,014 registered holders as of July 24, 2025. It details share repurchase programs approved in 2021 and 2024, each authorizing up to $60 billion, with $57.3 billion remaining as of June 30, 2025. Quarterly repurchase data for fiscal years 2023-2025 shows totals of 31 shares ($13,000M) in 2025, 32 shares ($11,960M) in 2024, and 69 shares ($18,400M) in 2023. Dividends declared quarterly were $0.83 per share in 2025 and $0.75 in 2024, totaling $24,678 million and $22,295 million respectively. A stock performance chart compares 5-year cumulative total returns of Microsoft, S&P 500, and NASDAQ Computer Index from June 2020 to June 2025, with Microsoft ending at $255.13, S&P 500 at $215.89, and NASDAQ Computer at $254.97, all starting at $100.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 6", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 6", + "level": 2, + "summary": "Microsoft is a technology company focused on AI and digital transformation, offering a broad portfolio of software, services, and devices. Its mission is to empower people and organizations globally. The Microsoft Cloud provides integrated solutions across the technology stack, benefiting from economies of scale in datacenters and multi-tenancy. The report includes forward-looking statements about business plans and objectives, subject to risks and uncertainties.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 7", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 7", + "level": 2, + "summary": "Microsoft prioritizes security and offers AI-driven products for security, compliance, identity, and management across multi-cloud, application, and device assets. Their R&D focuses on reinventing productivity, building the intelligent cloud and edge platform, and creating more personal computing. They aim to lead in AI, cloud services, workplace transformation, gaming, and Windows. Microsoft is committed to sustainability with goals to be carbon negative, water positive, and zero waste by 2030. The company operates in three segments: Productivity and Business Processes, Intelligent Cloud, and More Personal Computing, with changes announced in August 2024 to align with current business management.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 8", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 8", + "level": 2, + "summary": "The text describes Microsoft's reportable segments, focusing on Productivity and Business Processes. This segment includes Microsoft 365 Commercial and Consumer products and cloud services, LinkedIn, and Dynamics products. Key offerings include AI-powered productivity tools, cloud subscriptions, and professional networking solutions. Revenue growth depends on user adoption, AI features, and shifts from on-premises to cloud services.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 9", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 9", + "level": 2, + "summary": "The passage describes Microsoft's Dynamics business solutions for financial management, ERP, CRM, and supply chain, with revenue driven by user licenses and cloud shift. It outlines competition for Office, Windows, Enterprise Mobility, LinkedIn, and Dynamics. The Intelligent Cloud segment includes Azure and server products, with Azure offering AI services and competing via infrastructure and platform services.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 10", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 10", + "level": 2, + "summary": "The passage discusses Microsoft's server products, including server software, databases, and development tools, which are sold through volume licensing, OEMs, and retail. It also covers GitHub and Nuance Healthcare offerings, enterprise services, and competition faced by Azure and server products from various cloud providers, open source alternatives, and commercial vendors. The summary highlights competitive advantages in hybrid cloud, AI, security, and performance.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 11", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 11", + "level": 2, + "summary": "The text describes three major business segments of Microsoft: Windows and Devices, Gaming, and Search and News Advertising. Windows and Devices covers OEM licensing, Surface devices, and PC accessories, with revenue impacted by market mix, AI PCs, and supply chain. Gaming includes Xbox hardware, content, Game Pass subscriptions, and cloud gaming, focusing on expanding the player base across platforms. Search and News Advertising involves Bing, Copilot, Microsoft News, and Edge, generating revenue through ads.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 12", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 12", + "level": 2, + "summary": "The passage covers Microsoft's Search and news advertising business, competition in Windows, Devices, Xbox, and Search advertising. It details human capital resources with 228,000 employees as of June 2025, including distribution by function and location. Operations include regional service centers and datacenters focused on AI service demand.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 13", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 13", + "level": 2, + "summary": "The passage discusses the company's reliance on third-party manufacturers and potential risks from limited suppliers for server components. It emphasizes significant investments in research and development for new products and services, including cloud computing, AI, and devices. The company develops most products internally to maintain competitive advantages and protects its intellectual property through patents, trademarks, and licensing. It also engages in open source and cross-license agreements. The summary covers supply chain resilience, R&D strategy, and IP protection.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 14", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 14", + "level": 2, + "summary": "Microsoft sells products and services through direct sales, distributors and resellers, and OEMs. Direct sales include Enterprise Agreements and online stores. Resellers include LSPs, VARs, and retailers. OEMs pre-install software on devices, with direct OEMs and system builders as categories.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 15", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 15", + "level": 2, + "summary": "The text describes Microsoft's volume licensing programs for organizations of varying sizes, covering cloud services and on-premises software. It explains Software Assurance (SA) benefits and outlines specific programs: Enterprise Agreement for large organizations with three-year terms, Microsoft Customer Agreement as a digital non-expiring option, Microsoft Online Subscription Agreement for small/medium businesses, Microsoft Products and Services Agreement for medium/large organizations without organization-wide commitment, Open Value for small/medium organizations with three-year terms including SA, and Select Plus for government/academic entities. Partner programs include Cloud Solution Provider Program, Services Provider License Agreement, and Independent Software Vendor Royalty Program.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 16", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 16", + "level": 2, + "summary": "The passage discusses Microsoft's compliance with a wide range of U.S. and global laws, including those related to privacy, telecommunications, data protection, advertising, and online content. The company has implemented comprehensive compliance programs, monitors regulatory developments, and uses policies, controls, and technical safeguards to meet legal standards. It highlights tailored programs for specific obligations like the EU Digital Markets Act and Digital Services Act. The text also describes available investor information on Microsoft's website, including SEC filings, business strategies, press releases, and corporate governance materials, as well as social media channels for public communication.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 17", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 17", + "level": 2, + "summary": "Microsoft's MD&A for fiscal year 2025 highlights strong growth driven by cloud and AI. Microsoft Cloud revenue rose 23% to $168.9 billion, with Azure and other cloud services growing 34%. The company discusses its strategic partnership with OpenAI, industry trends, and competitive risks. Key revenue drivers include Microsoft 365, Dynamics 365, LinkedIn, Xbox, and search advertising.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 18", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 18", + "level": 2, + "summary": "The passage discusses Microsoft's investments in cloud and AI infrastructure, which may increase operating costs and decrease margins. It highlights dependencies on datacenter resources, third-party manufacturers, and qualified employees. International operations face foreign exchange risks, and global trade policies could impact supply chains. Revenue is seasonal, peaking in Q4. Segment reporting was restructured in August 2024, aligning with management practices. Key metrics are disclosed for performance evaluation.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 19", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 19", + "level": 2, + "summary": "Microsoft's fiscal year 2025 results show total revenue of $281.7 billion (up 15%), gross margin of $193.9 billion, operating income of $128.5 billion, and net income of $101.8 billion. Segment performance includes Productivity & Business Processes revenue up 13% to $120.8 billion, Intelligent Cloud up 21% to $106.3 billion, and More Personal Computing up 7% to $54.6 billion. Key metric changes include a new Microsoft 365 Commercial cloud revenue growth metric and combining Windows OEM with Devices. Microsoft Cloud gross margin decreased to 69% due to AI infrastructure scaling. Revenue growth was driven by Azure, Microsoft 365 Commercial cloud, Gaming, and Search advertising.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 20", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 20", + "level": 2, + "summary": "Microsoft's fiscal year 2025 results show strong growth across segments. Microsoft 365 Consumer revenue rose 11% with subscriber growth to 89 million. LinkedIn revenue increased 9%. Dynamics 365 revenue grew 19%. Intelligent Cloud revenue rose 21%, driven by Azure growth of 34%. More Personal Computing revenue increased 7%, with Gaming up 9% due to Xbox content and services growth of 16%, offset by a 25% decline in Xbox hardware. Operating income increased 17% overall. Research and development expenses rose 10% due to investments in cloud, AI, and Gaming.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 21", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 21", + "level": 2, + "summary": "Sales and marketing costs rose 5% to $25,654 million in 2025 (9% of revenue) from $24,456 million in 2024 (10% of revenue). General and administrative expenses decreased 5% to $7,223 million. Other income (expense), net showed a net loss of $4,901 million in 2025 versus $1,646 million in 2024. The effective tax rate remained at 18% for both years, with U.S. income before taxes at $69.2 billion and foreign income at $54.4 billion in 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 22", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 22", + "level": 2, + "summary": "The passage discusses Microsoft's tax and liquidity positions. It covers the OECD's Pillar Two global minimum tax (15%) effective FY2025, the U.S. OBBBA Act (14% GILTI rate from FY2027), and an ongoing IRS audit for tax years 2004-2013 involving a $28.9 billion proposed adjustment for intercompany transfer pricing, which Microsoft contests. The company expects no material impact from Pillar Two and does not anticipate a resolution of the IRS audit within 12 months. On liquidity, cash, cash equivalents, and short-term investments totaled $94.6 billion as of June 30, 2025, with sufficient resources to fund operations and commitments for at least the next 12 months.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 23", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 23", + "level": 2, + "summary": "The passage discusses fair value measurement of financial instruments, classifying them into Level 1 (quoted market prices), Level 2 (observable inputs), and Level 3 (unobservable inputs). Most investments are Level 1 or 2, priced by vendors or brokers. Cash flows show operations increased $17.6B to $136.2B, financing used $13.9B more, and investing decreased $24.4B. Debt proceeds fund corporate purposes. Unearned revenue, mainly from volume licensing and cloud services, totals $67,265 million expected to be recognized through June 2026.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 24", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 24", + "level": 2, + "summary": "The passage details Microsoft's material cash requirements and obligations as of June 30, 2025, including $397,045 million in contractual obligations (debt, leases, purchase commitments) with $148,106 million due in 2026. It covers a $4.4 billion transition tax installment due in Q1 FY2026, $13.0 billion in share repurchases during FY2025 with $57.3 billion remaining authorized, $24.7 billion in dividends declared, and planned capital investments in cloud, AI infrastructure, and facilities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 25", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 25", + "level": 2, + "summary": "The passage discusses revenue recognition judgments for contracts with multiple products and services, including determining distinct performance obligations, standalone selling price estimation, variable consideration for returns and credits, and impairment assessments for investment securities and goodwill.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 26", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 26", + "level": 2, + "summary": "The passage discusses accounting policies for goodwill, research and development costs, legal contingencies, and income taxes. Goodwill is allocated to reporting units and tested for impairment annually or upon triggering events, using discounted cash flow analysis requiring significant judgment. R&D costs are expensed until technological feasibility is established, then capitalized. Legal loss contingencies are accrued when probable and reasonably estimable. Income taxes are also mentioned.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 27", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 27", + "level": 2, + "summary": "The passage covers accounting for income taxes, including recognizing current and deferred tax liabilities/assets, uncertain tax positions, and management's responsibility for financial statements. It also discusses market risk exposures (foreign currency, interest rate, credit, equity) and the use of derivatives for risk management.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 28", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 28", + "level": 2, + "summary": "The passage presents financial statements and supplementary data for a company (likely Microsoft) as of June 30, 2025. It includes tables on risk categories (foreign currency, interest rate, credit, equity) with hypothetical changes impacting earnings or fair value. Income statements show total revenue growing from $211,915M to $281,724M, net income from $72,361M to $101,832M, and diluted EPS from $9.68 to $13.64 over fiscal years 2023-2025. Comprehensive income statements, balance sheets (total assets $619,003M in 2025), cash flow statements (net cash from operations $136,162M in 2025), and stockholders' equity statements (ending total equity $343,479M in 2025) are included. Notes to financial statements cover accounting policies, principles of consolidation, recast of prior period segment information (effective August 2024), and estimates and assumptions.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 29", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 29", + "level": 2, + "summary": "Assets and liabilities in foreign currencies are translated at balance sheet date exchange rates, with revenue and expenses at average rates, and adjustments recorded to other comprehensive income. Revenue includes product revenue from software, hardware, and services like cloud solutions, advertising, and consulting. Revenue is recognized when control transfers to customers, net of returns and taxes. On-premises software licenses are recognized upfront, while cloud services are recognized ratably or based on consumption. Software Assurance revenue is recognized over the contract period. Hardware and software are often a single performance obligation recognized at transfer of ownership.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 30", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 30", + "level": 2, + "summary": "The passage discusses significant judgments in revenue recognition for contracts with multiple products and services, including determining distinct performance obligations, estimating standalone selling prices (SSP), and accounting for variable consideration such as returns and credits. It also covers contract balances, unearned revenue, and timing differences between revenue recognition and invoicing, with examples like Office 365 and multi-year agreements.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 31", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 31", + "level": 2, + "summary": "The passage discusses payment terms (30-60 days), revenue recognition timing, and financing receivables. It details long-term accounts receivable ($5.2B in 2025, $4.9B in 2024), other receivables for server components ($8.2B in 2025, $10.5B in 2024), and financing receivables ($4.3B in 2025, $4.5B in 2024). It also covers allowance for doubtful accounts, capitalized costs to obtain contracts, and cost categories including cost of revenue, research and development, and sales and marketing expenses.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 32", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 32", + "level": 2, + "summary": "The passage details sales and marketing expenses including payroll, advertising, and promotions, with advertising costs of $2.1B, $1.7B, and $904M for fiscal years 2025, 2024, and 2023. It explains stock-based compensation for RSUs and PSUs, measured at fair value and recognized over service periods. Income taxes cover U.S. and international taxes, deferred taxes, and valuation allowances. Financial instruments section describes investments: cash equivalents, short-term investments, debt investments classified as available-for-sale with fair value changes in other comprehensive income, and equity investments measured at fair value or using the measurement alternative.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 33", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 33", + "level": 2, + "summary": "The passage discusses accounting treatments for variable interest entities (VIEs), derivatives, and fair value measurements. VIEs are evaluated for consolidation but none are consolidated; they are accounted for under the equity method, including a $13 billion investment in OpenAI. Derivatives are measured at fair value, with gains/losses recognized based on designation as fair value hedges, cash flow hedges, or non-designated instruments. Fair value measurements are categorized into three levels: Level 1 (quoted prices in active markets), Level 2 (observable inputs via models), and Level 3 (unobservable inputs using management estimates).", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 34", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 34", + "level": 2, + "summary": "The passage details accounting policies for equity investments, property and equipment, leases, goodwill, intangible assets, and related party transactions. It explains valuation methods, depreciation, lease recognition, impairment testing, and a recent agreement with Inflection AI.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 35", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 35", + "level": 2, + "summary": "The passage discusses recent FASB accounting standards adopted or pending adoption, including improved segment disclosures (effective fiscal 2025), income tax disclosures (effective fiscal 2026), and disaggregation of income statement expenses (effective fiscal 2028). It also covers Notes 2-4: Earnings Per Share (basic and diluted EPS calculations with net income rising from $72,361M to $101,832M and diluted EPS from $9.68 to $13.64), Other Income/Expense (net totals of -$4,901M, -$1,646M, and $788M for fiscal years 2025, 2024, 2023, including losses on equity method investments like OpenAI), and Investments (total $109,970 million as of June 30, 2025, with components including adjusted cost basis, unrealized gains/losses, and classifications).", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 36", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 36", + "level": 2, + "summary": "The passage details investments as of June 30, 2024, with adjusted cost basis, unrealized gains/losses, and recorded basis totaling $90,143 million. It discusses equity investments without readily determinable fair values, equity method investments, and unrealized losses on debt investments, noting a decrease in total fair value from $54,589M to $42,736M and unrealized losses from $(3,348)M to $(1,625)M. Debt investment maturities as of June 30, 2025 show total cost basis of $85,589 million and fair value of $84,259 million. Note 5 covers derivatives used to manage foreign currency, interest rate, equity, and credit risks, including hedging instruments and non-designated contracts.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 37", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 37", + "level": 2, + "summary": "The passage covers derivative instruments, including counterparty agreements requiring investment-grade credit rating and $1.0B liquidity, with no collateral posted as of June 30, 2025. Notional amounts and fair values of derivatives are detailed, with gross assets of $758M and liabilities of $(1,837M) in 2025, offset by collateral. Gains and losses from hedging instruments are reported for fiscal years 2023-2025. Property and equipment are summarized with net values of $204,966M (2025) and $135,591M (2024), depreciation expenses, and $32.1B committed for construction. The acquisition of Activision Blizzard for $75.4B in cash on October 13, 2023, is noted, with purchase price allocation completed.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 38", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 38", + "level": 2, + "summary": "The passage discusses Microsoft's goodwill and intangible assets, primarily from the Activision Blizzard acquisition. Goodwill was assigned to the More Personal Computing segment, totaling $119.5 billion by June 2025, up from $67.9 billion in 2023. Intangible assets acquired had a fair value of $21.97 billion with a weighted average life of 15 years, including marketing-related ($11.6B, 24 years), technology-based ($9.7B, 4 years), and customer-related ($0.7B, 4 years). Pro forma financial results for fiscal year 2024 show revenue of $247.4 billion, net income of $88.3 billion, and diluted EPS of $11.82. No goodwill impairment was identified in annual tests for 2023, 2024, or 2025. Net intangible assets decreased from $27.6 billion to $22.6 billion between 2024 and 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 39", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 39", + "level": 2, + "summary": "The passage covers Notes 10 and 11 from a financial report. Note 10 details debt: no commercial paper in 2025 (vs. $6.7B in 2024), long-term debt totaling $43.151B (2025) and $44.937B (2024) with fair values of $40.4B and $42.3B, and future maturities of $49.206B. Note 11 covers income taxes: total provisions of $21.795B (2025), $19.651B (2024), $16.950B (2023); income before taxes of $123.627B, $107.787B, $89.311B; effective tax rate reconciliation from 21% federal rate, with lower foreign rates (81-83% from Ireland) as primary driver. Intangible assets amortization was $6.0B (2025), $4.8B (2024), $2.5B (2023), with future amortization totaling $22.604M.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 40", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 40", + "level": 2, + "summary": "The passage discusses changes in the effective tax rate for fiscal years 2025, 2024, and 2023, attributed to earnings mix and tax law changes. It details deferred income tax assets and liabilities, with net deferred tax assets of $26,273 million and $19,652 million as of June 30, 2025 and 2024. Net operating loss carryforwards are reported for federal ($390 million), state ($836 million), and foreign ($2.6 billion) jurisdictions, along with federal capital loss carryforwards ($816 million). A valuation allowance is noted for certain deferred tax assets. Income taxes paid were $28.7 billion, $23.4 billion, and $23.1 billion for fiscal years 2025, 2024, and 2023. Gross unrecognized tax benefits related to uncertain tax positions were $24.7 billion, $22.8 billion, and $17.1 billion as of June 30, 2025, 2024, and 2023, with accrued interest of $8.2 billion, $6.8 billion, and $5.2 billion. The IRS is auditing tax years 2014 to 2017 and issued NOPAs for 2004 to 2013 seeking $28.9 billion plus penalties, which the company contests.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 41", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 41", + "level": 2, + "summary": "The passage covers financial notes including unearned revenue, leases, and contingencies. Unearned revenue increased from $60,184M to $67,265M, with $186,957M deferred and $179,876M recognized in fiscal year 2025. Remaining performance obligations total $375 billion. Lease disclosures detail operating and finance lease costs, cash flows, balance sheet data, and maturities totaling $25,481M for operating and $60,506M for finance leases, with $92.7 billion in additional leases not yet commenced. Contingencies include an Irish Data Protection Commission matter against LinkedIn with a fine appealed to Irish courts, and other ordinary course claims not expected to be material.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 42", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 42", + "level": 2, + "summary": "The passage details financial notes including $541 million in accrued legal liabilities with potential additional $600 million in adverse outcomes. It covers share repurchases under two $60 billion programs, with $57.3 billion remaining as of June 2025, and quarterly repurchases totaling 31 shares ($13,000M) in 2025. Dividends declared were $0.83 per share in 2025 and $0.75 in 2024. Accumulated other comprehensive loss was $(3,347) million in 2025. Stock-based compensation expenses and related tax benefits are noted, along with stock award vesting periods and activity assumptions.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 43", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 43", + "level": 2, + "summary": "The passage covers stock plan activity for fiscal year 2025, including nonvested stock awards, unrecognized compensation costs, and the Employee Stock Purchase Plan (ESPP) with share purchases and average prices. It also details savings plans and employer-funded retirement benefits. Additionally, it introduces segment information for Productivity and Business Processes, Intelligent Cloud, and More Personal Computing, with operating income as the key performance measure.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 44", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 44", + "level": 2, + "summary": "The text describes Microsoft's three business segments: Productivity and Business Processes (including Microsoft 365, LinkedIn, Dynamics), Intelligent Cloud (including Azure, server products, enterprise services), and More Personal Computing (including Windows, Gaming, Search). It explains revenue and cost allocation methods among segments, including direct attribution and allocation based on relative gross margin or headcount. Financial data for fiscal years 2023-2025 is presented in tables, showing total revenue of $281,724 million in 2025, $245,122 million in 2024, and $211,915 million in 2023, split between the United States and other countries. No single customer or country (except the US) accounted for more than 10% of revenue.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 45", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 45", + "level": 2, + "summary": "Microsoft's fiscal 2025 revenue reached $281,724M, up from $245,122M in 2024 and $211,915M in 2023, led by Server products and cloud services. Microsoft Cloud revenue was $168.9 billion in 2025, compared to $137.7 billion in 2024 and $111.6 billion in 2023. Total revenue including other items was $371,902 million in June 2025. The report also includes an independent auditor's opinion confirming the financial statements are fairly presented in accordance with U.S. GAAP, with an unqualified opinion on internal control over financial reporting as of June 30, 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 46", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 46", + "level": 2, + "summary": "The passage describes two critical audit matters (CAMs) from a financial statement audit: Revenue Recognition and Income Taxes (Uncertain Tax Positions). For Revenue Recognition, the company recognizes revenue upon transfer of control, involving significant judgment on distinct performance obligations, delivery patterns, contract terms, and stand-alone selling prices. Audit procedures included testing controls, evaluating policies, and sampling contracts. For Income Taxes, the company faces uncertain tax positions related to transfer pricing issues unresolved with the IRS, with potential material impact from remaining audits.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 47", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 47", + "level": 2, + "summary": "The passage discusses the critical audit matter of recognizing and measuring uncertain tax positions, focusing on unresolved transfer pricing issues. It highlights the complexity and subjective nature of these estimates, requiring significant auditor judgment and involvement of tax specialists. Audit procedures included evaluating management's methods, testing internal controls, assessing reasonableness of judgments regarding future resolution, and considering changes in tax law. The report also covers management's evaluation of disclosure controls and internal control over financial reporting, both deemed effective as of June 30, 2025, with Deloitte & Touche LLP providing the audit opinion.", + "chunk_count": 1, + "children": [] + }, + { + "title": "**Preface** part 48", + "path": "2025_AnnualReport.docx/**Preface**/**Preface** part 48", + "level": 2, + "summary": "The text is an audit report on Microsoft Corporation's internal control over financial reporting as of June 30, 2025. Deloitte & Touche LLP expresses an unqualified opinion that the company maintained effective internal control in all material respects, based on COSO criteria. The report also notes an unqualified opinion on the consolidated financial statements. It describes the basis for the opinion, the responsibilities of management and auditors, and the definition and limitations of internal control over financial reporting. The document includes a signature block dated July 30, 2025, and lists board members and executive officers.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "4.\tEnvironmental, Social, and Public Policy Committee", + "path": "2025_AnnualReport.docx/4.\tEnvironmental, Social, and Public Policy Committee", + "level": 1, + "summary": "The document provides information about Microsoft's executive officers including CEO Satya Nadella and CFO Amy Hood. It details investor relations contact information, the 2025 virtual annual shareholders meeting scheduled for December 5, 2025, and shareholder services provided by Computershare. The document also covers environmental, social, and governance (ESG) commitments, with the Board's Environmental, Social, and Public Policy Committee overseeing corporate social responsibility. Additionally, it includes a certification for mixed sources paper from the Forest Stewardship Council.", + "chunk_count": 1, + "children": [] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1 Annual Report 2025.png", + "summary": "image-1 Microsoft Annual Report 2025 Cover Primary Subject: Two colleagues collaborating at a desk with a laptop. Visible Text: \"Annual Report 2025\", Microsoft Logo." + }, + { + "path": "images/image-2 Signature.jpeg", + "summary": "image-2 Sarga N." + }, + { + "path": "images/image-3 Microsoft vs S&P 500 v.jpeg", + "summary": "image-3 Chart Title: Microsoft Corporation, S&P 500, and NASDAQ Computer Index Performance Y-Axis: Dollar Value ($0 to $300) X-Axis: Dates (6/20 to 6/25) Data Points & Trends: - All three indices start at a baseline of $100 on 6/20. - Microsoft Corporation (Square marker): Starts at $100, dips slightly on 6/21 (~$135), drops to ~$128 on 6/22, then rises sharply to finish at approximately $250 on 6/25. It shows the strongest overall growth among the three. - NASDAQ Computer (Circle marker): Starts at $100, peaks at ~$150 on 6/21, drops to its lowest point (~$118) on 6/22, then recovers strongly to end near $250 on 6/25. Its trajectory closely mirrors Microsoft's in the latter half." + }, + { + "path": "images/image-4 FSC Certification Labe.jpeg", + "summary": "image-4 Certification Type: MIX (Mixed Sources) Description: Paper | Supporting responsible forestry License Number: FSC® C132107 Issuing Authority: FSC (Forest Stewardship Council) Website: www.fsc.org" + } + ], + "tables": [ + { + "path": "tables/table-1 Share Repurchases.html", + "summary": "table-1 The table shows quarterly share repurchases and amounts for fiscal years 2023-2025, with totals of 31 shares ($13,000M) in 2025, 32 shares ($11,960M) in 2024, and 69 shares ($18,400M) in 2023." + }, + { + "path": "tables/table-2 Dividend History.html", + "summary": "table-2 The table shows quarterly dividends for fiscal years 2025 and 2024, with per-share amounts of $0.83 and $0.75 respectively, totaling $24,678 million and $22,295 million." + }, + { + "path": "tables/table-3 Stock Index Data.html", + "summary": "table-3 The table shows daily values for Microsoft Corporation, S&P 500, and NASDAQ Computer from June 20 to June 25, with each starting at 100.00 and ending at 255.13, 215.89, and 254.97 respectively." + }, + { + "path": "tables/table-4 Microsoft Cloud KPIs.html", + "summary": "table-4 Table defines three Microsoft Cloud metrics: Commercial remaining performance obligation, Cloud revenue and growth, and Cloud gross margin percentage." + }, + { + "path": "tables/table-5 Microsoft Cloud KPIs.html", + "summary": "table-5 The table defines key performance indicators for Microsoft's commercial and consumer cloud services, including revenue growth, seat growth, and subscriber counts for Microsoft 365, LinkedIn, Dynamics 365, and Azure." + }, + { + "path": "tables/table-6 Microsoft Revenue.html", + "summary": "table-6 Revenue growth from Windows OEM and Devices, Xbox content and services, and Search and news advertising excluding traffic acquisition costs." + }, + { + "path": "tables/table-7 Revenue Growth.html", + "summary": "table-7 The table shows financial results for 2025 vs 2024: Revenue $281,724M (up 15%), Gross margin $193,893M (up 13%), Operating income $128,528M (up 17%), Net income $101,832M (up 16%), Diluted EPS $13.64 (up 16%)." + }, + { + "path": "tables/table-8 Microsoft Segment Results.html", + "summary": "table-8 Microsoft's 2025 vs 2024 segment data: Productivity & Business Processes revenue $120,810M (+13%), Intelligent Cloud $106,265M (+21%), More Personal Computing $54,649M (+7%). Total revenue $281,724M (+15%), operating income $128,528M (+17%)." + }, + { + "path": "tables/table-9 R&D Spending 2025.html", + "summary": "table-9 Research and development spending increased from $29,510 million in 2024 to $32,488 million in 2025, a 10% rise, while remaining at 12% of revenue." + }, + { + "path": "tables/table-10 Sales & Marketing.html", + "summary": "table-10 Sales and marketing costs were $25,654 million in 2025 (9% of revenue) and $24,456 million in 2024 (10% of revenue), a 5% increase and a 1 percentage point decrease as a percent of revenue." + }, + { + "path": "tables/table-11 G&A Expenses.html", + "summary": "table-11 General and administrative expenses decreased 5% from $7,609M in 2024 to $7,223M in 2025, remaining at 3% of revenue with 0ppt change." + }, + { + "path": "tables/table-12 Financial Summary.html", + "summary": "table-12 The table shows financial data for years ended June 30, 2025 and 2024, including interest income, expenses, investment losses, derivatives losses, foreign currency gains/losses, and other net losses, resulting in total net losses of $4,901 million and $1,646 million respectively." + }, + { + "path": "tables/table-13 Revenue Schedule.html", + "summary": "table-13 The table shows quarterly revenue figures from September 2025 to June 2026, with a total of $67,265 million." + }, + { + "path": "tables/table-14 Debt Commitments.html", + "summary": "table-14 The table summarizes contractual obligations totaling $397,045 million, including principal and interest payments, construction commitments, leases, and purchase commitments, with $148,106 million due in 2026 and $248,939 million thereafter." + }, + { + "path": "tables/table-15 Risk Sensitivity.html", + "summary": "table-15 The table shows risk categories (foreign currency, interest rate, credit, equity) with hypothetical changes and their impacts on earnings or fair value as of June 30, 2025." + }, + { + "path": "tables/table-16 Revenue & Income.html", + "summary": "table-16 The table shows financial data for years ending June 30, 2025, 2024, and 2023. Total revenue grew from $211,915M to $281,724M, net income from $72,361M to $101,832M, and diluted EPS from $9.68 to $13.64." + }, + { + "path": "tables/table-17 Comprehensive Income.html", + "summary": "table-17 The table shows net income, other comprehensive income (OCI) components, and total comprehensive income for fiscal years 2025, 2024, and 2023, with net income rising from $72,361M to $101,832M." + }, + { + "path": "tables/table-18 Balance Sheet.html", + "summary": "table-18 Consolidated balance sheet as of June 30, 2025 and 2024, showing total assets of $619,003 million and $512,163 million, with corresponding liabilities and stockholders' equity." + }, + { + "path": "tables/table-19 Cash Flow Statement.html", + "summary": "table-19 The table shows cash flow from operations, financing, and investing for fiscal years 2023-2025, with net cash from operations increasing from $87,582M to $136,162M." + }, + { + "path": "tables/table-20 Stockholders' Equity.html", + "summary": "table-20 The table shows stockholders' equity components for fiscal years 2023-2025, including common stock, retained earnings, and accumulated other comprehensive loss, with ending total equity of $343,479 million in 2025." + }, + { + "path": "tables/table-21 EPS Data Table.html", + "summary": "table-21 The table shows net income, weighted average shares, and basic/diluted EPS for fiscal years 2023-2025. Net income rose from $72,361M to $101,832M, while diluted EPS increased from $9.68 to $13.64." + }, + { + "path": "tables/table-22 Investment Income.html", + "summary": "table-22 The table shows interest and dividends income, interest expense, net gains/losses on investments, derivatives, foreign currency, and other items for fiscal years 2025, 2024, and 2023, with totals of -$4,901M, -$1,646M, and $788M respectively." + }, + { + "path": "tables/table-23 Investment Gains.html", + "summary": "table-23 The table shows realized gains and losses from sales of available-for-sale securities and impairments for fiscal years 2025, 2024, and 2023, with total net losses of $17M, $53M, and $98M respectively." + }, + { + "path": "tables/table-24 Investment Gains.html", + "summary": "table-24 The table shows net realized gains, unrealized gains, and impairments for fiscal years 2025, 2024, and 2023, with totals of -332, -65, and 358 million respectively." + }, + { + "path": "tables/table-25 Investment Table.html", + "summary": "table-25 The table details investments as of June 30, 2025, showing adjusted cost basis, unrealized gains/losses, recorded basis, and classification into cash, short-term, equity, and other investments, totaling $109,970 million." + }, + { + "path": "tables/table-26 Investment Summary.html", + "summary": "table-26 Table details investments as of June 30, 2024, with adjusted cost basis, unrealized gains/losses, and recorded basis across debt and equity categories, totaling $90,143 million." + }, + { + "path": "tables/table-27 Unrealized Losses.html", + "summary": "table-27 The table shows fair values and unrealized losses for securities held less than 12 months and 12 months or greater as of June 30, 2025 and 2024. Total fair value decreased from $54,589M to $42,736M, while total unrealized losses decreased from $(3,348)M to $(1,625)M." + }, + { + "path": "tables/table-28 Debt Securities.html", + "summary": "table-28 The table shows adjusted cost basis and estimated fair value of debt securities by maturity as of June 30, 2025. Total cost basis is $85,589 million, with fair value at $84,259 million." + }, + { + "path": "tables/table-29 Derivative Notional.html", + "summary": "table-29 The table shows notional amounts of derivative instruments as of June 30, 2025 and 2024, categorized by hedging designation and contract type, with foreign exchange contracts being the largest." + }, + { + "path": "tables/table-30 Derivative Assets.html", + "summary": "table-30 The table shows derivative assets and liabilities as of June 30, 2025 and 2024, with gross amounts of $758M and $331M in assets, and $(1,837M) and $(800M) in liabilities, offset by collateral." + }, + { + "path": "tables/table-31 Derivatives by Level.html", + "summary": "table-31 The table shows derivative assets and liabilities by fair value hierarchy (Level 1, 2, 3) as of June 30, 2025 and 2024. Total derivative assets were $758M (2025) and $331M (2024); liabilities were $(1,837M) and $(800M) respectively." + }, + { + "path": "tables/table-32 Hedging Gains_Losses.html", + "summary": "table-32 The table shows gains and losses from interest rate, foreign exchange, equity, and other contracts designated as fair value, cash flow, or non-designated hedging instruments for fiscal years 2023-2025." + }, + { + "path": "tables/table-33 Cash Flow Hedges.html", + "summary": "table-33 The table shows gains (losses) on foreign exchange contracts designated as cash flow hedging instruments: $77M in 2025, $(14)M in 2024, and $34M in 2023." + }, + { + "path": "tables/table-34 Property Assets.html", + "summary": "table-34 The table shows property assets at cost of $298,619M (2025) vs $212,012M (2024), with accumulated depreciation of $93,653M and $76,421M, resulting in net values of $204,966M and $135,591M." + }, + { + "path": "tables/table-35 Purchase Price.html", + "summary": "table-35 The table shows a total purchase price of $75,408 million, comprising cash, goodwill, intangible assets, other assets, and various liabilities." + }, + { + "path": "tables/table-36 Intangible Assets.html", + "summary": "table-36 The table shows fair value of acquired intangible assets totaling $21,969 million with a weighted average life of 15 years, broken down into marketing-related ($11,619M, 24 years), technology-based ($9,689M, 4 years), and customer-related ($661M, 4 years)." + }, + { + "path": "tables/table-37 Microsoft FY2024.html", + "summary": "table-37 Microsoft's fiscal year ended June 30, 2024, shows revenue of $247,442 million, net income of $88,308 million, and diluted EPS of $11.82, compared to $219,790 million, $71,383 million, and $9.55 in 2023." + }, + { + "path": "tables/table-38 Goodwill by Segment.html", + "summary": "table-38 Microsoft's goodwill by segment from June 2023 to June 2025, showing totals increasing from $67,886M to $119,509M, driven by acquisitions in More Personal Computing." + }, + { + "path": "tables/table-39 Intangible Assets.html", + "summary": "table-39 The table shows gross carrying amounts, accumulated amortization, and net carrying amounts of intangible assets (marketing, technology, customer, contract-based) as of June 30, 2025 and 2024. Total net carrying amount decreased from $27,597M to $22,604M." + }, + { + "path": "tables/table-40 Intangible Assets.html", + "summary": "table-40 The table shows intangible assets by type (marketing, technology, customer, contract) with amounts and weighted average lives for fiscal years 2025 and 2024. Total assets decreased from $23,264M to $1,096M." + }, + { + "path": "tables/table-41 Revenue Forecast.html", + "summary": "table-41 The table shows projected revenue from 2026 to 2030 and thereafter, totaling $22,604 million, with the largest amount in the 'Thereafter' period." + }, + { + "path": "tables/table-42 Debt Issuance Table.html", + "summary": "table-42 The table details long-term debt issuances from 2009 to 2024, including face values, maturities, stated and effective interest rates, and carrying amounts as of June 30, 2025 and 2024, totaling $43,151 million and $44,937 million respectively." + }, + { + "path": "tables/table-43 Debt Maturity.html", + "summary": "table-43 The table shows debt maturities from 2026 onward, with $49,206 million total. Key payments: $3,000M in 2026, $9,250M in 2027, $2,054M in 2029, and $34,902M thereafter." + }, + { + "path": "tables/table-44 Income Tax Provision.html", + "summary": "table-44 The table shows current and deferred tax provisions for fiscal years 2023-2025. Current taxes total $28,851M (2025), $24,389M (2024), $23,009M (2023). Deferred taxes are negative, resulting in total provision of $21,795M, $19,651M, $16,950M respectively." + }, + { + "path": "tables/table-45 Income Before Taxes.html", + "summary": "table-45 The table shows income before income taxes for years 2025, 2024, and 2023, with U.S. and foreign breakdowns, totaling $123,627 million, $107,787 million, and $89,311 million respectively." + }, + { + "path": "tables/table-46 Tax Rate Analysis.html", + "summary": "table-46 The table shows a reconciliation of the federal statutory tax rate (21%) to the effective tax rate for fiscal years 2023-2025, with adjustments for foreign earnings, credits, and other items." + }, + { + "path": "tables/table-47 Deferred Tax Assets.html", + "summary": "table-47 The table details deferred income tax assets and liabilities as of June 30, 2025 and 2024, with net deferred tax assets of $26,273 million and $19,652 million respectively." + }, + { + "path": "tables/table-48 Tax Benefits.html", + "summary": "table-48 The table shows unrecognized tax benefits from 2023 to 2025, starting at $15,593 million and ending at $24,729 million, with changes from settlements, current and prior year positions, and statute lapses." + }, + { + "path": "tables/table-49 Microsoft Revenue.html", + "summary": "table-49 Microsoft's total revenue increased from $60,184 million in 2024 to $67,265 million in 2025, driven by growth in Productivity and Business Processes and Intelligent Cloud." + }, + { + "path": "tables/table-50 Unearned Revenue.html", + "summary": "table-50 The table shows unearned revenue balance increased from $60,184M to $67,265M, with $186,957M deferred and $179,876M recognized in fiscal year 2025." + }, + { + "path": "tables/table-51 Lease Costs.html", + "summary": "table-51 The table shows operating lease costs of $5,524M (2025), $3,555M (2024), $2,875M (2023) and total finance lease costs of $4,825M, $2,534M, $1,853M for the same years." + }, + { + "path": "tables/table-52 Lease Cash Flows.html", + "summary": "table-52 The table shows cash paid for operating and finance leases from 2023 to 2025, and right-of-use assets obtained in exchange for lease obligations." + }, + { + "path": "tables/table-53 Lease Data.html", + "summary": "table-53 The table presents operating and finance lease data as of June 30, 2025 and 2024, including right-of-use assets, liabilities, weighted average lease terms, and discount rates." + }, + { + "path": "tables/table-54 Lease Payments.html", + "summary": "table-54 The table shows future lease payments for operating and finance leases from 2026 to 2030 and thereafter, totaling $25,481 million and $60,506 million respectively, with imputed interest deducted to present net lease liabilities." + }, + { + "path": "tables/table-55 Share Repurchases.html", + "summary": "table-55 The table shows share balances for fiscal years 2023-2025, with beginning balances around 7,432-7,464 million, issuances of 31-37 million, repurchases of 31-69 million, and ending balances of 7,432-7,434 million." + }, + { + "path": "tables/table-56 Share Repurchases.html", + "summary": "table-56 The table shows quarterly share repurchases and amounts for fiscal years 2023-2025, with totals of 69 shares ($18,400M) in 2023, 32 shares ($11,960M) in 2024, and 31 shares ($13,000M) in 2025." + }, + { + "path": "tables/table-57 Dividend History.html", + "summary": "table-57 The table shows quarterly dividends for fiscal years 2025 and 2024, with per-share amounts of $0.83 and $0.75 respectively, totaling $24,678 million and $22,295 million." + }, + { + "path": "tables/table-58 AOCI Rollforward.html", + "summary": "table-58 The table shows changes in accumulated other comprehensive loss for derivatives, investments, and translation adjustments for fiscal years 2023-2025, ending with a total loss of $(3,347) million in 2025." + }, + { + "path": "tables/table-59 Stock Compensation.html", + "summary": "table-59 The table shows stock-based compensation expenses and related income tax benefits for fiscal years 2025, 2024, and 2023, with amounts in millions of dollars." + }, + { + "path": "tables/table-60 Dividends & Rates.html", + "summary": "table-60 The table shows quarterly dividends per share and interest rates for fiscal years 2023-2025. Dividends range from $0.62 to $0.83, and interest rates range from 2.0% to 5.6%." + }, + { + "path": "tables/table-61 Stock Awards.html", + "summary": "table-61 The table shows nonvested stock award activity: beginning balance 88 shares at $292.28, 39 granted at $413.90, 38 vested at $293.25, 7 forfeited at $317.23, ending balance 82 shares at $347.44." + }, + { + "path": "tables/table-62 Share Repurchase.html", + "summary": "table-62 The table shows shares purchased and average price per share for fiscal years 2025, 2024, and 2023. Shares purchased were 6 million in 2025 and 2024, and 7 million in 2023. Average prices were $385.10, $339.46, and $245.59 respectively." + }, + { + "path": "tables/table-63 Microsoft Segment Results.html", + "summary": "table-63 The table shows revenue, cost of revenue, operating expenses, and operating income for three Microsoft segments (Productivity, Intelligent Cloud, More Personal Computing) and totals for fiscal years 2023-2025." + }, + { + "path": "tables/table-64 Revenue by Region.html", + "summary": "table-64 The table shows total revenue of $281,724 million in 2025, $245,122 million in 2024, and $211,915 million in 2023, split between United States and other countries." + }, + { + "path": "tables/table-65 Microsoft Revenue.html", + "summary": "table-65 Microsoft's revenue for fiscal years 2025, 2024, and 2023 is shown, with totals of $281,724M, $245,122M, and $211,915M respectively, led by Server products and cloud services." + }, + { + "path": "tables/table-66 Revenue by Region.html", + "summary": "table-66 The table shows total revenue of $371,902 million in June 2025, up from $301,369 million in 2024 and $187,239 million in 2023, with United States and Other countries contributions." + }, + { + "path": "tables/table-67 Deloitte Signature.html", + "summary": "table-67 A signature block for Deloitte & Touche LLP, located in Seattle, Washington, dated July 30, 2025." + }, + { + "path": "tables/table-68 Microsoft Board.html", + "summary": "table-68 The table lists Microsoft Corporation's board members including Satya Nadella, Reid Hoffman, Hugh Johnston, and others with their roles and affiliations." + }, + { + "path": "tables/table-69 Microsoft Executives.html", + "summary": "table-69 The table lists Microsoft's top executives including Satya Nadella (CEO), Amy Hood (CFO), Judson Althoff, Takeshi Numoto, Amy Coleman, Bradford Smith, and Kathleen Hogan." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/images/image-1 Annual Report 2025.png b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/images/image-1 Annual Report 2025.png new file mode 100644 index 0000000000000000000000000000000000000000..6f60dce2ce0879f137837ec7f08f0c7696820eb3 GIT binary patch literal 524557 zcmeFYRaYfU%qEJv+s3^ccXw#qy>W-e-C^Ud8ya_acXxMa+}+(B&O6^Z>&#!6i@B;w zs!|tONj*s_6``aciHv}c00stzEG;Fb0tN;#00suB0SEIRXBJ7N;J*RfSw->}SoJjF z>3;;2xrn?77+4($@xuuEKN{XaO3N7x4COy5lmYM&`%+Udu=EdyX`%v_%SLTqJcREG2(R+2krX$<{zp8`*lIubEhf6A}WvXj|xJ zGuYq3NEmSNLFjM+Sh2OarS${VKPzbIGR_d8#MRB7w-lp-qr+JQ2O=5YDbe&_Gp4Pl;GU089qw(1OOs2A!3`bL;RisV-&+LC#w}VC~ zq2~YZ(tiy8y40ZE|9>x@BK?t}$Nv__jN-w~{$Gduf1<|!SM-0h@qg9yf0jvT@?f(1 znR$F3fBPOKQ_%BT)6voCr9D(;{g`nzFFDX9W=SsVA{^Cc|46ECHvPSJTGnW+-SX)7@f}Su*q`=}}B+ zjvkfOq|U*2XqL8_jO=?_vbm5W>!8)NflphZ_+X_lW)28Q^>$cI{h8X>19ki$F8UT> z6l>PIGSDq~X&AZnH3uqjQax3r{nBKlZ@KU4-%Hnea9&JcTFxcz9+vIhZEI`O1nRm|;al0Z z@43}6u&U{Ni)gMJ4&=3!q@ub!dCA>W-z0epAhG%0aQ8Lm>+@Tqo7UVG;7`PzIYYP{ zdBYyZ+J1E);hSDbINbLY0-kuJV#HFQBXf9$K2Z%eoH#yr+EW_%(i3eE0WMVpntcdR20CEG_X1C^q-hcy;**yCZNm z&7mjCAx;Nqg4w2M_A4|t@)BWfKw(=jacy&-!2CYv04al z?MR+p>+RSwsU0M+mT`kkz^fTUX|*8%YP}P4WY%*P4ls*ptYF)my~5I;iC?NtlwqK@ znrkFIVkl!uJ5^pL9+>vKMoxk@>h0&2+%e${!iVaVaoVVs#d&(gbrz4sAGqX0WP5YOhbIC&b2W?V4MH+ZED9oB5K5UqZ@ zC2{91lYg~0^B|Np6cT_sj0d+0G}mQ&u^ILT_x@)$RPzUq(R^zneZ1Otm=BWc4hZET z)IZ|LU)hQ@1+W1TkYpNqaFd@0v-h8youcExIuX?f&K;0@m)Ntkt{SU8L`^I3ki}wA zNd>k=z4a%Xx!(M3C-pdR9|a#SZdB6H+(|i!yf!wNJt|jrPHk2pJPyBolx2C5Ep1v!D#H5ozm2$ zltU=INqc=4zO=NY8Fw2yqup5M1$6at`lk>J+MKo$XHX_M8buU!y$n}ZY>0jL=dipZ<(qKo<-Q<;rQ%CCC|@$ zEM*)jU_S~l`1`k=2uLE;790rC(to=V$fLWD$|AnvA9HU>qsjswxMyo_ePfx#sYbMK z*2qKRT~XxV)2So5i#@Hgb^x!YR0}QopW#^V;NMXHj+KbT-qdQE(k+xeMUF@AM@`Onw2D%~%=j=x`tq6vO2Jszrq+oz=bCd`C?YWr;`X#DFx za|rW^P)stH|GQN1J89aZ(_^x)HbqQ{*jArf%TzJggRYM%!N_1=RGiK>Ml|&gxgeM` z{Z<;M($<}>^Uy1%eD3(D9C`+4B7wuHw0;>vEJ8EaV{2>*T42^*-XAq{fokWU7CYz- zbu{wS>xY~An6`;a>z$1`r0Z%h-aoODCgN?qvzo8}{tW63#zk^pPLvco=&&DWp+5rX z#Odj&D+LJ8{iuyy7$BDbj9X0|L@I|+76Kxxq)4TmI-o}^T+OUCLK#k-RWLh=Y2KK5 z6Pi>bJ@Jgl98W2Lk{Vajfn!}=}W(hUsdZ`qXMv8+ubarCUr0J@|(h@sK^s(AL-{Q;7dD+h=7x8@=zP?!@n&?U= zuo7`mu#3|}#cJ*KxSFLU(UdQR!Ye6Of^%zIA7s7KUeIVw8^c$$;BwXv`({c-t(+37 ztCF`U_R*)fxE>^GUDBEqDCCzVCoLu#kGbU4L08LgbNykfiy2Hz2t| z+2)$|#p%nmpm`T-*W9Q5bll#Lq19A7Cp2yyD)3ofW%5_C^2uA_`GWoCvmc%RVl@Y~mnBOs=GL4O7Yel4s&Pm(YlOyk z!zRs4sa0WL!kwdAP;r0^Z|=eFi`*=+CXUew#>82Kz-f)wuA6}(O+t!K`S<84@`KQX zLs90rvp{_*D(q_knDc3-Qumn5zu*HDl-IRPE&6J_B8zpNumZGO)j%jQ3WfpCtR%_q zwLV!V%E2)&{(y}A@&)L9a{4j?P4-2tC@{Xr+ZbYCD&hj}j`?u?7eOLa{wSwTJtD+4 zI^*5B`oN5eh9Oh1q6^DFi1Mp{`}9WYd)$zQV;%`jsSkMDmv7(J{}Dk`CVs}py07QEvbxD0n*YDBocjQV|4v(JURxG`KEh?W z12mXGBd~AAPwz$6@6oh|q>-~F;YtmIsFO_cA;48!v#c2PWZQoKdb|BMuay2`Cbr2& z08N4%b+(71{F3n@{vcJW%Oq#=06IIYg3iq zaMN3Iqsh9BC4Qmdn)Mv|pH4`9l?hS2&hzN|pVPBLncD%FAmK{F5ECEF1E3|JV6 z=3C%!!LKV5g+@5=A3Lcq>~p3(+CD)ZcF_dUutV#%gE}K6zS^XwHue2Z{C6S zGhQQo|1!E+olP3AY;@LhcJeltt^MGLbLp8!m0zpQ7pIudIexiDE2@^8nr~x0%q^!T zqn6;YL`1$jwrrofOaFv>!nUmYwL=JWlB1>H4~Tl1n4QSt8@>rT&lfx(eRUUoLu=YZ>Z8doGnUn#9OA%vE3-|IXQWDuMwkfpgw_R9Ir=F6SZ6N@G^>c_cq zQ~g3)#q#3nJ;|>S+qRcv8Gd2=d)+YvN%!T2qzmvX3lg8tqaz}qOd@YJkDphY3`GIxbOEN+a_?L0?dj! zwz1qM^RtRyqx?&5)7B`{_?oc9zf16%n|&WmSLwuE?EcXI^5NeQMjRohE+n;m4gp-HVQ1PL+;>T!m1Nx%4?pg#4j=*WYqR zn%SPF3)*V*M!ytR9-A{9^xUXFq;VLcw0=?$Eed@N~1GNZc&ZuO>!M2C*bof~W5y($Q`@v34 zrx|mjL^ki<=2B9c{P;8h#qUFm|DtF7-c-zER^{Z<# z8m|qz!y8M$PjIOx;>S97I)3L-jk8Qs8yBQWM|7j+F)^s&0X@zsA&D2NYHHpUYJ~`s zZ=Cz>=v7K+&3H5q+6MT^&z-o7-N9g!%Mb?Zo=}pyL_I2HzZ{O@V0os^tSgm7%Q!b@ zH|%aKk9;tN{VWQLQ~Y|AXGnVhGvq=FE1;(ex83Oo*XaFiZXw;v_BV6v!}00GyL(D% z_-q3a>8Vy)pdrYj2IL+Ufr20o3Z39|cu;$+%_f0A2Jpy}w7o*9>M?o-Rm3o!XDQag z{|#Tc+cMw>sx>!EdWcaC!jW_B`QS#W@5!b_kM^MYcwYRo_6zo%L0v6G@Y-Wq0hnm> zX-$*~Esc71>V28z$~Wn(5$ne@3mg?A=AZonE?83&G%fyFds2}~U2@>1tGn}-qhPZ*{(cOdpsbY_=If#xej!#QO|nVCqVos@lIEtYcPfG zzuPhTP+%N;({Qdb?V#)@3$gILhhn!0&-nHmHJB>ydYAxOdw+meh4&^bE#sDvxfOHA z9R2#&P+e3{0m9^0w-34h6rRF|c<;O3Ukw<Rk{eVPLq18_YYuc`CQev` z&7cdlB!IR^2J6`1!dbP`9#@mZj8kaBBW0f`tKuy4@nQC`^ zz86%8Rn*E+{BmRW=<`i_K%BG|agQ?h`tKL~h6Q+LvdK~;H*pp*wI$um@oRH)YhSg% zMK>4VQY<@aW%fyk;L<;ZtI8pweJ{qRNoNTnYsrHe)d>+6lmVh{ir{K`IHMWSqnk*j z?Vr1kX96rXY^O85Pt#v_A-`o(svGzBO z#ql8DS)JT-$K^HcPOrY50pB9;Qrwc-j^@tm;^J_DaJp=BM3N{xn>$JL=^xSk8`aR5P?^S?)uRVCxiGGIf<{5Brtheaf^Rz`3$740$kGhCvxv zG2!j2>mW#ykAf$aXhfPoWCi8AUaJDaJ5kTU;Q6W^vU9h^hdt6@HBz)O!k*{7+b&tM2^OmYl{W7KZ?devSuxb$$$4#z@HEz89jvJ+GXsHAAK`%^ z<)JRsdmF(pbD%&K==SmTUv$pj?*Ne%&aFNM9Y{wmA&u)BoC)eARXhj5hlQ}4i{u@b z{&=s=2Vd1}{8&DS8ZSXyKN4c%#-OfGx4n*`z6Rb1hYc7Nkn(H!`;r(@f}`9>)|I5l zO9_GZb-J0RgX$Ck?MX+s>#&IV+2ynF4v%abUdn)X)Hn5sf#BgUkCbo$ zh<~>O>jW~lElgW#T$QJ*n;>M6+3`+!LWjODj^d#^+xIc2o{^?zQCg5T)J`v`JsT=Q zb22}V%ZfF8Cs-h}k%3YwU)2%sO&?O36|neLfq8zggBH(w-bo2N?zHfaGLOYvFImpt zd9>g2Kp;Rd*)2x@pMcA~LI?iZ;RfzX~;qZ;Y{>6WGJ)nl|;} zd5GsE2%5~z6dCu8NXbOOGbW{{LyY!LKT0{gP@V=Xc7{Z25dN`{Sa)%ERp-tbq?0_V zgKxrJs4%$F4sD<|;MHK1BBG!qV?vL9JqRG7Tb3Ov=jJJ$CPcfR2IjS~ilSu-Dn4f< zKi#K(T=Ok&L%0VcdXx!qhD8xRIowZdr@0?SyCX0z^ylb0MEgST4?RSWM2uh}iGGd^ zMD=YTTJR(@klo90MKL)osk}}s>1PuI5-Vmi>SnuzUiE{E)}=@Ki*tFEgio9;cQo|! z#^bg7(ND3k!9=lfL7$yma6(w*55-YtPgsogMmtPAZI2B7HGOTZDN%D3?%WTfGWiD2 z8w8XOCJ1FaF(%r)^d>CSZfPwBVa+Sldj}aZ<>HCOktR5Tq`g~Xy|-hU`i<-wv%&)! zx6+cM3L|SKtqlr!Yt$Tld|@{g&UaPj$yX|{fGfFxYaBZJIZ6c=z1tj`Og_PRNl#$eL+6r7#}fwaeeyB5H#*lT57sw{qi$*Di`oYe*Ib} ze+(b}1t&LQlC7HrQ_cEA&}(Ndt~%@K*mDLrq7S8Btxv8Ra2EG-JCrszDz}>5v^yI7 z?|m4xn6F$v=Z^r!rqj{CrMmSrM!IgxuU);ZoFz9abtV3j(9OL~C#~U=rIK=!<&7Ul z8Z3z%>%xENlPKA1N8~U>7@LO$#j-p49XoGAUMs$=I7){x-{8_+qpMhE_C(a&KK&*{ zA<31D#l>g8xt!ELsUj;;?+4)U*`yMEbe1|E;VZwGuWS7 zB;a7)jYTIEd2FSkIsZ8Gj7o@Uc#G*yDL5Yk|6wJS7vd)PJ{Gd+QK@(LaHJSRD=*7e zH_LShFQZqGLMo$dP=EBJ2Tf2H^HeN=v1d`dCjMW-d+GFO5Ivw{*1Co_JZNb!z_v{4 zG{SywC!zjl3ldZHcrJ%>tQ)+UpVf6L8ZEI>8&WA)ynO`juVO5T$r6c)l&rbNwPD~t z>ufMeL#{M)5jL2vTjQPmbX4JLn^GD_j>43aW}PVi1P_FeZaBslsAt21!UEUcGaHS# zb9%r>-nQ6s#`-^q0Bx@B8`G&jmr~b+QQdfRxAgmBo0fXC1N^~a3p^v*=vCD>8;l8T z)y1krN|PFTC(s7X<~jqm{8sxHiS~F(@CH$PxAfRr7B300mwuKiX`KWp3z;@AHaJ}iPUg344j6|Q5;vx zKANWQ_yumcE_be6XE1;9*qgsRi#g|d<|!C*ka*kCC?UKUv!8zY2Mzsb4gVH_z^SgP zoB(OBk6dg5_s8l$vP+k9btr*J32sh$H}VL+fsay$7`AX3W>8 z1%n?&O_lqBlHs@CV6R@UzP$UNgYONposchD_zZ!L6cGv002oV#SV?Z`c=kmv{Pb|; z&M1exFWI+87*^UpNQ@EB3Y-$@sr7{Ap0{N}w0vSD4|pFJn>P^>H4eYX&&oTIR4U|oSxV5rVSz!iM!(Y)7v^K+6a`1C#UNIB1p?w~7)DG`Q zqg=#w9qSlT&@=BDwa&)TA%(b#M?%KlymUB2YBQcO(vB;D0$Ci~Dlk498BI`dmRn=XHuQ!NmND{rmv$CfH z0I&Z(@}$iY6<#O|p}o~koLq$iDo5R2q79MG!F|=SC%sWm(dU?=S`igk)(VOZmZwYj zUrQ?<_;K1qd9H~ASxPz|1OFm@PLHLoP&o;xE=AZ?`9^BJdQ=AP&|h%_LRo0ptbm}p z+{Col)h{qZ+#Xq@^8v!S2@qa^jOXSq0AerjIct>Lv*tOS3jCIAXk{^==dgEsuF8L9 zL&m(*#KNau?9+&+yxvnaB%qG=L%+i?_f0!c(6H2`#*0dia1PpvQS4+=2my)^ z07VpVpXnach&OfI$P{Z%%_LR6;yittW?1>C{ZEJx$n2OK36OWaDNx^m=nEv$V|`eC zWSjIuAN^1Qw2o#!W3PT24iARTHT(aje8l*3`PWs(a|fD1bQJLG&$#SLP8z_tCv6-D z^wJA#B750;4_#hbwk_b2hwVd8eN}LT8F@oRN_z(A`w?rz>_7%jLmw^|jh07+GT&C9 z-D7y}jKt8RIV+`JV+ax6X%+{jpu5wy4Jd$j8{|6;QxsfbjO3lB=w=8S`*5X>{3vzy z3A-I0j;;4d!roPr)b=X|Ih{(-Mn?Z5&n)pkXrWS_Pl&o@SU`$pcrg1Xj?p?4lCE+{+tsS4OgWIQtQ}x< z_`aZa31)@^?t+bLlX@g=jI1Ihv04U9wEDE-5Aq&T7MQ`xd3`;~w=z(3V8XZCsC(2~ z-;-6`7~-H1(RvFdmHF$Uz#jcnunuqMZ!-Nye9JoEf`G}g zP8Pdov*t-&5e}rnzLlVM^6>!WZhz??o;(|;!gIfDo#gV`9xB{U?v10r2;Q$eqfB#K z*cP?UbhqbvDI(ZIvWWhb69{XmU5d{UM&kud(n zh6-grdM}d?cq^+(j8t5d+>)n{%3gp5A&CTwljG{|68wM*Wmo_gxaPo7Glujdfhz>b zt)#iE!k+L4LiY?2n(kC|f`T%0>I4(RGI_=_j8VdwV0?#4sHhyPU27$@?~f8wP=!lx z7$O`fYeq(>5F%eAK3gdWzZRpCzg!+?JP2P5D;*fZ*M#;Nn9?|GV)a?l6zTPN$o4t@ z;IAWPnBL*cXiWH<&XykMIRF+h)Ff@5(tN};`PWT7cKkSN{sVXVHals_8}jbh-hSKq zkJwzdIgP1=(VSd&#xp$#lPlW--8QKZ#_a}J7wa8mrR2~aOxZVQR=lOgoEt}nt_xj+ zuN}5h>pxs729YJSDSPN*#ZrqPNR9*6z-4OKttU-h4~9S8>>*zl73X2;mOTD2+HZmR zJa`i7m51adJjOZ`q6bBGn_G*tCqAf(pqe4i!TeMHk#-rclO;c z7G8sQCZq;YTohLpH2Ya>5^az#=?Xm(;WUd`SC7savj%)p} zj$3ddyS8G39h=RH+%U&D+}}W}>b5#yp8>74#}?Cu>A#F>@Wr;cP$;{= zEUxLQ=Fn07rJAZqEWmddaY(wpuGg`pCd3wq`UXQovu=1YXJI54F z=)rIjf7gtT-w*aJOkwmctlC<~9fj^?lC4`gfM9_nY0D9(U=&`1VZ0FPV~#hJ_tN{x z^kh0-2@`LFOJmxpV1gk}C~LRBpZXgpuRV!v&~jR{P(Z?60o=Y1O2JN`Y!S3z>rdbm(x>|O5c zxqZ^d2xT=~g>Zner=m&1P_zAVJyeL=_=%M$BmUywh(JYsQJUUi_Qp1_MfBs zXr3)YQ!kdEf1dYb(-kzaaA@?m{49W|&Sc5KNHuZOv}+%M1=^o{L>AyL4u%?jvIm2R z?i##9PJBaCkrwIq#Xd#tMA?uysT6(&Z7OG7&$X~xf@^n>gwO{>I}CjTPt=wZ8xV|k ztO3vH(Q*RB^D5)Mi3_-yJ*D5lki;r26a3AVa(2>Iu;Exxaqbxke{xrL2$?H#*&%F3 zYtfh>tR>$v)jw}E^6^z)-SCi~7%^;_a=(6#7O1>B;Vtr5$K#M3=j!rrM{7#R(l0Rw zsntN!{2q6zOU7~=T=DPRPb5VViUDk$YW4x5WZ^QUrpU8oxR##9!nw$ZNR*fw%1Lj@ z&kk`;S|VE6z6?O~3iq$&)cFHol;WZe2&jlVzbQxuniLFfRK!`v-kSDZ&oYSKW#2*&wP+zuEUC~BmRcT_EjGn6W(zk z2=~|E_1Bob*o3K719TF}nHe1o!JnjkD4{L+xu~&v8WNLm#|48^lt@P}xy7Z+zz5bs zOeL%uPCDxGUGoXmBUp)19m4E}?BO}{Nj%Hnf3hapvg3e?am~?6JhrYQv~16dpFA>r znH-JX^}R2;6)E`6JwIX8{AA0wIO@0WrMhhwaI@rl%roXjl4w(!mpp`1^n}NTz3!#8 zgIgD0m{b@V$VhYnYV z)jo%ELWq4OKyFkvqcD+mLi0qeW(uUK=SVi5TEHTC8@x{v3DjLeHxnsn)jQb)vjHVqWLR% zCW7w5h?}nTQHmhc_^bOw&ZIjW?uj@?0YnlC_r!%s%_I^i@FrDc;G`l1R^`-z%(voaXo#_wD(JL1$hEcV?wlREoJu zbp+jRGisWV50Ph( zt{nL;&YUfMf(yD6%uDu$t7tDcDo4(+ca$k%=+joz8lnCSt zfX7lVIHadxFC>92qm;iRx?;VMW_asN(Gws+1^rkVyitqwTNPBiIrEaaDHA{ zGK`lS6&CL>A^91ju@WTy>$v`7ovahes`P;oqoO+8pfD@h#6UoC-fdPrNDGL>Eu{2nEJY_o$|W>w=3Ezh!o@`h zL|-v{)%hc7Sa;b94)B79r58{9!p=;CNCPRI3qym6M%7f1D^Bly$q-uqk(1!8Gl*ut)$V&}pI4ur0W1QVX)q z%Kn=-of7Yv#_p);23ryxHYu!<2@1r$hDENEkVJAeoTs=}y=VTZA@7)$ppsQr=mnMZ z19y}WmeeRg3RW*(Bv~uwSYuxyHm4_)X;;&|3w->15Zhh2DM=>O5JH@|Kz~D0IdK2d zs2Al6#zjl%aezc5@9QmL;y`QwHyN#YiG>)@?C6NNsJCr~Le*YbWc&h7=$^N#(zdBmv!t!=Z z1{edV%7&A)r;fTEnstE^{jxx0N8P4Y+58$~``!(=0Di*p`~Jc&B#?*xxBfK389Ub} z(SLwo(nb~|d-_mnn@JRMqN)stXL?>WXEV)4$P18sR_rS-KE zd#rY%Mj&}48m=mV@fp*s2x48%X$p0%=pK1p6lxtZO~IRocfODd*sdowMZ4&$&9{i^hTnQ;e-2 zs1eUm+Etgq174#Nzml=lLBM^heKXdFi~FkkJwKR(l0>NW zS=eh)0pSVs8rNi+D)d|Bc%J;FT~14;gpNw)UJ*bhrQjj^WpS#A$22=$ewO%D2)8Xp z+$AghreH>uSVa1IruL;JW5pGZt;gXaGC=0o4mFFP+mi&^ieM@PY)}_*H&b1R3$Ly zBdec;W93-RE8-2OWTrvN)zVU$ViYU!LP#z6n){=p5n^c|GhQVSsBSwNQGn?(6QOZC z)()bLntT65ci3e~^w1PAl)LI_pAJ}T@B7trNF^XxhL}xtjuIiBc}!VPE};jfR~l;{ zkxO^gPDv7aZbCmAph?H;cskh7-oC1E?NgMs`59^IDd^ErCG>53DJk^+`lKrKwfaz~Y?e6(Du*>IcetS#sqhAuZ z9K%O8M?_V}o=Xs7TyO<6?6|TLeT?f`Us9`kdzAW~u9P%$2o<5f* zP|oM_@~Vlc=Slm$r022d>4?j&Wz*aHcSl=WUawu1t*vd7d5(9v&j$sGoetmf`q}S~ zlSv1=IKi5P$@ipg-6_P!K3Ee4YVW!v{B%#om#`7d5e|(F6A+?jpkn^a1H2z`0e4yb zme$Uk&#+uC2?o2Q2*fX?6`(9I_%UkWO_G;&3^qq5m=LlJn^? zCpjA}2)CYDE)?sR+!1-TIpgV^Nmp9QqGVa~YQRMJovd0lEU$V#;Af@d3Eblja#NQm zc%4h}fzsfoW8WE|CBUG80D-O2uZzH<$J`1mb5LhUcUN#fl<>4iHcRc8z|0CveHZNJz_Ja2)_CAw}|5yBb%DVv%`MC{^=4uO(| zrseibm7{Ba=xE4pccR6E$UjC)vVY*3@JH^I`opbOHAbIhFKiW)xL!U4Bk zEuomwPD{IKR_j*Oku)#Z#_dw}e0nBqxIxf(YQH^Cs;{}&5Nl;Oe|P2AV-jvKES1vq zVt@Be{8T1J{v zUm@5ZQP;%m3c|82ibRd|3SWs!(#Wo6CIOh0K4cF-;NW`22m;0|u7Q#3A0c}?n>WDs zg8z3d+q$5~O>Z}qh4ZXa|0jI^Qzj1h;XDNt8 z-%S8kDetB|ItGB+2!zReh`@%(k z*V%4*-;YAuKdB8#2%FrDEQ&_^aBezjt;?KPA{DVkXj<;`8#y=Td$58-MTC%K&y88` zyHWX98o$U@dhEb#2#iRlDzpn$xxupbsN8#&xo?4ih6K;y89E)ro`$|}o8I!!prYTE zv+cL0D?;^Y!oH2SwJLa>h4}=DCMEUeispTbbneufIB|mTILR?oEP?YW`+1ixCF|RvYllhz~SCT#o5TV<` zGF82KKEa2d@5MDw7cKviA?}Fy@#!difZ3l_EexWH;Xs3T{+bPz{ti*52h4-)AX|^S z55=x>-DP+%b?h+qJ2{Jk@oG)DD1WXyI>3a@+EG2t>7fJ>A#OaX9D-o!UGRHpt@%HK{9Jl#^{ylZMcFDxc_-Tq)DGcYugF>6(Sb7yU-v;LMxl887--0Jk|t zJ+{%!P9HD@+AnIMfLa;VSZ-b$%VKJA`=D4a%csCwR%ze7Z18TizPf4&WO^%C_%py&RyuLZ z{am5Cy(;64ASg0d43lOUvxr|~Znm#UFf!~LCgAYq^Gn7I*1Kd$ZL9;8!!cKkXp)=f z&SxxniqjCozZBF?U^iPr%MGt$;1ldu-vh;uW~1K^D0IbeHQ?0}T=78*;LpfVEKRL? zDsxWRk$kbm0!B5MMr3D+1bGM-1Oxy{PT2r~-ZMAMSM#k2(| zdacExD6U^!`Mh7fHCdx0U0<(Dm}+Kl*>732TyH4Hb*|g#_v{WtH;ruW`uYrK8dLB` zN4a8JQ!P>2q&Rd| z?ZM)$CCdYYV4mJ3Stw`loq8E^#J8hCe6!#nCE$P6XDiuIEWPQd)_1yMM#k86Dxt3k zxAV8Ar-YnO8@`?M5cRMTk3?f1+XB6HU1y~WNi|>{Sq5H@ff8hrX5}RRrGTqvoq^78Te=BnyvIT{rJ8sV8P1{SBGaBT%>u?r>dVAO?TgLgkLxY0z4%ZCF zD81eVfn#QmBaAj*%epynKlsHqH|cjP4U@<_3#uXMhK0pv4TmY7CkXo+y%D+ii4R^o z2-UWGx7sZys8Qrc2XaLMjR2kA%m%rMkM}2 zmH;)i>|VWwXOw(dn+PS8vy2FdQ9< zkdTcDGBrr|cqJc-#88{*r>YHKY{7JQK2cI5rt=;^nGu!@f-l&%W?04{WO6o)59#%7 zC;eQ@{%_GFA1rC$ZhJwt!4IyJD3idZ2v*;4sE&^S#iF0zmdJNYL4!5~X0NZRIUlCp zPe-bkBYsEbYw;xY#q#`N@)`1a{08e=cbPUec4hj_W&MKFePX%z_x(MP9?}R9#^U3MC*c|3 zc>ukaD87BInLfWqoR*pC-)ZMGA!fh!9zY4X&8PLs?<} zhF|}AkY2)a6>nlsQ8ufzW^q;j`mx;zb{5&X2Hmt1nPaY-Fj2;K2cDC;yO?t?^;3QI zQmUa}dOSl*Do=rt8K?NmAXGme>v6n3bHCsl&SCd(QgP5kus8oG41eHF%hz8wI&94S zu}N_L?8gZ#1m!c@<24*+Q@w=SD|(d5=`> zP*ZhU!aJ5@&06*NYUFk-UXgzQ_7C^mTVY;8^WDdgdn}B)j_*^22l~Nmpd;nnHm&Ea z>#z&+Aj#)%9S=J|u)_AC`j}bQTRpoOP$^#nt1Cob;Z12ni;_tI3gRGMOd?qD3(Vat zh#K7?x;drC3s1#mjG$TIZBbLK5n1S%u+~p2;X!)(ax9}ixzAm2HS`L8+eEn>|LPYH z_r;or4OHx?{D~Cj9OtKzfn@dwn?fvEHbcza?QP!4sBIAJ(3*d&ng(kn_F7IhAxsu= zxqR{%op|+{7jf!5e#)8&qUXn66#1c+_vkxI#F-B8K-x}(7Qlq)*OUDj~DH{I~whoic3=v(f$j@|bY1W8CDPVgph5afN8Q$skw#A_#a zFi^ATJ}LfSxh*rJ?8}jPnU~Wz!C88_OBUcen)ta0>x6`XM+}=MCm5?$MXWsLbd1AC zHohcIVo3nX#v<8fNv60+WF&d#W?QB^U$IWl{~HtI+AIby|3$(9$S#kF8sB~KK&EVMu_P{-@PJ^3x)#1jMbQl1wRiXivG zg)lT6#i5!C`JJ&2TO0Q1o)*HoPLAfwrum4*?1R#OZL>zUcqFy>%rA&=hzY)mZiZml z6L|CU2yV$dahpbM6BF3{&9cHA)VxthBvOi$=u)0iZ_Jsm6eM+#H=Rj%R6y%2f9_<; zi;L*~rC1paM_8r`@x~cMxQeB8)@c*+aS}ONFD@v@rw@wI%|F(YKO3WS30A~-)kf&@vQ|8u{5xjsrW zAyepv(tt8_#K4wxmgWl%oEhVVXc^^`nuW*y3z0**_g{AlOF#^+9j0Ir!%r!py9bK9 zhB_elKbhIca3%?DX(vAAW;KG-=IS{V*i!|bF?TA82`FrC>EDLXJiFCwmo$;Ii;#0B;S$$ zUe_iW`jhmKa&i<=n~k_7vm+MB=CWdM@FgY z!iVlgLzHgO5OmW=M~uS(Tm@<>d;Lka#!d>q0#{^t9@RF;zxbtmD>2)2q+szxcAY}l zAe%yGj1PYhNIS3-(JlykBlWZL3`pKuTUqyATbZUx)&Y-W#PjU_^MK_SY?Bb@|C={% z;EV_;)mPNTc*6elA8SND@SSkU5Ey%onFrM)@1K3fZl@Zr)(vt81l8&<#sW{IvV(1v zv5`6RdXyfoDaNI|P0lj?;4|xxKZc78_2)b4YLC4~_P6qo?T>jP@4%b>rGw+aYg2#t zz!&*fa{idY%l)ds#6b)^kUJc#X~v0V2Xy%fA4rc`dP$W-Dp2KWc;>5QJIKZcUbgY) z3iFVSH1bmfZ>(4*vxv^DXsLeslZfldb0QruRG5ap6d@ggF50>xDxE8yh!f1E!D5|& z=s#2p5I6ohd1U&sc|A19AxITTq6R*ZA5)fQr=$2u=MgobD>%k7;en^jvU*_*UN1vG z(IH#*d#yNSJ#}T*PT^pgR5bcxN#7RgG%T}k8SgR-#&vpo4E5}WArq&^uraFhcz&NT z=^^bZi-zMhIRR|!%OlF>MSn-2##{Vxd^k8{qt^y*IZ?G8NGE|a9z*amyAfFVWWqWp zqEI$YB02$ZGUp_bCj^u7aRNafRz!QBh(ta)aiSBsEFqcXlPqoA&Vd&vMG>D8aGn!P zlJr7xGKzkxugY`Mb6=2jN1zu09p!-}8@4m1(1%{+LszX0{6U$(i#)obedVcF_@HYA zwMqWkD&GLhTe=aHArbSWoMeI{Qh^%+(x|?V)R+K*yQ!hK$hOf=y3hR8RI;6*k~^; z&m#JAKUt<9lltPvDmu_Z?uyF(|tZ_TxfN@^!}*x&6I8zYE1(Rg^?P&KcaMHasgB+HD#$a zW`oO?1yvl_0BHw~I?=3hq8Y{=@SHgpdf6-uGsRmlaOieu7wGMAF4qg}QZe?GGNiqp zXy_~`m3+9!Py#Rb_?8p)xW)iH$bp6gqwjGHwv&1aQJMObsu&!-rd(9UJOYbWoxtm4 z4HZghHSE!~Uxu(h+gA$MgtF{DqjM$Ru2d+Uq><;O8oFrFwRIY0g2K#1Qhk;>;ZfF| ztjk;nv#L$ZO~{38%q@IMOk31fBZT0303-hTA6|FHeiFaN{# z?ce#`?XBB)+RoNy`|w9T++O8|!U-HTEG+Vgr=-J3?&!dSO4S06Z*=SKyO_bAh;s&C#%^WIU?N zhxUO8xhiXCbN)OyX{xdQZC{bF&bK(AOL>k1#!&O)U{=R+m8(y>3KlO>c1gv!vmNpv z?6Qo+cRkR+T(rGg9E!9rdqcn76)ASSQy zoLB#&zeuBO`yE`+R($1mvY&`*$JJGs-Dbp$E}f*YORGShVLl)F&ITrn;TnN$Lr!9w>x*|SeM_~I`+P-I@LRu?R5>znP? zjT`N|zT;&F(Z?sNZEJg_ZHmtYS@qz~(-y$DdJN~L;v-Xg#UQFO_4+3mMmF9Rid~u} zAs;dW%!G|x6zIz$QsK8Fq?@mK!J-^$$8 z;IRe6eXm|h@$b@zaFz~l=V?>s;z-}Vcd19#`<3Mh;30cnI;X?>Ky0;v;(_xmM=TA7 z6T>)D*Qhbk^OdJazUx{T3%wmIHtIN8Iw0*pzpx^m))ivx=nSpB?#q`2Evc_|VTVpD zI?%}Jz;jJXNLOhqy=7OOXnKrYkFo*jA+IJ*;{`>|TQ@oprN+E0R9ho5RgO~%c0`T$ z7&5_{V6rA87{y+Yv8D|kpP2(}D)K|;H46cAsmp$N9am(BuKhW5?Uy0^p6wSfdN2B| zvZo0<)Goq;fKO=Rgt}@QeTOG#Teegxxy$Uo08>wm=ub|JE#yMh`>QF&T;1{Wzlwe* z!SKrFPCHt^(f;0Nex?2Lul~dKo!|LA?asaXZEJJAefYy4X)nL@vcFB@tvBCnpHMJN zVBnA2X!GX>B7W{?e@gwGwHH77(YCd<-F9wVvy5UBf$_>)We_E$b#k}eq6%QkJ;1hi z`D&w*UrW|(k;l)!WRt;JACdDwcb-)HK0zN0`{EB~3ZZV<8nWzQEoE$vGGnVIHqEv0 zBREALcgzU2?(Xlmx9;3-Pv82G6e_qtR>Hn3r#9TNzQ23B?J7`1$%+EE&8@Ar;@}it z#0?JiMt{IdZ8DyYAT)zR0y^vCHd@-os^w0Y14;+2st~c?#zjH&;k&TXgWhP0c4W{M z`ol+Yd{lnjx4-baB}5T8Z1R&|1!u?dv4edpv7TtFvGIyyQn#B>l!2!jJ8b+*8y@qq zuVpiLZ13E8;I9tcy|>@K`Ssi46EtgV$FUx+-P~zUJ+;$5^~qP;hd%VIv}Lq)j+0;0-K`PO{EJvV1J!Sm5p{>Jsjub3Sc+^#sJP?AW5mxG+467a$I& z5ksld@4O4a$>VV;^6+%N4to2fYlFSrFoy$t5pGS}5)LCATKKue3qi}Vey}O^^!^GK zp3l^b=m#tY!~Ohp*$JQK@s?E&l_k@2$qo1wYf13Bx=P@|`6A94-ZvD`#^V9| zQX{)^F)6F*Q9B2X5lqvU-%6*Q%dw{b>q=jS?pzP#okO8oU0j$cZl zkm|4_d`VqgAZF{KOV$$YPxJU&h6uhxSD&IJbTj4}ccCi`@^~*&&(iU+E{HeVt^(mV z-@4uY=3oCi?f?7ZKc|ZjU5FegIK*C^cX%%TrH{SP{>^{$f1*9+`F5~>)V8)Yr)^Vn zlxs%w>pSq$T|Vb$IhcOQZbxpPD}(jFy^A}Tn)M+;gy#mznC(H$S1PEYh5K5s<^~; zp~s~xFJK(~*w{IgE%XBjs;OU6r6!|}w+q_#hy8GTc+Qn~r{?g2fO5)D?5O^8%$yC$ zSl1}HEJuC_&fF$bC%xPr{3k@-2M{urR%`}&%K;rp+()!V-r>sP_>$>y#0 zU;flDw_p6FUsW*tz3u+)1D^z5{0PDDEA6ejciUajt1rD|UA&dQDqo%{kooFw{(Ad8 zzwi6ncYgQx*cb;u94Gb(R2Bxzr!9FAIa`hG_K!G|$wvyxW0SV3Mo;vyC>wCj^~=$> z3W$SVf~>3!+j?B24|&y9e}ny`5e$!}M25TRJR-qeBGw?i9xM2L?dxA}U;M&n6+952 zDIByePOeJipJ$$Xy8XW2`}^D4N<0Gz@I&moiQwfF` zFX$Ki!n%$C-Yl(5Z?u+2AZYmM{D?n7+ySsPVU;X$0gZ8Vx_Fwteevghc zR#}hCwt|-5`MZ8c`xF1ef4yyS_bwye7(?U%b~@ZN3r}LFh<;(6LvfMIY_`Sry6c;) z4&m8AR8(^Z!#RF{jjR>h>5H)CARw!ul(m$@1YWu^q#&_M_;T!@I?aUE3RiACuERy$ zO55Gv_fI(gdw=2o(*E|}{x93nA!SL_@-M^##+yCc(r(N5;7CL!6g7qn% z1CN-w?8e8%VT^GEW39(kq7Am|53|MAYF3(~L5zAAV_i-`To!&9*H!TSstT{WR8H9! zf;!J1oy07~9rf1s5BGc=@x1sAjdx9fZ=Bm!{Iz}?>stQ%tqwRE!zNW{-{er&fFnSlI6??N^Z=%oVg2OU74>oE}u{1RbI#O z-KG#VQ_#k;qw;{tZXTrxkSyt_pW2V_MyVkCf_$dsnMk2_@g0OTJ3~}jl2wT5;g=!?FJ*s=kc|=(N>Ngw6=FU1QpP( zZ0xju^qa4@uf6tq`_PA;_tz8gy}mbY+-x^*-tx1Sd^qIkr=GG;4*h(nP6Rs993Jep z-~Ky)hw5(nC!BeFYI|$jC&>*3OdIRk@Rjmq`80TuIlq%)p4q8vqN6-d3;8i~&?mEKT$kb)|gAcj!C;|aj5#t*A zEq>2ijR1k5l6}c#?6N-OV}2cMDCN`bUwi#)4j`By-nl@+k0HpbH*RgUPksF3K51@9 zKhL>xS9FWJO`=Wt5u(oi7i}ev6q)VKt+s91_`cUo`INEp$imi!bjrUcI+0`BmQL84 znm4|$l`ut?h_~+c5d9E#;ioO;Md`YZg-z+9yrHq1&@ZxGq}eZz?X1eL_}2bi-Hfu< z(uUU3hVo8T-eRrjVs2aMrtC2(sMypRMuya5k#6ku{NZQW#~5~0UOCa)-dJtF^2@)} zzWTMV`(nfA4-w}No_GI*{QJGX?Y|Z}H7j0osRw;Am+}#FCi21MJa-#8U&V0Q&mZIAW-1>JQxcD=Xb%4AKmFDA3qSv>+9Tqz8=sqHBcCt$ z#K%9{zW)#Yp|-uv*Q@Qjzqm`PxsXneLdwDo_LAR;l9c4IVH^*SY9LD+g69a@gW}_A z+NmUSL|!~TBgJ#!$EQb`7U44;Z|**H`JEJep+KFBNq+-{h=}MXgE4EquM$_=TL*mXSxrP%ug-K8 z9lYF_mn_FjdHIQycNx%rQAhC>>bmTpSLpN1l3S5>@>9gY7>GX3>Ch^z03&1qbtS}s z3vm#bkcUqj|Dh%p4LVr#MNtl(^T;b5bS)vhh(dnk1x0@s|Am}#=m6xkcq4E}7kug? ze~w%iZ7oT@$|Z9l4Wh!3S%Ss7<+9f4zs$P>YQ`CKf7N`Uu8c8fGU+4EH3#k|O5@nF zf3E{G# zw5mV|wLWxct+DV820^@)IFepm?_S(>mLYF~aq@LOQ$Ah3Dyp<-W#v?<`d>oUUHYsG z^yS)HT_Hrj^wE>zIum^bPkE)>ZVaB9%689%Xxzu>gT5lX@SX?ZS8t3;ErUsQ_erFa zF?)+^bY}Hes;Q5IL+)%M8+N0=Y=a#1aAGI0(aG2!$W#98OzlmCvwWrArt(4D+1fwa z^C#eW)|MawYx!{0@iBK{v4gL$<3msD?eLIq23QqQM}V`YU_rbi9tT-dAhM;vq^Hex zAmYwuPkaUgif2(}xP7B}(>jVgvqO+`*3@M?(7GYuIZ_=T|09^eKMt-a%V*?CsbAF< z06`DGgof{=BF-Vh+MspV!gKS-avtCCLCQ}@L&s+tyIjo&6aC>t5#M;?yZE9tDGuU7 zp@={20csZ)7U~Zf<@o?4UrR=~&CiSbZF*wk6O9*n)x-=zJb$e|G!JX`m+WwRv}qj5 zvjlZ1IMAAO_GNyI1`C|34x-OhmHDI0MF@8-E8Ub%%~sox{w<}ON}+WHg)%`o`DXWyRl8p}8Um#^;O? z4@pY%4!1V0we77NZAY|y?HOGVoVMNjdy>}$*5-AswT-ra zaM1P+4*gxoweqhTSKniYaq|_l<~bNMo3BNQ63=#%_CsZW(paxal^UPy$aG-3i&yd! zO|;4MZDBz?nui^v+OGs#nZ!n<8gFu(03-vm>vzlr+jeIf`5;?Ne$wJRLcEs=J^s@{ zb|G12)J_>&t%|y=)Ug$**|ZA^nH1^}ax_R1$9*lKcZz5_Z69VbC)x}pU@~tCTQs!K z+7eO)f7+)e>!*{PA{1SS>Ziu~u(YsPz6ibCpv(O^<8)dMTg)$}Q9;c}_P|hvfhJxx z|CQ4^f=X}Y-!`(n|0>Thl=&3O^gN8aoJGyIgT6)@D_jv|2`1;&au73~dyUDwu*iCT#H@q;`y1b4|!b zq~)QocGlFd@_d^A==h)=DX8SDqxO#u+QIR@@0|G#8^P!4zJfxBp6Xm9#XYM8oa$Ts zC>&6IEEz3>(669j_3S`F+`f9&#gYPxHGlP*Y?cVb{_i|Ik-s}uN{9;y!U>WN6@1~7 zE$%E!@0w&*Wv2r~tvf!fbfk58qOj&z^LDDVXzygb9Vke5V70;BXZAi^c%t>s1%&KW zU+e3d5Bci&Pn$`<<+y;)@T5a>v;73aOt> zwRi9jrCs(nsd$z8qCj@8YK^X#4wJ-_1Q!Yf&AtdR$rjnUr}@~IP5Vms6m{=uoO>Gg zsn!s5#`@8kIXT@JsA+ATop3S4cPeYrl{TgHbR<2zYsSSA#~Hu&xw^jHcCPKTo$LH! zqV5zKEI3gA(i`q;S4lKA~M3f=h%1+sVw(#RI4*Ew|_!**Pj&%juDzT`rvX`LEctAgsG={Hrvo1fR94_IqDwkVW5+fQk~5P-BVu^KwN!Dxj($o^|K8ZqHh$>$Hh9OqtH{K*5WCG^XTF0l(_l zQ2CO1(&*cH`>&H_x_pN)Pkh*=4ceQy#wZcBQyZm>fu`8fkC|BwLMD#uHEp89nCR2G zAoAk7-|!`{?@%s-c*~mp`MG5OM6TA(iG$fwzZK104fYIwpXt#tIqUFONuffz_#{13 zw7s>d6QE9r+SvVt`}Cue{W>QCsqx;SJ;cJQ^STfyLTOZ#J!LS1>O@@S6;2+q#^q>c z(5@&&lMv___#?^lqWKsD2hB+rm_XHk*nsr64@%MR8aDaSPg>Y}Cx z7IV~&Y3|Wf(kFb+k&b-=&;5O`x29l|!LhdJi?7K#JmT$pm1W1e*3{_=K|KRJw$Ea% zYNwEYBWTf|str4WF^aN(x@E`FDpDUo+`Q$T27K&#cq-#re&W}6XxFqFwcdQ!Ry`ss z<(B2VN9yO2mEX`B;)71xTEWMwM-HGJ?5by4;xGKMcdRiSb5}6ltzcYQ{`UP^63`Ir zMO!Gj*mpF4(s?Qw-`&T4)>aZ1C$s~czxnfi%;Irh5@%j%#~5VY(8nT4R)WfN_nLJk zF6T9$U#Lq0nfE$T3zLOLch#T^B#6nitkH zm#yr3{?H>WHO@V^)v?6|(aN#fHU)~h;8LAG@M(%iBTQ=P3oY?jO-^OMJlamwU--O8 zqqFUCocQ8ga!{4yxWSqv<#7i-;ewENP>^E|RNydQ@0V1{g(1f(V?5T{ zp)I;LMA!XX@OHa?{pq%K?V2v;4%_Xw-fH*m-_ zj!9q=aV0cGCp1Cuy?G22H50(pzYml|^bzeVS!|5HWw5%TjzXkUJ9Jox{AhznFYXgj zA8<(=U_3_`0j>@jR-kwsL_Uxt<2)&C1Rvko);hGvi(+8NQ6KV!vMXq@4VzN=g|ym+ zE;axe;I?{6g4Yqe#Xc9vUkxMj$ur6AsJA_7Uogh9DP*iEn@#G|KFBM{N>V1W0`)QO z@~{4dal~d&hzc?qR?IH53il< zD~#tqjx{4hqn+d`@V7@?=Sq{RL(csX>MBi2;jAYQl&3a%p)Oh=5*G_G7eOIcWlk2M zk9MgseHHv7%BkC~Eq>oF>Y*v3uc|YtJaqyJ-) zoU;O@=%if0iSAGHfG^z7d2-vLC2#?qZE zA4x=A`5^o;>=kanjveJ#5s>-KH+x7D_{ zZz^C`pv`AFt{5>Zj|DNoAm*an?ZOmdF(DIl6F5SX@7L5EtBhPsa_XTaYs6`HQ1%!W zXQCEl4|?Mdov}#e&`K>3@_n9H>#|%>1 z%7fkd^J~iW=!+qhq?eJVZG$)UH06@j0GviyBn}9_##pGw3)spjjJJXjq@G3u1j7lU ze5;e4$0a?S_-dXCCAJicm1Vk6!-1FlrQTQ+2~e49tjZ=0l%S)_~G(r*}PJwPlgXlpV~q3Rc2j1fm&? zMooyRTLK!&1UP9~M?fjE$z$B(NL7H*PrPLV{@@v;KQ3%IhU$Br<5vwSW0z!2ot|Zo zt~4XNWE`?0%drQLNf{gS2xPq@;WKuLnBUQVAohp!B}NNAKWnY{<`?^=*q$TtA;$0a z7I#T$=x|we^b-`G4MbW)BI9EsoCO1J*yQ;OT|u!X!%q2`M|r*A&jo_YHDcJ2Bt zKacaIwXxe3IwyEya1*cZK%3MnJD+RRNi)WfBc)Agjf3% z@=--Q5u0zxJB7!3@;t}*^`8`N#(esRyx-a!<)Xk{Q2eU0Xl#EGNh;H7Gq$6y=NBSh zI;M3!#=6i`=-Z7I=H>XvMxY+~5RgZ;0PM?Jvrf`~oCr~2qE06&C4dvf6j||Jp*j+5 z-vl-uBSDeIJ-Z<8_+~JxdOu>P6k2e>fZ+2I6oJkd)c6q+5&MF2_O;hF zZ*z+u?LgGAmc!qyXXOc^%ts`A4$u0{BGn(d$ct(HNi>6aWv6%)pf6r4ClEnK?xFDU znBe>|+Jk*cGdLvAoGzoWjrv4iPnUu=nRT9m+7wg$9v|^j&FWtPCVb{8zKd6RWJt@O zwv#|pb|vxbxPw;*uo1-iIc^bsPN{G{AGXnd=0(99q26QN3`9ad0gIo?o}{{8mWo3A-|KJZws7LU3<-&;(4 zYj2X7zzaiFH8*TXg>-$qD(R_5g$Dx$p-dMWJR;nx^avG;P z%8QE~lX~)`x46DOo7eq{*1fLZ%l9KFh&=h+6YV3!`k2`g48BjIxHS+-N~6{+ze(A|LyZ9H7k^ z{VEM@@D08BS!Xe+ceC!e8+$m9b7Cp#{;Ir;3WEaDUfH7qyREd>D@<&A+vppI(fwZZ zcI*xNuqVz30cJs{$_{>4<2^<^vJ(Lb89znbrHN@s)bq8Dx$4Q3MCakrw&GR2|WW^{n&sa6D>Jo{JkVGRp2HJx_jaU4$MMrc?9i zH^%_4be4ma5Uh}KfaP(kNK=whqxwjr>^`(v#4#J~V@C>#S?g&;%x@6=`Hy_cx-Rn< zQ((}u3l5Wjx8YBHR9njaY?Dhu z1;rxf!-2AE2Z<38*-i;7m12Q1eg&+>7%=&GsXa8#9efoa-&Da}?bscC6F@p*?paeY z=gv#O<<(pRS|@xn&N1J@B04-!Kzz)HQsbQ!zR87WrnlEN+pU|=wVmrXbc`LhJ8!+- z?%sLBc3manNVsA!9YsKX9B70J!(=YUIGKIJkS|?BKb`X0L@w5hIPzngC_6`?a?az# zhdNSnUQYgJ!D+*Trxg=HPw3?&>^lf7My2qxD%2#e>SOW2Q`vfBr*b>X#D>dHAclM_ z(xB`!m4^*vSlq6P{uYQkB1Jw>ltLBA6@~3Vwzn(m#eD*I+0|7)$V_-{V(H8$CiyHV zEx%a6w{U~d7a}Q}7)mJ@gg>FEMOy?vQI6LH3yL}~I%=}kvO)0&E9yv0RpZJm5Bz~W zp+L?RS=kx$ozJJKf^93Z=wl98Cvat?+eONY^~4LMm`COtKf1oiBBJQKWWwJ))tra_ z#smMOhE{C1fMO) zk4^Ltv+gt@<4?dV@WJ<1Vf{VmG_yZ{0W?2*OCkSfU>-g$CQohfa?<|h*Xx`S`UsIsE+U`BL-M;qv zZ?&&~dwUPseG~kB$v_8A z{e7-#&-Y#*5>W8Ht}Y+7gA3?5<&d%5;bA+3jt=7kp}b}7!zGV1jg{jJ#G2-tN@5=4 z?eaqfFh@@LWxmL(pA-1fM}Q-p@%4bD1avw^95ND+al9m7L||7`fh0sggl!>1&~)UK zXSVzH6GDdrcmX{w$NH)~R6%|aqDyp%`a7{4=Tlp0mvt^HMPI0-(*Daryjt(L-cJ<} z29S-^giOknh64Sd+i7*?e1N%%J(4-?NQy|w)V}D8A&7lc7hL`TTa~p3vzM*&-Ps%K zLAoIG_hl=;wYeFzMY_>8Mbsn5SMcp@ZTSPmn_Cz^?U`pjY&}{`F{mrViX1z7^NR4*U-$b%2zU^}B%7Je>DA6(9h^zZJY%BaOl*F7JEIq}vN3ngp}A-;x(x43$XeUi>R zp91AdkiNA}2fjZ@rBZ(aq#l?D(NB)OPRv@hetU!tW1+j2)*|ldT7mB(a7gh6S=cvP zXnDyjPQ8n$B0pw_w7^V(*n37DwULf{(pbkQQSCFae1}DdWR$_K!A@0{9uS^LsBYN! z;15m6gk1Ga;{HUvuU(%qq?9oawZ7|Oy;%>aS2F0xGU)Z1pXGXu`3YDy8}{v-8;VnY zA4e2w4yTaf(bv$8;o9Uyo^*rn{@vJatGi!shj0Ft zyu&B$H`@O8i|t1~`^EOfU-;$r$(KLg_736`(VzI(%k5*YywbjT`>nRCAm<|={g_;m z&kUng>FwP+?FawxA8H?a?nNJa9EBW_VWDKwPaMlmm2h&kNkUkbb+s7)`Q9QqZU@}T z#u_$D=;TMH5P4Q>U0eimUR(_?+bBDTOmt4c*h3bXi{%Er#&S7qYbPS<;w@CZN{Wv% zeDxb&Z(scRFSdKP?`dtY&iI5mL4^{@zVy;Z+xLIpA835Tul06-n-Kdlz-~_)9u>kj>qjE!%!fs4ZrOjK4@RRBYS`KSKIBw)}Fn# z>AS(KkLR9#x;^{!jkcnoj&GFUYq_}3%f3YLND8f~U(czAK|9t!tZT-#?!3Q^!t)s1 z*sJ1LuXA6tK}W1Yo?`}n78Lao;#2?Ec5bw1Z+)cw$A9BLY@hpypY!+7vNy2axa)oW z#*Oyv-~I9Sum13VxA6lZdk5KGJ;t3vdQ(4Q`WqA+O!2I!tq}2z7?hQli4Gk3lf1i; z*p@CVXv@aScR%y(1-{6Jr*R%1$72O_#YOO%Qh)uQ>ioC^`x~|}R*b`k=?~Tl7|g+T z`!4rFRq$vF#+8Z%|!uUsfG|^ zj!qOLZryBu^S}5@?Jxejf3@x3JMg0eJX1@bJJ&bc@BZE2*8Z3O>p$&a@KkkY8sF*4 zwgSZHGe(JjxY!u|=N~7*F9fq13*+M3E4-H~AD`<8-!gXg+!#M`O!|nb?1Mt+#-{|x zOd4N%$QtGQg~OkoOzFTXw}+*6)~$K-tr#dNFS@R6 zk}AfstJ!$T_quJ-K=tYq`#|Q1qV3q6V}*_02S7DV8SINRDyl8zsOxzyn>R(z;W$B# zF@T{B2SptiX7Lh5pdK<8m>zEdSe8oBX^SODTSL`(eSrl&O@>aWD@s`BMYss5mcC&3u&ObrR;}Z%D`0)pR41%xMGN})K!j}&$#jkpCS9nhW zf{5=qR(gCaI@xQdN>BOl-qC%h#|qp}M5l*$+ld17lY{$eN8JO->`MMYI}`Ea4qSko z`A(hebboyBZ|=|f_XpazBYkkxPWF!589z(SuZ!^m3nzyP5J?Yg%f9sQABYa`w*%R6 z;Pin4=sRt9|4w_bd%Hc@6X}BS!ESt??BUVA+w4KIasP0)9cXMW?>}g}d-sC&@0%Vd ze@A-m?7gM*%}n2Fw;y~{^rlF8cmp9TQlEGBMEh#5ew`QpwsajH+-pbp?fAaNzpFs? zzRK!Pe%n=kSNZ$0e^+wk9~|6m_YdAuDxQ42dj$Wk%F=bXr}q1rGw!}C=#_8$0~KfT zlh%OORjw&AvCLvG^od)1KSH6Uwy~zUSQFDE`4t7i+;Qh;RiqDFk2M#^r>q0g1Fbhw z&7;T{Sd#WWss-u~#nTUEb?iLAh^&K#cKY-D^L20jnnNz^9g>R|7Xhit>&1Jr^l&UV zz~!QWU|RVYAhN_K(xHnD*j(?#_e<-Tl1?5C;X-6h%Uk|zr>vYwNOh+R;8pHC`os)?RP+xB>L$#3t zM6%drvX{&k#-iG0uG&=HUA7(T26*JS#c*S+2bKE8G!L>O86b zY{$hwW!~UXT73nfwRFVz#c|$A@+1`9i!41^qiMxZ1{LJ?grdl&?b%oG-OQ-GWyyFF zawe!$HuR0KUsHsF@0bn@Am8VmnF4!L>`y}V#S6MCy$D(7=6yFdPKk9u`u8wHKLlvk zbzSNLeg0uUujL=si29<-!CLeDrflesaoObD*oJ;1^PZymHgsZ+?|+SV_ehTw9I}5q z7}govd#gnGiRW7ibYFex<@W6#f37w5M)7-ELmH)o$!O z)oxvXwmo(2Y02MG`jqOPYEO%vx}mb@*7ck1sp~hC-WbxGH?FlC(Dj{m^X86H^4G+_ z-k!R7vps$DMv~epg?CGJx0Jtm?RvX@{d&78vOa!VeCI~HzJ0Cnv*Gm&v@b#lj9H)g zN8R@3wRTTuRUFc*M~1tzkzif@I$BHZiaHq?ZSC{)BNwmV?!ASa~<$QX3m+UqEBGXGHZJE!g7WA@fvXJ zlqq$B}X|kX`kz|9#lOkb+J8!47`4<)yjU)dINnq#&I=d z*-2Md)dJ-j^F?5Nfl(rK*oLJDTcWS2jLq!$p7F|Ou+dvGZ9RqksJ7`jQ7_GU$ZYas zP9!Y{#}=qk%6L`qZ=X^FuZPxO=VJ?}Kyy$*F9Lft!mlxoZ)(G|vpZv@m z<$ZFTKmzHTJnFKIAbZOv4%V}2@C5sMvN z{Ab&}yAKEf;;yXk9>q@#?b&`u!j-Vo{Q{b)exnig2|?H&(B= zzx+4;Qv2*rd{#CQ;Lw(Ry0xvHo%X%o^QrbH{|EnCyS{N<;R=Br>m~?Sg~qVCz}%gk z#&f>x_i=}jM<8Uo+WD6Nm2MEUD&^RTcWuq4wbGCQek@R6eRBt$;E?{%?Z8fdc0d7D z*%qLiqR+V+b1)_3e*#Y(qteOxsvyz7*hau?Sz_n3MS1(v`V}Vn;sH{TAEa2_RN%)C z?eNQe@mR^)`d0gEfBUbtzx03m>+Qi?4|IV=pr=4kf$q+=&Gvi#>z``>*8le3ZqM9! zK7Kt;Ch|39tE*dTpJNoo^VoERhk0~i404%eo?#Eu*n3Fzimwf34u8w~Fx~a`i#lYT zMn9xl9qenWbHMCAE2XJY{c|x0JXZ@qH*3t$+}oCDDSz;zG-F+TD~2y7)CYa?O!|jE z^h(HF5zdd|F(KR(PG;h!h_b74LIzM?#_QpU5LnQnDn1-6AmmvXxGE?nHuN|UCL>~- zg6KVXm6o0)Dv*)nV5(DIlD9a91nuMPh|on7>mv`T?#u^;4swbTP-5ipePD+jq0jBm z63xiE59*NRtw3nFb0hLygn^(5rJ__8Qb)>DfEA@H2 zp>5!aye>@0R{xPNiut20V;l4S%l4l0B(DRL41JFMM>*`Ov`BKbmb3{M zxi5{!cMYhst+XvtTXgDA1B-P((OVSugns(z{ zAonTHGiPbAL{c5o*+KZ z1g$IB<=ZSod>{(qPNVwu9Y0@KXr#G@7;|`8K46`RM_{5bMSncciS0;vrgVD)*=;|Tvb`xe6FljW{HB7fP5F~NI=8mcKa5{Ct!w=8;YX2pK0M?LJ~zNv z{QR?g7#}JNinsmqF0IPmKX}ZAXP(i`SL`9Hw%(_F7n*O3*i?U8a#HLgfHle^Gz8xe zA9Cbkf%S2qb;36;@L8jy;|C&LjFBEa(30BIzNEEydZ39p)Epg4Kfh*oD&nz)WAWG< z;*kSBsMg5)qFezu!MXCMT&QS`vM%F8ko7nRau9)b+^JVV`ZOM0a;_2RE6870>K~lp zIFavo?BO)Yk8-$e_94=6V-3jH%+=P~mS{8QN+kc`EB`#YQZ6LOUo{@zo#8yefaMl+ zOkYV>J`x;!spS%&iA72l1`P)c5(DC7Gs0T3i8ly=0MPr#pf3dybwyciBP}^760Hxh zvkY98elDOn=qQLfiSfjy{$u=X-fThR37f;gO(j6v;#*JYtFa8V0*wUpW_Ey!sU&ZN zB*uBv(UqC=2uo-l($$I&)%oso2VbE|6{B)F3ByWssDbCavc9^lnoJSN3?IrF+B*}K z!NR`a)w6z!zQ?W}HKy|z*HE_-jJiQ%PS?Eb0>hfflt8ho&Sum0W0S~)ZICuH4<$7? z9qj|v)}1IHMPAd3=2sgEA~$la{1qvn|C$4iCGsH|S_ce{NLh5rzv1DM?b=vuoa?IN zwfZaA1~T-W_SjrrlE>S)Yiz>1Z?wa2{+3QEI*BWQc(C!|_9LJ9x%T-l{!;t4kAKpi z{wE;#wpU(lAO7ga+JpP|+P!=Cd}o+-Qo%E)i*G)7tNq{){$P9g`H#hERVM_V6|LE; zyI+Q?Z4dA+S4D_GgzZ)*y!_=NnHf;NJOJq5#j(C+X^?G+61 zjsbxbW^fOWeS>ocKhuqUo13y&Jo3I6RLb#iqQv&oicQvIB{}t-WoU*Pd^GVR!(pd9l- zwfa}_lsie14Lk5*^aFXm3d+}dV^r!?rfRMg_^G9Wa$UGwd#3&Mzw?*dzxU_=qqcu% zza48G;}L@HjqSDez2Ey%`!oN$Kii&q=Gk_epEX9a`t{t?KjoNPw$Ruo zxG2Z(eCShrd}1kCpgD|R{-eYgRL>YZCbR9Mn9D*`AjdQ9_}zDN86VEcIpe3HYpns~ z@h6X?ldl>+e8~kL7brX{uYsXgs$GTut@a|`xeUWh7zWe3xNex$nwLdLn z{>pIg^?#D?7Rcb#m@i?-48X-xZYnt08 z!Hk6YaDGt+KVOJw%|;#I&&{~0Gu+Sslh9m8^i^7G(i85x8^aMV5O2_1To z{K%!aC?h1Vm#Y%5>6rm>lE`}N&ocR>MfeVE=Cft|Yn>rJY2pIMN35NC^ue!sF;AOO z^rgJzgIq?ggKOZYJPhZ}&o#;kUE|`!&WXS$(%fuE{p7Q)Y-0`iE{k{^6(OYtYJ_%| zJqG?5C!9F$2NDNF-H56)KmJE~%F6qLPuz{*14w&!bj*q`yN`78`t`5A)?RziPlR1_t*|3zUuu|Z`jD4R%HA&a6F$J#8<+3 z4D^@2^PO7BaaWb+R^viLdj(#R-S>D?fyWbE`~R^ueD$N zl`pi<{`42y(VqHMKlm5$F4%J)y3v0B4}705lKe}C8oNJSsC@=om_L7rG6^0xko;jw z`$P6QJ(I8LSL31`e!_Nqr?J?_*vHu6CEs&uUyOt-HsAyMU3%#I#JcGte8i&-$c%S` zng8(TQTD;He z8ue_ow5vP9e!B)zA=Mk`XNrCQ8K6fLed1A~#X_JV4{Z=-CfXd}92S$unS|T8d ze6+K5Dpx<@Z~7|2Zf3iPC&jFrdZk6#g@~6xTsbSqNnN&OO@3ElIKfFjE$S+oKG4Cl zs1crl5~qZb*aNb|{P`eUjhxK*tZpP)2Yro(rnvW{1^lDvGOfZFd` z=#M&-)J31|L#gY4EMr+AKk$Qm95OkSl%T?LqEl|i*jf?BF#jQ%2!G+SbX!*+*2db% zyvSe01w}{VO%eKHUTZGK_J9s(SCsZGko~0gNU!_VUv%KN+i`~HOe+13oBNcHRL}m< z!#ipv*_0}U1noY7kmmVJ>+-~R8x_#$M7~SVqLV%T$93C$&#J8E z9Vot&bwftSTk?GE)@IvW<*oXic5~x;yCu4|cD-${Ueoc#TfzrzU%oyN9V#F@)HwDb zIpW|rj;Axe{lV*kwFb#vuD8^hWrsJ}L$w&}46hXlVy}N9+VyIvjDumf;Ddk)q%=P7 zMeZnk$R2mBzPZDy8qWoDjRN-r+@FpGrEvyAuYy3{6;a@`f`PV&|F|2da7XjYg+igunt7rIXo7fcok44*we9gri%nFK%|$zI_}tNzBGoFRUS)GFB%`@@hGU`u@qJp zrn3k50OK71-}U5pQ>qIcc00z*qZ4OJ*^iLtYrkw7rtr-UYU4wU;bR@^Uvlg#zVpqz z@J@#65L`XOZyJOA0P!8w+!Vp59A5_~^r1ZwUrNWxY3zM|lmq*6JSG*m+bhR4p=GdgF z4VwI==-{VfBudWzrcO@(5u~Ur|H5O_AM&Qq0+Ek-h~OxyC`WxzU}%r7p0a*yH^REz zPo}cPpSD(skcta5XxP+qx<)$_)Ukn6&xO0d*FLkC9UL+K$9R5+oFCzV$>6g=>3B;Vso zprjz0JAg-dCsX7fy5LTr0w%t*)<46`dQ=(|7b@5>`UqP%8fs=MCo*H|w~#E1evY;8 zSEmlnLY?I?jBldk zMT2TnR=PE&x`5~Ivp*C_^(Yd3rVJ*j+vG*Reg9D|dCF!dh5=R1z(sQEP@N28qKok= z4|zYea>OaU*qiPkEU5!^B=nC&#$iSah~)3gY(E^Ksahjpj@D|SjMCv zActdB(-D}IJ+pB#FL++hp=X9}XmOIfD3TSE9a6k%-58Y?Jo(5n=^b&!1rPKmZZ8oQ zAbN*^OaaG13hh$|;`ALhMjrufWUA=<9w;T2f!g@Xa;OdRi1Zh`PzEDd6$cOq(`~8c zIgr7H*L$`}Nem)bc~GLaz?ic(hz&w=q_p$holHw7n<4fwX1C$h`A92o$mW4Ju*oEK z^RejHC{Lt^S5PfOCjTf1b;mO{zus!oU&SD$7q0`dl(<{w@dw3?$DDsl)0c>ovv~GN zO`;HTNC6!Ws_g6W5g5i(1m8xGw@9FCY>~cBSjL+o+gordyR6;ix`Y&1E=FHvFA_zJ z>mqvl7<#ZRVF`KWGNGPS;`xDgK4cVyiaaL5o$jzzq{BXw=nM#_@u8o`S|E6yY2Y!xDwWBx?! z1Jdh@V+XOx7a@k-UXd;rpVUkrn#f@v`yt1y?POoHU*nxC?}PSnN9(%C(S?$~Qci8^ zycg#(?ncXd(oJxWeqcXigRJ7Y@?M|6YEAw6Yshk6j=c_BM*Ri1^IdP%i2@$9WEp4H z;cLs#2fCFeq2E6ymLJ578w||3^h=w4X|3qXaj9LbkQ@)diS9W1BlU+8Dw~8~mQYk*fD{tK zzM@X4WYWALwbW3^9;_hzaM6M@KigTH19IqJ430#l(JL>hd*~xJ%Ge}>VI7o*Miz@# z{;9WJRUx4e=+HBe&5gu=;^@WcKnske$cn0GT-qGHiHKmc;UK_knr=B5u|Bdm$mvh{ zfh=M14&jKX@Bc!NH#w0d5mk1$uo#f_N-aZs>x&xldVx!&;gUDrOb7qzFa4>Ckfm+v zV9u)yjVB2skmnB_?psXS`xOv`gr|J>s!|beA^Xmp`QtuA zT_kb$%CO9+bp*U+{Dc*CM|J8CeGkE)dOR-A;ce)fSeo-wn5rQ2aDDFx*-7;^+Cz^~ zVm_TJsrh{8i1s{($@8zgbxR<6UjZ(Uqw%c>t1H~mCFqECnwG%Hv>-j(iyrjXlpjDxjC0M6JpR=50kMFdr<)g08?uM523$GG_e z5fBrsxxz$ENuV3M;Sa&3?Td96IwdI{a+h(uIdGOv z^44t&NU?)j3U1)Kst24Vtw()CP{`Z%T<9}4o>`#p+!F{Q9Y_*v5nw5GU?*96!bX+R zkADgtD?ugWP*4_ol!9Qs{}!XM4V%=)!9Du(+YjJ*bkwOI?vvMkX0o01a;A$S?rL-4 zq`CDS?qlb1fbW6}<#)#6Z?tfp-jy;|+m1}vN45zmgoCK7JZxS4UWz5 z#?|)-!Y5O@5L9whMCnrCY&$&QjVSXyf5b(+;M3L%t;C#-0+H%(e89R8b(F?+fPgaz$^>;~8)J*{xeU4QpebA%f1xIMl@={v zk=5yqyDV9#aflIPP$?-+uvON7xH18sGESo6D&YQv(FvVZxiksxN`jY{r z%X4Uc)=MueT4a@YY@%5ws*_a-jCx0k3kDu(V8F>oD;p>ED%oqwOJSrb147SeC+&Qv zr}W`NVF++JSxF;h2g1CynFx`n8{{j9wIcDdW7ivH2`JtIWJk15S;~Dn(XkwZ=mZPQ z}IxLs;tdo2df|e89iwi@LCjwQbpsyq_u0`?n9)ZcE}^m{ZWTr z@}b+$q{d}Q#Y4qJvf-?VFpL|N7hCZHA<@j~#;5QyU z@mHbAe>{Q`0W85@v?E9!j(bw;qMU)*gr7ko!J`7*m{abQ#@t%A0%;R|K%QA5Weggd z))4(NNcITweEprjT1|BfopI96-_+sp(`Sso<}dGQ3Q{A z^rc+c33;s26Z%QRXI@ikW6H4%Z8UbqY<|ix4(5@y8^6$~fVlLkKGum+=_VM)kM#^Q zbtdTuIc2UG^nLZP*A}S9Q0_glD3s1mngCBSQNtu%=Gm|`DS^qZX5OWHG}g6k?{C^L4JMZIY7 zZdH6?4E7(62uTDTI6m;b=|n$+Ctc=XPMghCtuvDIrWX0A|^Tj7aLb>Qr_%+ zCoJK)h2+dfm}_7)YV1=#W7A$>%n#`$W~UhgN)@}&Jzyp zQYs_^zH63!&{y;sMNl@x8q{kB6s|+dSg4l~t{dfz!4A+@T$G=}!!H#R#h}Wh7wdRi z7}WH3zqc5yW;`NNlYCu)V~y1^?*bF=OBc`R@|XgF?Edkg=t%ag_)ak&oQ!8%vw<3u z<%3^*@XBAK=3qoz1=C)$wi#iLi+Ed^V2E%iK5WGcAV64eI~zNqYyLh~zMhIZ&J{Sg z!<=xMadGTpR{8i~R0cT&qVQejcpmdK5Bud$w5Slr*2o7OALvks)HalEC|I&R&gWNz zDM;jlVH*mP;yZk~`^VOg%|Yx15oji}iv3rNtM5_3VT%-|6qV-#b~t}{w19K#s1=0{K{DQhMvg$fs8IbZhilN~lDoJR zUhedpHhOc?#Dc+Y^OPqwOG?Qk)nBZ)kSU7IA{JaBRYlv%Q|zcm)JKgEyi5K@4V`>- zAUK~C67s-U4I^G$&p{Kc&=UuM57-V-u6i9X8B7&CbSbfq;afc~;U7*E?xcftS|gjo73qhGw6=c@Sj4Zs$ZNvk1Oj z&N0MsJ_zJ!YQp$e%E4M!eN@FuAluiN!|rGTsmA;u8JPLtp9ePg?e%{F#ap{oE@1oO z-`J`!^6YC!_Drdlni6ncFTIThK_`iIYCI~$O#%H&cnl;hr9XnpQ%wi9!1jyfoOk&= zsy|ChFW0@~%XU8s0Qqr-Q=PCk+l?)L2x8M8qPu^v+wSk)YxnN&wmp9bHQy+rVAgMY zA1ctvWaJCHPvg{d<0fc%SEs5BIEV_*NPXo<}PDR(gwP zDhV_z_=*6I^`mO#y^d9g;6hA}6UaeJP^@)=AM)$AsjcuW64EiOKNrI1B6Mx&7~4|+ zI~yCK&30{zAI#WNDpJaiU&N1E@Zmx16RBUGb)^df+jRV*i@hXICkKmXd3<4cZP~sRdFhI~ zw%EZ1o9yLJHt`^jk4cR89$CRRKH(d;;&C7jD!)snJfiGbjGt%g`EV#-XNOLIKfMHM ztqpq!JQ`jxYNziF{R?c?U-wq9q=lf^yBNsdzo^WIAG_gr^-ciG13vas<=C4h2@y$@CQ? z;`l*KGS!D{mq=+z1e(O9#r&v@34dv~D&K02uL z;mIa{c%*;^;yZEqswdi7+;G2~AHhCz9cVWjxAo!1v1XwOf*7p8Q>~BKMald-$xyk=i3mu*;(ld))Dru2TiEr+kgt z;epBr3c}Xgb2pxDH=cc|J@foa?PPPOy?$`mzVTqUJ=j(7dMx{nPTS$ZaogQH@UPMB z?(epHdk_2_#O&p(>Tj(nAW?wpwWc8)9y7Kxr+jr;9AB#n)LPqGUu#>u1H;$09o|&{ zeqZB0RGD8I<1Va7^S7yBYg746$@}?B1z`Nr+VyJ$j|6SHc+@xvXJ`{YdZ0i@^TiK% z&~}@)4srM+n29y-Ih71~o}DGQbHM6=Qb81?;K{!pMqs7>Jp?Z>bTChG$6E)2Y+yWE zg{QK6Wy2R1{;D|zX_UFsysr8U(ay$pyCz3m+f;sYO=W(tA@4FvqaVS*HqBj~|2SUP z2>4bM;I7CXrQDgMXM#NYNn_YHt+y?eH>~f>cPC8*u#8{D%CpsebrSn7@uJFJ?G0B1i1bpT0ky>n6Aocl@##sIZ)5k$D}_Dae~SM zP@nu2sOrm$s2Z)wP^0!`HycDrz>b%Kv<5SZYJn-iJC8WA1{Dx3@nFZ{Gh!@L-!)y5 z=cQ;w=`HHpEt+cE(x_ewc@ARui)}cKnUzgBKmeOYbv6Vzehpny-$h%KWutznFO%n2 z%UIex10LmNZ5Pnb#RQkMrN*^=(-fcBA)MDzoUx|wD%y+A^*_{%G>E9~VY>su*(m&) zMYC8FqHGL|swTLp>ZM7ja^CnnG@;J%D491;&2B2_)+y@YmXN z@Hu}Ihx}r_cvFiD$*`93s z0`Ws8CdnTvpgi2)ZwE>b6jUCo4IcMw)cnY&a)yFn?!@vV3p&}-&fnH>;9rTWzC4#A zE@<$PDim9Wt;dJQ>L-HOy9#pe9i6m03T*F*fB)d59V&=D-MH3nD8RVM9W@0&2iyhL zZbT4Bkj3K?oXij9V}4{}U;a7L+8Ez5hgdsWYg%{&nu?m*+NOdw-?3D%$*2e#70~hc zf$y%e$3uC2Ht3UKe>zl_79HA;^>6T*xBDDUA9GoOsTIz=;nR`p%_Ne!iFZ zh`pOJ4tKK&YPHUpO%)yJ5zPBnA*ttkc?Xh(nG$6B&Z)*kkjvZf^v5{qle@eGwGQCu zRlbkoQhBwLK8jEa>+z)U@k)CDr0#aFCBfL$US$qMkDaL=d4BKvd+Bb|&fK1A7+hV8iy_=#uP zzN-8xb%c+M@l5{U+paaotO>?|jupl$Kb&dIXQ!GUwPnmv&-lX+XOiQ4j!&fzvfZp% zQotUh@pC-w%O8jGhnBv|$nYz5vQ6y>o+0MJx|HWv^>oB*J?Vl|TyDU`8dF|tXq`1G zBB*v?eZ_bzzd(2QwSSU1kWRh?oOoC_Rj90aeRJjw;Y6OkoQs6iLPHIJ9cB0^wMNL+0oTQTZHX3^7}){~G(KAu#GhUuF{? ziy98PvVO=r8JJ@}v8Yc=wUZ#1)X7c82=z*ot#QsvH@$9~cjZ`wT@R(jE}m=mptc%zQ7yXdg1p8Xmj3F8LLK z3IO5s{X37ml)B#KLgb?@vYrY5jr=6aCjH62k}v(EK-cCRx!PjR3@gH%YVKUgR^5QC zz(v$lZROR{W#F=8Z7+XHSGlXjC;E54sg0RCXdCTNCz*frn_q2T{pxG&X?`A6n?88% z+2`8LTQ?Pa@oQ7Z?b@~L+Q11SWVeWT;NF9K?USGUWP9$ZXM6{eyH}^OPPGmgBFHF; z0K|cw>L#iFr23q^Q##UIkhryVr1w;CSugSJ1!Fruo2nGK;0L*#KyLaMZ+W_{b*P=V zW7`bAdpgx9{I;U;El?>X`R? zJiM??Wp<#=>)Y+smtIy|^(P+v_`7$JRRi`2e?6LX`+|YGGX=r?64#jm;nM?u;s36J z=sm52U4BLGgkO2v_uZ}g$EsIw&998{t6B%TKseROKQ1P?2sv!KdkVs}9`5bllMP1- z#x(ETdi4TRFrfYI%=@qDNP6PJPeG-E zqR;=*7us+A)33;P#OVZ;><`=xJZsmlZ?*6E-tWT z)B2cO`NHGo7tqv?=AVm{lT!t3GJrn$Mag~X*_X^d^Qe`?dgQyGof7!7Hchfgx(~&R z=fopDRJ#-9@gZ{9${P0fLnEiL&^J7LSU&yL|LRxTXMW z3qK^n7=0|Tj?f`Hkm1PUcs$aDAQzJCGuF*TH+^G^=5k2ozXJugmQ!n)=&{##WFN^^ z)_}T`J*pc6ax*?zx~*KrSD!w&TV{+z;j$1p3noi6hc(pc$1|tP{jXvDU zDP`wfnQfxVU`(}9PWgIa)hEzm9(l{A{$#pe{j>2p2oRq;UxxG}J>H!|x|Hc(mHu6H zp-dQM;gr6h0w%B-4>k?iX(`)r$rpY29z7zxv1qYtd!e#60RF!)2-kB>F;PC`N)gyt^&0E-QD)whn^QdPWaf(N%>2!f4Tj! zKl(%MJ3js?1u@*g;smrIcf{SGSUDrG=ThVtK#j{0i})^y=eKf!c$`#v3~j)WPn})Y zXrtC^O>&FAbH3v{&2GVP+zJu47Nobcfs=3SQzb*aU;Xwp!5=}(I_pp;$*;WeTKmN> z{(=s6c~i#`!HB{toxJ&_v6Xi7Lkc3k@B77%{^4;iO0!~Jfc{P7<*WU(eFrJX5J2te zr0x^E0@>bw+@0T-{H_9neMMxn*H&f7)L7@Kf!-^TYqWwsY%Q`F&60$Uv9*k{)jS z_--ukaHRVpaHp(!P~FoS2Z7Sv{X@;|s^(Ju&>A>!fE0mGJ{G;uE^}qR3?V~Th>PK{r&lG5$t+n+Ww|`_rda=ikY>D`V zL;Oe(e9ZY>#{2OPF2rOAesQ3wet2hyIb|vpv!1f?Pr^*x9xW8%yn?H`O3_K`je5?Z@bPV!| zv9D+Ys_#JdE0jNmdN<2<#T)F4&9KzETEqcErWSf2PBvEO@|XftmJQPy+Jvrw^ydvl zsLd5Him)FEwzoi?zPj_>@g_OXw?q`>2?w)L(jNqo7@1dz4N%Z1Npg%YgBa!>&HXW~Q=S?tsc;i*~M{ ztRg4N;Bl9xzYLqA`Xo2S+yRq9lYwh>`p5M`a+o!+H!Q0LjW968QW3T3ERl(f$`gH}Y z3ZzaI@MwA5KHP0T@)JMXzVhWa9bjMw$1gwN;k6~+hQ>rpWJuS#BERQ8^lW?e)mPeY z{Q9rAH@@+v0=w*TbKj91VZ3TEtGeI|j&5c_-?GOEn|8iTu_H;Y;on|fs6tF2poA(k0RkqFJ)O^De z7{|*WdAfb{`4`%4`Rdnx^UJn_$4xwsh#{*cwU&nSB2~Of2eJ-^l|Ru%0O0N?)dY0@ zAlT-HE>_OkD<6BY-FoU~`-i{&o9!EK-u3voc=ER3MpmmH-^LQ`^9g$Hx?I+Ur&wafB}&HzKyvqQv&T(}Yh_9b#h%_yAXGPPVPS#2D)m zeSXqY79!x1lmqs@HN5i76h0wLWJi>($Z$}3Q$mS|?Z|s<(va`eOar?@@uqdOi#;Va z!r{a$vLtB~lOOrO*y+H!2-&vKkKy9Rs<6=5515u74RkSf6SW62cP~};2`bw}h{jvu zS!cabA2m8r_ap`{HN6rAwEC_W&ie0;7nPC+5VT_J)?pYCMyn)00HA%9p&J|PXB+R+a@PK@|cJc1z44NiqaR=hKv zoe5s*`DWe@=FtG(5IvHdg6ESX15pwte)*^X<9qn-13ZP6$wRk+gSTww@{Q-FCpq$$W43e%nwW zOfV6)hRxLZ`Ead=z2m)hUvek1;c)*!+r9gyg4caNdU3{G*Tchhw13bJc6a^!^a*e2 zAFz&A+KB=f-i9a0*x$X^Zom0P+gAXkjnHetj~%FMzFEK@eB$Jxc5!hWK4e{<9%_;_ z7Uqciw7$R)cxIjT2 zeFszXvbU@8-Mg!c%DuL$i-vo9hwa|Jf_24PhleNaKxLBxx`W-rwtIgs+B^{5zwZmr zeYHK*xCjc5bTPuaE{94FPF8%^{=xoHyL*4X-MxF?7b>*n&N#NInfxc;$*<_x*TwDr zF3+W_{y^(RZTEM@6aP?o#=@APvV?d-2d5kD zAAjj35U?i0Pll@YKXaxzxEKlG;Y z`#D1VK%lOAPV|0z8~><3sC4nRsAbiL$4mTk!zVoFdE!rA+W9bl|z|Ydh_w7eAsCFrQi{P}tMSUV)8hkLQ3@K9pWY@8^T}?zKDj-)#5q-_Z&A zLA!nLt@gEVe6{`KFaP8A+Bd)4UVHP)?Q3uRX8X$5ztp}g`pWCS)$ZK;dOKL#Z^x^$ z(T`Z{wG~|*O@#lKI@IIq{CwRU|z}JSG(H} z_EhH1AIFT=hvtzkm`AnpIL|cRQ(b8JIM81_%ZqN-Hu;$U7-t+Wu`Qe`;6jHkG)@#0 zD~P*uSHYCz&rT0zvWRyJ=*Rwz;W5HB4pMZX$Hy5xi>G6Na~5@6ta0pd0mfq$+nZbn z;D7mImAPQG^5L*!f4c`C4)lj+_4n*kPqpWseZKwjul{`dna})O+gH%&yR&M+IcRfR z`mSATzx(%oU)#REqrT$!##$~Ahl_0;FX&^u@gzh%L(MxE>^1EE$j2BZsd3?Jej1ua zlzFViLlSYt$lib+>{t=0Baf5zGSW`s1D+g@N4(R(m}Q3|ZND?Yqeb;Be!NR2-=bUA z(8jTc7}()wl^L(pZfbpQZ!197akP89tNFU4~`rX^@-qBla?`*5xJJ@X>xplSA z41WNE;edU}E3@U)48ttITnm(O@S_1aErx7cUAT3pW3dd17u=Xs=aiz35_wBR6%1ZN zwv#Wo9c{eXOc8uqOUfB?rNfEaJE$~6^0E4c?w6tOe~OD+U8ZjlKDZtN`zSGvNvVjr zDI3EtNU+5vT*XWJM=v2A2%ki1xrO=xwFsPy5;2>FY0F6|Z7f7-$_trZUt$nbX2HI3 zeqwx!^-<2cM2H1H5NYTN|2aQYr(`M_+6^M|0(tqLvUMjJ-qa#Hq>;QMENv^d#ogWW zz>e_hLcNR2+Q8!!rL>drvEx)5{rXm0IeyTNAAHk1-o`&@o9&-``SteYFTd8Fx^+uI zT?AjxJo{|Bp#Yd~IUw-8e*H$C95@;3WTEu#{rd`rKi-~y<{6zJ)Q?VVob;D12VhJ3 zI>tY&QBH(P=!27%`ywTX&@p1UMT%Jsia$W=XN#@yJcZOK7j|)CRK2<4urB1gS`Xto zi3@$F=7k)7w#<+xjf8hXZ~v&>yZdIlefLhgE4rhTJaqTo zop$g3-F9CA(S!RB+FNhkZr}LEo9(STcm4Rr#-@UVofQQ~>+QzPt#Re0c0wI%VhTsraiouB~?K)^>YZ=~K6M+A~jIZ_j_| zDbdsI*=KLGEd@$j*uF+cc%ZSJXguMcmGwjQd1BuB2G89p;1^xj7@vOTDf@xD_I&5_ zw)}BT<6W0M8`8O^bW^;o>su;sv{m_Wb^S*bw@$4 zt^|*{Lws=1R`+kW)BV@l$3FbMn*XcA2jH3EfS&=}aYt<=o&3 zY~Dj~IH}W>3Xv5}NHTRM5#`u$3)M$S#j`QNM;#?^ppVTY+QIh;0JaL8U*e>| zwf+?7MTH;dATQl{YBEOksShCEm3y6)=R~p`qwYoIh?YZDb7Wmr$UNc*t{FRti8sjz+MO?mez=ewSK3cB89=U^D24_+`R znv)B&Y%sDwz5B{Gl}v?mE|NFS94Y=FNv>^B(T|Q~T_VJyO(55%6YdlcWMo~Rq!Ra# zd_egm`{>tGtb>rPKsACSdW_VsT^sve2>UKT%78IWGsV}sbUw&CZ(NWudu-88I=})zn zUi!F#ncMB~V6T1X`Hw`Pe0V6`Ckm1e+n3*Xt^Lp+{=@CNKJ{t8h06!%xXZfYdvwb& znjWjB%Iqmd^6jtW)tE+U$&k(Cy!U&}#lsPDAs>0}>Y~d(bfUjOlVmRHgS>0H`o;2q zZ0D_6^GQE`J{G3(UwZwG_Df&z>UEfrIw#C<8akn$> zZu4EKzxFG?(!TOrueWEPenEldQyK^#T9a0_<%XQ^>gmsS`pNk&q_UbOg(x0AQ?{au zu(P%8wxR%>&iV0!P4&qSP8=WY`rQe>Z^j@pWKo!#f;Y zFz{Wq_;g*-V$E006?a6hDe(R1GaqXI>EHcp?I-`iXEgTJc2mJLZ3xov^DPB<|H_~E zlkNI*&nb{okaEnoSnT^E5M6lON51B$)=JIwhJ5F*LX*9G`vCixpWCH{4#aisrOX45 z4y@+~_juRk-r*g;Wxw~}UfbWlFHLHZYn$(ec7OQR?Rf-*0Q{zc?C1Pddk=KM!K%7; z<7xTvn)~tvlzpP0kM*Fn!rjc{gI)Enpk20dp@Zo>dc)&8%pc$G5c{3>RQ8OG_1If^ z{$1_3GtQpkAN@GvPO4-lW%OfiP zz68VmTZAqv2QD6BLm@u~!%UOgDjhB&;CYT*a>`ia&<|Uf7x&oB4_OFV;Gnv^Vlh(e zNN96`9j{7N6)lpZms0O`)>ma$3zD`_)d-A3#T<=Z$GUN55-A(mzOL4mlSoTVuCXCN z639jSAl6|Rp}bhVEFwdR`pAsTl=TIev!m6OdT|RjN&V23gJCsrJNTuCO3`$DQ&f%h z0op7_HRs*^4V+A1F)Gu2nyL!2aZB9_0_35IuHI~NT9$WQ-5``MrV`S!7wUed;Y z=;szc{poLOuYU5=?aepeP(X0po_qeo+T0bK5fq&A{M}*u>Km`OKl(#|q*SF+tr+7S{i%{$5jVTgaa1mzc+@uEc0lNpC8uAkVkztYJcAIKm4DjajlMb2BdZ?`wU{-(bbf)D!;K(S}}Ep(+cVP5@qx-|JuHk~Qh z@i#Cq&jfj+BO9~1e>7?)-fIUd9Deqa}e6&xS_ z1vm$l0GL2$zr3~1_dY*pZ|#3wz9#t6Jo{_57&8q;@`=G?2slfGUVODKpYb6}e*la} z7IZPKVRZUne=z0ly`;v6?&q-{a^fz_V4}q|DfG}OMxWzp;Ld?uU~7ge&Xf! zXaC&)p}qL(EA8aqK)N)4YQy6fjM*^;bB>cSPd4*)d%n>2_Mv%@kGZMi5Ak7aVm$vL zk6PAT^zGwBD%C=5yqo(N@gq*uNkk){LqG4&ui5dg$^Ov;^?%r|v41Lv<=kyQvZm1@ zzl9y#gyfhFUvd%2v0(e8p0yvg$Kw@@$(;S4Y~T3C=i49sga22Rt_~l_U^u8IQQVx| z-cUGwnFGuVLbAghF(!$G1%(d%$CM|!MdKU`qH{;IiDd&W3WnndQr@qm^hbs?4kT;Q zztwB=Kv!=+ufRmEd^&p<*{Kos?~X1Tmz!t((f_Z?g)4R zwdws}WU$w_Gu#+R+F8cN_;p3A1Bo z|H5EcsCGIm66EPS))Hk`W&gp_iwMqer=%*23TlT$=z1HVzZ!Fg(VuCT(DRlw5rT;v zcGzUWL5<1;!y8+-+UkP{hPhj@xuIbA^hW#ppZQ|@X$8YCee7kQaQJY|r@#GE?WLDr zX|KKZH64Skz3{?|I`L|aaN;;QYxRGX6{q# zUivt4Df;e~$_{?S!=yd6B^c(kO?b#@-#?VWU0q4B!X4BRw2TQozaF&tOYt z_+|wW7up1n{LC$P=TU)hqMIFHdIZjpWk9jyQApGt*qzC(I_eWm@) z|NK8{pZT%R_+o>{L9mCPf94t84}bK<_QU_$ztL{qdZw*vj6B9%XQkTVjoeJ)F1YfX zzvMp$JSxgk`_6Ycv0JpVcIv<{#;JPE!HMYT><)%opKWG-=}kO-1cS>MW{zi-nQhh_ z@dEaEoJ#rQF{oo*UGeL1{(Qhn{OE`;K4@`rApLQ-njq0V+HpJr!uiL_4CXM_!C5=d zTpS(URsDf{#kY0v!zY_EgCC&S)l%5g9PfGDYnm74=Rf<~f1~~7zxY?%{WtFpK4Kre z_4KXw$|qlLf9B8q+4lU4&nxIujeN9_%2>@*vXG;M6j==pir4WUbf0EJ_yTK zr>)2*4%%6BX@1=-@b^_~{uxDlmozdl7Yw_=T=9yI2U_EN&o=KQ@IA%;wMRZY$~RtU zt{fBP-1C)p=sZ4_-IBH4nqTIFhj+w6S5ET-4RiBIZo6tdAK>LS?X#I&3&tH1w0 zdw&|V+mhXNV)+iwJDusCJ64rcsuCc95Dn1;wn0GCxWTv`;l>eex5I9K=m>}HPY!>v zgYXxd2%B&O(c!Se4jW@Pw4niY(}r$95`!i*RY@g;hAOF4ldA5mJDhoX=h6S)|DS8` z{XFmco_o%{bx)}St>;-grUbam14EMg%tbQF>4m7(*J zX=m*96A#=ePV+C~07Yw;Ke{EV%Qqmjl#%l~_3Sc%fL(YOf8ZsvkJCjX z{%IrHMi+VI1J`D2ZIiXc>oTTh(w2(DFuHO5m9{y%+SX;98fVwq#_UF0U%k;bg;zHS zula>0wqRpVTh=tM%Xrw3!LuP3zdv`Ze5)(hWze$>LIy0ewlZQ_2W?mMe2opG62r+) z;3O~lUBloI^5B{;+s%|loWF2~Jt&JxrlyP-)lucs|Exx#w+z%+%!CBXQ_umkyBK0< zUdkC!AolJK>HoA0t-y@F41Fb_FB|Qu*Rs&d3^3|-(}Kb1|TVKB-eO-|OyD8j~ zo~{a|6B+00GThfNpjDP{W+gZEt~T3Plip0n7XsIOQ?<5qD4~ud9EO|fVS^3*P;ct2 zvN6adk7u3Pg+qv$Euq|3tOws1@nR`Dk=TkqlR3(h0%n1^aXLHQ*J-Qqfh7RU{wqH_ zwNOX$1N2qj5FH-M#h|Cpa>I4hmd|+UkGCk{vn{?FZ}>pW5if4$g;QRZgvY`0oT@L!bOZsI@;SA@6xnq@c zc=Sl+??~P~wV7@NSaU2~w#!WkPd0R8==H^lXKlIo@q#JF`kMMH`(U#~jb#!il9{Ig zykyLbK5O8YmUh}(U%S_ic3+|77YeTHZwH{6qzWHL_vxTEzZ9O9_B>!oJP+F_-5_#6vpsMO_It4m#w5NH$I= zpS9*uFuf?p1}_51c#bwhU(KYb34vT^!i+ zD2Kzouu`sMwlRZW<%5u3jwy3$d^<0*c)rMkT(mo#(6gh)PS^8HX@V6;I&@qFyOGuT zm)sZ(WrS6pi$Ez#!!6y5e$Zq1z~6NITz^prnK^4bvX|k&3?nmM7)?CnUeyJfod8zk z2iIgQt*_QiOCWM3q-;vVc9PM#A%jX6`X%25cP0z62PP%%I7tP%?Y|VyMEGPuueNnf5LtepCFIHC$INrGqNtWEo@aP=*m- z$=hv5_|JGTPzIu0X7+e2&Wt>Ua6Lk2CLMXGD?QRTTJ$5ys-7U8mM(G_GvjFbt|-D1 zR_)1$I{0vn(8eOOij*tklqZ0vGN6uSK*`1E!kw*LRa>y^WL>qFk&nSE*_f5~ynlZx zk#*86580A?MMZB)PTiKBIko)$Bc2jB+1#I3e{@nlJZj6(UBmFMj2p{RCL-jQX^r?%NzP5cGE*?@;` zE?8=XJY4%$RiAaqxG7tFP5pCI@f+NtLr|Jj63Z>I^35W`_dy{ zbz|-O&f&hdujAdmuYbc%NYL#+c4m?AW4>-5^-x+(N%5<%VrSk+=S^yhHV(v)DFIRQ=80I_tl1b(*K@x^`xzwJZ`Hp?$@Me z)``#PXVNYoCGj~<-(z=*19pUACy>4Nu}^)Ub@n2`-Rs2z#p=X1(mCrsQYRcI6TQ^~ z1)ww!KphnWG6+JF86P}^eGux_QR>#vmoRyhmx}_QQS`+jk#l?jsGY;{c3fZ$fB^CU z980#ufs$uZCIek$5-t4>oNOf1-m4j$pp}PiO&4 zOL-QCcA9|b7&K{J=O7b(ud4x`NF=D+ZC3BHKQy#`5*f0~s%nF~CKmv3FY$W*r+sIZx>|@ zZwxH^P1p0uPZ`XV(wkWZ0P-nfe~!3g!pR4Ei!n2R(nZLbwlcH=b%#_b}Cqs5dot2@Zi%W=AgQ8PH=g~2it8=xUw^>x%KSd$^}^ z9<;r^`$~VTJTjO>&sXyJ#56_*ThAS;yljgn15Rn^iq9pdA7|X~Hd%}Z($}%dW>#?f z@JTyF7V$h(*%)6K*oTMPij$n|$idcjU1s_)>q>XUW=GfD9gRtQHp<`OuXdzs@9}$ZZ2va#(hXbl9jI>BH|K@qVLQPS z8N3*fkwJYtah#Jn#~EAc&;BZ4A2W1zDX-|&_IwQJP<>ff zNwtgWqB`;t@4ni9_wqFWZ}T6K{O z%(8P1`)0;xktw@>|R z|8?@-~+oNG}bC$0x4_&J$l2eJu0h{9vSr+i~f zAhbOIO!%2a_XFnHA*BZrUuoT)W4Z`QoAS%4)3eRxo%$C;i!yvvupooVDojKihFTyO zM+Pd69YkBk5`O2!c)`64I^9r?3PMj8W|`0w#HS;F(p{d=7q_?=TgFSk*>p&L)lF^; zIR0@j1XaA+G;kT=1mG)}>(lQGXq~}IjR`hh*R+U4 zLkB%ms|%bBG+ib6iWYdRi_MWP1YE=k^QJxq-#8NsF0{GAB~Jk~DbFDz$svgh{xF8n zGmt{RKxjYb9ZRsr7tsIA08Z75#*J8gUWfsBb=851%nWQ1+YIK;5nH6FGn zGC+2ZWWXL_;58Y9OY+O;!O)UiEJwf~%}lurEc{0@9FAAjS7i7f@}Y~k7+riJ(S1k; z1?ymq7_l;Bj@UCuC6i*g**vm7mSptOpJnK(91qdK z;zw7P4*#Y=h>LCcr!a=&8~84QWtz|J;umR$@M1}VIpT{7A-2csA`u?NXX5Yr88{T* zBK?-aq}&P2H52V2hr4p4{OCJ0q>VgC>%+nT#ioEk8mKla7>$DBIaVDYre>lH_Ue#- z4#*S1%p)J2a6)JVL!WyVs7w=>_CL=kPS`o>OK4gtE@d&yg$Zjuf;dZ$kNV^Q53Fa!FvP#D4 zUl=C5Eaf)fZ&$q*IqgmRuq_kd5l4ArG>b-TY-qDik;l?PKayU>m0?Jnj5xc>Q^twc ziB8aO-YY&&SkkT|VtmGsgs#xWCTT-@khOaljXXjg4b&E6ROqzJ5IW@>`C@F#OdNQC z7?aEtWfq>Z3d9fXvh-HSV|Xu(!mm6&N2E&MIQ!fSTJM0$iSgR8r-Rmem&xb8{N!V1 zj{Ttbnf2O}F}1%Nqe=$X9gO&cAKil#`38j>#wV^_yqEr8r8Gs>NDWF z@)BM}=V(do!!IFz0Og~%-bT0-#<{}&x~%V2Io>*sA%Jw1@2@5Dgh4-UVE9@gOIw(; zTsb=mwqy0ZEPLUFELFQ>Uk)QVW{Ss;9998^^JnH?=|g{_L#lksEx)(N^V17wQY^^t zsPT+oO|orBMz8sn5j<1evib@)Xyo7q3ZOpb(97jzJ%)}GyLV>pOk4IhQzy9F^EQS@WRv0alZ zFyti<6%dkMzbO!g{+U~4q9Erp{~j1Xnss=Lw492Zk_}lWk?@-jYjElq>M1OS6;lm) z2~1~&U5*aOR(c#g9UKl?cOzkUdfXxofPVo8v3qOG(O&|}V?C&|ViXh~W{jx=ddo$u z_ftocDGI@mOx;zc(p+zl#O@{>UWRf79#eYQc_xs&WBhddu|aO+f(GyMv;K5EcMACp zv$@2HhturBk;7#OTo9-Hsk8&#Eg1V1$B?3<=xbO;!37?5~c~@l+Go#K|$YwG&bF+bq5hs*zic*@4lO7&t?4jW% z1H=!%iK{raLL2!LkAWh387XBX^6A?Zx%5E{Q@_=(7VRdJ+u7xcCpRqU*xIq0#-|`< zsOm3N|FKRLFUy5?svD*t@~3X~WjYZ4tH?kdI&IaFn?2INAoyVp_%zg6ju76559-9Wd*8_z~tHgf$$ zNBss_d~9?d7CRaqaxf z7ab{{KGNgsS8*8v@Q1bxd$t37^yrapOdq$aSFg6KSF-Q-iv;xa!XZP`2sQ9O9cZKp z(0k%B!U8URI1F@e${Wid8oFXpc`E0;G(C;8Z~@YpFj^n_`5;Uw{gS_L#)a!@L4Y|D(R=)8}B?Cid#kT^kpa zshbL-ijpi9PImfCB9So$5LiT=u>2k7+7PHbi-gYr=XiBk)Y&vVcuU@bX`ZQmE`X^) zGQvf1mq6v$0BKx+;)FYg%IC5rAI?P>orCy0E`sBF`!3cM9s_XE36oG!ZV0#_0!2R0 zg_ap(QUFzAp%URoss1V5Ih3}D(TNi6%HV|)hDR`jd@(Iw5 z45%zO=qQ7Pm+Q(PKE&YV#UuH54-VR+ot^eThVH{BPjq2s%`L`oXm|E@+s^K;(jK?_ z4OfIqGvT`+4o6IW&}huK=?vy6q3 z$vRNk$bELkOIOS)_L)V{^Nz?M=9ULnw9+rzfh7+A*w5pq%~<>rL?6uGIxIaH6(-*A z6cjSK%cQFMWUd`5Tm69k&CF<)Pne_~s~lWZgoyM%bnF{F=>fOSYe{BT1n#>weC3Z! z$dI-I#mMxSm6pNkSHGseOMatxB%~j@g`e_(G>^-)OWV@=i7Q#u(ik^J>!8Phl#2}L z!+V6Qe7hnK9k&9sAvX!!kSQc1*juVeC%#pXL9L!2T-j`|eB#sX)z@EYE8E{_?|tFt z#O1{Ta=&nJhZK6i=S1_#L{kK$&k=_B#P%k0UfdBD!wW!ztFa1;Z54g)0trwgX&OY% zrBA}sG_mf7i+E1Mi)rrz;x+Qaonj!RLfYdh;v)o~p1qv0ZR`ZPEWSvA8UQlukbwYI zb#=}8FNX}@uA`7&V$Z`!)4{TgVdqvrVoY%ClfK(_Uri491tU9m`PDa|ur3U^UJFb%)r(7?~yPui46l9?q>8U_@5WmX*Jd7Y~JM9RjoSM=sV zT|LZ3MxzFQqx_z6!oc97-RX+fIsg}-%95}D5=U9pe{+86f}`^dybP#Uz+Ci44qW|p z;kpdPTqs)TLHxuqvo%G>2g}GF#((NbeW{PJ{D_b^&*sXAWme5kmTRQ`cl`Et1s{B{i*4wqUjdpb=cl}zszHz3W1F>AyPL|l zwVpVZQ>+s&+>n1m786^JZOoS2hAdB(WNgUt+*+G?o3RbKn**|yTefrC;_GLMW66vA zh{m%__@qDh+)}q#M zx{Z~qDsQuGuCQ&LbR>%nT$cs6uJ+mVQWd3vvrXZabh;`1331saL3zDf0$bTF%kO0n z(u?x1tA5bYUK`R6h_Me+o|h;v9!Re&m*B={l{K;I-~2qwFWR~RFkkwT_^wG;Y@e6% zRF?Yq`ZC+-DW7=dNdY%d7?f&v@!h0O;;}_Pk-Mq(qm8mOM{;$S@VeTNWi&JOA&})X z#I5U;U1tf5#AK;WZf2x2miw?&W=%IWYI~kifNXu|wpZV!Z_)qa_BLMIQd?eCJ1UO8 zEX0LoQ+ckaoR5F%54Z3A1AnA_^Yfo;Yx}&!`eK1|_}c)SaJf_&)hYaVj{%NO(+5JQ zDXfD#67d8{V1sLW=D<|(0-g38nV`z*H$J+H(#4MWPTYs>48lUz^f*Q`u+<7wlGsNB zykMlb=}fWf1Kmk~)S({(xVh;<>I)Zi%Ii0H)yb*naKTg7+`;KC3|S(TG$5 zAlNAeeT0Ac6Q6R(SE!>L&-A&?M3)G6(7Iz!wlTeE5#N=v8^?y4(zDH$X#FECMAQY9 zHs|#xHP(MI@7TS*b2Nf9^3L=Y*5uZDW_05A4(5M6wna&5%#1K@!6%WKEgIk$$ z%TGx#avKqv{8I5zR*_Olk9I$>bjYb+j413hI)MC^J!OELJ~6tEJcEdgu;Q^M7YYc- z0}U^&^z455#Xp}L`H%pQqM09IJzmz&Vl*8~4_T)tZjn_#vUw&*Jithe@}vXTAwMOk zyiqR%kr0ewh(R2qxO6ZD;*4|^S0{#$uA@S^#EF1Ms%AbV9*{U?un{WM1$Qu#2cUeR z^?}^%yDD9%ycl5QlbqcU@Qf%I_qwnnm-5p`-9~B)#$w*SXI<;*(V_bY<0%I4CfjVO zd@k@yXKeMvg?QcV!`tz@?80!IH~Ky6+GV62+0tomcc+~k9=5Y%@zk~D^yI8<%TRv& z=waJ=!b?8K?LhcMM)8yVqxN9$xZQrd+n#J6w&fFD_?L9SU(tmhV?;WdiN_Ykm1K%> zD}zR|Ud7mxQ3YO+0d+-2))o1$$?&==h*JaRew3o#9#>Q5=uCi`yZnl@?M~18GTS~ju zUb^y9yS90)ZDAB`T^Zb~&`HMYn_KPXri@?RBwPh`({OcN>AF$4x^_i8ueX~raBptj zlnggjx0|Aqe@m`tZ8XdH-IVcLMlpC@b-vn-?oE}$zQw$t#a_=g=wy(s%ShW;m)=FQ zA=+!I4|TexICQ_II&3N)gLrH8if?GHZKysQH-z$Us_v|-7XLMcH)Z_YR-mj?dP!_UpsogQQ)h94uS5Mq_M)eQIGTuGl z>wz0;<8`$ydtig!qgj21`m#kD#x@4^mJE9Op5r9Xns|WWkC7}I>4bI1%#@40`WRzF zAG_3F*mjT`GuFIwL%~fO{gys{O)_7T%piSmO}b${{kn9wr8dp}*XnO7i?!*DAy?Hu zu3!DQ_{`cH-~T79n->YH;%^VgpxLp*)Ywu;=Vrn-(mn)mnaKq$E&xX^$Rku4V_96Z z;!a@_>hjK~Qyww8boKIlQJX8PKy(+zFMvrn6n^W0<`|kpVd`GL$k;DZ1tOkGGEJkC zrQVg5yrcx?n)!o3>HPve(>VHA8R@PGt_FvMP2r=V%jf!XoFtKGUi}w{?;;tVA4cV% zr;B(-zw>l4o)Vip(~M(#(GruK-Kn%5GZM%jzL&wAc*^Itz97$(C6FS=*v0e@1QKRi zJ0PmySom~EAK`p0<5~Rp3&V!A7-!!7L4F2yuVIuQ6OOSYG`etTpbyT3d~|@XN`YL= zWXu?IVO!S4#k(Nr0yUj2o}2v~FmZVlPnZj>Xobm>a1h#l;{lRxnqDV2M$p;iCJk5h znM3G!_}gz3-S5x?FWhwg%AD~#E^%y8Cyn^=E!$LipJExOY*XdW*P>6Si<@ZhE2DS> zA|iSfA8y`6PkxLzzO)N}CN;QnpQ;j)NIYnx7|XM75p@b=!eT+!|yw4MDu<>j-tN+Z9wyCP0Tz5KlFNBYkGu3V7q z+a9YNwr$JSRV*K1`!xLCPEE)bU2MmOEK9m6W8LlQsp=)UF_LD|8M;N~G%M|fK9T!K zhBvzyWGgd;502YeUkkuXquyFeZJ2$e*^Pt7s+ZSfP)T>IN?R4LnGnBk@w!^}ZOBSxTX{pbb@qgfp@;%bER0N>h`oOm@UzuV-Gm zBhIt0>Nl&Boh{0iPY&%eV|*+cc0ynW1k0&*f|u)~8-peKI{WVWr9H{2d_M2s!zaH< zlOfEh#NN7$A>3$qJh5&BlG&fE*7?4ye#aL9WsgdRL*-{5X|*D&yM!n>EW*W^RDWe-AML# z9_ub&^;X|GQC~YzzijL*A|BxJq4KJ~oha{#^04c~34B!^JAG`kE!zn%t8TYL^;dR1 zIoN;Dc6aXT#`CetQCTX}s4|w6_r@Eqw^iL#J$UyU(a{S8joH;QFAm=N)gQ~bVz{h$ z^#HmM$(@9g1wX;0en3#>bWga*uMU@i8$W6Mia!xSX18!j!_C0a(-9OJdFi;tlQMj9 zz<@8odz${Fnvd(LUZ0x+IKZ;=$1fzwj~E4TwyQX1a%_5ZT6lg~;D)2K>yYw*(i`CV zRk%CcWRTYbm!S9{Gvif-yUKKuj=VkaOv0B#*I9uEF9sURFiDv5Viu8BPar<)zSATd ze1j$!WJPu^&n_Ue&^eJqATg7O3xK?|VtDle1~+-D-HE5&M^GSt)5NeLB1DAaa;#j+Qsyd(}-iSYvF^jQnBTe~E z#hd)c@DMi)#0L-Vdrww>ig<=;B{_7V-{0MFO}KHvfId9tLk9VP0b5{kb8$l#?wc>Y ztiH3_o@_sC+cF?%FP<7VYnp@gwXt zvXX>!=!we~e5wzIiBj2hf>}@GuyHEPk~I1sWyBbxLWFqiJzL#u*Ee2i|G$6qciU%v z;pWAgZ`Kqb9m9^k+q;5%Lh9Bd7(U>SuF1VZc}im$6W11&7f=QU8Yh`=ri5PRq;jYO zEO?QTbJQoI3iWu8jOf{uvrwgM0Qs`cSh6dQI@4GDqL|u%8$x(Mm!~k!<{L1|Va%aT z;I=Z`(*4wO`&)ne|I_~FU;jJp(d|ddCj%bCS4QyG)@J+k_k6tl#lQ5I+Q(jdRb}n! zCP~Jw()#;#^Ky0Iitq6uNiK_Y&JXB~PT)wG^MSEYhO(DL@Y!(mc6}OfZU&5DXG}c$fmi(D=ZuaIgbyb{58zCWPlsorIS;)WzThwr#7N=9NXpQzg491H zIaV}XOq~RZCx0iecp@t9X|ULJIk}O?x=3N+)9F08c1X02{Nkq?L*)o;nvW zAUdbxLQFV4;$k2LUSV6z`H#X`UtJmH8muhb1nHxs#~>Ei-&1fC$)VTXt>OM+C|6)3 z+Kp-L`C<~Nx^$i$hB}gONT6*;f6mb{Q~8K0J&vg&M=?)+I=HKXcAL#)76sqm28P4@v z0mh1F%qMAi+&gQh(lhlwRDPC9V2HhY|8D#Izxd^LypK_}qWoOEMPNRyU48Xh`_q5w zPkR>DpWEdHos%VYGBt1OPxYB$+$$gX+3G2?W7Lfq!Jrd!X0eW# z34_OM!{-ApTybglUa*Wck{zRBS;$LuU;oC}+O6B~`ZPKgpBvY%w@-fZJFVZZe)UV< zy+W2l+g#sJjAEsS@jOSTj`gwPksF;+Bjk(0jFFNj63q57Y2^~!6nx4rG{ z_U>D6wPUeBzup^L{4lP~Ryx>_LGoQ9Ioj=z$udcV9y?lj{e!szziI-n#SFT=f+fTOJ{d@OShKwZ}#HkOn_I!TVV-)pJ zJWFC&8bjaYwu8PvI=ztg_g>BJtDMsqVbuJrZL*x=#tpa61C{mU@gvm-W0!uLX8_*E zR+V5$BR6$U7jKq~_~xPi_|t!o=TIJGAyUP;OoFdpf2nO$ zd8|`(C71D$E%=zpX6XetZmLfEi2xS{e}PiUu-vDd%5i$JZ^8@zT37RKlE~kKy*Z%A}#$vNQ0px@#EY?Ru+< zrcOdW6vT^kFR5NHzw(MX@iCTnZoS*~_Bn?XLmBEHCa!;`+oOQZf=w%YggoqO%< z-n;GJ`(ONJYVyM1+hh#?q_l0svM zP94rZT?q?o@OGyc=Hv|!N0&S zTJq8d4d{))(3GO_{}N^cJSClSIzLLALe(GrSsXzm3Vm939;|02LqV2epK$V3Ic{qu zcGk}X&@UQ9)2@xt;r`ME4!ymuhhpqce6U9h4*4sZ5bFvZJBh3` z8q3WQeTHD63x#8ZHO_PCkD#P;444prC%0Vy(VM$y>|5O z=WPtnW>?ySvuo{#e)@Cm=YRf}+o!(kcgaZFYmXm2X|KQbTKmK&-e?cyfA!T@+sEGc zxC~{!_LR5Vd8GP{_r59P?B8g=?{|H#H{d^$G06kprN)J(PSz<%rY`mK|MvvW&!(PIqXz9*we#>LvQ=;&iI!mht`Q}k!;j*MhvkA5*I%BVnR7;xOkV{l-= ziAM3vu1N^{p_|Ew_5=#&3OZ05!o%mH{zkR1|A8gxz%l_0cYygji zqW{><*V^y-#P_y;`lJ7_{oFtOskU@VTgc#{d}vo>*ln%0fBV1x@3d<#U-iDs%p%ec zX|t`X*V?+u+*91WJ9lM>O~+r?O!`V3v+)>9j5&;L{8$=;92slSJ-TIfmww}>P|xIr zE!LA>Y1dwQNjD~|?csxm?cUw{ZVQi-FhMQS3*FC#TjvLFmb=sq3{s&7--OZ6pg~R& zNTdFW|F{gsO^k0Dr0?CjEo1+l+nM$C^e^;od%*oY_5w4dHXc>nzx!kDfBS#GKxJ`c=lB_a=jpjp}#<+{23#gD$es{!CUXF7NG-)sM!Kl{IQSuYB{EynPhzw#5^H-#?7 zL>&4V;^l5QitcS=_@I)b`yA%M9SH2@>eu;>}>t z`6M>SGkOl`PFn!U;#jBU;v@D3VmJ{;h4lQC8{Z}Hl++Ci$3m!}|oE&lK%5R4xm7p=~rGt32@b|;oN`4l&FKbNF@y$_nVV!5bP zTN=y**C9d5D&PWAK98_C3ZX4rUyK+KiR&3TCI)}v&H>7K8kAfepdQ;~0Thj#PVIxZ zjbSlhnZ(*m#_;ZF46m(SZTC;F4#x0zeb;xleHj2=Ch^AWGLFBqJ$}5S!Tm;i?e*7f z3}@!5*%?`d{)plmL7NgvZnw$gn~3ix8FxT)+V+mg;dqFpx|WT47mR*0S< zvKYfM{+G|%Te`XU`Okc|oyZuLNvjL@aSR&ioYC<5$6spy-T(HVmu_g~1M!uC*=MyZ zt9bKq33Qh7_~P5X%3#*Yf+!9nRAYb*sf+~F38M^o;bHmR$7f0-U1?Sc0sR$uBqRGs z#)|}ND~$Q9J5}1jp$uJ&bQu|EGPvQnxp`HY$xrzdk+$`^TQRngrTRrbYHx)+Q!BJS zov927d6ga+`pmg_;HJkbuZ&3>ekm-Gl82oyZf?BX{{BDw!S=KNs~jFZqX8LVdED8#0R6QmL+&UTZC3ydFg=4G|~o^mTcnBOfGtM zof(s?R2fX@7f!HNU39}qdBBPCdD#l|q>CSGX)TB1ET8hb6h-0k-YZWfdd(f_F;HH2 z57$+gB)JeQd*%~a&=qNJ2hycaiu0t6`xWgcBT!=sv#OMZyff+J+WPgjzO>f%uo3(%Z)7hK%7i+W+qV^q1RfZ@k)0j%6&XG}dvSiUwK${kVS;7P9<= zcMT9ElSk=8(q!sG02zCJ0<@<;6E7KArXVrZJ~C<54rqgGfNzB8J91gGoFztG2IT2w z`NoE#_$godrO6Oh>D_Csm%oAk4L`Xl@IX-Olr@WlkYTS0LZp!krgXM$vM?U_C zI>Q$Ozio_RI^a~0w0YuK>*N^+8r-cMpiw^v(0KYA>Ou`z6LcgziY{X~Ct18CcrcIG z{%19YiAyHshL|{_041_~f*QMHErMhq)`pGBg3>22>=Zw53TG8Z46T;dVC(34z|Et*S3&4;oO$+k2GZr zJ6|WC`l}8c82}ROaYa*McnP217$zc@u97)AoC3NgQvJYb=#aQ@2oL%k5$P)~M+Bbr z9T=wWd1RWlrN47Q;&d|EKnqJVk*~@PR9jLOA@d^6g_1?|eSogAdk%x^PkgVhVKKZ2 zQvoJ$p)qX!lw+wRe)KymlbEqx)YW#hcdH$~^J_YXyD@xrt^I?a_)PoU=YF+)`cuE# zGs6!bJZf*q`2DU=$rygL-HqWlJZO6zs_QA^?e^R4&;02>-M;^OzOS7u>w8MXe%2Ejh#Nw1=h*_dA9qK_dVx}~Fb zB;#^tPkn?hg^3+|+HT4CS(26P9XKQd8nS_bUMLf}tWX{C`$S>yT?|daJ`){BSujRk zD&{lrkkPy4`w{b&ha)eYg_|Kb0<{WQxYWQ?rI@Usj;9>#y- z6R)@b$^ZO+);6}*+otqjqf{-14x0Eaupq<+CnusU^Guaw-)BD7z;>GRlTH8t|MW>jK~!eE1h=XT z6h)jls|@#1_@O2NBYtjlC?MJ^103GzuVC_{mwoXVb<%NVAApda@WpAZFTgm#UdDV@H?7aLhN!v}6Sf!5hEh21gK zFV{uG*l>89r4q}hhwZM)_`NTG!W>={q*Y%ybT!e6fIBL#j=k|o%AJZ66oHEIagu_) zQ%}%98H@~WcogfwN8UP71_MI!!7<{c^K)(B1op*N>4c#sOXQ=o zgdf82`No`%DGPK685vQRBO8xlOfp5$azev9<%Rj9p< z?!MsntmuUSSCl%+RoER#E@{Zbi9#wD9bb4z?+oHkC>O8zp$Ah((GiKO2>g_1B)-EC zC*)*x8!FwpE#;F}kzVtspzBDA+g)XefQzfyn!Km`7w$mmB+zAYQqn69j`UB;k!K#} z+DW{X?t#r0C+?g-m8w<+7t^T5Rb2XHntf3Aj5p6;pxG0|35W;A(z7>WK zP~f zPY~kpD_u;Dovz}8iRbbJ`uyzl15XSy3~Zer$WZ;kkGNHL*boL0v-S#iWkd^Db@Ve) zS-C>kEm7}Y_w*m=aq*d%j_f8O){0(Z9Mq55B9>iy9ieH(SvME-Wv~0Ss4|p}MSmtk z7L{;*cR9y)K|j3crJEDA1?Py{L-fk+o1lEmfR9qJZ>$V0)eV=uVEIHcTjgQ!S$gGz zc1gmCLO$9DLDlA>TUpknU6oyK8*QfYuChH9TTw{|D>8CUj~J&`Rc_jzt*uyB>#e5b z_lGys9_$^<{;m0{9F^^5=U_LFtTC#YHRdHHjKtK9zG~x1bX?1ru<=ZsP+E8SEI&9p zJ+iUCELq1}EY4+&`W2lcGGv5o6UU2PXQD&-%V!wCGN6P^C}UJ&@MNq;MsZtNS6RqI z8@Ms0Px7$`@~myBP1Z%Xt~8zv%;Zw%w!|9k4e5W=GLXjhUF_d`WtAJ7L-m0p_bKX) z4k*uG;=@SxT3ppxvT&}E#tW&dE8;1cR(V+r`7~B?lSNC?mMUGkURRkiM#Xnadf8mo zxh~`UP<@1%b5V$ws8tqQovo|AS5*&oiCJISR6J{>Iqx^sSMj3(Zq(Esj3?3!
wb z*IACRB6-<)VgOe_sK0GRH{|PTPmM8qipN+*Z+v!nO>(WN4l~QoKFxj4VU{-qBeSR{+G9iuG>nk(0wb~UutX&Ncg?Yhl zN%cWDGuj4Cs9b9hy${G1ZlcqVn)5`2w!13B|C-w7I=g15T@Kan_SJ^$RB}SQtK5y5 zZlu&;N2=d7)#XJ4UFEk8FeF_F9z+G|Nceha4u-g*{g6;x7*OZHCwk<#sH`#SlDa6G zOU8wAUJ!LDbU79wF_)BmQGD@P!~x=GfStm(s^i#TxuD*N;CRv*)GP8SP&YNXv;jqr zdp@sc%gnQ1T=*?kR4|BV($HTHN$WktvP}E+x!ek*TF}|7?0|0!6+~B^cURn$YKYMw zQ!uD43XRuCU>YYEeRm||NLMf)L!R9D;PyxXMV5TPAVcM!^5)_or*IMGOMD7cdEKpE z@Kg|K$g7mN7@~e$EPxNO^Lfky!A+i}s z^RK&jfJ!I76A2_9NWF7&VM9k@3{LkW`RPf{!a@bS%qjZhh$N}zt(ey`U{aT#YRf%-)c@uwV< zg*v>u?C3r^Nqp#go5*1F8g|u>?bdin4lWSVH=%XA5$6DSGx3#9`%$;Wq=@0=ma7%9pO2dM}L*CfPr=ik-NIB*4bV!FAaS@0)AAK2rB)JFy<$)2M+8vvy zMLuH~Qx1g0Ck}3pdq0?zsC>O%cEhL6c?YHdP0nyZ%65JRC48Pf@LBxDD{(6RK>Lj#R4l!2Nd?g~5)m9~f-;D%o2 zxn2y;+Wmn^NOIMSN9ckST))sHNZ+A60y#u)(5YQ46RE-p+owVGLieri9|2y9HQy0k zhiAl`3-r+-;klA}5e8`Oq}@iSb}VTD(h28$%6U`y#EwV^EW%2*@0p(DQQhH{zF^CNYPQyEkK_>7xa{LqJJ z;4|u!oI%%tbn02IXqY+VBQ=@LrhXtZoNLn2hL=cS1|P_nI-pz?=lUVIAZ;0)h{K>W zkQ$V~`VIOj{RT-}vo6e6$xbnmf?G#hZG1c zPayJP8or9SsXDAk($lst!*owb{~7w6T#Qt=i^{h6Le~Yp@b9{??s_|>Tp{h?zNs=L z179BG>v%J@+_B1fazT)7*sQ+uMBkZ#7{5G zG3G>&7+<((6+vE#$gIYc>ixi-d;y4BjP^jXK%Dt4;!=-unF@%XR&I_@kNYrS7NZ_N z2u~!c`@Lx!*zr-&JwtNGDO8-Rd;u0*l-K16hZfeM@N$@N>iGw`<@)ii(mO?DF!Nwi zIOZhBe-X^fAZn8(uR+d`)3iNpt}JP|S7>Sk7jOEr358{B(JJu=lisOduQ<=rfY1=H zLy)u{07y4&!b4u-L2}zal{0jO7lETWh_A1s74F!ZBcO2cnTr=VTfBy8xl_ClE{Pt1 zeERp{B4GjTjt<-dx;~AH^R`Vw!sGfq5y8ay2WH&`rv9gZ$UGQN_|(x+0@M=&toj>X z%CErXz&Ql}Lc;uA<{Y@@@rf@N;Kb#ET4_KPJ;jn0m`~H8=e2j@I$jyc8-1t$($~H8 z0-bkljK!xPVF;9*mX7osCbT2;lzW2Zq&&>3@G!F9H-LBc~F#Ybj~ zjRTBbKJE~sC?^ACHc~`O*CCCk11!}d&S*KP5xl$H6luR(Z=j(ZnICd7z%iy(wrEbQ zKN*YkT^TpkXFb*!9W^oOTsI+G=kb^xV-dsbxUZ3BMw2B@JfQ+yo;m3#o!IB{G3KU*8u7GE z(i79^2veS7`II!q8RPY#*POFVLv1M;ylz&y#Oy9ZIDIGlERPT)l|GW0bcROSq~r}! z7Rwy`wKyBKqEq`F938Yhp*KxaCvZP?dEM<{WM+9$Zam~F&h^4?H@e7Ax>Y`weO#Au zdVR(+4~*94I_}5_-=WVyN&khXWQR}H%{ODx3p$KTJZh;DgvjRhbq?KBth7yxa4(hN zhGWUI-`j_e+T){#ZRcoDH#n?kCjX0toXIa57V9Zx+Ubhsl%t=>8|a?&ogUdcQie=o zHK0uR_Ve%qLuJ2=&0-Y{UA{r?ivSwZFCfYg_kJJ-)P0I`Qkr?Zf@w-Y^xHe`C0M}v zLGvql2RDZv>fu0nCsA&jix74fsKSu+B1+t1eXN2D8auk|nGEy(A%{Fk8QaY{%P%)G zDLtY?rH_z^bmgKaGe{gIA}s#MH*IgSRm|uL$kOu+ne7IXR$<5J5I=5o6C9%B1SH=y zY<(9f!595$zGCte7&v;6#zm?LUqR|0o<^1H3yb^%rOWQTo)w&>=b10^a;PGPY-EhV zJFP!p!-9*w{MI$2kI1uzRykd!;-`9i4w#;MfQw$x!N7W22L7kvWw4Nb8ENVoct4iF zTzbzi%{Qcvp6<>ArMd}*ek>?b1#`^^n}HQ@F*`?dSx!Hn5K)MEP1EMi#Pa|xgA^QP z7(mCUiqYXbjIkI->Ko=s6#2SQY@PQ0ST-lE2F~fc-;h&R=?o9LXC42ZKLG5y1K3b@iCytdWI}@>?VFM#{=Bw`B2H zp`QoPgws->exfTQ3xj_{1~@kw$VNF{PEa=+lqXsny3W({+7Kpw;2N2_T+tnEr8+57 zXz;TZoAOv!%$jv(G^t;7lQqEb=;eXrwwo~Pm>pJsVP+U3Fv~QMI6vR=QSK76w#>N8 zSTsRqP+8y2jRsm(tn-=&eGuM6xI}^ze-AU;jNci13UX3s)*tgS+nVZ1y^qv}yT^O& zNM&aWJX$?v#DyvxSfA~B(}r$OY_Z-;bbvA_gCB!iK9!m$BFs>$T^L6Xj`vmOfoI?M z*jeQT!{XTc0zoI06nqh+A*IruoX~lo1J&s{g*4I#q#DJy{7-{QFSm~)fjOD!3Jc)$ z`B>LSf8?#CF(+=}v%pw~!nwTizAvX_dMeD-`9g+^k*ZN|H(;8Y_@|>!GQ5JZUIET^ zPBsWS$pG_hUkny1s33jKv5W2Ja*HoVA@m+smB-|YE8zxM3>Qh&Bjv=)u@CY}%rMcR zybK;B_n^x)DrQdRMHthlAt4W{7iKZTcmWFRgp2fx=jL3f0&=ZabrvJ>I#7pdmuY-z zN13DmJuDdLo&sib0Zu_&y6^d5v1cJqX4tvdi6?BOFZ$uomnPtgLH7f0?TLOb0+KI+ zrw+=M2U+A40@Ls!R(sH`&q2x9>o|h^1eHFVAw9Xy$Hhv_MOENZ9Zb%<%4Axhqq?p! z==eEFhtAG*a$G_{w+jZW_yR&s^DpdTGU~O>DjvSGv%ao`Aa$^*&!vG`8 zIm|3Zj}12>MwT6ViIcd*gb;tH5Ff>9+~5bePLkbAG-RBnB#Mb^un-w*YL?GMMJYY( zm;pRXkG!TK4LL#sEquI0L-nR48*HMtA&z39mgDlAFPIrr5I@`k{$zIVwL^kjW+yTH zX3E2hUKpqtgZ?D9ey%5ScIn2fm}2@_2|0(!7lS$IIuIu$%x9N9iz{7{*JZ9|*$Xdg zVk~;Ognxm|e8=odojNKFm$W>w10|&nq~Yf#!Yg$tU-iTgW=R1bqu|AGQkfy<3y6G} zW}bh_pl);V!dyRk*n;X{fxz@9Z^9>?&z9N6f?Y#aRf99t@z86C{i$^H3bEIxGi!|t z3-PTV`f}I>sUA5W6=5h^&JB35jD;mHo_SYY(Zj*Xe%m|dGtRro^a6n%_iX|cmrlcp z7DWwuQ@DEo(Rr}4e?)gB2q#D!baVLyba0|Tj|Uyv728ipU8TE5?0TlaOfCj(>kP+;B7uh>)zF2Y(Wmck) z^K_25>t}E$P<=GzRf{DK!@_N%e8hW;Ij$3?>xLEyoUr`G%lu@N&1-}B?v+k$tT>Oj zD*<_3g53b*PLt21@)zcDF+3xu*tA`@|N^)f?rBY&5)mj_0# zk%cn5tTJranF7OxGLLl59LQ+P4ocKjXnl_cW7hP>;I#Dc14T>S;bZxFdG1$&tg+?= zD~u@C59ftCuUGbLEoCJiX6#J-z0|@8A5u67%`<$hKmS69UdAOAYD1N?p)#*{ORyCg zuO~9vj@nUbm`d75UEmYWp1GB~zCLq10cN35|(h?8r{ z2^-8gUI8*#Z}>Z%vvRaX*u}P5-qsGO+&2`K z<2veM9$lA?sJjrO3~)b@OS~M`Eb!Z4qOkFy%r3*&^NI&_NCd6NEr(W!FpR3E5nxf>m$N6sqhjUBC#ZOvZ;iUgx~vmNR1!yhelA zH*F2v`wlPL_&}=zO zA*H2A9rWk>0K}vB7_`i|^Mc!I+fun0w9K0B95R!6DEjCY`P^qjZ=(~wVrZkk-_$UU zxUGyz1yv^7^&uZ~`jib1bd_0f?;7oO*?Y&Vt7X%>K1M(-m-12(oOr5Fz%S6>; z%Z}fDDZ5PMCAUqLne8O^k66NxohDe4LO=E?GvV!{S<07A)l=gq4G{qx+TFFJ|6ml) z*436-8bV!PG<-YCj3z!}V8BU1Da=9q@mvTb4WbRLHZ-LH8naO68CfMZF&s0nj=uDp zLQ|Rl$bn!D*5xM7-Vp|2dI?avt;2uH`0 zN9E#oS)lsNSZct$DX23Uc$1f(((I=W&<#+OM4QqxeYLLE=eCm3+i)%xAcE5Dm}l}^ zUVe&7xnnwzFd?)d!%apMM!0tZP#FgrC;Z|bWs730X@ydbnZQM^>)unRE;J+|x#Q%A z7GsfFciM$=_?p@gpFvl-#Wygz=+0$vgQDodDR_&Na$@+8vsCu)+mn;u58FiWVaD^*=W4P=j99eA|MS* z8@O#?D0H7QD@qE#5kxN`t4{8*l@2eVjc}DmJ(;g%59ykfK9Z~%+d@lD+~PZc%MVx{ zdnW0Wub_3J@~9`DGG~_y@I}M?sPKXy>icMb!Ds-A0ejRWC7rN~kq)^47eMAomXm>i?5LlwrTxkx|eoU7->P%|zi49zv>5}%6-a@;g6 z($_#Y4~n}m-o>3$PX^G=S58EmC(4ah!1Ue4V|Pkl1ymmMpYMa_G2B%_r(UDrSii)c zgOO!->rz%S<>2HZ`T#)N^8*IH4$n(9Qe9S@bNdHfU6{-8qP`dxB=W%$f3!qtP&wBH zOfN3y8K9T#?AF^<`VRmI0#4e@@ImE!7FZ~&f_6Cw2frc0aIF4nn#oDWsgh{&hR1nG zT#}$30lsBeI_w)e+D?~u^xRn%fBVTJ8q6AoZqF7f!LzxFuz}psG6{U_MH7{AgmPV) zewY)K8#2z>`TDZ|K1GvJ{@s9Ox-u|Ie5y0BGJ$}IKMwp#5SXaveg=PaWAFs*0d&-Y2n0f`541X5@Uv#Z&pvqJs=!K7UFq?^~d}?F( z$S=9o%dQ5CN}l9%NQ@$TPwpTIP2)DVOUcmX$Hy(Uy!(QBgqScJ&5jmj^aG)!;}(#@ z46(E%Bs9pTK-!krNWu{<*)SC!KIzO&MqXCq3xY@OkYySwH@p>zb1A4+qsj%MFGUC_ ztIEd6r7c-*!*Ud6Yq{jmo@{3~tRWUePh;P71pTY}Sjl8U(t3CR=Q0nTF8CuK9bYs| zTkC}adMTKAKQwU7?x%|Y&O$drN=eFC;?P(f2v(F`pS+wNut~n0P@QiN+qJ)+U?kU> zQ<F4Re?&l`!jxFQr$eeB zDrB9|S349Cbj)oY=^FPujHqmESmV!6ckbo5%JxMSNCR(+}<+M?o~2JlXP zpyy0r0kY&E2;(Vl=j#X0Hgy<{U*x8&9uN?o))~IO(BX%+P~#;chJ7!0%1A(H;IxqT z%wSnm9)5UC_$0`ohkfiNDnKscEMI(10=Q_wAWIe?R=s)DsWglzqLL-$8=!HhAA9uk z!kmt6NX7f1pfjPBpW_U8E_h$w=PD9S8Kle7zU$84jxLpzuZMk9+anDR~BBI3lY z^rFw(`Z0*}LYqp~SXf2%+#M9kKj}J{pWO8GK5}uG0GBcA(Lfz#4C}n%Mj=Zic$lB_ z2PzxJ^5f2d%jo!wfP5H=-kVv56gM2%N{&bQl#9Wsay#FlkKlKl4!9F=+jM{RlY)d* zHyNd#IbGw6aLo9g+5o4nY`F$YNb8zTF62u2$Vj>24V@7`^p5S|G~%e|l3%Ff1-{uT z+qCidZMK!;gB8|rvbc;15{d4sgLP*L6*ahBKGwUVS6<@dYns#(-5wqt*eIsH=*zP6 zQ+dF&>;Ze6`wH?wzQ&$rs!n^NdZjK zVd5PCYQsgcPNH0%BJ$D<{FUfIS!g;D3};}Xoa*G{K$fHil8roL$HG6A*^vY4P`Chd zF$>Aeau8>ug^`WFWe0lVfFBi%{OhCviL+~OT&izxkDfx}E63{4S=!SB0#(ad3QbJME5^VkTU=^OZ`P3Sa5Cv6M|LSOv`Sgac`&o4B@4SXw) z{+R(-gbYahrjT=Jgbb+3&#^cam3d6ja)v5-iFYvIfZ`I0o2U!1xGmxUL)q~Isd}hf zQZJLJ30a4prXPHck9ASS=#}h*;ezY+!SGev=G+gz^cOP{bjD-Ea{)3Kdbw4H4-Cl0 z&oYYFfIaLc z|D30Q+mJRj#&A=N4LUB)8N0^_Eh@rvyeRjzsyt5B(fbD zTf9*UO9)(MCrr8Io1kKl8^jMOvA*3^Ab9{%HGXr_<98of@&_`m^TL^Qj1kPdJef2=T0Q_Nu3eYB#EYzmA`NjfZln0zdEoV!0{B-ccek-8#;Mxpz15}?pkF-9j==my;E5Cu!B21!T zZB`4UrSk!LxMBu>#V5eQ;Xt{7G{|5tpK}(C2h+e%%>hDPP-Qv>)%2N2>&Gs-N}Q=o z%%D02KKq}M+krWYM|tHmr8cA|7_eOB8! zih}A8XGpjhN|QrzqR+W0FMY-R2mal!{!tv$4oTw|1|t+tzmQZ2D6))X(rqmF+NUnX zvzImGO@7jbL?LIliznH%()A%?#jAL7m z1IQ=8W$RGO139JGmW~^RwHZ4FEc=BoFf-w4!x+M3foae(JCbm~=?Ysq<%36VP{w?r z;t@MLFUTPFha8sM2}=l$>T7YSUm7)XJLgbV=S8@C0VrK35FSovQbYBSVYza;(yqxU zeQ9k&I$riVUVqM5WpfiiPIN;y>hAKWhsGszJzHb%;TXTCy7A69LTMBOt-Y4#6dk+w zthQNOYa1&Y<|pm8J@)54+Vk=Sw@c4A>UVUU1_^W>QZ3}fL$dfv*IxR8eIVRYJ#VaU zv`zL)mTbD!P<=H%^2uoTDRkQz!bH60+OwC}i@=v-Pp5_I91NpTz83}4ap{GDtFMYL zFQws$$%&r3Bz(}|2P@4glJXoHl-Si)N-vP|qj;s#5bl&6N5;tyt$y5%m={o;MXqV^ zhoN9xB%gSN)tQ!My!ms8L@C9^$Y;@XuH(kBnn8i2SjVoOd<(EOFg*|!K{;nw|@vWJPI;YM;U(6&DnHSFjM-5sA`_MV=1!iDiNQ4&^*asb%ELLlg;8Y^Ko;`~ANq8U7QiD(#8>-pF##Rta4xn!9|WCD058#UkwE4irc1Bz zjt_5dv%T6Gdt8!N=NYpl`&$~ zaglNm%BP0gYluTn(G}nsD5WE_4$>thq<-$Jmw{xUZssqv;n4jQWgh%o#}UXw32r9{ zi6d`6XCiCo^F9Fos?AuV0QoDvWHG1=T8BnqymM11jnT=wduTjDNcMO2d+g`|8np}i z!9v3ft^2Vq&eV_mFs?oWSbr>YIb++bnfPW;T#thMm5(k|<1OvLjBb5U!{?!VG>dh{ z4h!(FGQ_9K*v+o9>GN~i7 z4WKZkXfzbU=TtJQyp^+dAfCv+CG!bA=@w7<`D8ME^3*)fj!$ebQ)l$aCJK=py(r4_ zla#?@YxE!*>BTY{NL5%QeC(G#eWC{8T5@e|&DX<2DV3MZ&X(B2g>y%e$dI&dI=ZIx zOE)aMQ@SE_OxWXYmkX%|D)A-%+S06TEpK?wS{rMsFY9?#Ng>N7k~`zG_|lLD>XSNn z?4oY?2aHB)%%g=Zx32d`%6BYjt}U;_r5&oA-Lr$X$8Hah3(3LW$!FHPG%CXx$QVeQ zNw417hCbZe{p6^}p)bZLn4!O8w=wv>Ayh=HBu5MKV#ufcxmp;Yut8yqp_%FnJA*Mv7!KIf$k@zcGgTLAYhUk^C{D1Z2+ zEI*Z@yt>Zo=P_IH^|imu3X_h^FBZ}&-zKOgnF6}W$q~PG)m^nD2d8xshXyWA@6i3^ zH359$d}uId9n7m3b~l|ie#1nh4+>#H*O`ZDbwd@L*Q@+=Jaz86j`}WFZVZ5nA)Rzg zznt-w~ z2#(Xij$;Ibjqn*WPW^$eoHFWh05lmxdt6WBSY%_)F<;m*4ktDFdtURKP*2LMar^R$5ub6t zZlpl$9*lER^VpeCm)-o)doYGT_(zZUl^42h+~~^TbRhYs0DbD#eLi}p2gPw?=Z|r8 z19YgLnI54ys0oCnCddg2<$AGhF)yH4tD!+EW(Vi2pIN5{wBRpU^la)XspMAwgFnW=Zx z{UQi$7v=>6n*D77*j`|D^YT|1uyRMu3Zm{2amMhpgL|kMgC7otG6G&AVK?QB{=h^G zLvk)eJ_S0*3v!Gy#NWdv9K$}b@XH+rVmjoaU=u)h(`0g4il=qte%Cn<(-tg<48Sn! zY3yZm9N;N(0&w$%ZD1Q-Wh6V@KEdF*7%~v`z&WTBJn%lOE`>$@7nPSPKfT@n*FL+Q z)Pt@)4jSt5BrR9b?o9+Hbnpl&RVN_+B1;0Ez+C^pZwH$Ax!~x~aUK@yYd$@1Y1Khy ze~0YqCnZydWSK-<2xA_*Q}Q7^&lmY-GJ6v`A0QPi$cQ5Zi&j;`_ zgx~wItqb0_4667S1Q$teeHYD&&SFGkh4s%-U1lAx5Iw z%r}eR2LN)7WzJ*PnX{wC%;i!hUs2<$ZIsIi%xrLU&3Jg;{E6}ZblEb)Nr@?swcvbU zVqL}|YiilI_e4hN;Th|e>7z2i#WG_^!hY%^Oy5ea;Ah=;M32zMElkN!XDc$+XR|ji zOJR=N{z+bL>zl3_xB)Ob{6m=l{k(fktvo~BC4*=aE?!88!Oz<4d^L_+0q;Y6eAxC5 z_uGN|d7?#*=qH6l=Zi3v#T}5Vlsn>i+K(F0?u!T2zm$*@nkb*b_H&&2LFnvHw0WT9 zd1I<8z9SsZLFNCJ0Q~77$fYB~;o^1qi7E(n@cn?nWb7yb=bc8$UoeP2G^y8e;VNkb9Hrr8GN6r^Da~LHKOASrhdVqUJATjTCZDZaFcy5i<8F4 z;#bCV2yjLh8V5P}-{XifQ6=T)zax5T>C@q!*GOWgxe%VByn%I;95?(I!D&vCei@mG ztDl0&%Rrqp%cVGnOuzs~`Uqp*9w8`tTzCe#0L|xx{}h2KA44T9K0D@CyOF*Q>n+sbQ{ zpWP&Etf@?1E@M{!AUhvG({t%CmXQ1ez4L^l>EDzss!lfQN#Mwi%79+8wl^Qf0LkyI zxjbWv!E~fLk7EMmyKkkyOQGsFEe~{F{V*`uqn5AJu??Bd;I=CRasT*02Jx}7kU5f9 zQDLV*$N9c^9&!#~>|+XM7Mktdwr1-pcdc!(r?9wEChK@L80f}&&Ep7pxxtPhTt%k5 zEXNUAAb@d4Vfste)%DJGRks=bVjLgG*j(c&){5$QBpo06BN-UQzVSfbp0jj>KguMz zQUe*QyAVCe_K14`iy3IonRUFezR^}@GcV;hP<>guY-8MgnY`$edZ=$!=Uo7oZyD03 za<0!`FJ1&(BHoLG8cwpF2SX?@^$#|RwOg@M-xMG&9Gx*dL_9ZCd6wCHgpMqpaY>J) za_I!1Q=Sis>t&gTbV151QXNU`1nFYkRh(%gQ=Nnt0%=2fxV}G(bgsMF3-oo%C*HI0 zfhPH;fUHw#r~Y$X7pg^YA!>ldc+W6{J;CJ;dMHSGFlj|m@ww1u_~=8~fb)=;%&L#j z**+Cek&OoV%N_JQ(H}n07hT6FtUzIC(@r(}TG;=<9~>d*L<2xX-jq@4=9NRj4(eERVIJ%>S%b`4!!yr%AP1Y~ zd*-zZ<)c}szLcvVe9K7tuwb2Zf1oSlY2Z2ps^5ac`1za=J&k?lgYtPl=qgnWdtZ=0 zn0|&29p~ju=~_CjF9ZUESPEKZ!v(@B3V!s{jL4zoNDJl{9 zD!0QlG&gqGsNoA=aHFsAA|{XUF%Tm`xg+2gb_W=S3P&SH*Gtc99SSjAqFlD+%FHEY zv)0o~2YUHAXTsFFNtPgm;8mf%DFmu~f%R$a(Mv$lxojTet6UjOd@RBrZtzxM`AQs5 z-2FmdFSlTE99#s;sxnev*Ve!Mxw(iOtf^(k1!jozAr2Xdl+XEUVj1|t-h~F3RQYl! zqCs}t%*11yvf~2=IqkyuVLIih@gs8MRiE-nReXB7{XK3dV;<13lXOAPtn`|U@l7wA zP{(SAGE8@5nC`JmMCZy03$>MjwxuV#?669F93SGy^;o8m#VYD9kJ9=02s>)OV5BMKmf|^_@d$4ZVUr%el#WYEN=%lxG)2gBAnLKqqmDO z$-M!Em2JZNg8^I@Vak)TIttRQ=AlG*CdEVBSzJ_d2+D;b=z&Y8JjlHn_C!q%&+~+1 z(HKJLghV-U@Z)k}#p9?BN$wZ2j)^^w7p0Y4PAV8Kj6DdNC>H?74=#reAuNyorSEa( zF%h)4*B#1rA0&$oT)QJv*DZs8&kH=Y{oW5O+p~k~{9j49SV>XwXfSc4PuTz?uIei# z53_ka3LvVYs~syj=Wl?X;Tt(vR-AdOeKp*8DKC``dC~?9vEj}YFZ`*m z_vIC?tPszVi7c1Mx@E47@KC;C##K1@F9)|3F65mDauAOk%))cy;U@slxs4@3syP_I zHO819eE?paCeQB-VEOF_uN%=m_Mydh)U+v%xpho@{W$MO_*n#i&V7cI;a@AEc<%Ot+1Kh4hg zsNltySTJZ@-BL=T6BSGiVqhAYPkA2bITv_;6j1#Ps4n(Akj^k+GKEXzA3N$;VzC$o z@wu!|mn<_NV}`k`m%^CGC*Dv*kDu^+7R}U2G0pWfu<>;nybqK43*88p#T!U+GJ|2K zt9lA6F2SDOVIepNNqY*xWiTD=00yzao58SV)6PSW?{r4fPU#bG8bZ?~L51TcfPb+j z3)xIFlsV_0V(3E8;2QK`o@f(5i|e3|6L9=AcJvo-JUm9s#Va?G)BwIG$2?}`g55!J zMQ2&0uXBdl2WdxO$p3ucvi~avXow5ZMWRpZBh3yihC8H9Xx{JKN&gRhjV)w{m)meC zA=jkPjBSoPJ(tTn#smm66Sx=;6hBKMPmI~rZLFbSQceIF=O}Clc20OYpH8Q!qNhZ5 z3-F_T<;h1lICqrVi!q(Y+N2$!qU?eDn^o{r6axn~9QUKqKDcl4Bb)BkZ6M#7l;)qM9O-hHUT$I;kumtKzl*UIn{NzEjELHH= z;#}tuNOK?T{Cx9H;iCYPqI!HUYXA|@f4qP865GA8XoDp=%PLbhT0(EB=X|1{<`vM- z=D`yAm(oWGUKA`!^unMzgcYn-p~LhQGk$!*37Zo=CwReBrPY`K`3oIZnti4rpb#f9 z4L=7Epa$|Trp-f|Dv!;$Gvee=IVm;WvX{@V52Pt0H0d1u0$vs99V0x1rs>=|$*>hu zGLCrxdHqtg9v zoCr}o)9zj03fH|eFti~9QExK8}W%T3iaMn4VVJLfFs?)Z%3Jq<9TpH0Lf-r67EwU6zen}bK(x$Y(XwX%*ZbNR) z5#o_W0E0`wx)%DPzkt+i{uf<=HrlM7$fp7gc{H9rgDlo5xCgq%V=``gG`e$6cXLoYCTy7oTW*#wp4{ zS1egzou}f&gmVw@x<{pZX@lAZndv8Rqp(^E*U9Z-2(n)*?F~I`2^Y*3f7V;N+&{@% z_-weDdP)!Qr_Get)evt#ZdV#Q;YBG*sBs{`d5FvvQl4e%an@$8JiYtO%O$o}wnTTT zYlHGg-z&>&=2E7c`#v!egBQZsR*fAB4)_R-Sb1$LeL=baBLhY@SMRP9id5gvr832w%_LEGRrJ7)z_#W>x#{;a00MAIpu{ditqY%f28l|ES5~Gw5yCW>diLw zTdLbjtLtq|^*A`&7yn%uYkOY1ynB{Cb7?!aH1oD*Zt&>wGE9GRPqZB93B{mWj95*&7ca5jYOY9 z4}FXfpLNyg%51Z3t!;Xy_T-e0a_qM!Lia`AgoHv$Pdx%2cx zsqhhgTRM|Yh^kk_C@H2VQBI)vzI8H&mkS**It^yM8*y?B&|u~k5Z0fFF)~2B>c*f2 zA6#xeMpWJ9GWg;rgAHlLkyE%2f}O^R9SL7#Fyk{4t#tSo zhR-y9avlo8g?bcDA#Lsa{Ny1&=r&Q9i#G&S9}ol9<@ucNB`mU@!i9-0G<;q90@+g; zerq8$2#h`SBZ6tYTyW~%m0{fzliSY3N6!o3^W0emG>AW5pLZ@Q=bg<{=$tCo{U-gj zV&f-oc*CON3g)yU0{>jrTuO&5l>jpWbz?+0$9Iy|V=J)8&t=450_xlX2LDhFUdm;L zSmAC|XMcS7m?Vy~M(^h1auPR|@O}`=jFHS5u4aZ9HL7fHKc@D?G{dF@Rw-t&jva_~ zo6*J-no%B81cyl4VO56anqP9;JLZKtjBfhQq2!S9DH*s)V)+O#UfVCjDJYMkoX7gE zg6Uh7otbHLfb8C?Q6$H@v8Zc=BdVEu!ea1c(?8I z;-ZJAw240v-unk^7EAZ+HHiF#6Fq?YtB~`CRQA&6OLbRN&vo@H%6o9Q@0S=)Wti6! z0j`za?WE`N*9*VR@a>mifYHE3t=8>17@K0TQ7!~^cn^U?M~m~Jz^pLn$}gziSmitg z=yZUObRuUN2bje{_EInzT!iuyM1N$*03+IEd?b+4&LbsOo$wbb)_0M(sq3>=^KzK- zPZ{H}BPZ|#1Ngfa;3Vu3J{E1Knbg@b zb#sFUNB+G8<$-^{!GnQdX#39B{1lZ``bqi96=>XFI%Ce6nRtv+wmnn5Wegv%{TN2E z`1wOY!nvgj^17Zq2+&+HES08&^i7}yiHxx&gd$`su+5dVwz0hC*~mQ^qW*f?$)0%d zL?^S#e$u5}V?CY@))6j0dEzIYQ~~XatQelmkRvqFh-Ke|$E3`u|57kXrxQJP4%Y}; z1|yqo&v>D3QwA~dHABqjo4sC}G8hx&2XY&Sy zj#c)y%6xRV*Pa~ixXtO$+`eIm!-6*9DU0D*>i!YIZyRIyG8K&RQCyDC1M{52ga$GaD`z6Bq_SPJBjkg^S7qWjqcCE|9GQ z@pA*^WrS1cLz!QX$#AhZG7 zvEl}rX&U({sCadB*m>;p5qh8}5BaA6f47R_F@!3QRjq?B*z%X^i(|EO3^dv>=V#KN z0>x;6$rvF|CF6QF??b?1U4e_@T(E^76VVCXH}K))xD0IA^m5a;dK+=Q?&6eDj~keG zNA2NG5ukO)5FWR+{_yk<-k_89tD8_)8SW#m^N{@R19GQ82rSJpDZ}Y$dn-h^?~Gvwk+jSJ+;T7x=QngpF{;P(7ph$OE5C zFx=%Ay1ns|v5F-uFu!<{PZ7gSuA&HgmYZY_WO!y!j$Y1%pfqklxB&rKKTY3aJue0+ zGvh&m54L_g&dpg0 z^b$}D5oQy7vGCi*7%uRI%}SBq$3Tw?Jfoh5JmtCpr(8b>bY^BVFNi}bwD%MHN0_Ib zl8{oV=LV$($f-EeRKH}6@R&r$4$b?p`__UQkM&C~U%p&W(#h22&#Qk>)CYy3!XE%~ z^3Q=IkiC0xBm4j*G3(4sWMWYDtGM@pxmt_<@;0Bw=VG?V-(@}p7{bn9U?$+C40P3( z?C)h)JQv>l632ykyyoLcJV#u|=RjmX#zp2_L^_I5VJ`BWw`j}n@(aUdKEGTbN4Y&d zep8S19^5B9O;v!7E3HQq7LG-t?>>T38i7vJMxgaz8j+)>W!n z_Qx4up=yTr8>2o9pwB5dSh8_lmOt%{f1Y*`#^lJGJl$g_E^U;4M_-u#jY^Y?KUq4A z0eqdEV94D=^8iO?#rWhch5|9w7vpAjc)njv7 zQ-rbS0LuO+BHmon27M2G5?M#dfXLqi0PDLI0S(3(d-AeYo3FJoLx?e4U#26kXSIug zGLf(NyI%ZFBUcGf7LL^O$Pt&9%eGcm+nVcr)()g&jKb3OQ11Ie^jQA_^OA#M}bG+e>$b62E=$;YZky2reCP75!Xl zyRg@MP=E*grr0^B7b!+=@61UB2={?V1E616A-ZJiC+D+3a-K)C(qZI6T=0c~Mo}J% zP-MwzBSM3pBPZ}-Wb2$0SOC;GB7^vd=y|@}(I1{d!pPkDXT4YQ0X~MXS~U@B(}brm z9~vR?m({^3LsC4W%mTbUXvZ>7(vZ5B zW}}mM{}LDe-M@gyf*g#eEYZMqSblg9*vnkx1x&-1%dSv)ENkWMGy!xLz46$(UX%)5 z&fOCU#0w~S2_@PA$_+PUO_)42V{$2&c1jZ-wZ@R9oNN0J2vZjH+3*hk5hAxGh_|9% zJh)sP)B%u%Pr-7Lb=%P?d`?e=>lN|2y8sJyul0%KM+cnW1=Ze+Wzf1Pl}S#ouat2S zx&W&{RKMT=<*T$m4<& zKb>_p6Ekw_+bE@M;--4cj5QQWp(x{&CTZtky$x!=bxfKGP=bwZMHGzot50TciBdwx@R8_uj`k*L^+%)gO}g zB19(u^vL>g%jz-U1|6F&jfyBx0wqfBkjD#LbH0E@+&ju=(?D0qPK)=CJov9IJ zZ?3GjtE-#s>e{Bu*gHCC+ef=?S1vCSX1qdPE@P4n`@}jJfwYpHHNIe#o3=28^KCBW zCBgmERi$jS8*5ua8OE!;CYrC+QTCp6?|t0Qj*)ucG2)Ab`5M06vgMFukR!K(wj)KCm6E_X->5L}2dIGa+;>5QQw{(95e zHddz{n38D@dR+k6h|c9r3v{w$-HGuLg=UJA`r=Uprb%3Xm)HPDuhZi~E+`Q8ZPbdaL~+oTIn;BO?YcZ`9K- zDU1FH%?bTvS>;RrJjOjdhaiB7sXT2FRzQ2 zhJBTBq6L4EA1jNhV`!~?HC8K;8+kBh=zuPCNsRN;bpP>7ZpP3(!D)iIANSM|t zgiF?f_w2>nJi{wS2sv}4?v}6TsXjbJfCb&}e3Rb*(>y}=;-#6+0TR^~9?)nLrI<-R zt?Tv@X>PuWQ0OgaGOp`u@VC0FOtRK}$ zUB!K}UWPnA-nYA|I#3XOV+BJ}ZL_8^$Xy?#E(+u+pz1(9il>SGK%#fqj>cROcrLPuk;y$8A^rn0~q}`OlcK6=R+>Pp@$R^{6r4^;80K1T2wI z{Z2VVM(H9*O2BoC7lLuHR%97; zgJCBaeSr$k8Gty#1bg5@`~XlgbtGKB=jYk!;c31JkM&6W$lsBJn>t5bQ<(2_9p>UD z{tHMi1lp#11B5il!56R7>z8oBVi+hHE599k`_IX7(`TNBChhrxj<^6xx*W&+6~<%3 zL&mCa*ldUh#f?K|M$!(nN#*q}Fd{fOWJZGvE=`$MG+%))g%}Qo@Bu33LU3Z2p_es4 z&tKrne#LddqObe_Bt?c1^1M)dE`yjKmq0I=>oIhN?*}6VI!r%2Iy&^5_k0zQn}p2T zk8L&rWsLeBVX@q_7Z*Ne&A1t74eQw6(U))b9Jkc=5B6mk(e^S}Bn#!be5G+?#LI7| zy69slovH8aAM9FA-_SajYR?56UdtRKoF-g=lIMoQ-=$6yD;i?aDC|B%l&7S91)F!Qh+4hm8jy7wNy zE~^`{_(WSvCQ^2R2usij6ot!}Kn1JlN<|aCX_U|V@hYxnr7^%TXx(1?=2;pph?;}p z)L$U18CQcCqx-elT7p7#mt^TA^z{+Oe9)9$N*%7Fg{_nv$wxUuMdW3MdZwEIzEp=y zJOMj6%4ef<6Com10#WnQ-+a4QnI9B<6Q-LW+Mjd!@c77IsymiJt_YnU+&ZzO=EcKr z17o}1bkFp_mhe+5*K5{vs!dY zB_v5l&@l-YR-D~QJW0Xw=c0EJE~Yup@C*NQSag@*VC>>9Y9v);yPH~3^%wmNeFb{d7E6N}sRD)zZ1LWE;nTL|22kfWbmw}t217t-?F*Xv- zBXJ)kq)W^f#96zD&=p<=ZfnU7SReA2PVjqn5A_l5NSnro(_NG>s2DvIpPnwT`J*7` z9KewavkHmx~qIun)>pjsp_1$-vHHz6K((=gRdV;R_ZHfBp&iJx?u=!tgd;t z3GXN(T_2q6x4lyt#9jZs9?3;K=e680h%T$V=Yl@=QSQ2K0=N<496mfbY`ez?*?Nus zCjQ<-+2y=g=(T*YaITuxQ(`Oa+(~{RkTF<;+qV{qXD??`1kyc=gqHywv-h;lYh-R? zkF*Ox9a2ejeNMLI13#=p&jBu~%36?9(IkzMP#2sD-Sc+z1@gMVz6e~@C1{w_YnSaZ zkep*)9r-(u*tF3Q*})!^7i~wgeLJ0orB6ZpKHYA6BZwPyI+InMffxzzxj$ThA-Ijn7P7L zrjf^&dAt~-3^HA;6<2bUp~RkxYa*rd}1#1#sS88hJkz7U>p+D1dGeCom1xMb}z1sV<5#B$q((rv#IehBL|RzTk^# zkLXZsHh|0MoLop|0vE-nulOg|#_8!%J3Kn{>?KC)%G#P|jEQG9_iR~x_#For>vgV3g#q0AD2^d)%EF3L$hSJ{We2{Mcj zR|t@oH1*vc1JpR-8*QZ#W(gIwW3V542;%Q^?NGSnxcSK~fueTkN`ltBlr1YZn z;Ral3y5L}t^U3B7J_ypEICzH%)tx46RYlCIZxlwJ-vG6jO(ip`c=P$GI`9$_i zZO_;0sQ-cVuqUHUVS+J0ZNLfoM=~AaNZAnr{Wn zms1ew1fxJ>U;7pSr%z@JI`3X!t%0?lqUHLHd_0|Ygy?P_O?U}@7?|c?N+*`<0Z77g z0&p=i_KQ%R$iVQ}NZ=XY?p_GWR{i7YCrZ`pBJ5x#Rv39ZblH3yln-r@cYmtl$UJ5g z_F13M_Bay?iw=a18qTx{^6CIrc_!bgOZ2C_HU@Js-(_bYq=((I0`ctse;hV-?h)TQ+0gk@3%9FUBHXs)$5Ni5pE`sDim9=IY zAbw4T5MgGA>v5(H?Vj#TUH^o8R)c9K2)LV!8Te0Nk{OsJdQ6xwcb{9N7fj24KM40Z z0iFtjINej1VF3S-et{6vG?oj*3*AWTy{(xs(}mdvs*MF^ZfUQfaCuX*^*_Mqk56>w zO-LgZYmf!BFP|!AB*?5VH#sLVOuXF;9tjT=FXMRoK>ovhzeMEqkunPR z_jgsULTXp)cce7NdEX2u4%Nh!q@bmr))x?wqKLIlxZ^G;-8zbX4&sNFKHU2rWokTw zG&-k`r)Pol5C)x%0AI8mNSI$BTo}VKOE+h0ZEJN+zU3H&*0$QgvZT$E-z}zN#?*|h zK=_XspdZxFWtzU%p`Rx<4EY#`+Mb)11GNj=j_s+fZIF_8n6Zf7!=aa?aHQ*Ubvgs> zBQN^Q7?AezxS`6M7pR}#O;B5(obfWBG{6m&%1xgZtr(%7Dx0EQGi0Qmlw(SMlX1xO zm8dk8ix&9!$cE~?FPqYwidA397*(GqwSm3sELj?ytcVuwPOldITp3cuS zWX*-l7t!t?<3@D8_d(NE8OE{vDK7G!M=s`~!cn(XBXR9EmtYf|on zOOd|zhnNCi9x(Cxzcx|6Mn9faURN8raZrhCet!> z=y=(>Iz3?= z)|OWNrL=YLF2T$>AH&d1$Xq^XBQH?nAmQ}2k+-B6emR@sr$3{DP{d^g91mc!VRbEdmSxY}aqBMTw28{LO^h5SNK2q6-CkO3N zAn_;5t-s(QFyFqmsFqDEX0_cEMlKaND0OtG5$TYtBLaI#UAZekyeS1Jci~~gm zVmDupv9vmAOWGR9fW`A=5ja6}H>z&h4b2hAd=Wkk#dV%(XETZrD&?27o0#vEuxN>;vC=Z2XKZoV=j1H+@tHUU3lSRMFtRl`m$&2 zM2j&@K4ynAPEGW!bik|@p{I>vQhtxtfs1v)Mfodk$Rpr9DPWY!OMn%gFuV|4aJ&V@ z_&Fe1l5f%_4DUBaTLQO1!a(F!K8!MXxzIC<7GpSTY8Sz6WZ!%mq{O(5@yQy1pSSgL zd#3%}-h4T0B_Ap98s7a~8LkIy`|+c;y1v%lk)eD4&Ykwv-~3|x*`NH$_Di4nZ2RgL zzu4aT+HbXA{l$OLe))61*gpHuf4cqLkNsr(%#Zze`~1)ULVNpbUvIa+`F4Bq;9)zF zp}V@m3sgL&mmEij?MTM(f#Q#3j3X-sH?zNG7#<%{XBkbx7^11y`3^D(qdF6U1QIz! z$&1>wNHHVzT=JFVxLenYH9)I`s#n7BN`HcEXFjdJ1M7uOQMS zjDK~;i(_2JbnBtMWq!Uc2E1d0z~hN@pBp{JQNEK#{-_|bbs9s`#xTAy4`AK7?$D6R z`f1iq&mjveXaRLmIsXnYBBQl^^oxq7V_~I$;9hHH8W}m-4$mb{bW& zL-^skW|RZMUxky$eHs$WDjM?kx~CoJ+y$yPS#cEtNAmEaq@nJnq1?1*VZ=-S9cB$W zkNi1s%s`4^Q5)V@{=zB1!m6W~L=E6RM=X7z^Dz-9MzkMoPIiThoQP zo$f1+?YGW+q2>fR6(O{{5GwuTMEYhnOKphU-l|Z7&!qn)@z*Vk^03rlKV?hF?gto{ zYAMb?W_NdYcH8#$PTSqn1zCa|ADySuZ-TP14t+yVwTW>#-;e5Y+<|lul{qrCCC+(+& zKk|3}Zu{~7;-9w9f999k-M4PH{U>{Ee~(!($+mwaqkCUQ_uY1MwCfi(IY&>=WURX% zst>MiN_O=LsgwTg^`5m01-(RdORG`IRXObSN`28Wt(ug=73Djj&3ePkraV*#W6Vq) z^O!O@ml%2EPi`H_Cz>+s@-(GCF@Ytt^ly_RL&;%IKEE(VH)Q#0(ys ztLe`(%2C+;Sx|N0$jF&?u5-TBHkkA)^h!O{3Dr)KDE%vA6aC&}1^rfZwOr8GNAyYm)dhT}@ndVYrTVU@ zO%K`=(H}^Gtl>olo)T!Zf`j@swdF>tj}Ma7=c8QsM<+x|*Hl!rY)?pT%Ov9ybi1g1 zSi`%eG4RT4%`X)msl1(|ed%F4=ep~M&Q!WC^}cSW5QA$il0vsRWfJdg@<8f-CY_@# zovSK4`dw=4DwA2_O_s~ZFz1|P|K}GCk@(vQIZNIL>h!VNC-|)gx6ku}>R^f7(J$a} zk*JpyB-->5c@}l^Ou9M`mCcS0KEFN`XiddAhyB$2M zw;(&AsqtBE4+NhAPVQuLwSmu&AD%rt!00csUb3NQTL6_XRGIG4S?y;=uAbZonb=9EM~6P3U{_g?QW_XV>;x=#9M0CUNR{_Lhql+g2%pPlA$% z%*@6Naw$JIZfmod2j5d&fDd(nSz5(#IMYR8wO!lTXjj%Z+nNUGLyT;Atgnf$bf9uQ zyCXd?D)I4xV;LmYjlwh!H>eJCg<1gjB{!r1c{-S!-z4rFD5=l(_e%#Z&>`^=C2c>A@_ zey;t-FaC1dd;G{BY+GNyrt+7Re^-5yn_KlY8Hl?5!=}-V7dL}C7i2W>WiR@opA5_? zx^$ax#R7BVjp(OAj30n8XW*c?fh~Q73@X>-(loQ6@)*d})2mN&bQwZYDz_&>XkX4> zwLxyygTPeg`4XQM2B|uOa3PlKHhN^22@G9|q_41Sf%0w85-FK&S5H_0o^S(&u^fUC;XMy`2vt9y<*No$?Ut(4tcCQ zANTA~l`ab zq|fCrLXt>xq7eP-M?p#C;6ckJ(4i?e(W1*N!H}20zz@>Q9u^Jpcq_f?;F);RNhM8x z>#&#SRPqX|*2MW3PiXX;Xho1vC+Wcng?M}Rqssx{qaV7@R)k@a?R*o@(a|c8+&pVS8M>IrP`!WWwjXWN7B*D&_ciz(wMTbvw|l}P@3qMs z=}Novc&|O#*%P0Qwt4kR+mK<*Idt>p%YM;heSM?7`r2#C!)%ZGoQ&ZE8C;Kd9<_h{ zkNm;*Cx76-XB8e)vbdmUl_*ii@$Zp_>wJd_MWT-`)Pr zfAi1l!giIxy3=xw+wV>btBuK9aHe z@bTmJ_>qj&C)JKVnbr2-{-d_H%gi!H(~|h@nHPGvcmHAA zm2v#aOE0yLzy6xij@nEY_I>p;`sc1(>b)WQ9r31JIi0Vne{IQ_hR@ZPuS*xp?drAb z?bE;e)9sBjiS~+)henEu1q8=7 zO!}&fo1gr}pKL$+U;bD-Inen8FDYz!Mf}8P>*Xu$-~IRhz4p7l>-Xw>(@m%7;ozH2 z)rBzN4B8`54i~RnotCF|Y;DHZ(RPF^oABfF6?Q$yHel>K!G6w^ zMVauyu!ill?=J%KxGam>so4CkTC*^g*DabvuHDj&n2HO2e7aYe1)`^NW} z;zhx?#~A*+jA2wE&C-=Q$wzHUwQgY0B!Z;PNoulb#04;6X*M0~pL7cNl%Ywd?;ic9 zadMTXjL6Dgoy#ui;*-b!1;iCi=x`CO2BrX|Kwxj^^g~ZYSLx4d410jWqr>wy_fTI! z8P2p>2~c&Tapvm-iCfue7PP~}8(tkoV;I0+X}Hg33{L^_X_irIfVJ z6GbJeC~~C#Eb<1XZ9(eLwsNaw2a29H=1`t7vAdT;u||wJ9^JhNxcJ5})CE@k@v#ex?1`5B;;Yzqi+}ZLGH&*RQ$V&BL zUu?JJBA;x5cBnq!b;A4GGH`d)KVn=m)9ZdF(XU^-s=g%tsxN7e9zSWDGSXjr^<~cl zujpjJ$VFE4_he_M?d}B$p}`x zF_>Qa&QG-O{=MJR-uSNXl0mK;H5qBwuiuQJs5bHqu^NN*&eaCFc?RSv&V5kj`?^BI zDZDD<CF#XWOXnHi55 zInd~N@U9dtD>o1-zJu~9%uS9#@fbpEFZTFg$IF};b+$IP+PcmmW=+>P-&Bt*HAuOh zWquz(&YBSNxio>?u8N2HKjRl)V|&8d?b8F%vj6jHyE0=51z%1>Bbh}_b#u-Zz&jf0 zx)w3U@V46X^I!hu_J5W!{OHbOpWo;nU0%7mrMATw{$IC0_n*4s`HOIL7CvlZ$%-h3aK9W^IQ<~hB576&+&%s>z z^DvhJooTrsTtezKEZGpOd`xTGZ_U=l@k(g8-2 z;-?t@drH5{4#Unzkkn2?n$rsl;_p2iy=o3C&!pJAo&;LI*u@uW?(kuB6`qA z$~@!kdbXQ7^~~wjtxZK8w+FXxwO{$gUub{-KmU8}AOEu-ZJ+xW|FZqs7rxlO`nBI` zJKNiy@!JY_dwY8aZFiUD4!iB{gU9}&*!ud+Gb*er-FdRp zRA9S#^Jd#zUvD!Z@~~!@daQ4)w~u}NjrPXt-{~0?+GKe--{PNBl@F;mM^{OnaXY7z z<{d1SnS~T52%JAm7SElyBhLVFl}?`S=}=_`nBC&ZOCFWh%%We=arx=dkt2*k(isfV zNqc`7`xTJcBZZkklq{#x-tzKUyZi87d++UcT01#zH?Cc2uf6<|AJy|>+3M;_yLxS- zz4G$M+k?Az+E>2#n_f2aO|>=4SDxIz-!?WkJ+r;1_FS2*xxd_g>#cVG_TBb{U-`B6 ztN-#>+Bd%bW_$GLN#n&eW{tO>JZ^ityY0!7hwb5`N1hGc-#b>{$Sm`F>SN62@+g^Q z81%oFUwX-V)$+vQ@bJ)Uh_8r-KE;y*W;5A#jgNLPbI9_E9rd&Ijdjo7`U_}E+uPYw zy`=%x2aDIy{!zPkPda$#-S*zCJMH-3!0TE$N7iTSeb8_GX+ zk>TJZ@u^8+JBsT;_W)*w(HqO3Jaa2>UzB&+ckoue^vMgLlMi*9U=q{k;g$JtW+{0; zL2aXU;2fO!=}5s;MQpPya&OR~wVB_TFdD@~!I|?N9v4 zA84rB~it12$n}dtuo(7_yX2DNG&ht)~M+H5oaPK1hpf(!iwkxL_zCJ0S7twfv0HJk^ zU3_@(#VsgrEp2&#USDvNz!19@g4P&fad@mfS3@*ryI&Cb8Q#d_&cI zgGRaNa!={>SB!Uz-s7VaFaJ4^yl<*+Z{L6D?bx^h+`M*`yB6yYKE8R9j3Bb6HG2Em zFDSjsrJOQ`&vYI@QHei@7E7)n;Umyan-9QEfj$*eIaIxpA3 zXW{x@Ub*aAf$_^t2r{}ZO5#T6_Wj%KvoeNvxA&aITbT)XVQf>z@DKcnKPF@N6E1Vr zr<`(%AN`44R0y*~!uE+|VC{0WGeU*ZGYdG+piMn!wn0}M?A9_$Xt{BRR^_N%zN|=o zzs4t8u1O&Jh`y#}1s3%U?<^ltUPt2-y07%5E{a!iX=lQU_*cUizWb&tPh+OLTmqsQ zRk{F|CQJaDPFLi(E_W_}H6C|9G&TOjIl5!+j3P1dbz)S*j~5On1|8pdnNh-|Rlb6v zvmqk%Nw1Dkbd_dVjNzGN4#BBh&g=9`!0Ev*-BpIVomTnZ+*S}{LlPs z``dr>@3bG1A^aO({8GFB@Ns)2gZJLONA2M7*t2!D{g7Xl5)U03Tc|yJydy(b#s@FI z$)Mpfe|W&lYZydJ&JP}oc7Ol4J$|y=9_?;R$IES7hRi-MV@WQIng{pqw!8Ok%UF?d zv$o;wgP4)r*x1yCaa`lbj1l$XkiIhl<3F>)@I=|GD{Ee#itM@gORAn_+GqOp#A%c0 z4%VR2Ng^5s%2>Sr;DHR-C!X2;#+%=04<9{l2QoN$F_IVM zp2!e==bc;a&fRE3?# zUH6x*%`H_`eMoi-#_rvF4?OG3lYrT5O>$k6k;*LZwqKyzmZ5`7JurZdWW+M`-BH<_ zS2k@}-@1L*eP$*@o|_*&&5Zj{eV6{mjlhFPGN3WgA3e&JeyeNk!Gni>dGGe^JMI46 zhwA@tw{Lv;%k80f^Fk=AU-MaLK6$%pqmg}lX+y7vU5R1Em`10uE5N||#(QtJul>ev ziHug)PeRk@=yofr-}n82?{BZX{IaqsA8k+5(bOH%Hv;Je`pM&d+A%af7~J|!iyZOG zRlhJ5eL&WcZw$Ro8#?7mMkS{YnpTkOV$d>9!<{D76C^zh<&Qx~|K{d~7roFCX9DK~ zdh(WPqH{LWhDXB4mNN?-idyvIDX|X^_uG4S-)o=!+0SZ>IB-v=zf&b%#${&skN@#M z+CKK$cWAueWm|3rFsON7AciC__r+Lb%{DtfP$(q{x(uk2^Vi}y#i>eDoFXe_fCxT7 zj8Gu@m4u$brcKzwFMAJ*Mqw_YJiS5()&sfZxSY|uUgma5p5!V7MmI+5B~U_j#-&f7a&bZrWFla; zro!`J34;Sc=h8c3QTQmn(L7#UrQ&wSG(Ejw=+BT1fdD+_xb|L4%tJ7~)E% zu6M@?h*H3urjhVV?7(+XE{H;rtWyN+>X3dBbX1QUb9a~qp&2;!1m!ug9+jet0sBZD z%24HT^p1=JuIul9;~VX-{6GGm?H~M?|EPWDvp?Uy`c)q9vYW(AMyU*v-2)k`NA2b7 zS9M`IX%BSq;Su{xM(Q{3JZevN4^^tl(7EQ9U*uv09|(^`Px1Ak10eH`-%`XS0=d^#-&u%yrSzbUg5L2Si8}bsBN@Ms zA9}`;S=;++-^aR%c`W1go%imvCy%$Y%;iW%<=&o*+Z``cSdxy}z2S+BT@0%2?Oo@8 zB;%4sH&5A$eZuI-M;ecU-o%KKVi(K88L=dH)ft0W@%6@ zyQuyM4VYF$B|197oJAuqg({q}q5L>Gx8f=Z=sX$A0l2ce0mdj%BR$vC5V<7DYwo?Q z!~&g0aTPno0Xi=qEd#&0W-~)9BM{ylB4wvX9{W^=-;+*R#K40~*ul|3yYujF`^-=M zoQz@J>>;C=V|3GJwr*^;Kl(@i8|~w-eIj*LeU5ZK*dSG1R~^>P4()HgxmnRsw22|) zzfUrBG|KCQ3xW7bJ5LjXSiDsPwPI}Y*r=X$q&loDF-9p}Sz~ZRJKjtil375Ee&R9L z9y+8}0n!Tol`w`M-+8n9q6}0qs^A%6)Za5fk;hE~$J=9QNT=pVa}7pihq=2#=$&T_ z53ce#gkKe0G=>Qze-C$nE*V2Vmkhm6fW#+n$~n(xA!0$fbIDE%-w`5T;yZDt^Kp@X zho0I=JB=K?<816a&hwi3fyHnjClIA19Ug>04-)baSMk8qAD9oJNlT=Qg=KQGN^uc7 z5H}!kQwXmr(*&j|!V1%;VpzDeO1=O_83rdF^PhKJZbF99UpW&K+3e}{nunrxG4TrI zyMT5cdX70d!``>+1>_N6aZSO+M z`qR^PC_2wP^HceyCoV*uO;ebQ3`@sMrOU}&nRHBDl_!d2`HvVR@O_w<;N~ZHZEpt@5q))?pusYN;4nhgkcwgI>d81&Pxm=3_C9D zTxfL<;CAuM?e=3Cd+*sue)2#@ulmB{Cp+z){OECeXPew+fDxE4$%&JU_{ZElssFe> zYHvP+%#1F}Vwj<2`3%O$rrM7@v_E49GpWeIn%|GV@w(&h-GAV+nH9!Z#&Gt>R-~tg zGN`vUH(Uno#Lf?_O*c=K&l>ep$#?(WeYYo1D-ijKUXr)Y-3e70{3NrZ>fJDS`K7mX;FRr z!Q18ONALRjyKlBHf8k3x`IMjb0DTUwue9&`{omhizWj>ncx?W@LE@Sg?fdAqRY0Q1 zQd3bP&KdtNh)G0f+^%kG>rd-cZC*u$@pcC5Lv&N)}hMR!WN1bD19Y@HA!Xn=kFQOeG zZ4Hr)T$RnuGV-jdOwyRC4h!l_dzfnw9XJii@=K7kc#F=yi7+qP7#@4$3xz=?A1Tl# zZkGuJJA)~s=LYg!gp2UBv?=uTOXSH3c^($2$FS%XU`+A2()GLKwy{)!Y2_!^pt+l@4pTz!bv@Dzz(DQ>0u?&)DK*y&!Tf~ z{&5^GzEf~H9Sf!;x=ys&8ivN1AA8ScOKoFwt^LB!{`2;K`M>_(y*Bi&42VP4jOrr4 zyR+Bs?;OYwI&Ifwl)Q3vQx^>xa;wYj(;s`i?d|T!NZ!%Foh1|d%*@HSsPZvF?0TS& znn*xX!ickkbCw{G9DamFk& zhB0e-W%jh2H*TmuuCyyMhOb}0CN*Y1Ym5Wd3P0K26Z;qxI~d|>cjQ}Bo(I*RsmT6Mb$OcuIrF%n3s0rj`at>3{x1phbcp#aiCGXTfLdu!|VIjYZ z`kLdazCHaMkCbFwOer89`63VlZoFaT!#Oxtj3N0ig<^U^X5WpgV3#fQVA%cCY4H0O z7`P}sEJJ5-1gIFdPx&dQQ>o9y^}mD4b=Jfb-vOMZ%cjIJWkS=Tb2o%d=!6~y4X2@eOtF?EA9H#D>fh)lwZ2C+4f~r@Z^&H za%qeM8Lhko$7~<7gM*l&FFkmsLjyiW3+O)61tkw_M3VdME^RP~7vm)} z!K}L_4@U28-Kgx#*ye(c;kvW4ZQZ}}%F9^_vAyFNWsF^3lG~R-=_dvf(QCxT2l-iM z;TJO%UQ@rj#_kb9#vW$0>A_E)Ja!&l@MOKYI)(bfk^3ZTfQ>@*Cs)68O{ zlTUxo_sBrrYa2@Eh8~^4&mg~N^Mw4~>@&mkN9gp2$n{xXUunPh_kW*s^osjt`eWz} zG^hO4s`RB!01HZvzk~A$?N+8RsVqM4Dva*E7Fs|0CdM&&^TY%Y9Y;>W0g__2E<*=1 z7cD0oOp7_ukO0JWcNo$k6Kf35$Z*owdl=n#XAl2GqeEs;$(78R8Rn_PLA(3#Zu|M4 z`E1+Y<|~JcN@Softuq^Eot;&THccYQZ2~enf+{jb-^frTrlH!Js zJ>8v$tZkZ4OPSzhlpVT`s#x*1<;u zvweh}*HGae6p&Wsq;C}e(3*9pPiY;X>Fz+#!TI$I-(FV+f1qf)OTPl`k7KAfU}Qfj z^EsfXYu0EEO0Ee&lb7YV(6siWKk_5(KlxjKyM6uZZ+V9Aw|?u*w!OR8HvMHbUTn%H z@yr6jo{wV#-NK)3e+P;^anqToz1N+Z=J>#Zf+B$ zGYca-s~yO&^5=0ih_h~!HOa5ObX|trL3_aCaM53pVa}uQ$B(z$BNtFqH`>ij1x!sdaF)mq#f}zACV#?x0G`>bh-7#|6nSq-VmMZY|KJ>oBTHmYJ)US@) zD>q+iue^N2H&48TxwWy@u3p`;k-9Bf&*)0Dt<6oh3(HutwH_bI*t3z25e%;@TYOLh zqgt{k51(S@X#pQ)b(YIa~ao%>VO#2UJtEyIXskcHe2;m3XFGV*4aAl&Nsf< zHgta6xN)-`ijKaI;m(Y+4Jh>o>X;Yk-fCa@!k4jddRwPWY)4=%)bQ2Wm&eYZz;b zTjfk4WzOqw;17i0axh&mIYMtEu=w$lm+G%XYeZS|3C~-PcD(%-kFD9im8%H5MC`E~K@`mFZ0W*&g_ny| z_C7CbbfO=wZF{5{?ix#q_l|$zl<0JN?-I4LT?HN2Cv2Sh7+I1PlEZ4Yq=Z=i& zO`o&O9vfruoVp^GuhjRUO(*`<{2Vz@TNF?`_qAWf|!B zP&ZMJA8z~QKi003pM8h#%iZ4DS3hMI`bc`n7uPW6>GzM7ex&#-TN~yD?8{JoA|v?r zo%=GN_uGoBCnguRb^hGDb*J5W=eBPQuDI`*5rDhG%Z$zw2~3_aZ_!^7ne zCx6lGYHaZGg}&~Z1ZHKIT|Q-@JLFTn^v4*!b@xvD?9Y6rJ-+`)G#K}M?UJVyYchs6 z+Q0E9{QS&cBhd$qbTua{5m>4^u$4}1E#%*ubYpObqW_5q5!cM z(E}`pu!db%StgtV^33>BoGN#$^NHDV*V=8fqCN#b@{ogeiOvc736KQN3~=%QQ06$| zi#CR<7rt0<{nLo)%AmVY_m5(YpXgP*lj)RrE`Tr;M#)8ozVX4|y7xft2*(3n_${Qv z45y3otL@#0NL#cH7jGYOlF{fkgez0YVA@`wXqGhSCmgQ9Y(gGm5E_ydZx8Wy1{ngK z=L81bOrp6E5}l-x1LbiCnu}TQutGXKtq~@r5xLx%3m2g%le-|k3^o`(xaX1P1v8LQ zDs2$-mQI!(zCzQ{CvsBaM%5{W#GG8|Bd#w&;x+gX)HysLValfM@B?rfe#0OwfSFP3 zICzZNJ#MGn70UBHwG|WoG*gZA$51bR zM!WOit@hXd#(&g)CdQu%8yUv00yanq0WF*=w9yn2{!Aff+4s^I$2vF)%yDUbjvh z5dO+$X6P$n%CHe*9jZPu3dz5JAS2O>Gf&ia54;0IV`q?i58AtL-}1|I1Mf69jR?K`^ZIy}&g%{9M3cx7!v((1;5Wie7L#;{t##uDYR#G|KGpsPuw(-YFju0GyxG|{cc3-YON zk{3;i+T(_5sk}1TB{zF-(`PWg#LDcEi9QZvpS!qpK6$AH*9iwb>rD*(Tp4wf!nMy8 zv#j*fl?>Tpu8~8cvjj&np(E*Q1}xDQ0fP zIrq8oqEARj-_-2y?yK*qSk-6gK&T}30*R6Hds)IAs4UO_iVq*b;C0(39q_646&cZJ za8EbS`$B2O^|eODfOep6r&1DiI7wv%a+QQCq7LGrQQeKXR=X`#7xmb-D!(|1iXK5* zs{Ey`wj~{|udXR}sqO2WJWyXa(0P5R%$yI@J3Ejd3x+f89R7&I4Fnud6vm$=S}IpG z;ERSAUn0;iDjegJ2A&z7LK|iVyqp{O>LLcnu)8ovO zvVF7+juv{X>M|}C<6J$jjWEZKkR*>}{t+?d{ai5Q>p9JCfs29`aWNin6o+}eLra(t zsg_hSdnQR23evd9Q+Wn&;=cZ+FSWn+SN>x!hhTZb?R$4!&8t_o+8b}Y;k{0IS?&I# zCo(W)^lompd-Ag>KSmEPTQMt)vBbSI{-ZzEb{;=&XUEL3 z$ha0CuJD`d*L2Opu$2*aBzd`DaWU1!X7@lBQC;{z8>Aic@*R)jb&15{5Cw`~(p}%u zsxCCT+REG%9(8xbvlU&$g-z(09xn0AY@c+bRDQfPbfAkIxbNi*C(9TlGHP|9R;uC; zjjc&uVQXv+CRtCp@euv#UfY+kd!YP>a!)8rxV*6gPS4g$S=6* zrimtGMwb19Z+IOy2I+CTc5SoWyutp#D>7mqx(&9rvbQk64i;>Ucb#>@y6NJ{#CLqh zCoRL5zw%WX@4Qqb4a&vnW_B96*fjuymYHVy9L5z-9#|5wCHVlB)i9IIQ5J;54<56;76ZlMwulgZ@u-da>+?5H1(e) zAv!H=OG@u)%PLC}2r;FT$1(^pB0?Qpe{>VX( zJX@06!+gMnS#$FfEfZ|?k5$E;y1z>H<>YB{$)melg%u~hC(Odq9~EMS&Y5K8%-L7` zzRrn#m3MHa@-QN0wBxFah+2Lb))?~uKHtC#k9A{mEFyo{L+uIfsQ>dsfSaneyspzx zKDGC1A^k=??Dx%|gi$-up8L|r&dGtwM(0As;WDsz8JP02rd-0?Soi7WDk+V-AnK1E z)P6jnVf^q-40(O25e@C4pL(Hjram!~yll0|xkZ24J=&G7R6p9BHl$5eBadP7txBh> z((RhsbyaCI{3>%<`qK!fc0P6-_@aSoetUqkjh1{ypev6s^)G-}($$R&96hpp1#=v_ zEF%9Wwu0oNlyT|DC>bRkOfMSNOduWx3u7j$v~kX|#@&jQavn$zw)IIAmh zfmAvHF9QseGG+kkT!=tKyBPQd$}66ICCT6U|54@+OTDPkQgFR?L>vDMkJjDou| zj@jRr8DWg#`}gm8o3YIeW*4ynSO${aAGWqOe3L`}q5W7N%wECkGEy-p_V&A{J1Qu0IBr^lJUcU=lI|Nxt+) zwW-R$Fu!u;nr{%_dFO81la6`m?ulgo>_7g;ZBH_j@5sgyUr%Hy4ljf0JW$!pE&A+G zS?CBqW&1iJBZ?;<(w#BKu1OEb$c_;hmvUFgCv#bdJj)oSE6n5zm!)H+ue0?b z@?z}5pLUi3xQ2lzO)_sQDMY*6uF}T3*~$%)Om&SDo_(dAWuPugziX0B_l4p?UiD41 z&%jV%KXH;>wAWTFhJ! z`S$IJO*UIJDUy;Xi4tU4mW~s2dOFCDA*WUHkylT~|RaL9% zsiz*>MIkL|2&)P${Qw!I{duyK?Z)n{u*w{195^oKL_9uK7;T9e7Oy)aGchqXs|Dwb z798VN=6k|%M7xSVNDH;GOsKtLfSqLW(_Y!;mJz4nBr@73fSA zGOmx0g#6hi*!~1R4liyX9~+UbNnx7%M*;{DHBG!gZcp@mZ`yWOPul!eWe{-v1m{{B z5BT7H$p@7LXR>zU=R-ebDc@GQ+fd2;Zcs>5yQjbCFD(SvM5dkflH1-EHBRoI0yS+y zm4U2(%2x)FF5{4gbhI)GGHpLBvoXM6rR*m3%O`ysO+OucSKd3Uyb!9gaJ9ooanEtRbI-4gbG8kQK!f98Io*Kk+i)N==e%9!W|1 z*}wAN#M|F`*EQd}NAc@%RCbKV`}E?~YIwk`$d+O_MxTI%5=Db2#apfsFIL?L3Q zk1MF4vnowYeCtvM_70By;$2UZJPJ)uiUN3`hiXBR=@KN}cAf**rKAT?=ux7Umu9?M zxi~)^OAE8k`)j}Rjkxn*#h>w=nVpnkK4#?yB?Hv+b2EP7ZfSlYe&%QXY<&4get6hA-qMk`Ym z&bv}B1aEI| z`X#)%=?O2a#%$)5{m{Mzm<#-rkMr$!=_It;e*K zT|WN7V_iO$!3`w~SLUL}3vPQyF)=wFTN`;L8=v=P9G{+?j!iAz*r7fYon4i`wX-M1 zd(ZV_LC4E^9VyH+Gc)G@{reBRGfqFD?Xn$VU*BB;FAG9;u;By!YD#px{>n?SpvBjm z3E8)Z=apYl?s`5}G9gmmP$enkpMI+e}-$S>x z+{dC*H0GifJVsE;Q#@mk z*X_jn4~vqWzxcoWpU02>X99UvP$I@F!+3eCtPIXuqHO%1@V=7XzP~GHh_s#lWDn3KWL|+*Gy& z@U}pc^c*N%i5T?CEUEz7ka0-$={HL8EF1v+Be&R^i zq%f6Er?948xaA>PExX3mtXRfV0}52din3VM8*dp%@`p(BAp9jSWto;*N*~%G=SynG zWlXoN7WdBK>O&hRGM60{_xa@bkflcF%cC6@%2iz1VmA=3)1^cI|0u7~ZuGHj3+%*6YzBEZc+B^n%FnW+R>Y8opcgd0fd zBn40UfvO}|I6pMB`S#tR;=Y)x`q^cwGi2}o9#0yhnQZ{`5@1w_3R*qvz zN(PhNfxs;ly@}VlpZ9fD4D>s+o+LdJ~ADx(;9+$#374Lj- zFMji@Z^z=|qG&|181+56V^WHl0O0kJCL(t8*s=Vhf8;0PjnBOiAAa;v@JZkU)xo>k z%ExXiFNh6!!sYmci732G>9tD^`}ns^WINzyQo>JK*Q?e|xTrfm*FQ5TpLkQ!faJU^ zfQK~XsNo6$Ewfz#T;_2}#~H%8`8g@SQWSbp*pz_<%xf>*h=uuC#h=Cc%6feCA-9KY zTNzg0ig(J7_ef_F%#p=gf5Sl9l?6Ee-uw1uf@s1k>`1nr3fFv z(Sqa3@}kB&@mPF6I^)QRz{ggUh5eDauv0+j;`dh#4M^&z~4 zi!cJgA(L&Ik(n`6)W+S06}fi zS`I^Jx)jG)p!k|6&j%5fR>i{$78Fv0+cfn{O6bQWtxu;&s5G}B2b2Ul4y0Eu!xO`0 z8CzvEBCK?6?$zKvoX68dHwjyLi#vxD961nFz-4ctjTJ4SHJAU>V{l#>tcmk2} zfKmem3{u8a5qPX&Fw5{7Owr)QU-NtLS3u<6K>}#UjKkK(^Y8K)sZSJgX}3~ zCk+*z_BY%!R5!ARg1GYdas0i%{qymi_dir$Whd-mtZ#0{6)D18e}y8%*WtKQi`_B> z#oY&wyyL=|VBUyl7mHn2UQ*&^HJ`auLI!ib`t@XeC#J{8tN`($3tp_Fp4>D+7-}Hb z8F9UqCk6p|*=4CmB9JNuXC~gtOHUm7Lt&kt8jDw7c{LX1CSz`D+-Fc%RPL|*##g0~ zOW~Dbh7ydThJwSR{Vt!FmO{tv1;Xp=wvum8!Il=8B(y`^^K4&|IUQKJ2bT=eA^ z#H`uDNz)5IQ6ex7RWQ-1Xk=gK&shZRaV*E&Ivy}NGD#^OuMs(!=t*4FU%}bHE*rco zG$_b_vQtsRJ`3Ky=M&YO6{oM8UZq)>n~Di7d{o)k+SrbJl7riK?)v4&iSDF7e1YOh zQ9f@D&2sbRJqsQbVQ_UiBUX$jJ6-X0Djqy~sP>qQQ}KOfa@=1Z92*&No^0y@U%R)q zwdGx2aKmSI=Xaz?-nsWs3Q9NT=NE!c&+<9t4?p_Q3pHc|SzBCMv;w)Yy=j@C!_Cjm z1RwoaUE%9;8&ZZx;?37y_0tVL+JW40zwo1nTKGvh?(jrIia1vTLYKc3CmOAg$WPnA zCjfazNufRAH+89U2@kMo3wDS{CHGTfqp>U{ovZxbdgE2gD@QY?=V$yP>D>n_T5Rs9 zZdPqnPAb0uMGg*S*_7^`K9!a2pkMu)i*+?m2Q3uDr9Tp)7_h)^Km&^^ebdSttVVY{#Bua@Vk z8aHWD3H5v|N}Wf4CPZZV)6K>uBfK2bA^pLj7V4ost@dKzXS!g~GAiUF++D7IG04kt ze!?X_D~>bX*`C=SDa*xx8@>oHb*Ej3wERjLrj0ATTToGwS< zehb~5g3dsqOm2MiV4%rHDgnZYr8IqvnXJ!+gnoR|id_fhHe}p+Kp6&V6n4eW+1h7A zKdz}W1zafe9F#$$odPUYcz3rX2_0mhG4YqKUgxzJ7 z>)ZG5tIV<3QoTnc9pSEkOp+csML|NLM0vip zIOnjNH?POdtLi&awy$1Yj4k!MZ@l|q97@(WJIq4MN=hluM=p?u;^%&BgK5D+2IWTf zEE@XBW~ck)d062fJsF%{ZRK1g3_N7AjCj0LfXcKbKjQ!%=t9w|a|-z_(+US)rje{d zV=2oxl!m7d_&xs-&cAGUoJUHez$tArPCFfPuX+)Vypu`wqdZ>G<5m!MQbW1t>krvv)%A?Bm*Y(IJytMYvFjqnjXz?`cF^6y#cm52w#sKJ~ zUy|Q)0jVlHU+z@=!`%HcE;G2_ zP!GILkA+JC=lH_}RnjE1;#yz~ttbYt%k#)xf}xrrD~qdX(`MHb7bg2S?dd2hV3z^) zY=iqTpb9Xlrc?F9C@u0))0F6Qz#F%8EC9Z@-G)L?KQGAoh#Gp!ulXpC#R^;{z{+7k zk}a5ZbqF$yUBKa%YeEb02IUb-w|2?UZpS`;Cu-s<9_*UJZalV0oTL&|d5~$T`+2{U z7f;36$DLrD&xC#)GUXbUvT-{^G_d0Zo)IYvzxWUT`}mb#`8D4SJF3BYU&??VQ>)G> zO1$-s^1#k1FCOv8`aAEx7jv_7u`HzsJSa}I3p>8t8o+f|49wK`!TpDxaM)es945Oc z2m6QKdEwRxc9!oyco^FX8_=K)jVwbs`^S9B4la$t#Iq-*Wxpqdoq=0SWunsDqyc|G ziiMPoD+{r-u%L2A`;mqTy?}T2o$)Rw{BTkcp_!& z#`SA4Jv|+t|J+;g>T9p~eq<)I{Gypa%(Of+QNd14yfg_r*tE094H@{7lObVDu{kL! zw^*A>JhbHmPsVIp_e0@xOA?$>c6f|a1mZ{-Y4XVt{T20A`4#D89;8Yd{jzInTnc9j zgrCS?hj!rji1_EHDx!g74cl6nY)OgWgBeVWEF@BXR4*X{*Rk3G#hG*BySv*lG0rQ4 z#7TCfh|ro-S_I8bbYf|GEIL|%=G~}F(8IC0*Ne4XDV@TPyzrUjf#Hs?W#b$syggQ% zbKIh*vevhEV`X(iW%m5i9t)e{(Gk(6MUoUse>$1R;bQHf+RaLZ>O(y@*0}<(7YhrE zKC^nDbXTq~(;Zv~?l<->r`m<{?7f41xbC)XL+yXGbMm1Y6h60rq}SS>heO|d$JZE+8tzjh^3h1Sz!Q-bSilbkW-ZHIlXia_fD>O=fdB4!K=Fu4zH68nT7u)ibLpC9tTdD~ z$aMd%=JVhz=c33|u~w)FS_xeq4u1N1XphPRFVLNcxC?Dq(z`w_yrus=1+JgdQto+7 z7vYmA!^x9cUeZ^gvayEbK*)ca(vr&oHC%QkwHr^m?UXq$*9n=2gh_?|&M_g~jZzHG zf+TS|#|3`@Pml?1o{^@;f8)dhqCx6^O<887Z!#@nmH^-hSuX{;0w7(vq*r;)4d93+B;t zM~W?q1tieU8(Uj(to)qSL;*s{hKHQVPB~b)Q{ew zfg2&VxspwTG>Yel6szv&pcQsL*nn(uJH)J%bQEu1&bl%;qk0W{$MNpt)wsX9<+OZa z`1lCrWWtjqS9cMki$G_dqxOdA4DjSp(<5*py#?UR-up)mWyi?U6em%2ZwA$Oj0 z&vs{~p$y+Ct};jmKcSr>Ia3Y&9^1qx@-hd|Kbbc;%l;zaQ~3hjR_k-j6-s*w8i&~l z^h0o`&mAOPHhq0h{^~|DGP;2bLI|=!__3=rkb_DXW#F#h&gAg)XBrhpI{YOf^OZz` zwwnxiGb~Mwu7iS^p+NA~$TrTjvP`?24&m&__xah!e;Lprn|#Jt5*sI`233BG4|JLa ze?n77*(kyINrE9JlQ+l=jw=Ht0E!D}4OzdM_k0Sn*JvC4xlRJp&~gKHa>ie2Y=e`v z@I4otHWTHrmO&L9Et6$jED6ZAF$(C?|HU#h80`IsQU#|yhqWL}T0oY0&m*mM24jX- z-qd7i>jzt&cKl@~AsxnQr^?a5*G|YTD9XUk|E<3jD^H%dKQZCd2^YEIi(9!PJneIi zC>P)P;5)InxM1a$H~CS7sRLgXLuq8Ex$ZP@^8`u{in1pM{qd3-JA6w^S3K~8hcAWs zg9*yRgv_qQl9brx#W@Yi`A8BzudS`crj(}1v5}aTveg|MiYX};$Q?HUFe#%9Q{MDs zexes;eSCB{cBJg=OHtxQDIWXr_#I`I$-*!DscgPZ$ByW=t5@UEBPr+Q1+TgJ**Me$ zThBi;dDVUgdy~&he%SVTf+9^CCh6>l_G8lw=83Ca@@m|0SmJS7;~}02kusFQ*c7H8 zY?tK-5S3W@nE=59U>UO<7{0c}gvX0$@E8T-_*fIY+Mn~?{si%1Pm2&uo|;%^rlw*YNxZfIyn|g)1A07H6jIhFha*g|QVSp4!?amrC~4r9^zAOCOuV1ZI_H5rzfyo*>7JOdniF3EXXE`y9$x>iCvFQL!yU1(?nou^6N%6AU+C&bsz z^F5H@y4ak{gGc;qx&3Fv#qabv0NE)!2QNGldjC4+Ul^E@R=xjuL4Eu~ooo?!u`2ef z@M;6els*X0#h?TN3x7-expZP8e7sn-n&i?tC&~tAwio5siM8hLLoE}#G58x(|DFdW zQw!Mnq-S}jL<+|@fB*O5+wZ-nI-ObB^%XrTjGb80?H}yN(&BtfqXa4L?)?XH&qPd!>t z$x&sxg>BLIY#HEfj+xmQ?w6cZm(3iV$0)=n`X~uVkHufjE91#YrtX5}&RZ#qyfRF7 z&4kWOrmVDSr7>;tCSDXkUKBf4{rt&mP*d1X^~o)|pRqKUJId`!c;Ri>SL zc_-uM(qdehorVpidZtVK01UyBlzkTOk+`WA_?mu{7e(?Ee7E7DkTjl|a zJ7g6FF^@X9mE4Zc1JWg*%WkN)d;Z-uY^AtR3OE3OzTse}XMOCcgz1BCen!HUml^rLVm(~tXs>|88x zBp_uocSQ?-l*jVv{L1byl*d%=KvCtBJn#k_$Qj`_%4o&X>5*e7G8*;eu~dMCkDX19 z#SqaS9YwGFd})p{N|5ndsaBAcUyo;!X~kiXD}RCVR7$OQZy8WJBB2w-_E7m5W2j5^ zJ1LtK0S;N&UfIuhlEkqL%EVNeK(43_8w6bQ$JgR`im^|6_^EK7rg3)H9|qx_4#_P_ ztH%MA>2tlZ(bw5AW)WrH3J)IvVUvPu)(%~ME*K8?RF-UR*5JrmPvh+Iq3YVpg~m~F z!#Ca|dC~Bx9Fw?Rj!7UDZaVDJP`D9{;wka^aZB@`=?wLrl>$yD$o%@L`IM|3rzxR$ z(o%{2GK(9*MgaTa3ZuiNE7Pr(9Eq!D5K!-o)9<{G`t843VqQR4+2INI) zNj#J?Q_2fd9AjW&@}#1|x234Co5RaYC{sMb?u?AZ>iW7A%_UDX zT`A9NQbc+2$&ckFACx&WH5ub$BYtUbXLsMbd+g9ojE~0p=DO>`7qBMBC#;}SLykpI zZg;X9lPU`(oAb@Uq{>IQ?~DH1Qg$}C4q{AsH#T=P(H!`t7Iw3#KXssQ!v|gf{QPG> z6W6XT#~ZJ|9(-n-T|Lc=mgStlRRA((z9+{ycUYRiZQ@Y>pp>XjsP6C=rGW@f?(#!F z?bFXoJt0Te7NB=wD3*@N4Iz(i?fV7JKOR2 zu@(SwvuljfTl+0<<8$){lPfP%@nyep7CtKP+QLjsjtxnn9f|JHS=^GcyEw@?afngh zky0@>Oxd}hfMk>xlvwEC2BE{_f!NtUjossj)vf*5Jvxi611SSXT6D>Mph=le(6f8` z(ZeUPzO!o{_$y-SvrkqxV|pr|V!ZwCyRt{)iQ4*&S6+%c_wM^lG;QOT!BjUEagSHG zV@fo$KwDW|k3B66P)y((GB-TNi&RoxC5r(3u8?h>;C7&@C8)*`Uftkxd&sbWLk_Mj z&BxM}x%kTG-|$ln&I|wUJ0I{F4Ydaf4YHzElX21DL-vZyd2H9e%fUfWN=F-$FO}ew zt-6_jl4ObsU6m&Hd8HvAaqw25j>|DS>rR<2&vGHW5&&c*k0gjrr-Bsa7X9RBTmU!i zL>-jSvgJG)TM5cz4oOpvLx=7oD0}1w4+}yRK&PiI<*y1mbfE-NK7A8fPzF(~Gad9u z-qiMW;YdB89cA8Qe5ns*@gySaEtw^s(m)SND21>SP2Yr{29>95#%V7&SWqavn?-e~ z`UuJ~$1coU`T=R2z8rHGAvsb{4vycJNdl;R)#CocrwFSS>VK0 zvYH*-49j?l${%vml?6(WX;Oe1S_D-{f$ApOxSYwk8sWo0cRQccI4NQ2<|lEKbPg=G zc~Ed;HPP~8;c`U3jI3(%4rrBr0gTOTI>(@N6gvS~O`wgy=uE|sJ?mN9$yUMTj!vJ# zomLaaB+-VZ9K~0K+I+-_rc@~V;69{eH6Ll}L<&7^x1rEiPz!2-=OM|5$2q{cUWI02 ziBD7&5q!xQw@h(62Dg4{oZXqZ1qu_X&f_S%3{IMn0%0gT@Zjrov|*;FUWubWK`4>k z@QFHb{go#lr*sAbb*+Fa>8x%C+mX&1P`X(uEUlDH6i8^}#kt4#@5ayl_5XJ~dHmQr zf|D~dR#56753kt`wW6c8x_$4V&yV@dd?pcf!<$j{d1rte7QrMVo0MXbt1F4)1Pyy-4q!pkp1(T|4Cc>JhIf~rWNV(Et>me)*4 z31-j-lTZz!?2Hb_lxX3j2Pnq03+czk$6{@JCm!&L+=m;nBPEN^L8HX)?Wt@j=Ij_l zTON-X@u!{nxWmHSd|bPBEpFVr5u3X^el*Dh#>8t`5C!fF>WjIXC!S+W8MJA#ReJjr zUR#YYCOeNAf;y>#&tb-4uaMBTDf?A6h_#DGZy?=XSb>6{0#(*AD8roZ}i{A9&#xV5vRs-1WtgJMiuPl-pgDY%z8IwED5mlpYW#*L|wxUn!E-Qm-i8y|^< zNln0fQg>j)RPw~x&w*;U(Q2GfoASC=2#88r6)%NJ{NvEG-3sKeOqlL#gKR1 zcNN!Dx<^~PYKsH4;ht|lLBUzu=9AO;c=>~mKla&SIPmf-uX}Ew%l~=upg0 zkGoI2e04rvy>&fi=N9A9=6-zZqkG~-;!6dUozzsi>R6nLvK&f-;$y%~LHai>{7Y`2 z*>hT=;@)!*H%qOyoot-9 z6C^K$6{mvy#7c=!d^tCa+b_fFuPUavsR7{nK_IMB+R8)v9SU@Ue2hVoXOwgBF;+6R zn#0Nmo}43jI;VO|K91Gzl3Xp)wJ2l!_1E*1MnS@#ZIp{@XzhRyX|m+_+D(@}=M@yoU=(`^?4H#uUSsmdQKLKRA? zR(y>kQ{Rb4O>HNPt&tWjHGGB%Y92L06IQ?t-3qfH6@S~dT*`}^KZLu5DHshtCVD-NI}=V~K$n!RFT2QK|JXyhhJrj}lDwjp;Oz)fB-kiLME zW*9%X7)(6m3l}vpRfQ(wZ``(=$F{BF_D|cLHu-D){0py&zmYrn8KC)euLo*Ps{fy# z{n_~1+i!a?gudDNIlH-{YG!)cJ2E_)<7;7e?mdV{QjlJfA~-xe>c?;t#^-qX#yva3 z%S%^$w1N*ZaLp0tZgKOn8D}%$7dKQ4j10#0<>gpiU3WWjjg`;hO0n&BMm1Rs#G%sB z{+%%mPl#+wHm(;wMV`&G+7O zo#=xo1(r+Uf*!)pwLxhHCYnk!HLx-$oA&nP2M*cpv-)?O@nwIt?0C{+qNbnZxa9s& zc8xvR2}+TwH;?3b|Yy>oHU<5@3-?*RAV*(2Q$2UJSfTxh-#!|Qo&8?CcW*6rj?UuY+Gg~W z&M#DscGNyw@sSp9@aFBikIkPS`P^%9s`%ghI!A&crpJe+6szq-#eo(@{yeo5YUmgm z9gDNFE62-jJaMp+MMeSPZD{&gfp+}GxKcgKEA`{$&~foH7bHXR#>+S2OP_nwJHx;B z)o&>cyVKlx3m)-8{fQe*P}sfsB!v92a7A(U0S{(xjKtAWMo%;l{6|VtD60QO%=~i0%Wkd=_}p^ zRs7gUN3@C0%7@~~*ikvP zCl${!mL&0pc07|sH$P<3*K{+|5s!t17vuiB4h|~BKmYZzI`*D9~7@Yy!!pBo8 z+JHLoPKW7NKGL$o&7zLQo5vgaw!(*G;0-v~QRhhsczFVNrvArW$yrXe544(9rbia- zXF64H1t3dS%0)Bfk!Ef6#lArIsVT!Qv=r@MHF$dI0tFOcADk&bTtl+$8i!roPNhGD z0OptU0yl*5m9#a=ovY?gppn{PD06M7aSXFK@p}LkE`VRCKzJ3CVZ@QHtqfZn|8q}l>E$+DDpM9ka z+&P3?4rFyw>+x6q@?Z8NZ$6H}7uqyy_~o>Y6iJRe z?Ck9&-$nbkzw<%N&dyj#h)`;g|nI{N+__r4CR-w#OM9J9q#}$pr2`?91{bkx}(vNpR=NT!Wq0Uuhz|P zRLh3*^lf&7N2RngkfUJGSD9?M9fk>PYws}LzrPyq+`b=csy{o5O!D{dJ&G+&?!5fP zZt?W&Z2a(-e<0L{dcpi0I9!v{#A-_4!S&?hnqWri) zGoAaf+b*MQBixE4pYviPV+wVYpzb!gOr^zkQpuh}81oe7BMAEQhXp{8KTebGJO{9)tZ#OG}}Bl%=~TM8a`5+BFojcvaKw=ZS#MELkT?e<r zeaGVUo6GS)@~COvyTpg4i-iNquohA%$0~@$7^5*~hv6|S$9@*{+q>Tnf$GPmztAS! zZ9JmI1Xsk}x;E$Y(A+5TN?vf0C* z?*mtO-b#>VK|!PZKnK?d$i>1zRaMOh^sm#AhYCWg z#3c^0=D3N!DPP7%6o@}3*P@a1cDKe?C7@(swtN#oK4gc!4lli!0+S0M0kU)A>%iIy zEyu732kC*vAMm8o-DLAu=Qz5>3%raK&|!UndO4lO66zybp^aQtWD9M?XK?bo_C@nV z#yi#myd`_)(;rGS>39*Dr*Mu3Kl#NZ>VT(R?2~X zWB~ed9Lgsi0C>P}UMUmp2z|f#8^7)N7X_cXGF)vO=fVu7F$&WG;u?}|*SKt;1gFwp zRbJ^d>B6p!X(*`FA|<>JjZhV^K(vz(U+n5mVI;5s;o#0#yKPrQ&g=iit&*~86&f7G zkmg*Z{v}lAQwgcK3uMk5e1Iy%03qqiU;Oa*X=S)C{1p$7Bo}-NX9#k%BB&~ZP)4Qj zo3>VdFn=N-TX6!V#3ry+rRM-0EorHK;JYcqKwaQ4dF7zXaH-!?ZsNq`Aku)G=Oo&B zWj7U1evzW)yJ&l$u`9vpMWbm`oO~HxezGnZhSkth@!DD4HLmZzTp@-h`jfMtx0vFm zZEcF1aQe)9-~3wq?Z5SRVpobMO6jc|*J5i&it+kJEH5qi3@y01*Y#s5jQFoDU-3&t z-mOzb=VoU7xO8i0+e**&_MUA9T|V85V$B)kNhzisDc5^aG7t9lW9iCOe++>M9j1?s z4*A8QA(6blvlBzA?94>RA66L9Ht1-wBNkY!K$sw;Ql zPIkBzLycf2_AA!6dqH{S5q z_P+YnuUQd(`R2{|(LeZOF*Y>;i>)*<#uSeGH$iR*W1VP7LlE@2CyVG~DMYkRwhe(f zE;cern10-z`bWMvw!mcOnh>uHerLcFS5BN165%J4LYT0qF8Kck@ND+rFdz1G3Lhx;^x#yyn1yeX1h{+hmL*zc4$C~?h!XR zWHiUsAL921iLn>&FyVnq9}ljU8tqL3ab9XPnQ%1WB}Os}KvCmfehl^$*E z#@0T&iJT=qj*ZP-U%SVdRFv(zQhZl-MaxJhwx!IjZtlbf57y%4D|2yWVJ_A-Hr@VP zQm8p!&WBbwV~(tF&JwQG9Y*+_T!nwM6MViL;Bk#__eJVMU3eP8{j>1)Pz%e+iBYw~ zam>#x#kHlmc;l7pF+MRJzwn#iif_I5zLho=?PDCZ-Jy|l-ZyUDvV4NSlH|6>CV(=P z-(!%xzC6$T%sC43DvjhMkCjxl3)x+s06C`q{O6dPHnLacwU(z}L&~K5bWqCCAM;q9 zuaHaEgLqy55fm=6qwH}E zqSg)icoIUn?yrKBH@W+S3Gp$HhqbXpMK{(M}ypSl4MP$!^V}} z_!B}4Wt+~Ys`E_9!2^?>UTBKjgLg^g0i(DXOQ4(jI85moK*(3?NO&Effy#plZGYpI zozC=^F{_EkZo5lj149Tt2a9e(DmYwWZnK)sY5Gv*z-8D~pj8J%>_RL74s*GV$`lnJ z{D31i?kgJTq)`5ofA|mMSAO+ZeV6d`)KpxR0{oqiZkw)!xw%+bTle);yHXB5_~7Gs z<)xe6C8N)>1IMFO6kV?768C)_88a<6Lh!||btxIVY&1SLCWUz%jWKqmY){F}&K~OJ@#RnR=ZG6&qTYDiVK|l?KZycYn z+^FhoevLBdGf`=>@PwiGstlvdZyuh;$B#GS-jfyYBJTD?Q->QZCZw$ISYcl;(_2+{(@tXVT3+k=HV18i0OLCsJ;-V8@+>!_nb>+)Se6k^e~ zwH-ILF#YW7H++-HuYc|1_>JHGs?uefR(x&Tpa0UA)HYo&Sp5R&g>(Sk925F@Qce;Y zIyFDL4(+j1^=0s}Oglhn(W;%UMkpZG$jNhR3)pl@VZ(7;m)5)hU=14nKS z(0>@?v#bINPug=qQ5A0L?I#NmqcFFY{>^vrvj9~d(!!^=zws^Se^KzMD#I?g6kn-m zlanbh$!0g2UH$YmE+Hr2%p;=)OD~O{&a@E`IcioHk$xJw0*%!6*VUWD9M?~80pX8^ z#$UoLj!s+h5M#fS+3myvly}OsyQ!am^un7l*tI_3!6&-t6a|mVYJt-jUqxClk*?O= zfWKWZ(dnEj5U(P3YQ#>;T<2ArqIe87WtS6vm3+@MTRrd3n= zh1~D}<(6_$z;=3Pv2}P9E1TPK|H*nh+}!a_Bf8p@76M$SrEET{y1cj)*Kgd2d;To( zT72d+Z^kRHy>5Iw(i9n`9aMhdf9+%WTf+ELc%2+ckMc;IjY9Y#9&{zdbz(v^^_IcJ zQ_CQ2+L<0VQBJ8gLg_5NGVx{nsdzr8g-Az>6+UKyLPcM@|M-zVpFOR`4QD+4@}cU< zozkCKnva(k$K%cAftZ`rI%k9lQ;QGDR(|f6F@&;`7MH_aQ3c#SY|RVI3n{ZGFSHSl zhRH_#Rb4-YpwCgQ6V60CmlTt;LGR-39vu2u$F|zYpVdZbfj_5Q6?G6jwJ*mcR;4^^ z?GwAgx2>?l(_yvqkiT^I{-aIr4zfUlp1NCyTqDl_Z3}NvegM-ge+{o;`ga>&3_kri(1D&&=^<|}W?Vnv%Q{dNvX|w!Yy~$$OUay4bQOYj7yt{Cq1l`%@HLOZ+)^2rkb;?MFvKA+eVi5^ z16T_VK`kXGjRXb{8n^+wGtA-GjzBQ7qtyIPX~9c7!%KSSPYUfY$VS)%Bt6$Eq*ov* z-3ctnO4dO{I+Gx2x(u3_rVQEXeF_|{h;!ob97tAVV(6hh6>nP^_E>cZP(OI%dI?XZ zuMXru<#$C!N_+A@f&B8Y8+8$k+c@B>twJXQmXj#HUtUR)11x#%%Q{qA2%}y4{)Adi z1I{m;4zde}aAg|Ph67#|Ss(>;!p|JwB0hNM?f4u2?cea=I6pV*7bdv^jTgw~HL-Hd z8p_|jhrD=n8uOZ{d6{ZTiV$av*&*|nnS_fkesOfdXX}I?Wr1Dd$Lkw@9Lon9>f`6J zu@Qd?cWivzI7X$|b`^HGyAzKeJ&D(@F2;lVx8u$0i*a*l#^o(9ElW|D^&?{xalY2Y z`QpJrDUeoFPGe7sJiEv~zTsU&6swasI-xg8)rnf`#4^i@5bkDNg}(_Y9J)>N2zT0(z0s*bof+oDeQvny@0s=v1;kOkhMqh?2yV55Bqbg~^)wj+A!eH(`P1ogNpA0OJ zB|id5;Rw(71DEDfV2bDSpiO@t3$y`N%mLGykSkyZHR+OU+?Sh*xg>WefyLWE4gcg& zF=WU5Bv7l=fHU^-80_sQto3{fo)LT*ko2;uo#V;#ppr(ynpljl&c=QQt+}|kJPsfx zWir8T3=^aV+xOo6He8MgDW!-21K^sJAU?aiv$Gdh78bk{>|MjTX{E^lp09vG66YP6 z%>9PF21*KI!XpeNH_Oe(z$n7x1;^oGFZ@0Jf&AQOhd1^$nLK*@#OEZB5BB4&o7dtq zFJFx-SLR}IVKx@#XXDnj<+ye0n#|R>DusD^x)Z}((RC)};OLNJz1}Tm*LVa)S*0_P z@zptQ3(%M^YR;q>X+YQ5&j%cs1fZ7}@;aK3<|fDep$Lv-+`PIJd&dK@DJA>f-P=Ac z!GycT2N!?^D@Inu1 zxW$0oJ=@|DUx8y+GI!dUaHZhWYGqQRpQ4a>w@@}p5{fcAs~l&cEOt7H^CMMpAeA5{ zOtqP0fzNPr4w#*M&Q&rovzrfXYBbm737=r@a=XS*jHzh&bA5R-X6Ncrl@Sk4oTwZh zWpd>aEwbij*AH){OP1tL1%SZ3+OHlvVB$t`Md2gP%8U^3vLkQEv$IV;@EuB7<(x4e z7dV!3b0j66{c;pwR1MY8FJV$3FY57yL3%>Y@4}-r^HXC|ocB!68>-6>zNQ6_gkn;{ z@!H~K+?eUcHOajdhyo}fsi>uNLrWU!qfk0Vff`2sHV)F%on>sEiqO7$hsRdr_x849 ztTW#2_Bs!;I-$V175c}QJ~?ZquF;iY(NXvDBFRo!)=%D7=A$^&Q8Qt5%_O1BkS zc<55nFK4EqaO%OXF-IE;E=4-M!Ss+0+VJv2MB>XI7B{J60pNjF_<=I!vPf_I(2(g$ z2?mrjaJW4Ay@(6d;vQ+*Mo65Lig-9NK@3~?ZLyz;sK%BDZjAKW+HKjo)C zm%dBi#?NAoO#$+On?6iC(zi?_bP2pbfVM&_`00&=*dq;$lk;YJF02Cv+SUMTc? z&kF?e(T#*~D4=v=2A8rGm4K52d1?R>=8jhJD%iI!(4aOXuYb5xXcO>xOg`6H(o}F{ zI!kaLZyFB-wneWAPaxcMDHXy}f%-WIaY{f8!wv%~Hsu*Nwul2aG`3J1APWwqa7$2( zvOiFHrOH5aPI8UeM%_<_6}?iMzs1<9c`_eYS2+zts5}&jHkf~+!SOQWVp3yQra|jw zEPkgaXS$)8FX_M!pD#QI_(iY6F-e*Jl+WeLZC;VI1RD27i*!-f{>_|F4A)H`Aq4w92Dh!I7qhmuw{-U*^jfACwZ3 z6ES7WWNXNf-jT^G|LUJ=z*EP$fHavB$X^3-1k&VaHTOy4+@jMGj` z_#73m!X`Q5OfY^7aVTkg8X0A0a%x%x=8gwv)gG^|c}L>2FJFsMiSE@PHYBA$6OR-H6l(ZQPpIvbifKN|R?Wo&)k$=ELX|hkTz9u~>uv=UeHyDR|d^}`3Ca0(USh3e*H~Jt> zXh88YZ>qopUg(grCpXGE!G#RBnI{&vHOh(>5GXACDFOW(-lPJShBy~PdKr`vKHbZ1 zJ4&o^s}8(CGp6?8ZTNB7UL;8A9UR}#*Nds|KPs+ z40YuCv=457=!M3$*@^hl%Zu@aYZEauqPAA{Gf=R?P}x1k{?W_)OQFHH~bt=;@8B1b%E_k(?ICpqvUke5yehR$MNvi_OZZ~BE& zRC)Tgjd;WSVP#w8c`g%rKT(i?S+tx^h0zAGt0HDPl%Kh=p{T^6EPE^yKV6TKf?d_z zjRda%9I69~6(kU7g-m#rRbj{iG}TflGt)InRdvfI$_f`a-G-SD{*VTFH1AQ^isFil zu$Z=oe;F&tZu`O$+FvwUQ3kI}DyI;TN^mYXtHPWLaNBVAxfNk4hbX)#lss|S(}Ie- z)EPgWRthoYqkK~%=ta3S4bUl?u&I}C|B$hQD}RnrN&+uSJNsT^(W8|4WQ+QVUh2l! z03NiKGs!vp;wn2Uy0)cYiynqJWQ3_;CXQcNb>AtlzHAfojyHK*--ci$_1Y023@QVHMXW}@HdX{0%ZGS zq(Wpv z*;V9BCLdj(Z8*2f?GyQ8#cBBZArv;@aoH+}ftK-uYom}a?-rYJv@g4L$KI7@=aehE zzVgM-#s?qYjfEL0zjHJGTr#`QbMrH?xV#h-lau~L@<~xhxHpq)(9Wdz4alZ$(1KjB z+Y3Oe2XE?)qNMOtR8tXQ612jnF#?>%D>_j+uFOySXvne@v`r~ao+QtX6+h&!;i0hb zVjv%RV3EKj`0RkQnBYaARP^CF_}KxKB07}D3i1Fuswjady_)Jo19YH#noej^T~J79 zLmx|#;e|uWK|w_c#*ZRQ7`yP*aQuXJVRs*!F9dpVqSZ$3oU2vLgA?y;;f6EZXUfZu zD2gaV<74@9ARjcDlXA8sWq5X~<2HsyKCcb^=AOckCsY%rZxmh(U+ZJPE9;SM;QmW` zHBatr(_ZjR{7^kBu3_=@^pu+@IDRpzIO;22s4lc2{K3!P;TxTvg^$aCwxWk^`)tJWnsb_0(@4yDvy*^?FuV~`0NVbH;n~-s~9fUvKsj!xM>d#q9Hp1&Bi^Cn5c`BYNi`h2q`e0L^RB){X z3mDV9!|C{>gr_Sx)}?ihW-;lalbIw zc>HlXlD8=(W?KdPg;BU1Uj{;%+7TSy&y>-C!ZJ@kJ>cJ(MgR^)Ha^@oAl*QZPR-}zL%keI&^3a!4$Jb~1rK~k)=Qh-$$Pcqp31W=GrT;LyW?jc4&vf?5|f+G?r z+w4Mb?j6Rl^5wl$ zn9O(MP_}hd|@+a9uA8Fw=KDjNw z#t*v-oidiEmYwpHvQ*^@gF9s`^JaQyaoRdRB>f6>nVBHxo|ZfdYC5;C@{!i*1y$+l zLzRyV%Y3Cya=nFMe%Mgq`eg_w95HV*1X zLy|A-#6nX(=`wxl!yF&7M?m>mCdzJd1FqnrKcXB%s&|BCvum6{ilpDEoj2*O?UePB zqRnDh;~(Mp{q)7*>U+xPu|eq>rzwQ~i5yqOm(;$wL-|PqT`0+<$?-yTicgS`^~tlv zD8=CZi+}My6owZE)bfQxK6)?*(#Y*5qHSy!KZ)JM=Y>B(VWfMWDJkVlySS_WG9Z_u zotU)XBA~=MTtColCjr#K{~1>^2wF}<<1fiFe+#d%VERBXU5ZIQqcgsp%JJzu2T3Kr z={gT!AXgug(aX3jK(<$E>a2w`PXU8?9mxBTD|AfFc`?al6g>^*dcjc5_Xm_uwJ@d0 z41_<+=-=s?6#ZDRi;>&LB9abiwOh_e14Q%(~36?j0|wQ+uQN`zx%s!AjOVz!5to>^YM(G-PqpR z_GSJP-L90>ZhZLBN5UaxP-##$xy^xrk{u5}hSVg)ZVB?W$1M;dhPv@+ob!zg&?pv6 z6bzzPgf$T`pmPNjHxh7uO)=5i*@~0H{rH@e;+u;zzG7{0c`>fux)Ea&lU8VWygw*f zMkgm@R0=BJ<45@$Ry_xWw@zv(+wA1v_oGkIdMsM={2vpcMlI1mU9u^16oX#<`?2Kc z;91N{Ntx@8#JH5c8PUp4HIGC0c6Z|T$G3fdGM{ID{P=O~?(7Ih6^NYFu~Gcsy_Hhh zglO>r9(XsGg@P0vZtSQ#$37aNg$Fw^D4SMzWk~OxVexWQ${_i8fpLi4X7b8-*AyJG zIm%+=o#@=TAC^DUX);xQ4Ed=yZS7rQ)e#zbasYmKH#;>Quiv;DuUwgl<+-sK8$uBf z|0HaxjjF0iSLSp8Wfi)|RoMmj1-ewnkuQ5lX&pe$$Dz;K3-jE!629aQ~> zdArDi>F{jlYc(n{*|t`ujA2mpu3rzx#yCq@N&5 zb=Og)I$!1?Gm0DarLNg;vkxP)gt;8oM#))l$ZvMbP9;)d3Q*F-Njky+dBKMQTa{w! z$SpDeim&n8PCsKmHe(HSr7fLK=}Ak!_R|m1?fSXR^6Q0_o($*73VmMuD2sS_14Ssj zegY&qYQL|GS$50W1qLXOv4JtdFEN+=D7Te{Vej~6UjP@Ms;9jT3S-X5BTa>F{Lmyf zmGXN5@Kt!(O@7f(D1Fh;@3a-FKW!M@hYV5}ra<3GYlXStiA1^c6^C+h}?Z&E6q9gB~b9ujZ0TVE3h@wK5pe-yco|u2VbB1-MA=1skYUNW2WD?bypeNR_6aDiPSvPl-r8#UsFP zQ~V832N^zqz$DAA)SVAL^gX^CGRO=A9J^4vQt;Rn=8-tYw@R$9d*WuX<;T;cm9osE zNCs^t03N$@jt-!Fu8YD37gunxqdcats+^)cL+g~3;j@!tE0KKcVRvUcUcR;%bJLTa z0J@5wnwhrp!!9G|lbKBX8D0(8u%Fup5CWu@{Tiehg%<)S1Nul;_>o`128vj+CVd4l$qge2#YY$(km#ylLRc%9_*hp^&w;RqAONtU#b&cPDm(z zqqYa7+zrWKqxi8{V6v)8FX`(`NsyJ}0l4+AfK&nq07~2AFJZP+--Il{}$asv& z^gJpWaXI3xvqM~h7qc`GEXcC#X= zw#1ETE2k;fsld_R{zD-mt(vULJxWo^RN}zYxQtJHL?qsV+iq+bWSnrzI$@NNI1^Wi zSAg5cI7mcAKusL9W_&FVnng7KO&-EhkWjGhW-2 zmGbKPDr+61UNl_f-lqbEPkn&Hv>Xl^l0PvnnFu5k-N?n)2K>Zb1`Ueu3BI>Xl;Irh z!rCY&G?5Q-xOkb{@p{X~=Y!8=%@b@THIf=6rD{WQhJqha%8GKM&m? z+pUi;SG*~1&uE%Upd_sHG<1Dp1m}R<)%4+f7kEl;sLCaO73?EV6O{gE)$x3IddxB7 z=;y0A6BmK$HxSnM^z)=u@>Eat6Ho&Mgcs)p&v|$8H03pjDjI{Qm_VVN1VVp2Tf?p=;; zsQR(7*NZ1oh`EJ;bQ@b+F(aDSHrK%-(%=gUM;eM{g3C!oM(*KZCOv4O@1dlzW6h+? z8NcH+e&I&pHd5$h*9p;QE2|&sl^)yWswiQU`^30izKnNyRCeD^{zDIw8)uKfM;K?p zS5`Nq6mxul^Wk}yGvDWmmGvt}O~J{l(2I@S;P8$+}V8B@iCll$Iw7 zT$^_+oAyKwrn~v!mPyr%n<(HSJN&fyh?uo=a2nU9QF>2e3jJH6fl`E|jKRhq_LETG^x0kY zNy_jUVx`NQA-GW?6=f^4qLDF;v)j;B7ct=V*>q?W?vsWkm) zM0lYUe3)9EcW;%~{J>T}GvCQ1a60;j{G?@U=UA7=bot>2yTvSWu{}1ZOn9KIrh~9L zhFCs$^OP|IoZ=}@o@o0@6F`2_zGxUw<*N+7F!43Fdd=fDytVJd#?x;m80>4AI%H8^SpB{}a^2TTw7v)*kWu9{B<&k~u2en4)L)s)am z^tqh!w6i<~&X75>UFJ%9xP0m_$oLrdix zxP)}O3W{ZN+F4(ZzwxvGZG80cZ7ajmvoo=Lb;%#PxPAAozbtnB+Evs1?swkz9j6>; zfMs1Nrp(mq8ykMx80CP$n1L5%7^R5%Fp#pN44)V0=S0AnCh>gSzP`ET7n1no^O)$_ z-Pwq}t?hVyX*&McS3Vb`Qph+rIX68K*I&66-RT)AQJkTaBBhCmbH>3W=Ye0mmiusj zR})oUtG36N;=B`h>SG$*VVh@$a|Z>bKb4vUw}o(_h;W>O?#Fpkc2jmSniLhaX9{3u zbo-hdMBon_CWDVuFzf~z}e3Bs+U_~B12i*c==6-mf60Q?||w6kjGTi52} z;qANe+rRzQ*y1zVE1PQHk@&G6`${a$OvaCW{#JbD)w$>lG60gmM$11!W)Y!4538ZqE= z@4U#kvAOAEK0UQLyh1^IvZ*-oD;?px2Pz-sManyEvwPT6JPVNI0*c4l_Fimi!NZs5 z_`LX(`UKa?9iHXU2GVT^51)vxd~`bt%==7aX!~ng0A&tkyq>cs|8!Ri1HrWVz^3ru zy0#FvZrzO6Uwbu%J0tPV`*(cf%Fgy~OwY`?-p5BLQsAYnONp7ApVcB#ZUOb7Jme#r zAbDvD6rS=aBhpC6u^@vioEXT?xf2gKktsR(Nmu#BCzt+_d7oWH5EX;p`AEc~w#h>P z_~lkXltnv34`2nb#9M&|Cw~1=H`mYU5@+H_gK9?}8B81iGLXqL4xDw< zz~M5aaT^FBebqQ*{_g+xACd8e!Kb1OfB4m3D2+nnCZy=eb^rp{g5qgNb<$+)MSzb^ zfZaq_2Myw1xic&qF9D}Ss|oAd=_W^ZBEtJ|5<$D|QFvY90Cl1;jtk+&SR$R%`2`I0 zhT<1gj?}JBo)r8{G4t{FQivJ)~{Xj!`%Rl>M2(Y0uXn<2-;FM~2~GAFlz$r>vzHMblu0WWPH&igkG49-d5DQX-XJ=+( zduP)Mey?{DE00#<%JP+1y0Q=_hf*{@{El~B|L9jf8((|^0!M+Q1}HaEPh!zjSf4HV=3-2?Yo-&i4qi`ytFFG1|H^Xv+# z4&XdKK8}N4e$k8<$abW3t!y2{_TFK9_-G@3|GfvXCS`@k@7#ktH#;5IuP^&k&i~VY z@869-{^$Sgf-lQpT)|CP?RW0ca;LIVCOetZ)ve4DZ)Kz`zESGEha-AXnpsSc-;*+U z^-r}_3-87!WWNIVWrtmiv|k1L<-goLX6Jo8b~o?GSAX?a<9C1kH{;_EKaM*eKZr>w zgFo^kUyfT+_<#Jf*W)wG9pPuk3|iPhlJcv?%GnUNOO)MRYN&FoKA9HvJ@fa(lXTuO zO}R>j;!B;7Q?(Cw1D_uEG_F7=eB-D_whPrfDT**h2Dl=M;{>$rBk>2bbxcdEr!52=ps*l1jxn~Y7-y*N7=*REZQ zFaPil#b>_wR{VGW;Fsh5J9lDKv_r2yxvT{Tw{*-eFT`K`U;PI$H9w>ABX3hcE^^E5Aj%;WrtQi+KDzLJ#)14>EDW&;*oWRVfyK?`UK@bo5ys8LlAAj zZ7-fPg|v|8GOZv}w!@2Gc^L;!)s7ol^gg`vDE|7-{#>C|G^me_jj+g9NusGcJsSW1 zfAH_c^_$Dl8%f?-pK#ey6e$e4Lw!Ov`Mm>69o)wxXQC0hNCy27guYSxut5ooM!*d& z>S{`*9|3ufR2GZ&&BGnP*vhBU8S|)@6kOwcQ475--wwIXpiRxZ6rGeg+Fu2qF5fw82}HVIbowZtBva zALFbHq_}H3K`E=rskj{c?aN>Jb3d*CY2>XmRsPps+xlsjL+E+DRmMwQOH@slFK7(K z-~YLvi(mQmf9l&EIMa*rJEMvI@yesv+S-ogr6pfu#noj`*4BMTEuVPxgsJkkx3{EZ z@JZf5PhOmtMfpGx=9mPN%li6;@U6sEDZNvZlTw~HMb$uTZSMFe#W825D$-dxr#eA`GVPTJlWc} zGJJn^D}L+UJF(R}jVGIXF+bgr690zyI2G$!B>c%g_0#e1|409kWeVjr`OZsGoMX=Z zYk5%aQ1qzI!7--#;=SK@P@eakO4{((2H#4Gbm=vrJc zLrZ~9Wq3S>WTd93eNaEM;y zhPJGI)NL$i_t^?Hj&@#}9`_LrZnz*RXY7|}JHE+cduvll*Uk8If9_w8*T3+&`1zmz z2XXuULoF5t^13g{d}nwp21keEjnBOv|A+tm|1pN-J~_>6!|Ib7ipz03Wsac?{_wR$ zjv!HQZfJoXc8Ez2zd7ekoN}XxI6sDSxfPzOpSpQrI;XjMB&WbDN5G)fea& zLD~58IFtd}9aHG)6Oi6QLs|-G!EBea?Jo{$Yrb%()oyG#Fqgx{2yu@87GoLWU#6$>86Varhs;0eWk36h1WL;#h1mEl#j z16wI-yhBo`FM}$x75WonAKZuA#9cJsCnQb%{Fjw&9y|m3MJMmtP)W0q-z#z=REF&f zz+jaF&C@h|QmBJre^4wSl(p=BoABGFBT?7Y-zKcE7IQz7dtXW{CFVhrA7inu* z4lp!vl`I09_{3k`eQ3LZ%*4d8C*||w+qYE^J6Tc|*@Y`RFr@OW3>sLwV;w0EJ>Mft z#PrmZ6|TBVL^vdI4%d%|B^2DR%L{ir!sf-i?s(S|I+Gg;P(I%7NjaYIxlCz5KGHDR z8I8}pv>c0bQ~q)p=auKLUWwuH3E@^UnQXtL5!z9l;8xweVgmHh2oz3-+ftl<)S*y< zERRG`*J2}qbzxMnJ0E}_y5Sq=p1E>tqLXH7vg@yu%}tKQh7=~0RCa%^EG@+9%Bpv_ zc!aJhOCjf&1`4>!I%3CD_6hco%1!_P|MW>jK~zShk>*J5lGb59Vo^=0g8vX7=t2?vTb~d8y_Ew2U{G| z==m#rDL=%V2k_Unt5C;;8pj`sy7X81U1bCH>m6zVCq;)Z9gcT9F}E;J6T5x+U=2#U zk4LC}?9B{{7Rn$GyK&k7qE`R?=OsT&c=&kt5y=s0m&K$f#o9)$PNqaG>7(|LZ|$@T5N6~ z^t32BihY&8qdIIW&1R1;GVW+B(ztTGZ~F)%_rChfuIf_FPODbX1FLp|3ErW&8Lm0ZOFs&Y{SGq^-npdR2cl7i0> zSq8vF*X+{DGS)I;*7TQ|X9Bn?K0$j2xSlhrfc|Es;Q|o;yn@&v>645?nWh5eP;Nh* zg3_ue+v)#BfU(yiRL7*ZlAhAA^uG zl3!9Pj|G(h1vJBtl#UlIhCH#NtZ)RwAD>V-Wploh9Xra%FexyaIF1i?Qia6~g+0mrIBXRGFQ!gcftw~Q-+KRF0NIp`3)o+w* z$D}nV%!;NpP2{!ow!*Ip*@qRVw7zCdyx>-eaZQHYCcu-=>F#Ju$!u!5 z{2JHNC|rt#CrA_^O_(UMEdH%XLac1+e(ZeN2fY}OeJsOn-2V24`HK?zgJ1b!D+C8p z6gdBjY)nZhV;7I}iPXnRuW-OCu4J+TAtjP03!|Nmca8H(v|-ULRf!#ADR^hx95Fm9 zB~E-)oA?SiwXr|94871@3^Ze-P%(~kjKVv_(Cju$86c@gPSJ&Y;?HZ=ihLw0%Lx4t zd7bKxMOTZP4lfF!EX(l7d0etG-ch_1`1$$yn3|gQ`CjTeH!~eGs@lVK$+#9oEbQPR zJGwcR_o1mn0piaAhgRwMgXUEfisez z7BKNqY-~Pp(^FZ*?TfzsqRj1!#|JXfuhUj}J(YMEQr<9UQs6{W`mxj98;}{qsIEhj ziosLy9=dq}joslrjT^h>g?T0eCnahA8*o~VRHUg?o~b;9^Nhg*nUV2v4aNhNH`wIM z^qGgFIpU`llZJ zbpb zbWJddN16Aom+Dy`&(`*>eHYtD5X?k&6!VR4c0j2MU&7-hC(aOV?H$F={!!e#x+uJw zOr>YDL(H!D@xB%kl5BR&I0nG^+CeEN)=5PRws+IqHu<<#Dx!GFFb-sYQz^o2BfnbPXTwqKo~*9K=YIIh z@#p`Y|8acb^KYpQMpXB4w;K~D$}m$Q%CGNGrcFl2V_>u!XCvJ>9UhMp6yec{h_MMN z#2uMberZE)l3>@9wxunt3{WT9Rc)F(|J0Rs5zY7&k4@j8f!tP7Hdt8jRW_aq(1yNY zLx1p&ehl;;)ug9WqON&GL zFHZojFE0ci$auVqay1~{j7qj8Cn|$9ZGg^aK`!vpOiB5!W5n6nkdGKmcchqiC$$Ko z6~!^NFZU$V@BC<7eFk+1nWkU*Si_*g1~}s{dr(@tY_*U-WHStp$t#+qy2qHx8drS? zAMo)M)kpjr)Pih?^y2-9(();6>=9MPeqqOqNQS^8HL@7-OSED}Ma!t#88d=WowPhw znLN%PlV37(sx%6dz=&qvJz$q!3og+;L>&~rul79@eJJ8R@reU7qE7KF9x$1AK=?5# zU4GgDer9{ru~w@w$q05DdYx*aJfm@Wn=J=3p0>8mgUpU=E(N&-Jd1p7d z<`~KlkUno~L1{;dNyE58pJsgTkMYQ4ZTBJSLLD%2tJJKQ)1|zS55A9Tz!}rHG(IAJ zs_l8wdZ;n+K<(J`0w}M3MD}a_ZRCd{o}kon0W0z1gu{rpQJs)q8PaKpOF_*ClF)CV`g#8Dn$#FfDZ9Q_g!YASce2LE}ed}|wV zsKkDMa8VKN`}=j|4?Cev8?>f77RySH^2^Q~U#F`d(g4Um zIWb{n5XF&~u4q4SqYyKh@WK&ijQQN{aql4h%#VI4UcI&)1G@)t_3G7_xPCLnqnm$k6s>qO zfr}>a9BX1b!klOUa>nfo91T!i6(%)8HOvX#lVI^n{cyaErUUkuIOlny$()gcCssDZ ziDZr%9GzsG6<(*wZj6e}i$|O%9ua+-dNl#dV9T7IjmD4`BIA>jD(5toZ(NNZ`r#jp znVD(D=A&Ug&rH3z^~3VToxxIWPNjsMjY#<&os84YRGi4>SOnLr@p2tsp!4p%WEp-i znbHLg4wAsGlUt1896{nH-5R1J6;AO|A=fp zmcdRihNCWXQ{!IPa7@PvdBKltv&+UVEX~ZhamHU3CF`5pl4Z^+AI9vY76(c&F2WBE zQI-!(PmZC~p^m4|2W0K&gnlFiPxa)AH6G)$Lwxnx^_ZBLis{)!^{)fJ9LN{PxG@6p z<#XBYSMZXSmJ-V%D5Mav@bM7_wW9*`i;SaE=eP=EIG(~?*J>QmA=%Ln=XTR(d{-LD zXsWs_O6fOJe`VW%9l1n?Q8eY3DB4gGPc4pRe5P7us#MuDYxZs0v9xr$kVVyn#V1c1 zJa5-sQ!DG(Qir9qgOhOqSu}3rq^`^vS$_U0PJHk=XX@;Dg-SW)IE!27-NGY(b2*p?%7CB>ZXfqTzd=W&e2q3_?2Xtm5Nf3kgC+1F5^) zAdJb>l1ac|;7%Z}r83v<02NYzT?TQ+OWEL~17S)$;VP{S=Y3f|c^MqtF}OcaLJt>R z2RFW^iIUN2y zRDcvu8U|!CG&~_fkdJBXG>PtHkj3|v`?GMAX7%24GUL zLU(W?c6?0;FPz`TR43q2-AImdD~@J;)ejJrUT;6X{`+6G!U`^KTi}r}Z=r*E1bWo4 zoOK1s*q8=bDYPzLh6#$t-lLP#zNTulGp-4F+&hL$#(9S?XNkvE?ulQ38y)M!+Qz0y zJ32ay8LnqjULJ)X9qh#yU%eSOUwyOl@kk79s)}gw z-cuZFLLTO8W~v-A>+^inTOAbz_=vNzhnk!;kw`sIHYve;uz{CMI{B41COs>LQm!lC z{im{PzfMV3ENSU~YFw0nF^wzv7&_v~`gTmsO~#g#x(O*)bYWh6Iu&oA52X{uPZBO0 zBPFFg7G{@G|0=uHCsYOt<1`94KWN+Bu|wgpjFJxuaR~CvCbbn_1tC)@FYFH0-5mL# zopI55kvk@T%&_Ws6iZW!abx+ql-_ak@8+wo#`WvhT>lAfKCnz_Qq>@UqA)b5#egQ> zq%)O4D1v_K>@%lQ! z`kQy&C19?H*E4Gu}(HN6LiVTfOUQi0b zBaGg0#~=sWQb>9FfGp2;CG(>LF+Z8dEvP%=9rQ5vq3C)HxBxXQH)%L#a4I|~HUm<` zkAtt~an5&ad^E0Hy{QLj{2bc>WQ*O=Va>&Tdu%%M^MqyfO*uX2YB)@4N zvK5+f{asrd$TN+MHJtAr_EU&S#tMaV?v_O!eNpUE9x3LM585!lK1mrU>QE&7>NhEa z(4ulYwkt2~fnRm@O$#bZ4l_etGG&tmmD-wdfg=-?2fyKo`zz@ch!TA$8~V@>%unH_ zo{U%UOsypPaxpfn{KM*_LmD@DO62}(ziXwkQ&ub+GFi`*bGdRJm5XhGp4==lPL;pI zcqGG4u{ROaZYb)9Cx@yF#0$5^4)3Qk{&<|AOn4(h^-TkDB@Cm6w$Yzj&+XFR)lT!G z;l-C&fF|hMMPMtaF~sIeprmT7Y}|e$JFCp)xtnn3=cgbt5O+SGLrNawZ-IkL@{7f# zxBM#7F6^4AhCKx(v{-b4%uhfXD>*FyrB{0Dh^g$7rbHGnjuM#Z`N^~nqo7X+W`l8* z?2TO}TN@xF(pSTt4c{$AA=T-cvK#2PWm_IDnt#=ysX}p_=RtuIXY09j&WR-f*Ja3n zOvNKqc-TB*W)hU2u!nc=#FHm0o}78{iCw4haVf4!$Ly5!Dy1IpW&^2!#jH3rAf`VW=+1#6ndG_ebY^BQmapB2>FHUw8A>E& z!B=ouu~vv*f?{`Gi)fBeaC`+n;U02Cua(N8tfmaIt1LW1$ysrpKJ+AmT#;uw-G2EY z1twvRyufpm5x|Q_x!DzD0msfLJF9udeaOdgcmhHDS%FqII1Fn1Zo^`GA*(nRhwRFZ ziyqGZE>2JT>@Z4c9p7jFLw ziXSje5#~pe2s$018h3}xN1B?8?72glb}i#s#-+%rZsSPQ1$P0`r^D$hJdH00Oy|{) z!*Sc6?u2Bv+qeYQy4hXKv!Kb-O3A?Cw$~5cEX%*#{L28}4;_`qRJ6)^{#*_B{6o9(=K%VnSDNVQz+^ z{Agkoo>ZukmvGE!iyz!(S}|6p>*OG?EGr2+Jv@>p&q1%}bDS(FOs~=%9_)$MRI;I2 zmidq+VFNBbZEFy+njaVMy7!^&1}^#&9IF1?&|P@rKadh;BPcWhgm9%C?Y=NKA2XBF zF*PwA!=s&8xUv|tGgIQ9aGjla(nPL~{elp?zdXuj5ytGCJ5N56qQy_jAqM{Ef~!vW zC_3hg6bg2hQK0kY1}1R$mJ1M+XLf{9c4;D)PdhW9igI5!CpxGz%1f=M%aA{nm#iah z!njF-onX#eqG;y}beWdN^xQr%CApuSlI$KG#QHXUg3oSG#O|>cdWY~pioI}b9k3Ws z`IM!)K?l`Jll-lKjC;&2;W>x!wZ{$7GUAsq2}9B1Q{e34fB8p#D7evKLiEtjP?(_w z+L1HMH0_KjZ3}w(J9bJPf9HX65wYYEkJ4EjbF@MIPkcx^OJYOyP!6rE%%qbq-N;;; zng-Ffk?%b8Luj5xHjPXeL=$yo!2%uJy1`Qkc1ClLl%$oO3Mu0$Z+$sPsBv-@deY<{ zVcBlTRZD?XhLnkvzq+7i0mc(lt~cWaSjJSIgpLxFMX5YEL=#(6DZCQBgZ`Dq8)puT9}%q#9AB$ZyLa|p*a2=MqKN!54hXr#R8oA zl%Q6)-&iz4y7(82AwGv2e=!Bx(HSiI3L1N5 zTOoB!Le1U?GEfpUX&5P*T>WP03sD-ERe}7@qVOWX_z8L*IF0j^0E7GtoD8PL7*mr7 zn>^<-WoK;Dlw`DUGuYrKQ8ft^kywd$B8OLiB~844GoAm;kHp7dLO6r`zZ&rU(Mffi ztIEnC*Z^V0X*Y=pu{(nS+%+oWYCuBfD?n6nKrtoM!Ig>Hk99XT;=>O=jy;qN#SBZa z;0VOz^t1>6x;wyUH_lWgqe%*95jft-U145^V?l_r(~+5|R?5l|dpmdzg;a@Ij-b;z$Yf+jholpA~}rHI*i3RlJ#ZiJP^)feHMz% z^>UQ0zIMn0C9(K7F)0s7l_z7eoSYAFrwk$;&7WnTzLUx`1|XfsAM%Q5{_Im&(nMS3 z3t|BM`cR7JvE)sXR1|&sHMmM{+x0JTw?o>0Zj-EpX|dmZmAcRu==Y*s36LK@*${s3 z$Scj5WYR}=B*SbF9CDPWMHgv3u27onH0?Hk3Vzc7ZMUJ&*Pmz=TMRs9Oj7TV&Eie+ ztNz=Ofp5rPPc$8=9ifX+(HaB&LUH<$F9AwSZ{-8#DIk^VU^`y^1jFd4uOME&rJ;p-8Un>l-z&d|HlcD}S2hLO6PeDcyPZ{>t#Km@;C&6^5JQabk z9~ueC1gY%ebV%AIyzY)7NH&?PnC#+J!6r6wmvj|R$&faF#TVnem$cFov-u_ht9X;%~cpcRfB+@S5)b~{p_7@P-V@zzVegLdyQKe}+h&a%?; zndk??^;mTCRlJS$G<+U7Uu=@XaC~HgVs)hamVXG+541G9%aXggbI4H&Kk}AIc@?jO zi-}4<=1p=^eL?P=z~GZlN6ILTjK>gqx{#o_FewtYBieZ3YI|!-at5Yi0jY7Mo zZR4_tqVKf@=owJyv!x#@QSv2TDuqpOA7AaGnbet!>8D{K&xvdv`=YRp4R@rBO~n}Z zMN8?bKwF~#aE=#c_@uX=HxY3C)5)O}V9p31MsIJAhHzW+)j$>p?EYF&R?)nhz(-P0 zUPr9x3I&RYl@ak>lRIZ(d0}!;3el;kCh+5P$?u?)!a*sXEbzuB#=Y~*7xuc6 zJ6H>xaA~=quNxWOMiIg8IwSOgHYX9$E(3 z9iEw*ifN@~SC}isMn+J&ISMoo9oeY}CGS1o33&PPhi)yzU-;pd;(ziZFUMc_(a*(a zug|!tFuUplJ*_Ed%R%+O$(fm$km2lhoJrx9BK)JDe>J}N`psCJnT(zmUfg#*DO&d> zbCiSm1fL_>-m*CO66_yV4wwu_Q_c^4$^S8-JRDQIPI5{8C1V7tVId1_}?( z#n2xzt4&p;`K^X8?6Fw!vmq2aU- z#1-?RA$!!P0WixBOGPn2+Wytg^O!;-U4-+DGy?u@D!Abrw}0)fHepQ)y9q1)1{_*U zg7abctHE;Ayj6LwVNJO6R6}^}gc!-3>Oq{c^Q*pCQYs$55F57@J(qz|a5aRH!{KcM z{=WYl5Lf;bDMMKCx4==(PZI26O3Lc~-oQEgxj^2xj5C#G=r02S`hevj^ImSAhEJo@ z^`VyiU1+cX7q`ImWVU@OAK}Rb$24`TbuV^Crm0NLz=R7t!3WO`vKS!iuEfg9Y7XF1 z9!Q81fuehG#IBoEQcc{>ru;nG=B{AN1o(u9>$W)a3%*pG&!i+s0TvE+WOMO6t?EHKzEG)*{+^oOqG$Ex1 z)@^EXUt2>7-i*~JEAjCDgIIa=*vBk-Qnq}iPZJLOK~d)h37_#JvkHa$ltI`awM}WD ziFDZPEJL??{JJ}G|y=x1JVbSm17}FOIt`8K%XESB@@N4z8vHi5=A>%IGZlcHWY#%{_|8~ zQ1S<#d4UN!hE-l(vBw$OVaexcj7!ml7781ZGCV4!H=lM0uhKGcqX2^wchB)U2}AXbTZ`O?&xWhYlT~3R@D5o76tO}scku8G1-L|;61ZKytXI#I@Y2~h4jQ9 z7L|Y(pbC5I+I;+5KX5Dl^lS6+#{6i!x-=O-e0@Cr{7-x_{?L~{>lf#^kDC|%IHJL> zbw>s7Z>`4f{^EZh1BbiunOoQ6hVcB@555&&c=cL*MT+r{efe{-pnk(yVrpumyjGHu zygt-AVi*3KAF{o`fai)sA%t})ZWobc&ZdlFr&5Nz3lIblU=)sVcxFrk@+=_zM9Ob^ zlyO2FSR%#2FgZpWqD<|2UG=_(XF6)Xt;vA05#q#b-d$vJo$L*EDbl!+Ah+5u2b zD^#Ir65FauUV@cbyySIqFlD~XMtI8@e>Ej_t4zQ+OM3m3(5yekA&%p}|L6abuWby) zGT_eSWZ?HHXz~fFD}(sSkiP3_s43cb`Z=CuK4oZ$;P*_|CrVZtu)?|W1mcy0{8JEO zT}*cg)V^5z5sKr^*0y)b*vaC>HzwyPxw($VFIo7E?~voBCnu~N%uG*#&BdY=az=P} zpDzw+z}JM$+0xzJyl#zn%HsvH{oOr3s;a80w zJl>7&L^q}t&b`DlvL~k}Vs>FZmbhnCN;GA1B2 z3ygzatgfyKe@~19Pb$;iT`O$@Q(cBB%-;%7pp@w;;B1~itH+LFo`y_PaiFB6jcH0d zhX=@#QC!g&xJiK!FLYRND7_z{7Cguh=~-ZSr&!7ZZaErU)RA3&j!FngNdvybnF}@+ zSF{mm#cwU1_+k)dpq!R|OWMDS4+&dWzk2=BaTSld7Ko!Kb+ek!6e@{VCCuc8IuNkj#d zqc6R*5I_Cd<@nK8W@2VxGR}s&QlQV`;pT3f^?LD#Kl^4ZPIbf&jw^B2lY4$8_v>r% zYrpsl@pJ#pUyZ-}*Zz9E{d?aKeH>33iJQwa@z$;Rn3)*z?Ja!iuhs{x!fw-Nd#-E2iZWNT1s%i}2FUFg+C9?83%g;8#P7F#{inC9ijG?Xb*aU4x z8Wv9OyP^-PWGj%kY*SYGYr-oH!q?_SUDzb9DCh+xuU&08@B400@tIP2=!uC{F7QiOvn_c_L0>`!Ps-Wm+^w$Qx>tk+Orz@N z@Dk^A#sp-!HSR?Mb$#&wFT~6$F!Fhl$)#qQe8d#kiZKHl=Fl|huUC;3gf!%91!Nq3T8LS`qYYK#+3IurFfI?RBkOtaVrPZs8N)E?+SHn9SMwV!BX9^O@ z21D^DL21fWEg4nPx{P#JMb?wC{`|0X0Ha9(TYs*1G1(6l$^bwD5?qvoZ3+Dhv0m`8~vKw+7 za_-M(#xw!9DMsd?CF$rRpqA%X-%XT2VG4IW4vgk+J@87*^#rF8{$S-1z zvr8Dq@uk;aisi}CSeWX@@Y!*6Pxj(;e=Cl5*W&DOGkRMqF?6^SU8P-^;KK)98hDJ2`jZejZc)@tWJ)*P=ZOn&8PzE19SrLy< zyfEPGHN~PHafOw~AdLiE_1KntLJ6Tf{%FsHbU3Brs+{_qtHM(`Qk-<= z<8*9V3AF%G+T(Ht67CO=%)dM4)Gxu)#&K# zzza5xOHe;BxoyS2Qwo&dM)AKb59({A9^h%bv;cG1DzRfyZCI zQ(4@b-4VY>l^&&YK*G*>=0h#^I6bTpf!xB0;w;$`7UeGGxd_LZ7);YP0Q~5ik~)pU zzy?)tjIot6WE4(M8Rq`s!LIMY?yB!`)P(Ea4zx($JLAb8FFtcTM)_4Vb!K5oS%lR? zVK(?N)VatsvWXt5lY@=|${>WT_)=D;l}%pZsY#o1uuJ+ZOZCIJ ztn!B&tMx+1-rc(_YNajK}hgD>1V$ z8)H&JM#l0HAhkwz4)*qZ<`^Z6NAd6%#TTHie0hvI9!XY?l?ATB&Sb3Z9LEO_*5m8%-;ZDX=Kc6bU;8M2<(-e>gO!cAy|EW> zKiG(`-dT-rYSHv%ZSe~1TeaR;M zi?80XOE)_^sWPP0%d9au3CgEA#QnjcbzJqbkv@I?xWNK)KT?IYFsJmgv746WHnG zPpO@caUaFp;DU?EpYkhP4(d_M#>y^A7)pQ6Ev^r=Kr0GhW>lKYh%7?ah?L@S@r{Kl z{6gu3HfL9c+=#-bye_EZ_b(v7?Cky8pkg}Q+Gm_fxkPc@c=RYXAKZ`q?JW<;95WayG!73BXaHATl<)XpKaNq5 zceY~qtQS|7X5-e&H{-@jH@p+g5fBDRWQ)47V-1_&6qA>af=HR-jc=4=UTQm1+I=l> zcje!ea<;L*7i&^ho=Bm8pvA!b)wOuIu7$zIR;+Ap`RigmrDdWVMcGr`*#R`~#A|ki zxyPB^V49Cb1Lb=MNdeeaG@k>{Z`n2&b2=+fXbH**`$GIvp#D*kq+Mu32<;>=`wI9t zGd{O;#V@CM2Z^MlKpBv-#!-^2r{dIxG17^Xp^+Gn(#qre(4=f7R#kcRT#JBD|iH{#|NvYH#9{On?$r(JLuD&Z) zJV>R&q*>;949-h+lcMXDo6CL~6goL0&dn236XP*6%8P;7hk(y5cjK*VlQG>H_4)Vp z_08DW+Kct=JuNavnS+fd@z%{H?=pgeBRK!~7yn7T^X^A+!bfI0 zBe6I;9WULu8Y5biQCE1#`R1khY1^)a6(*C=0%}(QmkfzSYWn(r;)oQd?E7jc(IO^F zVNZoP$5sC*zh$75*DteioCnL6xZav48Yn!Iqu3RaB53gq6MottWnJ#}*!i+A+}-}6v0 z^`U*!RQvBmL$=6^2JQE^gFzmbhp64{jhfS zR&WCdAi=e)6iEHbOL#sW#NQ8dD*SdQxHDi|_^}mVjQ$I*J`THycjs^U9d^;EyI8@M zxe#|A+Hs`K^!mv-|EAY;xx7qcxATDqD0xcSjA+GIf0M3dOe=Pi*r^iWE<6QAD^JT< z$yQ1mxr@uc3gF+S0W;kWr?2Kb6iIMg=6^a(8;EEyoqa7LWQIv^82Fc6$UItI{ zkBo8@V%l8sCqLyvapqIU;}c!)faXrUlwK*NQ63`>+rY0t0P-&Pfj-?FK zVRm;9Vq;B8u-ts9ZFqblhPsn6I>9l7@#qK-FGMk+93Ae*iOli-ZXECJO4Ho;WIZ!8 z8HD8TFjbKC|xd(x10ME}apUfkW-j@wdF9;|Q0M-QLG zw?4if@7%s0-@f}WK6vmrK6&aLVX8Y-xIvAxrC+i@0|PerqM zfS>TWKT%tD zUo?huo^wDbWmrj<;e$PvJ804$ZJu=HUqZ!as`>^#oDWG*xar2`&t>Ey5AD*}oVS%< zJj&FmJaZEVH%$0TcnaeMUzANBw^B;dl0Eq&utWJ}+o&9e*~+kVlT5X@qe7vj9aO-{ zZSoJAc%hD)FK~mV$G9q4^f_YFmi24rYh(JC1wI{RxW$Cp29+glqw2W;wZWbd{S-J| zlQ^5efTM-03BL@oCDQ;B9H2I^ql_yRwFJURkpq5a_fO_4fzFTH-gITMf;WG10J0Ij z3`&F7@nlTXk863J0kzDFkmaZGKqdTX<5vRJH`^j7H<#Mu3ufYBpr4eIJK?X!4aeR-FS139k9X{z zAz3#~;G~?|Z@|kUz-=kx$K_0-D&O%+ zM^uyEzfPJz9h1>nEhM*4&#`x59bkwf!Id{NP5%$pr z2@CnGNP6N%IaZhyHFotm>N3#ViJ?<1hE5Ko>>fxiq{W`}q$qC1(AkkD>SHaA6z7G{ zwiZ5Han_RztDWBa=t2DQ+jrutcUIz_6xoMrk5wt6`+Iz~?$}Bu{HE8hOR0Rkz7adT z@Oh@ion&dI+bKr!jEsqH6nB(xE8q%1=|#yxM)hCnjT^a@QY=~JqcM9t*^si%Mhf>2 z^OE7p>WUWtGaa6!48;%Jnu}SL$I}?~q}bQuYjt~D%7+$;;^F-CL`+Y0V`gD0x|5?Z zCS~|c^%#(xU!9v){}}bcb#?1F-hBNP^}~ty+P6OR*>R6m{Qs1fgo|q zrenxJllfO@B;DYE#w6K%<{d@d*EzELTIkFC>2hAxL6!}r+wItd>dyFT_i>#&w+kIg zkS5c5!KpI1^NlN%VEappY>6k>oLjjK@Dn3NBAao)UW$d-7~V8*5qOJy^q zC2QKhg;nWKOLlFLjCq~`?ADL-7l)lR362YpahS^@+hOg7<(JW2VF;$Yeqls2w*55~ z;Tjw>EOQZexOlGb8ZD_nWiZOlX^7P9{8z^Znn1+Hr68&eL$#SV_KxvDs!o{R%^k?3*3=sw7 zOP##_;Z6aO&zyP6{4+x#piPIN5oE=gB^ML)p@~A?$s)gBILXk0C*vGeBbo%~fx6S@ zIQPfoy|%JSZY!)l9w38^v5V!mpV`?cM4%LK4H^?5lQ0S{XCT>O=I-3}jSYW&?zndt zx2|1{<)tMLsyie!h5jptP^PbUn5<)?<5fU2(&)p~-&#SQEGu zhU+)4#UQ>@xE)jj8UKWP=+724bK4m6hho;-a5Wu((^8nT#KO>qfkJb0MzJPsRMy zr25urJQU6Jxf`N~V-^e3)9Mc$e~EE+Y0_t{nRAaMi}UJ7o$0A~@NgxTrn_1Q9>uom zy(by__6K)jV{I+gWi~f9)m={F&YefGvva8I`Qn=wYmzxDT5iP(a;uJ(w;HbYb*dt8 z?6BvHk33qZ&YZocMN{EqfosLHh>xxQPl2aOO=G^IQ(%*(|CfU$i^)vq9eNw-9m>N( z#ELLR;m}82X&dItq>ZsZ?ZiV?VTTVG%BLP2G`=I#yZ<7ajV^y!t~& z;)TrNfe|U5qtTI(Ac-y%&ZLiIqz7jOE3*WGh4LNhg44+DSm-KgD-X{p!UE7F{S=*j z7Ne0qMpQoMnmZoXMI)OCn!6dZ4!nb%^@N|$(l*T>xG$I&4MYElptxTctUtP~-y9VA zuU5kx8m`Th!o-VzP3E>CJ;Ky5vlUk$8J!q1UElQamSTYtRCj59Y)}A+Ykr%8x3SfcZF2Iqa!h{x11?WC z@otw3;c4=ZDkMJDuSu6-c3cF9L3H0ZX-ST$wBaVg;a*Hsn)K)JjALc&H2fBF%_|a@ zzV3MaG=_4W=ez;zR>hid{b2aXdf`cEXG>XDB>v^cFfH#_*^1OuYJX%>`Y9|#R>@4GHjvvV;sGZ%aNyH*^p%wLVU#VfIL=lvLya>P#7t>r7eSwebt+<5ErF?szn z!je0Te8hpTmmR4*lrUb(+t}WXoxMZp#8O6kqG7iuk4Q~im{ zM);H4E&LvfDOb}bc$L0C-)ByggQ1~L9Br@1@X59#KYq0BkErY(jB0_hCmu_N#M}MN z)i_vt7$b+&*3>MgB#SCmN=CkXi6VRR#$voAxfqu+?M()^oAQi|NwM!qS?{Whqc|Dw#NYp| zZ^xtco!F5~J$NF;clS{7Imy&jWVv~Vke zs^z0Y@G4&lOnG2laoH}PkmLgrld?z%@w^|Po-f(si4hUV7FkfRc=|`$x;RQ3T!nAX z1@Iwrna{`#m8L9D)VfS#G>w!ko_TYQ2_)w@Xd#=M)=9IzkPT3E7vsjogK>=8GRN3q zjTODPbNj>iwO{>Bqi0b|Kl2EzGI$!aurQ;({cmdFJf^(r|C9j@vfu!_!X8tl2>bKV z(5SpFK??K0P>vbSD8k@b{sJoW5^$eXdQ87(C%|s}RLOpPg{NS44tOWV!b^@y5C)HU z0?0>9r~>Vc+_RvtjwC-eK$sJmN|$CcG1evUg+cP@MFUKLAvlXU@3{c&xRxnsYz%f( z&+`DzTLeoZTn>G7>EO5Iva5!bM25s%00i=6%FCfiYX?6nfzCt65@g+)RDyndKix(D z^B~jPl)P3VZ8C@sY~@tK+}_Eprlw6OZo10CmmdxwYSLCQ{kDXuPJeQr}cVxZ?m9jIhim>0ixc6VcT zdfI0%IrhNK3=`dM?Cxgn?L;|o0UWruD{3K6@>+&sY!180PJY3|zcQjE&Kqcnq3 zOa{lt;?dq-tn`jzOv>imOE+U^ZYD;gaGninqLJcqs)_zoO6TsumdiPnQpB}N>}qq& z<50>FJJlb3csD-$&c{BR+wFE-C!b-JGU9<=lMynF0*WHZSL=2*cjI91AdaP&oUuzz zy-$vnIwvQ%$PVDBCcL~sVANM+>*u?u}r2Yj1g0F z^D)w$)WV_{k5&%i@!C;5(!yX{qW46K+5N|Bap%F~`1tm{`1;r1j(6XFC)QWjVv8pW zT9lng2|hm9mMz&l(jrTWoEpLk8=uWK-*_6ZvaUQ^{(50g%J16hij+<*jJDR}bZ=A2 z^>&;I*OBA3a1-sz(-ZOP(tKQ*R6D61ht(z=%{deQd6|>z(71&FRe;m5qf&_ZGTyP` zdBKu}23++yQS(AjZ9(NHAKXB>LJ6Cn`(bQ@YA?Coo@#y#Dqei|`6lKraF840Z^2b7jN`K2k@6)wtB9|*4|ANFo^PtVIxYv6b?g+A6=j`;yl?T4Kj)lOP?7LdH|2Kv z;^vorq{86|^-MZEg2>7+FCO&o;}SN`tTOt2o&YamOGU`f0@d`9RED6h6FwDI-4BN>lhh4r}8!B55U=>eR0;Xt)* zU{>3d7&Ezo80A6rlbDj_lR#;PXM^xTOX9ZEG62+Wn(+Kmn+7+IoGz>sIWVDd7YCiq zo^z<3Q*lolD`rg!nS%ej8vRtIF0VWzU*_ep937GgB+FXSqjRr6yf*3{XzWh?|may9<9X2y}R-7-FM=hul-)U z|L%Ly+ue`-y)EJ2wPN7U@k&wJ+257&-V==|?J4UAQf}6zFs(@OQX(k;yd=m@NAeY= z9R-Jva!?=M=07@+!o#JsJl9tqcI>!ole=BVxiU?P49YIYLr^+YkmW(sp?CA zO-!+Av$n6F7o>KB+hwXzoIg|4bOht+%R4v-(8M`38dF!U#qscDJlQ;m9qF;X!EUVY zsg2~`*zLuWjh(pvXf+ ziavRKMqCZ)U4CAR<8I@@(?co4o6b9OB&D#oD~<6Wy5cX#A-?eH_4wRNSL6EpY%EPp z`MfT#g`Nm6GRso}KEdtd0N5;s)F!ks(#m-0JE%RT(yj_q+ICZ*7!uwY9f_{wg*K*7 zO-)U!ojGSjSV|s2ued+7*3xE4> zidfMu**;SJIM3}P1r$VUx*ktMBU>aahNCa+pl98oM{^9e>ey`1I40q8%$I<0D{J4- zZyNw_aXu6HOf$#9@>mB{u@Ltztl}8YK;n_8R>#?@Cync?`Xx{3pINsoo?pUKv8DeZ z8?=YxJk}-jd50&%d_EW6$xq++XU@+U!-R*4fqRXG#$ArxSa3Y!uhMau$#I}h^0ANY(}Yh&8E)6U&jh*PT(KHxyDvbB zW10%*vKCwyjeh%o_P3y)h-3|vYzda;^rWy9K!)jQHYH5Z_wrjK+U}OG{4|Q)1*qy_ z+t9rlWe`JHrQ^FmCNlp?C)g=5d4kZ2z^Wdm* zvaU|)U!YG*-|fy;I$S^g`9?oag{5&+-?+@Q0b~e|Qkqwu1V1SlSvsEr3Mt?+1T}fP z{EN`amY}RD=Qt{-+mP6a%D!{~cgf?yO9Lu*;WFScsPRab$DBh_-tr|m=@1(9;6rmZ zSQ9Cq5@x_>67kh)8YEGakEEEuW9}73iDrk^pX60qKGBODfOK@MV_Z8sJ1%p4V#4P+ zj}LZZS_f8ooj2%r+lIA-TDXr#~?8}Gf$kwZI z3Fl;lG&G!}Ge2P$fl0TC=eHzzDrJMa$W_~zo+QpR*yemKp^}xU`Pe#)Sd)UgFJ*Qc z#dUB*3l_O|54>PP!P}D@pyaM?Zb>O#_myJ2?Bz%HgA$sPp7@(LMW7IiVcxYxVFkxN zA1%>BhP#L*MKL7h`FM9LhNJ)w3dgXBqpjY0?MD3MSH2Km{QT?j@~byv`Py=Hr3{Zq z9wjKL9EtCo;~s`V@^ikszD$Zjn`0wovqPN*<%PB-JvKPt4?Ak~E1n>5)f&J?GV;Wg zf@5lH@}L}ZR+x{$oTxl4)na$+AlBEm;&A66_Qj8#%{|rg!2Oe#5>e&`MECZA7J|Y@ zx!69=A)s8Y`NPldH0Sw0e!Lc2ySx;ua>dq(&Pc4SJdXXHO)K}#uYSRCDL~}&Yj*|L zvw-DH@%1nH)_Rdnw6>LCe(g4j48wl55vUA`aZzN<%$I``wV}`_dWHwp-!!~40!NU*)Uc4RR(35hx(_T ze1^O5E|uYt5>6`~g5zZShI2%sAP}F7^FEu4y+Zc*NW%scr^xId_V$Rbk@Kr<^LpoGh3G@Afi*qX>gz zlbt&$BbuVg9 zVC*?l9w}z3|CuI;p7OJRn7Q^+ynp9meC?g@nD6c%t)|2O@{OhV*M9sfv3Tu<;_6tG zldIG4=P=r_7(jTu_^fdPZ2Bt`7K;^%Vj)TNsCbSdfU@EiVulwUk#){=A0G6~A7q)2RQSb4)r2t%9-3d}Cl6c2tqZKi!pl>yO(izVx}5 zV`*tI-h2OJKj@$iFKHn_9T%4tVrG8MFO;$fhIsOW%kAklEiUXP9E)_gQVuM8q7Jc0 zMHn{F@In0Aul|O&RJ$HVl}hT$f2Y zMrY!~4#sO+Qb?)zC&*rKjee2kBc$Cbr}I9ywaxv`=6!YePwkQAwo6e>PxJ3BodpLzXO{DB|(Y`l7XE?)WU zoAIaq%%6z=@PGDa;(zh){PFmipZZFCLGcfttjGP8t@zH})%fMNZ^u9Qwcm@Ctpo2u zb4&pd3H^z=+jOUa*^nf6&|76ZHh z<(JURYaX=nXqg2BGM#6RQH=SNIrp#fiDCdg`jKpb9D~A&+1~Dsu%sf!&Rp%!`F4Ib z8-Dt$eK;%hg{I^w@lId-hJL!IfO-(!;LWsnQ_khWh&ND#ao;LkG64rW>EZ{Ec*P4w z&#SYYxG^yh*CvPKt?LuZ*!+`C=Mj=k5)Oiyb^aFuEpKQ8}UFgw7R($JNx-D4&Qj8hV_@$FvxPu&#p#%u8jrbYUd)7+;W(3Ffq6VI94k`%v&?J)Ye(*E5VtIZlUcNdX zuiU&IQxjZgxEotL8^XO5_wPQ8-}uE}ieLWwe^-LY=hyRzPeS6V&~E3xJL`?Ss3xwD z4V^*IKaP-n!}yY1yFD*oX3Wcy&T`f&u;z1`e!7HqIv^uCdJBcTQ_$fnYf@V2pS=?; zfW&`xh)pAQG3U@;fVvPwmVnCl3hb!XG=wUpXnqDp>LA*N^fw~fkQZL8Bjcgk6JQH_!-YVP=JPOUV#@nE~2vpfB*Pz-K@^ZIh>(Ou&zg;~iA| z&0nTy^65WsC1_&KLCVt84et=!frL{>QfbS*jDI*%1 zCc0pr3!pDXXiE7x_*a>zKsgMGneZ8=0Im#5`VxPk)vuV0CmmL{cm zji?U{#J#m`<)8ISZ8Nh{G-f7U_SWWBtgmgjZhYVXo}mDslp=?mA?`>??{>OA<^f;$ z=mJVUb>5W{?>k(@KX%qR_Q1K#>=#njG;zUCPpT{iiegZqwta0E_(vaOvNR6y(P8JC zQ6TJ9PcdYhJtYDH%ZZ{(QB9f8gfgl3;e2|07g5k{&XEHlRO(C%j-g4(j}*DdX)S7o zV{-H;7N!PbZm1W_6NB-F6u-~Dv>dNWNqb#w{qpQ&EJ`V!=?q7AR7#5$7cA6HrEI1G zkWvi|UM!#uXfmd~INoyDlTxEL!uWI8s?Svvz?YYOJIdkSPAp$tic!g%KlY(CgObPP zEA#R4&8zY94Q?-4h`9-#CTP)jbP^j_$!*#DQp9_`*5iK~M5It{jCapU zn-IkUk|Pj2qUR{ixa5f|_@<-`b4NFjt)1tK7v`pXWMh1$6W4EEi3QbRMDlxz0)kR4 z*+eej1;=kJ$D)IUDP?mk2%gYS>DLE`M?Nq7!R@u%G?cWk_kECcPT?Ps&;Hm*l-ElHT6R0{ULH+X%De~;(C4I?Tp~3eYXeDo; z26^tbk?dYT*7u1H;*0XkF$5m@vm0+3wQxgu%f;sfT7jkvYSqW);^@% zWZ?&=kr9`g`umAU zHR2sN5;5ZC%ZSjK%DO4cN`I>GeRAlZ4QRxELspZ-SOny68*;S4xoKQSCF8w{BFKg? zc@^Fm3v(LdHJ+!43y;%X7K=N5u4m&* z$lVTJ&Om?^gE?c@E<2?m3ohi8Fu-mZrxoqCOT685AQaggywhaIuLdQcp~x`we2HK=v!bY=ich=A;LJk1qQh6nfODRDvu%n86p`I7iDDw|J9&C&ni_ zew@lRWGH*w6mhCPxF&Cp4yQyy1JHdS7m0hJhDPa4m---I|n8N*uTuD6o8jr(|+2s+?mTkbx z1VdX7Nohv88I&S4IKZURYIBVI$WvaGCbwiLCv)5>X*|ux+exAA$MD{G3jztLdz5>aZCse&kJ8wYi-tThnF2W+uM}lyhTh?dc8J2GGcS{ zE7qsQ$uvh$hECgwN1nFnXI`-57ap^7Cyv|5kQQwIghTbCI_j?WN%@Q;H9eB;b@5%U zODS%(T?yu#GRG@=s2A!}xVy{luoqqiyWxztc|VOE;&{|OB)O?d5gr>Fw5lqvzYn=k z>+)1V?!Jl^Z4>mNKI>Kd!I5b@cm5gCR;_n;Q{kLNAGXcbhV>1M*m{$^)uZiH*JHzj z_QJCl>?=R=Nqh9%NxQkyvez%&^p{6_RVke5@7Drve0;)hVc1Ibs&BH>7Vk5#xo7gXP7_mv$()tur{R~pyg zgFJRgfW**c%Mm*~Rd?bS^E^c(J+EM=!9Q~3xlliezQQju7~?)BPb&Org?c1x7^}ES zu3P!@-CI}YB{yBai*=h^QTrq?pD{J>+t~oQ(CaB z55<4)!GV;w>{e)j%18Xjw|s=yt14%|>WVibqy~Gt<{jdGjWxSc*f)8iqx{gy$PGVF z89wCG^ZIZ>qVxR$8n_G$s-eV2!vM576&}sg`SGU^{yh+&D5!#2atK_)?nM`jg!G26 zM+YZ+Wi06SA)TkhqrjBrKFB#dfH|Kj&&d_;G7k^OC@4M_WNPPhPo99o{5nAVj2+8< z@Q#wOL*U&e{ADTA1YJZr!pGDGBVNjf{=j z)X^hS3MQ<-#!hq$X{0weD5XXDSz27Nci+8Y?_H8&d~41&8e7Uo#YTpQ?98!gDZyv$ zlaHUXW1~_Sme*})Ldx|c7j5F)lQ#C~vo?BO=9!DuJ#s|lmpoDl$`|dxjxS&R8W|q8 zTl4GoYk%;%{r!LO+xFl7v){D8_pkn~{nFRJWqhB-+1oj{{o!W^PfBp0%0Eq6hG3v7d|G3eOd6j zP@#A^jS;`m;OaOMY59(9yE0Fnw2x60s%b#Yt;1w<3_rotL1tD>X7k&o`aE?bkLI5IM1Coi0_ZpluqPYO(X z+4^@jY+-KBzWLUN_O(BG*S_}3TlTGYuG+WWy<=})S+d`I^_u<4A78R>y?tGZ;;IyG zE%YSwwfDA4)pF^hRn<}gHFx9tYQ9jfIJh!yK zT4?Oq-13IiH@BoP^8t?`Tb81{s)gWNfAWgjrfx63_-Xsh7r&%2nK~3cP;QtDlQjYK z$PfNuq6{&|#?c_(h>4sz#zKNRI|wiiq(Ax}bU8mMVdo;h=sRHUK=%iq<7JXR6i@1& zr$7Ml{pmyIk=pa{dWbv#(B}<_(077A*^I1)<-zrtl=)J;sJpCGQimKr&>|THRpDOl z6c_NM5CIea+qeR+Xy zb^d0ul_4sC2F?``4V@aD!64alI1{RT_Wz(M%u@V}_z7co;$DO&4Jb#)U$li^=1tlu zJQ_P_z#?AB&*i@4#-1N1+>2f~{_O`3N^vDL+4P`8$O~}VWCw5%PiA_uIrj;Isc2YTGzzM zIY_^Kq1p78;rKKylk$lZr~FRX*}J!Gv(@w+z|*HsiFdCpFD%-|`kI|Sb<#FA8n(Q; zZb!x^tVfFE(xtcTr3)wQ`00~&;=-fWJ5WqoP-Ti|6QL<`7UjB-dYjty5zH`eKq+GAAZ}_8gj)~L; zhU{D4xoU4+xnT<{>ni`If7bW%oke@~@=g1rcTt=^viEK+*q!BuYO+W5)o0zk{kGBE zk`g8bMb+$IA{!dB>+@^&+B@$nS>-*oPc~`)V@D_L(=WbYCoVpwsZUCv1jv7?d!HSbJtcD7yRrm z=eK()=iasy7A9iloiF{dOV8H_YZFST*ZOSJpH{4{Z1@cot^jJ$Mmz3o#ixv4`t);N zk2f#hu-9IG)qd?af5-mSZ~USC<~LroufKlD=G$F%XMNl1JCsgos!Wp;Lw5G~m_2fG z%8pMA+li?mn;ISTg%6v#?DV0C`;8NnTBg@eQGCV>@se!P1Sl=E6~`sKE#&5UGt!W=!Rb8%a*zu_W&&N8Sq2AjO|9%@BsQNi+cEnGef6V^in}298{NShUr62rZ zWl{>b^2itS9K_Rql*@m3CAZ3hojP|x5qele^gj?;cdjkAXyv`s?I6LoAL`dP0inIt?!Rv+ug99_5D; z>gmA84bkvd=1sJ)IbiF!4hQE`g6#A~Olb`AHq;`58s!{2^38D$zf}UoS^3TR!;G{a z{tI;97iD;D;Z~~1j6lcjVf;}s*&Bbi5FV}w>_A6hjv#2~srF zDT;zenI;ov_u)WLfg;Qua*P09u;7<4$8md5L~1z?>f8t0qU+{NokalN65bK6EohFzf@93J&^#vC7LZmiq<(xP3sc;1d5KkmDFckgh!MC>ll zEiP*??UjDJZDWIdc6@i;o_y@Qo&VIQtui{{c>pyp-?>EWmXAL z#5v-@IcD-QIy~g(gHN12WoMpz!p=Ycf>p*wY?r}XgLp-Qb`N*gZf@A#_NFb~zHPJf zd~I&s6+i4G(~q9DC(oa>CmuOs=Z}rq`I8eiIa;%`N5p$_+^_hWU0$(oe{k3S z@ZIb7{?%Ky-i~v}=n_*CBQ~zQ^p8x~wb>OZ!UcKY$B6qg-Q z6iLo5$9=C*LYK!p_f$BOQsxti1rW+Fk7zjyNo0NJt6_I;-I0PfqdKo^0nujzs;8blO`ejg z?d@G_G}~I7v}|RyWt$Doy>g|P6v(QSMaf5X@)0STYV)p^>=H@vdZ^cQov&aWv& zm@5HCCTsTD7az9|-~PaU=bLZZ$>(0M#l5Qi;kW)kJllS&MSFeKcA6XZ=&=bqH94#W z+K?R;)=*wlC=PYze5dJFP+!-P+LX%M8&*j(f*h=e&%A^r$mb*#Gf<1)k*$Z5P( z*{k{(s>bJhRQL!YsCr8M^` zygD*ur_ViZBV$LcuKJMpiC^rPA|n*XWzT)0Oc`8;Qn?BD$Nq^DGn8Sjf5SsEjO@vu z{p&1$l<%e`^f&}+pmjk+xx!jEYtx-QbmqW4BC(V_)q(yoqFn&Wj};ALY`10@rb%|xe}m| zq7b{ntsly$kKqNs%rTi!lTq$SL3RfGgas{bSwrBno%jhMlgYpm9Rok;3W^eKa;sA_kUK)_X(-I&iC4-) zkta;W1Tvp8veF0jSD?EyZ8@I&!F`JQ0iGgAF#_Bx;m3u2^qB4urZ;<;S=n9!CogOr zcjP#rV4HjkR2hP82pN5dvN3>R=NN-4gHlea{oEu`m0Jp{@96c};LxCd;VU1Dz^qhw zT&&4XN-BzOAJ;$)58ET>E=XB!#@qAa$?o&~{Jih%4Gs>v!ou+h-|1`BZEUFDhNU$3 zjZR5Z;DZ|Q=yQ+R7oR$3 zKk)bw`>{`)x34^V-hS%Y6ZS(FCv0M{VsBr%Y%_BU)<=D+({c6JR&&#arSy=$GPqm` z$U74p3lCmi%HxwansUYhH@8_pZqfaqDoM}@gLgURgMjMk@DO<9h+X!^PL`m*ANY7E z;JUr89&YX6{B_lu&29Veqgg5G-8OaV5qshDKWyip`=kx=0TwMxc2<^cYh&4ZcU#uq zE4l31^Sg;RH+O7trD3aecAwN=daKqwFmBz0;}Vyu)vCQxwk3~h!-|y4kzroKoUqf= z)3$VD)_&(VzGc?fu=68b_S4TiVlPT59+ey(9UZbyJa*B3?ngdnpL*=19Ure~;WKDs zHNIrnHZ`e~7|ELCdP=mz14>K$`OpegQw$4Kh}zS~3&{?4AKO{jNdDsmD-p_z7wI@g zME_yDL*apq-+QYC9*ew3P93!!DYdhA=KNf7x9V=kF_vkVcNu}Rbi z!aW<*u_NOdh#mh3zaOOB6?D`U%38O^go^S=UdbEnz%IIG!64oT!6OP2`M^p(s26Wb zwQ0_m&t+l}^VfrVOsOuSqEaoyttY&I3u1?iFIaNVuRA=R`1>zxSuD!{+t3Og1S#!O z9vmd!U3*EjELh}F+L&89IOD9vw&F5=a?Tla1>W=Y-O~Q|58oGMI3_Cuo(f8pBeZm((l5n_ELIsg%7LHL+~=2hBxF+Z2^2gK=BOVHdJuE?9v$CNcDdzQnEUt~ zL{C#>Wiy58Uk5)1bmS5GPB#cB@$m3)XNtqnr?#}!8-knutD}{<7l=O{Tn^kogvCw( zgGvsI!Ky=l4@AVE2}_F8@#9C1+z-1fKIr%$gMS%=hF|0*A?_khGDQn@p#Eomgex%8 zPZaLPe?NBOAH>5Tcq#$a6txA0f6@FNz8~G2e@+Y!|L^vWjsG$gnC*u_4&iz+%8Cby zDKFQsyBZTUN(GOz`&`lGr8jnpnZS8lU83s~d}LP>+YWcia$l>aIQg%vtcj-ACdMZG zi*ySMi?*VPmN(wVCMIl%30DI|r{eKMg1z^3}~wwA#|ox^C~;?Ci4DMkeg&sk1gTG;T9@ zX6^NtU$Ni$?Qhtx{Q5WSzyFo5+b{j*x9s1(@{V1;dB^Kze00ziD=oGBs6gCVDi zYUEK4DY+l98P~J27>{+Ko+HnqRT?)eV11>G=4#J$de&Uv)NIRXVCq zjz@%Y!VaD4+s6`N9zVQ)t1I~KEH7d)=Eq`_ymGrqMY0=xRV^%Hlv%z4xmj=7?98lv z@WF@Fq>U#J~7V@oo*Vjq5Z&E~W~ef)_h{H!&{p12>{=cKY4QI>%N;^5;)0_HR1#(SW=aUt8t zJ$;I?F4s}qKHz(a9=8L;!Vnka7W9~8L;o{61Z{TZs4mBcU}vrlI{C`U%Wle_{F&oP zgQqF^c8L=AeGpmj61h7AN%*{Q;0K>D?+-zhc>^$$4AGvFm0tBFu3clNnEKn)xaC*Y zIi@JGGmZ1vaCiFu1Rcf?1j-O*QBhen@U-FM?O3ECnUjCNFDleWWg-i_G|3YI6nEqh zeAw`PIVQ2bbX)(aygaFgFy{&ldyrrj;YAq2=#J!9rU?_se)`uz2}}4#xLV?#a3ADl z&+#)daAbRQj(ch1oRZ9x4^aP-9c#(IMDy{GL4tIN84&(%DbSrIt*uq+rwu_YNd~*dB%vCCBSPX}sMBLCOCzK`wWM z?#qhossBcr@-N3MgNHk@_|f(`VhP|2>-+IISWep6ON9%Q4R^GX?YRk`YNO{)6VNeRr#ORPc{>VxHpuzaquubq$g=44euHyai8}E?I zl-ro(aAA&()$D@eJ@v^KtU5RnFP5q-OoGvuBc98H=%c_`^O%J*#u30LNLM;`yw7p> zGqhE;1tu5y$L4|}Bt#j=c1Q8h|9#_fK-hz<<9jcVe(V};&VHb}URT@g+1<5G`{>$T z`{QrFVZZaum+hbY`tRDy?_ag|Z_V2~Hy5m-Nwl}WX2VkAuHKmQ4^W)Fc*ag$xF~!~ zX=-89KWshKk=RLZEZOqhye+P*yOQ9KsFg47ljeoXr=NS=Ui|!L?aX5rZE#q1+au+< zYu5(4HBoOYD^HSjj`pa1Z{D~moU^Ugj*Sis+8HU3pLp(~O->D4O*CH1F8}6rnkvET zJ9h8^LZytVA8;#)hvOr;q`y$Bq2SO4TvNzX5D|#qwv^$8*+tRss;`XN=`*Kg>(PRA z)BE7akC z)1O@qC1k$g1jobxH|aUw6qd`h-%Kd#INCt|h}prMFRBR&=rezpd*#o^Foos1WE|`9 zr&HmMbZC1|SmAsAftITuUA9-g{hHSgn*xOS7(_o)8&6M-*q46nr?eoJEOXd%zn^>HSDx0I#np%8pzvHErvXAUG9-G%41vdNRsRTdM4%AmXgdz4>EF2l^tG+$SyZ+ zy(NWYr@JD>gdGcZk}>5cWqR9>NhEn!{)k8&`i`IGK|W(UIy7KY6O;DD`Lp(kCob62 zkDs$sN2iO=GM_j)Wyhu`+&-c*Z*g9ATT_IOKPu1Gj@4D(tk_d`j8VuOz?n_jcUww+ zYh%r}P)5j4K~K9zJj#`ke&%-hllA=&V^I(SC~PvstLUAj_@1Zmb-)d{5>6cQ;>4mV z!or#naFVcknK(zO#*}h9;EG$L(Y7le+_IUOC7T)@vGd1I*o6}(jScW!)zgJzllFxd zp0}U++$Ze^KlQx*<-hQ=_PH;8)&{4K*x1yR*C977#LgFIs`sp~8czpEuOivtCAV(T zwWa8`*B5PL{;mXi!J4acwmNs$HWuz!^X@HMxpCd*u3opR?_aSuUVX>D^Tr2u`}U%p zI6i5gd+{mz=^y*7{nXF=uq)F?k51aL6H|8L*bzH%WYV9S(6-D6Uav)63Mn@2DddQe z*R8_Yg#>xR0{TKnJbe67Uu*D*>c*x`iU08v$E>P_CwFVN)|bSmWdozbwkzd~GwrY3 zSW-XJf-dDDw;^!Qq|c(XUB9#7w{DDT!8S8{*Dk&H9(S$TlP`V7rlybiI|1|`zhOr4 zTz&+3+yt^bxL8cWVVAbR>pfXxI!Pm3_m#?f=?Max{{G6UnJrmkf3c8(dSbX`F(kddIXzvJPr#O`G$Wz=s+~XfMi6KY`3hnyXh<+ zCQ#!40D#5{6PphP$(A&# zj_#5CG5KY;27T1`dMmC7cZt_F6M+U(C8oSZS;zwm zhkVZ~N~Rh>c~+|Ah4jd0pC&aSkCyuehK+kud!%T0^++LBUR^2nfqKssP?!_X`?h|5 zCROF&3swD6RCydOtLKAb8-oLbc3Mi@siViF2p_TI(^FE6Cv9SE)W$}*aihnTAnFK( zN6dT|IqG?T8<2@sswMA9&VYkfQn5{_IcK z|KV@_75h8C@aOG+_ve4Ye&(|;*zr--rEp|w(q(31VnRb%+cul4YWq#;aGSQ%uFGEc zy5<-PUu~<>_Of%vcXX&~QzJFoYd5U5wrthj9X~=r8TOz#@XAPcH;anU#RZ0-FC+h>F{#Y(NDb`%2*m!Z23d}uG5j5qV0j2; z37r>qAE6TBpX1yQ6&}hof|mVBH34k+6=A|G>=~7|3#^+RK4JGPBpyNIcKr%k&x7Jh zfI?|b{PItB{Y&K`ya-AD*@#Z*_2fJg{}i5zlK@-CQ(ZV7h>l4k)_R>y}tFmO3PC7flE9jko5FlT_V zGm#=|Tz7@y!Lb1*^MP8;CdS9r6y1u~FIsMQ(1a~z1X*;R8W@|crqbK>ui9~rl;a2z zV}ESBp^15Pbl7@_$6PT$NkwtmQ#^K2ypE*^_!(N6v_-SgQr~<2DNT+fy|=VuS8m>Q zMcC&V_^4rvv9Sc6x72n> z1d1j@3m?ruX=jPAF`sA!Pq|b5b9)XZ@&JFzft>i_hOvvjpeT`Y{tqoBJpklS2vtUa z@rgS0qZ_J2>JkJ{HrrB$V<($DlSf92K)ASG;gwqU??5N5+^%63)#N}en)Vw-tIWx>`wy}s9I1T;-f|n7Y?c6hXElFf#FUTH#+{wBN?4I_+^5C2rpYocAPGJ z_+{^XjsnYR+>2CK*!8dQ2iyzO2R)(t;-Be)CWy-Z41$614=nq4gxmcy|BSdxeg}E0 zN0j_?m`oJ~-ue@Egcf%|gz)nS?)ITaw#3sRn;!#|OdfDLfct)Ue}K(lm?3Acuf#N3*A9mHrCVTpl1cEb#3 zf=>IzV>Ym(j0HapPEuY`YHNIqV1U~rcww&BZ;aqG%FwY3i&Bhiuu}rP56DRwW(S#_ z)JC)AFQ0MOFM}iZ6ZeZ(yV;QPTGv3hZNtMun(Ugz~O&RSI|oyArGlSze%+DT>My$~?-H zictXkL3WM?hijfc`Xn?=TEvO=3VPqsDF8>F%HsdM5albto!;wH6zU<3J^aKnu5XGs z@xNr|xSnzMl4p6mxEK8EVw6FJLOE4sGi*tTlv@)ckL}xP|B=BytMzVKcY9sR>XPJb z(@q{8vC*zg`{<3=?D~fv*&~mi_koSwUUqgh zDrJx71-!I5H9n%Yl3X^{tVfD;)wXO%c^L*PT{b$(FlWJ?P{ z|M-U#RzAz^#{~q8;~d+eYAB{FOw^Xw4ecNK@wN=Fs5iM$8hG(2jxp@m>PE{N?OmH< z7hGXm?YQk^XPY~h&1%&F+q8=P{@Yi@C?NWB$y zjzDq0s`QFFd_Y=3n&c8YKe@4`+DI7n^4$Ou=D_%fL@Zd~<}ZFp8P4Zlot5kX)CUno zsXydA@`J-I$*p+$QIAfWpkxhzS4QU<=yS8sWi97X?St~}bt!tr*X>=&YEiF=2T)2U z*Uoo?C`r(ImSjf#0wvy1I$6r6{HZ$@6&#bGF3BtML;fE&Ahq9T2*KQk-l57!#iao` zqiB5MOBkz}OcV4z`3rqyl^gWOaa6cEz9Vte`6KRwQU8Lj$e(9Fr7zJFpZvI^ldp{U z!-se&-02f4jGZPoM(_l>Dksswt6Go3FIAzX{Wpq^s zHMyj|=JJi5{>CmIZq9|P{4#k2-_OpSlwQ9o%H>fD1W|F8kKRHI>XAISYzX-xQ$q`Y z+}CJ78D5SWU#}CAhPP4bW9*M{l~FjEaH26YkKmjjieCUBkf$i_K^(@NJk}NdqgZDc;)B zlD+fJHM?@-u60k0*%$x(pS7bWPh0=cxOHnWq8{rm^ zBY9zWou;p^H>E5{cKHnP#EA9v_o~kOt&g+C9OEIcDq^=}hWe>gs%qZ~s+<3I+uPP$ z*Md%F$R0;t$Zd>4Q4JB7_bGX)QpI(7MCoU@E-TNzLrncrCo+*SJ{qzsS*(tZ+FE%>GCvWAv$nC$#eIWl20DMY(0qF42QcHbY z3nB048rSq=M-yc;7SWlf{`=fPTc>i!q@LjeO`fkpN#>`#dfq$WzT!(C?Btn$`k!gt zlh}+)>|nDP*SH}W^`i(1W{lzN@YY`2^Xv*S;~bESDt_?`mKma0w^nN)2lVoF=RufPg0< z_g{(!dVR}+$ax1+<1wfZSnhm91pFU>St-Ttt7yeXAz+X!3kHDQJGN+iLP+sYmP27q ze*Q}llD&f|tnktQ#OEF?D4@X}KOFJ>_Cj1ET>m_843eIg{TSRJ#2x%QaKPnhIMW=f zfP*m)#trx3SRU*?F6y6Swr77=h*kFNcFgYdqiz6s6&QZu4|Kv9*ibeZND?2Jt~C3L zB^qQzATKJJ6p$TjOIfMcr8G-9YAT-#^NUi_w@tc4 z9EYH;#9vxXoIRF2Zu{boVI-Z!6EX<`QCl%SFz~p7D;`W+yu{^-&$h}cl?>NgqOO@7?b@hk%+sn7cX{I~%GC8#GUf{fTGslI@yNPb4f ztvWnm<4ub6@UQVw{2gY8io839`O8*1A-ztC~ z03YwizKPRq)xiN9oSd*8^%JY~`)9sCxPI4`mzMp=$jZu+{mHFW`@j9_H|^!OFZol7 zfaud9Phl?uZhx`L`#x!Mo`28o&dkZ&bY--z{xmi=Y1P_*vZB6D+ejJocJ`dY#|;OU z6JaiQXWWVK!vT8aK;uRi=W_X~BK*me##MOuFY=s~$TDR1m|mW67C`+vkTJlSCCq@l zW_0bLgL^*LOR7|E)Jv31jw6t_*c~R1imHD+g<#b21+?gKlV{)TQn|bx%IONc-aL;H zkW1kHMLo+mVP_7>awJR!-HDNbg;bR*C}oOLawlb2V;jg=P8#?}+$bmcz6X>JkgG1X zdISAG;hjML@A<^^|G}B@klSOt>4ZFOVw0d>ihHf6-zt7H$HN7r?4bj7Rn%7rs7EH^ zT;UM;>M_wf;w=m%PBu!Akg*w+FGg(E>hX90f#rOP?P)TT-Idx-G~a#5nE4-$kuT+l zgxn+_o@FxUyD35@DRHL35k0c$J4z+=L5ZI;?)a;V$P_wnp@cUp&`f^3RKSh~d@&28 zvk}LU^3nlV8Il02cQA#*U;goSXU=27M3wn^6_XG6CDfnSeP;ncy@Zk-Ks~}Q(K-(y z%~GN%cj&|L;{q%roxJcN^=K`2$^7kN_?1{H1s_0)^l-~x!-)@Sj`*o@9Ys?{XQebf z9ruek&WQm>d5MLXIz1>Fe+k5qyrldjWPmL4LfAeixr?|5@(Ie{bD(^290ByfQGe8t zx2w#UTm(5C!l^&;0(VgWl^+InV)AlXSD(tHtT#0IZnr4|kCX@e1vYkjQ8_rb2+K_= zHpoY>l!xlzu*c{6H3n;rN35@|Yv4mRyR9XyquJQ9eofk4yDeK@Ua=i1gmhQ##^h^p zi%aXaI5%%Y70JJ})y?j0+flsA&Zc#1Am8$xUul=JE3)tCa*Y<}lG&j{s*wp-f~6R< z^SrXOVynxGw%w}xV9b53OjKLlgSOKwy<%>r2iC^I^f0SAAko_J-SBSGf2B>$&!!AuY$O1V+N zTUeX4q2Vd(>7SHBJfMZpfb3jhH(*1O$6+n7I7_~_+mMV&DH%Q~g>^i&LBCbIIIm7A zls>z|QmnWOm=`ZAO3!ntY%xwY8g;vVcfr=1QiNM=$z?5%c_LMd=3i8e(hb$m3^iZCtiOlOh##Ua8Q+>&kK7yQL z`rIOvu!ZtosYwqXwm~WA%xdJLcl?Mg?eyC3f3Re~e|gn@n zvE^QX2E~}8DJ--$TO3nqDu0`H`Oc#q4ar4I`3fQHmO`iUd5yZ{ibnA@|8g`* zsrXz){d)TdDHn21W4I~wC$D*8;!h^o^l=>zO!@>RqfUiAN*^5%e|iUb;RcLqcbx0z z1+eCJLvbZr>i^s7-;{A{XUoS$f0@loALVC+A{~ywu;}q|63*mF5&cO|%0Dqc*9#88 zT;=gna^yeX*r43WF4^U_9GPrDutUsFF=@JddYKq6l}G9#qd1AP|Ih8`x{AuqnBuyG zf(c=+b-QO*xFRkK9frY_bn$X;k>cl%P9es}&Hdm>ve|AZZSoE^IZ zR@?`WdIB@UuqVfTfYAG48NBT2pUcR7vGdDN*quQCIzizsplt5|15}P0Ksi6gPXlpA z2Ib(P|A|r9v4lMa!oYAlk^Yq+k@$kLfMmy?c>3kt30LxEcT>2n$scOuGF}$LA6iZD zn$%GU*Vb13%TsMBy!A#?VwU7kyqQ3GQ4Ag^1>8f*jS?uOoc+za$BdGE%9*Rxq6n%p zS!z;tHkF?ZDXYBbgx_Yf;R?L(3RnBBxlxxwBjuSl+Qn}d#d~|xdYEXaM=8)vDZKU8 zmen`6U7>Dlw*2Tpn`xV?U5W>F8tDd^ax@xHGG1y8J)^me4! z&(6=uKO(>*_c4xoa5mJHOllb#wN<1mJ9Hi^B?d%N2}=GMeC&xb<#aNHTjP?qae%Y) z^bUuDN5QL%804Gylq>8pmL)3vgS{&aLalzJ#A*^%IjJW$8n}(5qD2xDIrsB&`$wPJ zvZt@#zc{$t-cXxKnN~ZZP({*WD#k4-y6o!nw1UJK8~Zr|q3~@f;7uvPTqEcU8Yu~! zOXg_^FI%&XdQ2(Uv+bWkY`dmgB`#Q2xhLDAh31C%9$pwTf z{4rw=MFv%JaTag2i(1vo=Y50fqRyyZZ-ZQa2@?A-;7=aIy`OTw%hL~Q2Rt36t$fA2 zOdSR0`QGy=Ko+>UA@URT2|dnYzQ=Uhl(xd5bJFlB;GmnbABGPG@~WCtKG`Jmv{i0u zl+79vhKDci%k_a@vI04of`SaYkxc)8KS*_fjXV?>^5b^ZA^5Q2;ip)FR3{GD%kDCG zBNr_o6FMg@V***77e6|6#_$XhtmUv07UB9=_V)ooKLlSMPz8-6^Kupq2~LVn92a^&hoG9PNbcG0p3;}dxWVx z$$5Vy-hM|Cv_u1eW#`K9pi;%jtF)+hvAx5nds6k&MzF4;E9+pfPX$ER-xIp)y4XKPXxI1;e2 zzG3}+72~tcWQ`ZqP>NjknSISAdDP?1e4{JyX-fGxZbKTmF zbtxiJgzGI^-)Kr1mO|W?BCPy!1fo%I+S1CppI;su9=5UZNt-xv+Ri@noK2rTW35Kr z_L^(9ySXksh}~)RFBA@Cq|(#nFTl}LcEK3*$5muvmv zMhT8fuxR0l3&#M|wi=s!fmAAcESjSqGd||rMLr@gjHTiG?-!7B>akz)!jS|P^pw1< z^wD3t7JdY0EctsFm86CYy8XXAkq1x6QatB#Z^PaL&@k{M9zMWj0e6hFb2^b0?nF!c z94@#2a#%U8&^u5OH~gTDyeGaq8cX(!`sbM49GxI&IX{6~7h3;&dVA5^0p+4OWX zFX%lV*h?tdoQ@;;^s+^Lx?6Uy!onTs0bm4rC(3@_kL*0c>@Xt%hDwZ5*a1A~l_Bh%ykX==i;mdKD9>z%!0A1vAe9wx zzaZI%K~y3XAOmp9yP5Gx9H#96+*Rg@7hc!P0wG3@J>L9I(a&X9sVVR zxAPAOx`6PF{vM|UnV^%6$G56Gf9&1Oii62+3ZE3-jj}NGz(zUr6rD`K1!~W9T@W_x7P(2e(xV_c#wuhGS za#u104~~Gis=cRtN2J0UnJi=ym`rYposSLl!SMH{&w{hR^fN10rvg@9Z|awwn&Aq55Oa|CcGqbOqHi;~Gp zYMgoH3^7Tyxe`tYEn}Z&*ZYCz!1L+ojSSByGjTnyLT^W}x14|arKB#$J%A4eDZ3I{1)RcjmlB+N2!$2-IKXEn#*mM(2lz%s&+p z`5sizLArgY|DuC*A0#;`pzJO>vIYqMUqbm~Ce}Wv#mc`kI32|O9I|b{JuA8X%W;eF zgvnp2`GD@fY$h7Q_~qT=*m)qN6F&o1f14bO%EJWBi$>fv>N}94LCIssYk7IuZ;F_n zp7zJ6@fqNdXY$R9a)fagFW0a6trD6DP=MjJxU{5vNEwjws+6qV4kc6bg>6YW=I!k9 z(NR}e1_uYNCBCKFgI`4KfGagG->l8 zg8AhoTUcDTrIotx5)X}y*zprbY;1bmhK`@Gu_r!ho5Lrp-9KSnwK41N8x)TCLe(c& zDPn%3gwo;Vs`}c76jjdH?Mac@@hh$-#gpA}KJPofylyLN4Xd}>DjU~oQ8p<`Ox`M7 zo3qU_soQPULrZm=cePM<*hRcCzv^w^{q6wx@e9T0ZGO@m9vyR@u!tTdGflRRsec#E zypJd0ibI&tM=3Yyzw{BsmAjaz!$?5*;rKvEJc6!+an=(%>Ws`@7Dkd!jzR3Sl`r`# zAHFNAuuA`c^;8GYL2XY9wcYlnDo09(l;NG`g7VKIMeQfW6Mozva zah6wnMb2l7heigKbgw`8SeRS3xw}iYwzy_{T7+>ny|J+-an-`B&AIuyWL4?4;)w)b z1FWyD+S}yZNk-VMMq%>tL%9aLPE{|M z(T}O0Fd5r@r&yKH*QIh0;^ASI`}X$&<#7RdJ2Ew9pZ?_Y_VkmF*<%+k+LwOf$L#j) zo3?c8ri~2LtXkE$-yOTNKJJ6cBa0MT6rcxgv>X@YHAKvGbOY*2iL$zCrE6$87 zPjR2G+DG{&3Ca^ZRbu7W>n(**23PRq@Op8`CJ=V7W9sY>d>|;8w0M`QPxtml^p<_#8`>9i41Hr^Bk35J&^*$nXtbhhgTG!Tn2V0GS7!3A1>FBJ>x|5^av* zZZekyActpL#LzGC^L&ExE9@DSwxd>5`d_4*>&o-wyg)e%A_GgR19UF#o|*oo$T^1<(WMlmSLBX;DCPUiPR-uYw~7`k|r z%r04+spKp#6V67R9oG%f#XyOY22UP46V|9Vqy$&}7KvK5>Q5HdR@aSl!hEcQ^Ny7Y zACHhy+-lg`+Pd=GV}nCOHYG)x$G*HI#;(+=!d4U)#d}SP$@`Z-ve&=;x~^N90~_3S*4paXAWg8OUR_7ji+nL`%jsQ<}8Y9{#vhG?`A3^cPBj z`)hn~yZ=7`DIOT(NbZ%TxR~MN|04&7L*@l-#1i5b#y%=}iN&~Y1~At6SdJOIdEpXZ zj~Kv@;}Cvyrmx?2M9WbFUM}lX+gAFdEb(;0Dq76M*8^n{TXx%37gDO@4q%Q)@FG(_ zJhA7uji`a_#L;mnej_$Iszt^?pPyC#;F9F*H9oz3$*x|xX?v2>#`1#Iwa79p;C40} zQiSVLjuaK=y&_IF-A=$E5HdoXqLBVSR3x>I+SbkF@5fp+7O(?L zy)ZsfF9iFGs7c-d@7E5*jrc&vpQy-OC!dZf%sD~YAs3{1R6klGY=BCjhIJo#xI!#t zm>25c$x@Y_S0_fs1$kT`KlEMSHTLr4d_mPo;olcfn;}E8t8eh}o{H*|d(i1?j3?ae zgECD0epd;)Pxi`sM6T!TJ}eU9VKUa`ir-iH4+Xw!c^?MsD|V`w0PInj-mw@olRF^n z;m)-9cQ|^rCT9@~qTEp75%!4uv7kdq-ZbdKkLoP)7>ysDq|7@^-?in@7pBl@5K8Qc z%29dG2zWi>+=rU%b;}?od+3f`T zpfjui9l9uLgt%{qr~s#9z|%y`#LGmAot-nT`9k3Zc>FFQavtnLwVF-e-9UeK?p-ZURhm}a?CDT(-%$Trzu5pd2P)qwLw>g z`7AT%iAi@&6D~YZWL7uo_TKxK?b-()+VbL(E!>{7yEkWSW@gbAmR4rkf@6Nmymg`ysNZ#MkBJ|~Nzh-lD z3-;At|8={2`;PCH=k_BD!leatX>6^C$c$FKIB>X3h$g z9~@md2Fs8$QUc190Y116#Bx4|`X_!q2N!I!CCq$w*uSQxl7)Q6qQ~>+5mSAM4*8Gv z&oY2KQ0fN*@ejiX0i})nL>?Hcx!Hm|<3k-Hvo3E=K|1>9pu8)b`1b>79n~fU!dsXb z9n4~Om;FCZS^k%R2al5f3xRhCDpg34z2J^95a$Za7#&l<@h_2aZOESNiQEx8m5(Jp z2Ekw#1AOtOr%d0WR9Yd>DJ}hC6^cDS77)VDF)VRrRx}CW4WAgSMFVf8iJ2qsqfg2& zX)2Id1vhJXUwCCdp_f6X5-Di;G^M9`5IqkM##B`S?v8ZJ z^&Rq1LI-3?HJ&OS*$TP{#e~$wQzHW2Imp4|@ee3P%BCVSkc1-CL6oCVkO)is!IY6A z4LlEkmsb!uj~{8_>yIPl1d$H+)6c0C7K2yt&*=Wi5id+cL3`P;We2WD$nKmbU(QEr zbk;=Pmh!r26}A0ve?^MIn$nXZ*T>Fi9BUAvajnqE@PM&TH!?V6Yb&eX&V0F!9i^e6 zAypAOJZgWHt1YE&L&`0e)i)Yz{-vwUt)}(&R&8Q@%tj}s{nN~wQe6A{d#qpa>7ff! zRKEW0x9qFG{Z0G&pS)?`e)p2S`QcUj;MQ$hYc;IWTd|@3K^q;Lu%lB`Ha<40d=1&4 z@-!*;P_<(9tM6KE>8kB5-L(4L9ow3pwe8uDY-i?*beyO=* z>r#dmR@Yr|XVJOC?G93+xvzKwrI|rZB=~}MF<5+fnxkYg-#7fDMAQ9K!R;X_4aX+c8W_VSpyM>5b|^*r<|a__GGX(6-2kp;=k=IV^? zHfPKvJ6mkTNxm3M{L6mb6)l3a5R#l$s{{T-f-C#RM~3Xgu?ahS@q~5tsc%Z&yZ2O{ z-5$Sp33ZpvfLa7E=!5#po#K!dQyF7xF*GZ*Dk;Jt}V>Y+cTg3wEeli{1x=pmM6l*W(SnpcpXX&h%q*}V z`#@(DO%j}lFyPwDZZQ2$X;sxO1B{^x1Cq5)bHGvx(Uqs z+rx&CjPC;=r(Vfk(Lgr6+B%qNQn2d(P+-7k0t&_g(Nn=LhN&%@SrKmk>ViWZYn&tAH$I-oQ^Wca10(EQu;+0$N<)rC*8`w8?oW_&|F&9u^HhFe|4%C;%xMle{ab;p?96E@7q(gqELsd;Ee2 z1WGNJfibDreMq}Y&ICwA)I*NTK*^)K?QPC7##`zPl-%`OmomuYvAVkEyTt2K;&4OH zo8^>|cX+-lr06WE9)I`EOTODnlh6*dNe21?e!Ehpn21J4My)bD zBIQacsZPuCyszI-A54B?#p1(J4lPaH^a+g~q5Ah1^!WLNnkG<}Blxg52IVLJP*O7D zM$wR+^c5#erJc}6QPGImKBP^%QFk76+19pl){Udo^IyPcEs)I|nmh4xg z5dZ6c`-J+Nbg74O@B{OW7= z>dSxVbi-qV_SN6`n*Gv0`c+$5V*#+?a_b8$KZvEmf!JATmyd%LPyxK*ttlk$=WHZMfr z>0@IT7N89oYY8UZmg>S4Lm?f*^-0=Th&^8G&2`AJAuo9v>yf2Ec%XA5Zt;{Mc+6%o*_J=nqd|yho?9CR50G;=B5ybOj%Tp6zCXbBhR8qfHm{)kiseba+rQlXBr@U+_#`B2oh0*mr?&y zc-cSx1vo!`&^WDo9)kS4A4sYcobi63VZ{Fj9uD-C`-hg`J_*s)3J!8WwXd!clqW>r zY0_XsC~kJIh}&v%lY}NLSCW~~HR%XBR)Hdn={r4|AQ&V$<^U7Uw{k{!b$wk*fs{QB zjNCuW5eULjq!=t)E%BEEi4rg{Fz7qRypcaFMRSM`F38^%wqBWL=K7@Ea@|?m=B4DWtgPAg7B5Jt+?y>g6Lr#8>Gvz)I6}e^4PKr?=BYQTlgH&K zy-fHVqX5|%;)zYGxoI~RmhI9FEduBx338hNC94#GZt>ai^R7(5OoU9xeukL#VKPCn zK-m@@XBKIX?G(;iQkMO4_B7t&PX`uyj<3x6OGd+ z)Grodk=}kgD88sh%A0jb0qdzL&9?H=mh!7QZY`T7rFg5cV$HRAzXoq}W7T%l9*iwJ z?0l)+rE%G=^1s(VVqGIAtY_$?S?#FpR;I02$}S(T7#h(6t5)^THp>z>qo|D<>kTb@ z=Kbao&USCK+P0y#=S4UaXm*yTCdO=hc*w3?o3n4e@{Zp#U8z=GX-1KmU)k`ND0vd( z8x=klhQi=|PX2)JN-C@mxt3f~Z|=`nnaHRocGz4NJXityfO@`Had7$vLWI+YjvgIJ z$C&uEdK{6OU2ogbsc~D8a^KK`b!BeJ=7hsqNWAp7Vu8%q1-j$pKIP z*tv{3HOfG{F<>UR; zujfMsNN{8&7(IMn#H!waYMAwNScf7$tue9-svZmr9c#+^WoC#vYa-fk^2xV4Af zV%kY1Ae?j(Mm{oX{O^Ll>`G6giSo|765F=1)wVjQaW&(H`Y3))l@Dxa{A`LAY^cvS zmG^`Au%R>ZeqTT>Q9^zlcgLV3AvoU264KEQ1rNyK*&f6>PQi;mH&TF`2z$17AlO2D z2!1R$80WiVk?#_gq3{=7D4XED3a|QODIC9uQ-<&>V~U$R_PdFYk$-?{l)>%*ZGoZ) z@{S+C033rqH0f6g_y77Ae!<`A%b4h$f5zl@5cY>-;!g1k?9g>=2l3;=y}ZldqE`IN zxc@)k(e%&rR`5WMOH?^f=n9Ar@Db#IeIQnid-a_h{GpbvSrmagcg&|55 z$`)_Ua~G@MHX%CrqZBiQ4h|3dxz@N}xS~mxFLA}~5GakSYpYh5a?POb1GxUz8|>^d z845Kp^P(Mb`Pc>@!JuQ(H(e=a;N}H9rON}ze(o=>XhN6bUXjAm*ROO3`ZW>DuQDKt zKI`q5DH)0{D`}!);Ad>1yc(#xHE35fps&u&*?X_QZtuPNCwB9rt9EN<-roP{y1o0s zhj#PMtj$YtU0htYxp|IHtZLGYNy8uYiYF7^iWGV7GiC>vGo)T;ngrUaD_*`sQDR(c zD12pY#ooL+YnxK0Xagq2+%8$E@wSnlKi*WCn`)0{qi#KF8^u!UQe;Gvv8~2$D=L#J z$dy>Upw|{M&1@WKB!Ms2UBDtEtacUN2iM#H5Z`aG=ZpjKMtmj+cZq4#+Q$d%2&va6_AZzJO7>x$U8Y26vVV=R^f! zPcrjy08;uHZF~C6Df`6pPuTE4P35ZCg;V1;HB!-Ha8P~~$?ldNJvMDercT(^YuD`f zBahgzQzzABHthZHykWCLMd(dqs*baNx|IExZnt_#)Ud2ZYg%_qH&bnX%j-T;k1#b zEs%bZ&XMQ&IodSFvRII+O@z?-EhCD*CfEn~-DIA344CAk+BwxqBn#qqR= zEA;f>B@;Jc3yrnFjWwIsLN?=~cy-bWeAu9IoMw2qaO={`Zn|K&jn1jjLLd=I18@R& zMpzyXa-a8{Cr21*a7n;U5fQTRxgX*2hs*zW%nZ0iFIo@LKWqmTiH{2rE5*$`lplVF zLh_C=ZhxAr1~WkB^(8n!75~-wJiSsJz(e)V;SZp%EyA0GUowiFtFMHql_ zPg)-D{=ndIKiC(uu*%o%ov?#R288F~VF4u$$-Ect7k_`LIxUlGrv$0>@GQwiVF$|Y z1Kx*m%-~0c?gK>(@h!1*r50sqTPA}vCjA9v`1-r=+c$pixBV)fijf0p?Ajq6tDg&HXhpZn}*r6_Bl-r#cnWhuoqo0*-rFa79`+J{mw|H1$FAKG90 z>%U-EZ+s}Fq^7jH#Ano27V9=SIc}f)sFsxPLGhY8ddyCrK4H_7qc(Zuh&}fMpRw-IY1`@EwZ6Sg|7gdS z6e4y~_B2@Y6+3Pm*sjO@!YH^OUAcLSh1Tkvp(4)^E_Ck2K{J=tcd8?*su+Z zOt^AE!%(YL)!EQs&FhgnlUX>BM^}uc1faz86}|b!u3fru)4uxJhn_}x{3?%G*ip6z z2YT(?sblu!lNash|Mvghjy(Ci^=MM&JS<;T^S+qk#be=+5e)L>JI`JO{S*4RO?K{N zk2kqvoWs^bANTkMW&x>!t2dhI>;i2a^T9mR6pw_ z9~>RI`kD>i{X?sCt4>g2Ya^nS;?H?&)f1oezH#}clz}aK@rOQVqsLE5LF%(DP3+xL zSooZF_wbb3r`NWY?)VXkc755}cW&F2%O4r{PuH89HlW4Xv13R5x;kDAySunz3o8wq zU6FF2_F8RdVpm$D!(;Y|Cm;949oMt%qO?iDSZi`G^p=ebRqYYU=hWzs%}Qo?5puQO ziiHnl)Z&dr2fNSy$XM;br}fc*sE#a8eNwW!dgB8lBJy2TKCa={7pi@cC!WT*JgMwF zwfXS+P5bpfcwJQ)a+1vl0kWDOMV~r3X+QO2U$Wo*+8@}JTeCJG1(@C1(Z${`t`Ffy<~g=5XBBTi!bE$_)Dc(rjZ9% z`t?WlI1=HKQDimflCS&kD~E!6rL6jHDNe{!IJForhLz(SG$TA^$HCK}9^^?r3G*iv zaK!CYIbN9MJmqi?OW{C{8A)p?t~TS``yc(@zp=mf|N8%EOrZWz<>ALJc%(OdcG~`{ z|K`8b;#6as${HzUzBx6plLsJ@XIDGQ;^orM7n~^DnJ$?D2l&ZC+5BsW_{q&~J$_N- zsGo?zlMZ&FBMt$%e)#M>_PkLQnaMnZro%Jrj3yD%F~9WOOUd?d;QOWwN2TNnOoL{w zC{>~l5fTtG7lXP7#a6Z)P(YLYGp}$aKKS8+U%`HVoC1<($;Tm9?-9JsZ~lM_BSfpu#3I@pO;u-wEx2=!hpT%Cx`vQ!E*IVA&tH138}aNN%4? zabYs(xij+Ok8*QbibN=J%1T=bE$Hs8yL&u3pr8odPm_}j2%-ZN2WSGmTetAg`_SAs zX0FlEPzG>o5MFXX3EWKjL4p`v?Ev|7Sn_Ge2opu3fUJiAmdR?%IWmXY9`PyPAxv_Tr15vDNJkGs#ch;6xw{3iKz%EE}KCb*f zdH$k(`HNq&{;6^68LHS06ZJsNcKbAWsPp;<8`=%qTwSu6o44)OjagfiLcF?O_m|~3 zs=>m*ubV@zRZkpwpstwohlU5G#0>auIcMg0+&DTsAp z-&(NWdg~)wSKiALMV5zbqP$@5>+Q0~&K|dO=T6yI{=#3flh3_qd*a_MWtihBd^$Sh zC@H?dP|B+(=gJggT7EPuhyGAgbN`jM(xF$f8Sm*6TJ%#fQXKN*icNyhfko-{<6cFc z@o)trm8T52NMUDBa@VzM13i*E=$0L>e!@z$*-4v!|kR>`;>t`EpX~)>kCq%+LD#Jt$6M#LR+=J z&xQuH*qdqCM{`Ryzq%qNcE{=*i%?kxYCU#zbi_^_nYQz1&$yD}U#DaDP>OOxiaEQS zDApWX85^$J=%5s4xmipRhI+_La{1*klnoR^URI=uqP@KK#9vCVlwlsL$89cZ8fh>cZ1vx7$~hzK6XZRWcIhrd=wa$!pVUj6L4DQ*+59~1LKGed%*$h%JbyP6?rcMb;%rz9cQFn@Cq-NcX(u( zbwB-cIWE(|Pj-$HxH6Z18G>hL&cNw@tQhcAyJm2{^24a=Uv*&AUjBpMu)i;5xLdO9 z#|qRhSR}E?<~MSD)c(u=`tO<=+RILSPYM%WDKFe#O#k-KG*RN_Nk=X>T{1tBBF`+g zJU;mpc*#Tyd0BDwqdB7YFqMn?B<;|{C?9Ug4FiN=ON&EaP^WaesXONzW03Q8P`osb zz>gfb2`!%R$Ck+d%%?sVmWKpA|0hARKQvfdywzC=s3N>ta(z)9R8GQJF;l5R4--k$ zpPRmY>h4S*VIIyQclQs0=jRQ2;3uN=r=JEWAJ9`gkLxtp5%7D_BS-A`DNY9Bkj;Y- zoifT*HYvkF`UnXOQOa@X46=LKq7#!|2g$QSa)48WJ=Yh%aC< z=en>*XZPrC&*^~<8=*kMzSY zx%j&WbNmi3cP8*KcRTSu#QT1DKj@^Kes}!% z8M}S^uGKf{_S6#>ZE0o2u3Wil73FcWx#3?;;v)b<6?R;uU{tzoTM7tg<~SbFD}`gP zwQY+VP3x}qTHpAPwfehkyMNcVZr-s+7B_AC@*UfK_p-I$f6unA-L#$ScdUndW2OA; zb?r#eKr!Tzc${fQ>GiXy^^W3xQH_q`)xQ@+5;@814|R)%snbU>9p~ zY1TG1uW^e6MLd)zVV^0tc4=%lbzD=4wHa!fv~ES9?tb~o^on@#CIZ3 zpF3wGlSh4*F5ju^Gai|h$qfW?q(!*3>-r%&+W1$bP}(&C4V!Gqn>L`Qu^7{i4oY!)4&;g*|5 z_oP)fX04Zxo~WIBn_8IEx!-rs=4MxIPBQoCV;5|AY|Ix%Ta7i_YOVXnFt#Okd#yF= zs?S(+amCh_H*9Tv)2`oIve&Q7*o+kFR(r?hHa6{RZ(X;qzIoF=m|3&sx|G;=LNUm4}Vsh5^?$U}B;u%lRxg9WpgMR*^oXZ;u z)2_|VFWV=dKJOn7A=ik|nImI%=8-eDyt-x!GwPRG@E+A-wAvRhwJxo0_%1dZCw&7W zk{@*=Ef97jH!d&8lw_SY-P+MYPzwikw71lj$OLW5)rl+ z;87R2kuQ1*PZfA6k8&Y@zN;xBFQ@!t_SR2+$);F@=?@<02=GyT_#CV{g$8@351m1E zMNIMSRh)bs8GKRD-6ej023m`YirP3hlbU{NzdkMYZeP1*-}>X%R5os%Q9Ub+UExsp zC39L-eCaD+kvwzjj{1duq(vV5Xqd(vSL{g~zNxI9NA*>f9--hXb%oKlAZ6}jX24Fk z-`pVwV>o4{?)cH}qzXUe)a%g88ga978}&hbi@(x!Wm<9l#aZeDQxs0Ch{OPcKzzSo z{`aCzzwM>mBO=dZ_0av%{Y{2mv}ALm>==y_@$y@KprI z;)mV+X!Z_G_z6{?47Lv0Zy!?MTEog3ZkZ}e)Hi-ncg`cx!_#S~ME6eV9faiXh!gbo zO<2T%CRslQ6g-LKO!bfYcOICBU<5v}&4Hjg1P_y?Ux57;?kOIy!5Uwy>#_c7zxN03j-|Qcd@#x&iXjtxoRem^#CL|J3^cSzSZOtF zcBO7BQqanPf*IpOKyE9P3tqT`cPvIlqW~21K6VjlJbV?G{?e%@rwq5-h{h-zbfEsZ z+yt$D*fWpRP7Z{NHWpZ(TJa;UAGuI|BA(I_oj;;gnLt;*eSjr(;z!xSB17m(h9bL? zp!(kKs>rj?THM6Y)}ly)uq2t9U0Ji*(2$LcjVdC$trjmxRt6QnM+z1n)Yz5G?%2(l zb-Q|J&2G%D+sxXA?Rb%U?dDR$UViVEUAnuhyzfeh*|VWub_-3)Z_Uo0Ic_IT9JQg* z5oJMn(4wKaz2z?gtxGOB-^zlgt6S}$_#=ZAn;adqxH^roR0gJsmmfVN%hU-4_P(le z^54y{cpaf+c$<3~F?dq1C=A#gQUsn(0J)tGn)})?0Y{wj%@F`JufOrWUv)S(+;2~v zJ#MpF@ba=$e}A7}bNKADPuY><$E>P$sVV-J6zBK<C5Bc$nr1nAf3R#Xti}RD) zpOQd8!a~nc{^xWN2f!Y?{W$_=fb&!sCgl&MQThd@$QWDwHVe6Z;XoaUg_qmoq%tav zqKqZ}o>vdwkJK1EP-s)rdd@Py!Sfb2$HGqn8)>RcN)G|{P&WlKa~|<=)9uMjdGJL! zSOR&g49=wpE6Dbf%8xP}@O%hB&-%WC*TsPf$<7B=O6qRWh}3kl>wm(490ru_2f@SQ zkA!8aVG)P~yjh7K9bg~oC(8pd;w6OM-N83tzq_ElA4208b_9^2CNSG&s_H-mr+7J5 zxC!u`2O+!YnRhfu(1NbqBFG8Z3CIes$B|#==WgtZM+eRWowry)>!b-gafs!R0{4Z1 z;!~je1!)A8FMFtWq7cfypU5cl4CKFU99uuYYJeUHw+6P1yLw z$86=%6LxK)-!6~TZ0orT{wDL5cy6soslT~m>))EQh1c%b=C8eLtH1t+-Tc?zviYyQ zY3=Xaww=ohR=+ZDJ9ACzYMJj2acc#)NFYD#hHXj7=h`gZ$VOqMeo=mTBcI*wwRP?x zR$5wYth8FT(uf!9n1HE&mrX%dP%;n(9dG)h5X0I(j-q-Lv3E=F=y?$J{+y8Z7GIy_ z>3j4Y-+5#*?etH=F>_ze=s-Iu4?!1xxyQKj0sn-4+lMO%NkKz_5jvm42cWV(P?Uhd%b5$7%@dDTJDST*T+&^#JOtNBKed^48HZVAB{aUQ_ zRBE=W-yD@_XmPWq#UHXiI5?oR)3FHEHT}u!MRiBp zs)$}!n6w#1r9NCyR$a5>i=1bfK-u*tma;>GVucCNo;-!fw?VjHFn%C}LKo-IGu+YQ zY(RPEF7W&^>qpmbsN8Wg%F@cZl;s`&!sPAi*QFF!{Z<&h0>}+3EdKndiI)voQodoV zywdL#K!!bkDW|H3kiT-C^J2m2`Jd=rmQWI9r>yuZs!ZnQ@IxM{=Op>Aq>+9BZq4b^ zJ}STXz}KIWK|;N|%{e&c|KT6te&j9Xy)gMl9>|=)WnH|(2xXP01>5S+v_5X~LriwV zp@CKGpu#`ZPq0sZG8M1L6aT~rUS4~orC+W|O60%93%`h&{M;UXzNkeXNNx}Dw1i?D zZvD?~is5v!Qx4CcQx*L^;&=RA9&%U$>dEWs`vlQi4;d&gMc!BJsH2YoB|<74!>w8T zGn5*{{~rRK$ZpC1J6(yO%qQH3Lxk-wll*c#=YbGrK?#ce5b5L&^guYkz*AvB0X?}$ zBV$kmjAUCD0jx}pZOOccjkQ%i?ECg zhF;#XKV&8MK`ud{zf{NWOmsQUK8$)OupEZlSsWJEy$;7n_FhzA$DMxo5h?kFS%MTN ztjXqij`F!*_Af!F{JH%E>XaySZIo0wnN%3CRu&dq=|sVa;er>}dVS}Y0UX7T;}Ix% z$S#lbnIJhnu&B7>eJ%!I&WKRaU*4S5=pGRqg`XloP*6gK3EtFZb8N9^1)&)GAd{sDXTQ=hc4 ziQ^KOs`YhOti}@s$<4@6j~$;Hu}9AwwTn`KFC05&6CdS2uW)(d9C%iY}CqSS(Oy)D`XQ&5^3eCByrH?+Amg3`;T6FBIoc zgq?=A7mc@-;?vjgFRC*^v;oOQ0Pbw3Zx2>E29$R>BYb(aZp$l6R zX~X`o;rnxDc;R-cWGc<8*DH<+ba#Q_rXjsCpv+;2eu*Z)DT+ES1j+50CP9i@ObRKO z9iQMEBo4AIbJ`1EnL-XfZ6FM8aIk3J^lc-?01-mv%Hdry4$f>^gr3|0O5Clor) zE}~FZ`nqhqzZzGtDXqckpp7-mj*U-R_t=0f-kG(D(GfdxWXk#nhV0mdC#+k_)TT9T z`^*J9`4|7H^}qC_)oU|0x3z5c%p=ws7w_7Rt#)l&@6>?p3`s#B?zQfbs=YV6Vs|$- z?7EcY<<$*aURd=Pq}bg)b9~y4O;7rp^1M{ol=8kM^fQsFWxsN(y=!;YH*J2a%dcq5 zP19+k!|lH7)zf8XkB$2o(np_p#->hR@Ns|z3SX#1{z485rY@uGP9h85x2Yr#3c8>T z)2T4=m#=?hL-}SR3xCpJx3V|lfl5Dy2>oXeND6}cX9WO$2@BZgCT*n$y%tE^R>4AR zZ+*pj>x=T=wr0I4Ww_-E>e%?O^;K&&AVs*J^Vw1myH&$Go7{IDkL*z@+k2Affl2AV z?A9s;mFMi~RmcwGEqHZgQf-25US5=aIOmT?tZ=)SqHONqJe!MctvaQ(JKd zvCkHoPpD&#>?vF{+;7~ae#!X8F@bF@K3TXiF0h#7_{OCxx4ex=t6Zo4|A3pcBTteU z{f>W@rNc!!i%`MV7FG7m!QoIJl#WkDA} z2yWyOBkWGBKMzgmxxWJmCOh;vMbECY(~?(tA^!o19-i39A9+gQZn{0_eC|no&5uCP zz_=8zqVeI@H9l$ms=w$)95xkvmtN74*}>r<`;o7F#kfs`v-_?Kc+x39o{Z9ui) z`r_|dEb@a}9_|d9Zn-H})-TAT$HfVHC8q8xng}PK1Y?&;Ab&C`C%mE^GrbeZ7IlQv z{b23wncYE)HU4!Vk0y_vIOFESg71qmyt;5p|D!@MDY{oju9I9zgvdg89^k_}{0kak zj!K90fJEqWp_JS5grhN5%Gdw4v+jqB&^v#*~6C1fPMa%w)|gy z4@M^|++GJV0wWyX!1QyveLni_4|gHfF9f;S?dlr|`}k@JDJ<7PqQ0FN}_ zxIKlJ?S&sC?(fzEjuOZ($H@w;JUtDp?j}rF&|xOI$>BMyK!y3PwnMmsKH`AxUvlSo z``jM`fsVZ#hJVyoA?4XnAT{q%+zk0ljxYtm(`~8ibok@;flP6+@~PU5wG~@mUA8;7 zZuu_nts6IOfln1{;AVGdZeiXUQdEbzMM26!ZE(O9gSr$%6p=9v=(I6cl=WA7eOH$c zOI-Qzs=f5-7nN3@z4M*-?9`c4)?DX|a@(p>va17C+ibL@FmBr;=gw-v-?ht^F4?Wy zcYU|-$)_$_ce`QZnzVbli~z_{TFb^ZS~mU6MeC80 zbLF)+Y+-rTo_y|k+wSSNuKr;=^YkaBL`v4B>@4nW+U}X7cJ8Ns-j09zX&btD!j=ar zHum@htDKm&-Xo*7H8yI$_U?x^FXf9{m+zuPi00_@gny#;S-H0&V-?GCSXmr z%x7xZPFxoWc7r3X`{l53gx$?f9|8O}17rhAG#LlN^&fsx4B_Y~I?RZfK!lVpmk_>A z=e#@>xrae;_=g$WptEZ)9x%?5W?t&dxbMZ?u(d|JDIu;1#NJGMo=Y7xg#iOojc zdb?1@q`>krmfA=fv)?(~ZZ)kX#d^C{_p*#mAG4`b=j@S3AGgoF_?&&>{Be8o!Xx&? z`P267W(N$DP((+ zU*+l4*$ehVKlhhx=;%p*iIzpP->9NI`sl<>CSE4^yTr&}c30A3oFPB>J9-{u0cH0* zdR%eCEpHFwpPSIApw7r$LE~v*%d(~VQ(ZFIw`0Aq0?iwKZK`ayuW zz)Swl(Y;Zeqp#+=^|alsf2m;E4|#`nmb1avKyijA^ySOKz4FE560k& z9TqanKP3~7G~$>+ zU6|n)Sfo`#{VVW*^W#BEiwKyE3AC`I?77@QN1ezTF=_B*DMJw|2D>?p#k8VA*TZk<;*4RVW$;^ZF|$!r4Ua{PDlZ0 z*+F2j&-nB*Z-aB3 zf^)P06DSJf0N0UKrAUjGGsQf$?9n7kJl;ItkPo_ykRejWQ^Kp5p_!9lq>EA2Nx$PeX|$&TpIb$EuGj3N=f zYbDyUY-A|>l03QFdBx-(a^nO_Gty-eg%BGlX4!RG(#d=TIbUSQGnD+o41SFEd^yiw zY?Axt%$&_Ft@#dVpX6^q$^v=l>Ff3{zui(@-oCr&Uwo4ab&qZ zP~*kNicMArZAc2i*x-mwj~|ngJ8b>7X%*Gopz3t2Rc0!m;4}S9OdAukj_0xT0MeAvbKR+eHT$h3(1usmeZrwNW@|#dfeB&iiFy*%d^x2?Q{TFGZc5>PX z!;zZ)9*F3ak5XdIWexK%fOXD_=cy=5`0^h<(WjHXCC@w_c|T9~NJ9S!4_g2oPXGnH zgJ=BfnBoD^0U!>)?>D3tdALC3Q01Uk0S$wvkV@q3T(>0f(|1v$UZI519sHAu8hAiP zn;;1x45ZQFK+HfFGCCBEmj-KBilEafm`0(=yxEi%N+oIejxmj?PVH~BVoK;q&(GlT z6JG{{P#U8M{^%;D9fI3wcBnFCCq@h=_rpw+Cnu+;%;g{cbT%)eI+olFMG;>yR4B$Y zk$M|wJa)|CB9MP{$bBe3KLaG*eit;!yD4VmNA_fO-j1FE)0FB(JUqzri&9FxhToy6 zbc*^+ZFLxChz^UHJOHn!K(3#XHsJa79FS3Ap?qpm@ia@w#rRx9CyXoo5)Z;sIPOeK{(NQc{_FzEPko?^F5%@4>E+y*VtARI z#_KucJTCfI+p0nRv9IEARt|Gpe3~0gyM5)R4GkZ$w%nY3+>|0bq)D=i9V+_YPSYo2 z>3ou#IBVCIGTD@(yp7pLnUShFGHRRc9X|)!)hlIqpk~eOwq2XK<-5k4Qn&|)hQw=2 z`R}so-nOl5EQp7cMDgPD%-rogFf?p4ON%xmrKeZ)o4al6+1>K1)@Xm)sgKViOG)hJ zb_bu@T#?NQwGr-gaJn!=|TB+1Q2WZEaR#PM9Pr=zq<4ma( zy55SE1hp$i7`l{az7D%0MdpLqb>~r*-TOhN$qFW)N~KMA+iL6HYUEV{;0s%|F@1}E zNw4#-+$o(3kEorCY;t9+h& zscEm?YT5N>m$g(jcCtr?rL0p`QhfX|vRZ;(NmZdJ`5mmPd$p)Mcv`Th#TYN5#ttmi z71xYyd0hWmpMF~!p!sZgyRi{RB$)4#ep&15w~3K4@53PA=k#UIa)Z1fmt6aocZiW` z{JCPzU(A$T9X&GQFUWCze|l_4^*(4*M^D(`^tf%SeEtb`+Ke$sWvTF0J(+%V#3vPi z@+!Z|Bh8503zqUn4&gy9AU8fF5uU)u(x6H|g_EY3;ZK=j9LGg1sVbrNG?;6A_VSQC z|B8W+p;_j9UZHLjn8(N@d65_W5e3L~O25M9c_Casp3wM9c*#F6Hpv@vCVYFvAA0U_ zh6nTD*5^7dbaAd<>5_)*5l?<3=lN1uy?w+SIfoWm z7y2y~*oY_EqC6SI-?L4)@L|KlPw@ay9WHFNio<~lx(B>jz6bc;sQ1%%>_ztQbBFTb zMhZvXKt8@lj4904?zl3|LGbVtnA6IT(>xU9O&!2ED4dhq_CpMKK+ZoVt;qRL z&NFWDmi+39jgF7`8|uxKCHs?C{=_z0O&_?q=aU_*dNaP1#!F1>g3>09R$LK=or%_W zwuDT$@NRJZk`y;4x}jQCljM%SRE9FmM9R5icGK4DOLVN8@2P# zyeMV0YJd34|D%2DfB1WL?OVTXumABkRDdo2@|fQepggzMW^HY0#%5*~?Hez@Z?9gv zWkX73T%!Eq*nmBI_PC8sj=IuC-LRv^tq)|Mx+PC7R+`G+o)!<^xwY(yve$iipf~_I zyiDV6nfayxCKJw#=7n1*U!qe!P&~!L0a|bS>~R2cIvJ^0H$6<~wgcdv6f(z?HO0$u z0rv-GkK{Xf2supdVoVXZSMWg*c=`Pswo;dJsBr`ZxuV5RcfXY5YQ?VHxMP3t%G-8J z3h}OFad>b{3l5X4iG^gcOAEQ)){0hni;}xtJMqMmcI4?7t!H4+>a$mEUh;7H-COpq z6z+Gg&f42o@7imZZ`wy%c+IajeaCrcw`CLKH9ImsXy;B%+r^_3c6xl&9=&+lnka%w z>PW5EUc7L`jtq||-#xxtOEExXp0c0}$BwU*PbsvV7mZ!;XfF`ii5+4o!rUw%{=N&W zI4$)71UY8H`HyhzC=zf5q{9c!o$@wEFLLljs5{FJnu%9E!SQ^n^Y3 zFkNwVci4j{Lm=ryYL>fB;k$w~<#e;qy;w4k`#K{BfewvgHzmk8a z?y{RQhd*gX7iQHTP>?pwE&Rvy0o-rckixRrBbtN zZ@ytyUwzFMuHCfxJM(t;)*ZX^(KUPL&G+rbN4M?S!J0i@>$7K19J8}eoU^gB$5g(4 z)lIJ}GJgFT%8txU-?deKw)gx>u2*j^+j~mKCp$s3U+SO-z}qdEv==5%C`P5Yj4D08 zJa+Ec=WOcaX(>lizS$*5{uPH!8wc8%c+`!{MeuWkGSTHRg7Cx47;i#W@QnD_NiW<$ z@Cjxp>UmOglwXeL6p0Te5%JIc2V?|eL3a7Qx9ktT^_rd30_M?UT83!x#8>0kyJrWA z4{0o|t^12!moMM4%U9;@+DEgtaC_O77dHGn@5=m=E#JCri??pr!u-5{#cpkO#_nAH z&_292V;|gEv>R*7{^{_>@~Z7g9w)W9nbu-&O7e43I6g9J&q#L9ec~xUkG!?31<};B zbt%r%r;pi7PoJ|RlLJ;kCViJ%WmX+PTTIQC$62-`uwMKe`ch>eUc-i1t^$>?b8=dOEcrX zZ?m&UM(yath@Fv1OK{uC)Wn1hanHB%^!DYe>i+}w%!{9KxlYoX(xR=LjsKIc{<*8j z3vO)gRv*MH#&O!5u_(KRfX_SpxikEI{v<5Z5`wAt_(7bE#FH`H9bQ-N2gsc<^Koyd zl|>bn(76|2$MG-A%R?{N=>W7!@Rn834zVd^V-T{pSLyP!| zec{J`-2Bt(_&G!a=4SMGsVGXt|9c+v&!4`^9{Gp>4}&uMV3;~C^dpTgmzW~yAOm0E&$f*C`$qvUN1LU;C z<6h_ldfpy{W&Y$=ei^VH!~)y{Gg&jAvRjtIk3ql(ie$Py4^lD67HsiL^%0Qmm}TU@ zOpm0lOFVaWcq)|`q)J$TTlvRC&7)Y(A8xuLOg-~5-2B{(wYQu0#;Y&e^3uF3zZ*>{ z%2IS`iZ)l3$-=$9ynsVEAEg+P!po70R;%r2HhG!N-=Nn7fJrv`IoGJT{c@ueu$#;W zDH>8f*zp<|7?hF+ofN*YaqIUFD0JJ>%90HXR_xJ>7wr1oIakW71H3@DW3@_e+9?xC zfH5hdbZRl`wZ4<%Huc%(tT{EHiRg|k-I=qUg%ulk<+}C%<3F;VfB6kNb?Jr;Ng*2P z?zcfj+*BDFe4%V^*}8AdTlYsZ_WVw_{nX%?J*TwKK61p4K6lb4j*r-g+`Zj>zRR6I zyul>mXPM>VF>P17W%C)~cP7 zdVNNDdTO9*pBf&pmoA>RANpT?#eVW{{S|xaOE21_7CvV_{RumI@tmD_@}dnNJ7Qgm z`?+T>+Gk&Q!q2kz?e!^NgErjD3v(44l;0qC@mBhzMDipgKF7^{*&ugmBQv}dh`pr+ z6HiB2P_b+5xY@K;qi*ZVEB=C@v|{T*hLl&oUsv@+O-@a66&#=c?$jwJinfmn)TQbX zxq<)q|!`K04_`%GCITJ#qe=O$-m&O1)uMv=~IDhK7bcE=?1SlUp#r zWr&}f8t0T&V5IHGa&V`-$cS(y8h(tA5g$7QjAcH4hByQe5ku(loDMg?1oDGVhK@}@ zm>Vgs&=(z!K>}nixATvJ3t8XL z5~$5Irb3r@dwpRk^hK28j`xvV|B;Wts9P|Sj6NC}DCix7Cfv!AY#{aGWhfx*D!rqJ ziJ{XwU58&J6mB?{!}r-C%l|!$bffNq?%_g=$`2hn4O8e!sWz}9cl&|rFTuS8-xJ(Z z&xs!mm-BHiQ9zM$$}6xhJ036MxJgin~AuxwKrpyGm= zzy9`>6F@oTMLGXGsua;}DY)#=LL-OGUi--2dHe0472&^Uw0Su+1HhRo5?QQslvriHQbaZM6Ew$7v2?LC{GWJg#^fD}qTSioaM zKI=+;YqhGceD8Ef9JU3Qaaz*_= zNC`d&xvybh^E-EC(jGi%@CH@@ex9xfa~zJJ5KbP-kaFSwv3F#}U5ZnzY902{0l!|XTR_TJNswvO7y?q?cxm1A|n+(%B2<#G4ZCL@6e ziR$6WC$o~f9PeR5w90<~;BA>Jqy#Qd$=@p|+r#+N!+k581Of|z%jn_z!Sj4rKn4+z z(!HtC5f4YQc0bHCbU?=ph1-e3bY4Q7hl0ZBz|?UfN+nO7A#ow$6ar>p&8yJ7`p>&T zj?nr<7#R5o`VOW%-v?uGQIegUI7XfsWTA5ceBCUO-GdPaOJI=?=T}1LGbGx>a9><` zWq%*U3ru-T{$)itqt~?$rlqj#F8S%-A=t4h8=9;`dB|b>E2jzYae}qw#y27uf1+>zVauwFgIuQx)f6Xb;r)DCUd0E3N9+qIe5 zIWO0xYqwP|J+3H2?=q%vlw}^BtF(TOQ<{gr7}v8UMdjFpjePDIt6e;1&Hiqi>)Nzg zDHto=JGRoJvhfz;{D$=}Y}n-Tri}<|%^e$7S(;KN*IHZF=0=Ir<2E$FjRw8el2XK? zVta2}2BO5Y6IcNG@?9>_ z-GP+dcl_0VP##t{n)c8B-~+o|@A0ps?X($Fx@~-`v+`$;iioqnY8K8e$kHq;1})mXI`>LfB47j^w0iPoBlI@ z$;N*C=WO(gKWc+dzhFJX$J9<*Bo0p5)Ty)9wcBmg?QT0Eh4+;8*|-$*it^A}U$;KB z&Dhw8J#+D#J@fRVcIx<18{z6ff85RDa7Sa}uCm9f?(4Qk&YYx59Z-}xXU~lqG(j{&mz|v&u$LY`V-rWF z?T3Eyr|fV3;@`Hv{v_0Qw!ym5IJQ5nwL|VPlkJ2Gy#oB z{ygoFgH9}q*xDN4a718$Flyry{NaUT2UOLVTNqWN*O-;eok9*mw)g|razugUPG}h z?Z!&Mh|bVas3M*~btJ2sbmD#C2QWkMiZt9jfKH!ivBeL22~%DRMG>ClTZ%~JQ&{@t zn4XWE#{CdV=Ox4xuXuD*#!R2Wil3*VUx~)^=*g;{qQb&2)v-(<;!;1sGuKZ>^1@@z zeRU;21-Pj8b?@0=ciU?98LKQ@wcb|4Zyw;t1XBptrukdKQk*y^J2*USW0TWX8yK|a z#)iH6hu^SQzx}dZe*cnvaOsL)?2m%Vo8|20a-|w~2D6LBkqLjhzhBCrCiAA!LgD4y zFmEiA*QVmt8x7mosQY+u)TwzgV+ zd}331>Cwc#vb^G7$hvavuHBkhuqi1YwVD(i)f^YfP$&MO2Dwqp`}^W$tM)dZ_uaPs z)1x;0g=ek$W6xXr{G{!$1E#R~_O7k&?Aa})Gq<&4*QAJi&~Do;DIix`O}lGb)+Ht4 z$P>q`diI$0ReP+DYE@mdrEFKE6!XCkUgqN_4R+F+D*Jk4)2`32+3&x5L-oi`H;an& zVpB5j2ef%MlQnLVkjW&wr?G%C`q^~kLTSh3D>o7-KeUFBGx1P7{wpMz{7U>2B1?Hn zB_Kw;Kxo~K%n}KDPm+J3SfrSNiQf6UUmkmdM?t6fK90J7G6h~7pCF>2{MM`Q+dulv z*KBrGi?Y2Qm&J-?a&)*~3h{uAY2nnPHtRcj+?C(qO3iu~7p+&a)ir+9y2g%M_wh5f zH8O3@z6sl&IAdK$&f3o8q;*w9!#%|+$H>_W);}_4J)3p^-1fNIcJAtRTYvLSJG!=J zd$Wst3enn9R5waR#6lH3TA1{=8P0_byve;ztin72$*ZV#G#8+ST1*!TC4RQ9wlueAVb1qJD6Ruuh z&p-Z{{U={|+P?6Kb9PP&^D|FAVnG0t zibm~0+lrnxXD;Un7VX>CqJulEQHaYg)#d8Lq+VmuNGK3_qHkpIl-*Aj&lGm$4RJ;oK%Q>WyoiLGccqCub_=O5r~<0Q=!WU;prdwb?^w0zo-$SQ7awraiECs z@S#(a4-qJ?7r9jKXr3a(4Htkkzd5O*%JOXpRz!0XxF*x4`#Ty}YQ$8J#OWqM5fWBiK29UZqCYp#E zzMhxDPbO~t1EYhdUn)D{xeX4(oQPZ9Ocp`r>vpFTDjcuF(3xaRd@~euxpz~R^n;ep zypIlas)X_=ejY!dV{w4plZSiaWv(?}Hwz8?*_V!v!%4tK2DLcG_Lwq5w z>h@b7U6msI$M)9iZ`!r%*KKZo&UbUkMpH^1is+D(3`CjpzdY*gmSVc0FqAA5GIqr% z69c;sh8pzS?WXT8(ayX)$F3JIdiD2d!0+wz^Tn$xtFm+b(w>yzA%Am!Y;@Qa#`^k( z?X;yVD7~fmMSq+9_{meYAcc5tWzFw--Ig*^sY;<%9ypc}Z`w<-P+nOWv7qTyz4a@9 zC=4yGZ0p*w>X9+)ec>@{zjV>IK5^2vADgnqi6L7X=(8ooSys6k)n3~g>$mD7lXmij zi#GX*i?(y$s4HuL9F(N1AKxP6H%c568|i_@pCdaY2y$^ zdZveXSm$a<6w}W1miak&c_STS2mfH_{RB6NY`duipV~4OeLQ90tA6RJ1jk)epl^S0 z)Bb@J;?=GJ>)O%cOLD|x{gJ^wE$q5|klQo-vop0G2 zAKtV@m7n85&2~%lu`9o@u%PmBxAvw#<*2LAAtRg@jwc17kRUV6YaE?u^QrDllv~NV zms83zN|X5e2R5`Q<8EsBQWuQ<)ID`h{UQr~bA~xNtVAUSV8W*xYK`O0(A< zd*Vs^%qO3>iWa|{UAuPfsi&=~aiKOaU_G9nU0c*blRKG*h6cSK4l0aKjB~9XN-`~= zjEShWO^^Tp|MW>jK~yN>av?mqCte&s|B+!*i}Au{ApL7Dygt)=N;Q*GeP5}>pRYwbg@T# z%7DzG&Qd17(#^+O)w6ieF3?lwj`)dJ(a)$HTE=OU>DE+MS)p@3(xx`t9rTok#lEBD z#;=-G-SO4Ucp@0clLRnd+X!@YA?gLrGL?6b&wRV-iv#+78Up~T1=XMY2^D<_8G)}V_H*ohyHWIi5bh50ay9fM9z ziWE{l!bp|$?+j6vgFxvH{Yj|rhaSu8Fd*9x`W12L80>_52O)k5OL01o{V)TapX#eX zC&^|9BHafeG=WYJI=}s8%=8&LaU*j4)P~{gpz}(!DX)n^m?g+*1Tr+Let20;6FPUB z_Elt;qIP7fZkz9a#~OEUShL>niIyF*zW$m9;ij!F&e(_VzG;`y|BR z0%tO22bjr>iIzc$xD0Uqsb+~4UwGrKcV*P7{ncWS8UBFZ%Fwb9FfATNkgNi-C0`o zk1??0&4(3Gu8|e`KZ9zXVCm*rF7az6J z&pvL`pMK27UN~=~pEz%|i^r{de9-!8Qkb;hfHw*^^~2=IwO0MT6)D1P<-2Yx^)=hs z+O*mAroD1?)?b49_W_iPHfN`wJg_L)k|Ncuw0rydT=}P-V-XPOawJ;sGA>uqce1Qe z$C%X9{$KJ@TOWo9K}1g`JR}P~iOHQ2f5MT{pocCn_k(c9gh<(3;UfS+6Ak*8IJ=U- z3tvrX6t7&qXi$%W{54u4I33P>OR`e2`>M^01|Pk;ZB}*RGn)-?r}7s?v%@ z5g(b@7W(o4az2cg#NrDa=N-H0b4TF@!$UIWozIb$v*SPMtWsUePTq_G`Y zbt*4Ab&_`-L03AEJ^A@u03!kO3LkIDM=3^!;w8FhW6bpcwy-JR-Bf~C-^7Fc*{3^7k=jDky^}In3kCb2cS6oW(C_8>(XSY}J89%Y;-?rMH z_V)TBp3-Ez%kjFnKi<>pl+)`?c5qMe)%cz-SNKsj)lM!aDrYViN;qfSd0LUjj_7No z2}XH1Urrw4noVJL!vnXF{CoS6M_QctUf)r#K3{tp_$4=pJt6s^?Hx<)v0pv{A2z63 zzn_qLG?07uJs|%BK?qUVEJOAw6p8}wr9B)D#=94?kh*;z>i@yKN-h+8+{M6lC`O%?@i4#aPv?g((*b)4Vb7(LqX2m1fg=zm zh22TQl#h4TvgvludtkxGe?-L{1J=QG$_Q~N91ME7yeWMeCO3=I<~S*S@(=nHN8nCD zdGaejfebV#IDAxPQ}Or;+m;5Ft{TeGt_^GAuU~r4zbG{{F`)sj+uEBAo4s}0ZeIDw zUisFy?d`YT)&$=0x7JaF$#+GA6n%?Fcf8%sP8W9y=W}jZ`OWtZw=@}}gnJqExh21%jp+0=j8vw{%HD{`udM-!3n zuE`!|iYL^c*VjGB$<=MJ^!* z+m?RoS8Vg$dpXEB3_sC+(9je$J}Hlm3L{ z+KpTG-t4S3`qUUwn7Mm+dzU-5l_#ao0&9M8(Qe!bsZClNT1Gal|2`$M<{$(q!IGRf*m* zUZ=EkEH+S%8Ea{C&uc25=RK9h6-VswB}{4ZsrvwLK#;#mf6Z^|;XMuSze!$b18*B8 z272Cd`zYKM(oE+_U5iHPyD7KVwbEh1&3NZwo(7WyEAbVX_S!a(_{)hZf$RiuOTm=Aw#R~i7m~CnP^I)K?wAx$UPR*70H3s!7K1&L6G}EO7|pND`!4v7^1ZOJ z<1fqsd2IS#;b1)4VRMUX?4n;xPK?{g*qBc)73GaO=0(JeOrqSq+?hyutl2S{R0*d& zLm6Q5Vq#+=3O6zr@`{=5Lg;*wBd(MhP0p^|6bggFd?E~Gf{EQH&qCqp*u4*{Jki}G z=Q3He`65pWA4?nB(`i&Ee0_~FEw4B1Ywz6A0@t*==3wLiAj``@tDD{pDxFlaq3$=cF}b!kB}A?^tvXvG;Dy*qY>PS6zw)Pj|QHOzqG8&e2~{O0VC%?MHfelEIl~+L2>G)G-1{tGa9^ zJ2I33H!}$JM;yuzUsqndZWP9U*&#-W%Ewh4RX4uN>uGglh&Wq&-F90GwA*t{^?51f z%PZoe{w-RL*{H=OQ^TGHe+pDT-HE=c{JOj&PksGbyz(>%Wd;Q)OvV6LnAC?^7~^(7 zDT;(=g|bkLF;ke!NFg^e#nVv=Fv_R%K-m;#1q-<;E6>QE#~1p03&>l+M`@Mh4Y%(^ zLPxzK&lDSz@h*o6bNRdsj`Bw_7k`&)xgC}7C~rcy!-sm4QYm?K8K6ysEbORPCF*HV z}oKH$o_r$=2y`uW|VP`;yV(SIX9B#{@aQ6AzFMjjO}p)7t6?Ao`eF>KB8B*Dbplr#! z%Wc_#PF@0V=343`8F4eMhkJ1($-?jra78!=gNaEwo#f@Na*m+s(8s?lOUQGNl(+#6 zVtkr*ufA$MGap#%qjzm*ean^_J2tn~W2?iXHhA)s%`C3it(kedb?3G%E-krI8)q6d zY45Z(OH09(oi9`^tSrX_tU-2gbj;43yWnRR`9f8_(R2ljFD9`Q`_!WsZFQqz*Q5xr zL-EL&)BZx)^2(YmEv?w&Pd;fkZ{N1%`FZ>N=YP=ZQWF00|L{wa%N;v)?zEjgGHw0U zK|6YM+D3+l?VWetwU>V2MO#=}u({b;`@OG!!(M&+eS7Y)^R~P&?@H?EP}PQOefE_f zc+M`KIpK;2lM8Kv^1~gkyr4&2p&T%w5{~k(naFpY`5;0Zd62@Q`b71Ly5JUyEwzK{ zekWR_B{wfjt*vkPq(3mgcYGO>%`uyLkokX%;fg_(*W`H zC*_>Q$hJLpVoZ~6kDa;jg#G;A`b9hT*rQT{wtc47yeyxV>T3-Yy%gjat{pgwGW_ zEiLpmm8Tt4L@nC<77%e!eRb3B+bs?F68kRstId1;Q^#6x$zESvRlCmH^=r3mX}xYO z(XH=nT4P)Goh>c&+A7wr4QLUfl6n4jceyus+iy%^;l#^xsyIvxW>9_vcNpFV1TCW{BI%dD{zy0g>!snh-x)nP!eZ+q6cfM-B{6GFf>s9wS zaYPFSEgIjwwqR>oY;q*z^I!P9{gq$*&umkRQ(D_!0uw)!H~JIxjm*TZvO-YO>5uA* zfXjmDJWpgxBpkzVam<3Cc%d9YLx4bGK7J-}rC)yhT;>um$N4-bFMP{!NW@_&@(2ZD?qyls-n}KLcS=M&!;rg|f`mY+;AEs4K58DB*!EaiY8! zq~mGf1%32Y#!b$xd)p|5i8I` zfI2Fd72<4vpfW57!Hs&(?z|&$D5SwH3w*>2KT-%W%qeC64it6nKz@NzcLXrh*bi6Q7Cf}cMl-=3brl7CJs^F0j6BXqn>hKbuJ zM8)gdYT9r`6Ql@bJTCmj*UbV{1imgf+zODIwwX4tGTw1daSL!x2ILwOAcQ?s4 zA23+o*sywi&3dIA@Cjntetvn$#)k&|PT%qAX?yIkC;jpN(&B<&xi&sN?l0l-HLK^J zdCKn2E!wQGSCik%o# zC#NT+%njMO3+MgziR;&I*iZiCPuN>;yUNSbnN<9_vlB6&X)Sswk9R~ z+1+NG*yKtz(Qj_GY-6+GyOhI2!~V!ynX%RNb-OL4X?de*uijX<>x*kirec84ZTIg2 zt|*EH$2}gEGF*{jG%|hMe(taT4VyfB+IE!JUbXR_773(9YMx?>2T*T5wxu!A{rRUn z#xa}{svg*tWG5CzaCDmB>L9&LWup%SKTMPaczSsg51#~`BqskNJOMV8BA;j#hw*Y- z;Z@1e(W6K0<*$F;e)U(ss=P3cGGWT_v56Z}`t2uv_UG&?QtGPn@7jsVmh}%#NNKED zuiCyx3y9sNP4$sJ+v(F{NQ!IE#%2$>@fyN&&ci zcg}9FubI>oYe{jXju+OKZKb{83&&m2RJ9oEk}~N^oeH<>XNXaTcm0a8HcGL|)KL8O z)~4!1Zk`_OqQFoy>=>);+nQ5kcJm}cavM9zUCLLljSUW1O(u_`+1-TBM=sS9M@FcF zEp`y8N%5XpY1nVPcFF4)ME=-;tyYhFU*FzSKD2Q7)C-@oU;HnA!N#X1?5PV++taC`I=Cd!^mwx6aZC!Q36CaKxDahr@%i(>9 zMU?!MCJW8{(yU96=xJkQMn!-~b)b@fG%84dWXQ)(P-OTsZsNc{$Ad3UrxB2cD6^Az zJvqpo=YP+i;^w;Xg)sCAOC$lrQCql|0*amX?BN;`o@9VUCiizPR~j?C7NZAh zGI_{Q9)-ES_Eb(UmWQXd)gQ^R2=3&|?7#gl|E`-43%)PP@P|@{0jdL`b>VRzxv_hL zLna@1&+t2^wz=Omu+Jz3%Nh9)wU+(C@Vfnci%4zMlC(9v_YA*h}QHnJ4s+xF&G?D}ZX=oTASVx;ME za{{}#gS}jmzhUcZD^_jp*|DSJHptuaO1izdY5nab+x+M)+q*kst(9H7y0L9b%Ew1t zqqb2!W%JE-8|>S(snID-ki#~&x?s1je`M9YwyiGSwre+U`oJx1S^4Sq&%B;Ee#)MD z<{4X*!mzX~MRaA=rj8x6>cEioO8GqT$SJ#ddq#sLW$Ck#@iDu2=8V1e@@uv(1@&o7 zymNDN){?UR+_O*nm+$_`FaL@rfu=q6Y<%11u@+_CDMv!6_M+sK|KAI|+Yr35#)Ibdtk z71ByqMU1v=es$3w*KgL__AlREaz(q?JyHAz$@X`HJSJseVkD2xpF83bI6ig6e(1-) zY(pnbTSJQsCOFR5`UEOunxv;-k`H#m4oo22#QaR@Wa9} z3xAZ%1T7lUfPd_KGMS<*#RN*4K6Z*Pk87dz2^+p1ciR^sq|YN-o<{IU{PCxsw(orV zW&7$^f6KbInXr4koXU#lWk(9jPyO_NVy7RyXw}B7eP(*chScuklI@n1=N_(BlZUI__$4y4KKfcA~D7?uToVkLp6#p=iE8V zIV}DoTF7Cio>825P(gBDpF$R&ev62>Y)TI1R_pfLZ(a2?GEi@|fm2FzPm8#Qb?vTM zcV*cA!q5MN{ru1WxSf6UDSP8vuh>8MhyT4zj}5y_bNq-AiwfWo`eUDX(th9vKX12h z-B2Id_BN&7sIRy&A{IZ$0E@1ulUQJ?WZqyQ3j)fDAi01={kwMvgUYW(hQcUA>_B@g zPlMwXNDlP`Aje_Tf7&u^&Ke%X^kfKjlAwvE0N#K7AY$aIC>o7O|H{|6=1sUN{6$;zJ%?Ih8%i6lqgN zD5z)pFVW#2iY$GbVA{nEr=vdLgU*I5U`Af#t3R(l5rd@b`i1%o{31U5MXGxDF~f5Z z`Te{9)&JYghXtw0A2vi`3n)3uAz|15)R+Nf`@OI~6j%14$+9aA#0mCOd^ZRxpG^26 z6>f8*0K7dYL%9t5%JZj#dsysCtjLiIs1x5e%>(gLc^($zj&c}&e4uRiD_;l79KSdI zQ71&%mq!K(^$m&^8(kXEnc)Bb%>8GOWl5IbiJegI#S@`bMrxg`m&5|FT{Ds&XyiwJq2UAf3yo_utYJA=V3)w`C``|wRaIA4RhP)j%1W(5i>Kqg zcoD(>@0{x|USwrfcMoQ==@I^}U)O&8_&I;~>(?*9Zje(OIW6vFfh_sr6r6#wH)N%kY%rJM*yTM<$!Vp zt8#1-q(%XvH26A}({p6VMkQcbh1lHOaG`o?vSh1`U4fJx`}RjSY<;Wd0v&eWVFgTz zpHJQbKv)GQPZ#{U_w5v9jDY6~?Ug5jn2E_gAW!$MCd6-Xw>LL^wXRmJNU7E}8K_x9 z_w}kQ*LvM;P2dj8(QoW&^?35zUaI0W4<=Ubr+YsJN`1_vfv;p?ogd`i zZ<(yom+*yezwtEc3;YP{rAy&lNgHudjPP2di?p=g`wg2ZPk4&SFtA0iNcFDUdcAFr z>I0(DsM7(^E0(bX(Q8jyj&r8>2Gu`>l63}#tko4T+#eK3?OLV6@nK9-ht+nCmxDno z@WQ3&&8NaQ>zh{9B$BN!ElpODsYrkC8`UGaST*rI<(gC}oRw3&9V94T;sahv6Duli zw2Ku;47g{vgCB9!Mx|$)1bi4LY`_bVr-KvZ4RjDaV0vhLApYa!z0VkBN}bvFPextw z@ z^mhU1fBIH3$3IQ(yuj8a%2q!rFX|fgjVpC#ECwBe9$3Z^PlSBTs^0j;YG&{OZ3r3! zd~wRUcZ@ek*pZHS*|I+~W$ZpiXC~1)w zW$wNb00&aQ-927p8MeKFc*wna$mC96s2rKsKvA1w&r07D77wh(= z&_BKLiT%+Z{wwACTlVQEAK8OPcdb(0uxe$~TD6KT-@9jToI7d%```F0_J91J|GNF> zf8#IOuf6fI9iN!+QwmUQ2vyqat1Ds_z;RCC7vT+u@nt!H4MH1`Gch@C!`Vx4&1OyM z1lS9s4w6g7yfrFYw!OXOfCI(L7v*RJFLmMOqy`8Hxb`zGb=_)q?86&3ZGLXfN&+jK zbaCzGWBa}De<&r~wIXNBs(y9(v$E2X!r`;G?BQfZ1)zmurXBD*pqW3~M!;H%RN{aw(g(+x3z3{Ks-2Bu!3xHtQ6-|tKQq1 z6m(RV4lCON&2>#iA8)SO%}4j_?#d%uuWngGe9U>rqpBk>VnSm88$1d?z%VrY)7+{% z+dbG$!|IxYCh9I>?Z8YEWx5x?Smg|=P>bDcem~Pq5trY-?4B0!S_^G`g7P%CO}s4l#2HPO68I!0Rp_O zj*pQznlL6f4(uqy|H7m`a-NVcba7tczQ-qFSs-wJqQBK}=dX;n;*<(_F481m6|~5B zPKD~C_*|o$C0^lSL>B<2u0bQ{5b**`#!GvQm((uG4X~>G zsjKpey7Od#aYefh>5Yw-{Swd(TF_?VGbSl7(TBWJI`078D^Zjss{juRnr1X~8fWRf z4Wv^t39E>wFy)i_ub*$9ry2S`$z<;5W;WLzMuX+{Ce!fj-f#Jffb5>&6zWm=jsN7| zyBzwvm-8O3lxJq#edy=$-NyY<6P}1GWhV!}?g|4llaOd0H7h&IEO~pB@al8?q=s}+ zwtrB8Kjp%TF&}UzcO#8~z?1ekka!lnQG&(l3rqP#Os!?y$Pls9RkiaBFSV z)*dh0!@KwG=FQvo5ci`cyZ>O>nr#748?~mCyXDJLbn4lmf{jhg+tk8oTR3*czW&9R zZGNWUKo~H@>dfGtK(#<_w^|db+p&79W$OZS-xOf{M1Xg5^|5VltXfNP*OpgoXn)tv z&raF@@<09$>}y|n)#k@WZL&CG69Nlo1hikje90CUPuga^>7bIs;W_JfU%Vro@5iW3 z2r#qdqn^E721uTtnUNB0SfyUK+36`;SzhvwBfRkPOa3_V(`#4l#_c^k!?uzOAfn3$Sps8IQ0z zc|pL3mjaaz*Z_~9%mKu#+97n~2sr_Y3~VtRQ1YC^%n?{y^@ddizWD@lv%M>K+cvhV z_Mpl?sRbSAKfY&vYH()IDKd57Hsj%X&WlA zg0Hv=j05aL`EeyZt3Ixrh~o+p#h6**;-@S1L~f*Z@aJb%E4{Bad0wOscO1FIoBj&V z^7pnXk5INzUQ(dY9}q6bVHsGB5y}najZzGSNFLMEv-bG@9lQDIEwv}os#>0o{$e}Q zsk5hS{`i7`?w|l|$B)T7Rvh-H5jl;$J++gU;@Ue6@sh5siZ$vRwzPELRyS5`b!)>O zXdK?%TD2QX5ABA4?{!^2esIt3$iLnZ0A@m|E-j1}Y`j!-0OD1U{DpI(z46GJm2KE5 zO^yJ>ynNZ(>1d1~XGCW;gKk4tR-$^T* z_&)$t@K-OL%7}J>!%}$ga-@sA9BBD{Zz~* zc87T-Fz0U1FO?hlI7k#hM>SUY_|bR2^L^hU!6Fu8h*fm2i_(pj3iiS)SNy~m(ve3@ zKo5awq60Ke;T}f$Y8QT_rHvTsLYzb|Pe&T)oaI41se6Qlf3%4{JiPb*@VsG;=wHGT zO$g(I96kSxcHzb^s~>rJeIgHEXvy*qnh)AHe(@V#|L0|XieNaqak0y9m_#J!6u&=o zMC4+A$9*E98;_5T!IBal)+x*AnTB{N&X1esPhy@dqyL)f zo=k+%hR?IlHiUiFs>|I)RZ?;f$=sMjr(B}>;C{>Uxj2Ki(`wtQ0PfA}*X`r?e`p_m z^r?OH!7cmr#$$orO?&)!!*1VyXj|JmwzVa&+|Jp+a7o4S812Avxp9HIU8`(uT2%nB zUfZ@-vn8P0v^TDtu<6-x3Azw4{${U0;NIq%4M-_U`&zZtwFj*Md+Tw{HtRb!B2bYp zj@jy3#qx6h>YFdw|M|c8d$zDRV~suKBLL(d1dsykO7RV=J#$l2)|A5D5ZFa~ZWCtO)nG*LKl`kXbZRa;wMx3|CdJ>PaQHaTGj?UuC!ym_?CnX0h(%+!RhF7sLE zrpoK6UWNRSPYybwA%KJNPPw5JSz!u6d8_)4!U zmaL7Q2E``DBW2_KuW&>biiWsxMKMJ6m*OZs zikGs<$MaP@S8($4doP>%5MS4CvRdYI;RjuR5lvGe|5{wFUbDx`%XV86&wJ|v!WvWS;;B{f+5_EJ z*d`)yStu86Y;4TN$H#42ZYFp0nzT*GuOzSvAZDIAq424RaaBzdFt!J1?C|t}8uEmK z7wOm*)F!g(ulb<2K~46p5^O7AwH5k|j11V6#!8v}q~dKX{!HQ=pvj&1%&Fns!_yAR z5U{ORZA0y^>;^wV>lB~WxbA`aXake!nSl(>dg^OpbTG_I?CY+SX3NTw` zSb;u${*;}+^rCHVZF%1TzM?PV*u~VpfGMo=mcnB)CI0bn0WrW)#t%MWb6$Z@6z;%@ zr)uy${1JT7<4cd90W&xv@i=+ArR;DAXp8ygCUy0^{a(DA)s53o&rKZSS2Gqjm<3#C!J^77># zZK6Gt9p!tSWI2XX#mUWhODo6FB8!*R!`p(tEOO54OS?Vq@Fx?0$qlRc(I>P?Da0pE zSLGivCF+yf*lUNE2Q4TM8Xng7$W@m7eG(TWbG~sf?mQ52xcnr;?z2jyy2JR%5RaA+maKl+*9kLlr}sQ_5dKZ1H!`G#9zr9#eXm8Lzo zyKL{gbJebW`pDV`QqFT{Z0_tUws`3cJ9hpxn?LuG&78h$bEhuZ+~R4=kImS@kN~ul z5MONLxG?e)gtdkF<2E@n?}D=;VAhoa|Jv)9Y-+OPr!>S{-KwL2+iNjYkD2rtJ8cv(}U{ZZ@h?z89@k7a$b?{qEau+gi0La5G|j9{S9} zj4f}~d>?UP_`t@?0_25)nj|2^8O6R8LZlN~f`;r>^%Y^2iv^@e-TDNPRboMFKT2&| zblno*XGPc_e@kF<4l)4aalLJ~H>&pjy{!PLeddr);`*my`tt$^r59Ny3IZNd9Alb9 z96Np5_M~)p8UfgG1ti7NQ=D#tUmbubAHO<43AY1O^+Uwvn(?^8l2UR2;%QO3Qd}rT z${_(p@g3Z_0!6tgt=At#nkhKNMZx1uyabHh-|Nz+H04MACNv3qeE*)^zI8`(l=y(b zeV>FWrAxT80>g8Qi@rL_8N-5=zF*#7q$_f?p6Vl>WCdiNE8bq18Mjl%=EP@ZUo9_}3ZCCcK5rM#ofSx)wx+-+VgRh^;W2nXSLct! zU)0It9`c#%DZ;XR0$3$J34hO*c*JoVc?OM?pTYu|1)Z=SpDX@ePSA!jeOnH7%<2hP zCtr`3wTT4u5q_BL;(5B&pQmM9xn};^mo*7Pk7utKNICX-F}XtY2LiPHul_~Z({KE8~W>6_OL!7WHPJ2 z$VWEIZ+^^b4ayWjuNKK%H$)jLJ&<`-?ZaM}*0U$*kpf@c3YDbigV7~WB$u5}NF zZNE5c6Q|GF@fR=Knb+R1i?9EJEu4A521W}KUL*EUpy2MUPyGcaUPc>1K<#&V&ME;u zVEYoN-A>y))`m7mrzGYu$$0 z>^(KBcL%JKFPe>wTUX$LPuH@7Tif22;@ae9c zY}A{+y<$b-%bOKn{pOt5mVgYvX1&_5j~{Q^J2#i?`?uG9qT+q6I=GT}hDkj3Szi0` z400tY4)0&dB*ru=?Hx^Il(VnGyMmJiG{kf$%5^!RwW^#lz*MF~~6( zPA1wF<>*JS7a7QE7heJ6#m$_}6w9`?KP*tF{uMY3Sv)AvH^5<4mK*Ha;=-i8eC3pV z>9vdYo2#n0&{11RuWmpr~)0WFmc$&%S85^6J@bS)P*YkXU173ko;1TLiAFJMg zrC5QCl}qFV*B);LgLQ}M%=7iCeCFIkwyMm^tdN zI`CCJ46>3qydMEb_d3!xXa?|f+C^KvE^er6lzlkHGs``El=%nnS9%W*{6IDiV4Y@> z7d%26Swx^+zK}rORYtPJ@r4q2JX_g@pIEF(v~Y{N|6X#;WtiGZ+_03TvA~lRjSF|n zK5z3=SsC6~eV|_v-a9lH4Fc-l@9_?n2{$Y1;r2WKg_**U89F#~dYE6yr+@aLI9ya< zVgyg;LHY>~eq88_fBe4Z9UkuELrN9)V@s4@tople7AFRtm+wCEOAeW+&ofLA!V^H3 zUIsE=Z{_QL9!({Rk;8pxdim?(U*6GQ(3pskCxZ>Q4;*aVW`+S2bn_}?aq#ETOXKc{ z|Kr&w&9nTH|1;9DneB*oiq~(GNsV$9;`Xrq-1|CVW-(kD@F>I!QRX>qpfPIYu zZhHcogRP3~H@D3cXLqM-+nWM`oju#Ig4LCFMBwt`v1$81{O5n$zWnATs|g75NniiC zM6+sb(MS;51!i-5(`w7>{%G{m=ALy1hyAEAgch@Ss5UV^W_)Fi6A3t}Ajg@D4=n|21bT#PHjE0aLiU#9tqGk?A4cF zx2?*y1nH37fB4YW);IlZ<3R!B*<@n`hpB3Y}+<8gZVEg#RCZHx_j-R&?jRWUX@uWXqrcTN>#y06A zoB$LNiibDYBaFFIgvSR+SV8A1MMqv(-7A4BM33v&J(qiNpUDs9#~Hf-gQPrNS?b~MR0SBRSU$i`KZJY_g0-gz=WUK3|##a-eTSqdXu6n?GHG%S4HO{t9 zPZa2fwrHf`>8HVz3q_loDcYDOfJFgco&-2x6cEbi;*;k93FsUeCg|-Qn9`5 zr8-zsSMieS%x%E5s3$Ae@P+e);y7=@PwPI}MDe>q-^;wEXxx|I{ zoJLtZ!d!N!-p&ik-}_4GJ%6vmp8bVi{tKeX^D{r4J;ST<0-c*9mu!$TQ2owt!Z*do z_(1kCl*P%eS)86b8)%1Vb>pK?_x*A88ztg=!e94UNO+Rm4L1=+haQ>`G#I~-9!K9b zP-nz>mWeDqgGiH^C`rGH(4(Q>GMEVM^Y;LfClE(ixZIEEJ(HD_wS_zm^9d6{{rNv> zj%0YmK%=9{xC#0t`&skskT9PYhDqU{ACG!LkJRSWKCb)w+ygGyliT&HKXTpIy$1JC zZqM%B`owO)zH|+MUoA&Xw8#XmLq4pQ-*urs51ST|Tkb)IBy>Q`@pKDC{ckkY@ z?Tt;}o7`=;HAy((k8hjREgLV5sw@Bor)%VGssu1Pu$gk%6)Ci4Y;aT*%D&#%v6^TC z=-sM_CXKpm0O9Rs-L@3|xGE5Q{jpttuyXv7fJ|#8OkMD;<*w+`H)V2C_YzWitB*c{h*!N zF3}Q-3LEiag-i6{Pj1R`utVM`IoZSes26b^RA+IbzRKL|!hM80NK)<-ld(_$)N_$_X&@xohe`d)V>-5N6_c|K2@&ATV6fq!Ky< z##`dq%1+ZZw|6wgnocXEGb{W6pxQ29G(524h}o1TzcZ5sn-n1CtBH(x;xPVMiH0`x zy|2Ezj0zeA4daD9CVJchBKO;7k2P_H)}JxTE18gS%2RYZP@;W~fYjuk?%=~9zW;eo z?cJ>@nfj?yv&zHbJ8i}$FCMBa<%NYu0rouo@ktT<$ z_IU2jcuB2uK7YD5kDmlzIQLDq0>>ex4JEQvU0UD_*Twf3vK7aY5)5bx&=m#C#P!HME_dd3=`gnzu-s|Vh zQG4W0Kpfun@;n?lmgUn+9(iMkPaly!crNfmYHzeN#s;)ctJO{mm61Tc-z!URe`FA4 z=t|k1kLcvOS8=euS6XS2f4};iU!Y4U>A9O8d+Cxo!EgGE$5EL3_FpqM$$B;m^GlyQ znN&v4w?Ew7N8=xwtSqeJJsAr&>oi7uu@B#Q$L`;MV8_p0vC`CW+ZA{&OYjd% za_{X_ZD6-)V^VnI=@?BfLZ);MX%a2#=?(O?__2XN1=k~g72)u4p>b_dJvrEsa z@2G{!n;sjJav1QlWanll?9DH|WT($BasY2{mW~1$;PbL7?_jrWIVt%80n$QIU{m>g z;l!l<+7~X_|L*VoM|SM=g23U9jSPvV&8m&;RIFRwbRb+`ePr7YZri%r)DlR%y0mV$ z1cvujmIx4vNeTYpZD}Z#@%=ENPO-=i18pky8n2)`hY?ojKiMQpq zx3}dNZ@2Tv6%c~bXte@FaJ(7aF)i>>Diu9nwg;3a#x;=;KoOk>1zHMuwpb`%fgB#o z@;oIie6A9yp2`mcV8Spa0qxKRWlH=)q2zp_$O288K!}$*!+E>ASu;%~92^M= zOGIe$r!l&_$CvIn=5WL`+=Nt>R^xMEM3YseI~Wuw9Vu8&K$*PA7jO(%_7ynsdR4r; zy1s5(+co#2{d}IBEjUa<8*B|ymD}n+R)8TAtHB!L;z#rCN~Mx`mr1bmt@>8+0U&t+ zZNMM>Nb`f9gy~9C64mujJSNu#WjZRJCnV7lLRf zeZZvVFkM{n_r4d9-fOQ|yLr!+Zoh3Gyz@P~di|;ekM;)?!$sgW}c<{K=i(cfR2&y+d z6yqzssFrjwPc+DixZWqo6!y}~jvBX7!JJnHIk(zsIos(-XiSH@3PCVsKS zBDSq~!LrC-U)0IsMH@rTMO!29p3UM!JAzKo!C#h>U+nGnoL5}PpM1z_4_={kdtgbKPZ0&SHHXq~xVV4Mk(np&KWh^6KRw;E(niT{Jbk+Fk01VF`rZ8|^H0G~+NT-< zV!+d9F8D+jlpew;UKV4OXB`L(awsY_q*uR58(0N2!2f*@a(mMYtx2?*v(Rv4eM zx%p!b{;IWxH9KaNX4f9BY}u`c+xF?*b-Q_Q)gCTw`7vI5QsH}hJ5ttrR@qqhJ&YI6 zE&4|w=+@z3RJ@ zmXvsH&5YILJwCM1vH@NB2nhQNmuoHi=<%9ucBRavu+!sO1T-tc>>H$CIbLn7Ty`N0 z2l2Ljp-^-_@<+XLbLCMnFC9%zPWsA^8xqJO4B2xGNDbZ{^!z8rCp_Gh7lIdlR9kcT z5j%PEglNFtWB~xxOVZW>2#>V=@#0?BMpbVNfHJ@w{m9s4^-Q4VARY~pr!o=X8nF9o z*@p`l6SCCR$9Ip1|LhHAC7=u40mslNCt$X#ss>+1U!)a7S5|5xVM%F)4TY4Hj<>`A z^@ktnGTZM?M=yEXNvC}LO5Icrb5FX@^Y%RP2&GOR_UA=9pP-Y%qtcfV$LlA*EI@w9 zpDaJ>sNV$0Tz-03@Jt->mb1}f(RmK#&pEj&)cNw(+P1G2b2Q(O_{8@`Ym5%&iZ)UZ z(3Q;<$~H1u768s^tO@-2r>zBy6`xNxQzWawO6=`r8_(9prd2ByTie+3ed@j^SUkp- zsM-!6D-qxpFU9Iww4$jA?P?P#K44R$BQ`r#uyNHp&V3e*^aqdg5+9Q{Os~137iIh8 zFPbl&#YcQCazRVBbI@)FxgPdxlfu1E=qnYnf6&ss#@~S^qU|;7v{pSF6Jyd{{+s&4 z>y=&U_r3P2J-&OxHn%sOFMW%i==tcun!WqZRr}zB+ji~a`*!`~2X^bzReN;5ZncWq zdcezi)82=YY)T{R`~je%ePMebP@ahw-69vb&#XM%`ua#OPHF=t?%~b2~!=~6t;kNBQfRtH()5|H(({}V6bM}0gWh}~xtEhTk}9*Dl6AA-(swr;!* ztQf~FPg5dq>H)pe1jhSO=|d)GZBZg{NMMoYYmT<~c^TR)(<6TJkm-@Y8AUd`2D7*; zi<9m%oBc%Cljf*zO5?^8hUq!U$0IQa!av=IBQvUY47NUdcyW`Gc=GVdvPgG7+Q4Ky zI!rl-;CZ6!vg}|!21+tVn97;0SBoR{JZhdE;x+IA5R9AZk^Hh4Pql}ChrUsl%!(-K zF8R^tnEtdXOdHBw2 zTzJ)niwjn(a|UnQALF<7hirADZny8O*vB8=vuoFH+ntBYwpM8f7?!L&F=dTb-8Q#Y zrF{3TC2-r=6)#Z;DPv1Pw(1}rfidBlxr|u;R+d^XS@6OA~}<%U6yOk{qXoA+NjuPhR;pq(rvXH~dR>GqY1F zZP3c1g?}VKLY9~F{86jY;da1LplX$BLly99Cvt`+~ei-*40;#)lOf@ z$^Zb5O!33`WI*?xzfTO1aR3iZR>V7iuDm8s{)W5wjPW22wXGeq?It5dphovzH36-` zVLtv+u%VGrP4q@RPxe^GlMdee&xr;C)*>K}s?qJN$TxR7zM9RW{YtfB)p{I520Z}G zoZnrob!@d3W13deYpf7&t1f)xWpsGY<^+Vx`J5Frj*6PF@xmPKF7u+E{As7tOm(5% zP7_^&icWnGqkj0gRrnq&shK;Oes{k~It0MG*X@M-U_z}i;VZMM1E)|BFM;9!&xiS? zd#=9!c%-DsM!~A04g8yz%qo@hHZwP)YuQhOSzcbTJ9qBcoqLaMb**9RYw~;8w6*0O zfoUc;fF{TF;weC-0Mn2reM5Qq0g?gO@D243UiACKEUJ$KI$8W3bSX2hr-A^HKyJT- zKQZDj{7I*<1kQ-(R`)4p%EMQYbxmN-^Y^-tUjA7gu}T{&#a_=WPnGWZAqRQ)0=`S- zk!Ny8TujuDMpRW5G*{!qxeK*Wn*{CfUe(mj_fE*c@b?+;JL?=ru`;8ey@lV$Z8bEE~d zDPy9@g?dPsDC1P8A89;8&&Y3+amAF`UZFUXpA79ye-S6j^Q`*rvzwlwTGZPWp8h<2 zmPZVL7+83aSHFFl-x14z^g1&5T*}FxN7vLh)hMf%;=7f>CkA5-+NdKo>J!&P_feDG zyFZD@i+0^8uO^#3cr+yI@Q`55^cc-#x?X4}_sjb;=v*t>_J5aEV z)_`5Rvuba>f5UFxd}JGqu5C5iR+yOe7lB5H_f)z-Y~U&s$zR{#(m|Nf^B?T@bBvYmku z>vs5@wE99|mq*p?+Y<~I*yXdg2fmsESmyJr(8s|Eyf62t{Bt7$a00f?M#EMgKk`Sq zyc|_3mF@APC1oMuCpvSi5HIr~mRNx&pHi{xz+~z1l9e-)c(0kJ$M12^%d; zh`#;2LF(scqFg2e12!Sv;#3e;uIVe^P9pws#VeljuhlU&MR!-$0)^C<6-Fi@#P=7# z6sMF zd&B0(q0C=>#Y}anxvl=h6A|Ev7_zXD5^KELJ=7;uereK*&)US~nDY=5 zUtaVZ7C3^x8jXrAEic)_drJbv4ePW@k~sp^17iYEB?nRP2E{qAB_4wJML*w40Dlvv zd;EhBb%p=ny^J^0l{CDY5$zC1$SBjwHGT{qIzSW4h z5kK;z&T0#F5+B5UmbdKlHtMaPj#1b96-DkZP3RsAvoIJBL8Gj1?uUuwl#D}Ac!-2p z7Q_8Xi}AEBGy}Z4mBJ%V7R4RO*>e-x%5oX zm#a#3AL`7Rn9(=@EURU(Oe(nFkUnd_TLw-BDt&WPpN9n^l+&a>)tT6XXDJ=>`5 zS*vi?mbL^41v&@!+5&hIT)foP?%KPzDz?16YvU6W_VUG9d;Nt|c4}e9#-sqD<$*xm z2>Xa9t_s0WQUMd4~>kB3J8yAT#3HCS>D{S z@$o4as;p)Q$nWaPThKecZ&=K&I&;cF+bYeL-CPn7k}R5@S@g4uhlk@>yJV16@b;gz zp$vGPq9_yoK#{2h{t*P-id|831J)7Wjtr_VHGbgnn8Yxc)-j6FG|P^QRHW0`V0+!OKPLTV=m| zNx+Fo4lk7g<{Fij-Mw|+e)|vJw*USQuZedQf2U@P;Re7>B? z+TL))HrbNS{;``PfKtKI1`tWj|(?xVZr1O-++pQb! z5mRY_VPEaZi3bJ_hV2&1&oUt^!}Pn8Nq$M!h8wweX!iwX`?sg0xcxR6!X&g*(YhV9 zH?6Yvsa=2n9sA+68}@j0RsFOtc_6T?{u?Wf+6BI7H$7#OMNJGeiJ2ZR+U(4v+O}^G z?{3)ggNCg<*tPpNn^xNtPuZ0Cd_ur@!g8Y`B>c?3_-PQ}DOv^y4iFb$k%_f<4InDs zBaZXEt^hYIE8uLOA+B$sVbGBZp5_Xlk`7*m-uNqyuYjuWd_OQUnEbsA{AH;J;1#*# z`_Mywx}H<-@k^gNec@@DedSAk=M-;GSN*GAQ7(D29SAuKgMJ(SMV`N3u$K1OXDB?toTf(T-__&L9!}<{Pa)Zlg{f$pz0%kKRpDwAsqsU z=ILDe02Jw{XVgdeXKfo$IQ_#@Jib&nqI)Jt4*0!oDo6F{f$HaIqQ{@RNrNf7?HTzc zSLWb~llTb);~4`n@^I^J*(52RYEpcU?FotcInlGQr^is=9U2!D$v!RD!~6bZKgt}* z;_DwwZQ?hV+=qcsVPoK4Q1utvQhMJO=5J7i61S%li26zsm-wc3`S$IByC z6fhsMIKDV%Cl;psrLW;Z6fgS^Rf?F%SK6S1lp!lItUMs7`3e^=l_B~7=X||rYirFm z*H`>88$y0RH)dTEa2pzvvKY249^;<7VBv@H0MD}`S`uZi#RBA@;tOrm-r*gP#wPL07#>fUKZt0hbTvm ztUQz)VX0j1BTmmx)|EB)ic6hnIyU^G9!?OY^V3~mV%%p=uuUD?bD;r)qa9)SU;K5Q+lRSlQwRllaV@`RV7JfTEHzqzFvjU(Y zYfJicEfat=>49qAXny>Lz89^+#|_uSmkuICkCy7m$7Apd+CLwY_!ZOd-rH|jZTW`X zz5b!yy?MiSBp*42V}9|ZCcZ3i0s7;KPN_U@CDDd0RJPy`%Prtkfxud;k&e2Jg9m-PiJ<(KkO&Ym~h+kDXn z{2*WIK&41azJONzJ+3S~lUBLC4C+q)y*P5n^y)%>+*2-TynNan>EJrAm;PM$WZ>4r zR94D|^v+qS02z7p%Jp#BR6pp~KZ%OA1X%XI0|e{p6KnjkEB1Mt=;`Ndf{s9eLxMu6 zWWwd-J#=NaPE_{|w`&PmoLQ9c=ifDPGKoP52qyZiW{z92JmKK5Qf=Bsy(_zG9RW0GU9Z-xSStD>z_yTZyQ8)VY|cyy5YLxwer7BV z)|b#8(9GH&Zz2FBJofhmN;@st9chwXM{0aMi=)#J9tVoU3MU|d`wsSI{(u1Cz5s;A1>k{IM~9S#wY)>>@%%*=yy#<)bBa9%@wgM zJjWmIo)v$&qQ$CeRF8YYxVj>e^+%>j5uxzBU9yVft7Lx7%8{EeSB|pe>rx=AJYnQT zcv$j@_} zU2ANuTXk*OC$_wpxw6X3t*ch4wQQ|Y7ca$=t%g7wA86STPa-RBu85x+`vRr{xT@q_ z;lO66N3Eo>!x3)@B+-`~rN*QeK+7sIbc3Qjy7D78ckqJ~MArTF0}a$0Xh7T6b#QA6`j{n z8>Ql;uXxwhZ<`z2YVv?pH+O9LzGz+K*>BW z`Uw-iGU`*d2+)bF5Hq=CYy!louP4lm>q%X`ylnhN{V7K_jbSe@(s`TQKl;Y+6`tyz z#=Y}f!b=_}=zu%@K^tZ6{y~?B>-X-A_+iL{GO|3B&g+_m_3A5|K(p7&ZIu7KP1d5H zl7aTwpvT~hyTjiQxXGPN)_F1>=2nct=+TYG?Oi=uX2>hEXi?Hdtiy;|y3EOM`hTV& zUP_MX??!j`!f;S<)bmNx%cn0xXE3h*`)m{JczFFuOlt9A%Q58hL|&<0J%&FtsXrna zak$U?WY`yqIZR3%BtVK6aeKGPxI3$hC&u_6n4h%i z@q(Xn;DDHYixL{STm)nHf36+ql=H?7lTUvAfhtv|-<@ww{Mu3P=q`_{Spkw9<5 z4vw9&kxOp~B91n$I&Iv>LXj>z?YiCxFW_Y=dlBPC>btM->JFWR}>flZER z!mz6e!)Sb}7GcKM;`l;WmlxGg1L|%yK?)3DCkpF-s#x?Zf}NM_Mhhih@$nZ%l^#J! z->`4+;p4{&E7;uZtW|4u`HRLP1=}24cFF?mZ~!jw^OyV7UoNm^nEjKqP^#w20vYHp2$7JBU+W~AhXb=L4)wtVZ)g0>cyL}OE}ZGD^L8ixC!LQ!#YsRo z*|c(xo0MFmf2oTnb<3ADX*)pI)1~~f@#9mV6j^RBAHT@Se}gB|*hAPxTo3h2xKrKS ztpLh!T6m;bETK=RCB2Ge;l}f)UZSOsPtS)oQ6EJje|UjEwUIcn0vy)kC3O^x&IuWD8((E^jqq`J`d(kge)jYm?#xH&(@%#AAg50Dm zl3!f&X1 zZVNNBa@Tx9QdAvw8dWPw$?j^>F(UdQ!~xPMFW>*Ee#eh1y^F%-i%c7p9eYsSw?>>* zW|Q-$Y;t9VSoh08Y!i+ez>Jn)UYb#;ZE zt$bg6@dYc7Pl!));=7K((!Tw{?|#SL`@uEqG#YmC;sy0*P9S4Pec81SZa%cS z_~`U}+3Kxbt5qA;YF4aXsavzrlzYd|4_WG(R+p9f#t-0~N^uHJrN)nRjM?lz*8}vZ>i@h$aWVb&R7G+1^0gIf{%EWB`k}w=`*4(PU+F zlNTHhG_h&e`byQ79&cEy(N(#5<;8y75i5@seN5*^Ii(|>9_4tV;6y#7w}Vby;axmq zJOtnQYA?JlKL>z{>*L7dg!S~||DZu!)7XL}d?=`*$<^i}FfpEeTq><<;dM&+5RJMex#a12po~SWFXN(np(RvwOy9k_GY9De@&g^o!?mO;=kSjFX;vXHSrR z_A6iaIy^5EEq?Ciuru61K>mEti|XoZXdgCa!>BP<1qkLRcq#VSB%AU;UZh8i#PfYR z?oP8;JhOasdo)~E4~lEV!DeB7@qCpcs}uM7jqphMgz-q3&Ac+hZNH6Q%Fso$3G)Qa zo?=q{l3^h5_jx9wAC_^LJoD?7Pe8KIFg}2YFUyGV=$YFM9tiL3I9=Ol`pMxum1 zFC+-?4yz3I4|jJ3T*c#@cgnGCokq<+J&fSslf3Qm1+)1p0;bdM&tt+4FP{yLs(u<9 z8GTVa5>3`+1*SY{9RJo{eQfQ`RXf;i+Gt_eI#TAV>s5Q{_=NrKU;T>x-~P$pw}15a ze#8Egzx3zqwHMFYq7>Qa||Cj^Hnq$3zp+sIuklsO_|EO+a^fbe7mB@e@aeRe{tE?+F z@>IP%OjneiD_QzPWdOQ8E{f69Q3uuAZLDC?T2Z$b-?wlmAFhNjCh1Gcqu(J9{Z3u{ zbFgCft3SME@4x$rZLZ(f#7uPKq!V7+8xW8X3HAnu?YDmaTbh`T2ng5gfyV4gb^S8TN+&@3R9*BIrL1m`D}t707F&CV2TW@^+X zHKyk#b2dFbViP6NPxN2~xCjVlq9se;#Z#V0A0L^JL;qoDoOm00;Fg7NF)n2lk2_J^ z4n(VMO~z}xhcD0lY4GHaYf@QJE^=LN$%Db+F`JsdV6VLSx}6ai&Z&)}gurUcw@NfM zvEb!a2lVP^`joyc7AFN<`Phl%uJYT~WOTzCdV5Vzkh@W-xvkhsIGjr(Lmy1kxfB54j)i1!Y%2Cjf@-cn4!3s}= zA2F$0JfmoqD`ouG+H|Ku#M{T?ZAW zVr{?CkP_du;o`6rrpq=snzK8fJ`j)>;E}_U3%2os^Hcps8Xx3USTie>SdadT$5D=44nXk!JD>c!v*&C0nhKvb> z!WSIX*5FhPK(0UnK$dVigk#uHnl;5?RgIN40@z0k>!nY+@V;m;lf`))5 zUas4$)V-f31@NbiowoAKv=!L0At262Q8;TDxq>i{2`}yt-0AvfX`pXfpUQPVtml(J1N9*;$}rM|jdGHLc5u-bM(yYy+RevWIPrmr&edxI!b@v5 zu-CBsc-fyu?DJwA@zuB8ws=yLRh}fASe&u3iLyX0FTW{VwB@M==UYEs-LR{79@_h# z+_O6m*KMoTQVxRxyVLgOia&AXu0X!W2SICc8d34y1`zvsgfPilNE$h<*5Ji6RzxFSiu=4w z(s#r8S(`ofvK>GBC4u0t*&8pvV#j9}oEGqBvsJgMz;H`)kS$CYKh7(wZ+UDy19lv* zs0oSyFkh2o^_VluJ8fhtE2W}ynWGl>lqWK!W%q8}vio=L+nV~WUTgXuBk7Y_X@|4bfnM5 zO>dlJWx)S2PBqO@c|NI<>|)a96Ii9=T@BjE>Np=b0uaZfG{7)*3mfAhY&HQ_-l;qX zm+lT;!gA;1(9>q^%;I<*_3I#2VX0pgLH8=7k1m+rif$?cz9a|FKdXoGAQ#-prgD8u zlZV?h`3?KLO-RD$Y<%@CbNizTS+41i-b!EBNkj~HJnH06YEE%PYylH25`tsEA{AE{ z573ZS$jBpX%9)N%z|K2SOi9Ou#Xn5M4{uB?bY}U;(_?x&^zUIP`o6sR^EP@E4uL|btQEdU*L4iRY+pyJOPuCqOTmV%LaJaK)+f}}pCS@l8 z%V%DPtYs&rMr^Xcit&H|;GPfg(cFL(=e`5v^`-!u2L4q3z|Jb}%tXn$O1VGCIi{k? z0jEfmd{vYK@N=m{r0jO2Kk5~M*^ZQnTE)r~=j2Kgs*VDuQXnFp%2izs;w3wdnPTrV zM^JTl>Qc}G^d$+{#gh`O<97d@8@7Gpo{i3o+3R0^UD$WVI{T{6NYVC21T@qO`@8Hj zR#~~CpC`HMDM4YgVZ?1YCBlu5vh6xG1IIaRq)Wn;TX{X)+#|XsA08?!n6J zCxsmQ%8!@L*me@DiYV9aUenK4UT+WD;`E$NEu3cmi2xM}hW*F%Ys)!>5gIS$9U6El z0T)bCN+F{NPrRkDA0g2D<0jLe4^GZX=CSe2cv1P8RWw;Z8Sd$(+vP>Cu5p5f19w1$ z_{t+vHhHK~C@z!@JbpyED-D16`^s+y459|aWlfTEe&dX7zCwuQI5p+$G;rmZEb&Q< z5~O^ttau6#qDcLu2E{v&AIE`VkpX-$?m%fhzVhTn(OtzIhDVfPUgLn{{FDiu;^U7a z@CfU6Rd1yMn;R=OKX=StzIfVR zzI5E0%CDx0?ybkm_Wf&j?B3FrUAz9kKE3(C*6S^s(PZftf9~t{_y5js*nj#r|Em3G zf8(#%S6;sCdx4P|#eAHn&BtMUHBXb|yey}j0E9<%9ps6aCas(s&7)$D0i=)oq!#g+ z_X|9n(Uu$J6Yc>1k|}^{4-;?ben9#1LS?Nj8fX5{x(8zZ1SWXzX;uYghDkdnh0%}G z^Ox+kH@{*pT)ASezHrIT9Mc3-69aXpPm&uNS8OL~h@LD600K*V$x<|9tnV@TR=trm zt@^fAHKAq|dABXvC_cw{q0L1e(N}UQTV>~|Pjja!`_NXFKD3pU2R1S|V%^rhWW}~^ zt*_hW`nKfKfPMV_O}qBqEnB&}ZY2R?`hoXZcW4zG%X4kT0k z3y1Km?t{nGx9U?~7Dg9%67IVLepDkg`lk1HR0)K~edpSZf} zr+tc)R+_z^xiY4`o#Yd=U>j4^gT32I10~@>Wz?G(C=P8TPtRNTJ_PV_8X||ddNI%4 z{8aV~Z>&D_D@5>Xa%2Wl_KXKU5rG&G9M0`a?49Lt36LdD&p#Rbbf5Y8Zx+uDLGcs* zeQA@c7u%!pgG$lGz4;+sxKsLGnx6S=7}pRud#|!#;;s_}GBD1yTj zN7EiglRrV;j|bM3aNKvHCm_t5$I25(P2qS2YNS8RQG*|xVUcIm<;ftNiS5lEPvo%0u+u6=Ud9zA+& z^;TPoVAo!H`6Zj3o^XXyE*5NNanV*+1Z-5+%bmuwry&1-sa9-G6QG?FgN!u&Y4j5z`4)CrULFoiwMq%f{ey16AQP9P{r5w^r% z<*|~lromfhFP^ullV?3HO41cEN>_}`1?80hu!rkDl$GwqORg|cK=>2x<6Y&-&#m$x zPKpYYaKXYe~Wxag)Mf%WesSNre%2FO)mS{`3|H+U}X%-h2?B?~4 z?dr!jT}zBYV{dQBe(=@@_Tl$G63;Ykc2QvL^jVu6ANLasI=sXx z{$LMqML>9NYQo-p;e_Q212!SheCEt)#Xn(Z&z!O=7th(3-+aZs{?#|_7ry>Q`?X*G zs{Qhx`;xu+%4M6GQP)iYk32S-rcvZ?`_Z<@xbBZB7DOS*cigd`j^~G?oQQC4Ze);pIV1)S0}f z;Na0115_mOvGaV!ztr0&li8R-UdQ-}E0e>VWXJl-viZW4#Qz(_} zi(mb+*Bw41ZvtX`rLh&`CdT-o<=z3YY+Td$QyrCNL_Fr{gk7h9Fat?oUZ>Je>& z)@qNZC7&!E`FT3ggLrPC3LO;utO#-jU6fdsCS>=XKMbGTq>nO^+(_X2E3bdq<2*0( zQvt)9yg-+kAc;(JI89+eIb_^hqb2UbI8E`;l@R}ni0;X+ST5)Wdy(DZ1sK1jk>_DAL2Z~DvWWtWO@ zzvPDGl!H*hva-Cl&%I{SLu_AAMEsPA0|Us-Qjdtk)Y)^}`cwPItFBQ`fvR=Fd}bI-pL*V$)wE@7s$2vGJF zN;J3$azCuP4z*i0(A>6z>Wb}dJhI*OhqkM+#~1Iq z4<6aphgYqjyysuPWEcL%-?pKNIkl11E0wj^^{o$kt(NSLD>3q~Y^^IyZnc+_E;wdt zQv_LS57@Upy=Oy(39D>x+RHD#VjNG!Tl<$z9*-5W(V|Vy9kZ>is(o_vx__>DyVi7Y z@Z#kwc9ECsa-&jg0-w5ayxP8Ku(tHrRt35@1%~;c0jEyPFU-3F9pe+P)eQ-Mkw|T) zfB1NWALS-GbXl2D8(Gc%=uyM!?O~fae%>ZeoVPv2vthMWiTBz9_?@nQwa>?$z{}*+ zq_65VD>ZBH?$}syOu%NungZXm0zjuup71_uh(7&e-B*HSe$@DXmPs-O!>{1JK+%8z z{Fpy37BktoOBZeC)LG+LCgStq0l%tj-`}S;g>qK=;3X7Hrp(yC8lY9?CD;4 zuiE?XU3H*B`Z$Lc_n=*Q;dMowutyKJ?86T}vD>%qSVKILADgt(r_b5^%&c)L2rI8$ z#w(LLjm6c~`_^c*>@WS?s{$SL^@!?1dj+z_Cv0YR)+T4BZEAW_JTUGjAOObq_c=FP ze5J9$DJx8JCnm~v`O<0oxv#upUw!?uZ!_Aiw8V?B7I{K{+5=5QAMur+hwXqO`V0B( z@4fhBn3@Ewd`!#DN;32z9v~#PmB`8uTHLRR2lppz?(7RTF>}o7TP^WR!_)pr4E^R$ zD6{mb|GB4p50l@}$dp}r@hkSm>tA#b%tQdXF!6{lnW_KP9@^w!n4@152c8-ZaEaJG#<>f6~daxpKwqtX%lLDi>6sfkU?!&{p zv^J_bv$cit87J_4>N{QCF9UFL`xuhj$BE+dNpHp)^x`SS>e8~k^_{mQM|i1|Z6fq5 z+FMuk_h_kTU-;Tr{G4^hHhILj3Xq$?HHI)iZ;VNmCBE{p?d4`Ri{s~S!*^_P^YPD9 zi|k_=l5sI+V;-hMZ|Fz5(h zhnsZN*RRDFbWo-&2c|$VJ-XTR8bhA&=8K~ z4ag;pB!j}uMio{r=WJ_d-+p+lV(q;HTbLb{@}IDA3Hl+`VQ-&DNC*neSY5!wv1Yf1%&t7!s8u325wkN`#^Q5t!-OvX-zFr`_8;*Cw~1m z?abf(2X^8={kxVsc3SO|uvB^e5sn=cD(CXDcX!YDLLR3!bQ;^XzWl%*J$ztG0;nre zC`;YE{b0FjpRTQYopTau;|DwT+TyhR&7b>{ePLnJE{^5xSYgDKQcN*%1KOJs=qqdkO;p(e?>e;8a0gT;$Dg#`lQx5IijFB$o0U5P%Tvheh)|%e|W1 zuo^tQaN174^d$k|3%;d*EeOO@B^-G3fd)0 z__@-Awz-`9q3D#rEq&u{#7}e~FVaw+mks}Uy=3L*asA1NqO0G1^-%sZvkUgxYhMy5 znztLD-m%Ago$-f1w6-QZym&c3zhEM|r{@8Y8fv zB!F+F5zkYqaDyg0no9?NU59s@9tl8N>EVKmS$xi@*F; zd-429n;sWmNRF_-ok!g?duW)K#IktsSNBV<&hesCe!i&FTKl}54+0I|M%N0jc@&Ymb&YvTE^-g>#Sza!V+RU+eo1U1lLcVBAj~?15*FLi4r7iK#9s8Z% z{;mLZT_8|0O#BOfdYNh)3YcRCpqI<~Ryy zXmH=ACQA;ix?0M8sxFsyzc^?S$2M4UxHaeu3WSSt1VkzYgxH4V8@P4*t~#a zk*{H)_&LEsf?|J=l`CO3RuYs?Ti|Z98ZXuP17`WLs^A~!Af`ZUcYE3P8e1ljYlD*u zmOpjTreA*3PJZ>z+4L{}x|P5D4I6s>m#n?`lC_5>t>Y(JD4)HK?X{cAqHS#{mNxs1 zMIW{s>@+IYsBHQ+T)*2nV*HE3t=7byz4sv-!`r7%^_QtU>`^w2#`^Kvm?7~FB zUY;r0`H4}R5-8wIV@RG}paBFSsM#~fQAvym0G;pGmD~T6fR7viO#N+R6o+WTWMnv3 z6bLH$adLd|i!+^hu?Y~oU8&pEJL^_hxMT~L-n7E>qVHjKU0(H zi5Cd|e2LF7OYuk{pj3BNZ;sx=v|3W&A~la$CnhEwyzti(>N50yu~5wJjvCr{$b*UB zj~e7gTQq{5s_k8$l(3Rh)PxD;18?~g1knL{IQSKR`KoW`?^gG&s5~sdaEeDhy?7{E ztbix0KgrNPkxzhB6j@3Wz#x=c-1mcp4Af==kbG28x#8ycbYAQ1=OGhbqSuqpc`C9sEnj{9Ox-c z0!3+ZLYt6S#9|NjaCyc?i)9Cr^#)&fYMnUU8f-HQFr4DC^4{4M5cVw<@l=g=BNLcx zk)s1k)eTY+Mmpy;v`w3m+Xe$W{4`~|Xuf1};O_k_jRqMm8Yxxea1eA??Y z&Dw_C@Xr~O2iD`qnDepfdBa1IuY+CHiSz0M__9YG-ps&h2B3YF7ppivN%r&r?BFNH zx~~E&4j|WoaR!(Hx@-l*0)9D4(809>?qPRRHd|35|0ts{%JvlU2*4a5oG~Lg3f)w0 zk~LhRozo_3W3tcNJpU49VtD8`#(dAf*e)J2|0F1qt9S1ah!@x7?ydax?ll;FzaWV) zF<=ww_LN8!4Dmf)mNrWhd3vz>`wBB|!ZJVZ`15NP_M~C3_8X7mX}N|o^61s_GZqto z9wC#Z?U^U<`%)34ffpuS`^=L`pOJTJFQSF)de;AbeB=4hH+|*wNYV2`_-4+ZAor-t z(fDM8jrtN6HtIo`0F=jJB|`!a0q0S8p-Z_bA}<9vpwN}Nw|DpbjcDHd<{Zo8$7hvb zPyjL%K6Qxvqu<%vM|)}Y@G#1^E1@CbF`Bc-D|H*u;0GWdn=je?Jdfj*en0@xq|gU> zTRvz0$xZ<>RtdMNQWPwPIFZPY!gQ5jfoB2e@mOrs+Ji;g9Xn=&r(dxAl`q;*Zpw0M z$6kHgM)x~5hHzBggB;7oi+1cQ-jz~ig;|I(K9a#>OqJPIxPZxaYuDbs@xbnF)qT%x zSLygP^0~Pw`?asUV)Ns9n-~)ykx(2Km>kOu*@U2ZVW(oH{kD{-fUv3@D;|S3IYHc< zlwaGGIxqbpq*0bUcEpb@3Va5eF}Sv|?O+sT@M(>p4BQ&qUdxF-0lbnv&BN0OK=x#Vrs%a@WE1fp1p?xP!kigF0d(|Pi=2* zR@BS9+%jt8;(;d(FF=kee(>C9jdz`lD?*gJOn+R!5r)U@011>0%7=-WpP8(TwYK|{B;jKl|Su|__6|I(36$m0AA_~r`M?M=QWvCecL-N-}BD4 zfY3`nxuFcFm(u48$21Pks&5bX$-hl6$)n7c z4V8nuW)goOSqMKdQSL~lt#5DI+UABeJ3BTzf6_Kq#0MX}XDe%uZE0oAe)nI0OMWZ} z@Zq6p@l4tKgns87U;rm$B_@+xMm)E#&9~2a`Dn!_9!tVnkTa8Daf-Jk2n8sTGfNQ_QHT{>aa10&_BW;!5aI+h^RX-%oDO z150|3aOjuv@O$DvSyT7yGjShD0bG77lhvI;C41C(|=~y5(6L z8lSU#dBTe0lQtru-c}p;2BnyWILmi9>LuVgu&eR~jQKiUr@CR?M$JkBf@7Kx3~^F~ z%857GhwR#&6?^y7$M$%$ZnI-!HrlD#<@s^@3txKGzIbZJ<|hOU)$aD#jO9*W@>45@ zSQXgavAwm2c42JDN)no_4y&G_Faa`vcLeOtZreBuxw^e!+iUAK=&N;kSkJ-z;_=fq zH8Cy4R92&7b(rI@?k}&|{k5uX?~dBw{1uyj{a5Yvzwozg`ob5Smb_@!1#GJ=tg?2x zoVvk2?T(bS0J!=cp41G$U%X;W@f|(qM>M3snl$B0Fgj$zkS`7 zmRJ0=3m&DCckqt+D%6PRfzF9v`)#?_ap` zs#R+p``%k`+v?JqRW@o4s&3!9ExG`9Vs&tCZeEI7{jYL-{6b$<#;xkZ2__t|MjyPl zvSt7Dzx_S&n&k>nh4}t-+8=$0p5X;Ym8oty zP&vhN!M^a)Mf=+8FW9S>&e*vV$Hcp1@z`2G!hx?WZ7x9D3>47cUSF63o>b9HfE-q- zTviT!b#D|PJghMwAgqaEb^W0LpC+;{Z=Y?TOR`TI=$RiW*{iSfk)2E8lf3&AcX9E! z(}x!V0sX9=h89E1LJ~k6lVgEU;z1*&1C+5ff~_m`HH!m{MkD%2=>W>?8>ZpDhgf5? z)ofb5R&jvLWZYkzbY6+4MqYE5)$l*O&kC++Tjt~yw$%t=XOo+j#`#0ZhxdN)uD$>6 z_w9%8ykl2?_<`?B9?l*7ZnFBX zE)qCa-;L&drIU#;e9u^fr-QEpkOmlzv6)tS;W1=3W61qwF+5qKzZu^E#gxy0_`(6O z)0OcPlUfJkvWMd*vz|^?AW;{h~N(W2J&`CoIX$i_)AX5?e8XhKC^ea02$c%;%4o zKO+;M(fPh-@H;ezZoFZjLqizDE(4uMY=B-l{GO=2F#Z%4V3o(dDJ&Qkh{d#|%S27Xm|HSX}`m9NG zdm;kNp9Vn6D7eks_ke%xmFZX%8Ywk!{5LA*rYUe)yD7kPV3 z&m+=2&7?Gu%TePe1!N)KFIdG-W$o?rh;mShSN8ykgFOePK0Js+9^C1co}e460=)a3 zhQJDYUoj!zET$zu~zB7@eI(qZ4E5Sk5)j;r2EFW|UVZCkSq=%PNV5H(XFenb^7eiX|# zB=EZ@aKDehRaj5(b#`n(N`fr{2Rw#k^#+irI_L5Q8=F67#hC>w&Myc|k63Ga$!vAm z4q7c+-)h){2dj2T3jH^~c)@;VcEH|T$lEtwIc>l6@}ezF3!q8hS)ph{GxH`SE}-4D zPJuPN6A_AIlZt$-1NsSkR~jvw zotbrTTHD#RrAo&}Cr{ex7k|bseeJK>xv&3qD;$5pc9dtW)v!&KhvMxB^sq|Bq-58( zM)0x@FXwPd2xrWS7APbsAXcpamdlSGTS?{QB+y}ieJOx~+QzCfAZbQ4g8ECV8@?)s zp-+OZ_HBG>LVa3vzTvTA^jGj;GAMTcIEcKGKi~|*7X`@&e;y|T7)Qk~9TkcqMQQk+ zU=_#;o_{e4fS{`bVkyg5^$F16N7yM32c}YFUQv%vJoN@VqWDiu@ep` zQ#zETD{{)DEZiwy#UU*d3no(VpjxExJ=J|s{wP@!2o#62`sv_F<%$e^fTlb#V?X<| zzhD!S#~hH(iryod1eK&n{ltOFBNaC!+VIJ5v(ngv07}uz^A%>=oY4u;;;U;;*^tq{ z^_`FH{SR&lFbU+UP(BVLhEVyMAST(Q@eO^|bEJ}&zaem%TpIo%pUc78S^TtcEFWV~u$efN+q`qoD{mhCTt8j1v_W@d!B_8@2 z)|iG50685i3iKT~AJ@058p{<$!UP{Z+XNL7zd6syru%c3zF=eJag9dWEny|&bwQ56 za{@&Hq$9q6K!mF-pdFqg7_h{uB#+qncn$!K(;+IAZ6EXT6ll+enLWs$R~bi)uT8!> z*=jlu@^l8i;w3ad7;=NEQa|nqqkiL)i_RPH1R$Dm&nXjptmD!BPcV9qud8os@P1@_oW{P&QDUzw{ph_4 zO_i4qP_B$!T^*cLj{IF#$lv7w?NgMH1vFgIMGM9h)>oQUz5{1n-B0zvG6{DXlhFzL zyv_43vETS$Iy5m3lONX*0{sT1((5gR)f0vR5f-<5cwA4HZv6+L#wClJrR|SH_N0q@ z++pGxgV8N{xTHF2BAgsZ6X6cL-FcWVK@l(hB6heAP3G@$)BTS!KTe!ynMmsj66+n2 z?7yGs^=UN3P5AZY-4~WzKdLpc{@QOMzeFaozBDfI*|nWo-N8FxrNSO-0q#nru3GXq zbKFmBC`dRmexh$!8Og_^RD`fR5tOV9@c{||7<-4e8cp9zT_{i4@he}n#TUPB^H;uR z2a~7m(RSU|x3{bTSmH%MK(3E>#t-8!#?iiL$ki1N!kpC{T@lJ0>%_5e+Z%w3;C}=) zlbe(eE7PoW@rXLc2i>7Knc@6%E%i<<#3Y{b`Xm{|PiFAL(3jAW$Bc3Gp30T7 zN2&NrN4od@mXxo$V*&JJ#Yz${bxCW99Z+IXH_G-^D#fGi4whXh!%K2|TU0;7{9gG{ z_sos;6=+>uDdNuZ#*lviaN3lOA>HRjONyVDyP!VhO@ECY@x3&z#)f!ji0vpPO*SS} zo1ESL zjwaNYbY6&hcIEi!lZV`B0;CBWd*n4v$OGE@(-2*glJ=EaQN!~F3@bnQ%XtjpfIo7A zK2hFm!EkVZ43&kq0Eeu&&&^EP3+K-W5TCX40?AY3Wt$i)Sh)y|R5#^=fnV864A@C8 zc8R(^G46_<3|Bu1MdK^_lA;fJ>FMTZc#Mzet9)_6re@}{_C#A#etu*jZIlHJ0pL6= z<5%+ai5czH9d*S5#NAhkG`a>$T>gaaIYdc#|v?6lW|F^{-IBC z)33T_(XATF_lWv&82V3UdyHAEuz(Df2Dcl3xCO>Y)VwrFFB>~`W4d(Fci^HGGgX~Um zdSgB_1Vy;61p0knz7e74=9WlF*{ZvQ&2koRV)J9htQtSo*#^RPTQzDP+>P7TN>?f2~N)lco!*%3Rrkhk*b89VheU$KcRuiNm{ zg6$Q`W}}>O%4d>MKwTT8{q>qFB);FpQFLv!VY9L0N^xpp(n=EcRe@FqEiSN8U>#LU zV82xtco=lRNxNLwie?SAMYMNqv)T|?nzgy}Z`j`Gl`^V;~NsU7*sRsZZ5r z{h-`9Re+UZxD3*sTaJe$JDb zfnQK$oP`RoXR8GJ<-CjZ1K{qDk9qRm^HAKb`f^u=0HAz@)|JYjA1&odDH}Vk$h==E z1cf8FrwIk)%14$`qNGeo5(W=sJdW{j@?oSXjH};MZ(SL=p=b%q!ex^D>j47KL&oC* zC^fYkvrqbW4n9rk$$3U|Phlw=#?h7XcGX8rM#{xGKN^q6^^6nG zQ+Y7vw^mnew^LIZYDb#~ZT`e%n>}_??h##2*tsiTu-W66WEV6c$XP`b>U^y2b!uc{mCXWx<6XK5!%8=wYi{sm&+JK8e65fPh3KfXX`H9=2aiE}U#ZUg0{n*c`YC?m0bdMK zM*J}`Q6%|h4U*g2L4T@Vq=nADqMMb$)s5#DR;B1R($asudSut+_ANDi_IVpx`rOSS znWEd_fMBjelLnR#79S{apRC6bqw|}r=Y`ATAZ-1v=ZB|<(b>uEfsS}HKW-xpCW{gN z;oPI++S*9h)2k=mnRlb<_alDH=0aos`W%(_We#ZAPMA$R}tcqvY1dTEb153^04 z&a}z&yEBs#r1;M=KO*>9=E-cMI(?0Niun;Wex?a!O@VRc53s7ytv+Rim=~)Qrh%JH zuMX?SRsIAwnk@-nz-7zk=ca6Nab5zxZJS#t=Rq6ev3j1zxD6?fUGP&3cx3`0W(5^R z2j8*xkSAkfIRTvfklkKt+WVhwt4ag5uu!(svl9}Msv|2G2u0P?m3zMIM}Z{)gK=6o zFKfIoW+WszUah;+^sBP;7b}GPZC46_NB-1la7Q3`ZN+jM8@3_P!D{f{_L}9o4SRI! zu6^_Ozh%pxF4-@9{gR#i+RJw4=YG+Sef`(0_~O@WaQc{aM~cd8)b}tmR(ZT!+upR= z%41u9_`n`MdTdPrvA%y-%^Y>mk$3Bll@Zv!(gHw|AeZpwaUm~X zjTV$v{ZiDmTCaL{ccrwQ=BnSQz$Z$ajxDV&S*>2NApyu^$LHH)Z5brX{&B_CY7X>Wb+BY!mQTOjCi`P2XC zm;gIi($f~69t`3_EC5dGye^~W&P%I0e@Gs7ZM|mQ|SIEZ4*-Od(Na_n@f9RLOFm?KY@sojMR(3J;Jrs{szTz=Foyk+7TC-+b+Z)zut*Jgd zEu*amcJj$edvf+Cy? zxXhi|HHD@6x+^(T!!qN-$ioif#zj|eN_O9u&Xu6wE3B6>SMIY5`S#MNd>B_Ap5GI7 zc)B5@WWw!j_djKk2}SbqD)^w#V9+0mD5W`S5Q35ZXebpr>O-E%MzGRFLF!|HF!`~X z$-ZBdFJTOR!uKU40jku7bnU$XtJK=|XlY$>yZ(vX@v(vgN$fYSRH`;25X?4*(NfuV zIU`d62se}~pe83FFgygu4+zZe`b%6W$h;IHtJ9UvfjwT?vh&9#ZE`H9_$ss6R@^<` z&M>5Woa`e7TP(Hh)bVn!*$9wus_r}nM6t7{Q9zw;)|K1 zaqFIS)P~l90AHmp@Z7aK-~Z4aK3ucesj~gTue@%T|MGumcHw7iX!e{9mN*+ic?o16 zNa3&#d9TICN19T0yVh>E{0vdf9)&J-DYvGScmDWkdr;f4m6bKykzc#j^3~xhFTZG0 zvoqFp&}TL~H)q45|I+fRlv#X0U}JsNrpCvuJTYOX&zv=hRGXZfwmpCB&S$f|r(7Wq z=$^;?oGK82FZ$3|SP{eu0}KN`l#PQ#9%;`=A##eqZoBQr$L)wGnCt*@IQp%!RZ&xG zcJlNEo1R;cV!UgW?RAx^8me!iEWZTV`HbRUY7U*+K7>ymwW|D z*01_=`ucr<$Rr~@TyA577AI{_wBu?%TurckI=R=WK3q#`kZFLf)6W zyopSbpvnMDY=Q8VVfoQY+D;RpC&v3?AV4AQhR1y0q~ruE#>2`}-pXI?ksM*$03Z<% zNFMMVUtA&Li&|R zlbbQw(pV=R@tOSkI3Qk-Pef6{5-qQV^!1)P*+_LY_8>5Cm}Zr#x)K z5)CKDCvA3STJ$Q3mK{x?sq#74l@3`!hf$|!3I2?+&}}{j zsEg+qUzoOiqpAkiV8phc^{=@}oK=vEaCJ2;WMW-Xi z)A(%s(0X2d5#ye}%*}xH{}kUZeFnb>bwgaz^cfG=AdXpoB*!mv^x|`M8@Hb^_`J%5$CJ%d^6Om?)c=C6tY$n zy)}Kssw-hw862@qR-=`EPm>Dzf}W^18de@3^F4EX9HiB3i0+y|dzzjkXa#5NFK^F+Jd2_w4&`y>HjAUblyj9td=83j7u9)QNL;{o052*0;ZDD|g?wPV=4} z>}}f8z4z?a^>5qe=DMw}E@>=QqYjy@U>ws&@BnAj;+N_MAVAS`Pu;04DIz_t0sHQ! zTlQc7^WW|5XIFsY$Npg^XbwgsDxk9i4$*@C3~+(+52f$-fD~4O;Xlq`#R>>>S_)e+ zNblQp+VMnzvCpIZUDa=OOJirJZQ~mE$c1WC<40{`kNDj5q%9nqu}c?D*}2n;Ds$8t znk28SuKGz5K?_><*@@QEO~iq2q#Mpn*tjNGInfGG0tja{GtmjTvQyvG80iWuj3;oF>fN=1?17eScu8;Wej00%om*O!|02K4pHu8+<^tHm3 zCuQ-$i19K{(Q>{MjB(7?kxj`ofN;LhwbtZeLVck+XJFD#eUO_vszkLb zR$l?OZWz$2H=o&7#e70UeW)MfY;1DU#wMmb4}dN*ix-GXlM#)9CZNy7?G2UH{aP70S z3E99q0}hbY_FmpuS&1G@R$W)AZwCi5x#VFH9x@{6=lu_FJ!b^Ypu!BEUO+{ue$>X<`WLPjNKV-94aqg9Q2* zCyW8CFsuf#AFq=#ksIX}iiGII%T^6Mg1i-nA>pNKK$FJbQfB4A|A?tFsjrxYxDKdO z8QfzP0UNF8w;rlGD{Nmro`$wW83TL`&9PEek|*Ma+ssc~fQ|m(UU@N?eNPn8h!@F1 z7{NB&Wci4|l+$mVt|az)aLwY&IshcZ@2Ol=K)LiR@yH$fG!r$=%21@ZmsQ3z!<&+& z_WqTPN13F64c*PYc6Kyy2W)0^MBsJA8tQ}fIxC~Qesmh=G5X4Aqv?uOYFdgdXQL7_V|fYjk=XOr z?6z%weAtdFUu9=EmTET4r-$>yc4Dd^1WlFT7x9UzyKt!P6<33v%Y+TLupzK1OyPmj9CQa_STY(LgTq(Dw7x2 z*fP-+P1v6-ii)m2ndBHa#T$`AU{znT-a!edg6iF9>8U#R0TwCXfEVWSgcUI(^KU=q+p7C1=KAK-gO;GQ#TQ49$r z1DILK>B+xm81iDJh0)B(8t^tp&Y}F-b3QJ5`H^oE6aMj!vXpX3_hVxdHimzBQhsCJ zz6mMCN%@Z#$HnhCDM~5HlG12$GAj3&+>_(8vNJZJbYseEY+U*I2S~;hrzCKMyDUXq zls{opeg^BP?u#lXFCO75UtRfBU=k?z^9846O-F zR5YpKCDV@b9kn-KecjHT5)i(A&3^dqySCR}i;*g&2$cXvm~wH<>gpG0lfaXY>4Uat zoU;+h5gsi=S`=z1VB~9@6ag?sGtd{Ynx@o>+!k%W_4Y^hTmRzUny`sa;C7WJuRI3_ zc*?R1s&gO+Fr!ykbtWJB2$4^}_!fqouIe9-TPzm%2m$x1mw1uBgs<~z^Jqzvup*~y z99TtsQk9G-i8qf;jrkTQCgug@R~B!Uin5~V_*h;v7_pV+sJ(Z;Y7OygLlfH_pB$if zyhYjd+2&9tNEK@#3Pc&>g;1{G7s= zfI)xfdHuIFcF?1^0aZANqryX*$VqP!pLi`zq5(?OkC!Wpe72cMxcV%%0%?5Hm&yyU z9AA1Huu;hi2?UMRmaVUBSaoBm*G|eDpEx1jEcy#{V^cFWBzjjRrzF`lo;x-%GjFp~ zi?+U5)uc-kujLinT3;8aUbRM}7AwWdXLYG6**##B0C36UlEw(1e;yV6paCFtl#c-k z=mBUaG)_t*qbeon$UbKO#JJkZHY;#Ln z_&6v@o|QBv8S@3X3o3&NB{FT4tE~5@^5h8_E9ZGp( ztpCKNXP&sv+&)0vWY^v;0yX*dBFhPYJ!)KVcoeUN$LAW=k;~H~gb0jV09qECpl~U` z)1>lb4qZnLAU-G-#?#0~((J0>h{Jtfo^k`glUIaA0te=9QkuhbZhC3-qdFu*Tlk9` zuCl3s=-k0R>lpI)5445+_*>q=TabIX1)Z>IpFpX)1m&B_Jo zi4&_I02~RxEh#?<=~HJHt=6d9`ue8rb$E<7Ex|SHUw>jMHpnWH%5_DojCi};w`2%R zj|w!iN<`j#`k9Y+w3Wwz!0etBE_(}E?WvPG=+i~wDroaKW6vIq33HhKQ>_J zPpIAk@V+u8Rm;nEY`5SmZ%XLO#kx%g^ampnglaM`UFsT2tV7Y01{sH>_TjyVbTYyz#o7I)2iY9^UuW1%xO{kuTC+ zzkWkXOyEv{i&HuljxE@+yRclqByo^t>*RQR#yoo=kda18j9Uo-BUG#&W1R(2= z;zS7NP?S2nqYk65eLtkAS?3tJ~ktUj1j|LsdLE=!pW*wpltP05cfAocAn2M5rZ zd;rgM5qpg1=jY)ze+-V|ORHhd6KdO!GJxum_=Uc0X>u_=#)1D*mg7acc;&K``K+Id zD|Qm-6l7Psa%zLGS>2H1)2DCd#w>L3uG+u;7ypxe?>q02T!GpcpA1GmnJi^2{(GHW@hvau z39JLo)lPs|t+^vP!QR|qwQ0=eW=b|WsqvW;2MUn}r(T+fI!OY+(nx?VWTM`* zm2^&ff&2IrcEm-m&@WIj=zwI(A@)ftb0$6TsORbX zkkv`59xI}0MRBvTWe@M&^1Z^eEzE$OzVNbxQ)CPiJLzd%>(){^oa8exG3n!Ic4pH4{D1gYMDtynm}Ww#$)zTfC01d_G$Bqesl*X;njVV48E%sZN zWn(O>99_!-u53{u4Ep-%II|VAci~{>>jQQ%FZJugTMTcll=z2NI`yh z5H&^0uH>EWvLEV>ak(Jl%DZR6eF)Z%nkWxYBsY1>a&@;8Tj8JS>PX{2(m`*8bDvCF z74{O2G&GccO5?_f)o-#eFGfGTx>Kjj>4DEMzS40Rewdv49)by{ZEEL{Hl{iz13iL9 z0O77crVVNa%BquuQ@cALV7Tj#wd1IpAuDS3&zZOTgZuWu^=+Fd?%J7iC+yviZ`$p< zs}4AhpP09kC+DnDtJ}s_gQY(Sc(y@E;0io+SV8Qt3OZuPkBw>OjgmJBSShp3hOKS# zXm(UK4(uNu9<>{HH+@y;ul>q#0k9$a{?!e;^Kg^(CsH= zN@&lD^NZG=zGU00%VyOjo1LF=B{A4-Smog@+m)gyFPyOM=(x4lSFByHTT_aYJ#r=zH9FbAl-emY@a+_w!iw<{;K`L&-|=i z|KzIrXTUC8yyyz?*3Fyt&2RpZ14{%f3hIkr{jz}Iyt-}Q9t)uE4UXCD@e2Z2@wrg8 zY_tS=4mwTmV-z@FqC#n+Y*|Hyf7sXTFX8}_cI9SyPi5zZ5A5;kn%%ztz-FeV?82#& zQoLRJ^u{$C68Ib!7+ySaToaax-MM;IigQ)z4h@QyGA~}f;`=~v-T1^;A=9=8*cI?| zI5t0P9H8ruW7&4qPCo8(>Ew*B2%kGKXX6WtwsW8`wU0BR0nDtT0}gp9kQH-42Y>-k z;-^BWtRxd$mQjv`4UejMB+mF|MW1qslfXNmh-xzY2y-RQ@bf$l)Gkeb1Tc2R<6~n5 zJ9qv?f%IW}`+M)(($b299ss2x8Rs{XUpkndJ)5lDw>3tp+Z9`1Thm0dVKrXj>{cZL z*h&B(ZrP+j(SXM7#_A2-Gvq@cK*Ncti}vapU$kT56D9>1=z**NNU)WL?Gu}tz<&7N zcdXk|S&a=fLG4ptur(O7C~@%fFQ2ym>;L5+h?fAPOr|6s)OznJcm#e@e!T37FZG}f zX@cV`pNhl&XC}g4O^r|9_~%3#wqZd#0G)qbQ1P}aH3842Z?jri-4IaUbI?4i3F_k9 ztgoJzCIuSChwSRZLHmFHn;W*aGh~lMuiN6wdgDQ_9{qHAmPv8L9sT&~>%VGqbBobe zv`pbkE064>_kZ8x;YYkLeEBcf<%_S_POI)T;mdA(D1$I~AFzp1NB&vv8DxT<3OzV zV%R>j*X=mqWt-J+|Eu4&|M!3M&lQIWR<=$|f64MG@XIe;w14zp{$l~_2@#0}6ebdJ zgdaKuxx)D2Gvea7xQB^m7_6F_&y5;x(3gvTU2LXNMHsi zL*9-JL?dJaeHD|!(7*mwRBtDl(*%{LyqRp14ta_mrM?(%?*LyORX(CeOk@>V|FjJm z9VYU7>B95fGyGEn!{E2s1T=IL0 z=>x-2@07n^!laC{N_oAxcYkqCubeQ7p+U#0EK!F2MJpat;tz0Qm9|!03}5RNyWC&xzZ#Bt8%9I>69YJgvXSp4&nL3M%b zT5H!9XC|a@#?@a^l;X>JbKmYSReXQjhy<>*qw2FSg=u#4?uK1h9J62knbUUf(U!e? zb={g0Fr14i6>Tp}?b@IJ+NgvzV3-w72`?%9y+O69wQc*Ys`HWqJ^^7HowSkGrtRGM zfqxNhVDgxN?6Pg&`P3#Zykw&Ss+{LKG+MI#(v%$xjhQsHtvtA6r4fM<&FcAj9>Q)| z69|s|Vufz`<7?Kbw|swXT_E_k-@9#vnFV|AqmKk^cI@>RUb0gYC42SEqD_^DY)>F{ zQ0;Bk+M;8{{_w^0Gjj9yd zYc?@G?|b1sx$&trcZaMgx?i~bx-A}G5d9lcf~u$5&DX`8kJ)0Pd%y^**?8@;pIM-W z7ybZg9f4yM0p~m4ee^(l&~zYoYH`un>v`|~9ksJzXU?AW$E%!O`pNr05GdY=GGx-~ zFbetNU(SC5VW&h&a_iU}EI;ijOJi2dp?rQw`Dk^mdllbvjrdMBG ziF<6}MLTo;vIA6BCDmGQ%gXAKee}aWh_+Z1_+cW&%bg;Ex0Oi_e3W=k zT_jo97EiPYc3QDYNq@i>FG-dL z-FV3l-r>uY4Ynd_ES8EzueTV$-uvKPdvN=EiW>luHqIS?!Oow5!PpYTB#|d=Vmf

sn>YYU|s!vbAO#E4LI+^@e{{@1c=Nn?G^ZCRz0?=Izy2U$V8e`*!c% z4cigFsxc~uCm3;z9CcxoAVW9?jWWxnal3!_p~fTpAvvn{@MHp70i4-NbMn*)vhq6P z52j&ah1^K}#i9=5*1@o7M0o(p0KXB2J8^{fw&+HGxIg*OrBRUAld|Alf3E~kt5jSS z?1reMOLcNmXC)-rL48>5hIfb?{d!n7nMAud-9l-y%zezblbYatVBh`LJ9s=d^HbR- zvALpIZ02TW=58pR86O~-`}2)rkw5^awK;zOHb>UET2RBtd-$jL4M&+#xu}3<|=bY zn;+9qU^1x!$&K->vi#gn?j!X}@iM<&Sbv;;_vabnMO??kS#kc;UpF`5KH?=qbRM7k zKH5xZy07P_!Ko{sL#5w1S%Os!gaM~TtXKAIVtmv%5e=|_B5z6Ia~3VLT=tH(1fqwA z_brD|mLGy212+hhOk*rXja8a6T^a3wIaJCL{i5drc7E5>;n*=OZz-*%+zcK1fDU9VVfw`JW{ zM{RFeySimtw{O_J4?nSu$Lm&?Lgxjug^4l6>)6?`yuEf})_&>DbM}j0f8EXq4D;gI z+_B?Un4YjfDao?N&MOzsS$#kXSmkbRtXO$`+$JU_?BSz(w!OU}uvxQP%AZp$c+_4L z5aeuHlr~CY6A)3`vD!}8PMkg~fD&KQqfdFP=<-Bv6eGf!E2MCsBn8B1@ofYMa9IHO z#`?N%2jImz6sID|Iv!AEN}X02wtp` z)@ya!5X9Zv1fN!V2eJc2Gj$SCIO`7w1PoRDAzV( z<&nNy-~(q4@B>&hYZaTF9=FSv&pU096#+W{yXqyVfh4;5_@mv*4?k$dm}5^U zFG&#w&kj;QfzGr=$H6+Ya~Xd+p+4cOg{Mv(w^CXCtv;Wa7`7i?uiO9qx9?hgcTkg{ zt}R!#RZmuww^Y7Bx!Q+uey6a{F~N($Bgc-tWD9IN5>LTHJXxS|@M%?aSYNx>%ai!? zi|1@6JARG#6Z}!X^Oo8xx(@RO$i3?;+m(tYHd|Gn5Z0=3LPfRKP*~ealEqc^88RmTqvVq6=6okQ z`FktmEt$rap4~RbIZEEP+R{g?g!Ae8t|ns2dwa|E&vvD1)vcPB!zbdIP*k^TKK5u& zRrlN3J+idQdEJUv*;YFF`;~hSlhwVa)&2Uq#%`7TE7s6Gr|IxTKVFh1oU`=_Z|GXt zuGq%frq^LhwBuizhY7gUlbA$cY8=7SCN%alim%XL?4UKzF7}}6~b8SOBw{HLV zAODx4%JVZW>z==f&KIZp-O~-sn(Q;o6X`vsypm7=G-Zp4Ztwy13J0Nj2}{$V{IeknNf^K7#La2TK9!!(EX(6}%{Nj=488DueKjwbb> zsB;YL&qR3Aq`aOaefVUHnSB0J~+WVifN%qW&20mh+; zS=njZml7Hn3{97M=Xg~PwTlVPQLu;;V z*`urXtgdUPA>h7Kv$@@hU7j@ih4a()mw)aRd+EzxvVr+C*33^@d#GT|Jxxe--IY>f z)u_{~+o&cB&fXQgFY)cZbE2}H6 zY@s78g0f|`%RgDHcKKV*C}AGUN*SSqLn-q@PR^C>_U4wyVJ{tfXW2&#kI-i&DMFNb z90A9p-yQD*2Tagfg|b%;zpaXRK?&@&^9o!y~iP zQ?@YA{?n4~*>*5w+wHRb>mMvzO|q!QN~rI@W>SJaA=luOXlL-x=NZBQ^^?;Vd~(XT z^2wm`aDbY*A*AaJDf zp#|nA2vO%a5^z^^Y`I({wi-@jZ9wCf7rJ?mg%x=G851r`_2=)=_}GLN%2WoJG?C3=hRH=eqa6Ln(l^l7Z5AofIvJ0ud{eCnmPFN{ zlw)t#K``I9=S80b37@uv5VKuYt8!A>tcVRuJQVU~GqckIaYdD@JS32s8uP2mYd&ih zJW;-UHkp+n1Y; zBk;UGXmhAVJSf9gnT7k75lq_&j(xhcXfM2e(q4US z(I(#bf{kB!)sJICk`AdK{Ro|c0Mmg0fGffQ-!ICGcL+;_J$oy8W0@7orhp`Sv3dD} zwjt1Kx(Db0a%W~{tSHc0C{0M&wtb%;$Nlga8(<5)P!6FiL&>8k0+^sInLNa!`~U$G z{HlY$d?I?33cPtO`l(!i3KN>HlpCo5xqzQ!BF_PTo1FLnZ2-aA_O|EgyqWp`C__C^ zHkke~UGFHp*I6W$5(P}a128pTK*LZb{U2-rv#VTp1_1T|m_`!)>U4^*}h}y%{i}}r#o5{h*0%5I4^o_ z-NDe3c&f4?Fsw;Xvr>`buiN&<0|DPPe>A?=S<$tkp~I62lzd)F6hP3GpUN+svf|Wv zo12=mqWHBaaLI#xRwz-ll;QlZiiaFvm6ox>?t3Obynq!iEnx}EiH4=ZfGtdp+T@sk z>&TEjyuWIH@cZ9(8Uu`cdjr51n(Lkq>;NDbhY-~lNkk|3Di$6h|0I9tQ|KFGR+h0Q z5X|0SpQMT3#>U5NdS=R|C0G0iElmOekN@L0Ke3xzrU^jF78Xyau9^%C3&5(LgQ8Eb zEZVGV0&&kWOc;lTC+)bRr5*rbUP$Z0^O8|)x2ROBDo4EEZF^iydFr&y zpL)%oYVb}1?E>U)Y_8jfAAhLH(}Df9zw$Tj!nqf0Vsh5sU!W}7&65G@i}h_0$S1j( zgtD?u{bM2^lQomRV6SzvQS)syqNvk=ZiSBUj|J8i9i zj_8vBe3DN7ZihJ7_sX-8fo)H|QAPD391xpVd142@r*~qjQG6tdmrLnqw*DxIo7k%# zw8rv+IP!uPI@p#Q2}MhJNs|Ra+D+W7^=|M((mZeTQ`si5vHB2_1rcCwX8`|kLs)d1 z!k)jPgq!jsTVfC&Kfm=d(_b8HO<2%0B6x~t znrI{S)+I}Fm;iy!HS^F-l#%LNPqPGKXdAOodd>)GWA^C+tt8F)Id~8gD zspSKeGi+HYVjn4AhvLXQ7yA3WY!`x>`)Z@jZiWzS5`3IDT*z}o5|8>+I6`q?#Dyl0 znfR!~u1f9-C`!Jmi-Mm1bv&ibucM#aOSk#zkS2<<)VKL z>e0gow!72zk1?D%e?h&=s(oG)j$KQ`@3nb2+vgt`(dCNYVsIcl!o*c*LXVKD; zZlJUJL}ihuzqPOS?D@)M&YBx7%lk8)mYto<+r+pgz&O_75&!F zH~sd|Pm|&Nyj^+mRRMoK6;6KxLKG(J{G#^57d!&r+`R8;F~l3mjoC?o=s3cTwgdK9 z{Z?CKkaz4Q=YvDYwwh!|MH5)IPOz_6Vp6n?lTiTQOinS>kZn&qRgv`__3K=8!Nuz zdhyCjqU%eVl&=U_H(gIT(R{p_O-)bwxPfi$H_lFv679K zx_KE?bQKQ+86iC5+QGE`*>_CeF|nuZEbI`Zp4AU!;?fv^CMpxsYZ~EcsE2P?@i8x1 z=LzW*`BOhqW>-MH7bwFXY-r3R!DXHP*y=&~r1K|nKE?sr%7Z+!xJBB;xK#VNk9Ql8 z^k)C$pFH1f5^iss@^orQ@DJ`v9yxXB}>!4R=8&oJBv^AVgS^>ps;Pb=$X;g=%HWI=9H zo&8sSpDE4_OqMy5@jOYu|6w4liaaxmTLU%1xK-@9^{_}AwinA4T2?`s0Hf}5%Wt$R zQg{;CxYIBG$zu8&oSDPmEKbz!@Rn=lrp|=&Aq{}AM!$q5tLhSbu7_1$>Kv8wxbO$( z)dGN~W+x?}n*R80N5YUtr+`{MwOSl2yE3k|I=0gVWGS5nx35mQFw-Cx=;IN&^n?|P z98tq#dx;_WAp~nJlt-*uP`+fR!UN>+V$Vu-*Ula<+ScZ-gnrkCl;4H9j-6QGk#Pvt z0fC?a-X>;Cz|N-aR_@#2_zCOg#%*uow(Z>h*eZ`UtZ?kOjh}v1BoL4&mg5}dy&cbliCQedK2Nq~IlpluU#({|<0|C$xQ_G>mUdD3z- z$2=c|`-tkp=aBciQU>f1l;HN&#%{-Ut6NrIx^Itf+_QVDTXy5|hW+zzy=`sL;P|=o zezaO~d|Ut~=SrUyiraVY#Zg59eY}Z3DBwD+@@7=O&W>m3 z00Kaamx8u8mYug!-ke{I;e?0yJoJbIYm~@Xf$t0|5Ag*i$&#lT=hwdQREF5e&Miz> zp`=NWln9UW>kZXMlQQ^>rvU(N##@q3o{xh7WQp8lmF4GGn2p#9b0Fjj4^_KoxzVCs zy8MbwOfK5e^13~Iw4|uR{*oJDmpzDnQh>(B<45;w{qa?+Za$WMB-%D?I6r2c)`r}y z)P=(1%@qeqbWiNl>H5*W!**4{6xr~FH%2yU%Vh^!c*5!jQ5U4ypgH8K}IHOe{aDcO3 zRR)t$#<=%~%n0Wbt4+LIH$E|B|KD$K+rM~g%kDhZL~(IJlUB7wU7 zy1gcMQ{_cQ^kc}5s2?VU4f*kOC+yt$I35juU$M>jf>mtl%;^OH8uoZrDvv#FGJ`g1 zXV0EfT4Xw)hW9D}!f=iF2HB@P*%rZzWosLpE4^h+0pgOtH^=$063r7GJ`Mu_qppBF zUu|7yFYKo3!rt=%$=;JTDVd;7_Ooj#2O#a`(=0WURbBuU{J6aG*eY9(oCfd#{c!f| zYj*m~IUhi5Tl?X=@3|blaN&|4*T_qFv3lM0tzx_r!spiggB2yoPtl^Si8<}5h!?v1 zP01qm!B+%&MKkp=6Qa8E;&`>Qn%LjF_o=VYvii@YPsH*g+4#C3?TU#6{^|qC6@NM) zSx^54*v<5*epum!x1bL|nNQb~AALk0z;n^>imtr8agxS< z$Q$Ctl}Qz8|KWe}k3HSbe}otC!qc=yB7M@{6{Yi3qV@4lA|2cB34vy`R;lz(LodIM0hGE!Q~Jm?)25f$oL(qA7M?#vxuT0=sc*X%SdCAQi%xh~K637yE0|)}8 z8Q7D8(A5Q3)RT)lWn8fflyhF}cLQX;(aLoC~>~|a3kuvPYRs+B>$Bs$dNdcE6 z30u~!ED0R1n;LC{`I2=7QU(MhNISUKw#MUIw)TVf1qf>b!pHq3yxQX>+jy|#=Maxb zIP)BBVrIf7P6$k2dC3M(zF^Jvfp6hpAMy~Zcq(hJUAOK|-G;ZUV7<-fa`*tzc44oKV=s$F4(0@^LF;!jGaG! zOp1R|e#hNjlH2X2<96lBDZ6m-lmOBRU1cv2_oTuu*-J0HEOyM>gU74B73I{a)2hX; zJ-T_dGM}1zR%a<9;@C=0%Jl3IBZrPiO1rxI(-rS!Ak9cHCr{SuHU!X z@+a2Vx-War+AThk5a(&f9&gTcp4QmP8z1w?iziT;^fQlb0mAeX{NdYAL{q>nkJt)?KXW(SInWd$2RO#L;cRR<588h%XXinQ#etD$M)18K&S>OFVO-d)R(Wzw0*{#;;KkWb$r+*Y7kqnYSkqR<`q(UMk1qc!V2?5|PuzqKEXUFX9^q8LZ{k*OBUirHI z-nvz#@8_J%dv8_0p4nMIZxn%3l{eisJGv`5ILZT8_!Di5ozP=Z6)JxwsWQkZ>mE(OhK1M!ta!nE?| z!mR4eSBVuT^oOQ}i33e0l$K~eF{b)s+Yw(Rs92PnS{{mjD&4*Pp31|a%C_zC9?6Ies^)@PRWLPe7hM zeH^09F<}5N;Si^LNyiDoW;eHwT8DX{;f|(*Jy*(rR)R=Jc5xy)ObVVN4Tk2B!oUC= zC!$K4Vkd4a^2~Wi4iQ!I2EzBKAND9(4jVvv0}v_OUU+r3$6~}AZ>IQq&8rEQ$Q|Z8jVUV;D_T07i#_Wr)%-en8-MY7F z=fuGN>K8Srh^?@-VO-1_#wA{CrRhcMKDucK%?DPx^cN(gY+CE1uUY-tdy1!ICx8A| zZ1TdVZK}H?A+~Pgowhyt*4M1|V9gGcuZ0U|MFR<+5{x-TyvrN@qS@Taye*zPW8;-Y zo19s;^1?|mXfbUGe;P5oTu!u z?%V=Fn{a_d{$KsQf9ChpZfvgm3f{um3dCW7Hwp+*7Ekm>MaMVxHA}_t$siDOIu_mg_TzzQdkZtfJbraJUcK`$KB;&MVaJSRtL5&?HEnQY;ipyonOTZ4X8 zLT=aI`tB`z|ATAxV0FU}$aI=D)yHGi;hjO=iQmgELB(H*W4q2i8+QKOMdRoxduvy-ELjGt_fmGG(eDkoF)M9T$o}>G&p4o;y0}1K@WVG@bOG953aCn zfSU@c{nFpCx!GmsalS1_yoyPpH!Uf)SFEkC zTXSdE8oidSJ-IKBaXWqHie1u#s-iTj66T|C!haE__ib}?!*1WaXZIdl7i~7Z4ZryM z7wzQY3DK==WeIyc>cZrQJaRWJ`+Rx4guc!snyd4+w{Aq6QyYQKufF_qnlxUPy=dQg zTN5`m`XBq;XVo9s9?_EG+4MNz$3Z&|Wg^V8TFk95AzV&cO>L*M+p?N?VxR3Hw0TVy zr`0wWRu=uP=W1)M(5IHPd~!By7K>4Wl;0|)h>H^la@ zq6)*r$H)1EgPE}_&Nv{@GKNK7NfQjfs5qMSsez03i>E+QVz;{2*X35;pezy3*}mb;S)y&8VNCb z7;pUvJjf6a%nUSIU?315!5ad3jJtayY|b+}Gr6aS!k*R3%Z1I!_x|BS_pP$HAn1|t_IXE(g1V?c*jJBOawmcvV?~hxD=XJ&8n9I)3 zFgV2Q7G}b1*YOp$E#WR!onS~kJCxwWSJK^j&+7H221X8Ai9soIYg;Y5cH^GnjMW%o z@xvUfn8-#u@H$kMz~{;>4$CYpRjg6#`E3nbwWb(Q2thD=KgF&YdaKd1FTOEv8#_I_ ze`ni1ccE{WPAXOjGy4+KCVE_>CdRD+w=%tCy(c%6Zrh3yCJxu$wVn6gwCx8E>~Q9U zo&Ssfmi5c?Hc@+IeF@5Y+ne_A?RTv(zi10ruGmCn-grjzu(fN2vA)$c*c~d~lG2)( zJ7HE{lwddQpNUxkmb9&SntreE?#8wl>7hT6Aug@Hyl=hwu038|vzx2(lhAd4vnHYS zJ-fBGVGr)y^VQ(bYhd~-f8&2&(2=C>{qa@ zamT3Q^8Km|&WfAEA?(%TGdt-a3>R2MRy$#WMY~}|C-x6N9T0^11CJG<80-Rm;m2OF zU;aD4YL!zLmH%nwQ?yq3_C3o7_AdSzL2p`dZ{n=~99eb&n_!aEN zhmY(xfBknQT(NrFvHN##*<%SOdp%9MG!W;NR{BTueI~*5TLxuT7l*=Z;E+Mr4igdIfdr$H3vw#1n$*a<>sIDrivjg>e z*8V4_rtC96@oD?ZfAjCCKFE4k6HsnS=tP;7KPF832rUwZ*_Y}!2q~_D`a)@D(YIKz ziacJHVC-KtRDKtRRsOw}k0-@`!`{8IYX9YLT(jT!{kQD#)=j4e?Gt(|E}ZcfFK^v^ z*Xe%o^2>JW)D@X0tj>W>O-%3Id{cc;yb2nQ^Rn1!E32KC%Y3fy*`mU;Ws7#gYiVrf zL7-(1EcZK4Yl1a5JMa0wbNiazx%H;!f%pKvxP0+vf7Mo&Pe)CN*4!P8z=xT}ZQI}wNfuGn)IuSn>YfIc_lyp{T+EC^6r>QndbKCqj&ZYZ7yinnU7 zf9_{(VP;NsQM9rqDLgZbP)EIzCl`h_0jjq&S=-sMn;*Vs&DxzPKaD`SeC_pLursI6 zTDx~(-}~;j?6uE+-j?Re;;U_+Jo&9QD2)kx3G}ZT5|XK-5I|L5ii#`ZI?ax^pN7&k z->a<&LQ(WpKb)y@4c@HkX~ur}um6Jm%HRG)>(;k40iLjy_`v`4PyVs}7ys4&Tk(=N z1W5P=ZI{X6p}p|ZCHoKm)BnV#B~bUfOqxTM>GI`I0iYLsu*W1!^k#hFi-u1qu^e|EB z{EN_G#67?(_=p0>kuye82wTH&0bw6t4#FIRk$8tyk-+`jkgQpE&0!EPxM(A|u9AJI z@kfioMk{e=*M@OrykvJhI=3^Wk&MzvT`>2s#Fs)jPK3|VW_xJ-2p9Z>K-8{#qm&!Y z$qf|fKbI^35kXcsxmW7vf3uk9?r$t06!K7Jit+ZwT14Kj{RM z$#^+VgbRdWUY>KEH6U>}_H%Z{bolJe>2L%Em>>0n>Wvq~$oD0{Xck-G&RV4=#xW(v zLK55s=`TyEUw0*7uW#)tPCna_E0~g)m(wK$MC?`?VnN0yFZP)A#GosRj{I;pWe?7j zXa_SXu(x<$Li^m5Elrp0e6?d&R)nWA!0>FZW0M+04|{b9vh2GoNI<@C<5dlMV;V4S zziG|Yb?fw}Y~hvHtaSEyo2;$bq59$7_usYk2dg%7;dwjr*`KoM#Zxw=!J2}Om#58U zPfAEW>rCECa}qu(HYVYBPXksH%X7o*?x7vs41azX+XlZgAbZ`oJB_LlvR zzxSs7&Re(bQBMNm@>$z#iRKalIFR_{@e^x{abA^hdFK3C+j_EQGxH0o`w0n~n-UUw z{)!p zIZ~ZW%=;6vGR*;^oKJvke zfpdRPEooegT?5nHbX9^i zzF~Ih)SO*>?ve%_)vemlf$D=BIriI4U$NZRK*>!Fd@<-N&!|RoR|9B6!dg2CU((fX zcv5#?f?ZqTS08TLcfWbfzV!QFw~s!$ZZ~h2Copt5E7zKF?a*PlK{Eu zC;Du13S7|`{P8KCSEY|4s3B1@*b^7!i!uVi3sgVzm+j>jU-8K)<0+1Q(};xL@>J{x zr*CnnCqL=!ymp)*jk?dBV;wO>2TMX_XS6 z?KL5s;?S1jQks1UKNl`t@O#1!MZ0~TIhWAs=`rE*c7~9t^i&-3#;0nwNKBTi!9Q3T zmJqKJxsVaOT=*i}?rzKen_vIP{@XveZf|{f*LdNu*WcN^I;XYdKn&qp)6SaNr zV4t<+m6NuhvqDEJpSKHVkeTT7uy|l4P zih#*YG=At&?TG`iY}cY~F|nqNGsdvI!Ad$`HfYEFzS^FW^A%7jzUDBoV4Km(%96$e z1jRAC{?YsPr{DaJ=LbI@hmLEcG?1;Z=!r!ck zruDkmKDY3w?@KUmv0~k<$!%ACT0Ee~)A;KD(z|l+w&maA+PJz@&G$OD8`}!E?R4hr z;hx%Bzq1o@bZc@~8VU;^;WbYVXLDX;Ozmt;Z6CgIAhqXTNE(`yDDHm83d*}UMQk|| zkDgMaz|46T4pNKpi09IeMNJNaOK~zuDw|HD zA3B2E;sv#=0%+vPG@=6BrXO%ZW&5a`W4Q`*K=?6}QtM4Pe$9tWFt%Z+1Y|HXB=3Cg znjG8^DExRRm>hO@ND)Q6VG`iaAm@1Q>@!b{K2ZAq@Gt!jlXSoZO@r&eJ<3Yj_(mR5 zU`G$b5IxM1M)Et7Mg|;5J+hf{<;b%NVBmKJg9w*|JH!-g^BA7?=Os_X$@Q7blbjvM zZ_bUcbh{rvIbGRg?DOKB(u0393uRTP)7`h41X_eI{}d(0$4r}5y&ijtyJGATtZqEm zwo^+LtBxZSO!+;Fd`)PzyH=}rtkF5Nod%!MH9#tJ`HvYdEE(nv)AjWjWror*AOgH% zSCbIjL@*Z%InGUYr}-Cry=hYSzXp%hnQ8 z)i7>jE3ep?1ewN1-?WAX*6HV7vYF5RWgBm8+5Yxj33eJd)^^4COLp$+6>Cb69qZPu zx1+%pVTm{-xNJYt*A&JcIxyQyL;<~KX=Rk!HVSO6A8<$woR6bHajO__{oN^jIEqlF-}?+L{k^ECTFaU zfTPOWm(X(WgH^kD^{P!vsOd>i;GR%giF|#sLtP*|^G5!y55Hr7_?!ROni2#Kr>YV( zU+_9)C2@X^yo`%hq6@--ghmb(u?oqv*$7{4O=$Z2U4PS`Pp#ZEF*_?^WGde6XV9vt z{NAe7SY04jg<%HJeX^!H%x)Q6kdqrK_k~Lm@}Gb4MQ;z>k2}Q*jr_*;)s7`pxUR_t ztJ$uEz zG)S}hMSmmhNIyfg-|+Ko5pwZ8xd|`lnSNrLl#Q{4;-KX?{$&t>n6i04&1nq7efl$B z@Xyt(#u2}NR;6iVUtXEbr`t6oJLpI9-8R$es8 zW>wk0`sqKF#`DFBG?QR|7F)FfpJJO16DI`g$-;sCzy9SN`}Rj`)*0ha>!IDb^N#2x zj*;k+V>}MNiYcLmgLIo44}H~ec6wUE?`fM+J4C4Ob^3Pd?0M&TTiaU__A4HaYrEJM zQIzsF&&suMSmkYKLUsS{H5X*zL(z+V2;8}ipANdhX9UcL~qlF4l- zY?B}j4tDW%$|pJq;$B}N3^&!DxCfpIsB>qH2ZyFMCV`mopwZOCMD-^L(i)=mE1!DK zzcL{3z5Cuf_U1R<60I3OXe1gRpeyZ5CF^%E~;qEU9n6w~2p7ZqwMg3t!>Oq$wP> z8Yx@8R~bH4l&Z?TBI&s7;myreWvnWFwmcPXq$hRi;-|d+o)ypve`J7AqPFGW?ID>5 zj{nLKgUB%6NCTFV1vUNXAN7M$9brVtn0Zu=C(I*w2@rbuL%5)Ox+s*xh9-zSWEb$C z22Y7zWujBzM;xIt1Opo>8D8L$CuktuXR{;mlJ@a<0&vrBQUMuv)a_p%kp0p6I}Ss? zk#Zg{GaxsZ4|#+EW9P)Q?;|~&vL{b!Vqyr9bPAsF}A*pw|w!OO> zt5CBuzS6?yWa!3r5rltMU0Cr$Xyra%1TC(wLeQXnGMNC_7uu9?Of9h$gVjMkXZn>v zqG_>Mw(9hZubkYueM@vhI1@4XcX}&M^%qWz7pG^fu7T@IU;15-oAz<~>_yQao$N-4 z@f9fgEdCNLHDE3+uV~PeAXS+3y?ri>sciYvBF_=iHfRS7;)P=DALSJ~R)1(4A^heR z6766V@DHIq@J4ZqC0?Pc*kZM0CuU1_W^vk1FHGB6>1XDvc51$2C+Es`TI$SP#of-& z&PkuNOUny(PGJ@%d$v$KwC&A3`;C9`75l^A`HEe8^L=YQ+P3QAlE_ZEM+LEuLvSGLnbF8I`K(1<;vd?kxjS3G>Hk=*!{UaQwM;j2ZTPhqlP#TJ|=nv{0;6~-qz2mq=# z(Vsn;Me17VuB~s`*S`Fw^*abn44?FE+OGsglW?VQenlB=LV1hw1dm17a^B!&;ie9r zZtrw4RD90xQ>EyBpuV=sRhTXaKp54CozTLlbL+!Dt z3Eou|WkUUJ($1Z`Z<>oTi83bs~fF9PF&pdYv3LQ`|9uK+#9HH^o1g2v9yx zl{SFJ$}ubQ5-im(i7$Y#?GRwyCPn;i10rtygfI@h4(R8KPjR91=X_It;Kf%w7C1;& zbWoV6lK^iUf&-s`Nn90eXO|ZJ{%}7429CGQICP2d%QcEKGn&ZKE|d>ehv5r;h)i|n z;e_9o&@FLT`Pft3=;GnBGwEcdv?2b=-SoF^-LtQJ>6_kGxqF!0alni;0lp}ZH!p+Q z8ttbbArirvRnnsJ!%LVmGwSoRs2Q7+nwhR@?5X%u-Kc7{Dx7KYPA1^0SJf{o??;r{ z8wWzC6ftQ`$;=qX!EJ6gs+2XJxjv#4HxmOMABv~Cv*MaJ;$SIYeXdChiR*6y~sozr)b`n9|SpQ-GUk z@JS5<2hjh3NLTLOOT+DxAUd&|2Ds|Vc!XsH{6!b(2t@ET!nzQe@|Qz|E%f1Aa}=_ll!9`|%}=+AY5HoF+}Z!usP$}ftFHP{r&bf;@EV^p8T=Ti zxp#EBQnbzOrUW~Mg|RUPGoY!%`ymu(7DaV*b{OU~BVB0A>Jx3i9R`_4g+EuADO*xv zP;^V-a~c<67y-E~!S>v=1Tbs*CpuP{e07a@U0@hDPN{Q1XHpEvCgyCa*D#ad%gWF4 z3!k?A+4FWNVPL$rZN~GS^_H20I<~!4iV~V$devsmJ!exZ=WR@kc3Z+dSNxnfeac?@ z^h-9sv|y#`tW~%_c(P=*Jq-uS!)CMXTOYpm{$2a6uYbo%d;%2{cVS(MdSfMp12)^c zJ3c7gy?@`I%go0vIeKk?Il*(NKq zE}V0Z^Mn+2O}$W;kek)XvIZUoM^=4dDaCTpKQVK_Ye54Qe)Lfn#FQw5t7^;#*n$QP z03j#%9`zsnD4PLM&Ifk%3>n`~u`5r@^RqTbn$@D6oGsh(Y|&QcO13+tzBf?BbR4cKOB2E)eYZ)|E#DqTnY_2e~Jq zcECL(Yz^`$h*dxAibpM%^0;6{SoaA!pb%b26Z6?~7o2urF^CdHe3Fmwd4I4w2?-%9 z4)72k!YkLV#SI`a5K9;r&ABzf_tOfWTTs9!JZZM%zTHy2AqaW=?EOZ#NARZAm;}qN z-Mjn9K73a~W|LKOK&4mPJ5=2C@puk8wsCmg2}wc7o9GEXhcbJ;woe9&<*NP0m)Nj+c$JmI*ry7+uFKockjJtJKGQJ!MzV{W9^>$%%;`r zyVh1;Vqu0`b?WM4+?QMARsu~ZSy8WVZ7b)L`Y5mh%;eQ4@d$ybjiACU4o7i!FK*-q zINV1lg?}oKUUAf)bk8sKOS@3~Qd~zjHpwj+^cU5HJmA~Z3IZ(gvc1W}LsKG9Xy4*C z5Q15JDA?u8uehMKv9WH;D<@Ph2!Z1L;+2H*vLs&4;X?R+p6Gb|((|JV;5-1%WmpHS>^HC@Zz{L}YDckx_uKOQvV_9ExoDigxKf7vLi-6CL0 zKpkTX!d^%DJhYM~V(zZC1P@`+zy(I^Qn+UoLMR6y*@hBZJm>@H2)r)1Dz1{oG!%mH zr1Df$SjIiBybC=k390lQCLT;0nNa!GGq;P5$~P3^tPOp{7dYe>bb}iB>rYm%JJB`o z!5&qVi+bfyZ0@hm8b03Ap8+_X9D5;f)p>xvlLM5}(amJm4 z{~(|jxt0tgK)=jP%2}Vx;=}G1_AqCGm{@XWcK4*hXJ`fovkO(2iO8LabqJ2{D1n~` zwBzQC8|o$mUsCe-Zf(b^ZX^#?wy=7?N;Eymqs7OxL51#-}&K%hMjM}IM{BaKX zh!IO-Arsyn{DBS$Z!Q!Qt>l8SrUtxP-KytbvC@TCtx>PrL{GwSdB!H$&nkge(ydLO zyJ%Ow@Jlv-`EwG2SFF=$+d)gXqSGO9mZoiObKBnk=(c@u{f^zd|HK|{?pPNlF6Ozh z+qPf(<8RwP_>C|5O$n3A3vIyjMZpCK^=u*F?qG!C&Fx)(*PcDw9Q^Ss%2+WF18234 zJ*5c92t6=)1Q!J3Ql%_1Av`l=sy@XuVWN8yWEe^bThNSVdSy-mz_tVo)sfuT3ka2A z((GYnulMfehQ0gUKb3^DV^^Mk!A_jHV(p#;Y7J)GG0a`C;}UYIALz=JZ`4Vl6oQsd zfYdg4q8Xk-K5(NB$vdki45F+q#Q>rFrapgsW%xLRKtqcOoNNbL%KJ{eZRPpXHeNkp z2jeroipe0({i{5O%!3h3N?2)Pkn($Ti&d3r+y^)tU$Ec*(g*hbwL6;REc=UR2=z=h zc2@6u8~QGRnjA>--dupb-b=SXEtlDb3~ARgM6-}5a++w>=!Q#$fe-&W2L zAk}xtnxeDy&4zt+ch}y#@veRIJAdryWPE_~{va5%7v%vx(;jn2efid{_w2!g52W6; z>mUA!-M{@kyM5zByY|65670WgYZ4TB4C2ni+t!zGwEpCggvSrmS34dT@%yT5$_tbG zajz2!MXdCXJt2ld>~AM2u(dc(FI=Yc^|>!>3H}6ZG;s=UjfcE6b`@e z&s<*kY0818ZwCnBF@}v#E!&0j7u6SvzI|c&M#rWF62onKPlS+SWp@lq?=WNFbBPMxx%^C+`!d(U~+<0rfJ-8bKs z`>f)XmzW6cfW7dwpj@@dHI?kCMG)prZ4UL7HMwSc3H_G^4&1n6kFQ}q z@snE!$RSKp#!)+4f2PFGSae`p49Zu-r}#?1tw_WZx+|H9H+Ws>t_X`x+Ui&6C5_dTfB#aFw32z<0>dYCrBAw9SHeD^*OeIb`Q(V#yhONlN)P*H8 z%#T8(toRQZ)qujK~z;% z&e`;d(^g(s@O^XLja$}VyJ7uD*X{oM*X?_6-?s03@4CJH&ifjy*Q~~qXBypn2P`tobGxOh?m?z}U0Sb$+O(7t}82T@|?@M06=C-*;9w z?LYm`{;B=ffB#?lz1OWqLv6ffj~=bt!`l*s59*#4@U%RJ8}%bV=un%JtAv}XboC+m z<<+^=$b~OmW#DXvAE0Bq1fL|_FUL<=Cv0Kmf)A1imcDg}aL`M16N2!@d{W}8N^ebX#I?RZI7Sk?d6Ct>q=4WEnCx3!>MA!EpOK?{i{AD)vDFj;t zv~k6i^UCVjt=o_6+K2aSYrUp!D!xIce1bq*h&zSplPK~|IoP5BU+^1iG`XVP|Is&Z z+xvGP+4ny9y2>17i24pXe?R!7`ay^b_b3->0<1#f&Oy9x=dtSkzJy?|A?w?5Od=nBdPnK88kG~~e#=d#gf#M}?$TC?Jv85|FYI76fUp&erL6R*G2#5%dCGL}+w~F(n zS-J6oA3`evFkdB;pej6YQKd51=FP-SB2|ScNmwngPzGKGW+1Yxo- zc5DUfTD7omKYg_zL0p6Ufd*OzSrzi2v2K$ZcnIS{!bDN!(qMs5-QKpt+5-LU49 zEvr3VwcUHSt+uftyeV6H^~Y@D%vBpJPJ6jH=+UpQ+wR%}dwA!*UAunIZb)F-?ewiJ zVl}(0h*IuxYwjJ|lUmEZ@xg8T7k~7p_6Ohmo+bv$H-jQ%SKfTTAcI*DexM@=o?Kb+ zgL7}c|B!cVhtJ1s>$3e)A-EFIF?nq$Xwu_gqND!?mAAJo*vHjt@-}nuC^Q*t(SFCYZ zY~kc-`|&UQl7F&h0ARZX10SC=>!M><>C_Y_y<=%8=5L==v&kVEwlEIVJ z4F>#O37G3!J8D}qB5_~BzXSvom_tH<5AX;@sqJQfh7A1N@j2>;J7F%KJ*k#DWouha z+uY%i0Zo=JpD_j*SAwB?yVkQ$zp&&hYX~{)!T0SJESw-%(a)K@sa!l&zP+hQ zPn$0R5|H>hL|#R6wlYbQR(nea>y%d);=~We*pkH|xhFmQ=YMd`-n)I>9zVXT(hvE4 zz_9**5JbI18yN-MvJU-*f+zea@DC1)cICObiIZnkH+-U} zT_~^eqaM(y6Ar0%*t)RKLDV@vL`gly{?k|?raodE*B{~4Mp*gXl@Q)gf8FOxgn&uc z^z;b{)pMT4;{2@o;(_PU@7-128S{uU6n&2^6tSub@I{CTF;}pG3vYwxA`sIJ@PqHO zT0S9O#8xOK1PI?Q=qlZ2r){@y-t^~~|Mbl_{ZRs~-drRbUTRqG^2m9f>5l)3Qe(+!C%kB>fm130MOdi7LDU&zT}5qPlEF;lk2+MKW4G19bT&LuhRiUYJ*=9;(61+64Z!u(Z+F!zMEymB~bB78tL$arVG6Ylfut5#SO*^ zgJz|o-{X)_*yZc-7sN;BXUleBdEXglQGzgeX!lxH7@P8y2*MH{k$YPE5~lkwVs&ml zMT>du_jYVg4EMxmzhI~S!e6xFsh4eE!b`DG_7lMc3E)reKD3X%b<6JG*sw_|e@5_DA1(+rItYHG8nO={p> zGO6JqinZH!{PU+ezfuZ;iBtT%jFr>p6Wzhtq@jW4GL~*gI$I1;a zQz2ZzFZ}8}XXf$jG7|>q!D=J<=K1c24<%$$&q@zLjbw@5=|Y~+J-s2gFhIRN1xUB7 zHn%UKVQ2k`?QX0qUj^TLy|2OjP`P#6)DHllKwrP?|N5W(xAxxq zx2)BDZ0*Ja(WeJaPT z7UI8818{f(*HFs_#&(zL{O(*mr9C#!)!644xuSR6!Fkj*#BOu@2YO1 zeRzE2R`pk8pq3qh`vzBE5?>@43-??eqn*2=B zafQ#I&Q<_Ls~`UEPTr4FO&6bS_0j)CRR)H%bLv1+2X>IonBhDQ%g%e>EenksvAKLnmb*X z5bKX^T|CkP4D(0d6S9E1fiK9DBt0Lw54a#o{iaEWugp@<#FJEvp2w?>6npqfNC#;fSwsFw^Tr|C36S{o20ONAa-Q zOPQh#WuX!QGZxGZ!lH!=7})5r#>FD@w)0F4ck4U2a1q{w*wY`f_%S z_(E||o5Oz-Q<#a1x{NmF+k#Ljf8Zf74+Lxv7kXCiC!MVl+3i^arF`b#du6i{xuIbY zKdVfBDWT9oLLVwt;~;EObd%il17KiICFv;gzz^XRPf|GWcNDOLn3Q|xl!q!-p`NaCc|8f&b9z!a z3bawH^5=H{Iq=mrwnQKlunG?2rk>fGRT0CVEgf2Cf6O|4F(@%W4uuuQG%FYF_Qn_x z$9%NPuwMx!hFV=&wxt(8Z|6Sq(>8zRs#O-B6W*ki`wi_RD|w@7YiO(qFe<`CtA!R-Rjs z@Z9y)8BRenSaS%azu)%3up_#_tXXkA)BrQHFrxvE!<8j5a)cI zU{7>Mak$<2(U4xkiofIlO|l=5-GOXoBE{!*m=5zY+9P*PbHDRG2QbIgel-9h*s-sZ zRm|DRmi_jZzHQ&Ve#d(IyB<&QIPnYmgz99#bCO^92V{G8a~=-Q(4>r;B3NnlEi0m# zn2qR-0OOdXAf5iK{7caCE7eB03@{L0q$WjE;-8qTdfnzdqyBjt9Px%qMdg#5EFO-2 zPOAH=w-WaNj~{3<()K;~`|8t=9z3#px9-};`a?}lZb{%6_x+sQORJu)HmUZbzp@5p zV6}FrHp;7t5^xSD6c71e+|i&d+}LKt9jiRLAwfxb+u5nxwRdmYlw^Ijig>w|R|T~$ zam20=<0F!Xm;0L1snaf*#3tHj$oA1Ta2)W}ExrzN!mfke(-a^g~8{#ux72Cr(zL9ZyOY zr-|?HR{BgPTzQU(SMlo)Y0CSC6mcmLi0}ChfG(s*n}~MdK3=GlM`D#D^vmt?S;DhV zQQ#Hjib5wHf+Qk1A)N9_MW`cg-rgNAlnZO>uz4E;l@$Wq#}#P}bi#W02K0yuh zeN+KzYeRFce?f9me!va$4+W!<31AMf(s)!o74QSs&4fz|v((2|p6Hk%OnV^AKGT-2 za53|am6wDd6>vPJQTTI}TR>jPH;j|K3_-N?`V0Dxsx0&ablIreG^0(7b4>AaeG;p_ zb6f=@A(q9QF-}RhTVk|h8mJVP6sy3MGFvAOY~NLz!PBAHX1 zCzt1KWpUbOs$x14WcQ%C!Xso-j1VM(`B8V|QEWtiqOq9j*g?_ChaLO;8z&@OFDO5b z?yH~dA!g;5l}ea9yNh9l&WMR`PfT6HbWei#L3PFoE2nJ!!UdbtfFP;dYFm%2{p4Ni zuHUr7tvgn`@sZtq>!v+?QnRgQ+cw%A+ZN+yrHapkZ25Rn@7R~$yKQfNaKl$S`Q*#1 zMSLa$ykB9L@|9ry0oZ+&SUSv~J=$yQn}qO{0stC#U-Unj<;DsfH-9kv&CV}K%{jv? zmeZ53l8zC2eUKFM<-l8u`*VfE=l=5xW&2P55C2>H<i+1AlW!u=?v0ANR^9w7=*OYH5Kp=AlEy8yYW~Hi?84ZN9Di5D;7knj& z6`#5JC8tSWwCSjAFo3}jxQCRoG3arH8+#!cOsT8fXVH@J2td0GAM4+;@)}py3Ys)2 zE!wwm$`u4P+LRiM!m}cD(C6BuX}f-Z&;Eyh{_D26&PsWFLe3C;kky}cU zwterTZ`oUKeMJ-7O`EPR+3C|~ebuqP+pr6lU$&R8d{)gxeYd=5fAZEt+o*9slGFWt z39unFkRSH6Ptcx}AD=wJ6IDJ0r`0Fx_Itnc$9DVun=Y7*7t7Ih5g=7JJn-Q{kacWz z-Ry%KkL~`gk0hi&a$5Yb0L@~xS7lebLJ#+$IoDa(N86Gn-=#&n`qCHe(xq2@iw!Gs zw0~8wvQ2q4DbxhPUr_Y2v!8v+IN-!ru|a%ZzX!e|+!p^BJ78abq@B}e(uF_c4nh+> zKK6ZwFt3KlkLs2_qP(UuFbQ%u;lXo#fkxgDw}MEad6~Uwg(G37c!ql6q}9Z`%bKvY zH38aZaY88Eiow0p2>;xD%vFsXNJF4x)fdH<8ovn$9u6-PUEuSa0bmmBtIbM>eZ$q+ zSxv;MN(bIXUKmHjBgAvK35T?|HDUVcpZED#`=oI?nwk z08+&d!A9{%K>mu&ecV@Zz_XN~`YTHDqKl{cR-mNZoqA6A@O}6)V;C||#EUuV1eLNf z03X8P1>kFo_HW8Kp~!88|6@VO5qek2ud*a94F=u{PTR^tRdpn$ zD?x*mD?Tmp`P}a!_I9EcC#DEf8JD0sK2er{qrA!C^xT+TI5DogNbuE6n}Nq)@JZVV zs1u)oPnA7hR)Zy(*@1*YUJGLNcVe=vfvqY=yl?G#%??}JR_JWn!Op7f+<35dIg9o;`x#6D=hxI8^s6LoTaPQ@AZCiX;@u}_cgzSkED;|em*QUJE zU%AI7lc2U+^aC*4mE?Q4$^tRTb__m7qYK z)3z=X>W&@ovc91~mdBNS$M9E0S?bwA2SM9aeUK7_+c@-Bqb^%|IvmY#i^0<|umqL2p zLIu4Bkj?1s@0N&Aiu#PTb-7|6xMNSjWTtMwVK_k@Du0eI9eTSGCU}9PHlN)IOBnsc zSgVL4g@`{2<>^waz$2??3`0)zBLwP&(x>_fv;PKQ^FN5_9SjAobQK}1K#oYf zQ)HC@Q&zwLg$C&=dHk?2NQZQ@F$@^9`ws=Qo(wXN*p5JE0b< zb05JVd;+&eT0`M?h!gHZ9CTO7{fAY9iW)$U+oKGeoCd#Wnb3pE3_Hxg ztj2PcO<^iN@$xwkd5@Jz;=^8G_AaTIy}=CNymZu&B;3;|%xQNHd9fgcki)%=#lnOH z?J-+gn6@bixs0Z)mia+CrOmy;gH`#|PFO+keOCG_j5Y1@*(G@@ET7HDFU*e>VP6$y zbzbxr1NFUE$|G02<*U}Hn>aN-?uWTJjKRc{RpW7$nS&{nm6J9(H*Z@zExUVHg5lP- zfA-(ogVl{$OJ&+>?c1G=nl)86(qV<0eYgzxtb%c0CEycEDZnW#BYM=&5L_7KN;CEa zCpW@@7S{_Ypz5HlxI0RNtr-s=uWMq$YECfQjs*U7<#l6g-Rcq`TCIk)TRXP3{!jw) zrfqCKws)@Iu;2R9AKAQw+c#dmp!(4uwEo0yfA~GS{?0c%-IbMjJAdJVUAl7BPM$d@ z5<^Q3Y8n*%#slabLx-D7obWDJygB(`sKFWm$XDf5Z-vr~=)=GW2J7YTJec8uko4U0teb}!Agd-nY03lgdiBos-o(L|?Etl0Qu$tFwF z_MO#-_T_gUs}4CZr?BD^)F*w6trd$?-vf9reJ+SNpK zQWF)$xv#i@mL}6TA3m{7wb|HG5bi%LMBPOF4C)U)Bek&jvOV|GkJ*`vuXsDnc(n10 zSAWW0{nTfThaA`k?<=np?1e^;e(j4b(2LquP~X2R9!M-48e?z1yq&(D3+hd6u`9uo zd_|f=$nA(`L0;iuKH*fyjum(L%g@uIPpSRl<_Fmb1OL%v2>OyQz=bgKA~*kfOCRHp ze0#lj#N#$7jD+E6`|xR2rrA$U-0Wi|9r8~;s9U1vl<=^8%F1Xs{5mIIG8GG=%=4yW1>}tZSaS` zJX|?GZcc)%ghkjthjJL0lU)%0$?~rPsj&?qM>>q0HW&@}C}v`3A6E(2@#$qAE|fE= z?2k*-O$a8esQfl-4arvVjBZ?!HRG$t-Nv@tnS5p##P<*ac zNqjX-gjZSBmPH3vzo1n?0vVr?*&j<=-~k72hCpaLd-|MJB`hPHArP@T=37Qm`2lDY zg0$!<81*}DcG?_s%W7QtV$!8Xp%y@0NB~ov@znq^X0}qXqS|YxQM3Q*KmTv+n{VE- zt*tFD3+*FdG+%O~P0^0BJ?U9xyMA5QG)rd1vETK{Vo9|07uF6rtjMG<{Bs!wW$_7H zJn`)N9tSuoljvTVD%nAgtv*aDV};lc*r*NC@8U~}`XFgbh`V?Hkv+V>VfSy}x6Ma) zYt8L#D{WqNcg(TrLl1GtSyc&+Bd#; z*Z!Zsea(LB+iU*ZI-64t`z;Bx4fU-M29bdL6SLBbgF^^*%AVZUA8*+8cdw~l)Xqg- z9(rI)%G!Fz@9SQDbkFlb`Tsl+^+UZynYbGkesl8tr|c(x>Mz)p^XF`CdRoGo+=YiA zJiB<+&YivFE9@RrKRgD3P^?@MsmJBpLxqCj}2in1qJ(v-oo1=uTrGr|XLX;o`p{xdZ7TBva;X z3B#0)K=fJqJ@F6+3S9Ogvps@5GBISU2dlY!{fhi>U{gxr*v>#bB1m&4#QBJDV%r?? z!K(QJL*(k~DjN^1aIZQqSKigcv|OFDlc&y5au2|la?&O&3drLp?&OpBs3(y5^M9|v zj|2Dw8R%~2Hz3T4MS1o(9C7F$Kt}Kn<=}Yk>VDZYfP^=UDi!7k3`2Nyo`+-}dOdT9 zmi3GwSGbyhlZwWe&C%IH4F%pb=!46j`(XpZ)qTM8K_uJPQax@!vlmsHH^I#68plXhpdtr3RdBnB6}7%WwqGe`HtV^clCi#peFX++d_I$;|i}Oj)t?PhlGH5kUO;Pa*rE3Q5154BnF<%t~>m*SD^8mp?V&l{D6~ zIviU|Limn0677)u1(Q)~hY|!P4(*jIXYBO6!ff61mB#V00@&iN2tKi8#ao#KbV4F@ z`%}^Vj?!U2q6PpB#;kNvF5k|A&?M%~>MpO;DQkAnY+CKfhHY=utls9+<)n30)aNg33bT#(BUeJwt>LKh(VaP(sYU=(556N_MgA38!i?zG7R37XwLZH8MvBmik} zY&W~Ua$k|)KQ`948R@_Kn}1+mee*5b(L|@+?y24@YSdL5pPJPey&yF!>8z6MCvCiN z+9nE1vd>Fj98t?QQ9R*lyl_Ie%T}taSao*EW|mLd+=&w!+^eF`fpOnxv2sFX6fSWp zo$*Pq5uCneOeSCIp8i14Tbm49CX z3y-~Ezj6DC-CKQPkJcaA?c3LF?eR^ys&@&;c{_q9X3{n!3K_R)>& zcH{c{LOHP0XU^H;(vq*LpL^kT36(F|MyF`M{oVKNzyGcG?Zy)cNqarDeQpe){RY3I z9WVwsAC#Tj0N74JUtGPvYVEB}P4M=uSYj+Xw1aA_18*Smw12K-NKP4Wy zC!woYIcGoqxu5c{83>Dc1sA|RJv(ouDYYZ<2839o0JaAag|Fs9C)yqSFVT|Vv_F;7 z&E(lv++mt)bH2ybV;aJfhaae%>6`_%Qc#(w2Kq8QPX%XH89)GLf3Nd*@lxo)O$U69 z$SdRirA_(wFBWB1)?>?r97Bl~2m9Q!tb8l(xKmqvRite-nzkz%Jzm?etIxe?1(jEs z5-s8NWQukwbabWdJj3f&j*MBX-k^KA6O9Qlz)O(b_;Wug+NSD1`$h00Qc(US3*!k9Af%#*!)Enm=|i3Jge@~ z_IS})VREDqOn~@dM0_IRl`0P2IFlB>Mj=NTALk^l7@0HgoL4eF%8-wh!GZcEt^rjN z4C{Qf=qX17KNtnLL;6S`L?g31CqFa-`oAobWl>AzQYEp1%Ze_H(^qfFuN3>iU=+{- zjeTAog}l2Gc)@K-aB8;u)^?$n)8hv=Q!PpGnN)s-4E-i~>8ohvvIJX%K{2yO+iiRE zo%>d=_vK$wT3z8QPSKxLUdW=5R#F*Xxx8$zzkJ%h^J~9lXUiLQ<>!A|HD0#yeoMmZ zj!hlZeGjWrQ@DcKo$8R4;+}-oZbR*1&&!Iy%fNxa=?7J)YyJ85hW?%%w$|+7M>p-^ zogG`>=~`3i@06CUE5QXJmqCyf@wJBHmH;}gG`I7ug!%&*Vz5 z;dQl_Uwhpj-B^44zzW4V7h2Yy+_H@)4{Tp`v?#%f=aw5A>sISZa8cdP&sA+^am82u z+qG?5keQ3%5yGZWvxqXdpVyAkB=TsJ#7~^ik;Ad9UL%xr+cl}Wbz8f3 z{k?D6#_Fmqubi`$6DK9C&)cKbHLIQy5!C&KNTuPcs^o~yAsWnSaFSGCosKUZQJ1tiwjpUktu_vgJN``ijkMdNpB8aq7(*9rxTdH<8?2!b#hqv#F z$F2&EHVEW)6J;q+U$US3g}-9cnmBTk!r(9Yb5}9f9RFk?$Y zv=4Z7v@i0*V?I0L7u;cf@8-K|7dxJ==Rk4KFI}}yfBMILqRNCGp&HTNxfEq0F2%dE zv!(P~>Yprdw0$MGTICu_3A@5U7)IxYnV`4{KlVx2>*5vl9owy|&-FG$Pk20hSUeQ$ ziL&}`)lQy1Ve@Jauf6huee2s_77u%2%+3`1jri0>(octcJmCyVag5=db` z76ZvA5ta#~f~v!R#M!)HZ}cGy*n#XQDgE)2n?1XA~a~ zpE+d!f8dWF8B10{_Cxz5A`&EnmvIo`&?B5kEBx4R#%hxttgBPsmkXb9Ps@K!3`B$2 z{sYOit5(~2B<3|~^DAGl!ql9iP$Se}$Cngl!JN2<1~g4?-iJFHP`0hx;Vw)L{hSjc zoR%l6bqqX_KHCN)21;j-qy`!Z5)xx1z_CCHa|}l31CGM@%pT)_)>x#uHM<3~^#Q{1 zB3$+b2$IR;!2`{O`*1l9*_~Pq{}|jgxMF8emN3jcxoj6`cVnL)tET=5m4QlV{w!un z3<0t)T$%y69GZjiz#MChjx{(~C!w1CudImr7J>s-F2yKWjaB_r$`dv%##WMW+?R0r z!R-yZdvD$95-!L6<_ZSMc=mU{7tg}JcIl+O`r0`wD$Lja@xQQz&NaLIsY|wW_OkWM zB|B{H+W1~8m@0z24+>)7otgwoDGgj?x}X7Lk_RRv#IkSDrb;%(LmjNvs9kXz#rD11 zcJr-w?e6BD6;8cir(XM_^%kGA%~mf}8GVp?%6$PrN`!J69S3mpz;_J5&+Q|{6;#HV zUx1c5jCaC6ohJZ?ykPb`MSKxMjtPJ8Kw+ZhjVy5S8Kg6W?B)Xdpo}Rtl1S|>l1kw{ zK7nXX+_VXY6rO$Npe8KGXRyX4gaL%c+|ClH|D@76eg>!42n_2XASTG+HYf`?a)TD` zCVS>R44N^pf-6Uvon`{h(|}3gKFkM)oJR9C3htWyQXa6At6`of`_sc80dSl~DIUQf z9Qg}V;3XV>DTCMFND|N?=sp6Yc18yY9(`a!1Z|nX zs&YKfOkkEyowk4Q%{}|IFa3u2NX@gAAy!+XKY86K?^$P+n@|5!g)@pxagRKZGO6+5 z+=WYa_3EeW#PYl*2nU*=wKPU-+Ra;c?7@TgG#AsPT5(@};pgnimCG(*`lo(h+z?Aq zyJDihr!tUt?o@X^sXtameUGgM0n+0O6m;}!&J^b1le_kC^|5VAXH_)k`8Z6^U$I~K zg}-W538bt_gGXQT`V_ujW2ib|J3*`2vTX^c+$&z4)`U-V=5BP_0QE<{m%5_G8y}bIW))^K^x$y`>ox&?e47Fes^7A>F-KIA~$0){Fg5?JW#W|biyYC zCr>We!a~8m{*^D;{QRuv>Gjty+yCW1`uD7)@WGoz8E^9b6ZAr{$l)#kgvXg_O0{~eX66L6E2odDtK7)9VH;82YctZhbK;&hV9O0JXX{35;R0%*I z&4Zf;cH+wMB{%uGNzRiJiaPB-vF)wjmfH!NoBvtO9N*B)za_@}b#=rWJ{w(MYuLGq zf6Hd)Uoh^eRp0baTqwU+SDE0|wDB=kQ|?F>+7y#nw^DVeBx!5#`kS`Z<5<#dfmr=<>qBdc_02HfC%XF_xgf?qSNqU7DFJK(U4)L4wU~- zsOxYG#b00XQ+5Q6vGJk{!`(fYUEjAdkZwNx>st7|KOfQ8h5hT&b{CAl{bWs`u^s?ZCD~t+EFuJIhde_;!xS5?nVKJV~*gVg}gQZT%j5C^a&TlcaM?q$>|7W zf9Dzc6OPxH#}j-bQVekPoOtpl+DJDN81NgS36lU5tHP}MkP>(~L*Yl_IUWy}gwh-b z?&K|}2Ouy+Jnrgl4suJe#4q^nb_B}-w+t?bt9+5i2t!zgG{OaLrZx^>MU0+AHYfj_ z&rtwB%?Yr${RoJ+gnP0I&!bUUKN&*E3mCyuT?|Hp?3^P`)$}wN(l^r2X4gae7#z?~ zG*}asRZ;pnlZ(@*&uU`Xv+sQQw`^x~!&ki9?YghLmuF^uf#I$1eoKApP=fO*Tf2GP z{_2mPwyP&5)QyWQ`ddk?6s{?5BJ!~BRD2SF5uuN?m8J{2g?alIA5GaGe*HJZv(?s8 zTg~OkLN9GJ`_X<1g=u^FwJ+Gx@>zSZdfPsF=MP1?ik-duX*+r9l+8$us1GV zv9I(S67FhtIX!}vVxuvW2^FRMr{547z6i7S18qx1U z$VSLS7)HR{+1;^e3BncSwIadPUt&wbJ7MCm+ky2YKsTEWd%{848}D1U6R+PzKZobf z*-!nz*_53~w$%Rq6Xzd;uVdoV-1g&rMZ)*Q)=K?40JqcQweEM4`2k6%)vLX zX8-VupuV6e@L!)fq);*@uMxms>8AQ2uO5f=doxec6;0kJ^0qiVzwNM0$fGs=cWc%2EygpqMcfq)9hbDsu&_;055j!Nk}JC=T0uz?DV7t60RE? zx9a@7?T<}auidxx`wwjUqxY@1y(vMrAR$+S2xZWq%Ar5X4tTw*z9}PjLH1$5MfQjm z{8Q#ZPYk}kZQJ+m+Kp=utY2EP7k~CwtXZ11uYC0#d$ji04})>5#R0;W!XK*8e%Mfq zcWhh(rV* zA4oI^qdY%MT*#B>4YS&~icXmsy1iCp=B5qsJiPQ_7|Ii2xs738;fceq4x+utja7Hg zhX#JaaOjZF;fJylJ}Zw$e(a$RobVsbKZh3m42_~2=&oo~O4s8jE(tz_Nji4SD9VuI zcSU|3hrCCg(GL|LpKsjX!-6-oXSbkRb`KnH;F`p_!c6>>ne8}E59w0quEu5euMa~z zI#-o(8^hC--%Fg-I6_TU3g-{fQ2&`gn58teRezO|+d#>Vdk~a16nZwZg@~)e{K;9& zS;;=7J~3ISxB%;X09vRI-MMqi8WLU$)meoXJ><)ak5kIGzaS=|=cSikwbN&>SY>w6 z7S3I;_Z~DP)XX`#STz)z5qZ_7)Q95slTh9-NlnqR-GJA?{89;piB(|LLGH)2K?!H0 zFi|;e^D8gd3$Ol!UA%Zj!dcZy64U?=R8=Z767uH#&RYaAR=j=7l;R{W9&`G7b2wIX zA(r&uU7oD+N3&;=+PXXU|toZYm z7wAkLP~-t2m2oWc!Q(uvM#F>UDI5=%!zvKY5l_T%pnR|Zlp#+h?6BjD6uvz|I{9EP za0tR&BN>PP+RdiN!A%y^5}a8PE)>;v6b1_L#~0WB0W0Hsevq=+s994&``+FIr+)~l z%9s>?MFUn610vw^{Eg){mkc#9*(+_{y?h10aJ^KvJ zxotT>FuXO&2A6f!zqj5E?wdLrU$MJ>&uT!!gNTRpHvyRM;Z!AAOR~O z`2gLN1}%SwUjht!yHGr($w^}9$S1Q9h{dE8QD4#-pBuSGtk>fvgaezKE7{`WtirJh zb6|TtRMAKI-) z8|Duds4R?Q?A@gOvCidgDCK8d!o~!5*^U=o=c$lzXJs$`d=iN+1<)*xs$jtJ@H~ebI`ze>8Uq~zGIf~yWTYLsgU6?`CM6W_ zFY!uRq$Mpu&I31U0G9B{aH9lUH!1BD`Ws@n!eO|1ghk z_OdB%2J{gXlXK;ja-i6giQN@p!;d}kl-0{S&@&!5DnD-OMgH{!A~cWpw#k57mWSh1XP!N_lnQC}_=CvCbiEkS9*<`x&N(J$D;O$k|W8x;g0 ziS1@61Wy0@L|jtvh!Btutkdt=ZhJ?(As(i1oMS-#;D1+M{i6NU&-|jj{Mu)1er^eA zPkCwksy2rm5BhD}-Ch^p(&SiefP<>U2X7)R+99jkzD-KN>teJ^?^F8Y<^jJk(_*3K@7!zaxG?$AwWlQG7FD1~ryKu-SLVN80 zO+Xo_Tjj&^3hliAMLd~eop~& z98?Fg`ypr1P(nCm-r23&h4YuaJ_#oj&!-m%+j*A*VCJ1=P{Svs-sJq)y&|QXs zA7*r@8#Hr0oE)G5<`{F5qrlB0_JFW^Pb6<{lYPYOX(Z(nBB^H$6aObcGCwoe-FoZ< zb73(NU0f`PC>?^h*}W2jdzwnPDE%b#QU4(75x1m#pd3{acn0RD;JcEaQi!xh)TlMQ zhYs$9?ur~HNdBmye<)rQ^V7X*EaPLRtU7Z=%zDOJtqs3ZwbNeL0I+BiwqV^RpJ(2* z&hDz(bWRs)^Y0aR68ZypMG{daJkRu#X z%0=p71oNNWf)lgB^^E;r6T}M#^hit39~uX46p1+RnZPrbz#C%KLEo-dax4YNw*y^JlYfhAH?z6vQ*LDu=-n~0EErze5TlrHIzE6~bDnBmh zG9a?jgRmMjL15x9S^_A6ZVi6L&}*M&9OPKq(72Mp6re}oyh zNg)ghDVh-Z2!%b$n8M_Az#Eux#~$I`k2n{gB#&-A2AR7KCTGQkEeM= z$zNWcKlFryeLz=?@^pO92MSX*58cB};S@gmdHuP&+(;+6dmOmo=PGEyPx5j#D1&=u zcZ3-UjpD5cBnW_f_V#u{7$&_k>};&tn18da`@+d;UcKV3MTc>I0Vm-Fl;6J_Jfvfu04p1kCb&(|J0BDC41rdS8ZHj z4kUPS^&D6CaT^ZXwurhezRz~K(`PT+?DUNHMe@tJ2OcQti3jodpI7%{5;JK_ODAn{ zVZjAGA0PRXPW`zMB?WJZ1GoKw8R3uk5PZ2Ww-~o67g- zljGFwgk8S!DaBDz9$2}TK-yRS6=t`#YdbrPt6eLW3VwicVPVmqT8G|#a5f3B+zGv} ziNMZw%@6mktNv@X$I45`TcY2gE&e(-QB|r;TDh%dUIO;K&CV`Ku%6M_fKa^Wd7_w0 zOAA)iJo(Av$9D1jN&BT=_<5Bxcv0j&5zp%>XcTn9J=%CGW;7*PDEIO91aA1rsD#lm zMqOpR6d!i?Lnl&HnBC100^ePU08{Gp9#>M~14J0lw-h*pb~ox&VF^#&;#aD!c&KLu zKQduB1k0e8)5w`cB8~vv?GQ>9@UlIt0rCjAdjjGNLqsqLOGg52R3*p^^l6Hpq#c25 zP0hi7GYrx^uJB9hI7}AAwEL`hEL)*`MvbGa4qOv6dg4rEZ|{*hr)X$3&H6XhA-`iA zj~@Gq0r$7Inr+S65rX+FSXD!1m3Ug6deLTPUlX%DqxiT9q3U;$j>*5%+LC+07M3rI z0X??b_V*NDU2fZAs+&rOr%P9qZbR{<*}5wvDEbzq44GHFZFe1JNI2|aC>dfmf=b?owG1;nd^aOp4NfPkUE%Vzo?0)DJ2 zgdV9RB;pRlfuh!k7qeHfBjZP30vG?tBlU|Z$za$qK_)4v6mS? z&D~{|7Huf(u7FXyl+zF|DOO2o)2s@5L8NGp2mML)Z}!8Ae{QeYzJxr#LQ8nn68k@i zK)AlD$hHg7M&TZBZAj4EwX2IIry$RlbBhf7U`ZaH(hs_S{8(z=UUogKsdO=C&JnoqCz*!yz$L*SCmp z-I#=56x$*s{31lQ+R{a*TB~98tzFTO8yw)V0i)##vuJYIv!xRkeY*mO7K1cv&f4T9H~z#dy2I?{Xo7|-5b2*Vjj`{I??ecJC>bADFVGnNK6+Ip*cJX)R zt1xv|^>E1+RS(tboP^*x#WiQA&z!O4(%&x={e6N?>*&FBk;)v7Z^OS_A_&^bQK4f-O z!RUpD^7O1X6Zd0B8v6adTD$tz5apjDr zuo8w@DduzU;o+nUM-tV1Rh0c&K`-uoR9U3hLNMGC;bM>vWL$S9jxE2wm{Txs0NlZV z4+_K*xYgYm$iNHSP=PZsrugJR+|0arB#mtihcI5D@bVMS40A^=osV|L4n*1n;Cm~T zhQ=uQ;UAxS#}y}oLtO&!R_o9nYnw#!OM>%dy@`LV7TZPy+= z?%J(K4QnV9CFT3IOAGe$b5qvp)RdkCR~3kRVkfNVOtPoR5Vv4-*@H>hgwr{g^ygt4 z`{TCTJ+Q}*ckK3!Rl9rVv2Aatj+Nd-Vb-=xjC``JxDKt{Z(HNh6RWN7hH&nyJzeEb z10&HY5R{mJ#?Te|Vnn4s2im*SpDqUTQd+)Z4r3z*Ik>_kB!wsbtcbi6oxi-g=*(?Fh^V& zcR;>m<|9_5IaI7-j^cfiau}`H<94Z0{Me(5Gc-z$g2_qleR5QKxjW!>!L*-zQ!=U$fJ)m68bATEK7#)v@HleFDwwiUK-({uASRnlZy>4tEr z_|!Q09e)mf|KO4GPrQ8zy%+3-SDyC+&n2#n z6WvRylM45b7pF9#ikConYMhCb-&mx6?$3HRC15MxrRkE&rn2iddXh3qVWoU=M*nOj z*;k!0BxHpLJSo}|;iTL>Ab8)<0i!G7Mepz!6F zjzDq0mZy@_O0>t3`28y@sB-ccPEto9IV2B$y#iAs$&KZcA%$aL_llLBe-DQ{D#evlhCF~d?f^gGBz>UrALfMb zag5+Z!wtn}>#;FbD7fii+RD?T%+nChZB_N3d()fJnW zJufD(Y`dGY)^74?QH_^$$6#W;V1E7ym`@uB#p*gJF+DfK{F0f~3YZ2j0Kq_@tqT%t z4;W_)#+35Lem*f5`AakmX6cMvsG^g1>OmSZyOO6vn1^;xKy51M3-sCh*NZze+56@= z@^|5mI&pd@7D2h>O&Wf>TTH4YA-Yy;xoS%oZAsNyecP4$R&(Fh>if3UD_OfYVLSCb zsY6@e>e~|u&X3o(t=S9aJzpNP^B2!ZfJU%mFR27nF*y#`;P0zWo^LYX|B3M(`YC4y z4)7Te*blw6S+l#hAKIgbYc?TzRW%~>fJR#bS-0J^jkO(LA?IW>arh)_Pjx5VBUf6K z8SpI#g7_VQ%;3>(Z7$T0*ct!y0P>opV zgQ#rJfN{hMWT%fQg+tGf%}4V-Y7WTx&4B4R*3p;+3EzJ)9|<4W*)PW*D%#bMUrD7E zw&=s5FuTO?XM#vO=RmQCu5>aC&FDjw{{pf*F!UdVjFA=K=0W*F>E~l^+1=GBWP57E zzT&7qR&wDBe0ibm6P{H<5y$xfdsE?A6UD0aB=COag9rBQ2U}Lxgjw9lO8PBMAnbPS zaB|Fc_b2S;R@Z*}?HigzH^sNdtuRrx;#4)_O^y20`Pc2WH$ErwOR!M?j_n+A;EYvL z4)Sp}fr66<4&p|h*zyte$FNQrn7k7P9z4bYLqWNuaYX6CJG}cR9smbK>fyd#!m$F% z)og90$Lg;iFw-P|w_cMw`-Iz8Yc~BR1+FAB4qhLpEt%w3?w*cdU;Mw*YO5SQ-+K&? zXCVX3%wqDm;WQVmeRbA7!~=y3EvT0~4rK7L19v8aDQ)~CU*rLI(NeshJ;ptR`3{R1 z@Oag`=*QOq?iJ?6L)rqjXCQ3zMX=F=9=yz1w^nOYagoP@#^lR(hTAzR>{Vt`A$NS0 zAGWNN%T>|6?1DWnY5Jm{vVF)_9F2*4noQJo+qSTukd)7L`|q^iKg$JpolsTC&j zM0)|E?4vZufXgSjKNW&R0niHtFMuv;U!Gr&6ffFY&cjoHJm&&)V$5XC(+C^`R0_gjjw1J8&M(-x%dcy;I%}J2cdXlbUvuM#wyfW4D*dWD?nyBx4%>-2OvItA@rebcd0N?#pnS+B_06=BAjJx7T1}!Z4tNzo zkZDahf-%ca%!KSHhX15&saAU>Jb`(Bp^;8U$l>mH*t1 z?$Ns*O71X7vX?rm)7?!hXLEBGm4kOlqGx`V$v8PxAQW*wmWf+8ybaFHwSK0Juh0)MJnDZa`NjdT| zIA0k3%lX06(HtomlY7#S;EY1Dd*5?o4ij!J+m>Z{AtAuijj>Kf38?DU0~m6H0T;$naQ;_?gj!b`9FEe2jIV&wFHwkWs&DBi#- zB{wYiZ4b&T5BWH+Q~s%E+{vR0jqncTm5D!FBzW?@%r$ZRrE*d({1J}f`P{eLV>^}v zRklTNIIbxnxFNx}DIvHnAsEFfHNtF5f_=ArbbAC4Z3+H4ge*XDV67#->5CGY%=eW~ z7mDFWj*P?F99>;)Z57QpxMTC(l%GQ)%D^=gx(S)X` zdLH*zB+J~>J{kO%F`6%z2jbbSc1L_)e0_7<)*oy9?AJ7=p0|@{U$u)DFZmPl(4V|0 zyS`FQoaJ)Gw_Fri754J0EUFtOJJhM)V=d<1Xm)IRX2yk2ZW7`v1vH>us7oIYRCLDn z{34$N&BT>{pc9pkIr$xd?0yVzb^#r;w;n0z?2(}W`bg+M2qIVxo6^Ki-*&f9xeqF_ zRH2&X3y}>n(G6yB%%zIdqSOrk<3mcT10kMV;c=eZ3SvHWM}rORNO{k7zRN! zbnw9UCP5l0U5^Wu;u(fbDyM_epX0hc;qvG9A3`CWCwfq)Un^!toWg| z8oQdMtK1V6b>6BvXG?KNmBeht5ZF3#a9GnU8GTEPsbW*bxyYaAkwWTUG@zr9cj$zf zymAqGL&8{Xd&_Ed4k)yICf@BxmzkHv2*ZkTFf&$CSOs7vOoWsAHl_R%EyDvX04wo&S}G5N1y>X| zC2($R?Mm0&U2(H7a6)l#Sdr(N3z!Mz`4bAfPx)b(;KM({?ZuUf{nYErvhit<0gX>Z z3L|9p1~T*K@s6EG5AC4Cfk;Gh1o>DA4(1qxh9AHnCDF|9|5UqWPw}269(x+FCYn2T z@7gtcu->q?+D*@e=zU{8GkM|#3C9)d^g71tg{-C!=Ky3GwNsu9;f{-_MOh)1BVg7q z{YeJuE7~(*96_G+wbWk1<*2TohyxWq5eGGs%&y!k;kb|6KL!K*)LuQNq5ko|h-uX2 z`~Kzl6(&LkI5T%XA`O4%AAy@3Re|7)gu$F}M%B~Y2<5uspbbU%~Ur8lgAKcwP z-0>LoAFAY@>haH_fFt%%^T03iB-=4H0y&)W%3pR%0f**MHvy&reu8Wtb&pt&{t=H0 zS@b9Ip|!iWtSjMq)NB8XUR#SrUmVG4k!GjIE zCL#I#dz;pf@YwDjT4{=fjeR?R`3?KA4Lb!lG5eJCE#0zLe! zR^Jf)cmkPpb3YifOV54EN`9`*`k3^%{axY zlDoB#D+dzQIjf_zsh6u|y)LaR+jCbg+82K8Q>M{dd58%M`A5%mj7f#JT(krjl3NDhW?BZw zoa_Tv>{*4wh^GKXEeHRDe+a0pki$ba+q~WBb51H{R3)D+|qCF>MKJ!hw0iU=d8DSAjDqj-3I%GB@N3R=2g_dRG46v9{KfX^@HHdv#o^fa&S#?D!Vo| z71t6WuyT4?Y#5nX%#g_gArJf-9S=Gp0%Zwykg6_bV0Yt6j~W;W z%eK6<;(IVV5`Y^LV(SeF!4hhDnT_pKGqa20;Z^5X2>H%`6d!^af*NNZc6Sv<6KcZk z)*2cUl%{CWX?2xeN8>_q5XNcC3MTgp`<`^jiDT$N*JyRwjE*qb>5LgKaKh{1N#utE z-0)@Jssfpm2HQ2L8~B{BX3N8`P?RDrgigOTL-hupKS90wM1i^UIMSm%U_O80W!RPk z;75M@#m18-z9$$VbU}jf{Nl2elqVKxc#^rg50+(=p zN+i^m=MWQM!>xWKZF=bQ-)!lMj(!_($S%*8tq%m(Dbg#FI>_^!(?(3CDS?yYQDgR;W=JbszYHf`-`* zn1?F+hb17JT=12}nYzIkkhIhp#FV+?)vPEcI3-1n5s<_f?YXfVM zT-U@9uZh9TsRK2vUcYXSS9ffCV_qHmtY)Z-*4vlR>3iSCZDzV?t(`S{a{rp`Zru|j z)(o~k>7Q}B5?H(KO{opr>+LE{9_I)e=hcN6gn~)A04F96M~AuPt8@^^VqY!+nfb@( zIS!y94D+^oZD&IhrVY)^C2)#~2u=BSbY}!GK*IToji-%Y#0~XwK3w>Ylmv8UF)hW% ziD>`ys4%!W!*+r*6NgzN4D&gPL5&+BVrv2`xUT<99U_hXQxk8ZE0c3Y^v<}VE-6xQmH%m3ZLicz-E?D*ti%o zhw3O(EsNF#hkSv?M%_r*-Y;dwlfUk!o9DgazKQoY4QW-oR{kfTV-9P#-W%4Y9 zhvPW>AUp7b{EtD-m#gG9C_D6};3GJr=7CbCgHQSYQ9t5w$8gW)u!ozwCm5LIHc|!Q zA*H4=L~^r!TxDmc+|!lg57i)#Sy7JT=2SKrU8O`EGVZ832K>-{@|XTdf0UL)7r_@Hlsus$ zAk%-DeDgJ}zo#6CPqTH47v%T?2kYP(G#(96tYpPkApLd-36JpCHd`KAZNGu2US~mq z)ld%0wX7@qzWAi4C1JI|0s}Nw#D4fs@%go#DkgLprEkzJ${(}~b9V3Vt^4Y;Z&y<} zeFYfhE4}b47V0%_NI1^hD=4>*E7CJ<0eHQB1B*zzcY&czZ_^= zm|w8D**Pn)*r71pu4o{C78Q5}kCk4=mif8Uwz$XxNwYRxE&D1q6CMOz?j2@A0mMFG zo^hVhWUA;Z;f&R7(Gpr89&|!!62Y_C@NoJfTTvc7+O&W2&;HQvu60x%77H~&Q&EWD z%cXLvH#CFU*PKA>Pn1BQ(lmy~&DFn%hDKR2e4B1j3H8#dQlGkjf z_El?i-mzk7MV*wH_mVB1_-R{Qc|o((IWd4Wb;b$X+AK)HG&y<83g@11$L!pNlm6MS zRojs;{6qqE-HL^hRVpWJe@}vPb4x<;ZN;&vPOMJtt0gL%&mMDq1U=})lU!I{so1oC z`qSVc0glVgT~HO8@6k(L4>y>fDnaqrt-iZqjoq5e+}{X^G@uCIKNpHw!<<~@&O}`# zJpF{i?gJ!kfx{up#wF-F(+$_a^?-x0je1*XfY1KTx7gR-*Sf^8gIqFbZS4HbOK(2$`Oh!+KlzuQ|+)o009bu#Ua85ME@F5 z^kHw40(2BOuFw=Z<>C9malC9l;^q|)YT$kh$AN|i;1?#xaXiPxJ^hWiXLI~PgN(~> z*atWSksCi@r;jDTGdz-CnB8pv+3k}-A;WN_GEz7x_a8BPItg-`NppMf$HXH?RQL_e z9&P>`@pr$0o8l2q^6>>bI`$O*ait&q(^g;#X|gQ=Fbl)2R$PUJP=%nxsw~2}#1uQ| zHig@<307N+@hV)bxH2}V-}5++gyjSAWBwRp+7fKW#`iQaUb4dYr1P|iQo+hn@JGI= zA@FglL!-d~wWj9*Qj;h0%1UA834A$3z-7NLx?g>!Fx)T9y}R6Zez?!05-La91KeU8 zh0?{JtrVDDD3&mcAnjLkDt5{b-}ODPQqZw9J8NaBVx??FjVt3*OF_qk4`e&wv^}CL z&l*1{XHRs6-h@M#g+2jMHgMxB7%SPS9m{JFy~0IW@OPD8c^^CACC?Bfft(k9hzgJ0)07n$gLI4R!wC@V2_0t|8l8ufUPh+rM=s%+xs`~*+2M) zziC?=n<`V^zmhUGdR?fA(vC8jC|M%;Cj#L{>*EIo7=ftI_<|o}y}Qai@v)}?VV;@> zTzu$0L?M64_#=cL$u^1FHm|jW*+0?p$7fpHl7|ybW-8H{n0pFRhXNSn=^(LkAb)V> z?)Y+7)Q<0+(BqKCpW}g(T!V<7Xv+PeQ~SOqTpmFbS*Le<@0~f?y^v+x=mF)EC)3$PIS$W%#AiQCH77`BFyE<#t zrK=KR%M!;otzElqk5|9uLfClWIpuQSUrgh{2=WrbJLUEQDn1FrrK%N+ye&U1Ay4s4 zmLv#r%}m@b0V*)scY-dg%5t)~r!pWka*=#r46Sap-Cg++Ha=U%pa4b1Bs5_tR%c}w zon-f?Lp5x1@8)i;r-4WUv<8$3UUDl4S6ODiuV-0`8GN(HfF+&PgPsIo(TSB@{d-*g z^eC~@Zp8dpDJrNuWeLMGCAGT=3I3{=ey{D@6W9Xy(&dx(>cwd>PSHw<`87sfe~L)? zbcQuq@aK@JQ?^SioI7vLJqhDmy!THVVt+APJQ$Rsy^v|Yo{))>6m`nr2Lp$O(Pmg( zaH4uS{Sahz$-1-W#LPQZFbVqdVE>*+>6jycpIo3Y<;D*evQB(J4ij$SNObPv;oZmo zD2D@O2G{W;#lXs#c!kOTQoJlyxUN43%INVx2f`$UJ9xyA!{$)9xy#W1lT?O?k3Zr` zagE|8cOcsbW~p2k(qtbh$4NhKcO65biI*!pF?+oW$#J9t3r-VhMtTaPgkr)XjD!pE zT+?6VBfE!+uwmvJv7tW^XGMPr_qfW<6Ll|syxzmx_3Wk`$t|Vf&S8$Qw627n`~jct zdYJHcTq!TBbnIEVyYp*V!KK(9WEh{D9U;`fPr5&cy|-s;_ikv;+qUZTv`@GZ7!eYQ z4*?6Ihx>8av)h$U-{K42#@a*my{4zbea5B2gq4dDs8w+IjdwwnsMkHTYH7x1=2mQT zd&~I`4`#43$)|Ky5SbXWO1oX(5igQ}r;sHHMQjN`K=!+qcvP}tK+=<)UTUXk_4|?K18e${}!p{zr7CcVEp~|>juiK6UX!ZyDf`b|eLRg_B zVNCpOO2U4L1B0@Y7Zy0E*gZ||5B7JIC2Bae89~-VpmI~v1IEW@ZJIs0nlSjS1yVdI z4NoTw%2##$>!&Cu0nE3M2*?+!-~h3528Bm3S{gIDEe`bUNyUmjK;9ayzVF*^)VK}e ziJEcIUVQ1(wtV7bjQ^rZOSEN$9J;Tqab2AG{lKSAo$=LU9;tFcRF%rBc8t+n4cBgV zHI}k9V%k@QOGWW=4HykgVs>lW_TbJf+jw$Yd14WXFQ!2!?gr&R|-cHVj# zOF#PXzRf7jGpFW+&(~kJQ;7eo<_{1QZk9KkGi+8CiTuH$=z_{rvRODPL+6{&144|x*Olfgk|@c5xZiDTf6Aba0{P?W z5J3bX7AtAf7*Zb8DS6(O87Ef%%r8N6V!Y&nuCsftD5Ik%=EDF0vJ3Q{E{eRXB#JwP zW8x=GX3Ts_DoQA;sAE00!o-SZrYqLiecLvl{IRV)`BO1>_5okA%FK#7tr#y`BgV>N zawS_>UY2`ROsH=Mdu^MWU$B+an%PztY_c%t_d?d2?@6HEw%U_1b@XXlIsa2C6MNUx zb@-IU>W7blFff&$j-)(#n^)FZ$tX(j<__etX3p%XiVN4}hfv8NMsIeAvX6@?LI6HT za?%&zQxr38J|PkY=7=7Bi`C^q>=*VrhwhrSi*q16Z`Zl3o;pa|7{uTU{=9LB7cc?# z`gOYdzPFe#J!R3$LnXNG)!?hK+|B9F?W#_&msK`aYO56q#$^T%l~>ugv{JDbo;z*h zY6B4i?L(T!6^1OdASiRN#8;+N?s+j`UNNez-ZjbY!lV7DO?lhYK&!!mywM&QJhlg8%~APrL(ticf|#g=LvjkpZpW4P zv(zRDkn@t%i18@6+h|+~$A9p4OZrVe$#CRPZiM%~o={@qCpY>F?gM{Ej(+EU$~UNB zCp=dJJKo`j#oZly2yS-|(opd|&Xg_=?&Y!gA9w_aU-vy`3iHQ7?u5t>;Py#M{hI_51-zN z1Y=fr*#pZ2sysR6d@HZ~(k~EX+5g|zS+!oL?+2tg0JpxmX-4l>Wq&H4&Hj1~DBT9P0vUSDh0?eXJPdvN=f zby^RU9fV-X0c=?#L5-IOeeUkpoHd4_B$|%=5Asf*1wY<2f2p_dYGNQZdMp{w@b+*+1x-kPiP-g9tRMSxzXbq z9Y`>e5h?02_eTeC6L@?=8*%zKWX~`JNkoal2k{ck0^|__LV%m}7rAa?id6*e z`F*6EY^xJ7Tjw(fpUSy^kUdCU;mhhYA$GJVNVwoBchV+5Uao=_%Clk$yVh*nwvCN9 z?eYCTw(YeCR@+p##=JGRHH&4RuOcZ|7X4Gt$uk#hx;kyO-5uLl+f;|;8A)~g^0amK z>sBaL6}>w9VZ$2iKW+A`$O@!pvj~$2f81EXimgiL%+QC9Y6|Z?o5ljD#a)` zK*uD3tJi$L7))3=K+tZ|1!x3e@ZvCq1jb3xKxu#*3TCNvvJl`9bxk~O^LSlYJ5C1ZRdFUR(X!x@N9Oj{KV&op5OvH)iPBMEM zLp!E$2{#81J5qCc8`yekU`f64nIf} z*B`|X9y|(1>f$MYFwNPA7*|DmrArJs9QnwryS@OVw8$U4*!hb51CRP(-!bC|e4mx% z@bh>bHx>slqw~nqoP>NI!(?VFN~6w!amL>%KghVWxNLK?^Wy16Pm_GIg3IbVcLQ^% zknMAd(5llF@%@TViu_4vjd7cs>$djbJqgY{*PJF&4$8;Ag!KcyB5)%|%Vy@z+Dosz zV#|vw_GI;;ZLO_oysg{GlgkPhd1jL0c~5k9ewRt^B$Gw{nDN4xJO*gMs5yBD9P!BE zzdsNp)X+2p_{PUw2S6qhWaEg_GlyI|KN1MRB~{i45eDOOhE%~`Imi8u0gVMU3LZ5H z(R04^vrr;gtx>f*KKqCiGB3J zH|_R~d-m{N+nSpa$oHx?!3kCg%=OxO914&CJ2Sti&Nd|m+EwS}MWK>4+fz1Ke$fsm zd9dS874J8!#X}Qy<)=SsGt*1XoY?%&KmijXFWxy7YtT=|OmMQV3?l^Bwl{5OOS4`v zeFS4>*p%IwuyAr5XdFy{;UTy+#i%kqG;_u|z$Y;n9E>wQLj!|m;CR78aFYz{f+s>e ztG}Hddux@RRF7>HVw|jaAas{X@wpQr*UcqPJu?uHVjPs=USL*HU{eJRq~p|)>g1r` zv)O9VW~X`EzTMwwkp_{q(ghBFT3+cn%$98XcWQ&!mq3UJ*~ zk_%?F_l%UI9*ql!pqot23qy_C!9nE|2Y<4T*yKkFp?uXFCxOS4aNR!e3zw`MvSHvL zNC+Qu8g(E|CXsTaz51%J(&ZUsZhL^gc>k0i6KaGZ&LS{TEth6JzkGe+ixo|u^8qL( z&oh;Rgsg&uu#)h5#!Wk{EF(PgEO1TyQA35t8>^gst4z;$|7c1`Ai8`x(`1~3)gh$DxTM*=*AZjU;X%U=c|N&c(-@^VxQAFmBol$c zOvQ>4erF{taxGX@Lh;OW)n=+y=R3Xy1bWEKf&f&^{XNFNAgK5E?*n8hgxp-C377k`W|sS$4o%8@2pO8oE~>7{xDM(Z$OSkau-=yf?0Q_#4~O&YP68rpL=Fp8&} z@h2Q(OWcVseyn<7k1uCU_e$XOWH96&$CpX3>*pT89-0w>=edl3W++;9-P=A@g*g| zF)4QuMZ!nF{14Cwd<^6k?oaXafJ66a6r__h!iLd*eN2%_?GKx~9)~>hbRp|N@v#w$ z{l0442$vxkBY+^ltxNd0Wu4{+noYhZ{Vm(wct;Ef#?`dl^#@{P_pG_ID(0k_rWmE~ z4blu1#uIyk#gM1C#)Va9+4(#t;fkUlkMz0?G3HHmI%cU$Ha#QRu)g8ibw;y}yI4mau8m|wr zj}4)>RA6uMxM)coVUf7K5u)KW#7UPg7Bo5l>rn+9Fr= zo}*`lhW5er|9(?il5o!UQ-Ztwv?NBu0lI~mqMckQ+gx?Z zri(GjZ?+}e`hlvh)f?P$&=$SfkF1HS(%~~ZdwS~_VFPYlX7ISkPvmL1Dh$8QV?kf_Bzh}6V+^bBgysAP&ncHnC6%uveji))*xJI_ z!+|;`99-qbmOrjfE0r}at3hxb-Hs+K)a&DicWuAN*(ll+E7punA(WHvoR=3~`kXDy zF4*S!n%%v9+cr1Y;&JG=Rs7P={e)e9?u^o6#h!(hSP_?T2*j22&_d4;X(bD7oO+f+ zP(~EObAZE7AP>SM|EDS^af)d`ZrpS17{g<HJ>Rb23(C%8R zwrK~wx^?^(g(vDrE!*qVCBUxvJ(7*xb=zy#Z9>e7j>IfAUJQ~Ds?o4iT96w8>a=F5 z7c^s)05!{-?HxPw!e6w>!o0N_2$L!!4|42_7ECb5i&m=gvfY6-8#Tol|p|vlBMDJc2d~|LEk3z&S(gjiv5OrCBjFUee!{zFeMC-U~8w6agVPR!?EFY)_aH^XJnPbyrYYQ;L@f$ry~@lcS!MEB-yJVlhfD z+5%T7jagZO^wNCARu(j2P@Mj?0H%!DqsVR2gGgS+kG#aN#GuHYX3`~>?M~l{b4#{x z?xIarrmfZF{!g_t;jx0`*Ca{#s<7gr4?!M(wpluZH(Mn-8ocY8=(P6>HhcaR<0UC) zh=TadfQ<~IkwMG|NbVWn0YqbG&WpjdGL%l-7(DWhn+7C)G_wto;{@eF#*Ma zih4(Zn=#T~az`N;sS(N(3;~QOnO#6l;X_530ZwvrcLI3aL*?$-?0z_9yL-F4`(>lz z_HAnk&j-|G4>#%x+65nsKtmJ1&et8?e_VN7_|j&TMhHHrgeP|}vK?HRTz6deb59>d z!Y}MY_o1DyZ#a*_z^?r5by`+if8uxR`s;EMiV!aS>N5JT{%99|t)cq7^E2fUVGcoh zYwMw};PF+7ty8pF`ed_nU~TcBdMmatEzDHy^3|)>6wl#qUvk-%z_qn40eNRz6X1B= za%+9l`Esu(A?dKEG{i&MipMzPE5CfmqL z#HknU^*4UX$E&s`B+xo0G#VGu7>k{S4OaeHwdYY1@S&S;i%=aBC1cX|?ydydbz6V5 zYP*}aG&Xaf&{K0g(q?issjwI9;-y#Z)S2hq|G~q1w!N|8i$9D570sA_=BGbv|Mvgp zZz&&4z92fI0E0K}N###v0;t2B7VRz4qg60QhS?R&8wcEi6bdnfh$7*nIKuT4RE`(r zcK+SuegXJrJ-cN%Za?Z$o;i_B&jx-Z!f<_i-3L{XT$Kyq)o5Az_v!=UcL2i8pOKT* z6)a#tJ^ZBHO#tI|CMZbNlNtc7;YWNiK*=9J=?D$P5d)4K@skTVVzt)IsiViL0V{KS z7Q;u1S^L3b36LL2@co85>0{gPaPvc545Ox)C+uLaW1Fi@JM5|x`-<+CI(gF>5Z7%n z<44Gy(#&*b;hYt!%j!TUtXP>>c+Y|}CpPt$iWQ|jY5QV+rP6uj=c=Fj-8@e5X6zG&^*s)X|$<%{Q<_f4G4J1F@>NQXW!76jpz7*w;tTkdq^5Yi|E z%=_>_bY#*ZeCX;E7%@74gL>05b2dFYCnmwbqhuI(q+AeIYB1%V=*gkEVtLLgRVD?o zy~3}abM7HQlCra6TQJSU+26$-p;R$=+=vmUs}YKcmlm=wCd8*LglPoIF%CZ{I^uL; zUy}-@?(VF-lkI}iYW63sG~TnbOJeX!w<|_Qc@eC6CL3XJTy4f%tIU9JVeoc^B?d3Y zkOQj%T^2j4EgYmnKV)&jFtZR+;OGw$B18GhP$f_*^Xa;x4&#oI=$WFLrcmh zaXAnC$xK+!S4PT4xxrsth)iA%WLjYnqCBqT@9ATfyJ}DP&`GYKIflS-36sJKhBY@- z{+r?^&rYL+?;6Sz)K88pdGO~ocqgG$puEtAb>A3j)M3V5XlI5{R#LTDB^J8wr^s*w!M1KgAQL-j_Lpz}Ff0O6Dz`FSvMq*?#qJ z{$+dq$|+l4+qI3YhSeJ#+ilgYwX-9^Ov13@U<=p&{IX1?gXI5?D0pi6P9Vlb)Ov(#?rCo3p95kDfKH-9YQNpiZ*Qhqg;~Z1sH$F*M z{(GViizt3iyZ8-zmsug*-r|u97CMH+9!>9)VE?daiziN4Sz{Oiw@Tq--soOr@96x) z&sApM!lov`o@Ktov4xHmThO>unTbTdw<-F^>OKG+$P4kIFD+lSU;3qA^~uV%1o?iy z>v3{Z2Ic1YW$K&#!|{Mg-5VkY>Wbx}O-m4F+YI9jw?)(%bz6J#$ktZh7Hvp7>M7Et zfdHgGW~a`-Vdu|Wk$gbhJ$tnJP~o?{t*9n^v1VqrY*(*bu%G>zH@qFH$m$md%0l2w zsH1pa$@C>1X`F5GMH%mW-Z1o+`w-)cXy}*`)_rBgFC1}411G{y${|Fwz)6^ap>UBx z4kM78x52QaU+IpU!vy5eL;WZ`Yhf6&J$p#R5A8S!z8^dVmq-Q#PWlMJ2QcE0l6M%L zd1i7XemZakaPkJAzkvXas~QU{KRJOh|E$C7XS~!iZujr~mKbeALddT5ZPR*(H*LJQ zV`nZdsbie9`Q?g)+>RZNKeFxZ57a3hN%)gss@XDvIX4INd+cRuiAn8?5p$?)N*$BI zKnULP$OBs!c!Ibr=5U~y=9X=&Z`;nc1g6@aI@r7|ubi~m*#(=f&e_t!vfn7tX|=7f zQ@7o%Etw^hizyZ)gm(8zw!XfvPB&$h*#qTq%1V`K+uVHQTODTRPB;^3w>AcJfhK;< zm>7lc1(xDko^rJ+L7nHG!3?GV-RUwsXPjr@Fmo4TRqz;iVzmv0fX^)y!K`_3V#1jx z^k+qePhBuft_+F+L3v6FTuvd+h>4ZO%-H7!^J5=hMfsayUvD)Y&Zv|$5Qy=_VMA_h zVE9#xiktGda4%JB_IwZYgz9*|hp^i8E6tuXHCXI!ieXOsWWe_yDh8M<1L&Y-=FAD4 z0qMt;$_|zA!yS5XG5tgV=6hgGrODQs>ddrVdj0ces>|ID`{gt_QrnyK&!0V;lsG&! zu0c@k10e;{^B_k{&4oie++pgg`SQyHN7PP0%9=wkU6@?P2`qPazi=Ogk>JMwpP_GV zH?G`1>YHN?9h}fZ4a_4UK9-SDzY#d@8f8rG{7AW=i_sHNgNacFjL;GvKv;<2?&yDH zjK4I7IJntW_&bc~ZW_@;VL1jl3^0n1mQ;54-{hCe=-@_bGW>?LRfwd%FJy0Y$CVny zk$(c*G_a>Mvt3%YG2Bq$}=}zv}$qO&YeGFiwm>j zClZeNa;SK@TfMQlDK{Q8;z0)WO&2iP`%4_qB8@ZTE7#?qPGjQEmaM?bd@#x|TKBO< zBJ6w_6eo1EZ>JKUaz4u9fcTy3Dt}+@%BZ48IO~b$v5FaE=P-{yEcAUL$BUyheKI9i z-%jB4V|y9#BM4KLEFi0N$_v6Bv}ZM$i3zmkT*)2}AEk0fT@3R^99N#Z>bQPxBE^xv zEJ6bwIbq9#-|^gS`{baiv1ev_#wt|~$xV9t?QYMSn#62x*EAtm8}LzJG+oM5m?+t) zbDy&L#S;>s>wX||x3(odtU6Pjl0NCs&M52cyZ3Bst)VeW^`l7$w+wMUgM;0BYG>r1 zc!O3-M~ZQf=8=B<7cx-wj4MaL=_9v59RSYiDQ+IN8FzFTxEu@jLxBQ+Jovv#g#UWr zN3$}#vpytd(u45@sXs;*IYi?QmHpiL%&SeR)<6W9q&FvO4e z-Qt~QbfqN@Oz{bghEL~KN0#u~X>Z!z!9(%QMG0e?@lMU#*c6|WBv3VYU}4tACikq< zyQj|aK!WCD#ow_)xk^Oph+`6j+UlHj>vni+o(sb{u%X$on9<-RHC9Kv;BZ+upZB&qDk5=*gPG zPDp@WuqxXv#4t{r{*=uvKQG~2rIqTmw6N2wp-t`dmC*#qsha$vnvutVd#t@X(a{kf_|vI z0~nzkQd!79dE~Z=nI-m=neFb@6|krBF$hW&CLZDMi9s`6^1G3h5AGOlsg4_alXm*0 z*GwU7oWY-r=1dK8m>g2t0ioj*kPK+<75;%fKoZbj!W~5&MQNxfyodGVea#(@LbxZ3 z0CT8tj4TW?;(C}y{Q-VSKVURoG`I8TkyBscpK-v-c;OfxVM=CcAxz5t%KOw?C(2_^EXU71~FM9q-cp#2xWprU0vn}Mk9C2Kb_l2!lX*8eL&4qKsO-Yy zDf+Ros^W<2I)^WX($7Z9@2;#Ev*3U`tCB@u4TkQ%T})+-iM)6U$*P>({l3RnD8?Ky z9uA)MT5A%-*_QCb0pFVzm(SXCnVUM&0tdWZ*0RUYfR+t%)J9i8f1 zb&R0v&-#lF+|~hYN~)LFKl7Rf2iipJZ)OW0Y3Hw_l#cAjOMR+G%Hd_loq8RDmlHiK zuE&>@kbCGqu;b_99gW{G`!66?1ZjBd+inxG@v=Ss_AJ z2v0Qdy{=~Y`x5GkT#vG7vl1lP`_t`kty;~dXHMEgflgDGkcF_v_5x{>ib;%!gK8XV zL%@w^8^%s=^>g5Q-Q{XX(;Doy~ zm^6HlA<+nZYU0pWcDNIhXHPi+&VbQsOR#MBt;qusjP!fWnhV2z7*=+kQ1zaoZ7lXZ--7@`Lct1iZerE}T=Mn90a>Zea3}t>Oi?ZbVls*F zJJ{|%KEQJ#U-w1tf@w{ER@=C;wkO_kD5bXU5 zyV{QPU9MB(0S)dTW)XzG*HbzOqI(*X*&eaAvtt{N@5}v>m*2-R#$#wXwQO@s7j5O_ zIh&n3;fDqJdg(WYXgureaqyAx9@#-x2hma(guA9J$YsDc8VSr*TXI0h65(o+z5d@k+4KN!-|1$ZX3Xn_#ATF z88^l;M;k!4pbu49ZR1(Ok4P9^OXVFQno}n=UjH3JRInIBf(RN|SkWPBs2+JSA zWzAb01`3V^9-dT`yC)p4w4w5_;K>|-O^iswioZf9q1Z(owHO}3K(H$pmuzO{bv58c z`5~xDC>7HKr#yXCow+3D#XW{|R;rwkpm{>Gdv%^M;ZH8gy{x#VBseeH(#jci<`owL z%O%ZJi)9Idr4W8gi+;~#b(;6ks}fe1#8}R$Bfq3ZrML z+9TzQ&qi$*Zf(48JGF+2IcaN;x9rBXhsv+wuU=N3lrE=ii<8RRv;^e^TYGTBmX$Z+ zho1RX1pp?BeUeZ3>clXxI1Ma%!91MK2#PQQF`$__l}8L(41&X5E(E7PKAVNme`rN9 zDfW0VTSrY!$zI{*IRsg?BWEy}aRKxzx2Qi^6-RtxQXVzy7AupE016w08OvQi?1N?{ zwya4@c~W^+IJSGVBmm!kB4$|Mx3#UhCJ$ni5=4LIGw1E>$*Qk{Q!a2R52uMqceup{ zK*&{^`eDzoo5j%d?~IMZalQ(rdO!d!Ofs;tYAt$E9JRS^KGQXNR$e@9=dQkF6SGU! ztem!;sqcSv==mXlv&xk(6!)2h9hf#GjTsB9j?Be||`3-+l9Qlv9+?hU| z?2g4GBjNiW#-SNJIKaRyjPZjTWlnU80Lkc1;D>TtZwDC3L+S^v!p#9qab_WMR1M<+ zZ|HVh=|_tH_{%1D!~|aYQxlX^m;tZzK^#ZbFmB9f`XP)|yj&G-S>;voQ8#bI**tL1 z^`~~P_A`pF{}BbwC}1DD1qDmRhs{#_-wMvo^DI!A_hwYqJujm2ulsU2`8a&pRI|zm2x?$N@jU z%_H&&u+@$FrlvX6-R$mbqIAdJedn6Sx}xZ`U9sZGN4+~Tt8W5ce0@U~TE6u+4IUN4x>%E@Ird-gfIaN)=8u>`;CA6*mk zY5FY^2*fg5u+J(e!lfoJid4*J%y_$k z1A{JvN--e70%(rZcUAKi9@!|dA}oQu);hF@Yc0F=ux6XPJ=@xCS)<2u&xiKn`HKC8 z*OvWY346!)lqR1dxz5cq$tOv|}D`PV+}x#@MA zqN_*ifEk)OV2}fo1kVfqvgdy1iddvm+K&O=qV)hk@i>#4%pNYqi?Hc`9+$k{K6vpt&b{AmK!zF>|+VP@HftS+@ zuL$=sr60$0v)jUu;HVv2{6&DHYVe0OgUo~M%aE1Z9XFX1-VajxP59`<;9BwnhHirn zoBYsnT{*Z=CwumnVW`4AEBDW#vwPSEDq6<~45}sKOcrH}j7jLJ?~0$T`qlvd%B22= zpfx#pU?-L<_UX^OV88lr{Vn_aPkqjIw;$O*`N#jAKS*@)^f~c6vpoqtw{N}giL&QgJ(Y<& z)4M$SA)#=uk9|wLiq%bSvmlA1@&8!>`XkVmrkb(5}4~iPs&M40+v3PBr=a(BCR_yw=kokED zv~!HX&@jd+1YsW0;Bl($txel_vMw6!%CD|`Gqz7#b@sI4J*l!x`_?-iIEjOXecyB5 zQaO0tknJ44C;XtJF($s2`lQ7tKAKEvY~tZ3((U#*1b<*pHkx+r#y$J+gS+-{b<<9q zJ!i8FOe!1ux&4cw#=^YEntXe-RINkwop3y+34ho};@80gMDN1?ID|Lhj{4<3HT=yu z`gQ!^_mNLj!c5^t;ok^hm}|mPeF#Jt4YYm;A-NtWixbgRwkMqA9>(m5hC3g{lil%> zyW3#|M+ISUV&zw%GJgC72F@P(4atZZ`Mu&)ry&WsngzGmjE?2Z?>%T_b*0$Uu| z!Z4}X=SiEMJ8PxVk_6&&s*PDYP#*ZBvBW;rs+B9#zUs-HzT9!kexdkePFgfogs3<@ z%nX}Ofv_8JC}FN5+@1?lekU&bq$Z-Fj~z_e!ikeM&r_cDwj31Scu7K9#}C>;MPA2w z?xokQQk}N@_qZNzTf+RJn1_V4V%6Ha4{i0%4VypxoOLCfaqxoM57@&D)x6;=P4@V5 z?HR(X-`^;5g6s96lwbgUXoG_wN|%Xp5I*S6elo(9cm+vUl zBt$)|H_}JQ?5TdT0Dd5*gCN@NOAr<_%~eg3T$9A9UM`idOdV)KAzC5`H@mjJ)wWeJ zr-wUY#vL&qHKLXzYQlc;GiPkMs`!lOIO7wa?=#!uX-}@R126KGAq&1#2JJC~b(w@m z&7w2e;d86+XVw7iD**CqwYq+QV^R|n-uUOmsp*p1h6b&|Sl6yzea>bj#CPjE*6y^d zC*ioM`koW+sY}l*%(OG?zOU>kE#=`z_8=ZWUhrd5lJYcA2?l*7hf!p-|L+Ymq52R;IMj(WS<0}NcPE>D#y^l zA=R9l2p%O3Gx%s=xE($H4S~KLdZLt;sDS&1%#%1ceNpIFI}t2T?jj;Ay4weI?zobC_HssAw^cI}(P;xLT(5 zB&_Z{SrxltG9O>4xNT-;x@70iuh`kM=k43yc++ltblu+gu}|BdV!rlWyZYkittkG} zY3|tV+aGwJt(4<)_UZFy?6uE*+P6gPOCaJoWmXs)ji%jy@WeK^cEnz$l{NKy7lxH5 zbSBZP0*a?0_s5e%KR-ho5m@d1qy1-A7L!Gu_iO z+&Ok;xJzQW5}SXE35y3RJHqTl z#VQ&*IeSxLHU$01uks1a_`2EaG=1D*JRF=tL$N+>QOb`F4szto<8txT z-GEUT$H84;aOIjV55S=$nepQOQ3nZPd~#=XmMsH?nVJg(aUg{9nuMd@SY`tpK{Phd4So zuQb)z{^2)FOj}h#Iak3!Kgv%2IXJ?8XdgTX=ba74!WIF_%{6kY0JGvCMkNlyp|buF zG!UlhtDvf?lnPc=J(LP!Mw-ABxPEBB=eQHQ|FCO!AGK|>e(2$Pif_88fpY59zVO_v zz4}5eR(QM&v;wMCaWl}$DeBv6MfpZRhcS^)4K^;QE2Qv(xv^zOVWeRGGhnKBCKLa_^vXJt8B~msGmrSl;Jh{rJnsS z>fA}_ktVapopkpiO_^v5pn>N;BZ&CoFND2-oKN_d4}{qDM+lTt%V`aOn~cfHCnd>0 z{PdIJbN(Qg%=@z0>&gp4`*1JM!}+Zc5tn<0pJ)*7$qTcqRA-nK9~d47e{R8^+{gaB zs>!E&CI#f;j=?aLaB?1U`NB8aU0(vy{^nyBhVf*Z1S^np;y*9G{JeFVoAy8a5C6g* zKi;(e@bCRc_J@D)@9g1&`*z{NOn3dq3^2-?wuCqK<-cFSkdwmgd8Jl8q5_wW&%IA@fSJaGwBS7Hp zNg(zGBux}J+^CSwtLu%n#^M!w?lR-ul+soG&rI1A4}U3&XbZGxfqtMv)IIKKAW6pw zcbUDvyL0pjZzJfmKWONv;9^2k+lHqJh=vd-cZnz+a$bg*@h?bj_~U?vpL+)O&&i(Q zcywbNpa~>!G9-cnPE7u#k6Z-_$K72;rG*|jLPBUtNVx1D~X37wl$mkHoJVqPL#&d`UTDMi$2pXOmlB#)hbn~N=1xl;Gf)1 z#LT=N#CVE4+@XPkeNgOE%Y7l}%#}B+u1xuXz^YVyMpNk}P*;uFdm#a@s!;DtP8ZIrB+R`A#|-c?;%>J^Msi!^982;AC9Ng9niPKuff76;Aeq8~lkvvm4g`1Q zXNc$i{AF*^$j#pF-6e2PV|~{BlFH`LJ;2aB!X25z9%?)t!0Vs<@as6qEBZ)cs9v29Og(kr2Z{R33(S`PCcZUmd^a_437ad;Rs-?H~W+|Bvl#J+bGW{{(5OtG>kf8W0=L5+H&Jgn@700v)yCe$ZX%zywDLUoKQ z3P=11Y%VDCA8}2jE7(|EVf8fUWjsGW2@Jyxcs!_RhhoTVF*_B1W1)a;5vLmEdsTMhlEN>QMUwlu$Jb93{`8#kD+hr4Gz<%`9o6OFyU z@f*LSF;#Wsd(Bl#l$E%pNWs&=tacn}qLMqhS9SPhm2qeX*quw`r$6YKy~lAqY_fqr zE@OPE5}wOx&u~XC*H(!BbA!Jz!Z4I{(t0Nz0Yu4-+5I!Pd-e}K`@1>010JKSxJBji zKk^68hXFE9)(IO-h6r+F!Ovs;OaOMIg#d?LF3#0M=y5;&`O9Oz=Qb1@>O**wct-Lt zqX|B@&_Ffd<4An&m(9fIE#BkA9*qx%s1#*~S#ZFOPhQ~CkP$%3m06pepZ9}$3yZ4~ z%FAN*61)^VE=4a(D86Pdyz+)^?`+zKAAf95w(dFOTD@@93jUdn6$urORc>XTTOXL| zk~R-oKs!Et@%e~5A2}H;Cl`46L|NpZo|sf@ji`El8JJjg4%(<}iS#fQuCI%IQ@BjH z@XM#A5Q2HF$TZtN^?kw!q{z_K^0d_?2(#yw)nF6@kYARseDAaecJJYV-Pvec^GGyQ z*^8QmSIR{V4L!TCFk^4LvSRBCqUq78!X5jNMZTh+M;$;1&y{o-eV)i*_PRxCb^oX* zS59Z;Q(=gU@`KMyRT>jq_zht%;^`fp*g?H*J?>TOcl;(2wtTQUv9`YK4F2xD4KY_< z$BGHY#Po~?nR{_tRlmUe_WnA_?~>t4Eq*8u!m;SNQ8+EyFQGox>U8 zj=*g=Cv)~4x(h$z7`LM*Gk=LEWaTwEG!riHhF{C1#%YvFDG-jk<-bu~dpv+!^n-E` zUbbe1n|?htZ!_-b;H*(KI#M|r{t$aXg;#SgAQhw#lfXF(;Df2Y1DVfc}RPx_wv zwD<`h9_n&06({X?{@&lW|L8yYd-kO-zG8p&2j8;)_W$wU*p=(QU@yM-sz=Oy!FTWe z(1qc3^|`No^%rcQ&Wn(R5X`6d{rw%`Jh8olyviz`CxK=#QBoA@*aP;J@|8^dQ2d1a z#w0n$j1WphEyf0aTq6W%F9X{brYbTkR2BpgUeQze0VdIGE%WUe)S`#?l#&i0m_a>i znc5*f#pR#*Bktj(_@@N&1dZ8~j6iv!3I3sYSyzIuZ+#HIN9T1-zwM#jky%1vzpIH! z*E+41Xu7ZQ|lFa<4sFZsBWaH3rjv0K^0!S^VMUvn#p|Fk#H{M`|C+Zrf*@q#w_+rRzG z8V9L^n56J(DrJp&M#oh@IJ<%dxXG1zJ!8({Mu44K#uc--J2y))Acf8S(Ze@zGi1N> zE)K%;xRO(hAN0F$FI9hDg<%NcR1rxtbp+WBedH?0_Jo@&fZx$fD8QMlthiBOuG3Nz zjx!qm+&T)EQ_MQr5GOf(W*}fvIe)2XB?Wly)LGSG>9p9?>e+rVIAv9s!wTFXNuK$H zMkmT0iT{W<)5~3?)3^xZf^bkH{IROJGvKf)j?<*1BDO0WvI5EsT}+}g%R!@KI^U0YmUlhCNNlTJ#CmzI?dGiM2@Wp1&Ey>SS`glBaaKk9O-z^xo5#iO_s2f{AGBCEjq z4}nuOfawv2y|-lvwY5?Rwv(bK9|j;1a%tyy+8V7xyYr}R2M2sIbKayu>qKD@Ca0|^ z^Np*Ew!Sish$Xc9ApDI{o`cyQNkS|&;&M*3&yeSpn?#K7N{qfFyS@p`RSv(lx zZNyU-oZMYu&xSFdaT|QVO^(ce{Q2Ze z3Y=jGGzYn4Kk4HF8m{h_)F(mqZ%I(ylQ2vtjdLsm?|)3v*s^pQ`dFJD}>tJg03Hj7?QlT!p5gyDmRJ=ka}2?UwQ zD|{7Ui2bsxZUP~IN+3-*WIynwBnVG%aL#!@W#+3|5}qYYWr#9!qeR3(_$VuJOZP1m z0Kf3Y5FP=9@4Q!8+)6(vgkiQ!qPzyjEn?_{BmI_|&yC(710Hx^ssyFW@2g(^gopU|hqGMUG$j`Y%YBuc*oyV>PuA zkI(ra9xt-MYbmNxKe%UwgJXt45Oi{ro4p*^+&S`xc|1%237m;3;U!3eO2m`kj3s~= zvY`EW3&X+j{)Pagj*6oJCs1X4B8(GLXpZ>|0-uLqDCi+{epVPY6cQU#p9VS1Gw{6W z@b4(uA12IpgxC&2{!UH~)bSpuV|*x~h1CGATVuZ$43xmB3@NE0qZ`JpigsZqEb2E& zDJ@@lg#p5tS(P~Um(^kwmrl!yF+!)mxuiO1Q zx9q*Q|5Ss3ghLG$Q<}Y!5hsDNz#%>aSwEO2;2>OWj%^SeB2gBU4xJl8lKU1}9cJ}_ z?Eu_^2{YsH1WYL6h4F;049g9sfB?)hsA8^ci;!Z!FgH7tl;=uu(kc?x*=q}vqW>Xm zvbXr~blUcs$98Xnr&5`fvjVHUOhs58)0jBv+v?JsU09!2LNIM*9l;lRIi<+~pok$m zNbN03ff>dssv2{WSDjza4VRD(+qsDbZ48rK$o+dFznkXo(vNF1` zFzZjOvVU#k;Um#}%C20wDEcy>5qC@~9Fz!5mZUPAF?%M?!c#7(j3dYn=RFjXhAHJt zIk`R~#sCn;qZ$Pe5I!8@^BRUxoH^faPic<(|5$Kz;b-M;n8Pnr*zu+BffpVb+%th% zhWpEu&^(782QdUHe&cE!NQ1VDV&I@}kAGDXMj*N^+{bMCOOQ~ro9mwc)Fy`h)K@Ym z6~pd$JI|01rTjPHM;0O__mP9UXKc4;C7PfT9?Ee%ijf_E|I7Y(InD^r<8epyWB|sE zgm*vnNA(qAW5UisUv3znLvdcA(K@nw_qOa0{_xxOU;j7%Z+j$x_={ipRbj^qln8MC zsoePvSCSpMK;wN=LJgn!IZ(ywCVOCmZ$*3YFIosov{!^s|MX4T?CnM9MPN#RzXV|y zT;bV-@qMQ<10FBDHA2MiSb+uZCczdVEd*&4<4CMtBAkiqiDo%8;RDBU36uNQI9-1N z$l>xC7V>?1!eO%=x$fD?;hqHH$D-f93%`d)JEBotcn7K{snaA3^97Rp4q6(%xaEPB zSJ|J+U)bZoG+P7fzT(`aX7`!O-k%T=Bupdjd50j^?-+1gv0fA z+%ge&&zrB7fB$!?P-eg4Zlhp!dZ5bM6H|Dgkj#1!Qj3a)4yC|3 zYLAOXON3*Q3fyhzL;zWQq=>B%2%WxSs}9Hg#bx&MDhyi_99u#x*RmlDvX6H#fN4u; zm*12a7L!^@z7-}BG1+2XK)(Dsuinb--MuxJ4;F`t? zu7fKm$f;)U2o=6}dNARSQas*j*mj-M$_nQs5`$Ad7$8|NDoqdU>cu&mD^J>tgydqe zEGDHik4B$9Q-(44UjR%Z1mSpfPP)uYFwVd*tcP$N9AA-8c^Gv0bjU!<>I88dipj(1 z8OBPgJGM};{bY0Zz_t(SwzuE3)wN}x{P;<6wKJ~^SCn7r zAKLH)XH0iB{0qo>PGej#!9k7klp$;pU-EYqLRQ9itZW!Me1dycnWqxDbT1R}c}CO; z+&xfn;(+id&8XkN2pF2(A3L~MlRcT;fJ=5Cs>p+@@b-Egg$eL@T@M9>KukFBL62|| z)=?Z>ppl&~R`efuVzF4U>6sZTmuq&==-OVrZ9stCvp@x)PokFxl+n0|N)I}>{Gbvyv+zEF|FFoBhStUw}k1q^AS zIE3W7uSh1hJSIjzmqEgpuk^YQ3>sV55O&=r1s*G#n1thbKo9)CEad#em7ozPg0>5e z5<9NGsDPa7dT?g*)PrUGY;^gj>5i zcG$b8Xj#S8m@~yW6b%-{=jqe_GvX6k6n{yDO)3~zrrVHG881V!#30~C1>w&PVN<`LUZaB$Z`!Au1T=om{5hI*`4KKYo2AId zK|*V|CqIu9Gc!blI1;AeOP#VW7kq>_Vg*O(t3&&~UEjPZ0eFhWshRFz&svSU*6(bo z12i>Le(0ZUrl%K`_b5;>11XQ6a=}Q$%t1Tu8-|8{6%L(I?p_{WVL$-nv&)o!HsZQA z;%%$L6-os$atX%%f*hZg6i#7?g``>GMOY;N9ONLj$tjr9gsrSTXAd8|ZN=)T6^c9} zBBr6)tzVrcfz9*aLSg!(r1EO^R6>3gQCEY65b}n5!tIM`ahPwaz^AK{ zgz1WeQt03mHExRN_D_7jFyk8q_PxHvDG9?8W@To(1D~5bZqbLCb-g{XyN{c;z1LB` zC>MvY#FivP4@5sGhXDQDr3JfjWlntVP^Xi_Db^7yP1!m=ml^XahpAH#(xz4T8V0(iETS6e6$8i(~zr-QsF@-g`^Lr{`Xww>dJ5GLnh7Smde?k9zoINL!q(PSoxgS@0Q0PTv!(vH+470H^D^&qx-YNts--VJyK9+$6wBh~(8uHw&D$UT_ZUGaLyu?1$FoktFYil9{#d0tVhpR)PX~yi?50g`#!7PQFr8P(uL%h(4XO5f>Hhg z7rT!$W9h*=PmU|rW#2JZtaT-PpX}kObc<5r@2Ar~Hgb&_Awv+2E8sXg!&t=jy?9V6 zR?DdiR)+ny4du`8j3y5#k9S^i#^MS2C@F4UY^9BRU-EI4auHs2fSYGqZSMHQ57aYB z<1YTk51-g`&t0+Exw5xo^21ocO(W!m_`KXItG6%z_sHDrT0}n5{g?deS3l_P79L3_ z@6?M@^>E?orsS5l2>PE2=jN=RQ-sXpdy3`fdCzbn3ja==5e)6FC9@hMgdKkEh6%<Spj}N=HyuM-|eegrYacawpYcA{}G_cCzw?xQ1 zGuc&^6rTjQZm({C`ltWQ-u>WvR$Qps^J}mBL4rdrAMa=eFQ&yI3--bGod%k{dwf+Z zm1cby5r476BnTrgG7wRFP`g|#ShXz1CTW%@QekF()tbuAy^ed0V|%pOl<>$Q2*QX~ zDz`GH{Hj`}N;ai3uFg){tIw}$JoH8=D!_PB8-m1CumGA+MsKFkq~dl9_WUYpYJ~j3 zzzFQ^QMjRK4-t-SA%`M2xE$DEZ*Ze?Y+E}Ge+Yz6x~yb*I#V;&?wzD3KToX5p^kp+ z?IvPYl&M1wW$@1ZOl8h^8HqGAc(O2c|I1k){~ZCrT#;G#C?YmRs1JcQ_nD93c2 z-gg!+TkwHCL0f?K9QNq7K`^-w<8_N0BXn}&Pwtcv|G)+*xhJgnKl{hO`+`gVk&YuT zl)1^{(;xlBapdX|M%hAejA-1S@J}iCB#?s}JL3JT zfn5FoQJXD-epn6RjVJy!ZhAJAI}R4IErR^8$2bZ2z}a}t0&R#qgL5YDlN>ZG*x-0y zMQwS!{-smyq|3Gi_KK4M?ha2L6Pv#w!9P2skRJAv z;KV=a&{5TR6+N@n!AaeYkDf?~dtV)?Az@`sjbmQe@%VzbC>Mz72)UEeVGt%^!RnlY zx*8>d+o~GVwBm!I?%D3vwmrP_U3>iKN7gu~+wI%$lC1m@;ttihN;3C6Uvby31ZxS1 zP7WcmqROW`p0b1<+<(b!5eM~05)65GW5zEa=TkeKtKDq+$z87ZA;)w!_Rta5Une^1 zizz@QRy$ZlX4ZRbmFk*Zzwxr|ZQt{{;Ire@v}Vx~-YCDXgiC1WF%O<_?MQGdRTq2@ z*B|`be`Ejp&;Gf+ee1h6U7E7_+Pp2!&wKgmjXllOMNbaQov4%P55bhXgAs;@dxR;o zr^8ij5`t&g(hw`ce2x-~?Lx!C@0IV&%u#n95o#DjnL;pl>USHM@3s z#UHmwEhC7Uz-cVV?JTrlKvXfGM}5yV&Ixco4}@oc#;en11*+#aV{mY8pnW>n>{$)L5uf4y z3T^{b2BJn%}Ne_Ft4^?=CUsmBHzntIkFt`UT-IBn=I}Jwo*?&~>1j#>E z|gvM>v7SUVMK(6H33{tCt>nIHYg#GTWi=c-qqzM#F!9%Z z@6dW(O=6oN2m_JdoOambPV#~JX&l}HBe3{Yyo^IG08`FXR<8po#uH{fich7JMqUtR zNAbrP=v-V5#05T_4uLCyuM$U?0MbUF8wsC(bwHk!AB}H*m~V1XycZ#uMHb4#s8Ee?V+)*Y*zIyi}xzT)Cfk@4@v_M(TAL%2Iu z0!%3!Dj|}u*b|)X{6<|wJA-ED6y;*veWoI84j=0ChQB%Y3_~n8Bz+8Wg9pM0VKIw{ zjvdN>ou~o0cv#_$uO0+`8r0kh9)<|xx`*@X&o2J^EO6XYX0*}hVNG`0kg&O7-R71W zAura@snxOQ^kC2chLOvV8k3iahNhY2l6n;v=pJw@X8ZY2Ya{GieBnTnU#g+|EUwNM5l?ee4fw0Zt&i(hS zRIW%hXKS1@&(CKzZM_ zE021nf=a413e()4FlAKu^@Nw)LQnof{b^9E06E!Gm^1bf;5_uo?P%oSX8+5&{)Tueila(} zJNB%*^N`No3bJ4RlRZM?T-KvD{UYmXdFtwh{qVX z5|Zn_;>(NQ0V~0n*Evs)TH-mARxVE2;=-J-;Bn2jM6`r3xWfxc+68G+De*K0QxClb;4+1s!2H z4%;CZvu$Kzf~)38Nt{&7H?MB8JpmpWw?o9LE_4k(1f6Ir9F`Nm9SZNs5Ab*-V9zXS z+!U`!!nD6_ph=W}UzD9Mw(w_o|LG}bgMuH%Guw<8Y=fZuteh*(fkG;uEF@J*WqbM6 z=jDe0uN-KsVIhPt(2{t(ttg%mkax0jwm!p3dLw^~KG(ue_Ru1}jE~tT8p$t1l;e}Z z<8VFfz|ZZ<@Tb8zfkq%l+F4c!5?}mFLNo$+NC+!5z)y}i?~MV^ud(qh4s|>1KJe`V zN4+Ovn0M7#cIDqz^u)!XEp-45j{cd%Ke;nA#GS^=40K-1w=c%qR1h)O@+GS+R3rc{ z*wV^%37usLv=uAY79=Pg$j(8yfjaJi;@}y}13NhQ$m%<9i*av=sW%nBI_iMVtk^zUl`-O__H(!F;F0 zu;To{sZ}cRVqB-$k$`;SE77b7(@~%=ou0#8+nXOrsN0ngUbfZsi?&d$+KsC>?N`3~ zx9vaq`~QXg@|S^nLIuJAN9!*FP?%IA`f+JVd$e&p_Z4SBRSag}uEc?RB ztUb40wS`JNxG~%}d#+ii73eoRO852ytvJNT9%-?2Inz?)?8?zXbMUrAObWvQ%;z_@ zRvb#Xu0glo>nd+Yz7?XYi9!8v(l+;-E-25=EvRhMjz>#kuVS~&XTX6b1vA#~wtNMG zI-#5hXrwpHfoCcMEq~lT%Ih%iNrh{`$P_T{!=KacS#T!p44K&`zcX5jiAMO0uu&g) zB|pj=rYO6sWJutQ*_F)64Z!T~VO5`0IdZffKC7W)Xl@K1;r0}V+mmv)QOtR4xfS z{;6JqOd6s+3<8CexQSexgGhxN>19B08i+;B;>8Hi|lK^!kp-?nCJ=s?m-jf+Ym^6J!LaF8QdCK}>K_?5wpAEe3lgf6=$n*8S z>P$J^kL?R!FcDBzocG3a%-m2xeo2?{-YMd3MY{4!yJ0a3ng{)YZW>&b{ep(ciVOS* zuy{>!Xq+3HBoq_M_xehC3=FJ_BLsGpR}&YuSdey(21x{QYY>RAojSE+ns@{B0NO$loqDIAB9Up;xH2QV4C5b z@$ivhXpbt1^rAyP3;a!Z>hB(R-8LfPStwdx=FO6&aMtBPY@jl3ns$mQU4sF>ON zrP~8_ZZRX@m#dEB3{f*xwiFy69Vm`DE6iL{xCLLSohj6;TDzt;PrfA7C_WCl9V6(f z6U`LnCH(xrHXr?uwzct1<>Q_m>^-o?!6Pq6b(RBdihlwDP2%}s(?0s)_wDkPS163_ z?>>^Sz3vACpz+L1$qMRN5M}eph6H>Gh{dwfJrNEwQ?BSLD&KOKu!WEf5z5tgJ+FDN zX$3KZsTs-^f+2S&@<#uhaMo8P@&j{mg=ODs8 zO9lE+4e?UGP&sd|z>}_gpUG!ZIiK=~-*f&rpNab^<-uIfD#B#*2sZ-)ACSWjy_6YW zfy;j>9BNcw9;yF2(Ep^lzx0exxpEkQHsCQ2-BPOv?*s~^^bs?;W4KjUAJSxztzmh) zg!lJ^%PJwlPh6SSk+5`ZZS~2cE;mPT?bng`&%m!YtW`?HZu@)t{^iMqld=2(*-R^+ zp4NCURZ|$+4cU@_Cca8L@|@F_RHu2hn8gf7bk0M(!sEak3K#M{N-CvAy~dtl7c5;6 zh6ZTqHXQaYizgtE`awOJ$M*}5=l`b#q9A^L$%_k)t{59x6;J-2rj*L!dJ2OY7ovU~ zA3!sg$dst;rYc(911(7 zxD_5jo12rSrX&Q54?Evi-o0|A|Y)&f$jf zMS!S4{5=E?{5vF<6eiM%K}`!xGV8kLZJjb zb%N=_i3E>_)@t6i=KdWCD|~L_Lh+_>_a&@u*xvSiYd3eT)!3JR)1NbDHJL+lJeNEe zOu9hB`|6eIqL?vExuKb6OC4lWVc(aq`oxa=cdc2!ZT-%%1as~wY+CPVUxHyp!tg~2 zT`&PYr#I}QkKVPFi_3O#?V8Qat=rDd$JSu~?rc>8WLfOCptNSaGq0>)ux@Wpv)Zyf zy8E#`esIU;7Uos9l79|7I%I1HpPwY~YA{(?TC>Lfo|QfCa;8p1%f0PwKcTy@uqLu< zaFJpamwK3z@XBf2L4P2+O<29YWBsG99kGW>0DC}$zYGe&yF9mKljRA&zIAKNS)D~u^ z?d6v*+x3+Rf1HE7&o8i-EoS}P#No_YW?DsV8N>}^qq(>uO$BCEm|G;`@SM_7xSY=n z9*#hA;}AvcE!V#$7M;$K$|eC_?ZFwga2e>Xtu5G>Ub}2xdU3(7ENP-6M*HyoeS5OG z-DkJ4#s|0bel87*@a85T_rGpc+0SABTffY+u8)<*kBf@qH2mU0h z#7BuL@I2CuIDs7BUj~M8c)q2Ro(p{d4$wlH>L4I=cu_9bXMs=3YuxM*i#+sF%1W!6 z;1Hb5kse#+LICAPCAK$kD-vU!uW}|H&U4Fb?}(&mi!8c8Kk5FZk-FmikK&S<^BdH; z_y3c)qd2Z-(|AF=OgwzT!k9;WuxLg+5jX7t`FoPap{yPaP9y_N+V<|Q6(>*a%`d#{ z@kNZM%Eh$IvI#xcVo-(nym$c^WQPiXHxD)nWhhYWiFOkaed_@G^D{X}DAS;Z;KVf@( zs!19PcquMNatJ-I;DlXaNi3mwif&GhG(uK^C+Q?Kb{G{*YsaQ1j;yuM!x-;KuqY@W zi`Hp%Bz(MUb4%A0j!$51HQ+7vZ9duGkU+blw0U=20up;kdu=h|$JTB=RM@&0;=tyY zt74E6q^IU2jIK$b6;A!W6>C*%bxPLi^)w@Vs4jaffpFG(%^#`nDUH(^36SbMg;`DV zC)APIzxJZ7tUhmr(xTjIwq4(}NB2LnWeKFuU4Ipxs-hm-{QSDzefWKKrYT#PUr~qa z(Dv0Ad3`Oe?5fQxj_Fz3-u}Qoy7i6~XIHIS;Jx-mn?CK?)PT+|L3(D|o;?0Yg4dBP zEsG8kXpfH@64-mTJh!S&e57*CdRm9ahxXvnhxYOPTXt{bk$v#-+xGF@_a)fflR&&J zfv{kC)x)faxzE=i!{{@?E5gbz3X6Xuf2TqzZ7h81)%d>4krkTg_D z0G<%z;Ij2&D`+6_yNJbHVQ}-MQ+xU5vdzJNV32ZW<+YghSkp+NebFXpJJCY&FN`g= zXCN?huy4ZpYBGmXY^`RsHW)LsazC$KD9=xYg?OJ z685Y1<(F3M($cZ*9CYpBmImf-!-~rI#x~n=VmnHyTvNL{vfa(cE`)Y`e`C)D^u_h7 zu5&X7f_!2}Q7+FSj4l|ycLkhMpTIl%RMto3fi}?61rzUk?&y9%^zAYHqz8_ar!5@m z(L42*;tGhcK1t<21ql6Ll5H+2s@_8n`;>_;YJ=Phs>V+K%hp? zgwK%t_4A1;#}Rk{Zq!e*3Y^^X=UT-u3I`pP+7md?#PgT@0Zfl$3!3%qP3Id3xu=?tGx4NPvt^B2Z`h0J+l*xI zc6yrRYI3Q`5l*Sb2YoGy>0m?`|f^rTr zBfrG~x^x{LGa-eiO)wckkG@BoUTcby?K0#a&OjR~$_+p95`Jk@9v}Rjg&-DTBoxby z@j(HMN3CKaL%nc|1@6#-cBC;x=^Y)h6;OB@KPu&tglx0=LET=sanmkeTCjrJL7%~A zMmRFlLpXp<5nNW18`}tdiw7k2qCtc6Gz~YxkXJv9oqdM=+im`rAx9H>4naB*p7f$i zjKX9nOQ!fd;b*rpe6H-V5`-ZWQxtMDZv414>2eR`Z~zi+_yto=q1`X+XCZJy&oDBE z!R?MG0}a!e4vx~jL)PuE_p#@{*WIv#`VVY%{i`-T!*yJTwz%?=Rc9|-eg7>9xj(RW z``cowcf`Y9ko#2$l`B@MEz7)YvkOah@!IpYdg)8Hy!N6ku3b~3WktEF_{!cw}85MGXoMtW>H<5L?qcU|PZ#9hgTq#H{;$t5z2!I4@`b zII{b9-mwQ;?^x$}*Y-MX>srGeJ-KZUAHHji=AMMjo;v4Cc6vzXo3+})sswUYLQm}0 zhd;0%{pdS(|KSJ9d)>D~%vWaZ!i5*@(#7ZPxoa=l_3N+L#fvxX>eZL*`g1SYg^TMp zH#@H~SCofEWq;BSjh(2Yrz$fpF-IU))rbai6#W>q+TCLb)jd0C^fkGV^Tdf?n8Tjj zqRRQ)rCD2FuZj)|1T93<;*2IZ#N)BU_@t#a0(gH;daUpU(_^qB-a}0!*ouMfdz_sC z2VU$Oli)}gB9`*O-MPf6pK9iF7~3M)($G>l+wHE+6iW8|l?Amk3A!yl)k@HBAE+%A zZR1#f7vV*Lg2FjT-V2<3Ul_s zp5nsn4uSJ5H4Zt9!?^#*khGB}VFJiE;Zq!PcQ03A7~uwtht0}eaF+^i|8YGFpcw+2 z_L$<9VkMV_xt&M%tbMSn{$BEaL9Nj*={M}rL{LHX#Q)eo%ynCxUdOskP4G`TVi(*3 zF%VyZ?u*dyUe4yMi2{L z<3bX`v3OQ=;EW?{5G4S?RwZ-)wCb7#Ko_!H@D98Cdr;E;YXa{eQ3fvuTQ|JC**zcx zc_(62e=`isjGTIAcE|0P>0V7FIUF{<3Lqf1mO4zDGI^-p`CW6bC2j*M@ zrB|v{eNxBEhhO>n7p>cE+DxV9+Y&kZ;@c_|4}0RtM;s1SI{H%-UY`otyV}2fH9Ec|w1yLVo;pambuBUAhV@!%la(avtB^#4Iv_e=XE{#{N zZccvc=?Mzer#fj_bQ)F~I-R;59BkO5hacL* zyZ4n2$*V(5vYEf5TAQ=V;te}dtJ~jyWLu9PC@X60B?+3-1>4>F*y;_g)LOQ#I)Ubo z>ck-cm1+yNx3ggn?!GHQ_^MsHa?^{&;})ckpf)W5s$5!9TvJxxdt}vxSr?r8hwNiL z@l(?N_(Qj@!Jwm#CPA##5NRfCbL*~s@X=d#=l%y0Za=oIy*qwDjab!U6Y9vP z65ft_OgM_Zg@Y%DyGN|3ig}7Itjv~K6;_^@Eqhg}4ml*(>Yn%&YHjI^KgVJ;90+8# zyE0$3n^$VSQvYPTZL{SGE0&7>ave-$xS~UR5vxOP$*zmJfJT#n-}R2zmJu_4_6rk= zcv!XdSvErOq}&N`4sc`dGr-=&j)dU8=*|I`&D_FHs@Y#Q>QMZ!H+mreBqkV4+ z;57DkeTxKxFB7U(yXku!k449oi&u@iiU)jiQ|@EME|+f{Qh_qKIqbQgW$#>$pvABk z4&8x~DIo0nbpw4NyM+0yl6%qvBLz#pYwsA4sP>2 zM>vOJ=Em(2GSdRLDCk#QCvy^|979ZT* zxNoifJ?8}o!|1HMaaAOHg#9LoBW{u45M8?7P`vN(kON>Gpv$XwV$)(g^liSR^)(45 z%`592>{XTv0yu36(j=Pjr8Ty=sP-l+YMUkJ@htijd2v+ZiC>#}I_W%qalU3R zy!gBw_m1MrhsH);Nd%8LA+5{fzmnP_R=<`&m%c7Dbx>Nxer53Sp~Yuh{1_R*b>ZC65AwrX!|WzMtm4?A$YEeFdWrSeRU*EZtcaEf+n9WVGcekKef5iq^&O% z?b5mgYYkoxpLG26_bwBA74fj*t(P)&jU?$Rh(DXmmKMc)!5g$7;0b zkHza)O7xCq8Eu_Q)W>1iD^qz`rEau()@W){b24f5rs|-7WQ)atteJ_>Xs%AkMG@%JStM|9X?0w(`W?zke%xEXW| z!!S=n)LHURdVuF5AeG^4c;&$HNiiwe{pvmcqhAFbGP_mNaGaQYWa*+A?TSb37+LP4*1jNK^f*vj-z(+SzuJ|$ht>*Cf!`N3|R3FJ=%26 zYvf6^##0vZh0!4y=$~gG*;%#Y)|VRx z`+I+9M^Q8teXQV3_|;p%1K1(~KSvN|5)6>vvu*3tYGw7`@|5$PUT@$le3PfscBsbS z!j{6gfaUEd*?nP#cqlJ%M_Y?_Elw0$pX+K#~8zA!RBpeI3?yKspoJaHo) z1U^2^pQGJh5&DXBjakpn(<)1_It{Cr+RR&*#6VG)L%Irqmf~G2r z++tGao(Y_$Dh>Y;mNuHgp$C$OC5{Zq$4x_{V3^(*EM0spnLZ!*8zT&7I+GbEi_a}fzR6P`QIXZ?mI0(@dWm`MDcH^a&Y<2aDz4^vh?8Pfzgx81;9qS!-ts?;*LA+9)wPOjcO5J$= z5#wgS%=+Vvr2X`%O)n)JWK0-fZmq11SE2;L?in$bficT}>M5yW8FtxrS$de^+tD z>wP?}{o0p*&1y>*)poYLd}m=~_rbsohW2Sir_FrlOtvLBp}FnU&6vp&O|r{ z^j#0*i6opMDg$un{+rChRZQ#(m(wI~zQ6Mf2uc62eXbh(-XHz0?LE9ti3PH_HAjf870UV;A7Yy82Ewz<91MqVdcGW!vWUs&Zl4OKc zwQr3f8XJ7II{{gB+Fj!h>JOJNk^L)U=CezR;$nA3+K5t?8XS8RS z+ZKWVs>`5jqD&YenM~m%81qBNKdJK&!mykCFO1UT-yuGA15d|+eb{lHm4m zx!l0Ltv#D6Xg1#$qm)oGIkBKdHX&hdLv9ixYxB0d$t?_5?ecTKY{k;51kgA7ZUT~x;tgHxWS3zuxBbU}0~DSs!%D|2)uxma z6EW42I`7o!jF|SEW|eB_DtB3C$^~;G?r!G?60|;6y01xKUblAZJBmvJvGO!mov{~R znvzg?VjBo*enX zxSQ7(Y_3*T*@UioIB06nYxn(=)1@_@wT?b7qIuhh)`3HYaLKDqF^kp^sYJNyt*3m& zz06RQRTyXplZh21gkj?I;N%NUKRzMSB#8ZG#7%e(>%fA!8jiYRkXH1*9~r^8=vUMo zi)ROhCfHgrANqg||`_|Ef z8{Wt&Fn0}eZCYL#ZgyJMYc*wWi+`|+N4l!-^tF(>rtPu^n3RxH5qQZLim?Yh^AXI` z>i_T|ZzbXxlM2h;(pa@if@=~60iM>^PbgyXHi<9oF2*PL)o@afAZVldMsW7<2v&&( zvis_#uWAx5WNav*DH%73gx!yl!mLw=OWbtVD`o?aG$rTl*iM7f^^+XuM{@Ok)F!LM#{^^*Xk2 zVclMP{Y`uE5f>lEpWm#W)28@>R^N!_xB2?vtohi;m^>8pu!4n-Flt-{t0?! zX2A-DsydKnW17vintQ^l*~0R535#nAL0n>R>O5TN&7M)rMKyN0NOAYF#Ny=$7Dla&&6#wuB%RsKArMaBcxq#l(HWl%LgKRz^>bm{}hC7Av+EP!}gA zK~&)mTL0W8dhgr9+^;K|6WiMRpT#t;T49Fbu|uDW-GqcpwniL z@2Oo}W>&5`_ok*=6Fu^{Fy1z(O%wnsgB=5EPr@+!lo{BFgdiN4ipsNVTB>r7j5^TlFcv?Y2T59o9| z)|c*sGnp0M;_5|}dC?!@Aic(}XBe4){LTTM32y{m!f*?3H|N!m zOul3lkYD2FFLIK@N4!$1lZZ8k7I>HuLA2qN0?$$I1N z2i9vg#KZf3FpkFycwD20z16a2{lJco8c7gVX<-H~_~_q-id+}*!z9+TEpzngYz`a* zVL~wRrfqQesZySiThYe=wt~fHcZ6Zm3Q+@gB8&bQkBUk=`AMK+r8-Z*&eXLa+6bTR zCFmug&R2c?ZhP5zMNMH@n4x^JLL6cLI^azHWOf0VvU(ns2dLu=SF~f*-6tIo556FJ zLrB5-a*o3%`Gq_2i%!0>PMov@UW~<$)pI~~ptSwr4duxXJToTJR|zXzUO;7EIZV@O z>v`uHd=9^}qALNnEJ3(jwvuoWYT^BS`8(|Pe5JRjv2Gf{mseml&Q+zC$|a3^QV7U= zffesQ(%5;bcEa9YR+`t>*6hX$H|_d$jkg-B>IXZvx4UDTPoCJpUfp(g_BEz;G`_X1 zbtpMSS!-xQ#~G97uHW$S6&h3I;FnMVJ@*xKH=NVKL8Ct0SwNc{S`wsq^fU62X$&pL zI}Se=dxkg0wn+>Js{{;dG)^)Vg9?gOFdAyuTq%q*!4%f96fO?pg_}5$`};~R?oD9^nBy9KF|JFMF zZJV3_nkGB%+vejxwdLin+tT_k+y34kN_c%s?uRx%`|qoff7zxh^O~{#vHjV%{?Ha? ztM=tDf6Xqe-<0q>Z`In0O|u%UJdqv;Ntgw9+g+JOAJK#B=GX^an_aOfO$eGD_5h!X z(Qr-Dw3XRstU;ltY)am?db4lajh^z!h4|c>acV0|6}!GxwS#)!b{dCDw`kWc6s_Gl zu^--PY7jiMufDQki_0^1EHittSV4!;`QaAPf%Zk-m;g9)Ww;}`EhK9_C%BLwnvj=P|_rche(<#JGWr8C9$_)YzY6V556dPdR-pQ z;HhS%rNA{!2(~vj?0~Nd5}es9TyN}Y0>WWU{VPpq_`)k+wzX?7+xE_5yj3hyY-vzXW!HyJ&?Nv_29<0cgR!_hl|L)HT_dWNBuxM>Ip!SmyF z6EKc94%wDj`iOp`oCZXEVZ;XvAo~lGvgYvMj-}=CyPtHKbEp8qWn871({UK31Z1y_ z9c(yq%74aov@_{M2n0>Xb!7OGqq*BKtmv3mHOW{0dIMx!aTx`lvTnCwZ~eP}Zd>>7 zTUQg}Zd(Gw$)TUjKkD!jT~mQL_>``>Qkn>(Fm4w>&k4ghsk1aV3*aFHb07N53{UZ% z+WMtM+Y_(c*9864uNz|<8Uik#@kzr&`X4K&v62e!K>$NYOXWP?P}UMYD;VNe6O%Pf z?ln$SxJe*Z7oGnf@5zt7tnhfoH&&-rUga&P%e8Cqj6G#y93|aY3IA!p`;b=x@uWBq zf}O`8aI$JEGk5qR1p9;_An5Uvr#(y)Gc=jt!41-va$UgW!)XZ}A8Ab%R6lH8kV1GZ6sJ_r2+7C|s-h!qDS_rJxCmckf`a+i zRs`E<^^8|zaZ1nIDA+2&tw9{@^8VvD*=l>bUb!r3Q zI=P)om&%i`Ob8(i3K51Yz)p@I+5XP|r2KzV4QECS?6Mk9-=^Ve)R6Uen{%d z)#seSGdm@3zBiN@)PTF)mfgF1&t}RsE7vaCqX&=e8-M(JwzYL=&)xhDbry9<%}Tk0 z*!QdX%EP4ZU1C2n2i3S6_f!nBQqinbv(AD#Pp8L)@l89{) z?Q3@X7wXw;g2#1lzw>xQv~7r%P21gmVvU2k?<-zdT~~S7KXu~ITWU;k zVUxN;=xugSBnTgBCfv79m-;l#x~J{Rg_^CclqFCOY-6)-{UZeY2?@~%)YG<9U&mYtUL9F?QA`v=~}Z@ z*JP<>ExrOoYx6b+1D>5uxzWAc9I_?XcX zUJ94vrQdlT63%&mCh>O`a{OoF1l&Ea`2}xi8*@*N&J#SYj3v#Pc^JYEd-e>2;|%3p zVC04G$m(&RCo_)Dr~<=(-8`~~DIl9;fXP(H9;w72A8p*-;BE0gMwJu`>7ZPYqn<|j z$l&G8A%}ipA6lhjbmivckRaUUn({mjj_(LbC2%a+ou)#FNd_vlr(|;&MiMZzx>MkP zn!p4aSQ__jaCB(5-~O)cY&;g9fnW5j-E8<_;lXi7e31z!0+~8B!kNnC2TqY{6cOYO zA<+~6Ob~fh`cm-nqWJCnLdBYzxO1S+4`?A6vT8dip6kMKRpU*~O63JB)|MpL&RKbO z-YT4m8zGlGQVmot7}$~FjcP2ioZ+XWG}2H%!QeG zn_FHN@1IxvCGl5fjxyS;x20IAr9Pyc5&S=UdB-8zg~BHK1wc;(UB=uHhNBNMR%Jn6W@Jj6R(uyzA{qiqQn3+J~=PM2AOMI??Q|s7_j(srsk=X$fPuK6=k~ z_73dEORw9~+=@EcoCKg5g^LEpWFj`?$34S*62E%mb*s)UNN{{)AKv@Wp1<*$t*%|M zxy7o=tqv{Ww^%HQp;oO@6=SWa)6Fd?uVSvnvN}}Px|&fQo^~ZH-m(W9AKJZ#@B6iB zCp`&QZ4DT`N6Oa&Yc(YBckgIYz#+Ge?@c|Duskz)P4wEb$M?Q%+Z!D*_?mah6=tRL z*TiB9R#AQ{Vk+B@@7SQTZ@=)1ziA(R{EqGHJha7y716L{k00E(`oXTRax=B$pbK|0 zGMQMttjU35$}8n134-)Z1Sf7wXta;5$+m~?k$3P)b;gz#%QjyfNbsGogVu>~G|?G| zS*fl|GgCGaGTSs0EPI80v?6UR|?Dj@h?pK`~3=-WKvL^>oq}`X7c=#)RfR*`oGC8^Z2>?heifLfCMV+tD4@ z?NVV$km4SPgyu&1jsD!h-G4j?-}52-97OggS@bt2NscdTu2UoV7f5LfoQkrHT+Tuc zM}OlHjay92x3;Y%;jphh)9!K}q#-`t7oLO#0(c$~prWoEJR&$k&^p7j1o3hIVZ8|8 z_bcK}vvXD3J&*v?TVwHZNg$&cFO2V2irSpHH|YNOAA%>iVi_3NGP@@r7&Q4*j!zvBXQL zxUqsMA<(zKv8K+#&eLW5i2z!X7wSRdXV^)HaT2m>65`JilMXjXFc}HK)%)!j4L=FS zX%4uZG`t>VCQf%l;KvQYF!c-Ja7E>Wgk0K5LGCOXu?-Dg%R~nAl<-)<;L9Y6yXo29 zP!=sGG%m5=L_B^@1bh}Tr2B%1(qZM8eZ{l0v$in5=wEp$G7BKwY0bFW>mzUUd~u}F zYPn#{*T7!8X>Co~x;>4(YDW^8H6DoH4_E-*P$JFH##IlalRkb~|JXaqC`cMrlP|Agc2p}UrU`A;fbL;YL{!^|+9 zaB3W}Iw2TR1wy(5KaS@*KZ)Q^3B%~#IWUcwQr5GZE02E^K7-q1>g#dxsX!u!R9kI| zy*-@pZ;ElySj!TC-<5E(A^+QA5Pb>3aX^hRctRZ;Mgo%=3|PtKlgfk|Ju_Z4R3WiC zr$lAP3x;*|KKG?L!IACsv%E zkg#8~1?8bSqZTKAw^Ki{o(7Fm37K(XmrqbIf+_1s0A8LeYLdWVLj>N$9MwLUcyP5{ z-X7rtqtcGaO|%|{3ItbgNpkeIB?aU2?GZ#LyW@)P)IW>|#Q+65t3^i)N(|1L03ZM- zS=hI@5I=c1QDNBM2xI2JTwV167B_G-n+-8Z(U7pjO}i5wZLd zAwef+VRj%xVQ@^W;m>dAfd3Nzm>EOvYr|g*V{xKJtFu6uBd^MHa?8JNa(@gNWO0XZ z^s>4M=d*F6vOVZ>Be2^;caUeq1K>kEIE>*4u(_||;H<|-Z-2)&9z3vm z1HofgeXE`lOTrS7`rcLSDFXfY2qw3jmrsLnoNPvif2}Of`s(oRE?)+Z#8Whel;&*; z0k>AO>fEdgz_Su)5qc4R5un(EiU3xUxj0>r;LJ1ZB|Feqiy*(a`n>q|qRr0|Mt%}@ za?hX%@MUsZ!q{~*3vEi_36DJ268<{|nf3=+f zW`ufZq~fTY8pSl(p7g9yL(1pcFloYcszcoM>?vCX8v*uCP?%e&6-M#aH zl`B1Yuv+I@y_R6Rh^HUAo$Dz1&CV%ptS84{q7z3zzKb z;cJ;z#mAh%x zQqPu_N;WrpECwsaa4Z^4%-P;9_i$1-D|Ykszh&!}U$CX+IlFLaUHKI=C@)KJtBPKl z^@=(4RM!1w)3y#8cJIM`yMONk#a*+DE0^uPcfV(kA3d;3moF=C&==-K{e~)67Hod` zqRPY|#9m?Lna-?ouz$GK;RcDG@84xmU0f*JVy&Q=@u@SWR%c*)^+VO4I=|`yCdK>i z^JT@&b`sHgvBoDa#U}>ghug%ICdEMUWU>GN|MW>jK~ykF4*?ay6nY`F(x!Yit-R${ zge&2kA;?a?r<52B1I7d2P}V#i^av+>AED%~f$3Of?`gur3srpDqyE_|*3l$^gNTIl z{eQ>Ds;{E$K6&JmqodE%tYNt#ZzO)E z4G&QjUVikG^8gC5^ZYT33O{=L+jjr;psjLJZPtVO#Fiyq@S2&YSPVX5B z3eMC=GDDkkaoT2QStuFU{(eV&8ewbIrX{>qs&h6w$0{(l3dG?mge_lvEFpyPQfUao ztX3jmmDIO=HL<&Am#)5SD-wol;=vr?f_DD;7Ti<$tazS_;hq^;-l;Fy?Kus?An#v>sAr|_=yWyqqamD~9Bfesc zpbHIIO;((&$WBR!W(;Cgm=$~%h7q0>Clhws3SW&`0X~x4z&Jwl;?Uk9tG_Pr9{Yj_ zhs8S@Gmj4U#q;+>yPyYxF!IU-6A0R)#u>I;pnUry`@xBnZGI{pUohE9aanc#m0x~S z@pU!987Tie9Zh)kbZ`H;e-buKK+rVWcy8R?EiF~TM@icmv*|HbYB9bhm4H2lEH@r0|ulMQ_O4O%>MF0p>%vP$Tpw5l=On9$c5O%e7+9dk6D_Ps}`J=EXH`j_?&k z#my?FcO=a%UZn|b@W*cFcq6yctef+d8;?t_KiRNUHi^o{J!n&?Wl8aD{R3oTzt;{;E#US zZr}gNUU=aJ7w{nhSFJ_ese6re(QpS*M@}MR*Ts&6NI9z zE|zVf#6ISNgzyR5-s}0QGK_=pbXsa^X|`mw5{D&3TY1C*Ntl_aA%T+vZ9M$J{m0y7 z;o}K8Q=33w$HDDV_>pg#7>Z`Y3N)|EamO%z-jrM#=AV!utWrU^voe3gfjkNQFhv+9 z_0jE$amwy@Rf>_a`fwrvy}q}r!JX63Cu%e77iO?h*~lM^S5eb`7>JnctXx1CZpu#D z0YS<58-n+fk==dW4T!MW%`s_IV_~wp!!TS1{8%l|g1VHOldl_?TDZ-B(cC@b%g7+_ z?Ifv$ImD5{-Tc4`c6SR1V@gv$Zq-kwO1Ov0=?!6MIqQxYt1DUZ%6F(#&#oL@i6uH_ zoHOx7>wr$r0(eRemw~oF!VmZH-=WQujkF>R?qNC$pW@>A6{Pmh&!`NjT(==C0{r4X zDp4+fxT}`J?D%f*Iz|t-z{e~71UO2@N$!{veKX)gDY-G`u-muy-uk9(KfVofrX}c4 zWub#Qmi`FuU-P6q*WP)@()#gm9uGV1f-PE0i}Mmn)o%}aih0`Rmao|S(z4CXF4)52 zf&}5XtF~5V+XRmhN$_PaDl5V2)99>Fa`t3?zAB-$WSbgKu3mq|R#(;}#7$`oD>;v3 z@*JP}DU!;fqOebOO?ZuZ!(TBiNYGxtxMrPt(~6anEzGh7BLqRN3G1|aDt=R?I7`>h z0<;H&uF0ucjdd5~!rtVXWZxB0wj@7ybR$qiFAwWkF+YP?R{O}vkJ`&Hi61|*tGAU1pS@PWPd%1gF( zLBeM=b%hCfI<>sdcJmWaGOUAD^)+m48o+>WUg7 zTPhH6BxJDyI@z`kE5GV+-R8a`M|jaJP=Zanz3)N@jrnwflf4MHMO$0GVe^{B&q~N8 z-^|4Mgx64KxpU_&;kE3-l~-+L^$m3}HNc~W9jG&Lc#nH}VK6gAF|Eq1AC$N!0ef@* zuKmuh{ck0B^Gxro%B4=l)_{T-l9=zIW}7ERM_#6)m@RaH)(Ff#ON9FY8p!&cZ>f`R zC@r3t-m{CBzNGRzQU3N+wncSlwpgs%#VcP`;6uCh?w{IwAHFTYw5J&{H&)C^U|x>B z!#uDdX2D9gL_e<+t}*)XgAeV4yB|x~>shr}v%mXy{{w~NwhS@Mf|zi(XDjQ^+5GH^ z4HTZcg_&f)e7M(_Ytr@(B*?eAs+%bZ;T2n|O30^bCB#17+_zSDpnPiJkPzSRi!rF4 zN`--4zaqL7naHTTs#`C=S_7Bva}RM-aWY$<=M^9^4ff438*aCIVvw4YXqJ97n358M zlpint93D^jc8uerz9uLzt%#o&(nw}T>9o0ZLhX<(9@7d#x(MB!!%5pc5YudT?YP@g z{->u@1H?rjj(uwEaTdd`%-gG9 z{xy|pUv;kb&!1N{xgrnr8TwE3iReH4=vT`7D2`(=?6=Mo1&$-6a88iZ!CQFh->@RT z#!enY>1;M52{1)^stv>E8)ZqnH&&*aPb6}(FBF2H)e zh>+aGpy-cXf^QyZ;tUUXuIxMVpP$v(p^nuS zZ|N#8?fQY@e5mwu+Y35)e-V5dT{0b}X4Y->;`8EfEs67bHN|d>lh= zLFlG^aLdF|?@)3IUp19im5VW->);rN7&l*i@uK~o|4;vgeeuh$*xuHje`O#;UP5Ka zw@k?Fi+}>!m-|xBxvtaiH0@Fl&XfwEpQC~Awr~cr-Rn!H6y)dPHi^c;mVSeDxzji# zBW@#S%&cNTAqa*CZm`Q?knqX|H;K(nNN`eEY`)^;jX7>uBI(s&bn5tgIvLLHASaM?r>uBSyR{*~uPh<+ zc?r+gtXaRUPE`_p7L@Or=u;FESP?T=m%ufvS?Z)+xNynU`uat?aPg|bfd_ME^^?=V zFemmi&)TX4nNM9D<~#HW2I18f6?q=Y;eIC! zxN0|j3B;#|T+bsng*lm+QJXxrdR>zpN%+LieZlM>QzbF%I}NMVS!*}zsux~B6jPQP z*Qv28OdEp6(brhXi2g%4VD@XO3&MoZPc`7rU^29+y@+rMP7JuGoI8YhG*UEc-n*; zgS@hzozLQ+|2PGB&Rp?FjM*pS@r!>p10#FnAs1^L&RT^l1aMNW10f-WrdS;(IS#{s ztCT#A?3V!=!n-2YfRP#O9CtRML*l;p;M;F~)9Txg-3`7R5_1Zl(@OSH=${rwA!!y> zP1@NO*X}f}(G*`4-??;!1K2CxXI(fGuYu>}{^)K9A1NwwhG2~Y?UyfC_a8p+8zhz_ zkRb%ICB*M7mym>T<7SPAtV;GISS>BD_`(B`afl24$Ao(0{zIEzm>17wdxXX(g(2OO z!G^n!kAy~qI^Y7~9{v1MXmscNgM8jB^MvR9Zfn#s)G{+{jZZrkqGmiYBO#nV+Bk}>$2=li5t zlsGY7aXpbK>R0ut!K*x5H4eM8&4O(heuG7+q;YuOYIAcgu*bFyl}Kjt%h@LKJ1GH| zF^X}?d7H`#zh(O-E7T9}+_&eSd(PG`uSw?U$E14=bUi~`>|q#y27t~K{}{V@ct7vmrl*_@i3vBjk;YUq=;_2eTLisu$C+L8od1foi% z=KGMQH7oV9iea&pWEA|Vh&eI&l`oe78) z02m~97}7}#>TF9Mrqij5A(WL@4miN1gQ3kxVC{F=g3~uB632 zxA}z?sb$Tqt9Il0oA%llzofbt*ewaer;}os5|)4S@BGK2!@3Ld?43sX^ohgClncSU z{Ki8UmC}sO)ubw7Y6@FWTHNI+{x3$zb_-VDxUPoR;^Ibz`8h5|*Fd7o!X_C&C?9!b z)qs6+3T&-LQ$jo|F_c$KO9DAx4om_$wC4c~KIyT#cOvG_;~PgwK!*(6D&fsp`8?9( z==fw(6RdsXjrrO6s&SYU}ZD+-hG0;r4VE2+<5#2tlag48U=AmfJzm=Vti!j^uM zA34JWlXe&VxP2@xhm24D!yuK7$xnaCk=P=5c~;$%5Axh}c7Dj+J8KA-u#K8koIU#ZAN@r zjedc%e#9A&F~+mA`F9|nCKHD#4%v}Af;#d{dn;DNrb^n5-ZC!|BT*jH?7;z z2qpjZi&t!MdChqx3l)5|3ggST`Td`&&052^rY3$;2{EP(irVC4rdds)TF=ile5tcmmdJUz^gnPt4 zqD-Goo9s_s=a<%eOq&z$ouAbhCSJ2kCA()3KHNH=5(Sa>a&-aG#Qg< z|D*IiOZvd1#`M6WE~oquAzxtp0GaZ|URxGSV$ZE`- z7niuYb6MjtZx#p%T1lAZe)uPwPi$xBj>7E9Ip;m{?1$J%XR4~Yp7nfi9%)A7*mMD* zoV~^f?Gw)PS(#!5xKiO@UTl-`x=~%Qn1?XWgobpOyhKT(-DU6zl}+33&VlXi?Alj< z^_RqBL|3&B?o-YMr5Hidh=qTrT~-MKMf7(q{GhJ;q@>T(jT^&0<|<)$Etx(a2u|}e zhVj8QgEOn-%xazY1PMe4u|d%xRYPGMBPC2B8W+qV8pnxdTnNWt6(rP7@wmo-Pmka; z_{1t9oq{W|CMLL+gq2(6;E2<`4c}MWsyF4&Ycki=IVx(TMRlS%%_x^`O9JfeTi^4O zo)<6OuwQuX>vmm2?!x@48tR;c*A2US`(4}H+qIi7{eper%fD*N%a?rh^YQ)LimzoI z_U;~bt=(zZL4C*S^=;eV-LfZ-?%ToEeS32Mj+IO6cH_k_t6T_ubSULg9Gz!8L#gqq zUFbToKKxiYHdyJDoiVjN4B0rIm|{K>1Q$C#PVdYztux@3!in!Z@BD`3&Wd{ZA znEW$Pva&m=GWs@%LfQShDhJyu7?45`PJ)eom=v)B!^?3hg7rvW%!q9k#`Ya@r?mS= z2)hceFs%M^z>cjgO`ecF;&Y))?ynCVdOzr1R(%> z3g;1JbAsXElTsO(&#M^RNhQjZRLXm(@%RG34g7#?ACV{qzzBB^gp>Uu*7(cdbjLT9 zX%t6bpD*h;1a5=}BCasy-}Aw}UCS}rHx8qIURRl$x1toQ#Wo=|$^KW` zF)AR({bTds{59@x=AQa#379i{{S)73BEdohW}t|AIBqE}^?et9qpksP*z;Q`tooQxRvEdQctYt@ z=^TtkadjT6!*iNA%+JLmSzL+Ai>=%QLR+E!r!@v5IJ4O0w_2#)SOI6upuKpRlze>c zoUrfw-~;;~|HZ#`I&fGrWC9@9DdwyPJtC5D@Vb{yn@T9&-p+udkGYN6`OWI{hM&;^ zx*cy|xzkW|?@av5O+RiQCrETP1`wDwN=%9YEBC=v2~I`~a%XVC?3?6;52{~L(rrjs z=*TN(+XybvFlSV8irpvP6YBJBF^Ps!p%K^BK~HUV;j0pER@IoUO0l9=u+H(KzYKQ& z{)e`^#q9^L+RYnZv^g=933Y<)_Llw0H~*<^?QQxejpfA!d;a>XR+9ithv%iEexKEV zrtKfFulPg)U0)rD)j~d-FgTw0p5A_smu18R)fu>=jqMg&TaWGHg=)aljv{fZqi(io6~n7K}}-2s2O4_zTKC~gQ(0)b|&ao9^yO{o6FL|HvX@O4J5GNQBM+HN-OpmE@z z=TuA9?;hEO%h&9%duj`dSA6B$SDw|mSsm~JL^Jb!UYHZJpxo7R!RBk6{-xz0=!#*9 zX?aUg`vZ8on(YYPt^{D^mn}23irN8X<=}zhrj57&BRa)ZI@3Ow@R^F4IE)ejm6d5| zNt~=IBk&#$rfg@QJC@r@E$;8E6s6BhXhOq52GL1r_5J3Hu1T})g9AI&0JOGPvDu;$ zh8b!AVzrk^4K$=K5Qt%xhjMG|?Rb?jpmn<)O}6%jD*+6!UcC-7&?eM3yp~0W#pQL4 z``n13cs)e48znIW7u1q_#6>(FRhS(dLvEriX*h0}GXy$*icfL6KJM>fk{x|$4(0wB zl5`vPls6nu9tIUg=QYVrxSXf4;sxL*Ew7wMUP2E+l{N!TiYu_gG)(iXAI?)EBv)M( z)epJ)k3DrWs#9)s@q?#KMj7-qsV=E+bCXF~gIP)bWeupjrdW`Gf^b4xi=-l*$YUtV z<9QqA-E9t-JlHvP-Z1?0@N&uVU~#mRPI$_k!nxfM+}*<@zibZ#?BD@o;j(gwl7xyB zMgR{AQ^pwjr!-u16F+C`Ip^w3;}+0w+v2@%zx6F^?%t0HnfhF$pC(2_MYJiApAvHT ztTQ2Bn6LW626xDIyYxjhi~f;azxJXp5cpUn-ot`|?;Te=&08kmNi%9=a zKo;uwvNtV$z*xf9#+l-j)#huqq;Y*#`9$SZp$T#ZvI=7gw*v9i_tb?9`loI*mpCCkMt@dI94I+cUjX$|sW@ZeIHE)?K@_zG|1Rt_Cjkq=X1@pMX$BP9^uvL}kU+2R#2I zLgk1_O|bnT1}DG?o9jSk)U<>x8pFw9#~QnLZEyQWw)6O1+uwWOvo#K)a7Q1^gFQjV zCor3ybj?bcB_DFPZAYDbSB+#(_%|i+ROBb7p|q!FBotR}+Pk;EC#KM_mv6pduRQ-1 zo3AL{=_6|&@mf&X4tDO_>im*@>CL}o%WP%PjH|$GQRz+<%6=y>oA_Z`>=Qng&J}2J zYC2YFX{<2hlT!(Xef-J3|7T}NWGju#OY1J;S)aXhPFY3PXzicF zB=GKOGDJNh2y@*VH<7Tfn0$Kus_kT^7R3(6hR{c878X{0kDPD+_ymY;Gg6Ar`&T&f zA7=keYUn1w?D#2+gFB|M;R+6ej@uj`J^Y7h%S`)>fMA@B0|CaJlif8}c({v<@DXB* zvS)A<4@{g1c?`fm3*+|4M?^Bh&A%0md&>-F>i0DbZWRskm2%0dY7?_n4e)9Q$LhCp zvvamMSF?FdR%@#8+477g)C|@YD`)_u{ZpUbk%qaCQ~)p0Fz@5;p8}3gTv-9fHj4|>H6UB1CK>HMUwU5xAg6uUO6hL^zGqM=<9Ju;*Bq?;4coT@G zO-(4-Ceb-~BGlAJgv%e}PsEZE`Afh!?J!fPGnxd=)=J`0Elr$HG$w`cba>dcwY7`3 zymHxvRu7}R^5qSG7lz?a`t$Kj?(lqXCrrqtGjYH3(VvM=%}Zcg_C*Z#xw3zi?RQMF ziH~?$5P0(Fp$pvWYiqs&%7O%A3xaX0{@9NC1GUMjKMsM&&iM=OUS^dwuX5&ceH!HP z#}9-aw{3p;iY+Wjpsdw=(SVgnRz-OnDsFM$Fg$!ryiob2DtM%duYxQN9UsQs>fu15_bj9W*h$8G4H75IQ8{&&Rq0W8H2)#@?$S1G2`b1g0pXT9<5!|rBm30!PU5pW} z{Y(<&l0SA{1x4spBt8}~nV1*PonKhD#g$8T?fOlPgXHCi>`bT<8UD7R_@-tZ@>K?{FcTkCIXBh&=RA5+?@JyBCp%zPJW$V zyL*I7CJ_un3IO1jRfZHlsyg>`B@ByZ@EoWN|3ZhzPW2Hu;V1c=s7jmQ4n0o5B8YHX zL*PePHB#_#Q)M6ZPpsG3x5nOGYc-!}&?$LWSxiZ=JC-n`iXBW8D+5iy%5s17XhRfNT`4X0gSF~=-fj>i+53l? zli#9>@+hh!;7T~ByutkN3W1P5BKE8SXibd17`T~xrFwTfsb~{Upm6s%a=%N6Ver4O zhgiq}5D+>kWT7-%pl!5wgpUZDv@iyMLnY|n@p1A<$VWzvUo=dtAbDRRoLr(kC!7TC zm?6@J?X4pKCwf6Vy```YC zbsL)*$@vHT(+@S`lw%u4^pgX$$bw>~3%B9SiZArl&-V_J(3||`=N2_yU39yDp`hPN zQeu$<-tAi>$gA2f3oP&zc=I&EoA@ECzjyDxCmuO%7cX7*vWpvdc&=vjC*EP!ir2&^ zkGa)BZF7EpR)Sa8>IVnD>g@bpe0@@re;6G_U|n%{=bpwMKKq|7(~kqtf=97!tYEc; zOSZVUWF-l7!~q|tY$3?WP1%EOBfRtSnnL?kWAyET@;>Mv$YnHkeHNJ3SU&4p82MVm zy}%NB#V1LdN@l?Up_0WIG7Rkp!tr`g5|v8TsuJG0W{q1mykRJx+};Bk#)1$2E9-(c{WYzM@L8Kl`9zJA3n)Jv{exWB@_4<8oz0z1k;&O+Z#ugya zSiFX9G0Te;d;N`%jc+xS30aMf&X1|urJ)r=&ZV5E-!ZUy%@snWW_j4f( z^XNtNg9?wZ5 zn2L!gUqwD0X#k&4p5jU`bwrhc&G=3lPjW)cZ$Uz^CIm+lHZMV-q`Y<7T^IP_%|+k7 zA_g^G7DF0nP+0SOh#3H2_(j!q%xI@n#%aHr>`-~BDXrDHBa;}aavqB26$zYe(Uuii z1TQ`{#+L=EfmK%O!nZc49l$_n11J5q>PpO5=}Z=sjs#|z{8KkyCJ4vNb?ntb;4Bp< zt#Hz}fAc5zY;kQ)GjNq#IJ0vlaXYnP(PCP4T$z%4p=dkJLpx|6QWf@ibK4IC`il9) ziB(IJcJYEHYs!OPK(BJ}V$jCpy28#X?@Cu~qOrHFHlng962IG$Ehb{B!EnGL--#im zI;DQOGt=urP^~R$;-@%8OYmIDJq8xS1oQWb<3IW5RuMMruzg2xH@~|S@L&if(dQ!# zH@F?&h>LuN&R=*&T<`%^PqKjmPl*1I0u8~P!sPjmoJTl!Px$<1!8cZ=$$j_*PWZ)N z2AK(SMrALV(a1WEDA63jc4cwaYHA;R&g3&JTPJ4A5{TInQJU67NRyq}nkLjU)@^m| z$=-o|bo-%w=dB;u@BW)VvVZxne%GeOlWMD%l}jdMkw&B_P<+rRY?1CTKl-CTgYTc0 zaUpNG^XujW+3a}4G3GBfTwTZC1M?sLUTe-v+!~O=fhRzWq#NB4Y>(3_S^Q zeFS*r6Y8^Pm+_HvHr)Lml@#zST8{&-^er!6QQ6gYl@2!pF+OOxaT-EX7ueOVV?l~` zKvgi2o$?zx_V)JFufri<5fWd{i@`f8!z7Ueborm&1O5D z8|gCwI87Nd`FhBW9h9Q6U$+AZ*a-WG>XYI{zUZUzmqosGp01vGjZWrfVxyk9xZ#}x?Rz+us|ir(#>39`VAs(jAEuw>GoqBs8~;gM#kBLdJF z^sCzOhU+TaG4ofLhRt?Vp>+w3n&ospv?mYVx6aD~3TDA23mqVm!`dvK?Vk zp&c*ElUIm{7vW;RvFi+JZhp}RNwy@w;0TXD$sZtxyj%7bIH?YnzpA*&5#pv7pDVwcK+I&U%U?5Rmk4{8Tm5;-6 zth}+OnCsO#y+bkIqOaB!C6rR8niwCSj9~uUZ%VQV{FAaHyfVzb^KsMu@GtJ!#pSYn z4|23Mz!{g6*CP!`x=*H1 zv?rD2;?)`+96o z%*hSFyjCsQ^UtkmveVY2tZxVPhP5|pDMN(>>GG}W<&86qxepOC-L3eX^j%jT5@p6;EroxONE9^V`_IKTKy zzh&2+d)?4*^cDW^pFG@`4d8LjUjUGMOuphni$9Cb`EB4|G zZ}^zSq(4946DzF}q!mGHui?Ve;>wb}{>saC@7`UJEuO1>@Zcl)9ZDddQyj-uTUfT% z!ENzr4x0aD;B7%b`QZV@$~AlQ%`a)ZXOcLhvLVFf3A*~0=)p?se*M4>8VwgJp%DTe zq-yMMNyvYqe8nnkhEJ=bnPSzhJ^xk3FJ35NQKPE@Ae>Y0&PcpwxXYHVXc>hkq#Hovf-u`J5y zE3wiM{^8xo2h`7@`aNIxI19qYQMQNh^%pq?nMMJHBl$tek`J~6(q2f{w`*t&LO5n? zAc8cDN+ig2ocntZm3yT%Ro2)tV^xhyv$cx7{L*#%NB`)5Z!g|lw@Jmp_~(m93SV$u z9SaEbIRt6%dx}5mR{oG4y79+Y!Nmj&JPzk|q8pi4F29jXpAY=ZR)&L0PQoCdzzOKe zt#0P8IK!c<@7UhgtH!C;qwW*iy!U~1n{V6h{)bj5&e+ju zQ!}3*N&va0xR}Whbx8#0(?u{l8Vi+3L-Xfsi#&16Dl1H;BB4!<%{QrY>jGO2)DBqn zRKumiL_dp8n{Ra>mA(W-9>AF4^NAXc&p2t^2?L>r^1|5rhdn#c?2@aoc>JK()XcEY z!wx;~NWLnkgfVl_i7+&D4nL9SZogqyufAj}>ldwBTd;+NMde!pvl632b^5$+vu!Wj zcpV3&tB&Awk2*kQei4m$Or<)BIts-vh*!vWNt2(v>JGPDR<$(R#ytB@U-2H>DO-rCWI1rwg1)aT8q8&szU>Jm_ zL_W)vf^k3}1{DqkvVRNVQS8ZQ>UpI;9)cm>l5Ol9*`IuS!~WuX4>UN;+J*H6TV1VL zNs~AZWz@yIVXzzq;f9Ir`jKt!HjQg{mgb7K+mygB`W0qGFymCWXsHQHWpTv{v-4^P zNB;0eb$L}ZRXfsPw7RxpkMDl0a&x-`hyRYPp#h@VW+j=G4C+*ssyxC7DVf{7>}jRf z84vs3<}_(s^frZmB98+#4>OCPqLp`|pa0!IwSV&e{g?Ky{?)&+fBmn2-~Rm1zi;pU;D>hm!9#oT^;hhz z?|ski-F;|HO-c}geR8B|Xqz0Gec|Po?DDl6a)i&w!!wk0L6ra5FrIJVy!|HuLJ~gz zIVkB&{{Z=%PeSbw$4qlzXd1&E#wl_+a>v8F{)kkg+3<--lRM$XhZh#-Js-Sm*y(p9543Gk zLVi^|ojZXMfG=LYVhd@ZOp!?QcJ^ttafCxVeLfH)(dQ0PV($ovNS_AY04En!ab>@s@A2dJB%CztKn;$ALa2TRVd%sbFHBlx zZcz=cZ1w#=(TwMh)qyrN7OI0Iz^J1gYj#2n5HE^6(BWgi)NDzOf=>eQZmwA4-dGw6 zD`$KT=gDF|Uo<7;@PZs0odSU#TwIA4 z6*#yBV?vz;NB1<6ay<4}Nnn#8<&y?Z0#D9ZYtXU#`*-Z!``@?s9=v6L^3EUFAAJA! z?ESlcZtvduGrPa_u07cOz&0DVZ13=a-P`<@eQ^7a?K|K9*Y@s@zF{BVf6u1pD|Re~ zGni@G&cP#ps?tBps8jQ;9HEqKIMY^s=~ysLzh_ZdI~4=zFhLfX8VwB+)SYJQOdfz3 z66u7pD~?Iu9s!e398z4#HiZD-lW`SQ8z8B2_{$E8Okwz#6<%r}$>)YtF1%n*1Df8Z3J#eLkC) ztRlgg*O0pG9}`3Bbh`eTeZDxb7hbrji9tbw9{U3&psVf?idpG_k>~wslrb|MsxSlQ z_*_o$89-sY)N6zo$959~?lT~}g7a*Q^Kdo=V8ti1Y$Rf?AYuW20f;{MKLC=94WS zB=O~7lgG)I5@qpNgC$?KP9^jV#LFg*)J8kvHxgjj!*{6q_bo}z9|XDVXCUY7S&A}5 z7)pUT<=~T5N&@`xzwrObz_Xj_f;r{iX(cs^8}F32jN`vK-K>Dm;x=m0C{F|Dr$Aic zMEp=pW4tc^Iv`iwcsgP~A$& z=scJOBwiLhIox#R_MZ~Gest-k5B60* zr*`qm1&<3J&7#=e{zKcZ-;zwSN3pD+v63&hClXqbBZ!9=0C)HH?D6)tZSNgu>}%M5 zyKBu}Jj2XlPeJ(<%doP>^uo*)(Tj5>ZH?q7k~?;k-iG4%&<=KPTSqbqE9vBcyznKH zeEQ;ncqwg(EjVn?W6R0X>IJ*`;xB3ZxnPG!bqVks((l@bw;$MtKf3EH-uV^GkU_kP zLigT<&W~*JciAdegLLQ!euT*$&j7qW#rv-deugW<(eWK1TDH3iXUzPuhh=h-I|uvz zcmtjM!qsI7Ib7$(!v`&E)%R7J`!>5=v7-UE0dSY>l2yuAY;N&2MYXDmXPI9WI$89F z1)HSvQ8`pF*Vp*X7BNjzm{8aepg2gy?FBr7ajZt{D^EUxH5`TnodAP&zJq zbvV#QgLI~)hCR4$?%Q-(Osdeb-Ge7qtgtG2*X}-iUmbnJj;0#6 zQ{S>b`O|OMlidxw|LB3;x&0$`=ts7FaMvC_xaW_G%rC85U-i!iF>cg=K0w@&to(>3 z{`!v0N2=c^n=PN5?C%|FGB6{$o=9k)w#p3k9;?69Ds_@eN1gZzr87a*w<4G>Nk~>5 zOD>f_JZ-(>N!#A<+joEP(7yHEJ9dAwV_nUlC)BPMW@}cgO<7Il<$mM2>Wtz!7IWpA zppxPqC~VITT0L7@nzxF^8y@=D-fJl*F!F-&Y$k)UGGYBDH%s(vb`GI=#(Is8O$~ZB zzr11}yz>JMYCNF~v!1Y)m}$M)NM)hjMqN_A_*CXK?7pn-BG`k+bgEcV+gjopofy;- zmWLT{4iRT|p=S~o^vobP+C~@y-wp1G13&%`ox`5BaE}yI-15j?sC;Pd4CFYH*~7pf z9H2usCp}wb@Yp?#5x!H&-A0h(IPad}yc_md<<_6_T;FTj*6tHc()p57vxS9OwdZ+F z*sGcl&C0%HU-*Szu!G%w^^<}vuFc!h!n`f4EZW@Coa%egme(%Y%dfs-3(FTohcV$J z_Q<$g9~Ov)b7M#4lB@z5sSGr^dVP@zWgfoIbc*+@2AHf%PH(r`-B5ykz5AE%5e^V3LJ=tk%9wO;~^^j=~Kdo`tSN2X< z{aO46H+0-EBkbLL;SDQQY8t=z%+2Bq2jBR#uUvXR4`G=Zd-zuBIni6joWv`lOKvL#Wq96L8@_jx-6ac z_)MyQDg=$AjF=IgR@a}`m|amB8Nd>sa9*!*RQZKBcO+s%g$G}abub?t)p--XpB zTf49#yn*PfJW6OkJ~@I((mqR!dxyTA3o+I4RF(D*80 zmU8AE=_8oRg+M%UT9OdF>$LN>!k1&2*-zf@?c1ToD{gP%YzvfNXLFELF)BR-V#WaE z4!$VxC1FM5ON+E65VIA9x>%g6+OPlZulvDy_H7ThiaL3-4!uby(>!q4Vh4D)AL8m~ z6v*ArLv{Z1T^arv3By|=KvWJypS)#{3biz9XL zmIOiG_126`63$XX4X@ zBFrlfqz4nBv(daso7acfPs?dfR;OborEaBs^Nx6hD^nKN9EbEKIY3t^DGm-l#LSNU zzzVEztQbnD6K#0{y4P#lWv(RSQ^#rCFflA-qBr@CGDjW*MznCipU}yT#?QVp4#zFb zF51qPW~N&QRx8x(xYJ_!*2={>m1)AZAJxS$I<~dBXS=%x@>2PyMBAeY`~LSow2mej zJ39yV-h1!c=H?TrEnAfU|AjC9EyZ)>d*cxP57~c3XBPM;L)wxu1Dz3cHxCZ&{{0<) z&8XMo#X6Ns`G6ktRqj?+uY|tPNx#QfF0BfuYYj~>Dsyu_+yCI5cYJFEpYT{2VO4>9 zLYY7@Py))b)5(Ql1{AMD(MkRY!w3dA&&+5dwY;V>(mw)+e&ThWln-3t8FUI`&?|$} z543Re_JrB3GC2cvuc6TFf@=yF5C-=<<1Qc%4zxhRWs}!9iZ4Hr?pO&`*U12Q zex!zlqOC_82V5W(xLw8}n$wQj$1%GCDIIB~>2YPZGXM{f&(JHy;r%P>53g}WOc_tv zOtzdhHnfC2o~2I&@KU%OUO+lgy08J1#C65LtBE+m@bTe+=XD(KuLDSvdZ0h@6f+0F zyF9j|BxyIfj?loldi`bbhV+EJ*dG&Z`gM#GfyYX(etmTpB)*=pU@$d_fL+jpxnuSH zTeiG@-Nzt+e6Ur5gWW=wgTiZk;Q$tb;nIruPpc{3RFbfFQ9_>ZPQ`0jP;j9$c=1mP zqc#$GfYz()H+`Qop3cW`7cJn8(1WdQZQlmKiZHy5@k}+VeDKJW^SV+d*aO! zp2-&iGm{%r7+!3hw3>wGJ>^t1LY@%KqOT$3HmuiTk%X0C_6P@EbN_X-+KDg1N%sd& zB!4Wgu2{Rlc8R|H)GTHu?6-gW-_v-;fy`qUnz; zdHB4A;h)I?x`!N~a{&fTm7zB0v~YUhn2kZG>1lQ98scnOh)NtMgZ8E1JwyMI{Nc&|l1}RGzU`=fFO`bIVpPEGWM9o_ote@E+lZXSL#TG zKWv}cwgmVeezYlJxT&;7FAX%s(u7r}C1@855>PbxQT+3>Q&y7zEew0%>Z)B?E%~9s zMxRO4v2`WvfA2>dqRqg*{tFt|G`YL8x$m?2J>`+Zc-(W$3UPg>ZfmQvcKON$3Asng zL)A;2AF5Aq=yRYx(6!C&Ex&DodS%rgpzi!&Bjrp$`}5l{ROiwBG!V}& z+x1t!ESj>{&e0sQG4yD^fDcT`HCBeBe?;4k_U&y%2{={?=k{dAV`xrs1zz+CAim5* zT6kq_VCbIy_@}^mPjR?OK?DyNb~m1f?4CSBJ7aN;{^p;u!)XX!deU-pq6M1VsN0X; zdE5T`|L(t!Z!+pP#KUR{`DehLDNft}@*n)K>^pz{r}puWZu_K;ff-sjU+^NF+TZzm zf5-mT@BFrZIq*2#J!cRMfrm@rd6FGG83P}8%(yFVcvggT7}g1bMwo_y!biFr2!kQ= z0ieNF#=(=(U#3UED8+D>%8>vWPj0wEr)&o~mnA!A7)FtF51rx|c@iuHNcVFkNA-yh zzWYu4C;#>T)%vZwN&`U=j|k`WpoE=nf{`Hjlf#h40^;L_p~d;4bvnly$D&UI=kb2= zYrkddmtOGv^8y(YY=kYfGT|ansa^0VceZvB8h^H8(dAE0ChhKnyY|B${GQ!->1(!r z;hK06XH)uqJ)}>rmCmGi5N&U7uWpUK+nT7oX;&{^u^-<0P&|6l_PM1=Ja-&2UB)5P zYownYAJ6(lSzrIc@Ax%@j7i+IB2OVJoPx2xnQO;bk>%R4lU`4Prpm7JdA}5YC3(?P60H5U~x@Cx$1&F;L!hAs^91OR^?au@>{GF>y*M)E0_Gd28$*LvhY^w zk&eXPUX!9u+AQAH51cM+Vc{W?(lmR4H5N)9VoSsB_J_uu{Lpxu*Jhfta8@j>N$_8h ze89o=v?!u>%t3V~7JLyUZ5Gq2)w1%^wY}|qTU@BvfBp~uzWtZ~hyTouyQ+J>&L$2K zX+g3`j|1`UAPmMCZ-`P+6_GD@N;;l7uj8zE&H=? z{A=6Xyl-n4ulbctgTqs6v>VpiyKn24Uy@M9p*}TaHGBkg_DJ)5vr3~#mG1$2oux~F zYOp6-%!gCNtlIW8+vQMSr`NPWw1hE=kzEAVz#$IA5}_eCc=W_1Ayd-T?Q-FADcd zc3sT(!i5{Qx_U*z?3@>vnJuSS*|^W7M4CD-k7!89)`*Ib*l2d_2k&z?#~u6TcOKjKez0xd{r<=H z_6HlbA%PgWl!^nZlm-&QXB6k0-wlenQsH0&bvNZqNtu`Y;nL}`Mm05f9jT@S8^zHFZQ&l6uU+v-Dx4vl)?%h}U zV-FbzNOt!3t<8#t>W>uz+B@x;wgnyv$O8idhvtOQj$nw>1zQxYVStb%Gp5mRC}&>D z4{$SXVF@e-C;!93f8e0-adYx>-=RJ99F{UbyM!2c{7Xlv#?zns#;@b39usbMk1Y#W zGrk)~D!GmPhb{?%ZVm+aBp|$_j_~Qv+gn@q{qMe|dWd$*iEKVw#!40g7ec_R_VE6O z?QikPGYP0f<8`KCM||q$%P-os=dUP&ps{CDfpcme<_wJ6bH0;XY75~!j)z$)#U161 zupUH4W#xQqWAL;K5V_1M!sj?5jWIZ5Pj#Zi2|4l^l{>qkqk?w&9WNt10yAD{=_g^R z-%s1t-8=T~xBgVl4E8y!(#dAQu>BYdg@YaKkYg{3Rv_s6#T@ExVW%Jj3#pXrlfRF1QPaE8FgFUq~X ze$)5T0{$sp%6;_1*nJ|wvZZmg+vzB<+?5X>b5(yXP|Aseb-YHG)Ahig0G>bTlM_mF zvaGS-kXtN$Y)9~9+ZqSlOH~P<<*Mk|w2vM58=`67?%a7`7nYapl{cREYxQz9qDP%lmza@@vYQ+5O4_J>xURIw9WF!s z1fTcH@Msz4bA&_+5n-(GP$Vmdk>6lY(=UhO-n45s|Gs8%|Cuc;{*G1WzGCyszvgOj z1=I6yx+o0Czx$zbIgE%CFYYPpAO)QXvQR21WvsG6k7b8-7Y@x^vV7#*Z^A?FH;#kd!s_@U*iqrQB zOMtIPD5_THCG=MnOMD(3K{O=Fx>Q-PuYBdV>~H_}e=1@1!0z0BS3P=8v&dOn5z~9{ z;Gt%Zb>T{IN8nZ3JfRr^2Lp$?s?tQDf>CnWH>=a^!-ZBH0Aq!lnRUV3+&{3sI{9DIPr8YEO!u{niPS(zNkTrNsgE7lOddiU0z{n0n?*!#D3?cU?M)eonvGbq?V z6Al<8w^7X1%Kmif{JduNRc;!gZbPV{?Jd-}OI8d)d1pYOG4)iw{RVrhHTfzQ)K>bo zvm?4_@L}~}d6qk-XZ#BXCy7fE^0{KJzPIPMQXn2Zy!)~3Dh${AFc8w7xE2W&j6eER zr~ureP0J-QNc{#5ezM0NIod!d59hz*kYaTI9Q(8I5(F_E!P|b=hVefe0?%=C0Kq7# z@Q)VY4({pkqr4x<-vdY`?b(2TD8jm)e100pv^!fmr|Qt1Usm&AEdwRS=TpxreqJ|H zvb&%2An*yNx8xGC!McrC#QTO z=X0^bBT(>{_;PL|#E;5t?x%qq9-xjfBKY~iB>81HAAhgYlHknsp4>da9#6Id@Wd~# zO|o5U%2)c>KEO&MtChn8RKYs^fc2#-*Zk}aUmX2&w{-XvtDrNQ$alr7{8ci=IyJqj zfCJmwdSt6Be3guCga}-|cpw<3MW#;%7N7v<-GUI*&f;VUzX=KE5*Sqm#za)CV6t>W zI94wu;LA?G;1&&IYol!72|4{%~>O=kf>LG)5drAnx3@ zMq|S|b+!+RCu1^nTV)X zDqpZt<%SwVJS%xRQFJvqJ*!}`f0p5f&-Zk2ZwLw(t1|H!0)d!I_UTwK01hEUXQdH& z{e%)OiZI@|5W0lRX9=FKWyQB8;rGeIN4C4YqZzkmYczaSdZASIJ;kv%n7Bs`fIMOG zKR2R}6pRg(g%}rP1jH+B$48G&1+F@<1k1_6kzKj`s=fKg22d#}(`l8|e zCl9PPH*41~zb2e{35l~lgN1p_&sBwA(ga2Jngl%whrUv;bk!XtK!)IQ*lUPcv{gn4 z=$cd%%LV^TSCr5Romhd02|^Ih1s;S<(PF|^RM@J*CFUpg@q>=ayl)rQtM=q^)2fvj zJ3#*C4otob2%nf>tiC5taZf3{#BRHLf5+Z=|B-$FhnseHt7~0oEur;vLUfu?eM{Ko zI-+W&YO^(!t-{S5tniCzyYNmuD2&=wO?(1;_mT7Q5i7$d*6i{rirXnBeM`ecp=_;Y zL(HXNylBc-gj#9J=BpBTPFNKOP!DJfxje#8F&vS7W#sZuT8~71O zX=WpCV-P-o@Rc36k)HP(cTH|?9pOt$rZGbp!5MSDSZV-~LEr!<2EO z(dcWF_~=wD(0md-5WqM94vcSlPi)z z<`-{jTw9Wm+HrxDE7dw}=?4dX(^R+JvZH>_xZ;d!R*@~d5=3ukp6fo3-&EeaFAakxB8AIK_?Plk%P{`FpUJzhPpV> zSlnpbw)=PQC_c$8?ib@DxnsOHO@6xVUC9&MHp`iyy{5hU&Mhi0lz?zUGH3lf$UU$F z_k6XasD7qE%lX;Bka%MRb<;VJBZAW4gslv5^$Sb+GJlhEV4HRyB_iV(!P z{JSgMYTCxY0hrnOWeJ9}>PXBc#VAz1sCfO5cvgTqesp>ScUGlga8fjH4m@zs3y~^~ z78E1O5akFRp;DcosPd`Mr$;S&{>JOJvU17({LlU;`-?ySr?$EC&=!~1?GJzNAKSOT z^Utj=oXO&x$|oiT{gsCaF&l|jNRAT7cv89IW#_Yxuh5F7tiHlm*~`@F?psTpozF)+ z?aa*`W?TnWv6-1U(WvZq@J<)2zW0}#I_9`DcyUn^nu^n(&ta^fQ|_6WvK7T>CnwnA z!mS>>h(_H*j|_=!(#c0K73g{_q43BaJZe~ZCg{`BjJisQ&j?D7Sl?GWectQ-(#Kc{c@f9Fe*H0CuDCPn)Z;HX)d#tw9?b-5t#a_66*;i}| zqMK%2YG3nWu#I$$4{aq*Jo72eg=(egOg8^{?b|jE!IZm_UZhIGnd9 z@HUstNf)?r%uch4eBc;yxtnx%iXWWaJhG0`9nZm7G(t<^o&lgZ+R%_w84ojZfnSWn zdHZLDQCZ!>Pcjcp0oe~?c;%pI=kO=h8GdRba*+w4n0_rQRV+K7SfeVz$x*n&u*g}F zR5k-Sxp8o7awA3J^+|W~@VNCaS2u^kFCZIfKe=5E9TS|hJ_VyNQ6`6Bpv*td134PQ zABPm+Y%pQFL(c!0ucB$4R_wV;OZK1threY1n}7c|?JGB5wuREV%}j9%P!c8-F!(JM zg7s_)pB18CMpy(+1X&*b`7|I50t-Ca!z$0nN$`_eZC1it!3AWtO}Mb=;|Sppa3)-^ zV;?cAz@?dz%yYhQlNXHODYQKfX8ZOF##jaRu~XoJsPLx?ONuXqS)YJSaA2GM8+7r0 zkN`8d+&qDqF%Dq7!(OReu$!-Z&0cuzSM2$hzi5}QJ?GmH>f+to+gtYd@gv*a-4Wkw zxo{ulR-B^Q1o>5kI{lO`xkS)+Ay#Eza`}@0JiK1N<$I2+a}~wSLRsAS5ZmfbZARnP z+LbF-)fhHaoVCO5u^({jb68IK5t2r3#z?M4A}vXo%FhP)az&*QCw{4m?q96;i)ZX?Dcq38v`(V z{ir^&yg}NilcUGBBVm_CJ`Tdef4sa*3CepMdTLY*M7-8BUEmWF3|)WHVJd7D$m;Z@>noXj&f*~i zu39TsYgR5+oEg!$DMze4E3w2x>9oGm<90kEPwtog_?M&OAk-!Y|ML+3gL%YA0|&ne zb%S_9lKqCv_?OqN*~;1#3HS$g>-O7r^SQ6sY<1Z_zWcGg{mwUR>&ZJ-F3(u0wxF_$ zF&>MtKrR~W1Pn!jofrXeam5aw+gzA|5g{mwS=@X0Lu)h-?753CD6DeE0XhDB0ucuQ zGBcetRtJ=T2+xUK7%R78aCb9%XlvZeQPeDax~NWA6HB3U(_zSu;+CS+v63lWb;t{J z<P@u=#m`-mQxdW{xHlsKdtsqq7gj40gi9JsD5vu0@u^Ir7x|l3oV9Yv zu3lVFnz2vs?&Fp{-i|BJ5TN<|I4#<-|C)QGIf2Z5ncD}#QQa2{8r){9qG!=p%xMoj z5&pd&{6K9kP0pMyrt)>{pxzSQqn(4xz~{{Zh9-^AZT=}SFhZ&P)DMG4)Xs#htY7wa zM>~~sf}ueWbr34KU!1AXZf)5+Z+)K;k3`&`+a{or@^)**L@$j?sV=s=bk&?H;*+Js(=CtBLR>g2m+j` z&}dmw%aY}G%XT_|NYnA=bZcM6(l*js~n+qZ=Rjw%9SfKSFX&Rnfq2;9NUa*3zhiz zBj@A!C(p%~pSm8;K5{jlxUdwL7MCO^r+kQ(ec%66fqX}wbXg82Gxn75zBr1+e%S>n zVl#~;_j$%BJ~2-+=x|`}>Enys{j7JNGU3d}9g?6V3{6B1xJtgL){+z_VUkl+ycTJo4DH zk~!z2Et$5fiT=*&R_yQWL`w=L<1vTNdKxE?lluqz8fW%>mV%p$Jn2%oCN*y4iZ%7kFm&9D!f@nvqH47Qtv~8u6g{Sk=5+B=;bZD z#$}Be(;N=3R#h(E1aw6U<4dT`O~y}u^$Uf1KT^l!RoKmS=UqH;1NIPPvIk%g%kW|E z8U9F=;l1tQ0XixQR-NiNCBp#uk4sQdpWNO+XIC97j+&^Hm5j+OMV)?vRlnkuwf=?BPB=ZU=4g>mJN=QSBW)&PdT%qy}f-u4qAKu3QeOv7w4Csh{qrORNTD% zUDNp!U-+^2sfCMw5qd-dCKOUf}{YirD{h#p>>K`fo_P>r7J+a|)_I@fHS>vhiW~MuTEIC@MQoO_0W2xdX;dPl(KjzeDuGTXgiPs&FNhB zR<~L`e?Gd-{h1p0r+r9)8EU3=b5QwQW;qc29Q2qNKa8rBSuUSn-9C(0-&l*OdM&P; zsm3#p&3G06-g-wuv>TJP$#~>kBNiHyvANl@0$r<4ND%kqTR&J+diHm-|5oJ}f9qU{ zrVc;Jz0(!ZEd_OUZZ^&?FGf|91b^v>cFDe7{pHGsl--HrIJ>eG^Ns14;!T2t$E)vd z#=X^zSeToM3+ES=&$tGJhWJ!b+nS2Gxw`tpejKugPj%i=_|DF5EX>Zv{G28$wR-%{ zKmJ_}Kzrt`ug;N>W{U*hv}ZvHVU}0-$?j$duaal16rl{4SAFr|l_DcBZI<3gz>4aT z1`CcMj5-i(7}@-^1z$&V%)$?s-L~PCZl?jtC%%O};|$#cz`u;6KWHAf9Y%XC!xM)2 z^5x4nV*e@Cf!dy{a9SQ$`w$&Zq5l2#u<%J#merW zyn=Fj)u?eeY29Fm1I-CWDL@q{ZAQnBpV?~2yo_N`IDNkaqc3bj`v<{j%!K4(nKDubu21J z?Rbny@$y{k^YI+X*#9X|)(r|*xiZW)FSnlp+|G)43E94gds)gP+Y=Zo7;o4ng(A#b zFRGfaud<@Kv$5@a-uVLN=GJCxZEr}S-SL4=^4n9p>@^RB3mTZ%YeG= zGH`ZYGNP`0B*PEeap}r+jUx4EEL@EG+CqK!lROJ7V($#yDzVq|Ndt+LFF9FZ`i3?M&L;u8T6{ zgFsRjkqR@ja9+<34?dEs)$ZJQO=VTKGIVaZ#|g|5%&fHJQ}MiF1?vhw$B|!2K5iRT zP$?28sAM%xW;NU;*lV-r%9}{zjaUCL9)0Sov2b=-Ga)sozLYjKh>&8)wv4`JS=?*O z$!+Rc%>em2&Qr0tbT(Q$tMP+3e#f8L{M3^_5$9GOjYExQlgf+j3YBT)p&2ep zA_{T4-PC|S5!0G1eEW^xj2ky@#jpJIpNlg~SHx${WYy6pq#&|d&#Ew|<#}Z^CB^cv zKQ2~wrHrmcQ;NU8WYcZNUUNU@7gj{ig;>y}gfH2#GH@asbYe;=gs#^^`6Kv!YBr`~ zaf-uelkxp`oAL51tHM)@%Tjtj^F%#%nq%?i4GjWonrx_T{q#>>i+L%wZ{0YE3+Luy zsWuTC-IMsO-+3=KT3vq`j?2xbB}{5EOtTPmp#c#J78UeR{w8?-kB zsmMlY7Uf;l8Lt_0KOJo`g);nPMH1zbeaIhs{-@kt+^;~yudRjM0+sfZco;M?fd9lx z)Q2gj<*H@IN#F8dlEGDMEbb#i7{+(F?IqR((z!RJ@dpsuSvp zxJu(m&q#Skp;Sbz`xf9VKM-guBat6~Y)cuQw2uOODgDxxF(1G;k}eO)Jq#457@$Y< zGoMseP>j}32_;RZXp`M2M)DdBr@uH%e)cco@iCpx9=;f7yYERcV+-8A_j|5yZpGwr zUy4{i-nqFJd%L@;ukkoIsl{);bUQYhTz7@ylJy13GQSVzT^S}-s%QheZ5&-0eh^Ar z3-e3ywXgjP>Z{xqA(^Z6$md?K>7q1d=6yEe-s=6hck7-eQjKWt?ggLoJ+$&9*`*ou z`o^l-%bwaT${I5FNE7}uF)LZjm1ZdE+;Txv^Px|TEfX3a>zcrE7&*Vjs{WvfH8&r4 zyPITv-}ZK_uD<8Rr&1A{Tf7pV{rs1_V8CJNt@Yiwy}mBl_p-yG3qG;nGhMq9bt&O> z^$B(g^2JdfDwEynn{YC=PeOogBNMnW+MDvvXs(g7vsJziP$S3 z!Be~qog?DzR71eOaATJlWt}S-Gs4iZFa8Cr>@dz=csjN=UXJsZ9`T)x zp@!S<-B!n?BWSkm85{~J`(+N*>D3VSwqot>+i~^!Q*r+K<7#X)E@&VZ8tstIhXIw( zE1TT^3VqC8dsZ&HZMGovER6j18yVbezNhi^BsdP$c`}`vx9IVg?j&@$=FFd=RGBN& ztTgh~yIOGb!<3XKX4NRA6)CV|6U_E6iRRlTAzbNsz6+ETy;}k6n8vYRVVoK0AZ>fWR%Y?j z6MeCDu%o))c4wPkT=J@YyVLR>Yd#k}J+0ZSd8H&`F0b6-81rWTw=IE7NLA>Ra?>CXdw5U|@MDloTh>bB1s_?4JgPy-wzV^ywVKe0ZfBH_(+?x&42t4Op&Z?1DK?NAbt}@$#D1}NG{ve=JoqnWha&^9rK(I zE!1mV@g;?6q8hucgJ?+}dUa3&sHXl{lgxUX2`Ok8=!=Z7lSk)rnwy*R?FR3@_iijNEouDVfO0!_w>BlKBnL#lbXHBi z?`dK!WlZ?E8G-bj_Ezlga&y6c>~8MHzGPGDpe049BRPE-yUmuK9m^kXoak_!m!3_< zVNBwazFf)X^B)>V`id(R-xIh`eeTPW@Ac^Q!PSxM>%@Mm6`kfrQT9|&;Ww}BkNXuV z{H%0J8Pgw=4T(m}M=u5;gB9OOBYcC0;LyN;>}4TKYL^w)p<IoNO5<#y9bl5$R!m7qGRf*c16wS#6 z#pg9Oc?!82E(culHKfTLfudXrVqY?!i{^?I()#s_{(720^X+)g6csOd!Y^qHyBZG4 z7=xh?2WXJw^NMVtU>1#{jX#h)!9Y{m_~F$&hw*sD!(OvPb=uwS)mUGDH)>N-;Lbf7 z+pV>@cl*`&iO>A1_{M$A(5mLH0-(s#NpnlcM11o*zZQE3x8j*6e>yH+d@`masJ0|9 zxyp={Z8qK`2G}>nUAnFPrUYlxN*E`Z5k_>o`*&|^=G~6@<@3IKxH8>PoT5d*IF$#X z1Td@iUM!V?&HMeMiI|%ii~H;C_`w^yF{KW?w8%k&llbhj62zkEwYOWbE@gCAb+|Y? z9pCu;`MAH?lb}9U+G@OUV>|xn!h?98olV5V%%1HXwxHkI~9&1P4OVo(Cz|^xJ!E?8A4f5 zz?~=Km2wGAV1H1-D*@U^VYFM4gD-9ARZ?1} zN*Kl-0)CXo|L@Cp2!jwZIGx!80B=l@UHqo|G?XbbHkRT!ihGnEx!pboDNlp>Ry+-) z6;Fqpa4Cy;C64SL2}8R3^8VWFowzo+Ci#-rYn@0owRU&o{{6kUvv%7Drn$?!*>1WW zuqXFM^C(`uvlZ{GZTW4{2eGN7akQSXdg3XtQq0Q32f*m6^Cv&~g?RMw&q_hlgxjlZ z;qwHX0omE!jkWc=&gcG}cVcVvwo)hFd>b(t)#{4mo0Pi#n&Vlv%fDQ#LB{)y1l#2_ zev$$5m)G&EPL7Y|K4ZT|7fu|WU<((w9KeU69O}o?%CZLHNn@^PlELaci&U-sJ&lX+ z%FXzqP}9OQmDY)=rI?vHC?k6>_k5`0KdVr$xzEPZrv7X%H;)N6CWW}rYqJc9X-iE=9!hGS-;gsY! zWnglGKOo*g$r$eBzt7AFKe8M-&Nc_`9p)3)DAFwbkVh)LS=Zn%;Zz$MWI&BbdrIE4^q%^&Bc=M-%fYjr&*QZ~pK8 ztyrFC!LZo?T}s1#^je2fV9vUMv=4SAEZVVr;fj^Cqht2fpiEF_YWQ?uH-IVSc)Aitfzz?u z8ym5=vt^|i(^ltX)zv3Xi$+?a;GrhzKidj_mq~(;gJygRniIc1pk{(H9AP@z=t~%z zk`mR8s7}LUDPQ7DZRT9eFI|a^?U&<6uY5D6)!O$C#RCbmrRj@O=$IKx5GXb$c{lFg zv*OP6KHvWS@5INS{!%>k*q6LYlCR*7RGxfo2SEsF%DJ(=CV{+e`j*a|^Q{|n2>8)# z?#jI=vlP>+6J9e>gQ-wyMoZyE109w6=9L&*1n93^GS5Whom&U-)_Ys=^kXY=?AiNx zEH6w-IUSQC+>f?|ZdZb8cmFV+xLOmfQ^K2r*6La-Zm)IxTCrN4jJg!zYL!)OUd?5D z;wEBVC9;Kpy~6C7LNIV^#kuAASeUIvRd&82MpI-}1f><8@XDL71??T=t2AcqO}1kA zlr95;_$8&86&uz~{jcs0dghPU5-+v!oZ5$s(F;{Ahm>}%q2qCv@r9mFeM>7ndm zaMLcRf7$|fkn%}z+7c%F20f#MY-vkpF2-bq57oeVVAy_grC9zp#$vZ^_M?9AD&3A$ zJTqPbVLY;p*vCO-JRW6Sf#RKVX5v8_{&@P4wroE`uLOCFx=VUTm~aZ*fAnYNyzm=f zH1da-2cRrR_C@iVzGOQC1$9%Fzf3!HXBcsZI2d$fvD@%qmV`ZojuJ{GK5JX1%)@_) z?m+`SV!h8Ukd!HG1u8&+CC7|iKjeqqLu`>a<7cA$Ez%I*9%U}52TDKiPbl4GalxIz z)Y-r>jDY_DXAx)w`5+#X{!3$Bd{i_kK`2G^^_eOMpdwak z6Y+`j^KnU&imw?+_k#8!sS+^y5gqV&W1B7QwT4nSMR0` zo}=6`KCzn2>w9oWX*-rIky6a-{l3Q4USH-=GN!W=?H+c%z}6B!+1fyS&W3R9=(Y+l z6V06ipBd`3SnZYZ-rnRBP_}>e@g5yERaUnDv7c3Bz$q+Z?wELsT%w+s;2*1Q0JdhJ z0JGnivLg$pYSZFR!}sU%ZGhg| zj$U777w=gJ>$Le29sELuN_J~7s95e&nXE*!I8{x)(+EyjK;n-{gwm+2KF6d#{0Pa0 zpmOmXL!m`!h9~m-01!38g`zFmkdy4&M&n^@)5L~)Q>)NZ{9s%?X3sBK6pc<$+AW_% zS1FSK%I2Gz#LJ^Y=?je+MN|n@&e@si`0VFC5oea>RF+|y#FusAr^_>Z4(uS}*Iw!7 zk$rS%KJ1lYsQ&1{%(2DJ0i@@{!M%dcs>)L|sv=GPk`&RWTGEV&3$Hs-t1n84*p9lC zF7^)7uZiuG{grD`tzT1g_6@H|nURv(+H+@|Sv;>e+1a=%i(RCwbh2NH*$;b_vUk=Uw=fFG2yVz4AbDc&@*i#&#)GnHD3l&TPw}X*mR=~x*~pYu`E=WA>|g-t!|u*b>~3#K8P&|W~9PYJ}f66e&3D`6B9{g||cQ;;d-2|YaVGYaMDddH{0fny}?`=JOtmHlbI zq;CkL?!hWB&(i!;g3CULpM3V9BNugTpfE!m zrM1AP|12VWApbJ_{l-7)3rY}jotSVxAdn0!4-HXodma;%;>PJfxd=5S$OpfXI3kz{} zVOf*l`BG*kU0{VR>x~5l z%94D{wvdY$BZPhclu`CsQ9am_+~B&)zDS$0f_&@Nt+;sQqWq)?seAx3J{2|cMxjkO zJnF{m+%=Ux6`><3w15?3DRrdx-ciC)ytt2E@c_F~oVoJ!xG8Tb&D05bA`88bl{5-x zuoWIy-f8W7afmBYCp5u7>>nt7xsrMsl5UhZ?omGwJ>n4+$T;dea*G=Kl#%DK#$%+) z>g`05sZc&t@{WBV^~{--c=YPy@k_t-mH2D_=AVlH^tE4@WnfM60L{JPgIHKx(F|1$vpy5`*|Sm>Xedq1=Gqn#bjHJ;X2^#J zan!mWH(&dqYEcclJ{yzOdE-9RjCYcoD)h$}fa2t4$>d)OO3tZ3uZqgX5g(Ql*uiTNM%O|am z5lSkvR71s4=Q--c#1VU4pNU7F__>%>d7FEyvAX%TX2j~Suu>BWbz`4!KWOj7`p%uW za{2M7Pjh2~!jvJWd?^hl>xK?Y3x4DTw zI}rtv!%Wb_1%f|pX-VF=GpnnyN3GXIurM2!h*9}S&;B-6?4b=|#Xe#_^0Bwy^VWrl ziD~1n*C%6o{4i=$eJQqMF<;NE4&3eNv=oYoig>0L<$()wX6jX?ttqb*!p9QY<^ZHi zh>KVVS277p#i+9fSus2iD4cZQLZ}>SfMNxlS9aNoPc&$BVry&N11L&5g4ev!pgTFm zK2q@v0b5mjVEw!!lA~@-Z5>G`ZqFXwQTjMH!>~ zExf^@zB#H01C;6BojdX3i$9F-{?WJN)fc`Od#m?itk;ei34&S4Uf#&n=IT+GQZ+04 z(n2E^8%*SLp<-!vIu@tLVnq|+OOhpz&Q{{N<(c@I$1cVPFcdGRWn57hTM^l#OZx@}I5bY6v>o+VrL%1c z*MzqSqAZt+*(hxQ{*2AuzHzvli#PqA(uf9lq4->(=IhL4`1CQ`!Dt8UaRpF5iO-fa zR&!BI*$e);&wf1qmB0Ka~u`XSAbPv#U)O+MJd!kF&u5e|U|n$WOa!+Zu$-aibKdTb-i)pFzo!c3j#j#CU%K#q^!FN3pL-)_2#U*J350P7kAnuR22>=V=_Ep<_y`zQ z6&s$to=A|=X}Prm1@`5az9r9YT)X~hHGB@KiMGi`Z0~Ny%P)Q_?ykNU&pr1uQX;GJ zS9wmVaj5*KjwfSmiUGY?z0yEB?$r>sDfFVdzoRnOC0vet+L$sdES!s(8PPFO5qH@C zn!*Fc(Ety!fF78STwc|e2xotCr_+zyciTRc_T*y=vDYH}BpQ|RSX!*d#x}3*s5&X4 zQu-HXrsAvyrw#`okGtaObZqV(#q!F0%+0VT{3K=?)!5nVNO2Wk#>77n#kLA=Y~X(2 z`I%~*J-evcKdZ$mrufR13O=FCy~nfD>}?j$eG7;Lx|HAjc0V>Zw|sC!m94tr5C*eu z)yZ_td5ozcaSHkPs2AH?o3XdO=~Zff$xdOcG;@Q-_@tC@2{y{fn+dO6D<8ixpd!T| zw}e6nvtN!+vkrhb;;Z@SC}4_MSU%^$G=;4(3;?fk&?*||AfDT+?CB<3{Ftm{)#06r zZ4ns)avg8R#L&B&XaKA15_{oYW_Z$bJRs`VzE4iofz-Krl-zKDpROq9O3D7FP z_7Q!OSLu>}Os17@V`41M)+%vjzG2zF%K@}rDp?8T^Wi`d2W?PK#|GSbcC`Q?rJVq!EW6Y_xGj$DB<45Z(?EonmQ`pm5RnmfH>P2>xV)spNwNUxTm8Z*)Ff5}V0_vFY7QAc=fg0ti zAKvWx5VK?{p(txsK2=V(d`*geE3d8wjZso6uU=k>KmMy5&@?@-qZpwgiO!%b6ey_-r99;VI zp+H_E8pN&Rees7?XXJqq_>rN=UHF3;g84My8_1Oh)R^b8YV0zf6^8uvN-^z`e9G}v z0G`r}0&dz3iukzdy;_}B8&Y}Y&)d$AJ@sfj`MBB-N`H~m+3JtCq_%^%6nF&?o_Xby zzqC8}6nxm*Bt9BtczewiSu%?{g&E0~hZt{WC$0pGBDx7EfX+TO*;AuaXEq&Rl5*Dt~4#vU?a%ihXElNUURb-Rf_D)T+Gfc#N6Cs%*`)J8C;3^dHFSF zTy!6l)@WUDN(FuX698^Zgb$U0F4L!(G}?|id!&+KLj8vNGMt3>Zp8< znb|R~zYx;!%^g>6I>wZgIBEFN5M7Pgx`azFmS&dJk!6>n`p!Gwj_>^8Zz$?feBp~< zi)-heG;QQBQKY2p=*ln;HCxPM{V6^IVWwUeEeNFXSX^50{x)V2T%bNSUW;id#H?D9 zl&u!nO5v$Xh zc(8KMKH+I8=}dB<89s7}`^M^iwEL>lX(_c*#F?mojg?zgKX@hRPev;*@a=7_#<&Iz zCQ_0$Qsjk8la1ckw3KfRaN+@{eBlfy-3G?lf9W3n>QC0v6rT1^E8%I`QcdxkS>k!9<$7v^btd!f~4+3$N zZ&@Gwmb9l$5kOG-g1C&UN5+>zB>yw5_hN3M<+o{k=%L?g#ooa|w0HO7-tD___tsrc zZjr%@^RxbRGm|w|Qh76la)$ho^fi7raaHrU!f4}yacWyeIQk4@4()-QN2q14) zR?hqDslCpjHdiZTA3pmJW#%2~aM86MFdnH)B(>R#*MLyyk69pOK zuq{Q}jO>&j(uRCWL%yO(Wz7(tyhW~Ly{l{}$HuL^ay!KkN6G7l0kN!SRT}=-6mMDC zmiN@tkHxc3Kf<6Y{E8(!)+>}Z3pj53`pd184weyJavQJncki?`8HqG7jro5bl;PbC z7q+aBtYBL|0Mv*aP9>@lB4B7Fs)y<7ff^e7hQ1fw&fA*hu12eQJ7#8QCE?zX^6-Xc zzi-HYQwmSt4VP8X%H$O_mRfY0cND)Bo0?fa`OG)sXgn3>gz{N7mJcHGgBw$sh6c_o z730-lUgJ}cm=^stb=aC$t;Zy6X6N$4{rQEds5NR)uZ5Jh8PUAphCEfJ)2NG}DXICf zw=@T`6oB|1azll@y?Rl++JMQdn&Z!FS)Yhi2SOpE<4j0+UA%NnlgE>I^~HY_KYa0f z@#Z@(#p>$a`1o^Qh%bKmmt($qmIJ*fR2@662H=T?X353*)6m4#^ z=y$r_-#WX%m+$!G^LR|x*n21X3pR34P!oAOhy9dJE4S-LvwU^#?%_nNZFOR{b|PNQ zXaGa8%oSTVO`$&FWD~r8jiF6hH@k`>hXq1i|tzOgH9S%84j?j`4yWi}_)-Efl z%=*bs{87SOl|JzvboNm%EzNp0n6KX*OBis-t%|HuTb-JalB!uW!g5<}?!7yEF}G5c zJnH#6wO+faFpvxPA^YCg$9&}d!~45i3g7btg?7kWh^`djV=hG(ZUz7ZB-@GjhL@9BO&lqvK_84A~_r_-gO`{>`# zw19gUjv2%uT|onU%W%qzeF*6b8Q+YhNpAi*h}{D|WFVO-*;#GWHKDJ?#@bfgy>rjE<7{ki zd%M)i${ESHTKEf|lBMA3siS$1HK5{qQYou?k^XY>T@&5~_2G*%GjT<-i!WWS9psxx z6t|S8EXJfA@S7$IB=8|DDWEemi}CI6e#;*)+TMKEGHmoyyB`GLZb~*?y!MD>sl;FJ zI69Jt;}hIt%VGt}lg3+*CPO?-;#k1&Dx*ErS7awf?t?89TULgxFjHz|4N9NFy+>F+ zlPrKpF+1k%EvbusBD?FTfI$(+2fTwV9m3$KkDQ?MN!C_)5>t6LVrm@~7bCb)SNqLaRF(w6CG;_{_MH9+_j-bTO(<|*N z>j^hTXl#J5EXJfVnPaOJVDl>Z=6v*%btFBB6>aD!^7X3xl*)a?ozy8mOP+r)Fp)xc zsQaox&e+gCpM2taeEiwRgiCZOUs|^R$={h3?T>sQ3`~xqA0WxLag!#=X>aKW|2!O^ zGgk_MRl0u+cq2EJ<`bD*+f~<-&rWhGUqLcl_fnmHLc-w@`Ku#NtgAk5DDBfXKs1tyDOiI~I%+5d^TAe+%veZm`NrwVsB~MdRi$frNw) zqV*++jy3z5S6L>aGDW;wQoQa(`{2FU-}#>CekTr(H{<@TZ^rt)m!sc1upmESmMkWC z-<&Xc##(Tf&cKH`P>K-{v^wwutW@wS$-+tA%6rAf|*xWbPQ&vtUU8^0R#vG3a^?P z^L$zqfuFpARpysFdfM&GCgTQ6SvinESY_hKpGg+!E-TPKl=3RO?lqOH6|aquPhZ|YJj%PrRzt%mRtxjDZ(lgs}NG}CH|Bj z41gIflJa5toi4eB1-ixW!KB1)4FuD9Pc8gme==pHuHjK3dR-S5UREBJhn~h+;2j7t zKvvxLv3&%D1AHZ%3M+tYER1C-@w%QFL{6nRo!a!JB_IBM7!Xo~W&;>81>mo`Vn9rK z8TM%}qRZ`dXh!3h{W9WK`Va7z&tbd_ODNm14Q~65JPQMq&x8_p31yxM!g?z4sd@qm zfB(wv_#<%IJ*3et;ui*m47pXB(eM(mj;2Y!p?lOk2=X2LorZ#+`hA!gTJ-QUVgF$?goXq{Jv`pmcq{3Aulwc`(qiq4X zRl;fdGhdXvRDxY`HmRug9q!weSp1Rkz5B>{+$hB&cT%6$7AA^&~O+G zzRk@}tG?%BcJ^AtBsxYxkBiJ|cv-eXjgza<>~Ni)2SqQkAwyjn%AxVkGD997XQ)5P zT)v=3*(NkuVBF|R5w{X-d6-YOR%|`^Z)c@H7 z!YWsvXe(+=<9s>iJvHK6enpSrD>^j#;}fndYbtD42`{TtJget<`SF=$4nM4@f~6E3 zZfOSCRR>yDhmkO>NXQGf^Bd`)3Dh8_tn`02m4a> z+`Mo@jkcvSwWHH|I}Ub#5C{9eFJkGVU@R5t*b6rebB^(cC0YUam=Y=4YI7hl&|7QY?{c(E+G?2`>8JA zDZEA@=B)tuD{S3R+2A9*qYZfSV4>AM0rxL@lB}0F*8<4Dpn|qkn~uEXlvZPdqH`8Ar7bZKBan z`=el+i;9Z@2XUbhzEYR8n#5s4Jf|7)d%%psC=K9C8XibLh*Lc6E?qzw6DVSq&H~C9 z12YniJ>7}1G{C6)L30|CCuvF;b?d1_rD8h_O9RicEoM9bSIG~x?xa!n(qz-mw8+Al z#*F(U{U4>6^9*OtEThe6vOf_u$(m;$zwSN1cegfUdvz=B-@Gk__?8r-jz5^f2Ucp+ z)wq0i*>{EW+MdtWa<@0Dl=M&iczT;3H;^2g8jHHj<8u@7mB-G;H=e!{*QJ;Z6c8Eu zYzd{k*KWt^{o4vp`AWS4C`!nW(rtP30F-GC_O_!jzYsN**UA?Yb;*Vj5(B^KCyDH| zewD9s7?0v}fXgexr2;9KY1DmqoN*xU#6{W5a1Pb+8E4x?CwY$MtBA!j=`_&3*1l30 zs>Ru}S2Z4 zi1$-d3(7%pe6vLImuJ~dJ_rcsEwbN47wHi1PBxMhKSi*e0Q%Bsjb=UyFK`Uw{?7wO zTH1#TRKmkL>BE4}B%_RsPw>iH6+Eul@W~IOx%azb_`c0_g$h=ssbO|{?`j5iH#*Jr zIP7wL)lbH`bN^bLyZ9GkcJV7x>Nq&ik9y;BEUZXbR)e3JecZFb|4BeOKFAD$X$WP9 z&lZFg#q30i@bS^j=pOtq4z|7>TN~er-L03TAYD^=cGb~d(6g;}t_IfUi*+0FdoSAU zyRvts__042Wl%)8fmICngOFMD{iOX2|f|MxpqIc zH{X^ZJBj)E+2|eYYEr?+Lk{xAnbNXv3)&GXUUgKFxay|tSd^1jdF4HX3^t}qPWBu% z=I3L6Q3DP5P|7S2qzJ3SS;+PW&@`B60Xiq(Y6WQz8lFm7+Fy*m6z6B+r-PI*!GDIZs)!9Nu|x~&7H=7>bg4+`;uaKxrJmuyzvw5g|U@vkF4^N_G zC@aBm3vl0bf^=sfes&E3_XFPPWsVI3>~=ump*>N^t)!!yXC`gu!G1lb0bpEPaNL z3jEAjZoF&?d*Lko6WcHha9VpQRN54S7f(mwM}h9@g*~7A5C8ZFyB`K{>0j|3E-<}i zgd(hT^RU6s_TuR{g(<)oh9^c{``RM;fjykuuzn43*-atBQpJ=9r?3IwG=hY%}I#WsOKdJ}Lzf4y+LzJuX z59j$1%GG*q*I>bn@e3u_={1T+bnED__2kj zg2$z(O!4X)B|w$IYk>Z`I$2c}CB>wAMvdY?i0%m2PPAGZ(e2-jgXWtOLZ6J~b6<;z ziW;sPe?)WtMK$t%?Cx-<sr_~w?EFD&ue}w=T?yq1`&YT*>s;(?eoN&<@w(`lS#$SZblUr|xbSS8z3>Go z_Gg1DW_Vri#_fj63Z;ax1g(`}0kreMkBzleUyH(LkhgZ${WYb7W>fBtI>d4%e zxHNwz>a}H2njc*}JW@wGR{1q(@MSp!0>W=h%9(zhFGARPDNoa)Vh#-}xjZET+0&p- zJvCjIz?8yg`bb0PL+BI&N^!h;scikq%6yhn&%S2T6Pj7iYNp>3?c2MDu_tYl18Y4^ z8qO|FMMHzl>Nc+pm9cuIzb&u4FJGLGNeM9oTjxme#f!#l&1IsD%}y^4+O!qQh`>VN zu~IAb7M>DQMYwQwPHk&EItSh8pG?O5LM<9ej zjZ&gH6gefs6Tz~#dEfVN`rw3UW%I;DUHn%&@U?mzloM}K36`+ego1&lRDfAEM(IUa zW^!{}#66OLX5TQ1CX*3v3}KL>DO>sU|Ck1Wg_vJf`-i{sr~PEbxN=-(@)AWwhVcd|1!cMtO55;R zBo$D|A&mN!^gEg3r!oc)Xsl)aj!oMvAXA^lJ3wMH1aP=K0)7cid@5%Vbr?R3S3-s* zUc$?;2cHjz_lJ&BS7dwIa{wQN`Js^wr*O+W=pKbKY(T>U0P9_L{S!A8;dGaYOTML_ zJdrVle6(Nj0P16had~VbX2zR7#LC2zm0c87CX{panOK_D1fKo9Qd;Ngsf66TzbS=u zKYGn>>>ccDj2V}bJFUKq+{ykyekimRRu1Khdf_uoD9B7yrEoAgN78t4L17$JTH$)O z<(YVFWhO2!&BQ{zBAF(oZMGVX%2-q-7wVD$%k@T_Z8YNCTsHSxFdf+No>oymI}C*q1!p)Yvd~l81$}pQn=M_U;u( zWWbC}uIN${uUMi)A9E;<$prmEJkWT|YBO@1Z6wH*qZ3v}*-OmV&3Ge8t+C3QwUA># z(XZ?-DpJbm_{3b=!9P)zJmO8BR6!|qTy;MR(x(P~;{4LUs!aoA3l zAIHb9M!)@@(k;2+He0*0KM@OOeo~XTS=ERd8kqx&!RQ{?^;k?{381TdzwHAIJGCUyHffr=|4%P)gL>qJ24}|HjnBO7tbX zX6HW{)yA_DW>N!Z)x^>HZESg8Z#2u z3+&;oG`zxzQZ`-XK4d<9DZCsE=yN?AuSd8Fj6G3Or1VEP3YhE2gNuxw?4zXo zesC}wR=IpkT?%>7f#6j_z9!@=Q7CfO$t0NQ{Cx?SMy=ul zFB?q`P;dd+NzB%!G+~&C%`HC6!5-rf?=($n9ZEP>V`g&7DlP)5-O*D@Vs&Op6dY=j zaH#f^LV_}yHu2Ge0}i#zmK5$OmHUy)suPt5xuo$csu}=RcY1N_&R*4#VGf{HfW^1TiF|$L?J4k0^x%Wlei>Spxrk|-U6i+Qr91u0DgZLyGG6JMP`b*eR~T&rSN6>k zpC0LQ7W&Doh?y3d5>JK?{U2sV)1^CY<}o&9>hic9-~;a!tRD<>W@kZ|o)Q;DI8r6! zSP^z=$4WojQHDG5Xmwxx^eF05r0Vn!HG`Szcr4A$#=^WNzTQq%iK_atZwF9^pQ%;7 zt?cH#`_bIli7_ca?W3-@SX_JLF)2T_Xh{zE4sa$&(tR3>RdH0Nsc@z1^3D2dOW>Si)YT&T$ko@Qf@^>0k+&KWFmdFQmj%R`*sZ|qD|)qQxYFNoo6b$ z)CpniH%6{`I|K^4fqNf+Q6o>ZZeoMH6%fIxS zUAREVhX!!(LkGVqRwJWh!1q)%hhW!8g?PnLP6QkWlqM#p)sSW+Xi$XrZH|uGYH;^r zd-FY4aJBNdI?0oo$^4Wm{YlN(*(Wrw%(z8iO_ey~@R`|?&>V+q;LepABw5q0{fd{A(vdMU<^#E)Ystm>#^6ZyGeK0CRywG%gQ+=&~v z?@Lj<8}GjJij?L`%q}lVDLhiXe2I?BxHVf+cx13>~5(;loNPs~W7}`<0-yPy|+`p^m6N%{Skx8dy|`*=kp6@cAU0f=*WaL(m%0xd7wsyUJB6{t3?8-`f_A}SwV;{Q^ zk6%Bl0dp_LeT#&IY;`7jJqb?XW19(|Smp{Lgd1h!v@xGur7ZrS2CE;UjeU)N`f!{C zwaUQjL1a!zBZ8ehbQR%5VP>B;dy|2D<(BQ$b*g&6POX>zidFn4QoQV+&f)aGG z(IFl^{hPRkP}rRnwAf|9KHPiE51tsMam5cjWCSXUedV+-Pkylb2U<#h8QQD;(x_!9 z9{GWVJTf8uEu*GK;Z4#Zl=i|iaTZQ{CEmm>aA3yE^593n`|Wlro=aEMy)fC7`3#|O zoc1&%{Y8+3^Knb#!v2OP_1ze+PWVd5g}J3zRR69^xt&&DVxVUqFpCW{tUyXZzIc8q zu3bFmFP5$E?5fN6{ki@1%}puMwK(DQR$b|Zi(7Q)Wn@augU zN(Xs-+yRA*XSw$n|159Hz$}lR;6($|ldWpt;4m>)!uk3~$r0r8r3T~0n8txt%+JnA zj->LGIVL|z3vJ13(KEr8t-Ohdv0?aSI+wskEkw#!GkPtBNp3M#Oo?B0@r%#dBTsYO z1Scz%l%uLKLoL5`8tbFDR-Y1=4 zE!$JltunHBbhsrQc2yi{s)==^qsUKn&fa3ioU-0>pD#QSE}`*?E&hSxogXb>p$YKT z43*!j^1QjEGs1gKQpU#l2u!)v0;*)JpmV5h*FGl-?UP*h>qIxgVL#wxk z=#MwhjK8dwGr_5dU(h9yzlS0%3f#{@4mI!#hn{X{b^`2a%J?YC*j-3tv4U5`W5)rjt$aGOFNLYyI*!HVC!V%DG?Y|bs zz1=u$HzZ(>VsmvXn!6`zh>Jd`QEx0s*d0XY;Jy^lh7_!3G&gR?yDxt;-g@niVtw_k z=yum6fX>A7nKSX^Qy+`P+2!c&93&4BW@?E3g*kd6NSI|$*LWQ`rAY+JSPEZW>v^`U z@^D~{4``t5(@9#LuI%|mpiZ~zZ5OR}JDL)R2hDbD@2_p!ZIrX zg;EBO#1y|u7ZuQ9&JEfE%B%#4;*>m1g&bz8#!zNeZ`)f<@iptOB4KuZWjb2jF>iS| zR9q%foaQ~ZSc}0qLWAuoZSk}&d=fF@S*yjK zakjaDcT(teSrw3RAAr$F1^UtpCuJG$ zHRDS6tr%-|eC;CVD){IPTba0_fHua#R^EiURQm5qp=fbCfE+c&nXgx)-8+m)33tvk zaF04~V#cMYJ@WK(G06gs`Y2ytJ5-+}Z@xfDS-_ERTj+Lbi}bmI|B#A}c(hWsFrgG@ zzepbVjDg3OESTQ}72_3X5*P7lCb{4EWb82d2R$F4o4x#ay!+N$(P=Y&64i-iNqi|e z=xgj%MGi@7ipG-@zBtF0kc?*_6FVhOr9iUS!hQKYwg3uGx6emGQ1&Ei*t?3%kOa>* zHN=x7SE8|aPVM4EWk$i|PG=NYDZpIyg<^(XPrfcUSwR-&!oX~8GQRY=PsH<|eJbV_ z7o@=QCYCpm$rCTENN)F~xbtzM)j1rp5xNraYu~rFp4Mu<)5~%@(QoFQNZs)2IZEM9FK`-fJaV89W&9%)#-ts z`Il`%0bnMP3;?cUVOKndE3I5?C*dPb9X`4uNCi0gpR`jB{51ABpJ`17xLK>WxQZ$N4f`3#VhqA(;&<2;zb-E8hkX{B=`cI zI521|G&H;AfI0RB(w2%aQ5fK5)c8>X5B9!kB^af0k{Rp7)#!DIdm&~TkEs(qulNhb z>y7WsoW^FA7DcKdh4Yd+(4tsR7%OpvyEX(vDiWwv{AeJqtHL+^%TAm#-V%qfoMZ#H zALXQB78PHano*c$awlwXKQHC|nv~buQh@(3+B>^aV)mrsor#?dRyk*+tSzWxs)I^M zA4&*yrHu7@TbdNyl0tnXzOH(r!N()6Kl#bH`ot&W+N00K{K_SNw)ePq9KB{sf@4Z} zda>U+i2Z|R>}>DE=Eg>Btjnye$HppKHg3h8TX*8tjhpe_+wa)C`}W&1@5Ea_dOco$ z?X`IQ)ikfX{KI(p#TVlTFa982{J~4{gCD#Y-}~MVWL}8xeD{U;-V5K4?|tXH@xu4g z|GVG$o)qI&eEid2H1A4ImruAJixpxD~!ME8&ZC3B#jeNq^Q z?8-%XvVwaO3$ruPkPvOQJ66Kqys;a*?fp1X_|n2mTs%{a)jRinm~KK7u1zW0hf<4C zpgGCW?;c25?Mk_=Xd-m9MW?k}~YqsOkf0ij@$B@3cAa^RdSciQQ|MV-u z(_!4$?Qbi;=@wdp?W0d;Y+GS6EW^_c;LaKy4I737Iq04CD}i)npB+NtzfDHUDX>maI!E}^~3!V<@qXcjbJYvUHKaStcs{>_3Xvk1|04 zc#fI^hBmy#&!bk1-+d(}dOPZW9S!~(6@+GTntRZvrChNhOkd=caV`e5rQFexptma- z!g+#e;p(Ul`&`6fPm0xET)g^dTz=$fPqu5*(_TSt?QTovu+>ZX)1QSS$wT3U)}%45 z`caumK2Wd7CHT+kG4UN%#?enHFL?utm)}rg%Pws!77MWK>uDf(8zKQ!13`yf+wn z#7mbr`z$!JElovZZY6jt#u%M%!nieptzx{1obn!RRw{V|30WwY6~(!FVIh9;>pu}I z7v_C+C2y2n1mT55#_GJ@vg30V`#WuolT%*Ixxcy>2m5LV$WSFh_xA)vPY%LybsGl< zy;TGSw}<983Osdjks+)Cr0*gxQqsCCg!2)v3srb3paRjOs<>+ELqM$8e>)EDui^mHp&oI+m9fr9@A~ zPk-feapA%lP3Dv}MzO$7`^x1d(#gjpK!cq6A-*w+2CBDgmxR%dlUHeG7eCx>5}I&9 z(uWK8ZvAM`A>F{_H<0;)fj;UG{i>3vXs$|`oao8GzKv-?y3LPmr6TFDv-ohWo;r^SkwYC}?TdPuRQP3~=u-tfD zQYk7DyyF(A-5x7`Qi@xxXzq2QBjMVUQ0AJSPD``U9@l>1*Yhd`0{T!Lomo8!AS=5_ zd~RRBkB*%y#3~=J4MitH)!RH2X5j(tio+A5e?8$xw(?xTl@(6`_FU;;#gg5;W3jj} zAAk1G|Hb&ySH9+b%|kxI7g^jP^iX0}6{r^mAM;#(-lwE`Vt=czmLcBMq}oF}Zr*Rl zLGL6ch2zVgT2^i+@y?xgZ0t$^%a2vp3rqEQ?rDV`cVn+Jrb&T#B_$eKc`HVquf4q$ z_qSN>hA`zPJbfQH$XPqD)mW9CsZPYjg-Sg2#C0pN&1O4ZdHs$CQFt*P_3^`a_R*RI z&3NqW9mVQKGj_Wt!KYYJR1dptPhj9WhwC=hH{;?X&q@$nR-ceCkx*nnq3rOm({1X> z3PAD&qP;S#vU!jZZ~Q54>O+3r=6>vR4b;YZ?CCgUZ>n=i>P< z{-l*=`kQNjAIk4On(Y-D%n{iMhYkj2@=C~f)U&9wKS5qS5a4e=3q$IAn9jclg2Ky8KBa60vR5l4lN2dQtH06#D8Xs@I}9xIE0GNI zCsF2vtgsXRf?43RydQ;;da>-T2t~*|@g6ASL!n%xlujRx95dT*x>+1HZf5ii4fKxXnIW$wm$pAFJbg6dz}OI&FOgdC%zsRuU(HyO^TLK9Jco2tyjJmJ8SnOr=TMzyvP{(Io-)( z2RHQb552Y)PyP5hO!Y?;V!$#2J~%FsQn8^On5W_e>_DcmMQ*><)&#yChqdLH{nRhU z@BZU&YSR9)aiP*Az|YbSt9T9HkOd_cxyhAkD7W%r?BvayR}lpwVSI8W&PZvRt5#xb zTe4`cDPExr9%`uFlkBKvnb>Yoz_7kS*N7`upLRJ>>^TI@N^NfKW6KCz#&UZEp9$v7 z5B9q1>`(pDm*N{=|7^@G&q*iXiZm9{iUkg#KkE0rO5N@3#_r~J+`qXQ(=%1^r5nHZ z%{St=f9K`cKahf4pNd|OymLH-$4q|M^3WFldR*;#v?DjleYUqOL)Hb5TL#bt(vK&8 z`Z1qK#cDI|+%l1Ph94P`8=~DGqztQFaa%=|5sy8dh>ya_RtgS6 za&yY0CKLNx??^v*Km8?M#_om1YjN@7`M9t$AOGRs{I}xj#j{>uAv~36Xs5*_4)RQE z;U+vSWMDslP`pN>8ZKY|WMTVo;3H9n@7{PVg{FW8T>^v*$tk!}R0z>jXBxOQgCFZf ztNE51BCDK_I@2M= z@`nDsea%*-09@4Y<_yfu;$|92hHByi(9(khT6Q%;| zDcVF&-aE`+He%CBWC3>K6So|Y%BM|%SE`W8P1)gN;x0RdU(+ielwtTwhvHS$m%sSM z_^W^I-|@#TM(e*Eg-wURY8Olu4%6a%2_p?NdbVkZzs}<*rlp+Txvv3gvlpHIVf@6i zXJUDFEbeX|#*MoNo|SUb#I$gJ?o&%q)1+#rc@nLzcq>8CmB3^5e7-Rqt2>=|=jK+l z_z;Wm!y^VfuH@qt6|-MHuE9Z>`s7hOe&t-u&CkY*uic1^)^SwEx(;t^fYeMygC1YO z8;>OovV94#z63E?x7Sxi6Q7Ol#)gzswvJqT`m+kBlzj<%7?EF<;(!8s5uwE^18m{I zB^EKsgZHp51(e8^X&>xq!nh-adpmZvHl(cY$JVZdzXm3nJOcx-6wj=jjn91fD^i9# zR;~vK#P9OTI_fp+qp%YNI;^!(d%5GUJO z#x2v9VM8ztWMX)K`1?U|9G|#l!iS~Hu(Ev)X{T)#)`tK+Wf*gazi94Rmb8M@k5kH{ zt&}(YkZ%cc6CYWlCo~Xaz&l(e9=qR@HP8v6{fyWo}1u}7<}fe3a{sF9gVxK{oPo-xfWB>57-9r=DTb0>%Z}Xcl9IFx+l6Z{@iRnZ6g`*DbZda^5pxi_HwVL&@B7u0}JH+1ESJ9SGL*^)xW(n~Z6 zf#05Rz<1X&iLxH@OMbL0`zp;y9`aRbozKJsSN~Bzld38y!jrS!%Z)D3Y2H>Ci&=!9 z1{R&FwF`0W`jvR>%7yp~fBxq*I#KV8gQB3!Pqa_EZ6s+7{ZNob!SG4tO#Os#`RnN| zh=Vwp|6lma|E8ZG7JM|y@O!U{3t4b+Mx{rALN!Q6m?C+*OB1fnGBI(aj-e`?Y{^TC z!`&C6e^iU5m9K$COr=7bu3$Epk4a%3KVb{Rp3RiG$o?N@nNs9V zltymXjtn|u+J{iG>r{hAA<&;IRzMVHk#xAWD!s$r*@J`VcDvD)oh!}`q?BxLH>LEl zsa|oU4DC0!^_0RUp?Z9**=MIGrAT; z(c)$Pkplhwc?lB2px1W*BNUVjzQA#4b{S<-u~x?;58Oh`lYK%cUly8*Ba3G49B9i1 z4IeCsC9JS>5_O7My98f-W;XuRpZqiNC;t3jjFT>(|77p7@FeRD@YWsI{sf5rf-e0F z$eX-a9g*-5|4)v(QIpWxJvfXvZteTS7@vD`F)poCV&i~4y!+AWAF0hKx_I*W=a-{V z8ISFQBTX=bXZ%pJ{fMs8FU(9uZ(=ImcylAR_Yl~$O=w{67<Q<9K3HL^0Ms1=aB~|qaOcBBl$KZF&BPnMUK<1;@cWms+0EBX4PUC{}=SDf~t zJUo!dJL@dnSo25VwAnTaAA#tVdnl+G=nIA0kp9%I+m*eBaUE`Fw$TzYo~5k-p5O;P z3^0Xn@q7?6ZN>+N@kD3gfCOlU^3!;R;U&f#@gZf2o&|d`I z-S^_H|HuC&{{DBq={WO}4OeHTW2sS%%NNeab5C52FMjUHn669l(&SNHKK7;ba_bCt zm2d9sS}EF7|3bcY4>aLbC+4iewEFPk<;P>`%tgOVIuxy)&Q9EU>xZ#>e=Vv~kUMlEfkI{2b!)~RH$=-}WPZoDiv6o)?F7tMPu&Udt@eb-)3DS;HVk>C)qpY#m0Z#l0r|+`(!-4^Rv>%NPV~cVA!ll^n zcH&@L%J4AayaG=6Uhbr($J!(PiQ5q1?7)1xx~%tr2-(E|uY# zYQ^%aD*I3ULB|(9^R^t6F(xV%)n8sS+CJEpGQ8@=hf2K`o2tLx z|J~Q)fBfwi;-HP}rDUSB#~6Y#480mt`~BEKhO-A+W8k=zdREL!g;;zjplo-*XgN#& zfuDJ;`0$sf{>H~RbSdS)X&8^h1C5=~;DdS^qsuWI$aIYJg`fC^AAFpctcyPB6W)F? zp~=Hi%r4Hy#@fv|?(dwU2`Jr*OOM6XYggj(x#gIf?Zxf;tFDJq?*Yh#o4srwdJjQp ziJ#ZYr5*vejVpoaF7p1ZfAlRsKP>o2_6%>V-g5nzaWH_6ONDblfu344i1IaGsv_1g zLhV--xgA^UH=^G7q5@IYq@bu%<{+3el}yR9zZ$4;^12_T8aGO?2YlOUCplm+r^z@< zEq^+XYftn*C;P~l#T&O8J_-$o%etBk9<-VggnPbA_TJq)ar@@Yc=zo$=w+=iM9e?t8c6#;v>Y-i>>4=gxX;Z0yFy=1y$w>?wUGcJ?$=mGZ>?DwN$rDQ>(P zVvmhC^otg+5__dsim9HynL}~&NtimUMn(i4(v$co;wY{tzr~@ zfyE(0Rk>4xE6291>+FC$l^b@);UB!z8Fi#OQ7|j8xS?%+X2Jthzkd?-+ITF?%~+^( z4?EH9oJ1cUXy7@sFdJv*Yq8(%g{DQ(-j|Ze%^Y>jz4Es|>1x1JzMmxVZ8O;jCz$FkomG ztcq~62Om`7un+A61?i~IYK7{Br-V1RX0)1`Aqcz=S?Tx?%9ZQaJ&*$yaKh>R`qYb@ z@Y|Wn!&XQ1ufqm^*ky7MPdo9P#_P^dv}>X{xbR}59ecAGfjE8#%UCkX$VfdJhNySh+o#JJe43ZCFIko1-Phl@Dg{# zI4n)#$@Y=)<)b*J1A^qkm~-6TkMqqR#@WfPCW=i@bXhoPsn6Y$jDGLV-8kwT$8{+J zEUqB`#wVwuI@5@;+Dyzg8u8|jUXS@X$&V_ZaIVB;y(;BO{ZW%#Div=aay2!eOs`>8tNRtl#fp77_G zmpG3Y%hZ@UQtGA`E=0d}IbQkEYf>ijFqOBT!7IfjufY?1rUA^VRr^5zliTEPUiZa{ zFY;2cq#Rc2OEE6l(~&~#>%*uI+JKcf(q;+WM@XsX*u-pHzxIS>7hA=gyZFvzgfGTX z9>z0&;Z8xcd6e_X$>aF?&paPbK7CDNG@mXepKQyt5A#H6S?J;Bq;b*P>u7v9IB*`* zT%E^&C4B2!2XSwGNAdElr^xVP6f%?p*vJds&N4ah>g@RJpv*wYJD`02(?Z5~sRigF z0y*5b1wO;V;uhsfpP?zT%2tuUih80%de9#-?y;2ON!7vhWG%ISn%D?)*`qzrwKOFc-l;PX&y{^W^tWX4abxWRNw%aO= z$DI~oFj5(4!fME@E*$om{my8Xeol>(y)QY#an*Z(R8%$Gv>Ul`0Eu_hZun{BeUXnj5I-n$<+CCuJ@>#ca>wKt^jz7y|B*}d`Jt+;vnUfjF4 z9{1NarR;9S+WJ;(Z0*F(?!IqUXtmmIJj$xT4;4AcmIFM43JsY!bV3irZau*uTEZ-cFN%g1oFk@ii8(y>l~!(MNbMZ zpH4ozT#LoVL^Qh+czYcwk9oxoS^KccA|Qw{dko` zc)*7;{KzB8e&Lji8ptGM0fBB#FpbnUJSA%x=LFS zXC&?bNK&{{ttuM>P{zyhyFr1lkWufd559|8-U9}MpgGX({HS8)-AG8l82A^xSkgfg9f6Gm}76clcfPbEt9 z{*ftbl_;gCCulSYu562^!Q(_>e;Ne0~V80$2nzW#DtYn;S&4feYS z?O1Ic##-+<_S9FJox@nYzZN%c-;L&8Q++<~wLR>jXsO>F_u{D4ia9CA=T}x@VSZ6Y zcqH>X9K>ZsShC=-yB}Nk@5B$j|6=U7n(nXDk`3Hu1C;}k<&O)wi;Sn8_|?AbKF=Y4 zBEyG4pDof^{z@ZbpDmu{o?`PXi)%#^Q6%vf)AJW&FXrPdjb;5_d9wi6lA2Ner_~t` z{H~m%EpmH%TvK@~vt8%R;2cRFlb~b+D7rz`UXBWkb9D^*tj6CsMHxQ4g6X;?k zx7}p_^GR^?#iVFvKkdoUc)b4lO^uBlR@XQo#j!8`dz_RU`TbXWnT__Xoj)L9XY221wkbDb3K22QADi`^S7pOT1GXZOkkxoP+1=;byGL zH1KP2`C=?A)Ww(axWBpvL5W9vBo5;U{AS|9xWT@8#vD)B0OIRG+swDkL;U4UNY?%A z8~;2h!&1I_d6)~1nt@^j45bo$(q3$V*|3a8kCJkHJSiqt)gjL4Nr%IzS}~w<#iUTl zicc7^j@Yw-kWb>4GRL7iRn$~{CcgWJ-;TH5d@F9=xDj`x_}-F130{>lj6!^WZ9UdE z^Of`N{z2^Rwk0u6<7-Gp_E*ND2FE4dci8NSDsZ~ z6rwV{v8q+ zmH0RQo&RmL_IHX-GiiE-&O89<@2(&4t0|~D1(#e#FbWE12Q(Xzi znb_Lv#Uq#I#n6M`VOlehtm7w4U z2b0sT4+1B&jo<7M3F%Igzez*sGk~&i#*^{Nm=A+8^)Ro~_(sc7#wjtTi#~y05&C1w zJqRy69TimN(~x*ZO)0UmwPb#U*Mr&F|9$0^ALSsb!_+Px281WNGQ23$Lm=^i7)SaU zAc%jN(OewO@uUd!_F{VT<*1yrHJKcbEA!KuDEH&NosJK^9ULCVM)xS*UEPVhH}A!} zx9`N9WZKHyq-6h^-zi=?bI$uOxssACPR;#2jTJlb4taUy5g+eI#ZY^Ql0ySOAGc7Pvg-8S;}w9kw)_1fTS0JY{PMhxB>Y z>yz=yk8Z@9ul+EZt(`d7+xAK)Zw9C{78)4K>NUQsHzi!i9?5-SL~qFtO))N^&kVsb z^&}gn{0MNdDulDvFo+QBDKUW z)d&Jk3ew0v0+On{rpR$ z!bO5~7fg`VI>7*0hM)SwJ7XJgmvIY6J_K;Qw2%4&MUIA1{~?&!3LA}SEQM_d_>|^h z0DgzdGHM7Y|A>*lX#nS}won;A%05=9sjtjPDVT^aJ$^2}_SmxHRiyBEbyf-9&?Ni& zZ?4DhfB)53-`tDY`kZ8mCYF=kn38)TTAzR|EF;4CTR=g&w*? zOG>sBy2l^69M69Iai8TtE+Z8IL^ZOeAQ!^@D2tZk4HFV*kba>uK;zuvOkBCL98FD@ zII!O8?MjBp6;+=@kC@=>t+HY5_A@_At7KFL)0j0wieCzS;XT{py-gUcTz`JtwYCL-Fa$LQ5 zF0Ng>8Y^d)q^L?T3orSxnsPkW7mvjhO)|I_@?iJCgTlEpi*e=Bxj4TvA9GULC&v!u zuR1}I7MGv_1wSvPuSRWV#s^2(pN~RpK}Y<8cLUDaUciXmHvN?KU>J#-Fd85;sOJQH z-JXkhDI`k)X|LhaK>7j)QNSld6lkW#)7XPwhZiY^FrXvzCT9BNZm&`n)iX@DQIYLpEenfpP}1OSTml3(_RE+aKTsp z6js*L@Xvv1Ck!O%(q9qyP{??WTck^qVWWTPQ}E;i(BYrh(hMrw35KS`HwupGyfe&O zRZcV!)?^;}$2pJ__0_s2pPzeVA^z0IuEymtK5eh=r#8ekq&3ZTzx%^m@kg)RjOKnj z<{Gmxu1O}>r!8uNdLkLvKbeYak3JEz=dZ`o@;NKU+^TTl()IYl^Pf)>p543TpjnRDAiD!bHC2_E;%Q?&-#wsacOdE2FAUbgi#Iq5?$6z_!x^I-$E z`k@1_=9xL{3)`7Nq#9;Z*8U1gm-~e0s)8A_pxnRwz*p)hXyuxY=9l&L!#Qh9Tlne)Fq0Y_T}9 z5{qKY>{CDb&#evK;(%N6SP@2v=2OdF0WLH*wws^`dv!qhpcH0fROPGTy3FD=tHA8T znyf%y3I>!Xm`!{X)4a+{v|8y+#n+0k2N4-5z}WFe=@%mucOqsqP<;M#pO2@W{e$ zwTj9lL!PYcGh^rQU0sSNZyMVD6HOGP%qnxXJG6RAq_WR1)K%9IrLrr`O1fwf8Ub5X zc*VydO%!45*+vlryo&TT2@SG`OiVNxS-XEfugeqh6$zZDo_;FM&P~K(eJp0H zRb4cqq543mvDyy~cn0g>Zwg0BklPQm+p%Y;0^<^Rn4p&(kI2aQ{>34wQ z%ToFkf5Ui0k&n#iV5fhXVgMZZ%e?qWL);QqxnMHHfhpqPFJ6v@0i|O3e#rXp>TVG? zL2msaCw)7NS6S)fk|FAM^!ZPmuf{KY`g%M)Cxt+0W+zT!Npj?7?>K(#AO9f!``>#x zwl}%|Uh-aj7x`S*B)Y2pd-nXLn9z7Jx6p{0h54A(Sh6s?7*9U&NL;ym!P_Lf*dvls zAxNGgTTYE>Zx#6=kbTn`C6x5)nP+&ADU2QP(M|}@r}5|Z1h&)FC6`uKo`|d09*=sp zPC{Q}#<)jg<`5j$mhDMLXm;AZwPR;{LyFs`6yx3KclP3m>*wOBk6p|29xTMr2}Wrb z4W_rC#r$;{3a`QfkGXishtrPxaZPgd$tOPMH054b4C(n+0dMY(HCfu*-tic^AVqkt zF%$E1vx-x!HWxa|gK&y}|9^zjZ9d_Bc{6D_>8U*J=!*f2JtEyu=ff$yH5DFT3o6#Z z3t!Kwh}iQi3X@94)oh1^sWrGheHAK3X~-F!1YGDI;N`kC(tCfl%!dsh$)4d`Z@nzW zKlz|B=mJAAqCwVPbHrMR%C;-Q*tQ>T*qA~8^4ku=`paJ3t`Dwp@|Z(g_@UUk!H8D) zh2n)GL^_mV8l#1xaOVUgJ8E`Xgb2!C9yCzVt(29k(X4`?{BjkFuR&8Cj zaL>ru3V|}OFusz8(#wn+p-K7K@-fW>K=x@S0Se=_JYNgCdgXHb>^FWc{=}dC7cGow z_d=|H@kX>#CI0vWv?-j$Px6LQOiVb;m$a(>RCAi_t7a8hPwJUM@kceMlR8FF<;rKL zd*oBA6QX5C?Wljmhd0`OO?Y;xp>!uwnD=9Qmz4`vebmuCKq-GI{P20EUh%2e{a)f0 z%Tx|l2@v`sMEsuS^tcAXF-|^12m)OE^Imw-jxgNa-i+0ItHs6hQetbh`1}`tGOk@Y zFC41Jem9j=8S{UGwMY4bC`IEK+nEa7`Uyc`91J=^ui9WRIGvB zOtGJn`ceMopY0$ia3&0&G#!UKo|{MkI0~y_=*Z}aqx2on@840oeox^6%J2e;OuafS z5SRQ5WY8cNnZiE6In3)c3?dZa2^qVLuQ(%k2B>8%Y0mtWqChdqpTtup!(K!hgq1N! zr&ySQ2meJ%WoDa^57S^zC7yOd7>Ve+gk{jpJjq=KCLhzr!N>Y=P96TOk)b=aD4qx z67Cqb1V=_6Ypke6eP+&MKz*iey4&qN$>@XFlRmJyxf=KG-i-UJw_{`NR%~wFiJhH$ z-lx5@w;H<#Yq8s0kBjFP6J$L`*aFwn-4{a`-f-LQ~Wt;E-6YMNj$7C0x3 zfeq;@<8gT0Q`uOlMn@1`jD4if25l($JIZGO_o!d#eptK*q4az335@!c_PhsJO060h zx{N~Pj#ye+@ztTLt9Rn|ty{6b&(~}BeUrrgett}4|Dffqf4e(7zGjdcBX)Oo<+tar z!xFwD`<~J^n@z>hliMVOV`q0Sws%zT^563ng$ME1|J}b~nm#P}NR;7M{^&RTF#x{4 z#h|{svF?@O^A|3}&iY!s{`%`0$kmYOlw!{2t+!$V;nY6Rj~Mt`>NNd! zOBKB5S@g9>9=Fm9on%h~r*n~|_cNIiR;{I&GRxu`7G{Q0iVy^>wsz&u^(^f5{p2rKUvoltKhYMNXsOEa#6>)t5P`OU9;7ZQadd{$$iv4pA?>zk~=L0_RdZ_Hn%w` z98s-EU`e?bv!g9~JF0^knx%5Nzta)v2tQ`Fd=-ypeluTsXiPz)VzOGpRs_GbfIiW) zx4#qj?nsz#?@9?(nUvR8zy5PlK+mhas7|Hm0w}PQ2PK%YxDDf>dZEqwnl=q?;Pn1x z@zn#5$_vm&C^sv?HT4aY>k4T^CzFgR33ild+9=BK!M^&T(ohE%E}TmRyAaeVAdCJ3 z2___;>1Uh&_;;A;Eevk`mGOq4^6M{at^j9%JNM-)+iv&RSV%*1{51IZ}lF;_jQ@3R2Gn=>ZFj1vo*^e&zbWOLeh!UH_C?oV@kS5&xL0~*Aepw~Ee>)Z$4eRKH$Zw?LF3OTkV}k>EHY@E zlUt?9`1<4i_jh~EP`4hhw(=(cgNH1VSVN&BUip~pxDKVWuh6RwtRLXuQ;fZ&7WzR-cR)fAE4A2|16%hoaizS@~ok4`V<3chX%vxtp2E z9|!bF(~@cWD~N>mti!ZA8P?JB!&Xg)Bg=;fSFb-3SFT<24C-KiJ8s;3CtebhZ@u@U zSkesdAAS1|;x~Td@5YUnz8m{*|1jFO-ih|jH{(RP8=#zwY7CAfl?ee5P-rc`0&XnmO6lR=zK-Z(Rhh4%we;ka zGS9A*5byO>hvpEQm0u}fIlo9NQdM++4#$Wi$|FW+LJ5uLi`vLU6bysv*LiVN?2F<+_`-_ z)-{QuzEN^N`Kixn@^>aGlT;qtX2c5=VFnksNyuj0gBF$Hh4Xg-4OFO#dJrKsi6)pJ~%yA<$de?O)Y)pIkXTS91IcoG{^DsY*A5Qrod@${!xDsFb8PO^K z!Y{qkT0C*51<&G9xC^iMLm3YKq4S|8fih&vB1K8kNCbmRYX36A=_pRi^TEcJ9ji#H zr`o&U+NmEiqn*;=KQ})Uvr>5dhD>7uUoGWut=o$FXI1^^EccYp&BRj+(=o69Q$a4t zytKI$|Mvg!{~G`OfA>3@9Z$z2&pZiuG(4J`?&l!-S|JWDt79?tW`4F!G;&ccXrP6{Vqlv63J^rtM%@&Ll1 zc$6vS7A67bO z`XJZdK_@nL5902!VN;vcj^lA@tbC;=30W2Ze$LMf-;T4P3KNNix&Iht=a@B&_tc8?>+=<-8Wp&qO6O{XO*pJ4E zVH8?c7C8_QtmH}wo}8{pR#&}(X~hmBC3;ew)UVl;pHw^v`N^1O+k^aBv*dO$^A71O z`jRfusg6V04$b@H`01beQe3%sDTShh$&nQNJl&g(AB7X+&R|}Zko9OISJo#AU!f4& z7jeS#{lLlz;pHI0_vhKxFgMo_&!i;wTQOgsh*`C%W3`8_+KumIR(-HK&+8%h>Qz_e z*_8s#!MaL?{e+@Y<@2h%gayh!Jg#UE!hLdFvq@jPEY3auF=UhCf_sm1#&8&^zR?F$k zuLN6@NQLAxJu@Ac0K!T^JOCpq(hLB1;WuI(;vA+)DvN0PDeYyR3HW64QQv=7D7aA; z^`8j_{Ig}?$mjb&{vR0jFc`sk%05J61*tIRjs25L51P|3EbK7+K?W45lo|OAOc`ST z4>m=t5g;*PYHTu&_rAYO<&?&eKpGoskk~WWGJ!tkNckwCqA`FiW84H$tJmV};#{1a zZ^ZIUJ(gx?SRa%Ol#G>O;#`q@SW>)4&n?IE7v^J@E~xzG)VDU**5beZwO@<>?7#f~ z#B1-|h%;Bt#_rbL=pWpR>exZtdi%}z-S58`o4dQN3;MISzKLAMEa=R|5h=KqHCYx` zrgPzht#ynA3YYQfwbHTo(rN}|4-cZbf6MuKzp#{J-h7Q6wOkjbO@BTM zWI)kJ!v_>1TuD2Id<%J4__N)FZF@D1VccTGLdUf$7vtm4J?%HOQ zPga}Efgkct&r%jqZ0ex|T35`v`9*lPBQqn zn_pjDi(mV@f6t%RWGSCUoqII%Fx#X$HFwAIYOAmCXPqcS>`f!+%miZLIARV%m)ocH-9PcXu!F}0-;B_ z^J{3S3}LL4tHHYb;sXuO+ZN~uC{Xg_ReZTe7=;Uk%t|YNQku;#g^5w#aKg$kn%6OW zRmX;pYW}WE$e@*@K%Z!!m{8|`?%Ai~ulz6na*U~bY@^swhvHS4sG*7qC61L44Kay2 z!=TyakYc9t^MMg>u9uhh)hQnj6y^!k4vL*PWb&0AYP=|v2NMrP_AK4|Z%?Egn_a8~ zk4w?*#r8f|$~4u{$6|J7TC-C2J&(sh_t^JEvl73uREhlqP2%Jy3S#`YC!U>{EsFo)=;@DsS$dgkGoVY(sU$D@px4+hNy5aUVN ziBrULqJ+#^Zu)D{{xVWN9OiV;E6@*Z;QdhgJs1X!4~zf)m=6Uam1YDI=V(CjG?58O zF(u$h`%c$r`9Bzjh|HSzfof0w4k7|Li--@6{!oC9)3hWC2F`4kCG`Z@R?;qawx16& z1q}E^5~cq@;xdl;T9hNb(tqG*1o-hR{fe~2DWcQXs|WYva{pG$G^X8VCe@c@i<&wV zZ@uW@oa3lf^OggypF}3p2hT6f#ZqG;<|lgb*m5JDf8t`y&CkWw)=s?n=6haH;bR58 z?e%!?l^@1`@wfj$eD}>;acN0PHnLX}XI{;)hy(4&IFw&_z_{fSxJ=@+pEAXhU-;(! zRtQ1MVtFG|kh6?T;73Mr(2F(hUT-}1s@LMJckjlnTQ4eZxk?K{qbc&tCt+UA)EI~? z%DN^TIm>v%*n)D*n3T9Cd25QoM_i>tIc73}o7HM7*TUuThH;ECD4olvIBAlGvJd%+ zGKmQ8a%S7x;{0rU`SYKOuYKdEV_}(FA2hb9{D@2Ei=ORI_a_dFC7Yh;03M&ad~Aa+ z$@RSYiF^foDBLn#_$1@G5vbMd#?70Xv2(DQd@(MC%Xs3TI`p-$#vt0=@fi2WJNeV= z>}l*lAyzpwJ_6-L!*rS7qr5l`_$&CI2xa|&%j3rgFEJ%N3|E==lM$iq?7AIc>_Z{ z7VXfbW^0<{C=dGJzxN;f2axiifn-9#u@4dc(|`BRsUc_|R4}t^YOleK z5a-XU;qiK-Y;fp$ilj>C)y8;bGXAUo{Qnle{*Ql0ig4d?RLxe#4y|PHdYY@TTG5sw z%(jSM{bPS7u0HXkl^F1$BywdLcgXfSQkq2%duNy}viI1l!BR|6PO$SOv>uAGo>qdD zmNM{JVDA<7YJp}?Lhh9TJ!vQiVdAl(P5HcvsB%DqJDr~NG-xjQKm(=00qi;TzEgKc zfC+*G8bSm^*@eVPC7q(cWd&St&_9aJt=(vK`l3Z7NIFeV*J5d9F24A=&%~ejGk?k} zD+rHfTN8j2DPW>^Ldqi~v&WYDCI4I<77pQtR+mt;Ff;5Qvs#&gjFnuua{CGU8AuIX zGT;FRcIoQssMF>huQq7{(2fG_4`RqYS7jxp9qr>O)o;Yk?m@hJe@9^_u{>9ci|3cU zt>X4(C+=-`V@mmb@~OpG-Dt)S-`kCza8-_WeLF>6LV}fN+B11@=;r9KE&6g*`8(hG zrVkqc976b+um60UyRhQbbYItF{;}s9Mg7=^>QHcLJFuHNH;+^oUQwWaRd5wmPs&Am z$;lEt<+HB&1*I^b+QK`j_$bH>ro2_?k4?opcXs2YmtKvRUVb@l-?ut< zfYZ2T$65LnG=Xstt)z1#-OvV%#5EKOk^mmzvOFl^1MvXEl)xyagps^x6LOr2QShRK zGF1tqVGlx?zLd?z&%?hDha9j`8q4%*XQX(5E_j}7Oxy#yNohGXvWrFKb96IC2QubC?J6OyA|JP=*%6Eauw$QqMj8MEo0n^)JT9KKqQ7t3xTXw1Khd zs^55d{1|pW(NMMp_=+qUS+utgcDcDmUc=50KHA!xPEP!|{0c?R)XJ|IQ!8 zZ~WFjicTLTnwF+IU<{nB3nvO;J?5p$OiRh*(4_IOFrfl-a14EbLt-ZyCk~Ih;`_9C zBHiGS)%9%mZkMuKp1SO1i%!Tkojd|E-q8PHcEA%E_$!{8E*Y*A z%!NMv1a18K;>h zokGg!SS7yp^`DFL*B%iG>;p>y#+s>Tn^JaB0M#3<{Bq(S#odFd@S&u66_%w@D3?Ff zAnbVY;?A$KptMp3ui~mq(+EA!QfX*3CfTPnP8LTeNM)mMQO^JqgNLcC`!Z^It@w)iZKEyg~{=8JMORd;=AAbQT);Oei*O4do!x_ z`S^?f@}G*I{K_xK6pHP{N$j^eQW&RIIi2XLRP5iK7TqYU;G&G?t#D^E6mC7(ckKy+ z=tGcm>j!rPt9S~VkPs4IT{qO7Xx1M%*j_TFe8GVt%GL~4m)VQUMCI=4mIPimo_h40 zI{BEtNO#iBy122`i-TrMR360@DZ>jh6LEj!r2W-tR|! zY&Yf_6_rPMvqeG#kyoduK}Y4^*C6`+7rz(VJ3Fekv3T8MvsoirW*D1Ij&7kx&=E;Ez{jl(6I#Ql4@Fd{~2T>Q7Ioyb@YFs&n=>OCI^6 zdVm8846w`}zxjtR#UH)!vIg%RABbQ-^ka`+iAz@>i{23f2z`L|Naj3qpf=OtxC@{` z)=%1x+6Q4aj6U#1^;*UyfCr6Ez`K;fA}GRDE>Uk!djn2G4dYAM(@is_{23wR)3^M@ zM|nERz-I_a&k)j8`q6#{xeuUJhAE`^&r|$Kw^BKLn9jB$&5#E;sK+6_r(l$>2f{`; z;`WUXFskUf%mr=$Y#9{_5^#^?Qp7FkL2*eSj%6S?)2oOxihmSS0rnd4FUujX2Z}U+ zjQv4@LtJJ410bw$mvlOgeTGmf7|2GVJ6sl?`POSZy5IQ+arYno=b^c9%++i0g)cl6 zAOHAcv9q-;xzLMoCgB=~Bt+xh#!lS4xfbqfY0}wLduYeIufHM%^Hvv8kFTV6%& z_fC8-?atnI+`WI#2cu8`JK|S$IzP10KjL-`wLQiS;hmaZkhkobg!DORCmy0~!W*`w zO-Rly&DZ0ROXuQW`E!3fe)<=_8q>4&Xm72>gm58C$7_o*F4@VIU}IBR7J#>dK)I-Z zGTh$Zl>OMBV_v;?Uv{+(aw1(s z@>}IZH}FTS*jgv07g^VsoAn|8x#dO6XKqjEb~~bxufTR=SCgk!r>Sy@RwJ3q9}O!@u&R!ut2f}sy|%#Z~sq! zLrh9Viv~$2=GFsuJ~4}xEDL!t4Dxtg!Kus&udgL!xcZ6??p4bIs3;-Gjsq^?0f}eZ z@+ZCzL@51ag?nRpeM35Bq=wYhjB$BH!ZnbCRwX^{oPG#k6YZH_5#)2;mY1f!K! zr7g?kw50Q@T~@Lc%@U!2vY~LWLY%A1%rJRN%2>m@`V>( zQ3sriU;LAQLBgfwM9i$)9cWXhm4E>6bcGuul76q!O2L{Q??@R{9JwhoTMwALu&PVP zJCcG4EtRRf;G4m)hrkiR1xd^1gv$bVsOM>XQ8Q=Z(cVbhDCc^BqpZ!ey{IC9T4JtWlVkL-4 zTUq}etVJs;>9k+K1B3iLNfE!;UPC>=N9vcFrmZkGRwcaHmqz(J8Z=SJCrKIcmb`01o=;K3_Y5YoT0eU`!#z8ROm3XXW6 zZrZ24&7aJNeo*=sfj;wj}K7>z5C*6Gl7t*dkM=Jysj;zW6`H z!EgPSvAeY%b#=Mr#pU?ISD%mPKL4>eX&%JRj^vaSp54|#tle9WyX&p^#23F3^_6py zACfW1SM|YLx898JfA_nV3v|{gDI;pQ<_iLV|^pucrL&+S$^khx)UkWVvRc=Xj z$uX|-L|?X zGR4h8;+5z{ocI14zotB<$E6t4Iyb+vV zSoBxyz&STRFA+@pfqv1TI%blB9^`Koh(?+K{C3TYFqC1&y*$Ln4H;@PYFE~Kth>m~ z3a!cpSbxzIC6p~d@WnXw~i0nxK9Vv~nvQPlQWZ zK?5MXuu?x35W+l zkkOEN#enjg3Il0U&=JRFE2)BZbrggfTNhHW(aE#i4D7ZmPv~_a6heF$VF95y-bX9H ze2QP{uZV39!SXs+6X>zqU80fn-cdAS%Gmo8(=bt$nhux;P-t0HKvA(()<5TsxdSzLu4&v3@{b+4>;xo@(l%P3|AHBC5 zFaPjmE4KA%wj|JQG?|hxVby0-g7v*0z7}u2`>yCbj-USOSL4t9D}O1*IAEf-qS=Gn zcc~1Q-H)^lD98W+|MW>jK~%$$p6no?s5t6GJffet&8t0#l|`GUHV|x@AWmr@rR|Ah zlodr-Wagw!@k*Ycs55b+o&3gce>cAMtsiL8H7VZjMOWo#HT&YlGjZ*)XS`CuiZt!8 ztfL{&wrD4!0Q%fsigs%++B;>CpJ>95HZug`Wtk{q8I~o2Cp3_3$5g6nR(AS8mfh?! zZxRoG^pDc6#|X$#@Ymq5jF9n8MJ^!w#KVF38Ax6kPn7G)1h&BJUjUjNZZGx`Yx- z=2oP^ZB#>i_?PuVl%ahX#~2FdsBgl9K58e04S0Wi0FHn)fjBnupf5OF?YH%#e-tOb z{a^Xo&KU`tvA#lPr{ZHzUyPsr+UG69wl-KW*^js1xe=|S$@u9%{x3#lb}}fWv6v4c-M@WXJV^PHZ)H=Kv3SFR4Q-~`>c;Q=;p@>7-le6*ICt?vzMK|0)}rIfpN7v7~Ea2||LdP1jVx4`6#066eVlY!4X^|YTK7JMYi@aE*a80xx$qiuLSvz(wi|d(0mnWDxvu5 zxZ($gC3s!KCxW|2gsVf1X(nEskJ{XFRA(AqwdN)UCKN^Xo_tJ=A8U3kLCvR;<;RT} zwFP)!h$v}zHk8a0nCV0#QBt{8Sug2>?dNbBGwqpDKmiRvzk3aQH{LC-@LP&;dAk&12 z0~E+Y(M9_)H)*2?3iSZ{$xwAX2bEH^T@S8By?6p3H%=63Uh@&1NyXuDt~S9*lqx0@ zDElbq8~5+UZ-4XExOIC?la&MUtnb%&2isdN!^+iXqpf(xD^BKC3k0LkW!4JI_GJZF zVJYYoh`$F*84uHx(<%5VfcC`-n5-@yVMC)7GT9yzCwy@69mPrPz)Ai?VGah6Q)WXV zvEofwfx^F_PAGj(h589V%P2T6+w@xmx!=7n-q3Pd*UNk|O{OI{qmM=`V1Q_7B~6)P z6zo&r#WgHrVjj|%0mh@Y^FA0vQ@jG-Fm1vprAbfeGt4P{hB5ISx`sAjkR@?W1&ul$ zq=DE1=P$RT;&*ynNcu9hWwhNG!emo>ZNK>Uqxr4B9cRui#@X|lAon#1ma^D7*pak5 zj@oQ3zWRkv$NcO}y!+-mvDZ3|pZLW;8B=F2MVA`{cD7?a4qnb3DV z9w@I|RFORPs$H&nBC7$vVZp+Wh$0@qwO-&MEpd^xgd-9#C_97|obWG-5pO9qPmV6* zLiad+xV{@dxZRHJZZE#^)t`&6e&vhmtGt2hOM%b*zbNS3=9En$$Dlct_v*FFaqiqo z%r4HmZ#fy0=J!fx55vI`gED67lJF$&6CGZuxsU{S1SvfXQTwyN^O zFUfAvi;T)c=bTX)i?j8Uc>L;eOwTqle<_ab-EHwoV}NA(*5*b`)*3N;_L_LjVn9!2 zor;;+CGmw%>x({Um+5z#Qij*%&)K1Vtlqk%{Bj`*g%){cJwkc`=RuPFC%#Ymo8Nv} zV^>G~>neE4`(PV0(=nxSk(F7#;Cl7a#d!YtPsh0{7sL_q$%<{tAMjYW5IOKac}J1p zC-ngvvp?v65N<`DMFY#m#0S7XCKO)W6K`1!Ht!)?&=sw;MZWllKjj+8aiKu@cb$?a zc>pRicDwNycY=MebGj5FZ?*h?e@Tof@NP_H?|h38CGc=gDZ2nL*7 z8O}q6ClZLq{`w7jfY;aeVtwl%+R!JZkxBXNNjIK*`f@yeX(le65l+1ZU;z!$U@92otR|Y~}iS@-S4i8F* zOd*unvLK!e1X%K%_IGrWGy9^jS#+F^TL>5U>v~N;c`XOwHg@9ejgxrwMl)`#ZbnUI@tX(r2fpUS-Ka(iU4l}; zd9_3Nx_#;C!GrcqIPqK%pnRqum;Pt@@S$yf< zE4XNTLb^*RX&r_GN>$()q!L@eRmKKfPepqBDUH(!)Pwy@I34C+-12vFJnS9;Ific1 z=Cg3=H)$KiX-fQ){)sxn%SQr^QyzBWGnyt>QHL5T|Z^z!Z|86|}^y9H| z?qc+g)wg=>=r?y_|IW>L>FpaarHSvK`IA2rpZ~-&aj>=@t2^De{u5u0%EChIYm&Kp z?^e9_>PzwJi?2ql!Q?mjVZMkSzwuJuJ}}1EAUoyupW=8-5lJ4i*~$v3she)}Ze-r9-Zxv}qkzvmXt#!r3zGx4|nlmA2TaUu4t zvd+&F)0 zr4IkKA}Jc%9=8o zGO#L*vYBI|_wDLQ8>25^r%nl{%q4x^vg9YN6t?gLvB)cU2$0@J_Hu$(`sH|K59toz z7cI&o@e+{nVXntB#RES6%(H%eSn!c3!#7?UQHG1kVHSIel7gWC9Sp|VUPPi&P$DIswt(D4C#6?ohY__c&6c+_^jcD!rK}&LV5|N#+(cdTqjM6n zPh4M$OXrv3sKXu8`^x?}Zr$IH7k=^z96a#%pX*j}6kfje zBeK&Mf`_Y%T*tHv`TMo3g{An!t5NY3T%0hc4GLwnCg$T?uk6LGt!6Z)r~OeFZn_vp z!50rk+Xr=I#n%SGi5p>>if|6hDUgdF8FEb=^sMYlWgCPFT(nclQl?QVKhj=8l9GNN z1V|%Z+dUH3d=fpTUG@@2%Thw=SJIe}d?hS#P<`M@+RGGW-q=eZ9eI`^C725N8;fa5 zTmx|1NXYQhIMXL%jQ$Sdk)|k1Nki#&MHeverXk@duzH2hE_w6kV5KW@7)pG^6e!I7 zmbCUO{0C2g!VP5_!u&tPOBQyW%Io0(*+5cDx~z{NkwI>J01D-U*J4A=v)V<1jI}HStA;dkdQA@dAYUIEo@Gig_avvMOZ|#&sr>Ze$z^ zbe@9|7WWtUOh5F(h1(KwIkzJYATyMHe4MiNV@&dfn&WUIc%VfDc-45JQHisQ<31?N z_B2r`IB3}6MH*r1@Dey z?e?uWk{@lNRD@X}rtNz1NQv0y!uJvW?%(}R{NMkZza6`+U6&o;29ZZDG~&$CnW)S( z;>shB#n*rCSL6JJi(YwVbvq|D@JD4()wB2l&G-ZGf&*}~jXCk0t#DJ~4H3l?<>NP_ ztX>N&Z}rUj7cY^=ikotn5b<3I@ncGcyarrv=DWso(ZvEF^}>_eA7&b}3WEo3riF|b zmq@4d;PYFGL}q{UK!(LK;*^kW6bn30THtLd4ggQY6h0Uu1)u)Z^L~C<@R8gzyt{hS zjD=$*gK205FKzf4Fc?-A0Wg!KaMG172gY9f5J${pRh50WN16@uH82}id>P}4Bd_^LsgTFN&iq_63R+-Zsb$Yh<_;@Jt{7%&lpaEvSQ1sEE62IHq@rMus&j@Htxe-47Sn^<9Nb* zBCHHT3>_6Z7&NE>%E%MN3C>%HD6T|x@O8M3>J)*FLMZVs1(}t9K5laCm1VXCFd+2g z--V^B`>iI|wnR)#N-$2tySDOCIiz!UmG4%w7x(UM#@6aubc751n|tlJcjsPo_jjV( z+=|DqUWi})h0jI5(~h03mT+@gydMWmu7znvQ^Jw&vU5nH-QxS}2fojjlf>W{GfOnvnhuI30Hca1Tv#%3zB!A9xI5SiT2? zM_q-TsAYNdEXy+p*JFTB4}1$(8ABdw0HhxPG317yWn{r<+Vmd@EMYXk03N_?XVzTS z*uxF=RJM;295R9qS(DBZ6X|5LfF4F{zWqWR+<9I7KuU$e$EPN(40jIpt;6`y+Frc1msf?>$LGDJ z>Z@P;e0=d|f68&_6JGdHdhA^9#bI4?_CYl9fCE@{DMuW@=JWPeWR!q$B*!-8@9d!? z=}RBUSMC~C9Dvnk@Udu;k>p7rUGJkq9eu3rjIRU6)+vqf#iF;N|dqZV@QyM5#nh7i^8z{+`O8Z^1lv z7jet-&~&Ih6jg^`HIc0Ue+vb{8J}u4X zpZQRSWW2@%e~66RHpF)yL{~munO8k=*p9u>9Jpg!3yVA`%q!OIm=SiSz&TpO)e-gM$it3Mg8xKCV1M_po!A#v57H=19udeMhK;{)5ckt@|#b4 z@bC}TbkjNUgOLK-u_sI9TQr1{t_$Zm82RDBN1_bxt-b?S#0jqtxseR7=tzT~+ukLUaDth{sAARkQO-oUm546hseQWD#HJF&a9Ei_V)XBT67{3!T>8~5OH z?b`Gdcj7X;6+O&KRYe;FStty)8Q3DK2V3y45Hr1C`q*dp63Rd29Hd%O` zcEn8!hX@SiKEcfv$5Ph$?)+5DPDPxlO~%>Um_H%iQ@Wk~R;+DGaot^yS6}{)1kq-E z_7jh4FhLPU=}o1Q>-ab!d@Ln={O~~TDukO)`MFdIB0#x&=HFVVl0)f_8 z8tO{K(?&GtG7w^SSmEcDQ7NgN?0FiGV1UPV|7IKmVTH8> zGl@7^4(tZ-ly*bL8-&}90gC4-IkH^HJC{RG4vV2g?rS3b_Ue9oe~0rJ$&3C;SMq5xzVH*Dh%bHZXT5m9 z)spGttJA#>U;ONPF@virxsnbAj4_5Uq@JxY`DfhW z{hxR!yeMEO+6OzEBH+Z|E2v5t9h;sBKKevcKu_Qk&iO?>xB;(Ke&OkM3uU+{8;Un| zM_IEjQl^22BIfN!8ds-|VrTmxUV7zqN{NM+&O|(f9&k-hS6x>X?g$oN zvFFvR#e$U5_zDkU7}xtPIaLjN|@d--ij34@FP;5aHeL|5h@{ zYyjxkrCnfa$uNbg2B@(2fh%|DPf;8VBYP?ii8+huniQjpG({L-CY0&rlH5-8*fX$* zD}@ByGUfFq5A!(yj_|qAgu`5LpfwXc-xc5r$BGlUZc3faL4FmO08F+`|fl@%w-1z#M`QIifM zkljiXf6xmaeig(uF&DYtP@VPmUOyU7{6s9Cy&@&Gn2i!o9UerYqWbOKJ36sGD2G8j|L-Ur82RC&Hc?> zCf?_2GPG^w+i7#%+hNR1w&TnK_b`fH3F_pN8Bjcjc2Nw$g_9bE1~(9g^N|erZx{Sj z^lVEC!&4rN=ia6xJN&193r`{>9%fB)!((ue9)Eyn@W2zBk`{oZg9kARyAT|XRpNy= z`tj!dw%okN%ZWvtXd-s)p(On?PRw z^Z@V4Gx_`=lzlWI@!Dnji*!YB>5re=$OvR>7nG470??iKSb!Lo0XyxJf8!aTum`_- znixXuZ<O%;2O@vl!$$q&W#II6O*NFidvo!34JKW}6Dn!L+BScH(t_9o== zXVTBT;QpYFDjeRS)Ou3OlSWXuIV@~Y-SXjrwdP^GdTT4*Z63!tDO}Z=nW$9fq?9ds zvEi40?q}jp|E0eq`m3?uo%A-clcPON@LBwkoE0i~5cGNFQRUX0X~|b9a41yU8{4tH zvs!=RlodtWXS;ziGU?@b>D6G~ZXqAMXdr%x zzR8nL%#62UdF()n?zj*0P1Tm7Qk#{Wk(}ONi`|`V>57%8i_VitDZ`Z+--^TPGW(vZ z)mh1>y7MpPEpOCzch-d?dAW7}UbLEf{%8^7Ol~_!1r#6+=N-mQxCN;fue|zR{MY}l zzaKBW_)>IKuO(34$FH4@#~;0{yejdD=f50Z{K=mbK9#E&^PwYg(>HUB=FLD&iabiG zvay_{J&>;#McnU5tN6uO;K>eSnjU#b{n%rp)0^(>`!uhzjZr6{f)0oAr>HKo;$&RV zVn^q33}vgDQN8kZordczIhtgYmk7iOEB4?LU5dljQZ@NPNoLhpb!s{(8*$h&G~BjI zUVN#KiO=KL9}%q|9(*Lq@QoLKo%#@@urmoPKYbu2IjELUl5D|F-4>9QnrJ+5Vu}bT zTqO|C?g5^BTG$kJu+U+>FGha+fsfTa?}?H^WPcettM~E9Sb&JKU61BhC2U#q5TW zQhl~%(g1=pHt3G@{>ZYKgym90DG(`wlwumOs1lx|mqUWU;}2yGj(Y)b8nn{nhuWod zyw{^>L%HD1h+c}q=Rr$gP$ItZ(MJ?e3=%fu0;Yg>6i!Gj)Nv@Izla=u$r?fix<|fs zjoSe(+&*~5QTOEgr5ioFbi1jti7w;`{^Il{vL$~3^s0`yO87C4F!cN&?45lHu7%gUil~g;r^kw9!2;T9PeX4 zd~y$Z4UOXb`X%xc#SuAABSaCs=MTT=t%6-LQxhDq76W)3J+~`lZ3-7(4^7HV!P*1_%!=-(v z(^vN{Kf1DonY$=ou?t>8yzZRFk)J?hzs6x=pS~m5@l)g!E(XK+kAIhzlVu*A#oJph=4vzVqqgJ*HPANV1fIOh(OkLHGr zch`J0uQ*w?Vs%~sy`rRs*4*Fqor{v`jE6lm(U=}+rG-Hc%~`mO><)_;Wr!%RgQ39i z{&o;S@>kp1uy(WlV3Zg+hir463hgRj0#%9L?bSN=t*^gi|LQmYja|NZ&t)I~sQBrIQ1m3iF>CxrIuU$Lp-u^~2DZ zCc^Gh#C`VUzXf3U+Sh;0E6|-Tp+W(d{15NFN6YU}$iedU=ZCWEd=k|*e;unLZ0st{i9RoQ%p;e6Kz8lXO7fwB*75L_yim?K*# zSfMg$6X~3O921#RBUFQD{!kmCv~LFjnXTO|HB@>oB?!p8%HXSnRYsB}FMtl>CC^;A zcYcKqTyatPyu|wNQF$0qL;k$Pe9|BH!hxuRxTz5hN17FO}Mc}0}e6x(4?n?3s*0`96dzd_TXK(Q7n;A{O+Y)`_60YQYcvLqMP!Gkdp}YdW`Gbd*#L3erge5 zSapUlMc@*b_X;aBqFzYXL+e@D3^_%qP*3C|v^gFq5e`3b{)i)X@m&nA-NBpDGhw_< zga?1kedyufWpLrU{>Pc&dwA=7eC##$$lpJ3kK=}49MK%njJg7KlJROqIk%orycpBRN>15=T8s4avubFhre)79{7%njh0aM;2sVW zzoi{RW4;BWSKmm$Z#;wH`W`Nh@dx=pZF^J?f7FX(&@Jbm+*63j({_a~fB5B2dE9>K zU)x~qHEZnd*zFrPZGLXjr$clkV~zxTnZ{Mfq#cCxC42UpUE8P&5N5M(=+g$qU}(iX z62Rf`AovKQ@ksI|B_gmb+jOO57nbriqw9lgi>~qqAaMYrf5@*+Pk;m2rFQ^WdY=Iq zd~-dc_nkrC_L;*h*jOx2+T7fnojQA7VEDWbd(*KXabihZ>AY|PtDw+2Hs+Ci=yZnYr>7@vYHrpzSOd@m$YRPFYbBTmJkF8DDk~?; z9kw=X)V=SAaurHdo19y;iA>%&(T(*J#mPyPRr-PBfCbWQJgq!3{XpeK?{Y8s0*wL0 zfZ?HJTfN3{%t5#BZ*N+s5l#(-FN5(PeqIncwiTrZmJ@E<-I3k6eB1uXul*d#? z@08MPell;*K5<$AZ_NJWpZ!Jq{LlU=r8iU=hdzCfc2S<3ihyRAih|yBz^Z@BCb20S zdXncDi$FV{R}D3xt-pkCbT2(YxzHc@BP`$;gJYn2LQd16Lrp*z)>Ta;IS!ggct?bU z6aOO1l`rBT90rfebvbm1;}N}Qmki1PHX*)4>5WV-XKfitOyx+aZXF2It2`b$)zB^Q z5l4nI<%3;47zv?Q99#srKX>Nef?6KBD5Xd!?ayzT<_geLz$3u z&GFCl0FPuHVJ!i~*gKaYEw>I8*5`?dNnx0AS*Z6fec$Svcf3(N0K<`y;AwegJj?tZ z&`9NG$Q&14V!!xtkg6-Mypcza=QswuJopmAlZTjg&;o!D!xw!qfL`j)gI=!?m(z)e zBs}~gfk96|M%ZkjC@`!tef#RVEq~<4ZQ=X{`31D$lg6d6#GxOJ0gz1lpO}!s8Omt5 z^M>VRG^NX|)XllDP36lH3TmjTo2G=~!7%3_iynCvwsjRvUo=M$FjN_iuok@>xCNoc z0*cW{U|L{v*zZ~;e`w1yQV`0T53AB+ELwzlz7Q%10;h}eVWGM~ zfikTDFquwbX+xVR4hadW4m^kT@Q2rM=ot>|!+Yo(6lt6&uup0tb>&9Bh+xc^Krk&#j$ES-x1Zgc=TtojIGVaOQn7 zC-5PVkW^cq5HLee4F&X=J}@q(_h|#lgHJP(o&X520}NTB{K!H&LU2O>hSJ2CCf~>& zitX*?priZ^*& z-8HLDR~1%(MpT`wmhIW6pR|vB;^TgB5+i)XL62xo?R{x^*;=iZ1G~w}loeEV^2pX} z$O`HSa75b#Xi;|Bm*+kDIUrS-Vj*X9iwgqPObdv47n!ZN1}*FDZv|O9Y}=^0Zo}@D zvCgMhs#tY#+4U;>=K|aSylja{XH+s*5+8&GbK{(*sk#i2iz7ta7$`z>PN4IJdUCa$ zEo;~JT$VZSDBs8f@*nsTaLGkrSoa*nvAbQjfAWw1mHpG-`aMFpJf{9;N;&(`Q|vU7 zw4eOBU$oEs_#am~7%~UGAPrerkdcwiN&gEYc=YZ<{?Hv%wAZz0<1z{ethm|k(x)&& zPtgaN8;>-J7B z>QemS@sU}j2c0lP=;03eMMtRqx^2;(d~{W|l=V7&fnot&)hX+n7*TBO*K9ao&bIQc zMmv)6lMzthsA%-b6OTRR`F_9L2eN=}?fUoh7m-dSg^F?}jhwTbh|Q16;Z695Qho3v ztPdwg@pzrfdtfRxIK7BP-h(T@pcMS4Km0s5y7Bblj88y<0oQd7VeB;*bb7r`}g|q>yYY|4JLP^4vMNae_@*DGQ!LUkw z`xVj@Ga?>5k4lOBg!kbcE--qah8O!G=Oyq)tZM{<;v>zQg=+IzFtW za05UdLP7#4{s!HgRthT$`8AH89tn?AKX#MIu^o<-%Ukz%ta{?SRp*xkN+qP^yrIiq zY++IjISIE6o6ZY7zWvSLwEe5k+i>f)HP>%hd-bZNx|?=txnv9TWvk}LZMtN(To@PV zKC-zyJ<&%toip8!*?fU{T!&Uj9Vs6p%T63xdR#(<22Cjh{?elVWO2fl=dxDHj$2BZ zATJ0*ENcV~J)a1Rw3Hf)BmjdP7D5kc{G4NfM84qfnERzbe!xXdLbldP3aHatL71GM z8(*E7wA*W4yT4JlHU0MM1M3LLvjyEzGH271i&m zpAx;&vcl6@S>ssb{!TyT55vr-=ui%o$&cj}tpTn4sDDoX+uZNjmFt_{Rw=dpFD)G;XSgw zb^9e*qx)7aPx-#&2-J+)kSl4m7csg(Z^}UnGLHilXm>6X1Tb0bw0Q2K)tW7vo|&?| z!ZI2~{4!f5%LEjpKYXDt`{)9SgmqaYmds${k9!9|CYyJ_)D{S3XdBSzAVUa@Tw`eC z{LmtSjby%PNy&jsv1kc;u9a?HK#?f}Z2O2*rZ<~Kfg^fQ+ig4ScN_!_8#{J)ecLwo z_N`W{+3wc1oj9>#KmAibWgq#}Cmdh`hH<_ljmpX*I?9bM0cbNNa4689mV7~G`{HV* z5Gt?89rOGO65WpmcGwkoVh>>X z%g|Y%uPTSBjq|wDaWIMJiy{AT=m-uFN3X3LEAqq7 zfu{Qp&xdr5B!ZkF7~j{|cI<^$E_ob;CBErm&Sqz)ZEj}4p8NEt9rPoEvz)G1E?A*d zRG!2mksS7pCHRH#(AzVtG<5L=nYzY)$UD8>@=R$%`E&;@eD1aQg|rw|0Q7ngWUIo{ ziNP*M_=U&lY-CGWI$T%})2ubE-#M^ui`_q{3-)SG*!ta7+g#rih^^cD>Y7yMz{3ET z*Y4i4J2&65+c)2~x2|4SJ@lmyv#OsXd*NFz+S}Lf+MT!V+to{N$|&nuK}IC`zw`E8 zd;7*Md;8{X+uvQck=kIjk`vDkZD#I-a{d0f52R=K#ee-vevOk3f;-|&oZdg2sdr~F z_UUTp9l>De2VWm{p(#FaLHv`pUr3U_4*(&cGiFX21HV+>dlY>dgcys8W+*1fo|J^q zA0Of-oC5+erv5o2%Kp8`^sLR5ZHP#i>p;zIcWurqq*-hs6U#{!TT6f0mz1SWQ>kw0GO zk8UuQ7+G;%CA+%!ucEi#=?~t+J$xUYoag*)^dtXXNAGmU;oK1>IQZGy3ir|b_?a|< zVv*CU5J$)lK7(S99q~MT4bFW(^ulu+J$v0m*LcJRAK@(yfBd5Nx&*`XLAdBvF1~-@ z&!1use$}`5mUwj&O7bu33&)uLkP;ZBt|evcP=05YPk4G0lCxtePC5&8g`Xy? zQw~aYx3_`4{+>8p&IrI3E0fmR-}i!&O=J}9GLsF4_JE#DrO8w#FTXb@rA<$;@)IePj2Xet&4~R@2&&C2Z?>D6lryk5s3b|KmGH* zyFr&@ikr;+WJp_bO6k)zGdDf$qh+;v-R|GHXK%dzhQ0mvZOP@5Zy!bvZ%H7pTrT^O z%IsI0%jc~+J!7S6)s1w|qw)@Y*oJOefN?^!0Qe)XlN0FHo;7yvTjHQ8VBY{@ zJ%#C6TIHKttRj&7aFl$x>N;k3b4xNsvawXP{M3>qb5+HcR9a!$9b^lKe2{$L3wpCV z?S1b_73F=e;#Pgn^MiAmNBPwsJ&&pOq|Zm}6g1)6aWbXswO8J<|NAffvTez*n3*bC zL1nIp9%YsJv!D5#eddq6rOq6Y^%mPBYFomfg8?$10VI zl}kB0eeS&T;P%#*%j~t)Rj1AAvuAB_`J{NJZ>x81OYR@ql}ndwc1AQ*na`X*FL{+v zy*D6=t=_#WqoL)!$xlA@wCe#5wgLpd_VNpEFtH8X<4-*1(<&K7V0*Z$uU@uEc%xWQ zxC^?{VZ^i!88L%3JA0yPz1qOm?{C(*=Ev-aMG_b#W71~!>X z*~)y?9(&>uE6+`P4+~p_4PhgC6{vI38G*~Z*Kr97Umbl6(D=hWT*E|v$S8`TbcE{N z7*bt~5*7c3fA3)n0Y&FPPiJDi$BW0I>jY^~7QGm$>}#^3>$RHVl`)adTd_JVdSGoQ zC42@FYCEYwR`s7}&1bKu>g~&RDrVAtQ=a?uC3t zR<#sSF{k&+zuEx#8$V>Ri*hV*-RuyR^ibfZlUqlkLTNwL28*1rAyfBd@=>Md)u zyQtKmJhp?*U-m>yIR7={OiY8QF61?OGG z(Un!5Or@QhC~v-yO+|amfY%>U5Q|4ei?H%hss3Z1vu% z-MW3pzViYjy{aS83f%-S&5IVS@1Pfyg;8aJ{V9RovOw9)+`KI>ov^2$xM)v2`=m|J zl&vdS!?b~+>IP*t*59|z?y4nACDE>6wL4dA|L&_cY#rFlQr;F%&e*6Afb4khZ*5m6nWSZ95Iu7KqA#4Aotd%? z=>fp!#>SQ{F3vlzedJT0c3x}97-97I_19k)fR@e~PuY)t@;R3e7&Ln_lwN)1bph;> zFRXm(=_kB*IiWb1n|uB0HQV2>TT^j7`sjH(d2(5}foNL|2BW<6((Ar=J-mJP^a+(S=<`~wVQ<}DwaH4=S^~Bwm*yQ;fO}{a8@qL81zAmH zKe@apdW#-168gh|wVU{X?p>Rdkpq}EeWJ# z&JM)LsZ35vpKV!kvWg}6Bp6_v)?nm9C?zJxvdb0A4hmLpy=9wg_ig^fY2SZwZ+}nW z7`_3BC3;S3s`=s9p#mSBv`1ykPdMW z-oX~}(a;r7TUMPndeMPBbsd!APYeej_nt@<91s=ef+`)D{e->c}Wd^X@0@ZojhTuE}pSxKK!&A@RaBx>vnoU zLQLhBcK~bXom6-=-X6W7de>E+yQ(inEfMkz!O<8864pm*yorOJ?+B3+a;LX|pD$QZ z`fOmI6)p3r0MOu!LN1=_pm*^Vt6zpbez#qFSKDTP>+k$CyRo(-CBYu0N^N|oIHhQE1)t-`eLk9=kycv( z1Td5fZJp=!gGSyn58vyGqqW73`VD9OlLx)fAN<1ap$-&}-{7EThzGx}lsrz}EAKGg zJ6!zCl|0BPF0OGN;+TVr{Ri1Wx-l?@HX$GULRztmdiV->#xEh94&;$%eUp82e6Ne& zd${8?^jdUww$nR`-V`!IAGviM0l_vT2 z7vXK}pce7QY4}3UC|$}CKyLWzDe+C^e%QJ1D2zK#D}?w1?qP4x>DcLm{s&qSUG$6} zrzd6hPw#M1t-E7{CiumpfDCgGS`{xqg?)Dch84DDQ+t2?<(C9v z?!u-5pmn=_`?kIE+G~mz`NJaM2mEwQa=DD?Qxxs!Db7od(0jSKvSL5-nIEyop15G= zE?y9QCjbq@+w zS$@hY^XHXVS#p;X)FkJIosbXm6HFwhTkP180lM^LKmY=pd)xZ$svJR%;V@3n4HN}g zfd1v)_uy5mzCJ}V{cB%+!G8O9f8Y0;X20NKHfQJ0owrZ^=#Sd@(@XL@?mfLPz4)qC zWpLFSO`Bhs7cj1R`(aH|SBA@#H?NDH(n|uf&wS`#a|ukZ9!*}*lPC*F%KzMun#ZT8C2_sX&BrcPfND?_Kul-=K4 zlm20wxuRV>le3RMRX{r3A^~E?r$RJE47*r|r2-65s5@aP={%KjFfS!?< zt1Dr__GnuYa7|fH%-?E>;pnZPd<^}w00B!qfW9EaI6VdsjPPKB`kPlTSyzDgQ=fRw zuB$Qp`mg`Gz(~veGc5roVTy*Itma@fG)$=F3p${Io?|-FdtJ??J)c=M7B=^$Cy(BQ zAf%PH-irXojUMF(lhfF7VMQ$qLJ~f-aergSuD|uR3vzaPAn(wXH0c4!=Q0vf^EM}8 zwID0<hQMD(b#sK^)-y3A6weWW`6z4Meh~V?!8Ya#BUCOahXPxaA>{Z+(NuQIm>KJN z8BsuNaPB8TiB1T>TxQ_PFC~Wat%BsAg#l~_gOY-BK|zT>MRNpG?=VeKD2%>|QK1oF zLcS9cpzsZ+3?UdD*o$8$GLx2(ft3=iM)JQet9)(szBMIidWpP*kYv1!v{W`}#iD>| zsc3~XJ>UnTi)duwSumIYKDBh2N#}H-e@lf zvR&DafAepD(Z2EeO&fGuzK|%Hq5Lvnq!6q60fOnd4!y*HVK%c5!?#Xfh4J?5H0BCW z3@)^9IocSs5YHj}0?5(5*Nx&4-x441`GgD4geOj731OlbBZr5FpM#5hKjiaWVIKaX zTQ7_5{fllwO$T4(Pa@$s1b+RU^5XAQes4Q`(~q`F91miQo<$d~UIhK2^x<9bi3^_s z4g4BjA3NUD{+xbF$o=!r+f|?!FEMs7bp4rEeD{Y?2kt4(UwrG2;UNHS!sx%}kNCOg zweEH0t>+*L@6S!Lp|(b+3$|{6$e?t5gjz+Pn-+yxqc`PdW@kVfjw^j z&mYlOJdaGtC&Yhx?vD$scJ?iK@9X}ZcARCjMf?ZQAQn0ph=C|sACS$kI^I)uxzLxq zXE%hNWD=t<-2w7m;FtELKt}=|EWpc2rp#BTY$y<%&X)biW+@Zzu*s`UacsNOi+n0A zwwS``Mm9Nrkrz51Brn3T$h7;eUzyj>d%n{V4M$5b<19V?G@{F+tR;DYY6ml>fV(r zS80*f!WdWn>1cr7&NGp7T_GCKyY9eV@h_iTvDvvLt4c?l6fj87OkG9|eR zI2UM(wk{)AjtlgIk)Sb~<{k7o{jH&QqE~&_>bpD6XUJ}Ei^w8x&kCa}$3+vE-94Dr zK_?SMy>j`kef6tfx7ROSaXuY$Y7>LjEn!6C%?A7sXS+-^F(hM z35-&(WhuS-8=E`we&9Q-FggLI8PDeJ#Obs4%1f`>xO51Bm}zxB-&wfoi6@@2*DhbS zQmSWv_Gd2IWMR+lU%F|notpL9U7zxUp#gsm#H-C39dy$3td|k3|AYV6HRql8%Y7hV z`1`;0xAiwPmH?yxbk3j7w8$CwHU3WJ!z{x?frpU*VOO37au5Pi^&V$m0tf+*H6P~Q zNFWX6U-bfzYRlp(mkRdjk3V6(`o0}Z6s#~aZU7 zowbx2Us?^hC%`+P=MpA!(v4Y8z<^=v)%*8tYgdgGK%?)TAHXzx=L{mg)PoBO2`=yL z1PCE`)%b6}b=@AjaL(??+Wwcn@kMW-iIjvT<(IhUD941)e}YhojEMCGA;`R71Q^o= zBx_|A=ItX-U9g4ustuDBfuVxsR1W5R=JOdBW(-5Kn+4IpkQ}|p+%ji$%BS~})soK> zrjd0z2$u-~|1J3q6Y%K~WH_E38(0hnz-H(sqcHTqBb-r|(}l7c=#0(GEvTIcNXmlc zM_v(*fJg@w+^eBNy8|@}kw)c5c!y~LIPbOAGwF+x$`idN@+*PCa6UbfC^|qaLwwXH zi$NNWixrkUQ9l4weQz^QM(;_}Ep(KHXbN3Ce*K*g$ytT~z^2!Z^)SADlF|c2)0=~Q z@gc&D!a^I$BAx*_h9OM>WtBZ8P|gtLn1nS83RzflkSO}R;qg>fwp&_dm~aCMrMIzh z*J{l%s~_a7t!jb4F=#3R#~Fs?B*0Rc1Op#RM@mfe+N^cdLnUpg07JWozmy;p)^&!m2w zCPfberyfxUE@1UT@ww9@b&49|KKkR1w`_YCk8&JOl~6pcfIJ@fK%QZDF8px^$RUR4 z1#Ka|mkWvuAASk#C3f8ET?}4>bDwhI*}*(dFu=PUZ~X1?H`=$-ItFlh*1xc4Cw?(d z!+T!D;rVtP{-}3BV^v?CZ=&K;Qgzz3m zbgzsA&=NwB!}z@ii2Tt`{CyN}d?ggYX*VHT`V(c(la>|m=^aU4>OS9GwQTK@hvA3(VG2aI z+wrsjiIjwV(Y}4pT{)&v1mFkl>7am@k0dYi0w%j^qXNy8W8%=BJhNaY=hf!&ib((p z(hd_@tBvV*$Q)gh9#PsZCjei_w6OZ0_~8c!3IZ?u=neKx3+Y8qS9wHkqj$&`S9Z|x z5kY}e@8KoW@=trqOMrdsLr>b`(xM_yO8*LMN!B#GeQQY0H#_VbY*v*4&9u4u_qPPZ z-gfYoQG1_e{;%X-vr)Gy9hc67IDhBl_~9F0uJ6I;1;Tbz)(BfZNGAk+nG zP$XDIJeLdP|Byj{rK|Qv{UTc_<=$S;ZU_i}^Sdwl5zQ<}X8!d@pZl17M!17WkcB{j zH*eh1H+EMT+SyZ!;vW|0D;E84i5W&mDO=xI7ayi=WobcS2Cl>E($5XHVO9OiRSQ-W z)6kg~7G#bqEL+oYWOCP*7w4RpI(mm5D$lOi>YZES>$Z=6Fj|Oy1G~&GpR^m--n4qX zVKdc9)h+pC%h4k{eeMx^`>oq{@BTeIcY4L^6`U|85IwuNY*#N`^`oLsEzSF4$1WI2 zv^jU-qP_O&t5!JLu^<1)l#Lp9q__Ka*kz8hbepgdBh3Thu1;HcsG(=EG$R+3=dXV8 z29@`|xeo*k|M{Q(tTzS{M#%G^;`=>)&?*rn^M z?r*;iI17SVvOoYuLQ4vpGs0QXIWseDOAB*$VWlD~dCv+b9=EYv)p{d=IT79!lUhzL zopnJAaPgr}K$w_ao-oTjH?EjqY!smUb_B!;4+HKp_eFvbE6^7mP&@>c8XSU{649&R z!pgzCS!J?fOD9j*`n}tBfAy|idh?!r=}TX7fQlnEX!367!aK^(@w3up`0g_w|FC`H zqt6InG0$1{uf8$KJu6^_5CaUb$`~L{I8Q_Qis{aD0~OkPJ?ZIYo)U9%*;{r**SZ)s zrN+RX$;`2)k@gRQ!r&kOSyjxm6o&7TxvT_h$>!!}tU5KLhLp4KedlGXwVOVi&khV6 zQ^{frdPInaJo``~KwUx?E0*lVmOaG958%|nm<$KDEgM&!iG#f{ONu)v80O*9V@HoK zU<2?5fCbQ#mU69odSOs-09nH5i;{Bcc#{6pKkuCp-%zK_#fx_tfo2{?3dd=oA4Jt( zuOQ!bK|+`ld{lF=j|)Pj=JE1O`uM?sqE0db&r_AW_-(?5sgkvZN%7o~wIx*c8(p{R znY)&B5G5g|#Eu4CmFdXJ<(v#5c9js6SN2#DzqjA@;WCEy;K9H@Oev8fFJ$fZy)FBD zzxoy1ZT7@VqMK;QZWfc3NuPE>FBm;pY~h5$_8xP}3EjMW(1_ydCjPl|Na1%3IvyNK z5Q;}o^1MeG(Dn#TSPUfIqqAZj_c(Oxv3R`DiEr_D!lJbF!vV5*9Sn9=Qw}w zcGMq1;<$+}`1QD9KEIdqOhwQBDawQHkpXeZy*&P%t8;(@=)36iaN42l z(9|Eov%dK+ik$iUUHOrOQ5jP<=+&*Tf5i&pYA=cputuHG`wOTb7l2J-b|9y?g7K?5 z@^*bBS&=k`u>nOWaby`swC~kBUha>dO4yHmSncAlX?+2cRfnus0oVX~UGC6HpeuF2RF0wY;JhMPzzz>=xdsq!H9EGp zyX&J-bQUbm&DqcX^r!9o$tfFl8cru)+o3vb^m^8ozS!L9T5XWG-FDrl{q5~F#Rwzm zkF<9W1s!_qxRSS)6d)d%%t;V|1e|uxu;c6Y3tYTT6^!FC7O3^EGs?1g2YtC64_-- zl}<^Y6l~brlPqq#Jf#-9ntel-2jLKU(-IR2ll*gHiq{6_g=TJe(kH`i35A&{G;}TpZ=VK zU-$=qOkc!+S-rPmi}N#9D&<^X>XT2a?DmeVCYtSRZQ7$3&l|gLupd4u^T3Xapb=XbBj3ibdgzl{QlZs z{mb@${40Ob{^76urpNhD{?)%x#b@o*i4(ThIkLweyJ(Ne@~zG+Stg&gHd~1ZCG6k2 zWSQ1&0nk|+Wu`4PbK26=%K{K;AYz&!QydhZQ$>ILq%8p_gdS7E5qJQI1LYIToN^p- znw&sg!bUDon)Q8e_#E|!bHU}HnqF(l${wx)Qz|EdHk-`|u+Qqz*rW0DYyDw;2Jw^9xh9D4;klr7=JdsgYq>PULfzmY<#)G($!76f0^k6ACUU zPJ|`p1-Ph=MOWxRgY;p4J&y@kBj6b_3#+x+yOX_wBcsB@5Qt#Z)R^Vi&lshuXQopi zOoQ?svqBPoOf3ntc0sCZP1ZVl3j5~dw1+;*51>7=%6P}luS{BUP`9oGNvk`uMrUL- z3Ei>stj%9|Oal7QYU}rT1FcB)|^tUzdg0w;qZiTe6?}`Ok~~^h9S> z!4s~ScDFZtK}3DKA%M`dR&T-j#TiceF(9MaS-6Qc{dBHNJDe>Nfz_55QDLN^R zKu=sKr$4FulrMbKPgj~1U-_b|_lo!TUNLRl&tp0_t@-Mn_?;ywXD5ncOm%)1|0mY|hGQ<;k;;YJt zY#_XRINbuR$UyK)c>hjVFGnb)`-(=PD7%ABcM8Ql7yqW8(g>iKA#(4EbFe`9i3gt6 zALU2%z@GxgfH=@T&WT@+>y<0d-kZyN{BuuueM2@;z7R*qxko|TG5yK=Lw;C2Fp`{_ zIH+4~?Je>99a}6WY-+0N^3-W7e{?92sEt>3NN&dkTHzxbT%1ayi| zd+>Uv?98E*?oivu=nWLqkEt*EhrA46IQV5gaJ$j6-MzZ4uif`ybNm#tX`2^l{lv3R z+h6*{&)Gp^-BQvUbTE1Ks(h5GwzqCuJ59T?m9pys!EC2?*d5yLe$zXcF1_)F?bm8H zGd<-~r!tbqzVKW5rUQbW@t*W3psGAM>CeSVL1pOK=H`~&y?0;v+O)UdzGa*D*R3p} z_|rf7A^Qt|=F@g!NiwU|6j16*&*UsyUY6cXSf*T229uU6E=!NKt<$)rJWD^Pm#i@R zgvvj%y|tGG&}y_w>79Ts0QBgGPNFxjT^ElsO*3sz31l7!43dwoK>7Dyc-79IJ!3Dw`nrAS znJ4^U5&$UmbocgMH#Wc&AO7$&GL(+o$U+x?`};5ZXdOJy&ORqrmb|B#eW8i_ErkX6 zvMWfbRPuJqoaycDZ2{q?!Z06xWRsC@N9J@R%gqWyb6xny_m>{w<(dQi=gk#%JLP(YZrJQ!e1PI$Vty^hstic7RD z7t{9FfA8Odo?!%4oU9ADKSCDphX4&g$?QM)g)i9O`tSaV1X0nSUwr;0fv!C(Ei77U z>YRmaQBXN?|s*IMwlorS#I&7B`S+nS3^$-3{x=-M_`=| z)vU0>uu`e~ES3y-GC6IT zTtW8$HH57inS>NW<4el(?&gkt?JHlCP*FZdJ#Q!|dGEm&jp%)3@fb!xE?4yx)kRss z#d6+PMaWHA@vm^q-*#h$ICY_? z)Tgh&gUV|))v$uXjZ5*4jUC8<8Tq`(yLW5$-B+*MM5$s+r_b5S%8Cq*qOEPOTWz!E zbj%&@+v#l0Y5B&R8+LP}Wz$c5*nt9vy;Q4HHZ?OX#Z{0&)3-74+hEYOt*wS_@2uIu z`npZqzW1uXcw^soj;1X;cg~KaRK^8_DIfFT$h-GS$#8{l^eEZ{kPK~b-0QFZ=%K@a z!uSaaUUZDZ;RCQETJq?|q-d&I&@-|l#0A}%&&S@F?orNViGI+|74#|cN@ z6pucMf1qCg!w>smU?!wb+R!I-DiAwqhZd|S8E(i96tK9;`5nI=i2Nx$d=D6K#*5qo znkawVN9|q-dj93ZWfx$Ce=#8S=}5{~?2t<_ZJa)O7ndU#3gn-(2+xmKpWbH&R@Ag5bzSZC+9^=tPnDIIbC)QYcl z0N@e?x-bpN<(L}^o>fl+`tlCHOkqopk8Dn0ZmO8KMr&mM;+t3PPD|l)^p;Q0Oxom>egagiLD&!gyMN=Zjk??RXa3ZO?fm%#>kFVa zw{J;T_Po4@<3$_GS8ZzLaZBe#ce@utqtbnkfydTvq(?~>b-O~v%Jdn<*?fcyDl=Mp~g@HGap4$|a z$83`6A^ADg)6jQ3;GiTHdm(2MsiI{IWg8|+_T3lXv@d=6H*7RuRMqU+4?Sz2{KQAR ze$Zv;7l0{OI>(mg=crPrV^28ZmJEiB_-ju_2P3D2VwiGBxzOv)M$PJtwomb-X@=2j z8D!#Z7NEoXEU1T%nQP25vX&DIkw^3-R6s<^ zABjni*xr^JqZn_T-ed_Ggs{Yl?d&%MuzGHx`?=$4WJf4}m>EHek`M#Y<4O+|<>OQ$ z2*MC|Ceao0Lg|euZ8cB{D+QHb8jfyMWxjKOg3wD2AjrxhWf}uk;n~L@vFiK@-{wuy zP#}(>a1lvC5lUF^*9C;zowf^7Kblf$!3y-!1439a3bQ}{%(D&#SP-Hce?O5B&829o zi?g=3x?wMW>-R0GM#v2Vo&1!;#}^^b%qy4i*t88Uz?+n-&lzZsvzk9{>u31-$^&199?9FP1L| zkN|bVMM8;sm3R>7Zh0LspNe5YtXb$lI5!;RpZ$rMzeQarL7SSI7SJ!tAfZP`#gGzU z#{^fQii@63>YBxCW0KS?;B*z2$toX#4h|jwZzx@*L0T6kQ}*b&SxXNZDnm-(QSViz z-A31b_Zu(Ro%L-A%Zg1`r|rbaleT#Jlpjr+Po(YO{;QVku8Wojwzb!_Z+-8E-EC3F z2`RFi58dU9lQy%k;tH=OgX8>TkJ|L|vRQk}wywQwxiNuc(f-n%ntkK)+qQV-ygl;x z6E-s?B~>Z=6pAsG4?6Abx9x>jUY9YrtM({`b2x4-77D73pnNby81-QYx7Wq+2?_;e z;$TvILOr_y$fy8(Aip?2jZx`1SQ2f*E*}VJ32w(b0-o?mptyKdfJQi-9ueLk?5T5r zBYuzzp)O;9p<@kt{IAQo-`r0{MCUHL4U z;SIjxsBz}es|EQE0Y^vCVlbtJA|xG;QL>Dk6DT=3=Q|EsLO*!6+mSL8@FI^v2`XLc z$xnt9-QZW`oDyeojgRW--&j)Vpm?cQ=<75H3Y7F1&48BUv_<*xW9{{0G=_Ltc*gwC zLzKGmLM4ynpNwPV9|=qz0&sNiU>hqz`S3#)ww0K_DjVWAl(=51~7$#~rn@CtK^0i4BBL2|QiRT;}^$!9t$lEa=| zT%Nbjf9jO|YrkMp9K*vnP`)UU)mh*Y5v}s11+jVQVYS!<_aFA{hp>=;CU81&~mp(0I1>*+R z-n{J{Gu7E8%T?z?=hj%x>dl_*?kYTUUez9s#PhArn7wgn)n0w=W#y-1yL(LmMA2HU zmfaqLkvI&!%uT@S$o9GbS2{?75e-1xsW-)AW711uN}SJ=7QM5a(xx_oEZ7wwjvYKn z+eW6vU{(M9J_fh(m2b!nms1b<3-Q2Q$P)ZVZA%__4&j7RREH;LC+&&H&f8-bPgy>% zdOWi%-SL=hsttVS`4{c}>ODKVSdk%d-^%H6TbwG`)Lhk1O>7CwGsTbHmFSo}eeRqL zo(q;K&s%YfsoApLy>;ofHCOk1apnQSBr9dDhBcBP7!#mq zZLSKK$y%K*+mw`ku~fC0xf#pMR&D2g-EO`7y7l*^03=-L$z!f2TVFl?lRx7dq>D9F zSxM_On_@@0{+kjC0#3Wz*1UVuMvXnal~9o~qGy@j3Bs~{=Q@fB0i{A7> zia^>sDIHa(3RD z`x3YU!Z83OO+XimDPH;hi&AneS)@}EoPZfARDn9?1|#5@xRaBGe_j?X3nH-o7(zo> zGdDGr&RQ;?vAuh%ws!rB-Pzo;J8S!1RvHq3FVv+3tm-hO{8uNd0>}XNw9U=W+ayDi zq6JF9orE$CoS`%}1!vVViV3iZ!uQ@=UFnTfAn`)R<`?Jes$&USKr{-S;Ug(*8b8~i^<;qfvCAVs7oaw* zN@i#YzChujJQxyVnhj;HihhrrnzhdUj;(L*+J3z)A=t88_c!dmc%IXyo_z8NyZFc> zki|B(HZ7Me+N>1fPyOgqR*)6WVI*Jq=8N__-+I+<-n!+BNh<=eETmv(3hHA_3h0HG zUUr2_U7T5%u_sS2SV2_@y!q0L*X{bozVk9Y@hdCK4yc(jLLD59ZmMkx z@CqQaM`cIG1$$5T#itC@jjKIA`P4HuJu~ORmNum%_%FJ{3#K+Hp(w!E_VyL#V=~pq zjRgtus*Dfs0qHc=wg4oendnA4qgQ28#sQ0nIQb8dM}46n`G;&NtG?JK>px zq3jm&Ro~;3wHy9T|JYwW&|+j$(^F30t_&g4k7Ywp*hjuTfR*qy@wLzU);sc#FKAe; z&glKX>M9?-qGm?RGiH05a#_Zc~jR{O2 z+Q~}8{>R_EXJ7z@yf4hixcdf*j_9DKsQ$@Fw%EVA~-!LWSN&S+muW5@`oo?b}U(@(+c zNo%c%cKbI3$d?7+7i?T|3{afRmEGVQHP)=Peno()=Em%V0PuKj-mG-SQhDis6l+d6 zd5*pQ(VY`&4+nuKls^t#av30AAg_!RbeenCsmqW66zhlK?#7eqYMh;{4!Q!&`>ODw z&73%E`ROw<8uTpe!SoN}fAh*6yLG2-V`=H>{Ime@z`pd`zim6)n?6E!?be3x0ZiH9 zC7Rl7x#*n&vg)kX4s+I-(uW>kgtOV|3ox@~Ul`4dny!y5(w@X#Zuq#~paSHI*|aZ$ zL{>S^>W7SS_mI+ERv#u6E0mn4kOy?Zz8`QQ7JHTO16c8_#b#qz>(rO8=8 z!y#Rq5?LoLTM!ly?RxY{$Y3CEk39OE=kNV;9|#!!AOFk$*a;r24l0a3McGXvCV-2T z>*a!!v=lYU763LbMP1us!}6xx6Hu8dXJp+U+2c>0_c*&!6gwN6Hj_WFwaYi`^{-#I z1Id%J0PgI$$8DG{*-@`0Abi(4TlZ~DRA*7qcy_|Z^C|ntC!Vp>A319qw|DKz*PplM z`kob*mTmFr&-%Puj%h3|owTWic{R4s8_SW9d)r&ioTGHnGE&Ibzw&SF+E;$ll80S2 z0yV`1);+y9C@`2zj4Pgm4^vF=*9|(3@-sg)&1q!I=WJ&Al#S0W$TH2@Pym8?&seDm z2`?lMO8%~__gbqZOG?(3B5g`g0)~I&r+>od&$eZCGT8$G3$2p^7W7PZq+FOYd}6+0 z&pghyE>fsFI|9Q!351?!ZlbWPU%~3i_l_DW@dI|)8V4`}prse6arcgOch~GNQ?{0@ za#u+D7gp2r3k}Ps4=6W&mOlblH1Iik0{Y|+K;%kRAe!E3R<&X-ARM|ZWKlfy&asmL zJw7O@s+9HVb7z&8qzwc}kJOL{hp7;k(&i}9zLc!bLx#-)*J?3Z}ZYse7bFp#o< z@Vpv%QT5!kev<<-LZhY^fwT!u{t!TZRb0%|Vyib6si5$oO{F?zNeKji3q2&LcDD0l zw-6Ou0IVQEFzHG}(Yo@Lcyus~FaiWYUpC5T(ICREBQR3mShw~2_pK=f2rU6A9Ba9U za#TJPp-oE2EH2FnobAcrsoS{9{+Isr=j@TwC#)`o_8b4^TlV7ByHY%l*&`P(*!}x! zcIWPW3GSNh$@rrimFY1DQm(7A>gf@^cyh(&WOxo$hp*n*usQK1AStE#r5>nfeG*NW zwUcw;@naV^u5=%a%s8Qcn88M2mitTl6pRJ3QJvf*rv&+X~i84fga4&F9@t`Z@>ex z=u~Dl_y?>*WfsXXq|d&xsc~d)+T$f0J%2fYV|r%m;z9DGEGUWqDyWB4CU4W@JifN#)kc)7w%h8@v#HZV{<9{tDjd|Ccr-~U8a9*1Ig|)T`4HmvD|q5l5J+5 zvHiJ^c=@S@&>J@{z7GdGHES)~+}v=v=dwJRuqn~+1H%8j1KdNA2BBK}pfQWo z%q}n4ihwX9aL^fp5@U)H9Aq|mr|psv>YIE*L*DyIn#z$I`?{k(65}!$^W%1UY0Rc8 zoa#5QVPo4njIxz;mMopHp5z8|g#lcB$?JprFWcVgW$6VdDG)Dto-WKbIBa6LuSh{-w)z zj)5+G4#p`7H?%|-)`5^-;B7{GCsf{q_ZZU|$edhsX#AbspqSbbkDO8-;taeWWEc+r zp{<8!T%5kQWAQQz13&x8kK6j%s_Ju0pm)g+Zkm#g`pA=KZRvEyre~)l4~1`73j^I` zn5lxLqYlN7YWu{c_ads=$jCb`o6;xsojtF&>(}pELGdwCoSmHZdEAL~p4{jsd_Vz2 zFGh5z8g`??t(<0&-}~o25HS3A|N4LCSCj_=5>S*AA?SK{>sBC)IhXWKFrTe2g5MYD z+FV_eU|{dv1AF4avMjNCx}Nm7%9HHHH?k9px=NwH@U6?Xwb8aI0m|9)PuM6~u*^Z* z_U_)a&D&RP)El^klbtTu;>mfds^q5U8pR~mMqD?N$iTNaeMW?<5dAojKH{QG<5VK}m`}?xgWS!?Z zB3Zyo3{}+q&;P|g>wDEUYc&CzhEhcdOF+n4hW*%cav;#WvN){KS$cudAe zIj4%VmXIN`&oH+0Ilx{@UOyBr(?yO1_KqYt2K9AokCz1k^WHm0z52>oH9GH+L8uc* zIoXqPWHLVhmmZyVlg;1(-~i|_gw5WJGcz-4xSU)kekkT`etFpnYGmnrUczh4YTNq) zi*4_1loS+gj|5yc#WN!*@Sf=16Yq2+5D?5k_%57&M2&mJd%;PQhRqzd=~>a3p-PmbYt|d6kqd|cQkjau)CoW|LWIR2L+1af`Ot)1@Mqb_S5~i#7SaP8pb*(V zm^!A691;>MTk2vWHR1bOvXYk`6ngpq!w5QhLp_eM1ItN3poskd7bzin%jx~**51(JMk3M$N!S>ZVo3__Bd-6lih=+sVbcHLPWBSZEvcQCl$CeCBMvpiNlTGtcYN-37z5~vkw^^jbyh1>sgNC7Hasm+o z)?EQzz>Em!ZCnX?9GP4~yuq=+HKmzQJ3bIi=%SZlr?N0tllIZ9%ji-5fKcLA|3+Jf z4<;uoethak<%ic%Sd0#_D26g{T3WLPPzr5R8PoqD+qlloNvSnj0^>5a^bTO^fL{*Y zVQ~~I#(lnG41oo0fXAeC&g^8_KL6oq z8@0BbH>5CaoN~!0;5(B(u+@8OD$9Yr@V4Y^vt?7_=gRDi?@mA)WJ(@#kWpPmrl{}Z zsW&1Jh8uaH&N%su_KQIq>Q8b);aDdzm+jlxlR2vtnMo~Opgh^cl;x%_SfaFGZOP-L z!W6{s2LtJi)mLrj-fNGivJUDr*`mhH^> z83E{h`}X%YB{xqo7& z%d>s8>7AWjTfKct?Yr%On5k&lbiyv2Tl9`4^a6%-t~70l!i*)0RXa-NEGu0`%Q=$q zN}V7r1*2%au!g^ZH*kw*pYrSbFfRk$6P7pOyfYT(E|OiTX%2QauNcg zLEZZMw=69n!+yqq4_~PrDSFK5Y`N-li(8GlCDgcXe&hG-{?$u1k|obs^$v28#MHBD zor&-=?xpYmsxSh>SOS|;5alzETVZ<9#wt^mLJ0JRu9UU~nyy@V%fXJ%9YZLIQIAAJ zrcki2CB1e(`-{I|dyS?K%^w|xA#Wrx0t`jNd?|X4SU^Ru5Edbg5v5E%T=A=ZkTfGT z7z8FeDX`Bn@&KSM4>l&@JJ`B!yZ3L1UXxayo>v2)$6fhRUWTHBPf1Xn(l}ZY=+|%F zw3>ti^KCl<#N>^1@yCy0(+d|Kv9ssSNm*A_Z)0juIa%MTSEWfKMhF9Jno=a2YrATQ zhc4`QHt$*Luuax1E4t<-RDdTbS)rVbpF8!4WeXF^BR!P}yR-z4z_ptqG)ksU?EC3| zw{O=4AgO11o|@KeZcEVg#RIA%DZP{kf$`uQvy-6Gfo7t#^^Q2jFzX;w*xZVhs`xwW zLW|A!p)pdK9@u_Ie7>=6gAOMUiajDf#OsPpaYAlr?*dFn4W&(bPeJ#zNE&99vF-lme2VoqfQgtKtQ#TeyMeFiG1ETSFy(1(2Ght1s?irw9> z+r3SCio=k!@2sIT*{P#iWq<0T4b_WjwnB6RNNRkJ}+JE^+4$ZPyjK?gv%A?$|4up-@7Fc zN*rFdiL@JWLj7)t79_Wr64I3-g^mazxDeh5yezOQAjIiVQn3mfKsKoWHZpboSTau7D6_q`FXl1-idh!ti|sC|Zxf=;KU9HQ5hzhuefakfxW+H|-m5Bf%ZOQSoGV}z(Hec5a6 zNEWvww;7ETpQs$gJbS}8tg*jsy{3!`$t38#xzqBgY)i`m$lfq>yJ{_@}O-3WZcepMtD!WrrV0VKND*YB;%8kZ22<>QTdJma7=Q>=u_ zo#iZ4bkg$LI`rEd?wk;(%Z9nlB{uBH77ygvJ_2w&nGST$HBeu7F&-!c_ZS1?M zhLKX`qW`627%J@yJBSAF#jMM^Oqb^^B_Z?bSANGfzxQoxtnWxyiE+|mJ}Cwof+UcD zEc+Lx1kT0KYOK}M=WO!aBch3{rtz#>gsb=N+SbOdZES4&0tpty1t1rOw5Dd~>=Qrw z8OsRV2twI}gwFECM{I3#TMfJ|MwStw_z(<8R~G;hNK8FJ$q$6qpea8uk?I72MK1@! zjfMt?&^rboCJtrYhcd=g#=RR?tg&@V#=IJ)tZ?EdZ@%D6X(GH~FedyX4B~vy!$xnB z&nHs?^o9fgnDV)B4tmvS$`x7Ok3M4gQr4|f8owW~p{%g2&E|a>OnWVFL z%xlK5_q}#yAWY88TA_lqEVEvYIjiKAd6TLVDZIAw1)%79&+uDs?@3^YCdxDOAScEI zge8!fM!-S`g#D4~(R-Z4qhTJHU;y+&NXQuQQ4I+aH^fvPZ)#|I1c1^W_67D-M-tEa z9)`-G9b2}s5*-EN(>sU*IQj_rP4X? zada^3b28_LYqT@^^)WOaF2aKSt(BM%pr@r6h9y(C&5eeqNkmnRg|`aPM*^cpppkE}^#3LrBIk`=8QEde7dNmpWEozb#s9d6PTn=})p~^N)Pk@-wHTm{KBd7|CPWA36?x zV2D0nX3i_qHUf=2@8Xd#)vaUQX3gghv&fG1MJllKL0-5alaj2JOgQY1EG2+C>a~O$ z?%Iyp=$HU3cw#hOkWMd42P3;>bcru+-`^0`nY}P zw#pEuGA+!?fPy~ai%e?5YIF{82#M;XB-sz20!mr`GFi^rU;gvYd7IqY5;$%h+CVhx z4^>~Hl~2`C8*SEhjI9LI;>C*E-CQ|kg~=JwwB#LO`ASvSSu2*Ol^?M=?L~QGswTbQ z9GrG&bP5Go!rs|n5qN z&ARv0u;4~jBv4k6(vYwe*c_Jv4nvWE5%x<|^BZ;r1{yYOh2fG)C2g}y6`LxmPy+!0 zG2@s7o!YJ)OjIm8tA-+hN&`YLA_UmY0bwQ8ZR2VXhocFbJay5=1eiA7e$ytIA9?N* z#`Y+C_pYdcsFCzr$_HEJNQk^sIv9+6ovjD{Cka=D7_yV616bJf+MQI!*+Nw8v&%gwW;v|4M zJ+7E9IA~ynwx5YFE8aKmw2jAoN%>|GF=5viQh3TRw9_idjJ2|EU2g9CJVEMMxa=)#xSo zlnuZXTkG_s%r@?<2^7otkg#R*aX=`;;)epAu1w|E4@S_{8?(n8_CO|IC=8S+y;}^u zLOfNQd;kR4MWOBU(2o>1fRy@YItJgN9le7L$&KgIKJ|sgWGomWlNkwTdRv(4f`CSe zHB^TitE<-D-w~K?+l1;K0FvgQ5rm8WF>jVRb!qWkUU4xD4cKo2KE(%E?|^sqWR|I6 zXHL)gG!Bprkn2F{Fp7g<}`n=!QRs-0UuJAgXVw)rk~Jj4DaxbX)y>~N(jBF6MFtJI+Y3i zl!BLAZQL`;t3YSc_TFW@%JJP7-iUvg#)8A2e}`vvlk{FV@t5ily2nNkqXpz0BaV6L z^n#)IFdAtCOs^pxk<#0R14reL_8dSa!py6GdIn3t}25{jE+;e*jiO%jRPzrL|6 z2>^{bZY92AxR`mN7+ih|m;Au5j0ym3kt5L1fgtUasP$t?l01Z|CZZ)} zAS^WY`Q0K1GKFoNFd#f$@fu}@edxtyfhG&==+t1IJMT%;)6x(5LO$5BfnFnM=_6EJ z0T=?x$l3kf0EWo}d2<7b8d9T zx`TAuqT(ft8|6xi@^Oz`1Gu>CjnaT~eOi$CIHR(Koq@zpN{4oUk-|1#)E4{u9&}qi zl%A9hVbLVUUYP5xGG&S`1DU;nA?)*TB`4{afL{P)8RA0M>?8A%wHo11&K%>v!df#n2r7##@p zFC$Vs#%?|==JU}IN=IGjA8mn-4|u_i|Al&fDr!#gH>|Vynk5eQWYEcoXzf^S{f=Z`UD1z8UMc?Uq|Kau z#&T2A@selMH8Q~YTQs5G*+P@H;)}HagSscYcQQz}_gPOQu*V`g+A6vnu*M-y$k0L# z{dwSeaeU^4tzNxkH($Tvz0mv3w(4Joc)nug>ZEirC$AEftJ9^HV_KUC|bL!y`<2$Yhx^ zCw|Z~9Vr+A(81S7sSN?_>4>EMSaX4~{pGKG(|fq#sbmtYB|}(7SRpUHEdHLLog50c zh!zEbVBe8SbSO;C`@(Y;zA-vL9sBec<(Kkd2+^LrZBhoM>7zT~Etdg$E&dNqElTMq zWO$TYKK*ob_x`vK1PovQ{r|-Y4#Nyu`*yIsF5n_TijqM&s6ibG2pmacLcl{Lf-?kj zzN8_g=-Ggro--!U<6Kw^EFbn|l}l)FAV5AXf=hWNG8L(IgpC@dn1{tx3ZTl6AR=zB z7^+eXsYJKl&s3Zn7A6Y^IvByVt&FhqAE7#iS@chSrAy(FoR{KvkJE zw6T=;4D|ob)@7ajHfFV0qrT9S)5lR6_cys5uQ1R^g#6lR@&Q}cKbVD zvu}L)_iaXvda0VVaxv^Iz!c3w=L6e*n#R( zWU@mE?`J>%bN1-Z{DPFdgo~5~TidZ!4SDKGh{E{H-5asDV8$Q!+c%AZY(StoU$vW$4i+aeiLeu$nC z7p|lQKnr4t_8Al@wG~!92N?k$;fLNsj6{Guih(>3F9w(cRkZ-#p~z6mF(9TVLJ!_% z2}d0ej@>?}Tfha+07#z-B}bZx1V9)d6?*F3I8;`#J48M)66WbU zO#xjfDl(&*^*mEn@rmj??ER~HQe3{MRCNt-2YjTZ&?tMLU*JcVWvVaQ2pJ+D7&uN# z{Yxh4m0~v)`Gsf1tG*4LXaId^vrwFj0~YBu27uFJ7Se)GA|NZDhvIRZPw`L~H+Z2} zR35-RyS-pQ5`xH)yWU%>_(%)cOBm{jvf)sYvGZP1(VkszLcOVGbR}*Nr^LL6iTLz$ z1gA3DGF(U&>?TIx900Ex@;WQAvr1v=RiMG@kDh`Er+_3&syX^D>%7xnE%xg2W(rgG$8ysW%*1?YN0bF`~& zkLg1l*Eb19P^72W%>X%*6W=LXl|SsSPmd}zb^t6{&Xh;COY_4iq(|YQk>X_+p0xNs zp|a%i6Sg?d;UCbRsVCyaVb4ab4du0K$?2ypSw5pQWNaTLmA`!(Hf~yCup@(|X3g4t zYwfQ~4znOnT4kbWsr;;Er_cF1m%t0iBGz@#?$!2Gey?}vCIL-{3sdAUd|?)P%X>0b zTD4v4HkkvDjt;zl-c$A+OVDG?JoBSW!EWEUZ97|g-ZtCP;Ve*=Rb%Y3P?;?IbO`Xv z@Gv3z4J|7g(Gnf_k{QJdA2E8vVo2vH{rsD*@bXanQ&GH`tLgbb8PJF&88Q?^H zc9slho%^aV2^;yP!m;GoOx+tCsVm*E;73A6jl>R`z9@)Z8s7CK9H?k|2RL4lefCgb z2xS-@*wk0P$Q#UmFpy%A%w>@aG)W0uFcl)p{30p%UcG7M=&38(iI05B7C!rV zt4ZK-ScCEM9;*p+X6!`ANIwoD;!)3YahcZQ{v6O;hWbt0-&(c43>OwX<*U;noz||cUb|wqzWOB_)EjOc4^_9`e3dqPmmXjM zFfia5RJS`jh=gV9D0F>yL*QaaQN_YTzLkNa4u zx$@9|_bA7x41g7X_y-JoJ}8~uxw6zRLGPmcF$WS5M9(pfI%e8{(*yrXoVsVEC8j^1 z9zO8AA=?zcSkAw}ul`42$x)yKX0~Xn@4O2jHqhKVGjt_Rw$yU|r19$-ySQh}=^@8= z{<)g>1mzTGO5Xup@T2#}D__B26ut1{3Y8q-oqQ_~4?n|K@=yFchf$K)>5jN`msXn% z=_{PlL{ZDD(}K}}U|3RIx0SpUs{Bf4vMg3SKK%5xL@Ene9 z6@6PgH9mI3b?(%&Q-`!vN4^)o8edXvLqbG`x4^axbQVQY&%Wofyb}$E8q*gryqIf_ zTKIoFjg2=r!8T?(W}UuTY9QlTaovJlGs{{9@~-o zqi3Hj=6uHxg(W}mD)mLU0t@`a_sOFJn<)qgpPaSojPlmok-WMiddRQJ$o%;U$-L~s zhwN}-E~+=txW8p%?OPI?8`c&^=(g5u&~3`dPHFy{B5gu+B&AQ2AqfOtxHQ`#wekTl>38V^_LGAfIjm`G=o!0tDr9T6ty^wPuZ0 z&$f4JG9rR*W;X~w2x^qD$qK#0DhHivG8mKTy!36+`NKPaM0eykjri#{JaIzW6xN&(3SosqtHSjIqof6qr3ksTP2 z<#JIVU3HseM4n#ewD>_Fm@~1$sJ> zT&IF9AeR19-<;s)(mHYw9Odmp!FqP<-hqSsszES}0WT5EhYV_wL*wk#*7UpII zP;;)VTu^&mNHB_y4A}upXs*;pe&M7&^QZs342~&lA?Ia0y!davq=va`bBpsfCgDBk z4y{_5^;7fGd0F#0)rabpMKTT$mCi^2X8*0rwsYkTpD*hRjs(_De)6YGlF5n^W{Cld zQhGLEL6&F<*JI@mntQglv+e?`GR1`dqJv@UHEeJFo^5^m8&==iRFX0#6ffmx`>UkD z1t14NF`^e=3Ss`j)2h>)9kzEZRW7N)3$X9ktg z`t2`U?akLMCnEttE|ROQzvC!C3|Z}2!Jln16Hz^{F^9iLk^MsE}C zDPbI?osltX>6}gE3Ib~kE2~~~ZTAoD#@%gO-L83W$bkR>3wM&0Ia_-CIqQtrpIgQm zLXw887*To%XYvL>c9EoCxP)tX0A%Q1g+ejFOI{vDgYcxi^BwvG?*Vi?0I>wsACFEv zLOi;syig@pDt;8Q?&D_%ggi%l;}Q8K_#rGKCNXfti^~^1<9q}If25;7UOu^p(}=Gg zTHon8cLz<}ABChI_07YFlsuHaJ9zWf7V_)q@YS#Q3~9O}KK*mh9ooq-fpjzAIPHK~`H~+I!kF()&u^$h>1CzGt_0K{`)r5lO|-XAhYYo|fOxK$ zwC6thlua!xsenzZZ{M`e>TA+5Y#XMy#tN1yowAAP$5gS*)24S_daH5Q4(m5;*x9sp zV@C{8_Yn@no-Az36iz7IqD)a4*K8@Lc7qI|H`}MRL2Sy1&QPE+fP~5gmdF+gG8&31 zulxdZYk=Wx@tF)_(Vfu}90y9UXn!RAxwpM-twvjNZfISpefIZf;UUN8tHzx+Ip%vy zZyHleWS}y=jSeVy$&IC$T_VQh9e2)l@Rl4jL_XtaN#XT_c7coKh_?xtTHlU&nAq^^ z{zG`NhZ`^4Ko1-k2rNT~5qjW2Sb=;o4_XnsuA^-;SKX&IiJzGI!swy5CefF)C;|6` z1siz!s#h*#Gj4N?qN=!xgON8ri_i7*QAh>%LzH-{hsARr4uRe$_rZYSfAO~kNY!{{ z4Yu#SX^mSii`kjrEikN~Ox>vRoWv49Fme!#IAK(w7-X|9R49~5Ftc`T^@Hm|*D|MDOG zW9zQpu}?p7(yFslHak7-bL=Q2^N=VnLMYDy8HM2)OLH`ywzHr7QJeelXS}gd!7OUJ z_2yN#&N;)pCqRE7fn6?5+uXvU<;w-jG51$M4IKnu(362eIbsoGYhQIKfT?oQgE>)I z7JZdp3Dt4=BOL@j`N2S84`Eggk4cd*6q+wjNeE;m(ED~cXxTtCYu)~{T#tl|)4 z^g2w4{wPHZH=iz`-$Z5F7C!Y8);dhv9`}51zkSz_XRTiRlubVU31hCJV5tKE_!Lk8|5Go3Jhoy35Yv-z&~IAz zt(R@|))iNl&Vb%fpm}r)fBKz%ijQUjpS$qlm{9L0|qx-3v3E(c-$o^~AoA@S1IU3vR6fJywOa)hgb()K;1 z+Nj>cKQ15S@IxB$Jpdtm3-W?{6cEbQmBF!7+i`^!dR#;Lye1`nA&;)y_~~PuCePSx zA-{U&p7(m@9jYd16g~zTxDwQl_(Goilj@GVLs=M(E9p|`kS14>^<;=cEdu}-8ZNFAFqfh{sH~Hta9#;;0jNA$e954f`LGOLpeg69?tX1dwzi2 zKqGOObPr4&_z^!Cz#c~T#1|Vl;{FF=@T+tgT<4U%h4H2>%$03wwrCw>8M_GOGS=+~ zjI|F0QZPuv@G?pW8Js@xxF00a+uydAzxn%;K@)cN(TkE_<97Kw->}u)aJnj>PbKtr zuIPd+As%>|yeU0+Y+UjiU<@tLVE|-)^hUsIX6eCJea~9^ zn<}sJuP4zgJivPa|2TLE1mzE3{oYw4(&gNHdkGPWe?mPR#}{AaTyj7o z6&_iJ5ySK=-@;ZHKozevBbg+6fmzr-mFa2Bv-Y(ha>zA!0{KVmT*(JK?)=YHk1nTq z)`Pbb{n-4A(SNv{&(8P!6esI}sF!$(;iDh^5x>4)?gIhC*Z-IQ-dN2{MeN-Y7{2?8 zCAuvK7t-w!6NH8cY z8LN~lHaR(KCqME@+dWKLZFkGq_#0z^A*DfQU_0wuwz0YGn}9QC6)P5C$g#O&2OXQu zD;+(T1%8)LEc&hpl3q4-`mC+4Z#u&={UixTi)OXkckRV*e%;FBhjx0QXr*jX7I0RL z13(I?$6XKrOvkcWOXVbdiY3fbUo=uac|pRj?8=KRW?0<{N1Jl?p$TY za*Dr)kt2l4+))-XjE|2b^txvCyLRiF-?IMteK$^MsPVkt%)&{Ved5DbT0X06r7c-d zSyd;>fpsMu;TL*9>HTM_LBH9s&gF00)|;2a09fGS55+liZr)BkD`hb1+PDPnUQ>0y z^jVt_I3HJenA7G%rzlmZp!NWoM(Cgz+=?GbsC8@BU%PHwFFfxDN5o(cYJYN z2QK(=JuLdfVUQ7_r5(RI{(O9e0)7(&Z!88$J0gqw#i7_;7Ku<83+`9+4}hC@qN&g; zbxHWxFAh3*esuNdB4y<~Ogw(oCC)L=x_B5AQ*=c_`7^RmJnea*Y@S}67C?o2WF+7D z#upDA{Cb|_GT{~e(s$z0Gc*bGlON*s58*;sWTo?!`~*il#LvGdB_j5)!H3^?+{DAZ zd?^kL+6Q?j4~oOX@P{0zZVAJ4_^&^KejyGnAzbX7u0%{Iy?1KrKlcvK-{*Vt|0uYf+p?-Mfbx*zMg-`Q}${g|$ z5~PfMy#{#4*>HM^8Hdu1V{3QCV;uoANxGio1ZlEIt_)Ye9Py(MFqGKuT_{X?q#&N8 zM_u*iKuAzgMlSqwd#!D++!nAKO9`Yg?5g-ABj}lWP*2xl=@p=8C$c}oLkRX0O{R|!J2fG_C6N*@Xgb0`nl zf)321$Ip?EWKH<~$Q-mpTVfhXr&+W5&V2#nwoiS77L$48h3i1T)?d~P}*=6S)g5!d-Cgbt{=t|Bgxb++n%yrV6WS=J<*QonryR*3}u(FY-P%h zQZnv@6C_0;iPzOH{LCMZa`XPU4+IQf{`>!xm^kIbM*Fwkw$A<64b2H5>E(gUFaty( zb1twFXebzs3Zp>}rBvl7sy)<&^Md~QjsvQAPJGj?!ZHkm(hs0WvB{&tBIKfUfs z=%HpP>Og6CmFA%s5&_2Ez4^I$%TyPY(a@ucFTs)1%?^YIx3d!?R)pVH^9?W{D-4)-z)R-tCzywz7lSL6y`8xv0MdCX$;dd zWm{QU^j?Ld!?xA7c5SLUW2Zj+aeMpDJqeb!&tDVaEyD`+W?cZZCop(gpmo~D(<<~h z$5%^mqy3DUebNmO-wLhY^F34nSkUC4-B~n3_lAvLZTjpqB?Z&>+ZCn1ZZ#@&Jmq12nkrx0==*bzgOaj3- z?nhqZvc-z>$$r6NPSHA*uc*#cUyA2Q#>q&^m_>$15^VIuLu-1O4#i8`w{P3M`@2@_ zOHrua0nIEdQZnwsX?ZC-b>X}{^4xRA?kOqWM*^I7&edtFK-8CmExixlbnAb@cNe_= zKKSLH-CRPL;1@pu;L*0p3j!9o62bsZQ1FcYc-m1sI0uvD16csE<@`H0i~8_ zg2YFCc<;Tug`sf16Uc*w5{5`TFL9axAOAr9<@`POx)O)a*H@Z2ls0V^qnhDP{F0Z* zyZ0jU&U*;Ok0>Yn4_)zLP=~zZ_!geYdwA!CaVJmEPzZKxnVOgn_o+F(KzY!4bVYA^a%~vI4(C=Qtf`_-;q|hergU5hrbisWI@7 ze>vtIWIyeOA0OQ)P#B`dKkq`A5I4^K@Q-KR2S2Vw{JjU3BOJPe$6&JQ2D&K!-a{`B z_=Mh+D>%i)U;e0`n5W5bwGtI*Q=1k4B1PH*weQV)c39i7ae=&X<_@#)tg)lto_Kj6 z+1B^=d6bmQNM{5xiwTf7 zN|OD^8hTfmOUl6mNu@89N>&tRf(s zN|%%m$y3QZUJWXCAU70mvs;6Twdi9xK>@T=#Yb_ zYTNe&-WtZ%oIb}_ya}+)W-xvSP7C%+XYpL88@AA-$C)((=qT^q6use(aHKZ(`XwvA zKYIp4Mc@3O8hA)Q=oCBW!N{R2;ajg9=N;!B|0alW94>?jd_+_6_9GXuuUrU!=xasL zZD6P);RWf;dd>F5SCq}?!z)t$Av0;CbV9OtJ5!L+pa=oc0qin8M7$$?O6MG4m%}&M z{&i=2-;dz0)f;{kJ$s#V00>iWI-Ky@8%WQpPE<#X#xmtDlPy?2!%^C4pKH$I+kg0L zzZT{H{c#@%82f!ITV9H_!4 z`a=RqBan~ z@o)OkQ#v`f?7La8Ac(n|Edi0{-nP|iE#LnYaJ6#gj6L!66IPj-vFZ6abTZ{5BlW^9USQHUzH{_SsW+m*}L+)I_ldFGt0JoiakUEgq;_aq3*0?TEAem2@J%Bm~R%(+6N2iqf-i>d_%^Q#XPtDIh` za#BMQL`mzZ-rF5DlD6QR1o*;i&em^TQ(OrvPP33gjWSktqCa(uwMipni#WaXw7h{L zoT*n72$e6z9D2}HFM94vxHj&+WovhCdjsXfw%O?^yZF&hTYhrd21fu5DI&IL0|@IL z!RNaPAZ4JW;^Ie-0>YHJx)N@^YVn$axli&?gji93d71JDcF3UQX0Gv8$Ln)CILU~Nal*+>3i2%AeFNRNYq#Xy-pD4Hw z?L`l9l+ndq0A~2-?Qvut&dKfn;Bs9b4yXrwdx+pMJjGE&VZ*(+=;5&McJGMz2fO3A z{{Q735YOEoAxohikNxS!U;H5s|3Ao^-oJN}KjcgFDJ1>CA3El}+{5*PH1$Y$_3lDL zlIYeKM8B6Ko~Z23yOHD0-|L4wfd_nwg5=teU6R!P0p=Lj%**tVIvJ^lt(v~~>`?NM zcTAl^E+PYogYd`-pU(@7h zfg3h6lec_U%CaJ(fAX}Ar-~{)Q!XU0R3`RLW`wV{`jYM6ds$`bINiq6Ggh9gNG508 zKkk50F@aRJjAY6MBY@-j4h8_kciMH2H=X9J0e~?3e6wvCJ;{O9<2iEcW^UUv?^B4v?5dSJG9TI_P2BD6d9c6>YGT1Ux6z ztj=YVwvZjSLM06Kq~(WwUA>oBVFLTo6DokLya;mGNEtk&vE|B^Dph^1CELT*WM%D1 zXt7v;Epgc9Y<_9kp1AmwEu1O?$6tu-Xm1$A>h zGHIJF=KQp~;^dU9UMv7HKFsVT#F4h&xbwz9L*#+0^2Bp8Gb5Tx0IKmccQ^fnAscmk z{_%KL_gL)+9D2v;rQYATV=sQ=Yqq|= z=KROu3DZ+E_SlC%ZZpfLya(EQQB^O8iVuLtob7z6B!fZ16~2^Xkr2R~9yo?3$s5WG z=f;;?;Cd&}N*VxTgbY2(^rmpQj0&l`Q&})r#0$yjBp=djHoKNB%*x1^_xfdSbNCi| z`{+GV?21yrQ*>be(1X^V_4e0<|D{k$^Io4=-jp9O3{YhcQ}Qt``u5i@DX%hM%1bs( z=AFhz(R_4H6+);_h`-1WhspG0NQ`9NbCMpTB^(^nTwk}BzVcPOcjunsay6*7rE<`t zLLQZNO8zixp8D7)r7%~#zI^XY>eE?-a*KDNAH1oj*ttG&KfzJ4~JcOe(Nj&l)* zj9gTP|8E_1eNf^D_dm#mI{1U`hlPQU-s$Am3;XW(58Qi~=)KeNx|Q=hMs)n4`I37_ z-Fmp_HgZvva`K08$Ni9e$KK<-Aiu&jr1NlYj^qBr>QCPvR=wQ2Ss&$tc$oeL=x~6g z_Q$kBWHyV9Vtzi9>}RctgoJ}PU+>~FNpcF1OCACG4wA$N3e;{i?E7E(r?$JtK_UkN zGCl7gNqOhSi7;|0@9ZHCh(jOxZY1JW{P}!hQWqVNfgHZGaPq9p%r8i0gkI8ahduTs zFAn=QS;|WP+>p#Tu#4yBY%+Ub`SgKhCTA@(_pBufONtvE5{eIb!~U}Z-@9+x_U)G> z>$?J^Y0KoNEF;WKrylaw1C?HaHVtMN%T?ZX;9heI3ex8uXI3> zC%(d^p@Shj#l7Yr1$YU)!D2k0nuG2W`etEgr>A<0%JqJ^4+ISV>d#*=)`G#vH25wc z7|;l3sUcF4NsyWN-Fp+|=XeV>{Tv(V57rID#f57Cc+82EUF`l21b!PwS*z>U5$Nq4 z99gY@XtjaKq8ZB4>HAz zNm3%#-u}M#@G)<9aeB&%nUwAC)vY1{_Q>Op+T6LvY^}yAR)>~Tx_QxjOkj9%X42*t z<|QQ95{yO#01PfDnD9hOepUYH-KSCdR2TgKw8K%+@|ChkHL{zpf8Xj`x9s$}Ggc5l zrw$zeNiaGeD7IKo1b5^hxZ7;GwMu>9haqH!mzfhrBRv>&tiHQtuYc`#?c2{kFF%J? ztrm=9c}wN0oj!fRmLIub=|a`FYh&sK4VwAffJlVbxP&Nq_Vg(qfSc4L7#(y2CW%=z zC%l7wc$D4;l!rdjvq_@|sP~y8KL}ebk!L@lO@fAedoQa1YISZ=0u(SE_J{RpA__pB zT)^;?MK%EID|^B2zBV;}jre2GTNGx_lBd&$Gqoigd~bPZ(TuiyFg0dUll^F!?T_U>mc z;T;!`kw+!WJ$2*Hkt09H{vYJX-9yQG+1%k({$pNzC}-hsTtGRAe0kM-csY9M;f~V_ zZ{sU2eEGrd0ssFYPX7L|_Lzs;QYI;&X@ecYj+{q@Fw^P*8l*1xO;a!9`_hV?;o(Z zp7sB@*H7%t4Jjp(lO+YJpXz;2rAQg$oyLacS0OCR4U@ z`h?`j3ESOn30QTll#~1)>{@25ZOOT3EVuB4WJfOKKjH^~w^V}{bZoe9-ThUiBN>Jq z9=3g21c!ex^~Zsw7zyZ*NaY21S-8eDGA5VMK7&5@dQ$>%z2hjO(?@h<4`IsBlmVtm z@R1QXrnAxcz_H){6L4WvcYAYJX^O@!+V#)tB=AxIgFFZDN9`c@Z9FZVQ?+EKc+4x_ zi9+3yp(t;mR)NtEIpWg=A6zF0#?8AN@q}399x5xIdEymxfOpva!{-Gn6!XwAwtPXI z>WOm?ngYbVmUK(6@6$Bcn}Z`wm=62;7eT`xE~Hz^LGjs z!gydewCF5kSCp@R|E94FBw-75mV9)s`!4-0U=vVJlz*DAv7nsP_nu(@T|9 z#i;@M5Cs-7eietGlPJfwiBmL+GJ(gr3BXc zn5^z`R_!ZNjvOXmSZls0Oh6`IofKWh?Ty!8bAd4_Eq zV#pgy5H~kHZI50&ZxsRRs(|Q`=rBK zA|5Zl>a){mN;tBIFF;GQ5N+tub?|yLuugr)?%%#?*Dt?ed%OEKJ5{l%$+Gvjv)e%` zk+$ijWy@9P>`=Vs`PUWxApi7`W@Ush9L5kN=hC~76$er-gfAtg{LC9qjx()7c@@pb zFv8L2O3R?|URTu%y%}yGh(^>ey~BamLKqfa(5swGGS4mWA#vaWScBFOL&~hRXPw%X zb=wUq&781wbyE6rPg^dRu=d`T z&p*tUXJwUUW$YYUW*p!?VfU_Fw>xamzF(6O*o%f)1;~|`af|}htIx+(T3zu(HkY^O zKJytXSEry>peGGOb@NWg7s|yg_xyqjq4awgmBxPE1;4SPI|i(BF=+PraQY6--H`S6 zqa5*0&s2Hr-jNLM;g53f5Sbx*;cdxv|EPd_Y=DDfW@=MPQCi4x^2x<7XB73^K!9(M1h`LR>mJpS_Tbl%JT5Pxxz zLT_wH74N+FZ@LdW8s6j2!OtJTy#t&@SKdU?9^b(}$Dl!BV!zJmLCJU?U^%{#OMH1a z<;7S3>mHei{Hg73+MT!G^m(2vuI{zEk_C0^H1~Z@G1CqL=wWvm$sM&D)>5!PIz6tG z&-Jg;WyA_V%rp$<#D3&sAGO8hX`3mTm8a)yW^vh4`6(Y+vV*oA^jfmv2bMeWAxq>J zm4;-J4J+FE9exxIUmJuQ_IJwk6Upje8R6FVjU{b+Cf z3n!K$JA>SeT}(ZAm7Z0m4bVa&EdK0~-cgGlpdc&iHkNfv5hJ5}Y$3FMLsh*WC zb&Gq5^Zo9azwkqt@Y3ZU*umc`eO^;$pJs|NgwE0T#JKLIN6@pb_uwUPkiKzBX3Sgg zCJtFQV7-i<(UJ7_QV(?Sa8FgR=o^3y@TE?{WK0iamkvfzTl$WS{+s{vKl3=>FZY3f z;h%eU#inItW<~I<8d#BDUpIBAkWeW}f#6JrT?WL20!oOMz%$;&Wd*|gX(?mo>zJ|r zvpGuGc5lpf)#f^~X8VHx76HNDFjR3=H2br%C!^}k0Rt7wjsx@#vmYe+KoB^H^?%_R z1rNaV;T6Tn_F7E#E|*JoW-f0l`6DZj^(~7s)(=LXP~2=+)29JRQqlL!s^OLz1|Yp7 zBjAwT76dw2gtoM>WYxv8O;yTPE|qP5c2PjP;KG8rmP{!iFU;W~ALPvi+*sPug}ejp zn>XK-u;}{mv^+Z~<21WBFJJRh?l}6ksyx%Po>oKW?EA@T)n*qKe6l=y4IU`|ff#dd zYtx6}*{+W5@aXkS3+$~d3JlM2oG){{=;x-9KrcD@^)?rS4gUr3j*uYVG(c!stJS?H zmpNIDoh{#DZ)$$o=gay$V;U+IFKT+jrje0^((-+!RS#{zu)rPyi9;wDw$Vp&ViN+x zt^GAyyL-#p-djEJ%xswa9>1s(IMhl1+HhlOMe zpeR@Xz>rf|Elf{|04Ksan-NJE@`EBy1^q-9XanyIyMCe;{*xTXEg&7di9UqU2byFy zpY^ztGIAJ3a;NuRew8o$bU*0Wu)S~H#;&zxz2~Q;c*^q%M}DMI91I^?m*c7>>~ZB% znnN4~qHFpI4EMwn8G+%sbK;4Sl#P@DWikPB@q?E`bn4gl?aEibX7#mo7p&~N+!Uz? zQVjGKv4B$fEOB@ee zxQFMFJ8?N5IO;Kkf7rnTIB59)sf)9D9OFZAhy48j_XqOzkb8)(?{=XrMh}rYrr&!w zzIy%e5U1;D{E$5GDayw?d@B5Vrbbz#`2BC3mb>`=q0sO6{sUZ;J~@B4*9TP?*W4c- z-Fv$@U`Q|KYvKzMH9YV>d6adC$uo@VL5SjEVak9I@xeMp5SV&nmIk3w7 zjCZuqp@5v`*lXlkb$Q7OE6)gAm%O?H7-pNyae-f;SdgFIlF{9iTUij?l}cm5GQ08U z1wZm8kCE2VBcYB2tQg{#%w;~KJC&QZ9R&$~1-g|1^vH0~wY~K_z7QFaz>)WC4Tqdy zq>bra=+uOCHFKr6)^~hb9eRvj)`!8cl2ZPerXjj}+K~Z~WP4EQ8O0ImS29{M!n5u^ zsawKFS2^dbD8KG>aS-kADFg4LZ_(BN`seU1ls7ypW^O`qaai;TFFal4lRW8}dFXue zQ8`^b3_gj1P8G`Q`Gm)GFIo%^sGAUvcLJ&0K6O#weU~j={l|Abv$&KR8unNWIkI{? zbmaYq|J%PtV(**#K)~>?|DFHB>S|~m0U=KJVha=-Q)5zAr~)i&=B4HfS>LofEynbt z6-7X{FzO3H4vr>lqjzMx-Jmo8mQ8*wtlH2}sX&MdNbz}83QyzkgAxE3N{96H!YycT z-~xKQ12lVH1SU|H-s`7)c+Z*uBvuZLMNi~pDPt@86~}v)krm4-;1TnUdrAkEQBv)p z8ut;+eawfH15*HKlm~$-8l_Jl7g(crnmMg(-u>947wqwi=j{CHd8<~k%0Sq>8(apx zT-d2xTkG4laO#Bj48J8X%-*le0i)c^<4h|Jj>g0yW2lFda>7xNNh?<-txzfoz_4m~ z;0)Z9&}ubXqNglym6KD9SzQYtDW-F_yf|Z(sY!3}2m}lk=90oq-oBunh#Hh%C?g;E z<_%IFF+|$!mgt%Cd8w`4EeB}j>3Ic~fgoD@s#ej1<{yW1AV~eHJ^~nS_^=u89sI~2 zJp^RtptWu5x8Ahvjr(3!j#P#Y%-e|I<{*)yezFW5^0{$wn{YXf>uWn^XZMdh#L zg3WnK{4)^GG9MPOgK^^4v3Q~&;5;rM?;{NUQ+kRo-U4aVZTb`zltYT)GJ##38!Q5p zE^qWFpXWQ<_OKO?_|w5XQ;D$D;cxM=c=)K_w4<)tl7MQh-n3L{){@03=S%1ndO*c* z$}iI`5QV<2p?DL8XNLl<3AUxn*}=HdOwn^hTcmeJihwCRx?_>)t(RZ4)i*9#QfUJ& zTcTf6{5w*}i z#*gp$#FmA@QOOV7agfM4!1XxQK9r|8?w}F_P2f3{TxfeC0{oFwP^Q88mxr?@=R6qQ zyW^v10-qnh_W&UfzT><1bkV19iT%Hqd#|sA=39Jy;NYnr?C|$q@x50VLcCX)_jZA5 zQOJi0`bTmRf!zHeVMB#S?tx$K)Pkd=ACB*byC74-z28K6_t(xe4_wSk$E^BCbnw&z zzIza--|1V}{U@|(qUJqed{+wIkuChn(~bS7PSJSz&eHbb8b1)`6E4*AxP-orS40ORKBEvxNq+oan1vUCj{2Q7ij zf~?`}dMy?&yXM7j(1s0eO*R#7|%tP@KxM z`$66|?rpju;hkj41G&(R%WxbcrVRN*moe5VA+vo^^{l0`ECQ5%5OXPdzhYS6`~pq= zDRNXWj{KANIFaam;0mtqbTL2j1(_WlsVC^ILLT#6h!1&5B-EqhlW}w^InnnB(7OE7 z;{>|M!$3#fd7hD7stb3~JM@LZL;J8uPClaei3qw3kA}i9yS88JSgph9s(t%E{g?l3 z6wmwPJ`gZ`>3jd&`eM|h{?Jl~3F{p7WL*aP7UoYeEL|-4+33tuvLl9_L#5Ss_U!hp zRlEB3s@=SG+xBV=f6v@s7_r@I_{v5E%NQ#^sdDO>Vycp`edoHp z_QLbFv3cLp#6uk_3k=0&1?qEUdf%rkRVZ1Zijc~=U?guS3l3XmUTrE{gaN$n4g_p_ z5}1d5&h76y7-yv|7D{z`+KRJ_mQvi=e93#7yY0IFn7hnk4z{ad(Md90_Fe{6w&Iph zWL~oO5Fh%m9AKCQpG*~Ua4mXx4~9T31|9JowD)X`!#0|A+udteqEPj{lL67LJVke+ zSHck8SzfKf0X5TdSWL=%!o--;Q(B>IKuP91vof2;b7-||Z`ie$Uh+dkR0GyJ7+FUg z-;qDx?4G=d*I8M-zO`xXZdYXxePv`Uoj7F|FPxRpA>&rck=5K2s&7~9&K!Y9LVc4s zd=&Fe^vwTA2Iw{agY$3_wgll(5z!`Mr&5F)y{Fk=R6rNNu!ClPd{YFTs_s!t9D;M(e;05W%&TpZqNB)o>4;13$_MssEKe{;1aQ%bso%wjDqwe3s z#nCC~J1O+u?%gyyrmxfAU;Y3Wf6$fND9nSuf6%>WRB}$+NCuJqG2MSiAoKw)@~K}O z{-GR11&H5xZ{Hu{-d!i}h%a)U9{yl3U1FVtPxr5P@z#6D4c8~gUD|-cIe3Zv^Ga39 zh?-B`l8g4zsM9sD!; zRo_~ZE^gcY+FMo@icTdtuw!HuWLddn`3sNw-olAg+4a=;nDi;ey$o`8hX)M9edCl} zzmd{qFJV8%-59bDqjydVrAb(qohl?{L;JE_7}Fmdj5>)zI}h~JKd+9eq@c{PO4*Kp%07Cc-el<-T;p zQKBS3Jf%3;+D-Hj?ODu5{A7uBR9>%2lz4dHO5KMiui|4D{K&t`?qS_Wcpu7*6Fp%7 z<%;R1UdQ_9OaZ+GNIe*@^p3E|M^87>0S>@J)P6lWQrV%q9z!oL28zOaHXriHFUkkw zl-?WfuX)eSp|zzyUwz@Sy>)lPHg+4{Tm1k0m0uyq`{q6nFnsOBF9kxwl#xPW(|`6v zWI{YYpQx{#kv)D8hhdz}^)0)7`>wtE+GX3@+jT#;REzIFrn4hy;KI>b^@Q%4m z{DP7U4FQG-E+{u2!VvR05cYRUTQwFBSJHqEol#&!8c_HSqlbnGL-35{6T~Hc=*i<9 zJ<>ksQy+Y*5g1zx43gUQ`C{*+LQ_2fQ7+@Y7}bEl-kP_oURHF z?U>>10@_?Ti21E?s&;AR9^X9R>7=cX-L zs90`#(eeVs2te8>4Hhs(ZL*b^w{fgoe$Y)<1OUL)D_eO*xFE2JgSpyd79#`tx+23U z;HUf{WC`Pj8~Z ze|iT$JgB4KoTl>2)E`$C%!O8cbhg#ru3WN>>(>Rm8)_#*YqJP-Fp|-xxKswB>Zy$E zdAz;1=X*?2_DXr$R+bj+kqhT-c5&8sxJb$%Nvb|Yq>yeb)jdzjqpn_N^2zJtawvZh z{Zx?fP65KF&>sC>zsS2^RU+LHvFb&J6{8#B%6om+dvC{zz-1CIRnc)AC<0GY4x^7- z&l~MmE~G2j5M90P>+1RQ@5C7VQs2mnzy~3tq1VN8NY>rC^7QcGTkPUGgC>$3o@Y-- zSCnA%SzmPHk# zrj2{iAOHStr*Hl_+(!9~9C5}k^pT!TuY1~noY#r|!SAF>If%gJi>_pwo=DmpBYBMC z`KT7{7QjS~j?>q@;ZsK;-WgC-La#6mX)*pe9bHJNkFY<<^d6Q<94q>Z@xu zSl>{ZBN@Un?z5~%5ulttWfP@EJCdxoWZI{UF*N}>PT9ygauRML=UfvG-r>7YLa#gh z&{gCp(183&pRq+I-{QJ}5T0-SQ~7zvD|);EwR9NNx9(ZHw&h?yn@-z-(&yw&SW6Pt zGSE#jHebwJW3R4qGP1@lL5wylkBT2EKx6M278}7oOga;kmrl#c&?s0k%Sozv>3Q=J z@`Qk8u|x+5z5k+t^6yxL40?VeY@+nz-rLxDktv?uyZ+$6esSH0ILKF!oAktbf#jUm zbW0QuT;OfwB9ewOq5Ek+@gi_yJ{qGgh+lj{9!v`@%tOd+Ftu-S7M5J`gZ`<(03h0?;Zb85+F^Cb|p-#+c&J>j}V$ zi>Nq2L0<}(JptJ*F_A}xmc4gv5E?M&ornC5*_jh3?F&EkC+t7^-~UA^oUGlt zcGX_^?)PnN?eNWC+$O@{giF2-M6*%d)C7uR$P>FKrf&&6q<(ehIEsooDBOQ ztb-LOU`|=&2chJ1k`;%82Y&KHm{KljCl}{zBA-)!6;{zP%;61$vWJ!t@~&@A2Z<8j zo`(jaG?XM2E+GLB4h93qgu0L#PpKXyAZ39gt;Yl|xe^a?qCA{-;tStYz18$GC9^ph zL3x`{Sn{!U_51eAzx22KnCHBNa59_mX*a1<*3$A*ELChypnh&jR(h#w#n}}r%`OhVQ&fiQkEt=^^x< z-+%2z+mNwUQ=6klNb1FQ+5_8Tqy9tD!KMDtws&`Iufd2+UjSY}p}%#Lv|_O+17prk zom{rrxmlZ@pZ7V?%sCen740TOE42r;3ECls1N8-+;J+{>L@s2^FQIoPDl@#`bYQ_G z-vfY;!G7%gUO#v?>@XrDg}A8?zAOI@nAGk`)3>7X4*+JxOPxa>(KzsD$ftYJ2kJ7U z$#Lz3#~|?}oiCgxC7&2s!4cMZS82k_OoejA0H1hfhVSJ=bAz(?i z7bI@BUWCTPPkxZGepGuDCMuR+kRUE7E;+sWws9m6hc;15bS_{pC-N0Lk}N(<=WLjiyw6l@B3JOyJ$k3vE|eZq+N1t^ z=LpcN;)s=tCvY%bw?%1ugZph z=n?ZLpTc|OmkcVeNS?%&5R>AE;9k#hTCt-}8R-s(69s+i(^-;uRTdAA?4w{)dVgv7%lrDP&o;ZEps@2kQ1rg2#iYkD8 z6Ibuuv9EsZd8^gM+-ekj_liRBi!odQC|t}q^c(=V!@_`|{9pjm<({}XMw8W#fZG6! z0j<#laD5}+Fr7yn{wM~$SK25f(uIRy2AEiK$MfXzC_gS}#O8bhe6rCw!{iI|3--Cs ze8T?BFaCm+W~S`hzxxfla`}>VnQN`aoRU@j@BO8}Y>zzkq+P#u$?mM)Q6r3cBSj!T zsZ9l;*NO;75jU?^h*-n2UbvCllin20cGN=Y_?<-fnbiE zo||2=2|cGL=PX^QC`}1jg-0<^C-itEm3;>eDhhe}(6FIN}u@l{Xw| z@-C}#48(WLA3RvQW7od=6{`#A`Vp@3Q|k}xUajN%bJLqbdi#4jcKgm<_mkxe`&`b> zo;q#k&z!NT>9QZEU941$gI!Q)UUB-3D3j9>nz8y@SAkF#{pf?|+sgtSbPxXxnJb6^ zE&sF=KF~`AC`9-kmout2fG;OhVR*Cs9UJY3US5TEJbF)jQ5ighz6W{)85Ka5uEa~>+!y|cQDOAxaqAf#c16W1%gApW z20F9yn;w2b5GM}05jO^^d=sO;=UvFP{6z1Fhhziak%Zz^8w>B~FbH|_zxqRdJfyCK zAx+8)`#Q*Rdg41cWV7>{=R@3~=eQlRT3&y`H$9Wj@Sa>#wd4ujkQ|D5!%x!GGqMPf z<+M`zgb4LayM-3CTSgp6oBYJ(jgwS9_0Ky{i6hcMp)3Lrlz}ojZ^(m}E993o33L{C zSwt_-xAOWht#RBt%@vjR;1UgQp`qeK&$xl%!}Izs8LEhatW&-@c`ei#`BHn(J$WZ8 z>#VNY4AUbH>y}O&NQcs~g|Vsh$WuZvWCsJ~Z6F}rY3xg~4BOaRjBEo3z;_(gxvv(LLwH&hmFh-XfxM0exh9m4bDo+*jGs9s?xR%l_; zyYO~Yr~X#=&Rb5e5EuRe

q7BMEWjDfwaUG}s8a>)VjR&*WWK2g2k@b%5-1tR~&h z2syh@fuG>MWG-X%y*=BxziOA?xMctJfAimY%I}x^K)~=n_~rkJl%3!~_@IO!dJryD z6y4@95%UYj1!nN$LQ_JQA^zo)r)*}5lh0JSa)gKZvCe2PY$E-o-}-IeKpJp7R7C+u z1L2A8L9ro$TYWGNRZYJ5?*D_o_#GAKJ-jpwz4Jv?%(K!r|G94h>L=!;cwR<5drz3Y zl1~_y?=TRJEDGiCvD8r=Ob=LEoU@O8_!;}jKk<`(*v4zmzihw%mETii>x4P5Vy;v& zWB>7g_J6S16DRGBS6{Td_isB5U`!Z~q$ry!SR{y?zt_70EBsaV7>p0~qns!%A4*oH z=wS#JsUp(1$$Zk5<^&q)L50EP2QbXk0T(V_R%n1SK^TQ_fu=N8h$5goWI8A#MyezIZ{vO*6zr9}Z=``Q=nSAO~b z;X;&yN3uDA;bKlexM2Bw*`{Y_?c}-BHaEX$#vYaVs=|j;-(0yVR>B715`+W1Nzr`k zKPpq{(#RbQuO(2)4`k~%*QF2_6Sg!06BpsjfxbY&sQ>hs^S`&TY2uz<0Au+7|Bt!rzIZosn& z$FZ7MuHUk)?Jd!8)(SE<1OlR|=Ysz>0pqCV0^0w@GK$tRo z%4YZi8!T>=U`?ii;HPY~c|R&!?F_}{^VCH{{Z&26Se75>UsaA9Z2(#-igE&aNeg3_ z-Y6+CRPKXye5HE(LABQU$!`c5{ zPSKZXMWLsMLsn1(Ea(Ys%X{Y376L$FTSp8(dU6NiISfGB5u--Thi94xN?oCuHbk2h z4Z_q7lpTD@9@&&V&NFlo_(%y<98ebJLD5oXetc)UC-}f=rZN#0rAvgd(dUFjUh5rn zaXF@XaGuhSwDCe-e2-UsaPsq0eAG2u7Dg)+gmxqkAwH;qe|^B|>LWz@=fF7PUFs$1 z2;z+V5>9C%HwZ5Bc>Yt$EAhcA?iD^KPte7|a1;;UqWrVpE_L;w&EOwdgX+3x)}%X2+S~2fFmjD(C_ttj16rx=q%!-E^vTEdb9E4(K-KHg z8$JpALq3$6CqSNYPBUG-47zt9>*+I!h8!bLpp*#qL7%=IjzT97ZOdOmI|w7cBk2l1osYeiCHLBmnlGwGF1Jz{{C( z!RF5{+dyewf9;BG-`%i5v*&9@dIGrYz~o2nAAk+uZ>1$V_ZempjSqM){$MS}k=kEc z`iS1**_E?)_VLGUW@$;3MMfq4bXle}vO|&cn(D!2B1KTGdD~JM;(5E=!*=dRB!RCK z%-_QoV*=OE0d@%(CVV4n%=yOA3F9C1!*4=qkT$dh>(H)TUMc+e1P6Z+C+!SAf~VnA z+N1JEe32{;JgF)rJ+Biw!dRcgl)uoP`l31hSgg-ZCO`hUpVs^L&wU_Z_<#8C|GF=j zLP%$&Bok5|&TPsGmBaC}wf&kmNVdu<;5QuAIUSR|xjN?j<8}3y(uAG%Nd_-l%&V{tUlYg(&z=XUHpeFd`o(jbg zE&ih~t{{2GuxjidA8HykBjtxnlTKKfC+@WkU*sFdvP&Fl93 zSH5LiJ3CUqh&vd|teBg!|IL5?pIc>i#$J2rd0XFFlMo5{jE8c)7ee<8E0aTL3$62o zg69PRPtQCVVrUn@5EQ_e_YfmU5z>^Sm^`$Zau(^Wd{4O0<%nt(!)b{U^Npya_ZUl<97y;aN(j*y=xxGs%@+$+EEk<7!^E@zY{B>VMOZ7to@XTz>P30$%ewA0 z8){s#9SqsBWravFB!}pb9g<~BmO^ysK|zJX5!Qp=751Ye z9Db2x*%7j>;G__OrZ`}{xD)!`?$h08s%h8WHJtMA_g|U2b{9_H3HP1@0%z^|>dRr} z%9SfuX0FUHzx+mC`tc#R$Pw^vhn)!zZR?pY;(hry+gJXT)^+k=mvwqO3@d+qRa#^6so#d!SLpZT%&$N$KWw|74KHg$5*9(?$u&Ft&Y zlaC+oF+kP|>K3}=OM-s!&XYiV%Co(-(cXWspZ2@O01)F58F-H9l*#eI!88VhdgzH4 zTbq;i1D|`VO^+_xAp^0-ZKS|t#N(fi;nCGR2NB0UkMG;r*Fkx9`YIpxDZA+1?R$6H2k(E7qo_Qo#r@N^h1Xtt zEl+8jDBityyB*{hKtIPl$%>4Z`T1nbzRpLX$Fzx4PoIG=_wQSF2cWDT!ry0q==Ju_ zXI`g#N0euW_<8%vS3aPPI#{ZSwqg2A>U3@WdV7t2;R1!w_*Gx__x9V9r~967Y*V1HL>p_AlCpW`_4M!<{jJC6Wn|h1x@X+UVlkqzckvmbV4 z(R5{~(u<)t%4~xCJ&S$aM^y40F;?%C)p*-6`C2CX4fQxr-yowgtE`Q)5A#XPI^%#Q4T;<=>Zwl~MCybuSCa|}Dr5Y8u_k&{Ns(d)lqaVZf{d%6BIIj2s56^vfLBa(m zPx)E#I5_fZCr{h8bJ{oIo>~0Nmzp)wsh8*y4A}lM=Finl`)S$}dT~y_loetw4bM4s zf8y3m+M4abRb7qQXQz(c^nc0?T+jM{?Q{3qANe2tVmq51wO{$0|1}2jSK9Q^6nVJN zRucw@Qm3SW33b69-@946+wQ&lh4z`x|G{?q^*1Oly1cR3 z?%liDCcfXZwVmyMwrOdqPtPcKO;$=rGI7N3QKFVThBJSTw*XY#UcQ}^!W2|~7owKxG{q--mul?5h8MM^SdTZ8>p_E|k)wB2^G3by_cLu6c zDm-&7d}n>6@ART5CK@!Xc=3s|gL5Eph1@?d$Ko*XB8+)(*A|RTcKe?rZH|`d|N>S$+F&e(CSD zkLjFFF&2j5?luQMOyS^o|Ml15XOaO-kmbBYmol`rfdO#y-o5O#e)Q2tAb7HYe7oXN z9sK(FQM*ouG)I{Upnj-Bs}6y9?()XQR_;Cu$%|gdTI1Sd$Erb9XPuQOJ6mUnXBx#c z?(1vFf=22p5)IEz(PPa#lUF5>UprN6V+=ayzVtVnqAVH#Kq>s*l?`ks|TzZvw z2f-kNQ+i+d3gIF-exhkFG97#gJ|cr3IyCFz_`z?r2VeSf+kfy;JD)SSpce-iVxHLY z-rlqwoSd~o28#XtqxRl6K4{-~v=`kM%u9XTrt|g;YCqi4m|*y*&{HWUbm&eJl#+un zg-Y8HG)8O}uCRI>p+=C~E|ib&4mhCXNr!FDlJRHfZ`0gj&^Vsfd7?(xUwABBI&^pw zClg&RrC9xktIm>l5I`eo7xqK8(|X)nowu~)L9kvN6f)UM()`C?>3um>!_|Jp(@*48 z3=P~9zu-m5Et3O)dD?HJe&Ht*Yw1}rT$V)~e7UEu>0tg_oAf8NUk4HmGpl|dBV1a> zJGe;S%hkp$4LizZJCxrXcyGN}@1b3FVVRYWpIKKn8Z}NZj?e)cxP7T?*2Gu?k3l=e zAcH*yl%u}81R{I-VU48}37V+`Xr%0SmaZjGw6B>lOJ%nwL59=yy3RA75e^T-|%1@@C%qX;nRt$$1bjBPw>&ozTVb8{z|($dyjGNj5g+Rp32Cf!xx=BF^WESd#9~z z+-Pm{ZhN-A8+ty4i5cM7?RZ9gT)$wFc}n?CnV4pUn91z?2!nT@v2i~JGlK%gD+x(# zTfTTk|DaA%_7e@VCo0kIm?Z^J< zKiM99?Y;IZf9vnHFaP8B+sWQ6CrN}=F2U~@s9cCMcgILy2+sn)`oZm-uygu{wOent zpZ~@GN&AUk_{FTWu3PV1c0bk6U^FwHS~nWLwu>3<*FMnIY}<97tdU^v;S(k(kK5D7 z58L74J`+0ZD)dZUbRysamrSZp>Gy2+$ZEWBZ<}0pwJi2UKT1JX0Z7K6UK2hZ}WqS>LHuVP>^$ooP-9^;U= zbp2wEimy9+cI@Qo;EmV7&)yy=J$&#WpG6!>GeJ>mg%TbhXU=w5f9jw4=V{DG?H~Tv zf2Zv|{Wuc>qzgT8?ez8z+a`nbF2?rejoWSe_G@kT&JFU8gaCQR(8Z8-8_LJuc+fuh z`j;uEoj1lh9oC5rYvoznzRv8NhT|Scm~A_^d<;&b#`u8>Mc_XTV0L&ISvp8zf#kmdb1+vaQUw2Q4f7#p>+bKTC0GDn30jai{ogmA0F^6Bro61_k_OAsTx zYDM>)UUcU=o%X@g41T^~vO;~RizXFHAIp<=z!hEhF=N7r3*FYXv0p@&PYXIhE|0$Q zrS|3j>hHGc;Zd6qe0p}=e(UjG`})Jf_GF&{@Zz|g9iO!?z5lR%@#_z0uNpcGe!abn z_Mk4`L4WSQ;hD{z&&=dx=Oz<}o9GsU9`RW{KsRfZ%6_2p4;}jAxYGue&81L~Tdl}f zxXHw;&fKCbw_{}-pv* zchmbb3D+lNJ{k}C<#;K}lkz)WxPW3h4QmEk-Re2$Ci5jdpE8!g?uVB&;7wnbMoBW4 z56do1@bcIol&D4{@fcf2QL<;iI1kHEGFL7LBA?W$3Zw11{r{XV1Hu zdF}c|d-L9;{nQU&q*A_F3?Y!$F{UT?+Sbl(jH0c4cxY?Kj|$m|sozxx*1s$3Fi{s- zp)KjVKYluI^I!gdw$)$z&lxu`Tvz<8G5t63t7}_Kkm1MiW&2iJ-88?uZO%C5$r6qi z?z6RjOkHzL&i62enfRR?v=t2DXLCRETn6#{$T&|WqTXEj))1!L8i6Z}pQlW;HMo3G zkV&?fKEXKFNF+>S7(?0fmNkgyl;e1g4*4wu$9?$ti9h#;+fV+&A8%j%HH_i^-!HYV z{-X!&m@=vt*Wu5j)a2I{(7X&>RTi*KJhOix%*)BcsdaVZ?e=H?M}MXL!9Vj0^p_ic zq_6GmKW!`L>V7RI-M#-hvUehZ9YNx?)k(IZI7V8B)`k0p|G_`{jrR1Tuebe=-*3~S zr|s6A+wGkn`$Kg$GCZjpB~N6;rS7O1r9Zgzs4HCNoBs9WxUCa?K6qs z#Tfp-{7Zkdgu|ubV#C!p+A}hZ_>4Eu#-?x7S3top2XM3Ns>;P+1E?YIkwhgQ09 z!qT}*)AYLY5=M2(t|W}XUX;1KN7+!ikkjzqY_HvM#o(lU_}~K!yu&EEMRd?tVb6TL z?Tibmo_)J}^Jcqo_b!DRMuEK0(bL#8ByCIwG)Ozj>x5?&iF!~bs~Cb5ATw?0yP~~L zBVGYlb;=_nA2EV&<8^i988gbf#I&(V=ZCz~;W6tpUrWJFnJxRJ9eQ+du0XiAD=Yex z&An$DIIiG2aCy9#Gg=#)4c?DC*pQz*dV~@0BT(@?PA=55QlvrWYNa3E&?f z^;i|`wg(3{<>nd4kACY*ZTk4*cIWnH+rIZ&Ti?CcE;c>;SviAGcbjJ%6rg^J%Kz4R z>4wxxnF%+HYfW1i<}zq@uf@SZr8Zvq1z79qM&zp zv+vJv{^;l1HD~L}V9G#pM7h5B8|}ldeGyr^@^IQ79Uix@J$%|89voxno93)NdVJLW ztA7j*^0#gr?9_SnbHLWtR(t*Ky*veQwadxX+V*bSyy@z`TOVAFKvtCB`l+^GWsOYW zzo2ABMDCWWx8RO*89v;ffRt7FNq^Ap;iOTI6JR)UtSED!ERqWp?bKzzH1_0G<1MUp zP;X-#B~R)Y`r>)TkO61V7~#_Q9=d>T>kCH=xn(H)0=bQWCRNc1Qx!^p=((}0scq5cRaJb(O);US8KnQ<6c>aBtjj#vweQCe_N$v0 zGy1}lqpS?K(__N+a}tEcu;Y}*cwZXBfi_9k~qG zAoeN75r*<~<`a(EUSU7|6aSq*+t+dQAPqf_Xf`5`4gP~jH@=j}_s{^j=N|K@MEecz**xkaqCcYo;J z_7i{fXY;goed88muCgTa4&uljN>tp-dSVL66?L~QOrJc;3+h|DH`*rs`UaCeCxxP^ zm!n@Zu^KnpTNr5MBi@&EJ> zq!P?|c?*OvgjoN3<4xyO@eyi5b$jpl+;m0o1WQ-J>n@DK$(_L9jtF3oAD?4R-taVX z2z@GDlJX+EB|7={F_s;U*eZpchY3`8J6h=%Jt2c8cy>|)s{bQxf;2EHz2P-zRDS0g zUE=vmypsjOplv6RN4)2!;YGte2E7ea!&M^C@(tZ*D1Qvw@NeGv3;3ANfx-de`gJ-K zm^Bh%2&06=ORg=K^>`M1-x+}_U`1~7Q}c!X)wiMzrgjTANIlSj+t zWjdJ(9lDePB7EcvZ}JnlE(TOr52_%hv3%o^v$Rbkm?nb&$$ZSay7Oj?eYa$!Q;58k zFY*T8vc{l2r$as5&mOex?MVz%KcjukqudzaQdZKT3An3Fh8Y5pTs1gy877EG_zY)w zrh~LjYtSwEty2f{eb2N!kl}nSZyjS;!`*mk3|$;<%xmrr2o56dw?k%3mV9jGxA$jH zpSJnYVLN>Kq@7Zx!-o&sAswopE=5nf=BMvZ}% zb!f76leCL_r7hia5YoUHR4zfhs5@!K|MVGLzFEHv;jz;-KJ@TwXjea3GPX>ke+@su z#UrEM!6{lrLmo>8qkrjivrtfmw2(R`t|X%05c?B7MaTAHfo!{B?*Pl`Zk>iHWP5Sv2926T^}ou+Be0+q}{rUqTt(mClv& z${)Y?>ux!bSBze$Z?2xVA9#Jze)!GPwmI=#o(X;14)L4qY(CEsmPd~(;(V=6_9=EgUN$om{?p^bTpGhU!``A=j~KT( zkKKLw$0u!ufvYTCOnrvY>+k5u?GYNimBbg{+&bZQiaGxI$+I@YJ=tDsf98MuPqjCG z>PIj-zu5la-}$BX{;z)APWCPHwRXBDpQUfMuSgr-8G2(^*pK=(SdS}n%Y;wn>|fVz zey;ueU-}o@PyCrb-*$G^+Zwv#dl;&sr}N~=a_U|N;fwls@ZQ7ro4@+&?cnKNJ78>Z z75Ip2_1RH$$*0b~c=wslzTNKKzZduR{U4zWJ8*$ofD?7VlSzZ^LGVs^bc>(H`!!_t z!I!?+w(j1_X%{z`5UryJ8ylWBWfIETby^)P8z5dT?GxKlr;q>4u?RiMWKvY(T!6{M z6}f37yRh{$zaO?q{Kx;?k5+|njgBP(Bvf`{EC!)E9_BLYRD}uL^?HiIBB-YsD>e}_ ziBxvoNiK>B3bp@Rx(g_4b$B>?deb;|*m3tUB!gejxm-D;EhM}$@s1~%6<*2$y`bpN zze*Df|KUkemB+i6ok2$mA2#LS;SoV)1BWgpe#y9EYZ7JQsRtJ_DB}<)Ioq{Jmqsez zMFyy{Mz1ltefwJ+5h!~25YOXrf-T8J8fSjul3&lR8gSSmSiW^w20>ajoKA;-mP5C6 zOmBxJWDhSqd6x&t=51V%mep5xJ{r;8Mp@kOoxEv?Ocu)D%L3ukybw@fq#zu3r?End zwKkNOrLp7i?xW7rd^J&{-hGpV8bLxaOfSvUxeRwB*AO!1Wi@r&+@AT?we;y=>}YZtgDdB*{2uM+v*qq_bk+>7t5IZ!XYRiA@%CG9 z*QziE@CJjbe6DS7V-Rm;f26BCnOLaTk%zJ_p7T#xB@8bzN*&-gFFVhaNi=b!dN-{49&uu9at}%4ymv6vG>PNCv6jhmu=V%9q#3+aKN9C6uS&~*X(iHd%T_$4 zUO@;Q`Kcs599-K%@Z!afoQM+iNAG(a|8d(kd*1Lf($-{F$qrn@yo!hLoBT@`#x?r- z@KnjtUSdUzK-2LnSxd9>G0Hut!x(n;ON1Cwx*eF6L$C`U62RaAKW#zdSC4*T`lK3N z316Mp^e4(9RA6$XDSz;lRb->CBc~XHVVFOSLY(5o5311Pq+$EeUHR4BeTYJXD8{_Tx(Kqlv zp%*=&j^IxNp8nXE#at}O3vkp^CQFWs8r&F)XR||Oj1jux6Hi~D!wBRu-uG$TKBvyO zC)n+M8pF@%`%X?U!ZdafsQTk}i0RaQ^ltl87s?pMS$8hOiKb6mZGYk~{A_#YkN)BI z^A+Xn-gcp zZ~_vXHlFr|9~{S|AN|3168ECuC9Mp9A)S+gvWcbxQ)sY~6+48Ls!&cSlY@`==`M&* zu=17IGWP^Z&rSpqR;fCQVnn`4W1Dh?sSMM~Lz;u1B<$`e1j;(H@ED@nGfbS!09g^T zpNp>)Y%o_1V8neAC{`j;{^V6*UcR|n%37p~K>np6t3swh|HGSQ(Ma~Leq{w)Wb!ub z{Dd$&60J*lGLCy>YQ}h^qd_kiROh`aOBy+|4#VWhFhir?`bG2NpAYlIuR*cPGhhmN z(J7 VcPZ5+N&pgj)p_X5Ck$xT!Pld72<@k@2tiF@QA1r;+>*A#e!uQ=v)I=?xy! zCKu4d#3QXo>)4q|rZ-h0O`MYIi(VQZ`4c7o6=gmI&@tuCzoA|6O`O_ExVQfFB z9JnNA9z=%Tr+{&iP1Ta7XK|VClS&i_pGCZTq+k3S)O)mqhhie|4!ndJEKUW7Z~Vn; zkP8px$j3M;LDKkBHc1^hh1?>oZNsfD>dGe!GW&|Sgrq2XT;0oGmAeMeA#=hyjX)KW zc&`de*6}AQe)$?8&lO(ui`U9K?IM2D)_oZ1DJ5mJEv1|Wa#g<4?CH`6I8Jb|ovKCqDg% zxLIAPI6@Lv`SpB4(|ju~Wa2mP@)wWhS9psw@pm5ZJ7G1>MW^dbqF2tR$c5|dfQj?d zcKGB`J9zrIO)y}iD|9-pG*8w&on@olzkRFSy?rO!;ns6)W*dd?hIoeXg#KXj{%709 z{U1o5=h(6Wt*QOdlLzhm>4)uN`n0W_crwuAwtj8id~wSyd($JoEpW(K=OhIqmMS`$ zVjxmqu9$i@z3ZSJ&Hw|cB?rN2?v&aeAq)QQJRzLx6?m-gM);P3ci(thU8{>Ap=Klf+JmjaO2 z@%*A4JbK!mK787a4yM$>!}jFi`)xMcr$}=bj zK~(uMZduOj>l^L0H{NP*yz{ws>;Btq^Tv&84b=T;2P&OvC8W6s>WXm9cCg-l_3!@e z_N9OLOYMy}UTd%a;19K1Z~s8MdFLj4I59$hp=ljtF$&$Sc4t`EU47~sd8ungXW|{8 z?r1Yv3?j}aq8h`Wf9K5}|6<^!7{h<#FMPgaP0nNBmkBFGL3xjnNtni`gSRh5*yGyi z*r6%6XHMTb_bEjxm2?VRnlAA}EJhyjRRP9D=AD*NDn;R?B0D|ftjo{_+()D*oywqv zTh;;{(f@WD60j5_KI!-?PnFj^t0N71z_N`n`7)ncHI%l*Y<=))-%YTGkQ;A4X6vj_ zn_a`g(%L+-(qwAGX-XLFDBR^;!oL+Cg{{CV6;=xgHgczjY%AfgE*h<^lzSVa9xWdF z6+g;c{pyv%pIYa{iLHh(L)piF;VNLCUS}RW2bkqQ~VAEb>xEHQ<;ULpZ_W_cX(BO_I>7uqaOO zC@c5kui8lPlV&MD!19C*>xcAZ{HY%muPlb)Ho{!pvklreCXCBdV@ADxGLPax5;`od zX%?tJWNaYp2S84LMr91Nupl9z(wevNrn3WuPx`Z!ylmyfp!~H$j2z=H`A4thCCu9= zIq(m;C5kYQZS%Arf-EbUs`sWZ8S&F7k{K@3RG1B0T{8ebX)SZMI^fW_OM2saJ(;p} z-LMp{^t`*h%S)|Z8$vmZ=kmN)J9Tj)BQ8oByehv&;9-og`QctWnI5!_)2D5^|0wS+ zoX_XLnf1`6)E!ZD-r_;3sVl!|~DMee$tvD_N;Mrtez8Q1-=c%V!^8J68`j z$bSc3G^9OQL+NM}`$bkqFRU#3LX2i6=Q}rVgwEmKG*8Y}D9hP=zpc>|-r7J{sgoxd z!~0K;+WxUm&*(25zc9W%%F(aRt!XP#*{62o)X@bpeYWCLwaw;9#nx{$+40Z1fac!e zanjn~xJ_@yq~O+_x7s`J{un&HgW>+H?LB(j9zOm}jN-?rs7EWF=c{#R$j5Q`$TN{W zN)Gyt8J_s!i*&2V*Lu8p`*!=x=YF95%%A$x?FWAJ3vIH+q=d;so~WwN1^J9`EUJy+LwRx*Dn*NQ;ataa!Fs0-r2XDpUm4&{DB`y+7|^c#TfqI{h4urj;hQk*t&{;J`2VV|LhW2IJO&qperRhKtp37hz+Uwlv$u$p7O2BxEmuXm1Q zoB=V>+;fRsowAS@hvXv>#h<^R@!z~E?NYfi&JYGe>4=D{7?i;@B+||^Ts(>OC%X%o zC@4FF%1E5f5sgBZTOBlfRVL3GH89y1(d^&y=$F+skB@=}e!l6dNOTX`$iLvtW^A50 zt%1IDbxm0#b1oaUY3!^u?$+&(aG&_XgNdaF6#wOxAr-0Y{4-M8v0k=e+HkHWKC<`c4A#k`LYLwvic2@Ix_!x{svCq}I!@ z(n;jQ|JvS!J9G;t+aiqY91~JtkPG_F$&Xu`I`pOsESO4+fZ|{Hox7yCm z&Gy4T`lIdMJMXj`_g~Mdb0#_Vv#Bp-ZxgG#_2ue$WuPo3$-mm8I_q9y@`EowH1tP) z{3i_}+j!hw69;1M` z?fCuTOdb!Y@jX7q5pKE>Xh1ogROx}oylDvM0ypL9{Fs9Vq;7e(?DF8xIAx|jwCv5lam|M%oZHro5M2xm)Ix#YZg;o^Eb0W6GyQrq5nWKqFRzdhRMc=_Nb( zla_|B1+gqCf6qI-bI*&3fD?#xe$ZRfxa)Yvn$~zd((I!}Gl?C|o8XUCPL%E*dAg__+YX zTnNgFhOztvg@XIHDu=vEz)vV$Q%YCHzA==&zwCmXqXd1~WK60rC?$Q;;%vTDx)k z_4fK(pUJ-F&CQKEL1Lou>z+T_vyU)6IfS<8;5-Zi2l(g?v_d!Ztx$?#u`Hm0Ke#8!0RwZ)fUk#n|bInyu!vcc*w7l)#2W9!_>1u!K*-Cs{a|QelbF+k{D2ee&{G` z^GV;vpY<+AP0VZJX1>YGaB!0sGS46hj8i3id|6D$A9$2iZ|EzFp4Psk>bBB|H~N^` zGcW(fd-Xi=dr8~LQkc<4knxxK$>*itd*PaQ(Wo|@d)nv~5f=*i8rcKd0G52KbR-+NpPS*GzM-Nnr7bHwLF}If0%N6Ny<)L!?Jq zX-F%Hgml_c(U?Sz=yg6%OjyPxQT%V34kZslp`F!V@W)HJ#%Xk`6Rygo4oH(P4(h;f zI~6L%c3S0^#p@LP_~c1Dc>1W#o*uN5Cy(0E;}6=?2VZN4Pd~&-mSDVLvwzjE(j4r#j;CStI{Op+i z5Pg876Zn4egh=wdcN-4j%hzq599D&Eps)JH$WzZ^p0$rLbUk?jbohrpeY5y}oMX;q zK1r$kh8(%q6I;byVPpKMz63B`53B!@_#}-1+9~h8cE8=n0RG$;ex&{MAN#p{S@oVr zv2A-wl6!eg17noYHX^@5)D^e!X&V7_#A=IEM|uw0gA?%TKOTMbVf*TrzS#c3|NH-y zNrlGun{R)vz4rFI?X}Om+tzn?YYPgp*04XD&r-kX$Ek=jZ9Xn1jUEw4?W2#5J^dXP z#w543{M^s}tX#Y>cqzv4pZ{TIKM}G9YP})pmWGY>jWm>n3UI5{SgR1_jv`b=IqP#K zAA*p8AiAS7Ud~5cNL!O9Br>o-YWA)=DH*VZ* zx9{C+J2!7qCk|i}I>B|^^{7}wQr#Seu>!EHVH&q(&EOx}{zpF+K0t(G1F+62rnvB5 zh2`stc%>~p+eyL=9l9Fyk)B`LXX!=~6IJn5z6z7RwInSQCS}sZpZLnt6^$Mmwxv|G zk*WR7lHYWTwEp5v3oab)RlBKL+Ed9YtLaqEIFcyw@}~?3UGDj%Pf^#4fAg<4E?&6} zi>LkypdoGZsZtx$Wm3^x@@yCZ(!pKzTAT}`9O1nJDvZDCvq=NnBW>D_SEoN9{V08` z@vdUpw-l=VkUHg`oXQ6Y_!lGDU*hVr4JivB@mATCZI#i}ibB+hLJVUK_ihM#fx7_w z#XU}G`Xi%ELZF{Qak(m*bi_p;2~!qH3Ez%oDJPd%dyg!70Omj$zxs+(K(-mvn+NIA zj))^os2NXK@gQ!=+xuetFh2SA>-v*|qlZoHAKH)4)q~*+Ozh8```oh;aVn7B8&}x8oGS z)o&W+opA47%MclN80~jC>?(+JsqYtUf-G)r?zHXAt@hSC7@(j1Y`gQu8}0SaywmQy z_FB9D#+&WVoqKsgXP-MrdpEsRs;6fS-=*7y#()2aAD7@c#%fyiCl!XL&5n=Tul)VL z+5W-b{=4mXe-GPjt=+nPt9|ANKi}T^fzP*_x9_%X&n3QjJ11MHo05dHEC!B`=%imE zPCdz2LywNy#~**#o;-Zm{^h^&FIIjp2wsXYycJ=054mqJh-`HU5gCmxp8j4+CCZvUpHU(CbR_3>D(B+z-WKqdl)en+%OojZdT8aF|J}Pl7&`ts4+R=o!{{C5ALY{!67#If8jj3{+w+FKbk~H}u+o#{-OCxkU-$ZF zykRjs(V5EMpsL?I)iA&W>Xa}lLnrC%8)JO-w81k^(^@|7o?Js@87D^5cbF%5>C41P z->O?W6>cF@PZH*GVi7s$CBA~;6(hcYbiFJ^hs(u@m45E9|B^fH7>=+36(Il%-Syj)L%$XFJT{-l_;UKJE=v2#eHa_dupB208UqYZ z{Sm$#Q$|eD{si>&19#AMR;jVF;3=U6>6o~i#^n0bcstUF8&H#ka!bqA#Vn?F{DG@6 zzeyju-UmK+4=U3w6t}i}^v)3)GmI$Xd{ll!ry?J&GN!ZOT8FlD%wvQN{!Aln`yq55 zH~h#WOim^~5|PikA`UL^I&Hr;ekBsQ5V*0au}{ZFN4T|V7!94Dhs8sfKUW3FmyR|i zBgWPkM5Z&}Dt2(fzlwnlKN=y*EUWJFZpXgdnc%FIcyKc?IB*~# zXr5_UW0-4TJKL=U(5KAeX^=v?(s0k?dt?z~ivUWH__$r=0@M@n<+)a|iPrcBQ@8b< ziI{b5Ck~#Mb>Kv3;+LC9@8`IW$^XILynX#(-VTtXkMvvt@dLJ&$If0*7ehPw;Q-8k z8RSVLZF92Qws&5`5Wd|ucP4r3+mjnlSme)@2}i|Bt2+~xJ#OWV zKW=H$p{Gv?ZS_f`Lcfz~1{668OPBe|w@I>kj_jo8;qyrco~viprCtZ_BusjSXyQt% z20h)n-ZS&`rx%ZN0#Ih3Qsz z!f)`Zjtozp@DSM%Hu}kuWshG#)wwz~>N426p-mN?K<&iov61?4JT?~cD2FYxwxcKfx@wAbFa-(G*~t@iF8_|f+E5B+eyAiI0xRz8g7dC*QUijF*7V@&Ille9zg zF@Nh;z4EJwW3od$x-CCh*jeyY|Eg^Ech#@$?>%Wp`%HGIL-&fG%%*Klzjl7kXhJbf#VRW4!z*5Ss1)*t%eX;Uzv*!rmRjrSwhZ7d>IMwB3LuMiqIQ0+wdxno$6qm0 z=_uXqWq(VTZrhK+ZrT{Y+&92?C4a1_yJvbK-+s+H4JQJ|*zY?LabK8WKHifDG6}EgO1&hGw0~Gg$FFFJlUz4*s^BDI>v6%@zbH+gHM^Pn{{7E1INL@PF7jF zxabQe;<}=BFuT@%Yktv~ z-^ytbbuK6aGaWe$@#p10sxgej;+7T##bIu#|IohVL)l~y2G6+2Y4Cale_7T=IlF&3JXe^y%wYk8zWG&>z>uYb_<~O$d*njX{jz9h%49;Xj>k=1 zODOT;#;EIM%DV=p0WXu*h3wE8^XhSeR8rFLAV)|`ySV$!-XdPEamG3f0`jv;i{G%RRfh@k#ggbKlonT?=Ur$WMW;7cA7rZdZz9T zGd_Kyx^H;zIWsXNd_hKh!jy37NiQ1V{7YErv-uc*>DC~OzL*FIj~##DjcpKVpVFr; zN4=OnZ7AvO=ZAZV?-Zr{>b2@I_Ivx!mow9UVdPpiPo>cyb`SCS;j|s^KWdNO`}Owt z8((fGdyjHd+w9<=UGu{d{Y$0F&U476cb!Hxp|7d^z0b1k(5H%Ot|v=PPc(Iv7A5FB zkPU{_c{`@A6f<33dt5s(3nwY6A6FZS`Q9grgg@Z|eq|z0j;mr?kH#C8EJf>=(7EZp z@(Z5`gT~P3K?s1{Oxgl9GHHd>W471HM`O zh+D>AhEL01e3J}?eI9%oZ_6-nmhS_Wd0Tc1pN12S*3etfL9+vZ!C=mYi$9H)jos#ryRN^g+2L5^P3ktY9U!P>{ z6R@10u$a^=@iYwg><#WO^JCPBV@?IuS zdgCgcuIhwMwm>7eIyiY?KqAOyB`@}O!W_PjxSS~1>5sZt$N|K$BH$odM#FE1k!ExE zUVHtmx7)_nR@+>!&usf;BL}ezvM`VpC)|rix1@E2xR|0e`0bo!q1Kfwr1X;7kVfz+E41V0FAx&O>9CA262Fn$ z<%_t`N-ic$WI)v51$sW*FEK7-6UM*Fs_HO2dwO)U44b>XGF>c$|_ zZ}qS6g$k-Lkqo!VB|RZ&10z25S0xP?-0AxmT&HYN0_3By%S8)u0_n*z;mTlUv?XxFkAa%F ze(%Qaz?H_3yJ1D2@U+Q}r$5(7S0WNdUiF30kjdrK4#|pEPd;$lXC%+~Vq9aEA?MS! zelcq^MCXf%$47hZ;e)TY{l_1*j~{%{o<4loPCxuwJK2AdGoDSAZFH{qw$-*t-ku0? zftBsc*XDV_EN4FTbOm2>eAZ4e!jz#asRwh6-@~&!^>T$+bNMmcx(2S34y4^q!xQO|lQjIwo-#al*#|$dDTkAm z0{fNX%?@bsZb}{B}V;FDm-fEjS?zXk98*OdV4~5)m_uqVzHgmy*=BOQc zlpm9%|Jh&t*NUeX1TW1PzM89Ss4(9QRHjmHyPe;jX7~-N6{b?U5oVsxOCuO|)xCTl zpmWf?3d^{I)_`SRmh%yo{a5o_rn4L_Jde(&;XDtXm+mT!-@OjxGz zt>AfSmcxXr{sC9h4p^ooEV~Dcu&e2W<+y}_lAEZxQ+DG!ZgrwbTAd5`Poy`zyS*>N z2pja0HZcXsBaTY9pqKxb5WFRv<@`oHdG9ET_)9l4{Pm5tyM4Fax__r_-`LHkT0Ppx z$G>*ct_+Vsc^JdtHhB=%9}`0jdPW)NUp#ds!j8D|3=hFma=8tZfH$ZG&z~U_B{-&! zLD2CkArK<+)`@f~%aE869ohl`q^(}(5{mzk#z*NwH$F%^_V60-N?hbDs(j?nvR0Z3 zGghwyXXmryC4NBo(qH8d2JxSm%ov_Dg;OcaueeNd0e|smX)fu+4Qkb!q$>V0x%6=m zFm!al;9cEHx}K;Hx3ZM9hKGb+dB9VA7+8EKy*i}3m(O^Oy-bL^-*^okZj8yp*zO2R zkSA*?A5qd~uy(72=jFy$2p=+HUgALu1Fb;bDv$WYCr>l6&a>@-vTH<2L!J^xn=oE? zWYP6QNNI8F_ml~B#Hjc+UNea*$%O`$QgOBQ$ow<$41IIp(hbho`jxgBzDfs2*!T*a zViG@fK(QlB_xXcg^?^a;R>Zmmvq+itZ%bEwrvo@Xi(Tvl){CzY%Z4dY6?FbG$W>O>AOP%oq3#Cw6?vpwG zD^uM1tLDpHdL*8AT_rzZgbCJ(xFbGtBW!?i8*6JAxHoI1_vX!Z^Um$|CdTiZpLw_4 zyZ>gpb?0u|+1bu}7`}Vq_A+}h>%+R!L`|`|)~UAYj?VA%C6-}(??Af^MRGP<<*x#wFvgPVq8YVclVcG?0JaC!@g}^5S zjBd(v0EJ0CugT*SX}DtNu{7dJYfd^h<=0AEDQr4OQjJH>O2jz#-^ikQyE-_+hmzOK z{*z#gs{*dH#?a`vg80&Zu2K9PT7#Em7-s;zm-1*_xi$|D`2!O^Wn-PVB#PM z6W;nNG5TabZ7$I;_))a7oesQr*y@0^9R%fWG4aj6;#XcB2I7soa4nC&lvAF#*Y=^R zJHv-zQG-X(6E0mA16~tXw8C@uFYs`fUVn@~GZ0_DW#t|Ng1hGGVRhI-tT5p$z zcg0Iy4ug|SrKO{%;@=`qJXib0Rmc(*88O-EThRDL;M`BKg`T!OeXe z@$bG+Wxc-cYV%4vp3QR3tp=2nYR^O7SX*sB`p%8^KmUtA)vjY)z5n%x?Y;Mp+T(-c zcJ5@7{^&w05L!7M2tI9N^Etd=Xg@oz4`cXo3i*3_h%ro@=MTHb_>8#Ysh?jyX@}J5 zp6^j;3n;iqLg`(esv{Wk-GF7Z+}^h~ChfJ`TkX!h`|ZuQKil4Z_uaO6V<#WNxcB&kK?NvXH>J{Cy7?zvHYeF7fF)xCa^UIW#ztEIBP z?a6!EQTl<2J>?dZ;}hy_PW>O{1-B2s_8xWgrS|yoM{V!v!)%*4JlxBP5XVOcOolu+ z`W$=hq8-ipaeMH6Z?|C(@3ft*jXJIlUEY{D-mIYmyY24nop$%u&Gx_jkN$7T{YAk` zF^2!tk9{VCbF!v_Q_t=4?Q%O>XMR?&D0Xr}4PFdzRH|+}fnfyr`9A5ncTu#=5HGl( zWoIO=74GuA2WGnuOd6S8jShm#>UkaA#+A_lt?4h0f&5YN$*#^#bE9 z)7%4b;ZMW69@A8~e$~x}=%b-B#t9?b?#VKV*Nav>aMu_bhLE%j6VJ};#Bvhi;BTB9 z36?2zxywV0S;EI-eNRO2WC>xNMiE|!(wIM*pS4X2bIO!@oaH-Cok-pBSNQVVhYs9Q zu9y32G9q;+BZiy5N94Gl*HaQ~r1KnqR%J2o!Mij~Z@6brZ_~;9O?MAsHy`8e5}XK` zPAI*p@<`NReO13DL*+YUA|2iQ1#GOA)t531Gw&EC#6?!bdv3L9G`LKE(mSAK<9ERa z{&Ym0LCdULul$Seeej1jpu2Rt=#p(@t=)KnyiBKz?IcYz=xcb|elnQRPULHpQ$E%* zcoIkXG>9+rsd8>@Zny2tjkdLWgPHtByS_#TwXwyumbeW%Y|&Fr`0@qBF@Q@YFZkHG zlK+4_0*0?-=0#e0Og*5C@}EkAg6YyGp((EISKfux15V{WfSmcSZdDzIFGIkMP7X z4wYKncfXj6pOeR_0&zU-hanTw59HN?j$s`hds_9f~Zw-^&-#EPS>=;7WUO z05L6YkGwiLp0#Hhl+zi;jmGamJ382JXUC`Q>E5HZe}M5jonl;_w%Kf&$L7B5CfIDX zQ>EqO@tvIs2CyICfluS$r50MsXtw3rHda@`Bkm8rbF2LufB8?eH}CGYlcV$Y=rPyB z{dPE`zgpdB^E1!icJDh#yQ3G-r7v>Ku!~WauE-xu58M0*8Sm}2Cl4RBLv-Y$ z4!82+0Q(dL@op^I-VuFscL3 zo2p46Zl%-hprPB!=s}AC*Gdr_gpz+psZIno=rEIq9S^@8B%g>vlTRK2k&p5)Jn<-v2chRN7jc$tywz<7F2tZBy?j&%%1)d!;20V(BClyMPAANx z5JSt(EAc8LjZA-aSk588jxzM1*Xc`GyfIig0~Oa!$TAiWHcC1@ zj9MS(c=>S#4KfeemX?q@g;_;qBlqokJ9@ujw#7`^NzQCOO{rsiQWnoAlp$B|Ndg~A z1EZ0C;fx6hwqLG%2@M4EveV4R1i+LhS3sQHi07G!kr#N=l~!Jr;-hu?9$|&MXM}r_M?fe{?WWPI?@$K>$WwJK+?eeEQ2HUVhwZ zDUq9C8TAZ38-~ePa+ar}QQ_8w=@%IQq7M+qpRm9)O#Y#vH^u~4@f-xMa+SeLNH1@~ zgID*__N^cXt4dq($(eBR>;!e{g|BIqYtlQ!uH9Cb3} zJN*8aviaDD=BY?}x=uHaGHO%*5p zPce?CxLpBsmDRdD$@Yh{jAQHcKN>Yo1pE@LIc}`Cwn@LOe6}_x;XOJZor72Ml{NQN zdyLrU-`r_`^)LO2_QUVqZ#&zbTfPHto{mEqHg3Z69rB*F(-XI3oac*i)5HCCwD+(* zdHk^L?;W%Qcz?2Y*gkrE(4HO~w=?>}82KkN?(-ORo-Fj}=}CJ;G33`5uQ!p&`^;yPP_B^>+Sm5M!SaLyLO%a6Sp5B@;wXt&b75k+uYpDl~Z}p zmGz0M#9M6_qi`LA+HrDS1q}$**_=kMJ_CN5xip8^^LSEDxs1%Y>m@v_+>Y>-czkxc7ufK)wwx zaw0`W;iJpJv}RvQR^=G-D4J3tE{#P6*YFyhDfwGY^GTX2OY!7NcUUla!1Terz;NWA zyutVJcUCWGY!_!W8OT|c^5CFIdT~pZy}UVg2R=wEKh7ja=PtgDu;k&{!!iIZe)6V~ zKL#pki5otkWBM>$e`2I;DJxgW(=u0i>>$n8FYdUC;mU=B(YdEhbei7%1&S#7RtKT6 zFAs*-09<x^%D%Xd4@pwN@?GH+*B+&Y4ae{B*uqq3A0S@DZKi zvN-t{-r<4&RX0^WCx5~slrZtUM^5;1OWh__bTeTBVM9+sPaMPerM(jtTA(U_A$f(? zg32dg&=P_X`gI==6$7MGE3ai>2-cZABwZy;_~30oj7oU~7I7mD?&1+A*C?Zpe2c#A z5t-6oB#x_esnQhxg;&xZZZ6U*3Q?pPhY>=jm9|N{G3}r^g5DO_sPSL z+JpDM(VjkeoR#3c{ihg~Ph$vsR@DKm8(UM<3|E)Qj`uc6|7x&5m5HJZeuqdepxD-UsdB6Aa~}c{@Hn!eD*^&T*T~ z+*)zo4ra5qH$Tk6f|LEp+O_uJ@k#sTkLT^kaV91D_95d)WK;0wUHE_PK6LD3eZN8j z)=$-U;}zj}LXV%W_hsaD9duu zE6HxJu&&jSF^0GRWz|z5si%MnC3JtaTU}lM6I|&W!=w9&t#Fa-o6gRNqb_Ug@p0=kRf(%i_@p6n6pmhvLYqn4_~{{_v2;%dm3ze( z-O4kp5s|?}bOU?7F}Cqmxc@1;gA6TN8oGV^S)-@2gP(MkIEk+mK<8!1N*;oeaPz1R zE8#tnom<7r5p|J{mO%*m20y#7GZs^vQ6H7q@>phl)j3yOm0LQ-q@+onP9%XLJAb3$vL!~bb5v`k!0~8$*4HKuFw!fYock8s%&qa~Iy`)m` zNq^;fNt0Oag)_=!`MoN75QuRna_|dB$mBQ4DvexK6#w)SFJ1AA_u`-k6%7%KX4ehM z?!LynKxb^)HQ}TDNmglUJSyI!a{R&@{usgDnY3hrVY0zP@Joi`j=GEy62HL`^s0W} z^tBRP#;`@tUSN%!j+%r57+X+rj^CZ!Izk3)b@BP-- z+6NzerA-fxY9H?GB#*>3fGtt_4Vm#ShtN_clC*9z#*$BrVdN<9ZZ)*8m`pU(Oe40X zqaPyw)B0p=jAi?tH*asYAAk3Ld*k*_ogmA*VkMjUd51JiLtvi*$Iv-?5nOOZ)qH_hvsCmk^9yLGF$frJZMqhZi`EP z$hdUFYXD*T)t|2S1?ptSOaJ7ZD`>Wa@|!GAr_n_zsSEap9wiwR%6(1+xtBa+5PIVy zQjgs`Ihp4y;TaRHjfAiP=N*aqf zS4YbLMx$9~MVH<*Dt>;o=(a;{rRR7}gMRHO@URBCD9rma1bL ziFK6%Bwv2R@Eb~Dr{Zd9hOQo$_$X3P0_ijfmvjtjvI$`NR5I!>9%kV84ieO|?>3(rWoG7zV#=5JmF&x@juKKdLdzz9mc{gfeAP-Sp=96@ zL(cN#TL|xALjFjfk)%3fWo={c83s4ypgegD51z(OI%Y$?+cC;f|4O6zn_G zw394(^{>!L%Osue=!23Y&aiwjM;zjeiF~X#u8p-#I+)seF+KxcIfF|cp()>9NmpMJ ztUctH%#Jdl#$W1a%1u=?WiXaWrffgPOR990k@j%bjQBg@-6EHa` zJu%Gs7NVcYR~{-~vr9)_0!z8V91$HgGAJlIMMSjXM@Zf37$|`B)#2QYmoM(3=%&fy z)>phpEM>9=DQo%yp!lrx*OXzWk$chs(o^3PR{l#~I+KA4?4%?Aw0m<#4(3@rC`0)* zo6aA>)qkjAChmMJLHyyN_!a8^z$_UUXozuDTSDa-{LY7R2qtMX5)u{J7(_nm7xkR9 z-Cr-Z4d^_kUr9h9ea({JCr{Ru`8qb_D|7Y_4&d^Vy0+(%9zXL+o@7s4=~~wxVrm?* zZJR>*RZ1l{`b%Zct>T7^1@HXHp5>{JofO@QM&D_J`t!0MMv^*e+p!;vvCLh*#7Q0J zLhB<#S6zF(b9Y?SZ+-cMYmOn}5m|>%AGJpxe$XC#{cG)GjFPj{6B|_wL(A*7H(#RD zaLraT@~?gCmaz<9nri74m*sO6xyISr@|RUm%ILd0(zMXi>0Vo-j2hf-W01eC_3Q03 zx3}7BH+L}X)-izZw3|0?WMy>w#=W+^z1#d!ty^1^WuCwx?RC<*b;EizPmdbQNfWbE z#y`i#c{@1TZwJ$pHg(04F~C*rr;nbthmQ~Pj>y@`MSFB`+`h8EtXJO|SoYB#EU)tL z?$3Trb}A4o0o}g?)k0YhegakQZfIpV$|BhWtj`T*v4oetVNT*sgS*#-JhW z!o>#0_U`s}+w>ie%^eKs4Gd>b|Iy%fkwe`Kli^RE{9A-~P_9DCLEWa$w;={d)ywF| zs~lgb@q~G$zxF*U>7ui?VGZJQ!raD~BV5%T;$mx<1>vcy=|S7yKWPW(!r}gY`*;7< zztZ!3G4N81;eY#I|5y0mg@G#_L}$bQvJ`>q2B(y#!V?G>{|V^n)4_4KQ*aO(^sLOP zuuj4{(be-B^v1vxL^1GI2ZNt9HRxBP_go$ggJ+h_AXf3i_LsDw7{ zOWdR@62Zv<@FM^)UODD~avHY{&e{I*?99$f+72=)hkkyRKRI>YEPE*^;VQM^qY-pg z<-tc`sWWJ0P(}vX?h%8akd;~RBRk?c4ti$BZhU??5aA^CKiXKd9;G7ci)aIRzND&C}5JbeO1p$}ye%^2h4-WR^=McxMfNiJXd5@JMYv=XO5=iXNHt_iwFtWC4}tN^f2^AfmY z)f?L@?beO0d~*55wyUF?d5_`N&6{m!_h#GO!Pr}O)!fBg>#?@X`Q@?`%5#3^2O*{y zR);xm&8M8!(^F5FI67{#6AfnZ&)b7X)ArG{wx12-Re4{wq3<0?SCDqbVZg0tzNg`2 z!L4c<%Wj3RFViTF4S{93vF(e97{|ozY+*dFV+>C)e782*8V2(!@>pBjgw_V~+F|mb z5xm`Q-n!Lp+`N~#)%9A`P#e-erJUTv8vK%~ZlP~p;^)P_q%Pj$l0;%H zf31$#87repeIUByCY8JcW+j`6j(gyD`aNOAms`-4l87I-fl{urMjscv7R_=C()D)P z2?@0~%P>84GiDfTqr3qe)}!D|JjO6##!K6%a1wp%?ZY^VtzVy7gK!BY7vy1kv!fGl z!Sj#@@kyU3@Q0^ZPoCl>t_Ny(Tk7Zu5NUv@ju9u`6oV75WsNtUO8_nOdU_%0JMH-5wG5Xy72fmWFXgY`+>2({=L*jM z$k_PgS8|D7XlUaqdR0d8(gr)v@?QJ?%m3RTJOV=&`Cq&QOs8|XtP4`)iQD`~tgER5 zwffoOZy}464L&=OPv~bbb1P8kIt1qvWl$y~ZAQEz+m{E&QG$PBXTuAY9oC#EXfOj z(19j0Hjn*dSKI5P0u3dPGrfaxcXQKi1siQ=yY|caa+zn)?(W{mX*c6Yu9Tfi`zz<& zo}R=Y)~Gs~a#0qIVQ1e@50Bf2kB{2n@d?KF^>*ZW#3x+`UJzIpBR}2tW2=5z*_GkR zB&)dfQZfdXzeo8Nz2Vj{n)e_-Dm@VenFn;lJ_M|Bc);kZ~y-4HdLOm0xB(le4O5;{OG0q$NP0)pNydQa z^a?sH{uyp^{ishV)&jHWn$IXe=zyHZgWNA^R@}G`yurWzVc6Ljxk8q^ofqMYbn*fJ z!x*j%ieCo;a|qvhNf)EiL80eY2$d{;T-KL$AWeTuPV|vq8O+MA4BCQG2Ay5D z))yhQs$G+$+A~-2)Jhtb^hen%AM4t3StqiJqw`)+{OW~F-T9X~mI4VgU62Ry5*p3| zwFfqsDR0qo5+Walc7TqliieV$zZ!tNidR=uMj4VU|4E}h0OAQg?yJF=F!J?L<*3sX z0oJx!|w2e=TPj!=AuW5-T2^MCps$G-Moq*9V1_JK)_F48s;RI2Z*+Snh<6 zvdLcEadbK2NEwqK;mTCmn?$~mQWBapoiAW=L#`2BbznB7f1%A4emYN;Uyw{_h@Pqo zBYNDZ=b_UjyTUC}-V0}d{=(_$xTISgC$zh0BgV^b#}|@>yEGQ3G=oiXkP_>c*P9acTXjiPTbY5!kcAOKZsX0*RT8SG@+;I{hD;? zHBhB8gezY{Nt^cxLd$-z(<4rL6+gmM#V``Lx4aB|j**MOJDVP~;{#6sc-s6ZhO4%R zhx_etdK7~>3oWkR{rOq?ysNO#FUm07-%A*DtFF{p>Sle?wkVeYc>&Jt4j90( zg;G`GDnZgFKpkk=U%ZLo?2FAFdAW&kyt(Cy@&wuJB;1$lG?;JRzSZu%{uc2!+Krp9 zW#4iwB&ZK%Z}5+sZESe`SCDs7mSkg}Q-IuEgh*Qb#(A?&468N~o)ZiGjO*$t@%;R= zF0{?GT|8aLlcnrev}yjrfBTP1;f29VF^2yS|K7ieun!p$?pt|>{2-IM42B&H0 zI#iV9ib^`~39?SoKn;&Efsh@5REA_uxUMn44g+@xq-cs+G_Dipy?Cljyt^t}{t63) zPDg`ftYY!6I&S`AOmQc=%7A-xY@@8YgA=}sH;qDq5@!Z6;Oz+LU{oaCnO)&G|D**q za*z+NkzNP#7_`hYt5rR(@Fi~UB3C;3%!&YwHU@8Ivcy4qq*11I^vf_>q~GZ!Z~3ud zTz2M0``F8d-~g4up|b61y{a)S_~UPM`bkfDM?U7E62uq$G7vKe85b1muZ+FH3nUav z(zya+M+i{&CpB^fH^?yAzt4*p=b!>hPFg$daXTAGChQq3}VNX@;JqE-UK}rN7G)%nE>64w;jBCX&Ec;Ay!wvQYb*$*TGK*E_8YhDwc6fsHC6# z1}~=9-9c}1H@_eZM9WLwW|Z<5kM$^;U*|*o@C{mMC7esW@WwwWMuxs7mnbG zDha}C9uh44D#NAWTCx(Q`VaBst71b$*~Qh}DML*}gTr64l5so(@3LH-T5jT8-U3}7 z9bus>^5eBkyNvvDk209fl!K&cOuyiJ#2cm?PjI8BmM6hnmA_#BUJ?8Sx#Gh|M=O+y z6em1GIdU`%cj()O-aZg{{^nXiQ+9OB{ivOu zU`V>3_~@w3T_JXb_4I`JgY4}*JE_&Xu}{|%9cCwfh+_t6+Q``{eRA<`-{sluzQ8Bn zCubgYh9Nw^z*u$CU+l=1ZvS~M{r38`c4N|V>{#~FuCK(<+s5z9X{L}~Ue4k&pX`M` zndcp$Q#y7HZ%;M&=kcr2FW|H8fVP~st?`*d1t&O&YduwZ}-MRaEyZ_c_+xG5F z;%{ZowcA>R^dZ&}#QiVCO`ggipCLE<#Hd@urzuo>6iQ@Q;_TrnQD~_L?jJUhJ-$60 zw`Oq}E2pZimPtcbiIpx+q9dU5c7{P+Kz2&5{aI@SoZ3-&4~UjL7Q zDxGLa)965yES-xDA6opcbQPY?8-qZ|05Ou{9v!2i&{g!ov-9b3LLOSk0yl%r)6U0J zE26YUo0hBeqTC{fzkVOESg94jldgFWjM4GMK-t7!}c>vP;f5l3_V5dm?E}A=5pz;a+l%T z0Y~JOu--0lD+%Kk3qC^hpu0P*$dI2jA{&95bW)bk(%3CRme)aKkd3S&YiX2j2SBLu zmA+L1X@(@h=0{NdtGYEGeAUO4Y!XU+hqw5PRMkO}=3km*xTia5Yj$76guafZHOk{6{_@ z$Tt!34|#a`8!xUZJ9#Yw9$9BX+Us=@FIb-bL`DF=%6zHM%RCvEd~i$3d*mZe`f@Ml zFouV0eXPFX$<4>FFJBrKLo2gNv~mDPwGaP9At}uMlVL~ zIR@pi7ydKGlaraBI-W7M?BkwtcPmAGkxaj5@Vj5w7vb!aPUn6R4!My>o^tupZe&e; zn->Q0v_rc@$0~Ynb-Qz4B|J{E*Y?hooj6RfIKwrn0 za^{=Ojm>K=m6!h-ZE9@U7lo?>gcqFKA29K zEe49gqVys=J9z^|=0?6LN8Hk^y74MT(VI12SdecnQ6t_e4-AEul+}ABDH_tk5gz&H zP5RT%&}5olOQ@B@Ul;wxRL?k`KPSRqX#8D`SW5z5{^DFrw0=wZeA3qAg&zRI__2O#7r266Us6mf4l;Jxe02v>jfEM&tKEjACX z#2d#Yz%KsM^Ys5f(ut$Pq)-r&qK6Nei<01qEgb|#{zA1YeHG_Dbx&9f9_5_Avt;5G z+SGO2^6G>=hV!7>ap0qL2LI6M^i{g$$1$wZr0hx4Pj#?92+AoF#@T0R6d`k;O3mC) zJac;lmj-e6SNoN{3lWB6+Py1&@{x~>XqAiMn{{hfPx!P8hyUqsn`LF(}E zs69DG_9&n$x$@v=mOT%BgRQGl}~igXN13LcdPyq3$e~V zaC-_K4QW4@@6*uDP4^6M=18@hH*dCEckf^X`{JC&@U3?5{##u4!4Yq#P1YxwXbezq z%NQ=X3Fe_(^+%uO!}|hA71U^_jt*UBsbj;dFJCr>pDn&4AY#{d>mS`y2kbB4D8}%A z@Arc-{D1lzf4vkk@+!gW9Guu->XC3kI-?$!hOMAYzQlTU1V&i_h8XANEVDX^4EWr| zGdzFgiSpYKq#-2fz^(XnFrzawi13Ah_X#@Hv6|j84kN=d){NHj5l&?rmaP%o#SM*~ zM&9K~qoxQJ54vsqRc&^-mGAjQ?HAUtnK`?UQh@=DG3ZvuS!GCGrRPOnx-p;%H{=K4 z?!`+w8*1vzF#gq-&n$zofFt?roQ*-laz+L;V(&4Ykd@h49+lsEcV_Ii3+c#SfVjRSh|SJ}cBd}Od7KiwiEPxqA+_qE|abu4~`hBACe zsAw7PKY7wnHN)~tU$^DdP+g@nNQ1|n!I!#;%*5&S8eP!e%ka#vV}-VE%Vyo;F+kj6 zD)fvuEONl@hdPvp%uClgRvk2EtOHk+OWz=pdaLAh@y{}t_fneR8s6`Qsr3WDsEDt( zaFjO`t4!dfLllfJN5b_gn*pg8Ad)zD;^JxvIfI*32Gf~^KXpR(@=wy{Ti#@|6mG!K zO{1z@1Vh7L`Fi^bZiQFcMD|;S*?}vMY<&>=O@|*nS>%}?pr1H@aRSwL$}sUKPla78 zB|7axJhDoN{vgljCTTJaGVLJPVU%S+SnMd$FE9U-FPM?V5|Xj%YZs)U^&*V?CCEhK z-+f)C{za#C7}L{^{#9B5@e-C)Nv)og7dJA>ig2$x=?AP+`#JGK zt3Qn!^wTF5&9o)l)$bBTI(d`U*rq0G@he8ay=hxZ>;<42|O~1d#6;ll0@GL#~+u zt{3*mV@7}Ei4xoE7j1WEJG_q zk=65-{fzS#2JxM{cXJfo?u{J`W4|K0({9|joqfZTiTy_3Yi$OAI&L4Mj5A*La@V3F za;5Hs@QnWxa>AWcA`I%{Qoa1=F@T9D(x(JA`(Y5DGM0sYexWpI37ESNULcW9Cp~4F{;nL+A80kQ` ztCz}m)ad{_UWima$|7-vAnw_@#|bg<>Vr@XhL%S|xlam^qsFmi$zw+ACUO;3W4LrD zOetp=FunXlF7RH1g6TDegV^awL^+jl>^*cAIObL5st7@2y6UNyr2d0b`iNgQH~p71 zm7}sNd@<#-6xuW+wBmzbF5JqdQl`G)q4a_45{djYn5w>lFOEC~jeHik6()^Z$*qV& zS8n8>?os!nUK5$Jbe)K<6eM-$H!_fhLEzT5iR!Nmi*dyuky0#lD-ooZANeL#Z@=+Aim3@+Gm0K z?YW`qRrnE?c-7uw*^hZdlscf1WSlzf1ja9J37PnDCn)JsNEy^EO&rC8`f36B1G%JS zI{C3o_>$E6DR=nh62nP&aLvesDTYXfRlk<=t5=*v8buzIOZ2K)E@TiMxqRWQ@-F^D zpYX8J^`r=iBCp-w+iv824TD5OC-Dj8Le54hz%Fw$RUZY*JM_@u{TDJy zxD#=VWf!*mh{q`gug9L%(UxZz)P4=_i1ef}ALM>^G9wP1A#=x6+n#aJffQVwEZxkH zyX78>Ek0(?*m`lT&B?=+;C$QxL%jM2`c=X_y+S#hkf(czGk)PWjddVR@k%ZcN;`F2 z24v%TkU9f~oR>k~d;+CG(;Tb>(?QCwr`T26*o!#K`lhOxj8X>8;p z8+D?@WCH^lW7iLpc)Xo-wzjw17DjMB?d%J5UT8XY?0|5DNh)kaoyI)$!@2I7w?7`l*V{U)qiGuJ?t!N^k$WKll4! z&+yOx{Le>N3dhE^5HQ!GcT9hSkVEBr-P%2RC1BaS>s zvkXU%(HeWR@~9I?W8#vAd8YFxz8+Z>-BAz9#*T4tb)5mgdkiyO;&IO*_}+_E(#mzE z65l)p>1Uvb+dRM^QDmYYUS%g=CL~Nyp7I%4Buy`K&&*rM^gK$ARldy3dpRo)@g&pB zX9JGxm1!cNTy*{R{E(0F{)?ZV#}Jv0NMuq2ls`WfQmBQSygR)N6d_c6f)#!cVh`uy z9uBxK_|?r7I^CU>wr4eP7_N?n9}+k*F-(vHOy7AMm?>2Z&f;T{ zr1M4g#LIsVHv(Ft6V@+};X%^iF+5dF@G8H6?T~a8&b{kaAb5peGL=5knLrnLr@eV4&URy7Rga}J=1a_JeVVsq zYLcRFUswK`jEyIJ>X<(bwet70;G3q}3hl{A|H%uN<%^ML`mWb0GifjJk%@UjyYR|* z3Qx2#HAkY-S;2@e8kJ5uWiV5J2{o+Z!=JPx0|K~)VOa97Ye4#^UQZW(Q8oNkdcisp z|AIcC#*C5HH2l}lTc#^5-NsV~8a%=%n`O3-PXEhYnP%coM`hpV6T4AAi#BRHt_#mC z&RE&yY#7n&_DQ{85>Gu5GVXCj=FlXn@}qC6MIY!}Zu2ADx_2vtpXGJp=Y$OSvOLC{kKw)tX!7#gQ;Q}CX%m`Umx%IkiL*`w7ombp(faMuVUZC+Hz_+6WLyj*>~&NG`Qs~FDN zcU))ZTR$4SzDQRu*HySL8hY0dcM+vj0J-#f6&5eLY;lTJL{XBXGAmBpXq={|M~w`?k@^niZT4t zKmC&>R76n)FX5<2hN}P`@Sl#&&eV$O#-tUN#;k-%7XbsK!;w_n+|!K^AK3?Ip!7kH zyLk$B3@U+!QFWlBvr_TFOlN?CqnO?`8f|#;KVH0&R`R4{N=FiZe-uR;%hB?jZii52WgJMtC`5^z}56BljA}3;|!_ z9l={xJE+SQ>QCBcICRUuv7fF63X@l!lDT$W{2CNAf`ZG>mnjF^Lx!UuRV04VeQ8=( zJz9Q7nec=UXxIKp|LPI%;NLh>#68L^B;0ff9P+r7i;jvLGzfyOliO5o2X!I!;Jq<$_+*qe zmp%nZzXc*xYT$>8G!rj6clE=5)$|%KwrSH;n<_ppf&8JVc_@Lj6XZ4M#bChTl(`tO zJwNfH1J)~d^BBgI_k3#BHVLl%fH75m+fwO3;hN66p$%)01LmFl#qp2(DSWu9i13KB z93>BnGV)F7pp`038LQk9F)s87?^%bR;Tgt$icPq6?0vaE6U0gUqHGN6T?V5}_B--l zlWcPiL&5j1Dn47Or+J}kUig%!bXNVuuExIPlB~)+^uWG3?VEf>kr$svh89=Gs?LjG z8xTi+O5cm4e8QRVY{!A7DKbWNerjA2&pM;e!1!@5v9emU8}b>e#(s`D#&A}NU0LmI zgKKtjkk#STH*~UE?8|ug%|pCeWrn`xLT4yRWV*O|eyMu>;=zmV=nE-)fi47rl2O&Y z?xf*LosrHuw!b%tZRzw>TIO0$7+n~8c44`k_|*!$w5uMF7w*Ukx5lkTu5!-mE^d#8 z^L$`sW!UESR*dGtvyafm@J!)N;;kq2%9Y^FZQQOJt7jO)8#gd?*ORA3vvi7oQGeJ^ zsY7|AeaXYh^|G}Li(Mscb(zbpBsF=pJrSNg&Cu~YOVZfp>uGc8dfKDoqp|@dlOF34 z*+E?NX2qX&Xh`VbH{Cd6*!|zaKmR}Y3%S22cqzv4-P<>!fC^^} z7|O>Z%6vJ?*^J-T_7U3*;jI%E;c}N*2Qn+iib!K$fRzRYVOC6xQV5r#Qo*Db>@HN; zC^C@F$qFHDIyC5}0|Cwmk!rOqhPDk0H#pLnRgqe%9S{XR%b4CuOtefZfUsY7exk=Pw7uE^9Zh&HGES5%ci_DXb~Qz4L_Dq z1s9n=9i&NvHnAvNX-q)^mo#M{UZxrD)T<;!O&oH@p|LMsIv`*%uwa-e|KL`VgbUV_ zgl*i`X9g5$C>zt}D%p>Y$h~N5X~;wbt?E+9Q9~vm!mxA}#})g^&TF(8M&sy-ey=eJ2%fwz@#Q5%50R@7BZ)h?#;OOh`mii{OdP^Jj$6qc z*(pO~MmpWS%-O=!<5I_#3FJJD3Uf{>z@u+amda~rEgk5CUSy#z*{790S^TP7 zacJW4a=bMRCLvm zYZwu}HNPI*&?mj0)YTBHHeF?N#m%Fm)FkDZEm-hi-P(t(PD&waJQa)3@@GZbLK;wbrl{=N4PNf zC@MlS8W{qV9`__KHk0Pzt`7FUNMM~ znFggsZq?OA+UzO@uUir}c zz=pmrMz5^ecfgCQ!6FkY-ROWTO7Na~CD`_$PDa;}amtJ=B74i`y{EN2>6=PG-L^tE z$y>R2&UD(3?X}f}FPJ7Ez9yF1c<@O+>Wny!)v=X`)Y!H?s~f`q{@?mv<^H1Jr5M8= zr0sHe85~p!6{qa%pq$NGDb5~FtQhdEyl@#^vs1NkQRyhfIIvq`ORTEE4%LdOiXTNB z6;uO(`4-*sRsLq-3{XYY=yvv&LBt<`9LllVNsF^zdLt5zZ9Pi0hzEVtN}lV8oy~2v=xxuqD zG%f=nv36Kl)$1}dFXO|z6CdgD6F1Z#uA!eggkFq({o-+nhXWcq0PqJ5c^Nn{_PCF7 z)|I|3c?63x*?uccC`%&l(pjNsOJ~qZ&8PFUll0!}1wIR)Cmk{y`L1A`)pj7#q;4Fn zNXgyRkJ711r_$?#TkuQX3GZN?TOaB0!@DU<{+1~$lE8sY+|mW|)7TVyL5w@NzTj!l zz)Q4gZHpACR`ukq+Ng%XS!lVPr`oFnM2rG;ZINfSQ}Ik|W*U{*Qbii&;6PNL$}OD{ zltosD)F))B!7DBElx~<#*(s~WPuCaY$^e50WUC?R7rbmMF@}}DJTXvfFj2zrX5QA7 z>8%eTX*)lHpNlpR+hfU%{K2)qEd;{wn(M$1q8A8{*6KmZra-J7vg-1On3w%l@%qq^h|9Jme-=4pAW4C0BY z9ICxZ$2wKLFD^P?qz|rna+%gPWEx^3m!eg9Mj!BduQnB~BxrWI*H(w>Gm{ol*0J?b z^pOR@8qSm;Vf>{|EC+v!cH!!<2Df9rJSjT!4_{pJYTuxaXt;*c&Wm|lZWjjfNp0h} zEVuPO@C0#flf_HP@A`^;5Hz%tpy5+2ip;~-##yY*madq;%er~v5AE!rI>S6er4=0Nu9-0H(4 z(}TnIbZSyoRx&c})zyxXZ&=T!A+xcK%$`pbx>!IRH8%121U zR$p)z@4cM5`I*+vOt7>yC{^Rn00^HNFw)6BItC9{7jE3R*)}&P;jucmdI81GRxoYJ zNuBl}_enaQo^=5pamGMcD*!zXf5N3>=Z0I_)iGU`!=HIbQ#lv$!ms>=iVr;?JgV5l z=WTGp!%~--ko2JeDzdhHD(?(NxsoDo&rcqsVLEKX{Z~POl1DhHyc0J5af|a(9@b0v zujrsz7Xwe6VxjA!1q`}zJrQxmrx*I=e{tgj2(VniL-szwu>Xi5XddWI%H9WKN)>;- z$58b23(>8(032fwb@kUW%dc|R5Dx#)Hc#V*u9`5^_*Vn$g5o70XUk-oQAgoFe*0{~ zmgvZrhLL%C4Fk!v)xPytI}v2XHfTGmG?KAwrE`MeP9*$bLsqc}(^yy-4t;<&Wq1@H zJ&WEN%H1~OB+B^gpQKGi9`eg&|65;wOG!d6tDWJ){55KjV`r+;p_lUA^OTP1#8BSj z0~2Lz%3rnp7)a=(aaCs1cf8J{G9?iD@?$v&BV0Zl2ZkO9m3PT+7#3bx+3oSt5;T0I z<3f|9OV&i&^uh6RoGQ0rRc}dc7`H{vmW^9vrQT8>1oe2!QFZLu;DYig9SgPieQKI? z_!ZIUlU?A6EXC)~iAv>dnxYWnRhrzDncFI}Juq}#Wh}k$Wv)?o=A%p!9{T7LeeB>t zsFh>lo|W-u`%K+(H$Q3TsSo@{UZh#n2l+(a-RlC42CICR&N^-@Uu_)1QTDqp~} z{DN|`O&Jz^cr_2pUz2`vLMJVgb*EA7%ZM7e=DkXPqYQIw+r}p4U4tL))Yp_r>fymr zn{juH_QN3Rq$h>A`l~!bpMT!fwbW7C0s+7mf8$rn@q*x`7{fPrx1tnwv~~_*p?3sU zq>$Cu&YAmB)cq*B@l}^}I#e8uHvmEzEe@s{F{1;nhG$1qN_agLF$UO?pMEQRPj5VDV>uQtg;>+9E9hrQo8y_hX}!}WSUDFH!c+ZXpmB`(zJ8S%Mau~^|1Mql6ws@4i-hzj?;Ad zJvJwX73rmsQ!j`UJZi{tMK`#sw{~iFVDhNmZf|e54buB4gjc{A^QhPjdB4qx<-1O026HlXQA{ zlm*`&gX3{XkqMoyqpK~^748^eTy(WkTqScjPigS$?ghM=E- zRTUx+$_Vq3&oO0IuA`hokH%f7!MkKR;4*F*@6&bAIFm2y<2Oq?^1KSq%i|Tig6{^@ zrR+z)RO{$gF4~4*AEN=PK{JeY9Q{_t-~hF<`YI;@7(?_?!a(t!{!BmQ#aERzN}T{` zR2h)``vj+qdfSZ?B@8**x$RXF_AMG|rZMi^7aWNzGEk5>m$>ppD~Fc(7Y~Mcw+t57 z633_|O|~vVGe;UqV1RTNZP-z=!!iyaj*D37y+F60bgT#(H01qNJAOIvqUHWljNzTF ziRv9A&n0>d5D(JU07(M?CmjGk71f5|<(ZY4EjnoRhAv+2H0(SkIH{Rk9*a{nquoj& zp5ZnejmCg3PJ}4X1qNhX$8bK6$YzlJSjgKc~1IV|jjjlG*Nt2aS`NIR2avlp%x< zx~_B_?j5uPXgaI6jAi&FyTB-`_W{H3Uk>|z!pQ$BANqx>d<+9`z}5I2=8ZYBM%zHwu|h8WAf)@&Fo^xgQL;EwxSU~$&-^A@iEZlfq1Ju z?=PMoYp}ftb9geZtUM7nCORT!@HIiea@U}0KI5L4j4nI%Nh3DzIeAqsWGnvui zEYlb`-!3et9eH|>k&JtpzOanws%LAJ$y~64`gZSF@ zw!5>FRo7#TC0B;;+`AWp%a!1){Dl?#0hp3 zvO44--}(veGP{H!w-*H?zpFf5_#2p|@^BTtC0K`vlrS5*h3BYZsCx7-ujyfo$6 zHY{W!DldJD5`^0E^fJ}1+U;`!?K=V)!FB@kMBgcmuc%< zrgfE${%?hI6}}T*z8Ngjc^+TS%j27c3*QUAUctA97cJ+PvrXcL+a$=AKLY;)+U zDsVDDYizlX2IGlzhI-$#FaGTA zq+~^!yzOvYT{ErpH4HRvW`0|qFvV_}W4uH*We5n7xn&XQ`v6yI>o;qIItwv=G%hXo zsM~c7qK%EVfx+n`J-0d}5t7a-B8d<7_uBymD3OtahUk=XyPE1*k9o1?nydO6z_p_5 z3a4_5ukWFYMgp5pVp1Wy!+e6>avrG?0fAX zU(5QXREyW%rXz^Hy4wBT^8-)!XuJBQl@kJ(^jSBaZ;SAw2fm1FUofY=D9;$^;LqVb z``Y}$$EeGqS%Mq+UUiSWrDvH3$cOZVrQmN12JKhy3cg!d)=^>T^HmV%YTUA(4_L-~ zUi^T8ca`o)a}}0pKF=+DE5EQz|9QB}X|B2lJTL4jEYn_wPowz?UL3q=xxW-+`0lNG z5lzS&y>P(T2x#1t(dXHK8XFmeQA*zo9Y%{rgoBpbA7U&6bmsS}k8iz$=xYs7v+ub|2+t$u@+rZEe9BBQn`fN7MV`7b)7(b+wzI^I% zJA*5}4(2{`mu^c;PedRP@@}z)BRjH{KTe4y`UWAZ`@x5Xk1x->(f>7 zRg9c$w#V@G1%vFxg&z&iyqrkcvK4}G4SfwaR|a7?Mz1S_C%U*c#7Nz}c_YT~ z{*(Ro=Kst%Y&eVxyalUMTOxA*i(d-CKUWixE7*t$i;tu$`Ykj~NJv`we3 z+@c#~s}bjZ+%*kZbjfWSdC7rvmR*R!tNvktJGtB6dz$p}t&Yv6)8r?NH0JH3!*qie zA?N8LAB|+e^sdO~je7jnlem6C2}i;#r_4?<#>dtZK|ISUh%+W`Bg{#Yko}azZ&)We z0&KO-j^|XEli7I;Xyt6O=h4fHeU#;zXLJ&-UK*Yib@-AUyj`a5d8E#uAtev(&zH6c z8?;}+EBIc6`nU|w)3xQc@$KRZ#x1))4VLK(%XtjvXuH*S#rdf;uhRQ=;d$x48E?Xf z8{y0T?}Uf%6#o@`_wb_S{8EhJx9;DLF=NJFs%Q3aVx)O#NN8wA@fc(rfRGHh#j3`L zY`Yb}4?qZ>QS9I~jdEFP}Kn1+}r~H7(UK3P`mGUj*;rXpFN}5-)ycL zLeJVU#;e7)jHb6d#cK_?u2j?4#6ZkGX$E9h{bD@wb3&1G3PVyOrsUy1+H1-;JYe`8 z&)Om7@Vw5^$X{e*cNKhd$Cc9S*>jt%BCBOgX;fv)3A{KF@`yRXDpV)3f{KASN2YFR zkyc*#!hi0{CI;F%GUf!dFS^MbYk5g_ThFa>Vj5^D0*f5lqLAeaj>DPed zJBBxT=D0iL?pc!>yA$7@jZx)_tOl|Mm#ebwA+K_j5r+Pf&I+|nsK3tY>cD!#C=Ye4 zfjZT|1Kd;YgiAhlDB~KF2VVr#h(0*-^S9^#`FbRlIWBwZELpKKKmzJQchRihfQy#!r+no~!SE_^= zPG-ng!_~SskL6WoVOc+QYwWvqMyOTy724`H@*zLk@+oc9m14_f9b3Q4>20U#$p-0s zNm1U6HctF}K9~UIWc?^R!8<6K80Tfm`3dF2kf*N3gyj{yg5QlW+VS^ON3XV1LH}@j zyV}P-4+j2KxXSx-e%}o&^Cf&5Z@<%gUct8yFIvto#Tfp~8+S7}X)L&Eq=LD(aFapZ zf!W8M4n%6Ex|u;&H@B>+W&lTF88nB|I%rq82SCqbl+NcqK6OBao{ycuu(%q$mbxxe z6b+4G$7%;i!!#~6hVS0HoBhK%x(Iv?k(}%Rk2x6t{_=tx_!<bHjQ9Jl3<0gC}B-tn<3S?(KlTLYXpfcHUM+^sEF`s#78hEeubu6Cn}96jIF zts5mT?$8`Zxo8Z#3hB#T@^d(w<#YlEaKSvjAMm_RWD|pZ4f(o##3N#ySZFLN=gDL< zylV(ra~d%Sfe8imit*uAfJ2R1!ne2TAaiBsX$#i7dHK@U1mo0IeOE<&obFL;r=(Xl z8d`o6UKKhgjq+BXKCQzW;ojXYp#EyS`huACqhGLo)I0fK^|PA&*qLnqaAHHgPFtkq z_6E-^u9ab5AX_r3;6(#$Iz6IX7^H-mM~q=E(|LM?A9QeHb&|cj;H;YGBz9(P)u|(O zvj1E^xjPA{)oVql(d!DQ+cc)!F^IF`sk~h+F1ZYY*SqP4EWD36{es&vhGA%4;wbNd zx8b%Ic=BTk;y6)rzxxb>*t)h24#RbswtivRJwW+7G1yL@LV9?2;_1tKPMnp)&~qV% zH2Gv7H?rK`*{ZU_x3q#!z8botZk?=4!ij|WYYflnTbx5H$C`N8lbvP5`xU%`7X_mY zely)$#`!e&r@?4@SMMWESWf?~XfCJ!UctA@bHKn8zSlgwg5Mdu==yjm#_(%*c5`>H zs|Jj#KtAI2ZFR;yps%!|s1Ax6%-LHkr#;NOcn(-XR*(pDaP!f*17R)(rlW&{tS)Lm zXf!*(jZ%A_Zw-nzNCvxXkGQ~ix!!U}JI2cGJ9lDmXk_|^y+*KSL*^6-2g6x?Qqq^V z6z8O$LZN}-*9=b?Of_^)&Q9tvJo&9rX59-T7rs67c3m3K_t>-(iZl!rd31GDV`{RY zVM`pw%K2Ft85$gu4U9`y)Ve&pl&ve(?ypp~G0@3tj^Ih(7?li36nuHqxO?>IVcS1EqU^QJ zVU_e6)N^D^bZ~vd@5;P}s)kHfB?0ehPV@!H>Mk-OQRu5Lvb|is=RtH37{(PvB!jIn*9Ha94Ri&nF2nr~J!=)W31+=A)#s?$QYJ-H7Y`MJkEbq-wB8{v8Rc{9jx>&1F^^5-#A z8?5{Q>LN zH1rMf5&c7E$S;$)&?>%{`F{nk;1#@rSMYpz(f#C0F@|s5+KxfuKlh0Wbw+PJdd$`U z2iy!e&oWSIz^$*j&#?CHIk0P#I#>!CF%JG&!RbY+$LzisbV!h^Jg%^6Xg|Yy=~Oe4PXgqG(tp3HodTzhj(=KwE_n9N;E#(+LRgU0yq zIgkKRmeXiNm~I$ABh4CoN8UcF#(>j+^!x1+z^sASS}k?8P8n$|J$kyA1HF|=RydKn zyvloCPD3{=n6z9i4KI{eKGYj|SY8FzfHJ-Oe!I=Vbj!HX$R1@4)SEYBFsTn3OUL_% z_32phP}|jc(<=$|flrAgp8jZCdzH~wPB`x8ZSj7F? zWoNj4@B9UpafU%`oODZG))_Le9JKAx#=WdF^YkuGR$(ctbcEp^cWL{5|4rx`X1*TZ z=!&~U|mQAlctCZG1M zF}&O+hCGH0mff%56}*C1@Cq*BMfZ&_#TdSSXP3dmmCib;!PQoc!t5EwUn^Rg0rluO zgJcGtdhsjg4Ko^LI|T#h00%DzIfo_3ruDVj11K%`&>9|tLF1v9dKlKyZz~w27*JV- zV-WR&2^!SIW^g_}I%WW@y`X*x%Ry4{dDS4SsUsZ*XTq|wH>)iho+n(ZyslD}(c)br zD(QPQIKUdNk>{$VCL;)0?NVlz0VntxAj8-&)bwQpc>03It}hU21iFR7!8>_^Af^Ua z@<*l`Vs0lm(OAd07V@H>TQ8t_d^|5XkdLd`gGXT)Bc1)oT`;fWt^Af}Ya63?)jgOP z9>@dsFybd%J`t$AoGiF1oCy#_r|w^!9wfbWr7<#vcaPZ<%-^)i!u`7D7sFEhARkX? z7^}fPDG=n#$&)<(PGOX>KIWk=FxlJLZacSb#W+@uISZEv^~J~Po)ob`oW_+p6T^&2 zL0&$CSgl5S99iviJw?|9Co}4vWpi7Id8u3ON3Y2b{6KGeW2a3v-KSgU1S5cKdx#9c zvz>^i?6xr8p>FgC66$bbaFb?ZwEWXlVVddMw0V43uk5sNWQxsW5@RsUWzgN#y!8}b`7JzW62m``oQOxsmkcXXk%c>>M-Cb3l5xu`xrH%H6&y` zMjxQj=-1-b)+ZS}z1(}c;i3PhCt0n{Do%>a;D}M?0BZWI5MlUfG&A%%n#JPu9N>Ct zT&uep(HoPk8iZY0G#}yyuNo;D_O7mF>xAJXcLjIG;OPr=+;j-dj7o^IhwUpu#M zB%NWV$@#P3Du(IdA;#Ph1`P%ZbmZH*(pdE5gONujP?5iK(^k>YyB2wz_y}7$6Csb* zmGUV)C|KrG(7zo2&A@sQ-;YG7!_wc`+G<-{o7wtfC1?=)Sy1=IDr051>7+n@C_{8$ zbtS9g8flq4VGOHJ>eVy1*N`ULE6B?e6SmN2w*sh>`-jtZbaWVgEXriVtr~vHlk%U| zr!%D|uVp+|`Fw%RkC#|_CqfmDUXqWeB&gf!i1j1B*D&rD?GgWw)psh#)EPhPCG;q- z_@=X-K_CV|MxIWgVVV_WAbYEcEJK|#BXhqDSQ9l3U?zp|ZC@cx`-`*&?KwnD>W8(vat=0Mr5t(WJQ?)Ox*Lcc?MyLYv}p1QeKvW-p zkw9Fn4tGNxURn}s2jNIiaFwOGdDXxR9V4M(RnBOXL z@dF|9sv(y1Z26hL=}zE%Bdgn#h5VGGv~-_h49BqTd5;x3)5+5?a?9{03}e|75_~kD z<1JmKJ3uI>#i%l zWz~{PL)?{U%b{jk9~yGhdR7iKyfQ%nPj+K4A;*h+#9za(AYYIJs!o)uxabV>bRwxC z>yrbyIGENanzOnJqK2fCmvib#eq5=v&PKgPRv_wkPrTbP)KAM$a;z2j)P2GW48 zd0$5k>+tAy2kEI>d2&R$%F%ZGtgV5goKGbtasIZQe9N2iBHniZ_GrKLk;tRb1$L*C_MuKH3{MZp0WaqluOZ)ba#G_}I3F?@J<(2fodp-={+A1%>HaWxqY86y>%%33++=s$kSM<{)*yj}gJ24cXaF1Zx_vuhY? zlu21-RhIIMeY9@fiNVa}v2iirke^@Tvm|qjVe4zOr)6WSCLlhsC?19`XfP{J<)l74 zsWDG^^2(%v-&pNeNA;^Grq!qyd_mEPt1xt^bReuJOn#!LxYe8J6=CXur-^J(r|a1- zPP>Pf98G8*X^(vs8d}rC+9&QR{8{JMHY_b6GA7J+C)mE6XlEjb-fK+f%Z!BO<0H^< zQs!!JCNacMDUbWLm5t>!tdoy>w6js4#_3;{xiI{LH~s2~(W!U*&s#Nq1+U;0ynlL7K{|F3d=hB2%G?-msKa=VNx!0wM$ zu7ZZS`3X*bG=#I_2_20^WulxS)m~;nSt`d-e(&GPND<$38fOD{&~V#{yz4Pu>9~Jc z&NP%g-9cUQMM$^nVPm|ENy}n*L>qH~_lDh$G z8HRo+Pho^x{@oiJk)P?LA>Xs9Mk0vn9X!a#&^6PfZCH2ko&DT|iSO#Hpi!?uY5E*5 z#x)E~c^MOfVGz1Mdfk^k!EukU#;S5OPYqF#%*T9{S<(@1zUqs5U>~69pWz?n^}qb5 zesE{2k8%Azwpwia1PR*ecj4Lx_z{$gwtqCu!gi$ZyleA-X9A|Kp zmo&#DX~@br?>VoWeEk@R?JRs7`8x#j8SuTJ_X@syh)(r781Ch;t8f*^`+%!yuet|} zuw@wWpGIRje!!>24gY}UG%o_K@+G`z9$vwVgBPw}Wcb3h>0UR6T{UEwU{FHI8B83# zG-6_OF^I^T1B(ag8|@z67*+Fg@Sho44x)CyfBcdWPm0x$eO{+}}%H;%kUxg_nS_GUa|%jUMxLCDQ@b z@@8d>G=hWWSc$XTg1k#tFl{b{12{O#<1m^9X$*S87=VRAD`)y@v|Czb#N~+^;%WTK zll)hj8pt&i+?L=fenmO(Gw4HOJjPeXe)z%tv94$;Cv(1dcGfnw3HQ-AkkdGrWO$CO zCD9;opR(naL01^-=(zeindjs>5h-23P&~$%1}CR7kj9feWWm+ZtjJOZ`E&JDuskO! z4d6s)=C%;hNMA^Kd%W@$Ecb9aSm}+uqu0E6BoC>3q7J zI?9AVed7{LpE}B9N14@x&G_Z~#jz}z*eCl|XvIT9~n6oT`GK-k&0DVA$zXwl9Bk$g&Inv6{)x3>2 zd^xWnSLNwF#&YCR<@{d2$X7X!au1$g!HWguzYO214n}>cSEcn;kE4$m?RCJwznahU zU>QYST~0IjdLF&yID<#p5&j}z@H5!@w08aqe(%AH)HC6w8N=B!(ES<2@u={w3~8*Y z`!SAD3C@@vmjhU zS3|=UM+bgawoL0s6Jm41%j|fDkwrdIB5l3~Cm+%dA426Xn`IowD|+y1Fw`Kc)xu#Y znBL!V;}<`ZhF&H?dZthL@#wb`HLZq9$v}Qeo_^S4(8-HMYo7012CI+XLlgY00E2g( z$wTPFhcc2MNX#b6srNo35SLo zbwoW_pXTRD8%oS_<%1|4l;bz+c$LkN`M3+Zmtn+T1;cw1;>Z0GZ#mrob@ z+3o<}8gGI+6hjW&!I$+uKsthDyUMfj7JORb7KTjl(q_QpPyP1DFx#Sg$Jeh{`&9px z`$=AMtW{wWnIyX43#jUfID(TL!_C%hFw(f@#I@t+L>BT&TeUqv&&i8^X{bx`s3Q7t zl~Ye{q0Cq7Yshf9oXcUO+_t$b&m{I`LUAdhw4&xU5QLBQBvaW(CC3nR~E_+HU|1>ZYR--fniQrvB>@m9dqok9SOFS z(H|^Vxouxuf9cQFEq=p4=np-)svB2%c%EOln(lWB-;C~v8}Pl(`xX4|g%@4dFK3$s z#xPpWmA_mW5WC_t#v+J9@w`)4jO-jvFEm!bBgH9We+=lPC?hz-FNthG4l>fL_@Zs+ zGYo9(qJ(P*=EE1{p%FD2-^>H}F$5f(#c^=1PRy6NFtETKtH>JmdBF+4d`s86SMgH$ zR>xFf8ln#Fg_lGrV+>ISS%c))ax8~;w8+o(D9P5dp=#o^+h-jWxd zvEZt&Ft*X?S4W3I>%Xx&N#a_m=dLdJ$!5t(e|=JNv7StmZ5YPsBi>U*Y?o+BCa~G} zPh7GENxks1=OAYiWA@d)j9bl#5l)DlXlfXEOrSi3Z}m_kT>O+ro=cze0;ROT^|(3D z%T{+R!;r;*=gCgcFBmqA2f-I_UG3i4-bx#Dk}*A;)+8#%OXh+cwo_QdS><&KtMfCl36pbiNxHG?rnR z-g5l!B%fFCPaeGJ`hF?K@SFE`7(mNy=9*#}vl_v^j5P+97@UOD^gv{ws>Y{AXt52Q9v`~hZ9aBJrdgDmyZMPz9jSQ^ zW4H{%u|hmx@GCgTYYfPHRwDSxhjfxhaHTDO8YgtX(D&uE!JCwFv>M?Wq;73-Mb@YM zdBJUUEi1hm#>tj{((}^bk*=M)FFN`XSw5wV(cl@;DF<{{D8uP&hQB66mdX9HnP?af zT~~)KrAC@7)7d`^FUFgf{2q5B%%h5srH8du%Ipt~GD>U;Q?@dp^B;pIMYSPp(o)^HHF+Kt`XjzIk}!SPex2~$?; zQrDg6G+!z1ievtZr#9*k1kJ~lW}gPQt>Kt@)l#&i0#&Rj}3G(m_8g`%S)-!pp6Vj8=;M6!NI(&ZqmA? zY=!og{r#PXYH&K0Zlm+*h-&ZqIpYbR+1cmMXs>hM`7n-OjW^%E&Mxbv zF!CL4``3Xp@-QrZf1TI(alg{--`nt_>-VJ?!*9KIBda<#a+*?ID>PaZ-DYd!u0ci| z8e;}!8uE&RumhMwvaN@!v!ApicI>imaiB6x{4^yDtTLF0!)-%0Wuhg8M}xEmchb2s zg&Rgp9wQ)zqbsc8mtZ>s;=Si+F@9y#mN6QmQJUrfj1F=%WNBm%FK*8Swwz;N^%g?} zx9RO9#21oxXjFh58dGVy6qt~LdN;H zb7`71<-ow8{wCmWY;MJnJUTqco^W4m^ur*6N3?mOg!J7$;a*XVDEZHcAHd1XecKp4 z&|b%QHeUYoQX+X>yH=A7W#+1|;7fEe<3|>hOX*D+#Euc_PF5+Ersd_bUe0}?P8pe5 z>J&W-e^q9$3_@lIxQADK%WXbG?)_GEC4C~H79KryK&V%sOOQV z{^x}YBW~Fp{s4n*LsgbCR0TOR|2m=$tseN$%~`zMJ%K}Tf@6GMyi}&W4atb5l5g{H zHPw?oMz}mgOvX_s^6b+TytRKgdR-Hmp_9@Vh9O{lozjBKOBsMGjbVfyqbuei3^@#t z2WiC(j`A&iua!B`&DWLak#xv!;_-LgDBoIL&(^5$$WPhqA0DRqGQr|kX>+i)w8x$V+L4sPk6)xNZerOBlI|IYSyc=TzH z@wOco3*27k7gdjr+@c0O;=F%a`@0GQuisK8!=w>E;mSlFJP+TkQsysj_6?FZ595`? z@V|oJyYQmx_azy_yBIkrHjT5CzE<;Wp}rZfTej*EJ<5N`Zw zl=7qRNkbc*k%Ot7gvO@f8bZ#3J?+6T-_UpU*VRlr7~AVGAjV*wP}37utC3^ik!E!` z(i-CQziDDHgOHAec+&>QFw%rR{CJM;WW5YyUq(A)(3OTVn9balVR>A=#gHM)xO`+o z!wRsIh(S%fI59N9laraJX-HEc1k5{KO3EYuu6!p3x4bSJ67ugVx#E^~%zJPwn@&4n zXzA!NV5zjRa%*1F((gpWttXyaEPqw5au4|^>FW#(x|}%4uScm_F6F3E4I?oo z%+s>w3xpU@@?}|FnaoE!LUX|R(KrzCM7oyooCL5@&kYNVq@i z)B$}HW7079-sY4B_{ks+m{)X)__QJ1kz?1ZluLR&UI?w^C07NOefZ6HeWLM!fxu5) za4)d&y2Xkf+TO`8^{0HL15Q*QB)^vEq|FJrAK0;cLx=NJf_hUwysG`HC&lC7MSOKY zp2mdJEAhr52W4QrSgz;6ay<*C9lG%Cz`!3oEvNlXX?;7eoX_*}{Jjj;)iCPT-1>ul zxhS%cbF`fpq93T6tFv1eo~19P-$(g$I(o_FxM*KvpT32$OyVp+c+Pm`V_!%cP_DRr z;kk5Jjj#U8trFrz$5d$Av%IPIo{*Gq+-}*Bj6A!TA(R2CF=}7E*4C^?`1kYpdGdEM zk4}D?{)~E8{{{Pl0dgo`#(6r>0Mm#rkAigb>q);89UA!yBb{|4d|Equ1;4xCJE`yQ z1-ukv_?u9|G|g6^#tbRK~FMHBz#QtZ0ajF@a-r zJn3JFQM6?k!jrRY+{ss&LC20*deU~K7$@6kTY-L^p4W1Lmmlmk4|i9e%mJZIPw2aNMN)fhn+EKBjR(r(|} z4LxNbBp4G8UV$e}Gm6xba5IlPtnP7wu zIR+?q>XW)R+> z5w71axe3xVy?huKeDV)p;k#DNQxek2`-N=}+VU3jH)>*mYrBh%<5``-9lb&(rc~-$U&`)^xKuQR)7-hFB~71r%#n**J*?M_tqtVIgaAHH zj!$Y&x@i@YX}OzLJ<>ngilB^?^|H*zUg*3~*h?(P>)1k(x{>fdArN^3gZJV7{Q%bW zFoxIZuWY~SgFoAi*O(BxV(fX_ZksrrpD_8ZBMPGr+)iHxWX*YPHJt# z`HH4}E_a{yq}@^GQN5$itB{?&Fd|BMZCo{wq@i;c--c|sk8T8wUa-M; z6ornDLgS3~r{g(0^O4ZVr%qL{aSCvgj>5;lt|l6%0pwfr)xpS2(KCN>rE4dWETBvG zl6*A8?TW%D^rbcO8peqo9eq=!1zW#^aN_aUDWqG&KRSrTBYZoqj*~dKBeBL~(egiZ zEDWIR;`hJw^qUuSYGu`qv2g9QHGm1~P`sO;AlfiUl#897xTaZOt3;$uhp9m(f3ART ztl9BdZp#loG!}I{Xmc4>D~%N3w)7A)f;E+ebZZ~CZnB{xG;K55BEZ&MgYuX9MnAB)!I#4D?AyEIBNDw{-E6d?# z7UdZJ2@8wRw@s?w8kWa1+XNF#@-*GpZsXu@z5^_`{A_L3%aHD2bY)8ePaa)~7u>t+ zigiBF0-X%F))V|nPhFMHfMHChoZyuXslUj8-{8S=l-?I^{S%NG^`(6L#Xuo0bSSHB zMgv)J;v#-cp41&C-nLfJtNEo)s3+tiy+OnJ4BHe# zog0(U=&->({dWqf_gCH} zrK7J8D7I}46t;T ziuKi1_t}OAJ4R@l59N(vjZ+8a6Fq=-a`cXa8`woViZH^s4AQXH0M}@>Ou9W%s|>-Kd;}4&PNXRqE{=Sfmyq&v z&x*4QC9GD*t&fYo(kMiCNte1NgE64^31JO$^9j#f(pMHH851z`mh5N)W-OjO*M90c zAwt@u3}4Dw9`h*H0nN1dt*^CY7BVSr2EpX5f#<6532_+o=z$Z70{Bu6AZlQax(-ds zl`jo4p`6d=ZHA6(JR0q)k}y_=)MeeK4++40<<;`o-o%UUfOpw;<+ay&PdD15=?&Ab zA>yUcoQX4$(o;XHw3SBMla1OVte(5)omlAaXC}1dnNMjGpQGDiEEEi*-ttIi$ip&}Yzu}Hluv~v^x#QPSf(?=UIdJMUu2#y z8b*DMiI_UKHeu{Au!rmP^Z)77k%h*hI$(k-%h=L& z$D<2ZaJ$ekWIpsNK$!TMe3IS|`Z&I=67E>%(~h2uJZLv=>#|MJ2WYfAi5Be3?YX4k zlP`Hm|H#jNM!sa-F~>BvEuXO1Ux+7IZrh0MW%vi5MdMOmEX#0zr!eGT`epYk_}vRH zx_(}YG5k|M@gsSOU^L{sR7JRkqm9l%Uv~^f!t?elT-r!RgOF))2m5JUDmi`|n_z!b z4a+nd12teu*TKa+Y{=3n!^fsw0}p{k54yp@Z3iR0(BU_X(eRV9^fUqS?1ZXA$^hO6 zbRiu;_a{ugxXVWdK64?s&g3m4HXWW+iLcJ2%2oF+0}Nm9vLUWU_At2Z{2Y*_CqM45 z9V-a6LSFLq?jFf9_Nx4LzGZYK`-F*Gt7z__B!BZXu6K5lG98?kF-F{~ID}bd_iY>R z%0dnPkP=dsDw7{!aD`5K=4A)3apVV1+#?*mdWX7ttscv{qAyo5ZXn`hN@H$*QnHdx z8S<{$R{6?c>h%GZ@dX+>Ww;w+ex@5~gxaQ3{F^!}^myc1JpJKA$_7 zdJx8dX^tAo(ye+a8G4$7`d5Zs;aC2}y91gI^LO8?ZsW7!&d)7Bu69dDP<9!VjqQ0^ zP8pV-|Kvro2Fqt150&A&%ja6>ixB)yTzrJD_tsg21DFw zU&vJ*$%G^7PMM*h0hxhWh5*y0UMz=+;mLc|Se4y4q3T6C6(+jrYp~}IP5Y)=%`!{s zk~sA)XHw($I6CReUlNm(WpIKJBbW)B!Kt6rP2B~{r-4+faXxA;gJgjGT?O+pyvp`` z9AU&S!wCA6B!N$(_uatrc(@AR%lv*f!l*CXg!;SYyC6NxqDN2gD=GCV{jBloD}UkN7s zcgV|6laKx`w9MZ&MOty}!$(|qEOcHp+p7yNb0*Z5K}eb5$2K!?2Sir{_LsJ7ywd8!t;Z^IJpSUU<-a4DFN=~w6jWHjR2yboi&WGF3VU-5>82jp4hDY+VM8RScuga+U-KoWH&Um4db1f+;o5_&AJ zWwDgT8Ci5IUpwGrZyttQp$0^{=8Z8=oL4Upua#9KUPq-L=25GvYZ(6W;}`i#&y-;ebn;#+(EtjZ z^n`m40q6|+V_&nA3-z=nGM1-gpF!6YxJ|6#ZW{CP4S4euGEpI|;gK`6Tyc*vi=1Z+ zuxM6r2OrAUywx4sK(YYRR^*MmJ%6~2gPLSmHjUxz(-x0Q!`js=%b3-D%VWF49Rt98 zi5qQnYs%FEmi0X@ZYNdSCx2kIgc*QxMGP(K{9B6A)q%7b< z6573Q;#aARgHQCwvJFsv`WJ1pgJC1?y8+XF8hkf2KaHNjgWOfOp#Qrq!|yiucIrd@ zvn`A|R}i|ChvSOhzueL2hTQR(lBjf^&Y4=p*oNn5PdnP(htMcs~Vf(>42oouLes1I(A>iNyGCV zC7u0&+-th#&VbWQ{GF?_gE)1KS?(&Tb_zZ|UbGjAw2~ zScMnUSx*ikwK}dbtUNWg^MMia7`h^_5{;fIHwY#_%YYje0=1Iqr^o%wvbt0Tc}}~) zpZ&e2=jTg+=B4rI7aN@zR^1gX^Y_!m)`{9KPwLKXP?0St^Q<$4vrowQ2xFK%SF-TXPdV&~^G}Cd%u|o$~iRDmPvN&#rA}`yN zmwcL#d0so-Ho~6%(9!c7*Ct;;3UfrEOMk?f@WMT1AaAha2*QMGjabMV^{_bfpg>)*SS48t8IDRr_Ikkg7%ZL8OJ%_Xvdl` zrOsyAj6ceKt+Rwt!=YhWxG37eoB~&`OEev2EKW9q2F=%mWwBzG}-HDh@9AxG^-uZk&#PF+L z@!ho|L}6@|0kMXO$WYN*1!o^=g?(aL%V;gwSjmo`BXO=)*eYuJ)Fj5%B^7H+maoE> zLRXg+wValZhxW&vE8-yPZwjsyZ_$b6>KCC9v+^8{MvAfD=#ZD3wz%xxy%pa(zq>o# zj7!8XD}R{z7ss-Lxz3Y}bm2*Z2{a z+&D_U#KyJZ&s$zLY`%Q(Md)S`C%2p$@oV6!X5Ja0QnUs9*N_K^3zFo zgl8UigQ*TBeSs zh0_m%WBk<|sp0pJ_5H#OeSbKVWwA%!UbEX^|4921+eF}|p1Hv>=!co* zGawzLhcB^lob%)v(HhPFew;>+HMb4?#P0_CwG4G)vnx~mctCU4{NuN7-ISvE>-zJH}#o@>xgL$gzl zMVk`YPB{H~*y7#jrvF0GJ#7W7x0eu`n<d``MN7)^B@4^Ci!Ue_pLbp<`?nm4E24!! zn7)z6!c44lE@C8@4s+{6JIY=6Y%JS%?1^KncYH7#uJg}7rz{SC87V^yeh=Y@=RP3rUtoL2wI4}S5VV=S?UThfkr#6JoILC5LzGZK;4mi_8S9l;I+~IVTI74);4uxogCa^Nz?YADvxJNX)9Xc*LLu^$)ItS^0L^a}4 z^EBhlxv)_EdpHFgUz}^pisjt$u#*jV{31MS!(81&sxls#&v+HY;edvpeB)>|qWp$M zZZkQJ9~d~y{Kkw=*4tYP3!4LKcBJ3iN`JfBe zl#jqr6sZe@OKDMc>7^6R zDOO=M)65qB5e59_FU3Q;i{nHwB9fgL*E6m!9VH{uSvSHVU+_vnj{ArQ_Oe`rujCbW z>4@^CHyqUVB5qnET>RhzCfNp1Zb(AsS4N6>b@|Hfk;C|pU$IA6q>E9=tQ%cfA{;JI zwH5w{L(UTxr|>B4$Hp%Qs1x|lhJ6la{ow{C{_vV?j_73C+W+aIGvYUP@i*%vVZ`|` z#s9TU$dUW_OFQ@%9%8QhwN1_ZiY-6B6UXtGaESo+hzDUd-!pCa$-%T%(}PIJ7`%Q_H$QJu0qhW<@U=1u2U-H);!Za7Qe)B)_KP>Nr2hjcVQ@+lclf)2KJEbK;k zZX>zr-=VPIiIg$OquS24bmES5)M-*Ro%l~=GPW6yxMq2U>PRfdXqO2mZ{E^Q)S=kV z_{2sy%zdr`g%E@Au!dXnAGh_|)Ntytny0wPusIofe2zT@c^s>L#5Ev{tS=Ggs5`(D zQNrdt_PL1%;*@lLkF|BxFT%~60`|i_Z3W>nIB?o&aV^Q$@Q>tV9Eo@Q5f*lF*RBsy z&V0xJFw@@GYsqz{Ja*^F`ixS|+ZJcaFqy?_I)szgBgl`RkOyI2xR~ot-r_#uoOxI4 z0R4B?b#Bt*JZWVOYtHe`0pVx-V{0!aH*z(RAmbO{N6RO`T;pB$>yupup=zE(Ey-Q} zhdcJM&EQAIDeOfyWIS-fFI002tL7ay_oSuW6T0)vG3$lrXW zeDU@>LwqhTt5&TJ`=^y{TQ-%gTeg->4{i#fx})5G@BL-VmM!Uqcx46X3e}QjOUiZ6 zdRlq*Q=e3R`bWPfjz@#gWhMX95Wpiu@{uYk!cZRCx~=@xUw^Lrzkm9Xa`yu}j`$6i zUUq3&9{wU4azr$a?byAyY=3Bbh~d3yr*EG^R70z9oI>@?XFjw1%76N`hy#zM4wZce zLr90X$+cZ0-x-j+8jL&-`s5y>p;>tPp zY3#BwjkqA1Pd#m6IX!MT4kO~iKg9bZtkMzgu$K*J{MNi89LfQ|gcxNot^?QikwdJ^ zfcPR`qhly_Zf%B}o`roHp9mx8q-HW|9Z6Hsxie0ZZ>(L4P^f9-hx$|< zX2(J>VLm;qEQ*$}jdak_ABwo-^`wJq_K6E=iiqX9+d=1T_)A=B**G0n&5B}?K)cJY$X;jG`fFvG+iw<4Km zkx#W8kWT&|D>MAn@)R5KPyf@H`Bg_$=uE@E=+yV{(qNY1a7Q?iXzB(xClEeo{^fNe z?eRyC%wMt5g@ad1K0C`9_ROorVy4T?BK;1O1cnn(k4s>!cKkyi3P zg2}jHZ*Sxo@nC&h%2Aa3JkJ04%QedzsAxZ1nWPS8+G9(*ajfPq7nAhWG*Omwv_0&n zxC;>nxFxS50HT}w)jO_KO@HQR`W4qT+6mHu9Uy$>$7j3RxjU6FTZa95GaUIJd5-%U z!RB{biL${IgUFL^r3CEd(q9p7tNX;8r*>m;rhL~npEA(uREEVgp!m!7CfpJhbrsT0 z*kz9!rp$}5#|v-#vV2+ntbDyDJi)-foF){0a12pGe0a*iv3&;*gs`at6w)c9xsmh&v~+OTH5K)4C^d+1ID@upe=!?1TsD<-)L&FY~6Tj55Tv#chNesf_rN zzFKywdzv!ZH}Zh|t^FclVkzr3*CpeM8|n<#pE89x6P|fberKHH79>p#@ zS!OuS`u#XLxIfz3s2^lv)=R?1ZMshnvVBh@X@m%eCvW68anEawA@5*w53Da+w{9;RHf=1s zcJ3hb7zbSw4;g6MXt>2k$@Q>dME?QSsu2@zMMuWXC?urnMTet5hJ3>HL?`lP= zc``7(-1N)|#l#C1zA9d6#0N1nEig5aLx^>KfQULLha#bU1R-o3jsjy#&9g6&6S zc8wWfAQW?y6+hw^e@Hj+L>Na1dCjx)jIbyxHRZK&K%CNGXWYV1qQgFMVnhu4IKUO* zvrD=N2S37EjYFKsP6<2dPvqm?{kgmtqvuP`SXoY4v^d?6uPGwJ&lEFJUb7JjQ?2#{ zm<2NYI*<)bzMUOn<6z`vTQMXX=7>O+A0ig!EWwoXEWWja#k<@h+|8RF$b9Al6QqNA z-C0({?deBeDnx|QS>s7uHyZDV@A5dlddcD(jUkK!Ax;^^=SGZjz!$OT&~lz`)Ty{k zvDNZJ@cf^h6Tg}2Q_6~^i^|ncys%Dd2!9yq-HU(G(PgI?0+cntN!df2H->nP^haJX zwSuxjgU!5OI?)wNmX<|1VjkrruF3vzzjyzE@FOJr;>BgfnXBSP@u9M7+t#vb^~$gl z=feGgeQ{&9XK#M3>+op{a-HM2AZ1|);nnAyTP}U#6U)jo*OYTtuP)0YF4R`sF;_p! zP3&hr2jm^sj^k{(-<{0z6ZuKLa^syMo%*zVdD*yeQ@P`=J0lPGhCRC2x?yKgl$+cl zFY==caWje-0?IG*trtdoSDv||eC`W3map7)Yq{^fJEGnlE=%G@Z1J))@_KR&s{riU zo`?f^l^=qM^K>?p_bL>s9m*Dh5PPlDWVDRyk@Y9+A~-SQM_5a{;WXW2zzu0e@NujP z8rpEm2seb(qc9%~|00j@n=E5(_Y{=jhQACO@AF(u4{4!ut?N-onIZ~dpN=wb%(0An zHahAw*ILGhl86U>6AXTG4m&(;0OCWS#Dj8Ae$Y0iy{xn0hJv0{yWBLePImgkjb@}b z=YT`FA+I=x^?X_oEQ~VB4F_$04h}@Qh``cbwio>3Cj2K4Xdk&=ln>I{+9QN{Uc1<1 z&T#FAI(MWm{>V4#P7VM>e0XQ1iqc9-;B;wayD?S1h+10g;aXMhC9)livlMvUB z8++nG8VHAU@@_+h$Mq+_;vadEBAYnHf5boYA>xXatoLy)Tb~G*^BG>$tGXXn;}|#1 zbtfDJgh}L%h@{v>gd-yPCEi)S^SWV<`arzXel2z6D{WTZBu2jAHA0%}m;Fh^Guzt; zo0hEhnUR*mh;4$W3{pmFeaS$%DG2u&YIsOk#ACr}i{dxZ%e4zTsZS`s_#Ma7UR;aX zKgE9du^{S1e$Xf4o_6DxL6W>;UD$jTa4BB~BpumTQ17`J!&wc#mW%K|j*=eQWv)*i zuMs8YBq!lF^N203bM>Q^U+QlDW{%Kr;F@|)7!+gN)Ta;G1`rPZhx?i@HBQ7Q|L{Lu zhJOrD#&4F92(aF`@N^Hc9(KFk%P!H5!{1%fFKx83^H=-_TCfAR4f%8g&!n3V_rshk(CJGU%bz9cFsBe67Or~$Ah9Afxg(W!E*R?hl$GxD2C@<>j*U=*B@*u_ht0gm|5aO{iD zJ4H!!$hgb9;jB~xN*v1uy+$X6S{@8YOIEx1N7%J zRG`K`MKzTlKf654X&uCexOINixFRg--T1b+Svn9Vr=?6XpEeCf8^ zawLKdmWGrvv^vV-;@CMOMD@Ano?VtNS(bjYj_IBQ`^$ryx8=)e%T}x`YtKHXoWA^w za^n|18}(*sbaJPai{pm!j&Iyq?%K4aEDnFyEL&Wz`OashIQ{xpZ!Z_GSzDg)lqZ*G zzwG7Z8P{D~&I!SJC~nXwYkZZC>s`ePDX#elk+g2Nl^gP*<^^Sl`b$SZx#ni+fBnT@ zmisqs$`v>2laZQ-r$@W7B=UXn;t=`^m&T3g@+d#c%93TVAEKQvfGu9UC~hhu&HEoI zzyGfHly83XYvr3?za{hcjO8oKnzd);^e*ZimxH>UWh~_ePha*$S)|U_K31a zoj4TrI;V;Rv@x}lk75>fDVHRW`j&*5{3o0&-|?5Cx1A4&SL@U`j$O(db_s@k;5>HQ z@m6k{ID&SLWB8R%TGAd>0ujOaw3YTaPFT%u6Tk>d_M4x~0nTB4WyPediJ8|Ics zZphPZ(Ci_m(yusPZzkfpMY}*dPL*D`s0vr6B-GVS@Hz53-0~EIU5atWKaP<*5z!h> z7>N`0pUTNG{O1(0z>GgBN;^n5Zm>XIrXOnqDtL(f%=bty>4SD15LY$-YreMn#ql~d z0>8s`z&`}gJzCu!KfZ*Pi!%J!b! zyYkv1kO_}^L2~lR8Gra4Wg#qdVUy?dC1F1zwrb6}d2HXleNpH4mfd@H#c|3IaYx*8 zb5Tc!bD#W+vaxK%ig127+O{omF0LP=+u`oCvaqaNb!NHpiYvl_Co?GcySXHr7Q2i!+~b2_#vh6yKg zDW~4y_DJ~no2KBDns=P4mpIO)`_A7no8?!3{hgiZ@!^Ds;Rn8aeZ39|oloqLk(oP0 zivRKur!%8!fvSR&C{zB&s>T>uc_?hS%FJcsg0fLcYNr3;|M+zI+*iL{mM%M^oD&^n zRKv1q%cin<5^<`!<9y*qQ#0Q6t6T```6~?<~*1_UiJ_U-!M)*yW?0 zv{FNbd@*nVBN`N@{pE|F`$D<>>)$S)yy45`OJ94iY~S5J=X$tYbjgMJ#V#J3P*D(n zn?nfi++GFwC>(2ryPBzFkCZc4tSSHboxfeytvfFtJw5yg^M(;%wa`+~5fq3X=8Dnz z(pc<^h5WowV)sa^0)cIf4 zEotMr@>rE)=}%NP8W&#dqqKQ+rBaCXdHCp*dp1bu(6O9+uw{~3NCVn%nk*0lNhxy!$R`Td`#j3R-be8An7>=tr zCJwl%*AqYJAczOw_ums@X-^32P{U;bB0NfBG`|c35nASdILI(FLikPNR1TB{2RFrT zZMd;JM%^cUxNncJNnFi~$~hqHE8{@Su~^;uuH3^jjaJAGq(n^4q`l&a!pK?&#pQ zmfdj^&^~OymUE(`gQaXDlKINz!q^st5MH)yS=PtHr^FWJ@~dCDt!%jG_P9CU8uk{I zGgh4y!vBnn3-1HaR_ximCoTjcZEsmXd0;ugh#Pe}EhgkfH6Mxp2&2c_e*ZXT=#_adq z^^d_sZu|6)5g#_%(n>9pQ2iXM_eZ&jc7gqzE)@Oap{~8Jf|d3k5xqalT|YcVRv(V` zk9G|)*Lj^~oR`qCmLspUXK@a3PJ2n1#D(~^{DXn_e*x%aMb$N%(0A%-K~%c4D4wPsa$$%~&~Uis=D z$bRn>v_y6<>2kXYk7r=|QC=h)%DvHq5JmbSB=tNy_&U-{$t`*?BcF9n_XabxF( z+w!-jos6rUOqWNM_^lbm<&3{5lz9QeGMPx@k0s2xiyz?$-f%oGPI^G4kK=KH02>sJ zFZ|sd<$=xHa~{^B=s=_Rl|6f+0PoyccJA6$cEs>&W*SOX)O~yRb!j0Q*d;1;zfX56h2*ooVf>E!F3cNK-?XO0NP zIqY*soKqx%9N`MTm>-*s7!6*;2zR(YHQXT9X(ZxAW($wLb0n1uiJk21sU6`m?DB{d zyO~MprpBMj$0$MOAIIwHG^Z#aeqpZZj*6Xeu6}YT%R+aaoa9aa*sEEZZfcoebRpe@ zUBap&Ij#wD3>)MP*M~=@*vk>%*e8v(5f8#mHdL|EXfh9#5r(BpmxjFsW%K&`a%6pJ zh_cg8TNHsZCo_DfP=uR@?mJrdU(fuG@!zQ{)!<%nOL;euv(oc~W`BpPD%{3q4a9OitY0mHEMY=~GIau~bIorSYK-nF6!7^!d z3mM;xW%X;!{n2oI;iARmtZ2M1Sh1|^i@f4y;MDN<{(IL)Ieet-Jw!Qsq%7XEFX!Ap zux(pLjk1&tdIyh8_lLc6L+qXrb+d#pUK|}VbJ|af&iSFZdDtEC;|6HUw(Vu@nP-$W zXPsT{yzl<9ooNZ-6(u(3zlVR@qf@@;{s+pL3m2E?z4Rqz*TxOyOW*ot+4S{qmhIcO zl*^v})DYIo<1Q#}Fv5Ntkt0o~hDe|c5XZV-%Y73QT(fHC%0-&P{X;uE+%#zFGF{+7$VZ7v_Da&AfJAP zJBrMq1N)-=kM@hUl6F17AHq7gpBJ%fz2I0bVVUQnooBv&xH}Lx!sJu+gR?cyT6;AX z-{YT`%};)bJ0f!Lp50~XiZk=c6lWR8AX?iV+7WH`j_CVnH(K6}#c$mOPbh28KDTV$ zx+(YLWS4L4x8@^xamJbFMEkZNZcg`?{rh%Af5MGH_3zQtd?0^!ggD>+(3Z|#72xzw zbQ?^ky5h1Yl&4?&w48c_u;*r)C(}%^%8jFThmD9C4 zH<-~uOw(gsbv6dB+6QCuH|(+D5BqropXRzTH7D)mK5p99j#oc(S{U}RmtnIX?RvTk zbLa%-v$I`x`I9oI9xv#9A9pAqDNy;%jSeQZ5ISvZ=isPwbV^;vG~O2fr5g%u?l(Bk zA1Wlyv3_Yz0q8^!FTwUBh{H}NSYc4o98pHkNj<0Tax3Em)W3SHrjFurAY(|U2YS?4Mk5S?f=_!iG;yn?>vD)L86jZp; z$q$ue>}tL+2aGzwKyk}I2$y-j`ZeYU$H+h4Sf+6#UvquoDAU|r5og|@Zf`V`GRn}s z(fMw;|NbKaxaCFO%;G2rtrwld-g<+!FVY+Nw9aJdTL!inDdYu{k&^%tfd(+Q$DVE@-=1KmIuqhy_>T0I{Unf^JU-A zy*c%Q`Ov)hx-{*5S6EYD)9>Go1pyHODG^W*1wv7QP-3B|G!Ye%5~%^f&`W?MDpHgt z(rbihP(+a4i6T7)D&5E4Lo0)&wE#rM5A7vH%#-}9V{bKZTkuJ&HDX3xxT)^FA` z3nn4n!F(^UNj=-EIL%mK0hPo@wgM5Ry~0sK7>Zz1f=^|KQ<^{Q*`h^Al(_qttbARK zjP5d37zi*KtM|OIVvXJLGszI&@OVb9}z!O{`Brge*?9ls>?u8%^=GKYJIEOG;8 zcYb<)oJVIp*zhwdoX~j<=Iq2Vo1E6zw710$sqhX%=C6L+62y*NMz`DdIA!G{CDlQe z`{%r6H_NWnD+m;v`T-nPzV-GEm&AT-&$cDx=T~=G!hVsE34u9|p^*-Xy1}K@K|?ww zc~#B9yDA|TO3zD*WT83#JaRe9h@g&x*7X!~KIT4J0qBjI$0lSi!RAhyE}M30z0QWo zAED}4Z)d1E-Y#pMp#>}x%av^*!Nv`LUMf-b+v)v!#ifo_B@SX;PV5=+&aI+KeUOXd zm#QNJGsfeadHw3Pt?2{CNm?7ZMtx3Fs8BtdtL>|tEPW{OX9&oH{TS9p{~T}Obn}DnSuNF5(UA7UdM%3F#J=TG z^}0iP7Qc&i6r1M@59*l?Xy%LDV^3oV)A-1wHLX@>?Mi*ZcC10aLKt1`ScS~Gyv(@& zOxeWcw_s=VXI+!=9+@To%ok63n0`0n_psWlV9sYaJz#cxV!fzg*FRvt_@-(pDPO5T zP>0{EL?pZO!}U;ykHRJ10}mYg?LwVKngZ<`7Hg&OEBU$P$he0;L=xT0-iY{5HOlX{ zcm3jdGrS+SFZiX=3-4E262EB&ZvJGt=y^GozkeXu6o;-S z-tjL!h&vD_2Ia~eni?INX$h5QY13N~Lzzvbu*nO0S5U5uJ-@3KZV2R}6e0JCh~e+IEhT6V@@rmQ>CB zWT=1dr#}4c(nM9u(s+Zyfm30Rs>CDxFvy7(G;O# zKZT>#t=?=Gd{{6p?pd`c>ywtj?T0g&kzUmGH#4x&#)}t>#9ltXR*>-c7D0FU6j^vC zPmiijek_K*n^kuxRr{nQ`oSyR0vEM`VbW;*&~^%)O?rOD3tQw4qK|eCu_%6l$&R}} zxW6gzE^2;8ZBzjH5H@T#XqiTxt4s>7Mw=>@#dRtkF~T z+qX1K&monoj-(F{E(?G5SFwH8otqnH>kufrUTVn53!v&kGg+;Oz>z9%?+^**S>^u0 zD{p`InU_*@z^|{^LDN`2qc@G>7Jc53N8KI>+0+&dBC<@w9S!@RHDj$!=DWk1`|o76 zu(mP`BDT^I*0pwXCpB^5xaZm@72$n(PLpRL(!tU61UJ8Yn^Mh2nU!x3;%dLS--!Dy zCg=I#7lG7AO!A=qyymY}2W3nyJrSbKgLg0;@#2|x<=h7J7<~a>7mCyNh^%Vaep1bZ zAn%TUc6tSCa2n~3X?Woh>HeZ7BE$!{xM~j>at!xJc%`_8PB_B2eRDtViKQ*P#cDfJ zopN-3LhyWQjRsq64~M6Zunh_&Ts ze1)gI&d|*>R(Fe@Q2agJ(>OlbTdU^|FqUVoPhK7>4sFgmvFumvCH$tEpA%qW`YH?|f_K=QJS>5JcH=O%WZ4{br9~`z3{168XHn{*a zty7jf`2HO{!HX5-AbEaQvcmD^b3RwAZz;RDf}{O?vIQNg^gAetxb06arIYR+9cJK? z%Qx23w&bnlZvK(+QYvtX=dG3E!wx-`>FI!gz8vm1D0iU$EY7(2@m^-l$Kec$S!V3E zqxDc;LrLj_ilWaaCMC@>-6QR4PGt(j-wYzf5Gapvo0*hQ-dZZ9986CS7-(U1P8&2- z|17c*H{IS8(NF-@pAqcKp7v+X===iiRqj(L3ktY<`BsI5#uDx6Mi+2j^dYC@#K&SZ z^3MRPplzX;QzpUHA_GU_r`pic9TuS?0O}Fd2WcVOn5%U(iB@B-->^Mh@@wpxv+QmB zSf20kM0t(-$L9jFOLYpIjk5In^EdTBTVCDTL-fxU#Owx#0`pX-hz2GG`^3Cj&jc2P7NyiwoZ7m&@)LEWFq?+imw5a;Xjva6Zv}#~Wp;0( z?yYdPw8@R{QsZthq<)4ab!4XI!PfihO=B+Z>h~{G*4|gF5vy6|Wd=d#Od<`E*mf1Q zWmhXgBf_eSOLcia>Tf_KDU~||k9G;ojU(F<8fAiA)?d4)XxEtECUt^%hn(>}W~zZ{ zf`%!ETkY@D8+=Shk}Hoc6+dyTS79t*?sHQ$jqK}AnZ97TV{@u-Nn3(f@~FrA^etXW z%}dL9G<5G7xha#>>O1{f3-h=Xu}9h~3S`0nobRMf`HkzZo(n~>`aLm4LayN=+9^Hv zoz|ONm?M>L>a8uI)Ul&&iKa zGY@7{L^Knn6IGY)4KEFGLLT|x^tR6axLqd$`7Zx_tk+}FG>KYqFydU%a`n68A|fIw z&bNIfjg9!v{y3NW{~F`}GyXrL2kH@DW(Pzi(=osc#^ZbgtR%gRi_;HL(UA`FNP**r zKH2ln!rA|54qEsT@O7W~TRWz0$PW8o*kv%baKmNe3d=Ol&%2Y!cgXT6kJ~IS(DFByr+x!a?YRf5K~1esdbZ-35eYMSUh2Q zjD%uVfrpnZ06*4-AY2N}$_+Mi;b1o7`>qReM}pK6YUBF$_X@ep{6g{4DQM1y)!h7> zBYQ@V!MdFoqk#O|0^h{Jyy)h^oSrmUlgG-o0{EWi>VB@r$ZAJYwM1X0GPr|)j5+-|-HC#{>#zn& z25m&j&>l%EMpo(tu(wylTjt+uu2rr^z~OfV`r?7D^`-W8YQ)pCL9 z*6&z-?ssn|8@1-OHtBcn;or~>C{9ZFKQ;!-JKJ4gCnPP=_s>=mvaup$c{Hf&cr4!v zn%I*~qZ z4(}w(#o&U6^YgtTh3zped%09CbjS+%;B%3_Vb1Sl8Ig}Rj$I>P;&Z=j#G>d_Z7YyP z%yzRXte({tQgUgIvl;b|7d)Zj2yuVEoR3t35xL7q zSjgxc=Ce&AEg-K>&83+ipq|Zpbo9%ju7*N1yo(skvvijJGA>HG|qd6n}7j)xjC|aj>NQ5bCFxGl0 z+w$NyU#gh)=nok2>jE^OvQQ!ee)zmebr&|FdhPdXlSc5}XX1`V9(YgqXpOfm^wf&NAOSlbg3{ z@;{^jP!o$FoBYFF0<`t#{c&}5@PpBU^!24TS>oMzEduJ2Np4aoC`r3LH*AY+0iv>g zKI7Wcweq!lE=H8nXDg0IExycJWLagSJ@*FHRtE`(B9(FhdZg_pd*5aAWaOy2*!V1? zSS#;+Lle2>tx7=U)T3jXz(P#qP|d(R-=^bpEjRVbkhgQ3v3M7Y>Y`ow8z>}eOlj#j zmcAc!N^vxIoO5PM_#f0-0Ym?hSk!)*nF03FfF<*6-73%Lo|~ud&u@2pmjY$NtW%gH zslgdULT!ydy=B){yg$E!60?5eT9h+y=5lD z*5cBV;)8F#AijtR3b=MPPZ2Nf3BLxKlPU#|-*1hCLvtj4=;S8@s`cT+dZb@6*@g)} z4evQmF&H>Dh1gh@c4OwUZk9zuu{3=;mVD;9)n1Ik8o8x?sC?KeoR+Fb+-O*7Tm%u< zWh6Z;$R^-~o~)IfHhKENj?aB&+?fGhvzCm3YSC90y5P_*gACI$9LF9PJge$&GAc3R zyb+sdJ=eOn->BW(Y5bag&8Cs0m+S0Y6d%J(+aBV0S!o;UuS1Hw6mBM>w-?a8l)Xm)69u-;cGp4&3Re0Co42EZ z`Yv;bRfKP7&?DqJ!PjtR#ng$Oqd*Pi_UkDfS(GAecUn(R!{iOwP*g}-iwAcBz4PJQ z;&Az@j|_ni(6q=Zd}qb_JS!QF-T0Q3fRqrW9DapGPQJp44i2zJ+7bHomk3t13s-Y? z*15^JosSdQ>wKyt)zRvW{^l|nslPKUs;z3tww&FsGn@<_>oVM`SXshICW2u-O+6Cj z4wff6=Qs2F2%gzF%fXQ3c=`Sm&I(o1 z{o;wWdTVbdyQ^pNy^;l|c{*j?0pKN%)Nfr)ezk=sp4DIBHK3W1ZVqOIj7Ut+&;mjoOk9Ddx7-Xa_<(_ zzK$wyesH1A7{^Qs?K`~7N|IbPBnWd$>WVM9%$KLnXYN&e-dTQ7E#Ed@byozDdQ3Wn zcXO;O7J|N$*LSp=3$-i=J|`-m(Qyj`QWE)x5%SqLx{@cjk^N08k!;eo@0tfUObYX$ z^y*dUqPb@7#f&#PDiNVD;FD zRN^U__KzA9<>0^h12LN_uA({&uyYTJ0G$uXi_;|_Br1#r>Sz`XQ=N~_@%iN(f2gZv zC<@cOfwzqcH)=VSEp#@<-Dw>P*trD-JqWXG2g8ZEAREq|KS8pjt-^KmA8UU!UJX~O zWr6N7sWWYud3hdW<8Ek>5Gpy7t6XSLA%QDb`gwC?qv-J zfwXUFDFYL^&|cHxVfEX;v5WuU-x6Zxa8vv8y}~i$HWPjQcO^u)J=#e%=i|OI)f)+$ zQ!h+;CaVOwd&PmS#xf5_E?zQ8VM!4YS;vAFm+Ri?bdvUO{`s4T)G29({)PFco&3t^ zHk)>513YkqlR{ZgpP!iI^WSx#Iz98-?;y?tHJ6XAe!)%OGIwVv~fc(0&K| zp@97Unug_%kY)2ts+Z!h=#JTAi_P@%k!>MPz=-(DcSuRRk}^@|m+Zb!j*kS(>qFJe%ilcuI0M1U;@eW8UI|U)j=R@ zbsD;9Nfm^R{7y}=kkSspu1g|%HTW0Xf|C0c&>w`6&5Qq7wcS@}0ly{){#wP&8Fbwh zU=$LT2jpUrlH%F%d|R0G;~e1DYyGm!J&FL0G3U^vh7!6@@%aF^N8J#Ry%F$Pre$Ws zv!ETNa7((;sUs|@V!GnF2OhmQJ9@R-!9+ua{%n2n30jp?bHzR6^L}5*Ug8L?O;Tn>9M)J7xmc!0 zf0wh>CWtB561k~#$mr-y%h&5b42ig};wKi;7m_=$G+Ogfi{L8^i+tT^F!3g`&Pzl~ z=B{HM%#oiFzSX}N!k`6HdeiyYxCvy991{fcHUapW_)2UojS#bqsIo7ev*~LyHz8p) zB3*oaii0HHJ$!2?O9;a5Qy-`S@7uk89g05;4XUp~?D8)A@<@;){*}8rdEgp$_gB(i z5h7G1#2$)92bQ|jf+;So&xBWi;Eb!=oK^H@wLp6y>d`kJBWF~4N)NX3)N>XQR(}Pd zk2Q|#3+3y4j`^>2y<{z`QkAYXk!ao9`bD=ufuJP%7bVR_wHg&w6>!7_A3)f{CF4&G zN4%KgEzBhSM+HmN54_uB^Zdr?xg*qRDj}DjywBxr*_!L#(-?yYy0-QmT+x~YH52We ze9!IJNO4z&?Ehvjg>{E4QJX1-n5$VSKCoNHZjb(oAt?r_lA&xzluOvZOiZCNt4&sY zvR7ikClP}eWM}PA9Z{?B9!3!ML&N4!c@VZi2=jzfe2~|`uuin@*A4aA>2l?~SQ?=^ z^QgW-H>%9kTB@~vdC3vsAaz-cJGm%c?TUiu>uA7wD z=RIt;M%-(X5NlKVvk%wl$!hc<eT&%<&)D4X!;Y}^?4 z0%kRwOV_h;->6`j56{3h7KuTbhIwY{I|GEvu8Qcl1uwaHH?5W!jyRHqH`-zf`9|17 zOF}HGg=4s)1_dU4j0Ky30E|S2t_2El^W;xgri)mjKdO3_G09os3pU96vRfY3C}eox1_-i2><-rNV0%g~lxG`C&U*8r zN$ahK-S7QSU#8H_>xuab?N^$<^Ng55ttZX}$n(Q@UUbqMoFQf1a~8!8E?+d~@2cqs zjT}%;^Fex(3vu<2DjGzxIeC8}Ds*hPk#_K!r z!(@K|ouCc)^`>>c1Gr*d1r566(|FatbrA5Pket7?PrdyJB;neT4dV@?4O4!-TpTGk zVYRm?QRu{Fl$%ehItbKfs@W;(j0sGr=mPbnQUv{Er~8HpD~KR{k@xpdvaKFkJwZa%)ItL?qI%wwgb z18FHpne7+aEnYuz{rPAI;Z5xIc49!nBUb{^W7$c7SwY~O#T+$iqq_F(m4<+q&-;Q; zNFN1Nb>}_Xw>qY814;q!d^zwVvUG#ghz;a?v_fhU5|OU`mCjmr?+cQmgslIOf0?dg za;5!L#zT3%)UU&nCri!rKrZ`(90W?aguTf~`1Gd{rLsE2=0~PZ)d*_}#ak0uKU6BN ze@Gz%kf~?ZYyIwOy2XTt(PC1IMKMj~Vlh81c z-JEoRCjt1lsdQT>xRLNFb-Nn!VKgt&aJTQ4(oJ7n^C^`bnD7*DLkTHA$GNoY=A_oM z2{f(kq6_@<2U!kdQFRYH)KIk@vTYccKZ@|@UJ#xgp8ZG)ah8{(bb zGG}se42JAQQaxJRyx6%Pj?>>0lJ#Q1oj3#?c-w;eSj$q^ygyIvlM=utf zDmWTfM2hUv>qH!x1OjId0$mBG6BZ6RU%o^0MBe#n9=qUB?ke^T;}W3n!9PEN@2=3a zllFK3ga$B;tB?duGc$@s|CUg3lN&))9d8KCY!dXXW0q~7tz|Qw- z$k8)LmaL!~ufzJ>@*T=V+u~{8(OVjl?POm=`T{wCtSR<3D!2AS?}S54xCzoC9S0W` z^~yIF0Ly+pJ$mQybwKZ)<)iEO1n`kxQ=GF?)p!5o-6oowzQVd_@>p~80v?*APX@dB zjEjWTbM-$H8E135``y%U6Q^v-i3@%bf>6sxs0nB6Mg8d)R(;GA1bByQmA?8m^Id&w zB+{J#8ANi)_n2P%5OWpT7npKEK#+IJlH!? z*UJ2qFkFF!S0_lHC#t&syof!QID{{b>`8}I)ski~W1_|}iwY45+FvC1_S58+>X$efF;)Or!u-4pZQX&~UZ5Ny(6s2o?m zM8e_p%}T)52sAyjbt8f%v2giId|Y1X*%)J^q!d2WZdYW~=D>n<-h!MQggVwnwNFJ5Z$`G>oU*492Z970q)NezFO#!(62R8 z>}T*?s^0;A;lOr<6o6N45JIr{YZNWhQju3WlS~196EJ_ioLUeRe5l(ZS5~DwVj?+1 z3A}Iv&TrqfIZSZ{u#5_Pi?+e+UVA)_8C~~Q^0#Sq>!RbTT2tuiO47ZZkW5wkjJfqi z3J^6?zXfcVouMU9t_k6bjN^NGj7bLIOuIir2*Pj`jfoyxz?Xb}nXW=^LE7YDnp^I=3oV zIJTtF3R!=-YE{vv523Q;!$jJoS$sCHxHorlI~gyE-hRGO++fU}IRVZci&%9;Z|NRr zk3t}&mJ~XTjVv$my))o;ToS z&?Y8#$oI$~|2(ACz~i7poi)o;BXFt)4D}ey$DdI{qcsx+ zeDjXtN0M2IhmUr*iNjhpS9^;MH?uE50xiUHMEzF{7*rsp;(Wi30)afZ*GshViTYy) zc!wA>{vX4|(i&5fP&>+uHj*x5N|5w8T@UmuZ~Rljeq7{Wsw3Z_cNkS8*LvlEt}Tbc4)G@)5TYW>mS#|u0Q-)xBdagxibna=@4&ppD1hn z)2rWkU7dJ}%N1&tPZli}J8%0|(vXuar71lwKRMldUTuKX^Th&ko$oCw|B`Y~DPb&( zmbzGSEcNTpxJxb1(v>E+Y9Xq;0U(E!H>ob0utRH;{;V&xWmlA#;%O3X*OZm~^mFgN z^VZF*P!xFi*na1lWKZD}(^Dr36UO+k&xFI$TL0-H94G@wxt|cxwbR+meAcfbAMj(% z7gt-ZWzh#zm9bw9y{*QL?!PsHnz}sFiil{KdPc{KsHe%G$j?S}5gT4BKEUE6f1K7BM`Fj_o= z1qeGe;TXJ`?rz6$k?I=IbBBw1Is0Gct1ftfQ&&7qFs?HU2aHM-Tt8 zga3^mFysYF{`u#h^&3W4?)?AaD|_Ch@c<{bBbh=S(@Ffr|8v9G(x~j}!)N~mZwYsg literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/images/image-2 Signature.jpeg b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/images/image-2 Signature.jpeg new file mode 100644 index 0000000000000000000000000000000000000000..948132a1348ac587a80ae592bb659c97e3e009aa GIT binary patch literal 23982 zcmeHP2{=^k+do5=?4nYM3F%GtWoQz~UJYf7gvQKZn9a;sQj|)xPzY^OsbndkM6&d@ zmy*g(lomq!cx9XKoHMrR_5aWN|K9iezU#ZLId#r??&Z0k`}sZhv(DwP8`&Lz%o-be z8vubo2v!MJ0RX#imfiXQDwE0xpwf^A+PZ+rYJ1F##Q=24z|ScUScZlf@5bZ^m=dSa zi4>+|2$L4V1Q|jQb3KFU>`SGRU=4c;lR%(^kRbt-hTfJWIz&sud7X*jkj82v6UqaL zIW`G&2P~Q3V(sn%vN(ltGOVXlseaA`W(dvEHxLI3C4lt+9iW2$`~W223=jY&5CYI3 zQ-Bz4LK~!DGpk8V%9uwIz9B>slSqM0g7TTLwk5`6EeDMikO=`T@Btyhabnh#8avXM zL@I>=%I1PMU`b27!Vu>0VmtlmY_F+kV{T@Boa>z@m)pjfJpeVrBnXITTS}2-fBAD#g)Re zQFS6gZ98$TCgG9O@tqhXCd^+SPFh8;8rK&KCE%D;x)qj*g=zphx!!+0V>~LLG^7i@ zA-92b1V4_MUy5}B<8-=WopB@%e?1+y$^)_hUotp4m9{Z}05&cdp+t)RIDAr&zzOte zHI>Pvl1WsG|5z4@aS7zY6u#s*--+lSFjXKvE&zR-(1-epU?0Z^!7A{L0Kh)SdBgok z9ZUm&6#RrA<~spSyh+034DcBaYtEaXm0+viVR|O~1Rn@E*Ah-|6&s6BkPYkpfVE}; zMzA6j!@@9!BmDzTD}v*|w3lGlm`9?p-dOYttO9!jc#p}HpcDXk%kh3bloEu>qW==s258 zd1e5W9$`Tu`i~}BGnrHxINdP_Q^x@@5;$P|i7Vj?4={h?i-S!Z@T)r9-+zM&0_QU@ zoXx+$X-?0Qh0_R5f1e=_Q&2z)z_}oOK=c9lfV#?Q@U{Vy9RRR<70Ay66aWOfk=+MK zSm3F?1RqNbNPhyPU<$zR1VCkW3$O=}5ET`jDJn5@ri84xn7Hh0X$cAG*$Q%Uv*qLz zWF_Eh?8PB|eIg{q#U-UAWu&BJf z-3IntG4>sVIq0PjxOxFE&|&Bj1P}r!4(EIlEM4Olsq;Anke`NT5?t5wxzJKgd%Fr@FQwG>&LQ>1vcp#dHMFo`D=IU-m%>LLieU# zD85(!zTen(qkmXzTGsXZ4IloH1q1{@Z)d7RUAXr6CNIyDJ9`fLp zQ|%W6HQ(3tT}hVCvvCZ=uZx8t2Sn73et!-+kvSJ{?TTU&-b}dguRT~Y=>oeMkPw8+ zlm*NI7P{whPKzr>;?ho|g8_(30W$U1GP|zjs`mK!EEG9crge#>uSv_DU-YwnMOV!? zw?~~fhD2NkKR=J~9C?YYAAal}u5Rg3HvIBPo_Tgtxu;~QkNe`1kN4+dx)$7#+@4Y(9@$lx1feKt<;KxYf zFiZ1#dC^dwn(=7%;+1rJwWK3S_G;(VlI)j)=`~JV42^-GjayMI4lApRTf<1D@yJw@ zsG#ONwJi2Cmg@t8h%XyBcyevV`K?DCvSWX)GZ4?YThbnnc~^$X3wpfW80gDQ9ULi> zdYV|@XmsnbC=kT_W3i8+_2Yh2|1OPczjq)--F_cV|*kg ziQL)dtxjHg*@Yc)06}q|IDaF+Xa}`k~l+-H}#{BkoYsy(ZJ?iH55`xYv)b;^(wj}T->&D|bdA`BIGnJdW&c=kLOZ8w)BBQu z_lG~f(>6HUeVAgz6+3U!spk};`?f&P3Lv>MCe(Zhyj)~I@jtq5L#xgb&%t^0o zws&OLvQ_~yCLTLH2C01xyN~S^DQ6{nd|^cP9s2GTR_~C>x-Vjs(5+zP*O4mOJan54 ztk@Qn#tPM}-*IqVtHDA>X8ZnE{RLTD*ueE!&ZR7a?9u0gW&48Iz^U^UF@2`7b%W+% zKXZe(vVl79vgh4{wJ8UpDWmCiHThP%+S2ZKYtX_zj|P51?_K+qM%NF@YaI-XjK6dv zZO*Qanm)BGvuP8g+mp=O>;SX}#Codn1u96Yy3mac_1&8BH`xlCzj7omA zW@}B}4tKM{A4)fSP9*d`O(tVzyVZ9onWc5FYHcx&`j)rjo~=vH?!?4reFOoK9ItCh z;c3fr%QyRN%w}cwc$I53WHh#^iQt_K=Xy1zvH|t$YFX($KAWP7LZXfjKl6-!etl5S z^Gv_#P`MTxz>Kz^@6O3^zOlsPqep0E*5=;%s(v;>R>|$p^D@q{fmfcD8zYo4PoDHB zyA8(sJ<`?RYolp$HNNB8(ApiwUE1HJ#pjC5J6e)be0-Tqz2Pc*N7=(sHzQrcE9+KL zv!b@GjqFoqk-)A$N)G(d{xzm{aIc@4qznG~5@cU<(27mI4j)E8e@k;ZyDV=7EkNq4 zpv#emU~h3e1!nPIUg?frv~Zv5XzQi0g^m~AgsfXeW6Y~Ku}8L5=_zrz!N}!fyW5S8 z8>{bBKf$F_WYsTS))Ka3y4U^~d9o>c>4%Ek5zMW@n?n@B<%bWil3CtB^TR>OppC1+ z0gU~e<+@Iz)%|gUs9JPNSD2EAsk%#Xg^ZbFX*tC2iqNAB=x43sy^V z0*AgM=aTFfFAZ*YsdO7`U>%aG`9Lbc8{dCnsyrA}usN)r*k$hirJ`tO)P35}k_`z@ z%%5!j{BiXfp9@9L*?_``eWnFPRnvT)_?68GLMd&ddis6Yw^)@~FWvi(&5(QGMgL_; z1;^MO>=!JTv`5)$Ss%|$_d0KTS?cJg2_6lgAcLGWl?qe>LK}kt zG{A%?AcB|$Z<)cR1jwHcA7wyG4PV59iF6?1w=o@XyMfSyHyzL}h>p;Q)6mN7a4Zqt zY(RO>f%h{&>s}QAfV;G@H@xmbPM*D3I-`)kZ6s3g)G)?UKwnqS7_Do9)D}{*-*P<{~IY!3q zSP>XFIuSlkLqc}Jv!W2@N);8Tpowd&US!%tLDS+lVHwP53Sxq`q!X~<5tIv+N;3Pm zfn%!ktNLX5la>hJSr)+z>Zw|~CR%zZ7fw&r)7R29MC8UX6jrO|%6KJRX3X9FNl>kO*V~ zg~P&L~X$Qdu#b73UoHky_WS!~ajKQ=H z1w-)P^;8CQY@p*rU{FKoID&O3SY_4eO*s)Jd&jg+aGL_vCp*Yk7-KMKdkTYzrQitm zR%T$mw24GK+Q!<%$i@g|VTm%)w>H+(TW+Ciz1nKIp0Tyfa$_S)V>lYQvvS7~D6^vC zLZIdvGmD4JTK+Y&Q?mZaEaxl>OPZnsdHtW5pT@+Wim@Xyz|OsOJc2o+1suoDgkWy& zcz~0La9q={bOr%x!j&rHZ8$YnIhLS2a44X$IOw?F3Arb!v8i0Wf z{hRPx@e1z04gcgY{(l4iWYppUuoQm+-b`maV8i9(Wk1eey2Dj#R%i!wIw z)54nQFV`}}`ub`48l&{UoeCa@L-`pQpbQMBpYSH+^0hoRyLW_nX| zPqN3FW_s{BTyU#uCI&X#_%``Zl^*Y2|LA)%X8uvL+{XBYaO1&);={!QAZ{@qE^a({ zP<*&}0K_fk!^MpU4~h>L4}iGEe7Lyr;6d@>;sFr1m=6~>9y};MTs#2c7W3ia#)AjN zhl>Y5++sdl+<5Sy_;B$6h+E8uiyIFf6dx`g0C9`?aB<_ogW|)*10ZfOA1-b@cu;(} zcmTvL=EKE}2M>x57Y~5A#eBH9@!&!6;o<=hx0nwXHy%7FK3qHi;uiDa;>Lpq#fOUr zK-^+JT- zE#||;jRy~k4;K%BxW#<9xbfgY{ad)iCjai7Kmq^QJPiD+a`4Mgme5xt|JDT=`CA5{ z%$%=7837T32=H@CoUcRO0TjWH69@=`;_=TBfS>3R13wBSgqXp;1870tei8sLgn*C` ztRaAazrh5E%Fdc8w@7}r*c?^84JsB+!HJ-{{8Td>gAiACit#Zn*Y}hvOJ#N@$|B>= zCdz7n&#g|2eMIBpPPuOUvhSz2&<>WL!SU@^sVl#y_bGOoob9T}enK%)zfc!$W=o8< zY6&~qQhDw{5$VR8eFLGPA;hJdDT;I6X9|2f@LTktt(zT5Di-?;efyhqi`2=FxhF4p zN8dN{E@)<&+*_OI@lJKm`+2KJjAa6QFP%y@ic4<#5$U^f(Oqj<6(_a+4hLyO6-`lO z@*(KhKj%%2(z6qV{V9r`5oFAj-k#a{=38oZ@AG`HzcU~s1@(%gDDPO4*g<0hcY;hV zD}T*x2x|PMdgoHw&c^pm*2#|~_P4k4WY_4^d+z?_`Le{$(VdZN0&9!(|=c5#vJ8_ag%FJ*%anM8Q zQ@8tidQUw#6}dg@`l0WgLjxU=9VSwOJ3AN@Hqh7LjV{&kx&3NLKh~?PDF5+-+fVym z5D$p!k`==~ZMQlRBY8yqZg=6nk|zt|_m%9oud0t6=@Z7(C1j8;KMF$UQH=_2eeC|G zAfv6E)HnOC;R#Wk?(eSnR4?6awNZ;&TcypK3m3MzrW+kl zRjGJYLBRJ*%32ICI+xFJRy*)c-V6IezGSh%>$bMZ_37Mad54c)tNu~J zYHT)pfut<2G*c2Ij4wow|9KrCo8`D0Q}X zo!in~ohj~jC4{Z4-1D>JTHRdAH^`jY*0b@gNp|m#STAOZU7d?Tf?%{_ z9sJ;CzP4Z+u8sKmfAL{ehSeM#NS%rFmj8f`f zRhPbQp#DjS-^2SuS`{lWWyROGsL#k%yjO7ZGKP|~6!ol)I2@|2@;oZZda=D(w0JN1 z>=NI?fdc_?iZiJlK|SdOdBMFk3md{pM^&9q5grY6Dlmrz8=Dq?yhrwm)aY(IY;YkG zv;SD6W?HnhPw#=MMpH0YFSCJ1pDOCTVtxVx{XO1Ixg+gjjp48QVu*^0Zg`s$WDhTo zwFIG#Me12@Ph3vOvjLSug`&P&ax@f|UoU>|dT!KgXyj{Jqh{QDveeDj2L8m&ySm?z zM=~-=7w#X6cgzdjFtc&ZX>33CA)zq1-d(*#XYo8XuqvtCS^q$``TWL>+gJ`MSx>@Z z!ZsR_*4AS-JNT)%=2*T3610Tm7`35-$Rd&dvb-kY|E}QF~034O%#iR zxU{D~c9T)>+9a*Z;_s!e?5YZN{rX*Srl4hK}wCz_r=->fpc-M&<}2@_Lc2uib=gC zz9q3Lou-LsNO=0up}MO&=t@NAqAg#<2aloj&o!6and#ob$_NTeIkv#&>va!wO23+s ziBi}zq5Kj>)H+3EzR$h8Hma?TN25N4Ih|77*!;()1L1BZ50w1-V$}yd*9X79ye!=i zHTw4c#XQMEQu%S;ro!Q+VqaE>7hYF3jM!atLZv-#;KKD@;pj$fr>DOYi&p%WVw*Qd zQ$5qY9*JnE{t;zY^8Jw9_Y;xw-tqx2qXi5@Gak=DhrS4o-`nKoWqy}qmZt5Xn zv&k7wov6>xt1FV%-(f9%Ca8J9Q?_p8xGC5boUS~JEy}J+IFxWXIn1-dtK@3y_u^A8 z?&$R{L0(;#Rk&5qO|~#?^uEttSN9bgflgecKDzW{wU7IegGnb6P`(dL<|tmPRn};e zsa{q{x!HHX;b&J9H3vT!ORKq@(`2;DYH#O2z1>Ag(?C(XrXyEbj5_80JwZKHN0iQf z_O7CCk|_MJjTJGJ=M<>b+0c2lRNd1#x$$kHn0b@NU4iVqC-&S48`f1O%9?*@w3eM4 z-7qZndiZcrsJXuK?4;9HM}GvThpP9;Du-Ml>WkIX=Oz@-5CwiegU-1aADgsTe)x-8 z)=g!Fa{teFQw6t`E~6sx;SvR+O;U?ZHOQ2LqOCqB;@ZseZX|UR%11WLnjNt;IpXi6kTnpa3=p=c0e;xOdh?dgnTP@<^o2WB1zQ{2>e`%uDJIG zoci{zk~vchv;?aPlH+y)y3ck3XGP=oi-CctBBlaeQ6@59E6r5W0zPeThzK10(7(fy zQ2OM~!CjBC6dn`2d}X!H>y$nAu-5(^u*_ewVdjmxtz8AX2W6u#zL5lMjVTcp-;D+G zeeU1u?-y#>pKe+3H9h5$epB7GM!yetb@aNIu`ZUeZpWfcyY{|aYm%+9jXuxZU(uQ5 zDsbm#>cT5w{vK4_pl!tYyZw60?N~y0uPwbEFe;S#*1g%fD5AV7_{y;%>m#vulgagU zRb+X^Szk6wEgX115dW-F`Q;yTYR{!MwR)_j+CB)4a&(<{OrpW8;^_?xXm!9}*BTB-V529p^sxQTyczCN!^(HIKCp2pPd#{3s+Pxk|f}#&J-=TFp zto8-A7GH32_BeX+QNEw}+dQ4_=e92;GA%R)4K6o3UUHB9rhc$Np}zR;Q^%h3Zk>m1 zj6ccMU1(e;zEs^|&6VS1Lx4`W*^{HPukW?2iPBthvE-|ZJ;H3D$-PnTn6UK4@72eT zSVh)p<^(r=JCl(g^DVxj(WEkFSX$t^=f<_T=Caq7S)QNg6RsDVtt>aJ#1~fAS}T=R zJTw?i`Tp>2V~vl0yYZ&NOWB47EW0n(*Q{K176n#&EIMWGs4f57;Y|gXcH>F66RQ0x znrb&Cy&fU$6e*C)Ip=MfGwK?y#0J1shv`GZtKT(|Eo*b~Q;2SdB*_Qkf?uU0)WpQQ zrJUoFYZG5v*+wNxGp$yC?~LyLGPzRGGyN`;+o<5+x!bEyf9bbj`iBKYTV6a^A@*MI ziyC6d(AnajG>uC~{od{0?$jTYxj-c-*m#L!f2wk9jMWFBGV!(`Vo3mnrue8d*@f2{nxKtatB zZMf=xup`_L{qE%BAEg>u{ERY58ALL*jVh&*ie`I0#duRRoVQ2e$=U``NW9Q`Hj!~6FGe#g(p%m4R>zc2ZZrmP%nY#h8?oLv8UF zgjf!SAW#T|h2vl1u|WRM4+|?BI|nBhCvXbw&Q>DwW3JiKD!5|UC%$||bIPw43C>7Ow$G(Try zX?5QEg5#B|PR=gZTs=I!ynTHA{6p`Bg-1k2-H(5mkeHO5lA4qI_(@*=(}Kd%vX|u* zm9MI*o0?l%+uA!iy9Ne_hDS!n-oBff{`_TT_UjyuNLpU`_I>ro+B#)-@8>V-Z{W}V zU%gl$Z2zGac>fQ@{*zuppkAzh=YZ?4UM#E;;Ke4y&T&kO^T=5Tu3N#v^4br$MgGYy zY3%1wIDMHcdOKv2S4>fds6_dz+P^gW_Y{l!pVI7qDE5Epg@f?3v4E4uCIo>&_RX2R z?FSH!Prw0$JV=VjM0@Rgg%M>(N@|ft9a9sTsAD&5`P|mBbhm2%Svuoo>9nUVWEn&zxyvtIYr zjy%0LnRD@KYN`?L&Weok0|oRVwZvTvcn}2B!Fm2bf4Zl-KJzzxUThA zV_3>d9%Xku{vVQ(FI>*o*7iLrajO%=gdqR}7&UK(ZqmS+EUr?__pGZ)bP1VK8VyFPZ}uIClW?oDhmMqUw@II`jpF;IszTF{sj|4_rHfX$6!;<&%dUhp8VFnC$V)+ zAcJOl`t!~KJA{j_#>!cc&p~-w_tJ>37=T&O6h?GzmhEL zO$g8D>Ad2}@FpTo{%R;*9p??8h)2M|~Ox2m;pnaKE|UfEA8pN)?bW@a_TEckBbzYV$;{bKu+ z){pzM(#6R86h9!Xlx_^flkaZ~?g9@^qML)TZO6KtBNg^!&!>A?E}bA%-OXg% z{!Gn7Yg7Ly44pm>#j>7>Yg&E3E3y-={$*b?DgY53G&Ae8c0!ANR6dc5U8w)J2=MsT z$b!AI767^$euo1H7eAdJIJS%cRPqOFFx)hRRpU1Xx zD=MlUk27`}yPP|L2BrY@(=R(fVZ4$qyn3GUt%0t*@BSKjEr3*+F@r-K7V%xVtkvVz zqJEQW2U2zHvgPD{lm*KN)bTlVzc!4}2yMrwGWqCyphHOYa~CFK2QYCV7OU%prez$% z-+NyxI>d?{hawwf)FiSWo+w$#4LEX`Aqsk@P_TtJY8o#B^Ti~>$oA;3oO@P?6fED2 z1dJ#-SH&uC(Vr4B* z0rZ#yNPHgXi4k-2)G`Wmvw0a7Uu2qpg^W#=Hvg;};IcVhlCAi4s(vkAWIN<^<`=5| zRss!9*JY%^ylt7<6x>i@(+!9*g=&+wNzObX zLiJ`#o56X28rzn_b|K*8`F_{BS+54 zWd3Gw-MI@;AqsH8ffpoO zL+_+3Sp&+)eEiOg{IDI$puOdc5~_&};bFfSi%N5Vvx1VEzzBq$uY*(oD1i1$Z0a^Z zfEjZ3%OAZbcIj_yWB-)KMZR@Zg4kGr-!(u znZGao9n@Ni{;?l>r5nBAh*|~}yEmcCrhC5r4qW{ExOaNkeFW~-$3;5E!|aH|McO%o z`QH4G=>o&p3y?6;Z}?WLtr(gSY6K%t0d&%==yr|dnKWq$Vsz=a?!vrHl>NEbw>l36 zyNW4q`@VPD;SM0wZQvG7nmICyX~U+|HEaEsq*`02wqeabJC?;mlzx)UC0Q+bx%WQ5 z*64Fg|HRen;0~wBGN*e58I3US30~B+JJ)2ZfW9>)ya=IQ8)lF3O*oe!2Yv5-{af@A zwdseA?Uw&i)hj*B-=awQ2P5Vc%TLF16#d2fVGt1qi~qKl=n7Ws0J8GkR2V&L3wp)# z8~?ZNN997;zwLdh@3*}v_w{Y|$7Q^!4IC{9z1vtHqdjyVI`-w=pL`~-BN&_x1LmL? z)+CQ68f#MEB=NK!(_T?*;rfUj-+G+r(|50Gn=RkHRav<4GQD9uQ}O|1pF0vXQg$Dv zuz=$C&l`EMjpk z5&G5b0(x$c$!klQrUrKaf0hG72MS?vNKJGpj((2vZKj$!E6)0@(zj^aQmrFl41Ttd z29CxKJNgr`98DO}WS+pPVf~?i5@rAf9LLhd)Rx6^!Lij$?J+Jo0z?z9se6*ghk^^} zKepMpPG&sPqh#F&lapk7tOtJI0GSx3|pg+UL*5VH1VbQRpcTiIY5HYLwpZ)U|&?{0j43(Raf|5h` z!%(bl%n2wz@)}7tz$X537nT;0qM0zn^c;>*?PJwpu{_KlW@25O`{yILBPY-W-T-ew z!w_OL>DI|juJ)2GtEzz1Gs~CV-lkn?&y3ht6t+2L5+{u7ZTg!8Fc)$8TcFjl`!XIg z*^tHsJ3VfcfXmH|Ga5%OwzNxGsw<|%u0*v@naju1kK;vKX_82tsa_p)KMH#X5s%xH z^U6x?(O(vj--Yd1g&N;|&OG#$9vuU4oazaWO5%bogl%16Y#A9;+WFz$700wxJ27*y6C7%?=H{Z@(k{fS4E zVxGRy_fy@q4i!SCG%=jpj|M==XY22TNCV|j6i0&o)&YbDTQ9UrF_m5W9l8cPT!+-u zt-opCi*$ZycDMWzI?MvGw zpaef5@K(UX*-$HuZM*zH_D4FA*A&xjR0MaEW7YBq$~cT5Y6MEYx&vq;=pspjhQ9~V zb76F~D3T!m^^=wLkrm5!%}troR%VZ`RgdW*a?TZ`$>eZdy?Pr}0nS_&y59|U2R(-$ zf$^dV&oNCq2|8#o)TEsHC!G>foxhDL^Sxh;&IRr`)^q-@7I~L<_4D~}SqtP?WrL38 zdD|oOfG}mBFkn3MrN=OxK~4U3gfk{ZHw7HQ%QLN6fQlwvd;^SN=pv4!hWJ)Bd}7y% zShj7|oa>)BF*ix|Acc!WKH?QqqpWHkiXe{K`P?&l#b5`Aj@x|p2Op9+FPQEG) z|B9vFW;}s;A3(m=FnOW0(~+Y?GhK*89fm2M)Sj@j+}8f-q(|l%%{&9sh^N&J&sL1x zWLZUxlHT!-!pUjNP`b=9^0Q@FT$tOI8*ySwiF^PV(!YTFS;U5YopLA8T;=J&w{$+~ z&5+65H!>x%oxc!(2AC$XjUGT=XVH&gn;SN}$?H%d`b{81K7(^_Xai&ZxuWmWgSO6|+9>7BqP)5w*hYP)* zpsQS$6a%X2n-!b?1jKbemiN|Yc`bY1VAu|vOmNSOn?jAk+LAVp;+v8-AJqA~>ri}# zGCXXuXomQIIx+)I&(tq>@|ssp+}$1JSCbNEHa;`v@L=*f)6M|1RPZJ7UQ9VUAMiwc zj&AXCqtF;PT=*L1+41+|duYniV=r9hh`!okJ6G1xN0Is91|QcqkM3vRuB2Ec_i%xT zi633M)HK&4K!gw3D}7wQFe|x#??Xj@lUp!jZQCNo0^Ha;T$l@MG@Z5mq>Nv7huZt1 zblzM=3j5s2&alwERjvPg`f$mI=!B2jqt!pxZdeV~neI7)f?e4pH8%UbgvUx)XQuC*q>mG9d5-m7KnweL^m zN=ran|2cIi%AkM?{MO>Lb)D%^S%IXmQjvS)gEwxMn&;F;SQC+*uA^U9}o z&wrrZy`Hf6oR=ZXXo2}WgW|AMTOfgExGt;?CqD%&e_v{UG5;-&NJ7ZNV{-lk?zB%~ zrOeltMHjN8YBMTe{s#~N=GQJpQ6~{A9vSYU;zJHxaYeem_@JF~tfOmQ zC`$1<$AYAd>eV|5_Qv^4zN?_|pa2X_IC8Kb(|7>cj3&&>qL)>&dg3dG(L6;OhCG!t zV~J$ZBX&Be>#3vun+yF9uEI?G@np9N+ys-C2Yi7cS)~sFDCsW#!o~)&jxNiZUndVa znn(M4Id%4C%i7GF%S2z5{*babFY#V}_Bhz$2N|Dn0P%nW>Vo}XIMdcmlz!pXCpPt? zrph*`yLTeYEA)8b0@cY^(lh-z=iNczkXh%*{%brWY6O&&E; zI8wemAr8zByWVS@bNS&c)LJ?+oMUXr>G$%{OaJ70)SpfLsC>HkAOV&N^M(OR6PuxA z=&0n4fI&l4NkEO&_n_T++q$y1a?uIiCt!=k>M>T9-jR(6EPlubNy?N_51)rmD2?EE#T?cf+(d1lLGJf+* zD8bEWhi~1|?lAJ)tPT)3Wv44L?>9qX_Z`g0G3vBDlVj>whH65(w6rxd9`slQ2JpTN zrYm;2EW4+2KJ8-xF+@B$NSrPjO1;KN?&Xi%AyjVJko=Gue$|oLVH1Lpae;<@fk6vy z79!N4p;{-4UE)n{$e*O_2WC)DbiWvkP>*hj*y5vEMB^ z-l8VQ%Nl_%J;_Nvg$Yv@o_+V0f_GQ`pwq#b{TI%H1&(uVB`s^twKI7XUIK74a#E~! zj)~1k;ii@~bs#6oJ=(bXUG^9;?@XCxh@FXFC*tfA)TlZ0j?*w`dM@Fx`k#WzM0cVP2Z^)wbpX z70g|Z>0cWerhf8|=B-szzFJDAhYp{~R_y2Ul%#(v?aF#NT6mCrX3WyUkgpB zF%|oCn}RBDMXHgulywqLD*1lMwxppW)I-_5GA($`W7D-mj3=m7V0U2d(SL;ie$$Gy z1n*JIw+@jJE>YiaobDGB$6f70`&^*v{nm_9|7aR~q#zTT_1yE6bil_I%13a`aJ&E= zVgl%?Js6KSTlun*K*#H5(K}=8I)mU3Ia_)CZPa(6tjD&dN3+8DSkFseS#4qz)sWy?JmrGw6QqcYI5l$@C|(N2Sx#w@SL*lE)OX9P;8b*T1m85 zP6&auw!L{E)cQt^%rFcqx(O>T6eZi)AZ^3fDb`1 z?f`N)tbqgY7{LvH!KU>%5NCY12Q0pcwp-`(7g%P$t2rLi@sWjVFM!EANC3T!a~zX1 z@*;B>(*jRJX&}$N`I5Ih)O&n3tfre#(2M_7`t+rSm@*lYvAvyzEtEeu9!O zA3){=ncR+~olI;?Z!_W9gp}X?Byi}795m(Jz-VtB>x$`H>2`b-qmQF<<<1`6GFZk) zf)kacQLhWe$G<_m%~o0L*@cXIZ)UTrP8ct88=EU+6?lZ0gwIx@`)y!%G-F|{zvP>nWilw_rE^?C(J*W>I2#j6cD8D z0COd4`x9wB7Uhg*R-f+GsBQ-S61aSA)n(Sxnu`V+GuR6EKdpy!~I8ICNv(@bJjvwC_=G+1S*MC5y9-AJL}2lRhfvNUdQMnkv&# zpiUmWY(OOu6(3e(OV9+KukXjCds}Bbi0-H;xDqm8vOfOf*@RiZJcUR@)*@F7EgAW zKY+}$F?qx(r>SFL?llh}Ncw|Ve*DSSigE6a_Lk42_C1fnQwI9fin4IDpDmXERdur5g?|exX<{(*u949|JHIhnGpHpX;#7f z4nL_A4k)#6c2k@Q82VuwY7sC@Q5&3fy^0;m;#;}_gn!)C&d`0RcPVb=7_Ca7IZ#oc zKO#iZdcU11;L!dMT+$lf|Cs(B8o$|&-45hn^OJ&-5m|Imn`Jm%#HLTqm@K%Z;&FRP z8wicMt^DQQyl#6Yb7tkDIMQ#oHkmrbITvf*z*faL>hK`lnFOUvxu-!#(YV;+k@*a1 zU#^4$$b~QW1^4!O89l|Z|2|6m@1n{7-#yo$1c1@1iWd19vnv3zZ{^j%lWCWKeyml zCe7s3XeUv9_-6EGJgMPuB%Uyu81^bGOT=xyzuq7BJFQjLo%H>Te$;t)D(QAfL_1$g z=*IxutI}BBJJmPB4>9~c+jH>|(R1AdySQGZb+|;25;om=ID9_MXYiaRM zx+MMW70#I9vj)bGPGBEky|U);#QY%>bJFY}_Cc@MC%Qw8`rzf+pKrt@g0>${spq69 zS@6#2=iIGnaf7u$NZOr9OHj{mPY&z zb0XV6gj}*cO|aiXPm7X*IqVY^C$ z?vjI<++WB_mXU+?dvRGdVi^n(mkB%K@cGei{V{bvvTY43z4o{lvu>ZK);AC_efWnM zPNbNrEK&uyRT3F-mUoZ_wVb})-X1u}f8i)3$1pc1x^UxF@gYGX*XBG~jf#x7Zv({1 zh{2*GSyyuOKbMiWb443)g(m&{&mc=?;r+%L;~TDcVN-A>feE5?77P6v`A{xNLu}(& zhNkN@sq?|a?aLMl3XL*Zu4}J~1r-1d8V_pp0MeY;sgl~ujv6-=Ah%U%H@&U$?czVR z5NLAcJm)vAj>|0d6QqcKd?Ho~sTNGgN;`lE-a`og_^l|}Qirh1{LXo6vTcleZ1aBR zNLQgM+c5JGxXO*Ox9vbmdw3GqBg}5r5zvP=?4S2nuN6TPRMJc?m})1<&dLN&r~cBE ziBuVY3-xG55wNn6Zgw**FaCj@{C)Auohj1NJ?6&f(G!E~YObje3OWHzwzK=1OF*aR zlb{2db5?ytva_Piay1o^;oPgq`(?|jF|(q0q&)y~=*x%%7?A_Ol6R~qdg;@OyMFh- zrAM5SblmP8W%mq?$8}Y!vK37-r6Qq&ODyexTQQ(%(@I{f$58oVIHe_Kri^ZM8$S@! z+ngU+xS2{s@h}JRO%a>(r1=S1IxFKjU5O~`ak)J@ddudYNm6^B%e5~bEFSyZ@^r#+ zOFH~eoB2`9nK0xAVahlc(nA@USg#$Nw~Go`K0$*4cEh`FC*MbBL|Y`1T=FMxpGTYZFd}l5I2v!N z0^DBhhYIz`mTwfAkREnV{R)uSzFC0AS+h6-@tLJvysB)YsKcl}Xk)}CgERrynew9E z_i~dzwtif6wkFfH7L@OA#dKZ-%^-E%REs%&00DUrUmGxM>i2k(bCv+x|SuA-xUJrzQP`QrKgrFJ$am$Nt{Z7luj8;pft&4EP=)dC;4{N$~U zuq@<=@X|p`i2EB2YB&6B_?k9?;RH9k5-?YgQBxWns?s%4F{dmg$z*(cCVid`4EhAQ z>RrCtRn;)VMGz4FJTrUUxFVOPvUPbPQoYBF3&+y5&%37UmEFECgz??4p z61c~+`hrmJci>k!Qj)wNr`NS`vgevy7Vr9z#B|HK=K`y$tVL05qM%=psT{_{;rUFB z0Yh{WUXm%}%T)ceUEmt1cKvnccQJ77bR0ePOG2Tso^KHUFZ&*lW9tIoM9kc2BBLRx zVf*s;-_}`!dd!P^M~}(&J9gBcm;AUuai6q++%?r;6foIpetA4ZZcL#9>{y4@mg zwUXaZ^F{JbSsfm_kS)}zhd%V_)ES9dC}_j*CfEa}Fr3a$ZrGe7%tN`>^HajA0B>1N z!N#SF>9tvkUg6!G=Z+Rxe)-vOy)@PU#Fdd}$UcJz&N;dXnd_m>miRx@C8xgn?gv(z zBiKE2926W{ofXL@c6CuEvvlFpRCX+7g0B9;oO%WbRc@zd!DwcH(|mAl_}0oD+@8laoafU7xi+V0^v2 zNOeG(c#D9JhaGA`^Zhu0_y=TdzTvXV#0ly)c2bYB1>-u;d=%??r^*%w*wS>69Du+p z|3XYZBHmP*{AQ%}_4=N#p|MllefvhcQ^g$Q8~Y$1IvZ^v0&)0nd^zXP!3-<+DPVg5 zxuHA3P-t-d&w+!P-0K&|KYw2P%XF*$|JVP@81&?m@bu@S1Bm^#di0v?{V!3P!`MFt zSr75Qra|yz$`JDF;o7B&$_xK48|uzQ3kThzHQEafs9Z`TG9su&l=AQS(A1Na?UNcm z87FP3f?P$N9uv9VRm&OYn7yxqM1K?gvjQXh5CjzGZVZxqQl&c-`rEHFAESrtp=4|9 z{_!H!1IU2h23?iyr^eIx2iKP{lz3uW9YyUjU{ns^tve`|S(Wp6|0QVRAF1t+Gu~dQ ze)1o`IjeqEsW&YEtQ^4Xr5H1T)nq6-LhJ0&{i_kDq$a6q2Wd z8z$_4R4~ucF@OmpK?xs*%>vFd&_=M<2aJxk0Zr&Ad~RpR*|jps^Rsf3w!6#QK<(@1 z?Z`msSWC9!+uZ)_Q9g7LA0{7^cIArtaGW#jUlu zKe$DIE(QO6`1p-d+m81qPhf_zzzI9LP$-l47jcIVsmtgv6^m>pEOnNxzt?R@X~8zX zMId@lOZUw-j`o;YI~ft}rA~zyj9qPk-R-fWNDrV6lWkXr&__0A!tcwysCfGd?w?~v zds=rHrd6Ra?R&P{Wc1>S^LJ)BI0bg7(O~~}*S2_?G1BRav9QN0M$X1)_&%#fp&mio z0TZ$s@~Eyg`PS(57zYZzAK$DwhyawmJIM_s@sWE|O!0ZTCUY1e>F=Coq?9(#eH3$l zc_N%c??n38-I}+e^fl%pU6yAdWvKpMG4aA2Hk*$xa5`)eZ+Lxp!DaF@!1F3sHv z9y#0Z#C77GnU~j1v!khiGqn`7+dSCqojG%A9K~mId3h%hW!y7^gjTmEmV{P(b<^$1 zIgzL&S#jBFe1UgAl*uzqI~nOnm1ATZ%hHbikkh>pvfLl8DOMNt9Wv#}LcZ}CZMUkog734{tW4-MFgIrR>FdT<*Yn<+p4yX$_zbxwz`D-ZZFKbRq8; za)d?Pz}ub&T6XGE{$X{BVN=t#lj9>x!`VG>wm4{vMv#t!w~@g8IBQdsD6J7NYQN!V z6;uC1PB?(GXvs^cyyD{d--o6iz3aX-B`u*ju~WJEhS(qxsZaF+qChMyIn=n2nM=5# zJaIk8a$!75^s%*;O6Q(vXR%`TTht=99no(GsATmOzIohU0^N2bP`b{HPsu z=1jw1fTzbex?vr|#}q()WJ*srXe!PcJ4Drt`}pv^Rm?fd-TuH{Rk2e)6e4mYnys(i znTV!~%oCuS-zfQD49blq6l2NeS^RWK@=`1R;mFY8u7>zP&z@N2A)A!p;m6)FwO8K# z=wDWBv@|I8JL&jKjS41Uammb&P_i8&{uAAsQQC7hIKU_7ZX51oe&zh1-(AnYMMT{l zeRKw1Q1J9bru7;|hykJ#1ndE%QG9Ee5(0~59!eW>7o~Wu+}S^wPkJx%Ak4O=d!`izp+`4Nn3N2HD{mqjbhV!L>qCFIqG>e4vv?RRBIS*;;| zM(M)KOulC{DRjSusr&+63J74lA>t*>gJ#8;P>MzB!0*?uK4h>bf5@~yJj~;S3t?et zV7)+v(<@ zbk(*-RFEBsqocsu1u3S;8g*Y|rd%+oz7_F$ppg#)Je%-4bF+g+eb)$en2BwqwOUw7 zH|z0iVfOYvKR)u`xoYUYBK!Y_3Zr^BDXSk18m-j5`4|Nd2C?NDn#s$R46F;=Iw9Rb zLf&kut?rWv&Q4EVR|WHZ_*WIG5??$`fv*2IK9~TDhd!ZiHhg^)#>i$$gCMOmxLu8G z%pPwVQL;^pg`GeXUfl(4;!Lj;^DD^id0~=a+{Qu#J09GUW(MZExpKVZ_{{`YA5P-dSfDSm2tn#KDN`J%co&U^gj# zWS@R_14M$Ha?a>kN^c3_>9y7^PGVsI&q|)ll{W*L1^k_oi-&_U3Tg372~gt$NSPIY z00V|V%mGBY!8sL@6sq@~Nj-6a$2bFf6Z7>kSkXR)7XWvs43O|&4qJ*;Adbc|MI!w* zxsiHg@BCuFQ_T`Oi&6tdz6#z(DLhgMlEy^_x}O}qvc_*z&g!uIIYzq!s(TB}N_GGh zI9)W`R2(&9c>6l)W8sSQ&jKAh-^jGIbzq-g&t}l`{xi1}{X@TDx-Koy2f-CI%K(`U z0?6yMa`oBV5T_cFO>-NNE}wNbd|D~uiFByx(;26^4>fnkMJzpz{t-LXBs&_c_5ibp z$v`*f_u+Bo18{O2@;-3280faT_OS6Q5O0(DP5u#oR7#I*R;K<*Y1xf&CRkOkjwKYK zS&eT5F7JS5(*%9+Gsb`9tTLwBsh15EwBydozeW=(U{EX}1=a$k3k&y& z!J6?Y(}P&y9>LPN%7^y^R0gL`mWe#HqNW-B(JOttB#FDd84>pb8tlZ3ZQbFDj2^x( zicg?BzjXC43@Qow1Yg@vsJ0dOaOc$6kG`5JnZsw%ZgIWiTa}rNnMCk zq(N9sK<;(l{g+;)Ip;UcM}}ArnTO6`L$fw#$*S~#c1Bs!z_w+0GfWTpt&nMt z-~|X5F1B-oE(H2lA7&qj(cn{WQD#AeNTsjqP@{jSPtCO-!L36oUle0LB*{H=?om~P zT-GC^PpW3<@q0hFv{Y3DG?~00+RJg1l0+Al9>E_%gyH9kV4{(N!&0$XE)?f^3F`_Q znKGoq=<6EWmF;YnFn;^=UkZc<;0%2Y5UUE?543g z`q}ebZ`dx4u7DJ-3K+#?!ES5L3sWyJ+Iq~(Z-99W*)o^-;`C5qd6yVys+L`uKF%I} zek2v|%^R&y zC07c$4e`m`gytJP5T!L!YzV7^ni;*kJph6d(Juv0(GfT+iqMIofJ(wPEsEznq{6f@>@s#S7ledu~-Mq6fID^|F=gYOI1j=1sev#$6xXjo;&Q6iGE~ z`wo$j`UVR?q<}Cin$QF%!!_d!5u4fd`YH5dgpC_A8}3KjTJwy81HI6CWl<_0tM4AM zNInA5*3KkhgH5%Wpx+0_$QR9SOI}{i9}E;9E|mEw`*ShcHuJmI(T{T?XN+V+?ry|s z-?a>Q!FT~yJk8lq&%5>0^ypmViTL@;I96`kb(Y2rr5WZvGD*XLU{p8ZHfK7RHAW<;pH z`QXj#GIcZ4H?&~XJQQ5YH({~j_?C!~nn0&OF1ybfcfMy?wF+NykKlf&roH01Hvzd< zEa+X@0iRo5#wK9>vL5t`gX92te`(i5d>Ci!nwKwEBABl|c=X+l{UzzoKB<55soz0Q zw#giZk?lY)IEnfQlk%i!%=F@Jv~ErN-uJf?D7>`&s!1#Rd|*vaY$flz(2uHYTNh4b zK4|bmsgi%?r|1~16iqlZWq$ui(&O;VJjb7Dl6-XbE!eL;SURY(83?ATOB4Z6nA1So zl^J8lSFU^=$m7H`(T?s^wt=_Deap{j(VI#f|QV z?VVGm@~)QSd)Hx!q?_@xtRZ%U)9^^EIqpyMmgA3BWuKAhAWtU^()s_0w*V5d$+Yzuo|s zp?xKod%i6b)#$PmjmRaCP{AAr|!Dyb>=AF`x#g) zr)37MxZ?a4{~AgcbEB$&!I4pgd?lFbv1pjesJ`=`~Zh62r!f7mY4AdnC$AO1f}bIlXH-#sBiK znJz9!t_9*?N!+mRME9t&_?Ct2Pj=k@#9|?xKajl*X>{8j?zsp}=(Dn4cN2LF|g!})#=fE>G9btfG{wB!d zvHzkdVz_kbIBBDpBge#=;gLR8mtU{t?OID{gayYnj(K<=m*P@t<9$*K4u!%K}^b36A^F68kUM;*KR#M=4tyFMA28VE-gvJ`8C zQU;B60#3e|l?bkbdX${&I!cG4^dbR2*D5F)i5LvfE8yEcaXv~p(^}*($}>GaD?dW? zSA87~!Td;tHHH1&YJCZ;%eoQ2WF@Z59%?TinfQ*Y2ocl6EqR}XYs9@qJdO=fl`Yt4 zab7(HCUIOB!4B1bX;IqejXwc3l@E=kcoWX8xdTAB#|UuFub+0JE(To_WCc zp9(bI`2VN)0CIoeUZtaBxnV4?Dp>vS|8#lgs{8N@ezR+MiY8P-$rl&dn7nQTgg~VJ z4MwrC9{~nfwjQr9%^hervxxp)TBc5s;o;qqmR9gyJ7d%Yarc=H(C*rHKU?niPkJi~ z9J8V0OB`FE-xUU0ftPUl4KU^9d<95~gGqi|D$)d zfZT@8vHLaB?I~>-)4ErYX)ewBdMF_>{iMRP%Q=VNuU?qa9%WZdh%)`dYp?>Id(Wm_ z^qp)Uf`V1LPQ7|iS(fjNh8tbYJ)gXv$<+9nUKlv0l2LsYLVtN!0J%y##vJLD0{cIT z?mw&34kS?0TqsttOzlXmpVF{ocTLrJ>sgKMyvR+Ze5Kpgq=hk;j0w5PceCqPx4U1i z;jVPGmA{&wpBoJFDqN3>vy!>#+3680a>&}4m&aI`%EG8YMp2e8P<#epU{(6`Jwz&= zPj_O|4(eBUE`26~e@v2fn!5XG$WXld&^&;yO-lBSL_93$BLvtEBUT=7?vcm1Cr=d z=a}kfdleu&^9fC~6C88l?ybiVhX74o)0icURYPY33^tC?B%;mV{6uEhzV`3bh$Ppsm=TEI_q4 z3p55<5S_jE3znSUWFem310!W{SyP7N+AhGYLNIHP@XRnyi;ehBbV6el=uwv2T0dh6 z-|@H$@78rAzbCm{dw7k7(QMnU#2KP)pq>I)CSL|ki2e^KizjN*6#O3r23|N(1>^Bj zg&KC3ruXM-A6@5R4lRh{5yws4Vu<{)p5IdVQfQ6f%Nr;`55P(}MKlQc%0xCh*qCTq z0d4`&RX41E--=VU>BiiwwnZI{bfY=}pcmlinnedp0ICwNd)uE4&NQG20ebIw$}{LI z(}||E`e#A6kD}jSBbo3tG@t{q^OxXr@l1hcz<}I7c)BSvjnp33!eAXQ%N5{=%^L4{ zmbv7b935&J9Xi4UK^d}n1E9{81oIXYT!;Tn8;gaITA%1%mF5i#k;fIgGwM9M+tZ)C z@c5^~Q6l5|5xX&Qz?Y+Zq8Uq=N6*XcQOm&*2wyAWp@g?nu|h~Skjr|Au+On%tTkO- zy!}B?;V^1LrQwk+QW(_4m0US02q-sz=iBYUb~#`eNDXNL%E+rj!d9W>?`z$9^0e|^ z+jzG{T`+WAIOk9(eS( zy|?cw5L>qEN;j1_^>{pRjJDgcsrCbr3MISQ;n=7r!TzcuYp9;!*W5^ti)nUGHy9Z2bRI+!!IjTg;uafuz`ZWF*IE)&5i7n1ZI=6uwghW z^47~mW4-&WFA|=M{kFVaeKh)q!=QWzhwGJOHQ~6^?%VWf*a$0zk;DW6{1E zRMsL;_S=Vo9HQfe7wnnu47^9I*{T$s-(kt_f>}_q)dWs_5G6W^CcNn7nbd`)I4>P@ z`rL|O2b`8KCF7d3`64V6l?tyZdY%wdg;L~rW_3-f1B56D##m*lpZ!BNR*(PBjQ?Z zZg>!va=Mam#=3#uZF=8njQ5=U7W(A@XZNdaz`cA6nnmuBWEw^Yf6Bo!0fW8|sX=nn+M4!oz%YQCJ!^c+i`>lQDJMVs&`S{? z(i8&8-x0T#KMrh+B`eBhVsK;eaz*#MtOo zxlL4**6xa1%Q^KSVyIX-pQy>XOWKyp?&uPv7NZJH=!TM=?NZ?_(9J5!Z7BO7!KkC0 z+@7W>7TBhjH}xz)_8E&;o_*5?2o2k_lL{V{{8xpy47j)|;Jhfv1nm}c8rycYVc3_> zIjMNJb~Z4oa(==vF;C?&zkH|Vts=f^}wC=7% zsf_>0Ko3iOFDbVO4h;>C3Tl_LE0E9qHqxZ-U&b4rW9gE+o3vF!#6!{j{xHG=CWKMk zD{jH)T;5@MFU#yMpQq1?w~IY`>-XwhbthxvxLa|D(ql+*o_!Q_E|hubFnKir)bL3m zqF+;hewK`nV;Un*fBMOgzSsuDmoJ$A{Q^q%HVfCZik029}g(5Er3+hA528 z{h*%lI>F{sQS^aPjRs2jnTod zFK+&NWEvF|Fm4A4WyB(H)Gt^OVzgH++Md~D{?nKnt%5`N$Thv`8a=GCojjt|*}VUj z;=|+_rAZ^NF<$6G=h9pl7`mnh@!7a+tp>X3r*GiuG|IMd0wMp(_j@`q@8W7mTxc#l zA4a}g*$+>{HgVBK5JUuBbecGur2M5vzrtV2j3SQ%cPI0n*BwM<*r4oqr^}OlZ+)(u zlHPP+6S$Q2mQ>Gf72 zUCDcUb=3p~1;BC~4zLbBZX4Zy9!r!5z|Dl6%y@1LEu)|L6b6zr&Oe_NsR+@x&8pqz)ON7r-?qB;E^NZ zi{HsT`jN`r6z>w@YuDD=d_qPApD2D%EV2TLMzgC1?mUt(vbinSFvxh|>7tnBzb8Sn z2IP3(Ord1(0>O8Jw#$Ul3L3Vr5|c6&ADK(rC{OiECfeXa`qZl z55K=};(KDD0==0A4*P3Ro+0KJso%+9)QU)gHrqY;rI}GA`}w587ZAu83%s{@0*A9q zBrLe?;Jkz(v1J?vv`UC`JGx&URG}8%Ryl&5j{wixUc%1r^)aFCl!gTL?eAGmWV*bJ zZJ@mRqoyjYu>={z3*8|Q!m2JPIr*%r#wcA3LmgI+*V@2+fshDWH{vk@gcGGFAV3q>o3gQjEBcaV)-*)EGsQ(9hZypcj|Mm@&otUy_ z9ib>&M9MZwwj`3QF_k?bAxmQhWz7A_>jhTsFA6QBrZhec~liT?k%IrV)1&%%#(&il*2r6bw8-LnzA|t7po!B}8L-quT zkb!C7erbpxo!Cxx`8o7@?ro~-6Y<+h`N@*g%H2oVPHmlIJNSAOQV%AVQ6y;HV995! z5j|vmW=0L;Vkv~YG`10}yP)o}QSm0nG2Ss%(o#TGA?kM1GKY%H{b|086(mDO5r>{n zV)4JBeA3G0iafuQkb+AyE{Tf9QYC@%oJJ6o zn>)!mcsB6pEC6FH#Y6Lrn=H(#CF7gjgi_ioGa0G&w(KI&Es>Q0d-iG=bOPr>u^^jL z{!~_4(YqqoS9|fhH-4(SSxy+5ktn1dWzNpRHY;lNBGgs#M4N}@|>og}2 zMRf%sh%Yg~BCjC8#>d6dK))Lz)!3Rt4|XJ+E(Ek*%99I!lsM~Gpq^tKsQ})cx0?l;waDzh4gIRk(HkAR{P!D_Eg>6|E{_#@^afQ ztxb~#W8muIud+U3D9)L)$faI_@Ly~Jn(=aS6x*&~;Yc4F&+Y=$EovduHkmftJMv0m z-t(5vkMp6;^88GWI2X?zPfGPD_spy-@(CoaJ(JhvK{>F+%5DYfaRWSc#>*uM;T3QAbWaD zG!XW=_)}*GqQ|RMUlG0QY5pg8n$P^o=|@mPl@l54OZmniaig8BYy3of+!L~kaqXjb z1|R#Zqm`$i)SgI#N0;M;v@H4v2}K&rCIqn&z8CKs%!@j63?mQ>jO*67C9cuLM(sS; z2ggD^288ad_V#wyw~yzaB4acTN?w}!Y8!{}hRsxwMtd4TT$C9nCp?bQTjurp7`QEf zQMkwfZ*#}D9sfRaFh27kPzrbnmkfJDNYZca1tFC3N_{b8u;`TDfEDyB(s>0vEML#{ zi8*ES(h~2puyRmr?(mIUp1L~@2KA%uXd4V*WnRTWo`12;8W6eL>AKgn{V;r}K)Pk5 ztE~vb;2vUT*0Duw#xJ<~!to}N{;~2p+}ktz*}SH{89E5-gEAIu70VKkpcv4~|1gU{ zdY+Dz(Z$AJh^!(7_cJe;y+KOPbbIL=%&zC^m43Fc`+PS2K=XZZn-|}dy!Smkge=8S ztp8$5#Zt~}&W)3H`l<*sBBa7d<=!H_y4<r?9zJm(&J}y&X3t z#UwNiD{xVH&rDns^X!0;k(_L16J6oCgF6T}M#ibU4cRgCpG)!udX* zgD@ta4q6YJfB(L6Ad`G^@TFKh;o7S=nQgJVtPZwSc`)<;>3+>ju>|9qm_^$_mqv4{ z&3}n`RdR*0nY68XZ)g~1+d9GZO|?f@BHges1b=wfV0iq|$Bagb!J;zxXE}Oqc9gtH z_r!D^L~En`s1N(P=RxG`K(;LuPxLQ$>(16`@)t&k`Sjq{;k}qSP%Lkx zpT>0Q7SaltKySkPHZBn%+JF}y7Am6<`lZe`=S|Hz_%{p&$jkW)m(j6bN143y4X3ZTxboh(=(=YNrl zGFaJiSGcMjcW1xuS8S694& z^nYEv!uq;ht4Jd;Um&gn>sGJmBCxZ%34(y)f!JfHxaenS77#;)_sXe^c{Ppew2;hC z*}I7BgfM`{%MAu6`yjys++{8VGYjvBo35bsGM|7LsFc8Fyc?LRaGMbY*X@%Gf;hv!yQ`IJay!!YrLmFrLe=qLoo+DTMCK z-Ghsp;?_5tLeAFJH8w^OLc_Y{m8Md#nOGQ02R#mYSkxJNHHwETI$j1DFFdmj-r2}F zmd6pXK>mibFuZzB5~Jl|L%p)=&tA;eoreB#=?|5gGy2^>0@nTkb9s4cMD2J~sX4@{ z?#)aLEF7Kyn%TIXK{-9Y8;mc~a;n_;uH5&PCvH{q?v}Llmfv%dZEE;W1xW-n`Ab(x zWB4MI?4T#rk*7Xi^h3kHA;f;(e!_K%InC-Hf1s5Hdq6p&x&b>c1Y56_mQWiv4(TqfGxezMFAWq za~w#U6^Ri^VR+d9mM?88wR-!f^&CO#A@=?Kn{%m^h%+tJO`69aM#D8*fw@RkS_))T z965BRzD?p*ld=5x1_nU{sBNaH{JIl;KrOIQ1g!@N**UOm-^U*62NCo$^LL zb!yQ(k9rRT&z-Q{(49WleMxk9uYN{BO_c8!tCe2QyhQtK)Vq^!4O8prr~fc%U}S&N z9EP7YhU75RTT)d$rhc_`rQp@zHOroT<)p>8#EmOiU2k#LrLo2_5O^m!tAXW3q2IHP zBrU?b5Hm-+^)=eB_0D|b9;^)tbpGk*_2fyZCv#bzK2?H3Gaur8k+Wqzlg2=)$XLkF zSbIEdZnSaQ`5R5Z@P_TnibuuUfz{?Qg3(-{wGs_To17SMevhv^AJ>l|Uq zEoyXw3FH*sV}P`4nQ&0#o z_?|J}M29rPn$nojQvB^$P2h_){fDn}1)e+K%&15dYX+S|2zEf}g3avE+$ik?5bgP9 z-i=nivwfSs(%v>6rmDqx^pdtxpsEV7#qy)o`YDwS)%WP*z)}wa1~R-MBJDd`0kqa< zGNh{ln#ZCF604gI&fXsQ{=;y0qL0zcj!0ZiC$3DTSOaB2H^+40L3Ie8`B`v`q3TOF zo{ba^lyjeq9_^tN#4Od4-`5-WOGsx@EhVWR6JoN6qk>;Q3sP#aB$$x@kZ*;*V~EQd z?FFy%<&<3tyJ;8gc*Yu4ig*9Twimk4W32fBXQ;KlmXt1` zOzHUAYCcXt8C&TH`2d{TJ&-n8!(fpk2)IA?B_Y2|KRaHrn0Yvj(cFmZZde-KL_djpt1|?RZQK*gJ6?&>DEcAefK8g&s3D%2~wp zXU@R8^{0|G4!8vC2ES}JMRdb=xuP^7 z7Yf}1>}J>1-ye%Um|igu9I)GWlU<6f?Wdtvw$-50)Cr0R?GzJKq`d22U;$+blZ289 zAIx>X{8c}?i`MKu=R(~{o+mtsHWFoDg$v-lY-ciwNximv819e-MX#qwmWsG{FgI8@ zpuppC-KA^xR)-4A9}cLwo^oDod9H()k)U}|Faxrtv_}+N=cW+&RC{#q`u(((a9w-X zJf-rkiv^qzwgksf%PEpp2+D%5EfhjA_~62f{^)-izazH zvw`{a(^+@2BnlOeR$u<{C0@dCNq6Icok5W`{T#X*^Gbr8=kMuDKgSwn$x-9Rwshv6 z*iZbTy^AhAo8aYO^x@d}yyIB}&?B8zAXreI#qof)6u%7y?8Z2Bj(2-0{HsxN0%;Dh`;YtQ3`6X5S!?DFJP~dky5f4G|Lo2E zPMOCg`8#E93OZAe8!2>clrhAQ@FwsxgeV(u71>T4GR}F>i z4(*eB_!ZFr+lZqdL4TT@bOa$ix7{%M%I^0pr7Q?U#01?=MkEsecfRnM z924YT^ardoLyI_P^{``vB|oXqbUWVQbXwP!U@y)OJjGtoA1q~U(B9}VJS7a=&?@4W zr=M_LZ;+y>bCZguw8(7(Sl!x2Pep!cNW%+&H)L6DZz8dT1__322px^Om}{d*2;SW`;$1sE_C7= zK$9?}quBX7rltr-zk(i{nteLS;qGV5(L~p6#njU(yP|krr@+=sTvfhc{6`^!fqwmg zd=hSYfacL$mus#n)V@?T9Y@MJiHzF-Sv)5Q2IDcTewx+*MU&=DAwOY6F5s?x zkTtcxA2&tTGg#2S`TSz&$O+wTXPc}jKYa}@_Tl9}n&e$-VPtC-uNiO>M?L}9MVKWl zViQMls7JeI9io$xuj86r*IFNE&l|+@UK=ksMK2jppGL9J_;tKR=M$5y?a21Y*XvQ;?L&crw;8@`W5rK4oa8I&3%z=M~sqE z1xvGqzCRg_E=Xnbx&hG8%>NMJ+8y$kU2pZSJYH#KVVzOPkYITCx-~bmq4#r!UG58b5X-Ws;AA{ow?9R?fIC zAK`Paw2C8(fs2ZPxQAE4Cg~kK<8)yn-?^r? zZmGsB&=Xy(H$zU#%nV$Qxi$3;Uk`#pE^sybpe3=UW~zk+rMdfw#uBU<*{0VTU-|Xy zo_g|emU!6Z$X~X$A7L{Xmf$6391G?;yN-byS}bBxrDoZJIu^p2*k^Ad1OkJ9kd1oY z6X|varY*m0sW~hG5XW)FF<-YKmG3fOR14=Z&j~&^BoE{8M^}QtxL;&%o;}R{$1$TS zHzE&1nhgHh4^+LZKz3Krhv1rzG#~kOHxGRXs;q9EA0{s!%3T`~GGDf{rN$x0;GKYt zPK#hZYSRRHchUx(V%uXZI6fzLT~7AnV$;z*N2JV8wwe!6PbhzW^HKE5t*C(J@xWax z8|YK4AuI)?{~H6uxi_|Z(4Ua9Z6}v1O9Hq23U*sPoqoIUX`cH@9&X+*vmJepH3EWB zpk`4pP$A9Kh-CF1CA1_( ze8GHed9C4XwfsGv%51ZNmc>}(xAfEdn~uiXE(rDq9PMKnQ!IKmVeXM{20Apx!`u({ zA3d;AS> zc@F+7Qp95)*}=1N@95N*K}lQ^-d}=QxM^q`ph^Bi``8C`6YV2u)=%e%ws$48pJ=Nl zlCJ8f&qldfy~}(xmw&up;;^OtJ%kJ1e-k=!fyI|YVO#L`5TF7{%A?N7dN~!4qOGvW z=3};?XNSbcFzEuzH}yiW_W_uN1Go}HjIan#@TBG?p!YzEAY%x6F4=$)M%lGx7|`ER z(2s;*UWESFyGnz9IrqpW;M`jaCEwX+oDb<_(dmeJ)VDtFrb1mLZ zN8hoJaAuQ@CaAGS<6E%)&x#ZH08KfwV(8xDX8-7H&!S`5A1WCEB2j)1C%qxQzaT&sqd5bG80&3ark@XIdfgUFrNZ+_bDwdeF5Vfki zGu4B;gBR$4`DsO7AZhm4!XRJkgk7c;5XK^YWMAjD^Fxa@cQa9u+sSwMxee@r=Epxm zp8x-oH2-%GVECUWne~4Y`v2~0^R$&>E|JG#;g7??W{)et2W~>Dhz(6T$Fx!N04Dty za_+MSgibt+*$PIcydfk3D zy5M{_JwoSXv*x4W-R@1eGhQITbK9AP!(mRcoVi?XEkRHXqTAnXq`wKmI63I2tZjXXC=C@?DLvWq9}7c zdDMm-tvlHL>gw2`{5L`x$%e80ZDmAk-{_351=Rropg3X@sBkdd7MnA!Jn?LTnyH?) ztPkSaLR=qGZJKE1GGf_I2@XF!KJD;q04$e#9JRl&R|^F&@A!1aH7a}nCg*Ocp{p8W z>hL=AtV{I9{P5~L8zv>6oejMqq`G~ZgzpYcLhvxO0*JT-54MjCy|S6or?q({Yjv(& z!f_w%b$7)zwZb2I$bnM|vk35Jj~34*ETlS0o~d5&x6+=^Ol0vD9_c155aNROa(q#{ z|91Lo&!!Ur{|e3!fdkR6WMak=EfN$e+!Y-x(*hT*AE7^0zO9h157$^@3nE$ajJ5wx z<$FyysIf*a+9sAn^Q#n}{O>Ih{#O+L|6oq0w zfsdt=iau_RkB}*6$cnjJ0hx&8_k_gIsd3$>N-MexsSL1Qdnx@7EgWeF}M@`|#?^8+y2@LTylHwN7qTa#t^#a&XIW=IpsFSFzla<$Q zG~2(r;`=!3anvDE6Vx;FAW*TNLo@>_<~U7^8N0FBfJr9o&Fhsr(&%0XNpz>>xyI$u zv!8UUUXRdjeH^y?2CPpT1u&>TfsNOJOk3y#5ztM2aCd&+5r0iRuw*S>5_8Rh^7fs^ z0j0}v{HN~5tXVF$143#PI0j_(UFsNOHWCKf#WAc8QcXEE0kh5l?WT)*y>rR=_Dcr! zfk6a?V^J3`W?#jm6iQNqQ}Glx{G0)cUu?+$3uZg;zM&hfUl>qYnY#crE?UDi1dY{* zLsIzf)1mxDvlX4?D4sSI`=xAE@T^p()0mS=T?WTAM1D?EfcJngR*Le)~H9%=liy0I<;nP zSNZc?eYIyNl3s?M?q-zkCv4Z7eAss9uxkFOqki!7$jCd<_>sxUc?%h>QnG^?8<*X2 zFbQzYhe}9Xa*C*b-*wNT`Uul-cW+!nq@rc~2?Z##r*rtIqHJp{$9G zrMQF(!|0^19a0}rC!V@}^?;A7Z(OfYbS4L?5uM8tc-SvqU^kIv6N?y13p*3enQk_% z1tcB+p;=mgbN1MOIhKEMEag{C-WGE`|5Bv0@7BO%@x;%2>QZBY77necpFe*V&y8BS zuV9|4{r>$qHjZrDE699g6NVuJ^T$v{Sdt*ZK2PaDrej5d3oJ9Lp1BPg=Tsc&RC|M1 zS`CUg%Dtz!OQV-u42xzyd6t6sQ>wpPJ?#j@eeJZhigXnRa3qh-#o#$#+6jtN22T%S zAL|oVC=9z-k`ptnk4>+ZZ`&P`aHOrI>%=}zuhmpWHuE9I6AK-R$ixK0g;6l(b40+s zA$QDTQk)K`*3Bi5<&#S6M#A?Pez;i65!1v>vlSEnR-Pc zkOb9pv6ovG?xg)$`A+zBbuFWOe7-qfw!PQ(i1X1#Z}QL<#NI&#k0tk3~(oM?{-m$T~iPmn!lr4$#s2#Y6BPHWEY03d}bOrZTi zKc`Va6PgdM+DMQzlQ!G#-9A2o2|pAU@0KPbmZYuc_&{cA9oa>QNAR=QA@LqkMHe>7 z6_FSwsJ&M#Bd^T4T;u8~A=v{;MOMlM>t)T(thGwE&c0pOLB`GbkHqo+-rxRL zjQ=l#;Iba648AC+Z7=h4_4+;8SlQ+88z5jE8Xl@$9%9aIeN-fY?WWc7Vu5q(qKw6k zU?#}#|4e5ZdI~?A6jtC#b@H@!LP?Gvg(dbIvtMntt2u+>E4*G zNW%!~o5j0?IwWH+jx<3KZi{*DZWR!-^n8OsRy}(@cB|{LQb{zk)P6t5!yh-f!jWKz z%O1t<4gu;*W;NKIjE{gVV32yxp-n2`-M5?4?V2`)FVeGD$@csEB|FV#OZHrMEWWiK zspbJtw0Mw9cXhSqTmwUxHapl@!#S4X1oLzmWB`p`|GMh(d^XSLn3uaN9Z)i?JoKNw zVA~lt+AazYLmpe7)T0`6%&H_k+&y|nWZku_!r;3G@9y~@4P0i3aI@Q&rhYpm=?n({ zwElM6+Q8Xx&?i6lI5EX3SBZEymT$0g7hVN%eG^i1B8$-jQ8!wdu^8gL0@2w?oKm(M zRKGZe`z`l`%|Mo1e&V!AC@Xtr{zjdu@V-0qKifTNFPT`34;IpxtRaqyFs@R5XWqjL zgtcog9-9!kl~uhqCGtqU7N&qO=6h7T$ajOAT^{>k=+MO_^_8GO`iu+}iAB~)Fr@qF z3Md}Pv&xKa_w^WFqC-Bib@Jrms90o{*TeH)0=Lv3-O?1`W&>6E(7lhJ|Kh3g|5M{d zh(K=ENKS?!xPck053{hlOKE0-q-m79Z`kC^xP(;cxvCREi)Jz^E^+5O<=Iq3PK?gs zU7MH{h#*)jqHeUurDOtU2%`$aQjmLt0=uOb4t9`0= zrIRZo=4RaAt=jyrn@G80*Zn2C1&2Mjs1+=JX}U2=nE8^yL6?Xez0hhahB{TB8*|L> zqfJeTG$Z-D40fS`ELtT2*GfrhXS@AVH2Fv>cXY&4(Chj)lpz02x8nb}y5TEk8txYy?pI=P}UW>JWqeY_VzYw8F5PgUd43tU5C@y ziap~xi!HKQ{fL*o8$)z%hDI~wNk5J{(fESR*&|X*~(%= zcfz?FVQKgVi8Rmg>XP&vi!tJ_P2GT4?z!8M18U1z**BgGt;p}P_yYW&?F@$218z$q zT62c=3Ewn+&0`nyN!XK<;2~+#su2B9I3_ z;bI@krX4y&SMPfP8kHV}J#ibP>ZR2>Jne5wWj|~(@Mweg%C+*+AjgN9L&)cZI?WA9 z`UxOt&V3y%gdX(}$h2)v5&LQ6>h}rP7?6e1OeKP^Kc%3~jEIuT569uG&omW$(L+bd1>V5aY*TpEdnLd{+k_vl?u0 zOSaI#osE944~D8#@o$R}DSD!rrVu8W_5k*nZRe(FoOxt-J)Ulu`fKsqKi83d0)tO` z2%@_UBf69+1=-zu^{ZrKR`6?$7OygeC`mx$h~Tp(5pkZ~!G)FDRIzNb0Ym zBA0m5beJ4Hzttb-xh6h+?yh$Jbn??oHFM`Bb_*pUo0S+iA7(nZAKQf82&L-LDxegl z0%RX9!?h8A55*A`_Qr>jrXRIHdaD@x%g0e=%<33XBvj!sW3EFU0%QRoF6<3nEgsNx z)CF1~a12rN;N)mp=@?Wsko&u%krt?G@iFkptuC2n`$Ip1o*COGoSwa>=(!I!XJkMt zV`eoB>&t4|FiVYPdjf)`27-fO_o^={rtLobT=)hUIR8ONW$zvOF7!k*rg30nfJ8`j zrQOxRikPmf8K3xYiyWwYsUeh$Txoyvjkn`|TLa0MlelZRI23TmMBD~2Q!e{L3m}Kk z+@s$d3+ALoz9(NP51{R+mO~?B@&aP%qq)sZT_(qWt=!|>YTTJT`PJ`V0+@dVT>e`b z-~R~x|9`!c{4f1)hftd@_q8Ttm6Te6AbGn=XRJlsutXA`RMNrocj?{Uws zX&iSvY`JxLq37)UZrPHzvlv7NP)G;;PdJ9eg~fA)bh|%Jr!mHswAuA0upUB?G;)+C zb3D2S_7@zzBlRl0B}cmhb8GJn6Rkg3O);iTbijVeuFN<6R8n@eVPDG{qcC8ilPLQPW31d^(q0 znqo3=a_7-cwVTes)#SLFcuEDrd%l?nJ&b?0gggY+vO^GG5b9SMBG#nG#`~%BgSf~c zM<-ieNUr~~Mk8PRveo{cm?Ey@@6%34vp)kj`SUReZ9V54AuT<@qD$~71|gkXp{UC8Of zO8hJwZT@O3Hc_rU7a7Wml2oL}J4o^c@jT=fCB9`hF*+wf+Ygp>VSOY}?6GON7|MCj zg&k2C!dX}SnGG^iiLAL!RBzcmxtz1xaM?6_%Qvh#O7QpEQ-AT+(^@mruBKdVTFHO- zdM{J}^g;w*w{NXEiqA+qxW)KkZzul`6$lay92%WkC$7{b)Z&z!OVX{ab?C)rO*;M&6+%)lnVmeBNjKp?95sF`@+D8ZJ7v zr_47^=Z@^R3J`8O*BXvRS#mnRG zRL`-$Rg2yOo03KLx*m-4JXleC?S{+T>GGkF2$c*SjuszzSnpIHwlR+(10;1gKQeQ} z5B=fZ3p_=IFY<~eenhduXc^KGzGRz4T!kj{3C{cJlXAN%`;%AuP-jTLf$LJn|0^YxaQsehxl;q5~dj zZkxvv!6iEmglgJq9eBNWZuOIbbs}f@_pmlC0P#~<6NpAaiaHC3Lb&3IC2}g%<_8fy zL<0Y3{n^>Nb#a%xcWE;BhXh;YuqOr()zmeLsZ@{f;TLRs#cqg3v(G-Y?3P)H1?0O8 zK+@!V!i>dV{YCzXjV07OC2jm13YSBsG+KwmRC1&VH;a6aOp$@BJvaJ+G*jcxiaIZU zkt4nBYI$kA?LnxI&W{Gq*Y;lC00EBh&SZ-ma?AJ>=s9U@99Sp%=KEpwl8H=LnIgEOvm+ySg;y)G=}ERco7_29+*I}9nB8`hL&i^+!D)0;X^ zF4$0P9_WeZ?S1K$dqVuk+x5;po$RXglR&6**G*i=3PUzq61(+ zuiA`%{k&C>+P3uV$?vH8}C)J)hYhOvcS|k3D#R+ZWv!|s>h$V*AqP{ zm0S6gVV}4jc+5Kes;spQHv{-aY!6epckj>_n=(RV>E{!rInSoJIvhvsrWMZK0(y3Dg~ z$cCo!DNFf1feS0Y^5^P<4eHZE3m6_C@&k|k3^N(@LJ%m{gnRJ4m_SW2f0h=MA0bqN zNJ53vTFA8rCZ1k$Ik{VX{|_bq4u3tKx3@gISK5v)je?kQ_5uk|M^q^lq|tk*lQ99x zy}`2XYq?{?ihRqI37Zg|TN?u_T5Va+xNlsZeO$RLbYVx2gdo46E1@E(=>t8ZfQa0H zy<=!w>spoHP$SX1=hC@dzEQl@E_HXZkEX7(R2e>i0IbM>kzpU~+QxN*?vBHo8?qD(i_eB)N^@g^_sLt;0524|YBi-6$gsk> zt=qr(rnhU%AV;U9kR!BDTg*UVd@*s}LV^o$svgyB4Z32eHo%}W4&rqL~FC90EQQ1rW z-;&_+_UE(ugyRY=*n3C`mLy7mB2-DWUIdM04VyxMG0=EB-Spz8AV-x_GtQrgsEi|` z-jD9HDJ{n8TOBmXcR_3x*`R%auQvx-LF<&(ZGgmTC$%__ph)#zo{luvHM|a4Z2FRn-oZxrzBwj!%j3JC1K^P=V9{hXms7 zuZ3T{gvzg*pPFty<$YhIt@Jy`HV=yrKx~{f%mO^I01I>L!E!N-WLJ~J1spPaJ#;2Y zxxsKeMb7xpj#N&q^ru;cYQ)V3SDKf0fyfM*t;T zQ0Tj7I|H+8X6E3%+C?KbmsV3Bx%kV$#~1ZDPWziH7KI`4`xjdFj~WOVOGr~N#2Ny{ zMK+iOOAGv?=?N)+&IKRzN8j1p^XZz`DUjd0=>_>OuB(1(dz1HSZhHMPm&O6-2Rg^3 zw{556KYycTF-yWXp>t@%$TuS`RK0t;0*Nhj_QyGD;@_7-k9z&8hZmf(l(r)m`X_pGu^*e&rj3}PI$;fTza(V!)7#a+TY}Gk+bE?vQSnk z?}y!7w2Jd4u_2v*S7Z|(GWdMCHACmeKI;5$JI=z`t}c~d!l6sv#t-)r?@{8Zl6ZXH}tIYfIuR_v-@h(`mxrhhxiR zvTvhxw=m=jz(;{FGd`I0)-mEf%9L5Q{lwF`{z7w)F)WyDy|xd-zLk=ySaeh=SzDim zO9xY2S6Kog;3#efAm$1rfD9nt?&fU1a99$Hhq-L8adNBRc-pNQ8PePczlG^=jQc-j z1MC>$BmAqSne}Mp4QAjX*<<+5Kn=n}ES&Y*a%>pPZtKsA{6mp$N4fLp?) z&TO`K2mO4&3OHvc3q~jnjzHV2_e!Zr0s3bW>$^bGKFVj;n}x>zUS_&1Jpdm*-h{Jp(|+m z&UqPHs*>EyAmWmn>)hPjKrknN)__#CUsdc2ZZ?i)u>+#~OQ*GjPFs(9=#M`@)s>38vrof) zg{S$X!%m0g8VwxRCs0h8Zp5iB#4LgKj9G>E$*K7YkacV#L&K(qRetbT{iQkH@6v-u zg~xRKnw5OD58DX045_1^uz-56?PwgLnb7TTEQ50RaGluZHguvIWhIxxIW=j`HSZ)S z8my3Asgl-9lo8E(GgbZ-&tVMvN)S|nj$WbkoJ{I|<;pAyhMAUk=W6#ySKieagWa&u zkr)>FH8Kgzx(riq%Mjh9t>!I7fJq#SwRNP;yG48NH^;KXze>6HO++Mn8#wwTNn1H^ zyF^7Se5t&69(ftuj~j;86o9e<;wWb{Wfjce-0h#IpA^rMj9jx{|Ew-s&FZTdxK7(@ z?Q{^!P|hYO?dBW`vk&;mhaqC|FUKhOR}wuKhT?TRv~a6~5723HNVq#+pcu*yspw9r z-_f?rW9d)V&YuiWv^?^q)QsO+NXSC5hW&kJHwHSKN8Y5|VeyYr+`x{Xh47qTIcIx5<&mWQOQ=SR&%z> z>cw&I_#-8q-kK3%x2o8NWp)!iv*-cBWbPLD9YXz1Ct>Etshl1-2YN3Q(XEdf`<3Gw ziLp%mq=fPe<(&vRzND->ZD;jxk-!;gJGnI-H zW~e{`lsQ@<NwAj9Fi?)Zn_QLt%#T&2$+k>)BqJ0BPzJeM%lo-wv z#}?)ykq%iwI?TPJ?@LW!57km0^L|j%%1RK7{q#E_AaJ+%H~!Nl4wp4Pjz2>j=AZt; z5@7H`)^r1g!5B*fB}zG1NWu$!r44aqJbdO_9r7IBJN;$;r|(sl3@fD{?pNi{B|Zk7 zN4kW^KJg|kV!yyv&E*9y0FOG82{d*nHUM8ea1RhxBT_Wr=RP$K%bm`K@Akg2j>|6n zV!pM}XZF%u{YRPx7z^Ob;-8}i4`9UISbTSYSE*@irRw#iN2Oy^4KF@DJjj!{sNUp| zdY0sKk!Sc(`b%#2^0&5m+kih5%Hlf!d=XzGQBrf$JF^cd(q>+SNoxy_5?b?m*x}RE z>@%dqo?;YXNjQpaEmkQ3X9vhayfFhr7xA|sjip7N^i2R1>@&Q;KfDB!^w7#L&Fb)N zYK)N1H6lcf4yy_d_luz?&8ACEuAM5aJm9n$vl;Em_b-%)zoA%pNKYf=wS82|cw{lp z`K)i3g8MTwvD|py(cOCK9mh}?KcJM$`zx3R9-kDO(x)-}oAPzEOIS~25dpLx$dKq7 zUyTnT?_%ndyYb?To79-`M`S0Haj*DZeE4pM(PN0U#|J{*Msq(L9nU|o_t{3~NSiMW zuzTX~pa<}E=mK5%>;*96hoPl(|5xs;`vP>Eo;CWr%6Z3He(AY;SqpB&kz&D+i2)Dn zREy>kLUau@3ir$<@N9@*r>r~e%^BzQSSuf}+d3)96KiXWWafe0V&*Tl=>U>(2U-a| zrjJiC){T61OW55VdiN{4cd$#77CEwJ8%(e|yd=ZV6ST@lm1YUqQ*8&ZVpfzSkf71` zL=G+Xcf9LrY`3<}RW-Qsm=E>H^h9RG=+VyNpPQHFu^`5RG=mwU&sOx9Er{y|h&#qy zujRZz$EyKXOPJ^Tn`VArUuhVOh8!HzR>Vzef*gvUC52L;^cutd^o*KQKIieuc|0lg zuXF8g^EB|ZJDR}^Kq+wCZ{URtXy-to1ZYAQ5b8Zjf)_XUFtl5sti|Kf5wzxBgSZ%p z{4NdkTQw~wzEkUDPm1kuCz0T=NKfm`)0CC0fQp&!HUpXd4p6VBX zId%W3r}6!B=cP`ID4XS??z{umyBaGM37j292&Epglqm-lj9Pr$9&&s=0SaTJEAg`j zBsIP-b8!&Bd07Nm6}%BP%$Eod@k`+FSA!wNjC4#n#j9mmJ0iKoSR=?>gC08dIUX#bZx{EK<<>uYFW?pMUdY%!1qe z@?AV7-sPo0z0?k721|2>MCeKkg|R+-6O&{81jNaj!+3o~-~RLHQ{?x};?J$}D1?IsCPcMh6d8Adx|y3XT?RY0Un zkKqFbQfD%R)2wP3W|OtXJxiK0YuAT^Nz+>NuZ#I;hcKRAxJ$lQK)wb|b7`sE$9nX5O%DBr=^ za7;?$6Ic zu$=3KPxE*xFUm(K#2UpdJ5Plc;3@<39Wy;ODqgJbnVUORxbnKlE;&^{+p1IS*GZ}} zOXvW!M!x_xA>%<(W0^UP=_TyZ4Su6{4`o#`B=+P*PMmU%G+QNWeyg9obJoVo8GJTL zV7patiXGVzAqxKMhzOb^z@UizOHeRT@a(TuXAEViZY$DmGBXf9wlk_d zWD=PFLCj1-!GpD*x8+!jXDPp#`!IaWa4?zNUMRMfTxc-XaA>g zHpKb02Si0O0we~&gm@$Lmplqu`v$9Q^*w~{nNV3yCTxm%;I&@xhJpBhg}`gDNNqA zU>rI^j`d=!(XBCEE?~|I1S)5J2BXu3EZ#v<06kFYl8&(&^f5}nwI?v(U47?B@tnEI zbuA|2j(}99{L~cUPw`7(X9L8joZl$Fp-D5O6v}>L$#=?F{@b3QLe%57TQ6pgn1>Dy zX8oEUDCVF3^@b)3E>j8R#VqMib7g%7XqBz_rO3Jdo$Q!;FuN-JNDRW_?8@=)d#)n# z;?>KT(}WMk_T z^_BH|vE!tb?3_6&tHM4d5-4{S!BMNa1>#WlU;Rk}AL2)rrh|9Hb}cemc(cXV;R zn#l!$Z{1=JPan5hOuYfieZisD%)#mF6%}+E;4YfbA?yHVu576RPdT#*DKWOzr5HLuR0vjGG(@Qus>hrb5sRxH9_Z18Q^oU}c@)21y3q6=73TMx)Bol6OSrfbJ9;VV{VaoBFCjS zuDn^Xil<0NOaKRY9yvRR@EV;dbD%~fsv#N;18XmtmY@tNjKKVG*&yYgYm-dTf6o9XKh`ny9JdyV6>1PVb|dpXdVT-uySV0ljrao z2l#zEo@T5_&0Bt3j zt4JiWlciI*h)vUb6o@h+UbdXJi1HPc5oSE|;^``>pO##o_iSEpr!#2P&@mDaj zqbz=rMLLX)v3o+Tp4;k;^zWxV{@FLPThr={=wqI67NtkOX5!rX4QV5J^177n39b6g7nod6RS5kPZwJG z9G^8Y#xdLR#HYwtfPYXys-*6ad6tq%ZLN+{@Puuem9CMi?e`wOD+xSoZ@+k6N_+WM z>{lfTHbZCe1Kf*85PD+^HPbJyOJ#NuW?m5N0%~}=a?O@nrxbX;+^A`I2Wr!K`3~}{ z@6whVZ|iLUbeMBLl^aJrOrXsHU-L5ZdvGdN3|CI~0@~&p#-9NC^W^QTT_wI@OMB(T zsaXj^yBg70R~MI=9@(c)%R|hh5;Bf|I9$xZ#-`79ytpimnUC^dVU);TUIMna^H$1jv3)zXOh_YsvZAd~YqNv1FLI^3cjTuYG zIw2{_5Tb@8@~$=XmbpKJMS~`+n~~e*ZY;V8(mq zeO+^2=Xt(f=j#;-L#V#>?aR@nvl=7+4IxfmRI#x%*l6gHurCR`2!0A&2Sun2&py^qN_|xUF52W!mI8_tbMXOhdj@8r19T`g@`-a{(m%T zEcXg#+r77O5r@_H&WjLQ{=ig!57I%-6<2}oLcaETMP;I)nX>7nGu;bTb7uPz7Ag(U zKO}$LihZY-AT`^{F^VTWM^Z0fAT_P0a{$ruS)MfPz$Z1|9pGf1q1`HcT-58l|K!J^ zeLc4{hJBp3ka`^3XBg+vL|kpyvgOf_nSyZjB0FzpC?4qY@*|hfi_Yy`mIXlwYU5qjfjl39dt6Mz(fqBYnIDyIb zZHI5)aab2PGXZ%~{Y+s;A^e1U>Y}B}cd{?odJgx%Zg9jVA{sbY2_eKJmxq#u3Nlb; zBvJg;ruCN)nI|Vw=SxJJ9tYht{qaHFPV)`_xA7k$fZB*vJsfsJ=HJjR{jVHpKWvt- zs@U%HEwDa!CaXVO97exftnS~B@Y9+Wpxy$WGGr-+#z z@%?7?q+RNWoO`kUHw>`+5f!7yv+B(cbuvIY*dA!UeTjyO!=w+J?=5L3Qv+}yXakj(+|(nt(i&{XPT8Lg zoPPq_e{v)Fa12MgqW&6tH#IELly`0k&KIt0SvR~2ebRUwEf)LJ)D>#O)B+^9k$(E^jZsE zC@t1tY%-u=*8h-MY&YrzS{*z)97C!jdqWy?p{%@CA-|JDgev4IrHeW9H9z|_zr1-O zIxCT=JJKImFUTmP{{k_!7u|t;t&kqPEDjFn|f7zpb?2jqX?SZ)tPZT&8ivchLPLZ z^xYt#ME`yJ3k!n;1r_QjQOq<{fbTiYA>GsM-tc#mwEI&}x@BgvWSZWXt`6#NIbF1o z9WE8nlKz~1J%oy(El z88utNs4H>GBgd!YV)vD@Nx&`Uwig$%DDuqEmp*q6&=IA%URS8Tle+#lEBGI{zW?gS z|3$gOa5(!jrjfW(T*jW3&0WSY_M!(AcG2!`3wN>*UE4jSKFBlyX~nMtCUcJgYi6E#no_;IGzZ!VIsPCjRlhFhS|`#o z!Hxt-K)W^K%G!?e%urS>-V>43tkwC3X?>*(N^+q#bwz9nEE!cPpYP9flk?hLo7Onf zX+((pjYYBmK0$hkqavT42ui}#B9rvBVu5e(K=-=UOK9E6CxhYI;PN}SAG^cJW{1F? zN!&v_+urJC1gVxSPF>3Y*a-~M>fY99EiL&;gqw=_yzt%1OiEmsJ>?-0{?tM6RY{yM zx9oOBr-#f?i(>Hw=yqqGSrzVzCuh5lk5^!gwcv@v<@X9z%jIKWce;6QN5BohM={2e zvM{v>rdU*m_xPMj{PD0^)QK zEfkpe4TKqkEZd7H`150{I_=3=>!qo6t-Yd|@3G1zH*y?Z>=KU_R!WqufSbyzj=GVt z@N4`c7*II~A)TD}+<4h!zCZ7Xuu6FzrTXaQCh6ZECv#bG5FZ19QUS3@5>OI$wC)U7 zAat^2hAJP*$s6>T6-AtGin>`it9;$pX@g{~atVpIx2dbIC4hFXYqxo7PIu z2rHn|$3r||aMcN)D!MT1%MDqmI}{IT zHZdjhmVA;l5Oa^%d|z#F4JSveU--Rl+G8jsmUZO!F%b&%h#Wh?u%d5*ElUJKnXgTo zW@TuEu5K{ka^qKV8(ZXBbL#~YrGXZkfI*~oJjak-8$zm&tOaSQFeR&uZs-Xe!Fwa( znu{w2Fe>?yrR5h4_vPHP&yRDxU%oe?^VT;{1QOi_z8jW<6ApCg#PRMs=p72CwEkNW zBcMSQoSK}4Jiqry3T&6{UYR5FgA(+TU>yn5niQO?40BmXW-R>dK>=l3dW$!Bxs1*z;Nm zEbE!A1P!?Y@C}|hsFz{SJey570-I1BrSTSuB~N+=motxL_3LLmidk7NvP*jC=)*Qw z*{@FGzJFMykrjg=5%FBi!?XtciW~SJJTx$=B-47>!=B!ktl#U^JiO7DW%t{~2|c=$ zos@I*(7d7(N8xoE(9^^*CGefIhTvN1Oj}kZKJfjnX;M`tc`%{bJ*jirEOKlfJ^tLP z{E}Y4LOrJS)_|H6|I6ELuxqS!;Qg2nUWddC%|j0CZO|7^cO8-fQrhU}wAQjY4FgiA zNQQDjkKsi8v$PD@#^RYjuny3o`~y3x0H|1Tz{)R=8^y~2zi79{U&T*_Spz_}f+V%V zR|FMziJ_vY=SGcP9@gnq3L3X)sr`xxIM=IU$}x<$uGl}`rodhM*1k`R$M^7nTwuoa z`X4zDw&SGsj=leHnoiCh*FW#R=3%z<4b<;8ObG%V^gQFdnWj47&o)t(G=X#V-c-Pj zHlIU(3EOAlbb*%5^CqxX;y`0cva8f3hvyq)I- ztd(n|ROOBco-I&n_Uez(IVNEjJkIXHP)`I+c>-D{t#zEuuML(jCn`U|_ZhAY!@JP* zYehqYLwNVs_9xame>>c7FSrr-|r946qgw+Mxs)hYNO2N30!0#r|pnU4ssH}~P1 z6S&oP5lN=|s?*V+1;w)fR&#`}S`WqU$MA%DUoXlm$&)|5J(^~CG}n;h!KT77O+MJ` zUW6yR9gyh|6Rz}Kd0LWC)ZBXbU=REJIx~QBaBOw%+9!KUF`x|&M$4ok_DJc|uN{=s z=ZSQ5Euw|})Al;ZipO8EPe=3^!(*p=S`F*fv}x684^18qQ)G0eyRn*IS9^IZZz9q* ztLwU7W@LQX|Ln!bN)7INsWo>}F8)vEg8wrC_J8%-|4PAkEZd)Ff?%GLP5$d$i3AW+ z;+p9d3RZM++dQYNV+84U@#&N$w`gpi$y@%y$Sk#A75^|<=Fy+wjb&Yv7*9B~AKTG4 zDB+(gB^2}ub+)DUkZ0LZqlkTCnjgliG*#O6qB1{RIK@>%2V_Q9Ha~%(48o96%RYl5 zT2S}Wi}06Zp(&}D4c7~yYIH?d&LVfjlG0s4!Gqo%xWO4Nx(C>P&oJ#6rain%Yihj< zL{Kig&xnrwGf2wey7=qt)VSpp2h08%d2o!$-6ZLye1qB3bjfyxoDoilBe-| zibD@FZ7N}lo-RVP4f|MTVwUQC{iWWXj(eJPe*dr1>cKtw`K~41X5e3v+ zq=s{rg;^7uALM8}J=C@3EpjO9K5#J_6w>Mz&dB?k?&<&h2aWY|`skazE4tht@~&*% zg~hzPz3)u>Ut)wiSPZdH3;p1buq2-^}3dF2kgaXh*T ze}|P$3-P0a{_G%gHy8ajcy(a7U8p>#Bu%9&;PjwRGBf%`kI+D$_E(z4u2U^MukUbP zsLAy{Qe9mf>nSZrU{NeqHJWRWzJvx*&Wk$Yp88f7X$rY>GdcnO9;217Uafq7-#zoS z!|2(WFOMI^6n}l;8CNm5*F79=Erf!YiOX&AS8LiL5oj$F#zi z!r&6j2rydjQ%|jql@CBSN9f&3}guyZ>U@oss$Y^Y0;)uMYcQjM& zfFM~YH-P?XlF?sh9J_0ke89rSLv3-hAzAT!Zo;`Xah`E2kk?FTv3Eu@cCybvjr#FGPlZ3Qlab`acgc}BLJBM^B}47rywN50{*pq@ ze|KPZmi2NvK1LZX!nuwF1Qyl2d7p8lO7~mGFl^Yfir3jf@_KpJs-gR|8%;vBasKPS zS3t7x{tmlmZ;8nWake~4iFyjyWNRhQ>W`ohMT?E>XnJLZdx?mZG(mmU;`?4O%O5)c z3)ykLy*)G9c>E)RdS(z(m-qE%2awj^9D;`_!Ft$o9CN*T3J`OsZ(0u%Pc|8PD1+htRCjYqg?|05 z`KKV))hO*tLp?LSZ&{BTB7xmaWb;RX3xA`T=hK4dOqymXHHvicgO;9F;_NJ6y#WRblVx6a896Oi2RGq-@6Czi4K_KShWKz+}wN3{gTmFH}(k*sflFv zywU0X72ht=J_)|$s%#@D?<<5S`!FSt^q6K^9eSPN+&MM>@NsFomV3?O(A1-@%am^_ z+&3fhU<{jnt0BU74^ak)w(ZA&5*wsa)(I>)5cRHDmx0xsTeGBE!X4Fd z3Igw-WLgsWcPsA~c&cDUvXA z_Zms+9@{8TW)>-H4LNuxs4VvFrOtTrm+V*!;1N=Pscpy-n>Y>AWi4R0z z;X8pUG;&J!5_J|!9dsJH!Iov3HsXJ2G&q0xKxSP{wGy-*KOfO*nl52C5T52!lUs-7s1wX|F?kje0;3$O(e7*_sZ%ogZO@ zbBrLo4!l?3GZ*aGhg|6pC`=|mkk+* zmv9X(nWXc0vLaHQE2c9nFScDB^3XbW%q__RGY&VMlo z=zR5HrnD${rnoMtnz>txL?Bd#wKXGB^dK$MpsH~63ivvs@}YIuUBGh#pTx8eC7U3iqnj8$cj1O5N#&VRUE3ess~656 zoxQ#4<=wNE&+AP2R81C61nu&*NdyH+m@XYbcKv|~h6qF--DcfAlWw^Z81T($pv@dr zzAGw^r)UaTe}8}%Ru7ze%1|L`Qx zbN-u^sgP{I@vBwhVJvL=);O8iv+h-q`iXrDRlG|8y+60Pm`b*u&>=|J)RS) z80ck0$PX1)8Mdll??w-Gyt6sMW!Xh;$tREa&>umkg_#@Uh>eVQm}CzwgecQQCDE;0 zb}XZMKYqGi!?H3^3f>_f0o#&a>Mn(h@B#?mwZz-jG`Vi$t!cUQ?yX8=CGSKMYE66O zRI2!o**;Mps*vXyBn zXh0##`@mbhQcq>)lLs>tI2vjoO!@U+R`hcL$_u_Y`+iTq{$A9tgxT(v){-2qO~FUv zqzAmzH4FHzaC^W*6ahEp5wac!rbbbL&|DPRAr>R*LMFDgs?la&D~*~eR=9PHV7xt$ z_E__&^G*)?z8(1TO!YhsR~mo`Kujtu2KzD`nWowB8Cp&zIFTCGL04UCuDUa5+NOud z4Y2OW`czCPc@1tuMu=TF|23e1(3uzF^nRvOXC0PPU=o5Mhnd{2sUZviU;@zp+Ybeb zq0D#DUb($x5z`K^UtU+UT5}Lac|)$`O&e-%SM{!+sfwsFhq1zZ#HkGbVz;=?^oHNj zprNlXMi+SN9mI*c4853ZWbddOubiL4tdwx$KD|8xK4=VJbs#1O*hGc^xXu-(mlyY* zN$7UBlSk+EH$S9pOh11J=gPlnRxY5p*?nGQm#_aXoggfR8X~+L#^sJW!74CSq>b3r z^yX1I=`^!%KTKXuNtdbzw~1(fS5nBCWP+0+{uvP2ed^!eq1P=bZ4)cu7OSv#*9()PDBF}I@ z@dE5e+5?4T6`@rY3#2})zwsY9`i)B$4BF0@X1?3cNuVm@)mlO6+I|{AD#lSSz@>lU zU1pEm`4)QbMLhthGI{(bh{;Lf8dZL9-7;)b<+m%ZqZ_Ba|`&_|2-%GImmpR zEusSM$K7?Z`}pd{S`u6Ok)5&aX|WY2&AA%_H^t)>G5jKYa0Mp7@Z$bvu677geXG`p zxf!PYEEDqrWdgZqnG=hcdp?Iv4XWOrf0*19A;TT7G+nm$HK$p7(io07O^tya!7Hud1^&Kt2FJ_(>2u5FL8L|_ASdWYdG^sE1v`%aQe9k+m2?tgn^>@ zV>A!D4=JR97GZKg0nZ_1Cmoe4Zy?eda`$qt+LQWWksa>6o*50q{X566Bc2+;j&T0p zC|I%4k_C_cllYXx?fHGFpM1ZZvDBP?1H)t^^YBgInNsVP)Fdo^E)yuvD)3^`;L`eP z_}5kjxpJ3$o3G9AeZ&w(Q{!`7&W{LQba?Or%z#~mqV&u`=@dsnSvS%~3q8V%Nv9LE zVz1?qgLW?*SBc&eUQAb^RWH9U^~3*F6FLOC3F7f* zhcLdW6&EzwGGL%|yI$A*q6Y^fU#YQI%6Ff+{q(V9&XP2l`~yi}0b+nw1NS`~+xcF_ zo@NN#GY!v-Yi+!XD4{8a|D+Gg9Z|ZcZJ&Rj$;apT4Mfgfr*m=NF5g%32UEBx>$&r6 zBXnQuZme@d6Ai)C@{jiZ z$I_oI!TCe4*Q;T9AVY_1e_+Wdnnu^|DZPpF8s5j%_q?k5ZbG#nEIKUvf;bw-ibY%@ zYB@sM40GRG{wZ&=yFK<%`6k8M->E<;>^P6r$8+5!(7u!yjYKuK-@q``eXK( zhL)eAr#m~+toxHMuZJlMEw95}@@xco`CB-)T}00U#>MuT8r`SyzY32V#-kO&73ani z)tM4h0K$yPDqzbx_Prji&(~{eYD&ZK+eq^lPw-FJp4-8p_w%3#M{LUfZ1U{i{JVcd zb85i;;74s)Ykk$rh8OAw}P_qUHYA^#uvdmp9W?f{I=VuNo?o0Nj&DE zOk*|N77a3;*SyOE3>Zo`dk(w&g}QB~AlEAy;CZINyI|eviC{$C$v-e%(d>rfqi43x z-SVHd{m>2bEakavh6OCDK)`fNS{?*PplDKuHhNgryZWZ76h?pHo?i5Y$~(OW@0tI4 z8J2eRN$~J{U3x6*sktzF6wkLA6bB}Cg?-3cg8q)b!=?geNjke+j)m?k&UEOnD$yyh ztRkoIne(y=Fr;bFfKwh^UZSBpDUvyJz#aYB(($)avb%CA8lgezdXqU#iY02h`S_dE z7II-R6JNPp-=k+h&~X6r1^gN$IRuP7?6D(jiJ^Kt+iE-{zp-&Ngm8oRmp6?03ii<` z-CN6`@5p_H^>{N#-ruPoMA+m?-&c*27PRjrf#*bz>(pMV zEnk!2&BwvO5k7;LXMa{OZ(5>mylOqYj89miiLQeQE}R3DM;94=U+;Un+HbG0(A?FKZoXtVVs=n| z@BzjPARab=kwQm|tRUyUFVB}OUHsBq{g{lAQK3KG_=Ldf#P6}0QnFl`ZGBlc5>U?w zUr{g2kbdS^W3!v*?3p18F5s7WHHFV~5Xd?{0FnjcNP+K=8!go)&+W|3+(?fX&%eq= zvG)HujH>9f8x2ehL4+o~;1;npmncSmKvJCtxqtXm8Er)BSpk#E-Qr+wvOpbj@GbYh zGZb=$o~dJPQr;%4u6f%0M&362=sB6BE&nsK%w*#gEuG2hTsgs++Zw^PASL9TsNebQ@MIii4r?^Wyg|cjunG?^tui za4<5LgR-GTcMQDn=;1rW9c5wfAKE)DXnEGBb%)s^$60lCju%~v z9I}8vgzrI+CW8`ZC1|I{yRndp)`}rTIIdNB!A(uN!EQBVBU|ZSSAKWbb=y64eHn)j z9}Yi=AeDj&;G!0E06ieb=67Hm#-4Jr2SzGuf(h7;dKSZlS@ zXNV6a1$5=5ES;)rZOH1O^W@27W}Po}{P^?u&b^!53J)Z1e@9N>*!)clA@ppc!6H?P zaw^rocKUaAxYdO#J};-A2=#qC(w`qyZ#`c0X=iousV-06CkmUtw-~oD?ajoRq!mRq zlr0Na+zjc`cF$5)jDzl>#9Plje$%_JwBG6E5%$58CwK8F3A%oY#HQJeW$iS>Z@=7(o3T6A#o$EBfr zkxhef0ApL8p?UP3&AIAae%C zZD64XnAebc&_9Rt(&y4fsDctY)^dsV2UZ9PciuYCQ>q*J=F@~&*}y*T4ae^(oP^PF zjw1tB^1wmc~tP%o!|s>ng#F-(wsx0!VX$3djQDrt#J*~`pD0V z4mCDlJBQx)F@IN)==Z5W-ae*c3)Cvqhb5tB79jFgqVKTeFO<-zRN+F4Im=Hy$z$KO`?8i z)zQYUe$5%ij|2JRIQGil6V=%ve%VnECC315iFQGNW+SKL$r`yUK47`3pgs&)!){@H zgLrp4V71T8T<6xloT7dEJz$61dkKSTS^k^Opv;1EFyYh$Tnrf3LY(>>a>^V?@8^_E zz7nx0^yE*0!E;rzvj+lR@t5t(UoZRo|H?n+E5+$)&hSKFfU@~@G$Giz8S>I8tT$+70>cW^?gB=k1X82z z9r=Uwq9y0FHg^_%V?FjFPklM=&?oybra(83e|{BJ2X3As$h~6mpu2H!7g+Y4E!^9{ zWGo?9N+hh#*Fzo{xe4`(l)wazde5hreb zI9v)atOfj^`Bsx1B*N$&dg}%A_{zL=uINR6M5P z)=rd^>Ym+^^6Ar(O(wI?ZA2+d=c=21+I?HzzK0gVp4;}Pv=4y*7Wy9-I{0^{oBu)) z1%nG&5?{Z!9^|k;Xrz89>(X?$nEinc)$%^w!?0-ooiy}V6V3uZ_$O@-5#5&Bf+ngLW z9U-OZ!;pBMJ@+Ll-Tc!JTBF?+o+|l>{+k6mJ9S9o8BCDM#HoV8MbOc>CrPZwG9@c0 z4`0xwN9~H3iqsWSike>F9owts?yu$f3C%9-Lza73;27X$g4dm4e*o-#EgsEeDxz&@ zCOnROSRX6rZ(H{>cV%?p68fjhld(k%>hEHIX|#f*BSwiUq4(bqXcpp;*5|GI5#E zI+9Cg;yyx^v}pjJ*WvF5zV;qLnfJ<5fvO+tk-dq5?)@$8LuN@Q_~#Ke_yDb`ckRR) zFt1FmPT$_BL3;dX_LqlSWRUna-M02=R4Q7o>)NRuf5I7|#QzI7=|krop|D~$2V;o! zl}{p*Qp`0k5`o2B^+SDMl`;JTGiqEPScFwFmX}{+M4qGi(F>XFDH zz~DuXxFddKm062?g#ndtuYtxkvlAe)4Jhn*SH%!-zbNB zd+uCjW4mRvHOE%kNeeLxzYxlpxh*h}R_VHxr@|D;CWsW@!rvI#Ok+3NS%6JAi)r@? zDO=ZHgZ8OpELb!8yvmliXdh}Z^T^68!M!D+-g>--nZH5cy_K|LLn7u<5KK`AG9hUU z0YPu1#_`$NQbAk2p9iM=@e{E%oQgAGj_l00yv4kji&faVDNLwt|B*@$)~>YV~h3 zqaW`+`Z4+*cO*ZVeGo(gRs3H<PfI);&y3Q0Kdze3Pn5l%Y@w#XM{J?tQeN+I%%g zO7C`jm3cX}%H_@s)8AJUVg;1XTJh6)RWsuQ-S+XBtiXFRV7vff;11GJ-^Ydy+({Pp7%$xG_ zt^TIC-wT4CMXB;eNWfR8UIH!RYPu+t9mLSYwBKOw^plkz5mms_oR5w$`|P21`1{QL zJj$08uEj4n9UkJe^S}cnMqZAw$6t0jT$ug_vZM^g7pOD^OE8Ulvc~V+H#{nHU!%+C z2!jeU3$K0^V6Fk~xIUir0wEO0mYfa;uqv_pl$>Ki>zBfYja%+k1mZ6{x708hvNiul z2SFiNR9FWt6(Q7W5H9QhJ*MIak;r=+LQbLvxL3YX$$DZ%@>Yiw&F-1Tx3_b<+PufK zsDZpOm$^Hh{x_D|LaP*d8hBuiav5#8K%!@s(WKh(qL<_nvlPgBN*ZIl9Ep7iuq(LI zXfqIh!Ft3ku?H}Xt7SpdAaZ&>Qww^_(_TMnDoQ?nxN>0liO$8oz3wabF}LQtxIO(n zm3BF=@gJ(#dpkzB()d5dut~#Sypg5ud7H~4f?>`6g0E{@&L_Pv-;W-`Na-Q@7A;dc z4JyN592kx9=}Sc5bt%x9^9)DcqQ{=d_xrIuEvfYPtY*9yZX&2HfW}MDhT2P^$e9PM zREMk}ooyEWZP1O!Suu@}O`2bNZ3c;7oGF$o@of0jnHDoxzJxPi~?&9+>4+ znz_gwjck7PHg3Z60E+xe^uyg?_wUO&pzr@JTscT~iamsuvL)U&k8BDo z=G(=g-?>E)@4SrM!6);>Q!$0V_Gbhp8D#S%U_YJ1fGQz5s8t3mx`z2~4{$fHR*f~6 zbz^sqoq{`e za6a41UlC&j;~$qlDA9HHn0tAw%}EaXk;1ZA_9VElSa;O>Ek&eVE{@S-ZVlRsL3d+!>U1cwPx7{m< zeugx+PT7Z9zNI@~n%Svvw(j|>K|z`WwXe-`arq2na|(K_)fFV>mKEO#n!TeQu?E34 zRkeGa3-}Yg_k6o1id6F>QcnP_3*^d+Vh<+BjN*DR?MGeP&;kv2zs`oMbQ%s>`FQyi zYkUvTp6q~eEnG3#gBrYln_|q{vo>ip=h`L7$^vwE>_h<^EZlAvjppheoVYJ$>UbNA zSZ}cfS30o^5$>{+%X8;%?s(gvmG>Zt?*-R+jk1L9ph8Qb+xDz{6mR7Yq_9DGhl8lU zHf7E2RqyxOOhEDhwyFAN2UNIGpgQFKt9u=HkUazJ(*~spz^00j9Qt0kZEGu`LF-v(?#_0$a=XNKJ_A5E?X7Hu&n?lyEAHUxx=^xV4>n~yEBYn5 z&!stQ>BPt?-ypLAHlqW7_tr&^E#1QbFWnBfE9+Hq$KcM8CAapgo6~JYPWpR*Ckl*X zdXJ-24GHra%M?S>&w%$72Jy9kN)Kh|X5Uvn>hGaET4P@J^6-!0PZqy#KRF)jD0cs? z60WeNWVIeJKVe*8`~0gQo`3EC`Ik)AC5#F_l=y(U2~nsxdKeV+0%R^DG%g1!F>xc+ zJMRz71q^Q6$YylD2r~mCRi16!pLtz12oGc)l?(K0OyQ02q+%k~qgoIJr?s;(F_#^< zv1v$|h8_Or146Mt% zwRq9zP!47K%~+Hc$6)w@k&yeR=k_ox?D%h*-mBq!I=RlPzn%RoJk0uY$h%XDmLr#I z>S|-J)YsJ|c&6+ey>dnzCSmbo2hSzzGhb}x?OU~({)G3O@}~fh>?*{tU=QOtK!BI(P+F+Y7q!_mR zm@KRS#YLDl(FphS#xptwj3Wx{rpk4P-5z_M!M^Dt!O$hlqfiA&`0lJ+GcC7kP1e_3 zVOTvkw5P1nPjuQMer6_ycM16v0P4cbJ)I0a$h8|w4WcVRNV0Iote&^_bXk)6JodDX zPoU({M7uV#lflc1#n~Ph>JnRs^*18k7?hc20wz@4NA|?UKd{tlF(FS9TmV@+H4)!y zrO6$v<>)lK*MWCHkP&|gLXFE*nnHh=g488IbAna^D&k^)XRd<}Q#EGohqe1Gm4+y6DAHM+w9 z6Kf-}hN}%?%K3nZvdF?6K<;Lqq4vC>`F9Q59(ePG84>w?RG9K^Z9Ao$m~hthcE@<{9xVf}HHiE9O4sA~k^QribSn5oZy z1qja#rkI|hsX3)L+62vM?64HavcBYf|LF3)d6l>Ey3px&tgt;DU)4htfdu?`5=65^e68jTe6mcbJH&YBMBLz*bQCiLe1%V-?|j)IvRT9FWv_1*e`$-$QX? zy)~C2Ed7B=MAu7Hp`VhG0{Sm|$bk{6Adw!aoQ;xL9F@D~*h;%(nUiE0*vt9KnzPkL=Vb7xSbnA*X~Fu}k^nn7Vnw61oc*~lVJ zkUfd1#ipOAj>r2taQNMTrW(26&g>~M|=yn-z45BTqP^URQT9MdFZJlD2YUs_ws=xV8^20YB^E@1>^n$sK^th+sti2? zr`q7dWVaH??-NHdN=<(#r<_hI?Lj<^H0(q`#z8=_mfZJd18CNuO{wyo^0h)y zof^i+q6g}13GCm6`Zl@l=x!V}ICoh{_9y7WuB}9rqC%aZAhK7M4naWg&y z{_E4C+v`EAUrj4(3|G`H*7HF+VC7n|`MN3C7`%k1mzzVX2iM5yQymolQ$e?1yC2)R zhwoaZ#qT{g{ps&WKyMXKhp|eoBs#?78vejyHvN-hCZG2LJ?tR7oLgMBaJSl6i~Srj z2`=8E7Cg5m6^=PhQ%EmA>e*a78nFLe<6Gx;7aPGU;i3VKezX9)6GOeZ)s2{N?l!=H z{pX8Irh~SIWdFGkt$VU!ADeq{LHD)QUG2p`omIH<)atv*fzTez&A(VpHOXPcmpV00 zCM>VebWhOCQzK+^nrrekxH55?*FPy7;tjd^%KXVV0S9JS{DD>E`EHb{MqN4s z1yx)M^lG7+tz`#RXYe&VFg3wUTAwx+9cw)}3^Lohpe%4jfXR}s)anc*Q1zKow5Qz- z?&?{$O~}9*R0NiVWd?hh)QfGWop4iBtAFxY@+6Ds@-fbqdB~qONJoR}pbDiw_ujWv z0;IVJFSuy9aJTFYqW70Iy`3c?Dd&uW40?VX=&s>KC@;z_B8%a*gG}*Ux&o^+;PBgCu2|=9{6mSjFyWl*B;o=#s+I^DM>tXP#5JSIq|L{hkaLwcg zqiPja879yHoS_EPX{e0;0GcU;T!htIB)7)hs;Ywjz)UMCLV9JBXY-gt80erDQ}Pdt z&5w^Z&joA3U_F^8;W{7V!+`mXKo7sm&uf8(1icU9v?m%ZP7-F<_P9Q0_YtW9D$Bo% zmR4xlbS9KfhG!HYMwJ=IE6m?U>G%#PN9yR-j~gvEyH452i3eNS6fmtpJX>IY1|S8G2OW{7fYZ&&P6+aR_fgi)f@zK${U z4k(N?pk*->0v&TPK8-CS1Ox=eKPw3?Kc~><@;RON7(9mqO)#N$&IxzI<5~slpfGB7 zTOKlT!$hL{dG@o3(iYLk`{l4+)2s-U7y3glxfhSmv#BLIdw(wd{%reG_tIb&J`K*_ zy6cB?cm;X9Yk;d``x5`f0IM$9E`yuTVnnC@;jdRiN1E7tBh)N5hMx%^McL6j zFQup0_>*Qd#-=XtQd5&VDgas}?nj~zz|d)S%AYJ1p1%!B!N|3WqeQ680u890rf5m+ z$=RK)TvY9sd9k~#7xtl3MLq$2ZVEm2k4Bz&YA6wyTN?Rd0ALckCKpikp4_EPY_Tbc zT^;F18|%5WNB1r?BNLF+8=q2XT3y9tB2#>v-bN$zI7pmoD65NidN}*h{Fdjsx35a( z?U5@97L?0NmY3dtYNCVGH~tCUvziNf+$xRk_28g6^sGtPp19aK7hr6A!{ZEdzpqq( zcEO^k)ns~v>je1Legke1EW%V^pdMy2YHnqmPtdJ!bak)U_WNnWw+tPH&iCjC?0$R= zV!4k4OQ%P2S6t68e!pbkl`r2^L-6Zk)!~DPllF9eDi_L< zRm@gh+%^hDOm6Gxs82n7#(F5=)t?n9*=GIWYv=x$&%fp90A7ms#azWD7{3F+YI%!g zHJAa*8J&%$7f}S7>~#&D$TR2bOSW`-m7{q=fkhOU!rr2Td^^qsa?*;0?`+X3S!&fu zXAhPzHJ~tmar*%6-!@Yx)#j(?UmMJo-k>H`qTW@aCy`WFIHW;uh$dksblatf+-N=% zIWPxsZ)z6kT<*-UT7DX#RL!hw-fiA_`0H1A};9dQZPcFNbb6ZQFh39B+r=34{+R`r?5K;bBqJsa9|MQ=e z&+@&0>y6%YDpo0$V2(e7d;;hbz}-0j+m4jQbjqRIT1A0OZZWOTp`hL;gEaY~M*-C_XaTUMn80dC)2Hhxf(}roxMx4Nb7Rd#rV`#zZ8`$9 zzcodDaLe(wN=%kHeR5TBSdd4w+{vt(&DR93r1k!@`Tpvac93M6(y&YoMB3CF`wtr| zCzAW(wqTjy|Ha;$z(d{l`@@zbBof(8q9|JkS%(VArAZ~bse~jX$~Mhdifm0Nii?pY zDNOdVja`yt$(~t6iD8C}AG2J~xBI!z|8>rF|IhuL>puT;Kj%Ep^?D^!#{7QsozM64 z{Vea#`_2Ep!6)a4ee{>^x@~1qYnv__GfA_EdGil}WYE}-hx4-do`K?;4XH`DeOi9- zf>Q`v#imxYQ|!JGcP|ilSMM$X0`JZ-Yo-0W!s;w&9?_u`jQGO&h=6P``|E*3PL&>N zmHPf+@#yo-RAd&s7Qk}Ol3(^K*fO<2yr{+5fdtin1f9n-!_`bfFfdT`VG5ra|3>h$ zt9JXm+6`49SUysQV~Ubx=a}$?(`{#P9k*55n&m5C^G3}z5f6?^{V;c8ErqvfI#;yl zZN$@oRMQ&$E10grhBJ*Zt&$k;+pgVNjcT>OaZ$FuDST8pv~ohM6mj?X(!ST1x6qWX zzWmt{v1Hn#CpbnRknjbJIY_GPi?FQq_mby{LO)E8_109LP1Bf7ldouNxL%r3pdz(u z(}N#|#UQJQK%%#|1sspU&0APH1C44=C(17kRlDlN>0k4Vyx=u9lX1yueSZ}Ea5ERC zwE?Zqern<>@|QWd1a?6U9yE~})2X{4xZTKptX);l+o%uOQhD>Ulcn2Ux1|~4hY)vw z`iB^63+*l;3cVer2jmh9T=GtYeCGe<{)i~oLhT_P#4SUW66IA_m(hD6aKU*;g+25g zc;Di2m##s#X`-9D{n6N$#ueP=>w2_IYAom<$#OtTt;%vCOofcSr`fC-);f2n6`(Cm zzMc2z-8YF|P^Q;v=~aEQ zis#3}_tt-bzFoS%jqNsC1QJEsHOyi|KGf5!!I>mTnmE(jJL_aUs5PIDx<+n&AUFHS ztxw+wwJOm~J1tid;K}j@FQ0?n9oEp};C+=vrOaZJ1VbGky*+_b@vj1-$|m>LGD=fcG$Rz-PW+=;>opX?LOausLe*#s2 zcDlBX3H+r4M=4Bmj=O%;?vwJbN5{UUXp|0+Bjq|sn+lIMKUqji<&}G*ExU-co7RiP zqE!S;JY2py+e|4}uV(d&#mDOz!N#5?7rv^L90rLw|8DdMuF9d~JOFGEpnb6WCMuVt zG`jCq>pfBAH+g9L{0@Rx)D7kl*9eRBugX3Pc3EvBbCR{W5@BE5Os?g<6gjH8!;>D* zsASi|Nz({OZT*g74Ky@EJxeAe7hnTYXEbcvMSP!+SMXj7M(}4#ymyC(D!r1U9z5l@ zY5-ojQmb8S{21k01TGn^GR8Uyc?CVscS4$D4w&#*9&^?-U`N`s42Prx}442OIG}eTN^$A3uf%i#`gjHO+1e zAdponYHrmGdRG};-i-MQ_t8o4kW`UVee8fDAX?4VqaLAg@PvuyeH6xGK zw^SAeA7H_sd)$sc{a{n0sjElsH?Df&+uJk?IGWm@8U#64FjugAVMvO{9lhVU?h|Q8 zkxnPKg(Yk-;ixuqqHR6@*3{dvO zB)gdGPR@r3vCIdbzK+pDZur#tdZ)ZP0>l9ezD4lNE@GT`PMj4BVG_&Gsl3x4#B@R( zagYbySAAd|Gd1mD1sp*l`22=x0n5+{jQ?^R?cstPdViebZ(Plu%D9%(yM09HkrYR2 zA05v4jf<|e_=c&=3ZNmPe;@@>1_P{PK{qLI?%q>a8RJg)-DagUyYX_e7y)7&IKe9YB&gdTtHrgU|ff?Jt_Snt+dd#iECNq zIdVyo69vP1)?uxw=hC*^yRp>}->ABi*58E(6{OZv7%veqmM-18Q)B{&BqcY(DTE}B z3@xLgU>+yvJrp3^T!hM4v61_FI@9k(h0?6a{k3(;U-J)H_e}y}5RTAWvqOYMJSl{o zdl%1AA<~*qFaSwIGUED`n2yxnxV|-e(;P2mU+`~9mf1+tg7#uq5_Jqhln1!D1pr~l z+|9+Ic-_PNTKbfZi?1E6Nb(Enh8;9wX7PJ25hminNCH?@h@GnZdXS^f8CH<4`Bb)2 z31Ms-BsXIS>oP#Kupsh?3xw%}(SicImnX724Lo?jY9TPKrmTx}LEBp^b9A2s^)I%z zJ|+jk%6gTrM#N=>nASQrqrs6u3kYS2F&S4+1g>e_$;61(fEmsavprY0u4+*IWx4~l zhR^8-GE#0=kP&=9C!7->NJj8No```T4TmRxa`*af5iK{`YsK??liQp<9si{oN&5w4 zKvcrwmn$vt;x&aWihpg^pWO7D1mvD#Rpk(H$X-7bPytr zf6!}+m>%QsSZ{3(fogyyEQ7ayiY1?jq!K3ikA5mwI%A2uyI5Xhqa3&tSDWw=w*FWo zf;V_!!B47ekL#74=MPJzNSF&6Zx;;B%|nf^>usE+U?8JdGAv~+yitoO$LWNNqm{l3 zPAd7TpP4WS=uqBD`31&!BvNy63OxZZEMKW8OclllJ7(LCRp-CC<$9|3C|2gxlZ28+k;y!fJJO68(vvXBGN(}7nv`h zB#uP!;y$J>km-ddmc#rTZ%sswPr{iepL)GU!8R|&vR}-8`PrdnGDAv+8l_3_ZRmal z`R0xNjl&tL`26J65zE)Ft-C`K_4{uwAWLw9K)+oQlE?$VIxQNK>=VyYE2h_vwr~D& z{sr{Swb%YN^ViXHBLkI-hMm0i@PnM68}XF~AAPAx1PeleJYsEN2O??WOR{oP9cLh8 z;&b7LlN(hw>Wgn?6bs%isx5)?Y&YPU1;WfdGA2gwMLH97R@%MAC)rPTJgy|#EUn*I z%IR+~RH}*8i54A;jdlwKlfSzF#Q>F5(!7^-1HwIp&K|#mUbP=t+Y+M_l=JfS>?Eu$ zFMZzJDUCQq|-38Z8~ZUWthst`4lXXr?z;B5- zQAZ?{VNc?>2)PC;QR-Nd{ET5X0d>0s^|MNG5dA^s(OOg8(oL!a`zAt$Vq~yvMYs$8_PD zlS71e)ST`Qn6R&c%mDQVe8>E)TK!uU>gQ`Su9pTKkGgO$Kk5ZnH`k6wg}T#G+AaBi z4DRxOj9}&u5`g`~TrqY&6G4ewxBPTFr`2i8F^ziRlzxe@nBHH8LYTK+HxKIxRmTX0 zJ@m@?azjFVhwZND>DSnaOhyeP#FRlP4V&q420Tn_XiKq|8qWL0z0VAm*mogDd(}?u z0$DurSFRJR^g7&6a^cvA=)IkSn5(Eg?1v#`E*5HCV-0l4hm$AzVOO%Fv^n`xxu|Zp zL;UW{vp`e|^a=wZHOWpUVGQKM^TkAG=;rwosGIbx?~-+iYMv|FaRJ+RDqJ7i)%UB_ z3vO>fY$Xs6oO2R82ZR}Bg6C_lLz*7$0RzKcRiafb=9szbPQk=Pj&q-lWARVbgD;kszflS%bw z265iQUVUx91*J1>QKu_t)Fgg8vk@y|=<~M)`;D(iYU%EPq&!N^=VHnu-j@k>*Iahk z)K{Y*Ak|`)q$9(U4a5|opkiVl9n{gvdyzcQshe3Rij=;%mORv%Ogpi0XywK`_DB_X*z&l4P{du-SH&aPI1&3$CN@s&!p+bKREkY%lEG4oAb7vGfn!kGCFS zf=&#mG;gs60!SCOf|1tIleE>aNy_%`X5%xKXnUYOC>GFq$jCEvWLd#I-sj z;@g%XrNG6TFU+w12IEW32gBa=9oJuhQMM7k(Jr7`Y?SY}ODBiy2|(uon&A9qK6Ial zFpk3CjZbnBxf2rUG(NrRI-EA8&;L0E(WjRCHu~`ouF<*CE-Yjy+X0hpv@vMu{|RUU zyf8@$2OHAg*3fr9E$rWo^UF0M?%`I_#3Xu0pSzHIqy!Cume#saqKpXw){TMGLaA^H zEGhR;eR_XA?eqOIlL-F`6*I@&lkdLs%U*KYY5lGs5IKndsh`>!IKJAZ&*6dOX#iKh zYk+!gTQ!^C_Wt;hrAe;o>C$nK;)$W1eK9mkBy?buC608j230bNNygxC2thcZ`6EGy zC2~FP$cNy%FL1BBQ5&Ima_B7gg=N?nlOy2F+>UNPY79;t0769#s^5H`4;~3v6-!Ru zwBpBao_iC*U;pfr*`psf#dm!164i<9lh)Glx{WmHQmg0X``|q@QfE3_8Ioxq) zNzFr^3F;oLrWmil$yH!MvFJgfP}$m}I!76g3fNMOuy*!K*D2M{3Xyo#G1-TsZ$L7g zEQ4SCSpbXzRwmjLL?MpzcGkfZ%y`D8rI#A{UL0LP`i>MO+G(Ssz0y*JGM$9j8YnYb>`D?$zF!b9OCI7e4rSsl{;PtVncPl;hN3 z_gwkZXeUUO5s)Hs(UBS922Uw(G@0h6uP8F%_ffI|wd-Owzs9{uxK-A}5A`?)wLIYB z5;Wb_jicGXq5X)Cx{R-#$X9jkaHC7Vq=Q~uT}5iblE){{OSe~X5v#XDIl>FzwyeXV zPk{2q$%>k(*S~R@kK$*#$zQV|s+j?%^#+ENxcjR$v__f%2)msFU(65*rOIh`R&AXK zVw!)MiLrUluF?wpXR*6;&5P_fIH?Nr%8|OlsF!RLNz$H2tJY)HNCb(!W8Yn02{;W` z*w6uESPt|xk$2j}wwpln12iXuu3Ol^Jjk^xSMo_!?dmbQXO##fJoroe3O=*>5T-RJ zs_N9jUyQ);!Eyhd4`3UBOZpp zXWL;&1s<}GvRiWvpkT_08N<_ApE?WRwfo|ES351qL})LdeKH^g5UDU|pFtbuUPWi@ zS0wcEgBF%?X+-1k*Kxniqxn&DnP+KV0V4%{fr4a-wa;L&&06Sj9L*LqkWfkkEIV?_ zF6*JiW-&AywqUcrh&F8AjQJ)iIE6 zHLsO}F+k&RdSC~xcC<}aBbadpdOAY{?N^Zt>7DDgv~|mxCObq&>z9~!(?NRg6+P<^ zIwV}ShO%FE%zIOL45@Hhn-@k~#7#ufY3HA80HDdK7(^<-MFRJ8AQTI~h8Mm<;KN+P zj%ChRu6N?OQ0HiM38P2pxc0~-HRA&s<>*fVT4QgYjToYuWsXLbQ)Ica63q5tTE}5x zN}u^#wC->OY|Iw)x^kKP<-%*dj?alKTOeeBgAVw#!zKbr%Zp0%{LwzPp(ecvADeu!;FM*-Y1t)01s|9BNHHd~UA! zKI9Ben4`e%9Z`yh`pL^wp+$3N1|FApI>-YTqLh7@PO{GXJ5cgLp}X&2bm}6UShq52c>Cp`>YB131ZNZhO9eO;QWmmmb^1y! zFMXs44CL;N$(TS!Edu5XtKBH9t$v2KX+!8hsddv*;M7 z4pMR(XqJ2F<)ON|Zzpu#Eq>k--gOHt2>fC_Qj~B74*ScP+6BN#(1f8B zCpcTw(#q3IY$a0C($l4%GzV%8s)|Vj7(mPrL2N^cpkPJcj2sFtYJ~Jj zBJK}P^U3pAioF2IvTJM+cW@FUN#iArE`BDdA@34+nx(Ex@1p%=<+L6;>?$6k)BUpc zqtKhBMcZ9H%A?*p&ASd<`L7jL_&*Ix{fFSBV*F?Ad_}di61n!>I{xlHgntBVu;^zT zH4Z>>6|~_6jMz=>mr(ney8gQ;foIJNeKtKU-V3uw4K05Uz4Vx{^PR(kvg)e+~1~{OO!t49!k! zg2)l#rH3Nejc`)==A)%l((8bXnHD0S#%#vt-IKFks|6RfiyHJcsPt$1ex5D=h;cp4 zHXtvz5_sC5&1Y;P<4p|m#_o|5PP&ZGGb--K=36D2MG0;YID)d^R>FT;2I*H1Stqjt z(3={(q`M01Bugmoefrh2ziu~)JyMsPp0eU;LfvkqM(TqxLcm*j1=azzwv83YJP)NY zNj{6qp{RWoH2o~R1f#-RcF*IY_LCjl2}|PE4Ra?>Qev>q>`HJEk~^{sF1DYY1|Fwk z1eVZ;SjXR3-DTPOT^jT?3jXRF*6uGUE8;Aj!QK6SW{(2^>5`&66rmjYUgBR!X`5J{ z^-#l1$wNEX9pzyKyMwLSk8GTI5^8zAK3Ae3E z{F9k_Sl8k6$3I66yd|}A%ipqQBxD1>h%N6HvZHFBv7D`ji({nIQ!FQ0BT5yqF+Xzp?DPB z&Q-6*C7{bT-zeZw%+4q!`_UgsZ5-K;D5a+pkWbsD#Ix*YV{H$d<{A1g ze-i$-)?R7Xw%*5=L5Y8vjNhs53Gw{)5Aurge$OEm_jul9nlI; zoK~QF(5BBb!$d#oWJ$p zi?r7bqsv^}%2B!=TKTMSu!q`E79Zm~0);r81Ob-XvlGq!#j%ay^u230H$1=dW*Zy& zUrddbI{$!bV)lV(2KIsp%a`ec9tD`={llPcP2f6=*c#SP>TfSuBo zKpJ~Mp%g&Rn_+wAXwu|?88i3U0&StW;?JIOyfNtb^B1FBnsK32)0wH=q z;bQ2_!fCoedw;_D`q1%^n>?#d1Iy%G*=%;on)~`>^o!?}IK|(NzGWH{zKqj6MOn34Fl4#>}Gp)0uEa0C=eOy?%QYcFH}B zB5y%L1}pbogw0W&Apkn7b(A3Qh_0a^?*a%73AcPKa9?b-`KDCoB*-D$;WE4r*_R{Z z>~jF4!dW_&u?Q#wd0#fEokW%5M{;lQ2b$Q;a1uppvJ$?C9G7KC4mR(&Vo=G|tVJ*Mq>5UzNrAN`a;udi zRzL8Au<>QaUUnM6OO?ox9%Tg4u(5M#rs-g|R?hzONy6uFnpe=8r^kzuxBD(nlsWZr&jbe7qks711s6z^mmLG3fQpz&A&bMVM+4cA~ zWuL&${Z}LM&VsJ^TFDh4CH@CXe*Ybw|3}M?yQn8R?kYz(=AZ|_ga)h?Ij&3geo9Yb z#6s~G$pn%4g#<0lv8A-5Iu({uy}f{UZI627b{=#7pt6Mt>-g&NwWi23F(0}HKhHHb z)<5|A1)6~(;a3}R^Zh~!Y>D-qVtz*DaqS1*Q{_|g-Cp0rX-+7cbSK8{^-R0^V?Ea)N(komc4yXJ1|I$SDEI<(7BOvolFlU zK`SNWQ{!JukDo+6zfWEKqH5Md|H)wjeEt`3MMac$Ejy8L5g9iC0bq}-o5mAByji%q zF43YeN;*N8WZT@@;x zd>-3@l+s&(?^WBw3cUAL&?QD-VrA=-S0|5e+uBI%gjW=I2Trt1aHe3i>o}^jNV`aU z>U09jn1&v;chz5sR}<_={^fX-m8eY!Da}Q{%b;2WkPdY%r_}C318_nhd{Ipyo!q3y z=|C%fcQJCE7+ad5ax1f^u31T$_Sz^N=t&CgHSa$SDS`Ji1QrD(F|B(JM`9*n96?v6 zxo@*ERCJ@yaJ11&*JEpwQZT?MPuMj3y@`e|OdXD0MlZcVNMTwp5lH!fPXH7KkgUDn zRrJvB>o&FG{J${wKc)M1GVHrviD^Gt<#nAD)j67e`1RM`W1XRxN+1&dgxnX}{SiF? zLZRjR6UE(d`4cbIH%%tMTo^uXGiFRF}Qa-R#*NQewO^EpVC(Ac1XXqD<+2EjWqBT{Am&BY2=#a$Mof z{h7s?qA>@Qa4H2>i=~GwS|u1N;~$rL4d6uwCp+Q~!Z&|#(~<3nZRn5iIwIAMjF||Y z32gi9NA7RK#GOGz(uiP2{m!vczxwyDs^Jr(fQ-WXLB~qzp$Z*u@8G(pOe^EV_qAoH zf@a@ekOV`=XFe=#e7I`S(#fSKS{!lnzd5$^KZ+Oqk77WNeB?gi?6@p*8Enx4TkVM# zI`7|e@enjXw(|^5Lyn;gW`wZgl0f&_(`N>oaK=WX89&2+*0f7qx|@p>I-5k z#@qICRaJTVbKTYjM;iEwd4ui)c_3zM0xGx!CmJ?%2Fp1>;Qlr$I9gX~EoPIvA^Mf$ za;W#lPWL$eIqSrqur@}{usY^vc6%M!rA@-bXH0vIwiBXjD-RoF-MP%6?yWk%nkqZ{ z!92@i^~8XpFjDm!yav&RAIEjqA*HK@Jw_Q(?82L}C7Nr9LW%$a-D3u%i z8PF#5cQh#4y_{9sF7p-&jHItK79lpMUVA(uiHLEAVNzPpRuxt~RtaS3kDh1rx87NF zOo`x2&8s^Vy-(P}_wem(I_Q@0GB0w7ND+}}Fn}ll4w%bW08XpD)D{TyTep4I6MD6P zZC7O-cWXj0$#+-9Qa{^_mNkT2~lxi*P>du!*d$@vJ4 z7z-)tYbtNB!s58MsZE|^#YP#^1iW-ZvnH5K-2t1h1(`eZ7~uJ6ber^m2~J#1D}Uni zVfnz9Bi`l_*4q*fMZU%lA@2f&zZ^<}W}4ZKxHHW&$kHRTbxgC+sRiRjO^r$=ZF_i? zR}g1TA@m0A&CBKaU|0gxmW2`bFA^UIU# z3<~=h_}t+su&T@!WD<>K3PsY+|HgH!cDU?Xw)fNWX+){BOX0M)Nnc5jky*#V?7gQ$ zCG|EH!72HiUdIw|x1;%_n8{#TZ5u`SCrp%H97an(DL$spu^kx)-Ep79iT4X#28sH1cagVypLsWD-t zA6I|!`ga4ZnZUMa&4#c2%-?uOTGkOUm36u-M?i;us}Iz%eU2?d9DpsA$@&AS6hjqn z05hm6$!RdzgxhYIZInnt2S>d+;}Qrn}##X&hnq z3}N0pcCovpq~X$=w7t>!&s>VX+9e$;NPeVppD%q`Y_I`JnO*~tjBkUb4&nihQAj6E zSj8>n$hvQyZr9VVI%=AI_d4@(PGQlET(SUxE<<^93(V-En5Pi&poncOZuY%yDdgm$ z5pb&tSR!8$Z2+Cc`64R=gUV7@i5DLC>u=err1LyJ^ob7ECj&S_7&@LK=t#!Rcv9L~ zx<0qVdyt3618ImN?=Qo$4CU6AJW2wOVaG*{FMDuCkB?Eh(ZJs{=nK0X5d>~!5TM4q z66)zK1elZ8@8FtRzg157MLFXOu9sxO_NM!eJSuYQC7n@#S|Nv~P%Zo{$taF+kl32w zbL0fl1Sp`b4`GB?=n~jm!?_=nq4P^6{dIaaSEtVEyh{zleD&*5Afqm!lAx^^maLN_ zP{51>`WaLpgSI$D94hYc1a%_-z|O~mbxd`EYBa%4)soA~W@+Vym*9m1tqlAC6adRg zc$g$;2vJ>Nos?I&sn{0m<3l|`=oj~Q6n0JjG+bORJ|lm=>uC;j`;>19*U(Q3><8Rz z=Kw~bo7bNe@exKU!nLFOh&(L)E;{n*q8be~R9WCW)dh`vp@$6@*3D>pDv)n?dG|&4 z>z_MoxmVW#GYtxA2%CXpi8lY1x`+Q($LWIsS2b({6Ez#mdG{an{U;}CSc)q<5Umvk2z1R!U0|2vbm4X= zyae?x{OUhl3GS0Eoqm@`f2}g(tJ_OXy717`9i@Jstx2+O?PZCv3t+)wv$bF#NWkst zf|4-QjFgUcmU_2xdz1M!u@2vX=l7mG+V@s|Noi%Q_Q|fyy-&pWK{Lh7F8;vKPNUG? z{v|=CgQNJ~bqRZ(JtzFW1XPcvLUuVGx$|Xu;}8u+`IFntgE;%q+K6>Q|k9=H3%mNJa;TrPG8{a=5I>=frIG?l)S7i9Y1)J`1k0 zeNQj)aOe?7I8d=>vCUb)*_`NFsFgYB?!C=wd;E5wWlc!UvI#eV#KL_jT=C4nIkbAPR8So5>*{feDz zd6QSupRvBphpV-Yi zkSO@<4unz0evAty%!E_)A7?#7obQ@2sZ+&$qBg3Pes9!!l4Ktm5F;@ZDW9Pf5Uon- z8fT1xdy<6ebpj3R0kOFz5fyBRhPLo0A?)ZlN^rKpQWKCI8!X*j@Fc{wMZA84L(vKq z$;XsDRdZD$zu@A3kL{{Gqm$`K^y5A)WM{|Xxf;l2K zk)s(l7HZjwPpK!hchi<9KIBycxaAX%; zN6!=cP|b$v2ubyMz#AnU?;PniVDwmk)XP;3*Gy{2+@29 zmo~T32YNuLvs?e7=BKaYzZ=X#D_LKe5G+R=8 zs>Vmw!6$tmzbl-Id8a;q;=?EDk?hewWoXUc*bit0t0;q>tNrtn%(wIx$3U0L1+sY7q&A{?7J*;aC@BT z1gb3=Dj6TV)`6_UB6J`2d3>aP`l;oxbpJyhpF!XGl0xNUk(x5|ZHedEPs0mc^J_A# zKPA_KoX7Ms$fW4A4h+iz)&Xtb?cj8lAHBK9jAJ}u1?Dj(k3uX185<)_ zTai_WL7B~BExG*KOmp)+2o?4V7^!jx@*)ZhY{=_tU=f#>8z7Lu0CG_hcoU@_?=n=1 zOjIe02~WtE(azPh>$xg^KaIyByDGtZ7yie=jx&ODNchi9*6W&L-fOz_8qyys2mrib zSIj>-C~{MV(+_*u-UH?1jDZDjYABhQKIX%+p&Ou8iW}n-BQG2IdD}0qWnDz+aecwRuhqFw z50yX?;mw3pSi}*22Je*V5oe@&d_n(4xOdRA?+pyMkc_K?r3fm&%s8U!F@G)g9d_21 zBV-K9W(QQokYJfOJ{0LGPKaX~uJE-X0^}j%di+jIyU;l07rApb?L~_BaT8u#K0@n+ z4BptfV|qJ)4re3*$u+3nA&d|#_?un}FOAbgE2VJg)wI#9ANvlpPb!8jJ{TPcf8e(F zC69)c(xb%pwpfgc!VMuHw#5EV+~;*7VM9YmZ4oEp!6Q_(MWX+KW^DW!}z9eA_B zE%z*2%L5SZ<(06XgIw>C{RnxWmqbKE1-s1Y&&jN5i4gz&V-c->Dn7L$XOFbd>+_0i z-{;pGiJklk3tgX8WW=*yb}81dVC)KxqIdj5hflvBwE8IO)>Ydf)-@Ev;Z#Ct>O zQ4IV6h%{h(jx42k?GPByOtJI2xi|#NspN97_bj--OddXDy+4S_pWXsc^~ZiH8NK{z zsJivwfSvu;`9kSY4uMGX#Lk5?!r0A71(qfN!Xqya5X^Lrm!x>q>@KVevlbb@(9%s? zn8-~vy@$+LUsjvNr0cn$7Z%(!=`A*x_^g3&X3x?E z?@ixts??B0!+qK1mHC7|eTt#j_ZZ*7nupcr16oQ(o=cEMq%)si(N9xN;wP#YH&2_2 z3I6b~mY-O9IZ@SAVzTCK!O;_lzKpu&iAwDy5jxM2LVuLP&e5Y4O7i{~_W?z8h{T_3;x{ls_QmwniG04!`B&;{{;z)+)4QqjP9;?zie>|P=P1gY#3IR{KuPj#bBQ)Ei5EE{O;j5ONdP+7E;LRLaT z+Un`#KUo`{$}b116!_#Xyq^$SaOvbgI8!JHD(rlhV=RrEqE(ZrZ>2+#{rJ1g*4ZQc zdbdMB;gX}k{9gA$qi-j_kR-x~tU(&T7o^2WCEyWAA9NyQqhwG)qbke+%bea zFh;Ca*Ip-&N9r~S_*Lxh^E1MyrcSXsVXAl z2pE~FBF#4^DIXcpvq`v`T*$z z9yr?TUlAvx_)T&GB$69Vh_7B9h?o;-i?Vp)E0dFU5MB;@@t$=9HEYHJ^>C(mLt`)St>vme@5WTZ%~I&I7mcab z{u^D?h3utA!O~lxq`aVSPkN5(jf9a(tQ9_5g)qKinO{*?VO>ZM+eL{Es2~_x~Kk&4AhE zLGbg5;O8eJ!1O6~lg3Rl@41Vxh|T!*?Pn|1o3ZGqm7=}bvq zMUwNt@G`v}E&*p;1oSDxN33NE&NS1G4E`*?B@$)?e;b$c1)`Gi`b~!?2uFM*u#Nbo zc@GNE)~ZPcn{~&Le6+`n-DZ>35$67MyM~6IoKpQpjeMIK3k!=CX)3fC?D6A+&i&yC zEj4oRI}mOwJ2GYuLgy1wrIlr^K2H4-p7rLmS=LD4gBS3<@;69^=iXX<6afL2lmVx? z;r5_qm_3>G)KHopu>TU)NDWQ&%yG@hK%Q%V2qTr35kr^UeAiyAwxR z)&ZiKPSBZE#%_QceEdyvXwNEfrdXj3d-YwoL1mzLR=C3@wUvG)dHE$s8stF#^mlWx zb+LeC*t-K!j(N%Cw( zyDZLfU2tIYyaXFZm57bN)%byUCq~34ukuZBuxDJpKjM3#&v zWAY!@#Sa=MsSMK&TO&RJ><2B8#c|PEi)fRvj!~Y_{W-Zm^X>JVUwg@kT-Xpz&{+ec zIDTJ*72WCk4n!23`2@bZiuR!@r9Dash?FUP?`Y5Sc`o}M7l)oTawcZuS@!g|l|Di; z<^_r>OUHmYMAAJ3w3!A*l2(3ZDc8%)@pA@EdhhkG+8)2;bxLiBQga=7i9McsUxWd_ zi<7Vtq&@HTE*^m0<>qT$UiO{m=*4#8&wIhqcqJD0VjdN4MEW77^r|0%M#XTjgq^BXe#!<;3KVYtztYwaK^ z{s02NkeL4)gE5zWw@W90A|~l?C%0)Bh1IVp8_Ahu@QXLP2;C9cy;s;vJAPyu#LpP( zrt0>)dR<4DM&H(kUxhULhZ*~dG2$Qr1a?cKAxDr69=eO*E}?;Tj!>u=OR|8$WMjJa zkr%vg@BVX!c_T+^%)+5zWXxH7GblGCP{1ZggXh2x0qlfB!Emw+l#WAp31?*Vi+FAv_ z-Hu({4P${EPZ9=+N68`O5PB28u_6Hfpl7l5!Ko{gh_e0KuU-a)8ugIsuc>e`TZ_S2 zE8zHRtWb8HVYdut0@qyJuMEXM)&{(ucZR{m5J#hwin>+D1Fz?m2jwNTsT`><8P==s z|C)1@t$0vkFrzAuC#xlsKZcGXi>=Kw&vT|Qt>}J5!Y)*=97mBe%n_o2wTpFhYeQVG zr9pX2U47(Bsn+l;-aN%?5hpT^x@j-%71E7iYy&>bYd)MQxOhD-6b)!m)2ZRK=L632 zKu~Q+TU)6|4f!GBEpjAEuOVmAzGS&6Z%bSwUH&5cMX0!c-0e`STt`J@WY-z%09x1* zLzP{Dx&-z;tT2*W*iR5TI;k$%Rz3lB0IH|?2P!N=fA zx*>XI7%Ph!XHR}wpW6q7Yz{Km-H&&15o4h`%RJV67$ll)za>#8c zycUqBh1&qUv?!1b<3Z@uE_@C}N`9oaG+0x%3aj(wGz99u?B%`d`Uqk?b4CgOHHr$- zz&vwc3&164F07eIeGR_-1Kh+(u5&~Y=Iu0CW|v~Z^|on+*6clw0v*};*EaM=mT68J z_QJgPc9mPpu|UL5N=Eg%fJ8J#2(0-S&ZRT#Ish%&I%%@2Ib?44n+E+sHk*FwCe1vo1BA|1f6xl)0&3Mw;Mb3|)MdJI*SD^8pXe2nO34Fxs7CE@zW&WB=O1BE8$J&yP zADEs*%KuQ&{IvFy8k^<*rYqlJAa0jK+ptI~w?mgnZS+PdJ4$9vF&eRb)uQS^P;}>a z{7K#s)De^->LPSrkum#{hVF-pp?W32BA5%yYka*PpV4g=TDroBEWSQ=NBcEXQ+)7> z!U>ho>qeG&Da_Bemng6qCZnJg+!H23hFuO$+%?&X-UWD)UXR7DD@VYa*{{5QxbE4C zm@iw-{$=&{&?@@VL1C^O@EO}G6?-}2mJt64axZ73g8)7-z!7W(9{?Nc)}}#OlDY#p z@h%;@EhVk4E+odbuHkvV;k<$W;u?>m*OU0Kipc^1_lmE_bP5QyE?>G_LDBhaEh{RpSNDqTiUZ9(Vy$(MVm)dA||rANPAvu#xQjl$ZOuk@8sB^ zI_Q5GU_P1T&Y6HppaO`qhD+cEC*Fi-b;Z!^doBITmY^w|{b`3ZCa42@8 zbH*w(C7}Cz8WnX&o@>)gutN#gy564?)eEQDi-AnYOJbsx#&i4IOAP_Ow9`4Y3qLNK zdzpDF<{Z(?*htg}DLh#mDuUVT#ykMFnmxq6D8H}|EaV-uiP?QGC*5=lw9*8`*a(0QUX_Wh`_1{7S8rl2Uiv0@rS6{T@Mfvkr z5Enzq3+kKp?GsuXOO~mRqT6S`L>^1;$9LWn4y5UBUPls8?I^Uh=r`0$)EyM*4o(yy zO;;eS6)9kb7J~b>?S3cY#ocdV_kSHNH*(-p9w~Sv8+57=zkb2R&3VHp;m9}HgIh5m zK10}#!9Hae&otRQoh8pS1YnFpK!Z%=>aU0&ChMt)eyUdr8#^}E`n!7ABma}mNR-0- zm*W9JdygKozyRuhSd#OMbeaqG&k}526^ZXCUR$DpsMB{g^d_ob*>+Hp_}Smj;nHw{ znWjLMic85kjRWTt6QI}+_5I5xCm!;9^N)UHIkS_%Lfbm97wo}h8gM3vG#9uEXK?QG z&EQNEl=!N@JiSFWugkK!LeFffh;g9y@OeiCtMc|;LZrk}wg9+=#wwF%uUEs&!uubJJm`)d+D(P+@+Y4M+m|#xy+H0oSdJrg5oXgI zxp42C$yUtWmhdvO4a#Vevr^3Qv1W}6cX|%0l1Lv70%#R~IpCY~@gN&4I{~|d6^_qz z!lhh`pCVG)?{?{m%r`_-o)@8t@!M&LL9mq+*h;fHlM3^}`5uLYU12%{G0O_z)zyKRHA>ta7B zIIk{8nb&6N@!bmHu@lzAs!icNL*iz{d$0UY{v3+_V4S480pkdP0WeQKxJQ#9Cv0W> z1hk}kAm2_qEo#_*GrPSb0+aTOIgeWGNVrw<*!xD%zLUNh*U7IbD7Ah-KZ{`^Z;7}X%%qB-zREO%$ zG>w-$M*GvJ96+vQc+UBi(Y^FygA3y{&m7P^zN%n-RLwDZgCw-E1DSE-2-bjh{|=mg zgQZ$Pit2&o?m)PHvJRWInpH<2j#d#z=mhm4Y}@g35tS#pB%(9fXK5`+HT2PD6+Ic; z94g>$m}tN&o8UPsg5E(r$s0F*9vU*XF%pxG&vXG>GF<6G-~Y&BW*Nr5?EDW0#$B3( z*RoZFA6>YauiFbFl|pH*pvX{~8$2-v5e&Fi`-RFXL}Ja#8slTm<|(Y7E;;>em`bY8 zlaAHUDDqu)zO9)chsEQBP`zA#Da599sv;!{4Yna0&ZqZxE4NPnb_AoWlv_UImviOy z{XOyzhu`fRmg+k)@7|WguMz2>s<8?O_KC5?%^9Hjlak_ZxlWm4d3*pD3hzLK76xr~ zIIQV&%S;dRtccdusEFKMApv3E`)i^iJ_r3bbRtJ z=1@K!o~w&i%MSf`(tcE!WLd9I+*Vf0bxcA@D10Br$*D$qB1cOl3o%1(`LxiWf8Sj zkHh^f?zOZVXCSYRxY}NGeV_{g6E)fm>>8Rv6ybd8mSn&DRi=>8b(n${vn~Je=SB5d zqH9}%!a+4|;jcFDWpyykc@v)BVXQp?jAa1D^g;DncZrd28xKrTDeLJSqBZGOu1{N% zse4-M!+9hNVdlkp4wdGoe)51C2%_Uj>HT7J=H$pG8higb*kqQn z;kz{PZ0?yP;Ms)4gG78Bb&?d7ZANV2r~a>QzB8=JZCe)wr70-VTU4rqqJV;e#Ig_& zk>ZjLK~d>Vs(=Zi5Cs7d0R<%}A}UR4=%Hf)BV9m9L3;Z_3DG3qI@7h!-R^tNe(t?L z?yvKY=OM{AzA?v~bBuSq@4$nomZ9q;}$RX8y;tM2y{k-r3IeZVk55o zk~^(0*0GFnSWK?YG_m@86WNmfMs?ROQo4;l+L8SZ1ZEsSp77OJ0HV

kU+$E?qZ zKtQTnKNrduY9WrS&WMI^vIe>9Ghg`MiBK-#{`j=dcdf&Rd zKh@#Kr|2KHkdtt?J5XZp?xja|pxTQCjS&PCglCJUu3(UuR(u7VNJZe@1nu|IvGj14 z{q(=4_9Ld;Tzrs`sXN#Et*}OGP zPKIi-v?q6Nk1K!_!X0PYet{EM%bbrSS3*S{OJn2ZP_Rx_@kT@TxrA~<-;(@;eCm80 zuG|O5vIMtD9fpW(aXlD}8%SX)UrWhja<^arw0Gwd4VO?nvW^hCC*@X^sANl<@%Vi4 z=_Q48x^edYC-kgzOKy^CZ+iLA#f3+auuu;j5u44x^*0H4RYyvsF$SRbU)JMljxfhN zo|bdoL72IAT>0=Y8Zt&RHO_+@bememAS}XmgZv10FzOMu13iVFjHsbu>ll5L9G(YH z6b{u2&EM_p}ivV&4#ZoDZ>=tP z+kH|kwd2!%5u$FkY|VX zI?4E?*9Gr6_~TXcJy8Q$Iw_I|lWL6yMnaAtkckOc3jP=|4hy>D+0>=d#3n5!O*Hdt z|H|`+J~xdHTs_#_BYLjbu>4cHX?o_exO?pLCYM1FwPj?;(UPe=+>R1hAVU{HME=ph zjG`^1Xe}MnXnZ^22uWoe@STZh_&TiEro9|^u;^gV2}Jdu*_lI44!~TH#kQQs2*b_f zviqp2^mg}lI`fv_WymN!APBSr{hXa!uO-fOeUQ8>7*snnTrNIhd>u_q+Q9zmk*^+NNVsQ=DG_(ES`p+|u*SDU^+`n5TO`*k5 zyZn*aB?b2>dp#%|K#ce?G%OenX)y|FA=IK2OWq-3teOA$sD)6i-fjPt#PPZ`UCb3b z>k^5n$pFkc=x=DgAaY?nwV3(}upp78`YFGKWdeU{;(@MYk)LqVWHFSdRx-Yf5My)@Gms-(@daoaG~`OFid8JDJom+z=~PN4JsaV_ z{qb`CJ)3sfT_<>={QjNP?*E&>m?!xBJ}Ct-$4X3s@l?hC2wrD7W-{m4pRW7+iVu$~ zPZ|?bLL@R)c-lA_E(Pn#M9C#sl^qH#KYu>wszMhc1YBZmUATad^L==3;Yjd8uhU3< za`c8gBAI5anAT8;730Eu0sMW(C(>e}69wttjuSf{d529!_bgQCYu~u-d2|k)E)576 z4Hm3J(c73HecYBbpggEz;X%u&rJ?R-MGuP{G?E#<6s2)Yd)6?Gji$i`L84NS|c}pwglazX3K<@SH}x_1b#Lm`43x8-Hxl-@wJOfevGvr z9y1c-lo`sr0X;Umwxp;(Am8w7TEY=|?b@TpLXFa&)Gc*)za*)97E7?JO+QT3@IH;= zWi%2V5Kh^5`_?y*_5H>cavsC!OcEqaF}dYWq&z{DkK=BIrr_#a8)*m;@481hg3KpM zQ$$>JXsF1yj_}UCF- z?eX{I6CcT^bS#eb#MC;Ws!q@7f9#skq&3xi4nk{Sx+2Mo8jxTIdQUVrBr&Sp>KI#U%ka>ZHUb zLzc?Yp*-e-t9WDqTk&C6saK9-ENox-Ol?#w`})`ejb6XrbJyFa-XB%xLOGEgDiWxE zSJTD5K6u;5Wk0De#h6YF9Y5wEt*=Q8kx&0oP{J0b15$PZwhFss z5RmJg)zv}9xr+6d7)gSh#LHRtm}VusC&4abqSyJf;)+&=(G}mv?%ZPSUXF-GIcka+ z+#$c0^Vmpv^c!=s9lxt`#PFhgChd)8NmV^|v9P+nW>8`6api|F>(giTkcM{;YIE@} z_}ct<@*}8{_RPZgBCB%#?_!Xx1X98J4gN~(s4!)Fsh`ix2fw`eG!s73ij@BnpX+hc zj?gf0y9J|`CVhNlvCy2x)eQZ?)11N!Gu&_k2LCLW428uD17g8Z06-aqgLx7z9_X5^ zV=9c$8gAHL1Zh9g%W1A*+Px25dn6+ z8wb>v-5Rar_on3`jcZC;o0DbN?hk(tXN!2if*VxQ*9z)TF$;lhIj_4SKf)|*!|bm+ z#LTvd_dB&RgPm}kQ0$rMq<@q}vZ=R@@=raaBw!&X$c#vA(J`83EJh04K(5%!W2O8H zj@;iIJ>k86k&AP-rD0!qeI5^g$l&{sV7Q`vXUn~-+mZl-xn&%qHeiWs;$ds{l}d$3 zl^jNXU!B%`cuc9`P}3s&Yx0>EMb6~#VHy^4EFEiO_<*yaHDZ&BU)%tv6^F_FCL+7D zhqhfi;;kQ_EQqrGCPDp7awQnlsylN<9Jdp=5G;67wQ6^?C6(CV^9z;>XUk5p4tSv}>~I(G+C8Z-xwnQ`_=@Az@_GI&qO~wOGWs?!M1q3Q3pC|;2Vp~v# zW|J%1m@+N%3!ya5vH8jKHC5GF_7;rihX@aRty0fC$g8G(Rqssi#@(e2ne+^w;3K6GAK5Dytlq8GP;?V|(PvBga&=(dGt9;Y*=dm7v_<~voj-0nC^PF=n5-6Y7n zEcZH}$aa#tHcvRpzi))+X5;Vaai47eUm{uA*;Uue>FE zWBD}1cif#Z_mt4uhu+AxoN$jxTf~p{!+Zpm0%ibTIZb2*z+dRzIm!bXY^|NunN$ar zU32fUJ)Q3<^6{jJbcGGM-R4#sq?6cEpUtCnn2R?f~A11(UgzPVCou#sxa5eF?=Vp_4#^`NaZ$m!XWb ztEZ8oj~*-Jx)X22pSo9cv+bX~ZFy(OeV7UArESr_AmChlwbD#Xv>59q0RPD?h`4vm zR4t9GXdbfq!;GCW->%Jlaqf}bdBEN}@AF#f+E&yl6y+TjvcUraK7T3eMzk)So%JXa zUlCt9l;_gF$Tsktb`ULhJ>l=Dqh?-~5iwwTt1QM|JtiW6Enxr)o@R-|-_T_pM4-u2 z*@^CQRik^{)%{V-CJOZ+7@i3he{t1|sNT?&s&0KFxwj}e{Qhe<5tLbj8EYq|1w|k) z;viB>6B4S3AWFCr{khstk|zYbm7?VD-m6io2OzP-PO%!TDT8?q;dmon)A zFdw)@c+N3VP}a>uc9Yv{^W^l*a8_=;v;Oea_BB(b5c^h-j<5CKN$F(Q3C?n<2RH?NzXIqKH^{IdS2Uda^i z0LLEOe}13=1gH;X#Kq&mG_CuiE}dyKt$Wg~3&k(K1!*Lu#)*tvab`c%o*J?2u5UIl zRWrH@9ts9tv%7kLx)+&UYbRiaVMV9bN%bbjCY+|yBf|ck|1rH-0S3zA8Z5^ixSoBVmU| z#3oxdgO`1jM7u{~H_oDd-YCy?9!mdQ2=nbr171pGYHs<9>j`G^FD`R`$~=y!YhV;H00ui6U8Nd&989_e*KRo02M=q8(Re))6qD`U&tJr=um5h|}n& zGjmaaK5Hl{s^d8L7r-vJjWF}VStgF?5%yu`Gx_I1uNE>ggthbYH%7@$AK&H$|Ik(# z)8`H}NbWzOS#T3{!c9{t=&?NvcV@-&7J_|Q!-Eb(ccIn^c+`F%hk4D{19h{EbeLsU zI{y=M>-j{)ZKnov9|J*;n);g5cV&MhI6VEy_4a&}9R+YT!&xUIfbzN&n1NxcKotzD zz73b)W~}h{_yrgGj?v(@J*NmhLE(X?xq4d_OsTekMXN`++B7hau`S`Sp*OLj0I;oT zU5V`D!BgIm)oI&rT~%JTTiQG{`mvcFZ-4bk%VM*Y$SfuRZZi@cL{aL=kORo8Mi64_ zmQ`#8rgf;~m-Q&>^vV4%lXb^zZK;IE({<@LTt+U$UFF@3R}_?vo8|qqgnkBut}D$w zdQ>ngc)^4YkW|lPR@a2$K&a2XCm4_9qe)M(G2&hFxigKJ#28UG0e*phKWvX^^S4U> zGIkR4kI7HF|Kb9ppH4`lp@7BS5lkPtx)isZ{iy*rIo00hL;y1gqsEKZ_4P~p-A3ln z55cu_rfd?y+)reSsn~FcCUK=z0Wn0`xJFz%q5qPG3a-kkS@2qZRLl|>#kZS@%VXx9Xgu-7JL~Xjk9<_2-;`|lnJu_LRk`HP3HQ%G{*k1G zM{K^{wYtSHVJ4tSjIB3-aM2?QS9%ets&V`2Do63h3g47%kZ--R@!IACPJt| zEcoqw-S;M;zR@-33iIRF=8D|4$z!kG-gS=!@8??HT75#OO?L1mimKQ4OnJt*whu;l z4y{J`k5eJ`gSj%KS8APa2%(wtv79Uag6IA^lo4vh`qwkKCaY3LRwya=j1VnQUOSrC zQ0oRY%GHl;5d7k#(tQ=hseK2k&YKx-8R$=Gw!DppO2(+tY$fPD^u`(6dZbPraY^B5 zS89caD`svc=LMj@{{H24bPQ@cTaRiWyp@TiqT&}r{4_aQl|5?xr-5X|{H5YcdZ&c< z$6EbBoh*LidVYYI2YkR?_5&@#skkTXcWw>)OUUjgBKj z=d>i}uoBTb=_-?Uo!A``t z4k(6|mi}JtEtf5x=(9SXom-oPH{^jWxxBr<{}vLYtt{5E4B3%`l}q5XYF-hDDBm=- zbx~eucibWI&>`N<{!HHiRsM@AJRv2ke1E9DzAJVys47wb{ZF^q*;sCED40xHM6nd$ zzDOLSv@2yKyG35TVN_yibQ9;dtilzt^^6M}+t}s3@%wvg!ng2c0gHJoowI9fF3!_v z8Oo1*XN%z62#nwj=SV5B)2SPW=YmhH21>l{3=etpcXa(7!Gbv|bPt%ngC)ogq4VTu zFdJD4(Z}iA-{?qW+YCf&_-=}nE!0^&PoeqLGmo>t%u^129lo%BHo z%|aucU2^tk@>IS}*CYP|Ey70xTZl}3tBMO`Txx#BpK&Y1q^S6Z4PSOB9>k=t$#>zqO2iR#i$Lr(?wLSh= zflbto2)mO|SXuy;EI1n$>5uo2Oz!*KI&vvN6w&%RMlB_`@ORLN=UtWK)Q?&Ue2{%@ zWBT3d;v?uJ?`wssAS{61ui<0J6^;2Dna$YP2`R|;Skc{|kBp>xFI70bI?*C+dBE<5 zRri*oo%74lzqda(O{!n@A4{+*6rT`~e89YUF zB1UjPNAcXY^H;x*$53NZtl!~tPLpD8m{^MdlLGkx5*W_X8g$f=go7>-B3WAk#N;OzC)-WcIWJuk^J_AgGCD> zG3{H=SRT%&BZ2;tbU#O^mKjBNZ_cv2yDl?*e`2<72roBSIi)C~%>2I1$tUho>Y#B5 z=6|glUpq4tA_FckKU?W41ra|X+#ez*Y>|{^*{7SROCVI+w5OYX*=A*?a%*5n0?3x} z;Pvt=O#w0fS2-8o;8D;#mRm&^PubbmKK}7W%%agGZ^H8LfaNANf%XouqHFO}zS(n_ zgy)V%QpTL(dj>Fay|k*e+V`67R?ai{?BhOBG*Zh7$aH1d@-P>(1==*GpSh>e9&GV) zVbs?yYk&PMuPJ8odw=i4_i zIwB~Jl=a;D`kPs^oP{5lZwn4;)8XB3AMWHj{#f-x=_jD0{Zx7vMWNtFGL3sl&qCdy zE_Bwal3{y_-mO{8E7;L`MmOgv_VP-6I7$@jNxqjAfD4(ue2qY;t3h}*g-Wd! zQ+cC5M@U))bRvgq|qLI9MEU z{iv-~vut#a`pdvAdlC6Yh7fNoMvqhDJCw?%6ml+YqyxZp^9B5z!Y?{#H1f8^@gsrlDCI zxdX!O4^CZ|<=6!p8J11D0Ej@nfJwQ*3d8o!v%wIww$!&+`cKu|+`m6rkDA@UXn;EX z{UD0TK|o{_!Gj}T+CgL=l`a4pV4(RzQJaYVq@TypiCE_K!ZUv`K3n48)O2!0H^RL7<|5tZHtD-@#~R9ad4V^ z_r|j^U|=|~@F(a0%|h&dwQ%_VXyMGixMuAiu5teDnt!pd_P<*wf%;bifBeHhP3(X5 zAnAYe;D6eO|8E`){j`v_bc{^w1oDq&WIHbA4e~GjrHZ`WtcyN$WI$L1zuKZnj j%K5U&WSOXlA^m-Er(Sv%#$Du&;i3qr<~~`*{x$L+*hKj8 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/manifest.json b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/manifest.json new file mode 100644 index 000000000..19c96aaf3 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/manifest.json @@ -0,0 +1,96 @@ +{ + "version": "2.0", + "job_id": "job_922992fe94d2", + "data_id": null, + "source_file_name": "2025_AnnualReport.docx", + "processing_date": "2026-06-16T03:37:46.586494Z", + "processing": { + "page_count": 72, + "billing_status": "charged", + "cost": { + "micro_dollars": 108000, + "credits": 0.108 + }, + "timing": { + "started_at": "2026-06-16T03:35:08.048494+00:00", + "completed_at": "2026-06-16T03:37:46.549470+00:00", + "duration_ms": 158500 + }, + "stages": { + "timing_ms": { + "document.profile": 0, + "heading.merge_pre_pass_llm": 1080, + "heading.hierarchy_llm_call": 1606, + "heading.hierarchy_llm": 1613, + "document.cleanup_unreferenced_images": 0, + "document.compress_images": 0, + "worker.parse.document": 158375 + }, + "token_usage": { + "prompt_tokens": 97895, + "completion_tokens": 5293, + "total_tokens": 103188 + } + } + }, + "statistics": { + "total_chunks": 122, + "text_chunks": 49, + "image_chunks": 4, + "table_chunks": 69, + "total_pages": null + }, + "HIERARCHY": { + "**Preface**": { + "**Preface** part 1": {}, + "**Preface** part 2": {}, + "**Preface** part 3": {}, + "**Preface** part 4": {}, + "**Preface** part 5": {}, + "**Preface** part 6": {}, + "**Preface** part 7": {}, + "**Preface** part 8": {}, + "**Preface** part 9": {}, + "**Preface** part 10": {}, + "**Preface** part 11": {}, + "**Preface** part 12": {}, + "**Preface** part 13": {}, + "**Preface** part 14": {}, + "**Preface** part 15": {}, + "**Preface** part 16": {}, + "**Preface** part 17": {}, + "**Preface** part 18": {}, + "**Preface** part 19": {}, + "**Preface** part 20": {}, + "**Preface** part 21": {}, + "**Preface** part 22": {}, + "**Preface** part 23": {}, + "**Preface** part 24": {}, + "**Preface** part 25": {}, + "**Preface** part 26": {}, + "**Preface** part 27": {}, + "**Preface** part 28": {}, + "**Preface** part 29": {}, + "**Preface** part 30": {}, + "**Preface** part 31": {}, + "**Preface** part 32": {}, + "**Preface** part 33": {}, + "**Preface** part 34": {}, + "**Preface** part 35": {}, + "**Preface** part 36": {}, + "**Preface** part 37": {}, + "**Preface** part 38": {}, + "**Preface** part 39": {}, + "**Preface** part 40": {}, + "**Preface** part 41": {}, + "**Preface** part 42": {}, + "**Preface** part 43": {}, + "**Preface** part 44": {}, + "**Preface** part 45": {}, + "**Preface** part 46": {}, + "**Preface** part 47": {}, + "**Preface** part 48": {} + }, + "4.\tEnvironmental, Social, and Public Policy Committee": {} + } +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-1 Share Repurchases.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-1 Share Repurchases.html new file mode 100644 index 000000000..e28e005ed --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-1 Share Repurchases.html @@ -0,0 +1 @@ +
(In millions)SharesAmountSharesAmountSharesAmount
Year Ended June 30,202520242023
First Quarter7$2,80011$3,56017$4,600
Second Quarter83,50072,800204,600
Third Quarter83,50072,800184,600
Fourth Quarter83,20072,800144,600
Total31$13,00032$11,96069$18,400
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-10 Sales & Marketing.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-10 Sales & Marketing.html new file mode 100644 index 000000000..0e66437fa --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-10 Sales & Marketing.html @@ -0,0 +1 @@ +
(In millions, except percentages)20252024Percentage
Change
Sales and marketing$25,654$24,4565%
As a percent of revenue9%10%(1)ppt
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-11 G&A Expenses.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-11 G&A Expenses.html new file mode 100644 index 000000000..bc22dd5d1 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-11 G&A Expenses.html @@ -0,0 +1 @@ +
(In millions, except percentages)20252024Percentage
Change
General and administrative$7,223$7,609(5)%
As a percent of revenue3%3%0ppt
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-12 Financial Summary.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-12 Financial Summary.html new file mode 100644 index 000000000..aeaf9b293 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-12 Financial Summary.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,20252024
Interest and dividends income$2,647$3,157
Interest expense(2,385)(2,935)
Net recognized losses on investments(349)(118)
Net losses on derivatives(260)(187)
Net gains (losses) on foreign currency remeasurements171(244)
Other, net(4,725)(1,319)
Total$(4,901)$(1,646)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-13 Revenue Schedule.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-13 Revenue Schedule.html new file mode 100644 index 000000000..ac444decb --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-13 Revenue Schedule.html @@ -0,0 +1 @@ +
(In millions)
Three Months Ending
September 30, 2025$25,191
December 31, 202519,733
March 31, 202613,742
June 30, 20265,889
Thereafter2,710
Total$67,265
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-14 Debt Commitments.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-14 Debt Commitments.html new file mode 100644 index 000000000..7e17603d2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-14 Debt Commitments.html @@ -0,0 +1 @@ +
(In millions)2026ThereafterTotal
Long-term debt: (a)
Principal payments$3,000$46,206$49,206
Interest payments1,50925,52727,036
Construction commitments (b)26,8595,29032,149
Operating and finance leases, including imputed interest (c)12,798165,903178,701
Purchase commitments (d)103,9406,013109,953
Total$148,106$248,939$397,045
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-15 Risk Sensitivity.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-15 Risk Sensitivity.html new file mode 100644 index 000000000..74aeed8b7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-15 Risk Sensitivity.html @@ -0,0 +1 @@ +
(In millions)
Risk CategoriesHypothetical ChangeJune 30,
2025
Impact
Foreign currency – Revenue10% decrease in foreign exchange rates$(11,596)Earnings
Foreign currency – Investments10% decrease in foreign exchange rates(17)Fair Value
Interest rate100 basis point increase in U.S. treasury interest rates(1,415)Fair Value
Credit100 basis point increase in credit spreads(436)Fair Value
Equity10% decrease in equity market prices(1,213)Earnings
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-16 Revenue & Income.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-16 Revenue & Income.html new file mode 100644 index 000000000..153a16936 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-16 Revenue & Income.html @@ -0,0 +1 @@ +
(In millions, except per share amounts)
Year Ended June 30,202520242023
Revenue:
Product$63,946$64,773$64,699
Service and other217,778180,349147,216
Total revenue281,724245,122211,915
Cost of revenue:
Product13,50115,27217,804
Service and other74,33058,84248,059
Total cost of revenue87,83174,11465,863
Gross margin193,893171,008146,052
Research and development32,48829,51027,195
Sales and marketing25,65424,45622,759
General and administrative7,2237,6097,575
Operating income128,528109,43388,523
Other income (expense), net(4,901)(1,646)788
Income before income taxes123,627107,78789,311
Provision for income taxes21,79519,65116,950
Net income$101,832$88,136$72,361
Earnings per share:
Basic$13.70$11.86$9.72
Diluted$13.64$11.80$9.68
Weighted average shares outstanding:
Basic7,4337,4317,446
Diluted7,4657,4697,472
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-17 Comprehensive Income.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-17 Comprehensive Income.html new file mode 100644 index 000000000..4ba680766 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-17 Comprehensive Income.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Net income$101,832$88,136$72,361
Other comprehensive income (loss), net of tax:
Net change related to derivatives(5)24(14)
Net change related to investments1,574957(1,444)
Translation adjustments and other674(228)(207)
Other comprehensive income (loss)2,243753(1,665)
Comprehensive income$104,075$88,889$70,696
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-18 Balance Sheet.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-18 Balance Sheet.html new file mode 100644 index 000000000..0f9e9ef74 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-18 Balance Sheet.html @@ -0,0 +1 @@ +
(In millions)
June 30,20252024
Assets
Current assets:
Cash and cash equivalents$30,242$18,315
Short-term investments64,32357,228
Total cash, cash equivalents, and short-term investments94,56575,543
Accounts receivable, net of allowance for doubtful accounts of $944 and $83069,90556,924
Inventories9381,246
Other current assets25,72326,021
Total current assets191,131159,734
Property and equipment, net of accumulated depreciation of $93,653 and $76,421204,966135,591
Operating lease right-of-use assets24,82318,961
Equity and other investments15,40514,600
Goodwill119,509119,220
Intangible assets, net22,60427,597
Other long-term assets40,56536,460
Total assets$619,003$512,163
Liabilities and stockholders’ equity
Current liabilities:
Accounts payable$27,724$21,996
Short-term debt06,693
Current portion of long-term debt2,9992,249
Accrued compensation13,70912,564
Short-term income taxes7,2115,017
Short-term unearned revenue64,55557,582
Other current liabilities25,02019,185
Total current liabilities141,218125,286
Long-term debt40,15242,688
Long-term income taxes25,98627,931
Long-term unearned revenue2,7102,602
Deferred income taxes2,8352,618
Operating lease liabilities17,43715,497
Other long-term liabilities45,18627,064
Total liabilities275,524243,686
Commitments and contingencies
Stockholders’ equity:
Common stock and paid-in capital – shares authorized 24,000; outstanding 7,434 and 7,434109,095100,923
Retained earnings237,731173,144
Accumulated other comprehensive loss(3,347)(5,590)
Total stockholders’ equity343,479268,477
Total liabilities and stockholders’ equity$619,003$512,163
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-19 Cash Flow Statement.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-19 Cash Flow Statement.html new file mode 100644 index 000000000..dd8d9296d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-19 Cash Flow Statement.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Operations
Net income$101,832$88,136$72,361
Adjustments to reconcile net income to net cash from operations:
Depreciation, amortization, and other34,15322,28713,861
Stock-based compensation expense11,97410,7349,611
Net recognized losses on investments and derivatives609305196
Deferred income taxes(7,056)(4,738)(6,059)
Changes in operating assets and liabilities:
Accounts receivable(10,581)(7,191)(4,087)
Inventories3091,2841,242
Other current assets(3,044)(1,648)(1,991)
Other long-term assets(2,950)(6,817)(2,833)
Accounts payable5693,545(2,721)
Unearned revenue5,4385,3485,535
Income taxes(38)1,687(358)
Other current liabilities5,9224,8672,272
Other long-term liabilities(975)749553
Net cash from operations136,162118,54887,582
Financing
Proceeds from issuance (repayments) of debt, maturities of 90 days or less, net(5,746)5,2500
Proceeds from issuance of debt024,3950
Repayments of debt(3,216)(29,070)(2,750)
Common stock issued2,0562,0021,866
Common stock repurchased(18,420)(17,254)(22,245)
Common stock cash dividends paid(24,082)(21,771)(19,800)
Other, net(2,291)(1,309)(1,006)
Net cash used in financing(51,699)(37,757)(43,935)
Investing
Additions to property and equipment(64,551)(44,477)(28,107)
Acquisition of companies, net of cash acquired and divestitures, and purchases of intangible and other assets(5,978)(69,132)(1,670)
Purchases of investments(29,775)(17,732)(37,651)
Maturities of investments16,07924,77533,510
Sales of investments9,30910,89414,354
Other, net2,317(1,298)(3,116)
Net cash used in investing(72,599)(96,970)(22,680)
Effect of foreign exchange rates on cash and cash equivalents63(210)(194)
Net change in cash and cash equivalents11,927(16,389)20,773
Cash and cash equivalents, beginning of period18,31534,70413,931
Cash and cash equivalents, end of period$30,242$18,315$34,704
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-2 Dividend History.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-2 Dividend History.html new file mode 100644 index 000000000..02b0ba965 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-2 Dividend History.html @@ -0,0 +1 @@ +
Declaration DateRecord DatePayment DateDividend
Per Share
Amount
Fiscal Year 2025(In millions)
September 16, 2024November 21, 2024December 12, 2024$0.83$6,170
December 3, 2024February 20, 2025March 13, 20250.836,169
March 11, 2025May 15, 2025June 12, 20250.836,169
June 10, 2025August 21, 2025September 11, 20250.836,170
Total$3.32$24,678
Fiscal Year 2024
September 19, 2023November 16, 2023December 14, 2023$0.75$5,574
November 28, 2023February 15, 2024March 14, 20240.755,573
March 12, 2024May 16, 2024June 13, 20240.755,574
June 12, 2024August 15, 2024September 12, 20240.755,574
Total$3.00$22,295
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-20 Stockholders' Equity.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-20 Stockholders' Equity.html new file mode 100644 index 000000000..cf97d7d55 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-20 Stockholders' Equity.html @@ -0,0 +1 @@ +
(In millions, except per share amounts)
Year Ended June 30,202520242023
Common stock and paid-in capital
Balance, beginning of period$100,923$93,718$86,939
Common stock issued2,0562,0021,866
Common stock repurchased(5,856)(5,712)(4,696)
Stock-based compensation expense11,97410,7349,611
Other, net(2)181(2)
Balance, end of period109,095100,92393,718
Retained earnings
Balance, beginning of period173,144118,84884,281
Net income101,83288,13672,361
Common stock cash dividends(24,677)(22,293)(20,226)
Common stock repurchased(12,568)(11,547)(17,568)
Balance, end of period237,731173,144118,848
Accumulated other comprehensive loss
Balance, beginning of period(5,590)(6,343)(4,678)
Other comprehensive income (loss)2,243753(1,665)
Balance, end of period(3,347)(5,590)(6,343)
Total stockholders’ equity$343,479$268,477$206,223
Cash dividends declared per common share$3.32$3.00$2.72
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-21 EPS Data Table.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-21 EPS Data Table.html new file mode 100644 index 000000000..98e1c072c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-21 EPS Data Table.html @@ -0,0 +1 @@ +
(In millions, except per share amounts)
Year Ended June 30,202520242023
Net income available for common shareholders (A)$101,832$88,136$72,361
Weighted average outstanding shares of common stock (B)7,4337,4317,446
Dilutive effect of stock-based awards323826
Common stock and common stock equivalents (C)7,4657,4697,472
Earnings Per Share
Basic (A/B)$13.70$11.86$9.72
Diluted (A/C)$13.64$11.80$9.68
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-22 Investment Income.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-22 Investment Income.html new file mode 100644 index 000000000..b2d8283e2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-22 Investment Income.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Interest and dividends income$2,647$3,157$2,994
Interest expense(2,385)(2,935)(1,968)
Net recognized gains (losses) on investments(349)(118)260
Net losses on derivatives(260)(187)(456)
Net gains (losses) on foreign currency remeasurements171(244)181
Other, net(4,725)(1,319)(223)
Total$(4,901)$(1,646)$788
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-23 Investment Gains.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-23 Investment Gains.html new file mode 100644 index 000000000..499ea82a1 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-23 Investment Gains.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Realized gains from sales of available-for-sale securities$40$22$36
Realized losses from sales of available-for-sale securities(65)(98)(124)
Impairments and allowance for credit losses823(10)
Total$(17)$(53)$(98)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-24 Investment Gains.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-24 Investment Gains.html new file mode 100644 index 000000000..32cd8ea40 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-24 Investment Gains.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Net realized gains on investments sold$83$18$75
Net unrealized gains on investments still held536146303
Impairments of investments(951)(229)(20)
Total$(332)$(65)$358
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-25 Investment Table.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-25 Investment Table.html new file mode 100644 index 000000000..54456bdd2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-25 Investment Table.html @@ -0,0 +1 @@ +
(In millions)Fair Value
Level
Adjusted
Cost Basis
Unrealized
Gains
Unrealized
Losses
Recorded
Basis
Cash and
Cash
Equivalents
Short-term
Investments
Equity and
Other
Investments
June 30, 2025
Changes in Fair Value Recorded in
Other Comprehensive Income
Commercial paperLevel 2$10,880$0$0$10,880$9,939$941$0
Certificates of depositLevel 22,653002,6532,3093440
U.S. government securitiesLevel 152,87871(1,462)51,4874,74246,7450
U.S. agency securitiesLevel 22,686002,6864962,1900
Foreign government bondsLevel 234924(9)36403640
Mortgage- and asset-backed securitiesLevel 22,55810(27)2,54102,5410
Corporate notes and bondsLevel 210,763124(101)10,786010,7860
Corporate notes and bondsLevel 32,51165(5)2,57101112,460
Municipal securitiesLevel 22071(7)20102010
Municipal securitiesLevel 31040(14)900900
Total debt investments$85,589$295$(1,625)$84,259$17,486$64,313$2,460
Changes in Fair Value Recorded in Net Income
Equity investmentsLevel 1$4,577$1,045$0$3,532
Equity investmentsOther9,141009,141
Total equity investments$13,718$1,045$0$12,673
Cash$11,711$11,711$0$0
Derivatives, net (a)282010272
Total$109,970$30,242$64,323$15,405
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-26 Investment Summary.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-26 Investment Summary.html new file mode 100644 index 000000000..cbf2c2109 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-26 Investment Summary.html @@ -0,0 +1 @@ +
(In millions)Fair Value
Level
Adjusted
Cost Basis
Unrealized
Gains
Unrealized
Losses
Recorded
Basis
Cash and
Cash
Equivalents
Short-term
Investments
Equity and
Other
Investments
June 30, 2024
Changes in Fair Value Recorded in Other Comprehensive Income
Commercial paperLevel 2$4,666$0$0$4,666$4,666$0$0
Certificates of depositLevel 21,547001,5471,503440
U.S. government securitiesLevel 149,6034(2,948)46,6591446,6450
U.S. agency securitiesLevel 21700170170
Foreign government bondsLevel 23193(16)30603060
Mortgage- and asset-backed securitiesLevel 29443(35)91209120
Corporate notes and bondsLevel 29,10628(318)8,81608,8160
Corporate notes and bondsLevel 31,6410(1)1,64001401,500
Municipal securitiesLevel 22620(13)24902490
Municipal securitiesLevel 31040(17)870870
Total debt investments$68,209$38$(3,348)$64,899$6,183$57,216$1,500
Changes in Fair Value Recorded in Net Income
Equity investmentsLevel 1$3,547$561$0$2,986
Equity investmentsOther10,1140010,114
Total equity investments$13,661$561$0$13,100
Cash$11,571$11,571$0$0
Derivatives, net (a)120120
Total$90,143$18,315$57,228$14,600
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-27 Unrealized Losses.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-27 Unrealized Losses.html new file mode 100644 index 000000000..67335c6d3 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-27 Unrealized Losses.html @@ -0,0 +1 @@ +
Less than 12 Months12 Months or GreaterTotal
Unrealized
Losses
Total
Unrealized
Losses
(In millions)Fair ValueUnrealized
Losses
Fair ValueUnrealized
Losses
Total
Fair Value
Total
Unrealized
Losses
June 30, 2025
U.S. government and agency securities$2,569$(51)$34,608$(1,411)$37,177$(1,462)
Foreign government bonds43(2)106(7)149(9)
Mortgage- and asset-backed securities841(4)189(23)1,030(27)
Corporate notes and bonds1,107(8)3,105(98)4,212(106)
Municipal securities00168(21)168(21)
Total$4,560$(65)$38,176$(1,560)$42,736$(1,625)
Less than 12 Months12 Months or GreaterTotal
Unrealized
Losses
Total
Unrealized
Losses
(In millions)Fair ValueUnrealized
Losses
Fair ValueUnrealized
Losses
Total
Fair Value
Total
Unrealized
Losses
June 30, 2024
U.S. government and agency securities$529$(12)$45,821$(2,936)$46,350$(2,948)
Foreign government bonds79(2)180(14)259(16)
Mortgage- and asset-backed securities201(1)409(34)610(35)
Corporate notes and bonds1,310(9)5,779(310)7,089(319)
Municipal securities38(1)243(29)281(30)
Total$2,157$(25)$52,432$(3,323)$54,589$(3,348)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-28 Debt Securities.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-28 Debt Securities.html new file mode 100644 index 000000000..3ada4788e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-28 Debt Securities.html @@ -0,0 +1 @@ +
(In millions)Adjusted
Cost Basis
Estimated
Fair Value
June 30, 2025
Due in one year or less$35,108$34,952
Due after one year through five years42,46041,481
Due after five years through 10 years6,5306,424
Due after 10 years1,4911,402
Total$85,589$84,259
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-29 Derivative Notional.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-29 Derivative Notional.html new file mode 100644 index 000000000..2b7db3ae1 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-29 Derivative Notional.html @@ -0,0 +1 @@ +
(In millions)June 30,
2025
June 30,
2024
Designated as Hedging Instruments
Foreign exchange contracts purchased$1,492$1,492
Interest rate contracts purchased1,1501,100
Not Designated as Hedging Instruments
Foreign exchange contracts purchased15,2147,167
Foreign exchange contracts sold43,30731,793
Equity contracts purchased5,4344,016
Equity contracts sold2,1892,165
Other contracts purchased2,7692,113
Other contracts sold1,242811
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-3 Stock Index Data.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-3 Stock Index Data.html new file mode 100644 index 000000000..1aaf90de8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-3 Stock Index Data.html @@ -0,0 +1 @@ +
6/206/216/226/236/246/25
Microsoft Corporation100.00134.41128.48172.01227.51255.13
S&P 500100.00140.79125.85150.51187.47215.89
NASDAQ Computer100.00150.44117.59154.73223.97254.97
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-30 Derivative Assets.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-30 Derivative Assets.html new file mode 100644 index 000000000..03a30921f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-30 Derivative Assets.html @@ -0,0 +1 @@ +
(In millions)Derivative
Assets
Derivative
Liabilities
Derivative
Assets
Derivative
Liabilities
June 30,
2025
June 30,
2024
Designated as Hedging Instruments
Foreign exchange contracts$89$(44)$24$(76)
Interest rate contracts150190
Not Designated as Hedging Instruments
Foreign exchange contracts248(809)213(230)
Equity contracts385(983)63(491)
Other contracts21(1)12(3)
Gross amounts of derivatives758(1,837)331(800)
Gross amounts of derivatives offset in the balance sheets(258)260(151)152
Cash collateral received0(99)0(104)
Net amounts of derivatives$500$(1,676)$180$(752)
Reported as
Short-term investments$10$0$12$0
Other current assets20101490
Equity and other investments272000
Other long-term assets170190
Other current liabilities0(1,639)0(401)
Other long-term liabilities0(37)0(351)
Total$500$(1,676)$180$(752)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-31 Derivatives by Level.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-31 Derivatives by Level.html new file mode 100644 index 000000000..60dab157f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-31 Derivatives by Level.html @@ -0,0 +1 @@ +
(In millions)Level 1Level 2Level 3Total
June 30, 2025
Derivative assets$1$474$283$758
Derivative liabilities0(1,832)(5)(1,837)
June 30, 2024
Derivative assets03274331
Derivative liabilities(1)(799)0(800)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-32 Hedging Gains_Losses.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-32 Hedging Gains_Losses.html new file mode 100644 index 000000000..5a9cd0914 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-32 Hedging Gains_Losses.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Designated as Fair Value Hedging Instruments
Interest rate contracts
Derivatives$5$(23)$(65)
Hedged items(45)(25)38
Designated as Cash Flow Hedging Instruments
Foreign exchange contracts
Amount reclassified from accumulated other comprehensive loss103(48)61
Not Designated as Hedging Instruments
Foreign exchange contracts(938)367(73)
Equity contracts(266)(177)(420)
Other contracts21(15)(41)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-33 Cash Flow Hedges.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-33 Cash Flow Hedges.html new file mode 100644 index 000000000..1ca1c4547 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-33 Cash Flow Hedges.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Designated as Cash Flow Hedging Instruments
Foreign exchange contracts
Included in effectiveness assessment$77$   (14)$34
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-34 Property Assets.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-34 Property Assets.html new file mode 100644 index 000000000..aadbc30e1 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-34 Property Assets.html @@ -0,0 +1 @@ +
(In millions)
June 30,20252024
Land$9,338$8,163
Buildings and improvements137,92193,943
Leasehold improvements12,1179,594
Computer equipment and software132,83693,780
Furniture and equipment6,4076,532
Total, at cost298,619212,012
Accumulated depreciation(93,653)(76,421)
Total, net$204,966$135,591
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-35 Purchase Price.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-35 Purchase Price.html new file mode 100644 index 000000000..0a73589e2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-35 Purchase Price.html @@ -0,0 +1 @@ +
(In millions)
Cash and cash equivalents$12,976
Goodwill51,001
Intangible assets21,969
Other assets2,503
Long-term debt(2,799)
Long-term income taxes(1,946)
Deferred income taxes(4,676)
Other liabilities(3,620)
Total purchase price$75,408
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-36 Intangible Assets.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-36 Intangible Assets.html new file mode 100644 index 000000000..3cbe10c02 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-36 Intangible Assets.html @@ -0,0 +1 @@ +
(In millions, except average life)AmountWeighted
Average Life
Marketing-related$11,61924 years
Technology-based9,6894 years
Customer-related6614 years
Fair value of intangible assets acquired$21,96915 years
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-37 Microsoft FY2024.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-37 Microsoft FY2024.html new file mode 100644 index 000000000..5b12eae6d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-37 Microsoft FY2024.html @@ -0,0 +1 @@ +
(In millions, except per share amounts)
Year Ended June 30,20242023
Revenue$247,442$219,790
Net income88,30871,383
Diluted earnings per share11.829.55
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-38 Goodwill by Segment.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-38 Goodwill by Segment.html new file mode 100644 index 000000000..a510eb2d6 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-38 Goodwill by Segment.html @@ -0,0 +1 @@ +
(In millions)June 30,
2023
AcquisitionsOtherJune 30,
2024
AcquisitionsOtherJune 30,
2025
Productivity and Business Processes$31,359$0$2$31,361$0$96$31,457
Intelligent Cloud25,6760(28)25,64804125,689
More Personal Computing10,85151,23512562,211015262,363
Total$67,886$51,235$99$119,220$0$289$119,509
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-39 Intangible Assets.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-39 Intangible Assets.html new file mode 100644 index 000000000..1fad0d5e4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-39 Intangible Assets.html @@ -0,0 +1 @@ +
(In millions)Gross
Carrying
Amount
Accumulated
Amortization
Net
Carrying
Amount
Gross
Carrying
Amount
Accumulated
Amortization
Net
Carrying
Amount
June 30,20252024
Marketing-related$16,502$(3,901)$12,601$16,500$(3,101)$13,399
Technology-based22,560(14,959)7,60121,913(10,741)11,172
Customer-related4,278(2,050)2,2286,038(3,051)2,987
Contract-based217(43)17458(19)39
Total$43,557$(20,953)$22,604$44,509$(16,912)$27,597
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-4 Microsoft Cloud KPIs.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-4 Microsoft Cloud KPIs.html new file mode 100644 index 000000000..0a8230155 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-4 Microsoft Cloud KPIs.html @@ -0,0 +1 @@ +
Commercial remaining performance obligationCommercial portion of revenue allocated to remaining performance obligations, which includes unearned revenue and amounts that will be invoiced and recognized as revenue in future periods
Microsoft Cloud revenue and revenue growthRevenue from Microsoft 365 Commercial cloud, Azure and other cloud services, the commercial portion of LinkedIn, and Dynamics 365
Microsoft Cloud gross margin percentageGross margin percentage for our Microsoft Cloud business
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-40 Intangible Assets.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-40 Intangible Assets.html new file mode 100644 index 000000000..459a5f11e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-40 Intangible Assets.html @@ -0,0 +1 @@ +
(In millions)AmountWeighted
Average Life
AmountWeighted
Average Life
Year Ended June 30,20252024
Marketing-related$1310 years$11,61924 years
Technology-based9129 years10,9474 years
Customer-related00 years6604 years
Contract-based1715 years384 years
Total$1,0969 years$23,26414 years
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-41 Revenue Forecast.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-41 Revenue Forecast.html new file mode 100644 index 000000000..c8e710b62 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-41 Revenue Forecast.html @@ -0,0 +1 @@ +
(In millions)
Year Ending June 30,
2026$4,594
20272,901
20282,034
20291,851
20301,382
Thereafter9,842
Total$22,604
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-42 Debt Issuance Table.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-42 Debt Issuance Table.html new file mode 100644 index 000000000..6d7eaae08 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-42 Debt Issuance Table.html @@ -0,0 +1 @@ +
(In millions, issuance by calendar year)Maturities
(calendar year)
Stated Interest
Rate
Effective
Interest
Rate
June 30,
2025
June 30,
2024
2009 issuance of $3.8 billion20395.20%5.24%$520$520
2010 issuance of $4.8 billion20404.50%4.57%486486
2011 issuance of $2.3 billion20415.30%5.36%718718
2012 issuance of $2.3 billion20423.50%3.57%454454
2013 issuance of $5.2 billion20433.75%–4.88%3.83%–4.92%314314
2013 issuance of €4.1 billion2028–20332.63%–3.13%2.69%–3.22%2,7002,465
2015 issuance of $23.8 billion2025–20553.13%–4.75%3.18%–4.78%7,5559,805
2016 issuance of $19.8 billion2026–20562.40%–3.95%2.46%–4.03%7,9307,930
2017 issuance of $17.1 billion2026–20573.30%–4.50%3.38%–5.49%6,8336,833
2020 issuance of $10.1 billion2030–20601.35%–2.68%2.53%–5.43%10,11110,111
2021 issuance of $8.2 billion2052–20622.92%–3.04%2.92%–3.04%8,1858,185
2023 issuance of $0.1 billion2026–20501.35%–4.50%5.16%–5.49%5656
2024 issuance of $3.3 billion2026–20501.35%–4.50%5.16%–5.49%3,3443,344
Total face value49,20651,221
Unamortized discount and
issuance costs
(1,155)(1,227)
Hedge fair value adjustments (a)(36)(81)
Premium on debt exchange(4,864)(4,976)
Total debt43,15144,937
Current portion of long-term debt(2,999)(2,249)
Long-term debt$40,152$42,688
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-43 Debt Maturity.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-43 Debt Maturity.html new file mode 100644 index 000000000..2790ea890 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-43 Debt Maturity.html @@ -0,0 +1 @@ +
(In millions)
Year Ending June 30,
2026$3,000
20279,250
20280
20292,054
20300
Thereafter34,902
Total$49,206
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-44 Income Tax Provision.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-44 Income Tax Provision.html new file mode 100644 index 000000000..e59a4c4b2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-44 Income Tax Provision.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Current Taxes
U.S. federal$14,086$12,165$14,009
U.S. state and local3,3422,3662,322
Foreign11,4239,8586,678
Current taxes$28,851$24,389$23,009
Deferred Taxes
U.S. federal$(6,250)$(4,791)$(6,146)
U.S. state and local(1,087)(379)(477)
Foreign281432564
Deferred taxes$(7,056)$(4,738)$(6,059)
Provision for income taxes$21,795$19,651$16,950
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-45 Income Before Taxes.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-45 Income Before Taxes.html new file mode 100644 index 000000000..757bb5778 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-45 Income Before Taxes.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
U.S.$69,212$62,886$52,917
Foreign54,41544,90136,394
Income before income taxes$123,627$107,787$89,311
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-46 Tax Rate Analysis.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-46 Tax Rate Analysis.html new file mode 100644 index 000000000..fbf526a2a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-46 Tax Rate Analysis.html @@ -0,0 +1 @@ +
Year Ended June 30,202520242023
Federal statutory rate21.0%21.0%21.0%
Effect of:
Foreign earnings taxed at lower rates(1.5)%(1.4)%(1.8)%
Foreign-derived intangible income deduction(1.0)%(1.1)%(1.3)%
State income taxes, net of federal benefit1.5%1.5%1.6%
Research and development credit(1.1)%(1.1)%(1.1)%
Excess tax benefits relating to stock-based compensation(0.9)%(1.1)%(0.7)%
Interest, net1.0%1.1%0.8%
Other reconciling items, net(1.4)%(0.7)%0.5%
Effective rate17.6%18.2%19.0%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-47 Deferred Tax Assets.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-47 Deferred Tax Assets.html new file mode 100644 index 000000000..2d79477a9 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-47 Deferred Tax Assets.html @@ -0,0 +1 @@ +
(In millions)
June 30,20252024
Deferred Income Tax Assets
Stock-based compensation expense$909$765
Accruals, reserves, and other expenses5,0504,381
Loss and credit carryforwards2,1141,741
Amortization4,1184,159
Leasing liabilities12,8746,504
Unearned revenue4,3243,717
Book/tax basis differences in investments and debt3039
Capitalized research and development16,89111,442
Other529426
Deferred income tax assets47,11233,144
Less valuation allowance(1,169)(1,045)
Deferred income tax assets, net of valuation allowance$45,943$32,099
Deferred Income Tax Liabilities
Leasing assets$(12,696)$(6,503)
Depreciation(5,699)(3,940)
Deferred tax on foreign earnings(1,148)(1,837)
Other(127)(167)
Deferred income tax liabilities$(19,670)$(12,447)
Net deferred income tax assets$26,273$19,652
Reported As
Other long-term assets$29,108$22,270
Long-term deferred income tax liabilities(2,835)(2,618)
Net deferred income tax assets$26,273$19,652
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-48 Tax Benefits.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-48 Tax Benefits.html new file mode 100644 index 000000000..353246b5d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-48 Tax Benefits.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Beginning unrecognized tax benefits$22,760$17,120$15,593
Decreases related to settlements(240)(76)(329)
Increases for tax positions related to the current year2,0661,9031,051
Increases for tax positions related to prior years4684,289870
Decreases for tax positions related to prior years(300)(464)(60)
Decreases due to lapsed statutes of limitations(25)(12)(5)
Ending unrecognized tax benefits$24,729$22,760$17,120
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-49 Microsoft Revenue.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-49 Microsoft Revenue.html new file mode 100644 index 000000000..61b433ec8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-49 Microsoft Revenue.html @@ -0,0 +1 @@ +
(In millions)
June 30,20252024
Productivity and Business Processes$50,567$43,599
Intelligent Cloud14,02213,683
More Personal Computing2,6762,902
Total$67,265$60,184
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-5 Microsoft Cloud KPIs.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-5 Microsoft Cloud KPIs.html new file mode 100644 index 000000000..e5c6240b4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-5 Microsoft Cloud KPIs.html @@ -0,0 +1 @@ +
Microsoft 365 Commercial cloud revenue growthRevenue from Microsoft 365 Commercial subscriptions, comprising Microsoft 365 Commercial, Enterprise Mobility + Security, the cloud portion of Windows Commercial, the per-user portion of Power BI, Exchange, SharePoint, Microsoft Teams, Microsoft 365 Security and Compliance, and Microsoft 365 Copilot
Microsoft 365 Commercial seat growthThe number of Microsoft 365 Commercial seats at end of period where seats are paid users covered by a Microsoft 365 Commercial subscription
Microsoft 365 Consumer cloud revenue growthRevenue from Microsoft 365 Consumer subscriptions and other consumer services
Microsoft 365 Consumer subscribersThe number of Microsoft 365 Consumer subscribers at end of period
LinkedIn revenue growthRevenue from LinkedIn, including Talent Solutions, Marketing Solutions, Premium Subscriptions, and Sales Solutions
Dynamics 365 revenue growthRevenue from Dynamics 365, including a set of intelligent, cloud-based applications across ERP, CRM, Power Apps, and Power Automate
Azure and other cloud services revenue growthRevenue from Azure and other cloud services, including cloud and AI consumption-based services, GitHub cloud services, Nuance Healthcare cloud services, virtual desktop offerings, and other cloud services
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-50 Unearned Revenue.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-50 Unearned Revenue.html new file mode 100644 index 000000000..125c2d5b8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-50 Unearned Revenue.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30, 2025
Balance, beginning of period$60,184
Deferral of revenue186,957
Recognition of unearned revenue(179,876)
Balance, end of period$67,265
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-51 Lease Costs.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-51 Lease Costs.html new file mode 100644 index 000000000..5667bfcde --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-51 Lease Costs.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Operating lease cost$5,524$3,555$2,875
Finance lease cost:
Amortization of right-of-use assets$3,408$1,800$1,352
Interest on lease liabilities1,417734501
Total finance lease cost$4,825$2,534$1,853
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-52 Lease Cash Flows.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-52 Lease Cash Flows.html new file mode 100644 index 000000000..fa3f736dc --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-52 Lease Cash Flows.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Cash paid for amounts included in the measurement of lease liabilities:
Operating cash flows from operating leases$4,931$3,550$2,706
Operating cash flows from finance leases1,372734501
Financing cash flows from finance leases2,2831,2861,056
Right-of-use assets obtained in exchange for lease obligations:
Operating leases7,8266,7033,514
Finance leases20,51111,6333,128
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-53 Lease Data.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-53 Lease Data.html new file mode 100644 index 000000000..ab70e756b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-53 Lease Data.html @@ -0,0 +1 @@ +
(In millions, except lease term and discount rate)
June 30,20252024
Operating Leases
Operating lease right-of-use assets$24,823$18,961
Other current liabilities$5,424$3,580
Operating lease liabilities17,43715,497
Total operating lease liabilities$22,861$19,077
Finance Leases
Property and equipment, at cost$53,876$32,248
Accumulated depreciation(9,861)(6,386)
Property and equipment, net$44,015$25,862
Other current liabilities$3,172$2,349
Other long-term liabilities43,00024,796
Total finance lease liabilities$46,172$27,145
Weighted Average Remaining Lease Term
Operating leases6 years7 years
Finance leases13 years12 years
Weighted Average Discount Rate
Operating leases3.5%3.3%
Finance leases4.2%3.9%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-54 Lease Payments.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-54 Lease Payments.html new file mode 100644 index 000000000..80d103dd9 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-54 Lease Payments.html @@ -0,0 +1 @@ +
(In millions)
Year Ending June 30,Operating
Leases
Finance
Leases
2026$6,111$5,008
20275,2375,157
20283,4955,187
20292,4194,521
20302,0174,382
Thereafter6,20236,251
Total lease payments25,48160,506
Less imputed interest(2,620)(14,334)
Total$22,861$46,172
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-55 Share Repurchases.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-55 Share Repurchases.html new file mode 100644 index 000000000..f6cccb6c1 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-55 Share Repurchases.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Balance, beginning of year7,4347,4327,464
Issued313437
Repurchased(31)(32)(69)
Balance, end of year7,4347,4347,432
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-56 Share Repurchases.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-56 Share Repurchases.html new file mode 100644 index 000000000..f332cf012 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-56 Share Repurchases.html @@ -0,0 +1 @@ +
(In millions)SharesAmountSharesAmountSharesAmount
Year Ended June 30,202520242023
First Quarter7$2,80011$3,56017$4,600
Second Quarter83,50072,800204,600
Third Quarter83,50072,800184,600
Fourth Quarter83,20072,800144,600
Total31$13,00032$11,96069$18,400
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-57 Dividend History.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-57 Dividend History.html new file mode 100644 index 000000000..02b0ba965 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-57 Dividend History.html @@ -0,0 +1 @@ +
Declaration DateRecord DatePayment DateDividend
Per Share
Amount
Fiscal Year 2025(In millions)
September 16, 2024November 21, 2024December 12, 2024$0.83$6,170
December 3, 2024February 20, 2025March 13, 20250.836,169
March 11, 2025May 15, 2025June 12, 20250.836,169
June 10, 2025August 21, 2025September 11, 20250.836,170
Total$3.32$24,678
Fiscal Year 2024
September 19, 2023November 16, 2023December 14, 2023$0.75$5,574
November 28, 2023February 15, 2024March 14, 20240.755,573
March 12, 2024May 16, 2024June 13, 20240.755,574
June 12, 2024August 15, 2024September 12, 20240.755,574
Total$3.00$22,295
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-58 AOCI Rollforward.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-58 AOCI Rollforward.html new file mode 100644 index 000000000..fc748dc40 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-58 AOCI Rollforward.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Derivatives
Balance, beginning of period$(3)$(27)$(13)
Unrealized gains (losses), net of tax of $20, $(4), and $977(14)34
Reclassification adjustments for (gains) losses included in other income
(expense), net
(103)48(61)
Tax expense (benefit) included in provision for income taxes21(10)13
Amounts reclassified from accumulated other comprehensive loss(82)38(48)
Net change related to derivatives, net of tax of $(1), $6, and $(4)(5)24(14)
Balance, end of period$(8)$(3)$(27)
Investments
Balance, beginning of period$(2,625)$(3,582)$(2,138)
Unrealized gains (losses), net of tax of $411, $247, and $(393)1,560915(1,523)
Reclassification adjustments for losses included in other income
(expense), net
175399
Tax benefit included in provision for income taxes(3)(11)(20)
Amounts reclassified from accumulated other comprehensive loss144279
Net change related to investments, net of tax of $414, $258, and $(373)1,574957(1,444)
Balance, end of period$(1,051)$(2,625)$(3,582)
Translation Adjustments and Other
Balance, beginning of period$(2,962)$(2,734)$(2,527)
Translation adjustments and other, net of tax of $8, $0, and $0674(228)(207)
Balance, end of period$(2,288)$(2,962)$(2,734)
Accumulated other comprehensive loss, end of period$(3,347)$(5,590)$(6,343)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-59 Stock Compensation.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-59 Stock Compensation.html new file mode 100644 index 000000000..dcd465207 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-59 Stock Compensation.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Stock-based compensation expense$11,974$10,734$9,611
Income tax benefits related to stock-based compensation2,0271,8261,651
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-6 Microsoft Revenue.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-6 Microsoft Revenue.html new file mode 100644 index 000000000..fe8c56c73 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-6 Microsoft Revenue.html @@ -0,0 +1 @@ +
Windows OEM and Devices revenue growthRevenue from sales of Windows Pro and non-Pro licenses sold through the OEM channel and sales of first-party Devices, including Surface and PC accessories
Xbox content and services revenue growthRevenue from Xbox content and services, comprising first- and third-party content (including games and in-game content), Xbox Game Pass and other subscriptions, Xbox Cloud Gaming, advertising, and other cloud services
Search and news advertising revenue (ex TAC) growthRevenue from search and news advertising excluding traffic acquisition costs (“TAC”) paid to Bing Ads network publishers and news partners
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-60 Dividends & Rates.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-60 Dividends & Rates.html new file mode 100644 index 000000000..f0a18a757 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-60 Dividends & Rates.html @@ -0,0 +1 @@ +
Year Ended June 30,202520242023
Dividends per share (quarterly amounts)$0.75 –  0.83$0.68 –  0.75$0.62 –  0.68
Interest rates3.4% – 5.5%3.8% – 5.6%2.0% – 5.4%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-61 Stock Awards.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-61 Stock Awards.html new file mode 100644 index 000000000..11dd03991 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-61 Stock Awards.html @@ -0,0 +1 @@ +
SharesWeighted Average
Grant-Date Fair Value
(In millions)
Stock Awards
Nonvested balance, beginning of year88$292.28
Granted (a)39413.90
Vested(38)293.25
Forfeited(7)317.23
Nonvested balance, end of year82$347.44
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-62 Share Repurchase.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-62 Share Repurchase.html new file mode 100644 index 000000000..53691c34e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-62 Share Repurchase.html @@ -0,0 +1 @@ +
(Shares in millions)
Year Ended June 30,202520242023
Shares purchased667
Average price per share$385.10$339.46$245.59
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-63 Microsoft Segment Results.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-63 Microsoft Segment Results.html new file mode 100644 index 000000000..5e2cbdf5e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-63 Microsoft Segment Results.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Productivity and Business Processes
Revenue$120,810$106,820$94,151
Cost of revenue22,42219,61117,202
Operating expenses28,61527,54826,875
Operating Income$69,773$59,661$50,074
Intelligent Cloud
Revenue$106,265$87,464$72,944
Cost of revenue40,17129,61124,109
Operating expenses21,50520,04020,424
Operating Income$44,589$37,813$28,411
More Personal Computing
Revenue$54,649$50,838$44,820
Cost of revenue25,23824,89224,552
Operating expenses15,24513,98710,230
Operating Income$14,166$11,959$10,038
Total
Revenue$281,724$245,122$211,915
Cost of revenue87,83174,11465,863
Operating expenses65,36561,57557,529
Operating Income$128,528$109,433$88,523
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-64 Revenue by Region.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-64 Revenue by Region.html new file mode 100644 index 000000000..af25a9250 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-64 Revenue by Region.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
United States (a)$144,546$124,704$106,744
Other countries137,178120,418105,171
Total$281,724$245,122$211,915
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-65 Microsoft Revenue.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-65 Microsoft Revenue.html new file mode 100644 index 000000000..3177208a4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-65 Microsoft Revenue.html @@ -0,0 +1 @@ +
(In millions)
Year Ended June 30,202520242023
Server products and cloud services$98,435$79,828$65,007
Microsoft 365 Commercial products and cloud services87,76776,96966,949
Gaming23,45521,50315,466
LinkedIn17,81216,37214,989
Windows and Devices17,31417,02617,147
Search and news advertising13,87812,30612,125
Dynamics products and cloud services7,8276,8315,796
Enterprise and partner services7,7607,5947,900
Microsoft 365 Consumer products and cloud services7,4046,6486,417
Other7245119
Total$281,724$245,122$211,915
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-66 Revenue by Region.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-66 Revenue by Region.html new file mode 100644 index 000000000..11b53e966 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-66 Revenue by Region.html @@ -0,0 +1 @@ +
(In millions)
June 30,202520242023
United States$230,069$186,106$114,380
Other countries141,833115,26372,859
Total$371,902$301,369$187,239
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-67 Deloitte Signature.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-67 Deloitte Signature.html new file mode 100644 index 000000000..20b07ca86 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-67 Deloitte Signature.html @@ -0,0 +1 @@ +
/S/ DELOITTE & TOUCHE LLP
Seattle, Washington
July 30, 2025
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-68 Microsoft Board.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-68 Microsoft Board.html new file mode 100644 index 000000000..e62c1b1b8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-68 Microsoft Board.html @@ -0,0 +1 @@ +
Satya Nadella
Chairman and Chief Executive Officer,
Microsoft Corporation
Catherine MacGregor 2,4
Group Chief Executive Officer and
Director, Engie S.A.
Carlos Rodriguez 1, 2
Director, Automatic Data
Processing, Inc.
Reid G. Hoffman 4
Partner, Greylock Partners
Mark A. L. Mason 3
Chief Financial Officer, Citigroup Inc.
Charles W. Scharf 2,3
Chief Executive Officer, President,
and Director, Wells Fargo & Company
Hugh F. Johnston 1
Senior Executive Vice President and
Chief Financial Officer,
The Walt Disney Company
Sandra E. Peterson 2,3
Lead Independent Director,
Microsoft Corporation
Operating Partner, Clayton, Dubilier &
Rice, LLC
John W. Stanton 1,4
Founder and Chairman, Trilogy
Partnerships
Teri L. List 1,3
Former Executive Vice President and
Chief Financial Officer, The Gap, Inc.
Penny S. Pritzker 4
Founder and Chairman, PSP
Partners, LLC
Emma N. Walmsley 2,4
Chief Executive Officer and Director, GSK plc
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-69 Microsoft Executives.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-69 Microsoft Executives.html new file mode 100644 index 000000000..263fe36bf --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-69 Microsoft Executives.html @@ -0,0 +1 @@ +
Satya Nadella
Chairman and Chief Executive Officer
Amy E. Hood
Executive Vice President and Chief Financial Officer
Judson B. Althoff
Executive Vice President and CEO Microsoft Commercial
Takeshi Numoto
Executive Vice President and Chief Marketing Officer
Amy L. Coleman
Executive Vice President and Chief Human Resources Officer
Bradford L. Smith
Vice Chair and President
Kathleen T. Hogan
Executive Vice President, Office of Strategy and Transformation
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-7 Revenue Growth.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-7 Revenue Growth.html new file mode 100644 index 000000000..a7fadd7cf --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-7 Revenue Growth.html @@ -0,0 +1 @@ +
(In millions, except percentages and per share amounts)20252024Percentage
Change
Revenue$281,724$245,12215%
Gross margin193,893171,00813%
Operating income128,528109,43317%
Net income101,83288,13616%
Diluted earnings per share13.6411.8016%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-8 Microsoft Segment Results.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-8 Microsoft Segment Results.html new file mode 100644 index 000000000..0b8185c57 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-8 Microsoft Segment Results.html @@ -0,0 +1 @@ +
(In millions, except percentages)20252024Percentage
Change
Productivity and Business Processes
Revenue$120,810$106,82013%
Cost of revenue22,42219,61114%
Operating expenses28,61527,5484%
Operating Income$69,773$59,66117%
Intelligent Cloud
Revenue$106,265$87,46421%
Cost of revenue40,17129,61136%
Operating expenses21,50520,0407%
Operating Income$44,589$37,81318%
More Personal Computing
Revenue$54,649$50,8387%
Cost of revenue25,23824,8921%
Operating expenses15,24513,9879%
Operating Income$14,166$11,95918%
Total
Revenue$281,724$245,12215%
Cost of revenue87,83174,11419%
Operating expenses65,36561,5756%
Operating Income$128,528$109,43317%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-9 R&D Spending 2025.html b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-9 R&D Spending 2025.html new file mode 100644 index 000000000..2c51f9d5e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-microsoft-2025-annual-report/tables/table-9 R&D Spending 2025.html @@ -0,0 +1 @@ +
(In millions, except percentages)20252024Percentage
Change
Research and development$32,488$29,51010%
As a percent of revenue12%12%0ppt
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/chunks.json b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/chunks.json new file mode 100644 index 000000000..31f42872c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/chunks.json @@ -0,0 +1,39384 @@ +{ + "chunks": [ + { + "chunk_id": "c802bf1d-1f54-5641-9291-845298a9844e", + "type": "image", + "content": "\nNVIDIA Corporation
0\n[images/image-1-NVIDIA Logo.jpg]\n", + "path": "images/image-1-NVIDIA Logo.jpg", + "metadata": { + "length": 63, + "summary": "image-1\n- Subject: 3D rendering of the NVIDIA corporate logo (green eye symbol)\n- Visual Style: High-gloss, metallic finish with a vibrant lime green color\n- Background: White geometric grid pattern composed of triangular facets\n- Perspective: Angled, close-up view emphasizing depth and texture", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-1-NVIDIA Logo.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "7d72c6f9-d7e5-5902-8ae9-d82879f2d987", + "type": "image", + "content": "\nGeneral-purpose computing has run its course. Moore's Law slowed. Dennard scaling ended. Accelerated computing became foundational.
0\n[images/image-2-AI Technology Stack.jpg]\n", + "path": "images/image-2-AI Technology Stack.jpg", + "metadata": { + "length": 176, + "summary": "image-2\n- Applications\n- Models\n- Infrastructure\n- Chips\n- Energy", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-2-AI Technology Stack.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "7a05906e-9a35-58f6-b3e0-261534609b91", + "type": "image", + "content": "\nEINFRASTRUCTURE] --> F CHATBOTS0\n[images/image-3-EINFRASTRU.jpg]\n", + "path": "images/image-3-EINFRASTRU.jpg", + "metadata": { + "length": 66, + "summary": "image-3\nEINFRASTRUCTURE] --> F CHATBOTS0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-3-EINFRASTRU.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "401d5609-6dc4-5055-84ef-273c4be183e4", + "type": "image", + "content": "\nENERGY --> H DIGITAL BIOLOGY0\n[images/image-4-ENERGY --_.jpg]\n", + "path": "images/image-4-ENERGY --_.jpg", + "metadata": { + "length": 63, + "summary": "image-4\nENERGY --> H DIGITAL BIOLOGY0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-4-ENERGY --_.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ddde7898-41ee-5425-9f32-0e0cc32cda1c", + "type": "image", + "content": "\n
ROBOTAXI0\n[images/image-5-__details_.jpg]\n", + "path": "images/image-5-__details_.jpg", + "metadata": { + "length": 54, + "summary": "image-5\n
ROBOTAXI0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-5-__details_.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ed17a528-af8b-5fab-acef-15f743819928", + "type": "image", + "content": "\nCHATBOTS ENTERPRISE0\n[images/image-6-CHATBOTS E.jpg]\n", + "path": "images/image-6-CHATBOTS E.jpg", + "metadata": { + "length": 54, + "summary": "image-6\nCHATBOTS ENTERPRISE0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-6-CHATBOTS E.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "40d2a04c-e080-5d69-aacf-bbdef64c3747", + "type": "image", + "content": "\nROBOTAXI SCIENCE0\n[images/image-7-ROBOTAXI S.jpg]\n", + "path": "images/image-7-ROBOTAXI S.jpg", + "metadata": { + "length": 51, + "summary": "image-7\nROBOTAXI SCIENCE0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-7-ROBOTAXI S.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d50f0edb-f21d-5dd9-a58e-a8b4a7e8a199", + "type": "image", + "content": "\nENTERPRISE ROBOTICS0\n[images/image-8-ENTERPRISE.jpg]\n", + "path": "images/image-8-ENTERPRISE.jpg", + "metadata": { + "length": 54, + "summary": "image-8\nENTERPRISE ROBOTICS0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-8-ENTERPRISE.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "4787df69-198f-5be7-91ee-c874d0b00eba", + "type": "image", + "content": "\nAI AGENTS MANUFACTURING0\n[images/image-9-AI AGENTS .jpg]\n", + "path": "images/image-9-AI AGENTS .jpg", + "metadata": { + "length": 58, + "summary": "image-9\nAI AGENTS MANUFACTURING0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-9-AI AGENTS .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "df55c5af-3f58-552e-b0e5-e5feb62c93ea", + "type": "image", + "content": "\nSCIENCE AI CODER0\n[images/image-10-SCIENCE AI.jpg]\n", + "path": "images/image-10-SCIENCE AI.jpg", + "metadata": { + "length": 52, + "summary": "image-10\nSCIENCE AI CODER0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-10-SCIENCE AI.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d774d9fe-826e-5407-8017-f5108240b1a9", + "type": "image", + "content": "\nROBOTICS
0\n[images/image-11-Acronyms in AI and Machine Learning.jpg]\n", + "path": "images/image-11-Acronyms in AI and Machine Learning.jpg", + "metadata": { + "length": 54, + "summary": "image-11\nLLM (Large Language Model)\nVLM (Vision Language Model)\nVLA (Vision-Language-Action)\nMMLLM (Multimodal Large Language Model)\nGPT (Generative Pre-trained Transformer)\nDM (Diffusion Model)\nGNN (Graph Neural Network)\nMOE (Mixture of Experts)\nSSM (State Space Model)\nLBM (Lattice Boltzmann Method)", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-11-Acronyms in AI and Machine Learning.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d56a4f93-2adf-51c3-8390-5a4cf1477435", + "type": "image", + "content": "\nMOE
0\n[images/image-12-Stacked Server Racks.jpg]\n", + "path": "images/image-12-Stacked Server Racks.jpg", + "metadata": { + "length": 49, + "summary": "image-12\n- Three-tier arrangement of server racks or storage units\n- Bottom tier: 12 visible units with illuminated vertical panels\n- Middle tier: 8 visible units, slightly offset from bottom row\n- Top tier: 6 visible units, centered above middle row\n- Units feature horizontal ventilation slots and glowing amber/yellow indicators\n- Dark background with white outline sketch style", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-12-Stacked Server Racks.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "dc6bc028-50f7-5a52-9ccd-34791f5c9152", + "type": "image", + "content": "\nEnergy becomes computation. Computation becomes intelligence. The amount of intelligence a system can produce is bounded by energy.
0\n[images/image-13-DSX Architecture Overview.jpg]\n", + "path": "images/image-13-DSX Architecture Overview.jpg", + "metadata": { + "length": 177, + "summary": "image-13\n- **Reference Designs, Best Practices, Methodologies**: Includes Rubin DSX Reference Designs, DSX Ecosystem, Omniverse DSX Blueprint, DSX Max-Q, DSX Flex, DSX Exchange, and DSX Sim.\n- **Chips, Systems, Facilities**: Features Power and Liquid Cooling components.\n- **DSX Libraries, APIs, Software**: Integrated with the architecture.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-13-DSX Architecture Overview.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e52b445e-e2fa-5846-b89f-ab6ea829a71c", + "type": "image", + "content": "\nFor agentic AI—low-latency, long-context, multi-step reasoning—throughput per megawatt determines cost per token.
0\n[images/image-14-NVIDIA CEO InferenceMAX King.jpg]\n", + "path": "images/image-14-NVIDIA CEO InferenceMAX King.jpg", + "metadata": { + "length": 159, + "summary": "image-14\n- Subject: Jensen Huang (CEO of NVIDIA) standing on stage with arms raised in a victory pose.\n- Attire: Black leather jacket, black t-shirt, black pants, and glasses.\n- Background: A large digital display showing a championship belt graphic.\n- Belt Details:\n - Central Plate: Features the text \"semianalysis\", \"InferenceMAX\", and \"KING\" with a crown icon above it.\n - Left Plate: Displays the NVIDIA logo.\n - Right Plate: Displays a logo resembling a lion's head (likely representing Semianalysis).", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-14-NVIDIA CEO InferenceMAX King.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f10168a2-7335-546e-a811-3049f7476c6d", + "type": "image", + "content": "\nOpen models have reached the frontier.
0\n[images/image-15-NVIDIA Leads Open-Source AI.jpg]\n", + "path": "images/image-15-NVIDIA Leads Open-Source AI.jpg", + "metadata": { + "length": 84, + "summary": "image-15\nChart Title: NVIDIA is World's Largest Contributor to Open-Source AI\nY-Axis Label: Repos (Repository count)\nX-Axis Time Range: Jan 2025 to Feb 2026\n\nKey Data Points & Rankings:\n1. NVIDIA: Highest contributor, reaching over 400 repos by Feb 2026.\n2. Alibaba Cloud: Second highest, reaching approximately 350 repos.\n3. Hugging Face: Third highest, reaching approximately 200 repos.\n4. Ai2: Fourth highest, reaching approximately 200 repos.\n5. Meta: Reaching approximately 150 repos.\n6. Google: Reaching approximately 120 repos.\n7. IBM: Reaching approximately 110 repos.\n8. ByteDance: Reaching approximately 100 repos.\n9. Baidu: Reaching approximately 90 repos.\n10. Tencent: Reaching approximately 80 repos.\n11. Microsoft: Reaching approximately 80 repos.\n12. Z", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-15-NVIDIA Leads Open-Source AI.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "38097bff-717c-54e9-8afd-4460c4bbd5be", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
MonthNVIDIAAlibaba CloudHugging FaceAi2MetaGoogleIBMByteDanceBaiduTencentMicrosoftZai
Jan 2025~10~10~10~10~10~10~10~10~10~10~10~10
Feb 2026~450~350~200~180~150~130~120~110~100~90~80~70
", + "path": "tables/table-0 GPU Cloud Growth.html", + "metadata": { + "length": 977, + "summary": "table-1\nThe table shows projected GPU/cloud service growth from Jan 2025 to Feb 2026, with NVIDIA leading at ~450 units, followed by Alibaba Cloud at ~350 and Hugging Face at ~200.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-0 GPU Cloud Growth.html", + "keywords": [ + "NVIDIA", + "Alibaba Cloud", + "Hugging Face" + ], + "tokens": [] + } + }, + { + "chunk_id": "40c19c19-1b43-5d78-8ed0-029ea1b798b0", + "type": "image", + "content": "\n
0\n[images/image-16-Collection of decorative enamel pins.jpg]\n", + "path": "images/image-16-Collection of decorative enamel pins.jpg", + "metadata": { + "length": 56, + "summary": "image-16\n- DNA double helix structure with chemical formula labels (H3C, CH3, NH2)\n- Globe showing North and South America\n- Cartoon character waving (wearing glasses and jacket)\n- Spiral galaxy or atom model design\n- Red and gold robot figure waving\n- Front view of a car with blue and yellow accents", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-16-Collection of decorative enamel pins.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "8e52c8e9-8cc5-5a07-b189-931f2e5f182c", + "type": "image", + "content": "\n
0\n[images/image-17-Nemoclaw Agentic AI Architecture.jpg]\n", + "path": "images/image-17-Nemoclaw Agentic AI Architecture.jpg", + "metadata": { + "length": 56, + "summary": "image-17\nCentral Node: Nemoclaw (Agentic AI)\n\nInput/Interaction Layer:\n- Multi-modal Prompt\n- Files (Structured, Unstructured)\n- Computer Use\n- Tools (CLI, MCP)\n- OpenShell\n\nCore Components & Connections:\n- Memory\n- LLM (Large Language Model)\n- Sub-Agents\n- Skills\n\nLeft-Side Infrastructure (Hardware/Data):\n- cuDF, cuVS\n- vGPU\n\nRight-Side Software Stack:\n- Nemotron, Nemo, Dynamo, NIM\n- AI-Q\n- cuOPT", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-17-Nemoclaw Agentic AI Architecture.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "1f67a0aa-68b6-54fe-b597-ba2b4d9ddbf6", + "type": "image", + "content": "\nRobotics, autonomous vehicles, and intelligent machines combine perception and reasoning with an understanding of physics.
0\n[images/image-18-Future Robotics Exhibition.jpg]\n", + "path": "images/image-18-Future Robotics Exhibition.jpg", + "metadata": { + "length": 168, + "summary": "image-18\nA panoramic view of a high-tech exhibition hall featuring an extensive collection of advanced robotics and machinery.\n\n**Heavy Machinery & Industrial Equipment:**\n- A large yellow CAT excavator on the far left.\n- Multiple industrial robotic arms, including white, yellow, and orange articulated arms on the right side.\n\n**Humanoid Robots:**\n- A diverse array of humanoid robots displayed throughout the scene:\n - Several standing on the upper balcony level.\n - Various models (silver, white, black) arranged on the main floor stage.\n - Smaller units and service robots interspersed among the larger figures.\n - One distinct orange humanoid robot near the center-right foreground.\n\n**Vehicles & Drones:**\n- Three modern luxury cars parked on the right side (Silver Mercedes-Benz, Grey sports car, Blue sedan).\n- A small drone flying in the central area above the stage.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-18-Future Robotics Exhibition.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "1cf6525f-8f8a-5aa4-8317-41c0a045570d", + "type": "image", + "content": "\nComputer graphics is no longer limited by traditional rendering.
0\n[images/image-19-Signature.jpg]\n", + "path": "images/image-19-Signature.jpg", + "metadata": { + "length": 110, + "summary": "image-19\nJonathan King", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-19-Signature.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "0d23efd3-9ecf-5549-a307-eb234859c9b2", + "type": "image", + "content": "\n
0\n[images/image-20-Jensen Huang Presentation.jpg]\n", + "path": "images/image-20-Jensen Huang Presentation.jpg", + "metadata": { + "length": 55, + "summary": "image-20\n- Subject: Jensen Huang (CEO of NVIDIA)\n- Attire: Signature black leather jacket, glasses, dark trousers\n- Action: Gesturing with both hands while holding a presentation clicker/remote\n- Context: Appears to be speaking on stage during a keynote or product presentation against a dark background", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-20-Jensen Huang Presentation.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "356c6883-2d99-56e9-8bac-bf484a1ec2ff", + "type": "image", + "content": "\nAgentic AI has arrived and is delivering remarkable results. Generative AI creates. Reasoning AI thinks through problems. Agentic AI uses tools to perform tasks after thoughtful planning, research, and learning. AI has crossed an important threshold: it no longer simply answers questions—it does useful work.
0\n[images/image-21-Poster Gallery Event.jpg]\n", + "path": "images/image-21-Poster Gallery Event.jpg", + "metadata": { + "length": 355, + "summary": "image-21\nLocation: Jose Civic\nSignage: \"Poster Gallery\", \"Reimagine What's Possible\"\nScene: Outdoor gathering with a crowd of people in front of a green exhibition structure.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-21-Poster Gallery Event.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "9fc7b388-4417-5637-8671-02c354c96b19", + "type": "image", + "content": "\nEvery successful AI application pulls on every layer beneath it—from energy to chips to infrastructure to models. Applications are where AI infrastructure becomes productivity, growth, and national capability.
0\n[images/image-22-Inference At-Scale is Extreme Computing.jpg]\n", + "path": "images/image-22-Inference At-Scale is Extreme Computing.jpg", + "metadata": { + "length": 255, + "summary": "image-22\n- Chart Title: Inference At-Scale is Extreme Computing\n- Y-Axis Label: Throughput (Tokens per Second For Factory)\n- X-Axis Label: Smart AI Fast Response (Tokens per Second For Cloud)\n- Key Data Points & Trends:\n - Revenue Curve: Shows a high-throughput curve where revenue increases with throughput before declining.\n - FLOPS Factors: An arrow points to the curve labeled \"FLOPS, HBM Memory, HBM Bandwidth, Software and Architecture,\" indicating these are key drivers for performance.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-22-Inference At-Scale is Extreme Computing.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "1908a1b6-46b8-5f35-a5d2-1704c307d0d0", + "type": "image", + "content": "\nFrontier model services are excellent for general intelligence use cases. They work out of the box and are continuously getting smarter. At NVIDIA, we use OpenAI Codex, Anthropic Claude Code, and Cursor.
0\n[images/image-23-NVIDIA GTC Event Scene.jpg]\n", + "path": "images/image-23-NVIDIA GTC Event Scene.jpg", + "metadata": { + "length": 249, + "summary": "image-23\n- Primary Subject: Outdoor scene at the NVIDIA GTC (GPU Technology Conference) event featuring attendees and autonomous mobile robots.\n- Key Objects: Two small, black, wheeled robots are visible in the foreground/mid-ground. One robot has a sign reading \"Welcome to NVIDIA GTC\" with the NVIDIA logo. The other robot has a tall green banner attached with the NVIDIA logo and a QR code.\n- People: Several individuals are present, wearing conference lanyards (green). Some are looking at phones, while others appear to be interacting with or observing the robots.\n- Background/Context: The setting is an outdoor plaza or park area with trees, white tents, and string lights in the background, suggesting an evening or late afternoon atmosphere.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-23-NVIDIA GTC Event Scene.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "0a66d713-0c32-507c-a97f-2c2744b24e39", + "type": "image", + "content": "\nThe next wave of AI is physical. AI is moving from screens into factories, warehouses, hospitals, vehicles, and robots. To operate in the real world, AI must understand geometry, motion, causality, and physics. It must reason and act safely in dynamic, unpredictable environments.
0\n[images/image-24-Silhouette of Man Gazing at Earth.jpg]\n", + "path": "images/image-24-Silhouette of Man Gazing at Earth.jpg", + "metadata": { + "length": 326, + "summary": "image-24\n- Primary Subject: A silhouette of a man standing in profile on the left side, looking out towards a large display.\n- Scene Description: The background is a massive screen or projection showing a view of planet Earth from space.\n- Visual Details: The image depicts the curvature of the Earth with blue oceans and white cloud formations. There are patches of yellow and green overlaying the clouds, suggesting weather data or satellite imagery. The top left corner shows the blackness of space and the atmospheric glow along the horizon.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-24-Silhouette of Man Gazing at Earth.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "881cafa8-719a-5a87-966e-32b6704576ff", + "type": "image", + "content": "\nVirtual meeting admission: We will be holding the 2026 Meeting virtually at the location listed above. To participate, you will need the Control Number included on your notice of proxy materials or printed proxy card. Timothy S. Teter0\n[images/image-25-Virtual me.jpg]\n", + "path": "images/image-25-Virtual me.jpg", + "metadata": { + "length": 270, + "summary": "image-25\nVirtual meeting admission: We will be holding the 2026 Meeting virtually at the location listed above. To participate, you will need the Control Number included on your notice of proxy materials or printed proxy card. Timothy S. Teter0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-25-Virtual me.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "805808ab-5e92-5b33-8a17-5c146ec3269f", + "type": "table", + "content": "
2007 PlanNVIDIA Corporation Amended and Restated 2007 Equity Incentive Plan
ACAudit Committee of the Board
ASC 718Financial Accounting Standards Board Accounting Standards Codification Topic 718: Compensation - Stock Compensation
Base Compensation PlanPerformance goal necessary to earn the target award under the Variable Cash Plan and for the target numbers of SY PSUs and MY PSUs to become eligible to vest
BoardThe Company's board of directors
BylawsThe Company's Amended and Restated Bylaws
CAP"Compensation actually paid," as defined under Item 402(v) of Regulation S-K
CCCompensation Committee of the Board
CD&ACompensation Discussion and Analysis
CEOChief Executive Officer
CFOChief Financial Officer
CharterThe Company's Restated Certificate of Incorporation
Control NumberIdentification number for each stockholder included in Notice or proxy card
CSCorporate Sustainability
CSSCCorporate Sustainability Steering Committee
DEIDiversity, Equity, and Inclusion
ERMEnterprise Risk Management
ESPPNVIDIA Corporation Amended and Restated 2012 Employee Stock Purchase Plan
EVPExecutive Vice President
Exchange ActSecurities Exchange Act of 1934, as amended
Fiscal 20__The Company's fiscal year ended on the last Sunday in January of the stated year
Form 10-KThe Company's Annual Report on Form 10-K for Fiscal 2026 filed with the SEC on February 25, 2026
GAAPGenerally accepted accounting principles in the United States
GHGGreenhouse gas
Internal Revenue CodeU.S. Internal Revenue Code of 1986, as amended
Lead DirectorLead independent director
MeetingAnnual Meeting of Stockholders
MY PSUsMulti-year PSUs based on 3-year TSR relative to the S&P 500 with a three-year performance metric, vesting after three years
NasdaqThe Nasdaq Stock Market LLC
NCGCNominating and Corporate Governance Committee of the Board
NEOsNamed Executive Officers consisting of our CEO, our CFO, and our other three most highly compensated executive officers as of the end of Fiscal 2026
Non-GAAP Operating IncomeGAAP operating income, as the Company reports in its SEC filings, excluding stock-based compensation expense, acquisition-related and other costs, and other. Please see Reconciliation of Non-GAAP Financial Measures in our CD&A for a reconciliation between the non-GAAP financial measures and GAAP results
NoticeNotice of Internet Availability of Proxy Materials
NVIDIA, Company, we, us, ourNVIDIA Corporation, a Delaware corporation
NYSENew York Stock Exchange
PSUPerformance stock unit
PwCPricewaterhouseCoopers LLP
RSURestricted stock unit
S&P 500Standard & Poor's 500 Composite Index
SECU.S. Securities and Exchange Commission
Section 162(m)Section 162(m) of the Internal Revenue Code
Securities ActSecurities Act of 1933, as amended
Stretch Compensation PlanPerformance goal necessary to earn the maximum award under the Variable Cash Plan and for the maximum numbers of SY PSUs and MY PSUs to become eligible to vest
SY PSUsPSUs based on annual Non-GAAP Operating Income performance with a single-year performance metric, vesting over four years
ThresholdMinimum performance goal necessary to earn an award under the Variable Cash Plan and for SY PSUs and MY PSUs to become eligible to vest
TSRTotal shareholder return
Variable Cash PlanThe Company's variable cash compensation plan
", + "path": "tables/table-1 NVIDIA Definitions.html", + "metadata": { + "length": 4584, + "summary": "table-2\nA table of defined terms and acronyms used in NVIDIA's proxy materials, covering compensation plans, committees, and financial metrics.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-1 NVIDIA Definitions.html", + "keywords": [ + "Definitions", + "Acronyms", + "Compensation" + ], + "tokens": [] + } + }, + { + "chunk_id": "53e91c02-007f-5a5a-9c31-a6924ad6dcc3", + "type": "text", + "content": "2026\nNVIDIA Corporation\nAnnual Review\nNotice of Annual Meeting\nProxy Statement\nForm 10-K\n\nNVIDIA Corporation
0\n[images/image-1-NVIDIA Logo.jpg]\n\n
\nnatural_image\nAbstract 3D geometric design with green spiral and rectangular forms on a grid background (no text or symbols)\n
\nFor most of computing history, software was pre-recorded.\nHumans described an algorithm. Computers executed it.\nNow software is trained. It learns from data and generates intelligence in real time.\nGeneral-purpose computing has run its course. Moore's Law slowed. Dennard scaling ended. Accelerated computing became foundational.\nBecause intelligence is produced in real time, the entire computing stack beneath it had to be reinvented.\nAI runs on real hardware, real energy, and real economics. It takes raw materials and converts them into intelligence at scale.\nThis is not a normal upgrade cycle.\nThe world is going through a platform shift.\n\nGeneral-purpose computing has run its course. Moore's Law slowed. Dennard scaling ended. Accelerated computing became foundational.
0\n[images/image-2-AI Technology Stack.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"APPLICATIONS\"] --> B\nC[\"MODELS\"] --> D\nEINFRASTRUCTURE] --> F\nCHIPS --> G\nENERGY --> H\n```\n
\n\nEINFRASTRUCTURE] --> F CHATBOTS0\n[images/image-3-EINFRASTRU.jpg]\n\nCHATBOTS\n\nENERGY --> H DIGITAL BIOLOGY0\n[images/image-4-ENERGY --_.jpg]\n\nDIGITAL BIOLOGY\n\n
ROBOTAXI0\n[images/image-5-__details_.jpg]\n\nROBOTAXI\n\nCHATBOTS ENTERPRISE0\n[images/image-6-CHATBOTS E.jpg]\n\nENTERPRISE\nAI AGENTS\n\nROBOTAXI SCIENCE0\n[images/image-7-ROBOTAXI S.jpg]\n\nSCIENCE\n\nENTERPRISE ROBOTICS0\n[images/image-8-ENTERPRISE.jpg]\n\nROBOTICS\n\nAI AGENTS MANUFACTURING0\n[images/image-9-AI AGENTS .jpg]\n\nMANUFACTURING\n\nSCIENCE AI CODER0\n[images/image-10-SCIENCE AI.jpg]\n\nAI CODER\n\nROBOTICS
0\n[images/image-11-Acronyms in AI and Machine Learning.jpg]\n\n
\ntext_image\nLLM\nVLM\nVLA\nMMLLM\nGPT\nDM\nGNN\nMOE\nSSM\nLBM\n
\nAI FACTORIES\n\nMOE
0\n[images/image-12-Stacked Server Racks.jpg]\n\n
\nnatural_image\nStacked cylindrical containers with transparent lids, arranged in two rows (no visible text or symbols)\n
\nAI Is a Five-Layer Cake\nIt is not a clever app or a single model; it is essential infrastructure, like electricity and the internet.\nWhen you look at AI industrially, it resolves into a five-layer stack. Energy → chips and systems → infrastructure → models → applications.\nEvery successful application pulls on every layer beneath it, all the way down to the power plant that powers it.\nEnergy becomes computation. Computation becomes intelligence. The amount of intelligence a system can produce is bounded by energy.\nWhen applications scale, the entire stack must scale with them.\nAI Is Driving the Largest Infrastructure Buildout in Human History\nData centers are becoming AI factories.\nLand, power, cooling, networking, and systems must scale in parallel. Every company will use AI. Every country will build AI infrastructure. This is the next industrial revolution. We are only a few hundred billion dollars into it. Trillions more still need to be built.\n\nEnergy becomes computation. Computation becomes intelligence. The amount of intelligence a system can produce is bounded by energy.
0\n[images/image-13-DSX Architecture Overview.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"Reference Designs, Best Practices, Methodologies\"] --> B[\"Power\"]\nC[\"Chip, Systems, Facilities\"] --> D[\"Chips, Systems, Facilities\"]\nE[\"Rubin DSX Reference Designs\"] --> F[\"DSX Ecosystem\"]\nG[\"Omniverse DSX Blueprint\"] --> H[\"DSX Max-Q\"]\nI[\"DSX Flex\"] --> J[\"DSX Exchange\"]\nK[\"DSX Sim\"] --> L[\"DSX Libraries, APIs, Software\"]\nM[\"Liquid Cooling\"] --> N[\"Liquid Cooling\"]\n```\n
\nAI Factories Manufacture Intelligence\nIntelligence is now produced, not stored.\nModern data centers are AI factories, converting energy into tokens—just as power plants convert energy into electricity.\nThroughput per megawatt defines production. Cost per token defines efficiency.\nToken output and performance per watt directly determine revenue.\nInference Defines the Economics of AI\nFor agentic AI—low-latency, long-context, multi-step reasoning—throughput per megawatt determines cost per token.\nInference runs continuously and directly drives revenue.\nReasoning models generate dramatically more tokens. They generate many more “thinking” tokens before they answer, and they have to do it fast enough to stay interactive.\nSemiAnalysis InferenceX measured GB300 NVL72 delivering up to 50x higher throughput per megawatt and up to 35x lower cost per token compared to Hopper, with the greatest gains in low-latency workloads.\nInference is not a chip problem. It is a systems architecture problem—GPU, networking, memory, and software designed as one.\n\nFor agentic AI—low-latency, long-context, multi-step reasoning—throughput per megawatt determines cost per token.
0\n[images/image-14-NVIDIA CEO InferenceMAX King.jpg]\n\n
\ntext_image\nsemianalysis\nInferenceMAX\nKING\n
\nOpen Models Power the Ecosystem\nOpen models have reached the frontier.\nResearchers, startups, enterprises, and entire nations rely on open models to participate in advanced AI. Most of the world's models are free.\nAt this level of capability, they do not just change software. They activate demand across the entire stack.\nOpen models lower barriers to entry. They enable deep customization. They allow teams to build domain-specific systems on common platforms.\nAs developers build, customize, and deploy intelligence, token generation grows—and demand for AI infrastructure grows with it.\n\nOpen models have reached the frontier.
0\n[images/image-15-NVIDIA Leads Open-Source AI.jpg]\n\n
\nline chart\n\n[tables/table-0 GPU Cloud Growth.html]\n\n
\n\n
0\n[images/image-16-Collection of decorative enamel pins.jpg]\n\n
\nnatural_image\nSix-panel illustration showing molecular structures, a globe, a person waving, a spiral, a humanoid robot waving, and a car (no text or symbols)\n
\nAgentic AI Enters Production\nAI now perceives, reasons, plans, and acts.\nAgentic systems break complex problems into steps. They evaluate alternatives. They use tools. They retrieve information. They act across long-running workflows. They coordinate across models, tools, and environments in real time.\nOpenClaw showed what happens when agentic AI becomes open: every developer can build agents that put AI to work. As agents move from chat to work, they will access data, use tools, execute code, and communicate with other systems. All of this must happen securely, privately, and under policy control.\n\n
0\n[images/image-17-Nemoclaw Agentic AI Architecture.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"vGPU\"] -->|FILES STRUCTURED, UNSTRUCTURED| B[\"Computer Use\"]\nB --> C[\"NEMOCLAW\"]\nC --> D[\"MEMORY\"]\nD --> E[\"LLM\"]\nE --> F[\"Sub-agents\"]\nF --> G[\"AI-Q\"]\nG --> H[\"cuOPT\"]\nH --> I[\"SKILLS\"]\nI --> J[\"NEMOTRON NEMO DYNAMO NIM\"]\nJ --> K[\"AGentic AI is moving into enterprise workflows and crit\"]\nC --> L[\"TOOLS CLI, MCP\"]\nL --> M[\"OpenShell\"]\nM --> C\nstyle A fill:#99ccff,stroke:#333\nstyle B fill:#99ccff,stroke:#333\nstyle C fill:#f0f0f0,stroke:#333\nstyle D fill:#e0e0e0,stroke:#333\nstyle E fill:#e0e0e0,stroke:#333\nstyle F fill:#e0e0e0,stroke:#333\nstyle G fill:#e0e0e0,stroke:#333\nstyle H fill:#e0e0e0,stroke:#333\nstyle I fill:#d0d0d0,stroke:#333\nstyle J fill:#d0d0d0,stroke:#333\nstyle K fill:#d0d0d0,stroke:#333\nstyle L fill:#d0d0d0,stroke:#333\nstyle M fill:#d0d0d0,stroke:#333\nstyle N fill:#d0d0d0,stroke:#333\n```\n
\nAgentic AI is moving into enterprise workflows and critical systems.\nPhysical AI Puts Intelligence in Motion\nRobotics, autonomous vehicles, and intelligent machines combine perception and reasoning with an understanding of physics.\nThey operate in dynamic environments and act in the physical world.\nFoundation models are trained at scale, refined in simulation, and deployed at the edge—creating a continuous loop between virtual training and real-world experience.\nPhysical AI is transforming factories, warehouses, transportation, and healthcare.\nEverything that moves will be autonomous.\n\nRobotics, autonomous vehicles, and intelligent machines combine perception and reasoning with an understanding of physics.
0\n[images/image-18-Future Robotics Exhibition.jpg]\n\n
\nnatural_image\nInterior of a modern exhibition space featuring a large excavator, humanoid robots, and a car with a child in an orange costume (no visible text or symbols)\n
\nAI Reinvents Computer Graphics\nGeForce brought AI to the world. Now AI has come home to GeForce.\nComputer graphics is no longer limited by traditional rendering.\nWith neural rendering and DLSS, AI predicts pixels instead of computing them directly. For every pixel we render, AI generates the rest in real time.\nNeural networks now run alongside shaders and ray tracing cores. AI does not replace graphics. It enhances it.\nIn the era of neural rendering, graphics is learned.\n“AI has crossed the threshold from generating answers to performing work. The next industrial revolution will be built on factories that manufacture intelligence—and NVIDIA is building the platform to power them.”\n\nComputer graphics is no longer limited by traditional rendering.
0\n[images/image-19-Signature.jpg]\n\n
\ntext_image\nJahibing\n
\n\n
0\n[images/image-20-Jensen Huang Presentation.jpg]\n\n
\nnatural_image\nElderly man in a black leather jacket gesturing while holding a handheld device (no visible text or symbols)\n
\nDear NVIDIAs and Stakeholders,\nComputing platforms shift every 10 to 15 years—from mainframe to PC to internet to cloud to mobile-cloud.\nWhat is happening now is bigger. Computing is undergoing a complete reset not seen since IBM introduced modern computing with the System/360 in 1964. We are moving from computers as tools to a new kind of infrastructure that manufactures intelligence in real time.\nNVIDIA is driving this platform shift, building the computing infrastructure of the AI era.\nAI is transforming every layer of the computing stack—shifting from human coding to machine learning, from running code on CPUs to running neural networks on GPUs.\nAI is more than a model—it is a new computing paradigm. Intelligence, unlike information, cannot be prerecorded, stored, and retrieved. It is generated in real time, taking into account context, prior knowledge, and instructions.\nAI is still software running on chips. But because the computer can now understand, reason, plan, and use tools, it appears intelligent—hence, artificial intelligence.\nOver the past year, investors and enterprises questioned the ROI in AI—and for good reason. Early systems, such as chatbots, greatly improved access to information but did little productive work. Over the past six months, however, the landscape has shifted sharply. Enterprise adoption has surged.\nInference—running trained AI models to generate responses, reasoning, and actions—is driving cloud demand, and demand for GPU compute is accelerating.\nAgentic AI has arrived and is delivering remarkable results. Generative AI creates. Reasoning AI thinks through problems. Agentic AI uses tools to perform tasks after thoughtful planning, research, and learning. AI has crossed an important threshold: it no longer simply answers questions—it does useful work.\nThe first major enterprise breakout use case is AI coding agents, now sweeping across the industry. Within a few years, AI will generate much of the world's software. Soon after, AI will produce orders of magnitude more software than all human engineers combined. Limited only by imagination, the pace of innovation will accelerate across industries.\nSoftware can codify nearly all enterprise information. As a result, AI agents can accelerate or fully automate every business process through the software they generate. Organizations will rethink their operations end-to-end, from customer engagement through internal workflows and across the supply chain. Every business process will be enhanced and accelerated by AI agents.\nCompanies that adopt AI agents—and build the infrastructure to support them—will out-innovate, out-execute, and attract the world's best talent.\nThis shift is driving a fundamental change in the economics of computing.\n\nAgentic AI has arrived and is delivering remarkable results. Generative AI creates. Reasoning AI thinks through problems. Agentic AI uses tools to perform tasks after thoughtful planning, research, and learning. AI has crossed an important threshold: it no longer simply answers questions—it does useful work.
0\n[images/image-21-Poster Gallery Event.jpg]\n\n
\ntext_image\nReimagine What's Possible\nPoster Gallery\nReimagine What's Possible\nJOSÉ CIVIC\nBough Discoveries\n
\nCompute is no longer just a cost to support software; it is a productive asset that generates intelligence, productivity, and revenue. Compute is now the infrastructure of this new economy.\nA computer used to be a tool. In the age of AI, it is manufacturing equipment. AI requires a new kind of computing infrastructure that runs continuously. Traditional data centers were built to store and serve files. Today, they are AI factories—systems designed to turn energy into tokens, the basic units of digital intelligence, and tokens into answers, reasoning, actions, and work. AI factories are not cost centers—they are productive assets that generate revenue.\nNVIDIA delivers a full-stack AI computing platform that spans the entire AI lifecycle—from training to inference—and every domain of intelligence, from language and video to biology, autonomous driving, and robotics.\nNVIDIA is a vertically integrated, horizontally open platform. We architect with a vertically integrated mindset, executing extreme co-design across chips, systems, networking, storage, and software—spanning GPUs, CPUs, LPUs, DPUs, and full AI factory systems. At the same time, our platform remains open and integrates with the broadest software and hardware ecosystems.\nOur platform powers AI wherever it runs—across clouds, on-premises data centers, telecom networks, medical instruments, vehicles, and robots.\nNVIDIA is building the computing infrastructure of the AI era.\nDriving the AI Platform Shift\nGTC is where the industry comes together across every layer of the AI stack, from land, power, and infrastructure to chips, models, and applications. This year, what has become unmistakably clear is that AI has moved from experimentation to production.\nWe laid out the roadmap from Blackwell to Blackwell Ultra to Vera Rubin to Vera Rubin Ultra to Feynman. AI infrastructure builders plan over years, across factories, ecosystems, and supply chains.\nToday, we have visibility into more than \\$1 trillion in cumulative Blackwell and Rubin revenue from the start of 2025 through 2027. Since GTC DC in October 2025, we have received additional purchase orders and further diversified our customer base, including Anthropic, Meta, and a growing set of world-class model builders such as Cursor, Perplexity, Thinking Machines Lab, Reflection, and Skild.\nAt the same time, the scale of our cloud service providers continues to accelerate. In addition to the significant buildouts already underway at Azure and OCI, AWS and Google have announced plans for major NVIDIA GPU infrastructure expansions through 2027. NVIDIA AI cloud partners—including CoreWeave, Nebius, and Nscale—are also experiencing exponential demand and have secured significant long-term offtake agreements.\nThis momentum reflects the economics of this new industry: AI factories are revenue-generating infrastructure. They produce tokens, and tokens produce revenue.\nAI is a new industry: The Five-Layer Cake\nLike steam power, electricity, and information technology before it, AI is forming a new industry. AI is more than a model—it's a new industry with a five-layer stack: energy at the foundation; compute—chips and systems—above it; then infrastructure, models, and finally applications at the top.\nIn the industrial age, turbines and generators produced electricity, priced in dollars per kilowatt-hour, distributed globally, and enabled entirely new industries. In the AI era, electricity powers GPUs that produce digital intelligence, priced in dollars per million tokens—transforming existing industries and creating new ones. AI factories are not designed to store information, but to produce intelligence.\nEach industrial revolution created new infrastructure and companies. Power plants and utilities emerged in the age of electricity. Data centers, chips, and software companies defined the information technology era. Today, AI factories have emerged as the infrastructure of this new age, alongside entirely new classes of chips, systems, and computing platforms—and tens of thousands of AI-native startups.\nAbove the infrastructure are models. While large language models are widely known, the frontier extends much further. AI now includes models that understand biology, chemistry, physics, finance, medicine, and the physical world—bringing intelligence into the core of every major industry.\nAt the top are applications—where AI creates economic and social value—such as software development, customer support, drug discovery, industrial automation, autonomous vehicles, and robotics. In the years ahead, companies will orchestrate tens of billions of digital and physical agents working alongside people to help us be more productive, more creative, and more capable.\nEvery successful AI application pulls on every layer beneath it—from energy to chips to infrastructure to models. Applications are where AI infrastructure becomes productivity, growth, and national capability.\nFiscal 2026: A Defining Year\nRevenue grew 65 percent to \\$216 billion. Operating income rose 60 percent to \\$130 billion. Diluted EPS increased 67 percent to \\$4.90. We generated \\$103 billion in operating cash flow and returned \\$41 billion to shareholders.\nDemand remained strong across cloud providers, hyperscalers, AI model makers, enterprises, and sovereigns. AI infrastructure is now essential, and investment is accelerating worldwide.\nFiscal 2026 also marked a turning point in adoption. AI moved from experimentation to measurable revenue and efficiency, with software development, healthcare, customer support, and industrial automation among the first domains to show strong productivity gains.\n\nEvery successful AI application pulls on every layer beneath it—from energy to chips to infrastructure to models. Applications are where AI infrastructure becomes productivity, growth, and national capability.
0\n[images/image-22-Inference At-Scale is Extreme Computing.jpg]\n\n
\ntext_image\nInference At-Scale is Extreme Computing\nThroughput\nRevenue\nTokens per Second For Factory\nFLOPS\nHBM Memory\nHBM Bandwidth\nSoftware and Architecture\nSmart AI\nFast Response\nTokens per Second For C\nfor C\n
\nInference Inflection Driving Compute Growth\nMarking a fundamental shift, inference has overtaken training as the dominant workload. AI is no longer just being built; it is being used at scale. With that shift, the economics of AI have changed. Demand is no longer conceptual—the AI flywheel is accelerating.\nInference demand is surging as multiple exponential forces converge. Models are becoming more capable. Agents built on these models can reason, plan, and solve complex problems, generating tokens at every step. As these agents prove their value, they are deployed across more workflows, by more users, in more industries—generating still more tokens.\nMore tokens require more compute. More compute enables smarter AI. Compute produces intelligence, intelligence drives demand, and demand drives more compute. The supply-demand flywheel is formed and accelerating.\nIn fiscal 2026, Data Center revenue reached \\$194 billion, up 68 percent year over year. Major cloud service providers, hyperscalers, AI model makers, and enterprises have deployed nearly 9 gigawatts of Blackwell AI factory capacity. Our networking revenue exceeded \\$31 billion. Our sovereign AI business more than tripled to over \\$30 billion.\nComputing Architecture Defines AI Factory Economics\nAn AI factory is not a conventional data center. It is a production system—a revenue-generating engine.\nIts purpose is to manufacture tokens at scale—with the highest intelligence, the highest throughput, the lowest latency, and the greatest energy and cost efficiency. In an AI factory, token throughput determines how much work can be performed. Cost per token determines the economics. Tokens per watt determine revenue per megawatt.\nComputing architecture determines it all—directly shaping the performance, economics, and revenue of the AI factory.\nBlackwell changed the economics of inference, and the results are reflected in our FY26 scale. Independent benchmarks show up to 50x higher throughput per megawatt and up to 35x lower cost per token than Hopper, with the greatest gains in low-latency workloads. NVIDIA delivers the lowest cost per token and the highest revenue potential for the AI factories our customers operate. Our extreme co-design across silicon, systems, networking, and software is what makes that possible.\nInference is one of the hardest computing challenges and is mission-critical to our customers' businesses. That is why we made it a company-wide priority. With Grace Blackwell, NVIDIA leads across the full lifecycle of AI—from training to inference. Grace Blackwell—the world's first rack-scale system—integrates NVLink, NVFP4, TensorRT-LLM, Spectrum-X, and InfiniBand into a single, unified computer. Dynamo—our inference operating system—optimizes every query, every token, every watt—maximizing throughput, minimizing cost, and increasing factory output.\nGrace Blackwell delivered a step-function leap in LLM training and inference. We built the Vera Rubin platform to run agents. Where Grace Blackwell introduced rack-scale systems, Vera Rubin extends to pod-scale production.\nVera Rubin is not a chip. It is an AI factory platform—realized through extreme co-design across Rubin GPUs, Vera CPUs, Groq LPUs, BlueField DPUs, NVLink switching, Spectrum-X and InfiniBand networking, storage, systems, and software.\nWith the addition of Groq technology, the platform now includes a new tier of ultra-low-latency inference—built for real-time, agentic workloads.\nThe computing demands of agentic systems are massive and complex. Vera Rubin has arrived just in time.\nCUDA Installed Base Powering AI Flywheel\nThis year marked the 20th anniversary of CUDA.\nCUDA is a revolutionary programming model—and the foundation of the NVIDIA platform. Over two decades, CUDA has created a powerful developer flywheel that continues to compound performance, adoption, and innovation across a global ecosystem. Its capabilities attract developers. Developers create new algorithms. New algorithms drive breakthroughs. Breakthroughs create new markets. New markets expand the installed base. And the installed base attracts more developers. The flywheel is now accelerating.\nCUDA also extends the lifespan of NVIDIA systems. Our platforms are architecturally compatible across generations, and we continuously optimize the software stack. Improvements in libraries, runtimes, compilers, and frameworks benefit the entire installed base. Customers do not just get the hardware's performance on day one—they get continuous performance gains and runtime cost reductions over the life of the system.\nThis dynamic is foundational to AI infrastructure builders and operators. CUDA attracts developers and customers. It lowers the total cost of ownership for operators and users. It extends the useful life of every system.\nIt is why prior-generation products remain valuable. It is why customers build with confidence across Blackwell, Vera Rubin, and beyond.\nThe confidence to build on NVIDIA—driven by CUDA—is one of the deepest moats we have created.\nA Global AI Ecosystem\nThe NVIDIA ecosystem now reaches the market through every major route: hyperscalers, cloud providers, AI-native clouds, enterprise IT, sovereign deployments, and industrial systems. Our business is broad and diversified.\nAWS, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure are expected to be among the first major cloud providers to deploy Vera Rubin-based infrastructure. Meta expanded to a multi-year, multigenerational strategic partnership spanning on-premises, cloud, and AI infrastructure. CoreWeave, Nebius, Nscale, and many other AI clouds continue to expand AI factory capacity. OpenAI, Anthropic, xAI, and new frontier and open model builders are scaling on NVIDIA.\nWe are also extending that reach into on-prem and air-gapped deployments. With partners such as Palantir and Dell, we are enabling customers to stand up AI platforms in any country, region, or field environment where privacy, sovereignty, resilience, and control matter most.\nTo help customers design, build, and operate AI factories, we provide reference designs, blueprints, and a partner ecosystem.\nThe Vera Rubin DSX AI Factory reference design and NVIDIA Omniverse DSX Blueprint bring together compute, networking, storage, power, cooling, and operations into an open, modular architecture supported by partners including Vertiv, Schneider Electric, Eaton, Jacobs, Phaidra, Procore, PTC, Siemens, Switch, Trane Technologies, Cadence, and Dassault Systèmes. The broader power-to-rack ecosystem now also includes GE Vernova, which is using digital twins aligned with the Omniverse DSX Blueprint to model grid behavior, substations, and AI factory loads before deployment. Across industrial AI, Caterpillar is using NVIDIA Omniverse digital twins to accelerate AI-driven manufacturing and automation.\nReindustrializing America and Creating Skilled Jobs\nAI is driving the largest infrastructure buildout in human history and creating a once-in-a-generation opportunity to reindustrialize America and bring advanced manufacturing back to the United States. Manufacturing in America strengthens supply chain resilience, reduces geopolitical risk, and creates high-quality jobs for people who choose careers outside of traditional four-year degrees—helping build a more balanced and inclusive economy.\nBuilding an AI factory requires electricians, pipefitters, welders, ironworkers, HVAC technicians, and networking and computer specialists.\nThis buildout extends far beyond the data center. It spans semiconductor fabs, advanced packaging facilities, server and supercomputer assembly, power generation, transmission lines, substations, and the vast upstream supply chains and downstream services that support this new multi-trillion-dollar industry.\nOnce operational, these AI factories create permanent, high-skill careers—data center technicians, critical facility engineers, and site managers—where degreed engineers and skilled tradespeople work side by side as peers.\nIn the United States alone, this buildout can create opportunities for hundreds of thousands of skilled craft workers in the years ahead to build and operate AI infrastructure.\nNVIDIA is helping lead America's reindustrialization. We have begun domestic production of NVIDIA AI infrastructure, including chip manufacturing in Arizona with TSMC, advanced packaging with Amkor and SPIL, and AI supercomputer assembly in Texas with Foxconn and Wistron—ramping over the coming year.\nDemocratizing Enterprise Agentic AI\nEvery enterprise software company needs an agentic AI strategy.\nOpenClaw showed how quickly open agentic computing can spread. Agents will not replace software tools—they will dramatically increase their use, operating at accelerated machine speed and requiring tools that keep up.\nEnterprise software platforms will embed specialized agents trained on proprietary data and domain-specific skills—capable of securely accessing sensitive information and operating in regulated environments. General-purpose and specialized agents will work together as a digital workforce to orchestrate complex workflows, from designing chips and running cloud services to operating factories.\nNVIDIA provides the open, full-stack platform for the enterprise agent lifecycle—from build, evaluation, and guardrails to deployment.\nOpenShell and NVIDIA Agent Toolkit help agents access tools, follow policies, protect data, and operate inside companies. NeMo enables enterprises to build and customize models and agents on proprietary data. NIM packages optimized AI models into production-ready services. Nemotron provides open reasoning and agentic models.\nThe enterprise ecosystem around this platform is expanding rapidly. Adobe, Cisco, Palantir, SAP, Salesforce, ServiceNow, Siemens, Synopsys, Cadence, Dassault Systèmes, and other industry leaders are building domain-specific agents and AI systems on NVIDIA.\nOpen Models Accelerate AI for All Industries\nFrontier model services are excellent for general intelligence use cases. They work out of the box and are continuously getting smarter. At NVIDIA, we use OpenAI Codex, Anthropic Claude Code, and Cursor.\nStill, open models are essential.\nOpen models advance the field of AI. They expose architectures, weights, and training methods—enabling comparison, validation, and rapid innovation. Breakthroughs compound across the ecosystem.\nOpen models improve safety and security. Cybersecurity companies fine-tune open models and deploy swarms of specialized agents to defend against cyberattacks. In regulated industries and national deployments, enterprises build agents to enforce compliance and operate within strict controls.\nOpen models enable sovereignty. Nations build AI aligned with their language, culture, and values—and keep data within their borders.\n\nFrontier model services are excellent for general intelligence use cases. They work out of the box and are continuously getting smarter. At NVIDIA, we use OpenAI Codex, Anthropic Claude Code, and Cursor.
0\n[images/image-23-NVIDIA GTC Event Scene.jpg]\n\n
\nnatural_image\nOutdoor scene with people using green IoT robots near a decorated park, no visible text or symbols on the robots or background.\n
\nOpen models enable specialization at scale. Enterprises build domain-specific models trained on proprietary data and deep expert knowledge.\nThe open model ecosystem is large and accelerating. Nearly 3 million open models now span language, vision, biology, physics, and robotics. NVIDIA is a leading contributor, with six families of frontier models, plus the data, recipes, and frameworks developers need to customize them.\nWe build open models across the major domains of AI: Nemotron for reasoning and language, Cosmos for physical AI, GROOT for robotics, Alpamayo for autonomous driving, BioNeMo for biology and chemistry, and Earth-2 for weather and climate. These models enable domain-specific intelligence on a common NVIDIA platform.\nAdoption spans industries—from cybersecurity to drug discovery to mobility. CrowdStrike, Lilly, and Mercedes-Benz are building specialized AI systems on NVIDIA. Researchers in Japan, France, Sweden, India, and Israel are building national models.\nScientists worldwide are advancing research with open models.\nAI as National Infrastructure\nIn every industrial revolution—from electricity to telecommunications to the internet—nations build infrastructure to turn new technology into national capability, economic growth, and strategic independence. Every country will deploy AI infrastructure—whether with global partners or domestic companies—aligned with its language, culture, and industries.\nWe saw this accelerate sharply in fiscal 2026. Demand broadened as countries and regions built their own AI capacity, cloud services, research infrastructure, and industrial ecosystems.\nThe ChatGPT Moment for Physical AI Has Arrived\nThe next wave of AI is physical. AI is moving from screens into factories, warehouses, hospitals, vehicles, and robots. To operate in the real world, AI must understand geometry, motion, causality, and physics. It must reason and act safely in dynamic, unpredictable environments.\nPhysical AI is a new frontier in computing.\nNVIDIA is building the full training-to-deployment loop for physical AI. AI factories train models. Omniverse, with Cosmos World Foundation Models, simulates worlds, robots, and workflows with physical fidelity. GROOT advances general-purpose robotics. AGX computers deploy intelligence into robots, instruments, industrial systems, and autonomous vehicles.\nThis ecosystem is already taking shape. Agility Robotics, Amazon Robotics, Boston Dynamics, Figure, Uber, Mercedes-Benz, and many other robotics, automotive, and industrial leaders are building on NVIDIA platforms. Industrial software leaders, including Cadence, Dassault Systèmes, PTC, Siemens, and Synopsys, are bringing NVIDIA-accelerated design, simulation, and digital twins to millions of engineers, researchers, and manufacturers. Biology is also entering this loop, with BioNeMo expanding AI-driven drug discovery and co-innovation efforts such as our work with Lilly.\nPhysical AI is no longer a research project. It is moving into production.\n\nThe next wave of AI is physical. AI is moving from screens into factories, warehouses, hospitals, vehicles, and robots. To operate in the real world, AI must understand geometry, motion, causality, and physics. It must reason and act safely in dynamic, unpredictable environments.
0\n[images/image-24-Silhouette of Man Gazing at Earth.jpg]\n\n
\nnatural_image\nSilhouette of a person standing on a platform overlooking Earth with cloud patterns and a colorful sky (no text or symbols visible)\n
\nThe Buildout Has Just Begun\nAI has been advancing at an extraordinary pace. In just a few years, ChatGPT awakened the world to AI. Reasoning systems allow models to think before answering. Now agents can use tools and do work. Enterprise adoption is surging, settling the ROI debate: AI can perform valuable work, demand is accelerating, and AI services are delivering compelling economics—because tokens now perform valuable work at scale.\nAhead lies the largest infrastructure buildout in human history—AI factories are being built around the world, powering digital and robotic agents that will supercharge every industry. Building a new industry requires resources—but it also creates opportunities. AI will demand energy, computing, networking, software, and advanced manufacturing.\nAI will transform how we work—and how the world works. It will automate most tasks, reshape every job, eliminate some, and create entirely new ones. That is why it is important to distinguish between a task and the purpose of a job. Coding is a task; innovation is the purpose. Reading scans is a task; caring for patients is the purpose. AI will automate more tasks so people can do more of the work that matters.\nAs tasks are automated, the purpose of work will rise—AI will amplify human potential and expand what humanity can achieve. New companies will emerge on this foundation, extending AI into every industry. Companies and nations must build now. The future is arriving faster than any platform shift in history.\nNVIDIA has spent more than three decades preparing for this moment.\nWe are building the computers, the networks, the software, the models, and the factories that manufacture intelligence. We are building the computing platform for companies, industries, nations, and all of us to thrive in the age of AI.\nTo our employees, partners, developers, customers, and shareholders around the world—thank you.\nNVIDIA is the result of a decades-long vision and the life's work of our people. We carry this opportunity forward with humility, responsibility, determination, and optimism for what comes next.\n$$\nj u l t h y\n$$\nJensen Huang\nFounder and CEO, NVIDIA\nMay 2026\nNVIDIA Corporation\nNotice of 2026 Annual Meeting\nProxy Statement and Form 10-K\nForward-Looking Statements\nCertain statements in this document including, but not limited to, statements as to: the benefits, impact, performance, growth, and availability of NVIDIA's products, services, and technologies, and related trends and drivers; expectations with respect to supply and demand for NVIDIA's products, services and technologies; expectations with respect to NVIDIA's revenue; expectations with respect to NVIDIA's third party arrangements, including with its collaborators and partners; expectations with respect to technology developments; expectations with respect to AI and related industries; NVIDIA driving the computing platform shift, building the computing infrastructure of the AI era; NVIDIA building the computing infrastructure of the AI era; NVIDIA helping lead America's reindustrialization; NVIDIA building the full training-to-deployment loop for physical AI; NVIDIA building the computers, the networks, the software, the models, and the factories that manufacture intelligence; NVIDIA building the computing platform for companies, industries, nations, and all of us to thrive in the age of AI; and other statements that are not historical facts are forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the “safe harbor” created by those sections based on management’s beliefs and assumptions and on information currently available to management and are subject to risks and uncertainties that could cause results to be materially different than expectations. Important factors that could cause actual results to differ materially include: global economic and political conditions; NVIDIA’s reliance on third parties to manufacture, assemble, package and test NVIDIA’s products; the impact of technological development and competition; development of new products and technologies or enhancements to NVIDIA’s existing product and technologies; market acceptance of NVIDIA’s products or NVIDIA’s partners’ products; design, manufacturing or software defects; changes in consumer preferences or demands; changes in industry standards and interfaces; unexpected loss of performance of NVIDIA’s products or technologies when integrated into systems; NVIDIA’s ability to realize the potential benefits of business investments or acquisitions; and changes in applicable laws and regulations, as well as other factors detailed from time to time in the most recent reports NVIDIA files with the Securities and Exchange Commission, or SEC, including, but not limited to, its annual report on Form 10-K and quarterly reports on Form 10-Q. Copies of reports filed with the SEC are posted on the company’s website and are available from NVIDIA without charge. These forward-looking statements are not guarantees of future performance and speak only as of the date hereof, and, except as required by law, NVIDIA disclaims any obligation to update these forward-looking statements to reflect future events or circumstances.\nNVIDIA®\nNOTICE OF 2026 ANNUAL MEETING OF STOCKHOLDERS\nDate and time: Wednesday, June 24, 2026 at 9:00 a.m. Pacific Time\nLocation: Virtually at www.virtualshareholdermeeting.com/NVDA2026\nItems of • Election of ten directors nominated by the Board of Directors\n• Advisory approval of our executive compensation\n- Ratification of the selection of PricewaterhouseCoopers LLP as our independent registered public accounting firm for fiscal year 2027\n• Four stockholder proposals, if properly presented\n• Transaction of other business properly brought before the meeting\nRecord date: You can attend and vote at the 2026 Meeting if you were a stockholder of record at the close of business on April 27, 2026.\nVirtual meeting admission: We will be holding the 2026 Meeting virtually at the location listed above. To participate, you will need the Control Number included on your notice of proxy materials or printed proxy card.\nPre-meeting forum: To communicate with our stockholders in connection with the 2026 Meeting, we have established a pre-meeting forum located at www.proxyvote.com where you can submit questions in advance.\nYour vote is very important. Whether or not you plan to attend the 2026 Meeting, PLEASE VOTE YOUR SHARES. As an alternative to voting during the 2026 Meeting, you may vote in advance online, by telephone, or, if you have elected to receive a paper proxy card in the mail, by mailing the completed proxy card.\nImportant notice regarding the availability of proxy materials for the Annual Meeting of Stockholders to be held on June 24, 2026. This Notice, our Proxy Statement, our Annual Report on Form 10-K, and our Annual Review are available at www.nvidia.com/proxy.\nBy Order of the Board of Directors\n\nVirtual meeting admission: We will be holding the 2026 Meeting virtually at the location listed above. To participate, you will need the Control Number included on your notice of proxy materials or printed proxy card. Timothy S. Teter0\n[images/image-25-Virtual me.jpg]\n\nTimothy S. Teter\nSecretary\n2788 San Tomas Expressway, Santa Clara, California 95051\nMay 12, 2026\nNVIDIA Corporation\nThis proxy statement contains forward-looking statements. All statements other than statements of historical or current facts made in this document are forward-looking. Forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the \"safe harbor\" created by those sections, are based on our management's beliefs and assumptions and on information currently available to our management. In some cases, you can identify forward-looking statements by terms such as \"may,\" \"will,\" \"should,\" \"could,\" \"goal,\" \"would,\" \"expect,\" \"plan,\" \"anticipate,\" \"believe,\" \"estimate,\" \"project,\" \"predict,\" \"potential,\" and similar expressions intended to identify forward-looking statements. Actual results could differ materially for a variety of reasons. Risks and uncertainties that could cause our actual results to differ significantly from management's expectations are described in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026.\nDEFINITIONS\n\n[tables/table-1 NVIDIA Definitions.html]", + "path": "2026-Annual-Report-Web.pdf", + "metadata": { + "length": 42608, + "summary": "NVIDIA's fiscal 2026 annual review details the company's role in the AI platform shift, moving from general-purpose computing to accelerated computing that manufactures intelligence in real time. Revenue grew 65% to $216 billion, driven by Data Center revenue of $194 billion. The document describes AI as a five-layer stack (energy, chips, infrastructure, models, applications) and AI factories as revenue-generating infrastructure converting energy into tokens. Key themes include inference overtaking training as the dominant workload, the Blackwell and Vera Rubin platforms, CUDA's 20-year flywheel, open models, agentic AI entering production, physical AI for robotics and autonomous vehicles, and AI as national infrastructure driving the largest infrastructure buildout in history.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "2026", + "NVIDIA", + "Corporation", + "Annual", + "Review", + "Notice", + "Meeting", + "Proxy", + "Statement", + "Form", + "10", + "details", + "summary", + "natural", + "image", + "Abstract", + "3D", + "geometric", + "design", + "green", + "spiral", + "rectangular", + "forms", + "grid", + "background", + "text", + "symbols", + "For", + "computing", + "history", + "software", + "pre", + "recorded", + "Humans", + "algorithm", + "Computers", + "executed", + "Now", + "trained", + "It", + "learns", + "data", + "generates", + "intelligence", + "real", + "time", + "General", + "purpose", + "run", + "Moore", + "Law", + "slowed", + "Dennard", + "scaling", + "ended", + "Accelerated", + "foundational", + "Because", + "produced", + "entire", + "stack", + "beneath", + "reinvented", + "AI", + "runs", + "hardware", + "energy", + "economics", + "takes", + "raw", + "materials", + "converts", + "scale", + "This", + "normal", + "upgrade", + "cycle", + "The", + "world", + "platform", + "shift", + "flowchart", + "mermaid", + "graph", + "TD", + "APPLICATIONS", + "MODELS", + "EINFRASTRUCTURE", + "CHIPS", + "ENERGY", + "CHATBOTS0", + "CHATBOTS", + "DIGITAL", + "BIOLOGY0", + "BIOLOGY", + "ROBOTAXI0", + "ROBOTAXI", + "ENTERPRISE0", + "ENTERPRISE", + "AGENTS", + "SCIENCE0", + "SCIENCE", + "ROBOTICS0", + "ROBOTICS", + "MANUFACTURING0", + "MANUFACTURING", + "CODER0", + "CODER", + "LLM", + "VLM", + "VLA", + "MMLLM", + "GPT", + "DM", + "GNN", + "MOE", + "SSM", + "LBM", + "FACTORIES", + "Stacked", + "cylindrical", + "containers", + "transparent", + "lids", + "arranged", + "rows", + "visible", + "Is", + "Five", + "Layer", + "Cake", + "clever", + "app", + "single", + "model", + "essential", + "infrastructure", + "electricity", + "internet", + "When", + "industrially", + "resolves", + "layer", + "Energy", + "chips", + "systems", + "models", + "applications", + "Every", + "successful", + "application", + "pulls", + "power", + "plant", + "powers", + "computation", + "Computation", + "amount", + "system", + "produce", + "bounded", + "Driving", + "Largest", + "Infrastructure", + "Buildout", + "Human", + "History", + "Data", + "centers", + "factories", + "Land", + "cooling", + "networking", + "parallel", + "company", + "country", + "build", + "industrial", + "revolution", + "We", + "hundred", + "billion", + "dollars", + "Trillions", + "built", + "Reference", + "Designs", + "Best", + "Practices", + "Methodologies", + "Power", + "Chip", + "Systems", + "Facilities", + "Chips", + "Rubin", + "DSX", + "Ecosystem", + "Omniverse", + "Blueprint", + "Max", + "Flex", + "Exchange", + "Sim", + "Libraries", + "APIs", + "Software", + "Liquid", + "Cooling", + "Factories", + "Manufacture", + "Intelligence", + "stored", + "Modern", + "converting", + "tokens", + "plants", + "convert", + "Throughput", + "megawatt", + "defines", + "production", + "Cost", + "token", + "efficiency", + "Token", + "output", + "performance", + "watt", + "directly", + "determine", + "revenue", + "Inference", + "Defines", + "Economics", + "agentic", + "low", + "latency", + "long", + "context", + "multi", + "step", + "reasoning", + "throughput", + "determines", + "cost", + "continuously", + "drives", + "Reasoning", + "generate", + "dramatically", + "They", + "thinking", + "answer", + "fast", + "stay", + "interactive", + "SemiAnalysis", + "InferenceX", + "measured", + "GB300", + "NVL72", + "delivering", + "50x", + "higher", + "35x", + "lower", + "compared", + "Hopper", + "greatest", + "gains", + "workloads", + "chip", + "problem", + "architecture", + "GPU", + "memory", + "designed", + "semianalysis", + "InferenceMAX", + "KING", + "Open", + "Models", + "reached", + "frontier", + "Researchers", + "startups", + "enterprises", + "nations", + "rely", + "open", + "participate", + "advanced", + "Most", + "free", + "At", + "level", + "capability", + "change", + "activate", + "demand", + "barriers", + "entry", + "enable", + "deep", + "customization", + "teams", + "domain", + "specific", + "common", + "platforms", + "As", + "developers", + "customize", + "deploy", + "generation", + "grows", + "line", + "chart", + "Six", + "panel", + "illustration", + "showing", + "molecular", + "structures", + "globe", + "person", + "waving", + "humanoid", + "robot", + "car", + "Agentic", + "Enters", + "Production", + "perceives", + "reasons", + "plans", + "acts", + "break", + "complex", + "problems", + "steps", + "evaluate", + "alternatives", + "tools", + "retrieve", + "information", + "act", + "running", + "workflows", + "coordinate", + "environments", + "OpenClaw", + "showed", + "developer", + "agents", + "put", + "work", + "move", + "chat", + "access", + "execute", + "code", + "communicate", + "All", + "happen", + "securely", + "privately", + "policy", + "control", + "vGPU", + "FILES", + "STRUCTURED", + "UNSTRUCTURED", + "Computer", + "Use", + "NEMOCLAW", + "MEMORY", + "Sub", + "cuOPT", + "SKILLS", + "NEMOTRON", + "NEMO", + "DYNAMO", + "NIM", + "AGentic", + "moving", + "enterprise", + "crit", + "TOOLS", + "CLI", + "MCP", + "OpenShell", + "style", + "fill", + "99ccff", + "stroke", + "333", + "f0f0f0", + "e0e0e0", + "d0d0d0", + "critical", + "Physical", + "Puts", + "Motion", + "Robotics", + "autonomous", + "vehicles", + "intelligent", + "machines", + "combine", + "perception", + "understanding", + "physics", + "operate", + "dynamic", + "physical", + "Foundation", + "refined", + "simulation", + "deployed", + "edge", + "creating", + "continuous", + "loop", + "virtual", + "training", + "experience", + "transforming", + "warehouses", + "transportation", + "healthcare", + "Everything", + "moves", + "Interior", + "modern", + "exhibition", + "space", + "featuring", + "large", + "excavator", + "robots", + "child", + "orange", + "costume", + "Reinvents", + "Graphics", + "GeForce", + "brought", + "home", + "graphics", + "longer", + "limited", + "traditional", + "rendering", + "With", + "neural", + "DLSS", + "predicts", + "pixels", + "pixel", + "render", + "rest", + "Neural", + "networks", + "alongside", + "shaders", + "ray", + "tracing", + "cores", + "replace", + "enhances", + "In", + "era", + "learned", + "crossed", + "threshold", + "generating", + "answers", + "performing", + "manufacture", + "building", + "Jahibing", + "Elderly", + "man", + "black", + "leather", + "jacket", + "gesturing", + "holding", + "handheld", + "device", + "Dear", + "NVIDIAs", + "Stakeholders", + "Computing", + "15", + "years", + "mainframe", + "PC", + "cloud", + "mobile", + "What", + "happening", + "bigger", + "undergoing", + "complete", + "reset", + "IBM", + "introduced", + "System", + "360", + "1964", + "computers", + "kind", + "manufactures", + "driving", + "shifting", + "human", + "coding", + "machine", + "learning", + "CPUs", + "GPUs", + "paradigm", + "unlike", + "prerecorded", + "retrieved", + "generated", + "taking", + "account", + "prior", + "knowledge", + "instructions", + "But", + "computer", + "understand", + "reason", + "plan", + "appears", + "artificial", + "Over", + "past", + "year", + "investors", + "questioned", + "ROI", + "good", + "Early", + "chatbots", + "greatly", + "improved", + "productive", + "months", + "landscape", + "shifted", + "sharply", + "Enterprise", + "adoption", + "surged", + "responses", + "actions", + "compute", + "accelerating", + "arrived", + "remarkable", + "results", + "Generative", + "creates", + "thinks", + "perform", + "tasks", + "thoughtful", + "planning", + "research", + "important", + "simply", + "questions", + "major", + "breakout", + "case", + "sweeping", + "industry", + "Within", + "Soon", + "orders", + "magnitude", + "engineers", + "combined", + "Limited", + "imagination", + "pace", + "innovation", + "accelerate", + "industries", + "codify", + "result", + "fully", + "automate", + "business", + "process", + "Organizations", + "rethink", + "operations", + "end", + "customer", + "engagement", + "internal", + "supply", + "chain", + "enhanced", + "accelerated", + "Companies", + "adopt", + "support", + "innovate", + "attract", + "talent", + "fundamental", + "Reimagine", + "Possible", + "Poster", + "Gallery", + "JOS", + "CIVIC", + "Bough", + "Discoveries", + "Compute", + "asset", + "productivity", + "economy", + "tool", + "age", + "manufacturing", + "equipment", + "requires", + "Traditional", + "store", + "serve", + "files", + "Today", + "turn", + "basic", + "units", + "digital", + "assets", + "delivers", + "full", + "spans", + "lifecycle", + "inference", + "language", + "video", + "biology", + "robotics", + "vertically", + "integrated", + "horizontally", + "architect", + "mindset", + "executing", + "extreme", + "storage", + "spanning", + "LPUs", + "DPUs", + "factory", + "remains", + "integrates", + "broadest", + "ecosystems", + "Our", + "clouds", + "premises", + "telecom", + "medical", + "instruments", + "Platform", + "Shift", + "GTC", + "land", + "unmistakably", + "clear", + "moved", + "experimentation", + "laid", + "roadmap", + "Blackwell", + "Ultra", + "Vera", + "Feynman", + "builders", + "chains", + "visibility", + "trillion", + "cumulative", + "start", + "2025", + "2027", + "Since", + "DC", + "October", + "received", + "additional", + "purchase", + "diversified", + "base", + "including", + "Anthropic", + "Meta", + "growing", + "set", + "class", + "Cursor", + "Perplexity", + "Thinking", + "Machines", + "Lab", + "Reflection", + "Skild", + "service", + "providers", + "continues", + "addition", + "significant", + "buildouts", + "underway", + "Azure", + "OCI", + "AWS", + "Google", + "announced", + "expansions", + "partners", + "CoreWeave", + "Nebius", + "Nscale", + "experiencing", + "exponential", + "secured", + "term", + "offtake", + "agreements", + "momentum", + "reflects", + "Like", + "steam", + "technology", + "forming", + "foundation", + "finally", + "top", + "turbines", + "generators", + "priced", + "kilowatt", + "hour", + "distributed", + "globally", + "enabled", + "million", + "existing", + "Each", + "created", + "companies", + "utilities", + "emerged", + "defined", + "classes", + "tens", + "thousands", + "native", + "Above", + "While", + "widely", + "extends", + "includes", + "chemistry", + "finance", + "medicine", + "bringing", + "core", + "economic", + "social", + "development", + "drug", + "discovery", + "automation", + "ahead", + "orchestrate", + "billions", + "working", + "people", + "creative", + "capable", + "Applications", + "growth", + "national", + "Fiscal", + "Defining", + "Year", + "Revenue", + "grew", + "65", + "percent", + "216", + "Operating", + "income", + "rose", + "60", + "130", + "Diluted", + "EPS", + "increased", + "67", + "4.90", + "103", + "operating", + "cash", + "flow", + "returned", + "41", + "shareholders", + "Demand", + "remained", + "strong", + "hyperscalers", + "makers", + "sovereigns", + "investment", + "worldwide", + "marked", + "turning", + "point", + "measurable", + "domains", + "show", + "Scale", + "Extreme", + "Tokens", + "Second", + "Factory", + "FLOPS", + "HBM", + "Memory", + "Bandwidth", + "Architecture", + "Smart", + "Fast", + "Response", + "Inflection", + "Growth", + "Marking", + "overtaken", + "dominant", + "workload", + "changed", + "conceptual", + "flywheel", + "surging", + "multiple", + "forces", + "converge", + "Agents", + "solve", + "prove", + "users", + "More", + "require", + "enables", + "smarter", + "produces", + "formed", + "fiscal", + "Center", + "194", + "68", + "Major", + "gigawatts", + "capacity", + "exceeded", + "31", + "sovereign", + "tripled", + "30", + "An", + "conventional", + "center", + "engine", + "Its", + "highest", + "lowest", + "performed", + "shaping", + "reflected", + "FY26", + "Independent", + "benchmarks", + "potential", + "customers", + "silicon", + "makes", + "hardest", + "challenges", + "mission", + "businesses", + "That", + "made", + "wide", + "priority", + "Grace", + "leads", + "rack", + "NVLink", + "NVFP4", + "TensorRT", + "Spectrum", + "InfiniBand", + "unified", + "Dynamo", + "optimizes", + "query", + "maximizing", + "minimizing", + "increasing", + "delivered", + "function", + "leap", + "Where", + "pod", + "realized", + "Groq", + "BlueField", + "switching", + "tier", + "ultra", + "demands", + "massive", + "CUDA", + "Installed", + "Base", + "Powering", + "Flywheel", + "20th", + "anniversary", + "revolutionary", + "programming", + "decades", + "powerful", + "compound", + "global", + "ecosystem", + "capabilities", + "Developers", + "create", + "algorithms", + "New", + "drive", + "breakthroughs", + "Breakthroughs", + "markets", + "expand", + "installed", + "And", + "attracts", + "lifespan", + "architecturally", + "compatible", + "generations", + "optimize", + "Improvements", + "libraries", + "runtimes", + "compilers", + "frameworks", + "benefit", + "Customers", + "day", + "runtime", + "reductions", + "life", + "operators", + "lowers", + "total", + "ownership", + "products", + "remain", + "valuable", + "confidence", + "driven", + "deepest", + "moats", + "Global", + "reaches", + "market", + "route", + "IT", + "deployments", + "broad", + "Cloud", + "Microsoft", + "Oracle", + "expected", + "based", + "expanded", + "multigenerational", + "strategic", + "partnership", + "continue", + "OpenAI", + "xAI", + "extending", + "reach", + "prem", + "air", + "gapped", + "Palantir", + "Dell", + "enabling", + "stand", + "region", + "field", + "environment", + "privacy", + "sovereignty", + "resilience", + "matter", + "To", + "provide", + "reference", + "designs", + "blueprints", + "partner", + "bring", + "modular", + "supported", + "Vertiv", + "Schneider", + "Electric", + "Eaton", + "Jacobs", + "Phaidra", + "Procore", + "PTC", + "Siemens", + "Switch", + "Trane", + "Technologies", + "Cadence", + "Dassault", + "Syst", + "mes", + "broader", + "GE", + "Vernova", + "twins", + "aligned", + "behavior", + "substations", + "loads", + "deployment", + "Across", + "Caterpillar", + "Reindustrializing", + "America", + "Creating", + "Skilled", + "Jobs", + "largest", + "buildout", + "opportunity", + "reindustrialize", + "back", + "United", + "States", + "Manufacturing", + "strengthens", + "reduces", + "geopolitical", + "risk", + "high", + "quality", + "jobs", + "choose", + "careers", + "degrees", + "helping", + "balanced", + "inclusive", + "Building", + "electricians", + "pipefitters", + "welders", + "ironworkers", + "HVAC", + "technicians", + "specialists", + "semiconductor", + "fabs", + "packaging", + "facilities", + "server", + "supercomputer", + "assembly", + "transmission", + "lines", + "vast", + "upstream", + "downstream", + "services", + "dollar", + "Once", + "operational", + "permanent", + "skill", + "facility", + "site", + "managers", + "degreed", + "skilled", + "tradespeople", + "side", + "peers", + "opportunities", + "hundreds", + "craft", + "workers", + "lead", + "reindustrialization", + "begun", + "domestic", + "Arizona", + "TSMC", + "Amkor", + "SPIL", + "Texas", + "Foxconn", + "Wistron", + "ramping", + "coming", + "Democratizing", + "strategy", + "quickly", + "spread", + "increase", + "speed", + "requiring", + "embed", + "specialized", + "proprietary", + "skills", + "accessing", + "sensitive", + "regulated", + "workforce", + "designing", + "agent", + "evaluation", + "guardrails", + "Agent", + "Toolkit", + "follow", + "policies", + "protect", + "inside", + "NeMo", + "packages", + "optimized", + "ready", + "Nemotron", + "expanding", + "rapidly", + "Adobe", + "Cisco", + "SAP", + "Salesforce", + "ServiceNow", + "Synopsys", + "leaders", + "Accelerate", + "Industries", + "Frontier", + "excellent", + "general", + "cases", + "box", + "Codex", + "Claude", + "Code", + "Still", + "advance", + "expose", + "architectures", + "weights", + "methods", + "comparison", + "validation", + "rapid", + "improve", + "safety", + "security", + "Cybersecurity", + "fine", + "tune", + "swarms", + "defend", + "cyberattacks", + "enforce", + "compliance", + "strict", + "controls", + "Nations", + "culture", + "values", + "borders", + "Outdoor", + "scene", + "IoT", + "decorated", + "park", + "specialization", + "Enterprises", + "expert", + "Nearly", + "span", + "vision", + "leading", + "contributor", + "families", + "recipes", + "Cosmos", + "GROOT", + "Alpamayo", + "BioNeMo", + "Earth", + "weather", + "climate", + "These", + "Adoption", + "cybersecurity", + "mobility", + "CrowdStrike", + "Lilly", + "Mercedes", + "Benz", + "Japan", + "France", + "Sweden", + "India", + "Israel", + "Scientists", + "advancing", + "National", + "telecommunications", + "independence", + "broadened", + "countries", + "regions", + "ChatGPT", + "Moment", + "Has", + "Arrived", + "wave", + "screens", + "hospitals", + "geometry", + "motion", + "causality", + "safely", + "unpredictable", + "train", + "World", + "simulates", + "worlds", + "fidelity", + "advances", + "AGX", + "shape", + "Agility", + "Amazon", + "Boston", + "Dynamics", + "Figure", + "Uber", + "automotive", + "Industrial", + "millions", + "researchers", + "manufacturers", + "Biology", + "entering", + "efforts", + "project", + "Silhouette", + "standing", + "overlooking", + "patterns", + "colorful", + "sky", + "Just", + "Begun", + "extraordinary", + "awakened", + "answering", + "settling", + "debate", + "compelling", + "Ahead", + "lies", + "powering", + "robotic", + "supercharge", + "resources", + "transform", + "works", + "reshape", + "job", + "eliminate", + "distinguish", + "task", + "Coding", + "Reading", + "scans", + "caring", + "patients", + "matters", + "automated", + "rise", + "amplify", + "humanity", + "achieve", + "emerge", + "future", + "arriving", + "faster", + "spent", + "preparing", + "moment", + "thrive", + "employees", + "carry", + "forward", + "humility", + "responsibility", + "determination", + "optimism", + "Jensen", + "Huang", + "Founder", + "CEO", + "May", + "Forward", + "Looking", + "Statements", + "Certain", + "statements", + "document", + "benefits", + "impact", + "availability", + "technologies", + "related", + "trends", + "drivers", + "expectations", + "respect", + "party", + "arrangements", + "collaborators", + "developments", + "historical", + "facts", + "meaning", + "Section", + "27A", + "Securities", + "Act", + "1933", + "amended", + "21E", + "1934", + "subject", + "safe", + "harbor", + "sections", + "management", + "beliefs", + "assumptions", + "risks", + "uncertainties", + "materially", + "Important", + "factors", + "actual", + "differ", + "include", + "political", + "conditions", + "reliance", + "parties", + "assemble", + "package", + "test", + "technological", + "competition", + "enhancements", + "product", + "acceptance", + "defects", + "consumer", + "preferences", + "standards", + "interfaces", + "unexpected", + "loss", + "ability", + "realize", + "investments", + "acquisitions", + "applicable", + "laws", + "regulations", + "detailed", + "recent", + "reports", + "Commission", + "SEC", + "annual", + "report", + "quarterly", + "Copies", + "filed", + "posted", + "website", + "charge", + "guarantees", + "speak", + "date", + "hereof", + "required", + "law", + "disclaims", + "obligation", + "update", + "reflect", + "events", + "circumstances", + "NOTICE", + "OF", + "ANNUAL", + "MEETING", + "STOCKHOLDERS", + "Date", + "Wednesday", + "June", + "24", + "00", + "Pacific", + "Time", + "Location", + "Virtually", + "www", + "virtualshareholdermeeting", + "NVDA2026", + "Items", + "Election", + "ten", + "directors", + "nominated", + "Board", + "Directors", + "Advisory", + "approval", + "executive", + "compensation", + "Ratification", + "selection", + "PricewaterhouseCoopers", + "LLP", + "independent", + "registered", + "public", + "accounting", + "firm", + "Four", + "stockholder", + "proposals", + "properly", + "presented", + "Transaction", + "meeting", + "Record", + "You", + "attend", + "vote", + "record", + "close", + "April", + "27", + "Virtual", + "admission", + "virtually", + "location", + "listed", + "Control", + "Number", + "included", + "notice", + "proxy", + "printed", + "card", + "Pre", + "forum", + "stockholders", + "connection", + "established", + "located", + "proxyvote", + "submit", + "Your", + "Whether", + "PLEASE", + "VOTE", + "YOUR", + "SHARES", + "alternative", + "voting", + "online", + "telephone", + "elected", + "receive", + "paper", + "mail", + "mailing", + "completed", + "Stockholders", + "held", + "Report", + "nvidia", + "By", + "Order", + "Timothy", + "Teter0", + "Teter", + "Secretary", + "2788", + "San", + "Tomas", + "Expressway", + "Santa", + "Clara", + "California", + "95051", + "12", + "statement", + "current", + "identify", + "terms", + "goal", + "expect", + "anticipate", + "estimate", + "predict", + "similar", + "expressions", + "intended", + "Actual", + "variety", + "Risks", + "significantly", + "January", + "25", + "DEFINITIONS" + ], + "keywords": [ + "AI infrastructure", + "AI factories", + "platform shift" + ], + "connect_to": [ + { + "target": "c802bf1d-1f54-5641-9291-845298a9844e", + "relation": "embeds", + "ref": "[images/image-1-NVIDIA Logo.jpg]", + "position": { + "start": 120, + "end": 152 + } + }, + { + "target": "7d72c6f9-d7e5-5902-8ae9-d82879f2d987", + "relation": "embeds", + "ref": "[images/image-2-AI Technology Stack.jpg]", + "position": { + "start": 1109, + "end": 1149 + } + }, + { + "target": "7a05906e-9a35-58f6-b3e0-261534609b91", + "relation": "embeds", + "ref": "[images/image-3-EINFRASTRU.jpg]", + "position": { + "start": 1349, + "end": 1380 + } + }, + { + "target": "401d5609-6dc4-5055-84ef-273c4be183e4", + "relation": "embeds", + "ref": "[images/image-4-ENERGY --_.jpg]", + "position": { + "start": 1422, + "end": 1453 + } + }, + { + "target": "ddde7898-41ee-5425-9f32-0e0cc32cda1c", + "relation": "embeds", + "ref": "[images/image-5-__details_.jpg]", + "position": { + "start": 1493, + "end": 1524 + } + }, + { + "target": "ed17a528-af8b-5fab-acef-15f743819928", + "relation": "embeds", + "ref": "[images/image-6-CHATBOTS E.jpg]", + "position": { + "start": 1557, + "end": 1588 + } + }, + { + "target": "40d2a04c-e080-5d69-aacf-bbdef64c3747", + "relation": "embeds", + "ref": "[images/image-7-ROBOTAXI S.jpg]", + "position": { + "start": 1630, + "end": 1661 + } + }, + { + "target": "d50f0edb-f21d-5dd9-a58e-a8b4a7e8a199", + "relation": "embeds", + "ref": "[images/image-8-ENTERPRISE.jpg]", + "position": { + "start": 1693, + "end": 1724 + } + }, + { + "target": "4787df69-198f-5be7-91ee-c874d0b00eba", + "relation": "embeds", + "ref": "[images/image-9-AI AGENTS .jpg]", + "position": { + "start": 1761, + "end": 1792 + } + }, + { + "target": "df55c5af-3f58-552e-b0e5-e5feb62c93ea", + "relation": "embeds", + "ref": "[images/image-10-SCIENCE AI.jpg]", + "position": { + "start": 1827, + "end": 1859 + } + }, + { + "target": "d774d9fe-826e-5407-8017-f5108240b1a9", + "relation": "embeds", + "ref": "[images/image-11-Acronyms in AI and Machine Learning.jpg]", + "position": { + "start": 1891, + "end": 1948 + } + }, + { + "target": "d56a4f93-2adf-51c3-8390-5a4cf1477435", + "relation": "embeds", + "ref": "[images/image-12-Stacked Server Racks.jpg]", + "position": { + "start": 2071, + "end": 2113 + } + }, + { + "target": "dc6bc028-50f7-5a52-9ccd-34791f5c9152", + "relation": "embeds", + "ref": "[images/image-13-DSX Architecture Overview.jpg]", + "position": { + "start": 3378, + "end": 3425 + } + }, + { + "target": "e52b445e-e2fa-5846-b89f-ab6ea829a71c", + "relation": "embeds", + "ref": "[images/image-14-NVIDIA CEO InferenceMAX King.jpg]", + "position": { + "start": 5052, + "end": 5102 + } + }, + { + "target": "f10168a2-7335-546e-a811-3049f7476c6d", + "relation": "embeds", + "ref": "[images/image-15-NVIDIA Leads Open-Source AI.jpg]", + "position": { + "start": 5828, + "end": 5877 + } + }, + { + "target": "38097bff-717c-54e9-8afd-4460c4bbd5be", + "relation": "embeds", + "ref": "[tables/table-0 GPU Cloud Growth.html]", + "position": { + "start": 5920, + "end": 5958 + } + }, + { + "target": "40c19c19-1b43-5d78-8ed0-029ea1b798b0", + "relation": "embeds", + "ref": "[images/image-16-Collection of decorative enamel pins.jpg]", + "position": { + "start": 5994, + "end": 6052 + } + }, + { + "target": "8e52c8e9-8cc5-5a07-b189-931f2e5f182c", + "relation": "embeds", + "ref": "[images/image-17-Nemoclaw Agentic AI Architecture.jpg]", + "position": { + "start": 6882, + "end": 6936 + } + }, + { + "target": "1f67a0aa-68b6-54fe-b597-ba2b4d9ddbf6", + "relation": "embeds", + "ref": "[images/image-18-Future Robotics Exhibition.jpg]", + "position": { + "start": 8549, + "end": 8597 + } + }, + { + "target": "1cf6525f-8f8a-5aa4-8317-41c0a045570d", + "relation": "embeds", + "ref": "[images/image-19-Signature.jpg]", + "position": { + "start": 9577, + "end": 9608 + } + }, + { + "target": "0d23efd3-9ecf-5549-a307-eb234859c9b2", + "relation": "embeds", + "ref": "[images/image-20-Jensen Huang Presentation.jpg]", + "position": { + "start": 9692, + "end": 9739 + } + }, + { + "target": "356c6883-2d99-56e9-8bac-bf484a1ec2ff", + "relation": "embeds", + "ref": "[images/image-21-Poster Gallery Event.jpg]", + "position": { + "start": 12988, + "end": 13030 + } + }, + { + "target": "9fc7b388-4417-5637-8671-02c354c96b19", + "relation": "embeds", + "ref": "[images/image-22-Inference At-Scale is Extreme Computing.jpg]", + "position": { + "start": 19056, + "end": 19117 + } + }, + { + "target": "1908a1b6-46b8-5f35-a5d2-1704c307d0d0", + "relation": "embeds", + "ref": "[images/image-23-NVIDIA GTC Event Scene.jpg]", + "position": { + "start": 30424, + "end": 30468 + } + }, + { + "target": "0a66d713-0c32-507c-a97f-2c2744b24e39", + "relation": "embeds", + "ref": "[images/image-24-Silhouette of Man Gazing at Earth.jpg]", + "position": { + "start": 33950, + "end": 34005 + } + }, + { + "target": "881cafa8-719a-5a87-966e-32b6704576ff", + "relation": "embeds", + "ref": "[images/image-25-Virtual me.jpg]", + "position": { + "start": 41497, + "end": 41529 + } + }, + { + "target": "805808ab-5e92-5b33-8a17-5c146ec3269f", + "relation": "embeds", + "ref": "[tables/table-1 NVIDIA Definitions.html]", + "position": { + "start": 42741, + "end": 42781 + } + } + ] + } + }, + { + "chunk_id": "223d3418-74ca-5ed3-b98d-8f75d3ec1ff1", + "type": "table", + "content": "
RevenueGross MarginOperating IncomeDiluted Earnings Per Share
$215.9 billion71.1%$130.4 billion$4.90
up 65% year on yeardown 3.9 points year on yearup 60% year on yearup 67% year on year
", + "path": "tables/table-2 Revenue Growth.html", + "metadata": { + "length": 334, + "summary": "table-3\nRevenue reached $215.9 billion, up 65% year on year. Gross margin was 71.1%, down 3.9 points. Operating income rose 60% to $130.4 billion, and diluted EPS increased 67% to $4.90.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-2 Revenue Growth.html", + "keywords": [ + "Revenue", + "Gross Margin", + "Earnings" + ], + "tokens": [] + } + }, + { + "chunk_id": "b1ead365-0df2-57ec-a55a-e10e1e97e489", + "type": "text", + "content": "Fiscal 2026 marked another year of strong growth for NVIDIA, with revenue surging 65% year on year to \\$215.9 billion reflecting continued momentum across our platform transitions to accelerated computing and AI. Growth was led by Compute & Networking, where Data Center compute revenue grew 59% driven by demand for our Blackwell platform, and Data Center networking revenue increased 142% driven by the ramp of NVLink compute fabric for Blackwell systems, as well as continued growth in Ethernet and InfiniBand. Gross margin was 71.1%, operating income reached \\$130.4 billion and diluted earnings per share increased 67% to \\$4.90.\nFiscal 2026 Results\n\n[tables/table-2 Revenue Growth.html]", + "path": "2026-Annual-Report-Web.pdf/BUSINESS OVERVIEW", + "metadata": { + "length": 739, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Fiscal", + "2026", + "marked", + "year", + "strong", + "growth", + "NVIDIA", + "revenue", + "surging", + "65%", + "215.9", + "billion", + "reflecting", + "continued", + "momentum", + "platform", + "transitions", + "accelerated", + "computing", + "AI", + "Growth", + "led", + "Compute", + "Networking", + "Data", + "Center", + "compute", + "grew", + "59%", + "driven", + "demand", + "Blackwell", + "networking", + "increased", + "142%", + "ramp", + "NVLink", + "fabric", + "systems", + "Ethernet", + "InfiniBand", + "Gross", + "margin", + "71.1%", + "operating", + "income", + "reached", + "130.4", + "diluted", + "earnings", + "share", + "67%", + "4.90", + "Results" + ], + "keywords": [], + "connect_to": [ + { + "target": "223d3418-74ca-5ed3-b98d-8f75d3ec1ff1", + "relation": "embeds", + "ref": "[tables/table-2 Revenue Growth.html]", + "position": { + "start": 656, + "end": 692 + } + } + ] + } + }, + { + "chunk_id": "5db4f7dc-635c-5760-ada2-9fbc56b95f0c", + "type": "table", + "content": "
Compute & NetworkingGraphicsTotal
Revenue$193.5 billion (1)$22.5 billion (1)$215.9 billion
up 67% year on yearup 57% year on yearup 65% year on year
Operating Income (2)$130.1 billion$9.2 billion$139.3 billion
up 57% year on yearup 80% year on yearup 58% year on year
", + "path": "tables/table-3 NVIDIA Revenue.html", + "metadata": { + "length": 513, + "summary": "table-4\nNVIDIA's Compute & Networking revenue reached $193.5B (up 67% YoY) and Graphics $22.5B (up 57% YoY), totaling $215.9B (up 65% YoY). Operating income was $139.3B (up 58% YoY).", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-3 NVIDIA Revenue.html", + "keywords": [ + "Revenue", + "Operating Income", + "Growth" + ], + "tokens": [] + } + }, + { + "chunk_id": "14155530-037d-584e-8be1-bc73bb1bb53d", + "type": "text", + "content": "Our two reportable segments are “Compute & Networking” and “Graphics”:\n\n[tables/table-3 NVIDIA Revenue.html]\n\n(1) This figure has been rounded to the nearest tenth of a billion. Due to rounding, the sum of the figures does not equal the total as presented.\n(2) Segment operating income differs from consolidated operating income as certain expenses are not allocated to either Compute & Networking or Graphics for purposes of making operating decisions or assessing financial performance.", + "path": "2026-Annual-Report-Web.pdf/BUSINESS OVERVIEW/Fiscal 2026 Reportable Segments", + "metadata": { + "length": 509, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "reportable", + "segments", + "Compute", + "Networking", + "Graphics", + "This", + "figure", + "rounded", + "nearest", + "tenth", + "billion", + "Due", + "rounding", + "sum", + "figures", + "equal", + "total", + "presented", + "Segment", + "operating", + "income", + "differs", + "consolidated", + "expenses", + "allocated", + "purposes", + "making", + "decisions", + "assessing", + "financial", + "performance" + ], + "keywords": [], + "connect_to": [ + { + "target": "5db4f7dc-635c-5760-ada2-9fbc56b95f0c", + "relation": "embeds", + "ref": "[tables/table-3 NVIDIA Revenue.html]", + "position": { + "start": 72, + "end": 108 + } + } + ] + } + }, + { + "chunk_id": "bec5449f-43fd-5108-8163-5854e1ec226e", + "type": "image", + "content": "\n(1) This figure has been rounded to the nearest tenth of a billion. Due to rounding, the sum of the figures does not equal the total as presented.
0\n[images/image-26-Server Rack Hardware.jpg]\n", + "path": "images/image-26-Server Rack Hardware.jpg", + "metadata": { + "length": 192, + "summary": "image-26\n- Multiple server modules installed in a rack chassis\n- Vertical gold-colored connectors visible on each module faceplate\n- Black ventilation grilles located at the top of the unit\n- Industrial data center or telecommunications equipment", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-26-Server Rack Hardware.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "70fe8e83-b575-5ec9-b7e7-7581e10d0a2f", + "type": "image", + "content": "\n
0\n[images/image-27-NVIDIA RTX 4090.jpg]\n", + "path": "images/image-27-NVIDIA RTX 4090.jpg", + "metadata": { + "length": 55, + "summary": "image-27\nProduct: Graphics Card (GPU)\nModel: GeForce RTX 4090\nBrand: NVIDIA\nVisual Features: Dual-fan cooling system, black and silver chassis design.\nBackground: Abstract green wave pattern typical of NVIDIA branding.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-27-NVIDIA RTX 4090.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d7b79090-7714-5962-b8ae-db48ef0e7402", + "type": "image", + "content": "\n
0\n[images/image-28-Photo of a laptop workspace.jpg]\n", + "path": "images/image-28-Photo of a laptop workspace.jpg", + "metadata": { + "length": 55, + "summary": "image-28\nPrimary subject: A silver laptop displaying a terminal window with code or command-line text.\nNotable features:\n- An orange fish figurine (resembling Nemo) sitting on the left side of the keyboard.\n- A pair of eyeglasses resting on the desk to the left.\n- A gold-colored external hard drive or speaker to the right.\n- The laptop screen shows a dark interface with green and white text, likely a coding environment or system logs.\n- Background is blurred, suggesting an office or home setting.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-28-Photo of a laptop workspace.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "a5911826-592a-5302-9332-3f63a208b3da", + "type": "image", + "content": "\n
0\n[images/image-29-Automotive X-Ray Visualization.jpg]\n", + "path": "images/image-29-Automotive X-Ray Visualization.jpg", + "metadata": { + "length": 55, + "summary": "image-29\nPrimary subject: A modern sedan vehicle shown in a semi-transparent, wireframe style.\nVisual details: The image highlights the internal structure and components of the car using glowing cyan lines against a black background.\nVisible features: The chassis, wheels, suspension system, engine bay area, and cabin frame are clearly outlined to demonstrate the vehicle's architecture.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-29-Automotive X-Ray Visualization.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "b1aaba9f-bfea-50e6-bcad-6a95580a5358", + "type": "text", + "content": "Our platforms address four large markets where our expertise is critical:\n\n(1) This figure has been rounded to the nearest tenth of a billion. Due to rounding, the sum of the figures does not equal the total as presented.
0\n[images/image-26-Server Rack Hardware.jpg]\n\n
\nnatural_image\nExterior view of a black server rack with multiple drive bays (no visible text or labels)\n
\nData Center\n\n
0\n[images/image-27-NVIDIA RTX 4090.jpg]\n\n
\nnatural_image\nBlack and white electronic device with green abstract background, no visible text or symbols\n
\nGaming\n\n
0\n[images/image-28-Photo of a laptop workspace.jpg]\n\n
\ntext_image\nDesktop scene with laptop displaying a webpage interface and a cartoon character, next to a bookshelf and office items\n
\nProfessional Visualization\n\n
0\n[images/image-29-Automotive X-Ray Visualization.jpg]\n\n
\nnatural_image\nX-ray diagram of a car showing internal components and wiring (no text or labels)\n
\nAutomotive\n\\$193.7 billion revenue\n\\$16.0 billion revenue\n\\$3.2 billion revenue\n\\$2.3 billion revenue\nup 68% year on year\nup 41% year on year\nup 70% year on year\nup 39% year on year", + "path": "2026-Annual-Report-Web.pdf/BUSINESS OVERVIEW/Fiscal 2026 Market Platforms", + "metadata": { + "length": 1259, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "platforms", + "address", + "large", + "markets", + "expertise", + "critical", + "This", + "figure", + "rounded", + "nearest", + "tenth", + "billion", + "Due", + "rounding", + "sum", + "figures", + "equal", + "total", + "presented", + "details", + "summary", + "natural", + "image", + "Exterior", + "view", + "black", + "server", + "rack", + "multiple", + "drive", + "bays", + "visible", + "text", + "labels", + "Data", + "Center", + "Black", + "white", + "electronic", + "device", + "green", + "abstract", + "background", + "symbols", + "Gaming", + "Desktop", + "scene", + "laptop", + "displaying", + "webpage", + "interface", + "cartoon", + "character", + "bookshelf", + "office", + "items", + "Professional", + "Visualization", + "ray", + "diagram", + "car", + "showing", + "internal", + "components", + "wiring", + "Automotive", + "193.7", + "revenue", + "16.0", + "3.2", + "2.3", + "68%", + "year", + "41%", + "70%", + "39%" + ], + "keywords": [], + "connect_to": [ + { + "target": "bec5449f-43fd-5108-8163-5854e1ec226e", + "relation": "embeds", + "ref": "[images/image-26-Server Rack Hardware.jpg]", + "position": { + "start": 233, + "end": 275 + } + }, + { + "target": "70fe8e83-b575-5ec9-b7e7-7581e10d0a2f", + "relation": "embeds", + "ref": "[images/image-27-NVIDIA RTX 4090.jpg]", + "position": { + "start": 455, + "end": 492 + } + }, + { + "target": "d7b79090-7714-5962-b8ae-db48ef0e7402", + "relation": "embeds", + "ref": "[images/image-28-Photo of a laptop workspace.jpg]", + "position": { + "start": 670, + "end": 719 + } + }, + { + "target": "a5911826-592a-5302-9332-3f63a208b3da", + "relation": "embeds", + "ref": "[images/image-29-Automotive X-Ray Visualization.jpg]", + "position": { + "start": 940, + "end": 992 + } + } + ] + } + }, + { + "chunk_id": "e7e65b35-4c1d-5dcd-972e-bd08101e9d93", + "type": "text", + "content": "- Launched and scaled the NVIDIA Blackwell Ultra platform – built on the architectural breakthroughs of Blackwell, delivering 50x higher throughput and 35x lower token cost versus Hopper\n- Extended our inference leadership – sweeping MLPerf and SemiAnalysis InferenceMax benchmarks – by leveraging our extreme co-design approach, where GPUs, CPUs, networking, security, software, power delivery, and cooling are architected together as a single system, rather than optimized in isolation to maximize performance per watt\n- Introduced the NVIDIA Vera Rubin platform – including new chips and racks – built for agentic AI and co-designed to deliver up to 10x lower token cost versus Blackwell\n- Entered into a non-exclusive licensing agreement with Groq and introduced the NVIDIA Groq 3 LPX, an accelerator for NVIDIA Vera Rubin, designed to meet the low-latency and large-context demands of agentic systems\n- Announced new and expanding partnerships to cultivate ecosystem growth, unlock new markets, and accelerate NVIDIA platform adoption, including those with Anthropic, Meta, Nokia, and OpenAI\n- Turbocharged AI market development with an accelerating release cadence of NVIDIA open models including Nemotron for agentic AI, Cosmos for physical AI, and Alpamayo for autonomous vehicles\n- Announced key physical AI milestones including \\$6 billion in Fiscal 2026 physical AI revenue, NVIDIA DRIVE-powered L4-ready Mercedes-Benz vehicles, and an Uber partnership to scale the world's largest L4-ready mobility network\n- Advanced quantum computing with new technologies and NVIDIA-powered systems including NVIDIA NVQLink GPU-QPU interconnect and ABCI-Q, the world's largest Quantum research supercomputer", + "path": "2026-Annual-Report-Web.pdf/BUSINESS OVERVIEW/Business Highlights", + "metadata": { + "length": 1705, + "summary": "NVIDIA announced major platform advancements including the Blackwell Ultra platform with 50x higher throughput and 35x lower token cost versus Hopper, extended inference leadership through co-design, and the Vera Rubin platform for agentic AI with up to 10x lower token cost. They entered a licensing agreement with Groq for the Groq 3 LPX accelerator, announced partnerships with Anthropic, Meta, Nokia, and OpenAI, and accelerated AI market development with open models like Nemotron, Cosmos, and Alpamayo. Physical AI milestones include $6 billion in Fiscal 2026 revenue, Mercedes-Benz L4-ready vehicles, and an Uber partnership. Quantum computing advances include NVQLink GPU-QPU interconnect and the ABCI-Q supercomputer.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Launched", + "scaled", + "NVIDIA", + "Blackwell", + "Ultra", + "platform", + "built", + "architectural", + "breakthroughs", + "delivering", + "50x", + "higher", + "throughput", + "35x", + "lower", + "token", + "cost", + "versus", + "Hopper", + "Extended", + "inference", + "leadership", + "sweeping", + "MLPerf", + "SemiAnalysis", + "InferenceMax", + "benchmarks", + "leveraging", + "extreme", + "design", + "approach", + "GPUs", + "CPUs", + "networking", + "security", + "software", + "power", + "delivery", + "cooling", + "architected", + "single", + "system", + "optimized", + "isolation", + "maximize", + "performance", + "watt", + "Introduced", + "Vera", + "Rubin", + "including", + "chips", + "racks", + "agentic", + "AI", + "designed", + "deliver", + "10x", + "Entered", + "exclusive", + "licensing", + "agreement", + "Groq", + "introduced", + "LPX", + "accelerator", + "meet", + "low", + "latency", + "large", + "context", + "demands", + "systems", + "Announced", + "expanding", + "partnerships", + "cultivate", + "ecosystem", + "growth", + "unlock", + "markets", + "accelerate", + "adoption", + "Anthropic", + "Meta", + "Nokia", + "OpenAI", + "Turbocharged", + "market", + "development", + "accelerating", + "release", + "cadence", + "open", + "models", + "Nemotron", + "Cosmos", + "physical", + "Alpamayo", + "autonomous", + "vehicles", + "key", + "milestones", + "billion", + "Fiscal", + "2026", + "revenue", + "DRIVE", + "powered", + "L4", + "ready", + "Mercedes", + "Benz", + "Uber", + "partnership", + "scale", + "world", + "largest", + "mobility", + "network", + "Advanced", + "quantum", + "computing", + "technologies", + "NVQLink", + "GPU", + "QPU", + "interconnect", + "ABCI", + "Quantum", + "research", + "supercomputer" + ], + "keywords": [ + "NVIDIA", + "AI platform", + "Blackwell" + ], + "connect_to": [] + } + }, + { + "chunk_id": "ea554773-c927-54c0-a3cb-715a8f313e39", + "type": "image", + "content": "\n- Turbocharged AI market development with an accelerating release cadence of NVIDIA open models including Nemotron for agentic AI, Cosmos for physical AI, and Alpamayo for autonomous vehicles
0\n[images/image-30-Growth Over Time.jpg]\n", + "path": "images/image-30-Growth Over Time.jpg", + "metadata": { + "length": 237, + "summary": "image-30\n1-Year: 32%\n3-Year: 822%\n5-Year: 1,349%", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-30-Growth Over Time.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "c97f7b16-ce2a-55eb-bba9-75d1dce5efa7", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PeriodValue (%)
1-Year32
3-Year822
5-Year1,349
", + "path": "tables/table-4 Returns by Period.html", + "metadata": { + "length": 361, + "summary": "table-5\nThe table shows investment returns over 1-year (32%), 3-year (822%), and 5-year (1,349%) periods.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-4 Returns by Period.html", + "keywords": [ + "Returns", + "Period", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "3ca7e906-a610-59ed-8ca4-3c297533187a", + "type": "image", + "content": "\n
0\n[images/image-31-Financial Returns by Period.jpg]\n", + "path": "images/image-31-Financial Returns by Period.jpg", + "metadata": { + "length": 56, + "summary": "image-31\n- Chart Title: Dividends and Stock Repurchases\n- Categories (X-Axis): 1-Year, 3-Year, 5-Year\n- Legend: Purple = Dividends, Green = Stock Repurchases\n\nData Points:\n- 1-Year Period:\n - Total Return: $41.4\n - Stock Repurchases: $40.4\n - Dividends: $1.0\n- 3-Year Period:\n - Total Return: $86.3\n - Stock Repurchases: $84.1\n - Dividends: $2.2\n- 5-Year Period:\n - Total Return: $97.2\n - Stock Repurchases: $94.2\n - Dividends: $3.0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-31-Financial Returns by Period.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "9d9e5c0e-c4c6-5233-b7b1-7abc0f650a57", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
MaturityDividends ($)Stock Repurchases ($)
1-Year040.4
3-Year2.284.1
5-Year3.094.2
", + "path": "tables/table-5 Dividend & Buyback.html", + "metadata": { + "length": 461, + "summary": "table-6\nThe table shows dividends and stock repurchases over 1, 3, and 5 years. Dividends increase from $0 to $3.0, while repurchases rise from $40.4 to $94.2.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-5 Dividend & Buyback.html", + "keywords": [ + "Dividends", + "Stock Repurchases", + "Maturity" + ], + "tokens": [] + } + }, + { + "chunk_id": "ef7676b7-6744-58ca-b8b5-c9abfadb51e7", + "type": "text", + "content": "Total Shareholder Return (TSR)\\*\n\n- Turbocharged AI market development with an accelerating release cadence of NVIDIA open models including Nemotron for agentic AI, Cosmos for physical AI, and Alpamayo for autonomous vehicles
0\n[images/image-30-Growth Over Time.jpg]\n\n
\nbar chart\n\n[tables/table-4 Returns by Period.html]\n\n
\n\\*Represents cumulative stock price appreciation with dividends reinvested and is measured for the applicable fiscal year periods based on the closing price (\\$187.67) of NVIDIA's common stock on January 23, 2026, the last trading day before the end of our Fiscal 2026, as reported by Nasdaq.\nReturn of Capital to Shareholders (in billions)\n\n
0\n[images/image-31-Financial Returns by Period.jpg]\n\n
\nstacked bar chart\n\n[tables/table-5 Dividend & Buyback.html]\n\nThe chart displays a single bar for stock repurchases and a separate purple bar for dividends. Values are explicitly labeled on each segment: $41.4 for 1-Year, $86.3 for 3-Year, and $97.2 for 5-Year. The total value per segment is also displayed.\n
\nPlease see our Form 10-K for more financial information for Fiscal 2026.", + "path": "2026-Annual-Report-Web.pdf/BUSINESS OVERVIEW/Fiscal 2026 Shareholder Returns", + "metadata": { + "length": 1208, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Total", + "Shareholder", + "Return", + "TSR", + "Turbocharged", + "AI", + "market", + "development", + "accelerating", + "release", + "cadence", + "NVIDIA", + "open", + "models", + "including", + "Nemotron", + "agentic", + "Cosmos", + "physical", + "Alpamayo", + "autonomous", + "vehicles", + "details", + "summary", + "bar", + "chart", + "Represents", + "cumulative", + "stock", + "price", + "appreciation", + "dividends", + "reinvested", + "measured", + "applicable", + "fiscal", + "year", + "periods", + "based", + "closing", + "187.67", + "common", + "January", + "23", + "2026", + "trading", + "day", + "end", + "Fiscal", + "reported", + "Nasdaq", + "Capital", + "Shareholders", + "billions", + "stacked", + "The", + "displays", + "single", + "repurchases", + "separate", + "purple", + "Values", + "explicitly", + "labeled", + "segment", + "41.4", + "Year", + "86.3", + "97.2", + "total", + "displayed", + "Please", + "Form", + "10", + "financial", + "information" + ], + "keywords": [], + "connect_to": [ + { + "target": "ea554773-c927-54c0-a3cb-715a8f313e39", + "relation": "embeds", + "ref": "[images/image-30-Growth Over Time.jpg]", + "position": { + "start": 237, + "end": 275 + } + }, + { + "target": "c97f7b16-ce2a-55eb-bba9-75d1dce5efa7", + "relation": "embeds", + "ref": "[tables/table-4 Returns by Period.html]", + "position": { + "start": 317, + "end": 356 + } + }, + { + "target": "3ca7e906-a610-59ed-8ca4-3c297533187a", + "relation": "embeds", + "ref": "[images/image-31-Financial Returns by Period.jpg]", + "position": { + "start": 733, + "end": 782 + } + }, + { + "target": "9d9e5c0e-c4c6-5233-b7b1-7abc0f650a57", + "relation": "embeds", + "ref": "[tables/table-5 Dividend & Buyback.html]", + "position": { + "start": 832, + "end": 872 + } + } + ] + } + }, + { + "chunk_id": "0326129a-da62-560f-a0e9-258426926818", + "type": "text", + "content": "This summary highlights information contained elsewhere in the proxy statement. This summary does not contain all the information that you should consider, and you should read the entire proxy statement carefully before voting.\n2026 Annual Meeting of Stockholders\nDate and time: Wednesday, June 24, 2026 at 9:00 a.m. Pacific Time\nLocation: Virtually at www.virtualshareholdermeeting.com/NVDA2026\nRecord date: Stockholders as of April 27, 2026 are entitled to vote\nAdmission to meeting: You will need your Control Number to attend the 2026 Meeting", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY", + "metadata": { + "length": 546, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "This", + "summary", + "highlights", + "information", + "contained", + "proxy", + "statement", + "read", + "entire", + "carefully", + "voting", + "2026", + "Annual", + "Meeting", + "Stockholders", + "Date", + "time", + "Wednesday", + "June", + "24", + "00", + "Pacific", + "Time", + "Location", + "Virtually", + "www", + "virtualshareholdermeeting", + "NVDA2026", + "Record", + "date", + "April", + "27", + "entitled", + "vote", + "Admission", + "meeting", + "You", + "Control", + "Number", + "attend" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e9b71c06-735a-5ada-b262-a09363a145a9", + "type": "table", + "content": "
MatterPageBoard RecommendsVote Required for ApprovalEffect of AbstentionsEffect of Broker Non-Votes
Management Proposals:
1Election of Ten Directors14FOR each director nomineeMore FOR than AGAINST votesNoneNone
2Advisory Approval of Our Executive Compensation37FORMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
3Ratification of the Selection of PwC as Our Independent Registered Public Accounting Firm for Fiscal 202758FORMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstN/A*
Stockholder Proposals:
4Simple Majority Vote Standard61AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
5Report on Faith-Based Community Resource Groups63AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
6Report on Workforce Civil Liberties65AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
7Greenhouse Gas Emissions Disclosure67AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
", + "path": "tables/table-6 Proxy Voting Items.html", + "metadata": { + "length": 1888, + "summary": "table-7\nThe table lists 7 proposals for a shareholder vote, including director elections, executive compensation, auditor ratification, and stockholder proposals, with board recommendations and voting requirements.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-6 Proxy Voting Items.html", + "keywords": [ + "Proposals", + "Voting", + "Board" + ], + "tokens": [] + } + }, + { + "chunk_id": "ffc5a168-5109-5823-aa02-ccb6328d5547", + "type": "text", + "content": "A summary of the 2026 Meeting proposals is below. Every stockholder's vote is important. Our Board urges you to vote your shares FOR each of the nominees listed in Proposal 1, FOR Proposals 2 and 3, and AGAINST Proposals 4, 5, 6, and 7.\n\n[tables/table-6 Proxy Voting Items.html]\n\n\\* Because this is a routine proposal, there are no broker non-votes.", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Voting Matters and Board Recommendations", + "metadata": { + "length": 411, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "summary", + "2026", + "Meeting", + "proposals", + "Every", + "stockholder", + "vote", + "important", + "Our", + "Board", + "urges", + "shares", + "FOR", + "nominees", + "listed", + "Proposal", + "Proposals", + "AGAINST", + "Because", + "routine", + "proposal", + "broker", + "votes" + ], + "keywords": [], + "connect_to": [ + { + "target": "e9b71c06-735a-5ada-b262-a09363a145a9", + "relation": "embeds", + "ref": "[tables/table-6 Proxy Voting Items.html]", + "position": { + "start": 238, + "end": 278 + } + } + ] + } + }, + { + "chunk_id": "c97a9561-8496-5564-a21c-1f109574decc", + "type": "table", + "content": "
NameAgeDirector SinceIndependentFinancial Expert (1)Committee MembershipNumber of Other Public Company Boards
Tench Coxe681993AC, CC
John O. Dabiri462020CC, NCGC
Jen-Hsun Huang631993
Dawn Hudson682013CC Chairperson
Harvey C. Jones731993AC, NCGC
Melissa B. Lora632023AC, NCGC1
Stephen C. Neal(Lead Director)772019NCGC Chairperson
A. Brooke Seawell781997AC Chairperson1
Aarti Shah612020AC, CC1
Mark A. Stevens662008 (2)CC, NCGC
", + "path": "tables/table-7 Board of Directors.html", + "metadata": { + "length": 1235, + "summary": "table-8\nThe table lists 10 board members of a company, including their age, director start year, independence status, financial expertise, committee roles, and other board memberships.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-7 Board of Directors.html", + "keywords": [ + "Directors", + "Committees", + "Tenure" + ], + "tokens": [] + } + }, + { + "chunk_id": "c3058156-0633-5f1c-93d9-c992f60af7c2", + "type": "text", + "content": "The following table provides summary information about each director nominee:\n\n[tables/table-7 Board of Directors.html]\n\n(1) Qualified as an AC financial expert.\n(2) Previously served as a member of our Board from 1993 until 2006.\nThe Board recommends that you vote FOR each director nominee.", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Election of Directors (Proposal 1)", + "metadata": { + "length": 354, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "table", + "summary", + "information", + "director", + "nominee", + "Qualified", + "AC", + "financial", + "expert", + "Previously", + "served", + "member", + "Board", + "1993", + "2006", + "recommends", + "vote", + "FOR" + ], + "keywords": [], + "connect_to": [ + { + "target": "c97a9561-8496-5564-a21c-1f109574decc", + "relation": "embeds", + "ref": "[tables/table-7 Board of Directors.html]", + "position": { + "start": 79, + "end": 119 + } + } + ] + } + }, + { + "chunk_id": "b3ba5747-c0f7-52e1-a256-01284857aef9", + "type": "table", + "content": "
Senior Leadership & Operations ExperienceIndustry & TechnicalFinancial / Financial CommunityGovernance & Public Company BoardEmerging Technologies & Business ModelsMarketing, Communications & Brand ManagementRegulatory, Legal & Risk ManagementHuman Capital Management Experience
Coxe
Dabiri
Huang
Hudson
Jones
Lora
Neal
Seawell
Shah
Stevens
", + "path": "tables/table-8 Board Skills Matrix.html", + "metadata": { + "length": 2210, + "summary": "table-9\nA table showing board members' expertise across eight categories including leadership, industry, finance, governance, emerging tech, marketing, regulatory, and human capital.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-8 Board Skills Matrix.html", + "keywords": [ + "Board", + "Skills", + "Matrix" + ], + "tokens": [] + } + }, + { + "chunk_id": "1d3a1057-62a1-579e-80a4-79a3cf3d7241", + "type": "text", + "content": "Our director nominees reflect a broad range of competencies, professional experience, and backgrounds, and contribute diverse viewpoints and perspectives to our Board. The Board seeks to balance the benefits of continuity and institutional knowledge of our longer-serving directors with the introduction of new perspectives and ideas through the appointment of new directors. In furtherance of this approach, one-third of our independent directors have joined the Board since 2020. The Board, through the NCGC, actively oversees Board composition to maintain an appropriate mix of skills, experience, and perspectives aligned with the Company's strategy and evolving business needs. As a part of this process, the Board also periodically refreshes committee membership and chairpersons to promote a variety of viewpoints and broaden director engagement across key areas of oversight on the Board committees.\nOn May 7, 2026, the Board approved an increase in the size of the Board from ten to eleven directors and appointed Suzanne Nora Johnson to fill the newly created directorship, in each case effective July 13, 2026. Ms. Nora Johnson will join the Board effective July 13, 2026, due to a prior professional commitment. She is expected to join the AC on the effective date of her appointment.\nNominee Skills, Competencies, and Attributes\nBelow are the skills, competencies, and attributes that our NCGC and Board consider important for our directors to have, considering our current business and future market opportunities, and the director nominees who possess them:\n\n[tables/table-8 Board Skills Matrix.html]", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Election of Directors (Proposal 1)/Recent Refreshment and Nominee Qualifications", + "metadata": { + "length": 1675, + "summary": "The text describes the Board's approach to composition, emphasizing a balance of continuity and new perspectives, with one-third of independent directors joining since 2020. It notes the appointment of Suzanne Nora Johnson to the Board effective July 13, 2026, and her expected committee assignment. The NCGC oversees skills and attributes alignment with company strategy.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "director", + "nominees", + "reflect", + "broad", + "range", + "competencies", + "professional", + "experience", + "backgrounds", + "contribute", + "diverse", + "viewpoints", + "perspectives", + "Board", + "The", + "seeks", + "balance", + "benefits", + "continuity", + "institutional", + "knowledge", + "longer", + "serving", + "directors", + "introduction", + "ideas", + "appointment", + "In", + "furtherance", + "approach", + "independent", + "joined", + "2020", + "NCGC", + "actively", + "oversees", + "composition", + "maintain", + "mix", + "skills", + "aligned", + "Company", + "strategy", + "evolving", + "business", + "As", + "part", + "process", + "periodically", + "refreshes", + "committee", + "membership", + "chairpersons", + "promote", + "variety", + "broaden", + "engagement", + "key", + "areas", + "oversight", + "committees", + "On", + "May", + "2026", + "approved", + "increase", + "size", + "ten", + "eleven", + "appointed", + "Suzanne", + "Nora", + "Johnson", + "fill", + "newly", + "created", + "directorship", + "case", + "effective", + "July", + "13", + "Ms", + "join", + "due", + "prior", + "commitment", + "She", + "expected", + "AC", + "date", + "Nominee", + "Skills", + "Competencies", + "Attributes", + "Below", + "attributes", + "important", + "current", + "future", + "market", + "opportunities", + "possess" + ], + "keywords": [ + "Board refreshment", + "Director skills", + "Governance" + ], + "connect_to": [ + { + "target": "b3ba5747-c0f7-52e1-a256-01284857aef9", + "relation": "embeds", + "ref": "[tables/table-8 Board Skills Matrix.html]", + "position": { + "start": 1574, + "end": 1615 + } + } + ] + } + }, + { + "chunk_id": "16ee725d-7348-5a9f-93b6-6bffeb5ce87d", + "type": "text", + "content": "Our Board is committed to strong corporate governance to promote the long-term interests of the Company and our stockholders. We seek a collaborative approach to stockholder issues that affect our business and to ensure that our stockholders see our governance and executive pay practices as well-structured. In Fall 2025, we contacted our top institutional stockholders, representing an aggregate ownership of 34% of our shares, to gain insights into their views on our corporate governance, compensation and corporate sustainability practices. Highlights of our corporate governance practices include:\n√ All directors independent, except for our CEO\n√ Independent Lead Director\n√ Proxy access\n√ Declassified Board\n√ Majority voting for directors\n√ Active Board oversight of risk management\n√ Stockholders can call a special meeting\n√ 75% or greater attendance by each director at meetings of the Board and applicable committees\n√ Independent directors frequently meet in executive sessions\n√ At least annual Board and committee self-assessments\n√ Annual stockholder outreach, including Lead Director participation\n√ Stock ownership guidelines for our directors and NEOs", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Election of Directors (Proposal 1)/Corporate Governance Highlights", + "metadata": { + "length": 1171, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Board", + "committed", + "strong", + "corporate", + "governance", + "promote", + "long", + "term", + "interests", + "Company", + "stockholders", + "We", + "seek", + "collaborative", + "approach", + "stockholder", + "issues", + "affect", + "business", + "ensure", + "executive", + "pay", + "practices", + "structured", + "In", + "Fall", + "2025", + "contacted", + "top", + "institutional", + "representing", + "aggregate", + "ownership", + "34%", + "shares", + "gain", + "insights", + "views", + "compensation", + "sustainability", + "Highlights", + "include", + "All", + "directors", + "independent", + "CEO", + "Independent", + "Lead", + "Director", + "Proxy", + "access", + "Declassified", + "Majority", + "voting", + "Active", + "oversight", + "risk", + "management", + "Stockholders", + "call", + "special", + "meeting", + "75%", + "greater", + "attendance", + "director", + "meetings", + "applicable", + "committees", + "frequently", + "meet", + "sessions", + "At", + "annual", + "committee", + "assessments", + "Annual", + "outreach", + "including", + "participation", + "Stock", + "guidelines", + "NEOs" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3d02077d-5d69-583c-8b12-3979311aef95", + "type": "text", + "content": "We are asking our stockholders to cast a non-binding vote, also known as “say-on-pay,” to approve our NEOs’ compensation. The Board believes that our compensation policies and practices are effective in achieving our goals of paying for performance, providing competitive pay so that we may attract and retain a high-caliber executive team, aligning our executives’ interests with those of our stockholders to create long-term value, and achieving simplicity and transparency with our compensation program. The Board and our stockholders have approved holding our “say-on-pay” votes annually. The Board recommends that you vote FOR Proposal 2.", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Advisory Approval of Executive Compensation for Fiscal 2026 (Proposal 2)", + "metadata": { + "length": 643, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "stockholders", + "cast", + "binding", + "vote", + "pay", + "approve", + "NEOs", + "compensation", + "The", + "Board", + "believes", + "policies", + "practices", + "effective", + "achieving", + "goals", + "paying", + "performance", + "providing", + "competitive", + "attract", + "retain", + "high", + "caliber", + "executive", + "team", + "aligning", + "executives", + "interests", + "create", + "long", + "term", + "simplicity", + "transparency", + "program", + "approved", + "holding", + "votes", + "annually", + "recommends", + "FOR", + "Proposal" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "42e15964-8518-5d56-bb53-07e534b3c1b3", + "type": "text", + "content": "Our executive compensation program is designed to reward performance. We use compensation elements that align our NEOs' interests with those of stockholders to drive long-term value. Our NEOs' compensation is heavily weighted toward performance-based, \"at-risk\" variable cash and long-term equity awards, which are earned only if the Company achieves pre-established financial metrics, with caps on maximum payouts. In recent years, over 90% of our CEO's and over 45% of our other NEOs' total target pay have been performance-based and at-risk. Additionally, 100% of our CEO's equity awards have been granted solely as PSUs.\nFor Fiscal 2026, the CC kept NEO cash compensation constant as compared to Fiscal 2025, but increased target equity opportunities for all NEOs to recognize the complexity and scope of their roles and responsibilities. The CC also set Fiscal 2026 target equity opportunities at levels that aligned total target pay for all NEOs other than our CEO to reflect their comparable, significant contributions and impact on corporate performance. To motivate executives, the CC set the Base Compensation Plan revenue goal for the Fiscal 2026 Variable Cash Plan in line with our record Fiscal 2025 results and the Base Compensation Plan Non-GAAP Operating Income goal for SY PSUs granted in Fiscal 2026 in line with our corresponding Fiscal 2025 Stretch Compensation Plan goal.", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Advisory Approval of Executive Compensation for Fiscal 2026 (Proposal 2)/Executive Compensation Highlights", + "metadata": { + "length": 1392, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "executive", + "compensation", + "program", + "designed", + "reward", + "performance", + "We", + "elements", + "align", + "NEOs", + "interests", + "stockholders", + "drive", + "long", + "term", + "heavily", + "weighted", + "based", + "risk", + "variable", + "cash", + "equity", + "awards", + "earned", + "Company", + "achieves", + "pre", + "established", + "financial", + "metrics", + "caps", + "maximum", + "payouts", + "In", + "recent", + "years", + "90%", + "CEO", + "45%", + "total", + "target", + "pay", + "Additionally", + "100%", + "granted", + "solely", + "PSUs", + "For", + "Fiscal", + "2026", + "CC", + "NEO", + "constant", + "compared", + "2025", + "increased", + "opportunities", + "recognize", + "complexity", + "scope", + "roles", + "responsibilities", + "The", + "set", + "levels", + "aligned", + "reflect", + "comparable", + "significant", + "contributions", + "impact", + "corporate", + "To", + "motivate", + "executives", + "Base", + "Compensation", + "Plan", + "revenue", + "goal", + "Variable", + "Cash", + "line", + "record", + "results", + "Non", + "GAAP", + "Operating", + "Income", + "SY", + "Stretch" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ddded897-59b4-55e7-8be6-65923a97881e", + "type": "table", + "content": "
PERFORMANCE GOALS
Variable Cash PlanSY PSUsMY PSUs
Fiscal 2026 RevenuePayout as a % of Target OpportunityFiscal 2026 Non-GAAP Operating Income (1)Shares Eligible to Vest as a % of Target OpportunityFiscal 2024 to 2026 3-Year Relative TSRShares Eligible to Vest as a % of Target Opportunity
Threshold$100.0 billion50%$46.0 billion50%25th percentile25%
Base Compensation Plan$130.0 billion100%$71.0 billion100%50th percentile100%
Stretch Compensation Plan$160.0 billion (2)200%$96.0 billion (2)CEO 150%Other NEOs 200%75th percentileCEO 150%Other NEOs 200%
", + "path": "tables/table-9 Performance Goals.html", + "metadata": { + "length": 965, + "summary": "table-10\nThe table outlines performance goals for Variable Cash Plan, SY PSUs, and MY PSUs with thresholds, base, and stretch targets for Fiscal 2026 Revenue, Non-GAAP Operating Income, and 3-Year Relative TSR.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-9 Performance Goals.html", + "keywords": [ + "Performance Goals", + "Variable Cash", + "PSUs" + ], + "tokens": [] + } + }, + { + "chunk_id": "1699db01-aec5-50a5-85fb-57568480c5da", + "type": "table", + "content": "
PERFORMANCE ACHIEVEMENT AND PAYOUTS
Variable Cash PlanSY PSUsMY PSUs
Performance Achievement for Period Ended Fiscal 2026 (3)$215.9 billion revenue$137.3 billion Non-GAAP Operating Income (1)100th percentile 3-year TSR relative to S&P 500 (1,055%)
Payout as % of Target Opportunity200%CEO 150% Other NEOs 200%CEO 150% Other NEOs 200%
", + "path": "tables/table-10 Performance Payouts.html", + "metadata": { + "length": 514, + "summary": "table-11\nTable shows performance achievement for Fiscal 2026: $215.9B revenue, $137.3B Non-GAAP Operating Income, 100th percentile TSR. Payouts range from 150% to 200% of target.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-10 Performance Payouts.html", + "keywords": [ + "Performance", + "Payouts", + "PSUs" + ], + "tokens": [] + } + }, + { + "chunk_id": "d84c65cb-b726-5faa-8681-511b273630e2", + "type": "text", + "content": "As described in CD&A, a significant portion of our executive pay opportunities are tied to the achievement of financial measures that drive business value and contribute to our long-term success. The tables below show our goals for the applicable periods that were completed at the end of Fiscal 2026, performance achievement against the goals, and their impacts on our executives' pay.\n\n[tables/table-9 Performance Goals.html]\n\n\n[tables/table-10 Performance Payouts.html]\n\n(1) See Reconciliation of Non-GAAP Financial Measures in CD&A for a reconciliation between the non-GAAP financial measures and GAAP results.\n(2) Our original Fiscal 2026 Stretch Compensation Plan revenue goal of \\$190.0 billion was automatically adjusted to \\$160.0 billion, and our original Fiscal 2026 Stretch Compensation Plan Non-GAAP Operating Income goal of \\$120.0 billion was automatically adjusted to \\$96.0 billion, due to the imposition of additional export controls on the Company's H2O products during the first half of Fiscal 2026. See Fiscal 2026 Compensation Actions and Achievements—Performance Metrics and Goals for Executive Compensation in CD&A for further discussion of these adjustments.\n(3) See Performance Metrics and Goals for Executive Compensation in CD&A for a description and further discussion of revenue, Non-GAAP Operating Income, and 3-year relative TSR.\nRatification of Selection of PwC as our Independent Registered Public Accounting Firm for Fiscal 2027 (Proposal 3)\nWhile not required, we are asking our stockholders to ratify the AC's selection of PwC as our independent registered public accounting firm for Fiscal 2027, as we believe it is a matter of good corporate practice. If our stockholders do not ratify the selection, the AC will reconsider the appointment, but may nevertheless retain PwC. Even if the selection is ratified, the AC may select a different independent registered public accounting firm at any time if it determines that such a change would be in the best interests of NVIDIA and our stockholders. The Board recommends that you vote FOR Proposal 3.", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Advisory Approval of Executive Compensation for Fiscal 2026 (Proposal 2)/Financial Performance and Link to Executive Pay", + "metadata": { + "length": 2090, + "summary": "The passage discusses executive pay opportunities tied to financial measures driving business value. It references tables showing goals for Fiscal 2026, performance achievement, and impacts on pay. Adjustments to revenue and Non-GAAP Operating Income goals due to export controls are noted. The text also covers a proposal to ratify PwC as independent auditor for Fiscal 2027.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "As", + "CD", + "significant", + "portion", + "executive", + "pay", + "opportunities", + "tied", + "achievement", + "financial", + "measures", + "drive", + "business", + "contribute", + "long", + "term", + "success", + "The", + "tables", + "show", + "goals", + "applicable", + "periods", + "completed", + "end", + "Fiscal", + "2026", + "performance", + "impacts", + "executives", + "See", + "Reconciliation", + "Non", + "GAAP", + "Financial", + "Measures", + "reconciliation", + "results", + "Our", + "original", + "Stretch", + "Compensation", + "Plan", + "revenue", + "goal", + "190.0", + "billion", + "automatically", + "adjusted", + "160.0", + "Operating", + "Income", + "120.0", + "96.0", + "due", + "imposition", + "additional", + "export", + "controls", + "Company", + "H2O", + "products", + "half", + "Actions", + "Achievements", + "Performance", + "Metrics", + "Goals", + "Executive", + "discussion", + "adjustments", + "description", + "year", + "relative", + "TSR", + "Ratification", + "Selection", + "PwC", + "Independent", + "Registered", + "Public", + "Accounting", + "Firm", + "2027", + "Proposal", + "While", + "required", + "stockholders", + "ratify", + "AC", + "selection", + "independent", + "registered", + "public", + "accounting", + "firm", + "matter", + "good", + "corporate", + "practice", + "If", + "reconsider", + "appointment", + "retain", + "Even", + "ratified", + "select", + "time", + "determines", + "change", + "interests", + "NVIDIA", + "Board", + "recommends", + "vote", + "FOR" + ], + "keywords": [ + "executive compensation", + "performance goals", + "financial measures" + ], + "connect_to": [ + { + "target": "ddded897-59b4-55e7-8be6-65923a97881e", + "relation": "embeds", + "ref": "[tables/table-9 Performance Goals.html]", + "position": { + "start": 388, + "end": 427 + } + }, + { + "target": "1699db01-aec5-50a5-85fb-57568480c5da", + "relation": "embeds", + "ref": "[tables/table-10 Performance Payouts.html]", + "position": { + "start": 430, + "end": 472 + } + } + ] + } + }, + { + "chunk_id": "4bdc6ee8-da6c-524c-a98b-1ad2f85d57e2", + "type": "text", + "content": "A stockholder is asking our Board to take steps to replace the supermajority voting provisions in our Charter and Bylaws with a simple majority standard. The proposal is advisory only and approval of the proposal would not, by itself, implement a simple majority voting standard. At our 2025 Meeting, the Board proposed and recommended that stockholders approve an amendment and restatement of our Charter to eliminate the remaining supermajority voting provisions, which did not pass. The Board recommends that you vote AGAINST Proposal 4.", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Stockholder Proposal 4", + "metadata": { + "length": 540, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "stockholder", + "Board", + "steps", + "replace", + "supermajority", + "voting", + "provisions", + "Charter", + "Bylaws", + "simple", + "majority", + "standard", + "The", + "proposal", + "advisory", + "approval", + "implement", + "At", + "2025", + "Meeting", + "proposed", + "recommended", + "stockholders", + "approve", + "amendment", + "restatement", + "eliminate", + "remaining", + "pass", + "recommends", + "vote", + "AGAINST", + "Proposal" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2a2412d9-172e-5242-881e-24352cbb4cfb", + "type": "text", + "content": "A stockholder is asking our Board to evaluate and report on faith-based community resource groups. NVIDIA is committed to ensuring that our employees are protected and supported through enterprise-wide policies, practices, and reporting mechanisms. Our Board believes that our existing policies, practices, and feedback mechanisms already address the issues raised in Proposal 5. The Board recommends that you vote AGAINST Proposal 5.", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Stockholder Proposal 5", + "metadata": { + "length": 434, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "stockholder", + "Board", + "evaluate", + "report", + "faith", + "based", + "community", + "resource", + "groups", + "NVIDIA", + "committed", + "ensuring", + "employees", + "protected", + "supported", + "enterprise", + "wide", + "policies", + "practices", + "reporting", + "mechanisms", + "Our", + "believes", + "existing", + "feedback", + "address", + "issues", + "raised", + "Proposal", + "The", + "recommends", + "vote", + "AGAINST" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0404129d-38cd-5b02-8c8c-447998e26be2", + "type": "text", + "content": "A stockholder is asking our Board to evaluate and report on civil rights and non-discrimination related to DEI. NVIDIA complies with applicable law and is committed to equal employment opportunity requirements. In addition, the Board maintains strong oversight and compliance mechanisms to ensure these policies and practices are effectively implemented. Our Board believes these oversight structures provide sufficient and comprehensive governance and accountability for the policies and practices addressed by Proposal 6. The Board recommends that you vote AGAINST Proposal 6.", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Stockholder Proposal 6", + "metadata": { + "length": 578, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "stockholder", + "Board", + "evaluate", + "report", + "civil", + "rights", + "discrimination", + "related", + "DEI", + "NVIDIA", + "complies", + "applicable", + "law", + "committed", + "equal", + "employment", + "opportunity", + "requirements", + "In", + "addition", + "maintains", + "strong", + "oversight", + "compliance", + "mechanisms", + "ensure", + "policies", + "practices", + "effectively", + "implemented", + "Our", + "believes", + "structures", + "provide", + "sufficient", + "comprehensive", + "governance", + "accountability", + "addressed", + "Proposal", + "The", + "recommends", + "vote", + "AGAINST" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b19f6723-a820-5b1a-8cb5-51d293df11eb", + "type": "text", + "content": "A stockholder is asking our Board to report and disclose GHG emissions from the use of the Company's sold products. Our Board believes Proposal 7 is unnecessary given NVIDIA's history of robust climate reporting and targets, and our future disclosure plans. The Board believes NVIDIA's management is best positioned to make decisions regarding the specific content and methodology of our climate disclosures. The Board recommends that you vote AGAINST Proposal 7.", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Stockholder Proposal 7", + "metadata": { + "length": 463, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "stockholder", + "Board", + "report", + "disclose", + "GHG", + "emissions", + "Company", + "sold", + "products", + "Our", + "believes", + "Proposal", + "unnecessary", + "NVIDIA", + "history", + "robust", + "climate", + "reporting", + "targets", + "future", + "disclosure", + "plans", + "The", + "management", + "positioned", + "make", + "decisions", + "specific", + "content", + "methodology", + "disclosures", + "recommends", + "vote", + "AGAINST" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0a8d89a5-6c73-5f05-9d87-10f5ea5898a5", + "type": "text", + "content": "ROBERT K. BURGESS\nThe Board of Directors and the entire NVIDIA family mourn the passing of Robert K. Burgess, who served as a valued member of our Board from 2011 to 2025.\nOver his 14 years of dedicated service, Rob provided exceptional leadership and unwavering commitment to the Company and its stockholders, offering invaluable guidance throughout a period of significant growth. Beyond his many professional contributions, Rob was admired for his warmth, candor, and sense of humor. He will be deeply missed.\nPROXY STATEMENT FOR THE 2026 ANNUAL MEETING OF STOCKHOLDERS - JUNE 24, 2026", + "path": "2026-Annual-Report-Web.pdf/IN MEMORIAM ROBERT K. BURGESS", + "metadata": { + "length": 588, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "ROBERT", + "BURGESS", + "The", + "Board", + "Directors", + "entire", + "NVIDIA", + "family", + "mourn", + "passing", + "Robert", + "Burgess", + "served", + "valued", + "member", + "2011", + "2025", + "Over", + "14", + "years", + "dedicated", + "service", + "Rob", + "provided", + "exceptional", + "leadership", + "unwavering", + "commitment", + "Company", + "stockholders", + "offering", + "invaluable", + "guidance", + "period", + "significant", + "growth", + "Beyond", + "professional", + "contributions", + "admired", + "warmth", + "candor", + "sense", + "humor", + "He", + "deeply", + "missed", + "PROXY", + "STATEMENT", + "FOR", + "THE", + "2026", + "ANNUAL", + "MEETING", + "OF", + "STOCKHOLDERS", + "JUNE", + "24" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d9da096c-e1fa-59ce-85d0-e7644766ace5", + "type": "text", + "content": "Your proxy is being solicited on behalf of the Board for use at the 2026 Meeting. Our 2026 Meeting will take place virtually on Wednesday, June 24, 2026 at 9:00 a.m. Pacific Time.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting", + "metadata": { + "length": 179, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Your", + "proxy", + "solicited", + "behalf", + "Board", + "2026", + "Meeting", + "Our", + "place", + "virtually", + "Wednesday", + "June", + "24", + "00", + "Pacific", + "Time" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8987afed-952d-5f25-accb-1cade69300fd", + "type": "text", + "content": "The Board believes holding the 2026 Meeting in a virtual format invites participation from stockholders regardless of location, while reducing costs to stockholders and the Company. This balance allows the 2026 Meeting to remain focused on matters directly relevant to stockholder interests in an efficient way. We have designed the virtual format to protect stockholder rights, including by offering multiple opportunities to ask questions, publishing answers to questions received for the 2026 Meeting on our Investor Relations website, and providing a replay of the webcast after the 2026 Meeting.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Virtual Meeting Philosophy and Benefits", + "metadata": { + "length": 600, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Board", + "believes", + "holding", + "2026", + "Meeting", + "virtual", + "format", + "invites", + "participation", + "stockholders", + "location", + "reducing", + "costs", + "Company", + "This", + "balance", + "remain", + "focused", + "matters", + "directly", + "relevant", + "stockholder", + "interests", + "efficient", + "We", + "designed", + "protect", + "rights", + "including", + "offering", + "multiple", + "opportunities", + "questions", + "publishing", + "answers", + "received", + "Investor", + "Relations", + "website", + "providing", + "replay", + "webcast" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ef2fbe20-cacd-5af5-a589-3d015bb21949", + "type": "text", + "content": "If you were an NVIDIA stockholder as of the close of business on the April 27, 2026 record date, or if you hold a valid proxy, you can attend, ask questions before and during, and vote at our 2026 Meeting at www.virtualshareholdermeeting.com/NVDA2026. Our 2026 Meeting will be held virtually; use the Control Number included on your Notice or printed proxy card to enter. Anyone can also listen to the 2026 Meeting live; if you encounter any difficulties accessing the virtual 2026 Meeting, please call the technical support number available at www.virtualshareholdermeeting.com/NVDA2026.\nA replay of the webcast will be available at www.nvidia.com/proxy through June 24, 2027.\nEven if you plan to attend the 2026 Meeting virtually, we recommend that you also vote by proxy as described below so that your vote will be counted if you later decide not to attend.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Meeting Attendance", + "metadata": { + "length": 861, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "If", + "NVIDIA", + "stockholder", + "close", + "business", + "April", + "27", + "2026", + "record", + "date", + "hold", + "valid", + "proxy", + "attend", + "questions", + "vote", + "Meeting", + "www", + "virtualshareholdermeeting", + "NVDA2026", + "Our", + "held", + "virtually", + "Control", + "Number", + "included", + "Notice", + "printed", + "card", + "enter", + "Anyone", + "listen", + "live", + "encounter", + "difficulties", + "accessing", + "virtual", + "call", + "technical", + "support", + "number", + "replay", + "webcast", + "nvidia", + "June", + "24", + "2027", + "Even", + "plan", + "recommend", + "counted", + "decide" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9bc4f3f1-f4e3-5970-b414-1d6913c6b3b1", + "type": "text", + "content": "We encourage stockholders to submit questions through our pre-meeting forum at www.proxyvote.com (using the Control Number included on your Notice or printed proxy card), as well as during the 2026 Meeting at www.virtualshareholdermeeting.com/NVDA2026. During the 2026 Meeting, we will answer as many stockholder-submitted questions related to the business of the 2026 Meeting as time permits. We will publish on our Investor Relations website answers to questions as soon as practicable after the 2026 Meeting. We intend to group questions and answers by topic and answer substantially similar questions only once. To promote fairness to all stockholders and efficient use of the Company's resources, we will respond to one question per stockholder. We reserve the right to exclude questions that are not pertinent to company business or are otherwise unsuitable for the conduct of the 2026 Meeting.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Asking Questions", + "metadata": { + "length": 900, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "encourage", + "stockholders", + "submit", + "questions", + "pre", + "meeting", + "forum", + "www", + "proxyvote", + "Control", + "Number", + "included", + "Notice", + "printed", + "proxy", + "card", + "2026", + "Meeting", + "virtualshareholdermeeting", + "NVDA2026", + "During", + "answer", + "stockholder", + "submitted", + "related", + "business", + "time", + "permits", + "publish", + "Investor", + "Relations", + "website", + "answers", + "practicable", + "intend", + "group", + "topic", + "substantially", + "similar", + "To", + "promote", + "fairness", + "efficient", + "Company", + "resources", + "respond", + "question", + "reserve", + "exclude", + "pertinent", + "company", + "unsuitable", + "conduct" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e6deeb82-d6f2-5154-bad5-73055385dcf8", + "type": "text", + "content": "To hold our 2026 Meeting, we need a majority of the outstanding shares entitled to vote at the close of business on the April 27, 2026 record date, or a quorum, represented at the 2026 Meeting either by attendance virtually or by proxy. On April 27, 2026, there were 24,220,525,225 shares of common stock outstanding and entitled to vote, meaning 12,110,262,613 shares must be represented at the 2026 Meeting or by proxy to have a quorum. A list of stockholders entitled to vote at the close of business on the record date will be available at our headquarters, 2788 San Tomas Expressway, Santa Clara, California, for 10 days prior to the 2026 Meeting to stockholders of record for any legally valid purpose germane to the 2026 Meeting. To schedule an appointment to view the stockholder list during this period, please contact us at shareholdermeeting@nvidia.com.\nYour shares will count towards the quorum only if you submit a valid proxy or vote at the 2026 Meeting. Abstentions and broker non-votes will count towards the quorum requirement. If there is not a quorum, a majority of the votes present may adjourn the 2026 Meeting to another date.\nFor Proposal 1, you may vote FOR or AGAINST each Board nominee, or you may ABSTAIN from voting. For each other matter to be voted on, you may vote FOR or AGAINST or ABSTAIN from voting.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Quorum and Voting", + "metadata": { + "length": 1334, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "To", + "hold", + "2026", + "Meeting", + "majority", + "outstanding", + "shares", + "entitled", + "vote", + "close", + "business", + "April", + "27", + "record", + "date", + "quorum", + "represented", + "attendance", + "virtually", + "proxy", + "On", + "24", + "220", + "525", + "225", + "common", + "stock", + "meaning", + "12", + "110", + "262", + "613", + "list", + "stockholders", + "headquarters", + "2788", + "San", + "Tomas", + "Expressway", + "Santa", + "Clara", + "California", + "10", + "days", + "prior", + "legally", + "valid", + "purpose", + "germane", + "schedule", + "appointment", + "view", + "stockholder", + "period", + "contact", + "shareholdermeeting", + "nvidia", + "Your", + "count", + "submit", + "Abstentions", + "broker", + "votes", + "requirement", + "If", + "present", + "adjourn", + "For", + "Proposal", + "FOR", + "AGAINST", + "Board", + "nominee", + "ABSTAIN", + "voting", + "matter", + "voted" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "57cf9e4a-a066-5c2b-8456-dbc68eb54eb8", + "type": "table", + "content": "
VoteChange Your VoteRevoke Your Proxy
Virtually attend and vote at the 2026 Meeting
Via mail, by signing and mailing your proxy card to us before the 2026 Meeting
By telephone or online, by following the instructions provided in the Notice or your proxy materials
Submit another properly completed proxy card with a later date
Send a written notice that you are revoking your proxy to NVIDIA Corporation, 2788 San Tomas Expressway, Santa Clara, California 95051, Attention: Timothy S. Teter, Secretary, or via email to shareholdermeeting@nvidia.com
", + "path": "tables/table-11 Proxy Voting.html", + "metadata": { + "length": 835, + "summary": "table-12\nThe table outlines methods for voting, changing votes, or revoking proxies for the 2026 meeting, including virtual attendance, mail, telephone, online, written notice, or submitting a later proxy card.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-11 Proxy Voting.html", + "keywords": [ + "proxy", + "voting", + "revocation" + ], + "tokens": [] + } + }, + { + "chunk_id": "64898a7f-adbe-51c1-b583-5b969812c9f6", + "type": "text", + "content": "You are a stockholder of record if your shares were registered directly in your name with our transfer agent, Computershare, on April 27, 2026. You can vote shares, change your vote, or revoke your proxy before the final vote at the 2026 Meeting in any of the following ways:\n\n[tables/table-11 Proxy Voting.html]\n\nIf you do not vote using any of the ways described above, your shares will not be voted.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Quorum and Voting/Stockholder of Record", + "metadata": { + "length": 429, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "You", + "stockholder", + "record", + "shares", + "registered", + "directly", + "transfer", + "agent", + "Computershare", + "April", + "27", + "2026", + "vote", + "change", + "revoke", + "proxy", + "final", + "Meeting", + "ways", + "If", + "voted" + ], + "keywords": [], + "connect_to": [ + { + "target": "57cf9e4a-a066-5c2b-8456-dbc68eb54eb8", + "relation": "embeds", + "ref": "[tables/table-11 Proxy Voting.html]", + "position": { + "start": 277, + "end": 312 + } + } + ] + } + }, + { + "chunk_id": "3836697f-4775-5728-b264-7e8a6f82bf82", + "type": "text", + "content": "If your shares were held through a nominee, such as a bank or broker, as of April 27, 2026, then you were the beneficial owner of shares held in “street name,” and you have the right to direct the nominee how to vote those shares for the 2026 Meeting. The nominee should provide you a separate Notice or voting instructions, which you should follow to tell the nominee how to vote. To vote by attending the 2026 Meeting virtually, you must obtain a valid proxy from your nominee.\nIf you are a beneficial holder and do not provide voting instructions to your nominee, the nominee will not be authorized to vote your shares on “non-routine” matters, including elections of directors (even if not contested) and executive compensation (including any advisory stockholder votes on executive compensation), and stockholder proposals. This is called a “broker non-vote.” However, the nominee can still register your shares as being present at the 2026 Meeting for determining quorum, and the nominee will have discretion to vote for matters considered by the NYSE to be “routine,” including Proposal 3 regarding the ratification of the selection of our independent registered public accounting firm. If you are a beneficial owner and want to ensure that all of the shares you beneficially own are voted for or against Proposal 3, you must give your nominee specific instructions to do so or the nominee will have discretion to vote on that proposal.\nFor Proposals 1, 2, 4, 5, 6, and 7, which are “non-discretionary” items, you MUST give your nominee instructions in order for your vote to be counted. We strongly encourage you to vote.\nAny NVIDIA stockholder whose shares are held in street name by a member brokerage firm may revoke a proxy and vote their shares at the 2026 Meeting only in accordance with applicable rules and procedures of the national stock exchanges, as used by the holder's brokerage firm.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Quorum and Voting/Street Name Holder", + "metadata": { + "length": 1906, + "summary": "This passage explains voting procedures for beneficial owners whose shares are held in street name by a nominee like a bank or broker. It details that such owners can direct the nominee how to vote, and must obtain a valid proxy to vote virtually. For non-routine matters like director elections and executive compensation, the nominee cannot vote without instructions, resulting in a broker non-vote. However, the nominee has discretion to vote on routine matters like ratifying the accounting firm. For non-discretionary proposals, specific instructions are required. Shares held in street name can only be revoked per exchange rules.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "If", + "shares", + "held", + "nominee", + "bank", + "broker", + "April", + "27", + "2026", + "beneficial", + "owner", + "street", + "direct", + "vote", + "Meeting", + "The", + "provide", + "separate", + "Notice", + "voting", + "instructions", + "follow", + "To", + "attending", + "virtually", + "obtain", + "valid", + "proxy", + "holder", + "authorized", + "routine", + "matters", + "including", + "elections", + "directors", + "contested", + "executive", + "compensation", + "advisory", + "stockholder", + "votes", + "proposals", + "This", + "called", + "However", + "register", + "present", + "determining", + "quorum", + "discretion", + "considered", + "NYSE", + "Proposal", + "ratification", + "selection", + "independent", + "registered", + "public", + "accounting", + "firm", + "ensure", + "beneficially", + "voted", + "give", + "specific", + "proposal", + "For", + "Proposals", + "discretionary", + "items", + "MUST", + "order", + "counted", + "We", + "strongly", + "encourage", + "Any", + "NVIDIA", + "member", + "brokerage", + "revoke", + "accordance", + "applicable", + "rules", + "procedures", + "national", + "stock", + "exchanges" + ], + "keywords": [ + "beneficial owner", + "broker non-vote", + "street name" + ], + "connect_to": [] + } + }, + { + "chunk_id": "1de5cdb6-4047-520f-9528-5a02b5939789", + "type": "table", + "content": "
Proposal NumberProposal DescriptionVote Required for ApprovalEffect of AbstentionsBroker Non-Votes
1Election of ten directorsMore FOR than AGAINST votesNoneNone
2Advisory approval of our executive compensationMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
3Ratification of the selection of PwC as our independent registered public accounting firm for Fiscal 2027Majority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstN/A*
4Stockholder Proposal: Simple Majority Vote StandardMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
5Stockholder Proposal: Report on Faith-Based Community Resource GroupsMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
6Stockholder Proposal: Report on Workforce Civil LibertiesMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
7Stockholder Proposal: Greenhouse Gas Emissions DisclosureMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
", + "path": "tables/table-12 Proposal Voting.html", + "metadata": { + "length": 1652, + "summary": "table-13\nSeven proposals including director election, executive compensation, auditor ratification, and stockholder proposals with voting requirements and abstention effects.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-12 Proposal Voting.html", + "keywords": [ + "Vote", + "Proposals", + "Directors" + ], + "tokens": [] + } + }, + { + "chunk_id": "5f7353d4-8cc8-5503-a433-427bb228bfa8", + "type": "text", + "content": "On each matter to be voted upon, stockholders have one vote for each share of NVIDIA common stock owned as of April 27, 2026. Votes will be counted by the inspector of election as follows:\n\n[tables/table-12 Proposal Voting.html]\n\n\\* Because this is a routine proposal, there are no broker non-votes.\nIf you are a stockholder of record and you return a signed proxy card without marking any selections, your shares will be voted FOR each of the nominees listed in Proposal 1, FOR Proposals 2 and 3, and AGAINST Proposals 4, 5, 6, and 7. If any other matter is properly presented at the 2026 Meeting, one of the proxyholders will vote your shares using his best judgment.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Quorum and Voting/Vote Count", + "metadata": { + "length": 734, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "On", + "matter", + "voted", + "stockholders", + "vote", + "share", + "NVIDIA", + "common", + "stock", + "owned", + "April", + "27", + "2026", + "Votes", + "counted", + "inspector", + "election", + "Because", + "routine", + "proposal", + "broker", + "votes", + "If", + "stockholder", + "record", + "return", + "signed", + "proxy", + "card", + "marking", + "selections", + "shares", + "FOR", + "nominees", + "listed", + "Proposal", + "Proposals", + "AGAINST", + "properly", + "presented", + "Meeting", + "proxyholders", + "judgment" + ], + "keywords": [], + "connect_to": [ + { + "target": "1de5cdb6-4047-520f-9528-5a02b5939789", + "relation": "embeds", + "ref": "[tables/table-12 Proposal Voting.html]", + "position": { + "start": 190, + "end": 228 + } + } + ] + } + }, + { + "chunk_id": "11c5042f-101c-544e-b4dc-fa836db3074e", + "type": "text", + "content": "Preliminary voting results will be announced at the 2026 Meeting. Final voting results will be published in a current report on Form 8-K, which will be filed with the SEC by June 30, 2026.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Quorum and Voting/Vote Results", + "metadata": { + "length": 188, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Preliminary", + "voting", + "results", + "announced", + "2026", + "Meeting", + "Final", + "published", + "current", + "report", + "Form", + "filed", + "SEC", + "June", + "30" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6709a18f-28aa-5de0-888a-e4edede4faa3", + "type": "text", + "content": "As permitted by SEC rules, we are making our proxy materials available to stockholders online at www.nvidia.com/proxy. On or about May 12, 2026, we sent stockholders who owned our common stock at the close of business on April 27, 2026 (except those who previously requested electronic or paper delivery) a Notice with instructions on how to access our proxy materials, vote online or by telephone, and elect to receive future proxy materials electronically or in printed form by mail.\nIf you choose to receive future proxy materials electronically (via www.proxyvote.com for stockholders of record and www.icsdelivery.com/nvda for street name holders), you will receive an email next year with links to the proxy materials and proxy voting site.\nSEC rules also permit companies and intermediaries, such as brokers, to satisfy Notice and proxy material delivery requirements for multiple stockholders with the same address by delivering a single Notice or set of proxy materials addressed to those stockholders. We follow this practice, known as “householding,” unless we have received contrary instructions from any stockholder at that address.\nIf you received more than one Notice or full set of proxy materials, your shares are either registered in more than one name or held in different accounts. Please vote the shares covered by each Notice or proxy card. To modify your instructions so that you receive one Notice or proxy card for each account or name, contact your broker. Your “householding” election will continue until you are notified otherwise or until you revoke your consent.\nTo change how you receive proxy materials (electronically or in print), or to request receipt of a separate set of documents to a household, contact our Investor Relations Department through our website at www.nvidia.com, by email to shareholdermeeting@nvidia.com, by phone at (408) 486-2000, or by mail at 2788 San Tomas Expressway, Santa Clara, California 95051.\nWe will pay the entire cost of soliciting proxies. Our directors and employees may also solicit proxies in person, by telephone, by mail, via the Internet, or by other means. Our directors and employees will not receive any additional compensation for soliciting proxies. We have also retained MacKenzie Partners on an advisory basis for an approximate fee of \\$20,000; they may help us solicit proxies from brokers, bank nominees, and other institutional owners. We may also reimburse brokerage firms, banks, and other agents for the cost of forwarding proxy materials to beneficial owners.\n2027 Meeting Deadlines for Submission of Stockholder Proposals, Nomination of Directors, and Other Business\nProposals to be Considered for Inclusion in Our Proxy Materials Pursuant to Rule 14a-8\nStockholders who wish to present proposals pursuant to Rule 14a-8 promulgated under the Exchange Act for inclusion in our 2027 Meeting proxy materials must submit their proposals in writing to NVIDIA Corporation, 2788 San Tomas Expressway, Santa Clara, California 95051, Attention: Timothy S. Teter, Secretary, or by email to shareholdermeeting@nvidia.com, on or before January 12, 2027.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Proxy Materials", + "metadata": { + "length": 3132, + "summary": "NVIDIA provides proxy materials online, with notices sent to stockholders. Householding is used for multiple stockholders at the same address. Stockholders can change delivery preferences. Proxy solicitation costs are paid by NVIDIA, with MacKenzie Partners assisting. Deadlines for 2027 stockholder proposals under Rule 14a-8 are January 12, 2027.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "As", + "permitted", + "SEC", + "rules", + "making", + "proxy", + "materials", + "stockholders", + "online", + "www", + "nvidia", + "On", + "May", + "12", + "2026", + "owned", + "common", + "stock", + "close", + "business", + "April", + "27", + "previously", + "requested", + "electronic", + "paper", + "delivery", + "Notice", + "instructions", + "access", + "vote", + "telephone", + "elect", + "receive", + "future", + "electronically", + "printed", + "form", + "mail", + "If", + "choose", + "proxyvote", + "record", + "icsdelivery", + "nvda", + "street", + "holders", + "email", + "year", + "links", + "voting", + "site", + "permit", + "companies", + "intermediaries", + "brokers", + "satisfy", + "material", + "requirements", + "multiple", + "address", + "delivering", + "single", + "set", + "addressed", + "We", + "follow", + "practice", + "householding", + "received", + "contrary", + "stockholder", + "full", + "shares", + "registered", + "held", + "accounts", + "Please", + "covered", + "card", + "To", + "modify", + "account", + "contact", + "broker", + "Your", + "election", + "continue", + "notified", + "revoke", + "consent", + "change", + "print", + "request", + "receipt", + "separate", + "documents", + "household", + "Investor", + "Relations", + "Department", + "website", + "shareholdermeeting", + "phone", + "408", + "486", + "2000", + "2788", + "San", + "Tomas", + "Expressway", + "Santa", + "Clara", + "California", + "95051", + "pay", + "entire", + "cost", + "soliciting", + "proxies", + "Our", + "directors", + "employees", + "solicit", + "person", + "Internet", + "means", + "additional", + "compensation", + "retained", + "MacKenzie", + "Partners", + "advisory", + "basis", + "approximate", + "fee", + "20", + "000", + "bank", + "nominees", + "institutional", + "owners", + "reimburse", + "brokerage", + "firms", + "banks", + "agents", + "forwarding", + "beneficial", + "2027", + "Meeting", + "Deadlines", + "Submission", + "Stockholder", + "Proposals", + "Nomination", + "Directors", + "Other", + "Business", + "Considered", + "Inclusion", + "Proxy", + "Materials", + "Pursuant", + "Rule", + "14a", + "Stockholders", + "present", + "proposals", + "pursuant", + "promulgated", + "Exchange", + "Act", + "inclusion", + "submit", + "writing", + "NVIDIA", + "Corporation", + "Attention", + "Timothy", + "Teter", + "Secretary", + "January" + ], + "keywords": [ + "proxy", + "stockholder", + "voting" + ], + "connect_to": [] + } + }, + { + "chunk_id": "7b9c94b1-a059-5591-aeb9-ce1ffddc129e", + "type": "text", + "content": "A stockholder (or a group of up to 20 stockholders) who has owned at least 3% of the voting power of our outstanding capital stock for at least three continuous years and has complied with the other requirements in our Bylaws may nominate and include in our proxy materials director nominees constituting up to the greater of (a) up to two director candidates or (b) up to 20% of the number of directors in office on the last day that a submission may be delivered. Notice of a proxy access nomination for consideration at our 2027 Meeting must be received following the above instructions not later than the close of business on March 26, 2027, and not earlier than February 24, 2027. If we hold the 2027 Meeting more than 30 days prior to, or delayed by more than 30 days after, the first anniversary of the 2026 Meeting, for written notice by the stockholder to be timely, such notice must be delivered following the above instructions not earlier than the close of business on the 120th day prior to the 2027 Meeting and not later than the close of business on the 90th day prior to the 2027 Meeting or the 10th day following the day on which public announcement of the date of the 2027 Meeting is first made by us, whichever is later.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Director Nominations Under Our Proxy Access Bylaw", + "metadata": { + "length": 1239, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "stockholder", + "group", + "20", + "stockholders", + "owned", + "3%", + "voting", + "power", + "outstanding", + "capital", + "stock", + "continuous", + "years", + "complied", + "requirements", + "Bylaws", + "nominate", + "include", + "proxy", + "materials", + "director", + "nominees", + "constituting", + "greater", + "candidates", + "20%", + "number", + "directors", + "office", + "day", + "submission", + "delivered", + "Notice", + "access", + "nomination", + "consideration", + "2027", + "Meeting", + "received", + "instructions", + "close", + "business", + "March", + "26", + "earlier", + "February", + "24", + "If", + "hold", + "30", + "days", + "prior", + "delayed", + "anniversary", + "2026", + "written", + "notice", + "timely", + "120th", + "90th", + "10th", + "public", + "announcement", + "date", + "made", + "whichever" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4de836b4-a103-5d29-b699-a3cbdfc3c647", + "type": "text", + "content": "Apart from Rule 14a-8 and the proxy access provision of our Bylaws, under our Bylaws certain procedures must be followed for a stockholder to nominate a director or introduce an item of business at an annual meeting of stockholders. If you wish to nominate a director or introduce an item of business at the 2027 Meeting that is not included in our 2027 Meeting proxy materials, you must do so in writing following the above instructions not later than the close of business on March 26, 2027, and not earlier than February 24, 2027. If we hold the 2027 Meeting more than 30 days prior to, or delayed by more than 70 days after, the first anniversary of the 2026 Meeting, for written notice by the stockholder to be timely, such notice must be delivered following the above instructions not earlier than the close of business on the 120th day prior to the 2027 Meeting and not later than the close of business on the 90th day prior to the 2027 Meeting or the 10th day following the day on which public announcement of the date of the 2027 Meeting is first made by us, whichever is later.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Other Director Nominations and Proposals", + "metadata": { + "length": 1087, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Apart", + "Rule", + "14a", + "proxy", + "access", + "provision", + "Bylaws", + "procedures", + "stockholder", + "nominate", + "director", + "introduce", + "item", + "business", + "annual", + "meeting", + "stockholders", + "If", + "2027", + "Meeting", + "included", + "materials", + "writing", + "instructions", + "close", + "March", + "26", + "earlier", + "February", + "24", + "hold", + "30", + "days", + "prior", + "delayed", + "70", + "anniversary", + "2026", + "written", + "notice", + "timely", + "delivered", + "120th", + "day", + "90th", + "10th", + "public", + "announcement", + "date", + "made", + "whichever" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4e9d31ce-e24f-532f-9251-ae0d0cbac472", + "type": "text", + "content": "We advise you to review our Bylaws, which include additional requirements for advance notice of stockholder proposals, director nominations, and proxy access nominations. We recognize the importance of stockholders' ability to nominate directors to our Board. Accordingly, our Board will consider feedback received through our stockholder engagement process regarding the process and disclosure requirements of our Bylaws for nominating directors and other proposals. Our Board will engage with stockholders of various holdings size regarding any proposed amendments to our Bylaws that would require a nominating stockholder to disclose to us (i) such stockholder's plans to nominate candidates to the boards of other public companies, or disclose prior director nominations or proposals that such stockholder privately submitted to other public companies, or (ii) information about such stockholder's limited partners or business associates beyond the existing requirements of our Bylaws.\nWhat am I voting on? Electing the 10 director nominees identified below to hold office until the 2027 Meeting and until his or her successor is elected or appointed.\nVote required for approval: Directors are elected if they receive more FOR votes than AGAINST votes.\nEffect of abstentions: None.\nEffect of broker non-votes: None.\nOur Board currently consists of 10 members. Our nominees include 9 independent directors, as defined by the rules and regulations of Nasdaq, and one NVIDIA officer: Mr. Huang, who serves as our President and CEO.\nAll of our directors have one-year terms and stand for election annually. Each nominee is currently a director of NVIDIA previously elected by our stockholders.\nThe Board expects the nominees will be available for election. If a nominee declines or is unable to act as a director, your proxy may be voted for any substitute nominee proposed by the Board or the size of the Board may be reduced.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Additional Requirements and Information", + "metadata": { + "length": 1927, + "summary": "The text discusses stockholder proposals and director nominations under the company's Bylaws, emphasizing the Board's willingness to engage with stockholders on amendments. The main voting item is the election of 10 director nominees, including 9 independent directors and CEO Jensen Huang, to serve until the 2027 annual meeting. Directors are elected by a majority of votes cast, with abstentions and broker non-votes having no effect. All nominees currently serve on the Board.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "advise", + "review", + "Bylaws", + "include", + "additional", + "requirements", + "advance", + "notice", + "stockholder", + "proposals", + "director", + "nominations", + "proxy", + "access", + "recognize", + "importance", + "stockholders", + "ability", + "nominate", + "directors", + "Board", + "Accordingly", + "feedback", + "received", + "engagement", + "process", + "disclosure", + "nominating", + "Our", + "engage", + "holdings", + "size", + "proposed", + "amendments", + "require", + "disclose", + "plans", + "candidates", + "boards", + "public", + "companies", + "prior", + "privately", + "submitted", + "ii", + "information", + "limited", + "partners", + "business", + "associates", + "existing", + "What", + "voting", + "Electing", + "10", + "nominees", + "identified", + "hold", + "office", + "2027", + "Meeting", + "successor", + "elected", + "appointed", + "Vote", + "required", + "approval", + "Directors", + "receive", + "FOR", + "votes", + "AGAINST", + "Effect", + "abstentions", + "None", + "broker", + "consists", + "members", + "independent", + "defined", + "rules", + "regulations", + "Nasdaq", + "NVIDIA", + "officer", + "Mr", + "Huang", + "serves", + "President", + "CEO", + "All", + "year", + "terms", + "stand", + "election", + "annually", + "Each", + "nominee", + "previously", + "The", + "expects", + "If", + "declines", + "unable", + "act", + "voted", + "substitute", + "reduced" + ], + "keywords": [ + "director election", + "bylaws", + "stockholder proposals" + ], + "connect_to": [] + } + }, + { + "chunk_id": "ce90156d-3fcf-5cb1-9416-38060a7b2814", + "type": "table", + "content": "
NameAgeDirector SinceOccupationIndependentFinancial Expert (1)Committee MembershipNumber of Other Public Company Boards
Tench Coxe681993Former Managing Director, Sutter Hill VenturesAC, CC
John O. Dabiri462020Centennial Professor of Aeronautics and Mechanical Engineering, California Institute of TechnologyCC, NCGC
Jen-Hsun Huang631993President & CEO, NVIDIA Corporation
Dawn Hudson682013Former Chief Marketing Officer, National Football League & Former CEO Pepsi-Cola North AmericaCC Chairperson
Harvey C. Jones731993Managing Partner, Square Wave VenturesAC, NCGC
Melissa B. Lora632023Former President, Taco Bell InternationalAC, NCGC1
Stephen C. Neal (Lead Director)772019Chairman Emeritus & Senior Counsel, Cooley LLPNCGC Chairperson
A. Brooke Seawell781997Venture Partner, New Enterprise AssociatesAC Chairperson1
Aarti Shah612020Former Senior Vice President & Chief Information and Digital Officer, Eli Lilly and CompanyAC, CC1
Mark A. Stevens662008 (2)Managing Partner, S-Cubed CapitalCC, NCGC
", + "path": "tables/table-13 NVIDIA Board.html", + "metadata": { + "length": 1925, + "summary": "table-14\nThe table lists NVIDIA's board members, including Tench Coxe, Jen-Hsun Huang, and others, with details on age, tenure, independence, financial expertise, and committee roles.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-13 NVIDIA Board.html", + "keywords": [ + "Board of Directors", + "NVIDIA", + "Corporate Governance" + ], + "tokens": [] + } + }, + { + "chunk_id": "77e0197b-7168-5238-9bff-7ff26e2706eb", + "type": "text", + "content": "The Board recommends that you vote FOR the election of each of the following nominees:\n\n[tables/table-13 NVIDIA Board.html]\n\n(1) Qualified as an AC financial expert.\n(2) Previously served as a member of our Board from 1993 until 2006.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Recommendation of the Board", + "metadata": { + "length": 302, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Board", + "recommends", + "vote", + "FOR", + "election", + "nominees", + "Qualified", + "AC", + "financial", + "expert", + "Previously", + "served", + "member", + "1993", + "2006" + ], + "keywords": [], + "connect_to": [ + { + "target": "ce90156d-3fcf-5cb1-9416-38060a7b2814", + "relation": "embeds", + "ref": "[tables/table-13 NVIDIA Board.html]", + "position": { + "start": 88, + "end": 123 + } + } + ] + } + }, + { + "chunk_id": "5808bb05-858d-5dda-a8a1-12b65dabdc49", + "type": "table", + "content": "
Senior Leadership & Operations ExperienceDirectors with senior leadership and operations experience provide informed oversight of our business, and unique experiences and perspectives. They are uniquely positioned to contribute practical insight into business strategy and operations, driving growth, building and strengthening corporate culture, and supporting the achievement of strategic priorities and objectives.
Industry & TechnicalDirectors with industry experience and technical backgrounds facilitate within the Board a deeper understanding of innovations and a technical assessment of our products and services.
Financial/Financial CommunityExperience in financial matters and the financial community assists our Board with review of our operations and finances, including overseeing our financial statements, capital structure and internal controls. Those with a venture capital background also offer valuable stockholder perspectives.
Governance & Public Company BoardDirectors with experience in corporate governance, such as service on boards and board committees, or as executives of other large, public companies, are familiar with the dynamics and operation of a board of directors and the impact that governance policies have on a company. This experience supports our goals of strong Board and management accountability, transparency, and protection of stockholder interests. Public company board experience also helps our directors identify challenges and risks we face as a public company, including oversight of strategic, operational, and compliance-related matters, and stockholder relations.
Emerging Technologies & Business ModelsExperience in emerging technologies and business models is integral to our growth strategies given our unique business model and provides important insights as our business expands into new areas.
Marketing, Communications & Brand ManagementDirectors with experience in marketing, communications, and brand management offer guidance on our products directly marketed to consumers, important perspectives on expanding our market share, and expertise in communicating with our customers and other stakeholders.
Regulatory, Legal & Risk ManagementOur business requires compliance with a variety of regulatory requirements in different jurisdictions. We face new regulatory matters and regulations as our business grows. We are also subject to multiple lawsuits. Directors with experience in governmental, public policy, legal, and risk management areas, including cybersecurity, help provide valuable insights and oversight for our Company.
Human Capital Management ExperienceOur people are critical to our success. Directors with experience in organizational management, talent development, and developing values and culture in a large global workforce provide key insights. Human capital management experience also assists our Board in overseeing executive and employee compensation, development, and engagement.
", + "path": "tables/table-14 Board Skills Matrix.html", + "metadata": { + "length": 3991, + "summary": "table-15\nA table listing director qualifications: senior leadership, industry, finance, governance, emerging tech, marketing, legal/risk, and human capital management experience.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-14 Board Skills Matrix.html", + "keywords": [ + "Board Experience", + "Leadership", + "Governance" + ], + "tokens": [] + } + }, + { + "chunk_id": "134f1d43-659f-57e1-8323-07504fb5bab1", + "type": "text", + "content": "The NCGC identifies, reviews and assesses the qualifications of current and potential directors and recommends nominees for Board approval. In accordance with our Corporate Governance Policies and the NCGC Charter, the NCGC considers a nominee's background and experience to ensure that a broad range of perspectives is represented. The NCGC may conduct appropriate and necessary inquiries into the backgrounds and qualifications of possible candidates and engage a professional search firm to assist the committee in identifying, evaluating, and conducting due diligence on potential director nominees. The NCGC has not established specific age, gender, education, experience, or skill requirements, and instead considers numerous factors regarding the nominee, including those listed below, taking into account our current and future business models:\n- Integrity and candor\n- Independence\n• Senior leadership and operational experience\n• Professional, technical and industry knowledge\n• Financial expertise\n• Financial community experience (including as an investor in other companies)\n- Marketing, communications and brand management background\n• Governance and public company board experience\n- Experience with emerging technologies and new business models\n• Regulatory, legal, and risk management expertise, including in cybersecurity matters\n- Mix of backgrounds, experiences, and perspectives\n• Human capital management experience\n• Experience in academia\n- Willingness and ability to devote substantial time and effort to Board responsibilities and Company oversight\n- Ability to represent the interests of the stockholders as a whole rather than special interest groups or constituencies\n- All relationships between the proposed nominee and any of our stockholders, competitors, customers, suppliers, or other persons with a relationship to NVIDIA\n- For nominees for re-election, overall service to NVIDIA, including past attendance, participation and contributions to the activities of the Board and its committees\nThe NCGC and the Board understand the importance of Board refreshment, and strive to maintain an appropriate mix of tenure, professional experiences and backgrounds, skills, and education on the Board. The Board seeks to balance the benefits of continuity and institutional knowledge of our longer-serving directors, with the introduction of new perspectives and ideas through the appointment of new directors. In furtherance of this approach, one-third of our independent directors have joined the Board since 2020. The Board, through the NCGC, actively oversees Board composition to maintain an appropriate mix of skills, experience and perspectives, aligned with the Company's strategy and evolving business needs. Our longer-tenured directors are familiar with our operations and business areas and have the perspective of overseeing our activities from a variety of economic and competitive environments, which enhances the Board's oversight of strategy and risks. Given our growth and the breadth of our product offerings, as well as the increasingly complex macroeconomic and geopolitical factors we face, these experienced directors are a significant asset to the Board. Our newer directors have brought expertise in brand development and cybersecurity, familiarity with technology developments at leading academic institutions, and senior management, operating, and finance experience, all of which are important to supporting NVIDIA as it enters new markets. Each year, the NCGC and Board review each director's individual performance, including the director's past contributions, outside experiences and activities, and committee participation, and determine how his or her experience and skills continue to add value to NVIDIA and the Board. As a part of this process, the Board also periodically refreshes committee membership and chairpersons to help promote a variety of viewpoints and broaden director engagement across key areas of oversight on the Board committees. Since the 2025 Meeting, Ms. Lora and Dr. Dabiri have joined the NCGC.\nBelow are the skills, competencies, and attributes that our Board considers important for our directors to have, considering our current business and future market opportunities:\n\n[tables/table-14 Board Skills Matrix.html]\n\nThe NCGC evaluates candidates proposed by stockholders using the same criteria it uses for other candidates. Stockholders seeking to recommend a prospective nominee should follow the instructions under Stockholder Communications with the Board of Directors below. The stockholder(s) must provide timely written notice of such recommended nominee and the stockholder(s) and such recommended nominee must satisfy the other requirements specified in our Bylaws and Corporate Governance Policies, both of which are available on the Investor Relations section of our website at www.nvidia.com. Stockholders are advised to review these documents.\nProxy Access\nOur Board voluntarily adopted proxy access. We will include in our proxy statement information regarding the greater of (i) up to two director candidates or (ii) up to 20% of the number of directors in office on the last day that a submission may be delivered, if nominated by a stockholder (or group of up to 20 stockholders) owning at least 3% of the voting power of our outstanding capital stock for at least three continuous years. The stockholder(s) must provide timely written notice of such nomination and the stockholder(s) and nominee must satisfy the other requirements specified in our Bylaws. This summary of our proxy access rules is not intended to be complete and is subject to limitations set forth in our Bylaws and Corporate Governance Policies, both of which are available on the Investor Relations section of our website at www.nvidia.com. Stockholders are advised to review these documents. The NCGC did not receive any stockholder nominations during Fiscal 2026.", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Director Qualifications and Nomination of Directors", + "metadata": { + "length": 5969, + "summary": "The NCGC identifies and assesses director qualifications, considering factors like integrity, independence, leadership, industry knowledge, and diversity. It evaluates stockholder nominees equally and has proxy access rules. The Board balances tenure with new perspectives, with one-third of independent directors joining since 2020. Recent additions include Ms. Lora and Dr. Dabiri to the NCGC.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "NCGC", + "identifies", + "reviews", + "assesses", + "qualifications", + "current", + "potential", + "directors", + "recommends", + "nominees", + "Board", + "approval", + "In", + "accordance", + "Corporate", + "Governance", + "Policies", + "Charter", + "considers", + "nominee", + "background", + "experience", + "ensure", + "broad", + "range", + "perspectives", + "represented", + "conduct", + "inquiries", + "backgrounds", + "candidates", + "engage", + "professional", + "search", + "firm", + "assist", + "committee", + "identifying", + "evaluating", + "conducting", + "due", + "diligence", + "director", + "established", + "specific", + "age", + "gender", + "education", + "skill", + "requirements", + "numerous", + "factors", + "including", + "listed", + "taking", + "account", + "future", + "business", + "models", + "Integrity", + "candor", + "Independence", + "Senior", + "leadership", + "operational", + "Professional", + "technical", + "industry", + "knowledge", + "Financial", + "expertise", + "community", + "investor", + "companies", + "Marketing", + "communications", + "brand", + "management", + "public", + "company", + "board", + "Experience", + "emerging", + "technologies", + "Regulatory", + "legal", + "risk", + "cybersecurity", + "matters", + "Mix", + "experiences", + "Human", + "capital", + "academia", + "Willingness", + "ability", + "devote", + "substantial", + "time", + "effort", + "responsibilities", + "Company", + "oversight", + "Ability", + "represent", + "interests", + "stockholders", + "special", + "interest", + "groups", + "constituencies", + "All", + "relationships", + "proposed", + "competitors", + "customers", + "suppliers", + "persons", + "relationship", + "NVIDIA", + "For", + "election", + "service", + "past", + "attendance", + "participation", + "contributions", + "activities", + "committees", + "understand", + "importance", + "refreshment", + "strive", + "maintain", + "mix", + "tenure", + "skills", + "seeks", + "balance", + "benefits", + "continuity", + "institutional", + "longer", + "serving", + "introduction", + "ideas", + "appointment", + "furtherance", + "approach", + "independent", + "joined", + "2020", + "actively", + "oversees", + "composition", + "aligned", + "strategy", + "evolving", + "Our", + "tenured", + "familiar", + "operations", + "areas", + "perspective", + "overseeing", + "variety", + "economic", + "competitive", + "environments", + "enhances", + "risks", + "Given", + "growth", + "breadth", + "product", + "offerings", + "increasingly", + "complex", + "macroeconomic", + "geopolitical", + "face", + "experienced", + "significant", + "asset", + "newer", + "brought", + "development", + "familiarity", + "technology", + "developments", + "leading", + "academic", + "institutions", + "senior", + "operating", + "finance", + "important", + "supporting", + "enters", + "markets", + "Each", + "year", + "review", + "individual", + "performance", + "determine", + "continue", + "add", + "As", + "part", + "process", + "periodically", + "refreshes", + "membership", + "chairpersons", + "promote", + "viewpoints", + "broaden", + "engagement", + "key", + "Since", + "2025", + "Meeting", + "Ms", + "Lora", + "Dr", + "Dabiri", + "Below", + "competencies", + "attributes", + "market", + "opportunities", + "evaluates", + "criteria", + "Stockholders", + "seeking", + "recommend", + "prospective", + "follow", + "instructions", + "Stockholder", + "Communications", + "Directors", + "stockholder", + "provide", + "timely", + "written", + "notice", + "recommended", + "satisfy", + "Bylaws", + "Investor", + "Relations", + "section", + "website", + "www", + "nvidia", + "advised", + "documents", + "Proxy", + "Access", + "voluntarily", + "adopted", + "proxy", + "access", + "We", + "include", + "statement", + "information", + "greater", + "ii", + "20%", + "number", + "office", + "day", + "submission", + "delivered", + "nominated", + "group", + "20", + "owning", + "3%", + "voting", + "power", + "outstanding", + "stock", + "continuous", + "years", + "nomination", + "This", + "summary", + "rules", + "intended", + "complete", + "subject", + "limitations", + "set", + "receive", + "nominations", + "Fiscal", + "2026" + ], + "keywords": [ + "NCGC", + "Board Composition", + "Director Qualifications" + ], + "connect_to": [ + { + "target": "5808bb05-858d-5dda-a8a1-12b65dabdc49", + "relation": "embeds", + "ref": "[tables/table-14 Board Skills Matrix.html]", + "position": { + "start": 4259, + "end": 4301 + } + } + ] + } + }, + { + "chunk_id": "13908866-486b-5cd4-a3e8-5a75b86c2e85", + "type": "table", + "content": "
Age: 68Director Since: 1993Committees: AC, CC
Other Current Public Company Boards: NoneIndependent Director
", + "path": "tables/table-15 Director Profile.html", + "metadata": { + "length": 193, + "summary": "table-16\nA 68-year-old independent director serving since 1993 on Audit and Compensation Committees, with no other current public company board memberships.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-15 Director Profile.html", + "keywords": [ + "Director", + "Independent", + "Committees" + ], + "tokens": [] + } + }, + { + "chunk_id": "82b279d6-4b50-5ba3-9820-7f2c446d1088", + "type": "image", + "content": "\nTENCH COXE Financial/Financial0\n[images/image-32-TENCH COXE.jpg]\n", + "path": "images/image-32-TENCH COXE.jpg", + "metadata": { + "length": 66, + "summary": "image-32\nTENCH COXE Financial/Financial0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-32-TENCH COXE.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "070a6edf-52e7-526c-9084-2d01667c92ac", + "type": "image", + "content": "\nTench Coxe was a Managing Director of Sutter Hill Ventures, a venture capital investment firm, from 1989 to 2020, where he focused on investments in the IT sector. Prior to joining Sutter Hill Ventures in 1987, he was Director of Marketing and MIS at Digital Communication Associates. He was a director of Artisan Partners Asset Management Inc., an institutional money management firm from 1995 to 2025. Mr. Coxe holds a BA degree in Economics from Dartmouth College and an MBA degree from Harvard Business School. Governance & Public0\n[images/image-33-Tench Coxe.jpg]\n", + "path": "images/image-33-Tench Coxe.jpg", + "metadata": { + "length": 570, + "summary": "image-33\nTench Coxe was a Managing Director of Sutter Hill Ventures, a venture capital investment firm, from 1989 to 2020, where he focused on investments in the IT sector. Prior to joining Sutter Hill Ventures in 1987, he was Director of Marketing and MIS at Digital Communication Associates. He was a director of Artisan Partners Asset Management Inc., an institutional money management firm from 1995 to 2025. Mr. Coxe holds a BA degree in Economics from Dartmouth College and an MBA degree from Harvard Business School. Governance & Public0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-33-Tench Coxe.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "059b9879-b611-5f64-bb00-2cfac5c06aeb", + "type": "image", + "content": "\nFinancial/Financial Emerging Technologies & Business Models0\n[images/image-34-Financial_.jpg]\n", + "path": "images/image-34-Financial_.jpg", + "metadata": { + "length": 95, + "summary": "image-34\nFinancial/Financial Emerging Technologies & Business Models0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-34-Financial_.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "4236e4c9-8b46-5c5b-a545-566569c57184", + "type": "image", + "content": "\nGovernance & Public Human Capital0\n[images/image-35-Governance.jpg]\n", + "path": "images/image-35-Governance.jpg", + "metadata": { + "length": 69, + "summary": "image-35\nGovernance & Public Human Capital0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-35-Governance.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e5076757-ba0f-597a-bfae-552f1845ffd9", + "type": "table", + "content": "
Age: 46Director Since: 2020Committees: CC, NCGC
Other Current Public Company Boards: NoneIndependent Director
", + "path": "tables/table-16 Director Profile.html", + "metadata": { + "length": 195, + "summary": "table-17\nA 46-year-old independent director since 2020, serving on CC and NCGC committees, with no other public company boards.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-16 Director Profile.html", + "keywords": [ + "Director", + "Committees", + "Independent" + ], + "tokens": [] + } + }, + { + "chunk_id": "a08b4b94-3bcf-58fe-8efa-1cc2333e2ca8", + "type": "image", + "content": "\nJOHN O. DABIRI Industry & Technical0\n[images/image-36-JOHN O. DA.jpg]\n", + "path": "images/image-36-JOHN O. DA.jpg", + "metadata": { + "length": 71, + "summary": "image-36\nJOHN O. DABIRI Industry & Technical0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-36-JOHN O. DA.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "3ecee2e4-7315-5c60-88ca-d096cf01304b", + "type": "image", + "content": "\nJohn O. Dabiri is the Centennial Professor of Aeronautics and Mechanical Engineering at the California Institute of Technology. He is a recipient of the National Medal of Science, a MacArthur Foundation \"Genius Grant,\" the National Science Foundation Alan T. Waterman Award, and the Presidential Early Career Award for Scientists and Engineers. He heads the Dabiri Lab, which conducts research at the intersections of fluid mechanics, energy and environment, and biology. From 2021 to January 2025, he served on President Biden's Council of Advisors on Science and Technology (PCAST) and Energy Secretary Granholm's Energy Advisory Board (SEAB). From 2015 to 2019, he served as a Professor of Civil and Environmental Engineering and of Mechanical Engineering at Stanford University, where he was recognized with the Eugene L. Grant Award for Excellence in Teaching. From 2005 to 2015, he was a Professor of Aeronautics and Bioengineering at the California Institute of Technology, during which time he also served as Director of the Center for Bioinspired Wind Energy, Chair of the Faculty, and Dean of Students. Dr. Dabiri is a Fellow of the American Physical Society, where he previously served as Chair of the Division of Fluid Dynamics. He also serves on the board of trustees of the Gordon and Betty Moore Foundation and previously served as a member of the National Academies' Committee on Science, Technology, and Law. Dr. Dabiri holds a PhD degree in Bioengineering and an MS degree in Aeronautics from the California Institute of Technology, and a BSE degree summa cum laude in Mechanical and Aerospace Engineering from Princeton University. Emerging Technologies & Business Models0\n[images/image-37-John O. Da.jpg]\n", + "path": "images/image-37-John O. Da.jpg", + "metadata": { + "length": 1726, + "summary": "image-37\nJohn O. Dabiri is the Centennial Professor of Aeronautics and Mechanical Engineering at the California Institute of Technology. He is a recipient of the National Medal of Science, a MacArthur Foundation \"Genius Grant,\" the National Science Foundation Alan T. Waterman Award, and the Presidential Early Career Award for Scientists and Engineers. He heads the Dabiri Lab, which conducts research at the intersections of fluid mechanics, energy and environment, and biology. From 2021 to January 2025, he served on President Biden's Council of Advisors on Science and Technology (PCAST) and Energy Secretary Granholm's Energy Advisory Board (SEAB). From 2015 to 2019, he served as a Professor of Civil and Environmental Engineering and of Mechanical Engineering at Stanford University, where he was recognized with the Eugene L. Grant Award for Excellence in Teaching. From 2005 to 2015, he was a Professor of Aeronautics and Bioengineering at the California Institute of Technology, during which time he also served as Director of the Center for Bioinspired Wind Energy, Chair of the Faculty, and Dean of Students. Dr. Dabiri is a Fellow of the American Physical Society, where he previously served as Chair of the Division of Fluid Dynamics. He also serves on the board of trustees of the Gordon and Betty Moore Foundation and previously served as a member of the National Academies' Committee on Science, Technology, and Law. Dr. Dabiri holds a PhD degree in Bioengineering and an MS degree in Aeronautics from the California Institute of Technology, and a BSE degree summa cum laude in Mechanical and Aerospace Engineering from Princeton University. Emerging Technologies & Business Models0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-37-John O. Da.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f292ea45-410a-5948-b280-de2e2fab759c", + "type": "text", + "content": "The biographies below include information, as of the date of this proxy statement, regarding the particular experience, qualifications, attributes or skills of each director, relative to the skills matrix above, that led the NCGC and Board to believe that he or she should continue to serve on the Board.\nTENCH COXE\nFormer Managing Director, Sutter Hill Ventures\n\n[tables/table-15 Director Profile.html]\n\nTench Coxe was a Managing Director of Sutter Hill Ventures, a venture capital investment firm, from 1989 to 2020, where he focused on investments in the IT sector. Prior to joining Sutter Hill Ventures in 1987, he was Director of Marketing and MIS at Digital Communication Associates. He was a director of Artisan Partners Asset Management Inc., an institutional money management firm from 1995 to 2025. Mr. Coxe holds a BA degree in Economics from Dartmouth College and an MBA degree from Harvard Business School.\nMr. Coxe brings to the Board expertise in financial and transactional analysis and provides valuable perspectives on corporate strategy and emerging technology trends. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders.\n\nTENCH COXE Financial/Financial0\n[images/image-32-TENCH COXE.jpg]\n\nFinancial/Financial\nCommunity\n\nTench Coxe was a Managing Director of Sutter Hill Ventures, a venture capital investment firm, from 1989 to 2020, where he focused on investments in the IT sector. Prior to joining Sutter Hill Ventures in 1987, he was Director of Marketing and MIS at Digital Communication Associates. He was a director of Artisan Partners Asset Management Inc., an institutional money management firm from 1995 to 2025. Mr. Coxe holds a BA degree in Economics from Dartmouth College and an MBA degree from Harvard Business School. Governance & Public0\n[images/image-33-Tench Coxe.jpg]\n\nGovernance & Public\nCompany Board\n\nFinancial/Financial Emerging Technologies & Business Models0\n[images/image-34-Financial_.jpg]\n\nEmerging Technologies & Business Models\n\nGovernance & Public Human Capital0\n[images/image-35-Governance.jpg]\n\nHuman Capital\nManagement\nExperience\nJOHN O. DABIRI\nCentennial Professor of Aeronautics and Mechanical Engineering, California Institute of Technology\n\n[tables/table-16 Director Profile.html]\n\nJohn O. Dabiri is the Centennial Professor of Aeronautics and Mechanical Engineering at the California Institute of Technology. He is a recipient of the National Medal of Science, a MacArthur Foundation \"Genius Grant,\" the National Science Foundation Alan T. Waterman Award, and the Presidential Early Career Award for Scientists and Engineers. He heads the Dabiri Lab, which conducts research at the intersections of fluid mechanics, energy and environment, and biology. From 2021 to January 2025, he served on President Biden's Council of Advisors on Science and Technology (PCAST) and Energy Secretary Granholm's Energy Advisory Board (SEAB). From 2015 to 2019, he served as a Professor of Civil and Environmental Engineering and of Mechanical Engineering at Stanford University, where he was recognized with the Eugene L. Grant Award for Excellence in Teaching. From 2005 to 2015, he was a Professor of Aeronautics and Bioengineering at the California Institute of Technology, during which time he also served as Director of the Center for Bioinspired Wind Energy, Chair of the Faculty, and Dean of Students. Dr. Dabiri is a Fellow of the American Physical Society, where he previously served as Chair of the Division of Fluid Dynamics. He also serves on the board of trustees of the Gordon and Betty Moore Foundation and previously served as a member of the National Academies' Committee on Science, Technology, and Law. Dr. Dabiri holds a PhD degree in Bioengineering and an MS degree in Aeronautics from the California Institute of Technology, and a BSE degree summa cum laude in Mechanical and Aerospace Engineering from Princeton University.\nDr. Dabiri brings to the Board a versatile research background and cutting-edge expertise in various engineering fields, along with a proven record of successful innovation.\n\nJOHN O. DABIRI Industry & Technical0\n[images/image-36-JOHN O. DA.jpg]\n\nIndustry & Technical\n\nJohn O. Dabiri is the Centennial Professor of Aeronautics and Mechanical Engineering at the California Institute of Technology. He is a recipient of the National Medal of Science, a MacArthur Foundation \"Genius Grant,\" the National Science Foundation Alan T. Waterman Award, and the Presidential Early Career Award for Scientists and Engineers. He heads the Dabiri Lab, which conducts research at the intersections of fluid mechanics, energy and environment, and biology. From 2021 to January 2025, he served on President Biden's Council of Advisors on Science and Technology (PCAST) and Energy Secretary Granholm's Energy Advisory Board (SEAB). From 2015 to 2019, he served as a Professor of Civil and Environmental Engineering and of Mechanical Engineering at Stanford University, where he was recognized with the Eugene L. Grant Award for Excellence in Teaching. From 2005 to 2015, he was a Professor of Aeronautics and Bioengineering at the California Institute of Technology, during which time he also served as Director of the Center for Bioinspired Wind Energy, Chair of the Faculty, and Dean of Students. Dr. Dabiri is a Fellow of the American Physical Society, where he previously served as Chair of the Division of Fluid Dynamics. He also serves on the board of trustees of the Gordon and Betty Moore Foundation and previously served as a member of the National Academies' Committee on Science, Technology, and Law. Dr. Dabiri holds a PhD degree in Bioengineering and an MS degree in Aeronautics from the California Institute of Technology, and a BSE degree summa cum laude in Mechanical and Aerospace Engineering from Princeton University. Emerging Technologies & Business Models0\n[images/image-37-John O. Da.jpg]\n\nEmerging Technologies & Business Models", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees", + "metadata": { + "length": 6093, + "summary": "The text provides biographies of two directors, Tench Coxe and John O. Dabiri, detailing their professional experience, qualifications, and skills that led the NCGC and Board to believe they should continue serving. Tench Coxe, former Managing Director of Sutter Hill Ventures, brings expertise in financial analysis, corporate strategy, and emerging technology trends. John O. Dabiri, a professor at Caltech, contributes a versatile research background in engineering and innovation. The biographies include information on their ages, tenure, committee memberships, education, and career highlights, along with associated skills matrices.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "biographies", + "include", + "information", + "date", + "proxy", + "statement", + "experience", + "qualifications", + "attributes", + "skills", + "director", + "relative", + "matrix", + "led", + "NCGC", + "Board", + "continue", + "serve", + "TENCH", + "COXE", + "Former", + "Managing", + "Director", + "Sutter", + "Hill", + "Ventures", + "Tench", + "Coxe", + "venture", + "capital", + "investment", + "firm", + "1989", + "2020", + "focused", + "investments", + "IT", + "sector", + "Prior", + "joining", + "1987", + "Marketing", + "MIS", + "Digital", + "Communication", + "Associates", + "He", + "Artisan", + "Partners", + "Asset", + "Management", + "Inc", + "institutional", + "money", + "management", + "1995", + "2025", + "Mr", + "holds", + "BA", + "degree", + "Economics", + "Dartmouth", + "College", + "MBA", + "Harvard", + "Business", + "School", + "brings", + "expertise", + "financial", + "transactional", + "analysis", + "valuable", + "perspectives", + "corporate", + "strategy", + "emerging", + "technology", + "trends", + "His", + "significant", + "community", + "understanding", + "methods", + "companies", + "increase", + "stockholders", + "Financial", + "Financial0", + "Community", + "Governance", + "Public0", + "Public", + "Company", + "Emerging", + "Technologies", + "Models0", + "Models", + "Human", + "Capital0", + "Capital", + "Experience", + "JOHN", + "DABIRI", + "Centennial", + "Professor", + "Aeronautics", + "Mechanical", + "Engineering", + "California", + "Institute", + "Technology", + "John", + "Dabiri", + "recipient", + "National", + "Medal", + "Science", + "MacArthur", + "Foundation", + "Genius", + "Grant", + "Alan", + "Waterman", + "Award", + "Presidential", + "Early", + "Career", + "Scientists", + "Engineers", + "heads", + "Lab", + "conducts", + "research", + "intersections", + "fluid", + "mechanics", + "energy", + "environment", + "biology", + "From", + "2021", + "January", + "served", + "President", + "Biden", + "Council", + "Advisors", + "PCAST", + "Energy", + "Secretary", + "Granholm", + "Advisory", + "SEAB", + "2015", + "2019", + "Civil", + "Environmental", + "Stanford", + "University", + "recognized", + "Eugene", + "Excellence", + "Teaching", + "2005", + "Bioengineering", + "time", + "Center", + "Bioinspired", + "Wind", + "Chair", + "Faculty", + "Dean", + "Students", + "Dr", + "Fellow", + "American", + "Physical", + "Society", + "previously", + "Division", + "Fluid", + "Dynamics", + "serves", + "board", + "trustees", + "Gordon", + "Betty", + "Moore", + "member", + "Academies", + "Committee", + "Law", + "PhD", + "MS", + "BSE", + "summa", + "cum", + "laude", + "Aerospace", + "Princeton", + "versatile", + "background", + "cutting", + "edge", + "engineering", + "fields", + "proven", + "record", + "successful", + "innovation", + "Industry", + "Technical0", + "Technical" + ], + "keywords": [ + "Directors", + "Experience", + "Qualifications" + ], + "connect_to": [ + { + "target": "13908866-486b-5cd4-a3e8-5a75b86c2e85", + "relation": "embeds", + "ref": "[tables/table-15 Director Profile.html]", + "position": { + "start": 364, + "end": 403 + } + }, + { + "target": "82b279d6-4b50-5ba3-9820-7f2c446d1088", + "relation": "embeds", + "ref": "[images/image-32-TENCH COXE.jpg]", + "position": { + "start": 1278, + "end": 1310 + } + }, + { + "target": "070a6edf-52e7-526c-9084-2d01667c92ac", + "relation": "embeds", + "ref": "[images/image-33-Tench Coxe.jpg]", + "position": { + "start": 1879, + "end": 1911 + } + }, + { + "target": "059b9879-b611-5f64-bb00-2cfac5c06aeb", + "relation": "embeds", + "ref": "[images/image-34-Financial_.jpg]", + "position": { + "start": 2009, + "end": 2041 + } + }, + { + "target": "4236e4c9-8b46-5c5b-a545-566569c57184", + "relation": "embeds", + "ref": "[images/image-35-Governance.jpg]", + "position": { + "start": 2119, + "end": 2151 + } + }, + { + "target": "e5076757-ba0f-597a-bfae-552f1845ffd9", + "relation": "embeds", + "ref": "[tables/table-16 Director Profile.html]", + "position": { + "start": 2304, + "end": 2343 + } + }, + { + "target": "a08b4b94-3bcf-58fe-8efa-1cc2333e2ca8", + "relation": "embeds", + "ref": "[images/image-36-JOHN O. DA.jpg]", + "position": { + "start": 4208, + "end": 4240 + } + }, + { + "target": "3ecee2e4-7315-5c60-88ca-d096cf01304b", + "relation": "embeds", + "ref": "[images/image-37-John O. Da.jpg]", + "position": { + "start": 5956, + "end": 5988 + } + } + ] + } + }, + { + "chunk_id": "85e03b98-d4de-58c2-85e0-d166f977fcb8", + "type": "image", + "content": "\nJen-Hsun Huang founded NVIDIA in 1993 and has served since its inception as President, Chief Executive Officer, and a member of the Board. Senior Leadership & Operations Experience0\n[images/image-38-Jen-Hsun H.jpg]\n", + "path": "images/image-38-Jen-Hsun H.jpg", + "metadata": { + "length": 216, + "summary": "image-38\nJen-Hsun Huang founded NVIDIA in 1993 and has served since its inception as President, Chief Executive Officer, and a member of the Board. Senior Leadership & Operations Experience0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-38-Jen-Hsun H.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ff8e4a93-0064-537b-b716-6a7654788817", + "type": "image", + "content": "\nMr. Huang has been elected to the National Academy of Engineering and in 2026 was appointed to the President's Council of Advisors of Science and Technology. He is a recipient of the Semiconductor Industry Association's highest honor, the Robert N. Noyce Award; the IEEE Founder's Medal; the Dr. Morris Chang Exemplary Leadership Award; and honorary doctorate degrees from Taiwan's National Chiao Tung University, National Taiwan University, Oregon State University, Huazhong University of Science and Technology and Linköping University. He has been named the world's best CEO by Fortune, the Economist, and Brand Finance, as well as one of TIME magazine's 100 most influential people. Industry & Technical0\n[images/image-39-Mr. Huang .jpg]\n", + "path": "images/image-39-Mr. Huang .jpg", + "metadata": { + "length": 743, + "summary": "image-39\nMr. Huang has been elected to the National Academy of Engineering and in 2026 was appointed to the President's Council of Advisors of Science and Technology. He is a recipient of the Semiconductor Industry Association's highest honor, the Robert N. Noyce Award; the IEEE Founder's Medal; the Dr. Morris Chang Exemplary Leadership Award; and honorary doctorate degrees from Taiwan's National Chiao Tung University, National Taiwan University, Oregon State University, Huazhong University of Science and Technology and Linköping University. He has been named the world's best CEO by Fortune, the Economist, and Brand Finance, as well as one of TIME magazine's 100 most influential people. Industry & Technical0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-39-Mr. Huang .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e47e2e5f-8481-503e-84fb-67017af8874a", + "type": "image", + "content": "\nMr. Huang is one of the technology industry's most respected executives, having taken NVIDIA from a startup to a world leader in accelerated computing. Under his guidance, NVIDIA has compiled a record of consistent innovation and sharp execution, marked by products that have gained strong market share. Financial/Financial0\n[images/image-40-Mr. Huang .jpg]\n", + "path": "images/image-40-Mr. Huang .jpg", + "metadata": { + "length": 359, + "summary": "image-40\nMr. Huang is one of the technology industry's most respected executives, having taken NVIDIA from a startup to a world leader in accelerated computing. Under his guidance, NVIDIA has compiled a record of consistent innovation and sharp execution, marked by products that have gained strong market share. Financial/Financial0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-40-Mr. Huang .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "889b1346-36ce-54bf-a1e2-d92cb67e9e5f", + "type": "image", + "content": "\nIndustry & Technical Governance & Public Company Board0\n[images/image-41-Industry &.jpg]\n", + "path": "images/image-41-Industry &.jpg", + "metadata": { + "length": 90, + "summary": "image-41\nIndustry & Technical Governance & Public Company Board0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-41-Industry &.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "b8c14ae1-3dce-5d49-9adb-5c433cbf2a3a", + "type": "image", + "content": "\nFinancial/Financial Emerging Technologies & Business Models0\n[images/image-42-Financial_.jpg]\n", + "path": "images/image-42-Financial_.jpg", + "metadata": { + "length": 95, + "summary": "image-42\nFinancial/Financial Emerging Technologies & Business Models0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-42-Financial_.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "918a8757-d542-5d01-a385-9d77625d5e1e", + "type": "image", + "content": "\nCommunity Marketing,0\n[images/image-43-Community .jpg]\n", + "path": "images/image-43-Community .jpg", + "metadata": { + "length": 56, + "summary": "image-43\nCommunity Marketing,0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-43-Community .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "997509c7-bd1b-59bb-89b5-ab12564fd687", + "type": "image", + "content": "\nEmerging Technologies & Business Models Regulatory, Legal & Risk Management0\n[images/image-44-Emerging T.jpg]\n", + "path": "images/image-44-Emerging T.jpg", + "metadata": { + "length": 111, + "summary": "image-44\nEmerging Technologies & Business Models Regulatory, Legal & Risk Management0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-44-Emerging T.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "cd063f0c-f174-5d40-b01c-bce1b9a60efe", + "type": "image", + "content": "\nMarketing, Human Capital0\n[images/image-45-Marketing,.jpg]\n", + "path": "images/image-45-Marketing,.jpg", + "metadata": { + "length": 60, + "summary": "image-45\nMarketing, Human Capital0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-45-Marketing,.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "9a8d296a-acf6-51da-a992-99463e4057c7", + "type": "text", + "content": "President and Chief Executive Officer, NVIDIA Corporation\nAge: 63\nDirector Since: 1993\nCommittees: None\nOther Current Public Company\nBoards: None\nJen-Hsun Huang founded NVIDIA in 1993 and has served since its inception as President, Chief Executive Officer, and a member of the Board.\nSince its founding, NVIDIA has pioneered accelerated computing. The Company's invention of the GPU in 1999 sparked the growth of the PC gaming market, redefined computer graphics, and ignited the era of modern AI. NVIDIA is now driving the platform shift of accelerated computing and generative AI, transforming the world's largest industries and profoundly impacting society.\nMr. Huang has been elected to the National Academy of Engineering and in 2026 was appointed to the President's Council of Advisors of Science and Technology. He is a recipient of the Semiconductor Industry Association's highest honor, the Robert N. Noyce Award; the IEEE Founder's Medal; the Dr. Morris Chang Exemplary Leadership Award; and honorary doctorate degrees from Taiwan's National Chiao Tung University, National Taiwan University, Oregon State University, Huazhong University of Science and Technology and Linköping University. He has been named the world's best CEO by Fortune, the Economist, and Brand Finance, as well as one of TIME magazine's 100 most influential people.\nPrior to founding NVIDIA, Mr. Huang worked at LSI Logic, a semiconductor and software company, and Advanced Micro Devices, a global semiconductor company. He holds a BSEE degree from Oregon State University and an MSEE degree from Stanford University.\nMr. Huang is one of the technology industry's most respected executives, having taken NVIDIA from a startup to a world leader in accelerated computing. Under his guidance, NVIDIA has compiled a record of consistent innovation and sharp execution, marked by products that have gained strong market share.\n\nJen-Hsun Huang founded NVIDIA in 1993 and has served since its inception as President, Chief Executive Officer, and a member of the Board. Senior Leadership & Operations Experience0\n[images/image-38-Jen-Hsun H.jpg]\n\nSenior Leadership & Operations Experience\n\nMr. Huang has been elected to the National Academy of Engineering and in 2026 was appointed to the President's Council of Advisors of Science and Technology. He is a recipient of the Semiconductor Industry Association's highest honor, the Robert N. Noyce Award; the IEEE Founder's Medal; the Dr. Morris Chang Exemplary Leadership Award; and honorary doctorate degrees from Taiwan's National Chiao Tung University, National Taiwan University, Oregon State University, Huazhong University of Science and Technology and Linköping University. He has been named the world's best CEO by Fortune, the Economist, and Brand Finance, as well as one of TIME magazine's 100 most influential people. Industry & Technical0\n[images/image-39-Mr. Huang .jpg]\n\nIndustry & Technical\n\nMr. Huang is one of the technology industry's most respected executives, having taken NVIDIA from a startup to a world leader in accelerated computing. Under his guidance, NVIDIA has compiled a record of consistent innovation and sharp execution, marked by products that have gained strong market share. Financial/Financial0\n[images/image-40-Mr. Huang .jpg]\n\nFinancial/Financial\nCommunity\n\nIndustry & Technical Governance & Public Company Board0\n[images/image-41-Industry &.jpg]\n\nGovernance & Public Company Board\n\nFinancial/Financial Emerging Technologies & Business Models0\n[images/image-42-Financial_.jpg]\n\nEmerging Technologies & Business Models\n\nCommunity Marketing,0\n[images/image-43-Community .jpg]\n\nMarketing,\nCommunications &\nBrand Management\n\nEmerging Technologies & Business Models Regulatory, Legal & Risk Management0\n[images/image-44-Emerging T.jpg]\n\nRegulatory, Legal & Risk Management\n\nMarketing, Human Capital0\n[images/image-45-Marketing,.jpg]\n\nHuman Capital\nManagement\nExperience", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/JEN-HSUN HUANG", + "metadata": { + "length": 3926, + "summary": "Jen-Hsun Huang, age 63, is the founder, President, and CEO of NVIDIA since 1993. He pioneered accelerated computing and the GPU, driving PC gaming, computer graphics, and modern AI. He has received numerous honors including the Robert N. Noyce Award, IEEE Founder's Medal, and honorary doctorates. He was named world's best CEO by Fortune, Economist, and Brand Finance, and one of TIME's 100 most influential people. Prior to NVIDIA, he worked at LSI Logic and AMD. He holds a BSEE from Oregon State University and an MSEE from Stanford University. The passage also lists his expertise in senior leadership, industry/technical, financial, governance, emerging technologies, community, marketing, regulatory, and human capital management.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "President", + "Chief", + "Executive", + "Officer", + "NVIDIA", + "Corporation", + "Age", + "63", + "Director", + "Since", + "1993", + "Committees", + "None", + "Other", + "Current", + "Public", + "Company", + "Boards", + "Jen", + "Hsun", + "Huang", + "founded", + "served", + "inception", + "member", + "Board", + "founding", + "pioneered", + "accelerated", + "computing", + "The", + "invention", + "GPU", + "1999", + "sparked", + "growth", + "PC", + "gaming", + "market", + "redefined", + "computer", + "graphics", + "ignited", + "era", + "modern", + "AI", + "driving", + "platform", + "shift", + "generative", + "transforming", + "world", + "largest", + "industries", + "profoundly", + "impacting", + "society", + "Mr", + "elected", + "National", + "Academy", + "Engineering", + "2026", + "appointed", + "Council", + "Advisors", + "Science", + "Technology", + "He", + "recipient", + "Semiconductor", + "Industry", + "Association", + "highest", + "honor", + "Robert", + "Noyce", + "Award", + "IEEE", + "Founder", + "Medal", + "Dr", + "Morris", + "Chang", + "Exemplary", + "Leadership", + "honorary", + "doctorate", + "degrees", + "Taiwan", + "Chiao", + "Tung", + "University", + "Oregon", + "State", + "Huazhong", + "Link", + "ping", + "named", + "CEO", + "Fortune", + "Economist", + "Brand", + "Finance", + "TIME", + "magazine", + "100", + "influential", + "people", + "Prior", + "worked", + "LSI", + "Logic", + "semiconductor", + "software", + "company", + "Advanced", + "Micro", + "Devices", + "global", + "holds", + "BSEE", + "degree", + "MSEE", + "Stanford", + "technology", + "industry", + "respected", + "executives", + "startup", + "leader", + "Under", + "guidance", + "compiled", + "record", + "consistent", + "innovation", + "sharp", + "execution", + "marked", + "products", + "gained", + "strong", + "share", + "Senior", + "Operations", + "Experience0", + "Experience", + "Technical0", + "Technical", + "Financial", + "Financial0", + "Community", + "Governance", + "Board0", + "Emerging", + "Technologies", + "Business", + "Models0", + "Models", + "Marketing", + "Communications", + "Management", + "Regulatory", + "Legal", + "Risk", + "Management0", + "Human", + "Capital0", + "Capital" + ], + "keywords": [ + "Jen-Hsun Huang", + "NVIDIA", + "CEO" + ], + "connect_to": [ + { + "target": "85e03b98-d4de-58c2-85e0-d166f977fcb8", + "relation": "embeds", + "ref": "[images/image-38-Jen-Hsun H.jpg]", + "position": { + "start": 2088, + "end": 2120 + } + }, + { + "target": "ff8e4a93-0064-537b-b716-6a7654788817", + "relation": "embeds", + "ref": "[images/image-39-Mr. Huang .jpg]", + "position": { + "start": 2874, + "end": 2906 + } + }, + { + "target": "e47e2e5f-8481-503e-84fb-67017af8874a", + "relation": "embeds", + "ref": "[images/image-40-Mr. Huang .jpg]", + "position": { + "start": 3255, + "end": 3287 + } + }, + { + "target": "889b1346-36ce-54bf-a1e2-d92cb67e9e5f", + "relation": "embeds", + "ref": "[images/image-41-Industry &.jpg]", + "position": { + "start": 3376, + "end": 3408 + } + }, + { + "target": "b8c14ae1-3dce-5d49-9adb-5c433cbf2a3a", + "relation": "embeds", + "ref": "[images/image-42-Financial_.jpg]", + "position": { + "start": 3506, + "end": 3538 + } + }, + { + "target": "918a8757-d542-5d01-a385-9d77625d5e1e", + "relation": "embeds", + "ref": "[images/image-43-Community .jpg]", + "position": { + "start": 3603, + "end": 3635 + } + }, + { + "target": "997509c7-bd1b-59bb-89b5-ab12564fd687", + "relation": "embeds", + "ref": "[images/image-44-Emerging T.jpg]", + "position": { + "start": 3760, + "end": 3792 + } + }, + { + "target": "cd063f0c-f174-5d40-b01c-bce1b9a60efe", + "relation": "embeds", + "ref": "[images/image-45-Marketing,.jpg]", + "position": { + "start": 3857, + "end": 3889 + } + } + ] + } + }, + { + "chunk_id": "85e03b98-d4de-58c2-85e0-d166f977fcb8", + "type": "image", + "content": "\nJen-Hsun Huang founded NVIDIA in 1993 and has served since its inception as President, Chief Executive Officer, and a member of the Board. Senior Leadership & Operations Experience0\n[images/image-38-Jen-Hsun H.jpg]\n", + "path": "images/image-38-Jen-Hsun H.jpg_2", + "metadata": { + "length": 216, + "summary": "image-38\nJen-Hsun Huang founded NVIDIA in 1993 and has served since its inception as President, Chief Executive Officer, and a member of the Board. Senior Leadership & Operations Experience0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-38-Jen-Hsun H.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e3e3a672-24a8-52ec-9701-0199c48bb1b4", + "type": "image", + "content": "\nFinancial Expert Financial/Financial Community0\n[images/image-46-Financial .jpg]\n", + "path": "images/image-46-Financial .jpg", + "metadata": { + "length": 82, + "summary": "image-46\nFinancial Expert Financial/Financial Community0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-46-Financial .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "970a6445-ab2b-55f6-8fb9-100ac35ff4df", + "type": "image", + "content": "\nMs. Hudson brings to the board experience in executive leadership. As a longtime marketing executive, she has valuable expertise and insights in leveraging brands, brand development and consumer behavior. She also has considerable corporate governance experience, gained from more than two decades of serving on the boards of public companies. Governance & Public0\n[images/image-47-Ms. Hudson.jpg]\n", + "path": "images/image-47-Ms. Hudson.jpg", + "metadata": { + "length": 399, + "summary": "image-47\nMs. Hudson brings to the board experience in executive leadership. As a longtime marketing executive, she has valuable expertise and insights in leveraging brands, brand development and consumer behavior. She also has considerable corporate governance experience, gained from more than two decades of serving on the boards of public companies. Governance & Public0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-47-Ms. Hudson.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "205e6a85-8973-5495-87bb-be6d1172cc1d", + "type": "image", + "content": "\nFinancial/Financial Community Marketing,0\n[images/image-48-Financial_.jpg]\n", + "path": "images/image-48-Financial_.jpg", + "metadata": { + "length": 76, + "summary": "image-48\nFinancial/Financial Community Marketing,0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-48-Financial_.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f8622961-9b8f-5a45-9d33-33dc25e87e73", + "type": "image", + "content": "\nCompany Board Human Capital Management Experience0\n[images/image-49-Company Bo.jpg]\n", + "path": "images/image-49-Company Bo.jpg", + "metadata": { + "length": 85, + "summary": "image-49\nCompany Board Human Capital Management Experience0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-49-Company Bo.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "a7ea863b-df79-5769-8a8c-4a4602021229", + "type": "text", + "content": "Former Chief Marketing Officer, National Football League & Former CEO Pepsi-Cola North America\nAge: 68\nDirector Since: 2013\nCommittees: CC\nOther Current Public Company Boards: None\nIndependent Director\nFinancial Expert\nDawn Hudson serves on the boards of various companies. From 2014 to 2018, Ms. Hudson served as Chief Marketing Officer for the National Football League. Previously, she served from 2009 to 2014 as Vice Chairman of The Parthenon Group, an advisory firm focused on strategy consulting. She was President and Chief Executive Officer of Pepsi-Cola North America, the beverage division of PepsiCo, Inc. for the U.S. and Canada, from 2005 to 2007 and President from 2002, and simultaneously served as Chief Executive Officer of the foodservice division of PepsiCo, Inc. from 2005 to 2007. Previously, she spent 13 years in marketing, advertising and branding strategy, holding leadership positions at major agencies, such as D'Arcy Masius Benton & Bowles and Omnicom Group Inc. Ms. Hudson currently serves on the board of directors of a private skincare company and a private consumer diagnostic portfolio company. She was a director of P.F. Chang's China Bistro, Inc., a restaurant chain, from 2010 to 2012; of Allergan, Inc., a biopharmaceutical company, from 2008 to 2014; of Lowes Companies, Inc., a home improvement retailer, from 2001 to 2015; of Amplify Snack Brands, Inc., a snack food company, from 2014 to 2018; of Modern Times Group MTG AB, a gaming company, from 2020 to 2023; and of the Interpublic Group of Companies, Inc., an advertising company, from 2011 to 2025. She holds a BA degree in English from Dartmouth College.\nMs. Hudson brings to the board experience in executive leadership. As a longtime marketing executive, she has valuable expertise and insights in leveraging brands, brand development and consumer behavior. She also has considerable corporate governance experience, gained from more than two decades of serving on the boards of public companies.\n\nJen-Hsun Huang founded NVIDIA in 1993 and has served since its inception as President, Chief Executive Officer, and a member of the Board. Senior Leadership & Operations Experience0\n[images/image-38-Jen-Hsun H.jpg]\n\nSenior Leadership & Operations Experience\n\nFinancial Expert Financial/Financial Community0\n[images/image-46-Financial .jpg]\n\nFinancial/Financial Community\n\nMs. Hudson brings to the board experience in executive leadership. As a longtime marketing executive, she has valuable expertise and insights in leveraging brands, brand development and consumer behavior. She also has considerable corporate governance experience, gained from more than two decades of serving on the boards of public companies. Governance & Public0\n[images/image-47-Ms. Hudson.jpg]\n\nGovernance & Public\nCompany Board\n\nFinancial/Financial Community Marketing,0\n[images/image-48-Financial_.jpg]\n\nMarketing,\nCommunications &\nBrand Management\n\nCompany Board Human Capital Management Experience0\n[images/image-49-Company Bo.jpg]\n\nHuman Capital Management Experience", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/DAWN HUDSON", + "metadata": { + "length": 3044, + "summary": "Dawn Hudson, age 68, is a former Chief Marketing Officer of the National Football League and former CEO of Pepsi-Cola North America. She has served as a director since 2013 and is an independent director and financial expert. Her career includes leadership roles at The Parthenon Group, PepsiCo, and major advertising agencies. She has served on numerous public company boards including P.F. Chang's, Allergan, Lowe's, Amplify Snack Brands, Modern Times Group, and Interpublic Group. She holds a BA in English from Dartmouth College. The passage also mentions Jen-Hsun Huang, founder and CEO of NVIDIA, and includes image references related to financial, marketing, governance, and human capital management experience.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Former", + "Chief", + "Marketing", + "Officer", + "National", + "Football", + "League", + "CEO", + "Pepsi", + "Cola", + "North", + "America", + "Age", + "68", + "Director", + "Since", + "2013", + "Committees", + "CC", + "Other", + "Current", + "Public", + "Company", + "Boards", + "None", + "Independent", + "Financial", + "Expert", + "Dawn", + "Hudson", + "serves", + "boards", + "companies", + "From", + "2014", + "2018", + "Ms", + "served", + "Previously", + "2009", + "Vice", + "Chairman", + "The", + "Parthenon", + "Group", + "advisory", + "firm", + "focused", + "strategy", + "consulting", + "She", + "President", + "Executive", + "beverage", + "division", + "PepsiCo", + "Inc", + "Canada", + "2005", + "2007", + "2002", + "simultaneously", + "foodservice", + "spent", + "13", + "years", + "marketing", + "advertising", + "branding", + "holding", + "leadership", + "positions", + "major", + "agencies", + "Arcy", + "Masius", + "Benton", + "Bowles", + "Omnicom", + "board", + "directors", + "private", + "skincare", + "company", + "consumer", + "diagnostic", + "portfolio", + "director", + "Chang", + "China", + "Bistro", + "restaurant", + "chain", + "2010", + "2012", + "Allergan", + "biopharmaceutical", + "2008", + "Lowes", + "Companies", + "home", + "improvement", + "retailer", + "2001", + "2015", + "Amplify", + "Snack", + "Brands", + "snack", + "food", + "Modern", + "Times", + "MTG", + "AB", + "gaming", + "2020", + "2023", + "Interpublic", + "2011", + "2025", + "holds", + "BA", + "degree", + "English", + "Dartmouth", + "College", + "brings", + "experience", + "executive", + "As", + "longtime", + "valuable", + "expertise", + "insights", + "leveraging", + "brands", + "brand", + "development", + "behavior", + "considerable", + "corporate", + "governance", + "gained", + "decades", + "serving", + "public", + "Jen", + "Hsun", + "Huang", + "founded", + "NVIDIA", + "1993", + "inception", + "member", + "Board", + "Senior", + "Leadership", + "Operations", + "Experience0", + "Experience", + "Community0", + "Community", + "Governance", + "Public0", + "Communications", + "Brand", + "Management", + "Human", + "Capital" + ], + "keywords": [ + "Dawn Hudson", + "Board Experience", + "Marketing Executive" + ], + "connect_to": [ + { + "target": "85e03b98-d4de-58c2-85e0-d166f977fcb8", + "relation": "embeds", + "ref": "[images/image-38-Jen-Hsun H.jpg]", + "position": { + "start": 2178, + "end": 2210 + } + }, + { + "target": "e3e3a672-24a8-52ec-9701-0199c48bb1b4", + "relation": "embeds", + "ref": "[images/image-46-Financial .jpg]", + "position": { + "start": 2303, + "end": 2335 + } + }, + { + "target": "970a6445-ab2b-55f6-8fb9-100ac35ff4df", + "relation": "embeds", + "ref": "[images/image-47-Ms. Hudson.jpg]", + "position": { + "start": 2733, + "end": 2765 + } + }, + { + "target": "205e6a85-8973-5495-87bb-be6d1172cc1d", + "relation": "embeds", + "ref": "[images/image-48-Financial_.jpg]", + "position": { + "start": 2844, + "end": 2876 + } + }, + { + "target": "f8622961-9b8f-5a45-9d33-33dc25e87e73", + "relation": "embeds", + "ref": "[images/image-49-Company Bo.jpg]", + "position": { + "start": 2975, + "end": 3007 + } + } + ] + } + }, + { + "chunk_id": "d6438b1a-a908-5938-a6d9-86b1d8c0f7e7", + "type": "image", + "content": "\nOther Current Public Company Boards: None Senior Leadership & Operations Experience0\n[images/image-50-Other Curr.jpg]\n", + "path": "images/image-50-Other Curr.jpg", + "metadata": { + "length": 119, + "summary": "image-50\nOther Current Public Company Boards: None Senior Leadership & Operations Experience0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-50-Other Curr.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e4e8a88c-a35e-59db-973d-0896b2db8286", + "type": "image", + "content": "\nFinancial Expert Industry & Technical0\n[images/image-51-Financial .jpg]\n", + "path": "images/image-51-Financial .jpg", + "metadata": { + "length": 73, + "summary": "image-51\nFinancial Expert Industry & Technical0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-51-Financial .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "08e10045-0aca-5419-8a08-e36b1a83af56", + "type": "image", + "content": "\nMr. Jones brings to the board an executive management background, an understanding of semiconductor technologies and complex system design. He provides valuable insight into innovation strategies, research and development efforts, as well as management and development of our technical employees. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. Financial/Financial0\n[images/image-52-Mr. Jones .jpg]\n", + "path": "images/image-52-Mr. Jones .jpg", + "metadata": { + "length": 509, + "summary": "image-52\nMr. Jones brings to the board an executive management background, an understanding of semiconductor technologies and complex system design. He provides valuable insight into innovation strategies, research and development efforts, as well as management and development of our technical employees. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. Financial/Financial0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-52-Mr. Jones .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "c4e0235e-3772-5f60-bbba-62cd8300ec53", + "type": "image", + "content": "\nIndustry & Technical Governance & Public0\n[images/image-53-Industry &.jpg]\n", + "path": "images/image-53-Industry &.jpg", + "metadata": { + "length": 76, + "summary": "image-53\nIndustry & Technical Governance & Public0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-53-Industry &.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "bac91c52-1acb-5361-b50b-315f56d31814", + "type": "image", + "content": "\nFinancial/Financial Emerging Technologies & Business Models0\n[images/image-54-Financial_.jpg]\n", + "path": "images/image-54-Financial_.jpg", + "metadata": { + "length": 95, + "summary": "image-54\nFinancial/Financial Emerging Technologies & Business Models0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-54-Financial_.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "7fdb99ce-98e0-5792-b4fd-9cc332173787", + "type": "image", + "content": "\nGovernance & Public Marketing,0\n[images/image-55-Governance.jpg]\n", + "path": "images/image-55-Governance.jpg", + "metadata": { + "length": 66, + "summary": "image-55\nGovernance & Public Marketing,0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-55-Governance.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e4f99059-620c-521f-8458-994ba68a7c5a", + "type": "image", + "content": "\nEmerging Technologies & Business Models Human Capital0\n[images/image-56-Emerging T.jpg]\n", + "path": "images/image-56-Emerging T.jpg", + "metadata": { + "length": 89, + "summary": "image-56\nEmerging Technologies & Business Models Human Capital0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-56-Emerging T.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "a38babf0-20aa-5efe-974d-22625f86e7c1", + "type": "text", + "content": "Managing Partner, Square Wave Ventures\nAge: 73\nDirector Since: 1993\nCommittees: AC, NCGC\nOther Current Public Company Boards: None\nIndependent Director\nFinancial Expert\nHarvey C. Jones has been the Managing Partner of Square Wave Ventures, a private family office, since 2004. Mr. Jones has been an entrepreneur, high technology executive, and active venture investor for over 30 years. In 1981, he co-founded Daisy Systems Corp., a computer-aided engineering company, ultimately serving as its President and Chief Executive Officer until 1987. Between 1987 and 1998, he led Synopsys, Inc., a major electronic design automation company, serving as its Chief Executive Officer for seven years and then as Executive Chairman. In 1997, Mr. Jones co-founded Tensilica Inc., a privately held technology IP company that developed and licensed high performance embedded processing cores. He served as Chairman of the Tensilica board of directors from inception through its 2013 acquisition by Cadence Design Systems, Inc. Mr. Jones holds a BS degree in Mathematics and Computer Sciences from Georgetown University and an MS degree in Management from Massachusetts Institute of Technology.\nMr. Jones brings to the board an executive management background, an understanding of semiconductor technologies and complex system design. He provides valuable insight into innovation strategies, research and development efforts, as well as management and development of our technical employees. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders.\n\nOther Current Public Company Boards: None Senior Leadership & Operations Experience0\n[images/image-50-Other Curr.jpg]\n\nSenior Leadership & Operations Experience\n\nFinancial Expert Industry & Technical0\n[images/image-51-Financial .jpg]\n\nIndustry & Technical\n\nMr. Jones brings to the board an executive management background, an understanding of semiconductor technologies and complex system design. He provides valuable insight into innovation strategies, research and development efforts, as well as management and development of our technical employees. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. Financial/Financial0\n[images/image-52-Mr. Jones .jpg]\n\nFinancial/Financial\nCommunity\n\nIndustry & Technical Governance & Public0\n[images/image-53-Industry &.jpg]\n\nGovernance & Public\nCompany Board\n\nFinancial/Financial Emerging Technologies & Business Models0\n[images/image-54-Financial_.jpg]\n\nEmerging Technologies & Business Models\n\nGovernance & Public Marketing,0\n[images/image-55-Governance.jpg]\n\nMarketing,\nCommunications &\nBrand Management\n\nEmerging Technologies & Business Models Human Capital0\n[images/image-56-Emerging T.jpg]\n\nHuman Capital\nManagement\nExperience", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/HARVEY C. JONES", + "metadata": { + "length": 2917, + "summary": "Harvey C. Jones, age 73, is Managing Partner of Square Wave Ventures and has been a director since 1993. He co-founded Daisy Systems, led Synopsys as CEO, and co-founded Tensilica. He brings executive management, semiconductor technology, and financial expertise to the board.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Managing", + "Partner", + "Square", + "Wave", + "Ventures", + "Age", + "73", + "Director", + "Since", + "1993", + "Committees", + "AC", + "NCGC", + "Other", + "Current", + "Public", + "Company", + "Boards", + "None", + "Independent", + "Financial", + "Expert", + "Harvey", + "Jones", + "private", + "family", + "office", + "2004", + "Mr", + "entrepreneur", + "high", + "technology", + "executive", + "active", + "venture", + "investor", + "30", + "years", + "In", + "1981", + "founded", + "Daisy", + "Systems", + "Corp", + "computer", + "aided", + "engineering", + "company", + "ultimately", + "serving", + "President", + "Chief", + "Executive", + "Officer", + "1987", + "Between", + "1998", + "led", + "Synopsys", + "Inc", + "major", + "electronic", + "design", + "automation", + "Chairman", + "1997", + "Tensilica", + "privately", + "held", + "IP", + "developed", + "licensed", + "performance", + "embedded", + "processing", + "cores", + "He", + "served", + "board", + "directors", + "inception", + "2013", + "acquisition", + "Cadence", + "Design", + "holds", + "BS", + "degree", + "Mathematics", + "Computer", + "Sciences", + "Georgetown", + "University", + "MS", + "Management", + "Massachusetts", + "Institute", + "Technology", + "brings", + "management", + "background", + "understanding", + "semiconductor", + "technologies", + "complex", + "system", + "valuable", + "insight", + "innovation", + "strategies", + "research", + "development", + "efforts", + "technical", + "employees", + "His", + "significant", + "financial", + "community", + "experience", + "Board", + "methods", + "companies", + "increase", + "stockholders", + "Senior", + "Leadership", + "Operations", + "Experience0", + "Experience", + "Industry", + "Technical0", + "Technical", + "Financial0", + "Community", + "Governance", + "Public0", + "Emerging", + "Technologies", + "Business", + "Models0", + "Models", + "Marketing", + "Communications", + "Brand", + "Human", + "Capital0", + "Capital" + ], + "keywords": [ + "Managing Partner", + "Director", + "Semiconductor" + ], + "connect_to": [ + { + "target": "d6438b1a-a908-5938-a6d9-86b1d8c0f7e7", + "relation": "embeds", + "ref": "[images/image-50-Other Curr.jpg]", + "position": { + "start": 1722, + "end": 1754 + } + }, + { + "target": "e4e8a88c-a35e-59db-973d-0896b2db8286", + "relation": "embeds", + "ref": "[images/image-51-Financial .jpg]", + "position": { + "start": 1838, + "end": 1870 + } + }, + { + "target": "08e10045-0aca-5419-8a08-e36b1a83af56", + "relation": "embeds", + "ref": "[images/image-52-Mr. Jones .jpg]", + "position": { + "start": 2369, + "end": 2401 + } + }, + { + "target": "c4e0235e-3772-5f60-bbba-62cd8300ec53", + "relation": "embeds", + "ref": "[images/image-53-Industry &.jpg]", + "position": { + "start": 2476, + "end": 2508 + } + }, + { + "target": "bac91c52-1acb-5361-b50b-315f56d31814", + "relation": "embeds", + "ref": "[images/image-54-Financial_.jpg]", + "position": { + "start": 2606, + "end": 2638 + } + }, + { + "target": "7fdb99ce-98e0-5792-b4fd-9cc332173787", + "relation": "embeds", + "ref": "[images/image-55-Governance.jpg]", + "position": { + "start": 2713, + "end": 2745 + } + }, + { + "target": "e4f99059-620c-521f-8458-994ba68a7c5a", + "relation": "embeds", + "ref": "[images/image-56-Emerging T.jpg]", + "position": { + "start": 2848, + "end": 2880 + } + } + ] + } + }, + { + "chunk_id": "a1c64e36-21cd-51ec-bc09-de0a69d6dd13", + "type": "table", + "content": "
Age: 63Director Since: 2023Committees: AC, NCGC
Other Current Public Company Boards: Conagra Brands, Inc. (since 2019)Independent DirectorFinancial Expert
", + "path": "tables/table-17 Director Profile.html", + "metadata": { + "length": 240, + "summary": "table-18\nDirector aged 63, serving since 2023, on Audit and Nominating/Governance committees. Also a Conagra Brands board member since 2019, independent and a financial expert.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-17 Director Profile.html", + "keywords": [ + "Director", + "Board Committees", + "Public Company" + ], + "tokens": [] + } + }, + { + "chunk_id": "05b7e8ce-f10d-5241-a72d-85029a585b13", + "type": "table", + "content": "
Senior Leadership & Operations ExperienceFinancial/Financial CommunityGovernance & Public Company Board
Emerging Technologies & Business ModelsMarketing, Communications & Brand ManagementHuman Capital Management Experience
", + "path": "tables/table-18 Leadership & Operations.html", + "metadata": { + "length": 906, + "summary": "table-19\nThe table outlines senior leadership, operations, financial, governance, emerging technologies, marketing, and human capital management experience areas.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-18 Leadership & Operations.html", + "keywords": [ + "Leadership", + "Operations", + "Governance" + ], + "tokens": [] + } + }, + { + "chunk_id": "3d6c4694-fc8b-5f15-b503-740c0a021dfa", + "type": "text", + "content": "Former President, Taco Bell International\n\n[tables/table-17 Director Profile.html]\n\nMelissa B. Lora has served in several senior executive roles over her 31-year career at Taco Bell Corp., a subsidiary of Yum! Brands, Inc., one of the world's largest restaurant companies, including as President of Taco Bell International at her retirement in 2018 and Global Chief Financial and Development Officer at Taco Bell Corp. Ms. Lora served on the board of directors of KB Home, a homebuilding company, from 2004 to April 2024, and was a Lead Independent Director thereof from 2016. She was the chair of the audit committee from 2008 to 2018 at KB Home. She has served on the board of directors of Conagra Brands, Inc., a consumer packaged goods holding company, since 2019 and is the chair of the nominating and corporate governance committee and a member of the human resources committee and the executive committee. Ms. Lora previously served on the board of directors of MGIC Investment Corporation from 2018 to 2022. Ms. Lora holds a BS degree in Finance from California State University-Long Beach and an MBA degree emphasizing Corporate Finance from the University of Southern California.\nMs. Lora brings to the Board senior management and operating experience as well as finance experience gained in a large corporate setting. Additionally, she has decades of public company board of director experience in a variety of industries and board governance roles.\n\n[tables/table-18 Leadership & Operations.html]", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/MELISSA B. LORA", + "metadata": { + "length": 1589, + "summary": "Melissa B. Lora is a former President of Taco Bell International with a 31-year career at Taco Bell Corp. She serves on the board of Conagra Brands since 2019, chairing its nominating and corporate governance committee. Previously, she served on the boards of KB Home (2004-2024) and MGIC Investment Corporation (2018-2022). She holds a BS in Finance from California State University-Long Beach and an MBA from USC. Her expertise includes senior management, finance, and public company board governance.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Former", + "President", + "Taco", + "Bell", + "International", + "Melissa", + "Lora", + "served", + "senior", + "executive", + "roles", + "31", + "year", + "career", + "Corp", + "subsidiary", + "Yum", + "Brands", + "Inc", + "world", + "largest", + "restaurant", + "companies", + "including", + "retirement", + "2018", + "Global", + "Chief", + "Financial", + "Development", + "Officer", + "Ms", + "board", + "directors", + "KB", + "Home", + "homebuilding", + "company", + "2004", + "April", + "2024", + "Lead", + "Independent", + "Director", + "thereof", + "2016", + "She", + "chair", + "audit", + "committee", + "2008", + "Conagra", + "consumer", + "packaged", + "goods", + "holding", + "2019", + "nominating", + "corporate", + "governance", + "member", + "human", + "resources", + "previously", + "MGIC", + "Investment", + "Corporation", + "2022", + "holds", + "BS", + "degree", + "Finance", + "California", + "State", + "University", + "Long", + "Beach", + "MBA", + "emphasizing", + "Corporate", + "Southern", + "brings", + "Board", + "management", + "operating", + "experience", + "finance", + "gained", + "large", + "setting", + "Additionally", + "decades", + "public", + "director", + "variety", + "industries" + ], + "keywords": [ + "Taco Bell", + "Board Director", + "Finance" + ], + "connect_to": [ + { + "target": "a1c64e36-21cd-51ec-bc09-de0a69d6dd13", + "relation": "embeds", + "ref": "[tables/table-17 Director Profile.html]", + "position": { + "start": 43, + "end": 82 + } + }, + { + "target": "05b7e8ce-f10d-5241-a72d-85029a585b13", + "relation": "embeds", + "ref": "[tables/table-18 Leadership & Operations.html]", + "position": { + "start": 1462, + "end": 1508 + } + } + ] + } + }, + { + "chunk_id": "c6d6c572-3ed6-5b82-b17b-e1d83de4806d", + "type": "table", + "content": "
Age: 77Director Since: 2019Committees: NCGC
Other Current Public Company Boards: NoneIndependent DirectorLead Director
", + "path": "tables/table-19 Director Profile.html", + "metadata": { + "length": 204, + "summary": "table-20\nA 77-year-old independent director since 2019, serves on NCGC committee, no other public boards, and is Lead Director.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-19 Director Profile.html", + "keywords": [ + "Director", + "Age", + "Committees" + ], + "tokens": [] + } + }, + { + "chunk_id": "4e86c10a-27d6-5c75-8d02-cac700312646", + "type": "table", + "content": "
Senior Leadership & Operations ExperienceGovernance & Public Company BoardMarketing, Communications & Brand Management
Regulatory, Legal & Risk ManagementHuman Capital Management Experience
", + "path": "tables/table-20 Leadership & Governance.html", + "metadata": { + "length": 785, + "summary": "table-21\nThe table lists senior leadership, operations, governance, public company board, marketing, communications, brand management, regulatory, legal, risk management, and human capital management experience.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-20 Leadership & Governance.html", + "keywords": [ + "Leadership", + "Governance", + "Operations" + ], + "tokens": [] + } + }, + { + "chunk_id": "89d460cd-ac1d-53b9-a9ee-8e8ad9f7936e", + "type": "text", + "content": "Chairman Emeritus and Senior Counsel, Cooley LLP\n\n[tables/table-19 Director Profile.html]\n\nStephen C. Neal serves as Chairman Emeritus and Senior Counsel of the law firm Cooley LLP, where he was also Chief Executive Officer from 2001 until 2008. In addition to his extensive experience as a trial lawyer on a broad range of corporate issues, Mr. Neal has represented and advised numerous boards of directors, special committees of boards, and individual directors on corporate governance and other legal matters. Prior to joining Cooley in 1995, Mr. Neal was a partner of the law firm Kirkland & Ellis LLP. Mr. Neal served on the board of directors of Levi Strauss & Co. from 2007 to 2021 and as Chairperson from 2011 to 2021. Mr. Neal also is Chairman of the Oversight Board Trust, a perpetual Delaware special purpose trust. Previously, Mr. Neal served as Chairman of the boards of the William and Flora Hewlett Foundation and of the Monterey Bay Aquarium. Mr. Neal holds an AB degree from Harvard University and a JD degree from Stanford Law School.\nMr. Neal brings to the Board deep knowledge and broad experience in corporate governance as well as his perspectives drawn from advising many companies throughout his career.\n\n[tables/table-20 Leadership & Governance.html]", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/STEPHEN C. NEAL", + "metadata": { + "length": 1352, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Chairman", + "Emeritus", + "Senior", + "Counsel", + "Cooley", + "LLP", + "Stephen", + "Neal", + "serves", + "law", + "firm", + "Chief", + "Executive", + "Officer", + "2001", + "2008", + "In", + "addition", + "extensive", + "experience", + "trial", + "lawyer", + "broad", + "range", + "corporate", + "issues", + "Mr", + "represented", + "advised", + "numerous", + "boards", + "directors", + "special", + "committees", + "individual", + "governance", + "legal", + "matters", + "Prior", + "joining", + "1995", + "partner", + "Kirkland", + "Ellis", + "served", + "board", + "Levi", + "Strauss", + "Co", + "2007", + "2021", + "Chairperson", + "2011", + "Oversight", + "Board", + "Trust", + "perpetual", + "Delaware", + "purpose", + "trust", + "Previously", + "William", + "Flora", + "Hewlett", + "Foundation", + "Monterey", + "Bay", + "Aquarium", + "holds", + "AB", + "degree", + "Harvard", + "University", + "JD", + "Stanford", + "Law", + "School", + "brings", + "deep", + "knowledge", + "perspectives", + "drawn", + "advising", + "companies", + "career" + ], + "keywords": [], + "connect_to": [ + { + "target": "c6d6c572-3ed6-5b82-b17b-e1d83de4806d", + "relation": "embeds", + "ref": "[tables/table-19 Director Profile.html]", + "position": { + "start": 50, + "end": 89 + } + }, + { + "target": "4e86c10a-27d6-5c75-8d02-cac700312646", + "relation": "embeds", + "ref": "[tables/table-20 Leadership & Governance.html]", + "position": { + "start": 1229, + "end": 1275 + } + } + ] + } + }, + { + "chunk_id": "8585b0ee-7cac-50fc-a272-cb734523ca16", + "type": "table", + "content": "
Age: 78Director Since: 1997Committees: AC
Other Current Public Company Boards: Tenable Holdings, Inc. (since 2017)Independent DirectorFinancial Expert
", + "path": "tables/table-21 Director Profile.html", + "metadata": { + "length": 236, + "summary": "table-22\nA 78-year-old independent director since 1997, serves on the Audit Committee and is a financial expert. Also a board member at Tenable Holdings since 2017.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-21 Director Profile.html", + "keywords": [ + "Director", + "Tenable Holdings", + "Financial Expert" + ], + "tokens": [] + } + }, + { + "chunk_id": "63742e4e-b33c-5400-8928-3163ab5f3632", + "type": "image", + "content": "\n### A. BROOKE SEAWELL Senior Leadership & Operations Experience0\n[images/image-57-### A. BRO.jpg]\n", + "path": "images/image-57-### A. BRO.jpg", + "metadata": { + "length": 99, + "summary": "image-57\n### A. BROOKE SEAWELL Senior Leadership & Operations Experience0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-57-### A. BRO.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e3e3a672-24a8-52ec-9701-0199c48bb1b4", + "type": "image", + "content": "\nFinancial Expert Financial/Financial Community0\n[images/image-46-Financial .jpg]\n", + "path": "images/image-46-Financial .jpg_2", + "metadata": { + "length": 82, + "summary": "image-46\nFinancial Expert Financial/Financial Community0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-46-Financial .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "56563371-ace7-5276-952f-dd3ed725cf85", + "type": "image", + "content": "\nSenior Leadership & Operations Experience Governance & Public0\n[images/image-58-Senior Lea.jpg]\n", + "path": "images/image-58-Senior Lea.jpg", + "metadata": { + "length": 97, + "summary": "image-58\nSenior Leadership & Operations Experience Governance & Public0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-58-Senior Lea.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "56d48886-48ea-58d3-afd8-5d3de8bf0c26", + "type": "image", + "content": "\nFinancial/Financial Emerging Technologies & Business Models0\n[images/image-59-Financial_.jpg]\n", + "path": "images/image-59-Financial_.jpg", + "metadata": { + "length": 95, + "summary": "image-59\nFinancial/Financial Emerging Technologies & Business Models0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-59-Financial_.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "cb4c343a-9b10-5dbd-b9f2-7332d61a41cd", + "type": "image", + "content": "\nGovernance & Public Regulatory, Legal & Risk Management0\n[images/image-60-Governance.jpg]\n", + "path": "images/image-60-Governance.jpg", + "metadata": { + "length": 91, + "summary": "image-60\nGovernance & Public Regulatory, Legal & Risk Management0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-60-Governance.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "dbe5b8a2-52ca-5dcc-ba2b-a6a2e86a6325", + "type": "image", + "content": "\nCompany Board Human Capital0\n[images/image-61-Company Bo.jpg]\n", + "path": "images/image-61-Company Bo.jpg", + "metadata": { + "length": 63, + "summary": "image-61\nCompany Board Human Capital0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-61-Company Bo.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "95c6247f-7648-54a1-9885-b99bc5c21ae2", + "type": "table", + "content": "
Age: 61Director Since: 2020Committees: AC, CC
Other Current Public Company Boards: Sandoz Group AG (since 2023)Independent Director
", + "path": "tables/table-22 Director Profile.html", + "metadata": { + "length": 217, + "summary": "table-23\nAge 61, director since 2020, serves on Audit and Compensation Committees. Also on board of Sandoz Group AG since 2023. Independent director.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-22 Director Profile.html", + "keywords": [ + "Director", + "Board", + "Committees" + ], + "tokens": [] + } + }, + { + "chunk_id": "9cfe5873-d793-524f-a713-95b26ef32a0b", + "type": "image", + "content": "\nAARTI SHAH Senior Leadership & Operations Experience0\n[images/image-62-AARTI SHAH.jpg]\n", + "path": "images/image-62-AARTI SHAH.jpg", + "metadata": { + "length": 88, + "summary": "image-62\nAARTI SHAH Senior Leadership & Operations Experience0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-62-AARTI SHAH.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "4776dbfe-4074-5db4-bdfa-366d83bd878f", + "type": "image", + "content": "\nAarti Shah serves on the boards of various companies and non-profit organizations. Dr. Shah worked at Eli Lilly and Company for 27 years until 2021 and served in several functional and business leadership roles, most recently as Senior Vice President and Chief Information and Digital Officer, as well as Senior Statistician, Research Scientist, Vice President for Biometrics, and Global Brand Development Leader in Lilly's Bio-Medicines business unit. Dr. Shah has served on the board and as a member of the science, innovation & development committee and human capital and ESG committee of Sandoz Group AG, a pharmaceutical company, since 2023. Dr. Shah has served on the board of trustees and as a member of the audit committee and distribution and technology committee of Northwestern Mutual since 2020, a financial services organization. Dr. Shah serves on the board, strategic planning committee, and global and technology committee of St. Jude Children's Research Hospital and American Lebanese Syrian Associated Charities (ALSAC), a non-profit organization. She also serves as the secretary and a trustee of the non-profit organization, Shrimad Rajchandra Mission Dharampur USA. She served on the Indianapolis Public Library Foundation board for the full term of 9 years and on the Center for Interfaith Cooperation for the full term of 4 years. Dr. Shah received her BS and MS degrees in Statistics and Mathematics in India before completing her PhD degree in Applied Statistics from the University of California, Riverside. Industry & Technical0\n[images/image-63-Aarti Shah.jpg]\n", + "path": "images/image-63-Aarti Shah.jpg", + "metadata": { + "length": 1590, + "summary": "image-63\nAarti Shah serves on the boards of various companies and non-profit organizations. Dr. Shah worked at Eli Lilly and Company for 27 years until 2021 and served in several functional and business leadership roles, most recently as Senior Vice President and Chief Information and Digital Officer, as well as Senior Statistician, Research Scientist, Vice President for Biometrics, and Global Brand Development Leader in Lilly's Bio-Medicines business unit. Dr. Shah has served on the board and as a member of the science, innovation & development committee and human capital and ESG committee of Sandoz Group AG, a pharmaceutical company, since 2023. Dr. Shah has served on the board of trustees and as a member of the audit committee and distribution and technology committee of Northwestern Mutual since 2020, a financial services organization. Dr. Shah serves on the board, strategic planning committee, and global and technology committee of St. Jude Children's Research Hospital and American Lebanese Syrian Associated Charities (ALSAC), a non-profit organization. She also serves as the secretary and a trustee of the non-profit organization, Shrimad Rajchandra Mission Dharampur USA. She served on the Indianapolis Public Library Foundation board for the full term of 9 years and on the Center for Interfaith Cooperation for the full term of 4 years. Dr. Shah received her BS and MS degrees in Statistics and Mathematics in India before completing her PhD degree in Applied Statistics from the University of California, Riverside. Industry & Technical0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-63-Aarti Shah.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "1ead651a-1e11-5cab-ad39-ba912ccb0b53", + "type": "image", + "content": "\nDr. Shah brings to the Board executive leadership and senior operating experience. Additionally she brings expertise in drug development and technical expertise in the areas of information technology, cybersecurity, advanced analytics, data sciences, and digital health. Governance & Public0\n[images/image-64-Dr. Shah b.jpg]\n", + "path": "images/image-64-Dr. Shah b.jpg", + "metadata": { + "length": 326, + "summary": "image-64\nDr. Shah brings to the Board executive leadership and senior operating experience. Additionally she brings expertise in drug development and technical expertise in the areas of information technology, cybersecurity, advanced analytics, data sciences, and digital health. Governance & Public0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-64-Dr. Shah b.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e521f2ba-a363-5c85-85ef-7b67120eef78", + "type": "image", + "content": "\nIndustry & Technical Emerging Technologies & Business Models0\n[images/image-65-Industry &.jpg]\n", + "path": "images/image-65-Industry &.jpg", + "metadata": { + "length": 96, + "summary": "image-65\nIndustry & Technical Emerging Technologies & Business Models0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-65-Industry &.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "5dd61a61-0d26-555c-af5c-cfb0b69a6da7", + "type": "image", + "content": "\nGovernance & Public Marketing,0\n[images/image-66-Governance.jpg]\n", + "path": "images/image-66-Governance.jpg", + "metadata": { + "length": 66, + "summary": "image-66\nGovernance & Public Marketing,0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-66-Governance.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d3dbd45d-4197-5adc-81a3-dc14e39fb8c2", + "type": "image", + "content": "\nEmerging Technologies & Business Models Regulatory, Legal & Risk Management0\n[images/image-67-Emerging T.jpg]\n", + "path": "images/image-67-Emerging T.jpg", + "metadata": { + "length": 111, + "summary": "image-67\nEmerging Technologies & Business Models Regulatory, Legal & Risk Management0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-67-Emerging T.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "be8755f0-bc1a-5235-8669-b7a5757224d4", + "type": "image", + "content": "\nMarketing, ### MARK A. STEVENS0\n[images/image-68-Marketing,.jpg]\n", + "path": "images/image-68-Marketing,.jpg", + "metadata": { + "length": 66, + "summary": "image-68\nMarketing, ### MARK A. STEVENS0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-68-Marketing,.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "1df28ff6-f211-5bf6-b15e-5f6622046d20", + "type": "text", + "content": "Venture Partner, New Enterprise Associates\n\n[tables/table-21 Director Profile.html]\n\nA. Brooke Seawell has served since 2005 as a Venture Partner at New Enterprise Associates, and was a Partner from 2000 to 2005 at Technology Crossover Ventures. He was Executive Vice President from 1997 to 1998 at NetDynamics, Inc., an application server software company, which was acquired by Sun Microsystems, Inc. He was Senior Vice President and Chief Financial Officer from 1991 to 1997 of Synopsys, Inc., an electronic design automation software company. He serves on the board of directors of Tenable Holdings, Inc., a cybersecurity company, and one privately held company. Mr. Seawell served on the board of directors and as audit chair of Glu Mobile, Inc., a publisher of mobile games, from 2006 to 2014, of Informatica Corp., a data integration software company, from 1997 to 2015, of Tableau Software, Inc., a business intelligence software company, from 2011 to 2019, and of Eargo, Inc., a medical device company, from 2020 to 2022. In 2024, the National Association of Corporate Directors named Mr. Seawell as one of the Top 100 directors in the nation. He also previously served as a member of the Stanford University Athletic Board and on the Management Board of the Stanford Graduate School of Business. Mr. Seawell holds a BA degree in Economics and an MBA degree in Finance from Stanford University.\nMr. Seawell brings to the Board operational expertise and senior management experience, including knowledge of the complex issues facing public companies, and a deep understanding of accounting principles and financial reporting. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders.\n\n### A. BROOKE SEAWELL Senior Leadership & Operations Experience0\n[images/image-57-### A. BRO.jpg]\n\nSenior Leadership & Operations Experience\n\nFinancial Expert Financial/Financial Community0\n[images/image-46-Financial .jpg]\n\nFinancial/Financial\nCommunity\n\nSenior Leadership & Operations Experience Governance & Public0\n[images/image-58-Senior Lea.jpg]\n\nGovernance & Public\nCompany Board\n\nFinancial/Financial Emerging Technologies & Business Models0\n[images/image-59-Financial_.jpg]\n\nEmerging Technologies & Business Models\n\nGovernance & Public Regulatory, Legal & Risk Management0\n[images/image-60-Governance.jpg]\n\nRegulatory, Legal & Risk Management\n\nCompany Board Human Capital0\n[images/image-61-Company Bo.jpg]\n\nHuman Capital\nManagement\nExperience\nAARTI SHAH\nFormer Senior Vice President & Chief Information and Digital Officer, Eli Lilly and Company\n\n[tables/table-22 Director Profile.html]\n\nAarti Shah serves on the boards of various companies and non-profit organizations. Dr. Shah worked at Eli Lilly and Company for 27 years until 2021 and served in several functional and business leadership roles, most recently as Senior Vice President and Chief Information and Digital Officer, as well as Senior Statistician, Research Scientist, Vice President for Biometrics, and Global Brand Development Leader in Lilly's Bio-Medicines business unit. Dr. Shah has served on the board and as a member of the science, innovation & development committee and human capital and ESG committee of Sandoz Group AG, a pharmaceutical company, since 2023. Dr. Shah has served on the board of trustees and as a member of the audit committee and distribution and technology committee of Northwestern Mutual since 2020, a financial services organization. Dr. Shah serves on the board, strategic planning committee, and global and technology committee of St. Jude Children's Research Hospital and American Lebanese Syrian Associated Charities (ALSAC), a non-profit organization. She also serves as the secretary and a trustee of the non-profit organization, Shrimad Rajchandra Mission Dharampur USA. She served on the Indianapolis Public Library Foundation board for the full term of 9 years and on the Center for Interfaith Cooperation for the full term of 4 years. Dr. Shah received her BS and MS degrees in Statistics and Mathematics in India before completing her PhD degree in Applied Statistics from the University of California, Riverside.\nDr. Shah brings to the Board executive leadership and senior operating experience. Additionally she brings expertise in drug development and technical expertise in the areas of information technology, cybersecurity, advanced analytics, data sciences, and digital health.\n\nAARTI SHAH Senior Leadership & Operations Experience0\n[images/image-62-AARTI SHAH.jpg]\n\nSenior Leadership & Operations Experience\n\nAarti Shah serves on the boards of various companies and non-profit organizations. Dr. Shah worked at Eli Lilly and Company for 27 years until 2021 and served in several functional and business leadership roles, most recently as Senior Vice President and Chief Information and Digital Officer, as well as Senior Statistician, Research Scientist, Vice President for Biometrics, and Global Brand Development Leader in Lilly's Bio-Medicines business unit. Dr. Shah has served on the board and as a member of the science, innovation & development committee and human capital and ESG committee of Sandoz Group AG, a pharmaceutical company, since 2023. Dr. Shah has served on the board of trustees and as a member of the audit committee and distribution and technology committee of Northwestern Mutual since 2020, a financial services organization. Dr. Shah serves on the board, strategic planning committee, and global and technology committee of St. Jude Children's Research Hospital and American Lebanese Syrian Associated Charities (ALSAC), a non-profit organization. She also serves as the secretary and a trustee of the non-profit organization, Shrimad Rajchandra Mission Dharampur USA. She served on the Indianapolis Public Library Foundation board for the full term of 9 years and on the Center for Interfaith Cooperation for the full term of 4 years. Dr. Shah received her BS and MS degrees in Statistics and Mathematics in India before completing her PhD degree in Applied Statistics from the University of California, Riverside. Industry & Technical0\n[images/image-63-Aarti Shah.jpg]\n\nIndustry & Technical\n\nDr. Shah brings to the Board executive leadership and senior operating experience. Additionally she brings expertise in drug development and technical expertise in the areas of information technology, cybersecurity, advanced analytics, data sciences, and digital health. Governance & Public0\n[images/image-64-Dr. Shah b.jpg]\n\nGovernance & Public\nCompany Board\n\nIndustry & Technical Emerging Technologies & Business Models0\n[images/image-65-Industry &.jpg]\n\nEmerging Technologies & Business Models\n\nGovernance & Public Marketing,0\n[images/image-66-Governance.jpg]\n\nMarketing,\nCommunications &\nBrand Management\n\nEmerging Technologies & Business Models Regulatory, Legal & Risk Management0\n[images/image-67-Emerging T.jpg]\n\nRegulatory, Legal & Risk Management\n\nMarketing, ### MARK A. STEVENS0\n[images/image-68-Marketing,.jpg]", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/A. BROOKE SEAWELL", + "metadata": { + "length": 7116, + "summary": "The passage profiles two board directors: A. Brooke Seawell, a Venture Partner at New Enterprise Associates with extensive experience in finance, public company governance, and technology; and Aarti Shah, former Senior Vice President at Eli Lilly, bringing expertise in drug development, IT, cybersecurity, and data sciences. Both have served on multiple corporate and non-profit boards.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Venture", + "Partner", + "New", + "Enterprise", + "Associates", + "Brooke", + "Seawell", + "served", + "2005", + "2000", + "Technology", + "Crossover", + "Ventures", + "He", + "Executive", + "Vice", + "President", + "1997", + "1998", + "NetDynamics", + "Inc", + "application", + "server", + "software", + "company", + "acquired", + "Sun", + "Microsystems", + "Senior", + "Chief", + "Financial", + "Officer", + "1991", + "Synopsys", + "electronic", + "design", + "automation", + "serves", + "board", + "directors", + "Tenable", + "Holdings", + "cybersecurity", + "privately", + "held", + "Mr", + "audit", + "chair", + "Glu", + "Mobile", + "publisher", + "mobile", + "games", + "2006", + "2014", + "Informatica", + "Corp", + "data", + "integration", + "2015", + "Tableau", + "Software", + "business", + "intelligence", + "2011", + "2019", + "Eargo", + "medical", + "device", + "2020", + "2022", + "In", + "2024", + "National", + "Association", + "Corporate", + "Directors", + "named", + "Top", + "100", + "nation", + "previously", + "member", + "Stanford", + "University", + "Athletic", + "Board", + "Management", + "Graduate", + "School", + "Business", + "holds", + "BA", + "degree", + "Economics", + "MBA", + "Finance", + "brings", + "operational", + "expertise", + "senior", + "management", + "experience", + "including", + "knowledge", + "complex", + "issues", + "facing", + "public", + "companies", + "deep", + "understanding", + "accounting", + "principles", + "financial", + "reporting", + "His", + "significant", + "community", + "methods", + "increase", + "stockholders", + "BROOKE", + "SEAWELL", + "Leadership", + "Operations", + "Experience0", + "Experience", + "Expert", + "Community0", + "Community", + "Governance", + "Public0", + "Public", + "Company", + "Emerging", + "Technologies", + "Models0", + "Models", + "Regulatory", + "Legal", + "Risk", + "Management0", + "Human", + "Capital0", + "Capital", + "AARTI", + "SHAH", + "Former", + "Information", + "Digital", + "Eli", + "Lilly", + "Aarti", + "Shah", + "boards", + "profit", + "organizations", + "Dr", + "worked", + "27", + "years", + "2021", + "functional", + "leadership", + "roles", + "recently", + "Statistician", + "Research", + "Scientist", + "Biometrics", + "Global", + "Brand", + "Development", + "Leader", + "Bio", + "Medicines", + "unit", + "science", + "innovation", + "development", + "committee", + "human", + "capital", + "ESG", + "Sandoz", + "Group", + "AG", + "pharmaceutical", + "2023", + "trustees", + "distribution", + "technology", + "Northwestern", + "Mutual", + "services", + "organization", + "strategic", + "planning", + "global", + "St", + "Jude", + "Children", + "Hospital", + "American", + "Lebanese", + "Syrian", + "Associated", + "Charities", + "ALSAC", + "She", + "secretary", + "trustee", + "Shrimad", + "Rajchandra", + "Mission", + "Dharampur", + "USA", + "Indianapolis", + "Library", + "Foundation", + "full", + "term", + "Center", + "Interfaith", + "Cooperation", + "received", + "BS", + "MS", + "degrees", + "Statistics", + "Mathematics", + "India", + "completing", + "PhD", + "Applied", + "California", + "Riverside", + "executive", + "operating", + "Additionally", + "drug", + "technical", + "areas", + "information", + "advanced", + "analytics", + "sciences", + "digital", + "health", + "Industry", + "Technical0", + "Technical", + "Marketing", + "Communications", + "MARK", + "STEVENS0" + ], + "keywords": [ + "Board", + "Directors", + "Experience" + ], + "connect_to": [ + { + "target": "8585b0ee-7cac-50fc-a272-cb734523ca16", + "relation": "embeds", + "ref": "[tables/table-21 Director Profile.html]", + "position": { + "start": 44, + "end": 83 + } + }, + { + "target": "63742e4e-b33c-5400-8928-3163ab5f3632", + "relation": "embeds", + "ref": "[images/image-57-### A. BRO.jpg]", + "position": { + "start": 1857, + "end": 1889 + } + }, + { + "target": "e3e3a672-24a8-52ec-9701-0199c48bb1b4", + "relation": "embeds", + "ref": "[images/image-46-Financial .jpg]", + "position": { + "start": 1982, + "end": 2014 + } + }, + { + "target": "56563371-ace7-5276-952f-dd3ed725cf85", + "relation": "embeds", + "ref": "[images/image-58-Senior Lea.jpg]", + "position": { + "start": 2110, + "end": 2142 + } + }, + { + "target": "56d48886-48ea-58d3-afd8-5d3de8bf0c26", + "relation": "embeds", + "ref": "[images/image-59-Financial_.jpg]", + "position": { + "start": 2240, + "end": 2272 + } + }, + { + "target": "cb4c343a-9b10-5dbd-b9f2-7332d61a41cd", + "relation": "embeds", + "ref": "[images/image-60-Governance.jpg]", + "position": { + "start": 2372, + "end": 2404 + } + }, + { + "target": "dbe5b8a2-52ca-5dcc-ba2b-a6a2e86a6325", + "relation": "embeds", + "ref": "[images/image-61-Company Bo.jpg]", + "position": { + "start": 2472, + "end": 2504 + } + }, + { + "target": "95c6247f-7648-54a1-9885-b99bc5c21ae2", + "relation": "embeds", + "ref": "[tables/table-22 Director Profile.html]", + "position": { + "start": 2646, + "end": 2685 + } + }, + { + "target": "9cfe5873-d793-524f-a713-95b26ef32a0b", + "relation": "embeds", + "ref": "[images/image-62-AARTI SHAH.jpg]", + "position": { + "start": 4547, + "end": 4579 + } + }, + { + "target": "4776dbfe-4074-5db4-bdfa-366d83bd878f", + "relation": "embeds", + "ref": "[images/image-63-Aarti Shah.jpg]", + "position": { + "start": 6180, + "end": 6212 + } + }, + { + "target": "1ead651a-1e11-5cab-ad39-ba912ccb0b53", + "relation": "embeds", + "ref": "[images/image-64-Dr. Shah b.jpg]", + "position": { + "start": 6528, + "end": 6560 + } + }, + { + "target": "e521f2ba-a363-5c85-85ef-7b67120eef78", + "relation": "embeds", + "ref": "[images/image-65-Industry &.jpg]", + "position": { + "start": 6659, + "end": 6691 + } + }, + { + "target": "5dd61a61-0d26-555c-af5c-cfb0b69a6da7", + "relation": "embeds", + "ref": "[images/image-66-Governance.jpg]", + "position": { + "start": 6766, + "end": 6798 + } + }, + { + "target": "d3dbd45d-4197-5adc-81a3-dc14e39fb8c2", + "relation": "embeds", + "ref": "[images/image-67-Emerging T.jpg]", + "position": { + "start": 6923, + "end": 6955 + } + }, + { + "target": "be8755f0-bc1a-5235-8669-b7a5757224d4", + "relation": "embeds", + "ref": "[images/image-68-Marketing,.jpg]", + "position": { + "start": 7026, + "end": 7058 + } + } + ] + } + }, + { + "chunk_id": "7a56408b-a1e4-5cd2-8489-f0c7c01b06fb", + "type": "text", + "content": "Managing Partner, S-Cubed Capital\nAge: 66\nDirector Since: 2008\n(previously served 1993-2006)\nCommittees: CC, NCGC", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/MARK A. STEVENS", + "metadata": { + "length": 113, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Managing", + "Partner", + "Cubed", + "Capital", + "Age", + "66", + "Director", + "Since", + "2008", + "previously", + "served", + "1993", + "2006", + "Committees", + "CC", + "NCGC" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1b878d4e-276a-5117-aef2-884adf63c8d8", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/MARK A. STEVENS/Other Current Public Company Boards: None", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b647b222-560a-5cce-9402-7c63d892c410", + "type": "image", + "content": "\nCommittees: CC, NCGC Industry & Technical0\n[images/image-69-Committees.jpg]\n", + "path": "images/image-69-Committees.jpg", + "metadata": { + "length": 77, + "summary": "image-69\nCommittees: CC, NCGC Industry & Technical0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-69-Committees.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "0ceca5a9-4ade-55bf-8189-84f36e67ca7f", + "type": "image", + "content": "\n#### Independent Director Financial/Financial Community0\n[images/image-70-#### Indep.jpg]\n", + "path": "images/image-70-#### Indep.jpg", + "metadata": { + "length": 91, + "summary": "image-70\n#### Independent Director Financial/Financial Community0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-70-#### Indep.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "493282b7-2220-5193-ac56-ad895802fec2", + "type": "image", + "content": "\nMr. Stevens brings to the Board a deep understanding of the technology industry, and the drivers of structural change and high-growth opportunities. He provides valuable insight regarding corporate strategy development and the analysis of acquisitions and divestitures. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. Governance & Public0\n[images/image-71-Mr. Steven.jpg]\n", + "path": "images/image-71-Mr. Steven.jpg", + "metadata": { + "length": 482, + "summary": "image-71\nMr. Stevens brings to the Board a deep understanding of the technology industry, and the drivers of structural change and high-growth opportunities. He provides valuable insight regarding corporate strategy development and the analysis of acquisitions and divestitures. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. Governance & Public0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-71-Mr. Steven.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "cc4df33b-dcb2-50e5-ab4d-aabf6d5b30be", + "type": "image", + "content": "\nFinancial/Financial Community Emerging Technologies & Business Models0\n[images/image-72-Financial_.jpg]\n", + "path": "images/image-72-Financial_.jpg", + "metadata": { + "length": 105, + "summary": "image-72\nFinancial/Financial Community Emerging Technologies & Business Models0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-72-Financial_.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "779c6ef2-3440-53d5-9a30-577a4be0dd8f", + "type": "text", + "content": "Mark A. Stevens has been the Managing Partner of S-Cubed Capital, a private family office investment firm, since 2012. He was a Managing Partner from 1993 to 2011 of Sequoia Capital, a venture capital investment firm, where he had been an associate for the preceding four years. Previously, he held technical sales and marketing positions at Intel Corporation, and was a member of the technical staff at Hughes Aircraft Co. Mr. Stevens is a trustee and the chair of investment committee of the University of Southern California. He was a director of Quantenna Communications, Inc., a provider of Wi-Fi solutions, from 2016 until 2019. Mr. Stevens holds a BSEE degree, a BA degree in Economics and an MS degree in Computer Engineering from the University of Southern California, and an MBA degree from Harvard Business School.\nMr. Stevens brings to the Board a deep understanding of the technology industry, and the drivers of structural change and high-growth opportunities. He provides valuable insight regarding corporate strategy development and the analysis of acquisitions and divestitures. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders.\n\nCommittees: CC, NCGC Industry & Technical0\n[images/image-69-Committees.jpg]\n\nIndustry & Technical\n\n#### Independent Director Financial/Financial Community0\n[images/image-70-#### Indep.jpg]\n\nFinancial/Financial Community\n\nMr. Stevens brings to the Board a deep understanding of the technology industry, and the drivers of structural change and high-growth opportunities. He provides valuable insight regarding corporate strategy development and the analysis of acquisitions and divestitures. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. Governance & Public0\n[images/image-71-Mr. Steven.jpg]\n\nGovernance & Public\nCompany Board\n\nFinancial/Financial Community Emerging Technologies & Business Models0\n[images/image-72-Financial_.jpg]\n\nEmerging Technologies & Business Models", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/MARK A. STEVENS/Independent Director", + "metadata": { + "length": 2136, + "summary": "Mark A. Stevens is Managing Partner of S-Cubed Capital since 2012, formerly a Managing Partner at Sequoia Capital from 1993 to 2011. He held technical roles at Intel and Hughes Aircraft. He is a trustee and investment committee chair at USC, and was a director of Quantenna Communications. He holds degrees from USC and Harvard. His expertise includes technology industry understanding, corporate strategy, acquisitions, and financial community experience. He serves on committees including CC and NCGC, and is recognized for contributions in Industry & Technical, Financial/Financial Community, Governance & Public Company Board, and Emerging Technologies & Business Models.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Mark", + "Stevens", + "Managing", + "Partner", + "Cubed", + "Capital", + "private", + "family", + "office", + "investment", + "firm", + "2012", + "He", + "1993", + "2011", + "Sequoia", + "venture", + "capital", + "associate", + "preceding", + "years", + "Previously", + "held", + "technical", + "sales", + "marketing", + "positions", + "Intel", + "Corporation", + "member", + "staff", + "Hughes", + "Aircraft", + "Co", + "Mr", + "trustee", + "chair", + "committee", + "University", + "Southern", + "California", + "director", + "Quantenna", + "Communications", + "Inc", + "provider", + "Wi", + "Fi", + "solutions", + "2016", + "2019", + "holds", + "BSEE", + "degree", + "BA", + "Economics", + "MS", + "Computer", + "Engineering", + "MBA", + "Harvard", + "Business", + "School", + "brings", + "Board", + "deep", + "understanding", + "technology", + "industry", + "drivers", + "structural", + "change", + "high", + "growth", + "opportunities", + "valuable", + "insight", + "corporate", + "strategy", + "development", + "analysis", + "acquisitions", + "divestitures", + "His", + "significant", + "financial", + "community", + "experience", + "methods", + "companies", + "increase", + "stockholders", + "Committees", + "CC", + "NCGC", + "Industry", + "Technical0", + "Technical", + "Independent", + "Director", + "Financial", + "Community0", + "Community", + "Governance", + "Public0", + "Public", + "Company", + "Emerging", + "Technologies", + "Models0", + "Models" + ], + "keywords": [ + "Technology", + "Finance", + "Board" + ], + "connect_to": [ + { + "target": "b647b222-560a-5cce-9402-7c63d892c410", + "relation": "embeds", + "ref": "[images/image-69-Committees.jpg]", + "position": { + "start": 1297, + "end": 1329 + } + }, + { + "target": "0ceca5a9-4ade-55bf-8189-84f36e67ca7f", + "relation": "embeds", + "ref": "[images/image-70-#### Indep.jpg]", + "position": { + "start": 1410, + "end": 1442 + } + }, + { + "target": "493282b7-2220-5193-ac56-ad895802fec2", + "relation": "embeds", + "ref": "[images/image-71-Mr. Steven.jpg]", + "position": { + "start": 1923, + "end": 1955 + } + }, + { + "target": "cc4df33b-dcb2-50e5-ab4d-aabf6d5b30be", + "relation": "embeds", + "ref": "[images/image-72-Financial_.jpg]", + "position": { + "start": 2063, + "end": 2095 + } + } + ] + } + }, + { + "chunk_id": "f44a2da9-aad9-5a2e-b500-eff41d35ec6a", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fc18ea98-baa1-5100-9f2d-40cf056412ec", + "type": "text", + "content": "Nasdaq rules and our Corporate Governance Policies (as further described below) require that a majority of our directors not have a relationship that would interfere with their exercise of independent judgment in carrying out their responsibilities and that they meet any other qualification requirements required by the SEC and Nasdaq.\nAfter considering all relevant relationships and transactions, our Board determined that, except for Mr. Huang, all of our directors who currently serve on the Board are, and all of our directors who served on the Board during Fiscal 2026 were, “independent” as defined by Nasdaq’s rules and regulations. The Board also determined that all members of our AC, CC, and NCGC are independent under applicable Nasdaq listing standards, and that each of Mr. Seawell, Mr. Jones and Ms. Lora of the AC are “audit committee financial experts” as defined under applicable SEC rules.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Independence of the Members of the Board of Directors", + "metadata": { + "length": 909, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Nasdaq", + "rules", + "Corporate", + "Governance", + "Policies", + "require", + "majority", + "directors", + "relationship", + "interfere", + "exercise", + "independent", + "judgment", + "carrying", + "responsibilities", + "meet", + "qualification", + "requirements", + "required", + "SEC", + "After", + "relevant", + "relationships", + "transactions", + "Board", + "determined", + "Mr", + "Huang", + "serve", + "served", + "Fiscal", + "2026", + "defined", + "regulations", + "The", + "members", + "AC", + "CC", + "NCGC", + "applicable", + "listing", + "standards", + "Seawell", + "Jones", + "Ms", + "Lora", + "audit", + "committee", + "financial", + "experts" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a6725bad-b335-5a7a-94b3-5f48fa856644", + "type": "text", + "content": "Our Board ensures that each member has an equal voice in the affairs and management of NVIDIA by having an independent Lead Director, rather than a chairperson. The independent directors consider the role and designation of the person to serve as Lead Director on an annual basis. The Board recognizes that different board leadership structures may be appropriate under different circumstances and its annual review includes consideration of whether having a Lead Director continues to best meet NVIDIA's evolving needs and serves in the best interest of its stockholders.\nOur Board believes its current leadership structure is appropriate because:\n- The active involvement of each of our independent directors, combined with the qualifications and significant responsibilities of, and strong oversight by, our Lead Director, provide balance on the Board;\n• It promotes independent oversight of our management and affairs;\n- It effectively allocates authority, responsibility, and oversight between management and our independent directors; and\n- It provides the right foundation to pursue the Company's strategic and operational objectives as our business and operating environment continue to evolve.\nOur CEO has primary responsibility for the operational leadership and strategic direction of the Company. Our Lead Director facilitates our Board's independent oversight of management, promotes communication between management and our Board, and supports our Board's consideration of key governance matters. This arrangement promotes open dialogue among the Board, including quarterly executive sessions of independent directors led by our Lead Director without the presence of our CEO.\nWe believe that our current structure best serves our stockholders, without the need to appoint a person to serve as chairperson of the Board. Under our Corporate Governance Policies, the Board may select a chairperson, but if it does not, a Lead Director shall be designated annually by a majority of the independent directors and identified in the Company's proxy statement. These policies help to ensure a robust independent leadership structure on our Board. While the Board has discretion to adopt other leadership structures, including having the Lead Director (or chairperson, if any) and CEO roles filled by a single individual, it would only consider a change if it best aligned with the interests of our stockholders, management, and the Board, and it complied with applicable laws and regulations. If in the future our CEO were to take a leadership position on the Board, such as chairperson, we expect that the Board would continue to appoint an independent Lead Director to maintain a balanced and strong leadership structure and otherwise represent the Board independently from the Company's management team. Any changes to the Board's leadership structure would take into account stockholder views, including through our ongoing stockholder outreach, and would be communicated on our Investor Relations website and in our proxy statement. Our Lead Director may provide input on the design of the Board as requested by the NCGC. In his role as NCGC Chairperson, our Lead Director will continue to lead discussions, provide input, and oversee the design of the Board itself.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Board Leadership Structure", + "metadata": { + "length": 3277, + "summary": "NVIDIA's Board uses an independent Lead Director instead of a chairperson to ensure equal voice and independent oversight. The leadership structure is reviewed annually, with the Lead Director facilitating communication between management and the Board, leading executive sessions, and supporting governance. The CEO handles operational leadership. The Board may appoint a chairperson in the future but would maintain an independent Lead Director to preserve balance and independence.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Board", + "ensures", + "member", + "equal", + "voice", + "affairs", + "management", + "NVIDIA", + "independent", + "Lead", + "Director", + "chairperson", + "The", + "directors", + "role", + "designation", + "person", + "serve", + "annual", + "basis", + "recognizes", + "board", + "leadership", + "structures", + "circumstances", + "review", + "includes", + "consideration", + "continues", + "meet", + "evolving", + "serves", + "interest", + "stockholders", + "believes", + "current", + "structure", + "active", + "involvement", + "combined", + "qualifications", + "significant", + "responsibilities", + "strong", + "oversight", + "provide", + "balance", + "It", + "promotes", + "effectively", + "allocates", + "authority", + "responsibility", + "foundation", + "pursue", + "Company", + "strategic", + "operational", + "objectives", + "business", + "operating", + "environment", + "continue", + "evolve", + "CEO", + "primary", + "direction", + "facilitates", + "communication", + "supports", + "key", + "governance", + "matters", + "This", + "arrangement", + "open", + "dialogue", + "including", + "quarterly", + "executive", + "sessions", + "led", + "presence", + "We", + "appoint", + "Under", + "Corporate", + "Governance", + "Policies", + "select", + "designated", + "annually", + "majority", + "identified", + "proxy", + "statement", + "These", + "policies", + "ensure", + "robust", + "While", + "discretion", + "adopt", + "roles", + "filled", + "single", + "individual", + "change", + "aligned", + "interests", + "complied", + "applicable", + "laws", + "regulations", + "If", + "future", + "position", + "expect", + "maintain", + "balanced", + "represent", + "independently", + "team", + "Any", + "account", + "stockholder", + "views", + "ongoing", + "outreach", + "communicated", + "Investor", + "Relations", + "website", + "input", + "design", + "requested", + "NCGC", + "In", + "Chairperson", + "lead", + "discussions", + "oversee" + ], + "keywords": [ + "Lead Director", + "Independent Oversight", + "Governance" + ], + "connect_to": [] + } + }, + { + "chunk_id": "49e4d792-85ae-5aa5-86a8-f6d38fee8863", + "type": "text", + "content": "Stephen C. Neal\nDirector since 2019\nChairperson of the NCGC\nExperience\n√ Extensive experience as a trial attorney\n√ Has advised numerous companies, boards and individuals on governance and legal matters\n√ Helped clients manage internal and government investigations\n√ Executive experience as Cooley LLP's CEO, and board and chairman experience on the Levi Strauss & Co. board of directors\nThe Board believes Mr. Neal's experience, breadth of knowledge, and contributions to the Board position him well to provide strong leadership, oversight, and the ability to contribute valuable insight with respect to the Company's business.\nOur Lead Director may require Board consideration of risk matters, including adding them to Board agendas or as topics for executive sessions of the independent members of the Board. The Board believes that Mr. Neal is highly qualified to assist the Board as Lead Director in effectively overseeing the identification, assessment, and management of the Company's exposure to various risks as a result of his extensive risk management, legal, and executive experience. Further information on the Board's oversight of risk management and its work with the CEO to address risk management matters is detailed below under Role of the Board in Risk Oversight.\nOur Lead Director has significant responsibilities, which are detailed in our Corporate Governance Policies, and include the duties listed below.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Board Leadership Structure/Our Lead Director", + "metadata": { + "length": 1429, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Stephen", + "Neal", + "Director", + "2019", + "Chairperson", + "NCGC", + "Experience", + "Extensive", + "experience", + "trial", + "attorney", + "Has", + "advised", + "numerous", + "companies", + "boards", + "individuals", + "governance", + "legal", + "matters", + "Helped", + "clients", + "manage", + "internal", + "government", + "investigations", + "Executive", + "Cooley", + "LLP", + "CEO", + "board", + "chairman", + "Levi", + "Strauss", + "Co", + "directors", + "The", + "Board", + "believes", + "Mr", + "breadth", + "knowledge", + "contributions", + "position", + "provide", + "strong", + "leadership", + "oversight", + "ability", + "contribute", + "valuable", + "insight", + "respect", + "Company", + "business", + "Our", + "Lead", + "require", + "consideration", + "risk", + "including", + "adding", + "agendas", + "topics", + "executive", + "sessions", + "independent", + "members", + "highly", + "qualified", + "assist", + "effectively", + "overseeing", + "identification", + "assessment", + "management", + "exposure", + "risks", + "result", + "extensive", + "Further", + "information", + "work", + "address", + "detailed", + "Role", + "Risk", + "Oversight", + "significant", + "responsibilities", + "Corporate", + "Governance", + "Policies", + "include", + "duties", + "listed" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d07f38db-18ae-5473-b871-74310aa7ae82", + "type": "text", + "content": "√ Determining an appropriate schedule of Board meetings and seeking to ensure that the independent members of the Board can perform their duties responsibly while not interfering with the flow of our operations\n√ Working with the CEO, and seeking input from other directors and relevant management, to prepare Board meeting agendas\n√ Advising the CEO on a regular basis as to the quality, quantity, and timeliness of the information requested by the Board from our management, to provide what is necessary for the independent members of the Board to effectively and responsibly perform their duties. Although management is responsible for preparing materials for the Board, the Lead Director may specifically request the inclusion of certain information\n√ Coordinating, developing the agenda for, and moderating executive sessions of the independent members of the Board and acting as principal liaison between them and the CEO on sensitive issues\n√ Presiding over Board meetings when the CEO is not present\n√ Convening meetings of the independent directors, as necessary or appropriate\n√ Performing such other duties as the Board may determine from time to time\nIn addition, our Lead Director may represent the Board in communications with stockholders and other stakeholders. The Lead Director makes themself available for consultation with major stockholders pursuant to our Corporate Governance Policies. As Lead Director, Mr. Neal has participated in our annual stockholder outreach meetings and we expect this practice to continue.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Board Leadership Structure/Duties of Our Lead Director", + "metadata": { + "length": 1537, + "summary": "The Lead Director determines the Board meeting schedule, prepares agendas with the CEO, advises on information quality for independent directors, coordinates executive sessions, presides over meetings in the CEO's absence, convenes independent director meetings, and performs other duties as determined. The Lead Director also represents the Board in communications with stockholders and participates in annual stockholder outreach meetings.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Determining", + "schedule", + "Board", + "meetings", + "seeking", + "ensure", + "independent", + "members", + "perform", + "duties", + "responsibly", + "interfering", + "flow", + "operations", + "Working", + "CEO", + "input", + "directors", + "relevant", + "management", + "prepare", + "meeting", + "agendas", + "Advising", + "regular", + "basis", + "quality", + "quantity", + "timeliness", + "information", + "requested", + "provide", + "effectively", + "Although", + "responsible", + "preparing", + "materials", + "Lead", + "Director", + "specifically", + "request", + "inclusion", + "Coordinating", + "developing", + "agenda", + "moderating", + "executive", + "sessions", + "acting", + "principal", + "liaison", + "sensitive", + "issues", + "Presiding", + "present", + "Convening", + "Performing", + "determine", + "time", + "In", + "addition", + "represent", + "communications", + "stockholders", + "stakeholders", + "The", + "makes", + "themself", + "consultation", + "major", + "pursuant", + "Corporate", + "Governance", + "Policies", + "As", + "Mr", + "Neal", + "participated", + "annual", + "stockholder", + "outreach", + "expect", + "practice", + "continue" + ], + "keywords": [ + "Lead Director", + "Board Meetings", + "Independent Directors" + ], + "connect_to": [] + } + }, + { + "chunk_id": "62baaebb-e9a8-570e-80b3-96d95b3da6ae", + "type": "text", + "content": "The Board has three committees: an AC, a CC, and a NCGC. Each of these committees operates under a written charter, which may be viewed in the Investor Relations section of our website at www.nvidia.com.\nCommittee assignments are determined based on background and the expertise that individual directors can bring to a committee. Our Board believes regular committee rotations are a good corporate governance practice and introduce a variety of perspectives and ideas, more fully inform its members regarding the full scope of the Board and our activities, and benefit each committee and the Board as a whole. Since our 2025 Meeting, Ms. Lora and Dr. Dabiri have joined the NCGC. The composition and functions of our committees are set forth below.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Committees of the Board of Directors", + "metadata": { + "length": 749, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Board", + "committees", + "AC", + "CC", + "NCGC", + "Each", + "operates", + "written", + "charter", + "viewed", + "Investor", + "Relations", + "section", + "website", + "www", + "nvidia", + "Committee", + "assignments", + "determined", + "based", + "background", + "expertise", + "individual", + "directors", + "bring", + "committee", + "Our", + "believes", + "regular", + "rotations", + "good", + "corporate", + "governance", + "practice", + "introduce", + "variety", + "perspectives", + "ideas", + "fully", + "inform", + "members", + "full", + "scope", + "activities", + "benefit", + "Since", + "2025", + "Meeting", + "Ms", + "Lora", + "Dr", + "Dabiri", + "joined", + "composition", + "functions", + "set" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5186cd2c-ed58-5421-9744-d5c515399150", + "type": "text", + "content": "Members as of our 2026 Meeting\n• A. Brooke Seawell (Chairperson)\n- Tench Coxe\n• Harvey C. Jones\n- Melissa B. Lora\n- Aarti Shah\nIn Fiscal 2026, the AC met four times. Selected highlights from its agenda topics included: capitalization review and strategy, tax, treasury, internal audit, information security, enterprise risk management, and insurance reviews.\nCommittee Role and Responsibilities\n• Oversees our corporate accounting and financial reporting process;\n• Oversees our internal audit function;\n- Determines and approves the engagement, compensation, retention, and termination of the independent registered public accounting firm;\n• Evaluates the performance and qualifications of our independent registered public accounting firm;\n- Reviews and approves the retention of the independent registered public accounting firm for permissible audit and non-audit services;\n- Confers with management and our independent registered public accounting firm on the results of the annual audit, our quarterly financial statements and results, and the effectiveness of internal control over financial reporting, including those regarding information security;\n• Reviews the financial statements to be included in our quarterly reports on Form 10-Q and annual reports on Form 10-K;\n- Reviews earnings press releases and the substance of financial information and outlook provided to investors and analysts on earnings calls;\n- Adopts and maintains policies regarding preapproval of employment of individuals employed or formerly employed by auditors and engaged on our account;\n- Prepares their report as required to be included by SEC rules in our annual proxy statement or annual report on Form 10-K;\n- Establishes procedures for the receipt, retention, and treatment of complaints we receive regarding accounting, internal accounting controls or auditing matters and the confidential and anonymous submission by employees of concerns regarding accounting or auditing matters;\n• Oversees risks related to financial reporting and exposures, internal audit functions, regulatory, and accounting policies; and\n- Reviews and reports on the adequacy and effectiveness of the Company's information security policies and practices and the internal controls regarding information security risks", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Committees of the Board of Directors/AC Members as of our 2026 Meeting", + "metadata": { + "length": 2285, + "summary": "The Audit Committee (AC) as of the 2026 meeting includes members A. Brooke Seawell (Chairperson), Tench Coxe, Harvey C. Jones, Melissa B. Lora, and Aarti Shah. In Fiscal 2026, the AC met four times, covering topics such as capitalization review, tax, treasury, internal audit, information security, enterprise risk management, and insurance. The committee oversees corporate accounting and financial reporting, the internal audit function, and the independent registered public accounting firm's engagement, compensation, retention, and termination. It evaluates the accounting firm's performance, approves audit and non-audit services, and confers with management on audit results, quarterly financial statements, and internal control effectiveness. The AC reviews financial statements for SEC filings, earnings press releases, and financial outlooks. It maintains policies on auditor employment, prepares the required SEC report, and establishes procedures for handling accounting and auditing complaints. Additionally, it oversees risks related to financial reporting, internal audit, regulatory compliance, and accounting policies, and reviews the company's information security policies and internal controls.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Members", + "2026", + "Meeting", + "Brooke", + "Seawell", + "Chairperson", + "Tench", + "Coxe", + "Harvey", + "Jones", + "Melissa", + "Lora", + "Aarti", + "Shah", + "In", + "Fiscal", + "AC", + "met", + "times", + "Selected", + "highlights", + "agenda", + "topics", + "included", + "capitalization", + "review", + "strategy", + "tax", + "treasury", + "internal", + "audit", + "information", + "security", + "enterprise", + "risk", + "management", + "insurance", + "reviews", + "Committee", + "Role", + "Responsibilities", + "Oversees", + "corporate", + "accounting", + "financial", + "reporting", + "process", + "function", + "Determines", + "approves", + "engagement", + "compensation", + "retention", + "termination", + "independent", + "registered", + "public", + "firm", + "Evaluates", + "performance", + "qualifications", + "Reviews", + "permissible", + "services", + "Confers", + "results", + "annual", + "quarterly", + "statements", + "effectiveness", + "control", + "including", + "reports", + "Form", + "10", + "earnings", + "press", + "releases", + "substance", + "outlook", + "provided", + "investors", + "analysts", + "calls", + "Adopts", + "maintains", + "policies", + "preapproval", + "employment", + "individuals", + "employed", + "auditors", + "engaged", + "account", + "Prepares", + "report", + "required", + "SEC", + "rules", + "proxy", + "statement", + "Establishes", + "procedures", + "receipt", + "treatment", + "complaints", + "receive", + "controls", + "auditing", + "matters", + "confidential", + "anonymous", + "submission", + "employees", + "concerns", + "risks", + "related", + "exposures", + "functions", + "regulatory", + "adequacy", + "Company", + "practices" + ], + "keywords": [ + "Audit Committee", + "Financial Oversight", + "Risk Management" + ], + "connect_to": [] + } + }, + { + "chunk_id": "7864b03d-26ee-5af4-af85-7a1074fcc355", + "type": "text", + "content": "Members as of our 2026 Meeting\n• Dawn Hudson (Chairperson)\n- Tench Coxe\n• John O. Dabiri\n- Aarti Shah\n• Mark A. Stevens\nIn Fiscal 2026, the CC met three times. Selected highlights from its agenda topics included: executive, employee, and director compensation, review of employee support programs, executive protection, and human capital management.\nCommittee Role and Responsibilities\n• Reviews and approves our overall compensation strategy and policies;\n• Reviews and recommends to the Board the compensation of our Board members;\n• Reviews and approves the compensation of Mr. Huang and other executive officers;\n- Reviews and approves corporate performance goals and objectives related to the compensation of our executive officers and other senior management;\n- Reviews and approves our CD&A disclosure for inclusion in the proxy statement and annual report on Form 10-K;\n- Prepares their report on executive compensation as required to be included by SEC rules in our annual proxy statement or annual report on Form 10-K;\n• Administers our stock purchase plans, variable compensation plans, and other similar programs;\n• Oversees our human capital management;\n• Assesses and monitors whether our compensation policies and programs could create material risks; and\n• Oversees risks related to compensation plans, programs and policies", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Committees of the Board of Directors/CC Members as of our 2026 Meeting", + "metadata": { + "length": 1340, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Members", + "2026", + "Meeting", + "Dawn", + "Hudson", + "Chairperson", + "Tench", + "Coxe", + "John", + "Dabiri", + "Aarti", + "Shah", + "Mark", + "Stevens", + "In", + "Fiscal", + "CC", + "met", + "times", + "Selected", + "highlights", + "agenda", + "topics", + "included", + "executive", + "employee", + "director", + "compensation", + "review", + "support", + "programs", + "protection", + "human", + "capital", + "management", + "Committee", + "Role", + "Responsibilities", + "Reviews", + "approves", + "strategy", + "policies", + "recommends", + "Board", + "members", + "Mr", + "Huang", + "officers", + "corporate", + "performance", + "goals", + "objectives", + "related", + "senior", + "CD", + "disclosure", + "inclusion", + "proxy", + "statement", + "annual", + "report", + "Form", + "10", + "Prepares", + "required", + "SEC", + "rules", + "Administers", + "stock", + "purchase", + "plans", + "variable", + "similar", + "Oversees", + "Assesses", + "monitors", + "create", + "material", + "risks" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1c9939c2-24f7-5387-b057-5e1613655e7c", + "type": "text", + "content": "Members as of our 2026 Meeting\n• Stephen C. Neal (Chairperson)\n• John O. Dabiri\n• Harvey C. Jones\n- Melissa B. Lora\n• Mark A. Stevens\nIn Fiscal 2026, the NCGC met three times. Selected highlights from its agenda topics included: Board recruiting, governmental relations, trade compliance and regulatory matters, stockholder proposals, corporate governance matters, and stockholder outreach.\nCommittee Role and Responsibilities\n• Identifies, reviews, and evaluates candidates to serve as directors;\n• Recommends candidates for election to our Board;\n• Makes recommendations to the Board regarding committee membership and chairpersons;\n• Assesses the performance of the Board and its committees;\n• Reviews and assesses our corporate governance principles and practices;\n• Monitors changes in corporate governance practices, rules, and regulations;\n- Approves related party transactions;\n- Reviews and assesses our CS strategy, risks, and opportunities periodically, including related programs and initiatives;\n• Oversees and reviews policies and practices on governmental relations, trade compliance, regulatory matters, and related risks;\n- Establishes procedures for receiving, retaining, and addressing complaints we receive regarding violations of our Code of Conduct;\n• Monitors the effectiveness of our anonymous tip process; and\n- Oversees the Company's policies, practices, and investigation procedures in connection with the Company's compliance program", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Committees of the Board of Directors/NCGC Members as of our 2026 Meeting", + "metadata": { + "length": 1461, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Members", + "2026", + "Meeting", + "Stephen", + "Neal", + "Chairperson", + "John", + "Dabiri", + "Harvey", + "Jones", + "Melissa", + "Lora", + "Mark", + "Stevens", + "In", + "Fiscal", + "NCGC", + "met", + "times", + "Selected", + "highlights", + "agenda", + "topics", + "included", + "Board", + "recruiting", + "governmental", + "relations", + "trade", + "compliance", + "regulatory", + "matters", + "stockholder", + "proposals", + "corporate", + "governance", + "outreach", + "Committee", + "Role", + "Responsibilities", + "Identifies", + "reviews", + "evaluates", + "candidates", + "serve", + "directors", + "Recommends", + "election", + "Makes", + "recommendations", + "committee", + "membership", + "chairpersons", + "Assesses", + "performance", + "committees", + "Reviews", + "assesses", + "principles", + "practices", + "Monitors", + "rules", + "regulations", + "Approves", + "related", + "party", + "transactions", + "CS", + "strategy", + "risks", + "opportunities", + "periodically", + "including", + "programs", + "initiatives", + "Oversees", + "policies", + "Establishes", + "procedures", + "receiving", + "retaining", + "addressing", + "complaints", + "receive", + "violations", + "Code", + "Conduct", + "effectiveness", + "anonymous", + "tip", + "process", + "Company", + "investigation", + "connection", + "program" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "acf708e3-2556-5a9d-a82c-80a1e236f2ed", + "type": "text", + "content": "At the beginning of Fiscal 2026, the CC initially consisted of Messrs. Burgess and Coxe, Ms. Hudson, and Drs. Dabiri and Shah. After the 2025 Meeting, the CC consisted of Messrs. Coxe and Stevens, Ms. Hudson, and Drs. Dabiri, Drell and\nShah, until Dr. Drell resigned from the Board in January 2026. No member of the CC is an officer or employee of NVIDIA, and none of our executive officers serve as a member of the board or compensation committee of any entity that has one or more executive officers serving as a member of our Board or CC. Other than Dr. Shah, no member of the CC had a relationship requiring disclosure in Review of Transactions with Related Persons below.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Compensation Committee Interlocks and Insider Participation", + "metadata": { + "length": 676, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "At", + "beginning", + "Fiscal", + "2026", + "CC", + "initially", + "consisted", + "Messrs", + "Burgess", + "Coxe", + "Ms", + "Hudson", + "Drs", + "Dabiri", + "Shah", + "After", + "2025", + "Meeting", + "Stevens", + "Drell", + "Dr", + "resigned", + "Board", + "January", + "No", + "member", + "officer", + "employee", + "NVIDIA", + "executive", + "officers", + "serve", + "board", + "compensation", + "committee", + "entity", + "serving", + "Other", + "relationship", + "requiring", + "disclosure", + "Review", + "Transactions", + "Related", + "Persons" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e23c4ef3-5e47-5d15-a0b1-b877b0edeeda", + "type": "table", + "content": "
Board of Directors
Oversees management of major risks, including:
√ Business model, including AI√ Strategic execution√ Product quality and safety
√ Operational, including supply chain and sourcing√ Regulatory, public policy, legal, intellectual property, and compliance√ Financial and macroeconomic
√ Information security, including cybersecurity√ Brand and reputation√ Business continuity
√ Corporate development, acquisitions, and investments√ Management development√ Enterprise resource planning
√ ERM
", + "path": "tables/table-23 Board Risk Oversight.html", + "metadata": { + "length": 753, + "summary": "table-24\nThe Board of Directors oversees major risks including business model, strategy, operations, regulatory, cybersecurity, brand, and enterprise risk management.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-23 Board Risk Oversight.html", + "keywords": [ + "Board of Directors", + "Risk Management", + "Oversight" + ], + "tokens": [] + } + }, + { + "chunk_id": "67936fbf-a425-50bd-bb06-18878d424835", + "type": "table", + "content": "
ACCCNCGC
√ Financial statement and earnings materials integrity and reporting√ Financial risk exposures, including investments, cash management, foreign exchange management and insurance coverage√ Disclosure controls and procedures√ Information security and cybersecurity policies and practices and the internal controls regarding information security risks√ Internal audit performance, including auditor functions, performance, and independence√ Accounting and audit principles and policies, and regulatory and accounting initiatives√ Legal and regulatory compliance, particularly as related to the above matters√ ERM√ Compensation policies, plans, practices and programs for directors, executives, and employees√ Human capital management, including recruiting, retention, development, and other workforce matters√ Governance structure, processes and policies, including regulatory changes and other developments√ Stockholder concerns and communications√ Compliance program and effectiveness of our anonymous tip process√ CS, including environmental, social, and corporate governance matters√ Trade compliance and non-financial regulatory matters√ Board and committee composition and board evaluation√ Related party transactions√ Policies and practices related to government relations, public policy, and related expenditures
", + "path": "tables/table-24 Board Oversight.html", + "metadata": { + "length": 1412, + "summary": "table-25\nThe table outlines board committee responsibilities: Audit covers financial integrity, risk, cybersecurity, internal audit, and compliance; Compensation covers policies and human capital; Governance covers structure, stockholder concerns, CSR, and trade compliance.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-24 Board Oversight.html", + "keywords": [ + "Governance", + "Risk", + "Compliance" + ], + "tokens": [] + } + }, + { + "chunk_id": "265b7c63-d9ec-55a7-93d5-70d3409a39b0", + "type": "text", + "content": "The Board reviews risks, oversees risk management, and delegates oversight of appropriate topics to its committees to support the Company's long-term objectives. The oversight responsibility of our Board and its committees is enabled by management reporting processes, including our ERM process, that are designed to provide visibility to our Board about the identification, assessment, and management of critical risks and management's risk mitigation strategies.\nRISK OVERSIGHT AT NVIDIA\n\n[tables/table-23 Board Risk Oversight.html]\n\n\n[tables/table-24 Board Oversight.html]\n\nManagement\nManagement identifies, evaluates, and mitigates business risks and reports to the Board on them\nInternal Audit\nProvides independent assurance on design and effectiveness of internal controls and governance processes\nOur Board retains direct oversight of strategic risks to NVIDIA and other risk areas not delegated to its committees. Delegation to Board committees enhances risk oversight by allocating authority and responsibility to the committee best suited to oversee specific risks, as outlined in their charters, with escalation to the full Board when appropriate. Committee chairpersons regularly report to the full Board on reviewed matters, including key risks, and collaborate with the Board to ensure effective risk management oversight. The Board believes its leadership structure facilitates robust risk oversight, with the independent Lead Director and independent committees proactively engaging with management. For further details, see Board Leadership Structure above.\nThe Board retains full oversight of information security matters, including cybersecurity. The AC reviews the adequacy and effectiveness of the Company's information security policies, practices, and internal controls. The AC receives regular updates on information security from management, including the Chief Security Officer and security team, while the Board receives annual reports on these matters. The AC also meets in executive session with leaders of key control functions, ensuring direct access to management teams and appropriate staffing and resources.\nThe Company's ERM process is an annual assessment of its risk environment, integrating risk evaluation with the Company's operations and strategies. This process examines internal and external factors, risk amplifiers, and emerging trends to prioritize mitigation efforts based on the risk horizon. It evaluates the potential impact and likelihood of risks materializing over relevant timeframes, considers future threats and trends, and identifies the actions, strategies, processes, controls, and procedures in place or to be implemented to manage and mitigate these risks.\nERM Process\n√ Annual assessment of the Company’s risk environment led by management\n√ Identifies and evaluates risks across different timeframes, including short-, intermediate- and long-term\nIdentifies, assesses, and manages the Company's most significant risks and uncertainties that could materially impact the long-term health of the Company or prevent the achievement of strategic objectives\n√ Regular updates reported to senior management, including CEO\n√ Overseen and reviewed by the Board and AC, at least annually\n√ Board and committees have direct access to management to receive updates on risk exposures and mitigation strategies, and give feedback on key and emerging risks, including cybersecurity, trade compliance, risk management and the ERM process\n√ Board and committee agendas adjusted throughout the year by CEO and Lead Director to address emerging risks and key topics\n√ Includes interviews with senior management and Board members to identify major risks\n√ The Board, its committees, and senior management may engage outside advisors, experts, and consultants to help develop and analyze the Company's risk management and mitigation efforts, and anticipate future threats and trends\nWhile the Company does not have a Chief Compliance Officer, the ERM process and action plan are reviewed by the CEO, other NEOs reporting directly to the CEO, and senior management members. These leaders are responsible for managing and monitoring key risks within their respective functional areas. The Company's internal audit team administers the ERM program on behalf of management.\nThe ERM process is designed to align the Board's risk oversight with the Company's disclosure controls and procedures. Management involved in the ERM process prepares reports that are reviewed by the Board or its committees to ensure that disclosure controls and procedures function effectively.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Role of the Board in Risk Oversight", + "metadata": { + "length": 4600, + "summary": "The Board oversees risks and delegates to committees. Management identifies and mitigates risks, reporting to the Board. Internal Audit provides independent assurance. The Board retains oversight of strategic and information security risks, with the Audit Committee reviewing cybersecurity. The ERM process is an annual assessment integrating risk evaluation with operations, examining internal/external factors, and prioritizing mitigation. It is led by management, reviewed by the Board and Audit Committee, and aligns with disclosure controls. The CEO and senior management review the ERM action plan, with internal audit administering the program.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Board", + "reviews", + "risks", + "oversees", + "risk", + "management", + "delegates", + "oversight", + "topics", + "committees", + "support", + "Company", + "long", + "term", + "objectives", + "responsibility", + "enabled", + "reporting", + "processes", + "including", + "ERM", + "process", + "designed", + "provide", + "visibility", + "identification", + "assessment", + "critical", + "mitigation", + "strategies", + "RISK", + "OVERSIGHT", + "AT", + "NVIDIA", + "Management", + "identifies", + "evaluates", + "mitigates", + "business", + "reports", + "Internal", + "Audit", + "Provides", + "independent", + "assurance", + "design", + "effectiveness", + "internal", + "controls", + "governance", + "Our", + "retains", + "direct", + "strategic", + "areas", + "delegated", + "Delegation", + "enhances", + "allocating", + "authority", + "committee", + "suited", + "oversee", + "specific", + "outlined", + "charters", + "escalation", + "full", + "Committee", + "chairpersons", + "regularly", + "report", + "reviewed", + "matters", + "key", + "collaborate", + "ensure", + "effective", + "believes", + "leadership", + "structure", + "facilitates", + "robust", + "Lead", + "Director", + "proactively", + "engaging", + "For", + "details", + "Leadership", + "Structure", + "information", + "security", + "cybersecurity", + "AC", + "adequacy", + "policies", + "practices", + "receives", + "regular", + "updates", + "Chief", + "Security", + "Officer", + "team", + "annual", + "meets", + "executive", + "session", + "leaders", + "control", + "functions", + "ensuring", + "access", + "teams", + "staffing", + "resources", + "environment", + "integrating", + "evaluation", + "operations", + "This", + "examines", + "external", + "factors", + "amplifiers", + "emerging", + "trends", + "prioritize", + "efforts", + "based", + "horizon", + "It", + "potential", + "impact", + "likelihood", + "materializing", + "relevant", + "timeframes", + "considers", + "future", + "threats", + "actions", + "procedures", + "place", + "implemented", + "manage", + "mitigate", + "Process", + "Annual", + "led", + "Identifies", + "short", + "intermediate", + "assesses", + "manages", + "significant", + "uncertainties", + "materially", + "health", + "prevent", + "achievement", + "Regular", + "reported", + "senior", + "CEO", + "Overseen", + "annually", + "receive", + "exposures", + "give", + "feedback", + "trade", + "compliance", + "agendas", + "adjusted", + "year", + "address", + "Includes", + "interviews", + "members", + "identify", + "major", + "engage", + "advisors", + "experts", + "consultants", + "develop", + "analyze", + "anticipate", + "While", + "Compliance", + "action", + "plan", + "NEOs", + "directly", + "These", + "responsible", + "managing", + "monitoring", + "respective", + "functional", + "audit", + "administers", + "program", + "behalf", + "align", + "disclosure", + "involved", + "prepares", + "function", + "effectively" + ], + "keywords": [ + "Risk Management", + "Board Oversight", + "ERM Process" + ], + "connect_to": [ + { + "target": "e23c4ef3-5e47-5d15-a0b1-b877b0edeeda", + "relation": "embeds", + "ref": "[tables/table-23 Board Risk Oversight.html]", + "position": { + "start": 491, + "end": 534 + } + }, + { + "target": "67936fbf-a425-50bd-bb06-18878d424835", + "relation": "embeds", + "ref": "[tables/table-24 Board Oversight.html]", + "position": { + "start": 537, + "end": 575 + } + } + ] + } + }, + { + "chunk_id": "e893a611-089c-5577-b910-66998683cc24", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "bdc7503c-0e98-505c-815b-929573ada8a9", + "type": "text", + "content": "As discussed above, the NCGC and the Board understand the importance of Board refreshment. On May 7, 2026, the Board approved an increase in its size from ten to eleven directors and appointed Suzanne Nora Johnson to fill the newly created directorship, in each case effective July 13, 2026. Ms. Nora Johnson will join the Board effective July 13, 2026, due to a prior professional commitment. Ms. Nora Johnson is the retired Vice Chairman of The Goldman Sachs Group, Inc., where she spent 21 years in various leadership roles. She currently serves on the board of directors of Pfizer Inc., and serves on the Boards of Trustees of The Brookings Institution, the Broad Foundation, and the University of Southern California. Ms. Nora Johnson will bring to the Board extensive senior leadership experience, experience managing large, complex, global institutions and deep knowledge of global financial markets, risk management, and regulatory issues. Ms. Nora Johnson is not currently a member of the Board and is not a nominee for election at the 2026 Meeting. She is expected to join the AC on the effective date of her appointment.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Board Refreshment", + "metadata": { + "length": 1131, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "As", + "discussed", + "NCGC", + "Board", + "understand", + "importance", + "refreshment", + "On", + "May", + "2026", + "approved", + "increase", + "size", + "ten", + "eleven", + "directors", + "appointed", + "Suzanne", + "Nora", + "Johnson", + "fill", + "newly", + "created", + "directorship", + "case", + "effective", + "July", + "13", + "Ms", + "join", + "due", + "prior", + "professional", + "commitment", + "retired", + "Vice", + "Chairman", + "The", + "Goldman", + "Sachs", + "Group", + "Inc", + "spent", + "21", + "years", + "leadership", + "roles", + "She", + "serves", + "board", + "Pfizer", + "Boards", + "Trustees", + "Brookings", + "Institution", + "Broad", + "Foundation", + "University", + "Southern", + "California", + "bring", + "extensive", + "senior", + "experience", + "managing", + "large", + "complex", + "global", + "institutions", + "deep", + "knowledge", + "financial", + "markets", + "risk", + "management", + "regulatory", + "issues", + "member", + "nominee", + "election", + "Meeting", + "expected", + "AC", + "date", + "appointment" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "744a4255-a711-5af5-a992-268394101c88", + "type": "text", + "content": "The Board has adopted Corporate Governance Policies to ensure that the Board has the necessary authority and processes in place to review and evaluate our business operations as needed and to make decisions that are independent of our management. These policies include practices the Board follows with respect to its composition and selection, regular evaluations of the Board and its committees, Board meetings and involvement of senior management, senior management development, and Board committees and compensation. These policies may be viewed in the Investor Relations section of our website at www.nvidia.com.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Corporate Governance Policies", + "metadata": { + "length": 617, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Board", + "adopted", + "Corporate", + "Governance", + "Policies", + "ensure", + "authority", + "processes", + "place", + "review", + "evaluate", + "business", + "operations", + "needed", + "make", + "decisions", + "independent", + "management", + "These", + "policies", + "include", + "practices", + "respect", + "composition", + "selection", + "regular", + "evaluations", + "committees", + "meetings", + "involvement", + "senior", + "development", + "compensation", + "viewed", + "Investor", + "Relations", + "section", + "website", + "www", + "nvidia" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "68f99db7-2d46-5dbd-aa50-a8ad1882e085", + "type": "text", + "content": "As required under Nasdaq's listing standards, our independent directors meet regularly in scheduled executive sessions at which only independent directors are present, as well as in sessions with the CEO. In Fiscal 2026, our independent directors met in both types of executive sessions at all four of our scheduled quarterly Board meetings.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Executive Sessions of the Board", + "metadata": { + "length": 341, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "As", + "required", + "Nasdaq", + "listing", + "standards", + "independent", + "directors", + "meet", + "regularly", + "scheduled", + "executive", + "sessions", + "present", + "CEO", + "In", + "Fiscal", + "2026", + "met", + "types", + "quarterly", + "Board", + "meetings" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ed60d892-772d-5b55-aecc-dc740e47d762", + "type": "text", + "content": "We expect that our directors will attend each annual meeting, absent a valid reason. All Board members as of our 2025 Meeting attended our 2025 Meeting.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Director Attendance at Annual Meeting", + "metadata": { + "length": 152, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "expect", + "directors", + "attend", + "annual", + "meeting", + "absent", + "valid", + "reason", + "All", + "Board", + "members", + "2025", + "Meeting", + "attended" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ffd47b72-aa29-51d2-a2a6-02277ba9261e", + "type": "text", + "content": "The NCGC oversees an evaluation process, conducted at least annually, whereby outside counsel for NVIDIA interviews each director to obtain his or her evaluation of the Board as a whole, and of the committees on which he or she serves. The interviews solicit ideas from the directors about, among other things, improving the quality of Board and/or committee oversight, effectiveness regarding strategic direction, financial and audit matters, executive compensation, acquisition activity, and other key topics. The interviews also focus on Board process and identifying specific issues for future discussion. After these evaluations are complete, our outside counsel summarizes the results, reviews them with our Lead Director, and then submits the summary for NCGC discussion. The NCGC Chair and outside counsel then report evaluation results to the full Board.\nIn response to the evaluations conducted in Fiscal 2026, our Board determined to focus on geopolitical and regulatory risks, government affairs, supply chain, cybersecurity, AI regulations, growth planning, the use and applications of AI, the Company's strategic roadmap, and management development. The Board also determined to continue to focus on the Board's composition and process for Board refreshment.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Board Self-Assessments", + "metadata": { + "length": 1272, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "NCGC", + "oversees", + "evaluation", + "process", + "conducted", + "annually", + "counsel", + "NVIDIA", + "interviews", + "director", + "obtain", + "Board", + "committees", + "serves", + "solicit", + "ideas", + "directors", + "things", + "improving", + "quality", + "committee", + "oversight", + "effectiveness", + "strategic", + "direction", + "financial", + "audit", + "matters", + "executive", + "compensation", + "acquisition", + "activity", + "key", + "topics", + "focus", + "identifying", + "specific", + "issues", + "future", + "discussion", + "After", + "evaluations", + "complete", + "summarizes", + "results", + "reviews", + "Lead", + "Director", + "submits", + "summary", + "Chair", + "report", + "full", + "In", + "response", + "Fiscal", + "2026", + "determined", + "geopolitical", + "regulatory", + "risks", + "government", + "affairs", + "supply", + "chain", + "cybersecurity", + "AI", + "regulations", + "growth", + "planning", + "applications", + "Company", + "roadmap", + "management", + "development", + "continue", + "composition", + "refreshment" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f2527e3f-fd25-5fe0-bbd4-0f07f1792916", + "type": "text", + "content": "The NCGC and our General Counsel are responsible for new director orientation and for administering or approving eligible director continuing education programs. Continuing education programs for directors may include a combination of internally developed materials and presentations, programs presented by third parties, and financial and administrative support for attendance at qualifying academic or other independent programs.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Director Orientation and Continuing Education", + "metadata": { + "length": 431, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "NCGC", + "General", + "Counsel", + "responsible", + "director", + "orientation", + "administering", + "approving", + "eligible", + "continuing", + "education", + "programs", + "Continuing", + "directors", + "include", + "combination", + "internally", + "developed", + "materials", + "presentations", + "presented", + "parties", + "financial", + "administrative", + "support", + "attendance", + "qualifying", + "academic", + "independent" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "61bfa9a9-2039-502c-9eab-7a4924f918f2", + "type": "text", + "content": "Our directors are expected to devote sufficient time to Board and committee duties and to understanding the Company's business. The NCGC reviews the other commitments of potential Board candidates, and does so annually for existing Board members, to determine if this expectation can be met. In making this determination, the NCGC considers, among other factors, stakeholder guidelines regarding numerical limits on public company boards on which a director may sit. None of our directors serve on more than two public company boards (including NVIDIA), and none of our non-employee directors serve as a CEO or executive officer of a public company.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Director Time Commitment and Outside Board Memberships", + "metadata": { + "length": 649, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "directors", + "expected", + "devote", + "sufficient", + "time", + "Board", + "committee", + "duties", + "understanding", + "Company", + "business", + "The", + "NCGC", + "reviews", + "commitments", + "potential", + "candidates", + "annually", + "existing", + "members", + "determine", + "expectation", + "met", + "In", + "making", + "determination", + "considers", + "factors", + "stakeholder", + "guidelines", + "numerical", + "limits", + "public", + "company", + "boards", + "director", + "sit", + "None", + "serve", + "including", + "NVIDIA", + "employee", + "CEO", + "executive", + "officer" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d414c000-eee1-5712-bb9d-add2223f45f4", + "type": "text", + "content": "Our Corporate Governance Policies require each non-employee director to hold shares of our common stock with a total value equal to six times the annual cash retainer for Board service during the period in which he or she serves as a director (or ten times his base salary, in the case of the CEO). The shares may include vested deferred stock, shares held in trust, and shares held by immediate family members, but not unvested equity awards. Non-employee directors have five years after their Board appointment to reach the ownership threshold. Our stock ownership guidelines are intended to further align director interests with stockholder interests.\nEach non-employee director and Mr. Huang currently meets or exceeds the stock ownership requirements.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Director Stock Ownership Guidelines", + "metadata": { + "length": 756, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Corporate", + "Governance", + "Policies", + "require", + "employee", + "director", + "hold", + "shares", + "common", + "stock", + "total", + "equal", + "times", + "annual", + "cash", + "retainer", + "Board", + "service", + "period", + "serves", + "ten", + "base", + "salary", + "case", + "CEO", + "The", + "include", + "vested", + "deferred", + "held", + "trust", + "family", + "members", + "unvested", + "equity", + "awards", + "Non", + "directors", + "years", + "appointment", + "reach", + "ownership", + "threshold", + "guidelines", + "intended", + "align", + "interests", + "stockholder", + "Each", + "Mr", + "Huang", + "meets", + "exceeds", + "requirements" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2b5ea90b-74d3-5f66-ab3d-d377ae92ebce", + "type": "text", + "content": "We believe in extensive leadership development. Our internal leadership structure ensures that many senior managers work directly with our CEO to execute our corporate strategies. This approach results in strategic alignment, exposes leaders to a broad spectrum of corporate activities, and presents the Board with a pool of excellent candidates for future promotion. Our CEO also selects senior leaders to engage directly with the Board on key initiatives and provides periodic updates to our Board on management development.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Senior Management Development", + "metadata": { + "length": 526, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "extensive", + "leadership", + "development", + "Our", + "internal", + "structure", + "ensures", + "senior", + "managers", + "work", + "directly", + "CEO", + "execute", + "corporate", + "strategies", + "This", + "approach", + "results", + "strategic", + "alignment", + "exposes", + "leaders", + "broad", + "spectrum", + "activities", + "presents", + "Board", + "pool", + "excellent", + "candidates", + "future", + "promotion", + "selects", + "engage", + "key", + "initiatives", + "periodic", + "updates", + "management" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4a9d676d-7c6f-58ae-a009-42647ce3da56", + "type": "text", + "content": "The Board and each of its principal committees may retain outside advisors and consultants of their choosing at our expense. The Board does not need management's consent to retain outside advisors and the principal committees do not need consent from the Board or management' to retain outside advisors.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Outside Advisors", + "metadata": { + "length": 303, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Board", + "principal", + "committees", + "retain", + "advisors", + "consultants", + "choosing", + "expense", + "management", + "consent" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cf879c0b-479f-5786-b330-a95ad8a4f38a", + "type": "text", + "content": "Our directors, executives, and employees are expected to conduct themselves with the highest degree of integrity, ethics, and honesty. Our credibility and reputation depend upon their good judgment, ethical standards, and personal integrity. Our Code of Conduct applies to all executive officers, directors, and employees, including our principal executive officer, principal financial officer, and principal accounting officer. The Financial Team Code of Conduct applies to our executive officers, directors, members of our finance department, and all employees involved in the preparation and review of externally-reported periodic financial reports, filings, and documents. We regularly review our Code of Conduct and related policies to ensure that they provide clear guidance to our directors, executives, and employees. We also regularly train our employees on our Code of Conduct and other policies.\nThe Code of Conduct and the Financial Team Code of Conduct may be viewed in the Investor Relations section of our website at www.nvidia.com. If we make any amendments to either code, or grant any waiver from a provision of either code to any executive officer or director, we will promptly disclose the nature of the amendment or waiver on our website or in a report on Form 8-K. Information contained on our website is not incorporated by reference into this or any other report we file with the SEC.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Code of Conduct", + "metadata": { + "length": 1408, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "directors", + "executives", + "employees", + "expected", + "conduct", + "highest", + "degree", + "integrity", + "ethics", + "honesty", + "credibility", + "reputation", + "depend", + "good", + "judgment", + "ethical", + "standards", + "personal", + "Code", + "Conduct", + "applies", + "executive", + "officers", + "including", + "principal", + "officer", + "financial", + "accounting", + "The", + "Financial", + "Team", + "members", + "finance", + "department", + "involved", + "preparation", + "review", + "externally", + "reported", + "periodic", + "reports", + "filings", + "documents", + "We", + "regularly", + "related", + "policies", + "ensure", + "provide", + "clear", + "guidance", + "train", + "viewed", + "Investor", + "Relations", + "section", + "website", + "www", + "nvidia", + "If", + "make", + "amendments", + "code", + "grant", + "waiver", + "provision", + "director", + "promptly", + "disclose", + "nature", + "amendment", + "report", + "Form", + "Information", + "contained", + "incorporated", + "reference", + "file", + "SEC" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "be8ee362-e05a-5ed9-947a-f6e31f61dc11", + "type": "text", + "content": "Our Insider Trading Policy governs transactions in our securities and applies to the Board and our employees, contractors, and consultants. We believe our Insider Trading Policy is reasonably designed to promote compliance with insider trading laws, rules and regulations applicable to NVIDIA. Section 16 directors and officers must obtain preclearance before trading in NVIDIA's stock. In addition, we comply with applicable laws and regulations relating to insider trading with respect to transactions in our securities such as open market repurchases of our common stock from time to time.\nUnder our Insider Trading Policy, hedging ownership of NVIDIA stock, including but not limited to trading in options, puts, calls, or other derivative instruments related to NVIDIA stock or debt, is not permitted. Additional prohibitions include purchasing NVIDIA stock on margin or holding NVIDIA stock in a margin account, or pledging NVIDIA stock as collateral for a loan. We allow participation in certain exchange funds for portfolio diversification.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Insider Trading Policy", + "metadata": { + "length": 1048, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Insider", + "Trading", + "Policy", + "governs", + "transactions", + "securities", + "applies", + "Board", + "employees", + "contractors", + "consultants", + "We", + "designed", + "promote", + "compliance", + "insider", + "trading", + "laws", + "rules", + "regulations", + "applicable", + "NVIDIA", + "Section", + "16", + "directors", + "officers", + "obtain", + "preclearance", + "stock", + "In", + "addition", + "comply", + "relating", + "respect", + "open", + "market", + "repurchases", + "common", + "time", + "Under", + "hedging", + "ownership", + "including", + "limited", + "options", + "puts", + "calls", + "derivative", + "instruments", + "related", + "debt", + "permitted", + "Additional", + "prohibitions", + "include", + "purchasing", + "margin", + "holding", + "account", + "pledging", + "collateral", + "loan", + "participation", + "exchange", + "funds", + "portfolio", + "diversification" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "137f8aad-e765-5f07-bede-a93b9b0ff465", + "type": "text", + "content": "We have established an independent corporate hotline to allow any employee, contractor, customer, or partner to confidentially and anonymously submit a complaint about any accounting, internal controls, auditing, Code of Conduct, or other matter of concern (unless prohibited by local privacy laws).", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Corporate Hotline", + "metadata": { + "length": 299, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "established", + "independent", + "corporate", + "hotline", + "employee", + "contractor", + "customer", + "partner", + "confidentially", + "anonymously", + "submit", + "complaint", + "accounting", + "internal", + "controls", + "auditing", + "Code", + "Conduct", + "matter", + "concern", + "prohibited", + "local", + "privacy", + "laws" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4ce5e644-e436-585f-bf20-f61f787bde33", + "type": "text", + "content": "Stockholders who wish to communicate with the Board regarding nominations of directors or other matters may do so by sending electronic written communications addressed to Timothy S. Teter, our Secretary, at shareholdermeeting@nvidia.com. All stockholder communications we receive that are addressed to the Board will be compiled by our Secretary. If no particular director is named, letters will be forwarded, depending on the subject matter, to the chairperson of the AC, CC, or NCGC. Matters put forth by our stockholders will be reviewed by the NCGC, which will determine whether these matters should be presented to the Board. The NCGC will give serious consideration to all such matters and will make its determination in accordance with its charter and applicable laws.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Stockholder Communications with the Board of Directors", + "metadata": { + "length": 776, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Stockholders", + "communicate", + "Board", + "nominations", + "directors", + "matters", + "sending", + "electronic", + "written", + "communications", + "addressed", + "Timothy", + "Teter", + "Secretary", + "shareholdermeeting", + "nvidia", + "All", + "stockholder", + "receive", + "compiled", + "If", + "director", + "named", + "letters", + "forwarded", + "depending", + "subject", + "matter", + "chairperson", + "AC", + "CC", + "NCGC", + "Matters", + "put", + "stockholders", + "reviewed", + "determine", + "presented", + "The", + "give", + "consideration", + "make", + "determination", + "accordance", + "charter", + "applicable", + "laws" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "64d7db42-438e-5066-a90c-02d825981c6f", + "type": "text", + "content": "Under our Bylaws, in an uncontested election, stockholders will be given the choice to cast votes FOR or AGAINST the election of directors or to ABSTAIN from such vote. A director shall be elected by the affirmative vote of the majority of the votes cast with respect to that director, meaning the number of shares voted FOR a director must exceed the number of votes cast AGAINST that director. If the votes cast FOR an incumbent director in a non-contested election do not exceed the number of AGAINST votes, such incumbent director shall offer to tender his or her resignation to the Board. The NCGC or other committee that may be designated by the Board will make a recommendation to the Board on whether to accept or reject the resignation or whether other action should be taken. The Board will act on such committee's recommendation and publicly disclose its decision and the rationale within 90 days from the date of certification of the election results. In making their decision, such committee and the Board will evaluate the best interests of the Company and its stockholders and shall consider all factors and information deemed relevant. The director who tenders his or her resignation will not participate in such committee's recommendation or the Board's decision.\nIn a contested election, in which the number of nominees exceeds the number of directors to be elected, stockholders will be given the choice to cast FOR or WITHHOLD votes for the election of directors. Our directors will be elected by a plurality of the shares represented at any such meeting or by proxy and entitled to vote on the election of directors at that meeting. The directors receiving the greatest number of FOR votes will be elected.\nIn either case, abstentions and broker non-votes will each be counted as present for purposes of determining the presence of a quorum but will otherwise have no effect on the vote.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Majority Vote Standard", + "metadata": { + "length": 1908, + "summary": "The text describes director election procedures under company Bylaws. In uncontested elections, directors are elected by majority of votes cast (FOR votes must exceed AGAINST votes). If an incumbent fails to achieve majority, they must offer resignation, which the Board's committee reviews and decides within 90 days. In contested elections, directors are elected by plurality of votes cast. Abstentions and broker non-votes count for quorum but have no effect on election outcomes.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Under", + "Bylaws", + "uncontested", + "election", + "stockholders", + "choice", + "cast", + "votes", + "FOR", + "AGAINST", + "directors", + "ABSTAIN", + "vote", + "director", + "elected", + "affirmative", + "majority", + "respect", + "meaning", + "number", + "shares", + "voted", + "exceed", + "If", + "incumbent", + "contested", + "offer", + "tender", + "resignation", + "Board", + "The", + "NCGC", + "committee", + "designated", + "make", + "recommendation", + "accept", + "reject", + "action", + "act", + "publicly", + "disclose", + "decision", + "rationale", + "90", + "days", + "date", + "certification", + "results", + "In", + "making", + "evaluate", + "interests", + "Company", + "factors", + "information", + "deemed", + "relevant", + "tenders", + "participate", + "nominees", + "exceeds", + "WITHHOLD", + "Our", + "plurality", + "represented", + "meeting", + "proxy", + "entitled", + "receiving", + "greatest", + "case", + "abstentions", + "broker", + "counted", + "present", + "purposes", + "determining", + "presence", + "quorum", + "effect" + ], + "keywords": [ + "majority voting", + "plurality voting", + "resignation policy" + ], + "connect_to": [] + } + }, + { + "chunk_id": "4444c2b5-de9a-51c1-ae02-533aadd90fa2", + "type": "text", + "content": "As part of our Board and management's comprehensive review of current corporate governance practices, our Board amended our Bylaws in 2024 to let stockholders owning at least $15\\%$ of our shares, and who have owned such shares continuously for at least one year, to request a special meeting of stockholders. Those requestors must also satisfy the disclosure, timing and other requirements set forth in our Bylaws intended to ensure that stockholders receive adequate, timely, and accurate information in connection with a special meeting. Our Board believes that this special meeting right strikes a balance between enabling urgent action and protecting the Company and its stockholders from narrow, short-term interests.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Stockholder Special Meeting Right", + "metadata": { + "length": 723, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "As", + "part", + "Board", + "management", + "comprehensive", + "review", + "current", + "corporate", + "governance", + "practices", + "amended", + "Bylaws", + "2024", + "stockholders", + "owning", + "15", + "shares", + "owned", + "continuously", + "year", + "request", + "special", + "meeting", + "Those", + "requestors", + "satisfy", + "disclosure", + "timing", + "requirements", + "set", + "intended", + "ensure", + "receive", + "adequate", + "timely", + "accurate", + "information", + "connection", + "Our", + "believes", + "strikes", + "balance", + "enabling", + "urgent", + "action", + "protecting", + "Company", + "narrow", + "short", + "term", + "interests" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "29c30f87-5e18-52b0-a7bc-8f48243d95f2", + "type": "text", + "content": "The Board met five times during Fiscal 2026, including meetings during which the Board discussed the strategic direction of NVIDIA, new business and strategic opportunities, the product roadmap, and other matters facing NVIDIA. We expect each Board member to attend each meeting of the Board and the committees on which he or she serves. Each Board member attended 75% or more of the applicable meetings of the Board and of each committee on which he or she served during Fiscal 2026.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Board Meeting Information", + "metadata": { + "length": 484, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Board", + "met", + "times", + "Fiscal", + "2026", + "including", + "meetings", + "discussed", + "strategic", + "direction", + "NVIDIA", + "business", + "opportunities", + "product", + "roadmap", + "matters", + "facing", + "We", + "expect", + "member", + "attend", + "meeting", + "committees", + "serves", + "Each", + "attended", + "75%", + "applicable", + "committee", + "served" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d65bfff9-c2c7-56df-b4d9-d32c6b798b13", + "type": "text", + "content": "NVIDIA invents computing technologies that improve lives and address global challenges. Our goal is to integrate sound CS principles and practices across the Company. Our Board and management believe that environmental stewardship, social responsibility, and solid governance are important to our business strategy and long-term value creation. While the full Board has ultimate responsibility for CS matters that impact our business, each committee of the Board oversees CS matters across our business operations in the areas that align with their respective responsibilities. The NCGC is responsible for reviewing and discussing with management our policies, issues, and reporting related to sustainability, including overall sustainability strategy, risks, and opportunities, and related programs and initiatives. Our CS team updates the NCGC at least semiannually on these topics, as well as pertinent regulations and stakeholder inputs, and gathers feedback from the NCGC on issues such as climate change and human rights. The CS team also reports on sustainability issues to the full Board annually.\nOur CSSC comprises members of our executive leadership team. The CSSC oversees and provides input on our sustainability strategy and program. Feedback from the Board, the NCGC and CSSC, along with specific input from our executive team, helps to determine the focus and scope of our sustainability strategy and program.\nMore information can be found on the Corporate Sustainability section of our website and in our annual Sustainability Report. Information contained on our website or in our annual Sustainability Report is not incorporated by reference into this or any other report we file with the SEC. Refer to “Item 1A. Risk Factors” in our Form 10-K for a discussion of risks and uncertainties we face related to CS.\nClimate and Efficiency\nIn Fiscal 2025, we purchased or generated enough renewable energy to match 100% of our global electricity usage for sites under our operational control.\nIn Fiscal 2026, we announced two new emission reduction targets validated by the Science Based Targets Initiative. NVIDIA's goal is to, from a Fiscal 2023 base year:\n- Reduce absolute scope 1 and 2 emissions by $50\\%$ by Fiscal 2030\n- Reduce scope 3 emissions intensity from use of sold GPU products by 75% per petaFLOP of computing performance by Fiscal 2030", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Sustainability", + "metadata": { + "length": 2365, + "summary": "NVIDIA integrates sound computer science principles and practices across the company, with environmental stewardship, social responsibility, and solid governance central to its business strategy. The full Board has ultimate responsibility for corporate sustainability (CS) matters, with the Nominating, Governance and Corporate Citizenship Committee (NCGC) overseeing sustainability strategy, risks, and opportunities. The CS team updates the NCGC at least semiannually and reports to the full Board annually. The Corporate Sustainability Steering Committee (CSSC), composed of executive leadership, provides input on sustainability strategy. In Fiscal 2025, NVIDIA matched 100% of global electricity usage for sites under operational control with renewable energy. In Fiscal 2026, it announced Science Based Targets initiative-validated goals: reduce absolute scope 1 and 2 emissions by 50% by Fiscal 2030 (from a Fiscal 2023 base year), and reduce scope 3 emissions intensity from use of sold GPU products by 75% per petaFLOP of computing performance by Fiscal 2030.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "NVIDIA", + "invents", + "computing", + "technologies", + "improve", + "lives", + "address", + "global", + "challenges", + "Our", + "goal", + "integrate", + "sound", + "CS", + "principles", + "practices", + "Company", + "Board", + "management", + "environmental", + "stewardship", + "social", + "responsibility", + "solid", + "governance", + "important", + "business", + "strategy", + "long", + "term", + "creation", + "While", + "full", + "ultimate", + "matters", + "impact", + "committee", + "oversees", + "operations", + "areas", + "align", + "respective", + "responsibilities", + "The", + "NCGC", + "responsible", + "reviewing", + "discussing", + "policies", + "issues", + "reporting", + "related", + "sustainability", + "including", + "risks", + "opportunities", + "programs", + "initiatives", + "team", + "updates", + "semiannually", + "topics", + "pertinent", + "regulations", + "stakeholder", + "inputs", + "gathers", + "feedback", + "climate", + "change", + "human", + "rights", + "reports", + "annually", + "CSSC", + "comprises", + "members", + "executive", + "leadership", + "input", + "program", + "Feedback", + "specific", + "helps", + "determine", + "focus", + "scope", + "More", + "information", + "found", + "Corporate", + "Sustainability", + "section", + "website", + "annual", + "Report", + "Information", + "contained", + "incorporated", + "reference", + "report", + "file", + "SEC", + "Refer", + "Item", + "1A", + "Risk", + "Factors", + "Form", + "10", + "discussion", + "uncertainties", + "face", + "Climate", + "Efficiency", + "In", + "Fiscal", + "2025", + "purchased", + "generated", + "renewable", + "energy", + "match", + "100%", + "electricity", + "usage", + "sites", + "operational", + "control", + "2026", + "announced", + "emission", + "reduction", + "targets", + "validated", + "Science", + "Based", + "Targets", + "Initiative", + "2023", + "base", + "year", + "Reduce", + "absolute", + "emissions", + "50", + "2030", + "intensity", + "sold", + "GPU", + "products", + "75%", + "petaFLOP", + "performance" + ], + "keywords": [ + "sustainability", + "governance", + "emissions" + ], + "connect_to": [] + } + }, + { + "chunk_id": "2b88307a-2d6f-5432-941c-dc0f2208ce13", + "type": "text", + "content": "Our NCGC oversees and periodically reviews our public policy engagement and accountability. We engage in the public policy process to advance the long-term interests of the Company and its stockholders. NVIDIA's U.S. federal lobbying activities are disclosed quarterly and publicly reported. We belong to trade associations worldwide. Management reports to the NCGC about our policies and practices in connection with government relations, public policy, and related expenditures.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Public Policy Engagement and Accountability", + "metadata": { + "length": 480, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "NCGC", + "oversees", + "periodically", + "reviews", + "public", + "policy", + "engagement", + "accountability", + "We", + "engage", + "process", + "advance", + "long", + "term", + "interests", + "Company", + "stockholders", + "NVIDIA", + "federal", + "lobbying", + "activities", + "disclosed", + "quarterly", + "publicly", + "reported", + "belong", + "trade", + "associations", + "worldwide", + "Management", + "reports", + "policies", + "practices", + "connection", + "government", + "relations", + "related", + "expenditures" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d1ee420b-e9b3-5f6a-86a1-2b1e854d0418", + "type": "text", + "content": "NVIDIA's non-employee director compensation program is designed to attract and retain a world-class Board of Directors while balancing our stockholders' interests, and aligns with annual service starting on the dates of our annual meetings. We do not pay additional fees for serving as Lead Director or as chairperson or member of our committees, or for meeting attendance. Directors who are also employees do not receive compensation for service on the Board.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Director Compensation", + "metadata": { + "length": 460, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "NVIDIA", + "employee", + "director", + "compensation", + "program", + "designed", + "attract", + "retain", + "world", + "class", + "Board", + "Directors", + "balancing", + "stockholders", + "interests", + "aligns", + "annual", + "service", + "starting", + "dates", + "meetings", + "We", + "pay", + "additional", + "fees", + "serving", + "Lead", + "Director", + "chairperson", + "member", + "committees", + "meeting", + "attendance", + "employees", + "receive" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f284fbb5-99c9-5d78-ac27-8911d198e3d3", + "type": "text", + "content": "The CC reviews our director compensation annually with the assistance of its independent compensation consultant, Semler Brossy, and assesses peer group data, compensation trends, and best practices. Following this review, the CC recommended the non-employee director compensation for the year following our 2025 Meeting, or the 2025 Program, as described below, which the Board approved in December 2024.\n2025 Program\nThe 2025 Program maintained the same target compensation as the previous year with a total value of \\$340,000—slightly below the median paid to peer group non-employee directors (based on the peer group determined in late 2024):\n\\$85,000 Annual Cash Retainer Paid Quarterly\n\\$255,000 Equity Retainer RSUs (1)\n(1) Represents target annual value of RSUs granted on the day following our 2025 Meeting, or the 2025 Program RSUs. 50% of the 2025 Program RSUs vested on the third Wednesday in November 2025 and 50% will vest on the third Wednesday in May 2026, subject to the director's continuous service.\nThe number of shares subject to each director's 2025 Program RSUs equaled the target value of the grant divided by the 30-calendar day trailing average closing price of our common stock that ended the business day before the 2025 Meeting, to smooth the effects of market volatility.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Director Compensation/Roles of the Board, the Compensation Committee, and Compensation Consultant", + "metadata": { + "length": 1302, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "CC", + "reviews", + "director", + "compensation", + "annually", + "assistance", + "independent", + "consultant", + "Semler", + "Brossy", + "assesses", + "peer", + "group", + "data", + "trends", + "practices", + "Following", + "review", + "recommended", + "employee", + "year", + "2025", + "Meeting", + "Program", + "Board", + "approved", + "December", + "2024", + "maintained", + "target", + "previous", + "total", + "340", + "000", + "slightly", + "median", + "paid", + "directors", + "based", + "determined", + "late", + "85", + "Annual", + "Cash", + "Retainer", + "Paid", + "Quarterly", + "255", + "Equity", + "RSUs", + "Represents", + "annual", + "granted", + "day", + "50%", + "vested", + "Wednesday", + "November", + "vest", + "May", + "2026", + "subject", + "continuous", + "service", + "number", + "shares", + "equaled", + "grant", + "divided", + "30", + "calendar", + "trailing", + "average", + "closing", + "price", + "common", + "stock", + "ended", + "business", + "smooth", + "effects", + "market", + "volatility" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b856fff4-5036-5bcf-b9ee-b61ab61ad914", + "type": "text", + "content": "Any new non-employee director who is appointed or elected to the Board with an effective date after the 2026 Meeting, or a New Director, will be entitled to receive (i) an annual cash retainer of \\$85,000, prorated to reflect the New Director's service, paid quarterly, (ii) an initial grant of RSUs with a target value of \\$255,000, converted to a number of shares by dividing such value by the 30-calendar day trailing average closing price of our common stock ending the business day before the New Director's effective date of appointment or election, 1/6th of which will vest approximately every six months following the grant date, and (iii) a second grant of RSUs with a target value of \\$255,000, converted to a number of shares by dividing such value by the 30-calendar day trailing average closing price of our common stock ending the business day before the 2026 Meeting, prorated to reflect the New Director's service until the 2027 Meeting, or the New Director Annual Grant. 50% of the New Director Annual Grant will vest on the third Wednesday in May 2027, and the balance will vest on the third Wednesday in November 2026; provided that if the New Director's first day of service is on or after November 1, 2026, the New Director Annual Grant will vest in full on the third Wednesday in May 2027 ((ii) and (iii) collectively referred to as the New Director RSUs).", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Director Compensation/New Directors", + "metadata": { + "length": 1378, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Any", + "employee", + "director", + "appointed", + "elected", + "Board", + "effective", + "date", + "2026", + "Meeting", + "New", + "Director", + "entitled", + "receive", + "annual", + "cash", + "retainer", + "85", + "000", + "prorated", + "reflect", + "service", + "paid", + "quarterly", + "ii", + "initial", + "grant", + "RSUs", + "target", + "255", + "converted", + "number", + "shares", + "dividing", + "30", + "calendar", + "day", + "trailing", + "average", + "closing", + "price", + "common", + "stock", + "ending", + "business", + "appointment", + "election", + "6th", + "vest", + "approximately", + "months", + "iii", + "2027", + "Annual", + "Grant", + "50%", + "Wednesday", + "May", + "balance", + "November", + "provided", + "full", + "collectively", + "referred" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1319bc32-f176-5d2b-9142-059a5c0d7e2d", + "type": "text", + "content": "Non-employee directors can elect to defer the settlement of RSUs upon vesting for tax planning purposes. Deferrals can be made to the earlier of (i) the third Wednesday of March of a future year (no sooner than 2027 for the 2025 Program RSUs or 2028 for the New Director RSUs), or the 15th of March of any year that is 2029 or later, or (ii) the director's cessation of service or certain change-in-control events, in accordance with Section 409A of the Internal Revenue Code.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Director Compensation/Deferred Compensation Program", + "metadata": { + "length": 476, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Non", + "employee", + "directors", + "elect", + "defer", + "settlement", + "RSUs", + "vesting", + "tax", + "planning", + "purposes", + "Deferrals", + "made", + "earlier", + "Wednesday", + "March", + "future", + "year", + "sooner", + "2027", + "2025", + "Program", + "2028", + "New", + "Director", + "15th", + "2029", + "ii", + "director", + "cessation", + "service", + "change", + "control", + "events", + "accordance", + "Section", + "409A", + "Internal", + "Revenue", + "Code" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a03b84fc-cfd5-535f-bef5-f3a63414b818", + "type": "table", + "content": "
NameFees Earned or Paid in Cash ($)Stock Awards ($) (1)Total ($)
Robert K. Burgess (2)63,750278,809342,559
Tench Coxe85,000278,809363,809
John O. Dabiri85,000278,809363,809
Persis S. Drell85,000278,809363,809
Dawn Hudson85,000278,809363,809
Harvey C. Jones85,000278,809363,809
Melissa B. Lora85,000278,809363,809
Stephen C. Neal85,000278,809363,809
Ellen Ochoa (3)42,500278,809321,309
A. Brooke Seawell85,000278,809363,809
Aarti Shah85,000278,809363,809
Mark A. Stevens85,000278,809363,809
", + "path": "tables/table-25 Director Compensation.html", + "metadata": { + "length": 1077, + "summary": "table-26\nThe table lists director compensation for 12 individuals, showing cash fees, stock awards, and total amounts, with most receiving $85,000 cash and $278,809 in stock.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-25 Director Compensation.html", + "keywords": [ + "Director", + "Compensation", + "Stock Awards" + ], + "tokens": [] + } + }, + { + "chunk_id": "370c36ac-7471-52fc-b136-1ddbe702c5ac", + "type": "table", + "content": "
NameRSUsNameRSUs
Robert K. Burgess (1)Melissa B. Lora (3)1,880
Tench Coxe900Stephen C. Neal900
John O. Dabiri900Ellen Ochoa (4)
Persis S. Drell (2)A. Brooke Seawell900
Dawn Hudson900Aarti Shah900
Harvey C. Jones900Mark A. Stevens900
", + "path": "tables/table-26 RSU Holdings Table.html", + "metadata": { + "length": 559, + "summary": "table-27\nThe table lists board members and their restricted stock units (RSUs). Robert K. Burgess, Persis S. Drell, and Ellen Ochoa have zero RSUs; others hold 900 or 1,880 RSUs.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-26 RSU Holdings Table.html", + "keywords": [ + "RSUs", + "Board Members", + "Stock Awards" + ], + "tokens": [] + } + }, + { + "chunk_id": "2b1fce4a-0910-55b4-a0a1-bd3ce51f2e3e", + "type": "text", + "content": "Our directors are reimbursed for expenses incurred in attending Board and committee meetings, and continuing educational programs. We do not offer change-in-control benefits to directors, except for vesting acceleration under our equity plans that applies to all plan award holders if an acquirer does not assume or substitute those awards, provided that the award holder's continuous service with us has not terminated prior to the change-in-control. If a director's service terminates due to death, their RSU grants will immediately vest in full.\nDirectors do not receive dividends on unvested, or vested but deferred, RSUs.\nDirector Compensation for Fiscal 2026\n\n[tables/table-25 Director Compensation.html]\n\n(1) Amounts shown do not reflect amounts actually received by the director. Instead, these amounts reflect the aggregate full grant date fair value calculated in accordance with ASC 718 for RSU awards granted during the fiscal year. The assumptions used in the calculation are set forth in Note 3 to our consolidated financial statements titled Stock-Based Compensation in our Form 10-K. On June 26, 2025, each non-employee director then serving on the Board received their 2025 Program RSU grant for 1,799 shares, with an ASC 718 grant date fair value per share of \\$154.98.\n(2) Mr. Burgess passed away in December 2025, prior to the date of the final quarterly cash retainer payment under the 2025 Program.\n(3) Dr. Ochoa resigned from the Board in July 2025, prior to the dates of the two final quarterly cash retainer payments under the 2025 Program.\nThe following table provides information regarding the aggregate number of unvested RSUs held by each of our non-employee directors as of January 25, 2026:\n\n[tables/table-26 RSU Holdings Table.html]\n\n(1) Mr. Burgess passed away in December 2025. Pursuant to the terms of our 2007 Plan, upon his death, unvested RSUs then held by Mr. Burgess accelerated and became fully vested.\n(2) Dr. Drell resigned from the Board in January 2026.\n(3) Includes initial RSUs granted to Ms. Lora in connection with her appointment to the Board in 2023.\n(4) Dr. Ochoa resigned from the Board in July 2025. Pursuant to the terms of our non-employee director compensation program, RSUs previously deferred by Dr. Ochoa were issued on the date of her resignation.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Director Compensation/Other Compensation∕Benefits", + "metadata": { + "length": 2337, + "summary": "The passage details director compensation policies, including expense reimbursement, change-in-control benefits, and vesting acceleration upon death. It presents a table of director compensation for fiscal 2026, showing fees earned and stock awards (RSUs) with grant date fair values. Notes explain that amounts reflect ASC 718 fair values, not actual receipts, and mention specific events like Mr. Burgess's death and Dr. Ochoa's resignation affecting payments. A second table lists unvested RSUs held by non-employee directors as of January 25, 2026, with notes on vesting acceleration due to death or resignation.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "directors", + "reimbursed", + "expenses", + "incurred", + "attending", + "Board", + "committee", + "meetings", + "continuing", + "educational", + "programs", + "We", + "offer", + "change", + "control", + "benefits", + "vesting", + "acceleration", + "equity", + "plans", + "applies", + "plan", + "award", + "holders", + "acquirer", + "assume", + "substitute", + "awards", + "provided", + "holder", + "continuous", + "service", + "terminated", + "prior", + "If", + "director", + "terminates", + "due", + "death", + "RSU", + "grants", + "immediately", + "vest", + "full", + "Directors", + "receive", + "dividends", + "unvested", + "vested", + "deferred", + "RSUs", + "Director", + "Compensation", + "Fiscal", + "2026", + "Amounts", + "shown", + "reflect", + "amounts", + "received", + "Instead", + "aggregate", + "grant", + "date", + "fair", + "calculated", + "accordance", + "ASC", + "718", + "granted", + "fiscal", + "year", + "The", + "assumptions", + "calculation", + "set", + "Note", + "consolidated", + "financial", + "statements", + "titled", + "Stock", + "Based", + "Form", + "10", + "On", + "June", + "26", + "2025", + "employee", + "serving", + "Program", + "799", + "shares", + "share", + "154.98", + "Mr", + "Burgess", + "passed", + "December", + "final", + "quarterly", + "cash", + "retainer", + "payment", + "Dr", + "Ochoa", + "resigned", + "July", + "dates", + "payments", + "table", + "information", + "number", + "held", + "January", + "25", + "Pursuant", + "terms", + "2007", + "Plan", + "accelerated", + "fully", + "Drell", + "Includes", + "initial", + "Ms", + "Lora", + "connection", + "appointment", + "2023", + "compensation", + "program", + "previously", + "issued", + "resignation" + ], + "keywords": [ + "Director Compensation", + "RSUs", + "Fiscal 2026" + ], + "connect_to": [ + { + "target": "a03b84fc-cfd5-535f-bef5-f3a63414b818", + "relation": "embeds", + "ref": "[tables/table-25 Director Compensation.html]", + "position": { + "start": 666, + "end": 710 + } + }, + { + "target": "370c36ac-7471-52fc-b136-1ddbe702c5ac", + "relation": "embeds", + "ref": "[tables/table-26 RSU Holdings Table.html]", + "position": { + "start": 1723, + "end": 1764 + } + } + ] + } + }, + { + "chunk_id": "47385369-243e-5ab8-9a74-38bec54e1e8d", + "type": "text", + "content": "Employees, officers, and directors must avoid any activity that conflicts with, or has the appearance of conflicting with, our interests. This policy is included in our Code of Conduct and our Financial Team Code of Conduct. We regularly conduct a review of all related party transactions for potential conflicts of interest and all transactions involving executive officers or directors must be approved by the NCGC in compliance with the Company's policies and Nasdaq listing standards. Except as discussed below, there were no transactions with related persons since the beginning of Fiscal 2026 or currently proposed transactions that would require disclosure in this proxy statement or approval by the NCGC.\nTransactions with Related Persons\nThe daughter and son of Jen-Hsun Huang, our President and CEO and a member of our Board, are employed by the Company. Neither of them shares a household with Mr. Huang, is one of our executive officers, or reports directly to Mr. Huang. Additionally, the son of Dr. Shah, one of our directors, is employed by the Company. He does not share a household with Dr. Shah and is not one of our executive officers.\nThe compensation of these individuals was determined in accordance with NVIDIA's compensation practices applicable to employees with comparable qualifications and responsibilities and holding similar positions and without the involvement of Mr. Huang or Dr. Shah, respectively. The total compensation for Fiscal 2026 of the daughter and son of Mr. Huang was approximately \\$1,232,000 and \\$1,320,000, respectively. The total compensation for Fiscal 2026 of the son of Dr. Shah was approximately \\$265,000.\nEach of them has received and continues to be eligible for equity awards on the same general terms and conditions as applicable to employees in similar positions who do not have such family relationships.\nThe Jen-Hsun and Lori Huang Foundation, or the Huang Foundation, a charitable organization, has entered into an agreement with CoreWeave, Inc., a third-party cloud services provider, to purchase GPU compute time for donation to university and other non-profit research institutes to develop open science and AI research, of which \\$108.3 million has been donated to date. NVIDIA routinely offers engineering services for no charge in the ordinary course of business to developers of NVIDIA accelerated computing and we plan to offer similar engineering services for no charge to certain recipients of GPU compute time grants from the Huang Foundation.\nWe have entered into indemnity agreements with our executive officers and directors. The agreements provide that we will indemnify them, under certain circumstances, for liabilities they may be required to pay in actions or proceedings by reason of their position with NVIDIA, and otherwise to the fullest extent permitted under Delaware law, our Charter, and our Bylaws. We intend to execute similar agreements with our future executive officers and directors. See Employment, Severance, and Change-in-Control Arrangements below for a description of the terms of the 2007 Plan related to a change-in-control of NVIDIA.\nDuring Fiscal 2026, we granted RSUs to our non-employee directors, and RSUs and PSUs to our executive officers (other than Mr. Huang, who received PSUs only). See Director Compensation above and Executive Compensation below.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Review of Transactions with Related Persons", + "metadata": { + "length": 3362, + "summary": "The passage outlines NVIDIA's policy on conflicts of interest, requiring employees, officers, and directors to avoid activities conflicting with company interests. It notes that related party transactions are reviewed, and those involving executive officers or directors must be approved by the NCGC. Since Fiscal 2026, no reportable transactions occurred except for the employment of family members of CEO Jen-Hsun Huang and director Dr. Shah. Their compensation was determined per standard practices without involvement of the related executives, totaling approximately $1,232,000 for Huang's daughter, $1,320,000 for his son, and $265,000 for Shah's son. The Huang Foundation has an agreement with CoreWeave for GPU compute time donations, with $108.3 million donated, and NVIDIA plans to offer free engineering services to grant recipients. Indemnity agreements exist for executives and directors, and equity grants were made to non-employee directors and executive officers.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Employees", + "officers", + "directors", + "avoid", + "activity", + "conflicts", + "appearance", + "conflicting", + "interests", + "This", + "policy", + "included", + "Code", + "Conduct", + "Financial", + "Team", + "We", + "regularly", + "conduct", + "review", + "related", + "party", + "transactions", + "potential", + "interest", + "involving", + "executive", + "approved", + "NCGC", + "compliance", + "Company", + "policies", + "Nasdaq", + "listing", + "standards", + "Except", + "discussed", + "persons", + "beginning", + "Fiscal", + "2026", + "proposed", + "require", + "disclosure", + "proxy", + "statement", + "approval", + "Transactions", + "Related", + "Persons", + "The", + "daughter", + "son", + "Jen", + "Hsun", + "Huang", + "President", + "CEO", + "member", + "Board", + "employed", + "Neither", + "shares", + "household", + "Mr", + "reports", + "directly", + "Additionally", + "Dr", + "Shah", + "He", + "share", + "compensation", + "individuals", + "determined", + "accordance", + "NVIDIA", + "practices", + "applicable", + "employees", + "comparable", + "qualifications", + "responsibilities", + "holding", + "similar", + "positions", + "involvement", + "total", + "approximately", + "232", + "000", + "320", + "265", + "Each", + "received", + "continues", + "eligible", + "equity", + "awards", + "general", + "terms", + "conditions", + "family", + "relationships", + "Lori", + "Foundation", + "charitable", + "organization", + "entered", + "agreement", + "CoreWeave", + "Inc", + "cloud", + "services", + "provider", + "purchase", + "GPU", + "compute", + "time", + "donation", + "university", + "profit", + "research", + "institutes", + "develop", + "open", + "science", + "AI", + "108.3", + "million", + "donated", + "date", + "routinely", + "offers", + "engineering", + "charge", + "ordinary", + "business", + "developers", + "accelerated", + "computing", + "plan", + "offer", + "recipients", + "grants", + "indemnity", + "agreements", + "provide", + "indemnify", + "circumstances", + "liabilities", + "required", + "pay", + "actions", + "proceedings", + "reason", + "position", + "fullest", + "extent", + "permitted", + "Delaware", + "law", + "Charter", + "Bylaws", + "intend", + "execute", + "future", + "See", + "Employment", + "Severance", + "Change", + "Control", + "Arrangements", + "description", + "2007", + "Plan", + "change", + "control", + "During", + "granted", + "RSUs", + "employee", + "PSUs", + "Director", + "Compensation", + "Executive" + ], + "keywords": [ + "conflict of interest", + "related party", + "compensation" + ], + "connect_to": [] + } + }, + { + "chunk_id": "5826a3b1-0ecf-5341-bc59-02685094c39e", + "type": "table", + "content": "
Name of Beneficial OwnerShares OwnedShares Issuable Within 60 DaysTotal Shares Beneficially OwnedPercent
NEOs:
Jen-Hsun Huang870,604,104 (1)870,604,1043.58%
Colette M. Kress4,443,523 (2)4,443,523*
Ajay K. Puri3,266,897 (3)3,266,897*
Debora Shoquist1,580,729 (4)1,580,729*
Timothy S. Teter2,687,197 (5)2,687,197*
Directors, not including Mr. Huang:
Tench Coxe30,555,240 (6)30,555,240*
John O. Dabiri13,88890014,788*
Dawn Hudson367,987900368,887*
Harvey C. Jones7,002,787 (7)9007,003,687*
Melissa B. Lora899 (8)9001,799*
Stephen C. Neal183,967 (9)900184,867*
A. Brooke Seawell2,505,707 (10)9002,506,607*
Aarti Shah25,620 (11)25,620*
Mark A. Stevens34,068,672 (12)90034,069,572*
Directors and executive officers as a group (14 persons)957,307,217 (13)6,300957,313,5173.94%
5% Stockholders:
BlackRock, Inc.1,805,935,550 (14)1,805,935,5507.43%
Vanguard Capital Management1,777,408,252 (15)1,777,408,2527.31%
", + "path": "tables/table-27 Beneficial Ownership.html", + "metadata": { + "length": 1968, + "summary": "table-28\nTable lists beneficial ownership of shares for NEOs, directors, and 5% stockholders, including Jen-Hsun Huang (3.58%), BlackRock (7.43%), and Vanguard (7.31%).", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-27 Beneficial Ownership.html", + "keywords": [ + "Beneficial Owners", + "Shares", + "Percent" + ], + "tokens": [] + } + }, + { + "chunk_id": "8a28c61d-cdb3-5f24-9fb1-14974a5fec14", + "type": "text", + "content": "The following table sets forth information as of March 23, 2026 as to shares of our common stock beneficially owned by each of our NEOs, each of our directors, all of our directors and executive officers as a group, and all known by us to be beneficial owners of more than $5\\%$ of our common stock, unless otherwise indicated in the footnotes to the table. Beneficial ownership is determined in accordance with the SEC's rules and generally includes voting or investment power with respect to securities as well as shares of common stock subject to PSUs or RSUs that will vest within 60 days of March 23, 2026.\nThis table is based upon information provided to us by our executive officers and directors. Information about principal stockholders, other than percentages of beneficial ownership, is based solely on Schedules 13G/A filed with the SEC. Unless otherwise indicated and subject to community property laws where applicable, we believe that each of the stockholders named in the table has sole voting and investment power with respect to the shares indicated as beneficially owned. Percentages are based on 24,312,141,810 shares of our common stock outstanding as of March 23, 2026, adjusted as required by SEC rules.\n\n[tables/table-27 Beneficial Ownership.html]\n\n\\* Represents less than 1% of the outstanding shares of our common stock.\n(1) Includes (a) 528,531,547 shares of common stock held by Jen-Hsun Huang and Lori Huang, as co-trustees of the Jen-Hsun and Lori Huang Living Trust, u/a/d May 1, 1995, or the Huang Trust, (b) 31,421,011 shares of common stock held by The Huang 2012 Irrevocable Trust, of which Mr. Huang and his wife are co-trustees, (c) 109,040,602 shares of common stock held by The Huang Irrevocable Remainder Trust u/a/d 2/19/2016, of which Mr. Huang and his wife are co-trustees, (d) 6,632,667 shares of common stock held by TARG M LLC, of which the Huang Trust is the sole member, (e) 6,632,667 shares of common stock held by TARG S LLC, of which the Huang Trust is the sole member, (f) 30,000,000 shares of common stock held by TARG M2 LLC, of which the Huang Trust is the sole member, (g) 30,000,000 shares of common stock held by TARG S2 LLC, of which the Huang Trust is the sole member, and (h) 58,683,830 shares of common stock held by The Jen-Hsun & Lori Huang Foundation, or the Huang Foundation, of which Mr. Huang and his wife are board members. By virtue of their status as co-trustees of the Huang Trust, The Huang 2012 Irrevocable Trust, and The Huang Irrevocable Remainder Trust, each of Mr. Huang and his wife may be deemed to have shared beneficial ownership of, and to exercise shared voting and investment power over, the shares referenced in (a)-(g). By virtue of their status as board members of the Huang Foundation since 2007, Mr. Huang and his wife may be deemed to have shared beneficial ownership of, and to exercise shared voting and investment power over, the shares referenced in (h), and therefore the Huang Foundation's shares are being reported in accordance with Item 403 of Regulation S-K. Mr. Huang and his wife have no pecuniary interest in the Huang Foundation's shares.\n(2) Includes (a) 4,000 shares held by immediate family member 1, (b) 4,000 shares held by immediate family member 2, (c) 712,934 shares held by a limited liability company, the sole member of which is an irrevocable trust of which the trustee is an independent institution, (d) 451,498 shares held by Grantor Retained Annuity Trust 1, of which Ms. Kress is trustee, (e) 451,498 shares held by Grantor Retained Annuity Trust 2, of which Ms. Kress's husband is trustee and exercises sole voting and investment power, (f) 83,060 shares held by a trust, of which Ms. Kress is trustee, (g) 2,032,048 shares held by Grantor Retained Annuity Trust 3, of which Ms. Kress and her husband are co-trustees, and (h) 271,592 shares held by Grantor Retained Annuity Trust 4, of which Ms. Kress and her husband are co-trustees. By virtue of their status as co-trustees of the Grantor Retained Annuity Trusts 3\nand 4, each of Ms. Kress and her husband may be deemed to have shared beneficial ownership of, and to exercise shared voting and investment power over, the shares referenced in (g) and (h).\n(3) Includes (a) 3,018,547 shares of common stock held by the Ajay K Puri Revocable Trust dtd 12/10/2015, of which Mr. Puri is trustee, and (b) 46,360 shares of common stock held by The Puri 2019 Irrevocable Children's Trust dtd 12/06/2019, of which Mr. Puri is one of the trustees. Mr. Puri disclaims beneficial ownership of the shares held by The Puri 2019 Irrevocable Children's Trust, except to the extent of his pecuniary interest therein.\n(4) Includes 1,424,603 shares of common stock held by the Debora C. Shoquist Revocable Living Trust dtd 6/13/2002, of which Ms. Shoquist is trustee.\n(5) Represents shares of common stock held by the Horne Teter Family Living Trust, dated February 1, 2019, of which Mr. Teter is a co-trustee and exercises shared voting and investment power.\n(6) Includes (a) 4,852,480 shares of common stock held in a retirement trust, and (b) 25,671,360 shares of common stock held in The Coxe Revocable Trust, of which Mr. Coxe is a co-trustee and exercises shared voting and investment power. Mr. Coxe disclaims beneficial ownership of the shares held by The Coxe Revocable Trust, except to the extent of his pecuniary interest therein. Mr. Coxe shares pecuniary interest in shares held in his individual name pursuant to a contractual relationship, and disclaims beneficial ownership of these shares, except to the extent of his pecuniary interest therein.\nDoes not include an additional 23,867 shares of common stock underlying vested RSUs that Mr. Coxe has deferred for future issuance.\n(7) Includes 6,933,280 shares of common stock held in the H.C. Jones Living Trust, of which Mr. Jones is trustee.\n(8) Does not include an additional 12,878 shares of common stock underlying vested RSUs that Ms. Lora has deferred for future issuance.\n(9) Includes (a) 30,428 shares of shares of common stock held by the 2013 Stephen C. Neal Revocable Trust, of which Mr. Neal is trustee, (b) 143,410 shares of common stock held by the Neal/Rhyu Revocable Trust dated 05/02/2017, of which Mr. Neal is a co-trustee and exercises shared voting and investment power, and (c) 7,142 shares of common stock held by the 2013 Michelle S. Rhyu Revocable Trust, of which Mr. Neal's wife is trustee and exercises sole voting and investment power.\n(10) Includes 2,500,000 shares of common stock held by The Alexander Brooke Seawell Revocable Trust U/A dated 1/20/2009, of which Mr. Seawell is trustee.\n(11) Does not include an additional 9,487 shares of common stock underlying vested RSUs that Dr. Shah has deferred for future issuance.\n(12) Includes (a) 7,399,771 shares of common stock held by the 3rd Millennium Trust, of which Mr. Stevens is co-trustee and exercises shared voting and investment power, (b) 15,017,750 shares of common stock held by the Envy Trust u/a/d December 7, 2021, of which Mr. Stevens is trustee, and (c) 108,650 shares of common stock held by the 970 Foundation, of which Mr. Stevens and his wife are board members. By virtue of their status as board members of the 970 Foundation, Mr. Stevens and his wife may be deemed to have shared beneficial ownership of, and to exercise shared voting and investment power over, the shares referenced in (c), and therefore the 970 Foundation's shares are being reported in accordance with Item 403 of Regulation S-K. Mr. Stevens and his wife have no pecuniary interest in the 970 Foundation's shares.\n(13) Includes shares owned by all directors and executive officers.\n(14) This information is based solely on a Schedule 13G/A, dated January 26, 2024, filed with the SEC on January 26, 2024 by BlackRock, Inc. reporting its beneficial ownership as of December 31, 2023. On an adjusted basis to reflect our June 2024 ten-for-one stock split, the Schedule 13G/A reported that BlackRock had sole voting power with respect to 1,628,565,130 shares and sole dispositive power with respect to 1,805,935,550 shares. BlackRock is located at 50 Hudson Yards, New York, New York 10001.\n(15) This information is based solely on a Schedule 13G, dated April 28, 2026, filed with the SEC on April 28, 2026 by Vanguard Capital Management reporting its beneficial ownership as of March 31, 2026. The Schedule 13G reported that Vanguard had sole voting power with respect to 238,857,870 shares and sole dispositive power with respect to 1,777,408,252 shares. Vanguard is located at 100 Vanguard Boulevard, Malvern, Pennsylvania 19355.", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Security Ownership of Certain Beneficial Owners and Management", + "metadata": { + "length": 8681, + "summary": "The table details beneficial ownership of common stock as of March 23, 2026, for NEOs, directors, all directors and executive officers as a group, and known beneficial owners of over 5%. Ownership is determined per SEC rules, including shares with voting/investment power and PSUs/RSUs vesting within 60 days. Percentages are based on 24,312,141,810 shares outstanding. Footnotes provide specifics for each individual, including shares held in trusts, LLCs, and foundations, with details on voting and investment power. Principal stockholders BlackRock and Vanguard are listed based on Schedule 13G filings, with BlackRock having sole dispositive power over 1,805,935,550 shares and Vanguard over 1,777,408,252 shares.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "table", + "sets", + "information", + "March", + "23", + "2026", + "shares", + "common", + "stock", + "beneficially", + "owned", + "NEOs", + "directors", + "executive", + "officers", + "group", + "beneficial", + "owners", + "footnotes", + "Beneficial", + "ownership", + "determined", + "accordance", + "SEC", + "rules", + "generally", + "includes", + "voting", + "investment", + "power", + "respect", + "securities", + "subject", + "PSUs", + "RSUs", + "vest", + "60", + "days", + "This", + "based", + "provided", + "Information", + "principal", + "stockholders", + "percentages", + "solely", + "Schedules", + "13G", + "filed", + "Unless", + "community", + "property", + "laws", + "applicable", + "named", + "sole", + "Percentages", + "24", + "312", + "141", + "810", + "outstanding", + "adjusted", + "required", + "Represents", + "1%", + "Includes", + "528", + "531", + "547", + "held", + "Jen", + "Hsun", + "Huang", + "Lori", + "trustees", + "Living", + "Trust", + "May", + "1995", + "31", + "421", + "011", + "2012", + "Irrevocable", + "Mr", + "wife", + "109", + "040", + "602", + "Remainder", + "19", + "2016", + "632", + "667", + "TARG", + "LLC", + "member", + "30", + "000", + "M2", + "S2", + "58", + "683", + "830", + "Foundation", + "board", + "members", + "By", + "virtue", + "status", + "deemed", + "shared", + "exercise", + "referenced", + "2007", + "reported", + "Item", + "403", + "Regulation", + "pecuniary", + "interest", + "family", + "712", + "934", + "limited", + "liability", + "company", + "irrevocable", + "trust", + "trustee", + "independent", + "institution", + "451", + "498", + "Grantor", + "Retained", + "Annuity", + "Ms", + "Kress", + "husband", + "exercises", + "83", + "060", + "032", + "048", + "271", + "592", + "Trusts", + "018", + "Ajay", + "Puri", + "Revocable", + "dtd", + "12", + "10", + "2015", + "46", + "360", + "2019", + "Children", + "06", + "disclaims", + "extent", + "424", + "603", + "Debora", + "Shoquist", + "13", + "2002", + "Horne", + "Teter", + "Family", + "dated", + "February", + "852", + "480", + "retirement", + "25", + "671", + "Coxe", + "individual", + "pursuant", + "contractual", + "relationship", + "Does", + "include", + "additional", + "867", + "underlying", + "vested", + "deferred", + "future", + "issuance", + "933", + "280", + "Jones", + "878", + "Lora", + "428", + "2013", + "Stephen", + "Neal", + "143", + "410", + "Rhyu", + "05", + "02", + "2017", + "142", + "Michelle", + "500", + "Alexander", + "Brooke", + "Seawell", + "20", + "2009", + "11", + "487", + "Dr", + "Shah", + "399", + "771", + "3rd", + "Millennium", + "Stevens", + "15", + "017", + "750", + "Envy", + "December", + "2021", + "108", + "650", + "970", + "14", + "Schedule", + "January", + "26", + "2024", + "BlackRock", + "Inc", + "reporting", + "2023", + "On", + "basis", + "reflect", + "June", + "ten", + "split", + "628", + "565", + "130", + "dispositive", + "805", + "935", + "550", + "located", + "50", + "Hudson", + "Yards", + "New", + "York", + "10001", + "April", + "28", + "Vanguard", + "Capital", + "Management", + "238", + "857", + "870", + "777", + "408", + "252", + "100", + "Boulevard", + "Malvern", + "Pennsylvania", + "19355" + ], + "keywords": [ + "beneficial ownership", + "common stock", + "NEOs" + ], + "connect_to": [ + { + "target": "5826a3b1-0ecf-5341-bc59-02685094c39e", + "relation": "embeds", + "ref": "[tables/table-27 Beneficial Ownership.html]", + "position": { + "start": 1228, + "end": 1271 + } + } + ] + } + }, + { + "chunk_id": "7246ddc8-f599-5668-8f40-9a80355ca5ab", + "type": "text", + "content": "What am I voting on? A non-binding vote, known as “say-on-pay,” to approve our Fiscal 2026 NEO compensation.\nVote required for approval: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter.\nEffect of abstentions: Same as a vote AGAINST.\nEffect of broker non-votes: None.\nIn accordance with Section 14A of the Exchange Act, we are asking our stockholders to vote on an advisory basis, commonly referred to as “say-on-pay,” to approve the Fiscal 2026 compensation paid to our NEOs as disclosed in the CD&A, the compensation tables and the accompanying narrative discussion. This vote is intended to address the overall compensation of our NEOs and the philosophy, policies, and practices described in this proxy statement, rather than any specific compensation component.\nIn response to our stockholders' preference, our Board has adopted a policy of providing for annual “say-on-pay” votes.\nThis advisory proposal is not binding on the Board nor us. Nevertheless, the views expressed by our stockholders are important to the Board and, accordingly, the Board and the CC intend to consider the results of this vote in making future NEO compensation decisions.\nRecommendation of the Board\nThe Board recommends that our stockholders adopt the following resolution:\n\"RESOLVED, that the Fiscal 2026 compensation paid to the Company's named executive officers, as disclosed pursuant to Item 402 of Regulation S-K, including the Compensation Discussion and Analysis, compensation tables, and narrative discussion, is hereby APPROVED.\"", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Proposal 2—Advisory Approval of Executive Compensation", + "metadata": { + "length": 1579, + "summary": "This is a non-binding advisory vote, known as 'say-on-pay,' to approve Fiscal 2026 compensation for named executive officers (NEOs). Approval requires a majority of shares present and voting; abstentions count as votes against, while broker non-votes have no effect. The board recommends approval, and while the vote is non-binding, it will be considered in future compensation decisions.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "What", + "voting", + "binding", + "vote", + "pay", + "approve", + "Fiscal", + "2026", + "NEO", + "compensation", + "Vote", + "required", + "approval", + "majority", + "shares", + "present", + "person", + "represented", + "proxy", + "entitled", + "matter", + "Effect", + "abstentions", + "Same", + "AGAINST", + "broker", + "votes", + "None", + "In", + "accordance", + "Section", + "14A", + "Exchange", + "Act", + "stockholders", + "advisory", + "basis", + "commonly", + "referred", + "paid", + "NEOs", + "disclosed", + "CD", + "tables", + "accompanying", + "narrative", + "discussion", + "This", + "intended", + "address", + "philosophy", + "policies", + "practices", + "statement", + "specific", + "component", + "response", + "preference", + "Board", + "adopted", + "policy", + "providing", + "annual", + "proposal", + "Nevertheless", + "views", + "expressed", + "important", + "CC", + "intend", + "results", + "making", + "future", + "decisions", + "Recommendation", + "The", + "recommends", + "adopt", + "resolution", + "RESOLVED", + "Company", + "named", + "executive", + "officers", + "pursuant", + "Item", + "402", + "Regulation", + "including", + "Compensation", + "Discussion", + "Analysis", + "APPROVED" + ], + "keywords": [ + "say-on-pay", + "NEO compensation", + "advisory vote" + ], + "connect_to": [] + } + }, + { + "chunk_id": "1c2227b1-98a0-5871-beb0-155f95a4809f", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6268fbae-e01c-519c-ba0a-62453be5b4fa", + "type": "text", + "content": "This CD&A describes our Fiscal 2026 executive compensation philosophy, design, and process, and how our corporate results affected the payout of performance-based awards. Our Fiscal 2026 NEOs were: Jen-Hsun Huang, President and CEO; Colette M. Kress, EVP and CFO; Ajay K. Puri, EVP, Worldwide Field Operations; Debora Shoquist, EVP, Operations; and Timothy S. Teter, EVP, General Counsel and Secretary.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis", + "metadata": { + "length": 402, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "This", + "CD", + "describes", + "Fiscal", + "2026", + "executive", + "compensation", + "philosophy", + "design", + "process", + "corporate", + "results", + "affected", + "payout", + "performance", + "based", + "awards", + "Our", + "NEOs", + "Jen", + "Hsun", + "Huang", + "President", + "CEO", + "Colette", + "Kress", + "EVP", + "CFO", + "Ajay", + "Puri", + "Worldwide", + "Field", + "Operations", + "Debora", + "Shoquist", + "Timothy", + "Teter", + "General", + "Counsel", + "Secretary" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "344281ed-ec77-597d-8c44-2f57073bdb07", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Executive Compensation Summary", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "306eb34f-d9f2-5029-a13f-6e1be8a8249a", + "type": "text", + "content": "NVIDIA's executive compensation program in Fiscal 2026 continued to be guided by a pay for performance philosophy to link competitive NEO pay with our stockholders' interests. $96\\%$ of our CEO's and $48\\%$ of our other NEOs' Fiscal 2026 total target pay was dependent on corporate performance.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Executive Compensation Summary/Continued Focus on Pay for Performance", + "metadata": { + "length": 294, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "NVIDIA", + "executive", + "compensation", + "program", + "Fiscal", + "2026", + "continued", + "guided", + "pay", + "performance", + "philosophy", + "link", + "competitive", + "NEO", + "stockholders", + "interests", + "96", + "CEO", + "48", + "NEOs", + "total", + "target", + "dependent", + "corporate" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0aa91e8d-1793-532f-8cae-788cc13b014e", + "type": "image", + "content": "\nThe vast majority of our NEOs' total target pay for Fiscal 2026 was comprised of equity awards:
0\n[images/image-73-Compensation Structure Breakdown.jpg]\n", + "path": "images/image-73-Compensation Structure Breakdown.jpg", + "metadata": { + "length": 141, + "summary": "image-73\n- Chart Type: Stacked Bar Chart\n- Components (from left to right):\n - SY PSUs: 44%\n - MY PSUs: 44%\n - Variable Cash: 8%\n - Base Salary: 4%", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-73-Compensation Structure Breakdown.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "794db717-db9d-5bf9-a157-26af722e35ca", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
CategoryPercentage (%)
SY PSUs44
MY PSUs44
Variable Cash8
Base Salary4
", + "path": "tables/table-28 Compensation Breakdown.html", + "metadata": { + "length": 435, + "summary": "table-29\nThe table shows compensation components: SY PSUs (44%), MY PSUs (44%), Variable Cash (8%), and Base Salary (4%).", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-28 Compensation Breakdown.html", + "keywords": [ + "PSUs", + "Variable Cash", + "Base Salary" + ], + "tokens": [] + } + }, + { + "chunk_id": "14d5d939-937e-5bab-ad92-fa2162728b76", + "type": "image", + "content": "\n
0\n[images/image-74-Compensation Structure Breakdown.jpg]\n", + "path": "images/image-74-Compensation Structure Breakdown.jpg", + "metadata": { + "length": 56, + "summary": "image-74\n- Chart Title: Compensation Composition (SY PSUs, MY PSUs, RSUs, Variable Cash, Base Salary)\n- Data Source/Context: Likely represents a total compensation mix for an employee or role.\n- Component Breakdown:\n - SY PSUs (Short-term Performance Units): 23%\n - MY PSUs (Medium-term Performance Units): 23%\n - RSUs (Restricted Stock Units): 46%\n - Variable Cash: 2%\n - Base Salary: 6%", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-74-Compensation Structure Breakdown.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "91e6c01e-ebd3-51f0-b90c-cf21aa23bb06", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
CategoryPercentage (%)
SY PSUs23
MY PSUs23
RSUs46
Variable Cash6
Base Salary2
", + "path": "tables/table-29 Compensation Mix.html", + "metadata": { + "length": 492, + "summary": "table-30\nThe table shows a compensation breakdown: SY PSUs 23%, MY PSUs 23%, RSUs 46%, Variable Cash 6%, Base Salary 2%.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-29 Compensation Mix.html", + "keywords": [ + "PSUs", + "RSUs", + "Variable Cash" + ], + "tokens": [] + } + }, + { + "chunk_id": "f2df63cb-74f3-58a2-a5aa-e80fc65793fc", + "type": "text", + "content": "The vast majority of our NEOs' total target pay for Fiscal 2026 was comprised of equity awards:\n• SY PSUs that can be earned based on annual Non-GAAP Operating Income performance, vesting over 4 years,\n• MY PSUs that can be earned based on 3-year TSR performance relative to the S&P 500, vesting over 3 years, and\n• RSUs vesting over 4 years (for NEOs other than our CEO)\nCEO Fiscal 2026 Target Pay Mix (1)\n\nThe vast majority of our NEOs' total target pay for Fiscal 2026 was comprised of equity awards:
0\n[images/image-73-Compensation Structure Breakdown.jpg]\n\n
\nstacked bar chart\n\n[tables/table-28 Compensation Breakdown.html]\n\n
\nOther NEO Fiscal 2026 Target Pay Mix (1)(2)\n\n
0\n[images/image-74-Compensation Structure Breakdown.jpg]\n\n
\nstacked bar chart\n\n[tables/table-29 Compensation Mix.html]\n\n
\n(1) Based on total target pay as approved by the CC, consisting of annual base salary, and, assuming the Company achieves associated performance goals at a Base Compensation Plan level, target payout opportunity under our Variable Cash Plan and target equity opportunities the CC intended to deliver.\n(2) Reflects the average total target pay mix for our NEOs other than our CEO.\nNEOs were also eligible for variable cash awards based on annual revenue, in addition to base salary and other compensation and benefits.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Executive Compensation Summary/Executive Pay Heavily Weighted Towards Equity Awards", + "metadata": { + "length": 1381, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "vast", + "majority", + "NEOs", + "total", + "target", + "pay", + "Fiscal", + "2026", + "comprised", + "equity", + "awards", + "SY", + "PSUs", + "earned", + "based", + "annual", + "Non", + "GAAP", + "Operating", + "Income", + "performance", + "vesting", + "years", + "MY", + "year", + "TSR", + "relative", + "500", + "RSUs", + "CEO", + "Target", + "Pay", + "Mix", + "details", + "summary", + "stacked", + "bar", + "chart", + "Other", + "NEO", + "Based", + "approved", + "CC", + "consisting", + "base", + "salary", + "assuming", + "Company", + "achieves", + "goals", + "Base", + "Compensation", + "Plan", + "level", + "payout", + "opportunity", + "Variable", + "Cash", + "opportunities", + "intended", + "deliver", + "Reflects", + "average", + "mix", + "eligible", + "variable", + "cash", + "revenue", + "addition", + "compensation", + "benefits" + ], + "keywords": [], + "connect_to": [ + { + "target": "0aa91e8d-1793-532f-8cae-788cc13b014e", + "relation": "embeds", + "ref": "[images/image-73-Compensation Structure Breakdown.jpg]", + "position": { + "start": 515, + "end": 569 + } + }, + { + "target": "794db717-db9d-5bf9-a157-26af722e35ca", + "relation": "embeds", + "ref": "[tables/table-28 Compensation Breakdown.html]", + "position": { + "start": 619, + "end": 664 + } + }, + { + "target": "14d5d939-937e-5bab-ad92-fa2162728b76", + "relation": "embeds", + "ref": "[images/image-74-Compensation Structure Breakdown.jpg]", + "position": { + "start": 744, + "end": 798 + } + }, + { + "target": "91e6c01e-ebd3-51f0-b90c-cf21aa23bb06", + "relation": "embeds", + "ref": "[tables/table-29 Compensation Mix.html]", + "position": { + "start": 848, + "end": 887 + } + } + ] + } + }, + { + "chunk_id": "3b742596-859c-5dc6-a594-e5f223dfcbb0", + "type": "table", + "content": "
Fiscal 2026 RevenueFiscal 2026 Non-GAAP Operating Income (1)3-Year TSR Relative to S&P 500 (Fiscal 2024 to 2026) (2)
Stretch Compensation Plan Goal$160.0 billion (3)$96.0 billion (3)75th percentile
Performance Achievement for Period Ended Fiscal 2026$215.9 billion$137.3 billion100th Percentile (1,055%)
", + "path": "tables/table-30 Fiscal 2026 Goals.html", + "metadata": { + "length": 455, + "summary": "table-31\nTable shows stretch compensation plan goals vs actual performance for fiscal 2026: revenue $160B goal vs $215.9B actual, non-GAAP operating income $96B goal vs $137.3B actual, TSR 75th percentile goal vs 100th percentile actual.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-30 Fiscal 2026 Goals.html", + "keywords": [ + "Revenue", + "Operating Income", + "TSR" + ], + "tokens": [] + } + }, + { + "chunk_id": "e554715d-7946-59ae-955a-d9eac0966909", + "type": "text", + "content": "- Cash Compensation Remained Constant: no changes compared to Fiscal 2025 base salary or variable cash opportunity\n- Increased Target Equity Opportunities: increased target equity opportunities for all NEOs to recognize the complexity and scope of NEOs' roles and responsibilities; by increasing target equity opportunities, aligned total target pay for all NEOs other than our CEO to reflect their comparable, significant contributions and impact on corporate performance\n- Continued Increased Rigor of Performance Goals: set the Base Compensation Plan revenue goal for the Fiscal 2026 Variable Cash Plan in line with our record Fiscal 2025 results and the Base Compensation Plan Non-GAAP Operating Income goal for SY PSUs granted in Fiscal 2026 in line with our corresponding Fiscal 2025 Stretch Compensation Plan goal\nRecord Performance Resulting in Maximum Payouts\n\n[tables/table-30 Fiscal 2026 Goals.html]\n\n(1) See Reconciliation of Non-GAAP Financial Measures below for a reconciliation between the non-GAAP financial measures and GAAP results.\n(2) Represents TSR for purposes of the MY PSU performance goal, calculated using cumulative stock price appreciation with dividends reinvested and the average closing stock price for the 60 trading days preceding the start, and preceding and including the last day, of the three-year performance period.\n(3) Our original Fiscal 2026 Stretch Compensation Plan revenue goal of \\$190.0 billion was automatically adjusted to \\$160.0 billion, and our original Fiscal 2026 Stretch Compensation Plan Non-GAAP Operating Income goal of \\$120.0 billion was automatically adjusted to \\$96.0 billion, due to the imposition of additional export controls on the Company's H2O products during the first half of Fiscal 2026. See Fiscal 2026 Compensation Actions and Achievements —Performance Metrics and Goals for Executive Compensation below for further discussion of these adjustments.\nAs a result of the above performance achievements, each exceeding the CC's pre-established Stretch Compensation Plan goals, our NEOs earned the maximum payouts possible for our Fiscal 2026 Variable Cash Plan, SY PSUs granted in Fiscal 2026 (which remain subject to service-based vesting), and MY PSUs granted in Fiscal 2024.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Key NEO Compensation Actions for Fiscal 2026", + "metadata": { + "length": 2297, + "summary": "Cash compensation for NEOs remained constant, while target equity opportunities increased to reflect role complexity and align total pay. Performance goals were raised, with revenue and Non-GAAP Operating Income targets set at record or stretch levels. Due to record performance exceeding stretch goals, NEOs earned maximum payouts for Fiscal 2026 Variable Cash Plan, SY PSUs, and MY PSUs.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Cash", + "Compensation", + "Remained", + "Constant", + "compared", + "Fiscal", + "2025", + "base", + "salary", + "variable", + "cash", + "opportunity", + "Increased", + "Target", + "Equity", + "Opportunities", + "increased", + "target", + "equity", + "opportunities", + "NEOs", + "recognize", + "complexity", + "scope", + "roles", + "responsibilities", + "increasing", + "aligned", + "total", + "pay", + "CEO", + "reflect", + "comparable", + "significant", + "contributions", + "impact", + "corporate", + "performance", + "Continued", + "Rigor", + "Performance", + "Goals", + "set", + "Base", + "Plan", + "revenue", + "goal", + "2026", + "Variable", + "line", + "record", + "results", + "Non", + "GAAP", + "Operating", + "Income", + "SY", + "PSUs", + "granted", + "Stretch", + "Record", + "Resulting", + "Maximum", + "Payouts", + "See", + "Reconciliation", + "Financial", + "Measures", + "reconciliation", + "financial", + "measures", + "Represents", + "TSR", + "purposes", + "MY", + "PSU", + "calculated", + "cumulative", + "stock", + "price", + "appreciation", + "dividends", + "reinvested", + "average", + "closing", + "60", + "trading", + "days", + "preceding", + "start", + "including", + "day", + "year", + "period", + "Our", + "original", + "190.0", + "billion", + "automatically", + "adjusted", + "160.0", + "120.0", + "96.0", + "due", + "imposition", + "additional", + "export", + "controls", + "Company", + "H2O", + "products", + "half", + "Actions", + "Achievements", + "Metrics", + "Executive", + "discussion", + "adjustments", + "As", + "result", + "achievements", + "exceeding", + "CC", + "pre", + "established", + "goals", + "earned", + "maximum", + "payouts", + "remain", + "subject", + "service", + "based", + "vesting", + "2024" + ], + "keywords": [ + "Compensation", + "Equity", + "Performance" + ], + "connect_to": [ + { + "target": "3b742596-859c-5dc6-a594-e5f223dfcbb0", + "relation": "embeds", + "ref": "[tables/table-30 Fiscal 2026 Goals.html]", + "position": { + "start": 870, + "end": 910 + } + } + ] + } + }, + { + "chunk_id": "c7d099e3-e835-592c-9dcf-86edcb371849", + "type": "text", + "content": "NVIDIA is building a one-of-a-kind company that invents the future, builds amazing technologies, and strives to achieve the highest level of craft. To achieve this vision, we must attract and retain a high-caliber executive team while balancing our stockholders' interests. While our CC considers numerous factors in making executive pay decisions, our compensation program is guided by the following philosophies:\n- Pay for Performance: emphasize at-risk and performance-based cash and equity for NEOs based on multiple corporate metrics\n• Competitive Pay: structure competitive NEO target compensation to reflect job impact, scope, and responsibilities, that attracts and retains talent\n- Stockholder Alignment: align NEO pay with stockholders' long-term interests and consider feedback from our annual stockholder engagement efforts and \"say-on-pay\" vote\n- Simplicity and Transparency: design a compensation program with simple, objective metrics\nIn this CD&A, total target pay refers to (i) an NEO's annual base salary, (ii) target variable cash opportunity, which means the potential payout under our Variable Cash Plan, assuming the Company achieves the associated performance goal at a Base Compensation Plan level, and (iii) target equity opportunity, which means the value of the equity opportunities granted during the year that the CC intended to deliver at grant, assuming the Company achieves associated performance goals at a Base Compensation Plan level.\nOur executive compensation program adheres to the following practices:", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Our Compensation Philosophy and Practices", + "metadata": { + "length": 1540, + "summary": "NVIDIA's compensation program for NEOs is guided by four philosophies: pay for performance using at-risk cash and equity based on corporate metrics, competitive pay to attract and retain talent, alignment with stockholders' long-term interests, and simplicity with objective metrics. Total target pay includes base salary, target variable cash opportunity, and target equity opportunity. The program emphasizes at-risk compensation and stockholder alignment.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "NVIDIA", + "building", + "kind", + "company", + "invents", + "future", + "builds", + "amazing", + "technologies", + "strives", + "achieve", + "highest", + "level", + "craft", + "To", + "vision", + "attract", + "retain", + "high", + "caliber", + "executive", + "team", + "balancing", + "stockholders", + "interests", + "While", + "CC", + "considers", + "numerous", + "factors", + "making", + "pay", + "decisions", + "compensation", + "program", + "guided", + "philosophies", + "Pay", + "Performance", + "emphasize", + "risk", + "performance", + "based", + "cash", + "equity", + "NEOs", + "multiple", + "corporate", + "metrics", + "Competitive", + "structure", + "competitive", + "NEO", + "target", + "reflect", + "job", + "impact", + "scope", + "responsibilities", + "attracts", + "retains", + "talent", + "Stockholder", + "Alignment", + "align", + "long", + "term", + "feedback", + "annual", + "stockholder", + "engagement", + "efforts", + "vote", + "Simplicity", + "Transparency", + "design", + "simple", + "objective", + "In", + "CD", + "total", + "refers", + "base", + "salary", + "ii", + "variable", + "opportunity", + "means", + "potential", + "payout", + "Variable", + "Cash", + "Plan", + "assuming", + "Company", + "achieves", + "goal", + "Base", + "Compensation", + "iii", + "opportunities", + "granted", + "year", + "intended", + "deliver", + "grant", + "goals", + "Our", + "adheres", + "practices" + ], + "keywords": [ + "Pay for Performance", + "Competitive Pay", + "Stockholder Alignment" + ], + "connect_to": [] + } + }, + { + "chunk_id": "982cc649-f55e-5a36-95c7-6248e5cab193", + "type": "text", + "content": "√ Emphasize at-risk, performance-based compensation, with simple, objective performance goals\n√ Use annual and multi-year performance targets to determine PSU awards earned\n√ Set rigorous performance goals and use different metrics for annual and multi-year awards\n√ Require NEOs to provide continuous service for 4 years to fully vest in RSU awards and SY PSU awards earned based on performance\n√ Evaluate our program annually based on feedback from stockholder engagement efforts and make adjustments when appropriate\n√ Mitigate compensation risks\n√ Cap performance-based variable cash and PSU payouts\n√ Retain an independent compensation consultant reporting directly to the independent CC\n√ Require NEOs to maintain meaningful stock ownership\n√ Maintain a clawback policy for performance-based compensation", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Our Compensation Philosophy and Practices/What We Do", + "metadata": { + "length": 810, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Emphasize", + "risk", + "performance", + "based", + "compensation", + "simple", + "objective", + "goals", + "Use", + "annual", + "multi", + "year", + "targets", + "determine", + "PSU", + "awards", + "earned", + "Set", + "rigorous", + "metrics", + "Require", + "NEOs", + "provide", + "continuous", + "service", + "years", + "fully", + "vest", + "RSU", + "SY", + "Evaluate", + "program", + "annually", + "feedback", + "stockholder", + "engagement", + "efforts", + "make", + "adjustments", + "Mitigate", + "risks", + "Cap", + "variable", + "cash", + "payouts", + "Retain", + "independent", + "consultant", + "reporting", + "directly", + "CC", + "maintain", + "meaningful", + "stock", + "ownership", + "Maintain", + "clawback", + "policy" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "bf5d6a47-be43-5553-bcc2-721b0379dfc6", + "type": "image", + "content": "\nX Give NEOs supplemental retirement benefits
0\n[images/image-75-Executive Compensation Timeline.jpg]\n", + "path": "images/image-75-Executive Compensation Timeline.jpg", + "metadata": { + "length": 90, + "summary": "image-75\nDecember 2024: CC determined peer companies\nDecember 2024 - January 2025: Members of management and the Board, including our Lead Director and CC Chair, engaged in stockholder outreach\nMarch 2025: CC determined performance goals and approved target compensation levels and PSU and RSU grants, considering stockholder feedback and peer company practices\nMarch 2026: CC certified achievement and payouts for Fiscal 2026 Variable Cash Plan, SY PSUs granted in Fiscal 2026 and MY PSUs granted in Fiscal 2024*\nMay 2026: CC oversaw compensation risk assessment; published executive compensation program details in proxy statement", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-75-Executive Compensation Timeline.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "7aa8a44b-a5fd-550f-abf3-841c4aa3de18", + "type": "text", + "content": "X Enter into agreements with NEOs providing for specific terms of employment or severance benefits\nX Give our executive officers special change-in-control benefits\nX Provide automatic equity vesting upon a change-in-control (except for the provisions in our equity plans that apply to all employees if an acquiring company does not assume or substitute our outstanding stock awards)\nX Give NEOs supplemental retirement benefits\nX Provide tax gross-ups\nX Pay dividends or the equivalent on unearned or unvested equity\nX Permit executive officers, employees, or directors to hedge their ownership of NVIDIA stock or to pledge NVIDIA stock as collateral for a loan (though we do allow participation in certain exchange funds for portfolio diversification)\nThe CC's oversight and decision-making for our Fiscal 2026 executive compensation program was a multi-year process:\n\nX Give NEOs supplemental retirement benefits
0\n[images/image-75-Executive Compensation Timeline.jpg]\n\n
\nflowchart\n```mermaid\ngraph LR\nA[\"December 2024\\nCC determined peer companies\"] --> B[\"December 2024 - January 2025\\nMembers of management and the Board, including our Lead Director and CC Chair, engaged in stockholder outreach\"]\nB --> C[\"March 2025\\nCC determined performance goals and approved target compensation levels and PSU and RSU grants, considering stockholder feedback and peer company practices\"]\nC --> D[\"March 2026\\nCC certified achievement and payouts for Fiscal 2026 Variable Cash Plan, SY PSUs granted in Fiscal 2026 and MY PSUs granted in Fiscal 2024*\"]\nD --> E[\"May 2026\\nCC oversaw compensation risk assessment; published executive compensation program details in proxy statement\"]\n```\n
\n\\* The CC is expected to certify achievement and payouts for MY PSUs granted in Fiscal 2026 by March 2028.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Our Compensation Philosophy and Practices/What We Don't Do", + "metadata": { + "length": 1811, + "summary": "The document outlines NVIDIA's executive compensation policies, including restrictions on supplemental retirement benefits, tax gross-ups, and hedging/pledging of stock. It also details the Compensation Committee's multi-year process for the Fiscal 2026 program, from peer company determination in December 2024 to certification of payouts and risk assessment by May 2026.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Enter", + "agreements", + "NEOs", + "providing", + "specific", + "terms", + "employment", + "severance", + "benefits", + "Give", + "executive", + "officers", + "special", + "change", + "control", + "Provide", + "automatic", + "equity", + "vesting", + "provisions", + "plans", + "apply", + "employees", + "acquiring", + "company", + "assume", + "substitute", + "outstanding", + "stock", + "awards", + "supplemental", + "retirement", + "tax", + "gross", + "ups", + "Pay", + "dividends", + "equivalent", + "unearned", + "unvested", + "Permit", + "directors", + "hedge", + "ownership", + "NVIDIA", + "pledge", + "collateral", + "loan", + "participation", + "exchange", + "funds", + "portfolio", + "diversification", + "The", + "CC", + "oversight", + "decision", + "making", + "Fiscal", + "2026", + "compensation", + "program", + "multi", + "year", + "process", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "LR", + "December", + "2024", + "nCC", + "determined", + "peer", + "companies", + "January", + "2025", + "nMembers", + "management", + "Board", + "including", + "Lead", + "Director", + "Chair", + "engaged", + "stockholder", + "outreach", + "March", + "performance", + "goals", + "approved", + "target", + "levels", + "PSU", + "RSU", + "grants", + "feedback", + "practices", + "certified", + "achievement", + "payouts", + "Variable", + "Cash", + "Plan", + "SY", + "PSUs", + "granted", + "MY", + "May", + "oversaw", + "risk", + "assessment", + "published", + "proxy", + "statement", + "expected", + "certify", + "2028" + ], + "keywords": [ + "executive compensation", + "change-in-control", + "equity vesting" + ], + "connect_to": [ + { + "target": "bf5d6a47-be43-5553-bcc2-721b0379dfc6", + "relation": "embeds", + "ref": "[images/image-75-Executive Compensation Timeline.jpg]", + "position": { + "start": 926, + "end": 979 + } + } + ] + } + }, + { + "chunk_id": "0f40b77a-274d-57fd-9f01-884fe801c3bc", + "type": "text", + "content": "The roles of (i) our CC, (ii) our independent compensation consultant, and (iii) management, including our CEO and Finance and Human Resources departments, are summarized below.\nFor advice regarding our Fiscal 2026 NEO compensation program, our CC utilized Semler Brossy for its experience working with our CC and with compensation committees at other technology companies. Our CC analyzed whether Semler Brossy's role raised any conflict of interests, taking into consideration the following:\n- Semler Brossy did not provide any services directly to NVIDIA (although we paid Semler Brossy on the CC's behalf);\n- The percentage of Semler Brossy's total revenue resulting from fees paid by us on the CC's behalf;\nSemler Brossy's conflict of interest policies and procedures;\n- Any business or personal relationship between Semler Brossy and an executive officer, or between Semler Brossy's individual compensation advisors and an executive officer or any member of our CC; and\n• Any NVIDIA stock owned by Semler Brossy or its individual compensation advisors.\nAfter considering these factors, our CC determined that Semler Brossy's work did not create any conflict of interests.\nOur CC reviews and approves all NEO compensation decisions, with input from Mr. Huang and Semler Brossy. At the CC's direction, Semler Brossy and management recommended a peer group for our Fiscal 2026 executive pay program, which the CC approved. Semler Brossy provided data on CEO compensation from the peer group companies to inform the CC on Mr. Huang's compensation and management gathered peer data from the Radford Global Technology Survey, or the Radford Survey, to inform Mr. Huang's recommendations for other NEOs' compensation. The CC considered Semler Brossy's advice, Mr. Huang's recommendations, and management's proposed Fiscal 2026 performance goals, before making its final decisions on Fiscal 2026 NEO compensation. The CC certified compensation payouts for performance periods that concluded at the end of Fiscal 2026 under the Variable Cash Plan and for SY PSUs granted in Fiscal 2026 and MY PSUs granted in Fiscal 2024. The CC also oversaw management's Fiscal 2026 compensation risk analysis.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Roles of the Compensation Committee, Compensation Consultant, and Management", + "metadata": { + "length": 2191, + "summary": "The Compensation Committee (CC) oversees NEO compensation, utilizing independent consultant Semler Brossy for advice on the Fiscal 2026 program. The CC confirmed no conflicts of interest after reviewing Semler Brossy's services, revenue percentage, policies, and relationships. The CC approved a peer group recommended by Semler Brossy and management, considered CEO peer data and Radford Survey data for other NEOs, and made final decisions on compensation, including performance goal certification and payout approvals for Variable Cash Plan and PSUs. The CC also reviewed management's compensation risk analysis.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "roles", + "CC", + "ii", + "independent", + "compensation", + "consultant", + "iii", + "management", + "including", + "CEO", + "Finance", + "Human", + "Resources", + "departments", + "summarized", + "For", + "advice", + "Fiscal", + "2026", + "NEO", + "program", + "utilized", + "Semler", + "Brossy", + "experience", + "working", + "committees", + "technology", + "companies", + "Our", + "analyzed", + "role", + "raised", + "conflict", + "interests", + "taking", + "consideration", + "provide", + "services", + "directly", + "NVIDIA", + "paid", + "behalf", + "percentage", + "total", + "revenue", + "resulting", + "fees", + "interest", + "policies", + "procedures", + "Any", + "business", + "personal", + "relationship", + "executive", + "officer", + "individual", + "advisors", + "member", + "stock", + "owned", + "After", + "factors", + "determined", + "work", + "create", + "reviews", + "approves", + "decisions", + "input", + "Mr", + "Huang", + "At", + "direction", + "recommended", + "peer", + "group", + "pay", + "approved", + "provided", + "data", + "inform", + "gathered", + "Radford", + "Global", + "Technology", + "Survey", + "recommendations", + "NEOs", + "considered", + "proposed", + "performance", + "goals", + "making", + "final", + "certified", + "payouts", + "periods", + "concluded", + "end", + "Variable", + "Cash", + "Plan", + "SY", + "PSUs", + "granted", + "MY", + "2024", + "oversaw", + "risk", + "analysis" + ], + "keywords": [ + "Compensation Committee", + "Semler Brossy", + "NEO" + ], + "connect_to": [] + } + }, + { + "chunk_id": "e41c9b61-8fa9-547e-837f-4b6353ffc653", + "type": "table", + "content": "
Alphabet Inc. (GOOG)Meta Platforms, Inc. (META)
Amazon.com, Inc. (AMZN)Microsoft Corporation (MSFT)
Apple Inc. (AAPL)Netflix, Inc. (NFLX)
Broadcom Inc. (AVGO)Oracle Corporation (ORCL)
Cisco Systems, Inc. (CSCO)Qualcomm Incorporated (QCOM)
International Business Machines Corporation (IBM)Salesforce, Inc. (CRM)
Intel Corporation (INTC)Visa Inc. (V)
", + "path": "tables/table-31 Tech Companies.html", + "metadata": { + "length": 546, + "summary": "table-32\nThe table lists major technology companies including Alphabet, Meta, Amazon, Microsoft, Apple, Netflix, Broadcom, Oracle, Cisco, Qualcomm, IBM, Salesforce, Intel, and Visa with their stock tickers.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-31 Tech Companies.html", + "keywords": [ + "Tech Stocks", + "Companies", + "Tickers" + ], + "tokens": [] + } + }, + { + "chunk_id": "8e6a9b08-dc07-5b2f-aa36-83ae87b39d63", + "type": "image", + "content": "\nFiscal 2026 Peer Group
0\n[images/image-76-NVIDIA Salary Percentiles.jpg]\n", + "path": "images/image-76-NVIDIA Salary Percentiles.jpg", + "metadata": { + "length": 68, + "summary": "image-76\n75th Percentile: $229.70\nNVIDIA: $113.27\nMedian: $54.03\n25th Percentile: $40.93", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-76-NVIDIA Salary Percentiles.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "59f52edf-da3d-5dfa-bb85-9b5f7e5d9977", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
CategoryValue ($)
75th Percentile229.70
NVIDIA113.27
Median54.03
25th Percentile40.93
", + "path": "tables/table-32 Percentile Values.html", + "metadata": { + "length": 450, + "summary": "table-33\nThe table shows financial percentile values: 75th percentile at $229.70, NVIDIA at $113.27, median at $54.03, and 25th percentile at $40.93.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-32 Percentile Values.html", + "keywords": [ + "Percentiles", + "NVIDIA", + "Median" + ], + "tokens": [] + } + }, + { + "chunk_id": "90c051a7-01c3-5efc-8e31-f4976e98dedf", + "type": "image", + "content": "\n
0\n[images/image-77-NVIDIA vs Percentile Data.jpg]\n", + "path": "images/image-77-NVIDIA vs Percentile Data.jpg", + "metadata": { + "length": 56, + "summary": "image-77\nChart Title: Not specified\nY-Axis Label: Currency ($) ranging from $0 to $3,500\nX-Axis Categories: NVIDIA, 75th Percentile, Median, 25th Percentile\n\nData Points:\n- NVIDIA: $3,471.92\n- 75th Percentile: $1,913.24\n- Median: $536.35\n- 25th Percentile: $242.70", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-77-NVIDIA vs Percentile Data.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "3b939b95-9bd9-57c3-b1f4-e523c6868875", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
CategoryValue ($)
NVIDIA3,471.92
75th Percentile1,913.24
Median536.35
25th Percentile242.70
", + "path": "tables/table-33 NVIDIA Value Data.html", + "metadata": { + "length": 456, + "summary": "table-34\nThe table shows NVIDIA's value at $3,471.92, with the 75th percentile at $1,913.24, median at $536.35, and 25th percentile at $242.70.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-33 NVIDIA Value Data.html", + "keywords": [ + "NVIDIA", + "Percentiles", + "Values" + ], + "tokens": [] + } + }, + { + "chunk_id": "13be5423-f2e7-526a-bc6a-d315cc07bfdc", + "type": "text", + "content": "We believe our peers should be companies that compete with us for executive talent, have an established business, market presence, and complexity similar to NVIDIA, and are of similar size, measured by revenue or market capitalization at approximately 1/3 - 3.5x ours. After consulting with management, and based on Semler Brossy's recommendation, the CC determined in December 2024 that the existing peer group remained appropriate for Fiscal 2026, with the following adjustments: (i) removing Adobe Inc., Advanced Micro Devices, Inc., and SAP SE due to their revenue and market capitalization being below our target range (and also, for Adobe and SAP, because of limited direct business relevance); and (ii) adding Alphabet Inc., Amazon.com, Inc., Apple Inc., and Microsoft Corporation as competitors for talent with revenue and market capitalization being within our target range. These changes resulted in the following Fiscal 2026 peer group:\nFiscal 2026 Peer Group\n\n[tables/table-31 Tech Companies.html]\n\nOur CC selected each member of the peer group based on a combination of the factors described above. As a result, while some peer companies were outside our targeted size range, the CC determined they were still appropriate due to their established businesses and relevance as competitors for talent.\nTo determine our Fiscal 2026 peer group, the CC reviewed our trailing 12-month revenue (reported through our third quarter results for Fiscal 2025) and market capitalization as of November 2024. They compared these metrics to the 75th percentile, median, and 25th percentile of peer group companies, as follows:\nRevenue (in billions)\n\nFiscal 2026 Peer Group
0\n[images/image-76-NVIDIA Salary Percentiles.jpg]\n\n
\nbar chart\n\n[tables/table-32 Percentile Values.html]\n\n
\nMarket Capitalization (in billions)\n\n
0\n[images/image-77-NVIDIA vs Percentile Data.jpg]\n\n
\nbar chart\n\n[tables/table-33 NVIDIA Value Data.html]\n\n
\nFor our NEOs other than our CEO, Mr. Huang reviewed market practices and compensation data from the Radford Survey and peer company proxy data for comparable executives before recommending their Fiscal 2026 compensation for approval by the CC. The CC reviewed the same information for peer companies' CEOs, before determining the components of our executive compensation program, as well as total compensation, for all NEOs. We compared the total compensation opportunity for our NEOs and similarly situated executives at the 25th, 50th, and 75th percentiles of peer company data where available, and the CC considered the factors below in setting NEO compensation.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Peer Companies and Market Compensation Data", + "metadata": { + "length": 2691, + "summary": "NVIDIA's Compensation Committee updated its Fiscal 2026 peer group in December 2024, removing Adobe, AMD, and SAP due to size and relevance, and adding Alphabet, Amazon, Apple, and Microsoft as talent competitors. The peer group is selected based on competition for executive talent, business complexity, and size (revenue or market cap at 1/3 to 3.5x NVIDIA's). The committee compared NVIDIA's trailing 12-month revenue and market capitalization to peer percentiles. For NEOs other than the CEO, Mr. Huang reviewed market data from Radford Survey and peer proxies before recommending compensation. The CC reviewed CEO compensation data and set total compensation for all NEOs based on peer percentiles and other factors.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "peers", + "companies", + "compete", + "executive", + "talent", + "established", + "business", + "market", + "presence", + "complexity", + "similar", + "NVIDIA", + "size", + "measured", + "revenue", + "capitalization", + "approximately", + "3.5", + "After", + "consulting", + "management", + "based", + "Semler", + "Brossy", + "recommendation", + "CC", + "determined", + "December", + "2024", + "existing", + "peer", + "group", + "remained", + "Fiscal", + "2026", + "adjustments", + "removing", + "Adobe", + "Inc", + "Advanced", + "Micro", + "Devices", + "SAP", + "SE", + "due", + "target", + "range", + "limited", + "direct", + "relevance", + "ii", + "adding", + "Alphabet", + "Amazon", + "Apple", + "Microsoft", + "Corporation", + "competitors", + "These", + "resulted", + "Peer", + "Group", + "Our", + "selected", + "member", + "combination", + "factors", + "As", + "result", + "targeted", + "businesses", + "To", + "determine", + "reviewed", + "trailing", + "12", + "month", + "reported", + "quarter", + "results", + "2025", + "November", + "They", + "compared", + "metrics", + "75th", + "percentile", + "median", + "25th", + "Revenue", + "billions", + "details", + "summary", + "bar", + "chart", + "Market", + "Capitalization", + "For", + "NEOs", + "CEO", + "Mr", + "Huang", + "practices", + "compensation", + "data", + "Radford", + "Survey", + "company", + "proxy", + "comparable", + "executives", + "recommending", + "approval", + "The", + "information", + "CEOs", + "determining", + "components", + "program", + "total", + "opportunity", + "similarly", + "situated", + "50th", + "percentiles", + "considered", + "setting", + "NEO" + ], + "keywords": [ + "peer group", + "executive compensation", + "NVIDIA" + ], + "connect_to": [ + { + "target": "e41c9b61-8fa9-547e-837f-4b6353ffc653", + "relation": "embeds", + "ref": "[tables/table-31 Tech Companies.html]", + "position": { + "start": 972, + "end": 1009 + } + }, + { + "target": "8e6a9b08-dc07-5b2f-aa36-83ae87b39d63", + "relation": "embeds", + "ref": "[images/image-76-NVIDIA Salary Percentiles.jpg]", + "position": { + "start": 1681, + "end": 1728 + } + }, + { + "target": "59f52edf-da3d-5dfa-bb85-9b5f7e5d9977", + "relation": "embeds", + "ref": "[tables/table-32 Percentile Values.html]", + "position": { + "start": 1770, + "end": 1810 + } + }, + { + "target": "90c051a7-01c3-5efc-8e31-f4976e98dedf", + "relation": "embeds", + "ref": "[images/image-77-NVIDIA vs Percentile Data.jpg]", + "position": { + "start": 1882, + "end": 1929 + } + }, + { + "target": "3b939b95-9bd9-57c3-b1f4-e523c6868875", + "relation": "embeds", + "ref": "[tables/table-33 NVIDIA Value Data.html]", + "position": { + "start": 1971, + "end": 2011 + } + } + ] + } + }, + { + "chunk_id": "914c2a4a-48e9-5750-bbc5-93ebb55aaac5", + "type": "text", + "content": "In addition to peer data, our CC considers the following factors in making executive compensation decisions. The weight given to each factor may differ among NEOs and each component of pay, and is subject to the CC's sole discretion.\n√ The need to attract and retain talent in a highly competitive industry\n√ Stockholder feedback regarding our executive pay\n√ The desire for simplicity of the overall program and transparency of the performance metrics\n√ An NEO's past performance and anticipated future contributions\n√ Our financial performance and forecasted results, as well as our prior financial performance and resulting impact on our executives' compensation\n√ Changes in the scale and complexity of our business\n√ Each NEO’s current total compensation\n√ The scope and complexity of the department(s) or function(s) the NEO manages\n√ Each NEO's unvested equity\n√ Internal pay equity relative to similarly situated executives\n√ Our CEO's recommendations for the NEOs other than himself, including his understanding of each NEO's performance, capabilities, and contributions\n√ Our CC's independent judgment\n√ Our philosophy that an NEO's total compensation opportunity and percentage of at-risk pay should increase with responsibility\n√ The total compensation cost and the growth in such cost, including from executive compensation, to maintain a responsible cost structure for our compensation programs \\*", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Factors Used in Determining Executive Compensation", + "metadata": { + "length": 1411, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "In", + "addition", + "peer", + "data", + "CC", + "considers", + "factors", + "making", + "executive", + "compensation", + "decisions", + "The", + "weight", + "factor", + "differ", + "NEOs", + "component", + "pay", + "subject", + "sole", + "discretion", + "attract", + "retain", + "talent", + "highly", + "competitive", + "industry", + "Stockholder", + "feedback", + "desire", + "simplicity", + "program", + "transparency", + "performance", + "metrics", + "An", + "NEO", + "past", + "anticipated", + "future", + "contributions", + "Our", + "financial", + "forecasted", + "results", + "prior", + "resulting", + "impact", + "executives", + "Changes", + "scale", + "complexity", + "business", + "Each", + "current", + "total", + "scope", + "department", + "function", + "manages", + "unvested", + "equity", + "Internal", + "relative", + "similarly", + "situated", + "CEO", + "recommendations", + "including", + "understanding", + "capabilities", + "independent", + "judgment", + "philosophy", + "opportunity", + "percentage", + "risk", + "increase", + "responsibility", + "cost", + "growth", + "maintain", + "responsible", + "structure", + "programs" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ead69b71-f8d2-5e35-b74c-42843b9ddcf4", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "881ab9a6-6bfe-55ea-8e44-d72577f631d4", + "type": "text", + "content": "We value stockholder feedback and conduct an annual outreach program. In Fall 2024, in preparation for Fiscal 2026 compensation decisions, we contacted our top institutional stockholders representing an aggregate ownership of 31% of our shares. Management and the Board, including our Lead Director, met with stockholders holding 30% of our shares to discuss a wide range of topics including executive compensation. Stockholders provided positive feedback on our linkage of pay and performance and shared their views on multi-year performance goals and long equity vesting terms.\nIn response to this feedback and considering our 92% say-on-pay approval rate for Fiscal 2024 NEO compensation, our CC maintained generally the same elements and performance-based metrics for Fiscal 2026, but (i) increased target equity opportunities for all NEOs to recognize the complexity and scope of their roles and responsibilities, (ii) by increasing target equity opportunities, aligned total target pay for all NEOs other than our CEO to reflect their comparable, significant contributions and impact on corporate performance, and (iii) increased the rigor of the Company's performance-based executive pay opportunities by referencing our record Fiscal 2025 results and Fiscal 2025 Stretch Compensation Plan goals in setting Fiscal 2026 Base Compensation Plan goals to further motivate our NEOs. Our CC believed that structuring performance-based components of our executive pay program solely around NVIDIA's financial performance goals continued to align management's incentives with stockholder interests.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Stockholder Outreach and Feedback", + "metadata": { + "length": 1597, + "summary": "NVIDIA conducted an annual outreach program in Fall 2024, contacting top institutional stockholders representing 31% of shares. Management and the Board met with holders of 30% of shares to discuss executive compensation, receiving positive feedback on pay-for-performance linkage. With a 92% say-on-pay approval rate, the Compensation Committee maintained similar elements for Fiscal 2026 but increased target equity opportunities for all NEOs to reflect role complexity, aligned total target pay for non-CEO NEOs based on contributions, and increased rigor by referencing record Fiscal 2025 results and stretch goals. Performance-based pay remains tied solely to NVIDIA's financial performance to align management incentives with stockholder interests.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "stockholder", + "feedback", + "conduct", + "annual", + "outreach", + "program", + "In", + "Fall", + "2024", + "preparation", + "Fiscal", + "2026", + "compensation", + "decisions", + "contacted", + "top", + "institutional", + "stockholders", + "representing", + "aggregate", + "ownership", + "31%", + "shares", + "Management", + "Board", + "including", + "Lead", + "Director", + "met", + "holding", + "30%", + "discuss", + "wide", + "range", + "topics", + "executive", + "Stockholders", + "provided", + "positive", + "linkage", + "pay", + "performance", + "shared", + "views", + "multi", + "year", + "goals", + "long", + "equity", + "vesting", + "terms", + "response", + "92%", + "approval", + "rate", + "NEO", + "CC", + "maintained", + "generally", + "elements", + "based", + "metrics", + "increased", + "target", + "opportunities", + "NEOs", + "recognize", + "complexity", + "scope", + "roles", + "responsibilities", + "ii", + "increasing", + "aligned", + "total", + "CEO", + "reflect", + "comparable", + "significant", + "contributions", + "impact", + "corporate", + "iii", + "rigor", + "Company", + "referencing", + "record", + "2025", + "results", + "Stretch", + "Compensation", + "Plan", + "setting", + "Base", + "motivate", + "Our", + "believed", + "structuring", + "components", + "solely", + "NVIDIA", + "financial", + "continued", + "align", + "management", + "incentives", + "interests" + ], + "keywords": [ + "executive compensation", + "stockholder feedback", + "performance goals" + ], + "connect_to": [] + } + }, + { + "chunk_id": "dbe76243-0eca-5e79-a793-d072d86200fa", + "type": "text", + "content": "In setting Fiscal 2026 compensation, our CC reviewed each NEO's total target pay opportunity and distribution across multiple pay elements. Our CC compared Mr. Huang's base salary, target variable cash opportunity, target total cash opportunity, target equity opportunity, and total target pay opportunity against chief executives at peer companies. For our other NEOs, Mr. Huang reviewed their total target pay against similarly situated executives at peer companies where the data was available from the Radford Survey, considering internal pay equity, individual performance, unvested equity levels, and the increasing complexity of their roles. These factors informed Mr. Huang's recommendations to the CC. The CC also considered the factors outlined in Factors Used in Determining Executive Compensation and its compensation objectives for Fiscal 2026. Rather than applying a formula or assigning specific weights to each factor, our CC used its judgment and experience to set total target compensation, the mix of cash and equity, and fixed and at-risk pay opportunities for each NEO. When setting pay elements, the CC evaluated them in the context of the levels of the other pay elements and total target pay to ensure alignment with program objectives. The CC established amounts and designed a structure to reward NEOs with above-market value from equity awards and variable cash incentives only upon exceptional corporate performance.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Total Target Compensation Approach", + "metadata": { + "length": 1444, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "In", + "setting", + "Fiscal", + "2026", + "compensation", + "CC", + "reviewed", + "NEO", + "total", + "target", + "pay", + "opportunity", + "distribution", + "multiple", + "elements", + "Our", + "compared", + "Mr", + "Huang", + "base", + "salary", + "variable", + "cash", + "equity", + "chief", + "executives", + "peer", + "companies", + "For", + "NEOs", + "similarly", + "situated", + "data", + "Radford", + "Survey", + "internal", + "individual", + "performance", + "unvested", + "levels", + "increasing", + "complexity", + "roles", + "These", + "factors", + "informed", + "recommendations", + "The", + "considered", + "outlined", + "Factors", + "Used", + "Determining", + "Executive", + "Compensation", + "objectives", + "Rather", + "applying", + "formula", + "assigning", + "specific", + "weights", + "factor", + "judgment", + "experience", + "set", + "mix", + "fixed", + "risk", + "opportunities", + "When", + "evaluated", + "context", + "ensure", + "alignment", + "program", + "established", + "amounts", + "designed", + "structure", + "reward", + "market", + "awards", + "incentives", + "exceptional", + "corporate" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "63604d78-7ff9-5fa3-9cbe-90d5919a7839", + "type": "table", + "content": "
Fixed CompensationAt-Risk Compensation
Base SalaryVariable CashSY PSUsMY PSUsRSUs (1)
FormCashCashEquityEquityEquity
Who ReceivesNEOsNEOsNEOsNEOsNEOs (except our CEO, whose equity awards are solely PSUs)
Performance MeasureN/ARevenue (determines cash payout)Non-GAAP Operating Income (determines number of shares eligible to vest)TSR relative to the S&P 500 (determines number of shares eligible to vest)N/A
Performance PeriodN/A1 year1 year3 yearsN/A
Vesting PeriodN/AN/A4 years from grant3 years from grant4 years from grant
Vesting TermsN/AN/AIf at least Threshold goal achieved, 25% on approximately the 1-year anniversary of the grant date; 6.25% quarterly thereafterIf at least Threshold goal achieved, 100% on approximately the 3-year anniversary of the grant period6.25% vests quarterly from the grant date (2)
Timeframe EmphasizedAnnualAnnualLong-termLong-termLong-term
PurposeCompensate for expected day-to-day performanceReward for annual corporate financial performanceAlign with stockholder interests by linking NEO pay to annual operational performance and ongoing stock price performance during the vesting periodAlign with long-term stockholder interests by linking NEO pay to multi-year relative shareholder return and ongoing stock price performance during the vesting periodAlign with stockholder interests by linking NEO pay to ongoing stock price performance
Maximum Amount That Can Be EarnedN/A200% of target opportunity under our Variable Cash Plan150% of Mr. Huang's SY PSU target opportunity and 200% of our other NEOs' respective SY PSU target opportunityUltimate value delivered depends on stock price on date earned shares vest150% of Mr. Huang's MY PSU target opportunity and 200% of our other NEOs' respective MY PSU target opportunityUltimate value delivered depends on stock price on date earned shares vest100% of grantUltimate value delivered depends on stock price on date shares vest
", + "path": "tables/table-34 NEO Compensation.html", + "metadata": { + "length": 2629, + "summary": "table-35\nTable details NEO compensation components: fixed base salary, variable cash, and equity (PSUs, RSUs) with performance measures, vesting periods, and maximum earning potentials.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-34 NEO Compensation.html", + "keywords": [ + "Compensation", + "Equity", + "Performance" + ], + "tokens": [] + } + }, + { + "chunk_id": "9b27598d-f855-512f-82dc-dd7f5450e114", + "type": "text", + "content": "The primary components of NVIDIA's Fiscal 2026 executive compensation program are summarized below:\n\n[tables/table-34 NEO Compensation.html]\n\n(1) Our CC considers RSUs to be at-risk pay because the realized value depends on our stock price, a financial performance measure.\n(2) Reflects vesting schedule for annual RSU grants.\nOur NEOs have insurance benefits and are eligible to participate in our ESPP and 401(k) plan on the same basis as our other employees. We also provide a limited set of perquisites to our NEOs from time to time. See Other Compensation and Benefits below for more information.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Components of Pay", + "metadata": { + "length": 624, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "primary", + "components", + "NVIDIA", + "Fiscal", + "2026", + "executive", + "compensation", + "program", + "summarized", + "Our", + "CC", + "considers", + "RSUs", + "risk", + "pay", + "realized", + "depends", + "stock", + "price", + "financial", + "performance", + "measure", + "Reflects", + "vesting", + "schedule", + "annual", + "RSU", + "grants", + "NEOs", + "insurance", + "benefits", + "eligible", + "participate", + "ESPP", + "401", + "plan", + "basis", + "employees", + "We", + "provide", + "limited", + "set", + "perquisites", + "time", + "See", + "Other", + "Compensation", + "Benefits", + "information" + ], + "keywords": [], + "connect_to": [ + { + "target": "63604d78-7ff9-5fa3-9cbe-90d5919a7839", + "relation": "embeds", + "ref": "[tables/table-34 NEO Compensation.html]", + "position": { + "start": 101, + "end": 140 + } + } + ] + } + }, + { + "chunk_id": "f990ed93-72da-53ff-a06b-bcd6863f7b05", + "type": "text", + "content": "For Fiscal 2026, the CC decided that cash compensation for all NEOs would remain constant with Fiscal 2025 levels and that the largest portion of NEOs' total target pay would remain in the form of at-risk equity. The CC believes an emphasis on long-term, at-risk opportunities drives results and increases NEO and stockholder alignment, while providing sufficient annual cash compensation to be competitive and retain our NEOs. PSUs and RSUs provide long-term incentives and retention benefits as PSUs require achieving predetermined performance goals, and both PSUs and RSUs are subject to meaningful service-based vesting requirements (3 years for MY PSUs and 4 years for SY PSUs and RSUs).\nGiven Mr. Huang's position as CEO, to even more tightly align his interests with stockholders, the CC determined that $100\\%$ of his equity grants should be at-risk and performance-based. Consistent with prior years, the CC split Mr. Huang's target equity opportunity evenly between SY PSUs (aligned with our annual corporate financial performance) and MY PSUs (aligned with our 3-year relative shareholder return).\nFor other NEOs, the CC maintained a proportional equity mix and granted 50% of the target equity opportunity as RSUs and 50% as PSUs, with the PSU portion evenly split between SY PSUs and MY PSUs. The CC determined this structure balanced performance over short- and long-time horizons while providing a meaningful amount of time-vesting RSUs as a retention benefit.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Components of Pay/Pay Mix Considerations", + "metadata": { + "length": 1475, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "For", + "Fiscal", + "2026", + "CC", + "decided", + "cash", + "compensation", + "NEOs", + "remain", + "constant", + "2025", + "levels", + "largest", + "portion", + "total", + "target", + "pay", + "form", + "risk", + "equity", + "The", + "believes", + "emphasis", + "long", + "term", + "opportunities", + "drives", + "results", + "increases", + "NEO", + "stockholder", + "alignment", + "providing", + "sufficient", + "annual", + "competitive", + "retain", + "PSUs", + "RSUs", + "provide", + "incentives", + "retention", + "benefits", + "require", + "achieving", + "predetermined", + "performance", + "goals", + "subject", + "meaningful", + "service", + "based", + "vesting", + "requirements", + "years", + "MY", + "SY", + "Given", + "Mr", + "Huang", + "position", + "CEO", + "tightly", + "align", + "interests", + "stockholders", + "determined", + "100", + "grants", + "Consistent", + "prior", + "split", + "opportunity", + "evenly", + "aligned", + "corporate", + "financial", + "year", + "relative", + "shareholder", + "return", + "maintained", + "proportional", + "mix", + "granted", + "50%", + "PSU", + "structure", + "balanced", + "short", + "time", + "horizons", + "amount", + "benefit" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "932f05c4-b413-5e96-82b3-d9f6563c0be9", + "type": "text", + "content": "For Fiscal 2026, the CC decided that increases to each NEO's total target pay were appropriate in light of the scope and complexity of their roles and responsibilities.\nThe CC raised Mr. Huang's total target pay by \\$3.5 million, or 11%, to \\$35.5 million, all in the form of equity, to align more closely with the median of the updated group of peer company CEOs. The CC split the target equity increase evenly between SY PSUs and MY PSUs to balance short- and long-term performance-based awards.\nFor other NEOs, the CC decided, on Mr. Huang's recommendation, to align their total target pay to \\$16 million each, to reflect their comparable, significant contributions and impact on corporate performance, by increasing target equity opportunities by \\$1 to \\$2.5 million. For these NEOs, where peer data was available, the increases resulted in their total compensation approximating the 65th percentile of their respective peers.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Components of Pay/Setting Executive Compensation Values", + "metadata": { + "length": 932, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "For", + "Fiscal", + "2026", + "CC", + "decided", + "increases", + "NEO", + "total", + "target", + "pay", + "light", + "scope", + "complexity", + "roles", + "responsibilities", + "The", + "raised", + "Mr", + "Huang", + "3.5", + "million", + "11%", + "35.5", + "form", + "equity", + "align", + "closely", + "median", + "updated", + "group", + "peer", + "company", + "CEOs", + "split", + "increase", + "evenly", + "SY", + "PSUs", + "MY", + "balance", + "short", + "long", + "term", + "performance", + "based", + "awards", + "NEOs", + "recommendation", + "16", + "reflect", + "comparable", + "significant", + "contributions", + "impact", + "corporate", + "increasing", + "opportunities", + "2.5", + "data", + "resulted", + "compensation", + "approximating", + "65th", + "percentile", + "respective", + "peers" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b44a05f8-0d6d-5e4e-b310-9ffcba0e4e4c", + "type": "text", + "content": "To determine the actual number of RSUs and target numbers of SY PSUs and MY PSUs awarded to our NEOs, the CC divided the value of the target equity they intended to deliver at grant by the 30-calendar day trailing average closing price of our common stock ending on the last day of the calendar month prior to the date of grant. They used this methodology instead of calculating the number of shares of common stock subject to each award based on the stock price on the grant date to smooth the effects of possible market volatility. The CC understands that using a historical average stock price can cause the ASC 718 grant date value of an award, as required to be reported in the Summary Compensation Table and Grants of Plan-Based Awards Table, to be different than the intended value of the target equity opportunity. The CC considered various approaches to granting awards and determined the process described above was appropriate.\nThe target number of SY PSUs would become eligible to vest if the Company achieved Fiscal 2026 Non-GAAP Operating Income at Base Compensation Plan. If the Company achieved Fiscal 2026 Non-GAAP Operating Income at Stretch Compensation Plan or more, the maximum SY PSUs eligible to vest would be capped at 150% of Mr. Huang's SY PSU target equity opportunity and at 200% for other NEOs. If the Company achieved Fiscal 2026 Non-GAAP Operating Income at Threshold, the minimum SY PSUs eligible to vest would be 50% of each NEO's SY PSU target equity opportunity.\nThe target number of MY PSUs would become eligible to vest if the Company achieved TSR relative to the S&P 500 from the start of Fiscal 2026 to the end of Fiscal 2028, or the 3-Year Relative TSR, at Base Compensation Plan. If the Company achieved 3-Year Relative TSR at Stretch Compensation Plan or more, the maximum MY PSUs eligible to vest would be capped at $150\\%$ of Mr. Huang's MY PSU target equity opportunity and at $200\\%$ for other NEOs. If the Company achieved 3-Year Relative TSR at Threshold, the minimum MY PSUs eligible to vest would be $25\\%$ of each NEO's MY PSU target equity opportunity.\nThe CC capped the payout for Mr. Huang's SY PSUs and MY PSUs at $150\\%$ of his target equity awards instead of $200\\%$ that applied to other NEOs because all of his Fiscal 2026 equity compensation was performance-based. With these caps, Mr. Huang would have the same total upside potential on his equity awards as the other NEOs. However, the other NEOs had $50\\%$ of their total equity awards granted as RSUs that vest based solely on continued service and therefore did not provide the same upside potential.\nNo PSUs would be eligible to vest if the applicable Threshold performance level was not achieved. Any PSUs determined to be unearned would be cancelled.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Components of Pay/Determining Equity Award Amounts", + "metadata": { + "length": 2768, + "summary": "The Compensation Committee determines RSU, SY PSU, and MY PSU awards for NEOs using a 30-day average stock price to mitigate volatility. SY PSUs vest based on Fiscal 2026 Non-GAAP Operating Income, with caps at 150% for Mr. Huang and 200% for others, and a minimum of 50% at Threshold. MY PSUs vest based on 3-Year Relative TSR vs S&P 500, with caps of 150% for Mr. Huang and 200% for others, and a minimum of 25% at Threshold. Mr. Huang's lower cap is due to all his equity being performance-based, while others have 50% in service-based RSUs. No PSUs vest if Threshold is not met.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "To", + "determine", + "actual", + "number", + "RSUs", + "target", + "numbers", + "SY", + "PSUs", + "MY", + "awarded", + "NEOs", + "CC", + "divided", + "equity", + "intended", + "deliver", + "grant", + "30", + "calendar", + "day", + "trailing", + "average", + "closing", + "price", + "common", + "stock", + "ending", + "month", + "prior", + "date", + "They", + "methodology", + "calculating", + "shares", + "subject", + "award", + "based", + "smooth", + "effects", + "market", + "volatility", + "The", + "understands", + "historical", + "ASC", + "718", + "required", + "reported", + "Summary", + "Compensation", + "Table", + "Grants", + "Plan", + "Based", + "Awards", + "opportunity", + "considered", + "approaches", + "granting", + "awards", + "determined", + "process", + "eligible", + "vest", + "Company", + "achieved", + "Fiscal", + "2026", + "Non", + "GAAP", + "Operating", + "Income", + "Base", + "If", + "Stretch", + "maximum", + "capped", + "150%", + "Mr", + "Huang", + "PSU", + "200%", + "Threshold", + "minimum", + "50%", + "NEO", + "TSR", + "relative", + "500", + "start", + "end", + "2028", + "Year", + "Relative", + "150", + "200", + "25", + "payout", + "applied", + "compensation", + "performance", + "With", + "caps", + "total", + "upside", + "potential", + "However", + "50", + "granted", + "solely", + "continued", + "service", + "provide", + "No", + "applicable", + "level", + "Any", + "unearned", + "cancelled" + ], + "keywords": [ + "RSU", + "PSU", + "Performance Vesting" + ], + "connect_to": [] + } + }, + { + "chunk_id": "64e1f819-5c66-56dd-b3a9-0f1fee075fed", + "type": "table", + "content": "
Variable Cash PlanSY PSUsMY PSUs
MetricRevenueNon-GAAP Operating IncomeTSR relative to the S&P 500
Timeframe1 year1 year3 years
CC's Rationale for MetricDrives value, contributes to Company's long-term successDrives value, contributes to Company's long-term successAligns directly with long-term shareholder value creation
Focuses on growth in new and existing marketsReflects our annual revenue generation and effective operating expense managementProvides comparison of our stock price performance, including dividends, against a capital market index in which we compete
Distinct, separate metric from Non-GAAP Operating IncomeDistinct, separate metrics from revenueRelative performance goal accounts for macroeconomic factors impacting the market
", + "path": "tables/table-35 Compensation Plans.html", + "metadata": { + "length": 1042, + "summary": "table-36\nThe table outlines three compensation plans: Variable Cash Plan (1-year Revenue metric), SY PSUs (1-year Non-GAAP Operating Income), and MY PSUs (3-year TSR relative to S&P 500). Rationale focuses on driving value, growth, and shareholder alignment.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-35 Compensation Plans.html", + "keywords": [ + "Variable Cash", + "PSUs", + "Metrics" + ], + "tokens": [] + } + }, + { + "chunk_id": "75b8c382-b8d4-5ed8-8a25-8143cfd14c37", + "type": "table", + "content": "
Variable Cash PlanSY PSUsMY PSUs
Fiscal 2026 RevenuePayout as a % of Target Opportunity (1)Fiscal 2026 Non-GAAP Operating Income (2)Shares Eligible to Vest as a % of Target Opportunity (1)Fiscal 2026 to 2028 3-Year Relative TSR (3)Shares Eligible to Vest as a % of Target Opportunity (1)
Threshold$100.0 billion50%$46.0 billion50%25th percentile25%
Base Compensation Plan$130.0 billion100%$71.0 billion100%50th percentile100%
Stretch Compensation Plan$160.0 billion (4)200%$96.0 billion (4)CEO 150% Other NEOs 200%75th percentileCEO 150% Other NEOs 200%
", + "path": "tables/table-36 Compensation Plan.html", + "metadata": { + "length": 936, + "summary": "table-37\nTable shows Variable Cash Plan and PSUs with Threshold, Base, and Stretch targets for Fiscal 2026 Revenue, Non-GAAP Operating Income, and 3-Year Relative TSR.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-36 Compensation Plan.html", + "keywords": [ + "Variable Cash", + "PSUs", + "TSR" + ], + "tokens": [] + } + }, + { + "chunk_id": "733bbe60-7812-5ce8-9015-d913da54952d", + "type": "text", + "content": "In March 2025, when making decisions regarding Fiscal 2026 executive compensation, the CC intended for performance goals to be rigorous, uncertain, and set at levels that would motivate our NEOs. As a result, the CC set the Base Compensation Plan revenue goal for the Fiscal 2026 Variable Cash Plan in line with our record Fiscal 2025 results and the Base Compensation Plan Non-GAAP Operating Income goal for SY PSUs granted in Fiscal 2026 in line with our corresponding Fiscal 2025 Stretch Compensation Plan goal.\nThe CC also considered the impact of potential regulatory actions, outside of management's control, on the Company's Fiscal 2026 performance. As a result, the CC approved Stretch Compensation Plans for Fiscal 2026 revenue of \\$190.0 billion and for Fiscal 2026 Non-GAAP Operating Income of \\$120.0 billion, which significantly exceeded record Fiscal 2025 results and Fiscal 2025 Stretch Compensation Plan goals. However, at the time it set the Fiscal 2026 performance goals, the CC also provided that the revenue and Non-GAAP Operating Income Stretch Compensation Plan goals would automatically be reduced to \\$160.0 billion and \\$96.0 billion, respectively, to the extent that additional export control restrictions were subsequently imposed on the Company's H2O products within the first half of Fiscal 2026. The CC adopted this framework to preserve the alignment between executive compensation and Company performance in the event of significant regulatory developments that could impact our financial results.\nIn April 2025, the U.S. government informed NVIDIA that a license would be required for exports of H2O products into the China market. Accordingly, the Fiscal 2026 Stretch Compensation Plan goals for revenue and Non-GAAP Operating Income were automatically adjusted.\nFiscal 2026 performance metrics and goals for NEO pay were as set forth below:\nPERFORMANCE METRICS\n\n[tables/table-35 Compensation Plans.html]\n\nPERFORMANCE GOALS\n\n[tables/table-36 Compensation Plan.html]\n\n(1) For achievement between Threshold and Base Compensation Plan, or alternatively between Base Compensation Plan and Stretch Compensation Plan, payouts would be determined using straight-line interpolation. Achievement less than Threshold would result in no payout, and achievement exceeding Stretch Compensation Plan would result in the capped maximum payout.\n(2) See Reconciliation of Non-GAAP Financial Measures below for a reconciliation between the non-GAAP financial measures and GAAP results.\n(3) MY PSUs granted in Fiscal 2026 cover the Fiscal 2026 to Fiscal 2028 performance period. MY PSUs covering the Fiscal 2024 to Fiscal 2026 performance period were granted in Fiscal 2024 and consist of the same performance goal structure and payout opportunities.\n(4) Our original Fiscal 2026 Stretch Compensation Plan revenue goal of \\$190.0 billion was automatically adjusted to \\$160.0 billion, and our original Fiscal 2026 Stretch Compensation Plan Non-GAAP Operating Income goal of \\$120.0 billion was automatically adjusted to \\$96.0 billion, due to the imposition of additional export controls on the Company's H2O products during the first half of Fiscal 2026.\nEach of the performance goal levels described above were set by the CC with the following objectives:\n- Threshold was uncertain, but attainable and high enough to create value; represented an appropriately decelerated payout for performance below Base Compensation Plan.\n- Base Compensation Plan was uncertain, but attainable with significant effort and execution success; included budgeted investments in future businesses and revenue growth considering macroeconomic conditions and reasonable but challenging growth estimates for ongoing and new businesses.\n\\- Stretch Compensation Plan required exceptional achievement; only possible with strong market factors and a very high level of management execution and corporate performance; for Fiscal 2026, included an automatic adjustment in the event of additional export controls on the Company's H2O products in the first half of the fiscal year to preserve the alignment between executive compensation and Company performance in the event of significant regulatory developments outside of management's control.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Performance Metrics and Goals for Executive Compensation", + "metadata": { + "length": 4266, + "summary": "In March 2025, the Compensation Committee (CC) set rigorous Fiscal 2026 performance goals for executive compensation, aligning revenue and Non-GAAP Operating Income targets with record Fiscal 2025 results. Stretch goals were set at $190 billion revenue and $120 billion Non-GAAP Operating Income, but with an automatic reduction to $160 billion and $96 billion if additional export controls on H2O products were imposed in the first half of Fiscal 2026. In April 2025, after the U.S. government required a license for H2O exports to China, the goals were automatically adjusted. The CC designed Threshold, Base, and Stretch goals to be uncertain, challenging, and aligned with company performance, with straight-line interpolation for payouts between levels and no payout below Threshold.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "In", + "March", + "2025", + "making", + "decisions", + "Fiscal", + "2026", + "executive", + "compensation", + "CC", + "intended", + "performance", + "goals", + "rigorous", + "uncertain", + "set", + "levels", + "motivate", + "NEOs", + "As", + "result", + "Base", + "Compensation", + "Plan", + "revenue", + "goal", + "Variable", + "Cash", + "line", + "record", + "results", + "Non", + "GAAP", + "Operating", + "Income", + "SY", + "PSUs", + "granted", + "Stretch", + "The", + "considered", + "impact", + "potential", + "regulatory", + "actions", + "management", + "control", + "Company", + "approved", + "Plans", + "190.0", + "billion", + "120.0", + "significantly", + "exceeded", + "However", + "time", + "provided", + "automatically", + "reduced", + "160.0", + "96.0", + "extent", + "additional", + "export", + "restrictions", + "subsequently", + "imposed", + "H2O", + "products", + "half", + "adopted", + "framework", + "preserve", + "alignment", + "event", + "significant", + "developments", + "financial", + "April", + "government", + "informed", + "NVIDIA", + "license", + "required", + "exports", + "China", + "market", + "Accordingly", + "adjusted", + "metrics", + "NEO", + "pay", + "PERFORMANCE", + "METRICS", + "GOALS", + "For", + "achievement", + "Threshold", + "alternatively", + "payouts", + "determined", + "straight", + "interpolation", + "Achievement", + "payout", + "exceeding", + "capped", + "maximum", + "See", + "Reconciliation", + "Financial", + "Measures", + "reconciliation", + "measures", + "MY", + "cover", + "2028", + "period", + "covering", + "2024", + "consist", + "structure", + "opportunities", + "Our", + "original", + "due", + "imposition", + "controls", + "Each", + "objectives", + "attainable", + "high", + "create", + "represented", + "appropriately", + "decelerated", + "effort", + "execution", + "success", + "included", + "budgeted", + "investments", + "future", + "businesses", + "growth", + "macroeconomic", + "conditions", + "reasonable", + "challenging", + "estimates", + "ongoing", + "exceptional", + "strong", + "factors", + "level", + "corporate", + "automatic", + "adjustment", + "fiscal", + "year" + ], + "keywords": [ + "executive compensation", + "performance goals", + "export controls" + ], + "connect_to": [ + { + "target": "64e1f819-5c66-56dd-b3a9-0f1fee075fed", + "relation": "embeds", + "ref": "[tables/table-35 Compensation Plans.html]", + "position": { + "start": 1897, + "end": 1938 + } + }, + { + "target": "75b8c382-b8d4-5ed8-8a25-8143cfd14c37", + "relation": "embeds", + "ref": "[tables/table-36 Compensation Plan.html]", + "position": { + "start": 1959, + "end": 1999 + } + } + ] + } + }, + { + "chunk_id": "1175208d-63ff-57e9-961f-31cdd9409319", + "type": "table", + "content": "
PERFORMANCE ACHIEVEMENT AND PAYOUTS
Variable Cash PlanSY PSUsMY PSUs (1)
Performance Achievement for Period Ended Fiscal 2026 (2)$215.9 billion revenue$137.3 billion Non-GAAP Operating Income (3)100th percentile 3-year TSR relative to S&P 500 (1,055%)
Payout as % of Target Opportunity200%CEO 150% Other NEOs 200% (4)CEO 150% Other NEOs 200% (5)
", + "path": "tables/table-37 Performance Payouts.html", + "metadata": { + "length": 526, + "summary": "table-38\nTable shows performance achievement for Fiscal 2026: $215.9B revenue, $137.3B Non-GAAP Operating Income, 100th percentile 3-year TSR. Payouts range from 150% to 200% of target.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-37 Performance Payouts.html", + "keywords": [ + "Performance Achievement", + "Payouts", + "PSUs" + ], + "tokens": [] + } + }, + { + "chunk_id": "845a3b9a-42b6-5631-ae48-837793ed3340", + "type": "text", + "content": "As a result of record performance on strength across all market platforms, led by Data Center demand for our Blackwell architecture and across a diverse and expanding set of customers including cloud providers, hyperscalers, AI model makers, enterprises, startups and sovereigns, revenue and Non-GAAP Operating Income for Fiscal 2026 exceeded their respective Stretch Compensation Plan goals, even when compared to the original Stretch Compensation Plan goals prior to the automatic adjustments described above. Our Fiscal 2024 to Fiscal 2026 3-year TSR relative to the S&P 500 also exceeded its Stretch Compensation Plan goal.\nIn March 2026, the CC certified the Company's performance achievement with the following payouts:\n\n[tables/table-37 Performance Payouts.html]\n\n(1) Represents performance achievement and payout of MY PSUs granted in Fiscal 2024, with a performance period measured from the start of Fiscal 2024 to the end of Fiscal 2026.\n(2) Revenue is GAAP revenue, as the Company reports in its SEC filings. Non-GAAP Operating Income is GAAP operating income, as the Company reports in its SEC filings, excluding stock-based compensation expense, acquisition-related and other costs, and other. Consistent with prior years, 3-year TSR for purposes of the MY PSUs represents cumulative stock price appreciation, with dividends reinvested, and is measured based on the average closing stock price for the 60 trading days preceding the start, and preceding and including the last day, of the 3-year performance period. This averaging period mitigates the impact of one-day or short-term stock price fluctuations at the beginning or end of the performance period.\n(3) See Reconciliation of Non-GAAP Financial Measures below for a reconciliation between the non-GAAP financial measures and GAAP results.\n(4) 25% of the eligible SY PSUs vested on March 18, 2026, approximately one year after grant, and 6.25% will vest every quarter thereafter for the next three years subject to the NEO's continued service to the Company.\n(5) 100% of the eligible MY PSUs granted in Fiscal 2024 vested on March 18, 2026.\nAchievement of goals for MY PSUs granted during Fiscal 2025 and Fiscal 2026 will be determined after the applicable performance periods conclude in January 2027 and January 2028, respectively.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Fiscal 2026 Performance Achievement", + "metadata": { + "length": 2308, + "summary": "The company achieved record performance across all market platforms, driven by Data Center demand for Blackwell architecture and diverse customers. Revenue and Non-GAAP Operating Income for Fiscal 2026 exceeded Stretch Compensation Plan goals, as did the 3-year TSR relative to S&P 500. In March 2026, the Compensation Committee certified performance achievement with payouts: 25% of eligible SY PSUs vested on March 18, 2026, with quarterly vesting thereafter; 100% of eligible MY PSUs granted in Fiscal 2024 vested on March 18, 2026. Goals for MY PSUs granted in Fiscal 2025 and 2026 will be determined after performance periods end in January 2027 and 2028, respectively.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "As", + "result", + "record", + "performance", + "strength", + "market", + "platforms", + "led", + "Data", + "Center", + "demand", + "Blackwell", + "architecture", + "diverse", + "expanding", + "set", + "customers", + "including", + "cloud", + "providers", + "hyperscalers", + "AI", + "model", + "makers", + "enterprises", + "startups", + "sovereigns", + "revenue", + "Non", + "GAAP", + "Operating", + "Income", + "Fiscal", + "2026", + "exceeded", + "respective", + "Stretch", + "Compensation", + "Plan", + "goals", + "compared", + "original", + "prior", + "automatic", + "adjustments", + "Our", + "2024", + "year", + "TSR", + "relative", + "500", + "goal", + "In", + "March", + "CC", + "certified", + "Company", + "achievement", + "payouts", + "Represents", + "payout", + "MY", + "PSUs", + "granted", + "period", + "measured", + "start", + "end", + "Revenue", + "reports", + "SEC", + "filings", + "operating", + "income", + "excluding", + "stock", + "based", + "compensation", + "expense", + "acquisition", + "related", + "costs", + "Consistent", + "years", + "purposes", + "represents", + "cumulative", + "price", + "appreciation", + "dividends", + "reinvested", + "average", + "closing", + "60", + "trading", + "days", + "preceding", + "day", + "This", + "averaging", + "mitigates", + "impact", + "short", + "term", + "fluctuations", + "beginning", + "See", + "Reconciliation", + "Financial", + "Measures", + "reconciliation", + "financial", + "measures", + "results", + "25%", + "eligible", + "SY", + "vested", + "18", + "approximately", + "grant", + "6.25%", + "vest", + "quarter", + "subject", + "NEO", + "continued", + "service", + "100%", + "Achievement", + "2025", + "determined", + "applicable", + "periods", + "conclude", + "January", + "2027", + "2028" + ], + "keywords": [ + "PSU", + "Performance", + "Compensation" + ], + "connect_to": [ + { + "target": "1175208d-63ff-57e9-961f-31cdd9409319", + "relation": "embeds", + "ref": "[tables/table-37 Performance Payouts.html]", + "position": { + "start": 727, + "end": 769 + } + } + ] + } + }, + { + "chunk_id": "55a5a8a2-b29b-5316-9be3-a8f54003ab70", + "type": "table", + "content": "
President & CEOTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary1,500,000Flat with Fiscal 2025
Variable Cash3,000,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($6,000,000)
Cash4,500,000Flat with Fiscal 2025
SY PSUs15,500,000Up $1.8 million, or 13%, from Fiscal 2025 target, resulting in 119,607 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 150% of target opportunity (179,411 shares) becoming eligible to vest
MY PSUs15,500,000Up $1.8 million, or 13%, from Fiscal 2025 target, resulting in 119,607 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 150% of target opportunity (757,360 shares) becoming eligible to vest
Equity31,000,000Up $3.5 million, or 13%, from Fiscal 2025 target
TOTAL35,500,000Up $3.5 million, or 11%, from Fiscal 2025 target
", + "path": "tables/table-38 CEO Pay Summary.html", + "metadata": { + "length": 1487, + "summary": "table-39\nCEO target pay totals $35.5M, up 11% from FY2025. Base salary flat at $1.5M. Variable cash at $3M with 200% payout. Equity up 13% to $31M, with PSUs vesting at 150% of target.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-38 CEO Pay Summary.html", + "keywords": [ + "CEO compensation", + "performance pay", + "equity" + ], + "tokens": [] + } + }, + { + "chunk_id": "e970d8a3-fb2b-580d-a91b-c53391959032", + "type": "table", + "content": "
EVP & CFOTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary900,000Flat with Fiscal 2025
Variable Cash300,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($600,000)
Cash1,200,000Flat with Fiscal 2025
SY PSUs3,699,807Up $0.2 million, or 7%, from Fiscal 2025 target, resulting in 28,550 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (57,100 shares) becoming eligible to vest
MY PSUs3,699,807Up $0.2 million, or 7%, from Fiscal 2025 target, resulting in 28,550 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (49,560 shares) becoming eligible to vest
RSUs7,400,386Up $0.5 million, or 7%, from Fiscal 2025, resulting in 57,106 shares granted in Fiscal 2026
Equity14,800,000Up $1.0 million, or 7%, from Fiscal 2025 target
TOTAL16,000,000Up $1.0 million, or 7%, from Fiscal 2025 target
", + "path": "tables/table-39 EVP CFO Pay.html", + "metadata": { + "length": 1614, + "summary": "table-40\nEVP & CFO target pay is $16M, up 7% from FY2025. Base salary flat at $900K. Variable cash at 200% payout. SY and MY PSUs at 200% target. RSUs up 7%.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-39 EVP CFO Pay.html", + "keywords": [ + "Compensation", + "Equity", + "Performance" + ], + "tokens": [] + } + }, + { + "chunk_id": "ba1346a8-84b1-57dd-aaea-715c9b461fc9", + "type": "table", + "content": "
EVP, Worldwide Field OperationsTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary950,000Flat with Fiscal 2025
Variable Cash650,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($1,300,000)
Cash1,600,000Flat with Fiscal 2025
SY PSUs3,600,000Up $0.3 million, or 7%, from Fiscal 2025 target, resulting in 27,779 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (55,558 shares) becoming eligible to vest
MY PSUs3,600,000Up $0.3 million, or 7%, from Fiscal 2025 target, resulting in 27,779 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (47,720 shares) becoming eligible to vest
RSUs7,200,000Up $0.5 million, or 7%, from Fiscal 2025, resulting in 55,559 shares granted in Fiscal 2026
Equity14,400,000Up $1.0 million, or 7%, from Fiscal 2025 target
TOTAL16,000,000Up $1.0 million, or 7%, from Fiscal 2025 target
", + "path": "tables/table-40 EVP Compensation.html", + "metadata": { + "length": 1634, + "summary": "table-41\nEVP Worldwide Field Operations total target pay is $16M, up 7% from FY2025. Base salary $950K, variable cash $650K, equity $14.4M. Performance-based payouts at 200% for exceeding stretch goals.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-40 EVP Compensation.html", + "keywords": [ + "Compensation", + "Equity", + "Performance" + ], + "tokens": [] + } + }, + { + "chunk_id": "5c8fffa7-6ed3-5a74-bf1e-f7ce0266a2d3", + "type": "table", + "content": "
EVP, OperationsTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary850,000Flat with Fiscal 2025
Variable Cash250,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($500,000)
Cash1,100,000Flat with Fiscal 2025
SY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (57,488 shares) becoming eligible to vest
MY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (40,840 shares) becoming eligible to vest
RSUs7,450,000Up $1.3 million, or 20%, from Fiscal 2025, resulting in 57,489 shares granted in Fiscal 2025
Equity14,900,000Up $2.5 million, or 20%, from Fiscal 2025 target
TOTAL16,000,000Up $2.5 million, or 19%, from Fiscal 2025 target
", + "path": "tables/table-41 EVP Compensation.html", + "metadata": { + "length": 1621, + "summary": "table-42\nEVP Operations target pay $16M, up 19% from FY2025. Base salary flat at $850K. Variable cash flat at $250K, with 200% payout. Equity up 20% to $14.9M. PSUs and RSUs increased.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-41 EVP Compensation.html", + "keywords": [ + "EVP", + "Compensation", + "Fiscal 2026" + ], + "tokens": [] + } + }, + { + "chunk_id": "d312e756-08e8-514c-93fb-4c2809072eac", + "type": "table", + "content": "
EVP, General Counsel & SecretaryTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary850,000Flat with Fiscal 2025
Variable Cash250,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($500,000)
Cash1,100,000Flat with Fiscal 2025
SY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (57,488 shares) becoming eligible to vest
MY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (40,840 shares) becoming eligible to vest
RSUs7,450,000Up $1.3 million, or 20%, from Fiscal 2025, resulting in 57,489 shares granted in Fiscal 2025
Equity14,900,000Up $2.5 million, or 20%, from Fiscal 2025 target
TOTAL16,000,000Up $2.5 million, or 19%, from Fiscal 2025 target
", + "path": "tables/table-42 Executive Pay.html", + "metadata": { + "length": 1642, + "summary": "table-43\nEVP General Counsel target pay $16M, up 19% from FY2025. Base salary $850K, variable cash $250K, equity $14.9M. Performance-based payouts at 200% for exceeding goals.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-42 Executive Pay.html", + "keywords": [ + "Compensation", + "Equity", + "Performance" + ], + "tokens": [] + } + }, + { + "chunk_id": "94cbf47c-9284-5a86-8e64-8bc271574d89", + "type": "text", + "content": "The CC's target Fiscal 2026 compensation actions are summarized below for each NEO, reflecting the target variable cash and intended value of the target equity opportunities, as well as the variable cash earned and PSUs that became eligible to vest.\nThe CC considered the factors set forth in Factors Used in Determining Executive Compensation above to set total target pay opportunity for each NEO, which are described in Fiscal 2026 Compensation Actions and Achievements - Setting Executive Compensation Values above.\nThe intended values of the target equity opportunities presented in the tables below are equal to the number of shares underlying each NEO's equity awards granted in Fiscal 2026, calculated by assuming achievement at Base Compensation Plan for PSUs, multiplied by the 30-day trailing average closing price of our common stock that the CC used in approving such equity awards, as described above in Determining Equity Award Amounts. These values differ from those reported in the Summary Compensation Table and Grants of Plan-Based Awards Table, which, in accordance with SEC rules, reflect the ASC 718 grant date fair value of each NEO's equity awards based on the single day closing price of our common stock on the date of grant and, for PSUs, assuming a probable outcome of the applicable performance conditions.\nJen-Hsun Huang\n\n[tables/table-38 CEO Pay Summary.html]\n\nColette M. Kress\n\n[tables/table-39 EVP CFO Pay.html]\n\nAjay K. Puri\n\n[tables/table-40 EVP Compensation.html]\n\nDebora Shoquist\n\n[tables/table-41 EVP Compensation.html]\n\nTimothy S. Teter\n\n[tables/table-42 Executive Pay.html]", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Fiscal 2026 Target Compensation Actions and Performance-Based Payouts", + "metadata": { + "length": 1930, + "summary": "The passage describes the Compensation Committee's target compensation actions for Fiscal 2026 for each named executive officer (NEO), including variable cash and equity opportunities. It explains that target equity values are based on the 30-day trailing average stock price, differing from SEC-reported grant date fair values. The text references individual tables for each NEO: Jen-Hsun Huang, Colette M. Kress, Ajay K. Puri, Debora Shoquist, and Timothy S. Teter, detailing their target pay structures.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "CC", + "target", + "Fiscal", + "2026", + "compensation", + "actions", + "summarized", + "NEO", + "reflecting", + "variable", + "cash", + "intended", + "equity", + "opportunities", + "earned", + "PSUs", + "eligible", + "vest", + "considered", + "factors", + "set", + "Factors", + "Used", + "Determining", + "Executive", + "Compensation", + "total", + "pay", + "opportunity", + "Actions", + "Achievements", + "Setting", + "Values", + "values", + "presented", + "tables", + "equal", + "number", + "shares", + "underlying", + "awards", + "granted", + "calculated", + "assuming", + "achievement", + "Base", + "Plan", + "multiplied", + "30", + "day", + "trailing", + "average", + "closing", + "price", + "common", + "stock", + "approving", + "Equity", + "Award", + "Amounts", + "These", + "differ", + "reported", + "Summary", + "Table", + "Grants", + "Based", + "Awards", + "accordance", + "SEC", + "rules", + "reflect", + "ASC", + "718", + "grant", + "date", + "fair", + "based", + "single", + "probable", + "outcome", + "applicable", + "performance", + "conditions", + "Jen", + "Hsun", + "Huang", + "Colette", + "Kress", + "Ajay", + "Puri", + "Debora", + "Shoquist", + "Timothy", + "Teter" + ], + "keywords": [ + "NEO compensation", + "target pay", + "equity awards" + ], + "connect_to": [ + { + "target": "55a5a8a2-b29b-5316-9be3-a8f54003ab70", + "relation": "embeds", + "ref": "[tables/table-38 CEO Pay Summary.html]", + "position": { + "start": 1352, + "end": 1390 + } + }, + { + "target": "e970d8a3-fb2b-580d-a91b-c53391959032", + "relation": "embeds", + "ref": "[tables/table-39 EVP CFO Pay.html]", + "position": { + "start": 1410, + "end": 1444 + } + }, + { + "target": "ba1346a8-84b1-57dd-aaea-715c9b461fc9", + "relation": "embeds", + "ref": "[tables/table-40 EVP Compensation.html]", + "position": { + "start": 1460, + "end": 1499 + } + }, + { + "target": "5c8fffa7-6ed3-5a74-bf1e-f7ce0266a2d3", + "relation": "embeds", + "ref": "[tables/table-41 EVP Compensation.html]", + "position": { + "start": 1518, + "end": 1557 + } + }, + { + "target": "d312e756-08e8-514c-93fb-4c2809072eac", + "relation": "embeds", + "ref": "[tables/table-42 Executive Pay.html]", + "position": { + "start": 1577, + "end": 1613 + } + } + ] + } + }, + { + "chunk_id": "da859402-631c-55b2-92b5-494980334367", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6f44a527-8a32-5fd8-8a44-11a0263f9de4", + "type": "text", + "content": "When evaluating potential perquisites for our executives, we consider many factors, including the cost to NVIDIA relative to the anticipated business benefit, perceived value to our executives, peer data, as well as corporate governance considerations.\nDue to the exceptionally high public profile and prominence of our CEO, and in accordance with the Board-approved, independently-assessed executive security program, NVIDIA provides Mr. Huang and his family members with comprehensive security protection. In Fiscal 2026, these security arrangements included (i) residential security, security during personal travel, consultation fees, and driver services, (ii) security monitoring, (iii) car expenses, and (iv) travel on private aircraft for all business and occasional personal travel, in each case due to a bona fide business-oriented security concern. The CC annually oversees the nature and cost of executive security measures.\nWe do not consider these security arrangements to be personal benefits because they arise from Mr. Huang's duties and responsibilities and are required by the Board's executive security program. Furthermore, we believe these arrangements and costs are reasonable, necessary and in the best interests of NVIDIA and its stockholders, as they enable Mr. Huang to focus on his duties and responsibilities to the Company while reducing security threats and mitigating risks to our business. However, for transparency, we have reported the aggregate incremental costs to NVIDIA for personal security arrangements in the “All Other Compensation” column of the Summary Compensation Table below.\nPrivate aircraft arranged by the Company is intended for Mr. Huang's business travel. Other employees, including our other NEOs, who are also traveling for the same business may accompany Mr. Huang on board. On those flights, we also allow Mr. Huang and any accompanying NVIDIA employee to bring guests. The flights taken by personal guests may be deemed to be a personal benefit for our NEOs, though at no or de minimis incremental cost to us. Nevertheless, any associated aggregate incremental costs incurred by NVIDIA are included in the “All Other Compensation” column of the Summary Compensation Table below. Our NEOs recognize imputed taxable income with respect to any such flights taken by their personal guests, and are not provided with tax gross-ups on such imputed income. To the extent Mr. Huang has personal travel on NVIDIA-arranged private aircraft, he fully reimburses the Company for associated incremental costs. As a result, we incur no incremental costs from Mr. Huang's personal air travel.\nWe also provide medical, vision, dental, and accidental death and disability insurance, matches for health savings account contributions, as well as paid time off and holidays, for our NEOs on the same basis as our other employees. Our NEOs may participate, also on the same basis as our other employees, in our ESPP (unless prohibited by Internal Revenue Service rules) and our 401(k) plan, which included an annual Company match of up to \\$11,500. We believe these benefits are consistent with those offered by our peer companies with whom we compete for executive talent.\nIn Fiscal 2026, we did not provide any other perquisites or personal benefits to our NEOs.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Other Compensation and Benefits", + "metadata": { + "length": 3301, + "summary": "NVIDIA evaluates executive perquisites based on cost, business benefit, peer data, and governance. The CEO receives comprehensive security including residential protection, travel security, and private aircraft for business and occasional personal travel due to security concerns. These are not considered personal benefits but are reported in 'All Other Compensation'. Private aircraft is for business travel; personal guests may accompany at no incremental cost, with imputed income recognized. The CEO reimburses personal air travel costs. NEOs receive standard benefits like insurance, 401(k) match up to $11,500, and ESPP participation. No other perquisites were provided in Fiscal 2026.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "When", + "evaluating", + "potential", + "perquisites", + "executives", + "factors", + "including", + "cost", + "NVIDIA", + "relative", + "anticipated", + "business", + "benefit", + "perceived", + "peer", + "data", + "corporate", + "governance", + "considerations", + "Due", + "exceptionally", + "high", + "public", + "profile", + "prominence", + "CEO", + "accordance", + "Board", + "approved", + "independently", + "assessed", + "executive", + "security", + "program", + "Mr", + "Huang", + "family", + "members", + "comprehensive", + "protection", + "In", + "Fiscal", + "2026", + "arrangements", + "included", + "residential", + "personal", + "travel", + "consultation", + "fees", + "driver", + "services", + "ii", + "monitoring", + "iii", + "car", + "expenses", + "iv", + "private", + "aircraft", + "occasional", + "case", + "due", + "bona", + "fide", + "oriented", + "concern", + "The", + "CC", + "annually", + "oversees", + "nature", + "measures", + "We", + "benefits", + "arise", + "duties", + "responsibilities", + "required", + "Furthermore", + "costs", + "reasonable", + "interests", + "stockholders", + "enable", + "focus", + "Company", + "reducing", + "threats", + "mitigating", + "risks", + "However", + "transparency", + "reported", + "aggregate", + "incremental", + "All", + "Other", + "Compensation", + "column", + "Summary", + "Table", + "Private", + "arranged", + "intended", + "employees", + "NEOs", + "traveling", + "accompany", + "board", + "On", + "flights", + "accompanying", + "employee", + "bring", + "guests", + "deemed", + "de", + "minimis", + "Nevertheless", + "incurred", + "Our", + "recognize", + "imputed", + "taxable", + "income", + "respect", + "provided", + "tax", + "gross", + "ups", + "To", + "extent", + "fully", + "reimburses", + "As", + "result", + "incur", + "air", + "provide", + "medical", + "vision", + "dental", + "accidental", + "death", + "disability", + "insurance", + "matches", + "health", + "savings", + "account", + "contributions", + "paid", + "time", + "holidays", + "basis", + "participate", + "ESPP", + "prohibited", + "Internal", + "Revenue", + "Service", + "rules", + "401", + "plan", + "annual", + "match", + "11", + "500", + "consistent", + "offered", + "companies", + "compete", + "talent" + ], + "keywords": [ + "executive security", + "private aircraft", + "compensation" + ], + "connect_to": [] + } + }, + { + "chunk_id": "f9376389-e7c7-54c7-8e7d-8bf8e608e1b6", + "type": "text", + "content": "The CC approves all equity award grants to our executive officers on or before the grant date. Typically, the CC completes its annual executive compensation review, sets performance goals and target compensation, and grants equity awards in March. This process is described in How We Determine Executive Compensation above. On occasion, the CC may grant equity awards outside of our annual cycle for new hires, promotions, recognition, retention or other purposes. While the CC has discretionary authority to approve such awards, it does not have a practice or policy of granting equity awards in anticipation of the release of material non-public information. In Fiscal 2026, we did not time the release of material non-public information in coordination with equity award grants in a manner that intentionally benefits the value of our executive compensation.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Equity Grant Timing Practices", + "metadata": { + "length": 861, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "CC", + "approves", + "equity", + "award", + "grants", + "executive", + "officers", + "grant", + "date", + "Typically", + "completes", + "annual", + "compensation", + "review", + "sets", + "performance", + "goals", + "target", + "awards", + "March", + "This", + "process", + "How", + "We", + "Determine", + "Executive", + "Compensation", + "On", + "occasion", + "cycle", + "hires", + "promotions", + "recognition", + "retention", + "purposes", + "While", + "discretionary", + "authority", + "approve", + "practice", + "policy", + "granting", + "anticipation", + "release", + "material", + "public", + "information", + "In", + "Fiscal", + "2026", + "time", + "coordination", + "manner", + "intentionally", + "benefits" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1151da28-3252-5c8a-8402-5d2f3cc904f3", + "type": "text", + "content": "The Board believes that executive officers should hold a meaningful equity interest in NVIDIA. Since June 2024, our Corporate Governance Policies require the CEO to hold shares valued at ten times his base salary, and our other NEOs to hold shares valued at three times their respective base salaries. Shares that count toward the ownership guidelines include shares held in trust, and shares held by immediate family members, but not unvested equity awards. NEOs have up to five years from appointment to reach their ownership threshold. These guidelines are intended to align NEO interests with stockholder interests. Each NEO currently exceeds the stock ownership requirements.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Stock Ownership Guidelines", + "metadata": { + "length": 680, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Board", + "believes", + "executive", + "officers", + "hold", + "meaningful", + "equity", + "interest", + "NVIDIA", + "Since", + "June", + "2024", + "Corporate", + "Governance", + "Policies", + "require", + "CEO", + "shares", + "valued", + "ten", + "times", + "base", + "salary", + "NEOs", + "respective", + "salaries", + "Shares", + "count", + "ownership", + "guidelines", + "include", + "held", + "trust", + "family", + "members", + "unvested", + "awards", + "years", + "appointment", + "reach", + "threshold", + "These", + "intended", + "align", + "NEO", + "interests", + "stockholder", + "Each", + "exceeds", + "stock", + "requirements" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "234f7a07-5800-5826-b0e4-90999d871a59", + "type": "text", + "content": "We have maintained a Compensation Recovery Policy since 2009, and amended it in 2023 to comply with Nasdaq's listing standards. Our policy requires the Company to recover certain incentive compensation from current or former executive officers in connection with certain restatements of financial statements, if the compensation exceeds the amount that the officers would have received based on the restated financial statements, subject to limited exceptions.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Compensation Recovery Policy", + "metadata": { + "length": 460, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "maintained", + "Compensation", + "Recovery", + "Policy", + "2009", + "amended", + "2023", + "comply", + "Nasdaq", + "listing", + "standards", + "Our", + "policy", + "requires", + "Company", + "recover", + "incentive", + "compensation", + "current", + "executive", + "officers", + "connection", + "restatements", + "financial", + "statements", + "exceeds", + "amount", + "received", + "based", + "restated", + "subject", + "limited", + "exceptions" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "06c62cf5-3044-526c-9f39-b2e40eeb071b", + "type": "text", + "content": "Under Section 162(m), compensation paid to each of the Company's “covered employees” that exceeds \\$1 million per year is generally non-deductible. The CC considers various factors when making its decisions and retains flexibility to provide non-deductible compensation if consistent with the Company’s executive compensation program goals and the best interests of the Company and its stockholders. The CC may also modify compensation that was initially intended to be exempt from the deduction limit if the modifications are consistent with the Company’s business needs. The CC also considers the impact of Section 409A of the Internal Revenue Code and strives to avoid possible adverse tax consequences.\nAdditionally, under ASC 718, the Company records share-based compensation expense for equity compensation over the vesting period of the award.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Tax and Accounting Implications", + "metadata": { + "length": 850, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Under", + "Section", + "162", + "compensation", + "paid", + "Company", + "covered", + "employees", + "exceeds", + "million", + "year", + "generally", + "deductible", + "The", + "CC", + "considers", + "factors", + "making", + "decisions", + "retains", + "flexibility", + "provide", + "consistent", + "executive", + "program", + "goals", + "interests", + "stockholders", + "modify", + "initially", + "intended", + "exempt", + "deduction", + "limit", + "modifications", + "business", + "impact", + "409A", + "Internal", + "Revenue", + "Code", + "strives", + "avoid", + "adverse", + "tax", + "consequences", + "Additionally", + "ASC", + "718", + "records", + "share", + "based", + "expense", + "equity", + "vesting", + "period", + "award" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "52aff8a3-02f8-5a4a-931e-bfdfe544b8e0", + "type": "table", + "content": "
Fiscal 2026Fiscal 2025
GAAP operating income$130,387$81,453
Stock-based compensation expense6,3864,737
Acquisition-related and other costs471602
Other56(3)
Non-GAAP Operating Income$137,300$86,789
", + "path": "tables/table-43 Non-GAAP Income.html", + "metadata": { + "length": 422, + "summary": "table-44\nThe table shows GAAP operating income rising from $81,453 in Fiscal 2025 to $130,387 in Fiscal 2026, with adjustments for stock-based compensation, acquisition costs, and other items, resulting in Non-GAAP Operating Income of $86,789 and $137,300 respectively.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-43 Non-GAAP Income.html", + "keywords": [ + "Operating Income", + "Stock Compensation", + "Acquisition Costs" + ], + "tokens": [] + } + }, + { + "chunk_id": "dd2094ad-7ac5-5e42-af83-a9c265a1f2bf", + "type": "text", + "content": "A reconciliation between our GAAP operating income and non-GAAP operating income is as follows (in millions):\n\n[tables/table-43 Non-GAAP Income.html]\n\nWe believe these non-GAAP financial measures enhance stockholders' overall understanding of our historical financial performance. The presentation of our non-GAAP financial measures is not meant to be considered in isolation nor as a substitute for our financial results prepared in accordance with GAAP, and our non-GAAP financial measures may be different from non-GAAP financial measures used by other companies.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Reconciliation of Non-GAAP Financial Measures", + "metadata": { + "length": 574, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "reconciliation", + "GAAP", + "operating", + "income", + "millions", + "We", + "financial", + "measures", + "enhance", + "stockholders", + "understanding", + "historical", + "performance", + "The", + "presentation", + "meant", + "considered", + "isolation", + "substitute", + "results", + "prepared", + "accordance", + "companies" + ], + "keywords": [], + "connect_to": [ + { + "target": "52aff8a3-02f8-5a4a-931e-bfdfe544b8e0", + "relation": "embeds", + "ref": "[tables/table-43 Non-GAAP Income.html]", + "position": { + "start": 111, + "end": 149 + } + } + ] + } + }, + { + "chunk_id": "cba61d4b-bbd7-555e-8131-19b2b860bf73", + "type": "text", + "content": "The Compensation Committee of the Board of Directors oversees the compensation programs of NVIDIA on behalf of the Board of Directors. In fulfilling its oversight responsibilities, the Compensation Committee reviewed and discussed with management the Compensation Discussion and Analysis included in this proxy statement.\nIn reliance on the review and discussions referred to above, the Compensation Committee recommended to the Board of Directors that the Compensation Discussion and Analysis be included in the Annual Report on Form 10-K of NVIDIA for the year ended January 25, 2026 and in this proxy statement.\nCompensation Committee\nDawn Hudson, Tench Coxe, John O. Dabiri, Aarti Shah and Mark Stevens", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Committee Report", + "metadata": { + "length": 706, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Compensation", + "Committee", + "Board", + "Directors", + "oversees", + "compensation", + "programs", + "NVIDIA", + "behalf", + "In", + "fulfilling", + "oversight", + "responsibilities", + "reviewed", + "discussed", + "management", + "Discussion", + "Analysis", + "included", + "proxy", + "statement", + "reliance", + "review", + "discussions", + "referred", + "recommended", + "Annual", + "Report", + "Form", + "10", + "year", + "ended", + "January", + "25", + "2026", + "Dawn", + "Hudson", + "Tench", + "Coxe", + "John", + "Dabiri", + "Aarti", + "Shah", + "Mark", + "Stevens" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "505e2586-b62f-54cd-a11f-52326f853864", + "type": "text", + "content": "Under CC oversight, management assessed the Company's Fiscal 2026 overall compensation programs and policies to identify any resulting potential material risks. The assessment focused on programs with payout variability and participants' ability to directly affect payout and related controls—specifically, the Company's variable cash compensation, equity compensation, and sales incentive compensation programs. The review identified key program terms and risks, and specific risk mitigation features.\nThe CC considered these findings and concluded that our compensation programs, which are structured to recognize both short-term and long-term contributions to the Company, do not create risks which are reasonably likely to have a material adverse effect on our business or financial condition.\nThe CC believes that the following compensation design features mitigate against risk:\n√ Our compensation program encourages our employees to remain focused on both our short-term and long-term goals, and balances incentives by using a mix of base salary and variable pay\n√ We design our variable cash and PSU compensation programs for executives so that payouts are based on achievement of various corporate performance targets to balance upside opportunity and downside risk, and we cap the potential award payout\n√ We have internal controls over our financial accounting and reporting which are used to measure and determine the eligible compensation awards under our Variable Cash Plan and our SY PSUs\n√ Financial plan target goals and final awards under our Variable Cash Plan and SY PSUs are approved by the CC\n√ MY PSUs are designed with a relative goal\n√ We have a compensation recovery policy applicable to executive officers that requires NVIDIA to recover certain incentive compensation paid in connection with certain accounting restatements\n√ The CC monitors burn rate\n√ All executive officer equity awards have multi-year vesting\n√ We have stock ownership guidelines that we believe are reasonable and are designed to align our executive officers' interests with those of our stockholders\n√ Our insider trading policy prohibits hedging, pledging, using margin accounts, and trading in derivatives involving our common stock, which prevents our employees from insulating themselves from the effects of NVIDIA stock price performance", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Risk Analysis of Our Compensation Plans", + "metadata": { + "length": 2343, + "summary": "The Compensation Committee assessed the company's fiscal 2026 compensation programs for potential material risks, focusing on variable cash, equity, and sales incentive plans. It concluded that the programs, which balance short-term and long-term contributions, do not create risks likely to materially harm the business. Key risk mitigation features include a mix of base salary and variable pay, performance-based targets with caps, internal controls, committee approval, relative goals for multi-year PSUs, a compensation recovery policy, burn rate monitoring, multi-year vesting, stock ownership guidelines, and an insider trading policy prohibiting hedging and pledging.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Under", + "CC", + "oversight", + "management", + "assessed", + "Company", + "Fiscal", + "2026", + "compensation", + "programs", + "policies", + "identify", + "resulting", + "potential", + "material", + "risks", + "The", + "assessment", + "focused", + "payout", + "variability", + "participants", + "ability", + "directly", + "affect", + "related", + "controls", + "specifically", + "variable", + "cash", + "equity", + "sales", + "incentive", + "review", + "identified", + "key", + "program", + "terms", + "specific", + "risk", + "mitigation", + "features", + "considered", + "findings", + "concluded", + "structured", + "recognize", + "short", + "term", + "long", + "contributions", + "create", + "adverse", + "effect", + "business", + "financial", + "condition", + "believes", + "design", + "mitigate", + "Our", + "encourages", + "employees", + "remain", + "goals", + "balances", + "incentives", + "mix", + "base", + "salary", + "pay", + "We", + "PSU", + "executives", + "payouts", + "based", + "achievement", + "corporate", + "performance", + "targets", + "balance", + "upside", + "opportunity", + "downside", + "cap", + "award", + "internal", + "accounting", + "reporting", + "measure", + "determine", + "eligible", + "awards", + "Variable", + "Cash", + "Plan", + "SY", + "PSUs", + "Financial", + "plan", + "target", + "final", + "approved", + "MY", + "designed", + "relative", + "goal", + "recovery", + "policy", + "applicable", + "executive", + "officers", + "requires", + "NVIDIA", + "recover", + "paid", + "connection", + "restatements", + "monitors", + "burn", + "rate", + "All", + "officer", + "multi", + "year", + "vesting", + "stock", + "ownership", + "guidelines", + "reasonable", + "align", + "interests", + "stockholders", + "insider", + "trading", + "prohibits", + "hedging", + "pledging", + "margin", + "accounts", + "derivatives", + "involving", + "common", + "prevents", + "insulating", + "effects", + "price" + ], + "keywords": [ + "compensation", + "risk mitigation", + "executive" + ], + "connect_to": [] + } + }, + { + "chunk_id": "3d9ec4e5-676c-5ad5-8c5a-6b7f385d5b23", + "type": "table", + "content": "
Name and Principal PositionFiscal YearSalary ($)Stock Awards ($) (1)Non-Equity Incentive Plan Compensation ($) (2)All Other Compensation ($)Total ($) (3)
Jen-Hsun HuangPresident and CEO20261,497,62724,800,5116,000,0004,045,691 (4)36,343,830
20251,486,19938,811,3066,000,0003,568,74649,866,251
2024996,51426,676,4154,000,0002,494,97334,167,902
Colette M. KressEVP and CFO2026898,57712,825,872600,00016,402 (5)14,340,850
2025893,73919,849,891600,00018,90221,362,532
2024896,86311,756,027600,00013,90213,266,792
Ajay K. PuriEVP, Worldwide Field Operations2026948,49712,478,9891,300,00049,840 (5)14,777,327
2025943,39119,277,0461,300,00070,46021,590,897
2024946,68911,320,3531,300,00048,40813,615,450
Debora ShoquistEVP, Operations2026848,65612,912,487500,00033,567 (5)14,294,710
2025844,08717,838,832500,00034,98419,217,903
2024847,0379,687,599500,00024,22911,058,865
Timothy S. TeterEVP, General Counsel and Secretary2026848,65612,912,487500,00015,240 (5)14,276,382
2025844,08717,838,832500,00018,90219,201,821
2024847,0379,687,599500,00013,90211,048,538
", + "path": "tables/table-44 Executive Compensation.html", + "metadata": { + "length": 2174, + "summary": "table-45\nThe table shows compensation for NVIDIA executives including CEO Jen-Hsun Huang, CFO Colette Kress, and others for fiscal years 2024-2026, with salary, stock awards, and total pay.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-44 Executive Compensation.html", + "keywords": [ + "Compensation", + "Stock Awards", + "CEO" + ], + "tokens": [] + } + }, + { + "chunk_id": "08a4759e-f885-5bbc-911e-ff94f765f1d3", + "type": "table", + "content": "
Fiscal YearJen-Hsun HuangColette M. KressAjay K. PuriDebora ShoquistTimothy S. Teter
SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)
202619,166,42418,034,3436,099,9937,350,4835,935,2617,151,9816,141,4437,400,4306,141,4437,400,430
202525,046,74833,170,2118,377,78114,564,6908,136,36614,144,9937,529,33213,089,6697,529,33213,089,669
202423,130,93722,666,27015,613,1881,937,35015,034,8111,865,42312,866,3001,596,47612,866,3001,596,476
", + "path": "tables/table-45 PSU Awards Table.html", + "metadata": { + "length": 1053, + "summary": "table-46\nThe table shows Performance Stock Unit (PSU) awards for five executives across fiscal years 2024-2026, with SY and MY PSU values in dollars.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-45 PSU Awards Table.html", + "keywords": [ + "PSU", + "executive compensation", + "fiscal years" + ], + "tokens": [] + } + }, + { + "chunk_id": "d14cbf5a-cd57-5823-b61b-8ad314be987a", + "type": "text", + "content": "The following table summarizes information regarding the compensation earned by our NEOs during Fiscal 2026, 2025, and 2024. Fiscal 2026, 2025, and 2024 were 52-week years.\n\n[tables/table-44 Executive Compensation.html]\n\n(1) Amounts shown do not reflect dollar amounts actually received by the NEO, and may not reconcile to corresponding amounts presented in other tables due to differences in rounding. Instead, these amounts reflect the aggregate full grant date fair value calculated in accordance with ASC 718 for the respective fiscal year for grants of RSUs, SY PSUs, and MY PSUs, as applicable. The assumptions used in the calculation of values of the awards are set forth under Note 3 to our consolidated financial statements titled Stock-Based Compensation in our Form 10-K. The reported grant date fair value stock awards with performance-based vesting conditions assumes the probable outcome of the conditions at Base Compensation Plan for SY PSUs and MY PSUs, determined in accordance with applicable accounting standards.\nAssuming Stretch Compensation Plan performance for SY PSUs and MY PSUs granted in each of Fiscal 2026, 2025, and 2024, and a stock price equal to the grant date fair value of the SY PSUs and MY PSUs, the value of stock awards would be:\n\n[tables/table-45 PSU Awards Table.html]\n\n(2) As applicable, reflects amounts earned in Fiscal 2026, 2025, and 2024 and paid in March or April of the fiscal year immediately following each respective year pursuant to the respective Variable Cash Plan. For further information, please see our CD&A above.\n(3) Due to rounding, numbers presented may not sum precisely to the totals provided.\n(4) Reflects (a) the aggregate incremental costs to the Company of residential security, security during personal travel, and consultation fees and driver services (in the amount of \\$3,975,533, reflecting the full cost to the Company), security monitoring services, car expenses, and Mr. Huang's personal guests who accompanied him during business travel on private aircraft, though at no or de minimis incremental cost to us, (b) a \\$11,500 match of contributions to our 401(k) savings plan, and (c) \\$21,904 in life insurance premiums. The 401(k) contribution match and insurance coverage are available to all eligible NVIDIA employees.\n(5) Includes matches of contributions to our 401(k) savings plan and imputed income from life insurance coverage. These benefits are available to all eligible NVIDIA employees. For Fiscal 2026, the match of 401(k) contributions was \\$11,500 for Ms. Kress, \\$10,708 for Mr. Puri, \\$10,083 for Ms. Shoquist, and \\$10,083 for Mr. Teter; and the dollar values of life insurance premiums were \\$4,902 for Ms. Kress, \\$39,132 for Mr. Puri, \\$23,484 for Ms. Shoquist, and \\$5,121 for Mr. Teter.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Summary Compensation Table for Fiscal 2026, 2025, and 2024", + "metadata": { + "length": 2909, + "summary": "The table summarizes compensation for NEOs in Fiscal 2026, 2025, and 2024. It includes salary, stock awards (grant date fair value per ASC 718), and non-equity incentive plan compensation. Notes explain that stock award values assume probable performance outcomes, with stretch scenario values provided in a separate table. Additional details cover cash plan payouts, rounding, and other compensation items like security, 401(k) matches, and life insurance premiums for each NEO.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "table", + "summarizes", + "information", + "compensation", + "earned", + "NEOs", + "Fiscal", + "2026", + "2025", + "2024", + "52", + "week", + "years", + "Amounts", + "shown", + "reflect", + "dollar", + "amounts", + "received", + "NEO", + "reconcile", + "presented", + "tables", + "due", + "differences", + "rounding", + "Instead", + "aggregate", + "full", + "grant", + "date", + "fair", + "calculated", + "accordance", + "ASC", + "718", + "respective", + "fiscal", + "year", + "grants", + "RSUs", + "SY", + "PSUs", + "MY", + "applicable", + "assumptions", + "calculation", + "values", + "awards", + "set", + "Note", + "consolidated", + "financial", + "statements", + "titled", + "Stock", + "Based", + "Compensation", + "Form", + "10", + "reported", + "stock", + "performance", + "based", + "vesting", + "conditions", + "assumes", + "probable", + "outcome", + "Base", + "Plan", + "determined", + "accounting", + "standards", + "Assuming", + "Stretch", + "granted", + "price", + "equal", + "As", + "reflects", + "paid", + "March", + "April", + "immediately", + "pursuant", + "Variable", + "Cash", + "For", + "CD", + "Due", + "numbers", + "sum", + "precisely", + "totals", + "provided", + "Reflects", + "incremental", + "costs", + "Company", + "residential", + "security", + "personal", + "travel", + "consultation", + "fees", + "driver", + "services", + "amount", + "975", + "533", + "reflecting", + "cost", + "monitoring", + "car", + "expenses", + "Mr", + "Huang", + "guests", + "accompanied", + "business", + "private", + "aircraft", + "de", + "minimis", + "11", + "500", + "match", + "contributions", + "401", + "savings", + "plan", + "21", + "904", + "life", + "insurance", + "premiums", + "contribution", + "coverage", + "eligible", + "NVIDIA", + "employees", + "Includes", + "matches", + "imputed", + "income", + "These", + "benefits", + "Ms", + "Kress", + "708", + "Puri", + "083", + "Shoquist", + "Teter", + "902", + "39", + "132", + "23", + "484", + "121" + ], + "keywords": [ + "NEO compensation", + "stock awards", + "fiscal years" + ], + "connect_to": [ + { + "target": "3d9ec4e5-676c-5ad5-8c5a-6b7f385d5b23", + "relation": "embeds", + "ref": "[tables/table-44 Executive Compensation.html]", + "position": { + "start": 174, + "end": 219 + } + }, + { + "target": "08a4759e-f885-5bbc-911e-ff94f765f1d3", + "relation": "embeds", + "ref": "[tables/table-45 PSU Awards Table.html]", + "position": { + "start": 1272, + "end": 1311 + } + } + ] + } + }, + { + "chunk_id": "60e36216-21ba-537b-8a7a-979bbe71ca6a", + "type": "table", + "content": "
NameType of AwardGrant DateEstimated Possible Payouts Under Non-Equity Incentive Plan Awards (1)Estimated Future Payouts Under Equity Incentive Plan Awards (2)All Other Stock Awards: Number of Shares of Stock or Units (#) (3)Grant Date Fair Value of Stock Awards ($) (4)
Threshold ($)Target ($)Maximum ($)Threshold (#)Target (#)Maximum (#)
Jen-Hsun HuangSY PSU3/10/2559,804119,607179,41112,777,616
MY PSU3/10/2529,902119,607179,41112,022,896
Variable Cash Plan3/3/251,500,0003,000,0006,000,000
Colette M. KressSY PSU3/10/2514,27528,55057,1003,049,997
MY PSU3/10/257,13828,55057,1003,675,242
RSU3/10/2557,1066,100,634
Variable Cash Plan3/3/25150,000300,000600,000
Ajay K. PuriSY PSU3/10/2513,89027,77955,5582,967,631
MY PSU3/10/256,94527,77955,5583,575,991
RSU3/10/2555,5595,935,368
Variable Cash Plan3/3/25325,000650,0001,300,000
Debora ShoquistSY PSU3/10/2514,37228,74457,4883,070,722
MY PSU3/10/257,18628,74457,4883,700,215
RSU3/10/2557,4896,141,550
Variable Cash Plan3/3/25125,000250,000500,000
Timothy S. TeterSY PSU3/10/2514,37228,74457,4883,070,722
MY PSU3/10/257,18628,74457,4883,700,215
", + "path": "tables/table-46 Executive Compensation.html", + "metadata": { + "length": 3136, + "summary": "table-47\nThe table details 2025 executive compensation for five Nvidia executives, including Jen-Hsun Huang, with awards like SY PSU, MY PSU, RSU, and Variable Cash Plan, showing estimated payouts and stock values.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-46 Executive Compensation.html", + "keywords": [ + "PSU", + "RSU", + "Variable Cash" + ], + "tokens": [] + } + }, + { + "chunk_id": "bb612be1-b121-5a1a-9b72-17e99e670f10", + "type": "table", + "content": "
NameStock Awards
Number of Units of Stock That Have Not Vested (#)Market Value of Units of Stock That Have Not Vested ($) (1)Equity Incentive Plan Awards: Number of Unearned Shares That Have Not Vested (#)Equity Incentive Plan Awards: Market Value of Unearned Shares That Have Not Vested ($) (1)
Jen-Hsun Huang315,570 (2)59,223,022
757,360 (3)142,133,751
161,085 (4)30,230,822
179,411 (5)33,670,062
286,350 (6)53,739,305
179,411 (7)33,670,062
Colette M. Kress10,970 (8)2,058,740
61,970 (9)11,629,910
213,010 (2)39,975,587
49,560 (3)9,300,925
53,890 (10)10,113,536
53,880 (4)10,111,660
46,399 (11)8,707,700
57,100 (5)10,715,957
95,780 (6)17,975,033
57,100 (7)10,715,957
Ajay K. Puri10,560 (8)1,981,795
59,670 (9)11,198,269
205,120 (2)38,494,870
47,720 (3)8,955,612
52,330 (10)9,820,771
52,340 (4)9,822,648
45,142 (11)8,471,799
55,558 (5)10,426,570
93,020 (6)17,457,063
55,558 (7)10,426,570
Debora Shoquist9,040 (8)1,696,537
51,070 (9)9,584,307
175,540 (2)32,943,592
40,840 (3)7,664,443
48,420 (10)9,086,981
48,420 (4)9,086,981
46,710 (11)8,766,066
57,488 (5)10,788,773
86,080 (6)16,154,634
57,488 (7)10,788,773
Timothy S. Teter9,040 (8)1,696,537
51,070 (9)9,584,307
175,540 (2)32,943,592
40,840 (3)7,664, 443
48,420 (10)9,086,981
48,420 (4)9,086,981
46,710 (11)8,766,066
57,48B (5)10,788,773
86,080 (6)16,154,634
57,488 (7)10,788,773
", + "path": "tables/table-47 Stock Awards.html", + "metadata": { + "length": 3667, + "summary": "table-48\nThe table details unvested stock awards for executives including Jen-Hsun Huang, Colette M. Kress, Ajay K. Puri, Debora Shoquist, and Timothy S. Teter, listing numbers of units and their market values.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-47 Stock Awards.html", + "keywords": [ + "Stock Awards", + "Unvested Units", + "Market Value" + ], + "tokens": [] + } + }, + { + "chunk_id": "415c9586-3cbd-5589-ae79-58059189e076", + "type": "text", + "content": "The following table provides information regarding all grants of plan-based awards that were made to or earned by our NEOs during Fiscal 2026. The information in this table supplements the dollar value of stock awards set forth in the Summary Compensation Table for Fiscal 2026, 2025, and 2024. The PSU and RSU awards set forth in the following table were made under our 2007 Plan. PSUs are eligible to vest based on performance against pre-established criteria. All equity awards listed are subject to service-based vesting.\n\n[tables/table-46 Executive Compensation.html]\n\n(1) Represents range of awards payable under our Fiscal 2026 Variable Cash Plan.\n(2) Represents range of shares eligible to be earned with respect to PSUs.\n(3) Represents RSUs granted.\n(4) Amounts shown in this column do not reflect dollar amounts actually received by the NEO, and may not reconcile to corresponding amounts presented in other tables due to differences in rounding. Instead, these amounts reflect the aggregate full grant date fair value calculated in accordance with ASC 718 for the awards. The assumptions used in the calculation of values of the awards are set forth under Note 3 to our consolidated financial statements titled Stock-Based Compensation in our Form 10-K. With regard to the stock awards with performance-based vesting conditions, the reported grant date fair value assumes the probable outcome of the conditions at Base Compensation Plan performance for SY PSUs and MY PSUs, determined in accordance with applicable accounting standards.\nThe following table presents information regarding outstanding equity awards held by our NEOs as of January 25, 2026.\n\n[tables/table-47 Stock Awards.html]\n\n(1) Calculated by multiplying the number of RSUs or PSUs that have not vested or have not been earned, as applicable, by the closing price (\\$187.67) of NVIDIA's common stock on January 23, 2026, the last trading day before the end of our Fiscal 2026, as reported by Nasdaq.\n(2) The PSU was earned based on achievement of a performance goal for the one-year period that ended on January 28, 2024. The PSU vested as to 25% of the shares on March 20, 2024, and vests as to 6.25% approximately every three months thereafter over the next three years such that the PSU will be fully vested on March 17, 2027.\n(3) The PSU was earned based on achievement of a performance goal for the three-year period that ended on January 25, 2026. The PSU vested as to 100% of the shares on March 18, 2026.\n(4) The PSU was earned based on achievement of a performance goal for the one-year period that ended on January 26, 2025. The PSU vested as to 25% of the shares on March 19, 2025, and vests as to 6.25% approximately every three months thereafter over the next three years such that the PSU will be fully vested on March 15, 2028.\n(5) The PSU was earned based on achievement of a performance goal for the 1-year period that ended on January 25, 2026. The PSU vested as to 25% of the shares on March 18, 2026, and vests as to 6.25% approximately every three months thereafter over the next three years such that the PSU will be fully vested on March 15, 2029.\n(6) Represents shares that could be earned upon achievement of a Stretch Compensation Plan goal, based on our TSR relative to the S&P 500 from January 29, 2024 through January 31, 2027. If the performance goal is achieved, 100% of the shares earned will vest on March 17, 2027. If the Threshold performance goal is achieved, 47,720 shares will be earned by Mr. Huang, 11,970 shares will be earned by Ms. Kress, 11,620 shares will be earned by Mr. Puri, 10,760 shares will be earned by Ms. Shoquist, and 10,760 shares will be earned by Mr. Teter. If the Base Compensation Plan performance goal is achieved, 190,900 shares will be earned by Mr. Huang, 47,890 shares will be earned by Ms. Kress, 46,510 shares will be earned by Mr. Puri, 43,040 shares will be earned by Ms. Shoquist, and 43,040 shares will be earned by Mr. Teter.\n(7) Represents shares that could be earned upon achievement of a Stretch Compensation Plan goal, based on our TSR relative to the S&P 500 from January 27, 2025 through January 30, 2028. If the performance goal is achieved, 100% of the shares earned will vest on March 15, 2028. If the Threshold performance goal is achieved, 29,902 shares will be earned by Mr. Huang, 7,138 shares will be earned by Ms. Kress, 6,945 shares will be earned by Mr. Puri, 7,186 shares will be earned by Ms. Shoquist, and 7,186 shares will be earned by Mr. Teter. If the Base Compensation Plan performance goal is achieved, 119,607 shares will be earned by Mr. Huang, 28,550 shares will be earned by Ms. Kress, 27,779 shares will be earned by Mr. Puri, 28,744 shares will be earned by Ms. Shoquist, and 28,744 shares will be earned by Mr. Teter.\n(8) The RSU vested as to 6.25% on June 15, 2022, and approximately every three months thereafter, such that the shares were fully vested on March 18, 2026, the approximate four year anniversary of the grant date.\n(9) The RSU vested as to 6.25% on June 21, 2023, and approximately every three months thereafter, such that the shares will become fully vested on March 17, 2027, the approximate four year anniversary of the grant date.\n(10) The RSU vested as to 6.25% on June 19, 2024, and approximately every three months thereafter, such that the shares will become fully vested on March 15, 2028, the approximate four year anniversary of the grant date.\n(11) The RSU vested as to 6.25% on June 18, 2025, and approximately every three months thereafter, such that the shares will become fully vested on March 15, 2029, the approximate four year anniversary of the grant date.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Grants of Plan-Based Awards for Fiscal 2026", + "metadata": { + "length": 5739, + "summary": "The text details plan-based awards for NEOs in Fiscal 2026, including PSUs and RSUs under the 2007 Plan. It explains vesting schedules, performance criteria, and grant date fair values. Outstanding equity awards as of January 25, 2026 are listed, with values calculated using a closing stock price of $187.67. Specific vesting terms for PSUs and RSUs are provided, including stretch goals based on TSR relative to the S&P 500.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "table", + "information", + "grants", + "plan", + "based", + "awards", + "made", + "earned", + "NEOs", + "Fiscal", + "2026", + "supplements", + "dollar", + "stock", + "set", + "Summary", + "Compensation", + "Table", + "2025", + "2024", + "PSU", + "RSU", + "2007", + "Plan", + "PSUs", + "eligible", + "vest", + "performance", + "pre", + "established", + "criteria", + "All", + "equity", + "listed", + "subject", + "service", + "vesting", + "Represents", + "range", + "payable", + "Variable", + "Cash", + "shares", + "respect", + "RSUs", + "granted", + "Amounts", + "shown", + "column", + "reflect", + "amounts", + "received", + "NEO", + "reconcile", + "presented", + "tables", + "due", + "differences", + "rounding", + "Instead", + "aggregate", + "full", + "grant", + "date", + "fair", + "calculated", + "accordance", + "ASC", + "718", + "assumptions", + "calculation", + "values", + "Note", + "consolidated", + "financial", + "statements", + "titled", + "Stock", + "Based", + "Form", + "10", + "With", + "regard", + "conditions", + "reported", + "assumes", + "probable", + "outcome", + "Base", + "SY", + "MY", + "determined", + "applicable", + "accounting", + "standards", + "presents", + "outstanding", + "held", + "January", + "25", + "Calculated", + "multiplying", + "number", + "vested", + "closing", + "price", + "187.67", + "NVIDIA", + "common", + "23", + "trading", + "day", + "end", + "Nasdaq", + "achievement", + "goal", + "year", + "period", + "ended", + "28", + "25%", + "March", + "20", + "vests", + "6.25%", + "approximately", + "months", + "years", + "fully", + "17", + "2027", + "100%", + "18", + "26", + "19", + "15", + "2028", + "2029", + "Stretch", + "TSR", + "relative", + "500", + "29", + "31", + "If", + "achieved", + "Threshold", + "47", + "720", + "Mr", + "Huang", + "11", + "970", + "Ms", + "Kress", + "620", + "Puri", + "760", + "Shoquist", + "Teter", + "190", + "900", + "890", + "46", + "510", + "43", + "040", + "27", + "30", + "902", + "138", + "945", + "186", + "119", + "607", + "550", + "779", + "744", + "June", + "2022", + "approximate", + "anniversary", + "21", + "2023" + ], + "keywords": [ + "PSU", + "RSU", + "Stock Awards" + ], + "connect_to": [ + { + "target": "60e36216-21ba-537b-8a7a-979bbe71ca6a", + "relation": "embeds", + "ref": "[tables/table-46 Executive Compensation.html]", + "position": { + "start": 527, + "end": 572 + } + }, + { + "target": "bb612be1-b121-5a1a-9b72-17e99e670f10", + "relation": "embeds", + "ref": "[tables/table-47 Stock Awards.html]", + "position": { + "start": 1667, + "end": 1702 + } + } + ] + } + }, + { + "chunk_id": "01650da9-f69f-5b09-8d50-f473c79a41e0", + "type": "table", + "content": "
NameNumber of Shares Acquired on Vesting (#) (1)Value Realized on Vesting ($) (2)
Jen-Hsun Huang1,113,555 (3)141,311,217
Colette M. Kress442,087 (4)63,881,560
Ajay K. Puri425,837 (5)61,550,468
Debora Shoquist368,879 (6)53,412,792
Timothy S. Teter368,879 (7)53,412,792
", + "path": "tables/table-48 Stock Vesting Data.html", + "metadata": { + "length": 493, + "summary": "table-49\nThe table shows stock vesting details for five NVIDIA executives, including Jen-Hsun Huang, with shares acquired and value realized upon vesting.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-48 Stock Vesting Data.html", + "keywords": [ + "Stock Vesting", + "Shares", + "Value" + ], + "tokens": [] + } + }, + { + "chunk_id": "174d6f1d-fcd0-5b35-993a-a27fe71397da", + "type": "text", + "content": "The following table shows information regarding stock acquired upon vesting for our NEOs during Fiscal 2026. We have not had any stock options outstanding since the end of Fiscal 2024.\nStock Awards\n\n[tables/table-48 Stock Vesting Data.html]\n\n(1) Represents the gross number of shares acquired on vesting. Shares were withheld from these amounts to pay taxes due upon vesting.\n(2) Represents the gross number of shares acquired on vesting multiplied by the fair market value of our common stock as reported by Nasdaq on the date of vesting.\n(3) Includes an aggregate of 552,105 shares that were withheld to pay taxes due upon vesting.\n(4) Includes an aggregate of 233.268 shares that were withheld to pay taxes due upon vesting.\n(5) Includes an aggregate of 211,139 shares that were withheld to pay taxes due upon vesting.\n(6) Includes an aggregate of 194,424 shares that were withheld to pay taxes due upon vesting.\n(7) Includes an aggregate of 198,486 shares that were withheld to pay taxes due upon vesting.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Option Exercises and Stock Vested in Fiscal 2026", + "metadata": { + "length": 1043, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "table", + "shows", + "information", + "stock", + "acquired", + "vesting", + "NEOs", + "Fiscal", + "2026", + "We", + "options", + "outstanding", + "end", + "2024", + "Stock", + "Awards", + "Represents", + "gross", + "number", + "shares", + "Shares", + "withheld", + "amounts", + "pay", + "taxes", + "due", + "multiplied", + "fair", + "market", + "common", + "reported", + "Nasdaq", + "date", + "Includes", + "aggregate", + "552", + "105", + "233.268", + "211", + "139", + "194", + "424", + "198", + "486" + ], + "keywords": [], + "connect_to": [ + { + "target": "01650da9-f69f-5b09-8d50-f473c79a41e0", + "relation": "embeds", + "ref": "[tables/table-48 Stock Vesting Data.html]", + "position": { + "start": 199, + "end": 240 + } + } + ] + } + }, + { + "chunk_id": "6c637dda-1764-5f8d-a86e-d05cf5866fa5", + "type": "table", + "content": "
NameUnvested RSUs and PSUs at January 25, 2026 (#) (1)Total Estimated Benefit ($) (1)
Jen-Hsun Huang1,411,679264,929,798
Colette M. Kress569,889106,951,069
Ajay K. Puri551,090103,423,060
Debora Shoquist500,14893,862,775
Timothy S. Teter500,14893,862,775
", + "path": "tables/table-49 Unvested RSUs_PSUs.html", + "metadata": { + "length": 479, + "summary": "table-50\nTable shows unvested RSUs and PSUs for five NVIDIA executives as of January 25, 2026, with Jen-Hsun Huang holding the largest amount.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-49 Unvested RSUs_PSUs.html", + "keywords": [ + "RSUs", + "PSUs", + "Executives" + ], + "tokens": [] + } + }, + { + "chunk_id": "8e5209ee-bc6d-5486-8813-17a36002e185", + "type": "table", + "content": "
NameEstimated SY PSUs Granted in Fiscal 2026 at Base Compensation Plan Performance (#)Value of Estimated SY PSUs Granted in Fiscal 2026 at Base Compensation Plan Performance ($)Actual SY PSUs Granted in Fiscal 2026 Eligible to Vest (#)Value of Actual SY PSUs Granted in Fiscal 2026 Eligible to Vest ($)
Jen-Hsun Huang119,60722,446,646179,41133,670,062
Colette M. Kress28,5505,357,97957,10010,715,957
Ajay K. Puri27,7795,213,28555,55810,426,570
Debora Shoquist28,7445,394,38657,48810,788,773
Timothy S. Teter28,7445,394,38657,48810,788,773
", + "path": "tables/table-50 SY PSU Grants.html", + "metadata": { + "length": 872, + "summary": "table-51\nTable shows estimated and actual SY PSU grants for fiscal 2026 for five executives, including Jen-Hsun Huang, with values in shares and dollars.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-50 SY PSU Grants.html", + "keywords": [ + "PSU", + "Grants", + "Fiscal 2026" + ], + "tokens": [] + } + }, + { + "chunk_id": "987ebafe-f0b2-5005-85fb-95b4b2860265", + "type": "table", + "content": "
NameEstimated MY PSUs Granted in Fiscal 2024 at Base Compensation Plan Performance (#)Value of Estimated MY PSUs Granted in Fiscal 2024 at Base Compensation Plan Performance ($)Actual MY PSUs Granted in Fiscal 2024 Eligible to Vest (#)Value of Actual MY PSUs Granted in Fiscal 2024 Eligible to Vest ($)
Jen-Hsun Huang504,91094,756,460757,360142,133,751
Colette M. Kress24,7804,650,46349,5609,300,925
Ajay K. Puri23,8604,477,80647,7208,955,612
Debora Shoquist20,4203,832,22140,8407,664,443
Timothy S. Teter20,4203,832,22140,8407,664,443
", + "path": "tables/table-51 MY PSU Grants 2024.html", + "metadata": { + "length": 869, + "summary": "table-52\nTable shows estimated and actual MY PSU grants for fiscal 2024 for five NVIDIA executives, including Jen-Hsun Huang with the highest values.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-51 MY PSU Grants 2024.html", + "keywords": [ + "PSU grants", + "NVIDIA executives", + "fiscal 2024" + ], + "tokens": [] + } + }, + { + "chunk_id": "d378f0a5-590a-5466-9bdc-993b3c6dccca", + "type": "text", + "content": "Employment Agreements. Our executive officers are “at-will” employees and we do not have employment, severance, or change-in-control agreements with our executive officers.\nChange-in-Control Arrangements. Our 2007 Plan provides that in the event of a corporate transaction or a change-in-control, outstanding stock awards may be assumed, continued, or substituted by the surviving corporation. If the surviving corporation does not assume, continue, or substitute such stock awards, then (a) with respect to any stock awards that are held by individuals performing services for NVIDIA immediately prior to the effective time of the transaction, the vesting and exercisability provisions of such stock awards will be accelerated in full and such stock awards will be terminated if not exercised prior to the effective date of the corporate transaction or change-in-control, and (b) all other outstanding stock awards will be terminated if not exercised on or prior to the effective date of the corporate transaction or change-in-control. These change-in-control arrangements apply to stock awards held by our NEOs on the same basis as our other employees.\nPotential Payments Upon Termination or Change-in-Control\nUpon a change-in-control or certain other corporate transactions of NVIDIA, unvested RSUs, and PSUs will fully vest in some cases as described above under Employment, Severance, and Change-in-Control Arrangements—Change-in-Control Arrangements. The table below shows our estimates of the amount of the benefit each of our NEOs would have received if the unvested RSUs and PSUs held by them as of January 25, 2026 had become fully vested as a result of a change-in-control, calculated by multiplying the number of unvested RSUs and PSUs held by the applicable NEO by the closing price (\\$187.67) of NVIDIA's common stock on January 23, 2026, the last trading day before the end of our Fiscal 2026, as reported by Nasdaq.\n\n[tables/table-49 Unvested RSUs_PSUs.html]\n\n(1) With respect to unvested PSUs, the amounts in these columns assume performance at Base Compensation Plan with respect to SY PSUs granted in Fiscal 2026 and with respect to MY PSUs granted in Fiscal 2024, Fiscal 2025, and Fiscal 2026, in accordance with SEC rules. The two tables below reflect the actual numbers of the SY PSUs granted in Fiscal 2026 and MY PSUs granted in Fiscal 2024 that became eligible to vest, based on our performance during the relevant performance period for such awards, as certified by our CC shortly after the end of Fiscal 2026. The values of the estimated and actual SY PSUs and MY PSUs in the table below were calculated by multiplying the applicable number of SY PSUs and MY PSUs held by each respective NEO and listed below, by the closing price (\\$187.67) of NVIDIA's common stock on January 23, 2026, the last trading day before the end of our Fiscal 2026, as reported by Nasdaq.\nSY PSUs granted in Fiscal 2026 - Actual Achievement (versus Base Compensation Plan Performance)\n\n[tables/table-50 SY PSU Grants.html]\n\nMY PSUs granted in Fiscal 2024 - Actual Achievement (versus Base Compensation Plan Performance)\n\n[tables/table-51 MY PSU Grants 2024.html]\n\nThe actual number of MY PSUs granted in Fiscal 2025 and Fiscal 2026 that will become eligible to vest will be determinable after January 31, 2027 and January 30, 2028, respectively, the ending dates of the applicable three-year measurement period for MY PSUs.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Employment, Severance, and Change-in-Control Arrangements", + "metadata": { + "length": 3591, + "summary": "The passage describes NVIDIA's change-in-control arrangements for executive officers. Under the 2007 Plan, upon a corporate transaction or change-in-control, outstanding stock awards may be assumed or substituted. If not, awards held by service providers accelerate vesting fully, while others terminate if unexercised. Potential payments upon termination or change-in-control are estimated based on unvested RSUs and PSUs, using a stock price of $187.67. Tables show estimated and actual PSU achievements for Fiscal 2024 and 2026, with further determinations pending for later grants.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Employment", + "Agreements", + "Our", + "executive", + "officers", + "employees", + "employment", + "severance", + "change", + "control", + "agreements", + "Change", + "Control", + "Arrangements", + "2007", + "Plan", + "event", + "corporate", + "transaction", + "outstanding", + "stock", + "awards", + "assumed", + "continued", + "substituted", + "surviving", + "corporation", + "If", + "assume", + "continue", + "substitute", + "respect", + "held", + "individuals", + "performing", + "services", + "NVIDIA", + "immediately", + "prior", + "effective", + "time", + "vesting", + "exercisability", + "provisions", + "accelerated", + "full", + "terminated", + "exercised", + "date", + "These", + "arrangements", + "apply", + "NEOs", + "basis", + "Potential", + "Payments", + "Upon", + "Termination", + "transactions", + "unvested", + "RSUs", + "PSUs", + "fully", + "vest", + "cases", + "Severance", + "The", + "table", + "shows", + "estimates", + "amount", + "benefit", + "received", + "January", + "25", + "2026", + "vested", + "result", + "calculated", + "multiplying", + "number", + "applicable", + "NEO", + "closing", + "price", + "187.67", + "common", + "23", + "trading", + "day", + "end", + "Fiscal", + "reported", + "Nasdaq", + "With", + "amounts", + "columns", + "performance", + "Base", + "Compensation", + "SY", + "granted", + "MY", + "2024", + "2025", + "accordance", + "SEC", + "rules", + "tables", + "reflect", + "actual", + "numbers", + "eligible", + "based", + "relevant", + "period", + "certified", + "CC", + "shortly", + "values", + "estimated", + "respective", + "listed", + "Actual", + "Achievement", + "versus", + "Performance", + "determinable", + "31", + "2027", + "30", + "2028", + "ending", + "dates", + "year", + "measurement" + ], + "keywords": [ + "Change-in-Control", + "Stock Awards", + "Executive Officers" + ], + "connect_to": [ + { + "target": "6c637dda-1764-5f8d-a86e-d05cf5866fa5", + "relation": "embeds", + "ref": "[tables/table-49 Unvested RSUs_PSUs.html]", + "position": { + "start": 1933, + "end": 1974 + } + }, + { + "target": "8e5209ee-bc6d-5486-8813-17a36002e185", + "relation": "embeds", + "ref": "[tables/table-50 SY PSU Grants.html]", + "position": { + "start": 2991, + "end": 3027 + } + }, + { + "target": "987ebafe-f0b2-5005-85fb-95b4b2860265", + "relation": "embeds", + "ref": "[tables/table-51 MY PSU Grants 2024.html]", + "position": { + "start": 3126, + "end": 3167 + } + } + ] + } + }, + { + "chunk_id": "8d7fd60a-f673-5db9-9cf4-73464c04f54c", + "type": "text", + "content": "We determined the ratio of: (a) the annual total compensation of our CEO to (b) the median of the annual total compensation of all our employees, except for our CEO, both calculated in accordance with the requirements of Item 402(c)(2)(x) of Regulation S-K.\nBecause there have been no changes to our employee population or employee compensation arrangements since the end of Fiscal 2024 that would significantly impact our pay ratio disclosure for Fiscal 2026, we used the median employee from our Fiscal 2024 pay ratio calculation in our Fiscal 2026 pay ratio calculation. We determined our Fiscal 2024 median employee by using a consistently applied compensation measure, which aggregated, for each employee employed by us on the last day of Fiscal 2024, or January 28, 2024: (i) target base salary as of January 28, 2024 (annualized for permanent employees who were employed by us for less than the entire fiscal year), (ii) variable cash earned during Fiscal 2024, and (iii) aggregate full grant date fair value of equity awards granted during Fiscal 2024, calculated in accordance with ASC 718 and assuming the probable outcome of the conditions at Base Compensation Plan for performance-based awards. Compensation paid in foreign currencies was converted to U.S. dollars based on exchange rates in effect on January 28, 2024.\nOur median employee's total compensation for Fiscal 2026 was \\$282,050. Our CEO's Fiscal 2026 total compensation was \\$36,343,830. Therefore, our Fiscal 2026 CEO to median employee pay ratio was 129:1.\nThis pay ratio represents a reasonable estimate of the relationship between the compensation of our CEO and that of our median employee for Fiscal 2026, calculated in a manner consistent with Item 402(u) of Regulation S-K and applicable guidance, which provide significant flexibility in how companies identify the median employee. Each company may use a different methodology, apply different exclusions, and make different assumptions. As a result, the pay ratio reported by other companies may not be comparable to ours.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Pay Ratio", + "metadata": { + "length": 2057, + "summary": "The company calculated its CEO to median employee pay ratio for Fiscal 2026 as 129:1, using the same median employee identified in Fiscal 2024 due to no significant changes. The median employee's total compensation was $282,050, and the CEO's total compensation was $36,343,830. The methodology followed SEC regulations and used a consistently applied compensation measure including base salary, variable cash, and equity awards.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "determined", + "ratio", + "annual", + "total", + "compensation", + "CEO", + "median", + "employees", + "calculated", + "accordance", + "requirements", + "Item", + "402", + "Regulation", + "Because", + "employee", + "population", + "arrangements", + "end", + "Fiscal", + "2024", + "significantly", + "impact", + "pay", + "disclosure", + "2026", + "calculation", + "consistently", + "applied", + "measure", + "aggregated", + "employed", + "day", + "January", + "28", + "target", + "base", + "salary", + "annualized", + "permanent", + "entire", + "fiscal", + "year", + "ii", + "variable", + "cash", + "earned", + "iii", + "aggregate", + "full", + "grant", + "date", + "fair", + "equity", + "awards", + "granted", + "ASC", + "718", + "assuming", + "probable", + "outcome", + "conditions", + "Base", + "Compensation", + "Plan", + "performance", + "based", + "paid", + "foreign", + "currencies", + "converted", + "dollars", + "exchange", + "rates", + "effect", + "Our", + "282", + "050", + "36", + "343", + "830", + "Therefore", + "129", + "This", + "represents", + "reasonable", + "estimate", + "relationship", + "manner", + "consistent", + "applicable", + "guidance", + "provide", + "significant", + "flexibility", + "companies", + "identify", + "Each", + "company", + "methodology", + "apply", + "exclusions", + "make", + "assumptions", + "As", + "result", + "reported", + "comparable" + ], + "keywords": [ + "CEO pay ratio", + "median employee", + "Fiscal 2026" + ], + "connect_to": [] + } + }, + { + "chunk_id": "ab26581a-3fd2-5f78-8c64-18ab2f0c2a6a", + "type": "table", + "content": "
Fiscal YearSummary Compensation Table Total for CEO ($) (1) (2)Compensation Actually Paid to CEO ($) (1) (3)Average Summary Compensation Table Total for Non-CEO NEOs ($) (4) (5)Average Compensation Actually Paid to Non-CEO NEOs ($) (4) (6)Value of Initial Fixed $100 Investment Based on (7):Net Income (in millions) ($)Non-GAAP Operating Income (in millions) ($) (9)
Total Shareholder Return ($)Peer Group Total Shareholder Return ($) (8)
202636,343,830162,180,93614,422,31757,721,5521,445.67198.10120,067137,300
202549,866,251344,188,02720,343,288125,583,9131,100.69168.4672,88086,789
202434,167,902234,132,30512,247,41185,558,057470.88134.7829,76037,134
202321,356,924(4,118,947)9,941,838(1,364,661)157.0594.134,3689,040
202223,737,661105,543,7688,910,80238,453,071175.98111.839,75212,690
", + "path": "tables/table-52 CEO Pay & Performance.html", + "metadata": { + "length": 1453, + "summary": "table-53\nTable shows CEO and non-CEO NEO compensation, total shareholder return, peer group return, net income, and non-GAAP operating income for fiscal years 2022-2026.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-52 CEO Pay & Performance.html", + "keywords": [ + "CEO compensation", + "financial performance", + "shareholder return" + ], + "tokens": [] + } + }, + { + "chunk_id": "975ca78d-b084-5ca8-8ae2-227f84a6dd95", + "type": "table", + "content": "
Fiscal YearSummary Compensation Table Total for CEO ($)(Deduct):Equity Award AdjustmentsTotal Equity Award Adjustments ($) (b) (c)Compensation Actually Paid to CEO ($)
Add:Add/(Deduct):Add/(Deduct):
Value of Equity Awards Reported in Summary Compensation Table ($) (a)Year End Fair Value of Awards Granted During the Year which were Unvested at Year End ($) (b)Year Over Year Change in Fair Value of Outstanding and Unvested Awards ($) (b)Change in Fair Value of Awards Granted in Prior Years which Vested During the Year ($) (b)
202636,343,830(24,800,511)60,909,207105,767,276(16,038,866)150,637,617162,180,936
", + "path": "tables/table-53 CEO Compensation.html", + "metadata": { + "length": 920, + "summary": "table-54\nThe table shows CEO compensation for fiscal year 2026, with a total of $36,343,830 in the Summary Compensation Table, adjustments totaling $150,637,617, and compensation actually paid of $162,180,936.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-53 CEO Compensation.html", + "keywords": [ + "CEO", + "Equity Awards", + "Compensation" + ], + "tokens": [] + } + }, + { + "chunk_id": "ab7bcdb5-2f33-5afe-8e35-319de390e1f5", + "type": "table", + "content": "
Fiscal YearAverage Summary Compensation Table Total for Non-CEO NEOs ($)(Deduct):Equity Award AdjustmentsTotal Equity Award Adjustments ($) (b) (c)Average Compensation Actually Paid to Non-CEO NEOs ($)
Add:Add/(Deduct):Add:Add/(Deduct):
Value of Equity Awards Reported in Summary Compensation Table ($) (a)Year End Fair Value of Awards Granted During the Year which were Unvested at Year End ($) (b)Year Over Year Change in Fair Value of Outstanding and Unvested Awards ($) (b)Vesting Date Fair Value of Awards Granted and Vested During the Year ($) (b)Change in Fair Value of Awards Granted in Prior Years which Vested During the Year ($) (b)
202614,422,317(12,782,459)27,759,18025,940,0531,776,820605,64056,081,69457,721,552
", + "path": "tables/table-54 Non-CEO NEO Pay.html", + "metadata": { + "length": 1062, + "summary": "table-55\nThe table shows average compensation for non-CEO NEOs in 2026: Summary Table total $14,422,317, equity adjustments $56,081,694, resulting in actual pay of $57,721,552.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-54 Non-CEO NEO Pay.html", + "keywords": [ + "Compensation", + "Equity Awards", + "NEOs" + ], + "tokens": [] + } + }, + { + "chunk_id": "01cd50ac-8071-5a93-8fa0-c013b1cbfa97", + "type": "text", + "content": "NVIDIA's executive compensation program is guided by a pay for performance philosophy and is designed to align NEO pay with our stockholders' interests. Accordingly, a substantial portion of our NEOs' total compensation is based on the Company's performance under certain corporate financial metric goals, which for Fiscal 2026 included annual revenue, annual Non-GAAP Operating Income, and 3-year TSR relative to the S&P 500.\nAmounts set forth below as “compensation actually paid,” or CAP, have been calculated in accordance with Item 402(v) of Regulation S-K under the Securities Act, and do not represent the value of compensation actually paid to or received by our NEOs. For a discussion of our executive compensation program and philosophy, refer to our CD&A above.\nThe following table summarizes information regarding compensation for our NEOs, including CAP as well as certain financial performance metrics, during Fiscal 2022-2026. Fiscal 2022-2026 were 52-week years.\n\n[tables/table-52 CEO Pay & Performance.html]\n\n(1) For Fiscal 2022-2026, our CEO was Jen-Hsun Huang.\n(2) The amounts in this column correspond with total compensation for our CEO as reported in our Summary Compensation Table above for the listed fiscal years.\n(3) The amounts in this column represent CAP calculated in accordance with Item 402(v) of Regulation S-K during the listed fiscal years. Adjustments to the Summary Compensation Table total compensation for our CEO to arrive at CAP for Fiscal 2026 were as follows:\nReconciliation of Summary Compensation Table Total Compensation for CEO to CAP\n\n[tables/table-53 CEO Compensation.html]\n\n(a) The amount in this column corresponds with the full grant date fair value, calculated in accordance with ASC 718, of “Stock Awards” for our CEO as reported in our Summary Compensation Table above for Fiscal 2026.\n(b) The equity award adjustments were calculated in accordance with the SEC methodology for determining CAP for each year shown. The amounts in these columns were determined by reference to (i) for MY PSU awards where the performance period was complete as of or prior to the applicable year end date and for SY PSU awards, the closing price of our common stock on the applicable year end date, as reduced by the present value of dividends expected to be paid on the underlying shares during the requisite service period, or the closing price of our common stock on the applicable vesting dates, and (ii) for MY PSU awards where the performance period was not yet complete as of the applicable year end date, the fair value as calculated by a Monte Carlo simulation model as of the respective year end date, for the listed fiscal years.\n(c) The valuation assumptions used to calculate fair values did not materially differ from those disclosed at the time of grant.\n(4) For Fiscal 2022-2026, our non-CEO NEOs were Colette M. Kress, Ajay K. Puri, Debora Shoquist, and Timothy S. Teter.\n(5) The amounts in this column correspond with the average of the total compensation for our non-CEO NEOs as reported in our Summary Compensation Table above for the listed fiscal years.\n(6) The amounts in this column represent average CAP calculated in accordance with Item 402(v) of Regulation S-K during the listed fiscal years. Adjustments to the Summary Compensation Table average total compensation for our non-CEO NEOs to arrive at average CAP for Fiscal 2026 were as follows:\nReconciliation of Summary Compensation Table Average Total Compensation for Non-CEO NEOs to CAP\n\n[tables/table-54 Non-CEO NEO Pay.html]\n\n(a) The amount in this column corresponds with the average of the full grant date fair value, calculated in accordance with ASC 718, of “Stock Awards” for our non-CEO NEOs as reported in our Summary Compensation Table above for Fiscal 2026.\n(b) The equity award adjustments were calculated in accordance with the SEC methodology for determining CAP for each year shown. The amounts in these columns were determined by reference to (i) for MY PSU awards where the performance period was complete as of or prior to the applicable year end date, for RSU awards and for SY PSU awards, the closing price of our common stock on the applicable year end date, as reduced by the present value of dividends expected to be paid on the underlying shares during the requisite service period, or the closing price of our common stock on the applicable vesting dates, and (ii) for MY PSU awards where the performance period was not yet complete as of the applicable year end date, the fair value as calculated by a Monte Carlo simulation model as of the respective year end date, for the listed fiscal years. Due to rounding, numbers presented may not sum precisely to the total provided.\n(c) The valuation assumptions used to calculate fair values did not materially differ from those disclosed at the time of grant.\n(7) TSR for each of the years during Fiscal 2022-2026 is cumulative, reflecting the value of a fixed \\$100 investment beginning with the market close on January 29, 2021, the last trading day before our Fiscal 2022, through and including the end of the respective listed fiscal years.\n(8) The Nasdaq 100 Index is the industry peer group we use for purposes of Item 201(e) of Regulation S-K. The separate peer group referenced by the CC for purposes of determining executive compensation is discussed above in CD&A.\n(9) Our Company-Selected Measure, as required by Item 402(v) of Regulation S-K, is Non-GAAP Operating Income, which, in our assessment, represents the most important financial performance measure linking Fiscal 2026 NEO CAP to company performance. See Definitions above for a definition of Non-GAAP Operating Income, and see Reconciliation of Non-GAAP Financial Measures above in CD&A for a reconciliation between GAAP operating income and non-GAAP Operating Income.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Pay Versus Performance", + "metadata": { + "length": 6019, + "summary": "NVIDIA's executive compensation program aligns NEO pay with stockholder interests through pay for performance, using metrics like annual revenue, Non-GAAP Operating Income, and 3-year TSR relative to S&P 500. The text explains that 'compensation actually paid' (CAP) is calculated per SEC rules and does not reflect actual compensation received. It references tables summarizing compensation for CEO Jen-Hsun Huang and non-CEO NEOs (Colette M. Kress, Ajay K. Puri, Debora Shoquist, Timothy S. Teter) for Fiscal 2022-2026, including reconciliations from Summary Compensation Table totals to CAP. Key adjustments involve equity award fair values based on stock price or Monte Carlo simulations. TSR is cumulative from a fixed $100 investment, and the Nasdaq 100 Index is the peer group. The company-selected measure is Non-GAAP Operating Income, deemed the most important financial metric linking CAP to performance.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "NVIDIA", + "executive", + "compensation", + "program", + "guided", + "pay", + "performance", + "philosophy", + "designed", + "align", + "NEO", + "stockholders", + "interests", + "Accordingly", + "substantial", + "portion", + "NEOs", + "total", + "based", + "Company", + "corporate", + "financial", + "metric", + "goals", + "Fiscal", + "2026", + "included", + "annual", + "revenue", + "Non", + "GAAP", + "Operating", + "Income", + "year", + "TSR", + "relative", + "500", + "Amounts", + "set", + "paid", + "CAP", + "calculated", + "accordance", + "Item", + "402", + "Regulation", + "Securities", + "Act", + "represent", + "received", + "For", + "discussion", + "refer", + "CD", + "The", + "table", + "summarizes", + "information", + "including", + "metrics", + "2022", + "52", + "week", + "years", + "CEO", + "Jen", + "Hsun", + "Huang", + "amounts", + "column", + "correspond", + "reported", + "Summary", + "Compensation", + "Table", + "listed", + "fiscal", + "Adjustments", + "arrive", + "Reconciliation", + "Total", + "amount", + "corresponds", + "full", + "grant", + "date", + "fair", + "ASC", + "718", + "Stock", + "Awards", + "equity", + "award", + "adjustments", + "SEC", + "methodology", + "determining", + "shown", + "columns", + "determined", + "reference", + "MY", + "PSU", + "awards", + "period", + "complete", + "prior", + "applicable", + "end", + "SY", + "closing", + "price", + "common", + "stock", + "reduced", + "present", + "dividends", + "expected", + "underlying", + "shares", + "requisite", + "service", + "vesting", + "dates", + "ii", + "Monte", + "Carlo", + "simulation", + "model", + "respective", + "valuation", + "assumptions", + "calculate", + "values", + "materially", + "differ", + "disclosed", + "time", + "Colette", + "Kress", + "Ajay", + "Puri", + "Debora", + "Shoquist", + "Timothy", + "Teter", + "average", + "Average", + "RSU", + "Due", + "rounding", + "numbers", + "presented", + "sum", + "precisely", + "provided", + "cumulative", + "reflecting", + "fixed", + "100", + "investment", + "beginning", + "market", + "close", + "January", + "29", + "2021", + "trading", + "day", + "Nasdaq", + "Index", + "industry", + "peer", + "group", + "purposes", + "201", + "separate", + "referenced", + "CC", + "discussed", + "Our", + "Selected", + "Measure", + "required", + "assessment", + "represents", + "important", + "measure", + "linking", + "company", + "See", + "Definitions", + "definition", + "Financial", + "Measures", + "reconciliation", + "operating", + "income" + ], + "keywords": [ + "Pay for Performance", + "CAP", + "Non-GAAP Operating Income" + ], + "connect_to": [ + { + "target": "ab26581a-3fd2-5f78-8c64-18ab2f0c2a6a", + "relation": "embeds", + "ref": "[tables/table-52 CEO Pay & Performance.html]", + "position": { + "start": 980, + "end": 1024 + } + }, + { + "target": "975ca78d-b084-5ca8-8ae2-227f84a6dd95", + "relation": "embeds", + "ref": "[tables/table-53 CEO Compensation.html]", + "position": { + "start": 1583, + "end": 1622 + } + }, + { + "target": "ab7bcdb5-2f33-5afe-8e35-319de390e1f5", + "relation": "embeds", + "ref": "[tables/table-54 Non-CEO NEO Pay.html]", + "position": { + "start": 3507, + "end": 3545 + } + } + ] + } + }, + { + "chunk_id": "973584aa-b010-58cf-bd77-8ab7443b93ee", + "type": "table", + "content": "
Revenue
Non-GAAP Operating Income
3-Year TSR relative to the S&P 500
", + "path": "tables/table-55 Financial Metrics.html", + "metadata": { + "length": 139, + "summary": "table-56\nThe table lists three key financial metrics: Revenue, Non-GAAP Operating Income, and 3-Year TSR relative to the S&P 500.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-55 Financial Metrics.html", + "keywords": [ + "Revenue", + "Operating Income", + "TSR" + ], + "tokens": [] + } + }, + { + "chunk_id": "3bafd74c-18a7-5102-be61-d423b9ac090c", + "type": "text", + "content": "The following table is an unranked list of the most important financial performance measures linking Fiscal 2026 NEO CAP to company performance:\nFinancial Measures\n\n[tables/table-55 Financial Metrics.html]\n\nRefer to CD&A above for a description of how each of these performance measures impacts NEO compensation.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Pay Versus Performance/Most Important Financial Performance Measures", + "metadata": { + "length": 302, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "table", + "unranked", + "list", + "important", + "financial", + "performance", + "measures", + "linking", + "Fiscal", + "2026", + "NEO", + "CAP", + "company", + "Financial", + "Measures", + "Refer", + "CD", + "description", + "impacts", + "compensation" + ], + "keywords": [], + "connect_to": [ + { + "target": "973584aa-b010-58cf-bd77-8ab7443b93ee", + "relation": "embeds", + "ref": "[tables/table-55 Financial Metrics.html]", + "position": { + "start": 165, + "end": 205 + } + } + ] + } + }, + { + "chunk_id": "b13127c9-c0e3-51e5-a810-e422898ecf4e", + "type": "image", + "content": "\nFinancial Measures
0\n[images/image-78-NEO CAP versus TSR.jpg]\n", + "path": "images/image-78-NEO CAP versus TSR.jpg", + "metadata": { + "length": 64, + "summary": "image-78\nChart Title: NEO CAP versus TSR\n\nData Series:\n- CEO CAP (Dark Grey Bars)\n- Other NEOs Average CAP (Light Grey Bars)\n- NVIDIA TSR (Green Line)\n- Nasdaq 100 Index TSR (Purple Line)\n\nTime Range: Fiscal Years 2022 through 2026\n\nKey Data Points (CAP in millions):\n- Fiscal 2022: CEO CAP $105.5; Other NEOs Average CAP $38.5\n- Fiscal 2023: CEO CAP ($4.1); Other NEOs Average CAP ($1.4)\n- Fiscal 2024: CEO CAP $234.1; Other NEOs Average CAP $85.6\n- Fiscal 2025: CEO CAP $344.2; Other NEOs Average CAP $125.6\n- Fiscal 2026: CEO CAP $162.2; Other NEOs Average CAP $57.7\n\nKey Data Points (TSR values):\n- Fiscal 20", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-78-NEO CAP versus TSR.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "8062c95c-5ddb-5101-895c-c6bf7fa98927", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
Fiscal YearCEO CAP ($ in millions)Other NEOs Average CAP ($ in millions)NVIDIA TSR ($)Nasdaq100 Index TSR ($)
Fiscal 2022105.538.5111.83
Fiscal 2023(4.1)157.0594.13
Fiscal 2024234.185.6470.88134.78
Fiscal 2025344.2125.61,100.69168.46
Fiscal 2026162.257.71,445.67198.10
", + "path": "tables/table-56 NVIDIA CEO Pay.html", + "metadata": { + "length": 971, + "summary": "table-57\nThe table shows CEO and other NEOs' CAP, NVIDIA TSR, and Nasdaq100 Index TSR from Fiscal 2022 to 2026. CEO CAP fluctuated, peaking at $344.2M in Fiscal 2025, while TSR grew significantly.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-56 NVIDIA CEO Pay.html", + "keywords": [ + "CEO CAP", + "TSR", + "Fiscal Years" + ], + "tokens": [] + } + }, + { + "chunk_id": "7b06a590-d8e0-5460-a3f0-904b80c0e5bd", + "type": "image", + "content": "\n
0\n[images/image-79-NEO CAP vs Income.jpg]\n", + "path": "images/image-79-NEO CAP vs Income.jpg", + "metadata": { + "length": 56, + "summary": "image-79\nChart Title: NEO CAP versus Net Income & Non-GAAP Operating Income\n\nData Series:\n- CEO CAP (Dark Grey Bars)\n- Other NEOs Average CAP (Light Grey Bars)\n- Net Income (Green Line)\n- Non-GAAP Operating Income (Purple Line)\n\nFiscal 2022 Data:\n- CEO CAP: $105.5 million\n- Other NEOs Average CAP: $38.5 million\n- Net Income: $9,752 million\n- Non-GAAP Operating Income: $12,690 million\n\nFiscal 2023 Data:\n- CEO CAP: ($4.1) million (Negative)\n- Other NEOs Average CAP: ($1.4) million (Negative)\n- Net Income: $4,368 million\n- Non-GAAP Operating Income: $9,040 million\n\nFiscal 2024 Data:\n- CEO CAP: $234.1 million\n- Other NEOs Average CAP: $85.6 million\n- Net Income: $29", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-79-NEO CAP vs Income.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "65562d3d-2479-5c17-81f9-47eb955199ac", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
Fiscal YearCEO CAP ($ million)Other NEOs Average CAP ($ million)Net Income ($ million)Non-GAAP Operating Income ($ million)
Fiscal 2022105.538.59,75212,690
Fiscal 2023(4.1)(4.1)4,3689,040
Fiscal 2024234.185.629,76037,134
Fiscal 2025344.2125.672,88086,789
Fiscal 2026162.257.7120,067137,300
", + "path": "tables/table-57 CEO Pay & Income.html", + "metadata": { + "length": 992, + "summary": "table-58\nThe table shows CEO CAP, other NEOs average CAP, Net Income, and Non-GAAP Operating Income from Fiscal 2022 to 2026, with notable increases in 2024-2025.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-57 CEO Pay & Income.html", + "keywords": [ + "CEO compensation", + "Net Income", + "Fiscal Years" + ], + "tokens": [] + } + }, + { + "chunk_id": "6ea3fab1-b121-5fa9-a3bd-ea4df85a9560", + "type": "text", + "content": "The following graphs illustrate how CAP for our NEOs aligns with the Company's financial performance measures as detailed in the Pay Versus Performance table above for each of Fiscal 2022, 2023, 2024, 2025, and 2026 as well as between the TSRs of NVIDIA and the Nasdaq100 Index, reflecting the value of a fixed \\$100 investment beginning with the market close on January 29, 2021, the last trading day before our Fiscal 2022, through and including the end of the respective listed fiscal years.\n\nFinancial Measures
0\n[images/image-78-NEO CAP versus TSR.jpg]\n\n
\nbar-line hybrid chart\nNEO CAP versus TSR\n\n[tables/table-56 NVIDIA CEO Pay.html]\n\n
\n\n
0\n[images/image-79-NEO CAP vs Income.jpg]\n\n
\nbar-line hybrid\nNEO CAP versus Net Income & Non-GAAP Operating Income\n\n[tables/table-57 CEO Pay & Income.html]\n\n
\nAll information provided above under the “Pay Versus Performance” heading will not be deemed to be incorporated by reference into any filing of the Company under the Securities Act or the Exchange Act, whether made before or after the date hereof and irrespective of any general incorporation language in any such filing, except to the extent the Company specifically incorporates such information by reference.\nWhat am I voting on? Ratification of the selection of PwC as our independent registered public accounting firm for Fiscal 2027.\nVote required for approval: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter.\nEffect of abstentions: Same as a vote AGAINST.\nEffect of broker non-votes: Not applicable (because this is a routine proposal, there are no broker non-votes).\nThe AC has selected PwC, which has audited our financial statements annually since 2004, to serve as our independent registered public accounting firm for Fiscal 2027. Our lead audit partner at PwC will serve no more than five consecutive years in that role. Stockholder ratification of the AC's selection of PwC is not required by our Bylaws. As a matter of good corporate governance, we are submitting the selection of PwC to our stockholders for ratification. If our stockholders do not ratify the selection, the AC will reconsider whether or not to retain PwC. Even if the selection is ratified, the AC in its sole discretion may direct the appointment of a different independent registered public accounting firm at any time during the fiscal year if it determines that such a change would be in our best interests and those of our stockholders. The AC believes it is in the best interests of NVIDIA and our stockholders to retain PwC.\nWe expect that a representative of PwC will attend the 2026 Meeting. The PwC representative will have an opportunity to make a statement at the 2026 Meeting if he or she so desires and will also be available to respond to appropriate stockholder questions.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Pay Versus Performance/Relationships Between Compensation Actually Paid and Financial Performance", + "metadata": { + "length": 3049, + "summary": "The passage discusses the ratification of PwC as NVIDIA's independent registered public accounting firm for Fiscal 2027, including vote requirements and the Audit Committee's recommendation. It also references graphs comparing NEO CAP with financial performance measures and TSRs.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "graphs", + "illustrate", + "CAP", + "NEOs", + "aligns", + "Company", + "financial", + "performance", + "measures", + "detailed", + "Pay", + "Versus", + "Performance", + "table", + "Fiscal", + "2022", + "2023", + "2024", + "2025", + "2026", + "TSRs", + "NVIDIA", + "Nasdaq100", + "Index", + "reflecting", + "fixed", + "100", + "investment", + "beginning", + "market", + "close", + "January", + "29", + "2021", + "trading", + "day", + "including", + "end", + "respective", + "listed", + "fiscal", + "years", + "Financial", + "Measures", + "details", + "summary", + "bar", + "line", + "hybrid", + "chart", + "NEO", + "versus", + "TSR", + "Net", + "Income", + "Non", + "GAAP", + "Operating", + "All", + "information", + "provided", + "heading", + "deemed", + "incorporated", + "reference", + "filing", + "Securities", + "Act", + "Exchange", + "made", + "date", + "hereof", + "irrespective", + "general", + "incorporation", + "language", + "extent", + "specifically", + "incorporates", + "What", + "voting", + "Ratification", + "selection", + "PwC", + "independent", + "registered", + "public", + "accounting", + "firm", + "2027", + "Vote", + "required", + "approval", + "majority", + "shares", + "present", + "person", + "represented", + "proxy", + "entitled", + "vote", + "matter", + "Effect", + "abstentions", + "Same", + "AGAINST", + "broker", + "votes", + "Not", + "applicable", + "routine", + "proposal", + "AC", + "selected", + "audited", + "statements", + "annually", + "2004", + "serve", + "Our", + "lead", + "audit", + "partner", + "consecutive", + "role", + "Stockholder", + "ratification", + "Bylaws", + "As", + "good", + "corporate", + "governance", + "submitting", + "stockholders", + "If", + "ratify", + "reconsider", + "retain", + "Even", + "ratified", + "sole", + "discretion", + "direct", + "appointment", + "time", + "year", + "determines", + "change", + "interests", + "believes", + "We", + "expect", + "representative", + "attend", + "Meeting", + "opportunity", + "make", + "statement", + "desires", + "respond", + "stockholder", + "questions" + ], + "keywords": [ + "PwC", + "Ratification", + "Auditor" + ], + "connect_to": [ + { + "target": "b13127c9-c0e3-51e5-a810-e422898ecf4e", + "relation": "embeds", + "ref": "[images/image-78-NEO CAP versus TSR.jpg]", + "position": { + "start": 526, + "end": 566 + } + }, + { + "target": "8062c95c-5ddb-5101-895c-c6bf7fa98927", + "relation": "embeds", + "ref": "[tables/table-56 NVIDIA CEO Pay.html]", + "position": { + "start": 639, + "end": 676 + } + }, + { + "target": "7b06a590-d8e0-5460-a3f0-904b80c0e5bd", + "relation": "embeds", + "ref": "[images/image-79-NEO CAP vs Income.jpg]", + "position": { + "start": 712, + "end": 751 + } + }, + { + "target": "65562d3d-2479-5c17-81f9-47eb955199ac", + "relation": "embeds", + "ref": "[tables/table-57 CEO Pay & Income.html]", + "position": { + "start": 853, + "end": 892 + } + } + ] + } + }, + { + "chunk_id": "d45cb4d3-c71b-5f8b-84d5-ca7462cbc5b8", + "type": "text", + "content": "The Board recommends that you vote FOR the ratification of the selection of PwC as our independent registered accounting firm for our fiscal year ending January 31, 2027.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Recommendation of the Board", + "metadata": { + "length": 170, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Board", + "recommends", + "vote", + "FOR", + "ratification", + "selection", + "PwC", + "independent", + "registered", + "accounting", + "firm", + "fiscal", + "year", + "ending", + "January", + "31", + "2027" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "92dcabf4-e175-55b1-a19e-731dc2b522d3", + "type": "table", + "content": "
Fiscal 2026Fiscal 2025
Audit Fees (1)$ 10,166,400$ 8,067,106
Audit Related Fees (2)1,716,820724,806
Tax Fees (3)1,076,103856,439
All Other Fees (4)401,999354,000
Total Fees$ 13,361,322$ 10,002,351
", + "path": "tables/table-58 Fee Breakdown.html", + "metadata": { + "length": 422, + "summary": "table-59\nThe table shows fiscal 2026 and 2025 fees: Audit $10.2M/$8.1M, Audit Related $1.7M/$0.7M, Tax $1.1M/$0.9M, Other $0.4M/$0.4M, Total $13.4M/$10.0M.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-58 Fee Breakdown.html", + "keywords": [ + "Audit Fees", + "Tax Fees", + "Total Fees" + ], + "tokens": [] + } + }, + { + "chunk_id": "c873c6fd-8d87-590c-8374-2a1e92313ba0", + "type": "text", + "content": "The following is a summary of fees billed or expected to be billed by PwC for Fiscal 2026, and fees billed by PwC for Fiscal 2025, for audit, audit related, tax, and other professional services during each fiscal year:\n\n[tables/table-58 Fee Breakdown.html]\n\n(1) For the audit of our consolidated financial statements, including the audit of our internal control over financial reporting, review of our quarterly financial statements and annual reports, and fees related to statutory audits of some of our international entities.\n(2) For a review of select sustainability metrics, system and organization controls reporting, accounting system pre-implementation controls review, and other attestation services.\n(3) For tax compliance, consulting, and tax audit defense services.\n(4) For products or services other than those referenced above, including a cybersecurity maturity assessment and subscription to accounting disclosure software.\nAll services provided for Fiscal 2026 and 2025 described above were pre-approved by the AC or the AC Chairperson through the authority granted to him by the AC, which is described below. Our AC determined that the rendering of services other than audit services by PwC was compatible with maintaining PwC's independence.\nPre-Approval Policies and Procedures\nThe AC has adopted policies and procedures for the pre-approval of all audit and permissible non-audit services rendered by our independent registered public accounting firm. The policy generally permits pre-approvals of specified permissible services in the defined categories of audit services, audit-related services, and tax services up to specified amounts. Pre-approval may also be given as part of the AC's approval of the scope of the engagement of our independent registered public accounting firm or on an individual case-by-case basis before the independent registered public accounting firm is engaged to provide each service. In some cases, the full AC provides pre-approval for up to a year related to a particular defined task or scope. In other cases, the AC has delegated power to the AC Chairperson to pre-approve additional audit and non-audit services if the need for the service was unanticipated and approval is required prior to the next scheduled meeting of the AC. Such pre-approval is then communicated to the full AC at its next meeting.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Fees Billed by the Independent Registered Public Accounting Firm", + "metadata": { + "length": 2372, + "summary": "The passage summarizes fees billed or expected by PwC for Fiscal 2026 and 2025, covering audit, audit-related, tax, and other professional services. It details categories: audit of consolidated financial statements and internal controls; review of sustainability metrics and attestation services; tax compliance and consulting; and other services like cybersecurity assessment. All services were pre-approved by the Audit Committee (AC) or its Chairperson. The AC has policies for pre-approving audit and permissible non-audit services, either by category, scope approval, or case-by-case, with delegated authority for unanticipated needs.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "summary", + "fees", + "billed", + "expected", + "PwC", + "Fiscal", + "2026", + "2025", + "audit", + "related", + "tax", + "professional", + "services", + "fiscal", + "year", + "For", + "consolidated", + "financial", + "statements", + "including", + "internal", + "control", + "reporting", + "review", + "quarterly", + "annual", + "reports", + "statutory", + "audits", + "international", + "entities", + "select", + "sustainability", + "metrics", + "system", + "organization", + "controls", + "accounting", + "pre", + "implementation", + "attestation", + "compliance", + "consulting", + "defense", + "products", + "referenced", + "cybersecurity", + "maturity", + "assessment", + "subscription", + "disclosure", + "software", + "All", + "provided", + "approved", + "AC", + "Chairperson", + "authority", + "granted", + "Our", + "determined", + "rendering", + "compatible", + "maintaining", + "independence", + "Pre", + "Approval", + "Policies", + "Procedures", + "adopted", + "policies", + "procedures", + "approval", + "permissible", + "rendered", + "independent", + "registered", + "public", + "firm", + "policy", + "generally", + "permits", + "approvals", + "defined", + "categories", + "amounts", + "part", + "scope", + "engagement", + "individual", + "case", + "basis", + "engaged", + "provide", + "service", + "In", + "cases", + "full", + "task", + "delegated", + "power", + "approve", + "additional", + "unanticipated", + "required", + "prior", + "scheduled", + "meeting", + "Such", + "communicated" + ], + "keywords": [ + "audit fees", + "non-audit services", + "pre-approval" + ], + "connect_to": [ + { + "target": "92dcabf4-e175-55b1-a19e-731dc2b522d3", + "relation": "embeds", + "ref": "[tables/table-58 Fee Breakdown.html]", + "position": { + "start": 220, + "end": 256 + } + } + ] + } + }, + { + "chunk_id": "3657fd45-7d4b-50c6-8361-ea7497787c9f", + "type": "text", + "content": "The material in this report is not “soliciting material,” is not deemed “filed” with the SEC, and is not to be incorporated by reference in any of our filings under the Securities Act or the Exchange Act, whether made before or after the date hereof and irrespective of any general incorporation language in any such filing, except to the extent specifically incorporated by reference therein.\nThe Audit Committee, or AC, oversees accounting, financial reporting, internal control over financial reporting, financial practices, and audit activities of NVIDIA and its subsidiaries. The AC reviews the results and scope of the audit and other services provided by the independent registered public accounting firm and reviews financial statements and the accounting policies followed by NVIDIA prior to the issuance of the financial statements with both management and the independent registered public accounting firm.\nManagement is responsible for the financial reporting process, the preparation of consolidated financial statements in accordance with accounting principles generally accepted in the United States, or GAAP, the system of internal control over financial reporting, and the procedures designed to facilitate compliance with accounting standards and applicable laws and regulations. PricewaterhouseCoopers LLP, or PwC, our independent registered public accounting firm for Fiscal 2026, was responsible for performing an independent audit of the consolidated financial statements and issuing a report on the consolidated financial statements and of the effectiveness of our internal control over financial reporting as of January 25, 2026. PwC's judgments as to the quality, not just the acceptability, of our accounting principles and such other matters are required to be disclosed to the AC under applicable standards. The AC oversees these processes. Also, the AC has ultimate authority and responsibility to select, evaluate and, when appropriate, terminate the independent registered public accounting firm. The AC approves audit fees and non-audit services provided by and fees paid to the independent registered public accounting firm.\nNVIDIA has an internal audit function that reports to the AC. This function is responsible for objectively reviewing and evaluating the adequacy, effectiveness, and quality of our system of internal controls and the operating effectiveness of our business processes. The AC approves an annual internal audit plan and monitors the activities and performance of our internal audit function throughout the year to ensure the plan objectives are carried out and met.\nThe AC members are not professional accountants or auditors, and their functions are not intended to duplicate or to certify the activities of management or the independent registered public accounting firm. The AC does not plan or conduct audits, determine that our financial statements are complete and accurate and in accordance with GAAP, or assess our internal control over financial reporting. The AC relies, without additional independent verification, on the information provided by our management and on the representations made by management that the financial statements have been prepared with integrity and objectivity, and the opinion of PwC that such financial statements have been prepared in conformity with GAAP.\nIn this context, the AC reviewed and discussed the audited consolidated financial statements for Fiscal 2026 with management and our internal control over financial reporting with management and PwC. Specifically, the AC discussed with PwC the matters required to be discussed by the applicable requirements of the Public Company Accounting Oversight Board and the SEC. We have received from PwC the written disclosures and letter required by the applicable requirements of the Public Company Accounting Oversight Board regarding PwC's communications with the AC concerning independence. The AC also considered whether the provision of certain permitted non-audit services by PwC is compatible with PwC's independence and discussed PwC's independence with PwC.\nBased on the AC's review and discussions, the AC recommended to the Board of Directors that the audited consolidated financial statements be included in the Annual Report on Form 10-K of NVIDIA for the fiscal year ended January 25, 2026.\nAudit Committee\nA. Brooke Seawell, Tench Coxe, Harvey C. Jones, Melissa B. Lora, and Aarti Shah", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Report of the Audit Committee of the Board of Directors", + "metadata": { + "length": 4446, + "summary": "The Audit Committee (AC) of NVIDIA oversees accounting, financial reporting, internal controls, and audit activities. Management prepares financial statements in accordance with GAAP, while PricewaterhouseCoopers (PwC) performs independent audits. The AC reviews audit results, approves fees, and monitors internal audit functions. It relies on management and PwC representations and recommended the audited financial statements for Fiscal 2026 to the Board.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "material", + "report", + "soliciting", + "deemed", + "filed", + "SEC", + "incorporated", + "reference", + "filings", + "Securities", + "Act", + "Exchange", + "made", + "date", + "hereof", + "irrespective", + "general", + "incorporation", + "language", + "filing", + "extent", + "specifically", + "Audit", + "Committee", + "AC", + "oversees", + "accounting", + "financial", + "reporting", + "internal", + "control", + "practices", + "audit", + "activities", + "NVIDIA", + "subsidiaries", + "reviews", + "results", + "scope", + "services", + "provided", + "independent", + "registered", + "public", + "firm", + "statements", + "policies", + "prior", + "issuance", + "management", + "Management", + "responsible", + "process", + "preparation", + "consolidated", + "accordance", + "principles", + "generally", + "accepted", + "United", + "States", + "GAAP", + "system", + "procedures", + "designed", + "facilitate", + "compliance", + "standards", + "applicable", + "laws", + "regulations", + "PricewaterhouseCoopers", + "LLP", + "PwC", + "Fiscal", + "2026", + "performing", + "issuing", + "effectiveness", + "January", + "25", + "judgments", + "quality", + "acceptability", + "matters", + "required", + "disclosed", + "processes", + "Also", + "ultimate", + "authority", + "responsibility", + "select", + "evaluate", + "terminate", + "approves", + "fees", + "paid", + "function", + "reports", + "This", + "objectively", + "reviewing", + "evaluating", + "adequacy", + "controls", + "operating", + "business", + "annual", + "plan", + "monitors", + "performance", + "year", + "ensure", + "objectives", + "carried", + "met", + "members", + "professional", + "accountants", + "auditors", + "functions", + "intended", + "duplicate", + "certify", + "conduct", + "audits", + "determine", + "complete", + "accurate", + "assess", + "relies", + "additional", + "verification", + "information", + "representations", + "prepared", + "integrity", + "objectivity", + "opinion", + "conformity", + "In", + "context", + "reviewed", + "discussed", + "audited", + "Specifically", + "requirements", + "Public", + "Company", + "Accounting", + "Oversight", + "Board", + "We", + "received", + "written", + "disclosures", + "letter", + "communications", + "independence", + "considered", + "provision", + "permitted", + "compatible", + "Based", + "review", + "discussions", + "recommended", + "Directors", + "included", + "Annual", + "Report", + "Form", + "10", + "fiscal", + "ended", + "Brooke", + "Seawell", + "Tench", + "Coxe", + "Harvey", + "Jones", + "Melissa", + "Lora", + "Aarti", + "Shah" + ], + "keywords": [ + "Audit Committee", + "Financial Oversight", + "Internal Controls" + ], + "connect_to": [] + } + }, + { + "chunk_id": "59e734e3-4109-5e04-aead-de929fc49176", + "type": "text", + "content": "We will promptly furnish the address and, to our knowledge, the number of shares held by the stockholder proponents of the proposals below upon receipt of a request to our Secretary for such information at email to shareholdermeeting@nvidia.com or by phone at: (408) 486-2000.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals", + "metadata": { + "length": 276, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "promptly", + "furnish", + "address", + "knowledge", + "number", + "shares", + "held", + "stockholder", + "proponents", + "proposals", + "receipt", + "request", + "Secretary", + "information", + "email", + "shareholdermeeting", + "nvidia", + "phone", + "408", + "486", + "2000" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "395f5805-365a-5d18-a818-ee84692af841", + "type": "text", + "content": "What am I voting on? A stockholder proposal to eliminate all supermajority voting provisions.\nVote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter.\nEffect of abstentions: Same as a vote AGAINST.\nEffect of broker non-votes: None.\nBelow is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 4. John Chevedden has notified us of his intent to present the following proposal at the 2026 Meeting.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard", + "metadata": { + "length": 743, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "What", + "voting", + "stockholder", + "proposal", + "eliminate", + "supermajority", + "provisions", + "Vote", + "required", + "majority", + "shares", + "present", + "person", + "represented", + "proxy", + "entitled", + "vote", + "matter", + "Effect", + "abstentions", + "Same", + "AGAINST", + "broker", + "votes", + "None", + "Below", + "supporting", + "statement", + "proponent", + "Board", + "accept", + "responsibility", + "This", + "voted", + "2026", + "Meeting", + "properly", + "presented", + "Our", + "opposes", + "adoption", + "recommends", + "Proposal", + "John", + "Chevedden", + "notified", + "intent" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "be03115d-d29a-513c-9a2f-7dbcb828d629", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Stockholder Proposal and Supporting Statement", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f7e01283-c75d-5088-af80-1559b8f56e59", + "type": "image", + "content": "\nEffect of abstentions: Same as a vote AGAINST.
0\n[images/image-80-Effect of .jpg]\n", + "path": "images/image-80-Effect of .jpg", + "metadata": { + "length": 92, + "summary": "image-80\nEffect of abstentions: Same as a vote AGAINST.
0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-80-Effect of .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d5aa1694-7614-5ef9-8c34-4650f17eb79e", + "type": "text", + "content": "Effect of abstentions: Same as a vote AGAINST.
0\n[images/image-80-Effect of .jpg]\n\n
\ntext_image\nFOR\nShareholder\nRights\n
\nShareholders request that the Board of Directors take each step necessary so that each voting requirement in our charter and bylaws (that is explicit or implicit due to default to state law) that calls for a greater than simple majority vote be replaced by a requirement for a majority of the votes cast for and against applicable proposals, or a simple majority in compliance with applicable laws. If necessary this means the closest standard to a majority of the votes cast for and against such proposals consistent with applicable laws. This proposal includes that the Company shall state in its governing documents that it shall not have any default super-majority voting standards upon adoption of this proposal.\nShareholders are willing to pay a premium for shares of companies that have excellent corporate governance. The supermajority voting requirements, like those of NVIDIA Corporation (NVDA), have been found to be one of 6 entrenching mechanisms that are negatively related to company performance according to \"What Matters in Corporate Governance\" by Lucien Bebchuk, Alma Cohen and Allen Ferrell of the Harvard Law School. Supermajority requirements can be used to block proposals supported by most shareowners but opposed by management.\nThis proposal topic won from 74% to 88% support at Weyerhaeuser, Alcoa, Waste Management, Goldman Sachs, FirstEnergy and Macy's. These votes would have been higher than 74% to 88% if more shareholders had access to independent proxy voting advice.\nThis proposal topic received 98% support each in 2024 at Domino's Pizza, FMC Corporation, ConocoPhillips, Masco Corporation and Power Integrations.\nThis proposal includes that the Board commit to not make any effort to defeat a NVDA proposal on this topic.\nThis proposal also includes that the Board commit to make a robust extra effort from a proxy solicitor perspective with the objective of obtaining the required vote for shareholder approval of this proposal topic as a binding NVDA proposal. As a matter of fact it may only take a de minimis extra effort since the 2025 NVDA proposal came so close to shareholder approval as described in the next paragraph.\nThe precise extra effort is at the Board's discretion. One proven method is to adjourn the annual meeting for up to 2-weeks and seek more votes. This seems to be a prudent step since the 2025 NVDA proposal on this topic received 65% support from all outstanding NVDA shares and thus came so close to the 67% votes required.\nPlease vote yes:", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 4 – Majority Vote Standard", + "metadata": { + "length": 2671, + "summary": "The proposal requests that NVIDIA's Board replace all supermajority voting requirements in its charter and bylaws with a simple majority standard. It cites academic research linking supermajority requirements to lower company performance and notes high shareholder support for similar proposals at other companies, including 98% support at several firms in 2024. The proposal also asks the Board to not oppose the measure and to make extra efforts, such as adjourning the annual meeting, to secure the required vote, noting that the 2025 proposal received 65% support from outstanding shares, close to the 67% threshold.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Effect", + "abstentions", + "Same", + "vote", + "AGAINST", + "details", + "summary", + "text", + "image", + "FOR", + "Shareholder", + "Rights", + "Shareholders", + "request", + "Board", + "Directors", + "step", + "voting", + "requirement", + "charter", + "bylaws", + "explicit", + "implicit", + "due", + "default", + "state", + "law", + "calls", + "greater", + "simple", + "majority", + "replaced", + "votes", + "cast", + "applicable", + "proposals", + "compliance", + "laws", + "If", + "means", + "closest", + "standard", + "consistent", + "This", + "proposal", + "includes", + "Company", + "governing", + "documents", + "super", + "standards", + "adoption", + "pay", + "premium", + "shares", + "companies", + "excellent", + "corporate", + "governance", + "The", + "supermajority", + "requirements", + "NVIDIA", + "Corporation", + "NVDA", + "found", + "entrenching", + "mechanisms", + "negatively", + "related", + "company", + "performance", + "What", + "Matters", + "Corporate", + "Governance", + "Lucien", + "Bebchuk", + "Alma", + "Cohen", + "Allen", + "Ferrell", + "Harvard", + "Law", + "School", + "Supermajority", + "block", + "supported", + "shareowners", + "opposed", + "management", + "topic", + "won", + "74%", + "88%", + "support", + "Weyerhaeuser", + "Alcoa", + "Waste", + "Management", + "Goldman", + "Sachs", + "FirstEnergy", + "Macy", + "These", + "higher", + "shareholders", + "access", + "independent", + "proxy", + "advice", + "received", + "98%", + "2024", + "Domino", + "Pizza", + "FMC", + "ConocoPhillips", + "Masco", + "Power", + "Integrations", + "commit", + "make", + "effort", + "defeat", + "robust", + "extra", + "solicitor", + "perspective", + "objective", + "obtaining", + "required", + "shareholder", + "approval", + "binding", + "As", + "matter", + "fact", + "de", + "minimis", + "2025", + "close", + "paragraph", + "precise", + "discretion", + "One", + "proven", + "method", + "adjourn", + "annual", + "meeting", + "weeks", + "seek", + "prudent", + "65%", + "outstanding", + "67%", + "Please" + ], + "keywords": [ + "Supermajority", + "Shareholder Rights", + "Corporate Governance" + ], + "connect_to": [ + { + "target": "f7e01283-c75d-5088-af80-1559b8f56e59", + "relation": "embeds", + "ref": "[images/image-80-Effect of .jpg]", + "position": { + "start": 58, + "end": 90 + } + } + ] + } + }, + { + "chunk_id": "efb05b63-c330-5611-9781-ab7366830564", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Majority Vote Standard – Proposal 4", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "872b3a2d-ece6-5af8-857f-8aa79ebc30d1", + "type": "text", + "content": "Our Board recommends a vote AGAINST Proposal 4.\nAt our 2025 Meeting, the Board proposed and recommended that stockholders approve an amendment and restatement of our Charter to eliminate the remaining supermajority voting provisions (the “2025 A&R Charter Proposal”). The Board also planned to adopt conforming amendments to our Bylaws, contingent on stockholder approval of such proposal. However, the 2025 A&R Charter Proposal did not receive the required level of stockholder support for approval.\nTo effectuate Proposal 4, the Board would need to resubmit a proposal substantially similar to the 2025 A&R Charter Proposal for stockholder approval at a future annual meeting. At this time, the Board believes that addressing repetitive proposals on this topic would not be an effective use of time, resources, and attention.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Majority Vote Standard – Proposal 4/Opposing Statement of the Board", + "metadata": { + "length": 827, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Board", + "recommends", + "vote", + "AGAINST", + "Proposal", + "At", + "2025", + "Meeting", + "proposed", + "recommended", + "stockholders", + "approve", + "amendment", + "restatement", + "Charter", + "eliminate", + "remaining", + "supermajority", + "voting", + "provisions", + "The", + "planned", + "adopt", + "conforming", + "amendments", + "Bylaws", + "contingent", + "stockholder", + "approval", + "proposal", + "However", + "receive", + "required", + "level", + "support", + "To", + "effectuate", + "resubmit", + "substantially", + "similar", + "future", + "annual", + "meeting", + "time", + "believes", + "addressing", + "repetitive", + "proposals", + "topic", + "effective", + "resources", + "attention" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b0fe0746-8d37-5970-8766-ff7c5731db01", + "type": "text", + "content": "Our Board recommends that you vote AGAINST Proposal 4— Simple Majority Vote Standard.\nWhat am I voting on? A stockholder proposal requesting an evaluation and report on faith-based community resource groups.\nVote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter.\nEffect of abstentions: Same as a vote AGAINST.\nEffect of broker non-votes: None.\nBelow is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 5. William Cunningham has notified us of its intent to present the following proposal at the 2026 Meeting.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Majority Vote Standard – Proposal 4/Recommendation of the Board", + "metadata": { + "length": 861, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Board", + "recommends", + "vote", + "AGAINST", + "Proposal", + "Simple", + "Majority", + "Vote", + "Standard", + "What", + "voting", + "stockholder", + "proposal", + "requesting", + "evaluation", + "report", + "faith", + "based", + "community", + "resource", + "groups", + "required", + "majority", + "shares", + "present", + "person", + "represented", + "proxy", + "entitled", + "matter", + "Effect", + "abstentions", + "Same", + "broker", + "votes", + "None", + "Below", + "supporting", + "statement", + "proponent", + "accept", + "responsibility", + "This", + "voted", + "2026", + "Meeting", + "properly", + "presented", + "opposes", + "adoption", + "William", + "Cunningham", + "notified", + "intent" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e529e84b-c2f6-59a5-835c-d45071c64215", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Stockholder Proposal and Supporting Statement_2", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d7d84e93-aba3-5089-8b25-0a7169c9d479", + "type": "text", + "content": "Whereas: NVIDIA is one of the largest and most high-profile companies in the United States, employing $^{1}$ more than 35,000 people from a plethora of different cultural & religious backgrounds. As a major employer, NVIDIA should support the religious freedom of its employees. NVIDIA is already required to comply with many laws prohibiting discrimination against employees based on their religious status and views.\nRespecting diverse religious views allows NVIDIA to attract the most qualified talent, promote a diverse and vibrant business culture, and is a key component to make sure it fully engages each of its employees. One of the best ways to promote religious diversity is through faith-based community resource groups (elsewhere known as employee resource groups, or ERGs). CRGs allow like-minded employees to connect with one another, seek professional development, and promote understanding and dialogue with the broader workforce. Despite this, NVIDIA does not have a faith-based community resource group (CRG).\nNVIDIA does not rectify this apparently disparity in treatment even though the vast majority of Americans identify as religious, and even though the Company recognizes CRGs formed around gender, sexual identity, along a variety of other criteria as per its website. $^{2}$\nAccording to the 2023 Freedom at Work $^{3}$ survey, 60% of employees were concerned that their company would punish them for expressing their religious or political views at work, and 54% said they feared the same for sharing these views even on their private social media accounts. NVIDIA needs to take proactive steps to address this shortcoming by promoting faith-based CRGs and providing them the same support and access that other CRGs enjoy.\nThis isn't merely a political or social point but a matter of brand value. NVIDIA is the highest-valued company on earth, with an estimated brand value $^{3}$ exceeding \\$87.9 billion, 67 percent of its more than \\$130 billion revenue in 2025. $^{4}$ NVIDIA is one of the most influential players in the American technological landscape. As concerns over neutrality on controversial issues, respect for religion, and equal treatment for employees at America's largest companies continue to mount, shareholders are right to ask about how the company's workforce policies forward this goal of equal treatment.\nRecent Supreme Court decisions in Groff v. DeJoy and Muldrow v. City of St. Louis have also clarified that religious protections for employees extend to all terms, conditions, and privileges of employment, not just monetary compensation. So failure to allow faith-based community resource groups may be illegal.\nResolved: Shareholders request the Board of Directors of NVIDIA Corporation conduct an evaluation and issue a report, or disclose analysis which has already been performed, within the next year, at reasonable cost and excluding proprietary information and disclosure of anything that would constitute an admission of pending litigation, evaluating the risks related to religious discrimination against employees.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 5 – Report on Faith-Based Community Resource Groups", + "metadata": { + "length": 3082, + "summary": "Shareholders request NVIDIA to evaluate risks of religious discrimination, citing lack of faith-based Community Resource Groups despite supporting other ERGs. The proposal notes legal obligations under recent Supreme Court rulings and brand value concerns, urging a report on equal treatment for religious employees.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Whereas", + "NVIDIA", + "largest", + "high", + "profile", + "companies", + "United", + "States", + "employing", + "35", + "000", + "people", + "plethora", + "cultural", + "religious", + "backgrounds", + "As", + "major", + "employer", + "support", + "freedom", + "employees", + "required", + "comply", + "laws", + "prohibiting", + "discrimination", + "based", + "status", + "views", + "Respecting", + "diverse", + "attract", + "qualified", + "talent", + "promote", + "vibrant", + "business", + "culture", + "key", + "component", + "make", + "fully", + "engages", + "One", + "ways", + "diversity", + "faith", + "community", + "resource", + "groups", + "employee", + "ERGs", + "CRGs", + "minded", + "connect", + "seek", + "professional", + "development", + "understanding", + "dialogue", + "broader", + "workforce", + "Despite", + "group", + "CRG", + "rectify", + "apparently", + "disparity", + "treatment", + "vast", + "majority", + "Americans", + "identify", + "Company", + "recognizes", + "formed", + "gender", + "sexual", + "identity", + "variety", + "criteria", + "website", + "According", + "2023", + "Freedom", + "Work", + "survey", + "60%", + "concerned", + "company", + "punish", + "expressing", + "political", + "work", + "54%", + "feared", + "sharing", + "private", + "social", + "media", + "accounts", + "proactive", + "steps", + "address", + "shortcoming", + "promoting", + "providing", + "access", + "enjoy", + "This", + "isn", + "point", + "matter", + "brand", + "highest", + "valued", + "earth", + "estimated", + "exceeding", + "87.9", + "billion", + "67", + "percent", + "130", + "revenue", + "2025", + "influential", + "players", + "American", + "technological", + "landscape", + "concerns", + "neutrality", + "controversial", + "issues", + "respect", + "religion", + "equal", + "America", + "continue", + "mount", + "shareholders", + "policies", + "forward", + "goal", + "Recent", + "Supreme", + "Court", + "decisions", + "Groff", + "DeJoy", + "Muldrow", + "City", + "St", + "Louis", + "clarified", + "protections", + "extend", + "terms", + "conditions", + "privileges", + "employment", + "monetary", + "compensation", + "So", + "failure", + "illegal", + "Resolved", + "Shareholders", + "request", + "Board", + "Directors", + "Corporation", + "conduct", + "evaluation", + "issue", + "report", + "disclose", + "analysis", + "performed", + "year", + "reasonable", + "cost", + "excluding", + "proprietary", + "information", + "disclosure", + "constitute", + "admission", + "pending", + "litigation", + "evaluating", + "risks", + "related" + ], + "keywords": [ + "religious discrimination", + "employee resource groups", + "shareholder resolution" + ], + "connect_to": [] + } + }, + { + "chunk_id": "6cc85123-ca2f-5f15-9e17-385e54e209d1", + "type": "text", + "content": "Our Board recommends a vote AGAINST Proposal 5. The requested report is unnecessary and would not provide meaningful additional benefits to NVIDIA or enhance stockholder value as we have already implemented robust systems and processes designed to minimize risks related to religious and other discrimination.\nNVIDIA is committed to ensuring that our employees are protected and supported through enterprise-wide policies, practices, and reporting mechanisms. We maintain a work environment in which all individuals are treated with dignity and respect. We do not tolerate unlawful discrimination, harassment, or retaliation in any form, including based on religion or creed and all employees are required to participate in discrimination and harassment prevention training. Our employees have multiple avenues to report concerns about harassment, discrimination or other workplace conduct, including anonymously and without fear of retaliation, and we investigate complaints promptly and take appropriate action when warranted. We also regularly gather employee feedback through surveys and other forums to identify concerns, track progress, and respond quickly to employee needs.\nThe Board believes that our existing policies, practices, and feedback mechanisms already address the issues raised in Proposal 5, and that the additional evaluation and reporting requested would not be necessary or beneficial for NVIDIA or our stockholders.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 5 – Report on Faith-Based Community Resource Groups/Opposing Statement of the Board", + "metadata": { + "length": 1440, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Board", + "recommends", + "vote", + "AGAINST", + "Proposal", + "The", + "requested", + "report", + "unnecessary", + "provide", + "meaningful", + "additional", + "benefits", + "NVIDIA", + "enhance", + "stockholder", + "implemented", + "robust", + "systems", + "processes", + "designed", + "minimize", + "risks", + "related", + "religious", + "discrimination", + "committed", + "ensuring", + "employees", + "protected", + "supported", + "enterprise", + "wide", + "policies", + "practices", + "reporting", + "mechanisms", + "We", + "maintain", + "work", + "environment", + "individuals", + "treated", + "dignity", + "respect", + "tolerate", + "unlawful", + "harassment", + "retaliation", + "form", + "including", + "based", + "religion", + "creed", + "required", + "participate", + "prevention", + "training", + "multiple", + "avenues", + "concerns", + "workplace", + "conduct", + "anonymously", + "fear", + "investigate", + "complaints", + "promptly", + "action", + "warranted", + "regularly", + "gather", + "employee", + "feedback", + "surveys", + "forums", + "identify", + "track", + "progress", + "respond", + "quickly", + "believes", + "existing", + "address", + "issues", + "raised", + "evaluation", + "beneficial", + "stockholders" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d53675e4-37dd-5d6c-8cd8-ad8ace628e4c", + "type": "text", + "content": "Our Board recommends that you vote AGAINST Proposal 5—Report on Faith-Based Community Resource Groups.\nWhat am I voting on? A stockholder proposal requesting an evaluation and report on civil rights and non-discrimination related to DEI.\nVote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter.\nEffect of abstentions: Same as a vote AGAINST.\nEffect of broker non-votes: None.\nBelow is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 6. American Conservative Values ETF has notified us of its intent to present the following proposal at the 2026 Meeting.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 5 – Report on Faith-Based Community Resource Groups/Recommendation of the Board", + "metadata": { + "length": 905, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Board", + "recommends", + "vote", + "AGAINST", + "Proposal", + "Report", + "Faith", + "Based", + "Community", + "Resource", + "Groups", + "What", + "voting", + "stockholder", + "proposal", + "requesting", + "evaluation", + "report", + "civil", + "rights", + "discrimination", + "related", + "DEI", + "Vote", + "required", + "majority", + "shares", + "present", + "person", + "represented", + "proxy", + "entitled", + "matter", + "Effect", + "abstentions", + "Same", + "broker", + "votes", + "None", + "Below", + "supporting", + "statement", + "proponent", + "accept", + "responsibility", + "This", + "voted", + "2026", + "Meeting", + "properly", + "presented", + "opposes", + "adoption", + "American", + "Conservative", + "Values", + "ETF", + "notified", + "intent" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8328e709-dab6-5bba-8af4-169f100650aa", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Stockholder Proposal and Supporting Statement_3", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a14d27fd-23b9-5003-bbfb-3c1024592187", + "type": "text", + "content": "NVIDIA Corporation has over 36,000 employees. As an employer, it is legally required to comply with many laws prohibiting discrimination against employees on a variety of factors, including race, sex, religion, and sometimes political affiliation.\nBut the 2025 edition of the Viewpoint Diversity Score Business Index $^{1}$ found that 90% of scored companies, as well as NVIDIA, promote divisive training and recruitment concepts like critical race theory (CRT) that replace genuine cultural and ideological diversity with a narrow focus on group identity. NVIDIA has stated that it has “increased [its] focus on diversity recruiting” and “when recruiting new talent. . . [it] strive[s] to build a robust talent pipeline that includes those underrepresented in the technology field, including women, black/African American, and Hispanic/Latino candidates.” $^{2}$\nFocusing on these metrics in recruiting is rooted in concepts that label employees as “oppressed” or “oppressors” based on belief or immutable traits, which can foster hostility and partiality instead of equal opportunity based on merit. A recent study found that DEI actually increases bias, hostility, and attribution of discriminatory intent instead of reducing them. $^{3}$ DEI also silences dissenting views: the Freedom at Work survey found that 60% of employees feared reprisal for expressing their religious or political views at work, and over half feared the same for private social media posts. $^{4}$\nCompanies are also facing increased legal and reputational risk for DEI programs that make distinctions based on race or other protected classes in light of recent Supreme Court decisions in Students for Fair Admission v. Harvard, Groff v DeJoy, City of St. Louis v. Muldrow, and Ames v. Ohio Department of Youth Services. State attorneys general and employees have initiated lawsuits and investigations. $^{5}$ Recent scholarship has cast serious doubt on the common assertion that DEI is good for business. $^{6}$ The Wall Street Journal recently reported that “Diversity Goals Are Disappearing from Companies’ Annual Reports.” $^{7}$ And numerous companies are voluntarily removing DEI initiatives entirely in response to sustained customer dissatisfaction and public pressure. $^{8}$\nOn top of this, President Trump issued an executive order seeking to root out illegal DEI practices, including in the private sector. $^{9}$ The EEOC has initiated investigations and secured settlements from major law firms to stop their DEI practices. $^{10}$ The EEOC and DOJ jointly issued guidance detailing DEI practices that may be illegal and encouraging whistleblowers to report these practices. $^{11}$ In light of these risks, the Company must take immediate steps to assess potential shortcomings and allay these concerns.\nResolved: Shareholders request the Board of Directors conduct an evaluation and issue a report within the next year, at reasonable cost and excluding proprietary and confidential information, evaluating how NVIDIA's hiring and training policies and practices impact employees and prospective employees based on their race, color, religion (including religious views), sex, national origin, or political views, and the associated legal, reputational, operational, and other relevant risks to the NVIDIA.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 6 – Report on Respecting Workforce Civil Liberties", + "metadata": { + "length": 3301, + "summary": "The passage discusses a shareholder resolution requesting NVIDIA to evaluate its hiring and training policies for potential discrimination based on race, religion, sex, and political views. It cites concerns that DEI programs may foster bias, increase legal risks, and face regulatory scrutiny, referencing recent Supreme Court decisions, executive orders, and EEOC actions. The resolution asks for a report on associated risks.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "NVIDIA", + "Corporation", + "36", + "000", + "employees", + "As", + "employer", + "legally", + "required", + "comply", + "laws", + "prohibiting", + "discrimination", + "variety", + "factors", + "including", + "race", + "sex", + "religion", + "political", + "affiliation", + "But", + "2025", + "edition", + "Viewpoint", + "Diversity", + "Score", + "Business", + "Index", + "found", + "90%", + "scored", + "companies", + "promote", + "divisive", + "training", + "recruitment", + "concepts", + "critical", + "theory", + "CRT", + "replace", + "genuine", + "cultural", + "ideological", + "diversity", + "narrow", + "focus", + "group", + "identity", + "stated", + "increased", + "recruiting", + "talent", + "strive", + "build", + "robust", + "pipeline", + "includes", + "underrepresented", + "technology", + "field", + "women", + "black", + "African", + "American", + "Hispanic", + "Latino", + "candidates", + "Focusing", + "metrics", + "rooted", + "label", + "oppressed", + "oppressors", + "based", + "belief", + "immutable", + "traits", + "foster", + "hostility", + "partiality", + "equal", + "opportunity", + "merit", + "recent", + "study", + "DEI", + "increases", + "bias", + "attribution", + "discriminatory", + "intent", + "reducing", + "silences", + "dissenting", + "views", + "Freedom", + "Work", + "survey", + "60%", + "feared", + "reprisal", + "expressing", + "religious", + "work", + "half", + "private", + "social", + "media", + "posts", + "Companies", + "facing", + "legal", + "reputational", + "risk", + "programs", + "make", + "distinctions", + "protected", + "classes", + "light", + "Supreme", + "Court", + "decisions", + "Students", + "Fair", + "Admission", + "Harvard", + "Groff", + "DeJoy", + "City", + "St", + "Louis", + "Muldrow", + "Ames", + "Ohio", + "Department", + "Youth", + "Services", + "State", + "attorneys", + "general", + "initiated", + "lawsuits", + "investigations", + "Recent", + "scholarship", + "cast", + "doubt", + "common", + "assertion", + "good", + "business", + "The", + "Wall", + "Street", + "Journal", + "recently", + "reported", + "Goals", + "Are", + "Disappearing", + "Annual", + "Reports", + "And", + "numerous", + "voluntarily", + "removing", + "initiatives", + "response", + "sustained", + "customer", + "dissatisfaction", + "public", + "pressure", + "On", + "top", + "President", + "Trump", + "issued", + "executive", + "order", + "seeking", + "root", + "illegal", + "practices", + "sector", + "EEOC", + "secured", + "settlements", + "major", + "law", + "firms", + "stop", + "10", + "DOJ", + "jointly", + "guidance", + "detailing", + "encouraging", + "whistleblowers", + "report", + "11", + "In", + "risks", + "Company", + "steps", + "assess", + "potential", + "shortcomings", + "allay", + "concerns", + "Resolved", + "Shareholders", + "request", + "Board", + "Directors", + "conduct", + "evaluation", + "issue", + "year", + "reasonable", + "cost", + "excluding", + "proprietary", + "confidential", + "information", + "evaluating", + "hiring", + "policies", + "impact", + "prospective", + "color", + "national", + "origin", + "operational", + "relevant" + ], + "keywords": [ + "DEI", + "NVIDIA", + "Shareholder Resolution" + ], + "connect_to": [] + } + }, + { + "chunk_id": "f55daf6e-29f5-565e-b40a-558945d30e28", + "type": "text", + "content": "Our Board recommends a vote AGAINST Proposal 6. The requested report is unnecessary and would not provide meaningful additional benefits to NVIDIA or enhance stockholder value.\nNVIDIA complies with applicable law and is committed to equal employment opportunity requirements. We do not discriminate against employees or applicants for employment on the basis of race, color, religion, sex, gender, marital status, veteran status, age, national origin, ancestry, disability, genetic or medical information, sexual orientation, pregnancy, childbirth or related medical conditions, or any other characteristic protected by applicable law. Our employment decisions – including recruitment, hiring, training, compensation, benefits, promotions, transfers, and terminations – are based on merit, valid job requirements, and other lawful criteria.\nIn addition, the Board maintains strong oversight and compliance mechanisms to ensure these policies and practices are effectively implemented. The CC oversees our human capital management policies and strategies, the AC oversees compliance with legal and regulatory requirements, ethical standards, and our compliance programs, and the NCGC oversees compliance with our Code of Conduct. Together, these oversight structures provide sufficient and comprehensive governance and accountability for the policies and practices addressed by Proposal 6.\nThe Board believes that our existing policies, practices, and oversight mechanisms already address the issues raised in Proposal 6, and that the additional evaluation and reporting requested would not be necessary or beneficial for NVIDIA or our stockholders.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 6 – Report on Respecting Workforce Civil Liberties/Opposing Statement of the Board", + "metadata": { + "length": 1648, + "summary": "The Board recommends voting against Proposal 6, arguing the requested report is unnecessary and would not enhance stockholder value. NVIDIA states it complies with equal employment opportunity laws, bases employment decisions on merit, and maintains strong oversight through committees (CC, AC, NCGC) for human capital, compliance, and ethics. The Board believes existing policies and mechanisms sufficiently address the proposal's issues.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Board", + "recommends", + "vote", + "AGAINST", + "Proposal", + "The", + "requested", + "report", + "unnecessary", + "provide", + "meaningful", + "additional", + "benefits", + "NVIDIA", + "enhance", + "stockholder", + "complies", + "applicable", + "law", + "committed", + "equal", + "employment", + "opportunity", + "requirements", + "We", + "discriminate", + "employees", + "applicants", + "basis", + "race", + "color", + "religion", + "sex", + "gender", + "marital", + "status", + "veteran", + "age", + "national", + "origin", + "ancestry", + "disability", + "genetic", + "medical", + "information", + "sexual", + "orientation", + "pregnancy", + "childbirth", + "related", + "conditions", + "characteristic", + "protected", + "decisions", + "including", + "recruitment", + "hiring", + "training", + "compensation", + "promotions", + "transfers", + "terminations", + "based", + "merit", + "valid", + "job", + "lawful", + "criteria", + "In", + "addition", + "maintains", + "strong", + "oversight", + "compliance", + "mechanisms", + "ensure", + "policies", + "practices", + "effectively", + "implemented", + "CC", + "oversees", + "human", + "capital", + "management", + "strategies", + "AC", + "legal", + "regulatory", + "ethical", + "standards", + "programs", + "NCGC", + "Code", + "Conduct", + "Together", + "structures", + "sufficient", + "comprehensive", + "governance", + "accountability", + "addressed", + "believes", + "existing", + "address", + "issues", + "raised", + "evaluation", + "reporting", + "beneficial", + "stockholders" + ], + "keywords": [ + "Proposal 6", + "Board recommendation", + "EEO compliance" + ], + "connect_to": [] + } + }, + { + "chunk_id": "c284fb42-c6a3-5c95-90d0-719e9a765c71", + "type": "text", + "content": "Our Board recommends that you vote AGAINST Proposal 6—Report on Workforce Civil Liberties.\nWhat am I voting on? A stockholder proposal to requesting a report disclosing GHG emissions from the use of the Company's sold products.\nVote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter.\nEffect of abstentions: Same as a vote AGAINST.\nEffect of broker non-votes: None.\nBelow is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 7. Green Century Capital Management, Inc. has notified us of its intent to present the following proposal at the 2026 Meeting. Mercy Investment Services, Inc. is a co-filer for this proposal.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 6 – Report on Respecting Workforce Civil Liberties/Recommendation of the Board", + "metadata": { + "length": 966, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Board", + "recommends", + "vote", + "AGAINST", + "Proposal", + "Report", + "Workforce", + "Civil", + "Liberties", + "What", + "voting", + "stockholder", + "proposal", + "requesting", + "report", + "disclosing", + "GHG", + "emissions", + "Company", + "sold", + "products", + "Vote", + "required", + "majority", + "shares", + "present", + "person", + "represented", + "proxy", + "entitled", + "matter", + "Effect", + "abstentions", + "Same", + "broker", + "votes", + "None", + "Below", + "supporting", + "statement", + "proponent", + "accept", + "responsibility", + "This", + "voted", + "2026", + "Meeting", + "properly", + "presented", + "opposes", + "adoption", + "Green", + "Century", + "Capital", + "Management", + "Inc", + "notified", + "intent", + "Mercy", + "Investment", + "Services", + "filer" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "799f8c27-6bb1-5574-8c99-a43fb727fe2d", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Stockholder Proposal and Supporting Statement_4", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "22218ef3-a720-5e7e-9b91-0e2cb47c81df", + "type": "text", + "content": "Whereas: Climate change poses macroeconomic risks that can depress returns for long-term diversified investors. Severe weather accounted for a record 21 billion-dollar disasters in 2025. $^{1}$\nIn its 10-K, NVIDIA acknowledges that “climate change, its impact on our supply chain and critical infrastructure worldwide...may disrupt our business and cause us to experience higher attrition, losses and costs.” $^{2}$ One-third of semiconductor supply could be at risk within a decade unless the industry adapts to manage climate risks. $^{3}$\nMeanwhile, energy-intensive artificial intelligence (AI) is compounding climate risk. Goldman Sachs research forecasts data centers' global power demand will increase 50% by 2027, and the rate of U.S. AI growth is expected to emit an additional 24 to 44 million metric tons of carbon dioxide by 2030. $^{4}$\nNVIDIA plays a key role in the expansion of AI through designing AI chips, systems, and software.\nWhile NVIDIA set 2030 targets to reduce absolute greenhouse gas emissions (GHG) emissions from electricity usage and emissions intensity from use of its sold products, in the proponent's opinion, NVIDIA's progress toward mitigating climate risk through real world achievement of GHG emissions reductions is difficult to assess.\nFor instance, NVIDIA reported zero market-based electricity emissions as a part of its overall operational emissions in FY25, partly due to contracting renewable energy projects across the grid. Meanwhile, it also disclosed a separate increase in location-based electricity emissions from the energy it drew from the local power grid. $^{5}$ Additionally, the company's total absolute emissions nearly doubled between fiscal years 2024 and 2025. $^{6}$\nFurthermore, NVIDIA's reported GHG inventory does not include use of its sold products. These emissions are likely a significant contributor to the company's GHG footprint based on peer comparison $^{7}$ and represent a key metric for tracking emissions reductions.\nPeers AMD, Intel, Onsemi, and NXP have committed to reporting or already published more detailed climate disclosures that include:\n• Absolute emissions from use of sold products and initiatives to reduce them;\n• Third-party validation and increased criteria for renewable energy projects;\n• Emissions pathways to 2030, including plans to advance climate initiatives across their value chains.\nBy enhancing its emissions disclosures, NVIDIA will increase the legitimacy of its climate targets and better demonstrate climate risk mitigation progress to investors as its business grows.\nRESOLVED: Shareholders request that NVIDIA issue a report, at reasonable cost and omitting proprietary information, disclosing the GHG emissions from use of its sold products.\nSUPPORTING STATEMENT: Proponents recommend, at management's discretion:\n• Disclosing emissions that account for major sources of its total GHG footprint;\n• Considering GHG emissions disclosure guidance;\n- Reporting based on reasonable emissions estimates, updated annually, and providing timelines for issuing or completing disclosures;\n• Outlining whether and how the company plans to achieve absolute emissions reductions.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 7 – Greenhouse Gas Emissions Disclosure", + "metadata": { + "length": 3179, + "summary": "The text discusses climate change risks to NVIDIA, including macroeconomic impacts and supply chain disruptions. It notes NVIDIA's role in AI expansion and its climate targets, but criticizes the company's emissions disclosures for excluding use of sold products and showing increased absolute emissions. Peers like AMD and Intel provide more detailed climate reporting. A shareholder resolution requests NVIDIA to issue a report disclosing GHG emissions from use of sold products, with recommendations for improved disclosure practices.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Whereas", + "Climate", + "change", + "poses", + "macroeconomic", + "risks", + "depress", + "returns", + "long", + "term", + "diversified", + "investors", + "Severe", + "weather", + "accounted", + "record", + "21", + "billion", + "dollar", + "disasters", + "2025", + "In", + "10", + "NVIDIA", + "acknowledges", + "climate", + "impact", + "supply", + "chain", + "critical", + "infrastructure", + "worldwide", + "disrupt", + "business", + "experience", + "higher", + "attrition", + "losses", + "costs", + "One", + "semiconductor", + "risk", + "decade", + "industry", + "adapts", + "manage", + "Meanwhile", + "energy", + "intensive", + "artificial", + "intelligence", + "AI", + "compounding", + "Goldman", + "Sachs", + "research", + "forecasts", + "data", + "centers", + "global", + "power", + "demand", + "increase", + "50%", + "2027", + "rate", + "growth", + "expected", + "emit", + "additional", + "24", + "44", + "million", + "metric", + "tons", + "carbon", + "dioxide", + "2030", + "plays", + "key", + "role", + "expansion", + "designing", + "chips", + "systems", + "software", + "While", + "set", + "targets", + "reduce", + "absolute", + "greenhouse", + "gas", + "emissions", + "GHG", + "electricity", + "usage", + "intensity", + "sold", + "products", + "proponent", + "opinion", + "progress", + "mitigating", + "real", + "world", + "achievement", + "reductions", + "difficult", + "assess", + "For", + "instance", + "reported", + "market", + "based", + "part", + "operational", + "FY25", + "partly", + "due", + "contracting", + "renewable", + "projects", + "grid", + "disclosed", + "separate", + "location", + "drew", + "local", + "Additionally", + "company", + "total", + "doubled", + "fiscal", + "years", + "2024", + "Furthermore", + "inventory", + "include", + "These", + "significant", + "contributor", + "footprint", + "peer", + "comparison", + "represent", + "tracking", + "Peers", + "AMD", + "Intel", + "Onsemi", + "NXP", + "committed", + "reporting", + "published", + "detailed", + "disclosures", + "Absolute", + "initiatives", + "Third", + "party", + "validation", + "increased", + "criteria", + "Emissions", + "pathways", + "including", + "plans", + "advance", + "chains", + "By", + "enhancing", + "legitimacy", + "demonstrate", + "mitigation", + "grows", + "RESOLVED", + "Shareholders", + "request", + "issue", + "report", + "reasonable", + "cost", + "omitting", + "proprietary", + "information", + "disclosing", + "SUPPORTING", + "STATEMENT", + "Proponents", + "recommend", + "management", + "discretion", + "Disclosing", + "account", + "major", + "sources", + "Considering", + "disclosure", + "guidance", + "Reporting", + "estimates", + "updated", + "annually", + "providing", + "timelines", + "issuing", + "completing", + "Outlining", + "achieve" + ], + "keywords": [ + "GHG emissions", + "climate risk", + "shareholder resolution" + ], + "connect_to": [] + } + }, + { + "chunk_id": "d979ea57-c6e0-5f57-b985-6726c6544339", + "type": "text", + "content": "Our Board recommends a vote AGAINST Proposal 7 because it is unnecessary, overly prescriptive and not in the best interests of NVIDIA or our stockholders.\nThe proposal is unnecessary given NVIDIA's history of robust climate reporting and targets, and our future disclosure plans\nNVIDIA's current practices evidence a clear commitment to sustainability, climate risk management, and effective disclosure. For example, NVIDIA consistently publishes highly detailed climate reporting, including in our 2025 Sustainability Report, which is assured by our financial auditor. In addition to Scope 1 and 2 emissions, our public reporting includes Scope 3 emissions across eight categories, as well as disclosure on product energy efficiency, climate strategy, and progress towards achieving our targets, including our success in meeting our renewable energy and supplier engagement targets. In 2025, we also announced significant new targets validated by the Science Based Targets initiative (SBTi), which include absolute Scope 1 and 2 emissions reductions and reductions in Scope 3 emissions intensity from use of sold GPU products. In 2025, in an effort to offer additional transparency into emissions, we also began publishing the market's first summaries of ISO-conformant, third-party reviewed product carbon footprints for accelerated computing products. We continue to advance our climate reporting and, as discussed with the proponent, are planning further enhancements of our Scope 3 reporting, including emissions from the use of sold products.\nThe proposal adopts an overly prescriptive approach to highly complex disclosures\nThe Board believes NVIDIA's management, supported by our CS team and overseen by the NCGC, is best positioned to make decisions regarding the specific content and methodology of our climate disclosures. Scope 3 disclosures are inherently complex and subject to varied and developing calculation methodologies, levels of assurance, and credibility. The difficulty of preparing reliable Scope 3 disclosures is particularly acute with respect to the fast-evolving AI computing industry. Despite these considerations, the proposal attempts to replace management's judgment regarding highly technical reporting with a broad prescriptive mandate that includes no timeframe or acknowledgement of the significant challenges and judgments involved in preparing upstream emissions disclosures.\nGiven NVIDIA's track record and reporting plans, adoption of a prescriptive approach to Scope 3 reporting methodology is inappropriate and unhelpful to our stockholders.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 7 – Greenhouse Gas Emissions Disclosure/Opposing Statement of the Board", + "metadata": { + "length": 2584, + "summary": "NVIDIA's Board recommends voting against Proposal 7, arguing it is unnecessary and overly prescriptive. The company highlights its robust climate reporting history, including detailed Scope 1, 2, and 3 emissions disclosures, SBTi-validated targets, and product carbon footprint summaries. The Board asserts that management is best positioned to handle complex Scope 3 disclosures, especially in the fast-evolving AI industry, and that the proposal's prescriptive approach is inappropriate and unhelpful to stockholders.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Board", + "recommends", + "vote", + "AGAINST", + "Proposal", + "unnecessary", + "overly", + "prescriptive", + "interests", + "NVIDIA", + "stockholders", + "The", + "proposal", + "history", + "robust", + "climate", + "reporting", + "targets", + "future", + "disclosure", + "plans", + "current", + "practices", + "evidence", + "clear", + "commitment", + "sustainability", + "risk", + "management", + "effective", + "For", + "consistently", + "publishes", + "highly", + "detailed", + "including", + "2025", + "Sustainability", + "Report", + "assured", + "financial", + "auditor", + "In", + "addition", + "Scope", + "emissions", + "public", + "includes", + "categories", + "product", + "energy", + "efficiency", + "strategy", + "progress", + "achieving", + "success", + "meeting", + "renewable", + "supplier", + "engagement", + "announced", + "significant", + "validated", + "Science", + "Based", + "Targets", + "initiative", + "SBTi", + "include", + "absolute", + "reductions", + "intensity", + "sold", + "GPU", + "products", + "effort", + "offer", + "additional", + "transparency", + "began", + "publishing", + "market", + "summaries", + "ISO", + "conformant", + "party", + "reviewed", + "carbon", + "footprints", + "accelerated", + "computing", + "We", + "continue", + "advance", + "discussed", + "proponent", + "planning", + "enhancements", + "adopts", + "approach", + "complex", + "disclosures", + "believes", + "supported", + "CS", + "team", + "overseen", + "NCGC", + "positioned", + "make", + "decisions", + "specific", + "content", + "methodology", + "inherently", + "subject", + "varied", + "developing", + "calculation", + "methodologies", + "levels", + "assurance", + "credibility", + "difficulty", + "preparing", + "reliable", + "acute", + "respect", + "fast", + "evolving", + "AI", + "industry", + "Despite", + "considerations", + "attempts", + "replace", + "judgment", + "technical", + "broad", + "mandate", + "timeframe", + "acknowledgement", + "challenges", + "judgments", + "involved", + "upstream", + "Given", + "track", + "record", + "adoption", + "inappropriate", + "unhelpful" + ], + "keywords": [ + "Climate Reporting", + "Scope 3", + "NVIDIA" + ], + "connect_to": [] + } + }, + { + "chunk_id": "adfcde16-1125-52c1-b7e7-2df8087bd89f", + "type": "text", + "content": "Our Board recommends that you vote AGAINST Proposal 7—Greenhouse Gas Emissions Disclosure.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 7 – Greenhouse Gas Emissions Disclosure/Recommendation of the Board", + "metadata": { + "length": 90, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Board", + "recommends", + "vote", + "AGAINST", + "Proposal", + "Greenhouse", + "Gas", + "Emissions", + "Disclosure" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1dd1cd13-acac-5a60-9c73-2bb038680798", + "type": "table", + "content": "
Plan CategoryNumber of securities to be issued upon exercise of outstanding options, warrants, and rights (a)Weighted-average exercise price of outstanding options, warrants, and rights (b)Number of securities remaining available for future issuance under equity compensation plans (excluding securities reflected in column (a)) (c)
Equity compensation plans approved by security holders (1)192,391,7653,551,257,925 (2)
Equity compensation plans not approved by security holders
Total192,391,7653,551,257,925
", + "path": "tables/table-59 Equity Plan Table.html", + "metadata": { + "length": 704, + "summary": "table-60\nThe table shows equity compensation plans approved by security holders with 192,391,765 securities from outstanding options and 3,551,257,925 remaining for future issuance. No unapproved plans exist.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-59 Equity Plan Table.html", + "keywords": [ + "equity compensation", + "securities", + "plans" + ], + "tokens": [] + } + }, + { + "chunk_id": "f0eac104-383b-5370-8032-fd34ce82ac77", + "type": "text", + "content": "The number of shares that were issuable upon the vesting of outstanding RSUs and PSUs, and the number of shares that remained available for future issuance, under each of our equity compensation plans as of January 25, 2026, are summarized below. We have not had any stock options outstanding since the end of Fiscal 2024.\n\n[tables/table-59 Equity Plan Table.html]\n\n(1) This row includes our 2007 Plan and our ESPP. Under our ESPP, participants are permitted to purchase our common stock at a discount on certain dates through payroll deductions within a pre-determined purchase period. Accordingly, the number of shares to be issued upon exercise of outstanding rights under our ESPP as of January 25, 2026 is not determinable.\n(2) As of January 25, 2026, (a) the number of shares that remained available for future issuance under the 2007 Plan was 1,324,516,906, and (b) the number of shares that remained available for future issuance under the ESPP was 2,226,741,019, of which up to 2,825,984 shares may be purchased under the ESPP in the current purchase period which runs until August 31, 2026, based on estimated participation and contribution rates, purchase prices based on the applicable offering date prices, and the \\$25,000 limit under Section 423(b)(8) of the Internal Revenue Code.\nDuring Fiscal 2026, we granted an aggregate of 69,186,497 shares under our 2007 Plan in the form of RSUs and PSUs, 1,041,733 of which were granted to our NEOs, 21,588 of which were granted to our non-employee directors, and 68,123,176 of which were granted to our other employees. For this purpose, PSUs are counted in the year of grant at the maximum number of shares that may become eligible to vest. Also during Fiscal 2026, an aggregate of 13,226,923 shares were purchased under our ESPP, 4,655 of which were purchased by our NEOs, and 13,222,268 of which were purchased by our other employees. Our non-employee directors are not eligible to participate in our ESPP.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Equity Compensation Plan Information", + "metadata": { + "length": 2011, + "summary": "The passage summarizes equity compensation plan data as of January 25, 2026, including shares issuable from vesting RSUs and PSUs, and shares available for future issuance under the 2007 Plan and ESPP. No stock options were outstanding since Fiscal 2024. Key figures: 1,324,516,906 shares available under the 2007 Plan; 2,226,741,019 under ESPP, with up to 2,825,984 shares purchasable in the current period. In Fiscal 2026, 69,186,497 shares were granted as RSUs/PSUs (1,041,733 to NEOs, 21,588 to non-employee directors, 68,123,176 to other employees), and 13,226,923 shares were purchased under ESPP (4,655 by NEOs, 13,222,268 by other employees).", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "number", + "shares", + "issuable", + "vesting", + "outstanding", + "RSUs", + "PSUs", + "remained", + "future", + "issuance", + "equity", + "compensation", + "plans", + "January", + "25", + "2026", + "summarized", + "We", + "stock", + "options", + "end", + "Fiscal", + "2024", + "This", + "row", + "includes", + "2007", + "Plan", + "ESPP", + "Under", + "participants", + "permitted", + "purchase", + "common", + "discount", + "dates", + "payroll", + "deductions", + "pre", + "determined", + "period", + "Accordingly", + "issued", + "exercise", + "rights", + "determinable", + "As", + "324", + "516", + "906", + "226", + "741", + "019", + "825", + "984", + "purchased", + "current", + "runs", + "August", + "31", + "based", + "estimated", + "participation", + "contribution", + "rates", + "prices", + "applicable", + "offering", + "date", + "000", + "limit", + "Section", + "423", + "Internal", + "Revenue", + "Code", + "During", + "granted", + "aggregate", + "69", + "186", + "497", + "form", + "041", + "733", + "NEOs", + "21", + "588", + "employee", + "directors", + "68", + "123", + "176", + "employees", + "For", + "purpose", + "counted", + "year", + "grant", + "maximum", + "eligible", + "vest", + "Also", + "13", + "923", + "655", + "222", + "268", + "Our", + "participate" + ], + "keywords": [ + "RSUs", + "PSUs", + "ESPP" + ], + "connect_to": [ + { + "target": "1dd1cd13-acac-5a60-9c73-2bb038680798", + "relation": "embeds", + "ref": "[tables/table-59 Equity Plan Table.html]", + "position": { + "start": 324, + "end": 364 + } + } + ] + } + }, + { + "chunk_id": "4a3b7c5f-e220-5bfb-a76a-14bb78647efa", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Additional Information", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "483e15a5-d4a5-5871-9b98-b6d182bf2cfe", + "type": "text", + "content": "Section 16(a) of the Exchange Act requires our executive officers, directors and persons who own more than 10% of a registered class of our equity securities to file initial reports of ownership and reports of changes in ownership of our common stock and other equity securities with the SEC. Executive officers, directors and greater than 10% stockholders are required by SEC regulations to furnish us with copies of all Section 16(a) forms they file. To our knowledge, based solely on a review of the copies of such reports furnished to us and written representations that no other reports were required, all Section 16(a) filing requirements applicable to individuals who were, during Fiscal 2026, our executive officers, directors and greater than 10% beneficial owners were complied with.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Additional Information/Delinquent Section 16(a) Reports", + "metadata": { + "length": 793, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Section", + "16", + "Exchange", + "Act", + "requires", + "executive", + "officers", + "directors", + "persons", + "10%", + "registered", + "class", + "equity", + "securities", + "file", + "initial", + "reports", + "ownership", + "common", + "stock", + "SEC", + "Executive", + "greater", + "stockholders", + "required", + "regulations", + "furnish", + "copies", + "forms", + "To", + "knowledge", + "based", + "solely", + "review", + "furnished", + "written", + "representations", + "filing", + "requirements", + "applicable", + "individuals", + "Fiscal", + "2026", + "beneficial", + "owners", + "complied" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5f347b1c-d05f-5525-b19c-41a9d41a6400", + "type": "text", + "content": "The Board knows of no other matters that will be presented for consideration at the 2026 Meeting. If any other matters are properly brought before the 2026 Meeting, it is the intention of the persons named in the accompanying proxy to vote on such matters in accordance with their best judgment.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Additional Information/Other Matters", + "metadata": { + "length": 295, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Board", + "matters", + "presented", + "consideration", + "2026", + "Meeting", + "If", + "properly", + "brought", + "intention", + "persons", + "named", + "accompanying", + "proxy", + "vote", + "accordance", + "judgment" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "bd6bca4b-fc4c-5242-9fce-f500d0a68d2c", + "type": "image", + "content": "\n### Delinquent Section 16(a) Reports Timothy S. Teter0\n[images/image-81-### Delinq.jpg]\n", + "path": "images/image-81-### Delinq.jpg", + "metadata": { + "length": 89, + "summary": "image-81\n### Delinquent Section 16(a) Reports Timothy S. Teter0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-81-### Delinq.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "14129253-17df-57cc-94d2-dc020ea9708f", + "type": "text", + "content": "### Delinquent Section 16(a) Reports Timothy S. Teter0\n[images/image-81-### Delinq.jpg]\n\nTimothy S. Teter\nSecretary\nMay 12, 2026\nA COPY OF OUR ANNUAL REPORT ON FORM 10-K FOR THE FISCAL YEAR ENDED JANUARY 25, 2026 AS FILED WITH THE SEC IS BEING FURNISHED TO STOCKHOLDERS CONCURRENTLY HEREWITH. UPON WRITTEN REQUEST, WE WILL PROVIDE, WITHOUT CHARGE, AN ADDITIONAL COPY OF THE ANNUAL REPORT. STOCKHOLDERS MAY SUBMIT THEIR REQUESTS TO: INVESTOR RELATIONS, NVIDIA CORPORATION, 2788 SAN TOMAS EXPRESSWAY, SANTA CLARA, CALIFORNIA 95051 OR TO SHAREHOLDERMEETING@NVIDIA.COM. WE WILL ALSO FURNISH A COPY OF ANY EXHIBIT TO THE ANNUAL REPORT ON FORM 10-K IF SPECIFICALLY REQUESTED IN WRITING.\nNVIDIA and the NVIDIA logo are either registered trademarks or trademarks of NVIDIA Corporation in the United States and other countries. Other company names used in this publication are for identification purposes only and may be trademarks of their respective companies.", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/By Order of the Board of Directors", + "metadata": { + "length": 953, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Delinquent", + "Section", + "16", + "Reports", + "Timothy", + "Teter0", + "Teter", + "Secretary", + "May", + "12", + "2026", + "COPY", + "OF", + "OUR", + "ANNUAL", + "REPORT", + "ON", + "FORM", + "10", + "FOR", + "THE", + "FISCAL", + "YEAR", + "ENDED", + "JANUARY", + "25", + "AS", + "FILED", + "WITH", + "SEC", + "IS", + "BEING", + "FURNISHED", + "TO", + "STOCKHOLDERS", + "CONCURRENTLY", + "HEREWITH", + "UPON", + "WRITTEN", + "REQUEST", + "WE", + "WILL", + "PROVIDE", + "WITHOUT", + "CHARGE", + "AN", + "ADDITIONAL", + "MAY", + "SUBMIT", + "THEIR", + "REQUESTS", + "INVESTOR", + "RELATIONS", + "NVIDIA", + "CORPORATION", + "2788", + "SAN", + "TOMAS", + "EXPRESSWAY", + "SANTA", + "CLARA", + "CALIFORNIA", + "95051", + "OR", + "SHAREHOLDERMEETING", + "COM", + "ALSO", + "FURNISH", + "ANY", + "EXHIBIT", + "IF", + "SPECIFICALLY", + "REQUESTED", + "IN", + "WRITING", + "logo", + "registered", + "trademarks", + "Corporation", + "United", + "States", + "countries", + "Other", + "company", + "names", + "publication", + "identification", + "purposes", + "respective", + "companies" + ], + "keywords": [], + "connect_to": [ + { + "target": "bd6bca4b-fc4c-5242-9fce-f500d0a68d2c", + "relation": "embeds", + "ref": "[images/image-81-### Delinq.jpg]", + "position": { + "start": 55, + "end": 87 + } + } + ] + } + }, + { + "chunk_id": "a2fa6c16-74f6-54a7-912e-08600c236cea", + "type": "image", + "content": "\nANNUAL REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 NVIDIA.0\n[images/image-82-ANNUAL REP.jpg]\n", + "path": "images/image-82-ANNUAL REP.jpg", + "metadata": { + "length": 128, + "summary": "image-82\nANNUAL REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 NVIDIA.0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-82-ANNUAL REP.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "286d46ba-7e69-5a62-b5f1-0a0a29e6d05a", + "type": "text", + "content": "ANNUAL REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934\nFor the fiscal year ended January 25, 2026\nOR\n☐ TRANSITION REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934\nCommission file number: 0-23985\n\nANNUAL REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 NVIDIA.0\n[images/image-82-ANNUAL REP.jpg]\n\nNVIDIA.", + "path": "2026-Annual-Report-Web.pdf/FORM 10-K", + "metadata": { + "length": 390, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "ANNUAL", + "REPORT", + "PURSUANT", + "TO", + "SECTION", + "13", + "OR", + "15", + "OF", + "THE", + "SECURITIES", + "EXCHANGE", + "ACT", + "1934", + "For", + "fiscal", + "year", + "ended", + "January", + "25", + "2026", + "TRANSITION", + "Commission", + "file", + "number", + "23985", + "NVIDIA.0", + "NVIDIA" + ], + "keywords": [], + "connect_to": [ + { + "target": "a2fa6c16-74f6-54a7-912e-08600c236cea", + "relation": "embeds", + "ref": "[images/image-82-ANNUAL REP.jpg]", + "position": { + "start": 349, + "end": 381 + } + } + ] + } + }, + { + "chunk_id": "95cd45c8-5576-5c31-ab59-fdeb6c868a09", + "type": "text", + "content": "(Exact name of registrant as specified in its charter)\nDelaware\n(State or other jurisdiction of\nincorporation or organization)\n94-3177549\n(I.R.S. Employer\nIdentification No.)\n2788 San Tomas Expressway, Santa Clara, California\n(Address of principal executive offices)\n95051\n(Zip Code)\nRegistrant's telephone number, including area code: (408) 486-2000\nSecurities registered pursuant to Section 12(b) of the Act:\nTitle of each class\nCommon Stock, \\$0.001 par value per share\nTrading Symbol(s)\nNVDA\nName of each exchange on which registered\nThe Nasdaq Global Select Market\nSecurities registered pursuant to Section 12(g) of the Act:\nNone\nIndicate by check mark if the registrant is a well-known seasoned issuer, as defined in Rule 405 of the Securities Act. Yes ☒ No □\nIndicate by check mark if the registrant is not required to file reports pursuant to Section 13 or Section 15(d) of the Act. Yes □ No ☒\nIndicate by check mark whether the registrant (1) has filed all reports required to be filed by Section 13 or 15(d) of the Securities Exchange Act of 1934 during the preceding 12 months (or for such shorter period that the registrant was required to file such reports), and (2) has been subject to such filing requirements for the past 90 days. Yes ☑ No □\nIndicate by check mark whether the registrant has submitted electronically every Interactive Data File required to be submitted pursuant to Rule 405 of Regulation S-T ( $§232.405$ of this chapter) during the preceding 12 months (or for such shorter period that the registrant was required to submit such files). Yes ☑ No □\nIndicate by check mark whether the registrant is a large accelerated filer, an accelerated filer, a non-accelerated filer, a smaller reporting company, or an emerging growth company. See the definitions of \"large accelerated filer,\" \"accelerated filer,\" \"smaller reporting company,\" and \"emerging growth company\" in Rule 12b-2 of the Exchange Act.\nLarge accelerated filer ☒ Accelerated filer ☐ Non-accelerated filer ☐ Smaller reporting company ☐ Emerging growth company ☐\nIf an emerging growth company, indicate by check mark if the registrant has elected not to use the extended transition period for complying with any new or revised financial accounting standards provided pursuant to Section 13(a) of the Exchange Act. □\nIndicate by check mark whether the registrant has filed a report on and attestation to its management's assessment of the effectiveness of its internal control over financial reporting under Section 404(b) of the Sarbanes-Oxley Act (15 U.S.C. 7262(b)) by the registered public accounting firm that prepared or issued its audit report. ☒\nIf securities are registered pursuant to Section 12(b) of the Act, indicate by check mark whether the financial statements of the registrant included in the filing reflect the correction of an error to previously issued financial statements. □\nIndicate by check mark whether any of those error corrections are restatements that required a recovery analysis of incentive-based compensation received by any of the registrant's executive officers during the relevant recovery period pursuant to §240.10D-1(b). □\nIndicate by check mark whether the registrant is a shell company (as defined in Rule 12b-2 of the Act). Yes □ No ✗\nThe aggregate market value of the voting stock held by non-affiliates of the registrant as of July 25, 2025 was approximately \\$4.0 trillion (based on the closing sales price of the registrant's common stock as reported by the Nasdaq Global Select Market on July 25, 2025). This calculation excludes 1.0 billion shares held by directors and executive officers of the registrant. This calculation does not exclude shares held by such organizations whose ownership exceeds 5% of the registrant's outstanding common stock that have represented to the registrant that they are registered investment advisers or investment companies registered under section 8 of the Investment Company Act of 1940.\nThe number of shares of common stock outstanding as of February 20, 2026 was 24.3 billion.", + "path": "2026-Annual-Report-Web.pdf/FORM 10-K/NVIDIA CORPORATION", + "metadata": { + "length": 4051, + "summary": "This is a Securities and Exchange Commission (SEC) filing for NVIDIA Corporation, a Delaware corporation with I.R.S. Employer Identification No. 94-3177549. Its principal executive offices are located at 2788 San Tomas Expressway, Santa Clara, California 95051, and its telephone number is (408) 486-2000. The registrant's common stock, with a par value of $0.001 per share, is registered under Section 12(b) of the Act and trades under the symbol NVDA on The Nasdaq Global Select Market. The filing includes various checkmarks indicating that NVIDIA is a well-known seasoned issuer, has filed all required reports, and has submitted electronic data files. It is classified as a large accelerated filer. As of July 25, 2025, the aggregate market value of voting stock held by non-affiliates was approximately $4.0 trillion, excluding shares held by directors and executive officers. As of February 20, 2026, there were 24.3 billion shares of common stock outstanding.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Exact", + "registrant", + "charter", + "Delaware", + "State", + "jurisdiction", + "incorporation", + "organization", + "94", + "3177549", + "Employer", + "Identification", + "No", + "2788", + "San", + "Tomas", + "Expressway", + "Santa", + "Clara", + "California", + "Address", + "principal", + "executive", + "offices", + "95051", + "Zip", + "Code", + "Registrant", + "telephone", + "number", + "including", + "area", + "code", + "408", + "486", + "2000", + "Securities", + "registered", + "pursuant", + "Section", + "12", + "Act", + "Title", + "class", + "Common", + "Stock", + "0.001", + "par", + "share", + "Trading", + "Symbol", + "NVDA", + "Name", + "exchange", + "The", + "Nasdaq", + "Global", + "Select", + "Market", + "None", + "Indicate", + "check", + "mark", + "seasoned", + "issuer", + "defined", + "Rule", + "405", + "Yes", + "required", + "file", + "reports", + "13", + "15", + "filed", + "Exchange", + "1934", + "preceding", + "months", + "shorter", + "period", + "subject", + "filing", + "requirements", + "past", + "90", + "days", + "submitted", + "electronically", + "Interactive", + "Data", + "File", + "Regulation", + "232.405", + "chapter", + "submit", + "files", + "large", + "accelerated", + "filer", + "smaller", + "reporting", + "company", + "emerging", + "growth", + "See", + "definitions", + "12b", + "Large", + "Accelerated", + "Non", + "Smaller", + "Emerging", + "If", + "elected", + "extended", + "transition", + "complying", + "revised", + "financial", + "accounting", + "standards", + "provided", + "report", + "attestation", + "management", + "assessment", + "effectiveness", + "internal", + "control", + "404", + "Sarbanes", + "Oxley", + "7262", + "public", + "firm", + "prepared", + "issued", + "audit", + "securities", + "statements", + "included", + "reflect", + "correction", + "error", + "previously", + "corrections", + "restatements", + "recovery", + "analysis", + "incentive", + "based", + "compensation", + "received", + "officers", + "relevant", + "240.10", + "shell", + "aggregate", + "market", + "voting", + "stock", + "held", + "affiliates", + "July", + "25", + "2025", + "approximately", + "4.0", + "trillion", + "closing", + "sales", + "price", + "common", + "reported", + "This", + "calculation", + "excludes", + "1.0", + "billion", + "shares", + "directors", + "exclude", + "organizations", + "ownership", + "exceeds", + "5%", + "outstanding", + "represented", + "investment", + "advisers", + "companies", + "section", + "Investment", + "Company", + "1940", + "February", + "20", + "2026", + "24.3" + ], + "keywords": [ + "NVIDIA", + "SEC Filing", + "Common Stock" + ], + "connect_to": [] + } + }, + { + "chunk_id": "dace29a7-079e-58f8-a299-30a80389a3cc", + "type": "text", + "content": "Portions of the registrant's Proxy Statement for its 2026 Annual Meeting of Shareholders to be filed with the Securities and Exchange Commission pursuant to Regulation 14A not later than 120 days after the end of the fiscal year covered by this Annual Report on Form 10-K are incorporated by reference into Part III, Items 10-14 of this Annual Report on Form 10-K.", + "path": "2026-Annual-Report-Web.pdf/FORM 10-K/DOCUMENTS INCORPORATED BY REFERENCE", + "metadata": { + "length": 364, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Portions", + "registrant", + "Proxy", + "Statement", + "2026", + "Annual", + "Meeting", + "Shareholders", + "filed", + "Securities", + "Exchange", + "Commission", + "pursuant", + "Regulation", + "14A", + "120", + "days", + "end", + "fiscal", + "year", + "covered", + "Report", + "Form", + "10", + "incorporated", + "reference", + "Part", + "III", + "Items", + "14" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "484847d9-5615-5826-b7c8-80637c7ab34d", + "type": "text", + "content": "Item 1. Business 4\nItem 1A. Risk Factors 12\nItem 1B. Unresolved Staff Comments 32\nItem 1C Cybersecurity 32\nItem 2. Properties 33\nItem 3. Legal Proceedings 33\nItem 4. Mine Safety Disclosures 33", + "path": "2026-Annual-Report-Web.pdf/Part I", + "metadata": { + "length": 192, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Item", + "Business", + "1A", + "Risk", + "Factors", + "12", + "1B", + "Unresolved", + "Staff", + "Comments", + "32", + "1C", + "Cybersecurity", + "Properties", + "33", + "Legal", + "Proceedings", + "Mine", + "Safety", + "Disclosures" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b9d7cb6b-238c-532a-b0f2-b5930edf7653", + "type": "text", + "content": "Item 5. Market for Registrant’s Common Equity, Related Stockholder Matters and Issuer Purchases of Equity Securities 33\nItem 6. [Reserved] 35\nItem 7. Management's Discussion and Analysis of Financial Condition and Results of Operations 36\nItem 7A. Quantitative and Qualitative Disclosures About Market Risk 44\nItem 8. Financial Statements and Supplementary Data 45\nItem 9. Changes in and Disagreements with Accountants on Accounting and Financial Disclosure 45\nItem 9A. Controls and Procedures 45\nItem 9B. Other Information 46\nItem 9C. Disclosure Regarding Foreign Jurisdictions that Prevent Inspections 46", + "path": "2026-Annual-Report-Web.pdf/Part II", + "metadata": { + "length": 606, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Item", + "Market", + "Registrant", + "Common", + "Equity", + "Related", + "Stockholder", + "Matters", + "Issuer", + "Purchases", + "Securities", + "33", + "Reserved", + "35", + "Management", + "Discussion", + "Analysis", + "Financial", + "Condition", + "Results", + "Operations", + "36", + "7A", + "Quantitative", + "Qualitative", + "Disclosures", + "About", + "Risk", + "44", + "Statements", + "Supplementary", + "Data", + "45", + "Changes", + "Disagreements", + "Accountants", + "Accounting", + "Disclosure", + "9A", + "Controls", + "Procedures", + "9B", + "Other", + "Information", + "46", + "9C", + "Regarding", + "Foreign", + "Jurisdictions", + "Prevent", + "Inspections" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "44d30c3d-e4da-5b38-80a2-046885038be4", + "type": "text", + "content": "Item 10. Directors, Executive Officers and Corporate Governance 46\nItem 11. Executive Compensation 47\nItem 12. Security Ownership of Certain Beneficial Owners and Management and Related Stockholder Matters 47\nItem 13. Certain Relationships and Related Transactions, and Director Independence 47\nItem 14. Principal Accountant Fees and Services 47", + "path": "2026-Annual-Report-Web.pdf/Part III", + "metadata": { + "length": 345, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Item", + "10", + "Directors", + "Executive", + "Officers", + "Corporate", + "Governance", + "46", + "11", + "Compensation", + "47", + "12", + "Security", + "Ownership", + "Certain", + "Beneficial", + "Owners", + "Management", + "Related", + "Stockholder", + "Matters", + "13", + "Relationships", + "Transactions", + "Director", + "Independence", + "14", + "Principal", + "Accountant", + "Fees", + "Services" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "33c3fc40-f289-5b5f-90ad-d74bfd858463", + "type": "text", + "content": "Item 15. Exhibits and Financial Statement Schedules 48\nItem 16. Form 10-K Summary 83\nSignatures 84", + "path": "2026-Annual-Report-Web.pdf/Part IV", + "metadata": { + "length": 98, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Item", + "15", + "Exhibits", + "Financial", + "Statement", + "Schedules", + "48", + "16", + "Form", + "10", + "Summary", + "83", + "Signatures", + "84" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "37254e04-7380-5492-b360-095d00152a99", + "type": "text", + "content": "Investors and others should note that we announce material financial information to our investors using our investor relations website, press releases, SEC filings and public conference calls and webcasts. We also use the following social media channels as a means of disclosing information about the company, our products, our planned financial and other announcements and attendance at upcoming investor and industry conferences, and other matters and for complying with our disclosure obligations under Regulation FD:\nNVIDIA Corporate Blog (blogs.nvidia.com/)\nNVIDIA Technical Blog (developer.nvidia.com/blog/)\nNVIDIA LinkedIn (linkedin.com/company/nvidia)\nNVIDIA Facebook (facebook.com/nvidia)\nNVIDIA Instagram (instagram.com/nvidia)\nNVIDIA X (x.com/nvidia)\nNVIDIA Threads (threads.com/@nvidia)\nNVIDIA Investor Relations (investor.nvidia.com)\nNVIDIA YouTube (YouTube.com/nvidia).\nThe information we post through these social media channels may be deemed material. Accordingly, investors should monitor these channels, in addition to following our press releases, SEC filings and public conference calls and webcasts. This list may be updated from time to time. The information we post through these channels is not a part of this Annual Report on Form 10-K.", + "path": "2026-Annual-Report-Web.pdf/Part IV/Where You Can Find More Information", + "metadata": { + "length": 1261, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Investors", + "note", + "announce", + "material", + "financial", + "information", + "investors", + "investor", + "relations", + "website", + "press", + "releases", + "SEC", + "filings", + "public", + "conference", + "calls", + "webcasts", + "We", + "social", + "media", + "channels", + "means", + "disclosing", + "company", + "products", + "planned", + "announcements", + "attendance", + "upcoming", + "industry", + "conferences", + "matters", + "complying", + "disclosure", + "obligations", + "Regulation", + "FD", + "NVIDIA", + "Corporate", + "Blog", + "blogs", + "nvidia", + "Technical", + "developer", + "blog", + "LinkedIn", + "linkedin", + "Facebook", + "facebook", + "Instagram", + "instagram", + "Threads", + "threads", + "Investor", + "Relations", + "YouTube", + "The", + "post", + "deemed", + "Accordingly", + "monitor", + "addition", + "This", + "list", + "updated", + "time", + "part", + "Annual", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "64b09c92-ef6c-5045-920d-b4f242da3ef7", + "type": "text", + "content": "This Annual Report on Form 10-K contains forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the “safe harbor” created by those sections based on management’s beliefs and assumptions and on information currently available to our management. In some cases, you can identify forward-looking statements by terms such as “may,” “will,” “should,” “could,” “goal,” “would,” “expect,” “plan,” “anticipate,” “believe,” “estimate,” “project,” “predict,” “potential,” and similar expressions intended to identify forward-looking statements. These statements involve known and unknown risks, uncertainties and other factors, which may cause our actual results, performance, time frames or achievements to be materially different from any future results, performance, time frames or achievements expressed or implied by the forward-looking statements. We discuss many of these risks, uncertainties, and other factors in this Annual Report on Form 10-K in greater detail under the heading “Risk Factors.” Given these risks, uncertainties, and other factors, you should not place undue reliance on these forward-looking statements. Also, these forward-looking statements represent our estimates and assumptions only as of the date of this filing. You should read this Annual Report on Form 10-K completely and understand that our actual future results may be materially different from what we expect. We hereby qualify our forward-looking statements by these cautionary statements. Except as required by law, we assume no obligation to update these forward-looking statements publicly, or to update the reasons actual results could differ materially from those anticipated in these forward-looking statements, even if new information becomes available in the future.\nAll references to “NVIDIA,” “we,” “us,” “our,” or the “Company” mean NVIDIA Corporation and its subsidiaries.\nIn addition, statements that “we believe” and similar statements reflect our beliefs and opinions on the relevant subject. These statements are based upon information available to us as of the filing date of this Annual Report on Form 10-K, and while we believe such information forms a reasonable basis for such statements, such information may be limited or incomplete, and our statements should not be read to indicate that we have conducted an exhaustive inquiry into, or review of, all potentially available relevant information. These statements are inherently uncertain, and investors are cautioned not to unduly rely upon these statements.\n© 2026 NVIDIA Corporation. All rights reserved.", + "path": "2026-Annual-Report-Web.pdf/Part IV/Forward-Looking Statements", + "metadata": { + "length": 2699, + "summary": "This passage from an Annual Report on Form 10-K discusses forward-looking statements, which are based on management's beliefs and assumptions and are subject to risks and uncertainties. It identifies terms that indicate forward-looking statements, such as 'may,' 'will,' 'expect,' and 'believe.' The text warns that actual results may differ materially from those expressed or implied, and advises against undue reliance on these statements. It also clarifies that references to 'NVIDIA,' 'we,' 'us,' 'our,' or the 'Company' mean NVIDIA Corporation and its subsidiaries, and that beliefs are based on information available as of the filing date. The passage concludes with a copyright notice for 2026 NVIDIA Corporation.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "This", + "Annual", + "Report", + "Form", + "10", + "forward", + "statements", + "meaning", + "Section", + "27A", + "Securities", + "Act", + "1933", + "amended", + "21E", + "Exchange", + "1934", + "subject", + "safe", + "harbor", + "created", + "sections", + "based", + "management", + "beliefs", + "assumptions", + "information", + "In", + "cases", + "identify", + "terms", + "goal", + "expect", + "plan", + "anticipate", + "estimate", + "project", + "predict", + "potential", + "similar", + "expressions", + "intended", + "These", + "involve", + "unknown", + "risks", + "uncertainties", + "factors", + "actual", + "results", + "performance", + "time", + "frames", + "achievements", + "materially", + "future", + "expressed", + "implied", + "We", + "discuss", + "greater", + "detail", + "heading", + "Risk", + "Factors", + "Given", + "place", + "undue", + "reliance", + "Also", + "represent", + "estimates", + "date", + "filing", + "You", + "read", + "completely", + "understand", + "qualify", + "cautionary", + "Except", + "required", + "law", + "assume", + "obligation", + "update", + "publicly", + "reasons", + "differ", + "anticipated", + "All", + "references", + "NVIDIA", + "Company", + "Corporation", + "subsidiaries", + "addition", + "reflect", + "opinions", + "relevant", + "forms", + "reasonable", + "basis", + "limited", + "incomplete", + "conducted", + "exhaustive", + "inquiry", + "review", + "potentially", + "inherently", + "uncertain", + "investors", + "cautioned", + "unduly", + "rely", + "2026", + "rights", + "reserved" + ], + "keywords": [ + "forward-looking", + "risks", + "cautionary" + ], + "connect_to": [] + } + }, + { + "chunk_id": "e39ab3c3-f70f-5e8a-82ec-cc31a121a388", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part I_2", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "70e05e31-4245-5ea3-823f-e74bf108b032", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "310f6904-8a15-5b7b-bae6-c59143c2a015", + "type": "text", + "content": "NVIDIA pioneered accelerated computing to help solve the most challenging computational problems. NVIDIA is now a data center scale AI infrastructure company reshaping all industries.\nOur technology stack includes the foundational NVIDIA CUDA development platform that runs on all NVIDIA GPUs, as well as hundreds of domain-specific software libraries, frameworks, algorithms, software development kits, or SDKs, and application programming interfaces, or APIs. This deep and broad software stack accelerates the performance and facilitates the deployment of NVIDIA accelerated computing for computationally intensive workloads such as artificial intelligence, or AI, model training and inference, data analytics, scientific computing, robotics, and 3D graphics, with vertical-specific optimizations to address industries ranging from healthcare and telecom to automotive and manufacturing.\nIntroduced with the Blackwell architecture, our data-center-scale offerings feature extreme co-design where the infrastructure's chips, networking, systems, software, and algorithms are holistically architected and optimized to maximize performance and scale. Hundreds of thousands of GPUs can be interconnected to function as a single giant computer. This type of data center architecture and scale is needed for the development and deployment of modern AI and accelerated computing applications.\nThe GPU was initially used to simulate human imagination, enabling the virtual worlds of video games and films. Today, it also simulates human intelligence, enabling a deeper understanding of language, science, and the physical world. Its parallel processing capabilities, supported by tens of thousands of computing cores, are essential for deep learning algorithms. This form of AI, in which software writes itself by learning from large amounts of data, can serve as the brain of computers, robots, and self-driving cars that can perceive, understand and reason about the world. GPU-powered AI solutions are being developed by thousands of enterprises to deliver services and products that would have been immensely difficult or even impossible with traditional coding. Examples include generative AI, which can create new content such as text, code, images, audio, video, molecule structures, and recommendation systems; and agentic AI where systems of AI models work in concert to automatically complete a task.\nNVIDIA has a platform strategy, bringing together hardware, systems, software, algorithms, libraries, AI models and training data sets, and services to create unique value for the markets we serve. While the computing requirements of these end markets are diverse, we address them with a unified underlying programmable architecture allowing us to support several multi-billion-dollar end markets with the same underlying technology by using a variety of software stacks developed either internally or by third-party developers and partners. The large and growing number of developers and installed base across our platforms strengthens our ecosystem and increases the value of our platform for our customers.\nInnovation is at our core. We have invested over \\$76.7 billion in research and development since our inception, yielding inventions that are essential to modern computing. Our invention of the GPU in 1999 sparked the growth of the PC gaming market and redefined computer graphics. With our introduction of CUDA in 2006, we opened the parallel processing capabilities of our GPU to a broad range of compute-intensive applications, paving the way for the emergence of modern AI. In 2012, the AlexNet neural network, trained on NVIDIA GPUs, won the ImageNet computer image recognition competition, marking the “Big Bang” moment of AI. We introduced our first Tensor Core GPU in 2017, built from the ground-up for the new era of AI, and our first autonomous driving system-on-chips, or SoC, in 2018. Our acquisition of Mellanox in 2020 expanded our offerings to include networking, enabled our platforms to be data center scale, and led to the introduction of a new processor class – the data processing unit, or DPU. Over the past 5 years, we have built full software stacks that run on top of our GPUs and CUDA to bring AI to the world’s largest industries, including NVIDIA DRIVE stack for autonomous driving, Clara for healthcare, Omniverse for physical AI applications, and NVIDIA AI Enterprise software – essentially an operating system for enterprise AI applications. In 2023, we introduced our first data center CPU, Grace, built for giant-scale AI and high-performance computing, or HPC. In 2024, we launched the NVIDIA Blackwell architecture – connecting 36 Grace CPUs and 72 Blackwell GPUs in a data center scale, liquid-cooled design – for real-time trillion-parameter inference and training. In fiscal year 2026, we launched and scaled the NVIDIA Blackwell Ultra platform, optimized for agentic, reasoning, and physical AI. Building on the architectural breakthroughs of Blackwell and leveraging Dynamo inference software, it delivers a significant increase in token throughput and reduction in cost per token compared to the Hopper generation. More recently, in support of market development, we have accelerated the release cadence of our open AI model platforms including NVIDIA Nemotron for agentic AI and Cosmos for physical AI. With a strong engineering culture, we drive fast, yet harmonized, product and technology innovations in all dimensions of computing including silicon, systems, networking, software and algorithms. More than half of our engineers work on software.\nAll major cloud service providers, or CSPs, AI model makers, and enterprises use our data center-scale infrastructure and computing platforms to accelerate the services and offerings they deliver to billions of end users and customers, including AI solutions and assistants, AI foundation models, advertising, search, recommendation systems, social networking, data processing, online shopping, live video, and translation. AI model makers use our infrastructure and software hosted at CSPs to develop, build and run AI models, product offerings, and services.\nEnterprises and startups across a broad range of industries use our accelerated computing platforms to build new generative and agentic AI-enabled products and services, and/or to dramatically accelerate and reduce the costs of their workloads and workflows. The enterprise software industry uses them for new AI assistants, chatbots, and agents; the transportation industry for autonomous driving; the healthcare industry for accelerated and computer-aided drug discovery; and the financial services industry for customer support and fraud detection.\nResearchers and developers use our computing solutions to accelerate a wide range of important applications, from simulating molecular dynamics to climate forecasting. With support for 6,000 applications, NVIDIA computing enables some of the most promising areas of discovery, from climate prediction to materials science and from wind tunnel simulation to genomics. Including GPUs and networking, NVIDIA powers over 78% of the supercomputers on the global TOP500 list, including 9 of the top 10 systems on the Green500 list.\nGamers choose NVIDIA GPUs to enjoy immersive, increasingly cinematic virtual worlds. In addition to serving the growing number of gamers, the market for PC GPUs is expanding because of the growing population of live streamers, broadcasters, artists, and creators. With the advent of generative and agentic AI, we expect a broader set of PC users to choose NVIDIA GPUs for running these applications locally on their PC, which is critical for privacy, latency, and cost-sensitive AI applications.\nProfessional artists, architects and designers use NVIDIA partner products accelerated with our GPUs and software platform for a range of creative, engineering, and design use cases, such as creating visual effects in movies or designing buildings and products. In addition, generative and agentic AI is expanding the market for our workstation-class GPUs, as more enterprise customers develop and deploy AI applications with their data on-premises.\nHeadquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Company", + "metadata": { + "length": 8331, + "summary": "NVIDIA is a data center scale AI infrastructure company. Its technology stack includes the CUDA platform and domain-specific software libraries for AI, data analytics, scientific computing, robotics, and 3D graphics. The Blackwell architecture enables extreme co-design, interconnecting hundreds of thousands of GPUs. GPU-powered AI solutions are used for generative AI, agentic AI, and autonomous driving. NVIDIA has invested over $76.7 billion in R&D, inventing the GPU in 1999 and CUDA in 2006. Its platforms are used by cloud service providers, AI model makers, enterprises, researchers, gamers, and professionals across industries including healthcare, automotive, and financial services.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "NVIDIA", + "pioneered", + "accelerated", + "computing", + "solve", + "challenging", + "computational", + "problems", + "data", + "center", + "scale", + "AI", + "infrastructure", + "company", + "reshaping", + "industries", + "Our", + "technology", + "stack", + "includes", + "foundational", + "CUDA", + "development", + "platform", + "runs", + "GPUs", + "hundreds", + "domain", + "specific", + "software", + "libraries", + "frameworks", + "algorithms", + "kits", + "SDKs", + "application", + "programming", + "interfaces", + "APIs", + "This", + "deep", + "broad", + "accelerates", + "performance", + "facilitates", + "deployment", + "computationally", + "intensive", + "workloads", + "artificial", + "intelligence", + "model", + "training", + "inference", + "analytics", + "scientific", + "robotics", + "3D", + "graphics", + "vertical", + "optimizations", + "address", + "ranging", + "healthcare", + "telecom", + "automotive", + "manufacturing", + "Introduced", + "Blackwell", + "architecture", + "offerings", + "feature", + "extreme", + "design", + "chips", + "networking", + "systems", + "holistically", + "architected", + "optimized", + "maximize", + "Hundreds", + "thousands", + "interconnected", + "function", + "single", + "giant", + "computer", + "type", + "needed", + "modern", + "applications", + "The", + "GPU", + "initially", + "simulate", + "human", + "imagination", + "enabling", + "virtual", + "worlds", + "video", + "games", + "films", + "Today", + "simulates", + "deeper", + "understanding", + "language", + "science", + "physical", + "world", + "Its", + "parallel", + "processing", + "capabilities", + "supported", + "tens", + "cores", + "essential", + "learning", + "form", + "writes", + "large", + "amounts", + "serve", + "brain", + "computers", + "robots", + "driving", + "cars", + "perceive", + "understand", + "reason", + "powered", + "solutions", + "developed", + "enterprises", + "deliver", + "services", + "products", + "immensely", + "difficult", + "impossible", + "traditional", + "coding", + "Examples", + "include", + "generative", + "create", + "content", + "text", + "code", + "images", + "audio", + "molecule", + "structures", + "recommendation", + "agentic", + "models", + "work", + "concert", + "automatically", + "complete", + "task", + "strategy", + "bringing", + "hardware", + "sets", + "unique", + "markets", + "While", + "requirements", + "end", + "diverse", + "unified", + "underlying", + "programmable", + "allowing", + "support", + "multi", + "billion", + "dollar", + "variety", + "stacks", + "internally", + "party", + "developers", + "partners", + "growing", + "number", + "installed", + "base", + "platforms", + "strengthens", + "ecosystem", + "increases", + "customers", + "Innovation", + "core", + "We", + "invested", + "76.7", + "research", + "inception", + "yielding", + "inventions", + "invention", + "1999", + "sparked", + "growth", + "PC", + "gaming", + "market", + "redefined", + "With", + "introduction", + "2006", + "opened", + "range", + "compute", + "paving", + "emergence", + "In", + "2012", + "AlexNet", + "neural", + "network", + "trained", + "won", + "ImageNet", + "image", + "recognition", + "competition", + "marking", + "Big", + "Bang", + "moment", + "introduced", + "Tensor", + "Core", + "2017", + "built", + "ground", + "era", + "autonomous", + "system", + "SoC", + "2018", + "acquisition", + "Mellanox", + "2020", + "expanded", + "enabled", + "led", + "processor", + "class", + "unit", + "DPU", + "Over", + "past", + "years", + "full", + "run", + "top", + "bring", + "largest", + "including", + "DRIVE", + "Clara", + "Omniverse", + "Enterprise", + "essentially", + "operating", + "enterprise", + "2023", + "CPU", + "Grace", + "high", + "HPC", + "2024", + "launched", + "connecting", + "36", + "CPUs", + "72", + "liquid", + "cooled", + "real", + "time", + "trillion", + "parameter", + "fiscal", + "year", + "2026", + "scaled", + "Ultra", + "reasoning", + "Building", + "architectural", + "breakthroughs", + "leveraging", + "Dynamo", + "delivers", + "significant", + "increase", + "token", + "throughput", + "reduction", + "cost", + "compared", + "Hopper", + "generation", + "More", + "recently", + "release", + "cadence", + "open", + "Nemotron", + "Cosmos", + "strong", + "engineering", + "culture", + "drive", + "fast", + "harmonized", + "product", + "innovations", + "dimensions", + "silicon", + "half", + "engineers", + "All", + "major", + "cloud", + "service", + "providers", + "CSPs", + "makers", + "accelerate", + "billions", + "users", + "assistants", + "foundation", + "advertising", + "search", + "social", + "online", + "shopping", + "live", + "translation", + "hosted", + "develop", + "build", + "Enterprises", + "startups", + "dramatically", + "reduce", + "costs", + "workflows", + "industry", + "chatbots", + "agents", + "transportation", + "aided", + "drug", + "discovery", + "financial", + "customer", + "fraud", + "detection", + "Researchers", + "wide", + "important", + "simulating", + "molecular", + "dynamics", + "climate", + "forecasting", + "000", + "enables", + "promising", + "areas", + "prediction", + "materials", + "wind", + "tunnel", + "simulation", + "genomics", + "Including", + "powers", + "78%", + "supercomputers", + "global", + "TOP500", + "list", + "10", + "Green500", + "Gamers", + "choose", + "enjoy", + "immersive", + "increasingly", + "cinematic", + "addition", + "serving", + "gamers", + "expanding", + "population", + "streamers", + "broadcasters", + "artists", + "creators", + "advent", + "expect", + "broader", + "set", + "running", + "locally", + "critical", + "privacy", + "latency", + "sensitive", + "Professional", + "architects", + "designers", + "partner", + "creative", + "cases", + "creating", + "visual", + "effects", + "movies", + "designing", + "buildings", + "workstation", + "deploy", + "premises", + "Headquartered", + "Santa", + "California", + "incorporated", + "April", + "1993", + "reincorporated", + "Delaware", + "1998" + ], + "keywords": [ + "NVIDIA", + "GPU", + "AI" + ], + "connect_to": [] + } + }, + { + "chunk_id": "6f53b305-19a9-50b6-852b-471fba2b3101", + "type": "text", + "content": "We report our business results in two segments.\nThe Compute & Networking segment includes our Data Center accelerated computing and networking platforms and AI solutions and software, and Automotive platforms and autonomous and electric vehicle solutions including software.\nThe Graphics segment includes GeForce GPUs for gaming and PCs, and Quadro/NVIDIA RTX GPUs for enterprise workstation graphics.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses", + "metadata": { + "length": 401, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "report", + "business", + "results", + "segments", + "The", + "Compute", + "Networking", + "segment", + "includes", + "Data", + "Center", + "accelerated", + "computing", + "networking", + "platforms", + "AI", + "solutions", + "software", + "Automotive", + "autonomous", + "electric", + "vehicle", + "including", + "Graphics", + "GeForce", + "GPUs", + "gaming", + "PCs", + "Quadro", + "NVIDIA", + "RTX", + "enterprise", + "workstation", + "graphics" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2d38313d-d0b8-5e08-9b52-851e63238eb8", + "type": "text", + "content": "We specialize in markets where our computing and AI infrastructure platforms can provide tremendous acceleration for applications. These platforms incorporate processors, interconnects, software, algorithms, systems, and services to deliver unique value. Our platforms address four large markets where our expertise is critical: Data Center, Gaming, Professional Visualization, and Automotive.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses/Our Markets", + "metadata": { + "length": 393, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "specialize", + "markets", + "computing", + "AI", + "infrastructure", + "platforms", + "provide", + "tremendous", + "acceleration", + "applications", + "These", + "incorporate", + "processors", + "interconnects", + "software", + "algorithms", + "systems", + "services", + "deliver", + "unique", + "Our", + "address", + "large", + "expertise", + "critical", + "Data", + "Center", + "Gaming", + "Professional", + "Visualization", + "Automotive" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9e0ad93a-2987-5d6d-800a-cde3cbb66867", + "type": "text", + "content": "The NVIDIA Data Center platform is focused on accelerating compute-intensive workloads, such as AI, data processing, graphics, robotics, and scientific computing, delivering superior total cost of ownership relative to conventional CPU-only approaches. It is deployed in cloud, hyperscale, on-premises and edge data centers. The platform consists of data center compute and networking infrastructure offerings typically delivered to customers as rack-scale systems, subsystems, or modules, along with software and services.\nOur Data Center infrastructure systems include supercomputing platforms and servers, bringing together our higher performance, energy efficient GPUs, CPUs, interconnects, and fully optimized AI and HPC software stacks. In addition, they include a growing body of acceleration libraries, AI models and training data sets, APIs, SDKs, and domain-specific application frameworks.\nOur networking offerings include NVLink interconnects and switches, end-to-end platforms for InfiniBand and Ethernet, consisting of network adapters, cables, DPUs, switch chips and systems, as well as software. This has enabled us to architect data center-scale computing platforms that can interconnect up to hundreds of thousands of compute nodes with high-performance networking. Fueled by an expansion in AI and HPC workloads, the data center has become the new unit of computing, with networking as an integral part. In fiscal year 2026, we introduced NVIDIA NVLink Fusion to enable hyperscalers and custom ASIC designers to integrate custom CPUs and XPUs with our platform.\nOur customers include all major public and private cloud providers, AI model makers, enterprises and startups, and public sector entities. We work with industry leaders to help build or transform their applications and data center infrastructure. Some of our direct customers include original equipment manufacturers, or OEMs, original device manufacturers, or\nODMs, system integrators and distributors which we partner with to help bring our products to market. We also have partnerships in automotive, healthcare, financial services, manufacturing, retail, and technology among others, to accelerate the adoption of AI.\nAt the foundation of the NVIDIA accelerated computing platform are our GPUs, which excel at parallel workloads such as the training and inferencing of neural networks. These Data Center systems are extreme co-designed with the GPU, CPU, NVLink switch, DPU, NIC, and scale-out networking along with software stacks and algorithms to deliver data center-scale computing solutions.\nWhile our approach starts with powerful chips, what makes it a full-stack computing platform is our large body of software, including the CUDA development platform, the CUDA-X collection of acceleration libraries, AI models and training data sets, APIs, SDKs, and domain-specific application frameworks.\nIn addition to software delivered to customers as an integral part of our data center computing and networking platform, we offer paid licenses to NVIDIA AI Enterprise, a comprehensive suite of enterprise-grade AI software and NVIDIA vGPU software for graphics-rich virtual desktops and workstations.\nIn fiscal year 2025, we launched the NVIDIA Blackwell architecture, a full set of data center scale infrastructure that includes GPUs, CPUs, DPUs, interconnects, switch chips and systems, and networking adapters. Blackwell excels at processing cutting edge generative AI and accelerated computing workloads with market leading performance and efficiency. Offered in a number of configurations, for customers across industries and a diverse set of AI and accelerated computing use cases. In fiscal year 2026, we unveiled the NVIDIA Rubin platform, which is expected to commence production shipments in the second half of fiscal year 2027. Built for agentic AI and reasoning, it excels at processing multi-step problem-solving and massive long-context workflows, delivering up to a 10x reduction in cost per token compared to Blackwell.\nFor physical AI, we provide an end-to-end platform spanning data center infrastructure, open models, systems, embedded compute modules, and software stacks to train, simulate, and deploy advanced automation and robotics solutions.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses/Our Markets/Data Center", + "metadata": { + "length": 4252, + "summary": "The NVIDIA Data Center platform accelerates compute-intensive workloads like AI, data processing, and scientific computing, offering superior total cost of ownership over CPU-only approaches. It includes compute and networking infrastructure for cloud, hyperscale, on-premises, and edge data centers. The platform features GPUs, CPUs, interconnects, and software stacks, with networking offerings like NVLink and InfiniBand. Customers include cloud providers, AI makers, and enterprises. Key architectures include Blackwell (fiscal 2025) and Rubin (fiscal 2027), with Rubin targeting agentic AI and reducing cost per token by 10x. The platform also supports physical AI for robotics.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "NVIDIA", + "Data", + "Center", + "platform", + "focused", + "accelerating", + "compute", + "intensive", + "workloads", + "AI", + "data", + "processing", + "graphics", + "robotics", + "scientific", + "computing", + "delivering", + "superior", + "total", + "cost", + "ownership", + "relative", + "conventional", + "CPU", + "approaches", + "It", + "deployed", + "cloud", + "hyperscale", + "premises", + "edge", + "centers", + "consists", + "center", + "networking", + "infrastructure", + "offerings", + "typically", + "delivered", + "customers", + "rack", + "scale", + "systems", + "subsystems", + "modules", + "software", + "services", + "Our", + "include", + "supercomputing", + "platforms", + "servers", + "bringing", + "higher", + "performance", + "energy", + "efficient", + "GPUs", + "CPUs", + "interconnects", + "fully", + "optimized", + "HPC", + "stacks", + "In", + "addition", + "growing", + "body", + "acceleration", + "libraries", + "models", + "training", + "sets", + "APIs", + "SDKs", + "domain", + "specific", + "application", + "frameworks", + "NVLink", + "switches", + "end", + "InfiniBand", + "Ethernet", + "consisting", + "network", + "adapters", + "cables", + "DPUs", + "switch", + "chips", + "This", + "enabled", + "architect", + "interconnect", + "hundreds", + "thousands", + "nodes", + "high", + "Fueled", + "expansion", + "unit", + "integral", + "part", + "fiscal", + "year", + "2026", + "introduced", + "Fusion", + "enable", + "hyperscalers", + "custom", + "ASIC", + "designers", + "integrate", + "XPUs", + "major", + "public", + "private", + "providers", + "model", + "makers", + "enterprises", + "startups", + "sector", + "entities", + "We", + "work", + "industry", + "leaders", + "build", + "transform", + "applications", + "Some", + "direct", + "original", + "equipment", + "manufacturers", + "OEMs", + "device", + "ODMs", + "system", + "integrators", + "distributors", + "partner", + "bring", + "products", + "market", + "partnerships", + "automotive", + "healthcare", + "financial", + "manufacturing", + "retail", + "technology", + "accelerate", + "adoption", + "At", + "foundation", + "accelerated", + "excel", + "parallel", + "inferencing", + "neural", + "networks", + "These", + "extreme", + "designed", + "GPU", + "DPU", + "NIC", + "algorithms", + "deliver", + "solutions", + "While", + "approach", + "starts", + "powerful", + "makes", + "full", + "stack", + "large", + "including", + "CUDA", + "development", + "collection", + "offer", + "paid", + "licenses", + "Enterprise", + "comprehensive", + "suite", + "enterprise", + "grade", + "vGPU", + "rich", + "virtual", + "desktops", + "workstations", + "2025", + "launched", + "Blackwell", + "architecture", + "set", + "includes", + "excels", + "cutting", + "generative", + "leading", + "efficiency", + "Offered", + "number", + "configurations", + "industries", + "diverse", + "cases", + "unveiled", + "Rubin", + "expected", + "commence", + "production", + "shipments", + "half", + "2027", + "Built", + "agentic", + "reasoning", + "multi", + "step", + "problem", + "solving", + "massive", + "long", + "context", + "workflows", + "10x", + "reduction", + "token", + "compared", + "For", + "physical", + "provide", + "spanning", + "open", + "embedded", + "train", + "simulate", + "deploy", + "advanced", + "automation" + ], + "keywords": [ + "Data Center", + "GPU", + "AI" + ], + "connect_to": [] + } + }, + { + "chunk_id": "fc60efa6-e89d-5bab-a0a7-ffd8f5ebd1ee", + "type": "text", + "content": "Gaming is the largest entertainment industry, with PC gaming as the predominant platform. Many factors propel its growth, including new high production value games, the continued rise of eSports, social connectivity and the increasing popularity of game streamers, modders, or gamers who remaster games, and creators.\nOur products for the gaming market include GeForce RTX GPUs for gaming desktop and laptop PCs, GeForce NOW cloud gaming service, as well as SoCs and development services for game consoles.\nOur gaming platforms leverage our GPUs and sophisticated software to enhance the gaming experience with smoother, higher quality graphics. NVIDIA RTX features ray tracing technology for real-time, cinematic-quality rendering, and deep learning super sampling, or NVIDIA DLSS, our AI technology that boosts frame rates while generating high-quality images for games. RTX GPUs also feature NVIDIA tensor core technology making them well suited to accelerate a new generation of on-device AI applications.\nIn fiscal year 2025, we announced the NVIDIA Blackwell GeForce RTX 50 Series family of desktop and laptop GPUs. The Blackwell architecture introduced neural graphics which combines AI models with traditional rendering to boost game performance, image quality, and interactivity, as well as the next generation of our DLSS technology powered by a new transformer model architecture. In fiscal year 2026, we launched and scaled Blackwell architecture for gaming and GeForce NOW.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses/Our Markets/Gaming", + "metadata": { + "length": 1486, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Gaming", + "largest", + "entertainment", + "industry", + "PC", + "gaming", + "predominant", + "platform", + "Many", + "factors", + "propel", + "growth", + "including", + "high", + "production", + "games", + "continued", + "rise", + "eSports", + "social", + "connectivity", + "increasing", + "popularity", + "game", + "streamers", + "modders", + "gamers", + "remaster", + "creators", + "Our", + "products", + "market", + "include", + "GeForce", + "RTX", + "GPUs", + "desktop", + "laptop", + "PCs", + "NOW", + "cloud", + "service", + "SoCs", + "development", + "services", + "consoles", + "platforms", + "leverage", + "sophisticated", + "software", + "enhance", + "experience", + "smoother", + "higher", + "quality", + "graphics", + "NVIDIA", + "features", + "ray", + "tracing", + "technology", + "real", + "time", + "cinematic", + "rendering", + "deep", + "learning", + "super", + "sampling", + "DLSS", + "AI", + "boosts", + "frame", + "rates", + "generating", + "images", + "feature", + "tensor", + "core", + "making", + "suited", + "accelerate", + "generation", + "device", + "applications", + "In", + "fiscal", + "year", + "2025", + "announced", + "Blackwell", + "50", + "Series", + "family", + "The", + "architecture", + "introduced", + "neural", + "combines", + "models", + "traditional", + "boost", + "performance", + "image", + "interactivity", + "powered", + "transformer", + "model", + "2026", + "launched", + "scaled" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ca845846-902a-5a1f-8d0b-a56fc2856f49", + "type": "text", + "content": "We serve the Professional Visualization market by working closely with independent software vendors, or ISVs, to optimize their offerings for NVIDIA GPUs. Our GPU computing platform enhances productivity and introduces new capabilities for critical workflows in many fields, such as design, engineering, and digital content creation across a wide range of industry verticals. Additionally, the increasing number of generative and agentic AI applications is giving rise to the need for the enhanced AI and data processing capabilities of our RTX PRO GPUs.\nMany leading 3D design and content creation applications developed by our ecosystem partners support RTX, allowing professionals to accelerate and transform their workflows with NVIDIA RTX PRO GPUs and software. As these applications increasingly integrate AI, these GPUs are used and leverage the same Tensor Core technology found in our Data Center solutions.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses/Our Markets/Professional Visualization", + "metadata": { + "length": 916, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "serve", + "Professional", + "Visualization", + "market", + "working", + "closely", + "independent", + "software", + "vendors", + "ISVs", + "optimize", + "offerings", + "NVIDIA", + "GPUs", + "Our", + "GPU", + "computing", + "platform", + "enhances", + "productivity", + "introduces", + "capabilities", + "critical", + "workflows", + "fields", + "design", + "engineering", + "digital", + "content", + "creation", + "wide", + "range", + "industry", + "verticals", + "Additionally", + "increasing", + "number", + "generative", + "agentic", + "AI", + "applications", + "giving", + "rise", + "enhanced", + "data", + "processing", + "RTX", + "PRO", + "Many", + "leading", + "3D", + "developed", + "ecosystem", + "partners", + "support", + "allowing", + "professionals", + "accelerate", + "transform", + "As", + "increasingly", + "integrate", + "leverage", + "Tensor", + "Core", + "technology", + "found", + "Data", + "Center", + "solutions" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c84fe4a4-da19-57ae-8d17-98d9cb2db432", + "type": "text", + "content": "Automotive is comprised of platform solutions for automated driving from the cloud to the car. Leveraging our technology leadership in AI and building on long-standing relationships across several hundred automotive ecosystem partners, we are delivering a full stack end-to-end solution for the AV market under the DRIVE Hyperion platform. This platform consists of development infrastructure, high-performance, energy efficient DRIVE AGX computing hardware running an in-vehicle operating system (DRIVE OS), a reference sensor set that supports full self-driving capability as well as an open, modular DRIVE software platform for autonomous driving, mapping, and parking services, and intelligent in-vehicle experiences.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses/Our Markets/Automotive", + "metadata": { + "length": 721, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Automotive", + "comprised", + "platform", + "solutions", + "automated", + "driving", + "cloud", + "car", + "Leveraging", + "technology", + "leadership", + "AI", + "building", + "long", + "standing", + "relationships", + "hundred", + "automotive", + "ecosystem", + "partners", + "delivering", + "full", + "stack", + "end", + "solution", + "AV", + "market", + "DRIVE", + "Hyperion", + "This", + "consists", + "development", + "infrastructure", + "high", + "performance", + "energy", + "efficient", + "AGX", + "computing", + "hardware", + "running", + "vehicle", + "operating", + "system", + "OS", + "reference", + "sensor", + "set", + "supports", + "capability", + "open", + "modular", + "software", + "autonomous", + "mapping", + "parking", + "services", + "intelligent", + "experiences" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9aef1407-c688-5f24-ad3d-b500fbf1da9d", + "type": "text", + "content": "NVIDIA's key strategies that shape our overall business approach include:\nAdvancing the NVIDIA accelerated computing platform. Our accelerated computing platform can solve complex problems in significantly less time and with lower power consumption than alternative computational approaches. It can help solve problems that were previously deemed unsolvable. We work to deliver continued performance leaps that outpace Moore's Law by leveraging innovation across the architecture, chip design, system, interconnect, algorithm, and software layers. This full-stack innovation approach allows us to deliver order-of-magnitude performance advantages relative to legacy approaches in our target markets, which include Data Center, Gaming, Professional Visualization, and Automotive. While the computing requirements of these end markets are diverse, we address them with a unified underlying architecture leveraging our GPUs, CPUs, CUDA and networking technologies as the fundamental building blocks. The programmable nature of our architecture allows us to make leveraged investments in research and development: we can support several multi-billion-dollar end markets with shared underlying technology by using a variety of software stacks developed either internally or by third-party developers and partners. We utilize this platform approach in each of our target markets.\nExtending our technology and platform leadership in AI. We provide a complete, end-to-end accelerated computing platform for AI, addressing both training and inferencing. This includes full-stack data center-scale compute and networking solutions across processing units, interconnects, systems, and software. Our compute solutions include all three major processing units in AI servers – GPUs, CPUs, and DPUs. GPUs are uniquely suited to AI, and we will continue to add AI-specific features to our GPU architecture to further extend our leadership position.\nIn addition, we offer NVIDIA AI Enterprise—a comprehensive software suite designed to simplify the development and deployment of production-grade, end-to-end generative AI applications. NVIDIA AI Enterprise includes: NVIDIA NIM, which increases token throughput using industry-leading open and proprietary models; NVIDIA NeMo, a complete solution for curating, fine-tuning, reinforcement learning, evaluating, and safeguarding domain-adapted models; and AI Blueprints, pre-built, runnable templates that help enterprises build, optimize, and deploy AI agents while preserving privacy. These tools enable organizations to securely develop and run AI applications on NVIDIA-accelerated infrastructure anywhere.\nOur AI technology leadership is reinforced by our large and expanding ecosystem. Our computing platforms are available from virtually every major server maker and CSP, as well as on our own AI supercomputers. There are over 7.5 million developers worldwide using CUDA and our other software tools to help deploy our technology in our target markets. We are the leader in accelerating and releasing open AI models which enterprises, sovereigns, and startups can leverage to develop and run applications on our platform. We evangelize AI through partnerships with hundreds of universities and tens of thousands of startups through our Inception program. Additionally, our Deep Learning Institute provides instruction on the latest techniques on how to design, train, and deploy neural networks in applications using our accelerated computing platform.\nExtending our technology and platform leadership in computer graphics. We believe that computer graphics infused with AI is fundamental to the continued expansion and evolution of computing. We apply our research and development resources to enhance the user experience for consumer entertainment and professional visualization applications and create new virtual world and simulation capabilities. Our technologies are instrumental in driving the gaming, design, and creative industries forward, as developers leverage our libraries and algorithms to deliver an optimized experience on our GeForce and NVIDIA RTX platforms. Our computer graphics platforms leverage AI end-to-end, from the developer tools and cloud services to the Tensor Cores included in all RTX-class GPUs. Blackwell GPUs' advanced AI and neural rendering capabilities combined with NVIDIA's world-class AI software stacks significantly accelerate AI workloads running locally on PCs. Omniverse is real-time 3D design collaboration and virtual world simulation software that empowers artists, designers, and creators to connect and collaborate in leading design applications.\nAdvancing the leading autonomous vehicle platform. We believe the advent of autonomous vehicles, or AV, and electric vehicles, or EV, is revolutionizing the transportation industry. The algorithms required for autonomous driving - such as reasoning, perception, localization, and planning - are too complex for legacy hand-coded approaches and will use multiple neural networks instead. Therefore, we provide an AI-based hardware and software solution, designed and implemented from the ground up based on automotive safety standards, for the AV and EV market under the DRIVE brand, which we are bringing to market through our partnerships across the transportation industry including with automotive OEMs, mobility service providers, robotaxis, tier-1 suppliers, and start-ups. Our AV solution also includes the GPU-based hardware required to train the neural networks before their in-vehicle deployment, as well as to re-simulate their operation prior to any over-the-air software updates. We believe our comprehensive, top-to-bottom and end-to-end approach will enable the transportation industry to solve the complex problems arising from the shift to autonomous driving.\nLeveraging our intellectual property, or IP. We believe our IP is a valuable asset that can be accessed by our customers and partners through license and development agreements when they desire to build such capabilities directly into their own products or have us do so through a custom development. Such license and development arrangements can further enhance the reach of our technology.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Business Strategies", + "metadata": { + "length": 6205, + "summary": "NVIDIA's key strategies include advancing its accelerated computing platform to solve complex problems with lower power consumption, leveraging full-stack innovation across architecture, chip design, and software. It extends leadership in AI with end-to-end platforms for training and inferencing, including NVIDIA AI Enterprise, NIM, NeMo, and AI Blueprints, supported by a large ecosystem of developers and partners. The company also leads in computer graphics infused with AI, enhancing gaming and professional visualization via GeForce, RTX, and Omniverse. In autonomous vehicles, it provides an AI-based hardware and software solution under the DRIVE brand, partnering with automotive OEMs and startups. Finally, it leverages intellectual property through licensing and development agreements to broaden technology reach.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "NVIDIA", + "key", + "strategies", + "shape", + "business", + "approach", + "include", + "Advancing", + "accelerated", + "computing", + "platform", + "Our", + "solve", + "complex", + "problems", + "significantly", + "time", + "lower", + "power", + "consumption", + "alternative", + "computational", + "approaches", + "It", + "previously", + "deemed", + "unsolvable", + "We", + "work", + "deliver", + "continued", + "performance", + "leaps", + "outpace", + "Moore", + "Law", + "leveraging", + "innovation", + "architecture", + "chip", + "design", + "system", + "interconnect", + "algorithm", + "software", + "layers", + "This", + "full", + "stack", + "order", + "magnitude", + "advantages", + "relative", + "legacy", + "target", + "markets", + "Data", + "Center", + "Gaming", + "Professional", + "Visualization", + "Automotive", + "While", + "requirements", + "end", + "diverse", + "address", + "unified", + "underlying", + "GPUs", + "CPUs", + "CUDA", + "networking", + "technologies", + "fundamental", + "building", + "blocks", + "The", + "programmable", + "nature", + "make", + "leveraged", + "investments", + "research", + "development", + "support", + "multi", + "billion", + "dollar", + "shared", + "technology", + "variety", + "stacks", + "developed", + "internally", + "party", + "developers", + "partners", + "utilize", + "Extending", + "leadership", + "AI", + "provide", + "complete", + "addressing", + "training", + "inferencing", + "includes", + "data", + "center", + "scale", + "compute", + "solutions", + "processing", + "units", + "interconnects", + "systems", + "major", + "servers", + "DPUs", + "uniquely", + "suited", + "continue", + "add", + "specific", + "features", + "GPU", + "extend", + "position", + "In", + "addition", + "offer", + "Enterprise", + "comprehensive", + "suite", + "designed", + "simplify", + "deployment", + "production", + "grade", + "generative", + "applications", + "NIM", + "increases", + "token", + "throughput", + "industry", + "leading", + "open", + "proprietary", + "models", + "NeMo", + "solution", + "curating", + "fine", + "tuning", + "reinforcement", + "learning", + "evaluating", + "safeguarding", + "domain", + "adapted", + "Blueprints", + "pre", + "built", + "runnable", + "templates", + "enterprises", + "build", + "optimize", + "deploy", + "agents", + "preserving", + "privacy", + "These", + "tools", + "enable", + "organizations", + "securely", + "develop", + "run", + "infrastructure", + "reinforced", + "large", + "expanding", + "ecosystem", + "platforms", + "virtually", + "server", + "maker", + "CSP", + "supercomputers", + "There", + "7.5", + "million", + "worldwide", + "leader", + "accelerating", + "releasing", + "sovereigns", + "startups", + "leverage", + "evangelize", + "partnerships", + "hundreds", + "universities", + "tens", + "thousands", + "Inception", + "program", + "Additionally", + "Deep", + "Learning", + "Institute", + "instruction", + "latest", + "techniques", + "train", + "neural", + "networks", + "computer", + "graphics", + "infused", + "expansion", + "evolution", + "apply", + "resources", + "enhance", + "user", + "experience", + "consumer", + "entertainment", + "professional", + "visualization", + "create", + "virtual", + "world", + "simulation", + "capabilities", + "instrumental", + "driving", + "gaming", + "creative", + "industries", + "forward", + "libraries", + "algorithms", + "optimized", + "GeForce", + "RTX", + "developer", + "cloud", + "services", + "Tensor", + "Cores", + "included", + "class", + "Blackwell", + "advanced", + "rendering", + "combined", + "accelerate", + "workloads", + "running", + "locally", + "PCs", + "Omniverse", + "real", + "3D", + "collaboration", + "empowers", + "artists", + "designers", + "creators", + "connect", + "collaborate", + "autonomous", + "vehicle", + "advent", + "vehicles", + "AV", + "electric", + "EV", + "revolutionizing", + "transportation", + "required", + "reasoning", + "perception", + "localization", + "planning", + "hand", + "coded", + "multiple", + "Therefore", + "based", + "hardware", + "implemented", + "ground", + "automotive", + "safety", + "standards", + "market", + "DRIVE", + "brand", + "bringing", + "including", + "OEMs", + "mobility", + "service", + "providers", + "robotaxis", + "tier", + "suppliers", + "start", + "ups", + "simulate", + "operation", + "prior", + "air", + "updates", + "top", + "bottom", + "arising", + "shift", + "Leveraging", + "intellectual", + "property", + "IP", + "valuable", + "asset", + "accessed", + "customers", + "license", + "agreements", + "desire", + "directly", + "products", + "custom", + "Such", + "arrangements", + "reach" + ], + "keywords": [ + "Accelerated Computing", + "AI", + "Autonomous Vehicles" + ], + "connect_to": [] + } + }, + { + "chunk_id": "1ab210bb-e9ad-5f1a-b7bb-7214c4ed4bc0", + "type": "text", + "content": "Our worldwide sales and marketing strategy is key to achieving our objective of providing markets with our high-performance and efficient computing platforms and software. Our sales and marketing teams, located across our global markets, work closely with customers and various industry ecosystems through our partner network. Our partner network incorporates global, regional and specialized CSPs, OEMs, ODMs, ISVs, global system integrators, add-in board manufacturers, or AIBs, distributors, automotive manufacturers and tier-1 automotive suppliers, and other ecosystem participants.\nMembers of our sales team have technical expertise and product and industry knowledge. We also employ a team of application engineers and solution architects to provide pre-sales assistance to our partner network in designing, testing, and qualifying system designs that incorporate our platforms. For example, our solution architects work with CSPs to provide pre-sales assistance to enable our customers to optimize their hardware and software infrastructure for generative and agentic AI and LLM training and deployment. They also work with foundation model and enterprise software developers to enable our customers to optimize the training and fine-tuning of their models and services, and with enterprise end-users, often in collaboration with their global system integrator of choice, to fine-tune models and build AI applications. We believe that the depth and quality of our design support are key to improving our partner network's time-to-market, maintaining a high level of customer satisfaction, and fostering relationships that encourage our customers and partner network to use the next generation of our products within each platform.\nTo encourage the development of applications optimized for our platforms and software, we seek to establish and maintain strong relationships in the software development community. Engineering and marketing personnel engage with key software developers to promote and discuss our platforms, as well as to ascertain individual product requirements and solve technical problems. Our developer program supports the development of AI frameworks, SDKs, and APIs for software applications and game titles that are optimized for our platforms. Our Deep Learning Institute provides in-person and online training for developers in industries and organizations around the world to build AI and accelerated computing applications that leverage our platforms.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Sales and Marketing", + "metadata": { + "length": 2485, + "summary": "The passage describes the company's worldwide sales and marketing strategy, emphasizing collaboration with a partner network including CSPs, OEMs, ODMs, ISVs, and others. Sales teams provide technical expertise and pre-sales assistance through application engineers and solution architects to optimize hardware and software for AI and LLM training. The company also engages with the software development community to promote platform optimization, supported by a developer program and the Deep Learning Institute for training.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "worldwide", + "sales", + "marketing", + "strategy", + "key", + "achieving", + "objective", + "providing", + "markets", + "high", + "performance", + "efficient", + "computing", + "platforms", + "software", + "teams", + "located", + "global", + "work", + "closely", + "customers", + "industry", + "ecosystems", + "partner", + "network", + "incorporates", + "regional", + "specialized", + "CSPs", + "OEMs", + "ODMs", + "ISVs", + "system", + "integrators", + "add", + "board", + "manufacturers", + "AIBs", + "distributors", + "automotive", + "tier", + "suppliers", + "ecosystem", + "participants", + "Members", + "team", + "technical", + "expertise", + "product", + "knowledge", + "We", + "employ", + "application", + "engineers", + "solution", + "architects", + "provide", + "pre", + "assistance", + "designing", + "testing", + "qualifying", + "designs", + "incorporate", + "For", + "enable", + "optimize", + "hardware", + "infrastructure", + "generative", + "agentic", + "AI", + "LLM", + "training", + "deployment", + "They", + "foundation", + "model", + "enterprise", + "developers", + "fine", + "tuning", + "models", + "services", + "end", + "users", + "collaboration", + "integrator", + "choice", + "tune", + "build", + "applications", + "depth", + "quality", + "design", + "support", + "improving", + "time", + "market", + "maintaining", + "level", + "customer", + "satisfaction", + "fostering", + "relationships", + "encourage", + "generation", + "products", + "platform", + "To", + "development", + "optimized", + "seek", + "establish", + "maintain", + "strong", + "community", + "Engineering", + "personnel", + "engage", + "promote", + "discuss", + "ascertain", + "individual", + "requirements", + "solve", + "problems", + "developer", + "program", + "supports", + "frameworks", + "SDKs", + "APIs", + "game", + "titles", + "Deep", + "Learning", + "Institute", + "person", + "online", + "industries", + "organizations", + "world", + "accelerated", + "leverage" + ], + "keywords": [ + "Sales", + "Marketing", + "Partners" + ], + "connect_to": [] + } + }, + { + "chunk_id": "19c9901e-2159-53a7-8e5f-829c8019b4b2", + "type": "text", + "content": "Our computing platforms serve a diverse set of markets such as data centers, gaming, professional visualization, and automotive. Our desktop gaming products typically see stronger revenue in the second half of our fiscal year. Historical seasonality trends may not repeat.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Seasonality", + "metadata": { + "length": 272, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "computing", + "platforms", + "serve", + "diverse", + "set", + "markets", + "data", + "centers", + "gaming", + "professional", + "visualization", + "automotive", + "desktop", + "products", + "typically", + "stronger", + "revenue", + "half", + "fiscal", + "year", + "Historical", + "seasonality", + "trends", + "repeat" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3263b687-ff7b-5ff0-844c-87e2673052b0", + "type": "text", + "content": "We utilize a fabless and contracting manufacturing strategy, whereby we employ and partner with key suppliers for all phases of the manufacturing process, including wafer fabrication, assembly, testing, and packaging. We use the expertise of industry-leading suppliers that are certified by the International Organization for Standardization in such areas as fabrication, assembly, quality control and assurance, reliability, and testing. Additionally, we can avoid many of the significant costs and risks associated with owning and operating manufacturing operations. While we may directly procure certain raw materials used in the production of our products, such as memory, substrates, and a variety of components, our suppliers are responsible for procurement of most raw materials used in the production of our products. As a result, we can focus our resources on product design, quality assurance, marketing, and customer support. In periods of growth, we may place non-cancellable inventory orders for certain product components in advance of our historical lead times, pay premiums, or provide deposits to secure future supply and capacity and may need to continue to do so.\nWe have expanded our supplier relationships to build redundancy and resilience in our operations to provide long-term manufacturing capacity aligned with growing customer demand. While currently our supply chain is mainly concentrated in Asia, we are expanding into the U.S. and Latin America. We utilize foundries, such as Taiwan Semiconductor Manufacturing Company Limited, or TSMC, and Samsung Electronics Co., Ltd., or Samsung, to produce our semiconductor wafers. We purchase memory from SK Hynix Inc., Micron Technology, Inc., and Samsung. We utilize CoWoS technology for semiconductor packaging. We engage with independent subcontractors and contract manufacturers such as Hon Hai Precision Industry Co., Ltd., Wistron Corporation, and Fabrinet to perform assembly, testing and packaging of our final products.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Manufacturing", + "metadata": { + "length": 2000, + "summary": "The company uses a fabless and contracting manufacturing strategy, partnering with key suppliers for wafer fabrication, assembly, testing, and packaging. This approach avoids costs and risks of owning manufacturing operations, allowing focus on design, quality, marketing, and support. The supply chain is mainly in Asia but expanding to the U.S. and Latin America, with suppliers like TSMC, Samsung, SK Hynix, Micron, Hon Hai, Wistron, and Fabrinet.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "utilize", + "fabless", + "contracting", + "manufacturing", + "strategy", + "employ", + "partner", + "key", + "suppliers", + "phases", + "process", + "including", + "wafer", + "fabrication", + "assembly", + "testing", + "packaging", + "expertise", + "industry", + "leading", + "certified", + "International", + "Organization", + "Standardization", + "areas", + "quality", + "control", + "assurance", + "reliability", + "Additionally", + "avoid", + "significant", + "costs", + "risks", + "owning", + "operating", + "operations", + "While", + "directly", + "procure", + "raw", + "materials", + "production", + "products", + "memory", + "substrates", + "variety", + "components", + "responsible", + "procurement", + "As", + "result", + "focus", + "resources", + "product", + "design", + "marketing", + "customer", + "support", + "In", + "periods", + "growth", + "place", + "cancellable", + "inventory", + "orders", + "advance", + "historical", + "lead", + "times", + "pay", + "premiums", + "provide", + "deposits", + "secure", + "future", + "supply", + "capacity", + "continue", + "expanded", + "supplier", + "relationships", + "build", + "redundancy", + "resilience", + "long", + "term", + "aligned", + "growing", + "demand", + "chain", + "concentrated", + "Asia", + "expanding", + "Latin", + "America", + "foundries", + "Taiwan", + "Semiconductor", + "Manufacturing", + "Company", + "Limited", + "TSMC", + "Samsung", + "Electronics", + "Co", + "Ltd", + "produce", + "semiconductor", + "wafers", + "purchase", + "SK", + "Hynix", + "Inc", + "Micron", + "Technology", + "CoWoS", + "technology", + "engage", + "independent", + "subcontractors", + "contract", + "manufacturers", + "Hon", + "Hai", + "Precision", + "Industry", + "Wistron", + "Corporation", + "Fabrinet", + "perform", + "final" + ], + "keywords": [ + "Fabless", + "Supply Chain", + "Manufacturing" + ], + "connect_to": [] + } + }, + { + "chunk_id": "538fe95b-437a-50b6-af30-473f34bfa62e", + "type": "text", + "content": "The market for our products is intensely competitive and is characterized by rapid technological change and evolving industry standards. We believe that the principal competitive factors in this market are performance, breadth of product offerings, access to customers and partners and distribution channels, software support, conformity to industry standard APIs, manufacturing capabilities, processor pricing, and total system costs. We believe that our ability to remain competitive will depend on how well we are able to anticipate the features and functions that customers and partners will demand and whether we are able to deliver consistent volumes of our products at acceptable levels of quality and at competitive prices. We expect competition to increase from both existing competitors and new market entrants with products that may be lower priced than ours or may provide better performance or additional features not provided by our products. In addition, it is possible that new competitors or alliances among competitors could emerge and acquire significant market share.\nA significant source of competition comes from companies that provide or intend to provide GPUs, CPUs, DPUs, embedded SoCs, and other accelerated, AI computing processor products, and providers of semiconductor-based high-performance interconnect products based on InfiniBand, Ethernet, Fibre Channel, and proprietary technologies. Some of our competitors may have greater marketing, financial, distribution and manufacturing resources than we do and may be more able to adapt to customers or technological changes. We expect an increasingly competitive environment in the future.\nOur current competitors include:\n- suppliers and licensors of hardware and software for discrete and integrated GPUs, custom chips and other accelerated computing solutions, including solutions offered for AI, such as Advanced Micro Devices, Inc., or AMD, Huawei Technologies Co. Ltd., or Huawei, and Intel Corporation, or Intel;\n- large cloud services companies with internal teams designing hardware and software that incorporate accelerated or AI computing functionality as part of their internal solutions or platforms, such as Alibaba Group, Alphabet Inc., Amazon, Inc., or Amazon, Baidu, Inc., Huawei, and Microsoft Corporation, or Microsoft;\n- suppliers of Arm-based CPUs and companies that incorporate hardware and software for CPUs as part of their internal solutions or platforms, such as Amazon, Huawei, and Microsoft;\n- suppliers of hardware and software for SoC products that are used in servers or embedded into automobiles, autonomous machines, and gaming devices, such as Ambarella, Inc., AMD, Broadcom, Intel, Qualcomm Incorporated, Renesas Electronics Corporation, and Samsung, or companies with internal teams designing SoC products for their own products and services, such as Tesla, Inc.; and\n- networking products consisting of switches, network adapters (including DPUs), and cable solutions (including optical modules) include such as AMD, Arista Networks, Broadcom, Cisco Systems, Inc., Hewlett Packard Enterprise Company, Huawei, Intel, Lumentum Holdings Inc., and Marvell Technology, Inc, as well as internal teams of system vendors and large cloud services companies.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Competition", + "metadata": { + "length": 3264, + "summary": "The market is intensely competitive with rapid technological change. Key competitive factors include performance, product breadth, distribution, and pricing. Competitors include AMD, Intel, Huawei, and large cloud companies like Amazon and Microsoft, offering GPUs, CPUs, DPUs, and networking products. The company expects increased competition from new entrants and alliances.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "market", + "products", + "intensely", + "competitive", + "characterized", + "rapid", + "technological", + "change", + "evolving", + "industry", + "standards", + "We", + "principal", + "factors", + "performance", + "breadth", + "product", + "offerings", + "access", + "customers", + "partners", + "distribution", + "channels", + "software", + "support", + "conformity", + "standard", + "APIs", + "manufacturing", + "capabilities", + "processor", + "pricing", + "total", + "system", + "costs", + "ability", + "remain", + "depend", + "anticipate", + "features", + "functions", + "demand", + "deliver", + "consistent", + "volumes", + "acceptable", + "levels", + "quality", + "prices", + "expect", + "competition", + "increase", + "existing", + "competitors", + "entrants", + "lower", + "priced", + "provide", + "additional", + "provided", + "In", + "addition", + "alliances", + "emerge", + "acquire", + "significant", + "share", + "source", + "companies", + "intend", + "GPUs", + "CPUs", + "DPUs", + "embedded", + "SoCs", + "accelerated", + "AI", + "computing", + "providers", + "semiconductor", + "based", + "high", + "interconnect", + "InfiniBand", + "Ethernet", + "Fibre", + "Channel", + "proprietary", + "technologies", + "Some", + "greater", + "marketing", + "financial", + "resources", + "adapt", + "increasingly", + "environment", + "future", + "Our", + "current", + "include", + "suppliers", + "licensors", + "hardware", + "discrete", + "integrated", + "custom", + "chips", + "solutions", + "including", + "offered", + "Advanced", + "Micro", + "Devices", + "Inc", + "AMD", + "Huawei", + "Technologies", + "Co", + "Ltd", + "Intel", + "Corporation", + "large", + "cloud", + "services", + "internal", + "teams", + "designing", + "incorporate", + "functionality", + "part", + "platforms", + "Alibaba", + "Group", + "Alphabet", + "Amazon", + "Baidu", + "Microsoft", + "Arm", + "SoC", + "servers", + "automobiles", + "autonomous", + "machines", + "gaming", + "devices", + "Ambarella", + "Broadcom", + "Qualcomm", + "Incorporated", + "Renesas", + "Electronics", + "Samsung", + "Tesla", + "networking", + "consisting", + "switches", + "network", + "adapters", + "cable", + "optical", + "modules", + "Arista", + "Networks", + "Cisco", + "Systems", + "Hewlett", + "Packard", + "Enterprise", + "Company", + "Lumentum", + "Holdings", + "Marvell", + "Technology", + "vendors" + ], + "keywords": [ + "Competition", + "GPU", + "AI" + ], + "connect_to": [] + } + }, + { + "chunk_id": "404fac8b-d123-5ce0-b2f4-153c1d0d4a5b", + "type": "text", + "content": "We rely primarily on a combination of patents, trademarks, trade secrets, employee and third-party nondisclosure agreements, and licensing arrangements to protect our IP in the United States and internationally. Our currently issued patents have expiration dates from March 2026 to June 2045. We have numerous patents issued, allowed, and pending in the United States and in foreign jurisdictions. Our patents and pending patent applications primarily relate to our products and the technology used in connection with our products. We also rely on international treaties, organizations, and foreign laws to protect our IP. The laws of certain foreign countries in which our products are or may be manufactured or sold, including various countries in Asia, may not protect our products or IP rights to the same extent as the laws of the United States. This decreased protection makes the possibility of piracy of our technology and products more likely. We continuously assess whether and where to seek formal protection for innovations and technologies based on such factors as:\n• the location in which our products are manufactured;\n• our strategic technology or product directions in different countries;\n• the degree to which IP laws exist and are meaningfully enforced in different jurisdictions; and\n- the commercial significance of our operations and our competitors' operations in particular countries and regions.\nWe have licensed technology from third parties and expect to continue entering such license agreements.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Patents and Proprietary Rights", + "metadata": { + "length": 1525, + "summary": "The company protects its intellectual property through patents, trademarks, trade secrets, nondisclosure agreements, and licensing. Patents expire between 2026 and 2045. Protection varies internationally, especially in Asia, increasing piracy risk. They assess protection based on manufacturing locations, strategic directions, IP law enforcement, and commercial significance. They also license technology from third parties.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "rely", + "primarily", + "combination", + "patents", + "trademarks", + "trade", + "secrets", + "employee", + "party", + "nondisclosure", + "agreements", + "licensing", + "arrangements", + "protect", + "IP", + "United", + "States", + "internationally", + "Our", + "issued", + "expiration", + "dates", + "March", + "2026", + "June", + "2045", + "numerous", + "allowed", + "pending", + "foreign", + "jurisdictions", + "patent", + "applications", + "relate", + "products", + "technology", + "connection", + "international", + "treaties", + "organizations", + "laws", + "The", + "countries", + "manufactured", + "sold", + "including", + "Asia", + "rights", + "extent", + "This", + "decreased", + "protection", + "makes", + "possibility", + "piracy", + "continuously", + "assess", + "seek", + "formal", + "innovations", + "technologies", + "based", + "factors", + "location", + "strategic", + "product", + "directions", + "degree", + "exist", + "meaningfully", + "enforced", + "commercial", + "significance", + "operations", + "competitors", + "regions", + "licensed", + "parties", + "expect", + "continue", + "entering", + "license" + ], + "keywords": [ + "patents", + "trademarks", + "trade secrets" + ], + "connect_to": [] + } + }, + { + "chunk_id": "563a5b13-07c3-5f5e-b60e-3680434c2700", + "type": "text", + "content": "Our worldwide business activities are subject to various laws, rules, and regulations of the United States as well as of foreign governments.\nOver the past three years, we have been subject to a series of shifting and expanding export control restrictions, impacting our ability to serve customers outside the United States.\nIn August 2022, the U.S. government, or USG, announced export restrictions and export licensing requirements targeting China's semiconductor and supercomputing industries. These restrictions impacted exports of certain chips, as well as software, hardware, equipment and technology used to develop, produce and manufacture certain chips to China (including Hong Kong and Macau) and Russia, and specifically impact our A100 and H100 integrated circuits, DGX or any other systems or boards which incorporate A100 or H100 integrated circuits. In July 2023, the USG also informed us of an additional licensing requirement for a subset of A100 and H100 products destined to certain customers and other regions, including some countries in the Middle East.\nIn October 2023, the USG announced new and updated licensing requirements for exports to China and Country Groups D:1, D:4, and D:5 (including but not limited to Saudi Arabia, the United Arab Emirates, and Vietnam, but excluding Israel) of our products exceeding certain performance thresholds, including, but not limited to, the A100, A800, H100, H800, L4, L40, L40S RTX 4090, GB200 NVL72, and B200. The licensing requirements also apply to the export of products exceeding certain performance thresholds to a party headquartered in, or with an ultimate parent headquartered in, Country Group D5, including China.\nIn April 2025, the USG informed us that it requires a license for export to China (including Hong Kong and Macau) and D:5 countries, or to companies headquartered or with an ultimate parent therein, of our H2O integrated circuits and any other circuits achieving the H2O's memory bandwidth, interconnect bandwidth, or combination thereof. As a result of these requirements, we incurred a \\$4.5 billion charge in the first quarter of fiscal year 2026 associated with H2O for excess inventory and purchase obligations, as the demand for H2O products diminished.\nIn August 2025, the USG granted licenses that would allow us to ship certain H2O products to certain China-based customers. We generated approximately \\$60 million in H2O revenue under those licenses. USG officials expressed an expectation that the USG will receive 15% or more of the revenue generated from licensed sales of our products, but the USG did not publish a regulation codifying such requirement.\nIn February 2026, the USG granted a license that would allow us to ship small amounts of H200 products to specific China-based customers. To date, we have not generated any revenue under the H200 licensing program, and do not yet know whether any imports will be allowed into China. The license requires that the H200s go through an inspection process in the United States prior to any shipment to the customer. As a result, any H200 shipped under the new licensing program will be subject to a 25% tariff upon importation into the United States.\nIn the event that we are able to sell licensed products into the China market, we may not be able to pass along all or any of the tariff to our customers, and may be subject to litigation, increased costs, and a harmed competitive position.\nThe export controls applicable to China are complex and address a variety of parameters, including the total processing performance of a chip, the “performance density” of a chip, the interconnect bandwidth of a chip, and the memory bandwidth of a chip. Under the current rules and geopolitical landscape, we are unable to create and deliver a competitive product for China’s data center market that receives approval from both the USG and the Chinese government. As of the end of fiscal year 2026, we were effectively foreclosed from competing in China's data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide. Unless we are able to return with a product that meets the approval of both the USG and the Chinese government, our lost opportunity and the benefit to our competitors will have a material and adverse impact on our business, operating results, and financial condition.\nIn addition to controls targeting D:1, D:4 and D:5 countries, the USG has also imposed worldwide export controls impacting our products, and may impose additional controls in the future.\nIn January 2025, the USG published the AI Diffusion IFR in the Federal Register. The IFR would have imposed a worldwide licensing requirement on our data center products, such as our H200, GB200 and GB300. The AI Diffusion IFR would have divided the world into three tiers, relegating most countries to “Tier 2” status, and would have created a complex and burdensome scheme for licensing approvals.\nIn May 2025, the USG announced that it would rescind the AI Diffusion IFR and implement a replacement rule. The scope, timing, and requirements of the forthcoming rule remain uncertain. The replacement rule may impose new restrictions on our products or operations and/or add license requirements that could have a material impact on our business, operating results, and financial condition. For example, in October 2025, the Senate passed the “GAIN AI Act” in the NDAA. The GAIN AI Act would restrict the Trump Administration’s ability to adapt the Biden Administration’s export control rules, and could also allow private U.S. persons to review and overturn licensing and foreign policy decisions made by the Trump Administration.\nOur competitive position has been harmed by export controls, and our competitive position and future results will be further harmed, over the long term, if the restrictions remain in place or are expanded in geographic, customer, or product scope, if customers purchase product from competitors, if customers develop their own internal solution, if we are unable to provide contractual warranty or other extended service obligations, if the USG does not grant licenses in a timely manner or denies licenses to significant customers or if we incur significant transition costs.\nThe licensing process may not be resolved before significant business opportunities evaporate. Even if the USG grants any requested licenses, the licenses have already and may in the future be temporary, impose burdensome conditions regarding the installation, maintenance, and use of such products, or include financial or economic requirements that we or our customers or end users cannot or choose not to fulfill. The licensing requirements have already and may in the future benefit certain of our competitors, as the licensing process will make our pre-sale and post-sale technical support efforts more cumbersome and less certain and encourage customers in China, the Middle East, and other regions to pursue alternatives to our products, including semiconductor suppliers based in China, Europe, and Israel.\nAdditionally, restrictions imposed by the Chinese government on the duration of gaming activities and access to games may adversely affect our Gaming revenue, and even if we are able to participate in the China data center compute market, increased oversight of digital platform companies may adversely affect our Data Center revenue. The Chinese government has encouraged customers to purchase from our China-based competitors and discouraged customers from purchasing, importing, or using our data center products, including any China-specific product designed to comply with U.S. export controls.\nWhile we work to enhance the resiliency and redundancy of our supply chain, which is currently concentrated in Asia, new and existing export controls or changes to existing export controls could limit alternative manufacturing locations and negatively impact our business. Refer to “Item 1A. Risk Factors – Risks Related to Regulatory, Legal, Our Stock, and Other Matters” for a discussion of this potential impact.\nCompliance with laws, rules, and regulations has not otherwise had a material effect upon our capital expenditures, results of operations, or competitive position and we do not currently anticipate material capital expenditures for environmental control facilities. Compliance with existing or future governmental regulations, including, but not limited to, those pertaining to IP ownership and infringement, taxes, import and export requirements and tariffs, anti-corruption, business acquisitions, foreign exchange controls and cash repatriation restrictions, data privacy requirements, competition and antitrust, advertising, employment, product regulations, cybersecurity, environmental, health and safety requirements, the responsible use of AI, climate change, cryptocurrency, and consumer laws, could further increase our costs, impact our competitive position, and otherwise may have a material adverse impact on our business, financial condition and results of operations in subsequent periods. Refer to “Item 1A. Risk Factors” for a discussion of these potential impacts.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Government Regulations", + "metadata": { + "length": 9269, + "summary": "The passage details the impact of shifting US export controls on a company's business, particularly targeting China's semiconductor and supercomputing industries. Key events include restrictions on A100 and H100 products in 2022, expanded licensing requirements in 2023 and 2025, and a $4.5 billion charge for H2O inventory. The company faces foreclosure from China's data center market, harming its competitive position. Additional worldwide controls, like the AI Diffusion IFR, add uncertainty. Compliance with various regulations may increase costs and affect financial results.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "worldwide", + "business", + "activities", + "subject", + "laws", + "rules", + "regulations", + "United", + "States", + "foreign", + "governments", + "Over", + "past", + "years", + "series", + "shifting", + "expanding", + "export", + "control", + "restrictions", + "impacting", + "ability", + "serve", + "customers", + "In", + "August", + "2022", + "government", + "USG", + "announced", + "licensing", + "requirements", + "targeting", + "China", + "semiconductor", + "supercomputing", + "industries", + "These", + "impacted", + "exports", + "chips", + "software", + "hardware", + "equipment", + "technology", + "develop", + "produce", + "manufacture", + "including", + "Hong", + "Kong", + "Macau", + "Russia", + "specifically", + "impact", + "A100", + "H100", + "integrated", + "circuits", + "DGX", + "systems", + "boards", + "incorporate", + "July", + "2023", + "informed", + "additional", + "requirement", + "subset", + "products", + "destined", + "regions", + "countries", + "Middle", + "East", + "October", + "updated", + "Country", + "Groups", + "limited", + "Saudi", + "Arabia", + "Arab", + "Emirates", + "Vietnam", + "excluding", + "Israel", + "exceeding", + "performance", + "thresholds", + "A800", + "H800", + "L4", + "L40", + "L40S", + "RTX", + "4090", + "GB200", + "NVL72", + "B200", + "The", + "apply", + "party", + "headquartered", + "ultimate", + "parent", + "Group", + "D5", + "April", + "2025", + "requires", + "license", + "companies", + "H2O", + "achieving", + "memory", + "bandwidth", + "interconnect", + "combination", + "thereof", + "As", + "result", + "incurred", + "4.5", + "billion", + "charge", + "quarter", + "fiscal", + "year", + "2026", + "excess", + "inventory", + "purchase", + "obligations", + "demand", + "diminished", + "granted", + "licenses", + "ship", + "based", + "We", + "generated", + "approximately", + "60", + "million", + "revenue", + "officials", + "expressed", + "expectation", + "receive", + "15%", + "licensed", + "sales", + "publish", + "regulation", + "codifying", + "February", + "small", + "amounts", + "H200", + "specific", + "To", + "date", + "program", + "imports", + "allowed", + "H200s", + "inspection", + "process", + "prior", + "shipment", + "customer", + "shipped", + "25%", + "tariff", + "importation", + "event", + "sell", + "market", + "pass", + "litigation", + "increased", + "costs", + "harmed", + "competitive", + "position", + "controls", + "applicable", + "complex", + "address", + "variety", + "parameters", + "total", + "processing", + "chip", + "density", + "Under", + "current", + "geopolitical", + "landscape", + "unable", + "create", + "deliver", + "product", + "data", + "center", + "receives", + "approval", + "Chinese", + "end", + "effectively", + "foreclosed", + "competing", + "computing", + "compute", + "effective", + "foreclosure", + "helped", + "competitors", + "build", + "larger", + "developer", + "ecosystems", + "challenge", + "Unless", + "return", + "meets", + "lost", + "opportunity", + "benefit", + "material", + "adverse", + "operating", + "results", + "financial", + "condition", + "addition", + "imposed", + "impose", + "future", + "January", + "published", + "AI", + "Diffusion", + "IFR", + "Federal", + "Register", + "GB300", + "divided", + "world", + "tiers", + "relegating", + "Tier", + "status", + "created", + "burdensome", + "scheme", + "approvals", + "May", + "rescind", + "implement", + "replacement", + "rule", + "scope", + "timing", + "forthcoming", + "remain", + "uncertain", + "operations", + "add", + "For", + "Senate", + "passed", + "GAIN", + "Act", + "NDAA", + "restrict", + "Trump", + "Administration", + "adapt", + "Biden", + "private", + "persons", + "review", + "overturn", + "policy", + "decisions", + "made", + "long", + "term", + "place", + "expanded", + "geographic", + "internal", + "solution", + "provide", + "contractual", + "warranty", + "extended", + "service", + "grant", + "timely", + "manner", + "denies", + "significant", + "incur", + "transition", + "resolved", + "opportunities", + "evaporate", + "Even", + "grants", + "requested", + "temporary", + "conditions", + "installation", + "maintenance", + "include", + "economic", + "users", + "choose", + "fulfill", + "make", + "pre", + "sale", + "post", + "technical", + "support", + "efforts", + "cumbersome", + "encourage", + "pursue", + "alternatives", + "suppliers", + "Europe", + "Additionally", + "duration", + "gaming", + "access", + "games", + "adversely", + "affect", + "Gaming", + "participate", + "oversight", + "digital", + "platform", + "Data", + "Center", + "encouraged", + "discouraged", + "purchasing", + "importing", + "designed", + "comply", + "While", + "work", + "enhance", + "resiliency", + "redundancy", + "supply", + "chain", + "concentrated", + "Asia", + "existing", + "limit", + "alternative", + "manufacturing", + "locations", + "negatively", + "Refer", + "Item", + "1A", + "Risk", + "Factors", + "Risks", + "Related", + "Regulatory", + "Legal", + "Stock", + "Other", + "Matters", + "discussion", + "potential", + "Compliance", + "effect", + "capital", + "expenditures", + "anticipate", + "environmental", + "facilities", + "governmental", + "pertaining", + "IP", + "ownership", + "infringement", + "taxes", + "import", + "tariffs", + "anti", + "corruption", + "acquisitions", + "exchange", + "cash", + "repatriation", + "privacy", + "competition", + "antitrust", + "advertising", + "employment", + "cybersecurity", + "health", + "safety", + "responsible", + "climate", + "change", + "cryptocurrency", + "consumer", + "increase", + "subsequent", + "periods", + "impacts" + ], + "keywords": [ + "export controls", + "USG regulations", + "China market" + ], + "connect_to": [] + } + }, + { + "chunk_id": "410e5e54-bc48-5ba0-85cf-886552723035", + "type": "text", + "content": "As of the end of fiscal year 2026, we had approximately 42,000 employees in 38 countries; 31,000 were engaged in research and development and 11,000 were engaged in sales, marketing, operations, and administrative positions.\nTo execute our business strategy successfully, we focus on recruiting, developing, and retaining top global talent.\nWithin our workforce, more than 80 percent have technical roles and more than half of the workforce hold an advanced degree. Our employees also help to surface top talent, with over 40 percent of our new hires in fiscal year 2026 coming from employee referrals. In fiscal year 2026, our turnover rate was 3.7 percent.\nWe invest in employee development through on-the-job trainings and tuition reimbursement programs.\nOur compensation and benefits are designed to reward performance and align employee interests with those of our shareholders through equity participation and comprehensive health and financial wellness programs. We also utilize employee listening systems to gather feedback and maintain an inclusive culture where hiring and promotions are based on merit.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Human Capital Management", + "metadata": { + "length": 1113, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "As", + "end", + "fiscal", + "year", + "2026", + "approximately", + "42", + "000", + "employees", + "38", + "countries", + "31", + "engaged", + "research", + "development", + "11", + "sales", + "marketing", + "operations", + "administrative", + "positions", + "To", + "execute", + "business", + "strategy", + "successfully", + "focus", + "recruiting", + "developing", + "retaining", + "top", + "global", + "talent", + "Within", + "workforce", + "80", + "percent", + "technical", + "roles", + "half", + "hold", + "advanced", + "degree", + "Our", + "surface", + "40", + "hires", + "coming", + "employee", + "referrals", + "In", + "turnover", + "rate", + "3.7", + "We", + "invest", + "job", + "trainings", + "tuition", + "reimbursement", + "programs", + "compensation", + "benefits", + "designed", + "reward", + "performance", + "align", + "interests", + "shareholders", + "equity", + "participation", + "comprehensive", + "health", + "financial", + "wellness", + "utilize", + "listening", + "systems", + "gather", + "feedback", + "maintain", + "inclusive", + "culture", + "hiring", + "promotions", + "based", + "merit" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "62297b5b-567d-5aec-9805-9d09440ffcd3", + "type": "table", + "content": "
NameAgePosition
Jen-Hsun Huang63President and Chief Executive Officer
Colette M. Kress58Executive Vice President and Chief Financial Officer
Ajay K. Puri71Executive Vice President, Worldwide Field Operations
Debora Shoquist71Executive Vice President, Operations
Timothy S. Teter59Executive Vice President and General Counsel
", + "path": "tables/table-60 Nvidia Executives.html", + "metadata": { + "length": 550, + "summary": "table-61\nThe table lists Nvidia's top executives: Jen-Hsun Huang (CEO, 63), Colette Kress (CFO, 58), Ajay Puri (Field Ops, 71), Debora Shoquist (Ops, 71), and Timothy Teter (General Counsel, 59).", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-60 Nvidia Executives.html", + "keywords": [ + "Nvidia", + "Executives", + "Ages" + ], + "tokens": [] + } + }, + { + "chunk_id": "fc569cf0-cfc4-5938-83a8-e48b066a29f5", + "type": "text", + "content": "The following sets forth certain information regarding our executive officers, their ages, and positions as of February 20, 2026:\n\n[tables/table-60 Nvidia Executives.html]\n\nJen-Hsun Huang co-founded NVIDIA in 1993 and has served as our President, Chief Executive Officer, and a member of the Board of Directors since our inception. From 1985 to 1993, Mr. Huang was employed at LSI Logic Corporation, a computer chip manufacturer, where he held a variety of positions including as Director of Coreware, the business unit responsible for LSI's SOC. From 1983 to 1985, Mr. Huang was a microprocessor designer for AMD, a semiconductor company. Mr. Huang holds a B.S.E.E. degree from Oregon State University and an M.S.E.E. degree from Stanford University.\nColette M. Kress joined NVIDIA in 2013 as Executive Vice President and Chief Financial Officer. Prior to NVIDIA, Ms. Kress most recently served as Senior Vice President and Chief Financial Officer of the Business Technology and Operations Finance organization at Cisco Systems, Inc., a networking equipment company, since 2010. At Cisco, Ms. Kress was responsible for financial strategy, planning, reporting and business development for all business segments, engineering and operations. From 1997 to 2010 Ms. Kress held a variety of positions at Microsoft, a software company, including, beginning in 2006, Chief Financial Officer of the Server and Tools division, where Ms. Kress was responsible for financial strategy, planning, reporting and business development for the division. Prior to joining Microsoft, Ms. Kress spent eight years at Texas Instruments Incorporated, a semiconductor company, where she held a variety of finance positions. Ms. Kress holds a B.S. degree in Finance from University of Arizona and an M.B.A. degree from Southern Methodist University.\nAjay K. Puri joined NVIDIA in 2005 as Senior Vice President, Worldwide Sales and became Executive Vice President, Worldwide Field Operations in 2009. Prior to NVIDIA, he held positions in sales, marketing, and general management over a 22-year career at Sun Microsystems, Inc., a computing systems company. Mr. Puri previously held marketing, management consulting, and product development positions at Hewlett-Packard, an information technology company, Booz Allen Hamilton Inc., a management and technology consulting company, and Texas Instruments Incorporated. Mr. Puri holds a B.S.E.E. degree from the University of Minnesota, an M.S.E.E. degree from the California Institute of Technology and an M.B.A. degree from Harvard Business School.\nDebora Shoquist joined NVIDIA in 2007 as Senior Vice President of Operations and in 2009 became Executive Vice President of Operations. Prior to NVIDIA, Ms. Shoquist served from 2004 to 2007 as Executive Vice President of Operations at JDS Uniphase Corp., a provider of communications test and measurement solutions and optical products for the telecommunications industry. She served from 2002 to 2004 as Senior Vice President and General Manager of the Electro-Optics business at Coherent, Inc., a manufacturer of commercial and scientific laser equipment. Previously, she worked at Quantum Corp., a data protection company, as President of the Personal Computer Hard Disk Drive Division, and at Hewlett-Packard. Ms. Shoquist holds a B.S. degree in Electrical Engineering from Kansas State University and a B.S. degree in Biology from Santa Clara University.\nTimothy S. Teter joined NVIDIA in 2017 as Senior Vice President, General Counsel and Secretary and became Executive Vice President, General Counsel and Secretary in February 2018. Prior to NVIDIA, Mr. Teter spent more than two decades at the law firm of Cooley LLP, where he focused on litigating patent and technology related matters. Prior to attending law school, he worked as an engineer at Lockheed Missiles and Space Company, an aerospace company. Mr. Teter holds a B.S. degree in Mechanical Engineering from the University of California at Davis and a J.D. degree from Stanford Law School.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Information About Our Executive Officers", + "metadata": { + "length": 4028, + "summary": "The passage provides biographical information about NVIDIA's executive officers as of February 20, 2026. It details the roles, career histories, and educational backgrounds of five key executives: Jen-Hsun Huang (co-founder, President, CEO), Colette M. Kress (EVP and CFO), Ajay K. Puri (EVP, Worldwide Field Operations), Debora Shoquist (EVP of Operations), and Timothy S. Teter (EVP, General Counsel and Secretary). Each entry includes prior positions at companies like AMD, LSI Logic, Cisco, Microsoft, Sun Microsystems, and others, along with their academic degrees.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "sets", + "information", + "executive", + "officers", + "ages", + "positions", + "February", + "20", + "2026", + "Jen", + "Hsun", + "Huang", + "founded", + "NVIDIA", + "1993", + "served", + "President", + "Chief", + "Executive", + "Officer", + "member", + "Board", + "Directors", + "inception", + "From", + "1985", + "Mr", + "employed", + "LSI", + "Logic", + "Corporation", + "computer", + "chip", + "manufacturer", + "held", + "variety", + "including", + "Director", + "Coreware", + "business", + "unit", + "responsible", + "SOC", + "1983", + "microprocessor", + "designer", + "AMD", + "semiconductor", + "company", + "holds", + "degree", + "Oregon", + "State", + "University", + "Stanford", + "Colette", + "Kress", + "joined", + "2013", + "Vice", + "Financial", + "Prior", + "Ms", + "recently", + "Senior", + "Business", + "Technology", + "Operations", + "Finance", + "organization", + "Cisco", + "Systems", + "Inc", + "networking", + "equipment", + "2010", + "At", + "financial", + "strategy", + "planning", + "reporting", + "development", + "segments", + "engineering", + "operations", + "1997", + "Microsoft", + "software", + "beginning", + "2006", + "Server", + "Tools", + "division", + "joining", + "spent", + "years", + "Texas", + "Instruments", + "Incorporated", + "finance", + "Arizona", + "Southern", + "Methodist", + "Ajay", + "Puri", + "2005", + "Worldwide", + "Sales", + "Field", + "2009", + "sales", + "marketing", + "general", + "management", + "22", + "year", + "career", + "Sun", + "Microsystems", + "computing", + "systems", + "previously", + "consulting", + "product", + "Hewlett", + "Packard", + "technology", + "Booz", + "Allen", + "Hamilton", + "Minnesota", + "California", + "Institute", + "Harvard", + "School", + "Debora", + "Shoquist", + "2007", + "2004", + "JDS", + "Uniphase", + "Corp", + "provider", + "communications", + "test", + "measurement", + "solutions", + "optical", + "products", + "telecommunications", + "industry", + "She", + "2002", + "General", + "Manager", + "Electro", + "Optics", + "Coherent", + "commercial", + "scientific", + "laser", + "Previously", + "worked", + "Quantum", + "data", + "protection", + "Personal", + "Computer", + "Hard", + "Disk", + "Drive", + "Division", + "Electrical", + "Engineering", + "Kansas", + "Biology", + "Santa", + "Clara", + "Timothy", + "Teter", + "2017", + "Counsel", + "Secretary", + "2018", + "decades", + "law", + "firm", + "Cooley", + "LLP", + "focused", + "litigating", + "patent", + "related", + "matters", + "attending", + "school", + "engineer", + "Lockheed", + "Missiles", + "Space", + "Company", + "aerospace", + "Mechanical", + "Davis", + "Law" + ], + "keywords": [ + "executive officers", + "NVIDIA", + "biographies" + ], + "connect_to": [ + { + "target": "62297b5b-567d-5aec-9805-9d09440ffcd3", + "relation": "embeds", + "ref": "[tables/table-60 Nvidia Executives.html]", + "position": { + "start": 131, + "end": 171 + } + } + ] + } + }, + { + "chunk_id": "886de81a-e10b-59d5-a75d-c31c51a0b864", + "type": "text", + "content": "Our annual reports on Form 10-K, quarterly reports on Form 10-Q, current reports on Form 8-K and, if applicable, amendments to those reports filed or furnished pursuant to Section 13(a) or 15(d) of the Securities Exchange Act of 1934, as amended, or the Exchange Act, are available free of charge on or through our website, http://www.nvidia.com, as soon as reasonably practicable after we electronically file such material with, or furnish it to, the Securities and Exchange Commission, or the SEC. The SEC's website, http://www.sec.gov, contains reports, proxy and information statements, and other information regarding issuers that file electronically with the SEC. Our web site and the information on it or connected to it are not a part of this Annual Report on Form 10-K.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Available Information", + "metadata": { + "length": 778, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "annual", + "reports", + "Form", + "10", + "quarterly", + "current", + "applicable", + "amendments", + "filed", + "furnished", + "pursuant", + "Section", + "13", + "15", + "Securities", + "Exchange", + "Act", + "1934", + "amended", + "free", + "charge", + "website", + "http", + "www", + "nvidia", + "practicable", + "electronically", + "file", + "material", + "furnish", + "Commission", + "SEC", + "The", + "sec", + "gov", + "proxy", + "information", + "statements", + "issuers", + "web", + "site", + "connected", + "part", + "Annual", + "Report" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cbdcce98-0094-58c3-b45d-fde8f303125e", + "type": "text", + "content": "The following risk factors should be considered in addition to the other information in this Annual Report on Form 10-K. The following risks could harm our business, financial condition, results of operations or reputation, which could cause our stock price to decline. Additional risks, trends and uncertainties not presently known to us or that we currently believe are immaterial may also harm our business, financial condition, results of operations or reputation.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors", + "metadata": { + "length": 468, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "risk", + "factors", + "considered", + "addition", + "information", + "Annual", + "Report", + "Form", + "10", + "risks", + "harm", + "business", + "financial", + "condition", + "results", + "operations", + "reputation", + "stock", + "price", + "decline", + "Additional", + "trends", + "uncertainties", + "presently", + "immaterial" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5c62ad29-e62f-52ba-85c2-dddd20edee37", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors Summary", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "39080867-9e65-5b4f-91f6-ed2685088241", + "type": "text", + "content": "• Failure to meet the evolving needs of our industry and markets may adversely impact our financial results.\n- Competition could adversely impact our market share and financial results.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors Summary/Risks Related to Our Industry and Markets", + "metadata": { + "length": 185, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Failure", + "meet", + "evolving", + "industry", + "markets", + "adversely", + "impact", + "financial", + "results", + "Competition", + "market", + "share" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "db8e65d1-e3eb-5726-b779-378767928306", + "type": "text", + "content": "\\- Long manufacturing lead times and uncertain supply and capacity availability, combined with a failure to estimate customer demand accurately has led and could lead to mismatches between supply and demand.\n- Dependency on third-party suppliers and their technology to manufacture, assemble, test, or package our products reduces our control over product quantity and quality, manufacturing yields, and product delivery schedules and could harm our business.\n- Defects in our products have caused and could cause us to incur significant expenses to remediate and could damage our business.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors Summary/Risks Related to Demand, Supply, and Manufacturing", + "metadata": { + "length": 590, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Long", + "manufacturing", + "lead", + "times", + "uncertain", + "supply", + "capacity", + "availability", + "combined", + "failure", + "estimate", + "customer", + "demand", + "accurately", + "led", + "mismatches", + "Dependency", + "party", + "suppliers", + "technology", + "manufacture", + "assemble", + "test", + "package", + "products", + "reduces", + "control", + "product", + "quantity", + "quality", + "yields", + "delivery", + "schedules", + "harm", + "business", + "Defects", + "caused", + "incur", + "significant", + "expenses", + "remediate", + "damage" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c9fe2a41-8893-5e44-b075-6c78894e3ee9", + "type": "text", + "content": "- Adverse economic conditions may harm our business.\n- International sales and operations are a significant part of our business, which exposes us to risks that could harm our business.\n- Product, system security and data protection incidents or breaches, as well as cyber-attacks could disrupt our operations and adversely affect our financial condition, stock price and reputation.\n• Business disruptions could harm our operations and financial results.\n• Climate change may have a long-term impact on our business.\n• We may not be able to realize the potential benefits of business investments or acquisitions, nor successfully integrate acquisition targets.\n- A significant amount of our revenue stems from a limited number of partners and distributors and we have a concentration of sales to customers who purchase directly or indirectly from us, and our revenue could be adversely affected if we lose or are prevented from selling to any of these customers.\n- Commercial arrangements expose us to counterparty risks, which may negatively impact our business, financial condition, or results of operations.\n• We may be unable to attract, retain, and motivate our executives and key employees.\n- Modification or interruption of our business processes and information systems may disrupt our business and internal controls.\n- Our operating results have in the past fluctuated and may in the future fluctuate, and if our operating results are below the expectations of securities analysts or investors, our stock price could decline.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors Summary/Risks Related to Our Global Operating Business", + "metadata": { + "length": 1535, + "summary": "The passage outlines various business risks including adverse economic conditions, international operations exposure, cybersecurity threats, business disruptions, climate change impacts, acquisition integration challenges, revenue concentration with limited partners, counterparty risks, talent retention issues, information system disruptions, and fluctuating operating results that could affect stock price.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Adverse", + "economic", + "conditions", + "harm", + "business", + "International", + "sales", + "operations", + "significant", + "part", + "exposes", + "risks", + "Product", + "system", + "security", + "data", + "protection", + "incidents", + "breaches", + "cyber", + "attacks", + "disrupt", + "adversely", + "affect", + "financial", + "condition", + "stock", + "price", + "reputation", + "Business", + "disruptions", + "results", + "Climate", + "change", + "long", + "term", + "impact", + "We", + "realize", + "potential", + "benefits", + "investments", + "acquisitions", + "successfully", + "integrate", + "acquisition", + "targets", + "amount", + "revenue", + "stems", + "limited", + "number", + "partners", + "distributors", + "concentration", + "customers", + "purchase", + "directly", + "indirectly", + "affected", + "lose", + "prevented", + "selling", + "Commercial", + "arrangements", + "expose", + "counterparty", + "negatively", + "unable", + "attract", + "retain", + "motivate", + "executives", + "key", + "employees", + "Modification", + "interruption", + "processes", + "information", + "systems", + "internal", + "controls", + "Our", + "operating", + "past", + "fluctuated", + "future", + "fluctuate", + "expectations", + "securities", + "analysts", + "investors", + "decline" + ], + "keywords": [ + "Business Risks", + "Operations", + "Financial Impact" + ], + "connect_to": [] + } + }, + { + "chunk_id": "bf9fe9c3-da4c-5087-8378-11e62f97145c", + "type": "text", + "content": "• We are subject to complex laws, rules, regulations, and political and other actions, including restrictions on the export of our products, which may adversely impact our business.\n- Scrutiny regarding our corporate sustainability practices could result in financial, reputational, or operational harm and liability.\n- Issues relating to the responsible use of our technologies, including AI, may result in reputational or financial harm and liability.\n- Adequately protecting our IP rights could be costly, and our ability to compete could be harmed if we are unsuccessful or if we are prohibited from making or selling our products.\n• We are subject to stringent and changing data privacy and security laws, rules, regulations, and other obligations. These areas could damage our reputation, deter customers, affect product design, or result in legal or regulatory proceedings and liability.\n- Our operating results may be adversely impacted by additional tax liabilities, higher than expected tax rates, changes in tax laws, and other tax-related factors.\n- Our business is exposed to the burden and risks associated with litigation, investigations, and regulatory proceedings.\n- Delaware law, provisions in our governing documents and our agreement with Microsoft could delay or prevent a change in control.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors Summary/Risks Related to Regulatory, Legal, Our Stock, and Other Matters", + "metadata": { + "length": 1312, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "subject", + "complex", + "laws", + "rules", + "regulations", + "political", + "actions", + "including", + "restrictions", + "export", + "products", + "adversely", + "impact", + "business", + "Scrutiny", + "corporate", + "sustainability", + "practices", + "result", + "financial", + "reputational", + "operational", + "harm", + "liability", + "Issues", + "relating", + "responsible", + "technologies", + "AI", + "Adequately", + "protecting", + "IP", + "rights", + "costly", + "ability", + "compete", + "harmed", + "unsuccessful", + "prohibited", + "making", + "selling", + "stringent", + "changing", + "data", + "privacy", + "security", + "obligations", + "These", + "areas", + "damage", + "reputation", + "deter", + "customers", + "affect", + "product", + "design", + "legal", + "regulatory", + "proceedings", + "Our", + "operating", + "results", + "impacted", + "additional", + "tax", + "liabilities", + "higher", + "expected", + "rates", + "related", + "factors", + "exposed", + "burden", + "risks", + "litigation", + "investigations", + "Delaware", + "law", + "provisions", + "governing", + "documents", + "agreement", + "Microsoft", + "delay", + "prevent", + "change", + "control" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0fb251cc-06c9-50bf-9cb1-d26f33f1029d", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b5937384-ddd1-51a3-acc6-f82b01b61ff0", + "type": "text", + "content": "Failure to meet the evolving needs of our industry and markets may adversely impact our financial results.\nOur accelerated computing platforms experience rapid changes in technology, customer requirements, competitive products, and industry standards.\nOur success depends on our ability to:\n- timely identify industry changes, adapt our strategies, and develop new or enhance and maintain existing products and technologies that meet the evolving needs of our markets, including addressing unexpected shifts in industry standards or disruptive technological innovations that could render our products incompatible with those developed by other companies;\n• develop or secure access to new products and technologies through investments in research and development;\n- launch new offerings with new business models including software, services, and cloud solutions, as well as software-, infrastructure-, or platform-as-a-service solutions;\n- expand the ecosystem for our products and technologies;\n- meet evolving and prevailing customer and industry safety, security, reliability expectations, and compliance standards;\n• manage product and software lifecycles to maintain customer and end-user satisfaction;\n• develop, acquire, maintain, and secure access to the internal and external infrastructure needed to scale our business, including sufficient energy for powering data centers using our products, acquisition integrations, customer support, e-commerce, IP licensing capabilities, and cloud service capacity; and\n• complete technical, financial, operational, compliance, sales and marketing investments for the above activities.\nWe have invested in research and development in markets where we have a limited operating history, which may not produce meaningful revenue for several years, if at all. If we fail to develop or monetize new products and technologies, or if they do not become widely adopted, our financial results could be adversely affected. Obtaining design wins may involve a lengthy process and depends on our ability to anticipate and provide features and functionality that customers will demand. They also do not guarantee revenue. Failure to obtain a design win may prevent us from obtaining future design wins in subsequent generations. We cannot ensure that our products and technologies will provide value to our customers and partners. If we fail any of these key success criteria, our financial results may be harmed.\nWe have entered into an intellectual property license arrangement with Groq, Inc., or Groq, that required significant, nonrefundable payments. Successfully incorporating the licensed technology into our architectures and product roadmaps requires significant engineering effort and may not occur on expected timelines or at all. The licensed technology may not achieve the desired results as designed or achieve customer or ecosystem adoption. The economic outcomes of this arrangement depend on our ability to translate the licensed technology into commercially viable products and services over time, and we may be unable to recover the associated costs or realize an adequate return on this spend. If our efforts to use the licensed technology are delayed or unsuccessful, our business, operating results, and financial condition could be negatively impacted.\nWe entered into multi-year cloud service agreements to support our research and development activities. The timing and availability of these cloud services have changed and may continue to shift, impacting our revenue, expenses, and development timelines, and these arrangements may not deliver anticipated benefits. We also offer or plan to offer standalone software solutions, including NVIDIA AI Enterprise, NVIDIA Omniverse, NVIDIA DRIVE, and other software products. These business models or strategies may not be successful, and we may fail to sell any meaningful standalone software or services. We may incur significant costs and may not achieve any significant revenue from these offerings.\nCompetition could adversely impact our market share and financial results.\nOur target markets remain competitive, and competition may intensify with expanding and changing product and service offerings, industry standards, customer and market needs, new entrants and consolidations. Our competitors' products, services and technologies, including those mentioned above in this Annual Report on Form 10-K, may be cheaper or provide better functionality or features than ours, which has resulted and may in the future result in lower-than-expected selling prices or demand for our products. Some of our competitors operate their own fabrication facilities, and have longer operating histories, larger customer bases, more comprehensive IP portfolios and patent protections, more design wins, and greater financial, sales, marketing and distribution resources than we do. These competitors may be able to acquire market share and/or prevent us from doing so, more effectively identify and capitalize upon opportunities in new markets and end-user trends, more quickly transition their products, and impinge on our ability to procure sufficient foundry capacity and scarce input materials during a supply-constrained environment, which could harm our business. Some of our customers have in-house expertise and internal development capabilities similar to some of ours and can use or develop their own solutions to replace those we are providing. For example, others may offer cloud-based services that compete with our AI cloud service offerings, and we may not be able to establish market share sufficient to achieve the scale necessary to meet our business objectives. If we are unable to successfully compete in this environment, demand for our products, services and technologies could decrease, which may negatively impact our business.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors/Risks Related to Our Industry and Markets", + "metadata": { + "length": 5851, + "summary": "The passage discusses risks from rapid technological changes, competition, and evolving customer needs. It highlights the need for timely innovation, R&D investments, and successful product adoption. Challenges include integrating licensed technology from Groq, cloud service agreements, and competition from larger rivals. Failure to adapt or achieve design wins could harm financial results.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Failure", + "meet", + "evolving", + "industry", + "markets", + "adversely", + "impact", + "financial", + "results", + "Our", + "accelerated", + "computing", + "platforms", + "experience", + "rapid", + "technology", + "customer", + "requirements", + "competitive", + "products", + "standards", + "success", + "depends", + "ability", + "timely", + "identify", + "adapt", + "strategies", + "develop", + "enhance", + "maintain", + "existing", + "technologies", + "including", + "addressing", + "unexpected", + "shifts", + "disruptive", + "technological", + "innovations", + "render", + "incompatible", + "developed", + "companies", + "secure", + "access", + "investments", + "research", + "development", + "launch", + "offerings", + "business", + "models", + "software", + "services", + "cloud", + "solutions", + "infrastructure", + "platform", + "service", + "expand", + "ecosystem", + "prevailing", + "safety", + "security", + "reliability", + "expectations", + "compliance", + "manage", + "product", + "lifecycles", + "end", + "user", + "satisfaction", + "acquire", + "internal", + "external", + "needed", + "scale", + "sufficient", + "energy", + "powering", + "data", + "centers", + "acquisition", + "integrations", + "support", + "commerce", + "IP", + "licensing", + "capabilities", + "capacity", + "complete", + "technical", + "operational", + "sales", + "marketing", + "activities", + "We", + "invested", + "limited", + "operating", + "history", + "produce", + "meaningful", + "revenue", + "years", + "If", + "fail", + "monetize", + "widely", + "adopted", + "affected", + "Obtaining", + "design", + "wins", + "involve", + "lengthy", + "process", + "anticipate", + "provide", + "features", + "functionality", + "customers", + "demand", + "They", + "guarantee", + "obtain", + "win", + "prevent", + "obtaining", + "future", + "subsequent", + "generations", + "ensure", + "partners", + "key", + "criteria", + "harmed", + "entered", + "intellectual", + "property", + "license", + "arrangement", + "Groq", + "Inc", + "required", + "significant", + "nonrefundable", + "payments", + "Successfully", + "incorporating", + "licensed", + "architectures", + "roadmaps", + "requires", + "engineering", + "effort", + "occur", + "expected", + "timelines", + "The", + "achieve", + "desired", + "designed", + "adoption", + "economic", + "outcomes", + "depend", + "translate", + "commercially", + "viable", + "time", + "unable", + "recover", + "costs", + "realize", + "adequate", + "return", + "spend", + "efforts", + "delayed", + "unsuccessful", + "condition", + "negatively", + "impacted", + "multi", + "year", + "agreements", + "timing", + "availability", + "changed", + "continue", + "shift", + "impacting", + "expenses", + "arrangements", + "deliver", + "anticipated", + "benefits", + "offer", + "plan", + "standalone", + "NVIDIA", + "AI", + "Enterprise", + "Omniverse", + "DRIVE", + "These", + "successful", + "sell", + "incur", + "Competition", + "market", + "share", + "target", + "remain", + "competition", + "intensify", + "expanding", + "changing", + "entrants", + "consolidations", + "competitors", + "mentioned", + "Annual", + "Report", + "Form", + "10", + "cheaper", + "resulted", + "result", + "lower", + "selling", + "prices", + "Some", + "operate", + "fabrication", + "facilities", + "longer", + "histories", + "larger", + "bases", + "comprehensive", + "portfolios", + "patent", + "protections", + "greater", + "distribution", + "resources", + "effectively", + "capitalize", + "opportunities", + "trends", + "quickly", + "transition", + "impinge", + "procure", + "foundry", + "scarce", + "input", + "materials", + "supply", + "constrained", + "environment", + "harm", + "house", + "expertise", + "similar", + "replace", + "providing", + "For", + "based", + "compete", + "establish", + "objectives", + "successfully", + "decrease" + ], + "keywords": [ + "competition", + "technology", + "financial risk" + ], + "connect_to": [] + } + }, + { + "chunk_id": "8f3faa70-7262-5765-8368-83d07350c408", + "type": "text", + "content": "Long manufacturing lead times and uncertain supply and capacity availability, combined with a failure to estimate customer demand accurately, has led and could lead to mismatches between supply and demand.\nWe have long manufacturing lead times and build finished products and maintain inventory in advance of anticipated demand. In periods of shortages impacting the semiconductor industry and/or limited supply or capacity in our supply chain, the lead times for certain supply may be extended. We have previously experienced and may continue to experience extended lead times of more than 12 months. To secure future supply and capacity, we have paid premiums, provided deposits, and entered into long-term supply agreements and capacity commitments, which have increased our product costs and this may continue. We may still be unable to secure sufficient commitments for capacity to address our business needs.\nIf we inaccurately estimate demand, or our customers change orders, as we have experienced in the past, we may not be able to reduce our supply commitments in time, at the same rate, or at all. Significant mismatches between supply and demand have varied across our market platforms, resulted in both product shortages and excess inventory, significantly harmed our financial results and could reoccur. If we underestimate demand, and our foundry partners and contract manufacturers are unable to increase production or provide sufficient supply, we may not be able to meet increased customer demand in a timely manner, or at all. Our reputation and customer relationships could be damaged and we could lose revenue and market share. Additionally, since some of our products are part of a complex data center buildout, supply constraints or availability issues with respect to any one component have had and may have a broader revenue impact. For example, our ability to sell certain products has been and could be impeded if components necessary for the finished products are not available from third parties.\nIf we overestimate demand, or if customers cancel or defer orders or choose to purchase from our competitors, we may not be able to utilize on-hand inventory or reduce purchase commitments accordingly. We have had to reduce average selling prices, including due to our channel pricing programs, increase prices for certain of our products as a result of our suppliers' increase in prices, write down our inventory, incur cancellation penalties, and record impairments, and may have to do so in the future. The impact of these risks would be amplified by our non-cancellable and non-returnable purchase orders placed in advance of our historical lead times and could be exacerbated if we need to make changes to the design of future products. These risks have increased and may continue to increase as our purchase obligations and prepaids have grown and are expected to continue to grow and become a greater portion of our total supply. All of these factors may negatively impact our gross margins and financial results.\nFactors that have caused and/or could in the future cause us to underestimate or overestimate demand, and impact the timing and volume of our revenue, include:\n• changes in product development cycles and time to market;\n• competing technologies and competitor product releases, announcements or other actions;\n• changes in business and economic conditions;\n- sudden or sustained government lockdowns or public health issues;\n• rapidly changing technology or customer requirements;\n• the availability of sufficient data center capacity or energy for customers to procure;\n- new product introductions and transitions resulting in less demand for existing products;\n- new or unexpected end-use cases;\n• increase in demand for competitive products;\n• changes in end-user demand;\n• purchasing decisions made, and inventory levels held by, distributors, ODMs, OEMs, system integrators, other channel partners and other third parties;\n- the ability of developers, end customers and other third parties to build, enhance, and maintain accelerated computing applications that leverage our platforms;\n• the demand for accelerated computing, AI-related cloud services, or large language models;\n• changes that impact the ecosystem for the architectures underlying our products and technologies;\n• government actions or changes in governmental policies, such as export controls, increased restrictions on gaming usage, or tariffs;\n- our customers' and partners' ability to secure capital and energy and to build complex datacenter infrastructure timely; and\n• the availability of third-party content on our platforms, such as GeForce NOW.\nThe availability of data centers, energy, and capital to support the buildout of NVIDIA AI infrastructure by our customers and partners is crucial, and any shortage of these and other necessary resources could impact our future revenue and financial performance. Expanding energy capacity to meet demand is a complex, multi-year process involving significant regulatory, technical, and construction challenges. In addition, access to capital can be particularly constrained for less-capitalized companies, which may face difficulties securing financing for large-scale infrastructure projects. These limitations could delay customer and partner deployments or reduce the scale of accelerated computing and AI adoption.\nChallenges in estimating demand could become more pronounced or volatile in the future on both a global and regional basis. Extended lead times may occur if we experience other supply constraints caused by natural disasters, pandemics or other events. Geopolitical tensions in regions where we rely on suppliers, contract manufacturers, and assembly partners that are critical to our supply continuity, could have a material adverse impact on us. Publicly announced intentions by governments or other companies to purchase our products can further complicate our demand estimates, as such announcements are often non-binding and may not result in committed volumes.\nWe continue to increase our supply and capacity purchases with existing and new suppliers to support our demand projections and increasing complexity of our data center products. We expect supply constraints to be a headwind to Gaming in the first quarter of fiscal year 2027 and beyond. We have also entered and may continue to enter into prepaid manufacturing and capacity agreements to supply both current and future products. The increased purchase volumes and integration of new suppliers and contract manufacturers into our supply chain creates more complexity in managing multiple suppliers with variations in production planning, execution and logistics. Our expanding product portfolio and varying component compatibility and quality may lead to increased inventory levels. We have incurred and may in the future incur inventory provisions or impairments if our inventory or supply or capacity commitments exceed demand for our products or demand declines.\nWe are increasing our U.S.-based manufacturing and investing in specialized equipment and processes to support domestic production. We may experience delays or difficulties in scaling production as planned. Our ability to increase manufacturing capabilities will depend on the domestic manufacturing ecosystem's capacity to ramp production supply to the required volume timely. Delays or shortfalls could impact our ability to meet demand.\nProduct transitions are complex and we often ship both new and prior architecture products simultaneously as our channel partners prepare to ship and support new products. We are generally in various stages of transitioning the architectures of our Data Center, Gaming, Professional Visualization, and Automotive products. The computing industry is experiencing a broader and faster launch cadence of accelerated computing platforms to meet a growing and diverse set of AI opportunities. We have introduced a new product and architecture cadence of our Data Center solutions where we seek to complete new computing solutions each year and provide a greater variety of Data Center offerings. The increased frequency of these transitions and the larger number of products and product configurations may magnify the challenges associated with managing our supply and demand which may further create volatility in our revenue. Qualification time for new products, customers anticipating product transitions, and channel partners reducing channel inventory of prior architectures ahead of new product introductions can reduce, or create volatility in, our revenue. Customers may delay adopting new architectures if their data center infrastructure is not ready, which could affect the timing of our revenue. We have experienced and may in the future experience reduced demand for current generation architectures when customers anticipate transitions, and we may be unable to sell multiple product architectures at the same time for current and future architecture transitions. Our financial results have been and may in the future be negatively impacted if we are unable to execute our architectural transitions as planned for any reason. The increased frequency and complexity of newly introduced products could result in unanticipated quality or production issues that could increase the magnitude of inventory provisions, warranty, or other costs or result in product delays. For example, our gross margins in the second quarter of fiscal year 2025 were negatively impacted by inventory provisions for low-yielding Blackwell material.\nWe incur significant engineering development resources for new products, and changes to our product roadmap may impact our ability to develop other products or adequately manage our supply chain cost. Customers may delay purchasing existing products as we increase the frequency of new products or may not be able to adopt our new products as fast as forecasted, both impacting the timing of our revenue and supply chain cost. While we have managed prior product transitions and have sold multiple product architectures at the same time, these transitions are difficult, may impair our ability to predict demand and impact our supply mix, and may cause us to incur additional costs.\nDemand estimates for our products, applications, and services can be incorrect, which may create volatility in our revenue or supply levels. We may not be able to generate significant revenue from them. Because our products may be used in multiple use cases and applications, it is difficult to estimate with any reasonable degree of precision the impact of accelerated computing and AI models on our reported revenue or forecasted demand.\nThe use of our GPUs for new, mercurial, or trendy applications, has impacted and can impact in the future, demand for our products, including by leading to inconsistent spikes and drops in demand. For example, several years ago, our Gaming GPUs began to be used for mining digital currencies, such as Ethereum. It is difficult for us to estimate with any reasonable degree of precision the past or current impact of cryptocurrency mining, or forecast the future impact of cryptocurrency mining, on demand for our products. Volatility in the cryptocurrency market, including new compute technologies, price changes in cryptocurrencies, government cryptocurrency policies and regulations, new cryptocurrency standards and changes in the method of verifying blockchain transactions, has impacted and can in the future impact cryptocurrency mining and demand for our products and can further impact our ability to estimate demand for our products. Changes to cryptocurrency standards and processes including, but not limited to, the Ethereum 2.0 merge in 2022, have reduced and may in the future decrease the usage of GPUs for Ethereum mining. This has created and may in the future create increased aftermarket sales of our GPUs, which could negatively impact retail prices for our GPUs and reduce demand for our new GPUs. In general, our new products or previously sold products may be resold online or on the unauthorized “gray market,” which also makes demand forecasting difficult. Gray market products and reseller marketplaces compete with our new products and distribution channels. Our inability to accurately predict our demand that arises from new use cases may create volatility in our revenue.\nDependency on third-party suppliers and their technology to manufacture, assemble, test, or package our products reduces our control over product quantity and quality, manufacturing yields, and product delivery schedules and could harm our business.\nWe depend on foundries to manufacture our semiconductor wafers using their fabrication equipment and techniques. We do not assemble, test, or package our products, but instead contract with independent subcontractors. These subcontractors assist with procuring components used in our systems, boards, and products. We face risks which have adversely affected or could adversely affect our ability to meet customer demand and scale our supply chain, negatively impact longer-term demand for our products and services, and adversely affect our business operations, gross margin, revenue and/or financial results, including:\n- lack of guaranteed supply of components and capacity;\n• decommitment by our suppliers;\n- potential higher wafer and component prices resulting from incorrectly estimating demand and failing to place orders with our suppliers with sufficient quantities or timely;\n- failure by our foundries or contract manufacturers to procure raw materials or provide adequate levels of manufacturing or test capacity for our products;\n- failure by our foundries to develop, obtain, or successfully implement high quality process technologies, including transitions to smaller geometry process technologies such as advanced process node technologies and memory designs needed to manufacture our products;\n• failure by our suppliers to comply with our policies and expectations and emerging regulatory requirements;\n- limited number and geographic concentration of global suppliers, foundries, contract manufacturers, assembly and test providers and memory manufacturers;\n- loss of a supplier and additional expense and/or production delays as a result of qualifying a new foundry or subcontractor and commencing volume production or testing in the event of a loss, addition or change of a supplier;\n- lack of direct control over product quantity, quality, and delivery schedules;\n- integration of new suppliers and contract manufacturers creating more complexity in managing multiple suppliers with variations in production planning, execution, and logistics;\n- suppliers or their suppliers failing to provide high quality products and/or making changes to their products without our qualification;\n- delays in product shipments, shortages, a decrease in product quality and/or higher expenses in the event our subcontractors or foundries prioritize our competitors' or other customers' orders over ours;\n- requirements to place orders that are not cancellable upon changes in demand or requirements to prepay for supply in advance;\n- low manufacturing yields resulting from issues in our product design or a foundry's proprietary process technology;\n• suppliers extending lead times and/or increasing costs during shortages; and\n- disruptions in manufacturing, assembly and other processes due to closures related to heat waves, earthquakes, fires, or other natural disasters, electricity conservation efforts, pandemics, and cybersecurity incidents.\nDefects in our products have caused and could cause us to incur significant expenses to remediate, which can damage our reputation and cause us to lose market share.\nOur hardware and software product and service offerings are complex. They have in the past and may in the future contain defects, security vulnerabilities, experience failures, or unsatisfactory performance due to issues in design, fabrication, packaging, materials, bugs and/or use within a system. These risks may also increase when our products are introduced into new devices, markets, technologies and applications, or new versions are released, and when we rely on partners to supply and manufacture components that are used in our products, as these arrangements reduce our direct control over production. AI software products that we or our partners offer rely on training data that may originate from third parties and new training methods, and the resulting products may contain unknown or undetected defects and errors, or reflect unintended bias. Although arrangements with component providers may contain provisions for product defect expense reimbursement, we generally remain responsible to the customer for warranty product defects that may occur from time to time. Some failures in our products or services have been in the past and may in the future be only discovered after a product or service has been shipped or used.\nUndiscovered vulnerabilities in our products or services could result in loss of data or intangible property, or expose our customers to unscrupulous third parties who develop and deploy malicious software programs that could attack our products or services. Defects or failure of our offerings to perform to specifications could lead to substantial damage to the products in which our offerings have been integrated by OEMs, ODMs, AIB manufacturers, automotive manufacturers, and tier 1 automotive suppliers, and to the user of such end product. Such defects have in the past had an adverse effect on our cost and supply of components and finished goods and may in the future cause us to incur significant warranty, support, and repair or replacement costs as part of a product recall or otherwise, write-off the value of related inventory, and divert the attention of our engineering and management personnel from our product development efforts to find and correct the issue. Our efforts to remedy these issues may not be timely or satisfactory to our customers. An error or defect in new products, releases or related software drivers after commencement of commercial shipments could result in failure to achieve market acceptance, loss of design wins, temporary or permanent withdrawal from a product or market and harm to our relationships with existing and prospective customers and partners and consumers' perceptions of our brand, which would in turn negatively impact our business operations, gross margin, revenue and/or financial results. We may be required to reimburse our customers, partners or consumers, including for costs to repair or replace products in the field or in connection with indemnification obligations, or pay fines imposed by regulatory agencies.\nIn general, if a product liability claim regarding any of our products is brought against us, even if the alleged damage is due to the actions or inactions of a third party, such as within our supply chain, the cost of defending the claim could be significant and would divert the efforts of our technical and management personnel and harm our business. Further, our business liability insurance may be inadequate or future coverage may be unavailable on acceptable terms, which could adversely impact our financial results.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors/Risks Related to Demand, Supply, and Manufacturing", + "metadata": { + "length": 19345, + "summary": "The passage discusses risks from long manufacturing lead times, uncertain supply, and inaccurate demand estimation leading to supply-demand mismatches. It highlights challenges such as extended lead times over 12 months, premium payments for capacity, and non-cancellable purchase orders. Underestimating demand can cause revenue loss and reputational damage, while overestimation leads to inventory write-downs and price reductions. Factors like product transitions, competitor actions, and geopolitical tensions complicate demand forecasting. Dependency on third-party suppliers introduces risks of defects, yield issues, and production delays. Product defects may incur significant remediation costs and harm market share. The text emphasizes volatility from new use cases like cryptocurrency mining and gray market sales, and the impact of increasing product complexity and frequency of architectural transitions on supply chain management.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Long", + "manufacturing", + "lead", + "times", + "uncertain", + "supply", + "capacity", + "availability", + "combined", + "failure", + "estimate", + "customer", + "demand", + "accurately", + "led", + "mismatches", + "We", + "long", + "build", + "finished", + "products", + "maintain", + "inventory", + "advance", + "anticipated", + "In", + "periods", + "shortages", + "impacting", + "semiconductor", + "industry", + "limited", + "chain", + "extended", + "previously", + "experienced", + "continue", + "experience", + "12", + "months", + "To", + "secure", + "future", + "paid", + "premiums", + "provided", + "deposits", + "entered", + "term", + "agreements", + "commitments", + "increased", + "product", + "costs", + "unable", + "sufficient", + "address", + "business", + "If", + "inaccurately", + "customers", + "change", + "orders", + "past", + "reduce", + "time", + "rate", + "Significant", + "varied", + "market", + "platforms", + "resulted", + "excess", + "significantly", + "harmed", + "financial", + "results", + "reoccur", + "underestimate", + "foundry", + "partners", + "contract", + "manufacturers", + "increase", + "production", + "provide", + "meet", + "timely", + "manner", + "Our", + "reputation", + "relationships", + "damaged", + "lose", + "revenue", + "share", + "Additionally", + "part", + "complex", + "data", + "center", + "buildout", + "constraints", + "issues", + "respect", + "component", + "broader", + "impact", + "For", + "ability", + "sell", + "impeded", + "components", + "parties", + "overestimate", + "cancel", + "defer", + "choose", + "purchase", + "competitors", + "utilize", + "hand", + "average", + "selling", + "prices", + "including", + "due", + "channel", + "pricing", + "programs", + "result", + "suppliers", + "write", + "incur", + "cancellation", + "penalties", + "record", + "impairments", + "The", + "risks", + "amplified", + "cancellable", + "returnable", + "historical", + "exacerbated", + "make", + "design", + "These", + "obligations", + "prepaids", + "grown", + "expected", + "grow", + "greater", + "portion", + "total", + "All", + "factors", + "negatively", + "gross", + "margins", + "Factors", + "caused", + "timing", + "volume", + "include", + "development", + "cycles", + "competing", + "technologies", + "competitor", + "releases", + "announcements", + "actions", + "economic", + "conditions", + "sudden", + "sustained", + "government", + "lockdowns", + "public", + "health", + "rapidly", + "changing", + "technology", + "requirements", + "energy", + "procure", + "introductions", + "transitions", + "resulting", + "existing", + "unexpected", + "end", + "cases", + "competitive", + "user", + "purchasing", + "decisions", + "made", + "levels", + "held", + "distributors", + "ODMs", + "OEMs", + "system", + "integrators", + "developers", + "enhance", + "accelerated", + "computing", + "applications", + "leverage", + "AI", + "related", + "cloud", + "services", + "large", + "language", + "models", + "ecosystem", + "architectures", + "underlying", + "governmental", + "policies", + "export", + "controls", + "restrictions", + "gaming", + "usage", + "tariffs", + "capital", + "datacenter", + "infrastructure", + "party", + "content", + "GeForce", + "NOW", + "centers", + "support", + "NVIDIA", + "crucial", + "shortage", + "resources", + "performance", + "Expanding", + "multi", + "year", + "process", + "involving", + "significant", + "regulatory", + "technical", + "construction", + "challenges", + "addition", + "access", + "constrained", + "capitalized", + "companies", + "face", + "difficulties", + "securing", + "financing", + "scale", + "projects", + "limitations", + "delay", + "partner", + "deployments", + "adoption", + "Challenges", + "estimating", + "pronounced", + "volatile", + "global", + "regional", + "basis", + "Extended", + "occur", + "natural", + "disasters", + "pandemics", + "events", + "Geopolitical", + "tensions", + "regions", + "rely", + "assembly", + "critical", + "continuity", + "material", + "adverse", + "Publicly", + "announced", + "intentions", + "governments", + "complicate", + "estimates", + "binding", + "committed", + "volumes", + "purchases", + "projections", + "increasing", + "complexity", + "expect", + "headwind", + "Gaming", + "quarter", + "fiscal", + "2027", + "enter", + "prepaid", + "current", + "integration", + "creates", + "managing", + "multiple", + "variations", + "planning", + "execution", + "logistics", + "expanding", + "portfolio", + "varying", + "compatibility", + "quality", + "incurred", + "provisions", + "exceed", + "declines", + "based", + "investing", + "specialized", + "equipment", + "processes", + "domestic", + "delays", + "scaling", + "planned", + "capabilities", + "depend", + "ramp", + "required", + "Delays", + "shortfalls", + "Product", + "ship", + "prior", + "architecture", + "simultaneously", + "prepare", + "generally", + "stages", + "transitioning", + "Data", + "Center", + "Professional", + "Visualization", + "Automotive", + "experiencing", + "faster", + "launch", + "cadence", + "growing", + "diverse", + "set", + "opportunities", + "introduced", + "solutions", + "seek", + "complete", + "variety", + "offerings", + "frequency", + "larger", + "number", + "configurations", + "magnify", + "create", + "volatility", + "Qualification", + "anticipating", + "reducing", + "ahead", + "Customers", + "adopting", + "ready", + "affect", + "reduced", + "generation", + "anticipate", + "impacted", + "execute", + "architectural", + "reason", + "newly", + "unanticipated", + "magnitude", + "warranty", + "2025", + "low", + "yielding", + "Blackwell", + "engineering", + "roadmap", + "develop", + "adequately", + "manage", + "cost", + "adopt", + "fast", + "forecasted", + "While", + "managed", + "sold", + "difficult", + "impair", + "predict", + "mix", + "additional", + "Demand", + "incorrect", + "generate", + "Because", + "reasonable", + "degree", + "precision", + "reported", + "GPUs", + "mercurial", + "trendy", + "leading", + "inconsistent", + "spikes", + "drops", + "years", + "ago", + "began", + "mining", + "digital", + "currencies", + "Ethereum", + "It", + "cryptocurrency", + "forecast", + "Volatility", + "compute", + "price", + "cryptocurrencies", + "regulations", + "standards", + "method", + "verifying", + "blockchain", + "transactions", + "Changes", + "2.0", + "merge", + "2022", + "decrease", + "This", + "created", + "aftermarket", + "sales", + "retail", + "general", + "resold", + "online", + "unauthorized", + "gray", + "makes", + "forecasting", + "Gray", + "reseller", + "marketplaces", + "compete", + "distribution", + "channels", + "inability", + "arises", + "Dependency", + "manufacture", + "assemble", + "test", + "package", + "reduces", + "control", + "quantity", + "yields", + "delivery", + "schedules", + "harm", + "foundries", + "wafers", + "fabrication", + "techniques", + "independent", + "subcontractors", + "assist", + "procuring", + "systems", + "boards", + "adversely", + "affected", + "longer", + "operations", + "margin", + "lack", + "guaranteed", + "decommitment", + "potential", + "higher", + "wafer", + "incorrectly", + "failing", + "place", + "quantities", + "raw", + "materials", + "adequate", + "obtain", + "successfully", + "implement", + "high", + "smaller", + "geometry", + "advanced", + "node", + "memory", + "designs", + "needed", + "comply", + "expectations", + "emerging", + "geographic", + "concentration", + "providers", + "loss", + "supplier", + "expense", + "qualifying", + "subcontractor", + "commencing", + "testing", + "event", + "direct", + "creating", + "making", + "qualification", + "shipments", + "expenses", + "prioritize", + "prepay", + "proprietary", + "extending", + "disruptions", + "closures", + "heat", + "waves", + "earthquakes", + "fires", + "electricity", + "conservation", + "efforts", + "cybersecurity", + "incidents", + "Defects", + "remediate", + "damage", + "hardware", + "software", + "service", + "They", + "defects", + "security", + "vulnerabilities", + "failures", + "unsatisfactory", + "packaging", + "bugs", + "devices", + "markets", + "versions", + "released", + "arrangements", + "offer", + "training", + "originate", + "methods", + "unknown", + "undetected", + "errors", + "reflect", + "unintended", + "bias", + "Although", + "defect", + "reimbursement", + "remain", + "responsible", + "Some", + "discovered", + "shipped", + "Undiscovered", + "intangible", + "property", + "expose", + "unscrupulous", + "deploy", + "malicious", + "attack", + "perform", + "specifications", + "substantial", + "integrated", + "AIB", + "automotive", + "tier", + "Such", + "effect", + "goods", + "repair", + "replacement", + "recall", + "divert", + "attention", + "management", + "personnel", + "find", + "correct", + "issue", + "remedy", + "satisfactory", + "An", + "error", + "drivers", + "commencement", + "commercial", + "achieve", + "acceptance", + "wins", + "temporary", + "permanent", + "withdrawal", + "prospective", + "consumers", + "perceptions", + "brand", + "turn", + "reimburse", + "replace", + "field", + "connection", + "indemnification", + "pay", + "fines", + "imposed", + "agencies", + "liability", + "claim", + "brought", + "alleged", + "inactions", + "defending", + "Further", + "insurance", + "inadequate", + "coverage", + "unavailable", + "acceptable", + "terms" + ], + "keywords": [ + "supply chain", + "demand estimation", + "product transitions" + ], + "connect_to": [] + } + }, + { + "chunk_id": "e8c12b1d-e479-5aab-82dd-a41ce7a944aa", + "type": "text", + "content": "Adverse economic conditions may harm our business.\nEconomic and industry uncertainty or changes, including recession or slowing growth, inflation, changes or uncertainty in fiscal, monetary, or trade policy, disruptions to capital markets and the banking system, currency fluctuations, higher interest rates, tighter credit, lower capital expenditures by businesses, including on IT infrastructure, increases in unemployment, labor shortages, and lower consumer confidence and spending, global supply chain constraints, and global economic and geopolitical developments, including the implementation of tariffs by the USG or other governments, have in the past and/or could in the future have adverse, wide-ranging effects on our business and financial results, including:\n\\- increased costs for wafers, components, logistics, and other supply chain expenses, which have negatively impacted our gross margin in the past and may do so in the future;\n- increased supply, employee, facilities and infrastructure costs and volatility in the financial markets, which have reduced and may in the future reduce our margins;\n• decrease in demand for our products, services and technologies and those of our customers, partners or licensees;\n- the inability of our suppliers to deliver on their supply commitments to us and our customers' or our licensees' inability to supply products to customers and/or end users;\n• limits on our ability to forecast operating results and make business decisions;\n- the insolvency of key suppliers, distributors, customers, CSPs, data center providers, licensing parties or other third parties we rely on;\n- reduced profitability of customers, which may cause them to scale back operations, exit businesses, file for bankruptcy protection and potentially cease operations, or lead to mergers, consolidations or strategic alliances among other companies, which could adversely affect our ability to compete effectively;\n- increased credit and collectability risks, higher borrowing costs or reduced availability of capital markets, reduced liquidity, adverse impacts on our customers and suppliers; and\n- failures of counterparties, including financial institutions and insurers, asset impairments, and declines in the value of our financial instruments.\nAdverse developments affecting financial institutions, such as bank failures or instability, or concerns or speculation about similar events or risks, could lead to market-wide liquidity problems and other disruptions, which could impact our customers' ability to fulfill their payment obligations to us, our vendors' ability to fulfill their contractual obligations to us, or our ability to fulfill our own obligations.\nAdditionally, we maintain a portfolio of liquid investments for cash management purposes, including various holdings, types, and maturities. These investments are subject to general credit, liquidity, market and interest rate risks, which may be exacerbated by market downturns or events that affect global financial markets, as described above. A majority of our investment portfolio comprises USG securities. A decline in global financial markets for long periods or a downgrade of the USG credit rating due to an actual or threatened default on government debt could result in higher interest rates, a decline in the value of the U.S. dollar, reduced market liquidity or other adverse conditions. These factors could cause an unrealized or realized loss position in our investments or require us to record impairment charges.\nInternational sales and operations are a significant part of our business, which exposes us to risks that could harm our business.\nWe sell our products internationally, and we also have operations and conduct business internationally. Our semiconductor wafers are manufactured, assembled, tested and packaged by third parties located outside of the United States, and we generated 31% of our revenue in fiscal year 2026 from sales outside the United States. The market in China, where our offerings are limited by export controls, is highly competitive and we expect it to remain competitive going forward. The global nature of our business subjects us to a number of risks and uncertainties, which have had in the past and could in the future have a material adverse effect on our business, financial condition and results of operations. These include domestic and international economic and political conditions in countries in which we and our suppliers and manufacturers do business, government lockdowns to control case spread of global or local health issues, differing legal standards with respect to protection of IP and employment practices, different domestic and international business and cultural practices, disruptions to capital markets, counter-inflation policies, currency fluctuations, natural disasters, acts of war or other military actions, terrorism, public health issues, restrictions on international trade, such as tariffs, sanctions, and other controls on imports or exports, and catastrophic events.\nProduct, system security, and data protection incidents or breaches, as well as cyber-attacks, could disrupt our operations, reduce our expected revenue, increase our expenses, and significantly harm our business and reputation.\nSecurity breaches, computer malware, social-engineering attacks, denial-of-service attacks, software bugs, server malfunctions, software or hardware failures, loss of data or other information technology assets, and other cyber-attacks are becoming increasingly sophisticated, making it more difficult to successfully detect, defend against them or implement adequate preventative measures.\nCyber-attacks, including ransomware attacks by organized criminal threat actors, nation-states, and nation-state-supported actors, may become more prevalent and severe. Our ability to recover from ransomware attacks may be limited if our backups have been affected by the attack, or if restore from backups is delayed or not feasible.\nIndividuals, groups of hackers and sophisticated organizations, including nation-states and nation-state-supported actors, and other threat actors have engaged and are expected to continue to engage in cyber-attacks. Additionally, some actors are using AI technology to launch more automated, targeted and coordinated attacks. Due to geopolitical conflicts and during times of war or other major conflicts, we and the third parties we rely upon may be subject to a heightened risk of cyber-attacks that could materially disrupt our ability to provide services and products. We may also face cybersecurity threats due to error or intentional misconduct by employees, contractors or other third-party service providers. Certain aspects of effective cybersecurity are dependent upon our employees, contractors and/or other third-party service providers safeguarding our sensitive information and adhering to our security policies and access control mechanisms. We have in the past experienced, and may in the future experience, security incidents arising from a failure to properly handle sensitive information or adhere to our security policies and access control mechanisms, including, for example, employees posting company data on third-party websites without permission, and, although no such events have had a material adverse effect on our business, there can be no assurance that an insider threat or error will not result in an incident that is material to us or lead to negative publicity. Furthermore, we rely on products and services provided by third-party suppliers to operate certain critical business systems, including without limitation, cloud-based infrastructure, encryption and authentication technology, employee email and other functions, which exposes us to supply-chain attacks or other business disruptions. We cannot guarantee that third parties and infrastructure in our supply chain or our partners' supply chains have not been compromised or that they do not contain exploitable vulnerabilities, defects or bugs that could result in a breach of or disruption to our information technology systems, including our products and services, or the third-party information technology systems that support our services. We have incorporated third-party data into some of our AI models and used open-source datasets to train our models and may continue to do so. These datasets may be flawed, insufficient, or contain certain biased information, and may otherwise decrease resilience to security incidents that may compromise the integrity of our AI outputs, leading to potential reputational damage, regulatory scrutiny, or adverse impacts on the performance and reliability of our products, which could, in turn, affect our partners' operations, customer trust, and our revenue. We may have limited insight into the data privacy or security practices of third-party suppliers, including for our AI algorithms. Our ability to monitor these third parties' information security practices is limited, and they may not have adequate information security measures in place. In addition, if one of our third-party suppliers suffers a security incident (which has happened in the past and may happen in the future), our response may be limited or more difficult because we may not have direct access to their systems, logs and other information related to the security incident. Additionally, we are incorporated into the supply chain of a large number of entities worldwide and, as a result, if our products or services are compromised, a significant number of our customers and their data could be affected, which could result in potential liability and harm our business.\nTo defend against security incidents, we must continuously engineer more secure products and enhance security and reliability features, which is expected to result in increased expenses. We must also continue to develop our security measures, including training programs and security awareness initiatives, designed to ensure our suppliers have appropriate security measures in place, and continue to meet the evolving security requirements of our customers, applicable industry standards, and government regulations. While we invest in training programs and security awareness initiatives and take steps to detect and remediate certain vulnerabilities that we have identified, we may not always be able to prevent threats or detect and mitigate all vulnerabilities in our security controls, systems or software, including third-party software we have installed, as such threats and techniques change frequently and may not be detected until after a security incident has occurred. Further, we may experience delays in developing and deploying remedial measures designed to address identified vulnerabilities. These vulnerabilities could result in reputational and financial harm, and if exploited, these vulnerabilities could result in a security incident.\nWe hold confidential, sensitive, personal and proprietary information, including information from partners and customers. Breaches of our security measures, along with reported or perceived vulnerabilities or unapproved dissemination of proprietary information or sensitive or confidential data about us or third parties, could expose us and the parties affected to a risk of loss, or misuse of this information, potentially resulting in litigation and subsequent liability, regulatory inquiries or actions, damage to our brand and reputation or other harm, including financial, to our business. For example, we hold proprietary game source code from third-party partners in our GFN service. Breaches of our GFN security measures, which have happened in the past, could expose our partners to the risk of loss or misuse of this source code, damage both us and our partners, and expose NVIDIA to potential litigation and liability. If we or a third party we rely on experience a security incident, which has occurred in the past, or are perceived to have experienced a security incident, we may experience adverse consequences, including government enforcement actions, additional reporting requirements and/or oversight, restrictions on processing data, litigation, indemnification obligations, reputational harm, diversion of funds, diversion of management attention, financial loss, loss of data, material disruptions in our systems and operations, supply chain, and ability to produce, sell and distribute our goods and services, and other similar harms. Inability to fulfill orders, delayed sales, lower margins or lost customers as a result of these disruptions could adversely affect our financial results, stock price and reputation. We are required by certain data privacy and security obligations to notify relevant stakeholders, including affected individuals, customers, regulators and investors, of security incidents, and mandatory disclosure of such incidents could lead to negative publicity. In addition to experiencing a security incident, third parties may gather, collect or infer sensitive information about us from public sources, data brokers or other means that reveals competitively sensitive details about our organization and could be used to harm our business.\nBusiness disruptions could harm our operations, lead to a decline in revenue and increase our costs.\nFactors that have caused and/or could in the future cause disruptions to our worldwide operations include: natural disasters, extreme weather conditions, power or water shortages, critical infrastructure failures, telecommunications failures, supplier disruptions, terrorist attacks, acts of violence, political and/or civil unrest, acts of war or other military actions, epidemics or pandemics, abrupt regulatory changes, and other natural or man-made disasters and catastrophic events. Our corporate headquarters, a large portion of our current data center capacity, and a portion of our research and development activities are located in California, and other critical business operations, finished goods inventory and some of our suppliers are located in Asia, making our operations vulnerable to natural disasters such as earthquakes, wildfires or other business disruptions occurring in these geographical areas. Catastrophic events can also have an impact on third-party vendors who provide us critical infrastructure services for IT and research and development systems and personnel. Geopolitical and domestic political developments and other events beyond our control can increase economic volatility globally. Political instability, changes in government or adverse political developments in or around any of the major countries in which we do business may harm our business, financial condition, and results of operations. Worldwide geopolitical tensions and conflicts, including but not limited to China, Hong Kong, Israel, Korea and Taiwan where the manufacture of our product components and final assembly of our products are concentrated may result in changing regulatory requirements, and other disruptions that could impact our operations and operating strategies, product demand, access to global markets, hiring, and profitability. For example, other countries have restricted and may continue in the future to restrict business with the State of Israel, where we have engineering, sales support operations and manufacturing, and companies with Israeli operations, including by economic boycotts. Our operations could be harmed and our costs could increase if manufacturing, logistics, or other operations are disrupted for any reason, including natural disasters, high heat events, water shortages, power shortages, information technology system failures or cyberattacks, military actions or economic, and business, labor, environmental, public health, or political issues. The ultimate impact on us, our third-party foundries and other suppliers of being located and consolidated in certain geographical areas is unknown. In the event a disaster, war, or catastrophic event affects us, the third-party systems on which we rely, or our customers, our business could be harmed as a result of declines in revenue, increases in expenses, and substantial expenditures and time spent to fully resume operations. Our business continuity and disaster recovery planning may not be sufficient for all eventualities. All of these risks and conditions could materially adversely affect our future sales and operating results.\nWe are monitoring the impact of the geopolitical conflict in and around Israel on our operations, including the health and safety of our approximately 6,000 employees in the region who primarily support the research and development, operations, and sales and marketing of our networking products. We have experienced periods where some of our employees in the region have been on active military duty for an extended period, which caused limited disruption to our product development or operations. We have not experienced significant impact or expense to our business; however, if the conflict is further extended or expanded, it could impact future product development, operations, and revenue or create other uncertainty for our business.\nAdditionally, interruptions or delays in services from CSPs, data center co-location partners, and other third parties on which we rely, including due to the events described above or other events such as the insolvency of these parties, could impair our ability to provide our products and services and harm our business. As we increase our reliance on these third-party systems and services, our exposure to damage from service interruptions, defects, disruptions, outages, shortages and other performance and quality problems may increase. Data centers depend on access to clean water and predictable energy. Power or water shortages, land or permitting constraints, or regulations that limit energy, water, or land availability, could impair the ability of our customers to expand their data center capacity and consume our products and services, which may in turn negatively impact our business.\nClimate change may have a long-term impact on our business.\nClimate change may have an increasingly adverse impact on our business and on our customers, partners and vendors. Water and energy availability and reliability in the regions where we conduct business is critical, and certain of our facilities may be vulnerable to the impacts of extreme weather events. Extreme heat and wind coupled with dry conditions in Northern California may lead to power safety shut offs due to wildfire risk, which can have adverse implications for our Santa Clara, California headquarter offices and data centers, including impairing the ability of our employees to work effectively. Climate change, its impact on our supply chain and critical infrastructure worldwide and its potential to increase political instability in regions where we, our customers, partners and our vendors do business, may disrupt our business and cause us to experience higher attrition, losses and costs to maintain or resume operations. Although we maintain insurance coverage for a variety of property, casualty, and other risks, the types and amounts of insurance we obtain vary depending on availability and cost. Some of our policies have large deductibles and broad exclusions, and our insurance providers may be unable or unwilling to pay a claim. Losses not covered by insurance may be large, which could harm our results of operations and financial condition.\nOur business and those of our suppliers and customers is subject to sustainability-related laws, regulations and lawsuits. New or proposed regulations relating to carbon taxes, fuel or energy taxes, pollution limits, sustainability-related disclosure and governance and supply chain governance could result in greater direct costs, including costs associated with changes to manufacturing processes or the procurement of raw materials used in manufacturing processes, increased capital expenditures to improve facilities and equipment, higher compliance and energy costs to reduce emissions, other compliance costs, and greater indirect costs resulting from our customers and/or suppliers incurring additional compliance costs that are passed on to us. These costs and restrictions could harm our business and results of operations by increasing our expenses or requiring us to alter our operations and product design activities.\nStakeholder groups may find us insufficiently responsive to the implications of climate change, and therefore we may face legal action or reputational harm. We may not achieve our stated sustainability-related goals, which could harm our reputation, or we may incur additional, unexpected costs to achieve such goals. We may also experience contractual disputes due to supply chain delays arising from climate change-related disruptions, which could result in increased litigation and costs.\nWe also face risks related to business trends that may be influenced by climate change concerns. Our business could be negatively impacted by concerns around the high absolute energy requirements of our GPUs, despite their much more energy efficient design and operation relative to alternative computing platforms.\nWe may not be able to realize the potential benefits of business investments or acquisitions, and we may not be able to successfully integrate acquired companies, which could hurt our ability to grow our business, develop new products or sell our products.\nWe acquire and invest in businesses that offer products, services and technologies that we believe will help expand or enhance our strategic objectives. Acquisitions or investments involve significant challenges and risks and could impair our ability to grow our business, develop new products or sell our products and ultimately could have a negative impact on our financial results. If we pursue a particular transaction, we may limit our ability to enter into other transactions that could help us achieve our other strategic objectives. If we are unable to timely complete acquisitions, including due to delays and challenges in obtaining regulatory approvals, we may be unable to pursue other transactions, we may not be able to retain critical talent from the target company, technology may evolve and make the acquisition less attractive, and other changes can take place, which could reduce the anticipated benefits of the transaction and negatively impact our business. Regulators could also impose conditions that reduce the ultimate value of our acquisitions. In addition, to the extent that our perceived ability to consummate acquisitions is harmed, future acquisitions may be more difficult, complex or expensive.\nOur investments in companies could create volatility and fluctuations in our results. These investments may generate realized and unrealized gains or losses and we could realize losses up to the value of the investments. We have invested and may continue to invest in our ecosystem. Many of these companies may not achieve profitability in the near term, or at all, and there is no guarantee that we will realize a return on our investment. We may continue to invest in companies to further our strategic objectives and to support certain key business initiatives, which could be subject to delays and challenges in obtaining regulatory approvals. Our investments in private companies include early-stage companies still defining their strategic direction. Many of the securities in which we invest are non-marketable and illiquid at the time of our initial investment. To the extent any of the companies in which we invest are not successful, we could recognize an impairment and/or lose all or part of our investment.\nWe are finalizing an investment and partnership agreement with OpenAI. There is no assurance that we will enter into an investment and partnership agreement with OpenAI or that a transaction will be completed.\nOur investment portfolio contains industry sector concentration risks, and a decline in any one or multiple industry sectors could increase our impairment losses.\nWe face additional risks related to acquisitions and strategic investments, including the diversion of capital and other resources, including management's attention; difficulty in realizing a satisfactory return and uncertainties to realize the benefits of an acquisition or strategic investment, if at all; difficulty or inability in obtaining governmental, regulatory approval or restrictions or other consents and approvals or financing; legal proceedings initiated as a result of an acquisition or investment; and potential failure of our due diligence processes to identify significant issues with the assets or company in which we are investing or are acquiring.\nAdditional risks relating to acquisitions include, but are not limited to:\n- difficulty in integrating the technology, systems, products, policies, processes, or operations and integrating and retaining the employees, including key personnel, of the acquired business;\n• assumption of liabilities and incurring amortization expenses, impairment charges to goodwill or write-downs of acquired assets;\n• integrating accounting, forecasting and controls, procedures and reporting cycles;\n• coordinating and integrating operations, particularly in countries in which we do not currently operate;\n- stock price impact, fines, fees or reputation harm if we are unable to obtain regulatory approval for an acquisition or are otherwise unable to close an acquisition;\n- potential issuances of debt to finance our acquisitions, resulting in increased debt, increased interest expense, and compliance with debt covenants or other restrictions;\n• the potential for our acquisitions to result in dilutive issuances of our equity securities;\n• the potential variability of the amount and form of any performance-based consideration;\n- negative changes in general economic conditions in the regions or the industries in which we or our target operate;\n• exposure to additional cybersecurity risks and vulnerabilities; and\n- impairment of relationships with, or loss of our or our target's employees, vendors and customers.\nFor example, when integrating acquisition target systems into our own, we have experienced and may continue to experience challenges including lengthy and costly systems integration, delays in purchasing and shipping products, difficulties with system integration via electronic data interchange and other processes with our key suppliers and customers, and training and change management needs of integration personnel. These challenges have impacted our results of operations and may continue to do so in the future.\nWe receive a significant amount of our revenue from a limited number of partners and distributors and we have a concentration of sales to customers who purchase directly or indirectly from us, and our revenue could be adversely affected if we lose or are prevented from selling to any of these customers.\nWe have experienced periods where we receive a significant amount of our revenue from a limited number of customers, and this trend may continue. For fiscal year 2026, sales to one direct customer represented 22% of total revenue and sales to another direct customer represented 14% of total revenue, all of which were primarily attributable to the Compute & Networking segment. With several of these partners, we are selling multiple products and systems in our portfolio through their channels. Our operating results depend on sales to our partner network, as well as the ability of these partners to sell products that incorporate our technologies. We have a small number of partners that are involved in system integration with our key customers. As our system design becomes increasingly complex, system integrators may be unable to meet specifications of our key customers. Changes in our partners' or customers' business models or their ownership can reduce the number of partners available to us and harm our ability to sell our advanced data center systems to customers. In the future, these partners may decide to purchase fewer products, not to incorporate our products into their ecosystem, or to alter their purchasing patterns in some other way. Because most of our sales are made on a purchase order basis, our customers can generally cancel, change, or delay product purchase commitments with little notice to us and without penalty. Our partners or customers may develop their own solutions; our customers may purchase products from our competitors; and our partners may discontinue sales or lose market share in the markets for which they purchase our products, all of which may alter partners' or customers' purchasing patterns. Many of our indirect customers often do not purchase directly from us but through multiple OEMs, ODMs, system integrators, distributors, and other channel partners. We generate a significant amount of our revenue from a limited number of indirect customers, and we estimate some individually representing 10% or more of our revenue. If end demand increases or our finished goods supply availability is concentrated near a quarter end, the system integrators, distributors, and channel partners may have limited ability to increase their credit, which could impact the timing and amount of our revenue. The loss of any of our large customers, a significant reduction in purchases by them, our inability to sell to a customer due to U.S. or other countries' trade restrictions, or any difficulties in collecting accounts receivable would likely harm our financial condition and results of operations.\nCommercial arrangements expose us to counterparty risks.\nWe have entered and may in the future enter into commercial arrangements, including long-term capacity purchase obligations and financial guarantees, and have been asked to offer financing arrangements to support our customers' and partners' buildout of datacenter infrastructure. We have not entered into any financing arrangements. Commercial arrangements expose us to counterparty risk, including customers' or partners' inability to fulfill their financial commitments and secure necessary financing or infrastructure, the occurrence of significant project delays, and counterparty financial distress or insolvency, all of which may negatively impact our business, financial condition, or results of operations. Financing arrangements, if undertaken, may in some circumstances result in lower upfront cash flows associated with extended payment terms or payment terms made over a multi-year term and may increase credit risk.\nIf we are unable to attract, retain and motivate our executives and key employees, our business may be harmed.\nTo remain competitive and successfully execute our business strategy, we must attract, retain, and motivate our executives and key employees, as well as recruit and develop exceptional talent. However, labor is subject to external factors that are beyond our control, including our industry's increasingly highly competitive market for skilled workers and leaders, and workforce participation rates. Changes in immigration and work permit regulations, or in their administration or interpretation, could impair our ability to attract, employ and retain qualified employees. Competition for talent drives up costs in the form of cash and stock-based compensation. In times of stock price volatility, as we have experienced in the past and may experience in the future, the retentive value of our stock-based compensation may decrease. Additionally, we are highly dependent on the services of our longstanding executive team. Failure to ensure effective succession planning, transfer of knowledge, and smooth transitions involving executives and key employees could hinder our strategic planning, execution, and long-term success.\nOur business is dependent upon the proper functioning of our business processes and information systems and modification or interruption of such systems may disrupt our business and internal controls.\nWe rely upon internal processes and information systems to support key business functions, including our assessment of internal controls over financial reporting as required by Section 404 of the Sarbanes-Oxley Act. The efficient operation and scalability of these processes and systems is critical to support our growth. We continue to design and implement updated accounting functionality related to a new enterprise resource planning, or ERP, system. Any ERP system implementation may introduce problems, such as quality issues or programming errors, that could have an impact on our continued ability to successfully operate our business or to timely and accurately report our financial results. These changes may be costly and disruptive to our operations and could impose substantial demands on management time. Failure to implement new or updated controls, or difficulties encountered in their implementation, could harm our operating results or cause us to fail to meet our reporting obligations.\nIdentification of material weaknesses in our internal controls, even if quickly remediated once disclosed, may cause investors to lose confidence in our financial statements and our stock price may decline. Remediation of any material weakness could require us to incur significant expenses, and if we fail to remediate any material weakness, our financial statements may be inaccurate, we may be required to restate our financial statements, our ability to report our financial results on a timely and accurate basis may be adversely affected, our access to the capital markets may be restricted, our stock price may decline, and we may be subject to sanctions or investigation by regulatory authorities.\nOur operating results have in the past fluctuated and may in the future fluctuate, and if our operating results are below the expectations of securities analysts or investors, our stock price could decline.\nOur operating results have in the past fluctuated and may continue to fluctuate due to a number of factors. Therefore, investors should not rely on our past results of operations as an indication of our future performance. Factors that could affect our results of operations include, but are not limited to:\n• our ability to adjust spending due to the multi-year development cycle for some of our products and services;\n• our ability to comply with our contractual obligations to customers;\n- our extended payment term arrangements with certain customers, the inability of some customers to make required payments, our ability to obtain credit insurance for customers with extended payment terms, and customer bad debt write-offs;\n• our vendors' payment requirements;\n• unanticipated costs associated with environmental liabilities; and\n• changes in financial accounting standards or interpretations of existing standards.\nAny of these factors could prevent us from achieving our anticipated financial results. For example, we have granted and may continue to grant extended payment terms to some customers, particularly during macroeconomic downturns, which could impact our ability to collect payment. Our vendors have requested and may continue to ask for shorter payment terms, which may impact our cash flow generation. These arrangements reduce the cash we have available for general business operations. In addition, the pace of growth in our operating expenses and investments may lag our revenue growth, creating volatility or periods where profitability levels may not be sustainable. Failure to meet our expectations or the expectations of our investors or security analysts is likely to cause our stock price to decline, as it has in the past, or substantial price volatility.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors/Risks Related to Our Global Operating Business", + "metadata": { + "length": 35859, + "summary": "The passage outlines various risks that could harm NVIDIA's business, including adverse economic conditions, international operations, cybersecurity threats, business disruptions, climate change impacts, acquisition challenges, customer concentration, counterparty risks, talent retention issues, system failures, and fluctuating operating results. These factors could lead to increased costs, reduced revenue, legal liabilities, reputational damage, and stock price volatility.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Adverse", + "economic", + "conditions", + "harm", + "business", + "Economic", + "industry", + "uncertainty", + "including", + "recession", + "slowing", + "growth", + "inflation", + "fiscal", + "monetary", + "trade", + "policy", + "disruptions", + "capital", + "markets", + "banking", + "system", + "currency", + "fluctuations", + "higher", + "interest", + "rates", + "tighter", + "credit", + "lower", + "expenditures", + "businesses", + "IT", + "infrastructure", + "increases", + "unemployment", + "labor", + "shortages", + "consumer", + "confidence", + "spending", + "global", + "supply", + "chain", + "constraints", + "geopolitical", + "developments", + "implementation", + "tariffs", + "USG", + "governments", + "past", + "future", + "adverse", + "wide", + "ranging", + "effects", + "financial", + "results", + "increased", + "costs", + "wafers", + "components", + "logistics", + "expenses", + "negatively", + "impacted", + "gross", + "margin", + "employee", + "facilities", + "volatility", + "reduced", + "reduce", + "margins", + "decrease", + "demand", + "products", + "services", + "technologies", + "customers", + "partners", + "licensees", + "inability", + "suppliers", + "deliver", + "commitments", + "end", + "users", + "limits", + "ability", + "forecast", + "operating", + "make", + "decisions", + "insolvency", + "key", + "distributors", + "CSPs", + "data", + "center", + "providers", + "licensing", + "parties", + "rely", + "profitability", + "scale", + "back", + "operations", + "exit", + "file", + "bankruptcy", + "protection", + "potentially", + "cease", + "lead", + "mergers", + "consolidations", + "strategic", + "alliances", + "companies", + "adversely", + "affect", + "compete", + "effectively", + "collectability", + "risks", + "borrowing", + "availability", + "liquidity", + "impacts", + "failures", + "counterparties", + "institutions", + "insurers", + "asset", + "impairments", + "declines", + "instruments", + "affecting", + "bank", + "instability", + "concerns", + "speculation", + "similar", + "events", + "market", + "problems", + "impact", + "fulfill", + "payment", + "obligations", + "vendors", + "contractual", + "Additionally", + "maintain", + "portfolio", + "liquid", + "investments", + "cash", + "management", + "purposes", + "holdings", + "types", + "maturities", + "These", + "subject", + "general", + "rate", + "exacerbated", + "downturns", + "majority", + "investment", + "comprises", + "securities", + "decline", + "long", + "periods", + "downgrade", + "rating", + "due", + "actual", + "threatened", + "default", + "government", + "debt", + "result", + "dollar", + "factors", + "unrealized", + "realized", + "loss", + "position", + "require", + "record", + "impairment", + "charges", + "International", + "sales", + "significant", + "part", + "exposes", + "We", + "sell", + "internationally", + "conduct", + "Our", + "semiconductor", + "manufactured", + "assembled", + "tested", + "packaged", + "located", + "United", + "States", + "generated", + "31%", + "revenue", + "year", + "2026", + "The", + "China", + "offerings", + "limited", + "export", + "controls", + "highly", + "competitive", + "expect", + "remain", + "forward", + "nature", + "subjects", + "number", + "uncertainties", + "material", + "effect", + "condition", + "include", + "domestic", + "international", + "political", + "countries", + "manufacturers", + "lockdowns", + "control", + "case", + "spread", + "local", + "health", + "issues", + "differing", + "legal", + "standards", + "respect", + "IP", + "employment", + "practices", + "cultural", + "counter", + "policies", + "natural", + "disasters", + "acts", + "war", + "military", + "actions", + "terrorism", + "public", + "restrictions", + "sanctions", + "imports", + "exports", + "catastrophic", + "Product", + "security", + "incidents", + "breaches", + "cyber", + "attacks", + "disrupt", + "expected", + "increase", + "significantly", + "reputation", + "Security", + "computer", + "malware", + "social", + "engineering", + "denial", + "service", + "software", + "bugs", + "server", + "malfunctions", + "hardware", + "information", + "technology", + "assets", + "increasingly", + "sophisticated", + "making", + "difficult", + "successfully", + "detect", + "defend", + "implement", + "adequate", + "preventative", + "measures", + "Cyber", + "ransomware", + "organized", + "criminal", + "threat", + "actors", + "nation", + "states", + "state", + "supported", + "prevalent", + "severe", + "recover", + "backups", + "affected", + "attack", + "restore", + "delayed", + "feasible", + "Individuals", + "groups", + "hackers", + "organizations", + "engaged", + "continue", + "engage", + "AI", + "launch", + "automated", + "targeted", + "coordinated", + "Due", + "conflicts", + "times", + "major", + "heightened", + "risk", + "materially", + "provide", + "face", + "cybersecurity", + "threats", + "error", + "intentional", + "misconduct", + "employees", + "contractors", + "party", + "Certain", + "aspects", + "effective", + "dependent", + "safeguarding", + "sensitive", + "adhering", + "access", + "mechanisms", + "experienced", + "experience", + "arising", + "failure", + "properly", + "handle", + "adhere", + "posting", + "company", + "websites", + "permission", + "assurance", + "insider", + "incident", + "negative", + "publicity", + "Furthermore", + "provided", + "operate", + "critical", + "systems", + "limitation", + "cloud", + "based", + "encryption", + "authentication", + "email", + "functions", + "guarantee", + "chains", + "compromised", + "exploitable", + "vulnerabilities", + "defects", + "breach", + "disruption", + "support", + "incorporated", + "models", + "open", + "source", + "datasets", + "train", + "flawed", + "insufficient", + "biased", + "resilience", + "compromise", + "integrity", + "outputs", + "leading", + "potential", + "reputational", + "damage", + "regulatory", + "scrutiny", + "performance", + "reliability", + "turn", + "customer", + "trust", + "insight", + "privacy", + "algorithms", + "monitor", + "place", + "In", + "addition", + "suffers", + "happened", + "happen", + "response", + "direct", + "logs", + "related", + "large", + "entities", + "worldwide", + "liability", + "To", + "continuously", + "engineer", + "secure", + "enhance", + "features", + "develop", + "training", + "programs", + "awareness", + "initiatives", + "designed", + "ensure", + "meet", + "evolving", + "requirements", + "applicable", + "regulations", + "While", + "invest", + "steps", + "remediate", + "identified", + "prevent", + "mitigate", + "installed", + "techniques", + "change", + "frequently", + "detected", + "occurred", + "Further", + "delays", + "developing", + "deploying", + "remedial", + "address", + "exploited", + "hold", + "confidential", + "personal", + "proprietary", + "Breaches", + "reported", + "perceived", + "unapproved", + "dissemination", + "expose", + "misuse", + "resulting", + "litigation", + "subsequent", + "inquiries", + "brand", + "For", + "game", + "code", + "GFN", + "NVIDIA", + "If", + "consequences", + "enforcement", + "additional", + "reporting", + "oversight", + "processing", + "indemnification", + "diversion", + "funds", + "attention", + "produce", + "distribute", + "goods", + "harms", + "Inability", + "orders", + "lost", + "stock", + "price", + "required", + "notify", + "relevant", + "stakeholders", + "individuals", + "regulators", + "investors", + "mandatory", + "disclosure", + "experiencing", + "gather", + "collect", + "infer", + "sources", + "brokers", + "means", + "reveals", + "competitively", + "details", + "organization", + "Business", + "Factors", + "caused", + "extreme", + "weather", + "power", + "water", + "telecommunications", + "supplier", + "terrorist", + "violence", + "civil", + "unrest", + "epidemics", + "pandemics", + "abrupt", + "man", + "made", + "corporate", + "headquarters", + "portion", + "current", + "capacity", + "research", + "development", + "activities", + "California", + "finished", + "inventory", + "Asia", + "vulnerable", + "earthquakes", + "wildfires", + "occurring", + "geographical", + "areas", + "Catastrophic", + "personnel", + "Geopolitical", + "globally", + "Political", + "Worldwide", + "tensions", + "Hong", + "Kong", + "Israel", + "Korea", + "Taiwan", + "manufacture", + "product", + "final", + "assembly", + "concentrated", + "changing", + "strategies", + "hiring", + "restricted", + "restrict", + "State", + "manufacturing", + "Israeli", + "boycotts", + "harmed", + "disrupted", + "reason", + "high", + "heat", + "cyberattacks", + "environmental", + "ultimate", + "foundries", + "consolidated", + "unknown", + "event", + "disaster", + "affects", + "substantial", + "time", + "spent", + "fully", + "resume", + "continuity", + "recovery", + "planning", + "sufficient", + "eventualities", + "All", + "monitoring", + "conflict", + "safety", + "approximately", + "000", + "region", + "primarily", + "marketing", + "networking", + "active", + "duty", + "extended", + "period", + "expense", + "expanded", + "create", + "interruptions", + "location", + "impair", + "As", + "reliance", + "exposure", + "outages", + "quality", + "Data", + "centers", + "depend", + "clean", + "predictable", + "energy", + "Power", + "land", + "permitting", + "limit", + "expand", + "consume", + "Climate", + "term", + "Water", + "regions", + "Extreme", + "wind", + "coupled", + "dry", + "Northern", + "shut", + "offs", + "wildfire", + "implications", + "Santa", + "Clara", + "headquarter", + "offices", + "impairing", + "work", + "attrition", + "losses", + "Although", + "insurance", + "coverage", + "variety", + "property", + "casualty", + "amounts", + "obtain", + "vary", + "depending", + "cost", + "Some", + "deductibles", + "broad", + "exclusions", + "unable", + "unwilling", + "pay", + "claim", + "Losses", + "covered", + "sustainability", + "laws", + "lawsuits", + "New", + "proposed", + "relating", + "carbon", + "taxes", + "fuel", + "pollution", + "governance", + "greater", + "processes", + "procurement", + "raw", + "materials", + "improve", + "equipment", + "compliance", + "emissions", + "indirect", + "incurring", + "passed", + "increasing", + "requiring", + "alter", + "design", + "Stakeholder", + "find", + "insufficiently", + "responsive", + "climate", + "action", + "achieve", + "stated", + "goals", + "incur", + "unexpected", + "disputes", + "trends", + "influenced", + "absolute", + "GPUs", + "efficient", + "operation", + "relative", + "alternative", + "computing", + "platforms", + "realize", + "benefits", + "acquisitions", + "integrate", + "acquired", + "hurt", + "grow", + "acquire", + "offer", + "objectives", + "Acquisitions", + "involve", + "challenges", + "ultimately", + "pursue", + "transaction", + "enter", + "transactions", + "timely", + "complete", + "obtaining", + "approvals", + "retain", + "talent", + "target", + "evolve", + "acquisition", + "attractive", + "anticipated", + "Regulators", + "impose", + "extent", + "consummate", + "complex", + "expensive", + "generate", + "gains", + "invested", + "ecosystem", + "Many", + "return", + "private", + "early", + "stage", + "defining", + "direction", + "marketable", + "illiquid", + "initial", + "successful", + "recognize", + "lose", + "finalizing", + "partnership", + "agreement", + "OpenAI", + "There", + "completed", + "sector", + "concentration", + "multiple", + "sectors", + "resources", + "difficulty", + "realizing", + "satisfactory", + "governmental", + "approval", + "consents", + "financing", + "proceedings", + "initiated", + "diligence", + "identify", + "investing", + "acquiring", + "Additional", + "integrating", + "retaining", + "assumption", + "liabilities", + "amortization", + "goodwill", + "write", + "downs", + "accounting", + "forecasting", + "procedures", + "cycles", + "coordinating", + "fines", + "fees", + "close", + "issuances", + "finance", + "covenants", + "dilutive", + "equity", + "variability", + "amount", + "form", + "consideration", + "industries", + "relationships", + "lengthy", + "costly", + "integration", + "purchasing", + "shipping", + "difficulties", + "electronic", + "interchange", + "receive", + "purchase", + "directly", + "indirectly", + "prevented", + "selling", + "trend", + "represented", + "22%", + "total", + "14%", + "attributable", + "Compute", + "Networking", + "segment", + "With", + "channels", + "partner", + "network", + "incorporate", + "small", + "involved", + "integrators", + "specifications", + "Changes", + "ownership", + "advanced", + "decide", + "fewer", + "patterns", + "Because", + "order", + "basis", + "generally", + "cancel", + "delay", + "notice", + "penalty", + "solutions", + "competitors", + "discontinue", + "share", + "OEMs", + "ODMs", + "channel", + "estimate", + "individually", + "representing", + "10%", + "quarter", + "timing", + "reduction", + "purchases", + "collecting", + "accounts", + "receivable", + "Commercial", + "arrangements", + "counterparty", + "entered", + "commercial", + "guarantees", + "asked", + "buildout", + "datacenter", + "occurrence", + "project", + "distress", + "Financing", + "undertaken", + "circumstances", + "upfront", + "flows", + "terms", + "multi", + "attract", + "motivate", + "executives", + "execute", + "strategy", + "recruit", + "exceptional", + "However", + "external", + "skilled", + "workers", + "leaders", + "workforce", + "participation", + "immigration", + "permit", + "administration", + "interpretation", + "employ", + "qualified", + "Competition", + "drives", + "compensation", + "retentive", + "longstanding", + "executive", + "team", + "Failure", + "succession", + "transfer", + "knowledge", + "smooth", + "transitions", + "involving", + "hinder", + "execution", + "success", + "proper", + "functioning", + "modification", + "interruption", + "internal", + "assessment", + "Section", + "404", + "Sarbanes", + "Oxley", + "Act", + "scalability", + "updated", + "functionality", + "enterprise", + "resource", + "ERP", + "Any", + "introduce", + "programming", + "errors", + "continued", + "accurately", + "report", + "disruptive", + "demands", + "encountered", + "fail", + "Identification", + "weaknesses", + "quickly", + "remediated", + "disclosed", + "statements", + "Remediation", + "weakness", + "inaccurate", + "restate", + "accurate", + "investigation", + "authorities", + "fluctuated", + "fluctuate", + "expectations", + "analysts", + "Therefore", + "indication", + "adjust", + "cycle", + "comply", + "payments", + "bad", + "unanticipated", + "interpretations", + "existing", + "achieving", + "granted", + "grant", + "macroeconomic", + "requested", + "shorter", + "flow", + "generation", + "pace", + "lag", + "creating", + "levels", + "sustainable" + ], + "keywords": [ + "Business Risks", + "Financial Risks", + "Operational Risks" + ], + "connect_to": [] + } + }, + { + "chunk_id": "56b72587-e8b8-5dae-b090-4c31057856a3", + "type": "text", + "content": "We are subject to complex laws, rules, regulations, and political and other actions, including restrictions on the export of our products, which may adversely impact our business.\nWe are subject to laws and regulations domestically and worldwide, affecting our operations in areas including, but not limited to, IP ownership and infringement; taxes; import and export requirements and tariffs; anti-corruption, including the Foreign Corrupt Practices Act; business acquisitions; foreign exchange controls and cash repatriation restrictions; foreign ownership and investment; data privacy requirements; competition and antitrust; advertising; employment; product regulations; cybersecurity; environmental, health, and safety requirements; the responsible use of AI; sustainability; cryptocurrency; and consumer laws. Compliance with such requirements can be onerous and expensive, could impact our competitive position, and may negatively impact our business operations and ability to manufacture and ship our products. There can be no assurance that our employees, contractors, suppliers, customers or agents will not violate applicable laws or the policies, controls, and procedures that we have designed to help ensure compliance with such laws, and violations could result in fines, criminal sanctions against us, our officers, or our employees, prohibitions on the conduct of our business, and damage to our reputation. Changes to the laws, rules and regulations to which we are subject, or changes to their interpretation and enforcement, could lead to materially greater compliance and other costs, and/or further restrictions on our ability to manufacture and supply our products and operate our business. For example, we may face increased compliance costs as a result of changes or increases in antitrust legislation, regulation, administrative rule making, increased focus from regulators on cybersecurity vulnerabilities and risks. Our position in markets relating to AI has led to increased interest in our business from regulators worldwide, including the European Union, the United States, the United Kingdom, South Korea, Japan, and China. For example, the French Competition Authority collected information from us regarding our business and competition in the graphics card and CSP market as part of an ongoing inquiry into competition in those markets. We have also received, and continue to receive, broad requests for information from competition regulators in the European Union, the United States, the United Kingdom, China, and South Korea regarding our sales of GPUs and other NVIDIA products, our efforts to allocate supply, foundation models and our investments, partnerships and other agreements with companies developing foundation models, the markets in which we compete and our competition, our strategies, roadmaps, and efforts to develop, market, and sell hardware, software, and system solutions, and our agreements with customers, suppliers, and partners. We expect to receive additional requests for information in the future. Such requests have been and are likely to be expensive and burdensome and could negatively impact our business and our relationships with customers, suppliers, and partners.\nGovernments and regulators are also considering, and in certain cases, have imposed restrictions on the hardware, software, and systems used to develop frontier foundation models and generative AI. For example, the EU AI Act became effective on August 1, 2024 and will be fully applicable after a two-year transitional period. The EU AI Act may impact our ability to train, deploy, or release AI models in the EU. Several states are considering enacting or have already enacted regulations concerning AI technologies, with new state laws that took effect on January 1, 2026, which may impact our ability to train, deploy, or release AI models, and increase our compliance costs. Restrictions under these and any other regulations, if implemented, could increase the costs and burdens to us and our customers, delay or halt deployment of new systems using our products, and reduce the number of new entrants and customers, negatively impacting our business and financial results. Revisions to laws or regulations or their interpretation and enforcement could also result in increased taxation, trade sanctions, the imposition of or increase to import duties or tariffs, restrictions and controls on imports or exports, or other retaliatory actions, which could have an adverse effect on our business plans or impact the timing of our shipments. Additionally, changes in the public perception of governments in the regions where we operate or plan to operate could negatively impact our business and results of operations.\nGovernment actions, including trade protection and national and economic security policies of U.S. and foreign government bodies, such as tariffs, import or export regulations, including deemed export restrictions and restrictions on the activities of U.S. persons, trade and economic sanctions, decrees, quotas or other trade barriers and restrictions could affect our ability to ship products, provide services to our customers and employees, do business without an export license with entities on the U.S. Department of Commerce's U.S. Entity List or other USG restricted parties lists (which is expected to change from time to time), and generally fulfill our contractual obligations and have a material adverse effect on our business. If we were ever found to have violated export control laws or sanctions of the U.S. or similar applicable non-U.S. laws, even if the violation occurred without our knowledge, we may be subject to various penalties available under the laws, any of which could have a material and adverse impact on our business, operating results and financial condition.\nFor example, in response to the war in Ukraine, the United States and other jurisdictions imposed economic sanctions and export control measures which blocked the passage of our products, services and support into Russia, Belarus, and certain regions of Ukraine. In fiscal year 2023, we stopped direct sales to Russia and closed business operations in Russia. Concurrently, the war in Ukraine has impacted sales in EMEA and may continue to do so in the future.\nThe increasing focus on the risks and strategic importance of AI technologies has resulted in regulatory restrictions that target products and services capable of enabling or facilitating AI and may in the future result in additional restrictions impacting some or all of our product and service offerings.\nConcerns regarding third-party use of AI for purposes contrary to local governmental interests, including concerns relating to the misuse of AI applications, models, and solutions, has resulted in and could in the future result in unilateral or multilateral restrictions on products that can be used for training, modifying, tuning, and deploying LLMs and other AI applications. Such restrictions have limited and could in the future limit the ability of downstream customers and users worldwide to acquire, deploy and use systems that include our products, software, and services, and negatively impact our business and financial results.\nSuch restrictions could include additional unilateral or multilateral export controls on certain products or technology, including but not limited to AI technologies. As geopolitical tensions have increased, semiconductors associated with AI, including GPUs and related products, are increasingly the focus of export control restrictions proposed by stakeholders in the U.S. and its allies. The United States has imposed unilateral worldwide controls restricting GPUs and associated products, and it is likely that additional unilateral or multilateral controls will be adopted. Such controls have been and may again be very broad in scope and application, prohibit us from exporting our products to any or all customers in one or more markets, and could negatively impact our manufacturing, testing and warehousing locations and options, or could impose other conditions that limit our ability to serve demand abroad and could negatively and materially impact our business, revenue and financial results. Export controls and other restrictions targeting GPUs and semiconductors associated with AI, which have been imposed and are likely to be more restrictive, would further limit our ability to export our technology, products, or services, creating a competitive disadvantage for us and negatively impacting our business and financial results. Export controls targeting GPUs and semiconductors associated with AI have subjected and may in the future subject downstream users of our products to restrictions on the use, resale, repair, or transfer of our products, negatively impacting our business and financial results. Controls could negatively impact our cost and/or ability to provide services such as NVIDIA AI cloud services and could impact the cost and/or ability for our CSPs and customers to provide services to their end customers, even outside China.\nExport controls have and could in the future disrupt our supply chain and distribution channels, negatively impacting our ability to serve demand, including in markets outside China and for our non-data center products. The possibility of additional export controls has negatively impacted and may in the future negatively impact demand for our products, benefiting competitors that offer alternatives less likely to be restricted by further controls. Repeated changes in the export control rules are likely to impose compliance burdens on our business and our customers, negatively and materially impacting our business.\nIncreasing use of economic sanctions and export controls has impacted and may in the future impact demand for our products or services, negatively impacting our business and financial results. Reduced demand due to export controls has and could in the future lead to excess inventory or cause us to incur related supply charges. Additional unilateral or multilateral controls are also likely to include deemed export control limitations that negatively impact the ability of our research and development teams to execute our roadmap or other objectives in a timely manner. Additional export restrictions may not only impact our ability to serve overseas markets, but also provoke responses from foreign governments, including China, that negatively impact our supply chain or our ability to provide our products and services to customers in all markets worldwide, which could also substantially reduce our revenue. Regulators in China have inquired about our sales and efforts to supply the China market and our fulfillment of the commitments we entered at the close of our Mellanox acquisition. On September 15, 2025, China's antitrust regulators published their preliminary finding that our compliance with applicable U.S. export controls, which required us to offer degraded products to the Chinese market, discriminated unfairly against customers in the China market and therefore violated the terms of China's approval of our Mellanox acquisition. If regulators conclude that we have failed to fulfill the terms of our Mellanox acquisition or we have violated any applicable law in China, we could be subject to financial penalties, restrictions on our ability to conduct our business, restrictions or other orders regarding our networking business, products, and services, or otherwise impact our operations in China, any of which could have a material and adverse impact on our business, operating results and financial condition.\nOver the past three years, we have been subject to a series of shifting and expanding export control restrictions, impacting our ability to serve customers outside the United States.\nIn August 2022, the USG announced export restrictions and export licensing requirements targeting China's semiconductor and supercomputing industries. These restrictions impacted exports of certain chips, as well as software, hardware, equipment and technology used to develop, produce and manufacture certain chips to China (including Hong Kong and Macau) and Russia, and specifically impact our A100 and H100 integrated circuits, DGX or any other systems or boards which incorporate A100 or H100 integrated circuits.\nIn July 2023, the USG also informed us of an additional licensing requirement for a subset of A100 and H100 products destined to certain customers and other regions, including some countries in the Middle East.\nIn October 2023, the USG announced new and updated licensing requirements for exports to China and Country Groups D:1, D:4, and D:5 (including but not limited to, Saudi Arabia, the United Arab Emirates, and Vietnam, but excluding Israel) of our products exceeding certain performance thresholds, including, but not limited to, the A100, A800, H100, H800, L4, L40, L40S RTX 4090, GB200 NVL72, and B200. The licensing requirements also apply to the export of products exceeding certain performance thresholds to a party headquartered in, or with an ultimate parent headquartered in, Country Group D5, including China.\nIn April 2025, the USG informed us that it requires a license for export to China (including Hong Kong and Macau) and D:5 countries, or to companies headquartered or with an ultimate parent therein, of our H2O integrated circuits and any other circuits achieving the H2O's memory bandwidth, interconnect bandwidth, or combination thereof. As a result of these requirements, we incurred a \\$4.5 billion charge in the first quarter of fiscal year 2026 associated with H2O for excess inventory and purchase obligations, as the demand for H2O products diminished.\nIn August 2025, the USG granted licenses that would allow us to ship certain H2O products to certain China-based customers. We generated approximately \\$60 million in H2O revenue under those licenses. USG officials expressed an expectation that the USG will receive 15% or more of the revenue generated from licensed sales of our products, but the USG did not publish a regulation codifying such requirement.\nIn February 2026, the USG granted a license that would allow us to ship small amounts of H200 products to specific China-based customers. To date, we have not generated any revenue under the H200 licensing program, and do not yet know whether any imports will be allowed into China. The license requires that the H200s go through an inspection process in the United States prior to any shipment to the customer. As a result, any H200 shipped under the new licensing program will be subject to a 25% tariff upon importation into the United States.\nIn the event that we are able to sell licensed products into the China market, we may not be able to pass along all or any of the tariff to our customers, and may be subject to litigation, increased costs, and a harmed competitive position.\nThe export controls applicable to China are complex and address a variety of parameters, including the total processing performance of a chip, the “performance density” of a chip, the interconnect bandwidth of a chip, and the memory bandwidth of a chip. Under the current rules and geopolitical landscape, we are unable to create and deliver a competitive product for China’s data center market that receives approval from both the USG and the Chinese government. As of the end of fiscal year 2026, we were effectively foreclosed from competing in China’s data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide. Unless we are able to return with a product that meets the approval of both the USG and the Chinese government, our lost opportunity and the benefit to our competitors will have a material and adverse impact on our business, operating results, and financial condition.\nIn addition to controls targeting D:1, D:4 and D:5 countries, the USG has also imposed worldwide export controls impacting our products, and may impose additional controls in the future.\nIn January 2025, the USG published the AI Diffusion IFR in the Federal Register. The IFR would have imposed a worldwide licensing requirement on our data center products, such as our H200, GB200 and GB300. The AI Diffusion IFR would have divided the world into three tiers, relegating most countries to “Tier 2” status, and would have created a complex and burdensome scheme for licensing approvals.\nIn May 2025, the USG announced that it would rescind the AI Diffusion IFR and implement a replacement rule. The scope, timing, and requirements of the forthcoming rule remain uncertain. The replacement rule may impose new restrictions on our products or operations and/or add license requirements that could have a material impact on our business, operating results, and financial condition. For example, in October 2025, the Senate passed the “GAIN AI Act” in the NDAA. The GAIN AI Act would restrict the Trump Administration’s ability to adapt the Biden Administration’s export control rules, and could also allow private U.S. persons to review and overturn licensing and foreign policy decisions made by the Trump Administration.\nOur competitive position has been harmed by export controls, and our competitive position and future results will be further harmed, over the long term, if the restrictions remain in place or are expanded in geographic, customer, or product scope, if customers purchase product from competitors, if customers develop their own internal solution, if we are unable to provide contractual warranty or other extended service obligations, if the USG does not grant licenses in a timely manner or denies licenses to significant customers or if we incur significant transition costs. The licensing process may not be resolved before significant business opportunities evaporate. Even if the USG grants any requested licenses, the licenses have already and may in the future be temporary, impose burdensome conditions regarding the installation, maintenance, and use of such products, or include financial or economic requirements that we or our customers or end users cannot or choose not to fulfill. The licensing requirements have already and may in the future benefit certain of our competitors, as the licensing process will make our pre-sale and post-sale technical support efforts more cumbersome and less certain and encourage customers in China, the Middle East, and other regions to pursue alternatives to our products, including semiconductor suppliers based in China, Europe, and Israel.\nGiven the increasing strategic importance of AI and rising geopolitical tensions, the USG has changed and may again change the export control rules at any time and further subject a wider range of our products to export restrictions and licensing requirements, negatively impacting our business and financial results. In the event of such change, we may be unable to sell our inventory of such products and may be unable to develop replacement products not subject to the licensing requirements.\nFor example, the USG already imposed license conditions that limit the ability of foreign firms to create and offer as a service large-scale GPU clusters, such as imposing license conditions on the use of products to be exported to certain countries, and may impose additional conditions such as requiring chip tracking and throttling mechanisms that could disable or impair GPUs if certain events, including unauthorized system configuration, use, or location, are detected. Such government mandates in chip designs could introduce system vulnerabilities and expose us to significant risk and potential liability, negatively impact demand for our products, and could have a material impact on our business, operating results, and financial condition. Even if not enacted into binding legislation, draft bills have impacted and may in the future negatively impact our business. For example, following U.S. legislative proposals calling for mandatory features in our chips, China's government publicly questioned whether our H2O products have built-in vulnerabilities, discouraging customers from purchasing our products. We provided a public response explaining that our GPUs, including H2O, do not include such built-in vulnerabilities, and will respond to any follow-up questions we receive.\nOpen-source foundation models are rapidly growing in popularity with developers worldwide. Any regulatory control or other restriction that limits our ability to provide products and services that support third-party applications and models, including applications built on foundation models originating in China such as DeepSeek, Qwen, or KIMMI, could have a material impact on our business, operating results, and financial condition.\nThe USG already imposed export controls restricting certain gaming GPUs, and if the USG expands such controls to restrict additional gaming products, it may disrupt a significant portion of our supply and distribution chain and negatively impact sales of such products to markets outside China, including the U.S. and Europe. In addition, as the performance of the gaming GPUs increases over time, export controls may have a greater impact on our ability to compete in markets subject to those controls. Export controls may disrupt our supply and distribution chain for a substantial portion of our products, which are warehoused in and distributed from Hong Kong.\nExport controls restricting our ability to sell data center GPUs may also negatively impact demand for our networking products used in servers containing our GPUs. The USG may also impose export controls on our networking products, such as high-speed network interconnects, to limit the ability of downstream parties to create large clusters for frontier model training.\nExport controls have and are likely in the future to have a disproportionate impact on NVIDIA and may disadvantage us against certain of our competitors that sell chips that are outside the scope of such control. Export controls have already and may in the future encourage customers outside China and other impacted regions to “design-out” certain U.S. semiconductors from their products to reduce the compliance burden and risk, and to ensure that they are able to serve markets worldwide. Export controls have already encouraged and may in the future encourage overseas governments to request that our customers purchase from our competitors rather than NVIDIA or other U.S. firms, harming our business, market position, and financial results.\nAs a result, export controls have in the past and may in the future negatively impact demand for our products and services not only in China, but also in other markets, such as Europe, Latin America, and Southeast Asia. Export controls increase the risk of investing in U.S. advanced semiconductor products, because by the time a new product is ready for market, it may be subject to new unilateral export controls restricting its sale, resulting in excess inventory and purchase obligations as we recently experienced with the H2O. At the same time, such controls may increase investment in foreign competitors, which would be less likely to be restricted by U.S. controls.\nThe increasingly complex export controls impose complex and burdensome compliance obligations on our partners, suppliers, and customers. While we seek to strictly comply with all applicable export control regulators, reports of diversion of controlled products, even when unsubstantiated and untrue, may negatively impact our business, relationships with partners and customers, and our reputation. Incorrect allegations that our compliance efforts satisfy the letter but not the “spirit” of the applicable regulations, as well as incorrect allegations that legitimate and appropriate business is using supposed “loopholes” in the export controls may negatively impact our business, relationships with partners and customers, and our reputation.\nIn addition to export controls, the USG may impose restrictions on the import and sale of products that incorporate technologies developed or manufactured in whole or in part in China. For example, the USG adopted “Connected Vehicle” restrictions on the import and sale of certain automotive products in the United States, which if adopted and interpreted broadly, could impact our ability to develop and supply solutions for our automotive customers. The USG is also considering restrictions that would limit our ability to support third-party applications and models built on open-source foundation models originating in China. Such restrictions, if implemented, would favor our foreign competitors and negatively impact our business.\nAdditionally, restrictions imposed by the Chinese government on the duration of gaming activities and access to games may adversely affect our Gaming revenue, and even if we are able to participate in the China data center compute market, increased oversight of digital platform companies may adversely affect our Data Center revenue. The Chinese government has encouraged customers to purchase from our China-based competitors and discouraged customers from purchasing, importing, or using our data center products, including any China-specific product designed to comply with U.S. export controls. As another example, an agency of the Chinese government announced an Action Plan that endorses new standards regarding the compute performance per watt and per memory bandwidth of accelerators used in new and renovated data centers in China. Although we are already effectively foreclosed from the China market by U.S. export controls, if those controls changed to allow us to return to the market, the Chinese government could modify or implement the Action Plan in a way that effectively prevents us from being able to design products to meet the new standard, which may restrict the ability of customers to use some of our data center products and may have a material and adverse impact on our business, operating results and financial condition. Further restrictions on our products or the products of our suppliers could negatively impact our business and financial results.\nFinally, our business depends on our ability to receive consistent and reliable supply from our overseas partners, especially in Taiwan and South Korea. Any new restrictions that negatively impact our ability to receive supply of components, parts, or services from Taiwan and South Korea, would negatively impact our business and financial results.\nScrutiny from shareholders, regulators and others regarding our corporate sustainability practices could result in additional costs or risks and adversely impact our reputation and willingness of customers and suppliers to do business with us.\nCertain shareholder advocacy groups, investment funds, shareholders and other market participants, customers and government regulators have focused on corporate sustainability practices and disclosures, including those associated with climate change and human rights. Stakeholders may not be satisfied with our corporate sustainability practices and goals or the speed of their adoption. Further, there are state-level initiatives in the U.S. that may differ from other regulatory requirements or our various stakeholders' expectations. Additionally, our corporate sustainability practices, oversight of our practices or disclosure controls may not meet evolving shareholder, regulator or other industry stakeholder expectations, or we may fail to meet corporate sustainability disclosure or reporting standards or legal requirements. We could also incur additional costs and require additional resources to monitor, report, and comply with various corporate sustainability practices and legal requirements, choose not to conduct business with potential customers and suppliers, or discontinue or not expand business with existing customers and suppliers due to our policies. These factors and increased disclosure may negatively harm our brand, reputation and business activities or expose us to liability.\nIssues relating to the responsible use of our technologies, including AI in our offerings, may result in reputational or financial harm and liability.\nConcerns relating to the responsible use of new and evolving technologies, such as AI, in our products and services may result in reputational or financial harm and liability and may cause us to incur costs to resolve such issues. We are increasingly building AI capabilities and protections into many of our products and services, and we also offer stand-alone AI applications. AI poses emerging legal, social, and ethical issues and presents risks and challenges that could affect its adoption, and therefore our business. If we enable or offer solutions that draw controversy due to their perceived or actual impact on society, such as AI solutions that have unintended consequences, infringe copyright or rights of publicity, or are controversial because of their impact on human rights, privacy, employment or other social, economic or political issues, or if we are unable to implement effective internal policies and frameworks relating to the responsible development and use of AI models and systems offered through our sales channels, we may experience brand or reputational harm, competitive harm or legal liability. Leveraging AI capabilities to potentially improve our internal functions and operations may present further risks, costs, and challenges. Complying with multiple regulations from different jurisdictions related to AI may further increase our cost of doing business, may change the way that we operate in certain jurisdictions, and may impede our ability to offer certain products and services in certain jurisdictions if we are unable to comply with regulations. Compliance with existing and proposed government regulation of AI, including in jurisdictions such as the European Union, may further increase the cost of related research and development, and create additional reporting and/or transparency requirements. For example, regulation adopted in response to the European Union Code of Practice for General Purpose Artificial Intelligence could require us to notify the European Commission regarding details of some of our Trustworthy AI processes related to our risk framework. Furthermore, changes in AI-related regulation could disproportionately impact and disadvantage us and require us to change our business practices, which may negatively impact our financial results. Our failure to adequately address concerns and regulations relating to the responsible use of AI by us or others could undermine public confidence in AI and slow adoption of AI in our products and services or cause reputational or financial harm.\nActions to adequately protect our IP rights could result in substantial costs to us and our ability to compete could be harmed if we are unsuccessful or if we are prohibited from making or selling our products.\nFrom time to time, we are involved in lawsuits or other legal proceedings alleging patent infringement or other IP rights violations by us, our employees or parties that we have agreed to indemnify. An unfavorable ruling could include significant damages, invalidation of one or more patents, indemnification of third parties, payment of lost profits, or injunctive relief. Claims that our products or processes infringe the IP rights of others, regardless of their merit, could cause us to incur significant costs to respond to, defend, and resolve such claims, and they may also divert the efforts and attention of management and technical personnel.\nWe may commence legal proceedings to protect our IP rights, which may increase our operating expenses. We could be subject to countersuits as a result. If infringement claims are made against us or our products are found to infringe a third party's IP, we or one of our indemnitees may have to seek a license to the third party's IP rights. If we or one of our indemnitees is unable to obtain such a license on acceptable terms or at all, we could be subject to substantial liabilities or have to suspend or discontinue the manufacture and sale of one or more of our products. We may also have to make royalty or other payments or cross license our technology. If these arrangements are not concluded on commercially reasonable terms, our business could be negatively impacted. Furthermore, the indemnification of a customer or other indemnitee may increase our operating expenses and negatively impact our operating results.\nWe rely on patents, trademarks, trade secrets, employee and third-party nondisclosure agreements, licensing arrangements and the laws of the countries in which we operate to protect our IP. Foreign laws may not protect our products or IP rights to the same extent as United States law. This makes the possibility of piracy of our technology and products more likely. The theft or unauthorized use or publication of our trade secrets and other confidential information could harm our competitive position and reduce acceptance of our products; as a result, the value of our investment in research and development, product development and marketing could be reduced. We also may face risks to our IP if our employees are hired by competitors. We continuously assess whether and where to seek formal protection for existing and new innovations and technologies but cannot be certain whether our applications for such protections will be approved, and, if approved, whether they will be enforceable.\nWe are subject to stringent and changing data privacy and security laws, rules, regulations and other obligations. These areas could damage our reputation, deter current and potential customers, affect our product design, or result in legal or regulatory proceedings and liability.\nWe process sensitive, confidential or personal data or information that is subject to privacy and security laws, regulations, industry standards, external and internal policies, contracts and other obligations that govern the processing of such data by us and on our behalf. Concerns about our practices or the ultimate use of our products and services with regard to the collection, use, retention, security or disclosure of personal information or other privacy-related matters, including for use in AI, even if unfounded, could damage our reputation and adversely affect our operating results. The theft, loss or misuse of personal data in our possession or by one of our partners could result in damage to our reputation, regulatory proceedings, disruption of our business activities or increased security or remediation costs and costs related to defending legal claims.\nIn the United States, federal, state and local authorities have enacted numerous data privacy and security laws, including for data breach notification, personal data privacy and consumer protection. Numerous U.S. states have enacted comprehensive privacy laws that impose certain obligations on covered businesses, including providing specific disclosures in privacy notices and affording residents with certain rights concerning their personal data. As applicable, such rights may include the right to access, correct, or delete certain personal data, and to opt-out of certain data processing activities, such as targeted advertising, profiling and automated decision-making. The exercise of these rights may impact our business and ability to provide our products and services. Certain states also impose stricter requirements for processing certain personal data, including sensitive information, such as conducting data privacy impact assessments. These state laws allow for statutory fines for noncompliance. For example, the California Consumer Privacy Act of 2018, as amended by the California Privacy Rights Act of 2020, or CPRA, or collectively the CCPA, gives California residents the right to access, delete and opt-out of certain sharing of their personal information, and to receive detailed information about how it is used and shared. The CCPA provides for substantial fines for intentional violation and the law created a private right of action for certain data breaches. Similar laws are being considered in several other states, as well as at the federal and local levels. Additionally, several states and localities have enacted measures related to the use of AI and machine learning in products and services. If we become subject to additional data privacy laws, the risk of enforcement action against us could increase.\nWorldwide regulatory authorities are also considering and have approved various legislative proposals concerning data protection. The European Union adopted the General Data Protection Regulation, or GDPR, and the United Kingdom similarly adopted the U.K. GDPR, governing the strict handling of personal data of persons within the European Economic Area, or EEA, and the United Kingdom, respectively, including its use and protection and the ability of persons whose data is stored to access, correct, and delete such data about themselves. If we are found not to comply, we could be subject to penalties of up to €20 million or 4% of worldwide revenue, whichever is greater, and classes of individuals or consumer protection organizations may initiate litigation related to our processing of their personal data. Furthermore, the EU AI Act and similar legislation could impose onerous obligations that may disproportionately impact and disadvantage us and require us to change our business practices. Additionally, Europe’s Network and Information Security Directive, or NIS2, regulates resilience and incident response capabilities of entities operating in a number of sectors, including the digital infrastructure sector. Non-compliance with NIS2 may lead to administrative fines of a maximum of 10 million Euros or up to 2% of the total worldwide revenue of the preceding fiscal year.\nIn the ordinary course of business, we transfer personal data from Europe, China, and other jurisdictions to the United States or other countries. Certain jurisdictions have enacted data localization laws and cross-border personal data transfer laws. For example, the GDPR governs the transfer of personal data to countries outside of the EEA. The European Commission released a set of “Standard Contractual Clauses” designed for entities to validly transfer personal data out of the EEA to jurisdictions that the European Commission has not found to provide an adequate level of protection, including the United States. Additionally, the U.K.’s International Data Transfer Agreement / Addendum, as well as the EU-U.S. Data Privacy Framework and the U.K. extension thereto (which allows for transfers to relevant U.S.-based organizations who self-certify compliance and participate in the Framework) are mechanisms that may be used to transfer personal data from the EEA and U.K. to the United States. However, these mechanisms are subject to legal challenges, and there is no assurance that we can satisfy or rely on these measures to lawfully transfer personal data to the United States. Other jurisdictions have enacted or are considering similar cross-border personal data transfer laws and local personal data residency laws, any of which would increase the cost and complexity of doing business and could result in fines from regulators. For example, China’s law imposes various requirements relating to data processing and data localization. Data broadly defined as important under China’s law, including personal data, may not be transferable outside of China without prior assessment and approval by the Cyberspace Administration of China, or CAC. Compliance with these requirements, including CAC assessments and any deemed failures of such assessments, could cause us to incur liability, prevent us from using data collected in China or impact our ability to transfer data outside of China. The inability to import personal data to the United States could significantly and negatively impact our business operations, limit our ability to collaborate with parties that are subject to European, China and other data privacy and security laws, or require us to increase our personal data processing capabilities in Europe and/or elsewhere at significant expense. Some European regulators have prevented companies from transferring personal data out of Europe for allegedly violating the GDPR’s cross-border data transfer limitations, which could negatively impact our business.\nWe are also bound by certain contractual obligations related to data privacy and security, and our efforts to comply with such obligations may not be successful or may be claimed to be non-compliant. For example, certain privacy laws, such as the GDPR and the CCPA, require our customers to impose specific contractual restrictions on their service providers. We sometimes host personal data in collaboration with our customers, and if a breach exposed or altered that personal data, it could harm those customer relationships and subject us to litigation, regulatory action, or fines. We publish privacy policies, marketing materials and other statements, such as compliance with certain certifications or self-regulatory principles, regarding data privacy and security. Regulators in the U.S. are increasingly scrutinizing these statements, and if these policies, materials or statements are found to be deficient, lacking in transparency, deceptive, unfair or misrepresentative of our practices, we may be subject to investigation, enforcement actions by regulators or other adverse consequences.\nData protection laws around the world are quickly changing and may be interpreted and applied in an increasingly stringent fashion and in a manner that is inconsistent with our data practices. These obligations may affect our product design and necessitate changes to our information technologies, systems and practices and to those of any third parties that process personal data on our behalf. Despite our efforts, we or third parties we rely upon may fail to comply with such obligations. If we fail, or are perceived to have failed, to address or comply with data privacy and security obligations, we could face significant consequences, including but not limited to, government enforcement actions, litigation, additional reporting requirements and/or oversight, bans on processing personal data, and orders to destroy or not use personal data. Any of these events could have a material adverse effect on our reputation, business, or financial condition.\nWe may have exposure to additional tax liabilities and our operating results may be adversely impacted by changes in tax laws, higher than expected tax rates and other tax-related factors.\nWe are subject to complex income tax laws and regulations, as well as non-income-based taxes, in various jurisdictions. Significant judgment is required in determining our worldwide provision for income taxes and other tax liabilities. We are regularly under audit by tax authorities in different jurisdictions. Although we believe our tax estimates are reasonable, any adverse outcome could increase our worldwide effective tax rate, increase the amount of non-income taxes imposed on our business, and harm our financial position, results of operations, net income, and cash flows.\nFurther, changes in tax laws or their interpretation by tax authorities in the U.S. or foreign jurisdictions could increase our future tax liability or cause other adverse tax impacts, which may materially impact our results of operations, or the way we conduct our business. Most of our income is taxable in the U.S., with a significant portion qualifying for preferential treatment as foreign-derived deduction eligible income, or FDDEI. If U.S. tax rates increase or the FDDEI deduction is reduced, our provision for income taxes, results of operations, net income, and cash flows would be adversely affected. In addition, our tax obligations and effective tax rate in the jurisdictions in which we conduct business could increase as a result of international tax developments, including the implementation of the Two-Pillar framework led by the Organization for Economic Cooperation and Development, or OECD, which involves the reallocation of taxing rights in respect of certain multinational enterprises above a fixed profit margin to the jurisdictions in which they carry on business (referred to as Pillar One), and imposes a minimum effective corporate tax rate (referred to as Pillar Two). A number of countries in which we conduct business have enacted, or are in the process of enacting, elements of the Pillar Two rules, including the recently released \"side-by-side\" framework, which provides certain additional safe harbors and coordination mechanisms. Any such tax laws, or changes in any such tax laws, including those related to the side-by-side framework, may increase tax uncertainty and compliance costs and adversely affect our provision for income taxes, cash tax payments, results of operations, and financial condition.\nOur future effective tax rate may also be affected by a variety of factors, including changes in our business or statutory rates, the mix of earnings in countries with differing statutory tax rates, available tax incentives, credits and deductions, the expiration of statutes of limitations, changes in accounting principles, adjustments to income taxes upon finalization of tax returns, increases in expenses not deductible for tax purposes, the estimates of our deferred tax assets and liabilities and deferred tax asset valuation allowances, changing interpretation of existing laws or regulations, the impact of accounting for business combinations, as well as changes in the domestic or international organization of our business and structure. Furthermore, the tax effects of accounting for stock-based compensation and volatility in our stock price may significantly impact our effective tax rate in the period in which they occur. A decline in our stock price may result in reduced future tax benefits from stock-based compensation, increase our effective tax rate, and adversely affect our financial results.\nOur business is exposed to the burden and risks associated with litigation, investigations and regulatory proceedings.\nWe currently and will likely continue to face legal, administrative and regulatory proceedings, claims, demands and/or investigations involving shareholder, consumer, competition, intellectual property and/or other issues relating to our business. For example, we are defending a securities class action lawsuit from multiple shareholders asserting claims that we and certain of our officers made false and/or misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand in 2017 and 2018. Litigation and regulatory proceedings are inherently uncertain, and adverse rulings could occur, including monetary damages or fines, or an injunction stopping us from manufacturing or selling certain products, engaging in certain business practices, or requiring other remedies, such as compulsory licensing of patents. An unfavorable outcome or settlement may result in a material adverse impact. Regardless of the outcome, litigation can be costly, time-consuming, and disruptive to our operations.\nDelaware law and our certificate of incorporation, bylaws and agreement with Microsoft could delay or prevent a change in control.\nThe anti-takeover provisions of the Delaware General Corporation Law may discourage, delay, or prevent a change in control. Provisions in our certificate of incorporation and bylaws could make it more difficult for a third party to acquire a majority of our outstanding stock. These provisions include the ability of our Board of Directors to create and issue preferred stock, change the number of directors, and to make, amend or repeal our bylaws without prior shareholder approval; the inability of our shareholders to act by written consent; advance notice requirements for director nominations and shareholder proposals; and super-majority voting requirement to amend some provisions in our certificate of incorporation and bylaws. Under our agreement with Microsoft for the Xbox, if someone makes an offer to purchase at least 30% of our outstanding common stock, Microsoft may have first and last rights of refusal to purchase the stock. These provisions could delay or prevent a change in control of NVIDIA, discourage proxy contests, and make it more difficult for shareholders to elect directors of their choosing and to cause us to take other corporate actions they desire.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors/Risks Related to Regulatory, Legal, Our Stock and Other Matters", + "metadata": { + "length": 48881, + "summary": "The passage details extensive legal, regulatory, and geopolitical risks facing NVIDIA, primarily focusing on U.S. export controls targeting AI semiconductors and GPUs, which have harmed its competitive position and foreclosed it from the China data center market. It also covers compliance with complex global laws (AI, data privacy, antitrust, trade sanctions), litigation risks, and the burdens of evolving regulations like the EU AI Act, all of which could materially impact business operations and financial results.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "subject", + "complex", + "laws", + "rules", + "regulations", + "political", + "actions", + "including", + "restrictions", + "export", + "products", + "adversely", + "impact", + "business", + "domestically", + "worldwide", + "affecting", + "operations", + "areas", + "limited", + "IP", + "ownership", + "infringement", + "taxes", + "import", + "requirements", + "tariffs", + "anti", + "corruption", + "Foreign", + "Corrupt", + "Practices", + "Act", + "acquisitions", + "foreign", + "exchange", + "controls", + "cash", + "repatriation", + "investment", + "data", + "privacy", + "competition", + "antitrust", + "advertising", + "employment", + "product", + "cybersecurity", + "environmental", + "health", + "safety", + "responsible", + "AI", + "sustainability", + "cryptocurrency", + "consumer", + "Compliance", + "onerous", + "expensive", + "competitive", + "position", + "negatively", + "ability", + "manufacture", + "ship", + "There", + "assurance", + "employees", + "contractors", + "suppliers", + "customers", + "agents", + "violate", + "applicable", + "policies", + "procedures", + "designed", + "ensure", + "compliance", + "violations", + "result", + "fines", + "criminal", + "sanctions", + "officers", + "prohibitions", + "conduct", + "damage", + "reputation", + "Changes", + "interpretation", + "enforcement", + "lead", + "materially", + "greater", + "costs", + "supply", + "operate", + "For", + "face", + "increased", + "increases", + "legislation", + "regulation", + "administrative", + "rule", + "making", + "focus", + "regulators", + "vulnerabilities", + "risks", + "Our", + "markets", + "relating", + "led", + "interest", + "European", + "Union", + "United", + "States", + "Kingdom", + "South", + "Korea", + "Japan", + "China", + "French", + "Competition", + "Authority", + "collected", + "information", + "graphics", + "card", + "CSP", + "market", + "part", + "ongoing", + "inquiry", + "received", + "continue", + "receive", + "broad", + "requests", + "sales", + "GPUs", + "NVIDIA", + "efforts", + "allocate", + "foundation", + "models", + "investments", + "partnerships", + "agreements", + "companies", + "developing", + "compete", + "strategies", + "roadmaps", + "develop", + "sell", + "hardware", + "software", + "system", + "solutions", + "partners", + "expect", + "additional", + "future", + "Such", + "burdensome", + "relationships", + "Governments", + "cases", + "imposed", + "systems", + "frontier", + "generative", + "EU", + "effective", + "August", + "2024", + "fully", + "year", + "transitional", + "period", + "The", + "train", + "deploy", + "release", + "Several", + "states", + "enacting", + "enacted", + "technologies", + "state", + "effect", + "January", + "2026", + "increase", + "Restrictions", + "implemented", + "burdens", + "delay", + "halt", + "deployment", + "reduce", + "number", + "entrants", + "impacting", + "financial", + "results", + "Revisions", + "taxation", + "trade", + "imposition", + "duties", + "imports", + "exports", + "retaliatory", + "adverse", + "plans", + "timing", + "shipments", + "Additionally", + "public", + "perception", + "governments", + "regions", + "plan", + "Government", + "protection", + "national", + "economic", + "security", + "government", + "bodies", + "deemed", + "activities", + "persons", + "decrees", + "quotas", + "barriers", + "affect", + "provide", + "services", + "license", + "entities", + "Department", + "Commerce", + "Entity", + "List", + "USG", + "restricted", + "parties", + "lists", + "expected", + "change", + "time", + "generally", + "fulfill", + "contractual", + "obligations", + "material", + "If", + "found", + "violated", + "control", + "similar", + "violation", + "occurred", + "knowledge", + "penalties", + "operating", + "condition", + "response", + "war", + "Ukraine", + "jurisdictions", + "measures", + "blocked", + "passage", + "support", + "Russia", + "Belarus", + "In", + "fiscal", + "2023", + "stopped", + "direct", + "closed", + "Concurrently", + "impacted", + "EMEA", + "increasing", + "strategic", + "importance", + "resulted", + "regulatory", + "target", + "capable", + "enabling", + "facilitating", + "service", + "offerings", + "Concerns", + "party", + "purposes", + "contrary", + "local", + "governmental", + "interests", + "concerns", + "misuse", + "applications", + "unilateral", + "multilateral", + "training", + "modifying", + "tuning", + "deploying", + "LLMs", + "limit", + "downstream", + "users", + "acquire", + "include", + "technology", + "As", + "geopolitical", + "tensions", + "semiconductors", + "related", + "increasingly", + "proposed", + "stakeholders", + "allies", + "restricting", + "adopted", + "scope", + "application", + "prohibit", + "exporting", + "manufacturing", + "testing", + "warehousing", + "locations", + "options", + "impose", + "conditions", + "serve", + "demand", + "abroad", + "revenue", + "Export", + "targeting", + "restrictive", + "creating", + "disadvantage", + "subjected", + "resale", + "repair", + "transfer", + "Controls", + "cost", + "cloud", + "CSPs", + "end", + "disrupt", + "chain", + "distribution", + "channels", + "center", + "possibility", + "benefiting", + "competitors", + "offer", + "alternatives", + "Repeated", + "Increasing", + "Reduced", + "due", + "excess", + "inventory", + "incur", + "charges", + "Additional", + "limitations", + "research", + "development", + "teams", + "execute", + "roadmap", + "objectives", + "timely", + "manner", + "overseas", + "provoke", + "responses", + "substantially", + "Regulators", + "inquired", + "fulfillment", + "commitments", + "entered", + "close", + "Mellanox", + "acquisition", + "On", + "September", + "15", + "2025", + "published", + "preliminary", + "finding", + "required", + "degraded", + "Chinese", + "discriminated", + "unfairly", + "terms", + "approval", + "conclude", + "failed", + "law", + "orders", + "networking", + "Over", + "past", + "years", + "series", + "shifting", + "expanding", + "2022", + "announced", + "licensing", + "semiconductor", + "supercomputing", + "industries", + "These", + "chips", + "equipment", + "produce", + "Hong", + "Kong", + "Macau", + "specifically", + "A100", + "H100", + "integrated", + "circuits", + "DGX", + "boards", + "incorporate", + "July", + "informed", + "requirement", + "subset", + "destined", + "countries", + "Middle", + "East", + "October", + "updated", + "Country", + "Groups", + "Saudi", + "Arabia", + "Arab", + "Emirates", + "Vietnam", + "excluding", + "Israel", + "exceeding", + "performance", + "thresholds", + "A800", + "H800", + "L4", + "L40", + "L40S", + "RTX", + "4090", + "GB200", + "NVL72", + "B200", + "apply", + "headquartered", + "ultimate", + "parent", + "Group", + "D5", + "April", + "requires", + "H2O", + "achieving", + "memory", + "bandwidth", + "interconnect", + "combination", + "thereof", + "incurred", + "4.5", + "billion", + "charge", + "quarter", + "purchase", + "diminished", + "granted", + "licenses", + "based", + "generated", + "approximately", + "60", + "million", + "officials", + "expressed", + "expectation", + "15%", + "licensed", + "publish", + "codifying", + "February", + "small", + "amounts", + "H200", + "specific", + "To", + "date", + "program", + "allowed", + "H200s", + "inspection", + "process", + "prior", + "shipment", + "customer", + "shipped", + "25%", + "tariff", + "importation", + "event", + "pass", + "litigation", + "harmed", + "address", + "variety", + "parameters", + "total", + "processing", + "chip", + "density", + "Under", + "current", + "landscape", + "unable", + "create", + "deliver", + "receives", + "effectively", + "foreclosed", + "competing", + "computing", + "compute", + "foreclosure", + "helped", + "build", + "larger", + "developer", + "ecosystems", + "challenge", + "Unless", + "return", + "meets", + "lost", + "opportunity", + "benefit", + "addition", + "Diffusion", + "IFR", + "Federal", + "Register", + "GB300", + "divided", + "world", + "tiers", + "relegating", + "Tier", + "status", + "created", + "scheme", + "approvals", + "May", + "rescind", + "implement", + "replacement", + "forthcoming", + "remain", + "uncertain", + "add", + "Senate", + "passed", + "GAIN", + "NDAA", + "restrict", + "Trump", + "Administration", + "adapt", + "Biden", + "private", + "review", + "overturn", + "policy", + "decisions", + "made", + "long", + "term", + "place", + "expanded", + "geographic", + "internal", + "solution", + "warranty", + "extended", + "grant", + "denies", + "significant", + "transition", + "resolved", + "opportunities", + "evaporate", + "Even", + "grants", + "requested", + "temporary", + "installation", + "maintenance", + "choose", + "make", + "pre", + "sale", + "post", + "technical", + "cumbersome", + "encourage", + "pursue", + "Europe", + "Given", + "rising", + "changed", + "wider", + "range", + "firms", + "large", + "scale", + "GPU", + "clusters", + "imposing", + "exported", + "requiring", + "tracking", + "throttling", + "mechanisms", + "disable", + "impair", + "events", + "unauthorized", + "configuration", + "location", + "detected", + "mandates", + "designs", + "introduce", + "expose", + "risk", + "potential", + "liability", + "binding", + "draft", + "bills", + "legislative", + "proposals", + "calling", + "mandatory", + "features", + "publicly", + "questioned", + "built", + "discouraging", + "purchasing", + "provided", + "explaining", + "respond", + "follow", + "questions", + "Open", + "source", + "rapidly", + "growing", + "popularity", + "developers", + "Any", + "restriction", + "limits", + "originating", + "DeepSeek", + "Qwen", + "KIMMI", + "gaming", + "expands", + "portion", + "substantial", + "warehoused", + "distributed", + "servers", + "high", + "speed", + "network", + "interconnects", + "model", + "disproportionate", + "design", + "burden", + "encouraged", + "request", + "harming", + "Latin", + "America", + "Southeast", + "Asia", + "investing", + "advanced", + "ready", + "resulting", + "recently", + "experienced", + "At", + "While", + "seek", + "strictly", + "comply", + "reports", + "diversion", + "controlled", + "unsubstantiated", + "untrue", + "Incorrect", + "allegations", + "satisfy", + "letter", + "spirit", + "incorrect", + "legitimate", + "supposed", + "loopholes", + "developed", + "manufactured", + "Connected", + "Vehicle", + "automotive", + "interpreted", + "broadly", + "open", + "favor", + "duration", + "access", + "games", + "Gaming", + "participate", + "oversight", + "digital", + "platform", + "Data", + "Center", + "discouraged", + "importing", + "agency", + "Action", + "Plan", + "endorses", + "standards", + "watt", + "accelerators", + "renovated", + "centers", + "Although", + "modify", + "prevents", + "meet", + "standard", + "Further", + "Finally", + "depends", + "consistent", + "reliable", + "Taiwan", + "components", + "parts", + "Scrutiny", + "shareholders", + "corporate", + "practices", + "willingness", + "Certain", + "shareholder", + "advocacy", + "groups", + "funds", + "participants", + "focused", + "disclosures", + "climate", + "human", + "rights", + "Stakeholders", + "satisfied", + "goals", + "adoption", + "level", + "initiatives", + "differ", + "expectations", + "disclosure", + "evolving", + "regulator", + "industry", + "stakeholder", + "fail", + "reporting", + "legal", + "require", + "resources", + "monitor", + "report", + "discontinue", + "expand", + "existing", + "factors", + "harm", + "brand", + "Issues", + "reputational", + "resolve", + "issues", + "building", + "capabilities", + "protections", + "stand", + "poses", + "emerging", + "social", + "ethical", + "presents", + "challenges", + "enable", + "draw", + "controversy", + "perceived", + "actual", + "society", + "unintended", + "consequences", + "infringe", + "copyright", + "publicity", + "controversial", + "frameworks", + "offered", + "experience", + "Leveraging", + "potentially", + "improve", + "functions", + "present", + "Complying", + "multiple", + "impede", + "transparency", + "Code", + "Practice", + "General", + "Purpose", + "Artificial", + "Intelligence", + "notify", + "Commission", + "details", + "Trustworthy", + "processes", + "framework", + "Furthermore", + "disproportionately", + "failure", + "adequately", + "undermine", + "confidence", + "slow", + "Actions", + "protect", + "unsuccessful", + "prohibited", + "selling", + "From", + "involved", + "lawsuits", + "proceedings", + "alleging", + "patent", + "agreed", + "indemnify", + "An", + "unfavorable", + "ruling", + "damages", + "invalidation", + "patents", + "indemnification", + "payment", + "profits", + "injunctive", + "relief", + "Claims", + "merit", + "defend", + "claims", + "divert", + "attention", + "management", + "personnel", + "commence", + "expenses", + "countersuits", + "indemnitees", + "obtain", + "acceptable", + "liabilities", + "suspend", + "royalty", + "payments", + "cross", + "arrangements", + "concluded", + "commercially", + "reasonable", + "indemnitee", + "rely", + "trademarks", + "secrets", + "employee", + "nondisclosure", + "extent", + "This", + "makes", + "piracy", + "theft", + "publication", + "confidential", + "acceptance", + "marketing", + "reduced", + "hired", + "continuously", + "assess", + "formal", + "innovations", + "approved", + "enforceable", + "stringent", + "changing", + "deter", + "sensitive", + "personal", + "external", + "contracts", + "govern", + "behalf", + "regard", + "collection", + "retention", + "matters", + "unfounded", + "loss", + "possession", + "disruption", + "remediation", + "defending", + "federal", + "authorities", + "numerous", + "breach", + "notification", + "Numerous", + "comprehensive", + "covered", + "businesses", + "providing", + "notices", + "affording", + "residents", + "correct", + "delete", + "opt", + "targeted", + "profiling", + "automated", + "decision", + "exercise", + "stricter", + "conducting", + "assessments", + "statutory", + "noncompliance", + "California", + "Consumer", + "Privacy", + "2018", + "amended", + "Rights", + "2020", + "CPRA", + "collectively", + "CCPA", + "sharing", + "detailed", + "shared", + "intentional", + "action", + "breaches", + "Similar", + "considered", + "levels", + "localities", + "machine", + "learning", + "Worldwide", + "Protection", + "Regulation", + "GDPR", + "similarly", + "governing", + "strict", + "handling", + "Economic", + "Area", + "EEA", + "stored", + "20", + "4%", + "whichever", + "classes", + "individuals", + "organizations", + "initiate", + "Network", + "Information", + "Security", + "Directive", + "NIS2", + "regulates", + "resilience", + "incident", + "sectors", + "infrastructure", + "sector", + "Non", + "maximum", + "10", + "Euros", + "2%", + "preceding", + "ordinary", + "localization", + "border", + "governs", + "released", + "set", + "Standard", + "Contractual", + "Clauses", + "validly", + "adequate", + "International", + "Transfer", + "Agreement", + "Addendum", + "Framework", + "extension", + "thereto", + "transfers", + "relevant", + "certify", + "However", + "lawfully", + "Other", + "residency", + "complexity", + "imposes", + "defined", + "important", + "transferable", + "assessment", + "Cyberspace", + "CAC", + "failures", + "prevent", + "inability", + "significantly", + "collaborate", + "expense", + "Some", + "prevented", + "transferring", + "allegedly", + "violating", + "bound", + "successful", + "claimed", + "compliant", + "providers", + "host", + "collaboration", + "exposed", + "altered", + "materials", + "statements", + "certifications", + "principles", + "scrutinizing", + "deficient", + "lacking", + "deceptive", + "unfair", + "misrepresentative", + "investigation", + "quickly", + "applied", + "fashion", + "inconsistent", + "necessitate", + "Despite", + "bans", + "destroy", + "exposure", + "tax", + "higher", + "rates", + "income", + "Significant", + "judgment", + "determining", + "provision", + "regularly", + "audit", + "estimates", + "outcome", + "rate", + "amount", + "net", + "flows", + "impacts", + "Most", + "taxable", + "qualifying", + "preferential", + "treatment", + "derived", + "deduction", + "eligible", + "FDDEI", + "affected", + "international", + "developments", + "implementation", + "Two", + "Pillar", + "Organization", + "Cooperation", + "Development", + "OECD", + "involves", + "reallocation", + "taxing", + "respect", + "multinational", + "enterprises", + "fixed", + "profit", + "margin", + "carry", + "referred", + "One", + "minimum", + "elements", + "side", + "safe", + "harbors", + "coordination", + "uncertainty", + "mix", + "earnings", + "differing", + "incentives", + "credits", + "deductions", + "expiration", + "statutes", + "accounting", + "adjustments", + "finalization", + "returns", + "deductible", + "deferred", + "assets", + "asset", + "valuation", + "allowances", + "combinations", + "domestic", + "organization", + "structure", + "effects", + "stock", + "compensation", + "volatility", + "price", + "occur", + "decline", + "benefits", + "investigations", + "demands", + "involving", + "intellectual", + "property", + "securities", + "class", + "lawsuit", + "asserting", + "false", + "misleading", + "channel", + "mining", + "2017", + "Litigation", + "inherently", + "rulings", + "monetary", + "injunction", + "stopping", + "engaging", + "remedies", + "compulsory", + "settlement", + "Regardless", + "costly", + "consuming", + "disruptive", + "Delaware", + "certificate", + "incorporation", + "bylaws", + "agreement", + "Microsoft", + "takeover", + "provisions", + "Corporation", + "Law", + "discourage", + "Provisions", + "difficult", + "majority", + "outstanding", + "Board", + "Directors", + "issue", + "preferred", + "directors", + "amend", + "repeal", + "act", + "written", + "consent", + "advance", + "notice", + "director", + "nominations", + "super", + "voting", + "Xbox", + "30%", + "common", + "refusal", + "proxy", + "contests", + "elect", + "choosing", + "desire" + ], + "keywords": [ + "Export Controls", + "AI Regulation", + "Legal Risks" + ], + "connect_to": [] + } + }, + { + "chunk_id": "67ddeed7-c2ee-51fb-ada3-a387f94472c0", + "type": "text", + "content": "None.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1B. Unresolved Staff Comments", + "metadata": { + "length": 5, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b4217a21-af8d-5edf-b161-73981fc68401", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1C. Cybersecurity", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c3dc4bc8-f182-5873-971d-9f06cdbd8564", + "type": "text", + "content": "We have in place certain infrastructure, systems, policies, and procedures that are designed to proactively prevent or reduce the impact of, and reactively address circumstances that arise when, events such as a cybersecurity incident occur. These include processes for assessing, identifying, and managing material risks from cybersecurity threats. Our information security management programs generally follow certain processes outlined in frameworks such as the ISO 27001 international standard for information security management and we evaluate and evolve our security measures as appropriate. We consult with external parties, such as cybersecurity firms and risk management and governance experts, on risk management and strategy.\nIdentifying, assessing, and managing cybersecurity risk is integrated into our overall risk management systems and processes, and we have in place cybersecurity and data privacy training and policies designed to (a) respond to new requirements in global privacy and cybersecurity laws and (b) prevent, detect, respond to, mitigate and recover from identified and significant cybersecurity threats.\nWe also have a vendor risk assessment process consisting of, depending on the nature and sensitivity of the supplier and data they process on our behalf, the distribution and review of supplier questionnaires designed to help us evaluate cybersecurity risks that we may encounter when working with third parties that have access to confidential and other sensitive company information. We take steps to review that such vendors have implemented data privacy and security controls that help mitigate the cybersecurity risks associated with these vendors, depending on the nature and sensitivity of the supplier and data they process on our behalf. We routinely assess our high-risk suppliers' conformance to industry standards (e.g., ISO 27001, ISO 28001, and C-TPAT), and we evaluate them for additional information, product, and physical security requirements.\nRefer to “Item 1A. Risk factors” in this annual report on Form 10-K for additional information about cybersecurity-related risks.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1C. Cybersecurity/Risk management and strategy", + "metadata": { + "length": 2127, + "summary": "The company has infrastructure, systems, policies, and procedures to prevent, reduce impact of, and react to cybersecurity incidents. It follows ISO 27001 frameworks, consults external experts, integrates cybersecurity risk into overall risk management, provides training, and has a vendor risk assessment process using questionnaires and industry standards like ISO 27001, ISO 28001, and C-TPAT.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "place", + "infrastructure", + "systems", + "policies", + "procedures", + "designed", + "proactively", + "prevent", + "reduce", + "impact", + "reactively", + "address", + "circumstances", + "arise", + "events", + "cybersecurity", + "incident", + "occur", + "These", + "include", + "processes", + "assessing", + "identifying", + "managing", + "material", + "risks", + "threats", + "Our", + "information", + "security", + "management", + "programs", + "generally", + "follow", + "outlined", + "frameworks", + "ISO", + "27001", + "international", + "standard", + "evaluate", + "evolve", + "measures", + "consult", + "external", + "parties", + "firms", + "risk", + "governance", + "experts", + "strategy", + "Identifying", + "integrated", + "data", + "privacy", + "training", + "respond", + "requirements", + "global", + "laws", + "detect", + "mitigate", + "recover", + "identified", + "significant", + "vendor", + "assessment", + "process", + "consisting", + "depending", + "nature", + "sensitivity", + "supplier", + "behalf", + "distribution", + "review", + "questionnaires", + "encounter", + "working", + "access", + "confidential", + "sensitive", + "company", + "steps", + "vendors", + "implemented", + "controls", + "routinely", + "assess", + "high", + "suppliers", + "conformance", + "industry", + "standards", + "28001", + "TPAT", + "additional", + "product", + "physical", + "Refer", + "Item", + "1A", + "Risk", + "factors", + "annual", + "report", + "Form", + "10", + "related" + ], + "keywords": [ + "cybersecurity", + "risk management", + "ISO 27001" + ], + "connect_to": [] + } + }, + { + "chunk_id": "561bb45c-5f0e-5414-bdc9-154d0df5aef4", + "type": "text", + "content": "Information security matters, including managing and assessing risks from cybersecurity threats, remain under the oversight of the Company's Board of Directors, or the Board. The Audit Committee of the Board, or the Audit Committee, also reviews the adequacy and effectiveness of the Company's information security policies and practices and the internal controls regarding information security risks. The Audit Committee receives regular information security updates from management, including our Chief Security Officer and members of our security team. The Board also receives annual reports on information security matters from our Chief Security Officer and members of our security team.\nOur security efforts are managed by a team of executive cybersecurity, IT, engineering, operations, and legal professionals. We have established a cross-functional leadership team, consisting of executive-level leaders, that meets regularly to review cybersecurity matters and evaluate emerging threats. With oversight and guidance provided by the cross-functional leadership team, our information security teams refine our practices to address emerging security risks and changes in regulations. Our executive-level leadership team also participates in cybersecurity incident response efforts by engaging with the incident response team and helping direct the company's response to and assessment of certain cybersecurity incidents.\nWe have designated a Chief Security Officer, reporting to our Senior Vice President of Software Engineering, to oversee the identification, assessment, and management of material cybersecurity risks. Our Chief Security Officer's cybersecurity expertise includes over 18 years of combined government and private sector assignments.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1C. Cybersecurity/Governance", + "metadata": { + "length": 1757, + "summary": "The Company's Board of Directors oversees information security, with the Audit Committee reviewing policies and internal controls. Management provides regular updates, and a cross-functional leadership team meets to address cybersecurity matters. The Chief Security Officer, with 18 years of experience, manages material cybersecurity risks.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Information", + "security", + "matters", + "including", + "managing", + "assessing", + "risks", + "cybersecurity", + "threats", + "remain", + "oversight", + "Company", + "Board", + "Directors", + "The", + "Audit", + "Committee", + "reviews", + "adequacy", + "effectiveness", + "information", + "policies", + "practices", + "internal", + "controls", + "receives", + "regular", + "updates", + "management", + "Chief", + "Security", + "Officer", + "members", + "team", + "annual", + "reports", + "Our", + "efforts", + "managed", + "executive", + "IT", + "engineering", + "operations", + "legal", + "professionals", + "We", + "established", + "cross", + "functional", + "leadership", + "consisting", + "level", + "leaders", + "meets", + "regularly", + "review", + "evaluate", + "emerging", + "With", + "guidance", + "provided", + "teams", + "refine", + "address", + "regulations", + "participates", + "incident", + "response", + "engaging", + "helping", + "direct", + "company", + "assessment", + "incidents", + "designated", + "reporting", + "Senior", + "Vice", + "President", + "Software", + "Engineering", + "oversee", + "identification", + "material", + "expertise", + "includes", + "18", + "years", + "combined", + "government", + "private", + "sector", + "assignments" + ], + "keywords": [ + "Board oversight", + "Chief Security Officer", + "risk management" + ], + "connect_to": [] + } + }, + { + "chunk_id": "d57cc12a-307d-52e2-aaae-91534a8b56ca", + "type": "text", + "content": "Our headquarters is in Santa Clara, California. We own and lease approximately 3 million square feet of office and building space for our corporate headquarters. In addition, we lease data center space in Santa Clara, California. We also own and lease facilities for data centers, research and development, and/or sales and administrative purposes throughout the U.S. and in various international locations, primarily in China, India, Israel, and Taiwan. We believe our existing facilities, both owned and leased, are in good condition and suitable for the conduct of our business. We do not identify or allocate assets by operating segment. For additional information regarding obligations under leases, refer to Note 17 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K, which information is hereby incorporated by reference.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 2. Properties", + "metadata": { + "length": 886, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "headquarters", + "Santa", + "Clara", + "California", + "We", + "lease", + "approximately", + "million", + "square", + "feet", + "office", + "building", + "space", + "corporate", + "In", + "addition", + "data", + "center", + "facilities", + "centers", + "research", + "development", + "sales", + "administrative", + "purposes", + "international", + "locations", + "primarily", + "China", + "India", + "Israel", + "Taiwan", + "existing", + "owned", + "leased", + "good", + "condition", + "suitable", + "conduct", + "business", + "identify", + "allocate", + "assets", + "operating", + "segment", + "For", + "additional", + "information", + "obligations", + "leases", + "refer", + "Note", + "17", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4a9eb023-b993-5441-a9f4-5c08dc48fbdd", + "type": "text", + "content": "Please see Note 12 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for a discussion of our legal proceedings.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 3. Legal Proceedings", + "metadata": { + "length": 170, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Please", + "Note", + "12", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "discussion", + "legal", + "proceedings" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cc4c3124-4c66-5fd7-863e-f016f6ab1f29", + "type": "text", + "content": "Not applicable.", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 4. Mine Safety Disclosures", + "metadata": { + "length": 15, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Not", + "applicable" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e67d76ca-f999-52a1-97e3-4473e0d54378", + "type": "text", + "content": "Item 5. Market for Registrant's Common Equity, Related Stockholder Matters and Issuer Purchases of Equity Securities\nOur common stock is traded on the Nasdaq Global Select Market under the symbol NVDA. Public trading of our common stock began on January 22, 1999. Prior to that, there was no public market for our common stock. As of February 20, 2026, we had approximately 1,226 registered shareholders, not including those shares held in street or nominee name.\nIn May 2024, we announced a ten-for-one stock split, or the Stock Split, of our issued common stock, which was effected through the filing of an amendment to the Company's Restated Certificate of Incorporation, or the Amendment, with the Secretary of the State of Delaware. In June 2024, the Company filed the Amendment to effect the Stock Split and proportionately increased the number of shares of the Company's authorized common stock from 8.0 billion to 80.0 billion. Shareholders of record at the close of market on June 6, 2024 received nine additional shares of common stock, distributed after the close of market on June 7, 2024. All share, equity award and per share amounts presented herein have been retrospectively adjusted to reflect the Stock Split.", + "path": "2026-Annual-Report-Web.pdf/Part II_2", + "metadata": { + "length": 1227, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Item", + "Market", + "Registrant", + "Common", + "Equity", + "Related", + "Stockholder", + "Matters", + "Issuer", + "Purchases", + "Securities", + "Our", + "common", + "stock", + "traded", + "Nasdaq", + "Global", + "Select", + "symbol", + "NVDA", + "Public", + "trading", + "began", + "January", + "22", + "1999", + "Prior", + "public", + "market", + "As", + "February", + "20", + "2026", + "approximately", + "226", + "registered", + "shareholders", + "including", + "shares", + "held", + "street", + "nominee", + "In", + "May", + "2024", + "announced", + "ten", + "split", + "Stock", + "Split", + "issued", + "effected", + "filing", + "amendment", + "Company", + "Restated", + "Certificate", + "Incorporation", + "Amendment", + "Secretary", + "State", + "Delaware", + "June", + "filed", + "effect", + "proportionately", + "increased", + "number", + "authorized", + "8.0", + "billion", + "80.0", + "Shareholders", + "record", + "close", + "received", + "additional", + "distributed", + "All", + "share", + "equity", + "award", + "amounts", + "presented", + "retrospectively", + "adjusted", + "reflect" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "38a97fdf-fb8a-5908-aa10-7c4486338b18", + "type": "table", + "content": "
PeriodTotal Number of Shares Purchased (In millions)Average Price Paid per Share (1)Total Number of Shares Purchased as Part of Publicly Announced Program (In millions)Approximate Dollar Value of Shares that May Yet Be Purchased Under the Program (In billions)
October 27, 2025 - November 23, 20255.6$ 198.895.6$ 61.1
November 24, 2025 - December 21, 20256.6$ 179.416.6$ 59.9
December 22, 2025 - January 25, 20267.3$ 186.527.3$ 58.5
Total19.519.5
", + "path": "tables/table-61 Stock Buyback.html", + "metadata": { + "length": 727, + "summary": "table-62\nThe table shows a stock buyback program from Oct 2025 to Jan 2026, with 19.5 million shares purchased at average prices around $179-$199, leaving $58.5 billion available.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-61 Stock Buyback.html", + "keywords": [ + "share repurchase", + "buyback program", + "stock" + ], + "tokens": [] + } + }, + { + "chunk_id": "42f87c92-653f-5075-b6a3-a3c4bb010659", + "type": "text", + "content": "On August 26, 2025, our Board of Directors approved an additional \\$60.0 billion in share repurchase authorization, without expiration. In fiscal year 2026, we repurchased 282 million shares of our common stock for \\$40.4 billion. As of January 25, 2026, we were authorized, subject to certain specifications, to repurchase up to \\$58.5 billion of our common stock.\nThe repurchases can be made in the open market, in privately negotiated transactions, pursuant to a Rule 10b5-1 trading plan or in structured share repurchase agreements in compliance with Rule 10b-18 of the Exchange Act, subject to market conditions, applicable legal requirements, and other factors. Our share repurchase program may be suspended at any time at our discretion.\nIn fiscal year 2026, we paid cash dividends to our shareholders of \\$974 million. The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders.\nThe following table presents details of our share repurchase transactions during the fourth quarter of fiscal year 2026:\n\n[tables/table-61 Stock Buyback.html]\n\n(1) Average price paid per share includes broker commissions, but excludes our liability under the 1% excise tax on the net amount of our share repurchases required by the Inflation Reduction Act of 2022.\nFrom January 26, 2026 through February 20, 2026, we repurchased 8 million shares for \\$1.5 billion pursuant to a pre-established trading plan.\nRestricted Stock Unit Share Withholding\nWe withhold shares of our common stock associated with net share settlements to cover tax withholding obligations of awards under our employee equity incentive program. During fiscal year 2026, we withheld approximately 51 million shares for a total value of \\$7.9 billion through net share settlements. Refer to Note 3 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for further discussion regarding our equity incentive plans.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Issuer Purchases of Equity Securities", + "metadata": { + "length": 2109, + "summary": "On August 26, 2025, the Board approved an additional $60.0 billion share repurchase authorization with no expiration. In fiscal year 2026, 282 million shares were repurchased for $40.4 billion. As of January 25, 2026, up to $58.5 billion remained authorized for repurchase. Repurchases may be made via open market, private negotiations, Rule 10b5-1 plans, or structured agreements. The program can be suspended at any time. In fiscal year 2026, $974 million in cash dividends were paid. The table details fourth quarter repurchases. From January 26 to February 20, 2026, 8 million shares were repurchased for $1.5 billion. Additionally, 51 million shares were withheld for tax obligations under equity incentive plans, valued at $7.9 billion.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "On", + "August", + "26", + "2025", + "Board", + "Directors", + "approved", + "additional", + "60.0", + "billion", + "share", + "repurchase", + "authorization", + "expiration", + "In", + "fiscal", + "year", + "2026", + "repurchased", + "282", + "million", + "shares", + "common", + "stock", + "40.4", + "As", + "January", + "25", + "authorized", + "subject", + "specifications", + "58.5", + "The", + "repurchases", + "made", + "open", + "market", + "privately", + "negotiated", + "transactions", + "pursuant", + "Rule", + "10b5", + "trading", + "plan", + "structured", + "agreements", + "compliance", + "10b", + "18", + "Exchange", + "Act", + "conditions", + "applicable", + "legal", + "requirements", + "factors", + "Our", + "program", + "suspended", + "time", + "discretion", + "paid", + "cash", + "dividends", + "shareholders", + "974", + "payment", + "future", + "continuing", + "determination", + "declaration", + "interests", + "table", + "presents", + "details", + "fourth", + "quarter", + "Average", + "price", + "includes", + "broker", + "commissions", + "excludes", + "liability", + "1%", + "excise", + "tax", + "net", + "amount", + "required", + "Inflation", + "Reduction", + "2022", + "From", + "February", + "20", + "1.5", + "pre", + "established", + "Restricted", + "Stock", + "Unit", + "Share", + "Withholding", + "We", + "withhold", + "settlements", + "cover", + "withholding", + "obligations", + "awards", + "employee", + "equity", + "incentive", + "During", + "withheld", + "approximately", + "51", + "total", + "7.9", + "Refer", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "discussion", + "plans" + ], + "keywords": [ + "share repurchase", + "dividends", + "stock buyback" + ], + "connect_to": [ + { + "target": "38a97fdf-fb8a-5908-aa10-7c4486338b18", + "relation": "embeds", + "ref": "[tables/table-61 Stock Buyback.html]", + "position": { + "start": 1129, + "end": 1165 + } + } + ] + } + }, + { + "chunk_id": "32fffc28-0654-54a5-85bd-04e583ceae82", + "type": "text", + "content": "On December 15, 2025, we acquired a company and issued to a key employee a total of 174,676 shares of our common stock, valued at approximately \\$31 million based on our closing stock price on the issuance date.\nThe above securities were issued in a transaction not involving a public offering pursuant to an exemption from registration set forth in Section 4(a)(2) of the Securities Act (and Regulation D or Regulation S promulgated thereunder).", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Recent Sales of Unregistered Securities and Use of Proceeds", + "metadata": { + "length": 446, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "On", + "December", + "15", + "2025", + "acquired", + "company", + "issued", + "key", + "employee", + "total", + "174", + "676", + "shares", + "common", + "stock", + "valued", + "approximately", + "31", + "million", + "based", + "closing", + "price", + "issuance", + "date", + "The", + "securities", + "transaction", + "involving", + "public", + "offering", + "pursuant", + "exemption", + "registration", + "set", + "Section", + "Securities", + "Act", + "Regulation", + "promulgated", + "thereunder" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a5bdbb97-46c6-5188-8d37-67fdee33872c", + "type": "image", + "content": "\nOn December 15, 2025, we acquired a company and issued to a key employee a total of 174,676 shares of our common stock, valued at approximately \\$31 million based on our closing stock price on the issuance date.
0\n[images/image-83-NVIDIA Stock Growth.jpg]\n", + "path": "images/image-83-NVIDIA Stock Growth.jpg", + "metadata": { + "length": 257, + "summary": "image-83\nChart Title: NVIDIA Corporation vs S&P 500 vs Nasdaq 100 Performance\n\nTime Range: January 31, 2021 to January 25, 2026\n\nData Series & Key Points:\n- **NVIDIA Corporation (Green Line):** Shows exponential growth.\n - Jan 2021: ~100\n - Jan 2022: ~180\n - Jan 2023: ~170\n - Jan 2024: ~470\n - Jan 2025: ~1,100\n - Jan 2026: ~1,450\n- **S&P 500 (Grey Square Marker):** Shows slow, steady linear growth.\n - Starts at ~100 in Jan 2021 and ends near ~200 in Jan 2026.\n- **Nasdaq 100 (Purple Triangle Marker):** Shows slow, steady linear growth similar to S&P 500", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-83-NVIDIA Stock Growth.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e92632b0-98ca-5bea-9606-b440c923f4c2", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
DateNVIDIA CorporationS&P 500Nasdaq 100
1/31/202110080100
1/30/2022180100120
1/29/20231609090
1/28/2024470120150
1/26/20251100170180
1/25/20261450200210
", + "path": "tables/table-62 NVIDIA Stock Growth.html", + "metadata": { + "length": 845, + "summary": "table-63\nThe table shows NVIDIA Corporation's stock value rising from 100 in 2021 to 1450 in 2026, outperforming S&P 500 and Nasdaq 100 indices.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-62 NVIDIA Stock Growth.html", + "keywords": [ + "NVIDIA", + "S&P 500", + "Nasdaq 100" + ], + "tokens": [] + } + }, + { + "chunk_id": "92fd42d2-d3fe-5765-a448-df41ef9ee2ff", + "type": "table", + "content": "
1/31/20211/30/20221/29/20231/28/20241/26/20251/25/2026
NVIDIA Corporation$ 100.00$ 175.98$ 157.05$ 470.88$ 1,100.68$ 1,448.75
S&P 500$ 100.00$ 121.00$ 112.98$ 137.98$ 174.50$ 200.33
Nasdaq 100$ 100.00$ 112.60$ 95.60$ 138.07$ 173.95$ 206.01
", + "path": "tables/table-63 Stock Performance.html", + "metadata": { + "length": 543, + "summary": "table-64\nThe table shows cumulative investment returns from 2021 to 2026 for NVIDIA, S&P 500, and Nasdaq 100, with NVIDIA growing from $100 to $1,448.75.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-63 Stock Performance.html", + "keywords": [ + "NVIDIA", + "S&P 500", + "Nasdaq 100" + ], + "tokens": [] + } + }, + { + "chunk_id": "4979fd06-a3c9-58e9-b909-1bdee11e2e03", + "type": "text", + "content": "The following graph compares the cumulative total shareholder return for our common stock, the S&P 500 Index, and the Nasdaq 100 Index for the five years ended January 25, 2026. The graph assumes that \\$100 was invested on January 31, 2021 in our common stock and in each of the S&P 500 Index and the Nasdaq 100 Index. Our common stock is a component of each of the presented indices. Total return assumes reinvestment of dividends in each of the indices indicated. Total return is based on historical results and is not intended to indicate future performance.\nComparison of 5 Year Cumulative Total Return\\* Among NVIDIA Corporation, the S&P 500 Index, and the Nasdaq 100 Index\n\nOn December 15, 2025, we acquired a company and issued to a key employee a total of 174,676 shares of our common stock, valued at approximately \\$31 million based on our closing stock price on the issuance date.
0\n[images/image-83-NVIDIA Stock Growth.jpg]\n\n
\nline chart\n\n[tables/table-62 NVIDIA Stock Growth.html]\n\n
\n\\*\\$100 invested on 1/31/2021 in stock and in indices, including reinvestment of dividends.\nSource: FactSet financial data and analytics.\n\n[tables/table-63 Stock Performance.html]\n\nItem 6. [Reserved]\nItem 7. Management's Discussion and Analysis of Financial Condition and Results of Operations\nThe following discussion and analysis of our financial condition and results of operations should be read in conjunction with “Item 1A. Risk Factors,” our Consolidated Financial Statements and related Notes thereto, as well as other cautionary statements and risks described elsewhere in this Annual Report on Form 10-K, before deciding to purchase, hold, or sell shares of our common stock.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs", + "metadata": { + "length": 1783, + "summary": "The passage compares the cumulative total shareholder return of NVIDIA common stock, the S&P 500 Index, and the Nasdaq 100 Index over five years ending January 25, 2026, assuming a $100 investment on January 31, 2021 with dividends reinvested. It notes that on December 15, 2025, NVIDIA acquired a company and issued 174,676 shares to a key employee, valued at about $31 million. The content includes a line chart and tables with return data, followed by a reserved section and the start of Management's Discussion and Analysis.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "graph", + "compares", + "cumulative", + "total", + "shareholder", + "return", + "common", + "stock", + "500", + "Index", + "Nasdaq", + "100", + "years", + "ended", + "January", + "25", + "2026", + "assumes", + "invested", + "31", + "2021", + "Our", + "component", + "presented", + "indices", + "Total", + "reinvestment", + "dividends", + "based", + "historical", + "results", + "intended", + "future", + "performance", + "Comparison", + "Year", + "Cumulative", + "Return", + "Among", + "NVIDIA", + "Corporation", + "On", + "December", + "15", + "2025", + "acquired", + "company", + "issued", + "key", + "employee", + "174", + "676", + "shares", + "valued", + "approximately", + "million", + "closing", + "price", + "issuance", + "date", + "details", + "summary", + "line", + "chart", + "including", + "Source", + "FactSet", + "financial", + "data", + "analytics", + "Item", + "Reserved", + "Management", + "Discussion", + "Analysis", + "Financial", + "Condition", + "Results", + "Operations", + "discussion", + "analysis", + "condition", + "operations", + "read", + "conjunction", + "1A", + "Risk", + "Factors", + "Consolidated", + "Statements", + "related", + "Notes", + "thereto", + "cautionary", + "statements", + "risks", + "Annual", + "Report", + "Form", + "10", + "deciding", + "purchase", + "hold", + "sell" + ], + "keywords": [ + "Cumulative Return", + "NVIDIA", + "Indices" + ], + "connect_to": [ + { + "target": "a5bdbb97-46c6-5188-8d37-67fdee33872c", + "relation": "embeds", + "ref": "[images/image-83-NVIDIA Stock Growth.jpg]", + "position": { + "start": 903, + "end": 944 + } + }, + { + "target": "e92632b0-98ca-5bea-9606-b440c923f4c2", + "relation": "embeds", + "ref": "[tables/table-62 NVIDIA Stock Growth.html]", + "position": { + "start": 987, + "end": 1029 + } + }, + { + "target": "92fd42d2-d3fe-5765-a448-df41ef9ee2ff", + "relation": "embeds", + "ref": "[tables/table-63 Stock Performance.html]", + "position": { + "start": 1181, + "end": 1221 + } + } + ] + } + }, + { + "chunk_id": "18cf4096-81da-5526-88db-ef58fab1049d", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9339bafd-9d08-521f-8921-9469c46c8a2b", + "type": "text", + "content": "NVIDIA pioneered accelerated computing to help solve the most challenging computational problems. Since our original focus on PC graphics, we have expanded to several other large and important computationally intensive fields. Fueled by the sustained demand for exceptional 3D graphics and the scale of the gaming market, NVIDIA has leveraged its GPU architecture to create platforms for scientific computing, AI, data science, autonomous vehicles, robotics, and digital twin applications. NVIDIA is now a data center scale AI infrastructure company reshaping all industries.\nOur two operating segments are \"Compute & Networking\" and \"Graphics.\" Refer to Note 16 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information.\nHeadquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Our Company and Our Businesses", + "metadata": { + "length": 939, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "NVIDIA", + "pioneered", + "accelerated", + "computing", + "solve", + "challenging", + "computational", + "problems", + "Since", + "original", + "focus", + "PC", + "graphics", + "expanded", + "large", + "important", + "computationally", + "intensive", + "fields", + "Fueled", + "sustained", + "demand", + "exceptional", + "3D", + "scale", + "gaming", + "market", + "leveraged", + "GPU", + "architecture", + "create", + "platforms", + "scientific", + "AI", + "data", + "science", + "autonomous", + "vehicles", + "robotics", + "digital", + "twin", + "applications", + "center", + "infrastructure", + "company", + "reshaping", + "industries", + "Our", + "operating", + "segments", + "Compute", + "Networking", + "Graphics", + "Refer", + "Note", + "16", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "additional", + "information", + "Headquartered", + "Santa", + "Clara", + "California", + "incorporated", + "April", + "1993", + "reincorporated", + "Delaware", + "1998" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "75708579-9e90-582b-8906-5788e0519c9d", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Change
($ in millions, except per share data)
Revenue$ 215,938$ 130,497Up 65%
Gross margin71.1 %75.0 %-3.9 pts
Operating expenses$ 23,076$ 16,405Up 41%
Operating income$ 130,387$ 81,453Up 60%
Net income$ 120,067$ 72,880Up 65%
Net income per diluted share$ 4.90$ 2.94Up 67%
", + "path": "tables/table-64 Financial Results.html", + "metadata": { + "length": 701, + "summary": "table-65\nThe table shows financial data for years ending Jan 25, 2026 and Jan 26, 2025. Revenue rose 65% to $215,938M, net income up 65% to $120,067M, and operating income up 60% to $130,387M.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-64 Financial Results.html", + "keywords": [ + "Revenue", + "Net Income", + "Operating Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "ff80f4a6-7ab8-509d-bfa7-b9cd964c0bd6", + "type": "text", + "content": "Revenue growth in fiscal year 2026 was driven by data center compute and networking platforms for accelerated computing and AI solutions. Our Blackwell architectures represented the majority of our Data Center revenue.\nThe availability of data centers, energy, and capital to support the buildout of NVIDIA AI infrastructure by our customers and partners is crucial, and any shortage of these or other necessary resources could impact our future revenue and financial performance. Expanding energy capacity to meet demand is a complex, multi-year process that involves significant regulatory, technical, and construction challenges. In addition, access to capital can be particularly constrained for less-capitalized companies, which may face difficulties securing financing for large-scale infrastructure projects. These limitations could delay customer and partner deployments or reduce the scale of accelerated computing and AI adoption.\nWe continue to execute Data Center compute product introductions, bringing new advanced architectures on a one-year product cadence, including our Rubin platform. We began shipping production units of our new Blackwell Ultra platforms including GB300 in the second quarter of fiscal year 2026. The complexity of our product transitions and sophisticated system configurations has and may in the future cause delays in production and create challenges in managing supply and demand. This could further result in revenue volatility, quality issues, increased inventory provisions, decreases in product yields and higher material costs, and/or increased warranty costs. Customers may postpone purchasing new architectures or may adopt new technologies more gradually than anticipated, affecting our revenue timing and supply chain expenses.\nIn April 2025, the USG informed us that a license is required for exports of our H2O product into the China market. As a result of these requirements, we incurred a \\$4.5 billion charge in the first quarter of fiscal year 2026 associated with H2O for excess inventory and purchase obligations, as the demand for H2O diminished. In August 2025, the USG granted licenses that would allow us to ship certain H2O products to certain China-based customers. We generated approximately \\$60 million in H2O revenue under those licenses.\nIn February 2026, the USG granted a license that would allow us to ship small amounts of H200 products to specific China-based customers. To date, we have not generated any revenue under the H200 licensing program, and do not yet know whether any imports will be allowed into China. The license requires that the H200s go through an inspection process in the United States prior to any shipment to the customer. As a result, any H200 shipped under the new licensing program will be subject to a 25% tariff upon importation into the United States.\nThe recent rise in high-quality open-source foundation models is making advanced AI capabilities broadly accessible. Open-source AI is dependent on developer adoption and if deployed on our competitors' platforms, it could reduce demand for our products and services.\nWhile currently our supply chain is mainly concentrated in Asia, we are expanding into the U.S. and Latin America. These moves are expected to strengthen our supply chain, add resiliency and redundancy, and meet the growing demand for AI infrastructure. Our ability to increase manufacturing capabilities will depend on the local region's manufacturing ecosystem's capacity to ramp production supply to the required volume and on a timely basis.\nWe have made, and expect to continue making, investments that support our technology roadmap and the broader AI ecosystem. In fiscal year 2026, we made the following investments:\n- We invested \\$17.5 billion in private companies and infrastructure funds, primarily to support early-stage startups. These investments include AI model makers that purchase our products directly or through CSPs. Many of these investments are illiquid and non-marketable. The related early-stage startups may not become profitable in the near term, or at all, and there can be no assurance that we will realize a return on our investments.\n• We made investments in publicly-held equity securities where the value may fluctuate significantly due to changes in stock prices and could adversely affect our financial results.\n- To support the build-out of complex datacenter infrastructures, we enter into commercial arrangements, including guarantees with partners. We provided \\$3.5 billion in land, power, and shell guarantees to early-stage companies, generally over multi-year periods. If the escrow and the partners' operating activities are not sufficient to cover an event of default under these guarantees, we may elect to assume the underlying leases for internal use or sublease them to third parties.\nMacroeconomic factors, including tariffs, inflation, interest rate changes, capital market volatility, global supply chain constraints, and global economic and geopolitical developments, have direct and indirect impacts on our results of operations, particularly demand for our products. While difficult to isolate and quantify, these macroeconomic factors impact our supply chain and manufacturing costs, employee wages, costs for capital equipment, the value of our investments, revenue and competitive position. Our product and solution pricing generally does not fluctuate with short-term changes in our costs. Within our supply chain, we continuously manage product availability and costs with our vendors.\nRefer to “Item 1A. Risk Factors – Risks Related to Regulatory, Legal, Our Stock and Other Matters” for a further discussion of the potential impact of these factors on our business.\nFiscal Year 2026 Summary\n\n[tables/table-64 Financial Results.html]\n\nRevenue for fiscal year 2026 was \\$215.9 billion, up 65% from a year ago.\nData Center revenue for fiscal year 2026 was up 68% from a year ago. The strong year-on-year growth was driven by the major platform shifts – accelerated computing and AI.\nGaming revenue for fiscal year 2026 was up 41% from a year ago, driven by strong Blackwell demand. We expect supply constraints to be a headwind to Gaming in the first quarter of fiscal 2027 and beyond.\nProfessional Visualization revenue for fiscal year 2026 was up 70% from a year ago, driven by exceptional demand for Blackwell as well as the launch of our new DGX Spark.\nAutomotive revenue for fiscal year 2026 was up 39% from a year ago, driven by continued adoption of our self-driving platforms.\nGross margin decreased in fiscal year 2026 as our business model transitioned from offering Hopper HGX systems to Blackwell full-scale datacenter solutions. The gross margin decrease was also impacted by a \\$4.5 billion charge associated with H2O excess inventory and purchase obligations.\nOperating expenses for fiscal year 2026 were up 41% from a year ago, driven by higher compensation and benefits expenses due to employee growth and compute and infrastructure costs.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Recent Developments, Future Objectives and Challenges", + "metadata": { + "length": 7032, + "summary": "NVIDIA's fiscal year 2026 revenue reached $215.9 billion, up 65% year-over-year, driven by data center compute and AI solutions, with Blackwell architectures dominating Data Center revenue. The company faces challenges including export restrictions to China (incurring a $4.5 billion charge for H2O inventory), supply chain concentration in Asia, and risks from open-source AI competition. Investments totaled $17.5 billion in private companies and infrastructure, with $3.5 billion in guarantees. Gross margin decreased due to business model transitions and the H2O charge. Segment growth: Data Center up 68%, Gaming up 41%, Professional Visualization up 70%, Automotive up 39%.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Revenue", + "growth", + "fiscal", + "year", + "2026", + "driven", + "data", + "center", + "compute", + "networking", + "platforms", + "accelerated", + "computing", + "AI", + "solutions", + "Our", + "Blackwell", + "architectures", + "represented", + "majority", + "Data", + "Center", + "revenue", + "The", + "availability", + "centers", + "energy", + "capital", + "support", + "buildout", + "NVIDIA", + "infrastructure", + "customers", + "partners", + "crucial", + "shortage", + "resources", + "impact", + "future", + "financial", + "performance", + "Expanding", + "capacity", + "meet", + "demand", + "complex", + "multi", + "process", + "involves", + "significant", + "regulatory", + "technical", + "construction", + "challenges", + "In", + "addition", + "access", + "constrained", + "capitalized", + "companies", + "face", + "difficulties", + "securing", + "financing", + "large", + "scale", + "projects", + "These", + "limitations", + "delay", + "customer", + "partner", + "deployments", + "reduce", + "adoption", + "We", + "continue", + "execute", + "product", + "introductions", + "bringing", + "advanced", + "cadence", + "including", + "Rubin", + "platform", + "began", + "shipping", + "production", + "units", + "Ultra", + "GB300", + "quarter", + "complexity", + "transitions", + "sophisticated", + "system", + "configurations", + "delays", + "create", + "managing", + "supply", + "This", + "result", + "volatility", + "quality", + "issues", + "increased", + "inventory", + "provisions", + "decreases", + "yields", + "higher", + "material", + "costs", + "warranty", + "Customers", + "postpone", + "purchasing", + "adopt", + "technologies", + "gradually", + "anticipated", + "affecting", + "timing", + "chain", + "expenses", + "April", + "2025", + "USG", + "informed", + "license", + "required", + "exports", + "H2O", + "China", + "market", + "As", + "requirements", + "incurred", + "4.5", + "billion", + "charge", + "excess", + "purchase", + "obligations", + "diminished", + "August", + "granted", + "licenses", + "ship", + "products", + "based", + "generated", + "approximately", + "60", + "million", + "February", + "small", + "amounts", + "H200", + "specific", + "To", + "date", + "licensing", + "program", + "imports", + "allowed", + "requires", + "H200s", + "inspection", + "United", + "States", + "prior", + "shipment", + "shipped", + "subject", + "25%", + "tariff", + "importation", + "recent", + "rise", + "high", + "open", + "source", + "foundation", + "models", + "making", + "capabilities", + "broadly", + "accessible", + "Open", + "dependent", + "developer", + "deployed", + "competitors", + "services", + "While", + "concentrated", + "Asia", + "expanding", + "Latin", + "America", + "moves", + "expected", + "strengthen", + "add", + "resiliency", + "redundancy", + "growing", + "ability", + "increase", + "manufacturing", + "depend", + "local", + "region", + "ecosystem", + "ramp", + "volume", + "timely", + "basis", + "made", + "expect", + "investments", + "technology", + "roadmap", + "broader", + "invested", + "17.5", + "private", + "funds", + "primarily", + "early", + "stage", + "startups", + "include", + "model", + "makers", + "directly", + "CSPs", + "Many", + "illiquid", + "marketable", + "related", + "profitable", + "term", + "assurance", + "realize", + "return", + "publicly", + "held", + "equity", + "securities", + "fluctuate", + "significantly", + "due", + "stock", + "prices", + "adversely", + "affect", + "results", + "build", + "datacenter", + "infrastructures", + "enter", + "commercial", + "arrangements", + "guarantees", + "provided", + "3.5", + "land", + "power", + "shell", + "generally", + "periods", + "If", + "escrow", + "operating", + "activities", + "sufficient", + "cover", + "event", + "default", + "elect", + "assume", + "underlying", + "leases", + "internal", + "sublease", + "parties", + "Macroeconomic", + "factors", + "tariffs", + "inflation", + "interest", + "rate", + "global", + "constraints", + "economic", + "geopolitical", + "developments", + "direct", + "indirect", + "impacts", + "operations", + "difficult", + "isolate", + "quantify", + "macroeconomic", + "employee", + "wages", + "equipment", + "competitive", + "position", + "solution", + "pricing", + "short", + "Within", + "continuously", + "manage", + "vendors", + "Refer", + "Item", + "1A", + "Risk", + "Factors", + "Risks", + "Related", + "Regulatory", + "Legal", + "Stock", + "Other", + "Matters", + "discussion", + "potential", + "business", + "Fiscal", + "Year", + "Summary", + "215.9", + "65%", + "ago", + "68%", + "strong", + "major", + "shifts", + "Gaming", + "41%", + "headwind", + "2027", + "Professional", + "Visualization", + "70%", + "exceptional", + "launch", + "DGX", + "Spark", + "Automotive", + "39%", + "continued", + "driving", + "Gross", + "margin", + "decreased", + "transitioned", + "offering", + "Hopper", + "HGX", + "systems", + "full", + "gross", + "decrease", + "impacted", + "Operating", + "compensation", + "benefits" + ], + "keywords": [ + "Revenue", + "Data Center", + "AI" + ], + "connect_to": [ + { + "target": "75708579-9e90-582b-8906-5788e0519c9d", + "relation": "embeds", + "ref": "[tables/table-64 Financial Results.html]", + "position": { + "start": 5778, + "end": 5818 + } + } + ] + } + }, + { + "chunk_id": "77b61e6c-1968-5ea2-8065-3c1f9023a366", + "type": "text", + "content": "Our consolidated financial statements are prepared in accordance with accounting principles generally accepted in the United States, or U.S. GAAP. The preparation of these financial statements requires us to make estimates and judgments that affect the reported amounts of assets, liabilities, revenue, cost of revenue, expenses and related disclosure of contingencies. Critical accounting estimates are those estimates that involve a significant level of estimation uncertainty and could have a material impact on our financial condition or results of operations. We have critical accounting estimates in the areas of inventories, income taxes, non-marketable equity securities, and revenue recognition. Refer to Note 1 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for a summary of significant accounting policies.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates", + "metadata": { + "length": 879, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "consolidated", + "financial", + "statements", + "prepared", + "accordance", + "accounting", + "principles", + "generally", + "accepted", + "United", + "States", + "GAAP", + "The", + "preparation", + "requires", + "make", + "estimates", + "judgments", + "affect", + "reported", + "amounts", + "assets", + "liabilities", + "revenue", + "cost", + "expenses", + "related", + "disclosure", + "contingencies", + "Critical", + "involve", + "significant", + "level", + "estimation", + "uncertainty", + "material", + "impact", + "condition", + "results", + "operations", + "We", + "critical", + "areas", + "inventories", + "income", + "taxes", + "marketable", + "equity", + "securities", + "recognition", + "Refer", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "summary", + "policies" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "063344a6-9d64-51cc-9a5e-f4f1bd0baeab", + "type": "text", + "content": "We charge cost of sales for inventory provisions to write-down our inventory to the lower of cost or net realizable value or for obsolete or excess inventory, and for excess product purchase commitments. Most of our inventory provisions relate to excess quantities of products or components, based on our inventory levels and future product purchase commitments compared to assumptions about future demand and market conditions, which requires management judgment.\nSituations that may result in excess or obsolete inventory or excess product purchase commitments include changes in business and economic conditions, changes in market conditions, sudden and significant decreases in demand for our products, including potential cancellation or deferral of customer purchase orders, inventory obsolescence because of changing technology and customer requirements, new product introductions resulting in less demand for existing products or inconsistent spikes in demand, failure to estimate customer demand properly, ordering in advance of historical lead-times, government regulations and the impact of changes in future demand, or increase in demand for competitive products, including competitive actions.\nThe net effect on our gross margin from inventory provisions and sales of items previously written down was an unfavorable impact of 2.6% in fiscal year 2026 and 2.3% in fiscal year 2025. Our inventory and capacity purchase commitments are based on forecasts of future customer demand and consider our third-party manufacturers' lead times and constraints. Our manufacturing lead times can be and have been long, and in some cases, extended beyond twelve months for some products. We may place non-cancellable inventory orders for certain product components in advance of our historical lead times, pay premiums and provide deposits to secure future supply and capacity. We also adjust to other market factors, such as product offerings and pricing actions by our competitors, new product transitions, and macroeconomic conditions - all of which may impact demand for our products.\nRefer to the Gross Profit and Gross Margin discussion below in this Management's Discussion and Analysis for further discussion.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Inventories", + "metadata": { + "length": 2217, + "summary": "The passage discusses inventory provisions for write-downs to lower of cost or net realizable value, and for obsolete or excess inventory. Provisions are driven by excess quantities based on inventory levels, future purchase commitments, and assumptions about demand and market conditions. Factors causing excess or obsolete inventory include economic changes, demand decreases, technology shifts, new product introductions, and competitive actions. The net impact on gross margin from provisions and sales of previously written-down items was unfavorable 2.6% in fiscal year 2026 and 2.3% in fiscal year 2025. Inventory and capacity purchase commitments rely on customer demand forecasts and third-party lead times, which can exceed twelve months. Non-cancellable orders, premiums, and deposits may be used to secure supply. Market factors like competitor actions, product transitions, and macroeconomic conditions also affect demand.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "charge", + "cost", + "sales", + "inventory", + "provisions", + "write", + "lower", + "net", + "realizable", + "obsolete", + "excess", + "product", + "purchase", + "commitments", + "Most", + "relate", + "quantities", + "products", + "components", + "based", + "levels", + "future", + "compared", + "assumptions", + "demand", + "market", + "conditions", + "requires", + "management", + "judgment", + "Situations", + "result", + "include", + "business", + "economic", + "sudden", + "significant", + "decreases", + "including", + "potential", + "cancellation", + "deferral", + "customer", + "orders", + "obsolescence", + "changing", + "technology", + "requirements", + "introductions", + "resulting", + "existing", + "inconsistent", + "spikes", + "failure", + "estimate", + "properly", + "ordering", + "advance", + "historical", + "lead", + "times", + "government", + "regulations", + "impact", + "increase", + "competitive", + "actions", + "The", + "effect", + "gross", + "margin", + "items", + "previously", + "written", + "unfavorable", + "2.6%", + "fiscal", + "year", + "2026", + "2.3%", + "2025", + "Our", + "capacity", + "forecasts", + "party", + "manufacturers", + "constraints", + "manufacturing", + "long", + "cases", + "extended", + "twelve", + "months", + "place", + "cancellable", + "pay", + "premiums", + "provide", + "deposits", + "secure", + "supply", + "adjust", + "factors", + "offerings", + "pricing", + "competitors", + "transitions", + "macroeconomic", + "Refer", + "Gross", + "Profit", + "Margin", + "discussion", + "Management", + "Discussion", + "Analysis" + ], + "keywords": [ + "inventory provisions", + "gross margin", + "demand forecasts" + ], + "connect_to": [] + } + }, + { + "chunk_id": "5d56a12e-92ef-5131-9421-991e7ef7f12a", + "type": "text", + "content": "We are subject to income taxes in the U.S. and foreign jurisdictions. Our calculation of deferred tax assets and liabilities is based on certain estimates and judgments and involves dealing with uncertainties in the application of complex tax laws. Our estimates of deferred tax assets and liabilities may change based, in part, on added certainty or finality to an anticipated outcome, changes in accounting standards or tax laws in the U.S. or foreign jurisdictions where we operate, or changes in other facts or circumstances. In addition, we recognize liabilities for potential U.S. and foreign income tax contingencies based on our estimate of whether, and the extent to which, additional taxes may be due. If we determine that payment of these amounts is unnecessary or if the recorded tax liability is less than our current assessment, we may be required to recognize an income tax benefit or additional income tax expense in our financial statements accordingly.\nWe record a valuation allowance to reduce deferred tax assets to the amount that is believed more likely than not to be realized based on all available evidence. To the extent realization of the deferred tax assets becomes more-likely-than-not, we would recognize such deferred tax assets as income tax benefits during the period.\nWe recognize the benefit from a tax position only if it is more-likely-than-not that the position would be sustained upon audit based solely on the technical merits of the tax position. Our policy is to include interest and penalties related to unrecognized tax benefits as a component of income tax expense.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Income Taxes", + "metadata": { + "length": 1610, + "summary": "The passage discusses the company's approach to income taxes in U.S. and foreign jurisdictions, focusing on deferred tax assets and liabilities based on estimates and judgments. It explains the use of a valuation allowance to reduce deferred tax assets to the amount more likely than not to be realized, and the recognition of tax benefits when realization becomes more-likely-than-not. Additionally, it covers the recognition of tax positions only if they are more-likely-than-not to be sustained upon audit, with interest and penalties included in income tax expense.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "subject", + "income", + "taxes", + "foreign", + "jurisdictions", + "Our", + "calculation", + "deferred", + "tax", + "assets", + "liabilities", + "based", + "estimates", + "judgments", + "involves", + "dealing", + "uncertainties", + "application", + "complex", + "laws", + "change", + "part", + "added", + "certainty", + "finality", + "anticipated", + "outcome", + "accounting", + "standards", + "operate", + "facts", + "circumstances", + "In", + "addition", + "recognize", + "potential", + "contingencies", + "estimate", + "extent", + "additional", + "due", + "If", + "determine", + "payment", + "amounts", + "unnecessary", + "recorded", + "liability", + "current", + "assessment", + "required", + "benefit", + "expense", + "financial", + "statements", + "record", + "valuation", + "allowance", + "reduce", + "amount", + "believed", + "realized", + "evidence", + "To", + "realization", + "benefits", + "period", + "position", + "sustained", + "audit", + "solely", + "technical", + "merits", + "policy", + "include", + "interest", + "penalties", + "related", + "unrecognized", + "component" + ], + "keywords": [ + "deferred tax assets", + "valuation allowance", + "tax contingencies" + ], + "connect_to": [] + } + }, + { + "chunk_id": "c25659cc-e0ec-5efc-ac03-0e7f743425be", + "type": "text", + "content": "Non-marketable equity securities consist of investments in private companies without readily determinable fair values. They are measured at cost minus impairment, if any, and are adjusted for observable price changes in orderly transactions for a similar investment in the same issuer (the measurement alternative). These adjustments may require use of unobservable inputs.\nWe assess impairment quarterly based on qualitative and quantitative factors, including the investee's operating performance and market trends.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Non-Marketable Equity Securities", + "metadata": { + "length": 517, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Non", + "marketable", + "equity", + "securities", + "consist", + "investments", + "private", + "companies", + "readily", + "determinable", + "fair", + "values", + "They", + "measured", + "cost", + "minus", + "impairment", + "adjusted", + "observable", + "price", + "orderly", + "transactions", + "similar", + "investment", + "issuer", + "measurement", + "alternative", + "These", + "adjustments", + "require", + "unobservable", + "inputs", + "We", + "assess", + "quarterly", + "based", + "qualitative", + "quantitative", + "factors", + "including", + "investee", + "operating", + "performance", + "market", + "trends" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e084a500-918c-565c-a2df-6701b49938fc", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Revenue Recognition", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7370e0e0-4852-5a38-bd12-18e0773a907d", + "type": "text", + "content": "For products sold with a right of return, we record a reduction to revenue by establishing a sales return allowance for estimated product returns at the time revenue is recognized, based primarily on historical return rates. However, if product returns for a fiscal period are anticipated to exceed historical return rates, we may determine that additional sales return allowances are required to reflect our estimated exposure for product returns. Return rights for certain stocking distributors for specific products are contractually limited based on a percentage of prior quarter shipments. For shipments to other customers, we do not allow returns, although we may approve returns for credit or refund based on applicable facts and circumstances.\nWe account for customer programs, which involve rebates and marketing development funds, or MDFs, as a reduction in revenue and accrue for such programs based on the amount we expect to be claimed by customers. Certain customer programs include distributor price incentives or other channel programs for specific products and customer classes which require judgment as to whether the applicable incentives will be attained. Estimates for customer program accruals include a combination of historical attainment and claim rates and may be adjusted based on relevant internal and external factors.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Revenue Recognition/Revenue Allowances", + "metadata": { + "length": 1347, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "For", + "products", + "sold", + "return", + "record", + "reduction", + "revenue", + "establishing", + "sales", + "allowance", + "estimated", + "product", + "returns", + "time", + "recognized", + "based", + "primarily", + "historical", + "rates", + "However", + "fiscal", + "period", + "anticipated", + "exceed", + "determine", + "additional", + "allowances", + "required", + "reflect", + "exposure", + "Return", + "rights", + "stocking", + "distributors", + "specific", + "contractually", + "limited", + "percentage", + "prior", + "quarter", + "shipments", + "customers", + "approve", + "credit", + "refund", + "applicable", + "facts", + "circumstances", + "We", + "account", + "customer", + "programs", + "involve", + "rebates", + "marketing", + "development", + "funds", + "MDFs", + "accrue", + "amount", + "expect", + "claimed", + "Certain", + "include", + "distributor", + "price", + "incentives", + "channel", + "classes", + "require", + "judgment", + "attained", + "Estimates", + "program", + "accruals", + "combination", + "attainment", + "claim", + "adjusted", + "relevant", + "internal", + "external", + "factors" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ea39dd55-b461-5dbf-b1a3-a6b5f18f6c78", + "type": "text", + "content": "Our contracts may contain more than one performance obligation. Judgment is required in determining whether each performance obligation within a customer contract is distinct. Except for License and Development Arrangements, NVIDIA products and services function on a standalone basis and do not require a significant amount of integration or interdependency. Therefore, multiple performance obligations contained within a customer contract are considered distinct and are not combined for revenue recognition purposes.\nWe allocate the total transaction price to each distinct performance obligation in an arrangement with multiple performance obligations on a relative standalone selling price basis. In most cases, we can establish standalone selling price based on directly observable prices of products or services sold separately in comparable circumstances to similar customers. If standalone selling price is not directly observable, such as when we do not sell a product or service separately, we determine standalone selling price based on market data and other observable inputs.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Revenue Recognition/Contracts with Multiple Performance Obligations", + "metadata": { + "length": 1089, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "contracts", + "performance", + "obligation", + "Judgment", + "required", + "determining", + "customer", + "contract", + "distinct", + "Except", + "License", + "Development", + "Arrangements", + "NVIDIA", + "products", + "services", + "function", + "standalone", + "basis", + "require", + "significant", + "amount", + "integration", + "interdependency", + "Therefore", + "multiple", + "obligations", + "contained", + "considered", + "combined", + "revenue", + "recognition", + "purposes", + "We", + "allocate", + "total", + "transaction", + "price", + "arrangement", + "relative", + "selling", + "In", + "cases", + "establish", + "based", + "directly", + "observable", + "prices", + "sold", + "separately", + "comparable", + "circumstances", + "similar", + "customers", + "If", + "sell", + "product", + "service", + "determine", + "market", + "data", + "inputs" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "84829ebd-beb2-5c2c-ad61-dd122ca52e4b", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025
Revenue100.0 %100.0 %
Cost of revenue28.925.0
Gross profit71.175.0
Operating expenses
Research and development8.69.9
Sales, general and administrative2.12.7
Total operating expenses10.712.6
Operating income60.462.4
Interest income1.11.4
Interest expense(0.1)(0.2)
Other income, net4.20.8
Total other income, net5.22.0
Income before income tax65.564.4
Income tax expense9.98.6
Net income55.6 %55.8 %
", + "path": "tables/table-65 Income Statement.html", + "metadata": { + "length": 1051, + "summary": "table-66\nThe table shows a company's income statement for years ending Jan 25, 2026 and Jan 26, 2025. Revenue is 100% both years. Gross profit decreased from 75.0% to 71.1%. Net income slightly dropped from 55.8% to 55.6%.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-65 Income Statement.html", + "keywords": [ + "Revenue", + "Gross Profit", + "Net Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "c756046c-4f01-5225-972d-8ce992dafe41", + "type": "text", + "content": "A discussion regarding our financial condition and results of operations for fiscal year 2026 compared to fiscal year 2025 is presented below. A discussion regarding our financial condition and results of operations for fiscal year 2025 compared to fiscal year 2024 can be found under Item 7 in our Annual Report on Form 10-K for the fiscal year ended January 26, 2025, filed with the SEC on February 26, 2025, which is available free of charge on the SEC's website at http://www.sec.gov and at our investor relations website, http://investor.nvidia.com.\nThe following table sets forth, for the periods indicated, certain items in our Consolidated Statements of Income expressed as a percentage of revenue.\n\n[tables/table-65 Income Statement.html]", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations", + "metadata": { + "length": 741, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "discussion", + "financial", + "condition", + "results", + "operations", + "fiscal", + "year", + "2026", + "compared", + "2025", + "presented", + "2024", + "found", + "Item", + "Annual", + "Report", + "Form", + "10", + "ended", + "January", + "26", + "filed", + "SEC", + "February", + "free", + "charge", + "website", + "http", + "www", + "sec", + "gov", + "investor", + "relations", + "nvidia", + "The", + "table", + "sets", + "periods", + "items", + "Consolidated", + "Statements", + "Income", + "expressed", + "percentage", + "revenue" + ], + "keywords": [], + "connect_to": [ + { + "target": "84829ebd-beb2-5c2c-ad61-dd122ca52e4b", + "relation": "embeds", + "ref": "[tables/table-65 Income Statement.html]", + "position": { + "start": 708, + "end": 747 + } + } + ] + } + }, + { + "chunk_id": "15861a47-8ed7-57df-b6fd-113a1dd294af", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Reportable Segments", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a0dea15d-65ce-580c-af26-17254a3cbc3b", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025$ Change% Change
($ in millions)
Compute & Networking$ 193,479$ 116,193$ 77,28667 %
Graphics22,45914,3048,15557 %
Total$ 215,938$ 130,497$ 85,44165 %
", + "path": "tables/table-66 Revenue Growth.html", + "metadata": { + "length": 486, + "summary": "table-67\nThe table shows revenue for Compute & Networking and Graphics segments for years ended Jan 25, 2026 and Jan 26, 2025, with total revenue increasing 65% to $215,938 million.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-66 Revenue Growth.html", + "keywords": [ + "Revenue", + "Compute & Networking", + "Graphics" + ], + "tokens": [] + } + }, + { + "chunk_id": "b8be8e95-1677-5d69-b4af-279ff94ad152", + "type": "text", + "content": "[tables/table-66 Revenue Growth.html]", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Reportable Segments/Revenue by Reportable Segments", + "metadata": { + "length": 33, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [ + { + "target": "a0dea15d-65ce-580c-af26-17254a3cbc3b", + "relation": "embeds", + "ref": "[tables/table-66 Revenue Growth.html]", + "position": { + "start": 0, + "end": 37 + } + } + ] + } + }, + { + "chunk_id": "5a1403a8-efd3-564d-b1d9-cafbed87e667", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025$ Change% Change
($ in millions)
Compute & Networking$ 130,141$ 82,875$ 47,26657 %
Graphics9,1565,0854,07180 %
Total$ 139,297$ 87,960$ 51,33758 %
", + "path": "tables/table-67 Revenue Growth.html", + "metadata": { + "length": 482, + "summary": "table-68\nThe table shows revenue for Compute & Networking and Graphics segments for years ending Jan 2026 and 2025, with total revenue increasing 58% to $139,297 million.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-67 Revenue Growth.html", + "keywords": [ + "Compute & Networking", + "Graphics", + "Revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "db8b95b3-6b6f-5fa9-842a-2dcc813c700a", + "type": "text", + "content": "[tables/table-67 Revenue Growth.html]\n\nCompute & Networking revenue – The year over year increase was driven by the major platform shifts – accelerated computing and AI. Revenue from Data Center computing grew 59% driven by demand for our Blackwell computing platform. Revenue from Data Center networking grew 142% driven by the introduction and continued ramp of NVLink compute fabric for GB200 and GB300 systems and the growth of Ethernet and InfiniBand platforms.\nGraphics revenue – The year over year increase was driven by sales of our Blackwell architecture.\nReportable segment operating income – The year over year increase in Compute & Networking segment operating income was driven by growth in revenue, partially offset by a \\$4.5 billion charge associated with H2O excess inventory and purchase obligations in the first quarter of fiscal year 2026. The year over year increase in Graphics segment operating income was driven by the growth in revenue.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Reportable Segments/Operating Income by Reportable Segments", + "metadata": { + "length": 957, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Compute", + "Networking", + "revenue", + "The", + "year", + "increase", + "driven", + "major", + "platform", + "shifts", + "accelerated", + "computing", + "AI", + "Revenue", + "Data", + "Center", + "grew", + "59%", + "demand", + "Blackwell", + "networking", + "142%", + "introduction", + "continued", + "ramp", + "NVLink", + "compute", + "fabric", + "GB200", + "GB300", + "systems", + "growth", + "Ethernet", + "InfiniBand", + "platforms", + "Graphics", + "sales", + "architecture", + "Reportable", + "segment", + "operating", + "income", + "partially", + "offset", + "4.5", + "billion", + "charge", + "H2O", + "excess", + "inventory", + "purchase", + "obligations", + "quarter", + "fiscal", + "2026" + ], + "keywords": [], + "connect_to": [ + { + "target": "5a1403a8-efd3-564d-b1d9-cafbed87e667", + "relation": "embeds", + "ref": "[tables/table-67 Revenue Growth.html]", + "position": { + "start": 0, + "end": 37 + } + } + ] + } + }, + { + "chunk_id": "f2cce4cc-89c5-5e3d-a650-a13983a7f38b", + "type": "text", + "content": "We refer to customers who purchase products directly from NVIDIA as direct customers, such as AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. We refer to indirect customers as those who purchase products through our direct customers; indirect customers include CSPs, Neocloud builders, AI model makers, enterprises, and public sector entities. Our revenue is concentrated among a limited number of direct and indirect customers and this trend may continue.\nDirect Customers – For fiscal year 2026, sales to one direct customer represented 22% of total revenue and sales to another direct customer represented 14% of total revenue, all of which were primarily attributable to the Compute & Networking segment.\nFor fiscal year 2025, sales to one direct customer represented 12% of total revenue and sales to two direct customers each represented 11% of total revenue, all of which were primarily attributable to the Compute & Networking segment.\nFor fiscal year 2024, sales to one direct customer represented 13% of total revenue, and were primarily attributable to the Compute & Networking segment.\nIndirect Customers – Indirect customer revenue is an estimation based upon multiple factors including customer purchase order information, product specifications, internal sales data, and other sources. Indirect customers primarily purchase our products through system integrators and distributors. We generate a significant amount of our revenue from a limited number of indirect customers, some individually representing 10% or more of our revenue. Certain companies purchase cloud and related services through various direct and indirect customers. We estimate that one AI research and deployment company contributed to a meaningful amount of our revenue purchasing cloud services from our customers in fiscal year 2026.\nRevenue by geographic region is designated based on the location of the headquarters of direct customers. The end customer and shipping location may be different from our customers' headquarters location. Revenue from sales to customers headquartered outside of the United States accounted for 31% and 41% of total revenue for fiscal years 2026 and 2025, respectively.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Reportable Segments/Concentration of Revenue", + "metadata": { + "length": 2339, + "summary": "NVIDIA's revenue is concentrated among a limited number of direct and indirect customers. Direct customers include AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. In fiscal year 2026, one direct customer accounted for 22% of total revenue and another for 14%, primarily from the Compute & Networking segment. Indirect customers purchase through direct customers, including CSPs, Neocloud builders, and enterprises. One AI research and deployment company contributed meaningfully to revenue via cloud services in fiscal year 2026. Revenue from customers headquartered outside the US was 31% in fiscal year 2026 and 41% in fiscal year 2025.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "refer", + "customers", + "purchase", + "products", + "directly", + "NVIDIA", + "direct", + "AIBs", + "distributors", + "ODMs", + "OEMs", + "CSPs", + "AI", + "model", + "makers", + "system", + "integrators", + "Certain", + "internal", + "resources", + "party", + "complete", + "build", + "indirect", + "include", + "Neocloud", + "builders", + "enterprises", + "public", + "sector", + "entities", + "Our", + "revenue", + "concentrated", + "limited", + "number", + "trend", + "continue", + "Direct", + "Customers", + "For", + "fiscal", + "year", + "2026", + "sales", + "customer", + "represented", + "22%", + "total", + "14%", + "primarily", + "attributable", + "Compute", + "Networking", + "segment", + "2025", + "12%", + "11%", + "2024", + "13%", + "Indirect", + "estimation", + "based", + "multiple", + "factors", + "including", + "order", + "information", + "product", + "specifications", + "data", + "sources", + "generate", + "significant", + "amount", + "individually", + "representing", + "10%", + "companies", + "cloud", + "related", + "services", + "estimate", + "research", + "deployment", + "company", + "contributed", + "meaningful", + "purchasing", + "Revenue", + "geographic", + "region", + "designated", + "location", + "headquarters", + "The", + "end", + "shipping", + "headquartered", + "United", + "States", + "accounted", + "31%", + "41%", + "years" + ], + "keywords": [ + "Direct customers", + "Indirect customers", + "Revenue concentration" + ], + "connect_to": [] + } + }, + { + "chunk_id": "a95a47f5-d369-503e-9223-b6dec2234184", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025$ Change% Change
($ in millions)
Research and development$ 18,497$ 12,914$ 5,58343 %
Sales, general and administrative4,5793,4911,08831 %
Total operating expenses$ 23,076$ 16,405$ 6,67141 %
", + "path": "tables/table-68 Operating Expenses.html", + "metadata": { + "length": 522, + "summary": "table-69\nThe table shows operating expenses for years ended Jan 25, 2026 and Jan 26, 2025. R&D increased by $5,583M (43%), SG&A by $1,088M (31%), total operating expenses by $6,671M (41%).", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-68 Operating Expenses.html", + "keywords": [ + "R&D", + "SG&A", + "Operating Expenses" + ], + "tokens": [] + } + }, + { + "chunk_id": "283e02c2-5bfc-5652-a56b-290d1939a24d", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025$ Change
($ in millions)
Interest income$ 2,300$ 1,786$ 514
Interest expense(259)(247)(12)
Other income, net9,0221,0347,988
Total other income, net$ 11,063$ 2,573$ 8,490
", + "path": "tables/table-69 Other Income.html", + "metadata": { + "length": 510, + "summary": "table-70\nThe table shows other income net for years ending Jan 2026 and 2025, with interest income rising from $1,786M to $2,300M and other income surging from $1,034M to $9,022M, resulting in total other income net of $11,063M.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-69 Other Income.html", + "keywords": [ + "Interest", + "Other Income", + "Net" + ], + "tokens": [] + } + }, + { + "chunk_id": "e28cb0d6-02d6-5764-aa90-bb7a45669641", + "type": "text", + "content": "Gross profit consists of total net revenue less cost of revenue. Cost of revenue consists of the cost of semiconductors, including wafer fabrication, assembly, testing and packaging, board and device costs, manufacturing support costs, including labor and overhead associated with such purchases, final test yield fallout, inventory and warranty provisions, memory and component costs, tariffs, and shipping costs. Cost of revenue also includes acquisition-related intangible amortization expense, costs for license and development and service arrangements, IP-related costs, and stock-based compensation related to personnel associated with manufacturing operations.\nGross margins decreased to 71.1% in fiscal year 2026 from 75.0% in fiscal year 2025 as our business model transitioned from offering Hopper HGX systems to Blackwell full-scale datacenter solutions and a \\$4.5 billion charge associated with H2O excess inventory and purchase obligations in the first quarter of fiscal year 2026.\nProvisions for inventory and excess inventory purchase obligations totaled \\$7.2 billion and \\$3.7 billion for fiscal years 2026 and 2025, respectively, including \\$4.5 billion associated with H2O excess inventory and purchase obligations for the first quarter of fiscal year 2026. Sales of previously reserved inventory and settlements of excess inventory purchase obligations resulted in a provision release of \\$1.5 billion and \\$689 million for fiscal years 2026 and 2025, respectively. The net effect on our gross margin was an unfavorable impact of 2.6% and 2.3% in fiscal years 2026 and 2025, respectively.\nOperating Expenses\n\n[tables/table-68 Operating Expenses.html]\n\nThe increases in research and development expenses for fiscal year 2026 were driven by a 29% increase in compensation and benefits expense, including stock-based compensation, reflecting employee growth and compensation increases and a 79% increase in compute and infrastructure.\nThe increases in sales, general and administrative expenses for fiscal year 2026 were primarily driven by compensation and benefits expense, including stock-based compensation, reflecting employee growth and compensation increases.\nTotal Other Income, Net\n\n[tables/table-69 Other Income.html]\n\nThe increase in interest income was primarily due to growth in cash, cash equivalents, and debt securities.\nInterest expense is primarily comprised of coupon interest and debt discount amortization related to our notes.\nOther income, net primarily consists of realized or unrealized gains and losses from investments in non-marketable equity securities, publicly-held equity securities, and the impact of changes in foreign currency rates. The change in Other income, net, compared to fiscal year 2025, was primarily driven by unrealized gains in non-marketable and publicly-held equity securities, including gains from our previously announced investment in Intel's common stock. Refer to Note 7 and 8 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information regarding our non-marketable equity securities.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Gross Profit and Gross Margin", + "metadata": { + "length": 3123, + "summary": "Gross profit is net revenue minus cost of revenue, which includes semiconductor costs, manufacturing support, acquisition-related amortization, and stock-based compensation. Gross margins fell from 75.0% in fiscal 2025 to 71.1% in fiscal 2026 due to a business model shift to Blackwell datacenter solutions and a $4.5 billion charge for H2O excess inventory. Inventory provisions totaled $7.2 billion in 2026 versus $3.7 billion in 2025, with a net unfavorable gross margin impact of 2.6% and 2.3% respectively. Operating expenses rose, with R&D driven by a 29% increase in compensation and a 79% increase in compute infrastructure, and SG&A increases from employee growth. Other income increased due to interest income growth and unrealized gains on equity securities, including Intel stock.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Gross", + "profit", + "consists", + "total", + "net", + "revenue", + "cost", + "Cost", + "semiconductors", + "including", + "wafer", + "fabrication", + "assembly", + "testing", + "packaging", + "board", + "device", + "costs", + "manufacturing", + "support", + "labor", + "overhead", + "purchases", + "final", + "test", + "yield", + "fallout", + "inventory", + "warranty", + "provisions", + "memory", + "component", + "tariffs", + "shipping", + "includes", + "acquisition", + "related", + "intangible", + "amortization", + "expense", + "license", + "development", + "service", + "arrangements", + "IP", + "stock", + "based", + "compensation", + "personnel", + "operations", + "margins", + "decreased", + "71.1%", + "fiscal", + "year", + "2026", + "75.0%", + "2025", + "business", + "model", + "transitioned", + "offering", + "Hopper", + "HGX", + "systems", + "Blackwell", + "full", + "scale", + "datacenter", + "solutions", + "4.5", + "billion", + "charge", + "H2O", + "excess", + "purchase", + "obligations", + "quarter", + "Provisions", + "totaled", + "7.2", + "3.7", + "years", + "Sales", + "previously", + "reserved", + "settlements", + "resulted", + "provision", + "release", + "1.5", + "689", + "million", + "The", + "effect", + "gross", + "margin", + "unfavorable", + "impact", + "2.6%", + "2.3%", + "Operating", + "Expenses", + "increases", + "research", + "expenses", + "driven", + "29%", + "increase", + "benefits", + "reflecting", + "employee", + "growth", + "79%", + "compute", + "infrastructure", + "sales", + "general", + "administrative", + "primarily", + "Total", + "Other", + "Income", + "Net", + "interest", + "income", + "due", + "cash", + "equivalents", + "debt", + "securities", + "Interest", + "comprised", + "coupon", + "discount", + "notes", + "realized", + "unrealized", + "gains", + "losses", + "investments", + "marketable", + "equity", + "publicly", + "held", + "foreign", + "currency", + "rates", + "change", + "compared", + "announced", + "investment", + "Intel", + "common", + "Refer", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "additional", + "information" + ], + "keywords": [ + "Gross margin", + "Inventory provisions", + "Operating expenses" + ], + "connect_to": [ + { + "target": "a95a47f5-d369-503e-9223-b6dec2234184", + "relation": "embeds", + "ref": "[tables/table-68 Operating Expenses.html]", + "position": { + "start": 1630, + "end": 1671 + } + }, + { + "target": "283e02c2-5bfc-5652-a56b-290d1939a24d", + "relation": "embeds", + "ref": "[tables/table-69 Other Income.html]", + "position": { + "start": 2210, + "end": 2245 + } + } + ] + } + }, + { + "chunk_id": "2801a169-ac8e-5c6d-92f5-0590e951d16a", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
(In millions)
Cash and cash equivalents$ 10,605$ 8,589
Marketable securities51,95134,621
Cash, cash equivalents, and marketable securities$ 62,556$ 43,210
", + "path": "tables/table-70 Cash & Securities.html", + "metadata": { + "length": 376, + "summary": "table-71\nThe table shows cash and cash equivalents, marketable securities, and their total for Jan 25, 2026 and Jan 26, 2025, with increases in all categories.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-70 Cash & Securities.html", + "keywords": [ + "cash", + "marketable securities", + "financial data" + ], + "tokens": [] + } + }, + { + "chunk_id": "659b141f-d15f-558b-af89-4fcf1b1fb527", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
(In millions)
Net cash provided by operating activities$ 102,718$ 64,089
Net cash used in investing activities$ (52,228)$ (20,421)
Net cash used in financing activities$ (48,474)$ (42,359)
", + "path": "tables/table-71 Cash Flow Table.html", + "metadata": { + "length": 407, + "summary": "table-72\nThe table shows net cash provided by operating activities of $102,718M (Jan 2026) vs $64,089M (Jan 2025), net cash used in investing of $52,228M vs $20,421M, and financing of $48,474M vs $42,359M.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-71 Cash Flow Table.html", + "keywords": [ + "cash flow", + "operating", + "investing" + ], + "tokens": [] + } + }, + { + "chunk_id": "045a5491-38db-53e8-a692-7a2cae666739", + "type": "text", + "content": "Income tax expense was \\$21.4 billion and \\$11.1 billion for fiscal years 2026 and 2025, respectively. Income tax as a percentage of income before income tax was an expense of 15.1% and 13.3% for fiscal years 2026 and 2025, respectively.\nThe effective tax rate increased primarily due to a lower percentage of tax benefits from stock-based compensation, FDDEI, and U.S. federal research tax credit relative to the increase in income before income tax.\nOur effective tax rates for fiscal years 2026 and 2025 were lower than the U.S. federal statutory rate of 21.0% primarily due to tax benefits from FDDEI, stock-based compensation, income earned in jurisdictions that are subject to taxes at rates lower than the U.S. federal statutory tax rate, and the U.S. federal research tax credit.\nIn July 2025, the One Big Beautiful Bill Act (OBBBA) was enacted into law and contains several changes to key U.S. federal income tax laws. We have recognized the tax effects of currently effective OBBBA provisions in our results for fiscal year 2026. We will continue to evaluate the impact of these legislative changes as tax authorities provide additional guidance and interpretation.\nAs of January 25, 2026, based on recent jurisdictional taxable income and expected future earnings, we concluded certain state deferred tax assets are more likely than not realizable and released \\$711 million of valuation allowance.\nRefer to Note 13 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information.\nLiquidity and Capital Resources\n\n[tables/table-70 Cash & Securities.html]\n\nYear Ended\n\n[tables/table-71 Cash Flow Table.html]\n\nOur fixed income security investments include highly rated, diversified investment types and credit exposures with shorter maturities.\nCash provided by operating activities increased in fiscal year 2026 compared to fiscal year 2025, due to higher revenue.\nCash used in investing activities increased in fiscal year 2026 compared to fiscal year 2025, primarily driven by higher purchases of equity investment securities and the execution of a non-exclusive license agreement with Groq.\nCash used in financing activities increased in fiscal year 2026 compared to fiscal year 2025, mainly due to higher share repurchases.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Income Taxes", + "metadata": { + "length": 2327, + "summary": "Income tax expense was $21.4 billion in fiscal 2026, up from $11.1 billion in fiscal 2025, with effective tax rates of 15.1% and 13.3%, respectively. The increase was due to lower tax benefits from stock-based compensation, FDDEI, and research credits. Rates remained below the 21.0% U.S. statutory rate due to these benefits and lower foreign taxes. The One Big Beautiful Bill Act (OBBBA) enacted in July 2025 impacted fiscal 2026 results. A $711 million valuation allowance was released on state deferred tax assets. Cash from operations increased due to higher revenue, while investing activities rose from equity purchases and a Groq license agreement. Financing activities increased due to higher share repurchases.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Income", + "tax", + "expense", + "21.4", + "billion", + "11.1", + "fiscal", + "years", + "2026", + "2025", + "percentage", + "income", + "15.1%", + "13.3%", + "The", + "effective", + "rate", + "increased", + "primarily", + "due", + "lower", + "benefits", + "stock", + "based", + "compensation", + "FDDEI", + "federal", + "research", + "credit", + "relative", + "increase", + "Our", + "rates", + "statutory", + "21.0%", + "earned", + "jurisdictions", + "subject", + "taxes", + "In", + "July", + "One", + "Big", + "Beautiful", + "Bill", + "Act", + "OBBBA", + "enacted", + "law", + "key", + "laws", + "We", + "recognized", + "effects", + "provisions", + "results", + "year", + "continue", + "evaluate", + "impact", + "legislative", + "authorities", + "provide", + "additional", + "guidance", + "interpretation", + "As", + "January", + "25", + "recent", + "jurisdictional", + "taxable", + "expected", + "future", + "earnings", + "concluded", + "state", + "deferred", + "assets", + "realizable", + "released", + "711", + "million", + "valuation", + "allowance", + "Refer", + "Note", + "13", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "information", + "Liquidity", + "Capital", + "Resources", + "Year", + "Ended", + "fixed", + "security", + "investments", + "include", + "highly", + "rated", + "diversified", + "investment", + "types", + "exposures", + "shorter", + "maturities", + "Cash", + "provided", + "operating", + "activities", + "compared", + "higher", + "revenue", + "investing", + "driven", + "purchases", + "equity", + "securities", + "execution", + "exclusive", + "license", + "agreement", + "Groq", + "financing", + "share", + "repurchases" + ], + "keywords": [ + "Income tax", + "Effective tax rate", + "Liquidity" + ], + "connect_to": [ + { + "target": "2801a169-ac8e-5c6d-92f5-0590e951d16a", + "relation": "embeds", + "ref": "[tables/table-70 Cash & Securities.html]", + "position": { + "start": 1597, + "end": 1637 + } + }, + { + "target": "659b141f-d15f-558b-af89-4fcf1b1fb527", + "relation": "embeds", + "ref": "[tables/table-71 Cash Flow Table.html]", + "position": { + "start": 1651, + "end": 1689 + } + } + ] + } + }, + { + "chunk_id": "38e86e3d-7cb0-52f2-853c-42192c44db90", + "type": "text", + "content": "Our primary sources of liquidity include cash, cash equivalents, marketable securities, and cash generated by our operations. As of January 25, 2026, we had \\$62.6 billion in cash, cash equivalents, and marketable securities. We believe that we have sufficient liquidity to meet our operating requirements for at least the next twelve months and for the foreseeable future, including our future obligations. We continuously evaluate our liquidity and capital resources, including our access to external capital, to ensure we can finance future capital requirements.\nOur marketable securities consist of publicly-held equity securities, debt securities issued by the U.S. government and its agencies, highly-rated corporations and financial institutions, and foreign government entities, as well as certificates of deposit issued by highly-rated financial institutions. These marketable securities are primarily denominated in U.S. dollars. Refer to Note 7 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information.\nExcept for approximately \\$1.7 billion of cash, cash equivalents, and marketable securities held outside the U.S. for which we have not accrued any related foreign or state taxes if we repatriate these amounts to the U.S., substantially all of our cash, cash equivalents and marketable securities held outside the U.S. at the end of fiscal year 2026 are available for use in the U.S. without incurring additional U.S. federal income taxes. Our first quarter of any fiscal year (including fiscal year 2027) generally does not include any estimated federal and state income tax payments and our second quarter of any fiscal year (including fiscal year 2027) generally includes two estimated federal and state income tax payments.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Liquidity", + "metadata": { + "length": 1820, + "summary": "The passage discusses the company's primary liquidity sources including cash, cash equivalents, and marketable securities totaling $62.6 billion as of January 25, 2026. It states sufficient liquidity for at least twelve months and the foreseeable future. Marketable securities consist of U.S. government debt, corporate bonds, and foreign government securities, primarily in U.S. dollars. Except for $1.7 billion held outside the U.S. with no accrued taxes, most foreign holdings are available for U.S. use without additional federal taxes. The first quarter of fiscal year 2027 includes no estimated tax payments, while the second quarter includes two.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "primary", + "sources", + "liquidity", + "include", + "cash", + "equivalents", + "marketable", + "securities", + "generated", + "operations", + "As", + "January", + "25", + "2026", + "62.6", + "billion", + "We", + "sufficient", + "meet", + "operating", + "requirements", + "twelve", + "months", + "foreseeable", + "future", + "including", + "obligations", + "continuously", + "evaluate", + "capital", + "resources", + "access", + "external", + "ensure", + "finance", + "consist", + "publicly", + "held", + "equity", + "debt", + "issued", + "government", + "agencies", + "highly", + "rated", + "corporations", + "financial", + "institutions", + "foreign", + "entities", + "certificates", + "deposit", + "These", + "primarily", + "denominated", + "dollars", + "Refer", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "additional", + "information", + "Except", + "approximately", + "1.7", + "accrued", + "related", + "state", + "taxes", + "repatriate", + "amounts", + "substantially", + "end", + "fiscal", + "year", + "incurring", + "federal", + "income", + "quarter", + "2027", + "generally", + "estimated", + "tax", + "payments", + "includes" + ], + "keywords": [ + "Liquidity", + "Marketable Securities", + "Cash" + ], + "connect_to": [] + } + }, + { + "chunk_id": "ba140363-6d5e-5c16-b4e9-2d7d91bbeb27", + "type": "text", + "content": "On August 26, 2025, our Board of Directors approved an additional \\$60.0 billion in share repurchase authorization, without expiration. In fiscal year 2026, we repurchased 282 million shares of our common stock for \\$40.4 billion. As of January 25, 2026, we were authorized, subject to certain specifications, to repurchase up to \\$58.5 billion of our common stock.\nFrom January 26, 2026 through February 20, 2026, we repurchased 8 million shares for \\$1.5 billion pursuant to a pre-established trading plan. We may execute repurchases from time to time, subject to market conditions, operating requirements, and other investment opportunities, in the open market, in privately negotiated transactions, pursuant to a Rule 10b5-1 trading plan or in structured share repurchase agreements in compliance with Rule 10b-18 of the Exchange Act. Our share repurchase program may be suspended at any time at our discretion.\nIn fiscal year 2026, we paid cash dividends to our shareholders of \\$974 million. The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders.\nThe U.S. Inflation Reduction Act of 2022 requires a 1% excise tax on certain share repurchases in excess of shares issued for employee compensation made after December 31, 2022. The excise tax is included in our share repurchase cost and was not material for fiscal years 2026 and 2025.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Liquidity/Capital Return to Shareholders", + "metadata": { + "length": 1464, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "On", + "August", + "26", + "2025", + "Board", + "Directors", + "approved", + "additional", + "60.0", + "billion", + "share", + "repurchase", + "authorization", + "expiration", + "In", + "fiscal", + "year", + "2026", + "repurchased", + "282", + "million", + "shares", + "common", + "stock", + "40.4", + "As", + "January", + "25", + "authorized", + "subject", + "specifications", + "58.5", + "From", + "February", + "20", + "1.5", + "pursuant", + "pre", + "established", + "trading", + "plan", + "We", + "execute", + "repurchases", + "time", + "market", + "conditions", + "operating", + "requirements", + "investment", + "opportunities", + "open", + "privately", + "negotiated", + "transactions", + "Rule", + "10b5", + "structured", + "agreements", + "compliance", + "10b", + "18", + "Exchange", + "Act", + "Our", + "program", + "suspended", + "discretion", + "paid", + "cash", + "dividends", + "shareholders", + "974", + "The", + "payment", + "future", + "continuing", + "determination", + "declaration", + "interests", + "Inflation", + "Reduction", + "2022", + "requires", + "1%", + "excise", + "tax", + "excess", + "issued", + "employee", + "compensation", + "made", + "December", + "31", + "included", + "cost", + "material", + "years" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6734fda2-c9be-5b02-880c-72a7f4381d5e", + "type": "table", + "content": "
Jan 25, 2026
(In millions)
Due in one year$ 1,000
Due in one to five years2,750
Due in five to ten years1,250
Due in greater than ten years3,500
Unamortized debt discount and issuance costs(32)
Net carrying amount$ 8,468
Less short-term portion999
Total long-term portion$ 7,469
", + "path": "tables/table-72 Debt Maturity.html", + "metadata": { + "length": 557, + "summary": "table-73\nThe table shows debt maturity schedule as of Jan 25, 2026, with $8,468 million net carrying amount, including $999 million short-term and $7,469 million long-term portion.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-72 Debt Maturity.html", + "keywords": [ + "Debt", + "Maturity", + "Carrying" + ], + "tokens": [] + } + }, + { + "chunk_id": "a4f674e3-3c7e-5a39-b3a6-e7b4018959a8", + "type": "text", + "content": "Our aggregate debt maturities as of January 25, 2026, by year payable, are as follows:\n\n[tables/table-72 Debt Maturity.html]\n\nIn January 2026, we increased the amount of our commercial paper program, pursuant to which we may issue unsecured commercial paper notes from time to time or all at once up to \\$25.0 billion. As of January 25, 2026, no commercial paper was outstanding. We will continue to evaluate issuing commercial paper as a component of our overall liquidity strategy.\nRefer to Note 11 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for further discussion.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Liquidity/Outstanding Indebtedness and Commercial Paper Program", + "metadata": { + "length": 632, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "aggregate", + "debt", + "maturities", + "January", + "25", + "2026", + "year", + "payable", + "In", + "increased", + "amount", + "commercial", + "paper", + "program", + "pursuant", + "issue", + "unsecured", + "notes", + "time", + "25.0", + "billion", + "As", + "outstanding", + "We", + "continue", + "evaluate", + "issuing", + "component", + "liquidity", + "strategy", + "Refer", + "Note", + "11", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "discussion" + ], + "keywords": [], + "connect_to": [ + { + "target": "6734fda2-c9be-5b02-880c-72a7f4381d5e", + "relation": "embeds", + "ref": "[tables/table-72 Debt Maturity.html]", + "position": { + "start": 88, + "end": 124 + } + } + ] + } + }, + { + "chunk_id": "2ec313d7-5bd4-5c5b-a186-a81673937432", + "type": "text", + "content": "For descriptions of our facility lease guarantees, long-term debt, purchase commitments, and operating lease obligations, refer to Note 10, Note 11, Note 12, and Note 17 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K, respectively.\nWe expect to continue investing in our ecosystem. We are finalizing an investment and partnership agreement with OpenAI. There is no assurance that we will enter into an investment and partnership agreement with OpenAI or that a transaction will be completed. Refer to Item 1A. Risk Factors for additional information regarding our investments.\nDuring fiscal year 2026 and fiscal year 2025, we spent \\$6.1 billion and \\$3.4 billion on capital expenditures, respectively. We expect to increase capital expenditures in fiscal year 2027 relative to fiscal year 2026 to support the future growth of our business.\nUnrecognized tax benefits of \\$4.0 billion, which includes related interest and penalties of \\$374 million, were recorded in non-current income tax payable at the end of fiscal year 2026. We are unable to estimate the timing of any potential tax liability, interest payments, or penalties in individual years due to uncertainties in the underlying income tax positions and the timing of the effective settlement of such tax positions. We are currently under examination by the Internal Revenue Service for our fiscal years 2023 and 2024. Refer to Note 13 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for further information.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Liquidity/Material Cash Requirements and Other Obligations", + "metadata": { + "length": 1591, + "summary": "The passage discusses facility lease guarantees, long-term debt, purchase commitments, and operating lease obligations referenced in various notes to financial statements. It mentions ongoing investment in the ecosystem and a potential partnership with OpenAI, with no assurance of completion. Capital expenditures were $6.1 billion in fiscal 2026 and $3.4 billion in fiscal 2025, with an expected increase in fiscal 2027. Unrecognized tax benefits of $4.0 billion, including $374 million in interest and penalties, are recorded, with uncertain timing of potential tax liabilities. The company is under IRS examination for fiscal years 2023 and 2024.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "For", + "descriptions", + "facility", + "lease", + "guarantees", + "long", + "term", + "debt", + "purchase", + "commitments", + "operating", + "obligations", + "refer", + "Note", + "10", + "11", + "12", + "17", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "We", + "expect", + "continue", + "investing", + "ecosystem", + "finalizing", + "investment", + "partnership", + "agreement", + "OpenAI", + "There", + "assurance", + "enter", + "transaction", + "completed", + "Refer", + "1A", + "Risk", + "Factors", + "additional", + "information", + "investments", + "During", + "fiscal", + "year", + "2026", + "2025", + "spent", + "6.1", + "billion", + "3.4", + "capital", + "expenditures", + "increase", + "2027", + "relative", + "support", + "future", + "growth", + "business", + "Unrecognized", + "tax", + "benefits", + "4.0", + "includes", + "related", + "interest", + "penalties", + "374", + "million", + "recorded", + "current", + "income", + "payable", + "end", + "unable", + "estimate", + "timing", + "potential", + "liability", + "payments", + "individual", + "years", + "due", + "uncertainties", + "underlying", + "positions", + "effective", + "settlement", + "examination", + "Internal", + "Revenue", + "Service", + "2023", + "2024", + "13" + ], + "keywords": [ + "capital expenditures", + "tax benefits", + "investment" + ], + "connect_to": [] + } + }, + { + "chunk_id": "d1bfdca6-bd2b-5cd7-be72-a25856318f34", + "type": "text", + "content": "Refer to Note 1 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for a discussion of adoption of new and recently issued accounting pronouncements.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Adoption of New and Recently Issued Accounting Pronouncements", + "metadata": { + "length": 207, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Refer", + "Note", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "discussion", + "adoption", + "recently", + "issued", + "accounting", + "pronouncements" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d93b13db-0b51-5f3e-83fc-2b15b2b59f55", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 7A. Quantitative and Qualitative Disclosures about Market Risk", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "07621171-16eb-5452-88d8-327c2cbc25ca", + "type": "text", + "content": "We are exposed to interest rate risk related to our fixed-rate investment portfolio and outstanding debt. The investment portfolio is managed consistent with our overall liquidity strategy in support of both working capital needs and growth of our businesses.\nAccording to our sensitivity analysis on our investment portfolio, a decrease in the yield curve of 0.5% as of the end of fiscal year 2026 and 2025 would decrease the fair value for these investments by approximately \\$0.2 billion.\nAs of the end of fiscal year 2026, we had \\$8.5 billion of senior Notes outstanding. We carry the Notes at face value less unamortized discount on our Consolidated Balance Sheets. As the Notes bear interest at a fixed rate, we have no financial statement risk associated with changes in interest rates. Refer to Note 11 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information.\nPublicly-held equity securities are subject to market price volatility. A hypothetical 10% decrease in our publicly-held equity securities would decrease the fair value of the publicly-held equity securities balance by \\$1.8 billion and an insignificant amount as of January 25, 2026 and January 26, 2025, respectively.\nNon-marketable equity securities are measured based on cost minus impairment, if any, and are adjusted for observable price changes in orderly transactions for an identical or similar investment in the same issuer. Valuations of our non-marketable equity securities are inherently complex due to the lack of readily available market data and observable transactions, and impact of macroeconomic factors.\nFor a description of our equity investments, refer to Notes 7 and 8 of the Notes to Condensed Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 7A. Quantitative and Qualitative Disclosures about Market Risk/Investment and Interest Rate Risk", + "metadata": { + "length": 1883, + "summary": "The passage discusses interest rate risk related to fixed-rate investments and outstanding debt. A 0.5% decrease in yield would reduce investment fair value by $0.2 billion. The company has $8.5 billion in senior Notes with no financial statement risk from rate changes. Publicly-held equity securities face market volatility; a 10% decrease would reduce fair value by $1.8 billion. Non-marketable equity securities are valued at cost with adjustments for observable price changes.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "exposed", + "interest", + "rate", + "risk", + "related", + "fixed", + "investment", + "portfolio", + "outstanding", + "debt", + "The", + "managed", + "consistent", + "liquidity", + "strategy", + "support", + "working", + "capital", + "growth", + "businesses", + "According", + "sensitivity", + "analysis", + "decrease", + "yield", + "curve", + "0.5%", + "end", + "fiscal", + "year", + "2026", + "2025", + "fair", + "investments", + "approximately", + "0.2", + "billion", + "As", + "8.5", + "senior", + "Notes", + "carry", + "face", + "unamortized", + "discount", + "Consolidated", + "Balance", + "Sheets", + "bear", + "financial", + "statement", + "rates", + "Refer", + "Note", + "11", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "additional", + "information", + "Publicly", + "held", + "equity", + "securities", + "subject", + "market", + "price", + "volatility", + "hypothetical", + "10%", + "publicly", + "balance", + "1.8", + "insignificant", + "amount", + "January", + "25", + "26", + "Non", + "marketable", + "measured", + "based", + "cost", + "minus", + "impairment", + "adjusted", + "observable", + "orderly", + "transactions", + "identical", + "similar", + "issuer", + "Valuations", + "inherently", + "complex", + "due", + "lack", + "readily", + "data", + "impact", + "macroeconomic", + "factors", + "For", + "description", + "refer", + "Condensed" + ], + "keywords": [ + "interest rate risk", + "investment portfolio", + "equity securities" + ], + "connect_to": [] + } + }, + { + "chunk_id": "5457364b-eb6f-5905-9ea8-ba3417e2737d", + "type": "text", + "content": "We consider our direct exposure to foreign exchange rate fluctuations to be minimal as substantially all of our sales are in United States dollars and foreign currency forward contracts are used to offset movements of foreign currency exchange rates. Gains or losses from foreign currency remeasurement are included in other income or expenses. The impact of foreign currency transaction gain or loss included in determining net income was not significant for fiscal years 2026 and 2025.\nSales and arrangements with third-party manufacturers provide for pricing and payment in United States dollars, and, therefore, are not subject to exchange rate fluctuations. Increases in the value of the United States' dollar relative to other currencies would make our products more expensive, which could negatively impact our ability to compete. Conversely, decreases in the value of the United States' dollar relative to other currencies could result in our suppliers raising their manufacturing costs.\nIf the U.S. dollar strengthened by 10% as of January 25, 2026 and January 26, 2025, the amount recorded in Accumulated other comprehensive income (loss) related to our foreign exchange contracts before tax effect would have been an adverse impact of \\$180 million and \\$136 million, respectively. Change in value of our foreign exchange contracts recorded in Accumulated other comprehensive income (loss) would be expected to offset a corresponding change in hedged forecasted foreign currency expenses when recognized.\nIf an adverse 10% foreign exchange rate change was applied to our balance sheet hedging contracts, it would have resulted in an adverse impact on income before taxes of \\$124 million and \\$129 million as of January 25, 2026 and January 26, 2025, respectively. These changes in fair values would be offset in Total other income, net, by corresponding change in fair values of the foreign currency denominated monetary assets and liabilities, assuming the hedge contracts fully cover the foreign currency denominated monetary assets and liabilities balances.\nRefer to Note 10 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 7A. Quantitative and Qualitative Disclosures about Market Risk/Foreign Exchange Rate Risk", + "metadata": { + "length": 2226, + "summary": "The company has minimal direct foreign exchange exposure as sales are in USD and forward contracts offset currency movements. A 10% USD strengthening would adversely impact accumulated other comprehensive income by $180M (2026) and $136M (2025) from foreign exchange contracts. Balance sheet hedging contracts would see an adverse pre-tax income impact of $124M (2026) and $129M (2025), offset by changes in foreign currency monetary assets and liabilities. Sales and manufacturer arrangements are in USD, but dollar strength could hurt competitiveness while weakness might raise supplier costs.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "direct", + "exposure", + "foreign", + "exchange", + "rate", + "fluctuations", + "minimal", + "substantially", + "sales", + "United", + "States", + "dollars", + "currency", + "forward", + "contracts", + "offset", + "movements", + "rates", + "Gains", + "losses", + "remeasurement", + "included", + "income", + "expenses", + "The", + "impact", + "transaction", + "gain", + "loss", + "determining", + "net", + "significant", + "fiscal", + "years", + "2026", + "2025", + "Sales", + "arrangements", + "party", + "manufacturers", + "provide", + "pricing", + "payment", + "subject", + "Increases", + "dollar", + "relative", + "currencies", + "make", + "products", + "expensive", + "negatively", + "ability", + "compete", + "Conversely", + "decreases", + "result", + "suppliers", + "raising", + "manufacturing", + "costs", + "If", + "strengthened", + "10%", + "January", + "25", + "26", + "amount", + "recorded", + "Accumulated", + "comprehensive", + "related", + "tax", + "effect", + "adverse", + "180", + "million", + "136", + "Change", + "expected", + "change", + "hedged", + "forecasted", + "recognized", + "applied", + "balance", + "sheet", + "hedging", + "resulted", + "taxes", + "124", + "129", + "These", + "fair", + "values", + "Total", + "denominated", + "monetary", + "assets", + "liabilities", + "assuming", + "hedge", + "fully", + "cover", + "balances", + "Refer", + "Note", + "10", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "additional", + "information" + ], + "keywords": [ + "Foreign exchange", + "hedging", + "USD exposure" + ], + "connect_to": [] + } + }, + { + "chunk_id": "192cc19f-0786-5685-93e5-794c3c455aac", + "type": "text", + "content": "The information required by this Item is set forth in our Consolidated Financial Statements and Notes thereto included in this Annual Report on Form 10-K.\nItem 9. Changes in and Disagreements with Accountants on Accounting and Financial Disclosure\nNone.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 8. Financial Statements and Supplementary Data", + "metadata": { + "length": 253, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "information", + "required", + "Item", + "set", + "Consolidated", + "Financial", + "Statements", + "Notes", + "thereto", + "included", + "Annual", + "Report", + "Form", + "10", + "Changes", + "Disagreements", + "Accountants", + "Accounting", + "Disclosure", + "None" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f037bca8-e808-5bc2-9455-e20f0b8b7db1", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f8f0905a-58c3-536f-b081-52512d55b7b6", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures/Controls and Procedures", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7bf5d6ed-5edf-5dc3-9a50-e3f508d8c550", + "type": "text", + "content": "Based on their evaluation as of January 25, 2026, our management, including our Chief Executive Officer and Chief Financial Officer, has concluded that our disclosure controls and procedures (as defined in Rule 13a-15(e) under the Exchange Act) were effective to provide reasonable assurance that the information we are required to disclose in reports that we file or submit under the Exchange Act is recorded, processed, summarized and reported within the time periods specified in the SEC rules and forms, and that such information is accumulated and communicated to our management, including our Chief Executive Officer and our Chief Financial Officer, as appropriate, to allow timely decisions regarding required disclosures.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures/Controls and Procedures/Disclosure Controls and Procedures", + "metadata": { + "length": 729, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Based", + "evaluation", + "January", + "25", + "2026", + "management", + "including", + "Chief", + "Executive", + "Officer", + "Financial", + "concluded", + "disclosure", + "controls", + "procedures", + "defined", + "Rule", + "13a", + "15", + "Exchange", + "Act", + "effective", + "provide", + "reasonable", + "assurance", + "information", + "required", + "disclose", + "reports", + "file", + "submit", + "recorded", + "processed", + "summarized", + "reported", + "time", + "periods", + "SEC", + "rules", + "forms", + "accumulated", + "communicated", + "timely", + "decisions", + "disclosures" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d0f23225-49f5-5ae5-b436-9f495585a67f", + "type": "text", + "content": "Our management is responsible for establishing and maintaining adequate internal control over financial reporting, as such term is defined in Exchange Act Rule 13a-15(f). Under the supervision and with the participation of our management, including our Chief Executive Officer and Chief Financial Officer, we conducted an evaluation of the effectiveness of our internal control over financial reporting as of January 25, 2026 based on the criteria set forth in Internal Control - Integrated Framework (2013) issued by the Committee of Sponsoring Organizations of the Treadway Commission. Based on our evaluation under the criteria set forth in Internal Control — Integrated Framework, our management concluded that our internal control over financial reporting was effective as of January 25, 2026.\nThe effectiveness of our internal control over financial reporting as of January 25, 2026 has been audited by PricewaterhouseCoopers LLP, an independent registered public accounting firm, as stated in its report which is included herein.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures/Controls and Procedures/Management's Annual Report on Internal Control Over Financial Reporting", + "metadata": { + "length": 1036, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "management", + "responsible", + "establishing", + "maintaining", + "adequate", + "internal", + "control", + "financial", + "reporting", + "term", + "defined", + "Exchange", + "Act", + "Rule", + "13a", + "15", + "Under", + "supervision", + "participation", + "including", + "Chief", + "Executive", + "Officer", + "Financial", + "conducted", + "evaluation", + "effectiveness", + "January", + "25", + "2026", + "based", + "criteria", + "set", + "Internal", + "Control", + "Integrated", + "Framework", + "2013", + "issued", + "Committee", + "Sponsoring", + "Organizations", + "Treadway", + "Commission", + "Based", + "concluded", + "effective", + "The", + "audited", + "PricewaterhouseCoopers", + "LLP", + "independent", + "registered", + "public", + "accounting", + "firm", + "stated", + "report", + "included" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "48c95827-5699-5941-be5d-9cadbc490ad8", + "type": "text", + "content": "There have been no changes in our internal control over financial reporting during the quarter ended January 25, 2026 that have materially affected, or are reasonably likely to materially affect, our internal control over financial reporting. We are continuing a phased upgrade of our enterprise resource planning, or ERP, system to update our existing core financial systems. The ERP system is designed to accurately maintain our financial records used to report operating results. We will continue to evaluate each quarter whether there are changes that materially affect our internal control over financial reporting.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures/Controls and Procedures/Changes in Internal Control Over Financial Reporting", + "metadata": { + "length": 620, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "There", + "internal", + "control", + "financial", + "reporting", + "quarter", + "ended", + "January", + "25", + "2026", + "materially", + "affected", + "affect", + "We", + "continuing", + "phased", + "upgrade", + "enterprise", + "resource", + "planning", + "ERP", + "system", + "update", + "existing", + "core", + "systems", + "The", + "designed", + "accurately", + "maintain", + "records", + "report", + "operating", + "results", + "continue", + "evaluate" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "500eb535-09c7-5339-98be-e29a368babf0", + "type": "text", + "content": "Our management, including our Chief Executive Officer and Chief Financial Officer, does not expect that our disclosure controls and procedures or our internal controls, will prevent all error and all fraud. A control system, no matter how well conceived and operated, can provide only reasonable, not absolute, assurance that the objectives of the control system are met. Further, the design of a control system must reflect the fact that there are resource constraints, and the benefits of controls must be considered relative to their costs. Because of the inherent limitations in all control systems, no evaluation of controls can provide absolute assurance that all control issues and instances of fraud, if any, within NVIDIA have been detected.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures/Controls and Procedures/Inherent Limitations on Effectiveness of Controls", + "metadata": { + "length": 750, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "management", + "including", + "Chief", + "Executive", + "Officer", + "Financial", + "expect", + "disclosure", + "controls", + "procedures", + "internal", + "prevent", + "error", + "fraud", + "control", + "system", + "matter", + "conceived", + "operated", + "provide", + "reasonable", + "absolute", + "assurance", + "objectives", + "met", + "Further", + "design", + "reflect", + "fact", + "resource", + "constraints", + "benefits", + "considered", + "relative", + "costs", + "Because", + "inherent", + "limitations", + "systems", + "evaluation", + "issues", + "instances", + "NVIDIA", + "detected" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a28bc951-0095-5616-a1a1-36814e5285cf", + "type": "table", + "content": "
NameTitle of Director or OfficerActionDateTotal Shares of Common Stock to be SoldExpiration Date
John O. DabiriDirectorAdoption12/10/20253,984*12/7/2026
Colette M. KressExecutive Vice President and Chief Financial OfficerAdoption12/18/2025500,0003/23/2027
", + "path": "tables/table-73 Stock Sale Plans.html", + "metadata": { + "length": 457, + "summary": "table-74\nTwo insiders, Director John O. Dabiri and CFO Colette M. Kress, adopted Rule 10b5-1 trading plans to sell shares of common stock, with Dabiri selling 3,984 shares and Kress selling 500,000 shares.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-73 Stock Sale Plans.html", + "keywords": [ + "Director", + "Officer", + "Stock Sale" + ], + "tokens": [] + } + }, + { + "chunk_id": "2dc0f271-b865-5797-9fee-2cc8ec394e73", + "type": "text", + "content": "The following members of our Board of Directors and/or officers adopted, modified or terminated a trading arrangement that is intended to satisfy the affirmative defense conditions of Rule 10b5-1(c), or a Rule 10b5-1 Trading Arrangement:\n\n[tables/table-73 Stock Sale Plans.html]\n\n\\*Estimated assuming our closing stock price as of January 23, 2026. The number of shares is based on an estimate because the plan specifies a formulaic dollar amount of shares to be sold.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9B. Other Information", + "metadata": { + "length": 532, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "members", + "Board", + "Directors", + "officers", + "adopted", + "modified", + "terminated", + "trading", + "arrangement", + "intended", + "satisfy", + "affirmative", + "defense", + "conditions", + "Rule", + "10b5", + "Trading", + "Arrangement", + "Estimated", + "assuming", + "closing", + "stock", + "price", + "January", + "23", + "2026", + "number", + "shares", + "based", + "estimate", + "plan", + "specifies", + "formulaic", + "dollar", + "amount", + "sold" + ], + "keywords": [], + "connect_to": [ + { + "target": "a28bc951-0095-5616-a1a1-36814e5285cf", + "relation": "embeds", + "ref": "[tables/table-73 Stock Sale Plans.html]", + "position": { + "start": 239, + "end": 278 + } + } + ] + } + }, + { + "chunk_id": "87ba7dbb-a30f-540c-9ecb-7b6af1b559fe", + "type": "text", + "content": "Not Applicable.", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9C. Disclosure Regarding Foreign Jurisdictions that Prevent Inspections", + "metadata": { + "length": 15, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Not", + "Applicable" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "48d79f53-3cfd-523f-a804-c169481a5e83", + "type": "text", + "content": "Certain information required by Part III is omitted from this report because we will file with the SEC a definitive proxy statement pursuant to Regulation 14A, or the 2026 Proxy Statement, no later than 120 days after the end of fiscal year 2026, and certain information included therein is incorporated herein by reference.", + "path": "2026-Annual-Report-Web.pdf/Part III_2", + "metadata": { + "length": 324, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Certain", + "information", + "required", + "Part", + "III", + "omitted", + "report", + "file", + "SEC", + "definitive", + "proxy", + "statement", + "pursuant", + "Regulation", + "14A", + "2026", + "Proxy", + "Statement", + "120", + "days", + "end", + "fiscal", + "year", + "included", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1b80ab6f-e2a5-5054-92df-2270ea0cc964", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a551210a-2b9b-5fdd-8ef5-0228ba35ab0d", + "type": "text", + "content": "Information regarding directors required by this item will be contained in our 2026 Proxy Statement under the caption “Proposal 1 — Election of Directors,” and is hereby incorporated by reference.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Identification of Directors", + "metadata": { + "length": 196, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Information", + "directors", + "required", + "item", + "contained", + "2026", + "Proxy", + "Statement", + "caption", + "Proposal", + "Election", + "Directors", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "015a5723-8a76-5d6b-8612-fc965db48c3f", + "type": "text", + "content": "Reference is made to the information regarding executive officers appearing under the heading “Information About Our Executive Officers” in Part I of this Annual Report on Form 10-K, which information is hereby incorporated by reference.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Identification of Executive Officers", + "metadata": { + "length": 237, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Reference", + "made", + "information", + "executive", + "officers", + "appearing", + "heading", + "Information", + "About", + "Our", + "Executive", + "Officers", + "Part", + "Annual", + "Report", + "Form", + "10", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a7c8f3f4-20b6-5f2e-830e-9f0ca188f0f6", + "type": "text", + "content": "Information regarding our Audit Committee required by this item will be contained in our 2026 Proxy Statement under the captions “Report of the Audit Committee of the Board of Directors” and “Information About the Board of Directors and Corporate Governance,” and is hereby incorporated by reference.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Identification of Audit Committee and Financial Experts", + "metadata": { + "length": 300, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Information", + "Audit", + "Committee", + "required", + "item", + "contained", + "2026", + "Proxy", + "Statement", + "captions", + "Report", + "Board", + "Directors", + "About", + "Corporate", + "Governance", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8a7b4695-2595-5035-9968-a38e0671fcf7", + "type": "text", + "content": "Information regarding procedures for recommending directors required by this item will be contained in our 2026 Proxy Statement under the caption “Information About the Board of Directors and Corporate Governance,” and is hereby incorporated by reference.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Material Changes to Procedures for Recommending Directors", + "metadata": { + "length": 255, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Information", + "procedures", + "recommending", + "directors", + "required", + "item", + "contained", + "2026", + "Proxy", + "Statement", + "caption", + "About", + "Board", + "Directors", + "Corporate", + "Governance", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a0290fb4-eeb8-55c6-bfde-af50c8e44b00", + "type": "text", + "content": "Information regarding compliance with Section 16(a) of the Exchange Act required by this item will be contained in our 2026 Proxy Statement under the caption “Delinquent Section 16(a) Reports,” and such disclosure, if any, is hereby incorporated by reference.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Delinquent Section 16(a) Reports", + "metadata": { + "length": 259, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Information", + "compliance", + "Section", + "16", + "Exchange", + "Act", + "required", + "item", + "contained", + "2026", + "Proxy", + "Statement", + "caption", + "Delinquent", + "Reports", + "disclosure", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f602847a-7ba5-59fd-8388-fbd06c187732", + "type": "text", + "content": "Information regarding our Code of Conduct required by this item will be contained in our 2026 Proxy Statement under the caption “Information About the Board of Directors and Corporate Governance — Code of Conduct,” and is hereby incorporated by reference. The full text of our Code of Conduct and Financial Team Code of Conduct are published on the Investor Relations portion of our website, under Governance, at www.nvidia.com. If we make any amendments to either code, or grant any waiver from a provision of either code to any executive officer or director, we will promptly disclose the nature of the amendment or waiver on our website or in a report on Form 8-K. The contents of our website are not a part of this Annual Report on Form 10-K.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Code of Conduct", + "metadata": { + "length": 746, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Information", + "Code", + "Conduct", + "required", + "item", + "contained", + "2026", + "Proxy", + "Statement", + "caption", + "About", + "Board", + "Directors", + "Corporate", + "Governance", + "incorporated", + "reference", + "The", + "full", + "text", + "Financial", + "Team", + "published", + "Investor", + "Relations", + "portion", + "website", + "www", + "nvidia", + "If", + "make", + "amendments", + "code", + "grant", + "waiver", + "provision", + "executive", + "officer", + "director", + "promptly", + "disclose", + "nature", + "amendment", + "report", + "Form", + "contents", + "part", + "Annual", + "Report", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e9e3f451-ccee-5412-8aa1-53cbe0d39bd5", + "type": "text", + "content": "The information required by Item 408(b) of Regulation S-K is incorporated by reference from the information contained in our 2026 Proxy Statement under the heading “Information About the Board of Directors and Corporate Governance.”", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Insider Trading Policy", + "metadata": { + "length": 232, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "information", + "required", + "Item", + "408", + "Regulation", + "incorporated", + "reference", + "contained", + "2026", + "Proxy", + "Statement", + "heading", + "Information", + "About", + "Board", + "Directors", + "Corporate", + "Governance" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9f2cda86-0406-556a-941d-14365ca03fbc", + "type": "text", + "content": "Information regarding our executive compensation required by this item will be contained in our 2026 Proxy Statement under the captions “Executive Compensation,” “Compensation Committee Interlocks and Insider Participation,” “Director Compensation,” and “Compensation Committee Report,” and is hereby incorporated by reference.\nItem 12. Security Ownership of Certain Beneficial Owners and Management and Related Stockholder Matters", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 11. Executive Compensation", + "metadata": { + "length": 431, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Information", + "executive", + "compensation", + "required", + "item", + "contained", + "2026", + "Proxy", + "Statement", + "captions", + "Executive", + "Compensation", + "Committee", + "Interlocks", + "Insider", + "Participation", + "Director", + "Report", + "incorporated", + "reference", + "Item", + "12", + "Security", + "Ownership", + "Certain", + "Beneficial", + "Owners", + "Management", + "Related", + "Stockholder", + "Matters" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "137f3b59-c700-5dca-af92-6dc1e503bd97", + "type": "text", + "content": "Information regarding ownership of NVIDIA securities required by this item will be contained in our 2026 Proxy Statement under the caption “Security Ownership of Certain Beneficial Owners and Management,” and is hereby incorporated by reference.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 11. Executive Compensation/Ownership of NVIDIA Securities", + "metadata": { + "length": 245, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Information", + "ownership", + "NVIDIA", + "securities", + "required", + "item", + "contained", + "2026", + "Proxy", + "Statement", + "caption", + "Security", + "Ownership", + "Certain", + "Beneficial", + "Owners", + "Management", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "879665f5-c610-5888-8fd0-ea3902e0e659", + "type": "text", + "content": "Information regarding our equity compensation plans required by this item will be contained in our 2026 Proxy Statement under the caption \"Equity Compensation Plan Information,\" and is hereby incorporated by reference.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 11. Executive Compensation/Equity Compensation Plan Information", + "metadata": { + "length": 218, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Information", + "equity", + "compensation", + "plans", + "required", + "item", + "contained", + "2026", + "Proxy", + "Statement", + "caption", + "Equity", + "Compensation", + "Plan", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a326b714-2c2f-54eb-8c25-4d8c4e20bdc6", + "type": "text", + "content": "Information regarding related transactions and director independence required by this item will be contained in our 2026 Proxy Statement under the captions “Review of Transactions with Related Persons” and “Information About the Board of Directors and Corporate Governance — Independence of the Members of the Board of Directors,” and is hereby incorporated by reference.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 13. Certain Relationships and Related Transactions, and Director Independence", + "metadata": { + "length": 371, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Information", + "related", + "transactions", + "director", + "independence", + "required", + "item", + "contained", + "2026", + "Proxy", + "Statement", + "captions", + "Review", + "Transactions", + "Related", + "Persons", + "About", + "Board", + "Directors", + "Corporate", + "Governance", + "Independence", + "Members", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8851a51b-043e-53e1-b3bf-45eb181e24ca", + "type": "text", + "content": "Information regarding accounting fees and services required by this item will be contained in our 2026 Proxy Statement under the caption “Fees Billed by the Independent Registered Public Accounting Firm,” and is hereby incorporated by reference.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 14. Principal Accountant Fees and Services", + "metadata": { + "length": 245, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Information", + "accounting", + "fees", + "services", + "required", + "item", + "contained", + "2026", + "Proxy", + "Statement", + "caption", + "Fees", + "Billed", + "Independent", + "Registered", + "Public", + "Accounting", + "Firm", + "incorporated", + "reference" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "78723bfe-fed0-5972-a070-2f7b2e294812", + "type": "text", + "content": "Page", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules", + "metadata": { + "length": 4, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1eef2b72-f46e-512a-b26d-9f7f7bda8d06", + "type": "text", + "content": "Report of Independent Registered Public Accounting Firm (PCAOB ID: 238) 49\nConsolidated Statements of Income for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 51\nConsolidated Statements of Comprehensive Income for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 52\nConsolidated Balance Sheets as of January 25, 2026 and January 26, 2025 53\nConsolidated Statements of Shareholders' Equity for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 54\nConsolidated Statements of Cash Flows for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 55\nNotes to the Consolidated Financial Statements 56", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/(a) 1. Financial Statements", + "metadata": { + "length": 687, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Report", + "Independent", + "Registered", + "Public", + "Accounting", + "Firm", + "PCAOB", + "ID", + "238", + "49", + "Consolidated", + "Statements", + "Income", + "years", + "ended", + "January", + "25", + "2026", + "26", + "2025", + "28", + "2024", + "51", + "Comprehensive", + "52", + "Balance", + "Sheets", + "53", + "Shareholders", + "Equity", + "54", + "Cash", + "Flows", + "55", + "Notes", + "Financial", + "56" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c9e49b81-7603-55bb-933e-8640daaccda7", + "type": "text", + "content": "Schedule II Valuation and Qualifying Accounts for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 81", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/2. Financial Statement Schedule", + "metadata": { + "length": 125, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Schedule", + "II", + "Valuation", + "Qualifying", + "Accounts", + "years", + "ended", + "January", + "25", + "2026", + "26", + "2025", + "28", + "2024", + "81" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "81b07abe-3da6-5d73-8b51-cb4a93b1f0ea", + "type": "text", + "content": "The exhibits listed in the accompanying index to exhibits are filed or incorporated by reference as a part of this Annual Report on Form 10-K. 82", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/3. Exhibits", + "metadata": { + "length": 145, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "exhibits", + "listed", + "accompanying", + "index", + "filed", + "incorporated", + "reference", + "part", + "Annual", + "Report", + "Form", + "10", + "82" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "09173c86-702d-54c2-9748-c5cd1f32129d", + "type": "text", + "content": "To the Board of Directors and Shareholders of NVIDIA Corporation\nOpinions on the Financial Statements and Internal Control over Financial Reporting\nWe have audited the accompanying consolidated balance sheets of NVIDIA Corporation and its subsidiaries (the \"Company\") as of January 25, 2026 and January 26, 2025, and the related consolidated statements of income, comprehensive income, shareholders' equity, and cash flows for each of the three years in the period ended January 25, 2026, including the related notes and financial statement schedule listed in the index appearing under Item 15(a)(2) (collectively referred to as the \"consolidated financial statements\"). We also have audited the Company's internal control over financial reporting as of January 25, 2026, based on criteria established in Internal Control - Integrated Framework (2013) issued by the Committee of Sponsoring Organizations of the Treadway Commission (COSO).\nIn our opinion, the consolidated financial statements referred to above present fairly, in all material respects, the financial position of the Company as of January 25, 2026 and January 26, 2025, and the results of its operations and its cash flows for each of the three years in the period ended January 25, 2026 in conformity with accounting principles generally accepted in the United States of America. Also in our opinion, the Company maintained, in all material respects, effective internal control over financial reporting as of January 25, 2026, based on criteria established in Internal Control - Integrated Framework (2013) issued by the COSO.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/Report of Independent Registered Public Accounting Firm", + "metadata": { + "length": 1593, + "summary": "The text is an audit opinion from an independent auditor to the Board and Shareholders of NVIDIA Corporation. It covers the audit of consolidated financial statements for fiscal years ending January 25, 2026 and January 26, 2025, as well as internal control over financial reporting as of January 25, 2026. The auditor concludes that the financial statements fairly present the company's financial position and results in accordance with U.S. GAAP, and that effective internal control over financial reporting was maintained based on COSO criteria.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "To", + "Board", + "Directors", + "Shareholders", + "NVIDIA", + "Corporation", + "Opinions", + "Financial", + "Statements", + "Internal", + "Control", + "Reporting", + "We", + "audited", + "accompanying", + "consolidated", + "balance", + "sheets", + "subsidiaries", + "Company", + "January", + "25", + "2026", + "26", + "2025", + "related", + "statements", + "income", + "comprehensive", + "shareholders", + "equity", + "cash", + "flows", + "years", + "period", + "ended", + "including", + "notes", + "financial", + "statement", + "schedule", + "listed", + "index", + "appearing", + "Item", + "15", + "collectively", + "referred", + "internal", + "control", + "reporting", + "based", + "criteria", + "established", + "Integrated", + "Framework", + "2013", + "issued", + "Committee", + "Sponsoring", + "Organizations", + "Treadway", + "Commission", + "COSO", + "In", + "opinion", + "present", + "fairly", + "material", + "respects", + "position", + "results", + "operations", + "conformity", + "accounting", + "principles", + "generally", + "accepted", + "United", + "States", + "America", + "Also", + "maintained", + "effective" + ], + "keywords": [ + "Audit", + "Financial Statements", + "Internal Control" + ], + "connect_to": [] + } + }, + { + "chunk_id": "82ea4134-1144-5c84-adb2-a0c6aae9b5ad", + "type": "text", + "content": "The Company's management is responsible for these consolidated financial statements, for maintaining effective internal control over financial reporting, and for its assessment of the effectiveness of internal control over financial reporting, included in Management's Annual Report on Internal Control over Financial Reporting appearing under Item 9A. Our responsibility is to express opinions on the Company's consolidated financial statements and on the Company's internal control over financial reporting based on our audits. We are a public accounting firm registered with the Public Company Accounting Oversight Board (United States) (PCAOB) and are required to be independent with respect to the Company in accordance with the U.S. federal securities laws and the applicable rules and regulations of the Securities and Exchange Commission and the PCAOB.\nWe conducted our audits in accordance with the standards of the PCAOB. Those standards require that we plan and perform the audits to obtain reasonable assurance about whether the consolidated financial statements are free of material misstatement, whether due to error or fraud, and whether effective internal control over financial reporting was maintained in all material respects.\nOur audits of the consolidated financial statements included performing procedures to assess the risks of material misstatement of the consolidated financial statements, whether due to error or fraud, and performing procedures that respond to those risks. Such procedures included examining, on a test basis, evidence regarding the amounts and disclosures in the consolidated financial statements. Our audits also included evaluating the accounting principles used and significant estimates made by management, as well as evaluating the overall presentation of the consolidated financial statements. Our audit of internal control over financial reporting included obtaining an understanding of internal control over financial reporting, assessing the risk that a material weakness exists, and testing and evaluating the design and operating effectiveness of internal control based on the assessed risk. Our audits also included performing such other procedures as we considered necessary in the circumstances. We believe that our audits provide a reasonable basis for our opinions.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/Report of Independent Registered Public Accounting Firm/Basis for Opinions", + "metadata": { + "length": 2327, + "summary": "The passage describes the responsibilities of management and auditors regarding consolidated financial statements and internal control over financial reporting. Management is responsible for the financial statements and maintaining effective internal control, while the auditors express opinions on both. The audits are conducted in accordance with PCAOB standards to obtain reasonable assurance about the absence of material misstatements and the effectiveness of internal controls. Procedures include assessing risks, examining evidence, evaluating accounting principles and estimates, and testing internal controls.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Company", + "management", + "responsible", + "consolidated", + "financial", + "statements", + "maintaining", + "effective", + "internal", + "control", + "reporting", + "assessment", + "effectiveness", + "included", + "Management", + "Annual", + "Report", + "Internal", + "Control", + "Financial", + "Reporting", + "appearing", + "Item", + "9A", + "Our", + "responsibility", + "express", + "opinions", + "based", + "audits", + "We", + "public", + "accounting", + "firm", + "registered", + "Public", + "Accounting", + "Oversight", + "Board", + "United", + "States", + "PCAOB", + "required", + "independent", + "respect", + "accordance", + "federal", + "securities", + "laws", + "applicable", + "rules", + "regulations", + "Securities", + "Exchange", + "Commission", + "conducted", + "standards", + "Those", + "require", + "plan", + "perform", + "obtain", + "reasonable", + "assurance", + "free", + "material", + "misstatement", + "due", + "error", + "fraud", + "maintained", + "respects", + "performing", + "procedures", + "assess", + "risks", + "respond", + "Such", + "examining", + "test", + "basis", + "evidence", + "amounts", + "disclosures", + "evaluating", + "principles", + "significant", + "estimates", + "made", + "presentation", + "audit", + "obtaining", + "understanding", + "assessing", + "risk", + "weakness", + "exists", + "testing", + "design", + "operating", + "assessed", + "considered", + "circumstances", + "provide" + ], + "keywords": [ + "audit", + "internal control", + "financial statements" + ], + "connect_to": [] + } + }, + { + "chunk_id": "670c8fca-3925-5244-b848-02652e84bfdc", + "type": "text", + "content": "A company's internal control over financial reporting is a process designed to provide reasonable assurance regarding the reliability of financial reporting and the preparation of financial statements for external purposes in accordance with generally accepted accounting principles. A company's internal control over financial reporting includes those policies and procedures that (i) pertain to the maintenance of records that, in reasonable detail, accurately and fairly reflect the transactions and dispositions of the assets of the company; (ii) provide reasonable assurance that transactions are recorded as necessary to permit preparation of financial statements in accordance with generally accepted accounting principles, and that receipts and expenditures of the company are being made only in accordance with authorizations of management and directors of the company; and (iii) provide reasonable assurance regarding prevention or timely detection of unauthorized acquisition, use, or disposition of the company's assets that could have a material effect on the financial statements.\nBecause of its inherent limitations, internal control over financial reporting may not prevent or detect misstatements. Also, projections of any evaluation of effectiveness to future periods are subject to the risk that controls may become inadequate because of changes in conditions, or that the degree of compliance with the policies or procedures may deteriorate.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/Report of Independent Registered Public Accounting Firm/Definition and Limitations of Internal Control over Financial Reporting", + "metadata": { + "length": 1461, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "company", + "internal", + "control", + "financial", + "reporting", + "process", + "designed", + "provide", + "reasonable", + "assurance", + "reliability", + "preparation", + "statements", + "external", + "purposes", + "accordance", + "generally", + "accepted", + "accounting", + "principles", + "includes", + "policies", + "procedures", + "pertain", + "maintenance", + "records", + "detail", + "accurately", + "fairly", + "reflect", + "transactions", + "dispositions", + "assets", + "ii", + "recorded", + "permit", + "receipts", + "expenditures", + "made", + "authorizations", + "management", + "directors", + "iii", + "prevention", + "timely", + "detection", + "unauthorized", + "acquisition", + "disposition", + "material", + "effect", + "Because", + "inherent", + "limitations", + "prevent", + "detect", + "misstatements", + "Also", + "projections", + "evaluation", + "effectiveness", + "future", + "periods", + "subject", + "risk", + "controls", + "inadequate", + "conditions", + "degree", + "compliance", + "deteriorate" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "08ce51f5-def4-5e30-a687-c523ef580752", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Revenue$ 215,938$ 130,497$ 60,922
Cost of revenue62,47532,63916,621
Gross profit153,46397,85844,301
Operating expenses
Research and development18,49712,9148,675
Sales, general and administrative4,5793,4912,654
Total operating expenses23,07616,40511,329
Operating income130,38781,45332,972
Interest income2,3001,786866
Interest expense(259)(247)(257)
Other income, net9,0221,034237
Total other income, net11,0632,573846
Income before income tax141,45084,02633,818
Income tax expense21,38311,1464,058
Net income$ 120,067$ 72,880$ 29,760
Net income per share:
Basic$ 4.93$ 2.97$ 1.21
Diluted$ 4.90$ 2.94$ 1.19
Weighted average shares used in per share computation:
Basic24,35924,55524,690
Diluted24,51424,80424,940
", + "path": "tables/table-74 Income Statement.html", + "metadata": { + "length": 1781, + "summary": "table-75\nThe table shows financial data for three fiscal years ending Jan 2026, 2025, and 2024. Revenue grew from $60,922 to $215,938, with net income rising from $29,760 to $120,067.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-74 Income Statement.html", + "keywords": [ + "Revenue", + "Net Income", + "Operating Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "66d204c1-4792-5bae-8b7c-64445b5ea8fb", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Net income$ 120,067$ 72,880$ 29,760
Other comprehensive income, net of tax
Available-for-sale securities:
Net change in unrealized gain107180
Cash flow hedges:
Net change in unrealized gain (loss)43(10)
Other comprehensive income, net of tax150170
Total comprehensive income$ 120,217$ 72,881$ 29,830
", + "path": "tables/table-75 Comprehensive Income.html", + "metadata": { + "length": 798, + "summary": "table-76\nThe table shows net income, other comprehensive income (OCI) from securities and hedges, and total comprehensive income for fiscal years ending Jan 2026, 2025, and 2024.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-75 Comprehensive Income.html", + "keywords": [ + "Net income", + "OCI", + "Comprehensive income" + ], + "tokens": [] + } + }, + { + "chunk_id": "51e474a3-da2a-55a0-b70a-b206f8b23f9e", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
Assets
Current assets:
Cash and cash equivalents$ 10,605$ 8,589
Marketable securities51,95134,621
Accounts receivable, net38,46623,065
Inventories21,40310,080
Prepaid expenses and other current assets3,1803,771
Total current assets125,60580,126
Property and equipment, net10,3836,283
Operating lease assets2,8671,793
Goodwill20,8325,188
Intangible assets, net3,306807
Deferred income tax assets13,25810,979
Non-marketable equity securities22,2513,387
Other assets8,3013,038
Total assets$ 206,803$ 111,601
Liabilities and Shareholders' Equity
Current liabilities:
Accounts payable$ 9,812$ 6,310
Accrued and other current liabilities21,35211,737
Short-term debt999
Total current liabilities32,16318,047
Long-term debt7,4698,463
Long-term operating lease liabilities2,5721,519
Other long-term liabilities7,3064,245
Total liabilities49,51032,274
Commitments and contingencies - see Note 12
Shareholders' equity:
Preferred stock, $0.001 par value; 2 shares authorized; none issued
Common stock, $0.001 par value; 80,000 shares authorized; 24,304 shares issued and outstanding as of January 25, 2026; 24,477 shares issued and outstanding as of January 26, 20252424
Additional paid-in capital10,11811,237
Accumulated other comprehensive income17828
Retained earnings146,97368,038
Total shareholders' equity157,29379,327
Total liabilities and shareholders' equity$ 206,803$ 111,601
", + "path": "tables/table-76 Balance Sheet.html", + "metadata": { + "length": 2669, + "summary": "table-77\nThe balance sheet shows total assets of $206,803 as of Jan 25, 2026, up from $111,601 a year earlier, with increases in cash, marketable securities, and goodwill. Total liabilities rose to $49,510, while shareholders' equity grew to $157,293.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-76 Balance Sheet.html", + "keywords": [ + "Assets", + "Liabilities", + "Equity" + ], + "tokens": [] + } + }, + { + "chunk_id": "23e7af72-eac0-52dd-a5b9-ae90e956bb81", + "type": "table", + "content": "
Common Stock OutstandingAdditional Paid-in CapitalAccumulated Other Comprehensive Income (Loss)Retained EarningsTotal Shareholders' Equity
SharesAmount
(In millions, except per share data)
Balances as of Jan 29, 202324,661$ 25$ 11,948$ (43)$ 10,171$ 22,101
Net income29,76029,760
Other comprehensive income7070
Issuance of common stock265403403
Tax withholding related to common stock(72)(2,783)(2,783)
Shares repurchased(211)(27)(9,719)(9,746)
Cash dividends declared and paid ($0.016 per common share)(395)(395)
Stock-based compensation3,5683,568
Balances as of Jan 28, 202424,6432513,1092729,81742,978
Net income72,88072,880
Other comprehensive income11
Issuance of common stock203490490
Tax withholding related to common stock(59)(6,930)(6,930)
Shares repurchased(310)(1)(189)(33,825)(34,015)
Cash dividends declared and paid ($0.034 per common share)(834)(834)
Stock-based compensation4,7574,757
Balances as of Jan 26, 202524,4772411,2372868,03879,327
Net income120,067120,067
Other comprehensive income150150
Issuance of common stock160644644
Tax withholding related to common stock(51)(7,948)(7,948)
Shares repurchased(282)(230)(40,158)(40,388)
Cash dividends declared and paid ($0.04 per common share)(974)(974)
Fair value of partially vested equity awards assumed in connection with acquisitions2828
Stock-based compensation6,3876,387
Balances as of Jan 25, 202624,304$ 24$ 10,118$ 178$ 146,973$ 157,293
", + "path": "tables/table-77 Equity Changes.html", + "metadata": { + "length": 3543, + "summary": "table-78\nThe table shows shareholders' equity changes from Jan 2023 to Jan 2026, including net income, stock issuances, repurchases, dividends, and stock-based compensation, with total equity rising from $22,101M to $157,293M.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-77 Equity Changes.html", + "keywords": [ + "Shareholders' Equity", + "Stock Repurchases", + "Dividends" + ], + "tokens": [] + } + }, + { + "chunk_id": "263389be-7d96-5da1-b4e9-9077ed5aba31", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Cash flows from operating activities:
Net income$120,067$72,880$29,760
Adjustments to reconcile net income to net cash provided by operating activities:
Stock-based compensation expense6,3864,7373,549
Depreciation and amortization2,8431,8641,508
Gains on non-marketable equity securities and publicly-held equity securities, net(8,918)(1,030)(238)
Deferred income taxes(1,424)(4,477)(2,489)
Other(287)(502)(278)
Changes in operating assets and liabilities, net of acquisitions:
Accounts receivable(15,399)(13,063)(6,172)
Inventories(11,324)(4,781)(98)
Prepaid expenses and other assets577(395)(1,522)
Accounts payable3,0963,3571,531
Accrued and other current liabilities5,2574,2782,025
Other long-term liabilities1,8441,221514
Net cash provided by operating activities102,71864,08928,090
Cash flows from investing activities:
Proceeds from sales of marketable securities15,15749550
Proceeds from maturities of marketable securities11,22611,1959,732
Proceeds from sales of non-marketable equity securities841711
Purchases of marketable securities(40,616)(26,575)(18,211)
Purchases of non-marketable equity securities(17,502)(1,486)(862)
Groq, Inc.(13,000)
Purchases related to property and equipment and intangible assets(6,042)(3,236)(1,069)
Acquisitions, net of cash acquired(1,535)(1,007)(83)
Other22(124)
Net cash used in investing activities(52,228)(20,421)(10,566)
Cash flows from financing activities:
Proceeds related to employee stock plans644490403
Payments related to repurchases of common stock(40,086)(33,706)(9,533)
Payments related to employee stock plan taxes(7,948)(6,930)(2,783)
Dividends paid(974)(834)(395)
Principal payments on property and equipment and intangible assets(101)(129)(74)
Repayment of debt(1,250)(1,250)
Other(9)(1)
Net cash used in financing activities(48,474)(42,359)(13,633)
Change in cash and cash equivalents2,0161,3093,891
Cash and cash equivalents at beginning of period8,5897,2803,389
Cash and cash equivalents at end of period$10,605$8,589$7,280
Supplemental disclosures of cash flow information:
Cash paid for income taxes, net$20,288$15,118$6,549
", + "path": "tables/table-78 Cash Flow Statement.html", + "metadata": { + "length": 4047, + "summary": "table-79\nThe table shows cash flows for three fiscal years ending Jan 2026, 2025, and 2024. Net cash from operations was $102.7M, $64.1M, and $28.1M, respectively. Investing activities used $52.2M, $20.4M, and $10.6M, while financing used $48.5M, $42.4M, and $13.6M.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-78 Cash Flow Statement.html", + "keywords": [ + "cash flows", + "operating", + "investing" + ], + "tokens": [] + } + }, + { + "chunk_id": "a9a021e4-5ed0-5a25-a837-36e8b1bbb8da", + "type": "text", + "content": "The critical audit matter communicated below is a matter arising from the current period audit of the consolidated financial statements that was communicated or required to be communicated to the audit committee and that (i) relates to accounts or disclosures that are material to the consolidated financial statements and (ii) involved our especially challenging, subjective, or complex judgments. The communication of critical audit matters does not alter in any way our opinion on the consolidated financial statements, taken as a whole, and we are not, by communicating the critical audit matter below, providing a separate opinion on the critical audit matter or on the accounts or disclosures to which it relates.\nValuation of Inventories - Provisions for Excess or Obsolete Inventories and Excess Product Purchase Commitments\nAs described in Notes 1, 9, and 12 to the consolidated financial statements, the Company charges cost of sales for inventory provisions to write-down inventory for excess or obsolete inventory and for excess product purchase commitments. Most of the Company's inventory provisions relate to excess quantities of products, based on the Company's inventory levels and future product purchase commitments compared to assumptions about future demand including the impact of market conditions such as regulatory export restrictions on their products. As of January 25, 2026, the Company's consolidated inventories balance was \\$21.4 billion and the Company's consolidated outstanding inventory purchase and long-term supply and capacity obligations balance was \\$95.2 billion, of which a significant portion relates to inventory purchase obligations.\nThe principal considerations for our determination that performing procedures relating to the valuation of inventories, specifically the provisions for excess or obsolete inventories and excess product purchase commitments, is a critical audit matter are the significant judgment by management when developing provisions for excess or obsolete inventories and excess product purchase commitments, including developing assumptions related to future demand and market conditions. This in turn led to significant auditor judgment, subjectivity, and effort in performing procedures and evaluating management's assumptions related to future demand and market conditions.\nAddressing the matter involved performing procedures and evaluating audit evidence in connection with forming our overall opinion on the consolidated financial statements. These procedures included testing the effectiveness of controls relating to management's provisions for excess or obsolete inventories and excess product purchase commitments, including controls over management's assumptions related to future demand and market conditions. These procedures also included, among others, testing management's process for developing the provisions for excess or obsolete inventories and excess product purchase commitments; evaluating the appropriateness of management's approach; testing the completeness and accuracy of underlying data used in the approach; and evaluating the reasonableness of management's assumptions related to future demand and market conditions. Evaluating management's assumptions related to future demand and market conditions involved evaluating whether the assumptions used by management were reasonable considering (i) current and past results, including historical product life cycle, (ii) the consistency with external market and industry data, and (iii) changes in technology.\n/s/PricewaterhouseCoopers LLP\nSan Jose, California\nFebruary 25, 2026\nWe have served as the Company's auditor since 2004.\nNVIDIA Corporation and Subsidiaries\nConsolidated Statements of Income\n(In millions, except per share data)\n\n[tables/table-74 Income Statement.html]\n\nSee accompanying Notes to the Consolidated Financial Statements.\nNVIDIA Corporation and Subsidiaries\nConsolidated Statements of Comprehensive Income\n(In millions)\n\n[tables/table-75 Comprehensive Income.html]\n\nSee accompanying Notes to the Consolidated Financial Statements.\nNVIDIA Corporation and Subsidiaries\nConsolidated Balance Sheets\n(In millions, except par value)\n\n[tables/table-76 Balance Sheet.html]\n\nSee accompanying Notes to the Consolidated Financial Statements.\nNVIDIA Corporation and Subsidiaries\nConsolidated Statements of Shareholders' Equity\n\n[tables/table-77 Equity Changes.html]\n\nSee accompanying Notes to the Consolidated Financial Statements.\nNVIDIA Corporation and Subsidiaries Consolidated Statements of Cash Flows (In millions)\n\n[tables/table-78 Cash Flow Statement.html]\n\nSee accompanying Notes to the Consolidated Financial Statements.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/Report of Independent Registered Public Accounting Firm/Critical Audit Matters", + "metadata": { + "length": 4739, + "summary": "The passage discusses a critical audit matter regarding the valuation of inventories, specifically provisions for excess or obsolete inventories and excess product purchase commitments. Management uses significant judgment to develop assumptions about future demand and market conditions, including regulatory export restrictions. The auditor tested controls, management's process, data accuracy, and reasonableness of assumptions against historical results, external data, and technology changes. As of January 25, 2026, inventories were $21.4 billion and purchase obligations were $95.2 billion.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "critical", + "audit", + "matter", + "communicated", + "arising", + "current", + "period", + "consolidated", + "financial", + "statements", + "required", + "committee", + "relates", + "accounts", + "disclosures", + "material", + "ii", + "involved", + "challenging", + "subjective", + "complex", + "judgments", + "communication", + "matters", + "alter", + "opinion", + "communicating", + "providing", + "separate", + "Valuation", + "Inventories", + "Provisions", + "Excess", + "Obsolete", + "Product", + "Purchase", + "Commitments", + "As", + "Notes", + "12", + "Company", + "charges", + "cost", + "sales", + "inventory", + "provisions", + "write", + "excess", + "obsolete", + "product", + "purchase", + "commitments", + "Most", + "relate", + "quantities", + "products", + "based", + "levels", + "future", + "compared", + "assumptions", + "demand", + "including", + "impact", + "market", + "conditions", + "regulatory", + "export", + "restrictions", + "January", + "25", + "2026", + "inventories", + "balance", + "21.4", + "billion", + "outstanding", + "long", + "term", + "supply", + "capacity", + "obligations", + "95.2", + "significant", + "portion", + "principal", + "considerations", + "determination", + "performing", + "procedures", + "relating", + "valuation", + "specifically", + "judgment", + "management", + "developing", + "related", + "This", + "turn", + "led", + "auditor", + "subjectivity", + "effort", + "evaluating", + "Addressing", + "evidence", + "connection", + "forming", + "These", + "included", + "testing", + "effectiveness", + "controls", + "process", + "appropriateness", + "approach", + "completeness", + "accuracy", + "underlying", + "data", + "reasonableness", + "Evaluating", + "reasonable", + "past", + "results", + "historical", + "life", + "cycle", + "consistency", + "external", + "industry", + "iii", + "technology", + "PricewaterhouseCoopers", + "LLP", + "San", + "Jose", + "California", + "February", + "We", + "served", + "2004", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Consolidated", + "Statements", + "Income", + "In", + "millions", + "share", + "See", + "accompanying", + "Financial", + "Comprehensive", + "Balance", + "Sheets", + "par", + "Shareholders", + "Equity", + "Cash", + "Flows" + ], + "keywords": [ + "Inventory provisions", + "Critical audit", + "Future demand" + ], + "connect_to": [ + { + "target": "08ce51f5-def4-5e30-a687-c523ef580752", + "relation": "embeds", + "ref": "[tables/table-74 Income Statement.html]", + "position": { + "start": 3785, + "end": 3824 + } + }, + { + "target": "66d204c1-4792-5bae-8b7c-64445b5ea8fb", + "relation": "embeds", + "ref": "[tables/table-75 Comprehensive Income.html]", + "position": { + "start": 3990, + "end": 4033 + } + }, + { + "target": "51e474a3-da2a-55a0-b70a-b206f8b23f9e", + "relation": "embeds", + "ref": "[tables/table-76 Balance Sheet.html]", + "position": { + "start": 4197, + "end": 4233 + } + }, + { + "target": "23e7af72-eac0-52dd-a5b9-ae90e956bb81", + "relation": "embeds", + "ref": "[tables/table-77 Equity Changes.html]", + "position": { + "start": 4385, + "end": 4422 + } + }, + { + "target": "263389be-7d96-5da1-b4e9-9077ed5aba31", + "relation": "embeds", + "ref": "[tables/table-78 Cash Flow Statement.html]", + "position": { + "start": 4578, + "end": 4620 + } + } + ] + } + }, + { + "chunk_id": "feaf4ab3-cb6e-5109-b0ad-444a5fcca4ee", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5034ad6d-5bf8-52f5-9345-a1118163fa71", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fb24b70b-fc04-5e86-8a32-039669a234ee", + "type": "text", + "content": "Headquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998.\nAll references to “NVIDIA,” “we,” “us,” “our” or the “Company” mean NVIDIA Corporation and its subsidiaries.\nCertain prior fiscal year balances have been reclassified to conform to the current fiscal year presentation. Non-marketable equity securities, previously presented within other assets, were reclassified to be presented separately on our consolidated balance sheets and had no impact to total assets or consolidated statement of cash flows.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Our Company", + "metadata": { + "length": 589, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Headquartered", + "Santa", + "Clara", + "California", + "NVIDIA", + "incorporated", + "April", + "1993", + "reincorporated", + "Delaware", + "1998", + "All", + "references", + "Company", + "Corporation", + "subsidiaries", + "Certain", + "prior", + "fiscal", + "year", + "balances", + "reclassified", + "conform", + "current", + "presentation", + "Non", + "marketable", + "equity", + "securities", + "previously", + "presented", + "assets", + "separately", + "consolidated", + "balance", + "sheets", + "impact", + "total", + "statement", + "cash", + "flows" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a88b7248-2ac4-53c0-9924-03308619af6b", + "type": "text", + "content": "We operate on a 52- or 53-week year, ending on the last Sunday in January. Fiscal years 2026, 2025 and 2024 were all 52-week years. Fiscal year 2027 will be a 53-week year with the fourth quarter consisting of 14 weeks.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Fiscal Year", + "metadata": { + "length": 219, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "operate", + "52", + "53", + "week", + "year", + "ending", + "Sunday", + "January", + "Fiscal", + "years", + "2026", + "2025", + "2024", + "2027", + "fourth", + "quarter", + "consisting", + "14", + "weeks" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9962cfc5-d2f4-5a26-9f74-7a7ca4de7f96", + "type": "text", + "content": "Our consolidated financial statements include the accounts of NVIDIA Corporation and our wholly-owned subsidiaries. All intercompany balances and transactions have been eliminated in consolidation.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Principles of Consolidation", + "metadata": { + "length": 197, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "consolidated", + "financial", + "statements", + "include", + "accounts", + "NVIDIA", + "Corporation", + "wholly", + "owned", + "subsidiaries", + "All", + "intercompany", + "balances", + "transactions", + "eliminated", + "consolidation" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7ed41908-486e-5651-93cb-995ef65e37fe", + "type": "text", + "content": "The preparation of financial statements in conformity with U.S. GAAP requires management to make estimates and assumptions that affect the reported amounts of assets and liabilities and disclosures of contingent assets and liabilities at the date of the financial statements and the reported amounts of revenue and expenses during the reporting period. Actual results could differ materially from our estimates. On an on-going basis, we evaluate our estimates, including those related to accounts receivable, cash equivalents and marketable securities, goodwill, income taxes, inventories and product purchase commitments, investigation and settlement costs, litigation, non-marketable equity securities, other contingencies, property, plant, and equipment, revenue recognition, and stock-based compensation. These estimates are based on historical facts and various other assumptions that we believe are reasonable.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Use of Estimates", + "metadata": { + "length": 916, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "preparation", + "financial", + "statements", + "conformity", + "GAAP", + "requires", + "management", + "make", + "estimates", + "assumptions", + "affect", + "reported", + "amounts", + "assets", + "liabilities", + "disclosures", + "contingent", + "date", + "revenue", + "expenses", + "reporting", + "period", + "Actual", + "results", + "differ", + "materially", + "On", + "basis", + "evaluate", + "including", + "related", + "accounts", + "receivable", + "cash", + "equivalents", + "marketable", + "securities", + "goodwill", + "income", + "taxes", + "inventories", + "product", + "purchase", + "commitments", + "investigation", + "settlement", + "costs", + "litigation", + "equity", + "contingencies", + "property", + "plant", + "equipment", + "recognition", + "stock", + "based", + "compensation", + "These", + "historical", + "facts", + "reasonable" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "024a9c20-eaab-5187-b5d4-2313a75e838c", + "type": "text", + "content": "We derive our revenue primarily from product sales including hardware and systems. We determine revenue recognition through the following steps: (1) identification of the contract with a customer; (2) identification of the performance obligations in the contract; (3) determination of the transaction price; (4) allocation of the transaction price to the performance obligations in the contract (where revenue is allocated on a relative standalone selling price basis by maximizing the use of observable inputs to determine the standalone selling price for each performance obligation); and (5) recognition of revenue when, or as, we satisfy a performance obligation. Payment from customers, per our standard payment terms, is generally due shortly after delivery of our products.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Revenue Recognition", + "metadata": { + "length": 780, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "derive", + "revenue", + "primarily", + "product", + "sales", + "including", + "hardware", + "systems", + "determine", + "recognition", + "steps", + "identification", + "contract", + "customer", + "performance", + "obligations", + "determination", + "transaction", + "price", + "allocation", + "allocated", + "relative", + "standalone", + "selling", + "basis", + "maximizing", + "observable", + "inputs", + "obligation", + "satisfy", + "Payment", + "customers", + "standard", + "payment", + "terms", + "generally", + "due", + "shortly", + "delivery", + "products" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ba9a57f2-c1f8-598f-bab9-46bba6bfdabe", + "type": "text", + "content": "Revenue from product sales is recognized upon transfer of control of products to customers in an amount that reflects the consideration we expect to receive in exchange for those products. Certain products are sold with support or an extended warranty. Support and extended warranty revenue are recognized ratably over the service period, or as services are performed. Revenue is recognized net of allowances for returns, customer programs and any taxes collected from customers.\nFor products sold with a right of return, we record a reduction to revenue by establishing a sales return allowance for estimated product returns at the time revenue is recognized, based primarily on historical return rates. However, if product returns for a fiscal period are anticipated to exceed historical return rates, we may determine that additional sales return allowances are required to accurately reflect our estimated exposure for product returns.\nOur customer programs involve rebates, which are designed to serve as sales incentives to resellers of our products in various target markets, and MDFs which represent monies paid to our partners that are earmarked for market segment development and are designed to support our partners' activities while also promoting NVIDIA products. We account for customer programs as a reduction to revenue and accrue for such programs for potential rebates and MDFs based on the amount we expect to be claimed by customers.\nNVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Revenue Recognition/Product Sales Revenue", + "metadata": { + "length": 1548, + "summary": "Revenue from product sales is recognized when control transfers to customers, net of allowances for returns, customer programs, and taxes. Support and extended warranty revenue is recognized ratably over the service period. For products with a right of return, a sales return allowance is recorded based on historical return rates, with adjustments if returns exceed expectations. Customer programs, including rebates and market development funds (MDFs), are accounted for as a reduction to revenue and accrued based on expected claims.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Revenue", + "product", + "sales", + "recognized", + "transfer", + "control", + "products", + "customers", + "amount", + "reflects", + "consideration", + "expect", + "receive", + "exchange", + "Certain", + "sold", + "support", + "extended", + "warranty", + "Support", + "revenue", + "ratably", + "service", + "period", + "services", + "performed", + "net", + "allowances", + "returns", + "customer", + "programs", + "taxes", + "collected", + "For", + "return", + "record", + "reduction", + "establishing", + "allowance", + "estimated", + "time", + "based", + "primarily", + "historical", + "rates", + "However", + "fiscal", + "anticipated", + "exceed", + "determine", + "additional", + "required", + "accurately", + "reflect", + "exposure", + "Our", + "involve", + "rebates", + "designed", + "serve", + "incentives", + "resellers", + "target", + "markets", + "MDFs", + "represent", + "monies", + "paid", + "partners", + "earmarked", + "market", + "segment", + "development", + "activities", + "promoting", + "NVIDIA", + "We", + "account", + "accrue", + "potential", + "claimed", + "Corporation", + "Subsidiaries", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Continued" + ], + "keywords": [ + "Revenue", + "Returns", + "Customer Programs" + ], + "connect_to": [] + } + }, + { + "chunk_id": "e6eca556-c5f4-5d59-9b2e-28f16fb7b46b", + "type": "text", + "content": "Our contracts may contain more than one deliverable, each of which is separately accounted for as a distinct performance obligation. We account for multiple agreements with a single customer as a single contract if the contractual terms and/or substance of those agreements indicate that they may be so closely related that they are, in effect, parts of a single contract.\nWe allocate the total transaction price to each distinct performance obligation in an arrangement with multiple performance obligations on a relative standalone selling price basis. The standalone selling price reflects the price we would charge for a specific product or service if it were sold separately in similar circumstances and to similar customers. When determining standalone selling price, we maximize the use of observable inputs.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Revenue Recognition/Contracts with Multiple Performance Obligations", + "metadata": { + "length": 815, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "contracts", + "deliverable", + "separately", + "accounted", + "distinct", + "performance", + "obligation", + "We", + "account", + "multiple", + "agreements", + "single", + "customer", + "contract", + "contractual", + "terms", + "substance", + "closely", + "related", + "effect", + "parts", + "allocate", + "total", + "transaction", + "price", + "arrangement", + "obligations", + "relative", + "standalone", + "selling", + "basis", + "The", + "reflects", + "charge", + "specific", + "product", + "service", + "sold", + "similar", + "circumstances", + "customers", + "When", + "determining", + "maximize", + "observable", + "inputs" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d719bea2-a63b-52f6-aa54-e6b0f767a6ba", + "type": "text", + "content": "We offer a limited warranty to end-users ranging from one to three years for products to repair or replace products for manufacturing defects or hardware component failures. Cost of revenue includes the estimated cost of product warranties that are calculated at the point of revenue recognition. Under limited circumstances, we may offer an extended limited warranty to customers for certain products. We also accrue for known warranty and indemnification issues if a loss is probable and can be reasonably estimated.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Revenue Recognition/Product Warranties", + "metadata": { + "length": 518, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "offer", + "limited", + "warranty", + "end", + "users", + "ranging", + "years", + "products", + "repair", + "replace", + "manufacturing", + "defects", + "hardware", + "component", + "failures", + "Cost", + "revenue", + "includes", + "estimated", + "cost", + "product", + "warranties", + "calculated", + "point", + "recognition", + "Under", + "circumstances", + "extended", + "customers", + "accrue", + "indemnification", + "issues", + "loss", + "probable" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b3bc2f81-f1f7-5753-8c78-7f473530b57f", + "type": "text", + "content": "We use the closing trading price of our common stock on the date of grant, minus a dividend yield discount, as the fair value of awards of restricted stock units, or RSUs, and performance stock units, or PSUs, that are based on our corporate financial performance targets. We use a Monte Carlo simulation on the date of grant to estimate the fair value of PSUs that are based on our stock performance compared to market performance, or market-based PSUs. The compensation expense for RSUs and market-based PSUs is recognized using a straight-line attribution method over the requisite employee service period while compensation expense for PSUs is recognized using an accelerated amortization model based on performance targets probable of achievement. We estimate the fair value of shares to be issued under our employee stock purchase plan, or ESPP, using the Black-Scholes model at the commencement of an offering period in March and September of each year. Stock-based compensation for our ESPP is expensed using an accelerated amortization model. Additionally, for RSUs, PSUs, and market-based PSUs, we estimate expected forfeitures based on our historical forfeitures.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Stock-based Compensation", + "metadata": { + "length": 1174, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "closing", + "trading", + "price", + "common", + "stock", + "date", + "grant", + "minus", + "dividend", + "yield", + "discount", + "fair", + "awards", + "restricted", + "units", + "RSUs", + "performance", + "PSUs", + "based", + "corporate", + "financial", + "targets", + "Monte", + "Carlo", + "simulation", + "estimate", + "compared", + "market", + "The", + "compensation", + "expense", + "recognized", + "straight", + "line", + "attribution", + "method", + "requisite", + "employee", + "service", + "period", + "accelerated", + "amortization", + "model", + "probable", + "achievement", + "shares", + "issued", + "purchase", + "plan", + "ESPP", + "Black", + "Scholes", + "commencement", + "offering", + "March", + "September", + "year", + "Stock", + "expensed", + "Additionally", + "expected", + "forfeitures", + "historical" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "24f593ac-b10e-5c1c-b4bd-edc4366b9048", + "type": "text", + "content": "We currently are, and will likely continue to be subject to claims, litigation, and other actions, including potential regulatory proceedings, involving patent and other intellectual property matters, taxes, labor and employment, competition and antitrust, commercial disputes, goods and services offered by us and by third parties, and other matters. There are many uncertainties associated with any litigation or investigation, and we cannot be certain that these actions or other third-party claims against us will be resolved without litigation, fines and/or substantial settlement payments or judgments. If information becomes available that causes us to determine that a loss in any of our pending litigation, investigations or settlements is probable, and we can reasonably estimate the loss associated with such events, we will record the loss. However, the actual liability in any such litigation or investigation may be materially different from our estimates, which could require us to record additional costs. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the reasonably possible loss.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Litigation, Investigation and Settlement Costs", + "metadata": { + "length": 1162, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "continue", + "subject", + "claims", + "litigation", + "actions", + "including", + "potential", + "regulatory", + "proceedings", + "involving", + "patent", + "intellectual", + "property", + "matters", + "taxes", + "labor", + "employment", + "competition", + "antitrust", + "commercial", + "disputes", + "goods", + "services", + "offered", + "parties", + "There", + "uncertainties", + "investigation", + "party", + "resolved", + "fines", + "substantial", + "settlement", + "payments", + "judgments", + "If", + "information", + "determine", + "loss", + "pending", + "investigations", + "settlements", + "probable", + "estimate", + "events", + "record", + "However", + "actual", + "liability", + "materially", + "estimates", + "require", + "additional", + "costs", + "range", + "estimated", + "disclose" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "16976697-29f0-55a3-ba66-ce98e98cde7d", + "type": "text", + "content": "We use the U.S. dollar as our functional currency for our subsidiaries. Foreign currency monetary assets and liabilities are remeasured into United States dollars at end-of-period exchange rates. Non-monetary assets and liabilities such as property and equipment and equity are remeasured at historical exchange rates. Revenue and expenses are remeasured at exchange rates in effect during each period, except for those expenses related to non-monetary balance sheet amounts, which are remeasured at historical exchange rates. Gains or losses from foreign currency remeasurement are included in earnings in our Consolidated Statements of Income and to date have not been significant.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Foreign Currency Remeasurement", + "metadata": { + "length": 683, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "dollar", + "functional", + "currency", + "subsidiaries", + "Foreign", + "monetary", + "assets", + "liabilities", + "remeasured", + "United", + "States", + "dollars", + "end", + "period", + "exchange", + "rates", + "Non", + "property", + "equipment", + "equity", + "historical", + "Revenue", + "expenses", + "effect", + "related", + "balance", + "sheet", + "amounts", + "Gains", + "losses", + "foreign", + "remeasurement", + "included", + "earnings", + "Consolidated", + "Statements", + "Income", + "date", + "significant" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "df857b88-98a9-56fa-a0d9-1387b6699a84", + "type": "text", + "content": "We recognize federal, state and foreign current tax liabilities or assets based on our estimate of taxes payable or refundable in the current fiscal year by tax jurisdiction. We recognize federal, state and foreign deferred tax assets or liabilities, as appropriate, for our estimate of future tax effects attributable to temporary differences and carryforwards; and we record a valuation allowance to reduce any deferred tax assets by the amount of any tax benefits that, based on available evidence and judgment, are not expected to be realized.\nOur calculation of deferred tax assets and liabilities is based on certain estimates and judgments and involves dealing with uncertainties in the application of complex tax laws. Our estimates of deferred tax assets and liabilities may change\nNVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)\nbased, in part, on added certainty or finality to an anticipated outcome, changes in accounting standards or tax laws in the U.S., or foreign jurisdictions where we operate, or changes in other facts or circumstances. In addition, we recognize liabilities for potential U.S. and foreign income tax contingencies based on our estimate of whether, and the extent to which, additional taxes may be due. If we determine that payment of these amounts is unnecessary or if the recorded tax liability is less than our current assessment, we may be required to recognize an income tax benefit or additional income tax expense in our financial statements accordingly.\nAs of January 25, 2026, we had a valuation allowance of \\$768 million related to capital loss carryforwards, and certain other deferred tax assets that management determined are not likely to be realized due, in part, to jurisdictional projections of future taxable income, including capital gains. To the extent realization of the deferred tax assets becomes more-likely-than-not, we would recognize such deferred tax assets as income tax benefits during the period.\nWe recognize the benefit from a tax position only if it is more-likely-than-not that the position would be sustained upon audit based solely on the technical merits of the tax position. Our policy is to include interest and penalties related to unrecognized tax benefits as a component of income tax expense.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Income Taxes", + "metadata": { + "length": 2321, + "summary": "The passage discusses NVIDIA's recognition of current and deferred tax assets/liabilities based on estimates of taxes payable and future tax effects from temporary differences and carryforwards. A valuation allowance of $768 million is recorded for capital loss carryforwards and other deferred tax assets not likely to be realized. Tax benefits are recognized only if more-likely-than-not to be sustained, with interest and penalties included in income tax expense.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "recognize", + "federal", + "state", + "foreign", + "current", + "tax", + "liabilities", + "assets", + "based", + "estimate", + "taxes", + "payable", + "refundable", + "fiscal", + "year", + "jurisdiction", + "deferred", + "future", + "effects", + "attributable", + "temporary", + "differences", + "carryforwards", + "record", + "valuation", + "allowance", + "reduce", + "amount", + "benefits", + "evidence", + "judgment", + "expected", + "realized", + "Our", + "calculation", + "estimates", + "judgments", + "involves", + "dealing", + "uncertainties", + "application", + "complex", + "laws", + "change", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Continued", + "part", + "added", + "certainty", + "finality", + "anticipated", + "outcome", + "accounting", + "standards", + "jurisdictions", + "operate", + "facts", + "circumstances", + "In", + "addition", + "potential", + "income", + "contingencies", + "extent", + "additional", + "due", + "If", + "determine", + "payment", + "amounts", + "unnecessary", + "recorded", + "liability", + "assessment", + "required", + "benefit", + "expense", + "financial", + "statements", + "As", + "January", + "25", + "2026", + "768", + "million", + "related", + "capital", + "loss", + "management", + "determined", + "jurisdictional", + "projections", + "taxable", + "including", + "gains", + "To", + "realization", + "period", + "position", + "sustained", + "audit", + "solely", + "technical", + "merits", + "policy", + "include", + "interest", + "penalties", + "unrecognized", + "component" + ], + "keywords": [ + "deferred tax", + "valuation allowance", + "tax contingencies" + ], + "connect_to": [] + } + }, + { + "chunk_id": "ab7826a0-7f2d-508d-a202-7dd3237d0484", + "type": "text", + "content": "Basic net income per share is computed using the weighted average number of common shares outstanding during the period. Diluted net income per share is computed using the weighted average number of common and potentially dilutive shares outstanding during the period, using the treasury stock method. Any anti-dilutive effect of equity awards outstanding is not included in the computation of diluted net income per share.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Net Income Per Share", + "metadata": { + "length": 423, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Basic", + "net", + "income", + "share", + "computed", + "weighted", + "average", + "number", + "common", + "shares", + "outstanding", + "period", + "Diluted", + "potentially", + "dilutive", + "treasury", + "stock", + "method", + "Any", + "anti", + "effect", + "equity", + "awards", + "included", + "computation", + "diluted" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "89a1201c-c090-573b-adf8-ef46681d8ae9", + "type": "text", + "content": "We consider all highly liquid investments that are readily convertible into cash and have an original maturity of three months or less at the time of purchase to be cash equivalents. Marketable securities consist of highly liquid debt investments with maturities of greater than three months when purchased and publicly-held equity securities. We classify these investments as current or long term based on the nature of the investments and their availability for use in current operations.\nWe record our debt investments as cash equivalents and marketable debt securities and classify them at the date of acquisition as available-for-sale. These available-for-sale debt securities are reported at fair value with the related unrealized gains and losses included in accumulated other comprehensive income or loss, a component of shareholders' equity, net of tax. The fair value of interest-bearing debt securities includes accrued interest. Realized gains and losses on the sale of marketable securities are determined using the specific-identification method and recorded in the Other income (expense), net, section of our Consolidated Statements of Income.\nAvailable-for-sale debt securities are subject to impairment review. If the estimated fair value of available-for-sale debt securities is less than its amortized cost basis, we determine if the difference, if any, is caused by expected credit losses and write-down the amortized cost basis of the securities if it is more likely than not we will be required or we intend to sell the securities before recovery of its amortized cost basis. Allowances for credit losses and write-downs are recognized in the Other income, net, net section of our Consolidated Statements of Income.\nPublicly-held equity securities and money market funds have readily determinable fair values with changes in fair value recorded in Other income, net.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Cash and Cash Equivalents and Marketable Securities", + "metadata": { + "length": 1888, + "summary": "The passage defines cash equivalents as highly liquid investments with original maturities of three months or less. Marketable securities include debt investments with longer maturities and publicly-held equity securities. Debt investments are classified as available-for-sale, reported at fair value with unrealized gains/losses in accumulated other comprehensive income. Realized gains/losses use the specific-identification method. Available-for-sale debt securities undergo impairment review; credit losses or write-downs are recognized if fair value is below amortized cost and sale is likely. Publicly-held equity securities and money market funds have fair value changes recorded in Other income.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "highly", + "liquid", + "investments", + "readily", + "convertible", + "cash", + "original", + "maturity", + "months", + "time", + "purchase", + "equivalents", + "Marketable", + "securities", + "consist", + "debt", + "maturities", + "greater", + "purchased", + "publicly", + "held", + "equity", + "classify", + "current", + "long", + "term", + "based", + "nature", + "availability", + "operations", + "record", + "marketable", + "date", + "acquisition", + "sale", + "These", + "reported", + "fair", + "related", + "unrealized", + "gains", + "losses", + "included", + "accumulated", + "comprehensive", + "income", + "loss", + "component", + "shareholders", + "net", + "tax", + "The", + "interest", + "bearing", + "includes", + "accrued", + "Realized", + "determined", + "specific", + "identification", + "method", + "recorded", + "Other", + "expense", + "section", + "Consolidated", + "Statements", + "Income", + "Available", + "subject", + "impairment", + "review", + "If", + "estimated", + "amortized", + "cost", + "basis", + "determine", + "difference", + "caused", + "expected", + "credit", + "write", + "required", + "intend", + "sell", + "recovery", + "Allowances", + "downs", + "recognized", + "Publicly", + "money", + "market", + "funds", + "determinable", + "values" + ], + "keywords": [ + "cash equivalents", + "marketable securities", + "fair value" + ], + "connect_to": [] + } + }, + { + "chunk_id": "d1845171-61c7-58d7-8015-7ca0c36cb894", + "type": "text", + "content": "The carrying value of cash equivalents, accounts receivable, accounts payable and accrued liabilities approximate their fair values due to their relatively short maturities as of January 25, 2026 and January 26, 2025. Marketable debt and equity securities are reported at fair value. Derivative instruments are recognized as either assets or liabilities and are measured at fair value. The accounting for changes in the fair value of a derivative depends on the intended use of the derivative and the resulting designation. For derivative instruments designated as accounting hedges, the effective portion of the gains or losses on the derivatives is initially reported as a component of other comprehensive income or loss and is subsequently recognized in earnings when the hedged exposure is recognized in earnings. For derivative instruments not designated as accounting hedges, changes in fair value are recognized in earnings.\nFinancial instruments measured and disclosed at fair value are classified and disclosed based on the observability of inputs used in the determination of fair value as follows:\n• Level 1: Observable inputs such as quoted prices in active markets.\n- Level 2: Observable inputs other than Level 1 prices, such as quoted prices in less active markets or model-derived valuations that are observable either directly or indirectly.\nNVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)\n\\- Level 3: Unobservable inputs in which there is little or no market data that are significant to the fair value of the assets or liabilities.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Fair Value of Financial Instruments", + "metadata": { + "length": 1597, + "summary": "The passage discusses the fair value measurement of financial instruments, including cash equivalents, accounts receivable/payable, marketable securities, and derivatives. It explains that short-term items approximate fair value due to short maturities, while marketable securities and derivatives are reported at fair value. Derivatives are classified as accounting hedges (with gains/losses initially in other comprehensive income) or non-hedges (changes recognized in earnings). Fair value is classified into three levels based on input observability: Level 1 (quoted prices in active markets), Level 2 (observable inputs other than Level 1), and Level 3 (unobservable inputs with little market data).", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "carrying", + "cash", + "equivalents", + "accounts", + "receivable", + "payable", + "accrued", + "liabilities", + "approximate", + "fair", + "values", + "due", + "short", + "maturities", + "January", + "25", + "2026", + "26", + "2025", + "Marketable", + "debt", + "equity", + "securities", + "reported", + "Derivative", + "instruments", + "recognized", + "assets", + "measured", + "accounting", + "derivative", + "depends", + "intended", + "resulting", + "designation", + "For", + "designated", + "hedges", + "effective", + "portion", + "gains", + "losses", + "derivatives", + "initially", + "component", + "comprehensive", + "income", + "loss", + "subsequently", + "earnings", + "hedged", + "exposure", + "Financial", + "disclosed", + "classified", + "based", + "observability", + "inputs", + "determination", + "Level", + "Observable", + "quoted", + "prices", + "active", + "markets", + "model", + "derived", + "valuations", + "observable", + "directly", + "indirectly", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Consolidated", + "Statements", + "Continued", + "Unobservable", + "market", + "data", + "significant" + ], + "keywords": [ + "fair value", + "derivatives", + "financial instruments" + ], + "connect_to": [] + } + }, + { + "chunk_id": "9a9cf059-828c-5911-a009-a90d7f867451", + "type": "text", + "content": "Financial instruments that potentially subject us to concentrations of credit risk consist primarily of cash equivalents, marketable securities, lease guarantees, and accounts receivable. Our investment policy requires the purchase of highly-rated fixed income securities, the diversification of investment type and credit exposures, and includes certain limits on our portfolio maturities. We perform ongoing credit evaluations of our customers' financial condition and maintain an allowance for potential credit losses. This allowance consists of an amount identified for specific customers and an amount based on overall estimated exposure. Our overall estimated exposure excludes amounts covered by credit insurance and letters of credit.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Concentration of Credit Risk", + "metadata": { + "length": 742, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Financial", + "instruments", + "potentially", + "subject", + "concentrations", + "credit", + "risk", + "consist", + "primarily", + "cash", + "equivalents", + "marketable", + "securities", + "lease", + "guarantees", + "accounts", + "receivable", + "Our", + "investment", + "policy", + "requires", + "purchase", + "highly", + "rated", + "fixed", + "income", + "diversification", + "type", + "exposures", + "includes", + "limits", + "portfolio", + "maturities", + "We", + "perform", + "ongoing", + "evaluations", + "customers", + "financial", + "condition", + "maintain", + "allowance", + "potential", + "losses", + "This", + "consists", + "amount", + "identified", + "specific", + "based", + "estimated", + "exposure", + "excludes", + "amounts", + "covered", + "insurance", + "letters" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "416a78e6-100d-56f1-a59f-2f0abad7c632", + "type": "text", + "content": "Inventory cost is computed on an adjusted standard basis, which approximates actual cost on an average or first-in, first-out basis. Inventory costs consist of the cost of semiconductors, including wafer fabrication, assembly, testing and packaging, manufacturing support costs, including labor and overhead associated with such purchases, final test yield fallout, and shipping costs, as well as the cost of purchased memory products and other component parts. We charge cost of sales for inventory provisions to write-down our inventory to the lower of cost or net realizable value or for obsolete or excess inventory, and for excess product purchase commitments. Most of our inventory provisions relate to excess quantities of products, based on our inventory levels and future product purchase commitments compared to assumptions about future demand including the impact of regulatory export restrictions on our products. Once inventory has been written-off or written-down, it creates a new cost basis for the inventory that is not subsequently written-up. We record a liability for noncancelable purchase commitments with suppliers for quantities in excess of our future demand forecasts consistent with our valuation of obsolete or excess inventory.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Inventories", + "metadata": { + "length": 1256, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Inventory", + "cost", + "computed", + "adjusted", + "standard", + "basis", + "approximates", + "actual", + "average", + "costs", + "consist", + "semiconductors", + "including", + "wafer", + "fabrication", + "assembly", + "testing", + "packaging", + "manufacturing", + "support", + "labor", + "overhead", + "purchases", + "final", + "test", + "yield", + "fallout", + "shipping", + "purchased", + "memory", + "products", + "component", + "parts", + "We", + "charge", + "sales", + "inventory", + "provisions", + "write", + "lower", + "net", + "realizable", + "obsolete", + "excess", + "product", + "purchase", + "commitments", + "Most", + "relate", + "quantities", + "based", + "levels", + "future", + "compared", + "assumptions", + "demand", + "impact", + "regulatory", + "export", + "restrictions", + "Once", + "written", + "creates", + "subsequently", + "record", + "liability", + "noncancelable", + "suppliers", + "forecasts", + "consistent", + "valuation" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "51fa4587-33c4-5fb5-9f87-863fc5588d92", + "type": "text", + "content": "Property and equipment are stated at cost less accumulated depreciation. Depreciation of property and equipment is computed using the straight-line method based on the estimated useful lives of the assets of two to seven years. Once an asset is identified for retirement or disposition, the related cost and accumulated depreciation or amortization are removed, and a gain or loss is recorded. The estimated useful lives of our buildings are up to thirty years. Depreciation expense includes the amortization of assets recorded under finance leases. Leasehold improvements and assets recorded under finance leases are amortized over the shorter of the expected lease term or the estimated useful life of the asset.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Property and Equipment", + "metadata": { + "length": 714, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Property", + "equipment", + "stated", + "cost", + "accumulated", + "depreciation", + "Depreciation", + "property", + "computed", + "straight", + "line", + "method", + "based", + "estimated", + "lives", + "assets", + "years", + "Once", + "asset", + "identified", + "retirement", + "disposition", + "related", + "amortization", + "removed", + "gain", + "loss", + "recorded", + "The", + "buildings", + "thirty", + "expense", + "includes", + "finance", + "leases", + "Leasehold", + "improvements", + "amortized", + "shorter", + "expected", + "lease", + "term", + "life" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "af378bb1-cb2a-503a-a1ec-c0595bdbd200", + "type": "text", + "content": "We determine if an arrangement is or contains a lease at inception. Operating leases with lease terms of more than 12 months are included in operating lease assets, accrued and other current liabilities, and long-term operating lease liabilities on our consolidated balance sheet. Operating lease assets represent our right to use an underlying asset for the lease term and lease liabilities represent our obligation to make lease payments over the lease term. We combine lease and non-lease components for offices and data centers in determining the operating lease assets and liabilities.\nOperating lease assets and liabilities are recognized based on the present value of the remaining lease payments discounted using our incremental borrowing rate. Operating lease assets also include initial direct costs incurred and prepaid lease payments, minus any lease incentives. Our lease terms include options to extend or terminate the lease when it is reasonably certain that we will exercise that option. Lease costs are recognized on a straight-line basis over the lease term.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Leases", + "metadata": { + "length": 1077, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "determine", + "arrangement", + "lease", + "inception", + "Operating", + "leases", + "terms", + "12", + "months", + "included", + "operating", + "assets", + "accrued", + "current", + "liabilities", + "long", + "term", + "consolidated", + "balance", + "sheet", + "represent", + "underlying", + "asset", + "obligation", + "make", + "payments", + "combine", + "components", + "offices", + "data", + "centers", + "determining", + "recognized", + "based", + "present", + "remaining", + "discounted", + "incremental", + "borrowing", + "rate", + "include", + "initial", + "direct", + "costs", + "incurred", + "prepaid", + "minus", + "incentives", + "Our", + "options", + "extend", + "terminate", + "exercise", + "option", + "Lease", + "straight", + "line", + "basis" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "39cf6e9a-a6e4-5caf-b368-34b276a12493", + "type": "text", + "content": "We allocate goodwill to reporting units based on the expected benefit from the business combination. Goodwill is subject to our annual impairment test during the fourth quarter of our fiscal year, or earlier if indicators of potential impairment exist. In completing our impairment test, we perform either a qualitative or a quantitative analysis on a reporting unit basis.\nQualitative factors include industry and market considerations, overall financial performance, and other relevant events and factors affecting the reporting units. Goodwill impairments were not identified for the periods presented.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Goodwill", + "metadata": { + "length": 605, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "allocate", + "goodwill", + "reporting", + "units", + "based", + "expected", + "benefit", + "business", + "combination", + "Goodwill", + "subject", + "annual", + "impairment", + "test", + "fourth", + "quarter", + "fiscal", + "year", + "earlier", + "indicators", + "potential", + "exist", + "In", + "completing", + "perform", + "qualitative", + "quantitative", + "analysis", + "unit", + "basis", + "Qualitative", + "factors", + "include", + "industry", + "market", + "considerations", + "financial", + "performance", + "relevant", + "events", + "affecting", + "impairments", + "identified", + "periods", + "presented" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "de9d1295-a103-54f3-86dd-9a0b920eeebd", + "type": "text", + "content": "Intangible assets primarily represent acquired intangible assets including developed technology and customer relationships, as well as rights acquired under technology licenses, patents, and acquired IP. We currently amortize our intangible assets with finite lives over periods ranging from one to twenty years using a method that reflects the pattern\nNVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)\nin which the economic benefits of the intangible asset are consumed or otherwise used up or, if that pattern cannot be reliably determined, using a straight-line amortization method.\nLong-lived assets, such as property and equipment and intangible assets subject to amortization, are reviewed for impairment whenever events or changes in circumstances indicate that the carrying amount of an asset or asset group may not be recoverable. The recoverability of assets or asset groups to be held and used is measured by a comparison of the carrying amount of an asset or asset group to estimated undiscounted future cash flows expected to be generated by the asset or asset group. If the carrying amount of an asset or asset group exceeds its estimated future cash flows, an impairment charge is recognized for the amount by which the carrying amount of the asset or asset group exceeds the estimated fair value of the asset or asset group. Fair value is determined based on the estimated discounted future cash flows expected to be generated by the asset or asset group. Assets and liabilities to be disposed of would be separately presented in the Consolidated Balance Sheet and the assets would be reported at the lower of the carrying amount or fair value less costs to sell, and would no longer be depreciated.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Intangible Assets and Other Long-Lived Assets", + "metadata": { + "length": 1760, + "summary": "The passage discusses NVIDIA's intangible assets, including acquired technologies, customer relationships, and IP rights. These assets are amortized over one to twenty years using a method reflecting economic benefit consumption or straight-line. Long-lived assets are reviewed for impairment when events indicate carrying amounts may not be recoverable, comparing carrying amounts to undiscounted future cash flows. If impairment exists, a charge is recognized for the excess of carrying amount over fair value, determined by discounted cash flows. Assets held for disposal are reported at lower of carrying amount or fair value less costs to sell and are no longer depreciated.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Intangible", + "assets", + "primarily", + "represent", + "acquired", + "intangible", + "including", + "developed", + "technology", + "customer", + "relationships", + "rights", + "licenses", + "patents", + "IP", + "We", + "amortize", + "finite", + "lives", + "periods", + "ranging", + "twenty", + "years", + "method", + "reflects", + "pattern", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Continued", + "economic", + "benefits", + "asset", + "consumed", + "reliably", + "determined", + "straight", + "line", + "amortization", + "Long", + "lived", + "property", + "equipment", + "subject", + "reviewed", + "impairment", + "events", + "circumstances", + "carrying", + "amount", + "group", + "recoverable", + "The", + "recoverability", + "groups", + "held", + "measured", + "comparison", + "estimated", + "undiscounted", + "future", + "cash", + "flows", + "expected", + "generated", + "If", + "exceeds", + "charge", + "recognized", + "fair", + "Fair", + "based", + "discounted", + "Assets", + "liabilities", + "disposed", + "separately", + "presented", + "Balance", + "Sheet", + "reported", + "lower", + "costs", + "sell", + "longer", + "depreciated" + ], + "keywords": [ + "Intangible assets", + "amortization", + "impairment" + ], + "connect_to": [] + } + }, + { + "chunk_id": "1f2aa645-4a19-53c6-af0d-bf2f6b0711fd", + "type": "text", + "content": "The Company applies a screen test to evaluate if substantially all of the fair value of the gross assets acquired is concentrated in a single identifiable asset or group of similar identifiable assets to determine whether a transaction is accounted for as an asset acquisition or business combination. We allocate the fair value of the purchase price of an acquisition to the tangible assets acquired, liabilities assumed, and intangible assets acquired, based on their estimated fair values. The excess of the fair value of the purchase price over the fair values of these net tangible and intangible assets acquired is recorded as goodwill. Management's estimates of fair value are based upon assumptions believed to be reasonable, but our estimates and assumptions are inherently uncertain and subject to refinement. The estimates and assumptions used in valuing intangible assets include, but are not limited to, the amount and timing of projected future cash flows, discount rate used to determine the present value of these cash flows and asset lives. These estimates are inherently uncertain and, therefore, actual results may differ from the estimates made. As a result, during the measurement period of up to one year from the acquisition date, we may record adjustments to the assets acquired and liabilities assumed with the corresponding offset to goodwill. Upon the measurement period's conclusion or final determination of the fair value of the purchase price of an acquisition, whichever comes first, any subsequent adjustments are recorded to our Consolidated Statements of Income.\nAcquisition-related expenses are recognized separately from the business combination and expensed as incurred.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Business Combination", + "metadata": { + "length": 1708, + "summary": "The Company uses a screen test to determine if an acquisition is an asset acquisition or business combination based on fair value concentration. Purchase price is allocated to tangible and intangible assets and liabilities at fair value, with excess recorded as goodwill. Estimates involve uncertain assumptions like future cash flows and discount rates. Adjustments to assets and liabilities may be recorded during a one-year measurement period, with subsequent changes recognized in income. Acquisition-related expenses are expensed separately.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "Company", + "applies", + "screen", + "test", + "evaluate", + "substantially", + "fair", + "gross", + "assets", + "acquired", + "concentrated", + "single", + "identifiable", + "asset", + "group", + "similar", + "determine", + "transaction", + "accounted", + "acquisition", + "business", + "combination", + "We", + "allocate", + "purchase", + "price", + "tangible", + "liabilities", + "assumed", + "intangible", + "based", + "estimated", + "values", + "excess", + "net", + "recorded", + "goodwill", + "Management", + "estimates", + "assumptions", + "believed", + "reasonable", + "inherently", + "uncertain", + "subject", + "refinement", + "valuing", + "include", + "limited", + "amount", + "timing", + "projected", + "future", + "cash", + "flows", + "discount", + "rate", + "present", + "lives", + "These", + "actual", + "results", + "differ", + "made", + "As", + "result", + "measurement", + "period", + "year", + "date", + "record", + "adjustments", + "offset", + "Upon", + "conclusion", + "final", + "determination", + "whichever", + "subsequent", + "Consolidated", + "Statements", + "Income", + "Acquisition", + "related", + "expenses", + "recognized", + "separately", + "expensed", + "incurred" + ], + "keywords": [ + "business combination", + "fair value", + "goodwill" + ], + "connect_to": [] + } + }, + { + "chunk_id": "c0c4bb05-e784-508a-a502-450a1fbd744c", + "type": "text", + "content": "Non-marketable equity securities consist of investments in privately-held companies that do not have a readily determinable fair value. These investments are measured at cost minus impairment, if any, and are adjusted for changes resulting from observable price changes in orderly transactions for an identical or similar investment in the same issuer, or the measurement alternative. Fair value is based upon observable inputs in an inactive market and the valuation requires our judgment due to the absence of market prices and inherent lack of liquidity. All gains and losses on these investments, realized and unrealized, are recognized in Other income, net on our Consolidated Statements of Income.\nWe assess whether an impairment loss has occurred on our investments in non-marketable equity securities, accounted for under the measurement alternative based on quantitative and qualitative factors. If any impairment is identified for non-marketable equity securities, we write down the investment to its fair value and record the corresponding charge through Other income, net on our Consolidated Statements of Income.\nThe Company assesses its investments for significant influence to determine the appropriate method of accounting, including application of the equity method. Equity method investments were not material.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Non-Marketable Equity Securities", + "metadata": { + "length": 1328, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Non", + "marketable", + "equity", + "securities", + "consist", + "investments", + "privately", + "held", + "companies", + "readily", + "determinable", + "fair", + "These", + "measured", + "cost", + "minus", + "impairment", + "adjusted", + "resulting", + "observable", + "price", + "orderly", + "transactions", + "identical", + "similar", + "investment", + "issuer", + "measurement", + "alternative", + "Fair", + "based", + "inputs", + "inactive", + "market", + "valuation", + "requires", + "judgment", + "due", + "absence", + "prices", + "inherent", + "lack", + "liquidity", + "All", + "gains", + "losses", + "realized", + "unrealized", + "recognized", + "Other", + "income", + "net", + "Consolidated", + "Statements", + "Income", + "We", + "assess", + "loss", + "occurred", + "accounted", + "quantitative", + "qualitative", + "factors", + "If", + "identified", + "write", + "record", + "charge", + "The", + "Company", + "assesses", + "significant", + "influence", + "determine", + "method", + "accounting", + "including", + "application", + "Equity", + "material" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7da60aea-cf21-5cbe-98af-b56c7504d3fc", + "type": "text", + "content": "Recent Accounting Pronouncements Not Yet Adopted\nIn November 2024, the Financial Accounting Standards Board, or FASB, issued a new accounting standard requiring disclosures of certain additional expense information on an annual and interim basis, including, among other items, the amounts of purchases of inventory, employee compensation, depreciation and intangible asset amortization included within each income statement expense caption, as applicable. We will adopt this standard in the fiscal year 2028 annual report. We do not expect the adoption of this standard to have a material impact on our Consolidated Financial Statements other than additional disclosures.\nNVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Recently Issued Accounting Pronouncements", + "metadata": { + "length": 766, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Recent", + "Accounting", + "Pronouncements", + "Not", + "Yet", + "Adopted", + "In", + "November", + "2024", + "Financial", + "Standards", + "Board", + "FASB", + "issued", + "accounting", + "standard", + "requiring", + "disclosures", + "additional", + "expense", + "information", + "annual", + "interim", + "basis", + "including", + "items", + "amounts", + "purchases", + "inventory", + "employee", + "compensation", + "depreciation", + "intangible", + "asset", + "amortization", + "included", + "income", + "statement", + "caption", + "applicable", + "We", + "adopt", + "fiscal", + "year", + "2028", + "report", + "expect", + "adoption", + "material", + "impact", + "Consolidated", + "Statements", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Continued" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3991254d-1a44-541a-97d1-2df8c38a0cc0", + "type": "text", + "content": "In December 2025, we entered into a non-exclusive license agreement with Groq, Inc., or Groq, for its language processing unit technology and hired certain Groq employees. No customer contracts, existing products, or equity interests were purchased. We recorded \\$14.4 billion of goodwill and a \\$2.5 billion developed technology intangible asset, valued using a cost-to-recreate methodology with a five-year useful life. Goodwill, primarily attributable to the workforce and future development of the licensed technology, was recorded in the Compute & Networking reporting unit. Total consideration consists of \\$13.0 billion paid at closing and \\$4 billion, inclusive of imputed interest, payable within one year included in Accrued and Other Current Liabilities on our Consolidated Balance Sheets. The goodwill is tax deductible. Pro forma results of operations have not been presented because the effect was not material.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 2 - Groq", + "metadata": { + "length": 925, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "In", + "December", + "2025", + "entered", + "exclusive", + "license", + "agreement", + "Groq", + "Inc", + "language", + "processing", + "unit", + "technology", + "hired", + "employees", + "No", + "customer", + "contracts", + "existing", + "products", + "equity", + "interests", + "purchased", + "We", + "recorded", + "14.4", + "billion", + "goodwill", + "2.5", + "developed", + "intangible", + "asset", + "valued", + "cost", + "recreate", + "methodology", + "year", + "life", + "Goodwill", + "primarily", + "attributable", + "workforce", + "future", + "development", + "licensed", + "Compute", + "Networking", + "reporting", + "Total", + "consideration", + "consists", + "13.0", + "paid", + "closing", + "inclusive", + "imputed", + "interest", + "payable", + "included", + "Accrued", + "Other", + "Current", + "Liabilities", + "Consolidated", + "Balance", + "Sheets", + "The", + "tax", + "deductible", + "Pro", + "forma", + "results", + "operations", + "presented", + "effect", + "material" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "111e8adc-e23d-5b9e-bee6-4304c8ffc909", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Cost of revenue$ 261$ 178$ 141
Research and development4,6763,4232,532
Sales, general and administrative1,4491,136876
Total$ 6,386$ 4,737$ 3,549
", + "path": "tables/table-79 Cost Breakdown.html", + "metadata": { + "length": 512, + "summary": "table-80\nThe table shows cost of revenue, R&D, and SG&A expenses for fiscal years ending Jan 2024, 2025, and 2026, with totals increasing from $3,549M to $6,386M.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-79 Cost Breakdown.html", + "keywords": [ + "Cost", + "Revenue", + "Expenses" + ], + "tokens": [] + } + }, + { + "chunk_id": "2bc46377-3abb-539b-8c0f-d9588cdd71ab", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions, except per share data)
RSUs, PSUs and Market-based PSUs
Awards granted7089140
Estimated total grant-date fair value$ 9,389$ 7,834$ 5,316
Weighted average grant-date fair value per share$ 133.97$ 87.99$ 37.41
ESPP
Shares purchased133030
Weighted average price per share$ 49.13$ 17.74$ 15.81
Weighted average grant-date fair value per share$ 20.75$ 8.61$ 6.99
", + "path": "tables/table-80 Equity Awards.html", + "metadata": { + "length": 871, + "summary": "table-81\nThe table shows equity awards (RSUs, PSUs, ESPP) for fiscal years ending Jan 2024-2026, including shares granted/purchased and fair values per share.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-80 Equity Awards.html", + "keywords": [ + "RSUs", + "ESPP", + "Grant-Date Fair Value" + ], + "tokens": [] + } + }, + { + "chunk_id": "89ddd44e-f058-532a-897d-53f4b8c8fce1", + "type": "text", + "content": "We recognize stock-based compensation expense from grants of restricted stock units, or RSUs, performance stock units, or PSUs, and market-based PSUs, and issuances under our employee stock purchase plan, or ESPP.\nConsolidated Statements of Income include stock-based compensation expense as follows:\n\n[tables/table-79 Cost Breakdown.html]\n\nThe following is a summary of equity awards granted under our equity incentive plans:\n\n[tables/table-80 Equity Awards.html]\n\nAs of January 25, 2026, aggregate unearned stock-based compensation expense was \\$14.8 billion, which is expected to be recognized over a weighted average period of 2.3 years for RSUs, PSUs, and market-based PSUs, and 0.9 years for ESPP.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 3 - Stock-Based Compensation", + "metadata": { + "length": 696, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "recognize", + "stock", + "based", + "compensation", + "expense", + "grants", + "restricted", + "units", + "RSUs", + "performance", + "PSUs", + "market", + "issuances", + "employee", + "purchase", + "plan", + "ESPP", + "Consolidated", + "Statements", + "Income", + "include", + "The", + "summary", + "equity", + "awards", + "granted", + "incentive", + "plans", + "As", + "January", + "25", + "2026", + "aggregate", + "unearned", + "14.8", + "billion", + "expected", + "recognized", + "weighted", + "average", + "period", + "2.3", + "years", + "0.9" + ], + "keywords": [], + "connect_to": [ + { + "target": "111e8adc-e23d-5b9e-bee6-4304c8ffc909", + "relation": "embeds", + "ref": "[tables/table-79 Cost Breakdown.html]", + "position": { + "start": 302, + "end": 339 + } + }, + { + "target": "2bc46377-3abb-539b-8c0f-d9588cdd71ab", + "relation": "embeds", + "ref": "[tables/table-80 Equity Awards.html]", + "position": { + "start": 428, + "end": 464 + } + } + ] + } + }, + { + "chunk_id": "1a2c4d52-bcc0-58c6-836e-5581357fa71e", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(Using the Black-Scholes model)
ESPP
Weighted average expected life (in years)0.1-2.00.1-2.00.1-2.0
Risk-free interest rate3.5%-4.3%3.6%-5.4%3.9%-5.5%
Volatility26%-96%31%-75%31%-67%
Dividend yield0.03%0.03%0.06%
", + "path": "tables/table-81 ESPP Option Data.html", + "metadata": { + "length": 595, + "summary": "table-82\nThe table shows ESPP option data for fiscal years ending Jan 2024-2026, using the Black-Scholes model. Key inputs include weighted average expected life (0.1-2.0 years), risk-free interest rates (3.5%-5.5%), volatility (26%-96%), and dividend yield (0.03%-0.06%).", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-81 ESPP Option Data.html", + "keywords": [ + "ESPP", + "Black-Scholes", + "Volatility" + ], + "tokens": [] + } + }, + { + "chunk_id": "7a30a2fd-4045-56ef-98db-2eda226f8da6", + "type": "text", + "content": "(Continued)\nThe fair value of shares issued under our ESPP has been estimated with the following assumptions:\n\n[tables/table-81 ESPP Option Data.html]\n\nFor ESPP shares, the expected term represents the average term from the first day of the offering period to the purchase date. The risk-free interest rate assumption used to value ESPP shares is based upon observed interest rates on Treasury bills appropriate for the expected term. Our expected stock price volatility assumption for ESPP is estimated using historical volatility. For awards granted, we use the dividend yield at grant date. Our RSUs, PSUs, and market-based PSUs are not eligible for cash dividends prior to vesting; therefore, the fair values of RSUs, PSUs, and market-based PSUs are discounted for the dividend yield.\nAdditionally, for RSUs, PSUs, and market-based PSUs, we estimate expected forfeitures based on our historical forfeitures.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements", + "metadata": { + "length": 905, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "The", + "fair", + "shares", + "issued", + "ESPP", + "estimated", + "assumptions", + "For", + "expected", + "term", + "represents", + "average", + "day", + "offering", + "period", + "purchase", + "date", + "risk", + "free", + "interest", + "rate", + "assumption", + "based", + "observed", + "rates", + "Treasury", + "bills", + "Our", + "stock", + "price", + "volatility", + "historical", + "awards", + "granted", + "dividend", + "yield", + "grant", + "RSUs", + "PSUs", + "market", + "eligible", + "cash", + "dividends", + "prior", + "vesting", + "values", + "discounted", + "Additionally", + "estimate", + "forfeitures" + ], + "keywords": [], + "connect_to": [ + { + "target": "1a2c4d52-bcc0-58c6-836e-5581357fa71e", + "relation": "embeds", + "ref": "[tables/table-81 ESPP Option Data.html]", + "position": { + "start": 111, + "end": 150 + } + } + ] + } + }, + { + "chunk_id": "2bd62924-0aed-57a8-a3fa-acf9e508b0a3", + "type": "text", + "content": "We grant RSUs, PSUs, market-based PSUs, and stock purchase rights under the following equity incentive plans. In addition, in connection with our acquisitions of various companies, we have assumed certain stock-based awards granted under their stock incentive plans and converted them into our RSUs.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Equity Incentive Program", + "metadata": { + "length": 299, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "grant", + "RSUs", + "PSUs", + "market", + "based", + "stock", + "purchase", + "rights", + "equity", + "incentive", + "plans", + "In", + "addition", + "connection", + "acquisitions", + "companies", + "assumed", + "awards", + "granted", + "converted" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6c52f209-3b62-5f77-a6e7-962bff35c4f2", + "type": "text", + "content": "The NVIDIA Corporation Amended and Restated 2007 Equity Incentive Plan, or the 2007 Plan, authorizes the issuance of incentive stock options, non-statutory stock options, restricted stock, RSUs, stock appreciation rights, performance stock awards, performance cash awards, and other stock-based awards to employees, directors and consultants. Only our employees may receive incentive stock options. We grant RSUs, PSUs and market-based PSUs under the 2007 Plan. As of January 25, 2026, up to 192 million shares of our common stock could be issued pursuant to stock awards granted under the 2007 Plan, and 1.3 billion shares were available for future grants.\nSubject to certain exceptions, RSUs vest generally over four years subject to continued service. PSUs vest over four years, subject to continued service and performance conditions. Market-based PSUs vest on approximately the third anniversary of the date of grant subject to market conditions. However, the number of shares subject to both PSUs and market-based PSUs that are eligible to vest is determined by the Compensation Committee based on achievement of predetermined criteria.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Equity Incentive Program/Amended and Restated 2007 Equity Incentive Plan", + "metadata": { + "length": 1142, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "NVIDIA", + "Corporation", + "Amended", + "Restated", + "2007", + "Equity", + "Incentive", + "Plan", + "authorizes", + "issuance", + "incentive", + "stock", + "options", + "statutory", + "restricted", + "RSUs", + "appreciation", + "rights", + "performance", + "awards", + "cash", + "based", + "employees", + "directors", + "consultants", + "Only", + "receive", + "We", + "grant", + "PSUs", + "market", + "As", + "January", + "25", + "2026", + "192", + "million", + "shares", + "common", + "issued", + "pursuant", + "granted", + "1.3", + "billion", + "future", + "grants", + "Subject", + "exceptions", + "vest", + "generally", + "years", + "subject", + "continued", + "service", + "conditions", + "Market", + "approximately", + "anniversary", + "date", + "However", + "number", + "eligible", + "determined", + "Compensation", + "Committee", + "achievement", + "predetermined", + "criteria" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "840d1611-7aa5-5445-8a2b-6680995faff0", + "type": "text", + "content": "Employees who participate in the NVIDIA Corporation Amended and Restated 2012 Employee Stock Purchase Plan, or as most recently amended and restated, the 2012 Plan, may have up to 25% of their earnings withheld to purchase shares of common stock. The Board may decrease this percentage at its discretion. Each offering period is about 24 months, divided into four purchase periods of six months. The price of common stock purchased under our 2012 Plan will be equal to 85% of the lower of the fair market value of the common stock on the commencement date of each offering period or the fair market value of the common stock on each purchase date within the offering. As of January 25, 2026, we had 2.2 billion shares reserved for future issuance under the 2012 Plan.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Equity Incentive Program/Amended and Restated 2012 Employee Stock Purchase Plan", + "metadata": { + "length": 767, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Employees", + "participate", + "NVIDIA", + "Corporation", + "Amended", + "Restated", + "2012", + "Employee", + "Stock", + "Purchase", + "Plan", + "recently", + "amended", + "restated", + "25%", + "earnings", + "withheld", + "purchase", + "shares", + "common", + "stock", + "The", + "Board", + "decrease", + "percentage", + "discretion", + "Each", + "offering", + "period", + "24", + "months", + "divided", + "periods", + "price", + "purchased", + "equal", + "85%", + "lower", + "fair", + "market", + "commencement", + "date", + "As", + "January", + "25", + "2026", + "2.2", + "billion", + "reserved", + "future", + "issuance" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b148f1e4-a4f2-5cbc-9a46-7a80f90e529a", + "type": "table", + "content": "
RSUs, PSUs and Market-based PSUs Outstanding
Number of SharesWeighted Average Grant-Date Fair Value Per Share
(In millions, except per share data)
Balance as of Jan 26, 2025274$ 44.75
Granted70$ 133.97
Vested(146)$ 39.14
Canceled and forfeited(9)$ 59.29
Balance as of Jan 25, 2026189$ 81.51
Vested and expected to vest after Jan 25, 2026188$ 81.15
", + "path": "tables/table-82 RSU_PSU Activity.html", + "metadata": { + "length": 678, + "summary": "table-83\nThe table shows RSU, PSU, and market-based PSU activity from Jan 26, 2025 to Jan 25, 2026, with shares granted, vested, canceled, and ending balance of 189 million shares at $81.51 weighted average fair value.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-82 RSU_PSU Activity.html", + "keywords": [ + "RSU", + "PSU", + "Shares" + ], + "tokens": [] + } + }, + { + "chunk_id": "4f1fcd61-9bc4-5432-a6fb-43ff35e558c0", + "type": "text", + "content": "(Continued)\nEquity Award Activity\nThe following is a summary of our equity award transactions under our equity incentive plans:\n\n[tables/table-82 RSU_PSU Activity.html]\n\nAs of January 25, 2026 and January 26, 2025, there were 1.3 billion and 1.4 billion shares, respectively, of common stock available for future grants under our equity incentive plans.\nThe total fair value of RSUs and PSUs, as of their respective vesting dates, during the years ended January 25, 2026, January 26, 2025, and January 28, 2024, was \\$22.2 billion, \\$15.1 billion, and \\$8.2 billion, respectively.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_2", + "metadata": { + "length": 587, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "Equity", + "Award", + "Activity", + "The", + "summary", + "equity", + "award", + "transactions", + "incentive", + "plans", + "As", + "January", + "25", + "2026", + "26", + "2025", + "1.3", + "billion", + "1.4", + "shares", + "common", + "stock", + "future", + "grants", + "total", + "fair", + "RSUs", + "PSUs", + "respective", + "vesting", + "dates", + "years", + "ended", + "28", + "2024", + "22.2", + "15.1", + "8.2" + ], + "keywords": [], + "connect_to": [ + { + "target": "b148f1e4-a4f2-5cbc-9a46-7a80f90e529a", + "relation": "embeds", + "ref": "[tables/table-82 RSU_PSU Activity.html]", + "position": { + "start": 129, + "end": 168 + } + } + ] + } + }, + { + "chunk_id": "a09ecf37-2422-5b9c-b453-96eb4e4cb7ee", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions, except per share data)
Numerator:
Net income$ 120,067$ 72,880$ 29,760
Denominator:
Basic weighted average shares24,35924,55524,690
Dilutive impact of outstanding equity awards155249250
Diluted weighted average shares24,51424,80424,940
Net income per share:
Basic (1)$ 4.93$ 2.97$ 1.21
Diluted (2)$ 4.90$ 2.94$ 1.19
Anti-dilutive equity awards excluded from diluted net income per share4151150
", + "path": "tables/table-83 EPS Data.html", + "metadata": { + "length": 979, + "summary": "table-84\nThe table shows net income, weighted average shares, and basic/diluted EPS for fiscal years ending Jan 2026, 2025, and 2024.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-83 EPS Data.html", + "keywords": [ + "Net income", + "Shares", + "EPS" + ], + "tokens": [] + } + }, + { + "chunk_id": "6515313a-173b-57dd-b960-9f8cc99d8d41", + "type": "text", + "content": "The following is the basic and diluted net income per share computations for the periods presented:\n\n[tables/table-83 EPS Data.html]\n\n(1) Net income divided by basic weighted average shares.\n(2) Net income divided by diluted weighted average shares.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 4 - Net Income Per Share", + "metadata": { + "length": 251, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "basic", + "diluted", + "net", + "income", + "share", + "computations", + "periods", + "presented", + "Net", + "divided", + "weighted", + "average", + "shares" + ], + "keywords": [], + "connect_to": [ + { + "target": "a09ecf37-2422-5b9c-b453-96eb4e4cb7ee", + "relation": "embeds", + "ref": "[tables/table-83 EPS Data.html]", + "position": { + "start": 101, + "end": 132 + } + } + ] + } + }, + { + "chunk_id": "7b0266db-fb5f-55d6-969b-9475c1ee909c", + "type": "text", + "content": "As of January 25, 2026, the total carrying amount of goodwill was \\$20.8 billion, consisting of goodwill balances allocated to our Compute & Networking and Graphics reporting units of \\$20.5 billion and \\$370 million, respectively. As of January 26, 2025, the total carrying amount of goodwill was \\$5.2 billion, consisting of goodwill balances allocated to our Compute & Networking and Graphics reporting units of \\$4.8 billion and \\$370 million, respectively. Goodwill increased by \\$15.6 billion in fiscal year 2026 and was allocated to our Compute & Networking reporting unit. During the fourth quarters of fiscal years 2026, 2025, and 2024, we completed our annual qualitative impairment tests and concluded that goodwill was not impaired.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 5 - Goodwill", + "metadata": { + "length": 744, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "As", + "January", + "25", + "2026", + "total", + "carrying", + "amount", + "goodwill", + "20.8", + "billion", + "consisting", + "balances", + "allocated", + "Compute", + "Networking", + "Graphics", + "reporting", + "units", + "20.5", + "370", + "million", + "26", + "2025", + "5.2", + "4.8", + "Goodwill", + "increased", + "15.6", + "fiscal", + "year", + "unit", + "During", + "fourth", + "quarters", + "years", + "2024", + "completed", + "annual", + "qualitative", + "impairment", + "tests", + "concluded", + "impaired" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6c9c9797-7d50-5c4c-8898-f1ec7b7a4dd4", + "type": "text", + "content": "(Continued)", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_3", + "metadata": { + "length": 11, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c8334bb3-384e-572c-b4b1-79b0c73b385d", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
Gross Carrying AmountAccumulated AmortizationNet Carrying AmountGross Carrying AmountAccumulated AmortizationNet Carrying Amount
(In millions)
Acquisition-related intangible assets$ 5,656$ (2,580)$ 3,076$ 2,900$ (2,264)$ 636
Patents and licensed technology528(298)230449(278)171
Total intangible assets$ 6,184$ (2,878)$ 3,306$ 3,349$ (2,542)$ 807
", + "path": "tables/table-84 Intangible Assets.html", + "metadata": { + "length": 762, + "summary": "table-85\nThe table shows intangible assets as of Jan 25, 2026 and Jan 26, 2025, with gross, accumulated amortization, and net carrying amounts for acquisition-related assets, patents, and totals.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-84 Intangible Assets.html", + "keywords": [ + "intangible assets", + "amortization", + "carrying amount" + ], + "tokens": [] + } + }, + { + "chunk_id": "e181e1c5-53bf-5d0b-97f3-b6e368829143", + "type": "table", + "content": "
Future Amortization Expense(In millions)
Fiscal Year:
2027$ 923
2028729
2029592
2030511
2031468
2032 and thereafter83
Total$ 3,306
", + "path": "tables/table-85 Amortization Expense.html", + "metadata": { + "length": 383, + "summary": "table-86\nThe table shows future amortization expenses from 2027 to 2032 and thereafter, totaling $3,306 million, with the highest in 2027 at $923 million.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-85 Amortization Expense.html", + "keywords": [ + "amortization", + "future expense", + "fiscal years" + ], + "tokens": [] + } + }, + { + "chunk_id": "38f5d413-b6d1-52ea-837a-72c720863f6e", + "type": "text", + "content": "The components of our amortizable intangible assets are as follows:\n\n[tables/table-84 Intangible Assets.html]\n\nAmortization expense associated with intangible assets for fiscal years 2026, 2025, and 2024 was \\$488 million, \\$593 million, and \\$614 million, respectively.\nThe following table outlines the estimated future amortization expense related to the net carrying amount of intangible assets as of January 25, 2026:\n\n[tables/table-85 Amortization Expense.html]", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 6 - Amortizable Intangible Assets", + "metadata": { + "length": 477, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "components", + "amortizable", + "intangible", + "assets", + "Amortization", + "expense", + "fiscal", + "years", + "2026", + "2025", + "2024", + "488", + "million", + "593", + "614", + "table", + "outlines", + "estimated", + "future", + "amortization", + "related", + "net", + "carrying", + "amount", + "January", + "25" + ], + "keywords": [], + "connect_to": [ + { + "target": "c8334bb3-384e-572c-b4b1-79b0c73b385d", + "relation": "embeds", + "ref": "[tables/table-84 Intangible Assets.html]", + "position": { + "start": 69, + "end": 109 + } + }, + { + "target": "e181e1c5-53bf-5d0b-97f3-b6e368829143", + "relation": "embeds", + "ref": "[tables/table-85 Amortization Expense.html]", + "position": { + "start": 423, + "end": 466 + } + } + ] + } + }, + { + "chunk_id": "6c9c9797-7d50-5c4c-8898-f1ec7b7a4dd4", + "type": "text", + "content": "(Continued)", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_4", + "metadata": { + "length": 11, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2a7db966-8d15-5579-ae85-e26396e44f94", + "type": "table", + "content": "
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable SecuritiesOther Assets
(In millions)
Debt securities issued by the U.S. TreasuryLevel 2$ 21,635$ 77$ (3)$ 21,709$ —$ 21,709$ —
Corporate debt securitiesLevel 215,41092(3)15,49934515,154
Debt securities issued by U.S. government agenciesLevel 22,15742,1612,161
Certificates of depositLevel 2110110110
Foreign government bondsLevel 24014141
Money market fundsLevel 17,8307,8307,830
Publicly-held equity securities (1) (2)Level 117,72612,8864,840
Total$ 47,182$ 174$ (6)$ 65,076$ 8,285$ 51,951$ 4,840
", + "path": "tables/table-86 Debt Securities.html", + "metadata": { + "length": 1587, + "summary": "table-87\nThe table details debt securities by pricing category, cost, unrealized gains/losses, fair value, and reporting classification as cash equivalents, marketable securities, or other assets.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-86 Debt Securities.html", + "keywords": [ + "Debt securities", + "Fair value", + "Pricing levels" + ], + "tokens": [] + } + }, + { + "chunk_id": "38a6ef38-14bb-59d9-b67e-ab6b5762821e", + "type": "text", + "content": "The fair values of our financial assets are determined using quoted market prices of identical assets or market prices of similar assets from active markets. We review fair value classification on a quarterly basis. The following is a summary of cash equivalents and marketable securities:\nJan 25, 2026\n\n[tables/table-86 Debt Securities.html]\n\n(1) In the first quarter of fiscal year 2026, one investment was reclassified from non-marketable equity securities to marketable securities following public market trading. The balance as of January 25, 2026 includes \\$10.5 billion of investments which are subject to short-term lock-up restrictions on the ability to sell.\n(2) The long-term portion of marketable equity securities, which are subject to lock-up restrictions through December 2027 of \\$4.8 billion as of January 25, 2026, is included in other assets.\nPublicly-held equity securities are subject to market price volatility. Net unrealized gains on investments in publicly-held equity securities held at period end were \\$6.6 billion for fiscal year 2026. Net unrealized gains on investments in publicly-held equity securities held at period end were not significant for fiscal years 2025 and 2024.\nNet realized gains on investments in publicly-held equity securities sold were not significant for fiscal years 2026, 2025, and 2024, reflecting the difference between the sale proceeds and the carrying value of the equity securities at the beginning of the period or the purchase date, if later.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 7 - Cash Equivalents and Marketable Securities", + "metadata": { + "length": 1569, + "summary": "The passage discusses the fair value determination of financial assets using quoted market prices. It summarizes cash equivalents and marketable securities as of January 25, 2026, noting reclassification of an investment from non-marketable to marketable equity securities, with $10.5 billion subject to short-term lock-up restrictions and $4.8 billion in long-term lock-up restrictions through December 2027. Net unrealized gains on publicly-held equity securities were $6.6 billion for fiscal year 2026, while net realized gains were not significant for fiscal years 2024-2026.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "fair", + "values", + "financial", + "assets", + "determined", + "quoted", + "market", + "prices", + "identical", + "similar", + "active", + "markets", + "We", + "review", + "classification", + "quarterly", + "basis", + "summary", + "cash", + "equivalents", + "marketable", + "securities", + "Jan", + "25", + "2026", + "In", + "quarter", + "fiscal", + "year", + "investment", + "reclassified", + "equity", + "public", + "trading", + "balance", + "January", + "includes", + "10.5", + "billion", + "investments", + "subject", + "short", + "term", + "lock", + "restrictions", + "ability", + "sell", + "long", + "portion", + "December", + "2027", + "4.8", + "included", + "Publicly", + "held", + "price", + "volatility", + "Net", + "unrealized", + "gains", + "publicly", + "period", + "end", + "6.6", + "significant", + "years", + "2025", + "2024", + "realized", + "sold", + "reflecting", + "difference", + "sale", + "proceeds", + "carrying", + "beginning", + "purchase", + "date" + ], + "keywords": [ + "fair value", + "marketable securities", + "equity securities" + ], + "connect_to": [ + { + "target": "2a7db966-8d15-5579-ae85-e26396e44f94", + "relation": "embeds", + "ref": "[tables/table-86 Debt Securities.html]", + "position": { + "start": 304, + "end": 342 + } + } + ] + } + }, + { + "chunk_id": "828be8ff-3948-51a3-a7fe-9939224842b8", + "type": "table", + "content": "
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable Securities
(In millions)
Corporate debt securitiesLevel 2$ 18,504$ 51$ (29)$ 18,526$ 2,071$ 16,455
Debt securities issued by the U.S. TreasuryLevel 216,74942(22)16,7691,80114,968
Debt securities issued by U.S. government agenciesLevel 22,7757(5)2,7772,777
Foreign government bondsLevel 217717713740
Certificates of depositLevel 2979797
Money market fundsLevel 13,7603,7603,760
Publicly-held equity securitiesLevel 1381381
Total$ 42,062$ 100$ (56)$ 42,487$ 7,866$ 34,621
", + "path": "tables/table-87 Fair Value Table.html", + "metadata": { + "length": 1472, + "summary": "table-88\nThe table details investments by pricing category, showing cost, unrealized gains/losses, estimated fair value, and classification as cash equivalents or marketable securities. Total fair value is $42,487 million.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-87 Fair Value Table.html", + "keywords": [ + "fair value", + "debt securities", + "equity securities" + ], + "tokens": [] + } + }, + { + "chunk_id": "27971e02-74c3-5a5f-999e-a1ad964b9b3a", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
Less than 12 MonthsLess than 12 Months
Estimated Fair ValueGross Unrealized LossEstimated Fair ValueGross Unrealized Loss
(In millions)
Debt securities issued by the U.S. Treasury$ 10,666$ (3)$ 6,315$ (22)
Corporate debt securities1,332(3)5,291(29)
Debt securities issued by U.S. government agencies1,134816(5)
Total$ 13,132$ (6)$ 12,422$ (56)
", + "path": "tables/table-88 Debt Securities.html", + "metadata": { + "length": 791, + "summary": "table-89\nThe table compares estimated fair value and gross unrealized loss for debt securities (U.S. Treasury, corporate, government agencies) as of Jan 25, 2026 and Jan 26, 2025, with totals of $13,132M and $12,422M respectively.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-88 Debt Securities.html", + "keywords": [ + "Debt securities", + "Unrealized loss", + "Fair value" + ], + "tokens": [] + } + }, + { + "chunk_id": "2210d973-a95c-5c7f-816c-066746a7282c", + "type": "table", + "content": "
Jan 25, 2026
(In millions)
Less than one year$ 20,427
Due in 1 - 5 years19,093
Total$ 39,520
", + "path": "tables/table-89 Debt Maturity.html", + "metadata": { + "length": 238, + "summary": "table-90\nThe table shows debt maturities: $20,427 million due in less than one year, $19,093 million due in 1-5 years, totaling $39,520 million as of Jan 25, 2026.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-89 Debt Maturity.html", + "keywords": [ + "Debt", + "Maturity", + "Schedule" + ], + "tokens": [] + } + }, + { + "chunk_id": "2f245217-df73-50a2-ba58-473570024d02", + "type": "text", + "content": "(Continued)\nJan 26, 2025\n\n[tables/table-87 Fair Value Table.html]\n\nThe following tables provide the breakdown of unrealized losses, aggregated by investment category and length of time that individual debt securities have been in a continuous loss position:\n\n[tables/table-88 Debt Securities.html]\n\nGross unrealized losses related to debt securities in a continuous loss position of twelve months or greater as of January 25, 2026 and January 26, 2025 were not significant. Gross unrealized losses are related to fixed income securities, driven primarily by changes in interest rates.\nThe estimated fair value of debt securities included in cash equivalents and marketable securities are shown below by contractual maturity.\n\n[tables/table-89 Debt Maturity.html]", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_5", + "metadata": { + "length": 835, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "Jan", + "26", + "2025", + "The", + "tables", + "provide", + "breakdown", + "unrealized", + "losses", + "aggregated", + "investment", + "category", + "length", + "time", + "individual", + "debt", + "securities", + "continuous", + "loss", + "position", + "Gross", + "related", + "twelve", + "months", + "greater", + "January", + "25", + "2026", + "significant", + "fixed", + "income", + "driven", + "primarily", + "interest", + "rates", + "estimated", + "fair", + "included", + "cash", + "equivalents", + "marketable", + "shown", + "contractual", + "maturity" + ], + "keywords": [], + "connect_to": [ + { + "target": "828be8ff-3948-51a3-a7fe-9939224842b8", + "relation": "embeds", + "ref": "[tables/table-87 Fair Value Table.html]", + "position": { + "start": 26, + "end": 65 + } + }, + { + "target": "27971e02-74c3-5a5f-999e-a1ad964b9b3a", + "relation": "embeds", + "ref": "[tables/table-88 Debt Securities.html]", + "position": { + "start": 259, + "end": 297 + } + }, + { + "target": "2210d973-a95c-5c7f-816c-066746a7282c", + "relation": "embeds", + "ref": "[tables/table-89 Debt Maturity.html]", + "position": { + "start": 726, + "end": 762 + } + } + ] + } + }, + { + "chunk_id": "8418173f-081e-5c0f-8b1d-2608a19096cb", + "type": "text", + "content": "Our non-marketable equity securities are valued under the measurement alternative applying valuation methods based on observable transactions for similar investments of the same issuer and unobservable inputs such as volatility, expected", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 8 - Non-marketable Equity Securities", + "metadata": { + "length": 237, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "marketable", + "equity", + "securities", + "valued", + "measurement", + "alternative", + "applying", + "valuation", + "methods", + "based", + "observable", + "transactions", + "similar", + "investments", + "issuer", + "unobservable", + "inputs", + "volatility", + "expected" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "498716d7-58da-53d2-b691-4241bd598b7b", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025
(In millions)
Balance at beginning of period$ 3,387$ 1,321
Adjustments related to non-marketable equity securities:
Net additions17,4441,309
Unrealized gains2,369816
Reclassification (1)(848)
Impairments and unrealized losses(101)(59)
Balance at end of period$ 22,251$ 3,387
", + "path": "tables/table-90 Equity Securities.html", + "metadata": { + "length": 677, + "summary": "table-91\nThe table shows changes in non-marketable equity securities balances for years ending Jan 25, 2026 and Jan 26, 2025, including net additions, unrealized gains, reclassifications, and impairments.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-90 Equity Securities.html", + "keywords": [ + "non-marketable equity", + "unrealized gains", + "impairments" + ], + "tokens": [] + } + }, + { + "chunk_id": "343c1cdd-36fb-5721-8726-7e7268ec57cd", + "type": "text", + "content": "(Continued)\ntime to liquidity, risk free rate and security-specific rights and obligations. Gains and losses on these investments, realized and unrealized, are recognized in Other income, net on our Consolidated Statements of Income.\nAdjustments to the carrying value of our non-marketable equity securities during fiscal years 2026 and 2025 were as follows:\n\n[tables/table-90 Equity Securities.html]\n\n(1) Represents reclassifications from non-marketable equity securities to marketable securities following public market trading.\nNon-marketable equity securities had cumulative gross unrealized gains of \\$2.7 billion and \\$1.1 billion, and cumulative gross unrealized losses and impairments of \\$176 million and \\$105 million on securities held as of January 25, 2026 and January 26, 2025, respectively.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)", + "metadata": { + "length": 798, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "time", + "liquidity", + "risk", + "free", + "rate", + "security", + "specific", + "rights", + "obligations", + "Gains", + "losses", + "investments", + "realized", + "unrealized", + "recognized", + "Other", + "income", + "net", + "Consolidated", + "Statements", + "Income", + "Adjustments", + "carrying", + "marketable", + "equity", + "securities", + "fiscal", + "years", + "2026", + "2025", + "Represents", + "reclassifications", + "public", + "market", + "trading", + "Non", + "cumulative", + "gross", + "gains", + "2.7", + "billion", + "1.1", + "impairments", + "176", + "million", + "105", + "held", + "January", + "25", + "26" + ], + "keywords": [], + "connect_to": [ + { + "target": "498716d7-58da-53d2-b691-4241bd598b7b", + "relation": "embeds", + "ref": "[tables/table-90 Equity Securities.html]", + "position": { + "start": 360, + "end": 400 + } + } + ] + } + }, + { + "chunk_id": "9876cf7b-2a7e-5f2f-bfbf-06173d066c34", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
Inventories:(In millions)
Raw materials$ 3,807$ 3,408
Work in process8,8223,399
Finished goods8,7743,273
Total inventories (1)$ 21,403$ 10,080
", + "path": "tables/table-91 Inventory Data.html", + "metadata": { + "length": 396, + "summary": "table-92\nThe table shows inventories in millions for Jan 25, 2026 and Jan 26, 2025: raw materials $3,807 vs $3,408, work in process $8,822 vs $3,399, finished goods $8,774 vs $3,273, total $21,403 vs $10,080.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-91 Inventory Data.html", + "keywords": [ + "Inventories", + "Raw materials", + "Finished goods" + ], + "tokens": [] + } + }, + { + "chunk_id": "a869f5ef-fa55-59cc-b75b-71839d43efdd", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025EstimatedUseful Life
Property and Equipment:(In millions)(In years)
Land$ 777$ 511(A)
Buildings, leasehold improvements, and furniture2,8912,076(B)
Equipment, compute hardware, and software12,6197,5682-7
Construction in process683529(C)
Total property and equipment, gross16,97010,684
Accumulated depreciation and amortization(6,587)(4,401)
Total property and equipment, net$ 10,383$ 6,283
", + "path": "tables/table-92 Property & Equipment.html", + "metadata": { + "length": 828, + "summary": "table-93\nThe table shows property and equipment gross totals of $16,970 million (Jan 2026) and $10,684 million (Jan 2025), with accumulated depreciation of $6,587 million and $4,401 million, resulting in net values of $10,383 million and $6,283 million.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-92 Property & Equipment.html", + "keywords": [ + "Property", + "Equipment", + "Depreciation" + ], + "tokens": [] + } + }, + { + "chunk_id": "d43b1e36-d222-551d-a4d1-39ee59590020", + "type": "text", + "content": "We refer to customers who purchase products directly from NVIDIA as direct customers, such as AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. Three direct customers accounted for 25%, 18%, and 13% of our accounts receivable balance as of January 25, 2026. Two direct customers accounted for 17% and 16% of our accounts receivable balance as of January 26, 2025.\nCertain balance sheet components are as follows:\n\n[tables/table-91 Inventory Data.html]\n\n(1) In fiscal years 2026 and 2025, we recorded inventory provisions of \\$4.0 billion and \\$1.6 billion, respectively, in cost of revenue.\n\n[tables/table-92 Property & Equipment.html]\n\n(A) Land is a non-depreciable asset.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 9 - Balance Sheet Components", + "metadata": { + "length": 857, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "refer", + "customers", + "purchase", + "products", + "directly", + "NVIDIA", + "direct", + "AIBs", + "distributors", + "ODMs", + "OEMs", + "CSPs", + "AI", + "model", + "makers", + "system", + "integrators", + "Certain", + "internal", + "resources", + "party", + "complete", + "build", + "Three", + "accounted", + "25%", + "18%", + "13%", + "accounts", + "receivable", + "balance", + "January", + "25", + "2026", + "Two", + "17%", + "16%", + "26", + "2025", + "sheet", + "components", + "In", + "fiscal", + "years", + "recorded", + "inventory", + "provisions", + "4.0", + "billion", + "1.6", + "cost", + "revenue", + "Land", + "depreciable", + "asset" + ], + "keywords": [], + "connect_to": [ + { + "target": "9876cf7b-2a7e-5f2f-bfbf-06173d066c34", + "relation": "embeds", + "ref": "[tables/table-91 Inventory Data.html]", + "position": { + "start": 561, + "end": 598 + } + }, + { + "target": "a869f5ef-fa55-59cc-b75b-71839d43efdd", + "relation": "embeds", + "ref": "[tables/table-92 Property & Equipment.html]", + "position": { + "start": 739, + "end": 782 + } + } + ] + } + }, + { + "chunk_id": "2ac43a1f-f2bd-53bf-b2e6-93660a673491", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
Accrued and Other Current Liabilities:(In millions)
Customer program accruals$ 5,318$ 4,880
Accrued purchase consideration3,9219
Product warranty2,8071,290
Excess inventory purchase obligations (1)2,7392,095
Taxes payable2,669881
Deferred revenue (2)1,379837
Accrued payroll and related expenses1,146848
Other1,373897
Total accrued and other current liabilities$ 21,352$ 11,737
", + "path": "tables/table-93 Current Liabilities.html", + "metadata": { + "length": 806, + "summary": "table-94\nThe table lists accrued and other current liabilities for Jan 25, 2026 and Jan 26, 2025, totaling $21,352 million and $11,737 million respectively, including customer program accruals, warranty, and purchase obligations.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-93 Current Liabilities.html", + "keywords": [ + "accrued liabilities", + "warranty", + "deferred revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "95b1bede-01eb-54b5-9e13-b7b6f6252075", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
Other Long-Term Liabilities:(In millions)
Income tax payable (1)$ 3,958$ 2,188
Deferred income tax1,774886
Deferred revenue (2)1,193976
Other381195
Total other long-term liabilities$ 7,306$ 4,245
", + "path": "tables/table-94 Long-Term Liabilities.html", + "metadata": { + "length": 484, + "summary": "table-95\nThe table shows other long-term liabilities for Jan 25, 2026 and Jan 26, 2025, including income tax payable, deferred income tax, deferred revenue, and other items, totaling $7,306 million and $4,245 million respectively.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-94 Long-Term Liabilities.html", + "keywords": [ + "liabilities", + "income tax", + "deferred revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "bcd9722d-c4c5-599b-8bfb-47aadddafbac", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
(In millions)
Balance at beginning of period$ 1,813$ 1,337
Deferred revenue additions (1)11,1375,083
Revenue recognized (2)(10,378)(4,607)
Balance at end of period$ 2,572$ 1,813
", + "path": "tables/table-95 Deferred Revenue.html", + "metadata": { + "length": 434, + "summary": "table-96\nThe table shows deferred revenue balance changes from Jan 2025 to Jan 2026, with additions of $11,137M and revenue recognized of $10,378M, ending at $2,572M.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-95 Deferred Revenue.html", + "keywords": [ + "Deferred Revenue", + "Revenue Recognition", + "Balance Sheet" + ], + "tokens": [] + } + }, + { + "chunk_id": "99d3fcb7-5279-5c76-ba5f-d6f9e262c361", + "type": "text", + "content": "(Continued)\n(B) The estimated useful lives of our buildings are up to thirty years. Leasehold improvements and finance leases are amortized based on the lesser of either the asset's estimated useful life or the expected remaining lease term.\n(C) Construction in process represents assets that are not available for their intended use.\nDepreciation expense for fiscal years 2026, 2025, and 2024 was \\$2.4 billion, \\$1.3 billion, and \\$894 million, respectively.\nAccumulated amortization of leasehold improvements and finance leases was \\$519 million and \\$410 million as of January 25, 2026 and January 26, 2025, respectively.\nProperty, equipment and intangible assets acquired but not paid for during fiscal years 2026, 2025, and 2024 were \\$820 million, \\$525 million, and \\$170 million, respectively.\n\n[tables/table-93 Current Liabilities.html]\n\n(1) In fiscal years 2026 and 2025, we recorded an expense of approximately \\$3.2 billion and \\$2.0 billion, respectively, in cost of revenue.\n(2) Includes customer advances and unearned revenue related to hardware and software support, cloud services, and license and development arrangements. The balance as of January 25, 2026 and January 26, 2025 included \\$160 million and \\$81 million of customer advances, respectively.\n\n[tables/table-94 Long-Term Liabilities.html]\n\n(1) Primarily comprised of unrecognized tax benefits and related interest and penalties.\n(2) Includes unearned revenue related to hardware and software support and cloud services.\nDeferred Revenue\nThe following table shows the changes in short- and long-term deferred revenue during fiscal years 2026 and 2025:\n\n[tables/table-95 Deferred Revenue.html]\n\n(1) Includes \\$9.0 billion and \\$3.6 billion of customer advances for fiscal years 2026 and 2025, respectively.\n(2) Includes \\$8.9 billion and \\$3.7 billion related to customer advances for fiscal years 2026 and 2025, respectively.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)_2", + "metadata": { + "length": 1924, + "summary": "The passage details depreciation expenses for fiscal years 2026, 2025, and 2024 ($2.4B, $1.3B, $894M), accumulated amortization of leasehold improvements and finance leases ($519M and $410M as of Jan 25, 2026 and Jan 26, 2025), and property, equipment, and intangible assets acquired but not paid for ($820M, $525M, $170M). It also covers deferred revenue changes, including customer advances of $160M and $81M as of the respective dates, and $9.0B and $3.6B of customer advances for fiscal years 2026 and 2025.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "The", + "estimated", + "lives", + "buildings", + "thirty", + "years", + "Leasehold", + "improvements", + "finance", + "leases", + "amortized", + "based", + "lesser", + "asset", + "life", + "expected", + "remaining", + "lease", + "term", + "Construction", + "process", + "represents", + "assets", + "intended", + "Depreciation", + "expense", + "fiscal", + "2026", + "2025", + "2024", + "2.4", + "billion", + "1.3", + "894", + "million", + "Accumulated", + "amortization", + "leasehold", + "519", + "410", + "January", + "25", + "26", + "Property", + "equipment", + "intangible", + "acquired", + "paid", + "820", + "525", + "170", + "In", + "recorded", + "approximately", + "3.2", + "2.0", + "cost", + "revenue", + "Includes", + "customer", + "advances", + "unearned", + "related", + "hardware", + "software", + "support", + "cloud", + "services", + "license", + "development", + "arrangements", + "balance", + "included", + "160", + "81", + "Primarily", + "comprised", + "unrecognized", + "tax", + "benefits", + "interest", + "penalties", + "Deferred", + "Revenue", + "table", + "shows", + "short", + "long", + "deferred", + "9.0", + "3.6", + "8.9", + "3.7" + ], + "keywords": [ + "Depreciation", + "Deferred Revenue", + "Customer Advances" + ], + "connect_to": [ + { + "target": "2ac43a1f-f2bd-53bf-b2e6-93660a673491", + "relation": "embeds", + "ref": "[tables/table-93 Current Liabilities.html]", + "position": { + "start": 804, + "end": 846 + } + }, + { + "target": "95b1bede-01eb-54b5-9e13-b7b6f6252075", + "relation": "embeds", + "ref": "[tables/table-94 Long-Term Liabilities.html]", + "position": { + "start": 1275, + "end": 1319 + } + }, + { + "target": "bcd9722d-c4c5-599b-8bfb-47aadddafbac", + "relation": "embeds", + "ref": "[tables/table-95 Deferred Revenue.html]", + "position": { + "start": 1633, + "end": 1672 + } + } + ] + } + }, + { + "chunk_id": "9d1c0242-893d-50d3-81f5-692a56692597", + "type": "text", + "content": "(Continued)\nWe recognized revenue of \\$974 million and \\$729 million in fiscal years 2026 and 2025, respectively, that were included in the prior year end deferred revenue balance.\nAs of January 25, 2026, revenue related to remaining performance obligations from contracts greater than one year in length was \\$2.3 billion, which includes \\$1.9 billion from deferred revenue and \\$390 million which has not yet been billed nor recognized as revenue. Approximately 42% of revenue from contracts greater than one year in length will be recognized over the next twelve months.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_6", + "metadata": { + "length": 573, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "We", + "recognized", + "revenue", + "974", + "million", + "729", + "fiscal", + "years", + "2026", + "2025", + "included", + "prior", + "year", + "end", + "deferred", + "balance", + "As", + "January", + "25", + "related", + "remaining", + "performance", + "obligations", + "contracts", + "greater", + "length", + "2.3", + "billion", + "includes", + "1.9", + "390", + "billed", + "Approximately", + "42%", + "twelve", + "months" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6575dcb2-7840-5368-9891-3bb82517179a", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 10 - Derivative Financial Instruments", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e7d19787-3b15-5c8d-823f-3f4fc8bb3264", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
(In millions)
Designated as accounting hedges$ 1,765$ 1,424
Not designated as accounting hedges$ 2,332$ 1,297
", + "path": "tables/table-96 Hedge Values.html", + "metadata": { + "length": 296, + "summary": "table-97\nThe table shows designated accounting hedges at $1,765M (Jan 2026) and $1,424M (Jan 2025), and non-designated hedges at $2,332M and $1,297M respectively.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-96 Hedge Values.html", + "keywords": [ + "hedges", + "accounting", + "values" + ], + "tokens": [] + } + }, + { + "chunk_id": "b023364c-2e13-5140-839c-453ed9e7720e", + "type": "text", + "content": "We utilize foreign currency forward contracts to mitigate the impact of foreign currency exchange rate movements on our operating expenses. The foreign currency forward contracts for operating expenses are designated as accounting hedges. Gains or losses on the contracts are recorded in accumulated other comprehensive income or loss and reclassified to operating expense when the related operating expenses are recognized in earnings. In fiscal years 2026 and 2025, the impact of foreign currency forward contracts designated as accounting hedges on other comprehensive income or loss was not significant and all such instruments were determined to be highly effective.\nWe also entered into foreign currency forward contracts mitigating the impact of foreign currency movements on monetary assets and liabilities. For our foreign currency contracts for assets and liabilities, the change in fair value of these non-designated contracts was recorded in other income or expense and offsets the change in fair value of the hedged foreign currency denominated monetary assets and liabilities, which was also recorded in other income or expense.\nThe table below presents the notional value of our foreign currency contracts outstanding:\n\n[tables/table-96 Hedge Values.html]\n\nThe unrealized gains and losses or fair value of our foreign currency contracts were not significant as of January 25, 2026 and January 26, 2025.\nAs of January 25, 2026, all foreign currency contracts mature within 18 months. The expected realized gains and losses deferred into accumulated other comprehensive income or loss related to foreign currency forward contracts within the next twelve months were not significant.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 10 - Derivative Financial Instruments/Foreign Currency Derivatives", + "metadata": { + "length": 1708, + "summary": "The company uses foreign currency forward contracts to hedge against exchange rate impacts on operating expenses, designating them as accounting hedges with gains/losses recorded in OCI and reclassified to expenses. Non-designated contracts hedge monetary assets/liabilities, with fair value changes in other income/expense. Notional values are presented in a table. Unrealized gains/losses were not significant as of Jan 2025 and 2026. All contracts mature within 18 months, with no significant deferred gains/losses expected in the next twelve months.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "utilize", + "foreign", + "currency", + "forward", + "contracts", + "mitigate", + "impact", + "exchange", + "rate", + "movements", + "operating", + "expenses", + "The", + "designated", + "accounting", + "hedges", + "Gains", + "losses", + "recorded", + "accumulated", + "comprehensive", + "income", + "loss", + "reclassified", + "expense", + "related", + "recognized", + "earnings", + "In", + "fiscal", + "years", + "2026", + "2025", + "significant", + "instruments", + "determined", + "highly", + "effective", + "entered", + "mitigating", + "monetary", + "assets", + "liabilities", + "For", + "change", + "fair", + "offsets", + "hedged", + "denominated", + "table", + "presents", + "notional", + "outstanding", + "unrealized", + "gains", + "January", + "25", + "26", + "As", + "mature", + "18", + "months", + "expected", + "realized", + "deferred", + "twelve" + ], + "keywords": [ + "foreign currency", + "forward contracts", + "hedging" + ], + "connect_to": [ + { + "target": "e7d19787-3b15-5c8d-823f-3f4fc8bb3264", + "relation": "embeds", + "ref": "[tables/table-96 Hedge Values.html]", + "position": { + "start": 1235, + "end": 1270 + } + } + ] + } + }, + { + "chunk_id": "c213e507-8222-59eb-b307-90b1a8978031", + "type": "text", + "content": "In fiscal year 2026, we entered into agreements to guarantee partners' facility lease obligations in the event of their default in exchange for warrants. The maximum gross exposure under all agreements is \\$3.5 billion, which is reduced as the partners make payments to the lessors over terms ranging from 5 to 7 years. The partners have placed \\$712 million in escrow to mitigate our potential exposure. The guarantees, classified as credit derivatives with changes in fair value recognized in Other income and expense, were not material.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 10 - Derivative Financial Instruments/Facility Lease Guarantees", + "metadata": { + "length": 539, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "In", + "fiscal", + "year", + "2026", + "entered", + "agreements", + "guarantee", + "partners", + "facility", + "lease", + "obligations", + "event", + "default", + "exchange", + "warrants", + "The", + "maximum", + "gross", + "exposure", + "3.5", + "billion", + "reduced", + "make", + "payments", + "lessors", + "terms", + "ranging", + "years", + "712", + "million", + "escrow", + "mitigate", + "potential", + "guarantees", + "classified", + "credit", + "derivatives", + "fair", + "recognized", + "Other", + "income", + "expense", + "material" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2db61cef-309a-5bdf-ab38-c3f324d79ac3", + "type": "table", + "content": "
Expected Remaining Term (years)Effective Interest RateJan 25, 2026Jan 26, 2025
(In millions)
3.20% Notes Due 20260.63.31%1,0001,000
1.55% Notes Due 20282.41.64%1,2501,250
2.85% Notes Due 20304.22.93%1,5001,500
2.00% Notes Due 20315.42.09%1,2501,250
3.50% Notes Due 204014.23.54%1,0001,000
3.50% Notes Due 205024.23.54%2,0002,000
3.70% Notes Due 206034.23.73%500500
Unamortized debt discount and issuance costs(32)(37)
Net carrying amount8,4688,463
Less short-term portion(999)
Total long-term portion$ 7,469$ 8,463
", + "path": "tables/table-97 Debt Notes Table.html", + "metadata": { + "length": 1232, + "summary": "table-98\nTable lists notes due from 2026 to 2060 with interest rates from 1.55% to 3.70%, carrying amounts, and net carrying value of $8,468M as of Jan 25, 2026.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-97 Debt Notes Table.html", + "keywords": [ + "Notes", + "Interest Rates", + "Maturities" + ], + "tokens": [] + } + }, + { + "chunk_id": "1c4d6258-cf46-51f3-a1c5-088bcc512c53", + "type": "text", + "content": "(Continued)\nNote 11 - Debt\n\n[tables/table-97 Debt Notes Table.html]\n\nAs of January 25, 2026 and January 26, 2025, the estimated fair value of debt was \\$7.5 billion and \\$7.2 billion, respectively. The estimated fair values are based on Level 2 inputs.\nOur notes are unsecured senior obligations. Existing and future liabilities of our subsidiaries will be effectively senior to the notes. Our notes pay interest semi-annually. We may redeem each of our notes prior to maturity, subject to a make-whole premium. The maturity of the notes is calendar year.\nAs of January 25, 2026, we complied with the required covenants, which are non-financial in nature, under the outstanding notes.\nIn January 2026, we increased the size of our commercial paper program from \\$575 million to \\$25.0 billion. As of January 25, 2026, no commercial paper was outstanding.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_7", + "metadata": { + "length": 911, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "Note", + "11", + "Debt", + "As", + "January", + "25", + "2026", + "26", + "2025", + "estimated", + "fair", + "debt", + "7.5", + "billion", + "7.2", + "The", + "values", + "based", + "Level", + "inputs", + "Our", + "notes", + "unsecured", + "senior", + "obligations", + "Existing", + "future", + "liabilities", + "subsidiaries", + "effectively", + "pay", + "interest", + "semi", + "annually", + "We", + "redeem", + "prior", + "maturity", + "subject", + "make", + "premium", + "calendar", + "year", + "complied", + "required", + "covenants", + "financial", + "nature", + "outstanding", + "In", + "increased", + "size", + "commercial", + "paper", + "program", + "575", + "million", + "25.0" + ], + "keywords": [], + "connect_to": [ + { + "target": "2db61cef-309a-5bdf-ab38-c3f324d79ac3", + "relation": "embeds", + "ref": "[tables/table-97 Debt Notes Table.html]", + "position": { + "start": 28, + "end": 67 + } + } + ] + } + }, + { + "chunk_id": "57e228e5-e4ee-52a9-99b9-0d480259a056", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 12 - Commitments and Contingencies", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "bc3c771f-c000-56c6-990a-e1f81fa1c959", + "type": "text", + "content": "Manufacturing, supply, and capacity commitments reflect datacenter-scale production and longer future ordering horizons across current and future product architectures. We enter into agreements with our supply vendors that allow them to procure inventory based upon our defined criteria, and in certain instances, these agreements are cancellable, able to be rescheduled, or adjustable for our business needs prior to placing firm orders. Changes to these agreements may result in additional costs. As of January 25, 2026, these commitments were \\$95.2 billion, of which substantially all will be paid through fiscal year 2027.\nMulti-year cloud service agreement commitments as of January 25, 2026, were \\$27 billion, for which \\$7 billion, \\$6 billion, \\$5 billion, \\$5 billion, \\$2 billion, and \\$2 billion will be paid in fiscal years 2027, 2028, 2029, 2030, 2031, and 2032 and thereafter, respectively. Some cloud service capacity may be reduced, terminated or sold to others by the CSPs, in which case our commitments will be reduced. We expect cloud service agreements to be used to support our research and development efforts.\nInvestment commitments are \\$11.4 billion as of January 25, 2026, subject to certain contingencies, of which we expect substantially all will be made through fiscal year 2027.\nOther commitments were \\$3.4 billion as of January 25, 2026, of which the majority will be paid through fiscal year 2027.\nNVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 12 - Commitments and Contingencies/Commitments", + "metadata": { + "length": 1527, + "summary": "The passage details various commitments as of January 25, 2026. Manufacturing, supply, and capacity commitments total $95.2 billion, mostly payable through fiscal 2027, with agreements that may be cancellable or adjustable. Multi-year cloud service agreement commitments are $27 billion, with payments spread from 2027 to 2032 and beyond, and may be reduced if capacity is terminated or sold. Investment commitments stand at $11.4 billion, expected to be made through fiscal 2027. Other commitments are $3.4 billion, with most paid by fiscal 2027.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Manufacturing", + "supply", + "capacity", + "commitments", + "reflect", + "datacenter", + "scale", + "production", + "longer", + "future", + "ordering", + "horizons", + "current", + "product", + "architectures", + "We", + "enter", + "agreements", + "vendors", + "procure", + "inventory", + "based", + "defined", + "criteria", + "instances", + "cancellable", + "rescheduled", + "adjustable", + "business", + "prior", + "placing", + "firm", + "orders", + "Changes", + "result", + "additional", + "costs", + "As", + "January", + "25", + "2026", + "95.2", + "billion", + "substantially", + "paid", + "fiscal", + "year", + "2027", + "Multi", + "cloud", + "service", + "agreement", + "27", + "years", + "2028", + "2029", + "2030", + "2031", + "2032", + "Some", + "reduced", + "terminated", + "sold", + "CSPs", + "case", + "expect", + "support", + "research", + "development", + "efforts", + "Investment", + "11.4", + "subject", + "contingencies", + "made", + "Other", + "3.4", + "majority", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Continued" + ], + "keywords": [ + "Manufacturing", + "Cloud Service", + "Investment" + ], + "connect_to": [] + } + }, + { + "chunk_id": "a3975012-45c7-53af-9cbc-7951e74a28bb", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Balance at beginning of period$ 1,290$ 306$ 82
Additions2,4741,203278
Utilization(957)(219)(54)
Balance at end of period$ 2,807$ 1,290$ 306
", + "path": "tables/table-98 Restructuring Reserve.html", + "metadata": { + "length": 507, + "summary": "table-99\nThe table shows restructuring reserve balances for fiscal years ending Jan 2026, 2025, and 2024, with additions, utilization, and ending balances in millions.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-98 Restructuring Reserve.html", + "keywords": [ + "Restructuring", + "Reserve", + "Balance" + ], + "tokens": [] + } + }, + { + "chunk_id": "05922e34-1ec7-5dc2-90d0-22a948f0683e", + "type": "text", + "content": "The estimated amount of product warranty liabilities was \\$2.8 billion and \\$1.3 billion as of January 25, 2026 and January 26, 2025, respectively. The estimated product returns and product warranty activity consisted of the following:\n\n[tables/table-98 Restructuring Reserve.html]\n\nIn fiscal years 2026, 2025, and 2024 the additions in product warranty liabilities primarily related to our Compute & Networking segment.\nWe have provided indemnities for matters such as tax, product, and employee liabilities. We have included intellectual property indemnification provisions in our technology-related agreements with third parties. Maximum potential future payments cannot be estimated because many of these agreements do not have a maximum stated liability. We have not recorded any liability in our Consolidated Financial Statements for such indemnifications.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 12 - Commitments and Contingencies/Accrual for Product Warranty Liabilities", + "metadata": { + "length": 851, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "estimated", + "amount", + "product", + "warranty", + "liabilities", + "2.8", + "billion", + "1.3", + "January", + "25", + "2026", + "26", + "2025", + "returns", + "activity", + "consisted", + "In", + "fiscal", + "years", + "2024", + "additions", + "primarily", + "related", + "Compute", + "Networking", + "segment", + "We", + "provided", + "indemnities", + "matters", + "tax", + "employee", + "included", + "intellectual", + "property", + "indemnification", + "provisions", + "technology", + "agreements", + "parties", + "Maximum", + "potential", + "future", + "payments", + "maximum", + "stated", + "liability", + "recorded", + "Consolidated", + "Financial", + "Statements", + "indemnifications" + ], + "keywords": [], + "connect_to": [ + { + "target": "a3975012-45c7-53af-9cbc-7951e74a28bb", + "relation": "embeds", + "ref": "[tables/table-98 Restructuring Reserve.html]", + "position": { + "start": 237, + "end": 281 + } + } + ] + } + }, + { + "chunk_id": "93e07080-320c-59e7-9d53-52810fad99af", + "type": "text", + "content": "Securities Class Action and Derivative Lawsuits\nThe plaintiffs in the putative securities class action lawsuit, captioned 4:18-cv-07669-HSG, initially filed on December 21, 2018 in the United States District Court for the Northern District of California, and titled In Re NVIDIA Corporation Securities Litigation, filed an amended complaint on May 13, 2020. The amended complaint asserted that NVIDIA and certain NVIDIA executives violated Section 10(b) of the Securities Exchange Act of 1934, as amended, or the Exchange Act, and SEC Rule 10b-5, by making materially false or misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand between May 10, 2017 and November 14, 2018. Plaintiffs also alleged that the NVIDIA executives who they named as defendants violated Section 20(a) of the Exchange Act. Plaintiffs sought class certification, an award of unspecified compensatory damages, an award of reasonable costs and expenses, including attorneys' fees and expert fees, and further relief as the Court may deem just and proper. On March 2, 2021, the district court granted NVIDIA's motion to dismiss the complaint without leave to amend, entered judgment in favor of NVIDIA and closed the case. On March 30, 2021, plaintiffs filed an appeal from judgment in the United States Court of Appeals for the Ninth Circuit, case number 21-15604. On August 25, 2023, a majority of a three-judge Ninth Circuit panel affirmed in part and reversed in part the district court's dismissal of the case, with a third judge dissenting on the basis that the district court did not err in dismissing the case. On November 15, 2023, the Ninth Circuit denied NVIDIA's petition for rehearing en banc of the Ninth Circuit panel's majority decision to reverse in part the dismissal of the case, which NVIDIA had filed on October 10, 2023. On December 5, 2023, the Ninth Circuit granted NVIDIA's motion to stay the mandate pending NVIDIA's petition for a writ of certiorari in the Supreme Court of the United States and the Supreme Court's final disposition of the matter. NVIDIA filed a petition for a writ of certiorari on March 4, 2024. On June 17, 2024, the Supreme Court of the United States granted NVIDIA's petition for a writ of certiorari. After briefing and argument, the Supreme Court dismissed NVIDIA's writ of certiorari as improvidently granted on December 11, 2024, and issued judgment on January 13, 2025. On February 20, 2025, the Ninth Circuit's judgment, entered August 25, 2023 and corrected August 28, 2023, took effect, and the case was remanded to the district court for further proceedings.\nThe putative derivative lawsuit pending in the United States District Court for the Northern District of California, captioned 4:19-cv-00341-HSG, initially filed January 18, 2019 and titled In re NVIDIA Corporation Consolidated Derivative Litigation, was stayed pending resolution of the plaintiffs' appeal in the In Re NVIDIA Corporation Securities Litigation action. On February 22, 2022, the court administratively closed the case, but stated that it would reopen the case once the appeal in the In Re NVIDIA Corporation Securities Litigation action is resolved. The case has not yet been reopened by the court. The lawsuit asserts claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty, unjust enrichment, waste of corporate assets, and violations of Sections 14(a), 10(b), and 20(a) of the Exchange Act based on the dissemination of allegedly false and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs are seeking unspecified damages and other relief, including reforms and improvements to NVIDIA's corporate governance and internal procedures.\nThe putative derivative actions initially filed September 24, 2019 and pending in the United States District Court for the District of Delaware, Lipchitz v. Huang, et al. (Case No. 1:19-cv-01795-MN) and Nelson v. Huang, et al. (Case No. 1:19-cv-01798-MN), were stayed pending resolution of the plaintiffs' appeal in the In Re NVIDIA Corporation Securities", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 12 - Commitments and Contingencies/Litigation", + "metadata": { + "length": 4183, + "summary": "The text details two related lawsuits against NVIDIA. A securities class action (In Re NVIDIA Corporation Securities Litigation) was filed in 2018, alleging false statements about GPU demand and cryptocurrency mining. After dismissal, appeals, and a Supreme Court certiorari that was dismissed as improvidently granted, the case was remanded to district court in 2025. A related derivative lawsuit in California remains stayed, while two Delaware derivative actions are also stayed pending resolution of the securities appeal.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Securities", + "Class", + "Action", + "Derivative", + "Lawsuits", + "The", + "plaintiffs", + "putative", + "securities", + "class", + "action", + "lawsuit", + "captioned", + "18", + "cv", + "07669", + "HSG", + "initially", + "filed", + "December", + "21", + "2018", + "United", + "States", + "District", + "Court", + "Northern", + "California", + "titled", + "In", + "Re", + "NVIDIA", + "Corporation", + "Litigation", + "amended", + "complaint", + "May", + "13", + "2020", + "asserted", + "executives", + "violated", + "Section", + "10", + "Exchange", + "Act", + "1934", + "SEC", + "Rule", + "10b", + "making", + "materially", + "false", + "misleading", + "statements", + "related", + "channel", + "inventory", + "impact", + "cryptocurrency", + "mining", + "GPU", + "demand", + "2017", + "November", + "14", + "Plaintiffs", + "alleged", + "named", + "defendants", + "20", + "sought", + "certification", + "award", + "unspecified", + "compensatory", + "damages", + "reasonable", + "costs", + "expenses", + "including", + "attorneys", + "fees", + "expert", + "relief", + "deem", + "proper", + "On", + "March", + "2021", + "district", + "court", + "granted", + "motion", + "dismiss", + "leave", + "amend", + "entered", + "judgment", + "favor", + "closed", + "case", + "30", + "appeal", + "Appeals", + "Ninth", + "Circuit", + "number", + "15604", + "August", + "25", + "2023", + "majority", + "judge", + "panel", + "affirmed", + "part", + "reversed", + "dismissal", + "dissenting", + "basis", + "err", + "dismissing", + "15", + "denied", + "petition", + "rehearing", + "en", + "banc", + "decision", + "reverse", + "October", + "stay", + "mandate", + "pending", + "writ", + "certiorari", + "Supreme", + "final", + "disposition", + "matter", + "2024", + "June", + "17", + "After", + "briefing", + "argument", + "dismissed", + "improvidently", + "11", + "issued", + "January", + "2025", + "February", + "corrected", + "28", + "effect", + "remanded", + "proceedings", + "derivative", + "19", + "00341", + "2019", + "Consolidated", + "stayed", + "resolution", + "22", + "2022", + "administratively", + "stated", + "reopen", + "resolved", + "reopened", + "asserts", + "claims", + "purportedly", + "behalf", + "officers", + "directors", + "Company", + "breach", + "fiduciary", + "duty", + "unjust", + "enrichment", + "waste", + "corporate", + "assets", + "violations", + "Sections", + "based", + "dissemination", + "allegedly", + "seeking", + "reforms", + "improvements", + "governance", + "internal", + "procedures", + "actions", + "September", + "24", + "Delaware", + "Lipchitz", + "Huang", + "al", + "Case", + "No", + "01795", + "MN", + "Nelson", + "01798" + ], + "keywords": [ + "Securities Class Action", + "Derivative Lawsuit", + "NVIDIA" + ], + "connect_to": [] + } + }, + { + "chunk_id": "7923544a-3185-5f7d-a082-b3e60362b9fd", + "type": "text", + "content": "(Continued)\nLitigation action. On March 7, 2025, after the Supreme Court issued its judgment dismissing the Company's petition for writ of certiorari as improvidently granted in the In Re NVIDIA Securities Litigation action, the district court adopted the parties' stipulation to extend the stay until the final and complete resolution of the In Re NVIDIA Corporation Securities Litigation action. The lawsuits assert claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty, unjust enrichment, insider trading, misappropriation of information, corporate waste and violations of Sections 14(a), 10(b), and 20(a) of the Exchange Act based on the dissemination of allegedly false, and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs seek unspecified damages and other relief, including disgorgement of profits from the sale of NVIDIA stock and unspecified corporate governance measures.\nAnother putative derivative action was filed on October 30, 2023 in the Court of Chancery of the State of Delaware, captioned Horanic v. Huang, et al. (Case No. 2023-1096-KSJM). This lawsuit asserts claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty and insider trading based on the dissemination of allegedly false and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs seek unspecified damages and other relief, including disgorgement of profits from the sale of NVIDIA stock and reform of unspecified corporate governance measures. On August 11, 2025, the court granted the parties' stipulation to voluntarily dismiss with prejudice plaintiff City of Westland Police and Fire Retirement System. This derivative matter is stayed pending the final resolution of In Re NVIDIA Corporation Securities Litigation action.\nAccounting for Loss Contingencies\nAs of January 25, 2026, there are no accrued contingent liabilities associated with the legal proceedings described above based on our belief that liabilities, while reasonably possible, are not probable. Further, any possible loss or range of loss in these matters cannot be reasonably estimated at this time. We are engaged in legal actions not described above arising in the ordinary course of business, as well as regulatory and government inquiries and investigations, and, while there can be no assurance of favorable outcomes, we believe that the ultimate outcome of these matters will not have a material adverse effect on our operating results, liquidity or financial position. These matters are subject to inherent uncertainties and if the ultimate outcome is unfavorable, there exists the possibility of a material adverse impact on our operating results, liquidity or financial position in the period the outcome becomes estimable and probable.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 12 - Commitments and Contingencies/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements", + "metadata": { + "length": 2968, + "summary": "The passage details ongoing legal proceedings against NVIDIA, including a securities class action and derivative lawsuits alleging false statements about channel inventory and cryptocurrency mining's impact on GPU demand. These actions seek damages, disgorgement, and governance reforms. As of January 25, 2026, no contingent liabilities have been accrued as losses are not deemed probable or estimable. The company believes other routine legal matters will not materially affect its financial position.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "Litigation", + "action", + "On", + "March", + "2025", + "Supreme", + "Court", + "issued", + "judgment", + "dismissing", + "Company", + "petition", + "writ", + "certiorari", + "improvidently", + "granted", + "In", + "Re", + "NVIDIA", + "Securities", + "district", + "court", + "adopted", + "parties", + "stipulation", + "extend", + "stay", + "final", + "complete", + "resolution", + "Corporation", + "The", + "lawsuits", + "assert", + "claims", + "purportedly", + "behalf", + "officers", + "directors", + "breach", + "fiduciary", + "duty", + "unjust", + "enrichment", + "insider", + "trading", + "misappropriation", + "information", + "corporate", + "waste", + "violations", + "Sections", + "14", + "10", + "20", + "Exchange", + "Act", + "based", + "dissemination", + "allegedly", + "false", + "misleading", + "statements", + "related", + "channel", + "inventory", + "impact", + "cryptocurrency", + "mining", + "GPU", + "demand", + "plaintiffs", + "seek", + "unspecified", + "damages", + "relief", + "including", + "disgorgement", + "profits", + "sale", + "stock", + "governance", + "measures", + "Another", + "putative", + "derivative", + "filed", + "October", + "30", + "2023", + "Chancery", + "State", + "Delaware", + "captioned", + "Horanic", + "Huang", + "al", + "Case", + "No", + "1096", + "KSJM", + "This", + "lawsuit", + "asserts", + "reform", + "August", + "11", + "voluntarily", + "dismiss", + "prejudice", + "plaintiff", + "City", + "Westland", + "Police", + "Fire", + "Retirement", + "System", + "matter", + "stayed", + "pending", + "Accounting", + "Loss", + "Contingencies", + "As", + "January", + "25", + "2026", + "accrued", + "contingent", + "liabilities", + "legal", + "proceedings", + "belief", + "probable", + "Further", + "loss", + "range", + "matters", + "estimated", + "time", + "We", + "engaged", + "actions", + "arising", + "ordinary", + "business", + "regulatory", + "government", + "inquiries", + "investigations", + "assurance", + "favorable", + "outcomes", + "ultimate", + "outcome", + "material", + "adverse", + "effect", + "operating", + "results", + "liquidity", + "financial", + "position", + "These", + "subject", + "inherent", + "uncertainties", + "unfavorable", + "exists", + "possibility", + "period", + "estimable" + ], + "keywords": [ + "Securities Litigation", + "Derivative Actions", + "Loss Contingencies" + ], + "connect_to": [] + } + }, + { + "chunk_id": "e48ae7bd-545e-5704-997e-ec976e446aa5", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Current income taxes:
Federal$ 19,039$ 14,032$ 5,710
State1,218892335
Foreign2,550699502
Total current22,80715,6236,547
Deferred income taxes:
Federal(1,364)(4,515)(2,499)
State(885)(242)(206)
Foreign825280216
Total deferred(1,424)(4,477)(2,489)
Income tax expense$ 21,383$ 11,146$ 4,058
", + "path": "tables/table-99 Income Tax Expense.html", + "metadata": { + "length": 933, + "summary": "table-100\nThe table shows income tax expenses for fiscal years ending Jan 2026, 2025, and 2024, with current and deferred taxes broken down by federal, state, and foreign categories.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-99 Income Tax Expense.html", + "keywords": [ + "Income Tax", + "Federal", + "Deferred" + ], + "tokens": [] + } + }, + { + "chunk_id": "b7ec00a8-6c58-58c3-98d8-2f5a8beecd06", + "type": "text", + "content": "The FASB issued a new accounting standard which includes new and updated income tax disclosures, including disaggregation of information in the rate reconciliation and income taxes paid, which we adopted on a prospective basis for the year ending January 25, 2026.\nThe Income tax expense applicable to income before income taxes consists of the following:\n\n[tables/table-99 Income Tax Expense.html]", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 13 - Income Taxes", + "metadata": { + "length": 391, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "The", + "FASB", + "issued", + "accounting", + "standard", + "includes", + "updated", + "income", + "tax", + "disclosures", + "including", + "disaggregation", + "information", + "rate", + "reconciliation", + "taxes", + "paid", + "adopted", + "prospective", + "basis", + "year", + "ending", + "January", + "25", + "2026", + "Income", + "expense", + "applicable", + "consists" + ], + "keywords": [], + "connect_to": [ + { + "target": "e48ae7bd-545e-5704-997e-ec976e446aa5", + "relation": "embeds", + "ref": "[tables/table-99 Income Tax Expense.html]", + "position": { + "start": 357, + "end": 398 + } + } + ] + } + }, + { + "chunk_id": "1fd49f1a-6e33-51ab-83af-cb755a483899", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
U.S.$ 123,181$ 77,456$ 29,495
Foreign18,2696,5704,323
Income before income tax$ 141,450$ 84,026$ 33,818
", + "path": "tables/table-100 Income Before Tax.html", + "metadata": { + "length": 427, + "summary": "table-101\nThe table shows income before income tax for U.S. and Foreign segments for years ending Jan 2026, 2025, and 2024, with totals of $141,450M, $84,026M, and $33,818M respectively.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-100 Income Before Tax.html", + "keywords": [ + "Income", + "Tax", + "Revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "e3205cbb-0961-51d5-92d6-24d339e2b4b6", + "type": "table", + "content": "
Year Ended
Jan 25, 2026
(In millions, except percentages)
US Federal Statutory Tax Rate$ 29,70421.0 %
State and Local Income Taxes, Net of Federal Income Tax Effect (1)2580.2 %
Foreign tax effects
Israel
Reduced statutory tax rate on qualifying income(3,064)(2.2)%
Other1,6061.2 %
Other foreign jurisdictions7410.5 %
Effect of cross-border tax laws
Foreign-derived deduction eligible income(4,208)(3.0)%
Other(142)(0.1)%
Tax credits(1,933)(1.4)%
Nontaxable or nondeductible items
Stock-based compensation(1,475)(1.0)%
Other29— %
Other (2)(133)(0.1)%
Income tax expense$ 21,38315.1 %
", + "path": "tables/table-101 Tax Rate Breakdown.html", + "metadata": { + "length": 1266, + "summary": "table-102\nThe table shows a reconciliation from the US federal statutory tax rate of 21.0% to an effective income tax expense of 15.1%, with adjustments for state taxes, foreign effects, credits, and other items.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-101 Tax Rate Breakdown.html", + "keywords": [ + "tax rate", + "income tax", + "statutory" + ], + "tokens": [] + } + }, + { + "chunk_id": "5fabe41e-d069-52e6-9fef-cab76ac3020f", + "type": "text", + "content": "(Continued)\nIncome before income tax consists of the following:\n\n[tables/table-100 Income Before Tax.html]\n\nThe income tax expense (benefit) differs from the amount computed by applying the U.S. federal statutory rate of 21.0% to income before income taxes for the fiscal year ended January 25, 2026 as follows:\n\n[tables/table-101 Tax Rate Breakdown.html]\n\n(1) State taxes in California, Tennessee, Arizona, and Illinois made up the majority of the tax effect in fiscal year 2026.\n(2) Includes the tax effects of enactment of new tax laws, change in valuation allowance, and change in unrecognized tax benefits.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_8", + "metadata": { + "length": 596, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "Income", + "income", + "tax", + "consists", + "The", + "expense", + "benefit", + "differs", + "amount", + "computed", + "applying", + "federal", + "statutory", + "rate", + "21.0%", + "taxes", + "fiscal", + "year", + "ended", + "January", + "25", + "2026", + "State", + "California", + "Tennessee", + "Arizona", + "Illinois", + "made", + "majority", + "effect", + "Includes", + "effects", + "enactment", + "laws", + "change", + "valuation", + "allowance", + "unrecognized", + "benefits" + ], + "keywords": [], + "connect_to": [ + { + "target": "1fd49f1a-6e33-51ab-83af-cb755a483899", + "relation": "embeds", + "ref": "[tables/table-100 Income Before Tax.html]", + "position": { + "start": 65, + "end": 106 + } + }, + { + "target": "e3205cbb-0961-51d5-92d6-24d339e2b4b6", + "relation": "embeds", + "ref": "[tables/table-101 Tax Rate Breakdown.html]", + "position": { + "start": 313, + "end": 355 + } + } + ] + } + }, + { + "chunk_id": "db7da263-a04b-5b2f-9797-89dd29207694", + "type": "table", + "content": "
Year Ended
Jan 26, 2025Jan 28, 2024
(In millions, except percentages)
Tax expense computed at federal statutory rate$ 17,64521.0 %$ 7,10221.0 %
Expense (benefit) resulting from:
State income taxes, net of federal tax effect5540.7 %1200.4 %
Foreign-derived deduction eligible income(2,976)(3.5)%(1,408)(4.2)%
Stock-based compensation(2,097)(2.5)%(741)(2.2)%
U.S. federal research and development tax credit(990)(1.2)%(431)(1.3)%
Foreign tax rate differential(984)(1.2)%(467)(1.4)%
Other(6)— %(117)(0.3)%
Income tax expense$ 11,14613.3 %$ 4,05812.0 %
", + "path": "tables/table-102 Tax Rate Analysis.html", + "metadata": { + "length": 1170, + "summary": "table-103\nThe table shows income tax expense for years ending Jan 26, 2025 ($11,146M, 13.3%) and Jan 28, 2024 ($4,058M, 12.0%), with adjustments for state taxes, foreign deductions, stock compensation, R&D credits, and foreign rate differentials.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-102 Tax Rate Analysis.html", + "keywords": [ + "tax expense", + "effective rate", + "income tax" + ], + "tokens": [] + } + }, + { + "chunk_id": "1ff4ab87-e18f-58ef-beff-e90371794a12", + "type": "table", + "content": "
Year Ended
Jan 25, 2026
(In millions)
Federal$ 16,755
State
California1,049
Other1,041
Foreign
Israel1,287
Other156
Total income taxes paid, net of refunds$ 20,288
", + "path": "tables/table-103 Income Tax Paid.html", + "metadata": { + "length": 459, + "summary": "table-104\nThe table shows income taxes paid, net of refunds, totaling $20,288 million for the year ended Jan 25, 2026, with federal, state, and foreign breakdowns.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-103 Income Tax Paid.html", + "keywords": [ + "taxes", + "income", + "paid" + ], + "tokens": [] + } + }, + { + "chunk_id": "df95931e-4ddd-55c8-9e3c-458cf6b82d59", + "type": "text", + "content": "(Continued)\nThe income tax expense (benefit) differs from the amount computed by applying the U.S. federal statutory rate of 21% to income before income taxes for fiscal years ended January 26, 2025 and January 28, 2024 as follows:\n\n[tables/table-102 Tax Rate Analysis.html]\n\nIn July 2025, the OBBBA was enacted into law and contains several changes to key U.S. federal income tax laws. We have recognized the tax effects of currently effective OBBBA provisions in our results for fiscal year 2026.\nThe amount of cash paid for income taxes (net of refunds) for the fiscal year ended January 25, 2026 is as follows:\n\n[tables/table-103 Income Tax Paid.html]", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_9", + "metadata": { + "length": 643, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "The", + "income", + "tax", + "expense", + "benefit", + "differs", + "amount", + "computed", + "applying", + "federal", + "statutory", + "rate", + "21%", + "taxes", + "fiscal", + "years", + "ended", + "January", + "26", + "2025", + "28", + "2024", + "In", + "July", + "OBBBA", + "enacted", + "law", + "key", + "laws", + "We", + "recognized", + "effects", + "effective", + "provisions", + "results", + "year", + "2026", + "cash", + "paid", + "net", + "refunds", + "25" + ], + "keywords": [], + "connect_to": [ + { + "target": "db7da263-a04b-5b2f-9797-89dd29207694", + "relation": "embeds", + "ref": "[tables/table-102 Tax Rate Analysis.html]", + "position": { + "start": 233, + "end": 274 + } + }, + { + "target": "1ff4ab87-e18f-58ef-beff-e90371794a12", + "relation": "embeds", + "ref": "[tables/table-103 Income Tax Paid.html]", + "position": { + "start": 616, + "end": 655 + } + } + ] + } + }, + { + "chunk_id": "f52eb654-f58f-55a5-9273-94464be2a91f", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
(In millions)
Deferred tax assets:
Capitalized research and development expenditure$ 5,436$ 6,256
Net controlled foreign corporation tested income deferred tax assets5,3892,820
Accruals and reserves, not currently deductible for tax purposes3,6442,058
Research and other tax credit carryforwards718759
Operating lease liabilities554299
Net operating loss and capital loss carryforwards443456
Other deferred tax assets679566
Gross deferred tax assets16,86313,214
Less valuation allowance(768)(1,610)
Total deferred tax assets16,09511,604
Deferred tax liabilities:
Equity investments(2,227)(264)
Unremitted earnings of foreign subsidiaries(1,813)(891)
Operating lease assets(533)(286)
Acquired intangibles(38)(70)
Gross deferred tax liabilities(4,611)(1,511)
Net deferred tax asset (1)$ 11,484$ 10,093
", + "path": "tables/table-104 Deferred Tax Assets.html", + "metadata": { + "length": 1546, + "summary": "table-105\nThe table shows deferred tax assets and liabilities for Jan 25, 2026 and Jan 26, 2025. Gross deferred tax assets were $16,863M and $13,214M, with net assets of $11,484M and $10,093M after valuation allowance and liabilities.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-104 Deferred Tax Assets.html", + "keywords": [ + "deferred tax", + "valuation allowance", + "tax assets" + ], + "tokens": [] + } + }, + { + "chunk_id": "a2b52d05-e0c5-5701-ac18-42a1cde30171", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Balance at beginning of period$ 2,861$ 1,670$ 1,238
Increases in tax positions for current year1,9591,268616
Increases in tax positions for prior years574887
Lapse in statute of limitations(224)(27)(19)
Decreases in tax positions for prior years(157)(88)(148)
Settlements(76)(10)(104)
Balance at end of period$ 4,420$ 2,861$ 1,670
", + "path": "tables/table-105 Tax Positions.html", + "metadata": { + "length": 790, + "summary": "table-106\nThe table shows the balance of tax positions at beginning and end of periods for Jan 2024-2026, with increases, decreases, lapse, and settlements in millions.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-105 Tax Positions.html", + "keywords": [ + "tax positions", + "balance", + "changes" + ], + "tokens": [] + } + }, + { + "chunk_id": "483a9cf8-9ea6-589b-a059-edc14a801315", + "type": "text", + "content": "(Continued)\nThe tax effect of temporary differences that gives rise to significant portions of the deferred tax assets and liabilities are presented below:\n\n[tables/table-104 Deferred Tax Assets.html]\n\n(1) Net deferred tax asset includes long-term deferred tax assets of \\$13.3 billion and \\$11.0 billion and long-term deferred tax liabilities of \\$1.8 billion and \\$886 million for fiscal years 2026 and 2025, respectively. Long-term deferred tax liabilities are included in other long-term liabilities on our Consolidated Balance Sheets.\nAs of January 25, 2026, we intend to indefinitely reinvest approximately \\$1.4 billion of cumulative undistributed earnings held by certain subsidiaries. We have not provided the amount of unrecognized deferred tax liabilities for temporary differences related to these investments as the determination of such amount is not practicable.\nAs of January 25, 2026 and January 26, 2025, we had a valuation allowance of \\$768 million and \\$1.6 billion, respectively, related to capital loss carryforwards, and certain other deferred tax assets that management determined are not likely to be realized due, in part, to jurisdictional projections of future taxable income, including capital gains. To the extent realization of the deferred tax assets becomes more-likely-than-not, we would recognize such deferred tax assets as income tax benefits during the period.\nAs of January 25, 2026, based on recent jurisdictional taxable income and expected future earnings, we concluded certain state deferred tax assets are more likely than not realizable and released \\$711 million of valuation allowance.\nAs of January 25, 2026, we had U.S. federal, state and foreign net operating loss carryforwards of \\$747 million, \\$427 million and \\$503 million, respectively. The federal and state carryforwards will begin to expire in fiscal year 2027. The foreign net operating loss carryforwards may be carried forward indefinitely. As of January 25, 2026, we had federal research tax credit carryforwards of \\$56 million, before the impact of uncertain tax positions, that will begin to expire in fiscal year 2027. We have state research tax credit carryforwards of \\$1.4 billion, before the impact of uncertain tax positions, of which \\$1.3 billion is attributable to the State of California and may be carried over indefinitely and \\$132 million is attributable to various other states and will begin to expire in fiscal year 2028. As of January 25, 2026, we had federal capital loss carryforwards of \\$902 million that will begin to expire in fiscal year 2028.\nOur tax attributes remain subject to audit and may be adjusted for changes or modification in tax laws, other authoritative interpretations thereof, or other facts and circumstances. Utilization of tax attributes may also be subject to limitations due to ownership changes and other limitations provided by the Internal Revenue Code and similar state and foreign tax provisions. If any such limitations apply, the tax attributes may expire or be denied before utilization.\nNVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)\nA reconciliation of gross unrecognized tax benefits is as follows:\n\n[tables/table-105 Tax Positions.html]\n\nIncluded in the balance of unrecognized tax benefits as of January 25, 2026 are \\$3.7 billion of tax benefits that would affect our effective tax rate if recognized.\nWe classify an unrecognized tax benefit as a current liability, or amount refundable, to the extent that we anticipate payment or receipt of cash for income taxes within one year. The amount is classified as a long-term liability, or long-term amount refundable, if we anticipate payment or receipt of cash for income taxes during a period beyond a year.\nWe include interest and penalties related to unrecognized tax benefits as a component of income tax expense. We recognized net interest and penalties related to unrecognized tax benefits in the income tax expense line of our consolidated statements of income of \\$103 million, \\$92 million, and \\$42 million during fiscal years 2026, 2025, and 2024, respectively. As of January 25, 2026 and January 26, 2025, we have accrued \\$374 million and \\$251 million, respectively, for the payment of interest and penalties related to unrecognized tax benefits, which is not included as a component of our gross unrecognized tax benefits.\nWe are subject to examination by taxing authorities both in the United States and other countries. As of January 25, 2026, the significant tax jurisdictions that may be subject to examination include the United States for fiscal years after 2022, as well as Canada, China, Germany, Hong Kong, India, Israel, Italy, and Taiwan for fiscal years 2014 through 2025. As of January 25, 2026, the significant tax jurisdictions for which we are currently under examination include the United States, Germany, Hong Kong, India, Israel, and Taiwan for fiscal years 2014 through 2025.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_10", + "metadata": { + "length": 5016, + "summary": "The passage details NVIDIA's deferred tax assets and liabilities as of fiscal years 2026 and 2025. It reports net deferred tax assets of $13.3 billion (2026) and $11.0 billion (2025), with long-term deferred tax liabilities of $1.8 billion and $886 million, respectively. A valuation allowance of $768 million (2026) and $1.6 billion (2025) is held for capital loss carryforwards and other deferred tax assets. In 2026, $711 million of valuation allowance was released due to improved realizability. The company has net operating loss carryforwards: U.S. federal ($747 million), state ($427 million), and foreign ($503 million), with federal and state losses expiring from 2027. Research tax credit carryforwards include $56 million federal (expiring 2027) and $1.4 billion state (mostly California, indefinite). Federal capital loss carryforwards total $902 million (expiring 2028). Unrecognized tax benefits are $3.7 billion, with accrued interest and penalties of $374 million. Tax examinations are ongoing in multiple jurisdictions.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "The", + "tax", + "effect", + "temporary", + "differences", + "rise", + "significant", + "portions", + "deferred", + "assets", + "liabilities", + "presented", + "Net", + "asset", + "includes", + "long", + "term", + "13.3", + "billion", + "11.0", + "1.8", + "886", + "million", + "fiscal", + "years", + "2026", + "2025", + "Long", + "included", + "Consolidated", + "Balance", + "Sheets", + "As", + "January", + "25", + "intend", + "indefinitely", + "reinvest", + "approximately", + "1.4", + "cumulative", + "undistributed", + "earnings", + "held", + "subsidiaries", + "We", + "provided", + "amount", + "unrecognized", + "related", + "investments", + "determination", + "practicable", + "26", + "valuation", + "allowance", + "768", + "1.6", + "capital", + "loss", + "carryforwards", + "management", + "determined", + "realized", + "due", + "part", + "jurisdictional", + "projections", + "future", + "taxable", + "income", + "including", + "gains", + "To", + "extent", + "realization", + "recognize", + "benefits", + "period", + "based", + "recent", + "expected", + "concluded", + "state", + "realizable", + "released", + "711", + "federal", + "foreign", + "net", + "operating", + "747", + "427", + "503", + "begin", + "expire", + "year", + "2027", + "carried", + "forward", + "research", + "credit", + "56", + "impact", + "uncertain", + "positions", + "1.3", + "attributable", + "State", + "California", + "132", + "states", + "2028", + "902", + "Our", + "attributes", + "remain", + "subject", + "audit", + "adjusted", + "modification", + "laws", + "authoritative", + "interpretations", + "thereof", + "facts", + "circumstances", + "Utilization", + "limitations", + "ownership", + "Internal", + "Revenue", + "Code", + "similar", + "provisions", + "If", + "apply", + "denied", + "utilization", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Financial", + "Statements", + "reconciliation", + "gross", + "Included", + "balance", + "3.7", + "affect", + "effective", + "rate", + "recognized", + "classify", + "benefit", + "current", + "liability", + "refundable", + "anticipate", + "payment", + "receipt", + "cash", + "taxes", + "classified", + "include", + "interest", + "penalties", + "component", + "expense", + "line", + "consolidated", + "statements", + "103", + "92", + "42", + "2024", + "accrued", + "374", + "251", + "examination", + "taxing", + "authorities", + "United", + "States", + "countries", + "jurisdictions", + "2022", + "Canada", + "China", + "Germany", + "Hong", + "Kong", + "India", + "Israel", + "Italy", + "Taiwan", + "2014" + ], + "keywords": [ + "deferred tax", + "valuation allowance", + "tax attributes" + ], + "connect_to": [ + { + "target": "f52eb654-f58f-55a5-9273-94464be2a91f", + "relation": "embeds", + "ref": "[tables/table-104 Deferred Tax Assets.html]", + "position": { + "start": 157, + "end": 200 + } + }, + { + "target": "a2b52d05-e0c5-5701-ac18-42a1cde30171", + "relation": "embeds", + "ref": "[tables/table-105 Tax Positions.html]", + "position": { + "start": 3223, + "end": 3260 + } + } + ] + } + }, + { + "chunk_id": "133f94fc-19d7-5d48-9e2c-093ebead9d22", + "type": "text", + "content": "Capital Return Program\nOn August 26, 2025, our Board of Directors approved an additional \\$60.0 billion in share repurchase authorization, without expiration. In fiscal years 2026 and 2025, we repurchased 282 million and 310 million shares of our common stock for \\$40.4 billion and \\$34.0 billion, respectively. As of January 25, 2026, we were authorized, subject to certain specifications, to repurchase up to \\$58.5 billion of our common stock.\nFrom January 26, 2026 through February 20, 2026, we repurchased 8 million shares for \\$1.5 billion pursuant to a pre-established trading plan.\nIn fiscal years 2026, 2025, and 2024, we paid cash dividends to our shareholders of \\$974 million, \\$834 million, and \\$395 million, respectively. The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 14 - Shareholders' Equity", + "metadata": { + "length": 917, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Capital", + "Return", + "Program", + "On", + "August", + "26", + "2025", + "Board", + "Directors", + "approved", + "additional", + "60.0", + "billion", + "share", + "repurchase", + "authorization", + "expiration", + "In", + "fiscal", + "years", + "2026", + "repurchased", + "282", + "million", + "310", + "shares", + "common", + "stock", + "40.4", + "34.0", + "As", + "January", + "25", + "authorized", + "subject", + "specifications", + "58.5", + "From", + "February", + "20", + "1.5", + "pursuant", + "pre", + "established", + "trading", + "plan", + "2024", + "paid", + "cash", + "dividends", + "shareholders", + "974", + "834", + "395", + "The", + "payment", + "future", + "continuing", + "determination", + "declaration", + "interests" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fa14a5bf-5466-525f-8aef-a2741560a7e8", + "type": "text", + "content": "We provide tax-qualified defined contribution plans to eligible employees in the U.S. and certain other countries. Our contribution expense for fiscal years 2026, 2025, and 2024 was \\$442 million, \\$314 million, and \\$255 million, respectively.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 15 - Employee Retirement Plans", + "metadata": { + "length": 244, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "We", + "provide", + "tax", + "qualified", + "defined", + "contribution", + "plans", + "eligible", + "employees", + "countries", + "Our", + "expense", + "fiscal", + "years", + "2026", + "2025", + "2024", + "442", + "million", + "314", + "255" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8626abe3-3685-5e39-995f-1e43faca3759", + "type": "text", + "content": "Our Chief Executive Officer is our chief operating decision maker, or CODM, and reviews financial information presented on an operating segment basis for purposes of making decisions and assessing financial performance. Our CODM assesses operating performance of each segment based on regularly provided segment revenue and segment operating income. Operating results by segment include costs or expenses directly attributable to each segment, and costs or expenses that are leveraged across our unified architecture and therefore allocated between our two segments. Our", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 16 - Segment Information", + "metadata": { + "length": 570, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "Chief", + "Executive", + "Officer", + "chief", + "operating", + "decision", + "maker", + "CODM", + "reviews", + "financial", + "information", + "presented", + "segment", + "basis", + "purposes", + "making", + "decisions", + "assessing", + "performance", + "assesses", + "based", + "regularly", + "provided", + "revenue", + "income", + "Operating", + "results", + "include", + "costs", + "expenses", + "directly", + "attributable", + "leveraged", + "unified", + "architecture", + "allocated", + "segments" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "db1e05e0-e753-5263-ba3e-94df1b24e2bd", + "type": "table", + "content": "
Compute & NetworkingGraphicsTotal
(In millions)
Year Ended Jan 25, 2026
Revenue$ 193,479$ 22,459$ 215,938
Other segment items (1)63,33813,30376,641
Operating income$ 130,141$ 9,156$ 139,297
Year Ended Jan 26, 2025
Revenue$ 116,193$ 14,304$ 130,497
Other segment items (1)33,3189,21942,537
Operating income$ 82,875$ 5,085$ 87,960
Year Ended Jan 28, 2024
Revenue$ 47,405$ 13,517$ 60,922
Other segment items (1)15,3897,67123,060
Operating income$ 32,016$ 5,846$ 37,862
", + "path": "tables/table-106 Segment Revenue.html", + "metadata": { + "length": 1053, + "summary": "table-107\nThe table shows segment revenue, other segment items, and operating income for Compute & Networking and Graphics for fiscal years 2024, 2025, and 2026, with totals in millions.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-106 Segment Revenue.html", + "keywords": [ + "Revenue", + "Operating Income", + "Compute & Networking" + ], + "tokens": [] + } + }, + { + "chunk_id": "83448a0c-6819-56d0-b85d-175cc3fef4e2", + "type": "text", + "content": "(Continued)\nCODM reviews expenses on a consolidated basis, and expenses attributable to each segment are not regularly provided to our CODM.\nThe Compute & Networking segment includes our Data Center accelerated computing and networking platforms and AI solutions and software, and Automotive platforms and autonomous and electric vehicle solutions including software.\nThe Graphics segment includes GeForce GPUs for gaming and PCs, and Quadro/NVIDIA RTX GPUs for enterprise workstation graphics.\nCertain expenses are not allocated to either Compute & Networking or Graphics for purposes of making operating decisions or assessing financial performance. The expenses include stock-based compensation expense, corporate infrastructure and support costs, acquisition-related and other costs, and other non-recurring charges and benefits that our CODM deems to be enterprise in nature.\nOur CODM does not review any information regarding total assets on a reportable segment basis. There are no intersegment transactions. The accounting policies for segment reporting are the same as for our consolidated financial statements. The table below presents details of our reportable segments.\n\n[tables/table-106 Segment Revenue.html]\n\n(1) Other segment items primarily include product costs and inventory provisions, compensation and benefits excluding stock-based compensation expense, computing infrastructure expenses, and engineering development costs.\nDepreciation and amortization expense attributable to our Compute & Networking segment for fiscal years 2026, 2025, and 2024 was \\$1.6 billion, \\$732 million, and \\$457 million, respectively. Depreciation and amortization expense attributable to our Graphics segment for fiscal years 2026, 2025, and 2024 was \\$590 million, \\$372 million, and \\$307 million, respectively. Acquisition-related intangible amortization expense is not allocated to either Compute & Networking or Graphics for purposes of making operating decisions or assessing financial performance.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_11", + "metadata": { + "length": 2028, + "summary": "NVIDIA reports two segments: Compute & Networking (Data Center, AI, Automotive) and Graphics (GeForce, Quadro GPUs). The CODM reviews expenses on a consolidated basis; certain costs like stock-based compensation and corporate infrastructure are not allocated. No intersegment transactions exist. Depreciation and amortization for Compute & Networking were $1.6B (2026), $732M (2025), $457M (2024); for Graphics: $590M (2026), $372M (2025), $307M (2024). Acquisition-related intangible amortization is not allocated.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "CODM", + "reviews", + "expenses", + "consolidated", + "basis", + "attributable", + "segment", + "regularly", + "provided", + "The", + "Compute", + "Networking", + "includes", + "Data", + "Center", + "accelerated", + "computing", + "networking", + "platforms", + "AI", + "solutions", + "software", + "Automotive", + "autonomous", + "electric", + "vehicle", + "including", + "Graphics", + "GeForce", + "GPUs", + "gaming", + "PCs", + "Quadro", + "NVIDIA", + "RTX", + "enterprise", + "workstation", + "graphics", + "Certain", + "allocated", + "purposes", + "making", + "operating", + "decisions", + "assessing", + "financial", + "performance", + "include", + "stock", + "based", + "compensation", + "expense", + "corporate", + "infrastructure", + "support", + "costs", + "acquisition", + "related", + "recurring", + "charges", + "benefits", + "deems", + "nature", + "Our", + "review", + "information", + "total", + "assets", + "reportable", + "There", + "intersegment", + "transactions", + "accounting", + "policies", + "reporting", + "statements", + "table", + "presents", + "details", + "segments", + "Other", + "items", + "primarily", + "product", + "inventory", + "provisions", + "excluding", + "engineering", + "development", + "Depreciation", + "amortization", + "fiscal", + "years", + "2026", + "2025", + "2024", + "1.6", + "billion", + "732", + "million", + "457", + "590", + "372", + "307", + "Acquisition", + "intangible" + ], + "keywords": [ + "Compute & Networking", + "Graphics", + "Segment Reporting" + ], + "connect_to": [ + { + "target": "db1e05e0-e753-5263-ba3e-94df1b24e2bd", + "relation": "embeds", + "ref": "[tables/table-106 Segment Revenue.html]", + "position": { + "start": 1183, + "end": 1222 + } + } + ] + } + }, + { + "chunk_id": "66a4acdf-dd15-5138-ada6-e8c45f91e91a", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Segment operating income$ 139,297$ 87,960$ 37,862
Stock-based compensation expense(6,386)(4,737)(3,549)
Unallocated operating expenses(1,997)(1,171)(728)
Acquisition-related and other costs(527)(599)(613)
Interest income2,3001,786866
Interest expense(259)(247)(257)
Other income, net9,0221,034237
Consolidated income before income tax$ 141,450$ 84,026$ 33,818
", + "path": "tables/table-107 Income Breakdown.html", + "metadata": { + "length": 903, + "summary": "table-108\nThe table shows segment operating income, stock-based compensation, unallocated expenses, acquisition costs, interest income/expense, and other income, leading to consolidated income before tax for fiscal years 2024-2026.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-107 Income Breakdown.html", + "keywords": [ + "Segment operating income", + "Consolidated income", + "Stock-based compensation" + ], + "tokens": [] + } + }, + { + "chunk_id": "37ba5250-03c1-597a-b9f5-ee9b8992c5f0", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Geographic Location (1):Revenue based upon Customer Headquarters(In millions)
United States$149,617$ 77,482$ 31,533
Taiwan (2)42,34523,60014,912
China (including Hong Kong)19,67725,04812,330
Other4,2994,3672,147
Total revenue$215,938$ 130,497$ 60,922
", + "path": "tables/table-108 Revenue by Region.html", + "metadata": { + "length": 723, + "summary": "table-109\nThe table shows total revenue of $215,938M for Jan 25, 2026, with US ($149,617M), Taiwan ($42,345M), China ($19,677M), and Other ($4,299M). Prior years: $130,497M (2025) and $60,922M (2024).", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-108 Revenue by Region.html", + "keywords": [ + "Revenue", + "Geographic", + "Fiscal Years" + ], + "tokens": [] + } + }, + { + "chunk_id": "59d925bf-0019-57ed-aba9-cb23bfecaf6a", + "type": "text", + "content": "(Continued)\nA reconciliation of segment operating income to consolidated income before income tax for fiscal years 2026, 2025, and 2024 were as follows:\n\n[tables/table-107 Income Breakdown.html]\n\nRevenue by geographic area is based upon the location of the customers' headquarters. The end customer and shipping location may be different from our customers' headquarters location.\n\n[tables/table-108 Revenue by Region.html]\n\n(1) In the third quarter of fiscal year 2026, we changed to revenue based upon the location of our customers' headquarters as we believe it provides a better representation of the geographic profile of our revenue. Prior period information has been recast to reflect this change.\n(2) In fiscal year 2026, we estimate 76% of Data Center revenue from Taiwan-headquartered customers was attributed to end customers based in the United States and Europe.\nRevenue from sales to customers headquartered outside of the United States accounted for 31%, 41%, and 48% of total revenue for fiscal years 2026, 2025, and 2024, respectively. The increase in revenue to the United States for fiscal years 2026 and 2025 was primarily due to higher U.S.-based Compute & Networking segment demand.\nWe refer to customers who purchase products directly from NVIDIA as direct customers, such as AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. We refer to indirect customers as those who purchase products through our direct customers; indirect customers include CSPs, Neocloud builders, AI model makers, enterprises, and public sector entities. Our revenue is concentrated among a limited number of direct and indirect customers and this trend may continue.\nDirect Customers – For fiscal year 2026, sales to one direct customer represented 22% of total revenue and sales to another direct customer represented 14% of total revenue, all of which were primarily attributable to the Compute & Networking segment.\nFor fiscal year 2025, sales to one direct customer represented 12% of total revenue and sales to two direct customers each represented 11% of total revenue, all of which were primarily attributable to the Compute & Networking segment.\nFor fiscal year 2024, sales to one direct customer represented 13% of total revenue, and were primarily attributable to the Compute & Networking segment.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_12", + "metadata": { + "length": 2438, + "summary": "The passage discusses NVIDIA's revenue and segment operating income reconciliation for fiscal years 2024-2026. Revenue is reported based on customer headquarters location, with a change in fiscal 2026. Non-US revenue was 31%, 41%, and 48% of total for 2026, 2025, and 2024 respectively. Revenue concentration among direct customers is noted: in fiscal 2026, one direct customer accounted for 22% of total revenue and another for 14%, primarily in Compute & Networking. Similar concentrations are reported for prior years. The text also defines direct and indirect customers and notes that 76% of Data Center revenue from Taiwan-headquartered customers was attributed to US and European end customers in fiscal 2026.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "reconciliation", + "segment", + "operating", + "income", + "consolidated", + "tax", + "fiscal", + "years", + "2026", + "2025", + "2024", + "Revenue", + "geographic", + "area", + "based", + "location", + "customers", + "headquarters", + "The", + "end", + "customer", + "shipping", + "In", + "quarter", + "year", + "changed", + "revenue", + "representation", + "profile", + "Prior", + "period", + "information", + "recast", + "reflect", + "change", + "estimate", + "76%", + "Data", + "Center", + "Taiwan", + "headquartered", + "attributed", + "United", + "States", + "Europe", + "sales", + "accounted", + "31%", + "41%", + "48%", + "total", + "increase", + "primarily", + "due", + "higher", + "Compute", + "Networking", + "demand", + "We", + "refer", + "purchase", + "products", + "directly", + "NVIDIA", + "direct", + "AIBs", + "distributors", + "ODMs", + "OEMs", + "CSPs", + "AI", + "model", + "makers", + "system", + "integrators", + "Certain", + "internal", + "resources", + "party", + "complete", + "build", + "indirect", + "include", + "Neocloud", + "builders", + "enterprises", + "public", + "sector", + "entities", + "Our", + "concentrated", + "limited", + "number", + "trend", + "continue", + "Direct", + "Customers", + "For", + "represented", + "22%", + "14%", + "attributable", + "12%", + "11%", + "13%" + ], + "keywords": [ + "Revenue", + "Segment", + "Customers" + ], + "connect_to": [ + { + "target": "66a4acdf-dd15-5138-ada6-e8c45f91e91a", + "relation": "embeds", + "ref": "[tables/table-107 Income Breakdown.html]", + "position": { + "start": 154, + "end": 194 + } + }, + { + "target": "37ba5250-03c1-597a-b9f5-ee9b8992c5f0", + "relation": "embeds", + "ref": "[tables/table-108 Revenue by Region.html]", + "position": { + "start": 382, + "end": 423 + } + } + ] + } + }, + { + "chunk_id": "f1882c27-b91e-5800-97f4-a92cf782be76", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Revenue by End Market:(In millions)
Data Center$ 193,737$ 115,186$ 47,525
Compute162,361102,19638,950
Networking31,37612,9908,575
Gaming16,04211,35010,447
Professional Visualization3,1911,8781,553
Automotive2,3491,6941,091
OEM and Other619389306
Total revenue$ 215,938$ 130,497$ 60,922
", + "path": "tables/table-109 NVIDIA Revenue.html", + "metadata": { + "length": 824, + "summary": "table-110\nNVIDIA's total revenue grew from $60.9B in FY2024 to $130.5B in FY2025 and $215.9B in FY2026, driven by Data Center growth ($47.5B to $193.7B).", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-109 NVIDIA Revenue.html", + "keywords": [ + "Revenue", + "Data Center", + "Gaming" + ], + "tokens": [] + } + }, + { + "chunk_id": "0194b70f-150b-508c-aa9d-936070fefaac", + "type": "table", + "content": "
Jan 25, 2026Jan 26, 2025
Long-lived assets:(In millions)
United States$ 5,125$ 3,626
Taiwan3,2191,481
Israel1,471840
Other568336
Total long-lived assets$ 10,383$ 6,283
", + "path": "tables/table-110 Long-lived assets.html", + "metadata": { + "length": 431, + "summary": "table-111\nThe table shows long-lived assets by region for Jan 25, 2026 and Jan 26, 2025, with totals of $10,383M and $6,283M respectively.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-110 Long-lived assets.html", + "keywords": [ + "long-lived assets", + "geographic breakdown", + "financial data" + ], + "tokens": [] + } + }, + { + "chunk_id": "95709c3c-f540-5c62-aa6c-087eedf8a9c2", + "type": "text", + "content": "(Continued)\nThe following table summarizes revenue by specialized markets:\n\n[tables/table-109 NVIDIA Revenue.html]\n\nThe following table presents summarized information for long-lived assets by country. Long-lived assets consist of property and equipment and exclude other assets, operating lease assets, goodwill, and intangible assets.\n\n[tables/table-110 Long-lived assets.html]", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_13", + "metadata": { + "length": 383, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "The", + "table", + "summarizes", + "revenue", + "specialized", + "markets", + "presents", + "summarized", + "information", + "long", + "lived", + "assets", + "country", + "Long", + "consist", + "property", + "equipment", + "exclude", + "operating", + "lease", + "goodwill", + "intangible" + ], + "keywords": [], + "connect_to": [ + { + "target": "f1882c27-b91e-5800-97f4-a92cf782be76", + "relation": "embeds", + "ref": "[tables/table-109 NVIDIA Revenue.html]", + "position": { + "start": 76, + "end": 114 + } + }, + { + "target": "0194b70f-150b-508c-aa9d-936070fefaac", + "relation": "embeds", + "ref": "[tables/table-110 Long-lived assets.html]", + "position": { + "start": 338, + "end": 379 + } + } + ] + } + }, + { + "chunk_id": "e7fe0e58-c8c0-5713-8ef9-b3e8dd4ed684", + "type": "table", + "content": "
Operating Lease Obligations
(In millions)
Fiscal Year:
2027$ 493
2028485
2029457
2030381
2031314
2032 and thereafter1,494
Total3,624
Less imputed interest680
Present value of net future minimum lease payments2,944
Less short-term operating lease liabilities372
Long-term operating lease liabilities$ 2,572
", + "path": "tables/table-111 Lease Obligations.html", + "metadata": { + "length": 691, + "summary": "table-112\nThe table shows operating lease obligations by fiscal year from 2027 to 2032+, totaling $3,624 million. After subtracting imputed interest of $680 million, the present value of net minimum lease payments is $2,944 million, with long-term liabilities of $2,572 million.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-111 Lease Obligations.html", + "keywords": [ + "Operating Lease", + "Liabilities", + "Future Payments" + ], + "tokens": [] + } + }, + { + "chunk_id": "02acde18-aaae-5f9b-b7a8-7e55e4758bf0", + "type": "text", + "content": "Our lease obligations primarily consist of operating leases for our offices and data centers, with lease periods expiring between fiscal years 2027 and 2041.\nFuture minimum lease obligations under our non-cancelable lease agreements as of January 25, 2026 were as follows:\n\n[tables/table-111 Lease Obligations.html]\n\nBetween fiscal years 2027 and 2030, we expect to commence leases with future obligations of \\$22.7 billion, primarily data center leases to support our research and development efforts, with lease terms of 1.8 to 20 years.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 17 - Leases", + "metadata": { + "length": 545, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Our", + "lease", + "obligations", + "primarily", + "consist", + "operating", + "leases", + "offices", + "data", + "centers", + "periods", + "expiring", + "fiscal", + "years", + "2027", + "2041", + "Future", + "minimum", + "cancelable", + "agreements", + "January", + "25", + "2026", + "Between", + "2030", + "expect", + "commence", + "future", + "22.7", + "billion", + "center", + "support", + "research", + "development", + "efforts", + "terms", + "1.8", + "20" + ], + "keywords": [], + "connect_to": [ + { + "target": "e7fe0e58-c8c0-5713-8ef9-b3e8dd4ed684", + "relation": "embeds", + "ref": "[tables/table-111 Lease Obligations.html]", + "position": { + "start": 274, + "end": 315 + } + } + ] + } + }, + { + "chunk_id": "cf6a3987-8f8d-5ac2-986a-36f5c3fee832", + "type": "table", + "content": "
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Supplemental cash flows information
Operating cash flow used for operating leases$ 428$ 313$ 286
Operating lease assets obtained in exchange for lease obligations$ 1,439$ 877$ 531
", + "path": "tables/table-112 Lease Cash Flows.html", + "metadata": { + "length": 488, + "summary": "table-113\nThe table shows supplemental cash flow information for operating leases, including cash used and assets obtained in exchange for lease obligations for fiscal years 2024-2026.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-112 Lease Cash Flows.html", + "keywords": [ + "operating leases", + "cash flows", + "lease obligations" + ], + "tokens": [] + } + }, + { + "chunk_id": "7e964f0f-f433-5fd2-b22a-16b49278df80", + "type": "table", + "content": "
DescriptionBalance at Beginning of PeriodAdditionsDeductionsBalance at End of Period
(In millions)
Fiscal year 2026
Allowance for doubtful accounts$ 4$ — (1)$ — (1)$ 4
Sales return allowance$ 82$ 188 (2)$ (100) (4)$ 170
Deferred tax valuation allowance$ 1,610$ 31 (3)$ (873) (3)$ 768
Fiscal year 2025
Allowance for doubtful accounts$ 4$ — (1)$ — (1)$ 4
Sales return allowance$ 109$ 151 (2)$ (178) (4)$ 82
Deferred tax valuation allowance$ 1,552$ 58 (3)$ — (3)$ 1,610
Fiscal year 2024
Allowance for doubtful accounts$ 4$ — (1)$ — (1)$ 4
Sales return allowance$ 26$ 213 (2)$ (130) (4)$ 109
Deferred tax valuation allowance$ 1,484$ 162 (3)$ (94) (3)$ 1,552
", + "path": "tables/table-113 Allowance Changes.html", + "metadata": { + "length": 1324, + "summary": "table-114\nThe table shows allowance for doubtful accounts, sales returns, and deferred tax valuation allowances for fiscal years 2024-2026, with balances and changes.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-113 Allowance Changes.html", + "keywords": [ + "Allowance", + "Fiscal Years", + "Valuation" + ], + "tokens": [] + } + }, + { + "chunk_id": "4554987b-23e0-53c9-82e1-5858603d149d", + "type": "table", + "content": "
Exhibit No.Exhibit DescriptionIncorporated by ReferenceFiling Date
Schedule/FormExhibit
3.1Restated Certificate of Incorporation10-K3.13/18/2022
3.2Amendment to Restated Certificate of Incorporation of NVIDIA Corporation8-K3.16/6/2022
3.3Amendment to Restated Certificate of Incorporation of NVIDIA Corporation8-K3.16/7/2024
3.4Bylaws of NVIDIA Corporation, Amended and Restated as of March 12, 20248-K3.13/14/2024
4.1Reference is made to Exhibits 3.1, 3.2, 3.3 and 3.4
4.2Specimen Stock CertificateS-1/A4.24/24/1998
4.3Indenture, dated as of September 16, 2016, by and between the Company and Computershare Trust Company, N.A., as successor to Wells Fargo Bank, National Association, as Trustee8-K4.19/16/2016
4.4Officers' Certificate, dated as of September 16, 20168-K4.29/16/2016
4.5Form of 2026 Note8-KAnnex B-1 to Exhibit 4.29/16/2016
4.6Description of Securities10-K4.62/26/2025
4.7Officers' Certificate, dated as of March 31, 20208-K4.23/31/2020
4.8Form of 2030 Note8-KAnnex A-1 to Exhibit 4.23/31/2020
4.9Form of 2040 Note8-KAnnex B-1 to Exhibit 4.23/31/2020
4.10Form of 2050 Note8-KAnnex C-1 to Exhibit 4.23/31/2020
4.11Form of 2060 Note8-KAnnex D-1 to Exhibit 4.23/31/2020
4.12Officers' Certificate, dated as of June 16, 20218-K4.26/16/2021
4.13Form of 2028 Note8-KAnnex C-1 to Exhibit 4.26/16/2021
4.14Form of 2031 Note8-KAnnex D-1 to Exhibit 4.26/16/2021
10.1Form of Indemnity Agreement between NVIDIA Corporation and each of its directors and officers8-K10.13/7/2006
10.2+Amended and Restated 2007 Equity Incentive Plan10-Q10.18/28/2024
10.3+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Deferred Restricted Stock Unit Grant Notice and Deferred Restricted Stock Unit Agreement (2016)10-K10.263/12/2015
10.4+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Restricted Stock Unit Grant Notice and Restricted Stock Unit Agreement (2016)10-K10.273/12/2015
10.5+Amended and Restated 2007 Equity Incentive Plan - Global Performance-Based Restricted Stock Unit Grant Notice and Performance-Based Restricted Stock Unit Agreement (2019)8-K10.13/11/2019
10.6+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2021)10-Q10.25/26/2021
10.7+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2022)10-K10.163/18/2022
10.8+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2023)10-K10.142/24/2023
10.9+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2024) (version 1)10-Q10.25/29/2024
10.10+Amended and Restated 2007 Equity Incentive Plan - Global Performance-Based Restricted Stock Unit Grant Notice and Performance-Based Restricted Stock Unit Agreement (2024)10-Q10.35/29/2024
10.11+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2024) (version 2)10-Q10.111/20/2024
10.12+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2025)10-K10.132/26/2025
10.13+Amended and Restated 2007 Equity Incentive Plan - Global Performance-Based Restricted Stock Unit Grant Notice and Performance-Based Restricted Stock Unit Agreement (2025)10-K10.142/26/2025
10.14+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Deferred Restricted Stock Unit Grant Notice and Deferred Restricted Stock Unit Agreement (2025)10-Q10.18/27/2025
10.15+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Restricted Stock Unit Grant Notice and Restricted Stock Unit Agreement (2025)10-Q10.28/27/2025
10.16+Amended and Restated 2012 Employee Stock Purchase Plan10-K10.152/26/2025
10.17+Variable Compensation Plan - Fiscal Year 20258-K10.13/14/2024
10.18+Variable Compensation Plan - Fiscal Year 20268-K10.13/7/2025
19.1*NVIDIA Corporation Insider Trading Policy
21.1*Subsidiaries of Registrant
23.1*Consent of PricewaterhouseCoopers LLP
24.1*Power of Attorney (included in signature page)
31.1*Certification of Chief Executive Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
31.2*Certification of Chief Financial Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
32.1#*Certification of Chief Executive Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
32.2#*Certification of Chief Financial Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
97.1+Compensation Recovery Policy, as amended and restated November 30, 202310-K97.12/21/2024
101.INS*XBRL Instance Document
101.SCH*XBRL Taxonomy Extension Schema Document
101.CAL*XBRL Taxonomy Extension Calculation Linkbase Document
101.DEF*XBRL Taxonomy Extension Definition Linkbase Document
101.LAB*XBRL Taxonomy Extension Labels Linkbase Document
101.PRE*XBRL Taxonomy Extension Presentation Linkbase Document
104Cover Page Interactive Data File - the cover page interactive data file does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document
", + "path": "tables/table-114 Exhibit Index.html", + "metadata": { + "length": 8215, + "summary": "table-115\nAn exhibit index listing NVIDIA's incorporation documents, equity plans, and certifications, with references to SEC filings like 10-K and 8-K.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-114 Exhibit Index.html", + "keywords": [ + "Exhibits", + "SEC Filings", + "NVIDIA" + ], + "tokens": [] + } + }, + { + "chunk_id": "39caf71e-0f06-527e-af96-362a3eb94ac2", + "type": "text", + "content": "(Continued)\nOperating lease costs for fiscal years 2026, 2025, and 2024 were \\$462 million, \\$356 million, and \\$269 million, respectively. Short-term and variable lease costs for fiscal years 2026, 2025, and 2024 were not significant.\nOther information related to leases was as follows:\n\n[tables/table-112 Lease Cash Flows.html]\n\nAs of January 25, 2026, our operating leases have a weighted average remaining lease term of 8.8 years and a weighted average discount rate of 4.38%. As of January 26, 2025, our operating leases had a weighted average remaining lease term of 6.5 years and a weighted average discount rate of 4.16%.\nNVIDIA Corporation and Subsidiaries\nSchedule II – Valuation and Qualifying Accounts\n\n[tables/table-113 Allowance Changes.html]\n\n(1) Additions represent either expense or acquired balances and deductions represent write-offs.\n(2) Additions represent estimated product returns charged as a reduction to revenue or an acquired balance.\n(3) Additional valuation allowance on deferred tax assets not likely to be realized. Additions represent additional valuation allowance on certain state and other deferred tax assets. Deductions mainly represent the release of valuation allowance on certain state deferred tax assets. Refer to Note 13 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information.\n(4) Represents sales returns.\nExhibit Index\n\n[tables/table-114 Exhibit Index.html]\n\n\\* Filed herewith.\n+ Management contract or compensatory plan or arrangement.\nIn accordance with Item 601(b)(32)(ii) of Regulation S-K and SEC Release Nos. 33-8238 and 34-47986, Final Rule: Management's Reports on Internal Control Over Financial Reporting and Certification of Disclosure in Exchange Act Periodic Reports, the certifications furnished in Exhibits 32.1 and 32.2 hereto are deemed to accompany this Annual Report on Form 10-K and will not be deemed \"filed\" for purpose of Section 18 of the Exchange Act. Such certifications will not be deemed to be incorporated by reference into any filing under the Securities Act or the Exchange Act, except to the extent that the registrant specifically incorporates it by reference.\n^ Certain exhibits and schedules have been omitted in accordance with Regulation S-K Item 601(a)(5).\nCopies of above exhibits not contained herein are available to any shareholder upon written request to:\nInvestor Relations: NVIDIA Corporation, 2788 San Tomas Expressway, Santa Clara, CA 95051", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_14", + "metadata": { + "length": 2625, + "summary": "The passage details NVIDIA's operating lease costs for fiscal years 2026, 2025, and 2024 ($462M, $356M, $269M), with short-term and variable costs not significant. As of January 25, 2026, leases have a weighted average term of 8.8 years and discount rate of 4.38%. It also includes Schedule II on valuation and qualifying accounts, with notes on additions and deductions for allowances and returns. Finally, an Exhibit Index lists exhibits filed with the SEC, including certifications and management contracts, with instructions for shareholder requests.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Continued", + "Operating", + "lease", + "costs", + "fiscal", + "years", + "2026", + "2025", + "2024", + "462", + "million", + "356", + "269", + "Short", + "term", + "variable", + "significant", + "Other", + "information", + "related", + "leases", + "As", + "January", + "25", + "operating", + "weighted", + "average", + "remaining", + "8.8", + "discount", + "rate", + "4.38%", + "26", + "6.5", + "4.16%", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Schedule", + "II", + "Valuation", + "Qualifying", + "Accounts", + "Additions", + "represent", + "expense", + "acquired", + "balances", + "deductions", + "write", + "offs", + "estimated", + "product", + "returns", + "charged", + "reduction", + "revenue", + "balance", + "Additional", + "valuation", + "allowance", + "deferred", + "tax", + "assets", + "realized", + "additional", + "state", + "Deductions", + "release", + "Refer", + "Note", + "13", + "Notes", + "Consolidated", + "Financial", + "Statements", + "Part", + "IV", + "Item", + "15", + "Annual", + "Report", + "Form", + "10", + "Represents", + "sales", + "Exhibit", + "Index", + "Filed", + "herewith", + "Management", + "contract", + "compensatory", + "plan", + "arrangement", + "In", + "accordance", + "601", + "32", + "ii", + "Regulation", + "SEC", + "Release", + "Nos", + "33", + "8238", + "34", + "47986", + "Final", + "Rule", + "Reports", + "Internal", + "Control", + "Over", + "Reporting", + "Certification", + "Disclosure", + "Exchange", + "Act", + "Periodic", + "certifications", + "furnished", + "Exhibits", + "32.1", + "32.2", + "hereto", + "deemed", + "accompany", + "filed", + "purpose", + "Section", + "18", + "Such", + "incorporated", + "reference", + "filing", + "Securities", + "extent", + "registrant", + "specifically", + "incorporates", + "Certain", + "exhibits", + "schedules", + "omitted", + "Copies", + "contained", + "shareholder", + "written", + "request", + "Investor", + "Relations", + "2788", + "San", + "Tomas", + "Expressway", + "Santa", + "Clara", + "CA", + "95051" + ], + "keywords": [ + "Operating leases", + "Valuation accounts", + "Exhibits" + ], + "connect_to": [ + { + "target": "cf6a3987-8f8d-5ac2-986a-36f5c3fee832", + "relation": "embeds", + "ref": "[tables/table-112 Lease Cash Flows.html]", + "position": { + "start": 289, + "end": 329 + } + }, + { + "target": "7e964f0f-f433-5fd2-b22a-16b49278df80", + "relation": "embeds", + "ref": "[tables/table-113 Allowance Changes.html]", + "position": { + "start": 715, + "end": 756 + } + }, + { + "target": "4554987b-23e0-53c9-82e1-5858603d149d", + "relation": "embeds", + "ref": "[tables/table-114 Exhibit Index.html]", + "position": { + "start": 1447, + "end": 1484 + } + } + ] + } + }, + { + "chunk_id": "87ba7dbb-a30f-540c-9ecb-7b6af1b559fe", + "type": "text", + "content": "Not Applicable.", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 16. Form 10-K Summary", + "metadata": { + "length": 15, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Not", + "Applicable" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ca262699-fb67-59e5-8719-934302fc7c6e", + "type": "table", + "content": "
SignatureTitleDate
/s/ JEN-HSUN HUANGPresident, Chief Executive Officer and Director (Principal Executive Officer)February 25, 2026
Jen-Hsun Huang
/s/ COLETTE M. KRESSExecutive Vice President and Chief Financial Officer (Principal Financial Officer)February 25, 2026
Colette M. Kress
/s/ DONALD ROBERTSONVice President and Chief Accounting Officer (Principal Accounting Officer)February 25, 2026
Donald Robertson
/s/ TENCH COXEDirectorFebruary 25, 2026
Tench Coxe
/s/ JOHN O. DABIRIDirectorFebruary 25, 2026
John O. Dabiri
/s/ DAWN HUDSONDirectorFebruary 25, 2026
Dawn Hudson
/s/ HARVEY C. JONESDirectorFebruary 25, 2026
Harvey C. Jones
/s/ MELISSA B. LORADirectorFebruary 25, 2026
Melissa B. Lora
/s/ STEPHEN C. NEALDirectorFebruary 25, 2026
Stephen C. Neal
/s/ A. BROOKE SEAWELLDirectorFebruary 25, 2026
A. Brooke Seawell
/s/ AARTI SHAHDirectorFebruary 25, 2026
Aarti Shah
/s/ MARK A. STEVENSDirectorFebruary 25, 2026
Mark A. Stevens
", + "path": "tables/table-115 Signatures.html", + "metadata": { + "length": 1826, + "summary": "table-116\nSignatures of NVIDIA executives and directors including Jen-Hsun Huang, Colette M. Kress, Donald Robertson, and others, dated February 25, 2026.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "tables/table-115 Signatures.html", + "keywords": [ + "Signatures", + "Directors", + "Officers" + ], + "tokens": [] + } + }, + { + "chunk_id": "88b9ab0c-49f6-5dd3-b9af-03e09b828161", + "type": "text", + "content": "Pursuant to the requirements of Section 13 or 15(d) of the Securities Exchange Act of 1934, the Registrant has duly caused this report to be signed on its behalf by the undersigned, thereunto duly authorized, on February 25, 2026.\nNVIDIA Corporation\nBy: /s/ Jen-Hsun Huang\nJen-Hsun Huang\nPresident and Chief Executive Officer\nPower of Attorney\nKNOW ALL PERSONS BY THESE PRESENTS, that each person whose signature appears below constitutes and appoints Jen-Hsun Huang and Colette M. Kress, and each or any one of them, his true and lawful attorney-in-fact and agent, with full power of substitution and resubstitution, for him and in his name, place and stead, in any and all capacities, to sign any and all amendments to this report, and to file the same, with all exhibits thereto, and other documents in connection therewith, with the Securities and Exchange Commission, granting unto said attorneys-in-facts and agents, and each of them, full power and authority to do and perform each and every act and thing requisite and necessary to be done in connection therewith, as fully to all intents and purposes as he might or could do in person, hereby ratifying and confirming all that said attorneys-in-fact and agents, or any of them, or their or his substitutes or substitutes, may lawfully do or cause to be done by virtue hereof.\nPursuant to the requirements of the Securities Exchange Act of 1934, this report has been signed below by the following persons on behalf of the registrant and in the capacities and on the dates indicated.\n\n[tables/table-115 Signatures.html]", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Signatures", + "metadata": { + "length": 1586, + "summary": "The document is a signing page for an SEC filing by NVIDIA Corporation, dated February 25, 2026. It includes the signature of Jen-Hsun Huang as President and CEO, and a Power of Attorney appointing Huang and Colette M. Kress as attorneys-in-fact to sign amendments. The table lists signatories with their titles and dates.", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Pursuant", + "requirements", + "Section", + "13", + "15", + "Securities", + "Exchange", + "Act", + "1934", + "Registrant", + "duly", + "caused", + "report", + "signed", + "behalf", + "undersigned", + "thereunto", + "authorized", + "February", + "25", + "2026", + "NVIDIA", + "Corporation", + "By", + "Jen", + "Hsun", + "Huang", + "President", + "Chief", + "Executive", + "Officer", + "Power", + "Attorney", + "KNOW", + "ALL", + "PERSONS", + "BY", + "THESE", + "PRESENTS", + "person", + "signature", + "appears", + "constitutes", + "appoints", + "Colette", + "Kress", + "true", + "lawful", + "attorney", + "fact", + "agent", + "full", + "power", + "substitution", + "resubstitution", + "place", + "stead", + "capacities", + "sign", + "amendments", + "file", + "exhibits", + "thereto", + "documents", + "connection", + "therewith", + "Commission", + "granting", + "attorneys", + "facts", + "agents", + "authority", + "perform", + "act", + "thing", + "requisite", + "fully", + "intents", + "purposes", + "ratifying", + "confirming", + "substitutes", + "lawfully", + "virtue", + "hereof", + "persons", + "registrant", + "dates" + ], + "keywords": [ + "SEC", + "Power of Attorney", + "NVIDIA" + ], + "connect_to": [ + { + "target": "ca262699-fb67-59e5-8719-934302fc7c6e", + "relation": "embeds", + "ref": "[tables/table-115 Signatures.html]", + "position": { + "start": 1542, + "end": 1576 + } + } + ] + } + }, + { + "chunk_id": "66fd3a6e-cfd1-5191-a005-6bf8fbf72b91", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Corporate Information", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "33e6c4b5-9131-548e-a1ea-ec51e18733e1", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d1ffe601-0d50-528e-9f39-e8488cf13ff8", + "type": "text", + "content": "Founder, President, and\nChief Executive Officer\nNVIDIA Corporation", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Jensen Huang", + "metadata": { + "length": 66, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "NVIDIA", + "Corporation" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "70129e66-fa01-509b-9aa3-e966721d405b", + "type": "text", + "content": "Former Managing Director\nSutter Hill Ventures", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Tench Coxe", + "metadata": { + "length": 45, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Former", + "Managing", + "Director", + "Sutter", + "Hill", + "Ventures" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a13a6ec6-0580-5f40-b712-221587a1072f", + "type": "text", + "content": "Centennial Professor of Aeronautics and Mechanical Engineering California Institute of Technology", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/John O. Dabiri", + "metadata": { + "length": 97, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Centennial", + "Professor", + "Aeronautics", + "Mechanical", + "Engineering", + "California", + "Institute", + "Technology" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e9443ffb-99e0-51c0-83b9-5f43c66e7b57", + "type": "text", + "content": "Former Chief Marketing Officer\nNational Football League,\nand Former Chief Executive Officer\nPepsi-Cola North America", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Dawn Hudson", + "metadata": { + "length": 116, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Former", + "Chief", + "Marketing", + "Officer", + "National", + "Football", + "League", + "Executive", + "Pepsi", + "Cola", + "North", + "America" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8e8eba4c-77df-51ec-b565-db8100479e50", + "type": "text", + "content": "Managing Partner\nSquare Wave Ventures", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Harvey C. Jones", + "metadata": { + "length": 37, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Managing", + "Partner", + "Square", + "Wave", + "Ventures" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "73d53d88-f717-5f06-8d36-c5a19ea57bf7", + "type": "text", + "content": "Former President\nTaco Bell International", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Melissa B. Lora", + "metadata": { + "length": 40, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Former", + "President", + "Taco", + "Bell", + "International" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "83aeda4a-581b-5007-89ae-13a501c3c429", + "type": "text", + "content": "Chairman Emeritus and Senior Counsel\nCooley LLP", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Stephen C. Neal (Lead Director)", + "metadata": { + "length": 47, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Chairman", + "Emeritus", + "Senior", + "Counsel", + "Cooley", + "LLP" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ef8bbfb0-b8ea-5739-be25-cf724fccf9bc", + "type": "text", + "content": "Venture Partner\nNew Enterprise Associates", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/A. Brooke Seawell", + "metadata": { + "length": 41, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Venture", + "Partner", + "New", + "Enterprise", + "Associates" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8b0c79d3-4731-533e-909c-d0d096f9ca40", + "type": "text", + "content": "Former Senior Vice President and Chief Information and Digital Officer Eli Lilly and Company", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Aarti Shah", + "metadata": { + "length": 92, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Former", + "Senior", + "Vice", + "President", + "Chief", + "Information", + "Digital", + "Officer", + "Eli", + "Lilly", + "Company" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1c70ea28-c43e-59cd-9d8f-054086bb067b", + "type": "text", + "content": "Managing Partner\nS-Cubed Capital", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Mark A. Stevens", + "metadata": { + "length": 32, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Managing", + "Partner", + "Cubed", + "Capital" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "93a62017-8d8b-5994-b325-656dd58a8743", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Founders", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "75de691d-0a1c-55cd-94b7-654011df3203", + "type": "text", + "content": "Founder, President, and Chief Executive Officer", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Founders/Jensen Huang", + "metadata": { + "length": 47, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Founder", + "President", + "Chief", + "Executive", + "Officer" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b68f4399-665a-50ac-b561-71fcd80da3e4", + "type": "text", + "content": "Founder and NVIDIA Fellow", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Founders/Chris A. Malachowsky", + "metadata": { + "length": 25, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Founder", + "NVIDIA", + "Fellow" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "bac4b37d-97d3-5908-95fc-f91762ae85e5", + "type": "text", + "content": "", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Executive Team", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3c375090-1633-5980-a7ff-4eb2e9afc210", + "type": "text", + "content": "Executive Vice President and Chief Financial Officer", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Executive Team/Colette M. Kress", + "metadata": { + "length": 52, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Executive", + "Vice", + "President", + "Chief", + "Financial", + "Officer" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0b19815d-05a6-52b8-8fc2-b67d1e5ad73b", + "type": "text", + "content": "Executive Vice President\nWorldwide Field Operations", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Executive Team/Jay Puri", + "metadata": { + "length": 51, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Executive", + "Vice", + "President", + "Worldwide", + "Field", + "Operations" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "707e68b5-b697-5864-9ea4-36d46268ddc1", + "type": "text", + "content": "Executive Vice President\nOperations", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Executive Team/Debora Shoquist", + "metadata": { + "length": 35, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Executive", + "Vice", + "President", + "Operations" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0fb09e99-1dfc-550d-9c7f-0683f378b322", + "type": "text", + "content": "Executive Vice President\nGeneral Counsel and Secretary", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Executive Team/Timothy S. Teter", + "metadata": { + "length": 54, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Executive", + "Vice", + "President", + "General", + "Counsel", + "Secretary" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "dd4c5afb-c805-5756-963e-8094573dedae", + "type": "text", + "content": "PricewaterhouseCoopers LLP\n3155 Olsen Drive, Suite 700\nSan Jose, California 95117", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Independent Accountants", + "metadata": { + "length": 81, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "PricewaterhouseCoopers", + "LLP", + "3155", + "Olsen", + "Drive", + "Suite", + "700", + "San", + "Jose", + "California", + "95117" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "bec7500c-1956-5ede-80eb-f3f3b13463f5", + "type": "text", + "content": "Cooley LLP\n3175 Hanover Street\nPalo Alto, California 94304", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/General Legal Counsel", + "metadata": { + "length": 58, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Cooley", + "LLP", + "3175", + "Hanover", + "Street", + "Palo", + "Alto", + "California", + "94304" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7b016cf5-7fe3-5437-8866-31f9c4552ad2", + "type": "text", + "content": "Computershare\nP.O Box 43006\nProvidence, Rhode Island 02940\nwww.computershare.com/investor", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Transfer Agent and Registrar", + "metadata": { + "length": 89, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "Computershare", + "Box", + "43006", + "Providence", + "Rhode", + "Island", + "02940", + "www", + "computershare", + "investor" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "94059dd9-e86f-522b-9abe-a43bd84190cf", + "type": "text", + "content": "June 24, 2026, at 09:00 a.m. PT\nOnline at:\nwww.virtualshareholder\nmeeting.com/NVDA2026", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Annual Meeting", + "metadata": { + "length": 86, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "June", + "24", + "2026", + "09", + "00", + "PT", + "Online", + "www", + "virtualshareholder", + "meeting", + "NVDA2026" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fe190c0c-b2b2-502c-9b4f-8414a4e67091", + "type": "text", + "content": "A copy of NVIDIA's Form 10-K filed with the SEC will be made available to all shareholders at no charge.\nThe Form 10-K also can be accessed through the SEC website at www.sec.gov, or through NVIDIA's Investor Relations website at investor.nvidia.com.\nTo receive a copy by mail, please contact:", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Form 10-K", + "metadata": { + "length": 293, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "copy", + "NVIDIA", + "Form", + "10", + "filed", + "SEC", + "made", + "shareholders", + "charge", + "The", + "accessed", + "website", + "www", + "sec", + "gov", + "Investor", + "Relations", + "investor", + "nvidia", + "To", + "receive", + "mail", + "contact" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c802bf1d-1f54-5641-9291-845298a9844e", + "type": "image", + "content": "\nNVIDIA Corporation
0\n[images/image-1-NVIDIA Logo.jpg]\n", + "path": "images/image-1-NVIDIA Logo.jpg_2", + "metadata": { + "length": 63, + "summary": "image-1\nNVIDIA Corporation
0", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "file_path": "images/image-1-NVIDIA Logo.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "46abbaa8-a8ce-5561-b657-783c8484ea2b", + "type": "text", + "content": "NVIDIA Corporation\n2788 San Tomas Expressway\nSanta Clara, California 95051\nshareholdermeeting@nvidia.com\n\nNVIDIA Corporation
0\n[images/image-1-NVIDIA Logo.jpg]\n\n
\nnatural_image\nAbstract 3D geometric logo with green spiral and hexagonal shapes on a grid background (no text or symbols)\n
", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Investor Relations", + "metadata": { + "length": 330, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: BUSINESS OVERVIEW, PROXY SUMMARY, IN MEMORIAM ROBERT K. BURGESS, Information About the 2026 Meeting, Information About the Board of Directors and Corporate Governance, Executive Compensation, FORM 10-K, Part I, Part II, Part III, Part IV, Part I_2, Part II_2, Part III_2, Corporate Information", + "tokens": [ + "NVIDIA", + "Corporation", + "2788", + "San", + "Tomas", + "Expressway", + "Santa", + "Clara", + "California", + "95051", + "shareholdermeeting", + "nvidia", + "details", + "summary", + "natural", + "image", + "Abstract", + "3D", + "geometric", + "logo", + "green", + "spiral", + "hexagonal", + "shapes", + "grid", + "background", + "text", + "symbols" + ], + "keywords": [], + "connect_to": [ + { + "target": "c802bf1d-1f54-5641-9291-845298a9844e", + "relation": "embeds", + "ref": "[images/image-1-NVIDIA Logo.jpg]", + "position": { + "start": 136, + "end": 168 + } + } + ] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/doc_nav.json b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/doc_nav.json new file mode 100644 index 000000000..8d0804fb4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/doc_nav.json @@ -0,0 +1,3684 @@ +{ + "version": "1.0", + "file_name": "NVIDIA FY26 Annual Report.pdf", + "stats": { + "total_chunks": 551, + "text_chunks": 349, + "image_chunks": 86, + "table_chunks": 116, + "max_depth": 6 + }, + "sections": [ + { + "title": "Root", + "path": "2026-Annual-Report-Web.pdf", + "level": 1, + "summary": "NVIDIA's fiscal 2026 annual review details the company's role in the AI platform shift, moving from general-purpose computing to accelerated computing that manufactures intelligence in real time. Revenue grew 65% to $216 billion, driven by Data Center revenue of $194 billion. The document describes AI as a five-layer stack (energy, chips, infrastructure, models, applications) and AI factories as revenue-generating infrastructure converting energy into tokens. Key themes include inference overtaking training as the dominant workload, the Blackwell and Vera Rubin platforms, CUDA's 20-year flywheel, open models, agentic AI entering production, physical AI for robotics and autonomous vehicles, and AI as national infrastructure driving the largest infrastructure buildout in history.", + "chunk_count": 1, + "children": [] + }, + { + "title": "BUSINESS OVERVIEW", + "path": "2026-Annual-Report-Web.pdf/BUSINESS OVERVIEW", + "level": 1, + "summary": "Fiscal 2026 marked another year of strong growth for NVIDIA, with revenue surging 65% year on year to \\$215.9 billion reflecting continued momentum across our platform transitions to accelerated computing and AI. Growth was led by Compute & Networking, where Data Center compute revenue grew 59% driven by demand for our Blackwell platform, and Data Center networking revenue increased 142% driven by the ramp of NVLink compute fabric for Blackwell systems, as well as continued growth in Ethernet and InfiniBand. Gross margin was 71.1%, operating income reached \\$130.4 billion and diluted earnings per share increased 67% to \\$4.90. Fiscal 2026 Results [tables/table-2 Revenue Growth.html]", + "chunk_count": 5, + "children": [ + { + "title": "Fiscal 2026 Reportable Segments", + "path": "2026-Annual-Report-Web.pdf/BUSINESS OVERVIEW/Fiscal 2026 Reportable Segments", + "level": 2, + "summary": "Our two reportable segments are “Compute & Networking” and “Graphics”: [tables/table-3 NVIDIA Revenue.html] (1) This figure has been rounded to the nearest tenth of a billion. Due to rounding, the sum of the figures does not equal the total as presented. (2) Segment operating income differs from consolidated operating income as certain expenses are not allocated to either Compute & Networking or Graphics for purposes of making operating decisions or assessing financial performance.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fiscal 2026 Market Platforms", + "path": "2026-Annual-Report-Web.pdf/BUSINESS OVERVIEW/Fiscal 2026 Market Platforms", + "level": 2, + "summary": "Our platforms address four large markets where our expertise is critical: (1) This figure has been rounded to the nearest tenth of a billion. Due to rounding, the sum of the figures does not equal the total as presented.
0 [images/image-26-Server Rack Hardware.jpg]
natural_image Exterior view of a black server rack with multiple drive bays (no visible text or labels)
Data Center
0 [images/image-27-NVIDIA RTX 4090.jpg]
natural_image Black and white electronic device with green abstract background, no visible text or symbols
Gaming
0 [images/image-28-Photo of a laptop workspace.jpg]
text_image Desktop scene with laptop displaying a webpage interface and a cartoon character, next to a bookshelf and office items
Professional Visualization
0 [images/image-29-Automotive X-Ray Visualization.jpg]
natural_image X-ray diagram of a car showing internal components and wiring (no text or labels)
Automotive \\$193.7 billion revenue \\$16.0 billion revenue \\$3.2 billion revenue \\$2.3 billion revenue up 68% year on year up 41% year on year up 70% year on year up 39% year on year", + "chunk_count": 1, + "children": [] + }, + { + "title": "Business Highlights", + "path": "2026-Annual-Report-Web.pdf/BUSINESS OVERVIEW/Business Highlights", + "level": 2, + "summary": "NVIDIA announced major platform advancements including the Blackwell Ultra platform with 50x higher throughput and 35x lower token cost versus Hopper, extended inference leadership through co-design, and the Vera Rubin platform for agentic AI with up to 10x lower token cost. They entered a licensing agreement with Groq for the Groq 3 LPX accelerator, announced partnerships with Anthropic, Meta, Nokia, and OpenAI, and accelerated AI market development with open models like Nemotron, Cosmos, and Alpamayo. Physical AI milestones include $6 billion in Fiscal 2026 revenue, Mercedes-Benz L4-ready vehicles, and an Uber partnership. Quantum computing advances include NVQLink GPU-QPU interconnect and the ABCI-Q supercomputer.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fiscal 2026 Shareholder Returns", + "path": "2026-Annual-Report-Web.pdf/BUSINESS OVERVIEW/Fiscal 2026 Shareholder Returns", + "level": 2, + "summary": "Total Shareholder Return (TSR)\\* - Turbocharged AI market development with an accelerating release cadence of NVIDIA open models including Nemotron for agentic AI, Cosmos for physical AI, and Alpamayo for autonomous vehicles
0 [images/image-30-Growth Over Time.jpg]
bar chart [tables/table-4 Returns by Period.html]
\\*Represents cumulative stock price appreciation with dividends reinvested and is measured for the applicable fiscal year periods based on the closing price (\\$187.67) of NVIDIA's common stock on January 23, 2026, the last trading day before the end of our Fiscal 2026, as reported by Nasdaq. Return of Capital to Shareholders (in billions)
0 [images/image-31-Financial Returns by Period.jpg]
stacked bar chart [tables/table-5 Dividend & Buyback.html] The chart displays a single bar for stock repurchases and a separate purple bar for dividends. Values are explicitly labeled on each segment: $41.4 for 1-Year, $86.3 for 3-Year, and $97.2 for 5-Year. The total value per segment is also displayed.
Please see our Form 10-K for more financial information for Fiscal 2026.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "PROXY SUMMARY", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY", + "level": 1, + "summary": "This summary highlights information contained elsewhere in the proxy statement. This summary does not contain all the information that you should consider, and you should read the entire proxy statement carefully before voting. 2026 Annual Meeting of Stockholders Date and time: Wednesday, June 24, 2026 at 9:00 a.m. Pacific Time Location: Virtually at www.virtualshareholdermeeting.com/NVDA2026 Record date: Stockholders as of April 27, 2026 are entitled to vote Admission to meeting: You will need your Control Number to attend the 2026 Meeting", + "chunk_count": 12, + "children": [ + { + "title": "Voting Matters and Board Recommendations", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Voting Matters and Board Recommendations", + "level": 2, + "summary": "A summary of the 2026 Meeting proposals is below. Every stockholder's vote is important. Our Board urges you to vote your shares FOR each of the nominees listed in Proposal 1, FOR Proposals 2 and 3, and AGAINST Proposals 4, 5, 6, and 7. [tables/table-6 Proxy Voting Items.html] \\* Because this is a routine proposal, there are no broker non-votes.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Election of Directors (Proposal 1)", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Election of Directors (Proposal 1)", + "level": 2, + "summary": "The following table provides summary information about each director nominee: [tables/table-7 Board of Directors.html] (1) Qualified as an AC financial expert. (2) Previously served as a member of our Board from 1993 until 2006. The Board recommends that you vote FOR each director nominee.", + "chunk_count": 3, + "children": [ + { + "title": "Recent Refreshment and Nominee Qualifications", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Election of Directors (Proposal 1)/Recent Refreshment and Nominee Qualifications", + "level": 3, + "summary": "The text describes the Board's approach to composition, emphasizing a balance of continuity and new perspectives, with one-third of independent directors joining since 2020. It notes the appointment of Suzanne Nora Johnson to the Board effective July 13, 2026, and her expected committee assignment. The NCGC oversees skills and attributes alignment with company strategy.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Corporate Governance Highlights", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Election of Directors (Proposal 1)/Corporate Governance Highlights", + "level": 3, + "summary": "Our Board is committed to strong corporate governance to promote the long-term interests of the Company and our stockholders. We seek a collaborative approach to stockholder issues that affect our business and to ensure that our stockholders see our governance and executive pay practices as well-structured. In Fall 2025, we contacted our top institutional stockholders, representing an aggregate ownership of 34% of our shares, to gain insights into their views on our corporate governance, compensation and corporate sustainability practices. Highlights of our corporate governance practices include: √ All directors independent, except for our CEO √ Independent Lead Director √ Proxy access √ Declassified Board √ Majority voting for directors √ Active Board oversight of risk management √ Stockholders can call a special meeting √ 75% or greater attendance by each director at meetings of the Board and applicable committees √ Independent directors frequently meet in executive sessions √ At least annual Board and committee self-assessments √ Annual stockholder outreach, including Lead Director participation √ Stock ownership guidelines for our directors and NEOs", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Advisory Approval of Executive Compensation for Fiscal 2026 (Proposal 2)", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Advisory Approval of Executive Compensation for Fiscal 2026 (Proposal 2)", + "level": 2, + "summary": "We are asking our stockholders to cast a non-binding vote, also known as “say-on-pay,” to approve our NEOs’ compensation. The Board believes that our compensation policies and practices are effective in achieving our goals of paying for performance, providing competitive pay so that we may attract and retain a high-caliber executive team, aligning our executives’ interests with those of our stockholders to create long-term value, and achieving simplicity and transparency with our compensation program. The Board and our stockholders have approved holding our “say-on-pay” votes annually. The Board recommends that you vote FOR Proposal 2.", + "chunk_count": 3, + "children": [ + { + "title": "Executive Compensation Highlights", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Advisory Approval of Executive Compensation for Fiscal 2026 (Proposal 2)/Executive Compensation Highlights", + "level": 3, + "summary": "Our executive compensation program is designed to reward performance. We use compensation elements that align our NEOs' interests with those of stockholders to drive long-term value. Our NEOs' compensation is heavily weighted toward performance-based, \"at-risk\" variable cash and long-term equity awards, which are earned only if the Company achieves pre-established financial metrics, with caps on maximum payouts. In recent years, over 90% of our CEO's and over 45% of our other NEOs' total target pay have been performance-based and at-risk. Additionally, 100% of our CEO's equity awards have been granted solely as PSUs. For Fiscal 2026, the CC kept NEO cash compensation constant as compared to Fiscal 2025, but increased target equity opportunities for all NEOs to recognize the complexity and scope of their roles and responsibilities. The CC also set Fiscal 2026 target equity opportunities at levels that aligned total target pay for all NEOs other than our CEO to reflect their comparable, significant contributions and impact on corporate performance. To motivate executives, the CC set the Base Compensation Plan revenue goal for the Fiscal 2026 Variable Cash Plan in line with our record Fiscal 2025 results and the Base Compensation Plan Non-GAAP Operating Income goal for SY PSUs granted in Fiscal 2026 in line with our corresponding Fiscal 2025 Stretch Compensation Plan goal.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Financial Performance and Link to Executive Pay", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Advisory Approval of Executive Compensation for Fiscal 2026 (Proposal 2)/Financial Performance and Link to Executive Pay", + "level": 3, + "summary": "The passage discusses executive pay opportunities tied to financial measures driving business value. It references tables showing goals for Fiscal 2026, performance achievement, and impacts on pay. Adjustments to revenue and Non-GAAP Operating Income goals due to export controls are noted. The text also covers a proposal to ratify PwC as independent auditor for Fiscal 2027.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Stockholder Proposal 4", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Stockholder Proposal 4", + "level": 2, + "summary": "A stockholder is asking our Board to take steps to replace the supermajority voting provisions in our Charter and Bylaws with a simple majority standard. The proposal is advisory only and approval of the proposal would not, by itself, implement a simple majority voting standard. At our 2025 Meeting, the Board proposed and recommended that stockholders approve an amendment and restatement of our Charter to eliminate the remaining supermajority voting provisions, which did not pass. The Board recommends that you vote AGAINST Proposal 4.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stockholder Proposal 5", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Stockholder Proposal 5", + "level": 2, + "summary": "A stockholder is asking our Board to evaluate and report on faith-based community resource groups. NVIDIA is committed to ensuring that our employees are protected and supported through enterprise-wide policies, practices, and reporting mechanisms. Our Board believes that our existing policies, practices, and feedback mechanisms already address the issues raised in Proposal 5. The Board recommends that you vote AGAINST Proposal 5.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stockholder Proposal 6", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Stockholder Proposal 6", + "level": 2, + "summary": "A stockholder is asking our Board to evaluate and report on civil rights and non-discrimination related to DEI. NVIDIA complies with applicable law and is committed to equal employment opportunity requirements. In addition, the Board maintains strong oversight and compliance mechanisms to ensure these policies and practices are effectively implemented. Our Board believes these oversight structures provide sufficient and comprehensive governance and accountability for the policies and practices addressed by Proposal 6. The Board recommends that you vote AGAINST Proposal 6.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stockholder Proposal 7", + "path": "2026-Annual-Report-Web.pdf/PROXY SUMMARY/Stockholder Proposal 7", + "level": 2, + "summary": "A stockholder is asking our Board to report and disclose GHG emissions from the use of the Company's sold products. Our Board believes Proposal 7 is unnecessary given NVIDIA's history of robust climate reporting and targets, and our future disclosure plans. The Board believes NVIDIA's management is best positioned to make decisions regarding the specific content and methodology of our climate disclosures. The Board recommends that you vote AGAINST Proposal 7.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "IN MEMORIAM ROBERT K. BURGESS", + "path": "2026-Annual-Report-Web.pdf/IN MEMORIAM ROBERT K. BURGESS", + "level": 1, + "summary": "ROBERT K. BURGESS The Board of Directors and the entire NVIDIA family mourn the passing of Robert K. Burgess, who served as a valued member of our Board from 2011 to 2025. Over his 14 years of dedicated service, Rob provided exceptional leadership and unwavering commitment to the Company and its stockholders, offering invaluable guidance throughout a period of significant growth. Beyond his many professional contributions, Rob was admired for his warmth, candor, and sense of humor. He will be deeply missed. PROXY STATEMENT FOR THE 2026 ANNUAL MEETING OF STOCKHOLDERS - JUNE 24, 2026", + "chunk_count": 1, + "children": [] + }, + { + "title": "Information About the 2026 Meeting", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting", + "level": 1, + "summary": "Your proxy is being solicited on behalf of the Board for use at the 2026 Meeting. Our 2026 Meeting will take place virtually on Wednesday, June 24, 2026 at 9:00 a.m. Pacific Time.", + "chunk_count": 25, + "children": [ + { + "title": "Virtual Meeting Philosophy and Benefits", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Virtual Meeting Philosophy and Benefits", + "level": 2, + "summary": "The Board believes holding the 2026 Meeting in a virtual format invites participation from stockholders regardless of location, while reducing costs to stockholders and the Company. This balance allows the 2026 Meeting to remain focused on matters directly relevant to stockholder interests in an efficient way. We have designed the virtual format to protect stockholder rights, including by offering multiple opportunities to ask questions, publishing answers to questions received for the 2026 Meeting on our Investor Relations website, and providing a replay of the webcast after the 2026 Meeting.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Meeting Attendance", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Meeting Attendance", + "level": 2, + "summary": "If you were an NVIDIA stockholder as of the close of business on the April 27, 2026 record date, or if you hold a valid proxy, you can attend, ask questions before and during, and vote at our 2026 Meeting at www.virtualshareholdermeeting.com/NVDA2026. Our 2026 Meeting will be held virtually; use the Control Number included on your Notice or printed proxy card to enter. Anyone can also listen to the 2026 Meeting live; if you encounter any difficulties accessing the virtual 2026 Meeting, please call the technical support number available at www.virtualshareholdermeeting.com/NVDA2026. A replay of the webcast will be available at www.nvidia.com/proxy through June 24, 2027. Even if you plan to attend the 2026 Meeting virtually, we recommend that you also vote by proxy as described below so that your vote will be counted if you later decide not to attend.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Asking Questions", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Asking Questions", + "level": 2, + "summary": "We encourage stockholders to submit questions through our pre-meeting forum at www.proxyvote.com (using the Control Number included on your Notice or printed proxy card), as well as during the 2026 Meeting at www.virtualshareholdermeeting.com/NVDA2026. During the 2026 Meeting, we will answer as many stockholder-submitted questions related to the business of the 2026 Meeting as time permits. We will publish on our Investor Relations website answers to questions as soon as practicable after the 2026 Meeting. We intend to group questions and answers by topic and answer substantially similar questions only once. To promote fairness to all stockholders and efficient use of the Company's resources, we will respond to one question per stockholder. We reserve the right to exclude questions that are not pertinent to company business or are otherwise unsuitable for the conduct of the 2026 Meeting.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Quorum and Voting", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Quorum and Voting", + "level": 2, + "summary": "To hold our 2026 Meeting, we need a majority of the outstanding shares entitled to vote at the close of business on the April 27, 2026 record date, or a quorum, represented at the 2026 Meeting either by attendance virtually or by proxy. On April 27, 2026, there were 24,220,525,225 shares of common stock outstanding and entitled to vote, meaning 12,110,262,613 shares must be represented at the 2026 Meeting or by proxy to have a quorum. A list of stockholders entitled to vote at the close of business on the record date will be available at our headquarters, 2788 San Tomas Expressway, Santa Clara, California, for 10 days prior to the 2026 Meeting to stockholders of record for any legally valid purpose germane to the 2026 Meeting. To schedule an appointment to view the stockholder list during this period, please contact us at shareholdermeeting@nvidia.com. Your shares will count towards the quorum only if you submit a valid proxy or vote at the 2026 Meeting. Abstentions and broker non-votes will count towards the quorum requirement. If there is not a quorum, a majority of the votes present may adjourn the 2026 Meeting to another date. For Proposal 1, you may vote FOR or AGAINST each Board nominee, or you may ABSTAIN from voting. For each other matter to be voted on, you may vote FOR or AGAINST or ABSTAIN from voting.", + "chunk_count": 5, + "children": [ + { + "title": "Stockholder of Record", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Quorum and Voting/Stockholder of Record", + "level": 3, + "summary": "You are a stockholder of record if your shares were registered directly in your name with our transfer agent, Computershare, on April 27, 2026. You can vote shares, change your vote, or revoke your proxy before the final vote at the 2026 Meeting in any of the following ways: [tables/table-11 Proxy Voting.html] If you do not vote using any of the ways described above, your shares will not be voted.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Street Name Holder", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Quorum and Voting/Street Name Holder", + "level": 3, + "summary": "This passage explains voting procedures for beneficial owners whose shares are held in street name by a nominee like a bank or broker. It details that such owners can direct the nominee how to vote, and must obtain a valid proxy to vote virtually. For non-routine matters like director elections and executive compensation, the nominee cannot vote without instructions, resulting in a broker non-vote. However, the nominee has discretion to vote on routine matters like ratifying the accounting firm. For non-discretionary proposals, specific instructions are required. Shares held in street name can only be revoked per exchange rules.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Vote Count", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Quorum and Voting/Vote Count", + "level": 3, + "summary": "On each matter to be voted upon, stockholders have one vote for each share of NVIDIA common stock owned as of April 27, 2026. Votes will be counted by the inspector of election as follows: [tables/table-12 Proposal Voting.html] \\* Because this is a routine proposal, there are no broker non-votes. If you are a stockholder of record and you return a signed proxy card without marking any selections, your shares will be voted FOR each of the nominees listed in Proposal 1, FOR Proposals 2 and 3, and AGAINST Proposals 4, 5, 6, and 7. If any other matter is properly presented at the 2026 Meeting, one of the proxyholders will vote your shares using his best judgment.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Vote Results", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Quorum and Voting/Vote Results", + "level": 3, + "summary": "Preliminary voting results will be announced at the 2026 Meeting. Final voting results will be published in a current report on Form 8-K, which will be filed with the SEC by June 30, 2026.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Proxy Materials", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Proxy Materials", + "level": 2, + "summary": "NVIDIA provides proxy materials online, with notices sent to stockholders. Householding is used for multiple stockholders at the same address. Stockholders can change delivery preferences. Proxy solicitation costs are paid by NVIDIA, with MacKenzie Partners assisting. Deadlines for 2027 stockholder proposals under Rule 14a-8 are January 12, 2027.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Director Nominations Under Our Proxy Access Bylaw", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Director Nominations Under Our Proxy Access Bylaw", + "level": 2, + "summary": "A stockholder (or a group of up to 20 stockholders) who has owned at least 3% of the voting power of our outstanding capital stock for at least three continuous years and has complied with the other requirements in our Bylaws may nominate and include in our proxy materials director nominees constituting up to the greater of (a) up to two director candidates or (b) up to 20% of the number of directors in office on the last day that a submission may be delivered. Notice of a proxy access nomination for consideration at our 2027 Meeting must be received following the above instructions not later than the close of business on March 26, 2027, and not earlier than February 24, 2027. If we hold the 2027 Meeting more than 30 days prior to, or delayed by more than 30 days after, the first anniversary of the 2026 Meeting, for written notice by the stockholder to be timely, such notice must be delivered following the above instructions not earlier than the close of business on the 120th day prior to the 2027 Meeting and not later than the close of business on the 90th day prior to the 2027 Meeting or the 10th day following the day on which public announcement of the date of the 2027 Meeting is first made by us, whichever is later.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Director Nominations and Proposals", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Other Director Nominations and Proposals", + "level": 2, + "summary": "Apart from Rule 14a-8 and the proxy access provision of our Bylaws, under our Bylaws certain procedures must be followed for a stockholder to nominate a director or introduce an item of business at an annual meeting of stockholders. If you wish to nominate a director or introduce an item of business at the 2027 Meeting that is not included in our 2027 Meeting proxy materials, you must do so in writing following the above instructions not later than the close of business on March 26, 2027, and not earlier than February 24, 2027. If we hold the 2027 Meeting more than 30 days prior to, or delayed by more than 70 days after, the first anniversary of the 2026 Meeting, for written notice by the stockholder to be timely, such notice must be delivered following the above instructions not earlier than the close of business on the 120th day prior to the 2027 Meeting and not later than the close of business on the 90th day prior to the 2027 Meeting or the 10th day following the day on which public announcement of the date of the 2027 Meeting is first made by us, whichever is later.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Additional Requirements and Information", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Additional Requirements and Information", + "level": 2, + "summary": "The text discusses stockholder proposals and director nominations under the company's Bylaws, emphasizing the Board's willingness to engage with stockholders on amendments. The main voting item is the election of 10 director nominees, including 9 independent directors and CEO Jensen Huang, to serve until the 2027 annual meeting. Directors are elected by a majority of votes cast, with abstentions and broker non-votes having no effect. All nominees currently serve on the Board.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recommendation of the Board", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Recommendation of the Board", + "level": 2, + "summary": "The Board recommends that you vote FOR the election of each of the following nominees: [tables/table-13 NVIDIA Board.html] (1) Qualified as an AC financial expert. (2) Previously served as a member of our Board from 1993 until 2006.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Director Qualifications and Nomination of Directors", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Director Qualifications and Nomination of Directors", + "level": 2, + "summary": "The NCGC identifies and assesses director qualifications, considering factors like integrity, independence, leadership, industry knowledge, and diversity. It evaluates stockholder nominees equally and has proxy access rules. The Board balances tenure with new perspectives, with one-third of independent directors joining since 2020. Recent additions include Ms. Lora and Dr. Dabiri to the NCGC.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Our Director Nominees", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees", + "level": 2, + "summary": "The text provides biographies of two directors, Tench Coxe and John O. Dabiri, detailing their professional experience, qualifications, and skills that led the NCGC and Board to believe they should continue serving. Tench Coxe, former Managing Director of Sutter Hill Ventures, brings expertise in financial analysis, corporate strategy, and emerging technology trends. John O. Dabiri, a professor at Caltech, contributes a versatile research background in engineering and innovation. The biographies include information on their ages, tenure, committee memberships, education, and career highlights, along with associated skills matrices.", + "chunk_count": 10, + "children": [ + { + "title": "JEN-HSUN HUANG", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/JEN-HSUN HUANG", + "level": 3, + "summary": "Jen-Hsun Huang, age 63, is the founder, President, and CEO of NVIDIA since 1993. He pioneered accelerated computing and the GPU, driving PC gaming, computer graphics, and modern AI. He has received numerous honors including the Robert N. Noyce Award, IEEE Founder's Medal, and honorary doctorates. He was named world's best CEO by Fortune, Economist, and Brand Finance, and one of TIME's 100 most influential people. Prior to NVIDIA, he worked at LSI Logic and AMD. He holds a BSEE from Oregon State University and an MSEE from Stanford University. The passage also lists his expertise in senior leadership, industry/technical, financial, governance, emerging technologies, community, marketing, regulatory, and human capital management.", + "chunk_count": 1, + "children": [] + }, + { + "title": "DAWN HUDSON", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/DAWN HUDSON", + "level": 3, + "summary": "Dawn Hudson, age 68, is a former Chief Marketing Officer of the National Football League and former CEO of Pepsi-Cola North America. She has served as a director since 2013 and is an independent director and financial expert. Her career includes leadership roles at The Parthenon Group, PepsiCo, and major advertising agencies. She has served on numerous public company boards including P.F. Chang's, Allergan, Lowe's, Amplify Snack Brands, Modern Times Group, and Interpublic Group. She holds a BA in English from Dartmouth College. The passage also mentions Jen-Hsun Huang, founder and CEO of NVIDIA, and includes image references related to financial, marketing, governance, and human capital management experience.", + "chunk_count": 1, + "children": [] + }, + { + "title": "HARVEY C. JONES", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/HARVEY C. JONES", + "level": 3, + "summary": "Harvey C. Jones, age 73, is Managing Partner of Square Wave Ventures and has been a director since 1993. He co-founded Daisy Systems, led Synopsys as CEO, and co-founded Tensilica. He brings executive management, semiconductor technology, and financial expertise to the board.", + "chunk_count": 1, + "children": [] + }, + { + "title": "MELISSA B. LORA", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/MELISSA B. LORA", + "level": 3, + "summary": "Melissa B. Lora is a former President of Taco Bell International with a 31-year career at Taco Bell Corp. She serves on the board of Conagra Brands since 2019, chairing its nominating and corporate governance committee. Previously, she served on the boards of KB Home (2004-2024) and MGIC Investment Corporation (2018-2022). She holds a BS in Finance from California State University-Long Beach and an MBA from USC. Her expertise includes senior management, finance, and public company board governance.", + "chunk_count": 1, + "children": [] + }, + { + "title": "STEPHEN C. NEAL", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/STEPHEN C. NEAL", + "level": 3, + "summary": "Chairman Emeritus and Senior Counsel, Cooley LLP [tables/table-19 Director Profile.html] Stephen C. Neal serves as Chairman Emeritus and Senior Counsel of the law firm Cooley LLP, where he was also Chief Executive Officer from 2001 until 2008. In addition to his extensive experience as a trial lawyer on a broad range of corporate issues, Mr. Neal has represented and advised numerous boards of directors, special committees of boards, and individual directors on corporate governance and other legal matters. Prior to joining Cooley in 1995, Mr. Neal was a partner of the law firm Kirkland & Ellis LLP. Mr. Neal served on the board of directors of Levi Strauss & Co. from 2007 to 2021 and as Chairperson from 2011 to 2021. Mr. Neal also is Chairman of the Oversight Board Trust, a perpetual Delaware special purpose trust. Previously, Mr. Neal served as Chairman of the boards of the William and Flora Hewlett Foundation and of the Monterey Bay Aquarium. Mr. Neal holds an AB degree from Harvard University and a JD degree from Stanford Law School. Mr. Neal brings to the Board deep knowledge and broad experience in corporate governance as well as his perspectives drawn from advising many companies throughout his career. [tables/table-20 Leadership & Governance.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "A. BROOKE SEAWELL", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/A. BROOKE SEAWELL", + "level": 3, + "summary": "The passage profiles two board directors: A. Brooke Seawell, a Venture Partner at New Enterprise Associates with extensive experience in finance, public company governance, and technology; and Aarti Shah, former Senior Vice President at Eli Lilly, bringing expertise in drug development, IT, cybersecurity, and data sciences. Both have served on multiple corporate and non-profit boards.", + "chunk_count": 1, + "children": [] + }, + { + "title": "MARK A. STEVENS", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/MARK A. STEVENS", + "level": 3, + "summary": "Managing Partner, S-Cubed Capital Age: 66 Director Since: 2008 (previously served 1993-2006) Committees: CC, NCGC", + "chunk_count": 3, + "children": [ + { + "title": "Other Current Public Company Boards: None", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/MARK A. STEVENS/Other Current Public Company Boards: None", + "level": 4, + "summary": "", + "chunk_count": 1, + "children": [] + }, + { + "title": "Independent Director", + "path": "2026-Annual-Report-Web.pdf/Information About the 2026 Meeting/Our Director Nominees/MARK A. STEVENS/Independent Director", + "level": 4, + "summary": "Mark A. Stevens is Managing Partner of S-Cubed Capital since 2012, formerly a Managing Partner at Sequoia Capital from 1993 to 2011. He held technical roles at Intel and Hughes Aircraft. He is a trustee and investment committee chair at USC, and was a director of Quantenna Communications. He holds degrees from USC and Harvard. His expertise includes technology industry understanding, corporate strategy, acquisitions, and financial community experience. He serves on committees including CC and NCGC, and is recognized for contributions in Industry & Technical, Financial/Financial Community, Governance & Public Company Board, and Emerging Technologies & Business Models.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + } + ] + }, + { + "title": "Information About the Board of Directors and Corporate Governance", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance", + "level": 1, + "summary": "", + "chunk_count": 39, + "children": [ + { + "title": "Independence of the Members of the Board of Directors", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Independence of the Members of the Board of Directors", + "level": 2, + "summary": "Nasdaq rules and our Corporate Governance Policies (as further described below) require that a majority of our directors not have a relationship that would interfere with their exercise of independent judgment in carrying out their responsibilities and that they meet any other qualification requirements required by the SEC and Nasdaq. After considering all relevant relationships and transactions, our Board determined that, except for Mr. Huang, all of our directors who currently serve on the Board are, and all of our directors who served on the Board during Fiscal 2026 were, “independent” as defined by Nasdaq’s rules and regulations. The Board also determined that all members of our AC, CC, and NCGC are independent under applicable Nasdaq listing standards, and that each of Mr. Seawell, Mr. Jones and Ms. Lora of the AC are “audit committee financial experts” as defined under applicable SEC rules.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Board Leadership Structure", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Board Leadership Structure", + "level": 2, + "summary": "NVIDIA's Board uses an independent Lead Director instead of a chairperson to ensure equal voice and independent oversight. The leadership structure is reviewed annually, with the Lead Director facilitating communication between management and the Board, leading executive sessions, and supporting governance. The CEO handles operational leadership. The Board may appoint a chairperson in the future but would maintain an independent Lead Director to preserve balance and independence.", + "chunk_count": 3, + "children": [ + { + "title": "Our Lead Director", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Board Leadership Structure/Our Lead Director", + "level": 3, + "summary": "Stephen C. Neal Director since 2019 Chairperson of the NCGC Experience √ Extensive experience as a trial attorney √ Has advised numerous companies, boards and individuals on governance and legal matters √ Helped clients manage internal and government investigations √ Executive experience as Cooley LLP's CEO, and board and chairman experience on the Levi Strauss & Co. board of directors The Board believes Mr. Neal's experience, breadth of knowledge, and contributions to the Board position him well to provide strong leadership, oversight, and the ability to contribute valuable insight with respect to the Company's business. Our Lead Director may require Board consideration of risk matters, including adding them to Board agendas or as topics for executive sessions of the independent members of the Board. The Board believes that Mr. Neal is highly qualified to assist the Board as Lead Director in effectively overseeing the identification, assessment, and management of the Company's exposure to various risks as a result of his extensive risk management, legal, and executive experience. Further information on the Board's oversight of risk management and its work with the CEO to address risk management matters is detailed below under Role of the Board in Risk Oversight. Our Lead Director has significant responsibilities, which are detailed in our Corporate Governance Policies, and include the duties listed below.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Duties of Our Lead Director", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Board Leadership Structure/Duties of Our Lead Director", + "level": 3, + "summary": "The Lead Director determines the Board meeting schedule, prepares agendas with the CEO, advises on information quality for independent directors, coordinates executive sessions, presides over meetings in the CEO's absence, convenes independent director meetings, and performs other duties as determined. The Lead Director also represents the Board in communications with stockholders and participates in annual stockholder outreach meetings.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Committees of the Board of Directors", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Committees of the Board of Directors", + "level": 2, + "summary": "The Board has three committees: an AC, a CC, and a NCGC. Each of these committees operates under a written charter, which may be viewed in the Investor Relations section of our website at www.nvidia.com. Committee assignments are determined based on background and the expertise that individual directors can bring to a committee. Our Board believes regular committee rotations are a good corporate governance practice and introduce a variety of perspectives and ideas, more fully inform its members regarding the full scope of the Board and our activities, and benefit each committee and the Board as a whole. Since our 2025 Meeting, Ms. Lora and Dr. Dabiri have joined the NCGC. The composition and functions of our committees are set forth below.", + "chunk_count": 4, + "children": [ + { + "title": "AC Members as of our 2026 Meeting", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Committees of the Board of Directors/AC Members as of our 2026 Meeting", + "level": 3, + "summary": "The Audit Committee (AC) as of the 2026 meeting includes members A. Brooke Seawell (Chairperson), Tench Coxe, Harvey C. Jones, Melissa B. Lora, and Aarti Shah. In Fiscal 2026, the AC met four times, covering topics such as capitalization review, tax, treasury, internal audit, information security, enterprise risk management, and insurance. The committee oversees corporate accounting and financial reporting, the internal audit function, and the independent registered public accounting firm's engagement, compensation, retention, and termination. It evaluates the accounting firm's performance, approves audit and non-audit services, and confers with management on audit results, quarterly financial statements, and internal control effectiveness. The AC reviews financial statements for SEC filings, earnings press releases, and financial outlooks. It maintains policies on auditor employment, prepares the required SEC report, and establishes procedures for handling accounting and auditing complaints. Additionally, it oversees risks related to financial reporting, internal audit, regulatory compliance, and accounting policies, and reviews the company's information security policies and internal controls.", + "chunk_count": 1, + "children": [] + }, + { + "title": "CC Members as of our 2026 Meeting", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Committees of the Board of Directors/CC Members as of our 2026 Meeting", + "level": 3, + "summary": "Members as of our 2026 Meeting • Dawn Hudson (Chairperson) - Tench Coxe • John O. Dabiri - Aarti Shah • Mark A. Stevens In Fiscal 2026, the CC met three times. Selected highlights from its agenda topics included: executive, employee, and director compensation, review of employee support programs, executive protection, and human capital management. Committee Role and Responsibilities • Reviews and approves our overall compensation strategy and policies; • Reviews and recommends to the Board the compensation of our Board members; • Reviews and approves the compensation of Mr. Huang and other executive officers; - Reviews and approves corporate performance goals and objectives related to the compensation of our executive officers and other senior management; - Reviews and approves our CD&A disclosure for inclusion in the proxy statement and annual report on Form 10-K; - Prepares their report on executive compensation as required to be included by SEC rules in our annual proxy statement or annual report on Form 10-K; • Administers our stock purchase plans, variable compensation plans, and other similar programs; • Oversees our human capital management; • Assesses and monitors whether our compensation policies and programs could create material risks; and • Oversees risks related to compensation plans, programs and policies", + "chunk_count": 1, + "children": [] + }, + { + "title": "NCGC Members as of our 2026 Meeting", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Committees of the Board of Directors/NCGC Members as of our 2026 Meeting", + "level": 3, + "summary": "Members as of our 2026 Meeting • Stephen C. Neal (Chairperson) • John O. Dabiri • Harvey C. Jones - Melissa B. Lora • Mark A. Stevens In Fiscal 2026, the NCGC met three times. Selected highlights from its agenda topics included: Board recruiting, governmental relations, trade compliance and regulatory matters, stockholder proposals, corporate governance matters, and stockholder outreach. Committee Role and Responsibilities • Identifies, reviews, and evaluates candidates to serve as directors; • Recommends candidates for election to our Board; • Makes recommendations to the Board regarding committee membership and chairpersons; • Assesses the performance of the Board and its committees; • Reviews and assesses our corporate governance principles and practices; • Monitors changes in corporate governance practices, rules, and regulations; - Approves related party transactions; - Reviews and assesses our CS strategy, risks, and opportunities periodically, including related programs and initiatives; • Oversees and reviews policies and practices on governmental relations, trade compliance, regulatory matters, and related risks; - Establishes procedures for receiving, retaining, and addressing complaints we receive regarding violations of our Code of Conduct; • Monitors the effectiveness of our anonymous tip process; and - Oversees the Company's policies, practices, and investigation procedures in connection with the Company's compliance program", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Compensation Committee Interlocks and Insider Participation", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Compensation Committee Interlocks and Insider Participation", + "level": 2, + "summary": "At the beginning of Fiscal 2026, the CC initially consisted of Messrs. Burgess and Coxe, Ms. Hudson, and Drs. Dabiri and Shah. After the 2025 Meeting, the CC consisted of Messrs. Coxe and Stevens, Ms. Hudson, and Drs. Dabiri, Drell and Shah, until Dr. Drell resigned from the Board in January 2026. No member of the CC is an officer or employee of NVIDIA, and none of our executive officers serve as a member of the board or compensation committee of any entity that has one or more executive officers serving as a member of our Board or CC. Other than Dr. Shah, no member of the CC had a relationship requiring disclosure in Review of Transactions with Related Persons below.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Role of the Board in Risk Oversight", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Role of the Board in Risk Oversight", + "level": 2, + "summary": "The Board oversees risks and delegates to committees. Management identifies and mitigates risks, reporting to the Board. Internal Audit provides independent assurance. The Board retains oversight of strategic and information security risks, with the Audit Committee reviewing cybersecurity. The ERM process is an annual assessment integrating risk evaluation with operations, examining internal/external factors, and prioritizing mitigation. It is led by management, reviewed by the Board and Audit Committee, and aligns with disclosure controls. The CEO and senior management review the ERM action plan, with internal audit administering the program.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Corporate Governance", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance", + "level": 2, + "summary": "", + "chunk_count": 14, + "children": [ + { + "title": "Board Refreshment", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Board Refreshment", + "level": 3, + "summary": "As discussed above, the NCGC and the Board understand the importance of Board refreshment. On May 7, 2026, the Board approved an increase in its size from ten to eleven directors and appointed Suzanne Nora Johnson to fill the newly created directorship, in each case effective July 13, 2026. Ms. Nora Johnson will join the Board effective July 13, 2026, due to a prior professional commitment. Ms. Nora Johnson is the retired Vice Chairman of The Goldman Sachs Group, Inc., where she spent 21 years in various leadership roles. She currently serves on the board of directors of Pfizer Inc., and serves on the Boards of Trustees of The Brookings Institution, the Broad Foundation, and the University of Southern California. Ms. Nora Johnson will bring to the Board extensive senior leadership experience, experience managing large, complex, global institutions and deep knowledge of global financial markets, risk management, and regulatory issues. Ms. Nora Johnson is not currently a member of the Board and is not a nominee for election at the 2026 Meeting. She is expected to join the AC on the effective date of her appointment.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Corporate Governance Policies", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Corporate Governance Policies", + "level": 3, + "summary": "The Board has adopted Corporate Governance Policies to ensure that the Board has the necessary authority and processes in place to review and evaluate our business operations as needed and to make decisions that are independent of our management. These policies include practices the Board follows with respect to its composition and selection, regular evaluations of the Board and its committees, Board meetings and involvement of senior management, senior management development, and Board committees and compensation. These policies may be viewed in the Investor Relations section of our website at www.nvidia.com.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Executive Sessions of the Board", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Executive Sessions of the Board", + "level": 3, + "summary": "As required under Nasdaq's listing standards, our independent directors meet regularly in scheduled executive sessions at which only independent directors are present, as well as in sessions with the CEO. In Fiscal 2026, our independent directors met in both types of executive sessions at all four of our scheduled quarterly Board meetings.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Director Attendance at Annual Meeting", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Director Attendance at Annual Meeting", + "level": 3, + "summary": "We expect that our directors will attend each annual meeting, absent a valid reason. All Board members as of our 2025 Meeting attended our 2025 Meeting.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Board Self-Assessments", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Board Self-Assessments", + "level": 3, + "summary": "The NCGC oversees an evaluation process, conducted at least annually, whereby outside counsel for NVIDIA interviews each director to obtain his or her evaluation of the Board as a whole, and of the committees on which he or she serves. The interviews solicit ideas from the directors about, among other things, improving the quality of Board and/or committee oversight, effectiveness regarding strategic direction, financial and audit matters, executive compensation, acquisition activity, and other key topics. The interviews also focus on Board process and identifying specific issues for future discussion. After these evaluations are complete, our outside counsel summarizes the results, reviews them with our Lead Director, and then submits the summary for NCGC discussion. The NCGC Chair and outside counsel then report evaluation results to the full Board. In response to the evaluations conducted in Fiscal 2026, our Board determined to focus on geopolitical and regulatory risks, government affairs, supply chain, cybersecurity, AI regulations, growth planning, the use and applications of AI, the Company's strategic roadmap, and management development. The Board also determined to continue to focus on the Board's composition and process for Board refreshment.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Director Orientation and Continuing Education", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Director Orientation and Continuing Education", + "level": 3, + "summary": "The NCGC and our General Counsel are responsible for new director orientation and for administering or approving eligible director continuing education programs. Continuing education programs for directors may include a combination of internally developed materials and presentations, programs presented by third parties, and financial and administrative support for attendance at qualifying academic or other independent programs.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Director Time Commitment and Outside Board Memberships", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Director Time Commitment and Outside Board Memberships", + "level": 3, + "summary": "Our directors are expected to devote sufficient time to Board and committee duties and to understanding the Company's business. The NCGC reviews the other commitments of potential Board candidates, and does so annually for existing Board members, to determine if this expectation can be met. In making this determination, the NCGC considers, among other factors, stakeholder guidelines regarding numerical limits on public company boards on which a director may sit. None of our directors serve on more than two public company boards (including NVIDIA), and none of our non-employee directors serve as a CEO or executive officer of a public company.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Director Stock Ownership Guidelines", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Director Stock Ownership Guidelines", + "level": 3, + "summary": "Our Corporate Governance Policies require each non-employee director to hold shares of our common stock with a total value equal to six times the annual cash retainer for Board service during the period in which he or she serves as a director (or ten times his base salary, in the case of the CEO). The shares may include vested deferred stock, shares held in trust, and shares held by immediate family members, but not unvested equity awards. Non-employee directors have five years after their Board appointment to reach the ownership threshold. Our stock ownership guidelines are intended to further align director interests with stockholder interests. Each non-employee director and Mr. Huang currently meets or exceeds the stock ownership requirements.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Senior Management Development", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Senior Management Development", + "level": 3, + "summary": "We believe in extensive leadership development. Our internal leadership structure ensures that many senior managers work directly with our CEO to execute our corporate strategies. This approach results in strategic alignment, exposes leaders to a broad spectrum of corporate activities, and presents the Board with a pool of excellent candidates for future promotion. Our CEO also selects senior leaders to engage directly with the Board on key initiatives and provides periodic updates to our Board on management development.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Outside Advisors", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Outside Advisors", + "level": 3, + "summary": "The Board and each of its principal committees may retain outside advisors and consultants of their choosing at our expense. The Board does not need management's consent to retain outside advisors and the principal committees do not need consent from the Board or management' to retain outside advisors.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Code of Conduct", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Code of Conduct", + "level": 3, + "summary": "Our directors, executives, and employees are expected to conduct themselves with the highest degree of integrity, ethics, and honesty. Our credibility and reputation depend upon their good judgment, ethical standards, and personal integrity. Our Code of Conduct applies to all executive officers, directors, and employees, including our principal executive officer, principal financial officer, and principal accounting officer. The Financial Team Code of Conduct applies to our executive officers, directors, members of our finance department, and all employees involved in the preparation and review of externally-reported periodic financial reports, filings, and documents. We regularly review our Code of Conduct and related policies to ensure that they provide clear guidance to our directors, executives, and employees. We also regularly train our employees on our Code of Conduct and other policies. The Code of Conduct and the Financial Team Code of Conduct may be viewed in the Investor Relations section of our website at www.nvidia.com. If we make any amendments to either code, or grant any waiver from a provision of either code to any executive officer or director, we will promptly disclose the nature of the amendment or waiver on our website or in a report on Form 8-K. Information contained on our website is not incorporated by reference into this or any other report we file with the SEC.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Insider Trading Policy", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Insider Trading Policy", + "level": 3, + "summary": "Our Insider Trading Policy governs transactions in our securities and applies to the Board and our employees, contractors, and consultants. We believe our Insider Trading Policy is reasonably designed to promote compliance with insider trading laws, rules and regulations applicable to NVIDIA. Section 16 directors and officers must obtain preclearance before trading in NVIDIA's stock. In addition, we comply with applicable laws and regulations relating to insider trading with respect to transactions in our securities such as open market repurchases of our common stock from time to time. Under our Insider Trading Policy, hedging ownership of NVIDIA stock, including but not limited to trading in options, puts, calls, or other derivative instruments related to NVIDIA stock or debt, is not permitted. Additional prohibitions include purchasing NVIDIA stock on margin or holding NVIDIA stock in a margin account, or pledging NVIDIA stock as collateral for a loan. We allow participation in certain exchange funds for portfolio diversification.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Corporate Hotline", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Governance/Corporate Hotline", + "level": 3, + "summary": "We have established an independent corporate hotline to allow any employee, contractor, customer, or partner to confidentially and anonymously submit a complaint about any accounting, internal controls, auditing, Code of Conduct, or other matter of concern (unless prohibited by local privacy laws).", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Stockholder Communications with the Board of Directors", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Stockholder Communications with the Board of Directors", + "level": 2, + "summary": "Stockholders who wish to communicate with the Board regarding nominations of directors or other matters may do so by sending electronic written communications addressed to Timothy S. Teter, our Secretary, at shareholdermeeting@nvidia.com. All stockholder communications we receive that are addressed to the Board will be compiled by our Secretary. If no particular director is named, letters will be forwarded, depending on the subject matter, to the chairperson of the AC, CC, or NCGC. Matters put forth by our stockholders will be reviewed by the NCGC, which will determine whether these matters should be presented to the Board. The NCGC will give serious consideration to all such matters and will make its determination in accordance with its charter and applicable laws.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Majority Vote Standard", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Majority Vote Standard", + "level": 2, + "summary": "The text describes director election procedures under company Bylaws. In uncontested elections, directors are elected by majority of votes cast (FOR votes must exceed AGAINST votes). If an incumbent fails to achieve majority, they must offer resignation, which the Board's committee reviews and decides within 90 days. In contested elections, directors are elected by plurality of votes cast. Abstentions and broker non-votes count for quorum but have no effect on election outcomes.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stockholder Special Meeting Right", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Stockholder Special Meeting Right", + "level": 2, + "summary": "As part of our Board and management's comprehensive review of current corporate governance practices, our Board amended our Bylaws in 2024 to let stockholders owning at least $15\\%$ of our shares, and who have owned such shares continuously for at least one year, to request a special meeting of stockholders. Those requestors must also satisfy the disclosure, timing and other requirements set forth in our Bylaws intended to ensure that stockholders receive adequate, timely, and accurate information in connection with a special meeting. Our Board believes that this special meeting right strikes a balance between enabling urgent action and protecting the Company and its stockholders from narrow, short-term interests.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Board Meeting Information", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Board Meeting Information", + "level": 2, + "summary": "The Board met five times during Fiscal 2026, including meetings during which the Board discussed the strategic direction of NVIDIA, new business and strategic opportunities, the product roadmap, and other matters facing NVIDIA. We expect each Board member to attend each meeting of the Board and the committees on which he or she serves. Each Board member attended 75% or more of the applicable meetings of the Board and of each committee on which he or she served during Fiscal 2026.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Corporate Sustainability", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Corporate Sustainability", + "level": 2, + "summary": "NVIDIA integrates sound computer science principles and practices across the company, with environmental stewardship, social responsibility, and solid governance central to its business strategy. The full Board has ultimate responsibility for corporate sustainability (CS) matters, with the Nominating, Governance and Corporate Citizenship Committee (NCGC) overseeing sustainability strategy, risks, and opportunities. The CS team updates the NCGC at least semiannually and reports to the full Board annually. The Corporate Sustainability Steering Committee (CSSC), composed of executive leadership, provides input on sustainability strategy. In Fiscal 2025, NVIDIA matched 100% of global electricity usage for sites under operational control with renewable energy. In Fiscal 2026, it announced Science Based Targets initiative-validated goals: reduce absolute scope 1 and 2 emissions by 50% by Fiscal 2030 (from a Fiscal 2023 base year), and reduce scope 3 emissions intensity from use of sold GPU products by 75% per petaFLOP of computing performance by Fiscal 2030.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Public Policy Engagement and Accountability", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Public Policy Engagement and Accountability", + "level": 2, + "summary": "Our NCGC oversees and periodically reviews our public policy engagement and accountability. We engage in the public policy process to advance the long-term interests of the Company and its stockholders. NVIDIA's U.S. federal lobbying activities are disclosed quarterly and publicly reported. We belong to trade associations worldwide. Management reports to the NCGC about our policies and practices in connection with government relations, public policy, and related expenditures.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Director Compensation", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Director Compensation", + "level": 2, + "summary": "NVIDIA's non-employee director compensation program is designed to attract and retain a world-class Board of Directors while balancing our stockholders' interests, and aligns with annual service starting on the dates of our annual meetings. We do not pay additional fees for serving as Lead Director or as chairperson or member of our committees, or for meeting attendance. Directors who are also employees do not receive compensation for service on the Board.", + "chunk_count": 5, + "children": [ + { + "title": "Roles of the Board, the Compensation Committee, and Compensation Consultant", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Director Compensation/Roles of the Board, the Compensation Committee, and Compensation Consultant", + "level": 3, + "summary": "The CC reviews our director compensation annually with the assistance of its independent compensation consultant, Semler Brossy, and assesses peer group data, compensation trends, and best practices. Following this review, the CC recommended the non-employee director compensation for the year following our 2025 Meeting, or the 2025 Program, as described below, which the Board approved in December 2024. 2025 Program The 2025 Program maintained the same target compensation as the previous year with a total value of \\$340,000—slightly below the median paid to peer group non-employee directors (based on the peer group determined in late 2024): \\$85,000 Annual Cash Retainer Paid Quarterly \\$255,000 Equity Retainer RSUs (1) (1) Represents target annual value of RSUs granted on the day following our 2025 Meeting, or the 2025 Program RSUs. 50% of the 2025 Program RSUs vested on the third Wednesday in November 2025 and 50% will vest on the third Wednesday in May 2026, subject to the director's continuous service. The number of shares subject to each director's 2025 Program RSUs equaled the target value of the grant divided by the 30-calendar day trailing average closing price of our common stock that ended the business day before the 2025 Meeting, to smooth the effects of market volatility.", + "chunk_count": 1, + "children": [] + }, + { + "title": "New Directors", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Director Compensation/New Directors", + "level": 3, + "summary": "Any new non-employee director who is appointed or elected to the Board with an effective date after the 2026 Meeting, or a New Director, will be entitled to receive (i) an annual cash retainer of \\$85,000, prorated to reflect the New Director's service, paid quarterly, (ii) an initial grant of RSUs with a target value of \\$255,000, converted to a number of shares by dividing such value by the 30-calendar day trailing average closing price of our common stock ending the business day before the New Director's effective date of appointment or election, 1/6th of which will vest approximately every six months following the grant date, and (iii) a second grant of RSUs with a target value of \\$255,000, converted to a number of shares by dividing such value by the 30-calendar day trailing average closing price of our common stock ending the business day before the 2026 Meeting, prorated to reflect the New Director's service until the 2027 Meeting, or the New Director Annual Grant. 50% of the New Director Annual Grant will vest on the third Wednesday in May 2027, and the balance will vest on the third Wednesday in November 2026; provided that if the New Director's first day of service is on or after November 1, 2026, the New Director Annual Grant will vest in full on the third Wednesday in May 2027 ((ii) and (iii) collectively referred to as the New Director RSUs).", + "chunk_count": 1, + "children": [] + }, + { + "title": "Deferred Compensation Program", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Director Compensation/Deferred Compensation Program", + "level": 3, + "summary": "Non-employee directors can elect to defer the settlement of RSUs upon vesting for tax planning purposes. Deferrals can be made to the earlier of (i) the third Wednesday of March of a future year (no sooner than 2027 for the 2025 Program RSUs or 2028 for the New Director RSUs), or the 15th of March of any year that is 2029 or later, or (ii) the director's cessation of service or certain change-in-control events, in accordance with Section 409A of the Internal Revenue Code.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Compensation∕Benefits", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Director Compensation/Other Compensation∕Benefits", + "level": 3, + "summary": "The passage details director compensation policies, including expense reimbursement, change-in-control benefits, and vesting acceleration upon death. It presents a table of director compensation for fiscal 2026, showing fees earned and stock awards (RSUs) with grant date fair values. Notes explain that amounts reflect ASC 718 fair values, not actual receipts, and mention specific events like Mr. Burgess's death and Dr. Ochoa's resignation affecting payments. A second table lists unvested RSUs held by non-employee directors as of January 25, 2026, with notes on vesting acceleration due to death or resignation.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Review of Transactions with Related Persons", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Review of Transactions with Related Persons", + "level": 2, + "summary": "The passage outlines NVIDIA's policy on conflicts of interest, requiring employees, officers, and directors to avoid activities conflicting with company interests. It notes that related party transactions are reviewed, and those involving executive officers or directors must be approved by the NCGC. Since Fiscal 2026, no reportable transactions occurred except for the employment of family members of CEO Jen-Hsun Huang and director Dr. Shah. Their compensation was determined per standard practices without involvement of the related executives, totaling approximately $1,232,000 for Huang's daughter, $1,320,000 for his son, and $265,000 for Shah's son. The Huang Foundation has an agreement with CoreWeave for GPU compute time donations, with $108.3 million donated, and NVIDIA plans to offer free engineering services to grant recipients. Indemnity agreements exist for executives and directors, and equity grants were made to non-employee directors and executive officers.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Security Ownership of Certain Beneficial Owners and Management", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Security Ownership of Certain Beneficial Owners and Management", + "level": 2, + "summary": "The table details beneficial ownership of common stock as of March 23, 2026, for NEOs, directors, all directors and executive officers as a group, and known beneficial owners of over 5%. Ownership is determined per SEC rules, including shares with voting/investment power and PSUs/RSUs vesting within 60 days. Percentages are based on 24,312,141,810 shares outstanding. Footnotes provide specifics for each individual, including shares held in trusts, LLCs, and foundations, with details on voting and investment power. Principal stockholders BlackRock and Vanguard are listed based on Schedule 13G filings, with BlackRock having sole dispositive power over 1,805,935,550 shares and Vanguard over 1,777,408,252 shares.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Proposal 2—Advisory Approval of Executive Compensation", + "path": "2026-Annual-Report-Web.pdf/Information About the Board of Directors and Corporate Governance/Proposal 2—Advisory Approval of Executive Compensation", + "level": 2, + "summary": "This is a non-binding advisory vote, known as 'say-on-pay,' to approve Fiscal 2026 compensation for named executive officers (NEOs). Approval requires a majority of shares present and voting; abstentions count as votes against, while broker non-votes have no effect. The board recommends approval, and while the vote is non-binding, it will be considered in future compensation decisions.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Executive Compensation", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation", + "level": 1, + "summary": "", + "chunk_count": 66, + "children": [ + { + "title": "Compensation Discussion and Analysis", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis", + "level": 2, + "summary": "This CD&A describes our Fiscal 2026 executive compensation philosophy, design, and process, and how our corporate results affected the payout of performance-based awards. Our Fiscal 2026 NEOs were: Jen-Hsun Huang, President and CEO; Colette M. Kress, EVP and CFO; Ajay K. Puri, EVP, Worldwide Field Operations; Debora Shoquist, EVP, Operations; and Timothy S. Teter, EVP, General Counsel and Secretary.", + "chunk_count": 28, + "children": [ + { + "title": "Fiscal 2026 Executive Compensation Summary", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Executive Compensation Summary", + "level": 3, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Continued Focus on Pay for Performance", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Executive Compensation Summary/Continued Focus on Pay for Performance", + "level": 4, + "summary": "NVIDIA's executive compensation program in Fiscal 2026 continued to be guided by a pay for performance philosophy to link competitive NEO pay with our stockholders' interests. $96\\%$ of our CEO's and $48\\%$ of our other NEOs' Fiscal 2026 total target pay was dependent on corporate performance.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Executive Pay Heavily Weighted Towards Equity Awards", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Executive Compensation Summary/Executive Pay Heavily Weighted Towards Equity Awards", + "level": 4, + "summary": "The vast majority of our NEOs' total target pay for Fiscal 2026 was comprised of equity awards: • SY PSUs that can be earned based on annual Non-GAAP Operating Income performance, vesting over 4 years, • MY PSUs that can be earned based on 3-year TSR performance relative to the S&P 500, vesting over 3 years, and • RSUs vesting over 4 years (for NEOs other than our CEO) CEO Fiscal 2026 Target Pay Mix (1) The vast majority of our NEOs' total target pay for Fiscal 2026 was comprised of equity awards:
0 [images/image-73-Compensation Structure Breakdown.jpg]
stacked bar chart [tables/table-28 Compensation Breakdown.html]
Other NEO Fiscal 2026 Target Pay Mix (1)(2)
0 [images/image-74-Compensation Structure Breakdown.jpg]
stacked bar chart [tables/table-29 Compensation Mix.html]
(1) Based on total target pay as approved by the CC, consisting of annual base salary, and, assuming the Company achieves associated performance goals at a Base Compensation Plan level, target payout opportunity under our Variable Cash Plan and target equity opportunities the CC intended to deliver. (2) Reflects the average total target pay mix for our NEOs other than our CEO. NEOs were also eligible for variable cash awards based on annual revenue, in addition to base salary and other compensation and benefits.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Key NEO Compensation Actions for Fiscal 2026", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Key NEO Compensation Actions for Fiscal 2026", + "level": 3, + "summary": "Cash compensation for NEOs remained constant, while target equity opportunities increased to reflect role complexity and align total pay. Performance goals were raised, with revenue and Non-GAAP Operating Income targets set at record or stretch levels. Due to record performance exceeding stretch goals, NEOs earned maximum payouts for Fiscal 2026 Variable Cash Plan, SY PSUs, and MY PSUs.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Our Compensation Philosophy and Practices", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Our Compensation Philosophy and Practices", + "level": 3, + "summary": "NVIDIA's compensation program for NEOs is guided by four philosophies: pay for performance using at-risk cash and equity based on corporate metrics, competitive pay to attract and retain talent, alignment with stockholders' long-term interests, and simplicity with objective metrics. Total target pay includes base salary, target variable cash opportunity, and target equity opportunity. The program emphasizes at-risk compensation and stockholder alignment.", + "chunk_count": 3, + "children": [ + { + "title": "What We Do", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Our Compensation Philosophy and Practices/What We Do", + "level": 4, + "summary": "√ Emphasize at-risk, performance-based compensation, with simple, objective performance goals √ Use annual and multi-year performance targets to determine PSU awards earned √ Set rigorous performance goals and use different metrics for annual and multi-year awards √ Require NEOs to provide continuous service for 4 years to fully vest in RSU awards and SY PSU awards earned based on performance √ Evaluate our program annually based on feedback from stockholder engagement efforts and make adjustments when appropriate √ Mitigate compensation risks √ Cap performance-based variable cash and PSU payouts √ Retain an independent compensation consultant reporting directly to the independent CC √ Require NEOs to maintain meaningful stock ownership √ Maintain a clawback policy for performance-based compensation", + "chunk_count": 1, + "children": [] + }, + { + "title": "What We Don't Do", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Our Compensation Philosophy and Practices/What We Don't Do", + "level": 4, + "summary": "The document outlines NVIDIA's executive compensation policies, including restrictions on supplemental retirement benefits, tax gross-ups, and hedging/pledging of stock. It also details the Compensation Committee's multi-year process for the Fiscal 2026 program, from peer company determination in December 2024 to certification of payouts and risk assessment by May 2026.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Roles of the Compensation Committee, Compensation Consultant, and Management", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Roles of the Compensation Committee, Compensation Consultant, and Management", + "level": 3, + "summary": "The Compensation Committee (CC) oversees NEO compensation, utilizing independent consultant Semler Brossy for advice on the Fiscal 2026 program. The CC confirmed no conflicts of interest after reviewing Semler Brossy's services, revenue percentage, policies, and relationships. The CC approved a peer group recommended by Semler Brossy and management, considered CEO peer data and Radford Survey data for other NEOs, and made final decisions on compensation, including performance goal certification and payout approvals for Variable Cash Plan and PSUs. The CC also reviewed management's compensation risk analysis.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Peer Companies and Market Compensation Data", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Peer Companies and Market Compensation Data", + "level": 3, + "summary": "NVIDIA's Compensation Committee updated its Fiscal 2026 peer group in December 2024, removing Adobe, AMD, and SAP due to size and relevance, and adding Alphabet, Amazon, Apple, and Microsoft as talent competitors. The peer group is selected based on competition for executive talent, business complexity, and size (revenue or market cap at 1/3 to 3.5x NVIDIA's). The committee compared NVIDIA's trailing 12-month revenue and market capitalization to peer percentiles. For NEOs other than the CEO, Mr. Huang reviewed market data from Radford Survey and peer proxies before recommending compensation. The CC reviewed CEO compensation data and set total compensation for all NEOs based on peer percentiles and other factors.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Factors Used in Determining Executive Compensation", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Factors Used in Determining Executive Compensation", + "level": 3, + "summary": "In addition to peer data, our CC considers the following factors in making executive compensation decisions. The weight given to each factor may differ among NEOs and each component of pay, and is subject to the CC's sole discretion. √ The need to attract and retain talent in a highly competitive industry √ Stockholder feedback regarding our executive pay √ The desire for simplicity of the overall program and transparency of the performance metrics √ An NEO's past performance and anticipated future contributions √ Our financial performance and forecasted results, as well as our prior financial performance and resulting impact on our executives' compensation √ Changes in the scale and complexity of our business √ Each NEO’s current total compensation √ The scope and complexity of the department(s) or function(s) the NEO manages √ Each NEO's unvested equity √ Internal pay equity relative to similarly situated executives √ Our CEO's recommendations for the NEOs other than himself, including his understanding of each NEO's performance, capabilities, and contributions √ Our CC's independent judgment √ Our philosophy that an NEO's total compensation opportunity and percentage of at-risk pay should increase with responsibility √ The total compensation cost and the growth in such cost, including from executive compensation, to maintain a responsible cost structure for our compensation programs \\*", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fiscal 2026 Compensation Actions and Achievements", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements", + "level": 3, + "summary": "", + "chunk_count": 10, + "children": [ + { + "title": "Stockholder Outreach and Feedback", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Stockholder Outreach and Feedback", + "level": 4, + "summary": "NVIDIA conducted an annual outreach program in Fall 2024, contacting top institutional stockholders representing 31% of shares. Management and the Board met with holders of 30% of shares to discuss executive compensation, receiving positive feedback on pay-for-performance linkage. With a 92% say-on-pay approval rate, the Compensation Committee maintained similar elements for Fiscal 2026 but increased target equity opportunities for all NEOs to reflect role complexity, aligned total target pay for non-CEO NEOs based on contributions, and increased rigor by referencing record Fiscal 2025 results and stretch goals. Performance-based pay remains tied solely to NVIDIA's financial performance to align management incentives with stockholder interests.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Total Target Compensation Approach", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Total Target Compensation Approach", + "level": 4, + "summary": "In setting Fiscal 2026 compensation, our CC reviewed each NEO's total target pay opportunity and distribution across multiple pay elements. Our CC compared Mr. Huang's base salary, target variable cash opportunity, target total cash opportunity, target equity opportunity, and total target pay opportunity against chief executives at peer companies. For our other NEOs, Mr. Huang reviewed their total target pay against similarly situated executives at peer companies where the data was available from the Radford Survey, considering internal pay equity, individual performance, unvested equity levels, and the increasing complexity of their roles. These factors informed Mr. Huang's recommendations to the CC. The CC also considered the factors outlined in Factors Used in Determining Executive Compensation and its compensation objectives for Fiscal 2026. Rather than applying a formula or assigning specific weights to each factor, our CC used its judgment and experience to set total target compensation, the mix of cash and equity, and fixed and at-risk pay opportunities for each NEO. When setting pay elements, the CC evaluated them in the context of the levels of the other pay elements and total target pay to ensure alignment with program objectives. The CC established amounts and designed a structure to reward NEOs with above-market value from equity awards and variable cash incentives only upon exceptional corporate performance.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Components of Pay", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Components of Pay", + "level": 4, + "summary": "The primary components of NVIDIA's Fiscal 2026 executive compensation program are summarized below: [tables/table-34 NEO Compensation.html] (1) Our CC considers RSUs to be at-risk pay because the realized value depends on our stock price, a financial performance measure. (2) Reflects vesting schedule for annual RSU grants. Our NEOs have insurance benefits and are eligible to participate in our ESPP and 401(k) plan on the same basis as our other employees. We also provide a limited set of perquisites to our NEOs from time to time. See Other Compensation and Benefits below for more information.", + "chunk_count": 4, + "children": [ + { + "title": "Pay Mix Considerations", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Components of Pay/Pay Mix Considerations", + "level": 5, + "summary": "For Fiscal 2026, the CC decided that cash compensation for all NEOs would remain constant with Fiscal 2025 levels and that the largest portion of NEOs' total target pay would remain in the form of at-risk equity. The CC believes an emphasis on long-term, at-risk opportunities drives results and increases NEO and stockholder alignment, while providing sufficient annual cash compensation to be competitive and retain our NEOs. PSUs and RSUs provide long-term incentives and retention benefits as PSUs require achieving predetermined performance goals, and both PSUs and RSUs are subject to meaningful service-based vesting requirements (3 years for MY PSUs and 4 years for SY PSUs and RSUs). Given Mr. Huang's position as CEO, to even more tightly align his interests with stockholders, the CC determined that $100\\%$ of his equity grants should be at-risk and performance-based. Consistent with prior years, the CC split Mr. Huang's target equity opportunity evenly between SY PSUs (aligned with our annual corporate financial performance) and MY PSUs (aligned with our 3-year relative shareholder return). For other NEOs, the CC maintained a proportional equity mix and granted 50% of the target equity opportunity as RSUs and 50% as PSUs, with the PSU portion evenly split between SY PSUs and MY PSUs. The CC determined this structure balanced performance over short- and long-time horizons while providing a meaningful amount of time-vesting RSUs as a retention benefit.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Setting Executive Compensation Values", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Components of Pay/Setting Executive Compensation Values", + "level": 5, + "summary": "For Fiscal 2026, the CC decided that increases to each NEO's total target pay were appropriate in light of the scope and complexity of their roles and responsibilities. The CC raised Mr. Huang's total target pay by \\$3.5 million, or 11%, to \\$35.5 million, all in the form of equity, to align more closely with the median of the updated group of peer company CEOs. The CC split the target equity increase evenly between SY PSUs and MY PSUs to balance short- and long-term performance-based awards. For other NEOs, the CC decided, on Mr. Huang's recommendation, to align their total target pay to \\$16 million each, to reflect their comparable, significant contributions and impact on corporate performance, by increasing target equity opportunities by \\$1 to \\$2.5 million. For these NEOs, where peer data was available, the increases resulted in their total compensation approximating the 65th percentile of their respective peers.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Determining Equity Award Amounts", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Components of Pay/Determining Equity Award Amounts", + "level": 5, + "summary": "The Compensation Committee determines RSU, SY PSU, and MY PSU awards for NEOs using a 30-day average stock price to mitigate volatility. SY PSUs vest based on Fiscal 2026 Non-GAAP Operating Income, with caps at 150% for Mr. Huang and 200% for others, and a minimum of 50% at Threshold. MY PSUs vest based on 3-Year Relative TSR vs S&P 500, with caps of 150% for Mr. Huang and 200% for others, and a minimum of 25% at Threshold. Mr. Huang's lower cap is due to all his equity being performance-based, while others have 50% in service-based RSUs. No PSUs vest if Threshold is not met.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Performance Metrics and Goals for Executive Compensation", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Performance Metrics and Goals for Executive Compensation", + "level": 4, + "summary": "In March 2025, the Compensation Committee (CC) set rigorous Fiscal 2026 performance goals for executive compensation, aligning revenue and Non-GAAP Operating Income targets with record Fiscal 2025 results. Stretch goals were set at $190 billion revenue and $120 billion Non-GAAP Operating Income, but with an automatic reduction to $160 billion and $96 billion if additional export controls on H2O products were imposed in the first half of Fiscal 2026. In April 2025, after the U.S. government required a license for H2O exports to China, the goals were automatically adjusted. The CC designed Threshold, Base, and Stretch goals to be uncertain, challenging, and aligned with company performance, with straight-line interpolation for payouts between levels and no payout below Threshold.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fiscal 2026 Performance Achievement", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Fiscal 2026 Performance Achievement", + "level": 4, + "summary": "The company achieved record performance across all market platforms, driven by Data Center demand for Blackwell architecture and diverse customers. Revenue and Non-GAAP Operating Income for Fiscal 2026 exceeded Stretch Compensation Plan goals, as did the 3-year TSR relative to S&P 500. In March 2026, the Compensation Committee certified performance achievement with payouts: 25% of eligible SY PSUs vested on March 18, 2026, with quarterly vesting thereafter; 100% of eligible MY PSUs granted in Fiscal 2024 vested on March 18, 2026. Goals for MY PSUs granted in Fiscal 2025 and 2026 will be determined after performance periods end in January 2027 and 2028, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fiscal 2026 Target Compensation Actions and Performance-Based Payouts", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Fiscal 2026 Compensation Actions and Achievements/Fiscal 2026 Target Compensation Actions and Performance-Based Payouts", + "level": 4, + "summary": "The passage describes the Compensation Committee's target compensation actions for Fiscal 2026 for each named executive officer (NEO), including variable cash and equity opportunities. It explains that target equity values are based on the 30-day trailing average stock price, differing from SEC-reported grant date fair values. The text references individual tables for each NEO: Jen-Hsun Huang, Colette M. Kress, Ajay K. Puri, Debora Shoquist, and Timothy S. Teter, detailing their target pay structures.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Additional Executive Compensation Practices, Policies, and Procedures", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures", + "level": 3, + "summary": "", + "chunk_count": 7, + "children": [ + { + "title": "Other Compensation and Benefits", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Other Compensation and Benefits", + "level": 4, + "summary": "NVIDIA evaluates executive perquisites based on cost, business benefit, peer data, and governance. The CEO receives comprehensive security including residential protection, travel security, and private aircraft for business and occasional personal travel due to security concerns. These are not considered personal benefits but are reported in 'All Other Compensation'. Private aircraft is for business travel; personal guests may accompany at no incremental cost, with imputed income recognized. The CEO reimburses personal air travel costs. NEOs receive standard benefits like insurance, 401(k) match up to $11,500, and ESPP participation. No other perquisites were provided in Fiscal 2026.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Equity Grant Timing Practices", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Equity Grant Timing Practices", + "level": 4, + "summary": "The CC approves all equity award grants to our executive officers on or before the grant date. Typically, the CC completes its annual executive compensation review, sets performance goals and target compensation, and grants equity awards in March. This process is described in How We Determine Executive Compensation above. On occasion, the CC may grant equity awards outside of our annual cycle for new hires, promotions, recognition, retention or other purposes. While the CC has discretionary authority to approve such awards, it does not have a practice or policy of granting equity awards in anticipation of the release of material non-public information. In Fiscal 2026, we did not time the release of material non-public information in coordination with equity award grants in a manner that intentionally benefits the value of our executive compensation.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stock Ownership Guidelines", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Stock Ownership Guidelines", + "level": 4, + "summary": "The Board believes that executive officers should hold a meaningful equity interest in NVIDIA. Since June 2024, our Corporate Governance Policies require the CEO to hold shares valued at ten times his base salary, and our other NEOs to hold shares valued at three times their respective base salaries. Shares that count toward the ownership guidelines include shares held in trust, and shares held by immediate family members, but not unvested equity awards. NEOs have up to five years from appointment to reach their ownership threshold. These guidelines are intended to align NEO interests with stockholder interests. Each NEO currently exceeds the stock ownership requirements.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Compensation Recovery Policy", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Compensation Recovery Policy", + "level": 4, + "summary": "We have maintained a Compensation Recovery Policy since 2009, and amended it in 2023 to comply with Nasdaq's listing standards. Our policy requires the Company to recover certain incentive compensation from current or former executive officers in connection with certain restatements of financial statements, if the compensation exceeds the amount that the officers would have received based on the restated financial statements, subject to limited exceptions.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Tax and Accounting Implications", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Tax and Accounting Implications", + "level": 4, + "summary": "Under Section 162(m), compensation paid to each of the Company's “covered employees” that exceeds \\$1 million per year is generally non-deductible. The CC considers various factors when making its decisions and retains flexibility to provide non-deductible compensation if consistent with the Company’s executive compensation program goals and the best interests of the Company and its stockholders. The CC may also modify compensation that was initially intended to be exempt from the deduction limit if the modifications are consistent with the Company’s business needs. The CC also considers the impact of Section 409A of the Internal Revenue Code and strives to avoid possible adverse tax consequences. Additionally, under ASC 718, the Company records share-based compensation expense for equity compensation over the vesting period of the award.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Reconciliation of Non-GAAP Financial Measures", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Discussion and Analysis/Additional Executive Compensation Practices, Policies, and Procedures/Reconciliation of Non-GAAP Financial Measures", + "level": 4, + "summary": "A reconciliation between our GAAP operating income and non-GAAP operating income is as follows (in millions): [tables/table-43 Non-GAAP Income.html] We believe these non-GAAP financial measures enhance stockholders' overall understanding of our historical financial performance. The presentation of our non-GAAP financial measures is not meant to be considered in isolation nor as a substitute for our financial results prepared in accordance with GAAP, and our non-GAAP financial measures may be different from non-GAAP financial measures used by other companies.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Compensation Committee Report", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Compensation Committee Report", + "level": 2, + "summary": "The Compensation Committee of the Board of Directors oversees the compensation programs of NVIDIA on behalf of the Board of Directors. In fulfilling its oversight responsibilities, the Compensation Committee reviewed and discussed with management the Compensation Discussion and Analysis included in this proxy statement. In reliance on the review and discussions referred to above, the Compensation Committee recommended to the Board of Directors that the Compensation Discussion and Analysis be included in the Annual Report on Form 10-K of NVIDIA for the year ended January 25, 2026 and in this proxy statement. Compensation Committee Dawn Hudson, Tench Coxe, John O. Dabiri, Aarti Shah and Mark Stevens", + "chunk_count": 1, + "children": [] + }, + { + "title": "Risk Analysis of Our Compensation Plans", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Risk Analysis of Our Compensation Plans", + "level": 2, + "summary": "The Compensation Committee assessed the company's fiscal 2026 compensation programs for potential material risks, focusing on variable cash, equity, and sales incentive plans. It concluded that the programs, which balance short-term and long-term contributions, do not create risks likely to materially harm the business. Key risk mitigation features include a mix of base salary and variable pay, performance-based targets with caps, internal controls, committee approval, relative goals for multi-year PSUs, a compensation recovery policy, burn rate monitoring, multi-year vesting, stock ownership guidelines, and an insider trading policy prohibiting hedging and pledging.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Summary Compensation Table for Fiscal 2026, 2025, and 2024", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Summary Compensation Table for Fiscal 2026, 2025, and 2024", + "level": 2, + "summary": "The table summarizes compensation for NEOs in Fiscal 2026, 2025, and 2024. It includes salary, stock awards (grant date fair value per ASC 718), and non-equity incentive plan compensation. Notes explain that stock award values assume probable performance outcomes, with stretch scenario values provided in a separate table. Additional details cover cash plan payouts, rounding, and other compensation items like security, 401(k) matches, and life insurance premiums for each NEO.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Grants of Plan-Based Awards for Fiscal 2026", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Grants of Plan-Based Awards for Fiscal 2026", + "level": 2, + "summary": "The text details plan-based awards for NEOs in Fiscal 2026, including PSUs and RSUs under the 2007 Plan. It explains vesting schedules, performance criteria, and grant date fair values. Outstanding equity awards as of January 25, 2026 are listed, with values calculated using a closing stock price of $187.67. Specific vesting terms for PSUs and RSUs are provided, including stretch goals based on TSR relative to the S&P 500.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Option Exercises and Stock Vested in Fiscal 2026", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Option Exercises and Stock Vested in Fiscal 2026", + "level": 2, + "summary": "The following table shows information regarding stock acquired upon vesting for our NEOs during Fiscal 2026. We have not had any stock options outstanding since the end of Fiscal 2024. Stock Awards [tables/table-48 Stock Vesting Data.html] (1) Represents the gross number of shares acquired on vesting. Shares were withheld from these amounts to pay taxes due upon vesting. (2) Represents the gross number of shares acquired on vesting multiplied by the fair market value of our common stock as reported by Nasdaq on the date of vesting. (3) Includes an aggregate of 552,105 shares that were withheld to pay taxes due upon vesting. (4) Includes an aggregate of 233.268 shares that were withheld to pay taxes due upon vesting. (5) Includes an aggregate of 211,139 shares that were withheld to pay taxes due upon vesting. (6) Includes an aggregate of 194,424 shares that were withheld to pay taxes due upon vesting. (7) Includes an aggregate of 198,486 shares that were withheld to pay taxes due upon vesting.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Employment, Severance, and Change-in-Control Arrangements", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Employment, Severance, and Change-in-Control Arrangements", + "level": 2, + "summary": "The passage describes NVIDIA's change-in-control arrangements for executive officers. Under the 2007 Plan, upon a corporate transaction or change-in-control, outstanding stock awards may be assumed or substituted. If not, awards held by service providers accelerate vesting fully, while others terminate if unexercised. Potential payments upon termination or change-in-control are estimated based on unvested RSUs and PSUs, using a stock price of $187.67. Tables show estimated and actual PSU achievements for Fiscal 2024 and 2026, with further determinations pending for later grants.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Pay Ratio", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Pay Ratio", + "level": 2, + "summary": "The company calculated its CEO to median employee pay ratio for Fiscal 2026 as 129:1, using the same median employee identified in Fiscal 2024 due to no significant changes. The median employee's total compensation was $282,050, and the CEO's total compensation was $36,343,830. The methodology followed SEC regulations and used a consistently applied compensation measure including base salary, variable cash, and equity awards.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Pay Versus Performance", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Pay Versus Performance", + "level": 2, + "summary": "NVIDIA's executive compensation program aligns NEO pay with stockholder interests through pay for performance, using metrics like annual revenue, Non-GAAP Operating Income, and 3-year TSR relative to S&P 500. The text explains that 'compensation actually paid' (CAP) is calculated per SEC rules and does not reflect actual compensation received. It references tables summarizing compensation for CEO Jen-Hsun Huang and non-CEO NEOs (Colette M. Kress, Ajay K. Puri, Debora Shoquist, Timothy S. Teter) for Fiscal 2022-2026, including reconciliations from Summary Compensation Table totals to CAP. Key adjustments involve equity award fair values based on stock price or Monte Carlo simulations. TSR is cumulative from a fixed $100 investment, and the Nasdaq 100 Index is the peer group. The company-selected measure is Non-GAAP Operating Income, deemed the most important financial metric linking CAP to performance.", + "chunk_count": 3, + "children": [ + { + "title": "Most Important Financial Performance Measures", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Pay Versus Performance/Most Important Financial Performance Measures", + "level": 3, + "summary": "The following table is an unranked list of the most important financial performance measures linking Fiscal 2026 NEO CAP to company performance: Financial Measures [tables/table-55 Financial Metrics.html] Refer to CD&A above for a description of how each of these performance measures impacts NEO compensation.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Relationships Between Compensation Actually Paid and Financial Performance", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Pay Versus Performance/Relationships Between Compensation Actually Paid and Financial Performance", + "level": 3, + "summary": "The passage discusses the ratification of PwC as NVIDIA's independent registered public accounting firm for Fiscal 2027, including vote requirements and the Audit Committee's recommendation. It also references graphs comparing NEO CAP with financial performance measures and TSRs.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Recommendation of the Board", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Recommendation of the Board", + "level": 2, + "summary": "The Board recommends that you vote FOR the ratification of the selection of PwC as our independent registered accounting firm for our fiscal year ending January 31, 2027.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fees Billed by the Independent Registered Public Accounting Firm", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Fees Billed by the Independent Registered Public Accounting Firm", + "level": 2, + "summary": "The passage summarizes fees billed or expected by PwC for Fiscal 2026 and 2025, covering audit, audit-related, tax, and other professional services. It details categories: audit of consolidated financial statements and internal controls; review of sustainability metrics and attestation services; tax compliance and consulting; and other services like cybersecurity assessment. All services were pre-approved by the Audit Committee (AC) or its Chairperson. The AC has policies for pre-approving audit and permissible non-audit services, either by category, scope approval, or case-by-case, with delegated authority for unanticipated needs.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Report of the Audit Committee of the Board of Directors", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Report of the Audit Committee of the Board of Directors", + "level": 2, + "summary": "The Audit Committee (AC) of NVIDIA oversees accounting, financial reporting, internal controls, and audit activities. Management prepares financial statements in accordance with GAAP, while PricewaterhouseCoopers (PwC) performs independent audits. The AC reviews audit results, approves fees, and monitors internal audit functions. It relies on management and PwC representations and recommended the audited financial statements for Fiscal 2026 to the Board.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stockholder Proposals", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals", + "level": 2, + "summary": "We will promptly furnish the address and, to our knowledge, the number of shares held by the stockholder proponents of the proposals below upon receipt of a request to our Secretary for such information at email to shareholdermeeting@nvidia.com or by phone at: (408) 486-2000.", + "chunk_count": 19, + "children": [ + { + "title": "Proposal 4—Stockholder Proposal: Simple Majority Vote Standard", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard", + "level": 3, + "summary": "What am I voting on? A stockholder proposal to eliminate all supermajority voting provisions. Vote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter. Effect of abstentions: Same as a vote AGAINST. Effect of broker non-votes: None. Below is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 4. John Chevedden has notified us of his intent to present the following proposal at the 2026 Meeting.", + "chunk_count": 18, + "children": [ + { + "title": "Stockholder Proposal and Supporting Statement", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Stockholder Proposal and Supporting Statement", + "level": 4, + "summary": "", + "chunk_count": 1, + "children": [] + }, + { + "title": "Proposal 4 – Majority Vote Standard", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 4 – Majority Vote Standard", + "level": 4, + "summary": "The proposal requests that NVIDIA's Board replace all supermajority voting requirements in its charter and bylaws with a simple majority standard. It cites academic research linking supermajority requirements to lower company performance and notes high shareholder support for similar proposals at other companies, including 98% support at several firms in 2024. The proposal also asks the Board to not oppose the measure and to make extra efforts, such as adjourning the annual meeting, to secure the required vote, noting that the 2025 proposal received 65% support from outstanding shares, close to the 67% threshold.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Majority Vote Standard – Proposal 4", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Majority Vote Standard – Proposal 4", + "level": 4, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Opposing Statement of the Board", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Majority Vote Standard – Proposal 4/Opposing Statement of the Board", + "level": 5, + "summary": "Our Board recommends a vote AGAINST Proposal 4. At our 2025 Meeting, the Board proposed and recommended that stockholders approve an amendment and restatement of our Charter to eliminate the remaining supermajority voting provisions (the “2025 A&R Charter Proposal”). The Board also planned to adopt conforming amendments to our Bylaws, contingent on stockholder approval of such proposal. However, the 2025 A&R Charter Proposal did not receive the required level of stockholder support for approval. To effectuate Proposal 4, the Board would need to resubmit a proposal substantially similar to the 2025 A&R Charter Proposal for stockholder approval at a future annual meeting. At this time, the Board believes that addressing repetitive proposals on this topic would not be an effective use of time, resources, and attention.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recommendation of the Board", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Majority Vote Standard – Proposal 4/Recommendation of the Board", + "level": 5, + "summary": "Our Board recommends that you vote AGAINST Proposal 4— Simple Majority Vote Standard. What am I voting on? A stockholder proposal requesting an evaluation and report on faith-based community resource groups. Vote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter. Effect of abstentions: Same as a vote AGAINST. Effect of broker non-votes: None. Below is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 5. William Cunningham has notified us of its intent to present the following proposal at the 2026 Meeting.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Stockholder Proposal and Supporting Statement_2", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Stockholder Proposal and Supporting Statement_2", + "level": 4, + "summary": "", + "chunk_count": 1, + "children": [] + }, + { + "title": "Proposal 5 – Report on Faith-Based Community Resource Groups", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 5 – Report on Faith-Based Community Resource Groups", + "level": 4, + "summary": "Shareholders request NVIDIA to evaluate risks of religious discrimination, citing lack of faith-based Community Resource Groups despite supporting other ERGs. The proposal notes legal obligations under recent Supreme Court rulings and brand value concerns, urging a report on equal treatment for religious employees.", + "chunk_count": 3, + "children": [ + { + "title": "Opposing Statement of the Board", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 5 – Report on Faith-Based Community Resource Groups/Opposing Statement of the Board", + "level": 5, + "summary": "Our Board recommends a vote AGAINST Proposal 5. The requested report is unnecessary and would not provide meaningful additional benefits to NVIDIA or enhance stockholder value as we have already implemented robust systems and processes designed to minimize risks related to religious and other discrimination. NVIDIA is committed to ensuring that our employees are protected and supported through enterprise-wide policies, practices, and reporting mechanisms. We maintain a work environment in which all individuals are treated with dignity and respect. We do not tolerate unlawful discrimination, harassment, or retaliation in any form, including based on religion or creed and all employees are required to participate in discrimination and harassment prevention training. Our employees have multiple avenues to report concerns about harassment, discrimination or other workplace conduct, including anonymously and without fear of retaliation, and we investigate complaints promptly and take appropriate action when warranted. We also regularly gather employee feedback through surveys and other forums to identify concerns, track progress, and respond quickly to employee needs. The Board believes that our existing policies, practices, and feedback mechanisms already address the issues raised in Proposal 5, and that the additional evaluation and reporting requested would not be necessary or beneficial for NVIDIA or our stockholders.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recommendation of the Board", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 5 – Report on Faith-Based Community Resource Groups/Recommendation of the Board", + "level": 5, + "summary": "Our Board recommends that you vote AGAINST Proposal 5—Report on Faith-Based Community Resource Groups. What am I voting on? A stockholder proposal requesting an evaluation and report on civil rights and non-discrimination related to DEI. Vote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter. Effect of abstentions: Same as a vote AGAINST. Effect of broker non-votes: None. Below is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 6. American Conservative Values ETF has notified us of its intent to present the following proposal at the 2026 Meeting.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Stockholder Proposal and Supporting Statement_3", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Stockholder Proposal and Supporting Statement_3", + "level": 4, + "summary": "", + "chunk_count": 1, + "children": [] + }, + { + "title": "Proposal 6 – Report on Respecting Workforce Civil Liberties", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 6 – Report on Respecting Workforce Civil Liberties", + "level": 4, + "summary": "The passage discusses a shareholder resolution requesting NVIDIA to evaluate its hiring and training policies for potential discrimination based on race, religion, sex, and political views. It cites concerns that DEI programs may foster bias, increase legal risks, and face regulatory scrutiny, referencing recent Supreme Court decisions, executive orders, and EEOC actions. The resolution asks for a report on associated risks.", + "chunk_count": 3, + "children": [ + { + "title": "Opposing Statement of the Board", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 6 – Report on Respecting Workforce Civil Liberties/Opposing Statement of the Board", + "level": 5, + "summary": "The Board recommends voting against Proposal 6, arguing the requested report is unnecessary and would not enhance stockholder value. NVIDIA states it complies with equal employment opportunity laws, bases employment decisions on merit, and maintains strong oversight through committees (CC, AC, NCGC) for human capital, compliance, and ethics. The Board believes existing policies and mechanisms sufficiently address the proposal's issues.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recommendation of the Board", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 6 – Report on Respecting Workforce Civil Liberties/Recommendation of the Board", + "level": 5, + "summary": "Our Board recommends that you vote AGAINST Proposal 6—Report on Workforce Civil Liberties. What am I voting on? A stockholder proposal to requesting a report disclosing GHG emissions from the use of the Company's sold products. Vote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter. Effect of abstentions: Same as a vote AGAINST. Effect of broker non-votes: None. Below is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 7. Green Century Capital Management, Inc. has notified us of its intent to present the following proposal at the 2026 Meeting. Mercy Investment Services, Inc. is a co-filer for this proposal.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Stockholder Proposal and Supporting Statement_4", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Stockholder Proposal and Supporting Statement_4", + "level": 4, + "summary": "", + "chunk_count": 1, + "children": [] + }, + { + "title": "Proposal 7 – Greenhouse Gas Emissions Disclosure", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 7 – Greenhouse Gas Emissions Disclosure", + "level": 4, + "summary": "The text discusses climate change risks to NVIDIA, including macroeconomic impacts and supply chain disruptions. It notes NVIDIA's role in AI expansion and its climate targets, but criticizes the company's emissions disclosures for excluding use of sold products and showing increased absolute emissions. Peers like AMD and Intel provide more detailed climate reporting. A shareholder resolution requests NVIDIA to issue a report disclosing GHG emissions from use of sold products, with recommendations for improved disclosure practices.", + "chunk_count": 3, + "children": [ + { + "title": "Opposing Statement of the Board", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 7 – Greenhouse Gas Emissions Disclosure/Opposing Statement of the Board", + "level": 5, + "summary": "NVIDIA's Board recommends voting against Proposal 7, arguing it is unnecessary and overly prescriptive. The company highlights its robust climate reporting history, including detailed Scope 1, 2, and 3 emissions disclosures, SBTi-validated targets, and product carbon footprint summaries. The Board asserts that management is best positioned to handle complex Scope 3 disclosures, especially in the fast-evolving AI industry, and that the proposal's prescriptive approach is inappropriate and unhelpful to stockholders.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recommendation of the Board", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Stockholder Proposals/Proposal 4—Stockholder Proposal: Simple Majority Vote Standard/Proposal 7 – Greenhouse Gas Emissions Disclosure/Recommendation of the Board", + "level": 5, + "summary": "Our Board recommends that you vote AGAINST Proposal 7—Greenhouse Gas Emissions Disclosure.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + } + ] + }, + { + "title": "Equity Compensation Plan Information", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Equity Compensation Plan Information", + "level": 2, + "summary": "The passage summarizes equity compensation plan data as of January 25, 2026, including shares issuable from vesting RSUs and PSUs, and shares available for future issuance under the 2007 Plan and ESPP. No stock options were outstanding since Fiscal 2024. Key figures: 1,324,516,906 shares available under the 2007 Plan; 2,226,741,019 under ESPP, with up to 2,825,984 shares purchasable in the current period. In Fiscal 2026, 69,186,497 shares were granted as RSUs/PSUs (1,041,733 to NEOs, 21,588 to non-employee directors, 68,123,176 to other employees), and 13,226,923 shares were purchased under ESPP (4,655 by NEOs, 13,222,268 by other employees).", + "chunk_count": 1, + "children": [] + }, + { + "title": "Additional Information", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Additional Information", + "level": 2, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Delinquent Section 16(a) Reports", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Additional Information/Delinquent Section 16(a) Reports", + "level": 3, + "summary": "Section 16(a) of the Exchange Act requires our executive officers, directors and persons who own more than 10% of a registered class of our equity securities to file initial reports of ownership and reports of changes in ownership of our common stock and other equity securities with the SEC. Executive officers, directors and greater than 10% stockholders are required by SEC regulations to furnish us with copies of all Section 16(a) forms they file. To our knowledge, based solely on a review of the copies of such reports furnished to us and written representations that no other reports were required, all Section 16(a) filing requirements applicable to individuals who were, during Fiscal 2026, our executive officers, directors and greater than 10% beneficial owners were complied with.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Matters", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/Additional Information/Other Matters", + "level": 3, + "summary": "The Board knows of no other matters that will be presented for consideration at the 2026 Meeting. If any other matters are properly brought before the 2026 Meeting, it is the intention of the persons named in the accompanying proxy to vote on such matters in accordance with their best judgment.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "By Order of the Board of Directors", + "path": "2026-Annual-Report-Web.pdf/Executive Compensation/By Order of the Board of Directors", + "level": 2, + "summary": "### Delinquent Section 16(a) Reports Timothy S. Teter0 [images/image-81-### Delinq.jpg] Timothy S. Teter Secretary May 12, 2026 A COPY OF OUR ANNUAL REPORT ON FORM 10-K FOR THE FISCAL YEAR ENDED JANUARY 25, 2026 AS FILED WITH THE SEC IS BEING FURNISHED TO STOCKHOLDERS CONCURRENTLY HEREWITH. UPON WRITTEN REQUEST, WE WILL PROVIDE, WITHOUT CHARGE, AN ADDITIONAL COPY OF THE ANNUAL REPORT. STOCKHOLDERS MAY SUBMIT THEIR REQUESTS TO: INVESTOR RELATIONS, NVIDIA CORPORATION, 2788 SAN TOMAS EXPRESSWAY, SANTA CLARA, CALIFORNIA 95051 OR TO SHAREHOLDERMEETING@NVIDIA.COM. WE WILL ALSO FURNISH A COPY OF ANY EXHIBIT TO THE ANNUAL REPORT ON FORM 10-K IF SPECIFICALLY REQUESTED IN WRITING. NVIDIA and the NVIDIA logo are either registered trademarks or trademarks of NVIDIA Corporation in the United States and other countries. Other company names used in this publication are for identification purposes only and may be trademarks of their respective companies.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "FORM 10-K", + "path": "2026-Annual-Report-Web.pdf/FORM 10-K", + "level": 1, + "summary": "ANNUAL REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 For the fiscal year ended January 25, 2026 OR ☐ TRANSITION REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 Commission file number: 0-23985 ANNUAL REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 NVIDIA.0 [images/image-82-ANNUAL REP.jpg] NVIDIA.", + "chunk_count": 3, + "children": [ + { + "title": "NVIDIA CORPORATION", + "path": "2026-Annual-Report-Web.pdf/FORM 10-K/NVIDIA CORPORATION", + "level": 2, + "summary": "This is a Securities and Exchange Commission (SEC) filing for NVIDIA Corporation, a Delaware corporation with I.R.S. Employer Identification No. 94-3177549. Its principal executive offices are located at 2788 San Tomas Expressway, Santa Clara, California 95051, and its telephone number is (408) 486-2000. The registrant's common stock, with a par value of $0.001 per share, is registered under Section 12(b) of the Act and trades under the symbol NVDA on The Nasdaq Global Select Market. The filing includes various checkmarks indicating that NVIDIA is a well-known seasoned issuer, has filed all required reports, and has submitted electronic data files. It is classified as a large accelerated filer. As of July 25, 2025, the aggregate market value of voting stock held by non-affiliates was approximately $4.0 trillion, excluding shares held by directors and executive officers. As of February 20, 2026, there were 24.3 billion shares of common stock outstanding.", + "chunk_count": 1, + "children": [] + }, + { + "title": "DOCUMENTS INCORPORATED BY REFERENCE", + "path": "2026-Annual-Report-Web.pdf/FORM 10-K/DOCUMENTS INCORPORATED BY REFERENCE", + "level": 2, + "summary": "Portions of the registrant's Proxy Statement for its 2026 Annual Meeting of Shareholders to be filed with the Securities and Exchange Commission pursuant to Regulation 14A not later than 120 days after the end of the fiscal year covered by this Annual Report on Form 10-K are incorporated by reference into Part III, Items 10-14 of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Part I", + "path": "2026-Annual-Report-Web.pdf/Part I", + "level": 1, + "summary": "Item 1. Business 4 Item 1A. Risk Factors 12 Item 1B. Unresolved Staff Comments 32 Item 1C Cybersecurity 32 Item 2. Properties 33 Item 3. Legal Proceedings 33 Item 4. Mine Safety Disclosures 33", + "chunk_count": 1, + "children": [] + }, + { + "title": "Part II", + "path": "2026-Annual-Report-Web.pdf/Part II", + "level": 1, + "summary": "Item 5. Market for Registrant’s Common Equity, Related Stockholder Matters and Issuer Purchases of Equity Securities 33 Item 6. [Reserved] 35 Item 7. Management's Discussion and Analysis of Financial Condition and Results of Operations 36 Item 7A. Quantitative and Qualitative Disclosures About Market Risk 44 Item 8. Financial Statements and Supplementary Data 45 Item 9. Changes in and Disagreements with Accountants on Accounting and Financial Disclosure 45 Item 9A. Controls and Procedures 45 Item 9B. Other Information 46 Item 9C. Disclosure Regarding Foreign Jurisdictions that Prevent Inspections 46", + "chunk_count": 1, + "children": [] + }, + { + "title": "Part III", + "path": "2026-Annual-Report-Web.pdf/Part III", + "level": 1, + "summary": "Item 10. Directors, Executive Officers and Corporate Governance 46 Item 11. Executive Compensation 47 Item 12. Security Ownership of Certain Beneficial Owners and Management and Related Stockholder Matters 47 Item 13. Certain Relationships and Related Transactions, and Director Independence 47 Item 14. Principal Accountant Fees and Services 47", + "chunk_count": 1, + "children": [] + }, + { + "title": "Part IV", + "path": "2026-Annual-Report-Web.pdf/Part IV", + "level": 1, + "summary": "Item 15. Exhibits and Financial Statement Schedules 48 Item 16. Form 10-K Summary 83 Signatures 84", + "chunk_count": 3, + "children": [ + { + "title": "Where You Can Find More Information", + "path": "2026-Annual-Report-Web.pdf/Part IV/Where You Can Find More Information", + "level": 2, + "summary": "Investors and others should note that we announce material financial information to our investors using our investor relations website, press releases, SEC filings and public conference calls and webcasts. We also use the following social media channels as a means of disclosing information about the company, our products, our planned financial and other announcements and attendance at upcoming investor and industry conferences, and other matters and for complying with our disclosure obligations under Regulation FD: NVIDIA Corporate Blog (blogs.nvidia.com/) NVIDIA Technical Blog (developer.nvidia.com/blog/) NVIDIA LinkedIn (linkedin.com/company/nvidia) NVIDIA Facebook (facebook.com/nvidia) NVIDIA Instagram (instagram.com/nvidia) NVIDIA X (x.com/nvidia) NVIDIA Threads (threads.com/@nvidia) NVIDIA Investor Relations (investor.nvidia.com) NVIDIA YouTube (YouTube.com/nvidia). The information we post through these social media channels may be deemed material. Accordingly, investors should monitor these channels, in addition to following our press releases, SEC filings and public conference calls and webcasts. This list may be updated from time to time. The information we post through these channels is not a part of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Forward-Looking Statements", + "path": "2026-Annual-Report-Web.pdf/Part IV/Forward-Looking Statements", + "level": 2, + "summary": "This passage from an Annual Report on Form 10-K discusses forward-looking statements, which are based on management's beliefs and assumptions and are subject to risks and uncertainties. It identifies terms that indicate forward-looking statements, such as 'may,' 'will,' 'expect,' and 'believe.' The text warns that actual results may differ materially from those expressed or implied, and advises against undue reliance on these statements. It also clarifies that references to 'NVIDIA,' 'we,' 'us,' 'our,' or the 'Company' mean NVIDIA Corporation and its subsidiaries, and that beliefs are based on information available as of the filing date. The passage concludes with a copyright notice for 2026 NVIDIA Corporation.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Part I_2", + "path": "2026-Annual-Report-Web.pdf/Part I_2", + "level": 1, + "summary": "", + "chunk_count": 37, + "children": [ + { + "title": "Item 1. Business", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business", + "level": 2, + "summary": "", + "chunk_count": 18, + "children": [ + { + "title": "Our Company", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Company", + "level": 3, + "summary": "NVIDIA is a data center scale AI infrastructure company. Its technology stack includes the CUDA platform and domain-specific software libraries for AI, data analytics, scientific computing, robotics, and 3D graphics. The Blackwell architecture enables extreme co-design, interconnecting hundreds of thousands of GPUs. GPU-powered AI solutions are used for generative AI, agentic AI, and autonomous driving. NVIDIA has invested over $76.7 billion in R&D, inventing the GPU in 1999 and CUDA in 2006. Its platforms are used by cloud service providers, AI model makers, enterprises, researchers, gamers, and professionals across industries including healthcare, automotive, and financial services.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Our Businesses", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses", + "level": 3, + "summary": "We report our business results in two segments. The Compute & Networking segment includes our Data Center accelerated computing and networking platforms and AI solutions and software, and Automotive platforms and autonomous and electric vehicle solutions including software. The Graphics segment includes GeForce GPUs for gaming and PCs, and Quadro/NVIDIA RTX GPUs for enterprise workstation graphics.", + "chunk_count": 6, + "children": [ + { + "title": "Our Markets", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses/Our Markets", + "level": 4, + "summary": "We specialize in markets where our computing and AI infrastructure platforms can provide tremendous acceleration for applications. These platforms incorporate processors, interconnects, software, algorithms, systems, and services to deliver unique value. Our platforms address four large markets where our expertise is critical: Data Center, Gaming, Professional Visualization, and Automotive.", + "chunk_count": 5, + "children": [ + { + "title": "Data Center", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses/Our Markets/Data Center", + "level": 5, + "summary": "The NVIDIA Data Center platform accelerates compute-intensive workloads like AI, data processing, and scientific computing, offering superior total cost of ownership over CPU-only approaches. It includes compute and networking infrastructure for cloud, hyperscale, on-premises, and edge data centers. The platform features GPUs, CPUs, interconnects, and software stacks, with networking offerings like NVLink and InfiniBand. Customers include cloud providers, AI makers, and enterprises. Key architectures include Blackwell (fiscal 2025) and Rubin (fiscal 2027), with Rubin targeting agentic AI and reducing cost per token by 10x. The platform also supports physical AI for robotics.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Gaming", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses/Our Markets/Gaming", + "level": 5, + "summary": "Gaming is the largest entertainment industry, with PC gaming as the predominant platform. Many factors propel its growth, including new high production value games, the continued rise of eSports, social connectivity and the increasing popularity of game streamers, modders, or gamers who remaster games, and creators. Our products for the gaming market include GeForce RTX GPUs for gaming desktop and laptop PCs, GeForce NOW cloud gaming service, as well as SoCs and development services for game consoles. Our gaming platforms leverage our GPUs and sophisticated software to enhance the gaming experience with smoother, higher quality graphics. NVIDIA RTX features ray tracing technology for real-time, cinematic-quality rendering, and deep learning super sampling, or NVIDIA DLSS, our AI technology that boosts frame rates while generating high-quality images for games. RTX GPUs also feature NVIDIA tensor core technology making them well suited to accelerate a new generation of on-device AI applications. In fiscal year 2025, we announced the NVIDIA Blackwell GeForce RTX 50 Series family of desktop and laptop GPUs. The Blackwell architecture introduced neural graphics which combines AI models with traditional rendering to boost game performance, image quality, and interactivity, as well as the next generation of our DLSS technology powered by a new transformer model architecture. In fiscal year 2026, we launched and scaled Blackwell architecture for gaming and GeForce NOW.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Professional Visualization", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses/Our Markets/Professional Visualization", + "level": 5, + "summary": "We serve the Professional Visualization market by working closely with independent software vendors, or ISVs, to optimize their offerings for NVIDIA GPUs. Our GPU computing platform enhances productivity and introduces new capabilities for critical workflows in many fields, such as design, engineering, and digital content creation across a wide range of industry verticals. Additionally, the increasing number of generative and agentic AI applications is giving rise to the need for the enhanced AI and data processing capabilities of our RTX PRO GPUs. Many leading 3D design and content creation applications developed by our ecosystem partners support RTX, allowing professionals to accelerate and transform their workflows with NVIDIA RTX PRO GPUs and software. As these applications increasingly integrate AI, these GPUs are used and leverage the same Tensor Core technology found in our Data Center solutions.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Automotive", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Our Businesses/Our Markets/Automotive", + "level": 5, + "summary": "Automotive is comprised of platform solutions for automated driving from the cloud to the car. Leveraging our technology leadership in AI and building on long-standing relationships across several hundred automotive ecosystem partners, we are delivering a full stack end-to-end solution for the AV market under the DRIVE Hyperion platform. This platform consists of development infrastructure, high-performance, energy efficient DRIVE AGX computing hardware running an in-vehicle operating system (DRIVE OS), a reference sensor set that supports full self-driving capability as well as an open, modular DRIVE software platform for autonomous driving, mapping, and parking services, and intelligent in-vehicle experiences.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Business Strategies", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Business Strategies", + "level": 3, + "summary": "NVIDIA's key strategies include advancing its accelerated computing platform to solve complex problems with lower power consumption, leveraging full-stack innovation across architecture, chip design, and software. It extends leadership in AI with end-to-end platforms for training and inferencing, including NVIDIA AI Enterprise, NIM, NeMo, and AI Blueprints, supported by a large ecosystem of developers and partners. The company also leads in computer graphics infused with AI, enhancing gaming and professional visualization via GeForce, RTX, and Omniverse. In autonomous vehicles, it provides an AI-based hardware and software solution under the DRIVE brand, partnering with automotive OEMs and startups. Finally, it leverages intellectual property through licensing and development agreements to broaden technology reach.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Sales and Marketing", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Sales and Marketing", + "level": 3, + "summary": "The passage describes the company's worldwide sales and marketing strategy, emphasizing collaboration with a partner network including CSPs, OEMs, ODMs, ISVs, and others. Sales teams provide technical expertise and pre-sales assistance through application engineers and solution architects to optimize hardware and software for AI and LLM training. The company also engages with the software development community to promote platform optimization, supported by a developer program and the Deep Learning Institute for training.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Seasonality", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Seasonality", + "level": 3, + "summary": "Our computing platforms serve a diverse set of markets such as data centers, gaming, professional visualization, and automotive. Our desktop gaming products typically see stronger revenue in the second half of our fiscal year. Historical seasonality trends may not repeat.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Manufacturing", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Manufacturing", + "level": 3, + "summary": "The company uses a fabless and contracting manufacturing strategy, partnering with key suppliers for wafer fabrication, assembly, testing, and packaging. This approach avoids costs and risks of owning manufacturing operations, allowing focus on design, quality, marketing, and support. The supply chain is mainly in Asia but expanding to the U.S. and Latin America, with suppliers like TSMC, Samsung, SK Hynix, Micron, Hon Hai, Wistron, and Fabrinet.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Competition", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Competition", + "level": 3, + "summary": "The market is intensely competitive with rapid technological change. Key competitive factors include performance, product breadth, distribution, and pricing. Competitors include AMD, Intel, Huawei, and large cloud companies like Amazon and Microsoft, offering GPUs, CPUs, DPUs, and networking products. The company expects increased competition from new entrants and alliances.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Patents and Proprietary Rights", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Patents and Proprietary Rights", + "level": 3, + "summary": "The company protects its intellectual property through patents, trademarks, trade secrets, nondisclosure agreements, and licensing. Patents expire between 2026 and 2045. Protection varies internationally, especially in Asia, increasing piracy risk. They assess protection based on manufacturing locations, strategic directions, IP law enforcement, and commercial significance. They also license technology from third parties.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Government Regulations", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Government Regulations", + "level": 3, + "summary": "The passage details the impact of shifting US export controls on a company's business, particularly targeting China's semiconductor and supercomputing industries. Key events include restrictions on A100 and H100 products in 2022, expanded licensing requirements in 2023 and 2025, and a $4.5 billion charge for H2O inventory. The company faces foreclosure from China's data center market, harming its competitive position. Additional worldwide controls, like the AI Diffusion IFR, add uncertainty. Compliance with various regulations may increase costs and affect financial results.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Human Capital Management", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Human Capital Management", + "level": 3, + "summary": "As of the end of fiscal year 2026, we had approximately 42,000 employees in 38 countries; 31,000 were engaged in research and development and 11,000 were engaged in sales, marketing, operations, and administrative positions. To execute our business strategy successfully, we focus on recruiting, developing, and retaining top global talent. Within our workforce, more than 80 percent have technical roles and more than half of the workforce hold an advanced degree. Our employees also help to surface top talent, with over 40 percent of our new hires in fiscal year 2026 coming from employee referrals. In fiscal year 2026, our turnover rate was 3.7 percent. We invest in employee development through on-the-job trainings and tuition reimbursement programs. Our compensation and benefits are designed to reward performance and align employee interests with those of our shareholders through equity participation and comprehensive health and financial wellness programs. We also utilize employee listening systems to gather feedback and maintain an inclusive culture where hiring and promotions are based on merit.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Information About Our Executive Officers", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Information About Our Executive Officers", + "level": 3, + "summary": "The passage provides biographical information about NVIDIA's executive officers as of February 20, 2026. It details the roles, career histories, and educational backgrounds of five key executives: Jen-Hsun Huang (co-founder, President, CEO), Colette M. Kress (EVP and CFO), Ajay K. Puri (EVP, Worldwide Field Operations), Debora Shoquist (EVP of Operations), and Timothy S. Teter (EVP, General Counsel and Secretary). Each entry includes prior positions at companies like AMD, LSI Logic, Cisco, Microsoft, Sun Microsystems, and others, along with their academic degrees.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Available Information", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1. Business/Available Information", + "level": 3, + "summary": "Our annual reports on Form 10-K, quarterly reports on Form 10-Q, current reports on Form 8-K and, if applicable, amendments to those reports filed or furnished pursuant to Section 13(a) or 15(d) of the Securities Exchange Act of 1934, as amended, or the Exchange Act, are available free of charge on or through our website, http://www.nvidia.com, as soon as reasonably practicable after we electronically file such material with, or furnish it to, the Securities and Exchange Commission, or the SEC. The SEC's website, http://www.sec.gov, contains reports, proxy and information statements, and other information regarding issuers that file electronically with the SEC. Our web site and the information on it or connected to it are not a part of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Item 1A. Risk Factors", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors", + "level": 2, + "summary": "The following risk factors should be considered in addition to the other information in this Annual Report on Form 10-K. The following risks could harm our business, financial condition, results of operations or reputation, which could cause our stock price to decline. Additional risks, trends and uncertainties not presently known to us or that we currently believe are immaterial may also harm our business, financial condition, results of operations or reputation.", + "chunk_count": 11, + "children": [ + { + "title": "Risk Factors Summary", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors Summary", + "level": 3, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "Risks Related to Our Industry and Markets", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors Summary/Risks Related to Our Industry and Markets", + "level": 4, + "summary": "• Failure to meet the evolving needs of our industry and markets may adversely impact our financial results. - Competition could adversely impact our market share and financial results.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Risks Related to Demand, Supply, and Manufacturing", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors Summary/Risks Related to Demand, Supply, and Manufacturing", + "level": 4, + "summary": "\\- Long manufacturing lead times and uncertain supply and capacity availability, combined with a failure to estimate customer demand accurately has led and could lead to mismatches between supply and demand. - Dependency on third-party suppliers and their technology to manufacture, assemble, test, or package our products reduces our control over product quantity and quality, manufacturing yields, and product delivery schedules and could harm our business. - Defects in our products have caused and could cause us to incur significant expenses to remediate and could damage our business.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Risks Related to Our Global Operating Business", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors Summary/Risks Related to Our Global Operating Business", + "level": 4, + "summary": "The passage outlines various business risks including adverse economic conditions, international operations exposure, cybersecurity threats, business disruptions, climate change impacts, acquisition integration challenges, revenue concentration with limited partners, counterparty risks, talent retention issues, information system disruptions, and fluctuating operating results that could affect stock price.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Risks Related to Regulatory, Legal, Our Stock, and Other Matters", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors Summary/Risks Related to Regulatory, Legal, Our Stock, and Other Matters", + "level": 4, + "summary": "• We are subject to complex laws, rules, regulations, and political and other actions, including restrictions on the export of our products, which may adversely impact our business. - Scrutiny regarding our corporate sustainability practices could result in financial, reputational, or operational harm and liability. - Issues relating to the responsible use of our technologies, including AI, may result in reputational or financial harm and liability. - Adequately protecting our IP rights could be costly, and our ability to compete could be harmed if we are unsuccessful or if we are prohibited from making or selling our products. • We are subject to stringent and changing data privacy and security laws, rules, regulations, and other obligations. These areas could damage our reputation, deter customers, affect product design, or result in legal or regulatory proceedings and liability. - Our operating results may be adversely impacted by additional tax liabilities, higher than expected tax rates, changes in tax laws, and other tax-related factors. - Our business is exposed to the burden and risks associated with litigation, investigations, and regulatory proceedings. - Delaware law, provisions in our governing documents and our agreement with Microsoft could delay or prevent a change in control.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Risk Factors", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors", + "level": 3, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "Risks Related to Our Industry and Markets", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors/Risks Related to Our Industry and Markets", + "level": 4, + "summary": "The passage discusses risks from rapid technological changes, competition, and evolving customer needs. It highlights the need for timely innovation, R&D investments, and successful product adoption. Challenges include integrating licensed technology from Groq, cloud service agreements, and competition from larger rivals. Failure to adapt or achieve design wins could harm financial results.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Risks Related to Demand, Supply, and Manufacturing", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors/Risks Related to Demand, Supply, and Manufacturing", + "level": 4, + "summary": "The passage discusses risks from long manufacturing lead times, uncertain supply, and inaccurate demand estimation leading to supply-demand mismatches. It highlights challenges such as extended lead times over 12 months, premium payments for capacity, and non-cancellable purchase orders. Underestimating demand can cause revenue loss and reputational damage, while overestimation leads to inventory write-downs and price reductions. Factors like product transitions, competitor actions, and geopolitical tensions complicate demand forecasting. Dependency on third-party suppliers introduces risks of defects, yield issues, and production delays. Product defects may incur significant remediation costs and harm market share. The text emphasizes volatility from new use cases like cryptocurrency mining and gray market sales, and the impact of increasing product complexity and frequency of architectural transitions on supply chain management.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Risks Related to Our Global Operating Business", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors/Risks Related to Our Global Operating Business", + "level": 4, + "summary": "The passage outlines various risks that could harm NVIDIA's business, including adverse economic conditions, international operations, cybersecurity threats, business disruptions, climate change impacts, acquisition challenges, customer concentration, counterparty risks, talent retention issues, system failures, and fluctuating operating results. These factors could lead to increased costs, reduced revenue, legal liabilities, reputational damage, and stock price volatility.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Risks Related to Regulatory, Legal, Our Stock and Other Matters", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1A. Risk Factors/Risk Factors/Risks Related to Regulatory, Legal, Our Stock and Other Matters", + "level": 4, + "summary": "The passage details extensive legal, regulatory, and geopolitical risks facing NVIDIA, primarily focusing on U.S. export controls targeting AI semiconductors and GPUs, which have harmed its competitive position and foreclosed it from the China data center market. It also covers compliance with complex global laws (AI, data privacy, antitrust, trade sanctions), litigation risks, and the burdens of evolving regulations like the EU AI Act, all of which could materially impact business operations and financial results.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Item 1B. Unresolved Staff Comments", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1B. Unresolved Staff Comments", + "level": 2, + "summary": "None.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Item 1C. Cybersecurity", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1C. Cybersecurity", + "level": 2, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Risk management and strategy", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1C. Cybersecurity/Risk management and strategy", + "level": 3, + "summary": "The company has infrastructure, systems, policies, and procedures to prevent, reduce impact of, and react to cybersecurity incidents. It follows ISO 27001 frameworks, consults external experts, integrates cybersecurity risk into overall risk management, provides training, and has a vendor risk assessment process using questionnaires and industry standards like ISO 27001, ISO 28001, and C-TPAT.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Governance", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 1C. Cybersecurity/Governance", + "level": 3, + "summary": "The Company's Board of Directors oversees information security, with the Audit Committee reviewing policies and internal controls. Management provides regular updates, and a cross-functional leadership team meets to address cybersecurity matters. The Chief Security Officer, with 18 years of experience, manages material cybersecurity risks.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Item 2. Properties", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 2. Properties", + "level": 2, + "summary": "Our headquarters is in Santa Clara, California. We own and lease approximately 3 million square feet of office and building space for our corporate headquarters. In addition, we lease data center space in Santa Clara, California. We also own and lease facilities for data centers, research and development, and/or sales and administrative purposes throughout the U.S. and in various international locations, primarily in China, India, Israel, and Taiwan. We believe our existing facilities, both owned and leased, are in good condition and suitable for the conduct of our business. We do not identify or allocate assets by operating segment. For additional information regarding obligations under leases, refer to Note 17 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K, which information is hereby incorporated by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Item 3. Legal Proceedings", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 3. Legal Proceedings", + "level": 2, + "summary": "Please see Note 12 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for a discussion of our legal proceedings.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Item 4. Mine Safety Disclosures", + "path": "2026-Annual-Report-Web.pdf/Part I_2/Item 4. Mine Safety Disclosures", + "level": 2, + "summary": "Not applicable.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Part II_2", + "path": "2026-Annual-Report-Web.pdf/Part II_2", + "level": 1, + "summary": "Item 5. Market for Registrant's Common Equity, Related Stockholder Matters and Issuer Purchases of Equity Securities Our common stock is traded on the Nasdaq Global Select Market under the symbol NVDA. Public trading of our common stock began on January 22, 1999. Prior to that, there was no public market for our common stock. As of February 20, 2026, we had approximately 1,226 registered shareholders, not including those shares held in street or nominee name. In May 2024, we announced a ten-for-one stock split, or the Stock Split, of our issued common stock, which was effected through the filing of an amendment to the Company's Restated Certificate of Incorporation, or the Amendment, with the Secretary of the State of Delaware. In June 2024, the Company filed the Amendment to effect the Stock Split and proportionately increased the number of shares of the Company's authorized common stock from 8.0 billion to 80.0 billion. Shareholders of record at the close of market on June 6, 2024 received nine additional shares of common stock, distributed after the close of market on June 7, 2024. All share, equity award and per share amounts presented herein have been retrospectively adjusted to reflect the Stock Split.", + "chunk_count": 38, + "children": [ + { + "title": "Issuer Purchases of Equity Securities", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Issuer Purchases of Equity Securities", + "level": 2, + "summary": "On August 26, 2025, the Board approved an additional $60.0 billion share repurchase authorization with no expiration. In fiscal year 2026, 282 million shares were repurchased for $40.4 billion. As of January 25, 2026, up to $58.5 billion remained authorized for repurchase. Repurchases may be made via open market, private negotiations, Rule 10b5-1 plans, or structured agreements. The program can be suspended at any time. In fiscal year 2026, $974 million in cash dividends were paid. The table details fourth quarter repurchases. From January 26 to February 20, 2026, 8 million shares were repurchased for $1.5 billion. Additionally, 51 million shares were withheld for tax obligations under equity incentive plans, valued at $7.9 billion.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recent Sales of Unregistered Securities and Use of Proceeds", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Recent Sales of Unregistered Securities and Use of Proceeds", + "level": 2, + "summary": "On December 15, 2025, we acquired a company and issued to a key employee a total of 174,676 shares of our common stock, valued at approximately \\$31 million based on our closing stock price on the issuance date. The above securities were issued in a transaction not involving a public offering pursuant to an exemption from registration set forth in Section 4(a)(2) of the Securities Act (and Regulation D or Regulation S promulgated thereunder).", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stock Performance Graphs", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs", + "level": 2, + "summary": "The passage compares the cumulative total shareholder return of NVIDIA common stock, the S&P 500 Index, and the Nasdaq 100 Index over five years ending January 25, 2026, assuming a $100 investment on January 31, 2021 with dividends reinvested. It notes that on December 15, 2025, NVIDIA acquired a company and issued 174,676 shares to a key employee, valued at about $31 million. The content includes a line chart and tables with return data, followed by a reserved section and the start of Management's Discussion and Analysis.", + "chunk_count": 23, + "children": [ + { + "title": "Overview", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview", + "level": 3, + "summary": "", + "chunk_count": 22, + "children": [ + { + "title": "Our Company and Our Businesses", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Our Company and Our Businesses", + "level": 4, + "summary": "NVIDIA pioneered accelerated computing to help solve the most challenging computational problems. Since our original focus on PC graphics, we have expanded to several other large and important computationally intensive fields. Fueled by the sustained demand for exceptional 3D graphics and the scale of the gaming market, NVIDIA has leveraged its GPU architecture to create platforms for scientific computing, AI, data science, autonomous vehicles, robotics, and digital twin applications. NVIDIA is now a data center scale AI infrastructure company reshaping all industries. Our two operating segments are \"Compute & Networking\" and \"Graphics.\" Refer to Note 16 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information. Headquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recent Developments, Future Objectives and Challenges", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Recent Developments, Future Objectives and Challenges", + "level": 4, + "summary": "NVIDIA's fiscal year 2026 revenue reached $215.9 billion, up 65% year-over-year, driven by data center compute and AI solutions, with Blackwell architectures dominating Data Center revenue. The company faces challenges including export restrictions to China (incurring a $4.5 billion charge for H2O inventory), supply chain concentration in Asia, and risks from open-source AI competition. Investments totaled $17.5 billion in private companies and infrastructure, with $3.5 billion in guarantees. Gross margin decreased due to business model transitions and the H2O charge. Segment growth: Data Center up 68%, Gaming up 41%, Professional Visualization up 70%, Automotive up 39%.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Critical Accounting Estimates", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates", + "level": 4, + "summary": "Our consolidated financial statements are prepared in accordance with accounting principles generally accepted in the United States, or U.S. GAAP. The preparation of these financial statements requires us to make estimates and judgments that affect the reported amounts of assets, liabilities, revenue, cost of revenue, expenses and related disclosure of contingencies. Critical accounting estimates are those estimates that involve a significant level of estimation uncertainty and could have a material impact on our financial condition or results of operations. We have critical accounting estimates in the areas of inventories, income taxes, non-marketable equity securities, and revenue recognition. Refer to Note 1 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for a summary of significant accounting policies.", + "chunk_count": 7, + "children": [ + { + "title": "Inventories", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Inventories", + "level": 5, + "summary": "The passage discusses inventory provisions for write-downs to lower of cost or net realizable value, and for obsolete or excess inventory. Provisions are driven by excess quantities based on inventory levels, future purchase commitments, and assumptions about demand and market conditions. Factors causing excess or obsolete inventory include economic changes, demand decreases, technology shifts, new product introductions, and competitive actions. The net impact on gross margin from provisions and sales of previously written-down items was unfavorable 2.6% in fiscal year 2026 and 2.3% in fiscal year 2025. Inventory and capacity purchase commitments rely on customer demand forecasts and third-party lead times, which can exceed twelve months. Non-cancellable orders, premiums, and deposits may be used to secure supply. Market factors like competitor actions, product transitions, and macroeconomic conditions also affect demand.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Income Taxes", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Income Taxes", + "level": 5, + "summary": "The passage discusses the company's approach to income taxes in U.S. and foreign jurisdictions, focusing on deferred tax assets and liabilities based on estimates and judgments. It explains the use of a valuation allowance to reduce deferred tax assets to the amount more likely than not to be realized, and the recognition of tax benefits when realization becomes more-likely-than-not. Additionally, it covers the recognition of tax positions only if they are more-likely-than-not to be sustained upon audit, with interest and penalties included in income tax expense.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-Marketable Equity Securities", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Non-Marketable Equity Securities", + "level": 5, + "summary": "Non-marketable equity securities consist of investments in private companies without readily determinable fair values. They are measured at cost minus impairment, if any, and are adjusted for observable price changes in orderly transactions for a similar investment in the same issuer (the measurement alternative). These adjustments may require use of unobservable inputs. We assess impairment quarterly based on qualitative and quantitative factors, including the investee's operating performance and market trends.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Revenue Recognition", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Revenue Recognition", + "level": 5, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Revenue Allowances", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Revenue Recognition/Revenue Allowances", + "level": 6, + "summary": "For products sold with a right of return, we record a reduction to revenue by establishing a sales return allowance for estimated product returns at the time revenue is recognized, based primarily on historical return rates. However, if product returns for a fiscal period are anticipated to exceed historical return rates, we may determine that additional sales return allowances are required to reflect our estimated exposure for product returns. Return rights for certain stocking distributors for specific products are contractually limited based on a percentage of prior quarter shipments. For shipments to other customers, we do not allow returns, although we may approve returns for credit or refund based on applicable facts and circumstances. We account for customer programs, which involve rebates and marketing development funds, or MDFs, as a reduction in revenue and accrue for such programs based on the amount we expect to be claimed by customers. Certain customer programs include distributor price incentives or other channel programs for specific products and customer classes which require judgment as to whether the applicable incentives will be attained. Estimates for customer program accruals include a combination of historical attainment and claim rates and may be adjusted based on relevant internal and external factors.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Contracts with Multiple Performance Obligations", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Critical Accounting Estimates/Revenue Recognition/Contracts with Multiple Performance Obligations", + "level": 6, + "summary": "Our contracts may contain more than one performance obligation. Judgment is required in determining whether each performance obligation within a customer contract is distinct. Except for License and Development Arrangements, NVIDIA products and services function on a standalone basis and do not require a significant amount of integration or interdependency. Therefore, multiple performance obligations contained within a customer contract are considered distinct and are not combined for revenue recognition purposes. We allocate the total transaction price to each distinct performance obligation in an arrangement with multiple performance obligations on a relative standalone selling price basis. In most cases, we can establish standalone selling price based on directly observable prices of products or services sold separately in comparable circumstances to similar customers. If standalone selling price is not directly observable, such as when we do not sell a product or service separately, we determine standalone selling price based on market data and other observable inputs.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Results of Operations", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations", + "level": 4, + "summary": "A discussion regarding our financial condition and results of operations for fiscal year 2026 compared to fiscal year 2025 is presented below. A discussion regarding our financial condition and results of operations for fiscal year 2025 compared to fiscal year 2024 can be found under Item 7 in our Annual Report on Form 10-K for the fiscal year ended January 26, 2025, filed with the SEC on February 26, 2025, which is available free of charge on the SEC's website at http://www.sec.gov and at our investor relations website, http://investor.nvidia.com. The following table sets forth, for the periods indicated, certain items in our Consolidated Statements of Income expressed as a percentage of revenue. [tables/table-65 Income Statement.html]", + "chunk_count": 12, + "children": [ + { + "title": "Reportable Segments", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Reportable Segments", + "level": 5, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "Revenue by Reportable Segments", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Reportable Segments/Revenue by Reportable Segments", + "level": 6, + "summary": "[tables/table-66 Revenue Growth.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Operating Income by Reportable Segments", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Reportable Segments/Operating Income by Reportable Segments", + "level": 6, + "summary": "[tables/table-67 Revenue Growth.html] Compute & Networking revenue – The year over year increase was driven by the major platform shifts – accelerated computing and AI. Revenue from Data Center computing grew 59% driven by demand for our Blackwell computing platform. Revenue from Data Center networking grew 142% driven by the introduction and continued ramp of NVLink compute fabric for GB200 and GB300 systems and the growth of Ethernet and InfiniBand platforms. Graphics revenue – The year over year increase was driven by sales of our Blackwell architecture. Reportable segment operating income – The year over year increase in Compute & Networking segment operating income was driven by growth in revenue, partially offset by a \\$4.5 billion charge associated with H2O excess inventory and purchase obligations in the first quarter of fiscal year 2026. The year over year increase in Graphics segment operating income was driven by the growth in revenue.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Concentration of Revenue", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Reportable Segments/Concentration of Revenue", + "level": 6, + "summary": "NVIDIA's revenue is concentrated among a limited number of direct and indirect customers. Direct customers include AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. In fiscal year 2026, one direct customer accounted for 22% of total revenue and another for 14%, primarily from the Compute & Networking segment. Indirect customers purchase through direct customers, including CSPs, Neocloud builders, and enterprises. One AI research and deployment company contributed meaningfully to revenue via cloud services in fiscal year 2026. Revenue from customers headquartered outside the US was 31% in fiscal year 2026 and 41% in fiscal year 2025.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Gross Profit and Gross Margin", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Gross Profit and Gross Margin", + "level": 5, + "summary": "Gross profit is net revenue minus cost of revenue, which includes semiconductor costs, manufacturing support, acquisition-related amortization, and stock-based compensation. Gross margins fell from 75.0% in fiscal 2025 to 71.1% in fiscal 2026 due to a business model shift to Blackwell datacenter solutions and a $4.5 billion charge for H2O excess inventory. Inventory provisions totaled $7.2 billion in 2026 versus $3.7 billion in 2025, with a net unfavorable gross margin impact of 2.6% and 2.3% respectively. Operating expenses rose, with R&D driven by a 29% increase in compensation and a 79% increase in compute infrastructure, and SG&A increases from employee growth. Other income increased due to interest income growth and unrealized gains on equity securities, including Intel stock.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Income Taxes", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Income Taxes", + "level": 5, + "summary": "Income tax expense was $21.4 billion in fiscal 2026, up from $11.1 billion in fiscal 2025, with effective tax rates of 15.1% and 13.3%, respectively. The increase was due to lower tax benefits from stock-based compensation, FDDEI, and research credits. Rates remained below the 21.0% U.S. statutory rate due to these benefits and lower foreign taxes. The One Big Beautiful Bill Act (OBBBA) enacted in July 2025 impacted fiscal 2026 results. A $711 million valuation allowance was released on state deferred tax assets. Cash from operations increased due to higher revenue, while investing activities rose from equity purchases and a Groq license agreement. Financing activities increased due to higher share repurchases.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Liquidity", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Liquidity", + "level": 5, + "summary": "The passage discusses the company's primary liquidity sources including cash, cash equivalents, and marketable securities totaling $62.6 billion as of January 25, 2026. It states sufficient liquidity for at least twelve months and the foreseeable future. Marketable securities consist of U.S. government debt, corporate bonds, and foreign government securities, primarily in U.S. dollars. Except for $1.7 billion held outside the U.S. with no accrued taxes, most foreign holdings are available for U.S. use without additional federal taxes. The first quarter of fiscal year 2027 includes no estimated tax payments, while the second quarter includes two.", + "chunk_count": 4, + "children": [ + { + "title": "Capital Return to Shareholders", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Liquidity/Capital Return to Shareholders", + "level": 6, + "summary": "On August 26, 2025, our Board of Directors approved an additional \\$60.0 billion in share repurchase authorization, without expiration. In fiscal year 2026, we repurchased 282 million shares of our common stock for \\$40.4 billion. As of January 25, 2026, we were authorized, subject to certain specifications, to repurchase up to \\$58.5 billion of our common stock. From January 26, 2026 through February 20, 2026, we repurchased 8 million shares for \\$1.5 billion pursuant to a pre-established trading plan. We may execute repurchases from time to time, subject to market conditions, operating requirements, and other investment opportunities, in the open market, in privately negotiated transactions, pursuant to a Rule 10b5-1 trading plan or in structured share repurchase agreements in compliance with Rule 10b-18 of the Exchange Act. Our share repurchase program may be suspended at any time at our discretion. In fiscal year 2026, we paid cash dividends to our shareholders of \\$974 million. The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders. The U.S. Inflation Reduction Act of 2022 requires a 1% excise tax on certain share repurchases in excess of shares issued for employee compensation made after December 31, 2022. The excise tax is included in our share repurchase cost and was not material for fiscal years 2026 and 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Outstanding Indebtedness and Commercial Paper Program", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Liquidity/Outstanding Indebtedness and Commercial Paper Program", + "level": 6, + "summary": "Our aggregate debt maturities as of January 25, 2026, by year payable, are as follows: [tables/table-72 Debt Maturity.html] In January 2026, we increased the amount of our commercial paper program, pursuant to which we may issue unsecured commercial paper notes from time to time or all at once up to \\$25.0 billion. As of January 25, 2026, no commercial paper was outstanding. We will continue to evaluate issuing commercial paper as a component of our overall liquidity strategy. Refer to Note 11 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for further discussion.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Material Cash Requirements and Other Obligations", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Liquidity/Material Cash Requirements and Other Obligations", + "level": 6, + "summary": "The passage discusses facility lease guarantees, long-term debt, purchase commitments, and operating lease obligations referenced in various notes to financial statements. It mentions ongoing investment in the ecosystem and a potential partnership with OpenAI, with no assurance of completion. Capital expenditures were $6.1 billion in fiscal 2026 and $3.4 billion in fiscal 2025, with an expected increase in fiscal 2027. Unrecognized tax benefits of $4.0 billion, including $374 million in interest and penalties, are recorded, with uncertain timing of potential tax liabilities. The company is under IRS examination for fiscal years 2023 and 2024.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Adoption of New and Recently Issued Accounting Pronouncements", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Stock Performance Graphs/Overview/Results of Operations/Adoption of New and Recently Issued Accounting Pronouncements", + "level": 5, + "summary": "Refer to Note 1 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for a discussion of adoption of new and recently issued accounting pronouncements.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + } + ] + }, + { + "title": "Item 7A. Quantitative and Qualitative Disclosures about Market Risk", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 7A. Quantitative and Qualitative Disclosures about Market Risk", + "level": 2, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Investment and Interest Rate Risk", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 7A. Quantitative and Qualitative Disclosures about Market Risk/Investment and Interest Rate Risk", + "level": 3, + "summary": "The passage discusses interest rate risk related to fixed-rate investments and outstanding debt. A 0.5% decrease in yield would reduce investment fair value by $0.2 billion. The company has $8.5 billion in senior Notes with no financial statement risk from rate changes. Publicly-held equity securities face market volatility; a 10% decrease would reduce fair value by $1.8 billion. Non-marketable equity securities are valued at cost with adjustments for observable price changes.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Foreign Exchange Rate Risk", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 7A. Quantitative and Qualitative Disclosures about Market Risk/Foreign Exchange Rate Risk", + "level": 3, + "summary": "The company has minimal direct foreign exchange exposure as sales are in USD and forward contracts offset currency movements. A 10% USD strengthening would adversely impact accumulated other comprehensive income by $180M (2026) and $136M (2025) from foreign exchange contracts. Balance sheet hedging contracts would see an adverse pre-tax income impact of $124M (2026) and $129M (2025), offset by changes in foreign currency monetary assets and liabilities. Sales and manufacturer arrangements are in USD, but dollar strength could hurt competitiveness while weakness might raise supplier costs.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Item 8. Financial Statements and Supplementary Data", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 8. Financial Statements and Supplementary Data", + "level": 2, + "summary": "The information required by this Item is set forth in our Consolidated Financial Statements and Notes thereto included in this Annual Report on Form 10-K. Item 9. Changes in and Disagreements with Accountants on Accounting and Financial Disclosure None.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Item 9A. Controls and Procedures", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures", + "level": 2, + "summary": "", + "chunk_count": 6, + "children": [ + { + "title": "Controls and Procedures", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures/Controls and Procedures", + "level": 3, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "Disclosure Controls and Procedures", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures/Controls and Procedures/Disclosure Controls and Procedures", + "level": 4, + "summary": "Based on their evaluation as of January 25, 2026, our management, including our Chief Executive Officer and Chief Financial Officer, has concluded that our disclosure controls and procedures (as defined in Rule 13a-15(e) under the Exchange Act) were effective to provide reasonable assurance that the information we are required to disclose in reports that we file or submit under the Exchange Act is recorded, processed, summarized and reported within the time periods specified in the SEC rules and forms, and that such information is accumulated and communicated to our management, including our Chief Executive Officer and our Chief Financial Officer, as appropriate, to allow timely decisions regarding required disclosures.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Management's Annual Report on Internal Control Over Financial Reporting", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures/Controls and Procedures/Management's Annual Report on Internal Control Over Financial Reporting", + "level": 4, + "summary": "Our management is responsible for establishing and maintaining adequate internal control over financial reporting, as such term is defined in Exchange Act Rule 13a-15(f). Under the supervision and with the participation of our management, including our Chief Executive Officer and Chief Financial Officer, we conducted an evaluation of the effectiveness of our internal control over financial reporting as of January 25, 2026 based on the criteria set forth in Internal Control - Integrated Framework (2013) issued by the Committee of Sponsoring Organizations of the Treadway Commission. Based on our evaluation under the criteria set forth in Internal Control — Integrated Framework, our management concluded that our internal control over financial reporting was effective as of January 25, 2026. The effectiveness of our internal control over financial reporting as of January 25, 2026 has been audited by PricewaterhouseCoopers LLP, an independent registered public accounting firm, as stated in its report which is included herein.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Changes in Internal Control Over Financial Reporting", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures/Controls and Procedures/Changes in Internal Control Over Financial Reporting", + "level": 4, + "summary": "There have been no changes in our internal control over financial reporting during the quarter ended January 25, 2026 that have materially affected, or are reasonably likely to materially affect, our internal control over financial reporting. We are continuing a phased upgrade of our enterprise resource planning, or ERP, system to update our existing core financial systems. The ERP system is designed to accurately maintain our financial records used to report operating results. We will continue to evaluate each quarter whether there are changes that materially affect our internal control over financial reporting.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Inherent Limitations on Effectiveness of Controls", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9A. Controls and Procedures/Controls and Procedures/Inherent Limitations on Effectiveness of Controls", + "level": 4, + "summary": "Our management, including our Chief Executive Officer and Chief Financial Officer, does not expect that our disclosure controls and procedures or our internal controls, will prevent all error and all fraud. A control system, no matter how well conceived and operated, can provide only reasonable, not absolute, assurance that the objectives of the control system are met. Further, the design of a control system must reflect the fact that there are resource constraints, and the benefits of controls must be considered relative to their costs. Because of the inherent limitations in all control systems, no evaluation of controls can provide absolute assurance that all control issues and instances of fraud, if any, within NVIDIA have been detected.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Item 9B. Other Information", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9B. Other Information", + "level": 2, + "summary": "The following members of our Board of Directors and/or officers adopted, modified or terminated a trading arrangement that is intended to satisfy the affirmative defense conditions of Rule 10b5-1(c), or a Rule 10b5-1 Trading Arrangement: [tables/table-73 Stock Sale Plans.html] \\*Estimated assuming our closing stock price as of January 23, 2026. The number of shares is based on an estimate because the plan specifies a formulaic dollar amount of shares to be sold.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Item 9C. Disclosure Regarding Foreign Jurisdictions that Prevent Inspections", + "path": "2026-Annual-Report-Web.pdf/Part II_2/Item 9C. Disclosure Regarding Foreign Jurisdictions that Prevent Inspections", + "level": 2, + "summary": "Not Applicable.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Part III_2", + "path": "2026-Annual-Report-Web.pdf/Part III_2", + "level": 1, + "summary": "Certain information required by Part III is omitted from this report because we will file with the SEC a definitive proxy statement pursuant to Regulation 14A, or the 2026 Proxy Statement, no later than 120 days after the end of fiscal year 2026, and certain information included therein is incorporated herein by reference.", + "chunk_count": 90, + "children": [ + { + "title": "Item 10. Directors, Executive Officers and Corporate Governance", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance", + "level": 2, + "summary": "", + "chunk_count": 8, + "children": [ + { + "title": "Identification of Directors", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Identification of Directors", + "level": 3, + "summary": "Information regarding directors required by this item will be contained in our 2026 Proxy Statement under the caption “Proposal 1 — Election of Directors,” and is hereby incorporated by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Identification of Executive Officers", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Identification of Executive Officers", + "level": 3, + "summary": "Reference is made to the information regarding executive officers appearing under the heading “Information About Our Executive Officers” in Part I of this Annual Report on Form 10-K, which information is hereby incorporated by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Identification of Audit Committee and Financial Experts", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Identification of Audit Committee and Financial Experts", + "level": 3, + "summary": "Information regarding our Audit Committee required by this item will be contained in our 2026 Proxy Statement under the captions “Report of the Audit Committee of the Board of Directors” and “Information About the Board of Directors and Corporate Governance,” and is hereby incorporated by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Material Changes to Procedures for Recommending Directors", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Material Changes to Procedures for Recommending Directors", + "level": 3, + "summary": "Information regarding procedures for recommending directors required by this item will be contained in our 2026 Proxy Statement under the caption “Information About the Board of Directors and Corporate Governance,” and is hereby incorporated by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Delinquent Section 16(a) Reports", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Delinquent Section 16(a) Reports", + "level": 3, + "summary": "Information regarding compliance with Section 16(a) of the Exchange Act required by this item will be contained in our 2026 Proxy Statement under the caption “Delinquent Section 16(a) Reports,” and such disclosure, if any, is hereby incorporated by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Code of Conduct", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Code of Conduct", + "level": 3, + "summary": "Information regarding our Code of Conduct required by this item will be contained in our 2026 Proxy Statement under the caption “Information About the Board of Directors and Corporate Governance — Code of Conduct,” and is hereby incorporated by reference. The full text of our Code of Conduct and Financial Team Code of Conduct are published on the Investor Relations portion of our website, under Governance, at www.nvidia.com. If we make any amendments to either code, or grant any waiver from a provision of either code to any executive officer or director, we will promptly disclose the nature of the amendment or waiver on our website or in a report on Form 8-K. The contents of our website are not a part of this Annual Report on Form 10-K.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Insider Trading Policy", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 10. Directors, Executive Officers and Corporate Governance/Insider Trading Policy", + "level": 3, + "summary": "The information required by Item 408(b) of Regulation S-K is incorporated by reference from the information contained in our 2026 Proxy Statement under the heading “Information About the Board of Directors and Corporate Governance.”", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Item 11. Executive Compensation", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 11. Executive Compensation", + "level": 2, + "summary": "Information regarding our executive compensation required by this item will be contained in our 2026 Proxy Statement under the captions “Executive Compensation,” “Compensation Committee Interlocks and Insider Participation,” “Director Compensation,” and “Compensation Committee Report,” and is hereby incorporated by reference. Item 12. Security Ownership of Certain Beneficial Owners and Management and Related Stockholder Matters", + "chunk_count": 3, + "children": [ + { + "title": "Ownership of NVIDIA Securities", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 11. Executive Compensation/Ownership of NVIDIA Securities", + "level": 3, + "summary": "Information regarding ownership of NVIDIA securities required by this item will be contained in our 2026 Proxy Statement under the caption “Security Ownership of Certain Beneficial Owners and Management,” and is hereby incorporated by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Equity Compensation Plan Information", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 11. Executive Compensation/Equity Compensation Plan Information", + "level": 3, + "summary": "Information regarding our equity compensation plans required by this item will be contained in our 2026 Proxy Statement under the caption \"Equity Compensation Plan Information,\" and is hereby incorporated by reference.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Item 13. Certain Relationships and Related Transactions, and Director Independence", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 13. Certain Relationships and Related Transactions, and Director Independence", + "level": 2, + "summary": "Information regarding related transactions and director independence required by this item will be contained in our 2026 Proxy Statement under the captions “Review of Transactions with Related Persons” and “Information About the Board of Directors and Corporate Governance — Independence of the Members of the Board of Directors,” and is hereby incorporated by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Item 14. Principal Accountant Fees and Services", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 14. Principal Accountant Fees and Services", + "level": 2, + "summary": "Information regarding accounting fees and services required by this item will be contained in our 2026 Proxy Statement under the caption “Fees Billed by the Independent Registered Public Accounting Firm,” and is hereby incorporated by reference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Item 15. Exhibits and Financial Statement Schedules", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules", + "level": 2, + "summary": "Page", + "chunk_count": 74, + "children": [ + { + "title": "(a) 1. Financial Statements", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/(a) 1. Financial Statements", + "level": 3, + "summary": "Report of Independent Registered Public Accounting Firm (PCAOB ID: 238) 49 Consolidated Statements of Income for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 51 Consolidated Statements of Comprehensive Income for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 52 Consolidated Balance Sheets as of January 25, 2026 and January 26, 2025 53 Consolidated Statements of Shareholders' Equity for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 54 Consolidated Statements of Cash Flows for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 55 Notes to the Consolidated Financial Statements 56", + "chunk_count": 1, + "children": [] + }, + { + "title": "2. Financial Statement Schedule", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/2. Financial Statement Schedule", + "level": 3, + "summary": "Schedule II Valuation and Qualifying Accounts for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 81", + "chunk_count": 1, + "children": [] + }, + { + "title": "3. Exhibits", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/3. Exhibits", + "level": 3, + "summary": "The exhibits listed in the accompanying index to exhibits are filed or incorporated by reference as a part of this Annual Report on Form 10-K. 82", + "chunk_count": 1, + "children": [] + }, + { + "title": "Report of Independent Registered Public Accounting Firm", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/Report of Independent Registered Public Accounting Firm", + "level": 3, + "summary": "The text is an audit opinion from an independent auditor to the Board and Shareholders of NVIDIA Corporation. It covers the audit of consolidated financial statements for fiscal years ending January 25, 2026 and January 26, 2025, as well as internal control over financial reporting as of January 25, 2026. The auditor concludes that the financial statements fairly present the company's financial position and results in accordance with U.S. GAAP, and that effective internal control over financial reporting was maintained based on COSO criteria.", + "chunk_count": 4, + "children": [ + { + "title": "Basis for Opinions", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/Report of Independent Registered Public Accounting Firm/Basis for Opinions", + "level": 4, + "summary": "The passage describes the responsibilities of management and auditors regarding consolidated financial statements and internal control over financial reporting. Management is responsible for the financial statements and maintaining effective internal control, while the auditors express opinions on both. The audits are conducted in accordance with PCAOB standards to obtain reasonable assurance about the absence of material misstatements and the effectiveness of internal controls. Procedures include assessing risks, examining evidence, evaluating accounting principles and estimates, and testing internal controls.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Definition and Limitations of Internal Control over Financial Reporting", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/Report of Independent Registered Public Accounting Firm/Definition and Limitations of Internal Control over Financial Reporting", + "level": 4, + "summary": "A company's internal control over financial reporting is a process designed to provide reasonable assurance regarding the reliability of financial reporting and the preparation of financial statements for external purposes in accordance with generally accepted accounting principles. A company's internal control over financial reporting includes those policies and procedures that (i) pertain to the maintenance of records that, in reasonable detail, accurately and fairly reflect the transactions and dispositions of the assets of the company; (ii) provide reasonable assurance that transactions are recorded as necessary to permit preparation of financial statements in accordance with generally accepted accounting principles, and that receipts and expenditures of the company are being made only in accordance with authorizations of management and directors of the company; and (iii) provide reasonable assurance regarding prevention or timely detection of unauthorized acquisition, use, or disposition of the company's assets that could have a material effect on the financial statements. Because of its inherent limitations, internal control over financial reporting may not prevent or detect misstatements. Also, projections of any evaluation of effectiveness to future periods are subject to the risk that controls may become inadequate because of changes in conditions, or that the degree of compliance with the policies or procedures may deteriorate.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Critical Audit Matters", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/Report of Independent Registered Public Accounting Firm/Critical Audit Matters", + "level": 4, + "summary": "The passage discusses a critical audit matter regarding the valuation of inventories, specifically provisions for excess or obsolete inventories and excess product purchase commitments. Management uses significant judgment to develop assumptions about future demand and market conditions, including regulatory export restrictions. The auditor tested controls, management's process, data accuracy, and reasonableness of assumptions against historical results, external data, and technology changes. As of January 25, 2026, inventories were $21.4 billion and purchase obligations were $95.2 billion.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements", + "level": 3, + "summary": "", + "chunk_count": 66, + "children": [ + { + "title": "Note 1 - Organization and Summary of Significant Accounting Policies", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies", + "level": 4, + "summary": "", + "chunk_count": 25, + "children": [ + { + "title": "Our Company", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Our Company", + "level": 5, + "summary": "Headquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998. All references to “NVIDIA,” “we,” “us,” “our” or the “Company” mean NVIDIA Corporation and its subsidiaries. Certain prior fiscal year balances have been reclassified to conform to the current fiscal year presentation. Non-marketable equity securities, previously presented within other assets, were reclassified to be presented separately on our consolidated balance sheets and had no impact to total assets or consolidated statement of cash flows.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fiscal Year", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Fiscal Year", + "level": 5, + "summary": "We operate on a 52- or 53-week year, ending on the last Sunday in January. Fiscal years 2026, 2025 and 2024 were all 52-week years. Fiscal year 2027 will be a 53-week year with the fourth quarter consisting of 14 weeks.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Principles of Consolidation", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Principles of Consolidation", + "level": 5, + "summary": "Our consolidated financial statements include the accounts of NVIDIA Corporation and our wholly-owned subsidiaries. All intercompany balances and transactions have been eliminated in consolidation.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Use of Estimates", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Use of Estimates", + "level": 5, + "summary": "The preparation of financial statements in conformity with U.S. GAAP requires management to make estimates and assumptions that affect the reported amounts of assets and liabilities and disclosures of contingent assets and liabilities at the date of the financial statements and the reported amounts of revenue and expenses during the reporting period. Actual results could differ materially from our estimates. On an on-going basis, we evaluate our estimates, including those related to accounts receivable, cash equivalents and marketable securities, goodwill, income taxes, inventories and product purchase commitments, investigation and settlement costs, litigation, non-marketable equity securities, other contingencies, property, plant, and equipment, revenue recognition, and stock-based compensation. These estimates are based on historical facts and various other assumptions that we believe are reasonable.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Revenue Recognition", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Revenue Recognition", + "level": 5, + "summary": "We derive our revenue primarily from product sales including hardware and systems. We determine revenue recognition through the following steps: (1) identification of the contract with a customer; (2) identification of the performance obligations in the contract; (3) determination of the transaction price; (4) allocation of the transaction price to the performance obligations in the contract (where revenue is allocated on a relative standalone selling price basis by maximizing the use of observable inputs to determine the standalone selling price for each performance obligation); and (5) recognition of revenue when, or as, we satisfy a performance obligation. Payment from customers, per our standard payment terms, is generally due shortly after delivery of our products.", + "chunk_count": 4, + "children": [ + { + "title": "Product Sales Revenue", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Revenue Recognition/Product Sales Revenue", + "level": 6, + "summary": "Revenue from product sales is recognized when control transfers to customers, net of allowances for returns, customer programs, and taxes. Support and extended warranty revenue is recognized ratably over the service period. For products with a right of return, a sales return allowance is recorded based on historical return rates, with adjustments if returns exceed expectations. Customer programs, including rebates and market development funds (MDFs), are accounted for as a reduction to revenue and accrued based on expected claims.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Contracts with Multiple Performance Obligations", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Revenue Recognition/Contracts with Multiple Performance Obligations", + "level": 6, + "summary": "Our contracts may contain more than one deliverable, each of which is separately accounted for as a distinct performance obligation. We account for multiple agreements with a single customer as a single contract if the contractual terms and/or substance of those agreements indicate that they may be so closely related that they are, in effect, parts of a single contract. We allocate the total transaction price to each distinct performance obligation in an arrangement with multiple performance obligations on a relative standalone selling price basis. The standalone selling price reflects the price we would charge for a specific product or service if it were sold separately in similar circumstances and to similar customers. When determining standalone selling price, we maximize the use of observable inputs.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Product Warranties", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Revenue Recognition/Product Warranties", + "level": 6, + "summary": "We offer a limited warranty to end-users ranging from one to three years for products to repair or replace products for manufacturing defects or hardware component failures. Cost of revenue includes the estimated cost of product warranties that are calculated at the point of revenue recognition. Under limited circumstances, we may offer an extended limited warranty to customers for certain products. We also accrue for known warranty and indemnification issues if a loss is probable and can be reasonably estimated.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Stock-based Compensation", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Stock-based Compensation", + "level": 5, + "summary": "We use the closing trading price of our common stock on the date of grant, minus a dividend yield discount, as the fair value of awards of restricted stock units, or RSUs, and performance stock units, or PSUs, that are based on our corporate financial performance targets. We use a Monte Carlo simulation on the date of grant to estimate the fair value of PSUs that are based on our stock performance compared to market performance, or market-based PSUs. The compensation expense for RSUs and market-based PSUs is recognized using a straight-line attribution method over the requisite employee service period while compensation expense for PSUs is recognized using an accelerated amortization model based on performance targets probable of achievement. We estimate the fair value of shares to be issued under our employee stock purchase plan, or ESPP, using the Black-Scholes model at the commencement of an offering period in March and September of each year. Stock-based compensation for our ESPP is expensed using an accelerated amortization model. Additionally, for RSUs, PSUs, and market-based PSUs, we estimate expected forfeitures based on our historical forfeitures.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Litigation, Investigation and Settlement Costs", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Litigation, Investigation and Settlement Costs", + "level": 5, + "summary": "We currently are, and will likely continue to be subject to claims, litigation, and other actions, including potential regulatory proceedings, involving patent and other intellectual property matters, taxes, labor and employment, competition and antitrust, commercial disputes, goods and services offered by us and by third parties, and other matters. There are many uncertainties associated with any litigation or investigation, and we cannot be certain that these actions or other third-party claims against us will be resolved without litigation, fines and/or substantial settlement payments or judgments. If information becomes available that causes us to determine that a loss in any of our pending litigation, investigations or settlements is probable, and we can reasonably estimate the loss associated with such events, we will record the loss. However, the actual liability in any such litigation or investigation may be materially different from our estimates, which could require us to record additional costs. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the reasonably possible loss.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Foreign Currency Remeasurement", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Foreign Currency Remeasurement", + "level": 5, + "summary": "We use the U.S. dollar as our functional currency for our subsidiaries. Foreign currency monetary assets and liabilities are remeasured into United States dollars at end-of-period exchange rates. Non-monetary assets and liabilities such as property and equipment and equity are remeasured at historical exchange rates. Revenue and expenses are remeasured at exchange rates in effect during each period, except for those expenses related to non-monetary balance sheet amounts, which are remeasured at historical exchange rates. Gains or losses from foreign currency remeasurement are included in earnings in our Consolidated Statements of Income and to date have not been significant.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Income Taxes", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Income Taxes", + "level": 5, + "summary": "The passage discusses NVIDIA's recognition of current and deferred tax assets/liabilities based on estimates of taxes payable and future tax effects from temporary differences and carryforwards. A valuation allowance of $768 million is recorded for capital loss carryforwards and other deferred tax assets not likely to be realized. Tax benefits are recognized only if more-likely-than-not to be sustained, with interest and penalties included in income tax expense.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Net Income Per Share", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Net Income Per Share", + "level": 5, + "summary": "Basic net income per share is computed using the weighted average number of common shares outstanding during the period. Diluted net income per share is computed using the weighted average number of common and potentially dilutive shares outstanding during the period, using the treasury stock method. Any anti-dilutive effect of equity awards outstanding is not included in the computation of diluted net income per share.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Cash and Cash Equivalents and Marketable Securities", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Cash and Cash Equivalents and Marketable Securities", + "level": 5, + "summary": "The passage defines cash equivalents as highly liquid investments with original maturities of three months or less. Marketable securities include debt investments with longer maturities and publicly-held equity securities. Debt investments are classified as available-for-sale, reported at fair value with unrealized gains/losses in accumulated other comprehensive income. Realized gains/losses use the specific-identification method. Available-for-sale debt securities undergo impairment review; credit losses or write-downs are recognized if fair value is below amortized cost and sale is likely. Publicly-held equity securities and money market funds have fair value changes recorded in Other income.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fair Value of Financial Instruments", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Fair Value of Financial Instruments", + "level": 5, + "summary": "The passage discusses the fair value measurement of financial instruments, including cash equivalents, accounts receivable/payable, marketable securities, and derivatives. It explains that short-term items approximate fair value due to short maturities, while marketable securities and derivatives are reported at fair value. Derivatives are classified as accounting hedges (with gains/losses initially in other comprehensive income) or non-hedges (changes recognized in earnings). Fair value is classified into three levels based on input observability: Level 1 (quoted prices in active markets), Level 2 (observable inputs other than Level 1), and Level 3 (unobservable inputs with little market data).", + "chunk_count": 1, + "children": [] + }, + { + "title": "Concentration of Credit Risk", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Concentration of Credit Risk", + "level": 5, + "summary": "Financial instruments that potentially subject us to concentrations of credit risk consist primarily of cash equivalents, marketable securities, lease guarantees, and accounts receivable. Our investment policy requires the purchase of highly-rated fixed income securities, the diversification of investment type and credit exposures, and includes certain limits on our portfolio maturities. We perform ongoing credit evaluations of our customers' financial condition and maintain an allowance for potential credit losses. This allowance consists of an amount identified for specific customers and an amount based on overall estimated exposure. Our overall estimated exposure excludes amounts covered by credit insurance and letters of credit.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Inventories", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Inventories", + "level": 5, + "summary": "Inventory cost is computed on an adjusted standard basis, which approximates actual cost on an average or first-in, first-out basis. Inventory costs consist of the cost of semiconductors, including wafer fabrication, assembly, testing and packaging, manufacturing support costs, including labor and overhead associated with such purchases, final test yield fallout, and shipping costs, as well as the cost of purchased memory products and other component parts. We charge cost of sales for inventory provisions to write-down our inventory to the lower of cost or net realizable value or for obsolete or excess inventory, and for excess product purchase commitments. Most of our inventory provisions relate to excess quantities of products, based on our inventory levels and future product purchase commitments compared to assumptions about future demand including the impact of regulatory export restrictions on our products. Once inventory has been written-off or written-down, it creates a new cost basis for the inventory that is not subsequently written-up. We record a liability for noncancelable purchase commitments with suppliers for quantities in excess of our future demand forecasts consistent with our valuation of obsolete or excess inventory.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Property and Equipment", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Property and Equipment", + "level": 5, + "summary": "Property and equipment are stated at cost less accumulated depreciation. Depreciation of property and equipment is computed using the straight-line method based on the estimated useful lives of the assets of two to seven years. Once an asset is identified for retirement or disposition, the related cost and accumulated depreciation or amortization are removed, and a gain or loss is recorded. The estimated useful lives of our buildings are up to thirty years. Depreciation expense includes the amortization of assets recorded under finance leases. Leasehold improvements and assets recorded under finance leases are amortized over the shorter of the expected lease term or the estimated useful life of the asset.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Leases", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Leases", + "level": 5, + "summary": "We determine if an arrangement is or contains a lease at inception. Operating leases with lease terms of more than 12 months are included in operating lease assets, accrued and other current liabilities, and long-term operating lease liabilities on our consolidated balance sheet. Operating lease assets represent our right to use an underlying asset for the lease term and lease liabilities represent our obligation to make lease payments over the lease term. We combine lease and non-lease components for offices and data centers in determining the operating lease assets and liabilities. Operating lease assets and liabilities are recognized based on the present value of the remaining lease payments discounted using our incremental borrowing rate. Operating lease assets also include initial direct costs incurred and prepaid lease payments, minus any lease incentives. Our lease terms include options to extend or terminate the lease when it is reasonably certain that we will exercise that option. Lease costs are recognized on a straight-line basis over the lease term.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Goodwill", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Goodwill", + "level": 5, + "summary": "We allocate goodwill to reporting units based on the expected benefit from the business combination. Goodwill is subject to our annual impairment test during the fourth quarter of our fiscal year, or earlier if indicators of potential impairment exist. In completing our impairment test, we perform either a qualitative or a quantitative analysis on a reporting unit basis. Qualitative factors include industry and market considerations, overall financial performance, and other relevant events and factors affecting the reporting units. Goodwill impairments were not identified for the periods presented.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Intangible Assets and Other Long-Lived Assets", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Intangible Assets and Other Long-Lived Assets", + "level": 5, + "summary": "The passage discusses NVIDIA's intangible assets, including acquired technologies, customer relationships, and IP rights. These assets are amortized over one to twenty years using a method reflecting economic benefit consumption or straight-line. Long-lived assets are reviewed for impairment when events indicate carrying amounts may not be recoverable, comparing carrying amounts to undiscounted future cash flows. If impairment exists, a charge is recognized for the excess of carrying amount over fair value, determined by discounted cash flows. Assets held for disposal are reported at lower of carrying amount or fair value less costs to sell and are no longer depreciated.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Business Combination", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Business Combination", + "level": 5, + "summary": "The Company uses a screen test to determine if an acquisition is an asset acquisition or business combination based on fair value concentration. Purchase price is allocated to tangible and intangible assets and liabilities at fair value, with excess recorded as goodwill. Estimates involve uncertain assumptions like future cash flows and discount rates. Adjustments to assets and liabilities may be recorded during a one-year measurement period, with subsequent changes recognized in income. Acquisition-related expenses are expensed separately.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-Marketable Equity Securities", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Non-Marketable Equity Securities", + "level": 5, + "summary": "Non-marketable equity securities consist of investments in privately-held companies that do not have a readily determinable fair value. These investments are measured at cost minus impairment, if any, and are adjusted for changes resulting from observable price changes in orderly transactions for an identical or similar investment in the same issuer, or the measurement alternative. Fair value is based upon observable inputs in an inactive market and the valuation requires our judgment due to the absence of market prices and inherent lack of liquidity. All gains and losses on these investments, realized and unrealized, are recognized in Other income, net on our Consolidated Statements of Income. We assess whether an impairment loss has occurred on our investments in non-marketable equity securities, accounted for under the measurement alternative based on quantitative and qualitative factors. If any impairment is identified for non-marketable equity securities, we write down the investment to its fair value and record the corresponding charge through Other income, net on our Consolidated Statements of Income. The Company assesses its investments for significant influence to determine the appropriate method of accounting, including application of the equity method. Equity method investments were not material.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recently Issued Accounting Pronouncements", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 1 - Organization and Summary of Significant Accounting Policies/Recently Issued Accounting Pronouncements", + "level": 5, + "summary": "Recent Accounting Pronouncements Not Yet Adopted In November 2024, the Financial Accounting Standards Board, or FASB, issued a new accounting standard requiring disclosures of certain additional expense information on an annual and interim basis, including, among other items, the amounts of purchases of inventory, employee compensation, depreciation and intangible asset amortization included within each income statement expense caption, as applicable. We will adopt this standard in the fiscal year 2028 annual report. We do not expect the adoption of this standard to have a material impact on our Consolidated Financial Statements other than additional disclosures. NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 2 - Groq", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 2 - Groq", + "level": 4, + "summary": "In December 2025, we entered into a non-exclusive license agreement with Groq, Inc., or Groq, for its language processing unit technology and hired certain Groq employees. No customer contracts, existing products, or equity interests were purchased. We recorded \\$14.4 billion of goodwill and a \\$2.5 billion developed technology intangible asset, valued using a cost-to-recreate methodology with a five-year useful life. Goodwill, primarily attributable to the workforce and future development of the licensed technology, was recorded in the Compute & Networking reporting unit. Total consideration consists of \\$13.0 billion paid at closing and \\$4 billion, inclusive of imputed interest, payable within one year included in Accrued and Other Current Liabilities on our Consolidated Balance Sheets. The goodwill is tax deductible. Pro forma results of operations have not been presented because the effect was not material.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 3 - Stock-Based Compensation", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 3 - Stock-Based Compensation", + "level": 4, + "summary": "We recognize stock-based compensation expense from grants of restricted stock units, or RSUs, performance stock units, or PSUs, and market-based PSUs, and issuances under our employee stock purchase plan, or ESPP. Consolidated Statements of Income include stock-based compensation expense as follows: [tables/table-79 Cost Breakdown.html] The following is a summary of equity awards granted under our equity incentive plans: [tables/table-80 Equity Awards.html] As of January 25, 2026, aggregate unearned stock-based compensation expense was \\$14.8 billion, which is expected to be recognized over a weighted average period of 2.3 years for RSUs, PSUs, and market-based PSUs, and 0.9 years for ESPP.", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements", + "level": 4, + "summary": "(Continued) The fair value of shares issued under our ESPP has been estimated with the following assumptions: [tables/table-81 ESPP Option Data.html] For ESPP shares, the expected term represents the average term from the first day of the offering period to the purchase date. The risk-free interest rate assumption used to value ESPP shares is based upon observed interest rates on Treasury bills appropriate for the expected term. Our expected stock price volatility assumption for ESPP is estimated using historical volatility. For awards granted, we use the dividend yield at grant date. Our RSUs, PSUs, and market-based PSUs are not eligible for cash dividends prior to vesting; therefore, the fair values of RSUs, PSUs, and market-based PSUs are discounted for the dividend yield. Additionally, for RSUs, PSUs, and market-based PSUs, we estimate expected forfeitures based on our historical forfeitures.", + "chunk_count": 4, + "children": [ + { + "title": "Equity Incentive Program", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Equity Incentive Program", + "level": 5, + "summary": "We grant RSUs, PSUs, market-based PSUs, and stock purchase rights under the following equity incentive plans. In addition, in connection with our acquisitions of various companies, we have assumed certain stock-based awards granted under their stock incentive plans and converted them into our RSUs.", + "chunk_count": 3, + "children": [ + { + "title": "Amended and Restated 2007 Equity Incentive Plan", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Equity Incentive Program/Amended and Restated 2007 Equity Incentive Plan", + "level": 6, + "summary": "The NVIDIA Corporation Amended and Restated 2007 Equity Incentive Plan, or the 2007 Plan, authorizes the issuance of incentive stock options, non-statutory stock options, restricted stock, RSUs, stock appreciation rights, performance stock awards, performance cash awards, and other stock-based awards to employees, directors and consultants. Only our employees may receive incentive stock options. We grant RSUs, PSUs and market-based PSUs under the 2007 Plan. As of January 25, 2026, up to 192 million shares of our common stock could be issued pursuant to stock awards granted under the 2007 Plan, and 1.3 billion shares were available for future grants. Subject to certain exceptions, RSUs vest generally over four years subject to continued service. PSUs vest over four years, subject to continued service and performance conditions. Market-based PSUs vest on approximately the third anniversary of the date of grant subject to market conditions. However, the number of shares subject to both PSUs and market-based PSUs that are eligible to vest is determined by the Compensation Committee based on achievement of predetermined criteria.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Amended and Restated 2012 Employee Stock Purchase Plan", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Equity Incentive Program/Amended and Restated 2012 Employee Stock Purchase Plan", + "level": 6, + "summary": "Employees who participate in the NVIDIA Corporation Amended and Restated 2012 Employee Stock Purchase Plan, or as most recently amended and restated, the 2012 Plan, may have up to 25% of their earnings withheld to purchase shares of common stock. The Board may decrease this percentage at its discretion. Each offering period is about 24 months, divided into four purchase periods of six months. The price of common stock purchased under our 2012 Plan will be equal to 85% of the lower of the fair market value of the common stock on the commencement date of each offering period or the fair market value of the common stock on each purchase date within the offering. As of January 25, 2026, we had 2.2 billion shares reserved for future issuance under the 2012 Plan.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_2", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_2", + "level": 4, + "summary": "(Continued) Equity Award Activity The following is a summary of our equity award transactions under our equity incentive plans: [tables/table-82 RSU_PSU Activity.html] As of January 25, 2026 and January 26, 2025, there were 1.3 billion and 1.4 billion shares, respectively, of common stock available for future grants under our equity incentive plans. The total fair value of RSUs and PSUs, as of their respective vesting dates, during the years ended January 25, 2026, January 26, 2025, and January 28, 2024, was \\$22.2 billion, \\$15.1 billion, and \\$8.2 billion, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 4 - Net Income Per Share", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 4 - Net Income Per Share", + "level": 4, + "summary": "The following is the basic and diluted net income per share computations for the periods presented: [tables/table-83 EPS Data.html] (1) Net income divided by basic weighted average shares. (2) Net income divided by diluted weighted average shares.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 5 - Goodwill", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 5 - Goodwill", + "level": 4, + "summary": "As of January 25, 2026, the total carrying amount of goodwill was \\$20.8 billion, consisting of goodwill balances allocated to our Compute & Networking and Graphics reporting units of \\$20.5 billion and \\$370 million, respectively. As of January 26, 2025, the total carrying amount of goodwill was \\$5.2 billion, consisting of goodwill balances allocated to our Compute & Networking and Graphics reporting units of \\$4.8 billion and \\$370 million, respectively. Goodwill increased by \\$15.6 billion in fiscal year 2026 and was allocated to our Compute & Networking reporting unit. During the fourth quarters of fiscal years 2026, 2025, and 2024, we completed our annual qualitative impairment tests and concluded that goodwill was not impaired.", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_3", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_3", + "level": 4, + "summary": "(Continued)", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 6 - Amortizable Intangible Assets", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 6 - Amortizable Intangible Assets", + "level": 4, + "summary": "The components of our amortizable intangible assets are as follows: [tables/table-84 Intangible Assets.html] Amortization expense associated with intangible assets for fiscal years 2026, 2025, and 2024 was \\$488 million, \\$593 million, and \\$614 million, respectively. The following table outlines the estimated future amortization expense related to the net carrying amount of intangible assets as of January 25, 2026: [tables/table-85 Amortization Expense.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_4", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_4", + "level": 4, + "summary": "(Continued)", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 7 - Cash Equivalents and Marketable Securities", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 7 - Cash Equivalents and Marketable Securities", + "level": 4, + "summary": "The passage discusses the fair value determination of financial assets using quoted market prices. It summarizes cash equivalents and marketable securities as of January 25, 2026, noting reclassification of an investment from non-marketable to marketable equity securities, with $10.5 billion subject to short-term lock-up restrictions and $4.8 billion in long-term lock-up restrictions through December 2027. Net unrealized gains on publicly-held equity securities were $6.6 billion for fiscal year 2026, while net realized gains were not significant for fiscal years 2024-2026.", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_5", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_5", + "level": 4, + "summary": "(Continued) Jan 26, 2025 [tables/table-87 Fair Value Table.html] The following tables provide the breakdown of unrealized losses, aggregated by investment category and length of time that individual debt securities have been in a continuous loss position: [tables/table-88 Debt Securities.html] Gross unrealized losses related to debt securities in a continuous loss position of twelve months or greater as of January 25, 2026 and January 26, 2025 were not significant. Gross unrealized losses are related to fixed income securities, driven primarily by changes in interest rates. The estimated fair value of debt securities included in cash equivalents and marketable securities are shown below by contractual maturity. [tables/table-89 Debt Maturity.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 8 - Non-marketable Equity Securities", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 8 - Non-marketable Equity Securities", + "level": 4, + "summary": "Our non-marketable equity securities are valued under the measurement alternative applying valuation methods based on observable transactions for similar investments of the same issuer and unobservable inputs such as volatility, expected", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)", + "level": 4, + "summary": "(Continued) time to liquidity, risk free rate and security-specific rights and obligations. Gains and losses on these investments, realized and unrealized, are recognized in Other income, net on our Consolidated Statements of Income. Adjustments to the carrying value of our non-marketable equity securities during fiscal years 2026 and 2025 were as follows: [tables/table-90 Equity Securities.html] (1) Represents reclassifications from non-marketable equity securities to marketable securities following public market trading. Non-marketable equity securities had cumulative gross unrealized gains of \\$2.7 billion and \\$1.1 billion, and cumulative gross unrealized losses and impairments of \\$176 million and \\$105 million on securities held as of January 25, 2026 and January 26, 2025, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 9 - Balance Sheet Components", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 9 - Balance Sheet Components", + "level": 4, + "summary": "We refer to customers who purchase products directly from NVIDIA as direct customers, such as AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. Three direct customers accounted for 25%, 18%, and 13% of our accounts receivable balance as of January 25, 2026. Two direct customers accounted for 17% and 16% of our accounts receivable balance as of January 26, 2025. Certain balance sheet components are as follows: [tables/table-91 Inventory Data.html] (1) In fiscal years 2026 and 2025, we recorded inventory provisions of \\$4.0 billion and \\$1.6 billion, respectively, in cost of revenue. [tables/table-92 Property & Equipment.html] (A) Land is a non-depreciable asset.", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)_2", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued)_2", + "level": 4, + "summary": "The passage details depreciation expenses for fiscal years 2026, 2025, and 2024 ($2.4B, $1.3B, $894M), accumulated amortization of leasehold improvements and finance leases ($519M and $410M as of Jan 25, 2026 and Jan 26, 2025), and property, equipment, and intangible assets acquired but not paid for ($820M, $525M, $170M). It also covers deferred revenue changes, including customer advances of $160M and $81M as of the respective dates, and $9.0B and $3.6B of customer advances for fiscal years 2026 and 2025.", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_6", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_6", + "level": 4, + "summary": "(Continued) We recognized revenue of \\$974 million and \\$729 million in fiscal years 2026 and 2025, respectively, that were included in the prior year end deferred revenue balance. As of January 25, 2026, revenue related to remaining performance obligations from contracts greater than one year in length was \\$2.3 billion, which includes \\$1.9 billion from deferred revenue and \\$390 million which has not yet been billed nor recognized as revenue. Approximately 42% of revenue from contracts greater than one year in length will be recognized over the next twelve months.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 10 - Derivative Financial Instruments", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 10 - Derivative Financial Instruments", + "level": 4, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Foreign Currency Derivatives", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 10 - Derivative Financial Instruments/Foreign Currency Derivatives", + "level": 5, + "summary": "The company uses foreign currency forward contracts to hedge against exchange rate impacts on operating expenses, designating them as accounting hedges with gains/losses recorded in OCI and reclassified to expenses. Non-designated contracts hedge monetary assets/liabilities, with fair value changes in other income/expense. Notional values are presented in a table. Unrealized gains/losses were not significant as of Jan 2025 and 2026. All contracts mature within 18 months, with no significant deferred gains/losses expected in the next twelve months.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Facility Lease Guarantees", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 10 - Derivative Financial Instruments/Facility Lease Guarantees", + "level": 5, + "summary": "In fiscal year 2026, we entered into agreements to guarantee partners' facility lease obligations in the event of their default in exchange for warrants. The maximum gross exposure under all agreements is \\$3.5 billion, which is reduced as the partners make payments to the lessors over terms ranging from 5 to 7 years. The partners have placed \\$712 million in escrow to mitigate our potential exposure. The guarantees, classified as credit derivatives with changes in fair value recognized in Other income and expense, were not material.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_7", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_7", + "level": 4, + "summary": "(Continued) Note 11 - Debt [tables/table-97 Debt Notes Table.html] As of January 25, 2026 and January 26, 2025, the estimated fair value of debt was \\$7.5 billion and \\$7.2 billion, respectively. The estimated fair values are based on Level 2 inputs. Our notes are unsecured senior obligations. Existing and future liabilities of our subsidiaries will be effectively senior to the notes. Our notes pay interest semi-annually. We may redeem each of our notes prior to maturity, subject to a make-whole premium. The maturity of the notes is calendar year. As of January 25, 2026, we complied with the required covenants, which are non-financial in nature, under the outstanding notes. In January 2026, we increased the size of our commercial paper program from \\$575 million to \\$25.0 billion. As of January 25, 2026, no commercial paper was outstanding.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 12 - Commitments and Contingencies", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 12 - Commitments and Contingencies", + "level": 4, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "Commitments", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 12 - Commitments and Contingencies/Commitments", + "level": 5, + "summary": "The passage details various commitments as of January 25, 2026. Manufacturing, supply, and capacity commitments total $95.2 billion, mostly payable through fiscal 2027, with agreements that may be cancellable or adjustable. Multi-year cloud service agreement commitments are $27 billion, with payments spread from 2027 to 2032 and beyond, and may be reduced if capacity is terminated or sold. Investment commitments stand at $11.4 billion, expected to be made through fiscal 2027. Other commitments are $3.4 billion, with most paid by fiscal 2027.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Accrual for Product Warranty Liabilities", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 12 - Commitments and Contingencies/Accrual for Product Warranty Liabilities", + "level": 5, + "summary": "The estimated amount of product warranty liabilities was \\$2.8 billion and \\$1.3 billion as of January 25, 2026 and January 26, 2025, respectively. The estimated product returns and product warranty activity consisted of the following: [tables/table-98 Restructuring Reserve.html] In fiscal years 2026, 2025, and 2024 the additions in product warranty liabilities primarily related to our Compute & Networking segment. We have provided indemnities for matters such as tax, product, and employee liabilities. We have included intellectual property indemnification provisions in our technology-related agreements with third parties. Maximum potential future payments cannot be estimated because many of these agreements do not have a maximum stated liability. We have not recorded any liability in our Consolidated Financial Statements for such indemnifications.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Litigation", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 12 - Commitments and Contingencies/Litigation", + "level": 5, + "summary": "The text details two related lawsuits against NVIDIA. A securities class action (In Re NVIDIA Corporation Securities Litigation) was filed in 2018, alleging false statements about GPU demand and cryptocurrency mining. After dismissal, appeals, and a Supreme Court certiorari that was dismissed as improvidently granted, the case was remanded to district court in 2025. A related derivative lawsuit in California remains stayed, while two Delaware derivative actions are also stayed pending resolution of the securities appeal.", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 12 - Commitments and Contingencies/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements", + "level": 5, + "summary": "The passage details ongoing legal proceedings against NVIDIA, including a securities class action and derivative lawsuits alleging false statements about channel inventory and cryptocurrency mining's impact on GPU demand. These actions seek damages, disgorgement, and governance reforms. As of January 25, 2026, no contingent liabilities have been accrued as losses are not deemed probable or estimable. The company believes other routine legal matters will not materially affect its financial position.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 13 - Income Taxes", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 13 - Income Taxes", + "level": 4, + "summary": "The FASB issued a new accounting standard which includes new and updated income tax disclosures, including disaggregation of information in the rate reconciliation and income taxes paid, which we adopted on a prospective basis for the year ending January 25, 2026. The Income tax expense applicable to income before income taxes consists of the following: [tables/table-99 Income Tax Expense.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_8", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_8", + "level": 4, + "summary": "(Continued) Income before income tax consists of the following: [tables/table-100 Income Before Tax.html] The income tax expense (benefit) differs from the amount computed by applying the U.S. federal statutory rate of 21.0% to income before income taxes for the fiscal year ended January 25, 2026 as follows: [tables/table-101 Tax Rate Breakdown.html] (1) State taxes in California, Tennessee, Arizona, and Illinois made up the majority of the tax effect in fiscal year 2026. (2) Includes the tax effects of enactment of new tax laws, change in valuation allowance, and change in unrecognized tax benefits.", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_9", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_9", + "level": 4, + "summary": "(Continued) The income tax expense (benefit) differs from the amount computed by applying the U.S. federal statutory rate of 21% to income before income taxes for fiscal years ended January 26, 2025 and January 28, 2024 as follows: [tables/table-102 Tax Rate Analysis.html] In July 2025, the OBBBA was enacted into law and contains several changes to key U.S. federal income tax laws. We have recognized the tax effects of currently effective OBBBA provisions in our results for fiscal year 2026. The amount of cash paid for income taxes (net of refunds) for the fiscal year ended January 25, 2026 is as follows: [tables/table-103 Income Tax Paid.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_10", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_10", + "level": 4, + "summary": "The passage details NVIDIA's deferred tax assets and liabilities as of fiscal years 2026 and 2025. It reports net deferred tax assets of $13.3 billion (2026) and $11.0 billion (2025), with long-term deferred tax liabilities of $1.8 billion and $886 million, respectively. A valuation allowance of $768 million (2026) and $1.6 billion (2025) is held for capital loss carryforwards and other deferred tax assets. In 2026, $711 million of valuation allowance was released due to improved realizability. The company has net operating loss carryforwards: U.S. federal ($747 million), state ($427 million), and foreign ($503 million), with federal and state losses expiring from 2027. Research tax credit carryforwards include $56 million federal (expiring 2027) and $1.4 billion state (mostly California, indefinite). Federal capital loss carryforwards total $902 million (expiring 2028). Unrecognized tax benefits are $3.7 billion, with accrued interest and penalties of $374 million. Tax examinations are ongoing in multiple jurisdictions.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 14 - Shareholders' Equity", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 14 - Shareholders' Equity", + "level": 4, + "summary": "Capital Return Program On August 26, 2025, our Board of Directors approved an additional \\$60.0 billion in share repurchase authorization, without expiration. In fiscal years 2026 and 2025, we repurchased 282 million and 310 million shares of our common stock for \\$40.4 billion and \\$34.0 billion, respectively. As of January 25, 2026, we were authorized, subject to certain specifications, to repurchase up to \\$58.5 billion of our common stock. From January 26, 2026 through February 20, 2026, we repurchased 8 million shares for \\$1.5 billion pursuant to a pre-established trading plan. In fiscal years 2026, 2025, and 2024, we paid cash dividends to our shareholders of \\$974 million, \\$834 million, and \\$395 million, respectively. The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 15 - Employee Retirement Plans", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 15 - Employee Retirement Plans", + "level": 4, + "summary": "We provide tax-qualified defined contribution plans to eligible employees in the U.S. and certain other countries. Our contribution expense for fiscal years 2026, 2025, and 2024 was \\$442 million, \\$314 million, and \\$255 million, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 16 - Segment Information", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 16 - Segment Information", + "level": 4, + "summary": "Our Chief Executive Officer is our chief operating decision maker, or CODM, and reviews financial information presented on an operating segment basis for purposes of making decisions and assessing financial performance. Our CODM assesses operating performance of each segment based on regularly provided segment revenue and segment operating income. Operating results by segment include costs or expenses directly attributable to each segment, and costs or expenses that are leveraged across our unified architecture and therefore allocated between our two segments. Our", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_11", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_11", + "level": 4, + "summary": "NVIDIA reports two segments: Compute & Networking (Data Center, AI, Automotive) and Graphics (GeForce, Quadro GPUs). The CODM reviews expenses on a consolidated basis; certain costs like stock-based compensation and corporate infrastructure are not allocated. No intersegment transactions exist. Depreciation and amortization for Compute & Networking were $1.6B (2026), $732M (2025), $457M (2024); for Graphics: $590M (2026), $372M (2025), $307M (2024). Acquisition-related intangible amortization is not allocated.", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_12", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_12", + "level": 4, + "summary": "The passage discusses NVIDIA's revenue and segment operating income reconciliation for fiscal years 2024-2026. Revenue is reported based on customer headquarters location, with a change in fiscal 2026. Non-US revenue was 31%, 41%, and 48% of total for 2026, 2025, and 2024 respectively. Revenue concentration among direct customers is noted: in fiscal 2026, one direct customer accounted for 22% of total revenue and another for 14%, primarily in Compute & Networking. Similar concentrations are reported for prior years. The text also defines direct and indirect customers and notes that 76% of Data Center revenue from Taiwan-headquartered customers was attributed to US and European end customers in fiscal 2026.", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_13", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_13", + "level": 4, + "summary": "(Continued) The following table summarizes revenue by specialized markets: [tables/table-109 NVIDIA Revenue.html] The following table presents summarized information for long-lived assets by country. Long-lived assets consist of property and equipment and exclude other assets, operating lease assets, goodwill, and intangible assets. [tables/table-110 Long-lived assets.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 17 - Leases", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/Note 17 - Leases", + "level": 4, + "summary": "Our lease obligations primarily consist of operating leases for our offices and data centers, with lease periods expiring between fiscal years 2027 and 2041. Future minimum lease obligations under our non-cancelable lease agreements as of January 25, 2026 were as follows: [tables/table-111 Lease Obligations.html] Between fiscal years 2027 and 2030, we expect to commence leases with future obligations of \\$22.7 billion, primarily data center leases to support our research and development efforts, with lease terms of 1.8 to 20 years.", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_14", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 15. Exhibits and Financial Statement Schedules/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements/NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements_14", + "level": 4, + "summary": "The passage details NVIDIA's operating lease costs for fiscal years 2026, 2025, and 2024 ($462M, $356M, $269M), with short-term and variable costs not significant. As of January 25, 2026, leases have a weighted average term of 8.8 years and discount rate of 4.38%. It also includes Schedule II on valuation and qualifying accounts, with notes on additions and deductions for allowances and returns. Finally, an Exhibit Index lists exhibits filed with the SEC, including certifications and management contracts, with instructions for shareholder requests.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Item 16. Form 10-K Summary", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Item 16. Form 10-K Summary", + "level": 2, + "summary": "Not Applicable.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Signatures", + "path": "2026-Annual-Report-Web.pdf/Part III_2/Signatures", + "level": 2, + "summary": "The document is a signing page for an SEC filing by NVIDIA Corporation, dated February 25, 2026. It includes the signature of Jen-Hsun Huang as President and CEO, and a Power of Attorney appointing Huang and Colette M. Kress as attorneys-in-fact to sign amendments. The table lists signatories with their titles and dates.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Corporate Information", + "path": "2026-Annual-Report-Web.pdf/Corporate Information", + "level": 1, + "summary": "", + "chunk_count": 26, + "children": [ + { + "title": "Board of Directors", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors", + "level": 2, + "summary": "", + "chunk_count": 11, + "children": [ + { + "title": "Jensen Huang", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Jensen Huang", + "level": 3, + "summary": "Founder, President, and Chief Executive Officer NVIDIA Corporation", + "chunk_count": 1, + "children": [] + }, + { + "title": "Tench Coxe", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Tench Coxe", + "level": 3, + "summary": "Former Managing Director Sutter Hill Ventures", + "chunk_count": 1, + "children": [] + }, + { + "title": "John O. Dabiri", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/John O. Dabiri", + "level": 3, + "summary": "Centennial Professor of Aeronautics and Mechanical Engineering California Institute of Technology", + "chunk_count": 1, + "children": [] + }, + { + "title": "Dawn Hudson", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Dawn Hudson", + "level": 3, + "summary": "Former Chief Marketing Officer National Football League, and Former Chief Executive Officer Pepsi-Cola North America", + "chunk_count": 1, + "children": [] + }, + { + "title": "Harvey C. Jones", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Harvey C. Jones", + "level": 3, + "summary": "Managing Partner Square Wave Ventures", + "chunk_count": 1, + "children": [] + }, + { + "title": "Melissa B. Lora", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Melissa B. Lora", + "level": 3, + "summary": "Former President Taco Bell International", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stephen C. Neal (Lead Director)", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Stephen C. Neal (Lead Director)", + "level": 3, + "summary": "Chairman Emeritus and Senior Counsel Cooley LLP", + "chunk_count": 1, + "children": [] + }, + { + "title": "A. Brooke Seawell", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/A. Brooke Seawell", + "level": 3, + "summary": "Venture Partner New Enterprise Associates", + "chunk_count": 1, + "children": [] + }, + { + "title": "Aarti Shah", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Aarti Shah", + "level": 3, + "summary": "Former Senior Vice President and Chief Information and Digital Officer Eli Lilly and Company", + "chunk_count": 1, + "children": [] + }, + { + "title": "Mark A. Stevens", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Board of Directors/Mark A. Stevens", + "level": 3, + "summary": "Managing Partner S-Cubed Capital", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Founders", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Founders", + "level": 2, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Jensen Huang", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Founders/Jensen Huang", + "level": 3, + "summary": "Founder, President, and Chief Executive Officer", + "chunk_count": 1, + "children": [] + }, + { + "title": "Chris A. Malachowsky", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Founders/Chris A. Malachowsky", + "level": 3, + "summary": "Founder and NVIDIA Fellow", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Executive Team", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Executive Team", + "level": 2, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "Colette M. Kress", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Executive Team/Colette M. Kress", + "level": 3, + "summary": "Executive Vice President and Chief Financial Officer", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jay Puri", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Executive Team/Jay Puri", + "level": 3, + "summary": "Executive Vice President Worldwide Field Operations", + "chunk_count": 1, + "children": [] + }, + { + "title": "Debora Shoquist", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Executive Team/Debora Shoquist", + "level": 3, + "summary": "Executive Vice President Operations", + "chunk_count": 1, + "children": [] + }, + { + "title": "Timothy S. Teter", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Executive Team/Timothy S. Teter", + "level": 3, + "summary": "Executive Vice President General Counsel and Secretary", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Independent Accountants", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Independent Accountants", + "level": 2, + "summary": "PricewaterhouseCoopers LLP 3155 Olsen Drive, Suite 700 San Jose, California 95117", + "chunk_count": 1, + "children": [] + }, + { + "title": "General Legal Counsel", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/General Legal Counsel", + "level": 2, + "summary": "Cooley LLP 3175 Hanover Street Palo Alto, California 94304", + "chunk_count": 1, + "children": [] + }, + { + "title": "Transfer Agent and Registrar", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Transfer Agent and Registrar", + "level": 2, + "summary": "Computershare P.O Box 43006 Providence, Rhode Island 02940 www.computershare.com/investor", + "chunk_count": 1, + "children": [] + }, + { + "title": "Annual Meeting", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Annual Meeting", + "level": 2, + "summary": "June 24, 2026, at 09:00 a.m. PT Online at: www.virtualshareholder meeting.com/NVDA2026", + "chunk_count": 1, + "children": [] + }, + { + "title": "Form 10-K", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Form 10-K", + "level": 2, + "summary": "A copy of NVIDIA's Form 10-K filed with the SEC will be made available to all shareholders at no charge. The Form 10-K also can be accessed through the SEC website at www.sec.gov, or through NVIDIA's Investor Relations website at investor.nvidia.com. To receive a copy by mail, please contact:", + "chunk_count": 1, + "children": [] + }, + { + "title": "Investor Relations", + "path": "2026-Annual-Report-Web.pdf/Corporate Information/Investor Relations", + "level": 2, + "summary": "NVIDIA Corporation 2788 San Tomas Expressway Santa Clara, California 95051 shareholdermeeting@nvidia.com NVIDIA Corporation
0 [images/image-1-NVIDIA Logo.jpg]
natural_image Abstract 3D geometric logo with green spiral and hexagonal shapes on a grid background (no text or symbols)
", + "chunk_count": 1, + "children": [] + } + ] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-NVIDIA Logo.jpg", + "summary": "image-1 - Subject: 3D rendering of the NVIDIA corporate logo (green eye symbol) - Visual Style: High-gloss, metallic finish with a vibrant lime green color - Background: White geometric grid pattern composed of triangular facets - Perspective: Angled, close-up view emphasizing depth and texture" + }, + { + "path": "images/image-2-AI Technology Stack.jpg", + "summary": "image-2 - Applications - Models - Infrastructure - Chips - Energy" + }, + { + "path": "images/image-3-EINFRASTRU.jpg", + "summary": "image-3 EINFRASTRUCTURE] --> F CHATBOTS0" + }, + { + "path": "images/image-4-ENERGY --_.jpg", + "summary": "image-4 ENERGY --> H DIGITAL BIOLOGY0" + }, + { + "path": "images/image-5-__details_.jpg", + "summary": "image-5
ROBOTAXI0" + }, + { + "path": "images/image-6-CHATBOTS E.jpg", + "summary": "image-6 CHATBOTS ENTERPRISE0" + }, + { + "path": "images/image-7-ROBOTAXI S.jpg", + "summary": "image-7 ROBOTAXI SCIENCE0" + }, + { + "path": "images/image-8-ENTERPRISE.jpg", + "summary": "image-8 ENTERPRISE ROBOTICS0" + }, + { + "path": "images/image-9-AI AGENTS .jpg", + "summary": "image-9 AI AGENTS MANUFACTURING0" + }, + { + "path": "images/image-10-SCIENCE AI.jpg", + "summary": "image-10 SCIENCE AI CODER0" + }, + { + "path": "images/image-11-Acronyms in AI and Machine Learning.jpg", + "summary": "image-11 LLM (Large Language Model) VLM (Vision Language Model) VLA (Vision-Language-Action) MMLLM (Multimodal Large Language Model) GPT (Generative Pre-trained Transformer) DM (Diffusion Model) GNN (Graph Neural Network) MOE (Mixture of Experts) SSM (State Space Model) LBM (Lattice Boltzmann Method)" + }, + { + "path": "images/image-12-Stacked Server Racks.jpg", + "summary": "image-12 - Three-tier arrangement of server racks or storage units - Bottom tier: 12 visible units with illuminated vertical panels - Middle tier: 8 visible units, slightly offset from bottom row - Top tier: 6 visible units, centered above middle row - Units feature horizontal ventilation slots and glowing amber/yellow indicators - Dark background with white outline sketch style" + }, + { + "path": "images/image-13-DSX Architecture Overview.jpg", + "summary": "image-13 - **Reference Designs, Best Practices, Methodologies**: Includes Rubin DSX Reference Designs, DSX Ecosystem, Omniverse DSX Blueprint, DSX Max-Q, DSX Flex, DSX Exchange, and DSX Sim. - **Chips, Systems, Facilities**: Features Power and Liquid Cooling components. - **DSX Libraries, APIs, Software**: Integrated with the architecture." + }, + { + "path": "images/image-14-NVIDIA CEO InferenceMAX King.jpg", + "summary": "image-14 - Subject: Jensen Huang (CEO of NVIDIA) standing on stage with arms raised in a victory pose. - Attire: Black leather jacket, black t-shirt, black pants, and glasses. - Background: A large digital display showing a championship belt graphic. - Belt Details: - Central Plate: Features the text \"semianalysis\", \"InferenceMAX\", and \"KING\" with a crown icon above it. - Left Plate: Displays the NVIDIA logo. - Right Plate: Displays a logo resembling a lion's head (likely representing Semianalysis)." + }, + { + "path": "images/image-15-NVIDIA Leads Open-Source AI.jpg", + "summary": "image-15 Chart Title: NVIDIA is World's Largest Contributor to Open-Source AI Y-Axis Label: Repos (Repository count) X-Axis Time Range: Jan 2025 to Feb 2026 Key Data Points & Rankings: 1. NVIDIA: Highest contributor, reaching over 400 repos by Feb 2026. 2. Alibaba Cloud: Second highest, reaching approximately 350 repos. 3. Hugging Face: Third highest, reaching approximately 200 repos. 4. Ai2: Fourth highest, reaching approximately 200 repos. 5. Meta: Reaching approximately 150 repos. 6. Google: Reaching approximately 120 repos. 7. IBM: Reaching approximately 110 repos. 8. ByteDance: Reaching approximately 100 repos. 9. Baidu: Reaching approximately 90 repos. 10. Tencent: Reaching approximately 80 repos. 11. Microsoft: Reaching approximately 80 repos. 12. Z" + }, + { + "path": "images/image-16-Collection of decorative enamel pins.jpg", + "summary": "image-16 - DNA double helix structure with chemical formula labels (H3C, CH3, NH2) - Globe showing North and South America - Cartoon character waving (wearing glasses and jacket) - Spiral galaxy or atom model design - Red and gold robot figure waving - Front view of a car with blue and yellow accents" + }, + { + "path": "images/image-17-Nemoclaw Agentic AI Architecture.jpg", + "summary": "image-17 Central Node: Nemoclaw (Agentic AI) Input/Interaction Layer: - Multi-modal Prompt - Files (Structured, Unstructured) - Computer Use - Tools (CLI, MCP) - OpenShell Core Components & Connections: - Memory - LLM (Large Language Model) - Sub-Agents - Skills Left-Side Infrastructure (Hardware/Data): - cuDF, cuVS - vGPU Right-Side Software Stack: - Nemotron, Nemo, Dynamo, NIM - AI-Q - cuOPT" + }, + { + "path": "images/image-18-Future Robotics Exhibition.jpg", + "summary": "image-18 A panoramic view of a high-tech exhibition hall featuring an extensive collection of advanced robotics and machinery. **Heavy Machinery & Industrial Equipment:** - A large yellow CAT excavator on the far left. - Multiple industrial robotic arms, including white, yellow, and orange articulated arms on the right side. **Humanoid Robots:** - A diverse array of humanoid robots displayed throughout the scene: - Several standing on the upper balcony level. - Various models (silver, white, black) arranged on the main floor stage. - Smaller units and service robots interspersed among the larger figures. - One distinct orange humanoid robot near the center-right foreground. **Vehicles & Drones:** - Three modern luxury cars parked on the right side (Silver Mercedes-Benz, Grey sports car, Blue sedan). - A small drone flying in the central area above the stage." + }, + { + "path": "images/image-19-Signature.jpg", + "summary": "image-19 Jonathan King" + }, + { + "path": "images/image-20-Jensen Huang Presentation.jpg", + "summary": "image-20 - Subject: Jensen Huang (CEO of NVIDIA) - Attire: Signature black leather jacket, glasses, dark trousers - Action: Gesturing with both hands while holding a presentation clicker/remote - Context: Appears to be speaking on stage during a keynote or product presentation against a dark background" + }, + { + "path": "images/image-21-Poster Gallery Event.jpg", + "summary": "image-21 Location: Jose Civic Signage: \"Poster Gallery\", \"Reimagine What's Possible\" Scene: Outdoor gathering with a crowd of people in front of a green exhibition structure." + }, + { + "path": "images/image-22-Inference At-Scale is Extreme Computing.jpg", + "summary": "image-22 - Chart Title: Inference At-Scale is Extreme Computing - Y-Axis Label: Throughput (Tokens per Second For Factory) - X-Axis Label: Smart AI Fast Response (Tokens per Second For Cloud) - Key Data Points & Trends: - Revenue Curve: Shows a high-throughput curve where revenue increases with throughput before declining. - FLOPS Factors: An arrow points to the curve labeled \"FLOPS, HBM Memory, HBM Bandwidth, Software and Architecture,\" indicating these are key drivers for performance." + }, + { + "path": "images/image-23-NVIDIA GTC Event Scene.jpg", + "summary": "image-23 - Primary Subject: Outdoor scene at the NVIDIA GTC (GPU Technology Conference) event featuring attendees and autonomous mobile robots. - Key Objects: Two small, black, wheeled robots are visible in the foreground/mid-ground. One robot has a sign reading \"Welcome to NVIDIA GTC\" with the NVIDIA logo. The other robot has a tall green banner attached with the NVIDIA logo and a QR code. - People: Several individuals are present, wearing conference lanyards (green). Some are looking at phones, while others appear to be interacting with or observing the robots. - Background/Context: The setting is an outdoor plaza or park area with trees, white tents, and string lights in the background, suggesting an evening or late afternoon atmosphere." + }, + { + "path": "images/image-24-Silhouette of Man Gazing at Earth.jpg", + "summary": "image-24 - Primary Subject: A silhouette of a man standing in profile on the left side, looking out towards a large display. - Scene Description: The background is a massive screen or projection showing a view of planet Earth from space. - Visual Details: The image depicts the curvature of the Earth with blue oceans and white cloud formations. There are patches of yellow and green overlaying the clouds, suggesting weather data or satellite imagery. The top left corner shows the blackness of space and the atmospheric glow along the horizon." + }, + { + "path": "images/image-25-Virtual me.jpg", + "summary": "image-25 Virtual meeting admission: We will be holding the 2026 Meeting virtually at the location listed above. To participate, you will need the Control Number included on your notice of proxy materials or printed proxy card. Timothy S. Teter0" + }, + { + "path": "images/image-26-Server Rack Hardware.jpg", + "summary": "image-26 - Multiple server modules installed in a rack chassis - Vertical gold-colored connectors visible on each module faceplate - Black ventilation grilles located at the top of the unit - Industrial data center or telecommunications equipment" + }, + { + "path": "images/image-27-NVIDIA RTX 4090.jpg", + "summary": "image-27 Product: Graphics Card (GPU) Model: GeForce RTX 4090 Brand: NVIDIA Visual Features: Dual-fan cooling system, black and silver chassis design. Background: Abstract green wave pattern typical of NVIDIA branding." + }, + { + "path": "images/image-28-Photo of a laptop workspace.jpg", + "summary": "image-28 Primary subject: A silver laptop displaying a terminal window with code or command-line text. Notable features: - An orange fish figurine (resembling Nemo) sitting on the left side of the keyboard. - A pair of eyeglasses resting on the desk to the left. - A gold-colored external hard drive or speaker to the right. - The laptop screen shows a dark interface with green and white text, likely a coding environment or system logs. - Background is blurred, suggesting an office or home setting." + }, + { + "path": "images/image-29-Automotive X-Ray Visualization.jpg", + "summary": "image-29 Primary subject: A modern sedan vehicle shown in a semi-transparent, wireframe style. Visual details: The image highlights the internal structure and components of the car using glowing cyan lines against a black background. Visible features: The chassis, wheels, suspension system, engine bay area, and cabin frame are clearly outlined to demonstrate the vehicle's architecture." + }, + { + "path": "images/image-30-Growth Over Time.jpg", + "summary": "image-30 1-Year: 32% 3-Year: 822% 5-Year: 1,349%" + }, + { + "path": "images/image-31-Financial Returns by Period.jpg", + "summary": "image-31 - Chart Title: Dividends and Stock Repurchases - Categories (X-Axis): 1-Year, 3-Year, 5-Year - Legend: Purple = Dividends, Green = Stock Repurchases Data Points: - 1-Year Period: - Total Return: $41.4 - Stock Repurchases: $40.4 - Dividends: $1.0 - 3-Year Period: - Total Return: $86.3 - Stock Repurchases: $84.1 - Dividends: $2.2 - 5-Year Period: - Total Return: $97.2 - Stock Repurchases: $94.2 - Dividends: $3.0" + }, + { + "path": "images/image-32-TENCH COXE.jpg", + "summary": "image-32 TENCH COXE Financial/Financial0" + }, + { + "path": "images/image-33-Tench Coxe.jpg", + "summary": "image-33 Tench Coxe was a Managing Director of Sutter Hill Ventures, a venture capital investment firm, from 1989 to 2020, where he focused on investments in the IT sector. Prior to joining Sutter Hill Ventures in 1987, he was Director of Marketing and MIS at Digital Communication Associates. He was a director of Artisan Partners Asset Management Inc., an institutional money management firm from 1995 to 2025. Mr. Coxe holds a BA degree in Economics from Dartmouth College and an MBA degree from Harvard Business School. Governance & Public0" + }, + { + "path": "images/image-34-Financial_.jpg", + "summary": "image-34 Financial/Financial Emerging Technologies & Business Models0" + }, + { + "path": "images/image-35-Governance.jpg", + "summary": "image-35 Governance & Public Human Capital0" + }, + { + "path": "images/image-36-JOHN O. DA.jpg", + "summary": "image-36 JOHN O. DABIRI Industry & Technical0" + }, + { + "path": "images/image-37-John O. Da.jpg", + "summary": "image-37 John O. Dabiri is the Centennial Professor of Aeronautics and Mechanical Engineering at the California Institute of Technology. He is a recipient of the National Medal of Science, a MacArthur Foundation \"Genius Grant,\" the National Science Foundation Alan T. Waterman Award, and the Presidential Early Career Award for Scientists and Engineers. He heads the Dabiri Lab, which conducts research at the intersections of fluid mechanics, energy and environment, and biology. From 2021 to January 2025, he served on President Biden's Council of Advisors on Science and Technology (PCAST) and Energy Secretary Granholm's Energy Advisory Board (SEAB). From 2015 to 2019, he served as a Professor of Civil and Environmental Engineering and of Mechanical Engineering at Stanford University, where he was recognized with the Eugene L. Grant Award for Excellence in Teaching. From 2005 to 2015, he was a Professor of Aeronautics and Bioengineering at the California Institute of Technology, during which time he also served as Director of the Center for Bioinspired Wind Energy, Chair of the Faculty, and Dean of Students. Dr. Dabiri is a Fellow of the American Physical Society, where he previously served as Chair of the Division of Fluid Dynamics. He also serves on the board of trustees of the Gordon and Betty Moore Foundation and previously served as a member of the National Academies' Committee on Science, Technology, and Law. Dr. Dabiri holds a PhD degree in Bioengineering and an MS degree in Aeronautics from the California Institute of Technology, and a BSE degree summa cum laude in Mechanical and Aerospace Engineering from Princeton University. Emerging Technologies & Business Models0" + }, + { + "path": "images/image-38-Jen-Hsun H.jpg", + "summary": "image-38 Jen-Hsun Huang founded NVIDIA in 1993 and has served since its inception as President, Chief Executive Officer, and a member of the Board. Senior Leadership & Operations Experience0" + }, + { + "path": "images/image-39-Mr. Huang .jpg", + "summary": "image-39 Mr. Huang has been elected to the National Academy of Engineering and in 2026 was appointed to the President's Council of Advisors of Science and Technology. He is a recipient of the Semiconductor Industry Association's highest honor, the Robert N. Noyce Award; the IEEE Founder's Medal; the Dr. Morris Chang Exemplary Leadership Award; and honorary doctorate degrees from Taiwan's National Chiao Tung University, National Taiwan University, Oregon State University, Huazhong University of Science and Technology and Linköping University. He has been named the world's best CEO by Fortune, the Economist, and Brand Finance, as well as one of TIME magazine's 100 most influential people. Industry & Technical0" + }, + { + "path": "images/image-40-Mr. Huang .jpg", + "summary": "image-40 Mr. Huang is one of the technology industry's most respected executives, having taken NVIDIA from a startup to a world leader in accelerated computing. Under his guidance, NVIDIA has compiled a record of consistent innovation and sharp execution, marked by products that have gained strong market share. Financial/Financial0" + }, + { + "path": "images/image-41-Industry &.jpg", + "summary": "image-41 Industry & Technical Governance & Public Company Board0" + }, + { + "path": "images/image-42-Financial_.jpg", + "summary": "image-42 Financial/Financial Emerging Technologies & Business Models0" + }, + { + "path": "images/image-43-Community .jpg", + "summary": "image-43 Community Marketing,0" + }, + { + "path": "images/image-44-Emerging T.jpg", + "summary": "image-44 Emerging Technologies & Business Models Regulatory, Legal & Risk Management0" + }, + { + "path": "images/image-45-Marketing,.jpg", + "summary": "image-45 Marketing, Human Capital0" + }, + { + "path": "images/image-38-Jen-Hsun H.jpg_2", + "summary": "image-38 Jen-Hsun Huang founded NVIDIA in 1993 and has served since its inception as President, Chief Executive Officer, and a member of the Board. Senior Leadership & Operations Experience0" + }, + { + "path": "images/image-46-Financial .jpg", + "summary": "image-46 Financial Expert Financial/Financial Community0" + }, + { + "path": "images/image-47-Ms. Hudson.jpg", + "summary": "image-47 Ms. Hudson brings to the board experience in executive leadership. As a longtime marketing executive, she has valuable expertise and insights in leveraging brands, brand development and consumer behavior. She also has considerable corporate governance experience, gained from more than two decades of serving on the boards of public companies. Governance & Public0" + }, + { + "path": "images/image-48-Financial_.jpg", + "summary": "image-48 Financial/Financial Community Marketing,0" + }, + { + "path": "images/image-49-Company Bo.jpg", + "summary": "image-49 Company Board Human Capital Management Experience0" + }, + { + "path": "images/image-50-Other Curr.jpg", + "summary": "image-50 Other Current Public Company Boards: None Senior Leadership & Operations Experience0" + }, + { + "path": "images/image-51-Financial .jpg", + "summary": "image-51 Financial Expert Industry & Technical0" + }, + { + "path": "images/image-52-Mr. Jones .jpg", + "summary": "image-52 Mr. Jones brings to the board an executive management background, an understanding of semiconductor technologies and complex system design. He provides valuable insight into innovation strategies, research and development efforts, as well as management and development of our technical employees. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. Financial/Financial0" + }, + { + "path": "images/image-53-Industry &.jpg", + "summary": "image-53 Industry & Technical Governance & Public0" + }, + { + "path": "images/image-54-Financial_.jpg", + "summary": "image-54 Financial/Financial Emerging Technologies & Business Models0" + }, + { + "path": "images/image-55-Governance.jpg", + "summary": "image-55 Governance & Public Marketing,0" + }, + { + "path": "images/image-56-Emerging T.jpg", + "summary": "image-56 Emerging Technologies & Business Models Human Capital0" + }, + { + "path": "images/image-57-### A. BRO.jpg", + "summary": "image-57 ### A. BROOKE SEAWELL Senior Leadership & Operations Experience0" + }, + { + "path": "images/image-46-Financial .jpg_2", + "summary": "image-46 Financial Expert Financial/Financial Community0" + }, + { + "path": "images/image-58-Senior Lea.jpg", + "summary": "image-58 Senior Leadership & Operations Experience Governance & Public0" + }, + { + "path": "images/image-59-Financial_.jpg", + "summary": "image-59 Financial/Financial Emerging Technologies & Business Models0" + }, + { + "path": "images/image-60-Governance.jpg", + "summary": "image-60 Governance & Public Regulatory, Legal & Risk Management0" + }, + { + "path": "images/image-61-Company Bo.jpg", + "summary": "image-61 Company Board Human Capital0" + }, + { + "path": "images/image-62-AARTI SHAH.jpg", + "summary": "image-62 AARTI SHAH Senior Leadership & Operations Experience0" + }, + { + "path": "images/image-63-Aarti Shah.jpg", + "summary": "image-63 Aarti Shah serves on the boards of various companies and non-profit organizations. Dr. Shah worked at Eli Lilly and Company for 27 years until 2021 and served in several functional and business leadership roles, most recently as Senior Vice President and Chief Information and Digital Officer, as well as Senior Statistician, Research Scientist, Vice President for Biometrics, and Global Brand Development Leader in Lilly's Bio-Medicines business unit. Dr. Shah has served on the board and as a member of the science, innovation & development committee and human capital and ESG committee of Sandoz Group AG, a pharmaceutical company, since 2023. Dr. Shah has served on the board of trustees and as a member of the audit committee and distribution and technology committee of Northwestern Mutual since 2020, a financial services organization. Dr. Shah serves on the board, strategic planning committee, and global and technology committee of St. Jude Children's Research Hospital and American Lebanese Syrian Associated Charities (ALSAC), a non-profit organization. She also serves as the secretary and a trustee of the non-profit organization, Shrimad Rajchandra Mission Dharampur USA. She served on the Indianapolis Public Library Foundation board for the full term of 9 years and on the Center for Interfaith Cooperation for the full term of 4 years. Dr. Shah received her BS and MS degrees in Statistics and Mathematics in India before completing her PhD degree in Applied Statistics from the University of California, Riverside. Industry & Technical0" + }, + { + "path": "images/image-64-Dr. Shah b.jpg", + "summary": "image-64 Dr. Shah brings to the Board executive leadership and senior operating experience. Additionally she brings expertise in drug development and technical expertise in the areas of information technology, cybersecurity, advanced analytics, data sciences, and digital health. Governance & Public0" + }, + { + "path": "images/image-65-Industry &.jpg", + "summary": "image-65 Industry & Technical Emerging Technologies & Business Models0" + }, + { + "path": "images/image-66-Governance.jpg", + "summary": "image-66 Governance & Public Marketing,0" + }, + { + "path": "images/image-67-Emerging T.jpg", + "summary": "image-67 Emerging Technologies & Business Models Regulatory, Legal & Risk Management0" + }, + { + "path": "images/image-68-Marketing,.jpg", + "summary": "image-68 Marketing, ### MARK A. STEVENS0" + }, + { + "path": "images/image-69-Committees.jpg", + "summary": "image-69 Committees: CC, NCGC Industry & Technical0" + }, + { + "path": "images/image-70-#### Indep.jpg", + "summary": "image-70 #### Independent Director Financial/Financial Community0" + }, + { + "path": "images/image-71-Mr. Steven.jpg", + "summary": "image-71 Mr. Stevens brings to the Board a deep understanding of the technology industry, and the drivers of structural change and high-growth opportunities. He provides valuable insight regarding corporate strategy development and the analysis of acquisitions and divestitures. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. Governance & Public0" + }, + { + "path": "images/image-72-Financial_.jpg", + "summary": "image-72 Financial/Financial Community Emerging Technologies & Business Models0" + }, + { + "path": "images/image-73-Compensation Structure Breakdown.jpg", + "summary": "image-73 - Chart Type: Stacked Bar Chart - Components (from left to right): - SY PSUs: 44% - MY PSUs: 44% - Variable Cash: 8% - Base Salary: 4%" + }, + { + "path": "images/image-74-Compensation Structure Breakdown.jpg", + "summary": "image-74 - Chart Title: Compensation Composition (SY PSUs, MY PSUs, RSUs, Variable Cash, Base Salary) - Data Source/Context: Likely represents a total compensation mix for an employee or role. - Component Breakdown: - SY PSUs (Short-term Performance Units): 23% - MY PSUs (Medium-term Performance Units): 23% - RSUs (Restricted Stock Units): 46% - Variable Cash: 2% - Base Salary: 6%" + }, + { + "path": "images/image-75-Executive Compensation Timeline.jpg", + "summary": "image-75 December 2024: CC determined peer companies December 2024 - January 2025: Members of management and the Board, including our Lead Director and CC Chair, engaged in stockholder outreach March 2025: CC determined performance goals and approved target compensation levels and PSU and RSU grants, considering stockholder feedback and peer company practices March 2026: CC certified achievement and payouts for Fiscal 2026 Variable Cash Plan, SY PSUs granted in Fiscal 2026 and MY PSUs granted in Fiscal 2024* May 2026: CC oversaw compensation risk assessment; published executive compensation program details in proxy statement" + }, + { + "path": "images/image-76-NVIDIA Salary Percentiles.jpg", + "summary": "image-76 75th Percentile: $229.70 NVIDIA: $113.27 Median: $54.03 25th Percentile: $40.93" + }, + { + "path": "images/image-77-NVIDIA vs Percentile Data.jpg", + "summary": "image-77 Chart Title: Not specified Y-Axis Label: Currency ($) ranging from $0 to $3,500 X-Axis Categories: NVIDIA, 75th Percentile, Median, 25th Percentile Data Points: - NVIDIA: $3,471.92 - 75th Percentile: $1,913.24 - Median: $536.35 - 25th Percentile: $242.70" + }, + { + "path": "images/image-78-NEO CAP versus TSR.jpg", + "summary": "image-78 Chart Title: NEO CAP versus TSR Data Series: - CEO CAP (Dark Grey Bars) - Other NEOs Average CAP (Light Grey Bars) - NVIDIA TSR (Green Line) - Nasdaq 100 Index TSR (Purple Line) Time Range: Fiscal Years 2022 through 2026 Key Data Points (CAP in millions): - Fiscal 2022: CEO CAP $105.5; Other NEOs Average CAP $38.5 - Fiscal 2023: CEO CAP ($4.1); Other NEOs Average CAP ($1.4) - Fiscal 2024: CEO CAP $234.1; Other NEOs Average CAP $85.6 - Fiscal 2025: CEO CAP $344.2; Other NEOs Average CAP $125.6 - Fiscal 2026: CEO CAP $162.2; Other NEOs Average CAP $57.7 Key Data Points (TSR values): - Fiscal 20" + }, + { + "path": "images/image-79-NEO CAP vs Income.jpg", + "summary": "image-79 Chart Title: NEO CAP versus Net Income & Non-GAAP Operating Income Data Series: - CEO CAP (Dark Grey Bars) - Other NEOs Average CAP (Light Grey Bars) - Net Income (Green Line) - Non-GAAP Operating Income (Purple Line) Fiscal 2022 Data: - CEO CAP: $105.5 million - Other NEOs Average CAP: $38.5 million - Net Income: $9,752 million - Non-GAAP Operating Income: $12,690 million Fiscal 2023 Data: - CEO CAP: ($4.1) million (Negative) - Other NEOs Average CAP: ($1.4) million (Negative) - Net Income: $4,368 million - Non-GAAP Operating Income: $9,040 million Fiscal 2024 Data: - CEO CAP: $234.1 million - Other NEOs Average CAP: $85.6 million - Net Income: $29" + }, + { + "path": "images/image-80-Effect of .jpg", + "summary": "image-80 Effect of abstentions: Same as a vote AGAINST.
0" + }, + { + "path": "images/image-81-### Delinq.jpg", + "summary": "image-81 ### Delinquent Section 16(a) Reports Timothy S. Teter0" + }, + { + "path": "images/image-82-ANNUAL REP.jpg", + "summary": "image-82 ANNUAL REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 NVIDIA.0" + }, + { + "path": "images/image-83-NVIDIA Stock Growth.jpg", + "summary": "image-83 Chart Title: NVIDIA Corporation vs S&P 500 vs Nasdaq 100 Performance Time Range: January 31, 2021 to January 25, 2026 Data Series & Key Points: - **NVIDIA Corporation (Green Line):** Shows exponential growth. - Jan 2021: ~100 - Jan 2022: ~180 - Jan 2023: ~170 - Jan 2024: ~470 - Jan 2025: ~1,100 - Jan 2026: ~1,450 - **S&P 500 (Grey Square Marker):** Shows slow, steady linear growth. - Starts at ~100 in Jan 2021 and ends near ~200 in Jan 2026. - **Nasdaq 100 (Purple Triangle Marker):** Shows slow, steady linear growth similar to S&P 500" + }, + { + "path": "images/image-1-NVIDIA Logo.jpg_2", + "summary": "image-1 NVIDIA Corporation
0" + } + ], + "tables": [ + { + "path": "tables/table-0 GPU Cloud Growth.html", + "summary": "table-1 The table shows projected GPU/cloud service growth from Jan 2025 to Feb 2026, with NVIDIA leading at ~450 units, followed by Alibaba Cloud at ~350 and Hugging Face at ~200." + }, + { + "path": "tables/table-1 NVIDIA Definitions.html", + "summary": "table-2 A table of defined terms and acronyms used in NVIDIA's proxy materials, covering compensation plans, committees, and financial metrics." + }, + { + "path": "tables/table-2 Revenue Growth.html", + "summary": "table-3 Revenue reached $215.9 billion, up 65% year on year. Gross margin was 71.1%, down 3.9 points. Operating income rose 60% to $130.4 billion, and diluted EPS increased 67% to $4.90." + }, + { + "path": "tables/table-3 NVIDIA Revenue.html", + "summary": "table-4 NVIDIA's Compute & Networking revenue reached $193.5B (up 67% YoY) and Graphics $22.5B (up 57% YoY), totaling $215.9B (up 65% YoY). Operating income was $139.3B (up 58% YoY)." + }, + { + "path": "tables/table-4 Returns by Period.html", + "summary": "table-5 The table shows investment returns over 1-year (32%), 3-year (822%), and 5-year (1,349%) periods." + }, + { + "path": "tables/table-5 Dividend & Buyback.html", + "summary": "table-6 The table shows dividends and stock repurchases over 1, 3, and 5 years. Dividends increase from $0 to $3.0, while repurchases rise from $40.4 to $94.2." + }, + { + "path": "tables/table-6 Proxy Voting Items.html", + "summary": "table-7 The table lists 7 proposals for a shareholder vote, including director elections, executive compensation, auditor ratification, and stockholder proposals, with board recommendations and voting requirements." + }, + { + "path": "tables/table-7 Board of Directors.html", + "summary": "table-8 The table lists 10 board members of a company, including their age, director start year, independence status, financial expertise, committee roles, and other board memberships." + }, + { + "path": "tables/table-8 Board Skills Matrix.html", + "summary": "table-9 A table showing board members' expertise across eight categories including leadership, industry, finance, governance, emerging tech, marketing, regulatory, and human capital." + }, + { + "path": "tables/table-9 Performance Goals.html", + "summary": "table-10 The table outlines performance goals for Variable Cash Plan, SY PSUs, and MY PSUs with thresholds, base, and stretch targets for Fiscal 2026 Revenue, Non-GAAP Operating Income, and 3-Year Relative TSR." + }, + { + "path": "tables/table-10 Performance Payouts.html", + "summary": "table-11 Table shows performance achievement for Fiscal 2026: $215.9B revenue, $137.3B Non-GAAP Operating Income, 100th percentile TSR. Payouts range from 150% to 200% of target." + }, + { + "path": "tables/table-11 Proxy Voting.html", + "summary": "table-12 The table outlines methods for voting, changing votes, or revoking proxies for the 2026 meeting, including virtual attendance, mail, telephone, online, written notice, or submitting a later proxy card." + }, + { + "path": "tables/table-12 Proposal Voting.html", + "summary": "table-13 Seven proposals including director election, executive compensation, auditor ratification, and stockholder proposals with voting requirements and abstention effects." + }, + { + "path": "tables/table-13 NVIDIA Board.html", + "summary": "table-14 The table lists NVIDIA's board members, including Tench Coxe, Jen-Hsun Huang, and others, with details on age, tenure, independence, financial expertise, and committee roles." + }, + { + "path": "tables/table-14 Board Skills Matrix.html", + "summary": "table-15 A table listing director qualifications: senior leadership, industry, finance, governance, emerging tech, marketing, legal/risk, and human capital management experience." + }, + { + "path": "tables/table-15 Director Profile.html", + "summary": "table-16 A 68-year-old independent director serving since 1993 on Audit and Compensation Committees, with no other current public company board memberships." + }, + { + "path": "tables/table-16 Director Profile.html", + "summary": "table-17 A 46-year-old independent director since 2020, serving on CC and NCGC committees, with no other public company boards." + }, + { + "path": "tables/table-17 Director Profile.html", + "summary": "table-18 Director aged 63, serving since 2023, on Audit and Nominating/Governance committees. Also a Conagra Brands board member since 2019, independent and a financial expert." + }, + { + "path": "tables/table-18 Leadership & Operations.html", + "summary": "table-19 The table outlines senior leadership, operations, financial, governance, emerging technologies, marketing, and human capital management experience areas." + }, + { + "path": "tables/table-19 Director Profile.html", + "summary": "table-20 A 77-year-old independent director since 2019, serves on NCGC committee, no other public boards, and is Lead Director." + }, + { + "path": "tables/table-20 Leadership & Governance.html", + "summary": "table-21 The table lists senior leadership, operations, governance, public company board, marketing, communications, brand management, regulatory, legal, risk management, and human capital management experience." + }, + { + "path": "tables/table-21 Director Profile.html", + "summary": "table-22 A 78-year-old independent director since 1997, serves on the Audit Committee and is a financial expert. Also a board member at Tenable Holdings since 2017." + }, + { + "path": "tables/table-22 Director Profile.html", + "summary": "table-23 Age 61, director since 2020, serves on Audit and Compensation Committees. Also on board of Sandoz Group AG since 2023. Independent director." + }, + { + "path": "tables/table-23 Board Risk Oversight.html", + "summary": "table-24 The Board of Directors oversees major risks including business model, strategy, operations, regulatory, cybersecurity, brand, and enterprise risk management." + }, + { + "path": "tables/table-24 Board Oversight.html", + "summary": "table-25 The table outlines board committee responsibilities: Audit covers financial integrity, risk, cybersecurity, internal audit, and compliance; Compensation covers policies and human capital; Governance covers structure, stockholder concerns, CSR, and trade compliance." + }, + { + "path": "tables/table-25 Director Compensation.html", + "summary": "table-26 The table lists director compensation for 12 individuals, showing cash fees, stock awards, and total amounts, with most receiving $85,000 cash and $278,809 in stock." + }, + { + "path": "tables/table-26 RSU Holdings Table.html", + "summary": "table-27 The table lists board members and their restricted stock units (RSUs). Robert K. Burgess, Persis S. Drell, and Ellen Ochoa have zero RSUs; others hold 900 or 1,880 RSUs." + }, + { + "path": "tables/table-27 Beneficial Ownership.html", + "summary": "table-28 Table lists beneficial ownership of shares for NEOs, directors, and 5% stockholders, including Jen-Hsun Huang (3.58%), BlackRock (7.43%), and Vanguard (7.31%)." + }, + { + "path": "tables/table-28 Compensation Breakdown.html", + "summary": "table-29 The table shows compensation components: SY PSUs (44%), MY PSUs (44%), Variable Cash (8%), and Base Salary (4%)." + }, + { + "path": "tables/table-29 Compensation Mix.html", + "summary": "table-30 The table shows a compensation breakdown: SY PSUs 23%, MY PSUs 23%, RSUs 46%, Variable Cash 6%, Base Salary 2%." + }, + { + "path": "tables/table-30 Fiscal 2026 Goals.html", + "summary": "table-31 Table shows stretch compensation plan goals vs actual performance for fiscal 2026: revenue $160B goal vs $215.9B actual, non-GAAP operating income $96B goal vs $137.3B actual, TSR 75th percentile goal vs 100th percentile actual." + }, + { + "path": "tables/table-31 Tech Companies.html", + "summary": "table-32 The table lists major technology companies including Alphabet, Meta, Amazon, Microsoft, Apple, Netflix, Broadcom, Oracle, Cisco, Qualcomm, IBM, Salesforce, Intel, and Visa with their stock tickers." + }, + { + "path": "tables/table-32 Percentile Values.html", + "summary": "table-33 The table shows financial percentile values: 75th percentile at $229.70, NVIDIA at $113.27, median at $54.03, and 25th percentile at $40.93." + }, + { + "path": "tables/table-33 NVIDIA Value Data.html", + "summary": "table-34 The table shows NVIDIA's value at $3,471.92, with the 75th percentile at $1,913.24, median at $536.35, and 25th percentile at $242.70." + }, + { + "path": "tables/table-34 NEO Compensation.html", + "summary": "table-35 Table details NEO compensation components: fixed base salary, variable cash, and equity (PSUs, RSUs) with performance measures, vesting periods, and maximum earning potentials." + }, + { + "path": "tables/table-35 Compensation Plans.html", + "summary": "table-36 The table outlines three compensation plans: Variable Cash Plan (1-year Revenue metric), SY PSUs (1-year Non-GAAP Operating Income), and MY PSUs (3-year TSR relative to S&P 500). Rationale focuses on driving value, growth, and shareholder alignment." + }, + { + "path": "tables/table-36 Compensation Plan.html", + "summary": "table-37 Table shows Variable Cash Plan and PSUs with Threshold, Base, and Stretch targets for Fiscal 2026 Revenue, Non-GAAP Operating Income, and 3-Year Relative TSR." + }, + { + "path": "tables/table-37 Performance Payouts.html", + "summary": "table-38 Table shows performance achievement for Fiscal 2026: $215.9B revenue, $137.3B Non-GAAP Operating Income, 100th percentile 3-year TSR. Payouts range from 150% to 200% of target." + }, + { + "path": "tables/table-38 CEO Pay Summary.html", + "summary": "table-39 CEO target pay totals $35.5M, up 11% from FY2025. Base salary flat at $1.5M. Variable cash at $3M with 200% payout. Equity up 13% to $31M, with PSUs vesting at 150% of target." + }, + { + "path": "tables/table-39 EVP CFO Pay.html", + "summary": "table-40 EVP & CFO target pay is $16M, up 7% from FY2025. Base salary flat at $900K. Variable cash at 200% payout. SY and MY PSUs at 200% target. RSUs up 7%." + }, + { + "path": "tables/table-40 EVP Compensation.html", + "summary": "table-41 EVP Worldwide Field Operations total target pay is $16M, up 7% from FY2025. Base salary $950K, variable cash $650K, equity $14.4M. Performance-based payouts at 200% for exceeding stretch goals." + }, + { + "path": "tables/table-41 EVP Compensation.html", + "summary": "table-42 EVP Operations target pay $16M, up 19% from FY2025. Base salary flat at $850K. Variable cash flat at $250K, with 200% payout. Equity up 20% to $14.9M. PSUs and RSUs increased." + }, + { + "path": "tables/table-42 Executive Pay.html", + "summary": "table-43 EVP General Counsel target pay $16M, up 19% from FY2025. Base salary $850K, variable cash $250K, equity $14.9M. Performance-based payouts at 200% for exceeding goals." + }, + { + "path": "tables/table-43 Non-GAAP Income.html", + "summary": "table-44 The table shows GAAP operating income rising from $81,453 in Fiscal 2025 to $130,387 in Fiscal 2026, with adjustments for stock-based compensation, acquisition costs, and other items, resulting in Non-GAAP Operating Income of $86,789 and $137,300 respectively." + }, + { + "path": "tables/table-44 Executive Compensation.html", + "summary": "table-45 The table shows compensation for NVIDIA executives including CEO Jen-Hsun Huang, CFO Colette Kress, and others for fiscal years 2024-2026, with salary, stock awards, and total pay." + }, + { + "path": "tables/table-45 PSU Awards Table.html", + "summary": "table-46 The table shows Performance Stock Unit (PSU) awards for five executives across fiscal years 2024-2026, with SY and MY PSU values in dollars." + }, + { + "path": "tables/table-46 Executive Compensation.html", + "summary": "table-47 The table details 2025 executive compensation for five Nvidia executives, including Jen-Hsun Huang, with awards like SY PSU, MY PSU, RSU, and Variable Cash Plan, showing estimated payouts and stock values." + }, + { + "path": "tables/table-47 Stock Awards.html", + "summary": "table-48 The table details unvested stock awards for executives including Jen-Hsun Huang, Colette M. Kress, Ajay K. Puri, Debora Shoquist, and Timothy S. Teter, listing numbers of units and their market values." + }, + { + "path": "tables/table-48 Stock Vesting Data.html", + "summary": "table-49 The table shows stock vesting details for five NVIDIA executives, including Jen-Hsun Huang, with shares acquired and value realized upon vesting." + }, + { + "path": "tables/table-49 Unvested RSUs_PSUs.html", + "summary": "table-50 Table shows unvested RSUs and PSUs for five NVIDIA executives as of January 25, 2026, with Jen-Hsun Huang holding the largest amount." + }, + { + "path": "tables/table-50 SY PSU Grants.html", + "summary": "table-51 Table shows estimated and actual SY PSU grants for fiscal 2026 for five executives, including Jen-Hsun Huang, with values in shares and dollars." + }, + { + "path": "tables/table-51 MY PSU Grants 2024.html", + "summary": "table-52 Table shows estimated and actual MY PSU grants for fiscal 2024 for five NVIDIA executives, including Jen-Hsun Huang with the highest values." + }, + { + "path": "tables/table-52 CEO Pay & Performance.html", + "summary": "table-53 Table shows CEO and non-CEO NEO compensation, total shareholder return, peer group return, net income, and non-GAAP operating income for fiscal years 2022-2026." + }, + { + "path": "tables/table-53 CEO Compensation.html", + "summary": "table-54 The table shows CEO compensation for fiscal year 2026, with a total of $36,343,830 in the Summary Compensation Table, adjustments totaling $150,637,617, and compensation actually paid of $162,180,936." + }, + { + "path": "tables/table-54 Non-CEO NEO Pay.html", + "summary": "table-55 The table shows average compensation for non-CEO NEOs in 2026: Summary Table total $14,422,317, equity adjustments $56,081,694, resulting in actual pay of $57,721,552." + }, + { + "path": "tables/table-55 Financial Metrics.html", + "summary": "table-56 The table lists three key financial metrics: Revenue, Non-GAAP Operating Income, and 3-Year TSR relative to the S&P 500." + }, + { + "path": "tables/table-56 NVIDIA CEO Pay.html", + "summary": "table-57 The table shows CEO and other NEOs' CAP, NVIDIA TSR, and Nasdaq100 Index TSR from Fiscal 2022 to 2026. CEO CAP fluctuated, peaking at $344.2M in Fiscal 2025, while TSR grew significantly." + }, + { + "path": "tables/table-57 CEO Pay & Income.html", + "summary": "table-58 The table shows CEO CAP, other NEOs average CAP, Net Income, and Non-GAAP Operating Income from Fiscal 2022 to 2026, with notable increases in 2024-2025." + }, + { + "path": "tables/table-58 Fee Breakdown.html", + "summary": "table-59 The table shows fiscal 2026 and 2025 fees: Audit $10.2M/$8.1M, Audit Related $1.7M/$0.7M, Tax $1.1M/$0.9M, Other $0.4M/$0.4M, Total $13.4M/$10.0M." + }, + { + "path": "tables/table-59 Equity Plan Table.html", + "summary": "table-60 The table shows equity compensation plans approved by security holders with 192,391,765 securities from outstanding options and 3,551,257,925 remaining for future issuance. No unapproved plans exist." + }, + { + "path": "tables/table-60 Nvidia Executives.html", + "summary": "table-61 The table lists Nvidia's top executives: Jen-Hsun Huang (CEO, 63), Colette Kress (CFO, 58), Ajay Puri (Field Ops, 71), Debora Shoquist (Ops, 71), and Timothy Teter (General Counsel, 59)." + }, + { + "path": "tables/table-61 Stock Buyback.html", + "summary": "table-62 The table shows a stock buyback program from Oct 2025 to Jan 2026, with 19.5 million shares purchased at average prices around $179-$199, leaving $58.5 billion available." + }, + { + "path": "tables/table-62 NVIDIA Stock Growth.html", + "summary": "table-63 The table shows NVIDIA Corporation's stock value rising from 100 in 2021 to 1450 in 2026, outperforming S&P 500 and Nasdaq 100 indices." + }, + { + "path": "tables/table-63 Stock Performance.html", + "summary": "table-64 The table shows cumulative investment returns from 2021 to 2026 for NVIDIA, S&P 500, and Nasdaq 100, with NVIDIA growing from $100 to $1,448.75." + }, + { + "path": "tables/table-64 Financial Results.html", + "summary": "table-65 The table shows financial data for years ending Jan 25, 2026 and Jan 26, 2025. Revenue rose 65% to $215,938M, net income up 65% to $120,067M, and operating income up 60% to $130,387M." + }, + { + "path": "tables/table-65 Income Statement.html", + "summary": "table-66 The table shows a company's income statement for years ending Jan 25, 2026 and Jan 26, 2025. Revenue is 100% both years. Gross profit decreased from 75.0% to 71.1%. Net income slightly dropped from 55.8% to 55.6%." + }, + { + "path": "tables/table-66 Revenue Growth.html", + "summary": "table-67 The table shows revenue for Compute & Networking and Graphics segments for years ended Jan 25, 2026 and Jan 26, 2025, with total revenue increasing 65% to $215,938 million." + }, + { + "path": "tables/table-67 Revenue Growth.html", + "summary": "table-68 The table shows revenue for Compute & Networking and Graphics segments for years ending Jan 2026 and 2025, with total revenue increasing 58% to $139,297 million." + }, + { + "path": "tables/table-68 Operating Expenses.html", + "summary": "table-69 The table shows operating expenses for years ended Jan 25, 2026 and Jan 26, 2025. R&D increased by $5,583M (43%), SG&A by $1,088M (31%), total operating expenses by $6,671M (41%)." + }, + { + "path": "tables/table-69 Other Income.html", + "summary": "table-70 The table shows other income net for years ending Jan 2026 and 2025, with interest income rising from $1,786M to $2,300M and other income surging from $1,034M to $9,022M, resulting in total other income net of $11,063M." + }, + { + "path": "tables/table-70 Cash & Securities.html", + "summary": "table-71 The table shows cash and cash equivalents, marketable securities, and their total for Jan 25, 2026 and Jan 26, 2025, with increases in all categories." + }, + { + "path": "tables/table-71 Cash Flow Table.html", + "summary": "table-72 The table shows net cash provided by operating activities of $102,718M (Jan 2026) vs $64,089M (Jan 2025), net cash used in investing of $52,228M vs $20,421M, and financing of $48,474M vs $42,359M." + }, + { + "path": "tables/table-72 Debt Maturity.html", + "summary": "table-73 The table shows debt maturity schedule as of Jan 25, 2026, with $8,468 million net carrying amount, including $999 million short-term and $7,469 million long-term portion." + }, + { + "path": "tables/table-73 Stock Sale Plans.html", + "summary": "table-74 Two insiders, Director John O. Dabiri and CFO Colette M. Kress, adopted Rule 10b5-1 trading plans to sell shares of common stock, with Dabiri selling 3,984 shares and Kress selling 500,000 shares." + }, + { + "path": "tables/table-74 Income Statement.html", + "summary": "table-75 The table shows financial data for three fiscal years ending Jan 2026, 2025, and 2024. Revenue grew from $60,922 to $215,938, with net income rising from $29,760 to $120,067." + }, + { + "path": "tables/table-75 Comprehensive Income.html", + "summary": "table-76 The table shows net income, other comprehensive income (OCI) from securities and hedges, and total comprehensive income for fiscal years ending Jan 2026, 2025, and 2024." + }, + { + "path": "tables/table-76 Balance Sheet.html", + "summary": "table-77 The balance sheet shows total assets of $206,803 as of Jan 25, 2026, up from $111,601 a year earlier, with increases in cash, marketable securities, and goodwill. Total liabilities rose to $49,510, while shareholders' equity grew to $157,293." + }, + { + "path": "tables/table-77 Equity Changes.html", + "summary": "table-78 The table shows shareholders' equity changes from Jan 2023 to Jan 2026, including net income, stock issuances, repurchases, dividends, and stock-based compensation, with total equity rising from $22,101M to $157,293M." + }, + { + "path": "tables/table-78 Cash Flow Statement.html", + "summary": "table-79 The table shows cash flows for three fiscal years ending Jan 2026, 2025, and 2024. Net cash from operations was $102.7M, $64.1M, and $28.1M, respectively. Investing activities used $52.2M, $20.4M, and $10.6M, while financing used $48.5M, $42.4M, and $13.6M." + }, + { + "path": "tables/table-79 Cost Breakdown.html", + "summary": "table-80 The table shows cost of revenue, R&D, and SG&A expenses for fiscal years ending Jan 2024, 2025, and 2026, with totals increasing from $3,549M to $6,386M." + }, + { + "path": "tables/table-80 Equity Awards.html", + "summary": "table-81 The table shows equity awards (RSUs, PSUs, ESPP) for fiscal years ending Jan 2024-2026, including shares granted/purchased and fair values per share." + }, + { + "path": "tables/table-81 ESPP Option Data.html", + "summary": "table-82 The table shows ESPP option data for fiscal years ending Jan 2024-2026, using the Black-Scholes model. Key inputs include weighted average expected life (0.1-2.0 years), risk-free interest rates (3.5%-5.5%), volatility (26%-96%), and dividend yield (0.03%-0.06%)." + }, + { + "path": "tables/table-82 RSU_PSU Activity.html", + "summary": "table-83 The table shows RSU, PSU, and market-based PSU activity from Jan 26, 2025 to Jan 25, 2026, with shares granted, vested, canceled, and ending balance of 189 million shares at $81.51 weighted average fair value." + }, + { + "path": "tables/table-83 EPS Data.html", + "summary": "table-84 The table shows net income, weighted average shares, and basic/diluted EPS for fiscal years ending Jan 2026, 2025, and 2024." + }, + { + "path": "tables/table-84 Intangible Assets.html", + "summary": "table-85 The table shows intangible assets as of Jan 25, 2026 and Jan 26, 2025, with gross, accumulated amortization, and net carrying amounts for acquisition-related assets, patents, and totals." + }, + { + "path": "tables/table-85 Amortization Expense.html", + "summary": "table-86 The table shows future amortization expenses from 2027 to 2032 and thereafter, totaling $3,306 million, with the highest in 2027 at $923 million." + }, + { + "path": "tables/table-86 Debt Securities.html", + "summary": "table-87 The table details debt securities by pricing category, cost, unrealized gains/losses, fair value, and reporting classification as cash equivalents, marketable securities, or other assets." + }, + { + "path": "tables/table-87 Fair Value Table.html", + "summary": "table-88 The table details investments by pricing category, showing cost, unrealized gains/losses, estimated fair value, and classification as cash equivalents or marketable securities. Total fair value is $42,487 million." + }, + { + "path": "tables/table-88 Debt Securities.html", + "summary": "table-89 The table compares estimated fair value and gross unrealized loss for debt securities (U.S. Treasury, corporate, government agencies) as of Jan 25, 2026 and Jan 26, 2025, with totals of $13,132M and $12,422M respectively." + }, + { + "path": "tables/table-89 Debt Maturity.html", + "summary": "table-90 The table shows debt maturities: $20,427 million due in less than one year, $19,093 million due in 1-5 years, totaling $39,520 million as of Jan 25, 2026." + }, + { + "path": "tables/table-90 Equity Securities.html", + "summary": "table-91 The table shows changes in non-marketable equity securities balances for years ending Jan 25, 2026 and Jan 26, 2025, including net additions, unrealized gains, reclassifications, and impairments." + }, + { + "path": "tables/table-91 Inventory Data.html", + "summary": "table-92 The table shows inventories in millions for Jan 25, 2026 and Jan 26, 2025: raw materials $3,807 vs $3,408, work in process $8,822 vs $3,399, finished goods $8,774 vs $3,273, total $21,403 vs $10,080." + }, + { + "path": "tables/table-92 Property & Equipment.html", + "summary": "table-93 The table shows property and equipment gross totals of $16,970 million (Jan 2026) and $10,684 million (Jan 2025), with accumulated depreciation of $6,587 million and $4,401 million, resulting in net values of $10,383 million and $6,283 million." + }, + { + "path": "tables/table-93 Current Liabilities.html", + "summary": "table-94 The table lists accrued and other current liabilities for Jan 25, 2026 and Jan 26, 2025, totaling $21,352 million and $11,737 million respectively, including customer program accruals, warranty, and purchase obligations." + }, + { + "path": "tables/table-94 Long-Term Liabilities.html", + "summary": "table-95 The table shows other long-term liabilities for Jan 25, 2026 and Jan 26, 2025, including income tax payable, deferred income tax, deferred revenue, and other items, totaling $7,306 million and $4,245 million respectively." + }, + { + "path": "tables/table-95 Deferred Revenue.html", + "summary": "table-96 The table shows deferred revenue balance changes from Jan 2025 to Jan 2026, with additions of $11,137M and revenue recognized of $10,378M, ending at $2,572M." + }, + { + "path": "tables/table-96 Hedge Values.html", + "summary": "table-97 The table shows designated accounting hedges at $1,765M (Jan 2026) and $1,424M (Jan 2025), and non-designated hedges at $2,332M and $1,297M respectively." + }, + { + "path": "tables/table-97 Debt Notes Table.html", + "summary": "table-98 Table lists notes due from 2026 to 2060 with interest rates from 1.55% to 3.70%, carrying amounts, and net carrying value of $8,468M as of Jan 25, 2026." + }, + { + "path": "tables/table-98 Restructuring Reserve.html", + "summary": "table-99 The table shows restructuring reserve balances for fiscal years ending Jan 2026, 2025, and 2024, with additions, utilization, and ending balances in millions." + }, + { + "path": "tables/table-99 Income Tax Expense.html", + "summary": "table-100 The table shows income tax expenses for fiscal years ending Jan 2026, 2025, and 2024, with current and deferred taxes broken down by federal, state, and foreign categories." + }, + { + "path": "tables/table-100 Income Before Tax.html", + "summary": "table-101 The table shows income before income tax for U.S. and Foreign segments for years ending Jan 2026, 2025, and 2024, with totals of $141,450M, $84,026M, and $33,818M respectively." + }, + { + "path": "tables/table-101 Tax Rate Breakdown.html", + "summary": "table-102 The table shows a reconciliation from the US federal statutory tax rate of 21.0% to an effective income tax expense of 15.1%, with adjustments for state taxes, foreign effects, credits, and other items." + }, + { + "path": "tables/table-102 Tax Rate Analysis.html", + "summary": "table-103 The table shows income tax expense for years ending Jan 26, 2025 ($11,146M, 13.3%) and Jan 28, 2024 ($4,058M, 12.0%), with adjustments for state taxes, foreign deductions, stock compensation, R&D credits, and foreign rate differentials." + }, + { + "path": "tables/table-103 Income Tax Paid.html", + "summary": "table-104 The table shows income taxes paid, net of refunds, totaling $20,288 million for the year ended Jan 25, 2026, with federal, state, and foreign breakdowns." + }, + { + "path": "tables/table-104 Deferred Tax Assets.html", + "summary": "table-105 The table shows deferred tax assets and liabilities for Jan 25, 2026 and Jan 26, 2025. Gross deferred tax assets were $16,863M and $13,214M, with net assets of $11,484M and $10,093M after valuation allowance and liabilities." + }, + { + "path": "tables/table-105 Tax Positions.html", + "summary": "table-106 The table shows the balance of tax positions at beginning and end of periods for Jan 2024-2026, with increases, decreases, lapse, and settlements in millions." + }, + { + "path": "tables/table-106 Segment Revenue.html", + "summary": "table-107 The table shows segment revenue, other segment items, and operating income for Compute & Networking and Graphics for fiscal years 2024, 2025, and 2026, with totals in millions." + }, + { + "path": "tables/table-107 Income Breakdown.html", + "summary": "table-108 The table shows segment operating income, stock-based compensation, unallocated expenses, acquisition costs, interest income/expense, and other income, leading to consolidated income before tax for fiscal years 2024-2026." + }, + { + "path": "tables/table-108 Revenue by Region.html", + "summary": "table-109 The table shows total revenue of $215,938M for Jan 25, 2026, with US ($149,617M), Taiwan ($42,345M), China ($19,677M), and Other ($4,299M). Prior years: $130,497M (2025) and $60,922M (2024)." + }, + { + "path": "tables/table-109 NVIDIA Revenue.html", + "summary": "table-110 NVIDIA's total revenue grew from $60.9B in FY2024 to $130.5B in FY2025 and $215.9B in FY2026, driven by Data Center growth ($47.5B to $193.7B)." + }, + { + "path": "tables/table-110 Long-lived assets.html", + "summary": "table-111 The table shows long-lived assets by region for Jan 25, 2026 and Jan 26, 2025, with totals of $10,383M and $6,283M respectively." + }, + { + "path": "tables/table-111 Lease Obligations.html", + "summary": "table-112 The table shows operating lease obligations by fiscal year from 2027 to 2032+, totaling $3,624 million. After subtracting imputed interest of $680 million, the present value of net minimum lease payments is $2,944 million, with long-term liabilities of $2,572 million." + }, + { + "path": "tables/table-112 Lease Cash Flows.html", + "summary": "table-113 The table shows supplemental cash flow information for operating leases, including cash used and assets obtained in exchange for lease obligations for fiscal years 2024-2026." + }, + { + "path": "tables/table-113 Allowance Changes.html", + "summary": "table-114 The table shows allowance for doubtful accounts, sales returns, and deferred tax valuation allowances for fiscal years 2024-2026, with balances and changes." + }, + { + "path": "tables/table-114 Exhibit Index.html", + "summary": "table-115 An exhibit index listing NVIDIA's incorporation documents, equity plans, and certifications, with references to SEC filings like 10-K and 8-K." + }, + { + "path": "tables/table-115 Signatures.html", + "summary": "table-116 Signatures of NVIDIA executives and directors including Jen-Hsun Huang, Colette M. Kress, Donald Robertson, and others, dated February 25, 2026." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/full.md b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/full.md new file mode 100644 index 000000000..b6f765e78 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/full.md @@ -0,0 +1,4864 @@ +## 2026 + +## NVIDIA Corporation + +## Annual Review + +Notice of Annual Meeting + +Proxy Statement + +Form 10-K + +![](images/1bcd3dea5b7d24d3e9e78090772bb74a558b15906dce328c7af6093dfc4b6e74.jpg) + +
+natural_image + +Abstract 3D geometric design with green spiral and rectangular forms on a grid background (no text or symbols) +
+ +## For most of computing history, software was pre-recorded. + +Humans described an algorithm. Computers executed it. + +Now software is trained. It learns from data and generates intelligence in real time. + +General-purpose computing has run its course. Moore's Law slowed. Dennard scaling ended. Accelerated computing became foundational. + +Because intelligence is produced in real time, the entire computing stack beneath it had to be reinvented. + +AI runs on real hardware, real energy, and real economics. It takes raw materials and converts them into intelligence at scale. + +This is not a normal upgrade cycle. + +The world is going through a platform shift. + +![](images/b688d90e4ecee0f8e16d19de8f1eeabbfc5d2000f535e1a202ade9aaa5761aee.jpg) + +
+flowchart + +```mermaid +graph TD + A["APPLICATIONS"] --> B + C["MODELS"] --> D + EINFRASTRUCTURE] --> F + CHIPS --> G + ENERGY --> H +``` +
+ +![](images/1bb69ea27a5925a4e1f9600bb292fd537655d0021c8fb5ca7f8747c483195b8f.jpg) +CHATBOTS + +![](images/88e9b68208096e918b329616f838299fe777e00b059788e9ae9763514cf2c85d.jpg) +DIGITAL BIOLOGY + +![](images/baca9281e72f1b698f491602d475fb66e5aab66e5a1adebe539e392c6fcce49d.jpg) +ROBOTAXI + +![](images/d27c2cf79777fea997ed5d70aa0c82d4a9cf2782abfefc76f58c54316cff6796.jpg) +ENTERPRISE +AI AGENTS + +![](images/56cbfd4210dcc1a304b34a8de8891ec19928e0e2ada6efd30aca64dd8d195ac9.jpg) +SCIENCE + +![](images/9ca57804f077454d57408e2de8b90c5e97d7ab3344d88bb9058ee52c86251206.jpg) +ROBOTICS + +![](images/8c6319f7eaf8e6399b2fdd94ddeceb1f90de9ce14ec10ecb44fdb62640ae66e0.jpg) +MANUFACTURING + +![](images/54d47b0d9f687fda287d087d89bfe5a44f4c7dfe2ac915dd7f8ff057fb7f3c09.jpg) +AI CODER + +![](images/fd23c4ea94a99a74ad019e3f98834682635fbe83555774b3c22cce0b612e5b44.jpg) + +
+text_image + +LLM +VLM +VLA +MMLLM +GPT +DM +GNN +MOE +SSM +LBM +
+ +AI FACTORIES +![](images/40b8c8c8762018df6e3eaffaa335d6c5b891192f22443dbfa36dbb4351a36f51.jpg) + +
+natural_image + +Stacked cylindrical containers with transparent lids, arranged in two rows (no visible text or symbols) +
+ +## AI Is a Five-Layer Cake + +It is not a clever app or a single model; it is essential infrastructure, like electricity and the internet. + +When you look at AI industrially, it resolves into a five-layer stack. Energy → chips and systems → infrastructure → models → applications. + +Every successful application pulls on every layer beneath it, all the way down to the power plant that powers it. + +Energy becomes computation. Computation becomes intelligence. The amount of intelligence a system can produce is bounded by energy. + +When applications scale, the entire stack must scale with them. + +## AI Is Driving the Largest Infrastructure Buildout in Human History + +Data centers are becoming AI factories. + +Land, power, cooling, networking, and systems must scale in parallel. Every company will use AI. Every country will build AI infrastructure. This is the next industrial revolution. We are only a few hundred billion dollars into it. Trillions more still need to be built. + +![](images/e6b930fc6bbf6cc0ddf49a961380ef788588d50ccf493fdbc5545b2aa374312e.jpg) + +
+flowchart + +```mermaid +graph TD + A["Reference Designs, Best Practices, Methodologies"] --> B["Power"] + C["Chip, Systems, Facilities"] --> D["Chips, Systems, Facilities"] + E["Rubin DSX Reference Designs"] --> F["DSX Ecosystem"] + G["Omniverse DSX Blueprint"] --> H["DSX Max-Q"] + I["DSX Flex"] --> J["DSX Exchange"] + K["DSX Sim"] --> L["DSX Libraries, APIs, Software"] + M["Liquid Cooling"] --> N["Liquid Cooling"] +``` +
+ +## AI Factories Manufacture Intelligence + +Intelligence is now produced, not stored. + +Modern data centers are AI factories, converting energy into tokens—just as power plants convert energy into electricity. + +Throughput per megawatt defines production. Cost per token defines efficiency. + +Token output and performance per watt directly determine revenue. + +## Inference Defines the Economics of AI + +For agentic AI—low-latency, long-context, multi-step reasoning—throughput per megawatt determines cost per token. + +Inference runs continuously and directly drives revenue. + +Reasoning models generate dramatically more tokens. They generate many more “thinking” tokens before they answer, and they have to do it fast enough to stay interactive. + +SemiAnalysis InferenceX measured GB300 NVL72 delivering up to 50x higher throughput per megawatt and up to 35x lower cost per token compared to Hopper, with the greatest gains in low-latency workloads. + +Inference is not a chip problem. It is a systems architecture problem—GPU, networking, memory, and software designed as one. + +![](images/4fb16199cfe8bea5c92ea9cb8512b930c5573e743da0c39a790a6201f242dbb8.jpg) + +
+text_image + +semianalysis +InferenceMAX +KING +
+ +## Open Models Power the Ecosystem + +Open models have reached the frontier. + +Researchers, startups, enterprises, and entire nations rely on open models to participate in advanced AI. Most of the world's models are free. + +At this level of capability, they do not just change software. They activate demand across the entire stack. + +Open models lower barriers to entry. They enable deep customization. They allow teams to build domain-specific systems on common platforms. + +As developers build, customize, and deploy intelligence, token generation grows—and demand for AI infrastructure grows with it. + +![](images/a7788baba65aefccb6db071e2e23e3c3a3c1d70e5ea360ff0979f789de037053.jpg) + +
+line chart + +| Month | NVIDIA | Alibaba Cloud | Hugging Face | Ai2 | Meta | Google | IBM | ByteDance | Baidu | Tencent | Microsoft | Zai | +| --------- | ------ | ------------- | ------------ | ---- | ---- | ------ | ---- | --------- | ----- | ------- | --------- | ---- | +| Jan 2025 | ~10 | ~10 | ~10 | ~10 | ~10 | ~10 | ~10 | ~10 | ~10 | ~10 | ~10 | ~10 | +| Feb 2026 | ~450 | ~350 | ~200 | ~180 | ~150 | ~130 | ~120 | ~110 | ~100 | ~90 | ~80 | ~70 | +
+ +![](images/553aa8d5e1e596d103bba73c2ba79596c94ce1324499e55d4fe3f30b561fecf1.jpg) + +
+natural_image + +Six-panel illustration showing molecular structures, a globe, a person waving, a spiral, a humanoid robot waving, and a car (no text or symbols) +
+ +## Agentic AI Enters Production + +AI now perceives, reasons, plans, and acts. + +Agentic systems break complex problems into steps. They evaluate alternatives. They use tools. They retrieve information. They act across long-running workflows. They coordinate across models, tools, and environments in real time. + +OpenClaw showed what happens when agentic AI becomes open: every developer can build agents that put AI to work. As agents move from chat to work, they will access data, use tools, execute code, and communicate with other systems. All of this must happen securely, privately, and under policy control. + +![](images/0bf1c7408e52db866c6a75ab13f5cbf6a65ce57796a63e04627033522f122f8f.jpg) + +
+flowchart + +```mermaid +graph TD + A["vGPU"] -->|FILES STRUCTURED, UNSTRUCTURED| B["Computer Use"] + B --> C["NEMOCLAW"] + C --> D["MEMORY"] + D --> E["LLM"] + E --> F["Sub-agents"] + F --> G["AI-Q"] + G --> H["cuOPT"] + H --> I["SKILLS"] + I --> J["NEMOTRON NEMO DYNAMO NIM"] + J --> K["AGentic AI is moving into enterprise workflows and crit"] + C --> L["TOOLS CLI, MCP"] + L --> M["OpenShell"] + M --> C + style A fill:#99ccff,stroke:#333 + style B fill:#99ccff,stroke:#333 + style C fill:#f0f0f0,stroke:#333 + style D fill:#e0e0e0,stroke:#333 + style E fill:#e0e0e0,stroke:#333 + style F fill:#e0e0e0,stroke:#333 + style G fill:#e0e0e0,stroke:#333 + style H fill:#e0e0e0,stroke:#333 + style I fill:#d0d0d0,stroke:#333 + style J fill:#d0d0d0,stroke:#333 + style K fill:#d0d0d0,stroke:#333 + style L fill:#d0d0d0,stroke:#333 + style M fill:#d0d0d0,stroke:#333 + style N fill:#d0d0d0,stroke:#333 +``` +
+ +Agentic AI is moving into enterprise workflows and critical systems. + +## Physical AI Puts Intelligence in Motion + +Robotics, autonomous vehicles, and intelligent machines combine perception and reasoning with an understanding of physics. + +They operate in dynamic environments and act in the physical world. + +Foundation models are trained at scale, refined in simulation, and deployed at the edge—creating a continuous loop between virtual training and real-world experience. + +Physical AI is transforming factories, warehouses, transportation, and healthcare. + +Everything that moves will be autonomous. + +![](images/03bb2bcbd4abad24e62f0d4875cca914ad648d2f4aaa3899ec03fc77c94ed66f.jpg) + +
+natural_image + +Interior of a modern exhibition space featuring a large excavator, humanoid robots, and a car with a child in an orange costume (no visible text or symbols) +
+ +# AI Reinvents Computer Graphics + +GeForce brought AI to the world. Now AI has come home to GeForce. + +Computer graphics is no longer limited by traditional rendering. + +With neural rendering and DLSS, AI predicts pixels instead of computing them directly. For every pixel we render, AI generates the rest in real time. + +Neural networks now run alongside shaders and ray tracing cores. AI does not replace graphics. It enhances it. + +In the era of neural rendering, graphics is learned. + +“AI has crossed the threshold from generating answers to performing work. The next industrial revolution will be built on factories that manufacture intelligence—and NVIDIA is building the platform to power them.” + +![](images/ed7473da5d2adae0ab5da29f147c550228deb17f3c658b775500b690d580c412.jpg) + +
+text_image + +Jahibing +
+ +![](images/358933c002827415708bd873785bf10b0d80c0cca255861dbba2ef9dfaf4724d.jpg) + +
+natural_image + +Elderly man in a black leather jacket gesturing while holding a handheld device (no visible text or symbols) +
+ +# Dear NVIDIAs and Stakeholders, + +## Computing platforms shift every 10 to 15 years—from mainframe to PC to internet to cloud to mobile-cloud. + +What is happening now is bigger. Computing is undergoing a complete reset not seen since IBM introduced modern computing with the System/360 in 1964. We are moving from computers as tools to a new kind of infrastructure that manufactures intelligence in real time. + +NVIDIA is driving this platform shift, building the computing infrastructure of the AI era. + +AI is transforming every layer of the computing stack—shifting from human coding to machine learning, from running code on CPUs to running neural networks on GPUs. + +AI is more than a model—it is a new computing paradigm. Intelligence, unlike information, cannot be prerecorded, stored, and retrieved. It is generated in real time, taking into account context, prior knowledge, and instructions. + +AI is still software running on chips. But because the computer can now understand, reason, plan, and use tools, it appears intelligent—hence, artificial intelligence. + +Over the past year, investors and enterprises questioned the ROI in AI—and for good reason. Early systems, such as chatbots, greatly improved access to information but did little productive work. Over the past six months, however, the landscape has shifted sharply. Enterprise adoption has surged. + +Inference—running trained AI models to generate responses, reasoning, and actions—is driving cloud demand, and demand for GPU compute is accelerating. + +Agentic AI has arrived and is delivering remarkable results. Generative AI creates. Reasoning AI thinks through problems. Agentic AI uses tools to perform tasks after thoughtful planning, research, and learning. AI has crossed an important threshold: it no longer simply answers questions—it does useful work. + +The first major enterprise breakout use case is AI coding agents, now sweeping across the industry. Within a few years, AI will generate much of the world's software. Soon after, AI will produce orders of magnitude more software than all human engineers combined. Limited only by imagination, the pace of innovation will accelerate across industries. + +Software can codify nearly all enterprise information. As a result, AI agents can accelerate or fully automate every business process through the software they generate. Organizations will rethink their operations end-to-end, from customer engagement through internal workflows and across the supply chain. Every business process will be enhanced and accelerated by AI agents. + +Companies that adopt AI agents—and build the infrastructure to support them—will out-innovate, out-execute, and attract the world's best talent. + +This shift is driving a fundamental change in the economics of computing. + +![](images/6a6eff2daf02e63e036dd4efb570a6b1e7da7e90e91e7717e3a21fd3a74ddf50.jpg) + +
+text_image + +Reimagine What's Possible +Poster Gallery +Reimagine What's Possible +JOSÉ CIVIC +Bough Discoveries +
+ +Compute is no longer just a cost to support software; it is a productive asset that generates intelligence, productivity, and revenue. Compute is now the infrastructure of this new economy. + +A computer used to be a tool. In the age of AI, it is manufacturing equipment. AI requires a new kind of computing infrastructure that runs continuously. Traditional data centers were built to store and serve files. Today, they are AI factories—systems designed to turn energy into tokens, the basic units of digital intelligence, and tokens into answers, reasoning, actions, and work. AI factories are not cost centers—they are productive assets that generate revenue. + +NVIDIA delivers a full-stack AI computing platform that spans the entire AI lifecycle—from training to inference—and every domain of intelligence, from language and video to biology, autonomous driving, and robotics. + +NVIDIA is a vertically integrated, horizontally open platform. We architect with a vertically integrated mindset, executing extreme co-design across chips, systems, networking, storage, and software—spanning GPUs, CPUs, LPUs, DPUs, and full AI factory systems. At the same time, our platform remains open and integrates with the broadest software and hardware ecosystems. + +Our platform powers AI wherever it runs—across clouds, on-premises data centers, telecom networks, medical instruments, vehicles, and robots. + +NVIDIA is building the computing infrastructure of the AI era. + +## Driving the AI Platform Shift + +GTC is where the industry comes together across every layer of the AI stack, from land, power, and infrastructure to chips, models, and applications. This year, what has become unmistakably clear is that AI has moved from experimentation to production. + +We laid out the roadmap from Blackwell to Blackwell Ultra to Vera Rubin to Vera Rubin Ultra to Feynman. AI infrastructure builders plan over years, across factories, ecosystems, and supply chains. + +Today, we have visibility into more than \$1 trillion in cumulative Blackwell and Rubin revenue from the start of 2025 through 2027. Since GTC DC in October 2025, we have received additional purchase orders and further diversified our customer base, including Anthropic, Meta, and a growing set of world-class model builders such as Cursor, Perplexity, Thinking Machines Lab, Reflection, and Skild. + +At the same time, the scale of our cloud service providers continues to accelerate. In addition to the significant buildouts already underway at Azure and OCI, AWS and Google have announced plans for major NVIDIA GPU infrastructure expansions through 2027. NVIDIA AI cloud partners—including CoreWeave, Nebius, and Nscale—are also experiencing exponential demand and have secured significant long-term offtake agreements. + +This momentum reflects the economics of this new industry: AI factories are revenue-generating infrastructure. They produce tokens, and tokens produce revenue. + +## AI is a new industry: The Five-Layer Cake + +Like steam power, electricity, and information technology before it, AI is forming a new industry. AI is more than a model—it's a new industry with a five-layer stack: energy at the foundation; compute—chips and systems—above it; then infrastructure, models, and finally applications at the top. + +In the industrial age, turbines and generators produced electricity, priced in dollars per kilowatt-hour, distributed globally, and enabled entirely new industries. In the AI era, electricity powers GPUs that produce digital intelligence, priced in dollars per million tokens—transforming existing industries and creating new ones. AI factories are not designed to store information, but to produce intelligence. + +Each industrial revolution created new infrastructure and companies. Power plants and utilities emerged in the age of electricity. Data centers, chips, and software companies defined the information technology era. Today, AI factories have emerged as the infrastructure of this new age, alongside entirely new classes of chips, systems, and computing platforms—and tens of thousands of AI-native startups. + +Above the infrastructure are models. While large language models are widely known, the frontier extends much further. AI now includes models that understand biology, chemistry, physics, finance, medicine, and the physical world—bringing intelligence into the core of every major industry. + +At the top are applications—where AI creates economic and social value—such as software development, customer support, drug discovery, industrial automation, autonomous vehicles, and robotics. In the years ahead, companies will orchestrate tens of billions of digital and physical agents working alongside people to help us be more productive, more creative, and more capable. + +Every successful AI application pulls on every layer beneath it—from energy to chips to infrastructure to models. Applications are where AI infrastructure becomes productivity, growth, and national capability. + +## Fiscal 2026: A Defining Year + +Revenue grew 65 percent to \$216 billion. Operating income rose 60 percent to \$130 billion. Diluted EPS increased 67 percent to \$4.90. We generated \$103 billion in operating cash flow and returned \$41 billion to shareholders. + +Demand remained strong across cloud providers, hyperscalers, AI model makers, enterprises, and sovereigns. AI infrastructure is now essential, and investment is accelerating worldwide. + +Fiscal 2026 also marked a turning point in adoption. AI moved from experimentation to measurable revenue and efficiency, with software development, healthcare, customer support, and industrial automation among the first domains to show strong productivity gains. + +![](images/1ab5a0983868402542b357fbc2feb97d0b1cc63d53055ca679441e4ce34de65f.jpg) + +
+text_image + +Inference At-Scale is Extreme Computing +Throughput +Revenue +Tokens per Second For Factory +FLOPS +HBM Memory +HBM Bandwidth +Software and Architecture +Smart AI +Fast Response +Tokens per Second For C +for C +
+ +## Inference Inflection Driving Compute Growth + +Marking a fundamental shift, inference has overtaken training as the dominant workload. AI is no longer just being built; it is being used at scale. With that shift, the economics of AI have changed. Demand is no longer conceptual—the AI flywheel is accelerating. + +Inference demand is surging as multiple exponential forces converge. Models are becoming more capable. Agents built on these models can reason, plan, and solve complex problems, generating tokens at every step. As these agents prove their value, they are deployed across more workflows, by more users, in more industries—generating still more tokens. + +More tokens require more compute. More compute enables smarter AI. Compute produces intelligence, intelligence drives demand, and demand drives more compute. The supply-demand flywheel is formed and accelerating. + +In fiscal 2026, Data Center revenue reached \$194 billion, up 68 percent year over year. Major cloud service providers, hyperscalers, AI model makers, and enterprises have deployed nearly 9 gigawatts of Blackwell AI factory capacity. Our networking revenue exceeded \$31 billion. Our sovereign AI business more than tripled to over \$30 billion. + +## Computing Architecture Defines AI Factory Economics + +An AI factory is not a conventional data center. It is a production system—a revenue-generating engine. + +Its purpose is to manufacture tokens at scale—with the highest intelligence, the highest throughput, the lowest latency, and the greatest energy and cost efficiency. In an AI factory, token throughput determines how much work can be performed. Cost per token determines the economics. Tokens per watt determine revenue per megawatt. + +Computing architecture determines it all—directly shaping the performance, economics, and revenue of the AI factory. + +Blackwell changed the economics of inference, and the results are reflected in our FY26 scale. Independent benchmarks show up to 50x higher throughput per megawatt and up to 35x lower cost per token than Hopper, with the greatest gains in low-latency workloads. NVIDIA delivers the lowest cost per token and the highest revenue potential for the AI factories our customers operate. Our extreme co-design across silicon, systems, networking, and software is what makes that possible. + +Inference is one of the hardest computing challenges and is mission-critical to our customers' businesses. That is why we made it a company-wide priority. With Grace Blackwell, NVIDIA leads across the full lifecycle of AI—from training to inference. Grace Blackwell—the world's first rack-scale system—integrates NVLink, NVFP4, TensorRT-LLM, Spectrum-X, and InfiniBand into a single, unified computer. Dynamo—our inference operating system—optimizes every query, every token, every watt—maximizing throughput, minimizing cost, and increasing factory output. + +Grace Blackwell delivered a step-function leap in LLM training and inference. We built the Vera Rubin platform to run agents. Where Grace Blackwell introduced rack-scale systems, Vera Rubin extends to pod-scale production. + +Vera Rubin is not a chip. It is an AI factory platform—realized through extreme co-design across Rubin GPUs, Vera CPUs, Groq LPUs, BlueField DPUs, NVLink switching, Spectrum-X and InfiniBand networking, storage, systems, and software. + +With the addition of Groq technology, the platform now includes a new tier of ultra-low-latency inference—built for real-time, agentic workloads. + +The computing demands of agentic systems are massive and complex. Vera Rubin has arrived just in time. + +## CUDA Installed Base Powering AI Flywheel + +This year marked the 20th anniversary of CUDA. + +CUDA is a revolutionary programming model—and the foundation of the NVIDIA platform. Over two decades, CUDA has created a powerful developer flywheel that continues to compound performance, adoption, and innovation across a global ecosystem. Its capabilities attract developers. Developers create new algorithms. New algorithms drive breakthroughs. Breakthroughs create new markets. New markets expand the installed base. And the installed base attracts more developers. The flywheel is now accelerating. + +CUDA also extends the lifespan of NVIDIA systems. Our platforms are architecturally compatible across generations, and we continuously optimize the software stack. Improvements in libraries, runtimes, compilers, and frameworks benefit the entire installed base. Customers do not just get the hardware's performance on day one—they get continuous performance gains and runtime cost reductions over the life of the system. + +This dynamic is foundational to AI infrastructure builders and operators. CUDA attracts developers and customers. It lowers the total cost of ownership for operators and users. It extends the useful life of every system. + +It is why prior-generation products remain valuable. It is why customers build with confidence across Blackwell, Vera Rubin, and beyond. + +The confidence to build on NVIDIA—driven by CUDA—is one of the deepest moats we have created. + +## A Global AI Ecosystem + +The NVIDIA ecosystem now reaches the market through every major route: hyperscalers, cloud providers, AI-native clouds, enterprise IT, sovereign deployments, and industrial systems. Our business is broad and diversified. + +AWS, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure are expected to be among the first major cloud providers to deploy Vera Rubin-based infrastructure. Meta expanded to a multi-year, multigenerational strategic partnership spanning on-premises, cloud, and AI infrastructure. CoreWeave, Nebius, Nscale, and many other AI clouds continue to expand AI factory capacity. OpenAI, Anthropic, xAI, and new frontier and open model builders are scaling on NVIDIA. + +We are also extending that reach into on-prem and air-gapped deployments. With partners such as Palantir and Dell, we are enabling customers to stand up AI platforms in any country, region, or field environment where privacy, sovereignty, resilience, and control matter most. + +To help customers design, build, and operate AI factories, we provide reference designs, blueprints, and a partner ecosystem. + +The Vera Rubin DSX AI Factory reference design and NVIDIA Omniverse DSX Blueprint bring together compute, networking, storage, power, cooling, and operations into an open, modular architecture supported by partners including Vertiv, Schneider Electric, Eaton, Jacobs, Phaidra, Procore, PTC, Siemens, Switch, Trane Technologies, Cadence, and Dassault Systèmes. The broader power-to-rack ecosystem now also includes GE Vernova, which is using digital twins aligned with the Omniverse DSX Blueprint to model grid behavior, substations, and AI factory loads before deployment. Across industrial AI, Caterpillar is using NVIDIA Omniverse digital twins to accelerate AI-driven manufacturing and automation. + +## Reindustrializing America and Creating Skilled Jobs + +AI is driving the largest infrastructure buildout in human history and creating a once-in-a-generation opportunity to reindustrialize America and bring advanced manufacturing back to the United States. Manufacturing in America strengthens supply chain resilience, reduces geopolitical risk, and creates high-quality jobs for people who choose careers outside of traditional four-year degrees—helping build a more balanced and inclusive economy. + +Building an AI factory requires electricians, pipefitters, welders, ironworkers, HVAC technicians, and networking and computer specialists. + +This buildout extends far beyond the data center. It spans semiconductor fabs, advanced packaging facilities, server and supercomputer assembly, power generation, transmission lines, substations, and the vast upstream supply chains and downstream services that support this new multi-trillion-dollar industry. + +Once operational, these AI factories create permanent, high-skill careers—data center technicians, critical facility engineers, and site managers—where degreed engineers and skilled tradespeople work side by side as peers. + +In the United States alone, this buildout can create opportunities for hundreds of thousands of skilled craft workers in the years ahead to build and operate AI infrastructure. + +NVIDIA is helping lead America's reindustrialization. We have begun domestic production of NVIDIA AI infrastructure, including chip manufacturing in Arizona with TSMC, advanced packaging with Amkor and SPIL, and AI supercomputer assembly in Texas with Foxconn and Wistron—ramping over the coming year. + +## Democratizing Enterprise Agentic AI + +Every enterprise software company needs an agentic AI strategy. + +OpenClaw showed how quickly open agentic computing can spread. Agents will not replace software tools—they will dramatically increase their use, operating at accelerated machine speed and requiring tools that keep up. + +Enterprise software platforms will embed specialized agents trained on proprietary data and domain-specific skills—capable of securely accessing sensitive information and operating in regulated environments. General-purpose and specialized agents will work together as a digital workforce to orchestrate complex workflows, from designing chips and running cloud services to operating factories. + +NVIDIA provides the open, full-stack platform for the enterprise agent lifecycle—from build, evaluation, and guardrails to deployment. + +OpenShell and NVIDIA Agent Toolkit help agents access tools, follow policies, protect data, and operate inside companies. NeMo enables enterprises to build and customize models and agents on proprietary data. NIM packages optimized AI models into production-ready services. Nemotron provides open reasoning and agentic models. + +The enterprise ecosystem around this platform is expanding rapidly. Adobe, Cisco, Palantir, SAP, Salesforce, ServiceNow, Siemens, Synopsys, Cadence, Dassault Systèmes, and other industry leaders are building domain-specific agents and AI systems on NVIDIA. + +## Open Models Accelerate AI for All Industries + +Frontier model services are excellent for general intelligence use cases. They work out of the box and are continuously getting smarter. At NVIDIA, we use OpenAI Codex, Anthropic Claude Code, and Cursor. + +Still, open models are essential. + +Open models advance the field of AI. They expose architectures, weights, and training methods—enabling comparison, validation, and rapid innovation. Breakthroughs compound across the ecosystem. + +Open models improve safety and security. Cybersecurity companies fine-tune open models and deploy swarms of specialized agents to defend against cyberattacks. In regulated industries and national deployments, enterprises build agents to enforce compliance and operate within strict controls. + +Open models enable sovereignty. Nations build AI aligned with their language, culture, and values—and keep data within their borders. + +![](images/fcb8da8a2ab8a3853bcf2e1f157eebe1f153098ae0d6e9f7ca1c8dd5f7e64a1d.jpg) + +
+natural_image + +Outdoor scene with people using green IoT robots near a decorated park, no visible text or symbols on the robots or background. +
+ +Open models enable specialization at scale. Enterprises build domain-specific models trained on proprietary data and deep expert knowledge. + +The open model ecosystem is large and accelerating. Nearly 3 million open models now span language, vision, biology, physics, and robotics. NVIDIA is a leading contributor, with six families of frontier models, plus the data, recipes, and frameworks developers need to customize them. + +We build open models across the major domains of AI: Nemotron for reasoning and language, Cosmos for physical AI, GROOT for robotics, Alpamayo for autonomous driving, BioNeMo for biology and chemistry, and Earth-2 for weather and climate. These models enable domain-specific intelligence on a common NVIDIA platform. + +Adoption spans industries—from cybersecurity to drug discovery to mobility. CrowdStrike, Lilly, and Mercedes-Benz are building specialized AI systems on NVIDIA. Researchers in Japan, France, Sweden, India, and Israel are building national models. + +Scientists worldwide are advancing research with open models. + +## AI as National Infrastructure + +In every industrial revolution—from electricity to telecommunications to the internet—nations build infrastructure to turn new technology into national capability, economic growth, and strategic independence. Every country will deploy AI infrastructure—whether with global partners or domestic companies—aligned with its language, culture, and industries. + +We saw this accelerate sharply in fiscal 2026. Demand broadened as countries and regions built their own AI capacity, cloud services, research infrastructure, and industrial ecosystems. + +## The ChatGPT Moment for Physical AI Has Arrived + +The next wave of AI is physical. AI is moving from screens into factories, warehouses, hospitals, vehicles, and robots. To operate in the real world, AI must understand geometry, motion, causality, and physics. It must reason and act safely in dynamic, unpredictable environments. + +Physical AI is a new frontier in computing. + +NVIDIA is building the full training-to-deployment loop for physical AI. AI factories train models. Omniverse, with Cosmos World Foundation Models, simulates worlds, robots, and workflows with physical fidelity. GROOT advances general-purpose robotics. AGX computers deploy intelligence into robots, instruments, industrial systems, and autonomous vehicles. + +This ecosystem is already taking shape. Agility Robotics, Amazon Robotics, Boston Dynamics, Figure, Uber, Mercedes-Benz, and many other robotics, automotive, and industrial leaders are building on NVIDIA platforms. Industrial software leaders, including Cadence, Dassault Systèmes, PTC, Siemens, and Synopsys, are bringing NVIDIA-accelerated design, simulation, and digital twins to millions of engineers, researchers, and manufacturers. Biology is also entering this loop, with BioNeMo expanding AI-driven drug discovery and co-innovation efforts such as our work with Lilly. + +Physical AI is no longer a research project. It is moving into production. + +![](images/f5e38560d71f861827461501187a236be0acef2adc68574098cd6cdb1030c0ae.jpg) + +
+natural_image + +Silhouette of a person standing on a platform overlooking Earth with cloud patterns and a colorful sky (no text or symbols visible) +
+ +## The Buildout Has Just Begun + +AI has been advancing at an extraordinary pace. In just a few years, ChatGPT awakened the world to AI. Reasoning systems allow models to think before answering. Now agents can use tools and do work. Enterprise adoption is surging, settling the ROI debate: AI can perform valuable work, demand is accelerating, and AI services are delivering compelling economics—because tokens now perform valuable work at scale. + +Ahead lies the largest infrastructure buildout in human history—AI factories are being built around the world, powering digital and robotic agents that will supercharge every industry. Building a new industry requires resources—but it also creates opportunities. AI will demand energy, computing, networking, software, and advanced manufacturing. + +AI will transform how we work—and how the world works. It will automate most tasks, reshape every job, eliminate some, and create entirely new ones. That is why it is important to distinguish between a task and the purpose of a job. Coding is a task; innovation is the purpose. Reading scans is a task; caring for patients is the purpose. AI will automate more tasks so people can do more of the work that matters. + +As tasks are automated, the purpose of work will rise—AI will amplify human potential and expand what humanity can achieve. New companies will emerge on this foundation, extending AI into every industry. Companies and nations must build now. The future is arriving faster than any platform shift in history. + +NVIDIA has spent more than three decades preparing for this moment. + +We are building the computers, the networks, the software, the models, and the factories that manufacture intelligence. We are building the computing platform for companies, industries, nations, and all of us to thrive in the age of AI. + +To our employees, partners, developers, customers, and shareholders around the world—thank you. + +NVIDIA is the result of a decades-long vision and the life's work of our people. We carry this opportunity forward with humility, responsibility, determination, and optimism for what comes next. + +$$ +j u l t h y +$$ + +Jensen Huang + +Founder and CEO, NVIDIA + +May 2026 + +# NVIDIA Corporation + +Notice of 2026 Annual Meeting + +Proxy Statement and Form 10-K + +## Forward-Looking Statements + +Certain statements in this document including, but not limited to, statements as to: the benefits, impact, performance, growth, and availability of NVIDIA's products, services, and technologies, and related trends and drivers; expectations with respect to supply and demand for NVIDIA's products, services and technologies; expectations with respect to NVIDIA's revenue; expectations with respect to NVIDIA's third party arrangements, including with its collaborators and partners; expectations with respect to technology developments; expectations with respect to AI and related industries; NVIDIA driving the computing platform shift, building the computing infrastructure of the AI era; NVIDIA building the computing infrastructure of the AI era; NVIDIA helping lead America's reindustrialization; NVIDIA building the full training-to-deployment loop for physical AI; NVIDIA building the computers, the networks, the software, the models, and the factories that manufacture intelligence; NVIDIA building the computing platform for companies, industries, nations, and all of us to thrive in the age of AI; and other statements that are not historical facts are forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the “safe harbor” created by those sections based on management’s beliefs and assumptions and on information currently available to management and are subject to risks and uncertainties that could cause results to be materially different than expectations. Important factors that could cause actual results to differ materially include: global economic and political conditions; NVIDIA’s reliance on third parties to manufacture, assemble, package and test NVIDIA’s products; the impact of technological development and competition; development of new products and technologies or enhancements to NVIDIA’s existing product and technologies; market acceptance of NVIDIA’s products or NVIDIA’s partners’ products; design, manufacturing or software defects; changes in consumer preferences or demands; changes in industry standards and interfaces; unexpected loss of performance of NVIDIA’s products or technologies when integrated into systems; NVIDIA’s ability to realize the potential benefits of business investments or acquisitions; and changes in applicable laws and regulations, as well as other factors detailed from time to time in the most recent reports NVIDIA files with the Securities and Exchange Commission, or SEC, including, but not limited to, its annual report on Form 10-K and quarterly reports on Form 10-Q. Copies of reports filed with the SEC are posted on the company’s website and are available from NVIDIA without charge. These forward-looking statements are not guarantees of future performance and speak only as of the date hereof, and, except as required by law, NVIDIA disclaims any obligation to update these forward-looking statements to reflect future events or circumstances. + +# NVIDIA® + +## NOTICE OF 2026 ANNUAL MEETING OF STOCKHOLDERS + +Date and time: Wednesday, June 24, 2026 at 9:00 a.m. Pacific Time + +Location: Virtually at www.virtualshareholdermeeting.com/NVDA2026 + +Items of • Election of ten directors nominated by the Board of Directors + +• Advisory approval of our executive compensation +- Ratification of the selection of PricewaterhouseCoopers LLP as our independent registered public accounting firm for fiscal year 2027 +• Four stockholder proposals, if properly presented +• Transaction of other business properly brought before the meeting + +Record date: You can attend and vote at the 2026 Meeting if you were a stockholder of record at the close of business on April 27, 2026. + +Virtual meeting admission: We will be holding the 2026 Meeting virtually at the location listed above. To participate, you will need the Control Number included on your notice of proxy materials or printed proxy card. + +Pre-meeting forum: To communicate with our stockholders in connection with the 2026 Meeting, we have established a pre-meeting forum located at www.proxyvote.com where you can submit questions in advance. + +Your vote is very important. Whether or not you plan to attend the 2026 Meeting, PLEASE VOTE YOUR SHARES. As an alternative to voting during the 2026 Meeting, you may vote in advance online, by telephone, or, if you have elected to receive a paper proxy card in the mail, by mailing the completed proxy card. + +Important notice regarding the availability of proxy materials for the Annual Meeting of Stockholders to be held on June 24, 2026. This Notice, our Proxy Statement, our Annual Report on Form 10-K, and our Annual Review are available at www.nvidia.com/proxy. + +By Order of the Board of Directors + +![](images/f9c0ea10730ce28a637d5fe2372c9a3a75693932ffdce4ada8455ef92ec55352.jpg) + +Timothy S. Teter + +Secretary + +2788 San Tomas Expressway, Santa Clara, California 95051 + +May 12, 2026 + +## NVIDIA Corporation + +## Table of Contents + +Page + +DEFINITIONS 3 + +BUSINESS OVERVIEW 4 + +PROXY SUMMARY 6 + +PROXY STATEMENT 10 + +Information About the 2026 Meeting 10 + +Proposal 1—Election of Directors 14 + +Director Qualifications and Nomination of Directors 15 + +Our Director Nominees 17 + +Information About the Board of Directors and Corporate Governance 23 + +Independence of the Members of the Board of Directors 23 + +Board Leadership Structure 23 + +Committees of the Board of Directors 25 + +Compensation Committee Interlocks and Insider Participation 26 + +Role of the Board in Risk Oversight 27 + +Corporate Governance 28 + +Stockholder Communications with the Board of Directors 30 + +Majority Vote Standard 30 + +Stockholder Special Meeting Right 31 + +Board Meeting Information 31 + +Corporate Sustainability 31 + +Public Policy Engagement and Accountability 31 + +Director Compensation 32 + +Review of Transactions with Related Persons 34 + +Security Ownership of Certain Beneficial Owners and Management 35 + +Proposal 2—Advisory Approval of Executive Compensation 37 + +Executive Compensation 38 + +Compensation Discussion and Analysis 38 + +Compensation Committee Report 49 + +Risk Analysis of Our Compensation Plans 49 + +Summary Compensation Table for Fiscal 2026, 2025, and 2024 50 + +Grants of Plan-Based Awards for Fiscal 2026 51 + +Outstanding Equity Awards as of January 25, 2026 52 + +Option Exercises and Stock Vested in Fiscal 2026 53 + +Employment, Severance, and Change-in-Control Arrangements 53 + +Potential Payments Upon Termination or Change-in-Control 54 + +Pay Ratio 54 + +Pay Versus Performance 55 + +Proposal 3—Ratification of the Selection of Independent Registered Public Accounting Firm for Fiscal 2027 58 + +Fees Billed by the Independent Registered Public Accounting Firm 59 + +Report of the Audit Committee of the Board of Directors 60 + +Stockholder Proposals 61 + +Equity Compensation Plan Information 69 + +Additional Information 69 + +Delinquent Section 16(a) Reports 69 + +Other Matters 69 + +This proxy statement contains forward-looking statements. All statements other than statements of historical or current facts made in this document are forward-looking. Forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the "safe harbor" created by those sections, are based on our management's beliefs and assumptions and on information currently available to our management. In some cases, you can identify forward-looking statements by terms such as "may," "will," "should," "could," "goal," "would," "expect," "plan," "anticipate," "believe," "estimate," "project," "predict," "potential," and similar expressions intended to identify forward-looking statements. Actual results could differ materially for a variety of reasons. Risks and uncertainties that could cause our actual results to differ significantly from management's expectations are described in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. + +DEFINITIONS + +
2007 PlanNVIDIA Corporation Amended and Restated 2007 Equity Incentive Plan
ACAudit Committee of the Board
ASC 718Financial Accounting Standards Board Accounting Standards Codification Topic 718: Compensation - Stock Compensation
Base Compensation PlanPerformance goal necessary to earn the target award under the Variable Cash Plan and for the target numbers of SY PSUs and MY PSUs to become eligible to vest
BoardThe Company's board of directors
BylawsThe Company's Amended and Restated Bylaws
CAP"Compensation actually paid," as defined under Item 402(v) of Regulation S-K
CCCompensation Committee of the Board
CD&ACompensation Discussion and Analysis
CEOChief Executive Officer
CFOChief Financial Officer
CharterThe Company's Restated Certificate of Incorporation
Control NumberIdentification number for each stockholder included in Notice or proxy card
CSCorporate Sustainability
CSSCCorporate Sustainability Steering Committee
DEIDiversity, Equity, and Inclusion
ERMEnterprise Risk Management
ESPPNVIDIA Corporation Amended and Restated 2012 Employee Stock Purchase Plan
EVPExecutive Vice President
Exchange ActSecurities Exchange Act of 1934, as amended
Fiscal 20__The Company's fiscal year ended on the last Sunday in January of the stated year
Form 10-KThe Company's Annual Report on Form 10-K for Fiscal 2026 filed with the SEC on February 25, 2026
GAAPGenerally accepted accounting principles in the United States
GHGGreenhouse gas
Internal Revenue CodeU.S. Internal Revenue Code of 1986, as amended
Lead DirectorLead independent director
MeetingAnnual Meeting of Stockholders
MY PSUsMulti-year PSUs based on 3-year TSR relative to the S&P 500 with a three-year performance metric, vesting after three years
NasdaqThe Nasdaq Stock Market LLC
NCGCNominating and Corporate Governance Committee of the Board
NEOsNamed Executive Officers consisting of our CEO, our CFO, and our other three most highly compensated executive officers as of the end of Fiscal 2026
Non-GAAP Operating IncomeGAAP operating income, as the Company reports in its SEC filings, excluding stock-based compensation expense, acquisition-related and other costs, and other. Please see Reconciliation of Non-GAAP Financial Measures in our CD&A for a reconciliation between the non-GAAP financial measures and GAAP results
NoticeNotice of Internet Availability of Proxy Materials
NVIDIA, Company, we, us, ourNVIDIA Corporation, a Delaware corporation
NYSENew York Stock Exchange
PSUPerformance stock unit
PwCPricewaterhouseCoopers LLP
RSURestricted stock unit
S&P 500Standard & Poor's 500 Composite Index
SECU.S. Securities and Exchange Commission
Section 162(m)Section 162(m) of the Internal Revenue Code
Securities ActSecurities Act of 1933, as amended
Stretch Compensation PlanPerformance goal necessary to earn the maximum award under the Variable Cash Plan and for the maximum numbers of SY PSUs and MY PSUs to become eligible to vest
SY PSUsPSUs based on annual Non-GAAP Operating Income performance with a single-year performance metric, vesting over four years
ThresholdMinimum performance goal necessary to earn an award under the Variable Cash Plan and for SY PSUs and MY PSUs to become eligible to vest
TSRTotal shareholder return
Variable Cash PlanThe Company's variable cash compensation plan
+ +## BUSINESS OVERVIEW + +Fiscal 2026 marked another year of strong growth for NVIDIA, with revenue surging 65% year on year to \$215.9 billion reflecting continued momentum across our platform transitions to accelerated computing and AI. Growth was led by Compute & Networking, where Data Center compute revenue grew 59% driven by demand for our Blackwell platform, and Data Center networking revenue increased 142% driven by the ramp of NVLink compute fabric for Blackwell systems, as well as continued growth in Ethernet and InfiniBand. Gross margin was 71.1%, operating income reached \$130.4 billion and diluted earnings per share increased 67% to \$4.90. + +Fiscal 2026 Results + +
RevenueGross MarginOperating IncomeDiluted Earnings Per Share
$215.9 billion71.1%$130.4 billion$4.90
up 65% year on yeardown 3.9 points year on yearup 60% year on yearup 67% year on year
+ +## Fiscal 2026 Reportable Segments + +Our two reportable segments are “Compute & Networking” and “Graphics”: + +
Compute & NetworkingGraphicsTotal
Revenue$193.5 billion (1)$22.5 billion (1)$215.9 billion
up 67% year on yearup 57% year on yearup 65% year on year
Operating Income (2)$130.1 billion$9.2 billion$139.3 billion
up 57% year on yearup 80% year on yearup 58% year on year
+ +(1) This figure has been rounded to the nearest tenth of a billion. Due to rounding, the sum of the figures does not equal the total as presented. +(2) Segment operating income differs from consolidated operating income as certain expenses are not allocated to either Compute & Networking or Graphics for purposes of making operating decisions or assessing financial performance. + +## Fiscal 2026 Market Platforms + +Our platforms address four large markets where our expertise is critical: + +![](images/e6fa6681413b79669eda7af4657e54d96c6665a50ae60046f6c25af29b30d2b8.jpg) + +
+natural_image + +Exterior view of a black server rack with multiple drive bays (no visible text or labels) +
+ +Data Center + +![](images/26bc018077966022264eaf14b3def10bb3c2672e2708eeda47e78701467a61b0.jpg) + +
+natural_image + +Black and white electronic device with green abstract background, no visible text or symbols +
+ +Gaming + +![](images/4c565522ce1873fab7085d09093726bb3393e55ea8540c53b42eb1e321764c65.jpg) + +
+text_image + +Desktop scene with laptop displaying a webpage interface and a cartoon character, next to a bookshelf and office items +
+ +Professional Visualization + +![](images/830a146edd991441fd15dd54f561b20d75597e51ff4dadc6f3a8f781fbb15abc.jpg) + +
+natural_image + +X-ray diagram of a car showing internal components and wiring (no text or labels) +
+ +Automotive + +\$193.7 billion revenue + +\$16.0 billion revenue + +\$3.2 billion revenue + +\$2.3 billion revenue + +up 68% year on year + +up 41% year on year + +up 70% year on year + +up 39% year on year + +## Business Highlights + +- Launched and scaled the NVIDIA Blackwell Ultra platform – built on the architectural breakthroughs of Blackwell, delivering 50x higher throughput and 35x lower token cost versus Hopper +- Extended our inference leadership – sweeping MLPerf and SemiAnalysis InferenceMax benchmarks – by leveraging our extreme co-design approach, where GPUs, CPUs, networking, security, software, power delivery, and cooling are architected together as a single system, rather than optimized in isolation to maximize performance per watt +- Introduced the NVIDIA Vera Rubin platform – including new chips and racks – built for agentic AI and co-designed to deliver up to 10x lower token cost versus Blackwell +- Entered into a non-exclusive licensing agreement with Groq and introduced the NVIDIA Groq 3 LPX, an accelerator for NVIDIA Vera Rubin, designed to meet the low-latency and large-context demands of agentic systems +- Announced new and expanding partnerships to cultivate ecosystem growth, unlock new markets, and accelerate NVIDIA platform adoption, including those with Anthropic, Meta, Nokia, and OpenAI +- Turbocharged AI market development with an accelerating release cadence of NVIDIA open models including Nemotron for agentic AI, Cosmos for physical AI, and Alpamayo for autonomous vehicles +- Announced key physical AI milestones including \$6 billion in Fiscal 2026 physical AI revenue, NVIDIA DRIVE-powered L4-ready Mercedes-Benz vehicles, and an Uber partnership to scale the world's largest L4-ready mobility network +- Advanced quantum computing with new technologies and NVIDIA-powered systems including NVIDIA NVQLink GPU-QPU interconnect and ABCI-Q, the world's largest Quantum research supercomputer + +## Fiscal 2026 Shareholder Returns + +Total Shareholder Return (TSR)\* +![](images/8eefa7867f6386dde47b81fd3f854a19e04ea87f61100394b30a71b9162f1d66.jpg) + +
+bar chart + +| Period | Value (%) | +|---|---| +| 1-Year | 32 | +| 3-Year | 822 | +| 5-Year | 1,349 | +
+ +\*Represents cumulative stock price appreciation with dividends reinvested and is measured for the applicable fiscal year periods based on the closing price (\$187.67) of NVIDIA's common stock on January 23, 2026, the last trading day before the end of our Fiscal 2026, as reported by Nasdaq. + +Return of Capital to Shareholders (in billions) +![](images/1f7e42398ac9fdd768969e3fe6b52869b38c138d03b0472fbcef8ddf46bfbf5e.jpg) + +
+stacked bar chart + +| Maturity | Dividends ($) | Stock Repurchases ($) | +| :--- | :--- | :--- | +| 1-Year | 0 | 40.4 | +| 3-Year | 2.2 | 84.1 | +| 5-Year | 3.0 | 94.2 | +The chart displays a single bar for stock repurchases and a separate purple bar for dividends. Values are explicitly labeled on each segment: $41.4 for 1-Year, $86.3 for 3-Year, and $97.2 for 5-Year. The total value per segment is also displayed. +
+ +Please see our Form 10-K for more financial information for Fiscal 2026. + +## PROXY SUMMARY + +This summary highlights information contained elsewhere in the proxy statement. This summary does not contain all the information that you should consider, and you should read the entire proxy statement carefully before voting. + +## 2026 Annual Meeting of Stockholders + +Date and time: Wednesday, June 24, 2026 at 9:00 a.m. Pacific Time + +Location: Virtually at www.virtualshareholdermeeting.com/NVDA2026 + +Record date: Stockholders as of April 27, 2026 are entitled to vote + +Admission to meeting: You will need your Control Number to attend the 2026 Meeting + +## Voting Matters and Board Recommendations + +A summary of the 2026 Meeting proposals is below. Every stockholder's vote is important. Our Board urges you to vote your shares FOR each of the nominees listed in Proposal 1, FOR Proposals 2 and 3, and AGAINST Proposals 4, 5, 6, and 7. + +
MatterPageBoard RecommendsVote Required for ApprovalEffect of AbstentionsEffect of Broker Non-Votes
Management Proposals:
1Election of Ten Directors14FOR each director nomineeMore FOR than AGAINST votesNoneNone
2Advisory Approval of Our Executive Compensation37FORMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
3Ratification of the Selection of PwC as Our Independent Registered Public Accounting Firm for Fiscal 202758FORMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstN/A*
Stockholder Proposals:
4Simple Majority Vote Standard61AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
5Report on Faith-Based Community Resource Groups63AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
6Report on Workforce Civil Liberties65AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
7Greenhouse Gas Emissions Disclosure67AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
+ +\* Because this is a routine proposal, there are no broker non-votes. + +## Election of Directors (Proposal 1) + +The following table provides summary information about each director nominee: + +
NameAgeDirector SinceIndependentFinancial Expert (1)Committee MembershipNumber of Other Public Company Boards
Tench Coxe681993AC, CC
John O. Dabiri462020CC, NCGC
Jen-Hsun Huang631993
Dawn Hudson682013CC Chairperson
Harvey C. Jones731993AC, NCGC
Melissa B. Lora632023AC, NCGC1
Stephen C. Neal(Lead Director)772019NCGC Chairperson
A. Brooke Seawell781997AC Chairperson1
Aarti Shah612020AC, CC1
Mark A. Stevens662008 (2)CC, NCGC
+ +(1) Qualified as an AC financial expert. +(2) Previously served as a member of our Board from 1993 until 2006. + +The Board recommends that you vote FOR each director nominee. + +## Recent Refreshment and Nominee Qualifications + +Our director nominees reflect a broad range of competencies, professional experience, and backgrounds, and contribute diverse viewpoints and perspectives to our Board. The Board seeks to balance the benefits of continuity and institutional knowledge of our longer-serving directors with the introduction of new perspectives and ideas through the appointment of new directors. In furtherance of this approach, one-third of our independent directors have joined the Board since 2020. The Board, through the NCGC, actively oversees Board composition to maintain an appropriate mix of skills, experience, and perspectives aligned with the Company's strategy and evolving business needs. As a part of this process, the Board also periodically refreshes committee membership and chairpersons to promote a variety of viewpoints and broaden director engagement across key areas of oversight on the Board committees. + +On May 7, 2026, the Board approved an increase in the size of the Board from ten to eleven directors and appointed Suzanne Nora Johnson to fill the newly created directorship, in each case effective July 13, 2026. Ms. Nora Johnson will join the Board effective July 13, 2026, due to a prior professional commitment. She is expected to join the AC on the effective date of her appointment. + +Nominee Skills, Competencies, and Attributes + +Below are the skills, competencies, and attributes that our NCGC and Board consider important for our directors to have, considering our current business and future market opportunities, and the director nominees who possess them: + +
Senior Leadership & Operations ExperienceIndustry & TechnicalFinancial / Financial CommunityGovernance & Public Company BoardEmerging Technologies & Business ModelsMarketing, Communications & Brand ManagementRegulatory, Legal & Risk ManagementHuman Capital Management Experience
Coxe
Dabiri
Huang
Hudson
Jones
Lora
Neal
Seawell
Shah
Stevens
+ +## Corporate Governance Highlights + +Our Board is committed to strong corporate governance to promote the long-term interests of the Company and our stockholders. We seek a collaborative approach to stockholder issues that affect our business and to ensure that our stockholders see our governance and executive pay practices as well-structured. In Fall 2025, we contacted our top institutional stockholders, representing an aggregate ownership of 34% of our shares, to gain insights into their views on our corporate governance, compensation and corporate sustainability practices. Highlights of our corporate governance practices include: + +√ All directors independent, except for our CEO +√ Independent Lead Director +√ Proxy access +√ Declassified Board +√ Majority voting for directors +√ Active Board oversight of risk management +√ Stockholders can call a special meeting + +√ 75% or greater attendance by each director at meetings of the Board and applicable committees +√ Independent directors frequently meet in executive sessions +√ At least annual Board and committee self-assessments +√ Annual stockholder outreach, including Lead Director participation +√ Stock ownership guidelines for our directors and NEOs + +## Advisory Approval of Executive Compensation for Fiscal 2026 (Proposal 2) + +We are asking our stockholders to cast a non-binding vote, also known as “say-on-pay,” to approve our NEOs’ compensation. The Board believes that our compensation policies and practices are effective in achieving our goals of paying for performance, providing competitive pay so that we may attract and retain a high-caliber executive team, aligning our executives’ interests with those of our stockholders to create long-term value, and achieving simplicity and transparency with our compensation program. The Board and our stockholders have approved holding our “say-on-pay” votes annually. The Board recommends that you vote FOR Proposal 2. + +## Executive Compensation Highlights + +Our executive compensation program is designed to reward performance. We use compensation elements that align our NEOs' interests with those of stockholders to drive long-term value. Our NEOs' compensation is heavily weighted toward performance-based, "at-risk" variable cash and long-term equity awards, which are earned only if the Company achieves pre-established financial metrics, with caps on maximum payouts. In recent years, over 90% of our CEO's and over 45% of our other NEOs' total target pay have been performance-based and at-risk. Additionally, 100% of our CEO's equity awards have been granted solely as PSUs. + +For Fiscal 2026, the CC kept NEO cash compensation constant as compared to Fiscal 2025, but increased target equity opportunities for all NEOs to recognize the complexity and scope of their roles and responsibilities. The CC also set Fiscal 2026 target equity opportunities at levels that aligned total target pay for all NEOs other than our CEO to reflect their comparable, significant contributions and impact on corporate performance. To motivate executives, the CC set the Base Compensation Plan revenue goal for the Fiscal 2026 Variable Cash Plan in line with our record Fiscal 2025 results and the Base Compensation Plan Non-GAAP Operating Income goal for SY PSUs granted in Fiscal 2026 in line with our corresponding Fiscal 2025 Stretch Compensation Plan goal. + +## Financial Performance and Link to Executive Pay + +As described in CD&A, a significant portion of our executive pay opportunities are tied to the achievement of financial measures that drive business value and contribute to our long-term success. The tables below show our goals for the applicable periods that were completed at the end of Fiscal 2026, performance achievement against the goals, and their impacts on our executives' pay. + +
PERFORMANCE GOALS
Variable Cash PlanSY PSUsMY PSUs
Fiscal 2026 RevenuePayout as a % of Target OpportunityFiscal 2026 Non-GAAP Operating Income (1)Shares Eligible to Vest as a % of Target OpportunityFiscal 2024 to 2026 3-Year Relative TSRShares Eligible to Vest as a % of Target Opportunity
Threshold$100.0 billion50%$46.0 billion50%25th percentile25%
Base Compensation Plan$130.0 billion100%$71.0 billion100%50th percentile100%
Stretch Compensation Plan$160.0 billion (2)200%$96.0 billion (2)CEO 150%Other NEOs 200%75th percentileCEO 150%Other NEOs 200%
+ +
PERFORMANCE ACHIEVEMENT AND PAYOUTS
Variable Cash PlanSY PSUsMY PSUs
Performance Achievement for Period Ended Fiscal 2026 (3)$215.9 billion revenue$137.3 billion Non-GAAP Operating Income (1)100th percentile 3-year TSR relative to S&P 500 (1,055%)
Payout as % of Target Opportunity200%CEO 150% Other NEOs 200%CEO 150% Other NEOs 200%
+ +(1) See Reconciliation of Non-GAAP Financial Measures in CD&A for a reconciliation between the non-GAAP financial measures and GAAP results. +(2) Our original Fiscal 2026 Stretch Compensation Plan revenue goal of \$190.0 billion was automatically adjusted to \$160.0 billion, and our original Fiscal 2026 Stretch Compensation Plan Non-GAAP Operating Income goal of \$120.0 billion was automatically adjusted to \$96.0 billion, due to the imposition of additional export controls on the Company's H2O products during the first half of Fiscal 2026. See Fiscal 2026 Compensation Actions and Achievements—Performance Metrics and Goals for Executive Compensation in CD&A for further discussion of these adjustments. +(3) See Performance Metrics and Goals for Executive Compensation in CD&A for a description and further discussion of revenue, Non-GAAP Operating Income, and 3-year relative TSR. + +## Ratification of Selection of PwC as our Independent Registered Public Accounting Firm for Fiscal 2027 (Proposal 3) + +While not required, we are asking our stockholders to ratify the AC's selection of PwC as our independent registered public accounting firm for Fiscal 2027, as we believe it is a matter of good corporate practice. If our stockholders do not ratify the selection, the AC will reconsider the appointment, but may nevertheless retain PwC. Even if the selection is ratified, the AC may select a different independent registered public accounting firm at any time if it determines that such a change would be in the best interests of NVIDIA and our stockholders. The Board recommends that you vote FOR Proposal 3. + +## Stockholder Proposal 4 + +A stockholder is asking our Board to take steps to replace the supermajority voting provisions in our Charter and Bylaws with a simple majority standard. The proposal is advisory only and approval of the proposal would not, by itself, implement a simple majority voting standard. At our 2025 Meeting, the Board proposed and recommended that stockholders approve an amendment and restatement of our Charter to eliminate the remaining supermajority voting provisions, which did not pass. The Board recommends that you vote AGAINST Proposal 4. + +## Stockholder Proposal 5 + +A stockholder is asking our Board to evaluate and report on faith-based community resource groups. NVIDIA is committed to ensuring that our employees are protected and supported through enterprise-wide policies, practices, and reporting mechanisms. Our Board believes that our existing policies, practices, and feedback mechanisms already address the issues raised in Proposal 5. The Board recommends that you vote AGAINST Proposal 5. + +## Stockholder Proposal 6 + +A stockholder is asking our Board to evaluate and report on civil rights and non-discrimination related to DEI. NVIDIA complies with applicable law and is committed to equal employment opportunity requirements. In addition, the Board maintains strong oversight and compliance mechanisms to ensure these policies and practices are effectively implemented. Our Board believes these oversight structures provide sufficient and comprehensive governance and accountability for the policies and practices addressed by Proposal 6. The Board recommends that you vote AGAINST Proposal 6. + +## Stockholder Proposal 7 + +A stockholder is asking our Board to report and disclose GHG emissions from the use of the Company's sold products. Our Board believes Proposal 7 is unnecessary given NVIDIA's history of robust climate reporting and targets, and our future disclosure plans. The Board believes NVIDIA's management is best positioned to make decisions regarding the specific content and methodology of our climate disclosures. The Board recommends that you vote AGAINST Proposal 7. + +## IN MEMORIAM + +## ROBERT K. BURGESS + +The Board of Directors and the entire NVIDIA family mourn the passing of Robert K. Burgess, who served as a valued member of our Board from 2011 to 2025. + +Over his 14 years of dedicated service, Rob provided exceptional leadership and unwavering commitment to the Company and its stockholders, offering invaluable guidance throughout a period of significant growth. Beyond his many professional contributions, Rob was admired for his warmth, candor, and sense of humor. He will be deeply missed. + +# PROXY STATEMENT FOR THE 2026 ANNUAL MEETING OF STOCKHOLDERS - JUNE 24, 2026 + +## Information About the 2026 Meeting + +Your proxy is being solicited on behalf of the Board for use at the 2026 Meeting. Our 2026 Meeting will take place virtually on Wednesday, June 24, 2026 at 9:00 a.m. Pacific Time. + +## Virtual Meeting Philosophy and Benefits + +The Board believes holding the 2026 Meeting in a virtual format invites participation from stockholders regardless of location, while reducing costs to stockholders and the Company. This balance allows the 2026 Meeting to remain focused on matters directly relevant to stockholder interests in an efficient way. We have designed the virtual format to protect stockholder rights, including by offering multiple opportunities to ask questions, publishing answers to questions received for the 2026 Meeting on our Investor Relations website, and providing a replay of the webcast after the 2026 Meeting. + +## Meeting Attendance + +If you were an NVIDIA stockholder as of the close of business on the April 27, 2026 record date, or if you hold a valid proxy, you can attend, ask questions before and during, and vote at our 2026 Meeting at www.virtualshareholdermeeting.com/NVDA2026. Our 2026 Meeting will be held virtually; use the Control Number included on your Notice or printed proxy card to enter. Anyone can also listen to the 2026 Meeting live; if you encounter any difficulties accessing the virtual 2026 Meeting, please call the technical support number available at www.virtualshareholdermeeting.com/NVDA2026. + +A replay of the webcast will be available at www.nvidia.com/proxy through June 24, 2027. + +Even if you plan to attend the 2026 Meeting virtually, we recommend that you also vote by proxy as described below so that your vote will be counted if you later decide not to attend. + +## Asking Questions + +We encourage stockholders to submit questions through our pre-meeting forum at www.proxyvote.com (using the Control Number included on your Notice or printed proxy card), as well as during the 2026 Meeting at www.virtualshareholdermeeting.com/NVDA2026. During the 2026 Meeting, we will answer as many stockholder-submitted questions related to the business of the 2026 Meeting as time permits. We will publish on our Investor Relations website answers to questions as soon as practicable after the 2026 Meeting. We intend to group questions and answers by topic and answer substantially similar questions only once. To promote fairness to all stockholders and efficient use of the Company's resources, we will respond to one question per stockholder. We reserve the right to exclude questions that are not pertinent to company business or are otherwise unsuitable for the conduct of the 2026 Meeting. + +## Quorum and Voting + +To hold our 2026 Meeting, we need a majority of the outstanding shares entitled to vote at the close of business on the April 27, 2026 record date, or a quorum, represented at the 2026 Meeting either by attendance virtually or by proxy. On April 27, 2026, there were 24,220,525,225 shares of common stock outstanding and entitled to vote, meaning 12,110,262,613 shares must be represented at the 2026 Meeting or by proxy to have a quorum. A list of stockholders entitled to vote at the close of business on the record date will be available at our headquarters, 2788 San Tomas Expressway, Santa Clara, California, for 10 days prior to the 2026 Meeting to stockholders of record for any legally valid purpose germane to the 2026 Meeting. To schedule an appointment to view the stockholder list during this period, please contact us at shareholdermeeting@nvidia.com. + +Your shares will count towards the quorum only if you submit a valid proxy or vote at the 2026 Meeting. Abstentions and broker non-votes will count towards the quorum requirement. If there is not a quorum, a majority of the votes present may adjourn the 2026 Meeting to another date. + +For Proposal 1, you may vote FOR or AGAINST each Board nominee, or you may ABSTAIN from voting. For each other matter to be voted on, you may vote FOR or AGAINST or ABSTAIN from voting. + +## Stockholder of Record + +You are a stockholder of record if your shares were registered directly in your name with our transfer agent, Computershare, on April 27, 2026. You can vote shares, change your vote, or revoke your proxy before the final vote at the 2026 Meeting in any of the following ways: + +
VoteChange Your VoteRevoke Your Proxy
Virtually attend and vote at the 2026 Meeting
Via mail, by signing and mailing your proxy card to us before the 2026 Meeting
By telephone or online, by following the instructions provided in the Notice or your proxy materials
Submit another properly completed proxy card with a later date
Send a written notice that you are revoking your proxy to NVIDIA Corporation, 2788 San Tomas Expressway, Santa Clara, California 95051, Attention: Timothy S. Teter, Secretary, or via email to shareholdermeeting@nvidia.com
+ +If you do not vote using any of the ways described above, your shares will not be voted. + +## Street Name Holder + +If your shares were held through a nominee, such as a bank or broker, as of April 27, 2026, then you were the beneficial owner of shares held in “street name,” and you have the right to direct the nominee how to vote those shares for the 2026 Meeting. The nominee should provide you a separate Notice or voting instructions, which you should follow to tell the nominee how to vote. To vote by attending the 2026 Meeting virtually, you must obtain a valid proxy from your nominee. + +If you are a beneficial holder and do not provide voting instructions to your nominee, the nominee will not be authorized to vote your shares on “non-routine” matters, including elections of directors (even if not contested) and executive compensation (including any advisory stockholder votes on executive compensation), and stockholder proposals. This is called a “broker non-vote.” However, the nominee can still register your shares as being present at the 2026 Meeting for determining quorum, and the nominee will have discretion to vote for matters considered by the NYSE to be “routine,” including Proposal 3 regarding the ratification of the selection of our independent registered public accounting firm. If you are a beneficial owner and want to ensure that all of the shares you beneficially own are voted for or against Proposal 3, you must give your nominee specific instructions to do so or the nominee will have discretion to vote on that proposal. + +For Proposals 1, 2, 4, 5, 6, and 7, which are “non-discretionary” items, you MUST give your nominee instructions in order for your vote to be counted. We strongly encourage you to vote. + +Any NVIDIA stockholder whose shares are held in street name by a member brokerage firm may revoke a proxy and vote their shares at the 2026 Meeting only in accordance with applicable rules and procedures of the national stock exchanges, as used by the holder's brokerage firm. + +## Vote Count + +On each matter to be voted upon, stockholders have one vote for each share of NVIDIA common stock owned as of April 27, 2026. Votes will be counted by the inspector of election as follows: + +
Proposal NumberProposal DescriptionVote Required for ApprovalEffect of AbstentionsBroker Non-Votes
1Election of ten directorsMore FOR than AGAINST votesNoneNone
2Advisory approval of our executive compensationMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
3Ratification of the selection of PwC as our independent registered public accounting firm for Fiscal 2027Majority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstN/A*
4Stockholder Proposal: Simple Majority Vote StandardMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
5Stockholder Proposal: Report on Faith-Based Community Resource GroupsMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
6Stockholder Proposal: Report on Workforce Civil LibertiesMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
7Stockholder Proposal: Greenhouse Gas Emissions DisclosureMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
+ +\* Because this is a routine proposal, there are no broker non-votes. + +If you are a stockholder of record and you return a signed proxy card without marking any selections, your shares will be voted FOR each of the nominees listed in Proposal 1, FOR Proposals 2 and 3, and AGAINST Proposals 4, 5, 6, and 7. If any other matter is properly presented at the 2026 Meeting, one of the proxyholders will vote your shares using his best judgment. + +## Vote Results + +Preliminary voting results will be announced at the 2026 Meeting. Final voting results will be published in a current report on Form 8-K, which will be filed with the SEC by June 30, 2026. + +## Proxy Materials + +As permitted by SEC rules, we are making our proxy materials available to stockholders online at www.nvidia.com/proxy. On or about May 12, 2026, we sent stockholders who owned our common stock at the close of business on April 27, 2026 (except those who previously requested electronic or paper delivery) a Notice with instructions on how to access our proxy materials, vote online or by telephone, and elect to receive future proxy materials electronically or in printed form by mail. + +If you choose to receive future proxy materials electronically (via www.proxyvote.com for stockholders of record and www.icsdelivery.com/nvda for street name holders), you will receive an email next year with links to the proxy materials and proxy voting site. + +SEC rules also permit companies and intermediaries, such as brokers, to satisfy Notice and proxy material delivery requirements for multiple stockholders with the same address by delivering a single Notice or set of proxy materials addressed to those stockholders. We follow this practice, known as “householding,” unless we have received contrary instructions from any stockholder at that address. + +If you received more than one Notice or full set of proxy materials, your shares are either registered in more than one name or held in different accounts. Please vote the shares covered by each Notice or proxy card. To modify your instructions so that you receive one Notice or proxy card for each account or name, contact your broker. Your “householding” election will continue until you are notified otherwise or until you revoke your consent. + +To change how you receive proxy materials (electronically or in print), or to request receipt of a separate set of documents to a household, contact our Investor Relations Department through our website at www.nvidia.com, by email to shareholdermeeting@nvidia.com, by phone at (408) 486-2000, or by mail at 2788 San Tomas Expressway, Santa Clara, California 95051. + +We will pay the entire cost of soliciting proxies. Our directors and employees may also solicit proxies in person, by telephone, by mail, via the Internet, or by other means. Our directors and employees will not receive any additional compensation for soliciting proxies. We have also retained MacKenzie Partners on an advisory basis for an approximate fee of \$20,000; they may help us solicit proxies from brokers, bank nominees, and other institutional owners. We may also reimburse brokerage firms, banks, and other agents for the cost of forwarding proxy materials to beneficial owners. + +## 2027 Meeting Deadlines for Submission of Stockholder Proposals, Nomination of Directors, and Other Business + +## Proposals to be Considered for Inclusion in Our Proxy Materials Pursuant to Rule 14a-8 + +Stockholders who wish to present proposals pursuant to Rule 14a-8 promulgated under the Exchange Act for inclusion in our 2027 Meeting proxy materials must submit their proposals in writing to NVIDIA Corporation, 2788 San Tomas Expressway, Santa Clara, California 95051, Attention: Timothy S. Teter, Secretary, or by email to shareholdermeeting@nvidia.com, on or before January 12, 2027. + +## Director Nominations Under Our Proxy Access Bylaw + +A stockholder (or a group of up to 20 stockholders) who has owned at least 3% of the voting power of our outstanding capital stock for at least three continuous years and has complied with the other requirements in our Bylaws may nominate and include in our proxy materials director nominees constituting up to the greater of (a) up to two director candidates or (b) up to 20% of the number of directors in office on the last day that a submission may be delivered. Notice of a proxy access nomination for consideration at our 2027 Meeting must be received following the above instructions not later than the close of business on March 26, 2027, and not earlier than February 24, 2027. If we hold the 2027 Meeting more than 30 days prior to, or delayed by more than 30 days after, the first anniversary of the 2026 Meeting, for written notice by the stockholder to be timely, such notice must be delivered following the above instructions not earlier than the close of business on the 120th day prior to the 2027 Meeting and not later than the close of business on the 90th day prior to the 2027 Meeting or the 10th day following the day on which public announcement of the date of the 2027 Meeting is first made by us, whichever is later. + +## Other Director Nominations and Proposals + +Apart from Rule 14a-8 and the proxy access provision of our Bylaws, under our Bylaws certain procedures must be followed for a stockholder to nominate a director or introduce an item of business at an annual meeting of stockholders. If you wish to nominate a director or introduce an item of business at the 2027 Meeting that is not included in our 2027 Meeting proxy materials, you must do so in writing following the above instructions not later than the close of business on March 26, 2027, and not earlier than February 24, 2027. If we hold the 2027 Meeting more than 30 days prior to, or delayed by more than 70 days after, the first anniversary of the 2026 Meeting, for written notice by the stockholder to be timely, such notice must be delivered following the above instructions not earlier than the close of business on the 120th day prior to the 2027 Meeting and not later than the close of business on the 90th day prior to the 2027 Meeting or the 10th day following the day on which public announcement of the date of the 2027 Meeting is first made by us, whichever is later. + +## Additional Requirements and Information + +We advise you to review our Bylaws, which include additional requirements for advance notice of stockholder proposals, director nominations, and proxy access nominations. We recognize the importance of stockholders' ability to nominate directors to our Board. Accordingly, our Board will consider feedback received through our stockholder engagement process regarding the process and disclosure requirements of our Bylaws for nominating directors and other proposals. Our Board will engage with stockholders of various holdings size regarding any proposed amendments to our Bylaws that would require a nominating stockholder to disclose to us (i) such stockholder's plans to nominate candidates to the boards of other public companies, or disclose prior director nominations or proposals that such stockholder privately submitted to other public companies, or (ii) information about such stockholder's limited partners or business associates beyond the existing requirements of our Bylaws. + +What am I voting on? Electing the 10 director nominees identified below to hold office until the 2027 Meeting and until his or her successor is elected or appointed. + +Vote required for approval: Directors are elected if they receive more FOR votes than AGAINST votes. + +Effect of abstentions: None. + +Effect of broker non-votes: None. + +Our Board currently consists of 10 members. Our nominees include 9 independent directors, as defined by the rules and regulations of Nasdaq, and one NVIDIA officer: Mr. Huang, who serves as our President and CEO. + +All of our directors have one-year terms and stand for election annually. Each nominee is currently a director of NVIDIA previously elected by our stockholders. + +The Board expects the nominees will be available for election. If a nominee declines or is unable to act as a director, your proxy may be voted for any substitute nominee proposed by the Board or the size of the Board may be reduced. + +## Recommendation of the Board + +The Board recommends that you vote FOR the election of each of the following nominees: + +
NameAgeDirector SinceOccupationIndependentFinancial Expert (1)Committee MembershipNumber of Other Public Company Boards
Tench Coxe681993Former Managing Director, Sutter Hill VenturesAC, CC
John O. Dabiri462020Centennial Professor of Aeronautics and Mechanical Engineering, California Institute of TechnologyCC, NCGC
Jen-Hsun Huang631993President & CEO, NVIDIA Corporation
Dawn Hudson682013Former Chief Marketing Officer, National Football League & Former CEO Pepsi-Cola North AmericaCC Chairperson
Harvey C. Jones731993Managing Partner, Square Wave VenturesAC, NCGC
Melissa B. Lora632023Former President, Taco Bell InternationalAC, NCGC1
Stephen C. Neal (Lead Director)772019Chairman Emeritus & Senior Counsel, Cooley LLPNCGC Chairperson
A. Brooke Seawell781997Venture Partner, New Enterprise AssociatesAC Chairperson1
Aarti Shah612020Former Senior Vice President & Chief Information and Digital Officer, Eli Lilly and CompanyAC, CC1
Mark A. Stevens662008 (2)Managing Partner, S-Cubed CapitalCC, NCGC
+ +(1) Qualified as an AC financial expert. +(2) Previously served as a member of our Board from 1993 until 2006. + +## Director Qualifications and Nomination of Directors + +The NCGC identifies, reviews and assesses the qualifications of current and potential directors and recommends nominees for Board approval. In accordance with our Corporate Governance Policies and the NCGC Charter, the NCGC considers a nominee's background and experience to ensure that a broad range of perspectives is represented. The NCGC may conduct appropriate and necessary inquiries into the backgrounds and qualifications of possible candidates and engage a professional search firm to assist the committee in identifying, evaluating, and conducting due diligence on potential director nominees. The NCGC has not established specific age, gender, education, experience, or skill requirements, and instead considers numerous factors regarding the nominee, including those listed below, taking into account our current and future business models: + +- Integrity and candor +- Independence +• Senior leadership and operational experience +• Professional, technical and industry knowledge +• Financial expertise +• Financial community experience (including as an investor in other companies) +- Marketing, communications and brand management background +• Governance and public company board experience +- Experience with emerging technologies and new business models +• Regulatory, legal, and risk management expertise, including in cybersecurity matters + +- Mix of backgrounds, experiences, and perspectives +• Human capital management experience +• Experience in academia +- Willingness and ability to devote substantial time and effort to Board responsibilities and Company oversight +- Ability to represent the interests of the stockholders as a whole rather than special interest groups or constituencies +- All relationships between the proposed nominee and any of our stockholders, competitors, customers, suppliers, or other persons with a relationship to NVIDIA +- For nominees for re-election, overall service to NVIDIA, including past attendance, participation and contributions to the activities of the Board and its committees + +The NCGC and the Board understand the importance of Board refreshment, and strive to maintain an appropriate mix of tenure, professional experiences and backgrounds, skills, and education on the Board. The Board seeks to balance the benefits of continuity and institutional knowledge of our longer-serving directors, with the introduction of new perspectives and ideas through the appointment of new directors. In furtherance of this approach, one-third of our independent directors have joined the Board since 2020. The Board, through the NCGC, actively oversees Board composition to maintain an appropriate mix of skills, experience and perspectives, aligned with the Company's strategy and evolving business needs. Our longer-tenured directors are familiar with our operations and business areas and have the perspective of overseeing our activities from a variety of economic and competitive environments, which enhances the Board's oversight of strategy and risks. Given our growth and the breadth of our product offerings, as well as the increasingly complex macroeconomic and geopolitical factors we face, these experienced directors are a significant asset to the Board. Our newer directors have brought expertise in brand development and cybersecurity, familiarity with technology developments at leading academic institutions, and senior management, operating, and finance experience, all of which are important to supporting NVIDIA as it enters new markets. Each year, the NCGC and Board review each director's individual performance, including the director's past contributions, outside experiences and activities, and committee participation, and determine how his or her experience and skills continue to add value to NVIDIA and the Board. As a part of this process, the Board also periodically refreshes committee membership and chairpersons to help promote a variety of viewpoints and broaden director engagement across key areas of oversight on the Board committees. Since the 2025 Meeting, Ms. Lora and Dr. Dabiri have joined the NCGC. + +Below are the skills, competencies, and attributes that our Board considers important for our directors to have, considering our current business and future market opportunities: + +
Senior Leadership & Operations ExperienceDirectors with senior leadership and operations experience provide informed oversight of our business, and unique experiences and perspectives. They are uniquely positioned to contribute practical insight into business strategy and operations, driving growth, building and strengthening corporate culture, and supporting the achievement of strategic priorities and objectives.
Industry & TechnicalDirectors with industry experience and technical backgrounds facilitate within the Board a deeper understanding of innovations and a technical assessment of our products and services.
Financial/Financial CommunityExperience in financial matters and the financial community assists our Board with review of our operations and finances, including overseeing our financial statements, capital structure and internal controls. Those with a venture capital background also offer valuable stockholder perspectives.
Governance & Public Company BoardDirectors with experience in corporate governance, such as service on boards and board committees, or as executives of other large, public companies, are familiar with the dynamics and operation of a board of directors and the impact that governance policies have on a company. This experience supports our goals of strong Board and management accountability, transparency, and protection of stockholder interests. Public company board experience also helps our directors identify challenges and risks we face as a public company, including oversight of strategic, operational, and compliance-related matters, and stockholder relations.
Emerging Technologies & Business ModelsExperience in emerging technologies and business models is integral to our growth strategies given our unique business model and provides important insights as our business expands into new areas.
Marketing, Communications & Brand ManagementDirectors with experience in marketing, communications, and brand management offer guidance on our products directly marketed to consumers, important perspectives on expanding our market share, and expertise in communicating with our customers and other stakeholders.
Regulatory, Legal & Risk ManagementOur business requires compliance with a variety of regulatory requirements in different jurisdictions. We face new regulatory matters and regulations as our business grows. We are also subject to multiple lawsuits. Directors with experience in governmental, public policy, legal, and risk management areas, including cybersecurity, help provide valuable insights and oversight for our Company.
Human Capital Management ExperienceOur people are critical to our success. Directors with experience in organizational management, talent development, and developing values and culture in a large global workforce provide key insights. Human capital management experience also assists our Board in overseeing executive and employee compensation, development, and engagement.
+ +The NCGC evaluates candidates proposed by stockholders using the same criteria it uses for other candidates. Stockholders seeking to recommend a prospective nominee should follow the instructions under Stockholder Communications with the Board of Directors below. The stockholder(s) must provide timely written notice of such recommended nominee and the stockholder(s) and such recommended nominee must satisfy the other requirements specified in our Bylaws and Corporate Governance Policies, both of which are available on the Investor Relations section of our website at www.nvidia.com. Stockholders are advised to review these documents. + +## Proxy Access + +Our Board voluntarily adopted proxy access. We will include in our proxy statement information regarding the greater of (i) up to two director candidates or (ii) up to 20% of the number of directors in office on the last day that a submission may be delivered, if nominated by a stockholder (or group of up to 20 stockholders) owning at least 3% of the voting power of our outstanding capital stock for at least three continuous years. The stockholder(s) must provide timely written notice of such nomination and the stockholder(s) and nominee must satisfy the other requirements specified in our Bylaws. This summary of our proxy access rules is not intended to be complete and is subject to limitations set forth in our Bylaws and Corporate Governance Policies, both of which are available on the Investor Relations section of our website at www.nvidia.com. Stockholders are advised to review these documents. The NCGC did not receive any stockholder nominations during Fiscal 2026. + +## Our Director Nominees + +The biographies below include information, as of the date of this proxy statement, regarding the particular experience, qualifications, attributes or skills of each director, relative to the skills matrix above, that led the NCGC and Board to believe that he or she should continue to serve on the Board. + +TENCH COXE +Former Managing Director, Sutter Hill Ventures + +
Age: 68Director Since: 1993Committees: AC, CC
Other Current Public Company Boards: NoneIndependent Director
+ +Tench Coxe was a Managing Director of Sutter Hill Ventures, a venture capital investment firm, from 1989 to 2020, where he focused on investments in the IT sector. Prior to joining Sutter Hill Ventures in 1987, he was Director of Marketing and MIS at Digital Communication Associates. He was a director of Artisan Partners Asset Management Inc., an institutional money management firm from 1995 to 2025. Mr. Coxe holds a BA degree in Economics from Dartmouth College and an MBA degree from Harvard Business School. + +Mr. Coxe brings to the Board expertise in financial and transactional analysis and provides valuable perspectives on corporate strategy and emerging technology trends. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. + +![](images/b3ff9460978d478d862965e1ac7734b2e3e757b485445bb6c229768ba78ce7cb.jpg) + +Financial/Financial +Community + +![](images/e5de7003699e25a7c2de1d03a47ed521970ae36181219283b2348a0f243467be.jpg) + +Governance & Public +Company Board + +![](images/1aefa0884891ca6247e99ce13af05f3a8abaf614a52a68e9e7d3e4ac5a14c3d7.jpg) + +Emerging Technologies & Business Models + +![](images/c4844b75905df4790625042bdab6cf1e37e138b9c3d1fe244fd7b1aee2171c38.jpg) + +Human Capital +Management +Experience + +JOHN O. DABIRI +Centennial Professor of Aeronautics and Mechanical Engineering, California Institute of Technology + +
Age: 46Director Since: 2020Committees: CC, NCGC
Other Current Public Company Boards: NoneIndependent Director
+ +John O. Dabiri is the Centennial Professor of Aeronautics and Mechanical Engineering at the California Institute of Technology. He is a recipient of the National Medal of Science, a MacArthur Foundation "Genius Grant," the National Science Foundation Alan T. Waterman Award, and the Presidential Early Career Award for Scientists and Engineers. He heads the Dabiri Lab, which conducts research at the intersections of fluid mechanics, energy and environment, and biology. From 2021 to January 2025, he served on President Biden's Council of Advisors on Science and Technology (PCAST) and Energy Secretary Granholm's Energy Advisory Board (SEAB). From 2015 to 2019, he served as a Professor of Civil and Environmental Engineering and of Mechanical Engineering at Stanford University, where he was recognized with the Eugene L. Grant Award for Excellence in Teaching. From 2005 to 2015, he was a Professor of Aeronautics and Bioengineering at the California Institute of Technology, during which time he also served as Director of the Center for Bioinspired Wind Energy, Chair of the Faculty, and Dean of Students. Dr. Dabiri is a Fellow of the American Physical Society, where he previously served as Chair of the Division of Fluid Dynamics. He also serves on the board of trustees of the Gordon and Betty Moore Foundation and previously served as a member of the National Academies' Committee on Science, Technology, and Law. Dr. Dabiri holds a PhD degree in Bioengineering and an MS degree in Aeronautics from the California Institute of Technology, and a BSE degree summa cum laude in Mechanical and Aerospace Engineering from Princeton University. + +Dr. Dabiri brings to the Board a versatile research background and cutting-edge expertise in various engineering fields, along with a proven record of successful innovation. + +![](images/15b087846a63426149fe1b1e9a37dea7dc183e36acd51e12645f20002d845f5b.jpg) + +Industry & Technical + +![](images/e8655484589e875f1967b38c864fc1b3573fc72172686d4f11a8fcd5019ea5a8.jpg) + +Emerging Technologies & Business Models + +## JEN-HSUN HUANG + +## President and Chief Executive Officer, NVIDIA Corporation + +Age: 63 + +Director Since: 1993 + +Committees: None + +## Other Current Public Company + +## Boards: None + +Jen-Hsun Huang founded NVIDIA in 1993 and has served since its inception as President, Chief Executive Officer, and a member of the Board. + +Since its founding, NVIDIA has pioneered accelerated computing. The Company's invention of the GPU in 1999 sparked the growth of the PC gaming market, redefined computer graphics, and ignited the era of modern AI. NVIDIA is now driving the platform shift of accelerated computing and generative AI, transforming the world's largest industries and profoundly impacting society. + +Mr. Huang has been elected to the National Academy of Engineering and in 2026 was appointed to the President's Council of Advisors of Science and Technology. He is a recipient of the Semiconductor Industry Association's highest honor, the Robert N. Noyce Award; the IEEE Founder's Medal; the Dr. Morris Chang Exemplary Leadership Award; and honorary doctorate degrees from Taiwan's National Chiao Tung University, National Taiwan University, Oregon State University, Huazhong University of Science and Technology and Linköping University. He has been named the world's best CEO by Fortune, the Economist, and Brand Finance, as well as one of TIME magazine's 100 most influential people. + +Prior to founding NVIDIA, Mr. Huang worked at LSI Logic, a semiconductor and software company, and Advanced Micro Devices, a global semiconductor company. He holds a BSEE degree from Oregon State University and an MSEE degree from Stanford University. + +Mr. Huang is one of the technology industry's most respected executives, having taken NVIDIA from a startup to a world leader in accelerated computing. Under his guidance, NVIDIA has compiled a record of consistent innovation and sharp execution, marked by products that have gained strong market share. + +![](images/64101faaf0a86b91bb06ee9135d29ce6de68bd273059abe838ca6405a4e112f8.jpg) + +Senior Leadership & Operations Experience + +![](images/51fd4f1896a051b52304f8c651c238bff96380c3adb315c699bb75a9e1e44489.jpg) + +Industry & Technical + +![](images/48c9a814c52b1584696b4a2ca182f4eace12edc4c8cfa9d88c4e9d3e2f7e3386.jpg) + +Financial/Financial +Community + +![](images/45218c132d078443388752db3aa5f6cf48993df261313b814b8a931a8997b2a1.jpg) + +Governance & Public Company Board + +![](images/c4557488cfcae52a48f105e7dade13d3d7a8478c0beaa141b93aac2a8c974d2c.jpg) + +Emerging Technologies & Business Models + +![](images/8ab11363ff2ee93670eb704c258664e8cd6be1d67590bdd818d888fdacdc2ebb.jpg) + +Marketing, +Communications & +Brand Management + +![](images/b114632e338705e871e7f36c1dbff2d92150752cb1854aaeea322f7887feb583.jpg) + +Regulatory, Legal & Risk Management + +![](images/16c9fe7b05cfbb9518e7eb683e2ab2a6dfc24f63696a0c6b9fcea7427d3bf8d5.jpg) + +Human Capital +Management +Experience + +## DAWN HUDSON + +## Former Chief Marketing Officer, National Football League & Former CEO Pepsi-Cola North America + +Age: 68 + +Director Since: 2013 + +Committees: CC + +Other Current Public Company Boards: None + +Independent Director + +Financial Expert + +Dawn Hudson serves on the boards of various companies. From 2014 to 2018, Ms. Hudson served as Chief Marketing Officer for the National Football League. Previously, she served from 2009 to 2014 as Vice Chairman of The Parthenon Group, an advisory firm focused on strategy consulting. She was President and Chief Executive Officer of Pepsi-Cola North America, the beverage division of PepsiCo, Inc. for the U.S. and Canada, from 2005 to 2007 and President from 2002, and simultaneously served as Chief Executive Officer of the foodservice division of PepsiCo, Inc. from 2005 to 2007. Previously, she spent 13 years in marketing, advertising and branding strategy, holding leadership positions at major agencies, such as D'Arcy Masius Benton & Bowles and Omnicom Group Inc. Ms. Hudson currently serves on the board of directors of a private skincare company and a private consumer diagnostic portfolio company. She was a director of P.F. Chang's China Bistro, Inc., a restaurant chain, from 2010 to 2012; of Allergan, Inc., a biopharmaceutical company, from 2008 to 2014; of Lowes Companies, Inc., a home improvement retailer, from 2001 to 2015; of Amplify Snack Brands, Inc., a snack food company, from 2014 to 2018; of Modern Times Group MTG AB, a gaming company, from 2020 to 2023; and of the Interpublic Group of Companies, Inc., an advertising company, from 2011 to 2025. She holds a BA degree in English from Dartmouth College. + +Ms. Hudson brings to the board experience in executive leadership. As a longtime marketing executive, she has valuable expertise and insights in leveraging brands, brand development and consumer behavior. She also has considerable corporate governance experience, gained from more than two decades of serving on the boards of public companies. + +![](images/c8ae876bb12d23c40422ee996f4875e6530b8e92562d63862d771c270214c5ab.jpg) + +Senior Leadership & Operations Experience + +![](images/72a61bfa4f78cb5a8b2cf380603b0a2984e693ea32d519695d605889305bacf3.jpg) + +Financial/Financial Community + +![](images/0579ec03dc9085985a622013fb20dc505a204aa6667260da3c0451cc3dd20661.jpg) + +Governance & Public +Company Board + +![](images/f4b5d951557e72a7cc4d6de4cbb122e56eecafdfd0760a5729873931a74153c2.jpg) + +Marketing, +Communications & +Brand Management + +![](images/3613f2aada17fab3c6538dea9bceab5f47b26cee5b1be44e167618a5053b43c2.jpg) + +Human Capital Management Experience + +## HARVEY C. JONES + +## Managing Partner, Square Wave Ventures + +Age: 73 + +Director Since: 1993 + +Committees: AC, NCGC + +Other Current Public Company Boards: None + +Independent Director + +Financial Expert + +Harvey C. Jones has been the Managing Partner of Square Wave Ventures, a private family office, since 2004. Mr. Jones has been an entrepreneur, high technology executive, and active venture investor for over 30 years. In 1981, he co-founded Daisy Systems Corp., a computer-aided engineering company, ultimately serving as its President and Chief Executive Officer until 1987. Between 1987 and 1998, he led Synopsys, Inc., a major electronic design automation company, serving as its Chief Executive Officer for seven years and then as Executive Chairman. In 1997, Mr. Jones co-founded Tensilica Inc., a privately held technology IP company that developed and licensed high performance embedded processing cores. He served as Chairman of the Tensilica board of directors from inception through its 2013 acquisition by Cadence Design Systems, Inc. Mr. Jones holds a BS degree in Mathematics and Computer Sciences from Georgetown University and an MS degree in Management from Massachusetts Institute of Technology. + +Mr. Jones brings to the board an executive management background, an understanding of semiconductor technologies and complex system design. He provides valuable insight into innovation strategies, research and development efforts, as well as management and development of our technical employees. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. + +![](images/189eb87b08924f499f16dd3b4a0c806aab79e962e8f038f5cc3f3d7a84a22907.jpg) + +Senior Leadership & Operations Experience + +![](images/8abea20a4bd52119fe9e43446139455b69eab50c487b7d28d645cdd0a3e140ac.jpg) + +Industry & Technical + +![](images/134e6c851796306f2c52bd6a852eea3b8a8d4ba10f586cc3084b19193e3b5a44.jpg) + +Financial/Financial +Community + +![](images/195ba2acb11009a1506b5ca0def813b905b40974e3683abb995d218ddc7ea4b6.jpg) + +Governance & Public +Company Board + +![](images/e96dd3e0b3650b8ea241c0567b1a1960a8c07dc6e9f6aa275a587538abb36003.jpg) + +Emerging Technologies & Business Models + +![](images/58cf98b0f0897c49f52142a28ed61eaf65e522ce78cbd84b7556925acd8d392e.jpg) + +Marketing, +Communications & +Brand Management + +![](images/95acc2f4ad9438130fad4a0808ce84388cfa8a2ddecad02585164e07318a94ca.jpg) + +Human Capital +Management +Experience + +## MELISSA B. LORA + +## Former President, Taco Bell International + +
Age: 63Director Since: 2023Committees: AC, NCGC
Other Current Public Company Boards: Conagra Brands, Inc. (since 2019)Independent DirectorFinancial Expert
+ +Melissa B. Lora has served in several senior executive roles over her 31-year career at Taco Bell Corp., a subsidiary of Yum! Brands, Inc., one of the world's largest restaurant companies, including as President of Taco Bell International at her retirement in 2018 and Global Chief Financial and Development Officer at Taco Bell Corp. Ms. Lora served on the board of directors of KB Home, a homebuilding company, from 2004 to April 2024, and was a Lead Independent Director thereof from 2016. She was the chair of the audit committee from 2008 to 2018 at KB Home. She has served on the board of directors of Conagra Brands, Inc., a consumer packaged goods holding company, since 2019 and is the chair of the nominating and corporate governance committee and a member of the human resources committee and the executive committee. Ms. Lora previously served on the board of directors of MGIC Investment Corporation from 2018 to 2022. Ms. Lora holds a BS degree in Finance from California State University-Long Beach and an MBA degree emphasizing Corporate Finance from the University of Southern California. + +Ms. Lora brings to the Board senior management and operating experience as well as finance experience gained in a large corporate setting. Additionally, she has decades of public company board of director experience in a variety of industries and board governance roles. + +
Senior Leadership & Operations ExperienceFinancial/Financial CommunityGovernance & Public Company Board
Emerging Technologies & Business ModelsMarketing, Communications & Brand ManagementHuman Capital Management Experience
+ +## STEPHEN C. NEAL + +## Chairman Emeritus and Senior Counsel, Cooley LLP + +
Age: 77Director Since: 2019Committees: NCGC
Other Current Public Company Boards: NoneIndependent DirectorLead Director
+ +Stephen C. Neal serves as Chairman Emeritus and Senior Counsel of the law firm Cooley LLP, where he was also Chief Executive Officer from 2001 until 2008. In addition to his extensive experience as a trial lawyer on a broad range of corporate issues, Mr. Neal has represented and advised numerous boards of directors, special committees of boards, and individual directors on corporate governance and other legal matters. Prior to joining Cooley in 1995, Mr. Neal was a partner of the law firm Kirkland & Ellis LLP. Mr. Neal served on the board of directors of Levi Strauss & Co. from 2007 to 2021 and as Chairperson from 2011 to 2021. Mr. Neal also is Chairman of the Oversight Board Trust, a perpetual Delaware special purpose trust. Previously, Mr. Neal served as Chairman of the boards of the William and Flora Hewlett Foundation and of the Monterey Bay Aquarium. Mr. Neal holds an AB degree from Harvard University and a JD degree from Stanford Law School. + +Mr. Neal brings to the Board deep knowledge and broad experience in corporate governance as well as his perspectives drawn from advising many companies throughout his career. + +
Senior Leadership & Operations ExperienceGovernance & Public Company BoardMarketing, Communications & Brand Management
Regulatory, Legal & Risk ManagementHuman Capital Management Experience
+ +A. BROOKE SEAWELL +Venture Partner, New Enterprise Associates + +
Age: 78Director Since: 1997Committees: AC
Other Current Public Company Boards: Tenable Holdings, Inc. (since 2017)Independent DirectorFinancial Expert
+ +A. Brooke Seawell has served since 2005 as a Venture Partner at New Enterprise Associates, and was a Partner from 2000 to 2005 at Technology Crossover Ventures. He was Executive Vice President from 1997 to 1998 at NetDynamics, Inc., an application server software company, which was acquired by Sun Microsystems, Inc. He was Senior Vice President and Chief Financial Officer from 1991 to 1997 of Synopsys, Inc., an electronic design automation software company. He serves on the board of directors of Tenable Holdings, Inc., a cybersecurity company, and one privately held company. Mr. Seawell served on the board of directors and as audit chair of Glu Mobile, Inc., a publisher of mobile games, from 2006 to 2014, of Informatica Corp., a data integration software company, from 1997 to 2015, of Tableau Software, Inc., a business intelligence software company, from 2011 to 2019, and of Eargo, Inc., a medical device company, from 2020 to 2022. In 2024, the National Association of Corporate Directors named Mr. Seawell as one of the Top 100 directors in the nation. He also previously served as a member of the Stanford University Athletic Board and on the Management Board of the Stanford Graduate School of Business. Mr. Seawell holds a BA degree in Economics and an MBA degree in Finance from Stanford University. + +Mr. Seawell brings to the Board operational expertise and senior management experience, including knowledge of the complex issues facing public companies, and a deep understanding of accounting principles and financial reporting. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. + +![](images/914d32f28e8c4d0a0933ec9943583d632152fad26816d0c64324e3441f1d6791.jpg) + +Senior Leadership & Operations Experience + +![](images/b733704d849527d36c89e8b3c636432980dd8c734d4d2cf347ba3d1bfe517c7c.jpg) + +Financial/Financial +Community + +![](images/38c6abb844f545fb711ad29c889783d681c94f2b84208c7266fe1487fa47886e.jpg) + +Governance & Public +Company Board + +![](images/40961e24ce77ff62c2a29541cbea8c9effe9be39f1f7ac38762a09dd3df6ea80.jpg) + +Emerging Technologies & Business Models + +![](images/d8e7bcb5c77f85936fcc5263285cc69dd9510a15fecc7600e63b03813e4b9799.jpg) + +Regulatory, Legal & Risk Management + +![](images/22a645cc27e3ee17f6009003ab307ba64a3a0ab0968ca6f8f0fe036b87a2b7ea.jpg) + +Human Capital +Management +Experience + +AARTI SHAH +Former Senior Vice President & Chief Information and Digital Officer, Eli Lilly and Company + +
Age: 61Director Since: 2020Committees: AC, CC
Other Current Public Company Boards: Sandoz Group AG (since 2023)Independent Director
+ +Aarti Shah serves on the boards of various companies and non-profit organizations. Dr. Shah worked at Eli Lilly and Company for 27 years until 2021 and served in several functional and business leadership roles, most recently as Senior Vice President and Chief Information and Digital Officer, as well as Senior Statistician, Research Scientist, Vice President for Biometrics, and Global Brand Development Leader in Lilly's Bio-Medicines business unit. Dr. Shah has served on the board and as a member of the science, innovation & development committee and human capital and ESG committee of Sandoz Group AG, a pharmaceutical company, since 2023. Dr. Shah has served on the board of trustees and as a member of the audit committee and distribution and technology committee of Northwestern Mutual since 2020, a financial services organization. Dr. Shah serves on the board, strategic planning committee, and global and technology committee of St. Jude Children's Research Hospital and American Lebanese Syrian Associated Charities (ALSAC), a non-profit organization. She also serves as the secretary and a trustee of the non-profit organization, Shrimad Rajchandra Mission Dharampur USA. She served on the Indianapolis Public Library Foundation board for the full term of 9 years and on the Center for Interfaith Cooperation for the full term of 4 years. Dr. Shah received her BS and MS degrees in Statistics and Mathematics in India before completing her PhD degree in Applied Statistics from the University of California, Riverside. + +Dr. Shah brings to the Board executive leadership and senior operating experience. Additionally she brings expertise in drug development and technical expertise in the areas of information technology, cybersecurity, advanced analytics, data sciences, and digital health. + +![](images/a37676290dc148a01b5e43f7957fa15de70aba53ed3cae1631a61677c6356906.jpg) + +Senior Leadership & Operations Experience + +![](images/14d7df59de91dfcd10e240981c782e27b1b65202dde97fc06a76b07baa151e15.jpg) + +Industry & Technical + +![](images/d664bd35b08abadc75374eabf1c43fcdbf946f89e7c112d2c9880f0fc820489e.jpg) + +Governance & Public +Company Board + +![](images/1b52df194df3f24e6f47168ab4099203dc53a48ef93956c13438db9971704fd3.jpg) + +Emerging Technologies & Business Models + +![](images/7c6c6f0369f38785e8bc3bbbf9821947743c7843c31013f7a94a4d1bd15dbef8.jpg) + +Marketing, +Communications & +Brand Management + +![](images/eed1607bd9843856e9afc282017a32c59c42196e2feae37e17401ba21f6dae71.jpg) + +Regulatory, Legal & Risk Management + +![](images/70f2577d3e5f72a242b806197428f541c37934e735dfa7813befbf2133365012.jpg) + +## MARK A. STEVENS + +## Managing Partner, S-Cubed Capital + +Age: 66 + +Director Since: 2008 + +(previously served 1993-2006) + +Committees: CC, NCGC + +## Other Current Public Company Boards: None + +## Independent Director + +Mark A. Stevens has been the Managing Partner of S-Cubed Capital, a private family office investment firm, since 2012. He was a Managing Partner from 1993 to 2011 of Sequoia Capital, a venture capital investment firm, where he had been an associate for the preceding four years. Previously, he held technical sales and marketing positions at Intel Corporation, and was a member of the technical staff at Hughes Aircraft Co. Mr. Stevens is a trustee and the chair of investment committee of the University of Southern California. He was a director of Quantenna Communications, Inc., a provider of Wi-Fi solutions, from 2016 until 2019. Mr. Stevens holds a BSEE degree, a BA degree in Economics and an MS degree in Computer Engineering from the University of Southern California, and an MBA degree from Harvard Business School. + +Mr. Stevens brings to the Board a deep understanding of the technology industry, and the drivers of structural change and high-growth opportunities. He provides valuable insight regarding corporate strategy development and the analysis of acquisitions and divestitures. His significant financial community experience gives the Board an understanding of the methods by which companies can increase value for their stockholders. + +![](images/b2647d449082452aed943a473abd9b1d823086174a94a4cfb1b22f4890570bac.jpg) + +Industry & Technical + +![](images/2808b88672bd4491539e0f7225f59765981d10e2f5421fde2ce53eaa1b051f33.jpg) + +Financial/Financial Community + +![](images/9bb7f57759f818363f57021d82e5f3569d8ed7e54621a58dcba8741dee18f421.jpg) + +Governance & Public +Company Board + +![](images/2ecccbccc9eb8a66ee5c37ffba8c232afc691b20838176ef42db9a875d79e493.jpg) + +Emerging Technologies & Business Models + +# Information About the Board of Directors and Corporate Governance + +## Independence of the Members of the Board of Directors + +Nasdaq rules and our Corporate Governance Policies (as further described below) require that a majority of our directors not have a relationship that would interfere with their exercise of independent judgment in carrying out their responsibilities and that they meet any other qualification requirements required by the SEC and Nasdaq. + +After considering all relevant relationships and transactions, our Board determined that, except for Mr. Huang, all of our directors who currently serve on the Board are, and all of our directors who served on the Board during Fiscal 2026 were, “independent” as defined by Nasdaq’s rules and regulations. The Board also determined that all members of our AC, CC, and NCGC are independent under applicable Nasdaq listing standards, and that each of Mr. Seawell, Mr. Jones and Ms. Lora of the AC are “audit committee financial experts” as defined under applicable SEC rules. + +## Board Leadership Structure + +Our Board ensures that each member has an equal voice in the affairs and management of NVIDIA by having an independent Lead Director, rather than a chairperson. The independent directors consider the role and designation of the person to serve as Lead Director on an annual basis. The Board recognizes that different board leadership structures may be appropriate under different circumstances and its annual review includes consideration of whether having a Lead Director continues to best meet NVIDIA's evolving needs and serves in the best interest of its stockholders. + +Our Board believes its current leadership structure is appropriate because: + +- The active involvement of each of our independent directors, combined with the qualifications and significant responsibilities of, and strong oversight by, our Lead Director, provide balance on the Board; +• It promotes independent oversight of our management and affairs; +- It effectively allocates authority, responsibility, and oversight between management and our independent directors; and +- It provides the right foundation to pursue the Company's strategic and operational objectives as our business and operating environment continue to evolve. + +Our CEO has primary responsibility for the operational leadership and strategic direction of the Company. Our Lead Director facilitates our Board's independent oversight of management, promotes communication between management and our Board, and supports our Board's consideration of key governance matters. This arrangement promotes open dialogue among the Board, including quarterly executive sessions of independent directors led by our Lead Director without the presence of our CEO. + +We believe that our current structure best serves our stockholders, without the need to appoint a person to serve as chairperson of the Board. Under our Corporate Governance Policies, the Board may select a chairperson, but if it does not, a Lead Director shall be designated annually by a majority of the independent directors and identified in the Company's proxy statement. These policies help to ensure a robust independent leadership structure on our Board. While the Board has discretion to adopt other leadership structures, including having the Lead Director (or chairperson, if any) and CEO roles filled by a single individual, it would only consider a change if it best aligned with the interests of our stockholders, management, and the Board, and it complied with applicable laws and regulations. If in the future our CEO were to take a leadership position on the Board, such as chairperson, we expect that the Board would continue to appoint an independent Lead Director to maintain a balanced and strong leadership structure and otherwise represent the Board independently from the Company's management team. Any changes to the Board's leadership structure would take into account stockholder views, including through our ongoing stockholder outreach, and would be communicated on our Investor Relations website and in our proxy statement. Our Lead Director may provide input on the design of the Board as requested by the NCGC. In his role as NCGC Chairperson, our Lead Director will continue to lead discussions, provide input, and oversee the design of the Board itself. + +## Our Lead Director + +## Stephen C. Neal + +Director since 2019 + +Chairperson of the NCGC + +## Experience + +√ Extensive experience as a trial attorney +√ Has advised numerous companies, boards and individuals on governance and legal matters +√ Helped clients manage internal and government investigations +√ Executive experience as Cooley LLP's CEO, and board and chairman experience on the Levi Strauss & Co. board of directors + +The Board believes Mr. Neal's experience, breadth of knowledge, and contributions to the Board position him well to provide strong leadership, oversight, and the ability to contribute valuable insight with respect to the Company's business. + +Our Lead Director may require Board consideration of risk matters, including adding them to Board agendas or as topics for executive sessions of the independent members of the Board. The Board believes that Mr. Neal is highly qualified to assist the Board as Lead Director in effectively overseeing the identification, assessment, and management of the Company's exposure to various risks as a result of his extensive risk management, legal, and executive experience. Further information on the Board's oversight of risk management and its work with the CEO to address risk management matters is detailed below under Role of the Board in Risk Oversight. + +Our Lead Director has significant responsibilities, which are detailed in our Corporate Governance Policies, and include the duties listed below. + +## Duties of Our Lead Director + +√ Determining an appropriate schedule of Board meetings and seeking to ensure that the independent members of the Board can perform their duties responsibly while not interfering with the flow of our operations +√ Working with the CEO, and seeking input from other directors and relevant management, to prepare Board meeting agendas +√ Advising the CEO on a regular basis as to the quality, quantity, and timeliness of the information requested by the Board from our management, to provide what is necessary for the independent members of the Board to effectively and responsibly perform their duties. Although management is responsible for preparing materials for the Board, the Lead Director may specifically request the inclusion of certain information +√ Coordinating, developing the agenda for, and moderating executive sessions of the independent members of the Board and acting as principal liaison between them and the CEO on sensitive issues +√ Presiding over Board meetings when the CEO is not present +√ Convening meetings of the independent directors, as necessary or appropriate +√ Performing such other duties as the Board may determine from time to time + +In addition, our Lead Director may represent the Board in communications with stockholders and other stakeholders. The Lead Director makes themself available for consultation with major stockholders pursuant to our Corporate Governance Policies. As Lead Director, Mr. Neal has participated in our annual stockholder outreach meetings and we expect this practice to continue. + +## Committees of the Board of Directors + +The Board has three committees: an AC, a CC, and a NCGC. Each of these committees operates under a written charter, which may be viewed in the Investor Relations section of our website at www.nvidia.com. + +Committee assignments are determined based on background and the expertise that individual directors can bring to a committee. Our Board believes regular committee rotations are a good corporate governance practice and introduce a variety of perspectives and ideas, more fully inform its members regarding the full scope of the Board and our activities, and benefit each committee and the Board as a whole. Since our 2025 Meeting, Ms. Lora and Dr. Dabiri have joined the NCGC. The composition and functions of our committees are set forth below. + +## AC + +## Members as of our 2026 Meeting + +• A. Brooke Seawell (Chairperson) +- Tench Coxe +• Harvey C. Jones +- Melissa B. Lora +- Aarti Shah + +In Fiscal 2026, the AC met four times. Selected highlights from its agenda topics included: capitalization review and strategy, tax, treasury, internal audit, information security, enterprise risk management, and insurance reviews. + +## Committee Role and Responsibilities + +• Oversees our corporate accounting and financial reporting process; +• Oversees our internal audit function; +- Determines and approves the engagement, compensation, retention, and termination of the independent registered public accounting firm; +• Evaluates the performance and qualifications of our independent registered public accounting firm; +- Reviews and approves the retention of the independent registered public accounting firm for permissible audit and non-audit services; +- Confers with management and our independent registered public accounting firm on the results of the annual audit, our quarterly financial statements and results, and the effectiveness of internal control over financial reporting, including those regarding information security; +• Reviews the financial statements to be included in our quarterly reports on Form 10-Q and annual reports on Form 10-K; +- Reviews earnings press releases and the substance of financial information and outlook provided to investors and analysts on earnings calls; +- Adopts and maintains policies regarding preapproval of employment of individuals employed or formerly employed by auditors and engaged on our account; +- Prepares their report as required to be included by SEC rules in our annual proxy statement or annual report on Form 10-K; +- Establishes procedures for the receipt, retention, and treatment of complaints we receive regarding accounting, internal accounting controls or auditing matters and the confidential and anonymous submission by employees of concerns regarding accounting or auditing matters; +• Oversees risks related to financial reporting and exposures, internal audit functions, regulatory, and accounting policies; and +- Reviews and reports on the adequacy and effectiveness of the Company's information security policies and practices and the internal controls regarding information security risks + +## CC + +## Members as of our 2026 Meeting + +• Dawn Hudson (Chairperson) +- Tench Coxe +• John O. Dabiri +- Aarti Shah +• Mark A. Stevens + +In Fiscal 2026, the CC met three times. Selected highlights from its agenda topics included: executive, employee, and director compensation, review of employee support programs, executive protection, and human capital management. + +## Committee Role and Responsibilities + +• Reviews and approves our overall compensation strategy and policies; +• Reviews and recommends to the Board the compensation of our Board members; +• Reviews and approves the compensation of Mr. Huang and other executive officers; +- Reviews and approves corporate performance goals and objectives related to the compensation of our executive officers and other senior management; +- Reviews and approves our CD&A disclosure for inclusion in the proxy statement and annual report on Form 10-K; +- Prepares their report on executive compensation as required to be included by SEC rules in our annual proxy statement or annual report on Form 10-K; +• Administers our stock purchase plans, variable compensation plans, and other similar programs; +• Oversees our human capital management; +• Assesses and monitors whether our compensation policies and programs could create material risks; and +• Oversees risks related to compensation plans, programs and policies + +## NCGC + +## Members as of our 2026 Meeting + +• Stephen C. Neal (Chairperson) +• John O. Dabiri +• Harvey C. Jones +- Melissa B. Lora +• Mark A. Stevens + +In Fiscal 2026, the NCGC met three times. Selected highlights from its agenda topics included: Board recruiting, governmental relations, trade compliance and regulatory matters, stockholder proposals, corporate governance matters, and stockholder outreach. + +## Committee Role and Responsibilities + +• Identifies, reviews, and evaluates candidates to serve as directors; +• Recommends candidates for election to our Board; +• Makes recommendations to the Board regarding committee membership and chairpersons; +• Assesses the performance of the Board and its committees; +• Reviews and assesses our corporate governance principles and practices; +• Monitors changes in corporate governance practices, rules, and regulations; +- Approves related party transactions; +- Reviews and assesses our CS strategy, risks, and opportunities periodically, including related programs and initiatives; +• Oversees and reviews policies and practices on governmental relations, trade compliance, regulatory matters, and related risks; +- Establishes procedures for receiving, retaining, and addressing complaints we receive regarding violations of our Code of Conduct; +• Monitors the effectiveness of our anonymous tip process; and +- Oversees the Company's policies, practices, and investigation procedures in connection with the Company's compliance program + +## Compensation Committee Interlocks and Insider Participation + +At the beginning of Fiscal 2026, the CC initially consisted of Messrs. Burgess and Coxe, Ms. Hudson, and Drs. Dabiri and Shah. After the 2025 Meeting, the CC consisted of Messrs. Coxe and Stevens, Ms. Hudson, and Drs. Dabiri, Drell and + +Shah, until Dr. Drell resigned from the Board in January 2026. No member of the CC is an officer or employee of NVIDIA, and none of our executive officers serve as a member of the board or compensation committee of any entity that has one or more executive officers serving as a member of our Board or CC. Other than Dr. Shah, no member of the CC had a relationship requiring disclosure in Review of Transactions with Related Persons below. + +## Role of the Board in Risk Oversight + +The Board reviews risks, oversees risk management, and delegates oversight of appropriate topics to its committees to support the Company's long-term objectives. The oversight responsibility of our Board and its committees is enabled by management reporting processes, including our ERM process, that are designed to provide visibility to our Board about the identification, assessment, and management of critical risks and management's risk mitigation strategies. + +RISK OVERSIGHT AT NVIDIA + +
Board of Directors
Oversees management of major risks, including:
√ Business model, including AI√ Strategic execution√ Product quality and safety
√ Operational, including supply chain and sourcing√ Regulatory, public policy, legal, intellectual property, and compliance√ Financial and macroeconomic
√ Information security, including cybersecurity√ Brand and reputation√ Business continuity
√ Corporate development, acquisitions, and investments√ Management development√ Enterprise resource planning
√ ERM
+ +
ACCCNCGC
√ Financial statement and earnings materials integrity and reporting√ Financial risk exposures, including investments, cash management, foreign exchange management and insurance coverage√ Disclosure controls and procedures√ Information security and cybersecurity policies and practices and the internal controls regarding information security risks√ Internal audit performance, including auditor functions, performance, and independence√ Accounting and audit principles and policies, and regulatory and accounting initiatives√ Legal and regulatory compliance, particularly as related to the above matters√ ERM√ Compensation policies, plans, practices and programs for directors, executives, and employees√ Human capital management, including recruiting, retention, development, and other workforce matters√ Governance structure, processes and policies, including regulatory changes and other developments√ Stockholder concerns and communications√ Compliance program and effectiveness of our anonymous tip process√ CS, including environmental, social, and corporate governance matters√ Trade compliance and non-financial regulatory matters√ Board and committee composition and board evaluation√ Related party transactions√ Policies and practices related to government relations, public policy, and related expenditures
+ +Management + +Management identifies, evaluates, and mitigates business risks and reports to the Board on them + +Internal Audit + +Provides independent assurance on design and effectiveness of internal controls and governance processes + +Our Board retains direct oversight of strategic risks to NVIDIA and other risk areas not delegated to its committees. Delegation to Board committees enhances risk oversight by allocating authority and responsibility to the committee best suited to oversee specific risks, as outlined in their charters, with escalation to the full Board when appropriate. Committee chairpersons regularly report to the full Board on reviewed matters, including key risks, and collaborate with the Board to ensure effective risk management oversight. The Board believes its leadership structure facilitates robust risk oversight, with the independent Lead Director and independent committees proactively engaging with management. For further details, see Board Leadership Structure above. + +The Board retains full oversight of information security matters, including cybersecurity. The AC reviews the adequacy and effectiveness of the Company's information security policies, practices, and internal controls. The AC receives regular updates on information security from management, including the Chief Security Officer and security team, while the Board receives annual reports on these matters. The AC also meets in executive session with leaders of key control functions, ensuring direct access to management teams and appropriate staffing and resources. + +The Company's ERM process is an annual assessment of its risk environment, integrating risk evaluation with the Company's operations and strategies. This process examines internal and external factors, risk amplifiers, and emerging trends to prioritize mitigation efforts based on the risk horizon. It evaluates the potential impact and likelihood of risks materializing over relevant timeframes, considers future threats and trends, and identifies the actions, strategies, processes, controls, and procedures in place or to be implemented to manage and mitigate these risks. + +## ERM Process + +√ Annual assessment of the Company’s risk environment led by management +√ Identifies and evaluates risks across different timeframes, including short-, intermediate- and long-term +Identifies, assesses, and manages the Company's most significant risks and uncertainties that could materially impact the long-term health of the Company or prevent the achievement of strategic objectives +√ Regular updates reported to senior management, including CEO +√ Overseen and reviewed by the Board and AC, at least annually +√ Board and committees have direct access to management to receive updates on risk exposures and mitigation strategies, and give feedback on key and emerging risks, including cybersecurity, trade compliance, risk management and the ERM process +√ Board and committee agendas adjusted throughout the year by CEO and Lead Director to address emerging risks and key topics +√ Includes interviews with senior management and Board members to identify major risks +√ The Board, its committees, and senior management may engage outside advisors, experts, and consultants to help develop and analyze the Company's risk management and mitigation efforts, and anticipate future threats and trends + +While the Company does not have a Chief Compliance Officer, the ERM process and action plan are reviewed by the CEO, other NEOs reporting directly to the CEO, and senior management members. These leaders are responsible for managing and monitoring key risks within their respective functional areas. The Company's internal audit team administers the ERM program on behalf of management. + +The ERM process is designed to align the Board's risk oversight with the Company's disclosure controls and procedures. Management involved in the ERM process prepares reports that are reviewed by the Board or its committees to ensure that disclosure controls and procedures function effectively. + +## Corporate Governance + +## Board Refreshment + +As discussed above, the NCGC and the Board understand the importance of Board refreshment. On May 7, 2026, the Board approved an increase in its size from ten to eleven directors and appointed Suzanne Nora Johnson to fill the newly created directorship, in each case effective July 13, 2026. Ms. Nora Johnson will join the Board effective July 13, 2026, due to a prior professional commitment. Ms. Nora Johnson is the retired Vice Chairman of The Goldman Sachs Group, Inc., where she spent 21 years in various leadership roles. She currently serves on the board of directors of Pfizer Inc., and serves on the Boards of Trustees of The Brookings Institution, the Broad Foundation, and the University of Southern California. Ms. Nora Johnson will bring to the Board extensive senior leadership experience, experience managing large, complex, global institutions and deep knowledge of global financial markets, risk management, and regulatory issues. Ms. Nora Johnson is not currently a member of the Board and is not a nominee for election at the 2026 Meeting. She is expected to join the AC on the effective date of her appointment. + +## Corporate Governance Policies + +The Board has adopted Corporate Governance Policies to ensure that the Board has the necessary authority and processes in place to review and evaluate our business operations as needed and to make decisions that are independent of our management. These policies include practices the Board follows with respect to its composition and selection, regular evaluations of the Board and its committees, Board meetings and involvement of senior management, senior management development, and Board committees and compensation. These policies may be viewed in the Investor Relations section of our website at www.nvidia.com. + +## Executive Sessions of the Board + +As required under Nasdaq's listing standards, our independent directors meet regularly in scheduled executive sessions at which only independent directors are present, as well as in sessions with the CEO. In Fiscal 2026, our independent directors met in both types of executive sessions at all four of our scheduled quarterly Board meetings. + +## Director Attendance at Annual Meeting + +We expect that our directors will attend each annual meeting, absent a valid reason. All Board members as of our 2025 Meeting attended our 2025 Meeting. + +## Board Self-Assessments + +The NCGC oversees an evaluation process, conducted at least annually, whereby outside counsel for NVIDIA interviews each director to obtain his or her evaluation of the Board as a whole, and of the committees on which he or she serves. The interviews solicit ideas from the directors about, among other things, improving the quality of Board and/or committee oversight, effectiveness regarding strategic direction, financial and audit matters, executive compensation, acquisition activity, and other key topics. The interviews also focus on Board process and identifying specific issues for future discussion. After these evaluations are complete, our outside counsel summarizes the results, reviews them with our Lead Director, and then submits the summary for NCGC discussion. The NCGC Chair and outside counsel then report evaluation results to the full Board. + +In response to the evaluations conducted in Fiscal 2026, our Board determined to focus on geopolitical and regulatory risks, government affairs, supply chain, cybersecurity, AI regulations, growth planning, the use and applications of AI, the Company's strategic roadmap, and management development. The Board also determined to continue to focus on the Board's composition and process for Board refreshment. + +## Director Orientation and Continuing Education + +The NCGC and our General Counsel are responsible for new director orientation and for administering or approving eligible director continuing education programs. Continuing education programs for directors may include a combination of internally developed materials and presentations, programs presented by third parties, and financial and administrative support for attendance at qualifying academic or other independent programs. + +## Director Time Commitment and Outside Board Memberships + +Our directors are expected to devote sufficient time to Board and committee duties and to understanding the Company's business. The NCGC reviews the other commitments of potential Board candidates, and does so annually for existing Board members, to determine if this expectation can be met. In making this determination, the NCGC considers, among other factors, stakeholder guidelines regarding numerical limits on public company boards on which a director may sit. None of our directors serve on more than two public company boards (including NVIDIA), and none of our non-employee directors serve as a CEO or executive officer of a public company. + +## Director Stock Ownership Guidelines + +Our Corporate Governance Policies require each non-employee director to hold shares of our common stock with a total value equal to six times the annual cash retainer for Board service during the period in which he or she serves as a director (or ten times his base salary, in the case of the CEO). The shares may include vested deferred stock, shares held in trust, and shares held by immediate family members, but not unvested equity awards. Non-employee directors have five years after their Board appointment to reach the ownership threshold. Our stock ownership guidelines are intended to further align director interests with stockholder interests. + +Each non-employee director and Mr. Huang currently meets or exceeds the stock ownership requirements. + +## Senior Management Development + +We believe in extensive leadership development. Our internal leadership structure ensures that many senior managers work directly with our CEO to execute our corporate strategies. This approach results in strategic alignment, exposes leaders to a broad spectrum of corporate activities, and presents the Board with a pool of excellent candidates for future promotion. Our CEO also selects senior leaders to engage directly with the Board on key initiatives and provides periodic updates to our Board on management development. + +## Outside Advisors + +The Board and each of its principal committees may retain outside advisors and consultants of their choosing at our expense. The Board does not need management's consent to retain outside advisors and the principal committees do not need consent from the Board or management' to retain outside advisors. + +## Code of Conduct + +Our directors, executives, and employees are expected to conduct themselves with the highest degree of integrity, ethics, and honesty. Our credibility and reputation depend upon their good judgment, ethical standards, and personal integrity. Our Code of Conduct applies to all executive officers, directors, and employees, including our principal executive officer, principal financial officer, and principal accounting officer. The Financial Team Code of Conduct applies to our executive officers, directors, members of our finance department, and all employees involved in the preparation and review of externally-reported periodic financial reports, filings, and documents. We regularly review our Code of Conduct and related policies to ensure that they provide clear guidance to our directors, executives, and employees. We also regularly train our employees on our Code of Conduct and other policies. + +The Code of Conduct and the Financial Team Code of Conduct may be viewed in the Investor Relations section of our website at www.nvidia.com. If we make any amendments to either code, or grant any waiver from a provision of either code to any executive officer or director, we will promptly disclose the nature of the amendment or waiver on our website or in a report on Form 8-K. Information contained on our website is not incorporated by reference into this or any other report we file with the SEC. + +## Insider Trading Policy + +Our Insider Trading Policy governs transactions in our securities and applies to the Board and our employees, contractors, and consultants. We believe our Insider Trading Policy is reasonably designed to promote compliance with insider trading laws, rules and regulations applicable to NVIDIA. Section 16 directors and officers must obtain preclearance before trading in NVIDIA's stock. In addition, we comply with applicable laws and regulations relating to insider trading with respect to transactions in our securities such as open market repurchases of our common stock from time to time. + +Under our Insider Trading Policy, hedging ownership of NVIDIA stock, including but not limited to trading in options, puts, calls, or other derivative instruments related to NVIDIA stock or debt, is not permitted. Additional prohibitions include purchasing NVIDIA stock on margin or holding NVIDIA stock in a margin account, or pledging NVIDIA stock as collateral for a loan. We allow participation in certain exchange funds for portfolio diversification. + +## Corporate Hotline + +We have established an independent corporate hotline to allow any employee, contractor, customer, or partner to confidentially and anonymously submit a complaint about any accounting, internal controls, auditing, Code of Conduct, or other matter of concern (unless prohibited by local privacy laws). + +## Stockholder Communications with the Board of Directors + +Stockholders who wish to communicate with the Board regarding nominations of directors or other matters may do so by sending electronic written communications addressed to Timothy S. Teter, our Secretary, at shareholdermeeting@nvidia.com. All stockholder communications we receive that are addressed to the Board will be compiled by our Secretary. If no particular director is named, letters will be forwarded, depending on the subject matter, to the chairperson of the AC, CC, or NCGC. Matters put forth by our stockholders will be reviewed by the NCGC, which will determine whether these matters should be presented to the Board. The NCGC will give serious consideration to all such matters and will make its determination in accordance with its charter and applicable laws. + +## Majority Vote Standard + +Under our Bylaws, in an uncontested election, stockholders will be given the choice to cast votes FOR or AGAINST the election of directors or to ABSTAIN from such vote. A director shall be elected by the affirmative vote of the majority of the votes cast with respect to that director, meaning the number of shares voted FOR a director must exceed the number of votes cast AGAINST that director. If the votes cast FOR an incumbent director in a non-contested election do not exceed the number of AGAINST votes, such incumbent director shall offer to tender his or her resignation to the Board. The NCGC or other committee that may be designated by the Board will make a recommendation to the Board on whether to accept or reject the resignation or whether other action should be taken. The Board will act on such committee's recommendation and publicly disclose its decision and the rationale within 90 days from the date of certification of the election results. In making their decision, such committee and the Board will evaluate the best interests of the Company and its stockholders and shall consider all factors and information deemed relevant. The director who tenders his or her resignation will not participate in such committee's recommendation or the Board's decision. + +In a contested election, in which the number of nominees exceeds the number of directors to be elected, stockholders will be given the choice to cast FOR or WITHHOLD votes for the election of directors. Our directors will be elected by a plurality of the shares represented at any such meeting or by proxy and entitled to vote on the election of directors at that meeting. The directors receiving the greatest number of FOR votes will be elected. + +In either case, abstentions and broker non-votes will each be counted as present for purposes of determining the presence of a quorum but will otherwise have no effect on the vote. + +## Stockholder Special Meeting Right + +As part of our Board and management's comprehensive review of current corporate governance practices, our Board amended our Bylaws in 2024 to let stockholders owning at least $15\%$ of our shares, and who have owned such shares continuously for at least one year, to request a special meeting of stockholders. Those requestors must also satisfy the disclosure, timing and other requirements set forth in our Bylaws intended to ensure that stockholders receive adequate, timely, and accurate information in connection with a special meeting. Our Board believes that this special meeting right strikes a balance between enabling urgent action and protecting the Company and its stockholders from narrow, short-term interests. + +## Board Meeting Information + +The Board met five times during Fiscal 2026, including meetings during which the Board discussed the strategic direction of NVIDIA, new business and strategic opportunities, the product roadmap, and other matters facing NVIDIA. We expect each Board member to attend each meeting of the Board and the committees on which he or she serves. Each Board member attended 75% or more of the applicable meetings of the Board and of each committee on which he or she served during Fiscal 2026. + +## Corporate Sustainability + +NVIDIA invents computing technologies that improve lives and address global challenges. Our goal is to integrate sound CS principles and practices across the Company. Our Board and management believe that environmental stewardship, social responsibility, and solid governance are important to our business strategy and long-term value creation. While the full Board has ultimate responsibility for CS matters that impact our business, each committee of the Board oversees CS matters across our business operations in the areas that align with their respective responsibilities. The NCGC is responsible for reviewing and discussing with management our policies, issues, and reporting related to sustainability, including overall sustainability strategy, risks, and opportunities, and related programs and initiatives. Our CS team updates the NCGC at least semiannually on these topics, as well as pertinent regulations and stakeholder inputs, and gathers feedback from the NCGC on issues such as climate change and human rights. The CS team also reports on sustainability issues to the full Board annually. + +Our CSSC comprises members of our executive leadership team. The CSSC oversees and provides input on our sustainability strategy and program. Feedback from the Board, the NCGC and CSSC, along with specific input from our executive team, helps to determine the focus and scope of our sustainability strategy and program. + +More information can be found on the Corporate Sustainability section of our website and in our annual Sustainability Report. Information contained on our website or in our annual Sustainability Report is not incorporated by reference into this or any other report we file with the SEC. Refer to “Item 1A. Risk Factors” in our Form 10-K for a discussion of risks and uncertainties we face related to CS. + +## Climate and Efficiency + +In Fiscal 2025, we purchased or generated enough renewable energy to match 100% of our global electricity usage for sites under our operational control. + +In Fiscal 2026, we announced two new emission reduction targets validated by the Science Based Targets Initiative. NVIDIA's goal is to, from a Fiscal 2023 base year: + +- Reduce absolute scope 1 and 2 emissions by $50\%$ by Fiscal 2030 +- Reduce scope 3 emissions intensity from use of sold GPU products by 75% per petaFLOP of computing performance by Fiscal 2030 + +## Public Policy Engagement and Accountability + +Our NCGC oversees and periodically reviews our public policy engagement and accountability. We engage in the public policy process to advance the long-term interests of the Company and its stockholders. NVIDIA's U.S. federal lobbying activities are disclosed quarterly and publicly reported. We belong to trade associations worldwide. Management reports to the NCGC about our policies and practices in connection with government relations, public policy, and related expenditures. + +## Director Compensation + +NVIDIA's non-employee director compensation program is designed to attract and retain a world-class Board of Directors while balancing our stockholders' interests, and aligns with annual service starting on the dates of our annual meetings. We do not pay additional fees for serving as Lead Director or as chairperson or member of our committees, or for meeting attendance. Directors who are also employees do not receive compensation for service on the Board. + +## Roles of the Board, the Compensation Committee, and Compensation Consultant + +The CC reviews our director compensation annually with the assistance of its independent compensation consultant, Semler Brossy, and assesses peer group data, compensation trends, and best practices. Following this review, the CC recommended the non-employee director compensation for the year following our 2025 Meeting, or the 2025 Program, as described below, which the Board approved in December 2024. + +## 2025 Program + +The 2025 Program maintained the same target compensation as the previous year with a total value of \$340,000—slightly below the median paid to peer group non-employee directors (based on the peer group determined in late 2024): + +\$85,000 Annual Cash Retainer Paid Quarterly + +\$255,000 Equity Retainer RSUs (1) + +(1) Represents target annual value of RSUs granted on the day following our 2025 Meeting, or the 2025 Program RSUs. 50% of the 2025 Program RSUs vested on the third Wednesday in November 2025 and 50% will vest on the third Wednesday in May 2026, subject to the director's continuous service. + +The number of shares subject to each director's 2025 Program RSUs equaled the target value of the grant divided by the 30-calendar day trailing average closing price of our common stock that ended the business day before the 2025 Meeting, to smooth the effects of market volatility. + +## New Directors + +Any new non-employee director who is appointed or elected to the Board with an effective date after the 2026 Meeting, or a New Director, will be entitled to receive (i) an annual cash retainer of \$85,000, prorated to reflect the New Director's service, paid quarterly, (ii) an initial grant of RSUs with a target value of \$255,000, converted to a number of shares by dividing such value by the 30-calendar day trailing average closing price of our common stock ending the business day before the New Director's effective date of appointment or election, 1/6th of which will vest approximately every six months following the grant date, and (iii) a second grant of RSUs with a target value of \$255,000, converted to a number of shares by dividing such value by the 30-calendar day trailing average closing price of our common stock ending the business day before the 2026 Meeting, prorated to reflect the New Director's service until the 2027 Meeting, or the New Director Annual Grant. 50% of the New Director Annual Grant will vest on the third Wednesday in May 2027, and the balance will vest on the third Wednesday in November 2026; provided that if the New Director's first day of service is on or after November 1, 2026, the New Director Annual Grant will vest in full on the third Wednesday in May 2027 ((ii) and (iii) collectively referred to as the New Director RSUs). + +## Deferred Compensation Program + +Non-employee directors can elect to defer the settlement of RSUs upon vesting for tax planning purposes. Deferrals can be made to the earlier of (i) the third Wednesday of March of a future year (no sooner than 2027 for the 2025 Program RSUs or 2028 for the New Director RSUs), or the 15th of March of any year that is 2029 or later, or (ii) the director's cessation of service or certain change-in-control events, in accordance with Section 409A of the Internal Revenue Code. + +## Other Compensation/Benefits + +Our directors are reimbursed for expenses incurred in attending Board and committee meetings, and continuing educational programs. We do not offer change-in-control benefits to directors, except for vesting acceleration under our equity plans that applies to all plan award holders if an acquirer does not assume or substitute those awards, provided that the award holder's continuous service with us has not terminated prior to the change-in-control. If a director's service terminates due to death, their RSU grants will immediately vest in full. + +Directors do not receive dividends on unvested, or vested but deferred, RSUs. + +Director Compensation for Fiscal 2026 + +
NameFees Earned or Paid in Cash ($)Stock Awards ($) (1)Total ($)
Robert K. Burgess (2)63,750278,809342,559
Tench Coxe85,000278,809363,809
John O. Dabiri85,000278,809363,809
Persis S. Drell85,000278,809363,809
Dawn Hudson85,000278,809363,809
Harvey C. Jones85,000278,809363,809
Melissa B. Lora85,000278,809363,809
Stephen C. Neal85,000278,809363,809
Ellen Ochoa (3)42,500278,809321,309
A. Brooke Seawell85,000278,809363,809
Aarti Shah85,000278,809363,809
Mark A. Stevens85,000278,809363,809
+ +(1) Amounts shown do not reflect amounts actually received by the director. Instead, these amounts reflect the aggregate full grant date fair value calculated in accordance with ASC 718 for RSU awards granted during the fiscal year. The assumptions used in the calculation are set forth in Note 3 to our consolidated financial statements titled Stock-Based Compensation in our Form 10-K. On June 26, 2025, each non-employee director then serving on the Board received their 2025 Program RSU grant for 1,799 shares, with an ASC 718 grant date fair value per share of \$154.98. +(2) Mr. Burgess passed away in December 2025, prior to the date of the final quarterly cash retainer payment under the 2025 Program. +(3) Dr. Ochoa resigned from the Board in July 2025, prior to the dates of the two final quarterly cash retainer payments under the 2025 Program. + +The following table provides information regarding the aggregate number of unvested RSUs held by each of our non-employee directors as of January 25, 2026: + +
NameRSUsNameRSUs
Robert K. Burgess (1)Melissa B. Lora (3)1,880
Tench Coxe900Stephen C. Neal900
John O. Dabiri900Ellen Ochoa (4)
Persis S. Drell (2)A. Brooke Seawell900
Dawn Hudson900Aarti Shah900
Harvey C. Jones900Mark A. Stevens900
+ +(1) Mr. Burgess passed away in December 2025. Pursuant to the terms of our 2007 Plan, upon his death, unvested RSUs then held by Mr. Burgess accelerated and became fully vested. +(2) Dr. Drell resigned from the Board in January 2026. +(3) Includes initial RSUs granted to Ms. Lora in connection with her appointment to the Board in 2023. +(4) Dr. Ochoa resigned from the Board in July 2025. Pursuant to the terms of our non-employee director compensation program, RSUs previously deferred by Dr. Ochoa were issued on the date of her resignation. + +# Review of Transactions with Related Persons + +Employees, officers, and directors must avoid any activity that conflicts with, or has the appearance of conflicting with, our interests. This policy is included in our Code of Conduct and our Financial Team Code of Conduct. We regularly conduct a review of all related party transactions for potential conflicts of interest and all transactions involving executive officers or directors must be approved by the NCGC in compliance with the Company's policies and Nasdaq listing standards. Except as discussed below, there were no transactions with related persons since the beginning of Fiscal 2026 or currently proposed transactions that would require disclosure in this proxy statement or approval by the NCGC. + +## Transactions with Related Persons + +The daughter and son of Jen-Hsun Huang, our President and CEO and a member of our Board, are employed by the Company. Neither of them shares a household with Mr. Huang, is one of our executive officers, or reports directly to Mr. Huang. Additionally, the son of Dr. Shah, one of our directors, is employed by the Company. He does not share a household with Dr. Shah and is not one of our executive officers. + +The compensation of these individuals was determined in accordance with NVIDIA's compensation practices applicable to employees with comparable qualifications and responsibilities and holding similar positions and without the involvement of Mr. Huang or Dr. Shah, respectively. The total compensation for Fiscal 2026 of the daughter and son of Mr. Huang was approximately \$1,232,000 and \$1,320,000, respectively. The total compensation for Fiscal 2026 of the son of Dr. Shah was approximately \$265,000. + +Each of them has received and continues to be eligible for equity awards on the same general terms and conditions as applicable to employees in similar positions who do not have such family relationships. + +The Jen-Hsun and Lori Huang Foundation, or the Huang Foundation, a charitable organization, has entered into an agreement with CoreWeave, Inc., a third-party cloud services provider, to purchase GPU compute time for donation to university and other non-profit research institutes to develop open science and AI research, of which \$108.3 million has been donated to date. NVIDIA routinely offers engineering services for no charge in the ordinary course of business to developers of NVIDIA accelerated computing and we plan to offer similar engineering services for no charge to certain recipients of GPU compute time grants from the Huang Foundation. + +We have entered into indemnity agreements with our executive officers and directors. The agreements provide that we will indemnify them, under certain circumstances, for liabilities they may be required to pay in actions or proceedings by reason of their position with NVIDIA, and otherwise to the fullest extent permitted under Delaware law, our Charter, and our Bylaws. We intend to execute similar agreements with our future executive officers and directors. See Employment, Severance, and Change-in-Control Arrangements below for a description of the terms of the 2007 Plan related to a change-in-control of NVIDIA. + +During Fiscal 2026, we granted RSUs to our non-employee directors, and RSUs and PSUs to our executive officers (other than Mr. Huang, who received PSUs only). See Director Compensation above and Executive Compensation below. + +## Security Ownership of Certain Beneficial Owners and Management + +The following table sets forth information as of March 23, 2026 as to shares of our common stock beneficially owned by each of our NEOs, each of our directors, all of our directors and executive officers as a group, and all known by us to be beneficial owners of more than $5\%$ of our common stock, unless otherwise indicated in the footnotes to the table. Beneficial ownership is determined in accordance with the SEC's rules and generally includes voting or investment power with respect to securities as well as shares of common stock subject to PSUs or RSUs that will vest within 60 days of March 23, 2026. + +This table is based upon information provided to us by our executive officers and directors. Information about principal stockholders, other than percentages of beneficial ownership, is based solely on Schedules 13G/A filed with the SEC. Unless otherwise indicated and subject to community property laws where applicable, we believe that each of the stockholders named in the table has sole voting and investment power with respect to the shares indicated as beneficially owned. Percentages are based on 24,312,141,810 shares of our common stock outstanding as of March 23, 2026, adjusted as required by SEC rules. + +
Name of Beneficial OwnerShares OwnedShares Issuable Within 60 DaysTotal Shares Beneficially OwnedPercent
NEOs:
Jen-Hsun Huang870,604,104 (1)870,604,1043.58%
Colette M. Kress4,443,523 (2)4,443,523*
Ajay K. Puri3,266,897 (3)3,266,897*
Debora Shoquist1,580,729 (4)1,580,729*
Timothy S. Teter2,687,197 (5)2,687,197*
Directors, not including Mr. Huang:
Tench Coxe30,555,240 (6)30,555,240*
John O. Dabiri13,88890014,788*
Dawn Hudson367,987900368,887*
Harvey C. Jones7,002,787 (7)9007,003,687*
Melissa B. Lora899 (8)9001,799*
Stephen C. Neal183,967 (9)900184,867*
A. Brooke Seawell2,505,707 (10)9002,506,607*
Aarti Shah25,620 (11)25,620*
Mark A. Stevens34,068,672 (12)90034,069,572*
Directors and executive officers as a group (14 persons)957,307,217 (13)6,300957,313,5173.94%
5% Stockholders:
BlackRock, Inc.1,805,935,550 (14)1,805,935,5507.43%
Vanguard Capital Management1,777,408,252 (15)1,777,408,2527.31%
+ +\* Represents less than 1% of the outstanding shares of our common stock. +(1) Includes (a) 528,531,547 shares of common stock held by Jen-Hsun Huang and Lori Huang, as co-trustees of the Jen-Hsun and Lori Huang Living Trust, u/a/d May 1, 1995, or the Huang Trust, (b) 31,421,011 shares of common stock held by The Huang 2012 Irrevocable Trust, of which Mr. Huang and his wife are co-trustees, (c) 109,040,602 shares of common stock held by The Huang Irrevocable Remainder Trust u/a/d 2/19/2016, of which Mr. Huang and his wife are co-trustees, (d) 6,632,667 shares of common stock held by TARG M LLC, of which the Huang Trust is the sole member, (e) 6,632,667 shares of common stock held by TARG S LLC, of which the Huang Trust is the sole member, (f) 30,000,000 shares of common stock held by TARG M2 LLC, of which the Huang Trust is the sole member, (g) 30,000,000 shares of common stock held by TARG S2 LLC, of which the Huang Trust is the sole member, and (h) 58,683,830 shares of common stock held by The Jen-Hsun & Lori Huang Foundation, or the Huang Foundation, of which Mr. Huang and his wife are board members. By virtue of their status as co-trustees of the Huang Trust, The Huang 2012 Irrevocable Trust, and The Huang Irrevocable Remainder Trust, each of Mr. Huang and his wife may be deemed to have shared beneficial ownership of, and to exercise shared voting and investment power over, the shares referenced in (a)-(g). By virtue of their status as board members of the Huang Foundation since 2007, Mr. Huang and his wife may be deemed to have shared beneficial ownership of, and to exercise shared voting and investment power over, the shares referenced in (h), and therefore the Huang Foundation's shares are being reported in accordance with Item 403 of Regulation S-K. Mr. Huang and his wife have no pecuniary interest in the Huang Foundation's shares. +(2) Includes (a) 4,000 shares held by immediate family member 1, (b) 4,000 shares held by immediate family member 2, (c) 712,934 shares held by a limited liability company, the sole member of which is an irrevocable trust of which the trustee is an independent institution, (d) 451,498 shares held by Grantor Retained Annuity Trust 1, of which Ms. Kress is trustee, (e) 451,498 shares held by Grantor Retained Annuity Trust 2, of which Ms. Kress's husband is trustee and exercises sole voting and investment power, (f) 83,060 shares held by a trust, of which Ms. Kress is trustee, (g) 2,032,048 shares held by Grantor Retained Annuity Trust 3, of which Ms. Kress and her husband are co-trustees, and (h) 271,592 shares held by Grantor Retained Annuity Trust 4, of which Ms. Kress and her husband are co-trustees. By virtue of their status as co-trustees of the Grantor Retained Annuity Trusts 3 + +and 4, each of Ms. Kress and her husband may be deemed to have shared beneficial ownership of, and to exercise shared voting and investment power over, the shares referenced in (g) and (h). +(3) Includes (a) 3,018,547 shares of common stock held by the Ajay K Puri Revocable Trust dtd 12/10/2015, of which Mr. Puri is trustee, and (b) 46,360 shares of common stock held by The Puri 2019 Irrevocable Children's Trust dtd 12/06/2019, of which Mr. Puri is one of the trustees. Mr. Puri disclaims beneficial ownership of the shares held by The Puri 2019 Irrevocable Children's Trust, except to the extent of his pecuniary interest therein. +(4) Includes 1,424,603 shares of common stock held by the Debora C. Shoquist Revocable Living Trust dtd 6/13/2002, of which Ms. Shoquist is trustee. +(5) Represents shares of common stock held by the Horne Teter Family Living Trust, dated February 1, 2019, of which Mr. Teter is a co-trustee and exercises shared voting and investment power. +(6) Includes (a) 4,852,480 shares of common stock held in a retirement trust, and (b) 25,671,360 shares of common stock held in The Coxe Revocable Trust, of which Mr. Coxe is a co-trustee and exercises shared voting and investment power. Mr. Coxe disclaims beneficial ownership of the shares held by The Coxe Revocable Trust, except to the extent of his pecuniary interest therein. Mr. Coxe shares pecuniary interest in shares held in his individual name pursuant to a contractual relationship, and disclaims beneficial ownership of these shares, except to the extent of his pecuniary interest therein. +Does not include an additional 23,867 shares of common stock underlying vested RSUs that Mr. Coxe has deferred for future issuance. +(7) Includes 6,933,280 shares of common stock held in the H.C. Jones Living Trust, of which Mr. Jones is trustee. +(8) Does not include an additional 12,878 shares of common stock underlying vested RSUs that Ms. Lora has deferred for future issuance. +(9) Includes (a) 30,428 shares of shares of common stock held by the 2013 Stephen C. Neal Revocable Trust, of which Mr. Neal is trustee, (b) 143,410 shares of common stock held by the Neal/Rhyu Revocable Trust dated 05/02/2017, of which Mr. Neal is a co-trustee and exercises shared voting and investment power, and (c) 7,142 shares of common stock held by the 2013 Michelle S. Rhyu Revocable Trust, of which Mr. Neal's wife is trustee and exercises sole voting and investment power. +(10) Includes 2,500,000 shares of common stock held by The Alexander Brooke Seawell Revocable Trust U/A dated 1/20/2009, of which Mr. Seawell is trustee. +(11) Does not include an additional 9,487 shares of common stock underlying vested RSUs that Dr. Shah has deferred for future issuance. +(12) Includes (a) 7,399,771 shares of common stock held by the 3rd Millennium Trust, of which Mr. Stevens is co-trustee and exercises shared voting and investment power, (b) 15,017,750 shares of common stock held by the Envy Trust u/a/d December 7, 2021, of which Mr. Stevens is trustee, and (c) 108,650 shares of common stock held by the 970 Foundation, of which Mr. Stevens and his wife are board members. By virtue of their status as board members of the 970 Foundation, Mr. Stevens and his wife may be deemed to have shared beneficial ownership of, and to exercise shared voting and investment power over, the shares referenced in (c), and therefore the 970 Foundation's shares are being reported in accordance with Item 403 of Regulation S-K. Mr. Stevens and his wife have no pecuniary interest in the 970 Foundation's shares. +(13) Includes shares owned by all directors and executive officers. +(14) This information is based solely on a Schedule 13G/A, dated January 26, 2024, filed with the SEC on January 26, 2024 by BlackRock, Inc. reporting its beneficial ownership as of December 31, 2023. On an adjusted basis to reflect our June 2024 ten-for-one stock split, the Schedule 13G/A reported that BlackRock had sole voting power with respect to 1,628,565,130 shares and sole dispositive power with respect to 1,805,935,550 shares. BlackRock is located at 50 Hudson Yards, New York, New York 10001. +(15) This information is based solely on a Schedule 13G, dated April 28, 2026, filed with the SEC on April 28, 2026 by Vanguard Capital Management reporting its beneficial ownership as of March 31, 2026. The Schedule 13G reported that Vanguard had sole voting power with respect to 238,857,870 shares and sole dispositive power with respect to 1,777,408,252 shares. Vanguard is located at 100 Vanguard Boulevard, Malvern, Pennsylvania 19355. + +## Proposal 2—Advisory Approval of Executive Compensation + +What am I voting on? A non-binding vote, known as “say-on-pay,” to approve our Fiscal 2026 NEO compensation. + +Vote required for approval: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter. + +Effect of abstentions: Same as a vote AGAINST. + +Effect of broker non-votes: None. + +In accordance with Section 14A of the Exchange Act, we are asking our stockholders to vote on an advisory basis, commonly referred to as “say-on-pay,” to approve the Fiscal 2026 compensation paid to our NEOs as disclosed in the CD&A, the compensation tables and the accompanying narrative discussion. This vote is intended to address the overall compensation of our NEOs and the philosophy, policies, and practices described in this proxy statement, rather than any specific compensation component. + +In response to our stockholders' preference, our Board has adopted a policy of providing for annual “say-on-pay” votes. + +This advisory proposal is not binding on the Board nor us. Nevertheless, the views expressed by our stockholders are important to the Board and, accordingly, the Board and the CC intend to consider the results of this vote in making future NEO compensation decisions. + +## Recommendation of the Board + +The Board recommends that our stockholders adopt the following resolution: + +"RESOLVED, that the Fiscal 2026 compensation paid to the Company's named executive officers, as disclosed pursuant to Item 402 of Regulation S-K, including the Compensation Discussion and Analysis, compensation tables, and narrative discussion, is hereby APPROVED." + +## Executive Compensation + +## Compensation Discussion and Analysis + +This CD&A describes our Fiscal 2026 executive compensation philosophy, design, and process, and how our corporate results affected the payout of performance-based awards. Our Fiscal 2026 NEOs were: Jen-Hsun Huang, President and CEO; Colette M. Kress, EVP and CFO; Ajay K. Puri, EVP, Worldwide Field Operations; Debora Shoquist, EVP, Operations; and Timothy S. Teter, EVP, General Counsel and Secretary. + +## Fiscal 2026 Executive Compensation Summary + +## Continued Focus on Pay for Performance + +NVIDIA's executive compensation program in Fiscal 2026 continued to be guided by a pay for performance philosophy to link competitive NEO pay with our stockholders' interests. $96\%$ of our CEO's and $48\%$ of our other NEOs' Fiscal 2026 total target pay was dependent on corporate performance. + +## Executive Pay Heavily Weighted Towards Equity Awards + +The vast majority of our NEOs' total target pay for Fiscal 2026 was comprised of equity awards: + +• SY PSUs that can be earned based on annual Non-GAAP Operating Income performance, vesting over 4 years, +• MY PSUs that can be earned based on 3-year TSR performance relative to the S&P 500, vesting over 3 years, and +• RSUs vesting over 4 years (for NEOs other than our CEO) + +CEO Fiscal 2026 Target Pay Mix (1) +![](images/8ff263dccdb1f30c2d66efd936506ef91662df6b70562a2976690b5c1c54544b.jpg) + +
+stacked bar chart + +| Category | Percentage (%) | +| :--- | :--- | +| SY PSUs | 44 | +| MY PSUs | 44 | +| Variable Cash | 8 | +| Base Salary | 4 | +
+ +Other NEO Fiscal 2026 Target Pay Mix (1)(2) +![](images/f4a2d8a15f032761df7e5a2cb33419d9342eabb484627fac573a5949f957c8ad.jpg) + +
+stacked bar chart + +| Category | Percentage (%) | +| :--- | :--- | +| SY PSUs | 23 | +| MY PSUs | 23 | +| RSUs | 46 | +| Variable Cash | 6 | +| Base Salary | 2 | +
+ +(1) Based on total target pay as approved by the CC, consisting of annual base salary, and, assuming the Company achieves associated performance goals at a Base Compensation Plan level, target payout opportunity under our Variable Cash Plan and target equity opportunities the CC intended to deliver. +(2) Reflects the average total target pay mix for our NEOs other than our CEO. + +NEOs were also eligible for variable cash awards based on annual revenue, in addition to base salary and other compensation and benefits. + +## Key NEO Compensation Actions for Fiscal 2026 + +- Cash Compensation Remained Constant: no changes compared to Fiscal 2025 base salary or variable cash opportunity +- Increased Target Equity Opportunities: increased target equity opportunities for all NEOs to recognize the complexity and scope of NEOs' roles and responsibilities; by increasing target equity opportunities, aligned total target pay for all NEOs other than our CEO to reflect their comparable, significant contributions and impact on corporate performance +- Continued Increased Rigor of Performance Goals: set the Base Compensation Plan revenue goal for the Fiscal 2026 Variable Cash Plan in line with our record Fiscal 2025 results and the Base Compensation Plan Non-GAAP Operating Income goal for SY PSUs granted in Fiscal 2026 in line with our corresponding Fiscal 2025 Stretch Compensation Plan goal + +Record Performance Resulting in Maximum Payouts + +
Fiscal 2026 RevenueFiscal 2026 Non-GAAP Operating Income (1)3-Year TSR Relative to S&P 500 (Fiscal 2024 to 2026) (2)
Stretch Compensation Plan Goal$160.0 billion (3)$96.0 billion (3)75th percentile
Performance Achievement for Period Ended Fiscal 2026$215.9 billion$137.3 billion100th Percentile (1,055%)
+ +(1) See Reconciliation of Non-GAAP Financial Measures below for a reconciliation between the non-GAAP financial measures and GAAP results. +(2) Represents TSR for purposes of the MY PSU performance goal, calculated using cumulative stock price appreciation with dividends reinvested and the average closing stock price for the 60 trading days preceding the start, and preceding and including the last day, of the three-year performance period. +(3) Our original Fiscal 2026 Stretch Compensation Plan revenue goal of \$190.0 billion was automatically adjusted to \$160.0 billion, and our original Fiscal 2026 Stretch Compensation Plan Non-GAAP Operating Income goal of \$120.0 billion was automatically adjusted to \$96.0 billion, due to the imposition of additional export controls on the Company's H2O products during the first half of Fiscal 2026. See Fiscal 2026 Compensation Actions and Achievements —Performance Metrics and Goals for Executive Compensation below for further discussion of these adjustments. + +As a result of the above performance achievements, each exceeding the CC's pre-established Stretch Compensation Plan goals, our NEOs earned the maximum payouts possible for our Fiscal 2026 Variable Cash Plan, SY PSUs granted in Fiscal 2026 (which remain subject to service-based vesting), and MY PSUs granted in Fiscal 2024. + +## Our Compensation Philosophy and Practices + +NVIDIA is building a one-of-a-kind company that invents the future, builds amazing technologies, and strives to achieve the highest level of craft. To achieve this vision, we must attract and retain a high-caliber executive team while balancing our stockholders' interests. While our CC considers numerous factors in making executive pay decisions, our compensation program is guided by the following philosophies: + +- Pay for Performance: emphasize at-risk and performance-based cash and equity for NEOs based on multiple corporate metrics +• Competitive Pay: structure competitive NEO target compensation to reflect job impact, scope, and responsibilities, that attracts and retains talent +- Stockholder Alignment: align NEO pay with stockholders' long-term interests and consider feedback from our annual stockholder engagement efforts and "say-on-pay" vote +- Simplicity and Transparency: design a compensation program with simple, objective metrics + +In this CD&A, total target pay refers to (i) an NEO's annual base salary, (ii) target variable cash opportunity, which means the potential payout under our Variable Cash Plan, assuming the Company achieves the associated performance goal at a Base Compensation Plan level, and (iii) target equity opportunity, which means the value of the equity opportunities granted during the year that the CC intended to deliver at grant, assuming the Company achieves associated performance goals at a Base Compensation Plan level. + +Our executive compensation program adheres to the following practices: + +## What We Do + +√ Emphasize at-risk, performance-based compensation, with simple, objective performance goals +√ Use annual and multi-year performance targets to determine PSU awards earned +√ Set rigorous performance goals and use different metrics for annual and multi-year awards +√ Require NEOs to provide continuous service for 4 years to fully vest in RSU awards and SY PSU awards earned based on performance +√ Evaluate our program annually based on feedback from stockholder engagement efforts and make adjustments when appropriate +√ Mitigate compensation risks +√ Cap performance-based variable cash and PSU payouts +√ Retain an independent compensation consultant reporting directly to the independent CC +√ Require NEOs to maintain meaningful stock ownership +√ Maintain a clawback policy for performance-based compensation + +## What We Don't Do + +X Enter into agreements with NEOs providing for specific terms of employment or severance benefits +X Give our executive officers special change-in-control benefits +X Provide automatic equity vesting upon a change-in-control (except for the provisions in our equity plans that apply to all employees if an acquiring company does not assume or substitute our outstanding stock awards) +X Give NEOs supplemental retirement benefits +X Provide tax gross-ups +X Pay dividends or the equivalent on unearned or unvested equity +X Permit executive officers, employees, or directors to hedge their ownership of NVIDIA stock or to pledge NVIDIA stock as collateral for a loan (though we do allow participation in certain exchange funds for portfolio diversification) + +The CC's oversight and decision-making for our Fiscal 2026 executive compensation program was a multi-year process: + +![](images/eca982d830b24c33a7019da0651ac75d492064cc61108b74ab2b7f389391be69.jpg) + +
+flowchart + +```mermaid +graph LR + A["December 2024\nCC determined peer companies"] --> B["December 2024 - January 2025\nMembers of management and the Board, including our Lead Director and CC Chair, engaged in stockholder outreach"] + B --> C["March 2025\nCC determined performance goals and approved target compensation levels and PSU and RSU grants, considering stockholder feedback and peer company practices"] + C --> D["March 2026\nCC certified achievement and payouts for Fiscal 2026 Variable Cash Plan, SY PSUs granted in Fiscal 2026 and MY PSUs granted in Fiscal 2024*"] + D --> E["May 2026\nCC oversaw compensation risk assessment; published executive compensation program details in proxy statement"] +``` +
+ +\* The CC is expected to certify achievement and payouts for MY PSUs granted in Fiscal 2026 by March 2028. + +## Roles of the Compensation Committee, Compensation Consultant, and Management + +The roles of (i) our CC, (ii) our independent compensation consultant, and (iii) management, including our CEO and Finance and Human Resources departments, are summarized below. + +For advice regarding our Fiscal 2026 NEO compensation program, our CC utilized Semler Brossy for its experience working with our CC and with compensation committees at other technology companies. Our CC analyzed whether Semler Brossy's role raised any conflict of interests, taking into consideration the following: + +- Semler Brossy did not provide any services directly to NVIDIA (although we paid Semler Brossy on the CC's behalf); +- The percentage of Semler Brossy's total revenue resulting from fees paid by us on the CC's behalf; +Semler Brossy's conflict of interest policies and procedures; +- Any business or personal relationship between Semler Brossy and an executive officer, or between Semler Brossy's individual compensation advisors and an executive officer or any member of our CC; and +• Any NVIDIA stock owned by Semler Brossy or its individual compensation advisors. + +After considering these factors, our CC determined that Semler Brossy's work did not create any conflict of interests. + +Our CC reviews and approves all NEO compensation decisions, with input from Mr. Huang and Semler Brossy. At the CC's direction, Semler Brossy and management recommended a peer group for our Fiscal 2026 executive pay program, which the CC approved. Semler Brossy provided data on CEO compensation from the peer group companies to inform the CC on Mr. Huang's compensation and management gathered peer data from the Radford Global Technology Survey, or the Radford Survey, to inform Mr. Huang's recommendations for other NEOs' compensation. The CC considered Semler Brossy's advice, Mr. Huang's recommendations, and management's proposed Fiscal 2026 performance goals, before making its final decisions on Fiscal 2026 NEO compensation. The CC certified compensation payouts for performance periods that concluded at the end of Fiscal 2026 under the Variable Cash Plan and for SY PSUs granted in Fiscal 2026 and MY PSUs granted in Fiscal 2024. The CC also oversaw management's Fiscal 2026 compensation risk analysis. + +## Peer Companies and Market Compensation Data + +We believe our peers should be companies that compete with us for executive talent, have an established business, market presence, and complexity similar to NVIDIA, and are of similar size, measured by revenue or market capitalization at approximately 1/3 - 3.5x ours. After consulting with management, and based on Semler Brossy's recommendation, the CC determined in December 2024 that the existing peer group remained appropriate for Fiscal 2026, with the following adjustments: (i) removing Adobe Inc., Advanced Micro Devices, Inc., and SAP SE due to their revenue and market capitalization being below our target range (and also, for Adobe and SAP, because of limited direct business relevance); and (ii) adding Alphabet Inc., Amazon.com, Inc., Apple Inc., and Microsoft Corporation as competitors for talent with revenue and market capitalization being within our target range. These changes resulted in the following Fiscal 2026 peer group: + +Fiscal 2026 Peer Group + +
Alphabet Inc. (GOOG)Meta Platforms, Inc. (META)
Amazon.com, Inc. (AMZN)Microsoft Corporation (MSFT)
Apple Inc. (AAPL)Netflix, Inc. (NFLX)
Broadcom Inc. (AVGO)Oracle Corporation (ORCL)
Cisco Systems, Inc. (CSCO)Qualcomm Incorporated (QCOM)
International Business Machines Corporation (IBM)Salesforce, Inc. (CRM)
Intel Corporation (INTC)Visa Inc. (V)
+ +Our CC selected each member of the peer group based on a combination of the factors described above. As a result, while some peer companies were outside our targeted size range, the CC determined they were still appropriate due to their established businesses and relevance as competitors for talent. + +To determine our Fiscal 2026 peer group, the CC reviewed our trailing 12-month revenue (reported through our third quarter results for Fiscal 2025) and market capitalization as of November 2024. They compared these metrics to the 75th percentile, median, and 25th percentile of peer group companies, as follows: + +Revenue (in billions) +![](images/30f1b92577cb03839293b51f489b680d9aba31ded8d320fa7dab93d3ba7e242b.jpg) + +
+bar chart + +| Category | Value ($) | +|---|---| +| 75th Percentile | 229.70 | +| NVIDIA | 113.27 | +| Median | 54.03 | +| 25th Percentile | 40.93 | +
+ +Market Capitalization (in billions) +![](images/92aaf53cc61d37d0a4ec582485f07866a3ed017e38de1bf343acc9383aed5cf6.jpg) + +
+bar chart + +| Category | Value ($) | +|---|---| +| NVIDIA | 3,471.92 | +| 75th Percentile | 1,913.24 | +| Median | 536.35 | +| 25th Percentile | 242.70 | +
+ +For our NEOs other than our CEO, Mr. Huang reviewed market practices and compensation data from the Radford Survey and peer company proxy data for comparable executives before recommending their Fiscal 2026 compensation for approval by the CC. The CC reviewed the same information for peer companies' CEOs, before determining the components of our executive compensation program, as well as total compensation, for all NEOs. We compared the total compensation opportunity for our NEOs and similarly situated executives at the 25th, 50th, and 75th percentiles of peer company data where available, and the CC considered the factors below in setting NEO compensation. + +## Factors Used in Determining Executive Compensation + +In addition to peer data, our CC considers the following factors in making executive compensation decisions. The weight given to each factor may differ among NEOs and each component of pay, and is subject to the CC's sole discretion. + +√ The need to attract and retain talent in a highly competitive industry +√ Stockholder feedback regarding our executive pay +√ The desire for simplicity of the overall program and transparency of the performance metrics +√ An NEO's past performance and anticipated future contributions +√ Our financial performance and forecasted results, as well as our prior financial performance and resulting impact on our executives' compensation +√ Changes in the scale and complexity of our business +√ Each NEO’s current total compensation +√ The scope and complexity of the department(s) or function(s) the NEO manages + +√ Each NEO's unvested equity +√ Internal pay equity relative to similarly situated executives +√ Our CEO's recommendations for the NEOs other than himself, including his understanding of each NEO's performance, capabilities, and contributions +√ Our CC's independent judgment +√ Our philosophy that an NEO's total compensation opportunity and percentage of at-risk pay should increase with responsibility +√ The total compensation cost and the growth in such cost, including from executive compensation, to maintain a responsible cost structure for our compensation programs \* + +## Fiscal 2026 Compensation Actions and Achievements + +## Stockholder Outreach and Feedback + +We value stockholder feedback and conduct an annual outreach program. In Fall 2024, in preparation for Fiscal 2026 compensation decisions, we contacted our top institutional stockholders representing an aggregate ownership of 31% of our shares. Management and the Board, including our Lead Director, met with stockholders holding 30% of our shares to discuss a wide range of topics including executive compensation. Stockholders provided positive feedback on our linkage of pay and performance and shared their views on multi-year performance goals and long equity vesting terms. + +In response to this feedback and considering our 92% say-on-pay approval rate for Fiscal 2024 NEO compensation, our CC maintained generally the same elements and performance-based metrics for Fiscal 2026, but (i) increased target equity opportunities for all NEOs to recognize the complexity and scope of their roles and responsibilities, (ii) by increasing target equity opportunities, aligned total target pay for all NEOs other than our CEO to reflect their comparable, significant contributions and impact on corporate performance, and (iii) increased the rigor of the Company's performance-based executive pay opportunities by referencing our record Fiscal 2025 results and Fiscal 2025 Stretch Compensation Plan goals in setting Fiscal 2026 Base Compensation Plan goals to further motivate our NEOs. Our CC believed that structuring performance-based components of our executive pay program solely around NVIDIA's financial performance goals continued to align management's incentives with stockholder interests. + +## Total Target Compensation Approach + +In setting Fiscal 2026 compensation, our CC reviewed each NEO's total target pay opportunity and distribution across multiple pay elements. Our CC compared Mr. Huang's base salary, target variable cash opportunity, target total cash opportunity, target equity opportunity, and total target pay opportunity against chief executives at peer companies. For our other NEOs, Mr. Huang reviewed their total target pay against similarly situated executives at peer companies where the data was available from the Radford Survey, considering internal pay equity, individual performance, unvested equity levels, and the increasing complexity of their roles. These factors informed Mr. Huang's recommendations to the CC. The CC also considered the factors outlined in Factors Used in Determining Executive Compensation and its compensation objectives for Fiscal 2026. Rather than applying a formula or assigning specific weights to each factor, our CC used its judgment and experience to set total target compensation, the mix of cash and equity, and fixed and at-risk pay opportunities for each NEO. When setting pay elements, the CC evaluated them in the context of the levels of the other pay elements and total target pay to ensure alignment with program objectives. The CC established amounts and designed a structure to reward NEOs with above-market value from equity awards and variable cash incentives only upon exceptional corporate performance. + +## Components of Pay + +The primary components of NVIDIA's Fiscal 2026 executive compensation program are summarized below: + +
Fixed CompensationAt-Risk Compensation
Base SalaryVariable CashSY PSUsMY PSUsRSUs (1)
FormCashCashEquityEquityEquity
Who ReceivesNEOsNEOsNEOsNEOsNEOs (except our CEO, whose equity awards are solely PSUs)
Performance MeasureN/ARevenue (determines cash payout)Non-GAAP Operating Income (determines number of shares eligible to vest)TSR relative to the S&P 500 (determines number of shares eligible to vest)N/A
Performance PeriodN/A1 year1 year3 yearsN/A
Vesting PeriodN/AN/A4 years from grant3 years from grant4 years from grant
Vesting TermsN/AN/AIf at least Threshold goal achieved, 25% on approximately the 1-year anniversary of the grant date; 6.25% quarterly thereafterIf at least Threshold goal achieved, 100% on approximately the 3-year anniversary of the grant period6.25% vests quarterly from the grant date (2)
Timeframe EmphasizedAnnualAnnualLong-termLong-termLong-term
PurposeCompensate for expected day-to-day performanceReward for annual corporate financial performanceAlign with stockholder interests by linking NEO pay to annual operational performance and ongoing stock price performance during the vesting periodAlign with long-term stockholder interests by linking NEO pay to multi-year relative shareholder return and ongoing stock price performance during the vesting periodAlign with stockholder interests by linking NEO pay to ongoing stock price performance
Maximum Amount That Can Be EarnedN/A200% of target opportunity under our Variable Cash Plan150% of Mr. Huang's SY PSU target opportunity and 200% of our other NEOs' respective SY PSU target opportunityUltimate value delivered depends on stock price on date earned shares vest150% of Mr. Huang's MY PSU target opportunity and 200% of our other NEOs' respective MY PSU target opportunityUltimate value delivered depends on stock price on date earned shares vest100% of grantUltimate value delivered depends on stock price on date shares vest
+ +(1) Our CC considers RSUs to be at-risk pay because the realized value depends on our stock price, a financial performance measure. +(2) Reflects vesting schedule for annual RSU grants. + +Our NEOs have insurance benefits and are eligible to participate in our ESPP and 401(k) plan on the same basis as our other employees. We also provide a limited set of perquisites to our NEOs from time to time. See Other Compensation and Benefits below for more information. + +## Pay Mix Considerations + +For Fiscal 2026, the CC decided that cash compensation for all NEOs would remain constant with Fiscal 2025 levels and that the largest portion of NEOs' total target pay would remain in the form of at-risk equity. The CC believes an emphasis on long-term, at-risk opportunities drives results and increases NEO and stockholder alignment, while providing sufficient annual cash compensation to be competitive and retain our NEOs. PSUs and RSUs provide long-term incentives and retention benefits as PSUs require achieving predetermined performance goals, and both PSUs and RSUs are subject to meaningful service-based vesting requirements (3 years for MY PSUs and 4 years for SY PSUs and RSUs). + +Given Mr. Huang's position as CEO, to even more tightly align his interests with stockholders, the CC determined that $100\%$ of his equity grants should be at-risk and performance-based. Consistent with prior years, the CC split Mr. Huang's target equity opportunity evenly between SY PSUs (aligned with our annual corporate financial performance) and MY PSUs (aligned with our 3-year relative shareholder return). + +For other NEOs, the CC maintained a proportional equity mix and granted 50% of the target equity opportunity as RSUs and 50% as PSUs, with the PSU portion evenly split between SY PSUs and MY PSUs. The CC determined this structure balanced performance over short- and long-time horizons while providing a meaningful amount of time-vesting RSUs as a retention benefit. + +## Setting Executive Compensation Values + +For Fiscal 2026, the CC decided that increases to each NEO's total target pay were appropriate in light of the scope and complexity of their roles and responsibilities. + +The CC raised Mr. Huang's total target pay by \$3.5 million, or 11%, to \$35.5 million, all in the form of equity, to align more closely with the median of the updated group of peer company CEOs. The CC split the target equity increase evenly between SY PSUs and MY PSUs to balance short- and long-term performance-based awards. + +For other NEOs, the CC decided, on Mr. Huang's recommendation, to align their total target pay to \$16 million each, to reflect their comparable, significant contributions and impact on corporate performance, by increasing target equity opportunities by \$1 to \$2.5 million. For these NEOs, where peer data was available, the increases resulted in their total compensation approximating the 65th percentile of their respective peers. + +## Determining Equity Award Amounts + +To determine the actual number of RSUs and target numbers of SY PSUs and MY PSUs awarded to our NEOs, the CC divided the value of the target equity they intended to deliver at grant by the 30-calendar day trailing average closing price of our common stock ending on the last day of the calendar month prior to the date of grant. They used this methodology instead of calculating the number of shares of common stock subject to each award based on the stock price on the grant date to smooth the effects of possible market volatility. The CC understands that using a historical average stock price can cause the ASC 718 grant date value of an award, as required to be reported in the Summary Compensation Table and Grants of Plan-Based Awards Table, to be different than the intended value of the target equity opportunity. The CC considered various approaches to granting awards and determined the process described above was appropriate. + +The target number of SY PSUs would become eligible to vest if the Company achieved Fiscal 2026 Non-GAAP Operating Income at Base Compensation Plan. If the Company achieved Fiscal 2026 Non-GAAP Operating Income at Stretch Compensation Plan or more, the maximum SY PSUs eligible to vest would be capped at 150% of Mr. Huang's SY PSU target equity opportunity and at 200% for other NEOs. If the Company achieved Fiscal 2026 Non-GAAP Operating Income at Threshold, the minimum SY PSUs eligible to vest would be 50% of each NEO's SY PSU target equity opportunity. + +The target number of MY PSUs would become eligible to vest if the Company achieved TSR relative to the S&P 500 from the start of Fiscal 2026 to the end of Fiscal 2028, or the 3-Year Relative TSR, at Base Compensation Plan. If the Company achieved 3-Year Relative TSR at Stretch Compensation Plan or more, the maximum MY PSUs eligible to vest would be capped at $150\%$ of Mr. Huang's MY PSU target equity opportunity and at $200\%$ for other NEOs. If the Company achieved 3-Year Relative TSR at Threshold, the minimum MY PSUs eligible to vest would be $25\%$ of each NEO's MY PSU target equity opportunity. + +The CC capped the payout for Mr. Huang's SY PSUs and MY PSUs at $150\%$ of his target equity awards instead of $200\%$ that applied to other NEOs because all of his Fiscal 2026 equity compensation was performance-based. With these caps, Mr. Huang would have the same total upside potential on his equity awards as the other NEOs. However, the other NEOs had $50\%$ of their total equity awards granted as RSUs that vest based solely on continued service and therefore did not provide the same upside potential. + +No PSUs would be eligible to vest if the applicable Threshold performance level was not achieved. Any PSUs determined to be unearned would be cancelled. + +## Performance Metrics and Goals for Executive Compensation + +In March 2025, when making decisions regarding Fiscal 2026 executive compensation, the CC intended for performance goals to be rigorous, uncertain, and set at levels that would motivate our NEOs. As a result, the CC set the Base Compensation Plan revenue goal for the Fiscal 2026 Variable Cash Plan in line with our record Fiscal 2025 results and the Base Compensation Plan Non-GAAP Operating Income goal for SY PSUs granted in Fiscal 2026 in line with our corresponding Fiscal 2025 Stretch Compensation Plan goal. + +The CC also considered the impact of potential regulatory actions, outside of management's control, on the Company's Fiscal 2026 performance. As a result, the CC approved Stretch Compensation Plans for Fiscal 2026 revenue of \$190.0 billion and for Fiscal 2026 Non-GAAP Operating Income of \$120.0 billion, which significantly exceeded record Fiscal 2025 results and Fiscal 2025 Stretch Compensation Plan goals. However, at the time it set the Fiscal 2026 performance goals, the CC also provided that the revenue and Non-GAAP Operating Income Stretch Compensation Plan goals would automatically be reduced to \$160.0 billion and \$96.0 billion, respectively, to the extent that additional export control restrictions were subsequently imposed on the Company's H2O products within the first half of Fiscal 2026. The CC adopted this framework to preserve the alignment between executive compensation and Company performance in the event of significant regulatory developments that could impact our financial results. + +In April 2025, the U.S. government informed NVIDIA that a license would be required for exports of H2O products into the China market. Accordingly, the Fiscal 2026 Stretch Compensation Plan goals for revenue and Non-GAAP Operating Income were automatically adjusted. + +Fiscal 2026 performance metrics and goals for NEO pay were as set forth below: + +PERFORMANCE METRICS + +
Variable Cash PlanSY PSUsMY PSUs
MetricRevenueNon-GAAP Operating IncomeTSR relative to the S&P 500
Timeframe1 year1 year3 years
CC's Rationale for MetricDrives value, contributes to Company's long-term successDrives value, contributes to Company's long-term successAligns directly with long-term shareholder value creation
Focuses on growth in new and existing marketsReflects our annual revenue generation and effective operating expense managementProvides comparison of our stock price performance, including dividends, against a capital market index in which we compete
Distinct, separate metric from Non-GAAP Operating IncomeDistinct, separate metrics from revenueRelative performance goal accounts for macroeconomic factors impacting the market
+ +PERFORMANCE GOALS + +
Variable Cash PlanSY PSUsMY PSUs
Fiscal 2026 RevenuePayout as a % of Target Opportunity (1)Fiscal 2026 Non-GAAP Operating Income (2)Shares Eligible to Vest as a % of Target Opportunity (1)Fiscal 2026 to 2028 3-Year Relative TSR (3)Shares Eligible to Vest as a % of Target Opportunity (1)
Threshold$100.0 billion50%$46.0 billion50%25th percentile25%
Base Compensation Plan$130.0 billion100%$71.0 billion100%50th percentile100%
Stretch Compensation Plan$160.0 billion (4)200%$96.0 billion (4)CEO 150% Other NEOs 200%75th percentileCEO 150% Other NEOs 200%
+ +(1) For achievement between Threshold and Base Compensation Plan, or alternatively between Base Compensation Plan and Stretch Compensation Plan, payouts would be determined using straight-line interpolation. Achievement less than Threshold would result in no payout, and achievement exceeding Stretch Compensation Plan would result in the capped maximum payout. +(2) See Reconciliation of Non-GAAP Financial Measures below for a reconciliation between the non-GAAP financial measures and GAAP results. +(3) MY PSUs granted in Fiscal 2026 cover the Fiscal 2026 to Fiscal 2028 performance period. MY PSUs covering the Fiscal 2024 to Fiscal 2026 performance period were granted in Fiscal 2024 and consist of the same performance goal structure and payout opportunities. +(4) Our original Fiscal 2026 Stretch Compensation Plan revenue goal of \$190.0 billion was automatically adjusted to \$160.0 billion, and our original Fiscal 2026 Stretch Compensation Plan Non-GAAP Operating Income goal of \$120.0 billion was automatically adjusted to \$96.0 billion, due to the imposition of additional export controls on the Company's H2O products during the first half of Fiscal 2026. + +Each of the performance goal levels described above were set by the CC with the following objectives: + +- Threshold was uncertain, but attainable and high enough to create value; represented an appropriately decelerated payout for performance below Base Compensation Plan. +- Base Compensation Plan was uncertain, but attainable with significant effort and execution success; included budgeted investments in future businesses and revenue growth considering macroeconomic conditions and reasonable but challenging growth estimates for ongoing and new businesses. + +\- Stretch Compensation Plan required exceptional achievement; only possible with strong market factors and a very high level of management execution and corporate performance; for Fiscal 2026, included an automatic adjustment in the event of additional export controls on the Company's H2O products in the first half of the fiscal year to preserve the alignment between executive compensation and Company performance in the event of significant regulatory developments outside of management's control. + +## Fiscal 2026 Performance Achievement + +As a result of record performance on strength across all market platforms, led by Data Center demand for our Blackwell architecture and across a diverse and expanding set of customers including cloud providers, hyperscalers, AI model makers, enterprises, startups and sovereigns, revenue and Non-GAAP Operating Income for Fiscal 2026 exceeded their respective Stretch Compensation Plan goals, even when compared to the original Stretch Compensation Plan goals prior to the automatic adjustments described above. Our Fiscal 2024 to Fiscal 2026 3-year TSR relative to the S&P 500 also exceeded its Stretch Compensation Plan goal. + +In March 2026, the CC certified the Company's performance achievement with the following payouts: + +
PERFORMANCE ACHIEVEMENT AND PAYOUTS
Variable Cash PlanSY PSUsMY PSUs (1)
Performance Achievement for Period Ended Fiscal 2026 (2)$215.9 billion revenue$137.3 billion Non-GAAP Operating Income (3)100th percentile 3-year TSR relative to S&P 500 (1,055%)
Payout as % of Target Opportunity200%CEO 150% Other NEOs 200% (4)CEO 150% Other NEOs 200% (5)
+ +(1) Represents performance achievement and payout of MY PSUs granted in Fiscal 2024, with a performance period measured from the start of Fiscal 2024 to the end of Fiscal 2026. +(2) Revenue is GAAP revenue, as the Company reports in its SEC filings. Non-GAAP Operating Income is GAAP operating income, as the Company reports in its SEC filings, excluding stock-based compensation expense, acquisition-related and other costs, and other. Consistent with prior years, 3-year TSR for purposes of the MY PSUs represents cumulative stock price appreciation, with dividends reinvested, and is measured based on the average closing stock price for the 60 trading days preceding the start, and preceding and including the last day, of the 3-year performance period. This averaging period mitigates the impact of one-day or short-term stock price fluctuations at the beginning or end of the performance period. +(3) See Reconciliation of Non-GAAP Financial Measures below for a reconciliation between the non-GAAP financial measures and GAAP results. +(4) 25% of the eligible SY PSUs vested on March 18, 2026, approximately one year after grant, and 6.25% will vest every quarter thereafter for the next three years subject to the NEO's continued service to the Company. +(5) 100% of the eligible MY PSUs granted in Fiscal 2024 vested on March 18, 2026. + +Achievement of goals for MY PSUs granted during Fiscal 2025 and Fiscal 2026 will be determined after the applicable performance periods conclude in January 2027 and January 2028, respectively. + +## Fiscal 2026 Target Compensation Actions and Performance-Based Payouts + +The CC's target Fiscal 2026 compensation actions are summarized below for each NEO, reflecting the target variable cash and intended value of the target equity opportunities, as well as the variable cash earned and PSUs that became eligible to vest. + +The CC considered the factors set forth in Factors Used in Determining Executive Compensation above to set total target pay opportunity for each NEO, which are described in Fiscal 2026 Compensation Actions and Achievements - Setting Executive Compensation Values above. + +The intended values of the target equity opportunities presented in the tables below are equal to the number of shares underlying each NEO's equity awards granted in Fiscal 2026, calculated by assuming achievement at Base Compensation Plan for PSUs, multiplied by the 30-day trailing average closing price of our common stock that the CC used in approving such equity awards, as described above in Determining Equity Award Amounts. These values differ from those reported in the Summary Compensation Table and Grants of Plan-Based Awards Table, which, in accordance with SEC rules, reflect the ASC 718 grant date fair value of each NEO's equity awards based on the single day closing price of our common stock on the date of grant and, for PSUs, assuming a probable outcome of the applicable performance conditions. + +Jen-Hsun Huang + +
President & CEOTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary1,500,000Flat with Fiscal 2025
Variable Cash3,000,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($6,000,000)
Cash4,500,000Flat with Fiscal 2025
SY PSUs15,500,000Up $1.8 million, or 13%, from Fiscal 2025 target, resulting in 119,607 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 150% of target opportunity (179,411 shares) becoming eligible to vest
MY PSUs15,500,000Up $1.8 million, or 13%, from Fiscal 2025 target, resulting in 119,607 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 150% of target opportunity (757,360 shares) becoming eligible to vest
Equity31,000,000Up $3.5 million, or 13%, from Fiscal 2025 target
TOTAL35,500,000Up $3.5 million, or 11%, from Fiscal 2025 target
+ +Colette M. Kress + +
EVP & CFOTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary900,000Flat with Fiscal 2025
Variable Cash300,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($600,000)
Cash1,200,000Flat with Fiscal 2025
SY PSUs3,699,807Up $0.2 million, or 7%, from Fiscal 2025 target, resulting in 28,550 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (57,100 shares) becoming eligible to vest
MY PSUs3,699,807Up $0.2 million, or 7%, from Fiscal 2025 target, resulting in 28,550 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (49,560 shares) becoming eligible to vest
RSUs7,400,386Up $0.5 million, or 7%, from Fiscal 2025, resulting in 57,106 shares granted in Fiscal 2026
Equity14,800,000Up $1.0 million, or 7%, from Fiscal 2025 target
TOTAL16,000,000Up $1.0 million, or 7%, from Fiscal 2025 target
+ +Ajay K. Puri + +
EVP, Worldwide Field OperationsTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary950,000Flat with Fiscal 2025
Variable Cash650,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($1,300,000)
Cash1,600,000Flat with Fiscal 2025
SY PSUs3,600,000Up $0.3 million, or 7%, from Fiscal 2025 target, resulting in 27,779 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (55,558 shares) becoming eligible to vest
MY PSUs3,600,000Up $0.3 million, or 7%, from Fiscal 2025 target, resulting in 27,779 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (47,720 shares) becoming eligible to vest
RSUs7,200,000Up $0.5 million, or 7%, from Fiscal 2025, resulting in 55,559 shares granted in Fiscal 2026
Equity14,400,000Up $1.0 million, or 7%, from Fiscal 2025 target
TOTAL16,000,000Up $1.0 million, or 7%, from Fiscal 2025 target
+ +Debora Shoquist + +
EVP, OperationsTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary850,000Flat with Fiscal 2025
Variable Cash250,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($500,000)
Cash1,100,000Flat with Fiscal 2025
SY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (57,488 shares) becoming eligible to vest
MY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (40,840 shares) becoming eligible to vest
RSUs7,450,000Up $1.3 million, or 20%, from Fiscal 2025, resulting in 57,489 shares granted in Fiscal 2025
Equity14,900,000Up $2.5 million, or 20%, from Fiscal 2025 target
TOTAL16,000,000Up $2.5 million, or 19%, from Fiscal 2025 target
+ +Timothy S. Teter + +
EVP, General Counsel & SecretaryTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary850,000Flat with Fiscal 2025
Variable Cash250,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($500,000)
Cash1,100,000Flat with Fiscal 2025
SY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (57,488 shares) becoming eligible to vest
MY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (40,840 shares) becoming eligible to vest
RSUs7,450,000Up $1.3 million, or 20%, from Fiscal 2025, resulting in 57,489 shares granted in Fiscal 2025
Equity14,900,000Up $2.5 million, or 20%, from Fiscal 2025 target
TOTAL16,000,000Up $2.5 million, or 19%, from Fiscal 2025 target
+ +## Additional Executive Compensation Practices, Policies, and Procedures + +## Other Compensation and Benefits + +When evaluating potential perquisites for our executives, we consider many factors, including the cost to NVIDIA relative to the anticipated business benefit, perceived value to our executives, peer data, as well as corporate governance considerations. + +Due to the exceptionally high public profile and prominence of our CEO, and in accordance with the Board-approved, independently-assessed executive security program, NVIDIA provides Mr. Huang and his family members with comprehensive security protection. In Fiscal 2026, these security arrangements included (i) residential security, security during personal travel, consultation fees, and driver services, (ii) security monitoring, (iii) car expenses, and (iv) travel on private aircraft for all business and occasional personal travel, in each case due to a bona fide business-oriented security concern. The CC annually oversees the nature and cost of executive security measures. + +We do not consider these security arrangements to be personal benefits because they arise from Mr. Huang's duties and responsibilities and are required by the Board's executive security program. Furthermore, we believe these arrangements and costs are reasonable, necessary and in the best interests of NVIDIA and its stockholders, as they enable Mr. Huang to focus on his duties and responsibilities to the Company while reducing security threats and mitigating risks to our business. However, for transparency, we have reported the aggregate incremental costs to NVIDIA for personal security arrangements in the “All Other Compensation” column of the Summary Compensation Table below. + +Private aircraft arranged by the Company is intended for Mr. Huang's business travel. Other employees, including our other NEOs, who are also traveling for the same business may accompany Mr. Huang on board. On those flights, we also allow Mr. Huang and any accompanying NVIDIA employee to bring guests. The flights taken by personal guests may be deemed to be a personal benefit for our NEOs, though at no or de minimis incremental cost to us. Nevertheless, any associated aggregate incremental costs incurred by NVIDIA are included in the “All Other Compensation” column of the Summary Compensation Table below. Our NEOs recognize imputed taxable income with respect to any such flights taken by their personal guests, and are not provided with tax gross-ups on such imputed income. To the extent Mr. Huang has personal travel on NVIDIA-arranged private aircraft, he fully reimburses the Company for associated incremental costs. As a result, we incur no incremental costs from Mr. Huang's personal air travel. + +We also provide medical, vision, dental, and accidental death and disability insurance, matches for health savings account contributions, as well as paid time off and holidays, for our NEOs on the same basis as our other employees. Our NEOs may participate, also on the same basis as our other employees, in our ESPP (unless prohibited by Internal Revenue Service rules) and our 401(k) plan, which included an annual Company match of up to \$11,500. We believe these benefits are consistent with those offered by our peer companies with whom we compete for executive talent. + +In Fiscal 2026, we did not provide any other perquisites or personal benefits to our NEOs. + +## Equity Grant Timing Practices + +The CC approves all equity award grants to our executive officers on or before the grant date. Typically, the CC completes its annual executive compensation review, sets performance goals and target compensation, and grants equity awards in March. This process is described in How We Determine Executive Compensation above. On occasion, the CC may grant equity awards outside of our annual cycle for new hires, promotions, recognition, retention or other purposes. While the CC has discretionary authority to approve such awards, it does not have a practice or policy of granting equity awards in anticipation of the release of material non-public information. In Fiscal 2026, we did not time the release of material non-public information in coordination with equity award grants in a manner that intentionally benefits the value of our executive compensation. + +## Stock Ownership Guidelines + +The Board believes that executive officers should hold a meaningful equity interest in NVIDIA. Since June 2024, our Corporate Governance Policies require the CEO to hold shares valued at ten times his base salary, and our other NEOs to hold shares valued at three times their respective base salaries. Shares that count toward the ownership guidelines include shares held in trust, and shares held by immediate family members, but not unvested equity awards. NEOs have up to five years from appointment to reach their ownership threshold. These guidelines are intended to align NEO interests with stockholder interests. Each NEO currently exceeds the stock ownership requirements. + +## Compensation Recovery Policy + +We have maintained a Compensation Recovery Policy since 2009, and amended it in 2023 to comply with Nasdaq's listing standards. Our policy requires the Company to recover certain incentive compensation from current or former executive officers in connection with certain restatements of financial statements, if the compensation exceeds the amount that the officers would have received based on the restated financial statements, subject to limited exceptions. + +## Tax and Accounting Implications + +Under Section 162(m), compensation paid to each of the Company's “covered employees” that exceeds \$1 million per year is generally non-deductible. The CC considers various factors when making its decisions and retains flexibility to provide non-deductible compensation if consistent with the Company’s executive compensation program goals and the best interests of the Company and its stockholders. The CC may also modify compensation that was initially intended to be exempt from the deduction limit if the modifications are consistent with the Company’s business needs. The CC also considers the impact of Section 409A of the Internal Revenue Code and strives to avoid possible adverse tax consequences. + +Additionally, under ASC 718, the Company records share-based compensation expense for equity compensation over the vesting period of the award. + +## Reconciliation of Non-GAAP Financial Measures + +A reconciliation between our GAAP operating income and non-GAAP operating income is as follows (in millions): + +
Fiscal 2026Fiscal 2025
GAAP operating income$130,387$81,453
Stock-based compensation expense6,3864,737
Acquisition-related and other costs471602
Other56(3)
Non-GAAP Operating Income$137,300$86,789
+ +We believe these non-GAAP financial measures enhance stockholders' overall understanding of our historical financial performance. The presentation of our non-GAAP financial measures is not meant to be considered in isolation nor as a substitute for our financial results prepared in accordance with GAAP, and our non-GAAP financial measures may be different from non-GAAP financial measures used by other companies. + +## Compensation Committee Report + +The Compensation Committee of the Board of Directors oversees the compensation programs of NVIDIA on behalf of the Board of Directors. In fulfilling its oversight responsibilities, the Compensation Committee reviewed and discussed with management the Compensation Discussion and Analysis included in this proxy statement. + +In reliance on the review and discussions referred to above, the Compensation Committee recommended to the Board of Directors that the Compensation Discussion and Analysis be included in the Annual Report on Form 10-K of NVIDIA for the year ended January 25, 2026 and in this proxy statement. + +## Compensation Committee + +Dawn Hudson, Tench Coxe, John O. Dabiri, Aarti Shah and Mark Stevens + +## Risk Analysis of Our Compensation Plans + +Under CC oversight, management assessed the Company's Fiscal 2026 overall compensation programs and policies to identify any resulting potential material risks. The assessment focused on programs with payout variability and participants' ability to directly affect payout and related controls—specifically, the Company's variable cash compensation, equity compensation, and sales incentive compensation programs. The review identified key program terms and risks, and specific risk mitigation features. + +The CC considered these findings and concluded that our compensation programs, which are structured to recognize both short-term and long-term contributions to the Company, do not create risks which are reasonably likely to have a material adverse effect on our business or financial condition. + +The CC believes that the following compensation design features mitigate against risk: + +√ Our compensation program encourages our employees to remain focused on both our short-term and long-term goals, and balances incentives by using a mix of base salary and variable pay +√ We design our variable cash and PSU compensation programs for executives so that payouts are based on achievement of various corporate performance targets to balance upside opportunity and downside risk, and we cap the potential award payout +√ We have internal controls over our financial accounting and reporting which are used to measure and determine the eligible compensation awards under our Variable Cash Plan and our SY PSUs +√ Financial plan target goals and final awards under our Variable Cash Plan and SY PSUs are approved by the CC +√ MY PSUs are designed with a relative goal +√ We have a compensation recovery policy applicable to executive officers that requires NVIDIA to recover certain incentive compensation paid in connection with certain accounting restatements +√ The CC monitors burn rate +√ All executive officer equity awards have multi-year vesting +√ We have stock ownership guidelines that we believe are reasonable and are designed to align our executive officers' interests with those of our stockholders +√ Our insider trading policy prohibits hedging, pledging, using margin accounts, and trading in derivatives involving our common stock, which prevents our employees from insulating themselves from the effects of NVIDIA stock price performance + +## Summary Compensation Table for Fiscal 2026, 2025, and 2024 + +The following table summarizes information regarding the compensation earned by our NEOs during Fiscal 2026, 2025, and 2024. Fiscal 2026, 2025, and 2024 were 52-week years. + +
Name and Principal PositionFiscal YearSalary ($)Stock Awards ($) (1)Non-Equity Incentive Plan Compensation ($) (2)All Other Compensation ($)Total ($) (3)
Jen-Hsun HuangPresident and CEO20261,497,62724,800,5116,000,0004,045,691 (4)36,343,830
20251,486,19938,811,3066,000,0003,568,74649,866,251
2024996,51426,676,4154,000,0002,494,97334,167,902
Colette M. KressEVP and CFO2026898,57712,825,872600,00016,402 (5)14,340,850
2025893,73919,849,891600,00018,90221,362,532
2024896,86311,756,027600,00013,90213,266,792
Ajay K. PuriEVP, Worldwide Field Operations2026948,49712,478,9891,300,00049,840 (5)14,777,327
2025943,39119,277,0461,300,00070,46021,590,897
2024946,68911,320,3531,300,00048,40813,615,450
Debora ShoquistEVP, Operations2026848,65612,912,487500,00033,567 (5)14,294,710
2025844,08717,838,832500,00034,98419,217,903
2024847,0379,687,599500,00024,22911,058,865
Timothy S. TeterEVP, General Counsel and Secretary2026848,65612,912,487500,00015,240 (5)14,276,382
2025844,08717,838,832500,00018,90219,201,821
2024847,0379,687,599500,00013,90211,048,538
+ +(1) Amounts shown do not reflect dollar amounts actually received by the NEO, and may not reconcile to corresponding amounts presented in other tables due to differences in rounding. Instead, these amounts reflect the aggregate full grant date fair value calculated in accordance with ASC 718 for the respective fiscal year for grants of RSUs, SY PSUs, and MY PSUs, as applicable. The assumptions used in the calculation of values of the awards are set forth under Note 3 to our consolidated financial statements titled Stock-Based Compensation in our Form 10-K. The reported grant date fair value stock awards with performance-based vesting conditions assumes the probable outcome of the conditions at Base Compensation Plan for SY PSUs and MY PSUs, determined in accordance with applicable accounting standards. +Assuming Stretch Compensation Plan performance for SY PSUs and MY PSUs granted in each of Fiscal 2026, 2025, and 2024, and a stock price equal to the grant date fair value of the SY PSUs and MY PSUs, the value of stock awards would be: + +
Fiscal YearJen-Hsun HuangColette M. KressAjay K. PuriDebora ShoquistTimothy S. Teter
SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)
202619,166,42418,034,3436,099,9937,350,4835,935,2617,151,9816,141,4437,400,4306,141,4437,400,430
202525,046,74833,170,2118,377,78114,564,6908,136,36614,144,9937,529,33213,089,6697,529,33213,089,669
202423,130,93722,666,27015,613,1881,937,35015,034,8111,865,42312,866,3001,596,47612,866,3001,596,476
+ +(2) As applicable, reflects amounts earned in Fiscal 2026, 2025, and 2024 and paid in March or April of the fiscal year immediately following each respective year pursuant to the respective Variable Cash Plan. For further information, please see our CD&A above. +(3) Due to rounding, numbers presented may not sum precisely to the totals provided. +(4) Reflects (a) the aggregate incremental costs to the Company of residential security, security during personal travel, and consultation fees and driver services (in the amount of \$3,975,533, reflecting the full cost to the Company), security monitoring services, car expenses, and Mr. Huang's personal guests who accompanied him during business travel on private aircraft, though at no or de minimis incremental cost to us, (b) a \$11,500 match of contributions to our 401(k) savings plan, and (c) \$21,904 in life insurance premiums. The 401(k) contribution match and insurance coverage are available to all eligible NVIDIA employees. +(5) Includes matches of contributions to our 401(k) savings plan and imputed income from life insurance coverage. These benefits are available to all eligible NVIDIA employees. For Fiscal 2026, the match of 401(k) contributions was \$11,500 for Ms. Kress, \$10,708 for Mr. Puri, \$10,083 for Ms. Shoquist, and \$10,083 for Mr. Teter; and the dollar values of life insurance premiums were \$4,902 for Ms. Kress, \$39,132 for Mr. Puri, \$23,484 for Ms. Shoquist, and \$5,121 for Mr. Teter. + +## Grants of Plan-Based Awards for Fiscal 2026 + +The following table provides information regarding all grants of plan-based awards that were made to or earned by our NEOs during Fiscal 2026. The information in this table supplements the dollar value of stock awards set forth in the Summary Compensation Table for Fiscal 2026, 2025, and 2024. The PSU and RSU awards set forth in the following table were made under our 2007 Plan. PSUs are eligible to vest based on performance against pre-established criteria. All equity awards listed are subject to service-based vesting. + +
NameType of AwardGrant DateEstimated Possible Payouts Under Non-Equity Incentive Plan Awards (1)Estimated Future Payouts Under Equity Incentive Plan Awards (2)All Other Stock Awards: Number of Shares of Stock or Units (#) (3)Grant Date Fair Value of Stock Awards ($) (4)
Threshold ($)Target ($)Maximum ($)Threshold (#)Target (#)Maximum (#)
Jen-Hsun HuangSY PSU3/10/2559,804119,607179,41112,777,616
MY PSU3/10/2529,902119,607179,41112,022,896
Variable Cash Plan3/3/251,500,0003,000,0006,000,000
Colette M. KressSY PSU3/10/2514,27528,55057,1003,049,997
MY PSU3/10/257,13828,55057,1003,675,242
RSU3/10/2557,1066,100,634
Variable Cash Plan3/3/25150,000300,000600,000
Ajay K. PuriSY PSU3/10/2513,89027,77955,5582,967,631
MY PSU3/10/256,94527,77955,5583,575,991
RSU3/10/2555,5595,935,368
Variable Cash Plan3/3/25325,000650,0001,300,000
Debora ShoquistSY PSU3/10/2514,37228,74457,4883,070,722
MY PSU3/10/257,18628,74457,4883,700,215
RSU3/10/2557,4896,141,550
Variable Cash Plan3/3/25125,000250,000500,000
Timothy S. TeterSY PSU3/10/2514,37228,74457,4883,070,722
MY PSU3/10/257,18628,74457,4883,700,215
+ +(1) Represents range of awards payable under our Fiscal 2026 Variable Cash Plan. +(2) Represents range of shares eligible to be earned with respect to PSUs. +(3) Represents RSUs granted. +(4) Amounts shown in this column do not reflect dollar amounts actually received by the NEO, and may not reconcile to corresponding amounts presented in other tables due to differences in rounding. Instead, these amounts reflect the aggregate full grant date fair value calculated in accordance with ASC 718 for the awards. The assumptions used in the calculation of values of the awards are set forth under Note 3 to our consolidated financial statements titled Stock-Based Compensation in our Form 10-K. With regard to the stock awards with performance-based vesting conditions, the reported grant date fair value assumes the probable outcome of the conditions at Base Compensation Plan performance for SY PSUs and MY PSUs, determined in accordance with applicable accounting standards. + +The following table presents information regarding outstanding equity awards held by our NEOs as of January 25, 2026. + +
NameStock Awards
Number of Units of Stock That Have Not Vested (#)Market Value of Units of Stock That Have Not Vested ($) (1)Equity Incentive Plan Awards: Number of Unearned Shares That Have Not Vested (#)Equity Incentive Plan Awards: Market Value of Unearned Shares That Have Not Vested ($) (1)
Jen-Hsun Huang315,570 (2)59,223,022
757,360 (3)142,133,751
161,085 (4)30,230,822
179,411 (5)33,670,062
286,350 (6)53,739,305
179,411 (7)33,670,062
Colette M. Kress10,970 (8)2,058,740
61,970 (9)11,629,910
213,010 (2)39,975,587
49,560 (3)9,300,925
53,890 (10)10,113,536
53,880 (4)10,111,660
46,399 (11)8,707,700
57,100 (5)10,715,957
95,780 (6)17,975,033
57,100 (7)10,715,957
Ajay K. Puri10,560 (8)1,981,795
59,670 (9)11,198,269
205,120 (2)38,494,870
47,720 (3)8,955,612
52,330 (10)9,820,771
52,340 (4)9,822,648
45,142 (11)8,471,799
55,558 (5)10,426,570
93,020 (6)17,457,063
55,558 (7)10,426,570
Debora Shoquist9,040 (8)1,696,537
51,070 (9)9,584,307
175,540 (2)32,943,592
40,840 (3)7,664,443
48,420 (10)9,086,981
48,420 (4)9,086,981
46,710 (11)8,766,066
57,488 (5)10,788,773
86,080 (6)16,154,634
57,488 (7)10,788,773
Timothy S. Teter9,040 (8)1,696,537
51,070 (9)9,584,307
175,540 (2)32,943,592
40,840 (3)7,664, 443
48,420 (10)9,086,981
48,420 (4)9,086,981
46,710 (11)8,766,066
57,48B (5)10,788,773
86,080 (6)16,154,634
57,488 (7)10,788,773
+ +(1) Calculated by multiplying the number of RSUs or PSUs that have not vested or have not been earned, as applicable, by the closing price (\$187.67) of NVIDIA's common stock on January 23, 2026, the last trading day before the end of our Fiscal 2026, as reported by Nasdaq. +(2) The PSU was earned based on achievement of a performance goal for the one-year period that ended on January 28, 2024. The PSU vested as to 25% of the shares on March 20, 2024, and vests as to 6.25% approximately every three months thereafter over the next three years such that the PSU will be fully vested on March 17, 2027. +(3) The PSU was earned based on achievement of a performance goal for the three-year period that ended on January 25, 2026. The PSU vested as to 100% of the shares on March 18, 2026. +(4) The PSU was earned based on achievement of a performance goal for the one-year period that ended on January 26, 2025. The PSU vested as to 25% of the shares on March 19, 2025, and vests as to 6.25% approximately every three months thereafter over the next three years such that the PSU will be fully vested on March 15, 2028. + +(5) The PSU was earned based on achievement of a performance goal for the 1-year period that ended on January 25, 2026. The PSU vested as to 25% of the shares on March 18, 2026, and vests as to 6.25% approximately every three months thereafter over the next three years such that the PSU will be fully vested on March 15, 2029. +(6) Represents shares that could be earned upon achievement of a Stretch Compensation Plan goal, based on our TSR relative to the S&P 500 from January 29, 2024 through January 31, 2027. If the performance goal is achieved, 100% of the shares earned will vest on March 17, 2027. If the Threshold performance goal is achieved, 47,720 shares will be earned by Mr. Huang, 11,970 shares will be earned by Ms. Kress, 11,620 shares will be earned by Mr. Puri, 10,760 shares will be earned by Ms. Shoquist, and 10,760 shares will be earned by Mr. Teter. If the Base Compensation Plan performance goal is achieved, 190,900 shares will be earned by Mr. Huang, 47,890 shares will be earned by Ms. Kress, 46,510 shares will be earned by Mr. Puri, 43,040 shares will be earned by Ms. Shoquist, and 43,040 shares will be earned by Mr. Teter. +(7) Represents shares that could be earned upon achievement of a Stretch Compensation Plan goal, based on our TSR relative to the S&P 500 from January 27, 2025 through January 30, 2028. If the performance goal is achieved, 100% of the shares earned will vest on March 15, 2028. If the Threshold performance goal is achieved, 29,902 shares will be earned by Mr. Huang, 7,138 shares will be earned by Ms. Kress, 6,945 shares will be earned by Mr. Puri, 7,186 shares will be earned by Ms. Shoquist, and 7,186 shares will be earned by Mr. Teter. If the Base Compensation Plan performance goal is achieved, 119,607 shares will be earned by Mr. Huang, 28,550 shares will be earned by Ms. Kress, 27,779 shares will be earned by Mr. Puri, 28,744 shares will be earned by Ms. Shoquist, and 28,744 shares will be earned by Mr. Teter. +(8) The RSU vested as to 6.25% on June 15, 2022, and approximately every three months thereafter, such that the shares were fully vested on March 18, 2026, the approximate four year anniversary of the grant date. +(9) The RSU vested as to 6.25% on June 21, 2023, and approximately every three months thereafter, such that the shares will become fully vested on March 17, 2027, the approximate four year anniversary of the grant date. +(10) The RSU vested as to 6.25% on June 19, 2024, and approximately every three months thereafter, such that the shares will become fully vested on March 15, 2028, the approximate four year anniversary of the grant date. +(11) The RSU vested as to 6.25% on June 18, 2025, and approximately every three months thereafter, such that the shares will become fully vested on March 15, 2029, the approximate four year anniversary of the grant date. + +## Option Exercises and Stock Vested in Fiscal 2026 + +The following table shows information regarding stock acquired upon vesting for our NEOs during Fiscal 2026. We have not had any stock options outstanding since the end of Fiscal 2024. + +Stock Awards + +
NameNumber of Shares Acquired on Vesting (#) (1)Value Realized on Vesting ($) (2)
Jen-Hsun Huang1,113,555 (3)141,311,217
Colette M. Kress442,087 (4)63,881,560
Ajay K. Puri425,837 (5)61,550,468
Debora Shoquist368,879 (6)53,412,792
Timothy S. Teter368,879 (7)53,412,792
+ +(1) Represents the gross number of shares acquired on vesting. Shares were withheld from these amounts to pay taxes due upon vesting. +(2) Represents the gross number of shares acquired on vesting multiplied by the fair market value of our common stock as reported by Nasdaq on the date of vesting. +(3) Includes an aggregate of 552,105 shares that were withheld to pay taxes due upon vesting. +(4) Includes an aggregate of 233.268 shares that were withheld to pay taxes due upon vesting. +(5) Includes an aggregate of 211,139 shares that were withheld to pay taxes due upon vesting. +(6) Includes an aggregate of 194,424 shares that were withheld to pay taxes due upon vesting. +(7) Includes an aggregate of 198,486 shares that were withheld to pay taxes due upon vesting. + +## Employment, Severance, and Change-in-Control Arrangements + +Employment Agreements. Our executive officers are “at-will” employees and we do not have employment, severance, or change-in-control agreements with our executive officers. + +Change-in-Control Arrangements. Our 2007 Plan provides that in the event of a corporate transaction or a change-in-control, outstanding stock awards may be assumed, continued, or substituted by the surviving corporation. If the surviving corporation does not assume, continue, or substitute such stock awards, then (a) with respect to any stock awards that are held by individuals performing services for NVIDIA immediately prior to the effective time of the transaction, the vesting and exercisability provisions of such stock awards will be accelerated in full and such stock awards will be terminated if not exercised prior to the effective date of the corporate transaction or change-in-control, and (b) all other outstanding stock awards will be terminated if not exercised on or prior to the effective date of the corporate transaction or change-in-control. These change-in-control arrangements apply to stock awards held by our NEOs on the same basis as our other employees. + +## Potential Payments Upon Termination or Change-in-Control + +Upon a change-in-control or certain other corporate transactions of NVIDIA, unvested RSUs, and PSUs will fully vest in some cases as described above under Employment, Severance, and Change-in-Control Arrangements—Change-in-Control Arrangements. The table below shows our estimates of the amount of the benefit each of our NEOs would have received if the unvested RSUs and PSUs held by them as of January 25, 2026 had become fully vested as a result of a change-in-control, calculated by multiplying the number of unvested RSUs and PSUs held by the applicable NEO by the closing price (\$187.67) of NVIDIA's common stock on January 23, 2026, the last trading day before the end of our Fiscal 2026, as reported by Nasdaq. + +
NameUnvested RSUs and PSUs at January 25, 2026 (#) (1)Total Estimated Benefit ($) (1)
Jen-Hsun Huang1,411,679264,929,798
Colette M. Kress569,889106,951,069
Ajay K. Puri551,090103,423,060
Debora Shoquist500,14893,862,775
Timothy S. Teter500,14893,862,775
+ +(1) With respect to unvested PSUs, the amounts in these columns assume performance at Base Compensation Plan with respect to SY PSUs granted in Fiscal 2026 and with respect to MY PSUs granted in Fiscal 2024, Fiscal 2025, and Fiscal 2026, in accordance with SEC rules. The two tables below reflect the actual numbers of the SY PSUs granted in Fiscal 2026 and MY PSUs granted in Fiscal 2024 that became eligible to vest, based on our performance during the relevant performance period for such awards, as certified by our CC shortly after the end of Fiscal 2026. The values of the estimated and actual SY PSUs and MY PSUs in the table below were calculated by multiplying the applicable number of SY PSUs and MY PSUs held by each respective NEO and listed below, by the closing price (\$187.67) of NVIDIA's common stock on January 23, 2026, the last trading day before the end of our Fiscal 2026, as reported by Nasdaq. + +SY PSUs granted in Fiscal 2026 - Actual Achievement (versus Base Compensation Plan Performance) + +
NameEstimated SY PSUs Granted in Fiscal 2026 at Base Compensation Plan Performance (#)Value of Estimated SY PSUs Granted in Fiscal 2026 at Base Compensation Plan Performance ($)Actual SY PSUs Granted in Fiscal 2026 Eligible to Vest (#)Value of Actual SY PSUs Granted in Fiscal 2026 Eligible to Vest ($)
Jen-Hsun Huang119,60722,446,646179,41133,670,062
Colette M. Kress28,5505,357,97957,10010,715,957
Ajay K. Puri27,7795,213,28555,55810,426,570
Debora Shoquist28,7445,394,38657,48810,788,773
Timothy S. Teter28,7445,394,38657,48810,788,773
+ +MY PSUs granted in Fiscal 2024 - Actual Achievement (versus Base Compensation Plan Performance) + +
NameEstimated MY PSUs Granted in Fiscal 2024 at Base Compensation Plan Performance (#)Value of Estimated MY PSUs Granted in Fiscal 2024 at Base Compensation Plan Performance ($)Actual MY PSUs Granted in Fiscal 2024 Eligible to Vest (#)Value of Actual MY PSUs Granted in Fiscal 2024 Eligible to Vest ($)
Jen-Hsun Huang504,91094,756,460757,360142,133,751
Colette M. Kress24,7804,650,46349,5609,300,925
Ajay K. Puri23,8604,477,80647,7208,955,612
Debora Shoquist20,4203,832,22140,8407,664,443
Timothy S. Teter20,4203,832,22140,8407,664,443
+ +The actual number of MY PSUs granted in Fiscal 2025 and Fiscal 2026 that will become eligible to vest will be determinable after January 31, 2027 and January 30, 2028, respectively, the ending dates of the applicable three-year measurement period for MY PSUs. + +## Pay Ratio + +We determined the ratio of: (a) the annual total compensation of our CEO to (b) the median of the annual total compensation of all our employees, except for our CEO, both calculated in accordance with the requirements of Item 402(c)(2)(x) of Regulation S-K. + +Because there have been no changes to our employee population or employee compensation arrangements since the end of Fiscal 2024 that would significantly impact our pay ratio disclosure for Fiscal 2026, we used the median employee from our Fiscal 2024 pay ratio calculation in our Fiscal 2026 pay ratio calculation. We determined our Fiscal 2024 median employee by using a consistently applied compensation measure, which aggregated, for each employee employed by us on the last day of Fiscal 2024, or January 28, 2024: (i) target base salary as of January 28, 2024 (annualized for permanent employees who were employed by us for less than the entire fiscal year), (ii) variable cash earned during Fiscal 2024, and (iii) aggregate full grant date fair value of equity awards granted during Fiscal 2024, calculated in accordance with ASC 718 and assuming the probable outcome of the conditions at Base Compensation Plan for performance-based awards. Compensation paid in foreign currencies was converted to U.S. dollars based on exchange rates in effect on January 28, 2024. + +Our median employee's total compensation for Fiscal 2026 was \$282,050. Our CEO's Fiscal 2026 total compensation was \$36,343,830. Therefore, our Fiscal 2026 CEO to median employee pay ratio was 129:1. + +This pay ratio represents a reasonable estimate of the relationship between the compensation of our CEO and that of our median employee for Fiscal 2026, calculated in a manner consistent with Item 402(u) of Regulation S-K and applicable guidance, which provide significant flexibility in how companies identify the median employee. Each company may use a different methodology, apply different exclusions, and make different assumptions. As a result, the pay ratio reported by other companies may not be comparable to ours. + +## Pay Versus Performance + +NVIDIA's executive compensation program is guided by a pay for performance philosophy and is designed to align NEO pay with our stockholders' interests. Accordingly, a substantial portion of our NEOs' total compensation is based on the Company's performance under certain corporate financial metric goals, which for Fiscal 2026 included annual revenue, annual Non-GAAP Operating Income, and 3-year TSR relative to the S&P 500. + +Amounts set forth below as “compensation actually paid,” or CAP, have been calculated in accordance with Item 402(v) of Regulation S-K under the Securities Act, and do not represent the value of compensation actually paid to or received by our NEOs. For a discussion of our executive compensation program and philosophy, refer to our CD&A above. + +The following table summarizes information regarding compensation for our NEOs, including CAP as well as certain financial performance metrics, during Fiscal 2022-2026. Fiscal 2022-2026 were 52-week years. + +
Fiscal YearSummary Compensation Table Total for CEO ($) (1) (2)Compensation Actually Paid to CEO ($) (1) (3)Average Summary Compensation Table Total for Non-CEO NEOs ($) (4) (5)Average Compensation Actually Paid to Non-CEO NEOs ($) (4) (6)Value of Initial Fixed $100 Investment Based on (7):Net Income (in millions) ($)Non-GAAP Operating Income (in millions) ($) (9)
Total Shareholder Return ($)Peer Group Total Shareholder Return ($) (8)
202636,343,830162,180,93614,422,31757,721,5521,445.67198.10120,067137,300
202549,866,251344,188,02720,343,288125,583,9131,100.69168.4672,88086,789
202434,167,902234,132,30512,247,41185,558,057470.88134.7829,76037,134
202321,356,924(4,118,947)9,941,838(1,364,661)157.0594.134,3689,040
202223,737,661105,543,7688,910,80238,453,071175.98111.839,75212,690
+ +(1) For Fiscal 2022-2026, our CEO was Jen-Hsun Huang. +(2) The amounts in this column correspond with total compensation for our CEO as reported in our Summary Compensation Table above for the listed fiscal years. +(3) The amounts in this column represent CAP calculated in accordance with Item 402(v) of Regulation S-K during the listed fiscal years. Adjustments to the Summary Compensation Table total compensation for our CEO to arrive at CAP for Fiscal 2026 were as follows: + +Reconciliation of Summary Compensation Table Total Compensation for CEO to CAP + +
Fiscal YearSummary Compensation Table Total for CEO ($)(Deduct):Equity Award AdjustmentsTotal Equity Award Adjustments ($) (b) (c)Compensation Actually Paid to CEO ($)
Add:Add/(Deduct):Add/(Deduct):
Value of Equity Awards Reported in Summary Compensation Table ($) (a)Year End Fair Value of Awards Granted During the Year which were Unvested at Year End ($) (b)Year Over Year Change in Fair Value of Outstanding and Unvested Awards ($) (b)Change in Fair Value of Awards Granted in Prior Years which Vested During the Year ($) (b)
202636,343,830(24,800,511)60,909,207105,767,276(16,038,866)150,637,617162,180,936
+ +(a) The amount in this column corresponds with the full grant date fair value, calculated in accordance with ASC 718, of “Stock Awards” for our CEO as reported in our Summary Compensation Table above for Fiscal 2026. +(b) The equity award adjustments were calculated in accordance with the SEC methodology for determining CAP for each year shown. The amounts in these columns were determined by reference to (i) for MY PSU awards where the performance period was complete as of or prior to the applicable year end date and for SY PSU awards, the closing price of our common stock on the applicable year end date, as reduced by the present value of dividends expected to be paid on the underlying shares during the requisite service period, or the closing price of our common stock on the applicable vesting dates, and (ii) for MY PSU awards where the performance period was not yet complete as of the applicable year end date, the fair value as calculated by a Monte Carlo simulation model as of the respective year end date, for the listed fiscal years. +(c) The valuation assumptions used to calculate fair values did not materially differ from those disclosed at the time of grant. +(4) For Fiscal 2022-2026, our non-CEO NEOs were Colette M. Kress, Ajay K. Puri, Debora Shoquist, and Timothy S. Teter. +(5) The amounts in this column correspond with the average of the total compensation for our non-CEO NEOs as reported in our Summary Compensation Table above for the listed fiscal years. + +(6) The amounts in this column represent average CAP calculated in accordance with Item 402(v) of Regulation S-K during the listed fiscal years. Adjustments to the Summary Compensation Table average total compensation for our non-CEO NEOs to arrive at average CAP for Fiscal 2026 were as follows: + +Reconciliation of Summary Compensation Table Average Total Compensation for Non-CEO NEOs to CAP + +
Fiscal YearAverage Summary Compensation Table Total for Non-CEO NEOs ($)(Deduct):Equity Award AdjustmentsTotal Equity Award Adjustments ($) (b) (c)Average Compensation Actually Paid to Non-CEO NEOs ($)
Add:Add/(Deduct):Add:Add/(Deduct):
Value of Equity Awards Reported in Summary Compensation Table ($) (a)Year End Fair Value of Awards Granted During the Year which were Unvested at Year End ($) (b)Year Over Year Change in Fair Value of Outstanding and Unvested Awards ($) (b)Vesting Date Fair Value of Awards Granted and Vested During the Year ($) (b)Change in Fair Value of Awards Granted in Prior Years which Vested During the Year ($) (b)
202614,422,317(12,782,459)27,759,18025,940,0531,776,820605,64056,081,69457,721,552
+ +(a) The amount in this column corresponds with the average of the full grant date fair value, calculated in accordance with ASC 718, of “Stock Awards” for our non-CEO NEOs as reported in our Summary Compensation Table above for Fiscal 2026. +(b) The equity award adjustments were calculated in accordance with the SEC methodology for determining CAP for each year shown. The amounts in these columns were determined by reference to (i) for MY PSU awards where the performance period was complete as of or prior to the applicable year end date, for RSU awards and for SY PSU awards, the closing price of our common stock on the applicable year end date, as reduced by the present value of dividends expected to be paid on the underlying shares during the requisite service period, or the closing price of our common stock on the applicable vesting dates, and (ii) for MY PSU awards where the performance period was not yet complete as of the applicable year end date, the fair value as calculated by a Monte Carlo simulation model as of the respective year end date, for the listed fiscal years. Due to rounding, numbers presented may not sum precisely to the total provided. +(c) The valuation assumptions used to calculate fair values did not materially differ from those disclosed at the time of grant. +(7) TSR for each of the years during Fiscal 2022-2026 is cumulative, reflecting the value of a fixed \$100 investment beginning with the market close on January 29, 2021, the last trading day before our Fiscal 2022, through and including the end of the respective listed fiscal years. +(8) The Nasdaq 100 Index is the industry peer group we use for purposes of Item 201(e) of Regulation S-K. The separate peer group referenced by the CC for purposes of determining executive compensation is discussed above in CD&A. +(9) Our Company-Selected Measure, as required by Item 402(v) of Regulation S-K, is Non-GAAP Operating Income, which, in our assessment, represents the most important financial performance measure linking Fiscal 2026 NEO CAP to company performance. See Definitions above for a definition of Non-GAAP Operating Income, and see Reconciliation of Non-GAAP Financial Measures above in CD&A for a reconciliation between GAAP operating income and non-GAAP Operating Income. + +## Most Important Financial Performance Measures + +The following table is an unranked list of the most important financial performance measures linking Fiscal 2026 NEO CAP to company performance: + +Financial Measures + +
Revenue
Non-GAAP Operating Income
3-Year TSR relative to the S&P 500
+ +Refer to CD&A above for a description of how each of these performance measures impacts NEO compensation. + +## Relationships Between Compensation Actually Paid and Financial Performance + +The following graphs illustrate how CAP for our NEOs aligns with the Company's financial performance measures as detailed in the Pay Versus Performance table above for each of Fiscal 2022, 2023, 2024, 2025, and 2026 as well as between the TSRs of NVIDIA and the Nasdaq100 Index, reflecting the value of a fixed \$100 investment beginning with the market close on January 29, 2021, the last trading day before our Fiscal 2022, through and including the end of the respective listed fiscal years. + +![](images/990093ba798638a209b5eb02f0c43327022416b179b2eba7d885b5c4c7f47d9e.jpg) + +
+bar-line hybrid chart + +NEO CAP versus TSR +| Fiscal Year | CEO CAP ($ in millions) | Other NEOs Average CAP ($ in millions) | NVIDIA TSR ($) | Nasdaq100 Index TSR ($) | +| :--- | :--- | :--- | :--- | :--- | +| Fiscal 2022 | 105.5 | | 38.5 | 111.83 | +| Fiscal 2023 | (4.1) | | 157.05 | 94.13 | +| Fiscal 2024 | 234.1 | 85.6 | 470.88 | 134.78 | +| Fiscal 2025 | 344.2 | 125.6 | 1,100.69 | 168.46 | +| Fiscal 2026 | 162.2 | 57.7 | 1,445.67 | 198.10 | +
+ +![](images/1cab9054910bd37bbe3724d37a0c2b14bd8e426f1b0e2d105d1c8375fe88f8db.jpg) + +
+bar-line hybrid + +NEO CAP versus Net Income & Non-GAAP Operating Income +| Fiscal Year | CEO CAP ($ million) | Other NEOs Average CAP ($ million) | Net Income ($ million) | Non-GAAP Operating Income ($ million) | +| :--- | :--- | :--- | :--- | :--- | +| Fiscal 2022 | 105.5 | 38.5 | 9,752 | 12,690 | +| Fiscal 2023 | (4.1) | (4.1) | 4,368 | 9,040 | +| Fiscal 2024 | 234.1 | 85.6 | 29,760 | 37,134 | +| Fiscal 2025 | 344.2 | 125.6 | 72,880 | 86,789 | +| Fiscal 2026 | 162.2 | 57.7 | 120,067 | 137,300 | +
+ +All information provided above under the “Pay Versus Performance” heading will not be deemed to be incorporated by reference into any filing of the Company under the Securities Act or the Exchange Act, whether made before or after the date hereof and irrespective of any general incorporation language in any such filing, except to the extent the Company specifically incorporates such information by reference. + +What am I voting on? Ratification of the selection of PwC as our independent registered public accounting firm for Fiscal 2027. + +Vote required for approval: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter. + +Effect of abstentions: Same as a vote AGAINST. + +Effect of broker non-votes: Not applicable (because this is a routine proposal, there are no broker non-votes). + +The AC has selected PwC, which has audited our financial statements annually since 2004, to serve as our independent registered public accounting firm for Fiscal 2027. Our lead audit partner at PwC will serve no more than five consecutive years in that role. Stockholder ratification of the AC's selection of PwC is not required by our Bylaws. As a matter of good corporate governance, we are submitting the selection of PwC to our stockholders for ratification. If our stockholders do not ratify the selection, the AC will reconsider whether or not to retain PwC. Even if the selection is ratified, the AC in its sole discretion may direct the appointment of a different independent registered public accounting firm at any time during the fiscal year if it determines that such a change would be in our best interests and those of our stockholders. The AC believes it is in the best interests of NVIDIA and our stockholders to retain PwC. + +We expect that a representative of PwC will attend the 2026 Meeting. The PwC representative will have an opportunity to make a statement at the 2026 Meeting if he or she so desires and will also be available to respond to appropriate stockholder questions. + +## Recommendation of the Board + +The Board recommends that you vote FOR the ratification of the selection of PwC as our independent registered accounting firm for our fiscal year ending January 31, 2027. + +## Fees Billed by the Independent Registered Public Accounting Firm + +The following is a summary of fees billed or expected to be billed by PwC for Fiscal 2026, and fees billed by PwC for Fiscal 2025, for audit, audit related, tax, and other professional services during each fiscal year: + +
Fiscal 2026Fiscal 2025
Audit Fees (1)$ 10,166,400$ 8,067,106
Audit Related Fees (2)1,716,820724,806
Tax Fees (3)1,076,103856,439
All Other Fees (4)401,999354,000
Total Fees$ 13,361,322$ 10,002,351
+ +(1) For the audit of our consolidated financial statements, including the audit of our internal control over financial reporting, review of our quarterly financial statements and annual reports, and fees related to statutory audits of some of our international entities. +(2) For a review of select sustainability metrics, system and organization controls reporting, accounting system pre-implementation controls review, and other attestation services. +(3) For tax compliance, consulting, and tax audit defense services. +(4) For products or services other than those referenced above, including a cybersecurity maturity assessment and subscription to accounting disclosure software. + +All services provided for Fiscal 2026 and 2025 described above were pre-approved by the AC or the AC Chairperson through the authority granted to him by the AC, which is described below. Our AC determined that the rendering of services other than audit services by PwC was compatible with maintaining PwC's independence. + +## Pre-Approval Policies and Procedures + +The AC has adopted policies and procedures for the pre-approval of all audit and permissible non-audit services rendered by our independent registered public accounting firm. The policy generally permits pre-approvals of specified permissible services in the defined categories of audit services, audit-related services, and tax services up to specified amounts. Pre-approval may also be given as part of the AC's approval of the scope of the engagement of our independent registered public accounting firm or on an individual case-by-case basis before the independent registered public accounting firm is engaged to provide each service. In some cases, the full AC provides pre-approval for up to a year related to a particular defined task or scope. In other cases, the AC has delegated power to the AC Chairperson to pre-approve additional audit and non-audit services if the need for the service was unanticipated and approval is required prior to the next scheduled meeting of the AC. Such pre-approval is then communicated to the full AC at its next meeting. + +## Report of the Audit Committee of the Board of Directors + +The material in this report is not “soliciting material,” is not deemed “filed” with the SEC, and is not to be incorporated by reference in any of our filings under the Securities Act or the Exchange Act, whether made before or after the date hereof and irrespective of any general incorporation language in any such filing, except to the extent specifically incorporated by reference therein. + +The Audit Committee, or AC, oversees accounting, financial reporting, internal control over financial reporting, financial practices, and audit activities of NVIDIA and its subsidiaries. The AC reviews the results and scope of the audit and other services provided by the independent registered public accounting firm and reviews financial statements and the accounting policies followed by NVIDIA prior to the issuance of the financial statements with both management and the independent registered public accounting firm. + +Management is responsible for the financial reporting process, the preparation of consolidated financial statements in accordance with accounting principles generally accepted in the United States, or GAAP, the system of internal control over financial reporting, and the procedures designed to facilitate compliance with accounting standards and applicable laws and regulations. PricewaterhouseCoopers LLP, or PwC, our independent registered public accounting firm for Fiscal 2026, was responsible for performing an independent audit of the consolidated financial statements and issuing a report on the consolidated financial statements and of the effectiveness of our internal control over financial reporting as of January 25, 2026. PwC's judgments as to the quality, not just the acceptability, of our accounting principles and such other matters are required to be disclosed to the AC under applicable standards. The AC oversees these processes. Also, the AC has ultimate authority and responsibility to select, evaluate and, when appropriate, terminate the independent registered public accounting firm. The AC approves audit fees and non-audit services provided by and fees paid to the independent registered public accounting firm. + +NVIDIA has an internal audit function that reports to the AC. This function is responsible for objectively reviewing and evaluating the adequacy, effectiveness, and quality of our system of internal controls and the operating effectiveness of our business processes. The AC approves an annual internal audit plan and monitors the activities and performance of our internal audit function throughout the year to ensure the plan objectives are carried out and met. + +The AC members are not professional accountants or auditors, and their functions are not intended to duplicate or to certify the activities of management or the independent registered public accounting firm. The AC does not plan or conduct audits, determine that our financial statements are complete and accurate and in accordance with GAAP, or assess our internal control over financial reporting. The AC relies, without additional independent verification, on the information provided by our management and on the representations made by management that the financial statements have been prepared with integrity and objectivity, and the opinion of PwC that such financial statements have been prepared in conformity with GAAP. + +In this context, the AC reviewed and discussed the audited consolidated financial statements for Fiscal 2026 with management and our internal control over financial reporting with management and PwC. Specifically, the AC discussed with PwC the matters required to be discussed by the applicable requirements of the Public Company Accounting Oversight Board and the SEC. We have received from PwC the written disclosures and letter required by the applicable requirements of the Public Company Accounting Oversight Board regarding PwC's communications with the AC concerning independence. The AC also considered whether the provision of certain permitted non-audit services by PwC is compatible with PwC's independence and discussed PwC's independence with PwC. + +Based on the AC's review and discussions, the AC recommended to the Board of Directors that the audited consolidated financial statements be included in the Annual Report on Form 10-K of NVIDIA for the fiscal year ended January 25, 2026. + +## Audit Committee + +A. Brooke Seawell, Tench Coxe, Harvey C. Jones, Melissa B. Lora, and Aarti Shah + +## Stockholder Proposals + +We will promptly furnish the address and, to our knowledge, the number of shares held by the stockholder proponents of the proposals below upon receipt of a request to our Secretary for such information at email to shareholdermeeting@nvidia.com or by phone at: (408) 486-2000. + +## Proposal 4—Stockholder Proposal: Simple Majority Vote Standard + +What am I voting on? A stockholder proposal to eliminate all supermajority voting provisions. + +Vote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter. + +Effect of abstentions: Same as a vote AGAINST. + +Effect of broker non-votes: None. + +Below is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 4. John Chevedden has notified us of his intent to present the following proposal at the 2026 Meeting. + +## Stockholder Proposal and Supporting Statement + +## Proposal 4 – Majority Vote Standard + +![](images/caad12dd4fafe70e60e7a237a2965618279627326b441c35c7235ef6cd38ed5d.jpg) + +
+text_image + +FOR +Shareholder +Rights +
+ +Shareholders request that the Board of Directors take each step necessary so that each voting requirement in our charter and bylaws (that is explicit or implicit due to default to state law) that calls for a greater than simple majority vote be replaced by a requirement for a majority of the votes cast for and against applicable proposals, or a simple majority in compliance with applicable laws. If necessary this means the closest standard to a majority of the votes cast for and against such proposals consistent with applicable laws. This proposal includes that the Company shall state in its governing documents that it shall not have any default super-majority voting standards upon adoption of this proposal. + +Shareholders are willing to pay a premium for shares of companies that have excellent corporate governance. The supermajority voting requirements, like those of NVIDIA Corporation (NVDA), have been found to be one of 6 entrenching mechanisms that are negatively related to company performance according to "What Matters in Corporate Governance" by Lucien Bebchuk, Alma Cohen and Allen Ferrell of the Harvard Law School. Supermajority requirements can be used to block proposals supported by most shareowners but opposed by management. + +This proposal topic won from 74% to 88% support at Weyerhaeuser, Alcoa, Waste Management, Goldman Sachs, FirstEnergy and Macy's. These votes would have been higher than 74% to 88% if more shareholders had access to independent proxy voting advice. + +This proposal topic received 98% support each in 2024 at Domino's Pizza, FMC Corporation, ConocoPhillips, Masco Corporation and Power Integrations. + +This proposal includes that the Board commit to not make any effort to defeat a NVDA proposal on this topic. + +This proposal also includes that the Board commit to make a robust extra effort from a proxy solicitor perspective with the objective of obtaining the required vote for shareholder approval of this proposal topic as a binding NVDA proposal. As a matter of fact it may only take a de minimis extra effort since the 2025 NVDA proposal came so close to shareholder approval as described in the next paragraph. + +The precise extra effort is at the Board's discretion. One proven method is to adjourn the annual meeting for up to 2-weeks and seek more votes. This seems to be a prudent step since the 2025 NVDA proposal on this topic received 65% support from all outstanding NVDA shares and thus came so close to the 67% votes required. + +Please vote yes: + +## Majority Vote Standard – Proposal 4 + +## Opposing Statement of the Board + +Our Board recommends a vote AGAINST Proposal 4. + +At our 2025 Meeting, the Board proposed and recommended that stockholders approve an amendment and restatement of our Charter to eliminate the remaining supermajority voting provisions (the “2025 A&R Charter Proposal”). The Board also planned to adopt conforming amendments to our Bylaws, contingent on stockholder approval of such proposal. However, the 2025 A&R Charter Proposal did not receive the required level of stockholder support for approval. + +To effectuate Proposal 4, the Board would need to resubmit a proposal substantially similar to the 2025 A&R Charter Proposal for stockholder approval at a future annual meeting. At this time, the Board believes that addressing repetitive proposals on this topic would not be an effective use of time, resources, and attention. + +## Recommendation of the Board + +Our Board recommends that you vote AGAINST Proposal 4— Simple Majority Vote Standard. + +What am I voting on? A stockholder proposal requesting an evaluation and report on faith-based community resource groups. + +Vote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter. + +Effect of abstentions: Same as a vote AGAINST. + +Effect of broker non-votes: None. + +Below is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 5. William Cunningham has notified us of its intent to present the following proposal at the 2026 Meeting. + +## Stockholder Proposal and Supporting Statement + +## Proposal 5 – Report on Faith-Based Community Resource Groups + +Whereas: NVIDIA is one of the largest and most high-profile companies in the United States, employing $^{1}$ more than 35,000 people from a plethora of different cultural & religious backgrounds. As a major employer, NVIDIA should support the religious freedom of its employees. NVIDIA is already required to comply with many laws prohibiting discrimination against employees based on their religious status and views. + +Respecting diverse religious views allows NVIDIA to attract the most qualified talent, promote a diverse and vibrant business culture, and is a key component to make sure it fully engages each of its employees. One of the best ways to promote religious diversity is through faith-based community resource groups (elsewhere known as employee resource groups, or ERGs). CRGs allow like-minded employees to connect with one another, seek professional development, and promote understanding and dialogue with the broader workforce. Despite this, NVIDIA does not have a faith-based community resource group (CRG). + +NVIDIA does not rectify this apparently disparity in treatment even though the vast majority of Americans identify as religious, and even though the Company recognizes CRGs formed around gender, sexual identity, along a variety of other criteria as per its website. $^{2}$ + +According to the 2023 Freedom at Work $^{3}$ survey, 60% of employees were concerned that their company would punish them for expressing their religious or political views at work, and 54% said they feared the same for sharing these views even on their private social media accounts. NVIDIA needs to take proactive steps to address this shortcoming by promoting faith-based CRGs and providing them the same support and access that other CRGs enjoy. + +This isn't merely a political or social point but a matter of brand value. NVIDIA is the highest-valued company on earth, with an estimated brand value $^{3}$ exceeding \$87.9 billion, 67 percent of its more than \$130 billion revenue in 2025. $^{4}$ NVIDIA is one of the most influential players in the American technological landscape. As concerns over neutrality on controversial issues, respect for religion, and equal treatment for employees at America's largest companies continue to mount, shareholders are right to ask about how the company's workforce policies forward this goal of equal treatment. + +Recent Supreme Court decisions in Groff v. DeJoy and Muldrow v. City of St. Louis have also clarified that religious protections for employees extend to all terms, conditions, and privileges of employment, not just monetary compensation. So failure to allow faith-based community resource groups may be illegal. + +Resolved: Shareholders request the Board of Directors of NVIDIA Corporation conduct an evaluation and issue a report, or disclose analysis which has already been performed, within the next year, at reasonable cost and excluding proprietary information and disclosure of anything that would constitute an admission of pending litigation, evaluating the risks related to religious discrimination against employees. + +## Opposing Statement of the Board + +Our Board recommends a vote AGAINST Proposal 5. The requested report is unnecessary and would not provide meaningful additional benefits to NVIDIA or enhance stockholder value as we have already implemented robust systems and processes designed to minimize risks related to religious and other discrimination. + +NVIDIA is committed to ensuring that our employees are protected and supported through enterprise-wide policies, practices, and reporting mechanisms. We maintain a work environment in which all individuals are treated with dignity and respect. We do not tolerate unlawful discrimination, harassment, or retaliation in any form, including based on religion or creed and all employees are required to participate in discrimination and harassment prevention training. Our employees have multiple avenues to report concerns about harassment, discrimination or other workplace conduct, including anonymously and without fear of retaliation, and we investigate complaints promptly and take appropriate action when warranted. We also regularly gather employee feedback through surveys and other forums to identify concerns, track progress, and respond quickly to employee needs. + +The Board believes that our existing policies, practices, and feedback mechanisms already address the issues raised in Proposal 5, and that the additional evaluation and reporting requested would not be necessary or beneficial for NVIDIA or our stockholders. + +## Recommendation of the Board + +Our Board recommends that you vote AGAINST Proposal 5—Report on Faith-Based Community Resource Groups. + +What am I voting on? A stockholder proposal requesting an evaluation and report on civil rights and non-discrimination related to DEI. + +Vote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter. + +Effect of abstentions: Same as a vote AGAINST. + +Effect of broker non-votes: None. + +Below is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 6. American Conservative Values ETF has notified us of its intent to present the following proposal at the 2026 Meeting. + +## Stockholder Proposal and Supporting Statement + +## Proposal 6 – Report on Respecting Workforce Civil Liberties + +NVIDIA Corporation has over 36,000 employees. As an employer, it is legally required to comply with many laws prohibiting discrimination against employees on a variety of factors, including race, sex, religion, and sometimes political affiliation. + +But the 2025 edition of the Viewpoint Diversity Score Business Index $^{1}$ found that 90% of scored companies, as well as NVIDIA, promote divisive training and recruitment concepts like critical race theory (CRT) that replace genuine cultural and ideological diversity with a narrow focus on group identity. NVIDIA has stated that it has “increased [its] focus on diversity recruiting” and “when recruiting new talent. . . [it] strive[s] to build a robust talent pipeline that includes those underrepresented in the technology field, including women, black/African American, and Hispanic/Latino candidates.” $^{2}$ + +Focusing on these metrics in recruiting is rooted in concepts that label employees as “oppressed” or “oppressors” based on belief or immutable traits, which can foster hostility and partiality instead of equal opportunity based on merit. A recent study found that DEI actually increases bias, hostility, and attribution of discriminatory intent instead of reducing them. $^{3}$ DEI also silences dissenting views: the Freedom at Work survey found that 60% of employees feared reprisal for expressing their religious or political views at work, and over half feared the same for private social media posts. $^{4}$ + +Companies are also facing increased legal and reputational risk for DEI programs that make distinctions based on race or other protected classes in light of recent Supreme Court decisions in Students for Fair Admission v. Harvard, Groff v DeJoy, City of St. Louis v. Muldrow, and Ames v. Ohio Department of Youth Services. State attorneys general and employees have initiated lawsuits and investigations. $^{5}$ Recent scholarship has cast serious doubt on the common assertion that DEI is good for business. $^{6}$ The Wall Street Journal recently reported that “Diversity Goals Are Disappearing from Companies’ Annual Reports.” $^{7}$ And numerous companies are voluntarily removing DEI initiatives entirely in response to sustained customer dissatisfaction and public pressure. $^{8}$ + +On top of this, President Trump issued an executive order seeking to root out illegal DEI practices, including in the private sector. $^{9}$ The EEOC has initiated investigations and secured settlements from major law firms to stop their DEI practices. $^{10}$ The EEOC and DOJ jointly issued guidance detailing DEI practices that may be illegal and encouraging whistleblowers to report these practices. $^{11}$ In light of these risks, the Company must take immediate steps to assess potential shortcomings and allay these concerns. + +Resolved: Shareholders request the Board of Directors conduct an evaluation and issue a report within the next year, at reasonable cost and excluding proprietary and confidential information, evaluating how NVIDIA's hiring and training policies and practices impact employees and prospective employees based on their race, color, religion (including religious views), sex, national origin, or political views, and the associated legal, reputational, operational, and other relevant risks to the NVIDIA. + +## Opposing Statement of the Board + +Our Board recommends a vote AGAINST Proposal 6. The requested report is unnecessary and would not provide meaningful additional benefits to NVIDIA or enhance stockholder value. + +NVIDIA complies with applicable law and is committed to equal employment opportunity requirements. We do not discriminate against employees or applicants for employment on the basis of race, color, religion, sex, gender, marital status, veteran status, age, national origin, ancestry, disability, genetic or medical information, sexual orientation, pregnancy, childbirth or related medical conditions, or any other characteristic protected by applicable law. Our employment decisions – including recruitment, hiring, training, compensation, benefits, promotions, transfers, and terminations – are based on merit, valid job requirements, and other lawful criteria. + +In addition, the Board maintains strong oversight and compliance mechanisms to ensure these policies and practices are effectively implemented. The CC oversees our human capital management policies and strategies, the AC oversees compliance with legal and regulatory requirements, ethical standards, and our compliance programs, and the NCGC oversees compliance with our Code of Conduct. Together, these oversight structures provide sufficient and comprehensive governance and accountability for the policies and practices addressed by Proposal 6. + +The Board believes that our existing policies, practices, and oversight mechanisms already address the issues raised in Proposal 6, and that the additional evaluation and reporting requested would not be necessary or beneficial for NVIDIA or our stockholders. + +## Recommendation of the Board + +Our Board recommends that you vote AGAINST Proposal 6—Report on Workforce Civil Liberties. + +What am I voting on? A stockholder proposal to requesting a report disclosing GHG emissions from the use of the Company's sold products. + +Vote required: A majority of the shares present, in person or represented by proxy, and entitled to vote on this matter. + +Effect of abstentions: Same as a vote AGAINST. + +Effect of broker non-votes: None. + +Below is a stockholder proposal and supporting statement of the stockholder proponent, for which we and our Board accept no responsibility. This stockholder proposal is required to be voted upon at the 2026 Meeting only if properly presented. Our Board opposes adoption of this stockholder proposal and recommends that you vote AGAINST Proposal 7. Green Century Capital Management, Inc. has notified us of its intent to present the following proposal at the 2026 Meeting. Mercy Investment Services, Inc. is a co-filer for this proposal. + +## Stockholder Proposal and Supporting Statement + +## Proposal 7 – Greenhouse Gas Emissions Disclosure + +Whereas: Climate change poses macroeconomic risks that can depress returns for long-term diversified investors. Severe weather accounted for a record 21 billion-dollar disasters in 2025. $^{1}$ + +In its 10-K, NVIDIA acknowledges that “climate change, its impact on our supply chain and critical infrastructure worldwide...may disrupt our business and cause us to experience higher attrition, losses and costs.” $^{2}$ One-third of semiconductor supply could be at risk within a decade unless the industry adapts to manage climate risks. $^{3}$ + +Meanwhile, energy-intensive artificial intelligence (AI) is compounding climate risk. Goldman Sachs research forecasts data centers' global power demand will increase 50% by 2027, and the rate of U.S. AI growth is expected to emit an additional 24 to 44 million metric tons of carbon dioxide by 2030. $^{4}$ + +NVIDIA plays a key role in the expansion of AI through designing AI chips, systems, and software. + +While NVIDIA set 2030 targets to reduce absolute greenhouse gas emissions (GHG) emissions from electricity usage and emissions intensity from use of its sold products, in the proponent's opinion, NVIDIA's progress toward mitigating climate risk through real world achievement of GHG emissions reductions is difficult to assess. + +For instance, NVIDIA reported zero market-based electricity emissions as a part of its overall operational emissions in FY25, partly due to contracting renewable energy projects across the grid. Meanwhile, it also disclosed a separate increase in location-based electricity emissions from the energy it drew from the local power grid. $^{5}$ Additionally, the company's total absolute emissions nearly doubled between fiscal years 2024 and 2025. $^{6}$ + +Furthermore, NVIDIA's reported GHG inventory does not include use of its sold products. These emissions are likely a significant contributor to the company's GHG footprint based on peer comparison $^{7}$ and represent a key metric for tracking emissions reductions. + +Peers AMD, Intel, Onsemi, and NXP have committed to reporting or already published more detailed climate disclosures that include: + +• Absolute emissions from use of sold products and initiatives to reduce them; +• Third-party validation and increased criteria for renewable energy projects; +• Emissions pathways to 2030, including plans to advance climate initiatives across their value chains. + +By enhancing its emissions disclosures, NVIDIA will increase the legitimacy of its climate targets and better demonstrate climate risk mitigation progress to investors as its business grows. + +RESOLVED: Shareholders request that NVIDIA issue a report, at reasonable cost and omitting proprietary information, disclosing the GHG emissions from use of its sold products. + +SUPPORTING STATEMENT: Proponents recommend, at management's discretion: + +• Disclosing emissions that account for major sources of its total GHG footprint; +• Considering GHG emissions disclosure guidance; +- Reporting based on reasonable emissions estimates, updated annually, and providing timelines for issuing or completing disclosures; +• Outlining whether and how the company plans to achieve absolute emissions reductions. + +## Opposing Statement of the Board + +Our Board recommends a vote AGAINST Proposal 7 because it is unnecessary, overly prescriptive and not in the best interests of NVIDIA or our stockholders. + +## The proposal is unnecessary given NVIDIA's history of robust climate reporting and targets, and our future disclosure plans + +NVIDIA's current practices evidence a clear commitment to sustainability, climate risk management, and effective disclosure. For example, NVIDIA consistently publishes highly detailed climate reporting, including in our 2025 Sustainability Report, which is assured by our financial auditor. In addition to Scope 1 and 2 emissions, our public reporting includes Scope 3 emissions across eight categories, as well as disclosure on product energy efficiency, climate strategy, and progress towards achieving our targets, including our success in meeting our renewable energy and supplier engagement targets. In 2025, we also announced significant new targets validated by the Science Based Targets initiative (SBTi), which include absolute Scope 1 and 2 emissions reductions and reductions in Scope 3 emissions intensity from use of sold GPU products. In 2025, in an effort to offer additional transparency into emissions, we also began publishing the market's first summaries of ISO-conformant, third-party reviewed product carbon footprints for accelerated computing products. We continue to advance our climate reporting and, as discussed with the proponent, are planning further enhancements of our Scope 3 reporting, including emissions from the use of sold products. + +## The proposal adopts an overly prescriptive approach to highly complex disclosures + +The Board believes NVIDIA's management, supported by our CS team and overseen by the NCGC, is best positioned to make decisions regarding the specific content and methodology of our climate disclosures. Scope 3 disclosures are inherently complex and subject to varied and developing calculation methodologies, levels of assurance, and credibility. The difficulty of preparing reliable Scope 3 disclosures is particularly acute with respect to the fast-evolving AI computing industry. Despite these considerations, the proposal attempts to replace management's judgment regarding highly technical reporting with a broad prescriptive mandate that includes no timeframe or acknowledgement of the significant challenges and judgments involved in preparing upstream emissions disclosures. + +Given NVIDIA's track record and reporting plans, adoption of a prescriptive approach to Scope 3 reporting methodology is inappropriate and unhelpful to our stockholders. + +## Recommendation of the Board + +Our Board recommends that you vote AGAINST Proposal 7—Greenhouse Gas Emissions Disclosure. + +## Equity Compensation Plan Information + +The number of shares that were issuable upon the vesting of outstanding RSUs and PSUs, and the number of shares that remained available for future issuance, under each of our equity compensation plans as of January 25, 2026, are summarized below. We have not had any stock options outstanding since the end of Fiscal 2024. + +
Plan CategoryNumber of securities to be issued upon exercise of outstanding options, warrants, and rights (a)Weighted-average exercise price of outstanding options, warrants, and rights (b)Number of securities remaining available for future issuance under equity compensation plans (excluding securities reflected in column (a)) (c)
Equity compensation plans approved by security holders (1)192,391,7653,551,257,925 (2)
Equity compensation plans not approved by security holders
Total192,391,7653,551,257,925
+ +(1) This row includes our 2007 Plan and our ESPP. Under our ESPP, participants are permitted to purchase our common stock at a discount on certain dates through payroll deductions within a pre-determined purchase period. Accordingly, the number of shares to be issued upon exercise of outstanding rights under our ESPP as of January 25, 2026 is not determinable. +(2) As of January 25, 2026, (a) the number of shares that remained available for future issuance under the 2007 Plan was 1,324,516,906, and (b) the number of shares that remained available for future issuance under the ESPP was 2,226,741,019, of which up to 2,825,984 shares may be purchased under the ESPP in the current purchase period which runs until August 31, 2026, based on estimated participation and contribution rates, purchase prices based on the applicable offering date prices, and the \$25,000 limit under Section 423(b)(8) of the Internal Revenue Code. + +During Fiscal 2026, we granted an aggregate of 69,186,497 shares under our 2007 Plan in the form of RSUs and PSUs, 1,041,733 of which were granted to our NEOs, 21,588 of which were granted to our non-employee directors, and 68,123,176 of which were granted to our other employees. For this purpose, PSUs are counted in the year of grant at the maximum number of shares that may become eligible to vest. Also during Fiscal 2026, an aggregate of 13,226,923 shares were purchased under our ESPP, 4,655 of which were purchased by our NEOs, and 13,222,268 of which were purchased by our other employees. Our non-employee directors are not eligible to participate in our ESPP. + +## Additional Information + +## Delinquent Section 16(a) Reports + +Section 16(a) of the Exchange Act requires our executive officers, directors and persons who own more than 10% of a registered class of our equity securities to file initial reports of ownership and reports of changes in ownership of our common stock and other equity securities with the SEC. Executive officers, directors and greater than 10% stockholders are required by SEC regulations to furnish us with copies of all Section 16(a) forms they file. To our knowledge, based solely on a review of the copies of such reports furnished to us and written representations that no other reports were required, all Section 16(a) filing requirements applicable to individuals who were, during Fiscal 2026, our executive officers, directors and greater than 10% beneficial owners were complied with. + +## Other Matters + +The Board knows of no other matters that will be presented for consideration at the 2026 Meeting. If any other matters are properly brought before the 2026 Meeting, it is the intention of the persons named in the accompanying proxy to vote on such matters in accordance with their best judgment. + +## By Order of the Board of Directors + +![](images/11f775eea171c32e335802e170831464b6cf8ae2651336621162b6375bcc449f.jpg) + +## Timothy S. Teter + +Secretary + +May 12, 2026 + +A COPY OF OUR ANNUAL REPORT ON FORM 10-K FOR THE FISCAL YEAR ENDED JANUARY 25, 2026 AS FILED WITH THE SEC IS BEING FURNISHED TO STOCKHOLDERS CONCURRENTLY HEREWITH. UPON WRITTEN REQUEST, WE WILL PROVIDE, WITHOUT CHARGE, AN ADDITIONAL COPY OF THE ANNUAL REPORT. STOCKHOLDERS MAY SUBMIT THEIR REQUESTS TO: INVESTOR RELATIONS, NVIDIA CORPORATION, 2788 SAN TOMAS EXPRESSWAY, SANTA CLARA, CALIFORNIA 95051 OR TO SHAREHOLDERMEETING@NVIDIA.COM. WE WILL ALSO FURNISH A COPY OF ANY EXHIBIT TO THE ANNUAL REPORT ON FORM 10-K IF SPECIFICALLY REQUESTED IN WRITING. + +NVIDIA and the NVIDIA logo are either registered trademarks or trademarks of NVIDIA Corporation in the United States and other countries. Other company names used in this publication are for identification purposes only and may be trademarks of their respective companies. + +## FORM 10-K + +ANNUAL REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 + +For the fiscal year ended January 25, 2026 + +OR + +☐ TRANSITION REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 + +Commission file number: 0-23985 + +![](images/c7d5c43bbe8251611e07dda75a83fbba506eab1d2bf1a87739b06de4a60096c4.jpg) + +## NVIDIA. + +## NVIDIA CORPORATION + +(Exact name of registrant as specified in its charter) + +Delaware + +(State or other jurisdiction of + +incorporation or organization) + +94-3177549 + +(I.R.S. Employer + +Identification No.) + +2788 San Tomas Expressway, Santa Clara, California + +(Address of principal executive offices) + +95051 + +(Zip Code) + +Registrant's telephone number, including area code: (408) 486-2000 + +Securities registered pursuant to Section 12(b) of the Act: + +Title of each class + +Common Stock, \$0.001 par value per share + +Trading Symbol(s) + +NVDA + +Name of each exchange on which registered + +The Nasdaq Global Select Market + +Securities registered pursuant to Section 12(g) of the Act: + +None + +Indicate by check mark if the registrant is a well-known seasoned issuer, as defined in Rule 405 of the Securities Act. Yes ☒ No □ + +Indicate by check mark if the registrant is not required to file reports pursuant to Section 13 or Section 15(d) of the Act. Yes □ No ☒ + +Indicate by check mark whether the registrant (1) has filed all reports required to be filed by Section 13 or 15(d) of the Securities Exchange Act of 1934 during the preceding 12 months (or for such shorter period that the registrant was required to file such reports), and (2) has been subject to such filing requirements for the past 90 days. Yes ☑ No □ + +Indicate by check mark whether the registrant has submitted electronically every Interactive Data File required to be submitted pursuant to Rule 405 of Regulation S-T ( $§232.405$ of this chapter) during the preceding 12 months (or for such shorter period that the registrant was required to submit such files). Yes ☑ No □ + +Indicate by check mark whether the registrant is a large accelerated filer, an accelerated filer, a non-accelerated filer, a smaller reporting company, or an emerging growth company. See the definitions of "large accelerated filer," "accelerated filer," "smaller reporting company," and "emerging growth company" in Rule 12b-2 of the Exchange Act. + +Large accelerated filer ☒ Accelerated filer ☐ Non-accelerated filer ☐ Smaller reporting company ☐ Emerging growth company ☐ + +If an emerging growth company, indicate by check mark if the registrant has elected not to use the extended transition period for complying with any new or revised financial accounting standards provided pursuant to Section 13(a) of the Exchange Act. □ + +Indicate by check mark whether the registrant has filed a report on and attestation to its management's assessment of the effectiveness of its internal control over financial reporting under Section 404(b) of the Sarbanes-Oxley Act (15 U.S.C. 7262(b)) by the registered public accounting firm that prepared or issued its audit report. ☒ + +If securities are registered pursuant to Section 12(b) of the Act, indicate by check mark whether the financial statements of the registrant included in the filing reflect the correction of an error to previously issued financial statements. □ + +Indicate by check mark whether any of those error corrections are restatements that required a recovery analysis of incentive-based compensation received by any of the registrant's executive officers during the relevant recovery period pursuant to §240.10D-1(b). □ + +Indicate by check mark whether the registrant is a shell company (as defined in Rule 12b-2 of the Act). Yes □ No ✗ + +The aggregate market value of the voting stock held by non-affiliates of the registrant as of July 25, 2025 was approximately \$4.0 trillion (based on the closing sales price of the registrant's common stock as reported by the Nasdaq Global Select Market on July 25, 2025). This calculation excludes 1.0 billion shares held by directors and executive officers of the registrant. This calculation does not exclude shares held by such organizations whose ownership exceeds 5% of the registrant's outstanding common stock that have represented to the registrant that they are registered investment advisers or investment companies registered under section 8 of the Investment Company Act of 1940. + +The number of shares of common stock outstanding as of February 20, 2026 was 24.3 billion. + +## DOCUMENTS INCORPORATED BY REFERENCE + +Portions of the registrant's Proxy Statement for its 2026 Annual Meeting of Shareholders to be filed with the Securities and Exchange Commission pursuant to Regulation 14A not later than 120 days after the end of the fiscal year covered by this Annual Report on Form 10-K are incorporated by reference into Part III, Items 10-14 of this Annual Report on Form 10-K. + +## Part I + +Item 1. Business 4 + +Item 1A. Risk Factors 12 + +Item 1B. Unresolved Staff Comments 32 + +Item 1C Cybersecurity 32 + +Item 2. Properties 33 + +Item 3. Legal Proceedings 33 + +Item 4. Mine Safety Disclosures 33 + +## Part II + +Item 5. Market for Registrant’s Common Equity, Related Stockholder Matters and Issuer Purchases of Equity Securities 33 + +Item 6. [Reserved] 35 + +Item 7. Management's Discussion and Analysis of Financial Condition and Results of Operations 36 + +Item 7A. Quantitative and Qualitative Disclosures About Market Risk 44 + +Item 8. Financial Statements and Supplementary Data 45 + +Item 9. Changes in and Disagreements with Accountants on Accounting and Financial Disclosure 45 + +Item 9A. Controls and Procedures 45 + +Item 9B. Other Information 46 + +Item 9C. Disclosure Regarding Foreign Jurisdictions that Prevent Inspections 46 + +## Part III + +Item 10. Directors, Executive Officers and Corporate Governance 46 + +Item 11. Executive Compensation 47 + +Item 12. Security Ownership of Certain Beneficial Owners and Management and Related Stockholder Matters 47 + +Item 13. Certain Relationships and Related Transactions, and Director Independence 47 + +Item 14. Principal Accountant Fees and Services 47 + +## Part IV + +Item 15. Exhibits and Financial Statement Schedules 48 + +Item 16. Form 10-K Summary 83 + +Signatures 84 + +## Where You Can Find More Information + +Investors and others should note that we announce material financial information to our investors using our investor relations website, press releases, SEC filings and public conference calls and webcasts. We also use the following social media channels as a means of disclosing information about the company, our products, our planned financial and other announcements and attendance at upcoming investor and industry conferences, and other matters and for complying with our disclosure obligations under Regulation FD: + +NVIDIA Corporate Blog (blogs.nvidia.com/) + +NVIDIA Technical Blog (developer.nvidia.com/blog/) + +NVIDIA LinkedIn (linkedin.com/company/nvidia) + +NVIDIA Facebook (facebook.com/nvidia) + +NVIDIA Instagram (instagram.com/nvidia) + +NVIDIA X (x.com/nvidia) + +NVIDIA Threads (threads.com/@nvidia) + +NVIDIA Investor Relations (investor.nvidia.com) + +NVIDIA YouTube (YouTube.com/nvidia). + +The information we post through these social media channels may be deemed material. Accordingly, investors should monitor these channels, in addition to following our press releases, SEC filings and public conference calls and webcasts. This list may be updated from time to time. The information we post through these channels is not a part of this Annual Report on Form 10-K. + +## Forward-Looking Statements + +This Annual Report on Form 10-K contains forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the “safe harbor” created by those sections based on management’s beliefs and assumptions and on information currently available to our management. In some cases, you can identify forward-looking statements by terms such as “may,” “will,” “should,” “could,” “goal,” “would,” “expect,” “plan,” “anticipate,” “believe,” “estimate,” “project,” “predict,” “potential,” and similar expressions intended to identify forward-looking statements. These statements involve known and unknown risks, uncertainties and other factors, which may cause our actual results, performance, time frames or achievements to be materially different from any future results, performance, time frames or achievements expressed or implied by the forward-looking statements. We discuss many of these risks, uncertainties, and other factors in this Annual Report on Form 10-K in greater detail under the heading “Risk Factors.” Given these risks, uncertainties, and other factors, you should not place undue reliance on these forward-looking statements. Also, these forward-looking statements represent our estimates and assumptions only as of the date of this filing. You should read this Annual Report on Form 10-K completely and understand that our actual future results may be materially different from what we expect. We hereby qualify our forward-looking statements by these cautionary statements. Except as required by law, we assume no obligation to update these forward-looking statements publicly, or to update the reasons actual results could differ materially from those anticipated in these forward-looking statements, even if new information becomes available in the future. + +All references to “NVIDIA,” “we,” “us,” “our,” or the “Company” mean NVIDIA Corporation and its subsidiaries. + +In addition, statements that “we believe” and similar statements reflect our beliefs and opinions on the relevant subject. These statements are based upon information available to us as of the filing date of this Annual Report on Form 10-K, and while we believe such information forms a reasonable basis for such statements, such information may be limited or incomplete, and our statements should not be read to indicate that we have conducted an exhaustive inquiry into, or review of, all potentially available relevant information. These statements are inherently uncertain, and investors are cautioned not to unduly rely upon these statements. + +© 2026 NVIDIA Corporation. All rights reserved. + +## Part I + +## Item 1. Business + +## Our Company + +NVIDIA pioneered accelerated computing to help solve the most challenging computational problems. NVIDIA is now a data center scale AI infrastructure company reshaping all industries. + +Our technology stack includes the foundational NVIDIA CUDA development platform that runs on all NVIDIA GPUs, as well as hundreds of domain-specific software libraries, frameworks, algorithms, software development kits, or SDKs, and application programming interfaces, or APIs. This deep and broad software stack accelerates the performance and facilitates the deployment of NVIDIA accelerated computing for computationally intensive workloads such as artificial intelligence, or AI, model training and inference, data analytics, scientific computing, robotics, and 3D graphics, with vertical-specific optimizations to address industries ranging from healthcare and telecom to automotive and manufacturing. + +Introduced with the Blackwell architecture, our data-center-scale offerings feature extreme co-design where the infrastructure's chips, networking, systems, software, and algorithms are holistically architected and optimized to maximize performance and scale. Hundreds of thousands of GPUs can be interconnected to function as a single giant computer. This type of data center architecture and scale is needed for the development and deployment of modern AI and accelerated computing applications. + +The GPU was initially used to simulate human imagination, enabling the virtual worlds of video games and films. Today, it also simulates human intelligence, enabling a deeper understanding of language, science, and the physical world. Its parallel processing capabilities, supported by tens of thousands of computing cores, are essential for deep learning algorithms. This form of AI, in which software writes itself by learning from large amounts of data, can serve as the brain of computers, robots, and self-driving cars that can perceive, understand and reason about the world. GPU-powered AI solutions are being developed by thousands of enterprises to deliver services and products that would have been immensely difficult or even impossible with traditional coding. Examples include generative AI, which can create new content such as text, code, images, audio, video, molecule structures, and recommendation systems; and agentic AI where systems of AI models work in concert to automatically complete a task. + +NVIDIA has a platform strategy, bringing together hardware, systems, software, algorithms, libraries, AI models and training data sets, and services to create unique value for the markets we serve. While the computing requirements of these end markets are diverse, we address them with a unified underlying programmable architecture allowing us to support several multi-billion-dollar end markets with the same underlying technology by using a variety of software stacks developed either internally or by third-party developers and partners. The large and growing number of developers and installed base across our platforms strengthens our ecosystem and increases the value of our platform for our customers. + +Innovation is at our core. We have invested over \$76.7 billion in research and development since our inception, yielding inventions that are essential to modern computing. Our invention of the GPU in 1999 sparked the growth of the PC gaming market and redefined computer graphics. With our introduction of CUDA in 2006, we opened the parallel processing capabilities of our GPU to a broad range of compute-intensive applications, paving the way for the emergence of modern AI. In 2012, the AlexNet neural network, trained on NVIDIA GPUs, won the ImageNet computer image recognition competition, marking the “Big Bang” moment of AI. We introduced our first Tensor Core GPU in 2017, built from the ground-up for the new era of AI, and our first autonomous driving system-on-chips, or SoC, in 2018. Our acquisition of Mellanox in 2020 expanded our offerings to include networking, enabled our platforms to be data center scale, and led to the introduction of a new processor class – the data processing unit, or DPU. Over the past 5 years, we have built full software stacks that run on top of our GPUs and CUDA to bring AI to the world’s largest industries, including NVIDIA DRIVE stack for autonomous driving, Clara for healthcare, Omniverse for physical AI applications, and NVIDIA AI Enterprise software – essentially an operating system for enterprise AI applications. In 2023, we introduced our first data center CPU, Grace, built for giant-scale AI and high-performance computing, or HPC. In 2024, we launched the NVIDIA Blackwell architecture – connecting 36 Grace CPUs and 72 Blackwell GPUs in a data center scale, liquid-cooled design – for real-time trillion-parameter inference and training. In fiscal year 2026, we launched and scaled the NVIDIA Blackwell Ultra platform, optimized for agentic, reasoning, and physical AI. Building on the architectural breakthroughs of Blackwell and leveraging Dynamo inference software, it delivers a significant increase in token throughput and reduction in cost per token compared to the Hopper generation. More recently, in support of market development, we have accelerated the release cadence of our open AI model platforms including NVIDIA Nemotron for agentic AI and Cosmos for physical AI. With a strong engineering culture, we drive fast, yet harmonized, product and technology innovations in all dimensions of computing including silicon, systems, networking, software and algorithms. More than half of our engineers work on software. + +All major cloud service providers, or CSPs, AI model makers, and enterprises use our data center-scale infrastructure and computing platforms to accelerate the services and offerings they deliver to billions of end users and customers, including AI solutions and assistants, AI foundation models, advertising, search, recommendation systems, social networking, data processing, online shopping, live video, and translation. AI model makers use our infrastructure and software hosted at CSPs to develop, build and run AI models, product offerings, and services. + +Enterprises and startups across a broad range of industries use our accelerated computing platforms to build new generative and agentic AI-enabled products and services, and/or to dramatically accelerate and reduce the costs of their workloads and workflows. The enterprise software industry uses them for new AI assistants, chatbots, and agents; the transportation industry for autonomous driving; the healthcare industry for accelerated and computer-aided drug discovery; and the financial services industry for customer support and fraud detection. + +Researchers and developers use our computing solutions to accelerate a wide range of important applications, from simulating molecular dynamics to climate forecasting. With support for 6,000 applications, NVIDIA computing enables some of the most promising areas of discovery, from climate prediction to materials science and from wind tunnel simulation to genomics. Including GPUs and networking, NVIDIA powers over 78% of the supercomputers on the global TOP500 list, including 9 of the top 10 systems on the Green500 list. + +Gamers choose NVIDIA GPUs to enjoy immersive, increasingly cinematic virtual worlds. In addition to serving the growing number of gamers, the market for PC GPUs is expanding because of the growing population of live streamers, broadcasters, artists, and creators. With the advent of generative and agentic AI, we expect a broader set of PC users to choose NVIDIA GPUs for running these applications locally on their PC, which is critical for privacy, latency, and cost-sensitive AI applications. + +Professional artists, architects and designers use NVIDIA partner products accelerated with our GPUs and software platform for a range of creative, engineering, and design use cases, such as creating visual effects in movies or designing buildings and products. In addition, generative and agentic AI is expanding the market for our workstation-class GPUs, as more enterprise customers develop and deploy AI applications with their data on-premises. + +Headquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998. + +## Our Businesses + +We report our business results in two segments. + +The Compute & Networking segment includes our Data Center accelerated computing and networking platforms and AI solutions and software, and Automotive platforms and autonomous and electric vehicle solutions including software. + +The Graphics segment includes GeForce GPUs for gaming and PCs, and Quadro/NVIDIA RTX GPUs for enterprise workstation graphics. + +## Our Markets + +We specialize in markets where our computing and AI infrastructure platforms can provide tremendous acceleration for applications. These platforms incorporate processors, interconnects, software, algorithms, systems, and services to deliver unique value. Our platforms address four large markets where our expertise is critical: Data Center, Gaming, Professional Visualization, and Automotive. + +## Data Center + +The NVIDIA Data Center platform is focused on accelerating compute-intensive workloads, such as AI, data processing, graphics, robotics, and scientific computing, delivering superior total cost of ownership relative to conventional CPU-only approaches. It is deployed in cloud, hyperscale, on-premises and edge data centers. The platform consists of data center compute and networking infrastructure offerings typically delivered to customers as rack-scale systems, subsystems, or modules, along with software and services. + +Our Data Center infrastructure systems include supercomputing platforms and servers, bringing together our higher performance, energy efficient GPUs, CPUs, interconnects, and fully optimized AI and HPC software stacks. In addition, they include a growing body of acceleration libraries, AI models and training data sets, APIs, SDKs, and domain-specific application frameworks. + +Our networking offerings include NVLink interconnects and switches, end-to-end platforms for InfiniBand and Ethernet, consisting of network adapters, cables, DPUs, switch chips and systems, as well as software. This has enabled us to architect data center-scale computing platforms that can interconnect up to hundreds of thousands of compute nodes with high-performance networking. Fueled by an expansion in AI and HPC workloads, the data center has become the new unit of computing, with networking as an integral part. In fiscal year 2026, we introduced NVIDIA NVLink Fusion to enable hyperscalers and custom ASIC designers to integrate custom CPUs and XPUs with our platform. + +Our customers include all major public and private cloud providers, AI model makers, enterprises and startups, and public sector entities. We work with industry leaders to help build or transform their applications and data center infrastructure. Some of our direct customers include original equipment manufacturers, or OEMs, original device manufacturers, or + +ODMs, system integrators and distributors which we partner with to help bring our products to market. We also have partnerships in automotive, healthcare, financial services, manufacturing, retail, and technology among others, to accelerate the adoption of AI. + +At the foundation of the NVIDIA accelerated computing platform are our GPUs, which excel at parallel workloads such as the training and inferencing of neural networks. These Data Center systems are extreme co-designed with the GPU, CPU, NVLink switch, DPU, NIC, and scale-out networking along with software stacks and algorithms to deliver data center-scale computing solutions. + +While our approach starts with powerful chips, what makes it a full-stack computing platform is our large body of software, including the CUDA development platform, the CUDA-X collection of acceleration libraries, AI models and training data sets, APIs, SDKs, and domain-specific application frameworks. + +In addition to software delivered to customers as an integral part of our data center computing and networking platform, we offer paid licenses to NVIDIA AI Enterprise, a comprehensive suite of enterprise-grade AI software and NVIDIA vGPU software for graphics-rich virtual desktops and workstations. + +In fiscal year 2025, we launched the NVIDIA Blackwell architecture, a full set of data center scale infrastructure that includes GPUs, CPUs, DPUs, interconnects, switch chips and systems, and networking adapters. Blackwell excels at processing cutting edge generative AI and accelerated computing workloads with market leading performance and efficiency. Offered in a number of configurations, for customers across industries and a diverse set of AI and accelerated computing use cases. In fiscal year 2026, we unveiled the NVIDIA Rubin platform, which is expected to commence production shipments in the second half of fiscal year 2027. Built for agentic AI and reasoning, it excels at processing multi-step problem-solving and massive long-context workflows, delivering up to a 10x reduction in cost per token compared to Blackwell. + +For physical AI, we provide an end-to-end platform spanning data center infrastructure, open models, systems, embedded compute modules, and software stacks to train, simulate, and deploy advanced automation and robotics solutions. + +## Gaming + +Gaming is the largest entertainment industry, with PC gaming as the predominant platform. Many factors propel its growth, including new high production value games, the continued rise of eSports, social connectivity and the increasing popularity of game streamers, modders, or gamers who remaster games, and creators. + +Our products for the gaming market include GeForce RTX GPUs for gaming desktop and laptop PCs, GeForce NOW cloud gaming service, as well as SoCs and development services for game consoles. + +Our gaming platforms leverage our GPUs and sophisticated software to enhance the gaming experience with smoother, higher quality graphics. NVIDIA RTX features ray tracing technology for real-time, cinematic-quality rendering, and deep learning super sampling, or NVIDIA DLSS, our AI technology that boosts frame rates while generating high-quality images for games. RTX GPUs also feature NVIDIA tensor core technology making them well suited to accelerate a new generation of on-device AI applications. + +In fiscal year 2025, we announced the NVIDIA Blackwell GeForce RTX 50 Series family of desktop and laptop GPUs. The Blackwell architecture introduced neural graphics which combines AI models with traditional rendering to boost game performance, image quality, and interactivity, as well as the next generation of our DLSS technology powered by a new transformer model architecture. In fiscal year 2026, we launched and scaled Blackwell architecture for gaming and GeForce NOW. + +## Professional Visualization + +We serve the Professional Visualization market by working closely with independent software vendors, or ISVs, to optimize their offerings for NVIDIA GPUs. Our GPU computing platform enhances productivity and introduces new capabilities for critical workflows in many fields, such as design, engineering, and digital content creation across a wide range of industry verticals. Additionally, the increasing number of generative and agentic AI applications is giving rise to the need for the enhanced AI and data processing capabilities of our RTX PRO GPUs. + +Many leading 3D design and content creation applications developed by our ecosystem partners support RTX, allowing professionals to accelerate and transform their workflows with NVIDIA RTX PRO GPUs and software. As these applications increasingly integrate AI, these GPUs are used and leverage the same Tensor Core technology found in our Data Center solutions. + +## Automotive + +Automotive is comprised of platform solutions for automated driving from the cloud to the car. Leveraging our technology leadership in AI and building on long-standing relationships across several hundred automotive ecosystem partners, we are delivering a full stack end-to-end solution for the AV market under the DRIVE Hyperion platform. This platform consists of development infrastructure, high-performance, energy efficient DRIVE AGX computing hardware running an in-vehicle operating system (DRIVE OS), a reference sensor set that supports full self-driving capability as well as an open, modular DRIVE software platform for autonomous driving, mapping, and parking services, and intelligent in-vehicle experiences. + +## Business Strategies + +NVIDIA's key strategies that shape our overall business approach include: + +Advancing the NVIDIA accelerated computing platform. Our accelerated computing platform can solve complex problems in significantly less time and with lower power consumption than alternative computational approaches. It can help solve problems that were previously deemed unsolvable. We work to deliver continued performance leaps that outpace Moore's Law by leveraging innovation across the architecture, chip design, system, interconnect, algorithm, and software layers. This full-stack innovation approach allows us to deliver order-of-magnitude performance advantages relative to legacy approaches in our target markets, which include Data Center, Gaming, Professional Visualization, and Automotive. While the computing requirements of these end markets are diverse, we address them with a unified underlying architecture leveraging our GPUs, CPUs, CUDA and networking technologies as the fundamental building blocks. The programmable nature of our architecture allows us to make leveraged investments in research and development: we can support several multi-billion-dollar end markets with shared underlying technology by using a variety of software stacks developed either internally or by third-party developers and partners. We utilize this platform approach in each of our target markets. + +Extending our technology and platform leadership in AI. We provide a complete, end-to-end accelerated computing platform for AI, addressing both training and inferencing. This includes full-stack data center-scale compute and networking solutions across processing units, interconnects, systems, and software. Our compute solutions include all three major processing units in AI servers – GPUs, CPUs, and DPUs. GPUs are uniquely suited to AI, and we will continue to add AI-specific features to our GPU architecture to further extend our leadership position. + +In addition, we offer NVIDIA AI Enterprise—a comprehensive software suite designed to simplify the development and deployment of production-grade, end-to-end generative AI applications. NVIDIA AI Enterprise includes: NVIDIA NIM, which increases token throughput using industry-leading open and proprietary models; NVIDIA NeMo, a complete solution for curating, fine-tuning, reinforcement learning, evaluating, and safeguarding domain-adapted models; and AI Blueprints, pre-built, runnable templates that help enterprises build, optimize, and deploy AI agents while preserving privacy. These tools enable organizations to securely develop and run AI applications on NVIDIA-accelerated infrastructure anywhere. + +Our AI technology leadership is reinforced by our large and expanding ecosystem. Our computing platforms are available from virtually every major server maker and CSP, as well as on our own AI supercomputers. There are over 7.5 million developers worldwide using CUDA and our other software tools to help deploy our technology in our target markets. We are the leader in accelerating and releasing open AI models which enterprises, sovereigns, and startups can leverage to develop and run applications on our platform. We evangelize AI through partnerships with hundreds of universities and tens of thousands of startups through our Inception program. Additionally, our Deep Learning Institute provides instruction on the latest techniques on how to design, train, and deploy neural networks in applications using our accelerated computing platform. + +Extending our technology and platform leadership in computer graphics. We believe that computer graphics infused with AI is fundamental to the continued expansion and evolution of computing. We apply our research and development resources to enhance the user experience for consumer entertainment and professional visualization applications and create new virtual world and simulation capabilities. Our technologies are instrumental in driving the gaming, design, and creative industries forward, as developers leverage our libraries and algorithms to deliver an optimized experience on our GeForce and NVIDIA RTX platforms. Our computer graphics platforms leverage AI end-to-end, from the developer tools and cloud services to the Tensor Cores included in all RTX-class GPUs. Blackwell GPUs' advanced AI and neural rendering capabilities combined with NVIDIA's world-class AI software stacks significantly accelerate AI workloads running locally on PCs. Omniverse is real-time 3D design collaboration and virtual world simulation software that empowers artists, designers, and creators to connect and collaborate in leading design applications. + +Advancing the leading autonomous vehicle platform. We believe the advent of autonomous vehicles, or AV, and electric vehicles, or EV, is revolutionizing the transportation industry. The algorithms required for autonomous driving - such as reasoning, perception, localization, and planning - are too complex for legacy hand-coded approaches and will use multiple neural networks instead. Therefore, we provide an AI-based hardware and software solution, designed and implemented from the ground up based on automotive safety standards, for the AV and EV market under the DRIVE brand, which we are bringing to market through our partnerships across the transportation industry including with automotive OEMs, mobility service providers, robotaxis, tier-1 suppliers, and start-ups. Our AV solution also includes the GPU-based hardware required to train the neural networks before their in-vehicle deployment, as well as to re-simulate their operation prior to any over-the-air software updates. We believe our comprehensive, top-to-bottom and end-to-end approach will enable the transportation industry to solve the complex problems arising from the shift to autonomous driving. + +Leveraging our intellectual property, or IP. We believe our IP is a valuable asset that can be accessed by our customers and partners through license and development agreements when they desire to build such capabilities directly into their own products or have us do so through a custom development. Such license and development arrangements can further enhance the reach of our technology. + +## Sales and Marketing + +Our worldwide sales and marketing strategy is key to achieving our objective of providing markets with our high-performance and efficient computing platforms and software. Our sales and marketing teams, located across our global markets, work closely with customers and various industry ecosystems through our partner network. Our partner network incorporates global, regional and specialized CSPs, OEMs, ODMs, ISVs, global system integrators, add-in board manufacturers, or AIBs, distributors, automotive manufacturers and tier-1 automotive suppliers, and other ecosystem participants. + +Members of our sales team have technical expertise and product and industry knowledge. We also employ a team of application engineers and solution architects to provide pre-sales assistance to our partner network in designing, testing, and qualifying system designs that incorporate our platforms. For example, our solution architects work with CSPs to provide pre-sales assistance to enable our customers to optimize their hardware and software infrastructure for generative and agentic AI and LLM training and deployment. They also work with foundation model and enterprise software developers to enable our customers to optimize the training and fine-tuning of their models and services, and with enterprise end-users, often in collaboration with their global system integrator of choice, to fine-tune models and build AI applications. We believe that the depth and quality of our design support are key to improving our partner network's time-to-market, maintaining a high level of customer satisfaction, and fostering relationships that encourage our customers and partner network to use the next generation of our products within each platform. + +To encourage the development of applications optimized for our platforms and software, we seek to establish and maintain strong relationships in the software development community. Engineering and marketing personnel engage with key software developers to promote and discuss our platforms, as well as to ascertain individual product requirements and solve technical problems. Our developer program supports the development of AI frameworks, SDKs, and APIs for software applications and game titles that are optimized for our platforms. Our Deep Learning Institute provides in-person and online training for developers in industries and organizations around the world to build AI and accelerated computing applications that leverage our platforms. + +## Seasonality + +Our computing platforms serve a diverse set of markets such as data centers, gaming, professional visualization, and automotive. Our desktop gaming products typically see stronger revenue in the second half of our fiscal year. Historical seasonality trends may not repeat. + +## Manufacturing + +We utilize a fabless and contracting manufacturing strategy, whereby we employ and partner with key suppliers for all phases of the manufacturing process, including wafer fabrication, assembly, testing, and packaging. We use the expertise of industry-leading suppliers that are certified by the International Organization for Standardization in such areas as fabrication, assembly, quality control and assurance, reliability, and testing. Additionally, we can avoid many of the significant costs and risks associated with owning and operating manufacturing operations. While we may directly procure certain raw materials used in the production of our products, such as memory, substrates, and a variety of components, our suppliers are responsible for procurement of most raw materials used in the production of our products. As a result, we can focus our resources on product design, quality assurance, marketing, and customer support. In periods of growth, we may place non-cancellable inventory orders for certain product components in advance of our historical lead times, pay premiums, or provide deposits to secure future supply and capacity and may need to continue to do so. + +We have expanded our supplier relationships to build redundancy and resilience in our operations to provide long-term manufacturing capacity aligned with growing customer demand. While currently our supply chain is mainly concentrated in Asia, we are expanding into the U.S. and Latin America. We utilize foundries, such as Taiwan Semiconductor Manufacturing Company Limited, or TSMC, and Samsung Electronics Co., Ltd., or Samsung, to produce our semiconductor wafers. We purchase memory from SK Hynix Inc., Micron Technology, Inc., and Samsung. We utilize CoWoS technology for semiconductor packaging. We engage with independent subcontractors and contract manufacturers such as Hon Hai Precision Industry Co., Ltd., Wistron Corporation, and Fabrinet to perform assembly, testing and packaging of our final products. + +## Competition + +The market for our products is intensely competitive and is characterized by rapid technological change and evolving industry standards. We believe that the principal competitive factors in this market are performance, breadth of product offerings, access to customers and partners and distribution channels, software support, conformity to industry standard APIs, manufacturing capabilities, processor pricing, and total system costs. We believe that our ability to remain competitive will depend on how well we are able to anticipate the features and functions that customers and partners will demand and whether we are able to deliver consistent volumes of our products at acceptable levels of quality and at competitive prices. We expect competition to increase from both existing competitors and new market entrants with products that may be lower priced than ours or may provide better performance or additional features not provided by our products. In addition, it is possible that new competitors or alliances among competitors could emerge and acquire significant market share. + +A significant source of competition comes from companies that provide or intend to provide GPUs, CPUs, DPUs, embedded SoCs, and other accelerated, AI computing processor products, and providers of semiconductor-based high-performance interconnect products based on InfiniBand, Ethernet, Fibre Channel, and proprietary technologies. Some of our competitors may have greater marketing, financial, distribution and manufacturing resources than we do and may be more able to adapt to customers or technological changes. We expect an increasingly competitive environment in the future. + +Our current competitors include: + +- suppliers and licensors of hardware and software for discrete and integrated GPUs, custom chips and other accelerated computing solutions, including solutions offered for AI, such as Advanced Micro Devices, Inc., or AMD, Huawei Technologies Co. Ltd., or Huawei, and Intel Corporation, or Intel; +- large cloud services companies with internal teams designing hardware and software that incorporate accelerated or AI computing functionality as part of their internal solutions or platforms, such as Alibaba Group, Alphabet Inc., Amazon, Inc., or Amazon, Baidu, Inc., Huawei, and Microsoft Corporation, or Microsoft; +- suppliers of Arm-based CPUs and companies that incorporate hardware and software for CPUs as part of their internal solutions or platforms, such as Amazon, Huawei, and Microsoft; +- suppliers of hardware and software for SoC products that are used in servers or embedded into automobiles, autonomous machines, and gaming devices, such as Ambarella, Inc., AMD, Broadcom, Intel, Qualcomm Incorporated, Renesas Electronics Corporation, and Samsung, or companies with internal teams designing SoC products for their own products and services, such as Tesla, Inc.; and +- networking products consisting of switches, network adapters (including DPUs), and cable solutions (including optical modules) include such as AMD, Arista Networks, Broadcom, Cisco Systems, Inc., Hewlett Packard Enterprise Company, Huawei, Intel, Lumentum Holdings Inc., and Marvell Technology, Inc, as well as internal teams of system vendors and large cloud services companies. + +## Patents and Proprietary Rights + +We rely primarily on a combination of patents, trademarks, trade secrets, employee and third-party nondisclosure agreements, and licensing arrangements to protect our IP in the United States and internationally. Our currently issued patents have expiration dates from March 2026 to June 2045. We have numerous patents issued, allowed, and pending in the United States and in foreign jurisdictions. Our patents and pending patent applications primarily relate to our products and the technology used in connection with our products. We also rely on international treaties, organizations, and foreign laws to protect our IP. The laws of certain foreign countries in which our products are or may be manufactured or sold, including various countries in Asia, may not protect our products or IP rights to the same extent as the laws of the United States. This decreased protection makes the possibility of piracy of our technology and products more likely. We continuously assess whether and where to seek formal protection for innovations and technologies based on such factors as: + +• the location in which our products are manufactured; +• our strategic technology or product directions in different countries; +• the degree to which IP laws exist and are meaningfully enforced in different jurisdictions; and +- the commercial significance of our operations and our competitors' operations in particular countries and regions. + +We have licensed technology from third parties and expect to continue entering such license agreements. + +## Government Regulations + +Our worldwide business activities are subject to various laws, rules, and regulations of the United States as well as of foreign governments. + +Over the past three years, we have been subject to a series of shifting and expanding export control restrictions, impacting our ability to serve customers outside the United States. + +In August 2022, the U.S. government, or USG, announced export restrictions and export licensing requirements targeting China's semiconductor and supercomputing industries. These restrictions impacted exports of certain chips, as well as software, hardware, equipment and technology used to develop, produce and manufacture certain chips to China (including Hong Kong and Macau) and Russia, and specifically impact our A100 and H100 integrated circuits, DGX or any other systems or boards which incorporate A100 or H100 integrated circuits. In July 2023, the USG also informed us of an additional licensing requirement for a subset of A100 and H100 products destined to certain customers and other regions, including some countries in the Middle East. + +In October 2023, the USG announced new and updated licensing requirements for exports to China and Country Groups D:1, D:4, and D:5 (including but not limited to Saudi Arabia, the United Arab Emirates, and Vietnam, but excluding Israel) of our products exceeding certain performance thresholds, including, but not limited to, the A100, A800, H100, H800, L4, L40, L40S RTX 4090, GB200 NVL72, and B200. The licensing requirements also apply to the export of products exceeding certain performance thresholds to a party headquartered in, or with an ultimate parent headquartered in, Country Group D5, including China. + +In April 2025, the USG informed us that it requires a license for export to China (including Hong Kong and Macau) and D:5 countries, or to companies headquartered or with an ultimate parent therein, of our H2O integrated circuits and any other circuits achieving the H2O's memory bandwidth, interconnect bandwidth, or combination thereof. As a result of these requirements, we incurred a \$4.5 billion charge in the first quarter of fiscal year 2026 associated with H2O for excess inventory and purchase obligations, as the demand for H2O products diminished. + +In August 2025, the USG granted licenses that would allow us to ship certain H2O products to certain China-based customers. We generated approximately \$60 million in H2O revenue under those licenses. USG officials expressed an expectation that the USG will receive 15% or more of the revenue generated from licensed sales of our products, but the USG did not publish a regulation codifying such requirement. + +In February 2026, the USG granted a license that would allow us to ship small amounts of H200 products to specific China-based customers. To date, we have not generated any revenue under the H200 licensing program, and do not yet know whether any imports will be allowed into China. The license requires that the H200s go through an inspection process in the United States prior to any shipment to the customer. As a result, any H200 shipped under the new licensing program will be subject to a 25% tariff upon importation into the United States. + +In the event that we are able to sell licensed products into the China market, we may not be able to pass along all or any of the tariff to our customers, and may be subject to litigation, increased costs, and a harmed competitive position. + +The export controls applicable to China are complex and address a variety of parameters, including the total processing performance of a chip, the “performance density” of a chip, the interconnect bandwidth of a chip, and the memory bandwidth of a chip. Under the current rules and geopolitical landscape, we are unable to create and deliver a competitive product for China’s data center market that receives approval from both the USG and the Chinese government. As of the end of fiscal year 2026, we were effectively foreclosed from competing in China's data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide. Unless we are able to return with a product that meets the approval of both the USG and the Chinese government, our lost opportunity and the benefit to our competitors will have a material and adverse impact on our business, operating results, and financial condition. + +In addition to controls targeting D:1, D:4 and D:5 countries, the USG has also imposed worldwide export controls impacting our products, and may impose additional controls in the future. + +In January 2025, the USG published the AI Diffusion IFR in the Federal Register. The IFR would have imposed a worldwide licensing requirement on our data center products, such as our H200, GB200 and GB300. The AI Diffusion IFR would have divided the world into three tiers, relegating most countries to “Tier 2” status, and would have created a complex and burdensome scheme for licensing approvals. + +In May 2025, the USG announced that it would rescind the AI Diffusion IFR and implement a replacement rule. The scope, timing, and requirements of the forthcoming rule remain uncertain. The replacement rule may impose new restrictions on our products or operations and/or add license requirements that could have a material impact on our business, operating results, and financial condition. For example, in October 2025, the Senate passed the “GAIN AI Act” in the NDAA. The GAIN AI Act would restrict the Trump Administration’s ability to adapt the Biden Administration’s export control rules, and could also allow private U.S. persons to review and overturn licensing and foreign policy decisions made by the Trump Administration. + +Our competitive position has been harmed by export controls, and our competitive position and future results will be further harmed, over the long term, if the restrictions remain in place or are expanded in geographic, customer, or product scope, if customers purchase product from competitors, if customers develop their own internal solution, if we are unable to provide contractual warranty or other extended service obligations, if the USG does not grant licenses in a timely manner or denies licenses to significant customers or if we incur significant transition costs. + +The licensing process may not be resolved before significant business opportunities evaporate. Even if the USG grants any requested licenses, the licenses have already and may in the future be temporary, impose burdensome conditions regarding the installation, maintenance, and use of such products, or include financial or economic requirements that we or our customers or end users cannot or choose not to fulfill. The licensing requirements have already and may in the future benefit certain of our competitors, as the licensing process will make our pre-sale and post-sale technical support efforts more cumbersome and less certain and encourage customers in China, the Middle East, and other regions to pursue alternatives to our products, including semiconductor suppliers based in China, Europe, and Israel. + +Additionally, restrictions imposed by the Chinese government on the duration of gaming activities and access to games may adversely affect our Gaming revenue, and even if we are able to participate in the China data center compute market, increased oversight of digital platform companies may adversely affect our Data Center revenue. The Chinese government has encouraged customers to purchase from our China-based competitors and discouraged customers from purchasing, importing, or using our data center products, including any China-specific product designed to comply with U.S. export controls. + +While we work to enhance the resiliency and redundancy of our supply chain, which is currently concentrated in Asia, new and existing export controls or changes to existing export controls could limit alternative manufacturing locations and negatively impact our business. Refer to “Item 1A. Risk Factors – Risks Related to Regulatory, Legal, Our Stock, and Other Matters” for a discussion of this potential impact. + +Compliance with laws, rules, and regulations has not otherwise had a material effect upon our capital expenditures, results of operations, or competitive position and we do not currently anticipate material capital expenditures for environmental control facilities. Compliance with existing or future governmental regulations, including, but not limited to, those pertaining to IP ownership and infringement, taxes, import and export requirements and tariffs, anti-corruption, business acquisitions, foreign exchange controls and cash repatriation restrictions, data privacy requirements, competition and antitrust, advertising, employment, product regulations, cybersecurity, environmental, health and safety requirements, the responsible use of AI, climate change, cryptocurrency, and consumer laws, could further increase our costs, impact our competitive position, and otherwise may have a material adverse impact on our business, financial condition and results of operations in subsequent periods. Refer to “Item 1A. Risk Factors” for a discussion of these potential impacts. + +## Human Capital Management + +As of the end of fiscal year 2026, we had approximately 42,000 employees in 38 countries; 31,000 were engaged in research and development and 11,000 were engaged in sales, marketing, operations, and administrative positions. + +To execute our business strategy successfully, we focus on recruiting, developing, and retaining top global talent. + +Within our workforce, more than 80 percent have technical roles and more than half of the workforce hold an advanced degree. Our employees also help to surface top talent, with over 40 percent of our new hires in fiscal year 2026 coming from employee referrals. In fiscal year 2026, our turnover rate was 3.7 percent. + +We invest in employee development through on-the-job trainings and tuition reimbursement programs. + +Our compensation and benefits are designed to reward performance and align employee interests with those of our shareholders through equity participation and comprehensive health and financial wellness programs. We also utilize employee listening systems to gather feedback and maintain an inclusive culture where hiring and promotions are based on merit. + +## Information About Our Executive Officers + +The following sets forth certain information regarding our executive officers, their ages, and positions as of February 20, 2026: + +
NameAgePosition
Jen-Hsun Huang63President and Chief Executive Officer
Colette M. Kress58Executive Vice President and Chief Financial Officer
Ajay K. Puri71Executive Vice President, Worldwide Field Operations
Debora Shoquist71Executive Vice President, Operations
Timothy S. Teter59Executive Vice President and General Counsel
+ +Jen-Hsun Huang co-founded NVIDIA in 1993 and has served as our President, Chief Executive Officer, and a member of the Board of Directors since our inception. From 1985 to 1993, Mr. Huang was employed at LSI Logic Corporation, a computer chip manufacturer, where he held a variety of positions including as Director of Coreware, the business unit responsible for LSI's SOC. From 1983 to 1985, Mr. Huang was a microprocessor designer for AMD, a semiconductor company. Mr. Huang holds a B.S.E.E. degree from Oregon State University and an M.S.E.E. degree from Stanford University. + +Colette M. Kress joined NVIDIA in 2013 as Executive Vice President and Chief Financial Officer. Prior to NVIDIA, Ms. Kress most recently served as Senior Vice President and Chief Financial Officer of the Business Technology and Operations Finance organization at Cisco Systems, Inc., a networking equipment company, since 2010. At Cisco, Ms. Kress was responsible for financial strategy, planning, reporting and business development for all business segments, engineering and operations. From 1997 to 2010 Ms. Kress held a variety of positions at Microsoft, a software company, including, beginning in 2006, Chief Financial Officer of the Server and Tools division, where Ms. Kress was responsible for financial strategy, planning, reporting and business development for the division. Prior to joining Microsoft, Ms. Kress spent eight years at Texas Instruments Incorporated, a semiconductor company, where she held a variety of finance positions. Ms. Kress holds a B.S. degree in Finance from University of Arizona and an M.B.A. degree from Southern Methodist University. + +Ajay K. Puri joined NVIDIA in 2005 as Senior Vice President, Worldwide Sales and became Executive Vice President, Worldwide Field Operations in 2009. Prior to NVIDIA, he held positions in sales, marketing, and general management over a 22-year career at Sun Microsystems, Inc., a computing systems company. Mr. Puri previously held marketing, management consulting, and product development positions at Hewlett-Packard, an information technology company, Booz Allen Hamilton Inc., a management and technology consulting company, and Texas Instruments Incorporated. Mr. Puri holds a B.S.E.E. degree from the University of Minnesota, an M.S.E.E. degree from the California Institute of Technology and an M.B.A. degree from Harvard Business School. + +Debora Shoquist joined NVIDIA in 2007 as Senior Vice President of Operations and in 2009 became Executive Vice President of Operations. Prior to NVIDIA, Ms. Shoquist served from 2004 to 2007 as Executive Vice President of Operations at JDS Uniphase Corp., a provider of communications test and measurement solutions and optical products for the telecommunications industry. She served from 2002 to 2004 as Senior Vice President and General Manager of the Electro-Optics business at Coherent, Inc., a manufacturer of commercial and scientific laser equipment. Previously, she worked at Quantum Corp., a data protection company, as President of the Personal Computer Hard Disk Drive Division, and at Hewlett-Packard. Ms. Shoquist holds a B.S. degree in Electrical Engineering from Kansas State University and a B.S. degree in Biology from Santa Clara University. + +Timothy S. Teter joined NVIDIA in 2017 as Senior Vice President, General Counsel and Secretary and became Executive Vice President, General Counsel and Secretary in February 2018. Prior to NVIDIA, Mr. Teter spent more than two decades at the law firm of Cooley LLP, where he focused on litigating patent and technology related matters. Prior to attending law school, he worked as an engineer at Lockheed Missiles and Space Company, an aerospace company. Mr. Teter holds a B.S. degree in Mechanical Engineering from the University of California at Davis and a J.D. degree from Stanford Law School. + +## Available Information + +Our annual reports on Form 10-K, quarterly reports on Form 10-Q, current reports on Form 8-K and, if applicable, amendments to those reports filed or furnished pursuant to Section 13(a) or 15(d) of the Securities Exchange Act of 1934, as amended, or the Exchange Act, are available free of charge on or through our website, http://www.nvidia.com, as soon as reasonably practicable after we electronically file such material with, or furnish it to, the Securities and Exchange Commission, or the SEC. The SEC's website, http://www.sec.gov, contains reports, proxy and information statements, and other information regarding issuers that file electronically with the SEC. Our web site and the information on it or connected to it are not a part of this Annual Report on Form 10-K. + +## Item 1A. Risk Factors + +The following risk factors should be considered in addition to the other information in this Annual Report on Form 10-K. The following risks could harm our business, financial condition, results of operations or reputation, which could cause our stock price to decline. Additional risks, trends and uncertainties not presently known to us or that we currently believe are immaterial may also harm our business, financial condition, results of operations or reputation. + +## Risk Factors Summary + +## Risks Related to Our Industry and Markets + +• Failure to meet the evolving needs of our industry and markets may adversely impact our financial results. +- Competition could adversely impact our market share and financial results. + +## Risks Related to Demand, Supply, and Manufacturing + +\- Long manufacturing lead times and uncertain supply and capacity availability, combined with a failure to estimate customer demand accurately has led and could lead to mismatches between supply and demand. + +- Dependency on third-party suppliers and their technology to manufacture, assemble, test, or package our products reduces our control over product quantity and quality, manufacturing yields, and product delivery schedules and could harm our business. +- Defects in our products have caused and could cause us to incur significant expenses to remediate and could damage our business. + +## Risks Related to Our Global Operating Business + +- Adverse economic conditions may harm our business. +- International sales and operations are a significant part of our business, which exposes us to risks that could harm our business. +- Product, system security and data protection incidents or breaches, as well as cyber-attacks could disrupt our operations and adversely affect our financial condition, stock price and reputation. +• Business disruptions could harm our operations and financial results. +• Climate change may have a long-term impact on our business. +• We may not be able to realize the potential benefits of business investments or acquisitions, nor successfully integrate acquisition targets. +- A significant amount of our revenue stems from a limited number of partners and distributors and we have a concentration of sales to customers who purchase directly or indirectly from us, and our revenue could be adversely affected if we lose or are prevented from selling to any of these customers. +- Commercial arrangements expose us to counterparty risks, which may negatively impact our business, financial condition, or results of operations. +• We may be unable to attract, retain, and motivate our executives and key employees. +- Modification or interruption of our business processes and information systems may disrupt our business and internal controls. +- Our operating results have in the past fluctuated and may in the future fluctuate, and if our operating results are below the expectations of securities analysts or investors, our stock price could decline. + +## Risks Related to Regulatory, Legal, Our Stock, and Other Matters + +• We are subject to complex laws, rules, regulations, and political and other actions, including restrictions on the export of our products, which may adversely impact our business. +- Scrutiny regarding our corporate sustainability practices could result in financial, reputational, or operational harm and liability. +- Issues relating to the responsible use of our technologies, including AI, may result in reputational or financial harm and liability. +- Adequately protecting our IP rights could be costly, and our ability to compete could be harmed if we are unsuccessful or if we are prohibited from making or selling our products. +• We are subject to stringent and changing data privacy and security laws, rules, regulations, and other obligations. These areas could damage our reputation, deter customers, affect product design, or result in legal or regulatory proceedings and liability. +- Our operating results may be adversely impacted by additional tax liabilities, higher than expected tax rates, changes in tax laws, and other tax-related factors. +- Our business is exposed to the burden and risks associated with litigation, investigations, and regulatory proceedings. +- Delaware law, provisions in our governing documents and our agreement with Microsoft could delay or prevent a change in control. + +## Risk Factors + +## Risks Related to Our Industry and Markets + +## Failure to meet the evolving needs of our industry and markets may adversely impact our financial results. + +Our accelerated computing platforms experience rapid changes in technology, customer requirements, competitive products, and industry standards. + +Our success depends on our ability to: + +- timely identify industry changes, adapt our strategies, and develop new or enhance and maintain existing products and technologies that meet the evolving needs of our markets, including addressing unexpected shifts in industry standards or disruptive technological innovations that could render our products incompatible with those developed by other companies; +• develop or secure access to new products and technologies through investments in research and development; +- launch new offerings with new business models including software, services, and cloud solutions, as well as software-, infrastructure-, or platform-as-a-service solutions; +- expand the ecosystem for our products and technologies; +- meet evolving and prevailing customer and industry safety, security, reliability expectations, and compliance standards; +• manage product and software lifecycles to maintain customer and end-user satisfaction; +• develop, acquire, maintain, and secure access to the internal and external infrastructure needed to scale our business, including sufficient energy for powering data centers using our products, acquisition integrations, customer support, e-commerce, IP licensing capabilities, and cloud service capacity; and +• complete technical, financial, operational, compliance, sales and marketing investments for the above activities. + +We have invested in research and development in markets where we have a limited operating history, which may not produce meaningful revenue for several years, if at all. If we fail to develop or monetize new products and technologies, or if they do not become widely adopted, our financial results could be adversely affected. Obtaining design wins may involve a lengthy process and depends on our ability to anticipate and provide features and functionality that customers will demand. They also do not guarantee revenue. Failure to obtain a design win may prevent us from obtaining future design wins in subsequent generations. We cannot ensure that our products and technologies will provide value to our customers and partners. If we fail any of these key success criteria, our financial results may be harmed. + +We have entered into an intellectual property license arrangement with Groq, Inc., or Groq, that required significant, nonrefundable payments. Successfully incorporating the licensed technology into our architectures and product roadmaps requires significant engineering effort and may not occur on expected timelines or at all. The licensed technology may not achieve the desired results as designed or achieve customer or ecosystem adoption. The economic outcomes of this arrangement depend on our ability to translate the licensed technology into commercially viable products and services over time, and we may be unable to recover the associated costs or realize an adequate return on this spend. If our efforts to use the licensed technology are delayed or unsuccessful, our business, operating results, and financial condition could be negatively impacted. + +We entered into multi-year cloud service agreements to support our research and development activities. The timing and availability of these cloud services have changed and may continue to shift, impacting our revenue, expenses, and development timelines, and these arrangements may not deliver anticipated benefits. We also offer or plan to offer standalone software solutions, including NVIDIA AI Enterprise, NVIDIA Omniverse, NVIDIA DRIVE, and other software products. These business models or strategies may not be successful, and we may fail to sell any meaningful standalone software or services. We may incur significant costs and may not achieve any significant revenue from these offerings. + +## Competition could adversely impact our market share and financial results. + +Our target markets remain competitive, and competition may intensify with expanding and changing product and service offerings, industry standards, customer and market needs, new entrants and consolidations. Our competitors' products, services and technologies, including those mentioned above in this Annual Report on Form 10-K, may be cheaper or provide better functionality or features than ours, which has resulted and may in the future result in lower-than-expected selling prices or demand for our products. Some of our competitors operate their own fabrication facilities, and have longer operating histories, larger customer bases, more comprehensive IP portfolios and patent protections, more design wins, and greater financial, sales, marketing and distribution resources than we do. These competitors may be able to acquire market share and/or prevent us from doing so, more effectively identify and capitalize upon opportunities in new markets and end-user trends, more quickly transition their products, and impinge on our ability to procure sufficient foundry capacity and scarce input materials during a supply-constrained environment, which could harm our business. Some of our customers have in-house expertise and internal development capabilities similar to some of ours and can use or develop their own solutions to replace those we are providing. For example, others may offer cloud-based services that compete with our AI cloud service offerings, and we may not be able to establish market share sufficient to achieve the scale necessary to meet our business objectives. If we are unable to successfully compete in this environment, demand for our products, services and technologies could decrease, which may negatively impact our business. + +## Risks Related to Demand, Supply, and Manufacturing + +Long manufacturing lead times and uncertain supply and capacity availability, combined with a failure to estimate customer demand accurately, has led and could lead to mismatches between supply and demand. + +We have long manufacturing lead times and build finished products and maintain inventory in advance of anticipated demand. In periods of shortages impacting the semiconductor industry and/or limited supply or capacity in our supply chain, the lead times for certain supply may be extended. We have previously experienced and may continue to experience extended lead times of more than 12 months. To secure future supply and capacity, we have paid premiums, provided deposits, and entered into long-term supply agreements and capacity commitments, which have increased our product costs and this may continue. We may still be unable to secure sufficient commitments for capacity to address our business needs. + +If we inaccurately estimate demand, or our customers change orders, as we have experienced in the past, we may not be able to reduce our supply commitments in time, at the same rate, or at all. Significant mismatches between supply and demand have varied across our market platforms, resulted in both product shortages and excess inventory, significantly harmed our financial results and could reoccur. If we underestimate demand, and our foundry partners and contract manufacturers are unable to increase production or provide sufficient supply, we may not be able to meet increased customer demand in a timely manner, or at all. Our reputation and customer relationships could be damaged and we could lose revenue and market share. Additionally, since some of our products are part of a complex data center buildout, supply constraints or availability issues with respect to any one component have had and may have a broader revenue impact. For example, our ability to sell certain products has been and could be impeded if components necessary for the finished products are not available from third parties. + +If we overestimate demand, or if customers cancel or defer orders or choose to purchase from our competitors, we may not be able to utilize on-hand inventory or reduce purchase commitments accordingly. We have had to reduce average selling prices, including due to our channel pricing programs, increase prices for certain of our products as a result of our suppliers' increase in prices, write down our inventory, incur cancellation penalties, and record impairments, and may have to do so in the future. The impact of these risks would be amplified by our non-cancellable and non-returnable purchase orders placed in advance of our historical lead times and could be exacerbated if we need to make changes to the design of future products. These risks have increased and may continue to increase as our purchase obligations and prepaids have grown and are expected to continue to grow and become a greater portion of our total supply. All of these factors may negatively impact our gross margins and financial results. + +Factors that have caused and/or could in the future cause us to underestimate or overestimate demand, and impact the timing and volume of our revenue, include: + +• changes in product development cycles and time to market; +• competing technologies and competitor product releases, announcements or other actions; +• changes in business and economic conditions; +- sudden or sustained government lockdowns or public health issues; +• rapidly changing technology or customer requirements; +• the availability of sufficient data center capacity or energy for customers to procure; +- new product introductions and transitions resulting in less demand for existing products; +- new or unexpected end-use cases; +• increase in demand for competitive products; +• changes in end-user demand; + +• purchasing decisions made, and inventory levels held by, distributors, ODMs, OEMs, system integrators, other channel partners and other third parties; + +- the ability of developers, end customers and other third parties to build, enhance, and maintain accelerated computing applications that leverage our platforms; +• the demand for accelerated computing, AI-related cloud services, or large language models; +• changes that impact the ecosystem for the architectures underlying our products and technologies; +• government actions or changes in governmental policies, such as export controls, increased restrictions on gaming usage, or tariffs; +- our customers' and partners' ability to secure capital and energy and to build complex datacenter infrastructure timely; and +• the availability of third-party content on our platforms, such as GeForce NOW. + +The availability of data centers, energy, and capital to support the buildout of NVIDIA AI infrastructure by our customers and partners is crucial, and any shortage of these and other necessary resources could impact our future revenue and financial performance. Expanding energy capacity to meet demand is a complex, multi-year process involving significant regulatory, technical, and construction challenges. In addition, access to capital can be particularly constrained for less-capitalized companies, which may face difficulties securing financing for large-scale infrastructure projects. These limitations could delay customer and partner deployments or reduce the scale of accelerated computing and AI adoption. + +Challenges in estimating demand could become more pronounced or volatile in the future on both a global and regional basis. Extended lead times may occur if we experience other supply constraints caused by natural disasters, pandemics or other events. Geopolitical tensions in regions where we rely on suppliers, contract manufacturers, and assembly partners that are critical to our supply continuity, could have a material adverse impact on us. Publicly announced intentions by governments or other companies to purchase our products can further complicate our demand estimates, as such announcements are often non-binding and may not result in committed volumes. + +We continue to increase our supply and capacity purchases with existing and new suppliers to support our demand projections and increasing complexity of our data center products. We expect supply constraints to be a headwind to Gaming in the first quarter of fiscal year 2027 and beyond. We have also entered and may continue to enter into prepaid manufacturing and capacity agreements to supply both current and future products. The increased purchase volumes and integration of new suppliers and contract manufacturers into our supply chain creates more complexity in managing multiple suppliers with variations in production planning, execution and logistics. Our expanding product portfolio and varying component compatibility and quality may lead to increased inventory levels. We have incurred and may in the future incur inventory provisions or impairments if our inventory or supply or capacity commitments exceed demand for our products or demand declines. + +We are increasing our U.S.-based manufacturing and investing in specialized equipment and processes to support domestic production. We may experience delays or difficulties in scaling production as planned. Our ability to increase manufacturing capabilities will depend on the domestic manufacturing ecosystem's capacity to ramp production supply to the required volume timely. Delays or shortfalls could impact our ability to meet demand. + +Product transitions are complex and we often ship both new and prior architecture products simultaneously as our channel partners prepare to ship and support new products. We are generally in various stages of transitioning the architectures of our Data Center, Gaming, Professional Visualization, and Automotive products. The computing industry is experiencing a broader and faster launch cadence of accelerated computing platforms to meet a growing and diverse set of AI opportunities. We have introduced a new product and architecture cadence of our Data Center solutions where we seek to complete new computing solutions each year and provide a greater variety of Data Center offerings. The increased frequency of these transitions and the larger number of products and product configurations may magnify the challenges associated with managing our supply and demand which may further create volatility in our revenue. Qualification time for new products, customers anticipating product transitions, and channel partners reducing channel inventory of prior architectures ahead of new product introductions can reduce, or create volatility in, our revenue. Customers may delay adopting new architectures if their data center infrastructure is not ready, which could affect the timing of our revenue. We have experienced and may in the future experience reduced demand for current generation architectures when customers anticipate transitions, and we may be unable to sell multiple product architectures at the same time for current and future architecture transitions. Our financial results have been and may in the future be negatively impacted if we are unable to execute our architectural transitions as planned for any reason. The increased frequency and complexity of newly introduced products could result in unanticipated quality or production issues that could increase the magnitude of inventory provisions, warranty, or other costs or result in product delays. For example, our gross margins in the second quarter of fiscal year 2025 were negatively impacted by inventory provisions for low-yielding Blackwell material. + +We incur significant engineering development resources for new products, and changes to our product roadmap may impact our ability to develop other products or adequately manage our supply chain cost. Customers may delay purchasing existing products as we increase the frequency of new products or may not be able to adopt our new products as fast as forecasted, both impacting the timing of our revenue and supply chain cost. While we have managed prior product transitions and have sold multiple product architectures at the same time, these transitions are difficult, may impair our ability to predict demand and impact our supply mix, and may cause us to incur additional costs. + +Demand estimates for our products, applications, and services can be incorrect, which may create volatility in our revenue or supply levels. We may not be able to generate significant revenue from them. Because our products may be used in multiple use cases and applications, it is difficult to estimate with any reasonable degree of precision the impact of accelerated computing and AI models on our reported revenue or forecasted demand. + +The use of our GPUs for new, mercurial, or trendy applications, has impacted and can impact in the future, demand for our products, including by leading to inconsistent spikes and drops in demand. For example, several years ago, our Gaming GPUs began to be used for mining digital currencies, such as Ethereum. It is difficult for us to estimate with any reasonable degree of precision the past or current impact of cryptocurrency mining, or forecast the future impact of cryptocurrency mining, on demand for our products. Volatility in the cryptocurrency market, including new compute technologies, price changes in cryptocurrencies, government cryptocurrency policies and regulations, new cryptocurrency standards and changes in the method of verifying blockchain transactions, has impacted and can in the future impact cryptocurrency mining and demand for our products and can further impact our ability to estimate demand for our products. Changes to cryptocurrency standards and processes including, but not limited to, the Ethereum 2.0 merge in 2022, have reduced and may in the future decrease the usage of GPUs for Ethereum mining. This has created and may in the future create increased aftermarket sales of our GPUs, which could negatively impact retail prices for our GPUs and reduce demand for our new GPUs. In general, our new products or previously sold products may be resold online or on the unauthorized “gray market,” which also makes demand forecasting difficult. Gray market products and reseller marketplaces compete with our new products and distribution channels. Our inability to accurately predict our demand that arises from new use cases may create volatility in our revenue. + +## Dependency on third-party suppliers and their technology to manufacture, assemble, test, or package our products reduces our control over product quantity and quality, manufacturing yields, and product delivery schedules and could harm our business. + +We depend on foundries to manufacture our semiconductor wafers using their fabrication equipment and techniques. We do not assemble, test, or package our products, but instead contract with independent subcontractors. These subcontractors assist with procuring components used in our systems, boards, and products. We face risks which have adversely affected or could adversely affect our ability to meet customer demand and scale our supply chain, negatively impact longer-term demand for our products and services, and adversely affect our business operations, gross margin, revenue and/or financial results, including: + +- lack of guaranteed supply of components and capacity; +• decommitment by our suppliers; +- potential higher wafer and component prices resulting from incorrectly estimating demand and failing to place orders with our suppliers with sufficient quantities or timely; +- failure by our foundries or contract manufacturers to procure raw materials or provide adequate levels of manufacturing or test capacity for our products; +- failure by our foundries to develop, obtain, or successfully implement high quality process technologies, including transitions to smaller geometry process technologies such as advanced process node technologies and memory designs needed to manufacture our products; +• failure by our suppliers to comply with our policies and expectations and emerging regulatory requirements; +- limited number and geographic concentration of global suppliers, foundries, contract manufacturers, assembly and test providers and memory manufacturers; +- loss of a supplier and additional expense and/or production delays as a result of qualifying a new foundry or subcontractor and commencing volume production or testing in the event of a loss, addition or change of a supplier; +- lack of direct control over product quantity, quality, and delivery schedules; +- integration of new suppliers and contract manufacturers creating more complexity in managing multiple suppliers with variations in production planning, execution, and logistics; +- suppliers or their suppliers failing to provide high quality products and/or making changes to their products without our qualification; + +- delays in product shipments, shortages, a decrease in product quality and/or higher expenses in the event our subcontractors or foundries prioritize our competitors' or other customers' orders over ours; +- requirements to place orders that are not cancellable upon changes in demand or requirements to prepay for supply in advance; +- low manufacturing yields resulting from issues in our product design or a foundry's proprietary process technology; +• suppliers extending lead times and/or increasing costs during shortages; and +- disruptions in manufacturing, assembly and other processes due to closures related to heat waves, earthquakes, fires, or other natural disasters, electricity conservation efforts, pandemics, and cybersecurity incidents. + +## Defects in our products have caused and could cause us to incur significant expenses to remediate, which can damage our reputation and cause us to lose market share. + +Our hardware and software product and service offerings are complex. They have in the past and may in the future contain defects, security vulnerabilities, experience failures, or unsatisfactory performance due to issues in design, fabrication, packaging, materials, bugs and/or use within a system. These risks may also increase when our products are introduced into new devices, markets, technologies and applications, or new versions are released, and when we rely on partners to supply and manufacture components that are used in our products, as these arrangements reduce our direct control over production. AI software products that we or our partners offer rely on training data that may originate from third parties and new training methods, and the resulting products may contain unknown or undetected defects and errors, or reflect unintended bias. Although arrangements with component providers may contain provisions for product defect expense reimbursement, we generally remain responsible to the customer for warranty product defects that may occur from time to time. Some failures in our products or services have been in the past and may in the future be only discovered after a product or service has been shipped or used. + +Undiscovered vulnerabilities in our products or services could result in loss of data or intangible property, or expose our customers to unscrupulous third parties who develop and deploy malicious software programs that could attack our products or services. Defects or failure of our offerings to perform to specifications could lead to substantial damage to the products in which our offerings have been integrated by OEMs, ODMs, AIB manufacturers, automotive manufacturers, and tier 1 automotive suppliers, and to the user of such end product. Such defects have in the past had an adverse effect on our cost and supply of components and finished goods and may in the future cause us to incur significant warranty, support, and repair or replacement costs as part of a product recall or otherwise, write-off the value of related inventory, and divert the attention of our engineering and management personnel from our product development efforts to find and correct the issue. Our efforts to remedy these issues may not be timely or satisfactory to our customers. An error or defect in new products, releases or related software drivers after commencement of commercial shipments could result in failure to achieve market acceptance, loss of design wins, temporary or permanent withdrawal from a product or market and harm to our relationships with existing and prospective customers and partners and consumers' perceptions of our brand, which would in turn negatively impact our business operations, gross margin, revenue and/or financial results. We may be required to reimburse our customers, partners or consumers, including for costs to repair or replace products in the field or in connection with indemnification obligations, or pay fines imposed by regulatory agencies. + +In general, if a product liability claim regarding any of our products is brought against us, even if the alleged damage is due to the actions or inactions of a third party, such as within our supply chain, the cost of defending the claim could be significant and would divert the efforts of our technical and management personnel and harm our business. Further, our business liability insurance may be inadequate or future coverage may be unavailable on acceptable terms, which could adversely impact our financial results. + +## Risks Related to Our Global Operating Business + +## Adverse economic conditions may harm our business. + +Economic and industry uncertainty or changes, including recession or slowing growth, inflation, changes or uncertainty in fiscal, monetary, or trade policy, disruptions to capital markets and the banking system, currency fluctuations, higher interest rates, tighter credit, lower capital expenditures by businesses, including on IT infrastructure, increases in unemployment, labor shortages, and lower consumer confidence and spending, global supply chain constraints, and global economic and geopolitical developments, including the implementation of tariffs by the USG or other governments, have in the past and/or could in the future have adverse, wide-ranging effects on our business and financial results, including: + +\- increased costs for wafers, components, logistics, and other supply chain expenses, which have negatively impacted our gross margin in the past and may do so in the future; + +- increased supply, employee, facilities and infrastructure costs and volatility in the financial markets, which have reduced and may in the future reduce our margins; +• decrease in demand for our products, services and technologies and those of our customers, partners or licensees; +- the inability of our suppliers to deliver on their supply commitments to us and our customers' or our licensees' inability to supply products to customers and/or end users; +• limits on our ability to forecast operating results and make business decisions; +- the insolvency of key suppliers, distributors, customers, CSPs, data center providers, licensing parties or other third parties we rely on; +- reduced profitability of customers, which may cause them to scale back operations, exit businesses, file for bankruptcy protection and potentially cease operations, or lead to mergers, consolidations or strategic alliances among other companies, which could adversely affect our ability to compete effectively; +- increased credit and collectability risks, higher borrowing costs or reduced availability of capital markets, reduced liquidity, adverse impacts on our customers and suppliers; and +- failures of counterparties, including financial institutions and insurers, asset impairments, and declines in the value of our financial instruments. + +Adverse developments affecting financial institutions, such as bank failures or instability, or concerns or speculation about similar events or risks, could lead to market-wide liquidity problems and other disruptions, which could impact our customers' ability to fulfill their payment obligations to us, our vendors' ability to fulfill their contractual obligations to us, or our ability to fulfill our own obligations. + +Additionally, we maintain a portfolio of liquid investments for cash management purposes, including various holdings, types, and maturities. These investments are subject to general credit, liquidity, market and interest rate risks, which may be exacerbated by market downturns or events that affect global financial markets, as described above. A majority of our investment portfolio comprises USG securities. A decline in global financial markets for long periods or a downgrade of the USG credit rating due to an actual or threatened default on government debt could result in higher interest rates, a decline in the value of the U.S. dollar, reduced market liquidity or other adverse conditions. These factors could cause an unrealized or realized loss position in our investments or require us to record impairment charges. + +## International sales and operations are a significant part of our business, which exposes us to risks that could harm our business. + +We sell our products internationally, and we also have operations and conduct business internationally. Our semiconductor wafers are manufactured, assembled, tested and packaged by third parties located outside of the United States, and we generated 31% of our revenue in fiscal year 2026 from sales outside the United States. The market in China, where our offerings are limited by export controls, is highly competitive and we expect it to remain competitive going forward. The global nature of our business subjects us to a number of risks and uncertainties, which have had in the past and could in the future have a material adverse effect on our business, financial condition and results of operations. These include domestic and international economic and political conditions in countries in which we and our suppliers and manufacturers do business, government lockdowns to control case spread of global or local health issues, differing legal standards with respect to protection of IP and employment practices, different domestic and international business and cultural practices, disruptions to capital markets, counter-inflation policies, currency fluctuations, natural disasters, acts of war or other military actions, terrorism, public health issues, restrictions on international trade, such as tariffs, sanctions, and other controls on imports or exports, and catastrophic events. + +## Product, system security, and data protection incidents or breaches, as well as cyber-attacks, could disrupt our operations, reduce our expected revenue, increase our expenses, and significantly harm our business and reputation. + +Security breaches, computer malware, social-engineering attacks, denial-of-service attacks, software bugs, server malfunctions, software or hardware failures, loss of data or other information technology assets, and other cyber-attacks are becoming increasingly sophisticated, making it more difficult to successfully detect, defend against them or implement adequate preventative measures. + +Cyber-attacks, including ransomware attacks by organized criminal threat actors, nation-states, and nation-state-supported actors, may become more prevalent and severe. Our ability to recover from ransomware attacks may be limited if our backups have been affected by the attack, or if restore from backups is delayed or not feasible. + +Individuals, groups of hackers and sophisticated organizations, including nation-states and nation-state-supported actors, and other threat actors have engaged and are expected to continue to engage in cyber-attacks. Additionally, some actors are using AI technology to launch more automated, targeted and coordinated attacks. Due to geopolitical conflicts and during times of war or other major conflicts, we and the third parties we rely upon may be subject to a heightened risk of cyber-attacks that could materially disrupt our ability to provide services and products. We may also face cybersecurity threats due to error or intentional misconduct by employees, contractors or other third-party service providers. Certain aspects of effective cybersecurity are dependent upon our employees, contractors and/or other third-party service providers safeguarding our sensitive information and adhering to our security policies and access control mechanisms. We have in the past experienced, and may in the future experience, security incidents arising from a failure to properly handle sensitive information or adhere to our security policies and access control mechanisms, including, for example, employees posting company data on third-party websites without permission, and, although no such events have had a material adverse effect on our business, there can be no assurance that an insider threat or error will not result in an incident that is material to us or lead to negative publicity. Furthermore, we rely on products and services provided by third-party suppliers to operate certain critical business systems, including without limitation, cloud-based infrastructure, encryption and authentication technology, employee email and other functions, which exposes us to supply-chain attacks or other business disruptions. We cannot guarantee that third parties and infrastructure in our supply chain or our partners' supply chains have not been compromised or that they do not contain exploitable vulnerabilities, defects or bugs that could result in a breach of or disruption to our information technology systems, including our products and services, or the third-party information technology systems that support our services. We have incorporated third-party data into some of our AI models and used open-source datasets to train our models and may continue to do so. These datasets may be flawed, insufficient, or contain certain biased information, and may otherwise decrease resilience to security incidents that may compromise the integrity of our AI outputs, leading to potential reputational damage, regulatory scrutiny, or adverse impacts on the performance and reliability of our products, which could, in turn, affect our partners' operations, customer trust, and our revenue. We may have limited insight into the data privacy or security practices of third-party suppliers, including for our AI algorithms. Our ability to monitor these third parties' information security practices is limited, and they may not have adequate information security measures in place. In addition, if one of our third-party suppliers suffers a security incident (which has happened in the past and may happen in the future), our response may be limited or more difficult because we may not have direct access to their systems, logs and other information related to the security incident. Additionally, we are incorporated into the supply chain of a large number of entities worldwide and, as a result, if our products or services are compromised, a significant number of our customers and their data could be affected, which could result in potential liability and harm our business. + +To defend against security incidents, we must continuously engineer more secure products and enhance security and reliability features, which is expected to result in increased expenses. We must also continue to develop our security measures, including training programs and security awareness initiatives, designed to ensure our suppliers have appropriate security measures in place, and continue to meet the evolving security requirements of our customers, applicable industry standards, and government regulations. While we invest in training programs and security awareness initiatives and take steps to detect and remediate certain vulnerabilities that we have identified, we may not always be able to prevent threats or detect and mitigate all vulnerabilities in our security controls, systems or software, including third-party software we have installed, as such threats and techniques change frequently and may not be detected until after a security incident has occurred. Further, we may experience delays in developing and deploying remedial measures designed to address identified vulnerabilities. These vulnerabilities could result in reputational and financial harm, and if exploited, these vulnerabilities could result in a security incident. + +We hold confidential, sensitive, personal and proprietary information, including information from partners and customers. Breaches of our security measures, along with reported or perceived vulnerabilities or unapproved dissemination of proprietary information or sensitive or confidential data about us or third parties, could expose us and the parties affected to a risk of loss, or misuse of this information, potentially resulting in litigation and subsequent liability, regulatory inquiries or actions, damage to our brand and reputation or other harm, including financial, to our business. For example, we hold proprietary game source code from third-party partners in our GFN service. Breaches of our GFN security measures, which have happened in the past, could expose our partners to the risk of loss or misuse of this source code, damage both us and our partners, and expose NVIDIA to potential litigation and liability. If we or a third party we rely on experience a security incident, which has occurred in the past, or are perceived to have experienced a security incident, we may experience adverse consequences, including government enforcement actions, additional reporting requirements and/or oversight, restrictions on processing data, litigation, indemnification obligations, reputational harm, diversion of funds, diversion of management attention, financial loss, loss of data, material disruptions in our systems and operations, supply chain, and ability to produce, sell and distribute our goods and services, and other similar harms. Inability to fulfill orders, delayed sales, lower margins or lost customers as a result of these disruptions could adversely affect our financial results, stock price and reputation. We are required by certain data privacy and security obligations to notify relevant stakeholders, including affected individuals, customers, regulators and investors, of security incidents, and mandatory disclosure of such incidents could lead to negative publicity. In addition to experiencing a security incident, third parties may gather, collect or infer sensitive information about us from public sources, data brokers or other means that reveals competitively sensitive details about our organization and could be used to harm our business. + +## Business disruptions could harm our operations, lead to a decline in revenue and increase our costs. + +Factors that have caused and/or could in the future cause disruptions to our worldwide operations include: natural disasters, extreme weather conditions, power or water shortages, critical infrastructure failures, telecommunications failures, supplier disruptions, terrorist attacks, acts of violence, political and/or civil unrest, acts of war or other military actions, epidemics or pandemics, abrupt regulatory changes, and other natural or man-made disasters and catastrophic events. Our corporate headquarters, a large portion of our current data center capacity, and a portion of our research and development activities are located in California, and other critical business operations, finished goods inventory and some of our suppliers are located in Asia, making our operations vulnerable to natural disasters such as earthquakes, wildfires or other business disruptions occurring in these geographical areas. Catastrophic events can also have an impact on third-party vendors who provide us critical infrastructure services for IT and research and development systems and personnel. Geopolitical and domestic political developments and other events beyond our control can increase economic volatility globally. Political instability, changes in government or adverse political developments in or around any of the major countries in which we do business may harm our business, financial condition, and results of operations. Worldwide geopolitical tensions and conflicts, including but not limited to China, Hong Kong, Israel, Korea and Taiwan where the manufacture of our product components and final assembly of our products are concentrated may result in changing regulatory requirements, and other disruptions that could impact our operations and operating strategies, product demand, access to global markets, hiring, and profitability. For example, other countries have restricted and may continue in the future to restrict business with the State of Israel, where we have engineering, sales support operations and manufacturing, and companies with Israeli operations, including by economic boycotts. Our operations could be harmed and our costs could increase if manufacturing, logistics, or other operations are disrupted for any reason, including natural disasters, high heat events, water shortages, power shortages, information technology system failures or cyberattacks, military actions or economic, and business, labor, environmental, public health, or political issues. The ultimate impact on us, our third-party foundries and other suppliers of being located and consolidated in certain geographical areas is unknown. In the event a disaster, war, or catastrophic event affects us, the third-party systems on which we rely, or our customers, our business could be harmed as a result of declines in revenue, increases in expenses, and substantial expenditures and time spent to fully resume operations. Our business continuity and disaster recovery planning may not be sufficient for all eventualities. All of these risks and conditions could materially adversely affect our future sales and operating results. + +We are monitoring the impact of the geopolitical conflict in and around Israel on our operations, including the health and safety of our approximately 6,000 employees in the region who primarily support the research and development, operations, and sales and marketing of our networking products. We have experienced periods where some of our employees in the region have been on active military duty for an extended period, which caused limited disruption to our product development or operations. We have not experienced significant impact or expense to our business; however, if the conflict is further extended or expanded, it could impact future product development, operations, and revenue or create other uncertainty for our business. + +Additionally, interruptions or delays in services from CSPs, data center co-location partners, and other third parties on which we rely, including due to the events described above or other events such as the insolvency of these parties, could impair our ability to provide our products and services and harm our business. As we increase our reliance on these third-party systems and services, our exposure to damage from service interruptions, defects, disruptions, outages, shortages and other performance and quality problems may increase. Data centers depend on access to clean water and predictable energy. Power or water shortages, land or permitting constraints, or regulations that limit energy, water, or land availability, could impair the ability of our customers to expand their data center capacity and consume our products and services, which may in turn negatively impact our business. + +## Climate change may have a long-term impact on our business. + +Climate change may have an increasingly adverse impact on our business and on our customers, partners and vendors. Water and energy availability and reliability in the regions where we conduct business is critical, and certain of our facilities may be vulnerable to the impacts of extreme weather events. Extreme heat and wind coupled with dry conditions in Northern California may lead to power safety shut offs due to wildfire risk, which can have adverse implications for our Santa Clara, California headquarter offices and data centers, including impairing the ability of our employees to work effectively. Climate change, its impact on our supply chain and critical infrastructure worldwide and its potential to increase political instability in regions where we, our customers, partners and our vendors do business, may disrupt our business and cause us to experience higher attrition, losses and costs to maintain or resume operations. Although we maintain insurance coverage for a variety of property, casualty, and other risks, the types and amounts of insurance we obtain vary depending on availability and cost. Some of our policies have large deductibles and broad exclusions, and our insurance providers may be unable or unwilling to pay a claim. Losses not covered by insurance may be large, which could harm our results of operations and financial condition. + +Our business and those of our suppliers and customers is subject to sustainability-related laws, regulations and lawsuits. New or proposed regulations relating to carbon taxes, fuel or energy taxes, pollution limits, sustainability-related disclosure and governance and supply chain governance could result in greater direct costs, including costs associated with changes to manufacturing processes or the procurement of raw materials used in manufacturing processes, increased capital expenditures to improve facilities and equipment, higher compliance and energy costs to reduce emissions, other compliance costs, and greater indirect costs resulting from our customers and/or suppliers incurring additional compliance costs that are passed on to us. These costs and restrictions could harm our business and results of operations by increasing our expenses or requiring us to alter our operations and product design activities. + +Stakeholder groups may find us insufficiently responsive to the implications of climate change, and therefore we may face legal action or reputational harm. We may not achieve our stated sustainability-related goals, which could harm our reputation, or we may incur additional, unexpected costs to achieve such goals. We may also experience contractual disputes due to supply chain delays arising from climate change-related disruptions, which could result in increased litigation and costs. + +We also face risks related to business trends that may be influenced by climate change concerns. Our business could be negatively impacted by concerns around the high absolute energy requirements of our GPUs, despite their much more energy efficient design and operation relative to alternative computing platforms. + +We may not be able to realize the potential benefits of business investments or acquisitions, and we may not be able to successfully integrate acquired companies, which could hurt our ability to grow our business, develop new products or sell our products. + +We acquire and invest in businesses that offer products, services and technologies that we believe will help expand or enhance our strategic objectives. Acquisitions or investments involve significant challenges and risks and could impair our ability to grow our business, develop new products or sell our products and ultimately could have a negative impact on our financial results. If we pursue a particular transaction, we may limit our ability to enter into other transactions that could help us achieve our other strategic objectives. If we are unable to timely complete acquisitions, including due to delays and challenges in obtaining regulatory approvals, we may be unable to pursue other transactions, we may not be able to retain critical talent from the target company, technology may evolve and make the acquisition less attractive, and other changes can take place, which could reduce the anticipated benefits of the transaction and negatively impact our business. Regulators could also impose conditions that reduce the ultimate value of our acquisitions. In addition, to the extent that our perceived ability to consummate acquisitions is harmed, future acquisitions may be more difficult, complex or expensive. + +Our investments in companies could create volatility and fluctuations in our results. These investments may generate realized and unrealized gains or losses and we could realize losses up to the value of the investments. We have invested and may continue to invest in our ecosystem. Many of these companies may not achieve profitability in the near term, or at all, and there is no guarantee that we will realize a return on our investment. We may continue to invest in companies to further our strategic objectives and to support certain key business initiatives, which could be subject to delays and challenges in obtaining regulatory approvals. Our investments in private companies include early-stage companies still defining their strategic direction. Many of the securities in which we invest are non-marketable and illiquid at the time of our initial investment. To the extent any of the companies in which we invest are not successful, we could recognize an impairment and/or lose all or part of our investment. + +We are finalizing an investment and partnership agreement with OpenAI. There is no assurance that we will enter into an investment and partnership agreement with OpenAI or that a transaction will be completed. + +Our investment portfolio contains industry sector concentration risks, and a decline in any one or multiple industry sectors could increase our impairment losses. + +We face additional risks related to acquisitions and strategic investments, including the diversion of capital and other resources, including management's attention; difficulty in realizing a satisfactory return and uncertainties to realize the benefits of an acquisition or strategic investment, if at all; difficulty or inability in obtaining governmental, regulatory approval or restrictions or other consents and approvals or financing; legal proceedings initiated as a result of an acquisition or investment; and potential failure of our due diligence processes to identify significant issues with the assets or company in which we are investing or are acquiring. + +Additional risks relating to acquisitions include, but are not limited to: + +- difficulty in integrating the technology, systems, products, policies, processes, or operations and integrating and retaining the employees, including key personnel, of the acquired business; +• assumption of liabilities and incurring amortization expenses, impairment charges to goodwill or write-downs of acquired assets; +• integrating accounting, forecasting and controls, procedures and reporting cycles; +• coordinating and integrating operations, particularly in countries in which we do not currently operate; +- stock price impact, fines, fees or reputation harm if we are unable to obtain regulatory approval for an acquisition or are otherwise unable to close an acquisition; +- potential issuances of debt to finance our acquisitions, resulting in increased debt, increased interest expense, and compliance with debt covenants or other restrictions; +• the potential for our acquisitions to result in dilutive issuances of our equity securities; + +• the potential variability of the amount and form of any performance-based consideration; +- negative changes in general economic conditions in the regions or the industries in which we or our target operate; +• exposure to additional cybersecurity risks and vulnerabilities; and +- impairment of relationships with, or loss of our or our target's employees, vendors and customers. + +For example, when integrating acquisition target systems into our own, we have experienced and may continue to experience challenges including lengthy and costly systems integration, delays in purchasing and shipping products, difficulties with system integration via electronic data interchange and other processes with our key suppliers and customers, and training and change management needs of integration personnel. These challenges have impacted our results of operations and may continue to do so in the future. + +## We receive a significant amount of our revenue from a limited number of partners and distributors and we have a concentration of sales to customers who purchase directly or indirectly from us, and our revenue could be adversely affected if we lose or are prevented from selling to any of these customers. + +We have experienced periods where we receive a significant amount of our revenue from a limited number of customers, and this trend may continue. For fiscal year 2026, sales to one direct customer represented 22% of total revenue and sales to another direct customer represented 14% of total revenue, all of which were primarily attributable to the Compute & Networking segment. With several of these partners, we are selling multiple products and systems in our portfolio through their channels. Our operating results depend on sales to our partner network, as well as the ability of these partners to sell products that incorporate our technologies. We have a small number of partners that are involved in system integration with our key customers. As our system design becomes increasingly complex, system integrators may be unable to meet specifications of our key customers. Changes in our partners' or customers' business models or their ownership can reduce the number of partners available to us and harm our ability to sell our advanced data center systems to customers. In the future, these partners may decide to purchase fewer products, not to incorporate our products into their ecosystem, or to alter their purchasing patterns in some other way. Because most of our sales are made on a purchase order basis, our customers can generally cancel, change, or delay product purchase commitments with little notice to us and without penalty. Our partners or customers may develop their own solutions; our customers may purchase products from our competitors; and our partners may discontinue sales or lose market share in the markets for which they purchase our products, all of which may alter partners' or customers' purchasing patterns. Many of our indirect customers often do not purchase directly from us but through multiple OEMs, ODMs, system integrators, distributors, and other channel partners. We generate a significant amount of our revenue from a limited number of indirect customers, and we estimate some individually representing 10% or more of our revenue. If end demand increases or our finished goods supply availability is concentrated near a quarter end, the system integrators, distributors, and channel partners may have limited ability to increase their credit, which could impact the timing and amount of our revenue. The loss of any of our large customers, a significant reduction in purchases by them, our inability to sell to a customer due to U.S. or other countries' trade restrictions, or any difficulties in collecting accounts receivable would likely harm our financial condition and results of operations. + +## Commercial arrangements expose us to counterparty risks. + +We have entered and may in the future enter into commercial arrangements, including long-term capacity purchase obligations and financial guarantees, and have been asked to offer financing arrangements to support our customers' and partners' buildout of datacenter infrastructure. We have not entered into any financing arrangements. Commercial arrangements expose us to counterparty risk, including customers' or partners' inability to fulfill their financial commitments and secure necessary financing or infrastructure, the occurrence of significant project delays, and counterparty financial distress or insolvency, all of which may negatively impact our business, financial condition, or results of operations. Financing arrangements, if undertaken, may in some circumstances result in lower upfront cash flows associated with extended payment terms or payment terms made over a multi-year term and may increase credit risk. + +## If we are unable to attract, retain and motivate our executives and key employees, our business may be harmed. + +To remain competitive and successfully execute our business strategy, we must attract, retain, and motivate our executives and key employees, as well as recruit and develop exceptional talent. However, labor is subject to external factors that are beyond our control, including our industry's increasingly highly competitive market for skilled workers and leaders, and workforce participation rates. Changes in immigration and work permit regulations, or in their administration or interpretation, could impair our ability to attract, employ and retain qualified employees. Competition for talent drives up costs in the form of cash and stock-based compensation. In times of stock price volatility, as we have experienced in the past and may experience in the future, the retentive value of our stock-based compensation may decrease. Additionally, we are highly dependent on the services of our longstanding executive team. Failure to ensure effective succession planning, transfer of knowledge, and smooth transitions involving executives and key employees could hinder our strategic planning, execution, and long-term success. + +Our business is dependent upon the proper functioning of our business processes and information systems and modification or interruption of such systems may disrupt our business and internal controls. + +We rely upon internal processes and information systems to support key business functions, including our assessment of internal controls over financial reporting as required by Section 404 of the Sarbanes-Oxley Act. The efficient operation and scalability of these processes and systems is critical to support our growth. We continue to design and implement updated accounting functionality related to a new enterprise resource planning, or ERP, system. Any ERP system implementation may introduce problems, such as quality issues or programming errors, that could have an impact on our continued ability to successfully operate our business or to timely and accurately report our financial results. These changes may be costly and disruptive to our operations and could impose substantial demands on management time. Failure to implement new or updated controls, or difficulties encountered in their implementation, could harm our operating results or cause us to fail to meet our reporting obligations. + +Identification of material weaknesses in our internal controls, even if quickly remediated once disclosed, may cause investors to lose confidence in our financial statements and our stock price may decline. Remediation of any material weakness could require us to incur significant expenses, and if we fail to remediate any material weakness, our financial statements may be inaccurate, we may be required to restate our financial statements, our ability to report our financial results on a timely and accurate basis may be adversely affected, our access to the capital markets may be restricted, our stock price may decline, and we may be subject to sanctions or investigation by regulatory authorities. + +Our operating results have in the past fluctuated and may in the future fluctuate, and if our operating results are below the expectations of securities analysts or investors, our stock price could decline. + +Our operating results have in the past fluctuated and may continue to fluctuate due to a number of factors. Therefore, investors should not rely on our past results of operations as an indication of our future performance. Factors that could affect our results of operations include, but are not limited to: + +• our ability to adjust spending due to the multi-year development cycle for some of our products and services; +• our ability to comply with our contractual obligations to customers; +- our extended payment term arrangements with certain customers, the inability of some customers to make required payments, our ability to obtain credit insurance for customers with extended payment terms, and customer bad debt write-offs; +• our vendors' payment requirements; +• unanticipated costs associated with environmental liabilities; and +• changes in financial accounting standards or interpretations of existing standards. + +Any of these factors could prevent us from achieving our anticipated financial results. For example, we have granted and may continue to grant extended payment terms to some customers, particularly during macroeconomic downturns, which could impact our ability to collect payment. Our vendors have requested and may continue to ask for shorter payment terms, which may impact our cash flow generation. These arrangements reduce the cash we have available for general business operations. In addition, the pace of growth in our operating expenses and investments may lag our revenue growth, creating volatility or periods where profitability levels may not be sustainable. Failure to meet our expectations or the expectations of our investors or security analysts is likely to cause our stock price to decline, as it has in the past, or substantial price volatility. + +## Risks Related to Regulatory, Legal, Our Stock and Other Matters + +We are subject to complex laws, rules, regulations, and political and other actions, including restrictions on the export of our products, which may adversely impact our business. + +We are subject to laws and regulations domestically and worldwide, affecting our operations in areas including, but not limited to, IP ownership and infringement; taxes; import and export requirements and tariffs; anti-corruption, including the Foreign Corrupt Practices Act; business acquisitions; foreign exchange controls and cash repatriation restrictions; foreign ownership and investment; data privacy requirements; competition and antitrust; advertising; employment; product regulations; cybersecurity; environmental, health, and safety requirements; the responsible use of AI; sustainability; cryptocurrency; and consumer laws. Compliance with such requirements can be onerous and expensive, could impact our competitive position, and may negatively impact our business operations and ability to manufacture and ship our products. There can be no assurance that our employees, contractors, suppliers, customers or agents will not violate applicable laws or the policies, controls, and procedures that we have designed to help ensure compliance with such laws, and violations could result in fines, criminal sanctions against us, our officers, or our employees, prohibitions on the conduct of our business, and damage to our reputation. Changes to the laws, rules and regulations to which we are subject, or changes to their interpretation and enforcement, could lead to materially greater compliance and other costs, and/or further restrictions on our ability to manufacture and supply our products and operate our business. For example, we may face increased compliance costs as a result of changes or increases in antitrust legislation, regulation, administrative rule making, increased focus from regulators on cybersecurity vulnerabilities and risks. Our position in markets relating to AI has led to increased interest in our business from regulators worldwide, including the European Union, the United States, the United Kingdom, South Korea, Japan, and China. For example, the French Competition Authority collected information from us regarding our business and competition in the graphics card and CSP market as part of an ongoing inquiry into competition in those markets. We have also received, and continue to receive, broad requests for information from competition regulators in the European Union, the United States, the United Kingdom, China, and South Korea regarding our sales of GPUs and other NVIDIA products, our efforts to allocate supply, foundation models and our investments, partnerships and other agreements with companies developing foundation models, the markets in which we compete and our competition, our strategies, roadmaps, and efforts to develop, market, and sell hardware, software, and system solutions, and our agreements with customers, suppliers, and partners. We expect to receive additional requests for information in the future. Such requests have been and are likely to be expensive and burdensome and could negatively impact our business and our relationships with customers, suppliers, and partners. + +Governments and regulators are also considering, and in certain cases, have imposed restrictions on the hardware, software, and systems used to develop frontier foundation models and generative AI. For example, the EU AI Act became effective on August 1, 2024 and will be fully applicable after a two-year transitional period. The EU AI Act may impact our ability to train, deploy, or release AI models in the EU. Several states are considering enacting or have already enacted regulations concerning AI technologies, with new state laws that took effect on January 1, 2026, which may impact our ability to train, deploy, or release AI models, and increase our compliance costs. Restrictions under these and any other regulations, if implemented, could increase the costs and burdens to us and our customers, delay or halt deployment of new systems using our products, and reduce the number of new entrants and customers, negatively impacting our business and financial results. Revisions to laws or regulations or their interpretation and enforcement could also result in increased taxation, trade sanctions, the imposition of or increase to import duties or tariffs, restrictions and controls on imports or exports, or other retaliatory actions, which could have an adverse effect on our business plans or impact the timing of our shipments. Additionally, changes in the public perception of governments in the regions where we operate or plan to operate could negatively impact our business and results of operations. + +Government actions, including trade protection and national and economic security policies of U.S. and foreign government bodies, such as tariffs, import or export regulations, including deemed export restrictions and restrictions on the activities of U.S. persons, trade and economic sanctions, decrees, quotas or other trade barriers and restrictions could affect our ability to ship products, provide services to our customers and employees, do business without an export license with entities on the U.S. Department of Commerce's U.S. Entity List or other USG restricted parties lists (which is expected to change from time to time), and generally fulfill our contractual obligations and have a material adverse effect on our business. If we were ever found to have violated export control laws or sanctions of the U.S. or similar applicable non-U.S. laws, even if the violation occurred without our knowledge, we may be subject to various penalties available under the laws, any of which could have a material and adverse impact on our business, operating results and financial condition. + +For example, in response to the war in Ukraine, the United States and other jurisdictions imposed economic sanctions and export control measures which blocked the passage of our products, services and support into Russia, Belarus, and certain regions of Ukraine. In fiscal year 2023, we stopped direct sales to Russia and closed business operations in Russia. Concurrently, the war in Ukraine has impacted sales in EMEA and may continue to do so in the future. + +The increasing focus on the risks and strategic importance of AI technologies has resulted in regulatory restrictions that target products and services capable of enabling or facilitating AI and may in the future result in additional restrictions impacting some or all of our product and service offerings. + +Concerns regarding third-party use of AI for purposes contrary to local governmental interests, including concerns relating to the misuse of AI applications, models, and solutions, has resulted in and could in the future result in unilateral or multilateral restrictions on products that can be used for training, modifying, tuning, and deploying LLMs and other AI applications. Such restrictions have limited and could in the future limit the ability of downstream customers and users worldwide to acquire, deploy and use systems that include our products, software, and services, and negatively impact our business and financial results. + +Such restrictions could include additional unilateral or multilateral export controls on certain products or technology, including but not limited to AI technologies. As geopolitical tensions have increased, semiconductors associated with AI, including GPUs and related products, are increasingly the focus of export control restrictions proposed by stakeholders in the U.S. and its allies. The United States has imposed unilateral worldwide controls restricting GPUs and associated products, and it is likely that additional unilateral or multilateral controls will be adopted. Such controls have been and may again be very broad in scope and application, prohibit us from exporting our products to any or all customers in one or more markets, and could negatively impact our manufacturing, testing and warehousing locations and options, or could impose other conditions that limit our ability to serve demand abroad and could negatively and materially impact our business, revenue and financial results. Export controls and other restrictions targeting GPUs and semiconductors associated with AI, which have been imposed and are likely to be more restrictive, would further limit our ability to export our technology, products, or services, creating a competitive disadvantage for us and negatively impacting our business and financial results. Export controls targeting GPUs and semiconductors associated with AI have subjected and may in the future subject downstream users of our products to restrictions on the use, resale, repair, or transfer of our products, negatively impacting our business and financial results. Controls could negatively impact our cost and/or ability to provide services such as NVIDIA AI cloud services and could impact the cost and/or ability for our CSPs and customers to provide services to their end customers, even outside China. + +Export controls have and could in the future disrupt our supply chain and distribution channels, negatively impacting our ability to serve demand, including in markets outside China and for our non-data center products. The possibility of additional export controls has negatively impacted and may in the future negatively impact demand for our products, benefiting competitors that offer alternatives less likely to be restricted by further controls. Repeated changes in the export control rules are likely to impose compliance burdens on our business and our customers, negatively and materially impacting our business. + +Increasing use of economic sanctions and export controls has impacted and may in the future impact demand for our products or services, negatively impacting our business and financial results. Reduced demand due to export controls has and could in the future lead to excess inventory or cause us to incur related supply charges. Additional unilateral or multilateral controls are also likely to include deemed export control limitations that negatively impact the ability of our research and development teams to execute our roadmap or other objectives in a timely manner. Additional export restrictions may not only impact our ability to serve overseas markets, but also provoke responses from foreign governments, including China, that negatively impact our supply chain or our ability to provide our products and services to customers in all markets worldwide, which could also substantially reduce our revenue. Regulators in China have inquired about our sales and efforts to supply the China market and our fulfillment of the commitments we entered at the close of our Mellanox acquisition. On September 15, 2025, China's antitrust regulators published their preliminary finding that our compliance with applicable U.S. export controls, which required us to offer degraded products to the Chinese market, discriminated unfairly against customers in the China market and therefore violated the terms of China's approval of our Mellanox acquisition. If regulators conclude that we have failed to fulfill the terms of our Mellanox acquisition or we have violated any applicable law in China, we could be subject to financial penalties, restrictions on our ability to conduct our business, restrictions or other orders regarding our networking business, products, and services, or otherwise impact our operations in China, any of which could have a material and adverse impact on our business, operating results and financial condition. + +Over the past three years, we have been subject to a series of shifting and expanding export control restrictions, impacting our ability to serve customers outside the United States. + +In August 2022, the USG announced export restrictions and export licensing requirements targeting China's semiconductor and supercomputing industries. These restrictions impacted exports of certain chips, as well as software, hardware, equipment and technology used to develop, produce and manufacture certain chips to China (including Hong Kong and Macau) and Russia, and specifically impact our A100 and H100 integrated circuits, DGX or any other systems or boards which incorporate A100 or H100 integrated circuits. + +In July 2023, the USG also informed us of an additional licensing requirement for a subset of A100 and H100 products destined to certain customers and other regions, including some countries in the Middle East. + +In October 2023, the USG announced new and updated licensing requirements for exports to China and Country Groups D:1, D:4, and D:5 (including but not limited to, Saudi Arabia, the United Arab Emirates, and Vietnam, but excluding Israel) of our products exceeding certain performance thresholds, including, but not limited to, the A100, A800, H100, H800, L4, L40, L40S RTX 4090, GB200 NVL72, and B200. The licensing requirements also apply to the export of products exceeding certain performance thresholds to a party headquartered in, or with an ultimate parent headquartered in, Country Group D5, including China. + +In April 2025, the USG informed us that it requires a license for export to China (including Hong Kong and Macau) and D:5 countries, or to companies headquartered or with an ultimate parent therein, of our H2O integrated circuits and any other circuits achieving the H2O's memory bandwidth, interconnect bandwidth, or combination thereof. As a result of these requirements, we incurred a \$4.5 billion charge in the first quarter of fiscal year 2026 associated with H2O for excess inventory and purchase obligations, as the demand for H2O products diminished. + +In August 2025, the USG granted licenses that would allow us to ship certain H2O products to certain China-based customers. We generated approximately \$60 million in H2O revenue under those licenses. USG officials expressed an expectation that the USG will receive 15% or more of the revenue generated from licensed sales of our products, but the USG did not publish a regulation codifying such requirement. + +In February 2026, the USG granted a license that would allow us to ship small amounts of H200 products to specific China-based customers. To date, we have not generated any revenue under the H200 licensing program, and do not yet know whether any imports will be allowed into China. The license requires that the H200s go through an inspection process in the United States prior to any shipment to the customer. As a result, any H200 shipped under the new licensing program will be subject to a 25% tariff upon importation into the United States. + +In the event that we are able to sell licensed products into the China market, we may not be able to pass along all or any of the tariff to our customers, and may be subject to litigation, increased costs, and a harmed competitive position. + +The export controls applicable to China are complex and address a variety of parameters, including the total processing performance of a chip, the “performance density” of a chip, the interconnect bandwidth of a chip, and the memory bandwidth of a chip. Under the current rules and geopolitical landscape, we are unable to create and deliver a competitive product for China’s data center market that receives approval from both the USG and the Chinese government. As of the end of fiscal year 2026, we were effectively foreclosed from competing in China’s data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide. Unless we are able to return with a product that meets the approval of both the USG and the Chinese government, our lost opportunity and the benefit to our competitors will have a material and adverse impact on our business, operating results, and financial condition. + +In addition to controls targeting D:1, D:4 and D:5 countries, the USG has also imposed worldwide export controls impacting our products, and may impose additional controls in the future. + +In January 2025, the USG published the AI Diffusion IFR in the Federal Register. The IFR would have imposed a worldwide licensing requirement on our data center products, such as our H200, GB200 and GB300. The AI Diffusion IFR would have divided the world into three tiers, relegating most countries to “Tier 2” status, and would have created a complex and burdensome scheme for licensing approvals. + +In May 2025, the USG announced that it would rescind the AI Diffusion IFR and implement a replacement rule. The scope, timing, and requirements of the forthcoming rule remain uncertain. The replacement rule may impose new restrictions on our products or operations and/or add license requirements that could have a material impact on our business, operating results, and financial condition. For example, in October 2025, the Senate passed the “GAIN AI Act” in the NDAA. The GAIN AI Act would restrict the Trump Administration’s ability to adapt the Biden Administration’s export control rules, and could also allow private U.S. persons to review and overturn licensing and foreign policy decisions made by the Trump Administration. + +Our competitive position has been harmed by export controls, and our competitive position and future results will be further harmed, over the long term, if the restrictions remain in place or are expanded in geographic, customer, or product scope, if customers purchase product from competitors, if customers develop their own internal solution, if we are unable to provide contractual warranty or other extended service obligations, if the USG does not grant licenses in a timely manner or denies licenses to significant customers or if we incur significant transition costs. The licensing process may not be resolved before significant business opportunities evaporate. Even if the USG grants any requested licenses, the licenses have already and may in the future be temporary, impose burdensome conditions regarding the installation, maintenance, and use of such products, or include financial or economic requirements that we or our customers or end users cannot or choose not to fulfill. The licensing requirements have already and may in the future benefit certain of our competitors, as the licensing process will make our pre-sale and post-sale technical support efforts more cumbersome and less certain and encourage customers in China, the Middle East, and other regions to pursue alternatives to our products, including semiconductor suppliers based in China, Europe, and Israel. + +Given the increasing strategic importance of AI and rising geopolitical tensions, the USG has changed and may again change the export control rules at any time and further subject a wider range of our products to export restrictions and licensing requirements, negatively impacting our business and financial results. In the event of such change, we may be unable to sell our inventory of such products and may be unable to develop replacement products not subject to the licensing requirements. + +For example, the USG already imposed license conditions that limit the ability of foreign firms to create and offer as a service large-scale GPU clusters, such as imposing license conditions on the use of products to be exported to certain countries, and may impose additional conditions such as requiring chip tracking and throttling mechanisms that could disable or impair GPUs if certain events, including unauthorized system configuration, use, or location, are detected. Such government mandates in chip designs could introduce system vulnerabilities and expose us to significant risk and potential liability, negatively impact demand for our products, and could have a material impact on our business, operating results, and financial condition. Even if not enacted into binding legislation, draft bills have impacted and may in the future negatively impact our business. For example, following U.S. legislative proposals calling for mandatory features in our chips, China's government publicly questioned whether our H2O products have built-in vulnerabilities, discouraging customers from purchasing our products. We provided a public response explaining that our GPUs, including H2O, do not include such built-in vulnerabilities, and will respond to any follow-up questions we receive. + +Open-source foundation models are rapidly growing in popularity with developers worldwide. Any regulatory control or other restriction that limits our ability to provide products and services that support third-party applications and models, including applications built on foundation models originating in China such as DeepSeek, Qwen, or KIMMI, could have a material impact on our business, operating results, and financial condition. + +The USG already imposed export controls restricting certain gaming GPUs, and if the USG expands such controls to restrict additional gaming products, it may disrupt a significant portion of our supply and distribution chain and negatively impact sales of such products to markets outside China, including the U.S. and Europe. In addition, as the performance of the gaming GPUs increases over time, export controls may have a greater impact on our ability to compete in markets subject to those controls. Export controls may disrupt our supply and distribution chain for a substantial portion of our products, which are warehoused in and distributed from Hong Kong. + +Export controls restricting our ability to sell data center GPUs may also negatively impact demand for our networking products used in servers containing our GPUs. The USG may also impose export controls on our networking products, such as high-speed network interconnects, to limit the ability of downstream parties to create large clusters for frontier model training. + +Export controls have and are likely in the future to have a disproportionate impact on NVIDIA and may disadvantage us against certain of our competitors that sell chips that are outside the scope of such control. Export controls have already and may in the future encourage customers outside China and other impacted regions to “design-out” certain U.S. semiconductors from their products to reduce the compliance burden and risk, and to ensure that they are able to serve markets worldwide. Export controls have already encouraged and may in the future encourage overseas governments to request that our customers purchase from our competitors rather than NVIDIA or other U.S. firms, harming our business, market position, and financial results. + +As a result, export controls have in the past and may in the future negatively impact demand for our products and services not only in China, but also in other markets, such as Europe, Latin America, and Southeast Asia. Export controls increase the risk of investing in U.S. advanced semiconductor products, because by the time a new product is ready for market, it may be subject to new unilateral export controls restricting its sale, resulting in excess inventory and purchase obligations as we recently experienced with the H2O. At the same time, such controls may increase investment in foreign competitors, which would be less likely to be restricted by U.S. controls. + +The increasingly complex export controls impose complex and burdensome compliance obligations on our partners, suppliers, and customers. While we seek to strictly comply with all applicable export control regulators, reports of diversion of controlled products, even when unsubstantiated and untrue, may negatively impact our business, relationships with partners and customers, and our reputation. Incorrect allegations that our compliance efforts satisfy the letter but not the “spirit” of the applicable regulations, as well as incorrect allegations that legitimate and appropriate business is using supposed “loopholes” in the export controls may negatively impact our business, relationships with partners and customers, and our reputation. + +In addition to export controls, the USG may impose restrictions on the import and sale of products that incorporate technologies developed or manufactured in whole or in part in China. For example, the USG adopted “Connected Vehicle” restrictions on the import and sale of certain automotive products in the United States, which if adopted and interpreted broadly, could impact our ability to develop and supply solutions for our automotive customers. The USG is also considering restrictions that would limit our ability to support third-party applications and models built on open-source foundation models originating in China. Such restrictions, if implemented, would favor our foreign competitors and negatively impact our business. + +Additionally, restrictions imposed by the Chinese government on the duration of gaming activities and access to games may adversely affect our Gaming revenue, and even if we are able to participate in the China data center compute market, increased oversight of digital platform companies may adversely affect our Data Center revenue. The Chinese government has encouraged customers to purchase from our China-based competitors and discouraged customers from purchasing, importing, or using our data center products, including any China-specific product designed to comply with U.S. export controls. As another example, an agency of the Chinese government announced an Action Plan that endorses new standards regarding the compute performance per watt and per memory bandwidth of accelerators used in new and renovated data centers in China. Although we are already effectively foreclosed from the China market by U.S. export controls, if those controls changed to allow us to return to the market, the Chinese government could modify or implement the Action Plan in a way that effectively prevents us from being able to design products to meet the new standard, which may restrict the ability of customers to use some of our data center products and may have a material and adverse impact on our business, operating results and financial condition. Further restrictions on our products or the products of our suppliers could negatively impact our business and financial results. + +Finally, our business depends on our ability to receive consistent and reliable supply from our overseas partners, especially in Taiwan and South Korea. Any new restrictions that negatively impact our ability to receive supply of components, parts, or services from Taiwan and South Korea, would negatively impact our business and financial results. + +Scrutiny from shareholders, regulators and others regarding our corporate sustainability practices could result in additional costs or risks and adversely impact our reputation and willingness of customers and suppliers to do business with us. + +Certain shareholder advocacy groups, investment funds, shareholders and other market participants, customers and government regulators have focused on corporate sustainability practices and disclosures, including those associated with climate change and human rights. Stakeholders may not be satisfied with our corporate sustainability practices and goals or the speed of their adoption. Further, there are state-level initiatives in the U.S. that may differ from other regulatory requirements or our various stakeholders' expectations. Additionally, our corporate sustainability practices, oversight of our practices or disclosure controls may not meet evolving shareholder, regulator or other industry stakeholder expectations, or we may fail to meet corporate sustainability disclosure or reporting standards or legal requirements. We could also incur additional costs and require additional resources to monitor, report, and comply with various corporate sustainability practices and legal requirements, choose not to conduct business with potential customers and suppliers, or discontinue or not expand business with existing customers and suppliers due to our policies. These factors and increased disclosure may negatively harm our brand, reputation and business activities or expose us to liability. + +## Issues relating to the responsible use of our technologies, including AI in our offerings, may result in reputational or financial harm and liability. + +Concerns relating to the responsible use of new and evolving technologies, such as AI, in our products and services may result in reputational or financial harm and liability and may cause us to incur costs to resolve such issues. We are increasingly building AI capabilities and protections into many of our products and services, and we also offer stand-alone AI applications. AI poses emerging legal, social, and ethical issues and presents risks and challenges that could affect its adoption, and therefore our business. If we enable or offer solutions that draw controversy due to their perceived or actual impact on society, such as AI solutions that have unintended consequences, infringe copyright or rights of publicity, or are controversial because of their impact on human rights, privacy, employment or other social, economic or political issues, or if we are unable to implement effective internal policies and frameworks relating to the responsible development and use of AI models and systems offered through our sales channels, we may experience brand or reputational harm, competitive harm or legal liability. Leveraging AI capabilities to potentially improve our internal functions and operations may present further risks, costs, and challenges. Complying with multiple regulations from different jurisdictions related to AI may further increase our cost of doing business, may change the way that we operate in certain jurisdictions, and may impede our ability to offer certain products and services in certain jurisdictions if we are unable to comply with regulations. Compliance with existing and proposed government regulation of AI, including in jurisdictions such as the European Union, may further increase the cost of related research and development, and create additional reporting and/or transparency requirements. For example, regulation adopted in response to the European Union Code of Practice for General Purpose Artificial Intelligence could require us to notify the European Commission regarding details of some of our Trustworthy AI processes related to our risk framework. Furthermore, changes in AI-related regulation could disproportionately impact and disadvantage us and require us to change our business practices, which may negatively impact our financial results. Our failure to adequately address concerns and regulations relating to the responsible use of AI by us or others could undermine public confidence in AI and slow adoption of AI in our products and services or cause reputational or financial harm. + +## Actions to adequately protect our IP rights could result in substantial costs to us and our ability to compete could be harmed if we are unsuccessful or if we are prohibited from making or selling our products. + +From time to time, we are involved in lawsuits or other legal proceedings alleging patent infringement or other IP rights violations by us, our employees or parties that we have agreed to indemnify. An unfavorable ruling could include significant damages, invalidation of one or more patents, indemnification of third parties, payment of lost profits, or injunctive relief. Claims that our products or processes infringe the IP rights of others, regardless of their merit, could cause us to incur significant costs to respond to, defend, and resolve such claims, and they may also divert the efforts and attention of management and technical personnel. + +We may commence legal proceedings to protect our IP rights, which may increase our operating expenses. We could be subject to countersuits as a result. If infringement claims are made against us or our products are found to infringe a third party's IP, we or one of our indemnitees may have to seek a license to the third party's IP rights. If we or one of our indemnitees is unable to obtain such a license on acceptable terms or at all, we could be subject to substantial liabilities or have to suspend or discontinue the manufacture and sale of one or more of our products. We may also have to make royalty or other payments or cross license our technology. If these arrangements are not concluded on commercially reasonable terms, our business could be negatively impacted. Furthermore, the indemnification of a customer or other indemnitee may increase our operating expenses and negatively impact our operating results. + +We rely on patents, trademarks, trade secrets, employee and third-party nondisclosure agreements, licensing arrangements and the laws of the countries in which we operate to protect our IP. Foreign laws may not protect our products or IP rights to the same extent as United States law. This makes the possibility of piracy of our technology and products more likely. The theft or unauthorized use or publication of our trade secrets and other confidential information could harm our competitive position and reduce acceptance of our products; as a result, the value of our investment in research and development, product development and marketing could be reduced. We also may face risks to our IP if our employees are hired by competitors. We continuously assess whether and where to seek formal protection for existing and new innovations and technologies but cannot be certain whether our applications for such protections will be approved, and, if approved, whether they will be enforceable. + +We are subject to stringent and changing data privacy and security laws, rules, regulations and other obligations. These areas could damage our reputation, deter current and potential customers, affect our product design, or result in legal or regulatory proceedings and liability. + +We process sensitive, confidential or personal data or information that is subject to privacy and security laws, regulations, industry standards, external and internal policies, contracts and other obligations that govern the processing of such data by us and on our behalf. Concerns about our practices or the ultimate use of our products and services with regard to the collection, use, retention, security or disclosure of personal information or other privacy-related matters, including for use in AI, even if unfounded, could damage our reputation and adversely affect our operating results. The theft, loss or misuse of personal data in our possession or by one of our partners could result in damage to our reputation, regulatory proceedings, disruption of our business activities or increased security or remediation costs and costs related to defending legal claims. + +In the United States, federal, state and local authorities have enacted numerous data privacy and security laws, including for data breach notification, personal data privacy and consumer protection. Numerous U.S. states have enacted comprehensive privacy laws that impose certain obligations on covered businesses, including providing specific disclosures in privacy notices and affording residents with certain rights concerning their personal data. As applicable, such rights may include the right to access, correct, or delete certain personal data, and to opt-out of certain data processing activities, such as targeted advertising, profiling and automated decision-making. The exercise of these rights may impact our business and ability to provide our products and services. Certain states also impose stricter requirements for processing certain personal data, including sensitive information, such as conducting data privacy impact assessments. These state laws allow for statutory fines for noncompliance. For example, the California Consumer Privacy Act of 2018, as amended by the California Privacy Rights Act of 2020, or CPRA, or collectively the CCPA, gives California residents the right to access, delete and opt-out of certain sharing of their personal information, and to receive detailed information about how it is used and shared. The CCPA provides for substantial fines for intentional violation and the law created a private right of action for certain data breaches. Similar laws are being considered in several other states, as well as at the federal and local levels. Additionally, several states and localities have enacted measures related to the use of AI and machine learning in products and services. If we become subject to additional data privacy laws, the risk of enforcement action against us could increase. + +Worldwide regulatory authorities are also considering and have approved various legislative proposals concerning data protection. The European Union adopted the General Data Protection Regulation, or GDPR, and the United Kingdom similarly adopted the U.K. GDPR, governing the strict handling of personal data of persons within the European Economic Area, or EEA, and the United Kingdom, respectively, including its use and protection and the ability of persons whose data is stored to access, correct, and delete such data about themselves. If we are found not to comply, we could be subject to penalties of up to €20 million or 4% of worldwide revenue, whichever is greater, and classes of individuals or consumer protection organizations may initiate litigation related to our processing of their personal data. Furthermore, the EU AI Act and similar legislation could impose onerous obligations that may disproportionately impact and disadvantage us and require us to change our business practices. Additionally, Europe’s Network and Information Security Directive, or NIS2, regulates resilience and incident response capabilities of entities operating in a number of sectors, including the digital infrastructure sector. Non-compliance with NIS2 may lead to administrative fines of a maximum of 10 million Euros or up to 2% of the total worldwide revenue of the preceding fiscal year. + +In the ordinary course of business, we transfer personal data from Europe, China, and other jurisdictions to the United States or other countries. Certain jurisdictions have enacted data localization laws and cross-border personal data transfer laws. For example, the GDPR governs the transfer of personal data to countries outside of the EEA. The European Commission released a set of “Standard Contractual Clauses” designed for entities to validly transfer personal data out of the EEA to jurisdictions that the European Commission has not found to provide an adequate level of protection, including the United States. Additionally, the U.K.’s International Data Transfer Agreement / Addendum, as well as the EU-U.S. Data Privacy Framework and the U.K. extension thereto (which allows for transfers to relevant U.S.-based organizations who self-certify compliance and participate in the Framework) are mechanisms that may be used to transfer personal data from the EEA and U.K. to the United States. However, these mechanisms are subject to legal challenges, and there is no assurance that we can satisfy or rely on these measures to lawfully transfer personal data to the United States. Other jurisdictions have enacted or are considering similar cross-border personal data transfer laws and local personal data residency laws, any of which would increase the cost and complexity of doing business and could result in fines from regulators. For example, China’s law imposes various requirements relating to data processing and data localization. Data broadly defined as important under China’s law, including personal data, may not be transferable outside of China without prior assessment and approval by the Cyberspace Administration of China, or CAC. Compliance with these requirements, including CAC assessments and any deemed failures of such assessments, could cause us to incur liability, prevent us from using data collected in China or impact our ability to transfer data outside of China. The inability to import personal data to the United States could significantly and negatively impact our business operations, limit our ability to collaborate with parties that are subject to European, China and other data privacy and security laws, or require us to increase our personal data processing capabilities in Europe and/or elsewhere at significant expense. Some European regulators have prevented companies from transferring personal data out of Europe for allegedly violating the GDPR’s cross-border data transfer limitations, which could negatively impact our business. + +We are also bound by certain contractual obligations related to data privacy and security, and our efforts to comply with such obligations may not be successful or may be claimed to be non-compliant. For example, certain privacy laws, such as the GDPR and the CCPA, require our customers to impose specific contractual restrictions on their service providers. We sometimes host personal data in collaboration with our customers, and if a breach exposed or altered that personal data, it could harm those customer relationships and subject us to litigation, regulatory action, or fines. We publish privacy policies, marketing materials and other statements, such as compliance with certain certifications or self-regulatory principles, regarding data privacy and security. Regulators in the U.S. are increasingly scrutinizing these statements, and if these policies, materials or statements are found to be deficient, lacking in transparency, deceptive, unfair or misrepresentative of our practices, we may be subject to investigation, enforcement actions by regulators or other adverse consequences. + +Data protection laws around the world are quickly changing and may be interpreted and applied in an increasingly stringent fashion and in a manner that is inconsistent with our data practices. These obligations may affect our product design and necessitate changes to our information technologies, systems and practices and to those of any third parties that process personal data on our behalf. Despite our efforts, we or third parties we rely upon may fail to comply with such obligations. If we fail, or are perceived to have failed, to address or comply with data privacy and security obligations, we could face significant consequences, including but not limited to, government enforcement actions, litigation, additional reporting requirements and/or oversight, bans on processing personal data, and orders to destroy or not use personal data. Any of these events could have a material adverse effect on our reputation, business, or financial condition. + +## We may have exposure to additional tax liabilities and our operating results may be adversely impacted by changes in tax laws, higher than expected tax rates and other tax-related factors. + +We are subject to complex income tax laws and regulations, as well as non-income-based taxes, in various jurisdictions. Significant judgment is required in determining our worldwide provision for income taxes and other tax liabilities. We are regularly under audit by tax authorities in different jurisdictions. Although we believe our tax estimates are reasonable, any adverse outcome could increase our worldwide effective tax rate, increase the amount of non-income taxes imposed on our business, and harm our financial position, results of operations, net income, and cash flows. + +Further, changes in tax laws or their interpretation by tax authorities in the U.S. or foreign jurisdictions could increase our future tax liability or cause other adverse tax impacts, which may materially impact our results of operations, or the way we conduct our business. Most of our income is taxable in the U.S., with a significant portion qualifying for preferential treatment as foreign-derived deduction eligible income, or FDDEI. If U.S. tax rates increase or the FDDEI deduction is reduced, our provision for income taxes, results of operations, net income, and cash flows would be adversely affected. In addition, our tax obligations and effective tax rate in the jurisdictions in which we conduct business could increase as a result of international tax developments, including the implementation of the Two-Pillar framework led by the Organization for Economic Cooperation and Development, or OECD, which involves the reallocation of taxing rights in respect of certain multinational enterprises above a fixed profit margin to the jurisdictions in which they carry on business (referred to as Pillar One), and imposes a minimum effective corporate tax rate (referred to as Pillar Two). A number of countries in which we conduct business have enacted, or are in the process of enacting, elements of the Pillar Two rules, including the recently released "side-by-side" framework, which provides certain additional safe harbors and coordination mechanisms. Any such tax laws, or changes in any such tax laws, including those related to the side-by-side framework, may increase tax uncertainty and compliance costs and adversely affect our provision for income taxes, cash tax payments, results of operations, and financial condition. + +Our future effective tax rate may also be affected by a variety of factors, including changes in our business or statutory rates, the mix of earnings in countries with differing statutory tax rates, available tax incentives, credits and deductions, the expiration of statutes of limitations, changes in accounting principles, adjustments to income taxes upon finalization of tax returns, increases in expenses not deductible for tax purposes, the estimates of our deferred tax assets and liabilities and deferred tax asset valuation allowances, changing interpretation of existing laws or regulations, the impact of accounting for business combinations, as well as changes in the domestic or international organization of our business and structure. Furthermore, the tax effects of accounting for stock-based compensation and volatility in our stock price may significantly impact our effective tax rate in the period in which they occur. A decline in our stock price may result in reduced future tax benefits from stock-based compensation, increase our effective tax rate, and adversely affect our financial results. + +## Our business is exposed to the burden and risks associated with litigation, investigations and regulatory proceedings. + +We currently and will likely continue to face legal, administrative and regulatory proceedings, claims, demands and/or investigations involving shareholder, consumer, competition, intellectual property and/or other issues relating to our business. For example, we are defending a securities class action lawsuit from multiple shareholders asserting claims that we and certain of our officers made false and/or misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand in 2017 and 2018. Litigation and regulatory proceedings are inherently uncertain, and adverse rulings could occur, including monetary damages or fines, or an injunction stopping us from manufacturing or selling certain products, engaging in certain business practices, or requiring other remedies, such as compulsory licensing of patents. An unfavorable outcome or settlement may result in a material adverse impact. Regardless of the outcome, litigation can be costly, time-consuming, and disruptive to our operations. + +Delaware law and our certificate of incorporation, bylaws and agreement with Microsoft could delay or prevent a change in control. + +The anti-takeover provisions of the Delaware General Corporation Law may discourage, delay, or prevent a change in control. Provisions in our certificate of incorporation and bylaws could make it more difficult for a third party to acquire a majority of our outstanding stock. These provisions include the ability of our Board of Directors to create and issue preferred stock, change the number of directors, and to make, amend or repeal our bylaws without prior shareholder approval; the inability of our shareholders to act by written consent; advance notice requirements for director nominations and shareholder proposals; and super-majority voting requirement to amend some provisions in our certificate of incorporation and bylaws. Under our agreement with Microsoft for the Xbox, if someone makes an offer to purchase at least 30% of our outstanding common stock, Microsoft may have first and last rights of refusal to purchase the stock. These provisions could delay or prevent a change in control of NVIDIA, discourage proxy contests, and make it more difficult for shareholders to elect directors of their choosing and to cause us to take other corporate actions they desire. + +## Item 1B. Unresolved Staff Comments + +None. + +## Item 1C. Cybersecurity + +## Risk management and strategy + +We have in place certain infrastructure, systems, policies, and procedures that are designed to proactively prevent or reduce the impact of, and reactively address circumstances that arise when, events such as a cybersecurity incident occur. These include processes for assessing, identifying, and managing material risks from cybersecurity threats. Our information security management programs generally follow certain processes outlined in frameworks such as the ISO 27001 international standard for information security management and we evaluate and evolve our security measures as appropriate. We consult with external parties, such as cybersecurity firms and risk management and governance experts, on risk management and strategy. + +Identifying, assessing, and managing cybersecurity risk is integrated into our overall risk management systems and processes, and we have in place cybersecurity and data privacy training and policies designed to (a) respond to new requirements in global privacy and cybersecurity laws and (b) prevent, detect, respond to, mitigate and recover from identified and significant cybersecurity threats. + +We also have a vendor risk assessment process consisting of, depending on the nature and sensitivity of the supplier and data they process on our behalf, the distribution and review of supplier questionnaires designed to help us evaluate cybersecurity risks that we may encounter when working with third parties that have access to confidential and other sensitive company information. We take steps to review that such vendors have implemented data privacy and security controls that help mitigate the cybersecurity risks associated with these vendors, depending on the nature and sensitivity of the supplier and data they process on our behalf. We routinely assess our high-risk suppliers' conformance to industry standards (e.g., ISO 27001, ISO 28001, and C-TPAT), and we evaluate them for additional information, product, and physical security requirements. + +Refer to “Item 1A. Risk factors” in this annual report on Form 10-K for additional information about cybersecurity-related risks. + +## Governance + +Information security matters, including managing and assessing risks from cybersecurity threats, remain under the oversight of the Company's Board of Directors, or the Board. The Audit Committee of the Board, or the Audit Committee, also reviews the adequacy and effectiveness of the Company's information security policies and practices and the internal controls regarding information security risks. The Audit Committee receives regular information security updates from management, including our Chief Security Officer and members of our security team. The Board also receives annual reports on information security matters from our Chief Security Officer and members of our security team. + +Our security efforts are managed by a team of executive cybersecurity, IT, engineering, operations, and legal professionals. We have established a cross-functional leadership team, consisting of executive-level leaders, that meets regularly to review cybersecurity matters and evaluate emerging threats. With oversight and guidance provided by the cross-functional leadership team, our information security teams refine our practices to address emerging security risks and changes in regulations. Our executive-level leadership team also participates in cybersecurity incident response efforts by engaging with the incident response team and helping direct the company's response to and assessment of certain cybersecurity incidents. + +We have designated a Chief Security Officer, reporting to our Senior Vice President of Software Engineering, to oversee the identification, assessment, and management of material cybersecurity risks. Our Chief Security Officer's cybersecurity expertise includes over 18 years of combined government and private sector assignments. + +## Item 2. Properties + +Our headquarters is in Santa Clara, California. We own and lease approximately 3 million square feet of office and building space for our corporate headquarters. In addition, we lease data center space in Santa Clara, California. We also own and lease facilities for data centers, research and development, and/or sales and administrative purposes throughout the U.S. and in various international locations, primarily in China, India, Israel, and Taiwan. We believe our existing facilities, both owned and leased, are in good condition and suitable for the conduct of our business. We do not identify or allocate assets by operating segment. For additional information regarding obligations under leases, refer to Note 17 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K, which information is hereby incorporated by reference. + +## Item 3. Legal Proceedings + +Please see Note 12 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for a discussion of our legal proceedings. + +## Item 4. Mine Safety Disclosures + +Not applicable. + +## Part II + +## Item 5. Market for Registrant's Common Equity, Related Stockholder Matters and Issuer Purchases of Equity Securities + +Our common stock is traded on the Nasdaq Global Select Market under the symbol NVDA. Public trading of our common stock began on January 22, 1999. Prior to that, there was no public market for our common stock. As of February 20, 2026, we had approximately 1,226 registered shareholders, not including those shares held in street or nominee name. + +In May 2024, we announced a ten-for-one stock split, or the Stock Split, of our issued common stock, which was effected through the filing of an amendment to the Company's Restated Certificate of Incorporation, or the Amendment, with the Secretary of the State of Delaware. In June 2024, the Company filed the Amendment to effect the Stock Split and proportionately increased the number of shares of the Company's authorized common stock from 8.0 billion to 80.0 billion. Shareholders of record at the close of market on June 6, 2024 received nine additional shares of common stock, distributed after the close of market on June 7, 2024. All share, equity award and per share amounts presented herein have been retrospectively adjusted to reflect the Stock Split. + +## Issuer Purchases of Equity Securities + +On August 26, 2025, our Board of Directors approved an additional \$60.0 billion in share repurchase authorization, without expiration. In fiscal year 2026, we repurchased 282 million shares of our common stock for \$40.4 billion. As of January 25, 2026, we were authorized, subject to certain specifications, to repurchase up to \$58.5 billion of our common stock. + +The repurchases can be made in the open market, in privately negotiated transactions, pursuant to a Rule 10b5-1 trading plan or in structured share repurchase agreements in compliance with Rule 10b-18 of the Exchange Act, subject to market conditions, applicable legal requirements, and other factors. Our share repurchase program may be suspended at any time at our discretion. + +In fiscal year 2026, we paid cash dividends to our shareholders of \$974 million. The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders. + +The following table presents details of our share repurchase transactions during the fourth quarter of fiscal year 2026: + +
PeriodTotal Number of Shares Purchased (In millions)Average Price Paid per Share (1)Total Number of Shares Purchased as Part of Publicly Announced Program (In millions)Approximate Dollar Value of Shares that May Yet Be Purchased Under the Program (In billions)
October 27, 2025 - November 23, 20255.6$ 198.895.6$ 61.1
November 24, 2025 - December 21, 20256.6$ 179.416.6$ 59.9
December 22, 2025 - January 25, 20267.3$ 186.527.3$ 58.5
Total19.519.5
+ +(1) Average price paid per share includes broker commissions, but excludes our liability under the 1% excise tax on the net amount of our share repurchases required by the Inflation Reduction Act of 2022. + +From January 26, 2026 through February 20, 2026, we repurchased 8 million shares for \$1.5 billion pursuant to a pre-established trading plan. + +## Restricted Stock Unit Share Withholding + +We withhold shares of our common stock associated with net share settlements to cover tax withholding obligations of awards under our employee equity incentive program. During fiscal year 2026, we withheld approximately 51 million shares for a total value of \$7.9 billion through net share settlements. Refer to Note 3 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for further discussion regarding our equity incentive plans. + +## Recent Sales of Unregistered Securities and Use of Proceeds + +On December 15, 2025, we acquired a company and issued to a key employee a total of 174,676 shares of our common stock, valued at approximately \$31 million based on our closing stock price on the issuance date. + +The above securities were issued in a transaction not involving a public offering pursuant to an exemption from registration set forth in Section 4(a)(2) of the Securities Act (and Regulation D or Regulation S promulgated thereunder). + +## Stock Performance Graphs + +The following graph compares the cumulative total shareholder return for our common stock, the S&P 500 Index, and the Nasdaq 100 Index for the five years ended January 25, 2026. The graph assumes that \$100 was invested on January 31, 2021 in our common stock and in each of the S&P 500 Index and the Nasdaq 100 Index. Our common stock is a component of each of the presented indices. Total return assumes reinvestment of dividends in each of the indices indicated. Total return is based on historical results and is not intended to indicate future performance. + +Comparison of 5 Year Cumulative Total Return\* Among NVIDIA Corporation, the S&P 500 Index, and the Nasdaq 100 Index +![](images/70802190a84f971e77f9c815eb7ce0e35b1600f9ff0424b0eca22a46bcf9097b.jpg) + +
+line chart + +| Date | NVIDIA Corporation | S&P 500 | Nasdaq 100 | +| ---------- | ------------------ | ------- | ---------- | +| 1/31/2021 | 100 | 80 | 100 | +| 1/30/2022 | 180 | 100 | 120 | +| 1/29/2023 | 160 | 90 | 90 | +| 1/28/2024 | 470 | 120 | 150 | +| 1/26/2025 | 1100 | 170 | 180 | +| 1/25/2026 | 1450 | 200 | 210 | +
+ +\*\$100 invested on 1/31/2021 in stock and in indices, including reinvestment of dividends. + +Source: FactSet financial data and analytics. + +
1/31/20211/30/20221/29/20231/28/20241/26/20251/25/2026
NVIDIA Corporation$ 100.00$ 175.98$ 157.05$ 470.88$ 1,100.68$ 1,448.75
S&P 500$ 100.00$ 121.00$ 112.98$ 137.98$ 174.50$ 200.33
Nasdaq 100$ 100.00$ 112.60$ 95.60$ 138.07$ 173.95$ 206.01
+ +Item 6. [Reserved] + +## Item 7. Management's Discussion and Analysis of Financial Condition and Results of Operations + +The following discussion and analysis of our financial condition and results of operations should be read in conjunction with “Item 1A. Risk Factors,” our Consolidated Financial Statements and related Notes thereto, as well as other cautionary statements and risks described elsewhere in this Annual Report on Form 10-K, before deciding to purchase, hold, or sell shares of our common stock. + +## Overview + +## Our Company and Our Businesses + +NVIDIA pioneered accelerated computing to help solve the most challenging computational problems. Since our original focus on PC graphics, we have expanded to several other large and important computationally intensive fields. Fueled by the sustained demand for exceptional 3D graphics and the scale of the gaming market, NVIDIA has leveraged its GPU architecture to create platforms for scientific computing, AI, data science, autonomous vehicles, robotics, and digital twin applications. NVIDIA is now a data center scale AI infrastructure company reshaping all industries. + +Our two operating segments are "Compute & Networking" and "Graphics." Refer to Note 16 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information. + +Headquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998. + +## Recent Developments, Future Objectives and Challenges + +Revenue growth in fiscal year 2026 was driven by data center compute and networking platforms for accelerated computing and AI solutions. Our Blackwell architectures represented the majority of our Data Center revenue. + +The availability of data centers, energy, and capital to support the buildout of NVIDIA AI infrastructure by our customers and partners is crucial, and any shortage of these or other necessary resources could impact our future revenue and financial performance. Expanding energy capacity to meet demand is a complex, multi-year process that involves significant regulatory, technical, and construction challenges. In addition, access to capital can be particularly constrained for less-capitalized companies, which may face difficulties securing financing for large-scale infrastructure projects. These limitations could delay customer and partner deployments or reduce the scale of accelerated computing and AI adoption. + +We continue to execute Data Center compute product introductions, bringing new advanced architectures on a one-year product cadence, including our Rubin platform. We began shipping production units of our new Blackwell Ultra platforms including GB300 in the second quarter of fiscal year 2026. The complexity of our product transitions and sophisticated system configurations has and may in the future cause delays in production and create challenges in managing supply and demand. This could further result in revenue volatility, quality issues, increased inventory provisions, decreases in product yields and higher material costs, and/or increased warranty costs. Customers may postpone purchasing new architectures or may adopt new technologies more gradually than anticipated, affecting our revenue timing and supply chain expenses. + +In April 2025, the USG informed us that a license is required for exports of our H2O product into the China market. As a result of these requirements, we incurred a \$4.5 billion charge in the first quarter of fiscal year 2026 associated with H2O for excess inventory and purchase obligations, as the demand for H2O diminished. In August 2025, the USG granted licenses that would allow us to ship certain H2O products to certain China-based customers. We generated approximately \$60 million in H2O revenue under those licenses. + +In February 2026, the USG granted a license that would allow us to ship small amounts of H200 products to specific China-based customers. To date, we have not generated any revenue under the H200 licensing program, and do not yet know whether any imports will be allowed into China. The license requires that the H200s go through an inspection process in the United States prior to any shipment to the customer. As a result, any H200 shipped under the new licensing program will be subject to a 25% tariff upon importation into the United States. + +The recent rise in high-quality open-source foundation models is making advanced AI capabilities broadly accessible. Open-source AI is dependent on developer adoption and if deployed on our competitors' platforms, it could reduce demand for our products and services. + +While currently our supply chain is mainly concentrated in Asia, we are expanding into the U.S. and Latin America. These moves are expected to strengthen our supply chain, add resiliency and redundancy, and meet the growing demand for AI infrastructure. Our ability to increase manufacturing capabilities will depend on the local region's manufacturing ecosystem's capacity to ramp production supply to the required volume and on a timely basis. + +We have made, and expect to continue making, investments that support our technology roadmap and the broader AI ecosystem. In fiscal year 2026, we made the following investments: + +- We invested \$17.5 billion in private companies and infrastructure funds, primarily to support early-stage startups. These investments include AI model makers that purchase our products directly or through CSPs. Many of these investments are illiquid and non-marketable. The related early-stage startups may not become profitable in the near term, or at all, and there can be no assurance that we will realize a return on our investments. +• We made investments in publicly-held equity securities where the value may fluctuate significantly due to changes in stock prices and could adversely affect our financial results. +- To support the build-out of complex datacenter infrastructures, we enter into commercial arrangements, including guarantees with partners. We provided \$3.5 billion in land, power, and shell guarantees to early-stage companies, generally over multi-year periods. If the escrow and the partners' operating activities are not sufficient to cover an event of default under these guarantees, we may elect to assume the underlying leases for internal use or sublease them to third parties. + +Macroeconomic factors, including tariffs, inflation, interest rate changes, capital market volatility, global supply chain constraints, and global economic and geopolitical developments, have direct and indirect impacts on our results of operations, particularly demand for our products. While difficult to isolate and quantify, these macroeconomic factors impact our supply chain and manufacturing costs, employee wages, costs for capital equipment, the value of our investments, revenue and competitive position. Our product and solution pricing generally does not fluctuate with short-term changes in our costs. Within our supply chain, we continuously manage product availability and costs with our vendors. + +Refer to “Item 1A. Risk Factors – Risks Related to Regulatory, Legal, Our Stock and Other Matters” for a further discussion of the potential impact of these factors on our business. + +Fiscal Year 2026 Summary + +
Year Ended
Jan 25, 2026Jan 26, 2025Change
($ in millions, except per share data)
Revenue$ 215,938$ 130,497Up 65%
Gross margin71.1 %75.0 %-3.9 pts
Operating expenses$ 23,076$ 16,405Up 41%
Operating income$ 130,387$ 81,453Up 60%
Net income$ 120,067$ 72,880Up 65%
Net income per diluted share$ 4.90$ 2.94Up 67%
+ +Revenue for fiscal year 2026 was \$215.9 billion, up 65% from a year ago. + +Data Center revenue for fiscal year 2026 was up 68% from a year ago. The strong year-on-year growth was driven by the major platform shifts – accelerated computing and AI. + +Gaming revenue for fiscal year 2026 was up 41% from a year ago, driven by strong Blackwell demand. We expect supply constraints to be a headwind to Gaming in the first quarter of fiscal 2027 and beyond. + +Professional Visualization revenue for fiscal year 2026 was up 70% from a year ago, driven by exceptional demand for Blackwell as well as the launch of our new DGX Spark. + +Automotive revenue for fiscal year 2026 was up 39% from a year ago, driven by continued adoption of our self-driving platforms. + +Gross margin decreased in fiscal year 2026 as our business model transitioned from offering Hopper HGX systems to Blackwell full-scale datacenter solutions. The gross margin decrease was also impacted by a \$4.5 billion charge associated with H2O excess inventory and purchase obligations. + +Operating expenses for fiscal year 2026 were up 41% from a year ago, driven by higher compensation and benefits expenses due to employee growth and compute and infrastructure costs. + +## Critical Accounting Estimates + +Our consolidated financial statements are prepared in accordance with accounting principles generally accepted in the United States, or U.S. GAAP. The preparation of these financial statements requires us to make estimates and judgments that affect the reported amounts of assets, liabilities, revenue, cost of revenue, expenses and related disclosure of contingencies. Critical accounting estimates are those estimates that involve a significant level of estimation uncertainty and could have a material impact on our financial condition or results of operations. We have critical accounting estimates in the areas of inventories, income taxes, non-marketable equity securities, and revenue recognition. Refer to Note 1 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for a summary of significant accounting policies. + +## Inventories + +We charge cost of sales for inventory provisions to write-down our inventory to the lower of cost or net realizable value or for obsolete or excess inventory, and for excess product purchase commitments. Most of our inventory provisions relate to excess quantities of products or components, based on our inventory levels and future product purchase commitments compared to assumptions about future demand and market conditions, which requires management judgment. + +Situations that may result in excess or obsolete inventory or excess product purchase commitments include changes in business and economic conditions, changes in market conditions, sudden and significant decreases in demand for our products, including potential cancellation or deferral of customer purchase orders, inventory obsolescence because of changing technology and customer requirements, new product introductions resulting in less demand for existing products or inconsistent spikes in demand, failure to estimate customer demand properly, ordering in advance of historical lead-times, government regulations and the impact of changes in future demand, or increase in demand for competitive products, including competitive actions. + +The net effect on our gross margin from inventory provisions and sales of items previously written down was an unfavorable impact of 2.6% in fiscal year 2026 and 2.3% in fiscal year 2025. Our inventory and capacity purchase commitments are based on forecasts of future customer demand and consider our third-party manufacturers' lead times and constraints. Our manufacturing lead times can be and have been long, and in some cases, extended beyond twelve months for some products. We may place non-cancellable inventory orders for certain product components in advance of our historical lead times, pay premiums and provide deposits to secure future supply and capacity. We also adjust to other market factors, such as product offerings and pricing actions by our competitors, new product transitions, and macroeconomic conditions - all of which may impact demand for our products. + +Refer to the Gross Profit and Gross Margin discussion below in this Management's Discussion and Analysis for further discussion. + +## Income Taxes + +We are subject to income taxes in the U.S. and foreign jurisdictions. Our calculation of deferred tax assets and liabilities is based on certain estimates and judgments and involves dealing with uncertainties in the application of complex tax laws. Our estimates of deferred tax assets and liabilities may change based, in part, on added certainty or finality to an anticipated outcome, changes in accounting standards or tax laws in the U.S. or foreign jurisdictions where we operate, or changes in other facts or circumstances. In addition, we recognize liabilities for potential U.S. and foreign income tax contingencies based on our estimate of whether, and the extent to which, additional taxes may be due. If we determine that payment of these amounts is unnecessary or if the recorded tax liability is less than our current assessment, we may be required to recognize an income tax benefit or additional income tax expense in our financial statements accordingly. + +We record a valuation allowance to reduce deferred tax assets to the amount that is believed more likely than not to be realized based on all available evidence. To the extent realization of the deferred tax assets becomes more-likely-than-not, we would recognize such deferred tax assets as income tax benefits during the period. + +We recognize the benefit from a tax position only if it is more-likely-than-not that the position would be sustained upon audit based solely on the technical merits of the tax position. Our policy is to include interest and penalties related to unrecognized tax benefits as a component of income tax expense. + +## Non-Marketable Equity Securities + +Non-marketable equity securities consist of investments in private companies without readily determinable fair values. They are measured at cost minus impairment, if any, and are adjusted for observable price changes in orderly transactions for a similar investment in the same issuer (the measurement alternative). These adjustments may require use of unobservable inputs. + +We assess impairment quarterly based on qualitative and quantitative factors, including the investee's operating performance and market trends. + +## Revenue Recognition + +## Revenue Allowances + +For products sold with a right of return, we record a reduction to revenue by establishing a sales return allowance for estimated product returns at the time revenue is recognized, based primarily on historical return rates. However, if product returns for a fiscal period are anticipated to exceed historical return rates, we may determine that additional sales return allowances are required to reflect our estimated exposure for product returns. Return rights for certain stocking distributors for specific products are contractually limited based on a percentage of prior quarter shipments. For shipments to other customers, we do not allow returns, although we may approve returns for credit or refund based on applicable facts and circumstances. + +We account for customer programs, which involve rebates and marketing development funds, or MDFs, as a reduction in revenue and accrue for such programs based on the amount we expect to be claimed by customers. Certain customer programs include distributor price incentives or other channel programs for specific products and customer classes which require judgment as to whether the applicable incentives will be attained. Estimates for customer program accruals include a combination of historical attainment and claim rates and may be adjusted based on relevant internal and external factors. + +## Contracts with Multiple Performance Obligations + +Our contracts may contain more than one performance obligation. Judgment is required in determining whether each performance obligation within a customer contract is distinct. Except for License and Development Arrangements, NVIDIA products and services function on a standalone basis and do not require a significant amount of integration or interdependency. Therefore, multiple performance obligations contained within a customer contract are considered distinct and are not combined for revenue recognition purposes. + +We allocate the total transaction price to each distinct performance obligation in an arrangement with multiple performance obligations on a relative standalone selling price basis. In most cases, we can establish standalone selling price based on directly observable prices of products or services sold separately in comparable circumstances to similar customers. If standalone selling price is not directly observable, such as when we do not sell a product or service separately, we determine standalone selling price based on market data and other observable inputs. + +## Results of Operations + +A discussion regarding our financial condition and results of operations for fiscal year 2026 compared to fiscal year 2025 is presented below. A discussion regarding our financial condition and results of operations for fiscal year 2025 compared to fiscal year 2024 can be found under Item 7 in our Annual Report on Form 10-K for the fiscal year ended January 26, 2025, filed with the SEC on February 26, 2025, which is available free of charge on the SEC's website at http://www.sec.gov and at our investor relations website, http://investor.nvidia.com. + +The following table sets forth, for the periods indicated, certain items in our Consolidated Statements of Income expressed as a percentage of revenue. + +
Year Ended
Jan 25, 2026Jan 26, 2025
Revenue100.0 %100.0 %
Cost of revenue28.925.0
Gross profit71.175.0
Operating expenses
Research and development8.69.9
Sales, general and administrative2.12.7
Total operating expenses10.712.6
Operating income60.462.4
Interest income1.11.4
Interest expense(0.1)(0.2)
Other income, net4.20.8
Total other income, net5.22.0
Income before income tax65.564.4
Income tax expense9.98.6
Net income55.6 %55.8 %
+ +## Reportable Segments + +## Revenue by Reportable Segments + +
Year Ended
Jan 25, 2026Jan 26, 2025$ Change% Change
($ in millions)
Compute & Networking$ 193,479$ 116,193$ 77,28667 %
Graphics22,45914,3048,15557 %
Total$ 215,938$ 130,497$ 85,44165 %
+ +## Operating Income by Reportable Segments + +
Year Ended
Jan 25, 2026Jan 26, 2025$ Change% Change
($ in millions)
Compute & Networking$ 130,141$ 82,875$ 47,26657 %
Graphics9,1565,0854,07180 %
Total$ 139,297$ 87,960$ 51,33758 %
+ +Compute & Networking revenue – The year over year increase was driven by the major platform shifts – accelerated computing and AI. Revenue from Data Center computing grew 59% driven by demand for our Blackwell computing platform. Revenue from Data Center networking grew 142% driven by the introduction and continued ramp of NVLink compute fabric for GB200 and GB300 systems and the growth of Ethernet and InfiniBand platforms. + +Graphics revenue – The year over year increase was driven by sales of our Blackwell architecture. + +Reportable segment operating income – The year over year increase in Compute & Networking segment operating income was driven by growth in revenue, partially offset by a \$4.5 billion charge associated with H2O excess inventory and purchase obligations in the first quarter of fiscal year 2026. The year over year increase in Graphics segment operating income was driven by the growth in revenue. + +## Concentration of Revenue + +We refer to customers who purchase products directly from NVIDIA as direct customers, such as AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. We refer to indirect customers as those who purchase products through our direct customers; indirect customers include CSPs, Neocloud builders, AI model makers, enterprises, and public sector entities. Our revenue is concentrated among a limited number of direct and indirect customers and this trend may continue. + +Direct Customers – For fiscal year 2026, sales to one direct customer represented 22% of total revenue and sales to another direct customer represented 14% of total revenue, all of which were primarily attributable to the Compute & Networking segment. + +For fiscal year 2025, sales to one direct customer represented 12% of total revenue and sales to two direct customers each represented 11% of total revenue, all of which were primarily attributable to the Compute & Networking segment. + +For fiscal year 2024, sales to one direct customer represented 13% of total revenue, and were primarily attributable to the Compute & Networking segment. + +Indirect Customers – Indirect customer revenue is an estimation based upon multiple factors including customer purchase order information, product specifications, internal sales data, and other sources. Indirect customers primarily purchase our products through system integrators and distributors. We generate a significant amount of our revenue from a limited number of indirect customers, some individually representing 10% or more of our revenue. Certain companies purchase cloud and related services through various direct and indirect customers. We estimate that one AI research and deployment company contributed to a meaningful amount of our revenue purchasing cloud services from our customers in fiscal year 2026. + +Revenue by geographic region is designated based on the location of the headquarters of direct customers. The end customer and shipping location may be different from our customers' headquarters location. Revenue from sales to customers headquartered outside of the United States accounted for 31% and 41% of total revenue for fiscal years 2026 and 2025, respectively. + +## Gross Profit and Gross Margin + +Gross profit consists of total net revenue less cost of revenue. Cost of revenue consists of the cost of semiconductors, including wafer fabrication, assembly, testing and packaging, board and device costs, manufacturing support costs, including labor and overhead associated with such purchases, final test yield fallout, inventory and warranty provisions, memory and component costs, tariffs, and shipping costs. Cost of revenue also includes acquisition-related intangible amortization expense, costs for license and development and service arrangements, IP-related costs, and stock-based compensation related to personnel associated with manufacturing operations. + +Gross margins decreased to 71.1% in fiscal year 2026 from 75.0% in fiscal year 2025 as our business model transitioned from offering Hopper HGX systems to Blackwell full-scale datacenter solutions and a \$4.5 billion charge associated with H2O excess inventory and purchase obligations in the first quarter of fiscal year 2026. + +Provisions for inventory and excess inventory purchase obligations totaled \$7.2 billion and \$3.7 billion for fiscal years 2026 and 2025, respectively, including \$4.5 billion associated with H2O excess inventory and purchase obligations for the first quarter of fiscal year 2026. Sales of previously reserved inventory and settlements of excess inventory purchase obligations resulted in a provision release of \$1.5 billion and \$689 million for fiscal years 2026 and 2025, respectively. The net effect on our gross margin was an unfavorable impact of 2.6% and 2.3% in fiscal years 2026 and 2025, respectively. + +Operating Expenses + +
Year Ended
Jan 25, 2026Jan 26, 2025$ Change% Change
($ in millions)
Research and development$ 18,497$ 12,914$ 5,58343 %
Sales, general and administrative4,5793,4911,08831 %
Total operating expenses$ 23,076$ 16,405$ 6,67141 %
+ +The increases in research and development expenses for fiscal year 2026 were driven by a 29% increase in compensation and benefits expense, including stock-based compensation, reflecting employee growth and compensation increases and a 79% increase in compute and infrastructure. + +The increases in sales, general and administrative expenses for fiscal year 2026 were primarily driven by compensation and benefits expense, including stock-based compensation, reflecting employee growth and compensation increases. + +Total Other Income, Net + +
Year Ended
Jan 25, 2026Jan 26, 2025$ Change
($ in millions)
Interest income$ 2,300$ 1,786$ 514
Interest expense(259)(247)(12)
Other income, net9,0221,0347,988
Total other income, net$ 11,063$ 2,573$ 8,490
+ +The increase in interest income was primarily due to growth in cash, cash equivalents, and debt securities. + +Interest expense is primarily comprised of coupon interest and debt discount amortization related to our notes. + +Other income, net primarily consists of realized or unrealized gains and losses from investments in non-marketable equity securities, publicly-held equity securities, and the impact of changes in foreign currency rates. The change in Other income, net, compared to fiscal year 2025, was primarily driven by unrealized gains in non-marketable and publicly-held equity securities, including gains from our previously announced investment in Intel's common stock. Refer to Note 7 and 8 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information regarding our non-marketable equity securities. + +## Income Taxes + +Income tax expense was \$21.4 billion and \$11.1 billion for fiscal years 2026 and 2025, respectively. Income tax as a percentage of income before income tax was an expense of 15.1% and 13.3% for fiscal years 2026 and 2025, respectively. + +The effective tax rate increased primarily due to a lower percentage of tax benefits from stock-based compensation, FDDEI, and U.S. federal research tax credit relative to the increase in income before income tax. + +Our effective tax rates for fiscal years 2026 and 2025 were lower than the U.S. federal statutory rate of 21.0% primarily due to tax benefits from FDDEI, stock-based compensation, income earned in jurisdictions that are subject to taxes at rates lower than the U.S. federal statutory tax rate, and the U.S. federal research tax credit. + +In July 2025, the One Big Beautiful Bill Act (OBBBA) was enacted into law and contains several changes to key U.S. federal income tax laws. We have recognized the tax effects of currently effective OBBBA provisions in our results for fiscal year 2026. We will continue to evaluate the impact of these legislative changes as tax authorities provide additional guidance and interpretation. + +As of January 25, 2026, based on recent jurisdictional taxable income and expected future earnings, we concluded certain state deferred tax assets are more likely than not realizable and released \$711 million of valuation allowance. + +Refer to Note 13 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information. + +Liquidity and Capital Resources + +
Jan 25, 2026Jan 26, 2025
(In millions)
Cash and cash equivalents$ 10,605$ 8,589
Marketable securities51,95134,621
Cash, cash equivalents, and marketable securities$ 62,556$ 43,210
+ +Year Ended + +
Jan 25, 2026Jan 26, 2025
(In millions)
Net cash provided by operating activities$ 102,718$ 64,089
Net cash used in investing activities$ (52,228)$ (20,421)
Net cash used in financing activities$ (48,474)$ (42,359)
+ +Our fixed income security investments include highly rated, diversified investment types and credit exposures with shorter maturities. + +Cash provided by operating activities increased in fiscal year 2026 compared to fiscal year 2025, due to higher revenue. + +Cash used in investing activities increased in fiscal year 2026 compared to fiscal year 2025, primarily driven by higher purchases of equity investment securities and the execution of a non-exclusive license agreement with Groq. + +Cash used in financing activities increased in fiscal year 2026 compared to fiscal year 2025, mainly due to higher share repurchases. + +## Liquidity + +Our primary sources of liquidity include cash, cash equivalents, marketable securities, and cash generated by our operations. As of January 25, 2026, we had \$62.6 billion in cash, cash equivalents, and marketable securities. We believe that we have sufficient liquidity to meet our operating requirements for at least the next twelve months and for the foreseeable future, including our future obligations. We continuously evaluate our liquidity and capital resources, including our access to external capital, to ensure we can finance future capital requirements. + +Our marketable securities consist of publicly-held equity securities, debt securities issued by the U.S. government and its agencies, highly-rated corporations and financial institutions, and foreign government entities, as well as certificates of deposit issued by highly-rated financial institutions. These marketable securities are primarily denominated in U.S. dollars. Refer to Note 7 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information. + +Except for approximately \$1.7 billion of cash, cash equivalents, and marketable securities held outside the U.S. for which we have not accrued any related foreign or state taxes if we repatriate these amounts to the U.S., substantially all of our cash, cash equivalents and marketable securities held outside the U.S. at the end of fiscal year 2026 are available for use in the U.S. without incurring additional U.S. federal income taxes. Our first quarter of any fiscal year (including fiscal year 2027) generally does not include any estimated federal and state income tax payments and our second quarter of any fiscal year (including fiscal year 2027) generally includes two estimated federal and state income tax payments. + +## Capital Return to Shareholders + +On August 26, 2025, our Board of Directors approved an additional \$60.0 billion in share repurchase authorization, without expiration. In fiscal year 2026, we repurchased 282 million shares of our common stock for \$40.4 billion. As of January 25, 2026, we were authorized, subject to certain specifications, to repurchase up to \$58.5 billion of our common stock. + +From January 26, 2026 through February 20, 2026, we repurchased 8 million shares for \$1.5 billion pursuant to a pre-established trading plan. We may execute repurchases from time to time, subject to market conditions, operating requirements, and other investment opportunities, in the open market, in privately negotiated transactions, pursuant to a Rule 10b5-1 trading plan or in structured share repurchase agreements in compliance with Rule 10b-18 of the Exchange Act. Our share repurchase program may be suspended at any time at our discretion. + +In fiscal year 2026, we paid cash dividends to our shareholders of \$974 million. The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders. + +The U.S. Inflation Reduction Act of 2022 requires a 1% excise tax on certain share repurchases in excess of shares issued for employee compensation made after December 31, 2022. The excise tax is included in our share repurchase cost and was not material for fiscal years 2026 and 2025. + +## Outstanding Indebtedness and Commercial Paper Program + +Our aggregate debt maturities as of January 25, 2026, by year payable, are as follows: + +
Jan 25, 2026
(In millions)
Due in one year$ 1,000
Due in one to five years2,750
Due in five to ten years1,250
Due in greater than ten years3,500
Unamortized debt discount and issuance costs(32)
Net carrying amount$ 8,468
Less short-term portion999
Total long-term portion$ 7,469
+ +In January 2026, we increased the amount of our commercial paper program, pursuant to which we may issue unsecured commercial paper notes from time to time or all at once up to \$25.0 billion. As of January 25, 2026, no commercial paper was outstanding. We will continue to evaluate issuing commercial paper as a component of our overall liquidity strategy. + +Refer to Note 11 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for further discussion. + +## Material Cash Requirements and Other Obligations + +For descriptions of our facility lease guarantees, long-term debt, purchase commitments, and operating lease obligations, refer to Note 10, Note 11, Note 12, and Note 17 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K, respectively. + +We expect to continue investing in our ecosystem. We are finalizing an investment and partnership agreement with OpenAI. There is no assurance that we will enter into an investment and partnership agreement with OpenAI or that a transaction will be completed. Refer to Item 1A. Risk Factors for additional information regarding our investments. + +During fiscal year 2026 and fiscal year 2025, we spent \$6.1 billion and \$3.4 billion on capital expenditures, respectively. We expect to increase capital expenditures in fiscal year 2027 relative to fiscal year 2026 to support the future growth of our business. + +Unrecognized tax benefits of \$4.0 billion, which includes related interest and penalties of \$374 million, were recorded in non-current income tax payable at the end of fiscal year 2026. We are unable to estimate the timing of any potential tax liability, interest payments, or penalties in individual years due to uncertainties in the underlying income tax positions and the timing of the effective settlement of such tax positions. We are currently under examination by the Internal Revenue Service for our fiscal years 2023 and 2024. Refer to Note 13 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for further information. + +## Adoption of New and Recently Issued Accounting Pronouncements + +Refer to Note 1 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for a discussion of adoption of new and recently issued accounting pronouncements. + +## Item 7A. Quantitative and Qualitative Disclosures about Market Risk + +## Investment and Interest Rate Risk + +We are exposed to interest rate risk related to our fixed-rate investment portfolio and outstanding debt. The investment portfolio is managed consistent with our overall liquidity strategy in support of both working capital needs and growth of our businesses. + +According to our sensitivity analysis on our investment portfolio, a decrease in the yield curve of 0.5% as of the end of fiscal year 2026 and 2025 would decrease the fair value for these investments by approximately \$0.2 billion. + +As of the end of fiscal year 2026, we had \$8.5 billion of senior Notes outstanding. We carry the Notes at face value less unamortized discount on our Consolidated Balance Sheets. As the Notes bear interest at a fixed rate, we have no financial statement risk associated with changes in interest rates. Refer to Note 11 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information. + +Publicly-held equity securities are subject to market price volatility. A hypothetical 10% decrease in our publicly-held equity securities would decrease the fair value of the publicly-held equity securities balance by \$1.8 billion and an insignificant amount as of January 25, 2026 and January 26, 2025, respectively. + +Non-marketable equity securities are measured based on cost minus impairment, if any, and are adjusted for observable price changes in orderly transactions for an identical or similar investment in the same issuer. Valuations of our non-marketable equity securities are inherently complex due to the lack of readily available market data and observable transactions, and impact of macroeconomic factors. + +For a description of our equity investments, refer to Notes 7 and 8 of the Notes to Condensed Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information. + +## Foreign Exchange Rate Risk + +We consider our direct exposure to foreign exchange rate fluctuations to be minimal as substantially all of our sales are in United States dollars and foreign currency forward contracts are used to offset movements of foreign currency exchange rates. Gains or losses from foreign currency remeasurement are included in other income or expenses. The impact of foreign currency transaction gain or loss included in determining net income was not significant for fiscal years 2026 and 2025. + +Sales and arrangements with third-party manufacturers provide for pricing and payment in United States dollars, and, therefore, are not subject to exchange rate fluctuations. Increases in the value of the United States' dollar relative to other currencies would make our products more expensive, which could negatively impact our ability to compete. Conversely, decreases in the value of the United States' dollar relative to other currencies could result in our suppliers raising their manufacturing costs. + +If the U.S. dollar strengthened by 10% as of January 25, 2026 and January 26, 2025, the amount recorded in Accumulated other comprehensive income (loss) related to our foreign exchange contracts before tax effect would have been an adverse impact of \$180 million and \$136 million, respectively. Change in value of our foreign exchange contracts recorded in Accumulated other comprehensive income (loss) would be expected to offset a corresponding change in hedged forecasted foreign currency expenses when recognized. + +If an adverse 10% foreign exchange rate change was applied to our balance sheet hedging contracts, it would have resulted in an adverse impact on income before taxes of \$124 million and \$129 million as of January 25, 2026 and January 26, 2025, respectively. These changes in fair values would be offset in Total other income, net, by corresponding change in fair values of the foreign currency denominated monetary assets and liabilities, assuming the hedge contracts fully cover the foreign currency denominated monetary assets and liabilities balances. + +Refer to Note 10 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information. + +## Item 8. Financial Statements and Supplementary Data + +The information required by this Item is set forth in our Consolidated Financial Statements and Notes thereto included in this Annual Report on Form 10-K. + +## Item 9. Changes in and Disagreements with Accountants on Accounting and Financial Disclosure + +None. + +## Item 9A. Controls and Procedures + +## Controls and Procedures + +## Disclosure Controls and Procedures + +Based on their evaluation as of January 25, 2026, our management, including our Chief Executive Officer and Chief Financial Officer, has concluded that our disclosure controls and procedures (as defined in Rule 13a-15(e) under the Exchange Act) were effective to provide reasonable assurance that the information we are required to disclose in reports that we file or submit under the Exchange Act is recorded, processed, summarized and reported within the time periods specified in the SEC rules and forms, and that such information is accumulated and communicated to our management, including our Chief Executive Officer and our Chief Financial Officer, as appropriate, to allow timely decisions regarding required disclosures. + +## Management's Annual Report on Internal Control Over Financial Reporting + +Our management is responsible for establishing and maintaining adequate internal control over financial reporting, as such term is defined in Exchange Act Rule 13a-15(f). Under the supervision and with the participation of our management, including our Chief Executive Officer and Chief Financial Officer, we conducted an evaluation of the effectiveness of our internal control over financial reporting as of January 25, 2026 based on the criteria set forth in Internal Control - Integrated Framework (2013) issued by the Committee of Sponsoring Organizations of the Treadway Commission. Based on our evaluation under the criteria set forth in Internal Control — Integrated Framework, our management concluded that our internal control over financial reporting was effective as of January 25, 2026. + +The effectiveness of our internal control over financial reporting as of January 25, 2026 has been audited by PricewaterhouseCoopers LLP, an independent registered public accounting firm, as stated in its report which is included herein. + +## Changes in Internal Control Over Financial Reporting + +There have been no changes in our internal control over financial reporting during the quarter ended January 25, 2026 that have materially affected, or are reasonably likely to materially affect, our internal control over financial reporting. We are continuing a phased upgrade of our enterprise resource planning, or ERP, system to update our existing core financial systems. The ERP system is designed to accurately maintain our financial records used to report operating results. We will continue to evaluate each quarter whether there are changes that materially affect our internal control over financial reporting. + +## Inherent Limitations on Effectiveness of Controls + +Our management, including our Chief Executive Officer and Chief Financial Officer, does not expect that our disclosure controls and procedures or our internal controls, will prevent all error and all fraud. A control system, no matter how well conceived and operated, can provide only reasonable, not absolute, assurance that the objectives of the control system are met. Further, the design of a control system must reflect the fact that there are resource constraints, and the benefits of controls must be considered relative to their costs. Because of the inherent limitations in all control systems, no evaluation of controls can provide absolute assurance that all control issues and instances of fraud, if any, within NVIDIA have been detected. + +## Item 9B. Other Information + +The following members of our Board of Directors and/or officers adopted, modified or terminated a trading arrangement that is intended to satisfy the affirmative defense conditions of Rule 10b5-1(c), or a Rule 10b5-1 Trading Arrangement: + +
NameTitle of Director or OfficerActionDateTotal Shares of Common Stock to be SoldExpiration Date
John O. DabiriDirectorAdoption12/10/20253,984*12/7/2026
Colette M. KressExecutive Vice President and Chief Financial OfficerAdoption12/18/2025500,0003/23/2027
+ +\*Estimated assuming our closing stock price as of January 23, 2026. The number of shares is based on an estimate because the plan specifies a formulaic dollar amount of shares to be sold. + +## Item 9C. Disclosure Regarding Foreign Jurisdictions that Prevent Inspections + +Not Applicable. + +## Part III + +Certain information required by Part III is omitted from this report because we will file with the SEC a definitive proxy statement pursuant to Regulation 14A, or the 2026 Proxy Statement, no later than 120 days after the end of fiscal year 2026, and certain information included therein is incorporated herein by reference. + +## Item 10. Directors, Executive Officers and Corporate Governance + +## Identification of Directors + +Information regarding directors required by this item will be contained in our 2026 Proxy Statement under the caption “Proposal 1 — Election of Directors,” and is hereby incorporated by reference. + +## Identification of Executive Officers + +Reference is made to the information regarding executive officers appearing under the heading “Information About Our Executive Officers” in Part I of this Annual Report on Form 10-K, which information is hereby incorporated by reference. + +## Identification of Audit Committee and Financial Experts + +Information regarding our Audit Committee required by this item will be contained in our 2026 Proxy Statement under the captions “Report of the Audit Committee of the Board of Directors” and “Information About the Board of Directors and Corporate Governance,” and is hereby incorporated by reference. + +## Material Changes to Procedures for Recommending Directors + +Information regarding procedures for recommending directors required by this item will be contained in our 2026 Proxy Statement under the caption “Information About the Board of Directors and Corporate Governance,” and is hereby incorporated by reference. + +## Delinquent Section 16(a) Reports + +Information regarding compliance with Section 16(a) of the Exchange Act required by this item will be contained in our 2026 Proxy Statement under the caption “Delinquent Section 16(a) Reports,” and such disclosure, if any, is hereby incorporated by reference. + +## Code of Conduct + +Information regarding our Code of Conduct required by this item will be contained in our 2026 Proxy Statement under the caption “Information About the Board of Directors and Corporate Governance — Code of Conduct,” and is hereby incorporated by reference. The full text of our Code of Conduct and Financial Team Code of Conduct are published on the Investor Relations portion of our website, under Governance, at www.nvidia.com. If we make any amendments to either code, or grant any waiver from a provision of either code to any executive officer or director, we will promptly disclose the nature of the amendment or waiver on our website or in a report on Form 8-K. The contents of our website are not a part of this Annual Report on Form 10-K. + +## Insider Trading Policy + +The information required by Item 408(b) of Regulation S-K is incorporated by reference from the information contained in our 2026 Proxy Statement under the heading “Information About the Board of Directors and Corporate Governance.” + +## Item 11. Executive Compensation + +Information regarding our executive compensation required by this item will be contained in our 2026 Proxy Statement under the captions “Executive Compensation,” “Compensation Committee Interlocks and Insider Participation,” “Director Compensation,” and “Compensation Committee Report,” and is hereby incorporated by reference. + +## Item 12. Security Ownership of Certain Beneficial Owners and Management and Related Stockholder Matters + +## Ownership of NVIDIA Securities + +Information regarding ownership of NVIDIA securities required by this item will be contained in our 2026 Proxy Statement under the caption “Security Ownership of Certain Beneficial Owners and Management,” and is hereby incorporated by reference. + +## Equity Compensation Plan Information + +Information regarding our equity compensation plans required by this item will be contained in our 2026 Proxy Statement under the caption "Equity Compensation Plan Information," and is hereby incorporated by reference. + +## Item 13. Certain Relationships and Related Transactions, and Director Independence + +Information regarding related transactions and director independence required by this item will be contained in our 2026 Proxy Statement under the captions “Review of Transactions with Related Persons” and “Information About the Board of Directors and Corporate Governance — Independence of the Members of the Board of Directors,” and is hereby incorporated by reference. + +## Item 14. Principal Accountant Fees and Services + +Information regarding accounting fees and services required by this item will be contained in our 2026 Proxy Statement under the caption “Fees Billed by the Independent Registered Public Accounting Firm,” and is hereby incorporated by reference. + +## Item 15. Exhibits and Financial Statement Schedules + +Page + +## (a) 1. Financial Statements + +Report of Independent Registered Public Accounting Firm (PCAOB ID: 238) 49 + +Consolidated Statements of Income for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 51 + +Consolidated Statements of Comprehensive Income for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 52 + +Consolidated Balance Sheets as of January 25, 2026 and January 26, 2025 53 + +Consolidated Statements of Shareholders' Equity for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 54 + +Consolidated Statements of Cash Flows for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 55 + +Notes to the Consolidated Financial Statements 56 + +## 2. Financial Statement Schedule + +Schedule II Valuation and Qualifying Accounts for the years ended January 25, 2026, January 26, 2025, and January 28, 2024 81 + +## 3. Exhibits + +The exhibits listed in the accompanying index to exhibits are filed or incorporated by reference as a part of this Annual Report on Form 10-K. 82 + +# Report of Independent Registered Public Accounting Firm + +To the Board of Directors and Shareholders of NVIDIA Corporation + +## Opinions on the Financial Statements and Internal Control over Financial Reporting + +We have audited the accompanying consolidated balance sheets of NVIDIA Corporation and its subsidiaries (the "Company") as of January 25, 2026 and January 26, 2025, and the related consolidated statements of income, comprehensive income, shareholders' equity, and cash flows for each of the three years in the period ended January 25, 2026, including the related notes and financial statement schedule listed in the index appearing under Item 15(a)(2) (collectively referred to as the "consolidated financial statements"). We also have audited the Company's internal control over financial reporting as of January 25, 2026, based on criteria established in Internal Control - Integrated Framework (2013) issued by the Committee of Sponsoring Organizations of the Treadway Commission (COSO). + +In our opinion, the consolidated financial statements referred to above present fairly, in all material respects, the financial position of the Company as of January 25, 2026 and January 26, 2025, and the results of its operations and its cash flows for each of the three years in the period ended January 25, 2026 in conformity with accounting principles generally accepted in the United States of America. Also in our opinion, the Company maintained, in all material respects, effective internal control over financial reporting as of January 25, 2026, based on criteria established in Internal Control - Integrated Framework (2013) issued by the COSO. + +## Basis for Opinions + +The Company's management is responsible for these consolidated financial statements, for maintaining effective internal control over financial reporting, and for its assessment of the effectiveness of internal control over financial reporting, included in Management's Annual Report on Internal Control over Financial Reporting appearing under Item 9A. Our responsibility is to express opinions on the Company's consolidated financial statements and on the Company's internal control over financial reporting based on our audits. We are a public accounting firm registered with the Public Company Accounting Oversight Board (United States) (PCAOB) and are required to be independent with respect to the Company in accordance with the U.S. federal securities laws and the applicable rules and regulations of the Securities and Exchange Commission and the PCAOB. + +We conducted our audits in accordance with the standards of the PCAOB. Those standards require that we plan and perform the audits to obtain reasonable assurance about whether the consolidated financial statements are free of material misstatement, whether due to error or fraud, and whether effective internal control over financial reporting was maintained in all material respects. + +Our audits of the consolidated financial statements included performing procedures to assess the risks of material misstatement of the consolidated financial statements, whether due to error or fraud, and performing procedures that respond to those risks. Such procedures included examining, on a test basis, evidence regarding the amounts and disclosures in the consolidated financial statements. Our audits also included evaluating the accounting principles used and significant estimates made by management, as well as evaluating the overall presentation of the consolidated financial statements. Our audit of internal control over financial reporting included obtaining an understanding of internal control over financial reporting, assessing the risk that a material weakness exists, and testing and evaluating the design and operating effectiveness of internal control based on the assessed risk. Our audits also included performing such other procedures as we considered necessary in the circumstances. We believe that our audits provide a reasonable basis for our opinions. + +## Definition and Limitations of Internal Control over Financial Reporting + +A company's internal control over financial reporting is a process designed to provide reasonable assurance regarding the reliability of financial reporting and the preparation of financial statements for external purposes in accordance with generally accepted accounting principles. A company's internal control over financial reporting includes those policies and procedures that (i) pertain to the maintenance of records that, in reasonable detail, accurately and fairly reflect the transactions and dispositions of the assets of the company; (ii) provide reasonable assurance that transactions are recorded as necessary to permit preparation of financial statements in accordance with generally accepted accounting principles, and that receipts and expenditures of the company are being made only in accordance with authorizations of management and directors of the company; and (iii) provide reasonable assurance regarding prevention or timely detection of unauthorized acquisition, use, or disposition of the company's assets that could have a material effect on the financial statements. + +Because of its inherent limitations, internal control over financial reporting may not prevent or detect misstatements. Also, projections of any evaluation of effectiveness to future periods are subject to the risk that controls may become inadequate because of changes in conditions, or that the degree of compliance with the policies or procedures may deteriorate. + +## Critical Audit Matters + +The critical audit matter communicated below is a matter arising from the current period audit of the consolidated financial statements that was communicated or required to be communicated to the audit committee and that (i) relates to accounts or disclosures that are material to the consolidated financial statements and (ii) involved our especially challenging, subjective, or complex judgments. The communication of critical audit matters does not alter in any way our opinion on the consolidated financial statements, taken as a whole, and we are not, by communicating the critical audit matter below, providing a separate opinion on the critical audit matter or on the accounts or disclosures to which it relates. + +Valuation of Inventories - Provisions for Excess or Obsolete Inventories and Excess Product Purchase Commitments + +As described in Notes 1, 9, and 12 to the consolidated financial statements, the Company charges cost of sales for inventory provisions to write-down inventory for excess or obsolete inventory and for excess product purchase commitments. Most of the Company's inventory provisions relate to excess quantities of products, based on the Company's inventory levels and future product purchase commitments compared to assumptions about future demand including the impact of market conditions such as regulatory export restrictions on their products. As of January 25, 2026, the Company's consolidated inventories balance was \$21.4 billion and the Company's consolidated outstanding inventory purchase and long-term supply and capacity obligations balance was \$95.2 billion, of which a significant portion relates to inventory purchase obligations. + +The principal considerations for our determination that performing procedures relating to the valuation of inventories, specifically the provisions for excess or obsolete inventories and excess product purchase commitments, is a critical audit matter are the significant judgment by management when developing provisions for excess or obsolete inventories and excess product purchase commitments, including developing assumptions related to future demand and market conditions. This in turn led to significant auditor judgment, subjectivity, and effort in performing procedures and evaluating management's assumptions related to future demand and market conditions. + +Addressing the matter involved performing procedures and evaluating audit evidence in connection with forming our overall opinion on the consolidated financial statements. These procedures included testing the effectiveness of controls relating to management's provisions for excess or obsolete inventories and excess product purchase commitments, including controls over management's assumptions related to future demand and market conditions. These procedures also included, among others, testing management's process for developing the provisions for excess or obsolete inventories and excess product purchase commitments; evaluating the appropriateness of management's approach; testing the completeness and accuracy of underlying data used in the approach; and evaluating the reasonableness of management's assumptions related to future demand and market conditions. Evaluating management's assumptions related to future demand and market conditions involved evaluating whether the assumptions used by management were reasonable considering (i) current and past results, including historical product life cycle, (ii) the consistency with external market and industry data, and (iii) changes in technology. + +/s/PricewaterhouseCoopers LLP + +San Jose, California + +February 25, 2026 + +We have served as the Company's auditor since 2004. + +NVIDIA Corporation and Subsidiaries +Consolidated Statements of Income +(In millions, except per share data) + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Revenue$ 215,938$ 130,497$ 60,922
Cost of revenue62,47532,63916,621
Gross profit153,46397,85844,301
Operating expenses
Research and development18,49712,9148,675
Sales, general and administrative4,5793,4912,654
Total operating expenses23,07616,40511,329
Operating income130,38781,45332,972
Interest income2,3001,786866
Interest expense(259)(247)(257)
Other income, net9,0221,034237
Total other income, net11,0632,573846
Income before income tax141,45084,02633,818
Income tax expense21,38311,1464,058
Net income$ 120,067$ 72,880$ 29,760
Net income per share:
Basic$ 4.93$ 2.97$ 1.21
Diluted$ 4.90$ 2.94$ 1.19
Weighted average shares used in per share computation:
Basic24,35924,55524,690
Diluted24,51424,80424,940
+ +See accompanying Notes to the Consolidated Financial Statements. + +NVIDIA Corporation and Subsidiaries +Consolidated Statements of Comprehensive Income +(In millions) + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Net income$ 120,067$ 72,880$ 29,760
Other comprehensive income, net of tax
Available-for-sale securities:
Net change in unrealized gain107180
Cash flow hedges:
Net change in unrealized gain (loss)43(10)
Other comprehensive income, net of tax150170
Total comprehensive income$ 120,217$ 72,881$ 29,830
+ +See accompanying Notes to the Consolidated Financial Statements. + +NVIDIA Corporation and Subsidiaries +Consolidated Balance Sheets +(In millions, except par value) + +
Jan 25, 2026Jan 26, 2025
Assets
Current assets:
Cash and cash equivalents$ 10,605$ 8,589
Marketable securities51,95134,621
Accounts receivable, net38,46623,065
Inventories21,40310,080
Prepaid expenses and other current assets3,1803,771
Total current assets125,60580,126
Property and equipment, net10,3836,283
Operating lease assets2,8671,793
Goodwill20,8325,188
Intangible assets, net3,306807
Deferred income tax assets13,25810,979
Non-marketable equity securities22,2513,387
Other assets8,3013,038
Total assets$ 206,803$ 111,601
Liabilities and Shareholders' Equity
Current liabilities:
Accounts payable$ 9,812$ 6,310
Accrued and other current liabilities21,35211,737
Short-term debt999
Total current liabilities32,16318,047
Long-term debt7,4698,463
Long-term operating lease liabilities2,5721,519
Other long-term liabilities7,3064,245
Total liabilities49,51032,274
Commitments and contingencies - see Note 12
Shareholders' equity:
Preferred stock, $0.001 par value; 2 shares authorized; none issued
Common stock, $0.001 par value; 80,000 shares authorized; 24,304 shares issued and outstanding as of January 25, 2026; 24,477 shares issued and outstanding as of January 26, 20252424
Additional paid-in capital10,11811,237
Accumulated other comprehensive income17828
Retained earnings146,97368,038
Total shareholders' equity157,29379,327
Total liabilities and shareholders' equity$ 206,803$ 111,601
+ +See accompanying Notes to the Consolidated Financial Statements. + +NVIDIA Corporation and Subsidiaries +Consolidated Statements of Shareholders' Equity + +
Common Stock OutstandingAdditional Paid-in CapitalAccumulated Other Comprehensive Income (Loss)Retained EarningsTotal Shareholders' Equity
SharesAmount
(In millions, except per share data)
Balances as of Jan 29, 202324,661$ 25$ 11,948$ (43)$ 10,171$ 22,101
Net income29,76029,760
Other comprehensive income7070
Issuance of common stock265403403
Tax withholding related to common stock(72)(2,783)(2,783)
Shares repurchased(211)(27)(9,719)(9,746)
Cash dividends declared and paid ($0.016 per common share)(395)(395)
Stock-based compensation3,5683,568
Balances as of Jan 28, 202424,6432513,1092729,81742,978
Net income72,88072,880
Other comprehensive income11
Issuance of common stock203490490
Tax withholding related to common stock(59)(6,930)(6,930)
Shares repurchased(310)(1)(189)(33,825)(34,015)
Cash dividends declared and paid ($0.034 per common share)(834)(834)
Stock-based compensation4,7574,757
Balances as of Jan 26, 202524,4772411,2372868,03879,327
Net income120,067120,067
Other comprehensive income150150
Issuance of common stock160644644
Tax withholding related to common stock(51)(7,948)(7,948)
Shares repurchased(282)(230)(40,158)(40,388)
Cash dividends declared and paid ($0.04 per common share)(974)(974)
Fair value of partially vested equity awards assumed in connection with acquisitions2828
Stock-based compensation6,3876,387
Balances as of Jan 25, 202624,304$ 24$ 10,118$ 178$ 146,973$ 157,293
+ +See accompanying Notes to the Consolidated Financial Statements. + +NVIDIA Corporation and Subsidiaries Consolidated Statements of Cash Flows (In millions) + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Cash flows from operating activities:
Net income$120,067$72,880$29,760
Adjustments to reconcile net income to net cash provided by operating activities:
Stock-based compensation expense6,3864,7373,549
Depreciation and amortization2,8431,8641,508
Gains on non-marketable equity securities and publicly-held equity securities, net(8,918)(1,030)(238)
Deferred income taxes(1,424)(4,477)(2,489)
Other(287)(502)(278)
Changes in operating assets and liabilities, net of acquisitions:
Accounts receivable(15,399)(13,063)(6,172)
Inventories(11,324)(4,781)(98)
Prepaid expenses and other assets577(395)(1,522)
Accounts payable3,0963,3571,531
Accrued and other current liabilities5,2574,2782,025
Other long-term liabilities1,8441,221514
Net cash provided by operating activities102,71864,08928,090
Cash flows from investing activities:
Proceeds from sales of marketable securities15,15749550
Proceeds from maturities of marketable securities11,22611,1959,732
Proceeds from sales of non-marketable equity securities841711
Purchases of marketable securities(40,616)(26,575)(18,211)
Purchases of non-marketable equity securities(17,502)(1,486)(862)
Groq, Inc.(13,000)
Purchases related to property and equipment and intangible assets(6,042)(3,236)(1,069)
Acquisitions, net of cash acquired(1,535)(1,007)(83)
Other22(124)
Net cash used in investing activities(52,228)(20,421)(10,566)
Cash flows from financing activities:
Proceeds related to employee stock plans644490403
Payments related to repurchases of common stock(40,086)(33,706)(9,533)
Payments related to employee stock plan taxes(7,948)(6,930)(2,783)
Dividends paid(974)(834)(395)
Principal payments on property and equipment and intangible assets(101)(129)(74)
Repayment of debt(1,250)(1,250)
Other(9)(1)
Net cash used in financing activities(48,474)(42,359)(13,633)
Change in cash and cash equivalents2,0161,3093,891
Cash and cash equivalents at beginning of period8,5897,2803,389
Cash and cash equivalents at end of period$10,605$8,589$7,280
Supplemental disclosures of cash flow information:
Cash paid for income taxes, net$20,288$15,118$6,549
+ +See accompanying Notes to the Consolidated Financial Statements. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +## Note 1 - Organization and Summary of Significant Accounting Policies + +## Our Company + +Headquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998. + +All references to “NVIDIA,” “we,” “us,” “our” or the “Company” mean NVIDIA Corporation and its subsidiaries. + +Certain prior fiscal year balances have been reclassified to conform to the current fiscal year presentation. Non-marketable equity securities, previously presented within other assets, were reclassified to be presented separately on our consolidated balance sheets and had no impact to total assets or consolidated statement of cash flows. + +## Fiscal Year + +We operate on a 52- or 53-week year, ending on the last Sunday in January. Fiscal years 2026, 2025 and 2024 were all 52-week years. Fiscal year 2027 will be a 53-week year with the fourth quarter consisting of 14 weeks. + +## Principles of Consolidation + +Our consolidated financial statements include the accounts of NVIDIA Corporation and our wholly-owned subsidiaries. All intercompany balances and transactions have been eliminated in consolidation. + +## Use of Estimates + +The preparation of financial statements in conformity with U.S. GAAP requires management to make estimates and assumptions that affect the reported amounts of assets and liabilities and disclosures of contingent assets and liabilities at the date of the financial statements and the reported amounts of revenue and expenses during the reporting period. Actual results could differ materially from our estimates. On an on-going basis, we evaluate our estimates, including those related to accounts receivable, cash equivalents and marketable securities, goodwill, income taxes, inventories and product purchase commitments, investigation and settlement costs, litigation, non-marketable equity securities, other contingencies, property, plant, and equipment, revenue recognition, and stock-based compensation. These estimates are based on historical facts and various other assumptions that we believe are reasonable. + +## Revenue Recognition + +We derive our revenue primarily from product sales including hardware and systems. We determine revenue recognition through the following steps: (1) identification of the contract with a customer; (2) identification of the performance obligations in the contract; (3) determination of the transaction price; (4) allocation of the transaction price to the performance obligations in the contract (where revenue is allocated on a relative standalone selling price basis by maximizing the use of observable inputs to determine the standalone selling price for each performance obligation); and (5) recognition of revenue when, or as, we satisfy a performance obligation. Payment from customers, per our standard payment terms, is generally due shortly after delivery of our products. + +## Product Sales Revenue + +Revenue from product sales is recognized upon transfer of control of products to customers in an amount that reflects the consideration we expect to receive in exchange for those products. Certain products are sold with support or an extended warranty. Support and extended warranty revenue are recognized ratably over the service period, or as services are performed. Revenue is recognized net of allowances for returns, customer programs and any taxes collected from customers. + +For products sold with a right of return, we record a reduction to revenue by establishing a sales return allowance for estimated product returns at the time revenue is recognized, based primarily on historical return rates. However, if product returns for a fiscal period are anticipated to exceed historical return rates, we may determine that additional sales return allowances are required to accurately reflect our estimated exposure for product returns. + +Our customer programs involve rebates, which are designed to serve as sales incentives to resellers of our products in various target markets, and MDFs which represent monies paid to our partners that are earmarked for market segment development and are designed to support our partners' activities while also promoting NVIDIA products. We account for customer programs as a reduction to revenue and accrue for such programs for potential rebates and MDFs based on the amount we expect to be claimed by customers. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued) + +## Contracts with Multiple Performance Obligations + +Our contracts may contain more than one deliverable, each of which is separately accounted for as a distinct performance obligation. We account for multiple agreements with a single customer as a single contract if the contractual terms and/or substance of those agreements indicate that they may be so closely related that they are, in effect, parts of a single contract. + +We allocate the total transaction price to each distinct performance obligation in an arrangement with multiple performance obligations on a relative standalone selling price basis. The standalone selling price reflects the price we would charge for a specific product or service if it were sold separately in similar circumstances and to similar customers. When determining standalone selling price, we maximize the use of observable inputs. + +## Product Warranties + +We offer a limited warranty to end-users ranging from one to three years for products to repair or replace products for manufacturing defects or hardware component failures. Cost of revenue includes the estimated cost of product warranties that are calculated at the point of revenue recognition. Under limited circumstances, we may offer an extended limited warranty to customers for certain products. We also accrue for known warranty and indemnification issues if a loss is probable and can be reasonably estimated. + +## Stock-based Compensation + +We use the closing trading price of our common stock on the date of grant, minus a dividend yield discount, as the fair value of awards of restricted stock units, or RSUs, and performance stock units, or PSUs, that are based on our corporate financial performance targets. We use a Monte Carlo simulation on the date of grant to estimate the fair value of PSUs that are based on our stock performance compared to market performance, or market-based PSUs. The compensation expense for RSUs and market-based PSUs is recognized using a straight-line attribution method over the requisite employee service period while compensation expense for PSUs is recognized using an accelerated amortization model based on performance targets probable of achievement. We estimate the fair value of shares to be issued under our employee stock purchase plan, or ESPP, using the Black-Scholes model at the commencement of an offering period in March and September of each year. Stock-based compensation for our ESPP is expensed using an accelerated amortization model. Additionally, for RSUs, PSUs, and market-based PSUs, we estimate expected forfeitures based on our historical forfeitures. + +## Litigation, Investigation and Settlement Costs + +We currently are, and will likely continue to be subject to claims, litigation, and other actions, including potential regulatory proceedings, involving patent and other intellectual property matters, taxes, labor and employment, competition and antitrust, commercial disputes, goods and services offered by us and by third parties, and other matters. There are many uncertainties associated with any litigation or investigation, and we cannot be certain that these actions or other third-party claims against us will be resolved without litigation, fines and/or substantial settlement payments or judgments. If information becomes available that causes us to determine that a loss in any of our pending litigation, investigations or settlements is probable, and we can reasonably estimate the loss associated with such events, we will record the loss. However, the actual liability in any such litigation or investigation may be materially different from our estimates, which could require us to record additional costs. If we determine that a loss is reasonably possible and the loss or range of loss can be estimated, we disclose the reasonably possible loss. + +## Foreign Currency Remeasurement + +We use the U.S. dollar as our functional currency for our subsidiaries. Foreign currency monetary assets and liabilities are remeasured into United States dollars at end-of-period exchange rates. Non-monetary assets and liabilities such as property and equipment and equity are remeasured at historical exchange rates. Revenue and expenses are remeasured at exchange rates in effect during each period, except for those expenses related to non-monetary balance sheet amounts, which are remeasured at historical exchange rates. Gains or losses from foreign currency remeasurement are included in earnings in our Consolidated Statements of Income and to date have not been significant. + +## Income Taxes + +We recognize federal, state and foreign current tax liabilities or assets based on our estimate of taxes payable or refundable in the current fiscal year by tax jurisdiction. We recognize federal, state and foreign deferred tax assets or liabilities, as appropriate, for our estimate of future tax effects attributable to temporary differences and carryforwards; and we record a valuation allowance to reduce any deferred tax assets by the amount of any tax benefits that, based on available evidence and judgment, are not expected to be realized. + +Our calculation of deferred tax assets and liabilities is based on certain estimates and judgments and involves dealing with uncertainties in the application of complex tax laws. Our estimates of deferred tax assets and liabilities may change + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued) + +based, in part, on added certainty or finality to an anticipated outcome, changes in accounting standards or tax laws in the U.S., or foreign jurisdictions where we operate, or changes in other facts or circumstances. In addition, we recognize liabilities for potential U.S. and foreign income tax contingencies based on our estimate of whether, and the extent to which, additional taxes may be due. If we determine that payment of these amounts is unnecessary or if the recorded tax liability is less than our current assessment, we may be required to recognize an income tax benefit or additional income tax expense in our financial statements accordingly. + +As of January 25, 2026, we had a valuation allowance of \$768 million related to capital loss carryforwards, and certain other deferred tax assets that management determined are not likely to be realized due, in part, to jurisdictional projections of future taxable income, including capital gains. To the extent realization of the deferred tax assets becomes more-likely-than-not, we would recognize such deferred tax assets as income tax benefits during the period. + +We recognize the benefit from a tax position only if it is more-likely-than-not that the position would be sustained upon audit based solely on the technical merits of the tax position. Our policy is to include interest and penalties related to unrecognized tax benefits as a component of income tax expense. + +## Net Income Per Share + +Basic net income per share is computed using the weighted average number of common shares outstanding during the period. Diluted net income per share is computed using the weighted average number of common and potentially dilutive shares outstanding during the period, using the treasury stock method. Any anti-dilutive effect of equity awards outstanding is not included in the computation of diluted net income per share. + +## Cash and Cash Equivalents and Marketable Securities + +We consider all highly liquid investments that are readily convertible into cash and have an original maturity of three months or less at the time of purchase to be cash equivalents. Marketable securities consist of highly liquid debt investments with maturities of greater than three months when purchased and publicly-held equity securities. We classify these investments as current or long term based on the nature of the investments and their availability for use in current operations. + +We record our debt investments as cash equivalents and marketable debt securities and classify them at the date of acquisition as available-for-sale. These available-for-sale debt securities are reported at fair value with the related unrealized gains and losses included in accumulated other comprehensive income or loss, a component of shareholders' equity, net of tax. The fair value of interest-bearing debt securities includes accrued interest. Realized gains and losses on the sale of marketable securities are determined using the specific-identification method and recorded in the Other income (expense), net, section of our Consolidated Statements of Income. + +Available-for-sale debt securities are subject to impairment review. If the estimated fair value of available-for-sale debt securities is less than its amortized cost basis, we determine if the difference, if any, is caused by expected credit losses and write-down the amortized cost basis of the securities if it is more likely than not we will be required or we intend to sell the securities before recovery of its amortized cost basis. Allowances for credit losses and write-downs are recognized in the Other income, net, net section of our Consolidated Statements of Income. + +Publicly-held equity securities and money market funds have readily determinable fair values with changes in fair value recorded in Other income, net. + +## Fair Value of Financial Instruments + +The carrying value of cash equivalents, accounts receivable, accounts payable and accrued liabilities approximate their fair values due to their relatively short maturities as of January 25, 2026 and January 26, 2025. Marketable debt and equity securities are reported at fair value. Derivative instruments are recognized as either assets or liabilities and are measured at fair value. The accounting for changes in the fair value of a derivative depends on the intended use of the derivative and the resulting designation. For derivative instruments designated as accounting hedges, the effective portion of the gains or losses on the derivatives is initially reported as a component of other comprehensive income or loss and is subsequently recognized in earnings when the hedged exposure is recognized in earnings. For derivative instruments not designated as accounting hedges, changes in fair value are recognized in earnings. + +Financial instruments measured and disclosed at fair value are classified and disclosed based on the observability of inputs used in the determination of fair value as follows: + +• Level 1: Observable inputs such as quoted prices in active markets. +- Level 2: Observable inputs other than Level 1 prices, such as quoted prices in less active markets or model-derived valuations that are observable either directly or indirectly. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued) + +\- Level 3: Unobservable inputs in which there is little or no market data that are significant to the fair value of the assets or liabilities. + +## Concentration of Credit Risk + +Financial instruments that potentially subject us to concentrations of credit risk consist primarily of cash equivalents, marketable securities, lease guarantees, and accounts receivable. Our investment policy requires the purchase of highly-rated fixed income securities, the diversification of investment type and credit exposures, and includes certain limits on our portfolio maturities. We perform ongoing credit evaluations of our customers' financial condition and maintain an allowance for potential credit losses. This allowance consists of an amount identified for specific customers and an amount based on overall estimated exposure. Our overall estimated exposure excludes amounts covered by credit insurance and letters of credit. + +## Inventories + +Inventory cost is computed on an adjusted standard basis, which approximates actual cost on an average or first-in, first-out basis. Inventory costs consist of the cost of semiconductors, including wafer fabrication, assembly, testing and packaging, manufacturing support costs, including labor and overhead associated with such purchases, final test yield fallout, and shipping costs, as well as the cost of purchased memory products and other component parts. We charge cost of sales for inventory provisions to write-down our inventory to the lower of cost or net realizable value or for obsolete or excess inventory, and for excess product purchase commitments. Most of our inventory provisions relate to excess quantities of products, based on our inventory levels and future product purchase commitments compared to assumptions about future demand including the impact of regulatory export restrictions on our products. Once inventory has been written-off or written-down, it creates a new cost basis for the inventory that is not subsequently written-up. We record a liability for noncancelable purchase commitments with suppliers for quantities in excess of our future demand forecasts consistent with our valuation of obsolete or excess inventory. + +## Property and Equipment + +Property and equipment are stated at cost less accumulated depreciation. Depreciation of property and equipment is computed using the straight-line method based on the estimated useful lives of the assets of two to seven years. Once an asset is identified for retirement or disposition, the related cost and accumulated depreciation or amortization are removed, and a gain or loss is recorded. The estimated useful lives of our buildings are up to thirty years. Depreciation expense includes the amortization of assets recorded under finance leases. Leasehold improvements and assets recorded under finance leases are amortized over the shorter of the expected lease term or the estimated useful life of the asset. + +## Leases + +We determine if an arrangement is or contains a lease at inception. Operating leases with lease terms of more than 12 months are included in operating lease assets, accrued and other current liabilities, and long-term operating lease liabilities on our consolidated balance sheet. Operating lease assets represent our right to use an underlying asset for the lease term and lease liabilities represent our obligation to make lease payments over the lease term. We combine lease and non-lease components for offices and data centers in determining the operating lease assets and liabilities. + +Operating lease assets and liabilities are recognized based on the present value of the remaining lease payments discounted using our incremental borrowing rate. Operating lease assets also include initial direct costs incurred and prepaid lease payments, minus any lease incentives. Our lease terms include options to extend or terminate the lease when it is reasonably certain that we will exercise that option. Lease costs are recognized on a straight-line basis over the lease term. + +## Goodwill + +We allocate goodwill to reporting units based on the expected benefit from the business combination. Goodwill is subject to our annual impairment test during the fourth quarter of our fiscal year, or earlier if indicators of potential impairment exist. In completing our impairment test, we perform either a qualitative or a quantitative analysis on a reporting unit basis. + +Qualitative factors include industry and market considerations, overall financial performance, and other relevant events and factors affecting the reporting units. Goodwill impairments were not identified for the periods presented. + +## Intangible Assets and Other Long-Lived Assets + +Intangible assets primarily represent acquired intangible assets including developed technology and customer relationships, as well as rights acquired under technology licenses, patents, and acquired IP. We currently amortize our intangible assets with finite lives over periods ranging from one to twenty years using a method that reflects the pattern + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued) + +in which the economic benefits of the intangible asset are consumed or otherwise used up or, if that pattern cannot be reliably determined, using a straight-line amortization method. + +Long-lived assets, such as property and equipment and intangible assets subject to amortization, are reviewed for impairment whenever events or changes in circumstances indicate that the carrying amount of an asset or asset group may not be recoverable. The recoverability of assets or asset groups to be held and used is measured by a comparison of the carrying amount of an asset or asset group to estimated undiscounted future cash flows expected to be generated by the asset or asset group. If the carrying amount of an asset or asset group exceeds its estimated future cash flows, an impairment charge is recognized for the amount by which the carrying amount of the asset or asset group exceeds the estimated fair value of the asset or asset group. Fair value is determined based on the estimated discounted future cash flows expected to be generated by the asset or asset group. Assets and liabilities to be disposed of would be separately presented in the Consolidated Balance Sheet and the assets would be reported at the lower of the carrying amount or fair value less costs to sell, and would no longer be depreciated. + +## Business Combination + +The Company applies a screen test to evaluate if substantially all of the fair value of the gross assets acquired is concentrated in a single identifiable asset or group of similar identifiable assets to determine whether a transaction is accounted for as an asset acquisition or business combination. We allocate the fair value of the purchase price of an acquisition to the tangible assets acquired, liabilities assumed, and intangible assets acquired, based on their estimated fair values. The excess of the fair value of the purchase price over the fair values of these net tangible and intangible assets acquired is recorded as goodwill. Management's estimates of fair value are based upon assumptions believed to be reasonable, but our estimates and assumptions are inherently uncertain and subject to refinement. The estimates and assumptions used in valuing intangible assets include, but are not limited to, the amount and timing of projected future cash flows, discount rate used to determine the present value of these cash flows and asset lives. These estimates are inherently uncertain and, therefore, actual results may differ from the estimates made. As a result, during the measurement period of up to one year from the acquisition date, we may record adjustments to the assets acquired and liabilities assumed with the corresponding offset to goodwill. Upon the measurement period's conclusion or final determination of the fair value of the purchase price of an acquisition, whichever comes first, any subsequent adjustments are recorded to our Consolidated Statements of Income. + +Acquisition-related expenses are recognized separately from the business combination and expensed as incurred. + +## Non-Marketable Equity Securities + +Non-marketable equity securities consist of investments in privately-held companies that do not have a readily determinable fair value. These investments are measured at cost minus impairment, if any, and are adjusted for changes resulting from observable price changes in orderly transactions for an identical or similar investment in the same issuer, or the measurement alternative. Fair value is based upon observable inputs in an inactive market and the valuation requires our judgment due to the absence of market prices and inherent lack of liquidity. All gains and losses on these investments, realized and unrealized, are recognized in Other income, net on our Consolidated Statements of Income. + +We assess whether an impairment loss has occurred on our investments in non-marketable equity securities, accounted for under the measurement alternative based on quantitative and qualitative factors. If any impairment is identified for non-marketable equity securities, we write down the investment to its fair value and record the corresponding charge through Other income, net on our Consolidated Statements of Income. + +The Company assesses its investments for significant influence to determine the appropriate method of accounting, including application of the equity method. Equity method investments were not material. + +## Recently Issued Accounting Pronouncements + +## Recent Accounting Pronouncements Not Yet Adopted + +In November 2024, the Financial Accounting Standards Board, or FASB, issued a new accounting standard requiring disclosures of certain additional expense information on an annual and interim basis, including, among other items, the amounts of purchases of inventory, employee compensation, depreciation and intangible asset amortization included within each income statement expense caption, as applicable. We will adopt this standard in the fiscal year 2028 annual report. We do not expect the adoption of this standard to have a material impact on our Consolidated Financial Statements other than additional disclosures. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued) + +## Note 2 - Groq + +In December 2025, we entered into a non-exclusive license agreement with Groq, Inc., or Groq, for its language processing unit technology and hired certain Groq employees. No customer contracts, existing products, or equity interests were purchased. We recorded \$14.4 billion of goodwill and a \$2.5 billion developed technology intangible asset, valued using a cost-to-recreate methodology with a five-year useful life. Goodwill, primarily attributable to the workforce and future development of the licensed technology, was recorded in the Compute & Networking reporting unit. Total consideration consists of \$13.0 billion paid at closing and \$4 billion, inclusive of imputed interest, payable within one year included in Accrued and Other Current Liabilities on our Consolidated Balance Sheets. The goodwill is tax deductible. Pro forma results of operations have not been presented because the effect was not material. + +## Note 3 - Stock-Based Compensation + +We recognize stock-based compensation expense from grants of restricted stock units, or RSUs, performance stock units, or PSUs, and market-based PSUs, and issuances under our employee stock purchase plan, or ESPP. + +Consolidated Statements of Income include stock-based compensation expense as follows: + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Cost of revenue$ 261$ 178$ 141
Research and development4,6763,4232,532
Sales, general and administrative1,4491,136876
Total$ 6,386$ 4,737$ 3,549
+ +The following is a summary of equity awards granted under our equity incentive plans: + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions, except per share data)
RSUs, PSUs and Market-based PSUs
Awards granted7089140
Estimated total grant-date fair value$ 9,389$ 7,834$ 5,316
Weighted average grant-date fair value per share$ 133.97$ 87.99$ 37.41
ESPP
Shares purchased133030
Weighted average price per share$ 49.13$ 17.74$ 15.81
Weighted average grant-date fair value per share$ 20.75$ 8.61$ 6.99
+ +As of January 25, 2026, aggregate unearned stock-based compensation expense was \$14.8 billion, which is expected to be recognized over a weighted average period of 2.3 years for RSUs, PSUs, and market-based PSUs, and 0.9 years for ESPP. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +The fair value of shares issued under our ESPP has been estimated with the following assumptions: + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(Using the Black-Scholes model)
ESPP
Weighted average expected life (in years)0.1-2.00.1-2.00.1-2.0
Risk-free interest rate3.5%-4.3%3.6%-5.4%3.9%-5.5%
Volatility26%-96%31%-75%31%-67%
Dividend yield0.03%0.03%0.06%
+ +For ESPP shares, the expected term represents the average term from the first day of the offering period to the purchase date. The risk-free interest rate assumption used to value ESPP shares is based upon observed interest rates on Treasury bills appropriate for the expected term. Our expected stock price volatility assumption for ESPP is estimated using historical volatility. For awards granted, we use the dividend yield at grant date. Our RSUs, PSUs, and market-based PSUs are not eligible for cash dividends prior to vesting; therefore, the fair values of RSUs, PSUs, and market-based PSUs are discounted for the dividend yield. + +Additionally, for RSUs, PSUs, and market-based PSUs, we estimate expected forfeitures based on our historical forfeitures. + +## Equity Incentive Program + +We grant RSUs, PSUs, market-based PSUs, and stock purchase rights under the following equity incentive plans. In addition, in connection with our acquisitions of various companies, we have assumed certain stock-based awards granted under their stock incentive plans and converted them into our RSUs. + +## Amended and Restated 2007 Equity Incentive Plan + +The NVIDIA Corporation Amended and Restated 2007 Equity Incentive Plan, or the 2007 Plan, authorizes the issuance of incentive stock options, non-statutory stock options, restricted stock, RSUs, stock appreciation rights, performance stock awards, performance cash awards, and other stock-based awards to employees, directors and consultants. Only our employees may receive incentive stock options. We grant RSUs, PSUs and market-based PSUs under the 2007 Plan. As of January 25, 2026, up to 192 million shares of our common stock could be issued pursuant to stock awards granted under the 2007 Plan, and 1.3 billion shares were available for future grants. + +Subject to certain exceptions, RSUs vest generally over four years subject to continued service. PSUs vest over four years, subject to continued service and performance conditions. Market-based PSUs vest on approximately the third anniversary of the date of grant subject to market conditions. However, the number of shares subject to both PSUs and market-based PSUs that are eligible to vest is determined by the Compensation Committee based on achievement of predetermined criteria. + +## Amended and Restated 2012 Employee Stock Purchase Plan + +Employees who participate in the NVIDIA Corporation Amended and Restated 2012 Employee Stock Purchase Plan, or as most recently amended and restated, the 2012 Plan, may have up to 25% of their earnings withheld to purchase shares of common stock. The Board may decrease this percentage at its discretion. Each offering period is about 24 months, divided into four purchase periods of six months. The price of common stock purchased under our 2012 Plan will be equal to 85% of the lower of the fair market value of the common stock on the commencement date of each offering period or the fair market value of the common stock on each purchase date within the offering. As of January 25, 2026, we had 2.2 billion shares reserved for future issuance under the 2012 Plan. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +## Equity Award Activity + +The following is a summary of our equity award transactions under our equity incentive plans: + +
RSUs, PSUs and Market-based PSUs Outstanding
Number of SharesWeighted Average Grant-Date Fair Value Per Share
(In millions, except per share data)
Balance as of Jan 26, 2025274$ 44.75
Granted70$ 133.97
Vested(146)$ 39.14
Canceled and forfeited(9)$ 59.29
Balance as of Jan 25, 2026189$ 81.51
Vested and expected to vest after Jan 25, 2026188$ 81.15
+ +As of January 25, 2026 and January 26, 2025, there were 1.3 billion and 1.4 billion shares, respectively, of common stock available for future grants under our equity incentive plans. + +The total fair value of RSUs and PSUs, as of their respective vesting dates, during the years ended January 25, 2026, January 26, 2025, and January 28, 2024, was \$22.2 billion, \$15.1 billion, and \$8.2 billion, respectively. + +## Note 4 - Net Income Per Share + +The following is the basic and diluted net income per share computations for the periods presented: + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions, except per share data)
Numerator:
Net income$ 120,067$ 72,880$ 29,760
Denominator:
Basic weighted average shares24,35924,55524,690
Dilutive impact of outstanding equity awards155249250
Diluted weighted average shares24,51424,80424,940
Net income per share:
Basic (1)$ 4.93$ 2.97$ 1.21
Diluted (2)$ 4.90$ 2.94$ 1.19
Anti-dilutive equity awards excluded from diluted net income per share4151150
+ +(1) Net income divided by basic weighted average shares. +(2) Net income divided by diluted weighted average shares. + +## Note 5 - Goodwill + +As of January 25, 2026, the total carrying amount of goodwill was \$20.8 billion, consisting of goodwill balances allocated to our Compute & Networking and Graphics reporting units of \$20.5 billion and \$370 million, respectively. As of January 26, 2025, the total carrying amount of goodwill was \$5.2 billion, consisting of goodwill balances allocated to our Compute & Networking and Graphics reporting units of \$4.8 billion and \$370 million, respectively. Goodwill increased by \$15.6 billion in fiscal year 2026 and was allocated to our Compute & Networking reporting unit. During the fourth quarters of fiscal years 2026, 2025, and 2024, we completed our annual qualitative impairment tests and concluded that goodwill was not impaired. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +## Note 6 - Amortizable Intangible Assets + +The components of our amortizable intangible assets are as follows: + +
Jan 25, 2026Jan 26, 2025
Gross Carrying AmountAccumulated AmortizationNet Carrying AmountGross Carrying AmountAccumulated AmortizationNet Carrying Amount
(In millions)
Acquisition-related intangible assets$ 5,656$ (2,580)$ 3,076$ 2,900$ (2,264)$ 636
Patents and licensed technology528(298)230449(278)171
Total intangible assets$ 6,184$ (2,878)$ 3,306$ 3,349$ (2,542)$ 807
+ +Amortization expense associated with intangible assets for fiscal years 2026, 2025, and 2024 was \$488 million, \$593 million, and \$614 million, respectively. + +The following table outlines the estimated future amortization expense related to the net carrying amount of intangible assets as of January 25, 2026: + +
Future Amortization Expense(In millions)
Fiscal Year:
2027$ 923
2028729
2029592
2030511
2031468
2032 and thereafter83
Total$ 3,306
+ +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +## Note 7 - Cash Equivalents and Marketable Securities + +The fair values of our financial assets are determined using quoted market prices of identical assets or market prices of similar assets from active markets. We review fair value classification on a quarterly basis. The following is a summary of cash equivalents and marketable securities: + +Jan 25, 2026 + +
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable SecuritiesOther Assets
(In millions)
Debt securities issued by the U.S. TreasuryLevel 2$ 21,635$ 77$ (3)$ 21,709$ —$ 21,709$ —
Corporate debt securitiesLevel 215,41092(3)15,49934515,154
Debt securities issued by U.S. government agenciesLevel 22,15742,1612,161
Certificates of depositLevel 2110110110
Foreign government bondsLevel 24014141
Money market fundsLevel 17,8307,8307,830
Publicly-held equity securities (1) (2)Level 117,72612,8864,840
Total$ 47,182$ 174$ (6)$ 65,076$ 8,285$ 51,951$ 4,840
+ +(1) In the first quarter of fiscal year 2026, one investment was reclassified from non-marketable equity securities to marketable securities following public market trading. The balance as of January 25, 2026 includes \$10.5 billion of investments which are subject to short-term lock-up restrictions on the ability to sell. +(2) The long-term portion of marketable equity securities, which are subject to lock-up restrictions through December 2027 of \$4.8 billion as of January 25, 2026, is included in other assets. + +Publicly-held equity securities are subject to market price volatility. Net unrealized gains on investments in publicly-held equity securities held at period end were \$6.6 billion for fiscal year 2026. Net unrealized gains on investments in publicly-held equity securities held at period end were not significant for fiscal years 2025 and 2024. + +Net realized gains on investments in publicly-held equity securities sold were not significant for fiscal years 2026, 2025, and 2024, reflecting the difference between the sale proceeds and the carrying value of the equity securities at the beginning of the period or the purchase date, if later. + +## NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +Jan 26, 2025 + +
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable Securities
(In millions)
Corporate debt securitiesLevel 2$ 18,504$ 51$ (29)$ 18,526$ 2,071$ 16,455
Debt securities issued by the U.S. TreasuryLevel 216,74942(22)16,7691,80114,968
Debt securities issued by U.S. government agenciesLevel 22,7757(5)2,7772,777
Foreign government bondsLevel 217717713740
Certificates of depositLevel 2979797
Money market fundsLevel 13,7603,7603,760
Publicly-held equity securitiesLevel 1381381
Total$ 42,062$ 100$ (56)$ 42,487$ 7,866$ 34,621
+ +The following tables provide the breakdown of unrealized losses, aggregated by investment category and length of time that individual debt securities have been in a continuous loss position: + +
Jan 25, 2026Jan 26, 2025
Less than 12 MonthsLess than 12 Months
Estimated Fair ValueGross Unrealized LossEstimated Fair ValueGross Unrealized Loss
(In millions)
Debt securities issued by the U.S. Treasury$ 10,666$ (3)$ 6,315$ (22)
Corporate debt securities1,332(3)5,291(29)
Debt securities issued by U.S. government agencies1,134816(5)
Total$ 13,132$ (6)$ 12,422$ (56)
+ +Gross unrealized losses related to debt securities in a continuous loss position of twelve months or greater as of January 25, 2026 and January 26, 2025 were not significant. Gross unrealized losses are related to fixed income securities, driven primarily by changes in interest rates. + +The estimated fair value of debt securities included in cash equivalents and marketable securities are shown below by contractual maturity. + +
Jan 25, 2026
(In millions)
Less than one year$ 20,427
Due in 1 - 5 years19,093
Total$ 39,520
+ +## Note 8 - Non-marketable Equity Securities + +Our non-marketable equity securities are valued under the measurement alternative applying valuation methods based on observable transactions for similar investments of the same issuer and unobservable inputs such as volatility, expected + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +# (Continued) + +time to liquidity, risk free rate and security-specific rights and obligations. Gains and losses on these investments, realized and unrealized, are recognized in Other income, net on our Consolidated Statements of Income. + +Adjustments to the carrying value of our non-marketable equity securities during fiscal years 2026 and 2025 were as follows: + +
Year Ended
Jan 25, 2026Jan 26, 2025
(In millions)
Balance at beginning of period$ 3,387$ 1,321
Adjustments related to non-marketable equity securities:
Net additions17,4441,309
Unrealized gains2,369816
Reclassification (1)(848)
Impairments and unrealized losses(101)(59)
Balance at end of period$ 22,251$ 3,387
+ +(1) Represents reclassifications from non-marketable equity securities to marketable securities following public market trading. + +Non-marketable equity securities had cumulative gross unrealized gains of \$2.7 billion and \$1.1 billion, and cumulative gross unrealized losses and impairments of \$176 million and \$105 million on securities held as of January 25, 2026 and January 26, 2025, respectively. + +## Note 9 - Balance Sheet Components + +We refer to customers who purchase products directly from NVIDIA as direct customers, such as AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. Three direct customers accounted for 25%, 18%, and 13% of our accounts receivable balance as of January 25, 2026. Two direct customers accounted for 17% and 16% of our accounts receivable balance as of January 26, 2025. + +Certain balance sheet components are as follows: + +
Jan 25, 2026Jan 26, 2025
Inventories:(In millions)
Raw materials$ 3,807$ 3,408
Work in process8,8223,399
Finished goods8,7743,273
Total inventories (1)$ 21,403$ 10,080
+ +(1) In fiscal years 2026 and 2025, we recorded inventory provisions of \$4.0 billion and \$1.6 billion, respectively, in cost of revenue. + +
Jan 25, 2026Jan 26, 2025EstimatedUseful Life
Property and Equipment:(In millions)(In years)
Land$ 777$ 511(A)
Buildings, leasehold improvements, and furniture2,8912,076(B)
Equipment, compute hardware, and software12,6197,5682-7
Construction in process683529(C)
Total property and equipment, gross16,97010,684
Accumulated depreciation and amortization(6,587)(4,401)
Total property and equipment, net$ 10,383$ 6,283
+ +(A) Land is a non-depreciable asset. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +# (Continued) + +(B) The estimated useful lives of our buildings are up to thirty years. Leasehold improvements and finance leases are amortized based on the lesser of either the asset's estimated useful life or the expected remaining lease term. +(C) Construction in process represents assets that are not available for their intended use. + +Depreciation expense for fiscal years 2026, 2025, and 2024 was \$2.4 billion, \$1.3 billion, and \$894 million, respectively. + +Accumulated amortization of leasehold improvements and finance leases was \$519 million and \$410 million as of January 25, 2026 and January 26, 2025, respectively. + +Property, equipment and intangible assets acquired but not paid for during fiscal years 2026, 2025, and 2024 were \$820 million, \$525 million, and \$170 million, respectively. + +
Jan 25, 2026Jan 26, 2025
Accrued and Other Current Liabilities:(In millions)
Customer program accruals$ 5,318$ 4,880
Accrued purchase consideration3,9219
Product warranty2,8071,290
Excess inventory purchase obligations (1)2,7392,095
Taxes payable2,669881
Deferred revenue (2)1,379837
Accrued payroll and related expenses1,146848
Other1,373897
Total accrued and other current liabilities$ 21,352$ 11,737
+ +(1) In fiscal years 2026 and 2025, we recorded an expense of approximately \$3.2 billion and \$2.0 billion, respectively, in cost of revenue. +(2) Includes customer advances and unearned revenue related to hardware and software support, cloud services, and license and development arrangements. The balance as of January 25, 2026 and January 26, 2025 included \$160 million and \$81 million of customer advances, respectively. + +
Jan 25, 2026Jan 26, 2025
Other Long-Term Liabilities:(In millions)
Income tax payable (1)$ 3,958$ 2,188
Deferred income tax1,774886
Deferred revenue (2)1,193976
Other381195
Total other long-term liabilities$ 7,306$ 4,245
+ +(1) Primarily comprised of unrecognized tax benefits and related interest and penalties. +(2) Includes unearned revenue related to hardware and software support and cloud services. + +## Deferred Revenue + +The following table shows the changes in short- and long-term deferred revenue during fiscal years 2026 and 2025: + +
Jan 25, 2026Jan 26, 2025
(In millions)
Balance at beginning of period$ 1,813$ 1,337
Deferred revenue additions (1)11,1375,083
Revenue recognized (2)(10,378)(4,607)
Balance at end of period$ 2,572$ 1,813
+ +(1) Includes \$9.0 billion and \$3.6 billion of customer advances for fiscal years 2026 and 2025, respectively. +(2) Includes \$8.9 billion and \$3.7 billion related to customer advances for fiscal years 2026 and 2025, respectively. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +We recognized revenue of \$974 million and \$729 million in fiscal years 2026 and 2025, respectively, that were included in the prior year end deferred revenue balance. + +As of January 25, 2026, revenue related to remaining performance obligations from contracts greater than one year in length was \$2.3 billion, which includes \$1.9 billion from deferred revenue and \$390 million which has not yet been billed nor recognized as revenue. Approximately 42% of revenue from contracts greater than one year in length will be recognized over the next twelve months. + +## Note 10 - Derivative Financial Instruments + +## Foreign Currency Derivatives + +We utilize foreign currency forward contracts to mitigate the impact of foreign currency exchange rate movements on our operating expenses. The foreign currency forward contracts for operating expenses are designated as accounting hedges. Gains or losses on the contracts are recorded in accumulated other comprehensive income or loss and reclassified to operating expense when the related operating expenses are recognized in earnings. In fiscal years 2026 and 2025, the impact of foreign currency forward contracts designated as accounting hedges on other comprehensive income or loss was not significant and all such instruments were determined to be highly effective. + +We also entered into foreign currency forward contracts mitigating the impact of foreign currency movements on monetary assets and liabilities. For our foreign currency contracts for assets and liabilities, the change in fair value of these non-designated contracts was recorded in other income or expense and offsets the change in fair value of the hedged foreign currency denominated monetary assets and liabilities, which was also recorded in other income or expense. + +The table below presents the notional value of our foreign currency contracts outstanding: + +
Jan 25, 2026Jan 26, 2025
(In millions)
Designated as accounting hedges$ 1,765$ 1,424
Not designated as accounting hedges$ 2,332$ 1,297
+ +The unrealized gains and losses or fair value of our foreign currency contracts were not significant as of January 25, 2026 and January 26, 2025. + +As of January 25, 2026, all foreign currency contracts mature within 18 months. The expected realized gains and losses deferred into accumulated other comprehensive income or loss related to foreign currency forward contracts within the next twelve months were not significant. + +## Facility Lease Guarantees + +In fiscal year 2026, we entered into agreements to guarantee partners' facility lease obligations in the event of their default in exchange for warrants. The maximum gross exposure under all agreements is \$3.5 billion, which is reduced as the partners make payments to the lessors over terms ranging from 5 to 7 years. The partners have placed \$712 million in escrow to mitigate our potential exposure. The guarantees, classified as credit derivatives with changes in fair value recognized in Other income and expense, were not material. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +Note 11 - Debt + +
Expected Remaining Term (years)Effective Interest RateJan 25, 2026Jan 26, 2025
(In millions)
3.20% Notes Due 20260.63.31%1,0001,000
1.55% Notes Due 20282.41.64%1,2501,250
2.85% Notes Due 20304.22.93%1,5001,500
2.00% Notes Due 20315.42.09%1,2501,250
3.50% Notes Due 204014.23.54%1,0001,000
3.50% Notes Due 205024.23.54%2,0002,000
3.70% Notes Due 206034.23.73%500500
Unamortized debt discount and issuance costs(32)(37)
Net carrying amount8,4688,463
Less short-term portion(999)
Total long-term portion$ 7,469$ 8,463
+ +As of January 25, 2026 and January 26, 2025, the estimated fair value of debt was \$7.5 billion and \$7.2 billion, respectively. The estimated fair values are based on Level 2 inputs. + +Our notes are unsecured senior obligations. Existing and future liabilities of our subsidiaries will be effectively senior to the notes. Our notes pay interest semi-annually. We may redeem each of our notes prior to maturity, subject to a make-whole premium. The maturity of the notes is calendar year. + +As of January 25, 2026, we complied with the required covenants, which are non-financial in nature, under the outstanding notes. + +In January 2026, we increased the size of our commercial paper program from \$575 million to \$25.0 billion. As of January 25, 2026, no commercial paper was outstanding. + +## Note 12 - Commitments and Contingencies + +## Commitments + +Manufacturing, supply, and capacity commitments reflect datacenter-scale production and longer future ordering horizons across current and future product architectures. We enter into agreements with our supply vendors that allow them to procure inventory based upon our defined criteria, and in certain instances, these agreements are cancellable, able to be rescheduled, or adjustable for our business needs prior to placing firm orders. Changes to these agreements may result in additional costs. As of January 25, 2026, these commitments were \$95.2 billion, of which substantially all will be paid through fiscal year 2027. + +Multi-year cloud service agreement commitments as of January 25, 2026, were \$27 billion, for which \$7 billion, \$6 billion, \$5 billion, \$5 billion, \$2 billion, and \$2 billion will be paid in fiscal years 2027, 2028, 2029, 2030, 2031, and 2032 and thereafter, respectively. Some cloud service capacity may be reduced, terminated or sold to others by the CSPs, in which case our commitments will be reduced. We expect cloud service agreements to be used to support our research and development efforts. + +Investment commitments are \$11.4 billion as of January 25, 2026, subject to certain contingencies, of which we expect substantially all will be made through fiscal year 2027. + +Other commitments were \$3.4 billion as of January 25, 2026, of which the majority will be paid through fiscal year 2027. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued) + +## Accrual for Product Warranty Liabilities + +The estimated amount of product warranty liabilities was \$2.8 billion and \$1.3 billion as of January 25, 2026 and January 26, 2025, respectively. The estimated product returns and product warranty activity consisted of the following: + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Balance at beginning of period$ 1,290$ 306$ 82
Additions2,4741,203278
Utilization(957)(219)(54)
Balance at end of period$ 2,807$ 1,290$ 306
+ +In fiscal years 2026, 2025, and 2024 the additions in product warranty liabilities primarily related to our Compute & Networking segment. + +We have provided indemnities for matters such as tax, product, and employee liabilities. We have included intellectual property indemnification provisions in our technology-related agreements with third parties. Maximum potential future payments cannot be estimated because many of these agreements do not have a maximum stated liability. We have not recorded any liability in our Consolidated Financial Statements for such indemnifications. + +## Litigation + +## Securities Class Action and Derivative Lawsuits + +The plaintiffs in the putative securities class action lawsuit, captioned 4:18-cv-07669-HSG, initially filed on December 21, 2018 in the United States District Court for the Northern District of California, and titled In Re NVIDIA Corporation Securities Litigation, filed an amended complaint on May 13, 2020. The amended complaint asserted that NVIDIA and certain NVIDIA executives violated Section 10(b) of the Securities Exchange Act of 1934, as amended, or the Exchange Act, and SEC Rule 10b-5, by making materially false or misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand between May 10, 2017 and November 14, 2018. Plaintiffs also alleged that the NVIDIA executives who they named as defendants violated Section 20(a) of the Exchange Act. Plaintiffs sought class certification, an award of unspecified compensatory damages, an award of reasonable costs and expenses, including attorneys' fees and expert fees, and further relief as the Court may deem just and proper. On March 2, 2021, the district court granted NVIDIA's motion to dismiss the complaint without leave to amend, entered judgment in favor of NVIDIA and closed the case. On March 30, 2021, plaintiffs filed an appeal from judgment in the United States Court of Appeals for the Ninth Circuit, case number 21-15604. On August 25, 2023, a majority of a three-judge Ninth Circuit panel affirmed in part and reversed in part the district court's dismissal of the case, with a third judge dissenting on the basis that the district court did not err in dismissing the case. On November 15, 2023, the Ninth Circuit denied NVIDIA's petition for rehearing en banc of the Ninth Circuit panel's majority decision to reverse in part the dismissal of the case, which NVIDIA had filed on October 10, 2023. On December 5, 2023, the Ninth Circuit granted NVIDIA's motion to stay the mandate pending NVIDIA's petition for a writ of certiorari in the Supreme Court of the United States and the Supreme Court's final disposition of the matter. NVIDIA filed a petition for a writ of certiorari on March 4, 2024. On June 17, 2024, the Supreme Court of the United States granted NVIDIA's petition for a writ of certiorari. After briefing and argument, the Supreme Court dismissed NVIDIA's writ of certiorari as improvidently granted on December 11, 2024, and issued judgment on January 13, 2025. On February 20, 2025, the Ninth Circuit's judgment, entered August 25, 2023 and corrected August 28, 2023, took effect, and the case was remanded to the district court for further proceedings. + +The putative derivative lawsuit pending in the United States District Court for the Northern District of California, captioned 4:19-cv-00341-HSG, initially filed January 18, 2019 and titled In re NVIDIA Corporation Consolidated Derivative Litigation, was stayed pending resolution of the plaintiffs' appeal in the In Re NVIDIA Corporation Securities Litigation action. On February 22, 2022, the court administratively closed the case, but stated that it would reopen the case once the appeal in the In Re NVIDIA Corporation Securities Litigation action is resolved. The case has not yet been reopened by the court. The lawsuit asserts claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty, unjust enrichment, waste of corporate assets, and violations of Sections 14(a), 10(b), and 20(a) of the Exchange Act based on the dissemination of allegedly false and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs are seeking unspecified damages and other relief, including reforms and improvements to NVIDIA's corporate governance and internal procedures. + +The putative derivative actions initially filed September 24, 2019 and pending in the United States District Court for the District of Delaware, Lipchitz v. Huang, et al. (Case No. 1:19-cv-01795-MN) and Nelson v. Huang, et al. (Case No. 1:19-cv-01798-MN), were stayed pending resolution of the plaintiffs' appeal in the In Re NVIDIA Corporation Securities + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +Litigation action. On March 7, 2025, after the Supreme Court issued its judgment dismissing the Company's petition for writ of certiorari as improvidently granted in the In Re NVIDIA Securities Litigation action, the district court adopted the parties' stipulation to extend the stay until the final and complete resolution of the In Re NVIDIA Corporation Securities Litigation action. The lawsuits assert claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty, unjust enrichment, insider trading, misappropriation of information, corporate waste and violations of Sections 14(a), 10(b), and 20(a) of the Exchange Act based on the dissemination of allegedly false, and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs seek unspecified damages and other relief, including disgorgement of profits from the sale of NVIDIA stock and unspecified corporate governance measures. + +Another putative derivative action was filed on October 30, 2023 in the Court of Chancery of the State of Delaware, captioned Horanic v. Huang, et al. (Case No. 2023-1096-KSJM). This lawsuit asserts claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty and insider trading based on the dissemination of allegedly false and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs seek unspecified damages and other relief, including disgorgement of profits from the sale of NVIDIA stock and reform of unspecified corporate governance measures. On August 11, 2025, the court granted the parties' stipulation to voluntarily dismiss with prejudice plaintiff City of Westland Police and Fire Retirement System. This derivative matter is stayed pending the final resolution of In Re NVIDIA Corporation Securities Litigation action. + +## Accounting for Loss Contingencies + +As of January 25, 2026, there are no accrued contingent liabilities associated with the legal proceedings described above based on our belief that liabilities, while reasonably possible, are not probable. Further, any possible loss or range of loss in these matters cannot be reasonably estimated at this time. We are engaged in legal actions not described above arising in the ordinary course of business, as well as regulatory and government inquiries and investigations, and, while there can be no assurance of favorable outcomes, we believe that the ultimate outcome of these matters will not have a material adverse effect on our operating results, liquidity or financial position. These matters are subject to inherent uncertainties and if the ultimate outcome is unfavorable, there exists the possibility of a material adverse impact on our operating results, liquidity or financial position in the period the outcome becomes estimable and probable. + +## Note 13 - Income Taxes + +The FASB issued a new accounting standard which includes new and updated income tax disclosures, including disaggregation of information in the rate reconciliation and income taxes paid, which we adopted on a prospective basis for the year ending January 25, 2026. + +The Income tax expense applicable to income before income taxes consists of the following: + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Current income taxes:
Federal$ 19,039$ 14,032$ 5,710
State1,218892335
Foreign2,550699502
Total current22,80715,6236,547
Deferred income taxes:
Federal(1,364)(4,515)(2,499)
State(885)(242)(206)
Foreign825280216
Total deferred(1,424)(4,477)(2,489)
Income tax expense$ 21,383$ 11,146$ 4,058
+ +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +Income before income tax consists of the following: + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
U.S.$ 123,181$ 77,456$ 29,495
Foreign18,2696,5704,323
Income before income tax$ 141,450$ 84,026$ 33,818
+ +The income tax expense (benefit) differs from the amount computed by applying the U.S. federal statutory rate of 21.0% to income before income taxes for the fiscal year ended January 25, 2026 as follows: + +
Year Ended
Jan 25, 2026
(In millions, except percentages)
US Federal Statutory Tax Rate$ 29,70421.0 %
State and Local Income Taxes, Net of Federal Income Tax Effect (1)2580.2 %
Foreign tax effects
Israel
Reduced statutory tax rate on qualifying income(3,064)(2.2)%
Other1,6061.2 %
Other foreign jurisdictions7410.5 %
Effect of cross-border tax laws
Foreign-derived deduction eligible income(4,208)(3.0)%
Other(142)(0.1)%
Tax credits(1,933)(1.4)%
Nontaxable or nondeductible items
Stock-based compensation(1,475)(1.0)%
Other29— %
Other (2)(133)(0.1)%
Income tax expense$ 21,38315.1 %
+ +(1) State taxes in California, Tennessee, Arizona, and Illinois made up the majority of the tax effect in fiscal year 2026. +(2) Includes the tax effects of enactment of new tax laws, change in valuation allowance, and change in unrecognized tax benefits. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +The income tax expense (benefit) differs from the amount computed by applying the U.S. federal statutory rate of 21% to income before income taxes for fiscal years ended January 26, 2025 and January 28, 2024 as follows: + +
Year Ended
Jan 26, 2025Jan 28, 2024
(In millions, except percentages)
Tax expense computed at federal statutory rate$ 17,64521.0 %$ 7,10221.0 %
Expense (benefit) resulting from:
State income taxes, net of federal tax effect5540.7 %1200.4 %
Foreign-derived deduction eligible income(2,976)(3.5)%(1,408)(4.2)%
Stock-based compensation(2,097)(2.5)%(741)(2.2)%
U.S. federal research and development tax credit(990)(1.2)%(431)(1.3)%
Foreign tax rate differential(984)(1.2)%(467)(1.4)%
Other(6)— %(117)(0.3)%
Income tax expense$ 11,14613.3 %$ 4,05812.0 %
+ +In July 2025, the OBBBA was enacted into law and contains several changes to key U.S. federal income tax laws. We have recognized the tax effects of currently effective OBBBA provisions in our results for fiscal year 2026. + +The amount of cash paid for income taxes (net of refunds) for the fiscal year ended January 25, 2026 is as follows: + +
Year Ended
Jan 25, 2026
(In millions)
Federal$ 16,755
State
California1,049
Other1,041
Foreign
Israel1,287
Other156
Total income taxes paid, net of refunds$ 20,288
+ +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +The tax effect of temporary differences that gives rise to significant portions of the deferred tax assets and liabilities are presented below: + +
Jan 25, 2026Jan 26, 2025
(In millions)
Deferred tax assets:
Capitalized research and development expenditure$ 5,436$ 6,256
Net controlled foreign corporation tested income deferred tax assets5,3892,820
Accruals and reserves, not currently deductible for tax purposes3,6442,058
Research and other tax credit carryforwards718759
Operating lease liabilities554299
Net operating loss and capital loss carryforwards443456
Other deferred tax assets679566
Gross deferred tax assets16,86313,214
Less valuation allowance(768)(1,610)
Total deferred tax assets16,09511,604
Deferred tax liabilities:
Equity investments(2,227)(264)
Unremitted earnings of foreign subsidiaries(1,813)(891)
Operating lease assets(533)(286)
Acquired intangibles(38)(70)
Gross deferred tax liabilities(4,611)(1,511)
Net deferred tax asset (1)$ 11,484$ 10,093
+ +(1) Net deferred tax asset includes long-term deferred tax assets of \$13.3 billion and \$11.0 billion and long-term deferred tax liabilities of \$1.8 billion and \$886 million for fiscal years 2026 and 2025, respectively. Long-term deferred tax liabilities are included in other long-term liabilities on our Consolidated Balance Sheets. + +As of January 25, 2026, we intend to indefinitely reinvest approximately \$1.4 billion of cumulative undistributed earnings held by certain subsidiaries. We have not provided the amount of unrecognized deferred tax liabilities for temporary differences related to these investments as the determination of such amount is not practicable. + +As of January 25, 2026 and January 26, 2025, we had a valuation allowance of \$768 million and \$1.6 billion, respectively, related to capital loss carryforwards, and certain other deferred tax assets that management determined are not likely to be realized due, in part, to jurisdictional projections of future taxable income, including capital gains. To the extent realization of the deferred tax assets becomes more-likely-than-not, we would recognize such deferred tax assets as income tax benefits during the period. + +As of January 25, 2026, based on recent jurisdictional taxable income and expected future earnings, we concluded certain state deferred tax assets are more likely than not realizable and released \$711 million of valuation allowance. + +As of January 25, 2026, we had U.S. federal, state and foreign net operating loss carryforwards of \$747 million, \$427 million and \$503 million, respectively. The federal and state carryforwards will begin to expire in fiscal year 2027. The foreign net operating loss carryforwards may be carried forward indefinitely. As of January 25, 2026, we had federal research tax credit carryforwards of \$56 million, before the impact of uncertain tax positions, that will begin to expire in fiscal year 2027. We have state research tax credit carryforwards of \$1.4 billion, before the impact of uncertain tax positions, of which \$1.3 billion is attributable to the State of California and may be carried over indefinitely and \$132 million is attributable to various other states and will begin to expire in fiscal year 2028. As of January 25, 2026, we had federal capital loss carryforwards of \$902 million that will begin to expire in fiscal year 2028. + +Our tax attributes remain subject to audit and may be adjusted for changes or modification in tax laws, other authoritative interpretations thereof, or other facts and circumstances. Utilization of tax attributes may also be subject to limitations due to ownership changes and other limitations provided by the Internal Revenue Code and similar state and foreign tax provisions. If any such limitations apply, the tax attributes may expire or be denied before utilization. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements (Continued) + +A reconciliation of gross unrecognized tax benefits is as follows: + +
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Balance at beginning of period$ 2,861$ 1,670$ 1,238
Increases in tax positions for current year1,9591,268616
Increases in tax positions for prior years574887
Lapse in statute of limitations(224)(27)(19)
Decreases in tax positions for prior years(157)(88)(148)
Settlements(76)(10)(104)
Balance at end of period$ 4,420$ 2,861$ 1,670
+ +Included in the balance of unrecognized tax benefits as of January 25, 2026 are \$3.7 billion of tax benefits that would affect our effective tax rate if recognized. + +We classify an unrecognized tax benefit as a current liability, or amount refundable, to the extent that we anticipate payment or receipt of cash for income taxes within one year. The amount is classified as a long-term liability, or long-term amount refundable, if we anticipate payment or receipt of cash for income taxes during a period beyond a year. + +We include interest and penalties related to unrecognized tax benefits as a component of income tax expense. We recognized net interest and penalties related to unrecognized tax benefits in the income tax expense line of our consolidated statements of income of \$103 million, \$92 million, and \$42 million during fiscal years 2026, 2025, and 2024, respectively. As of January 25, 2026 and January 26, 2025, we have accrued \$374 million and \$251 million, respectively, for the payment of interest and penalties related to unrecognized tax benefits, which is not included as a component of our gross unrecognized tax benefits. + +We are subject to examination by taxing authorities both in the United States and other countries. As of January 25, 2026, the significant tax jurisdictions that may be subject to examination include the United States for fiscal years after 2022, as well as Canada, China, Germany, Hong Kong, India, Israel, Italy, and Taiwan for fiscal years 2014 through 2025. As of January 25, 2026, the significant tax jurisdictions for which we are currently under examination include the United States, Germany, Hong Kong, India, Israel, and Taiwan for fiscal years 2014 through 2025. + +## Note 14 - Shareholders' Equity + +## Capital Return Program + +On August 26, 2025, our Board of Directors approved an additional \$60.0 billion in share repurchase authorization, without expiration. In fiscal years 2026 and 2025, we repurchased 282 million and 310 million shares of our common stock for \$40.4 billion and \$34.0 billion, respectively. As of January 25, 2026, we were authorized, subject to certain specifications, to repurchase up to \$58.5 billion of our common stock. + +From January 26, 2026 through February 20, 2026, we repurchased 8 million shares for \$1.5 billion pursuant to a pre-established trading plan. + +In fiscal years 2026, 2025, and 2024, we paid cash dividends to our shareholders of \$974 million, \$834 million, and \$395 million, respectively. The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders. + +## Note 15 - Employee Retirement Plans + +We provide tax-qualified defined contribution plans to eligible employees in the U.S. and certain other countries. Our contribution expense for fiscal years 2026, 2025, and 2024 was \$442 million, \$314 million, and \$255 million, respectively. + +## Note 16 - Segment Information + +Our Chief Executive Officer is our chief operating decision maker, or CODM, and reviews financial information presented on an operating segment basis for purposes of making decisions and assessing financial performance. Our CODM assesses operating performance of each segment based on regularly provided segment revenue and segment operating income. Operating results by segment include costs or expenses directly attributable to each segment, and costs or expenses that are leveraged across our unified architecture and therefore allocated between our two segments. Our + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +CODM reviews expenses on a consolidated basis, and expenses attributable to each segment are not regularly provided to our CODM. + +The Compute & Networking segment includes our Data Center accelerated computing and networking platforms and AI solutions and software, and Automotive platforms and autonomous and electric vehicle solutions including software. + +The Graphics segment includes GeForce GPUs for gaming and PCs, and Quadro/NVIDIA RTX GPUs for enterprise workstation graphics. + +Certain expenses are not allocated to either Compute & Networking or Graphics for purposes of making operating decisions or assessing financial performance. The expenses include stock-based compensation expense, corporate infrastructure and support costs, acquisition-related and other costs, and other non-recurring charges and benefits that our CODM deems to be enterprise in nature. + +Our CODM does not review any information regarding total assets on a reportable segment basis. There are no intersegment transactions. The accounting policies for segment reporting are the same as for our consolidated financial statements. The table below presents details of our reportable segments. + +
Compute & NetworkingGraphicsTotal
(In millions)
Year Ended Jan 25, 2026
Revenue$ 193,479$ 22,459$ 215,938
Other segment items (1)63,33813,30376,641
Operating income$ 130,141$ 9,156$ 139,297
Year Ended Jan 26, 2025
Revenue$ 116,193$ 14,304$ 130,497
Other segment items (1)33,3189,21942,537
Operating income$ 82,875$ 5,085$ 87,960
Year Ended Jan 28, 2024
Revenue$ 47,405$ 13,517$ 60,922
Other segment items (1)15,3897,67123,060
Operating income$ 32,016$ 5,846$ 37,862
+ +(1) Other segment items primarily include product costs and inventory provisions, compensation and benefits excluding stock-based compensation expense, computing infrastructure expenses, and engineering development costs. + +Depreciation and amortization expense attributable to our Compute & Networking segment for fiscal years 2026, 2025, and 2024 was \$1.6 billion, \$732 million, and \$457 million, respectively. Depreciation and amortization expense attributable to our Graphics segment for fiscal years 2026, 2025, and 2024 was \$590 million, \$372 million, and \$307 million, respectively. Acquisition-related intangible amortization expense is not allocated to either Compute & Networking or Graphics for purposes of making operating decisions or assessing financial performance. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +A reconciliation of segment operating income to consolidated income before income tax for fiscal years 2026, 2025, and 2024 were as follows: + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Segment operating income$ 139,297$ 87,960$ 37,862
Stock-based compensation expense(6,386)(4,737)(3,549)
Unallocated operating expenses(1,997)(1,171)(728)
Acquisition-related and other costs(527)(599)(613)
Interest income2,3001,786866
Interest expense(259)(247)(257)
Other income, net9,0221,034237
Consolidated income before income tax$ 141,450$ 84,026$ 33,818
+ +Revenue by geographic area is based upon the location of the customers' headquarters. The end customer and shipping location may be different from our customers' headquarters location. + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Geographic Location (1):Revenue based upon Customer Headquarters(In millions)
United States$149,617$ 77,482$ 31,533
Taiwan (2)42,34523,60014,912
China (including Hong Kong)19,67725,04812,330
Other4,2994,3672,147
Total revenue$215,938$ 130,497$ 60,922
+ +(1) In the third quarter of fiscal year 2026, we changed to revenue based upon the location of our customers' headquarters as we believe it provides a better representation of the geographic profile of our revenue. Prior period information has been recast to reflect this change. +(2) In fiscal year 2026, we estimate 76% of Data Center revenue from Taiwan-headquartered customers was attributed to end customers based in the United States and Europe. + +Revenue from sales to customers headquartered outside of the United States accounted for 31%, 41%, and 48% of total revenue for fiscal years 2026, 2025, and 2024, respectively. The increase in revenue to the United States for fiscal years 2026 and 2025 was primarily due to higher U.S.-based Compute & Networking segment demand. + +We refer to customers who purchase products directly from NVIDIA as direct customers, such as AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. We refer to indirect customers as those who purchase products through our direct customers; indirect customers include CSPs, Neocloud builders, AI model makers, enterprises, and public sector entities. Our revenue is concentrated among a limited number of direct and indirect customers and this trend may continue. + +Direct Customers – For fiscal year 2026, sales to one direct customer represented 22% of total revenue and sales to another direct customer represented 14% of total revenue, all of which were primarily attributable to the Compute & Networking segment. + +For fiscal year 2025, sales to one direct customer represented 12% of total revenue and sales to two direct customers each represented 11% of total revenue, all of which were primarily attributable to the Compute & Networking segment. + +For fiscal year 2024, sales to one direct customer represented 13% of total revenue, and were primarily attributable to the Compute & Networking segment. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +The following table summarizes revenue by specialized markets: + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Revenue by End Market:(In millions)
Data Center$ 193,737$ 115,186$ 47,525
Compute162,361102,19638,950
Networking31,37612,9908,575
Gaming16,04211,35010,447
Professional Visualization3,1911,8781,553
Automotive2,3491,6941,091
OEM and Other619389306
Total revenue$ 215,938$ 130,497$ 60,922
+ +The following table presents summarized information for long-lived assets by country. Long-lived assets consist of property and equipment and exclude other assets, operating lease assets, goodwill, and intangible assets. + +
Jan 25, 2026Jan 26, 2025
Long-lived assets:(In millions)
United States$ 5,125$ 3,626
Taiwan3,2191,481
Israel1,471840
Other568336
Total long-lived assets$ 10,383$ 6,283
+ +## Note 17 - Leases + +Our lease obligations primarily consist of operating leases for our offices and data centers, with lease periods expiring between fiscal years 2027 and 2041. + +Future minimum lease obligations under our non-cancelable lease agreements as of January 25, 2026 were as follows: + +
Operating Lease Obligations
(In millions)
Fiscal Year:
2027$ 493
2028485
2029457
2030381
2031314
2032 and thereafter1,494
Total3,624
Less imputed interest680
Present value of net future minimum lease payments2,944
Less short-term operating lease liabilities372
Long-term operating lease liabilities$ 2,572
+ +Between fiscal years 2027 and 2030, we expect to commence leases with future obligations of \$22.7 billion, primarily data center leases to support our research and development efforts, with lease terms of 1.8 to 20 years. + +# NVIDIA Corporation and Subsidiaries Notes to the Consolidated Financial Statements + +(Continued) + +Operating lease costs for fiscal years 2026, 2025, and 2024 were \$462 million, \$356 million, and \$269 million, respectively. Short-term and variable lease costs for fiscal years 2026, 2025, and 2024 were not significant. + +Other information related to leases was as follows: + +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Supplemental cash flows information
Operating cash flow used for operating leases$ 428$ 313$ 286
Operating lease assets obtained in exchange for lease obligations$ 1,439$ 877$ 531
+ +As of January 25, 2026, our operating leases have a weighted average remaining lease term of 8.8 years and a weighted average discount rate of 4.38%. As of January 26, 2025, our operating leases had a weighted average remaining lease term of 6.5 years and a weighted average discount rate of 4.16%. + +NVIDIA Corporation and Subsidiaries +Schedule II – Valuation and Qualifying Accounts + +
DescriptionBalance at Beginning of PeriodAdditionsDeductionsBalance at End of Period
(In millions)
Fiscal year 2026
Allowance for doubtful accounts$ 4$ — (1)$ — (1)$ 4
Sales return allowance$ 82$ 188 (2)$ (100) (4)$ 170
Deferred tax valuation allowance$ 1,610$ 31 (3)$ (873) (3)$ 768
Fiscal year 2025
Allowance for doubtful accounts$ 4$ — (1)$ — (1)$ 4
Sales return allowance$ 109$ 151 (2)$ (178) (4)$ 82
Deferred tax valuation allowance$ 1,552$ 58 (3)$ — (3)$ 1,610
Fiscal year 2024
Allowance for doubtful accounts$ 4$ — (1)$ — (1)$ 4
Sales return allowance$ 26$ 213 (2)$ (130) (4)$ 109
Deferred tax valuation allowance$ 1,484$ 162 (3)$ (94) (3)$ 1,552
+ +(1) Additions represent either expense or acquired balances and deductions represent write-offs. +(2) Additions represent estimated product returns charged as a reduction to revenue or an acquired balance. +(3) Additional valuation allowance on deferred tax assets not likely to be realized. Additions represent additional valuation allowance on certain state and other deferred tax assets. Deductions mainly represent the release of valuation allowance on certain state deferred tax assets. Refer to Note 13 of the Notes to the Consolidated Financial Statements in Part IV, Item 15 of this Annual Report on Form 10-K for additional information. +(4) Represents sales returns. + +Exhibit Index + +
Exhibit No.Exhibit DescriptionIncorporated by ReferenceFiling Date
Schedule/FormExhibit
3.1Restated Certificate of Incorporation10-K3.13/18/2022
3.2Amendment to Restated Certificate of Incorporation of NVIDIA Corporation8-K3.16/6/2022
3.3Amendment to Restated Certificate of Incorporation of NVIDIA Corporation8-K3.16/7/2024
3.4Bylaws of NVIDIA Corporation, Amended and Restated as of March 12, 20248-K3.13/14/2024
4.1Reference is made to Exhibits 3.1, 3.2, 3.3 and 3.4
4.2Specimen Stock CertificateS-1/A4.24/24/1998
4.3Indenture, dated as of September 16, 2016, by and between the Company and Computershare Trust Company, N.A., as successor to Wells Fargo Bank, National Association, as Trustee8-K4.19/16/2016
4.4Officers' Certificate, dated as of September 16, 20168-K4.29/16/2016
4.5Form of 2026 Note8-KAnnex B-1 to Exhibit 4.29/16/2016
4.6Description of Securities10-K4.62/26/2025
4.7Officers' Certificate, dated as of March 31, 20208-K4.23/31/2020
4.8Form of 2030 Note8-KAnnex A-1 to Exhibit 4.23/31/2020
4.9Form of 2040 Note8-KAnnex B-1 to Exhibit 4.23/31/2020
4.10Form of 2050 Note8-KAnnex C-1 to Exhibit 4.23/31/2020
4.11Form of 2060 Note8-KAnnex D-1 to Exhibit 4.23/31/2020
4.12Officers' Certificate, dated as of June 16, 20218-K4.26/16/2021
4.13Form of 2028 Note8-KAnnex C-1 to Exhibit 4.26/16/2021
4.14Form of 2031 Note8-KAnnex D-1 to Exhibit 4.26/16/2021
10.1Form of Indemnity Agreement between NVIDIA Corporation and each of its directors and officers8-K10.13/7/2006
10.2+Amended and Restated 2007 Equity Incentive Plan10-Q10.18/28/2024
10.3+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Deferred Restricted Stock Unit Grant Notice and Deferred Restricted Stock Unit Agreement (2016)10-K10.263/12/2015
10.4+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Restricted Stock Unit Grant Notice and Restricted Stock Unit Agreement (2016)10-K10.273/12/2015
10.5+Amended and Restated 2007 Equity Incentive Plan - Global Performance-Based Restricted Stock Unit Grant Notice and Performance-Based Restricted Stock Unit Agreement (2019)8-K10.13/11/2019
10.6+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2021)10-Q10.25/26/2021
10.7+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2022)10-K10.163/18/2022
10.8+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2023)10-K10.142/24/2023
10.9+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2024) (version 1)10-Q10.25/29/2024
10.10+Amended and Restated 2007 Equity Incentive Plan - Global Performance-Based Restricted Stock Unit Grant Notice and Performance-Based Restricted Stock Unit Agreement (2024)10-Q10.35/29/2024
10.11+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2024) (version 2)10-Q10.111/20/2024
10.12+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2025)10-K10.132/26/2025
10.13+Amended and Restated 2007 Equity Incentive Plan - Global Performance-Based Restricted Stock Unit Grant Notice and Performance-Based Restricted Stock Unit Agreement (2025)10-K10.142/26/2025
10.14+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Deferred Restricted Stock Unit Grant Notice and Deferred Restricted Stock Unit Agreement (2025)10-Q10.18/27/2025
10.15+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Restricted Stock Unit Grant Notice and Restricted Stock Unit Agreement (2025)10-Q10.28/27/2025
10.16+Amended and Restated 2012 Employee Stock Purchase Plan10-K10.152/26/2025
10.17+Variable Compensation Plan - Fiscal Year 20258-K10.13/14/2024
10.18+Variable Compensation Plan - Fiscal Year 20268-K10.13/7/2025
19.1*NVIDIA Corporation Insider Trading Policy
21.1*Subsidiaries of Registrant
23.1*Consent of PricewaterhouseCoopers LLP
24.1*Power of Attorney (included in signature page)
31.1*Certification of Chief Executive Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
31.2*Certification of Chief Financial Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
32.1#*Certification of Chief Executive Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
32.2#*Certification of Chief Financial Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
97.1+Compensation Recovery Policy, as amended and restated November 30, 202310-K97.12/21/2024
101.INS*XBRL Instance Document
101.SCH*XBRL Taxonomy Extension Schema Document
101.CAL*XBRL Taxonomy Extension Calculation Linkbase Document
101.DEF*XBRL Taxonomy Extension Definition Linkbase Document
101.LAB*XBRL Taxonomy Extension Labels Linkbase Document
101.PRE*XBRL Taxonomy Extension Presentation Linkbase Document
104Cover Page Interactive Data File - the cover page interactive data file does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document
+ +\* Filed herewith. ++ Management contract or compensatory plan or arrangement. +# In accordance with Item 601(b)(32)(ii) of Regulation S-K and SEC Release Nos. 33-8238 and 34-47986, Final Rule: Management's Reports on Internal Control Over Financial Reporting and Certification of Disclosure in Exchange Act Periodic Reports, the certifications furnished in Exhibits 32.1 and 32.2 hereto are deemed to accompany this Annual Report on Form 10-K and will not be deemed "filed" for purpose of Section 18 of the Exchange Act. Such certifications will not be deemed to be incorporated by reference into any filing under the Securities Act or the Exchange Act, except to the extent that the registrant specifically incorporates it by reference. +^ Certain exhibits and schedules have been omitted in accordance with Regulation S-K Item 601(a)(5). + +Copies of above exhibits not contained herein are available to any shareholder upon written request to: + +Investor Relations: NVIDIA Corporation, 2788 San Tomas Expressway, Santa Clara, CA 95051 + +## Item 16. Form 10-K Summary + +Not Applicable. + +## Signatures + +Pursuant to the requirements of Section 13 or 15(d) of the Securities Exchange Act of 1934, the Registrant has duly caused this report to be signed on its behalf by the undersigned, thereunto duly authorized, on February 25, 2026. + +NVIDIA Corporation + +By: /s/ Jen-Hsun Huang + +Jen-Hsun Huang + +President and Chief Executive Officer + +## Power of Attorney + +KNOW ALL PERSONS BY THESE PRESENTS, that each person whose signature appears below constitutes and appoints Jen-Hsun Huang and Colette M. Kress, and each or any one of them, his true and lawful attorney-in-fact and agent, with full power of substitution and resubstitution, for him and in his name, place and stead, in any and all capacities, to sign any and all amendments to this report, and to file the same, with all exhibits thereto, and other documents in connection therewith, with the Securities and Exchange Commission, granting unto said attorneys-in-facts and agents, and each of them, full power and authority to do and perform each and every act and thing requisite and necessary to be done in connection therewith, as fully to all intents and purposes as he might or could do in person, hereby ratifying and confirming all that said attorneys-in-fact and agents, or any of them, or their or his substitutes or substitutes, may lawfully do or cause to be done by virtue hereof. + +Pursuant to the requirements of the Securities Exchange Act of 1934, this report has been signed below by the following persons on behalf of the registrant and in the capacities and on the dates indicated. + +
SignatureTitleDate
/s/ JEN-HSUN HUANGPresident, Chief Executive Officer and Director (Principal Executive Officer)February 25, 2026
Jen-Hsun Huang
/s/ COLETTE M. KRESSExecutive Vice President and Chief Financial Officer (Principal Financial Officer)February 25, 2026
Colette M. Kress
/s/ DONALD ROBERTSONVice President and Chief Accounting Officer (Principal Accounting Officer)February 25, 2026
Donald Robertson
/s/ TENCH COXEDirectorFebruary 25, 2026
Tench Coxe
/s/ JOHN O. DABIRIDirectorFebruary 25, 2026
John O. Dabiri
/s/ DAWN HUDSONDirectorFebruary 25, 2026
Dawn Hudson
/s/ HARVEY C. JONESDirectorFebruary 25, 2026
Harvey C. Jones
/s/ MELISSA B. LORADirectorFebruary 25, 2026
Melissa B. Lora
/s/ STEPHEN C. NEALDirectorFebruary 25, 2026
Stephen C. Neal
/s/ A. BROOKE SEAWELLDirectorFebruary 25, 2026
A. Brooke Seawell
/s/ AARTI SHAHDirectorFebruary 25, 2026
Aarti Shah
/s/ MARK A. STEVENSDirectorFebruary 25, 2026
Mark A. Stevens
+ +## Corporate Information + +## Board of Directors + +## Jensen Huang + +Founder, President, and + +Chief Executive Officer + +NVIDIA Corporation + +## Tench Coxe + +Former Managing Director + +Sutter Hill Ventures + +## John O. Dabiri + +Centennial Professor of Aeronautics and Mechanical Engineering California Institute of Technology + +## Dawn Hudson + +Former Chief Marketing Officer +National Football League, +and Former Chief Executive Officer +Pepsi-Cola North America + +## Harvey C. Jones + +Managing Partner +Square Wave Ventures + +## Melissa B. Lora + +Former President +Taco Bell International + +## Stephen C. Neal (Lead Director) + +Chairman Emeritus and Senior Counsel +Cooley LLP + +## A. Brooke Seawell + +Venture Partner +New Enterprise Associates + +## Aarti Shah + +Former Senior Vice President and Chief Information and Digital Officer Eli Lilly and Company + +## Mark A. Stevens + +Managing Partner +S-Cubed Capital + +## Founders + +## Jensen Huang + +Founder, President, and Chief Executive Officer + +## Chris A. Malachowsky + +Founder and NVIDIA Fellow + +## Executive Team + +## Colette M. Kress + +Executive Vice President and Chief Financial Officer + +## Jay Puri + +Executive Vice President +Worldwide Field Operations + +## Debora Shoquist + +Executive Vice President +Operations + +## Timothy S. Teter + +Executive Vice President +General Counsel and Secretary + +## Independent Accountants + +## PricewaterhouseCoopers LLP + +3155 Olsen Drive, Suite 700 +San Jose, California 95117 + +## General Legal Counsel + +## Cooley LLP + +3175 Hanover Street +Palo Alto, California 94304 + +## Transfer Agent and Registrar + +## Computershare + +P.O Box 43006 + +Providence, Rhode Island 02940 + +www.computershare.com/investor + +## Annual Meeting + +June 24, 2026, at 09:00 a.m. PT + +Online at: + +www.virtualshareholder + +meeting.com/NVDA2026 + +## Form 10-K + +A copy of NVIDIA's Form 10-K filed with the SEC will be made available to all shareholders at no charge. + +The Form 10-K also can be accessed through the SEC website at www.sec.gov, or through NVIDIA's Investor Relations website at investor.nvidia.com. + +To receive a copy by mail, please contact: + +## Investor Relations + +NVIDIA Corporation +2788 San Tomas Expressway +Santa Clara, California 95051 + +shareholdermeeting@nvidia.com + +![](images/6bca24f3070d879664467bfeeb13ca9debe7fb2824239a1710dba5a951cb315b.jpg) + +
+natural_image + +Abstract 3D geometric logo with green spiral and hexagonal shapes on a grid background (no text or symbols) +
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-1-NVIDIA Logo.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-1-NVIDIA Logo.jpg new file mode 100644 index 0000000000000000000000000000000000000000..459b68f179a0e53da1e86578a8967604910fe3f7 GIT binary patch literal 118840 zcmbTdd03Kb_cnY3;y}(pIp-{BngckcZPY+SQ6{y>+D;jennPOJurbSu)ErY04VBTU zG%an~HsDZUibH1BHY-cb+Gf^Md$*sLp6Bj(Du((+3I?r>R>)%uV zo(1%Y_z*k*0s#O>^#K0;3R8d_SKFm1Sw_PTZ2#)j+l;N~WqEX+;J z%q(qP94xIj+nAXly-=H7-OyHuc`z&c(BYii{FA2&PMBJg9#n;xk5J8*I@p?N)w<_Mgc9{{?p9|0=To9oYXH z*EFCF0jV|*VgURCd`VXzqEP39qCgrvAj=L;2H{jIgohNUp=W4Pfx5vmr8Ad4CA8+) z6h{@#@~)R{J|XKI110*XbGG>A!U_wi@-$OWa}xO;7r2D;DQWZM!0-#9q|^eA)vvc` zQLYNB>`C(_TP$NJZRFzq|0n_i+J zBJyTPL^~TfR{F^INakq~iNFNwp@P|-dDCJr@8JSgTkgX4w51*50ae!JS-2i$DjI~! z4P4v66hL!a9eYd4sb1ECO}QK`+c>50>P@Sww@}!-Eunfoa8OsVQ5x%8DRYIjgz4{RvMzHy^gt2dGG;jFcD}9klCIouH4{; z)k&?iYdU(LzE5PUbM+`${qj7QZGn3J4rpqFlOe?X>6RhQ7WzrDdIE|zDCKp%usjm! z6YCfP2l=<`)xablY+b@T%1KNZkOik;p;n?sjtMSsMO}kiUy_G4UNYwH(Lm(r!Jh#C zz`3s`IS!3Ye+)2K6}m#QT5l6(XFKAzGjvjJlL*-DOmhCXQ;~ZFcDCJ4wa4hhG|=lP zocwFW-}YX(i8eK|c_t$^(!tQe8nr| z)(xQVqCooI7)M@F9vmgm$|T-mHP&~FcS?$EM-_A(`7-jA*=?#K_ixTa>>e_k^xaJ<$o9G_V`@ zoKD-S6gJ#=;+y*X#`xvkmrD{S(W?jF?z>zW)CkkD*o{kZ%~E~0`nV&X=Ow01i5~)~ zyeP3PJ^5~H;=w;B*7>^*6@LaXyIs4ZS3U5ru!bJ=Fy?$Y9H^07ByAmu;yuLb8t%L0 zXEu>ODGr(%fpqXZ$E%y*AOm8m8_j^^gw<9T(I^JoNN>S97;D0iS7HLhPz}OW$!Nvi z*ycp`9RVJa)Y?J_cqW+e=4%Ub4XaH*Yo*G}^`#Vd<>rdHrvgn)Ir`eXzB~kRCn#Te0wn?o);)M= zxlFlvxS4ZvpXEwa0^dt#l4Go}If&Vcw888NYXjW3bvxrLz5~c69ujEe#+U%9NV&5y zL?%#oeT;~lLq0ndGrdbhyn1Bd5eW!Tt9_!;=3RWag)-In$s?%m2pO8vl>#3MJ2NGB z)nEg)Z{TCr+rMRyJBxY_b?ldi~X+GC(` ziUJN$aBzT19mQ&*(g`Rg*t1ZdeWO2`*Gt-~0H_w9dKkL7eJtN^WPr z>3-?3p9hRJR<~OtRF2i>if zmA)Jj?r&UTHC=IMV*^G7x!ks^6wCIL|_d&g>)DZ}yCIUT; zMRQs3?`}MRFe7Uz9V(@NSx*hF^yAehp2~QK?iZiMnzOxnC?ONv=coF6e8edj)92c2 z(mj9bt||Odyw~goJIS;(=)NK8sbQVy6x^A)p5~;k$um_RzdoXnysVCJ@9%$-Jnrmw zat$M=dHt4nn*3t_Bi+X4q2G$vF4)HVp!<`bILGu0FBFC+-=5BUj99%BiKc@kWPq2$Lv~ zK6Na>7cE7>A>;&gVGtLnN^wEbcv%1*hi(fEDI8IQG8xq(`v(Bx20fn_OFRq*dW3SM zBmoUFh#AW^>9|_87r+X*CGvns<&iu&)y%NDsT;S+)Q&FD_ zIQl*a%d$p{u`L-|Zf8HsmSqxb1?%i&!ZAo9brlO^^+@uF8#tPsCYSp0jwNOjC%pxl z!#e0TUoLQ}!lHZAqtn)H2*vgV9$?d>kuML^^y3QO1rsyEhVBT~Rn*C3mgx~|0(IT4 zMdnht09^Q=;~vihxrK14ED`4;Rx_$PTEm*3NTHx-$!aT)q6n2sW4dc2OpUtXbS^4m?VdhEt%w%Y($r0cfBeK9BdP>H* ziBk!!%8()Zx~_WRi2Au-@muE`QnqK)BDd3L^OYQnC!Jmal&&jY8K}R4MJ}UQL|tOs z!$=?NxZ#MmjO6Zh0&5QP{(Vo!&tW?BGHZ$T1s_;FOLG)32<38pGN55TDyV=*O$o)2 z^sqRp2}=O`XcVW2LXuaMmns$bl|(8SDQ;o*O>791aLar&1?#%#>~HWtGsD!-O4NKU zR>!(JN+h>pU;}YTZwX>sMQw=ibxFJbF7#sj-?vDpN?g)XKHET2k)0 z7mXO~u zIDM@x6(YFQ`e0?THOF?%OXB`wnngT3U&sBT;I>x`SEiI>)&*mAXw-b(IU@-KYe>x_ z=|mU_G}I(H$H&*e7q4=x4MyLL_feMxFgePN+exPWyBx0DIG!wOV#55a-w zU8hjr#1{mbbvW!oKJ7*%D^U(WxG@vyE+m2}8kl40I2#GnmIyTKdO#9Oh9(t`Z;zc& zHZ~l72ez}NKX{5TU{K`14)qI7(TZ&xc$Vx!gN!@^rc#w;h|~uCye|t~AAf!h4ir{G zf@$b?TgSJg-L3%ocIqMGeXePnJTA+uhX%YI)XN2?V^T$%p?lk2IewnMXf~hv_Xu=Sf*4m<fN#WA`n@8xmjOfJ8; z8@(~Y_|5{>6wKwKCb2#;7Kc`mNkF~x5olG3h>jQ;6xiDBx+eJ|_F@3X;JOuLumP`a zv$>ohw6W%RoEcycdu_4Y=7(J)5T=SM=(x$oi^;;mbM13E_Jdhu; zh7_{tDY0qkG-q>(qcMWLK5amRSJ+T{AYs<2IuQqCE@}NTwDu3S=heM%&392>tb6|m zOTk->UWxgzTUTqY{Vw7(+r|Aa_HWN)#!1q!KMuz8I`oK#u}H9o|Glr`<;vTu7l&J+ zMxtrtWyJkGHjwrxXQ z<(puc@!WWViBMe{$8`(EBS9x=nlc@9mQc;5a`#-zn+!*XVh*aQ0+ZnrAhqy~FB9jP z0Qp4r&5U`m2D)FpHXJ=ehH!*&;S3t_9qdEpI()Eo8$WkEaaJ1I zj}kAe|JSLw&51TQHo6_)d7HEI7u|{rT0&*BN4pg?4~a!%CjRzkNSj+bJj(d@io$&E zsSl!rp?1fb>+gc3shH_4x$WP%x@%>LN#<pLBZ%nrewVw4W8$b-~|PJ;oZ5) z*G&MsAU#|sU(O2bl97m(b;ZqE*YcvO5Ra#xe^#3ucTbFdHk+P)No~wG&gp_%X;`8p zKeX^kPQ&gd>4UkpQc5QonmkwE8Go8sC=;yf@W%M^Cl5w+O-#)g@PLc(9-%hU1qn(I zvm_@XOo>P!7FF9P_G@gpd2t;Fa1JO&$|zAtHeWS~zalVJ)(* ze#isb${d-0U`+diiOfO4y1v)6=9>qWX!$%KO+v$Ylffpe*Yt|JNU(dQF;yqrSBX#^ zqkP(g%$Vg%R_l9BKP!)pk`vn$05)~kWpNGvtK_`k6}xOkj0&9Ol%JtE{*aM12KMbu#e<|sEIxD*R7 z!=G)zn+jg36;61an{I*)X6RUY<_#){C$6-8$mC22PU>r-`}j0DUdP)6D69V5D|QoT zfs91b@s-f)oJ}F5=ONcQ<~;6pZPb0`dMv^y1wCDlHHyqgV*)~d6ibf?u-7DG!Hlf` z=rDHXEEW<}SkwGi|8q;h&u_M1U9a=7X=NKTm71t#vES;IVtX0p#s}IO95ia^hV^Ov z7e@27(Xh_05PGp-L~WdIf6zPzcJ^c3udm>s1LztAf0}P2CT-7(_|Slk(MfsnS@~)_ zD&Y5;>n@I-^@kR5PI>KPjcc)vT~&W>iPl)yS!DVk)?$WF`L*;1`waFfTtgCcQ(YPqo_BHt1&P+Zgt zZ@VM8M8rlh7jMXiaZD8#+hPvTiyoMn^qS?JX0)&5(%_vYYD`EdIamd#Dli+3OjPjJ zzLvNE9nv5MLM@YF>o~sS*GqTf(u+CcdK-?=74Eu1dPJ5<;p-@Oc|g|sG9w$yiIcOL z^i7mAW^_ZgKpWZaGk>sYy!hb$D$2Mr=Ya-ghn_UV4m^i9-(NZZ7 zZj0-R;s-;$Y?Y@hlX@gIQRSUBbv%;v^`DLow=;YvbK-TZlAmnu=(czt^Qi7^>G>DZ zhdzs}7A_=l?sDdVs3Ewjve+6Ze8vjLy`|-4*`R0NVd>H@JaRg@b`er}nPv8tw@1gr zZ|p!Q@H#n5^1(>VaSOBM9vwz%c-dDJ!p?Z1=Vs_K2kA+LJxj0}ZhxgmCqeG0Osa52 z_+zQ8u3kYYiBnL?0u(OkA%F?g&RDM310%s_MtLsaZivXvc_T24v{#7?&RC?-o{M#c zSI~M?cze2J-E;{Lg#_IbsVVotYZP1G0ZcfBN6@gTNV$9Ued<@FajPMFhn<_Z&)kj6 z-UFv)$MXwYhkI?+wCRicO(x%Y`n1`V82ox5u!}uVV@|0YQlLvIUHOwu?n&2>m1m^2 z@KZ8jxsqdY7-LQ1p2??p;sEBl(xrf&*4cqc5niwkAIOr}pL$JO!5Sd!n-W{pXqp1p zDS}W#U6QB>d=Mbe8f~U1B7?Y>+ju}W$}PO<7U1>pF}$DiJ+i`B7M24C)m@g!;B&2L zj6oO~lBe)9Ihh`zgEh<+^L}Hq;J^%3NgV+r6!;+;RbK~)ULfkCK6Kp`lA*0X)rbFQ zQMp74Rwu4J%!!w?gafccWOw}x#|4C`Jrh(i6MKr?rRlFjIlPciSu*r_IKW zy;G?zKNS}wsHqJF&u{bP7~#ASE-01O7v}D`;XEQ%MYKWDW)nA#oRF{BY zN`J1-XTO&~%GFGhhBNyKy$exo?^7P*eS<~D5C$PbRkf_Ju#2z=kG4JQ<$u?7#oruvMwxhQs4hHVD_Kx^)olx-GB3a zbs@D<=`d}oI9Pi|TPM1lTI{idwv!0AthNU)RBA?aZ}wt(*o$93nuNjBooXJC7Uk)IU{Rwf(cR%0nSIAGp?aeK-`Zd_onANANUlO)Lo7Vq4W6{d&h96Bg| z<-gx&=X*;kD=2pUVi!_+-oAj3t$j4pBc^iTv`|pA{qzkZ2`_G#si*KOx!V-teBTwL zrc6*TT%{bkyzY8s$JGH^aCp69VD+$sIZ#P!+i$?BTR4wH(S4gnv|~;LKSLbIq2`~_iA8% zE;W(D9~;$9^=T40tBVANg_Ka`Nr0(9S*jF~TWZFQ=XMlXHzg{>0E z4|u>xWAi|oFJ`B<8~ivM>>LY-q;;ZUMLq%W#KEH@QL%qO49BPO}Bn| z|M15&@OWeAmPm~4p`>sRZin6F0b-M_v*~MUybCn@W5kB#7fszgYAMHcevL7ZqoL_g zzaTEJ3f_;m0avB`0)Y| zU@uQIvi1dA1?5Lmv3IM7JaBLtAm%YN1#0lW6KPOl7pMIldqWuaI-czrfHi)9N)Lgi z)`MggzU}0E0&ceNGrBw6N7K87_)U`5v&fQUPEqxbjQ0vRB7EsJ2vAynIq9#k*dH&y zO;Ya%WNmbdy)5>0vy-dP%!=of3f$t|&RvlSB_Mf}SA1ThFq;n3Bc%9Sy898Y5AHE4VcVA6NRekoc+|bW=D>eaMKui{jii1F z(E>{8H9M|C3B{#)x$_qsg>Q1UT4lPH7uB>W5L%~`}I8h(*j<-|eQSBC4N?i4t>XUtHu83MaFIj)}d)Rye3hknyC z0v)oh4v}dYiA`)lJlGg3Q;vH=h6PM=0|*-HxkOVQHa%ztYM)1RWr@$bnir1XQTqI` z>Xit5yOmcg6;e(TnNe>Gyt&wP{9;hhG3;?XO1O$uFZq*oO`wi!Len`(7s`q&fcJAH zd}UHz_^dfnu?0RhSw5oCrqPBrVF*8P!e|kUW9VXbB8LG|W#&c@Q`lJ7$B5iYKu>#+ zwzAbLZ{9h*XMdnai4z#*`3vL8wF=+j{R^L354pcC_mtS+z2xDV7!o@x+2lL=L-C1B zWeqninXCCzcH1k%L789~BW+IT6z0SQN_9eboOG(XKnpcXj~zdU)dSaWD1nv0fYj+Y zMlHNr#Zao?Yjga$8YRbpCPzv0X?FxrF`qhL#xc5=ANks-WqVT7!9f9RGXZ0LHJFhs z#cdaIfi9%ji6#f&KXFOx?Fq<0{RER6hy=H0po96PQDxXyj%@qzZY}Mi$~n%KB<>~i zK5v~nvQ^cg%g|tWV3)8uc6DHF;F0af=Q9$Xr{cbdO&# z63l!PZJr}@@-7G}NsqF+=%*53YX4!t${HUJYsys5FS%?^LO;sC1#HVFV;<V`@QJCtLTG;$MB4n(y|{Z*#XR46^=Ah+^%yjIW4LKaI>UP681hKu zv%4cugRH8L+&cKs-~w5pk><)uNpka?Tp-{YQbXp28|EbVXk++df|Wif!r(knwby;v z9+3)L{Ok?~dPMF$?;zUsk#c=RJ*KoWnS13xdZ9YTW}sBrf=)D>6XJO#_bkU+e2uwg z>!VBWY0p;=jIi8~soF&OICYMGmgJRnweNV7$ki%Qy3_F`ygz0{{m8&Ol)(dwwe(j~ z3%@Ty>w_h(2s<)$ko_nF?9(_)!r7;j9$K6b4LeC=vIxGJlmzBOSNI=Y_dTi*{0{x% z_t6f6YeKv9Nco#yZ(XhCs@$%0^D{ZEjtR26eu}dS>u+DAHJM*8JDIPcUYi+0@du2< zjn67_Ruzq3tj4|EV6s#l*WGv3GxK%?F5qxq`GMtuU`N?Qw-r|@IC9${NdIoW4A4T?!jdmio> zT(jfge8aGNF%LR2@_38+Dc|+o=gr0%DNOn!P1^$WPpzDIT8m;Asm&GKv zI}k1{r43us*kdHAJGGi?rEe~=;(586JIo^PMM>b>40F@TaJ$)pXR+5+=5UX*EUdx_!K+hgy`#)cZwjIYa zIA1NZ8(bfG5dMXhx@$6DPk0p4C)6v}61T)=k`^;} zuHG8a)0m&Tc#*+QF=w^Utc3sF(W4iuslZ^-qh(07VpQh-Y)+;LF@V1s;Ek?b1n?** zLDRY^E}uwz@D6HSqRYBh8s=w0+C{blcW6@XB^eG$)Dc!DakPHZP=7zX+RSjRp0nTMOxO(q|CV*%Ez7;llAo z2?x+O{v|$DZ2F!CnLk{fikeda6Bgqb9JJvN zHu>hyW<0>5J!!MYiZ`nIRakC`0ezBD5`x{}K_KBx(TyCl9E2&X9G;>80?Hh5$x8P* zjvLmH8$*j0`H?{?9<``y^_IRik*?_HMxFgh%piV=CZ!DF;p^i^(y@jRURI#4BMuFG zSAnAS2vhM9dRGC*(eJl9>#)JJ90XH}*+aM`1>Yr@Wyv%c3TVj+@0!|QJq|peY>%c0 z{xpJ^rh-?6c`r%^RV>q&d#as$NXmPkEm(Ij@ero|Ny-#W`_nb7swHOqa49TDU$;@J zhq=Xin5r1(0Xy5t2V1`#FxpgMqTImET@Bxm{%;_g09RQd&_d|`uYJrwpX?OLA$5l=0 zWfjMQ1JzB&B0(k=kyvPmqm>!VL{*$4cjk{%FT-D!HYV@mjuvAVT{G$Reijw?OPda` zfpL*Eu9~18fx)%p@U5iVfu>LI7vMIwh%oow@1HMfj zFi>-ezZ2`rW5rEPXl0r1wb9lI*w$j3kg?ru=_hES+-}SyjQiDBOr9(q0(Z1$`7mtV zetC2R)_!gMj9G}smRB)X6bXul@Ta)z>W1U&>Y8C$8*Q^2V*NXm~z?~ zQ_UZU9+kJWGLD73?WySZ@)?pgK0op8ad$AfEq%+n^F(ln9^#>6G3|Dnb*fLIlWTle z=iNttod?DsSIiEfznzq}^sLZsny&>KiF#xHL_WAOJJPB@o5?%RGF%;*OnGVx6@Ka+ znr|N5vwR+aHt0QLf@4odo;#YT1Jz~sA z#j$07fGr-$yd$oz!M{@0SiSX`mta3sJ8&y1YKyLd(WTMtZiLR`t9cw&VP4{$1)6J- zzJ89dxBdsPe{Xm$P5UID{K=!ZT9A+uCOX9IQ2(A9f1c7QFf-9zNGlA1AL z!uMqG6*oj-jH?10sk3}XG1@>CYBXz228nm~hz<|&oK|zaSUoy#Qn-o7f8@3U<=pHc zlGJ<)ANZt^P^W7(+}0qnDbQq7w+1g9LMy^Ak@dI04+@_LYuwt1RU1)f!cdvqC{SiU z$^)Mn@}d^_814rH)C>Hj#6flBkSH#YPhGEY-VEY0Qkn1)UGyk_Y(WqH9&&?fLYh&2 zk@V(wK2}%WD%gQR~FJ_(LJ#%GJDELA-I;mQnm4P~qu4so038>Ue zTwjiGi8NnfMgyf2S-q-q&S`!Yj@v)=JPm{?bR@h^ObaG1-kIAQc=Wl(#)LMic_t~2 zD@izL4%lCM?Ru&A-jkvoD4TZo>oyt)%iL2v;se&#Va^4j!R8Y|R1F0XalJfU3<(zi zy!{!dDM1^s-Cc*+E3hXKWmr98CEL}F3j3+$vs8rESUgvcGZLtC)0$Kqm&-_T2CRe< zlN12m&h}Jw%W?yu05wtWC5lP3s9YDktZKNQ;w<7^ALak+6c7?Z#*COl@lW<$R+Z4A z!mMWe8wI+nXl6vMyo>ESVSku675xaXMCxU z?HS%81m5@o-)*kFMcr7vdfiX=CeBXfPb%)3j$Fd(#;JT@^Z}74$H{%%FnBqdJ08{r z!frTwCHiG;nlVPL6zr&t-e+B+@;5Q}&A5_wXjrTGxljzg#3*YrGf_Fc>{K2H9h>RK zgI{C|ZdmyAw*98G@oo3%vR8*NEwbItUE5#pGmAF(E#s@pp-);1F%xekj_&;H=C<3c zN_uv;TrSr2ov6W_hP4YeQDRc+?nFFyd!5nSMp#Ff`g>`9Wk(^-_|xivjw-(*^&9o` z@v~R&eC42DTJc>sIi7zs5xGxIm}C0|-VqiL4dnb-jiI!D%(?opr`^$W`C(q^AbKG+ zi4T46J=fwE|I|EWIm#jGge|Qk0>j_yu(7>vyRTtN_&nu?RN+qDd0+73)d)Ro`Cjj4 z|HVkYO+n<}$hV}7_nd<(F0&8q!+oT0>b^LBbTD_L)-+zQkJ{bJve^}XI{lt*Wd@P- z3El4AGYdenYX&e? ze5=mJL>U`9Y+_tk2C6c?A}DoDk1aU1yX^QEwjNPlF^*|CP;p4Q=`iT(Z0^`i@uz$F zlUHcjZ>~w#%U=r<%P%aGN(wa;hNnIS0AR<gw*k*9z(Bz?v3yp~3FQyvdUB&m? zXAm!Nr z-1*HZgFa|ZdikWNH*+@=dh!fWG$h>#_+=7X-}MY{Xq-8e)m+iES1cgTl(-S^(9|~k{#%~Ug+6rM-_*R$VnmND!!yIt#*{PFZ>ybyu$}U|#Ar7Q+z8R4Bn^?ziC4v4^1_a}a?%m0I)`!bD<1W( zdWq!6^D&9m=ep;}OjG~JJz3AD6xFYhb9W?cyo5fvSEkj(JNFqhFwMg=H74t(&&6o2 zsMs+rJTtr>N*s&R@#Dp;goOWNLQqfSAsf>2_%Gh;fppeJt591HA zS(e(=1YXSOi*uIXbkZ>Ev9fl$pbv$!b4d>bW#Eh+Gs0GhjB^5=t{0zs6?WRtXD>?l zDyVxL>zx?@=a9pZAh(ERk^+f#0>*I8_Bp~4SS~Fyl!uH3lBxizVLEC4+?3XaiaZLJ z##6a_kP16*dfN^_h<>jG@x&Y=V-FO~+YI6%UymwFU{LtWfR;C)TS#Xj&eBZz2`q2B zbN2hnuF0{F0gBA+38FonJ&t|r_uSrDba3C0^SP6)!V=Yvw}BX;27kNFM5!Rb;1ic@HQm;eL5Zqz`H8C zfUfB*^ayzV0#9cf$B>KC0h~Q<`BpPQPbJyXJ2LnRRvmRJGeTj72Ay5K)`mQFJ~{Gv zld_Nur5spf%{^zUpr%0cq@xOIBA3Heaw5!4;amI}__*19RY5J=Y^XuJy=^%<{Q??P z>)hkubVt_KUK3Ge`8PZ1yuYKyd47EtXH?CA$9T-mKEugj+-@~WG>QGMQBq~=a!3!w z0R)r?&$=h_@*7dx>u4XrzCxgzsGME#XWqD7)gyKd`a5nYv=|5Njd7o^qeo8LEIN*7 zoZY~&{V8Cef*C~0B*!8bma?_m4uGFby1)g_mNhW%AegHp6c2dJd|<{cLk)V{i<#di z4xKmSVpSdnDzC{*q^n1~TBZ&MS*na7u)_;sU=;+0bmP)pPuORq0h02(m}_S+;o6#q zKXQjN-$dJo4=Tpjw6?gIUaYWi@+q<*0K@F{-iW);zqSm4*Z(+69=ROwvy_V(9F zsx1wL_d95BTlnj4uiHTcGpB5r>+@i+(?Kn=X51{sx}@z1re?kDS#f4`%^T)TaY?dS z-miYO=g|&7+^l4d{feH%1a9(`p{+UtmIo~=u1G(Xav%$#T_~E-^8t)r?d)xibw9)U z+4XG6+0q0WC&nbn@A>lA-=0c4PWO9I-#om0Zd+nT)4ZS0M?%u}^ReUzne^J_84dL_ z_jBySCNi&Gebw`?q0DQ=N5(|!bT=MQEK%+kM%IB|H3$!}5_iAufM+sH+VC+Mwtv-9h;V07K=H=Y5yxZzGIJ8B0@Y^~1T^R@ZGf6d{zCjSS!yWo9LkNgZY46&$LP zTV{suG!Q+vw9nu4P#)G{b3f-EEWPkW9p^-lMaZ6t?TEZVvH6XwJcBQ8HoV6*nY%fO zEoQXdRFY3_)!qS-WmLs_J`BDC7XC&a+$Twy@*Uq-K3VZ>QB`q$_o4f4?8qau(i6~) zo(`51@06Y6>K%86Q9&PPxmAwprU;0(mj<_{oxJHv;aHHz(Z*8=g(l$|mTSW4ci<{i z-+WD3KcCZ!{$TTqt0*-k7qAD=Z_?CyfS|FdB{W5I$6*V~~OI-fWt@zn>SeZL`l@be?=($#=%kY+IyUN6eq@!0jN=wcKq@R@F>|#%4_x2w zr;R1BY~+<#J7Wb(ZjYRTz^esWBDq!!6kIYI5~OM~v$|eRo&=Y3lQP(NGn(7>5-3vG zSOP*fiR{4DAh>|K9?!fEx=+i*li+)1jYJBR8I8s1I;k8xG6Mc8Qh3&J^)*T0DmWz; zu5ycv;yn8f`%!pFtjzFP7gjJMi51W|>&?q3)3oV!uZVR1pe{c(`U}c$Qt64ya2rfW z>6`pfIr-Vx@5$Jk9ZQ$?dC9tcKBINqs;q^ZT-8gs-{Bt)>Z{EQ$v8h}f?>4|Bg@;) zlnJV|_C)olI?Ehq)Sx83F^I|q!l< z9nfP~EnX?3QSQ$Khm8_ch#AJGgcajTj&}W{qq2P}|2yYHV38bhf$C$^CyUv8|7vVf z-R<6O>d+mb>giAqFLXQ@Cde^8omFLHnqe*?HnMc#Gw$OClfyr;?)q zc-;b1pmF|22+@Z|Qva`Ug*U{@at0!q-{%{r2M?kCc3g=W-=YHL@@=mQX2*Z9oBEME ziD{}%cxnaB;GQmOb_DVlm0GOwz)@7S_b2Go#BCp)7Yos#a6%?Uo%tA(!<}Hkc-T<0 z-WFIuYrf~bI}*`9x=2Vgmt}dVc~i<}N2l7y*6VHRp$@i{Q?-{B`)C`kzJ{#wUheB* zB_<`5k#>%!zW5UV?ChS2FuVrirOeAS^Fegt_HxQ)?}ex{UqqkoAnML~eAD3Ny7g<# zi7!}tklklt7ih*BS>c~@Zhig*;iusr?_SK8IXJn5$cT;)2!SC7ECb*B(qk{jJ;7`)-`den zEbbb4uUQ(MNAteNTLG;?1KS_uNO!b&Mu~XuhZ~5+j>vndgK&C!MVmR<{l2o{>;|{T zxydJ66U^wLzvuJ~$%yFEROe^~>iHV@YRGG4DMu&7`b<@QW|+o#_S^LI>{uVkK5Wkd z3Cij77DWT9-s}QwQqaHO4^I!de2I}0c8L)}oYa2=&XE}X z6L*!TVM)989_^F`(%cv8k95rBrj`5Hw1MM_agmK*WAPf^4F%Jrr^(!-g)pwdE(roL zWdG{w_zFJmsBz1c&tvtp%&0k+heNn~{n=z4n-p(PoQoIjuovPSHb<)o-DD(GYycE?Te(3& zfE=KmvT7V+gAa+s$`KA~CcajoX0rbBx)&8R9LkRtrs3&EQRauwoo*Mnt0;f3ki!VV zGsS&jqoG`st>u79hGk?MoutwhbaXB4k!B{85F96cf%Gx79)-FLbCbi9N>pM**ia)c zc3bU>!cnf#Iy?%8TA-I`H6Qz60l7|x68oEIMP0OSoeYn07po; zF=RD6wtG*DEzOt-I+IRQ&Bo+MM8iQ|Y+~wS;5Z;0GnU$yacmM4)*X^cg=GvGvMNUk zV28mRQ||7r??|wJSA0Fr3l4Pk3L80Ylxfe$#2aQ09XgtqE!dU}PX)l_5ercF`4-)n zpAP!M5rFv-2WbO{+h)awB>x5gv2MkG2ZUIn%m|?zh-R*mD-J(%Opk_*83zc|xU-&! z9t?(lIM^-%ZTr!9`fo09Ei`N=SYb8GZjFR!@Z`0g>ucY{*x@VL$?2 z2fH%SK}xyx4qULu<4%kIjQZ@FbRTofZ^GE+RpDz37U;Sj++?78;?rNdByYiwK*Okl%T4LmvAt6}N@nurYq{9b}n_^${*Byb};-s}mvN5VVp-&hVJ5 zzKa`d&aeVjcsI5Mu8=(maNc^?6Yij~~v^+xh8~<;Ut*D&S2?PNrWng6;jr zPWIz=pV-nh9A(qB#Rl_eI>v~PANVmn)170Scc$+7()^P03x>`ZA=GC)8#=mAj9Sp* z`H%c9KmKR)huu?UJ|C(5V~oF7u6kv)+Sty|T>Tq+^0(H{Z~5be2_CvjT)gA*O})&p z=c|d%`?Acvz=W^gP9@j{tVy5w_(;$SE3<@!uzp+Mv&f#;2|mKqep9gABT95Nf9HM+ z<#0btEr(LfI_u%cIlk9yuKTOihg(m`h`=jmZ+5Iwu3UQ_Pe9nO_z8d9;huf$y~G_p zK>Fon-_gA_5888eK&ip{-Q@4R8(2M^&3k&t{{E|NB@#!N-t1v@bGZB zpaI{Uw0*PUD@%HRUoo?y_E?Fg-nP#_{?h0bBljH$;vMO~yO-KG-{?1a9b}jk_DUu> zOMmM}&uAZzTZ>;6&haYdzG07lqMva0XO+t0bNXb66R0TT&(c4m=NE#auEd61dHAhU zOZw2*Y#%pGyHM00RX>=N-0uqYY;V0iYx7%W0WwyiM%%HRnqys8x7;VY^$Z&U>!+_O zqJOI_My()<2HE@T{w}U-7LT3Sjg2qt_8CtE;XSoSe|P37?Do596($};pUJ+esUT=n zvxLh>$rIfm#VS7m{s(Vcg@|A(lcXD?pDm56xjbS8SyXVi;cFX@%s?$Kp?lRbL-=cw z_uo9|)}woy;PFM+Fp9EpG!$#j(^5fEMhmCHf|XACNQInf ztf)sdwR!cZ>ln#>%gzBOE|iOm8?yDgdgf|hRC!C%3smj>E`XXPP$m`h9b&T-#!YQH zGtgx`OG}QN=jii@RG-ER0;Cxk2ag19Fa;QUb{FtiRf*x^GLQ#v4+@I1<^dl=m1X5< zXCRyTF(fn$3Rh{MXt2!j*{RfpuQ$d~{!ZH3+@mzKi&5^ABt3)8eq*~3#yjiy$wo;M z#?aV@`&^$k*SojJ^ca{}VX^c^V_aw{xh9>`H8Dboy}Tx^Ss@!=33DL<{(tc3Wet|R zqX}?OqS@Hp@$^y}3?OwacXf3Ece|91G?Fz(pUMpg#p*WJsz%a`#JnskmC2iWHqX$ zXOVwZn*120{s!I77yW1a^d}nzm$W&zd*;^w>7&)5rRWV$ zqq_+@=LUuD^ce55Ilf`AgB2JB{geUE5a)Ng4M)vA0#4WgIa3lT8_H4>Isoa@_*n=7mB1 z#$K6?AAJRXH#N^QTBvyI-)!&%;n!AajL=hZj51ZTV&V41-PTUm4FfhEKG+6Sb8XJiJA=RE_N*NKjkT`r{^;0Uzd`=TZ|4J+C|3yoi==apXS#p?_-AuI zlda4tnb;_iIa3Z{#vC?>5)zV4$((ZNFp+8@=QO8b!<^-iW1$>U!iH_`R@LG z?(gp}^LUK?F}u9qufy|#xT!%a&&(w8%NLAa!f}x3ev`bnL$)fd-bohv+3?8vqkmjp zf9pLRr|HrpVh;s_%>+}yjS6*d5!J=-j;`bSr zlIW8rLTKs39Eq&=cSXY<;6h6esVV(-`l4>L13k#ek?L6#7gMEYJZ0=VzIfSFn7}+G z!&cnIm{|)2+pan0`zOQogB7{0-qj4{gF3?D!w+Q?&?CyJUJu@vGkxA*=Um#$W2lUi zV;HT#+&t%JA&5;5{j8cA-Q7*fXfye$blx6ZYUV}DFpPP@I%>$^Nw(fa!^)V|T}$LE z!=K66`lNq85iO`)(s;iec|~0UT%!KR~CGV#k2wyC#Ho&6qHI1Q_AiY8;EK+XtLE@f~)-LTaMq z8$(o)v;)LvIA|~Aar6Xm@Ox}q@7Mgd57!fxZf%yzCk_~g$p^~}phg$Z!GztqVWJR+ zDQae8e1?^huv3^V*b=-Qq5w)0e71Q>F5!z0&Gx5*K7NP^;UeAKCgv){Q8}&`miRzh zgsy{UVQT)1DIS1;CPrr7Sl}fAVbIXQbfHnpG@0IB7{R$t*xHJR2HP<&TZWiz*xJMZ z6ihr_3{_U$WrhxXgBB_QFYKOQ>{&%x!a8O$j1fp4XFh|aNn$|krqaP7P33XWl`$D> zpp&S37_A|UY}BG}wQ(iHH2>X`)SHoQcCkh3P>l?vg8JU=cgfA#^PQgUa_Ywyv-?u> zQY$REU1YUI6 zVxTQPg#zOP`)C9(%FZQ{!p!)fG(Ld%pLi%(n#D#74gv1r0!@~QNlSRo3+3-nkop<^ zk1e_e$ZA5mX4HH>Y+0wNEoOxO!RH%r(Tj_J$k;`92MKb@v6dT+z(+gVv=qE9)9UXb z@DRxeDu3H~>g}MBFwlxUV?#VMbQKv6>lw4fitR7ZUGFIg8t9KaP!B-|)ANdX|xxNiUK2%fT zTA9yx{5)|nEWdr7;HFOIoP^6gpUn~CFY&dbexV!mkyS#h#c+cCaN8uqGMnxpeKk(u z!P!zC=8}22<16A|NT+%OPikz8QzP9uqD@{VXd-9zh zA&}iJ@xRg;>}p-k(r

OY^(PVm=2XQQ%mOd)A($W!x}TB5rM_ zK!SAZqM2K~xnFvAO|3+v2ZR8M{lTa(TE0^@D!IV0^3KyesA#Bpk+sY|9osiM1x?5D zV4~tyr!HR)hot--W)z$3SJ^F1w>_~dda)q0Hri95$Wf;tGfA%%AU?>1r8>kC-3Zm~ z?8-+CKQYtMaX z0MT9c=!RLvi6&{NxY$11OFZ8BPmh5gr~GH0B}=C#YOHXP9Em zcjCA>Y_Igt;+TWwIQ#4@)XkM>?>B@DeuI}Ue+jdFzgoqUZTj8;-J(82fX;C4X&JJj zsO_yo`rEpQy(051zL=LB($}Q2Q$>+6dE&@87zaTe%<9V zWkmOrr%MPx(=%dD3;4v*4`f*1@NiWKv^xuE?=sMRB<7BTn?Wi43TVlP;fFD-2RJdc zpg^cBs+h^gEY4*28tXv`3p8^F<+}_?O_I2I8UHk1P4!L?02xAq;dEFAz#4`Th~r>D zorJb6F_Z2*s+XTPQl&JX(Fp^+_6IneH#d*yqZZ@g9B+dt?m7&kn8T_dRwy1=!dRUM zCuuCqO1{e&Gv#C*xPS@PVwTMjv85vbY(0=c-+xy?HaMC&B(rUK6=KVf#YLjreGa1# zOolAj$w9#a9_o4knc|J0_Y6&=YtuY*T>W_hVSj|dBa3Hkh%{^t4e8~pXf#8x_~|6H z_iG^?+jMa&C5&ukTI%v7UW61p$@hU60AwWR1&NNw>e?akPX0Pf1j6tS@S zL}kgMs7uB8?!EV33OuVNn8X}mF>I5hE_jYj&y)>hfwl+c6W2FS4B(Hmf>IT1K`~Gj zjm94WZD}xbg`oqtXlNF7odTXQOGxwHRqDv-gO}()J=Uz(CFuF2%dwEnE|u#j==3M& zmvDzO9|aIYS|!h%@M@Fq+976A_x}YnwlXG4ue+X4y1Ymyjn&A=o^{#9HH`)`80sp%!9Fu0S*%{* ze@9aASjy+nxOQi2js3mi3_lSDDO8E(udov^?AK+w6^9Q2u1{J5#oxa4;MlkW>^DKC z3w@+|?8q|(G^A|P&OXI8)nAdc+l5Ah{W=n=I91{S%kUSCOe}2DLf|z(k4wpaBh8hW zkdfAV?!UT#lDd`gsmQ0~U)aZxcvnReuHQ8q5cVOPKR?~$5M86Z{IMn=z;CxYIcjjx z;LYq_TzrP7YwmLWZR;%*;AGaTD#oKUq3Y|WTpYC?8e&A0P;u(2usqgeC!aIV@{yy( zm8g?Xopu{CHJ=f4v6&)I<-h7BTvd2^+CQMOnM}!VDaQ^YO1rK!AyF!dyIT~jgb4>& zv&Eg3#)3|zP@kCju`TAlTJs53qPkQR%kA(zSgHE@oDAdO^a&@yjmhuq3t!t^yB632xKRwPhZ{EQ$RhcR4%8_;WMh?o$%sot?L!GdYdvE zE;ruFtS`XEZVFy{crafV5wvyt)j+EW1($}&y=SU%@;5;;R9>Pl^n%2m~dH*ASNy9e{tzA+DefNLf@E}ajviQv1puKwU#ehWez^>-}pU*N9u zC-Tt|&k^23in}#shVa5~KPcaOvGfF=>vK0{4G)9&Jda2fAZFxrOZTSh?M458KH7at zP)mgG12gJpM9r2{)nc$0?q$Vm4e7sD;1P~|j8CE{<@u&6lqn=*w1!oju&NEuoK!Tv zj_JhPW;k@K$rQDB0+E|Nk23NIS-pmoq_Oc%v-5K!(Pi5F^ca^$h$Al_*d?~XS+tba zGy!GkX*_EackC=RDS8O8W=FGzMN^k~(C6xe zqnTyl%)Y=tUKS4~A8pReW`NZcGVYSaNDLv~TsglH(&#hlC@<{n4#{*1WI}0z%r$ev zvg?2|Hjss+xq+6^G7SgZBwz9H{DPIxwFi<2zIpwa9OekXVi1UvzH~?_ScCpWElq#e z1nk)J!kAxh6l?TU7l!y%+g>@&;<6yote6@InPkxHzu3~N0Wkz?!v+VYuzF#$yN#Z{ zTTImqOb4;^1k%1duSRcI(JCW2;Q`H0#D&a~c&9WRy6>_v!$`!gx+}~B z`wl)hJ-qZ(r#EJYrk^!9RYv+|-FULZ0ur zKjK1_Fvki*`GfEO{0gwIXo6V;U$4(&St{w*3&LJ zr}X07kgMEx{CFVXmEZ;S=gJjFLOotS)#+yi9p`Z}@eD|nHV@6HEs1e?Zn~jWGnV$; zwMA`4uk|_D=%IPo`<1QO{m7(2@sJ!;@$MLpRvtEY?OQE0nWyT4xnohC$|UI_ygl!KVyTKlRFMiM#gAFBO=WTWV+7tZB2 z`MJptE{7$L4ibDau>K2K&DKh-1UV%K{hfPXyPv~-KG~*GKD99 z!MIXN-T@vK&mnb353e$LmpiFeN(BBuXycc^+X}zp*W5AF9m&bx?^J73`*q!{0t<)E zaa!tIu3Q$W9zCj{_ns!QAZqO5cU8UhL{G)k%#o4!bo^OcS$)wm{-M+%Mn}O&Dq2>c;x8o_tPRNY0#HkDSWU4xqU`?@4Tzp`1!kGs505G z-0!{`^eDbiF|C0Eond1^0|rU;CqL1Oe^oEO_(Tw3*R)({y2$B@2w2wUa*wZTY2NVB z$UHu4@N;Qls`x8Jr6cJ>azsDc?K=}${?L+LomS$LuX>VhU`8*L58U{6 zQg-5reO+}s*426$J}Pfw3@ZuCkkw>7UrvqNt9VS(7o3cKFc2Lrv+Z!`tL5%QcR3*~)Y=r{iY6M4rpiLHOu z`smiNE^W@|BTXahZm&zM5Ne(rFL3!Aaryx{IlCcSRwDdBKt)tMSGHd98N!6V^3AVF zwJX+bUl->(bj6K6BN|O^-(MYUGmEG>_)@{zO10nH^})*Njy8IXfcA&$&1gqa^t`{R zScr!13xCP}ok}70jM^93E=UdTC%@YrNxiu4O*!D5;>3Zi4B$tNIF+@*xc zUuDU2%n&6QNz{-Wm_kIivWA4H(l#+pI@p-8pmOnERT2bqk$x3GVtoJ8O`&LDndYGn zn4&=4Qw+pQc%YyuyP{$saAg6u=_3l>{h=_>Su&-HL1_>sB;K4g#9u&F4P>&Q>0Cb0 z|2}L;&Ta#!ggoD@yZx}&V=(}D#^yV!0o;1n1kKznO%)~^a0fdNgOuZ9ctl_y_Kly^j^NM{+#|7nB-(+rRr1OI=QhTqt8oa739 zuL$E%4;5$bF+rJ><^i%#T#l<+omK|ZYdoeS?1A~R#Dxagtn^9~uU@5N{!(dH&NSaU zVw@U{JMvDvX?tbb%k;vStiRvT)Wq~m4gp}vi_)EpxuIGh1~YPoghm|^hyg6|mn6+g zI|fP)-*(AxN)8pkB=`pDDSAF7mv?W_`VAB|N}fL2`iQFZtokwpV|hw>Mtr;&eqQ%t z=s9g0ihcB3c^46c8a>t8RHKlnwR6K8UGds6CP(>~lQk$(+=Y^G{zR?|)oH&*$H`+j zYM)zU&V?wr5KjwpUX?~$!k<0^QQUp|Kxv-QO&oIQnQSfKM}P3Bc}yWB{wHPhQx?U! z`K=7fLOT?1x`}9IU>R8baDgDPgP7gWL#f$X&m6zr-Z1Ze!?SXkL1l) z8)x5-O!H6C@boqKS$XwdV1&d;3+-pJ&|4Z zS^B+7HDygJTb$Kld@rSxPU^=vQ>4=Dm07*|@}nV#rib`h%DFU6-qTDCLYgg2^MH~D zzE3?A%?fO(u|{L(AHqI4Q)UvAToMSrROKmCbi0+k)gibh%I(nel*EFEIyR%jD>yg0 zTTz#;=JU#7$!*_Az?N#{OB2{tS+2TuRyileJY3)aPC!t z*u~r0(Ip}B1hBE2JHrRwv@>%d@0j4l7l?)E`@R!!=Dgd+=Yha zEGKc6lVy50@!L2sD04|bx>xUEwZiI*xj9D4mZs3<-AHf>Y=K}h3AFq!?cSru$hc;z zwgdm80#XnCI#o=S6gZzqMfHLo^{EV#5mM(W_&ac7V0HLU5>aRwtO*6r!JGn%yAVu_ zGtCSIV&I%e;AdY&yAC<=gGoC`g za-$LLH(A#e%SP5{u2SP2ywR@>N_VSiszd_2VsKo&2}EfXh@C1e5Ccxobily_1{2M} zeE4#%otcc&>yf;RXm@U+*I=l6LKyo*bNGa*tN|BJ5w2OqYAQ7zC( zkKl*-bKoioisQl!r)I~aXtgMZMFKQ1K-5N=6_yQupt2!-;jC&LyTEqSs%z4eVr{m4 z`ej9D_q3wUD*BVA|3RHBMXn$x{iCFOUB&Trsagv$vmt z7xN>IS#?#zpIEIH=TnWG_yZf#Eg#8nzf4+pT9cRgG*NYuRe}an8XBj^va$Co{`52_ z>vE0Y%9&Z?3FD(Hy|_F@K~^s*X#Wj--=3%VUat4GlEF`5O&8A#mWgLZz#`cy!FX&L zc=%Hwb$v102kA9)IGyCO;go*0z@Uga&Kp4gl2}I9rl-Maa1&xyJ@NSOfRg(BtIO3l}e`XY9hCuzs9hf@~igJa=QSEM#KJB92#mdyKMN%!&{U>jddd z?Z+>5n3_~Sy%0EWI&}GHVp?h4tE%LrouL57irI5F)@_^181oGu43+ZYdZQ!4u(h}J z_x=SA?HsOepj6KoPG72&CyfPG=zq?S>9fv^3zHd?8n1Ykb06L<;lK( zd_PiLqafi^W{f-D&&X$=wffqLC9}>$Q2ApXuUk`7_5xxvRj4y=m!I>DOf;m0UWk^f zc2`p^ST3oFe4Q&xN5@ul*0F7t-uGO z(^xkq2w2+URRK9;I4}U>DTrmb+$JSBhe#tzV8!_<=o$YDByECk25jrNa^r0>$l^7V zx?Gw$gr^0gpp}~F1lJxa_t-%bmzI0Q2g%M!oGc}dZ%Jm_hRWU0D)I}X%rE*X^A$8t z;fJvAmhJjSn>gY!6OFP}p?*g+lDl?Ud?QPzX-n*Q)`@Ji{Vu#E`C%e#uIqb^_)+V5 z)!+TJvF5jDl?>s}-H&B|xu6zSIl!zM&^^7#vkZ?`b>|V)xD|d;ARrNMo8;W$7jGS; z2VSg-sSukO6vHp~~{F($te}Lp0=w5Pn@@Gf*=(N3pjcz}`(xO{)VU)SBN9RoC!B0z{7I$^a zjExl#rSKsY!WHkPL-4({tL`>95TgTU!4wVC(oD4_>XWf@Sk!@S2G#&cFb-&CB2E zfHfYt;xXK*wA8s3xPJ}#fISBqHKQ%V{;)N1Iw1LRY4zc$$8&q7yj%j|fC!fcc&Hy} zOO9QxJW=w_4J^2W6?7~-J2+V7cZaaK!i&P_A58uXPwV(?;Yw8o?S#j@OAX)e=^CA} ze)b$}<$NO%@W0*q`!Dd-qV4Sdn(eI`+#}nyE_oO83sHaO zuc{X;7YFzyzxYjj^w+C0zjNbtgnVx1wEygaVZa-sZ)JLVaSw7L$^c}V5JJHLU1K?b zgF-f)gN|uStEC>DjyiGiGyR^KlDPTsY-#<=?qLP?z$Zq`29!c2kgPMUjd(kUZo@}L(tj$)UQxvK@t@@Kaum6rQIpQJ zI;9kGQWwShT@Hiz1%yn?_<$LZMrtt1RWyY3gI#oA4fI9#tLOT>1}f{#m&A9E$Enjs zQft#8l}R}4P*yBk*ZDqAJkfsXU@crI-3Xg)sl4tR*m8+J*))-E&%fav;mL=+T)uTB%!XVm zHf1o!{oU=YL35(aoX7Dg@p5(uklBZ@%*;MxaP+hKP&2$kU?dV44LJZKEK^C7{L6VL zH?+7UYTV&1g|eY_5hlL7v+0{ zwq?TBCB;BMEklt6fY1jz-VHQwZspQLSg};FCX+A*uxwCWw97Enu{NHfv6@$a&~mfu z6V0IexBR4UiG@e%-|z+S*MnUjxFSC-wT?Pm7c;Uz!xo8^v--$hlmX@wcxYE~W5)+n z(9+JrAU?8D^duQ>1cO+>J?FtsR^J zIj>g{6<^K&*qQBrDB9|wo)p?)^UL)`n*X^NfPS#!MwWcxmz@Khwg<$)haH|~*H`i6yAK?NOy*4G_-*j=8@E;BU8c+PA?~|dR#&IBwb8`+r$n=QJn%3q&X5nz2SZX0KmfQ45C2#$SGSC=Lm4JG$U~pQzgs^#e*hj^wKS z^h;f6CXXbZ;kWxtYM2(4bW5LgyWLzr7^&8DXrGAo47c)9$qy@s79YW~`J90SoRM&p z`5I`XNXvk4JqZ1Bw#D%QhBk)G!ki)0{JmDoU50Swpm0!eT_1Y%>mwFMu*kI8<_XK) z_Ck}PfX12!6RJ(Sh2;r{QPV|k=Z6}c$U3B({>22}D#YuW8#}DCn0oDbC*a9qDxkqp zsXcV`FkpYbH^B3h;!;+dlm{hT7q=Gq9$hMcsF>DG9$i>etqh*gs2@Lq>}JXs^E_fBprO+N?G@z0cmYxBcDtsgl$1(+_KFBGrpf<8+X-F1oqrwu(JznR8FK zm8a08M0R=Ul80ViicitraQx`yrf+-}fk}T_W$Ewg1g$^8_{PMy%b_zi)*R{Yiy==c z`Ay`OU67ZFn9G=jsl{6+M_R3Sz602~%Y z(s7=tA5s8nH3YoL>^W`<-r6B(za^{{nh_bp3E#jvvaq*RA7z-Ure{dDGKUVss-PU% zD;Q~-1#AS(9HpBX(SpIOM|>xL(tPSJ6Z$`pA6W7uVSp4Jrvfp1{Hf=3IRKn4#Ace0 zbzrY;^TWW>|J}{!Q)PM>mrVBDZYK{ zZ67B1{Q@;MTQtJt{HdZE*RW;d_0!BSR4ups; zhlY}rHe;5FN?Fex!Uz-4#!g)xL=Ig7Q^lQlE_#(AYfnc#J1>kstbhl391z>t%rH9| zzvgjt<0l%a^b>T11mmp0lq8)%{A{YQm}XQdkKIJV3pCk@)WR5RmtiiaI$M(>xRz~% zF{Qv(NRDki7l#)y$q#dPQ&Hf<@}rZO&6hqzSGYEqbC;`p zv+qM>L7PWnu#jr({6EG#%|~H9>hF>{vHyD@R46le(bbS4GZbil+{)}Z&0aXo{)?Vb zwKwQXUm#bdT}*T0k4g#TaiXhtxN9f@pUGsv6Vx$v%*0u5(2u}rI?r8MKG{>uzJoS{ zm8{xyp2*2RyWz+@gcbw+IH>-E>xAqXP=0JgBP>P19;MhTtLI;vq7}SCODkUS>%{e1 zq+>3oWI|ZB&izIMz&0U(diZ{|+tY)`5609ZmPXoRxBhd^;_t|#Q%cd* zoF{YEttvKM&O3&y>J96uXLYLEZ+wZ~!499uTAu$Bz94t*=7!^XUl@5RYco9dqnC-px?FSIri_3 zGUJC_zAYEt8`VZVR}cYrsQzsV_t~h*E&|JQ3Gm7&>vE__WDMHrehVXl+;vK==Ny^U`1^qv|PYow^Y20!-~Gz+w}DF zn$yZ#%*usg$e)VOGm6F8QHe#Q#kzmi!w`GlN<##YwYY19XnWzfwDlk|JYSZEm3lK* z0uL56D3hoTA~)Nq_eO_AF9j=%0|sGdehc^#=JhK*OH*Wo6A;!oQ}+ZYGA z^2=8>T|Q{4e)gqFdHsgd^OITMGnY%!G;5b$IMu^L$r^-TVt137DNVe!1x@W&KDMtR z<$C&B*855w#ov|q&VRl*BHR~&gqOFTdHGC@?vrADNxlA;@TVI?#nbYFQv_4et8$MQ zHzO`c?y3lXWIrf55b$zo>y5`zq*I~X(GLO>H8vkF+y3f1upSh?A-&V6EZ!x2cm3)S z=A&FS2K>#Ax|inud$uNWp`%-|ci&DGm>^}r8^crQA1>*hp5Gt++@xQ6D4_&oq-~#Z z)bUbQ-1d3U^DAcM^;2Dy8wWW&^LC}ksmjGPE(fX;e;S0QbI)Di&%aLy^gx|F`aBK) zw|kUQXu7@+3^woJ&ScsBL20({7mq834ZitV`Y+I9%*=SuSXZt?^X^!opPM4yyddK6 zqO4~s_WE>b`P&}1ug%Dg<@{wW&j7kg#M0^bPvn1rM*2;`*v`AG&)4pTJi(IDCI_!o zW<8?%ozX|KvBR9JrY!&Q)w3HQXXyl^9AFYXyI|s3)Y^p ze0Iv*DA2Li&6M@#5+NnmY}7%!tByQoVP@D%#7Ihb+~XvS(-EAhpz=wv-XN=}bgDVL zdlR{p=kod75M-U=ec{^UW2Ygr!2+Eo?z$IiG<-}iO@FXalF?ms?ctXGvv_N^>Y-AJ@gMmVD1}q{rX4Hqr7Dtl6Xo-&E8XHx3z@omV zFnW@gB|x(TW#(hBmQoH)k{?y3m^md|#}AfKsr$2Y9o~?N5=mfwKs8Rd#^shqaNobc zk}$|hNjDU|Bv|uWQT=rE`%9h|{3>^z2=n;wRZX8w-k4}aEWVoboXFl>9goO4V?7mHiwEN3>-9$Fyru|7QVY?jpVK=sNXl z>$Vvv`Ze}<%F7sNe|3=1MnaIAFIoHqHc4n}D zfd=kkgf78A&#RmVF`(Nd`LoxJe%RKJ@|MC7ovu8UtNvVU!=?DQb)nu|=+pe<;FmJz zw_W{{7WhINtCWYD20cs7f40+d!mF~98CQ-Ye}K?kH*dM68@nU>QOs>Sv!4I5RlOkp z8DJOxE-f-u54^DkVk4WZ$u^tz7;5rET{)XH;uw}8EFo zKq07_}A-}jq^59co!Wovqq zzb;&6d}WwW#-BE&oX*P|yGNjgJv(_@RCrPMS|_=oVp}aNzxlglY&c63m(bO!4UCAdM1q*7+&#M0K$Lf+IHn5t##BSSmBsgc3OxlOy1kkjk|6+$ci zUqChT_lRJyU>`}apE#s|fea)c>JcqyeB|>Ev-+NK9wo2Hr}I!Hk7|l~Z1^s^3^>%Y?vTT~Z%pGQck={hu6 z53J0lUVbC9Fw4kZl3dDh)7`$idHE$#_3UWOL}R&2#U`(St=Ma)KaceXS47LoCRo{1 zA6`qy@SQE?qr_aAD!Y3{@9d$(_m0EJ7dCdo-4~y|mYOq!@BX}Qcpk0qx*8-f zSo0=P;e?u*o9PSgzrefH;(+UE4`K!%yhwC7s#h$Y5K7kTO!w^z<@i&EaeYrCU^6~V zausa{23J389~OOW+ES)h3#0Jenl+fNhdy@NSNk=-UyVbLT)aHrL3@;<>~$4ntnXX@ zF5L%TI{#qni;I4%?>jXv-C7Wsa){;lCsCtxUlOX8xAjBMUZLoG=sf1!Rq_PJkJP013T)p+!v1gtL$|a=>)e zuusRTt&ZnON-G0(^yPd8vJ^di!;==9xsLZMGxJzB?Cxtv6_?Tb)_LySOFSt{y(KQ#8$=NcCLfz;*VE6-k4Y}U)m0bp$zg6q z9*s0?Aa%YZYM`1A=byUhmOWC5a9~E3&R5$i3Y80%t0d7#Y{AkrGqK*o1{`9Gls@LQ zKQ{^Na$))~pzOxyAuL1mFe=IdGhj;;TNU2N%-?ANisNm~MCkTAW*{99T_R5)#tjrHNd;nGZbqWGlgo|fBB;J$)<3( z42?L?QatIjch01SdwsTFXXZ~8P?FcsVLJ)J{A$DL-8Zf#6So30vMX#Q9E!aq8s)} zs1XGdH-Zk!Md|m(w6uSgP$Qr z+{4#-Fd0OQ`bR@ zPvP|F&?KMW>x6Ft^zgfERrqrWXM4jpW(Lguk9Mc3eGWX8z)_~-sok;H$!5sp!GIBd`@ z1DaH9b@;k>&-zqTVM2IZ+2rGz5Am(E#GLjvF6c{Z#KS=Cgu5qR(|BO&?k12Yr{QRssFm;+?)3(nfG zC=eQ1cBQLVpUvuXglXK>?nPjv;)nPW`T6!kP%8`u%}Gw~4$iFw!W+TTNRo`<$k!n- z7vUbj9VO0!{K$&}yvAmi2yU86_5W5Pvw#WV(@PbxLV%>a1?7w;L!*5pA;jERkQvl^ zHg5#vx^CMJX;2$wQB1&5h5^B0`?Rgqsd6$h%>Edo-7-{{AttyBtQ@ClDo6;>XqnCk zgHE9NiSv_veS#Gajl7bl_ucW6H5xJc;^p$C={#%8_|a9xgDqwr`^8N)RDKSSM)LOuno?BzIZY!ApH!(Ipt#sqR{2 zn&Z(MLB0{HuIKL9!8{n_6lzeD*=LFISsWMe4k89S>WbDSQEV&km0)O=1ng}`T?G6Q z63Z+jIs{5Sx8EixX1>`SLnzA=hjsUTAsqGav%fb{grqlHlMi?j5#neulQH}uF-+!y ztt%xNGv)~XLm0T*>&UlQr`3w=>vvA(G(PZp)eBv4{M1IF}JiCQS{Q-Z%_bCi@=yr5O$XHn$ss+oOrwbx! zkeXQPayOG%c2uqoCQo!!eV8;t!BaCY7e8!L0cQ(}$VotLV#Ofd2p!S z!~5$*=$7D85CR885r8`hroc^MGR4{XR2-UOi|Yb$TSR~b%XFfm12^#{KJp}x;7I54 zG@&jtrLZLQjM&ie-^CeczhGs=gYMES<@{w=!+(#gL;eHR7~VPlEi5LNQ)nH`GsGKE z>&Vr>Gc0Eci}v6KcFUEPyY7hS)q>zi2tR#(CrH_5Hk=b;nV z>e2>^5mRd2Z&1*^!;p06)acz383w?ui_nGJNz5{MQ~&hj+8vt(Uj|hiZ^nn=aQ<85 z1lz%Hn@K(XwF8zV_TVA3^DsfSe5qb{wH(RyfwSx3WKSP)a(QH2-DzQf)An)U#4d-N zFqLQJBAqUFaLTD}FmHjAdZ5AnqU&NEJm>qEZ4c7rWwJINO zTUz=s#jAnXADA?c@)PZ!jyuCGFOnO;PhP0{uAkzD8s>guYt}Lv8?tB74Iwct2U&Vw z{|R6Gty^;K(}`#!H-OT5U+mrXIh~w<lMbjvOe55&**QsgmENYAv=tZ$?t1AeS3gFjT)k*YD{D5jrcidwCjTpio zdD0Nw0D9*aAam|zw_mY{v+QE2n@(^2+m2+n-})CAwxIlU!!eGwq@Qb7Y|`sa@`KB1 zjwZQ3amXZ9S_vaINe-ax+k*mLl;og&1Lt&6wzB77QLqU{0&S*OLv z9o$DRB`#ez5}l?3@~%1VwEJ)qU7bn$j5N1PHTb5U7K3ta8{AT!OG$wF>VADRMYzti zVm2sw^vWM_^3Nz{O&%$BA;#_1=DrvWXO5Cm=JOm5Ky&|CWMv)$mg~%vsi%VtVDeCk zgSDW8i#*LTe$y(8YR~hQ=H^uvpJv;uc8#zKW%*-^|5O^uv6a{^@^sBIfdng>)R|+y4ua3C*3Rj8X)|4V zr|1{$ez1-1NcG^eedWlBqL#W_H|fVA-)h}bPGsTV-|E+<><=a9iptKmjji03PVS0v z!+c5YSBl^E!k1MX0uj7;m`gf_mXt#F;}n~cs0?Bp6Vz7{O{DB8bVwOgeGK9C7A{p8 z%%Ii0inmGxjTJO#c_-3IeRL}=gaa&r5OY$r9T>lcO}3o**-54=@Ql zRy9@KIyYW<5O4d@rV`f<%WKt$7RlGb13oyV^CQ&Yy=QrZHWXgjZ*}uOG=Sf}P^CiU zza%@6@=?V(5x(S7WaQ}LtczlT3+aBj*#eFGOF?>A7a`0Hmf6bSTLo!dGvN15ldSi6 z7BpcR5@xPUcRPcl^;$g_71=O_2@=s z7m6wWpeH+(4+IZfn~L%6kgFx%4oJn|}-Gtaob$Yiyw7kZEwd?lDiC3HZ?3R8Ek!>qhuOJ)7t zFDG7Z7JiTSucYHgXSv*a2B(y%#~Lb9uyBvFlOvRu6(0iVmXFTvvC{{2*LC)o%FHL} z0~A?$<+oV&2ZNKzm#eVxw|PFZOL&y2hQ{rNMH(612AHMhSi@8G$*VEd;j;NPKV%k) zhTPIWs58a2ZExQ@TH~7!c>sL(-ZQw-6?Xcu+33?BUmuIl*G_rah7_E(TnQg0-{U45 z>|ovNgO;y;U051kG+pobo@P;dF?gDZTuGI`0WIbpn{Rxg*Sz_A0;F~={B+}wep#E- zzlW-+^byUN9svlxGulKB6Sc^D`5v)YXa1?xX zNY4CQ89W5}1|lmZQ^+!Se8`~$@+X7dmynR$qeb7VwR#yRFW_MXW(MrMh|K~X1rJ0as79DA=Qqho|{ z93$Z^WQK?oso$&5_mAJ@I*O~V4o=T;f86i4drtgQ2~x0*_>|XFP55YPClj{+d57+C z=YJVIDFGc|yC}M9Mk~Tk&)Su#A@_2nV!BsTQMzPFWpw{kG1+qqrSN1BN9rem&{GU7 ztTqf`yi@D#(W5Me>h4ZCWPe4$+0V|7Leq+J(PS>=k_f?sIW}wh+(|9@!XS{_N4PKH z*M_34M{lR>tT`+?{X3dYgv*J#*PQ#3RjYL==QnwFHcBk8j`z&>2%2XN%xx(Vqc#r| z6X4AdTL$VQ5S;aeL{;6D6oZMDdsqn1`$@+Euz?vXMJN%&x+WbD>=9yY&i^T}tK|xs zDdXX~!W+OEv6R?|3u1<()9GHNTDpPQ_c1|-4&7FmIKQID*bCD)YtHXNUM zo_2*o%GhUG>{0c+qCL};hDdBtK6Q4_Y{hwU6E!TZEmbU&5vtxxE!uUoZ5J9pHx^$* z^3=QvO!9cwT zvx{kvE+mqJt|7n2lOlBn>6<-_WCSz_GD#1pm+1RktbxXUDJlC_%WkIA$|A>g)Bpx3Qcn|xi7<93t?OR-9U^| z>fJ>*mTA^j`@rgIZzQu)jL)V*gS2=A)-g_se%fCVT`QB`9R%}3EH>N?Cwt3cVYL(l z?gE-u_c(?--kL{bT(SsWA++fgbzQIRi+Kv!3}o=8ML>MdfC~J8|;+_BwAd)nt_BjZ%grxC`7$x7^uE0I~=#ORI%VJ{Sq84@iO-0 ztCm%Z<-_RzL?ia~`ZQ+c^Rr`n0~Ex+G`L>Q@m!fNF8-(F7+zf!=cHtN9TNh4IEN-` z9IhJ+0^?E@Jb;#O!y#k4Hk5koTlOxGYxTLT;S69j8RT(bo2jk z|DaGw$Z6%raucjBz%K2um|1`{dpF>#>}xm0(PYrjz=)E_a+RpV12?bC<&(w-x{O&q zY_0A`4GRQLW+3@UdGV#4VD0?wcejziYJL?Cm!ZMj%pv1irT)f>qc|}RlKAPD82cVY z#urpsXb!+cKO0KNrD0co99B9Tqh%ko>fC?1Q?R|`i-k!!JgC)0khi#J4c;R>BE&X* zQ!-Xxw|)24N(>i%>Xdq(-67=7z0HTRR&1KK6Yk&Y3NI*hudAr{Tt?7wDSW z5==r#nYzplGv8KYryEY|zxubB8SES^pO^81ReycqA~-fJgMMRiQ7>cO{*nCN1+3}? zW5bJ#i<7VT!k?M`iSm@b=yKr>?_kkX16Dp$YLM7wQC1kuItk;(x5 zRn>@&sAnPpm$MpNmqvC1rShl|E#4>zQ=qZ;XUT~LQ}c^QZs`Ui-5Z3h;LrbPHGAw6s5@>_Bl7AJAP;iqit!|_>?Qi?<7d`=8HV-r07-2MHM#vgN!f52 zl@~Do_)GcJ3s(EnOeTt!JjLU&fA?RIw!Zq7+E3;Y*AK5~TaRTgFbgb`q`VTmxqhv` z%|j9z6{z~;aEE!1uJJ7-2TJgXZ7K5@^4-EoA?ZqP@=6*-V9^L!xmlXJaMFL|J>w6PiZqXxcHh1muUCN}EzK?sMV zW69doagHH-&ME=9O}!eVDtc0xm7qoCO!scQ0o3!HVHeDTO$qSoUdM1i0f!a@)N7#TNQds%(wp?Ie&HV~FHO*~USK_rp#6hWdj_5I|l#iZF zq@(ieqB_>!zLR_XA16RAavap5J)MlG<;|TKh5ouH$hTm7vIcR|Z*JE=DGe%fIY`-$ z7qo6LUy`H^v0c`OfDOZjOW#1Zyj-&$Sbs&~FENJIKaJ!$Briv);tC!?zFJP~a#ewD zVfjAYZPo8-_GtMV7W7}!guD{5V)O=YmT;=v3zN^2GIpAZ*#|Y5EJuS-hya#> zEvt`$SLSkTh)gv=bK>mx;a|6Yx&E4L_oUYs{zp!Rp#>Qr^xxTT7#X=BfTHdI3|h9O zuB%U80dLpUFkqOw#!~w8;pV+oltA0dQ-_BX?U4}M^ZsF~Lr{%B+^*U=-mH;zzaGE( z#ZSv_$x1UJ~NFvo*;rO=_A&5&mKsE);z2x|LRF{V*S zFD>R6O$L1SmL)A9Yj3xT!NTW?2oSffx`vRT^D+iqkQy;T?UFd#6?pF+ks`Ov*g}U8 z&5(Ww)maij{_v#6}&{x%qtgSBCES%RcBzPjFS8+{oE!GpQrE&=y40^3WX=hOH=#=F%>Qd z0A$hCUr;E`Cx6z|91}d%z+`hw{bf4AR2~Shlgbx{k& zS%}Vd=IXe!ZYcEh3#2Wi2g`l=k)d5ITyFmm`%E%yYr`eoA)=5Y?4S%gzG8jP($+mi z(EC1FurQ!W1yCh{Dkd%nkJFT@5CBKQz@AJPG4qvVmZX^|SQTk;T{gXp;z^^JLhQTY zL^t$xMTe?xCL`%*%k(6SLvN~_ymbQH>s+ew_4#yhL*!X8%&;mv zwS4~O@Z#z6^}iqy9||iYS$ut2r_o z|0>5rKOEfk>ZlC-eU$E1MvJJPb(TenWY-0&J)6v48cv&>rKbGUrIF=4>VcocnVDl+ zcD}gV??%yNBd|0P(mKmjcZa)5#$+emC@esC$ax2TqM%rj{&*Q*XNYGt5s06#F z6>UOSAnis|t)XxNf@7kuTWH7J(q>*ySk+oOMnImL{9^Ncubdu4a2Rj<Y=}4JCQG3Cf=+aOlY*X}T5ZyNscVRfj%PxG zdh5aO|L%NO$=xorMRi7!gN{61+lvdzC1~ho@Q09HBNl$^HLDB#$L;*`&cN|nlA$jI> z=_xZIMj*O=q1BYc;kB9wf!Z^I?`fBXK}LPfiJ+>o@eKRm{-Bi2^T9#vR|nRTfL`^c zMk5Is0nZ`jO^kmP-{t)4E(Tw+X&+`5Lu+cfKTE_u&0MWZh*b?joiF@Ec=|HGiQ0GW z3Y=bvT!c4{>wz9mu38s6XZ;=@8t)H zbyY{cL*3_*><-S8JAHMreXiQYIwS?(gE9N1H)n=``@`xc6R`437Lj)g>!x{=!b@OIWy^k$ z7YM9BIHx<U6m^C?wdt$Wk#HCM}fe=hpv7a19TUfEmjBkE-!5Ht$(^zmy`-!QRgxoJ;Si-h*; z@q%IB3Y(|zQewJCdFx1<-OC>(Ktr?H+iij6c0WK?mp><5554k&=Vs|d41SUFd1DfL zh4#X?ur#cq7xasW+E}QQnYyFs^~sm@)9H81e$uCR!9$&|Ggm90TAvhv9V~`0mh(;% znSHJ^#8<2BIoZebQI=Ab06ib{TW%SgYI?KhwO?^QMYn_FoaOeuy7A9Z8?**D*30~NqQy$jofLb+5mJxf6!&qS^cJx=?7@a zK>Mlk_V|gF8D#T85Y)VFrn{_PiQnC3$~xe%i;+N|Jr{4?3V;2K#dc7nsn_zdxq}}` z^1}=_!9|{P!V#f6Pw_Pj5sDF;W;D@822k~RP$SY`4@koDXNAD@8UQinUyRm`9zY0e z=!=2wlVIrp1H_OndF%iL__e}Iwu?zDpb&#_pt?jI04EnD2Xm-prqX$ZCHrOQi8Q@> z`1vHVV3hg&!lW;3C8PL@={@(o22$}P`?6x?T4Qj%I7p<)I#K`LSMl}>(z+qrcGG_Z zKDf-_B9=Obb=30wHJ>J4HxMX06M|mompoT*q*l9a9qe26)mV}#3(mLcJo)0LT@Z{) z1Kv#0&!|jT-7_MxEOwK`c-akyvQNR(3lK5(6wAZ{Fy)HIB3lEyW1Zb(Woo@Nx=xaN zM_b-wRt&UM@ky6S9trE>U9m0u_~M=$;09**%}cKtj$MPdUjf_3p83mBd6DQE=(S7&C0m`6Oy zJ>JgK41Y~74OsTUEM&>&UUuDwWt&|B+^-_sydA_&)D~Jjp!Hpb!<~RN9`%eUk}8;u zzpk1>keQ1UshF8u_qI1MWNt0{g1aUWTmBB_KiGX(z#&AP^ijN$mxNnr@`SNFARTT1 zIT7IN)ZLUVisH+?6|~uOo$K(-wr#(-Hs!YCUr_nAT|p*OBOrlGN6^aJ~PA zExUYXsnMux0Uk6^z>pXSIF>RoYO-u9BVFaecOgt_Rmh`2{PKIh)W|9Z4Xp|6}KwR&4>-u&$|~+*m-?0vyxs7 z0|hb#-C!58Y~i2by~>7>%1!I8_RgrHCI7m+Z-&v@Gw#?v)5j>ja*VzU?tV|UFc5q> zo1YSAqlFU-xvjH=^Gh6y$=;=?jRzWj98E7acV#kd5WTrU||xENJDOU2eprp4wMj;_>eLgu216 z%n`9vTE+eEA@ZE3Pth%{3U{SjZ+ro&7_>XL>{o0}$Td|%wa6zLq&-zTgW0<3pA2l( z!Cu;nwUbw*S(|Q{Lh`L{34=GzV0-kgh)p?cErEZm#K^`eGvY z6AQT;rZDfy!Vv2_Rp_0M#kZIx@O41gQ%G*7sWsB}4^Q_GGA&}N?{&#@_baHH54qpu zIR$Q}L~BmNir)`=RSx3%^^8Pi7*N!-S7w-V(t3WD*=x<1Vhz%NV*u~0&A-4ajM%Kc zlo#?{T8pNv&XIVitxtOceXo_D!G_2Xqjd?fhZcUN|J~$n?e&B_P=O7cp=cxnm=&pD zz@tM6hu_t(o~q~&rZ`d?U%o!UrI63eCm}yru1r36brZ@3|Frp#_4#FzCHAbN0Q&AGd2-OKZ)frJ6fYYlLD(k$V*$Yc^&dhFT6=b8G61t(!n+$_#k_ zd^><=+6w!YBv>^#4(lcA#cbQ1pQ^bXx~IV}%vDzN7>o}i&pH;oMOT2gOvOBMb+-(( zepL6pGvJXXaz~BGLaLBUg21I=fuv4Rj>S5j$rzPR7V;*#qmD?y2DKwf=j4!{Y__}< z8K6k|WFiD_y+%+xPmeHz0rUm|BmHX-fYh)IqiT?qjSOL6Zb&rgVzvPkm;Mw(Yh=d( zXPQpTHcF`**g9Bz(g`S28_3rw$*5uy1Px{_QDNGYLt+ZpRu!uD^Gat+V~ z_dM4+7jQ`NaHfDzm@(_sIm z+jt}j@U6<0fPq`w2uMLnpa`c0s7Ma-2fbtLD19az8x$I%3;s83kn1^}`3qVopL93h z8T$)r(uGw0{2q4-gDV)GKWBFBlQfA)&r+f(AFniG3^W+X4xTZb$RLZiL7zU-w*!i$ z_CKv%N@rjNG5|{5ALQR6c@> z*o&GIgSOM_hr2S8xnhf4SlZ+rF}^JZX7O@pdgxIfXR0o=*uUk_0cg4eH1rZ~S-Tja_r zkt5Z^EahHv^N5tREZ3?$p#HIH_*h}sSoX>FX5rlY&CJ;k{4^EZyCd@V1Bxec!u|?F zhf972iE;Sene^ZK7+(kiK!^gY3R!KS0vY&42CoWyg#49^jB6(>?uh&A1h45>UeTHS z&)^E5NKxC#3;65UaQR511St#Pf zGS0+#a9Mi0fihLv0pP;yLhRol~Tkh4*8yU9@&2UDSuMvLzHq)`Z8=L=lswcR)r|wd) znR&I3jeI_Mv`V}m78JGR#QF(+L*s)_pXLDx#4B?ZZ&s9n zVYfPU zj^A9JtWU@gt{_yO%P_TxA2$_=t}Z1BZ)Mi*vB>y!2>utnCvBluqu4A$(jM2ZBtXH; zowC(IPXTEIARPegSr7{8BZP7TI=ejyT4Qu+6zM?OqJ*xYsVp@=-?;R6!fun*y!U^| zMWXov_pYm9h8ER_+tJUwWcfd~~X|`Hl zv`I~Na|Lx!t(m6kmn!?SV!Q-YR%C{yDg>F`Oazua3uus?G6zND(f+AVK^WOB^uOB* zrE$CQQyRMtdvjNQ+|}~0jG2nREwwx}vk9PY7w}rP{Ly$uhH-!3*^dBSSO1o$9ALF{ zfU5pW2mf1RfshoyOhlOgJVvy|zY+$(YXqn?K&rYJ25Lly{qMbZJQeIS0-$)K7W7qy z`ftoxb#7^7EZr#$vT<)`_&mDoO3IJ&z^nE}0*PKWs46hj{UtVoYTc!p@zc0Z9kDKc3{OvOvA~l<^njBiQkFGpVlJw69eC z(E8d3prAExo84GT5}dMVNimvE7jygy#hLr=|3A#y-0vaxFFB zAqU8ZDySb81pGvV#=r!qDnj>M@1I(>VNvN~&ExM5+_^>^sCH)p@H&Xf7LlOFKmSo? z_#rpfhL^N=Bks5Fn-qp;*0Fj*(>xU$uRPl&>A$?av0-PGWOYmZP_eHY7qmc}pE;=X zQ6x5t3#~*<$Gk*q7PXUPPkl{h1CwOyA&Nkc;b7^u}FH)!W77 zmIgWRU1edNj2sx%RQ_tbebOG!RIy*(An~{Fs;`HP^;R7RaEZ-oauWmG*)X>~eH=)V zhXUc9HNRnisXZH3syr|xVu_4azKLo4ZX&@{V1g+T3>p!KoPz;|=zVAvHJN*v ziO77h$S=w$X@*MYu7Bn3>=pbyFfFHK7xFp^$O1La;Vrr*Km~XQB}{yLO~fq|1FV{> zo-POc3{HbQP@=Mjs3=r{2S5+c_F)#MwQN*{jWDWu8U>7_0At4jgpmb|jqXt}s*iNB zcj3i1xL#nXNcWpVyShD!kQI!^oB>U1k3iV&Qv%MRvXqDhqMxg0vN- zRRL>M7I2G35M)aj&JVRu#x(_A;YZBaJ`Pkp6Nz8i^*1E+3!6r^-E2%1R6XJ`w;D0Z zToKH3)5%yLspo&Ehi?$HND)Og2!#!4obA^ac1+ZP){z|aDZyqpxe@|3t-8>1ORw;& zX&?ai6~zcNG01ve@&B+TB#H#;?$BNGcE1-gvj2h};x-8JNb04H^(TN3{^jidNFTsy z(Hn0<9RmXk9N55z_z%kcgc<gEDs4PZ|U$nrhZzug=`T3L_!*O`e70TEM7veoJeK7Inl#8zgV}`U)M)gG8I19ofZbb=dbeX#OMh!4hyPKwrAg{q=*P zlTFa~8NG}Ab7YNvk~kV9Lz2@lo!?hBT|h1#l>E|l0$g%Z0+Nwvkqk{xV;l+cRZ%%U zpO)H6&&Z~9ADI0NjMk8ZHSvCQQCsoUux+znx(sUOBvw$ugxJ`LEx*QL@tyZ(kJ+k1 zloD8l&FRQ8GF~lo+9k)^{uAog!V`Pijh2my?8e-!+^mJ=M!){}XynT`FWl|D}S;^|8{EU%w<)hy7}z3&eb2sqthH5l$g zHC)mUOq13VS!!uC=NKXnOTSuaazVgfeSXa5guT*(FdMSgkUXm1{PF8zGap~BZl2>w zUq@vC-0}tDFUUD-^Z48X;YH+qb~VIpum9Rb!|zr0he!mtf6%hzavFD4wP`HZYLMOm zaH}AO{UBr|s$Tw@1s#ipd4w4)XGB4ou0(G%$jA(F>Qh{$bpU2IgN7HELT>L9!TJJy z%qRkAB+#CBI+(u^MiPH0MLC=E zzNIe_WYtwhbpJT2e7VKj(l*;GQFjQaN)?r++O8cfaQv!#NAfj|i`h3#)E`*==N*qX zuNU8S-oc&bpo)ZZEd}|GCwp)AzwfE&P$?B#ZOC9UCcYnz2lBMEK6_^N#y(ENxk{3O zQJiUV8Y+`wGPuc7Htcp(7U&bl_e^-m>cMJdHm=M*5PQk=DMzc@c+5hIVUW4dA<2!0 zDJ=z!>MnJYSQ%rbBLZbzz^!>7dNY<_6K_eUyA<71)D1MddJ18lP}lIDLYCY-@j_!& z|Nb3$p()jy_1!S<9Ksd0mn@Gkq&m|PaoAV-7H@*7QZtlPo^I$qJb4B#p$5&Z+psBu zJ)J{zk-3@fZ<#njm(y%3w;gO5@WDZ)7wQl2ZN6rWDy<$$@^#~%nN}r5n$m6uNa#`S ztXI68{87W#WNTAo-j@4tPLr(d~EdNMDclBt7{ z`UdJ<96SQH)Oq*a_2xgcJ+N^3rq}`brr5N$(OfLvD7!+g>2ezWYQ21P5hlf5cPSI- z5qJ7}!-quS2VhLb`eIl7H0rM*NAJ9O)Ko+4t8yRldJoFL9hTx_ES|j1yLOA4I&R?m zOb-DJ=FaGg1LSRBEd~1k>;|@=66&!f-4%7Tui%=Eo%F1)$DKp()_EhQHu!vWsX)u{ z^MHOmrm73el5DM6L2aL(2FXz(1oO2%O29u}<<)qgNtdd1N7-l*C`nIj8g|@b?@*IF zaA?0`=@ims>MLv*avSDNHhg&Dz}ah>v@ZQZkP&-*>JKvRwXf<#5XfprTQ z0fr#%Ts$DnfE<9`%D@~80M0$)!jg7tdCmNzUSVneDT^{ zgzT~2X{^ZnS(l@M7SN<2C=rq~qIj#7l}z^YxG zT;jPuuYS`L?l`hlHHd{07f&&4x!(BwydT~kF)f9}r6_%nY<~@m2)Nqj?iH+v_^g9C zrng99s{T3`&TUpwd~$(TyS2}+fOxm?k8m*%sW*! zgCHi#sdH=}CoU$+L#*81?HkqjN_UrMXcL=Vb4aq^wX(F8O-UoLT-&L<>w;DSu=Mgr z?z#dF_xIP@lE%N=4z8BHQ^Ov~Z;d#Ph_eB7+&+R2I1j3d#Khw0uuu=q#_tFt_Kn5v zQL!JeIv4%{grVp{#RvLt*z_O2dE)UP00S0JWj=c@PHTz1-MBW=1bVX9{XD*wh zL)Qc-FnS5CV#{r_GJmHFVb`9@4?j~GW)$PIItR5`_mx3@>o^j8$#`vjp)`=FU5F+f=^}+%@Vg0m;WKHL6M!xu76(mp888TyJPJ>V;fYQ;Y5!`^NlX`E6}w zYDg-^{snvFDnWP0cSVjCCUrNYC2+%xz2K~^&aaXmMfE1hBv>v=mr`6M1;IH9?HO^% zeh49GGl!VaB?UwSF588)ErTvarH4M(s>EO-Z)iC5r=+X-s|i&tfqhVBmxxHzL`jhK zCf#<|pgLS~tm-g-{FTn-*MG(i%IYK1Zr;1dYTf`~MBlXu2k)@Bmu20502Mf|-? zx%rh^^DB9*^x?m<#%LFOcQCXBBdDQwm`qr(c69b$wA{nEG3qm^M|^5BGuCR!0wIn& z8x|F?+rEkNftVHrDqqQ9;R2!y_a}psXlv1yDS=0!nx5RwfZuNvHshPCRr)bNj?H;c zaK0&Ql$sb3JYN`D*_+y8S-#eZ^*Kp z4b+4DCa?RVs?Hg>FBD`UZwoY%`+nMosp6yKJ&OMgek%N~h6)Nk zZ6@59u@#nzcvfz_0G(Iciqknfj}48x)K8nxTjpJBx%;uqJ8DD<@`MQwoS%_Q7S37`3V10Ye_3gcAuh@=vbEG$p9 zK+l)}a4CUFrW8GpUqI<{4?hm|ji;F(TiV`IJ{_-CMg0XOW!9K0+WI}`4TkEKeK#f? zI!LBn#JvlJjnD!}ZkMFqJqxc0+N{p=Rl3r7fee4xM;!UtRM?08?NAv2wB9H-7QR2g z_Sj+3^IX~6)P1$}o)UZO*Z8G#`!{#6G>3*7v@hZQgGWQcTERCRX~w##tjj00Wc?Qu zoGb$M3cF`=7DzhE6{iB<5?+R^K*Pxf>sM%ncSiji`dwp^m(k2q1*x0-e?c#bIoPH4I_zCUxdJcq zl$OY$W@WRbfSV-wWc+}ad^jrX+m=0lZEy7~&Y6pf{$Uq$fGvr?OL&eD07a#yyHGIIffn z#5L(or;YYNDIohc?ouh-kIL6CD0VO34|ph3?D#^E!P_>L2!ut+eeFG6{vY@1P0#i5 zlMk!pP2sD7RApJslgSIhYN$&o84{jh@K2aPxgP0U|7b;VH_*JhwQ2LmLS55aD8~VpO#B&Ir84T87G>6feB7wKY;3m%aUW@DZFY11!BhtKr6>{*+4G2 zt7&gxt%uqe^-=1JFmQyRS!{F&3ZYVPRFW1jz-k6zZ~0<@2u)dAtyD07XY=Dpf12dm zdCT8h>2D{i519u&A&=)Qr?*`4MHSLdH`ZflUoUs)Vf^#H)AjzeO|uLBnYg;~?(W@p zX=8P(wFNDhRlCooR37KdSEZkP)Jj(Orbi^MnUf+8lwbu1u4?KPS+GKdrq`yAl@w1K zdQRmF>fTak8NEF(5f{sZz%HI;Tjr4=C-pXB;R0J|%ygB1U?rmO{%Q>`y8J2G2OlQ* z$#c4|<~IN7B%UY!L02HQZE~%fg_jF(#(emqD9&D$FZUtDRm1kYiy1X*UGeTv64_~h zE{ESUcL&bhmK&e*^3O4x6RKp$`(aI{!sE5{m5{XIdZD1B1g-5XhWJrw<`aVggWiC? zhh?LvEd=6-`_9SrMWJ8u|5>6FOpQLl8-pWL?B^p7iV0MHeC?yJ*}n*v5KMVm0sFJk ze?eYB7H?9k02kvC%b$_%;=4SMvHM2{)tmNjr1N*a$R$8_LN`2@?96HGQ_g*_F2uys z@5q-VuG)Q|Kc~%wOkJE==&mVg_0{FOx%+IPqcueMpc=s-^gB2b)*wuY@RSN$jO%U`N&GjpT}E&s}QT zE?5adIo`;htnuemrre+0%t54mqBOF#Gec&k=}Cx0`Qh~HfwMNJslOtwCT;pMMA%+8 zO1ha4nTAwz8XY81=Nn}60i0c2G$zs(TmUHgfD(^hi??7YnfT=X0*`bWpGuC2(HCFB z17Og9p+P2T--`#2j+;hq&Bu>AXP?Is>*_8e8;c6!Lo4K6UZ+TsElD>EtcLJ|wUTss zc%RZ3BPJD~H_%cfL%<;u6xdT7zOlIUsvA-j2vu2nX)~`nTM*V=jZL2wFLpF`nk5uo zKJL-px?$qyKp0J5H8`G9e-V6h{()>FUbb3~5yIIRn|m; z@sGu@IjVOC?h#;<67d<4XGsgYP=1xV`T{KdkV9L_>ne|hcRF02{VnhJ(h%FC$!{PJ zUmlouiG_Knj(1q#W?ixh_Upazw+7Xc%EGKb_e~0yrPKEw%Vb*0TxIz@ztzjA>gya9 zz|M;lwl`Xt<~3o+>aiYm1qS}nzB{8gmH#*Ch6rBcHn~5}d`T8e>%XO!CB{~&@HIY2 zp#QT4zjyAJ7@l1!?$S`1ZU_6@F`Lj9#hnLDEL>I2rq?8n;zW^oejcJh%u_h!1lodY z)`dR@YYY=r)!qbmj^1B)cE?WT?8WHzXQHBq^-iJFe4BqkuXBP`yAOyLH|=if{BWHk z)INKrRL(qR5k#$lC3Fqm2F0Jtjx)mA196n=_=IAGi&_Cb9>1$+A9 z#ChEH@bYQ)K+q<(yL|KgEtM;C4lD1hzC6#(9=u~~+8{oyVkDmx(l!)9Xd*kmtSE5{ z?&Kpc3r+WVDl0aAZFUr*W>7}NWYmk|SH3$vNv+f4Q_eap!wq(CnFl{n)I2H)yzr$w zb5T)gCLmqIX^)zm8bP=}+vVnwxiQ1^qeX;XFeOl<-`5kBx^Ck$+_LdM?{ZAui?6l};_-c8`Pa01i zx&INPALiS`{BZ-N{xp|W0Dghq3p6th?{2bPF~6h+8rz z;T0Mq7PRW+{JwLn?Dgz1HXQGi4cJpA#T#bow3ZK+&LKmfizkz^wvLry)McKW>-3rDA;yW1WfL5p|6L@UdnY_Dfq1>=V-j}MCvi-My5yq}_1bCiB#sYKDx3VRHY zJz`!frL8jyXz9yLrRMX4KE;ODK(PZCMxg%vQF!w{K)x4^X60Xl%vyv4_zEC6MpB=r z$N;5Ip9AR@`SmO(q08R96JuC4Oqq_fx+Ve))jZ*?MApTJn|q>|c;I>+X)ehx1%#Z|b|{ z`o!orac5!@c<)tkJmUnb@6wLol4=vq0~ep4(YP9k^7VzU^r$DN`c^eLKIZ!LE7kLn zhW22Mf%2Og_r5EWU(G#zZ+{-*YIOGh{%qy&nx0c)UBwW2PE}P@{+irGm^SjE+2WwY zxwc=n)G{~lh|l~fTfnhG+;y-!{`Hu(daaO0_1#Krzzt0fder`ZgLfAXM#A=L!v^51 zr~%Kwp)`SQPJIY^hTmSs+i=9>_lEBaCW_P zWL10FalXl|_av^$+th#3D~Wlb36dPlI$C9Kznp#l9#SY`Ipybc+tUvVDb;Vh>ni3} z6hkQi7H(61nG-$tE%Dpcu(`}g%Qwm&R0k>tFAj=Nyt?5sXDzq48l%voGV|qk_t-Zq zls|>yx3DqPSP!;Ky&-q)2L64hlgTa9BD+3j*kBt|ZYrJf*n}5D4F1DAyMjlD>QLkeZ7Z7sJR#HfOh=2S-fa_P)EePrBr^GO=uuV)~l?E=#fGL6xyjL=Z}k7wdR4 ze5rNIS~X3kIc~Ob$%GBh(nXJNeG%ny3o)Op&dz43&x>P=QDl;jv~X2?KW`pn*SxG+ za0qXx$o++R)!w*}zIw{B@jxNxc9nFbh0m^&Q2_~fF*=Qj0}f4!7`-lO%OKJctU0b* za9Gr(AO%Us1qtxJ41N2CxA21;u~%ihD%M8X^ngii@C1wrpl z`7v^sGg#S?n4WYI0A61dd2z+sWm!T3QjEVZ0QqouiFp|=l(J%l#~3Y6VAsz?1fKEv z)AUm2zdKd_8!7Vk8dt+A5v>_8_{o1_*tCJ+Jb)0 zvL0kTgvTVK2KaOmca4UND?G}YhW7*UkJ^1IE9vp3uti#cTx@jAvWWM{?SmUJb%tS_uvm2xIIelC ze7ab`c7FV6>UdZ8_n_b92n+RIHaK|SQ54aCP+D7H5Spyw#oZ>`l2bKsCQUT-;Mkg~ zUEVPka3L-Bpx})Ac;EU5+h3MZBi@D*H$O6Cc5MrDZVP=|t}S3SRm)4%0DQK1AnOXc z4Vls)!Kgv@h}4-u&hzx-$`KFCI{|f<8Qb6mRF=sS&z^&7@BJO)wli%XG}!=?z`3^j zHedUQjq+-$6n^aujV*#}P8|GuESQ=`rU?7|#(;=!OBv~948R)Kd6ldjZHl&S<)D{g zN0irYR z)XhW94TEt(ATtzHCe;>w_Q4>;v)Q*}ESZ4$e>}Z=IMe_CKR#zEr%XA_Ss0PS=B!9_ znB(S@IpoZyKuh;LX_wV!lW0&f3UAa8B z$K!s#9pO)ds^#}@i)F{)YUe)hakGgIK>>&TAS@^)ubW!6U<6X(QIsv;G9le<4ADum z_ZTlE{FyJXF-r#gp{|w4HT{f9aV8W0v<0_x~;&{O)H)2=XP@=%bN(^#3v1$AoDw=LVz;*l!+f~ zT%X;7BjI0SCAFVo=>Rh7vce2eEN!n*+}$6)P32zVukb#r`>TLaAkmpbeS~5Z9~K3D zuYk**3$O4MzWX7{D#TS?gLhEu-QkUCL)mrI#3j+atyP}^U2bT zsgh=qEl^X?mcz@7f;aMi0{T11cqQM3tn`=o{v_|{0Sccr|wm2Gf{^I1)} zgHDP}?Y!gtHq1NThRf_VT4j7BldpfSNV zuz%NKMIW@>x!tHH{Upc+?~x2$ZWC@jZ3>V}irAg@uM+=GI!_8TzYiJp$SLZRo;a@u zqN)E_-!q3inW$Vqy$laETvEs<5&<4F7=$8Oh9E8dKs3l$ZaAQ~;sSyg3qhfUD0;j; z2Wu2vlv5Ui0jb$#`(Wy=<8u7VPW`?nBj*ZT!CNw?}_-JS- zj8_e@D)EC+K|=S{EaVLeg_J<5oF?4-qS%n$^Gud>ix#d^RNr6g&F%ooc<%Bf@xz^I zE_sWuB@PXIG8tWp7lO|U0vZIsm-qi(3|LOEH8c9pxT9+Ah(E?MvC$Wj>oDf6@cmZi z*^FeIy7@%4>0FmVj)U?MJpVPeddu@nb7mN?O{#+6n-P?PjRiP|`bnS9l;kfIuzm5I zG))37x_#QF?voBIp(3GzQrUL(&9m|kpRLcA3d$3WJqDvZ&61DoDBsIT=HIKVq1Rf* zmy5AUm4LHT4I}SI;oH_%)?9uH=1>^dc}4g?QtU59l`eKvTud`Pf83+bgOmEs8*K^g zO9#Niz?ci*CiVbTpd}Ec&%c@=y{&arnJmOMur=3YvH=$^cONKZT--Ea9E^>@g+o@l ze9ZVH<};p~E4}3JXU?9V%XkrxDSoib^$dlr6(GABezM$q@#b{+HZ)(NE%4P~~5eH%lv zTHg3s?w2wxbzpaWrcoP&7l+HTBs0Kw6de6IAJxj&+!)J)^2odHniK2GyhI1NUZeJkX&g) z_81FK7bM+sZ@2ZiR<($|!C7Ohps=Ul?U3smddA3~RBvCc;L8OCzd4gKHkH)N#gnmG zGyyP|LPCRnoPM9&h^3Bo4D@GgIOUxb_T2D63;q>6dByz_V?%I77abBo?bz6|jtg+zg3jHLcZ~P@3EOW(ssWCWzke(tXag zZ8c}+#p3L>cQ<1Xx0qLERx$s;`kLEsn&mibFwCsFRZ`+Rsa@piztj4Szw}B!XcA%+ zw%kuwbv-8Bu9EI-Z~F2rUn=%euJMYq1#M#bIjduCqZmCslG08bR?7<6K{33eK#`x& znY~P`1oWtdsfMr#0y?s738d?2U+Hpi*qf9d!EFaO=gwY)mge6no$?S9&YoAzg-rVv zfB5+&)=ILKF!@J&zq$-vc0aO4*M%D^q9M2`z8XqlW5QQsgJmA^MBC+;l7BaVqa8nfg)_Vbx0c6j%3hbCIJ|pB~tuSs9 zWudD=R<8)RJ|nxY^O3jK=Hhs&&B^2zRVNel_1WuD`0wsXWf;g9 z>z=pbW20aemjqC?^aBkPT3Ce;M+BV?UEA?PqiY{MDWOPsET7O zU|VK6&vXOv(<}4OxyrZi$)2AbLQ~I?0oU?Izh9~oxdEk_{(bUL8B0@^WVwb{M zqbR_jP)g$BRD9rQV^wZFQJmo(G|>bQo@9kIGbB)0n~KoTwf)pqj-MF3Zj`=Xnz8H1 zR_#9!z6JEn8}#|d^Y33WbZ7wFb2aWPfIH#;o#usOruv{lAV-xMohw1zGSOPEeK!QL zH-B^Af2*PS*E8ay_x!xDsSLJi@te~n9uM;~(aU;6mhywJzEzKh!bv3@Ve+^ad;XT+ zAo2Ktj5S?{`n2@*VG@nh68{8tZhtV!AZfyMzQ1&w;v*n4)DC)jmP#CnOT$#l~}y5(xm z0-PtIg$oo|72U%e&FBe+Smv+#(CD@T=UL_wJd*CS4JBSuH z?2meptXPN@2(JlP_Jp032)&4a_qqg^RjF1evex|!u_%$=H)7SEPH0OdHW6(fxy)NL z;Pmyd3qRnlCA`FHO*whX$kKER!CoO5#xQVLajXfsvzviJhLGfV(BH{B#^MM1Y;$7zoyKu$W+G&Yzl{A zjb;L4cGTV1ek2EK>$?31(l$TsHSQ;aXI2^2T(uK^rf;3B9Bm(`3Eup8AiTbX(h4rp z@JM)=gZ=vao!Cp1Q`)d_>6%_8pI>WtWd>sH3%TN3tu}csLTf(9hi}7$FP{bHq-x~d z5tPRBiaV;W-jT`8s}aI6KKUPLbS1erDz5Ea1$@I}NAuO6mzN9HrH^xLI!vWZXzBg1 zSIQ-{A~b+F=bF1gIi>K3y_-xKv9yrbF%%$=5z-KVdn%t>L=FhN_w52H zv2L?7w63Dj<2+nk*i+vq<)tx?i~`wU^xOJc@Az&sp&m)yJil{DwYX#+bJe*ZOF2@8 zYV{m7DP*cD$;K~JRDFSb#k;{=T^aM@HOoWW3MEgIezS(Yk}819`sM@|KIU_#rt8uL zr9Iv*&`i=`RIgyi6f`afn??&YZDR^RsH*fwVfl+M_Gy*H;4hQ+QA2}`YFcdI)~e55 z&yK%fFZVWuEhpPUsD*$}T7KpI*Pyu)6+W{}{2QRAEQwxX793?qOo-ibw7i&=D1@!R z3%R2e6xa!P2G}Q|t8#TBtPd+xO(Z>gwp?)PA4vz-z5G_saIwQ@H;-f69w>-qJESDP zc04F5@ztdt?wDw!)A~$nRbO*;$&p2(==h637!e@!2T8Aply4b)qx1{X^t#OtjQmy) zkRYk*Q=lB0;5%^eo#NEf`G@(3*0!fl=x{n|FH5ovPnEK>_G9WaJ$%+YqPGcKX$_6 zhDTD|_0aXgJIKjh^4Tl8bX8VyBUBeZA6i7LqGMmGj zPpe1*whDmj|I1SR^;iHdHw4H~W0)8IIB-^`U32&M7u!R){e}@|9(j6tt?9$$xH`ef z7z&kmqT(&tTyq!+hS*58su^wdFIq9|bRH+09g)0!r?z zF{w%x!l2+j>ieFq1Rle-DxaZIuOR26%JLE|M@7iWG^W_~Mio`Tkn* zPp8bildim~7Q}+!03go-62xISg)EdCozS+jON(F-&G=v4fO|QmMStf0K)?TzndP5WjfPrr!G4^xr~}fe>oI) zb`|}L&Op@q zS0`8Xs|a?cmcH{*a#VR8g$BV8^Q6mkA6?^sj+8d)@cSI4ntX?wz+Bm^Bo%GJYpLvW zMdcmtwdUs=ddU?s6PN$F_aT;2dY>!1q*>wC!q=~IyZD3OIg-G23#=Ie)p@Ok}iQkfx`P-I%Dn(l|qltJt)pSCqQ$RWo&?-^E^X+c(F3bW%iaB zkX#?Fh`isS<#Kgp@E@S^@JO=o{-|ekZnG())Y^WQZ0_-7f;nXUMLVX^xI9THVF^u8 z;|Egt55fGO>8TGqK&1s8<9L_O^;{Dsin#a{xHexazg`7Z=a7tM3SfPVUrkgt3cQ@@ zxZE?(hH2pz?3EBFSgTUa?dpvABiW?{1j;)LaMm0x3R-6e-^$G|7e~8A*_N-9M{pOj z3E!TVo11eZC%<+Jl)~IL@K2eQY@{oyNHOmLH=`(&#)hetMnqC^==CQ#>N-0*ak|Jq zA@;_y`~#a&X~sNlHTA;5OgA)KGblPLMQ_}e@oEMT2Tc}mI507uY~qDsKuiMoOuk=n zMl(oJ$*NiwkVazw`3(zTJ^-OCfrj(fQv6at<(Q4-dj<-4Ff<&=I5_(-%GBMjz@O=p z|CO@(s+R9R7`6Iy#9rd|N|J7HT6JE`?zcwR-)XDXy490Hd$dRRH`5}kT^9UywS$NU z!{p|>?TX``vgUvtTM;4y05kikNoJ1Jo;~Y-GiS3)K_cY}a%}R0hJ@Oya{g!6JzTaM zi=FVsKoUb!R|Om9^;HDR*wgM%5o$4r)hBaWF;WW8pp&89J_P%-*OFQDQ1SRwAtHkr0^BE3 z;R&?JekV)8FN*Dq?-)RPRN zzwbq$3WpKT3GRk+Bc17@M23LDZa41gn~&0n5C^3OVyfHsnq<1m{u$D*&Mqfza5(yh zF+BHHzWkHE$7f>KjMp%2G);$F8A|<9VIN!TOay)_oLkYqGbrajmJQvp?rO5s;#VSh zwMq^9G?4fPDjj&&xI$bhms7xnzYC|}8xMJk=879bf1O>5)SH>JK!0#??I`MMyU4~r z4D-GoKh+Ra|LjR(%(gX%A}#m5l#g^3+ziX53kBLBdIM3UdazFYHI&z3wSxcM6?`AX zj92C|eOfxm>09_rKR+ksxurtD+!>F~y9obm)A{^G9#GXlA%~YA3z8+0|3bLIrtL!# zV5C-?6{Z)b^Ze0-tAMi$5%7rT0V~Io1M8-DHV^v|?uq;^%YGvc6sAkc@`2c9u!DPS zva4~mxdT(+YSbtbc6jO;^5q?C#zn%MbINs^!*i{ZYUf7#nzrZJ|BS2I?*6kvzd~V@ zJ$d}S@Dz7TS;mrgTGaJ7Gw6(7aAt`rJJAGHQ&^#wwQ?t2l$-w`yk$f4Ohj=yf!@m2tb-MMqR|EQu%eUv zSai|WmVQM9xe+MTL1asU1+0E*Iq_U>c-c{Sk(F8*57LjA&l(xm@g#Y%3fdQ!jn5hX}K2@pXk0!;GEi*$f5pVR)(H4 z45CL!e-8$Qfj#+;Kp+&j{()|CNu?0UKX65%hdR64_r(M% zzWFLu2xtuUHN3Yk@>EJBKKbE}!|MP+fLDis_)PqY<4!GB`4{Z3U}e8w5TjEkqKJw? zWPXFET34Le~K-LD4^vg!OQ}fwJM!Wg4dq3bx7M! zxR(z}vy5Y1jddk|(shoz*;U&2$1~|)wqM1Ff1o#KoZdUTWNrvsPb;6uvz&lVG>Kn))YLj?M1th5<0GZxD#Bq*k-b@oVJvDcQfdy7DtYB8OptB zRbYAeX~}JzKUo!_T^OfkqD|6Vn0l&9=jFTKk57?*NXq_dDSxi-6EAYfyiIDys@rb; z0R;UA8@skr8|o)61t0iP{o z-wql9%ldjsDwwM6(O$j`fo0YI+ahwT>SO<8s1DG?lj|*xiA+p27LMoDkMZ2P>@`<9 zwKb;l$}nd9KQf=n#cO@?f4-2}7UpoRJbJj^#y-_FMuJJC^2zniRI}jE1h2`cuiIv% zTR(~4IaK8!V|d~NMF}Zb*PRoEP0nF)Y!F}__QC6eJ%A7O$!R)OUf)W}d}1s(oaUPE zB;2MfG`ebg&q>jFi4LYkDKXE~G$?P&zYX4~R_-<7iN8LyiD(iZ-?t;C*;E}*-I{k` zgd(8p>5bb=71$DrrrjA#z1w{=3TxG56+_bFD2^Drq2#rg>nt5^+_)n9OU_-x^_54f z6Ifc4%{)-^e7}*WiP!|V0fD?A;1>!U>T+ZT6Ku4N6UEkUPTz)Pksen4M=5421-Efq z{GUz*Lu4CoEH~!Y?~9RGgh=X`h#LHZjz^g*2jzxDqw%Pz^3*Rt)6v_IhaZzUnH1luLE@*>9KRW> z2mMjFRu`!M?-upFsN(teFa|I5sM?lEKls_`RV|u2UfZ5?K2Z@yU(e}a(@XHSzgyp$ zV`ZP~H22B==QDa+*{y!~ZJTzD^RflRK>1YA_u40Y_TMI5TC_^{x*7doUEHkcZKhmJ zzN~mCyUr+s9*cMxn^a+G^#n7_np=1D1c^fGYu&>c?Zndrd&p&@ENno?iRA<;5QzL> z!{|$pn>^s`|AFS%zzw+^)-POWgo4G3S2Cjo6_supYj`4&U*Q%Ppi6Y2Urba(QxZ4u z&dljYCYwwldYE7bruu>hb6*TOhpmk-PCcwh7qeV$w&VsJFx-n)L#as;`C8MVw{pWj zqdEHJ_ASSO7qj~=|)#Hv*VkshpuLrFqvtfKbTQxhvK}U> z487O;SX7Kf?o{Y-e9UNj434IcYVMdOhgeH0JbZJ zyE9{r{mSCe!4Rw_r7ddEs(5DHx7%)hrIsRCd^eMxnkOCs)3D&;J_A`sAX@f zJ8ee(Q2j>2^k>2yf!}KJ>(@Tr^aZ1t@`=rB1ao?z+X@6KytW8D2p|iqW>R8UV zrpX(gpHbrlS+KeN|2ynX4!r*zdI$hvFk!vU0UZlb{o{}0U9aF@hqnco*+mECw^E^R zvKsGpsktC)R;eyVV!6IkFZJ3@{WMwPt-Z>G^Q&ipEWjTG2+V z-{ad-DX@c13w@+veSCN=tDyC$4{lS)9Qm;=|$e)=S?&jzM5vkC~U zT^BL4&vYo90w29C{{x{g81sRtLgr~C(el{$^5~-CeJ8O7VRzXVZi%ZJQeG|ED3;wUQZ6g4dgyzYikYOFITjroK zpb!E`OF}asM7DrrLxvu}ED3(r%A&*yWy8oZ7^#?W$$=qU#BQ(ZUHe+T;fI$0m@5B+ zta|d5VaFlN74|+{tVcfMhDDl9(*?CCg>cM>1$h3jkaWyf2*3q^;S&TPr+%P3Vw`YY<^?^)H& z_!m$UAbEyD3p#s(mHmAI^QP{Y8=t^>R`SG&@F}#AW%sjr0CB}4VE~ z{Ng(&7g8m~E3DU1f(`S~M(jdU{*!o{h|(KR=CFFB@lM5rUHk4Eb3C2egm~vSXPZ~QshNf1>VrE`@ zf8tm(lkiVfG2AxqYbAr-8<$)km+7c*m%zA)iFFJ!%jxliKrD+6j`0wf<&gS7Lbj<9 zFo-D1V1YEHUL^q3<>F+jRs?q&8u?KQnSt~YI*2TDAK>r6KB=hD@xN>mP`LmcJz2?6 z&J%bu>?BTt_*aU}0Li_tppAGfH0qhPT!qY?9$or4oZ#yD4|~h!_wcYtYOjuk->Y9l z`=mn&zKV*6$T(xT>QRN~1@OK(c>j7mgY$%W?!?d1{AgQv&~>Tyzdk40KAL3zv9#_hqpEyRee>Px#N5oY_x(t6{R%`7G2>d8k%?2$c`9 zM+-?2HYhImi{EO5GPG6)R-QB4z8desB#Pp;uL^|Ls&}-jWR->nX7)NEdXZBtG>0n?RdC@jXqC&t!T)j6zssp| zO!^I%7H`^Kis9hWg4<#&WR-qY@y_+-HN;qmi$~{_EyKj`*$K@I?dFI!U5ztKiE^_` zrfdEOdce0Q-z~6Z*$r;(1*w}v$S>I&W!|i2{`Fqi`wG3CUo2$jI*&AY;T7H?l)ML5 z4R_!Rmuk0reTkEVAA55}KIK)dVok5s>hYw&=%14P8mQ}cmFW3#v!dsuf|09jY^2iB zRU11;FQ$WHn>7c|Hk`xkX_XbAwn687PIm4A=D2!llD8`4dgSW#un+tX`De5MZQ&tO zuDEkt`2<}6M-&rmKvG?T%PF|4a)%&FFXSY&iKfXw$gluZvNYhD*Z<298sp4IqpX># zpN921O-yU+7JN#@oxbpAACCLAP7mlxCH*olhP2wPhoN0JCEohQRP+_K+?pDD{tK=c zl~`nokbr@#{mAe1pdhr6NG7Xerg2*(dOK@BoljNya7+Jj9(1h6(!v|F@Gc^X$(`3I z96y7gx!b$2XGkL|t|lK&tQX9+uD-qgR84sqm}f%3>(%PC=4fdlo$K zEA(RcV8!Qu+JrXX388I~(fEPts#XoB^Qo0AzP8hrqU%O%;>`v(;P%c3!52E}{m3Qo z>;wP~8D*reBtzxib+LnDL0U9$VqC9qtWVZ`K$=xF!$K~R1mYVnGXBA)LkfsC9pfdp zlBtnBm5&=4(*z)2{Ho>%4DWZ9&%9Z%4YP6lS=&{FhC?e17DTWMrx&vj2Nz{se_4oE z6bYW`3RZEFOe^qtzdJg<_P~tyNut3~`OQ1VR(tjQ0kN8cyK^I3g}2)+ZO2(Rlh=Ox zmrPx}H;Hp(`^=Z-{QFi{;OjG*@P0Ie`YNdLqOo#Xur|F_1XqWyPD4`*zxXTU%!A`( zJ`dGF|e z2K&^fBv%+c8Yv5={{QZi#bjW$A&lkA)iD&~Rpmgt2yi$sndtygSYRMrYwRRP@UL$7 zT@W{zr_)@|~!GDHDrX)UIj$s0!xB`@?Lp|40y$Gwk$S(OYrK&h@%O7n>% zJ`t>-{{yCcY*++sWZsjJihbz)E{Cv7{d4u7%X`MUUWYa2hRDqJ7QYu~DPWZI-wEqv z$^rrw|EC)$0S07tEB=5t!4@{~djsyAQ9of?_2)lfOZJQ2 zGfWp>s9WXCueqC<;+@ezl0K(!9RQv;<@s#^)dW$12`hO*{A@B2P12VM(Fc|nRVV^b zjW7g4^#x#aYL=UfY3=B7=if~XxbFM?X{6`kAAE!J>q&iIbK@>5Hn@-HvscZz_l zUKvg$@GNlCUwmf~Me@2DdtXUl&HrcAL-aoNsFD=;ALuINP}&!^Ra@=;>i~LtuU4kL z#?ccNoenyJKMA5X2HkuXGvhV+kNolX9Kn|3`sK{t5KIhmiqYunYB;7ysVeBoFtZa| z`HqbNr7u>Hby?5C*~E^OU;mgx-}AJ>v-oyy4dRA~vtZ#GH);W^Um=IUTIL;WFZBPmFzBGN%bV<1|trb3ivnR=XRGQY3EO@5XPNFC!AT8`*5mUI1lA!J}zu*32XeW zi82@VjvjPALh5}Qy*N|jOmZ=GnWSGz$8|f6)poWq2l$s8F|+db)cKLmbUIj2T$7gB zX$KDR$*)_7bJ(ZfZRmd|13t3Irve;mctc}1xjGK*kSRibHc&h6JkDSnmrMAA1@ftu z0G$}HcgD)mWc>Q%7%aevS-{C82LdiTlyQ`sAf=@FBq$enRA`?p07oHbm;k~^qyP&r zl>U8!wl-oJ{LoU#pBZs8_CGN9=uPemPZS!}-NrO^C+bI3Xl8d~2fHhy+2Kb5PqB_% zrE5z)39xE=d+*OS8loI`MFe^g4BIuzDxSTu3>kD8v~!!nmMSeG1Rl{SHmpqU4I(PW?p< zPD~siES6L7r0f*1Uz|t|96Rqf5RY7Yrtv(&HdGI&00-Or=B+)1^ql}fMbIQyR9(<~v_Yj-_pfZyjL{@{Y1 zd4Q8$*)ExIU;e5l^?mC=cTZVbX6ZO_qEPX_%^2JNoWAFth>8Kk&jJoq;a%ZUFk4)>@`? zNa(R$<}mxBORiVUce#nAzf{TD*fqUS7@| zHvhHmBU@2hK=W^Dn9!F?@82%;MGbt%J?+fXZwX4xb?Pz%mgxXRr8-&NIE^*t0{V) z>9A#f-FKXFKuuB+d&&L7?#v{9NZ{T!!w1l;L@=@)T^GG|%PsZ?(_XPB$nLjyS*{;#PF~b#HJl+VD z%==2>^;t8j0T^0QqBtv-Q3X)FgHEK%!2P2;4)m81f=GsD`s~{D{3as0^>l8C0h1sk zV=4m{1C9N$s!}HhWT0f-zxE4J_wzqcvc!jZbZD7WgQmBpR9DzXm5kR`q$@$FjK?aM zT>fdQ{!k>vEXrwm;Vp{GNp{q3!O7gk@!FS<0n=YNuPYZNd$P<4tg#h+zTuwOJO1W% zJ{vMU>kTd9Vm~kAeAZQ zlx&|ueV~MBXSOwT9PoI^?l>QP{Fat!awO@VW@+$srJDIXC^<1tPF!U{A)>Gg53YUbA?!K2BIS;jnzhqqHx| z*4s*}vbB$4DtV&Nrrv@=HxW)#(` zxHx{1E*g)&Zy4*ZFODAwm-gwRW~J!)YK+p;J5I&ULhT&=J!O+{^LzvYvx{)c+z_2xHR1`6?(+1NG}7su8H?v&`ysP#)slC z2-KiV2EWGuUnu*(qB!s7#LOAea9V?;N|U`x6)EAhRq&_&%bAv)?c+A(bMls-fK#J^ z2fcrSkNqDY^!KZ&ISb;09DlIC^^oSST{Y|00cK|RhGT}!S`lr2libvnI4G@{$sa!V z=`4fpv0aV?XTJP{DeA*uflXkUgn#HwD`-eo;I)Ypp_5W;5g}8j;6iY zUfBxyQiNu|l>z>#U=0+9cx1g>1zNQ=Z-Vw?XgYq(1SkVgj)-ONsTA0!mt?BSBVbg` zO5CP~nzThVCdygUMn-!wQfi^U+8>lPr7*MaJh6o-iFs(LJs$~dGjI>V@0_(i>Fuoi zE-TQ1RfqTaaiN45xysRb-q)d3*UkjE8Wb|0?8E@g0@4vm&hH1$^_6UzX9Kw6)HTk4 z+zDpy?kZ;AYQt33zWQnfqh%kTT9xbH;wPrWT$>}%1#erWbU8{$KYm+SMpfs2jqq5( zTQ+@)g|e~M)k~yn7T-7W8~NMust>?LAt!}dek&*BAkl3Me~|%52Ox5sL=6A$y$A9k zMFt&}*<5*Yula?bE7c);YW@C~;sRk0#4(-WD;{x0`FMKsv}ZFv`)rL%96_0lwLn_x zvy8!E@dncy&Xko$VjDKRCTG^;Rg-vr+Wd1T;&}3zQ=<$z@T`8&C+uEa+hz=|vp!MS zi%A5K_`NjJAJk*>VkGGxWD=ok(P|w zJT$A?43+k>rWp6S9R$l42O|~ET8@p{?OJZth%x`p*>W`3UT(Jih3`GZR;|A+;D&!L z8H{5hO!2k*SOraKKWNcl5PTUP<6qy@;q>S=qe24lwx-m6l~By@7PSXHGHUbFtO^{J zN`8B2-zaTsTwdPt{55uWn$tU7hsScQ9V(9zC-egXj0P3QApy9qq>C8{y?P2G=52wN z^tYip$0$1fwi}1uW^pP9*NO=+G?g+vB|w(5mghL_ugHJ2MY0Wx@1PFoPc`AqXRd9U z<)6XMZh9b}YTNZ^22#huL*}acwzma^1NC}u_OC9aG%l7`i3Q+FD3 zxgI!7yVZ}r>2K8bgiH4H@Vo9UD2aQe`byyy_vm44;u=Bowr?DwT{yJh3ukBnzxj22 zTPtFJRvBNfKaf1n%e^S;96==qZwrL!`M;~;G77QOr8cm^#UVDpG~k*Uuja8UO6<}> z*COvJ$e7H>PvwIoNyg2f6+AY!wFG3-T8OVA5YmM~LFR$-H*JJ-@oxH{UCWZ&2A06N z%ftqIVDs)Shh_NcQKh*j`*QYcN%CqwHhYk!L#Qz@n(s13ZXQ;BxU78YXM&rFPIA)@ zc5cx-Y|{$9;VBoQcxpB;b<50d*VhqHJmlIQEj9c4ST~{OPEaHuDD<>ZG*ItR3p)li z8yGz4+SZfmCeh8>>Nxc+RH8&x{x^@A=LooPVA(#V zI1zAYEi^=i+c^8O2~<9Ozh#4C?(v1Vtwc!M!@Y-|eR7V@ISa}CPE#d*#VQP)1la;V zK5&tcDSsi<|IKTL^sMX!WhkZmeCXleMcj|dDYF)#29}r+j)KM_fxJNbN5!P!R!{J@ zvjYloYS&;?SEFKlP-x21(RB#IY z?51{<&9C~@cE%4&3k?(3F|YY9U+1-(We)mf_rN)v{Nzr>eTRo?PwwMW(F~SU0fs(# z8P_AY1Y?H7X`aL$K!k@-nI<7_LTnXb9v2`s2WcoKE4H<2k0HZlE7c{g*Q6B3+)64E zx0@SnN7l7x`Mr2+Ja>b%`erz*pdwnT0rr3cqp-)d+*HSu4ow@`ZyC?EYXsC$;=hxd z7*w@~G?(a08>FIN1J9~4$xF(w!;T8Cm8O1yMjfu_NObo?b;S89ic>>N;BRNMn;y`| zOboAo46(a(f%jDm!tqML2F5tj zB+)cc#Q*L^qCeuric^-sPlWhVO0}u{EkO}^-rV&U$Es)tW1M7{aM&}Se-FLv?#px9 zoQ^De8I1f7Bwy#}W@OxKG{C-I3hOK6cy(pbEg0x_uwqEw2i>RuqxPeDXMsGc6@S)h zM@GJXDQ^pvd{!mI+T^`)mx-m83_veWEnOXM4XwP*!#P+*Mdo!?YDuyd2cfSu2j@83 z1{*=#CPv5*8;Y3@#{S@Fv>X zO|8j9@#t+*7p|bqvW+XM#e)@kb0H!*SC7bB2!RR)KdJmC0J>X_r>BM$jxw$g#t3wu ztXRITOJE}Ux^8I=1Sk3U#rqn-CTU(Z@p zOiue#Ir&IycOu4LRrqQ04atvz3(xASXV&hd4Bgq&!;WuF9TP92t`4vw^8umH@m*G2 zB__}1*LRHsv(2|}8QFE?=h~`%-L6~o{%7)vm8X+x*>h5PvmZ+)TS6}GktQ&e%9w*=pX>!@BsfhrCKK~NtiAyIzT zRh|vAK{WOAy4^IHpe0yI_MjyFWF%g=B_QL}{D>A@H3ON9!EC&1lZri`EWg_#`JA-0 z>gFc&CiAh@<>q00LB~)#>DF0os5W|<=fMd;aiW1WKS3-=atwD)7EvH}0O)?UINcWe zog$M3a8gbmvA>XGDcST6xTmlu(;OMxjeYfES3yA}zI-7$Q`$O*OW~@XRya_MOjVN& zk|jBNEmNI)*6Mq&b_%(nCKO*I==HqPiRKkGHFw^mx#4^LUJV&R*>N{CU8wTcbl%I5 zzSN9Hj}|~{6~r7|ZAeYc^%RaZTwt^PDa|V3KUE_eMh*~!ESP>sJ9?IWI(^>=r#P{D z7{b!eOG6eljZ<)HGv=4m}1m= zmdSiS#j!af<9|zXx zgWFP_>*_umc|=WLs#FhnbhzApx@x`oL0mFRaQe_VawIC_#ud^Op_-*F&!|PJAJ$e# zKs!A0IDuXMVlAx6hv9##h5(;BpdJHJAb^XE0Wipv5YR#(=k@fb0)Izals+4dLHs^o z-(!7JJ*0XwHRReDv(D2Z>XM1BDDKmFgIH6x71u?VoOb&%8_(;uErf^HmxTqdbjXPf zQGGSbo14UDRR%ggr3yvm7G9n_QX!GnlScEtog=-5m1a~hO_h5p0VRAfk#I1_IlYBJ zO6#q6aqnlABwm0n}`z>`1)ym@cBMW7^qQe@OyODp=bh^V)L zTng&O7bCUECAS98M(z);7xx)$NA_$(xgt`KRgVaAr+1Dhmx;CgbOK}4cq?zJ&Ve~D zb)_`2bhlS0&TcwTmY<>O?Tbv8@mhZ3$6zn~4D&dIDb)!3$r4*u1|OLue3MFCcRFQQ zae*7R6S{y3@N~}iPY<{w`3&4JhZI!}ns?B5W0lkl?-px~+3m+Jg&j}v?>$v~KqQdug;uzxcu zg%Yg^nF{TiYkx!MS8RLpVOd{^H+&#h^?1BE)l@*TJ;wN2;ugm2o&)3s zpHj^5SD|sIz@z=CI3E;=PzD41oXGiT_q2LVs!W8>@b&7e4D0W(p-)Km?MiT!R*K$36!7 zr;`JW-fr!d@rmF3g1BDeZ{F$W{ndGYt{rx(TWDVa9hmuq55lP^UR*B#oG>S>)7S~g z1pIpcA_GDvP66NoLw^Gz0)WMk;jbni3Y@O+FD(ly8n<^T+vbo7pC>r8nojpU;WSa=%>N z)0Bogq3{XiHzK8H@e20xL>5SUYlZBW>2%q2SzBhx+t1H&OINg(Gc4U?%+vWpFZTwzY`_MgsfMg`*0Xdf!>4pU0 zSwOw|-(?yC{qmB%OEyngFWF5~$!+Y!aVX!~K8a4WTF|km+3%sF&%d8GTPEdkRhcwVJ5V+8CIPHs{bHe>wpRTRu)zC#?xI=HvDRE95!XPM1-^k)piiU+R zO{mdET5K{`vjzi2&0$2btq?VtdI1eHoM+P`Xy}uAp*2k6%1qz$iv$MydB1lzd4n7A z&6kwlGJMYspVC~Zsr9wqu(T?vDqPATArE*AMgRk43nQVxasoSm(9r0vWY;{fzYaO0 zw`ax?X1Ta9oC)`yLV>h{z+}Wl)A|zcew9Fs#}`Oc;72l;+x6|GcA*vOW<|sh^lb}&l?Wi$_iHC?B0lVJ9>Y> zARmIg1vcK^9DdY~0yaZ!N7?PaCt(ck62K}|aU>U5u{_PsIp2&gWtDc$B#hEAU$2!a zODWHc9*ElAYK+^iRJ^jr{bs(fi)JCyxo#TO44IsEtqGYzKRd^% zqK(!)E%)dGbG?oBM>i`AS-AA-PCk=zEFtF|t@U61w>Jh5_lNcfzCY_?r3w-UND*nk z%+65i1|7$~VNDAJfGv_rfS|r5-`=sSmR}(!%P4L;e?vSjUa;RnftT@wtug}5d7KAjpCvI}I2WqZM6i`7 z!P`I476}%z9f6SwJ+HJR4%PX?4wD_pkM=!5VIwQ(t~n;LhhVv1L$Jiudk{8;n=B!95Q5rb z=YFnBuaew9+@MSM`Zr}-nk2X|E)i49{jQJpp4Dz7{}r>Jd?DJ&pt3iCP?a_lmdeFz ztKFYk@v;VJ*uC}0AAyctc%Vkm9>c|m-g71K7?aC0c+8-IrEJY?2?%*^4q(A>G^;|d z#2>deKN9u_V4Wuk20)e13*e4sNFkG_xHozBPJ9$}a_|I!Agu zeN-g&tkqi?dWuNsxUR9-~Q)AH?!$Q#z9bGU%kESzY&CW*9 z7&8F-kS?CdxME3jde%=dZ2h*jd>h)q*C6NlW=gDc_2tQ}J5eGlt@YY^4ndO}15HyS zfM$Br36kwK);KslExVYA!NZ8+a9;tw@Q|2%O|{ouRvjO z+eJ$$3D@e+H$_lo3O)S93dFu)7;W);Z%Wg&0E}EKfaUN9UKph@V6OAWMomnKMUM>J za1UTO4OTx60D}L!?MLs&=JUdR+cP-cW|9zw;r;5O>jxKM2WX8!o5SAW9ghK)EEC8PQc7JjcH~%3&VJJ7%8uF|OGZ=xcKBhfm#g&~VAPG?B7d zCb_ehLRWosHzVs=vsa-Em&B9}4964;xDh}d42@6WtV_!DiVBKQeXu4u>giQnV;(}_ z0nKLL?`bs{^t~9+3?{t0$C=l9-J93V0L^Qx3PuAHWu#Im7n|i00lL5roM|QtN}hd# z;0^f0xal*2A?Q5d0_qZw#TxmPi~1gSekx}Yy9oMrjiELh`dA_C&27L$QfG3$xIs~} zAY7Y5DptD?v5xE*|6)XM&(pE8Or;lnnDHI%*EKUz_D9KCGvps7nmhWwe&_n}gxOn9 z^O9rx5F)8ql zhFHfvDcmU(g>Vx@gq?z@S?pS0bLe`pX4U%xGgck^z=$p%_cx=%`YuQS=DW`xrR;g} zFNqoJi4XD?%QL^2ABgdBH;}`v#M*QCJ!CmJQ|eLV_4QAchLN=i;<0Pa%9`2tyy))w z*SGT2Yhoa4KD67Fj-F%sP{6R*?VXAnlppAufi9Cn!oD92@Qg^vqosfn;6(>~C)y9T z1I0SHEZe=}aQ64lv_BR8@T>FW!pv0QNX?WFk4G+ql3{jl^TVBXWjDBsuTW_>1VW}Z z-NPCbw}tKC{I$PC8>)mHPGL43Wk*uP2$&bbI8z*X@*Un2z;8S)*9s7Sq>ajF);(Et za#ar+q0`$2RR`Zb?1h}mZ(4K2Ea)hkk1Vv+xV0#xS0nT{Ju3;ZGzB%)|jB% zuvG>7Xf^ajf)#-dW=+&WLL-Ga^-DN}Z&k0>3qQ@~%U7g+!#vr@=%`neJ8+9_ev#O+ z73X-G;U|TPc1c}|-20aF9w6*H>zzT!1ECNN{L^4Oa5#C7#B1#=jj zn5>0OnX=I{j8mX#tS@ImDtNF=+fRbYuRbXsQ>dzP*CNgzasme(J5*DqCEW=RRy!7u7VqT7 z^XfKMx&QjpNMoQMsN}khkkrqB@u-!h(~}PCn1aTS;KZFIm2S)}(%m#XU(8qP!{)J3 zf3p|Auq-$C?Azuk3dLzb+NSQFD#)-wg{B{PX4@tJnK+`(PfRk5lV+%YE!7B>nBlDG zO^KS8kU<{9YfT}QpHLVUtKIFc9n66KLHh}J01ayDV# zl}Y3eosAs)C{0s|;7m@uz}m?dd3p4s@jyt+pgZuyZ-i!*o=J9f@REIC^D@34Y;W;L zHV0@K0fQ$XssiRbM?-Z$z%d_QyahBoK-2U8$$%k1l2jlhyYT?7JyK~7Pa$4EtgN;k z48-Bbtj5mkiD_j7RS4eC74>9!3Fkiszs9Zi(P0Ci-!{?%MRNcP%SqbGgrq;4_xI(uQ)D})24y-b%%0xFD$D?HGaeYF|5=r)Th1Y)j zUCy;li?t_yDaAgD^=hrVo1esN7ok`>!w@lR^|{_#AwgECnUcjsYh7Tke7>SkIXGJ^bUc7MnvIDDFx8W_C4fn6CEmYeL zmDcNr9b_ndB;mJn>8&Y{p{fQvlYBu9%KekdT9aBbeM}9u3R*O9R7-sD`Gj=3I3UsY zg(*`yywYI$3GeJf%vou0z?s^j53YmSKjE2;!Q(b;pSX8my-oKuihP-+B+mslLYvL% zwK05iH9TYfmE7g_*y&U=8=QjiDt6n0**Ko_KK$fCO?`^w&LD2H>1{T9)hWYB#%J$d zna=8|1$ke;5BdZp?fWl)HxNin016q72Eg(F8}p4tmG{R4}qj zOa+n_W)Zw67-}x1$h8J!C~NIEk$13ArIaA+dN4NkEW(0}p*7e1H{m^I0h^*>7Z!QNkO1dSLeBp&oT^ptQF8?YG zK!Vf5y80~QF8z*3)2wSu&Rc(98!&bz%HQY}4#qR4rrfIqs8IbmqHJWaPj;6&P(B@f z{Bx37r2u22n~x%#gIgtc&1Tdyf6UTz#=qpYl?=7xKB;`EvSb#c;&ML^+MC(%yR^hj z^Q~lY_7Y6tJS!FO73|$f5oX!Liy;XPgogMCZSY`7Kf*G!r`~c7Ql`7>d@9pPXO)aP z5m{GK^ijDnxy#C`hc-TS3=O5VPUz@XeI+|sEaFlux;PA)&8jNL(4yjMIN2OMi1$X} zlaA`MVGi0?L+WuOhRQaJr)Hi+C^trgzKj*%xQ@$beKll!kMBiWik|l7D$H$a_9ZQ_ zMjo8w%ma8lB0v!jzdT6#HGwwwsmTA9-sJ%jm}>s6wyuaPNtI-11L z09O9ZP&-HPt&yojTTO_z+4ea_X7Lq84$E5buT@CfO~a{k=Z^9Dk-{6~EL=$IpKn`h zz#d{WTWHimm-y%!^Xw?M@P=R`+{tUIXbh{$hrAkJ*0TG48&wJg1IM+Pw8i-wXuB}y zdUGU7u|lXYAo>gQ{jHUDEBC%aPVO3I`em)$R&5sB>7#JF2#`sMLakNErGlGMHNgK7h%R?j<`$kCfzAnUC7TeYr@5W`G=m z3=xl!I{y#V#TW7bBQql-y;2drqz5q&Lf1Wn;|Rt; z=fVm=FPh?&N+T+|{_83bAV3Zjz>F(Z?iyFIGNUR0*1gi0=wSuHTmCH0=CBF0mYK0` z!%dqH40iC_2RP%m1aJ&4$(?s-yQl{&SD6d^;}#8NdnE`*4Z~ZSnvY{!YD|__7!Stx zt5G5uH$T^Yclo$!x^8B}QmOn6lkM80p&)9S{ZkEA^CjvyUqR|+Cq2j_@A#iu*eM>u zo>)_2j!ZGZe38!>Xw$9`j{C@#_Hz> zcq;<54{YQ&0EUy;W9{{3U+*#1GvrDZOQ5g?QWl_H1vu|?pGIEEHWX;aC*|GeImWV7 z9%IIgWtmP#R)kS6ihsia!ar%}W4=}fiFYs?yoxi^J6KCDh#mgj=C>21Dum=u$2_tN z<4@Jynk-@crR+9oOBgwxdZ72?mAT689saX8txKGsJ-33o%HDTnW;&@8B64UF|F`P+ zPQlC~3|*=E(iw_kq?Z&qJ+g03)+>LFyoJw|o|QytYxy@_f}BQ% zso^>W)2eFxg!)n62yVH>dKC9FF{TOq`X*x@O?PFqo{xY=ADe8u&cAvu+;KW0Jg?=N z`!C;z)MrUEZ873u0VTK%a&&O5XuI?lELE@Ww2AzFeQazg+I z*r^{%FKXKP=TL8xNRPmfI+?&tAG4)qZ>MYCQ`BkFZI%->vf(5PWs@+Wc~8|c4F=go zC%Ih;#OsRCt{bB!pD(fz?qM@uoOM7=Z@S-%c#2s|!@FW=BW)evjzY1AbM0*bdqupI zB9aKx`5{zKXHL)_aCEHX*aLcYmZeSEpGuHN{Jn3B8~iA^XIZkuGS27bMl2ij;Zetz zZ`|V7NF#J+pLL4Uh>)Y~9R+kO?|5b|INM{X;%4GoRH$W8S+?BtXZytaK?Nk!_f}#z z+cHxX=)+k-jx}f#l3!P{K`FJd5Eh@O-A!3@h;;}XV>@j*L!69aS&Ek=(DBzApr7d+ z{pUi;b~rskEd$hI`onef3CQ>pWB$ndtC^(N=dX6Qq48ce=UBuZkPHfKB<6;ALMPP@=oWH`yBl^++jDx5-0G#Q^OyIN;504ahzM*~!Y7 z6Zo710uW*2ES3al7_e%|v7Z4Txlk<&j*2dM02^Ckke*RNOxktY>en6XUu3z6W=4G= zfs8C55AG#D>E&$i%be{GNTIuqC}z!CfA+Vwrr4+=z~luaXSw@*bDRQ0ql)Gf9( zICq@6)=a}NH)Bm&EqBV+)e5O)O3h*tuTK=F|k3@{ob0W7rvZZ9yp0JW_` zRH8KdzqpG0CP6l8_5@BH=tDQ9*w8(}*IR_RoShnK8z z*|yzX>%;Hp(Zpe#CR|Gl^9Z={-OR+23L~6-be7O&UmDjB!3QuP`Z3 zrkw`!nA9u3R$ZKWM=CmnIwdEbXYN{N?z-+#-K$TY%SaToCe?nQ&Wh-~_(;v7`&P4QB6o@+fmba(h17ZkWYEwy*XW&76jYf)|XrPZh@Hr6{q7qh}fJ`E$HRz5fF> zocEZJOrgpbudl z6ZT_B4zu%H+E>iMm3*OtgUIOg`O4JWIG@Be9L@@sD^t=S)J=;VmlfT1p}w_u$(nQX zXFqGyHKxFbmbvS8X|8AVu>(e-Fnm|DJHarC*N7u5>--m$g>0`lwXS@b9rp>BAc;wA zV4ZIKzb#!}{Bb6e7wGBInvO`ib}Z&NB|U(%OE{-{x|oaNmbVy|9@jjbRJ!YAFX6P9 z8hKKbI;4!Lx7gypUE@1alh#y&kSN&2Kn|Q-x#5I$l-;S*nND&L*pOhXUjdv2%zuww z^0|Z}9Ois67{n|0&8es3%>K_lQU+S7yWfy^A2$lnnzB2SI8+!?1X_#CT?6ssB5a1SKs;E?F;j2 zo(DaK{>{xDJqg$Dt$Q`Ut{<*5IEDE*$H?FCczj8+4bb``#vkEvc=<)kdR4~#Jmg)Klaga?Q-wb^M{nl2iH+G;U zYGjUUzmHH7XM{G|Zk+RAdM-9GA3q|{^5BRFcH5F^ai~It_;%b4w|c@@b4>N9l7J2N zphcKgU9@&cYkhDX-$B-TsYib3NGOm#tYSAPAb@;#F`$hcUQXwE6qGnc4#9>(6L`x8 z_+?E+!p2-G;qk+HrxF0aBfXWTracQ`NCC5fb4OW!# zCV0mTx0U(pSa^EEDFlGjv71l-U9_W>Yi?N=c*CA-!eU-ZJ>d z{rs51_=K5wa{EMxh&waBa&kyKgu_6B&!JU_Dyj6w%D35&YB(9sNfKo<(|3xI{eukP9#MSqBu2&EHgoFxK0 zwQ8fTfki4r=cRT7PcyylGd*<*EyCf&UlC$*6DR@kGK#Z&{GUC{(Hsi-Z~F`Myg=Wp zf7BV-Td-#3FdcyP69J_4XFi$JRJ_F!YsVwGRawe|jye9bw)>`7{@%fz=*M;1ZUh`E zkfSVGaEYMsy|i=`_die^^K6fCbpg5BNukBtC9tWDQ|e_L7vyDhc5fcxv#_^NE~%+2EBebc=gD#j43zAyT{R@) zsCue7bXcUxHb8B^sU*n$6xruRO&T}j?VUm{hTrzPzYkdU>aRQ9QX}g(NAp_? zi(gs(g4M>9mTWR4oC%+!{Eh)z9H0$5nu4*70n`_uyFF%Ujx+hQ;$e*{&&w#{n*_0i zCA>A99_pUAIE_pc`E6aonA~<|jR);rw#~WHc4klN{ZdYB{7N1{c}_!h6{}rnkfr$hlV$-T{u2Jx*zs?Z>-UDb zaQvbF1MQuNQMwaZ1cC+Fn+n{H6*EV?yy}5u4gp9hd-REHD=+&qM}(fE#i2aFpJQ2! zs?5HtOA(_`J!*u6ZMDD4E#|)Z@}8idJnz;wR)Y7qy1u$F(g!#{(xPY4ki~%@I3M1F zU}Sv_EwUONq%CW4rA&>uM$KNd3;bRXR(vHDU+61Ah2g`I5Awg4Zx>IThZgufGyBT@5Irb zDz~NK>(UrQOzH-0;jJ>D;2t*HJ5Y0NC@1)an^z50eL7qxh>(K4emN2E$&pstP|8l6 zeYx@kUHMYNO-G43-2?tP`}oR#n>sZ>14EJufEWO*+JEB?xH zY>!bb@A6P*=?W^DQmJZc13{J3nm^d}4K$n(F-lfj)+gQ9DmkpKOhkoq<8S?%w`ZsqHNFIHagf*A|y>XkXq-0iO2f4D$MbkP`sL_LE_+TtV}PB?4M-KG&+ zA?5p=5VXtvE#*_3skr_S&8EAKmEI15+H#p9)9r*yF%0)x)^UY21E$1AVs_xvlUer= zR~x~u@DMLKf}X_JTG4mTyFH(2G99hW(wr*gZMfxk4o(%SmRisFOR&>pu0cMgZ0FA3 zomo3)DZQgS%19|ztQPgp%1{qZ0kra!fNt|o9ztqZ8S_|K1RhbY^Wu+2ha5*TI7h22 z=23SV|u_g4jdh@k&SJN&VDQ*iG=zsvVx_SPUKA^?K%%YX7+5(Zuho zFJ>D-72pmmLYMb@6QiD)n{sWHtWmxy!+>ohHWfgGS1fDor!yyq7&2474hzf9Ney0M+MDT7j{|oz zn+!UOQvxQ_N5CEQ3LqH4GM|6|l2R5h;OEIEfR=vTxf_aNU&;J!@#4;j(74>Wq~7+= zj^>akpFRQCgKxKCZ?+x1b+x0Vl;6LDh)^pS%4dKSR+fr_ucCbW)sGkbjXNc7gY9pl zzWLw{MIvGEfn{LwurI0chJr2}u4@CW?yw2!qwB0k;)fR(CTz?hN)R{BWv5bQoe%LE z7KjVrN!KSL2>9d}oRD`{od7;19i<5VFdaJpdC31ViMN9Im728$ukGv1!OpP4bu$OG z$A;b(Wn*r9AZTMFLf?hHl=B-e9z1h!tkXR+{jO0qVKY)+|Vj6I<88mu4O%-zl z5%+3~N@ME10{1MB5#%ddnJ)uWEZO7nk~{umo%6eWhzd+V}1|wtl0X;pbB1EX&0Y z(IxyV@FWmb{0iOTU9xkYq1ZzY+Dcyc<*{>j;6v+6FjNJ@juT+8i;_k9*#M>2Fe{R& zUMYu(b;e4G0JOoIINnWPS7{rnexH8rZMH6bJdKB($wm}%ouY`+HQm}&Qz*|_UhxMb zlp-VuYyQnjQDrhUbGRGg*Y34gq=(}DiNE2gT)l%2mL!Y;3=8%A&vt$SpYS2bXXyl z4jFiKo@>gSvFuFQn8+r>(QoBLeA&&an3yb<%stm$>E*wJDqJm=KN;~O!YKXWw3>hkrCeh$;l`*LoBbTd%^P!{2$=83|x}Q@{gi+}rcz7h-F?1N5jK*W&_N;w56y-fO(s;F)hBc#_(gdy<1l zZjtYk|H#f~gmqV2`$>Z=DwVjm9n^cM3(L!@Acrg!sZ)f6mCp=KO?}6kY->X3Ko5r00nnV&hNPk3k8tO(P2Z12!}Gwj-%#V$msQ;$PQ{X|}6;SwiccjB@1 zxF+=#6WEZ=ymTh~3VKDqOS8o^iQa(w7f8C7QD#-+?JWW&1L2?|=7HrHkeMogqT}DO zdJ+Sw3yf^gHz)psm>j`OfXhM&!2AH5ydwl?d`#sSD=CuI-qzl(!utRN_J7u^I&j-3 zm%Oj~KhVo`tw#!&JYt6q&f(GVm2=Z3zpo!-JFGob_t4Lz(Y?Iz0%w z6yo1@Lrs1ydnwnO@ahac(kkG?fIeyD`IKBocJ}?jMZ0Lo#!5Xtv0=d$HC*>ury$1u zXGR^1`5nepjo~r&q|cRgCqr`{!=Ik%*y%lgh&LB0uVjM@1OjP0W8&Yc=D#QaL;?UY z0_7oB6+Io%Ct*G3Kzx4nKj;=bKCbN~SJrx=x<#*%{-2S-nQMTg9*E4jI0wX$77%)&n>DieXb8qXukX zUXK#wn^f_dRr(~);4SkOrlt;c6M`jss`U*M#}sXG8~(YKk^XveYqL@c)x?;SAMcnx zgZOBM0_m~ylPQolO^#lQKNkRih~gyK09@fi-wEY~`N1$zXS=W+<%cuy$fnT!H3~V3 zvz!Dj=50^px30izhrN6x$wNMtPQDT3CXzrLPtRvQF_Eov>!AnxPbV7+##YGY=VR%i z0nJW!Zoz@9BcM{gsf4yFj*TK+IH=PLR-sH5PS$wKfk+W zlo|}`^U_jRH^g0U1^Yk$8uk*FlmC%rmi-nmAP^=b7S^Cr;Z78!pdN)T? z10b~h3-1zHIr1Ff7wre`pd{kY2?Xp}0OB<8Hld{eO%@Jd;Kzcjaa;$q(0F+JLv=lPJ@=-xHII0)y?4e8aE=q;lty;#X|`ZdwL z<{+}f{X2YoVYOLLz#Q!RmS8GoEJ0G&$V3jsMwb4pV0{ojX-O<;LS)&zBgEMEP(nl; z!#`Tv4~x+555hqjq$p7yVp8`CX`j#W=l&}M-=peyvOIN%-!a-kyG|y)*&#qtTa(}u zICgb||%;!np# zbdWD{(o4Qx8g_d(7+~fyPqdjO%v*$Tj(;) zRGHnhJEEBHzv!{I`O};)1n_`GCE{1jPl*ie7L z_vx_LZZ;^$#?#xbz`;i!)jR4eCY1o^zsU{?&{Tnsyh{Cwz`m>kr~)S}O;!@PZ8)(Z zj%w$B9DV^OvACI^migTbw;0qs=P>!kxiNS7S7lda-u)S&4t637!OY%a*+Cp~_@#c_ z>56@2@@dwp45xtQ@f0shONq&^P8M%v$b7i?e0UcpwmR%$(cdtbZQeO>aps9oOWNK& zZ6S!y)UD|?%O&Rk*{GqZ3Xr{doM#H{+*02H(H>*p7d8|3I`xHHV z1x0)nmiw6Q)|oN3c;3FI;V(S6{8R26Nq&;{Sq+eY<>fWI4`Q3RYW+ys#Uobg~CckqVj zwlzRd{?G&(^q4{;25^H?O_PZlh`V;{C|mJO8pZOsEFu19vjGb}EREkl6oS@EM>bGo zXrPL9s-NB!bR$TQs+A8e7-fqGK;KGLtSec009g)iKP-%_6he+!Yl1W^c!j%xZN|=C zUd>RCT$Ql9tSD8!K`z*fAHSlWQPVQ@%FcSnyWP^9vnGkOdcBU^avSJSkD*Ky?07bw zg$X4A-I)fq!X}GU%dg*BBTYUMPy9bg9XlXE9(BUNK+yqc)+_;5TVSOmP<9aXnJe+a zc?1)nR6&n}08x@8K%xN-$ekO*>K%G$M-lR6#jd?T69p3TD5=x9ET*FA(lE=&civZm z&4deFz#@DD`nApN`oJS>baMWBR+q>5bB`BDTh``YHL$#v)t*9UAvl!vY)#X z#GGbR73gY7PU#xnnva`81T>n5e04)9%@*bi#MuuE=NN`hAu`9Gyat!qXjweiMYE)n zK_Zga3%ur_I(X?7V@Uu}x)X~QD$#e?CuqG!f&x9u;|(Ua&*yy5sAW5=?d3S2f%y?m zt@IoOe~^*)?1iUjM#B~C|BiJes)y>00Z4&p77r&q>9>GPWKMxy3>B@p21MXWrVnmi z>ze5y)lEj9FsqRWO##I+zxb?c@@881FwC1d%WqyJYlO3(S>RF$0H zZ2sgct?^_JWO4E7!tLG(y6J`ExmOQ-p=p&EyayA^9;$`3M`Nw5Z5e!)nrO);TDwMm(z7~G&7MM+f7c;J;Oe!|qeoiy=u zR(0*nr4HK)-qjLM`&;_y<9}6F;}<`A(D)Cv)myz-*iXl<7xQNPlYz62E-&1GN@CwO zC@X`SnhF8R<{yeC&>$R*!houjb2Gq8!G(l^{&^e}vBKCSl*9VYK3`YpdRZYl zeAf3|#oMWwqMG*>0N%pz5WJD4d$2IYuXT?3tY}M$SsgB_&Gzz*%7Rh_){sYOk~ity%qyV9Wsw;(W;y7=S4sMspat0*E{Gk8PT zwTaOT&brAv-~9K%rSh9?>>A&tsu;(1tt5UCI7a>ll1P%97N>vv;L((3TW=^xj{c`; z>^iGNX7pe4F06U#4;R<)uCu4I7KXf>fsKC~ko=-QE89@|hOR$Rv2{ZVA`LJ;r+{fg z;J=F$D-sugsvQGnWgu`BP&ojvJ`xBh0Os2sAzBJz!V1L$bwAjaR<31${YpR-mNueN z^yyuRJFxdX@8rEAw{!>IXK8F?;O}SrEdBbG_MOkc90XFYh=&Hf$G(8CRGeTaD|4dY z1UFOMz@=3ZU3$#@ik{Kn_u%7|PuSXMVhlIa%nDCbA9g&J zG!PpoS34|+C%fIau)8L4llQ31%xILzUSily_|cVYr@i3=`(f#3pn6Ynf2I0IOLlu#~|hVgH0xRAr25TC+{iLP#eG)Z%hhZr@4%`9u9s z99=y1bqI8u@-@u(_;8HjH||ccSMB3VfiIIjmIRer2y1lHpxPn0&5{N8t_}!35hn5d zbM0StlQJEq_>|i7-(Lh-+L*lEbw%oE;rf01rNO1^CRHtay^=6?-83b(rJPUDs`yja zpAtqSHXPuxp1+)i`B*&CUOoSUM>?`Spt_# zUc9o0n+iQc9T%@PH=J3-`{tu{9t&)l@QSQ7g~^id*O<|Na!a`+2&b24lIf8%bA#us zBr2e~(^~kV7pImg{%fhP5eyG_4*@7DDiCu4BOT%dFbD&r zCxG|x|A3l+Lp2a|k6?!Y+S|B1!TLXp9k2IM7Y{-x`B0=u7ZD&-o4jX3JR)FWYgEB& z;b0rnE*6U%3gr*C?|b#=FJrx784{P$aw7D_Ww)&j3$;mMCH^-vh^1UP<4#}M`7VMk z)p3r{4|TC<1;1HP-(*~rY0%LF-xsi`eURn(`WtwEl0;=YyXcF&ik}5)%s_!IK#r|e zv3m<)kq!M^zKo2ue@j+x?T0-vLY-nUzK|^KWjUrBmm}5G%^}OTQEUO?%0_h8Y$Hz- zif1LJoyGNDyE#8;5mj1R7N8Iv-$hE>!E~QkhRu*Qsb8xKdPN2$(2xz1|L{8$ys2;q z;?aYDqqhlu(H&9YTR*IjKeWj>e`O(|&^^A8N<^eD(qAkh&gN;^%(CgupO_>7OnM^k z6bxD?(o>Pk1=A=I7SOF2R(76|es81YzXT70_^Q!#V4;zz%n)JjT=Z1cJI-zst+>dDOilbmwyZRU;%{YlVF+Ig(2 zU~5ehwxobM-;;jD)rz0z2Kaoyw4}BET?Ywsy!dV0$Se2q#M=+?X2Yuz!jxax1lgcB zuRm<&r^mi<;XPY%vEAU#1#B-Xzf|n^S&66LD+A77xj`MX&>l01J5>4~=x?zXuDGq< zi*>@Z56~#uM-S3JDXaf5PW?^eN{!`ex^jpcUT@qQ$1dzdF+fCTT|=wuY2G=f8OK(v zP=CS0SbpE$xzJes)#Qd&g^JQp4ag_k{+`@8ACyR)GE= z3&-==x;SzKn#ax&RO5A93Vxx?{|DkRP2_Iy|DuYz+N|Gn@<2ZfmbQ}p%3(;a*8Zoh z!RbZUa$uY}M}6DZ{^%zS%q$dp)mkMp{2Bh?w@(t?icl z%lMMvew;~Z%fxwll&k>R^|ki2ttzE!&Qc-VQpUPt;wjtsG*wlxXho^;SrXaDyvBW? zN44TFiQ93x_19No)a)3~&NvKAFV9BY>bFlzeXe!+TD#b>^3nf+&YlwLHWE2N>BJh9 z4GVh=8AZEQ$WJvCH8!3D>~BJCZq7b%*Jd~%%7m$N)cIcX)A>Q*$Q|7gfjj!I^Hwa~ z!|Hu^Y;bBn@s?hLHAsoPXGvoJck_2t9|6}NoO@w^*uMK_&iuRNJ36L5?$Go(gHVzmfvJ4#S~RXB+gKBWe8um`G0 z&lG;kjy2ZIS@=Z%6*mg3zG(gG6}+t*v1lrD(@6PcwqBuOKq8>jC$Dm&;3PDbwdpR5FDrz@x;XcZKui1r>S=kJ%Xi)+=P9S|mUtiVDv)M( zPE-}&_!3wkS`@e6WkfHEQn6I%e2Ma}<9L17TzhLe<@{Tz*1EG}pE`ey_=#f^5*|W5* z^pjraF^xF}$0yRn;V*x_0k7}{0K)~7gfV}M%9T=YH|wd243vUn1Uxzx!qcQ8}jmKls{RRY(nO!eu zWAHT-Il`nOcwh~URo8e|;%@$tu-Gsyi~yD3h#4w{L#)){Ao-5qbnT6mhJU7<(>Boe zeCyK?-rU*>VIIwS)e{*Fc=vJ7N_z2&XOS&Z;l(5kZ22@eAz|C5HsnZ<$ixbOO>zQd z3-n*HFf#BuLV>36h`tNRNKYII6>}c-Mt?>kX;X>urbljSgExB7=yTy9?O)HR20uFQ*or67yAa{b!AQx*IEOsz$a zvo`C1XJJQVW#5pm$TyUWkN^cdm@Wz3*D+%%nM%n@(vbT^hbgvxk$yKNe}Pycnqzp+ zRyp${l9R(AK}rDIYSeF)cIZ<)VBQD;$4mT;d!=PJ8h>}AxSPQK;&E*BWw*=|Yhdq< zyWI~!B}z&-W8F3MDuJH1#wm@uG=6jXNhed}VCg}E)o28~&^v}s)>zPt#pYHKMk9G(ap$5mK@K1@^N9vq*$4_}jfuV)qa+t2_9%)r%?^ECdUid8*++a}j=8W#JdW|!$C>mB&l(*l%;cZ~K0+-B400Krzq zhI`w6CrW_-Ru|5usAdS5uNBsq9W@k*M`04Ut^mVJ>2jdH8B>7X90y`$?9miArwMo) z;l&4LN^p=^kPZi5w_x1VoZrcJuKBmdzRF0(RXr^;bs~=nkwdZ&El!y!+m^o;?o9n~ zik=`c;1td0OYp-4;uw6aJzzbZ@`NVg?L6}?B(qFh^;Lms9~BpYt9cym7{AhSOc9V$ zD`41+X%GOzWFQKH_!SJ8lb#*uY)cQ+oX~T*l?-!Ih$syb^qTsu_*INSeYsfXnGG_6 zHf==DD|;m4vX3dv#`aX@Rvp%w&HKF%41Nt~;yOcD8K@5J>f8})k7>p#>gf;b=etF> z9&CR^Jxzu1j|~bbngDpZ?+<@0Eu0GLUKTX*C?Nc--i>>_`5W=R)~U6yDzmsvB@`zn zI79e!9~fpNa45#Q9d+YKz%iweT+03t0UTEQKey3(!dS8j3sVZPyvJEN7DtGc#scYK z9E5FZ$^lJU9C<;rid{rvt1`H;X=OPnTiAgFYk>e)W_QRwTlV)^J*{6wxe;eNW^cz` zxOu88U?OPXne^|3F(5G8uUbOZOa4c?U+h_`?zZn1H%TLXxy;3Qxv7a4 z8rf(+biiYr3doVyW8WeM4%B&R_MfIsuC50gOsAgTLtH!W@R{Z#E~h67@t`>rVG`CH zkWdLGD|f91AZf^bQG!5M(G_)jDDM~CB(2Rt?#HCPbK(0;XGzg=WKFPTEwCl)q1Ysm z(4vjU1Q7R_Mk{y)Z?Yc9D)IBP0vn&${r`b#XpR!0fX4WAhPNvRhOiT44*I0*=Y-wM zGnpg}v`9gmNDB7W9RbzAOvqTfDvJ)7gE+Q)nt!(b%7|_{==3%@6mjdfvRaOI>fO?b zsSN^+V<9vl1rm7t=4{T&zm+)+=zkDwf1bv_O)(%E13a4K|33WNG6S{w(em~GcX5YW z*sV$7d{x3~hs{@=_7NV6N(TKzzwelL?VMcmjb5T>bTci!GVbPj1LNDQ}lKu+&W@8@*n7zejcW?W*yfLAY)*T)6cMy=fqGvut5fQ2kTNdF6T>CQMy0c4{7cgZ7;A7 zfYFelRYEpd1c;Z!S~He0_7Dx*H>}I(cmUa(9ylgKS+-9=X=E--1 z>|rgxd+P&*&|t&GS}y8NT&0#Tt8qBBpK-ns+d<^}J`+B8i}DNxvl%Kh;{Ac2&mMiP zt?q@z6E?bD4(aUe*?=2(19jo!ya??A7UEQ9W%l!Y-c!EY%vy4DZQ+j}v#xRy_v6QI z*gVQ35Af&i!RJ0a3pH$g-ueH~_2%JFx9{KhFt%i^QP#3Ewk$*T24U>XFc^}Z8IrxC z#jYmV*Rf=o8S5ZfLs8ah!Wb&aEu^e%wAcN+bbs#8_xt>wUpA|c zpC8HpTJ0=6jN*&KBz$hjKtIC^LI&F)!v^A$HEh#YX@V%H*@AREvV^%_|GIDq?v-_p z{-xK3vkRqhw@-ptzFDhp9!8cFO$D7UX_McyM*64^^Qop(Icm+^jIG-30l#%wbPG$P zfjYRcfy4D)bOq}a$S$4)`+Etnqy^^iU_qMzNZdeZwpTFe5J04oU5vb>C`LY0r%8aT zaUV*X{!+x5me+p848&W9%5FRy4|ev8U|6fpE3wE;ttbS1L1YiU?tQp*7tiNje}*>P z=N}YR;XZ*Tcz5ic_mow%coanya1zVFxJ99JY<*72vH4b*YXGA&TaKMt4l)^VA1>qf zQ3H!AL7lY!Yoz!BV)o2PS;si`+~obp?p-0JW;+cIJNFw=o^64lIfoN5bput0i5Hf7 z0=La%d~BsFMv6nkxNHSoLnXEDJ&7IBi#1BM3<>-6P>{o_me7?m%ym2Az3~se)mx2D zCa1Q(eG?Lz$q5n7wZG5BA(@q6bZG-tK8k31u19WBBfqrLvx(4o(d`Q+WdSs3Gra*Z z8@Mj+$?(yPga79eF#5m?L%X}0my@Vz?~gzHdA#2vfK#s_YA$_ka5DsK%2e}6yg|d^ z?BDE%^=+z-35@MD z1)*t4>X($|Nz(VF##Rfnc>t1<8GCI3S)R6@DYan> zOC9{IU24>qopcH{9H&th*mCHQ^2u}d^bBv`Cg!1U8imj#Lc}}g>?Q-FIt7^pQe?2~ z2kQa2C$agh$?~U0$rXlXl0GA%yU9$v+rZ;fBX`t7gQoaR>*&aKPe18gZx2}iTD^3u zm;-|4QxIa{+|e5+e?G3JY>`sGmYfjvMzO=ir_A`YfJ$S4E-b={@GF}}aukV2nM=K? zqNXVa>22A`Ogb@&x^RA?u33aMTUOc{W~c7u#bNk2OsiYy_MropLiAj7<4Gi|{gN1_ zxae8d*2czL2j^`usIl5Z(7d@)(Q$*&RGWj}xQmCo{H^UaHO8PElyv(G=_Se+)H7|4 zP^mBRl$)BbQl3iGh%eQ>w2P@X4p7Ot!VNuTo$u4swHaDj(?eKxO`DSmp^oNFQAU(i z{VvEJQ!a{{pQ&OU73OA_C9r&{h4*lyL!-m?AMlSdvG0c6 z=Yvz(D&K6Ns*=kp;8kNXneDrJ4Yv49{wD^+a|^?eh@1C|#AE8ckuMTPY4>!KacaRpYy|0~MUg^eHDB{Qi)}87R&?pX z#nJ5%y}&x-3FFEokDD*=8*f6v(31R!;}YbiO|SG z8ng!2BO1Y|zNYU&@#KC}b}ckq7^W>k&2!V>Mcf);w7Rh6o~O9XM`WzFv|p!qoDCIA zKPvqVklcXp@m80Y_TzEYzwTjQ`BvQsJCJ{O~t%`16Mq^aD@+A8t< z=mc&gZ^O)0DDz0Lf8xVCImDEmrCx(Z#chG4zR%6RUmObOOk>MC@@srWwTjV^JE6%10wiPhJ;8M%Ze@aAd8BA@j$eAx8K)|6`?q z6RNKHpU|TZC)ox**gfILnLfj%N6070#g0*{8CzquiHS&V2uD(75&6T-X{jLn;1IedZ`jN@EE&SV^NVc?DJv2VV(kUc z0ru+uV#o(-rqu3;xuuVmjQ)}JUC+jL5FM>QH47d-^ zbbY?LjCA#yeUjGwER2|SH2$f{&%|b?WS8r>8iJCjJJx98TV1-8L{9hF42M^>nviADk#SJxY6S6|(xFo?S?WYh9%ed(y1ZzFQ<1I-2BfAD zRXt{^CdyJWif;3uOSxa5T<7Jp`q1-*Ybg$47ei@6{-cusuoIdQm=s|PSez^Eq}OH~ccam^LryKf=Gmp< ztb)r0?x_ov&+qt2+b>9k36g>rfSvfWs{ z#=5OcJ4<4C3!u_sE!I4I2c+s>5QLVf7`udZ08ye@Mg1^qH;@9>#u&TQ?qSq2u{k}o zeV|j(R|wCtP3UX$UqqMnIFdKh#$Mb2k)=G-1yP>pbLwHsPL~N&?LVG1OpWHbZcXFq zkxyc^Lc4D*lP+Ekx#X+1=W|yvAcR&V^Pym~@f0YB|zz7GM41J5%g!!Y6YWRyK7H zNX>5tjnSV6IC-zz{a-dz?fKL_l#fE?3_^j>82|;y;HhPxzN!<`+^{BN^>oA}NBAez zn?1Pp+@=OD!|%=+Rf%ja?%s;$ERlz-s7wuq-gIGY;AwvG5zYrR5vIon{n_j-AcZ*Xog7S2R>y!S^rq9 zuwL`JRgQC#$>%1IUX5^@kz&HNlZ6@R@CcL^dYmmsM*>whJ_uIEirSJ(CH8FDc*r9D zJZy&J<}gH3uhx%uT9sPT&Z zs#oG_K4elxELPWkqubFV4OV#n5B1O=^P>N(7WP&M2(W(V__tdB^U4xwX|EK#y<*Ec zL`)*$yL+ohHW6&|Cm!ODyoh-04|u~nCDR4I%8V)!mQpiwSoFq$N1R`#PdGV>1>e38 zxs;}~a&GHJ8^ql9UQiL0rNuY%tOYd>8`#hgUveL!Zx z>g)4twO{dJ+01hNjYgO9b!ih?XHcnelsyIE~~5_2fsFy|{~ zxPi|~+c&q~)QI@{q_6O(`OzB&!AI}InWBwS&-^m#kIc|DHR;l=lF3hJ@5DW3s(z%p z2nVeEn02Y5OYp2)mWT{^N6Z#jSfB>U(C#&Rdbs%g4`mo<0n`Mna+1?1n1ySCqdJC_ zVt9_KD}mS0MA>sF;gSxf*ME%U(A^G3BK(ii6g|aPQ=jk!i^F%RP+^I$TPC>hwLI&~ z?|{=95F$tl3n{MVCzgq($)DsW5G|za+!gg_^!TiV)UYQzk@M3!%wc>lK`%Aj%ULt4 zBo4cMGg|H{+>rPH>C#=FYtd2SfBMeoO~T61)EifG`RD-#Q@~ahdCA<)%iu;ik$Yct z_Dvj8?L7I3{`9#*U8lS)yIeQb?2eTUf!~WOq`C5a%vhqRUDnII1kFqA+6#)tAw5{c z@9oHu+pdoNz1fuf6G7>e1i?gWhC39Mlr_SV|4yxwqpdz~q%&Ef;mj2CgV7&BHk!HdgxXDwrwj;tmF zffW3H26`!bKS|4uLWii~sem)Yq3s&;Mmq#Z%++AtG31wty*#;>S(BZT^IXOeyS^Hp zc?FN0m>;MSj`v@5(}ZQ#hJEy8&3s1>ODM&7a_6{8)C>D|*EUKe9oMbT>cDdH z;;`7M)=y!j0e3R;Hnz|!)#*YJs;|s3MuRW7cX$2XDl0qPi|=W(7dr2<$!>tY7gB=q zk_HMfq_S=q>7(*&RIJD4EALtEUfnlNMmTw{>_K~behHxF?(N;JK<-iScy-1k#7+@N zR@4Fd@JCKLpjchh0g(fbDYP9ihXFk?W(p)&gTgCCWsuBJY*Z4?43HnV@7eR7++Xz| zdbiOzg;m`WHjpKm$%`t-r%CqF^t3?YU*(C(8|bc8LDzRZt%N5XdMSug+0Bp%{-YM#|Wx0U_kI5hUkng zoYPTlB^t!JJ6W00AbY{rNEz$M?V>y2esx$r)mKGbm$j^|fY2xH)m(|tI}Zef0L#l>Bf>subp(@-?V#%fzS<9MEjj+id3@syQ6RFxhy&Wh3#vPOx@ z&!;>szxwDE#yLpoQf9wApFT0$`%AO+Uzl8(jx3?BkUABB0BSEj9ek3<93QT+C*BYj zWMT&Ck?y(w;Xk4FCLlJzkt+sLvit;`A|MI#%BjHBxOEvYyx4?e{S&IW7nc0k6@rYi@r%S5^>WsbZMVN8hNUXx9`!%ZqCUW zu@{HEMUdeJ+ztDLlsq4n3qP&3)TQw%)K@o56&YeWR-AYz8=;eaH{(bPA9Bsu8JpSS z97duYxx1dpDukhEZQ0J*!>t9`eDR{Y!K*t;cvX>xh7wGR9!Fu`5B(4wRuP`b&l65) z;T5PHSMyhF%I`zyT>n54-ikU)VF6b%Bssq zDB16fY}{nzvFaA@Gxt!);I0|j%;eHOb@dvSugDYsc1!&xnoGTeBC>)OQX_ z;Osvk@-_lx_HD~1%d36Le$0y?A`@_?;nI4QIJ52U=eiGW)~-^wx+~=jx#NgAf?RZz zfTr-Xi&X0VVco>9QxWrl<&je;pKK7-H^-;WXQ}=8FM;}1WuK?J^-a5-&scFKuh)_% zU1eLtnpzH`)}xOa{2G!y2NFfks+=+TVLosnVkqL1`;Qn^6aKkoTaBaN0*)SiNbQ=Y z(MHDwUjACr44HiJ+6ngvbAGyqdtv?DreB&y5a*YKM}oWcm5j=3U3UZWddDK|PG_Bc zIdCMc;?WcSMKIsIf1flZrYNg{ghIU0OGUyFgu^aT;PTgd-){h;! zLOAFXrDx;;TPF}t%955_W-bMr+-HfgLOqr-9Z3*VNzyj%sMtK!h7ZUe8?As>x*EQY zf*V{tOe*kunF)7M+7vjLsZ0tRqo=m(>d_Rt#iwPiD%Msg6C=-*z=L{m8$3k%o3 z6)tY3UQpfGa(_Zl#Cyix3~W18v6$cG(6?}R=4L&CH$v*DqAn+xF7mdmo7&5<53g*^lt)|Se zOG>*r#bdrou!D3m0fs2X3dx1%g_A2y(da28@2Ixd5<_p0r)C3t%L>I8_{rh?z84Q? zWEZnTZk?YgVuQNtQ%gjL+AP#ZC_|ukeSxZb`8%w=i1p+*``|QVmavWzX&z>SmNkSE zvADp5=)_}+Y7rw5WtiJY9@&ZxFNcD`ddidYb<{PI@FC`VI6=2eQ`KfJLD>2Yn{%XT z6MV&gwZam#u}<;69OmZfX0z?6VZ;b+t{{QhrNYP6xDNz>jwV1Ii&1(amHXeCyg~e& z4O=xEM0|*l+%O;L!xeR4tr9oT*N$_a6M?1kqC|83=-T`=se|Ch3}FVbW~?Si=1$3O z5gVqg!k$~`(li*^b)&XSYi^G!?a2M7TqYB~(HcWu=VN8&WL9>xJwj)<3c}bg6qj{s z&2og@^lNpUu)Z8p(m7Lt;!_UXKNAifs8_T1nbchiJH#ovsztL{Hh%G`mMN%!IDol< z4JG(F?*Y6D;E#0R;tJ8S83qu2`(9B56%^8r-t(pbHjjIJLa;jfp*GB=?9r2!Gxtd*~Qy>&Wx%A@zfiCNPV+*fi! zj?ep2+kP<8B;SWLRQO%3Z6{)aC(r2JOPp!oY%kEyU@9A4jTE`3-%84OX#8Ggp`ZH# z^Xce;8Lt&~)H?5Rp=bij=)?iiX`PwxRYK1}D%scO>n={Fq!5&cht(HZ~POzSfF(d zXpGV3*O=Jum!w-GvcnL}SR-9u98fb zIiQa@nEj375)_G2(T*)@5PTuutPo`92m0ZN=Qk_(^^alCU3%^Rc;0nr_T8 z#n89fYahjRY>bm8FYWkTmx_8rF$^<;{dN%@j8MMV+VfLcWlFD z=)b={Q?U6LrVzE7zoh#-tCbewu%B>zK&w6W!xyy+wB)AtkhTS3pCffO+SVaIu08QD zC|wV!aI2$rh^And#qP(+wFhmT)AF==TgLy!Wm{#F1`}gbf_y8e|1qv0pO-xNHn^tP zxz@=_)!?v*V!xo5`~qr@-H>Q<lbRm_K+_C+qaO7otqm|4=Udz?X0NR_};vmcpr^Q8-2jH z8G#cV?ZEq{DAu26?XFb|^dF+;*h7tHV#R`ArEeWqSlKLp_SDF#P7-y7M!J5kx7)zS z{y}3rcTFAPSfpuIuG}l)2g<6csF}&$ppFU# z`ZWlij1Zp|8TdZQx)R59P{i@Q`3&Q7j#rv%LVRTYXOll>GBDbLe?tzuzr?Fi z^3nZ)pJPGc&0OVbasqe6zhio%JX~?kXNxj?8kIuM zrM4~$@gi(a8W@L8f9HdD(*} z3F79*2qDo+sj@_*>M??!I3``01hMoahCNZTuOXU7%qB>wrwQ%1OdiIm64j}oE-OBm zEtAMU*JzvEN`cdwXZ#-PcPnbe;0bJJ^G2;nn9wv9<@^}1HBJ(i!yKE$c_(vkdfKJT0)y(Tf}A~GOE(J>dJNL?p285 z7iU}&A7J|qH%f{V@8{AzYK1tSZ@o~mSAIAYkW<*IRbV{o*+RM$75n`*oP{1@b~!h* zV;Os&S$jtlU;SB;t`UHJXa6LGbpH?$%!PZ^^?!e`P=HtnXoUWw#sB}m{$YdtpJ>CM zvsEby%DU%GoRdt8h0F))6!IW#{-JL6FD*2+HsZ(y&UK}7Idp#E<8#NZ1ow*f1wY$` zJy-M2zxesq(Q0CSrefpgnG+?c&y8O|XP=xHKwph+T>Z11LH<|gl8YO%gi7B zZIt!K6#w1&%@^K@esdD;3Xb-3F!IVRyvY^j@i(Z>J8nlmY+gC&_{jwEyTkogWlv0z z)ibtO&{c4dOMa2$p(8K=duh45JW?q<{Sd`8u+*Wuwch9`(cm8GV z()#1Q`-}XU+oykf9A6}G_K+O+c!vtaC&K5nVhDxw>t0%{EsNN!%;o<#`5fHlysCQ_xG%Vk zHGm=r?1DhubU^%}x&&rSft27H#Qz*NiE=4QjnH@a8ZRO(Y2aL4$@4pothZU5M6@OY z%Udg~KwHKttwd)oeS7*1P-z1EKH| zdF3Pzoy?4xTdK1;tJ-)dOU_7X8z@%5APd2y#pJc{u3?-p&tuNp^HWTV_(h&8Eqe$y-{E)RCHvhJ*wOQs$NE+yw;a^=5gAJb?6wM!Fm?X{)q059+|qz=)KYe!0eK}GxLYM(GJqx z!Oro|nFA7=y~^|_fZ>nyrX3tCIS~ptY#^h7C3sj)E%869zluz=@WK5fb zSUk&%uFqaS>$pm38S2tsnzx=HF$&#kxf!In%Wm4Ibk6&d;+$x!-91U z&Xg06)#mPcB4)HxB?{b>r*7Pj#!zgOA*@e~!{z z*z!OQH2Uqe&!ioZ!A?p}76Nqd9KYR$+lJ(_XwdEJ>P46BH%7Rl5}tv&&q? zS62dG4s=+Z2~nLdb&A?Rzj)SOzxw#TZaS6_@Pyi0mGl_qmHMM2+*d;*sqJ?^>!W(Z zuEOt;h}iWb%YC)gzmyMzq?bIpeMq5)YM5H8w^~B_u?p7C#Q@!wYpc{TID9S8UOg_vhV#uJ339eSY$JI?udypW>j-b*Kp=>AUWoszfFEcg4(fs^VL?rUr^5n*fJx)knr-Jb(60va=Li z{O&SV*1g(b5E+40eIX;Hc=7?wzz`E*0{G)Vsc*}(ORYlu>YIcPRAwr|HZcOy5u^xM z<)WkbL6k2$IT{jp)6vtaBW_xdZS(MJ@jSOV3QJ;vRQ+tV24 zBDkFMp)+5{0#?(7RE6E>Hfyve;t@EV?+wN*y2iWWg&R_mM4Ehgg)1$5_R$*iS-PE8 zoB=uF0e7*k3@wT~Of?HC>C&xj|AedMNT&z-=)SyUi_dsMhHKq&novLh#_MAJLp8bp zxHp7Fwf5#fkSX=2vHdS$^53n9H$c!NL7P&%N3q=NV)n#Nm8^+iF8p8AO32;-&kjTT zf|L4(oB+%K_j((iz4xT;*`4t1pwIowI)bwUHk}iaOUZnYif-Y!6JA`D-ax&3In~0m zd8I1CzB01ib^QT*_UylGOnxtHfs@mPWT1Qa@ZdwT_m(;ZWt zXBN#B{k)a%VV?Nd;aDSlU*=o$y=KOAdtPdQ79#`pE{jR~B}sJy2()4=x<{ zI_dp$B&S)mz0K4&B)z`x*V?bUey@3X#+LoY z<#oOp5_T;64#5(vh?^lefJ0ij6$_apw#=fHz) zQUku9C%xqKjmM6v! z>^pwdL1dk6?RD3zEsgN{laP|CZ^9HPEaekjQpxJj0$%1!vtHwYW^-qsWRyj^%NT{V zy7SV9r6wrR57NG4`{a=aloj2zfMYXN_Dh zWp}M`9Ums`EGEErzaR};B!c7m7GWD;#i3=xmLe`1ebvtFiyi8Qklt#UQs03vAwO+1 z4&(!lMl0V;&3znAjNb^{Vn9dxWLeb`+0GPo)g87x+pe+UWWze9jue`3<+Z22En*ve zP;c>dwV86i%T$^B3}RWMg3wyH%KSNy<8>;u{w!3%;nplCXTO2=#EbO?ByVH6s1nnC1fZ3Xo&{!%8A)0jLm)Kz{~+A^p!y58Jb@H-`uV*#4ig z03L#W@g{Jt0j+SK%ti4d1*+I5O3~k*iU5*vjmV>8K1LNIOv)MI_1O7Ld-mf}aZxA+ zyNH6&PA%%=hUkhoKW?o5o(i|c%lK)v?YY3YKE(z?>g(^$)+amlb-TwmFvDm2?>a_d zB|GqOKzW$<+!soTdJ>Bh>GCPmxv%jF^OD=<>R{++p*S&k^Ldk03Agv`p6;7F9a5(A z4w>X|W%`h>kHquRPtqOJE{=Bb3VBDX)-NZOUTAC>IMej>(RHKvj==ea@hGWWQ>z^X zii?UB$dv0;i_wvL)cPj-*7s?ka#w6d|F25Pk-$RaJ^G)^oczc4+X9qRaKY}~@c^P* zvoco)hEO1Q4%QFM))D{?XJgBT*eexI0IEDtn46c26dB*Cf>0?*b0KNJ=02>S7drlo zv2d*YG)m@bG-rLCd~c~o!I>?S+Z`qk`1ZTjralUCS-{DlFrjO&Q`Fa79vG@_;n!m|eomg^a#MB)IYn+*d1)v=5;lI2P9U@45JOiZxoYwqmqp{EL z+z?ZxsuMpiPJ+}cd+FT7ec;(f?ZJ%yb|$PR^KM7iQ99?(S9_kgC)gQh2T)$=VgdGwXln zV&_lxvCqv(@BJPb@9GQ;`1|`(hSh^#bBzj17u8z7?4Hg=TED+sP+*kN`YD5$E+*(( z!Lg0ES2{ixTDDrYZ-jtkSzINBJHR%f!;+4?tO7ho2z^sgWiw?G%*+IuJiEsQR;w)B zq=c6H;_>@-AYTt`QkG{g%aoT~+-iAzTJF0QRr}-pMq>vU3ft{5=U0 zc*s`)y4vy=^HV+*_-x5c_5L5s3_eAO1l(!@b_~jqjN}QRyX0uNnwv-$jt$Ev(0EF5 z?SgH}+E4L~I*d2Trh?H<#9+NzjBpQGMw(=aSPTXl6j~(g4ah!ccP!}=cN{X8AkuTy zmEbD1iEZSZEnKVREq1#XLTd0v^;_2@Rwj=YYoAVcF=OUnk8sV4*EMLaJsf?#)T6YB zcc|jqN6PW4xgyZW8sIb7jBk3A^4dzetSI+>_tJh82p7 zT5A7xpMT7a{kf2)-qJjc9bB)<75vpoww$pBtFR|8kS(o)zQ1o-zl-Ua5&0y2r#>oG zBE)`nAmh-05aHoKl8|F>9E%5@ssgc!m%J8#3?1BZE#n}wNIK=vJNW}Jl}o9?a+Lm} zcgF(6Mo~q1w%xM_bo)vXA4k88vdtp$ZuiKpRFI#hUVZG@o>-bgxzeaT@UiB!lGK#* zEXH`*PBxFEOwTX-3sFJVS$Jg*bG~Z*CugEMj|NedQ&SuUhM=(g0g#dm=z-Zg|NPeu zfz8YwQ>Cz&T@H2xbp60b5P3S*pZUx!AhvTzB^H=F@~02oLv6 zz~et_uNuN%^-h;S zW9zsz#fgyoDO>*D7T7unhX&6!$$p2L*AH>v$5B9;nR#}HSQGI#<$*M6& zGq0oDcsEp$c$tI!u+DXLwPao-FH{0fsJ4Dzo||N}&ghN++^2RyYoK|Xg9i^obO2+5 z2ss1mu7Pn`o5GoC3Gl!8=j(U0HF>Ik!Ha5iR&R2u*H>kr_X<&lAgL5+cSXoc#U$M?=yD4V@^i^|SHDUkEuuvRSMpY!#k;JKBVkbWA?P9SIshl+An$zRq7tnss8;0$y*$0%iR;E8y?ELj=So%f{ z333_V+lyXJ%$OF2nyev#wg+X7#~86=Umqp*O9FkRivJ#Y(M*0(X+XPRfAL>=!+04E z2r-y*mpn#?(xf0!Bo57>`@0R3BLPK5sR!@9#?l02B`c`e2NefK@OYT_8@*b842$B3 zxtck5+%yZt{N#IbGP}9k0D=K*%$|qBeoYaZW~WUfQ)d5Si>B%)={xIEl|>@F==;8X zB6c03l79{nkMLyOQM!FUxnZfN!3RDOZIcw(elGZLNX%g4{$oNp*R}6;i7sobb#B(wk~Zc#RGQCR99t?3aQYb?UX?a~&_6!lte4#Df^{5wYv*vY~6qW*yFF>S9i0KP~tNNlAOm`#BFHWid{Sbsen zLd^vRHQua=vjDHeYnhAxTi^G-n@<#Lz^}jC||n70zyj*nIa{Rq(j=<)16z_imi?3CA>TcdDs}?)ShCXivOTI1wqg zvPx-ZsO?(aooW;}nbp;rURSoXp7LpRQn}aPcOnk^<|ct@6tecdTiD4%!jW3j{cIVP z;{;P;8Gu4C2}?yCLJPs>vL=H|x3tA9#4`)}wqY*&GajNwf}s|mJxja_#GG;vo$N$= zME*eH@-{s2s;)MncuX`~Fhdm1U0}?toP<~cBx>C@3VG0SuI}ZPQMy-F2f9zVn*(rE zROMI?WbX-@(4shNq&yrxQfvoNWt+Z0VJ&9YWm!YP@sL+&nyHfWx+1G`rYx-L>Ikpu z(*=yJ1iSiC*@$_yr%>aLo3dKCyB4;WXRaU?pNo0qptx_@=JEuQ3afhBM&C&-es^}9 zFWbyZG74s0P#l)j6QPi$6hWH_Y9$OZk3N#WZtKW)*iVPPL+)5*uMhIPzdTpfw#q+J zJpKWs_mZVF?1qi=V35Z^Rk%v7fw5RDjlhcvR@>f8%Vx>x2~94MoSnOgqcXB#6^+m*9+2(m2gpFc_cvcgA5rf2kk^E zW7d$PIY=lY&k&u?k#M-Gn&igUzWLn|imQnkqgz|xYlVlUU~*mr8~69I2Qm6kXR>Ln z3{5p9WaS*yI=n|hiPm@DWzIpGn*RzyH^@lI(zu_w+LFp$bxEnbA<*hC-uO3f5G2sCa*Cf{d`Y|W}Vqo$jI^VK1W$3mS8fvkPMC4-{=2^%vwDA?ozyq zDoxHNd>#(tPIl~1;frc@HGGe#y(2em8+mHvrR>og)(R8hgnW{7P=T)!=X9;fj~?Nt z;vY$8eNV}Td~1ItsXbA0jP?svD_p5%p$EZpcRmwn%uG@Q+t zc+g`Zx=Lr;VVPrr#D*@7TLamcpHnycl=t}%Mx*UYeDkygistfB%7^W3qS^90j+WRj zXrPyd^oe1wEMA!>#K1}uVS#^cW-TONB_r?;G;D#agfKt>WQ4&6Vqo;&%-(C8!LNZJ zLa8^}%p}wHQ|fopEJDe~-#~peW2qPi`y1jB-uRWt-+*^Kx?nH%HsrjwvP~t^Jt%w> z$D;bhL_0lV=}~amb&v4HbJ{l*h`aD>ZaIEO?%$~@82@S8Y(^6t8z_SqJyFsNQ=vUQ zemGd;ndlOw?{1ImqZp|PM?R&lyhD*gUA^K@PZ_qA#l+=TX5N1@G#Ahp?|p-kck4xt z+f!rtnXpKy2%ampq7!K%^4_hZUlVb+o9~FF6cjIVYwqRg&VUQD z6OH27jM&b&$guqSh-WVDyLkU^sslIlBV2TDOFTJu(9Sf*Fq;NHtQO-g9-6+A@g4ky z*2TwByR{J39avwN*>E(bMp)EhAMXe!#JBeP3(84*2(*H5KFt7|L6t3?8XEsnM2p77 zFeV^xDt6PJk3#h}%-iWf(U21|t~td%JC_RmcOZrKtI;~`Nr_5m zk?U>5r`>ONj!nAuBtLxdr9sTJU^k#n^=!=ZE)l z(7dtJ(0Mu{y{eHMfNH>AxorN0stIbmDlA2Q?^j4MS-Kp&vAB{L3uXu{7(>YxV+S#P zPfkd4yu$tH&Xi0FyK^E8~v%LoL7nj=Ig0j2ChOzY#!KK{$==vo&^V5l({gN;Y3Y^`z zshnpFmcFeOXQBx#HBHm*c(~8tQiAM;tx+7B_3+nwIJWD*ULm0@#H5q#1Z<(Ex|{O> zPYYd+yn{;FzM(C+*1T`mlP)y0&GViArPNcLHeqdg+=ty!K%e>I_An3!zdnrLVo+HD z+1WOrdN6((7sFaDP_`>yCbJ7#kvBi8OIOEc^oRO~PY@Cr3CdTW-^sHV{P}FyzAu&n zEt~Qd^o96Vz!WL(Q1eU0Towe8uhylyOAYgHX0wG}o%(|B?KmBrAfZGl`gu`rBM}#> z373-5do@8wF&LW1UPasUnj!R7xK-E+zeuHtC1cH?j5c?713~m~xQWnH6d;p@%4AF%e4`tmINyadCk@cFNq>7DX_`Do@*HxRS6E@q z$+#rcLT7o+)gUnY0R9=z?9`HYW2?3pUJIIEvT>Rw6(jMD{OZV{pGea?zcpp?xap|6 z(J`?`|FE1pAJ7fU6tf!6Nzb~0kd2)U7DQvCeY&Pc#3_y$BP3a)YW?~*?AiTxJBOt; zGVi4D3?}bTvapH8VQ6-Ie>?Br5S5}6))Yg5swM584RT53FQzMIrB1hHBy((kPq%M> zt)@#A`liYKY(%cjHtAivqEu8gbqV^sI-40f_BX^G$qem%g!QoP7!_2th~6;&8&YUZ zxN^JhGk+lPf@oK{_lBnM`Nq`~8})xfeA*R$p&9$dJpYC?;7&hh#&g~_)oHBV623_C znN@L#&vOXza2C-x)&(2eT#8kAB>dgwl4+Lr=irQu!ZBkv8)v(gzabL_8eeCxwx{s# zhI~U`2V=mU6G~IwYih|W7A8-R#ZiC)ovB@NchkgVTWR+(3qROFYd-N4PZQyP7K1$kW>^`&~ycZbwI^euK>#_ZQz zip9qR+cp~87i#Y}a6;v4p-{vOqC=E;k~=|>lVT8AwckNfe_2TEXEQ7-|I1FJ@JAFU zYw3}#^zcEtqMI?AOUtuXe``(ayMZCaA<)Y?pbGk$C!TYs(MjLN5zMc(I3%He0 zP*P1am)O>CHu!44xI&4{jKEzF|<(V{9W4ZtvK*1-^1`P(hP z)>w}+m#F6jiKHeBB2uIhOh#}-`LLc$Zs2~=Lf`YO)b5JEtn9iEA8yvYE;86m62{NQ zg*y8_SMA6M@Ctcuc-7h8e>nzgbvEV#X<_10b4BfuaN%Rv%^j~UqmvV#k_Mu01*I(t$o z%HDO3pu$*9ED|h8*Dk87JHUc5q{N9MS z8Cc%H=S7V*zi57F7Mwh|uaANw`x zhr1L!%&K?u_OMKGfP>m;uaU-y>iL>&qGy)%+V&agJEs>|a;DOq+^JSOYdZ9z>$l^a zyQ%V-zQJlsobgBRSN-~W^it!UeY`_H-QI+zIcLqt#$?AuW%0|;uf8t58RwGe=_lW{ z%5cAcAyMD!i)$`XhOkZssPwoO-qU&4=BeysoL9@`l@l zc9g5cHw!}E;c|5dO^Qc;jlXTQs$vWC4!4xwI3E_ARLP$*jolgu0jj zZGxe*6S&QN&|zZ@iV-zL>*Fj)j&S@rl0847e!>aKVP(whd4yg(hJf?ZAyUK6#Vf?_ z0T=k?U3q!6vV;=;i?fM~5p>LS!pCPnW+o#{S&}6p=xUmJLh|eq=c(-H%n8l!f4D3q z>`*rc-jJ}+6YTsC!wLR(0tN^Luh;?`5eg*8q}3s$UhPtwnc>AGlSaXXaOhNc4`L<9 zrRFHOYTSZ&G`SjtHd(Do4AN& zVHjjacrm&j2`h^jHW034a|sb|Rn8gmxXn4hO6mTfa1-~Ru4diQvy`~QD@GfZ+wjhu64 zbI4%|<$O8K*+xj^oRRY>L_?&7oN1vs%nVavp;9F0L(E8$+PeJZ` zS=S(k1SQgxJb#>zP=c19y@$PP6I0#a%WK#EX`tq`*PXc%fwJ2kspXS?>imu*`e0ex z;#4)n&uXB@=Gstwiq_N}DRw8mDsg^{%UXn`MfTq-qFYVm-_&5VNLZPPG+f)4 zqeHEUz#{N3^BY{_Wq^uPJpvqcOC9&Kfc7QnfAM&U+DZb{Or(eV;0cIa`E$}4$Ol{3 zhb><%ZYyWJ7cbKPQ{eIYj%DlFo6ReO?;_+@s#_;QcB`!zN7ZL`&qm4lCcL_LZOG4Y zE?^4r#`n;i{ne;TJ5k#{*z4`#zdC|~?8bwmR`5oevxe_TcCPW1vvE~!zxnY>7!`)tMQE$2p)0$u9qpO+;958qPs7=xew#vK~Fl=jm2 zoM4WbaUbShG$iXJbi~U0eeZ;ks1X~on^^^+QBy&M{5xl8anRH=wL{N%%7c7kGy$`Z zUpZl$FbI;7g*h|OfgFKv=!_zB-H!QQbHrW|J?j8F5wd9$#T}2E$NCAAxrVJ}B2bSx z(>l!DvD?fXVb5$OHjc1vx?V*Qr)j&MliXWU@)1#t$I8Cm0h@J#a5%5biU`5*f_{A_ zPtQ+ZiOuR6%UOH<%!LwOEI`nJ(d_;_6b&IyG&`;oD-h&H)NIXpmC#T%Qy%sHx@Na6 z17Afrlge2xhUlarmdfsR33}g9WqKxPoABi1h!s*i6&t?)h?f3{+RK?GzI^l}H0QSx z+iye+;u=u?*(w5eWDU#_))*`1ZkNL4Eu z_|6+8W`?FRY;zhSMuO0BayUCaCRgMA7t~B{x>^}e8J4qHsUhO<=Wxeqg%uKZcHv=Y zRHoLYkjV=+6BMCLbN&(RV2}6V>p>bePYm)~Ar1*qu9zqqB!VNdd>SErP@H``Y98a6bdn;gG26QDEaG3Wrf3YGF^l znkjiR!A_5jXX4V#k?P@FMuP5d-7E55i4(`2f6*Et5_ng0E%^B* ztXa1*@7uZ$g#IwH^TC@DUg70NBh~_R_WCgBEXTQ9!assIr!<2FUKzSG#J`$#>Pqff zWxnEw{AAtQak1wL`PXKu3v)wpZFWLLQDv_1xc-JZY@3S)L3b*h_5l;+rEDeP+iyoK zWR`G)ZKHw6qjWZn#>LrbI2--7^;3!KbzdAkD&RT`m1)0jm|k28;l&74?NE;}BJf9q z`#(7^Up-+jZ#84^e!dwYCT=B}#q`e?A*&do z{Y#tL#iM9K(8(-xx`}2y?&g8htI))YWoBV9dhsa(=|MvTD9+V4i?V zAp#zr^y;xf5d`y%`BpnT(=mmUA4Z2D%_94kaQxXR4bWT?F$tnm-D!z)&Ro=X5_(aZ z%?P>|ceFUF+D)K^k}~bqw2-ON4NzqwlyT&M{xwi4xj8~y|W~PHx zGWr5mtj}R=K$EEc58^X{9WKp8kBnaxZ4#Z^3`aqN(aXBR?`M+`=kcFi&1sjCkL(&| zB}($EIAEM+*V5D)9#htG{&D)YJTfxAny`Iz?Uc(qxjpaZog0}&752__==ovq%7*N; znoeY-RVgl5$q~n%M%Z+o^B!TCX|0ydhoOcYu8D64K%0zz*$>dZ-J+-hm#z){@%s&r zt+;J7ztgEAg4flzLppRvr1UtJjVavC`w#3P%b<1)jfIz zNHW(K%pp<^0hyL1bu{373-jM~TR9TOljBB>008C*!Ta5^c+2bSY^9ZXp^uFY-NU#W zGmDZr69Ywp3)9cgt<7LactG~lI#wAjBBZ>!)%_`DHAQyt7yatK)6v__%2jSPgiV&>q^d72`y~s-(ZXc5u?~U9#=RbqD z#}p%aeA0djZr%APP#P=Ah(P{&pSo>W(MYf4?{Y}0BnQjPS68B|hpJZgmdg%wFI zA`;6uE1MpJsP%redg>qFD_q5^2gBuJux#b%#?Qwbb0` zV;6l{=qhS+pyk~z5A+YJ4f?AtyC}@p9yHjUphq^%({rj_oGXa5&Wzf$fgu(rh(=uE zOc3Y6qdUKl37gBTi>cf8P|=lJ)u%yK@+$8sQcNx zX-*xso9a!@(JHsfWl(<4QpJ~&1T_Ojq=lV=%#ZWrr>t4lynR#tPn|)_fLvCu za~FN6SsUHPX zqbLS*zMii`3HEaOP>(%(bPc=ffO1y|<6eV%Tx_Tee?exhTUtcorb2`15m_C>G!Zz_ zh`l-yZ+4wt|8#C?qGVmc+ZbUV)X6vC%Onq)P?D(=3SJHf) z)t2@5`xbUB@v9WnP7jR3doa?*UgLnvQVjZ+PiDADKbLeo(eqT8hi~)_jq|b;?qU;S zCwT68{t?w^$b^Vie}yHxIB*N6`&zje8P!gF!ZBq-bU_fF>z&g~0Z$D`6cv-aAXQ#4 zUBG5Rv1++U5`EOnHd+`zc;&WUHV5dmUj{B;bTN0U+|ZkPn>Jl8y?No(t!}-DN7KEI z&1fY+yMMwo*(cHx?f%ji{<<0vk%V@@3CrOS28}1b2NtqERg*pwl%G+Y4Geg4gV`N_ ze5Wrx=S-Z*@Ov@|Nwvd<`uxI5QrFLCABE~QZo7+`^O@SRgqjn}na3ILBu`}D{Nz&k z`KP_&jJEFS@27tl{i^??ebt^B9QC$W@U3#<`q?7GtN>5}ccw2%=xI!^XQ4BTvV#ms z8c(Hv85F#EmAx_iK%hOYj}rLADD~E0(D8bM_uoM9Z`is-m=m8MYBr$rYd! zI>2mw10A<=43s@4PyPG7)1Jb^iDxcxjNljUtN6ZFqA%+@1VF>Bn7}{I*%Lo`KzPZi zq2zt7aGM08v3(3N}oQ4v=(YIe{*k%eU1q&emM~9gh^Yk>A5HIB3HMT{HkMuZYOLkKM32{#_otJR&%p+ncf%SAIU= z%haiFtCnS!3^N{;4_u3tdJq$S=EV(Eh{Oj{TzrG#&u+xEkcE822H+)ou@y*%QVatm;wXw}$Sf*K z1fA9XXom_GN%jPL4BqA&t;u`_LZC9FGvaYS1C)&fMFnL!B5@y8yu1dvS<22U7m98| zbAx>lv1vwz>h*N0xt%$l9Q-7!<1$^fpdqFA;z|{4ged*Wyh2-VlL{(1*~{AL(uZVaDUuF5RRt>kXr7~xNN4#6F+-j!txs~- zZ8gG~plT1;2ZufN7`%(N&EKK_3}5MOESWJ6vG^1LRDk0_zC)jr z{HSu)lch<5459RwS+mZJaeBpyldXZ{^lG8^`6QozabE@w>Ha+3a254B+y{O@Xz`P? z?~^_*0=J5g?AgsiVdD6qcv80Yo7JD&b^K;EryKg@SZZ@EohgaTi&xUVzEY^aWcCMo zmtw2TG8XxLoi5F}pvOO$DiCdO(f%VbqONZuf75yZ%|nrM%UXmTTtk}QRtcl`q(|dejOz-PxzUk=0c%jw^n6=$(KVCJ^&=hMv}N7s>XDQAD5evFn|QidE3ww!!$G{iC^sQ|xSKqp zWTcf%rAzfgzK$T@X6rGbaXQqL$j;hseg``(;*}7I2r*f7$!)nyt`gnC=+AZwT77{5 z&hH*~U?BR+nx>(`rzeCx#X$mxaLq_w+irEI>$Z_j7fqvlXS#2YL*D{D##L2Uo?i8^ ztl?u=(tB-<{(&0XV|R|(h(|YsBb;5)J@Qde>VtB&oh(KwWvhp+h(Cd}@VUVz^BKX1Tqx>TGf{W3FtNI% z=*qMCQ2#*B>j%<*nV?-9 zvG{GcEwuv@5Ec)T9gHy%9VoL)V;TXKW&1>5KvaR*nrVK@yvMM?XP$E4g#4cDx)4A2 zFz2*MQ)rdOY{+i?UR1l7t(>fRQ8o(;>BVOWqxal7&EtR>pe)hdQ@Zkh8Yg#P{DUq4 z^WOuOcrd+~0hm0uQvOF#4*u`>WVRhj1>ROl%+d^O9K@_J0c3W>2PUYh$GU{I4|dPu z3g!dYT4m{iUOC$TI1nMBfnl3g7!HU{8LuOlcCFk~G{u!9{ThSSQu+Wh5lE%L6@Uq2 zXD$M6W<{iR3!g1A3s>R2T~iwnnVS zwNO@i=C~zic_DihDzN9Dg%aLo!^bjoXKT~%*E`Pc2b4Mqw>X-y{iVt713dyHFB3gn z3o`*Lk{joJK+7a%;MXSF`gH4ke)(4ErEiUe?|BT8IE;qjL&C5*f*LB1+$zcXklih4 zf92c1$G^Tv$e#4qbut?S?9FH(&D*6^!+%{m3sG|ZJmV5+8(p%P05xXr>{ ziMn-Vt3q#G+wNn_=|ALUs0X=NP#3EUxcVhc>&eqj zl6JNT0)_)XZ7CrVwnFzFT3qNYQ~#m07ASb?;)QS( zab78!n2xwjdN-E1Osg!Ex9#uz8CyTtqZfg?wue8=7|CZZ_`RerdJ`(I>XN>7?$_;t zZcwt{t-JI{9Rd6SFr9*|TN53S$yu?IYt&oQlfN}DoU~7SoSf=-PC<>oirYWqjdzjc z+J6uVB=V2K_%E91hDOPfq56*YP+do}pWX5(X75bHxwG<ka>10J0)rzM!OuXk4)pfeS6@MpUkju2}vn^MGNVDWn&s1 zHGA{}rI{x ztdb?PEqwgULy!tO?ZK`@&#uK4O|895X~6wwfMD3Urrp?;jTO0OCkCgS{9c zeq3>ea7?Z)HQym5ZaP@l{qS2#@JnW6CA$YV#MeRPN;yNCA`7_Gs5y&AZ#(Msy1v;# zcj-4cmt?2z$EB~OPBURuaG89G^_++Q7Gy6}*Qj%r?q~AuyK!BUA(+7L5;LzaqCi7>9T#f&Vvu` zVmv*nYklq*D&xB06=XIN-j)*-VY_NFDA*u0qX6{okm}IK7^Toh0c;!_czitzC@TO- zyc%W+v~BxSSI~7I{wsQ?vhln7sRclL)!!m(H2E>kKG=GsFdBN850(EV(%t`bCUN+# z%m(4u9MR)2J|=!lx(si!H*tfFWETKYlt5Om06q^A3~eF%0pJG!@Q?#V~5wA!F;2c5F1nfM`S1vgFvZ zwlZS_>2qivu<1C$EFDIpaK; z#Cp~G-aoeaL6w1kUL{=4#Bzj7nlb ztIV`T&c!PGV$=1ACfwlp!Vum?nGV1)+fGMYS5WL80(x8T^Ya#!(M{ZLU9+NlQ;o!! z;Kv$*s&^)?@`Qe!??a!N5lW@K!(?wm zcWZ%k@@KYi_Cii}!!yByh8-Hh{dL`?wKCP*a78*q3x74xqb`9Y@wIcIip(CP7Qvdu z)&h!uQ88kFg6$2yA4--ImC@D)r3+#`sz*PM>b{&}baNsWQ)+aJuUt}( z9MLFrEM#fk_e>xrI^^{x>Y0;XY8k}^Bmh%a_qW|q?O9K~Ci&T&?%()g$Ci_b?{wfe zHTm=2dXMA8%fbe0k<*s91!*yN$e|%Yg75lSA67otb^ptRRW@8bw0<|V<#2D+XvW12 zePz$65v7XyvrWMRA+lwVwO@{W?m38T#!O2$~Lw~xf{4VP1v8cDT z)BYu0#=WlC=U-oau30T>k&?=@g<#<5W&MnW$|1I+;*z`4@YxbWZ{BCuWzEju3;{J9&&ZoFTE5NDU zx$5aFzG=kF%+-XNPh>jRPGNJF?d=1oTXz8Yfev>1+?dsba(+>`(x7XTq-E%hC*DZXTh`xqhX6+Pn3o`v7aPJG~+!DOX2j{K#3sftk_Z zPpoeWk0_T2-OS`$2vLyJ%BBglFxh6-wlXDxk}{_p@(>Gj0f#4UOMC&)uF30&jrw#` ztXygb#2|;c3aW0)zOvNO8n73UXFobQ-F<8%Sa4;fy5`nXJPS-^qk%06@IWDR?nCnF zB&4l8>@Vd;%MOm1#S%JK2$BkZ;?Bs3?T^~NJ2iqJo4nEp-obIB`ZCvYTy1ww7>g@z zUu+4w^GpXe?|cA<023r|yNJjOQME*AI30I~r1i?7$L$M}w(o$6H}FK;r^;gZfQbeC zb2pDpUEUk?#VhAdCA%OurpbVHFP7T%otC^qA-(n!%YjUu6Il#y)c9n1P$tzDTW+Sa zQ^Vo%BgA;O-ff}IE?iHvg@4ErzLH4_2~Z#$jYGVb(#My@M>;il^0ZQXx_Qd*5y8kL~McJlnJ9leW?m`2+8e zDq^XJ=c4ZoDDA8Tu>IEI9&LFxupMG-Wl_879@M`8N?{u{O_5@GUn_>prD3bF(&F?= zKk)gCl8BT8=fSgDt#7L=GH$*m%a$`9{lWbMYlY7}5RGbmRmjR}Z0Vdi>v;-SQJE2$ zId!G@PnPrf%~%&QuiJ5Fiy>U~WB*S$o9?M9e8)P}t6C*aTSOaLMDhiZ1G-+DHfpHD zFTU3-G1<)HZt+{n7 zI=(aNl*f}P5s|j8GLIe}kV(zH-d0!r()*`B$NN7=#TPe?gnl|XG(1pqOa_r4Tp+XF zn#g$)tail+E*nQ{@DbF&Enm1@Jk_{R@4;gUf)MikXZhwA=-xCpacfk<%Dpi*fNqQv zA$DabT)r%?O^-17K&RA4(|=tOKayHC4{W)bMm$Z0K4pXp(`cCMFmY{-+YWQN$6tRO z>o$C5z9;$=j_@$%9X{{Oj`pJ_c)CEeT{0fi!hoAG9f6>&B$+6 z=NPDWL-s_!)(>55X}_#+p^X2kz5C9?XybO^KKVtV^wk$tmcWYTqw?iIUjVEaXUyUx z!M=pDab5dCVMVcKk!HJi+6XtyFWGm?}2` z9X4l*34d?UcbPp$EuzZmZ~SbYJ;50ROt9yRB-#+r;u?30i`N`n^QT%_jq1+|y?c8U z`fNmB!-lgw#82n}!J{KFuxX98#|aD|fU;h|+6;nb9YsEc3vapbYP53_)TKgE;U9u3 zvQo)jvo0LZFSUizgYes-t$pDp0uh;Du=0Wpd?1a|BXkkFp4}2_=su5#Ck`6J8Y50D z*>{}bOu9;0`HeGIe52VN;Ia|iIo-95cWv*!d|(dG?P@r<)<3J!UoR=y_9E*5AZQDW zkJ;&>`*Ix)+hUK)4|%J{>`jOy0WiZ8y&s>cVI=Utzrwf2tq0v0z=nv>(N>b$seP@1 z;ZuU{%by-ubbs~PM*B6s4Jnw7Krh-6ic1jHxR|rl!QyF5Wqaf!~qX_ctNR68)|k6LALl4 z7UlLBcNsLd|4*8?{~d^mq3>yK|D^J0tvjxi@Ghqz%f}pcab69E;6PYD#YQ3s$hS&$=c2qw_JT zm6Bn5tZgNM|9R2>FQE84#YG+HJ54OFEb$9~5D4Kenu)f~ncJw*0O*@rG|JawKYMQj zCog{R11pZnzfc3j?j-GY537L>Z#SOA#tQL2^k_MErx7&?=i%}y)URdpWr-RWc$(>) zAHACW!2e<1+b3_lV|J0)u2}I!k(usVG_>N|>;E9D71col)1HIx%Fc|QVjeL%G;6IE z;p_A7Q62UD`DNIph){EEwKy z?}MX{%J-ZdQAc{XSzVfozNGH#N|uwn`w|r<+KZiVqnn*+JW^Xhs2FuWPYPJ8hRJ{rk3B_!=*&9 zVCzA^h4K4Au2GB35>6!wWgq+i?R*>1Z;Cq-HH?mMx`?^@V%@mrv-&SfO+aaaOh0Uw8VGVW&(<=8+8#v{brDFB<*aCde=?paZSFAzc*3*} zF-z~-O?oPc`(zi)Bg2iv;TciQKJiSMq!qg}l8p5EMW{J{;4ZKE!0`{|LXLR`9^bTA zUO5bf%H{UA`ROYol{G~*led@mL?@j@IPD$XBGu&k*8+EJM738bmN&VdsT~;Dt(8=z z7w8y{Qe0l1FZky0*21m;$`cW?_-~p`n=I*KVY|gWx;acX>pw^YqLh;wIg zp)|hZU%Tkgyd|M;KFy_YBB$3!PRfD=j%)g;$IyK2DS@&MRJ~5@^N7SWDOEE9C+f;X zL)7EP+5h|p=5sd9>aA%sWGXk+PpO$tNxtwYPN8Lf-0#53V4LIAqy9_5^kRWt ztqn81%ilXUx6YzmW*IOs>j#XWlmC)mY%&VkI+{ z2=v#2Mu=-)EQgQg0mF5d+@@oIy&(4rYCgGKr}I9~fA+WjhO25yM0ftCi#Z(P05jom z;6Ss|0V*N178h&IP^o^)&!hz+2cH@+6iEN7R5M{V1dG#5h((eTf_j&(K}6!of1>sv zn0=Map@Lp&u4RBoF3p0`4V@{*6knofb=8YrOCZZd3l;+%cK_>Me)P4dREDayr5}4N zV#5;wituwVl4DcMR9|-txqM4@xgPxr7K(=6s(q0sovXK)>+!q#7RQ?JXyiGLpU3yw zl2sD8iOg};j5tK7;x^el(_tX;dcL$iYLAc_2;tCh{8o?TDOtCE)-*q+Vf#Pf(!MDl z@KR#~uJI* z%sa62t{UiwRm~Ssfta8uG?eWN@d2u-{@@t)|4iUj<$(#!t-zE7^xf49MRqT6EWVb` zDnVB6z_JIAPgVS6XJ=+A$xX&;;u1mNSH133(9yx=c|(en+A~y;=9;(AH(ff{id8-e zOexQx7H0^@&l+s{UiwCh#O$ ztlk(~OIk+<4PZ;$5G1c2kl4-;y-GHGY;yc~A?Ypn52HkB&#abk9lh+#OL@>O)y3=cMS@dEk!fC0N}4 zF#77#G5>1DW4H8Ik{(rcqDF2#h>qH7h>}-<>`bYWzLLPr=-h(>zZA9UFnF?XWUkxy z19!Fcjq63RpC%S};RtA06b0;!jv!i`#=xe`3?eVGt5frYSZ7rW$%|QC#y}Eg@1OD! zBb7{T$$av8EdF(-p^v+Ao;!g*=T@MPmJpY3;Xk$Fl6+XExiML`t=mw70G8Mx6ozF2G&!g&|Q6Kt4v>2k5vPmYOe z*yWSM)IAr3j*6y1Ve0tp;#O3tI5Fw7dOXn^`R1$mAH)JiS{j6RviV(tzC(YTNpjNB z7TxB=&;OdPexY9Bv2{8DTgfe)V~2kqlbb2dYI>6%R$=9{?&#&}CHb#DtxM!zNYgj^ zV?+KS&GcyHbAshJf&;6TdFB}5HLo#>vKK3Cqat2At|?dE|J*})d_8JlNBc=`9gh-h z-L5yt&sK)zBqsR8@{Y5~9}uzG5PR@}-}?B^TNL}bYIcEi5y46`;gy9&`vUnc&9jp1 z(@Tj##%}UPgR#o7RBOpGBeAh;+!St#IsIqvMYHtJ&qw;+uUm4TL@s++R&6MgTk=O3 zS-;IZJwRDTJrdUtj@HLbW#2sEnRnuJl>;Yj9_KWC=%TE)cynG%uO{B>q2fwXOXm$2 z>-5#Mo%*KdP07hG>H0XOALd}K`b#!Mf3fRfci>Roc^8scmzKX(>hh$%T{xcDZPTYA z(NzsM(nT=DRUZjzO9B0VRX7&79`(kLX)MAkC_N?-z`aLYeKpmH+YC+2Bbh;rHAAwB zS!eX6W`^~ZDPFE;nm&|utREoFTzVQ@`CYlkqk)&tyLsS%eo#`$bHI_|sIJoHnP{*o zl8Sd=&wkHo1#yY`?qu^uW_{DhO>KHr`V@D&;3IQRIGQVEW?E+HR>>~9H7A`gAAx#Z zbDzU|DHEsywp8WOwPXcr?5l(&Y1d|8B7cac5MEp6k<$`(f`Ap z4_KN3ir*Z5kv($i4rms8G?*m(+V3pg903xf;y;KguO+hW_qw1u;Odc&%#`<8m|PVYE)vdZU9vNXW`);8s$F-E|3dI zc4{fre_c4iGsy8R1%&?09ok{>-Gy>D*?ArkB+x{os2Xqbxsvp8gf(H%0-@ZDCObJ1 zA$V~njjfD`7awD<=s@NfH6A!Gnj?<+-{ z!e6l@$tt)Fdg>Eq(4y)0j|v5CJbqlilp3_t6O=z{;Xd-HIf(YunDOihlJFM!WyPXA zo#CW8o7>|&@xhysleav(Ti-ZR8igP@IddM(cPvr>t=$2zeJ8W;!#n%K&BM6>UgBCH zXf1s_HQ8^smR!fo)TM3NdG~cdY>8k70g<9A3-kzA667v8PLbQ1(?9qnTUl60;1^{n zr`7cXV5%VsEq4YUA5;81w0WkB-BCn2iSWOx zAfg2;@f>)jRVO>5uaz}r4PV7LTim&Ka2;ETQaj`TI}B;N_~Xtyh80ocUB@W|WE|iX zHV&M=iQ&4+wJueQ_{pIoyNS$$Mq*={cq;B5(1o~peP z_Maf1KDmvY^SCWr{5E6na3XE|>DO7ukj$SktL2PlpNf&oiEz5VaS1WfrJ|(>TWEUA zeKYv9Rw$>aeOW!mS|{&yij=T;xC{BU2UV5&Z^9Oq79(w;@aWDt8Rjr=Zh9yd7q$92 zqKY*?^$oSiPPCGS!>E!hveUSW1`=F7JfQO+bFdS$ExUP#x)5I->?4;mj}P}TAd!1- zYx~O$-;Nq)#O{q($+PmBho8E7Wj~8=p`=@YywTE|j$?-&_pb$oP5D!gy$yc@JpM^^N?yzg1xr3-ke`#@ z&|zmF*hk0rploz>0`DaVt-1P;gtH@%wht@BZ+H3-H(AT3d<$(u%hE8ux*#gEk)!8Y zaA}Tu0NWP(7ZUyw&r0n`3{y8-?b#gJ_ti-g4MyBmOo~`c>5qKhJ=je3!=5ppz2Qz2~shSBIFHC7e{ea9IT-lDPcJhU{&)`nIO+^81El9Z8?fhXqQT9+^ z;MpsUW`dXaaOVEL8GK0tAt+P{T+#<^HSaHgpHn~im!1U|A!j+h#nEy1lXo=z>~(*?H#--1zG0XQspd@_6Lck}wb zv2inqm1RI>GGWtP7vvM`o>8db5{v|cM6toQdRRU1>%ad+2^`fOzi9?ijKP7fbB6)_ zBe<(RnvMMR(3U~U9Ue+LS0&NNPLDYb;-WyE;N}2l6FQf7qzgO_$X5Z_3pf({=l$37 z;4sM+;Q(7S`WFd##8FEs|(SNMiNj&TPHYuQixe$@bWmlFQpVIH&!CiL)#GyUD z`49ain!)?l6!M1bl!!dX#=G7dpUrYK>XV+;vJ)RSlM*T-A_@CuX6VUNEZ&gU;O+MB z>^Tc#gqj06A$SqlBmg9@VSxdIO16@2@UjRh-10wn02pt5_COCYSd#mH;uCmZCHyJB zWI?%Td$gwMWc+XN9*IVdZVxTPT8?2_LoX4*DGP;zWF@i}c*mi<@(3)XPy2S=KNQp0 L5rl*6f3N>Pc%Qi^ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-10-SCIENCE AI.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-10-SCIENCE AI.jpg new file mode 100644 index 0000000000000000000000000000000000000000..144d74cde230ad92e8915315a3f90abeeb17d95e GIT binary patch literal 3979 zcmbW3c|6qL_s8FpWe80)_AOgui!hTdTZK>*Kp+4B zF$O@N1}*?>tgK*G7B(;#%+AhsloQIu$-%)X$a4Y$6%`T}6BQB>IVr6KJ1MCsB_blL zE~j`#1rCRc!!&g?RL?1?!Bzhr0%B)p=j7lN;NlWcl@O6o{XZMM1K?%@Jb-j2kR)(~ z8^pv7qQ3{k8F{jR{xRUc40MEvnS~Y1#(tE85zufPI09l~I>OAv!otkVh{iD10cLI% z9tjl#))Po)uw)=qH74aHo7DN*PF|}o+tO+-L9y&d`S=9{g=9{_WaZ@Hh_mV%npzhw z8X6g2GBLesZDWhFyLR2x&Ha|gZBMV@5cHkUu<(d`_aDT?Cp=6{eflizdHRct%!0zA z;*!#`@>g|l>KhvIP0eq+y5E23>HXN(KQcN-7$;6lPLURumRG)hTU{ele(db-?f<0y zI{3>40+{}d#hCvM_8(l_46Y;0%uLMSzg(arp$uo@W@eF4VdXJEf}I0TNUFxLLC>eW ztnFl%QnT9TbqV@%lusH?lA-)X`-kj*2NwJPBKvQ!|KXYfIGI3<<}q;tdcZGQqHI3b z=83NBxoz&j;obs#`&5WjNZ%%^opgixa7oqBS0qtK#uBSYG+Vg%92@5-ZMegt3HG)l9T-~a{qyJV))$8e%5hAb_L*M%v?Lu^d=*B6hSfPWd-`<~{0jBP zn<0&}e6#x*Z`QvGr6C-Q&YX9emDPQEhE@-nf^`n@HijDBezE?RtVb!uc4bqPgum6z zw1-EJWQbPj@6ry;q%8wXnlJqEPV_}t20wq-thY~S@2Oh*{)+tb%ftsP zEJNXDI9?c~XBDU~DuU(MyJnWKRrS#Pn7Lk#HjUS))#jSVYx!>ZD`T!w6huz-?K#=) z4N0^0IBu0GA%)@qi$mtjLca7Gi<`>4B|pIDM{m-Y;GmJx&Uv0D7opyx0rP7}3LQwc zsoh3hpDNzdkyBd_$vd;FQ6yG7HoGRzQ>#Tn^ zJ1F>u;8@Ma&bDSvI#5zYgM*0Ju5t?BhTm&C&@s_YeN`*oIe((R`KpU?!wi;@{=9Ds z9mp$h|4u5cE{-bXhz!zA#Z!mDZghY>%2=>lHy6^j>~v^jn@);PC=}#yo8Ou8 zI}+Ar37cG=p1ip9>sIsFI6|Y7y==BjM?a&zC+kl|**87^w?j4A^Go`oJcn8oyMfq_ z60O=a)WO(3_E?ljc(_3N#z&A~xZlzOcHU;-Y)yIHrmxXP@&z?4&InI*C#qjv(%Tsx z2A4;9CkflNYP$+cA==7qWrHBIYrl|o?L40l$#GdHU)Bnn2U#%vRmB)X6%qb1$IZ5f zdM-81ukdx_w|~5?1n$R*SnWT0BQDpCba1R# z5H;^%3igoJdhl9M*@u*H2uAeRrMXnHNtHTF| zNXY$h$+-XCSmq4HFRviXNH9eBd}dtH{Q6ek$47tCdc|q408&{2K1wvd{oy&Jdr9?q z?XN~Oa#G1%#kn=UDT{lXO4X`pE(84gc&dvi9S}+GZ_eD%mGrhfNPfJ+XJCkXetMG- z;p?d2h-IIo4jF0n#!FjlvJw9%SS+aMzS5>Kss*k1NnAI7E_gDnAoFaJp@w^4EMz1z z4-;JvnvXKL`sW_$IZOi|xcj>mX%ppWM-4VnG+`-HcsjjOa986MaeM1ta1A+ds^bh~ zF8GUad1X>b8@tltHZf+g{yubUPdR4>eTfbn{3^@x^lTb|S2y6)y7s#-e>pMKY%Y_X zmmXTv7KBn)4;`t;qj*d=g3Q!+%XfWw=m6@DnL_BZvQOMJL?Jj{IcLH=b_z~f&RSvG zS+{C-B_hfFnR*WzxWcoq88peRBm>b1T*canwb8b=F5eW0WZTs)n$&13qVz@&PUby# zUM{AICHbbF*4IPSsk|C;w95=8^^U$P4`lYEw#fb=+12XCCL({vCJACbj^>hFx%K94 z>e;KKnqaaP$|k5$O#MwNc}{i{-IN}Z0^`uwV;Zl`>&i^eww!JeaTDlSaVE1Z*3p5H zdYr*Evk&Jyq-rYyt;V$_8#h_~j^a*I2ABxs{j3zy#1G_>iQQV|Pih2U6!L@E&Z*Imk zLh@Ma*qkl}8bJpV<~Lk#WqF-@bg9?SV3>B!60B(AVxkEvM}%G)SKie*uf3^DTF6y< zV!d~8KoFYV#>Z{7T2valv+_vnTu}?CJxMmnhZtsZ!4^y7s}agF@WJRo^(!ZZowG3( z05L^XU`6(lw1}HX;;y(*SdDqZ=+&-w0-%spjm%|E_=d)3jKUEA_v9)d0`h19)^Q#B z+dNbQww@^OFQ6g@9^tjf{n@kWm zRfjb)dw;i5-bf`~Yo@7@kVrgWEh1;S7gypqrGm8{KNKeF<}F$04=bZFhz-XR(EY); za_+wj9M^|P4(1P+92B~q*LnW9F&HTyaP8SPhSkvd-RpNsg!fMzKDoJQFFz8O5O<=$ zn@gw7m+F*PGiTFd>t7Q?E$-aXFjr;OYnkW$rVAVAcOX_PkEjS~lJt9dE0*jdvv8)` z35%{?pAu44p1e3foiYG3-%RRhC}yGTx%IW-s`>&+irg8ezX|9L=51G+t|}HH<{ZJ`IPy&#? z)T?bBQW*g#l+uCw^;2Qix8FK6HS9FEWMh6|-b|p^ZhU|OR;BY!RWe%2d*!t4e8>gL zlcCu9#y(Q*<4B}c*m{VgKx4e~kAX?sULT~sm|rE55N{dmP?MOv|2;lrHbO6B$-qB2 zDL0?##>Vb5KUYYYAfXSMhqoyx$lBU^&{ zE3$tJ4G|76_%33CH&t$!xWj$U-R)hzse|DPp8A?F<H#M-)>QLxtc229f?M+HZbDE@ZYE;=3IsMnX)4 zv@>osTj}m_mg#m7b*g9W$y2+db_R8F!fg5I%FH$R6_w5N{jBNVglgN}GGa z__Hn^PO9ap?9le$E$VYF(IsWZL3vym6Uv9u5gIUBF8`s|)Zv+WW65tHrq0WX+FmB+ zAhUn{2smvyCW+0S<}%hoZsi$-Hs#dhhYV+ z`;vESYbXB#xw2cq@xJLbA=z-{OPBC*}%j;6N z@9$MruU*W6#O|?M3{XnPz9uT<+7vxnHq6_Yg>`R9x2073ffWR<;t?qo-fz6jZH&_( hKf%~4sO>I+s5TR~`SPhD)VAmW6sgbhA1+Rx{1=a>hWP*h literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-11-Acronyms in AI and Machine Learning.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-11-Acronyms in AI and Machine Learning.jpg new file mode 100644 index 0000000000000000000000000000000000000000..0ea360b1e9f8c0ec0fdc7b93c5af5a415f9fd9a6 GIT binary patch literal 17685 zcmbTdby!qk*FHLQgCJdlf+8&?&4>sHh?I0mH%N!T(CAPi0s_+AAl=;!Lx*&C4mF&O z@ArPcbDh7=Im5NNVD`TDUhBEnTF;7m-%sB!fgZ_8%SwY#P(UCQ;0JU+2a*I~V`5@q zVqjxoVd3ClNO}CC(y$rdl(mwnj2D?+ z&kZ8-+*>;0dudtpC2xlfmoQC3P(1!Tb1imNZj(&PUaTZ|)j0gkio_f-@w=x-iyk%g z-Q)M5BKcqJhRS+RyO-$&6mq|;f|n(80wulv?4Wzg+=Cv=2z=Mv*ecF4Kk}6RIA_2M zb?<>75{#+?jcoRse|nOm-GfXs5_{O4_%yWcGM4d)5HWfV_aFrsgzAU1Wm+&IX5k+6 zcN9rloARlwZ&XjH`6&0MPH*&j0MeFtm$e+cv`v94_O|_&Qlyi#JQOPZ0@!`JrppJ2cdr)u&yA>vVMxPH7>{f(3nu9qJ_zMk7%xKrNhcF$ZFFOR(dLrUBm{| zT|e!gW(lHS6KPNCJ#JHyU7xr9@y^qXY4tSE>7A?KEL&+^>5)Q1J%w^(bl5rt*31_P6%{-K&qW0Fmn@h@N(0a+=M`!mqMeQE6t5k}3&VEqa{QcS(JQe|EDC-&3 z`zDS#*V;UMb&)@mOyq(2tnuw(x8hg|fXn(;acb+*;53D2u$b#T=sd4o@hC7fW}-M6 z&S!9ErHhm>j!Ca49TjC^^{<;Nspo{qX^e=neyc~4pnCh8ML3fKQs#Iwx3kI<*uBXp z4O|=FJ1fD{qaY~kC&$8A?Xo;geZY*(k{%rX`&ESw*tt+TuQB3lD%pfyV?%%p4<034 zJ)t#|*RaG?d3QZpoupmTDRST3zzHSCL9-*huJS0nau5Gnzld%;|A3_&>uQR%PZ?I) ziiNI}ncQmlBE8r$!F003GdA{Ho%Gr@%8q6ACzeq?6~q_Gpk$gxA8wu_ubPw8JEy_d z*Uy6m4IVuizT<|hst$!1|4y%al4|+RiR&xF9CE-UWm6$eZsaE@VMF$CjTR~_^j&jv zt7yIK=*$kiuCTbV?~S#T?JY&QMO}lD47)V;-HQ!8+90y2Zo{RPT5q|dl-Gb=*j(rn zC%-H;n3rU}jCX{vt8AX&(Z-OC($Po?;gxWHd5qJ2n)@L;nD<@eD-iUcH{m-}TT!I-g|7$2(8ACX7lndb^;rCC^Zq27Y zLvA><$15OsRwtofyj9%IA#OvISaVsNU+nz5X{4;IubE~sb1xigya@VM5q^brnFqyz zq^ip>*^PJUDnpyN8MVceTp^Dov7=QdIBa^UxNE7vMJ%#14w*CUqk8Y_Fu$P(3rs3S z-OBN>WXK(@5V{Ebx||C{Iy*6x4Jhl$TQ-rcBA=J19n=dLACn$ux@(5rgKTB*K|WiX z62{T#bt!vLd~>8^WaRalmBF4pE=oZ0JH-lTr_={KmfZFsSAwoMabLf+s;Hu;15ut^ zMwSJR0&9pt)<{A&9ArOrSj~pM6q*^Zwn;QZDAQN7Z+j zrq>?M1)1(jr36L&fY7=Bgsyj3j>K}yNI$}a{N+|LOpHdlg~w_af%_+c z+`mWDYPZ{{<*|G`MuZA%dnht9;MXCeArkH%e;1u2bPtOB8VuPkCB6C=$V+a}q?AMl z^&fSZ<1VeF#=RY#D6MO7-*Y73G$9wNW6Qh;1%8ne6P2(>yKn;_ly$byLg()z&@6CO z31#~xtr$y*iFE0iZ<(y-d5u`0Ars8bOl}#MER2}jb=rP<4-(#KV%QPj*~K$X;hcOi z`h$P_N9qr#@cP*~SMD5ZSTyS~ViJh-_|P^jL6ARV7`S*;_SQm}wvkj>ogp3e0+bG$ zD?Qq)MpDCFe#Wb@YpG>0#twpvl%;HT#lg`s;}!kg@g_QmFc_$7p4P~hJ2!8yNdNi* zPqT-JHxT`-w;%x-=RAHbO#pzIt`8kDjNwGGjy8OeB-Q0Owikq4VUB9oT%iMB$mS<5G`sHBfK_G9$#jFbYo z@TYiNTu@w+|9GU2L;Z>oPOF6B`6AsH7dZHCA#Tr-t8JfhgQ_Ue;MAeoVNyl<>35^N z#;3Uzm&7~d2cB4D;&l>JT-aNt#@Fz3u%vihBmsH#=s8>FGflroj3uJ?Ai4G3tPvc* zAfXzo46G?>=jl4ALe)a9;%7gVKmHHyj+|Fyj+iT$idG<7vYcXz@-2;iYmtJ(a8r#B zVTC;09~puUI2nVv(ax@X;BV!kk5{MGL|iDn40F0@f5$(paD5pe-l8PKpFsg0URu9@svD2C~5s!UZfHx zc4}Z=cT3kBY0b_VH|J534NNJz*nwNrodRMbTvhi7p>4IWU;BpTZmp*>STOdr7zw5B zlCc6rfNF}$G^pi|MswMnmg`;b>UDus%DrRt0J_TOYiqz2O4zo~B>yS6%75^{? z?7R_~wcLMaHO(er()Xw2Goi;F8;LJ2Tk6qppbHTx-emb~*-V>h)U-pleUbXnVTj$W zyf;Tul!36IwvE6sN(9#CYbsKeIzBsPcmYc*R&Ei5$U?)1X>$Lp;K|K)1d15N{0OEV z*~px`$_kW9Ab8PKcGB!?Zv;>6=2@Xmzgc-h!c33Zw>}b)6<>z*YK{g0j^V)K^Yx^+ zuH#J7JVGLVLU(M0fJD}sM}rrkIhNIzApWK_qQu7~65|1fgGM7LQqH^xh*H&&3H|eE z)G^D3OkGFM)3cS@9T1PVb<^0vkIUtYj7`}(Ytv~=&%EuFz8dN+QNjIZ$wd}KBs$~} zKk%wDzcZiHnX;5-#Du{O*Yfp1_#d6v*-V$#-8;5W_U;1kP#vA)(mkk6EPW2`3x zOP>_wqK`3~!YQ)6bpO<(OOj`N=xdz>IQWZHjCpSbn#7y#)68_=S7-9yp22P3+o|se z&}c$QuBqXOaC>y|=jK7ER7$+L=jX-t?_-wW`zSKj%oVKl6If?xJ9TqNWzj2z52Hnsvc7r? z3S1Wh2^5hf#VrHUJ0BCohl0KWb$~F)-rspE8*Vni*+TAVBkh5+r6>iCC#yXsJn%MX zU!R!(BA6uM=)` zNsGL@2gT>-cABWuNes$+&K>61Yf&OAaqpJTv#W?_5Oda++U&jrgP0%0T8cUA4#^8R zqccq&;WJWQYcQ%31=@f9F9NRGr~m*G%WQX#Zl}~!w$(Fg(Si+bzd(Y_-Pzs>AHr-o zG=0`-2IHISLVNS|=_XR%P8D%QXWEC*!IH6W!#(w@&8?Tbu|7`NtcCTKENSkGcH?;s z-ukBE)9ibDNgf-`ZqRc){Vi#iks9t!c?&8w??O+ZvZlHRrSwA27!{5ZF^=^U6mKud zMiHScP0!$>TKWC1s}fvX*CAP6?|;ezPfIoDp?R;kowOm0jj|({(*$K(NbO_$s>Q z2i4z9PGKavneVGx3g%lzZQGy`5w3IVC|aKaY-Wes53}pHU+g#z6(-xN2x=f6Y`^r@7a7;yVsv$uq-%vdpDE79df_xR=GeU-N(9xUZTPZRUF{d$&X zlN;?Cw!WJ_>PJsGsY7e4oYM8p27A*Y9hUartbi_Qz~JaR0+(&cP2ZCJ$xkS7p9SmYHR^C9A*KJ%Js*_F-b3)_cw;eEo{G_3&6G*7YS$@^=?+DL7Z?_62Hnkc)M=d51U?eA8)TUaN0&j=e!6S;O{dmc`D6 zPJI!3;79MLWn$0hT5B;MFd{A0GEufBf}iGFl-!aJAIP0qF=AEIdN&OjVn|iN@;WeC;Ao@P-qFpVm8Geg?qcmgdHeU^dNW-X~ zB<9Fk<+mD6^kHH-mWx2=HTcqQXTPl)s}ba}PFB0J|6GiwquRApxUtrrhg|i$a;?Ew z=@yW*JtS@~&)2xidM-o5Z`Z z0~(^q5VVg#E^P1|+F&i;u3VUKUh34sgbXe}sJbiGYj*SK6iy|oshtoQy63#5epNB5lXT(a$59Zu5?*UCPG3Y81M~r9)5AG{)cf zYO!6f_8rqWC zSwSxvX)K;OtBj5CWL%KT1r9*<}qZ)Qs^rTre+{Q2%DL-&}VW6(}hy zH}6V}Dv#-Y8rmL)9L9Lcyu4{Y7P>9aXk+){Pv@*oimM(ks?0lh3=tvfAR7A9z?iSy zOE4UAuB9#}eu-gfe`UhSb^g7fyoOzQFaf490SXv%do>XdKS*8T#Ilh+J1HML&1aKR%bQLjuEAGNrU^# zQskrUx-WkhEyr5z!R$laFwm}7=fr+d1FvH{1v#;-+2Acn(3W^=!m@KC`NmDh%WVZJ z*|2)LcQ-Y=36sYh9kF9uZZFF;vOr-ZJ!QgysX5mV&fgo8BX zm)Gw>-!Q8Xt&jZbo3%Fz3II<$irmdYZvMbpRTT>cs;3wE#CiFEs|XTb8m^ud)=fhNTW(bE14)Ft#? zPV(xZ(MzqS$|(evQdw;!{FXa!pj2n)9lokiON81@?d_mONSi+Uh)X$LAy50$HSc_J z=&qX)CBJloAS0sG(icnA-riSZAFNauu5b5ocGGbze8Bxtcq8TTE`vr z79j-s8*Xe$?PTo;Umm<#3v(^z?Q*-cUga=V-%)C|j`m#xFf1z46bQL}d>5a$*e-TE zg~z0R+(X-vN;kup?I?Ir_!rxXL|c2;At2^QXBUxiRKWSJZhL*tT+&Qr0qvB+(l7CG zr+kTuAG2P+C@-}^>;KJxAM)VJ>{#lgve$SX0aQhbRPV+R*~f{+d+JO)e5y>a&wVYh z$z$I5N@Zsa%bIW#hU4sv2Gp-d+_U$HeLDm%(alPiKE)B& zZWZF(?vC9`Wz4y!i^$gtMGu@1zlAyCfIcqX~{Dfo&6iEC9G_qm)qx2Oam#a~o=gP#&3fF!oHcKy3|RgT=w8K@76k5mT=wFRO0 z!UGfEYQ6BViPL308(OIOfDg=4G4LEPc1{xHM<>o@Go_jdu7lRci;4_ll0u!b_C-6c zu}G=Lv*XUYPA}Y~2HrxpPt$!wH`$qrxVwZOINn|VRvz#l82oX4U`Z}(XX>W z*Cc8qDl;lngtnZ1qZ%vV{D)w*!OfnkP8&6lkHx9@gz(j7%(D*&J4E%Nb9Q1$4Wm{O z1@Wii+~9}M`=>{?dp1Ot*_aO3C3yAMXvKe2|9ksXdkV7kfqNHdeg^7j02{QCSZ_wW z8y+t;Yzj(X^=_Iw8~T@Y?Wih`c|PKmxc_Sf+8%ae4cR65^Xrq>;_KzVuhnFZs>-Ni zU#UD2!?d=SQ$)zziskltSFs#Wm{=$dKyl;wCVgXbC_iXMzrJbTm3=M9F!6O)RDY~D zXKVPokp+{?XAn&mj??m|h8bC#!0DSCF5d+dn}L{U1N>Qr4jE%z z&5Yq?y+v_HvE@_@vjX(L;WF*QT|Dfj5UJ-WJiw(2WDXTY5`@%!4(DOxtD}`v(&}SM zVsIE|pi+7jR#4@S#GtS&Z`Xv!ZP)cR$`(s0`Y$#ljAG*-C8}11g#|DamCy()6yjun zQ0oQnLFTZL(OZt^Ny2}vK>3}qtxJR=32n#tD_?%~Q1);|D$K(dge=5^#hj_FNsnUl zDmI-`8BdFSs=Y)+6)%!_Jjex$$1TjZXexzlu-nqBIUAnk&-SVI5@Njx5hBC)p`I{h zCNJc^J776wDCe;#N*EMJ3Yj+cNJ~7oL?q)~xvR^E1#w1u4v$?-ZY*|*D=juqtq-Vd1W4*yA!8~JiQh}No z7(+}_zz5U?UjudJs`f!qf$v*ku1s#ZJAK%X)y@6MA{i6q!17;CdHZkO9)H&sq{(sp$J zq8?^i$lU#hI`W1>wacd+a(84=hexKq;hbcy^;@nH?8bOp5N&;3+0s!L-cNMBWd+&u zes>S*5AVLa;e(o3K~7iy`6k8dXW0c%qoK)ar?( z_XhzOOPNLyN9p-jFM^JIDrT#k`;RAw{+&7>trX~*!}BM);{`60t}A+UrH?=P_okEL zYEsx8s`le`s4=&_LU0|O^c0eK7@JOf8NRqs;~EC8%K+vJ&yxdWsH8~MzAnJEXxgdG z-7vbmv1rxFT#mORi`}Ki_Y6gM%}t$|(Kqiw6PixIk_9~5<9sO+$WLr-SYf8D9+Bvk z!D0E5K8rwLkHQ?O2k6#Bpf&z?1oI98q@`*}-UU25!gI?dH+?-j|p?(D-OMv=N zj?c)|^AjiVrF=q?Gl5#LPwQu_29u8k%Sn1=p+Z9Eq(BEbw)|aK*q@gyeZSAHmDx-9 zikQN*rq_R+Kd&MRtPG-h)xKSq2qe%|g_sI8VD(Pg2XnS6Gxv5fWRth%eY=;1$MxA0 z*5JioV2(PGc(-Ut)2`iysu0+{1$5(sj9Fndvmeq^Jo}csjrrPh^dqXup?@($Q`wYN zDhqc$kMt1LCAgP17)IH0)7KM-`02ySE@+ysF5-{~zVmUef7*Su)I-sKik`2N?NM|{ z`u25MQe7_vSRxE6>_6UC*V)fPZWju#2(Umq&Ov0aKWm093cbyXGEc8XOsi}V(4;qYzI1t1i8Q~$K~ zOz1AvAFLjCQA&^bPBE?xU!~NU@x2FmCf=3g9xPn+mLtzq(nA?IUgdc0X%%NN-GeIM z3sk@TH2?1g{xEjmU;4TcaU{TK(d>8irS;f`G7tgz^!IycPI`HXF zi~UyPE+r*`!kof&dk($X8mWHFXlnY|Tar2g z1}eYH@BngY9^eVmX+N}{2}R9B)(-N(*inFn6xwR+J6%4zp7y%mYG=ooMN}B}*_7LZ zW}|XF1b!6{*#L z?5hdC_=6^Mk)Od@g0z)er1uaQuh?4Lg9hFhjNh_AmzgJFtNSND zlDBM6VL}>)3$X|M^1E4VV(6(%;`%Rg!9<*jUB2JmSpSyvzT(Fdv;Fu6VNr=Tj{umrirfK;2M+EM`~k zk5_Hf{7ktFmjIqQYZcdfX}1nOV*S3W9W`Wcy1AIEEZ2i2(h#9FT~43*d)xx3CTuPB zU%=14ldx2`--AT|Mrl^4ScVFvS;QDyrm=>AJ_hnB&&Wy$5-G;~yH8k0IQA zi+tI@yylMqK1eyv761^gktN|q_T7WRKPk(w;+5;&Di=2EY=U32VomW2VbWiyh*zda ziGO5^BP{TdCV)3zPDtAy^!C<7Bd_z%lua;F_Gt!;_j#ayKSC|^iLTtEPxIk`Zug*y zLz&x#-mB5OJTK__j^a>0paNch;SSq_YT=^R*ML}YGZ48Uz^(W*jyVfMAl7Ere(G3Z zE?L_y-Pxm>e1NSH?v2|&+8KP}^o7BA8jvsMwae|EbQ1iBo;$b)ZEif+uqOboedYY7xL5o=$bIyp zAoA*v`R>e$n#IrhM(&dG=KU`qT4YiWr8jatTV2e3a6FXa9b=>c(;chSF@13c>iIEp z>mG#L59nK4+QdAeQPg6ePjr+6Cb!1Qn#4*~Mv{d}(H4{*3_=mI=aE;p_BZR$w%<*x zlS*s))psY5wzAxe7+1V^b}qZ`lH6E~T(6f%*fjJfGOOwnuuTT`6cd>*p)>mJbP!yH`WXTo zfq>^#47xk7l?lat)ZOi83gO)Ap5F+H8b7G#8d(XI zud#6ET1qmMe(O+ui?50mz#lLv7`OYPP_a@MI_v>GESH%zhn&nR-t3K9x)Bh>9r+Ze zoD4A@Jj0Ae5_2flR)Mu9%bG?Zw-X-@b2L6S-W zf~+IEiaU;<-nzJ!M-jP#9aLwhv45o2d$t{b9*kwOSI6d{*FC7=K~@Wdq&nKOCus0K zmYlZg9iB1mNQ1vc#&_xVSNnjOST@;&6jX0YWzF-wYF?)L4bnatFh?eGA>aZ&v~Uh1 zoQM{NF`LlC`_%hQ)b-sTPg(D=OKgFr|Y4$)~W&yg)sbJ%%=%Unu(t zg_!Tq^74s&y5GWGUL@$qYcWX)vSX1M10}1u;0^NLRqd%}Hg&EpWW6nW zJqESi@fe;n9iB{SpeEO8oBd*?=L4A_T;PAZ#oR2*cZzm@cbPs9?df!Ai znUYn46w}K1*q|RRu&I(=pW7uaAE&xQK!0SR8{I}dZ%$z5be;`uV*zFT%Ye$ns{%#~>Ep79JGjLcO##)ts*=tKP;C~MfR zXVk76^Mi~OkK45C{omsG?W`m>J(^Hd`wG6^V&mJ1>)6~^G2i__-IWRgTSvphmH_+0 zo#^xb_W0V5Vdc9!qb}pq8KTin6Y(7HC;fWw6sJg~@(o(%GU{U#Z{S6o%V|ElZ3%naz!6kSR4RH#j8l- zSCbSfWr&7-CX<%U-OU~iOK=N5bP@c>zTV-`c#u|!7dOmm@PmgEM=8i>0eQ_RDLZ9|4*x zGIVqytMZ#b&Lai}d_*faxDKJA4y^~-XP_xS@vO0NUVU)b;?)Xu626-;e9!mFIl)oB1#0(5g zwJI5VC=SP)muZ1IvCrUU4KAYtpK(p>bvVz}QEsrS=1D=20oGwbglE+$@d`0qU%$KWE{PCy|iHoz=TrqWqfPlQ(|9zo8U64St8x__*+COeqPv zA26wd_o>z69JU}zd?%A10lUTPt&r@gqO ztBi}$b{DVp#BtzCeoT{jDOxfJc;r$uI}xvDnFey(0m(l|wPE+C_PQ)P&S~P;odS8> z=aFsIRt+BfZm3y1Tn^HeC=kH8++|lZ2U++8EpkfN!jh%*CZKn2{!iHZ7n#o$f&#nGdNUwI5JTV-IBucQjv4r@EV~Kw2MN_h6V_okEAT003qG*Yd2WsrQ#d zt09spEc&gIz6CS*ea80?qf_7yZ5T1xGNDrbM zZUo#}Q#hRS#WUySm|;oTr*2D4D;B@uz5{lF-7#x=iZ3)0-sYd<)b$n=+0KhS>%c_w ziu}PD*Cto+YoHnd1MP|e~ZK+Y07*pYw*{y1B08Ls7jHIU6&`f7J$F4x|^#@23SpyY-tB@D`PR{4yp)fU=^M~|C%Lk%KyG9 zd4S3@I5-_){?Ranfv5neS+Iu+g=b3(hbw^opx8fDME2cieUKj@vk!YB+H#{3er40> z&IeZTp2F+j9Q6V>4HK#^pEWF#--roNOVuTLzF$RRC@NEQhqwUaVonvnVxC7eh~?+@ z{)t0g@v8z`E|4%j*wIfbiU6ekLsgJxC58u{7S6f70v_u{9bOR*^uLA-A8Pdp_~G|; zyG*QIRnGc6e6#IoPg+$t-&15CoJd}1c$9$aLOK(&-nELnOZd5ao$Ofsp{WJ;Tcx)| zqLW^a;$*m_gz0D{wr$hsvqFDv_`ayORk*!nGBcc`nCyAnklzp;tqHLj{|!^=By+H+ z$J>MVQ0KGE#KvFb4Rp;&l2EZEANxD>N&R$L_`x?Jai`5w{-UMwdk~rJSL87uU_?Uy zgoas(XH#^|J*YbzY#~erbhrOA$79WNQ&ax&0ppp18SfWhqstDx-9jq@6Yv95NSDq) zl3wbb(tbx+2XR!Z_<2AP!d~s@Ha{UeqXT>Z`DmZEWu1yhWGe~v)uObm@WrQ3sn>Q3 z!EM*EwK+Q! zlMZMYfj%Y!A4L2jOZda}r~K$>OuFuoN1wNCWtw->tJNPIMqE2jDJIw5Gn4(|N; z<((BJWOaodz#S?qjMBXNI9z%ba-bs(IUdzrcM%tPUDsGkM^NE)eVblurG!vgTn73p z*bh$|8LiM+a7AcqN!1?58n>liGF_03$}VBLKSK^rp54s-3f5j^!CZ}~_t-qjvvadB zAShHQn-MQyi_g^0BK=MQiCfnlim@z^2W-3GOCHnLJH(@^zkR;o3=4|cgVc*x4gsIP z4kRX_Y_R?0xk;MvTYiO(?Z#*Z zR9CH3T^4Z{m=XzfdQGK%{cK_An@fRo!G*Q?M@K~=F-QzV`RE)OpRHx}xJ!J{y*}MV z2rRi}ojwZ0LfmS-vzw0wJ>x?KBwE{bE=cXJojGH3q*gVhA@a)z)?;)o2M1d2u!b-) zpX{PtVW~-=DVMVdVh%k7IkHY&m|V+z+b=R*E8J4t(@344ce+Fi}>)=A)t^o~JkRG;6-hP*H8?Jp!Pq1gdkbxfg zfz@Z^nsQ=vrgy_#dgawk6oe6;u*v2o6S!zm7xO12p7g!garXny>SgZ#-TdifTQc)( z??J0OT$kOasd%%K0nk+4PK;X}YR%Y>jNkjDrOY()5Z@>|KI?7UxLFB%vGC^|jIM4d z!hO_Q-K}M9&Go7((AE~3q>(!jvV5R@pjjRplRj-#t&CpXgx)^PM0j>cgfLJ!W*X!8 z@wD^NKzU!QGFCl3TjOzZrl{o6HQrZ;i!Zh^dy_3h^_24zXQ;IXkM(-9G^e*-MU1}r z&NRL+c*CGO=C(vwr03G( z2Rbb#odU>c!JXXNf+_CCT=x6ZgEqZ1<}F^`Bp2-vUzdefT{`u_wsG&ePcUn> z|JSV+n^=*gr7V%L(gf#))KiW$ZCkv3!vTnmKP*qmzqADiR721A{tl=U*djhRI4<;r zpucWEC;KXmL&`A?j2+)UaF);~tI8J_WNd9|Wck7?iKS)vC#u^E&Yok!`Hpi@22*A! z28OeFr)KjI6iSuj4GN^yDut{)J<%_F!Xu^g&Lgqi1LS_~$SItOREN%KoRZ%B;Dv!U zSH?=RCoq4-j?uIs4PvS{7PevA2FnZA2SNM<;__!;{)BlJuS+`5n~%xl#}u<(dK?FN z&fj%)TSkg{^@)dGcAszL?Ib@wCJcB9e^c2ZHy3*{bK4=l9t1Mhwnzv+?;LDwEDY!W zLlMK>=lj|H)>^#21lJvQ&aY>z>#FK>VP-Dk{z=mu)_v>~Pf(mjpJtS4dF_971T9aN zLsE5uOe_;@7fW)bj7^DSS5yNf7Sro16S_{Ub%N&jIF+YHA}Y7Qgct5Ue~~22co*?G zI`Xdw4)5R_~9a;nxDN4_jPu zP8T8FDqY79v@6(pzG3PA32lE<+WUzDI9!&6IQaH)BI&pnO96^u?oKyqz4hT0Q-b-O zPp__R^{uxQ5`lC^fh#WSF~5&8oZv^S{`S>>k;YMaO+sL9olKAHBd;dEv~)9dOJQ}U z+z#^>p|{RE)(ANZITjt)NBC2shf%FQiZva!BamFV1NV;c^@mfh+lxZbJS{UPcG7yT_xU4pxxMAK6Z7JGZwIa1gPK*2#ardd}yax}^e;r2_xa)r{Jcq^H4GJ1GNdo@z#gR3ET~Fnz^n zx{@iTTn;G*E`&`E!7(HSdrv+;BCOQ*<^-cyOg%r?p{lH(zRg_@YR4u$kthBe=EL9D zyP{yF90D&UqP&@2DL09g)oq9v>aM3v0Bze5iqp*y!}XJ)n8lHA6F(gD7sY1f17w6D zSW2_TC;Vf-!0tROixN?RI0CaOGxb+HH*Ky2yVvk1@e-h@4y5B zzZWE5VWls9hv1&hRpNQ@MEFhnRK#BL&c1(E9h31LRe+^+BZKLOtk82`lK;}sDx9Cx zYgbVY4?v>|cV><{bSe!5@1~8-A+eAm78~rrCZXmUrcE9J@do&3;O&>lyXD($y6T?6 z9VdB;YdFvaQ$@=@o`hBuo{sO8ts<|;`ZY9GOAvRZjE$MlYwTi`bTBT>Jg`rfJun~W zLzbtDybPA&j^=b_2n)=gB=-+|LQmavG`ZY7%lDxly;aVzVh=mDbcHR=4>HWwV8J7- zVY2iBL^OX&w>~bx%EYt2I8n+LZ$!zMIa;8Nx#R?>AtF6E%FMq`v#}BPdj&-6dt~hV zuoT!9z8y`8)f|>lg;VFz!H?gWM3`$Z+AVo>XGy)L&v3|UH$UpvY%Fu}7CzY!2*G+r zNciz4A|=RCoaw7f)|L&{Vl^`)#;7>MJ(kF*ngmb7cAwlw#oLO7IrgK}4{1}0AoMd* zlxc)z#xcvEapovhm8ux+F%7}R)RVw_5LcGuF(2{jf|RCV07$0M@$(?lW2Q0@rmv2L zIq1~h3k#j+{qqZQV*{J)`|t!%@c+PpS0}+%phG<6t#ka8@?1j&n{t(3QSMV+Q5hqn zJ7tkQRa%Iih35LGw1VO+STdia{t8+P-wwO$3yF9!Isl63gu z2-0DIJP8sp5nL{JTP~!GC!W0%dv;#3WK{*(ZI|o}?}{0}x$*FXnfMJ6@edR&w9QuL z#&_VihIuSSEoP0M%kQ%;DH=O5kq~ZKpk*JGgz(GTO1XULm$hZL@%tg0^Cln-gnuqN zvG7V{U#FFNeEZ^(r0;TT-4MML9g^iFoC9-dv?3-?%N-Q`v}xD^{!T_7v%+B@DXXf` z6gy*jv&?E4JboGB9@j=GUe(7p#+^fb5e5?IUZ4ThkS0d>ZwdC&5IFDPyI* z?6S;E=X{`_0B`jn7+6y_uRKh3##_K?g$dim`?rxOGi8*Pc6dItQ@pAa4&~`bKT$VU zF2sRJyBXS4l%8| z#%6KeV5w$y?8>?!viJq+M(y(Sc_dgL=-T?TxW=Bedu2;p##rS98+4KJDU(y z>A1bople4PtNnFg4qMahW-Dy1#zEN>uPu&gdg>lim^`d)D8+Zef1=A8)zX^`DGp9Tkl~9U6L87Blp!+NF2TDW$DeUQc>sV93bF zJgobVnLUe8ZUsas_c88W zb8{s?yIr~MKJ^;x_Hnag{tdsJy3|e@*IgVMf8UNVmzRnQ{~tKW<`4`h8N2Kr1BMP) zQ^b1rApdk?gt2-edOU^wI8|=mQH02E3~|P)kM`1-@c_j;r86(6G7hraE_*h__NzG3 z;Mcr`C)2MUVjl8K%3{ZC5rQDn|GW}_Hpb2;CfNbSD=ez$iM2FraC&+G$`u=QN{9>K z*F&9U+g-IP`$lRjA7pmxH*D1c+gZV2Bj_m0+ac}&XT6f$_-Q1o*M0L|0p#Yn8FbF| z*qUd@_4(XhIjBw6{3E!CSEc69(P1Fa&ho$Fvgfo@`4ar~AmU{$io}7B@s$Q4Co@i| z0)xTpZZi50PK&<6hgABq)+_Zo_2z6$t|Na+&pWMm>7S1#wqM3wyY&`YA>?ygTE5~R zcEh8tG?GOlpI}Uvp;0mO&4Laq>T8Hv_l-u^!{oS(Oy2zEUcS*lldtuU!T;^8tL)jQ6j`x3T^R{^f#a%;8l$n+lc%0M~Ea$I1e1efc_gr&lxG~ z@rc#p0`FWB$**v0=q_(l9yci)<)<^-*B1=*`S?1_ z-be;`kt%Nv(XHxl!23uR+26sfvNYrQ z9^O#xuX!ApdqbYE1^Oq!LV=~*{z1M&ykHU+x zdS=vF74KKnTg}U2@a~1`3anWOvJrfgVl@}w;W|S{6eoAGM@`8n3xC&BR?~gw#MeB1 z4G=wafp$@bPRI&oogztxm_M%SF_x}HnAy9pVjqSo^t}TlQ-p0#&dYy8=XmA#`Y*Q4 zg|4NoZ4kQ}tfW^=gaM{vqTWR^N#WSTH>CP{0g}J1iNW@Z%tP(@SF9 zU;~PsvG4DrA~TIb`%5j1?|!UOoiIIUE3kKJPOrLoH#=)UX^s1mlLCA`2yM%b?7D7% zcH(&(JxKC~h{&S4TDy?PJE=qYrJETg2-kq}uVgOb$A&v4&*xe?99mkHMMS?=W3>gn z%e%k9#JN;tezMG8N>6YP8VE!kErpBxUvaJ}jB;I;&DLnwW?Q)lOP%P=mTIY{G4iby z7vfBNPn65y=LebUKM#wLIPX*;uBF0#cdemA(SbJ_o6t8TOQa>*q`3QD90Esao0-Ab z6V2C%`!urdDsZW|u25WvI5;n_8ycP?hGTU-Zb-}=FVva3G{3c61QrtTC)JgamVnIZ z)A=U=;F?~lxb)7`*k_#--o6{2V|(@oDSdTu$3~*2Bwd1jJIc~f?Qs4Hud=`&Vy^*j z(%gWR?&O?g7H5en;M>Oi(YRhoVh&w)I<@=^eh=fkLFrT0BTZ&mxa7Mr1s}j=Iw>ey zDk{Ji!}p*PslT6I<^p`PcbkWn)(_pI^DTo%^lp}uwkc|m)encIl<;+FXkF$VhF$wg ztY5WEmL}Rk7_>-E)c7+BnqnE%Sr}l@Y=A-ga^#3TfdT`D$!pt%fqq)o6Os?&+bfmO zWsZ8#zCXq_hb%?7i=YvA*LT*anHLQKZ#KgywYO(|5*Nup5i@A$Q($#2u41lDUfmwO zVW-)Gao8VBXC&&$Fzl%;CFhx`?&b4bYfmy{QvSTevx*Sw-#mE?k=p(K9vnHcX^uah z^p);eS5_gy4pFlUa-=+-9x-h}slPcfJM&v>t{?{?x5v@QE52Z=k0xC$UtkAQ-PaZz zNjR8?(zzzbRdJPJ+MKj!>@9vWB75XA1X9i=D_4{}obkqG^*lh~CYcg`AR9`+8UEADq1^6b#0MPB7 zyq9I=%o&mz zKK}(FV(FJS=lom=+I(Wfep00Bfl1E7UUiY}OBcuSOjWhde5sy%0N%o*4@sCmx(C6y z8bl8`+GI&nd<3X4OLW~+e2?o# zGwc>QDl4RyU(x>n<&J>@pZ1ua>s6%JZP>#SmvLqS%mn@ZaNFDX;i@$r87R!9CPJ6)sJ53mjb) z*dG#jen(ON00`E^8Wz4vA@QhABpUcjoKBF8UFxl{Aj!E7BX~O zc`mhBA&D88n}DLN_^HEd^Tr4wEJw$+TbsqEF#iDAnti>CIKhztEC;qtb8O7)S&1x8 MNj>OSZtM^L+1rgKN&o-= literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-12-Stacked Server Racks.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-12-Stacked Server Racks.jpg new file mode 100644 index 0000000000000000000000000000000000000000..c5280f3d02e69035051ebf207c5fa64bdca8fe06 GIT binary patch literal 43349 zcmbTdbx@p75H`4Y@IbJ|gS)$v-~N3 zLcu~sK}JRaVtzr#BElyjCc-BqB&A@YCM9DaCnTidre$DZ<>25Tq2?9jVH03t=V1G9 z5GX`ML=Lc_x26aFM7C8wmO<>eO?78RG2metnPH#9aix3vE2 z>Fw(u7#td&o|&DSUszm%Y;JAu?C$L!93EX>UEkc^-9J1&{f7$*0Q0}X`uP4|f&HIw zVSM0%hJ}TJMfeXF6twGyz+k|_ePxBm6jekpu>V5F=JyHfTU>5UHzGN^(gn7m!xRz@ z1;-}k<$uur2eSWf!2JJzA^YEf{U5j>02COg597gL0E7XLZ{ajK(Epc0=Tsk;F#21P zh z-R-zjb_|N_Upi=DNDwBc>YwnlPZxx*XXhfMIgv||`sfWLG-Jexi+pK9S8kLamHX-k z3gjNuB3tv9`wg2?3lj2Wd36{IxiO)4 zfJTx5YT;;Zj1Dp2meX`!r=w2U8tx@tI(tnD9e9HUm2gLpV}Iw){&vcI=e#Gp4f*xF zSB!sFQj36F8%%^|G1gfHULUV6JQ*E3pIP&zc?X#OG7>gPCOMnrXuE@rkd`9y)|&2| zw_%()X}qtM^4YCu*^J4#tJp06<=dqz9eg~Y1qfZzW!9h|2_i9Uc?V=+daF{XyT%Wf zB=DN@5Y9z(eA}Y_+-LU=AnoDu$hv|0p{bMoXh6LykMFhZ)cEHe&;^puck@^g-dAI2 z5d1zNwcQ#g%F82I0u4XI9BTqLfmsD*>$g#$*9|4i?3SzJkLN64`Vi$=YO%0ZNYj9&~A0fr2^-zERFFaFCwO`UXP$} z{n|5!$#{nQnEs)U>7UNzyOfZt&+%!K=gT}4SQ&~tmw^;sSI)P_`Ym%n;nlPo?_+LgnS{&wdn)5~b4AU)vkk?OfB#qn@ea9t`{c@O#`9v+YS z-5qB(Q(>tjJE9FGCR$4lQqjEg@t2r?DMt2l3cKn?^Uj|NFd1idJmv$Fw-p2NZ19n@ zx57dp`^z(jB6!{^fo4YDDnTD!pjOz{88qr3bJSD9iPzbJn$3|728jB5;=B2)0WoTb zo54YUJsw>q`1+P`ZrkYQTk4>Ea-K-(B@6|U*H5ea`)3uczN_fY`Y^@F{nK| ze#3;Zrk06}PL+^{N&AOzK&iL_V?kU_tiiS*kvldV083H)e@d(lNsRpa_m1{(;zhd2 zd-Q3ysjtcwY97alFebUlJ@20?5LN|>h|3t) zw7P!cs3hN)YFf*BSNIK|ARVc8UW)4e`7fk#zAH=vk+$YS z1jIQAKcp>pyaNJV;4Sf-9vflRwoag>0Xr`ZL9QS~v^f6FPx^?16Ecg_9O+(0y$L;x zn|{6YP3Gh0v8JVM^nTHo=51jZEX3I$k{3B-w2-(Kihh3@a^L@X{H7V9TuCKj zV@=yTyE(w1)F^~NUnvOk78GQ4%|(5@NNTJZ7;_pLlC2K3zA;Jc@!l>^*G2M?#SJ6fwNm(Z`>LCukwqFI+I>xJ0D zLh^$Ap@#)HItVQqADxSrlG50=0WB+Fx1yBel$YYOgfVq?3YYzPq$(bGTiDF62pQPI znkv{~uL5^B&EXH|z43VsZzf!x0!C?gy)(#T!g$MQ;wE0`@qC(}jxm~*ovPd4Z` z;0}WC(`Rrau(5Jj;vwe)GjTC8V4uRpMF|}GV22I@o&nyH`uNz9Qmn-E(MF0L2Q*Nz z9r;`T%nOK`B9Q5+emJWg{-T{==3vr-o!wrZ7;61H@gkCg0rM!?ek%1wnpQ zcXEi5PhsgRy6A6I;6s-u{G-NhvP{HSXzyu0;=er_D3co)RynV1w0CHAI+xJ25^5;< zM17=O&YA2IG0H7C2hXrk?*L}NWp<%y@!hx26!k8^Bc>03U$WHh5h7_=ICrMbddy9t zGAzkYg1s$o4GHEr_zGE2GEtDgI}Oir_C%X-B%{|wlUCDE(x`SE8x($K)S2TDB-F_l z8x@#E5lwLx4l?|uA&RBcE^o&IB@p?f))1VI)&ySE*3tD!`2?i|@A{cfmD|LQ-pBAT zAnG~B=f;=yNqIqL`U{F~08ADW!h_oGeV~>V7=mb=saS$FjLRd19!D`#6Y} zx|-9-Qn7jxpB|LPi?61TkzUPh^ai6DjEboJCOCBWtnr^`R;C)DzU3Batt#yFFrEC> zaj>kO6+q30U<%D~A$;vHcaUE4m&@4d$*bXadPL)Wv>A8@d{Nt#L)(|WHfxL(u&w!Z zN=|B&%kN%x&vnOD_h=^&LR#s%A2VF@k6yDLnaapA2WC5m9L$RIgOBteW}##7zsE|E zP5quqA-Zyd-Z6Uzh*lPDcd4;ziIZC}lC{FLs6$tu-#pU;{DGOnB#BK(i5zzXayXPi z#LYDs9f7$yFtDrslB7N^HLFhSQH|4nM>dilTFIHuf*;UaYD+0%?t(4B0o`dHhw}^4 zJ`CZ5=*ah;+4C6%`OTD$sf{r`zF!2m;o9I@U_M^sU{9%b;ga%hds{BD(Ap-5#Ch>* znMZJ#fz`CTV|~!m&UU9Ik&)DSAx5J}=;dJ_#~FOlI{G6;W#(5wZi1S~cfdL)ifvB5 zsVv*ejdrZ?oz_A9I^0sHpS4G81kWE^6Gh)*yo%#j=zqZ9y)A29K41JGg>u~bb)+#r zFHUMH5DWSIvh?g;KE@#{nQ+f5HF-vEgNZyJxG5?0vp&!5387#mb$p>FoZk=n&1zi* z;{Gz?p%||_hDR8^3TljRgRRsD&qs;QI~FLtT?)>HZ0hBE=|M zp5Da=dGWBOP22?-^K%+XXLaRMylAalA0;fn6~#1gW|7g*^X}qA9h0##+wm%)Szu%P z9Y9o}9hw{0v>YTZ{Lgp@d(O@ETZG=6vsnB^*GP6OGXuL?wIis`!MCd{YarRuVc*9 zCih=wNHy38dEo$>{|TA$S0s+zj!f+ayuw+a+Nk;#w%b&|3rVx(M`qY-*zHN-V66Sv zzou^1kQt!}%V{rlUCj^8u5xjpgjQ&*5Li61!A zV*FSRLmCUGm!+TUI(xi?M*6SkPAD}iOB`V1!X0ZyL7#5%s7b063$`CH;%)}mo(D4( zEC4|PwcBm?FJ-$)-ISSNaQtu|%e4&Oeg^t@<1H%u-t^M07vGGX9xg(x558PM*N)yE zuz`RTKT(y}D&tzbv*fPw2uB3W4luGVh@4GMni9&e{TgJw*Az=n%8P>TGv8Q`N;m#C zzw#q~EOKz99Y8>x)5kTf?d zs^?|8b$~L!G`bD17~OnR`kOY73QX8md&O>m%L62 z#WbT8%83DrHT0Z4@76`zD>qze94Sjp*?#>QtPbhfC8Cp!pfRXV1pXXtCKFFsj}>R* zS8dMnOeF(_pNNHN6GS5%s=x(Q%5sSID%IAI(kAn~(zSF3Ypiqi$&#b}64-cFQZPSA zef|vPPRhlqsIj84(zZNea~U61GglW7un-PQFU6QH7h7WWZ6IkmTH*V#a$|$NJy7*o z0-gR;2CY zXDwLTt+gP-bp5s`;*mVD-zQP)X9T;f(x!v?-KiVCzUt zlu6ew+RYppxi3gNeK^l0j&m*BN^=}_%tGMc@d{B_hr!Xr)L_oy`IpxW`TGq{J3uFn zOk?}{k1qg`3CsSb1hu;HSvRkSW||a%C=6CnfOWzjkUXoM%q_82dz4dpz1ez#g*Xg19%@E`K-bU?U z(iNn{*`^EzWs*HRdT++u4AVUY23dyrv}F|*e3<$XgYqr3U@=?!a@)!>U>G*`*Ux6H zD-XXH`B^{irG&dA%L=Y21}$OI$Fjc-6%ol`hT{W~Y5G+!&=0k7AVFDqu$Dzy5Z{6@ zGCZ5iBk*346*X7Q(h?h)KQNynBl2ky?rVvb^&{AX@pupYq8Rc^q%L%$k!HHJ)NpB= z9FoO3G_dO5>$5EU)}h@nzXwv{{aWj6$cD{J{Av!3ch(oQd-!=!D)#M)8X__wA6$rE zv7k8s0dno5((vpP|5+9f*KTWw(^7YLDT_wXD{9@j@Jj!_Cy5o>?clKLI8h81`-mCS zR$ENAdC%l!Vk*@G zf8w;dy-K_9^4@;fm73@w+rh*A3=?i-ViMZc`7{0rBsu}SDXa1%-usUH7($wyuq)t2 zU)FN5!uHKLKp~4}(MD&Ow`-u#Ax~DrR!muI%6nW)fyt}<56C%Gzg0C=L6;(O#>wdlKYL>vbC`pjhb~ngPU|A}C zkTBSFU+6`Ray}YjRm;#=Q(IjAdfs0s&up-iYRfm+6fch1->DVrdv?Z`)`V6gAj+5@ zwx!>J^8+pQrtBVP2zi8%dMAs6oSNUA(6obUqU*_dEz?GFNT_R!p3A{yyAU%idLzM; z*vVNZ)~wN;+i8{h9iUYr?Y_UpP9wEo3iws%)Or>KpElh_;c39Hg7}e}v6b1orP-pl zwg(t%suSSt!`jO;a^`Rv87a&XoNx^X8&hC3^0Kr5_wnvZOk3h?Tyzl$F%rTMioEEJ zdRjS@HOndMbnyE3te5Sxd)7*hX>E)2P!^MQzGS}T6KCD%>U9}2ZlrCkgx(dABn%x$ zYNgA~%4KEs+Nxn%!w~?9Z?PtguWl>^>d?T0t1RQn^Xw62*YsAkFBU84eK2ro_@c z1|I=88%AIg`Qk|zQMdT? zR1)-Z)zs!2fnXySirq4E(xyJwmfG-WUHPD`e{%lwr#}SXAlcxeRn)eZXFx_le9Io; z@Ll11@|96j{L|8}QM7G;q2N{@*ne1<3&?oDjk~=TQXS%xE+*WEI+lg^@+`a7-u z4)Bs4It`(W;|skr#djz{%y|ePps?1}>CBLUlin4tFU(h|L+!Nw>ghv9l62tUnc#6k z)2LeVFFvAS9_2GSkYplk<`U$EJI(+u?WeEoxOd`#9{KEd2o#o|A8@dW;9~W#v0KT3 zFvA;rVWhHc`Gns?&6&1|K1vYHNHw$AGSF-MG|PCb z9?oRP@2BQkJ31m<%sJG*S5|H9)fzzsm06C(26`8&f7m7nAmsZX_nWu@+wxW)6F_+J zp5o*wIJBVxn!9=Mw(^%^%O8t#4qGaqyboPCsJK+gbRod(*Mqcms_#2M?SUk$N~5_` zQGa_xIWC&M=Z3$>woGeK8z=$$=DeT)&X=qoUJo+UO8-;o&osE(uIR|%v0TF78<*uV z%cq3MM5K{fA+`MqiP4H*+g1Vn-WwXYn03Jw`Y}2z2^;sQF6JWEF`?Ml*bpPr7zT5S z1dmv$&>w|!ki@e7754X>H|p<(KtNnF=TAJnn}N~F=sra1<7$KfpQC~kzmbR*WgZK8&}eWnj+ z`bDQ5+V4Qxjrw+LWr2SzGxR)dnfYwWRH>72JrFZrgqT}LTh!AG3D^@j&G`{f=I@;d z{Nryo0A^|XGCS|j$)Ju-)*~#FG}fb2^CViYz1LJlpK{i$CRmU$#{P-3-PYd&2P;39>Ft<1I6B!GS(A*w=NBifa z$;cL_u4u30%Fg+oC2Lq>`p7xkACSZvy1%q~bD1171Wb@`Hk@^M6!r@p>mqn56J9f1Y!!WL+&TR;I~rAFa_w``5hO1UqhW z`Ulla4@>u|xxlezA^XO<_S0mzZ5|=JZW5Vke#A4+tVFuX>}qW|eID#Sw-R2oF;l=T z9f%`SU(yDKKf{$XyK+s^5&4?j3)QioQK~#+f9TmC*JcBO=*?blJ%9C44TDJ(N)o=E zavv4DjfTE@!{7aRsY8iQs#i|>OL1@{KoKo&8*#&;Ll|Mgz8GRwH?NU+&Fm?gk8K--KM}>=cZW-rPxQ) zud|L>^wR6e-xTZI@wFlH&j~z7>hFLmJrk_uTf`h|Kkk57)|YP`B&u_mva}-6(tW_732iB4chE zqG?J&sTn`7YA_jGUddM0j^zrcDS{Hm`uYDN54rKC%lJ4#|9pW8(X>={ak-{~zU5SF z1csH&#$FWM-7!sb$%C|X)xYBJ_TdV_CQ#di6tYkb*7_&%)9e>+>_*%@2 zTwSll@JOGhS^ve9<{;B>n&(3`}tIo;9I*zT9WVGJ{H~ACLd%=%wt-0&0ra^ zyB6mO`%n>lfe>)+>mQ!aJ3ccq@s zhSUuRrLPyI)+gsU&}A1dALLC^Viby_0!=jO=zb2im-j@G@u1q)er>>2?)@zXl*P5e zENUJ{cm%-(Uem8avsZt&slVa%sYGY z4)~{U88<|9q(PRpXB$3j%S#W>0OF6u6B)ayyn^N{5ONq~_zadqX+Xxwjv*d0s7|`r?0VyaUXaNgl9` z=SGkXIHDLbL_e?WKcbJv;uuedS2KE9Ap#kvkdL8eL^T)K)59R&y95z( zP8F45oU{Z^{mPcA=0#}V0gEDH#tE9$kaZ@poCRTZTil_B#z4O|o4Xee?>|ub2SaTz zO10z*V^|0EQmcbrN<)A%6YMx~fD>rb#gq0-f6mG{m=MTHIyeD*wFp0$%6<%6La*>o zEdmNn{4Oc{({t@O6Q@}ZRmP!W25C)<6T0$K+qmfHt(9Ut2UD2mUlWYP5+dy%y+xJD z7FfitNe-&@lpYMbkzb7gG@Gig*2i>ZrWEXqv8^#uQ;5_??Pr_IKGLVb>x$2+#V;PZ z=v=~9SCSi!8h=htY5pP&?*;SolDF4c0>=q| zPx7942@`|-T%@LIvmuQSx9ZeH3_ejRUOM-48~isYbVYJecJJb#6}zhRF-U^gcoKP>llYU_^nzHE3kw`0ykzdhIZzdi@=j z-wUjyS-6msBk$rgvvg{Ag6q64%c_uu7;bpTw_Y>iocogOp4K$3g+>#h7iC6gcOy~M zL3Ystd8AiV4KH3>m60M{j^a|kWi+9z`mVy)MCU!Me`mxuM05vcpYWoG9V{v$7?iML z*_w~g6{Uz~{qg(-51@cDHEMX|{-~6?+H;96s{?!XJ<)hd1TYldCUuC(tugI?6-BTP za{STI{~uUC>^a?)!)j6NKWYliVnyF-;Z_DRZYnjsFP;|E(&YL(wh9~lw!&Wi@EQ%Q zkbtRk2yZTI6KL{SU*~NsJWZM!zgB(?*Y|W!sqFWGI-(_ioGCil$);p5Y-n0H`Nl>McAv(_NjWP}#(y*+j%rP|%0# zh$!Bc>Jvwe*+~fmt9eO+tlT_%lAq?ra|JWvsFTTQQP||YI)l?e^wsQnvrbY8plRf$ z@=@V`**H}N#RxzC-68+SlPo89xxR_2+1at!BTnc%%`>#pgJ!6%uj$WvdPUxx{v<3i zjM1a2e@hr;J3Kl1oGAH+kW0}1r+C?e{o7lx?u2XWe`)RtZQ~_RXTo&vWA0(M9sC2Q zY*-6i$PcYRTPrgA$;hmZ$K3iRR84v58(y9m`5^RSh~$H6VuC0J0791#iGyoI@`TF! zg6(KZw428ENsyfSY2FHA)hU)VL>I)^jw{zfY6KkPm{~;g`&$j>_(U1-BaV1;cvyUL z9NukwEkn)bH`~uJj-db-dbhtXa{5D9@fwhC?9IbJ9s3=11>ON@=Xp?&z0VY?q+c@Z zrH2^1=~E@2KeIyjb<{b1@NH?h7uQnZPmZ58U?OUc-%e$zgGm$u!-^$#GAi6pW3&dc z#uZkYQ}M*QziorM1-+hTEnL!h%*ptbfObHJuYorY`B-);(yGDHS#}?pu6y=Kvxz? zl-^1_8=i4Nx5Czo-{rPrE5O?(Ck|RV3Fgm@TrUSIm>!O4B{kCUnVw$~I(Y#+0?yDW1R+vW z>GsbIErP-`ADhImk`MCJiZ#GoePxZAxFMk--K(ImTS26V_xA)>p%{l zzbGsnr~AEO*V$iO-N74sws|c#p*ao|YvsFcZX3Z5QAfCZ8m{!*pO26CWL=5}J((s8 zj@owwV7n^DIq`fPa@y3LCr}~cD<|0dQl$pUua;OqGu#pF6@npM>_d4B&wX) z@(HgSTS`xz4#LO=qutEPcxpYd{jYE!+JH?iQYb1QXTg=X#V(Ek|ypvM8cUS1g3uAi=*d47>_LP zKld=Df2$`AI#P{hq(laws?~2OUMg|m*QkmGtoYu^&Jq){d)ah1;EsFxH%@;W+IKezTHMjk^498&AIb%hNzY)?C}{~ey|^8! zwkM~5lY%0MIvd@O+?au0?hIUaRUm&``#76@QSfzh>ol$YQ}v8%@xfot3BmrAHUGWb z=8(=3$mEgnH1G8e2r_9*7*ceSlMbg(w?2fGzuoN>*8 zwPt$KylyOsV@InnrM~O>LYW}I?A}+(Tn!91GSwn{eX}Aox&T`e!C&`k4Qp!+hE9pZ zsl%p_)O%g3-d}#-hOeTZ#zw%0Mjr@^2}&Hve9-M_wu;8rcL2) zHbdMH#vNQ(W7N@vF>^0TAW^pC z;m7H3S@8`k+j^Q<>EbYf3f`_R8huyW=fd~b088?y+Hy1wnPW_U=?ch|E1m@^nKVDC ziBv!1X7SIv{2`9lD`2 zyZm3D_KtF_+w;H1ir=`UdIxHC?v2ODm`AKFEfacPhmA=aiR1E?R+)oFNS+8|nl)F} zgYMRAKS<|2j(A`J{EZYRKIV+I2DV>qwj18&!3bch2#_C8o%HTvU( zr8P~6&hm0f!DG@lZT@O`bOo6nFu%VP3$9J*?aMC(v(*IGb&k8xrt z(G&;lzRxZQH8m?3$_o}azui^#3TlLyjqljCP1DhyfhcX`&K0m6%L>zhjapfS|8%Bz!=K)X`}mLg|+Z;T>-(S$%H@F+Pv#DT$f=*-UXX?`zYi$9b9hqf$#a}qZ% zExmP*hdReGS9*!AI3M0_H!O+IJ3#kWK+k;>tmI$643kAFQxF=j<$q@Q}EO_B0?zzTUncq zYM}hg2*w*;y^kC{r;>_jg&0P!uXY<|2b)S-DNeQ{MS4&hO1}N=m>jGU%;}h{GdMkDIo*V3TuG0fTy(O5EY1PW*k^z+mzmXw0k!a>!kc zNt;EME{;Ef6W=+B4F=y#W$q-eMdE!s)lPJePE`RUKWF+Q4Ts#U*c2RV&P6cor? zE{gt3Elj-^oKp1Y(%PY+)Ueb7a=d5MZ6VmrQET>gWTZ_Nxa(I)N2$s$O?>-}O>Zd| zW>ocK=i>2GxR6W3i1pghhJDExc0T+8{g-`mFvnf|4`t&ErYanPf%^Zt+0j6MrEq0m zA#gw{to8R!fzOSnp+Y)lEO`Vx7Icpb(j8pft|cYFr>)uY3FdT@+M#(6s)}y*Vw6jl zA70iRMX{_1BLTv!e^z?W6vvcbzZ^!GV6;S~T}{#$WL7P6@=RL$L}=BLNFY4ThHbtI z#*9@#m^a@63pt<8UP>7XQqj-mPXIkdthz*hLz%!8d5zj51fiep^nP%m8`2{yR^yun z-+iQsqU~lvBqJ|GwL;Ao*iSqy)%6DpMMyl0Kpa`yoW*l;lU75pk~~}XZ`kP`hT7Dc zHb##3$(F$kUbNRB?obL4!>Sa27|{p6rTgccv2C1jYUP zf&F*sGhBuhz1I?7Cc_BVS8kF`%_dXs_WETufliC4GDp87IZV)kw}?WGOynoR=+2vN z5NG?HB_ErqH&kTwTgPDzGsvSLO}@eHa=b={k_W|GZ66P8nTRj{H;_dQLnf5NS&VTkF#$%(`qxR5nd6wZBOg$lUf|o zcW0!qdjx4TR)6x+?3M*{P&&oPCWEG1&(6bHtTzP@xPd?U%jo+ zJXLs6^Q{HF=MhQ#ipa}DdZ+tLP~fk|-R4}P_7RBM=6G)@S`oOprp%gCR7ftuEu?DJ z31~5wGO$;g2bUKPRVq0w2e(|g#-7KPrS>2G{wt)3J~1tu&s#8W`!a}zYbNxP)>2@B zsdZc`z#eM)xg^HzQDei-lJ4%`7P2jus#6Rp&2l%-ZIPBh-BHq;R-pIp$Y`A%eU4i zN$p2LxX?^=P&T4zImWoIj{4ofGf2N|w(Mvd?1?xB!kJ7K3pY_7yVVC|{->g{6U6*ssC5L*BFa^Jy!DNWG`RO2PLl=~^M zyfFNe0B;;<=_r25Fs~)E)NPOGC$Xm)jg);+foI_9)?jY3+VPoouzP$HQm=o}TeIZ8 zAkG+XE-LOgbl`YGTc5QNft&UY$P7V;j$a%eN-Jayms%Ms6~PFWpJ|Dy2Huu6Y5lPK zNDRv~0ycbg90xe@X3ofbSGq1`!aw;iHenNQ1tyDpZKiY5kRrKS`U^IlYczD~1s+o0 za4Mzv|6kr^p~->M)9|v)pR(FXR$0WzOr6qTzIU5A)Y*F7ysez*eg_Cuc4vfrrzMFk zq&DVoQ92mG`#fOWjkq5@H~7#wKMd^$-jPH=>AE3F^blWhsEA1BMNYe3$BRFE%wS@T zT;Sx@6Bmqj__xPlo)g$6VXD>>9YR+82WVZ`^m^9qFsjBn`#h^$hh&H`msx@9QSVi(Yt!ck``xqZa)d(!-fE(EjyB~8XyF%WMj&;Fne z$}znS;K=6w6A|wLfU_U(_hiM>Ehp5(UVD(|*VtgDZ0`;HDP_a%cd9k}AWk&$G@n!p zs#KGj3nz9pY%haMyD`trKdo$rl8;Rged_Z>D|R$|V$ZHy(Wm#pU4XQQFMA&(~Z+nazy6RH*0t04;TI zq}KXI6st4ikjmR}@A^*cw(#{+M%sTvd_fZqfXX|86?U7p?*zBsZjR*o0#zq)+NY~> zlUcQOLa%+LINkSJVl_C&&siCFel+isa5@RGHZ<2+iMoUbT@^HMwLH%_?|?p{iaixs z0vb5u^=O)HPo1)yIOYPv@lowAk0lld+%))*PT|lRm;Bs{RIRJEx>W&f%4Mn|P=$b{-Xc}N1=J1wRn{i5y z%7{+kTaCj~KSrJGl7hn8l-9B(TRzkP18_$%KY{MIUrA#=O@lnm?AB~-)I*nKUR$6sN_@4uxy7m^rmjP#TiNZgaZ1k}+#>_=JZDarb(d`j zhzHWnjQs+in>LJcPc>&VGsC5>y&OiKnJzfK=qA~y+lb(mH7yW(mr(a0cpCh5RdU(p zH0zh~p4O2^RO@HgWRGVu8h5ZBcni?#Ud5acyKqT!7GztS-@^4Y0r(=obV_b(YkgG9 zEA=H>6;PM(?v=Vx9c>W_P1YZ!#GK)N+?ANNtl}JNW?dNw)_Hx}DWxjIX%q;*gPl3g zH30D}ZWt4q-BdJSG(76IDOPp&xL|{;+-p{SXXtC}ywxT=CPIXl6-#&08-xp;B@KaR ze9rX_%egesIYyS84q41I*4C9wOlxbzpFeYNps}VMRG8AE+c|zz1VrdUR+6G6qty41 zcsY9zJ3a9ma`)yV^Cpy(*#vBH#0=`jHzk8(J7&t*$XSr%?FWcx8!xpPa647or?Sa~ z+pXRKpYO`EJ~xKT&YPh&aAooBhu;wQOXlmkInc|B$4W(Nv1r(C7rN_JnOYZnCDWCS z%5yI?S~zM(u|y>Z0SHNK3*D2$Hm6@g3!YIY4Lw2zZYpSp8kHq^P6Z$wID6m8!ou_% zqtRvN;y;}kPu(3=d2pg*WZ6ruGR&U?KHk zY_1YfWca($xf;jbO1R}WsSH^kSB{n4udvleqIwMnM@jDhvu~I6)545#%aiIUhx_9- ztsx}*6lT;-t9z|rlAnpE0M9-%2^y(8p7OFxe+HYj&1JX5O3b;Ps2;oGv$PG;rYwS( zhr2W(JMT>TnFiV5WI1EZghb&?cOCYTyLHB0E8FhEhf2o$ag?qHT| zIXPT){wa(c>v?SrZ6-;+_V}GxtU$8t*32#zm@6zWKQ5P26lXR3`D&_ckoSW;O(Y*G zmG2$UP6w(M@20lC%87%Vw!~1Dhz&o=D#Ask^>}|uU85_2t)x#{#b4H!6!mi+ z`4p4BsT#)Zb!;p~Z76gr%_nqpovKf$37@?~3;mgi`v zyWW3wQqKHNLQa_Yw2%e5>%OH~Kb@?@C2`vpSdcAUlZO~dL=ektweT3UgsZx!5J(A- ze3wb9HK39;R(PSXVVy{&Ch@CMySQnx*CG>d`H2$w>Rv@wpN^lJlJz6%DL@x3to-yJ zd*Wg*v6mdS(b(q*{!<6-mbfklh57ciJAx8xW;Ok8X&tlIP3(Q2~eXy3NA4W@YO66tRz^pbhELZDnDjU)v6kT)8=?R92~-5brbLU_wjIUAN_>X$&G<*o!i4aHYOU2YyvzZ*gvDU^QI|K)n! zF1ZIQ@DL2`tY%JpepPFoo*cc&M$r4arkGyP-e z?zVy7+hNIx4GN=;%EqIOaHbPu7+-eCnv$Lzb4p*ZB28{6YM`-JU1PMMoOIv{9#C}fTRu*?p#8M6_z zKWj4~dZE)3>dOVFGRqeBf?ZKK(9$vQ`P*xn<6a~3{gZm8c6EYs8$CJUIr(j6cLQ(I zxI(HlufPI;OBy1EObsi`(z72=&=0 z@Rt1rnYB>FvI2fhH-^_zX8HsYwzfBdt^C!H z2Gs-3bs5r1lPIZ_r2ZEGX+f607|At-N_UmYPeU`rVPlXY`zza(vscN<9qThwyuX&t>NIx4Ap-`v z(WolhqKUg!R;hkPvy_`8jAFT!FM8$k6)j@Bj?($=gAd$b7XJWvR-b~1Ys-wth&C@Q zewEGIOjg$D7AG;|lm5`FdIihFJc^ybC-W87_A@7}{g5^!5OZ}qR z=lDJ9^!k;At}RvU;D$L0=gcj^&+wYn)2~(yLdsDSFjLeVV~WyB=t%Y*Lgr?j_Pj@q zd9Nq&eAD^zf>!|cuAb{xh{(^e$v%x-CcikkxN?xol5?EoRy5$#DY8u_W)R*;!eJ`{Hr=)jF|1 zD((LOO;7cz^`Rj}6i@+06i@+06i@+06i@+0T1yFxLRm2sN25O(`g+u+drM^k-aj^8 z@=4S16agF>ZQaCs>$6qo655iyLBOR$GNRfak*~18vY3V7{HKa%{380~&!4c$SoYd#t>(3Pc092miV$)mLxKMQ>#Abv;2eph z&-=!JHOGZDq&YUMxH;ye{{VzLTV*nN3xm(ztU)yHzzHNI_S|aFMuVHFS&zkmGeo)Mjj*1DWH^29r^Pmlw{6Bp$D7O}z zkK&jLdDC??a{!jyl^vzxu(!AOkLOx8x^uij>h4nou72}%e`Ou$0@LYQwC(m_f#0qX zv&B;W&A3nF=L`g$9CeuU>N{{T~2W=B0(&;{kwt`b0e*a6OYkW}{e zH&M!6WVbuJd7`p%eN9=hvboeJ@~zad11x(e^*+=AsU$bA{{StmLFtb)R1@h}P_S9< zZB&t*8AfqR_cr=SG1@ZEbB(m8ix%uY!(tOl9WwqTi< z>0V$79DUgU>%+}7!Xv9$IUMCA1b%s>lftsZBQL9I z=r-YZ%mzXHYNJ@kp~ACY;qKjcM{l3G=v(|N9(##yC21sLSkeQgx6-GJ%e9AB({q!L z-;94cu=*wc04mzvoW8i2e~odOAm=o@oBPM}qQG=*;o_~9)@?KEynl^O{vQ!rEz4_G zg&l{?P6kKg?OEDhn-I1-o}}e0kl|B5G|m0d{c4V*s7a_?Zz>~+c#b9Lqtcig)^;#Q zcIzJaIfY*yphBqY&@xws`bi9S>wfKvlX-ppGPS5mVe{{YiL{xvB; z{{VYZv;P3TzxwshQAL30(?dyJW&YdKR}32lVg5Ax&ksT_OHHoWd5!aYtPl093pk9| zaYu}T!z0_hKT(xszm%%&P?q6Q`(DTDWgp{99)SxU+;OK{{ULV zqJTQ0Na;Ww zG4Qp-w$~Czs9MbxvH6NHKHjwzy|0%R-|b5wp3LJv&Y9soZ0P#J`M7IKfMJ~vLZ^?@ ztwG{lAn5wUkf{v>pS*>KLWhw^3ylzE^3JJehTxsuKU#zP4&4=)YgQ+YfyPhkSaegc z9R;S2*BrsASr$Y%Qy>J6{AQLbdvc{@)ZkL1`@;sX?d~MCxJjZobU78JtHA{E$EjKj zMK==S&UQygw|?h%TetI8dWG{cAyV zc6T=KtJ=XJ;E9ghAAFkR)<;$huE35#&1+lTTk2?S^%)9J{{SrRef=BSm4LN(p(JmK zHM ziY6H~1o}PU+>*;`tfvGSfFRZyvm*vpAdZAq^ctm}qi^ah7a#WB5Be@++AA}{>-AGz0>Bi3j2hxKiriLRjQs(tuQto-dvC;LVcvD zmq)*s<`~*e0~g}6gRtnvuh{C+TOopB=UTb><3G$E#(*V#67oE5n%}Sd^%Xn#e)1rW z*S15R0?rAju5`to7Q53WX`lly89zLKQR(kl8=L#b^PmoitncLDUE3Ug(@KeaF{#{T z(W7uq2@jE2TbtDWbu2eiK>Lx;xX%WV9VC7S*D%VzX5q(Wz~<3%f~#u+Jj6@4A35j|^*WxmyX{dgWA+QU3tKIjZ-xk$$5UdL2&N zQC*WB>SYQl%w8SS453Z!tg)Um9op_+(z%_oVabt3+;l#* zYDIf3L&y_}2GDm7oPd9=0Ccxr8q=(&Yj`!=CN zAN`eQYHd4eQiqvAL__m;`9&d~;@%L^&k@Uo;ChVE1(xvrh?FSTZ8rY^c>e(U>c{*d z+HNvDPpAH4{Az?(myHN!xQymPn^|*?Gn#z&W-UG`)u3sa%CZIM2jf5+wq6jjv(wvD zpI6ha7tHdaQz#3&8Oi7Ls{lANJ6HjZ%u{c_&!+ zLc5=4>~yKbuCgGO<}^ZB77NoE{{RXA(y;Jlwe`}%z9Q399!cH~2YxeG?zDY6J9n|O z@jZl41BU6kLOYY1&G5FPGx(AvB?^R>1g~F_star8mikqmNaH~x4clk|kKx-ua5d`& z{{VrH@uWT-wE!#!2lb^Fmr^zs<}IvS8D7Ib;Z@_*FJ)z2(Q*kH9-scI0J9&4 zVRj+4%`tJ@%sKsQqJzVhJ|D1x=xaJ;x0gyxQag>YE=b53?i@O|_ED1lYM&29Kz{v3(i9Vcv zjRttUyM3Q#-ut(Lzjqn;_Nvlf+(5B13t!$TZWpEr9=_B8V0dcK{{YOvrw{vRKgP5) zuLRrPTr(YeOlV_Ubo%02%88UU)=2_JJ()8FZEDHlR^89IiZKDR4cjjdfn=52>hO^U2`#ChG1 ze|m!d0AG?DlM>rlN}vO~%00$?XafD;hUL<<#I!nwogCK+B$Tzl2WjL1`WnxN!xr)` z={5Z_KW{h_{Oe1^I_=DNmeCtqM&ATv6B#6BN4;T9eLTi&!*Bbcll!s{U#$R5E`em3 z+pgKZJqYLZs0NJ-ece*e8U85%{6SZ zXCCwctEBisDx2!{XwynFKCI-5X3u|Ug8(xxIJ2J7` z1GmdKZUTr_V73wan3#X;)BtJT6ukR=yZyIL8)Z|?%KD1A;vWgi7P)ih zYMNwwn|2MtgUHWHp&q>?H*twH%OAWkgg(Dov8!u0S30Ea9+_(31>8hTalrSW3P;0n zhR62SnU7zzFZk5I@Qd4HKk$%gfA{!*jY2Oq83t2Ihac?1f5w_+)ETgJn2UOOC;Vsv z&-@~=cJT(4{{Vm2{A+#uG}^a=B)Xeduq~&`BNBPQf~U5B8s-Span6w#2mY9!@vVzJ zersO@-asv_hHk(p$Z^2*%>ZIapv4>DYW55NIa46Kn&D8 zDL$WjHMF<-czDl4Z~-H+{HmhpPI+tnJy+D+3fl1oy*;+SIvYzSKQGMDl6nt%Y}(br zNh(JjgU1|b+{Ag^jym?u0A-y3bQ-Rlf9>IFSaeBkr%5B#^wo7a1YEBbEZW_|N@M#D zoY==ABj53+wz+gJ3_oSl!BL!;$Q?85KpPj{7Pr(isMzwH_T zuHGB8QU3rT{XG37ALC3u9D^V(x2GKX%m?+TR_Wx178eouj$3*a`?$_Ol`>0-)nXz@ zU(8b2$4pQL&9{b@$fd$~n@WyOpe@3m&zijc6SUVH{hL?Q673MYo?L@tfzk? zjAySszgp264vzM!w>~4YxP0IQAxe*~PHUWw`u1QIr9ucNEGf4N@yM*R#@w$?4g`(-l&059uhs@)xxP^u)^5OIvwG*2k0QVIEJm!Bp;T z+TLkX?FlBx)$%*%9DcL`pq?MooSgWA(S0&M`t@eY2h-WoJ$A*Q&KVhl1|Lk;EwmvDU0@?hpecOu;C}jshSl~Sj!OslOTE#llf2v zm5dNwBBlNPl(CW*=5QT-!OdiAnqui1w6% zi6540nv+?wia$0N)Po7V=P zR)n69;Y&RZ{{X`=vRg*7#CFN@ZerbXYJ00aE*Q$W^3QDAll^PV4y7--u0KWcfHEFdsULwBepBJ+!nwF)Pg@+=e13XcM^z| zCPD+07!BMUR;tf#;tk(x4I;B-$T%ZD-qk}{mJhYtTic-#pai=dvy;wG=Zet?+|ZiD z_}H$g0C?UsP_57spaN^UzS49^oxG^+N)F2jI6s|vt=-x|;%UReWl189a2sg{V8I9L zQ;lK;f=L$dP!b^lhVk_DsFW`6ai?Q@QKYY>hjrn5d6L#0=r$iPZY8>!@~IiyObkE@(-J=_=q)wNcx{l}THU7T z#!M204URWpgXv0f=39bxI?K3p>veDRI90u%KjU1sqjQ-;gBc9NXb=&P){xqixVL?zf3kQq0#9}=Ul27%nUsICKGR+r}7h)O%)AOod+g4yX zw|z~K05StO=zrif6YBbOuZLUmrvoh@J%8XeF@s%3-O6|H+rF7B&m3k<28u|A(Tw1N zIrph;+R<(_6R1bJ)G6Mi^&5>Bw-#@tUnEN@lX|FW7#t2b#wsmDtTAYfBtlg3PE-5C zdE%@`sNdM>=&1IV5P3|XfRU4d+r4wTjP|3XFvRQ7n{t1xbEjFbjjYd-9B&7M)3?267&X*xMD3`Xo1+?=IdC$_AtNAl9eAp7MF8;y+S;d= z34xd3>3aIP?9175rqxgAiqu$zF zLwj!=ld0YsU%CfQ2d!^i+38RgKVh`TVj?F$t$E&)s6_>$E3EN4$UvSG)Dibj6`(b! zw-k!|P$|egI{sA-4ZGZ=^e$?*5GX!MTgHHLq65$4n!wZy66xTyNG2aLIt{s8;{;W~ zeP<$uxwyUo^R==LK>R8#UTrlr*#62VmV1s&vmd&k^aCE%X$YEEwH_zaP4btw5yiX% zjy8()jYCY)@5nxii&X{8R_e}oeWBE`6L?i5W66cnmZ|hcXBX2tV zlE)4Q%lK7gzK-SWZ6iiVSOQzk>keZG_doS$f3QEtBv z{{U4hwaX8l8^8PO{{XF18?7jj`^&;R{KNbyT747vP40c%e}w>0h2-kZ5mM}!iCZBFS%qRS7i?!8sv1*uW%SnobkV4s4Bfn2-m9=Qx zppNfQF&+UVa(^nzk>2+!n?d@7)`h&d+RNrVDTIUn0HIb^psJVlV8^mmWeDq0)tYNT$kafgFkLO-=^V%ngZezWT+BpbN@wOQV#t8hX+12gVW=L=KCyDRSJ*Omcu3f)}%jKLcK1B_(xRF#BhGS(IU|w!_fVn($!fN zRkyC@X2wbDURf&t0L3dDndIH{{{VoRzh|u5NgvE^pk+b4LgNfa0OKB%N<7FDvD{xZ zoL@Uxz#MlFAMvhNQq&|e;Y?&R4WL_*kJhTmtw!j}_M7rCkPDCSsM_mIM#aNltf~QT zoF2dM8c~Zt*$k6KHoIxEBuSOF$>oeIA6#~=PY7FSmmVj*8k7Q2y@X+p${Tl6S!PQ| z*DMy!)>u$5C8cb%g!kwwTMLVOd+l%SDC{!DHZ}^dkaL=*uB-_3Tm3Uif<_V9+p+_K zRw8hJI_C8$t>ac(c;buK{T%c9)>`Y<@qf#0L7!-5B0=B&15wGS*xYZG*HJMhH@BR* z=l%mx7&X*wm0oD%f^Rk$$>2EY&DprEp z`e_w%qDcW&&lx{2y;gYy9v{kx-2EV`C(4$~k!9-qTms%<^W z<`Q|MY4)RANLQVrBBYPbvo!wzDYa?LL6r<{R1?#K#c6+P>1B4UrSy_+!7VW4`eSWm zMR4=$6UebFWrK}__(l&Eq7mPGKd{j+yg4YE);GOET1RoX<2V@oD)rT^g|`R=SePB;<>_HB18tg^E`zw(E7V^}(T%(^LvH{N8om}H z*aTz_wdMN7$JzW*6x#!V6hNu&3IJu8$pB3egrl16yP z<|~e~n8^mTb_l~m_XqHe>%}IKs>cjbTQbcXf`5Ap?In3QJQ^h)WnfOnqTSkQ*5f~B z0kPTx{x#0(Pj9r9x@1gaXtX&Wj%z3TekPM*I=V9G!EnyCh*V=-mAThhge1!f&T z@0wAILu_WS(S4m_V2>(cmQ_uMX&ngsYA+D;cW!V;`ZQgC+!Xy%?#@dT7It%}cFRvY z?k$1PU-_6O_p@uxuy1J`m0vHZ;}x3@eCYAl zaQ=0%rK=^&x!fJ*Sn@jHk)P6lG@{T9F69}PW#s*KVP}H>jJ#^L+D=6AqrgCc^X{lVNCu^l!Ts$z^>JU#Vs{tDg zyN^5$c{N@r-W>x|iNQx%R5G8as%ssxPv)iJjg_*aXCRNpwdA_{W|^!=*k)l7EC*gx zX0{|{u7FLYrSnY`G03MsX>gf2AZ^c4S`v7CJ@ZE*RlNHf<3F^2`y{aWHQKfT+n?<=R35N2$~dc>BIr8IDhO@Q zF7u9XIIhda8a2_dCJ3bAzl56NwB0{gveYAy?WWqp3>+L4JQ{G9GPpHu3_~e4ix;Ej zwxH22S`RijCh}y9hLK8weZ6ZrV@1|hGMPbg20gcQuVT@37`0|8dWHP9$`IV3#{=@B zbLCeOS3IcM-X*g|_Hp+BVo%-YH1yN1H9MKsBvz54P^uf2BODLPy>D3XcAXugfu_pm zZw5>mI3x6_jYTbn<6|pNiS0aCBAOGEaFk8398X35|%=kluhZI9aY6t{uhWSZo(jJyD# zbp0zq6+g2sU>9)#5iE+lKTHu({f6#RzqteFQ^D*2AJVmMd^{}T0#PrQ)VKuX``0^v zZY`uD8DwQ0PihlPEMC-f>&t5y=avT$yn_x0%6}SmwIcnSL|{nU2PAbLIi~2hlKq2H zw0A8Ou;*bp83#H2D&3mLajW=#IhS-}8;7>~1YCo_{vf{6WYoX2u1d<_qdF0l)9oWNCA-ee8WIJt_yc)@Om$k(N0DRv07YH9n(hyHb){ZARTfpO+I1 z@m8*+VXnh8gfq)(a>MwJ!ZTe6gF8>MxVx}XZyfS%xXS>)AQJd9IvUegz zAY-T?ikjv*FMqN0%T{P3c*!F?7Rlzf{{Z10irUIbh9Kby>P=+o_BZ-%z_q`%mffSsRgp@PGhJS-rTvP@W{**lAr~Zod;T?(e}4Lh zhS_c|WVm=H9$QE`3P(@YoMf0ZMXkh>YE}`Az%&t$m2z`|&{bQD8;v|fZXuE}=W)DX zk@T&&*^^aS)Q$03DO2f{$Mvi$TS>3=B@#}u7Gh3cnqnhbbadD_l z^4&~JF<=88GtMf6Y&XNbaX;2IlxNZI05u!<6H<$rFErF1rs->+d1d}4oAF~&mX-P! zBkP&~>Knn*SuA%WRnw-DHdc6HoHGHqbmy9aO${$^XPWC$g3Ve&0hhn)RJ5#%s7cR3 zoc{p5TRtF-uQZ=F7#O$P2+EA`6p8@Pe-CNMIq?ml{{VG!{b>e=qk;2yiXeTUC;tFu ztkGd}jQxsIG1MuhG%Ta$wUAH-!f2Wa``#l_{{U{s{{V`sx0W#J9vswd674Qkq`ZY<(@WH0|PxUD=L%aDk|rc$)?>K z=%aWDeU5SNb5zZYmN)u(TqMaWI+jTgjC|d_c&^vRx=BlAktB_Fp1ZMJ*=eTHtpt*J zo=w_?Vsn6dW~vhAR})q#YfA6pY1Y{p+{uJcGD)p?{2#8{HKo+LUCQs=%nA{V=91b~ zwedEgaLtLMk1?3WQ6QS2lEcl_Av zl7GgcgG{`E1g{OsBEs96B+fpy(0HRtg(6QT^mGT5k#KdW`oE7N0AhC2$lHNf@o> z@T3vQsC7tkqjP-O&+A;it8R3wqE}UrGd5LGPamyHl3`-BD?C0RvxezqNup&}Nk$6e zwkgqB+~4ZjtX7c8E0%M$26*3AwF}XGd!yZXtt`bh$m}pl2LtpKt>H_DitkWVZJu?@ zBak|dPCrU**<2BcE|q&6rx0Wql}|kg0+LHB(G9}9QN*&Mj2Fiv9R8K-cK#95THHuf zlEAT#wof(4>$(n|XL{1Y9ROZK9OvG$jb$fdoK8OPO1-d#TZKmZGsK~!iU&HYmjTp;SEye z+Fe0yBv^<-fTcBh)FnkO0+wH#VE#4G zYqtLY>EUec=7Y-M7Ly#0!mIe854YMgCS_}6G9T|9qnhYW>gH^#pxE7Bi?p|lurX~X zw^3a!#)GTFWaVuS?=4a?Pq9<|DSR!b++RayYZNc!woX_0n?38T)I2jKipr)LS#U|o zJXA_{b{a>Y9TNIUF9DB}0;uP>+r?+yX){@9qDl1ol$mA+&p&f+I&|xtQ?(s7?%GK_ zu^>kzoe0qE@Bx(JlY7`TOW&i+l?_QOwXm*-}26#}b1v`R_ax2bt z`+FN{;_~8i3W(W1IzrtckZ(48%%Ww$?@UA`| z_3{4zyg!|FUKDmsS(yrf=P1KAI+0T?hL2Ranr4j-)C95ZSUw__N3m683`xc@U48AV zc~KQH@&46vUM9AkEq4vL-JAyYt~j>YZ1S0sWV(O!MJI9KsjX-&-E^%~$e{~I5ymn6 zYgt!@@(Y;`Tny(m)M_lPq1`z-K1@50{0VEc%*HnoMX19Nmk0pu?OxZU&VR7&l1YaV z=Y_8khwRhqP=}S2WpRbiVP4B>_RC^}&R_(1`?ZB_iiyhWw$NWqHu>9-dM$Cb5=Cia zdV)gx+h#W7y>_=I!EUD#TuN}H@Ri8CHa}#U?M%!>ib-yRE-LARCDN_D+v|ifq(m}( zXvp(3^1Rz+pj_Q>&`T-vg(&9BxL^i4hQ?9yF0t9Seg_ONz1QB`g>H?*FnIG z#+Gb!>4IL{jPcBXSDRUB7lLpa>KDmeZJZwU*Z9`*&GgxWMJ_i1!~X#FtAx;*PO%px zFPOLl`-*fGpcCALAsX6(dR+xS_tkrkh2lt=_gurbcL29NE5w@z#2Nnpr{B|cdY6ap zV_9w^xmI9hAquuQ#VYFY6n8p*5XwYl70)Ar4S8m@D|y5BjmYVm^&83TCbU9hoVGUP zIR5|&^G_0`ny__AvaS8^f2CtXdlbxhK=bSy)(Lt40KTiL@Os2_J0&20t#@xydBt*< z(ncFkWKyf-$-z1P^qG&4XQ1S41t(OKa@ zC6`ZG{{VLjBc@dB?m4c4-v0eC7Eo6`u~hs`KA}B?xD3n_Dt6a2TT-#3)f^s|tbL}} zP+N;@U`4sw%7h%}>rK(*{{V^RiMa4MADO6ZHCwT1e-yzS2(gD=nKb<|mK|qAd}ML| z0HGS|dl{bTZE+w|z&JgHWO$0^NkIZ7W(aT!`&98Y%7-Z&pLmghS=V20lHrR*Ha4H( zuP!&avEsUfgx^TMTBaEE7_FPtKX0)U^Bas9ZDKOn-QJD1B>o{d~LY zK`88pcdV+_lm`{Y$1kOvr@kNg7_epFV{W`1(4xcr3xUeS^{HA_v- zLtt#^4Z0|vbI-7M2yjv$hPvVacnf2DV~^F#LM3?HFAYlpWGH;Lma9A;2? z`qQV_7`OQIkOR{BhyFZk&^$R}BD0ys3rj0D593~ac1gZI+K=@$>-rUo&21~JPn9S5 zd(l*nGL$&{wLrG25t`mw#gex@whc|Bu_ab8K<5iwH;6SmhTnjgBaf1|^y}NlMPVT! z*92m+HOVyxjeu+pM<5E|r=*P;=Kla{ch+Dm6^RZpxBZh$N*CmQwGO=o z*{qOA8~q(l;q}Q|*LhQ#YNR1)jcCDtagpSr5*+RMU z!jL^H&b(cy+(CY#GP_t4_iM1UGsqh}ar{k+`uxJGUy+zV{{Wt2`BzbyRy_Hs zj=2@R;qw~bF`Td-YVyDD3d6S>u3{dp2kToFlC7NgZ1JpsI`!?y=92^2uV;Wo0P`QG zrE=aRcpBAnJ1EaN6`hZAo)J91D7MTPTgb5Mv_5&oa^4sFG1avXw9N4)Mj2CfPZiPx(mv9s`$k}LdcuFkv#GgJGJIcs zc9vihZg)RNu5(7cf9*H~*7k)^rmYc6a-#q>^Nxb|{{UvMK+`Ymqq(>eM>8~2^M)AYeJjr5A+8!<`gjNb z03NNxyD)2VAvj4KXX~_6nz|A_D&tYJFzt@sTpmH){{YsmQ(Jop6}Q7LJn-ZED)L{( zY?QQMmEiMO_xEr!0ptUoIW^5qS(eOM%q=C4maGs%D8n3|;-XZy`$m{2l%QC$pbYml z7t4N0L%tlC24>>MfRGhJjItP{WyNR`jwQ))BXwXxqkstOUn&0|^1 z1H=gPKNC^v^8setNyZIx^83Z#Lk=jcR@VH(bnJF=40!%3k}^Ksp%gsvpZo5!ERXsX z_xpqY074ad03Po~+no3R0KScM8J?#4gs}%zm4+Dftb4s$;baHx4nGy-2UW;w@7^H!ZLo5S)Q!{qJ$ z@vhqT?%yHs$8wAu`&JcI*>TwCyjwCATxaf{rnxv{51{IDWA3sM`HI)^9sRKjz9)Qf z!{ho^JT4{C^&<>yk%U482oFBzF7XqH(y4?%05s%>r(3S z-K~n5&T@Z-xvS|dmINMi6TBb0?^wbm3!^K2lf;ZZ=!tXx0Dl&_JEQ)P_~2Jnqm%uY z;$onYB?GMI>mh zNgIdONb5VFp6Vn200R{pBLV*aA1Z#0Kocrl2G7`( zeNfh}pjZ1QVnVB(oDZO`bh~HoAI`Ql0uG$Mqi_C#CmqC~ee^oYOH^syz1R^Z6RUnGBev2-sgPZL{g3^OKqG~lczO?ri2&Cgv$x%^Ff zmxVP6MYZ5Eu^ad~_OCF!-0LV+EwfW&f%P@rctYY+74ig8^KKcXRbC^G=G^!4>|!Nv z&b+t8SGNrE4?Xd?@K3#ZEyRc zTGAwnE1UTqa~Kx{WG;BkWvI>OpO(twkItO3d2#AQc9fC-0DV^E9;v2XT1NpzxRHvD z!{x?BRllAI(f7QHI*u!y@b;i(wu$w~S{{TZumrr0XBrSA^SOd4xq0>N*N!JuM@7y5B{{Vd|7y|zQYcT%+rb>T7T0Ml0 zwlfMJ`1mcKMXqPWFKW{&w9<9YO4aj`5EIGAt#f+Z(Y(L}=N)rd!D==nV%TZ7VkdSx zM^S;)RV%_vN1yI5@~d{zM%qo<2mo-Pm+#b6I+6uwInH8J`U=>RwZHs)#s2vIb*9oZ zUv4p-t+{deip;hD0FStT>!SYvI@fP0FK$Ha$c8{T9QL3N#>Z8;lS7Aljlg52VrsM9 z-9_>=I6Qpk6zfS~7BV8v`BU9%7f~>}jNC9F{{VKfiqxjTY;Rqc0^6@<6?$0AT5hz5 zsPcjQ*0CUt+{n4SaB?s<9)DWfmG)`6$~WHq);~3^NiipaM7{%KbW_0g71F@rp-u5k zz{di!-Yr^LR#+w*cCJ2Qo`SR>I^IU<2p{hc@Tp@(d(&zoZSmZU_aFYMtRB@Z;f^0W z=ViwK0Qc&}-NvuAluyV$qN&|nUA48ao=wCMdUxGT125ZLeUk3&Pga-=egd@w!_NmJ z<>b~(6U%R_$s~BrGb&OpUxqh?jftyj#wwf)^+Zq6I-le)*+Lg z^^{^;W^|1X_3~OwXPw5k^weoJ8OJ0os~@MzYYA>6E)N_KCH&1sRf#rkvydNr zDXoYp$t;uDX$CmVjS{yQ&+!voq_=kx*hIq2XK3ePt_~Tbj>6O|g+U}>eNAgFp>k{l zesqMd1XM{}Xs@lk?5A(ocq6Ztc&N0sG5B{{SKzcs@}9qaYK`poj^Kuu7^!U}Wzw%v zfd)v2C%?=pmcTxHg1UvPW26Cb{{R80G@HM(>IFg}0OuroRa?hDZMJg~pWP3X_u{0q z)FKxwe#aJQVmethi-Wl4Fb;YxW8UhMOv?)t`MnmdCCf=3<*|<$>sj|LboU6N+7wcI zRHQQYj@xvf5vTX9o9joS#au z5xXxN3w|b%a7WD{AB957G2B$1B4Zuu z!{n1saH;6?LEj)ND&v8%jTYsk`ZrZ?*SbK3m8BQh)HdwDg{04&+sK`l;xA2${HOz?(yyOSSOOKypbkg1Y_6+u_Kf8)y?$E5v`JRt&C)bwSab7Mgx8YA zX)Bv3ixYv2)>2xOna_BSQftfq0Gc??(SSgzt1s^D20rY{Y4?}WnSOgHvVOH2$g(}N zBnXS6D3#+i3{cq42{C?czA_ctwk zyQE{m&VH02W$Cv`rCrWC9YFK-%^h8ZBc1;jzT9EMFk^B|w@pvpYq*&1xQvO(%XIIW z?yhOmuI^^KJEoiI>9*hd{rv%NeCnLfIp;agV|D%E;`21;=jguWwlouy-nFdgc#p^Q zX7ZYxB?-)*l^l&eT@eZ|wV-uEh>zQI63pBu$DiDaWGR-pLca+g@y%Fj?bb)q2x&3! zE!xsiS4pQ%{n93ThlsW~OottBTMoyScVnE-65NbeC$Xfu3@fKQ&^xcGvnab#>n6k2IW z_Q^`~XhqY*iH!o6@@!Y9e&0TJENn?zUoYrqRrG5;V@L0e^cI#ZI}wg_g`pn+*L5NL z;jx;Ak8I+V^@cO+dRLom4PDxm8ljkfQI~G|v5AFmUjd0~J@n`>xbnU% zdU^WKpXv7 zXjYIyA*}vt<}`yN!cijX*A)7C+ zvU~T~6!#4pR!)CeXGV-?oD}VQ*dcLu-{@!Y`0Ke4js0*|VxXj7c(igoDMvP18F!q* z#Thy>>*sIYLkx0F^yKJ;3Qkk3a9jf)g=$Ichp`Va|0F}OFpzO?aF|MTFJIG}W04Tx`l8!f58*jlBs?+8M@F}dvD(%I+Y`gO-^ zM*Wu_r8(nb;j#@Yzd>LdR_tWvWq(=F=U&UX<24j$-7L^CBD6_Z%}LkwY?9qD&$|Sx z8oQ9=8Sc*c+t1#{_t&#E8Wb)o3N&K`(kTc7%6$D+_Y--)Gwd0F*+pja0eN+@2iAZ(~(jmveK zJk*%{>I*t1OeKfdyYqXQTtnI_Rhy0W(IGY)ASG_57TlP@21sWV zC6JJ0E7ww-=1EbaMbzY$dOLZ^7q0Re+;P0)whd!(9nh8PCjC-9C@5j z$w&A0Rz4)WTF7T9aDa{bRahW*-++=l?)c~AM0cn2xg|Tx0U?YyayijgTc4iE*RHyS z*lmYhoEI;Zn}r5YYw#IO@ZG6nFmg&3_HZq#z2GermRupn{oH|!MX76Oy!i+AA(nA@;yeqoiD+hNYSI+6|XDgnAZ?e^*o&(M%_Bf6;?~J$`uz3BRm-e@=-1_ov5#PHhlxuCl1C zU=~fVtu0wqqe&B}Kgmr|+SHFaOzm{hl$s z@kv(DcuS&M#GZ2Pah(YD+ZxlQQ7<+$dh1Nb8!k=Lx9mgzK=bIdcBd7W1^P^*Bqsv* zxXMvPVn{9*-_)zpP?=QmVj71|_{X+nK4Y&03wqTr!F{5sno(h$g$t+%`L}(7qnJs= zulPKkh`H0Ke#M{3tBdX+m>mo6OqbTpk>USExds)5EEoOub1{~|ImSCu%qA0yr|qP7 zR(mO;$~n4b2q5BUE-W2{a^Ap{hThGc55MEJp-nt=E`sS~ty0Tt0KtdnhIfNliI&sz zVWef)FFW+J!`iC65Qeyapqy2^afLVh7~#Ehzum?n0x`waAA4z6;0g8qRkS{@ls%@n zYASGXk6aH}DQwyIYCE_WJg4_Bi0F9=Yd$n;1=AWzn7+SUuIZ~b6#k+Nq?DaYyD7TH zqVSBv{(&|I_3WM{4B7b}O?62H{R8b(eN&%!`RE=CtZ?oNf+b_Ig(9Ui9SU9@?ejvW=ct7lY*CWRP?ab%mqjUTgc^KOijET`qgKDJ zBD)^^4>V!zvuQN*4tX4{Juqf%-M1FG@v)OJePC;op}FdD|4UpatDLo;F*m2+u|_-{ z|6IKOZ)G8%4oBB4(46gBf&zF#OP0C6!XdyrT!5!~K18@ZvaCVUJJSZc#cjII> zH2<`uPl%nJ)(LUpi(9d^SBhG>>Q-vp{iz}F*9WqjSzw|Civ8wyM^0GYXCrb#|HXuL z(U^|da)u~j;Ba|WBJg~~GGHfuW3ju{P!KgP+0(Occ}%N^=;k8sQgLnttgq7bRNo%m zZ=2BAcFM8D7clWe#Sd7Yo({9mCOi7YSGbM{v8N(gp~4bN4|SMV@3nT4=OaWcU#Z#L zm^7qD4)2r5yG9hg@F#;w4*-F@X0-1di$MeL_5Hjz4R9p?RAKM*%*7%mD zU12txU)>hna^X#>_m%ugnOQW8G8$dJ*emgti;+BP8XRE@rN}45A1P4vNp&N#q`R7k zt~&Q0i*k?{S%_2~?Z)oJVyESJKlU~y8vS>3U!~?3^|BnZOS!}bYT-Hvq+5+tV_Ij% z$9!_QkHjagz5_t<19#j@Sox9dj^=Qugq2~6K##u8`84lWe0V2yvnU&tMjduHqq9wW zDJz1{#I%uxLhN^?-ER{IdD!$ow$)+%+PC2sebfio14||`^;SRZ>Qga}wKk(_QEjCN z;~uX#O}Y`>Fp+}MA-8VOr~33B{bHt!_rErXm79eF<`_h-cqf_yWw7qCV-F&}diKGP z3&1Uf&M#C==7tw(8UGJX9F%kdB7g>vAJ#zp3PDRfS05d||x!YN75 zwq~)PS=0SI0E}Xd`q78wL^y>aYP8mWoY=_TfM90?zn$5%%_#K7HXU&rCZuL;NKl2stRk z3rnLeK5lnBMmUq6I`^F`Qa{Q@pI^@dW>NbtG^pm41?9~<6br7aAh9Y_{18Y_E}`F$ z@&ro$${jDu*1hTz?E<}|OpkHh_ffI})5beJ_q0)GQ#z`Y09KZ9dH#b*F3z$h2eYjA zkLtBA0>&<+MK#o@uf{TQ+#_On_g^1qo+=Pk6s}O1eB~OYbS%C;M-^&gR5V!s=l5?l ze)l1$>;rh&&{JVIf-5Uy6ML^Zp6k>2&#AYULT&DNptuG&dl6`l!%butfHCyfopP+m zw&1rAsYvP1U8L3Qm%?gmmC~&GF|gq4Kw}g`h;v7tImRRVWn{}QKSoh$xf^6-@zIG* zUriC8v_954x;Q-iH{Ba*J?sKcVH7y>KrZ1*yC1VqCgJZsVb1WIV22?ytoIuis8~6U zmgaBzn}bO%1uIiW|4}&@|8e)TCO)@{haA#SPXGY7y6BIL2}8dqO$6QWFtH#LKkLpo zFNEKJT8HV7aGbdmX*Ix|sf=ToOn(`VDT9w}JRkTClxh;OeuvaC`WJl+=1(U-B7utg zZqr4GgL9Ql#WOjHg2l%BFBjByi~7R$CSjU{8{6$nqEaCG^w+aVHnH-}?MFs@K_YUNg1)6Aa7UG>)_d7iiu z&?96#5%ucRtif~tNqyOasASE_W_4+Pf%;27U^%ohu7sT8cE5JW($kdRZ!muMc$@Y{ z;1oksIeTdD8QEhsn%hAQg(%f2LgPMDo&9TnXO)Q~^~bW$ahR7EU@~tU zbP^9KWI5-Ocf}Ehxq{ zNKGF3MgRcLX|02-JH-+riat$3!k`;4T2SwnPRF|IPzopIyI9*#Q=p9Bw=pu`4|XEm zKqRVIJG$RJ4=veD*pl8j8R5Y`#8{RL$}*r58=@~|GEG~c!+L+68uhHI=%^CT2)w?A zLy1ey@A@52oJ>L?4D{hJ9k%pjI_x)yDVqmpvk zrg)WHQ#u#Uvv2!WPUV}aV4Jeb0i-xhn-8_9toimTQ!Aq z+7fp=Qg@(nG+o@t^|hjhGN%8;j~Y4GE7pW?brKP;dCyy_oXCKQEn8T)tlBvrTvgTf z4XH>^nov51w%@lu?rnch>oic{T>LTMfwpU4&SkH;+i2qVvMr>B0|#b+qdkm$1rY=HD;x;;>K zk+^3@x8+zQm{s}y__yTQw3aM}jFXSGt?=b#U~`qIvru@0$M60!w4!=|F;_3E^k}`qtgqyf-fc!%?ZT%##%s4V#iNL-h~6x`T-7da?RQKDhdZT4Eh- zHDW#5`r_F}viPQCSiXzBI#L%Dc_4KAe7k3dD(|K0yv2O(m;*1BBsPvnW$AdY^%M@$ zeVtgK>hpN$^W`6TkT{OS!*UuJy z+3q(60;ebe%{xyVGeJ5I=K-wR=NglfRL1P3RPSBTOICQ%@Gnh=z$D~eW43vaT-{Kq zbSU<&R-o8F5Fno_7u+8wIu-WBvXf(RJ^vJb$Jw*mw>mUJdL#lUib8P;<~ZU@!oF}( zmNBmT|1#u>@M6`cUAC&fAmuQVIf}06x#sTipL+SgmD)O z6IhZq3ztBKU;S|WeMHO0nH26|XNI4&m%ILpw-XBJcPdb`S&X5 zb-RafkeWzyfrj+D#br|U^fCHCm9<}?)8@-CN_7$r_NZ1HvqbAG=t_UzuEIEa$>Bg> zc(vrJ?-al}GsGXer|YN+@4oaYFg8&;dn=aVFZ~ei{uE2P-uQ6<5P3ksOfx9vV0d5c z=Bx9yeB~!OHQm4?aJQvDUCK}LJLk&~n8%L~8GliK=LWO>=*G=Pmo(RL1sMlNjwr6W zyBm@Z2kfoQU40t<{w^H$n9Zl8aw_hK&!$-Q4lx|xqsS%m96<@!+kf}VjX6jTv&&fo zbD;;xuKw~`O;a{aDsEyea+qK-KfOysZb9u5r5<*A!m=MRZA4kZm^>)e(D;McI)8ugr9ZV=h__qGCdrTAB29|$S}Xg#ImDMXNyCEl zPsAzUR4Q}y_%k-y!lH7+=wde@yUz;Am36XpDE+Ow1n0Fr$@E`;e)c>Y9GO{EyUG_) z4pj#AVcUyc?`rT~XUE8>52#nAwNu*pLP-JxYgX$Yhf}~aJTklp?fa6vQ%G|lADHJE z?LUxgq7_fT%_!-Apg1v`U^mAJ8IurG4dF-$C^?C)>x3f{{Jci4uuuDEc;%`YeY)$e z&NqfIsiL2leRm=vK#+tXZV!fLx>@(=VJ;>1W>Ll4cH8liSKHNh;+O-$y(b6AWG-xZ zA@bI|ZKR2s$dcUJy?K2{>8$h{RIl~@=`5wzYjZmfz-H!!-tQxOzP>CTwFg97!|l=? zwn{)62%GWGT)%SRnV$og=rd-j!lw+k5F1!0Mb*j}D~BKu`os!&3g|km?Cg&aw4teu zU~r927BO{*9$^b86IX^*=Cef51|um$K-;D#xLNsOIy&B)D1yM~)Z1z2S7WBa! z3_IQVSRMYP5joG~nHbjwsAG#28bPLaqjzdg72qe7b|uw}eC>GZx3j`fa#YEUHMua3 zpBRiT&F=DnpXFYHtnk8u-MWkJ;ORtv!&2~OT2JOA&qFi`rB@3?1Y1N^ zc|}s?`wd*I*De!x0hQtw^%T*Z{dg%QjvJll*;n_-M-0RX+4H~mCKYV{CWUmIvK1Z` zDV?M00%&q?4sBu@;YEc})QQnLpQsyO)lCTYzLUuo+mo`eJf?Ryl7xX5=#U#7-h$)y zf7Uk8NX_C4uFl5P%rl?On4bhhLc$%F34&pHRnqY*Pl^DSYUgjSvEZsN6EQw!_Kd~a znbf8lZD`3Wd46}}h0MqF+_u|kl6eG>9Pp)b7*xOg3)W5{{e|rc+_5zgAPdGUD{^iQ z6b7=n>Dsa--(XG&Ny(85Co&XYiUD*QGEm7XYy~v0eUWrf3Lh&{<7AIJ##Gc{r19Z= zOB4A}hVtYl(?Gci*(kTkOl>wEuSfH0<$>)e^@d(92ri(!!)B8ikvcJm+>zZDVLpI9 zsa3R7|C5PVC85TVHYttX!w{~1xO>&ZMt&x5<c+Niiou!RY!yAt4BJn#pb>7?wtMsEUFW&9%kate2W^*6Q`DdlBw?T5JCCTA_gb2NM`AB zGBQivUVr2xv|pCl>t}A*sGh_RBZt&m^>LW>{UzKsYB%W@jaP%lhBHS|iqIRyP zgIFKzF&wn|;z#LILDiOKVKJ*Eln{!tB;h{J7*gnp$o>wc99()6^&LUzZPyOjP)xWn zOB_mgbLfr6RK8v`_}K6dw9GpI8V|S}e)`s|>9wy>&6|daGFleA6k2(2fU)th zsasPDQ;GKTg+fWlx2I|~e-{8MVaaB-)?l1f$D5)O=FJe$eknrb{;jc1aR=R3J#;r} zI`VDY3eMGv=!Ih~2N8omudq>ZS`?AiNwTGwD#5AlT$!MHCV!%}uWRU3D()zJ+{|UO^&7jd04?j-xY#D^($y|=e)oZ!V)>E%xz!<_}YEd zeIBB6fDMmcZrca~hLRB~fmxo}5M2u9!R3flj}o}||KPEbGlOjgj+K&6nKP8a=qL$g zQQH&C3R05M+fV7D5XTMbzwU@~Al2gshxpWbgJkmO!>+P{6%x!JPo?w$|lknq>GN`@6zJ)c_|)!jf< zjMM4xnkyT7(ACZ2nuE#Dl;Ku+A#B2iq2Ap2Oc!~t?S5!{8W^Ac_J`H=!Ll?vqqc$W zi8u0NA(VU&1*g7fG;M|a?t84UnN*%&Of!m=9XZ(YJ)3CQ8yhF$FL^5ZZsqn_=5r36 z?v|g`PkdXTq8*2;ntHk|i!y5Vm9eSZGo{U0h{_@jZ5p!U(^)|%Ia~2a{lnA~JnurF z=#eHOC6p5~lTb?k3o)Q*cReFYgf=hk^pKks^S8qnB-B&icZNkASg<06m&+~j*_P70 zTjQ%co~V^sXOMwZ0~Uqk=eASUScV<% zoti6R&oF`hFOzY3ypU1V@y$jR&Nx(zq69#Ja7@XCBw&jUoBtJ=pHLH?H(#?f|Fg1G z%m|)OjM^Z1Tl!WIeVzbvJU@o{x}keWdc60nbhK-k2H<64s(w`)BdN;RnZ+cgdpAp% zgTj0*-dF0g(S4QZq0pThS6cAW17;5;CENC*mQh;jgq zqop?QTBfS6y1dcv&D|J5cbFcp?OJq4YF?7DPyG+{DfW@jBh?4Neu^i3G#u{{BbpCwW`5wvQmpP$4TUxtA!Im+Yr}7v9Sd=EjGc+{{f0&P z7)K7>ZU#X7d4AGboy@=387;8Fb#`UN(yZb+$0%&CWUw@?hJr+IvA@7rYRZN2u}I%n zhCbe8;`nwYLO2l!3aW~Xlfu~UM6NK7aA3wGuV>u%Cg0vy`aNi-@&${J&H<&VW8X(! zv_jkU{TnJa3)RJiAMpcjp5W-d{QY#A_%eoCeoc=TmHC5$I~fM<+4#XWG(8J)@)6rS zk9=GwUK{+|)s@D3#*hstk(N$045OqRAmgAuDKtEJ-UZ&Ot%0aRHnb*t>Vmi)GV-a5 zneHvakAsfNG8`mqNzcQ%9?a1l))zWT+f1v_W-YqR@^^G~ms>9CI4DB7zK1>WH&N`VI7q6&BFv1@1I)D`#Llyru z2w|L#D4Nv%Ey-d1I!c;M8Od|3_v<>)Ra_;hu1#>!M@`MekO8O6RSf--Q?){~_6%b#VPT!{oEI1wOH6U=R}2K^wtUc#-*qf1ber zJOUNUcD?Btz8LwW-Qzy;M1!F%#8JD_pkszpsmbnk45N*}wMwNbd4@KP~JB2JWR7(H?7FB%8zHFmE zjmk<&48WK`bK}h_e8+(o(PAlv`?%Q97tuWz8)YpHSn$cT%K6uTV{@Uqoq4G1(pK&9 zAiEdMPqAnx42nbX<4FsO`BZJ5kOayOeIxbrAbXbiH1^Zj6GMSOI*y1|(|1<$tluN? z;W2n_#>3i?UGwB-F8;*}lMHP=8IZ8*E0DxmIQP5vcg4s{OHi!hiD`^MN5Ug@c~Czs z3!+$tBCrKw zo*78!eaUYN=VdzmeGO}uNj*~S%oy>eJ)vqI+dGrsk2l=Go?~8`qCVDi6nac9mT$y} z-I@IQSX1b+p+I}HEZh9nWTIBg9$wyltR5J`yGdsjP<`NmLY}CDI@dXv#Ww#QCS3}* z^`udUa%6#|Gc6V=@B9AyC<@#&GhX=O+Ez~Q>i>UHw9{h)w;OkCj>u@VePs0(`s1l$ zC*aNL^x4#eG2jLY4K$~C_kJ|j0o*&8wskH_^P}pF6b`JD7$j&ujK9q4Ra#z2b)ZA{ zW-aOM!=9z*hvqPXMp*3V!@K`p$HRCcHMMzEQw_dDk@cPxM6}U>_~L)`0)R6sw{LWM z{u+pvGQw`u^)_u+)5`THeu=ol53zaK0;&}P|CtnmIoIIUX$0Bi!DVoKtGXZ)M7IHT z9kD&B*&HGa*Dqr{Yvfi`+R#xb8_|AOSN8!A*42Aaz%-p-AJw#Bw;3fghSBe-;>1i- zg21w&gKTKPYM4p+XL|R5nhebhQ(cr0JHj_rXuAfv#^1aUdGH{-zX^B;Z2t5&2_v@L z!ff12bu{lEtZY?Jye{HS9%6)Db3B}Mu=9<5BWjYCIU9{LDzqbWr0*>%uS<{Y?grsJ zJv>{$8?f+SEgj6yI4d2E0}e>Y?>b?%oOWdB#s3|4>FrRdJ>oEM9I&y9Wh`KVcCpa^ zq-1gEx@JRDrEGvYjQqZ>#7SOpBw|QP{Nsp6p5ynBl!LXF`GSa3j51@-@ zyD~?=InhM13iU+|P#Pvg6qXjc=0A$A+Zx4_?^$MJ&d^?sCRXx|n%e~L+GcyruA2xg3(Y`NdQ?WFI zx9VyPa*#j(5%@@m6HDWO(ZEy&z*lEcIO{GTaPIDqw8Kka)2>|0@o_XE9l~}Vw6i>Q z3NvqTS-m8UdC?j^uj`aK4lD6%_H(zR2tv5A@bX`Kp}3XEEsIy017^RHZgaZcTaw-z zqq^)F!ZK8S@?v|1y7I3AmRc+D%+Uo!3_RFP$tU`D+TtAUcf;440+q>ox-0fy0Hboh zZoPUUS`xd{1RQo$U66^(;tane@OM0b^R}GrcHtTzq@W-j@?#?o+abH1xL$p+uDM7= yUDffm1u|Wt71}k?@heuT;t6Hs5&*OzuSGc&?{R-xLY(P=oWnq%_DuYLbN>U)%8YRU literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-13-DSX Architecture Overview.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-13-DSX Architecture Overview.jpg new file mode 100644 index 0000000000000000000000000000000000000000..53190524772cf2454f23697f948d33cd2558a2a9 GIT binary patch literal 240714 zcmd?QbyQqUvo|_eNC=wX?moB=kPsLM?!g8E!QCAaBuEGhF2UUyg1cLA9b5-@7y>~8 zxjg6nzUP&D&RyUA_uihhdbRA`UAw!cy82gDf9C(J1Mn4O2_Ox? zKto4IN5go`FfcHlU}8VTd`x)HpFhJUzMxQgSLP8XB6H zB=n5*)QptWG}M0$g7Ww(<`YcZr%!RI2?+?P|6iv+odANTXhF}6(NO3Cs01iz1So%c z0aSkxiTM|~e>sT1;< zILAIEVtAtwNX)Mpms2%K!YClE{lO(DzKcl^{LVF@dV^F*SVqg#%{@4G>Wn#YnnhN} zEU%{f{F(V@xsb<6$Wi`R`kMs6KWKRDlO%Yg4fT-*G*k@K$Mr);d+Z}XdqKyG{!*Qu zPm0hPgI~iW_DvvzX;l}IW>5~J^oPl_Cju^U#8c;g76I63kFOJ;5ddBR&Jpz7w&CI7 zPs0QLnPsc}RC}$)sxF(0yO&L9I4NaOtW_$ircsU+^hFJF8Y3fPB_o3_^ZNfW>qe!g zrKkTp3yS!=_|L}wDly{|5TFqJ|0pA1CDcNVSbUmG_&a?92!V{{*^nOOc%dG|_lAI*B z8&ncKz%491g(~4-=cYG*wsoe64vVl56Rt%z!O)4b=*0jUB!*I72KCQ750aH zn)He5F>Nq9sh`taI#I?gNNLM|1>B*+9iTX=y7d@vo6+&I7t)uTeK(pp#_TRaCgJ}T ze%mea^x=tB^5K5zf)u5i*wYgE-6Tx_X(vShyiDz^5N?Km#WwDLU6}#_cb+i@c`>7U zEi>#aY*D4-g`Y|7Dz21nn73%_!xF{|Dhgr-#JLX!@KDRobc&C~X{VZ_wQO(O6kS_} zFO)~V&!h&9e6UlF9sA_jdXi^eZu3Cy^l;j(f9&3KTNcbv-}vR5mi+N+Cu@pLOfGa9 z`ZP*~99OXVX+*R)Gi?9(!3(D1FN<}`n5fBHfZteSTzD4Uh6CS_ip1Sg%)QPlow@&!6_u zRJReD5$m%mtt1@XP46BF@4r`FZn1oFS(sbn3_Z-Rvu_n`XdYxJ&xO+DL9>q_x{+i(DBJsmhWfTQ>5h0+4@MEXHxc_+ zs9Zyzlnp99`@fwG;o)4s8Np=o#=Si@i_($=gh-Bw@3Mb@LrrS1)FB53ensnkS^*C4 zgOH62xx{Xl=US{oeCY9q*5$nPsdvhnd$4i~kS^PN7i0S8TNv#ohu!pVy^VqaJ;iL! zX8o;e4s(C9%2JqLcXs5-73c$1aEEqAG?{6aT*3{v%9LNaeVu-f+G3Wb>vmDFXL8I!2Q9=F4Ca}MbW7!cV@D#oVjpp&^WbXj0>gF>x9QK4%Z3Q&KrWN zNj=f>2k^3yK()wGh`r2l40Y5{j|k|eF-}>dA-ZL{Pb&Hr6uD_yz4GCPPnW@6i3oef zLb$g5d0MHlnKEXcDDNmkDBs(G489`Tp4yN%6<}I{027kaY!lw7lTO zBG)unmC-R4<}RTW4Fr;+H9n>FHW3nO7xyD{K7Ky6xi!^Q`X$abzV_D#91*kF=IUw) zGG@6}FluLPizsqa2c@L8J>Dbl`>maQ0qyLhmNr;pXCR#KqtbTsB;$&qhfYXDic~~KLfz{B z6UqQS7CQ>q){34!`=46Xzm_<1=?`}P^~`?yG6bD9f=-RaP4femS|PhyO?|A|P(+|c zEIJz^EO)e7i!c1Epo|Roe_kG919|H^Xwi(;I|_)szd0mGqz6jQk5CQ3viqR!sgM=7 z_A=M@^Q+)XO_W$d653yl(E#4&P2odr|F}bSo8~EbIYUJ{sFxY3E2TJ=0;f8-@x1%U z$2OoLEi}xdyyZR6L*G)fIhWMtd1SN&AAexzuASwmkmxCnZSisHRi9I3#&?Zr9Wdt< z{##S&29QI`YU!qZeHc_|sUY)`$>G~iS&do2bOdkwb$@Mu%XIKGh-gQcQe-^?x>Rg# zDLuVNVQ{wl_PqLDUD`WTk?1xf&PQ7RY{F~tR?YexjMX$Z0$9HLHRv5X5sDUb3RqZ)6iaRRD_8b6#mTk1JD&(({U&W zmEtrRJ=!7&75&`Q`0C^!5u!d(+bTU-Z)-nUF+GdrP4Dl!i0t;2ltl@Q15i05TI=N* z$d-oLM*OCw;U1Z6d%9$OYz1vK!l~a@UU$8`DY7l;L7_0lHQ5Zw*)y~(wC`YEQ7*LQ za^^eNV+_JFk|Svqos4mXLOfL?5M+|pBmQa3VQM#g2OM092vdiK!_RJnAHx{^%LDsb zchPb8B%GT2%PVeP%{f2G3l|hCg;h7famoZ@lSQGvuRC#gBTrEJbrKg{UK{#?e`j8< zmQ5vy#jOZJ*~Mnb=&%&fhqp0jnxanRymi+Ye04i3Pw5JjXLkuJZ%RljjB zF6en@_aW4DXsFp5r0d=KHB9oZiMAdwmY#^b-h0s13UTzfUyn`{RwS9@tLTq*|{klg*~!Y)=EdUnAidSWy%X` zGS~Di=rYPd|NFG6m!z^fY;T$QdF#^u1IO&xHMr2*&fZp5-u1haxWpD$mNwD|At)#i zKde{i{lPTz^ep8E+HI+w(~P#w2Od>Uylw8N?IKJD2GgJt=Ki_v=NUL@Uy_~|j1|<* zXu<9MrbS+mE)&E@{np79$uaB%dKR-?V(jY9h~;K}i%BV2T^}Ux_QUpGTW$T8#YAno zd#jb#Zp+1LWoX*_oC*W6^iDv#BvfR6^)MX?z;l)T9RT1F?p`df{f#r#T$S42S7})k zM8D5aKcHqb+BGaH?funLV$M$|$wgK3Tx}4zT#~b2#`41Jlj5hDbk*V#sKLFi8}YD- zX^X7vWN6a>WzKl#7;X=(yY3^%g;jsa6=HzX(+WbjFi*g%WfVj{ju{sREK#_q=CI3BY%jl|qb~cAkDfHuv9sF{d z?1!5EJUzV~0fErd@Zx`nG1gkM4JmXzkaFmy7gTQ1?(*_DGT@h{rkIL`k7a^gL9_0u zP&Soote+Jb#j+RmE1xELmK#J-A{oUA>p?FF3CaIS*?D|S0ZZ=T zj*qhVhQQoV?%!hH43!&Ipun@F^jo-)U7 z`Fk9row{hH6Rj!Az^D}^<_uTy#BXKemSNy*VzTHm@?ueTYMFOEqZOA&0m*^r*e^X} zG#?w+)qm@Z7xZSjeg#*Y=*_62T|@J_q-S1E<%w7xAD>u&-}8FB^99V(Yi-Z`FNifNoVZP&Ilg?D(t#bAA)8W5iTU+og>2SLB&L(nRE-cA? zL6th@0@?Wk05AOk=y{iJD#9ufW)`6}yF*^KIo6@A!z@#>o@iMT5}(U4fZgrT@9@dl znduqFpH)gb9Ut4)U4oaBxR?yMUa>0nM_QUo2vRX>xnZ1Q+fBV$e!G01YZ!Y^^r>(0 zV1=beSxWz&Z9ggrQ%cReB!-V!1RsJO=7O>ZUk`=Dg(|0()cI}DZtm1B#z%KY=Op3r zEe9EjUU}wBQ-1)i6-ALIRjN7+OJ^)kC8Us*9kYF1_E?A7i@EF^ky8>onQvAXqQ0cC z3;?!1*3DeYGL~$#B~R?9`Q{SZ*$bMcT+RHXT0nbd(zWynd9e!?TC2xHOjfI*gQEIyv9KA5#_n3~7H3f%UCjEvU*tRDOyU@S7&$3Y5X4^IxJ z@nfvN4w?qgbw|8;Dk+C$XRNMW9IDIg#^1-N!LxDf`_@%87SpLl=$NMTL;$sYt&m$# z>$90E5h2v%7e0@sDe6}yIn&m6LM1cx5jTC@XV9UHlcSD;z?D1rbN}}GZLJ|icPg-< z;(!M!1Gj7I*^~V$tB5vjfuD)oMWsbaWO*WSqU8EJS@gIYhLG|(MLTC@r@GlGFwcl; zDkg8`uNRMVP)1zM*~6N(ZUyUu3hVtyJlxJjPdg8s#EIHu_`1YnlwTr}go)dt9XV2E z=NU$Y*#6SS$}}`pb@UJA_DSbO4)uGhPIKBC z?L3=X&(PfFyK_65^tqCLWPhq4sWpJnY2I~-_C>l)+3M_h3ROM-g z>wIYXtSTDpZ&L?>!EuA&_GPfxyl*3gAyfE|f>8VS?JMoBmpfTXMjyUJarv%E)8e3& zfSCKA5$oR-aVN3+$e$87r;;9$nicX3C)c8pF(8PU)cyn61HXf2`ZO2smzczP7kF|o zL=}uVBLVIlypqvyr#Jf+(K7QJ54R7uAvlHPS5G< zz8;LzmnWSNN4~K$iPb46%!FA;DXKDFyghj*_wpZ{R$j3rZ{rJQEoRm#{Eg?jz5h7W|8w_cWH9ix2*|*Me7VH+ ze;@e&4l}~Szm}!^^M>IlqI&IuVX z@zd~#f4u|!8$isJht&H6NJ^7?ui^LyU5`VCFvrXMAgz~# z&mZ9T&g#m4Tz++__f@w(H7#xj_)^Y#6x0i4uxY2CTiAD) zkQU9)HCGJ&;Qgh-F202kRk0qB4y2Jie@HnWYTnS}$)1L1OVijrOtE_o#4KROL-=r4 z>FioX5LbZ_cO^gzkYRPIL1fKzpM+uv=jcb#8z!rzRz4JJ4hH_d6{;R`4jkr+W?d_w z#^NCv<59Mmc^yxTE19XhZ93KPL`HRG<7V-ZP&=iX!NC!s`;^hwl4IU>dTGddC))Sd zR#r~5-Y`XE2?7_|y1(=>WttgUaa7DdEqx`-EEUT+U!xa+0Wfge}`$f-~&f%<>Vp7rMh!2FDNc#}K29#6`HWTSb#T#UQEg z=gT;0amzlavGo|9k~<{p?YXatn&EYPN#=ovSlf{Rp?}`uZ4Buia@8(57* z8Sx3%hI7%OegSB2H+fN=RIMK-M!Sx9)}udJ4-nO|YYuh5+Y^ftR(6{t43c{X(bCUo z>Yez9{NklCdPkMJRy+%~>-LwZs4^dhu1)fMt+8=q#AL$=hGS&;4^2EdRB;6`QYRHH z3MV~$nw_7TT@pdR*2i|5j4_Sjk}+;3I*$pidPzv-6p{0NRheG3D2Eni{L~oKWTX4t zdqL@;+<66yF1&8(v4w}m9lkfPf|gVTs}mWTZp1>*ajmR($i{fIsI@oilN zVs<9#Mw2q2qMW9k;zcXX$6~NMujw@=dlu(vNGTBhsV$HH&54dnQj6Umz-jJ=xKT%6 ziC2_o%QQTq)Y@{pC`O>x!+s$f^M!NB-UFHK7^!^sbBthZ`2OO6=Ja=EIYV_0Y*iCl z>O+++B0p?PygHmyBfs=IbIFoof0M77tfdxcTtHjM91ZGDdgSGb+c8a#J&wuL)v%rZ z8@Nl%^uT_dc0ZRlBM@Vk+o{Lp?MaXh{bmEHo39(qx(;4;PDtD@MxpEFd3sABRqq~d z=gzf3M{!ZaNkH4)lHJ^5i^;4agvf8Vd zJq*DBOP7_HLo<^qF43QE+m1xLrlNV!_yj%N<^F8XzImhQ(x#u68oFW%&7UF>sn!WY zq9QzLQL=1uOiR{J366b|Y=y$|LSbw2jP$DDBmGs2mKTB*j4xwnPOEFDnja!=kf~4{ z@mqsL9B@JH$+v3Z8rM``HQ)I$Tz$DZcThoQysEY0@0mphxM^V$nR)W0{sv>dECPbq zoN)Y3s^xr7M}u9PjQaDT0Us~0*)5=rO_5vZu3uYNknFw5u;8wBKsNq%9_~8tOp%_uW8mos*+<6yb~7{ zp=IIvDy2Q$N!qj1h&fUBe!8&>6>?F1$n5%wlZPZ0?so4-iv6bRPoDNPc=i}o6YctX ztftn91r;Kq*1JE>cWlo1yKKM9K0P35sbe9z_LeR>d1L09neJDolSQp?;%twELR%&vgz)yXhns&^#v=byr^xcFDd ztn%v?Ywy9FT4yso&zH~&qD8sUtf7_=|g6YFXkw0!oDc|lgp!wmxXUcV^ z3HRfjB4NyI>D@GY7&~I_42)<&>mFAX z3h?ZscQBlJJI@wnas6P8)y>aZgEZ69q_MrBS=d#1r!K=*5^`R1uhZIM$fW016rb z>d?{+Xq-n%Qs`1o3z54>fm2lM6SCo86%-F&6}#)r=kn`ycW*VEN;emDDX@vXsBn;} zHhyG`*i2nZ_smZ^H!Kgm27AiOLMXLi}wHNH)IM7W6>ZN!WpCz>v)qerl<}TQc>T2y_+-)`-@>f57H#;PW z{mKC1^&Xs4^7*)kU*#ry$5lGOc{^DY7iF~b()k#Y98M`W{mU`IBei8(M^kLthPV56 z9ck(%-rG_5n)3QCA=NY7$Gd|aR9KK%os>DLF)&qMF-1L|SZ-x|5xI`^_A1Nit&6Hl z^4IaGlX0bDvYQxa_pR_o`-$;IFUsB-dRop+qU*c>pHt@N@uNh_9qPKb{cXq+i}qn{ zDAtJ9^!s`cMH~jNi-L;2SEWmgiWe^!EpjI7q88B>(d`37APD#&eC#eUP`WDpt- zIdae!erV7xKW|WTmdpxsnXawiY>A_c)=qD z5GPL;b@@e7f{;jH1^6|(8F$jaPa0vBaBbmUeQk=3&+&@cwb#uB)x_t$Ej}(~IcU0ZG ztHYVPk(bsJ)GCN=R~tk#e5p^(U*BgBk!adip`3v@rba(X_0L-4J!$gd2suxAPY$vX z#jiEP>-Z2Y-Qqfy!92zNU*qBe?r;H~tH=Q>vLA0n7^tb?{f@t1)Gdl^T25#h*)rue zsrwigo7unkPTbb%Mtj??SO#|%22Y>a^cdi*hKv_BPpL9tyI5GI@Q;Z4EmBSA4j!=8 zN^E}dkYsH!Sj;KPtrvodWbnUEympns6r~V3jz$emwT_N=$51Jqaat8bpIG3$AYI}| z)BhD(1#?Y*D}t6rfin*NI@5|(r6}jvv3D#dG1MdIrUeI&P?g)eHYqM503t{rI#-_= zA}RJ`6>?qvSW?DS-5p672NVWQ^ei)r-H#rZFxaOv-npx4S9<9%-5ujTzA)7=Eef! zgKhBC?#sfHCfW$pQeHG@_~c9~I} zQPi=|EnF9O;(S}SUbs#a@59)ve8#wpoQsdoG&R(vRHT zl!pN)Y|=Cjw)0A(V!qssyB0WGp;5%)yY7n+WN|mFZL``_ zSelGBH~?!+Qp1NsW?8_WX->`1zS-rw$9Pe(nengzgB)U|{+@@pEY-e)>Z?=ZgsC+WreB%-ZQfa6&Ft7V!17XhyIx!hth@Am}yt2=? zQgG>t7$p2=O7qJJeq6a?x->>Y!`zi0Rha+#J^epCBmT!Kk@cs~Jr6M-zqeHkWHK8y zIu^K%T5O2st5xpNrxG?WGY2V+X3YC#$3t~|awUx*rs()OM2O7tiB#V(;!KkLrJ57BwpNR6 z9VN8z&#%5+d%pSgLWQ@Ur$)%0$kgoW4iq5)+(>6;Y>o(7 zQgS}gU)uRb{xPXuNE5ObI4*Kqvzr&rQhdNCt|M1jZH3v;(!B0oefxpGcy34Ty|=$u?vT9z=((Tu7ZgG3swBUF;V z>NwfYTDS~C4t?2K7ocZx%(Vd$(7jgqQeae#4c3w=hjrOEWja$E>y=T;yspm5cL$kF z^FWez>UU~uR;AX8RLV{SikV}={S%a@t_^oFA0FiNKTuetmBI!4VzpMx-37@cl$C{i)&+MTcrekJ?H0k_atL%fNLA~?+t?r@-MAEZlU6V zF+qs;PX+3rqmJ;|=iBBVQ~e>%@8Twyp~51;49~3gBTYunheoB5t~up1wSvU=viok4 zKOIW_owr`pqQuVG;d?h#1(18}HE?mQ9qX(TsutP`j)u1X5;VhC=BK2Of*n7Rw;Dr} z-8%iQlOR?UT5qdO@{GQtUP13z1I#p7V;Qr@N$MP=8# z$~iH%w|^)>de0%dFpqL(c(-Mc&t}CiLvpv>b_C1uMezPu)nhPo_JCTAmqMKQx0;wp zUR)MoACw6shN(^JEV1smTCXkeS>4DFax&VdpB@{G58g#hXxg+Kx^LS`J0e{Nb$*a` zBF9EYovcoRmW{7X46n%i2?mB|>kw|_F5Bj{a%PRAj08QDnBStdogG|!S9qi@Ako`2 z`Fn1jL|nv&5PJ4!DMP7V18*14<0g+VX}J}hrWKqnRD{@lY?g?+J?ENju%`6(UxOG? zrA|dqW#@l%U(5~7=Cr zp#s;Om_C67B`|!P)=~7u3+EF|4y%b<*HB8H8J9zadw*ru;vi8&RdkcqprOqr*UKp+lWFpDem%_8JkMIx&GE&hLqUh3yS7p4jG!P{UF<81ADd`6T?5|p zXbcQ}ZV@0j9`3cRZ=KB;om{ESVNVF9YJ%t}m+7}3Q;N`MPK@z^&-BK>n6^2dhl}fj#|r;goG0$)h5thpWC|5 z^~SxMB(9qlC)_yBoi1XpXc)fKI-h~OfA)eWFz8HG$z{9ZS6tN^J83OfS=CbALt6Wi zfm158$4^5TQ*6;e(tSP6SO1}~A1NY`OVIpm({aJqWQIzw z5A}=e7SO~?t!uu3Uje*pEG)~lYjMD zmJLN7n%>teNxm9ReSLdyP?~mM(VSLX9>tcw#MNBUsjr&ec-gVsm!t} z%1c_)R1}|(A6ayzjgB2-$a9%0wP{m!t5aKX7)N|eD{63&RCv0cZ5Ee)GE4o->v)Lw zRO$!kZ=8I%IH=YJhrv53w_R@uV+bp>mM`N5CkFR3@vaC<->JhLwB-omemlR#< za-m58#tqobO)m>vo6pS2zW0by`)?Vnt!o?5cVU!pKHOH4eBt%nyn>PsSt zMJFU!Df`yehk1TAZ=GfZs0OdN)tq6YJr8U;x@ARkKZ*A$Y6~$#+xZu3V_cp|tOfOv z2)7tT9=p9F=GbFBhLtA9%yh&PT#VugxPZ{gR2Hr(lD~LDnT6FgwfAfVBQ^^co@2+f ztLT3*3?5w@d+6prex7{pXmr-TuUh0~!Ij8oy{JdQi0 zyd8ZlEQbP-1PIFP^;Wi$%$7*8kFcEW3sn!OJmqF&V+&)YoikTUc$FBO2G{PLVla;A z&mN;W%x_KgILoN5;OzJE#)!5Y_~p!KK~6J&T?yU@q{6czcD?O=Us*^Rq<$Qj_w`fo zq`Jlxw>eO4JAGGtNwj@ij=n*cc2l-(hH$Zai&JbUZSR}1@=JfJ(MXq;mO~POsI!wW z_JgvmeMQG<;;i&bl%g3OF<3XbG5NI$A7lf1Mcy@{;^f1b{Ol(*cuHKm+2~aKZ5<@9 zzLzn`gcr|uwcE^@MmIq5_1M-O@%Tr?lP@z4(-7JGb~)@MHpZabHSxx@$pR)AU5DQB z4lcFnaxc2=Pi0Q|vifyvr`y)*W};k9ueG46T;^cyU8 zs8ZKU!?oqYP7y4(bLXNB4nCx62AW2sa=&qA(NV6X>9u+8U!5BjBwIJ?Y&yS;ZMKC5 zO&J;(FLA5=#Bi9IoVN^a}P9Z4MEDq8?boY{R4?tv|kbu z0!LnmDFy8|ZHiJ*ryK-_p51NjS~l-D!J1R=bR7KK6+)N24JkfERf*To^_t;?Wod`c zwxZ<TWJvtCELPYDE}Hkyc`rsN%&49=)-t~fNb+V_&~1-Xf3&fhkN!NX1L zR`x|pMbiSKkiyRmCLwR8lknp0^S`zc&aq>adbWR7{8^NwLxSy%OfT%KJKQ?Y=0r?d zHQRdcJI)b5qYW90PW8A>s-Zv&j?8K-H^dvvNu*RN`=0i5O+Zs~K(zR1ap6nM3$pel z07OO1i!0DwTnxcYPz$TehmSz()y@^AeC!*&jUHo*F@-rqIPWzKtzohB*LNe<2?Tys?ibWDxb*D5X#H(z(WH$W~*Wh-Ef`L~56s+C1 z2UKi5)HH^IAx!SgE!yuHUH7d`y}S!;mm&S2AHrp=W0*mCeTUE(MD%m3x&&d6e}|Tu z3>eH9=hFOI!U6@s=bd&+rz_|fCr}^WSl;Ev^TqJ(xvV)Y?i^4A{iico_bGYiuYN^y zbA`J+Ra2gJi%&7k&Q#`xFk8IXP(@a9cLu=gz7NgJ|A;aCv&s7ge4?l3e< ztre*P;gPo;%F7}|#WEyfy3A$m-!}4H0}^;zFe5P^YQ zI5LyNNxdM*ujtUF@-N3(EK#_N-`p+-3?*)3&zr5RbGJ-;^%i#c(<2>x7v|TwMGwj( zRI+BYuk^d8)xm1lMtTP=n8jBF!M`eHsT7T0aTKm%?0hx|X&u;?_|(T4&xR#53V zi^Icn|IIDr->0Nq%B4qz`V9rq5`r#eq^MevIdNZp5yd>qh(BRF*LucM3iTUQQ;pE4 zSFPi3|8zwtiO6zRBXp7bqIMU-@f~#eIz0P}H|@|U+mL!0@4gnBWjAhDGdFsYb5MNjSaoydI#frWbGa5HY@D&810c5 ztK~TR(^P`-P$%*8=D@@1zg?37v6FHsH(3!utn_$~r1)tONWBcgr*vQT@+BmAveDB{ zXXM6RBsr++C`c4#s|=QTB2TwHcC?H#*Gp773s*JJ6ExR?rFv(Y#mP}7i=@YC6G~Cr zxm_2d%E`!JU#>ipc|ADh+HAHfx%TqMg68MMQ_f`tX%U>mMC@)y8`Ur4ixGjJ_Y)k$tY2<372~)1i_! zq6^~o@l5%(TbA?X9}(fgt32nDr@g}=)z!yx^0J5i+<7vK#z({#BxYMTi8;F^=V~$p zW4H@<(~?u!;q)pVV}y!rMTi2h8$(?7|8{5m_s#ro*y&$m+x~tyTm*?+c{aHAiUtw^0WE00yOq=1T~tzzk*)a{Z6?*{07yQ{qx*>^_S_r3YbJb6TUC{B71u-oDYR z+Sv4|{}B|UCPQOmDIrfMMMyAq|MnNN=hz>>BMp>ynjrUUR^~l=TtcCpE9SlUH;loE zfIEdyEOvSowe;((HYq}MCA#^I$8GvM?jtXYPa%jc`))?M;o#$O!pkmn*TBf5Z(Tps zz~~D!cMH-fcQXWn=*vr{tW?LZh}ca87_vS!d@LE=o|NJ}mJ$|PJ5d`C)gmzJc5uW_ z)JXEw92(RcdQTC&G%~UNR+J30SW6v*doG2Ht_;o0RHz^`EpF1Ri4*(2s)QQ`z>aE- z*h<~iHoQzKDw`2*y0d9cHVXjQk_*|sMxvpe=rU@)Vch+o+9ySzhw?XLy_Puxqf03j z8*j3GK4Qo~K7NDhYJn2r5%YMRH0YCOYM{&!L%LV-1*Iv=N%9lD{R4_oG|Dl{M(-@9_D`Lu?=K!1 zYU25v2RFE=PF=lwZ(&%an*lSs^Gv7GDJtYNHYeMi1;x~zTJROUx73<0%@JyK`)}w* zjX~CXz4GPxl-D9fWQ^&6wvv4w!Tffs0q%;`c}^8rb^X1(s_e}T9T9etnq-Z>3u=w; ziGz5eG7{)_CdcpY8_a0z>IxX!c$2&Zxfuy;?uE)do*0%{b3 ze)ZaLxP}x+31MmerMfluqs-K&Ou2gq%kO9Te*ogDw~a;}6^(q~T30?JQRd?KoXWFj zw2(ve_C($+IRYnE_pJ6?Yu7J^AG|-^qS%o90mL(5q@D^^7inveF#s-}QiR#Bw~aGx zl$&PlX@3@mF);ond5$3JjyQ!emF2r-d%QN;OvWWcH*w$SDFeRauUBqzZ`iW4E&fpU zRi`EH9cC@5K{}0~nyQpIH^&hUwz^9ZD0a=6wX*x20qut;HN z)ER`^8Cxdc=9gqq>)wQl9&f+8emVT$nRSF3{2+4?8n0*H07pI@;;r2I0EnNLTh?f;CDVDF9GBD=r-#ccAUydH#}tD;l1{kzaIT? z`B2(dN^cWYGb!-cy!%&k>SOc8;|L|W|7_O#_vXhzF#l_i2+1$muadamAC&8-G8ZpN zynOu_if8l*o|5?``%@D4tADnC;xy`j>*nmGqTN_r&OJmt&gwQ&c1rrGGgtMZ^AF$yVdV{>@}ui*081w+ z>Qx|$sEhE5kf&eersXq1n&;f=ldI$lB`o84a9L30l`qz=?Vj*368cO;d?>kpnp0b8 z%bB7kVytRe9 zYuk-Dezan;HN~aU7gsq1uaEe{zHec+?YW~A^-_=gG&pQz`&41Z*M;5W2IO8@ty<1# zI~3Xn6*+`$6r)U$f1h*fv-ktRU}A=Dc$e9-uW1sAVy4Q%&7y&A*V=l>kML-3H@B%Q zf-sPX)*8}(7U)DB#vMKDxlHVH4l2*czxH7Q^wEa5P@a@CIG(CiOf8F0&Za=H8+#d}yq{~N^q&q{6mS~rZm$*Dr9Wkv*+hbH~IEf9UHYIWe zUOp4pXX2mUbg)+lrEK_Sg5~misBvlbLlz{_Pz>de{_qnrLJfi=-^7fbz~;g=?7ABD zwAU3;R#N)6-n|H zN8xp$j6RSxw(ZWCnk|>RwyQRGf~Z4S)Eht%{qLNr{{ZqG-W?{6LEmYL<;95Xl65U% zE|-r}EK&(?C&bI1mA6Y#CO*d3tW+31AOG&f>G~5Ed@5T1%k)5amu|%WuB=2v-My{X zb}x6(@077_fY!c>PFcBALPyjDnmEAt)3-q}Ap!m&tH4bO$}6u3U0Ax|M-&NbfD5!I74q?HAf};EYC5-~GfW-X2}5ZvX`c&_dniISzR6mWYTCfz zdplcS-xP1^H6(=DpD7eTh*xtks~HnMYfjPq?0e zBRIOWYbK$n)u5<6B8fe~GD#91+5fxo2U~g6K+-|GB1HVF*0io^<~Cz>w}tPWX(%2> za@h@y(*u_0@P6{kAwJ2JO4%Ok{D@1zrTA3X+a+pMxYZffxc>aVc@YW0C-YYf6^(ULwOQLa1 z5Y9z>p^HVe%R zMfRQ#o_K_UG`9Bf@Vet?+K`2kt(!p-8*?<(5{L9K*T_K+gwFIVhq_vZuY zDw8DsC}pL!$NL@abfUJ4JPYJLBHU$TP3l)uT}us_te-0#?gXwYpMJW3U65afph{|v zsXHHivpFW&;KjJm@NJU7{%^Go%sOu6FQ{G0O&yFL6~sDkUYzkzu0*BU(Ft5M1mzSL z2}I{N%{GRE2*u|+anoTt%c|^yJyH9tL9f*!?DK7FUSJro8oN{@O@;5GQ z?}-g?W?WGgTD7S){Jxy&DSE_wpSuvlp>sPTcQXHZREvH2i_;;`DY_;iMQ}<%R39&Z zN*k`dcqx@TqFm{mUE;`T7m2)bz|1*+^n|EueW%+tLoO|&U$PmS;G*@z2n}K0=_=s0 z>mF5Eu;GUNS<~Qzh)<$dw4&^iccUTVJI_1?ArXVBFJ#BKx`qpOpWL zueXe9<7@jy;SZFySZRw}Tc8v#5S&0M#odb+cXxtQv}ka5CuooaX(?8MLvV)%cXv+i z`;qrq&wI}KGHcDuUNc{2_TJaeu266%V#70t@E6nkMgvEU4}3Y)id{&z#9=sBMt^oc zr!}5<7eXFueF^ad z7Tq7@-v>rXab4j3kZ&F-&QSsGQX0pmQBVq||8a3{TEfvqOi{Vopu~)b)muE3q^?ew z3+=~q6|FL>3?fQv>f2}jDL#!fwH6L`o$Nr3oNLC`C@;Nl)^dUEb){YS)PKCGcB~Bz zkYz}!4UwBh44(P`WzI7^=oji2bTY@ON-bTnj2P=+v|pVu{JP9Z0UBR)RsR#-c=3*~ zLKS`l3Qskp-zk~hkF!%S_1;WhBwv2JFN!zD7m%0|`HC-No!vFxmS#kkiS00C=j3XW z7Vpb7yJ2n#VE+JJI%2Uz^VasThK)q=+Yc^#YEIQquIV+Jud!Cn;uys2mlc8xDc(d5 zVK>`n$uLnI>+>K6!x~*zI0?$+X=o(f?Rj!c!6i9G?RB!({Fp7y52#9>P+!rqQgcN% z_i9VKtK5tNL3t(80S5^9X(N^u3;u+r5@O5r<(kz9pVT7t{R19_5e*aDewzV(aa{U_ zoXs~~S>YV}c9XRE_T^{yX<takNw9?@tZqF(7E1Y)nu- z3rUd!FP$oN1#ksJm>(49D4M5rt3Mi88x2zS<4EbQ`S^PF!93Y5-D?MM*xeB}UJf5m zrH)1nhPDTB7oLxC&h3Dho zv?l$D#^|R*-mY7drbS?P#?FU%obmTlUlw6OsDM4zf^9!>>q#_&rqDRZ7!&8FMn~G_ zj`KbpW=Sn_m*hsv&DdgkFQ}1yA~#m;uVY|1<#qplcg)V4W_)3Nu0m?-b_RRJEc)0TSQ498R>!P`P!5Vp1x*C=nuzAG))&_J%sy1v?|&Kf&o zq1=4N^wfBhHS_wNK2ut->lkF#+pAq4aj&|}LBV8$s^H=pN6gIh0spS>7Bb|1qtTZ8 zdE}_7_kl4(2*ENJQLt|${{m!&+L8ATB>!a{{S0Fsq?HfF0-q{s_H$UH~74jn9SaI2*#FmKofG z+FFwO2N7RKDybHp98g7@vJyjD)cq&4u5H`jrF00n;7GhRow;flO%cJ0v(oa37Ir9Y z1Z|IKh@Ow_i_lFro+TywLeJ`kUKOQ|8V|gCyQG)2ronH zM~~<>q$yQAYbg_T8n?N zXgS4qp{|7<&<&09Q~iY$syUge%-X-y>erJ1)OB0<34711bt0iH7<(X+wi$WhqAnIx z519nIntI;&zP6bv=A$_y=^a>}YGGSsK=+ptu$>goU~9K24#o{e%>uztI7NY=S)luF z3w<{UaDl^sJDX3jG3kn9`NXt#&UW6nO+U}D=s{q+F@;fWFvz0CQ(@AkWD{ zZwHtO&S(qhcjZHUQwK`ns@xir-SltFq3fR$TTFqAGd(&&H#17*=&4Nq$`XDd%es@E zEv0@2EVL+^a0R#Kqq>-CnDFFMWMA^hy6lkw?S0XKZgVvMu0am@Izd&!A?U)4ZA3lj z215t059=J*KJI^8kdW}{I%jQ`d=cl z0eKpcWWvzJa*;vQj?s*=xAOnfXz+U&aVS$1EK5)H0xn3c`}7Z?6E#Hi;YXU`{-j4GL=oQf5s~4U z_SZ@fBkr9DNmnZVLGeMlvGbvpRRo;-|E)r0>xh2Dh-u2bH8a`|jnMHU>wrN2g#eH# zZ-;)ku`Okv%g8tf9!*Rj`$X8y@ZE}wV+d$_;wRnas*ml>*)%k2*DPe#n3&QCu6?{J z&R#2l!6{OU)%Kjocu&)p|Rtk5=r85vLfETSja`|Njq5>PrSZoNV>*9>~c(|qN&p7 zw70BvI@l}VO2iWsT$Fjoy9U%$yplmsyiJTq;z;@lU)Az`A{R4p6f}+Jn?$WpVEc7+ zE=*BPnTsB{+`>sBw|iM2v&JFKfnmb`&;!)ttS&Eb<66+`r{eyie0n_VSj-3#Ot2{j zd9AN0V(c4^ksUU~>w8>sn3q7b=PDPc;@i*NlgwP+U}hcG`9`eJ*h7x zo{`6_<7O_jj`Sa}RFvnDu*3UQ5cS7wn(wkv3o$!%SDkyZZ4Rp-I4@z`&kmN)2-5B^ ziZ4-pZL3JgfXw_kO=NLF7)7V2#9MOO5t~7HLX}+%|9qASQ@B`$7V7(9NKOt)_3VH! zHRsydHjv1PyVk>W@EWJ8T}nq$i|J)Al2?+%-FCIP6Tr*k5n~6A|0?2K-`mtvNhQms zSXyG+;dIubY~wVRm6DzEdxl6D=BWL`h)xu{>fC=uT*1s$!qJn=W9*WW(Go=W735c$ z3cJtPau!Q=r#TTNw_y$S1E{DYdwWyrKObE6-p2m;#k%E*nEL2NvpkMbkNw~%{qBT4hI_5+vtBgV{|M+`H_ zr*cfT5h{^q<$>cE|{k@7!}2>9q<{1uG^+);9F5g{va2FDGJOkBzn2pDZY z0bl;-uh%vW*Z@U?(*G)|P zznAp?N&WpVCh-40b;zQ26s=0gbd*rd-5ZHI(B+ZWv#~67U)EKdE5{>W=~SoB9%z~o zgE$jDkKSbpn5huEFXhOaCt31(n?hxv5#xpt{Fdph%dq30{ZVTOj(v``{eqHdNFf`s zhMeJ4%b93B(hqaErmO_A*3Jv8{aN)2*t1^EwE{IQ<;-4VJ2!B>O8#sw?1Ql}rWwCU z*PYAhGNL5LI^qBNytq2HRb{rkY-f>?$L2jTQjs--`s2qz)%8_VPN@9-q}3b zR}&)!>QyW#Go_~iH&it&Y-em2H)}}!8cQWV4yqsQqPfLnP8{SjX}@>#>)~rp{yM`vmkBr5@>Pig-rkws@`Db=}uluiz8B=1mDD13IcT z8%R%nEy8j9L~2EphtJj7*Zwoqk>K!kK@s6M&rs8O`|5nSN>qhldDo5yXIzrNO8NIb zzw)L|^Kk&lG?uDf88iFq)4>Cm>*bNG`-GsQQv=b(qATZ5)}?uC;g+npezEbeIH%n@ zPsx4<(jPIVI_$jI4gBMDt@M9-5#N1$F7N_6Nkm(JMqRLaHMJ`u4yCmT&Nn8zZ-jiv zHgjZ3LGn_*l(3(4)*r0SitzK0Jj-)CYi3o6!fmx?_*Xc-w>W>L1ouhAq6BGDW3}3Q zguKgJ+?5Hwa7Kwc38Mc?X(un{SRTr3CDDi<^=OvrKLj8_a=>RN-^gr zy@W&45AyW64cH2v@mcnpe@hLr!Rsu)?+Y2)kdw%fG6_Yqe^S)`gjeu)dG2mhZv%?X zJcGRKw+jmE>;-^nomS*%r*BJ+A`cb|Ub@79(vi^sGw>9(QJQ|^<|EHwPBvPr$Ui|% zbXkdxeEbovW=J`0s@e>9a(yv>hBiZo=OF_^NG2y17|V)_@=|l_Kwl-|l#H_yLXrAz zE1ah3UzYRsbI(dGu3qhL*N=Kl?eZ!URepc~kYEFcq+xR^9cDtWEzf^JsjVPuvkd1Y zO7Y*79%)6@EjASql@%torKX<5?S&=WLYw9t5)}e62!tRCnK)={ZA*C^e;n)rT3P^2 z=T-=acgw8W>J;yE)q#a!So1%(Wm+V2su9Yd2H4y@KOXOK^MfW6{(P$0BB)wr{x$v0_i!3`FiBGsdy z!z)p4d?TXW!J2`oiqPMk?6=PfEwJ z-Qw5-YFjw1ef*PRnHl%CVaz){tXcA{^_z7553GH3oX5&^FYJyc>Z)v+jp8Q~z^iN7 zTGv8`mR}@w|9gUeU0CWAdY7TBMz^W{P+xt}AlUf&v?VO8EvQ+1t?Eh!b6hBiu2K00 z(b_)Nqbju_(wdstRS~J>#+I_K4(#8@7VvNisU@IgxfznB`Q2w^6X1m!|1FZURCq{N zE)|*Z+-e>}Bo8y(i}~z!QC?l;B2bvwHJ58m^m+Xxvxz&6ANv;-DU+DZ^f)Thf?3X+ zHK=H86gZYewSST8QlI9KMvA1jvXFT2Qi7aVMp!D>Shr_s^k!KqOUV>PMZ8kyl-h1< z2VrMKtg2i)Os?hd2OoGT^YovvQvP9PzX=-Gn0&tD+4i!};S%35zWS_(hGs-vS<)4D z!!G=_4`Xv3D*A@#)t=tG#Tccov>N>VLdiFrUkypkedYQGKJclz+{Mz_-2*bMe!`q) zK63rF-{gxCM}OpQX>8v)Ri<$}D0u?-FXWk6^<)>S02oAU{DKYS zFfD)YCz~PNB~t%Fw*L+5d$m-f(C2C1Z+o;YqxXOn5T4Ar=Imf8qJNy~&$I?e07QNR zyyKe$N&N$%aa^CrMida?z3VCMWAUT>$3P#B%;;@f!;Jx49z5L#`-ppa0w7SK(b2Wb zUUYDG_8(ku_I2ZHT;mx8tN-)|*L zzW*l|FFyIgT~0W>f5Sq$s1g6A@=R2@Fs;K#ZY6H`Wfg=5$GqU!IY+Rd%-tNy=e0EK zBE^N~wgbT3ciE;<@`@5RmTrJ?0~51hEC;o*>DgOtrAFPV;{ipI4hW=Ojl24pb@r0o zknKR41ROtfb3l;$z-@jtUr$e!U^`KTOl<&Ir15EgAQFGLZ>gxYF1FRjoKZcb!Z<2_ zoLV3}Cx2-fr^e=a>ky45BX6&uwSiopBqkP})cQnbHz>Ish>$TOiiBYn#bYup`9-15 zJ-+XQTD78KDn9ayM&eubg89AmA5MDhr|GnpNILRcwXFHzB~fspdHLKq{+FkuKd*}% z2%ibCrX!sVo8}$&w(-?~gD5W$x6nfZ)*|pl<-lFt-ryCi9MJQxi&frDF zyVVyLN7NOjV-DW1&ShO>I}zOUg>82AEtD^@Qdru7>5w~(0ID3WfA1{6 zZlhSthaSYw+H^SDI*xGw>s(IIcE>FoV=(k%OS5Qh&(&i%Zm^^b@e7J z#KUogjl9;@VF+UvlppNc#kgsvb`LPK^%8oT-lhe%Q_CD+j1xz7X9w)dH|20@*JHuZO<9wf;aS!p8e%~`}w#I z+hM(4)|4yqkJF`;G}3CuzY!&tTug5z(4{HUZ=PJI-aC2?G;NMLuP)*BykYU_o787s zH4`7#F?l1HRk)MT+R#K6@2}oy(HfbPwEBiy<+n3EEQ?S0)tc{=tryJX9lOwf6$byM zSWOd^`OnQuMADZp;$J?zLms`0dRi9s--5g6TTv7S`eQ186w2P9mGXZ}FsI3|P@fK` zT25xK+#wvgw1>QiSscLlz_ZAqY#2ELH54;|NAI<0&~7BDYF=IiZ+hx|Fx6W(`L-^0 zRBAQnn@<9w3Tj^9Lunw3jvk_JY}9`LM5DzNa0QHaJoW5ddh4AibNu`fqomR7CJL;o zDzfkTJ}M8k2ckJwQWkF_HBvkdQJXo`rkp7#EiY^l&am~92E@m`8I}D-E5Ercopfs$ z7vOIghMaCzvliOyq|rpc=c!OpEK#1lgLHb^h9Yg;QuAz*A=*Z}{>qBT34KPtsq{vU zVAtRc7k;>qmt9L7@m0Iel$NUbH9TxuVvzPiJ!J4ir2puG&3FyCEi=C%9XfbCJ1)5QQrwE&=E{OTMeM#eG=L*U}8=}GB?7OT~naN^EU3( z`baBFX{~)%|Av|b03f{)62wjrmnZwcaq7IIB_`YN&ED~Bpo+nhGby(Er}bB>{CX{T zyn(QgZgW`hX5TD7d;b>MrxrEU?HfG?K1Av~^Ey;rp13IH!1>_7TW85bNp=^u=yevoW(<{V?Rm-rtM6&>B$(e<8SlUQ zMvnTk1KbtKiA$VMq`Q3BHysj!c(Ot+ScW;R!Y1l<(<-a~De$R?z!C1QBEK`l?8KdV zYhstNtTykj<^f?=c1vgyYHBG_OkVil6hJB3Oa4ADQFnK9k+U*ln^E=G;m_U zVe-Pj+8vqn5c39}yX^LoKO=j8*jqwb{BMP`Sd8bQt;KcbMx}j0=mBARQr+tKdLGlj z49WW0U5okh&$t`Lzv)Vc^5kzO&NMX6+}&Tlt||!{yW^JCO(kzNA-wm^Xm%D z{dlb+&vpWgD+lWs?=2=S(x;X!So)HLj|Y4(khhwEHtA(o%!9&WMTaxwQWEH?E@Ww> z>s*)K?y}}IBOTip-xY_cbyAs@L~zAeZDBm?+;;6ttz;ZY(RbSoU<|lwtfYJ*pN)Yv z()vbs`Wb+uE^kc}jiy*s6qOd2zwBrGPSCu?lbCDLd6XKZ?N8mbGH(E%*F_vCjWs5? zhMjDZ)A`J!IVom8R~@OV4_0g`<+S4yK=wB;@vmQvp+jFvyxsaM(l&YduiSy0IP3$F z@ZWXOraSvd(D=mY6hMw;!ethn%dBmUo70TLvG(Et331n}jlU@=;uKD}yZuds&WmC$ zCq_hGG35ZmGIUCqU4>l(GMo>RjFp{C;JhH<1?e(yUOF##TQrDb7qe>W{JNwY-b2O{ zae{qQxX`k>b2LYSCIqwTta+8A> znBm{LDA4Z&mvk>}8gh!#VT?KDO_157Nz=UxCo-mAYVBx)P3i!$hqo&nNoR~>aeaUQ zwlE&H&>{eD-NF8y>@S^-oxa3UgInI16ng|fG{0rrG4a0l)G~K$cg6QNg||3Ys=`z% z{#BsT^L($L2D}0NYoLz&a(=w;%D>WATwnI%GK(&}e0EW7Kt^00f7Dwk&KLE46p8dW z$m&ba1k)ghQa*#@BBpf>$(8fcB|lr0l=D8xAG;S;5=Rk^C`8w2Z=bZEFFR-_{O-XL;p|JMIvr7tRN7Q$_5i$&!9wWPBw*_jK`geNpn*z~C97Z%4a7*Kz|a1rp7 zm)6g+Jk;rjhW;)HzT4W}f2Z3(mftu@Ks|QSoF?K$9b8iZNE&WgFbnP&pUJkj(+C;Y zN#LBarn987tnNP}u$Zr)JN1jw&F3*Ad&a_|YX-cM{lk9hq;)Xm(v(^2tNE@#udM7x zj$*bF4Vc>Zc$B?&x|ohcGJZtQa&H@Q5bMcTUQ}-$zxJkMl=aOzSKHrU}Q|8rjgZ#Brar=`T)}@lPW; z2f{DL*kZ_c_Y6MvYS~Xr`OrnC>{jR48~B4m>zgU|^eAkbc7ICdP>a`+{xTWA`>lX^ zrh2JOH(G^1p3ujZ*z#HDD@2ijJrJ~FoM8)X5P8HoRo+mXR>Gj zDF~>avdHt=T{UnkI=)x$4Q?ADZqzzuYD6l5I4B=^@|$@jA1=)X7<++l$8O~+#g#B! zlhN<2^ln#u>A(p{v4Z+JjA_#*0{xK$IQB8X;hEP_e&cSIok1rtK(j9}b2}hm`;$bSu;X6Gbq5Ywi-u zf8rQU+RDopfmX|O#r(F$`&-gnD^&I6BNk}>Nn*!*6+c;}y*l6GYtl5UN*LY_-5~xj zMc0>>`Al*FI-ki#V9H@(BclO$WnI~+$djSunG!$6Y*(l8U(jDrQNl1Na-rpKZu=8y zWjKN<$uLyC@b8Ui7X=|&nU!xrJ%qJYO2XeIX0d0&x1PImb*5XwKW~n?+1&F)?t=ZB z`rXWS#9p`^-3pqaWNm*jVAal7va7|sT^?@dq|Eb&9=;g+a2`TWMS#@(y^a_g^bOab zA#75_RP|}mlz$6mmT^-wmH#wiL|7rK(-qFUkf_v`61*5maUz!Qy7tjYgJ&hgZZ+R> z^;|Fx*}c$lAZ=`8ZZE?6h!G177)`woXGdEGNP}KI{f}Cs$^Uf#LJ7U%x4nDBu>Kb> zA$lh_(e;RNKpX%qZV#WzFj^T4_=F}VSOBG$=qJ#XYvP&@O1>5hhY1EhEeI5(WlmU6 znu+@0iv36YwaNxX@Alve^zC3*(DSDSVV`4Fyq8huj~MUJ%u8e6&Gg8tNk77C-$#r; zNg;HnEkAXy*R+oRmzCrHzsJ)5eV1wzi--djGc$5%eR}Yhb$ZTKRp>(=v1Vgc_qV{o zZ%+=hQ{x3F@f!cKcXpO!3LGDh>>?G}S`#?`4!{pb@!m70B z&|SeKYt_ZZ!VfCx7gT%eKYRQ=gOBIkQ;ZI7^mA&i+!A$AVsnC^w4Dd~QDSw(wpsdO z1NoCH8q6*DK}zkCEu%RXl0?xejv*pVdPhIC(fns5^dy z@;pmA(NCglD(`e}VobdpSNlf&E14aOGkcc@b04I%HqnWf(8bbKc5&f#Er}M(h&ocL7bk5ej963_#He9*fkjP-?OGdYevwWvOklrs7=)#{88@r}@f@LpV4&zzERv7bje?&)e~fIDQ~f!?|$ z2v$>f=1jIu=s5O*`K1*E@jM3o=}2xD1Y5W=23R>S(Av<&Yb8T`$W9ts2;ctq5u@4S zpW|}J1SD5S{GTAIQ@{()?VW0|lf>#ARLQnwd5WOnT>ZH2up&%qpw@7I`onuiTg4ai zf*`++^WF~C-^L}(-3Jk^6SDt^aj2AX_<9em+ux)AAB&^a0qWwGb+HG2&fAvfZ)eG9 z{+;iYrts#S>0H`H-oK72o}3A&-xZUUHQLXG-lEIPlXc9ki%s_}JwJCuPV9j47V1Q( zLq18rF81~rCl}2f&!m%Bf<2I33)((DrF7#}+yrZ{q}2baB;~|^Y}i`dXzf+W&Lecq znlAgvns!7Yp}h`;+I+A+R_x55Ty!J>oTyEV^}dB0P^rHjvRuPcUTf>We-4>Cc1fSU zs3%{w2d$NEx*16;T?CFl;Xfn@he8mOmJ>;CQ&PgJ$~=%sLf;yI3#Q7mbD1NtVKogM zz~0og_0OjzgziI)F^Qu&PNPQ$gK@NJh}N)O`4$`7?UYNACQ)sng*icb6#2I|euB$0 zLhNKera^&q6P=p3Q@Zl5CZ(R7tUH(1+-7bD!<$wUBrlBU$+p+Xi0LdOsrpM%^6b3VIdsn?KwSNsi?K3R$mW<)?%a^w}-Rf zv1>~M4Nb>Ancw|*%~4bK_{6-d_qW8YKMXDIyUmg)MdsHEIf<>by86n(5wycU8Ye}c z@R5cA5*QRfgD`>vFH7>6#7vT z@OpGanUPn5-h%u%`t;iwt7U8f*a{V#^?ACk2K$)|uJ7g&`n!8o@B|+F4f94j+!NMO zrn8;y@QcO z)QAzj``0B5^{?Bz>CTXfYU7L`mQY&y0lYH3eIJ(;JUoI2OLS^Tjcd%S zK8HR_SX{*uEti(PMXM&;;;FE(&eu-$;#M^&O!Bjr7N91s>-8=cEqzX6-AX-nPK$Yi zGi{L*vsg>lh~C@3*WQrR4TunXSVHP`;kzBywFIc+LEwSI2<@Bf`U>rNw|xcvt)2_3 zbXPZ}Z)g0u(ypH6y|J-DxRZA9!WMl`1By@cAxvs`KnQR&H zK>FCMt153DmNk=|m{OUy)_hY|E;_O(S&G6$>Mjf!e;Tp`e?8~UCwb#Y z=($6cJ+N-f<1o7RCn322t8mEh*8d^xBci?viQx^*7o4-E5qfX;ZLUo z4=IKHaY9UZ#=wLIUuJH4paeZ0!$q*zgFLaLtHF!N*x_dwAQ58nojI>rew(e#k=WgE z6-_9~>nh}M&L(y;S9>5o!3gMAGPHhST9Url$rUC`+hZ*8+9XYW_Eg}n+eNvi`<1x7 zgdQ(2WnxBVF^e1v*03hKmG4+*#L26i^Ok@ontQG>Kt>rR=l3MSUFi)ctX$oUu z&Zc^3karHvEt$-Taekd5^npIJ0XKv15(cNaOCAEk_nphnGTXlv+WrzbAuQS~7h52$ zcBSKeFKo*R%Fp9EZD|(l3)3WI$eKT!J#Bi#xHC6*)bie1PsqP7)M2)eZ?|u#?{ogl zL-Rf>I-1xX&89T7RgJJSoPSYZFN=(`F`WzipU050FEnbg^S?f_hJxb^Rvd?MRiOSH zN0<7lyA6ukO1tfd4Uk8_ zo4JU^nR;jr;-zQ@Fh{=n=tpS#&sV&Rv@K!65Ge}nSK%HxveX|zg#+Y^wz*6kBs+eL z6$U@?{|Dm#``hGc*;7qAw3vok!qe9ra# zq^Q$3%iYB;-MU1~7P2R-^k^!jw^<=!m+~_cno~+!W!|u51_!1|tkO$XXZMUBUnVAI z>G_1&g$?j&in1EQ04u4lc7j%;!O($?CNz&IZ!-Z%T;^=7Hy-gF26;dH35Y+TVsl4y z>26z9-{~Qb##Fz>rLPM5#@;{(u>G?j&OSQP``4zT);7hD7!g`ExsaxPveJsiC;3>( z$e#VN04(>rz;ei|AW3bdF)snHx`CPh+65;!hyNNsaB_NPd|vD8Q=`yF z&6!Y|DiIfbLN(S+g=-th?Jgj>h%tM{w?o!Yi6szu6`WHiP|0##XOrJXjFhHZa`-<)(!Xr6pCy1a92? zgxFW6hj{hJAkE!dLg>VK^DV$*Zoo|^?Cal6d%b6eYvPmb9!H4n-sPgk!$IFKS-^92 zmyahRP2a}q;;-P`PepwzP&@T~6e;Q>odP9Ve@?3NCiw3~XJ|msNxklME{}Z_$0d%k z@>fowVCa~R{Upgbb=zVV`-X?eR96qgUK7&DvuniA5&p!q$aF!LU_^Z%81Q*l>>cev zMK#F7(-uo2sIzlpegS^sit6c^w)FV;+dNmnd#U7Ec?CmBSGZt>Z$o}V;qKS2gdkK0xZG)&Y4gqVkO@5D|Az|fI*)Cfg15?z3QW_VcG@T zM(m03CF>P;YG>Yf{w?{8)iwPmv&{V?Dd_yuzg=+^ZQGSl2s#{lfj$pOy8Xe%JVK!t z_zT=Mx+n8hpJaFG{dP811#f8fT*a=AyvnVw1{69f_h*@j!)1l2id%+;Cv>=a;Yn0@ zr1NYKYJBE>Ni3W)ftJyi!;{%T2p4P8b?v-G-@kg&Wd&h4OcWBmDk!`4%-~e>DD~5< zaXFMgU7fHr$K!5UUW1BCk609%x!-iPQ$NS^U4*HlWsahJceACr<)*^>>s8u`6ZbD6 zOPXhT9SyOjljajVKUhv(#f5SFO82IP&zCYLtc$?a^~I|RUxurlx5zBU?w-Y#sx9q% z&KC`tcNFN#_`MYAkj0Y7SkL{3@B!7l#ZG1pT`I4-A=4e?;7Adu8FnLZT@|_d7I9`9 zbXlGY={5DyayA+mf@gGjOn6rO;5~>rb3glw?#X~%drkeau83rtU9qzgJ$L<}jIFr# zfmAn*DLu?VR-Q-BjTkH`Vs+<|sH-~R$r%DdC8xtLqc*k2zcBfd99WyxCB|B`w`?D% z)9Ll`a<@;+4Gau@m8d_9O4`YNqSC()Nd-OD&%e5uk42gHrHJaPeYpC-ss8WY9l=c+=;Q+6(}JlD5J2vJ3E-@pgbSrUg>PNnQwJeUW8#X>@amNNdeN!ov!2b+8g$liHo6bPapGscP6bm6x6>6qiMbAiw_D`_K4+CDAv`LQx$(x)=5- zQSDT!M$S<_!lo;}ymplVBKWB^eZ6afX?xdp@CUeqqTPjObM&^=PDPfUydvPQJF%2e zT%fj=Fu~NdSL~Q=K}JQBL=Ya;(tLGaKS)Bts#Pm)Jmrdci-mQoqHI>>2H{QJNA}Z` z)`Q^4qjMflZ=7p;>+0{0-eit-rNc*JtE!mJyx&yI``*rjcUK8HKg6n+wUw9}nS4&g zUby)gCUl6-vJF?^N7f5jeBOXsmMwQPdpQ?X@}`uAjY!F0Q!sh;>QH&QhWV`ES9hHx zPQBK+Xtrh%*wJxL81mE4B+tdwf|{AV_&g;lCw7vc9vyUQuH=mRqb)m>y}H`~*XOzf z%$HpArRv4lkx{|ELj@E*%w0UW&Z7MhzDvX9jNZH#^lMSEjRtOXya&&cOG$*sN$zF~ z7c|<7!u=sd$Y#Fzg-x;1QY(9UU1?5E;q2%DO{+Qksch*!0$FvcwJP(DP-v|}SIW;e z%1g{O{L zV;?;A9Z_1FY}ewmA%c#bTR(XDT9E{ve{z>aSTk)IJuYkgfV5*xBl62U;JEVzzLQ(e zQQB#&+BjAKfCJ_;ob5=*X4=C&|65gpq5kQ*Y0l{w1UvB(N39^N;@Ld@*{LM`F=CdU zo#}vebb+aHd!Nj>-Bz!MIuXeBGSC&Qh z;{Z=#5#Txr@&@ISoYVb`$_MC#^cM;F?bcROr24McpKp(gB@gza3$Al7F&@fW~t=e+FZ zR2kRho2Aa@`d0(~6q+!H)c2_-^;MElZvd(LbzCCG5x@PI2-x z5cc{ll>(l|e}UGRvnb;rAJkuaApGe&!B2$)SCX=Aly|y`5hUeZc`c>P?6CG;S!&M~v>o zM+`N0)$qPXaZ8+?8NwTzrGR5}b_246Y{2Lm;vjZa|A?`phLXai{j2vBNm_F%H+x^T ze3LA-!-_UU9oSdhqB0*bbhC8`Y@ZA0Nx1VSxsPdu2ds@eVnCF56%+4z8t8l*9x-;6 zDvx+Y*g9h9?lf%#{+#`RpPm06h|xBoaU+R;#Q67k^RtwS+WvA#Ld-MsnBj_;{Gk+I zXez;Fz^)F=DCn^4gYBPhLauTrLT@%j<(f_tVWlV_xQ{;@6kKGL2u z*c0-IHZC5?_)Nus{i<0twjvuo=1`Ks8VVFMeG|qc2jae<5F9p099{2pt0eW6W*C#f zqyJ;K=g9`6#dT_Qcs0OTihOQr^~y)YFCC7t1Efr(xhhD}N)!@Jn*gLNcVKOgwo#xtegN_sw+v@|5w6VC&@br-=NLKL@Ir&y) zkF={(V1F?*1S|4o!q=`iHhXxF%Iv#X)&SA43HubUYwK>$!~u@kw>id=QwY*=A){Z5 zBFw9~2?zX;!o)wf+PR9{)dg@&E=p~g_8~8P+O$!wHKlnH^-%|r3BY#D`;Vhrry6K` zjj$alx2>X>y(T`K`}Zt)C%e4jUlioXwc{y!t!M$h)h=F9#rKGxU3vrLR44cS`31hE z%)9SadKOP-_5ccQiX17i#nl0PK6XMN?O><&!`_ikjYlOYSO1uLSy`nojL|Sr3xQ3K z7vqH8-Zk_wp2g@Oqn&-h`$jZ@qAFBHN%;CHYo0%?b?zeu-r3>Bx!C=eJCO;x2QurC z{fdB_sB_hk2Xv~M0k-WBXWTXWMDD!CBgXmb03x@aj~JnF*IS$1hXQxi5$8h%j?gQ= z60tk>iGcqa1O3;S4BK0u{}|&f=aM}7h(UDruggA;|Gw-`w;h1?2%atf`_+ror^y%w zc<7e9@{br-+USslW^@e9&W9Mk;yW*$lH=Tgf4%*Mi5GeUO^&9#YSkYx>T6SkW|(Qz zw_j@Z&RQ>$yNhI%GLhpBwYur0v_YqK{2A5IyB#|cUgYS-N8Y~VkfbcL!$6)%+nWK2 z9d;%VUuHHL;65$6I`9IW1T=;1Wobsi4uq`L4bT@hQ?4=soKiL0fJ5xV-X629XBTW7 zbx5@xp+7rjMPa!OmT&gvFzM$vswk{s?Lm9_*+YVz_$=9y9ZgT88`PAkFgrOc{=xGq z@>S!LSaV)G7$|J(Rz4#hp6=+k z`1Tp4Sf+VMw+1C9e|PGNIQhUAb#Jy)36XTJpZie~FA_rZ30FMYuDqu*mrkA}x7vAz z+8@7YBPn2``(}=x}|lN+ysa z(@VMh4Nt7l{rinYMgcjy)?9;d5VuCx@I@fuC&3x&htL`=b!BOFMqUl&Ft2RH)zU|F zBT5o^MFPp;O-09IE8lY-KJp*|oM8c~D~FJ8(?Xr_6iOMBx_L-+Db)?Yw|H|fxdIYZ zD!C)$kV<9Ms0@^Z8zM^Ux-zNS>WfpQY|_eO6jE3&K97MezJ=Y1pE8m0Pa z7x=G45B;yw54`_67pi|qMp_G~{6?F;@u8cK7~y=J`#b12MZ=qZKVqndU+{rS${sPM z5=DL%3j_Av&6y*c)F&i$`rzn3edm?doKwv=aJ88cw}yZn-Sp+4npL~I5A%^izUMu@ zmS>f?i;&^%(k;CMu=2e;u#4qJRv_HimR)1)l>wc1wfFx+*;_@m6}|1gw9wKPD{U!7 z0<=IWUR;BeV#VE|xCD2nK!Fx5?!kk52qai3h2WCl?oM!*?)<-TzP-;Gd!LJak&KZs z*IE~Atjsmv`99C@>ApvSWk+kfSYMHSTCm+Q;=!@~JKX0bvQrqh@!yWs5Uv?(qqY*P6oI)A}n?h~?kQpvF#0*hHuABsl>updlbTrax55 zJToitP}j;OpgO+dDmL$Xb5iO<{2g5y7R^DjDxm_uj3W-oW5b11Rb$5V8YS24BjD1T zIZ1qvC~Fg&ZDd%}{+3`v@`8ca(7S$;f`9_`2fMLp8Vco3wWHUj^=MDJrDjybp1wa} zLis0I#oioRp&mCQG_j6=pk*lvQ^d#lqp7il_;RCf-MLw(SFk-;-ZH zK%y63Cq_WI@Lor~jb-LqCgW+ayM44JhUTo~$actmOcuC=;{Jl>8}!h6^f(h_K@>L9 zKQQ9E#4IeZAZ!`4J?Hx@;)i>K{bu;M_VPFE-igXzJ3$c*_OjZTA6~tO#@-{>We3(} z`*-xH59{mPpEu2u{D9^8V*@);WHYsBVVUt3c`CW-806<+Vxdb#%Z4KVnzN#2twa4% z=*K~daL7BKe>g?=XUhq#vfZky9nEdd5+zYorw7#>HoKWVOcnaw{}c`XZ%okt zBgFB)t&;yajp!uU=T;=*e)uDnaf!p|eo zyJ7Z?0-mqEi;@zA=14wFdq^0aD$g=zu8|x(GVDy16&pBv+KrzI_mAaVT3EO*F5uMR z#zI%I?;%Ka(3_=wN z@HK0?hQ(TtKqLiGeq(JhHol{4}n6GJKXT2q@MFaB!p%7v{j#;A+zxruxVsI-$tJdq# zH`ua}Id{v;s4&#NXA;O6<4f z++bpizNImE0?p(mI0N@ZJz`pw<~H`9`EoN_4{?UwuQ&n|FRer=`nemboOC*S+0v%r zcLrGDPoW;b2tZA2pF9wiJU=yM_@L+Lfy2kiG49v6uFW;8p2yUzV$@@OY$J~;t^(*{ zSF;S_oV^;nDCtZjaibmZ2lpw&sVh%>pMpIPj<1b#F3&{`y_dajRrayh?%+7!Y0Xzr zt~df2BRSZQV?phYm)%NRsWO_5YeM+fX+R6S1EwX8=07ljDTZuXCIFdFyUIx1@c;@JCMa2EyC^XDk;mqS9|Dz*7ub}Cp*_PQH! zRlWPRq^NGBAqoBh`FY|&`@}kcyj1*&O^(cU%G`dyuaZ;m9|b~g^?eGmv8!cc^D?iM zl;z8QxoD?Hn_qpj0nCXnlm)&JfUR4F9brkuX+ z1FC(f=MT)Q29n}(5X6#CoL(=bx%7zDTcwQga81}c6J@gsS_8pjeco;ChK+l;k90bL zttZf_=YvQGWk3GqR+BgqvkgjZCV979b5lR~oFTqLA7bwG>;(dxiyAN^d$3;jx=F+y zyIC-vNT?i>90$`H>oQ{Pq+alR!abMZOG~j+8GXcpBg+<57ln&}3+w_7!QiMVF$o(9Ef%G+jC^z_(n z6ugw851nswnKuIGpJ)&PXY}}h4Fj;IXJ7-9ij@cb{^zX2sVVu;jHKG}%Ai-$2;c`Nw{e<05yyzqN{A)qdcR{|`>b_PM_pprf7u43nZIQ&;}c1t;|MwVWvj54&+ zly21lI36Xe9|Z!$%zncS@AV^LT%8ePjJ8@G8 zI!G|9Zp1vfFH(j`O@3a4z`p$pf=_4f*Ag ze7=qB2IWBIj^-sl6=@%e;{7MwZq!%hslBESY-6LJ%>4BACmYgU;4U8!5XjN-aYS-k zcvE*&&MkwetBlpVjq_^dMn-UKr?SdCcLeiD_PPa*RAUS3aWnY$FG_70IZMRQqad5U zhQY7w>^rEZy3L<|aVNgWz8Vr^IZOpx5oT4oQcnD)C2+p-=~<@W1%}Y4+~A`IbC`mIQurewn^I6&`3^Kj zZf9Fqd>a0g@^XkkW#ZY)SqbQadwEG-Ns17F$EOfrvAauCjW2^?O&?ZkJIl`v5QPI(+}@KFLw4n7FeHsHZ)r7l-eg)4rELk zh;J=)$qt>}O!o6c`I@o%t!Cs^RN%Ti;KaMyoJmmHxc~*Fv@%>Mlv9v3eZ(g5`fTTL z+T}!HEh#jaaoHGW7_r$se1qlH8R@KT5s2E-k0Kn)F7kO9mOZAwv2IhlJqmaPDMqp1 z6=Y+rV@6;zeisVdsYZue9^SRZply|iE~gxX$7FYJvD-f!&#s$S;_f%q$M3Pv%zAua z7r3R3Tw^`od>;F7r<~ms8@0zn5B}k>N1ZcaGCjoCjDXlTlKI`w$()YT<53sPB?MTo z(wH;Py?gt!xc$l^s8-y)zxtTtKbNb$|NEc9=*nj5{5ulLK6k{iopZ#WgB=;Y|9gk3 z%ITTGzBW^3Z(de&Zu5!h%K-B}n-8#$&&k1aj)C8>W5VK7g`!gM#Czsi5|1|Vscb!e z%Gl6j58B(eemDo)@mZJd1bM?u1!D#)W$$ANzRxFV@e#5A@J^6FG`tAE&OgzC?lM1-YCANM(c>YZcjk95~6B$*NP_E%( zXiI8D2d_&V#O7M?*McYS*>9n~!#TK+Jr?(MaF0P3#|C)_-#xXj-|ZJGiTnbRhD{U3 zE4*|@$^mx1;j31hy{=A3xltgBQ!so`1)A$GtT}o7U}bbP+c!CM@*;{T(wp^w{Cejm ztBNr8u^_n+=K7nv&2XGnXhp;i#D8uWS*>jh^Gh&G>$^`#c+Br^WemfWL*1EgkV^FU zWZaLT1Kh$L+aW?4al4$H2(z&Em6w6-tAl#xW?u zT3MDgn;jP{D{5ZgGMM^u^ZHfdJoG%VusAe)h7S?4yiH8lG$xQ{<%89zJxF*_@{J!{@Pb2h5WVza^9rR@1{MwaOUryAb34MYrrXm^eBta? zqvd&#G9K;@ui_$v-NK(Ucj${@@S6ddxl>{QwXAxraE`jM`+Ca3kMDf^llAqGg0Jom znaq>_Uaoz;i5wHm_Abg%m>iy}O8{?5lPP?K{A#HA@|47j!{196j^%_Fk2jVBt2Tz5 zlbk*hgv=_KypjV{(5L{c;vp~J4DSvv0P84D)j z)9-v$ht|kZu@6qI3+oI(|%fQqOD96bfD5|wOHm@eb!3&59gjC zs8-!GA1+c-_zy=b?ws8*`U_iBwN}**a|9wir&i9fA)?y!E>zsiF$&P4{L+cfeK$-B z@k18z`o4?4Y{N@^7#{qH@Ll?_TdO@S{7XsS)=7+>4bvY4E_RoY&o?u2j@_}X45kzK zLduTU8A0B~$hzio0H1y{+vP(k@Vc39>5`Vwc{kbT&FM`3B0W#L-HDs?+;RGYx|px zH$>*++nmC^uzU3a_Qb^ABN%5tpJzbGUJ9$ z`VogavxJ-pcE5U=jprNRw)|aX%AfUMmloj_Ah97l%E-ytwjjj@R&TAjJ(#38GFDE3 z!K_{@>8R1Ggs!{L7M3b$NxD&6{ep)!t(>IX#CjUG%cleYeVmk(4PB0CnpXmYizjsU z#Y?w1<3|VWBHxrVuf_aruM2ej!+$cI8i^>x63UP-@^TAGEJpSAl`IVhg}%CEHASCJ zeO>o?$XYq{!GzH38SL{(Dp$SW!+-*o1)&WqKOveS(>{z2=>V_|lyxe==Ih$usgYw$ zKSC{4T(dKr+<+2BG_kUyPtYuk2*SYi6{<@sqEv-|3hI?bZ^CtBp%WN~I=qthw zl7}$i=U2UA#zMw{d@s*sI7YECIZ?PjY{k5{lb>v6@MCf4=X|TgAVUe~^VPDzD40f| znK7v8VjS@g#}N)<|2*q(Ew~b(c!YpFWH2{0G#sU)5REYX1!dQ8o#gK~vbL^(7FEG) z!+6~*LG^~zo`BWxMr8gm5a3}+om~Xb96O+YG%5N3MO4Q5y~JPf+j?aJ=FuQEI#G-7b;f zj6!l**iqwXbt9#i+Dj|vM=?a#Lx&<~#ws2wy9^J(3d!cn-6A|9Q7-Gzm8p`u@KZm9 zQg&V~S43VSaNN;v#)pIl|GmT;FF&C2%Tpio9R(=_5_W5{^t$ zZ0dwG`KP2aHI`>sY{su#HA|!6wdOKtRdxaPU6^vrZU&W_YN~0@OiBpSHMks`BSQ1> zQHD$g>~nk;7YmbHW-FSFPjKe;_a{$!8!K;2@yaP4`E>%Cn5dEWuCU={e-A>m;}4SN<)jxGZ0(DFG3#SC{W;;s{+E16ADEZMU3~_%6az zSs5B;JYkDI@Nx0Fn4z%IF%8+Q2n!2Z_M>qdY=gLYKfL2F)A$b6k480lXs72giKh-sVg`>|A$%$1a=t(@tTlb%WZ$E&kpJr`>8c89CZ z+u7+m&GS|D$3bJ)#>lQ`Gy5tZLc;4P@?&#(GHd-}^$*N2oGY|MHg@IosRm+Yddvgy zZ<=)^Xol6k3m!vrHolR^?I)2n3QuiSo*rwC^*Q&Xac`0lgb`^WVEQam48^5kT?sNr z*ua@*%e~tl*M=M~+>xnl<|F=eE3YHq+nM0_i4*RHUJw&F#fN?~?5IcH-Kl%k^`mxU z_YRoC;PmcM<|h>e9rc+UIq(->$6Z;)r-lk~a2ID~PoVaf{9e!EIw_x&Mb&=)1Gd<_ zL5$$l!^SuWlKQ$a$l`*#X?uG%rG6P3F84}H$1j=6(LITXxN+k5iijsqDq`cb=YiWy zrAe`E(^H$gJOplFakYJTXrj$bZ`Pqti_DWlLpcnvG;t$wbTtm-=VZCwcnh@U`-c-k zdHf6@S;7H@lEvd}GuvUEc}&&}ybV*{$9Se7YQYu$5aLHGjp+8in~QQMwM9M>5^U_Q`&nCLQ6r1*7u%WZ-n>T*S|n%s zl$hM-VywX~+SBsZD`dyBpKwCJwF5Q){7Y0bNL-+uhe{rnk= z4#d*)c%S?zvB>{V(vT(=tz-!e6o2NRM{##Weq(B;Z6fo~!>Fk3F^ugXuId%db}3vP zWk*m$##e{to|zaOpcab0V(7;IcHnW&77%`;IJxaBZ=^$&XDi@oeaXB&y*iNCzEDwS z!(52I=tqBeWAte3(_||W`g6*Wx?{DSdq}ufp17&3z@rRbnANnDPR_>?zhL2@;Tyj} zfw6~HHSgUXzwQ!X_ru}b%>s1FZ#BlG)mu%^ zGi$K(MTP}8xQIR_-Y22^ib2(R!gr+O6S}XcCqLnPq;HBMnUe17N(_pZ{a-nd+I?wS z1u?K&7ekj~eq5BNZ7f&3F>g@vCI-DhZx2oZ&d*6`?L4GG`ga>TgXcQgqia#xzGu(4@xJR$`I$sy^~f%}WK5FjbkYkSVqu~` zP^ig=IbJ^b@B(mx5c6)V%v-0oR>*-pO?LMTO~ca)_So#Cvx$o)XrGjyvQp?B{6shS zG1xJ(BE9?V*2R#1@Qg_I=y|jGRZ)|M>MT*$&b7{=hVJZNx9)>16E#IZ?5&eeui()T zdGutRn5rUlbjriBZDzt%c%E>NOBC@@Azb3H?+7u7`VV&R=M*>gLrN+k@#95oMA<0{wSXz!bKmF}wUlTRJD^68K*mmrq_`O!h&hcurh*2C2vOQp z{r^@8WO$A_o^52Jwzp=c=IM*FmMb4fDy5cp(%ZYt?aWMUKB#{#psr=ap{4ewjJ1q0 zuB-Z2d~&m{dsS9!M$rZ98Mq(~|I@~}>+7Whn7=PONWarQ9Pi+Mph z93PqkY?)|NmS;3HTn+)l>|s&0Z;cVg0ue0FUPNi z0&Eb5>{^fW=hd%$yqmn#}S znur7^*rjxD$ZkLXN%?0v_A9ZBYqA$cRaX+{3_95HRNvNq1n_$inARITRmAy|~E_VrBSOmEPX>-eV`Zyjt02itnj z8~HoNmblgX~FS% z_McII@ZGk3NY1FAsyvJ0A7aAH_$<8iXTH3t9QhkSBc?}Ki6=5w zmulcJloG!|G0)S}yM;z~&cH3nDeT8y4%;V8>oAN84gTu8H_y5%b$pR_Hr=$NAE;mE z$H&KRBoZdYu;JrWKG|V1R|em-hz?%j4GEYwL_|fqYgLIwS!b^oydk2Pw!nW81($-J z?r+$>9ijuYI2UU(cJj|wEb4f`(j1`eUjU6Z-1uF#?i1{Cy;Dj+CRU@wd7KKv7AII_ z=#z$?ZQR;#kGZStE*r_%h`{;%Jo>Biba<&m6&GClfY-4NW9Ncfv7UzSrhs7 z`gOJOcB`zjdsPh_%~g;+ggbuNqeFdpKHizAZ|1NHeol#++~n}GZLBfH1|5|ac|<}6 z+24Y_%xe-etyZyHM6cNaUt0~{HO+2>*8lQKo08ED?A7xnA|>Q*LJxJfHg!CY4x1GT zjqYbQjEMa;m$ytbVn|;cO_9gG7cN#~tR=R-my=8cHgI|Z%4P%=Z-UvQT+Tv{%2qut z9cNR0vEB3krgC9tHY|Ae|LP0UebPaDiTN#w%pcAQDqx+V7Jb!P1ciQ4v6v5IC5eba zRFgaF+zUaMi(l1e(VklRiYV+w!d9^x_c8kiMW} zxaHB&T~Nj9DcpT?{hC%`D?qlqvpNcZbh4;VXBam97C741ainz|xFKM+OuN!nb3It`Ad`QXQ!Sq>kdi$IcXPO`)}>d|KYG4gEK3eMzqJ2B$ay=3d{{n z$OQ=?wlO~C4yQ7dexF~|c*#2mVN*BL zwPHi&8*uWzuD`Q;49B6{f%sQNdfF}q`l!D=mZ3PC?}a|XH_6V1^2XWgk-mk+#pc!< z)>g=e`*9h=B$dl+eTsJ1)k|XKXqWX~aD&Ksv!kfst3_U^84Ah(x@+jZS84_(smd9K z{t-psFErZ|_`t`i%NIWKX^L5i)^cE5EP>eU$)6qmp|>Xau#PvDk16N&CI!TuO=PZ3 z)RGsdee`a`7&lIvJl8fmX&1JqO0xwV=a6v_>oT1#yioGhh3 z9s4U7OWf2ok@p#2Igj!4&vDp?)l(i%{K$|%emK#gAvWaH_MW~;6{9+LHBa65JehgN zF=O#rTJpaVRUW>}j&QzutBF#>7Gcu%W^z=H&dzR8GFKgf=T)-`-U(o|)z5g;Fc4BS zt`%qh{b@c^dIj#V3t~Blkwt%JcBr5Trq)(hr(jC@sr!VS#-SV7KT&H-a5%0Fi=+%z zBC3EV(?q+daMEFDdG_@)WFpi8EFD7C&Yw#w&t~O~Q6C5uIWap@i+buLXTXyt)2cAJ zk-ks10$w4hBhpV2cg;&WYcJSCG8OIgc#GhPx_0o*W;ZP~rP~n2j5U<$w`=)M3{Ppc z>)63nThWjM50$*?jLEW~>M9)GiFNnS?ZpM5&^y zjMEn^sS_~4{Nn107j^I3Td8!P*w7uc$@iV&ps-HZeU|PSNbH;KEDNz{xq`%c=iU>c zw=S&#N;c@3>14zPS_iLg456FOAe8_SOLtRVBCJ$SUwD6xK84o8R=*hTd==piOrFM{ z-eav&5wmElfr)Z!e=`spoJ9sknI!(8rp+sLfz+$YKPP1Ov-!Pg3fx#aeQROR#X&01 z!#r5uEVijaAY2|yR6m!xnk=;kGWRj;j613nN9`9IcSSW#ikGyi6;;zYsJ zfbUR`*+RS$RbOW2=1R$sxLoORDYgOl4ct&r_kE`5CjQFBwh^UFy88aF4oYHo=+D^}G$-Y3nJZ|3WMm1pA=5`^I5Y65x!5a@esJ>^X;?_3!M!FwmBtj$S zCRE?&-_mH{02%g)1-!CCrOF`U^I0Z*N$>XOa5ERP$60pS&y|EqOR0)D0L|sI?9Wf> zZpY%-Ag`YT$8M_$8bFFNG=9bsrvxcf04C%u13DeD%XXsb%Qx3b|m zY5+|be@tWS5*n5Ttzt^Tf>x!mWhJ(xRKg0g7Twu26Ay1Rkh&*S1lv8{Vx=3pLafYm z6+cfKxQD31@7^*c`4Co|e}j+-lW553Pg;)HFE>_Hl<&bxerMOcdGNNYc!%1zV9*+N zli#AfJ3n5Xz^vc1D#R5U@j4!Cx`AzJCAkkq=2awsE5u2Jn6gL_$U)!o;s7#wgS3rxNr{q|diM7g+ZaM)Pe$3SCHg?d9k6o_=Fb2-pS z>ZmGrj>EgWc|C5M(VA?!Wf%KMDC$S^f z%4^__5miK-Mp;v2-f)!b<9;e#%oj{S%NP|7sEVIrTj3rYGY|FN(GM`a@X!x9&mQN^ zOL)<}Nmp5}k~%set;UQId@X;6Z?8o)-&j|cfFIpZEm}O~^@QuWF{{J*TV*S^ReKJQ z2$Fp`p>NAbz^G(jv7)ltlAU9Z(`EfpClxgtnHH;6fxqMVsTMYFv`C+Tl4U*Pxlc2B zVsDP9dtbNfZoTeqzy{ipfvsE&+`^L-*ph0|;gO!y0+~T1t)_zIGikhN2uOd^-~nBbVucGfrbCBS zV7lPlyVL)<7$L>iq$#)}o7t+?$JS{aBtMf*!XI7{7yvY_|P*fmoa)3 zqlVC0s5(#E^&r#KCU_b2@=M2b4M`oRak@Pfyfys>WRK8eieg98TXoYYbNx2e@f#l-uN@*mFIm*$0e zP_(2I=5L?cA~>%VkV-V;vR2z7&mO8`{K2*mJS#8qt}54DWO}}(19K70;i2W%Fzfi) zwq)Kwb7|!|Bf{;t>AEA&!yvhC);+Hv)`rzIrCKGCLHc&ci?#mO;sdxL+zsu$?D@wI z0c!9}C@zU4|I=%~V%^hQXSwbv()ss6G;>vkRi^4Ti-6pO%s-rTuo1n!mz!tr@sUAy zHp*>s#vwn}S3iR`V~06o(y}!V@A)Jx?l{ODF`wb2fSzV({CvIR4GHnJy{UGi@%?z} zNfL5=HrFDe$Z$MS?fYEAW@3VW{ncJcZO@c>%$47|`J9x*QO~pu8RZ`ch?>6^C*l=h zLCJvZMTgCpag z#YAx9xRR#rGKldFK8#%G59xbJoM`IU2+NF9K>( z1KXJkw0Qae6^)5?gR3zoKL4E~s53OI3P%bN8ai&g7?uL9`w{i!-G4Uwuru@fU{dKO zCMvpFZ`q>ha*s+znD<%$y&Jm<=af@;Kr!xt!9|~V1k6ZC+G>o_Zo&2gw(9mRF`x(2 zXBX4EB{m^D>~r1&#K|6KSB{an@h9jXZk2BO2{rRWpHG_ivE_%!!n3q)DQ`K3K%-(mW$6qky&UbF64*D&#jit>R&h zk)4Pit3!(Vr3J+g1y*zS&;dRqqbKXYXy>s`K}m+#gy06Ob#Jx>w-?#T>XSF7uf8pB zVSzHLXyC2~rzUXpH`;CyFgVgs#lXJ=7K@2;V*E7Nz`|qv2H7Foz^WrDJ~Dga^#o>v z2$Bm{kOy0GLycP3W$&83VlUDM`M{Gy#Q8=?(?xc)gr&+C;k&~>A<1NA@-eFSxBH(p zQlyTwE9sB5@G1(4weP7jz-AeejJYz1Q#}mj9lzld-K>ax7(R=xKR=%uvFI+bCnN#l16vivGV20gwvq z9uG(;Sf8@nvsNhjh0FoB?s*n1jq0bVgKJrt1uL{Zf{GCUYi)3fl`HuEx%cF$pWy~^ ztRuJ1ATs61dK~Mw>qk4MD-R1!c}xDK8izMPU*oq45L?M?>-QQ1qnQ|uoQosKyyi_A ziOv^wl;l4hvvdZVZUX3;I>p4g<~_~=A+@);4k_gzyXoT1{AlQpn^@7pbfNI&Y1}TW zhJZoLY6D{(Ww9q8|A(^O(%%mtxl81L`YiLgh5+mL8bMJ4Hw;S&`}s#$DJh_GM!&Q3 z+C+x$mlWA~a%H!(UqShg-GtTb6e{CGBIq^8uug&PPF1RxnD!*JZ!{cgk1^a|DRJ4< zwo@PhU{XY+7s_HM9~r74<&XhX`ucOx&)vo_Uh_`vAJVJOtKLj=WLSOKu`R9uQ}$&j zMhIo%P|LWC#ndKq4X4sNwI?^NS|WU^HuD;!VI7aIVf`9JZTZH3n6nnDhp@T^7V<%Y zvWg__nda<-GB9h-A-6#y$MkPpodz$a6K54Q0SM=B}6@Pu0>_z{S6g<{pdzQuYfW41e$mGsJM%SQ`P-69*=qaOvc^@fpT^*dQ;2fDXvjgWZ?)o!@L9oEJabFY7rI;as#=t~ z(au~aeSIzwkIC(+MzS`=6kL`Yq#2P?Ch1Dfm2h=2wCci4X;#%Q=u*Ohs!;G&HD~28 z&0T^sMl{%N8Qg@>P}3Z-?VHZ8Na}j9hUAGbic16LT;Ps-De5Z;agGT?X{Phm~VPd-=)VtvTq3&4iDys#$X$?Vq-!yPCHY zM7-yrDhct<|1Z})%LkgsH|HhEmGyF zLlX|yAYxA&>i}MNNH|k5QJmB(Jhff0u#LG}+7QVUoXz$R1c5~-Kb*Sktd>;++d}#+ z^Xd)gVrmRl9N6k=n>G`FCCFpzA=aUDWnv{gr(4Ex@)G-rnNsj&Xw>JhqBZ;%o;E^9 zMmr$~(gt~DyR&Kpw&`Abt#47hd|kD21n{5T1a(CT<|Tq2_Ek0NYDQFm=Mii&kb_0k|>NuMP`O#xpSw4Ls|0@1o&Mz0U%X(}Fre#%cgxeg2( zDR~{Ua+K>bmemv>9~`E#VA(OEcfUmng#2QCY6Rx(sK2(`l?8|^$@+N2PE*&*zTEnn zjaFj#8es_sjG&&)2de68FIVxl!m4_KA0g|Jg-UE$wXZ$tsGD=e&L-~}`)MlDudU8r ze?m%@c$0pG4!XxKe!li3*HQ}q*75%D=2w+Qhz{4vBEC#QJ=$n*l4-7V!;SL6x|g6& zr{SN$A489U<<2X3js;iTGo^)oxZl}+Jl|_rQ)p*CI+^YCNNv&Xko;{#W}>?%^|T;< z4mk<3(^nQfe4s&i^|3j1kUMRi?=B-Sb;*bjdU|N5q(aoTt%+|g#}M;9r-tCp95Zcu z9rYdC6nwsYTByNqs!15H3D3%_uodm<1j=sD5>&gGNE-8L*VpVA5GiV?^2B#*_ulFS z3Hr%RWl@_~^&KCSIgt7LHjZHb)q{B@-`dkm6I!4R=_j4yCmd-fu09L5sf40hIV6G_ zmZ<9#s_fTAvCA&`KNk{ZEaI~G_SrYq080W#hS$5fyL{~w4X@NRpXm$x%r>HMxZn#&OZiu4Dqg`L0&5Gw~qh4s}*kS0lr;6=$p6=kT z*b2JMm=X$}&|R`AD;~l67ZoZTi)@8ELwI>-to*u@#^Q~MK6rtsEMg_(DBL8i%a&um zKlgvuG#)Q6-izoUROCpIXP%n4YJ&e}TO@*ZO@yMpVpG)&W5AgfL`+qp)d_~v@ z5Q`Kchq?H~k3OCHw1WeS+_K+}bOZua>>#W22=JF5eliA&O7ti7R(_!we=zB@w7z4T zYngQi_x;{SU#mnP)FL(egpIx;wf;(0KsJA#$e_eJo%P*<+k8u$eLI8D*xC-|+SRl#0fH;yZ0q5hGB4mh)str}{ShOzHY~))EI$>ceLK2DO5l zwYd++$9ZI4(!`^O2iw7BzY=7F+e)GPWq33_O)s9V3p&%icw%G{*4V>w`|gz02tiPU zim&`WqQqtYxnWcE65pO0x)i;xx&rX5Z`k&;gp`e&p~nP-dvs?eRNnH|)tunGk(hPm z=AYLa9>f@0Due^r&9;BD0;rfBR`t=H-sRC^PgoegyM;E1!h2|a4BumwKpCUnWiU;7 zyhM#JTMJvpG=F3h!~b%c7_eW5pQz4S<)Ko_rU)K$Jnu7P6yNH19k>d?F}7HhP8lt} zO!kFN?R-AvJ)^Gub?<2SO~Gw7M9!#{lV8u;rEh;YlY$SD@gr(W@c(C{yY5=xo=?9K zdC(JgZ+iXut)=@nd`m)3A6N zd%Wyq6xAG!1>MBlVFqYT#q1_Vi+Gc@1k0PIa!-+NT*tuTOnk^IaXj8={;J*#&!jdX zq3QYE_PqyuFF{Bnz7Mvvh?M97h6EsKy3_pffjVNhBv#kEG#_chE$+t?#WR6NLA^(ru>EXhyOxgH#sL!$yOzE)d#H^u1mA@aaapJd{H_LsYZgbj|1NrQ|MK@ z&#ycm>18gl&6v(^n~e<>G*(Sdc=l=k_VEJ2h@oCF3`H>CHZJ){#FCWU^|R17k+YqL z{kFja{7-*L_1icJIP=6aR9lj~)}IG!O392iU6U_D_;_EN0xJ))-l9WE?z|@%r@J zR*4B>j@I6>yJ1itFg6je>l`2n83Zur8yRRYq&!nKMhTeCvF&KJW^ne=K#l~SI0fcZ zLgz68pHV#yhYfMlv(5A!-tL?hxA|o97*hnZm6Z&<{8pC+WxqLMb>1}}GpV~*RA-%=h)U3P*LuA40qeCnJS`L?Q5PCJvD1H0D%uncLy_2cDZ-)g-r7VvCKQr!?N^r0@aBs=F+eYZ)5wwU4ArE4{?) zRz0uZ7PxMi_mNGQxm5eBA4@x?t%RNXt2|V{x)u7EXNZXuuLeaY4WW40&>rpz8Y_|QqyOqw;wFB;q>G0mc(58>-4Xl ziKZs3qApyWmp{%|@CD`|I)oDP^{sw)yL}xX_&F>=o?<#GAm7l?SGFvvBd@$u+#+1Y z<{2~81vd4A44Ox#ys*o%gSc$C6(0D%{sanm)dI1TGuHpOV1@gwEq4F-NXeW=6n*WP zJ~{JDOz%TpLJ{2CQKW6I9H2v{)}(1*W5b~p8>`cBP*m*J8%@(hL&a&ZSWr~>tzW@9 z)@&$5@PW{wzFx@6@0|<%GTRx8UVAQ_Jr8l$D{%XSWtY_()7EB}RqosEbylk!2M;$A zQw@0&k-l_lta#u%rLj3sHQjfqty;0`nI&GNF79EvZ% z5^WkRRu&ainDO76TWJ~Mfa)zW2A)&!bhz1za+yVA$A`;F~ zA3$LGGZPv-uosG+YYxwk6}GDd!rot~_-n{K_W$S%y&1|Z%RcpBhbs4_b5xYMSJ34C z#tR;+j^DYH6`8Cacv%AHP|POJ1^f6{}9z+-WsoXilsT%!vcoZnmc= zLSHy{Lq{dv4vbn>664^={H3L}F{TB6xrpm`4NkgcE-$lBAm#h48`FtEJ(=^<3BA>B z<$Q10)67Y_ zLY~#cBIS|h;tY~zSpR#9ThBT^q4d@wB?MeXLqGjpIaUbQNPQz^zp2B7=Q5&T^Gx`> zF0BoDn6q+;HE}nV2i$Wl-swGU@^ArBuNnpSsYE1pUHVNHN40cb~ zqaTX9Q>junEwuKNY2@#rw1o4O-{}IVTtC{$1IhcA?+JFjFPzx0tTn9L_O?DsdA_rV|L?yg_T8cVX+pF2@gi0t9mV2{@(pe5+R$|D1HM#;TvU$L;;zn$9FslJgjw z?X@=-4W3)fbMI&Gtt^T!Hb!KfT<2#O1{UQVPM|GXbS5QT+Eiz(V+H3rw4G_@y#EJ# zZy6O=u&s-dUVW+r)kxW8tH|53GM!VPHvuL3;Lz~O0R9qs>Bc^~Kzp76BalI5 z>&Z3J-^)LTDp|(wDGLDE^*gc&TZWo7Gy4+_x;D<2<;jX=u6=TDZ@w>NBa4UKVi~h3 zu|g&vmoM$F!=UE@t%5^h%8LQ=zY#1iv&LU6l)MJ`7v5vr_uAUfn8wb*FtIV2vn8M3 z93$vr2@{mF5!b3r*a}(ql{spuDfMKzsDC(dyVKPhM-Hb%F5 z7`u(O0Ua1;YG;is%W9Kg1)m>($jEP$CBe_Vl6N(LvUy&P3*oE>QUFktbvYAX84(P* zpi6zmw&JYWz5wriPPUTOZJxPSa(nV@nOSiRpuyCqgF4Ll(RJM*=DHgU>|5s!OTpy5 zehvnW960>igZhoDya`_SLb@1c+Y1lKDQ=iv+8>*9N=Yvyoj7v0WAlB1frIYyW`3?s zF(?Oi$;;KyB}sWgwb<%e#EvXS|U%Y7C$x9hAfEf^2xzQ=;tkueEd+kbmoneRI z#E0OxRwALa!xsMrk#tKK!TNn092MO|vq0|Fz}OJ+>z+o~l|!XnlYBuACk#Al9w^yp z%0t?^wH%FH*))`eAM%tw2?G^je`TLQJ%pY>X)$cHXZ^@{sP4E*PIFA<7A_4U4quSG z6C!sJJLcD#|8ekY&X;t~UhXxtkFR!o6<|k<-)zI$Q>Nv)V(#*eN2su7nc7jLGy8>@ zb4;dVA9|9fm>+h07S0-ff*RUVHzZn2rXxDZQwQ_bHQr`SyZa^xiXN})UUHHM4fXsn z*JNFC!BQ1Jw$of@x`DCuWMFhfb6Urgh=2X`yubIwTA$8Ep6m%{Nm4N3>o-+k{kJVjx7XJ&jY0LBbDzd zrIH7}t9tnQxayXTsHxNQA16KikF#2K9WyHxh3CFTs?1S=hjwZ9-&Ps^BhtCrxPYRJ zeO*>gF&FGrCNBqA?cPAKdLW3|vBoxC8tDL1QPXafrEYj=I!nCXe2V3)+H7+#SRP0o zIpJ3Qi@&b;1rI4|W4enL>aOgNMNv!*f83>O*Z4>H%JVZrw@#Kyu%0+3JVHB=Q%73C zvA#$sb#67rf#=9JSQhU*&S9>&P#jw z3RZe#v_hTvzTL?LfRF4TMv~huoH5}d?IQKFqT`3}B3-F(d`aM|UlPasoWQFtWd$9Y zu=t7LO&3>I?C*>1Yh^|}pHUg%@l37t=n|!Z4UV@0s`3s0$c zvqS2+ziP>HbZOz@+@B*kpI-Hl zz~#|z^bI+FjS{!@Pimc|Ruz}*TVNqqQd8pO|6Hd|FUAPb0yv=3hUF z79d-{BP?d*8S$SL-OV2_B(Gp`W)&6%D5Z(h-TSX>=Zu$as!^iwd6Mhd1i(9**l)dA zQN7s|yZi52^*8l>4^Jn1UK-E}l1+UGNFQOBTV!}+hFixAy?8T&ycL{8MqyS`viIye z_~}?cFHm8vAqDqV_RH35XJXMAK+4!NS9nnQoS^UOX3l8T+Me37Qt`j|F{wP;XG8)$={HeUlgD}e zv&U-)0zt4XbWW4IYnQf2bvph6dvBl01tzEM2>2x5lans?G1>Pbd1VGLQT-LGJ^+ zevr#Vx4beg4b5_>>HnPI3v!c@7h}?KC+LU=`F}!(>vD&FOwh`&;PbLV@Qir?a2RY@)y|#>cw4_TFFNKQn1$9ZzlxZU_s)-}$pgWM<@C(FhhGOvI5|EN z3Q+a(WO@&1;$I{m#e5Us>A~lg)k*Qm1i=$sHJhjH^?Yf&$5G^Wy@a#vF4<9e4*1em z1N%Qq%DQ2cXBIR$T(0czuhd;TwPv4&><44z+c1zr)kiix=N+&M(Vb-ZMfodhu zpIWf_By=lE@Lg7VNTFYbW{aI_@AEG#@k{HtN589m3|VDPB@O637TafK`;G9E)xf)b z)Tp4+0V^-APZ>$M%YnX9c53*o!l;`3&UJN5V)LQp=>hLDLyVT1wI#X@!_0M$>!E)} zcQ!|7Zd$h1ayOz`XJ67wrUCK&D%ASs#+zUFKmsxIF*h&GlHOXktGlwD(WZ1*gZ z8g>{-)M%w@Zw=#9`o~|>t~~YirpZ{RW=^L{2?IS6ScB_!4Efhlaz@|DWAboJhAG@Y zw%P{#Nl%s*JW=~BcAFlc7Li`+4a>4StO4%L)r4u(-r5-%fh(47G>SV-&_w{FL4Q8S zK8V#WQ^!mw=RCXR_cVGRIP~5t2ENX(FHZ{ z#eUWO>||Gysm1>(}rE9*X3TSCXiKn#T=`izyb?@Q(Af{cCi!~Al z@JywrLwGyh-H&`F(n{`xW@kuY- z`&6F~06*4;&4}Pb=#{RCqJoMi2yPK2Bz>!j3MoI~xeM?ITWNDQc-v46JUVQp3nv03 zEp!6I-y59Dwe1v&-u6cw^visc2p7ujNhCL@@Mq(#p1uw6bg!vO&u(=1ES^3rs+C96 zNpx*ArF~XX&FZ8*xH!b;tLeC62Y_3V;@@ zjV;{HlR}}(Z*6RxJ1Rd4?qrNy5}nhGzrR>wdG*!0|P8`Hd=nk?V5t%!|c^5ZgN=SKg5 zHUGB$O4u6?GLgdTD@$`VukD_pRO?b+a7jTf9?xya7CwWmv&CE($~v4#SCHIzZR>@+ zAWcT88l3>Vnb0ddye{Pb9CMDPJ&NMq04O1C_z`^eC0sA?m1{YxEEzKuK>5WOcf0BU z;&v(&iN049tNO~EHN+)O@I8C8(y`z=@~)oecx*&w_^oKv=Nf_(9M1TfVu_#JY9_QS zQ`@#=MBP^DDZ=ZPS+R#hi~KcF-=oPlzaV1x3|kBUaB-|?RpFso(692C6wJMtz8N3n z?~?ERy5%i5BI;Bwy?$j+6!x6dr+EQzMajrnc7Ymd9-gF_A!Q8f2d|cHHk>0)^$wqE zPAWP(w(rTy@(RD2mbv=b@@80LO*=S3d#;iijagJUCqzbYj7ay;0YOBQ<(|Dm%^P-Rs9{^iCh!k1W=NpB#&it= z2J~h!lHo-7Bp7!t^>Fb{fsGEVm|1Sh)S2KiJR{QJ#Mm?K5P|TaUeUFPtR5wqz)zM6 zvt4%SpSB#ldQ)d$qGv0@i#6*9%gOV)H`bgnE6Vf9F-`~D2qt_Ygs^t?d)X-U;~upc z=_ip2N`!J`;FbUfoiEze`_sX_;Zbk?X@~==Ez$(hd94|6+{lUV?c|zg|k(uA1C?&KZJoxA1SjxBCn8IQleOc$%U|EGn zUC3yBv%1}_;)Zmkp2PR7Ec5*?nYZ6i+$Vf03iIYMX}QS|UA92hA{iO+?IOuP<&u}% z)}1{70MLq6Ix&ozW?!?q#G{yMAf@25F=XaMB)-@9$yIV1v=lGt8zG_G0Yssc)K&eP zHztCUp_i_2CQ@^03AA`)s3#lt^u%w;A0#T`?m7JlWC=d{99vo8A@Weg4DlA_)7&d( z3Pe7%0Af^E(aia8tUol#lR|?ISiNS>boZd>vm|-$H+t|zu1^mjz7w$Az0)pC6(2$a?= zVxroeMxYtoX>#Us@+fqH0yuo*Y@=PV3X?-yS-OmkM>&20)OiB`AQ)x>(zd;lUoQ@n zwKMDul^1=}6H+tjku+3UlsrV0d+^Jw`<0v^>CmwrnxBs7XxBKZIi-M&mv1QQ_F{mu z#?Qf6+SGaT?PKd6pUwMfKX2{H1!v4W2kuZ9*QOjbT#lS3o42V!DD4xSrF|$rD+b9O zL8gXvwZUGhbyY%wHC8`gP0s1f>H2rex_V2P9)+|`lb=q$aDqdETi{+$;)I*)#&}{j zKRJ=|+DupL;6~3FR?c;~kAAnNG6V8(T*ziNhA(qz-WEK)HtcQbDG^E76%9Fc*&XNV z{IEFN8Ob%*jUGM!Q4t2&SKro@;o8YSq#N(<&11{M`26&-eA3Py&$E4~In92%l$1Ua zkU0J?M&$qGCz4{+44Lv2ay(vOagMpon$mE1Bi0O4yr|E@4(@Ia-p3bz90y<6y6{vV zj9??DQ5k4|JnXv_tq~DlD4ihm>B-1AH^A`eDHB6Wmb0}*hIR|B8fUVlmbV@Ozz(Qq zp`11O=nF#K*gn^chg&c71Dcm=&SzMAZLX>2oi`$ove)Ln3eFl#zTl8VmADAd9$Xy> zAmBQSl@4;j4t+|Me6TT1h`_(Ex0;PCQ6pR{Xewwrw0x+Oh{Mj-J3P{O!F=86D>Haa zS9q%5DcY5IZeFQXPx>V#lF~&io1R=4zgiAe$;ZsBJe)8pXa>dPkg+6P)s2n!(ACU) z>`;rY?P0X9Q5g0m$qJ*^0^4w#vJ{QhWKglGdrs2;h+;<5jc4JJ{}g*FM%hB~M<7Y>wBBStM;z*ZEo|EE z8W2;-uc%$ysy$n3q-8*cWgY?T^Saw_%hf93{V-kuEi_-$-mLFN9c@BTRf{X=@nkNx zUjBlkutC6TQfmJ^q=u=aEL>cMf$W2<(ImPmj(yqQ?)IQ<<0%a(;=cO2 zn7WQu^V{1I-F#Pya$Iv1VNYo7?yU(=4skTUErgA<@7>h!H|r0F7GA0 z-yer0so5+Tz5KLqr+-n05hDZ>Dbbj;mt9BwaJX^ zxGKWKDI{>{1|Q%|gOIzFV-fyX{AuqV!XGP4cdTsGq>7VHyJfPlXR~BzQ4-#6_a18@ z$27KfQ^$MNDZ^6~k(n0D-uMZonAqen?_-oR#=*hFW#!zKU2fTvZa>tvR? z#UBsK^8C9=NVfUrzmZn9v@fj-8sGXMU^n?s$P4^frw_l9Sz2q5S2qwD+yseB9|`#L z>s&yF)eyFA@+uMvur=ow;n~3nXXpGAyQ}k_#@Il(A!6%FwkGj6!rS23Ge{f1K^bo{ zZ>T~wX^l3DW!-+Bm*Ej*tMBR07UL?NMRB8`d{0{cQX2bpf5B&@gU1^BvdCaQ$F*>Yc+cAMH1F|VZ5!7Fe)As{Vk?i0$hCv9s zIuy?V?9w^yNz@(UV=fLZktvVpR{hVBPJ{IsU4caN>|UZKt(a|Pb_xcWQ%?FmcPOnQ z5p;d-QLgdu6>nWAnT~D2gx2v=)OtwKiZL%%SAESpdS z?>1`;x7uCq8sovQOZxC+iwiEkb$WrSfH_>=PZ(Wa}M58gq?eEOML z3Xy$QU%L7R*46}%4N*Uc*}IK5kVtV5=W5=&CFrh(d<%G-+ADUj{(L$OoxGT=>7S+K zl&rYE07#Jgp;jYy(opc*BUvdR~ZwC-kDn)Y%dC38$6hZvH zlA|K;YN8KFfNP)Zwb7 zk(qUFlkE__)vfv<1(R~c=R$MlCi+WtR=W*51-0=H4z4GC`d3815h#FK<`%%wW;IK4 z?A{3Zev-B@uB-YRVgBSdLPjp>112<4(89dQIekT?Wb9&ZVGE0eW`r-oMS|WwC$Y=uguIe@i%o2{+ktzwQ?R#J1?4U1 z(O7}Pv7-|*Ts^UeSNQpp#A(chU!t*`CpSov$>u|&r|UF5t%id!A#uH=eZ%Xft~>pQ zIc(W_dIm=M)>WlnQi$zjFXN|c-q4yNnMPxF^YuvgdQ4*+UhOu7_V9q5p6*WB z+l~vVuiNIk7l!y*S3l$qH&VkFn%(?-oz{y)+C&r7mv#>05ZR~) zRl4h10^5)9dl~lZ7-v$9RP zSE^CHw=>6D0`7&(7h#0uP=24ieeGKa=evFhe%}**(H;uhUD5Iaz_?KE%oBJ}lazh? zI;X6zPD+WfA}52wlw!^#l71N?Z-2{X`E#2A7E!fymT_Z2S__C%H&Xf~H#t$}+xw^+g; z?bfAYL2$;mJo1wE+Wvp-rhrH7Ux#n}KS!N^uaRT=o0Ha0oyTA)ZTC(}TFNbVJxK~l z8kqr4N>7)His4_X^WT-|fAtd!f>7bnQILo++9_qI5-_1uA#IoMm`w{UNvF?UO zemt2HVqj?2eGXK)8t(fthdbkH?UgAD^VUprrw(Yz&tRmO3;Tcy0}e+Ql&6H$ztj<= z=l`UpnByWs<<ZO}K*#NOSjIQo~*dVLyOtS#gKnXgV7 zVJGzvzM($;*m7{4Q)|({bIZ&bJSz>RGehUaxZE7>%P77*D6OzsxKGiY{|Soe>;c{ z4BxnBS7m1t>?6HCX*xJZR2nrvNv;z!km8jXi21bL87{DpQR2!4W92T^(*wk)`&QCC ze)u!a6EVea$M=N-)kuyYlJ!^J(L8*(A4&|x)JWeosS71EoZuY`e*YQ!88?2dX;vuR z4}0TNX>P-IR3TX3-h-tACsM4NW4c$>w`br3VK{-i&Tjf;dGq~haFep08oYn(tX4J- z9VJ12Rq8MJ&tJXE75*Ge4R%6>IxS&0p_$VsttO4f=I_0~VfuPN%oz1ZUmMO_%qzO~ zjm8rB%|7kD#fv&I{<}FZz2L>CTlXmdqKpO_rZ51o=tL2ACH@P}ew(WHFv1pgpZ4cp zc1|(Ee

s^Z0S=A3r_64t9T3s`-sj`gxH!Xvww_vLKFq7~BUox!}nD=Pi^+iw`!_ z{Rvk8T&(!#;(FPuH-O1YI~zPPm!7XOKgf6aA8v{!4nrQkVS_&wO)L_7Ep-7s_WbMm zAF+oYJ^pc*&x^RLF)T@w_4<*yMa+S)%)@}83&SgZ%xK@H8*JJlmZkiOEB`&}&<3lT zEb7#qRZ|dOwg@(btuA1w{L5g`YeRJ{uGzdAu;9+@aJ_x91rD=%{$~}K=d#369udr*%P&aU;N!}H(rYCTleBABDXbFR^-K`bOrw0N=CbN z-5JFV3r6lx2;n92K~!4t;#OnitfRM#7A&osc$)K@aF*3=4q3O@#>!?uep6Am8{fq% zC->{zN$G5cLnaQChKpCPx`lD-0b&vB4%Us`uW&LYcAD3(@y{m zc`I+B+&7bL&|CWKV+W%c3-?3^XO0KvXN0^U^$b1X^u6F3^gY*ySAEAfQ%(m*gTc}} zV0zLXqL>3$|5QU?Kb#SvWHU2po+rH};xF7Wfv)G#jfhDe#xDWEEhdP)@$HyJR_hDe0xp2%~)@z ztl~ujI1-Iv2eQ>C_`>P_oaUzLS!|Lw-h8z2`&X~-fthsf=Z>Uojp`!%jbN~ivezRX z>Sn%9Uwn~j!g-WQ56yP1^u%;*JAcVJv;~s&%OOR)g3hic)CmY!fBSMqmA~<%%Xe3l z${+_@^c8{(ANuyK6Pt=)Mdgx;L1?hU_-HVDo2qkIn8&S` zd>$`A)8_sjm4>$^LFY5rM9WTis;ot9C$LG?NQZ9WR9FDjHZanIZ$s)$7TFDg z**fl?{YE%1dED_^zVZBxFw+Cn92<>m)l5#d=LxCJD)lj*%I|})w7R@~bw?0sAHP8r zuau^MLb2G#ZgBt-y~sXz1JKmw6pzU-wVKig7JW?~K59?u5baXA3mUt}u3>+-`;7oI zZhSNfzT^bbnN%0K0j39b@w-<*zqw`e0NZXosKi1;N)l1g2v-z;f{-``gE0O#!Uyg6 zzjxmeua5nVpb)eJS-< zE8QQ(6aU`7w8P`T{l0_!QVr-?YSIY`sUH(P*or26*>6UVmYY?Z(t7srQJ-@y2zD~@ z0>~7i*njshUO#+#pZ(8~XyNng%c71-J96E?bQJ7kM*nFSI)8qEl3kJ<+tlu@r75%Y zH3Y$4wefw>#fS6=p())=lB#i(CbOT>=Biy7Q0+Y5GWvkC{u>KlHut6drfe<+N7rO! zTzs2($pFCOR$o&ROq zAS(GN>?UBEAWq*!qW87qWjnn++>)U%cVhDR2MHWLfVHG2Tpp8frAQ%rZ4^-1yt4BMp~uZ&H~`Q593n;HUL@D#qbTM@JxAUWsA5ORWFoc;^d` z;rlfUmS0i2hUU(Rh6oQ&6sp`sPAa1i@^~Hrgn3W&AZDp zZ33rzv4!2MP6mO`r;_5a=xt%Wxn{oM^*?ibZ=jD^@&#6?)LL;KHZqJMGI7s;rd z^M36&D=jFA)b=*+>5@S5+^q{09B^@$M3qepkCOG}JCMRg!p`XWxA#$0r3EYp@^o8v*5ygcs}?n;vCT z#E)F!0yHYg(kA1&hbgAyH0}{>7V!cEZUR2$JwSH4t7t^V?1aKHw%B6AF21lIrdNum6h8KgKQLDu z5J3~N5{$ay2Hfw*D(Cu$kA$|)xPlAAB_|Oss<`uK@+J+-dL?HAz<+g|sx6!V$q)Q8$SDwOslG1R4H zNA>Ix0I08|Ui^^_vfZx)c~MnOwSV7*fD9V-Sn2ynPpyW@V3=;XH7N69&>`tCeg@lR zK4q!xLr<(Y2)@z!q7MTKrf-h5g&U5CU{^cg1*okdlL<5KTxfL*%y5hjXJEIbf@1Q3X_lHf7E|Ndnsk0xi41MT?Y z!B`?BR+RZ4WZceeiY^!lGJ*>OA0NwAuc{ax_e7+Jg3;M1GQ{5p0iS;(2(wY(-n~@z zm~L2h{1|ts_HBy80pZmXd%uvmPSOf{tbH#PzU|i7yYds9i%HL|BZ_K?1?@H5h26;U zqm73M;g=}yYo6ypv#2pXfn~AZ-9XoW{YKzsye@R|G<-7PZaiW<`0?A>cC+eX3be1U zLosm+FSuGdLus1D#h({x&}G|z)bz1%I3>0UJ*twhn6+(75xFhnUG)S3ANuDBE-pqv zm#wb?``F=^_-v)lgBj*^ny|*NA$cIVWmV>&khj-(8Xw@Z09AbS)h@p)0u2?T?%6Ym?3CqTBTLVO9G!hdC#@3GEsAl9GZVPvPSp zzA`EmZgLrNgkxb=r(ur?NlmUAZ@56Yi}~(cCoQl14qS~;yQMSLq3Ow5Pz$$LkD4Y? z(6ZH#Gl;9m8t>wcdv@2Z!QF2GYC7D3=()Ro?z2iiU@qpLILBv%Xw#JfuQTBr#2#yp zIZY&$Jm_IqPPr-c0g4x}OIUAXBU6~*h9P!xH>M!%b?l9kSI4@-(>V3e&G6BMde-i| z>{Bas)NTEC75C(w_4RGnhGhowlJFMF{~<#;B>bK9d#Xq@<~Lz^mJ)y=o{Wo&_pjaY ze?A8BUx{D*|EjH2UStSXSr;y@^#30C|JCbGf$~q3?VmZ0|1);NKm4=*8O!2-xA-&8 zxJXjKdjGGE1%XPesvX*#m6qDfp(ls97aX5?Y1%*h6_Y`3w6uO%LRo^h#-xTa(%>Fx z1*^+lWIk{%OGi?op!t5v6hen)dP7NNx-KB_mUB`WIQIQt6vO@xsnP#8w2gd91;UPB zo=L}%7U5Ln2@!Z75GzXDouImaSBYH7>OvV4E`}Bk_-L8pB}7^mlepyj1vC4{Ttxbd zxPKLW$CrPd7GERVNv#VogZwN&^^c8AM6$M&!j%epk}!svg@>()q(p}Xv>qR(aY_)N z^(WdBtf1jv3cx*9>Yb-vi?0YeaJ7B1v=#!PLjy)WFZ~}l+yASi2gzU$`A{`+@#j=* zqJy_CYy2X!t&K$jIDTO^c}+6(1D5>WsE=by6T8(hfML9ELI>I1W_H9Uv{yy~S@J4< zI8`uK)Yq4zSq}ue5km5n2Y*)LP8#!~BuA+(=UVf=B+IhNGU=a6@P5*{o0Y2X5KL#p zP6i_hv})$uWn4KRQ89T%kLX4vy0+9#xxxwr@fZ&5gF5u|0wIwJa78213!VuMFUKUr z4!$kXfN33=jh~V1&pT&)MqbDAO*u1=j+F}qnBVLd$+#fp#mt65q}q=Au0Ad!)bS{7RC!_q&*e@=*KvxEK7f{O^WA&L;lsTFQ?I8>tcGz zumTizM8v136anB}&%jcE`)1hh0~fuVBnkJLGov6qA}bq(i$Mc;8XPEF<;j;;UY^TC zAa53kn8_^ZtBc@2C;z>(wKcJ6ee^z?v3O1Ke4qd3OJ!sTQ~DQ!xgjsP(-mrGi3|C@ zW6h@GZ{@t{a%8G`^^C1!5|ozjTYX{JT|dSG2X{V&(3ZGR z#n;r7o{Lr>&QtBk9sZXz1SQa_^bwlP^UZ|r#) zalV$#9`6z;YXkW-w+rwm=Rdx%Anmd^ZNmLD&V(gf?(^KWT5spjVM2Ru#;)9~=;Nq6 z&Sw9ic8fqyigd(;Az5YZiP&gvws#TOKH4$9{z+K;hb=)0SEEL4J(JEyt4*11BRLbtF)mCo8adrHw<6$ur5|=2^X>TlDg`NtEfe?(9qm2 zb)n#Oe#)1bkmso60C3a!`6$2PiS1EZv^{0C=#Up;u)ly)0RODnA+hY0{*Ca?o>yv% znuM0rbsxx^pnI*GK=AVimlbbBa+>l5yACF1VFyE+(T0R=nuLhOkKosfaVY$*4K98U zdlaYgs~3@1@?KLypj0i5dBFi((bho}KKHi!Vr%jbhz)zRr)7nXtT?2BipWUGkZ z!1RW1>XtMAp1m&WIIRBs<>9b#)8#sP{$f>wF@;2D8@@kGeL`ncvv$nVzGeo&96}!O zX43^xKmeWIWwp>qoGkMvb~bdam-4`Px?d_rG=-gU{z;aszz=U?S>%edc>(mUbkC`X zY8|zP21fnMr<udJfyr2F5&r zH>okDj9wV?<4w==sPA6B<7Z(L8oz&=H(h% zJdV;dI3uPSmgN^w?Ba3Su7ml}KSJPcl174iX=^<-@#DvYLTQYg$Un!%lK&Xj{zsbV zzdVTlANYzOAgB{(c*r3|71~}&hpkls6sa5{>*YZRothU+IMOaIRFa{r_5(v~2-nu4 zFP7b#49YO2+Z&2N>YT$!mFCqOaawtdxf5)L${{TuC+1!#Fg3HWlymiI-ylLZFG0Z( zn{3Od;`qZcn+My?Nqq=T0qOTWXsEbH4m@$$x-mZ_^wWnMSEGu6x7kG~aD+WGZ z=!cwmG&X1(9P%c`B)ymUE}Tj(+9X_#r>O-g@|~L+*#eUdga|$P^u*2Fyvc~UoZ7mqKQgQ>f8n{hfm0*1MJ0INL`#~wn{=Db z9X+fO)zcz!TPv%`E@Dwssz#X^WO&uoJS1k3>`i-kMB8Lg0MikR_9*t095BCa{?U^c z^0h4E?Y03+2z^HL$E~Dno9$x!$q;S$juz>GGqcfL3#BjO3N&7qSz1R8uSmBzIY9u& za5_#-)J^8DvhkY=<+do9q7gyK*!Y57S)XmG?G_4qH9XFsBYd~%G@J&5(R4q4l`r^G-MSW}3`tVx=VcC;#q!{IAUZ#K8ie(bfZfr-W5sm2#=6`V1r#a_sYo!>>Z}(MtRJPU+1Ay!}e)|oBMWlz82)i zPoAi%=8>q;(+HKTv@N|Ye}HRb$kXOkhoSW#PiaF<;;u;<350=(uX$X z4!!y}!e+-pF)4df=3~Iu-w05OnX29d!a$OfeH|CWD3{r^Y>Y4CMVu2q9yhMoH|DW# zM;{{YQVcs%98C6PFImfoR+7BQ`5(}lFcFbH?}TZnim*37Nerldm=r(M#@8nqHR)3; zt-Dt_^^||`F+$+2BuC5CBI8vQWTFCT`wA*XOl)<8;4N=CWdK1?09`)lnMQCX%>BXn z0nYC=M$p`usmNFq9OuoU@3=Xrr*WR5iDL3_a+2w286Emlkw@~jfW>}}4*`h^mSE*-f4Dc*hgANln>n2_Q55|@> zlx77Z`+a&%@MmXnSA#G#CS$N*$|z)I6EvXQk*_kcjq?Zm( zgu?L8`wQLEal%9$wDg-v`nln9gO2l zJzjXzFgl|?IE%p8sJBR8k>>rXG`BdGoN26jj?ZepOg*Js7^Umy&r4S zd)3f3XlC)PrGH_JfI2GKPvIqR6QT?`B>q4}yeP*$`RVp(w^e3z zc4F{m^29LCA>IXRWaXm6rhvGta#qW^!Dbdr&2VkggS42)n-+l)tUutCN@H#xW;gI#7#t28N@m*f`44)iFXEtmYtiH{Dp`i)RPbU5;~Hib># zMlYs{^OEaPMipq)_or*)vdM<){d?hTb@d*+8ju+YUEu}^n>1YqOhQprR2s7`ZGHgJ z_Lgf=ir?rn8Ng|C%^~4V@gO>(x#2yC8&)_6$`@Jz|mVP z^&~-Uki6RJQlKo%6|8}cJa(C00rxIu!|D6ePNn;A|L0l_fY&EOZc{LK>1nv7Vy3*s!wFPEpBzr^@=s zH~NL@xuqM{VkLC|T>O}Kyo0p13(H_7FiK#H%djXzvO3YB`{C+B;hDvK`fmgt=xKJi z*08%$!!2FUc&ZfE(*+0OkY)FXBa0Yp>k1MXj!HeHfP%v{BcRh@Kz1yax0i1Uy>As` z8vHA~cD-PgzGC@|oxZXW!sJnklXE}XM@yNYC+|ju1QtU#NnhZ1545sE+&S^W3Y^6@ zlga!4ioo_I)NQ&CZi37!hCT5Czlm$rk>heWg9C>wQI12~$(>{&(FqqA9r{N*FSnq- z(&AB3<-AhpxC#1{7SD6ZHTVo8zH@)S?%Bn~_Z#}vP=}|koYsJ|@FLm;G@iUE@ z#6QvkTgsf6;OC-pPwnH((%q3@=49<(CO!T^Y7NnGYe3OJxT3Iv^9!3;$5 zCRv+!3aA%Y56;mf#A`=Y86R0#Dju~}p3t}Ql=-n%rT)>u==)Dy@F&8#?&KVN7dG`UzOiSN0tSnYNP~bIiZZa{jFR4Wl@( z7t@VZWY=9Z9vxp>j0ZMEG`k(W4Q|wIM+`%cPI8x2+WPu&V5TifZflcK_C)o55(^L^ z@iQI0dP8If>7NDWm9LSV@3}|!C+9#}Q#;G(v`8tp0`pJI*$iD_cIFQA0}n~Sh<}%o z{<`66p`ZZuIoT-!s4)XJntypR>+W*)3yv0}uFT zQL0;$0n)?kP0VaGl*4g9YqGR#_cC^(Q7DUiP`)$x$Ia69uk%mOiy_%@@t^b zpruj==u=RfSC|p?#QB^5&;3M6KJ#NT#^E+%>4)*g- zP#5o+d|JcW#D4P&fn{s@U)eXMK#uMktBT{FXHySJeT1a0b?jQYo{}O1Z9%Of(jmJX+6B&`}58q6HfdwHn;fO@u z?{pYS0d^oAv7~XH)FlIO=QO%HoNo{Yvk3(1A?WG93o~+q2LE(eU!6ic^EoD`Ecl#J z-Uu;FM5t=gGdWI)J{L#5Ph2FaFei@d>Caz1e$GXE&ODc@_vw$%|Ges>-R@lMSP6+p zT+hoiRAxngkpz|52Xt&_*p!yyY@{uBcIgvluRgvm8_~bwlnZ(z4y34t>?djOWy}m> z+MM(j3K&?Gj%e{=Qdmv!z6c{pv>^K&zf#vIAdVc-7Xyp(S8iRR_I+esc2NPwCNXgZ zhP(-Dm^HMGW(oEXXNjuR+Ne~xy?0Za$gJ_TjnMBO2uN&6yL1Zzgb-NR*Unh@M$#0yRC|5XR8sX_4SSD}h~8ti^AH zzP{tXod$n*=2+gdNU1qDRSBvfGGN7|?T5zrJ6@!EN7?i}cTe!9Y3z(kUBcS;2FCQy zULJD3l(sr(G_3$6aTD1pSJL%PDJsL~EY?XXtxN3eSx3>pJ?MD>pqP)<7SX=tAPvSQ zBY4@y!ync5UA8tDSocORhFvPFUPJSfyRs^Z`e&h+{+UuSwz=issi8zah71$2nQ7Ze zb^czfG6cxTN*hU+Bc@t`RZj8C9-ichc%^X~!y%xE^ z^0r^OM&raTqTtc{ZM>V0fMB+)WCD`?=Y@A+U+O(LRaqQke!Q4!Lh){H)YWG$G?L1Q zDTvFsmVhn&MqsH^0ahFEf*8(s*w<_wl8())AvVl%!oWdkF-Mx?`6JgtMZ zL8jOSEoSy+6e(Cd-ds#S$pV`FT?dbL=cw4{MnQ!WNqk|_8cIwB_GHvM$Hd%|?|D5F ziuzoe4d$*31KUnfdx`eyi=SAN%cq3EMVhLBkpb+LkXWmuKYtch+uP*@(TTB5ezI)5 z%oJyIrA{|Hf5Qsw0W}7z1H$~*K;j+{tkE zl;)DdarFj1x9S<&jiI-RZ2yh2w~lJF>E1?ZDbN;cp%j-ED}~@5oZ`h@TM87{KyWXm z#idw^6!%axBzTeF7CgASyFDj;-uL|0IcJ@3t?&CYE3@vId+t3md-mSfbrshSAev_M z|Jxb(?+gLv=e!+F`olAF6zCsM$a@>A4R2X7cdW!I!1aqw7OMVQ$=@0MJ1Zz>p2?Xl z`wzPBH)$W+Mz9$#tTE7J1?bh{7R0wVcfnYxd34bF|E>qp>cA-HOV003MW@|By{1-c zx}0(X4*`ebm=LzN%)!C?0Beu(!L?owx&8H~|9)MtGiJ%-dyl0U0+L+*Oz_w460JWl z*7{(~pB}@U-|^tvajmZ@6Iy+D?wl4gru*lOcYS%vM>KNBDwSKa`kV#Ouu9$L1bJ;$ z<mM)Bxw2-HgS?_cMw@V|3Av)IC`z+Zp7(JNNA!);Cij=_qp%J0iJS*XTbt}Yp(H!fkf^`-91!4sy>^oA${>KXD)Uf8b0 zwlLuudi?sdLy4LfmN#}=M;YB_@HuMdZLu_0Q4`PVL5_{*=8^#)yN zh1PN47zruC@-hBoBcGHuW^MSRn5bv7WotdGN#m2&#Q$mi*NFaJoVGpDR&o&KUZ3!A zM5brwV3*HqA*F%vj*-a&UQE>#Hvr8jJ|p?p;toRo^?#ai+)EUH@=TB3e>Bzwhtwr z^)CPYn*a85?1Z^0CZ)*eo1T_$UnprYzN~*YXcR5Uwvjn$=WIw-Hs?jOve zbtzz-0(_3Svjy17z6ia9`8}Ptz;I&XVq=(R`tg|+-p#-&(?_k4*gRk0O2D#Ycot)2 zrh?y#e6YWB_Ky3wE(fTOW_WBkqD31g6a)UO4SHHEMe$&Ro~+nYFhcSF5yZuLQz5OS zZK_Yp%)|xpHlga`woItg-z-G6RMvmk)zKw9TTZTMM6J4T=VljrAUcbjhUt{+i_-{r ziOTk9jF!b>#Po_8BWF#NnD&?b)FVP|{u}Mtz(jiz59NQ}?xI?DKI|KnFw6u#V*3Yc za83jh{8_zyR`@q2v~fLolU$m(9+ZA2U3>pnoA~}56AY@jU|ozBx|%KwoM@S?)se0Z zc#I0nfOy{7-}PKFVgg3Cf7<|bpE-W`H!78Z5qtc9N2PS%QS1Nb5t5i1^4Y8q{)LUt zm}bV9X5(&}xRZ3>ii=&Xs*XOx)u`if#P!S$K0d-J8*2mb4~2Nm(Nxyg?z*(u(n@Ot zAH>LaBa`7!BKvR+1sU;DyJg^n;H|1h>X#+ypt--Q_ z!tD56_h|3{m^a^`!Ze`dYvpf&3U5!@bdGqx#fQtwL_(}qVYO&>G=skv7ZDhXm5FnR z?mHaq&aZy~Ce>0?;3w&Fd$0+*;50n$cSt#&?V3N5d@O8#>8zqFpxU7=_-eNGY>*dD z|J-pf@AR9 zq{Cq%rA6F|aBGiY1Q5scoy_~MA$%cZ89C3zyIAPbr`g9BSjBnl5C+gH6H3y&_@;Ef zBhZ7&Sd89o*7XyBx+* zj!hV_Qr16MHDiG_dUjA0R17YqpZ|;Y`wy`JwKI{p{p8*B>(+L+ zc#Jlh#GBO^zT^|4NML&mtmURot^^B=e#&2$evcW6X?JEc<@aef-}#G(A&F^B(A%0{ zXppdoE@YBEkEVkd2h?RXlhEX^1a)tp8ufCv9O9IJa95ggTU^0^(Zr=>sA?ZF)y8Ky z)!{ZQtF~j5n_}`fXEj0<%>sU8yh~Z}O8&9%Tk{ywKbC9U^Y}ZT zAUQFVa207HJQp$Y{7Yvj|0=jkj*F-)p^@KewLbA$KZ)^C_b6YSUyADJ!PqB?yQe~q z>|pF|3lH9Qnp!v0okkzGa5DqS!CqBtn>@cVo^a%trLXew2U1Rwk)x;kpX?w1I2MP} z;pf^;?y#BxmJyUVBEB~Pi&NNhcN8YDNX*_?gP%5Z6nX|&iQ%Kq7joS1b*a!W-sF-S zN4R}m9mR7q32I`tn}k@M#%0aUbaBr$Ifi$7xR3^r{(IjV)#=ru(nZC`;+ zul4n7kMPyBbrkqbM5{X1fvF;Cy~h)YLT!qw%|gP*L>km9850}$wZXo9=5eq(igpR8 zkeP2gvU%@p&Q@WKHZURA4E5nva9p#V$nvXtCwZQvU$*!I;^OC&C4d)64ukKVyKx*$ zyh!|Z9QQyoK9XMa4{N9uQ&1M%FfL4`%U}Sz1hC1tpmqXs`Z)ET#$X|MWgw|)sA6y2 z$ZW-`ZVTv9rYjdN@z71m%(FH_gAJrk`fXQ%llrWxuIbI5jJ=&|uTWJw_hPbR6sLj9 z(V2=^ty^uU(|DAlzpOZIOb4_*23ay+ukTsa&h6i)JUp9jnw(~~SGmugKF@Pfu4uz> z9&=|elt=`wx-YdVY&6jLA=jZsEAQT);?*Jn6KKxh z!Vg72TpxD&zWJWDa|fuw`^JQ6U-2iOCLo_+{1lLVMp)W;^E@!wI<2#7Iu{UY#5b`l+in7)F+V}yeA6^8Eub1;}0ZImdG^YIz z&~45&YGe78UptQHJ)?gvEmo(z6!CU4$=ltg(HIv*r{J$vqifEX3m}PUmtPY5)f+4L zecb(jdF6)X=hDs|?6H@Q-)_6hp*DnPXFKrScoH$t*hlF#w1a(lUGC?fyuncNg)+VM zHZg?6oW923h24cP>9xMm=|Zv(plcvyTkLQ3NW(u?8M9$B`NkemCXY4&fy`` zwt-Lj<3sVg_h=rkEzsddudKkG5YBH>*6_Tn+6`~Fn3-OUb8ziKfMr7oBd6mY>`tx` zk(Ol7LlddLW8LilyNEz8zxIrLWz|HQcGPOmbIQN~miLZTdp{;W)5V}fThvF5#0e2Z zC?>A6(9D&r(dx<_pfxUVKjrW8$BXpG&+YE--Rd2*`_|^*@!AdW$(b^U z;3^1PEb5Cw>04OV9{DrY-AsVq`*=k~7|i-M%^}ju*H}zTR0@A(mGRG64)^K)D(7-H zSbMckD|BK6>t)wdD5jM4uquHh2;z41g0Xw4CRRL(d1k&h;EF|6W2c-X(J@?WjJg-p zKWBB2!gC_xBTYfQ9nIpBt{@;5NUg-xqv?S2C4;s;{o(3ZReVom!Cr%-%l z8asXV3ZQ*S_sUq(h-eIJB7+O4I{D`1f&=-&`{+^m`=PWn;VEL8a_Vo`qE+<|>I1fq zR(K#^^$r@Cb1S-#{^H!N3qBz z7|Qsw>FY+D<0(=TtV-}zy_4p`4^%a1;$@7<`zbEhRTMUD`uG^J0{|tNq?0Sz>fk|hUd_GLV| z#ONaiUQ}0@N|2Ib<$m|7(~cUaRd0B6$AqLZtlZr)gm1ZV?W2%Xcf(6zp)7v{PG-T? z?Xf>;H;35V0?lr{8HZ>CjJMB7cST@7r&<%TZ$Y|kV0K=!>jnxdur1B^7k6kXHjuN(S?9JN*b``+IVXd88-lv}<_`Jl5)6a2wh=Y0N>iL91PDglxS) zaOKx4x3^})YV%T{r7Cohga;2@XUyg>t^`}M@B&jGho7hkWJPDpc+R<~X{<*TIKvb@ zG`-T!9&fi3tj`)wRrpXTA&n#giPddHfvB9dS)+DXaZL#)7A-b$q}Nr>YVX_@5&;+(r$l52@y( zw=0t=0`#nY9L{*u6*esZYHMM}QD@ri)RZ&{K<3bP{2zW>X4{z~<60?+2};K0zIKv$ z(42fI(f;D|=qmW3edwLO$FV6BWOvB_!hJxXxZ-s90nR8mNbKp%t4Ty%a*Qk4b#Zh8 zFVcLzT0LYP9GmV<_1YW=yz66{I)4te4Jz1;ReR)PH(ZI3z=4O<|b+l4q&>l?K`Cx zA9;_ohH<$c>+7p(`!}W9LMN3Yy0;cRKOq}FUzO9v%{2%$U2*sllnd8LRQk|E3%1*Sl zK(mV9MM?PR5Th}_1E*6@$JxQ$E(XchgS@23v1lw!i0`Poyp|e+J&e1EA^7 zigz%fz9ZU?YeT;6v#=Q9eqWjU-I(|$Q?@+Zr&|WnXdHJ@T8k2cVPl`-R%#;iupS4~ z7TZ&Xj2OWqW3fOW*t$9zSwH*H*VCYnCXgMFNi9)1Hd#8({HJQm1yYbhkvg^N3$vk9 zN&cDdyFr2|Uh4=^0w%zy>U8`%J;?w7FB>HHpJ50R1bcwoH{R+}f6|1Ihm1{Ldr5`t ziCIk5MfiJ`+7??o=x=uk8bmp_SN5aEtUy5Z!pBG}(#l>>#>J{U(*|<9u0;{N@YsmI z7DZk%)Vp9hYqwbPMOKsj-S3)uYir+$*PUR5ob~{jWfbGngAWhwY);_2SkvZ!j0z_a zs@)S}VjL73r3WwoIu`Yjpka*PaV5;dZCrn+?^hmTw&d+nVc33HEVv&Jiq~e#-y&)i z7VS#Cg1uPv($#S5s@Nike-Ic;PZs5Z^v~5>yRsdhv=7~;AJAp|?Jch!GzG|dZ-+kz z#ItZo%UrJMEvVP&nFEW&T%MOT{Tw|cP-S&pS_&(TYSwb{a^EqlAOE5KeX&`WSKcQE z?-2*#iKZ?JP3S&bt2Lg&3*vO!JrglVy>|y0iO>-}5q56V7+Y3vvtu}qy9dQD7f>iI zle1qLpbz-}l#v@1&xt{p#|5#hDP z^p^Q5ce;!%_ofC#+Tv{qe1maj41offC1L38!$z-uYYV^v$3CQC7|G6*(5cD@0W%}d z%a_G9Z4+!Pvs5&?*Pltd#~ezW^#WuHAIYEw>I4nf1y@bDk-JN0elOitx~_O?n_U@o zftp$afyvWKPlT;=V zJ}aFLU`orE?0KXeOKcz$+RR+MHVXo#hG;8a{vPNLr}6~nv|@CMPd4(0FGj^q0H-n>uA_f`7juyK**qCPfK7r#B(Y z-_#ML*yVGhSJ|O_`Nc88?lH@0&>t}Jvtke6vJ79Y_1&ZddECT_@Po-fNw;oSPJ@m7 zoP@WsyS1E;fFpv2ZV+!gLx&Yn>aLGZ_jc3<%;sFL{-gZo<&Q~J(Id0 z%9b?cOywc5s2*!}#cffxWQqkALBtQEwiI>_Be!IvtaE-)kLJL0<{{rI(L|I-pEK+! zqeq~vO(7_)(#7?_S9$8xeJBE&r>6}+@L=2R5ndD7i6zVXFtg;Dg%wA%L916F+^>c9 zt@D=*%nlm}T)blg+oeZ_2UMA2z3w3bHbmC zou~tTt`AP$a=&X)HQtTd@|bAdoc4E6&7ZKT*yQ_B-m*SH3xW zYIy3dChHXax!}1scH6y*P#MukcU>CztF;aiJ`APzLEYx;tnIcIp6np#dOQY!w-IMV z#0}atD0v{^`zvGOxXY-EGXuA{wwrop%T5H}=d*(KSy^VmxSO3~%eQxlYBQT=bIxJ^r+i)A5INYo}Dh_9wi~?9RzVrke6snFnpx zs&n&V8-b(Mb*6^W)R@p{J3rS2e4!RpEYc5cMQ8O`0q&EgFBm;b>!{MqLxnd?lZQIJ z9l$$Zz~w;c_t(AF+6bu)l8FY3=<9NJP16+^k|se@L2t4)mxP4dbrP%G|IhD^<;BHW z{tyK@5ELFwEApsmiWcuTD;-6TJAA{=j#gd9n*J831p7I6qc>?k0lsvNW|vYt;=&kR zK~f^PL16lV3{pteR=#F&r=58af$_A*k)qo(Kffg*5LEa&r_V;TjM>r|EJZ|Ep-Lv( zOT}-Niv4(}imm;;Y<<(+Tvw_p3U0u&1-uFEn^z+E+L+|)mS$f%C<_PvRb14%iSsPObnzk z(o}A)%avY-x3zqIsxS?+3~|=MGY#JClcv!=%4sOmB>+sTDYJ!jOPfm}ayGQ5FXXmi zOXZNT;YM4zXQcD`<5KUWH7&QkN{pcACw+~&ou`i>eEBvsQs^|yHK;tH<-3QzzqmqP zcCEICai@M>LuC&r9Sgex?k37(I31fCSiZJc{gPCJ^EAZlGmLXKiY_u)%&Q+pD_WJP zeBnDMjuCZ(jHnwdD-p+c(0DUf>uS`_Zm4QL&#F6^`EAs%=+xA;isd_RPc644PW2~X zZDJ+bq6D!P`(f0$#JnV28GW_v<%^+Rj&0H&n<=T#8*&{JdYt-}wFu{^wmu4r8%tts zXpyXeZlIm>*bx0e1VxG`M$^9w>$D)?N#1hnJx9}Cs>`7IE`{VhyS!s|MC`|Fo?}Q= z!ZuCtenxEI)Z)ruo5ncS8H!5l_qw3c9joB37Wy=$P@xE2Zd=3-P8Cv4mT*|`?RP3A zFm~{i^;9iD$*u5L8E6eAXG9zHC@)4kEGTmHrNMxOT3+`s!y{j=T%y1Y5hkj0m?SgiItf$Zq@84o~6Ec*QOXeqg8hK(< zd*v^8r0FY-%}l_I)CJ@m`E>?zvmM1a!i5RxyiMe*FLmG)VjE}B;KKgQXg3lP)HJ4S ze&a0b&pQ2(WVs;=pQhMc@$1*4-~R95e1vB8R?B64ansns{qn@ifkMe;z5M%V$$zlw zYYRg5f_jc=b)Lz2syfu`to+W&e1Dfr_7k>gs$%4ud_r*&w}e|2nLB)#v%?}Th{(k9 z!&Q6P>9j+vv(x*(#x|BRbav9`&vEr>_GO0VUY%6SEIL;|I&bA1^S=`keA`g|6t9v z{Dbw}=-2&O?M~JaQ4)r`G4TF%wO{o6vv=Z}Hvi}K)nt2+X=6`#n%;P= zvY6-T?lvyW=)J6vNqb=cp(JUMI~r^b;$6Hrq)bQ$7Hj<+%;}U z%qjXvB4f70t%m>Q%5)O!kZP-#tI;&SVVkYgE+6vsbEmk9B9V>lbYTb1@&c!@#~HXK z+-c%;EBlC8LC5$kX(C*G#)%ZYMTc$Ho%P1qg*4^CfuS1Frm;lXua*f@=U}7oh^)6ftd1@%iXnD@GBl2qyq+@>k$52`d?t9s*c3A{rC1TvYykr5lY?*O6 zKN{L$o+m^>9KG*C_eW~l;>_^W@N=VpSLIvZPxGjD#S2Di%h_1f*<7W6BI0`9-GfbIq{QmtH)*&Hy`M*;B>E>)(ziH-N_w?* zm2B&EP{D~D{6oHiJM|RP7AB*k(i9AfAx-GqUyO3-!b&4YeJX5K>MtP*@BwSC5^Nd~ z#f2i^nVITJ-a`qNd3?uDQ^=+&!RK5yj{#?)b=~56O_C)S)%80@YHFNm>~bG_x_H$? zPQfu(5;J!0yX1tjUe5-&u}&rw-gngpc`I-;Pd&1%5S39eqY(|QLtFV&VK|BZn7UfR zq)XYOlb0S5a^^!L^y-9O2ANC>QQ}d~sq1HY^^Cgg9Iwn;EdlyuP2Wt*_m<2kFRrM) z^V?^xAEX%xCp>;^G9hZXR{Y{fmT{eP?Ro8&1_^@`v$(Mdh$A_azb-rCKzveN&{e*T zSvo2136(cJrKG;!0g6Ju&J+iSp0@O0RDNocQgzGxvfcRA^s0Z`;C)qkK(4`^zI_UL zz>t8J;eJw*mevTm^sOD^*>VJ)$lgT_LD&;VPYV0!M`L8J<1xP_$OYB1ETbxPZ?ksx zF1Bg~ta9cYGfPy`N1rP68uue<;+HJlUXBysM97i#w(tqB(*EARohz)v95bP zIt!MA_NB!UYxzg08>YZVMqv@}ed?A?WS26ARM3tWZ+z`$j8bLuT_TCXmf#>0JgEoN z=>Q3|HDWe&05NB1K*786h$t~-p_hoftUjs2QS^h%sP2TVoo~gEyyDC>Gw2;d*Cdwb z!N&$?AWRJ>^m1DPCXOeP^8|VbbIV4X|ju#xHHQt2otL6?K*mWOQ+H+co#u(6>SGfh+AP zx3+9KEA_uIIq>3v%iPt_ls#>mYTGw@+w6c`BO`YJAY-TquJIHSQOqK2?cR`7cBTlv zLVE<7HZXOZv)oKZXMILg&2tjhP1JomE#lwb9XhKFQ=4C=K4Y_pEQ4qd8Q%s|P@4U| zk{KNmj6>9|wQt+nF+5bXw^{Qwm6_c)nbhbsAb3P$+z25KHn>Un!1j zY(-p+gy<&=uXyi_bVd@Y%UiutBhhi3FD$Bxx|(i-;#G{?Rz z#F-6H!`N_8a6_s5g4;5UH(GN68D1cU(}(&8OOn;(zI)wB6<(kDQ^2=nwMF$g>Sg6r z2#D_+2X=Z8&w&}qsI0&$X(bkyw_i#FLT~nbcj?9cQ1YB!lgYw+>+rgH!z)>YVJ-BR z-)Tt92c>{dwA6?d{36%n6>XOmQ_Pmn+Xs-nO83c_HmT4<4(7ypUemDPm`7Bh#96ax$vA3XA!I)J>3bXYudTrOT0;9 z*Jcr<2xYeLE*KKz>lJI-5uig}@fuO>@!)T7F+^DNdwy{YAh&Gs@hHfQn<>+nxAZ;2 zDD|y62b+Yhw$boxQH05TaQgEA1o3vgxa1u0$!*lt8WmESGtcJm=A-<5ytLZPqIWZ? zyH&mA^uW~^tSm}~vs$sf9YOs4&)a^InIfRh&3=|jKTiNjao z%As={GXIeVSK7?GkadH)?H{Z%$vv{oz=Q*E{c7|IWncJ3MPM1?&lWPgZ*-unAOzfb zqDpDQ;50W0xi=m~GCe+wEq)_)4PleILEp4;SUd$Ljz*$!rg_fFplWS@Z;i)xyPFXlLIf zcJ*57cLOK31)Kp|{HeWE8Vk#K^5EnHeImOEmuHm{YH8K^#}4I9DQt6PjTgv4xF3D{ zaFYKKPr&nP!brlDq%sBzN@K2GhtN*F-9Rdzn~CadZUJ^7#;yAaF@5yv4N%PM%9o_; z+K63_Ct^~0S?Uoy2m^!{diIuWZK<}q3ax@YJ8_XlcP}C7mMvrCsPoY+!S(k-n`#BH zk-vg^BdNjLr*HMzp?m}BvAdR-+y1A@t#*8=^0%;>%dyZb6xlj6BvTYzrKPPt5E}jf zwOBt^-o8(Hl;wC=T6*ZW?DKIYWK>KabtYc9dwgi`KmV=ygXJ#>#pd42MWl`bz4y<> z=v#8CD=T&MxOV&_o^SnLmhW3P5iI4qYo&#tLmtDIkI_JHsQGcDC;y1s2ED>{3bFKs z-qnpQvOLXW`xT8QH<5Ns1t_B;h_PVNilf zU7C%}2%mx5bY6X_-*#^y;oB4+oME4yeqwwT6ra4`=uwVORMmj@ZjwXRg z-L?)**bcOwHZz)_q?*`PLMJ*8H~6SmRSvM+2ueI3^AYgVZhq=rwI7QWw#A-tH)iBt zJ!-T`ao-_EE|hI%xfBe@_RHOu zNNJDs`y7${>{^yFiUTV_O-R!1c$bLil%YE(DtU2{@WL#51w%@{NQDbyzz>){Iv`u$ zHaT}^vWycGU-P|eN#Z}Tz7VHwd%KxBAuWJNFcA{#aE>E?k1`5T*m?72&Y_Ot9nNjz z&sBQ2)pT;Ny%{BJgiqF|`I=#y$>Cll*Hi7_D6gaw(9OY(xRzQ9?uv(somb=*QDN*B z@cnNggpl~d$y4^(zTD3kr#a`n)5G4Qpu`Hcc^~tQikdQ(h#!KOKwnVC6a1Kn%rqG zEtWew+6(K-Pc#UpCIl=udlueX`iL-Xdgt%chK4eJbjfH<5CTG*+91i=y6rD&TzXNl z1~^D5iRd@ts@kxdNu4ENyS;4>Mn8LcySaCza}B*v#1yK5-uhIB?%*BjJ?ewwC;OGY z{i$67AE4!UO*zwrq~=Biuk0CTfEW{XXJ$7()|LK3#+hua@vGF*+~zu~P_&(M(!~zk z{oF{f?cWPrnLJjYu%OR0!P%U901AfqAN$QIZy8D944ApVjw5$t2+^rH^QY3*i?i(w zQPI_VMrVOvmYww+<@jglwRCih%ZCyqv+bh{8AwFA*z8ic&7i+z{w%R8`hLA<9}#?m zrVAOn%}cPNxIJv#C`Mj79nyGi*z6pbiUY3M#ZcQ0vwW%Ax?LM>FMVmarl4e1 z$aBy|KNq4rrzcrGTD^P}lV2`=eAe(X=XsBER|4AN*5+Fmcrx``x3yn%>UWA4VkpV9vkeIWqym?$s(;+)PUYd=N=OHidv`NL+ z{c8#sB`b9tf(u`fd?%L2`Oq88{b-_Kp`jQ(86B; z?NKd5>!SSR%&Pn-nwq=tH$zP-9lnt}mxj*n=Mkc8OvQ=J(S5&h(WTciz0o|HWP{{d z!xXIvABVYTucF7B;o+*%ZP$9n?93&pW8I(P>M9eHCBZ*f)kl4tyksEeh5=Oo6kE-Cx==?sh_Kv)^+O5*#!(_+>_pKsc4d zjo~n(EJfY z=*hh&WF@+9e~>mx)mrFlU%g*0+RAaBWfWm8G`vJc3{3HUeoannZ%ggx`FY9TzZ>V_^(C#?o#r zVGIu;J)aA3Hd?aW3BsA8T<563bXKjzI&k9ij>EP4!pq~&)$w%qnW_7wceQ2<8A(hp z?+P{zuP7NPd)a>5Xsdyu=SY}=O+YxCD=TA&h{Kr$Ow(|7UNS4`E|cH%#cu^@X)Q^% zzrLV?sol?TgLv8O=H$d_&$8M>=KKFp&3f(Ut%%FARN7ioTL~~hbCZU6V4lF#UG;DJ ziBEkpL?_IQwnf#z7cT`R>vH@#dX`LCF5abl*HvWG;qC@X8#_%x zylm?z`cu?yH7kQxv0;$?nRgmU)rFGKjI!>?-$Oje?rA~B+u*6f3jC=DR2FMm&*;be7Jp%Q$$ z@OgSQ(J?kW?wuvBbte%nsGEpf-ydXSi>Y!}S9Jv~^xl7VbGu0aELBG`(S`}ftsq1b zk4!%w5F$zvA`e6KHMo_&Ren98r^lFe`Of|xk_7_m6*Xvi!tAbh?Gs?^M``K02xK75 zvRgT@O+GZ9de?pY_>yz(IMy*cWy%nL%{Rcg@H9L$(`ge!D&H~BMja48lCVL3A&Y?~ zgj`shgT2eHGn{3S@1}A{CS>}9L^I`{=Smayn(1o*RGOUG;sizR4GhQ+Rl< z97^CA5hYUZr1@E6~zRHx3E z5@=8LX*hja7Zny;wEjc(rTR49lVTkF0X>zr=lrm1%A@n*Q@&g>?2%;>A#}5TLrRVJ zQKD0=3df5&Yju^)izhQi)YLg^baUep+&U4_!2I+bQ;bm;q>YLS#<=-?KYDuhhbmfh zeJ+>Z-K2lrrQP#zs|8Zn*#BVcgq`>sBE^Nz{a@MCQ6E zAmrs*s4y16lDjGyA-gf9zpfaNsPcqt$jPxT^2=vDVOUaQ_C`Qtfel*$VU&XWv}@2~ z8{tc6<2}Vn!ldGGk^|NCg7Zs+Q7`NBwDe&<*tCg)c2i8_SfVPv(RSMJ*1;WxN>6co z_0Ngl&!_F}<_J54NuMzhSadtoXV$z3{G$8mOQW`r&Rmv|3CFd)@8(*f17UnzbMcps zQBX{HL$;C82rdI_H#}Hd;maH-se_hgLlxwt#rWUWV{taqB^NNdZJbJ9qRbPGL~dy z+>xK0%2wFbudJv}LuASxzwZGD&Z1TGc!BLGVLf5cMJ#j5ugHI}s&-aB9l%GqAN;=O z=E9iG^!(tUg?A>!TC&hPWMz;D6y)3CO@yie%gD{FC|kO>xm`NQ98U?+z@l zHt~~Xz1L0A3CdC%i{HB*|0(YJ4BR+j^c}ffc(#rag08=Iyvf?#>VrbotJF!i+fEJI zK-dkr*b-|^n-?fBA!pvVj4DO^Okr$oN?XRwpw>g<{n&f_UF3KDEQG5lJ#lIATITuo zPg8>3{JbAEimk@mvBadS^?Umz>Ak+Vf(Dl6j2XF%4vXHLNx#zQZHr9@Ria<~A|S3J ze_M}a5>I!7CH-NnJe;;i`NR2o$~!o;y0&s$3(eQh>hN=@ii!cjhoiECkI#bcA|VD3 zNZv^g={C>G(%S&m+kKl{^y!SxL>H&MvIQD#RVC?vVbI6rNy~~5{CS0736)qqU7^|_ zQ3&5a+%k0(G`FK0|MWcdS;%sC0Vi0BR-d&5z$Zl;KQAI;mI;{W+&Av>gONsdqBA8BQ&~g)X9~|wzhh@xU|Dc4b7L8;HG3^+$QGx?Oj5QQ`YRDsnrvq< zg8fbMoA*6A2piY);uKauQ5DwA^L)+{qR>Qp$m6{jKliv3ToU{s$=N#5>j=C>n2!92`M&O_Aw=!69x0Zruk3|>%;9{JQ9VF5jPFs~&08bhP zcT&SYIi&u;G7GE;Bc}a5AKlOofXC`{WSmnW1sw7Q?deK5(7PT*I zTpqoK^Cr8{owxpkKRB_fYQ=t&2wocURM?h)g=f08l|o)Li?>+2i?(=gkgJUdSR!T| zBHclqw@3L+v1_$q3Vv;RlNm_p=4EF!>|+o+vW(T>-!cm;<>2Sm235s>S@poimubCf z%S+txF24^7mLZbAl)%pt=<_c%@fsqWpC+sG$1k>Hsu@*c0>hfp_lDRTADf82ZX?Z0 z78L$w;JP&8$*zV+bf!U#K|7jWPZE9}i=~wa{^{QrkT4$wGgwQo;_tRSAJrN2X9hlg zgh|efm!G%rE?=BLaR=^ z5ieO|WBK;-LJfWcv-Ue;t)(De~ro4g-jEl;!lnhSLe_;^&TG(#K_ z?zAa=7!!URweyRt-K51n4%ph}7G!s+R?U7Jo)^JVUH3C7n!@G*^5cv1EQQSc@&pKD zY}H7T^wb{FgIkAr$-%efUzUH9uB|u6iZ}5pWIX{mzjzpdP2iXlb@Gh1eOHm) z-0NXgi4{2Nq8}pAvfA*BDo;^uQ(%9Tpaqj0=#tbLpMPL0ypSH9JemKSff5<39nvIG z$lg$bBnO}TGM3i%$^V)&hrJ>|@aSc9n>bOQ)Ut{D?gaVe$NZ4Yj1*2iB&iD(H@VbO z7tZ~UOP_av<)C{Oi$j!VTt9vK_jPemoLHqK_!`2k2SwpD80YqtVy@1`f=!G+3m$Nh;CND97#-^^fdO9R+{lwvLAFl5xa3g7B?E8q)t)>0S^!fvw z4Wpux)e5itA_t0(-7ed(%_>+}jIHQ&qz4npx1L?|0ABeA#_UvYmV!`PvIZN#mO11b zqc!Z0?bJGEid~9mk6JCHHV6Om5Z5Q8+xkWj9`OT2m;PF#rJ-xd_a8IK3Zui)3aa8I zteE4xaf3Q5AW-YX>ce5N=aAoh?bg1yf{p-1LU!BQxtF&E(XIabRWHjm?~qcuL{3pq z)+a$a+*&Q7RL((m3qkJhAri#YwAO=Df zEOD`bkcnkS$(-{9HAdfXk6`%Fe1eZqb|VFg-&)2fQJY}Y9@0ohT75=)_&2b%#lu># ztf|j&53TWvIq!lx>}y%wq^?vO0g)^6yP>i@j%9K(N^yz$TN|q%Tdl>G^Eel*hq-ry zjVXQ}+^L0{g?hY)x&i=0-hCoOA^AmLg||!77f}fmxH}dwTObX=bBRnPL8SMFhl47d za4RMmQd(&q2S?))tbKapK7Nmonz7e4?m61+rqA~=`a0*F3pllV>UG?PE=bb(VYlAl-G8bb7#G3QLcRH{GPny-(FMuC%v zxXHDdMN30Tc_&IoCiBxR3934_hNZ{k=s8(5_RYjJRI)=<|9p3y(qx>;` zShdrDx=>>|)Ys@F8$7ZvbF{S1s5vNb_{uj)=$Y&{k#DhrlYgt9(=^LwSImD$$X|0P zO+?E;MBPA*rJk?^e=8zkKM;#iZ7)>mOMq4g<;07(Vji#!;a{ zd=@Rs3LwkS^uqcbJM6pJ;sC+w1{n3AO1*q+x%K{l?tdgo`;b-x!3nx=F#U{02~FN&8nB*aWfC8 z(Y%wB9=})OSW-5+_GR@L#4a@a2dGEiDaOgiN3VuJl4(Pdu4M5r#mxE^!{ED9eGa8K zH2}j=C|xYp=-5|PSl~g?@(&8(euZnX-~+^u(c0RfWWk#1TISq#$~%=2cQN_J z`QkdwzA}!p7ceTXpVW|T@Bm#qEh^3xCet|M=2i{%2Y4f;bt|Ui<+oMnL5;P;3zC-+F}$dvv3pxm9zXj~cPPQ++@2&c49Dh;Gud_KK#lyvQYlk&VJvo{KXMoc+H7BAt9Q5Cfy2&=;%c`K(E($fGzl#TO_Q4*rI8W+IkAH)`s+E2~2kK ze(%1g$H{cqd)Ge2alM|rLEDv_HeAz5%@mbsA~3)TPB*s@*6tW!CW3$yCMqWp3-~>G z5W$mhZW1H9+0gimZ{pTZkw}tTJxyf&{?}jaOrko)1_~|Fct9M7@(yI}Q zhsMDa;!;c-C9q}k7iIdKJkzzN$xl;^H@9Brr|bO2OFlks3_?vlTG=;IVeK#EtBgu= zJ*`&_=<^160@OQAB@K~N9!UfOpNBT%wR)MNo`hm!n@HhNqGv0hvnC^xF>Gdb(7Nr2 zavqoX|BJKpjA|-+yF5r!X`&(^0aOswNR!?zbOb~M1OkE}olpXS&_aaIr zUd~x3Krc>)cUF#_(V&v$(5ld)e0R>|_ur4=C;10Dk$+PIJM5pRxy86c>t}a-Q>q`YZ^#M0XOs=+p(}B+ zIG_Z0g|kX#wHAS@U*?s)x~-<+p8Qkx6CIuK?I1!aIiU=mbMe_T@U3UyoG+xRWA+{~ z8@{(Q{8o~4y2bza&WJjnr6~L_-&sv_rK*S}klcy|A#H{{t~AfHu}TF#&7tK!hzRc? z3zyQ`ChU*`EY+`#^!y6w$~}ALHOgvTmZv>^KE>E^J__Hcw6#$fTERM3HsxWCliuA! zTJ;R{hTU3v{(6e96xvTFz1&U?!u0<i#2QW&qg@gBZLo(tC|!Wm57w1k!W275I(=7V>cdr$mg(y+8{;kyQ+N^NhC9 zn&!0wuI{Qx8|RlYGW|U*GP<(J1|eBYk~77NxR9i|Sb=RCuWyUL`B=bqk{5mwn^_GE6OvP)e*G*pRW~RvcaWX3bpC{}ok_|$ zur)uSxejscdL3!#fq}5COcVT4@7+nH&}@V4F>+GkgzL*Ai(+Y;8bVtUh5&}yo%~}v zK}Tp;$h+xSC^gLGrZZsnblv=*bsgRGJ@AhXd?*iI=CY-kM`^0_ETI0zU-Lf~!vE`n z4B>)WF1H5gg3RUmZq6l_SNqZ3-3xm46SeDK7;nU%@R3&F>A6S{5At5oM(ZxP*+Ii^Hm-iS_mfNog~Ou*Km2Q{B9aej!1)id z0)-2L;_E@meEKEKX9dMQo=3am(1{VbDa{CX;YW*=4ReoxDiz)N>}@Phw&nXo_xHb< z$Dw$I*&vt(h9_TU6VaOTc;(qZ!3gMn&&*lwIwG)$H`NDQe-`WQ1=Ltvu$-BFJTh-M z+@!$2bF>GvU9Pbo!qf)OFX(VG`0o0iT@gyze|4-6ZWX8gUT7<(A(-6~@UO)muHyHs z4o1|6Lo05{=pA;5-NL#`v!wkavM7Fk70=hRtR8C8zJm%ZxC91L#HG)!zNsvr`hJKvSAe1OdT@@Elu;vGN=p=jT&7u(VhAHLnDPZ@N$a(b%K`~+voK|GG z3#cb4VYQx`Q%%3`GbWJ$3)!A0=o}qqnR+wCz7San^pcT^H6q zj9Fnxza-hZobVHxxcUpa9xy!4rhatMM`gyI(wKU@m;(7n^vXqQaNIzO!sCtknCEbv z{ljYfg^dKd>!xF4T_XFw9yXT757+~^MVuHAU&Atp&?(5=TH&L@Q?bbFL=*ZN_-6v3 zmfv3X*Qo8OPZMByt{by&Gt#=yuDeZ%CtN)t6#l&Hk^MJ4`*TfhcVGLZnme5qXRZGu zx<>Kx!2627?A?3t9eM|fp1PR$RJv`zh=amhcCa0eyU~a3ZZKRsL0wsxtth%Y@@ml_1SvN>pRR2BD{XzR2HtO zpY^mV?EfR0N6o`~UHJ}$V?xn2eR5Rc4bGVrL2X;1#bfekr zQ=7#7wo~CE3+Bqgg~Pb6qozUhsXNST!vEE6+M?Bk^^=y9qT*I!wye)71B#+8z~9|h zJSFFY*PD%3NLhK4fyZ!#Q{J47q7s;^*@2U)fr_S->dTxQ_kTp59esMuP!JmDOckk4 zrA~|>#4s-zvU5`h6-;Wq83_ZG4(FElHj`jUv8&HSPIyK3c2r`a;%*6l0m9?#&O4m- zQskG#O7;~Ly03`H+U4zuHX3lo)aN4t6PUfVM)%f%j##TtB#!a|tJRxj>r#?4%-Ju1 z`Z|BJ_@oMK#ai6?ZsRr@$_u=&gzDm0YWYh~d0l4#3-LRC?Ze&PL{VgPTFAc-9gFmk9aD5rbGtT+66I#)6ee&>$F010u)l4X*P+aacEJF$vK#})SvC; zMokl^@4t!k_;EMKE61P|v3h%K4OY@4I~5r?MxBsVBz#z2huAj+XA<4zE_;h!CyZWc zrP^{Sp%l_2uFD$!+Aw;GeADt|WkvxELHIF(4oPkM;e*FxwRv2lW#ZlHeR3`|Y5LQT z%Ssm1GtxlH7UKuD{EaPtu(_f)xarqBmW3gV`z(@tC>^fnr-H}bN0PdoueF^AoxH zGhOqVdG_fO^BRJae5G>(D3*qMLg!g3=YDellEZx0JotF7o)wEiCGM0Dx$?l;TnH0` zN)%6>3>j;LuN?uCeGhDr-_L73cfTN21|IK-vE<@o8d7@9&DlK%Ccm7yJS?oKEbHk6 zga+99vp^Nc4(_SI-DMn_g45bS%+WAr;DG$jByUdd%;N)N+?v}f8RZgnG8WlGEh%rI zs6H09Z;$>u&jqimw7wG0_s$A0&5z6qC9dr{c*_-{h>=!&xgO6>+WT>2R)fP(beLE3cs*Pa7dWCYu1uBEqRdVhfs5(5hF4zFBt9&ZIPG%GSMB<6YN(&HpMFpxEps6Yu zvdUsv`}bA)T6%6%NV6y(dfvopqtf4gC5@qolUQ=8gCTjT?B>F>TJ5KUU_ZCmNq~dZ z%ezViG;gpYYJB2b^=|H91~3Q-abBt?ej24!8_&)u&{G%FI;W|tq&%aa@>#df&?=cN ztbyV6QZ%gq>}3)swjNj_eLU(6T**oM^8li;v*mV&XJB4}mrG=RJ8-(wBclNGyE^+$ z%015?56KE?rfLokK-!|XA4?sBCVDJAbqhVrl_IcttUyaf&dM8826r9S5b_gwPT1-` zGq0D#$%=gwq-ssO`FluZjg|=G4x01UydXHWZmc&qV{3hibo{*jr6TcUrwdSJ=?cP_ z4teqN+i2tKcp0kKH*pC&(OL>D2JTtsD7$osI}!uL4>qnhS^Hb*Db?aVegsAd3qX=n zFrHKsRcVD$IT82`S0J@|-&AI|ySh=|fvf9hC#wHXf;s;#VGKS9E5s z&17?|TJDSL&LOmMDdED0egF=s5BuQC`)+oPGjR?%_E0f z-=$h>cFNyWH9!s_7asmo%we-RnOBwldaEfWH_m`vYhvhIYnRQ{`p)QplritSb=6&~ zF1pi8nb>Rw(oyro1I6u37qI~~3ay2TXe%Z=NP?2`<7ZsS-Em=c-nAD7^e;96vIP$YJ)twgNkBB#8sbBnFM+gW_ZR?C)~nOjYW$CS{8 zrS$AwxOo(=LBtp`)vd+~l~Hy;x~;p#9E=Wsf@Ul8{b-ZBv&yqx3heGb72m%ufcFSb zXtCTpJA7c8U#%Hi0dQO-f~?e(23YJyLBmZx81&eSOgbCb;U50*JTX61Tut(nVNZ!L zn5B%U@Ppewa}kvwWer&fKbUtvmEiQhCw?2YnO8>MDn_DE&kBSMRRIf*-F}KyEZ^CV zu8fL3sBw~r$y(#=pOQVZc!}Vjk2w=Nq2l(FS3cCqGx?#~GQw;#8Pr_S z!pth{4>lV#Gu&~+`4#WsymVN2^1!!b&E#Kn!CuDGmhhlI_(*mzPS zkB;{?kym6I3|@cCRpRvVzV<*%wKD?Ov-;p45tTQcOdc`%eO;^{LP1!7;T#{c_3sI%~|xUN|ACKQ&cm2LwZ{EwDk&4MOcd&)^S{2&oVs1D~T`b!-bm#weV#=HPl8DM$3`3W9uIum3lXWBgGIzz*`! z`#!U`h~BCnH6i>t$;NidQ=VSyMbcLrng*T<~66>U@cq?4#o zF0*T~OYSzXAAM${wXe=m5%8#S;iG^z{UDP?PVnxG;%a~!l`6Xgp2E387u53T9s55b zFWj&5w@P%V$f*Y}{{;OKYv6;fHOz-g9avwAAM~44(?dP^Q`yT|2vPsc$ErMNX6FOF z=eOvc!(%m-28jMd+{IbyH;_f zJMX$?dF3qQ_-1w%cWqI|_zqYX*IxFCed~M77mPV>KcTJaKP|!#ed^mdX z+*M#e`3M_sTN99%9ZQ*_!nz2{%d&4PN1RUTlJVmk4 zyR#?tjjmFJs}EIiJ8n3%dJNu+L5i1i)1WZ~VxH?^>W;u=qnJqn(8qA{qn+i#{ZMRk z0V>vxTjkx1^RGIgq5@Hl``6W7tuunICc3@e!iG(n&eb$LTNO^2`x_Dt2$9%Bw|-{Y zM*V$gfR!iAF=mHx`|JGMMKy=zM~B(7^V1I|cvxk2^ojvtV$0NHP>SkpMt3#OAoNEb z>1YYqYo}H8dKj%t(|&m2iT3$h8P{@&@LCHa_)l?K1G}nIlJ|45ute;^-1b+11V)z> z?rSjQu0s>#qj*KZ_9r}d(A%JA`t8zcUNgxR0YuMcwNz@ScoVqBe*MmF1pcb_;CeV} zq`fY8(e>OIT(y65-{{GeREWw*VC>%+Z^&Y`wF6B}1)jsJ1CKv6DAB_1NO66HJh*w` zxJoY}{W^SkX0gtN?U#_=HxA;%4{-toB#Ez$u9@A6&DgrarEY$JAH7w5^iJy5+U$$& z8b3duhpBA+z?82co|L3CejQHWM*t8Btn|py_WE-J6U@zFZ|Sp_z>|6yp$*g`Z=8FR zm}wp7uQW-sA;ThaDW^Spb~L1-piXvFd!;^w!yZ;`tDddVR=Ao=6cA`2VsHqzz$TRe zAu01CpX{Tac_5q)V=ju?0Y6@k9&8#I0mIWVON+>T2RAY4305hVF$CmTn*hoPC8BbY z?#SyQNmLS(#OF;Pm9`VP({2!U5$7iU4SOrggp$0{n+I%xl@^j8apI#IZd2#&TRW+g zf9zK(t1m!jV+z=9(B-c{@j;DKzou3N8~0m-8Qa~z#ZZ$SO-UUkCM*<)UYDh(krmso zJvk))5i!bGjoc%_5eR8zRAzVp6~=5=Q>3|n?j6$QGIRfJx+I!x9)*re_`jA_-A*J;EE?&Jc( zI+>DrQUlK!PEC?_Z5~yhIG}sgI4ubFLrkeu#W5dY^k?3C=&G6qH>e)mJWZ`+3eJ+O zCGy@bdDNo8LHF+CW(7^|s@G$7J2A_zy}hn~_opiQ#_+MQ=efc`vN_BoRZ^G6u+-eu z>~HG?qY{Yz^%R=9wVuxXy5c=$*StXhq1?HN44<$y;Rj5$)jD$+@MQW&RO)`VQxYz1 z__bYUL0}FZ4TNJBE+0x;5@6Px+JyYW?quDuytZO7b>6B>cctGyBI1QoY3o{=s*9#A zmdgiul!7{DGY)KrcL^ey?|($jV*;`Laq(8l<--kM4E^ALE1RE=9xq%xSNJc|y9n1x8J zUz*ye=Y%yWwo(*%jIVPdeacYU6m|En1LI27U(aOckkLz!C_L$v7kRbwVsf}=Jlt_= zH7z?c7h+}kJb)FQQbc$ERHBRax#frF62*Xu3IR2}p^uuzh0q%^$Y6B#rtGe8S!URQ zQ~|Xl`j4*ga8I{M*y1masX}UrzeR6m{}J6@TX7Ps@CdaJ8)o&=#Kv%|XVP5~JLdFRUy{1jN=Rb{>I?W&{OeAPU;!@dDD%8nxPDUps=C-MO#XloI-@G-|u#Z8s?CSGmaQ*Cob5QBl_8)Zo8cKg@w$_>{T3 zBd&fTGQc~t_e#NY4sBUE&z_YIQQ8nB!etije?%_uD>*A;+r)*d4_Qq+yZ?xS{t`0E z=@0e(|Df7(#B}s5`72T5GG7g0@gw?Hzbhh>*@VWOrSq3#5*Hj&qXLz#2345*KC9Dh z*FO>NUi;?XbE7cStx;HfPnIOpGw;(7h}`S2tj$3^SbU%i?>W*WGjPdlg(NW|PHRju z@D>O?Gib_8R@5X_X%X2ro=%71vgO${%_R`QD30;D3ZK z58ht4t;-v<8~G_#{W66leu{;|rkS)sbho!hN$kJ?;xBGHzVlGm+Nn~is(55Q>Gk0k z!LIR-a7Nb0OTNM+62KH&X!Vq!HkiK_#otYC#+?0MTB^U zB+_rSWES>&@(iE70rblZ3H*6pv<$`7Kq>9T)RRBNF2Z8A~B zl>RRw`(w=*OvqBino_njmWyI*t*Ny#w;6yz!mck&ooo<%Q(wFGCTQ#cQNuARILfxo z6lWQu_Tq8w3-GQHL{{jH+E>~SPo>(niTr;qxschE zwHZ7ry;oKjn`dEWI@%c7&MrEsFz+CxP~t!PBSZr;)^A^-YlFy)4D)=-AI-*ZA?D$8 z)lfUAI_}dJ(s-${(XZf`9fjdu$9SfUOtCX_14H;aQLH)?MQ~!uk7nEj98%l8dxYvO_iWG3=QchZI|yl>T4WUtG0Z(&usd;|H9ZV%esuPhR?u z*U+LxE`@J8zW+*#IA2~mhyEin9h}Tz=#60hB67*;)e$N?#-x$U<8AwSVN3V&p?q3B zP0Yue7G&vn4Iy$PEB!XNFX(GOnWMj*c_@-feVLN*e?-cy&rzF-YD97a%22@@Z~|A^N0dS2T5S%RA| z>{+d^{}D|Q=Kn{^FHE$*6$?HNKrDUf{&SwL6C1?X-vSmP;sI9;0B(Sv;j^P&RXHF1 zG^JmyjyO2xNPngsFO2D^w*JZ0jXlm2-wYoXJkNXcGGjhf@|=vryKtt;>CZnRU8{?U zv7cCp^TL$*vWa6aG6{n*gwH$SUt*Dg5)fK8s|%WW9S2mDz&NIh&Icp%gX0S3^u?G%U(B*l z<(S$&X|_`1VR54cCxe}B`+#G0UA^<=^STc`Th3>%<5&Qt-1L&lYgiXHvW+DD& zJ5r+@4J^>85~~%EhFu(7-aiRC`PPjP?-s03Er=^~Ai6ce;y+o*qIxl-q;SZMl9-K> z{3T(CXZK)WIu{U5<>q2c&8aq{;74<97{necK?*Xnz%lh@(_asMKxPeD`GCxJ18Mju;*iIp9l1cppkrg8ccHq{*b<@@XR-%!)tx!k7+y--BpEK81r3D9R z=59VMO!~5Ums>7#hfj0z`c5A;=Lb&6mUc2*@ZDe5SmCw2uieZ6-M49&UFoYG!~;^} zWG7ImUDF>AR^<=E5e&W9G6*_f>_}&@2flRbOR*oAa}<2U+a1zm_}1Eo;1kMd7`kT@ zw$1Z|iO{d`Y2Xl&^J%ZW6r@X%p?+Nb%(>04cUT=Vr9HkhhYv4kz#oh;`Q$L4^zRdB zZ{f$ABmuJoAAj4>{@lvzY3+UjAsLLlVY!;Q)G>tBEI-O|oESOD?<+e2@V3K0Rt29KVfqR z4${|Ff8+Y408pSxkV+4+?B7e^*L8Rfah2kg+UZVuUn$*-0qtU;8v+48JYa=%!T8 zjLi?DUz7Q+Gm@sDeUDG6Okic2U_K&(9qhP&{kK2hn3;rzC9E{%#4uyY`9^*+h)$uO z3dYPfQoXrft!F-}^SQ|QJZb|ZFGZtXxOK{V+xqkPrb@I!soj=gOO;hpOiN-wu$!CL zfVjEiqbQo~g&){!Ba+I(AvSWjy#YKB>wv94Hy?B9uHRFv#=9y!HO#4y$108h+`@}* za4TMPAgM=nFPRv5nOGJZ85}WMRhRb?gBe+#h9pCL|KqP4gFREucrwTia!D=4KO027 zNWmhDJVSaPp#>8`&V=Vozi#*2dBB~(?@n}`MBlw-Kk=Z6|IpAp6K*yw;*)BpPcn^) zqbd(!GW@tZc7!LJ={F0laCLqT)&IcC^C?b7%B#A3wnl-Rrz1l6xBNh#U}V|a3AHZ7 zt2j7FbCi*a{O!A^vvV?^&jIn@T*{|-hYlGY1I9PU&sj%hlC_Y8%uD-|Tk0aUpnacJ zRTnnCu4nHRgs+u5EyPQyhjU^wJFhC}syjRpx<`_(C;gjYuSeHZB_}2QSawfe!x@vp zh(?y-0c>om$nR)fXL>p5)ZlE!jBOQw#8hY7E41E%RWX1+Gd7Pg{#0#wryosvfD}5E z5@9*bayu^`I;iBgj9i*p-JYwG%CEK8NbYw5f`m_b9!@y6p6k8YRf5t8ROy&3UiX%y zinbA(l+eg;&TZgo3?ye|Aj86|GHVovNra_mw&D1oq#1K;$TgC5)3Zr~5xHxPLq#zf z@%4(DT=*(%g9331{J9u`o)dC0jLFj+Y0AMX8D@kI7?r)FAW{;W^k=1|_o^{}begP4 zP!Ly(=1njJRrE|2BkXvX`Xjdte+^C4CYat2!s0`Tge9BOtjDo-KaY-wg(wZq$a3%O z5|hI9MR#ex@5g3C;J0_)9&eM&C?9i(BbJpwx4=Pjgn4oCo{(|0+KTi@N>Gg@eyY`kM->Kj4(k0|t}V->|X5&sJ7&Ti?@EQvkl znXn(LjGdN${34Aj!8>rpd81)IpAxdQ-QgTUe2`{I=Yu`;@Z##X=lVElo%Xx|9oh8P zh4y!fP0VkEM`RV>3#L$s=MJyDbU;SEkR#iT;T2Ti!-YM@%K2`iK>LMZwnU|1#rRDhG@PVOx*wBW7p~IMDlR`oP)))> zDak)qhRMfI(}vQ6m{t<;9kNvOeay<@lHN_60kGqde@Ui25ezq#u#^B3JGO;f)= zUW6>tFmPe}2lDa_$F(Kpp9_ufPlCybIvKR+zH*KW!+6V4iHm@Nj-*h8-92GYet5v} z8C=^^U*Kh!$sVntjos-Fli@*b$b|4{QOv0}uF!;}>IvTFoaZ`|O|1)-0s`F(#>-75 zOR)YR%o(T#k&Qv$N_9*vNLvp-exrxhzvP$KB{rss;4@)84%b-~YMJxemLn{4#}u~U z%&hy;FMb)*tdcGstu@Se<;J7(0Q{}5qI=XPdyZ`AEY3uK6ya2}W#?G`MK}qF|r@b@++GiSU(;w7^rh9)8Lr+MP@^CR* zl5HK(n8&S{OSn%oF&(K-wO$VlrpZ#9f)g8u!|Kz1;_MP-K7n2~-8NuBMfy@Y%_V30 zDz4XDj>oi3pu_Pn(`Z3p0l|>Ssv~;|5FN6v7zX@=}0VFFCUU$nWZAzYeD$ ztwEq(Sp{mU=D_SqQL#adwf2l+mocn-S7_GAk;s{qSf$2CCIf2v-F?snG$E&pI`vYg zq1V;X(A1)w9lv!V67T2BS0^86i~DQp z5MVDYUN=#BV4GSb(;giG;4Sj}3 zza&z~Z-@vdcf#g;LQV3}c`3_#Wf~h2#XH8iNzdyL@jYv4&wq;6W+DUS@cPG~m3~TQ zHV$Xfa%t9prfgqtmu6jQRmfZf;jf=B&c`~f~ub%l%i zRxfg!Zhu(U9XwKSxQL5R+WF7#SN*sDbhho==CzEUqx$RZwj@t(5sMvyj7L?Uug~_I zz>uj%Tdpvk-h(h<=tBIo7pC8~4mqM8QpH@o12UcunmMee>)Y|#6+UAny3}c6!!?}l zK&kINp$e?55f`X-;)_yVa8f;3Qs?;tv3RN1#a%8Ied~@CIW}*k1}=9XrH-WG4aZYY)Y#_ z!+nEfC8iV)t|k67ChfbV*~dZJKChADConpnZ(+_~Np3KU>h#XY5+~l z7}^?0j5X6901UDJcB}}8-RY$4Fi?vcae`>JsIwUhCWkK2a1HKgneq*0&+~ptw30T% z+uZjM2YS7xB-Im8lC$wB`#8yY@!q{5kl7C9R?M4yId^VnPFs|pn_C#RYh#k|hh=|= z=T|y2v&pZK38(=%iOMg~a{cOpYEU~mRM4^VG8I|rG)Ka%!Z=KJ&�(4AZgGSQ-rJ z&0b>;lLg3EX?7@)9xe6TC!lEu>+_Z$k5}n!UTc~5+gWm(?z%i_^IEJo3k>eM;hNzr zN*D%_%*2XF5fWMq^yBgo8;^^V+WbYbRHp)W-s&DzuNOL=TR#dO?+}$r`b8rR{G+jP zyODbDv6ADKOKt0jjji!t($bv9<&M$6>3*x-w}V-2t5N?U^Ngh|6`%B5gJk>@ z|CQC&+5BPuu~9Xw8F^vTA$*i}gHoY1!W{fYqSs-0b|4p1Jt58~pu6F>qo;~e>nU6; zF)o_%C{Fu5|KJ4XaG4UVS~{1^VQzL-n6Ke0s*{V5UuG*r7Do zq7;z2$!_u&|EWsqF*|(g>oQThw-XR=;-4JV2)!0}2-!EPF)otoHAEsN{yvI$?PW!M!{Q3 zEso>=9<=lIZ78)V17aO2Q0<5VG^Q?PG1i&FS_~v5Xj81It!qQCQHI2Rv_x1oq zoE%25Ts;FF8g|u#vwuKs^45a5k94O_I=Tp!1=geGxslRQ*WlhmKgeItZqVPQkXflk zl^z({Ha0LC@mvC{*LttNL~h)*G!6~;^l|(kN*G5OhzV3xr^8XyVKLw2!qvoV#fr8} zj}XWF2QZ#|6O;2p{MGI$&%5kb1g4p>cfAyC)LK#f;D`6?e}Y9MO(YkK0B17>uDiD7 zk~Ucd(6^dy{qG}c{NzMaUMPa7q`qmi##L;-`?_{1{Z=Y5={v)cwAx8a+WBgMD7K@Il0^BQ`A7M{d{`f+U zdpdQwoF3d`oZm{$m_@~L&B&O?(ehNASRc$?RJg{ZEVJsiQur>yTOC#y+@*RyRnKMy z2UeRp99I4$Ad6yEv<-8L7LkI(f>INl41``3+p;}}CC^#AQ#XMoIY$YBJ=_WwphE0MC<2K)m8-uvykA=;aqrtF=vr(0H3u^WwZ+4a zYuHCiqc<-fo!ikoI4@nye^%rJk9Z1ApG2cT!tACxxACxVo_}dUR^xS?cW_@h9^0J!*N=)B$o>(9gba6|h+d|6ZwX%Dxt`kUeUTapHAihYampP&16-1lQh*6dR|gzAWL=t@$RP2JJ7%3UN){pp6w%QrDOj6WQ8>~(byp# z|F%DJ2T8_PoHDdl_+;f>Yi_0Zl*aa5{IC?;q2qO2(sr2TgdW&_RB_SGJjGTVs5w0K zJuGT^?z5U=?>POi+&>~&*BM7W$=J=-Iz7w3lcrT=`H=c)VoFH@w6UYjJ<&EX{FCN_ z>r=19*y(A$o}hPIfMMWiG}y+KzN`fE`5kele@tdm zQ=ek9UVqbY;`qg<$hkL*{83eIuK3uy5X+b`TybyNY@E9YsO(@}@;SDz68`fwC*zWT zm8+^438~Yp=Q-$pzMfT%;@X^NKDT8vDZ_T(Iv5%BRnS1(+#V}pH8`QXmw{+l34gvhS@aUEW-{3H5pc9!dL={n$L zpR`<>`#;rJ^D7gP4u!pd6LC<^nSby*jJVO}EJFk6)nB`t1YM4P8;wVSt}Z=mUn&bs zsbhAz`*o^vPuS~&hhc=c#+a(}YTW}*xSf|OcSvK^DQriAD>UezbtAj})e zlEl2)9Pgq(j|-H80c|h2;?KM**EeQ_V6UIIk9cQh@gg8k+3ySbGzA|VoObhfDixN~ z>j+Dy3sc-*Wsul;HNgDWSuw}@F*#_fz?J)fu;EZ#a)tj}byd<(yvKL2N!)`P%H)>1 zzF`oj^Hgcv5&<*R=4|Zw_syqLgcn4E4dJ1c>0)7^_atlym)2f?cLbF-(!u=ojck}h z>yd-1Y*PSuU0dQ+x;GHO)-vzzDr-=%uO;o8*Gh)BB{mVBCBO~%m z$qExHCSU=R=!Ud^J5ol&bPpYcnemGszMw9W}5G$ZSs6R z((|yfZz%DBL&4E!KI84VE`=kAzkFU*Uar7Mo*6Q-q2Rt`{cjl68PMr7;r0+))3heT zw9F&XqSM>!rHJlIUMh4Pk9}coe%@LhncVY_M+Y~bUs4+Elj%SEXnMtLI18R&j!p#A zA3Og*{Y^EOVcklJi3@&K&Lny3*{lCfmIwX6pQ;8}Xw$}!r-BQG`I`FXG~q*5|C=s4 zcC^EtDnd&^JD$1!ivNZmwfASUL?$sM{@@erxke_`l3ltv=egzYaOOhyuYtkZu|{lw zLjPWp0^xKDaH`d9i=i8f+4DEhP6`6LS^0vJLI9;kLLswQiuFI-RRElPbx{qzpm_TM zrN-UN(_h}qE5o(G!W9_jH1i5kn`J9Ol+Y=y-CaV#PI~!>*RVt9QDMWIiq2Qe0)5Fz z)k%}u^4+h&;}wks0|8u6;{&0|*Gwz|qm=Z;)Pj2Ht3BcguPiSWDx;cQWF%G!)mOXV zX@3p%7%$!$Dm?)Z54)4BoWNY2+WN!i@1I@X-N$uzjA(|&9xU20l!_7wv>BabKHt?gUfL14IqtCX2K`)^(q3PvB`}Gg7rzB4RmuZ=Z2w6sw zQK{Y!Lgp&4Vypbg#%%N!<+fu#_GVC627&=GwU^IQwwITdiAa_ig*C)ZbTFDcepcI` zl^5hYH-DhNrx#8>JE&AEH9PoB{&Wcpm_t=p+Foh8s|6Jrowsf5itZGLqeV9O!?nNl zdR%17ECD;Er#7#jHQqTehIzukk8g)kQSnN>YwX#+iP)0K)JgOD7;oFWc<4et*+Xy$ zt{;iopNGr#VChZ10w@GJ9SMff=rlPMem&`?+b|sRA^$OUaeuJ^doafKz7KFM0M{^- zSz73Cs&v9{vj@(wbsKyr5%adOCwN8Uc)7PrK1biIPk5R1>M?}Ch$o3hT)C^B;&4ed z$-aRD#)%*g+ZdA1d|cl2H2Foc7XD$&kY*eW(5ywmHw^c^g(~PhA(OjvH&%P#%6~*L zz{vH9V;|2;KJ$IU?RymSX{v?mGpa}qHf;Ktp|a`)AE0Eo*ddG+cX+;g1+~v zI54|yHKdlGBN|vAEA+s@^b(|WwbzvBs)_HuKf^8uHXN240}s+2B)OyE(kL9Z*Nhq! zD0-aIG>;0;X}XgBpLsfaQfSqKAs6*uPo06xOx8(dM0a%I_XBAdC zkU7WwS!Zc$>>-?g8;#=S@W_ValoGB=udNM-!exGjS(30c{bKdPlw4&cD$wg=kI+Ag z*R=7Pe9_@K1Hy;jAkhSiXm1I(nRri^P@Li{%Ay{HXn3->+}LhfTeyFktDkCMnQ1xj zMi^bHTHJg+V2zH}t&+;}fU8_>!Ck5e=8$naWcD`E!2qPG(%JC-7k;Dqv&}4V|dIz$+3>r2J3tb!_}O?MC%PEtS~#Ad8xwW zT}wKTI#x`1Ob%+gqVAb+&&w1d$wsY_FV`Y_s+PXjDk$k& z6nJV0^cfGbGd=;|!mjO?_t_=(Lmerd`Vg15U0%?TkRpu?LzZm*Om2Vx^INirQ%$7f zp>xqX_{X?NH;0e+BI_snM(Vb(cbOL3V1)HKx%MOEpJPJ1bRp~8a2nZJVX6GVow$3s+xyU}7aWf8;@1%`V=)pF2b?orhpKk=5<(VXA_uX)0GkuNJ~hU{=^*h?i$!13F6)o#HN0p z>|Cq{PIWYaB~N!;P&AF<$&R%()S{Ae+WXnRqQAXmX5c&00R+11NG!=+s*EtM{;h|~ z`ouE2na|cqBT*L0o^q-Vai2vi3t!tmy!0V7>riF8l#jlNV@N%#4sYwSA1;Y-Ua$Qc z#`J=|IP{qIP@jT1yYgL8`_V}Eb1wEds^t2b2UXgYnPBbgiGV>34poP4IFCIloTHjW za`Jp;Sori{-NJBIutw5mrx9Zf825-+gd zJvscAVZMZsD?{va%Q?~V`hq8~cU|&L!`JUoOdCApG$5Ccu>BKW_1svj6SkwU zu29!yNwqM{2x^5IXOKMqXlf-E$T=S?JmA3~hd9mPtof{!BD-T@wk_sPGTLn+^U`Wg zLf`dA=dgj07U@W3=jYjb4yaqWb%&aT6T8^n9PTQ`VRv^{*gFZA1|8>NvL?SW86tpB z?M4kt8P1kLlYPaW)}^p}*oB+HYh+kYRY-27q~3n%PM5l7*xwG*gUJFl$dPb1B>iHM zO6GhCHmYgoUSA(^fz3N=Ox#T~8LiHDid#2z)`z8A8@KRwJpnh2=I2=~qGmJPGapF5 zJC*m;PeH6lQDI~gzpLb!wiVR0cDR#vm>SEuVQ>lLZJ22&gq z5Qv!!+f*1IDN*v=8gzpgob_E)jlJBd*75SEr>$2SoQ8u8~JZc7+D1T}TuOU)p7Rc@t;ScI;w<17>fsE8*N z^UT`D44qlgE|Ik#8r_98+}S%Vg0;nce_>Cdu~^4I9g>s_7BhY_+F!#W{Ee5laRa># zL+f{*J989UvP|pP{14{dIxMa&+ZTl-gy0a|ArK%~Ah-v2cS|6+OW_tIcqjtF-QB6e zAwdh5!rg+q1P$!F$nNgl{oePzd*11H`<;IVt5|c*HO3sjDQnIlM`ZEJq3r#v3eWq$ zbAPw*^4%pl6A6BfJIEn(EH8DZk;aX|LwEW{J1zQs_U-h<#_c;n;jQd#2Ablk5sbF{ z(@CMAuXTyZ$s>`PES^S9V~(*2B`Y^7+a*72k_p{Ysje78CTeA;`1QOza`$@_U$38J zVp(_Xwlg)nw~r}klL=NR-z-6Ed)pVn~X2v0&gAlm?lU!S8^^;0d zdPa;Yl|!icy_hyd$%uu*^u=u7Z0_|y)o~MFy25ZzEOMH|{5jg>uHQ5(C2OX%MZ=_Z zs^CST#Eq6vlt*UE+0w8maFC?cSnPUWNRx@1!DRZiX}{hner$Jv(Xmw4*xLns5nrZ6A5u&#jQ~LkV7Z{ zN2DP>Hh6l4l2I>IfFTNdt<8F$=}Tzu^WXqjWkp>(ym=HfoADmr1a+;+sRPF+cm@r~ zzVVW8Ru0?{d}EWQGgI6ZYQJA-d#f!saR+hXd}J=1c9!OaDD*L{#b<7B|0twMXF1L_ zr6%FDXM~o#+>i`8PGAz28N0S0VOhxv$}_U`<0jYEx)wu0no`@G;ET>FWIjOL6hP`N zzE<21eCbBMRdYE6oHA{F1Jzry%Y2BeFD}|7I;;^JJJzp=9>4S|+@&5-nDew7Ct5g2 z6Lc4G9*3YYDyR}!ZZ9m>nC)tT6fP=LH6M3d3o5=%=Rn`#l%2E3)j~m zrKOJ`3<(>v9E|YMfdoztD&AjmXRY_KPV@b+?(?r{7Ekt@eJO3|(V+-o3u&Z+FAs{W zHXgI~8BhWn7$~K^iBCbh50Nt!g+uu4_Z(N3Ut5lGiKi?E{Fpy9RH~U()t8o==PdWM zIit^`aJr;K8t|~6ZU7IbN4v*W%r9xnOoZqhoYT3SX=n@^_xSR;Im zBrx`!)goywQZ?77^j}`L@xx(A@yxbRBAD9PXJ)soUZ1Ye^|2%rhj*>`7lBBV7#jhn z-n2~1Ljv09K22wqcZ`7F9~8@1JPnRMvguH7Mo-9sKhY#2t2Nn{#@o&*#Z6}V!y)+P zNBbd`*ER~h`on`H9Ovnpp?FWWl&oYw`Xr^xa+C8ekX@g@QcJ?N`Qq^ig(3T&ez zra6Q*Vi4akbPB>Kwj5GH6oqWkoP48e!O}LMz_Z(jbufkW=`z(92{T z*qn?;ce-|IMYC;C>`JwhP?~0^N6sfBDLE$E3-d#kQEl* zSZesRe{4R((Z323COSq7e9y$ErTue|v^x zpKxiSHBTbS{CXkiz+g}Ay&d-OETzV3)QFP<_xybtct_sgHaErCv3EbZ)#Ph5liL!X z6lug8uA92^c0!s|?YcW)bI1$6Q#Cj1%!eZMrk?esfi3<`;2sD%HN;Y<+c_9*Plm-X zJRlKrV0{h~fWH$@U+dnjIq%(I-gRXW-k(nnYID;i@svI^ETXuI3!)U^KH1&A>|%b zNW5J`qESCiL$ni(5B{b(Uo4+fR2);=nrm+ucBVR>*sQ>%2-;2KlbMh4^2|Rfq!6fN z4S#pZbw0D$S}VFxal zhc(PXlRBy#!-{#~y1ZUtSDxAF*^sfrStP-=Q~#9$a}=;E&5PoSYcTv_4#kn<*Vv@9 zXWsKpv{^!i{lzNbZaAc(2#&|G@Lu0Fu3cN6hH|z`eoo$z{j$Q&PzadhqBHW>-oRf7 zHqufg>#ZgzZ-%yn;V#J6dmanBiyE+7zcL*6kTPq2~XCS0Ae?2?NYmK3AE zF+lf3@Ov#p$f_h*Kil#qCV9a8umZQ5EL3N$(pz-}e1jXUT9peJe&VzA*rW#0I6lYhjV0X_W*SPQP&=42 z1Ghj)2X6ISLHgJ>srppB!u397lF*rAcaFC=>T9)xbIsz1_|qhM$(eS(>KiscEIuN2 zeU5Imfh5{+g0{xriGJihseI?W>o?4y)}%;(mDT*go5!PcZ*WZ*y+)kK>AFMM_VDR| zVV$-9<6;TS3g-DIrxqV-678h(H&=huJv+n|u9}3sWWzBcvvt-^M1PYEGpOn533sEf za%U1~Eo$!{+=*WGH??w+G`=JZk*idkp zvCGm2>Xe8jGDHjREym7>aVd{P*WhY|6%#>N5s_+;%FeV_yN104KOx9DDf|i3J1Rrl z%L5Z_*IH%HiPJi^!?CnR57PCSMlT=cAIpNKp%Z3gR8iI@T7J|@zNyj7H zCH;e^=NKbwJ1Y%m{Hvt#h>ojEMK9+`=mn7zylYgEU0#F2v;{Vm{O`rvIy zdwJG)V+ACel_|M+WT=CHV7FPsn=$I%ccc4lj&VyFtT!q;yhsS&IsuL#!93RHnKJ%n ztztt??)Qr4-)a9qh{~V&C|Qc3fKv47Ryua@#CQ2FZO4Z~{R{$lI$9qef8-(bQ?7>f zZ5YGmgoP)eJ0Xc;C>ytzdey`xHOduQ(*+-c{Ga-veJlP|-Lb^$L4J0}detrA$9H0I z7XF0XtjPcM9|*)`0rmcdgrS~xr-}2b6i!D5d*M&gO^R-BeSWxIz1qJ21$L-oAF&ZG zE)sBc+NoJ=Q0Ng(Pdg}zGra+~>xt)6CL|sF46U|bpl8M#q@54|T@~Puq#oErAA$qA zu4XBb9G-})qa3U*X%qSyue6*vn~}wX&l{JqH^>T8iFw}zuYu+~SUh0)fsE8gisMt{ zgc*m=$0gqZ?N-30Yt%!UX4?ehb$Y+lc=o{C>( zrJKp_(1i|~NZy65D=U#B+q0fWvve_E#??vV1xnxidfGGWC;7Rh$#{vusT&frn?o$H zEHZd}(d;f?#fDr<^vkL{1Go+Y(c2C(a` zpOx9&7t)!p8D+wnYW{XA@aP2+!}66bjY*C5i}la2XUOw&_?c$vjs7DpuGZt{QCtdh zuhsh$NtZsD2`u(CT4K>NMzQ9ZFzEFsEz=&%+XbAeD@+YDrXx*x2BFmwAza)rth^I*Q=MY+UB|x zfO`v{%u|&6jW+!Tcc$i_&@4_@oZ8#|)`af71C}901!a)(Nezhit`lX$RfKDpm@Jm} zDf=*eui5+YnZvIQ1n^e>t8ZovQ`V#y&@5Foeznz@{?Yd6z>X;!#`+&>YMlsYF!no! z;d??ow-Jvz%{O%!CGt@$0QCxoP@-G&tLHW^<%(sN$9Ax$Zd+l)MRmaZKz22p9d)= zs>N6F1UH>aSt_n7E_=3V6F3&65PYXEWh|^iZMeU_S9H8^cCO+d1gWwe)CLYCg-xnl zOax>s1RHBL5ogIcrG>f*V3uyX2DcXW&1T`;GYiB7+z^fk5p>K2KHH#P(jnI#9eNq+ zlab`iQ`hV$v^UcCt9MDnrjOtES5rxd_FHS=bpNg?3rJ_k{B`L&TG=`4{>u~*v8^Jv zSHq&wv;L(!B{7P{MT|U-X8KCAo_i~KYl`Z7u7+kRwnNV+WBFiA{uj5Je;{m$`9@LX zFXokll%-ePoyNTB175SNrl!gC*2#R-J4v=@M8MqaD$J%|kKYHQnzg+4BfXdV3S_v~ zzFfbp?_O&W@8U6oQ+$)tN~$OK1C7K4&9U%#HT;sMoA;Uz`O}*RK9MjF4A5W6|CXfB zvRu~TUvulr?onZuml?4+#wL2)x@=nD&$fOUaVy2 z-6n88j<)=joLl0S6a{q-TabDKLtb?g_|+7+EGvB_AdkPKEf_Abf=KFAFWD}vs6F&r zx?nw`R%Mj10YR5GJJ^jy(j782t|nawEw3Bi{=pi*SkL~266xg4`p8nMhj zn%r%?2sJx;3pa&w^+yFVvvkrKxBT!qzSIh3{W6BNo8D`%?*H_CysR=etg=Ns4_6SV zT!I_`2@iID;^AwaD<^PJ(Q?gKRu8a5a|^%gS(#t5W+j6se2_9eL0@0qtZQRvA;&em zy{Ira-H7EtIBcB8@`%i_I}!HMSKQ>Ib9C_^MpF_Bwk3gu6ZlFOlpH`H@hANkfzxvO z{cOIGn)@FhB^-YtQp{b;}Z@{f2{Hl;{`KJY^rQ@Rs2xSDUE^(tg4nLYTqUv9)XV)@t>1AK4OK_YfrY-QzX2h}t| zN6M~rL+M|np7c7X#qKXTRONVk)SQh>zt+Ncg9b1*#&>Il71!lJrfp*gwx6ASvTD;J zVhXo5=f);OpWWp*Q(ezH8*#56QMqig5KSgqk2Tzi&(Ym2F|G`sIPIO~=^@kp#>~TS zz~0pC`v^L}n`xELF1FR$k{1#)j#62)o80Lgw1`_5m47{hS=y&nu2`0n_b8*H`nwpK zvv zPi!`q8e4eeGkPCp^2ymojkG@yNRFoJ(h=MpqP@|U+wL;lMlj=(#QRw(yPh?9T*cGl z7QdIw-l1{{xGI{tE-y)69B%6_KM7d=MK4=_vglkdh~Kw!neLxvU%!V0mJ2g^Q5jdC zDnN-1AyE^0A$XV7)D!1^WME3!`a#vnWE?&|J>I)}XiO^3C+@ZKQxEw$m@D3meGhA* z=_l#;qG;b!AU^AJCD*8r=dkyVdaXI-oC_2&>ix9LodG~ujLbbhDXb^T&@dC@c zWReCn?pCkSv7p@7KyQ1lPcEDp%hR)}^OR!C0=dgY=Z*-7V|j9IY&YnHaaqJE%!(`x z%8E`+jZH$ozSYXNj;t%e2{oa%Tlg?C;*$IaLT+&JEE!V!r@{^WZj3Q5dGXNPdQci>YXSn(ageDvs`s382&)$-WA@VlmpTOPDu0SA8Axskd{x4TgF-87=utF!qUlA;~d5V~c#QoV99xKN} z4Z?~WNT@H~hTec~Z_p2t=01weLD?V6Y2yZtj>T-gfk?fxO4aF0m8Zf$czU*F#&Im| zx7y`1+S%+Cn!SW=o5o zPW22-d_AvLQdQo_nC#D75VC)Y)kB$KQR@rRLPYrNQCVE)N$WD-?49>so$RpwAI<2GYneyK^`OYG>)H`S=^c4}AucLST+1xvw-uC^) z2BKqKSTwQwi-mF!V`;f}KC`PR^{0@1Zeo3<_^HHFQ-n@l!G=9tsKY0vX*DsPOVbe@wx{B>NevjTX%#ju6=i>$nBbk7 z7~Ybldn&tK(oN>k7nSO;8yr0Ck0B)Y*i+iptw#_SogLQ2Qa_-Z)2iLBHml42gpx{I z6yij$qAGq4;zS_*_5tpGoA^2xmd63?CJm^6*w;^*6YWSucRnM~7v!Nkkyh6ll{ZN! zL)9xznN68VS#oGWuP`jB&~#d|-dRsf;O>{Pq+oKiv=B5XvlPnW+dvbWG`+{3^1KMP zmKA(icWnk4E@~PTVb^%hpH_w6Jnkf-`nG*Qz*CRJX$JE1Hj!uCC;!S3*|kBvhV{LQ z&!eGr*lTukEe|tj0kqXR)z_})K+C+iz=S6JS3pY^c^z7^&zY0UU2)M7$U-pNo+m?V z=Qfi8=9)n;CL8&*Rnub0g#{wYIKlB6h6_H@5oDLtT

!s;ef@b1nlcQhQ0LyK2tI3 zK8>=kKqay?O;`-ubwUDV4xFDY!H$k^iNgzT6XNNP+*``ODu+Cqua#w7FguN%SuQuw znB>ImR#()aNf5S|zF_X1U;F`GH1R0Q7($vI7l0S0k9hYjd`&A3($VpbnxRCBn(XpJ zQ(Hf6_CYlc;T;~WTV}>U+FZqur?rz>zs`0@$e>={&LU4+r;g&=Pw~oh-gJ52p+8=- zWM>?bW)?+5w4L8c*3zA? z6sG@*x@{8b+(n7osDCw9MOsd(ZE_5*rL-+&Pa2X~nib+Iu9#3WXZ(CShq}1qkzAB* zMBYp`1+75n7!h&aym5R?-P^xFg|mKQxAV=_{AAsyd$+8ja>{*3Rr$R$(74uCIi{-L z5R5Q`9-9p*Zm#XzYL6twq6p@?)*Cbmn>Vc>o||OpJ$5tzrV>Rf?rkj=C+}Pv3H8tR zi2L&~(T%oof;y8fUzFyo$FS^>EvRjopFkI-oHCK}bS2EOJ@uS<`uhiag9VPKZrTt{ z<~oNHORAgZc660BagCCowGwDO{AHS{DhoSXm}b&d?qZddY)r2lLFb6Z2Si*tyfkR; z#6AUkxuYjoDqT}gl@_p%*7R2|nS()z@bim>ZeO~WkYr9 zSg%l7S}-pV4bGI9h!?}6UA4G8WZ5D=!_tsSaLD0R59(JGdQ1Cdrdu@8BFlNffkJo{ znnEbc@iAKXoU0HgK8tV^-S8QbOXRzHD1Z3qh@{5KFVjblPC$-Z4i>r5g8G-(`UCBe z=5IhCAw6XO(|5*`AJ=y)&r>T-in;C*_Ogp169oMU6WmquWJex|xm`xC2BcazXLm-Y zFK2DM*ZS5&Dx~x%_tg#gd)+5-MD-81(jwTZnSglz4T?q z2~rxWnv>34cOkts@2g=EizQ*#fg5)(^E&r}n>fz#qGN>$+F5$~FXZ@$rUeya$p%e} z?Pv*j7Ai}6_eIJ$zCn#XK<^bh(674N_OR@t;PTarX3$k27Q$PR-4-pzG01$@u%y@} zn=b>gIpz}6OKV_RH^teU#7`N}QK;Y-?J!lra4jjRLcH`MVl;2We&08D(ws%(&|7q? z=SLZ5X0-cx#K;^Oy4|xB!=?~XC7I0JXUj*7cIrU1u5BqGLL#iuu%*7*Z&9xdDjz!N z!h7MD?-cIMRiecQ=OJaj zZ13hm*}jD5XNDSFuEDWWKq4s2=ck@G&C%n*delO;9ohN?!Du*?4rdrSMyvMP{MD

&nvBTUJ^o^CSdVhhoAb&U{q&~IISe7G4ed9DKg$%K zNRDi5b`0#PVJ!2_CyL~o*fku~rpH5Rn2<;TGm`2J5u}vy6Jg$6Jp(cU)%tI@3Bt%% zrpE8*-w6>kJbK&|br&I(+({N#g$-n|R;e#_wAr4US(p}}G5%c1-)E>lFLF6Cv-$lf zJ}#?aVP-kEIL*k^c*E0Eh(~*4p2AK@^DX=UshH|8v~A)pZG|7(=ca;E33l-^E&Z@N z>G4R(q*;?;`jgMTn)12diQbU^npCvVWOj)=6{n4foeCt`w=jZswjZoJS$vJ1XYv`R zswv%hYeaABUvFb-&o$&hJ|fhTnDVN$)Z5@jwd4P?(}6+mb2pAPb)Fj z?_jFZqpZxX-|bC~%IF#!g(IPlhbxR7)RL)net5yEt%f+X3{y;$UJYYzWUs-7wZ?K< zAechd=>?E}c-#4QH*Ay?@d;xw(UWf;)3lcqcP1UJKYQoB!{gJWtg^5PmeRPeH=t!9gAzfZ3vfmjSk+hwqP5`Sf@^omb1*)- z-Xp9!3mK}b;*Tew?K0kXJ-ONE3A;<`!F8-cq$9x=Qsf;x{;E)zJb!i%9eA1Tc$tK zr+9@#pmuXxeUqFsXbW#Cz7Fzv;q2?zqhcLNfs|=5l?1rWoI`E&og8h7EO?2A_38|n zy7c5F6!9QXbc$iM24L(<;Fj47rF(Pfj!DwO%-JGQ&lCiUuAQ zmM_d-8;Fir*Vz~#t6i|Io0(-ZF-;RYed-7bu4S84mqf@0Gj4(W@5jGebfU@BzhgEo z2YcOk$~>kO#F*+nYpsyUGQ-%+_zaz;hafYB=$L)OOqwl_2CHfI70uCx$Bh3P!uCfT z^|()~e@w@G0M)U9wKdDkn#a}(kSX1LiHOb%$4i*!otG*&uM@u0iROl z;k3ptsR4VqDqOSX07*YP?$Y{IkP~V9DeFzXR&ta1Tu;hwgsS@Ef#QwlxHO2n+`B@- z&U@)M2acXxJSacxm5|wcga!!KlzznemRueCYTxq5E49{)z1_8D^emY%FyO|4=Kbv9 zH#4SRBL{6@6y_J?h3$LlCz=2@(*?@Ois>TE>HT-EmEO%>*BOKRd{u1K6)$15UyoEI z1ydB>t@|rERSjRMZQyX`ykLAG(rTBo^-DDW;~|iLE391Mqt`sw^i`xB7U16T;9FpR zO%nTp<+oG7e}v38m`*FJd*y_I6xFRVc#|&{t6|^qi%L^n%@mPLHRC}1#H;Qc6yt%Y zarzC|N>V&3`oU5Kk~ipz44DS?w%)rls$ULOPf(dWfJJ2`+HuB9T=3pE@B=uP8F z2bV_LE0p+VlI7?5EF4tP@*1aACSKEwnW0Vke82FD{y;Ev&e<+|mmk+HL%}B{Xi)JD zi?jZ^RQ5k{cU9z=D!dL zYKm&n4zz+Md?^G;kf?cl{{!JH)M2jK6}Ew^n$PLWofxVhs%lx1dxILHB;#$sLpdJDLMPnJYJJT+2uEOf|E;}Hiv(LXMxY6)3}fUW7mvCdB00n z_QE5L=bp-%1DTP`r-=j&XGsHOU*B7#1zk%Pv7GLexcGLS%xM)?cteLtOP?L32xmwQ zDdzo%)k=eCXz>k>gh)r6n=pfdd*)Pm^GLB#RcEAm3*wCpZ-PVT0`Op?vy7`9z_D z^jLcJp0U{}RR|WM%4lh~Q*8RLQG;Tg!%sJF1z2AVA90+#8a#sdy?VpBTM5;rxD+~o zwjMgMI>o2n$E%?-8S>rj-@3e;>R+6u+jO53!P^4J^F$h@9(to5rkyH5A?gyn>-EYU zJA6Akvh#g+4ildtUZ!YJ^8sbJlDlv238U?fE24GHBVq8&vy$3Qbqc{+ zPFC69?|vYRgHZx_uHZ08w_^cPAq$<`Pc2P|+wSi4lvWB!_>aXSpYoU8&K19F6nqdd zU0S}V(*fFXXqoI_d?B+?+D@a#HrHe%^fh{}kx-y*?5gmA6i>J`o$cnu#_I7*sWqX5 z&>F`#IG*aXTh-c1aIcBLIV;THJ;*-{keB>X{N;k2aaY)~lGx;YX4YkN8-E~VwBI#r z(>@6QJolm0s|l+~R)vB_K5%7hgnQ4D7NN{|z>jj1y185r(2II#XZCYW{2|Z9i|DRw zZ-9fiUC!OH3q{1T;sVRZ#`8R=W%xDU4?RkY=hmLlDqzBn45EK)_s=HF&4(-2Ol&2# z@gwu5#hTi0Xa**PJU++FiC&};+x#v0pUQP5aW!UhkZbb!Sb!?G#od$+nFPT{9%HfD zY+(U4w_{W+x!bF>P6!vZ?ebuD!?TO zD<7Hh(+yu@n91)lO5)XgbLVZ%iyw-75a@T40s`$|kLj=Sg{iV#he8dJIc*uKF5QL< z>VKmS8Gv-SQm#@y=WuL-oc83~ZBrs|h4OZtKM+z5Jbp=H)wWF*S*1FR@Pw7OrQ$Pf zK9RbX-FV$IB&;i7rl&2cDXGo%Qk^!!!<0FuVpwm*QM#KSoG2DS^a4ThB@@E2ZGhH+ zykU&yGnjHdB@~4DAn9+U0g}R5aYUVBzG}54xL3DiwE>w~B0jm@x>&G)!vs13C(nfk zoqaVY6pJ<2IP&n?_}>BnvW(LtpEMBdW?PX4 z(o#ZoC&tt?CiuNV;C)P&pvd{AtzUn2bXO9G<1y6^5S~kvu<-%v5A3C{--X~vL)Yp5 z(AjLEOoGx>^7%;0W4w%JZ_y>K57Q2NGw{cER1o3cV0q|;FM*a8k07EZikcX@6l84w zRWaxB_CiL)dh>V7{SA4n>8MRaO!>A@`-1o(vhq#G8AgTrQTBx~n9TSl8A0%0tx_<= zPh~@bN%zS+57kX>b1NZ-`-jmbOgUofFX+j+(xr~o%~yZLctv}Z$n#PYoRUBtg6f84 z;;)DNmm*R*w@nhsXLIgH{U%xTKa_@lnP?*H>HIQIwcB|4S1~|e19npv4LBH7BIBqH zri!zb0a+f%!6Vmy;di!B!fM0OF$G(+S8!EtL{d@^oME}MOcU+V*y)<{TUUPvRBN}a zB*95da_DshY6E(YfR5K)^{uur*28F!_FDnw=}(okJ83a}Si|odSSg`;zvk_bEsMYL zvQV`?48nh6Y1feFDkqSAfH*jF8r7sPx<~vj>#j8OuVM04t>R%#Mn;8HTW(d%5D~HC zjE_dl=xEhOj5_CaC{pe}bCu$PFawkByk3vyD zzAf#V;EV)jH@kq$DKv@S0Q{5fqVbQ2L3uG)d%-BP7?J|7wtQZPhq7RLbd3HUx&GeK z7w5%ot$HKkDA`s@A#dA5`jn8*fg;g!BKdp|gxhLV;40l!E45xhJtpy-an4JRc*CXk8EwoKtCQoRl_2bZ5}x)z(q#({ z$c4z>A?dh@=dMvsQ9>J+eS9|;TLs(<3Bc*3fAvQt)*$+QT&1!4r^b-5}gmK7VC=_T<6 zqsjl(0K1$^{=wznUEX6C+RVTR-{^#Lcq)6ap#3GidFq$!x+|D+MqVOjNvUz3W#{{tMZ3i8QA@p9 zZcn23kaGUIZ#(YgZn}*QBv44LXW1kzc?4csUR${eF8Gbm-vr=fwXDFZQ-bEEU7Fpu z=oC}h^IsWpD-KC0cn8nwrm2ov$O9(?gfiO#g{^)o@J}sDrNUT*9m%mr&H|Ic;Vtd4 zQcK36?g^2Deq(p9ev^svLAVMc#ARz~r4Y=hh0aq$vBg6FSEChkF0msJZAK$sw=$CVJ3WUb$36eij{r_)-lW-*V^fZ+k|)t%McStyM#}sXh_v?9@r5KhbZ-eo zh!%IX#>eU>u>r;WbE|U{?2{sFZvzghO0eoEkUp&uuMyZR_b++uTo&|YC7wQh^#_8G zhN;WY43YnZkd?td5K^h;Qtu<$uczP*IIA;gaHEEB)wa=xcx5%%LNXqPskd7XOI1L< zu(9nqH`~YaUvm4Do(_jmnmovoEaIucD!d$#E?=-e4I_Rci#?D0#6>TCn3}x9969vS znkWU|qTZRn;4f8492_7B@}D~df;f1&c4L87fWd8H0q9dl#w*9+{3y@YS4gGZoG~&6 zoq?iHGO;y>BKGvn8l~ThzP8}tUNEbmU~xP_?z7nZX2sv9C~5zdeeiC1d(z$n<%GJ8 zVk2W+O(19P-gS&}6KZjD8z=nttFJEBMRk?wF4EVfy7z5AuBRnFra&|I4)u0PN?0nl z46#4nKHS*QimjIh>^Kya^Gfp)q@PwtRk|6-_TUy-ZP(u^QiN9_YaE*{I91C|A%QK5cS~i8yqS0ZG5D(_AK5?z)D&4Ouu`v ze2U+6IS2iEa~{{EL2B{y38!%_;^x!GZA_B7LhPO;RZGQ(E_sAfju$uyy`6hpon6@Z zMAJXW&n}Btat&s!UTqY%>ZbUaHJVMV7Y8rv(R)kp>NWjuf(HDjl<{BCdKTt4=OveO z1-PE~aYhMuYPyL$i>k;j=e`Ba8ZrbwVsuK*>!!FuTbo}%sKqVb8F9}y33iq~zES0y7UtheO;M&Al`<-bKy-VmLDU~k$$69@-FW253QBV}@R5tb(RmhD^p_s~ z^H&USyqwQ8KeUW5>^$2FhcVMwK=wCd)}HCwxtiw%jU7!Hr%Lt`N+6CHQ~k^q9`*A4 z%<`VeYBbSHDk4h2i+(dT90Syp^m2O^dHM4#9!UvvLLOvN_ciXk4QcoXx8Y-Nr2`LP zs?XTfCI1qz;Qt;l;(x1qOdLGYcfrkS7n=O`KdA{y!3baeKo}~xsflJe$^6h^O{%CF zd!%Yt=D=LlOiWIOB_`NTTKY7WpJg{1Ft>($&vL`KAIEX!6NVZ*cJA;o7#p8q%NoOx z{2q!e`h*f6NfZbFKV;we?--J5*8v7^OMA-p*^!V8_tPp5?~y7d>PAEdvs{Ut56@U8 z2tB@)%PM_T&&+HvGu7QojpusFOrkc#;Mx}6HNy09i{s8BlZ_v%)f!2nA=5dq3W&A; z@9^{g!LvvNu`;O7vX+XlRx6UO>%_=?p%R8teo%$SAmTHSSG2jBpg6_U3Onhu^${6eI~mgnd=lUsvSImU&@Nh{<&Zy_`37*Lm zn%xF(p9vxf32mc)do2GyJ%%t3$5xLNy#`=&(Mf@}mAq8O9<^}kJF9N;I{P3l4pO_@ zJYh7yErWN`@DZg*xD?^!aq*?GP2xbj;o6R>bH$f7&*i~2Wo4ra%t-c%*Nz4jA@jDF z5n~PxX7a?s^>Z;x)A;2K&%!DK{wH)uq^j-IGv>)tQN5+5NgY6sxZgOi!tNax0--<1 z@p5^}%3fC+E!SmKO3-nC(Z1p8#iE6~YH3=5ZKux;%( zcg{ijOFWB9@(PNg#uP~--@(ucVkYor^8RhbvnMqh^`ygeaf$==tbPPq0wMnzpaR%k$tp}Ey!P^mHU?8L2`3S339z*Zy#|G=um`?FgZg^+vH=Puub zduRa=XGv4_rKChff)$y;Wy6n#uWMO=dba=nA1ePgL5T{g3BMwKSg$<;8DKu6$=0vS zib`o#ic*&Q&g!CQ>#j4y^ToyXyXQIb8pJFKuHCm!7*?8%c;c@sC_S0LvBw!46Gqcj z#FA<-vvcqh+o0Ux&FmriLGk;w5h&@mwDA>ayxYIMFi9bfqVY{>_X0Q3p}e4i3DUdu zk=B8f7yNj&Q+8oU6%{0VHqH8y-RZ4lL4K<1q%CRVbp-We_9P_hvij28F)f1J5slyB zgX>T(=}Gs7bK&Ex@sU8LfuW|7ap#O>eVm9K!i3g5zf7myvdMpc4gBB0+WbtYi+tEOGeN!j{5Rc-VLLh+-q1C#bhXSbi$xCJ3BWDKB08n!-km`54EwL(^sq^ zr?mVc%s@V8z(@?w$?dY}E*HRZ(;)IS^K-p8v@jLR0272+y_&Bz4sYhl@@^_4!TlPi z@jnKmZy#0$qV?O}@mz+?b8uDQe-<0@Jyrj) zW&da_@8t-q6K%7zvgVfwcdydLHSc)+TIS*5r|nVOB**1nwJ#MpPB(;`dbT*s-PX8X4$LiGg5kSbVObW!zvr#2|Lx~w z*nhqMMF86twh|KDFJ6%TeN*mVwf$B?QZXznq)S3ws!Kxr!#|Z)1fIYM3oEY1!FwX1 zF%uT{XFUiri@BMTJ%$w`~R)6J9{Q(-dWX#fZIg+wok2~J9lLe+JjR%P4hYE3^o9Obar#gct{ z>KST8MJn@A8XJ2>kl{;& zb*d=di~?OTq^JN^&Aty96samkfAH-qv15ES@#k-aM?fi=h!5CEu8R4R{2{m+a~kT# zYy}U`P_tDOLmDKx@f=eM;4jj}Q68T27yDhrWG3n2)$e*93KbQq& z#{FHmD(T^!($60K$%*&xM}7OEKUk;Ag@rZ$?hF82#6K{^A<}XOGOV&yL}f+rD!hU} zpw1t42cX8BR1v&N8lw+)Q#5d-Q9i--0vV#>oKv{t@n%#yEay-Hb8=D)vjdaV5!6v~ ztsIlojm5h;Q^iq&0XBf&*}?pDQTZK4MXQpWcC4tz3Q3Bj@}Jt(Us%o=(Bbuc8iO>v z3M`QLmp>|toObAV9pGKwhq4{1C=aC+@ZSIxwJT2arf?@Bv1$C%AC%xX#z5t`uH1OI zQiC~Q;KjC6k!?+#?)X^IaE1-5h>G;3vTA`Tct( z0*g7O@O$*3Rv%s0g9wr?cC0&1$+%MR!_?v!PCJHvFrF%>?PF9-VpTvqgju|Y_Hq9g zFAT7rus42(Yl1fw?|m zWa;%_;dOK4+pbSd{fOZ*J$wuyJ*c3ofd-W;JT@OVpaI$l1B+s!BgupuNB2pC!mH}I zjX!m54Dhh}Hdq+%8Nfw^5uRf@6C|vlU4dQ}06G-|pgRC+3=hQE@b}=SNH;1;MvdxH z0gQ~sk~3iuz^Dm~<_xJjxpHYPelhLM109X%p zp?dTR=tZRBe}fxJU_pn+1DeYv6}b#GG$3Qby7`VKL}RgAxK*f0#61IERb4}rg(4`7 z>}RbYc{`_!Y7RfE-s4k>pE2^<8r_A=%zL;}9w5rX&KHVJ5$4l{tWyg5{XxQ?({jC`sh zqvA1q)?Hpzi9hE(aVsT_-u(76QM`@_YDhk?^5-^Y%L%hXr^8dl$fxN;I8%Q?3Up6A zaR7K75rL$*&jdA8=1>{JsDm*kdIO16CB)6~9sxJlah03 zBFmBlj>PcBB!M;n)CXX&q2KudZLb+$0Tlg31Z@hpzv}N9)(!zr6?-O2EP}eq1(>v* za~@`G?Ej~IN`>FwaYi{0ry+mRzdr2fuyNRbBr%|}fRZ*FD_9(<0{K6CWQi@_?W*yV zjB)`Rmc!hk@5|W{wZ`@|-*Vd~ov>{CMoQ?C3%Qq88?H2{etRx)c&023xf!DA!miaj z9FbX;p3nl!nA!=M8_SS|qhC*j-L?pJWZ;*)FqTH{({&4b5%x@KH|#|;^{LpiW6>uH z4#8xUwAhl^f}{R{&kAguxfPx~&jDi@FswcSpnUr^4qWt7k{e5bo)JY=@zcigT%d-! z5ja}{ife_VM3UQ!J(qbPsUG*rQ$;Sq&j39m0x^%$8bwt}&ns#a{S$W#-bN$i|6=Pc z!`gm=c3~WfQ?xjx6o=w&rMLzH!Apy~1*Zfp6nD2`fgmmJUc5kYclY2<-}L`H@43#0 z^X+$KZFY9Ed+*NdJ)EZW#2d~L#iyUyLNHoru{D>^gMd0B^1OKl-~Rn?Mv>UKY+n#* z4EQC90JQQmk*Sw(^O8vncnbm$tr2`q#lp%*mc*uD!Ag6>kxdhH0a9Mbuzit>(*_C* z(DH=ygxBCrlFx$ce`_oiKaxyI4Z8WE)KItM|K@sDVT5vBd8qs%$5!y1Q}oqP_<6LG z`%eCS<0E;;E}g&_{KYLd1oKf}YENt3AFubhxfalEb2?+xJmxyQNf(!R_~Faq=3JdS zB5p(T85JA&YGanKLWo0PwC{4_i@xD@@RV23`z{l(WMRG0?)QXw{;$j_VLv@h#IdeP z--`G#Rh7Rz6u9MtJS&&nvP83%#Hy$%baH~X;g0XZYal^Bv3GBw7eM^vDaMN8f{XWH zE4ZWI&ypHhrJ ztBg+v>%BQ31Cd#y{31W zr6SI4MQ*kFfkmfVn>(-|NPh7n=@b$HTuuGCcQ?%nKC_7($%D`-je`yyI$amSoPBPr z&}9d)FKf7@>p86ob7 z4{QDF^IOpZ^r1BUj{j=XQRw<$WBeX6C?JLQUlL}>ycI%u^iDRVbgrrJ&rgr(QY*xZ zV`~Rd?;QOEHMfn;l&lx|XN1*Of^>DMuS3WB_w?UdOK>T^eIdGu2tH3S9n9b3U#>ufBmJ5MPv_FvBiZ%!2Gd&z5S@7UUTfORp#~i+m`k>$-+Ns zLYSufhzq2GVTg$!i|F~Ya;elbfKnU5r1xg(*|vQJf(iUYWSL#$j&x{SLN_f_^y8VV zIW{QgzXX64M$(ipA}Rs!kpw@bBQVjfgw8c*pK(_^5}9vVr7*e8H+Sx8PSmr0tb0-IRs;+~MkWCQUgk$cU?HV-hXX)}9a9yvj))(D zX3fW|h)88f-bcKh$WJF+!17^<9z|)ZS~xfb+4?%*h(NQu3I+>$`v;JlN2gQ&)EoHB zTeU1_&l_;&sc3FCSf7>=f2kCHNRiSpfjQlLJ0T9}YiS};+b1!zto230;i;?ySIJ-` z*9cQNAWM`NQ~>`X8OR=uc6xoXwN?80#D>j6RyaYQpO=rowTOlJ_5Yy!&<}X=njh|) z0M67Uh-H!jOI}lkL>emys~{jBu?)Q=K^h%EfX8mvsJtN7s`epFv*2koBK0(^H*yFd zAj8HQKF2~^LLcFdgrTVt^*R@c)lV7gbqr>>T2(b0RzWPg7_sQX{~en`KhPsU ztSsVO&&Yxh)H!V)?Ogy%OgNz|RBO0Zc#UhcQ**)c1&-y{n9yCHw}DTuD+}ru#jP_S<+@M|om@|{9E>^X=_b)dI|-Jm zr_Vw2_Wf|OXK&Eu9keBOSglj770a-IQ*aiX@3foj%u%248yh)S@^db9#Of{Q3#?UC z_woJ9X_%_W)Ek0*Qcv30+7_1&XU99N_kOdL0+bTuzOW(-*$=8_ebV5}xUy(Wzp=-~ zv9m~M_uLxPT+Q*f6ANaSre6i8%TXsXl)d)y_E~EezA-B*V$C)p-X|pP&LM(5hg+Ek ze#$9H-!Ase+MG>SU1qhAU>Vyb?S@;lkEphIx-Q?doUfTzLT_2k5O_3qj7_7RdGlc=p!v$C-15 zcnK5Ohf0(umJatIEq_32b^})=Lcc?H+43el=zXUYE{1&dY1H+0$__b;t53CTO426a zyh_4ITwjm5O}kLSgZ!3sdMl@8Y||&Sv{OVjG5yMf%iBhaGx!A_`~e#A4Vz@18k!pB z#U*vNN(_Q3aky!`s?D!3%h(mtnCyGE?kg7O$V89j(0vQOmWH0KZJ~y?HdZ|{YiHSt zNvOpfOEgzh95XT|k6f8d@Xivvd*4^rlDuqaT4(F~LH8iTN^;vN{GM^?cstzDjQLme z8QfrPBIP3`V}nuSQIlD(?$&mT85`#|1NfqtLjnp%jrvl@Iw>fHjUrX4bG<*cV~Vm; zz(#R$8daHo%WdwdOz^m^6RR>u+5J%(A`Y!XD899^YZ2Vp=Sr~BY*)LI>tuiesnlUB zLl028^tdq_7|>oUYY@g83x7Yp)(kYT9h?`{uULDBqTshx;V)`uRNabWCo3~|k=!x4 z&Ft~zw?UaahPXsHB@*L^WBj560g~Sm=2b*Hv}%5ZK%fJ95iwR1KTG|P*l$5A{eEv# zLoY_ks|B)4BI?uI&JwnpY=t2=D$>E|R{hO!aUE=FGtA|-$H_P#Ueez4$>uK-8Di+3 zbKl%BW_|izc*hwOsc2%mH?K`)N;DvnIr0`?V%Y3U&~VIg2koe9HTd`_-`?IVqb~}1 znwhM9`_QI*oV=k0@6!IkLXSO^x>hGadq{ub-FYMbw!7EDdaCEwp&q!;B;aylLgZY^ zQ!MKw-DT`D;rO@F_>m?Oep#tp5r)8i2sM0q!P-?@z^%DFW9wRl{Qp*Ej#FtXCDV=Vb|wHzlKr4WlDCG zYx9bC*vCRK<_3_0l%;aI$x_%#M*q8lLA<79hSAdiUWw{;e~ zpG?rjDclom@76X!a(!?j@|IHl_O|3aYK^e@0sJ*n!XcLKwn*C}W->4F!n;4$T6?j0 zGk(oHI^a5~bLa?F!KR*{+)yQ~`hCoVD1xpl@r@~bCoWl{HD>gaX*Fo&@S}=TbY*Svq7CM7fb+N)ujMd6rrtXUyEy*kMvL2=p+f;?-jsuP^aJ})u+Hse^NEmGe1IoIN zRPYv8T-16=Yz&_DX%A%ws$%njyJk(lhxfqKKh55GshSCpV&Gnj;Kp_Zs-K@`T>;Iz zFJ)EHRK<%e6(Paf4BC7O`yiU*iDSfarPB&ta>@ULOds6_!?!(j7u8VbwO4R*K^@hq zV4Bo@n+giI-2ga3D(d>*QDM%MEo6RW_kT z$am{oXnI;{Nu0#5E7U<>;l{RhF$kAY`moPu1lXMqG6n>BF85ow3yiECYd2h`Aci?1 zPaVWMw%Fw2K=31ew@u_B3u3FFr8HRTAyr&goZzSYgV;zx=gVeseVDK*mLtons zdU9aYE++RX+j{cPx-e222&+E#> zsYP+-I)B5lQB<*Mf-x35MhV-l=j!pM*!0`Hv2=~8k0IBm!#*+j1e9@7Kc25Jr+-po zcbyOzNI%5IXXVc1mUdt^_#5VxeY%3VJ-Gvt-+4_1p1O|r@l0Np>@Flut%kO$?HDxn zC~z7=M6@nv%rJiZMdF=^{I=(u|Kg0v@W81@a{112q%?5Z*6{AY1{%OqRENiPcFIW+ zb@is0XVUyh=T@S5sf~?eO||Ey02-N#VTmMv%dW~Op`0f4tMtCYpTMR~PFsCK1Qzhz ztft$SRbN81zG{F7+_tiZboR<12lhydXrev!71cDssCIS!OBn%1x(NfpFVKzR+ehG& z+O6F}2j_{wfUE+i-oUv}KD*3cq-6ZF@{JPQ3?C-9${Hwv=wq5pF9*Ff-kCSE1T$qF5YF~F?*Prn}d zPfTj8o~c4kS?@F9zG_31;YLEGNDtJ-0`GnvBWUBYZ5$TAOOdRK4BnRh%c1YrAZxIC zulIQ6I#fbCa?tRc@aG(Itu6I1tZ@~OiM`{%VYZET}tZa^A+I+$&=izM9cY7rXj^XqSbT_8ZT!2Qi?G| zs41U`N$5AkTnR>~aYK24S}Nr9iAiQU9=T5A*#c^Qmm3ia*gX5+44tz=rfiT4)@*H@ z%r+1r9rtosuHucO@>u(9gCV-y2Z(L~LUBJ6in<s7W7(A+kri_@S2mOeTT&H8h{Pvg4HST*XRU_&$i3Abw57IR|W>x z|17D5ZJ5r5{7Hhoy?icT1*&WD+owwpR3uXD3MbX(u-4faoA#IKh+Nh(pwzx<0dT`xwruk1^5 z;G%`{3Gw0pd;$p|LKM+Ke0DnB)@`ruE9Xp`K*^CRU}5h+ma%M<1qo*(-UQij_YM0} zDyWNPru^52hrS)yp%CgIZvUF%+Ds$hrb3-na>vmJ;Z<;`WhN1#cd=_w*BxU^IcUR1 zHO!&D?K~E7`W!!hU*Sx#zf(7~EMZw*o>E=;Rbh~^dem zr(*~hxfCN4ly<|_xXcJ$I`SbxAXiiso0%*sZ>d?f2mTwzgir54#*ZYEW6LRSr zLVM2DKW<06Y`*^zO-nwx7wV^^a|Ngyu#6 zWkjuiYIWUer_@E#wXJXb*pY{A%dfGB#KiOh=CpMS0bg}bqgukuWT{f^cc9_(jMxQibpPCLDaIbiv(wDK;Qyy9SB5S}Co3h=&_|PC6miB# zT#BZ$AjHgADRCf6xBcL%88A7C&vpXUSjh9jZwMNzrPZE;Hc_udPbsv4t;=2 zjZ-1iHSd_A6*l+U1BX{WC7}twUb`9biXFIo|1r8x{V}ImfG{S92-g(B)>hWb(|0^a zI^nW0rsG1{>|JV0hjxshWDFA;1T6NtXS9h}SCpr~^0UVh6wKo`N}pfMiDC<-LKMrv zhU-3AWypfY`U|3cq7(5??nDJgM!8+IUt0H%x7-z2k%+SL`as&E>YW%giV9a#XM2Ze z(Ze<-1eb}6gQA;B+4rvxV#a55TQJh+o->kcsK}p4`|tmi?dKd&jTM+qFRZ;2=67VK zbI?vy)JGOB@_O~4`~W_PF>c7<%D!ba-{1QoHKX%-LKk$r=g6FQ3T;s9=RCXN+%dAQ z;F%X#Y7ROkGk2b!s@=`I7I~KJZQidSHO(`t!*gf+h|Ye7#x17CvV^*WsT1C$x5w2W zazaj~BA|N89pAgG^cQJWF4v0bz%{xmzM)E%!JW9U+*`N^fy@09YiKf+Wl(z$Mybx- zMZA_a*`6n@OPgM@F4HP#ZZs4WRmiM%Y3#Bk)klc=4(A1qZ(R-5#CG^(+I|cIsTX#| z+C|hf&mD4By|6l3ekD9iliZ<3)#NhdlFJm+s3mA?#jAIhnUhU!*dM!|p1mmDd4VO3 z%I}E%uOwO0Lnpc=2A&Y`)AbS{ljcEz07uZ9_~_1)Vo=qGd0tR zE~4&SKkQyV%^zcb*|y}4&xpOUTeOXP%vtkdPUvWKjIPeGE2ZJ3t6E+DJD#AlE**;q zJ#&ui%IFn!3eC=bqe_AHA#*y;BxwrmC{vQo?Xfzvz$W4M(^3kg)DMK(TktU*LXuxz zcge&tfoNG!ks*xuDOmKY7tw#TI0u(M@FT&9j@}b$_b0fX2XFYm(xfL7f;5+oiaK?8 zR)zv!^@$TIQ)PDt+6)ero~T-&^9lkuX@FY{eeZ&DFn~VCp~AE*5(7-QT5%%bI-+@0 z2|@4QEt$FF;%@8o0jZi_I4_l$TIJ``_R_>@xjAc7Sz(tVm0)arLVx$iXN%q%&gKzQ zIu!JNDJR?j+9ycMCXAYot!@nm;H|Rd9ggGp#c6LZz6gve9hjFUkm7&2#^QP!2kleN zr2Rz_jW_LrBm7FG2|As1esaSb+!wON6kr5aRxZ@xLG663To) z>!yj5&X4$niY*}zB$4kyM^hypt zT80!2%O&~`Z+=ndqD5-G%|o&?fjvzW^X(u$=JHN?Kl@rWj~>N-GQ_@SiA$X}8mCFz zM4pmtOu?$wu+uHDvV@k4!c4i14>+}v{e7vOsBd+yWqtqnA?w7IXG#~x;srrj+^o7} zfrr(@^kZ_O5=9V3HOoELh9jBne36}PfIH~aEB339p?iM6V*^Ioy4~u=E^hMa#%2BJ zq=uy5y|w-1)o4RAO!t}y`KPDW&uYyNgNoc-7t3HZJD|$v0-G-8^kHe33#kUCS=Q6= z*uhooTj=^^q=#c3gDDa!TnAP944BENE=}i?UTxo~u6#I2ism56yittnl#v|9jO|h3 zK3UB6YbMm2RN72>+vXwN(?jw7%wSLD1hd4-XH| zE5P+;nqIt^R^!G0;Z1+8aMGQ+i%Ii9@r>P*E}dPk4zqz=WG;?J)42vTDbxJc8+nT8 z?~>{7Bw6Mk@IwTRC97G>U#u;^kM|?v&lSLTvd!cl81c+dr5y}D0j`2G>A&)=l^Zt) zWfrFsWbdF_Zvcc~K}baMRWP)-`$ZQglZmbNm`jC1J^{YvmS2VjN@%W?lE0=R-jnG> zw+Q;%#^3mar4UmgJH>MAScKw|G#S|P)-D9qXISrqj)b0xrN^R9OWK)EfiOV}1yj0v zjCRx~L+ymCYv*^lJ1v!8%9OVCjd?r-VK;&v9+v`mN3it=`wQCf*%NJdcarD@|IqSV z7OL83T&pgPQir$W&h~)8#fjf56DgLLDSUrooznEW;_F?ZfRcgwV@<^#GdkDw8{1SN z^r%}7rxsX+tdfLb&ubego4GkomE`7c?&Po6wfQ>D9b6uSm&&8XwW1`y7i3)`Ij;IN ze{blKm){=iJCVfh+jlWjia*F`G9l{fCXZf;8A}Dl+ao>9PgG=~dCgw4mthmy1DOQDzFc z()NhTPf{JeycCSMM+=56&BezvgxcmSpCyjb*lYfzZ7H>;EU`5UtA5cQb##-4r zB&IO#<8ygJqwz%cT-FK(a|*DyK?gKM1USF8{PNu_62pfjFg^$yx7Kr^ zxB*2twhg9W0yB=ESGcjS3T(WG<=Hq7$i$Hkr8J}_i2EiL-vD<4=W9^s%DpKlU6;@E z^jA+Bx-e8f+A5p@=;+`3J!sFLQ+z(miKLi=^KIno`F&GaG8yzP6${^BUMdnA`iWs0 zb$X)f7cAC*jUK~^hS!=>44h=TxH4?%qwsAb>5u?6&EtJD3o*YnV;wN8#;p~Nsb>6I z&2WR}QCQ~o4Gc)|(`AecMUnVcqgpkkBm4+d6u<+8064@l@@I}x8wZBHO5*3i0NC4h zEw{NU1FyH(DnBq%xhs923-~sx5zZ}NkNV6nDb9d_*6?LuUciVD56#^5SjJ&!({{UetN+LqB@SF$@??DUhNFm_mttMe ziHv(W@#S!OFc(DA+mGwt)&vsdnO)0z?wz9gT(yZ3#3cMve?Y4B_3cqB?R2c1n=Pen zMDg$&RW5Vf-MCofwI7tf`MJ74#Vj^pCLMw16yZvNPLX*E@pgR!#@a92(XY%!&Q96A zYND>>Yv0Gu?~(m_;snzelJBxHs~O4clkc*Qr|xiVd`)iQoiw_hmbYw}7OEt4J~!xt}*g`dB(|6@p4jiB|+ZDY;paDJ=Lb{>%2X)yFTj~`cLG>`UV&)9XGAn6EBes=M9Z4(fp za5Nv(&+Tdn3y-`=guAYLUrI$u7dz_ik`9;hn4q0AX{>j5~P!xJ7A+;6u?tR8ZI0 z{d&|N*Qn#5o%NQ-^)C|awrn!tp+Nirn(J{p;laR;`}l_%2Y+vg3cme|*(+CwB&dGP;CFXyXXo z^8F_vbZum;Q|?SXE4q&qQz!F(g@J$m->$}W=Y*q-KI1H7hUndtW)?Yy>=K!+WQ$Om zE(mEioapcDtCWffa*yh4P6)Y5DEhM-^V=QAGBDKZ>W<#-<@2@+ioRc^)vPZ8J8Z4h z3)h?)e#E_;b|};9HFPsK3Hj{j)OUvlR*EiiQJ-Y)WDTk4dvREOXt1cf;9wAx&$5^A zE*``C%Ox5SJ%M-D9FzNd_;XQYbgGKD{Al-Br*$r)cQ6Cm6M`^ps_D1=IKN1txmA~A z1q_<04ASmK{Xn#a`G! zy4XvX<1miixa&kxYQUNnAsl+;56i&H9X~*rcM0#QFkR#)U=-&lFgtFLV&I13*xWdB zTGj|FI$b9wuVi~c5V6Nf`w@1!jSDt^^gCvCJhFW5eb@S1nkMnft={wC&}-tDh$~7& zk{0hyEve!%Y_Q+F*O5_puV~A|10AuKgB_=>97iP*#6o^;kC;l4It(cG5!ZMoxEuOr z7)K`c2;<(e7PpLEpEnEIizlaYJw#8nHc4O`*mAYIC7wHAIVPG-f9kG1>W?1^Hrh@B^I{GAJ zRl@f&$zN~tDZqL$jfxpP4FWBOF^6oQc{vhImjUsc7JXjt*~~!Z@!h7lwBiF->@0nVX&HW@jcQ zl{MQKPHvmsE%aFpN`gy8?pdC(m->^rnOV<3KNtbPrMKSH5JGPeuwJh9;)*# z)@F9m(vvyFzEADgm)oQ*d;l>B6ZCk#JSa}AHl5F;cVeZLvBb&bm+phFc`}%0DLfCA zYVd73yJ4ZOln<-CzOJh7mff^=^V7kq)WLOTs95X_jVnk_=Ga=y9LRwgv{+eB+_qHG z(xR&@DK5(zP~FzM=HjwBwE=0X5YX=Y!?keLF@5k>QTMoc1;hxc9%gB^Iv zg32q)X#fHBJXWkmWnaPd@(L1j0!nO>zifEoZCXD6x?6DwWB11)Gg`V-w@~ACY*s(1 zh`j)wEkn(GD`MBGL?mr3ljU`8(pQ_zqy!YXYTUOhC59a^}LTsntHt}~8W*wHRL21ht5{y<|pYN)*vyekpqIoTM zS(kXFMu4`C#8Q~?nO>jFda8c1F>JMf=Zg@0_tX~go5R=&Gm2+JI|{L;x=YJIoQPRt z{^nyDiXQ13Hb+iR-Pk5Ae7LszoIO;BKl!(f1W8O_C~!~{ze{r>t~hofDCeEWP+9g= z)|`iOMC5m?I^|EMn6R`~GnbB_xDL~IK=Kq7d-u1})CL>UpjjLX3o@Dkqd$Va^7177 z!#^!=IJ;vkhZVv+&#Ev7J-c~$Qfjdqzn6ydR%0lyihG8w{B*xJU29)_2l<%qvJ8$a zDo!a#KcY#DmBtuG*KsU?T64u_M@twsHyu3n%j?~`iNbs$^uNI~;t$c6C^1>(D1C0${Xo`zQ-~ zf4-20t9RDbUB60kcx>WVto_@Ou^veiZ*UaoEkol8kU_2vKM00JV2S0Y(x7A$A^|i% z5K5CcXVQnU5yeEX1W^nHBCCeeVgq6#!1VN~6nh1sBr>=x{H`VWC`9W45nKldcK{Lv zAPYlC_iHSo=HO1a4(10Lc6s^t5kVxlEV!mYP^=(n3=-#jASMv4JG}=RNEV^PSI^=t zzX@YA`b6TbO(7>;;8T+S!oNDQxg>V?H5N+=aX>&`Y)}ATwTmbF5B`<~31)fv1=i~g zUJ}CS`n_OU7OGoE;uR7P_y;uEEGsdV*FYeV3|0gwn%oDLC_=v}7-K#be~txpOgND6 zL&5&iF2*4tkmK9m5qi>xs-W34|=G5wb|K^S0dNwUj&FMlJGBM9ys^fynG-dmENuc zGd7$vk^1*oU$~s5UcEcnog|V_y9Id!DJq z8(eET!(QLpHi3?4gO>!N~#&dvkg{SGj*UmeKQbh%ZA%&)T11$4-_! z{iVOzj&54E?Wg%Ki+=L2|L~u`L3g;t^8|nPX06^g9zqtYoSW-IrIZ*!{ut@D8JMhN z&imTKPTd4FHa21AFwAznIs9faRH0|F<$y|YNZrcpP4MI?=#pBvOu%fHaLB7~&St6d zSk}3AqEDYhhx3hTV}M4l!S7+*YEe+chDOUZx1@h=jMqok*=+Y^Si%UzfriS6;uG4| zX>#y>!_5!5WpGMZi>@H7K4I<7+uN0Ht>hWWdC$}b9l`kj%cw14X=JzjE@|L8x5#qU?o zv9)Yrwn9Ok0&rYro~~|wA}&! zWDHC*=3XD@jrbaZI$PFzu_s;OMq;gj9%vX5uv=>pC3TJ_s*IqYD{xx zze_XoZdov(dl0vvx z=skRX=r(+CHagt{sS?;|igE{qWC506+#u##QVV(1c01O=m%d(~pVlkbj+1iW!NpQ+ zv;5n)7-}?sAbN;8&@h={VV?rCNDr>}Mqk8fd^Z-nPNwTzd0|Bx3LaZYAz;q%biFT< z)|O(#8>F?Hv66$Ulom7v9y2K~Q*{ZoBms=tnbZh`_jl?~?LE>fIL*3t8h4Wa zl+(X+9-Ak{Ig7h&o~nACfNDf6poBI0Z_C{*l0=1N4t$d9e5m;e&ELrp`)S;iUuV?& zxOcx56+suW9%r0e)oFqoC&Cp%h~?)F)?21T+KgnDepocZJ#viND-}{5r^iZMVzNWy zcd1vHK0lFm-a{E-S3lN1qN6JE3%%sqO#b9GDUMsOU0+#=lgPM(cL2)d_`$TIo=BTB zCSY*6?cU7@82D6#iJ#nK_MqD@2W5p>kX!;eQHimLA@5ux zujMH3yNmF)Nuy)ZTHjx!Hnw&ynIhqL>Jl%@F7wVoBd4_3uLGY|W*L>>cCSe4DCj0+ zi-d(@*OHk}UA<`)NC+O-K1o0N6!|P?Xat)(C;-k}(`>l#t!Tdq-(0Y+(<=Ux0Jsu> zxDsh|x@GF>`isP$_H+VpOj!Gi^h5@UOtd{$$Wz+r6&g8VoEo}nobaO?mB6?6PS_?Y z^pcj^Nq98MR7j9(yIl_F9EDqwS4yEZzQ;g+vwP|qJ|WK7H|2)tu38y_qaAUPM|{KR zaE@AoebqT(^rmkr(a)&ZGGYBK!KqR^m&P@%fBRqy~Q;|^W ze#u!@`J{CQ1k0q_nX7}~xoic(($KPoAS-%c@>#OsdYKrgB=l_Vk+{wxHlcDu&ip{Y z*t~#fFrmo3ZPomTVT%$xwlBpgG6_F%olKt9s_QwS1(fgweZ8eh(b`->4^*ihzy3m) zKhbQA3t`f`6&q<|Qm8>6WH~q7%brZua%Xro460C1SlhYFFO%{H7Ebt4ZKF>KFvB-`@!H1YC%HY97`=&Tqh)6Fa4#TLBqtsFCB^ zl+YJX2p-fgi^-&pWMSMnZ8S&-L2q>*m>9WcB8%jlqdcHBo)w!>q>I#0d>YV*>-3m} z_-t{7_VmY+b#wEk2^42hf|m`MzeqYH73WGn$B#Ih8E>>Ou6A(>8O*QRHBSCsGvMua zxYrfjH0)N1rcM_1k7%3@i!F4f7y2@~Z6aRT?0aLy>N$ODQ>VM|b{sIyiy>GH3X@m^ zTZ5tt>O@cO+Gu$+)F^By=0~r~M)aL0-9Wq@>T=p7MBR)onTJvm^18?q0|Si3_!333 zk=Z9_A7!a|hTF65Pab&Pb%^>+bLvYlC3g2ZGM0|2w6s&JDp5roLGtyp&NVx>w(L)w z_H|LF$09O){_LzY{lTGm*la6ldl1|6bGN~iV$KIot`l|%#s={seaZNKKdy7zThY_I zXq$)FW3BW?VV}iAjU|Zfrnjg6be4xORcZDXNl=b_qew^?{{=37WXA>E`l^2Bk>)}B zFH%*_vsy)e(`ua+V&d-GEmm^772U0@Uep5o9|8oD?NP-yyVm^cLP@M%agO>aWSkP4 zG;%xS=w+hY^Lco2ZA(GjRY7puoK-amV?6B^%k>{-8tgy6Jb@&e!~`adH~@`wz><3g zDbPj>JTaIlFVfu+YhZr0PL-}Jck5s*xWjh8qdT7 z|1S;SpJ4aV*bwe{4EiUPC7@*hp{NJ;)^uS>slls`jMIf8UN}T*<9>C-^S|a4zPB7D zT?&0kUZLW~L^z!t`r|*r-kK0>K7KU3t9&YLIn&_6T@bSRuF3yTq%k|hD!27tY0YCV zsmEMB2<&trvtv>Xgylfh8m{S1RrQm-^}3brtU(ts8-QZ!h|Ycar+6WMHYatQy~+nB z38H5?&&tW*zexWhQohC7vs8fYvt5U+_9ZK5B{Q27LN z(eGNH_}X$|{GfBt@WkKN+IR#m%nDEf#cAczeI?g-aKdt1{$APN_&g1$TY3X8_&07- z2M7f=MN5b{u?=ax_pVwG?ci+*X(vg}MSalu4N@m<;Fi z1WtZ4LM+U)n1j}51~yFJ%Nv?8L}8c-vzs$cZJz1e;yektHScr?$N%s@PY=5PsxNUm z`B5dPw}*_2yy-_e8_D6#)vRJaiOGqF=Au2-CEqupyfOr%7k$u8uei+QJc`)1k4zC1 zBj}%pv*$0U@o|+y%j|L$#CozYjl9;KmX`QSTI#<|U8y%kaYG4v9x3g7cC?S6vh0>KON= za~NFMs6G7KrjT(?Ibgzw+4FHmW1>J{%Sk$Q;E@EcU8C9`XZB-|(OXQ~NQR4pB9FLr zL1g|!_c@EfYVjz}Q!2!+&q9frct96)H4Kh@Vf5lszy2^`OH?}~xY@>hEX61j6pz=| zQ)Ud;vFjD!2&~Aj~DP5CSLJ$Z;=4n@|cL=nv zPKf;5<}KzN0))gbNETa^bW4Bo5%~^+|dQaPram#Hfi!tT3e^bR*w5}qUn*i zwjvb_d^E~_CfYKSd5Z}S5xlVyX7E{ss{{u?&|&ls;|m(WhW=Nu+a}BeJfc1!O%iT8 zVDh1rjU;Z=^A!xCJs~A$E4RJm+#=E~Gx^8+B8eBfo&TVTv+347ja3I^A#a(XD*>4# zImCCVP8{L-I_FJ2p%SYDEGX!;6Z`agmF#JFhFJ-2COKtKAtgno(+J=~wE_g2Lqdv7ra(IT-`jQvDGGFKCB8bS@n#{T}(3Lz> z%QJNSo)D3qE$uUW4vZ(UVAfhB6mMWI!kj{T}3M`n{O zyg$H+DN$DaH@rX3n_}jIYJj9c*1w1wk~a%l$qNr1vdGIK5`24^evPr4a2`_yM`v%;`79{y(+yoGh|{crU;^q%>jmY+ z{ift^b*rF1cb5wv+}0W^bgJNLd()Rig;D2CZC_S;?U#faBle{HMa4pd@QsvbTAGcI3`NGpU(#5gI6MsX#V&8nOo}0P7PBo-IRw?~Eo8@2T0C!7`f%W z;~V>peF`K&-b0lYKdde18P`u`W^EQ=MkzVaT`exP&Ld{%*e^jKOt+MFrt7m}&Fm0I z?LKr_)9Bq@zPZ!Vfgn?7b6nrWsWupPnA?2*>EHXMJf`u&^W6o5tXVs#AdKMFWb^A{ zWg)`-%)giYk^mQJ$Dw&oT3AX^>T^v5w26Jx_LWJbN5FcqGiQS=!f{dRbFmks$0&|- zR{b<9_?rD_!Dz*6iJ3>t!iQv!mMXZdemC|qQ1J<;=TrP1IrlN~T;a&>_#`ammud1U z1i8^!Q1Br|!Ke)*e0}g0*~3?<;+)=(E?vm|##8vuDs9;WJ-=@pZitU|H9f-C;b_f4 z8{H|$4~OO-EYI63>&fYwl0H1a!8r?)M+I`+rvzD zVhNuk9>`3NuKKx(Clhrh((@uj3|5LNmsP=SmXz1p=3<{;?B8(pDw$94^Yqj3>zvi| zL}!uQV1aJ~+6h!%BFF85A7+ib8atScABBJW_?*eHXR8o;D0o*&eCY+%8i}#>Yfb&S z_YPlx90$Z#Qbc6Sfi@jEFRy|r6@EwsU zXfVI&mitA&3F66~dEY$9&&jJfIJ!F6@uT;X>W>KZ(R}h>kC)DUUxs7!<`>$$ws|8E zm)}fI+exQnU?awRkow`Mu;dgG17*!Ql%EW-MXdhA#oT}ADE)DTpPi9 zqECDZlEyNuU)5VRRgKieSH!G%h`#a$z9qHp%Cq{*S{lXciwX z^>K<~)}4~Rt?iiZU4!-h9EUY@d7t-fZqFFsYOIkE(tSn{+6Cm3Ag+0dnQPR1jR+sCnfDrmtgo`;d;?WN{<3XUj{s2<2og2S8mI^Sj2J_4FLj) z8L+WVLy_Y=Rqt5xXmc(lu~A(Kf+A3Yo58Zx=1EBT-zQ-fXaHH(B*mcDufq{yVh7P9 z-W9T4g53oY1u3oR*vF(%7CC)bj~H&eRb_vek3vs%Ke@vBZ$GDahOb9Y2w&p8GE~cy z(;21W^zG#^_V%wySyoV!ks;tj=XZ;;=hT1o6&S@?Fr83E1;gV_QFnDyCCh3%Rz3(G?7H za|`lb73w~Bu47+B&ru>^(IQrw(ZV=e9{J;Br_eodS@ zTwPqu5F}Fa!Gsr#PUt2nUn|Qk1x3mq6zn;+@lvo|Mt5G(?VO|=R~8UcC1NVbYU>_V zWtBh3S(b48`L!de$rN|VoL|AYFs;Gx)nLHj`s!7m!qzV(oH%%BpJLxn#cL z0=400bUn`S@S((5@ttRus0qSpJ98Th7;i3bYQdUMz)rZ*DYVYUijUWp=x?W+B49i= zP>*i`dbn3D9)5S5`(1ZN8}a~B`=g7)Cm)34L#oH`bXQji_ zE_QO>hQ=~BdI`R0EdE~7Q=&qc&V$Tg&}W>F7*|t%HzjAyQDdnt8q%wvgl=4;X2_85 zz&t*s!YIwq)YZ34mj3-@BHeCLj$ccneCMmKn0Vn`X;#xGAJ1nkGbx_wGOi5XuZ;=- z4K$ZsE^qivN%LX;6T>voIIp#c>+aVq+)LeNRE>E@zIJ2Tc@XHU$8ux?=3*>q*aqgy z3>qD)$r&$J`oPvil{co;G%_#46{N;VSfon^ur*$_N})XK^_NXpWB_j6)6ewy2mS?%Tn}Vv5Tm!_V`s zko{&Pf3z$S(`8@${i~qsAb)&d!+Q)tDo#fE=;l!Y3FS3k!d`=+h(=AmeF8ooEP zu#9*F#hQsuZ#SP9oW(}YyfiqFk($rZ^iqg_d)B}>y$#f|?Vf@+jhk4O= zbzL5f=T>35FaIK?2!xzpYO~hr;4BRdmDwSQ7fPyrdT+}~vo|p#iLJP4=9}h%P(`ve z_A-vXTZAzHgy#+(`{kn4BGSq$uIm)xbMgviIZ0LqCb=w+`c~|QM2}a0jzhM@t?s*e zQiJzYu@SLc#CEWVi?Z?;DR0+{V2pLB;m7xh%V`V|Xze@VfLWCwjf8bk$t&>&k-R#P zEPy@3R}vKXlWN=Dk!Bx9XCw*aprkZc`6_GO-{rZbh9v} zbX>Eo!V~NfvTa-+scK~IApUlVYCcHz&7h3|h{P*oOs!;6Fw)QY!%T{XrXDiIgP46& zOhVfj(~yst8c=)i(S(*);hPk?+vt3`=dMq`^BbDWz{x3aO|pwUE`;Kk*!|+Z|6gn(5p#(^lI4Tc)9;j(XPAmcYliqx@Vv z)@FH2yS;wvahmMQ(rJrSa7j}z4BIzw*wz_j*3A z(WYt{B_vZ?8!0CrPrf~?Rn+vle#NIIxy@B?o$$MOVV^E`{tM6dnu>~RrA0MF&F!L? zw7N5ck>QnTB}3$R<~+{9BLnrXspYA(BC9$@WlpHoD9;3r0 zWwcI_)Y4W*Ui(#KbgpJ|*&IsWFLLEKZ9*x?V}+s+uc)cddF~|d?DAfGJuY0i zce-ybd~nCZV7%4C+VaHQ4?K9SV)1Pka}lm?DdK-Hk2|=mzhvPcd;9YBYX^tDPa7P0 zoOxX@)^X_30J9z(fgPbZud1#BnrWGZ-TwHTq>(TVOJk-k9R90Hr@ey<4Y`Hsl{{X#Z>{>>O zo-s9ibP}=7^_wQZ4g8h^gw{%TKBo=Zv)gp#f~Kw25y?YU8(o&Qt{W;BO($opo}|M3oLHqT+WK%PBT#aZq^add1uXX>U~2!we7Uo>f&{Lk_JZyKd8?W?JDwe+xGte)iL6i z%f3yfw4?h@&MG!NgDwqlI3CqbJvpb?#ynC{Ih>`%<=EH`t5)H?lvG@5s;;dg(Le+j z5G5|(C(H#u7&TnB+L|3dYouruK>9=4+~PcA1IsBhpPPTmUOisTj>TltbN0@~6z%31 z5x*d8s9kItf3|UJ?eyWhV8{VnE)7dHxv^WVGQew}IBAHyH<7vhE2`*g3|Bj>^;g-TEmi)y#&n=}xKUrk6`%1MyBpi#A0wB+)-RV!q<`rIO$1)(dPwrlgM8v~`ai2eDh6lU@G+%6onFBWSeRbnNVNnFS2E zKT+8VjmE>L+V{VfjxY!W7mmJ5gS%8sZxLGU^-#rCEk`G096OBhIq@s1XsZ-8*O2?A zJEUzZ1i{aCTK6BUeajf~W#d=;t>csM+o|;RjuscoO-@(`F?M!SP}(%*OAL@hNi#Fm zrz=$A=Bbw6X|(;i0_iIN?VK66W_$;|a`x(q%Ufc#TPrEsQ7um=WFLaYrhJ!uzv@mo zXL#?}D(nuY?1`Yzo zR@yYnC2UVL5WIayJOWNv5RX&W(Oa1f1l*Fkd^>!#k^Z&5@ZN%xP~VwytflQl3^F!H z5=4(ZN%wgy>ysY#U-G>;{u|_W3p`Y^cGT3&kFROusoE@-Dc<>7$~u<%t>nQ0V0g2r zDk*5`Z8o@so|xg=NYRkiXT$X_WiF?wvo)5hbdbYX`k@n=*KEe0Z<^Yz^dOdSsO}UQ9(+`fX_B|0cu({-6dY7Y_;|=zviqi$M-wB#Z#{nT|=+9!;yLxwj>doVl^v1(OY17Z9F+opDD@iB1bNW^` zuTJUjrQ~fBV~pX!$Duw2avI%A`+QBHqqs*;RoezU2YR{ax2+Ry6*;A20Oo9*`(w>iv|4nm)RK07fl}UMyIruy zwQBr-)N@eTV6ob1g!L82mQGe@mOmVQ>x@)tLu9kVZ5IX!x!u~v*I+PbN4YL(h%zM`J4N>*B`BByC{MKt{HKXw-e)b*5A zcVkr)u|2$?0#A4!6^Q&QxY=p>1>)yT0dWl!6#^%p>3Oez9vN}TmcPw$%FE&C-4SVq zqMli9^s~pxa15+1Egy2t+%0-$g2P*Vsez)3k|#Vp+XtTpk2U0*9=O&ul%+OCAdDX!DZ_8KB%v=#bCJI>S`cszZ7IVg!@&))8*J9J+>Z-(fvA4^bhTVIc?7jn*T3;l zHe(TEq{pD23zdW+TnIzF3!hXFGhLHG3u^(&iYz>@g$Q;Fn2I{&V8Q6a?82f za`ZQ1?~%GCqUin1v@>I&&6sAB`Vo&AeTZAlyFKYH2pX)?!CRz}Snp?2Z=^OCd4dU0#os;@eNq;aX-wWHpnE9d=zwRPgBCo#BvlaQE~fYB&0}y7yS_VtePoX7mAUl@%3dJgjdcedTk< z8Oh6}%jx6Wa;uhFN|~UOEp%=?z<7^8a>qYVz{YdAvSY*D=C0?bs=rw!YuwXvrD4I& zJ7~XXT6mznxU@@6Mq0-54)=ofOs?#XI+9A*pQ)h?1&-=7xloT%#1GOmW6U{W4VJ0w z5jqNHHQP9}_$!LLTiZ}uOG^QSU<~zBS^S+*ni`gimpsKnT}@8FISFw-)rN}JA5B!* zY2gxa%q%8NkT}V5=?9^rs41 zNF5LWG`p`wGMN%fEgiIUscs?m6e~NB=GVj$eH1|Mqi#0?9_n6=ozRO(h?4z2V z0M2IfICaK%S2cF>DksFo~%t+YokMLEj2qzmsvDW~Qb)KO^r`ok>Z{v$qQcS4Qzt@eLFm znRj%bB^fQ^YkY*$K-!S-S%b1pvvHPoriYsWnH@8gGc_$K7=A%m811r2RREfnoskFS z{D*QCT8b$I?`$-b<}iDt`78B&oxT~dsVMv$RU@h52bHbQj!9UPZtX4aCY}b=wm@5s zgef4kzDmk-Q%eJ3=kM~J@UwdNsBa&Jb}dOvN751#@Q#ple*Q}dH57t(2_tph>BQETE9+H#;a zCd&6V6H^>{atktdnsyVrmUvRTO%A=>@71{*VfhC9a~vlQB_$JfbAuaml5T< z`P|%I-zCd^tG+o*vD4FZ+H6uZi=D>O#}8C?!)35qELMph9MVlgC=Ks&Ns#5`YecsV zO+noBikiUwQcFo|2S7;+GRxr8)6+pEVH~Hks3F8VW4Uj!c=`ILiq<$|;pvzxb{Xp@ zES8c?I{}S;OI6^oS>lPYdNns}LG_O7gz)E7$!*p@ILz4UC~8S-q~^vLd~xQwzYsMh zm#ov%8f&x$t&Qe`vVI?*5$+=^g`B^gw{gOF(_*&R?7|TrEmmWO zneq>+-PaJR)qgd{O&%?8g&$VklY^F22IPpv7I`u7@!%1+r&nP+BahuKZyG}laDrt>9EZGu8o!W7e?+WiiYFmph6t%IjlSee&oX;KLTkSpz z7u01h)wtZ@mUfP_HHyv3V{IifBrPQ5@3M^@j>Fk{L~v5dF^*tz+^i=OFB*B)x#*?3 zGLqsO%^xGM{ro=upL*oZ)U_0o(?@QR=7^qIm|xy!o;*6P`sB|Y<1KU68`fQ&%8t{g zXM33pG_pOuuxqpNR4z6<2zZ1=NfX`)%bkO_lH@ukQ&rGvdU@f3GaQmRwKyX#XMy+Z zy5_dFP+#=a?i|dhhCb3Kv({s=kPon_yrt>Oa<5F*@mx^ZY%Zb6j+Rl6SZ=ND)=LzV zNMeGDenINB*~h=haa*10*&($a_Y~`ZH74T5fdr2wNQRRLWzh za2(vs@X7EYbjlW%K_4Qt@l>TQu#R^yTwX(H5Qg z+Wf0qW}LCUuA$71hM}!?;gV#x-{iRE((8HCkD!WLkrgD}*Yin^Q{l3$p*2lq1#51U zW+`Kqqms0=Z1!iz%igh)+eR*f@e^f?+}p?SwJm5=62WrD|2$B3!uW|4;vjyLC&d6mX47kkE?ijDJ5+G}LM zvO^<)EkAm=(VDJX9aMLFafYrsnt=^wIyr}-&w*V1%P#WXzjm(_u9;t?Xe9K4(M3+! z&U4yy9@UMLO}*5P6ppoaO&s01oM8Tix$3J$vr=ibFVkH`lsDOv)Oj2X7rT{O@jjo8+@PcTZM{ zs}Aj_(fOKV5wmWwxNPs}sw|FrEXz;y zkb{>n?pm){(bh8aFIdyE3DV8MPFAdTwK`QI1I*nxgILsil$db9X-RbjEQ8`y@5iiFZ%jn}o0o~^B6XJ&jFeAg?Czm~OU;p_NqnqIQr zT@kgh%xP%ko0t>lkfn;zr>dwUrKcA8YS|eLmY3ka`6jz{2ThFoGljCg$ERkshr>&6mCbJM5^P7BwV|leJW|k9^|jWvvKimX1grt^%j`b& z%4<#MOV(FXMO5KwrIVIIyy6H=W%8m-2dsQ*Rp5ubhdJ*Qank!XQQCGgfQNttvhwHM=Cc)4dRA%|R8lsGeS;xJ{W0U z+84jF=E&+u>K44p%_~b$nz2a$l1w%^9m;ug;5=6_v+6Ua-Se$SqSVwq-0X?iEo(dr zhqZLN-D|?#U8fo~9CTFGEgq>2=Z)XZa9!}3S#6g80K1-3Rj}1RM>>twbo5m6v>s4c zcE_5MTHOIt*+O80=-Zq|H_wq=R$WI=XSqv3YQIH2u(vS=oM8B6T4`--akN)UTM(qJ zm=b4U?}rTdt~q1PgtsYV{oQexW!gI3ZRqxGpqKWN! z2naqmj69_F-hJ{{RHdJ1S$KW2xqUH#=kXLdn~oqI?d zrd=b%-6s2;*H5;UumW8v02s*XT1Me&XQ^ekU%Fg%;wPf5(9MoI)>_->&pXN4o=~xu zJI05u=kyzAr95XXjhz7TD@eI~`={o8H5GenEb>NQPTwKmO4sxzxx8C0=2|01Q7oDC zs16Tsex=77l<#-1-L8r7-CJLwDDF&cYZ})aI5o{Va3iMcH#FKz2E19%p`ThtGc6wV zm$6rAE+<$gs<~K&Bk4iyq%93}WO#=U1+jZih|-QZBxYOQ)T;rJpGHOCnd_s zg}?7v#~+y3>nG5)Mo~dVW|x;dj9~--I-e!Z^)2$rOH~~`9V`(&{{U8UTI2T+nk!q~ zXfFU;-$zwdAw9fK11@Ou%Ir3YbaxW$hFiv+h8by?T=`tU40!58hU>o{s+>1>{;cQ6 zZ0zAinrFF{k}yY3=VY`UkGW&&q@|9k8d+*%6V1$6)*Xl*2imqCBI%ojqM|yDKYA5* znHuLs%yQJV-0u#q0CO>ijs~34F7hz0PL1IA?&)Y>d3++}zmnA9BRhUoDi*7BzwZCSz(qZ7tp^g z>SK5JENGqaNMW6hjs5|{8C){LF`@ju|+dTk)U~y6|3l{ zfQQ8|Jb2$7Ld}+W?&X&om0h&;%Go1};W9I?J2wO@P(I!Kmtbu`chN}RYsVp4g)A9BMH>?Q1YnfD@`w5)Bq7*r6ypb{f`|jB ziYlHkqm8qb92ab&K@HsnKB zbA_aAP71R3TH7hFMv^YhmWBv`9a=Y49jPiD0CW)(;!ae)jS%{QSl?ldludI02c4rU z)Adiwms29$MI%NXNfW%r>n{$ruceAH6)iMCj)Az@U|mlTslaQ06`Uw)1~rqw#j6L zxzD)IiwWX0<6PWaaA{eeXu_>euIr(Af1twRJPu01R@y73rnbJGnnQ_mE;cw>H`!|5 z3xy+yJM20w``n&7nfCFYhE(xN^4^qC7LJ3LV5`&dKJ?~?b`;~pWG!0ZYke&=ovT9} zBPRs!m5H@C7Rr8NT;d5N_3EDYCiQB`89Rk!wbk=hPB8gh%y4$fpbmBiI5}=qTiJND zpS?{ykws4F8y|bXtoOcCOopD;H!UY}j;7xnZ)O6mIVx*?+Dw)YE zh$k)b-DtB%>2(d}IKWo7sy9bU*TnJ&bl?XjoOCYH1!lVx>)9( zF3MSONpSnh#?8eF`l=_>R9Yz{4m-Kc-vjFVmWg1Z?QN5}&&dyV30Zq}v9F|z;iQ(P zKA#KUNqI7D6ueWrQ!qshvWh5=)s8aX)Ux)zHs7t)uCArGQ?{x|`zOeGiF16qtui}A zK<+$!A-~j20+rfW$E#_>c)NCzFRF%6fQL)q16_OX9xd5NN@lu7YasF+hsp!^~!q~%zP~6%H3_g*!bIfwp;~$J-&~( z&W1z0K)47Kw?F`0A^D0U{CRfr!9CVgC zI3r-d3N60j|mGw8k6S@b*ywcqB{*r>%ouSVMP#j&)}_ot4As*-tQy*?SK7@l_o941Wca2l>lW8I06ts^01pH>$fO!k*~|epnV+jC5@Mi{}9$_6o7UBjfdiXmlcAz>6|4-!X-T)LLr_rS+GLP=;O=h&`WZ!r-edB7fPe#RDfNy}S(?}_zSddcLh zp{M4V?sSh}!6b|ojJ4`I8YrLhWv&?BIsoTqEYuI+aqHy`y;nBk$_DVXSi0 zj=5WL+i;Umo^H|BGb5aGOXVcz&#JeRQA^qaEYE2jF)9744IQ#h&%F0}?p38l8=F-F zTJyWWJFiArW%+QDGfkM`4aOH*E87_m#2Q_>FnHTB`jX*BfqBt?d~e ze;*f)Kb#jGf`#s@sA+q$){hlOMax%BSv6!cJWowgnDXP{{mWf>_L8$Pj-MYDZ^fK~ z$iU;dJm;syPB^E1z-h6!Zw`LHrEB_%!|CxysI8H>Q&k_Qh-7v$_>XbULe(qmw?|%R zgi+LXL>_Nq)#T4!PMF2Q7b0Eiuy1)1Fc9c~$DI(^B{3=59O* z-CQ=J(%3^^Zm4vhS;AZbvz57bw3AG8YNlc45=eP>^|t+MW73+&OGRM0(`lF`sXb+p z+2;1FT(x#C$=J12b8_P*euv_<=2E&xx$S$J&SQr26NSSpcHFMhQ^!qPHL?qq`1k?QKo3$Id37G>+<6ra=iO%rFx2}+DV>LJ4D=S)U`5p(&>6h zXXFg+-0uGXDCxN?Ri@TcL#V!c{1$=LZ9Mre+dmbh3v8z!NbXCmnP1(J!yLt2K7v#HV}bsW%D%=Kd5P|2^4w}Qy%jc~j_o~~kk>T!hn2?-B=54` zskr-cmdjCTii(B4lt9ljgTV3!E6>;-{8t->GV4yV+a2C+GU_7Y z>9ou>61BabYWL)pk8W1EQLSofZxLM|q#*5~cRK--Y@WH@bBO6>d)(nGuugqqrmCZr zw2yae;CWBhD|3qmJH2&u@Sg5c`ENH3T~}3eis?3#hqsC4 z?B$RFpFXQ~xAE6hQ(G$N#+j!kp`5*-mv#?k>x4;2_EytffW4-21 zS32}v`_DK&(y3kT7b&eJ_UNjmJmb_SrdFq?*(mm#4){Wt`5K zlFsfk)o}{Dol~ZgNhg>&&LzW~oaFn}$g|hBDEX=CdBb`Ta-kFy*7|oo^JuDgal}s& zau0G&EOEV=x!dKA6}1`9YcBZu%c^jTj5qxuVyd@W>~UO}@<(dm4UCSx=JV>gL#rfd1j_CGs^;HT?$z#e&N=JeSi{rt%iXNG zGOqchIM=Ey=1|=!9K6wn4rwEf(q<7TdG(evrZChUuhVb9MlPEU`uat};#XeLvP z{iR6Nz18gcLyQsCZu(p*d%0=7UpRwwx!g@%R+6S8NMS zU7vbiA3nWzTy{ywJ?E46t6I9JhOxb?_i)Z0ujzQ?@UxdKI*o4_H7je)RF+8OJkBj+ zHI7Ad<;FnHkE)RKnR@OqRNr$TsF{ zx~H*@H*r18tkq!TnZ%y^jt_xW)e^9|vCRpbIOGQ&iz8Rq!t&Qr$r&R!bGKNUNaeX) zVWw*iTu%N&D>u(i%3|7dj8VmIns!nOSoyLBbIgAAnu2o4TQP1)-AedNA%=n&h+}Zm zfz+;{Y0yGjd=eU2k33-Wc4uYm*GC?wYUlT4!AwXEIbpOFXhUyE(BPmcK(|ud2U^t0^XoH07*w4qSbUTwHgLkGGn(%=6vx zKkAJoD7D#}OBe8o$Op`m)T`4Ad8w_jj_Z5U`+Sr`3$Bw?%*Qf1Uf^rS2tIgFjqYtU z)qlp$NMJr*3*9l8%})w$vfo4SC`d)#K)%^>xXXcTT5F^*v4E*_bS5S zP4j8RD98_aNXg)?sxyQL}n+_s8M z4SKb*=S>|H)emswjp2mlr=j*N19)|+Lw2^+9D54RT~}6qSM%X)ZxgZ128sRthrvF0 zsFGn5D5ZOj`*|$=<3?&J=&kfnTIz93%)Q}b9=tx26I2S(@gLM0<;l8X9wzFu7E;mO7QqAnpmm%Gq=-q@%i0)KpvQOmV-w3rPx{ z(?M#Qo1H~ObQL_0ke1=6qJ3I77{^?yn;DzXo?6p_;Cg$|-KA&)(O&HM0<-q*8K|kO zR~kv}wN7Eh*Ybe-0u%d1A_ zyfw#Nr(-$I8~5lk6z1tv4j)AJJM6n+uA|Tzswc%WtR5!l;nW3 zx|>F7JC#*ERXwtHH?eD6IY{3M8dpW?dv(#$)Z1zdFNm~)!g_?&>e0WaZG7G*5t)4v z`0(8$yw2wH`WyOZuY$T1YA0LT_hS+0A3fo9M8ANMMZwB32OvdRWOBBjIowJ-KY zj!2x)@bKpK%GB@HDt6r^c5o!`fVQZuH;%;O-BdT&2#U(w^!>u3C?)y2eqhiJVmVpq zwraPJcB!Ii+eFErYs<3kV63&IkXx@D&&ydeL%Mja#*a_jp{Xxz1whY~Wn?aTW?K4a z-&1OqXE5MqBb+VFcaDxnBNuusel+|QM%}1z4`_chu8;T?5dQ!nJ0IwZow_QV{{WDk z{gGIlIsCbG{{Uw?ZQ_R*=Xd6nRhOC|J$HU-T@jiho5sZd0PIy(=z4v>z)t?fJ*@s( z+ndRHr7S#;c7LK)Lx**py~4By7X#+H1bQByAL2Lv0A+=JZ_$R0&V41GSlr=)TNzoH zXD;n)8>_XW;EuB1_z3leLGnHYTT;FC7SR6y`OPECPI~S8R@EJb;}`^6>M{O|tToDp zYensn)Ij*-s%{(}-d^{b_T!tLURG7k5z9|@h~F4e$uIX)isor(jUR7Y6E$TXES=uJw`m1b6Mzj*zJ&3MME?p?;Or^NC(Aey1s?i zYXPZ`B6?Plw%H5GFy0;fx@C8F8q2Q}$us3_vO~`D;Tv9XJ&JbSs!O)0n8aT~S~}%BPu6Wo{i}?ZF6)j%!U z#`ic$B-rK8QMp_F9~HGlHL4m|bX^=Y>>#(1z#RNnEqo=-Y<9a1eZ7Ns31lwjKE*=h z%KBOo`iSZyZa@z?11ndD48G`-Jm-gTq~m_9qqS7f$)#C2sU*5dC&}OMTCMMhH+#iX zXRXoDJjQ!~KW6Q|H(cI2sR0Da{r>=*RmiF(m~;FOikSL-J?*Q>m*LbXxVfllZP8n) z^mQz7#e&|<YmlwWV^5kP-mBeVOh^1id4@1!VRpd2~ z2WcITdXv*}O_!4@-E|JUyj|* z?%nwS8UVQ;FMq-n_NA2cE?NiIVJPT1XR+LwrTZ9@82*XdYX z`eGc%HeJcia(7&I!3?t5<(2gk7Y92Bn5+6qC%Tq5nIF4G{O5IW;^XnAZh3sV^}mX@ zt5qz}L!_yLCEeM7L*_SEmaaVN>N|_82A+y?VC)2Mle!!3&sE1;Ku$9w@%4q)m%!u>Ne9d<)q+or<1gdwoGrF@4rQm)4Wc(+o-DJ zcAmG=Q@9sOWq)6G>yG*ir2aM8&OLIWH8bF+l)hIuk$^d}r?tneSiO0CyL~Up7VB-e zM^_y!EkNvyLNM}L8{y)$?N7wX^>w|GR#<5(BWwAv$vIxrmB%7}2_?jDg?C9C(P<#o4Y)BH`l+34R#N2Mv71CKW$BV*+ zW!V`|A#aA+@LXM#5(c~J9n~YK5IQzMH+obCrCtbBY zxz_8Zn5T3GQy>g}3rRlo)0|4W*r{DjL8d9pFC&?+c5{y&y;nPu*nA%EiBi7W?m~N4 z3=De7eLfo1i#c-nc1U>bttqaQ*9KXb1v@(~%t_rX_hn)Ae-y5|u7XNCrjw?8Zg)OI z_ZjPr_FRg(2eVAeBW8bU0XYsnYBu^=UvG*>PU7b?ea`zSI5_dTk$UpKmqW5~65(W~ z6ttRVnen%A+7RRV)b|b~-RoZN*H6U2{=BkmS3PAD!P--gddFGmWs>Dy$IRjw@EQw_ z%6nL2mg4o@Yi*+Ys48j?qx{}zd53$yoUA#hnBx*lxATgrOxjvXqAMxnsbSv&LJ0fS z=cHJRd$zikyU)@;-n!||-^$X8vZe~iYFXMyVt9Wkal^fZY?12vxa%RS8g+(Po(o8R z$ICmc1;$!Q3{0V>mJ&uBUjA&WDQs6SM3vFQFeRk72gN#N?zrV-?CA7-PrSe$Sz>E< z$!}?}{VGQBR{l$Xva^)#ThFrRHme=X+oJ8^sO>i=C&gU8-8@7#K={sf2OzECbCb1Z z`R?dW;+_4>4Edp(bnox_)ROTo^Dir;vBq7{yPtFKT)Nu>-G8c%NnOeq2pQ*zZvM^JGIgkPWETQ`s#jXx53Hs1oOiv8?`0h#r;ZwyrZzIV$4OM!k78jU+tDVYn;nX8yq5NcjKMoBfX^lrJKWs z8j>vK%K2t5dg`vaO3CE1K?K!&t$d9=8afZuwXW7lT<7KP$t3QktDy8^*FO?7t}}&R zDX4h{DBdzR@2>L+h17DnT)eG0j(6w5K4)c{vTG`<6WD0&(HWQrG=yZo9Wp#sgs-L$ zJjDZ=ey@32Hr{HFY9Gp7)Ojssgvx5e9IiKVx?L9^HCz-k(FunBe9$ z@(|`*n7ut3hOOTALgumb+)C}_py*Wv;%A^hK!F~*O_75aD2yt<;#`A_H@GBqhYAkbw`!Pnimu5kkv`rqPNC2 zd^T9_P-rQOhCAg_uZe5mp!Qr$3?>m_}|4`fCtykwVPt50X9j!s8cb2xP^Zp%3&U3FAj&C?FWiWJ&XtZ0k7 z7x&`s-WDtF?q1y8DNb;Am*BxQxLbhW+Ar_#`y(egIh)zt*}1thb7!7MM}x6qAmNws zJjt5@^@cY2EoU%V@aqx#%Eo?%9eT5`Wd=Vd3nhB}1F&5$23Tkuw=+1};2(=bX3Bo#hyHiqTLXJupOwi&K)AEqM$6p^)R!w6zEO<$v z=YUjpj!?SCcC=95)lm|!IYLIO>Qi+Z@9cYCUQen%bk8#Fb7_j?w8OI6bbe_0Y;D%P z2Y_n+w7yQ(ii6U94~P`YTHoZwv6G#iN=|Ys=XJQv#GIT1$wjWI>zW$IHd7&>bQuN5 zzB{*C(PRZ+gfZP#L^*w50FX{Mz0Q|rW|ObZq$$;o!jqQ{jdDTv5kK6Q;24wM(rhBHr#?=gFb)!4nQmIONU+NgZ+JXh&nAvmvGrrU9sYS3mvCR@E$Pt$c z!E~EP$tf{y;gdHIp*_XQ>v^n@%D4u0k!@-11tFCBOaHj&^dKHz^acIay1^i`$(*uOnU4 zlTEd>SxtZ|Cu2(K3U?r>A-2GqDT^n?F%{;lnJ`A^)jY|E`*!a)wjUzVo?=@)fr{?Y z;ZZt9IbslB+`PWHBvGu~h#-GNu7Y*f|ZH<;{Xz7|n zpSp-I5%+P7)G%PHjF9(x-sd9A#5a>Cj!Q`_5r~>{n#yCUM&3*570{bSGvY|(;q7gR zv*M%@5K^gs3zGY7(>H@+pn2B8`^mteA$w3hH?7fEDkyLrSpL;bqion&(vhxG%)|7d z7t}!9ayri-|7{dc7Npra2*VFrzQ>I`tqqFBRp+WCpuA#)K;cAza_?SlSA#3cN>k)C z6pfQB`5U{iNx!?OSI2a$z+u(y&^E|eERg;_$q=qpp{{N10Ril(2@&%F39d} z153RNs(SypP%6br<2lTm=DVo{Q_?4>2R`6+M(ixjb+j`faB0Z&?jypX*Vme<1rb4u zi$Y*|^g)u`xBz;6YU1W$*t~@&eMG8Y>Vjjt&-P*E_Dq9y>eqYF=2;lL;z)%63I9gp z=y=P}{7e_X+cz;Lc&?!I*@Vv8nTa4_mjG4Om|i#104V(vhxvh-0BkxfX%)gzhcr$| zv}a6NT7pfJod@5>#dsgbh4P9HhOTYFmzJKWzs;+V9);dbtd?-YZD5Hxk~C_o*25&; z(AG-S1Cwt-itk5U8k2f{Mn)$zB<3b_jag`~^3jLq z9Z3!182#G(TxXe39Jr&(!*nFqvQB!{s1stKj8VmY*NmE(Uq8p>@-U^g0LlL3(9#J8 z@f__o-fF3uU0xcmmvCK=#~1UZn4uP=F!LI(Yqm3EB{cIce~w|q$=>2Gx3r@Hu9*9lvRhMusq^V`W$M5IvOf@^Wt;MOOE=+b7mPJOmB+d{BADFXm`y zf`NY7gpGnCRuuKqGjpLj{2D83NLk?0-)te}BN@&T*>H!zadC}s@f6osO;nmdpBdA^ zs*be}Px8niKOV|hX6W{phQSxK7q?d5p(T8u>7=Q@PV^X?&hFs|JOAnd^Sp zL4TQgp%3Y%eD6V9d{xPFIom0N3Zotvhu+K~r2=u*A=#sZ*5d4?QL8AN@^JDXu+HhL zi(JhuEIG+qEozraSoxTR@X34^xO zE1uK+-dX@ZgO*@LTX%ZCJwae$54n%ei=lX*&{}SBq(8><*dCNHV(lCOPEvrOhl;II zVYKg6^76li5Ow3RU{w@!j)+Peu{G4u`|39{UkW1+Hk|(M+Fr&kg|weD7=3JNg5JFm zBuz4YTxUovxxTmEtD0X-70RSwD23ucpjaG^w>Jrdl)E=lF}UUW%eK+sc?3tHq4MRl zK*yELicC&7?Mlm&cjZb7245a65Xik`?+j#0a=hC~a-*8*Wu*_dn^)IIUBh~boG%aS z7&ceGZECDYO zhC^~yeuHzuCATcRk!!#GKBGcP@K?rnW>9;T?1CO;l09i1Vosj3B_92ZVxfaBG|vVS zcR^41*{|M}h!#UWBUfkreTElWtDE;fDssYJma4kG9h1#I;oKRu{R2=>`{pinoSvNZ z4>??@JlwX!b~dw#$!1)%@`YWA%HRW3UUQcUo@yYE&#Mv_!nGcY2P@}e*nzR&J8Z(x zUTXMQH&61*h@FeH-7g*M(a>c7K+GFLySJZD0*I>D|bDJbs0bIU!A# zLR(bMGZGRq#0lI2nOg%sqke0i$BVpGP1AvVfpFD$n%=fnW1EU#*>3Lq)rmN%RPd=W zFy9!V{Ka0y{8Rm2!0Xe|*4m@0?pZeSnN{p4Wr^$<&MndmfSKUUQNin@|Z#Lndkk_S?}r2F8pLVvacOn z4;J^21>iF)t+^gNXaDyjdllER$C)UvBof!s$ywhJpDyepZiB14AFXg0bXB;@LnE7+ zyT8rAEkPgqz_S`ns#l)*M*G?PKdIhCRWSEXCqbr$P^fUNu*+&5{72Kz@S$V}mer%N zfdTCrCO)+H3_SOKg==nq2UpWiYzywmmh#xWu54(IULu3M&omxXE<|0*1*tb6*~G##b~K&>`mOPu zZL#OzQ)OaoGz_NDX=Tx5U{Lds1q`JBPh5Dp}lv)4smF2(rgCd*Z$5AGP zddvLos!|26`%e-x6_~F5|1&8&8n>6lDr39MQuuj&*}fUSZdQS7qD6QTF6{;Fs!~z< ztXu4V>pWYR!ne)&b}Zm?OxAWbPEtANXAf7ZUp9THjHA!!*CVwX(-*X}i+=!_HhA4w z+s*95x5Vq$ftN*Q_*wZOKH;O{%ho^J9jXx(DHDxyK>xc^6n3E{kPLOW0R==h%tp6!Jl z)ATJnOkdEi?Uk-GML84t_FlJ#EkXdM%8K z3C0N5YT4l)hY>c-ygIx@hO_6}#9BP!4ycYBbQH<-4_>ra%}mm(w>#MsJ?}i>1r6B9 zP*)G9iu*diUMz z{o$R2syX%BZlAxmWYN`RP*1W45e>VUbF{+IqwH5KhePy?%i<&uVT(EXn;sP}k98K6 zm$Sn%oaan#ofd8?*S5Z{L-?6$F(>#xa>~{WU1mL988UZ6RqN|$810OdI)CV%6g7%j z*tmt81e1h~`H2l^41qw`X-Myqi#RroYmRie*F5T)O-|P~yVpp?4SuSgHIL1EMJ01* ziWF406Z}2cx5`d6!%+ECm7GnllX#{TuQmHfXI^UMe}U{Ys`Fr1Ps?~)v##c6EA=9y z@{qBZ_53=%Ll!Qm$xJ_LzqV_uA*!V)DPIYP4stXT#&ni6!a%N3`$Hm1#!9sPF`o&q ziB!!mmghNK+x^67(#tTa%(N_Ea-Rc}rZuXiPAiQ@n z35IuThdwe}Ti}M!t-%N5B6L5;cWsO#%9i}hD@hU%Fl}Xc6D+;$_=Vq8E8@?h>`?6@>96V7 z!Q|)|bJ5{AvQ`BZ^+kjY0^O`u1-Rpy{;oQ&6=&x@S$4~=Q9UiZ4d%$7i9!`&BaZUi zL-$jK5Q{05N^dN;cbc%*qOmn#dTr&bP9$7gpXu9?5u?p&T<&9j<=={sXs|hA?Pl4C z3Q;UaA-I0TteWd6Qp+L(kCQB?KBrrJvkuRSs}4PGXk8cTW|O8ks7JU%_LDqBv`%tL zpDfgjyo`NR2^nOQQx~`@3zcCOJcVMKFpx<#fpjRlc}%O{5lMREam{k_NPRo*jqZXLSU zyfX8{n8w#=%5*wYDl!GTMVZy{+a@(M0X?<1T!x(_r-bp`)Fss2UG*zXQV7xZPoO#{ zV_tj|9CD)OWxpk+D{o!R=vM;!ZS%{4X$HZa^#MI$cRnX1#N$>?u4o#VXT#?KPNKuK3k~~dsM+|7`AW)64Y8-oUYF4`!t6P*1p&}>ZYO<4NR?& zgnm2SlnVXrrWwT%NR;%wF0m+Yw2q% zPME;h^{8?3(|^wu2$M?-I?>kpH2(+ zZ?Fexs6zTQJo1`4QZs|~AG2vr$=zC)Q*RgH9{XwQQ`KGSND9 z3ijRYmN6ORb>T;(n#M4%b^1iAmlaU8n)jkZ-z-WZ5MJrtF$h7dw~wsdLDsjq zhkF_^#mh8qGsEC3yrNO@x*68;9(#|-q>g2Y+a1$C5lLWy%5olGHi)L?36mhE>L8U$ z{EW0dgrmwE7hTUk>{q+cVB>Oe!QbX=3H|-n<=wpSH8PgS&`iC@MdDWW%IiVBV0Bj_ zxihv&BF9R=f;8&(@@9|J(H)avilx~a-ARZYgYlT093PO&4PRYdMuo8hCJVc3T~lfl z;~TDRA4Z>)ABi$$b{(b;L(mx8_i(I0&PYxZ%8~{xso+k1@PB?N{XDGim0*o~=fSeY z$0qyatb}C__XUTmEGjQ%dPk1}mRwJjRb2Ym*RSfF<^?Ck^JxkbEVZ+h zjwHs7YaqBTBvvB;_(Qu`jZjm4KzBpVx7Nu?^vqehfRP+_0k5&_H-zFpwSUdz3(qs=V_f-iJ?#9gPjHzI= zS{aC)*WRo)HT*?l(2Wq>*QaKE3@_-Ue%$N7Z%lGE_ssQ1cI6@)GogX1`%z|ZGFBJ?iAxD+oa#b zg$19P?_-f{P$l82zj9!f^3`Q|)-?;E@8!7CNb>6mEqLgK>*ZwQ2A#$ntd#jXIrmL` zUh+Se$1ReO*MSNa-X~LgY%omY**1KHM5Il612k{&=&~jXEuK<-5FTX)9wCtGXt4kb zF=-meaH~2J^KD8MjN%)W$1Mr3O#&ev6-BfOJW)0tJ-6kfWfuBV`>iGP147^M_j*u0 zC>yQAbB++emOFRl9uQqfOJ~mmu*!v%oUrYzt%NOnoXKkC)Pu=w{dTgrp)tEw0h`vB zTNA4g>6M?%ag;3{R%U?17d)opR(Q5$PKaLsB21lkl}#v$ACVqLYHt#G-ZFTrt6$_z z;vuQt##u`Li+t}N;ICR~3L4!8fmeV%NW${agiO7}jC+~>#{5XPX`zoBhzNz38cffW z2s?IuP*_BFW7{FL&`#TKtFX(H+EIDLnWFdyC=579f0!nD0!HR1_E={OdgZNBCiH`m zPQCoWbj!y3f$pS40*X;jQW~{0Zbr7gdrlMswJ>$Iy84mPCwEiAxG8+laq#+Eo;R zMC1>MexJUJz{KZ2Abj}lB8sGjO#kL<#M`f5(c#Uc^y)|oiT`2vhkQWF3)&3z#{Cd* zu238NCOjYn8xsi=pO}{SJ5H9f3dICc0VzExub3oXSwT3gyFR?iD>g_hb0dHb5Zx?7 ze9Bhol)DMg8D$dW2>NM2@hi5P_9kLM_6+cd;G&=i^L!Ak~@R6kgNoPJJg8 zicjhQdpBk*CDHw2D}5ze3uc(USF;i>cQd_aEtDoXhUb=?ciDmqH)H2E0!Ym>^L(T%fxG`OfE*igj^oDq2!Nl5l-DT#v_3;p*%@8Y2B zDa;?ebVCcNcyNx{{FjyVveG1P?X>dm9)zv!Vx$9Ld;TWzyGgMaNsf^3{2*RMqfE~7 zi8|I@fCf1fD&Sabq1IaT%dIcQo3+fzE}7KkR?DJDe)GC7GKW_LV5*8m1H^|z^clH6 z4rC`S+j3JlwH3}L9FUPA5rDt1;d<0aV=WV(UCh=M%ZR(S`&iK=RKbe@pnjah)jub>(uC95}<5Z(jr zES$2~y1tEO2>dat@_fS2waxh|5R_>fVf0zjZM1Az`pa1)F~ubBWnK2v@KVOz_x;8- zG$LG+64RRBSn6(IeTGzV25GGD37TRl4NCZbyCZK%6^sJSRJ++#^paj#-&?~qDiG}{ zzQUY{5g0~EGJRXZUav zoDI%~M!$siAQ3W545ge1;J-Z@$E}k@HV2>x?))0udZ@2Uhv9roM9*AvqJj2^ProQCL{@TQtP7Om+)bE%yy|~g`18!}le>OK;*Z1`5 zDc@(O!AV!c;+h53c|~6L)N5h_M5kN)?Ukjc6TB~?hgSu0C2MvF`yAa34W{M!;mnf- z(J-vzyjXrid5T*a2brGuENIcjpcb6e-PkQi%~{3OLlhaq3)m91xoiD**eQzouvjk+ zJ4~5(WCOkWYe1AuLS}wqY&y8p__zQ?)m2Aq(hgbDoPFCqWjgZx+0?U!uK9z>s;T>z z9tZ^mpdnyxi$7e_6DJw9xjmN{cpfV>NKrtO-#Kf^zv(Zj-a>9Suf?|0??Pi1sPgI0 zvht-`3cuhdUW47L6}8JzIKlH?6-=1|ok}llb>et8nVWXW@kTADED-j2z3{Y=|5l-V zmHeYp&W{s6-7e?lmf0R`DJNq3Ko+s0k(=7wPUF%#+YbU+38&R2UF#f#`->PiwoRVl z9^S4b>NlxxtoI4-hKI&*JmCdYV_Cc>3`Igq#O#M7Iknaq#}{j0R?EOvQYjoM}7 zZ_xr&aj5dQ{sJ&hZ`^6CGMo4rZoy8`b7EWXsRoU;OaRjCOml~nn1(dj%KXSr@<&=x zYhQ?a^$(uw2`LOePM{g*^+HPz({B=H_|3@DTehu9cSz6JemAt^+2bC$9>nr@_1Jcv z{ED||>DHZDSWiPk$AC%Bf4rIof}c9?cztsDfSh8bO70(qwE*V_$cq!q8HCL+Wac)+CJ8-d2fE{*RHJ84O8xJ%UHJ8BO;J+ujV7<#%OkarzsqysTdsT#y!U#NySB>bIB!3ctM`2IY+`qcX+6d{z@7(pa?H( z#b0h#6q2LzMQsKgQKd+aOQiJ8MUr4$$y`Go-4~AIY=+N1Vw+f_VBHju#;5nZ54&SA z@W&atFpDPi;hY)c_|Sf%YHRLrrun{~L*HL_r?dN)yz!EYFLA46eh}-=)@rm{f;jn6 zs}uCPQ3`F6T|VWzwPonfUM2L>C~sf_70W{o+BpP!D*cpUgCzq&Y4R}CpEETX@?=WP zNu3Y2-24nqL+Z0$xTY~6){P}#T>;Zbau5hZJH=y-A()9cZ_|X~SSryj`vwqNTj>8y zQ|EzmB4aolFy5ooD04*4^Fr>Hg3tcfJGNDdBJvWPWpX{bYDJUUwxsKbq7(tvGA?e? zwiM4y9fe?Xsmv>zILQ&f<3!Do-|TqmSVx2-OH9UUV-(Oj6yB}qC4j45>j+wORm^e9 zL3$l${Zavmi0_cH;GSii63xUkR}jk{BF&&yK~7ohqvT%6h>X`Am!4=@c!`J3U73L@ zh$B0Z)!!R^hARu~Y+oDF<3fqcoRK}PmjgnK+ZLq})M(x56xZmetLE*$qu-g)M&Xyl zH?)|V&>jhbNTSBkAKlRaWMY$ctjlB63}&yEYul>c@Hf7{avpLU&9{2X^YI-hV{gJa zm)|vbBxjLlq;K`uB=vAj)?QB3^+6k6iqd8iT}m9p&QX$)JnsPT0DiTA(YRC*<^Eo{ z65Es<$cfaLXCDy+HAJNZLQbU*n^Uy66K&&<{IPzI#P63&@XL8buHIFCTV=>Kj!nv0 z!hHAmG79~CWK@K3nINT#gIO%+_05C+FQP|UpZD!(@k((c2xn+3>kayJqm3Gn_6u7A z>*+4;J0fz{IZ-;w=oPsXJIZDFqdHvaMBXG0QeS#w_3?9t&r|%9V2yGW4JA~b6!uLe zH}v%*d*|Eb9b<=D((g==+B9h4TAR)9Iw5Zf6(Ueg*kspNxkgh&7#sbA@ zvw&-fHEQ7Oc9Io`xOU;lbmRBT{KMb_`*G^g^6c``(JK6;ys&z|cSd|%+L&PedUdb@8y^H&d( zt2)fpQEDP+krac#Y+iGM50P2j-*C1z2u#2I`H&K1rbhHKfIAA+Qc<%l3$KNJh?5d) zoM3P$ZT|5$t#dLeM~ky1JEGOmSb#CplN7yp$3iZmDZA(sUd?SKNvh{1-TOay zv7IIjjYIjm4};18czt~_Zol9{EvVh7Q-Vo>C|D1Sc1A%n?oQ8Kfuo{k@RcS#Hwtm= z#5dX3mC5~5J42byc6LfG2wCKe0<4%Ba++=vwOX>1=RV~;TTg5_MKOM# zd!9*&uF;fD9=Cq3y7M`c8Y2!i}<>mB`CVr<~s4z_iL z>F>r1%D!@Lq-XQJVfRtw!Evt33e$fQgf3UJ?qGtqiRc&b564y=im{_ph@G>_-jF`l zmJvO` zt>SFl7pLM(A@fH_wA!*e<&?Eh3~HI+2!f-&Lw|ej5QWf2n4$9Vef2_kqZ0MtH)7vGyu_|06^bQ)%7*V0t+ln6%DpgU-k~SzL(t;D zPc|n>-1c8_Q6HKcO?K_->BXi;zlkcY^nVIAnFdFb^Au9^F2_}lTI)KONdZm*5V|u2BL6bg~s3&CG66=LIzy%#}elv1AkUiHWTR=IZz>_c@Jl{biG#}A=;BP(u?_>a^(z;$iH*6 zJ`2ARO(+KaMk6GZqdf64^_Ni;0S6Au2)Np8m}b^b5-&>Rn(GerDb%WpGAa&4Ir#w7 z-0Yh)Lu+zBBsX-Ra@RX~be@o@SQ)QQV6-ZQF?f~Puj-aJs7=ZgyBVF=xsf^cFR09` zc7Ggj*xdZaU|m|AfD=w2=dE9M6*q%IJeg)c%z%%UD0pe{(2MFQ7$NMd*{^0%2proU zXZPYGYc;a6nO6@^r(m{Qv6C2M=Kq`y9-YP!e|^;X@YI=zTr!=&VJzL=#ERo!2#OpHplirQ?2Lpq|cXfO+}LsFfQ<|I}0U!lC-H*e7jp3 z$5!c`fD;366XU`PYm@$((SXkLvt?g!TX4*&uHT{#uPGtD+MZ~5fA?U$xP2{Ki_HNG zvB!f&k~HVMozRrhXs~M9DvCV?WrUeK{(y!)o)Ri^i^PoO!9yjjj~IGPqBYfz7BvgD zDb2}t)izMGzt&Y8ta_A+LWv$-IF$^(=z<|pyxhqt4R_~D_$oTG0i}ZqEL`GRAb4cC}48zuN4GqASP$eNzT^hd+?v3 z#q8EXa^vj|9ZyV$bE}RUqBe&eQI(YjPmXrQoT@GTSzQLqB^REk`1Y_1gStOp8Kned z#>5XXX3}iQkrgvH8ZInx3)5_co*g@;pv2kn-+RtknNG)63RIkC>6P*8Q>}Z_?<&)N zA*}HKYF};8P=P~oC|@6~gaGAwSECX~=(_auR(M+8TsCQ8w$wjM-tHAtIp$-n;VTmp zw%~L=F2^2~HV59})Xd?5!D=hu&DE6iIeLHQ{B?BwEa9DZ^;bi)WS4=<4@W%p z_k3BUJb)bD(c_8D=<(e9LF+pw+ShGvmUj|XuTuKOIY>R7x8F5fPMjd9s`P!XYqJ<$@@&lfq9ei-aA`dIfWSQ{9?7&ORO^W3)%_q)DQ9x_Krvx;sn85r&K z;KD_U7Hicfsdw4nmp$L12#xBBsPIjIA`iYZ4WjCb(2-cqIvnp2oJx)4qJo8V+P3PB z;{>LgyY)L0D}M?!V2cM@`+NOHJ3F}OR1_UghZ)3o(>#^?4^`AeJ&fbJdS^o?32m!q z{{g&v6(h3UUx~HoqN*SpiP@X~0Cyn8*k}6^yOFkwovyn7h|`JbDm@I&`ddQ-!(3mq zoc{rW6gK*Gk2cQw(z~fMdwKXa6c1ilRhQ~`WNx><;{|NuyV;EZk3CO(_ zfz*+We?&b3)yo%SI_iN-6TW4liUY{jlZNDT#tFjk{6h6496hETv`6a1mSfNt7JRcu z<;nm0u2G`>cue%(cb@fuTf-WqqS0F;l`MY$Ks<3;+K7D1Qx86bl&@;}Qg`G1{TI@d z*{d~RqfPBvrdK>-s;qZly;cKv35FB9xNoz+R=jFw%Yo3mYRb1d-z)Jjp3&nh`YqFb z162hBlA(w{DKs#r@fXNBIbIvr2DNAN*0^r}@_R|#inx|sqw98GMYRuIrA;rRG`lpVud%bS~U0Agw#%USpCm z>{gJXa>hp)fT~dbC=-$ugLkA?n_KBkM`O%~7D3jGU=qPe>d**!97GA8Ud+8?T(>EZ zKveW4cLIw_Rwv$Cal=uko+UK4s<1dP?GFM2h-2N86S-^UUP6V7(rLn@d60jFSZADM zv`Tv`G(*0vgC}`xs&of;KFaWPd5pjwZ~$q2nW?=i4?5=mK%~v97yG`G!}&pI%rR?3 z2w2$rm}m1iPavcyANHP$318!2Ou#PUC~`dK&(+pe2nQ_$Ba2Y3&|PQ`x&}@RwGZpFORB3o=cwF#Meh{>~jc{tdSAm*+#RnkSmz zXRXrJ9$MklBO!FgO=85#*VN5S&^|UZI%?unX$0bI(!t7qB4nLje-z!*`fl1Iz6upR z^}*+G^RS;19=nDwgP`|E(MvD+tA61){yNfYb z;(xs+jbk8hF8^|AeJbk-`)c(_3Hk|c=iO37)RV4UE-ZQgW zEe1*Te*NelfJ^-DZPBY<$$wIm%D48aGdFH=_78GPJLTPR6rP)v9!hy|Y`e*kyom)s39rvr z?p&cJlYKvZ1WKv!oR8uI!2oi8j2rHKpwfeoBs3#1akJ|3A-p;H=l4@*bpy%s_hL74 zuZov(4y6_xP<6@&z69YNs<&$H&wW}c2J)Ai8Z=_(RL69_3NZfSqULAZif-z*+%h0j z&^r7$8OCixR!{4`$)d`!rB6U>Hoau8;niQ%t@%YM7L6LM6~-8m{uczIv>}ek#)?ImcP%*kucvk1_%n!1kr6@YMz^S!_d#KK{3*lCFtlx)7m7!){Z>|Tr zsqb>kUO%du@uzZ`FcKsBwAS%#l>Ve>!X`zaq2_gTF*VY+?k|=~zG-mpm}6)4&=%Jf zEO)ORz zVhxkWgC9e!&9*y#2DD!JTcHW8hq$NJ;Z|uTGOY zWJX7{6Voz>M_N<{>;|Uhkc=NThAw4ai_!JDGjr6onk7!h=%Usxb}AP_swh#Sln1+6 z*DUAMKZJ%7i2GmkUGHfD3d>44^^%=(Rgy`-A)L#kT*ranXs=?|X2h z5sScUychEzIm1V)-p}Z8X>_(LsPHMlU?0Tsl?RGgSI4RITNlGGi?3Q|@i}u%jW_)< zChDWmXlvmEzc@0t$;ldK=){ChGm%?MyGMzH*sz{J;`bxD3k1hheN<_>0Zdka+b$2d z(=&k-0fd97`OFU3@dcMj4v3RafjMu|we_y{$ZgqT3{nWZGkOxUjz4p(PqJpE%b%y& zZSC=x?_Q^^raO$kH9vGSAgm6k;Y62)Eo`2h9=GrM(cJ55ORNVmk%)T&p_|gDV|v~N zNcXTx;~bs|Tlk%tg-ElX1NyOb(!;``UuKDpO!jp+1wZ!iRIAcTNbybn55GS2Z%>FC zgZLvWpFMpi7Blb@w{=^6_|J`?ucW@S?@Tn!jeYp_H`(R>>{(^3@KUw2O>_25>f<$J z;ht!>d|X=kygE;SI8sjH^dOMYvoXVJtg#}aBahR(t49Ftk) zThD-mE^fg*XF+U8Gedav!ovgY5?a*ePX2iV1Yp7*F65QJ^vkeQW3B`9AbdJwDPa)QJg(>E>cG=3EM;~sJ+eo%W%+sGy|;io5M|54u7sKz zm2(sGjc4`mIeSJ0wCF{kJekpemUkARDpBMM#aozV>t2kOS0r_gMy)u7UBk@QU_d|9 zh2EYWWzsCl%wRiqAe2*>#_szN7c$P>)H+Sb53$r+!B$;{mq!!&6wT(b6L$49q)uI5k8 zyt&+JoT?2Sk_eRUPnxl%bi{T}@kA@6PNN$do{yOU%n?H3-UY&Q>cyL%92F+*ICWkF zk4AEMT_G~YLv^azD=2S|q9Y*A9Q+J&%bb#8Kj&9-kq_x!6~tt+wn{AkH2hqqv^+?U z%Aet@&x|sTH;aXLW_pFhGFfGq$#H(H;~?lI=?$soi{r|S>2#X+VlMZuC^g6Gm?kEM z4#u=RXQiu)+T8vTa7ST0a2S2~vu-eTIfe{rJ!1zgZ*qai8E?yj=T`KGkAeUK%XS)v za&99FvH~fV^a8p26$SV2$?z$v)y*hS+u9W6fAclr?hqVqtUCo|Nkt~jal*cBVSn-E z7qQ10Tf8Bo2`dV@j9kOvnEKs(dLW=(_ zxX~$fYIwO-pAyi%iJ{^QTXET?T;)aTrM=gtx^C#A8&a2w1l=4YCn60wr> z)^RFgE@DX6(P8X(L+|CAO>PHB^4hVGf|yP&h08k%(L-KM#kC&7GTotUap_x5W8!ii zQ?@Q6+r@;NcSzCgvT{Fo*K(k}Iunxn&BX0H=RXF3n-pfeKKr3U-oVNF-NVrogsT@K z>C1Qeu%qU2ORcPX2jU4X{VgyJQOlmq5S>vePHhX}9t9~hro zWv`o~W80myqR{A3j5~PTinzVi_$&1{e|8KYSjl2-940I9c!;%GGJz5d?{ z!*N_hY$3u6Dx1Y@(i7Z3^hF3AkglA{35ikgFy`SY{K!#}Bu zy9K*7!U675yy}ErX}V#<-M+t~CP5&@1)+0QH`KfEdH)LB9OE zdk?eyWueiqrf$;WlevF@7E{`P0LK1@2-JY2?lD$hqfYHu#!M^lJl&Ob?yqV!->v_? zel@~EQ8jyZ!rm{@q!_Jg60W90SF`r@QPl$e1K0zV_R{>|9w&*W7`La@r{mMV|3v5)s;c z=_?aX`SlEUmnp``xP^NoNCT*S+x#Y!U-~TCGS`Iv^Kuo>nLLqlja^WcWp44yV&BmDrEEUW(`_oiT_|QX z@BjCvS(v=n&+7sMHW9|AJrnOxi(mDN@5whMxM6$T@CafUwwNdE-AuZND0T`(zu{=X zmPhgRE`C~p#{oy(sb$?<20Ji;CUfu1wxp?%=!{NjcML^EU!S7;FPlulA+XddxaZot zS0J+KFzky?>hGDhmV7MUqwDAgYRE-K?2+}`TTu#x2N?nl=1Qr9q7B$IZ-sHVrpJ+y$8e*kHrJ%paOVek;^Q!X+* z285Q@-ZD=ce)u zvwuT&^(P)1@C6>y!>vIKz8;&sMHT+js9Y9g2JEchQ_2#$zfvU~Dre2*_8 z)Q$yi&MwARxFd%Wf<525Hi(EQKr@)0>w~JF3APfXB;MOu+z$^AGgRx{qd(#|PFd>W z#W%rSc-gL&i@?7PGZQ4YYm^m+Ot#`qv6xl&Fo#!bV~zd|JF@~g#4m;c16N{tJNu3F z+3b&Wdns?_udCK3j*Shrt3~GCu|C}}Eo`VP2bu&VVEx`H1&8ZDfQ_kKU()UGpB806 zF}?)H^NtxSsI`+B?@c24<6G@IRbsf!DDsa}D!Jr-mRbzp#wp)U?xn+CSy0D4s$%{D z5))|>n?qIg2uRsK&&KB(U73NV_uIz0v3W&$4#({PcD3}&*TLiBp{<_!&zZX*c$#V3 z8*YDBs&X)=GAp#-Wz9T0zoap;7spigJ+>KLcTDALZ~`9jRx3Gg6#m?3SXMaOCZNJn zB|S2{!{}$fTO7Ke6b8<7E##~$#~@ujro*(PP8WlvN=}rd9Ec7`2al+|Gc9>VpP3u8 z+jed9&l6o_j>>I>KdlS1@pr5)-`kemDsF(r7-r0Nm$hau>SlBOp5|g7xaQq0SEcV- z;2E8uiP(^3m}+azMBug8s{(F1cq=5!3nFY;c%CTXKm)h5ijw7VxIl`@w4UX+LPXW# z!npcmFMvP8jQzq^?pK^QW(kfy-nsj>GS+DP9y^nv5Qu8Ej7zbBDGVQQ;V z-sf-dB-%|#KiFGafM1>0y&#ueU$Jh6J11= zh?@)F2MAVl;lm|SACNuKi@EYw)jx$Apm3DhAi}ed*oe?h$^3@gli*vYdnG5bR0>oc zzyUfI4Nu>R?WPZWzSK&Ee}F-BEqlqs1^C5-p>y>plC$9V-?TCb*cmBeis&c8Eg^Hw z!ol#wE4pYeLmt61c@x=5h)=~UR)?>C$+@xvl5zg=MoaRPOZVzGc>0g1loV~vVZD15 z`=+l8E~_HKDQTY4BPWJ^R@Q+1n&T) zVJP$6UKDZd(rbS8OorgMBf<%#{sD4wCe7f2((o}2i5>Bk%uuR&&OMs3y^@qrZ8yfg zorB@1q?;$8y-9Oh70F&?aYVvgJrYfMQOr1}nbtovF1C*7tG_n~r~8Eklo4O3e9X24 zZHSd(^eQR++taY7?MUHQJ?lyCpAasb0eR-oa1Q`v{6|UMj5otGtW?|2hq|fvR z@*r_88>H#P8fH|)5sk7Lm-jvTAl249JA$|mcg$C_at}r@9dDk%NT_i z!3fUesuG};nek>Ev15m!%X#k()tA!9>Hh=BKsUeM)_U=|!@KS)9eLBV)l|e!3{${; zUHicM3eEJl@VN5Eu?&J(`$l)R@KcX6rgojFppW;2rg-y{6o(ET}<*!xH z)YCh@!W-rWa;UVs)jPMu9m3lA$p8`$y;;;-DXLscK8fecqMhRx4fty{R+>7l{B7)? z=HE4it+iFsobq!=_M5&|?ejeBFB!{QY0|0f(%N$J&17H@W7I`+;Xmpr_HcITs z@%0kxk|)zh?K?Gpz702S!t!z21|@7$Lo! zel9C9MQo;nzsP2H+CHktF`xkFJ0NrwcE>6Fv&e6>tzG3hLX)*Dnr3Mj=?+ajh3CQF85m6qgcucnqF3dUR${P z*T%oaIe9X;+a!CZq{9inE9i6``R>ahJj19YkAk+U(U<1bwwzPQ=e+k_&w-yMLqlpk zO<$;~Afb-&b97&fkFq?6`pbIM7wu0SwxX`*Wv`~Cj&bPh48)DV-1|mXmzUN0oNlF# z*l_V#y3}YmajYupj4ZcE&R(u_yQko#Zx1y@zLCjD=Lnt3@*m>5o|x5omOE56QCRL1 z);O|2^2|foV{lGAWoLKd#U-AZp_+cO3!|hit%c*cj|?1jK1<6#iF&;6PC9DvamJ1~ zN3ESoCWeSTglK#!>t^GU+V`LItE|%T3szU@@n1uwYb1_#2Gg;TySnV}&;I3M zQN$~hH`)th>{kZXgUweHM|U1Pr_E;n01)-E>gi_(6`N>>gSy$a&r5wEFDLO`whk2v zUK8tDhxT#X&o(oYJ^=j-tviU6MW8{iH)A9~gvyVmD7EN1X=2McWyK8MnHOg788gAi9YNUsrP-D(y#@xPYO&i23g?)y4_%!u=_YTmuoHe|y zb91&q_ALJZ;$E{`?7yeu_;W~jEv}tX_CA-Dqt$`LnxM)A7mS#O?ec z*HtciU2lA4z7Cj$T5b(%$lk`bhb7)3bAj(TT_#>Q!yVpAD}9!xM^sD6Pt3wd_kgpN zJVVs5h$WXvS2mcicFru_hs|`px7JDDOwKMh8jph7_4PfI=!-Vp{*rlBcsL=hA6#Od z0LTXYps6Q6a_KsMi)TIZxm3>|Fs9{?*1_>t8jFmMw%Uvpwo1withnsMna;+K zGvu>9cdECib8GD3KMq?b1$FDCaBU+w218xxBxlI07w~gmu(atX`V0C zUudLtw@z2o$q6ya=;JtJ!0Xd-?p9aMCtmF|t*EcFRnfXhTE{c8^@oWbJFAzh^}pFC z7amSvI47=*0#gH;Pe%Gd%BnaauO`PV9G+#>MRsU-kyB9(Emgly+@!av#h98amX+N5>X&2XfiPFBUr(dWC0u+iLYhVt2|Xl&O@gwT@LHW*28_T-FVOg&$% z%f8`w@QzH*3-JxDbcTXQ06kwzc1oU34D0CVMMJ5iX+0U~4kQ)U?>g@9Yoa$Q+8do2 zcX^Ilx3$HjK5zZx9w1-Ba1qFbP2eA^two)n3Q0FBcf8 z+RFvXqB)6_O4zW+Fzx`(`CTWf^_LE=$qttRoT}^iGfbmYqNpxeUDIU_Z^*Y*WTVtzlR;z?jNWx;7 zjK)dmG2*d(Z>-0)*8c!@rxzQIGlSel8{BYpwMjx=r?GVa54! zIxZFQ255}Tif&SHN2NHb(D1K_#|;opK{)30f;afCkEZyyOiQuUoxNpYp!fAD1(&;dc>fVm+f=Ply$6#y#s^ z`G-=m?-I*9TN}9^?%mi0cAZ_tS_fL9tY@S$k^4GUFgqFgq-SCmp16aprny$Wp3hxS zE$;P855?S za!+xtd3H~b3d)8Ws_NlO0}JFQB$s3X_KETfu4pYCL14a7-fFc4mZJ>jdtETbPnJsh zIa|kt{MVjqJiGRc4w~Usrmds&`nX$36NfWc{{UEzU|UX_;mzfhOoi7;D464x;m#TR zfm^NLh!(3&x=HFQY57UzXJqjFpJLW*d|1-dmEE<{=Odi+(m2T@$$U)xU#+(99e=ak zj~4`N`)5%ggF{IH%LbwbnO#k*uYq zpsJ|tM(%TEc0l&-ukF=Ulu+WP<5^KVfXFjAf%;Y*vb0F!84sQQv&hH_cFD=ziTuv` z(jI|*YzuL6%R|3(nDimc1G{c%__&Y=4otRsr8)VsWz|D zDqCr`xOzj!GoSWb;;$zpIk@%<}GCoQ##QPp7qIvYE9$ol^F{ zpffmN{8l4R!KUr9^E&%qB}~|Hd{n%}r#=OAYUyYwDEkY&&g(S{u?Mx!rx_U?Gv>Yr zsr5X$qx#09^sBkMS zp(^b47U*ZwuDjGk+?kum3%34lHe2=AS~}lqj;Z$Roh2JNL6%5&13n>nIlV`$!!M^= zb(ryG<8DUX55z$J&CtjWag$PdrNc$>3Wd?OM#+f}tSS!6v}yfeM{=3s-)p)_8E|mt zi(-88u(!>5Lms7$-*IhhWzHZmi~ux$IQT42S?ev->EG_Y-yHJhvFNRLEanz=qCt+e z1cUlj)ciZ*EKr#Vd2s29qrd1~KATW7ncct5x0yRZnB5#X828~xRm2KPN7O@cq`4Vk zEpsAdIe5>Q9a8$gSuVbx@^o?E4pU8{c#}MaF$}FBXC|b8eaf9ykm4;YVbZ}iPe3XT zO70(qP@56BL2tUTfMxEUE>Z`%Vr^R9-+iGTi3P%G-7AM6XM2=+1jp6-%)ZW>{_Ng9 zIeWRC27uxvRTreO&Tf9A>IflH{0ZU{YD5G}l5xx`4i{v0wT_+_5ofx^9V}(?I#@8= z4{Elz@e0#&xJMKeH);EU*=0UPed=Fb>ZJTE;>P%M33z$LhRKHWIC=U>P-cMQGo|dD z?#%mYit1XMh&Ic0WHi-qSHNaH-d1FR?F$uP;uRIevea2=FLiNqmO5ihT+@!Y`}S7) zuUD?DY|?vpcxh5_=ZI0pPi@TM;M$T@l==sVa&r!TUQyB2F8zz4Pp|1Io!eagbb>h` zGDQi4mBbIzm6xK{6t$K|*3et-O=w{zSz#a#YM0jfx#q*#!p9zuM{xx62ZuGyI3d+8 z-Cr|@y7|SWlDvlupWg>%zvPFTwwj{BYoD^>rjE!^mWiOVi_1M9K`wi6#Zwa-nTYhEH%A3R9X*+MsXX9 zhJ1Y~*jG&9?z{|;zHU;{mVr4r_AZd^VNct=(NfpV69^6v#oA!`ZP8C%uPwIt1lKEl z&Y7=}#?tt)lZ@ak-&5-()z;3OFNo#QXkH+>yu&Rx+!~i>^(qo*?jbNZIvEKa2c(_X zS+MK63(YuvzS(qeEjV-FbKHP>YFrHzcGW+-rAD7i*;o^H9XT-QC(q$j1Q7oV&!W`lCjfz{?B-9 z#@R)8obwUPU^u|>S3g(kaqqZJEPfo$Z7;6ieIxtf`!*`1CZjQK2$cBN$Y_trrL z#u|1;NO^~^+@$rq^6M)$w~HUcj!R+IF;zAR89NC$4(-ohioBOb>((*T21xb+ovKMV zTlK$C+3Q0qdph|!yUYw=B%ff$R$@I(8(Q4k?%kd_oaaG+8n=B{te3u%qX!+J(>m6* z#j&z>I2>Wq5TS6=TH=|!(!($r#zjHFT?)g*jW;ds;SC)v)}`%#Ddnd)2g@qgTJalB zQGn}^xzY@CV8zUS;IHdFPWAM!w}kU6pF2zI5&;aH`#W;5*QhU+gUeezOh9#Tdx##) z_^ppn@h-urA`x5fmueaVv!!J&B<39eTD9oQacwKD_gcHXC_hW-uWSx$zoL4tPai+2 z{{UT^X3jnK+GuHSwXw2#d1~Jay;ldF{{TvomV)(HUFgi3tF#9NrS8sF)nvQU)7xT( z>jT;&Vas8vU|CH)s5I4*hObn$%#xThnl_mOq~|^aXY{Uj)%u&3vBSH5-xr%)vtG+z zr|KJNlI2k}?`CC>w0CzuRd2jI;WnX`#|6$><8PLwfyJPZK=Azwl-spUis8iO>OFB< zDd=BO+qiP%9D1z6^bHv1l#WZC-bX(rf&A{u_Ih4EUuPN1_hiOd%(J;q5x!I0>#6LbTH3kh zlh$aGLU%ndtz@XGiqs^ine$akkPDn#clY=zHAGh3qoZbfCPgr4=#PnpEg|>HOG%uO|}hHcIEyZR!@#*ckNeA)mK5* zvZVh2;-A{IMpiwisrahRuW&n_tHajx^tYz!n~@vURG<7)`&8`JHvoTzRQ~{Ke`?XA zxH~;nX4fgyvR3fjQF1@?+Mn$lt20sDag$k5{i*${MrCA&>tS5OQtY1oeTmdw&A#QR zt`S-1sBL9HaBczZO4Qk`ue)aDQgA;ia&cK-lM?u#~cO>aBLr&Mfl_|yAVi=w7v)zvivcEKNCB2`Xk z$26jB0!ci_Wvns168yR4$@OWw{uI|^_&_kwj&2SP-YVsmyu%xEbrGuS!2bXkPxs8>{purCRvpECK_BUz`&Tx0 zpkZN;uz;S+v*y9bKX2ZtRNh3@0#Me%0|R?DX9pfW^^bAB?w&PGY5xG9)c*i%@7kY* zHF(~>pbzx!{i~QV?A?iThI~}f-bFc-xv|P0A0X!kx^vVgTqLfTbxlq%Yp46CKebFX zO;$Fpsn7ane$~z7zqRqbqK*a`BN2yT-OPB8X;#d-3=Z-ADegtpbuYqt(tn3lXW28q zYM1b)uN&MZ&VMdfFE?Q$9MJQgnFQ>vpQsq_(Sn+o7ev^`#~_Sl-twJr)pT*Gs|fxd zQ~k5QxTuXyS~}NMe`@dAxt&etwx~0Pwg#4fBVZtH;-T}v113+qbnA<%>4#KR1HFAf z{{Y5t{`C>5>d4x@a($DZ(Yof5-dhtV4iUUcz|dX~aObF1X1}&r<&Ctk#oDo;J)q@) z4x`#rt{Se5_)S^=08Xj?+26HHbv0(+zp8z|PVd^edv2qbca!nTjF(nC7Kt1WvpXkL zPM1zPnzLXoGw1y={i-8U)*t!xLHEk<+PS;0?SE(uH8;{m#*L0zT6ZJbRXMKxtxM&@ zVGh6yU?2=9TwPa8I+nC`uc-dn-?cvps=>pnss?=KcK-mSb0c+cQzMVJeUxLo@Q|hYpQYno!_+vrmV;C`Op44ziQ@gx;4!qvhusA&Q*E! z0CAU=nwh<)FF<>wd3(}l7gjnb)b(ejbxHpK#VS99)u4Y4sn7ane$~xtZvtw`CK!Y~ z<=B%=ML_abwy{5eM5coJNBpHJzKh5Di6AI`&TuJ?i&=b4$e;3 z9ItkA!Z$egg;IO#Dj6t_C?gU*uI6AlW(MGU30JNi-56?W+~d<~5B@3rswY#{bv3mB zf2BXQa}jjM@ud4OR9|*xFeSmxMst$0)@r(G)b)Nf^M7taeiPMWBD$mdQ~OssmrybC z$(lM=^F59j2fL4|H}T4&S9dbb`I*ZQk^4r_aNi6o*D2rS($1%={{W{{e`Q^t~6LIq>#v3@aMSmH`%wE+{Z7(XWXAIfXz)) zovZ2(x>NgAM(R4e{{RiB-~O=v)uPP%gwe{@Lq``p-KR4ga?y`edorOLE-k@y(2zfz zrN=KiUbyJ+U#lEOYid8WKeb`ChM~Gs+x;q<`-%)7Se?$lQ{IIz{#jzf>epW2``HOr@^ zRsGUC_N)e^GSgg~;hxSx%+j&VuX|1dze@CEA+NK(w;r{?jss7d9 zwP}Ll{N$BUw+}M2xJumFsct88RZ;z;ziO)8Nnif}OQ`bA^odHZDB3Ra=NFfZU%bcPwu;4Y8fZ&n^7E+ zIMYeZ%gk_Eb~}CQN7Vt?T;jk3XOMgftQ^{kQ6@6qRpZPoMXNARQD~cXWSGhEY5j}Y zjojRF*1y~Sjh1`GR5dV6>Y1d4_-Mq-3uT-j>nFuATIO$)tVY!l|OpC0AuNpDjjT_yKSSH~q>bu^u(?S4ynzW)Ff zvfC?l4Rs4^ppvFqcN}wxPT!?qc#o@WwU!7iF~x9wWJF0Dwev++LksEZ@00d#Bo;-EtIx^L&xgX} z$A(`lT}=kBrgTC`st0I0zfN=TRc&^yZE|F;rKYR;BZmh}Mo-0UmApdI6&I*&@-pwv z)?uxD7(00twOe&94JEt*Z@a!XI~@Ltf^*Yv?_O)_{ZEN8boQ|R?n7Bmtf^(}=BBC# z{V~~HS?v1K>dt7Os%uBA=So7}HC@)tByowRxJG7wD_q8c{P|vM9lg$?ghCRoUQ(v z{c62QPTB!=moG0dvLU`a50cEuswky`NFQ&xQ@S(C<0O2eK0uv{W9ofYUNl}lIe#uI z=z8Et#(E`7i;pP8$FId?sx+?<;|y@7k=)%wY6-&K?bo|SeRR~6rsV1AWXqb!xBW{k zSFbG;b&y-@DXx&t@qpksFz)~zSbARoSh%9i%(`A=K+!pK^R!tyDqth z5xP0dz93W;*r(O7KAgz~1=KaWb2b7&_j#=%{{X~2F)dlAtCsakQxkxAG{wDodzGH2 z)c*iu+QF&N>Fs}9PAR91G~km`RJE2pdrMIqGSgH7;u_ZpjyL<)RIpdkTVZ{C4ZT%E zDCcl2$Z1kiYWr2*xA(Ml=2)aZLSgTF@U!~QQTXZq04KGJnpEkYA;93r)6_VY^8(q) z3d&1HyklBjOH@HP%O)fP?OnP}UrZe>G`7o}FNwU)k?~=mdFP_m#-z8@>Id4QlDv`| zE*T#%Ap1aD%hdX*H``3iFs%`qONeh{UD(Rfd<^F*wE8z+G%!HvWpN$wWh0dNdsj-n zYujy}HbGfOb%oAwXpq5>J-PDss^5mRLMUqK?e|)kUebA@hnNxPw^hsao~QdY?PKuc zl4<@SM=4~Kl_2^0K~-q9&cCXgvu!;?HqKo!T|N5Jgi=j3?6@?xCrZ}M7iR1Oz5(mi zRnr>6*K?vbi)}5|XdW^{VafO{=jwf4J@V9vFys6=n{;mwT>e8GfccrqlA}iP5)6|J zgkz|vE2Y|aiDs+1NlIC8?To@UMFi>o&TC0QhP$P5=`fmC@3iLkx~jN2n~5p%WdKK-3zR**(sZk*=*|^--Q)&7s4X zA7I&LuG+4a&1YmabT@MH!OJUP&ODPpSL*QXb-&%Y$H!-9D0YzI0hN(9I+9%8W0}b1 zR+4F+A;m9eB`rXAnW;D{q&Hl``Q)skb>gMG!Y491eqj@c@m3Z5LAJ+H6*b@Is3x7* zhF4OSu)E7DndlaS+s;JhWm`y2gst zv=?5|qK%hIksez1hIo#?3Y5GP2i5Nhx9M+6&s@i~>IO*AG+cGM7?+mU=k6S39iJJV$|ybEs;& z-3M%Qm3A~=5aoPqhAE!+GnvEE2;@FTb=Y02V4>4-RMp$=v&LimV_^>WEM2c!(Zf8i zqNlo4F|i3{kFw@?&$FuI`tMhJ&Y$wW))TXm)#<(=(o7?smZ7iZ*olcB6+()~uWKp{ z5z$q~?x4D3;rFi9U8!j+0I8aa?N0+wB(Ih(Ko8zJEJlje_Unb_X4~VgsFIz{wovK`oD~S$?ajS-MAgj zJzH<86xFfRvSZPra#d<-JxwjDON^4vDcIn{%Oi`qkB451eAaiX4ZKGs>{r7)llEW; zfp!m=@b|1l1(RByXzn+bTfpuJ=|hR%sa~t;HliT$iay7D6RE}+fyHW;z zD>O6({<4F$W&U7XH*Yhs`qpmb`ZwL(StZuVQ5%TvGN=>x4D6-5@k>eRnOj3yQX45| zM`StL2fQyAH>mZpO=9oIhXuXsRn)p#ejk~%U2_&kLeuOuz+y6+_mhlo-1K0_EGnS5k!2KZ7#JA`@3d$=t#!B0}=jfQFn{`F2y9_z^WeH~eL zafZ}8bMakr<;6=io-s^esppNM?<{99k9~u->aQ8ZihAlOt`_@!$<+AF?huE2$nx2I z-2G3i!s^U_%e>>_t}%4s=D(sacGNXwN^ssvhY~&es~Gf7w4SO;W>Qrd8L|3zV1AX` z?c7?jUhA7~qqW^CWQ;hN`z%ODnwaVuidiadb6YMjQ&N)oq;)Y34FjjN?i9n+`pb2C zcl)j5;*Lc(2)KkO6xDVF5jfac2`?Wx( zf|j~=%-(2&TfTfZSGBxAuCPr}RbOSfMH`sYPQeCo=gdOmeP^t~^ySmP-F>b*-aPej z;l8z_s&q43s=3ENDjae>o9?b`{3+MBAYTQxpfq*noxuLJ+}o>a?zYy}Qd?{59T*bH zBV))dKT)IYDm5Hh8fvAq)t1W0<*5cpB9SLnvw zKR6$jal!bj`8YwZ+f4Z#Zjcd-H_}c>U3%ZEu2yI!s;{tJ<`bUMIXeg<_BzPeC?iTN^^|p0MJ08y#^yYNDCfBobHI~^!hfrz`%F13E>zHIQ z$#0lA9fxjrUG!_4wY1M&S*L5+Kw{;LvMypt=;MzCiMDIX>#Y!y*o=60c7J_lU{V|n!W#W~6=E}*fdVZ>^;yHw@1G)!+3t9T_TZHPm(-Vh^ z_;L9-Kd*qmwaD!3&?*k&&#Iv}3pK?wf-`QF*%-@bBm?Euc00Ee>tU_7&`oQ&(^YeZ zRJg<4-^65AE*VdaU@vpmN>vG#kGfAJq#ej4^T^|vEu2DR{{Gs{BY z;p(tX;#{fH@Pl1c2N6K#xxZD@5V|Pi$3CEf87NTZSCqruRK$`>83$da?{n(Qx>(Z z=ZKznpAg~D4{FvQ#C>LVNh4Wsabr6<6g(~05=ihZlMYkLNtEL%>Ru1))>FDh+e@%| z3HSIerDfv1W$M{e6}G`rZZf^ajvtF_PpW;mQ(m}_sBKzm-foslly$t{zF!PQ*#oaA z0cZaJ6ZKN`-E93o9dZY7zgtHuTKi5``Qc)DDuDXkT>^^r5RrVo#}2=FJwlCwXGdcW-7>A2j=n}youcb2l)_dTO=+hYo) zCk(ZWw-!>#M(D%3$mx5nU2<+JUToTRmG=!PT*B6J&i=^Yzc@Gpd@NOyjY7SZ5W{4u zlIuHxq>K!(Hyh-w{{Z41v!|=JnD%(LrjJ#^J#kLgJ+|>IVE0}o-yZBN9cGEuuuNh3 zws$ygaQ>^b-8hG-?fO}+y=nT^Rx!oX7^BE@JOgp!t*Lc2)~U4j8-AX+%|%wnFc&lF z874L-Zi_ho02K9Cyw;cW+;hv64jb!XMd`mZ%h{IX{Wj{I27=Xfysd`cA$e~{yp7jO zn~WD5AQn5Nilx$lhss#s2+sJ<#NlKj*4LWp<=~5KjJ>G z3B069!N+_#6>ksqZ7hY7+b3@>&LQWjvHTCzrldyOZ0{s^xP{g(I@a|rqNby`TJ8-q zoW>myYa7=A$Ilp8El+>euG95Ww$E{e^zxEwAcUN=I{-XZxW9;ci>cXfr{j}ucUNq? z+8Nl~;pAXAAm`zgYff7Dk6>~dc8TZpDis=rvhz~*32W-4bGD9Xh%I-yZ1&13%36v_ zitRI3%gE;!v}L4#dj5sK{N7i{j{Y`rcyeXVuC@7e;=dNutAxT{L~NGl`ks-X`! z?uQIGdyk5}w^h_s*=p)8RBnQ3YPkzvunpUM9-nflQ3;H(nQmOr{zn|(cwqQ1rN``@ETP@^N>KI{{S)3Qe@zaf~vMh7ULmj zOXFfBD`OXmHLHf3 zrnuzxVU9mpYWIyVa-jvnE_nRH=&qh>XA>g>+0!jh`N@wVk0IIHl#SOutNy6ZM!Y!- znC6@~77^4Gb~A<3E>=rr$D1q>$bPU3RJ>_A+FZVznVI@j=$Gwhs?RNX6ki;X9$8vi zLB5a^~=3)g0mZZD_sC%1J>vO8ARJ~4>5KS z{kQD5TV0;bY{R03vrFS7o2&Y%qW}jI?DPX*w=kIbrK~l#Pg7?vi z*|U|Sr~{x?DQK&!X8H(-Z`yZ7ou<=t@nTxpOs~g5(RwiA%P$@r^5oN7riOZ_aze*A z`HqV*L#GJDm6C#U>IIx{qtnD03~gv0K}hL|*e`ze?0 zp;TVtH*o<|?FRTG-IeQ@N#VU=Tb#j285R9)(+ zZWVDH2$=k_pK|JwwkldTh+<@MvAZ>+056s5KZtZ*oMYLC1lM;O(!4FUTCOm^QA%rN z<^V$9P#-e7rA+iz=_u$a1W>j~149{rIvtlk*Lt#&E0$uYXb={;?&um8mg~)CtJN0D z<5eX!B_(r=k}>#nj1AawuMes96Si61{{Uxx9vk922NX3mpHFG(kDO&|p7wWP`qCdI z=DlOA=D7Mbx(ZfP%O}z?*zPnFfs^L99yDAc))&WHEzPQ_FhthGB9A4`_p4ZVA*gi? zB^$2QKQ_f2K)JD=`94R;FPE1mUNhn9#~FqG?o({4zxq`~ZRe<{tdp{hz<_h?UFSh? z!jA20hNo3}>;ck1$+rHn@LHR+eXC4esbi$C6_r#Zp2^Myr2AZYmA_hQ{{RQ6?zg$> zB&hktz_y$`y_EglnD?(IKQ2AGW3D)MWbS(VNNE>*&0)88)3L4`kVl5Jw0&dZ7aG6o zElI1I`-NM?RB_G891U&=Iq+Mam9SFZ=?(HfNt*P`#S27yjV?R$9{?6V_Y%0Zv)ylZ znPBCdP1v$f*xEtJ$Dg@QyPx|S#~%t5lImN<&fRae)Jphbgfgr;UQj*Zbg1;Uh|%;; zzTabHvl7=i@}7S8^IPJ%jl!yW>xIUW*;3d|$CjVc8Ql@M4_IS-?bU2tN4e7e4_b80 z(@f@4xq^B5d5oAIe*M=y%yN?!vX>=2f5#x(d!o|T`C}cgV#=dr7M#Z!asqnmkIOHx7RTXzO2Fs2ec8RnNhVQGRmDl=-8%HyXx05e(d@<$kTjfpPLGaE8 zQ_|38uarH)Hgi4sINi%r(eX(HXRa~EHqyw{xnj?q3ziG$8W9GFRD_`_tjS!!tX z#S5j92*_ib2P{nR*eW!)db?hOxmHV3_f^~~9vJ-gqBfm_K=v(vPieb-eWIjTn^{+O zsmyXRX#1_s<=OBVT=?_4)GizET4OI4o1NB<B`u`ho6!6T5C+aoL>xOdhyacJ|g&zI%1Hq42gz zjFUK2&rsFV z+HEkthK{L+o0+?~543ze%Q?lms(PVnps1E%LrWMlM-+U{FWUz`%zl-Lym0R4t0}1I zblZ&;3prgSQ->)d8-=buOLdHAqGR5)-8+7*w%V#$4Q)hoHV3jNFn3Fd>bgdP(9mkD zbv=-uLP^}`2TIU0j$dY0YeHS*i%?Tl zTx;U=N@!mkadyCGeVkk8;HG%q{WH4nvuo6v^Gs;A4LxR{WYv+d{`RM5axv{IgI~2q zpwv~)yw$nDHlRlsBbT!o*fcGYrn^tORMl41+#LGX)H(dv9EH2#ne$Yj;XTgzQCD}X zzeP_BvNiOy(bNX=`$_QkF1IZ5z7k_GiP?0zT;#V}D4jg4b9ri5_uvvwS04bn#+=Yv zKTF2_M{$A)Dk5(?=aY8D>}ytmeU3Y2)vTkdWA5x^a7QBmM8GzbkC9*Vj<6SH!mk~m9oQPp6hC=v1v9`Nj$RtGDz@Qw(cjuEr(2SV(F)jjfeOCh$?7MJAqZ+nA*KJZpUiiX*8yTLu$IeR*4 zX5yX}3N0;Vp^p1n z`Q1wQIws^f^1Hy}z%IFTWjB;3Yo^b`?j2a_T@IkW(M0;Z2eJa}vOIX7GOFwDD^c5D zRb|t3&Yq^45??!8hDGdZvf<@*+4c2zTclt`G#!K_%sV)7oDx00D?x9})K#@|z4dih zdZ#)!$k=(C%bmF!;<@3?amM184c*qZZWnQVhOVNfNMBVLoQ7`VbI`4}(8h2t7f_j-Cf#9n8^&3@pLY3asz?r6~3^&4F{;Muu+(*@0U7u zhsDp_hafcX;6mw_A6Jh$G0R?Eu8xyL@W)L}bdaz@F=`@W+zVAnA zzx>GBIGH6Fn3qjE@OhdD^VxUf%H3mT>z7Wh)bv$UbhH%_MRKc#phEu5og<4LC)%_d zjdjMlu+AoyHnrI^7x#%-+8WVouuBA$lRl!W*NM#p0q^QP%U`RIIi$M{Xz(2Z`yP|f zdF14eFE^;y(G(`CsBIvoGf3zGE-fzE&OR#IHkl%k-0W*(0K1>BUj=$-WtJ%Ck~av8 za@>7yYQ#xX*tVt?#z`XsA$H5ug35bW>Sbd4ELKeIofw47~j~mnXQu4 zK+@{?_Q-BY$o4GYse7Z2HpKJU^YW4HQ`mIO!IwYTj}=q0 zc_LYu=VPM)*`a1MY?O}Gwo0mzGbtHb89a}}TF^J<$CC5(ivUDoKU z&9+NxuFyqE>1=b`Oio?M2dv@Xw2d2Xxkai>voN+X+H*O>1dJYmcKubU=fCUtT*z4^ z1vhIA9ZQ{qBQw>*$!L`p_E1RsoVEUuQ_k>WkG;wcciX{!Yn#<_=WXQt{yra99vmjR zT9aP!P(>4^iIn;5-tB?+dsjx`%~j7=Tm_EpO+1pw(a6^xRCFFl<)*X-6Cm$Rhlt3H{~ zT6Sl`*lC!?b5zdl-2VW7>|BqC+M=gjQv*dwQx&?NX9hK=@xTM#ceBZCl=EG62M?|l z6@8rxu5%d#kAE->A5;%DkJR^VIj8gmW|p-2M0J%A2AY;IyS_)UdA-||nB{x5FxA{> z`hQYa(o|8~C~B(Zl2G8{&Dj{oir;u^rf79M(%09+B$SMBYo&X)sBk$ucr|PM8{uZE zx5p*wrl5}6A=yJQ^}Nmw<106LxxYeMskE(>#Zhsps^+b5m+>La8)wY#vd(yP-Nv^v znsiPe>nmoBwnGdw-Mv(RG75Oc13~Cdv2e@fRYP@Dw|SKFyBm&R=%w zB*URSuTs}nTblZzUn~r%F+j(E8cxH1vb#?YbhW<4Ml0oso}V*&&L?ILyrF7#o*Y+O zbhQl?-np-Wl(wRxKtLe3bHA3#t&d)d-lv_=xC>KYbabHXUUP!+<`kKGId57rlaevK zP1lt7r(AU9x+l7hCSq!a}y?3+rj?9Vm;@V%@>MWpvvlI%7w= zs=m0%Syxjdnoln->%5!e7Pejve zu-lHN)$&^Edub$SY_$_T*eq!tqeq{$b{nYCkAcBv9oS1?mV0O;?lf#R|8!%?VobTn5AqqR$?nsRd2f^*BC(55_@ad**inbw1< zZuNSm=?%_V_Hv87>yl zr06VGH=0K@@<`y!j!!OSf5L7Y>MBh$SSxD-*;$sy+8mzE!>IDvN&f&79qD^xTQ>;w zt)^zPRKAGE&)%huM`x6MvOLx26m`pOmDJjL-4|_DR7j|o!X^>EKb&*2!fH#@I%1yH zp=j%S86f3+@{_rY9PWJ1!B;dE$#tvsb!_Z0xq+;wAQIDz^JCl)&ftD=2vFpaM4Az|-I{r6iPHC*@F8$5R=)l^s3#u)=-?zS_Nm$><@3sg{QEkmtE z6|!kasj4ofF~^Qp=QuvhVR{1Dp=zNP%B#Hb$}NoTS{QNO18{u8bla?>cPTELUYWDo zG^U>v@nMZ%GEFPDFL#TI;~M_}Q%$UDB@CsJl$2F?qzxVCkT-E#EV|B@QPka;OF48h zMrmV@o0PDPobPVqs>5p7ZMGg58#M%lms*T+%+B-6ym8{MC#bjRV+=Zb)~QE(xKZj` zZI+JVS2S2Aa@_K?^}$=e4z%?Sq0|ypPfa7Ks3k6u?{?J_e>iy#t3=?=m%m(fja^g~ zjH{-KFxe*zEn{Ehzf!u};iloPpS#`b>X{{7WBHGJZT1|kT9{s~$Y*A>&v83f!=*Ir zl@!cu)zS{mnjzwG)Sn=@gjEk^_O9zt`iNa4Ow-EUxt)(yt8ryTuT%>^X>)7YPVe6( z=esFBuo>vI&NbX<>Z;l-6)$XX#^|Msmu_x@80w|ZlaoH3E;3_0r?6k=y;v$MZtky^ z7-#N4@_C$kG#_H^oG{ar_G?F{%M%5Ca1WWse0el`FuCsx^hN%+htX2ky|FT*z05`) zK|OrOv2D7$S6ei#;?1VzsA~X=o#VuYj0~S@ymImP(>d!d(SqWZtZmBiQKle_x`mJv zQ7Z=WKfn)i(JJmWb}CDBHrkrj$x|p2#~Z&C(Y8wH_S!o=H67afUr^q|mE=f9T^M(o z;*d~YYju?)X)9)_iPa_0hO~nAyP!Gn_O5e0xw5|woiiDyP^7j)e$-UD#Tw<#?{QR#ew|9AVLbbD0`9$XVViQBl*_Z5FFtWOOx-X^=$5?;!s5uJA_5 zscP>RCc31`Imw;xZ*`tcZ?$>xmy%lRZmXtnZpTF|^?HUl9>C`qDO&PREMvs`N|TRz z-09myR;mtHR@S?f!ycm#Uy93BvRx0un~h1R0(NuET<2^Y?*aYGOJz=-)w;HZ=NvLu zGB_F}5)SAr+_*Qcc02hmW?B1qIOC62TXP92D(pI5qTx+dQxxx)=;`7+ydJ0EwasT0 z(yLEUT$w#Y#PfEJk;fyVx#GpfI-N^>b(-c4JkI5XjA#N8o$#sBU%P!UZ50z2!G?yM z0Ow<=1$5x!%lV6zuV$A7l1XBgUfyzIc1h?x`C6W+n)_EP-AhKVV8Ni?0TuJ0aGj_K&JTtBu{ytmSW%n=T0 z9PiK;dg(jgP-SQYmu%l5;1}KWm`rf@a{N3z*>~ulmW#ybYN)DdOwu&AQI;{J_kHUp zW#Q$`^UzgKQxoZ>EquI(a5}5KS4l15^wM?!r9fK0UaGP3vROq1y8$9QL{TKyeFVVLb3?#lUI zua~QnrhXot9CbF!_MX)dHtNJxL^{adiks46{G)y=%hc?@WWD8GlAkq-6)=OmqvEG+ zTp(>IS8B{QcL1^$Zql;r1Pt+TC)_t$zox^C83(&7bK#6_o?5GqtjNb^)8R3X&4fmo&8n1YvAg-=$OoR2`dpvYMyB$iMZyGzy+^S zdii@Cc=U0$4KHtXzeqlIu)12DyJ5aDvcj>ouB34^bGg{T1~t&>3{%4(98vN z%D4k}a|hfPr!0E0#(X^Z_P6`FNS6NCnj39Ak&mPL6*P4<{JYxZ=K9#DU0hc}Mg*+K zl9*OQjBR_CZ6%5a+&__}T%4@#%bn0o!>t9((d9I3@*yq9}A*Zm&`<6n- z1hQLhcM6MfWYK`wm}2YqE{-K{izDA`;2@QuxbBAM9vi7UUF^8sIS{V6WBgA14<5l~ zq>fp%okelBmejUZwT$d{@bFzANZwol7zybN0~lc*Il^P9=B>rZq^P*a&dK(tfJhk- z5#+NvnmKLLQo%=TGBne)xQKbTT@XsaxeVLb6y38)!;20)b}LxU-Ko>v$|)-A zXm|zSWYeEDcc|M_L4KY(J867ejCRoEX9FvycFY|cAHXMU-T8w(=c>(}yX18Db8Rsc z_0`o9Ra?6;6b^HFp~gzZipzr?kD49V{*JouY|!8^zz|H&0MF2t_VSY(w>h>lTKh%G z(_2dUO+yN)MvqDbkA#Zsg{GkU?lZ%p|(I z+T43~az@nLVP|FYQ;#9#&tc@Ubg|6c9~FBor_xNsJ)Mr=dzVHsvpMd5g&x&_0rY&Q zd~B@LzDtd+PAjW}PAu+m_S3Ab$_ajGpt)1qzLTIL*Nc5c~e4Y$t9h^FQTk@;SHvDSNQ7X7MX zGgz5c+?PV_{{W&^o!Wvx8QF9zd~L6{$w)?9G(>&)+w`tuHB1!sGQjtM_}*IJe8TU` z+v;O1d#`!@sv=A)0bNa>#y8@l>Gb{+Fn#NnT1dz?Y$}3&oAg+}PNzL*XT?1_oAS}^ z?NhCFrf5{1w^f^5=J>67*BAiWW~y>rVJ+HhCs^rPMxa&DPZPSfMQwkH^zMe_#*q4F za>nZ>s3~Y7vy;`JE&dG*g1%dZ{W(>Z>QIH0Fx9Zb?x$T^-1FxKaw2gPNnpIvLQ z($<-(W2$41aszc6mX^LYIwv=sl7^0EW61d|Hn6yM?2;OK@C&(J46PUr z$JO>P2dm?|?DX;7M#WoktYavyVybD{A?*X7xsBu23)MU*)i7$UAxmA5Dq7IWhGW%m z{j1762^61(ZieYl>d2&wvN{KMgs&rPXWF`zn^aqLs|B`&-iuLFbB)hvIpqM1w;o;# z;@qzueL8xq>czA1J5NDJttl&66U9BnrLJ=veu`#{f*b9^(qz5fn|GGiQ&&{lXR+9r zJchB}4Z`i3vc*$*tGC=KUe3}O$Q;LEk(}XqzlfAK_Uj6|I+}q=IFIRKWj=i3I2|ZTZq&0kZJ!K`92_r3viKY2JTpVnI+i!#QtwI5d*i0y>f~ttswu-Kc3Qt^&3~;zZDMWz0F)u92#= zh02c2dYYaeN@S()3~Bm7oy(h^qJG*i|Yk zEn05vHB%#sDr$(!np!Y=ZdJ{iX!ZX94<6X614)t1W7u#(V~zm|`fhBq!r`Vf!&yw+ zmg{x6S?(748hXm-QM~d=%1A9~=w1HhJr(M%&vB@_+gU8NkQSH8Fz(92?ZEcoLyy{K z_d`!ZD1x14MP}<3z6Qv;f%>eoNE?l@}#g6QC$0?QU z?pBK1exb0(IBOpjOhcYBI%IAa57ZidEiJ33pre|2T;OUb-OOAa8Q8F&=Hw8zo0hbo z)0Vn&`J-f%l?`D7#y4zF-IbWs_Ia!I9VVcssHAKa6nu~Eo-lfHFB1jKDH(agJgxd4rBT} ze?@ieKdR_!wM_bY8o1_d#vawJFZ^EEa0k!WwY_h{?Lf^1aipVjrOCk&fEsNO=r|`R9>o|B|NQk0vCh$&Y9n->bpB7qq2*QH9cEdO;$^q;^5xI^<1wt z;m`ZGc;<0{;+Bu4vtMGqQAad%)lcOuh1u*TJqR8}X?Gj-&Rd;ruCAuLJ;NK^OIS~k zU2dADmxx+wDJ|6wbxp3MVcgivjq~;u%XssnuJkur8@yF-ihz*W{;ofP+IosyDHlwoHwU(K64NE!TnVitb-JKh} z9Mj-J=bBjBFC06gx6Lagu7+Op&u7>0Ir75E-L*s-d9+p4RSj$u4eXxQZ`w5Vc0G$6 zT=$PHNN0y!+nLo;U38P~RB%*O$5lu&=0p5@bKQ79YaLb|&`y%0-#Z(!1{ob-&J)53pRR9}P_~0tvYeByKUklY++ayFpS_z;$Jzyp?pvnx-}! zHdYR&`axVlU2JBjH|tyDY))u0!La5eY^^PtmXhmJbkVi5m&tSI^n6LM&sKxx6N}Yb zE(w@kTREo>q^f2mhF0#=@AF(!Sm~Or38>?31w}!Irb7Bxu=5v>7~kz1 z;H{oosVqD`sH38)h`rJ_NFLM3BWODetc5G#;w>Gr??EOmw!FtIZakwWXJmMT;8gub zxrVgOZ?T@-|+0i9M6DG6)?4=`hKrFLtVG>rGc3FqX=d@5hrasO6l|vlk z%LyHNcpnAdwH~^uu4B1oLO6SATx+ zY572iRWZ*ji9(?+pFm)rIwA)6*1$Je1nzI_?b&fSx-?=+FE)^WvL{& zujUThd>`&y=u8A{Zs0UvmGm&=na-!5EIL+NYKpRkqDzcbO$F>AW3CrY;X9jj)g}pW zL+$d7#?5dCb%#93a-KK!k|y&r z>%@Au+^g5n$jYZg_WC_z4&DH(aPaxLAM`bB`p+O z%6WCemwmHw$fzZEG(xU1+4O!+O51GrRzU+Gt*B(DU?j&ylD}z)Y7CW@rt(xc=2~;m z`&zW=X=-D0n)uut)3_sWycy1iXF;h9nw%Cpw2Y0gRVrIsJ4nQJK8oGs$?ajnUK8w_zKdok=FRmLVW4=V;4IO5^G{ zI7K^{9VNpPEtRxMt}STEG@+wSr+1}R6; z&z~db>|RvTPp2Bd%dB<%Ws_Z2U23SD%T>ycvzMA)epZa(@wcYx=Eb0HI+sMz$6&gc zXy@ji(i-v1d5-LG@LoLUUR_yDy5scYrZ`uvmYwnT(zY?}>bo(IM)5yP z7fVkaL~d_a4q0J7{-JTIJsUp`^wmeRfVIM`e}A1-6jZrnJQ+i$a6;j_=% zQ%b@Gc#zY={R+2h}yKvu$ z^|s0xDYXS7=Zcb15qxX~%?I1!6m@2psn_~q-Yq(ko`Q~)c5yNy)?;jC-Z=|fqp-!L zwAakqs_hj_%{=|AQiAE5or8U<%zAN`3~AMk?Ui;7O``OUzo2E#7+W1voAgzXjz{~7 z@gZ6| z?^fH*y0W$O^VQQivB2yHf*jlu^Wbyhv!6G)b)T8!!wWn#ST%N_wAIOViU*HHkTu1? zmk@f5{MTmFTt%^J1`9pvgwx9_0OgsPWSRX-Lev}^)fWr-)HK)5cRp^~N=LkQZ_iLM z?C@1J6m)ua=TB29mY(t00~1VHV=!^v3s){qJG>(69Q#(st@ve0Z>E~_r30zwVCVE; z?&%r&r@?bN>-FXv1A9+NnzEd|vooJOq@IO!JyCnEtAfjRw^W|c!sbaCb|RU@z9ZOH zzfGq<2`zVPH8p!^sxDGCX^SzVBQl3I zz8*OCu8uze_@6*ymXtcJk@IfONutlRZnO<0!aLTRxlC>~Zn;}3OGD0cBtyhw#bxw0 z-p^sw(p#uqD`~`!j$U?p&d%6(kX;|;c+LuIDLFhHpKFS^(>00nP zTjcewS85CN7uh9%IpQ*BV~N~L^QPfb)Qt27mJHE#3Lp|Bb!|L7_ z*XySn#iI8$D3Ps+k;Fq=7z3_8E1o$_^3#7-GrMTfoLiu-wo@}uTjhzh0nIP$^3#sF z3iDBL-%j5f4x-b1mioRz_{(;VO#PwZ_bqyD7Y&a`wfc}$cJ};oIsuU*P67MYQr&xs z_n>R6@KnB@qMezJTy}h*J8#jl-R9)NblKU@+s7-p<)X=7si+~Xw_K%}^{hPcG4k^c zQR36&sObJ6U$o^#G_t_?>1rN%q9Y`qw6=cI8k@X#3YlPKL+FhoLmnImaO>~N#M<~T zsI2!$Zg%U$m9PK{;HZ!mxMw?iR&bK;)p|3C4K}NVw=y-deyv0e?B3k(xba*! zh?Z?#qi#|eWi>P;c4Br*hP^>jnouYTg6_SWYRvBtPG9MkCoZn5>L5nnwHoq zbnKdLuHor4hNen`TO<4Amu0Q{x-Oro`oH2vCl;>fhYyPNQXu8J+vBLEY`JiF?|yfR z%H1ibFS@n|PaM>AQo_#5(Z|v?p}Ti{cU=bUz|pCtisfXyNlMl=r?UXIlD5R7CF7Ca|0YmJB*H*3mI{tf^9-B zH1zS*GM(;;)cpB>BSy!_`LCnt{{V|+gqY{u=H>J_<=+N?T0K{6?E11CWBMgpSJ1&y zIVEFWPlJXZ_Z7Q*!pnsNDLW>j#t^4_r|f+ zXWY`pbGnW_s^-4kTR`&Ks26#|5iHy$p3t(l(NyQ@!bz%4$ z^9{1Oo1t@LFDy*(-QbxkZ67#Xj)czsMuLB94Bd-D4B278uDiv6W&o6YI1mFKA~ z=(#s!d|K~ffQ0jI(pCL`FTw2K?1x$DuFwHnlkVAB>OEy^ZK`2xt>&uRvmkU_jn3t) z5OQGhfGDh0bmc zaJ+eEmHfNoWOl9?aTcP_c%-Yi)=k;6cDhbqtlu22%fz3QUAC^*B?gvxiPAR@o?f{2 z4-b0r&!st$oE)l7ta)xTjQKAnc-xlUvE-M`b!&#Jpsj1>xnE^{L&yR#>Hs&%^H-X8 zTiFJuhN|gyFnzsT8BHBWSnfI87Cy4yJYt9%Kb;VtGBx?ynW#vmfs_- zk~(^pHw`_G45!*k=6po?L*8{1uBWwEveQAhzn3^y#xl+S7Y|`eJ3_FzW zn0kC<)l12{w|Ak}T6T)9wO={5tbyr>+Bc`$lss_8vGIODA|?$=U$h{eHu@4|*t)H-spmZ}5RK;zoK@9+ytuyF?4YPr)~K6^D%hMv|r z*_IsjTqk!wlsCgCbS)rwt9?EecuC7=hoQKA#LFC7k@ISM{)p`)M#dr{7FFaUPzKcwTCbx2-odsm?~w;L{c!th;i zxMe& z8-C^Cmb8t-U7s~9(>0Q^A`V~e!jsbR;g?h|HOhB(o9__ceX6+KuUGjWT`Lb^8*|DJ zgGZ2D>Rn%`9`0vXKVz{AUF8|h58hP#qzv`S?}ihPzAkr?<=br7)wk@2dUJsu#eZ+s zcVg)cZN^g`Hp7VD;-_EhQvj4lJjVRg%sET*Y4uFa=7k!G|{OI*vVDk}&*u&Er) zX!Z{^eWUoFXOCI7s`aR(j@IXw7~&fbaPZ3Wgprmz4g@GhGVE`a!-hEY`DXHN?cSfnnrbU-)E4{9ZXDdqpOp5w$4+IlaBGTMI^OvX zxV}p>+WfVprNRF4ytz*1OU?scISx9iv5b;1?(bZ9tmWx=?mCfu+`eA!)juiS4?}1$ zX)-G6$RZ%<`6Fkvg~2NLn^%9kPjl3InZied*tRT&x7qGoM|@+I-+4}4obr*f>H2In ztuyY~cJ4ZZiW1jbsp%;%Fu=&b=VW;eb_ND_+^)I94l;16(Q2ovzTCsEtYFKGIpjV~ zUN0aAHJqn^YM&ke9nR;OV=qa?XKALr{GP$$ugYSU;X^I6Ma&>%jV*m_rw`xkg~AQ` zlA@{A_4k%kR!rjQpC(T+@?8Gd-g?%vlbgCu`}izY+eqD5Gdm%lf*9c!A=sc7n z)HXLB-fNUklj;m?)yaD>M$w!fHz}JFK>S;=?_FoE)l6`9AIaf-S}6p zbk(BxUwpX7=(Ee2Bx`c!Gwh76F!sBU9&o&nhS&t;vP2rWr^iCg? zM+mJpSmCu?+aY6nl+eY?gYEF&&2UXq#VK`mv31h#b&4ujwxxxhq)qk@KWg)pJySW$ z8ykV^)m^e+a5;K=R~&e7#~z<8-0zpFFWhLSsn9ffdec26B(RbnQH+8-!0;-+3b>n~ zZS%x+b#or*e-?QQn&)TMc6ycNfirM0++kCOc;znclY{oUCG`AuYeePBczdH?e5EwB zb~)hEHG-nx;mZ`Vvk|lT-rsWb{jXZl#a8Q;!td!7&Z%tkmz;@6IX~|5TyM6?AOnH6 z3%aRxAYrE?rt4nDHfocTULDr$#P{4aX}q@Y6DdI?Q*&%~#_O|iyNx_EwAw_MOR#0y zPsY2k&If~Aj|JjJ;B{tCj_Ru=Yg`&)cKht&6uy^^PSZ!bZ<_VL9C+(Rs3--SO5E8E z1fRmDc0=SJWO)UlSa^?XvKd~>gmn@MUdo%BA42St@LVO+0tqp=pJ`FIhQ76}^l)>Z zSmj&ku+D2Jlak}%+O)S3JsS4<%f0S8ky!r#r<9MyG@bcJo)@C5{CD9cn#mn4y68h= zZqYI`lYM`Yd%PZ_pRmW@wat}$z<&GzYO@5q^AeGg6le;`%%J;iW`xDY;t z#Qy+EQ9PK-qJLkst@PMiyB2fV+3h|Rai@pSMJ0Xy=*TWlcPj%~md77}ZI)7ZYj_7Wc*O+T79Ksr2d7Bw0a?$67#?LFAlMA=; z=2WMWD9Zz(cYW9$;*-+x&h{t0UiH!Zap5(p!yPw4Zt&(gUI2mo@#;o-SbK1Y+tGDz@R!;10u$FgOf+bgxD{{SJP zumRU*;brvRy1B)8qt%T(82ueXYKPduT+VBj0}Ks`2G1+`3W z8tB|e_U)BT^>Dr6u8G~Wn5_?MOCV#g9zhZS?(H!;{`O#g|O0ww;_xG1JlH>8a7zmcZ}rt#WLcrM|? zJ~8mt&81?3rt-+EUQCrZW=rHfclWOl(Y`-PXV7QfRoqpLY@4;4XB%`@`g}2;OeS|O z&rtERj24|#=q+}eglx4j;nKPL84tN=_sxHAq_9&_>D${Pt9*yjXK4-Y9<3jCR~VkZ z#i64E$yi@1Bl*?Q_94B-p5egglhfgp-CVgkdq(q7QE7DVWH)#_DKJknyV>zkg5yNl zc_h43b}_ZwjR)%=4EZi~^iCu;2NRF9rK;@d;B>F0acKT>^>oXQwb{n!Wp?|1rP6~$ zP+9A5;8aud<&6Nea~}nQwb=CK+PG6tU1Xi-yx`0NH(a`rjh+_B%6H)JyVq#Yc4~bu zqvgBRFfy{Gz0h|ccm(v`hmWf3#cwWdsoj59M?q_&7Ry_ucqcPC=OxF zXG%R|;HzkgIW938+0Naeardo|S|&A(nUes~oQ(BfL)7uh$2hU@^6=Snx#Ypytyb9{ z2!p!K&U_XNTGr*>j~?wC06Z4;cW9xZI1KFw%iKIxiEXyacfXF&r?X+^a=bh~#}vkS z%gNU{c{;C%$myCkS?q4DoN~SG<|ppiU0S0Qmbvaz;LT7uzhGC_SUoD^PiTiFjKk&f zz9X99}(8C#o>5o0i=XSAyhJwoJ+)JmZ%Zc&kcR-}Bf6RJf zS&0GRh3EQ3WlRpPzOpEeV5e_b*6(V!(SMyWDnrUN^Q=`!K++qPE*cF>Zu=nitCbHty|uXC}Ki56$yu z@lN<4x6hx&IeHbg%WwS#5&cy-wOdd8${)=NUrGJinY{7n6L`c_eF9&}mFR}q?Wg`> z59S3CWwsOi0uRj#Kc)Wdbw{1O0ziKfeT%c{YhbbH`gk1B$zI;o{_?hnEH>B38|V;u z=aefgRL~%M8ZqUC>h%0mid?IYCF{y!=T)uripR)z-DY&XD1CU4w=IdK#a5xVzm=uU zIPo}J1sxkHEc49Khdh3Tjx0Bx>!&VS!p4a`(el*H)-}b>kb#eS$5qqb>L!+0VVE(X zHyQC)aZyg@x@g)!<|J>A6<#X^{zrC=bb+q#;7YTOvbSztCw94AQ(JDTG_^00j{p<8 zg*3JI3b-h!=akDM@HbiNjWKkrs5K9XvN()nyDKTAY!!Cv;i{e%wakd}434BE;=Ny; z9bKA*b$rxSvQo`cDUvr~*gh(~uBM{pSsfib^14|L8ab8Seu=)++$X8*T12uJXPhji zjncPEHR0DPsGEqQcOWhYyj1q`xocgTxlW(%)+Mf*ioT+frlw;gQWhLLWqE8D8pvwv zVXBl5Ffjjy*tz}lLJ;EavJxbbDxQsqAl zLObU%{u4LnE9fMwt+UowK~W%OFw2Q|MI$(WeaHp|=h*PG^g4R!Z_%~3-pPihvRo0& zIwlzBzUJdpiu8LR4P`+gy-SttQ>*RP+G@(mnfWTR=)if{Db9zyvLIgzF^%xxv)ac_ zT<*5}TojCJqo#+I$D13Z!#6E$P#oEss>#E>{{`RRuf@K8FYn?*X}29COv&T(X<&)#A8RS5!kuOIJT@EPyr#cf(sPS@LMZq%~I1hr7|uyX~jMZ^2(ZXFv5>$BCkWhZEu+4)>h%U5!ute%pl za%rRIcYDEO}uzVo+_ES#M0)*D>+8J@K)^#ZhQ`^*wbw{fNYG4nIgd z3YQGJ`%cT%G`G1_RoX696qU6z^R>gF@G5Id?iX6wn?+SIY!De9((eGa9c`yBH*G{} zX_&*9fN|y+&J{fwV7ggvQcF`#!4*UfjmMZPIR3vKld5T~8`x>(HB*C}5zMdc&a1n@ zQpy?TGDR0T%ls6x=sVl0C37QTVR>^%cpXa4=sh!cvfbsHswQ?4!ee|=()s8Vf z26K(icqPLv>rG3K%1dPRw+ZBimT9DW8_mm}phNQMlxe4)A==A04il?-N zjbV7@$dd(Bue0hF9@Y#S`CZm#8-;sL^bh>ylegU+E8fXqbWdkcO4moFk&jFm z@xzr$8E5o~TImCva07|zrMT;|HVfn;<7cXmlP`z|wP_|75$XlJXyWRM>c}O@9Jo9C z7B;flAgY!e09x&a6Q^@-xZNbvGZ7F3;IOR=iWVbFswJKniHfbFR**K&is%-aXTwtF zmhA3}WvQWtve?OGfs#HW$^(XQ7WzHRbxupCoIM6y{wSxQ$@wGjUVv>425{?q1Y@XZ zRRzA*dfVar(7Aq-`?*e6p8W<-$sdAx6aN4vehbj;wc3c!ph5Y8L|Uy!{GSKrg+EFC z+V>uK6mK0EehPNb$bO+-g!!V?{GA8pg%;WEU&+vZU|GJC`>f5#_+Q#n-RFi)gV~S? z>bliCTT@P1j-zAbRak9S=%(!>p@EM3LAWaV2pHU807sJaabeLk-OgF2k*HyPC0#w$ zt;QT8pV~YMsb0>;uAbo@9UJ?SWArVW@o#NiJdwG%ie}^7R=Z`5)f6+@szxM)_HG?d zu&-7&Jpy=;n#wD@aoi%1Itk-`2-$8m8phvV*dF@m87Yi**v3o8)(5Y(a!qe#ZG%wFN+5S) z^Un)|w&f*aVXcZe3q)=X@$avX>Nv6LyN^a0*&g%Z&mJ^vK_?DpH{B-KZqos+iRHdzW@A=_=*x|Z*0x_%-~?%2x@Q13Ov zt8QGj`+qD^GziZ|?sxI?R`J~9$Rq%n`ffaP@VN5Z9{jiQ_lET}Et0D6+G$A3<9kWR zn&8|)*7_#rU2>zh)PruGrL_>tfY&kKTOQ7f&y_b8y@cZ-M@V{+o{*yia!PRqu^c#*P%-G}T=_UF=O9k)NXx_w(RK zx7xh(d*U^v)y=wFE1N@bZ7G15-GK9*i5}&|pHa-;#U8mEt178t4KH*~AZ~lzQ}pvM zy*i*voa&ini+Mn=u}W_nemF%&r3@(LDYv z{85i*AGD!U~Sw7ar*U$j=IrsMM_nfG$9 zpd*i*EV$ZT4a({o$yv)R62x2x#^HI#UhADW>$~F8cN0rV1jX=A(7bs#c;mrwBQZZw zzE-*6a?cLwbK+GE)+z`puFY#HqHu8y-OfE$-@zU@)9GrdEtjhQ z07lqY>ZmH9nzfGSc4Nm+ynlFD2;D$q@7yW!2Suw%3}J=FGqCVG zsVUmoq7n$*k885KVJ0(DGjg}hZ&bI~DKza|6w@%&^YcdVIt7G&r@cH6-;QbX0}C4j zZf!KA0s(H=R6}2Z%+faUT{^PHGbF6H;SJZXr1k5v>D(UT=+lu`(&{?i(wcSv#*x#_ zalpb)L+`@J-Ea#zm|Q&94qftN)0c-%J8-+j=`DN2J{w=(PTETG0FE7{5rFo%bZI^z zXcB5|JyT92p1v-4$>fxEK1+{>#jr6Pd-|%3?8`^cyW&;!+4~cx9}CHS%ew1P>N_Ni z?3|X>T}A;85d^aALduCdfdMe=k|ZQ zto{<&+|GB_#>t0AGh%(}%I#mK?u_kZtZY?HZqhJp{{VPg=)~5RgBy-^+^UUv}7rngifTbAWy37d(AU;nZU)qjeRCvVIB-Q$C`&?0|+sl{EF+3 zH{(fW`|Y}FTDw(cS8YRcW8b#U1Y_8(n^7NT(es3D4LN88fUGqnuWf?Ndk7;ThrHH| zxO-2E*L!IXtns%@1&sP|w;MMamn%D0T9;Rf>unrY8>=RsoxPYZZbv7U>uceU7OS*H z)u!EXj-uO>+EPoNS9`pd0&DX_G>-2UkG9t3n^~Qn9lGVjIy;41W_9S(wm3NfiMTE~&PwP!EBQuEWpp;(Lo{{v*6i8Z zNR^|*W3gUOYFm@qTYnAMfjJz?!d>!($`g|?D zQ)j!B@a^*P0g%@ll@6eERfYM>2P;7G9!t3J*X1u_b+NR%&UV8)Fws*BavnXQc#WeS zGNyE{jyF8#0A~Q>D$Wcs`(~E%WqjAKc-6$t6y5ETR%yG*iu18yf<8dw=2r;3YUfwL zAM=ZzXy~mNv@Jd&50?E6OH32RMzn7j6z2$Cp&{=bm=Wei=W)o8W0< zYou#Q&H|9?Thp=4T_b?#3C=z}KIMfu&cQJ6BD?$izctUNAH#Ul8v5}?Bb!rsd@KXI zzQp)~x5|zv+bW`_5#AdMpkV>biw-3CpEbZPV0Iw57a!EDn(-TXugDP1 z%e5yZmvPyxd`5;leO)z0m5a#Q94+FZ#+^H$ozN^xG0E% zTs%7aRMEG(4l$mo`d&TnN4=H&z1OCGQyL=SY;1a=rix5F;{8yNVKKU3PYhk*^8^eoTL#g3phN1248r(V9^FMOb^q6ONXN0`B z9?{e)_>pU^ir?B@qofqka-PaO66|>S?zRmJ@_@Z*>d51o4T_3rsN`@%g(Vsu)DrL@wKgA=3Xb5F6$hH7nn zrs=heGimGW^;X#3&6(SuxDO@3BYU3EPDt|4L~}WO&ONF6d@fy9ZXWHP9m)FsiM6#C z3On@F)oz=#XJqbHwE1Cl&K~@xC7=S6R7o9t17T4a<&b;)*O3_;i5LfQ;1y0;0U2o? z)n})|UeNa~^7jU><2MEGmgvd%O-v6h}pQKK}$h*pncJ@MOEE`Nn+-7 zV<)Gyg~H@)gyWY!+^S)YI3OGYw)?ElONEzk_p+WH^4rB(Y*m7NKdEKqAQ#6jc}ep3 zF3qDpQ(9uG-r9X$)>F(zA0#84j{d1$8K!$>Y=in3tB{ zx$15)@e71^n{%pk#pZfziRlbXEQtGY=DEeoh?JG~t6dqcf|$`U!Ow-9vOJ%xNb0!T zqbDS@-|kc?+Gm5bYk5H=V?IjG3}o@$FD74#>hIRERJ7ARzqV*`Et8WDcfTGhORLnE zV^tt!VeG?*ZrkBvZSc5G+>!vwk?kf&07)S3y;xkhrm=F$PD}MN!&2Uw21I>lRFlo~ zb`TK-0hJ;h6c9-O=@42(l-?nsH$#yQp?3rkMF>(tFG?r$-g^&Cdhdi@q<4Jt{Qe)_ zFZbDV?%nL{wX>VunYrfz(Mw`|Eykf-wO!YhsuIjYLNX8aT8AQWOmOA{+5Xz(>!RMU ze*pD9@v$>US9w8XQoDcKJsBqa@}%>8y#c8#>-zEuKBuK4vAyd_CeQTGCAW>o%sj5R1Un1dYN^`@aSA9bN%cLpN;8IsdMhuWdJ*bQlPYqxj!To z9-#|ArMnSc#E!xn^{G7@MY&V0o&73>?@!1sY?ZCAqh?OP(>7$VP z{JC6L+A}JMy5#fvUlRsQ#_kSQPH2Rd=cn{jxsQLb-ZmzU<-n2=J^_yY+jX%*CEK%w#n#kKkjmeSbwN zM4=8V#v6fJyGgG^woa$bWjJ#QeUc~xD~cIxSN+^|rGDBO+=Y|C!30I%WB@|DD5Mb^$ONq?u9mlDOdUW&i?@3 zA{(8sb^7iBr@~HZro;sOJ#9`m=MtE}V%xg*Knvz4IsmIbGid66f+-TEl$mF;Bro%- zS$6HlKsyL0?xG6*kQtzP!kpm)+BSh(KT=0|ge551YrNE`XXAQdggpdOgYH@PROZn0 zbnhegv|;Th<@7k$fPPJ|ll_8xz_zLTHq6HAjoZ^TOEK5BhB6UiEa8<~{yp!TOwcUP zy;G^(l>DKPWA^jXyTLeb;zEDjQ1bxGI5DcBuRcRfF;&Tok&-Ad15_ zk8_(8(Wp0$!MG;$#dyPnB;I_?f9fy?XF zi%OmS5vumLEKGSHvYKeLY*LloLsq)Cp&#AdXHgTL@TN$3K!K&PxKhxMpFuQqAa)!c=pB5N@-^cZjg6Mhvf!Tmti>K0?Z%<_6=^pB^h zuBdjIJ5^0!F|BtEgOtvt++H%GNH3C2`vgct(;R8^EK%>`ys$OD zHjp3PVNnpumW>QV&ND`*nzpK+Au!8!;uLGXr$f!924w~jAG_Zzk@6I+u!zX3OHkCt z8pBcjstg(D5Xy*OJ*gR23R}qb#REg$l7ethw{?5kNC7t9t!S}RE606Ac!>D9Idcl2{rQ92tj;GbX3pUqu73@SD!R_~LkI=_HqZaIEgWDatcbZVb0J1s zOWW1V(b(BZ!C#0xcKEKhLF$kwt@7!sI^V%#caG@dCTd6M?HD8QVB(-y?B-pB>T7wh!#bniIJ6ypKN3MP$N6mkwU)_E5BY@ua2>x@=D4`x119@>;4!_+ zPjQL8-NB2^M&~#h!D>+OjqpD0^sai>rJ#gKmRUxH+nRIDwGY)QY1}ehwoCfj(Yfm@ zOK!GY*SnY+YIsIN+e@oCaLyTs%7#Q>wDRapKPZG-!u`O3X0iPbi;`ph<%h*t%BR^P z(6F18e}KKY8dkkU6_16L*~?(4Ma`(kKLA7Lfm{Y3T)1^vky7wB6c#u#Z6*KOqiYLd zhh+DVUAc3=x$wQR?7BZv#UlWkg;!T(9ZAxvOX0VB!-s(XDMpqz&D3jV1Ub>sm z_x)~DRL36`@1L7-{oBd`YvE_AIr|Tgpt3l!mm}|3ZVP+*4-kfB^%4$N^5?gnBXz#m z;C}59oK>Ed%7TM}{_RMIKN43;UA)2gZ@DeDN zaVhu5>_WyluG6=S|LaE8;m(}xA7INZXN!~y9GqR)!gRI4ss{-!4WQB4h3D8}J0UKH zpc}O0870&Fa|M48KjaSwZd&>wC-v@2IgfdRU+qgm{u1$`^mfOY__w(B1?9!=!03|G|sBCbgFbbv8^&bda*BM(d%QjhiB$|B<*OtHx1;=K9mws~~)-gKP(1FS=FE%r1%?NdDuNvvM`4~1K{wn?5Kt-YIU#ERv2S}j~B-RqQ3JHkP$ zE==*7GdM+1z2zRI`iZ*(lh|qAT_)$;kT3g@9aP)%;vnF-&ajgH4}z@Q;gg)BlK*J| z4$hH1_02mfncb?wMV~>nBw5EIHSqReH_KK^*-FlLTFqaz5d%Q{@3h)Jb0O<7{}XNR zmV0KNz3tDRQJodSfqSQ<}48 zGKOzB;rRMGyJXWwr9bseJePYf=`gkN#coV+C*HW8&gL z?v8o9){+H-?t?v|ggU~Y0(XA;t~4_wj>h}_NHZql8THs*ir+U>j?`!MOTc+Er~aQtF$z z#GsJylpV(DZlIu-esHB`ofT-Yf4?T7@0;J|j6V3X-_J_A!|s2hYD@iZU(BEMW5}x$ z4h{%9{!)*dS))#q{kySThBs{owCDcT*ml{c6}SlgwCZD`?<+#F-4`8X+n_!F{(`f+I0UU_@Z!n2 zzb%%j75RJvA|iHe$xSU1p7pl3mTUmXR3{UAr~aSzEaDbX5l*#pmL?zjZyn)&mS@y} zj|oeNqY=lp`MZR_x%mYV$i3Ur)4ogl2SBM>`>#V6nn`yB79Stq@kU=;z77YstBRL|`{K*TmM3N2n8aoy-syX9HmS^?mZI$En zRLTaRO(U~(pT_95t;i+S#2+3x8Hr2EExXR!oNn1G+(C7Jpc+4dYT0P4oXK4Kik}Z} zNTm7`_a(plx})my=|QRqOB|!+Z$n}QtK0i_<#TVeg@<8s`!M96ujJw1@SJ8C{4?is z3~g)Xg0R?j16m(ZN+SvXOv}+`q4Bw(Y?6I(o?Y1wlafk^#00AZ&L)O##jQpHZF}cp z$|JsvN;g7ly@w1B4f?)S)tfh63ry%ft^n_gN5QC~zl%bvYL@fsoV8>g&`&U;gB_l_ zm@Ctd9TbZ=&OR2SOcr&-Y>eqne_oLY6`+pNl_GS|g7e z74xUm06PxmGrV8jM64F`K!d8<2CJ8mRiw#QFQ_V9{(w~tLl=iHQK)vsDn+M$n zl;K4&%g!m}@)DvPpoRCgi#w(q3HR$#=3!plZfr!b%4U!{cfZDDC>*xSrtbehwLP_@v>h?Lz6S|8oolVQ<<(UIP-UswNUDs$BK>?vyh{ zFlD|p-L}3M8ky;4H;xZ6%zt!xMASPwe@BW2&q@49<*n?OEQZ2t9~w@!cdCbsYxetn z+pmn6fo_O!wpb0`%LzcJybJJ~T3^JI+VIQS*SD-A7+|1$y>`Xbi2lutR)6wRgh-d_ zwVr`5HSVt=KMGoy!x-v3nY>$6M%i7ewf#+C1TPJ$SsAgQbcWSJGi%Dx=N&;xBTWh!1g39HwaL>K!ti_b@Yj=5mBe3On)B8;GyZrE;Nx#vaa8EV zT^;tC(B10wyiBA=tIUMPxPbF&0|0RULD+ql3(URA#kE)wW(Dq(-LUad?^U zW;BYOnEAbDU8wRFU=O8zNl=y@9Eb28g0<$n@?*XWMTXS>@yOSQk?@KI4SIocar|4Wt92QlK2f|+t8y$Y z+H)vtlJVjIsvcOmG5&^D_0Fs3cV{&C3*`Io15@fBEIyGmBIdvk$OO^Y-Vu$Vi|tJA12~b%!AES>Y1U`wwOmw@6Bi*-ipRb9C%kp3!t2?l zVf0uN)OL_zFOITEY0IFxZ+8y-6~-ExrFF1KF>BXfv>4u;S+L!@XY~(| zS+usfxJ6K)X+6M&r^1tm3^iG)U!~qpxJijV>%!=G;y+j*;=a)=U!iz;N|(mQswGd_ z)7v&(iC@~*2PrF3JXat0H)c9g4l!l z$-Yv~KU(^<1~`GHVOX5H1G7|QK~PJkbU)d^ov_^jgLms^wg4Av z0IBTvW4S(0YN1oDZc-Wy?pyc-7Q75*v}4VK8h^0Le*B?EXF!`c;_HN4bfZ+)d|AW# zixxE1MZJ(}oc$JGkv#Qgwj{BJ+;Y$IsJWdb{`%S4t)*Y3kyJnbk`$4?tuKm2hn^H` zF^Fx98~fdmK-9%2z24nuEL|>#DG8q6sx2S3VQt5}J@?Eru2)f+QmUT$bv)-@i>MqA ziEQa+X1bm?)Zk5jkVX$Ou;7qCx;yn%{frN|2BD1GJX#om;T_A7CMSrZah2NST@xl- zF4ad_qt&Q@gQco{x_ZjzKK|_TSD|C{l94$gFYJ6o?kL^PW;1U;xGyyuAUy{MIHuk# zN~rEYf8CKkkK=*ub8m?G!mq{$AgilPgTzDPb$*oS=HS8A)fXCbRj$kT)F1Y2S;E9>WT4^%WB|J;aGS@=a#|L z8#kV;YwIE2f@U;Er`=G|ncY@XjEPd9xb)(f{5&tuSVrH`?gd+5%w1HC(^>Cqlr=70 z39TEzhqAqd%>pwLIq*aP<;v^uscp16YAYO3iEtm3p{PfksIhR}cRfku)O#@<8>z`POwh~C~bh5oM>)7>v z?w4hKuc7>_m)7TLmNIZ#`C4Qk_{~tM;WYA%+l}kqY*_;nUv!fcgF@{uAydc9T|BEEqZt(T4&}S7IX@F%w2s@gt+Au=J)ms z8dtw>Do<^dum>?$?U4u_D1R&Skb$DOm8*r&mC=BMQJC>0{~1Qus)aB?1uB~J;OHPg zH8lD3yr6Hs@RgQ1ezhrfu@y=OR!iv~NuTqktWzZ#WQkew04kkAwT0`TRQp4SL7CUl z!JJ=AUT=tD6<_VOZP+;3y|_}!_vjY%z9}gZ6LJe-jw+0wxU%`WWNK;M03ZY8IH=_! z4Z<;bG!DWV1@+3mKkTU93OYY*Pdk9Yj=<~NV%z(S8XHTu=DARF@?ixbb^ZO`*yC#U zmb~wb3ei<50T+Sj6TOsh?!EhT-|%Rm5|6W3mft>#*f`Gmo=XVCO-Z>F?m%Z10R&)X#|oeuX(Wx$r4R z#B>J7DQ_5f&IJhR5}wNfFb1H7( z%gS4ZpoyjL?n&^hS(oX+ytg{1Lo z`HY0=-ataPie}%{KftD4m0qEAz5&cG(3B(o!4FnFW+MV~M^h$sy(id%KdvW!eKP+y zXT4V}vMz(k%WW?Og#a+-F?r2dvI-qTJTw|Ay0s0}itZjqTLO7aipSix)m zy_r{9lZz#$5qw@KaiE3XWs!IY8xXJe$p7k4CtvYeZg0Iiu{L4?QC<}}JLWw;=BtXc z)E8|v(@LIxF7!W$tMHoZwB!hSyvsc}4I}rNvNkf~CS3>X4W21|#IS_mMUHXK_e7*D z@ZRYAYm_8RG|Rr?@m3`nA8_s~Lb`c(2}S?a?ZRD(5d&|we!Pd*<;~NNDhB@1?+1sJ zGPjoy#hzpOtRz*4EB|g>8pD|6P0%MIiNnDt3qe{1|) z(9k$+GuPSt4Ns~R>60gH$O=B}RlA^30cVJvS82x}=|>;?A`138v-b>m{{*MV7YcG2 ze2b8JusEXrB+4IWXG$=<$NwjoJp zbL-M?A;RZ1kO}qh;MJG$c9G+TH5Dj@&+V}$vhq}QFI}Oe1J}NUcV@h4)Px(LNqDcE z(^|^YqO3><4}rD^{>gd$cO4z9KiKxM_`vPoEvgT@CxDo<8_6H|1^HCUwh7Q)tDzOA zTXV)f3FhuoRF`U00+MF4piL89dR$-PV>(Qh>Hel5UzAX=dsB7;g%Xh zuN3~S_No=KY9^BvJX?L)U#PG0{%M`;J2{6+Ex2ML{R?Uqv08a=PpWC5wwf%F?1-5^ z<-AG5gc5Q4K&nony)`NY&n zqYGHLd^PhrB-X2wCZNO*-8sIPg&~iAlV7?LHxGll8)$33lA&R*_==&LkKO#AK)_1W z2R*{as3G_M92c(s7h{z^_JHVz0EZop_x6K-!u;h8$G}bDuNI;2k{&!flg!mlXMmc6 zk|>5{HhIiM7_i!<9%Anw(lH)oxH;Oj{kZyIW^lMx%6njnD7ER0SRw3GPdF_5w8G=? zVz!~UHiV}|XEv4q^DXT;MITMrG@o{&q!Rvf&Cd%`q2InHE{AeS*w!sX#dq6~@r zJDedre~D(AFDMrOK=O>@mC!MEwm-Z1O{Z5S%VN6dTl5JHZp|ff*T>0Dv`%z(=?)-; z@?kkpeOh;qQ5@g@YX;@=T3iPXsc(JV?5|2ef>n>t>v;BA+EBj^; zw|is&^GLeiWwyP6!Uy;xZswa)^9EGl-$!Wuc4SmN%@b)ei%3prntwrSlo^x+pR*}F zklS*UD1vF{N!7J8nX!1&1qDs^n!9W=29Ad`e!h6`BgB;FxzC?t*S9Da9aRXtf%dhK zY*VJd@V$x@zLh^^$Iiyjl!#2CFE~zBLPL3@lwZnVVUs-TIV14#LSIqwcUGNV8MhCc7 zIFo8KT;~9f$dz{Nq;jlC?psRQh^$osI}mdO8p^Bg5g*?1^{*zmimwU196hu@!IP$9hDF_kVk3V?5>O>EltJwf{3wOg%6 zPTG$+GJ$d@o?^{VjjZkretbG0>g|<4&}Eip1~@BXYL0!wqM|LM<(Y;?1H*Tn-S$jHhL3q z`=bVcnZr^UI>0vbCz(QoNe&^Ne*eMZPl8h@fW7V2nO`dRLUv_Iw}ztT>J|ET zUVXLlg&>Q=ry*WrWUfmrb=I#F+uU?pAgw4f zhO(gp?I)3a!>W+qK8tFvkNf5aea1zE>?yj+i?)lHOV&7gscE#baB}LV-2TYHa=wSR zzI}qnY7TyTsK$(!iXP^auAeM&1E!FBnqrx45!n{Itez-^;JLrf-U>r&H{9a#8d_m3 zH@8vQ8e?IjJ+jjvn487>7#50QlMB$&uIA5zPe}u^#NUTp4F#%Wu2J2sJ_T*@^t)N4 z_#($L``Rhv5#ftX7TiMyE(*;ZPxpvm0*QH_-5zL-x5p;qm2eXaYOCY*NML^9_i9*x z=~l~9GC}uW%@14p9o5{T2lJ+)n6^N*;*iXOU6g1%J%`qm^3La0vLAdnd1NFR8<1bN1Q+uF z0eujwz#Zl;gcrdq8L6w<)30_**`)MoUV`DAbSS_d{s^|(sy9EW0ug*!mJ_o??DRDn zy#4bzU>a$86PoqAsJ!!FnfU7Fje&8k{Z;${B_B!}9-7fs%>ZO)y(!f3lj&@$W{$}& z?b~_5I-Q!&?W`^^$LJxRUNK5{V!-{2dY-uunAk`ICghcZ{{D3Gf-H^(+|-6)i-|Jt`GI#K2@$$9n&@_r_8b(@fHOtc0|Mx9XpI973rlf@(ooMgFmjCF^L_ zEub{>z6_;T)!mg>4yFc+o#4dh5t_EJJCz8hqvmwGxgGEqtyeg7QI|e4lRQzB*4c8WHI09*tf;Ldg-Q>oa%|DhiF=2egU6QUzIj{ zP8*rqAEMqUpn`jB;AX@El{ge#F?hw(O|^#%UEa6yW?V%Yg~DW{rbCv}Hy8g@vRjzJ zPO5@Z-EOyEHl{qXuI5d)|C5q{KTd4uvh{sHQ7QPQ-6@!Qjv<*o?>@YPK#Z=hAOEdx z?091|flB^`Kq^+f5h~i^OS5YK!%g>Tc%~CFRQB^}%-7mG66m|HO5Xon0?Q1=RCc)a~zrh~oIKO2`YC1I%>7m~3xke?xYz>4e$S6Vi zEPTh#k5`Gl!D~xcZiyr!*V$1<1>sQGIR9QS47f}N4hu9`fZ9znm0nTk(=chDuns2b zrq0HVnlRTv zzyhS}+YttGf9}{QXOkZ80q@SIA*$wgqc!EKewX$R8qGxpixk0;2cI;kjl4>7-AIRW zHPKn!)8*>!38Xu$5FAA4|k z3vV3yKfzVx#b|Bkqf;um*ug~d#^f7FZ80}bmXD5CNT7Z1qUj=mSD_(0rLuQB-dtY% z1K#yKK0bz2B_>3{aG1GxloiJ(8WIgTPu(=Gap}q23$*6F^(D3-+N(+DWC1uZ+VZ8! zXw0+B$effZMxc@ATJ6M14W;aEuzF<;7(1_fl*Yz5s9L6B_$n^u^l8V2n1qTK@NlAQ z#GS6-(^t(?sNB-e+jqX0Jx;SuADz>{l*-}TUj=kZN5ZB=H95{QVBE3UqO%npK1kLx z_1eqUWHtU-p3oy&A-XN- zxy<~_{V3Y0627yb@N-SRsuypB^a?mO(Zpq&Oc98B3M7NcEm42r_d-xr(@Xak+^$@% zb@7ZT`Jw4nS)@Z|E8-;?uEC3fZ}A^W{GKJ&2APV7=YMV|@tA&>?qO+SlMJ{|pHF0M z)e%tl96nb}0Q%N#xw=+54(oK%emEuKyZy3=o!*&1X;cJfpm*9O;>vaud*?@t7=3>b zd?d3`??xZ2Hqm83`w%RGiRoq+<_%vY;P`&Q(CvDpJfCB3A~ncc`!1Z29g_EcwV#c` zns+)n^#_IZL#O@V*mw(ePWqrGr}a{l^#>`QXUAXB1^FgmUdt=Yc!6^rfv`FI6iCdF zyVDu(1_mkktqi5NcY1mXp2y6V#2wkI$|fDDC?^_s>vbL53WHBkFzpUsTe3#DzOwKd zZaK+l42`{47^_pH;ni1-(D6|}joJ4$kh}kwzyjkLu<#rhnD|b*R+w?TsIF~iWTm$+ z@Po6N>PE!<2t}qODR@q+!jlQIV zI|s|;t|Tlmp`jloGFEcOke?Xw81rPsJR;1R9Tzk1`2M_x2Wel6D8gJgC&RD$ zXgREXJ|dFU$L{3v`2>2)R=S!^uV6wRi>!k`ADn-efXZ!ItS`JopCOP3K!ZX{80VK~ z-a@(%`DcS{k-x-Z$n<-q3R)QE(yWs;Yd2okxMj;OOplqMnDi6m9gIl^ z@8lZ5k6Ku`nB-Wgl6~THFMA(t8q^F7sw*l`CH6WO9yU(1+|f39xsYIlqVm1xTYrVw z7JvF`p4aXl24b*=EZ5zBl_y5s2#S>(E($Q4z9kvN_-SZc2(?ce|JdtnqG8H2Uzg~{ zL#Fz523Pt=?6c2MnXYcmIi0F*8`K<(OsErk-eRfbHuJp^l!Xr@h`1UcjoN^xVO*=r zF?^IY-Gkgrr`8+hcGZL2Q+}`6j-Pw|Ac}))whaF%H6O%lYn!0APV{a50pg#evosGA zA8#VJL*_pf`!;s6hksEShg}@k@fx}qQsRxsP`b9_wd8YND&*SWN@MjCK(FEgCaZ2f zIw7M_joEQXmVOR-M$@-=W5G}R(x5{5*2zTQ_|HPSt%Z-8LgI>TsT9g%#xzZQKb9tD zSgKw4@@@FxEN;4Gw7+97C_!&YsAD!()- z$0z))({*mp@Zt5)g@Q6JJoG$2x(23>cN`OPR{%;Wog~7<(sTz_7qbr!odniUcfW-a zf0dxv6%BR>k1iWkXH-=BV(+G{<-0BN=q4WK6>+1=?X?C%KfZb_=1)cO4`38=QX~U- z`%c&&ozZvmr&`kjLzsi~Vk+V>IdKx?b5vM!qKrf)(HXFPjgX;tHw=1KW)p?W=I!m4ev-LK7;d13NxB`NOu3g_Iwt2GV4#`M1UN zcB9YF6##xC2oD1vKIjFBDbBS-)Pzv@?MN6a(Hus!v_x5qv300bp6peSXnhM$qTY@T z6}s_)sP8aM6bdRH+By+lblp`tIp7Z?7j^C`&C$Ha`c-$l&4re4s%^sdd*2e8GWxYZ z#qu69hEpwS`DL>7aP>~d@@tAkP4xf|9qu%FlC#ldBmhmRQP9oQ3JA#iB|CsWfs|lG^&Erf;aJrX0V?f za#@aWrrGG~z>2y6u5s!cT<&z3HS1B9nWJm^Jr!)0b~;mIqKs(TM6M3+u>hStEwqzV z;%U7{+C+K85{zajFqui5gNIZ|T|6UKzfzzo9ybT@*^9FL4PemWaf~iLLxOhY-$dF- zWsPC_A`=fQ38_t(`J`J#_V6B2aL3(YasVl{57d6Yg9O#ses59j?Hy;ngISnGRD7Q- zGZ$WgzmpNWM|cvGuHf|F>o=ytYZy0j zKc(ye_C1p>F>VQIbU041 zjemY?0hXQU-wT#t^rudy+!koE>kYu0P8-zp*eL~!8j?#+{h@xz&e|)`6t&h} z9aFNp@O7E`(2fpVVsL04F{U9&;3yJl5vMzPs0>#osGl>x0V@8j!nS%ViI z?TLQC?qMMBMGkXj+7Y+5t02vr9;U^518k6&otcNN+rpn_HFx zEs-!gpgvl$Gj&o@jVQ@8L|$-=YnSa4p*=Fz;AewShh*k!DfqN9J9;ysRduR?{vp{y zG>E)xUE#u@nCw{Z zQu7)3Ccp4n=C~%D=<&kWT(O|%s+v>fanY=)`M})=;+EUV1sw3?KYF7prO{4$Ua@^h zi>DDAz_4O0{^nONS6E6|bF_Qwdnw*6x+Y>V!#9#n{{Z<+&$3(%5DLBY3mU`gg^i53 zc_9#A#MRSP=DWAlO}}KWMLB#K98OX08{$LZ%V>I|-$c|Ot<=fjjAQ+aCUuI;AX3<_ zxkI3zxvN@3v(nl54&;p5T4#lw57Zk*`*#|3ZS5WT>IrcKTW8AdBWtksuwxgEx@9Km z8dyPAQuLvvB=)Hy4_${Ju~-I|p=au;s%?c}_yEKSdEBc5Y`L=THDz}bsVGQI2_H7z zOkw>{@GLBfGUYZ~5o?n{Fu*THc%awk-e};utdLl`8-93H@%_PFr!%EK?LZ%*Fg-ik zw&NhM=0qE@Q3fGMMWb|JEz5z+XLJfrLl<#t|Seq9rf zy9ZFy@kc*UHOI+q7bvSK=C{0=;+Bn?)+csbCrTDnYiT7Xl>%$HG4PF6SFoEcm+Pk0 zs-U?U^q;Vh&MOMq5F64V!C(3E${68svVS}|81R#Q-xeF@XL)&5eIPd_BaDGdTHmFt z%PDq|QgQsGN|_QZNTGJjL2Ly#(V(emz{i>bCVAp2+Zpah7X@~8?eZ_aW@`Z4JryP7 zQOTK}ub~x*2>_Y{pkZ>F$qf~k^@mwd$sh0ojDzZ(b@^`QUw9MI%SV2#@{w7i+XfxjvH4}~A5(TwPNWxlq$ml7uFppVy5IMs36hp422pLJ!-*51@1L9GtI5V4tfc?DM_~+RAAw{IBC?F}GK!b)&GN#beXZ zRBcIN@8j{09c1|95CV~J5xIWE8q)%a^E#=cY3*#V>{Pkh8PB(lt=hhEyz6V=ztduT z&aq^HL&%*hk(@QUBlJS-pn*ABrQN?1g{L}MJ2@3U`y#8QMzF>d%c$U>V#S~*ff;L2 z5yMI7CbM2mpZ41ZodRr19!@=e7dNUSHrMX|@n_fSe~K#mcEuleLNwUQf zj;4&|P!>i==q-`V8EWUB4ubNDmq_FNLgWQpaw z1~P|Cfj>;0W!}GMLNr3Td3E=ni8s@s?g^=ra4|m(3Z@5HjBxTs5Loi}eKkNxO@XG% z9%y}yQB)ZjZsIG~*ZH1>e@`24jmYiz^Toy;qIh!vyNM7*Bibg80ParnaF|cAcy+Jov4FwIhHM%fMpmnGU1_?g$9(-YR>i%p}?p&e0Uw(}at46NyU_{(^ z^ZfbKT>!9Rtjj_{#xJ~K5fN>jGdpmpeTSgILb@O;nfPofaPj3P>vQR1@ebDVs!pGVCwgn4j*iTt%HNx&x1L|I<5 z8c$|V5k0Q7`Rz@Q7|0^^P&#m=(&wex#)G3k$KU-vL!DC3h0?_buRnlOB6M!Fo_WK;8ty34RySP}Ve_9zDl^8I=9TOx z$(Fze?3UCj(N3fj#u%}c-!r{uDTkVFDmXgfOLRrWV9Ma9smiQgwEgR|GT&++t@k6y zD9CY4uvX@mF|BTVmB&aDCQt3!nHT>DgV+0ifLQ+xP|`epVH`L&6|p(^2o`4QMlrvD z>^VF3Q~w7@&X^rY>uNrZ!5Q@eaHr9Id$zr-!6V2@N=0*w{{}}Aal~gL6(xE~LFEo_ zD<3n@YfOk0JC@;4wxpVIjFSazo%HQ&|3udja0K{L9b{mUIC=p`Erv$<6_^ z=rXvI8h@F6@_k4*^9Cy(Ew>E|k?3ynLMabZB|pjbmdm#Ho7Q&O({RH$1{0arK?O)5TBp=`&zrL|Q>yfN)`ae`Om}#_Yry)4?V>ncqP^~%$jf?qhMMvqX zX;rT)IorhX-@5Qo2r!kP)U&o!ejVn@oS(0pcUE?XQyPl@;Kg~c6xL&Fs{717tvlL# zC3Ct@rBx>#72mAwVgGw8l3zbxJ26v=k3@9UOCF*=PQR~-id%723r6tBkz zRp~IHzr~!Z^#1=w|0y#{W;BW)#rZQq>)yb2ETH;ckyw@#ciFzW@o)OV0 zMM3V|yvG!Zdx))MeRw~2aIJEYWA)$S)(U)iiT`LjH!|>DZc;_`Uj0hOOy8v}S)&|! z{H;lR*J^Zt*0@$)$dQgA5n~O~FI8 z9|cr-aRDH~F`?PX`qTK{vKEp?_QoGeOS<|@W8^l(A=wXj8r}NfUyCY+!10?3L7(ZI z&DDfo^N1B+a4NMwaxpgx_@;Y@}UMzjkY?+T^1GzY!v; z$HAMW(c>M`p6>&WtLeP8;kD3*C!P3L72UN*dfoKSi_V;N@UYRULpppzvtstCPUFN) zK11^wz(Nknp`ckk;n4;z_c@b)`YxJK2>XR3n%<%wcCu07JV?JxK=MI4i|wX>hHJ4? zEj_co9g&;onpja~AHeTe#l@gc!Z6nRf(yP3DYP4C`XeNJR8MpEie8grAtwYvRvVTU zzfaB7ufAt*Vh1m7W$>iBn#Gb?EDuahpMy7qzkqU3Gnq`^mPQVks^y5IzHUh^d3XOz6A>)2B)CdtMSXEUvnf81ml5 zL;UnDk?e-7+eAFy$zP?D8&^a@mT4o8(?gwElBT~&dtaV&M;~JWG45UMy;djVj^!Fe zzw0HKlMAq4n^~hOuh5Deak=>)VoL;r?o;9H7uZR@zXVauIa7wj8yLnp5d|ML<4^Ur z--X)7mZLju)Q6JThD0ur4QZPonyokF3m*&2W2>WtmXOMmYRJnEv*0f2x-af^%;(RJ zC(YkHZnCyf@lA^5gy#c8PbWphrHCwPZcf%5GAtTNLxw zeoiG~FY(lR`+?@{Hjw`Rwe_9>O*PLOD7{IQfOIJ$B!Gq@NR<*g1VR!z zN=G_`CQXXctMuLpNhs1gN|6?Nl`cw0nh2;U{%(B#_kOz{az32w+1)ccyZbzIW@a_C z5}h)F5q~~D+m;VkQXI^k@k@nSZiiG6pB=rtY$ZemX-(1o@M50E)XU@CRc}EyFiKCZ86*~ z+nsaD6#Jw;cI{5rG<+XbiqBYjw|*s{nM}?&eiZNEJ{s9`c_O+$AT{?R9gkEaW83#{miTXt>AS^9>4v; zf)It0qpp3O?NZ%Db;x^hODZ2C(sVzzy;?uDgIuRN$$B3pCH-D2sZaZ4+gG{w{LF9a zj&;_9QJ9A(ofOGMvrR85-g{#f-uM3+9=}5nZB_C&QHP`k6Q`U|^?CM}CC6O4SH1h? z+<(Jmw+c4p9o{>J$YzNYw0fUK;5ri&t>Im z_bDQ*!b*>NkHueKIsRz=;|g_CpSSx%?Dc0*^zzaCNn8u7%YI_rj*EP|;y zkijLWHo#&X3vxLctebK&=JAI=P&5!7RpJ9VX}!tkpsFWIOR2j`>y zhL3--S5(5kmWzyesQ7mAmShb3FPZlPIbTlaP3x34zPwK9RKJDJ8htvFC$X31Msm>q zTb2KXM0!EB$jwQ#zPuUd-E28ZBJm^LK_%_qvWp%_DTgyNkxyJTV?68)@Bx;U$EI^O z?Oa5P>hBL+cevz?qh;QE6LBjTYv^7Q*$1YcXc&1-rQZ}}?WhD5wdNTUB)Mdjc1x}% z+GFy`7s9qVZUz3dPp<5h&Ka7Sfc?mtXP#>t3@T8$%;&NAX5?-2WWFejj_H%U+i8vm z+xKl_8Jb%Q>h7Ww4Ie3nd?F{S%l`71ld1GXd1x*9Tj1(lgVpsRFbX6IbTY)0`q(Dh z1Ks_S>n~3JLurZW13JFt_4q2W&Fuubu-ykmkLbD)e`9dZg;d_H6Z=ujB(?qr}M7 ztqZcUTMEiW*u6)yS{pTLb?G*Q6S!^Lna82ySGYSaB0Zs;rSZd0xyPx;6tc}@3=Lxx zz84YB4Z=Lvdf}5Sn-aP3xkxVWY&SKbJ)G30JwTZ(I*{!C*X8ej2E2wZ?&vsNzwS#8 zle%-rI%ijmSvMPew3F~#R&h7&yOrkhuw9)87ridahyG1Gj71-ZE36|fIQ3Y+W0>|o zBIUynb}zZlHhzOES#=8lrYAYMdCELH2?OHmtL9l!InTU8_~x(jmdl^-?Af6tSTv=N zD6ZsVn%dhG1^g;2=OWp;%q>KcC6CSG9+5L0{oatTeq33kZxAE3-FZSDy=id4!e-so zTIQ@D|DIl;JR}=&=bAxlN{IsO)d2sRpvSDwlw6;BxkiqXGHRkxay=}nA!ns2cJ!hMdbArRQ5o*WuI z@r_-Ls_2A-@;H(Kuc-?b(9&DxSzTr~N1QMC4wJ$*-@NsFKZ+%R2CMBb>gbB1&7279DMavLdx?E5Ph5lmiuQ) zA`GPuz!o>Mjkj<8(jx5ap9L3ZM_Zc`+8)Y`2aqWIVr)IeQ#Y9WmQSeQR)GLJg4GIu zcQL{PM0=0PM7;hU`wmO_h*&)1#9=@{WsiPXvil!a-8l6sc)z*U%?{Pg-4in_KmV%g zywo8}HR4HD=GgT9V@(}xF{M}3LyE+(Kj;r51mQvTSJ-z?Ji!1-nRKXaWlq(qrMYI0R@Abq@k0a^aY-tpKlfB zH})UKH0qXD*xZvXKe3ef*!nf`x^>9su0*b(`jCESEN+6q3E=tL4wTl(gpWpec^8wD zlL9j!(+-kHcgY~r{QMLFuwA83@sS{B9w%$s3*|kAnwKD|NQHlRDq}Yw_h%*KQt--ckUes0oTcI4kN%y0*NHFfx+aAd?)Z8=insOj$&1{t0>UTScRS820cY{dx93-b(=F=>PDxT6TzoW|NTH zeQbNJipY)h?WcRTDy>HEG=uDy?vb1uL%%tNHZHiezxpE)Z)kEs z@RRR)O>suppuE`=VFE#cYu+lRO-oK3Lh^M^WcnAk{Kv2VBeRCIf(T#1s->EYm15V2 zs{wmL|8ZaB5qIK_maV@;9(wlwlN+UGd+4EJLWgWd>FaP+mwbrs{V=xE_cmy*Pkvt*{y$v;T+_3m;7lIG1$fx)xt#(*N{7 z-sS%vrDEY_hhKRN4S*){WBxxKsD zMefz(_dWW^H7$qKn}87bf7)EmWx0782?(b2v0KVRfB8SH@&8X-Z(fDp`zb7OM^4E) zPk%6o-Tortll*VwbC8qko?lT z43%3IPf+b!sLnXIbT_Bx|08OBDa=E-p2ok5rZ%LZtvw8kEkAwCpy>fVbv8ZOunqf1 zgF!*L+#-CGFO5;S(EOMO{!{9nL9zd@-2$C#R&XXyu=Nx+4^LRm!QSx7v)EV8UKLIS z6Q?HPej>^}xMz@7-||r#S&!&^!k^Q$NZxW@5#Ov5GZB|U(PfKzAJ+AV@boP|_-=d& z*0PU1-}donA}&=mqgW!p*=RlIw!+|*@4u@S9iA>V?^qqG*fbfY)WIDF=djJ9;Hqd)*Q1ZNkVIACaoY1 zV}A}~FjK@C65{?FIQ!_N-*T%7q?APXQnaK-qzEA76V{GwIaSX;Vo@N()f|3y3RIa=sn-TCX1V7_+3vR|c3 z1$6M#lizqNmT?~E5b*FD)MT)B3Rfrgxu@dQT|@Na06Ft&%-$5Xv8!Pg`Vde!(Pg~% zrE#ZrQdXyE0-P~Fuo0|GO;wf;1ctKO%#=C> zT3i`6ymycT&h~2S$_jMKVUt9Eu8~MIMI}w)O7WB$RqAr4FkJzQCdFR}`>O0_Pp{8z z3bl&n6+RXdhcy>)J*iafSL?-}5~cg_U*yF*p3B0dFY&&J13S@+*~X5FpHkJ-&LDYBn;6{6agAq1^63I- z{fN7~Vb}QWU0dk!%j7Ko`5`^0k9TXGT6Uz~D~vC;5zeAQVoeqK!5YEGH?HDu1+06D zPEA*@Glx_Da!U*!KKGv0O+g|6X>r8;6AekfK?=3M1?^VlOikcATQhJ>BwVkT?aSs? zYSSOWA1dQyZgUw|i?$#0wPl3I0JF0bBk~=xW@l5ja?(Eu`c)Yrh59Mu>n?h{ULM9( z^Fu&$0Xors2`?+8dugt%wu+xQhJnY%+P|6YHFR(d8TqVShgBux;Lcnto z+fWg%?^#%Ku9v!sEPkd$!%X!c#q__UKgA+48k#Te24R+RsF#4ppGNQNyQu@^#XphD zlu!JU5leVeq_&A>YEz<_e6-;D^50yDMH41K6>+o{( z3kUUvHYMErmYn%-XUE^vf>M>o1758>6Ym|=8aN8dvm@FOzHZg`^2bULiECABP0yzGkmu) z63`eEbNcXb6<+Bln#|k&;==4NAsz=|bxV2# z?+Y(?_dplfG6cxuCaS!i56FKI{rcIF(ofIEaebaJm&sNFqjySSMOn62?J}hnd=c#g zyD}}U6UY}{8RL5I+w-Mr;L_g>RZV1L-pj%@kwC1uWng~#3G`KEO;*84nYnlU6Y_c* zw(p17RjtKDR_u(iUfxpDUODyej9~V#`=^o15IR!gB)GrY2=KVB(UqV2J>g~!d^{{3 zwfx!+`%uZ_|Y!{Mo92GNg3!b=OEsRCGP59)`Ik<{Rt(V8} z9U6Zk6Wt?{5dv*!<*)FZ#y^#?U5HZX{*(~VXa7*DG4Brr7fMFx#LRza!)E=PNxD&F_e{ZS>+Sf0m7imJ34`X+SGAS{9PuCQ{50k1{QnF5;9#jJ~+% zvf?AaBTfd7qLQiU22;66R|_-L9Za16@INhb1>g@~d2LH_QKm5@8>8?%7J)7ES4#tO z?vdr+k;>_0;Lxvh2QAhqpKkj;UMF`q zYjw`wU+Ld7EX29ggXY%sHNQNj31O}tj_)kDQ3BsnHl=3 zNg-!PmR&dN{({>@BfXJUu>62^f{f zj$S#Gmz={@DR!1N8PyYMzZ_#Z@Kcri{l$P_K5$n~qj%o9yxy_MV8?gTeAczd`9kD3 z@ZJc6r+K$Rapis6Zq@HqQJtMO?Bcxe&w|FHo9WM_qnvv zg1Lp;ujjs zN)0_Y$b9y1w5$w|CMDLN*h4vIxIKzfgQ_W3@TGTwnDj4|kNgNFsD4y0(2xJFo!jYQ z6*8I`AFVxpdi^7 z&+hV6lU?I${>mA|XDlpS{(+HLmkD-ntQV`s?Rj6sg~{rH z4mEXh!(`=3mkWNq^UzBaajm7TX`a_Ir*gCZ8!gM4)?Z`nUH(e0W-W_|KdVsrfaomxC`*pc9zvAxpkSW7G^u9} zUcTgiL}nJP$yGB|+jh5t^&0~Y8%XM^i@%-!MP%2^;L(1X~<%PqG44!?c@nmkYlm4Z*N>nly zvJ@Q7qf2>!w5pP+82uruwSq5rB1sS(sxPx0hLBhlDBw(nf2paFnjJ%mdr)v};P8!g zRnwJ z&iwT5-q+ohZhy(dk^&8XPu0?WNq^EDnX0_L);bLM@6YrvWg?&s?*oQ|tMnIzxoy*9 zOA+`_@r^x3W3=EW885+3VYUsA{GiWo&Tifp)IMlw;V+W$uk+`=%xHb?DB7Gh;SrpV zrWWC2irN@WMPn{nAfBrnp^31vzJ`V}*0{H}*Co<6AYanquEj1mRGy`71=}xHUK0 zr*KhZ?*7G(8>!;>pC=W0INOnzd9Y94jm&4E+NaCtkqQikj7_q5wY3!Edq24YHgT%XXXO^5dLYt#`O?H4tTA1WH$v`nJ`g^ zv|Cl+?y<^p&(~QlQ~^3_J~Vhls`i9;#2Nn~K0b2Xe>QYSe`TKH)#?+>HU0#&YTWjj&JvYZSr(!`pzst(QdRCe92 zkeadoVJy|7r^Mt)CuMqxcr!bgo>kfkWHp@~yb4pzylB{RL?@%gSoNFT<=;fzYqlm$ zyWkV-4Tn+zRve1H!!tFFrn_RvbCsb|7 zANqIg;vOY@4BZfXDa(a&+HiW# zDBD0wt2fiYSCm<-2o<)aN1F9EhYLXx*abFK83A&Resc}QF)8P*-@yHETMN*{WK6$! zvJZ-Q*4awIw| z+E5?0llRs^u|!VHt4)yRP#{ywUIw@f4%Lh`4zg-Dn^Fqks#$K41|RKeM^!I=xX`LJD*6umnFt-qR?@q4o=B5y#{#8KbfFr7chb(3HY_CMsbd(*Cru1NG(PJ4?Ge(Q!z*vmdUGu+j3F#_*pU4B3@c!f;GK>Es zx-xU7^Y~IqNEg=dGO;g>LfXP=v9Rv*aZ9nIQ}JTww=;*D+_Cg-)P{*?R&iu^TSI>l z_5;&Ju=|1|H*|{c4C%b-=_LHuWhaHgA~`$n5yr+RIS|Uyn`}_8+2F{oH#5_YjFQ({ z7pfXnoTx8tv@1(()w>Xp-yi1>AnPgpOW^q^~FPF+A@jyK9@F@ohuatWR$?GPm@J z{+kmpqsCRt?b(2RG;MGZ^oi=EGzC`_Gj#q&N%sWYUDDj#vfN)q#GlzgxhQ)}!Fh#E zi$-I%dAR*9*^QF+$c0bMbnlto+OYt_O^FCkY&A^7`DLt&lXE1U{ zNk<{&%vt2Y3ZwL}`csf`M3X;!b+oSi<1ec#9g4>{1o<;oIT{4ec#{(Cw;kDMXX|?s z41Ert$hO0lpO81k3J7@th6;;_gbvOX7n_%gGy)BdEwX7%KMT$)Ryaoe{3-HKGi9*a zaT-RU>lGlp@n+!S!I}cA6WjLowdIOw@nmrFVVL|=KjrL^fJ$mkI)%K3V+QLkw@O@Ke6x2kX$yFFyf`L zo7-qHhtG~O#E`_t!n8zyNyX168ut2^2ugM4Y2jwKNR_WQc%v$w4*iDLw;!qZ+UO?= z#Q3r3xdBKD%ey+G@l3Q1DthJNd5CVDD2=VOp*3i z*-Ti!#f*gIV5Mk9;zcR7Z&)x1M$%W+3D*3d4gRKr9#G1f7|-PWe1c|mT4%+%c>m=p z-e7%_Y43OcM;!NFlry77j=B2po)ZH55d?q%6V(a2g! zh6cDWuX?N%{(-kGICx+K7kyH8^`j3-8e(W6+l4_LRd@7k@2Co+wtQx^EYdSkX2mG< z=lAL?nwhdlD0~~ebH$6M*dN-)b;pcqNSC15->VyfK}jyLE?C4b@68g6qWA{S_*WWS zk*f6IYD1}O0|RoBHQf9Qw`Cxc%hj%?(}X^B`o8sFFXTPXS>|I{#Czye(D-@NfIw|- ze)+%mA&;RORxey6%Fll?lQLhwwC+1K_V|TLSJE7@9z_uFm9e=tzS1WM)V{8qe5FJs zthe4MhkKnAnXtMIn0QVqNLI@(L}R9L`~z1)RC2XPPFC54<(kB`6RQl!s<@97Gm-Ne)$rF;di zHV@0j9saj&8#e9x6}~m1FSkm~$6RqdsOj>9M2BqjP z-$uc2-a78Ff#|9WXM6Hv7+i8tde$rQj?vu1NL6%b&MuKIRJAwmuqXE5upT_^)pnwi zThv+6s6mwzIB#KLNJ^Ac*cTPy;HQ#WPxlA0i>@GT9Ngxlm1Mq^&`bYdk_2;*`=+j#~Yj#G@q>C0R z2fFDGR+^d*S*r!zb)D$AL*fhO7noBA-}}=Q=uLW*a*70|bhx@>FjIQ=Al25=rYjH9 z$MCgRE^ZTaOHra}&HXo%fjWyf_iO0#Djy*qIuB@_0Z&viTH2*cS0l1&u$dkHo3$22 zj0|GEpdgoSbIlIDK#H-IW~mDV4(-L%DV|2JY`*msWU9R5e|Y=jJXne;@xu6A zr8uh*J2T2@=M**!a$Ezc<>`qkluuYI-mf^tCt+~E1%-H*f#~sjJ+;mcCM9s}=3476 zX^Zni99&?Z1%$g5jL?;Rx0`a6N!9wQ%(7pG$M0EL*^mqDx&a57P+5=#;unJnty|?z zhI?VP+URPTgz@&?Ei;x!3d@QL2yPRjkIWIt5zm6?a&$ISJ`$Wpojt~oS*+c!FLZA0mD91sk?HE!?}DsPL}>=&ww&&o*f zm{#;M5PC?0bC$oa&1t^aB`?a0#ZIYWwVxS&tY;#mSEXxe@svRdXSSmj1#7U5+fE>U zT*-^O9nVm7(T&Q^_I=m5N&Mz;_ZH^XZ5JFNx}tgWDAKt+rz3nbqH$s>kgmMNV=FXX zLL;kdqlMrC$g0@OgS%}!ugUIB#t`(eTRM5BpkhZQU8d>lH}g3VNEam;sbDyTn5kzo z2>wl%qt`o)s(mm;kACiaJw~kP4sj<7FFV9;upJQ7abg4SM+YtNx>M1weP;<+Y2$Ue zs5NFk_(T|DI9$Y`4t=NtHvkc@=PH2o#Es+IoVtQGt9&~5ZJfWY zyFN`fqVLl>TtQ2vQQd;4(WSjJd|n($fu9{YTO6Ma7!jcb?PW14Kc3vj$1R=Pd3+GH ztnA`6aoJGI29B4LdMDPJ%z?V&Iq+|V#Ga~z;qq6h@@CSAO=~ra^N%$YSKn;0@rnU*@xBNga>e*oWtj7;iJ18#&Y@%8S_K1 zRDY9;Z~mU7vb~67()|tZ0(qJ*de<83;F>($9W%8|J&Has4!4j}^6DD2#PBE-u?FuA zjb9~pP4h@AS-fF4%EHwS^U)aIj+0f4=fX8`x|LAo`Nqk@Cg7h{MrU~!%X7e0Z2c@j za4bZFB;^4~F}o^*`lE7ZQh!6QWrL{OjMzX!T3|Q~T3DtT)Tj4}g;Ao6f)x#1s9>fM zZkwGlVxs|OlFmHMI8^>wn{CW+CRlvK@T1mq&~4@IAI;}i>QQsFFY^Y+4qZXjPa?;D zfHZNqsm)o@@!+;BtBm}{_-t9_03+*(Q3-yri;hdGAbyeArhB!AW$5RJIl$RMZDm+5 zjQ6v8?PF*69`VzYQZK^QmUn0l19rN2D>xft?)&%GrjXIC1us@;4C4C!a5lB|LX5Fv zovWA1yK}7tp^+_K1KV$ff+mIgKW(Ie^f^VcNO)g-N30BTXc56tCpgr_vK~kDye{Gv zR6OcgCKw$>)X&}u;8>5Z<0M$MZAS6v*@$(cc0Xi2pw45&T|aS5gP39hjr50 zGV(D0JO)UvycCw$1B6LKVet91R-A!65IvmzJ!;;Pa! zIne2eI5rD6EwjJzB(HpI20E1b(Z;TF#$yPV(=SD?MB=LTrc^T%N@8;x+kxXH0(`hH zEvupfrbZQKvTd}(GmtVeor$E`Xr!vdny`2mKQF}%i*#G}jwRiEji*tTyX{(&?>0>v ze+(#Z7DIqXTW+9a8m@9GbtNR@f+6isf&6eLBi7@n`u)Z8cTImMPv`4jEZuIp*4tbR z%;8Rn{B&n+=q1O+#CYA4Ng(63GH#qw{K+JUmhFYv?QTa@WpJo|H`0vgSaDwiPJomr zNiMletUih3ge*5K&Xi#wOY-Q&!*Ur%00YDm*PW(#4V5-jM7_eX=#y2y;*~0`ny`3* zeOE_teCJ*WLXjB-XTC$2f8ns$EyHe=zyeH z(q52RTMA^*ZB966!%Q!ZY#UqZ?K<3G+Vb>b^Y^O_IR`UFp9WAZJfXPwk)9jN*eGLv z4hsXoP?Ls8cn4x#k~!eOn)v$AE2a)lbgnAD412Bz;wQ-|Cn@Fxe0jhOq z#`BKcYsoPSlsQE29K4S#T9Y#sJ~r^oBE(WZ>gv5TtKsWCGw|SMiils@{DK*9;V(TQ z_1SG@4VOSPH2Wumf?pJhnoJ}sX|39wH*u8la zFl9A7j!Pbr10H|ztv6id;UK|&x686B0gm*Vtj6fskawy-oJf&mo^b-SkBu65GFQIU zcYaXg+U>ny8M1y~MnrUz_V@~vvrPC44W4L|=lvz(v{K;ux@fS0hy=YUZ*u~T^=)`2 z{eA+Ty!372)NBpNMxk_Z^?7!nb>$ECok!64)`;+3)Vnok7Sq+| z7JC~iOHauOENy!XnU*GYblP|rltBv?f-ULY^&^d2oy!raRF4w((oc{z^w)X z#iMh4RPM{MxSDtd?s8EJ=|BI#A@eVux)C#Q#&Pm6 zinF8)ThdJED_x$xK6y&fxaWpD25$3qbmEef!QiUSnO2wf%1Lid&Az9MkRAnqrEE9J zM+n;$-1>0xl;%Mc(6 zeP$k;O$f`hi6Y73r^NB`2sf3tM=-${ zGO=$kWq_fI-pppWNBIodG52hRYj7zd_Jm>^$?^lN37nkI&$^d}SgEW^RTz`tuvq6%MD&)b74auZqU4 zYcz$CMdzl|Q00BB*wX{r0GbRLnlKWDxlDC6D%E`I*O2$9SaujplWi>xC1!G&Z;jyefwNmw{X_VQCcU0|UpXaT&qB}CS(UMVj zNsnRuHX3Ot%xBH8d%rWJZPaMwfAK$GArrQ917@J|klSCwIZq@{oc-ma5BLU+#L zF%3CFt4ss$Kr4)ULC7&}qk|>tPlyfyEU810s$m zw~Y6==e6tez9KvzG9_7KbSsluDs$^0xdH;Z15V#9vv5pKAL&6O<2PQ+WVD1bTP! zM#kGf{8Y%`c0fGg%Q0X{7S^jzt^CQ4u{;sYtV(|fm=KZ$vB4w4^B7-lfh=!V2@ndV z6Eei%ONjAQs1gr)q@Hjz28cS%B)ChU7U&Vv*;K?d4ie7^XK_WG!FGUADol(z7M<3> z3@=9b4xF5Y2XXr|7=afSq*JBIh0vB{Q&9kxAh$7?h)gc}92JVB^O8S1(3$Nh3~&=$ zkM56VRHGtnp8}&~;2ancCZdxXPOgv75f&U;(qnl7il(8G0Rba(g!wB3!fXo?g?Y27 zBhlaBsEpf$ang^>>8JL}p|K~tFyMnnLC-ehbGQxIkR+1rTo2grS_P;KSi|BYEqSNn zbHR~ph61DdiTu>c-wk>99OQU2$XU4VGwR|Iy*BEMOiW}Iw-I2(ex&7H7Fysuqi$pw z08ho6kHM5=lM~6VLrWr`R?$4B;q@rrtc%hM0PqssuDXsl?aYMi5QOk870ro}FOw|m zV)7qNjJm@V^#E>boQz|-K7rFmsdUNWspG@-O7-KtuXhju>H-v8RRQt$qv;yU=YbY8 z>Q=E>rN&&HPzP3C&DqGRZ+tqTUg&RXs_8%1u>Gmm`+3kx>#wJ9d!WUad}jV=R9pk) zC5@R}6!S+AWe9IRRY_T-w&n}qFd?d27A`EvjriH0K4_2$0n~8{r*~$K2xfnCfgMK4 zM{CeTi@%1n!=A&5GX$ve)JSP)9B4gm&*`}T2?Sv;BQtAHcc4Mo7A7VpO2SSjHDT}p0=?tQqbf6GNiOZ zDs84Xn+lOM+?qe1N>CNuPw}fHK~5+>md1_Y9)e0xxdoa7+Sd**BgPkJLS%ernsYz` zAX}RJOf$ODC`NjK8Y1X^y0aF!{_Fl+DM0~YVgiP>W&yzqQ%R~Ikamz8KVK8|YoZ-J zY;(IF@QqYfr8?JrX01r_Cm>XH<%$siN<^TBQC%VxD^~Mh#I_qC5Uy$|4E$yINUDLC z8jnL&fm}GXBXd9~X((d$UY}tflCbJZb!oruFV2*OYs@lba(|0N??4eVEfVpl_4$AI ztkH*6AQU>OLS3BAaEMUheZW$`jZ~Lz6y5x2WK+2_Jd8!+ekM^2E}pPTMFI2))0A*# zq8tXzOr0qF8Ucmgv832qJiC(&ZJQbQ@Vj}W`^Ydon>>=cg zvEi|?ya*U0gF4d;6B+YYG6-{WJUvADjK*exiOfKq;sRnGDS!+0g)njJ#Pd_JfcO7B za-W8ni44X>%=B++kHm&YFp(i>_^C3KsA6gOn27O&9}^=ZKW~9B0Ky1?FcCDJpNbp+ zA%iGiWBw~X`0r6dA>|T?3IL)Efsk>-81;fM{Ji}9BoPo5#@JZGYW)9mX$GYrITx6S z%J@n6d0&MP`mRiPh;U{?Y$V;GVHs~^4nIEvCxGJ@RU@>)_>-SEWzi4<%TUd@gdiah z71m2bh_Z5QB|Y2k&8Nr2lE9>KNqCSE z%?)E>RBFuO7a>#4@r6JzB@lIWCdM*;0scq-#;Jx-H4di7#HhCs#l*r0vi?5_#}R5* zgfWJzEikcAToQES->9k0{2LycNPf~F;U^e~dTbd#HBH2RY+2HNY%GR__$xnet_qV@+ zxhWY0MEE&{*|@kl|9S`@lItCFYCJ`qU85QTh`TFw%fcxszC!}-)IBEbqE*t_b z+@F2`<;yye;Qr|X{MQQ(9sv;v8RZo!8v4tCMjQY<90CG7A_5W;BI3(v|Cjp!L|i01 zDvl4xuhq>^sGadSgA%^IqLHZUA<&pPr{ywt2}VUDd_zP`LPyWQ$i&Re!^_7nASfv% zEh8%@ub`==t)r`_Z(#As(#qP#*3Q+<-NVz%+b85}XjpheBq%W{IVCkMJtMQAu&B7C zw5+`1dwoM=Q*%peTW??gz~IpE$mr}GWFES(xU{^vwY{^uw|{VWba8ogeRF$v|M2TC zyx;%`|BTkl_dg@{f58j)1uu9+LL_N;ucn0rQy;zCop%JK_jH)-lV(u3$=g3?EfBO!T*mi`)`Q-cf6JX7zl7L#6!RZhyfm+ zL5%tE|3QEXix;YWOU87pmfI*&%E|Xwk}u{p4X-YZ5BM7Z$J&CdgoBL&mhs6<{Q28@ zBQQnEXBti#U+o|8ukQg;)Q1e!7XpYV>(%q8v^>Aoo>~_(04OjPd_nsgK;J~1b}lwr zAfl`cd~QGILHFOj$ww>zF#`Sp|LPvV59Ixtdp;~9J);Eceg6Y62wdYa?z;H(_eF!J z`PbPGi1J+=w7BsP!1$gd^55gJe@9yXGt7lEpZab42jB)lc_E0Z^>fP~K${)%fh8Vh__!+TghW2etP}31FGKq0R)v9o;5?i;Qa?_H^;@X$4L{qwJXP~`u_8ULLk zzXpG0W{y3J>K+89Nc;gDlOEZ?HKx5dNUW8Cdnt8tlIjYN$LPdFF4%?mVSWe#Nrv zzZU)gydX^>d`#+N&#iv|>)e3Re^{3PZ6y9q-x{VE+afmlE$J_7qWsC(tz*09vaINZ zwO?E%jQ`11idF{!7>m*5f9#bQ3Z*^Pw*Eo`1cTbRW2D1$Ce20S=i39(kM2<5XBK{6 zfB-`UZ3h7qN3(CO7&C6buSAJVM*56pFooXN{g^R{79}vc$QmACe_|ft$ZM8`rlg-t z@fUP5`ms2a@?GrsLLAOp!vg2FpAOKuCT*%OO~bC@ZakeEtyjGew4#L2My~OZVpTw^ zfhfl#bILj0YZ`Wh=YuVIzKU&>x8u_^^q^+mG=GQZHmKeUGZ<-W6p`Uh0O0{}1E-p?3dg(?e(o{lvpjS7J6``#35&2WxFPw4azt3Pz(p!jMZ0TJ9KgC?UpM) z3eWC=`!+EH_R~+S&5lDrx^mnnZW52ZZT5KwPJ6hkV38+Zq*_YrAida>H0-`kb8+E* zMpg61Y3cJsb+i$lt>{>B{({4l^4tI>(+%d8FDTtF=vJ&g$w5kqw9xONM^9XJl9AQt zBA(iH=le2bvToJ>t0b(S?_$ZHoGRt-inb~ywsKTt=&UTCfDr0)lNg)#>xS`y%{+|L z+}eKBlCwEQLdv=Fe*mb2rP1E7DQMR3_VR={k>lkBnHdcpOa~MQNY4#SrH3Dk+;iCU zHSfB2R#*<0887fC`RsO4V`t-?F)RXC_3?RKyhx3u|LL?=dK z{KBh4yL@0)0;qsZd88KYFa=ovkK`Yu;WlUCNLkWAbdo8kAZ8h#fMN02n*q{JL)*O;OFYT67Q`zP@Zb3>Seil} zQS|5)Bk@o~WEH@Hn^E%Kv(;Z=OSX$70$Hl}zm$H>F}#F~e_8j%xIV!u&2pPpLw-qHo#VGjrLC(4970gEr@O`KugW%YsDS*?!=(l zlW^V*k zOgaMwmBo~T3KZpvwuzZC!JZZ3g}eO>O*7FNH&`b7w0Wqf8{&6dVJ6)S1l`46lG3&6 zw~Krb4X;=!6h5{MH$zN25F6KW5(0Wra`5Ka0xXU5sICx2s@ zo=-DfEe+;>I!gAbe^)#rrXW4bBfr8SnGl<0x7Hof@xjHVX-yWzw#@z8YQU&J=*lZJ z1Ab@jjbSFwunU)Af}hBRP}z!o9tcu-J~vP;29Xj&%X6aGzSS_s(7wPQBc@g&(eqPx zp8T}mX)k|1G+_VUVe1J=J882NM8&{+Xe&!+Gt5}_vz=b14R-iu zJ2KFF2Iun4>~}V|u`F-2^f9~4k11-~AH#40L=oyt-dymNm$Om1LC%L=_kwoEu zWLv%?%Iz^xa}_OrSt~ItaVXo{S80AEKu1Kh)WOQ{_bj0@FMx))IDbBMNL4hy={|!PF!e#JT38*wRXzHy zT0*}VOHp76;;OZiu@p;Wo|S|JBu896`Ie4CbiyKNtg^K^S}SHyjK=2KBoDSRjv&@v zksnq#DVmjk`(wv5|L{^SKcOLWv7s*a)FIWGt5qb8y4`_hGq&4L{j?~SU)A0fG^&j^ zLo%4BqvXy6k+17(s4sV@SO$bh3 zK+FFf-^oJ8hJjNbrpfORdAKhW4~NE9dFL9mRHa=mJ{zn)n{={7Lyg)k5SWA5!YIC$ zyrKTcvv|>ZeB}7cX$xjU)Vnv)3Vp2v(~|kNsoSWQlsimx^Dv$H(pZgALEQE4Y5jc7 zSbkBi&5*mQVuKhFz9(fGm4-inN^=;b55BLJ0R64q| z55IpLbl20}rO{7wj|@#&zuDE({R6;q1J0`S;EGIZCqr+Dis1KFPegjaq&YsLlia?+ zx}J?ZhS;noc|Is;(<`R^R}-=uYlX-_VClLyV1c;Yk7was5i3%`WZ`<}<&;iSw=r~* z$alU#{y4;d$y87Ah9(0{AD4Ke{6@2u0hVf_D*}Y~b-(%$?YAD>)!#95y(VscG(-Cb z;0;Wge4TpUQDr0QHH`lMD>FS%yuV_>8QrLb}Gbf}1T`xtEd?~umqjVZ& z5=oN;u{TUqSi;)Rw-3+0?+v<(j4=>20HqJsOEe1UDg=gaSc-=xhKlVBfM@uDT_&;$ zvHJ?q+5H>7Iu=U`^11@*)6nvarNaEE0oTUS4FXl81#R0c!$8I}Cyt0i6EmXA5vA=_ zNcsDachR0cNh%vfGY79fSX50aoimq$KOIpJ%vs317hq_@G&ej-L(4MJ?m3(=1VTCD zlfLZu%o|mfn628})sFr!93B;XPhQb<9Hj#|m0Un>U-7+-Qq{`TwnE4SqUa&$b!&T z6B(YeboF{s%-)aJZ};I_Bb|~HU!QsGRXi*pjl~9%_+Z^muO#*3RefPZTRa8wZfhl; zO2j@eblJu#_fqn~)lF;DSZ&_*{J`(%U$HFYj^|wbhm_ zm2g-Ps?UXd(E}T3Ahc|kFdrHG(8_pUq?IWeRR&B$Cy(ab7xSut@3iQB39-IPTSJ$u z9NT3-=#T0w&T<@0@ycULoHva%Zpa*rUcCYy|7Cs(`RL-7U{}xk> z8fyrp(Iy9!<87`>E_1AA_PRnRGTrXfF~g!t4-=R_I1n*&#Is2HR3t@Xo9TP{ti=>2 zTaV{PO{f`yB{ha)&?J2Cs!XVHW(B*c=ZVMGQ`fDI*L#T3W@FaF>e+Z&m14*UqlfA} zDmcBYUi|^g-Zf_m>_AuXiaZqhNSmMmvsG((?e{4oDkL>tlg)xLedIe})!!kAMOoiJ zG1OH;`Q)8(Ek^`;${lH?;@AgS6ukHqgy;PFRXQMGa*V}>M?-JrmI$Juw_5kjSK2!7 zTgqm)$LYdQs}9|977kxCk=+cKXVb75^PK?41RR~xqJT)-1jx4^b{m^`)q9htq5-IlNcUa6Cz z>KOwcZRi9qZvZJdrt2ol<89XLx>bs8z*&YT>+0yt_gwKawJE_`9ooIR8)co{>cZGp zIUy9EnSmqppkcA(flfdnK}2i^*l>?g&2E7k?}Ef#CXE=TQ|#5RWJm?`U-6*PGRwQld?BIz@Mn)2>NGdc!I;l;C4VcGd zona%ULHxE)4OSR?H(r-banqwwWNBHt4jkTQu$rNKqUoW(0$XyFlM|LHV%1qQ*(~h> ze$N?e*E-d|RoB|%q&q?wO&Y0g{b=@S4!Hk~qkj@6qBMr-a6jzS~*`NG|t-tzU**)iMiHQClH~&DE{#fi!$Eo8DnJ%Q{px+}Y#W5^Hq%*cC zjCfmi_?gO}7o3yyfJ^6oMzbSR7j^)LR(Hdb=J}DeXi$DX$n=9(^Tscu_H|obWE4Oi ztE|CAr>f+7%;|qw0Ly|@(eAoboPGv!%kwT63@7QWK=t)QM{b@=Le*p9Q zpd#%w#9(HxJ=f2%eJM$d{41YN)YLs)V*F&#x*|Gop1hDo+od z#6ZaY|9T{G6S?I1aT}WX85@NgkgT2!4a+K)FFUpLuP99d0gCZexr1;lV~ib0U^HyQ z0QfwSC_TkJ+MuRfW?Ogfy#}f<(h^s3a^x6qn_oup@^;0&+ul_YW40{v_}`3@1}uwD zhNH70QNTV7!NRE4Ta^f+PjXxds~UUi^C|ed1^gNAsI+ept~{L#f*Mh~wl9I2{Q;Co z;?R-?7hDU}vAz1~v|8`l}fUNdsGkvDXnm`(EJ$z6VduBt^UI5(B%} zI(2wck3BQKnUf9kMbMO!Wt66mJ_#Xl2DA0ExIbZU0)=zD%bgs8;{?|{hD9ojnE3^G zN`XLgNO|dK^Ql(2Wc?i0-FTHZD+5m@Mr|mCM)lmGR!!+ds@^Gc>qn+vbT`GY*2{ED zAJ9+qlZRCcQorbms}@)JxXwhLR32N7_C~GO0}MH2)*e5+2(Y@J+t_IxKl^GyJ?K~P z>Usi9d19FuG)71?>yw*#fdl*lxhr#@zK4;&CU)^VQk` z;(gW{2AoKfyP(MTRAedfFCW@`t`uOqix#=im#9$r6+r*J?*og1brD}gL-A%cPRNkK zRGQxGcV{2KHKvio6Awf$j!TnS#N!X(=BS?%?UdFcN2~T4;SZojm;I-$@9&)oFE9VX z`o@KZGC3BPX#WWVQMx%=KhlVEe`W9MZgjSdz$7hZY=`dOKD74^0d-GpAsl}Io-gy2 z1^U*MOvbSJ0xYKKx@WC~2F5>hFPW)xE+3&-tP;je^LU>MRf z8xL*kEBH;GK>2O_Gbs8nvw5K_thF+NHx+MmxC|NcDA$EEue>@wY@tgdZTMk>?UtS- za5vf0j<|H0r*pHLuIut3m6CexIrFjYZ<3lcaMNxo2%q0RI+jdJQ+)Q~NwD2ECcNke zUM3wI(-lFHu{qx;54C_(nF||?nl1(QCAd#lDGKS@c!))D^C+^kjJxONHdxZkBB7%yW19;_BeKonhJ=LyE%qJ4S*x+$OC4 z06reJ1sP6!l@mZxL6WaJCLBF<1kqXZZ(Nsqe%E_&jus$Ca)`)b5L!L}**OO{r3vJy z>Or=;32O}^SfMuIAja)IJeQ?N_q3gx)}41K}P&|Jt;U{rjw_ zV6u&*)asfo`n1x6Tsoj38G(u%Ig$&t3;h+o#d3tQ3bUMM97`SR$!A$cJuDdk9-W+= ziKJC%4CL0l=PgKVJ?M(BQ=i@v|9;Cm%id1!QBz-mLkAJZ`L7~Q^Z)J9>i?;P0o3d~O=JU2g(sNVWC6FB(s_ze|$W z9%tgU;G5{UIQQoE<`7hUZ_$>VFlw&%1%9_O!P(f(TvuFiM&OYA!oL-?_(`YjlMV|R z^uUq#rcC2C8e#K;Zi~r{_Njwo*axLg#Q4x-r>gh?^$BDcYan5;^l>zD%=D{LP6ZVh7FJ7 za5a3h_H^^r1e#bImq-?&d2K6v{k<@~x`IVjf472l91E0G_^zOH1HGOS;a5GD2z!95 z%#DT#s^B4OyRBu(F+zfMFF(WOVW?@q26o1eV?A>YircXG!ZA!w0hrWl1kbOc7S$dU5KFy#-x1@(B~`_`*bIfVfZ6dP}*sBN+&mIeFFDlm@>-e*25Yq-h+H$yx78Ud_ZJDYp7vt|--?yfC@ z#HRq05bPdKVQ*%Av!^}FKG)wgNR44vn(ds~>`^F<8)ehCv$11Uvug0}t-K?ml>3ik z?V2l3>pg~d*_o|#w{}zdo_LwbDP3=-p$)n64p50NpR`$EIz()U?{r-~${KP^DC_N$ zVeZd(cId@XY7m;_w`d)E?fa2YMXe@Pz1#+652-=++PjHN_jtd57Sje$#;LN?b;MSz z)*|OGJn0wN;2^hHTC4v+$-Ii1iXa(^a8zzeLDh?Mg?8K#RsJ+p#;W*SseV{+C4zOP zH+y}HH}<(i`^2CX({xMtPJ;)WEieTL(mD}-U2f+{+m#-(n#%v3RrBNl&V0ZbXe#uJp0?8eAx$ zqZ{ZvrAf(9q@6H!xQ;Sw4_JD&4Q)1?gKZGS(gJ+7dqaruK^)Kqtyzz{F~PZ9a02sX zfHuHaD?Iiy2S}nLlaBHL7K_DHxs=fU-pQRdOFed!_aXgh5xk;DRWD96scMvY1vXiS zJ$!M7%58DyGk1gjtDbVrOK2vTOdRgj208X?DiL|}Rg z>7otZv8`s6$B<#c?u9lJUTpcDjtKA+yW29k69_3}$!b}Mc(8J8Bf660^%e^UU1{L> zSk)!!9N!a{jA|_PF{jw1A^Pah{S4u*?A6aVUZ0=ycF1jL$F}R4X>V-+DhP3nyE}!O z;WP$aF6yA!X5L2ssBcws5jkGrO?~VG#eg@R8w*g5)eiXDA7|z<{L=cXy^F)+YFO(q zPlPl}UA7$nZ88=4P!>W(g@#-L3%P(2?cxh+BxDa;kLC(YK2l7hdIp>)|o5@hq z@HOjih6g&+qioo8-ozD&KU*Q_GT|fY!9&H~H{H=ZF{mA2d81OLOjuQieDSe?nckZE zQoyPA1mEau}l&9{u%r5&)GABQ{Kx~z)kt-e&RCTWw3X(uN0JOcH4#x88S7x zbgb*xSk{R1wgG8GIJdWxnSay{qU2E!+WLV{)C@nVDGvSPu}R$j7>mFgcs*)0R2Xc9s)d+W}jd7G)9DVSOgo{(OFsd_@&F83Wcw>*s>_T+1DGE4FA|k09`0n%E)8bp*roGN*Pxf z>?9rD^^JrI;y7n+uy>G)E$@)Hnb6nom9y^z11Ai!1XfKzMZu0*@#26a0^1}%l0lUd!}fs-*;EZv<&ocLPj zM~H&g2wwb}x4pJ=u`;t$2v*TFvOj^xzCwHBP!=xnA*gJ4rA}GbBu$ci^*7f^EE{+<7IXTU1kpf_tiw!@ViuzRw1g@| z4T%W0m%hcH5m#Huad3CZZ~A?hYZa`o9I%Cq3@DGYlf9s?m}rO5QksbmRev?NH&&^A z#MQxyD<1q-1c95n1JcXa@Gw1FCE8*;mW0pMsHc2tPLMxr(S(0gE?FFjh&S?W)Tf|p zj-xQics1mcIrSqmXSXGTf@kXdr4O(EAwbmqCe`@)cHJ0 zXIF=xcP#9HWtvA7#b(;}nAS37p@usXXe%+^=j7mjW)@H$6X+gzL7hyw%&qkTp^hDwuehKu8oX8K#zi2Z~a1mdx4fyKw2S2i%RjA=igY* z?#05K*dt;-Y&#Lulo}QpR)fQK9bL1bC=p+qsiEwR7{Q#mjh%>WZpU5){xLtyU`CBO zw=y8M zm1{(mmYS5f*co)bk*Kgu)43#K&{c>l zEeDa)yNaDptMbZHI)Bb;^VC>;Y(3Kkm7`9{)^UG?@UG5xMaeaWo|S%#ir(k1VY{I0 zXSgz$k~HOZTkU}I>=lFW2T%GOt5e^t!W|m_0fcs-SQkqqDh12OG4M={5eqF(H9zDe z1Q*osxyGC?1T|9P82xV6_hcUOfYxZt33HegXHf^o!@a;hzi6epcpZa|O+NlYze!lg(C;sj0VXm$koke(%qCHk(UO%jCgp z*Gg0^r5iGU2;VLT)3CSW<+Zu>u@_3mGkoRLU+hD{;)9;zMtlqk^Zmf<3N0rW-F&sU zL@hf?kdD3b87QcK#Kg`Ue_;LKfvfhBZf^k#Xwd_)_(pT7X1-`gy&v4 zA|Nw@%C8y6>`8pN{jK;1gD|8^e~@ue)^1+ov>QRxoU`ltqYoqt6(+x%rw}l?yvW8K zt@#;+he|@4N&<;Va+r-Ha$jITk~)l7_?dvqnpVqiyVPgUKPom1t3F@nFXe!U1?)-$54;ENig+@q#?Iyrio3J!fpldYiGB zY5gM}N`h2SfSJJtPzKM=alMq6K6zDBof;xNEBE%wwR^pe{@ho1%-fbdfjK)Bg&2(8 z9P?68J|6WY=LiOaY*E%PmDv)RwH#_kRLX$Y546?{bQEQ*=$_t9n9;ttjWus6lr54U zMXRDnl4IKTJ2kt`jc=0A%9oV}&vVAH-#%(Zm9d^68aQs3^%^HVsWl$J2&;JCCkR$jElGEymTqMckA)1In*l2O9nNb%T#Dp#-b%btgSzw@j5rsiC{1H zTuzk+4mcu8?WJK%@w+-fJT%VCxCO1rucmw~w$y#Rjf?=_v!GLR^j`vKcDnsED{8WA zXlk01wcC#gqMqZ-{iCNh&&o+M$dq;YR_nzDdJgJ9MqQDL;z`GxN z<%X*Yv$E_NS1i_<%UOE4N5Al`jTo;>e*EyKzvUO+3+naRs-&GAxFb(HG=53P6Dpe7 zX}{-pO3f(Hw>cBb@kC2s-1#(3$UoTGYL2eZA7@bayKEd^9xtBd`vpZzp$ht8U6rgz zoIz#jJ!Z8%X8-q%gdY=x=wv&~DGdXX24C_$xa0dujZWF$hfI^qs3IM4nGjJe%YxM) z9U}fE3M15xD-bZ8xkb?-?3K+8{;IDw>h&K0Q!i~Ll?aDAYtEeP-a9;e@{0ip;y#Sm zZShg8C2P!19OPY%#szMVqq3VivBh5HyBV(#qDj84Dg5->QA5Pn9zF$!SShOgjG?VJ z7!FW$$vbBTluPr6vIY%7rZveIuKspS{@-Z}{NK!FlRFbfo;wsL=4ZgqZ#K7OgjY_8 zd$9ebfh_IR+J3AV!@F741*uH`T_L0=b@kP18QTZox)v5^6Z&QqEyUF|`kA1-{J_Qb z_E;>3w2%TI46RNA7Ofs6dXGFl4+&2Tk@W*mma&+6pez-ns zu;Za9lc43ddV2!1SmvtCobRg-=N|)dn18k!9+aq+3);h6ua z5Q(qpe5Z+f%V9WA&;YgF(G@RQ&r*`z5B>bge^rydDu5n$f?I|wOvXpUFrC7 zv9t=W4>2W16}zUmj^4ca$`lcOL!-YSaKxkMhlz?Wv?L92*(3pIi}(pQ)LwhSd`SuZ zjT5b$fEh#=$*ts}z(eg0Pignt{F%Fa*8twm@&={Cdb1)d*f6JM5CETFJFO|+oTzBU zho>Z%8PT{Hb#gE2O$)mw7~>fQuO5Ol=5bXq!JWgtwI;hOcM^>h58Nv@J1x6Ezi`b~ z7%Z+B7}4DpicN;ZtT8{{o39wP#i{FO6q1F)^j!ohS%;WDi1}J5s>nIg>hDVEUA9(7 z(>hS#dsMp+E9dln*v-k|^U;;74IyG&144iFG84*XfgBIhGi#rxblRdV+d(z9$az&V z3~0hVkVk0IIVv5;xJ&w(kx8qdt*E)}6#?tr7nl)$%1H&^T-|)QN`-pVU3-F8q4D== zDlgag%71Rx+HrAbZx6M2-&$ehdv$|NMBj-)kBG{ zs}6-=t@GLy=&5&>#;>IsXC3<0`;li34M!^*$S~$`KbI6E6ub8dmt8R{Pprji2749K zkTW?iklXUPf$d@C5gYgv$%DynAADw0m-}_-J8KMVeEgbqHISD1M~#HAFin0&sxC2D(;Qt; z{Q=~LToq2f{eAJUFlw>2A0a-JH3SX0@a%u7M*{!0YvmxPxli&jow%pe$!M8M7Zb!- zja$c_>{6#Be3D!@iUTv=w{E?4-FtR(F&7G;@_F&w5RQ7p#si!mW*>$q{d7mW&zn0W zV=42DrbXGj;-j6WirJJ@o%+Sxf1Zn$bCLJm1t(1j@FxTcKIlE^U(gE#*oJQnMFdfH z41vkgtA9qkBmrJVR&t$Vr!$F8yFSuN95mbD5TL8H_QftSv3Nl9;qLd`$ivjT#pigtjo^?|3hAEGY;}P{_QKZCx4|snS zv63Zxkvj%04;zvHG>BQPJ01d?sUHAQ`P&Nz8jax}`*h(AJuI+J^6ZZN_{vv+@m24H zo5?tYUmM5VM88QLpzA?t^qybKyVWVs#!cn?Nk!2(CUhfPtWeV!Wysag)!IA_3+=2p z%lL(KY+#_SpDJJ_L`xG7%%0ZauQmVL z13|oBJ0*rpDKr)>SDBY|ShA(^tKNwx^LGkTa=VH7&iFdLMk1)p7RbX#ju>wsIcYX* z&?_I+Z(n;q(R;!s(2l4|gL|`HKTDAU2@m6=j~a57!8Z@mpYV-&{U@otL_6E120Kd% z+ad$JL94K;CsoXRtZD;Qu(bcyE625FmosR~xpOY#+N4d3(b?8Q6^z>%L8d+3B6dEn z!0<59SY&sdBfDbZtzK@?T*uJ!MLwH@*VmQRBY_4bgadx?oAbRK?693q@1L-@@F;e> zgc3q|lO?IQw$ztPD^4Xgh@ySUtq^Nm!*3SHDv*Zu7&d2&Z44Qkxke_p2XvX;fD{7Q z_1#rQsmr(xJa==d-dF*lIwVM4d$AEY4APrHOqLBHoR!5tx}?7$+X}+f(1xkpu$Sn> z5=6sAqk)Lb6H%;_S=hyWr;+STL=J|y_oGWVeWnp!+UHcomFfOghw$I&F8=4a3Ke=6 z|NJJB5RjjG!|Xd7&i)OJzcTa35<)VY@tffijPQn)0w=5QE|M8aeCM>OIq(dzgh=#H zOIqJfP`L1~;Kn9zogq&liEtsoWr|cjfX@Z!hlR<$I z4o1_gschu9mslb#gAX98A3jvtq0Blj*V~Q_4pmbfcfnQw*w7|cAtAJ=ujgKb@@vSS zDa8;QagbSkd4sVTTrGF3&YTp_{39XN<6{x>HvURFW_zp`jX*x~gX5denh7dlQmm0- zx|9l3$%)!+xS+`C%9MQ6)O8tMuO2iJJEWD#BNy&;=A!M+w@KvqY6zd5vz_HN@xa7W zrV8Pm0Q}9IPNs&NF*Cpj13(RBngXw>w&mX2esPVndPDp`Quo`vq?TYRXpDW_oQ`GE z55YFjX_Ika9Ep0_o2JqJp3Ro}0Bpr3I-{YDN)~wNM8}I#>q2no@~L84+^MorUT1uU?7J##oa2ia8FJ2_dr6-$P#l3_msx@q zL}2u(BIEn-+f>V>Zo;uym1VQnU_}Y>U<=lYxc2M>&BOQx9K&HXYKy(0*`}Xr-}QPp zYlPuUV#rvxCm5_;bo&*a>U3p{7ma0Be`lXKL)>_@Z^n*U)X|oQKdFD&RbdWmb$7Q2 zX!gJ*M!Cp8WiE~MbG0qC2>`z3!|FboQd4#rbk$6ISM>+LnIb~4?fyV;_N32K@xCQR zskD|yRyoC~TKLey3d1+HaW506^W_}u>m6Oc^d-LAfbVTg@rtB<76iT!RQff0FUiwd zp@X#{1!FT1R*b;{3#ye(&DXp<{AHND>+g1RNqf5_ciyyVI}oD zHkP>t(;$yp&WW~`b-H4 zhUa2@9ftZ%F_)`pK8c+|{qhK}Ly5E)DCWbVN1}N_)3qK4&sW2gn?s;!%Z4j`&j6eY zd}7YUT7>1!2~k#?$vz)k-Oc;(P2alR;lq|+39`@%hUN|IS9!8`c^s;?-u7~-ghuYe z6-BQh3-c}}UW8BBA~1Q2>6Q{oxO;XT9bV|PsO-81i)DdFdXST2szj&JY|f-a831~ic-u|3xtsRJG({L zNqj9bO()4$S9zDO;{}{P+EIL(oT(Hm~qy3ZjD=D=AAWe&C&bA19KYS>pq=; z6fo%nE^BqucwR5|`b*tK8+mFgf08 zt>CHru? zP@TFq`5VqVl=ndmuaEU}aTeYt(${uU?U{>D727ZOIGI)WXIie2g}dxGdS6E8-w3@{ zHWAeTXMN8Z3lX;R0;7stqNGaaR} zh|{WpCE2Z+a~A?uFml1{KY-|(d`P3AkQ?bT!x(mNixX(B<`zwg2JLfW01nN<89^x? zXI-CF`IF2veap1y(rvu97Vj*uBr9M>!FV_uF9dYr6-vjnksxm}vh^rR-YN96jU%z{&P~fHaD1qKhqvA8RL0l)k8hob0 z|BLTfj5Y_MPW&%C5}B9PrjgCJ*zaV$B%yM*k~~P1PqfhVHsAwN%t=WjSL~fR#1!V9 zdvjU4J~B?+eIN&#uk?~cMi|LW;Ok8u4Qg+s?p^e0`C~A4@Kh*qrVfGwd*6Uh)*V`Y zEl2{tttzb>kwkHjhuX5ml=oczF5X3`SP-sr5HcnmLiL{$KHZk-Zsbp03%adv!ic;} zNIPO#v_ExZ{md}4la9TGAU0d#?<&_ykKW0+lGI~;Af`abv7&{I1WppxN$fNUM7Vd= z$ntM0stZ|ihPK97Y~GezF!9!QoV#P77%UT2Xn1r+GlZ4Djw>k1g#v9cT&ZNpBg*QG znSNU2RASKtDcgG#@)zYJ5uIsLnS60o(`xr*^CrqimBS1I!C$`8JX0^U*XScI@0#}e zc6oCwhiJ|WR?A2v^+YKa`*`;z?E)F>Wjb`2dgS>8#{ZrO3Xk zgup_UT`j(&HG7hUQ3%8x{#F!gqlJUOxC5aL=Bpvv#s^G$?6@kGhSAWvaZg?YHlsh@ z6rIUS1LSv~+Jm;L4&U<(mcla_Z_Iyaw?*JTGiC_>7W5=(Z^b}bF27_{#xg?l6SmoE ziY(D;|MmK=nhpBvOyYv4t9_6Wshm_ zb=y8&%1zr+^yU%aC5?DUPZ=sTXeMY#usw95IW;e2m~BnH&`K30j+N5(L=!zf)b$gp zGPcVUZu>d_AAOEurl)|Nvf8pf^5zD8(e2>d_b@Gx{7>Qoo5Y--GEs zCr+3-M~KD|lrVU4Qs3gor2LNx*D8`M_S6+?m4KA-d(e_5()(xu8R4#mIsS;*8|6pMLMxFn#S<(*P#Bp>HAT4&8E6xcx~ z9f_N^ujbh ztJX8TCk?Ms!T^dG2FK_hjtH04){hANCe-Q1M|?`uuC0ZbL2)ky9d@HW&6dC2L=^Z9 z9lkBvQ4(AOooNYigoZodD+htGZ5LwEFJ<6hrFe-o-n}+bQ%oDE9sCMYI#$ts=4_lB zQzL;yoZU{2o%%dW&U7DbEVCHq(XX0`@*@rFu5AMxFSG$MsELzk%G`VF-MZYxQdw*U zgv|Y1PLBpR-dlCK-l$$5dt)_T5xXDfd+D95)6WF8)BuX)?KetrdcPfOTw4DNX>S?S zR@_GWA_ZC~(Be*uJH?$UUR(l!AjKuPOKG7%2@VBHaJK}CdyBh<5ZrX5_ndoY z?%cUE=ia#=GJ7Up_NV{e`+3${zxByVZGBAU=GhBY^|F7@rZ;zRKc`N^tU2^Q23%Ac zSHzUHyqw!8J>krJ81f$uA}&}Ut1A)tH~#x)!pB*EF!GWa6Uo-Hh^jZ9za+P%@n~cb zKGlkmmYNcu44*Jhopa2ky3o|}X=-acd$N}Cc>mVG%QU*n1Sxv=sYZhJDdyVYgHC`a zE1J4NUX#1EChnuK$ea^)$ZvwM1r(}246ud}No@Y+yOFEdMVn>{uPzEp{}V8Z^#vYB z6yIw`^hmSz3{|!DtWvEd5tM$9N;^E;io~rNHne4pSiL)NM|HOaE&K}&IEN+G>xsJ- zn`k{2pLE8!Q9LM2wwr!(6Q7Lcr8i1o!mbj+yYJGYTup=pH;u2K)lzp#tZF=GfVxbG zRJxOlIUw}&cz^=?ep0?v%%av#n9S+MByb8V75qBq7*bl$kTc#Y$K2IAUsskRYx zUE<{~_S|^oWd1-?GUuW)mvONix2F*?F-q{hsy^4s>p5plKYvc`pLiRMOnZS?+%*)! zEM~-aKT#HtE_*oby|SZSs|$tBo&OBRdUM>&BOOb5o#K*31E}|3?OvUl2q<~gTW;5S zj~S5!eAIuHkVQIE?E_orBm0oI-SL!+Evj~=X!O(^Whju;y)R2uyB~AKYleHBE^n0G z2*K*l??Z@_33-Kt(M>Cl^)=DM+)=P5-nT*Rqh5HAJ$2`zrc}$S zb{b4Vq4xIT+_Gz)xONR7XjS?joPr5wibINI!VrGB_a~zxE-8BqKayYE*0ez|al|8YDzp3n*+#x>vy3_SMz2@?n<~@0)lX=gY#5U^wp5WTea$F80cyXo7ZgeeF7D zEn|oKr0XrIl}xyb_A2`iG`R5w-91AGk9e@AcyQR?7(*cS#{xnnsM1E&j1r;GI()^; zaecABLA|!Dle2B~TuJNdf>I?Bfjk9eU;Ya?DhFRh|M;`a?DVQAEQHUBkL1(O+oAn8 zPcyCujmdb!65x+M{a@IsBL8K(LCkdUbb*0dU(5CZ`~<0OgQI~PUqMF+Cs2%Qqcizd zto`F4Bd^NG!5>jn_6kYc*x~N5k!OCQAQn}e8|lMe``Ue~{Ah!EH(jvk#?BgC%r_xT z#~+)8OfvCrGHGCY{4#R6?NwKm3A#Ac#}tV%)(AGuf?1H;@;yygB@cvz5}TSZwRf8Z z&00CGB}d7^52kYN$&?IIR3|^RzETO|NVcc5HvVhdAya)OWo$UNY0fMpR&QLFw8T*C z-uO2|1z3!2^o~@l*HvKZ+efm}NQ}PlE&EHV`p(4Yj zcloSw9|!`knofqV2JxOT<*x5Q##?VZUK%5$ErQ6jWC8E`-`jb$Q(b>NXA@BaVHrSS zM;Ev4O>sYSz7h8z#e?9IIw}2ct&;k{43lZmKeZxME9>p2)D5tB0$m8QSN;DH)KyG( zjNMTI)B07{Ox`|)V+7{5wA%hN!`^$vgc3t+662gb;ceB4#DaQz`_wF5k*Y&Vme%b2!u$!!qdOuM3>-w*(#o~e?N{?*@O#LcxNk%2lSDc7+0R>Wz z38t+g!}vMNd{Vok0EyhM6o;cadCT-Y3>P4=&IMIOvP2`{8ky-ztEza%Ywa{lnhr@; zV~}nDI*wgwYQaK+gbZhb8@lPkwd2@9I?HF#rXnEE3ij}7nCP4*DzxpJL)e}HB$|~fQ!eZGe+UCTn++vHOif^Y9&Qi zo8S+sRsP1-n!XlLN3^i~lajr(5y@nNwhcZN>QJRgpEiFWsT#syH$ufN@J|J>D%0(% zock4+i7$tgJ5pZLr=l{TF*fn#z|ZBs`;N-G*EZIEBmR3T9w+xoGcS{QVLw!R(g$R! zIZiFQvfrzVni&%|e`oskHkp4nddaNdZ$|(61l6v6K&TI!aQ>LjbXnf5zVyy`&(Av-Er@0(hCL8J-MM5y@Ot2GL@0HFx%?(2tAP%T&wJf<+(s-XX`b_@a{BBn}?a!n0%`BZ*14Wa)3!Y=pLti z*Mg?YUNMVpg;*Td*6@YAi3L6SNA8bLt)L*)$XVmoXxa!*LVi5s0%G5kL=P=7J-~-? zGNI8b)lU3yJ>|z;+_4v9%6TmH`GcLz6@!9ln%`XZ4F(rxq_RN7f3$UVe!bYB!e#HK zc!5@uIUtpIWl)}n^Md=r5~YST8})( ztkmU&+8xODEc%g={wuy+WBXla74zh7RKMe$xV1dP5U5Y1Wy&^)J>fL6C

xP1@a_AWYu~5cE4#C;OOl$O#nK5me=GH_yY7&dW~PH2QF39{4Kn?_ zJWGg720xt}dnetJ0{<^ru1pU* zZV2a7z{@D=WxE6C ze3lo#8fhCd{$9x7dsG3G5O%Ydf9h3nm4Z+B$IChP6bHZz&MBaznH#B|;o+upAsg)S zg~hl#0w2GrSg-DjsVSTCPy7aRqNHPhS`p0S6Q9g5`uiE(v&Zb^Ri=QG{I;JjY_IfOn;z@-NwBf+!{*X~CJsm$4g1;%%!ymY`g;>5MsPEQyFnc$9iAS1*02@l0 zjjai4i~+6CGRGpz8XoQxvsBy1elC0Z<%U4ihuMW#-$XSJu?pq(bBosxN;grGb)G0+ zy-&+pbm6Gto0wDF%O@6EwV})SVV$$-$!Z*DntqnpEMEDUObzu9g+Vu@Y_fLX#RN_; zZ~N)<1RdntC~fW+-a>q{J>-$$ram(bHa>N+N2-JU4}N!myqk#u-#b#Qb{JcEm59e< zvCgW#U;VvqGjiMe)wtFn1F)ne-=FB?J#&^vlu~y=B5A;nriJ6!OMG@Y6&r10T#Q<= z|L1$twtrJIOTV?yC2ijY3mc<9HNAl-5Cl+Hc4tllZrGO!Be!OLh-VW|C~GG@s#;3J z+Q6l125(jJ%JD4m!;iIb0NHx&CG$vFvdnsLM`Z^xv0 zvDG1m812V0HfWGuq6|mwmu1K>D0rK*ZQ!J8*6{Ur0GX9y;FOx8+zfX!NX8OxO%i zzUxQR>I2hp<>nL1(Q+QSjFUW9M?#H<__Hpm3OnycY5h3e>^}Ic@iWrJ8ND(J|a9Y66$cUn~p4$d~CQpz0f91HDK*i!e=MyDY8mEw09*7KT7?Z z`zeO4^>8VO_`-vaDYG(*E@ypXHQMh=?V zEQhhfG0M|$W|Z%@lc=x!5rn7O;j|mVtJ;Ud>G1Xw{L`$cJJDNco=6T>P49HC>DTHc zqQV7zA2+bnqfCL4zg7ER>|h5BgTzXMeMGId9vVILK;8!PBG#3zc8s z49ke$ircjr1=uVmbzPN9(1xKz6sDhj_mrt91vMfMGV(nw`$_Q447uqwQn3ufk}17f zhXFS-2idm)woPi=)>%1PPF3ycf(5s&9`|xrW`_LP;w0{XmVvGc(CSt z6#p|aIRN4sn3GwqoN4<}t$PinT)y67KGz-YDKhQ>(CScKF7@U`6zg(w+69lMO{=;8 z;&8FC{(?;gjQl0iLu{zDH~P=CQt%@mbk|kId)z#1Nk}3r)u_eqRx&Cb{)020e%jp1 zgDW6+kDa^B#{^}wR&|>6`*`4SZtrr{vXaG~yk-wFpkWaFR}y>mOROR`1qAW%*!~?w zQY;~4>SV90ox~tzWBDFGjPwbkYL~7C<(wgXY(XxE8ztC>ucwBp=l`H=>s|HXXi)bi z{^Ii_vPM*QxsjEqi3;fX((N0fkY;0#+@I3Y$tYmBnp&oXiC^_pzmp;S!t7TJGAR$B zSle3&=r;^lPdJowhnYG8c@qn(b(B<8 z-~NAvbc3WK9+*SW^gRN$f?r2ujua{ezG72J(U+WddV=g$+ak;%l}LFLG+eCPm!uGr zR0%30rk35#=~3O3$kb%)zB+XI8jNV)Jkxa71+MHNMp08@6=y0|AXL=ap^62P)c|Ff zL_S*919w}}DBbO&J zhFsZscp_fQ8e|*uU$EZ0|GH*yhpPHl*N;lQ@`TD@U(H;JLlhf^=ihO+N3u_4;hRs> z6uY)Q1)5RRiK@gEw473dq*u7L7?R87R8LJ~0tD?Gif~@7&4#RzN4-8tHCZw@lTl5H z=-SK$4mMAb9>Hth8Kgbz&&o50kVV_I2J@ebY}deJwjhn$lifi__bk%EjAd$-n)Y}! z28m^#yUIxZbh^vy+=-u&Z&`8o;Cgqq__|Zi(ed*m~pNSt&Z; zT$mz%mh_muYnOjqUiVaTy%z;_A+X*I$U_mcOzpSsI&2;~{M{+s&tXRd%UH&Vm30je zrI-hyUqfnA@m2cKGFV(Y{t=e6)W+1Pg z&?^~z@eefByO*IJ5a_2|M3vYn% zbqpZ%A3S+>LU9ZLcS^7+s^k^1r0@atE#ZGm^Go4^nIlYj9t|~P>e;-Cb}$)bEy*Be zRD;7p-`*oP+4+F(EC`~ZDQp;dTJy}agMY+(*{fHPs8`B2LE-lSy)g@cA3mt&)I5$u z)@9>RsQvs}8&NJ~;&sGpKFs{Ia+MAg3WhhP;wl9d8N0l!9tr)Txv9Y7y^fJ?m z$Or9~GSaA79LtzD)WP-<{(f5LuJ4W%@M3kf(*S9$&&{7Kh{!VTmO!VUi5TI-JsYwA z`X%_eZ`Ls}#m`i)hf*%?H*)EIS?;q)W$;|`Sb!wOLZN{_yY-5eIO5;;|0b(D)ruYe z{%BB?j!dm0hZMo6KBbf?)ZOyvZdTprd5M`|aThL!JSD)j;9GC?vs>e<%3y?%w$da* zGN#V-U?S(=3yfyze$c>DO4cl?4_+&dw8T*;h4)0Gnj4*23zH=v!&&iJ8{&`vIZm7t zPCXj|OWtznd9F6HqC#HJ4rYhCe6f6Jhoo)u>VXDMByc^oC3XwPKjjm+2VNCf>bA&; zPwy*}c%gnq-1lbbHXrgs-X^-sLX)w!1Kd{h7pf7C!J4L*fkAo!AB5Ds5 zI}jYW(c!^Sm2Rwv7|-6Dis6VQw%BBMf*|FXlL27k$y|H9Eurf50egeJQ9&sq6j3_C zPvK9vf3QNGJbSQMjFUB6dc$s#so&^dlwhBFHv4?7Jg*+%Dp=YmC@b=GuF?YHb3AvW$C;%@FUy`?JxYC&*frD%TQ97eqNOfr%c?XZ<$e|NiB@8mG>M#KC8l=?l2*kflNFtHXY0e@+?CMLnnc zieuLvpr)Aq4 zQQ;K!PlaP=yV^kctg%D}uO=d^rN}Y53`zA|XgTTy#LF6IJ5^m8qh2EvfB}KIYj!M} zd6uv$Q6zfZ@>`nte&$d3C(Z=ADh-`KVs$c{*5&~hRZf?woVG0u#6pE_b2;}qD-0?j zHD&wI)71?fJ*# zSak23pPn&(oQWzpK}88JZQtD-CygI8_&GjS!trLV?CS+wIVTI7!|oWwtu;YMlG3R- zhIIJxeBKFp81-n#5AwjXrRtW##K|b}lNR~`R60vZ`l`g!$R3uA;(-y>z#NAiH^Q&v zI!`t)FlGtgAC!WM_l!=JSacT&TYnpVJv1?Ai>Cx%HN}cVWufcy(51)x@q9=9y{zrO zAfE2VSXPj{V^pPVoBm9X!Fwxe^E!w2O@XLN=i<5a z0IHoEFKtTVstSEzik5DAS4~u&^EEWBMzB5lzHF&nVrGdnIZ@UZ7W!0$5;)|`PK=;f z=GTo!dwNX(p=WWopJ0w{v@f=y%1=sip6p;WavH`Bzr;?{2e8y{TozXAYa@f6{TO_3 zdIC=~y0v-r))1|7m1v`L>H` zyTfln`}yvnV<2EO&b{bTx>^9~Vmp1z@`NguGNCx=r_;D|nrY#7&77O9kpIE>b(gu+ zE8#h}A`TX+6!UEBip&5SRh!{gW5LKOgD|qkGK3xYUH^nFF}D6a%=j#qBe7kj$DPwp zV*VU$htL-9jfCRhMjou8V(OGtT|@Rqy#Fc($@JDdv*wm(xKQ8+g#CNj5{XuD4x?x& z6dbESG<8tt``E6K<5Qb|aFdg_M!{vv*MnyKx(BfltWxGGd!o7i+4|IuI#0<>a`iK{ z`j(Pez2OBfs&8l5;{s0Kj!X1Pf}H)Kz)&o#sCiZl52Uiy#{RI+UsP(MrSXGm#e27L zI4>4@Qu(1XgV0hwV-sZ^q{aGc$<<~NkIh+(B@>>7=Uf0G*-aynmD>|fPt zj+Bw&2eGEV&UW+9G-yTChoqJDg~jKjP8@rIs5y#YCO6Tc9JQT-TAM`McLoGOsUcAuEZ+daKNQL@<$Ju5GqZxhs(15uhV?2he-ZMJ_M-KtSW=0}h>y#2VK;^n zf<_9zTw{a#DnIQ}*Ve`_%&4!wu`DS1jkKM187UPAalBJ1YU14)RTK3xYaeS97Rz5* zLd?aeo@^EA!&hGWFXcEFC8J5CuS9gT-gA={>Z}OY6t9rCFiCKT*>Pd`Kk7m2)q{bk3sYJ_qocSW4>8z^m zs_drgz-b`51W`Xi)I0;CM+P_VA3|I`z6n!TMf6{91~E2imkyMQByB+GX^-vkskedY zap;rFuIYOJzD1TyE$57uHk7=z*js0TBur&@4TsNQlYQQ4IW6Pr{TUuIGOXa)QofZT_v_Z zc$S)+Vx!1g{EGJhm{pwKlg&rp_c`Cf3e7Z{Jv%{fo=P1fp_zp0GSve+B(( zIQb8b+ALhETx3w>e21-;YZ@FZ_0%1hIM(zie!GJh6(_+l>M*9R=^D^jdUDG4^IL{W z@N=`n)yV1t!)#i2W)X8F7`pW#8Apve*Q2<8Vk^=-q%v80zSPcj_ic_G7-q4kNuD8A z9#KmLz04*-*NeVOyZaXO3P(7e)NY^US>1D22c+5_e{$Q&=Hv$iZmaNH5DO|BBxw`G zJ*&UyJr8?$|0taXR$GySeZxnsDT_%?k>5L`^0*#jE= z0i%PgT}E>(x6t2q+|(%DWa%49PD$t(e9}dG@*L9x$UU5fO$^p z{BENY5D2L(b7!T{BR}j*UU-po*u|bN(F?sKXWl(_a~UMOB9ghNe#H(w?$*vpmOrpn zREJ+k%MyMLd(b=O6;~WcT|<&n-wIg=$qTgqK>{+4q16tbk^D@y_uHN zv|c~cWL*$<c?chO?$3EtcS`uv2PEdUAQ_1pDB*8A25a;qcax<`B6}>c50$JezacUC zH;B?*=^CHgwDdR})4eIqH?8f1&2x|NJni)!fYbOP(^_)?N>q9r|8EO2_?(SN$Ch4^ z)k<58A|;%)GQh`$sgb@}3Dr(GS1rfr6P0i4wKcCX(+Np$h6MHeSvzah`1gOmWN|sm z9)>9%8lFfdNdEC+DJ{#l{r>Ne-07kN=s^%|tSsG$mup&pDiuL1s9CKfJ!9AX8qeSN zpj0stG`ftR0QcmYhU5WJud&s)J4>!>u;eBv8{L<}hb`y%VnV zbqA<_vF|5of$m>0H(^)Thwnnn?Bit9=3uAcS|>U7>B4sgsjK?1T;<4PgBB=hOT>3Q zSr#$lrt5Ta_aw~etCon`UgA2BAjt#)qthBkTh<(B>+S{3OZYv=Oc};4v*ny51R5VX zXC@3UehYIvUC|<1jKt-xm{k%eFLtbr7Mdx-NhBdI_eHgm!qSVse^T7Vn0u8^2uyJP zT{^ix3Dkd`fDT2@F407HPY5aDWcGGnA1i6yV%Woju#)NB+s99*~1RIf>JX`0Zm zTtX|43o>;q|1^?_w-L|VQpV42KgHG`U1SWL$EA9sY6M?gSc3J*m+}OmE6vTSGA-Tc zJL3QneQKwvRvF1WKO&@;oTa;cKXiIoUK>pTOH(uOnm*O*R1{l2kjz0U!S` zl)U!k!4^@LO7#b!iv&deDZ{P!rE1bqFPf4SDlDiL9Y-rYo_-<2B|#PcTW*?7?EWS4 z-EtFhJD_uuUcM)oLGcQA125O{RNJIp^H(41EMY4%lf5vXo%uA$AN}2O@7*STeu6%B zg^uQz#pXn2v*RfbhNn-9;&uz?SWjW|4?%FVDpc0Fi2y;+vI+UVL2Khg2?EzWO!ZGd zu9hQd>!LIVm^+z&_&3dil7CfXZiOnINVgo*=Z3>;-^u{IA+pEp?9>q-#q*$hkgL7@ zI+>U-k+gmbIQy*yKMu<~ zW@aEP%%?80sdngp6Po1sew+o%f-7qSjNh^NI4JxDqY5o8*X5K4CFu-kCE9U=T0Sm2n)mxi*Q)G zmpaA@iQBw+uO2gaA<^(lo9xM+EI`hIQVnHK=D~P=0pD@*($Y}LAJc4kJ5Yfv_X3k{ zlDU4NsPf2}*;0I+tnZST^&NS7JQf7%>u419yG?h)y#P`MnA7>H3gGAe$mf`d3>`SEQV z_CR`%kKeH3z~-?iKQOS}8Nc zj#&|;-uvM-isf^|$bOqpm5breR7XOirU$v8N0Y0JBk#_|Ncm zJr$w#Joqo!e{j0*w68$Wb!jUx_y!#6f&!q~2fiogEP-S*yRwQo(2$>T z?qo96l+P{ZOYZ70CrFu3ne()VYpO4eXWrn!ut|4II{XgR-kvG}RWL3v`U{IZv6Du+WLKLkLk_ z>+yvwOK&tt+H2PgZCi3FehMU}?G3j}AOUOLD>XNpI+_H?f0jt`D_eFV<7m$QiG6xr zXpG5q>x0xdT7U32KXm>o`}3=vd@Jd1DUm`C@g2mt!qH1w?v6#0om_XQSi((z;_%qM zE=$8rM8({b9N4}vUJ5hsA|37PUz3b~nEkXM6{h?eZBOVi?Ot4qXlG?cw*zAZ*1zJr z?*jx3rH(z@?&$Tqk=m)6rpSvp(dW)Zi)E}`dUzs(8{%a>2Q@BTZ~h2(yh=&v|Ls)R zzNpaZG{xkujflPM=WP-9d@@*8Cr+!tOJa>k=uSB)wS0Q9rpMAjz^$0g8!(}6WX>GQ`O$nYsw4zxJBBhG^_ zerC574ly@v1SCpV2?7qTg&$R(B_TS@6nqis21c28grrK%>C6}f@@^IZM#X`^RS{`yaGhd8CrC!mcN!G= zJyJi6Oev5%GFVTDT|*;2FK{*^t~;ceC6bad)YE=)L8!*999T7d9^XFG!p`;nY&L1m zNeq(`y}qO`^CD*k(@N7pSm*r1s+~+J08{eACT>rE>$<>Lq>O{*53uKyF!$;wN$U6% zqV)+|C&vz4)L+k3*@!5~L^^<)I14V;zlPkLlCW&yWr8E(n8a-@m7WEZk^tT;28^4o!R!`sNSqxa3=J8|ut>o+#>^9HJ3 zQvqWnR67bU%OPR9wSmtur~$KSYpCi+%7s{_oVU~*M&}Y8T{&X12QTa!bWUay>Z4z@ zt3{K`QG6qwo~$lX4*n1m;4^~;U}?Qizae>^s{G9mUSceh zUTk=-`hm0>?X^}{lyq6SNeoj?{r?)Bx6Gi+CrclFG7aS)Ht=dbNptKFwk8lQj(Nd) zN5<)_A;Wd`CUEM2qM+JkkjV)6n_ZN{-~cnf=M*cTHU1NE9~Kq|?#P$@bldAn%)7cM zxuEbTP1&R$FTSOkw0JZtC+gRt(-bhZ{SENgBnb1Wys`FN*F<#g+ z?2VgH@X$tH8Vhd6#MUu^yH7PZgTZ)|vVkAghaNbET@Se?SZ&%5{>vaVle|k|)PX`x# ztu=9MJHUied{`AQ&Nd|QHTX4!j$a(hjTd$Pc1oU5KSmCabY>Hc|0AW|cKROu2uKia zY%&<<#g;T6=e*rkSzdL-5Rla#@2OVRO2=us;X^r31Rd?g|wvL zU*L()spvc9j(}2&KH;2gg`U6^B%lN12X$>L9`JCca}=4ot_qw3ydT}qhs1$W-pc%Usmb&0tpVU+7!l+R+dV`*l00e#wXT-+$41U&NwE5Y(#GE(cs8t34Ql9% zcn#z6lu~-m0fBJoD)WUE$0Ypz75~?=IR&|OG1p&*7M3*?R1Cs(8k0)Q#5T!-W=R9q z;6TPFm+ZJSBj8j*JKCFcs4reL2PtM{{2>I*uRM_E9i?N@?tRYN*71ckU6Bh zR;Q=@t0L|FO?;wE7^2-F?rLge+d>Dl|$ld5>`t z!Kvu~)4_GkNZZnjm~tshZc?tKdlzXZ>C4LFZPL`G3_;-_x&AW!1NKn z3+#v9lKQl(0#6Vnd-4X$#Si#(9H$!Ye-(I;N;n8C{Cm!Zc(@bStS?Y1Ts(**mwd!a znoDOMyYrN-nvy7F9C?p$vurtKE~4P9t5?lS8vt%E@OZbiWEaezW!d=ksgwk(rgd>5 zW~jd18hkbU4^HFZal_;O0$hS9Vh{7z0&B-$@=s#17xYDrV*WAuj;>s1)TW!h7gbjZ zchq@7Lw~&JZkbM7>nTPwCEZ6DD!MDP!p*c}8s>bEAla#ebP~_E;~^EG3WY#gx_UU3 z(M-G9LM&V?ONqWS7E3nc`f42jcM_>D+oo|#3gP>mu--VcL#$J#m2)|{d#4R?@$tSF z;5kvA8)B|lCy*Lp9?<(d%$=u__a7Ylc{d1~F?pNGnadnv`3odGd%3cx2_?y$8fa)s zGFVMUw63Kd7F@SS&SW+Vb4Xx~oIagQqiM zDAYKxlnUp$LUk;xH9@EWpl3G9Q%Kbe|M+bT3$;>%qzZ1Q94U|GzJ2pt7uxhUBP8~3 zvfKR{lHxtbhbPa|6$?@)4~D=(D+@|CpXivvF&1xv1#S|B-#J3V|6LckXajFo=_*@o z57Gb{0$RFV!-;_m{SBNFqfN@lipAw1u|vkp*CN5YH4_qj{=-i;5k;>yMLj)qgQZU& zWcvynufZ`t%sf9|`d*_@L(kJb2G}Q)2c9}jtDOr9w(RO>hS$y34BtqZLz#KmTc$Jc z5-y|uV2oD;l2-vUmMsX^$2nr>2UxE~W%DUb3_)y?kl3#{NW;L5oEnMX^109y$G3{6 zVmH9^C;IBbXX3@`>RLsq@z}M?3nUyxZxB%9KEJ2=8qiWZwH7r0ojjK3QX($E_0H#R zd9iA8Ko~NauKaJ&_~^(qm_PO<2D%xi#$7NwiO`9=&77=R4{^QQObU2z;PrE&46YD3 zG126!bPKtxD94=+lPkAjekDuX5$enOQTVZd_D5nwgJih`_jW9EN5v1CNb(3h4O5d( zsWHquaN+3sk{K9ug@Ev#91^OS>+bVhPVM{mt&sivGYXVSQkG@BXrR$gA6|s!zxQyS z?!ub$qqdVXFvGCvP}o)gx*hfJW&Q7fasFN5@2zDiYETg#{%N>B>A?tx>6=~WN&-F4 zj{BdV3aP1!GZK|aIaPNJ@chg6r{yRAXAI>f2=PTpq9%v$vgnOz26nj=nL6*PQXh5rN_ z{YyJk8e$yQ$WXeJA*XU)F)U;M`R;t*Jb*CM4{h2IHo?asy^I0b$G9Y1iMVgjVRZco z-T~WxJxmWwl3nQ5TaP!`rNn&3%-)!;i%UM^NS8_;;@OErw+fn*YQL`+HF=aQ!7AEh zW*8y{oXvYH*cX1Q@-;%H)Jc&$+g|B+t^Z5chylV@e#AG-dB-@=Gq-+onMwP5uoW2V zEK62t+)Sji&acYs2HN-1T)YwO%T2k@@MzgM~y zumn1biPGKQwry*_w&OsMo8LP*)rbVz^#_H9U&;%c->eUx?@2lNMy~^CDa^DGgS~Z3 zn@DlYD-M@HsKfT}442?(aEFqTz`BZlpP#I!P;Rl67X54%fVbH6_Igl`Ow$(#u3NDv zdK8=o+3Ztuna~$34)9{O2*3E1L{dsGxDWc$Kha-lY#O6s^Bw1vbS1mA#Mup`s%X)E zQbGLvO*i@n?gU%>7yDvny3;-{v!tG4VXCnkCNf?PuUOwaBPPa`#2ndaQ3amm!#Bo$ zX6J9$h#g$-H6O)`k}|*19V$wGElC>@^;D*etk9l$c0S`&WVb=aQf7d#YRpMaEGsl> z!DhNX?%9Z!sv;vshJ6@+7K(YhW|`~cZk2P{G|N|AKU%CV8d!YI)D|p__!(Nkq^ESU zuNh{LVU15DU6t!c1ZQl;(`LB{@oYL533_b)PUE^!ViXZ!L7Kzi z^l17EBgbWEG5Cqxy^g1LhY~@rlt+CcQ{J9qXPCzrg;8L9&*>smD?Gnw0+r;o>wr5z z$LvT>C;h<39^|(3STuM-{YM&+{FFV?n#?_LT6+?mqdFi{Qq#2|@t$sg`;iTR7ft$l zTZ*_DuN4+@)BwCA$eBvKGpa8g;uwyplkWjG(cW3U2%dth%~^fN)}tNgOCEE8sIma0 z6?1Bj4%w=Y=1x2;RKAtLt$W7Yubdo4%fZVV&C2iI6Vmm)h=v^P^7UoCKfGGYf$`Tg zSocHxRLULGDKtv$EQc3^P`O{L>=Sx9*^|OVHt|svF7pn}^5B1|QejX2gYzwBOZv8h zzt3AS(9T6s+Q*2lHR7GLWBqKVP>6SN($l@7q;cmg zM6gMjc#m4+mneC+zCaonjRq01vb*!JppSHYNa$$dm%bXPZam!8cRKjuV8W>OXe(Ahe_k6ysY9Oftx@)yD2amtOfOeuT2O}Zs5%!DWeI6^k zK9T-2TSfxzodAPaRe9Y~|SqfjvRdHn#ugK@~7JZ{gOugpW z@F>oGNjvJ|5o6=!p?P}6kOxZ=pK8}HZMOD5I!RwPA*)X_=+ck&Y(^D5Jz|kQpY&Vh zck(MPci{;yk9g>d3Lu||19|N5GW3-aRkqaP7-U0!=nt23z0$U}P`csoW}pf*pnaxX zVwsn+DzTA6f1*`gUe=-->Px2VKNh$hUapq^xc(Nt7{tWHlDds{2CBKX=A^O2Y_xVcWe_SN3B)`*hQzvXT6TL8F^IoB*ZfRWy*7EypU0g7ZH-n(+w0Pd~n7)}<)fJj$5f zrO|YAsTs%R6EXh%`gE)*(i>Z(n|`R(tTbkPsvUDEFIG^0mdIh-UFDHtA?zP+(KdU@ zD|BR*RkNT#U_=vB-nbk(xCPf$I@kF8y}Dwr3D|d#4$v}L+QAq&z~S88(Ej@TgM%&q z^SE^_{<(I?N|yx1>2jfi-s0*H^*1)BB|f}BwjeUTwBofz675SsL_!O76bd6{Gxo34 zXtt(j%2dojl#oqo`NmL9JhQp~qYY%nFnrHucg&k-O3T{@4125?nF*vP@G&PjHTcbp z1O#v`PmGEoT7bIx(!2wy=}(%~)L23OcE&!H;=dnJvA@9fgwBX7NK&{#4qf>>gGn~- zvxyfpp4cjyoWyqbh~j#dCTHfflc1} zNc88#h56%n7@zPjI*DjJaX0z3b_b|(TA&Gc{^~T6GWX0m3fo6iXVyOXG}>pOb+t)e4%5xSO~WE z!m`zwC%zSNQneiS69l^cG>+iA!74)45noilTgQ5tM-*$}g{k%j&Irb_1!_j03%*$% zq#pK1vE6AdP$;f%P)}iM2^%QbJ>s1id;fGyS2bBC*PH5PoQ?Llj4RMVY)$@XLL<8G zR9D9<4SxPFb??7$mjBJqdxcPEPgKnGuNBLX14Kl1v*Kde_=_GW&c9Ifj8{k zR;1Ev4Fc?$qbrlBPlBuJ$cSU{k(!LAcQu`Ram>AP-@^sb`l0Q(4^E2H%N-bs8SXoZWy5{$86Hl zlA4d)7=N0t49E1bkmk(Du>Yg%!b!uxE6lX+Y^A)@OxE9-TAJjqsMZA18_8?4R8uSC z^T~KkGi+hh;P|s2%LQI2RLb&DE!_yK)2>S}djLPVC{~82KK#sd18}3#RbA^>u?W0y z7ABiB>94R-W^aMXR2bZNt7z91Yoq<_$J27jeTe=CWoOwGM+0qZED#_-Ah^3rkij*$ z2Zup}yA3)(2ohv)4<1|xf(-610fG)raCZ;#o;knZp8KJz`cqd|byauowby!<=IUZf zJ`mn;QQBIXJ6=+SY%W{U!JEUVg7K@&Un{Si_onQ`wR2CqCgomhv#R=tScR@@+K1&* zkbijBSP*CzE-qa~nKWSgWh`w72VA@yls0ynwM0qXO7#1f8l}GdWf$_2ZoR8!M+uiu z`vC6EmL-NdP=DFl2%Y|+J{cuVYxhN*ro<0=8aI^d%LM!FLRpvdUV-Cq+|iMer9ImI=4k0S_K8!}+ZM=D_?c*&%@G!1=D2!v z*^HEAXy6I$`kB&<+@+H>D%kWKm#h<)nO*%5X*B1JG26H(E$z`+RdE4Ldx14Dmmcj? zT9)Lk)2zNl*!yrz1N5k5KO=s^zS{SjG8IRkvq;M+Yh!lH6HnA{tpm{?-ZiLn zDyZz$x!gJc#B7a2-#2{_YrMd{)!eA(%$NgGu2^0^?jlcSra+XF zD1fga5S^VAxSQ6bhU3-JL13mod7}|k6?+?({+O!j(rLWH65eiB?+?pQ!dqRTIjw0^ z^wlo(qO66D{FD&Wc|~mx;2yu*`{;jL^#}>Wlhrw zgB@s^KWsaW=C_5;8s-ZW(@AVJqawC&3{*20&(th_4ufwN_Y+JnR<(*E=y_8%gloAL zyqo6(s zaG2UV)K=fKuQvw;6#EsNlibH*(rK1B8U^ZKkM%7Zi+}t3H~6+=~)H;dW>xhx@27)|Ha^dRJ)c;n05*)s-JQxTki|LqDXDsFGWD+*pYGU^;~l zN5DBCW1v|)%}eoh*Kp~czxUPW1mg&0Ds{W*C}lwV?9Zo=X8>&nsH;ni9(@M`=R`*hSgo6lX?{OozK|_MFdYED6Tn%ThF_j&mzMq|3!vCiOly z?!u|(L(h}nq2}N*^zaTwbIARPG{1Jxr0ifFVh};}*RrzP#Of7GVLKx_fF!tunH;zQVYByqs4AHzFhmI>0 zvjT^MD$1&IN>7@EiyNjv>45&azMNaA*we8KH;*7fV1eTo{D)+3VT4mT9Hqb6v76`=hucxPlj0NJ^(!X zhsJckj4Z|_Fd>7XHfKmKQS{JqyZ07_Dj9>(n&OZ4^ww^&2eZE4gWisUBb3&5kb?4F z+=?xA>g64Lnsj8;PS!66x{%~n?&1vDikKl{g4LINS^|%(rIvIt<1e0P;?>raaFH!U z&w_&w*qI+#p!_)&-(+wd77$c6Xn7_lADVtQA?Kc0VTVJay}E2*e-ykaG8?!Tm#cuw zpQY|*4C|gATTz1M{hi-^Vyee=n7pvA|p8P=(w5SsnTg^g5peU$QS%kwsIAhZU`zyAM$9cHh)ObY)rFq zh&1u8uKL;VFiJp%q=P@`xg3#6oKiKxU74(jJ))9&3Y5hob6_T77f4aMiE$d?p~FK9 zdCi$=r&@F3>AyaN`*dG&j%dvZ0q59dEt+Lb^w`Q(hU$FhTs6gKNuOKR;85jZ@z_}- zzilGK7@raC45|H5%?~dD&B`?|J2Wq2vrc-Lg;W}2o@O&rVr^Gb8n2JV7LhV!-CWGO z;fUhvV>#(el%caxGbA-`rB7T2Bad!=H-&$>f?*mm3pL5xt>oV<25%6Z4PJlRAJt7@ zQl&nn|Tsncn+vVYZM`L@gN^rY)s%e*X@V8(Z^Ez##7 zKJoM8BwH=5Hx{Y8$RO{=7f70t-&55e!FR#gi4>2c1xsNS~1G+X!i1V^zOq$=!!lx?D zWo})6w2zE;B`EKYL`L@N)_hv2BZ&_FjT@P4G#Cw4kndRg)=IQ}%Fqsb**1%yE)&_e zEPoPvuO&}3@i)Mr?wM-FvUNW1%J3o14-n^Y zHLx7(UkrY(liyq;Ui%X6g@si4v6pJ-c*zPy?W-P5%xz1Nl4Y!a2nCb8;je@zMGfT) z-cbIg4)U1gxqBbQT6Iqu5eE=4=6+{~-&YsNGstwl`^@FrFkP@{&+G~PlX5cepKEQu ze60+XdcMkZgw&{?=$a!ERi3(27zJ>_nl&_HY<*`Kv`U>8Qe>Y27P$JIEcOYLy`)_) zmuKK-Z_^2%#u{P#e38HeRf82`uZ+%GCHy3*3qx>y97yM4uIBc4J~)w==ddkzSOJ=y ztcmHXDSUz)BU0l^n?~_tct^I{IE0zn=pfr!qh`xQT1$`ER<_kWgaRA*N9mH()1|Fx zp)KYas|zKkr+MGYlqJT)vG_piT^0{v+y2j5l_PcST-DB>YQT3jmBMyHc5YW9EWq*H z)5jmLESG@&C>SPF+Egm($!EX8FGFz*mGtSh!brZwNOefwfYV^CuT+d7-nsgGers6x zJ@|7bJZSEJ`qSYiHSOPDZN)@VU2yujWj~lbB-DhF6ijQY0A^ z$4;wl&NV}RI%FMv$wuq`jq5{jpFWV0UcsDix79Zo`1iZ$k4(_$fzP@C!~Tn_dwLhv zB}NodzE0~^k!xUQ3w<-!R}zQB?THP#Gmz<_vjyw$rdH>iR&aGH>bdDniTvhzUe@T%It9i*C@Zm+gtV1R6A_fMM)ofH!|5C{#1dz zxO^FrU6w7?v}i&v+HAGZaON{r0<0fFwQyw&Fpw>OP#B<>jz0;v-Y<1Z>wJ9I^f8`c zs}g5;%~mIFezISBjwu`Q1dvi3@pB9!9UaFjO(Q5YbCzi0!0>Qy*|O%PV(h2cM%#ui zx(87z(FJu?BxSm_;ZCPd#ST+xw<2|fYf8**)OzxS2F3ngG;|`+d`f4+AV{arfl#e|4DRR0804Hr}h{6;fl@2c^YQ*@pl z2Wdp+9D8zG2VRL{T}xO*0~HdfIi|h$Fc~g9vaaJCI!Zka&b@v{fWkxlic4H^nCL$> zdWiwL7hIqlXJX$rR3>ZSo3I>^(}sO6h*I{pBBjyjvq7?A>9L?s8*_>s5hR4+J?5Dc zi~hy!oIZl##7;Jf3UIht$@SH3NhhIal{|WaTb0$DJa{UV>a==xz*ApYm;6QrAxmEUVUo+x_VP>Bc zCGz%G=QVqZGEf-NOOR+f@i8r3bx)Xbts_Px8>RVRWL;O%e(Mho$Od6Xcgx9&6pi`3 z(&`|NxeVTp>LjjR_?>B$`KmJPo3IHy!HFC;1O;EK(=m7~V;->IM|a3d)`I&z40Zo< zS=e?~b5rSG=NEQzpt z+_}FeLi#4Ez0WSehOVwP2je+;@O^aN@j+>05uQyD$cq#SX7FWBrG|}4#TCYe{^3jT z(4wJIoTL*u9rUOk`rg=@4+~w(%?mofm8oIs@buX+Ztf;(FF3}~Q?SgIzAbw4)R3iauX$IWUcb5bXQ90>=BJkg8Ti+4HEeFjcZ3@`l#7k|9A_Z zS)`thTpQF8-O@yM$k=Rg(Q}krW4Q0w71YKMD8$v5VUX5pKG>BXq3qH|>>}O)7F7eo zOM`Q={oT&aWAV! z%L>Vro8D-~_0j5;iD0bBr1cFgoS_;2@(!sjQf^n~?_cewqjZje_i=7Js+`}D<6#Xs z`gKmSY1azs%C4Y1Zu;ZVMbNNtO+f!E7z&dS~kpQ_o83MNsUh8HMb)2IK`#w6_AV!Sg2uppkS!uJx%Pg zv0UO>ci+ej#SmvdGR)AQ-3D8 zw+=gc1%LFV4T(?P|6GqQ@Q@?gwhhx!9vSS!%L_E6MJQH%?nP(-=B9xI)w$f#;>!-L z;cd*fld7Q4$7oc9NWtNI%@Kb@*G{H5gS$3>M;*k+A6CVX@0G~@8L;=`ZUEP3KySMK zdbD*q{_7h~`u@spXH=}<1~bo2!~wJ03MV1|^{HX#y?gB7YcBCci(9;@%^6k(UF92T zZW=HdDV$JlFBGHU_dc`#qRm1cmNntx4)fr8kBS7=*Y9fHu;a1m{1Nyexj193fE5%hWxOE4IxSJE4LpiTNYR&65+d48x7x>Q`s?UX}I4J&920iGI^>JFJGg~$Onj%_FR>p{(`GD~F zE$pl<%jzcUD+2z$5bvuvxYoGW^5{~%@pAQ5GDMbPtP#`kDI9qc_D=i!&rzX3tQVvlF^V$D~aibpV6qJURd%|8NE&oFb3>N(-O4Bc? zLkcy##g6cqg^PVXc-7r@Jh$7(=SD8WatjC~MWqK^^pR zKSxW(b>cCLTc2p##-K0t46Mk~c2V1TrC3^AJ|ZB~Q~RyCUaK|d?SzfCyq(Pd-2^_> zCc-fMennzdI*tXsE6PdL&{=#~FE>pzPr23(2gB9&er{*5FXmzZ#NI8rRu4VSN0 znCT_Ee_YehEc@!kP-RfL1O8gJVM3CikD%y6Qa*PRzt_s@Rjv$f`Z0DMnA$0KPWnk< z_dwpoX@X&c*pqL@?zVFIMvIa7Bn)j^D$RHL;3>y1*Vr_cowbZ*gX*y+C90X+HLM2D z`Mx&jf}p?9kJgcqZ%jA67LA*m+htRp_6(&f|Kc^q#^dh@r>iwJgOSn2J5DFvBPl+)Fj;F6$Q;jmaGz^%)@o>AyEgdH1c3^idIjn_mIkSRR(B z4(UM^-)qfP)t$4v#Fpp?*ft^FD%4KspOt0po!D?925919?_6)|Lv#<61`U3g^tlu| zc=jVfX2J#&H`ZU`ekC*~`jukSNNI7?FoAWqm^3_1Vq}4?9x*!Hz0TMiN~(hVK=Wn& zHt%d@1|lpz4R$p9EsK&aXDgzNEyLcK9kgwBN6J>yEk*Y1UAwqv&4)1NcChowu7`7E z-V$Js5`jvGTfRSPV12!;-qrq%bmn0$9~u5d-SEnTFQt6zAJtU9CZ-sRJCx4pQs!AJ24* zP8}1q8kCKkmc7v+IepCPqM0t7fue<{=V@jsNGL-UE%wQ_av8i(iQdgre@u5qdIPSD zbr>3#i)UWl^=)T&^c}ZBri(b69~uSu2JpXs(L|{>Uz05I{q|K)T<2t)IQcB;t2P)u zVR@;EuRXpgQAwyrYFae{rD~jf!hRJLE_xtO~6hkGdok-zPNK}>Dl7WPBdy(fODrd~uYib{!rtOn~AsH_GNCW8KSTzvTropR1P>apBKR2aL- za_Up$QO}5Fi#BEA7C9b_VcOJYzta0jJqg;iMo3iR(fZbw>7Ml>>S;}eW5uP2p%%T< z8pyStj&~IwCc>s??M9fB5r&we(c+`}AuA3ttEOLoZj3xN_w*6?>8b-PaaBfFO!F%qFO{g6;2w>Q`DGlcFl;E1;Py9YQYmwmbT! z;AvmwhLPmPEE(xzc~mH$#zx<+n()dEHt6O1>l zl?G?_O%mAhTa*IMBIUQ8r3=ngURC5~@NmK@m5Q2VE}NY{g6;Wy5ACP}8MNFY)$9l2 z*6o3+9#bWto4h?Y_d*EU|84O8^GeM2I_oKd@GR_t75nltKKtGeA8LrjQD0ie>{jgk zoU=6j9`Ao!MI8GM6@I7(Fj#U+NqNu#OT-D(?TH?_`Rh{0nGiz>wG*qA>Dy1uEXB8U z5h0Yn0A;-NtM|xLU`aVuPK4wuKJ*|n;(*v3FUhfV zg)&U<=t)J_ddBqJp&w0JXTBLu{}@)WRRdJG!RTov%Y$x! zo1X$z$#|P4gRt3S@|_f=N&-m|q8K~sGT)!-wDT%&FgWtm9sM-cB4TnMGgw$IYrxS8 z@7Br3_Trluu9|#?9d})6ql~6@_r>s}zPp&zJt>Tu$LGvbNy}4J4}2t(eJ(GshE!&5 zTdtk)7^Cboy zZ@l6?*s{8Og){m)pvw&U<1GR2wL<3!UDx?M%(WjBHB@oP-}dyQkyRg!P7JZOx$8a_ zdxgk(gi~7RSSMLhTL}{4f>e}mCP1RQ12aeNMmxo7-Mm!;dUu35Gp%88?Wdztzf+DRFL9LdQyI&`9jAwLly!e5u-8HgCg2U=_hx(T<2Twnl5Mf8Q^2zNVe{?lO>eFW%xyctb1p#Rx=ntsj; z-ybXV@_C60u>#q)m$lc{kbJg&A54DTP3jn7{#K=Ek6F_Hu9kxY?44TJY=7fEt;e=f zcU?zXtNZ}mlF$&mtrcDdev+5~3-vt|yXcGKzVY5RKQkH+HSbdVQ^o(1;v>D)u$glI zXiGL>uS|R_X+TPpK#9q?L=lYv3tTOptvixo&gL>7<-@u7g|`Ujl}sr=YIdAm!;oGb z65eW4EwRWpq)wC7{O+|<)}Ag?H_4lqR!t7-h_fm-Z6??|*8F1=!3d)p=^cjZEM)zM zqz%KPj1bnm_&58k?CX5LP~RgmHDOFaKPPZ7Af0YP&0*vTYX9KCTA}>j&_uS#Ppc&J zV>zE!i z5A`b}EQvRG9~T;^HS!tqx6x}KT$RRMyjO+nh%y*M1QsKqnwy8u+1Dw(I)=Il0%Wra zCNTOSjGvvbg=Fq+v9FWDWYy;H8>AEc(wR6owcjPJ+7an&o}Rea_&RY=&D1ToR*`Tq z&CgFo7Qe&){EG(aFWynVckMOn9Yk436g8c>>RIrT zsw?N(&*u>GyZRHJmO+PnsmY@G6bO0Ag`V=8*mQJLI(?YKRYOYKg-drdH;v2Wf=4q|QE**w=lBr#|^<{dexH|u@81dxJ%6XA14!#oM z3JKhMgYJdkx{16zMhC&|->Aa19J9|N!C$tR(x2`IT2haiks>hbTu7-z16Q*^15-qJfaEQgIDacX(PX>!Gp=-}D zY5DcFWk{kEHcp|*uQSQUX$I>V1n^WE^Y6C6>8ji~Z_gA|jq<~so(9-MdgKt2r^_4@ z{=z2Yb@({%o%Bs)uhal1UQAc`5woBt53C&@Rzi!EaD1$SE~ed_#1QpQ+vA8|W`v5Z ztcf^(1jF}KS{S30O*9yXMdYc`cm%`suh#^IM4kFSERB5#*AB|7?z??q!ti%qr8_zs zKTo8Dr%*zbkF_#MNCy>ck+!~5oJ1r^seD@1^Ki#=Ong6{q&~o@n(yegsXhf`_M61v z!0M0gIaD9nhuHcPW^RXQjQCK3XYfZ!OB2-v@|3D%z8v@8)C3PCvBhOCH;aB0mP)gB zI?lEgmf&y{RmbJf8HX)MDv@seviJ`vA*$}>^7$BKy7^JvqR&Iz?xo=Q_=xoxV{H3B zqvQkdrp=CuwoZIv(zCvs=jVJ_)9% z*-_HjdtgV2N`zMYt)(GdR2UoDHk50q!~t8;rQ1phb)jgEVfJVEsKljg!j?OSsG+yu zoRETN`Fu_j9iPH!)S6n$4D`z6x#k5rNLLybYLnBx9@4 zxJ>wclg&oIaBF`WA|=EES-f1kG5c^aj?I(o`#I)3t8TNCdU8mv+lW7WC+j)^B087V z9d!yV`f_KuilQj1rOY;J1S!s|t!w_dqR0~-xDxpvl2YoK+rVsxxu}B~Nj=t1I<4JB zWx%4j;o^!?V8eT^__UbqL^t!d(t8!p_AtSNFE(SorkLjs?EjyirNR z_ZT}QJ&LCg4`B;X2&C!6c4kzn-z?RD)6%~_L!j7tHia&P%Av8U7a9*zlzqH?k{GTo z%DV3a898x~x)lCkvmfHkla-5bML$jX6kl-Za*C60v6rt{N?(xQp=q_D!tnyy^cRa|$#W_KCtnw4y-D9%79Rc*+|OER zm?5XII1k0zYe-#zC28MVIUw;cjt%xoD6j`v?axqlVPflUw=^~7aE8CTL2rb~@jLJ| za$|Ws>IyN!=wGiS4K9@@8@MIZKhG3i!&L1@MjSAR3btTuHELz$R-gU6g^m}xrn3GC zORbje_yguIx%~4rdn>sTA}u0!$)a`f`r3GK`e_bdIlL~6V<)2*-fPJlN309jH*`?!=SEmkjkF0V{7%8OVkDV=et1%k-&bw%Y^B2qoPX!oFiNI#9|dLWG9LT@ zWY^E|w2@cE6pVeR6Ugg#2d-}u#3Vh|QG=vwmum}Gc2O9v0KeOyRY+7EY|>M03oqV| zy9wUn@A=TayKwrnWtpyyMY8L$@Z(PQ#N&}u*QL0wk#83fOE*Z+>f3MDxSQMa$|oA% zRIT7DZgL>jc9*q`Qyhr#eKQwFw#M2QdsB72vaGn(9uQE>N$Qbg+{JJ)_&qNV{x+QC zscU|kggCX@PI%d;b@Bn!TR-(P<~WlI+w*O?+4Lt~KE(@*EB_30w@@Ahf8FRS)k*2q|ktAE#ic+hr{>Q9IJtyab6%a||)*fx9IO`)LK zkuCB=E)_^EIofepSf5t%T)DCM?|lezY5ooVn{^m*?}5vVd9~4bw}0Yq1{?yt23#gj ztPL{kGZd$mF#(6NGb4Mquj* zLo?xN%`ObIr%A!R@xcVI1O!`nmPLIfh<+=fYrv#@W_^zUGbe^Z{tsoFokE+R(p!Vo zH)cA$GmfelfoLlIc0Um|K>@8ZW*}( zR4KTWkQi+-LbRZ5W~-md-Ycuh4B+ZE%5hm(clkxLr*d6O}UFoeA3c}OJ z%quEaEOY>(ekxF_=N6Vww|($j`#F2}@<&;G4)bpWanpS>N$d{Npt{?-q8oMS5c&Y5 z1KDJ;bre8+<|HG}QLNoLt9{u2W|YdqQ1mWA)KKXt(zbtudhQ(WVaAV}Lai+lIh)#n zs_G9cyYr);t;Jb;V9Mw}{ma75Hyl4EVY_2VR|`3UM>}9%=;7zao>=c{Ln@6<4#j)y z9R15olm?F^vpC~xX8*ZHb^oldxQ!%|JkmX1lmIuKIhJD9NEXP0fln7pZ-%3uPdhmd z{37VhOVlY^yXlV((NsrK(d%z>g-#P6qde$i!+=&LtdgA^l_daeZob9sy5F)QoiQi% z+@pmDnoz#21p`g)D;d@)H(rRDW1Ba-wob)=zM+WWl@N)7Z9vBsM3UE5qW zv`w1KlhXsasK|rn4Ks$Lcb%U;pU>g;)k!sbv-ZAUOgZo{^39N>?H;jT!}wVPcE#4O zJub+pWW5)qdGBpY&&EzeHFES%gw&NIi(GHEqlOcNa~oH2)R(w+NnTwxR}gRcYo_#K8A1*pU<_qxmUsN64VR zVN`_*@7r*3vo12TPY%g-u zOivGQThDbUKOFZUED&U6_yUCzs5(i!6-#@HWQ4P9>nA}O!c<+G;JzQR^DWS;cr>hr(|1vuo*1hmlO~_3oX6~>W zKrzxCaIeK5^!)5SQ-4$O*qjv}!<6+86oEHX{DbxOI#BQlG|CL+I(A1EBM3WQQy!t2pATb z6e$B)>AJ+tE;D^8T(}@G;LPqFtyOG~?-SEc*Nx@iVGJ@E+{vU9jN!Y?OQS=mU>+c54E0_k#9i` z%5^H!T>X%!w8sM(?(= zWPVwpb}}I6-INJafezItHG{`0Xj^g{OMGmSfMxPT$57%%`gRKtyJe}KkIF- zc!gQB^zDJ9Mi%tVyR%`H+d~`pDJL3nB_oqx3I^oj;KZBM+t{D8P~XY-ohQdy;}ofJ zS3K0aN#yQ%OcSiznF_dK?{3VVrUXnjSDPp^MHKMaA?D-sPMAGF4nl@%6EbqB|EvzJ zJR;FiB(uv1U*A=+NHECZ!=(!(;&<}qV(}1I7hA5{zR15;k z^vWjOL`*ekAx(#{)c6*n-8_!YfK%sXCX}DG>@=lSDkI|4?_?`r2x0 zLqqCd4mx6bSxBYy|9#9da^i7o-46Znc@gc31H9b>9)v}%FviIlPMJBODBa04zWnvC zRn`rEB>|0l?c+roLRD~an1~881C&zbtBhMu=2xBaUFLSem!)mki_T)-TF3DJ%?H@Rg^0RO>-BOi*t+x#F?CYriof7+G}mMppJjWt%Ba{HfHAg zp@l71oBCy_F7Q|yQwvdAQ{yNHIXB(ok1{J^(AsvK*_NDhX{r~`G>EEeAD?Dn0hXO9 z6!rj}KM%5;+c@3a-npfYO2S5t5RiNdnPiHB2#%0RcD%O_ZT6>wxsVK*-t!QZoK(hL zCNH%Dw!Z1EpVBKvEQ{qpP8cb!sSfREJH2!lz(X}N^jk{lM5(($b{Y z8t(Lc5&<6k-BaMM9)Eu|?=Nvo`ZC)Km9YK~srYJ%C})T9%~H>>%(01{jyIRpbdim^ zpbzka?S$wJ)MvO`!ilB7d|yM^jk=$~A48@kt8qL**+@ zqf<8_P&OVLf!@T35uC-@=r;s%?fm;%nPQwwrM#op+lZH!JY$+z9Dm*!zh`#*{!*^-PJ{D>Me(awWfD!7t4l1tu< z>|52RIjJT!{J8wBS$34(+;rSJtYR}21rQNP{+HIO^zG5O#&JK&XI-6uEbSYuHThB5 zAD0!`=4-=`?2Ioar<^TV1>n?N=R}im!qw#|$ z!+$+7Kgtd~^LRQuN%?PzhW>$C>?hCygtdZkX1cLGSO2sCHhIzd*FgQ} zOO-}LyK(Hdgm_1qNMNM{c%_|H3{D9*^|+(RuHZWR7`hoP`B-Bj*GAQ1&vH)1^1W2` zg<3rx8yP-RGHo$ieX^T%QzVt0pFInRE;#_^M--(8SeOkCd1})CvTA5w;LbC!Gq!kO za+*r@o<0EUE<2ecdVXwb`@Ev)>yu;xE!a-=1eON7vg%E$gk52Xy0W+Psg`T=d#(!D zUNvp%0Aod7_RFT$I?s*_UKqB#5ydih%Fj_-9-KzTq}*Ycf5Lz!{*q(?S6pnsG(GK! zxLww0*B@`a7VqZHT>2??3P{jdu0x4m&^>zcW@7R_)Vr@bD8~{eVxyo^lx&youoR>z zxer-S`_#`ux%JEI)tew_!V4F5XpBFZqp5iy9!m;a^!)qxxW_NEfE%Lrh-mOW(PsDC zqPSZT=t_kAB*zMuC!@^~2bN1^?eF!H9D);5yLhJGv4jCLiXayb7v=ecpMZg9ExzVR zZPrphwR(DDok?tDlimC8I~QFUF4czX8Z*PBjA9q_hm4s6;AM4erRnN#^|xfD>VU$w zVg8!!KjfF)r>f^T#gdGjEtbvPLIvXQE{#7{{GMH3x{6Q0_Vcf^u59wjyZZqz@|_%ZODoY%IEjF zd3GiSztG1m$OZBEQHO=9tTx4#66_}eK|xaa0h~WCPP6_AZM#Kn*Zqn3L35vh`V+xv z)D+IO(wSl){Y zLomsOP%H0i@VzuYjC;CttqJJw+~3qHGj&v>_LTaUtRkU1?H?}e%v0AujdFTsHZSgC z_#QLU{+xm4@_|*GF+8x|jw5uL_O!Je{e&1}$6msClreWlctnQgG|^o-dq3k|bBSvko3jZSiw7)qah zLXl!>5OL#o^CZH}Ee)dOUOH3v&xmD8J~r~gdFou$C7RdBp=b~}noHIDV}1lDPfwm* zQ@aw9HiR=7XG~|CGJeuz-1~Lh6sxVVLj9FqJ2gXaQ=U+NScH8HL~3|7+~7hf>yoUF#SBQmiA_f;}wQH zp*iF(HQ6kYu6dgZhm@xbP!E3Dpu*0$p?EV22Sc}gTCAHFj)5$*I=nmUV$LUbn-fi# zzfS2w2fX!Ap%84i^Q>HCwAS`JTkY!VFZC_r$NT2ut}gh^Y`DG}wSg^ z6XX9$<8@PG{A9MG1GiMQpCfdY{s(W93@LrdagJ8mCaq8&j{Eg~Umu+wVKCWn`1eGNzwWR=e+==Nf_TdN_zs55&9O=lbL z=NNe_hS)=_;`xo&i9}kQv&yU5$re(=DU(usQJ2c2 z)J4#0)n3@UR=5-}@lhhQ6{+Wo{)e%*3~K8O_i%w0DcVwkTX6_dB)C(&Kyim+MN-^@ z6e;ddEJccI@Bl$earaQ%-5rYjZ_b@L_v^VI*4I62X3t)--}m=C-K{1&@jOMX6h`Wi zo;Z}Y1DaFyJeIFb8Zb z`Lt5fmI*)9qslRKvh2S*rDMP09-Ro3o@~AyB$t)IVxLpBvG|--Wxl;rv|fbxW@5}- z*31574Jb6BmHGaTKGut&mHdbRUVN92KjJ=>*D;AyKDM@KsHF-bX%A0FsFIN$Tcr&* zdKJ+9Di%E9PXsqKgC!(AW47^^+*k4uQ}k*7Ub$g(u3#is%tu+y>E;a_|G3<4#P-*g zqfX@OpU}5)^(|R6X)pYVrdyysdAw|P@|r3Ky?75il>Ik{V{#|CwpVN**VoYf00^1u z)-tsZ*;AgeCQ?~OOpZP^fg0*Z+OdMQW8#Xh3#P0 zX;(&D&SU&W3)K7P+j&Jz#7K z*Ti^E<4nk0p$|kxBjz2FaQ9E{%)=lZ^EOHN^@Luuq|jS#hdNtbMszDOx<5ek-yF%{ zJ>!=r`vy~?H#pBIoAeO;nYIkobRR7!fnG(K`zL9UXhWWMe1qtf9Bi@Tb-p}W1wb09>R({H^i-WBaBG2c<&ya{hlPCQQZIXitj z5_Gu{crECMr>IQEKdQ28snQ}~p3w|CoDPHmKv_@V7o8wJIDo@U%#dWUMV$ChueIh} zj=7D1WoMh@OKTYP@W|XgMcvcH7OUi}w=qN9)Xm9Cn_c`%>!`Q(l)!@~emUX#1nyBB zNRQ0OAZo^A_e+(};iK%-*P=J8_G-=4wscPpA)7rnEA%b%R*qXi{5i8z2mg2?9#ySa z=N%gdI(x81s&Mh%{aS(}#x}fSp(X0cll~uiXPi-L+w83wPlL7e-M^53`sO7ESRw?8 z5x%khh2EebDA!uli5A6+P7bB2aix0rM1xi7!5tr939$)fh9wglEj1P0nFY9M29&2o zF-kE&5+F^eI*001aP9Qhu*IHE;e^^sxs7F&TeTc>&TczNUF?wVB3lE4&WNkwc8Bd} z)zl#@@z76Ys~Z`N2dq2PWoieI&G~Y5h@o{9;52#NO^3MT$oh*$>iJqpB2%AwLgh(gpk~JP>@?&G9 zhVFnX@4gQ`GbN>zfClxaEjmN#4GD~+e?7TUx!qY`{Ql*Uzw6KM-LskGzV1)#YJ0Y&d96i+ z%^QX|D|IyTUVLeqZ~taCd`h>3Y3id-9i3d6zBmAP7%^nXcRIY!uUxt}arEeUL$e@r z6Llk`z_f2Ip940S*tzP-I{K=EHSQ0H4^NnSVNNg3VF9fz`z)Rc5p#mQZ;pQRCyB)9U7%fP510_<7XAK3juhopMfQh%!8#5L zRS5KwMnOaL*-)wtg$Kh!nC3V!Gm}I&Yzu!5U~kaZ?D1Imu$al0%k2x;r)21UXv7Q7 z!2*v6P>i^RS!7*Rf0&B0Am>mT)zW9DEhwrJ^>62xeU$#~E8(YS7F#eppfCzvDkbgb ztrc+V2os}|^w<6bn*F^C_&t~JuVa1(WjyXe)puu-xc>hCPUkL2l9fCvPh5|Idk+^o z$hntsS!n-58!Klt&dL4yDk5nN$glgED{y}tRA%Q+{u1>LuzUD^1ijp4i4Tv+bvPLX zHckhdSyhZ&nhT52(_%st5CSpH=rj6R4!Fbt{ikn7i5V1JU!u4;%#!-6?;LEDnU7h% zF=|HDCbdGZQ1^Q{9upK3ODWuq2y)3K{$(qG=mMrjWb{0eqm%BguuFbh3DkC*rP1NI zup2i%+l|_7tMS9@%u>M7jIJ$IwkX)2R`ywQN*5zSHiHS&xT;948FxVi-oMO!LZ99l zhBY6Mw(WPZRz+Uz4;>8B+`um25K!pjA6$w{wd z(xz>9vrM}R$(W40Pw#;#Wb%W{H=j$$dzqIm9j(=yJ}C6CO8~KZRKC@dt2`coX2EBw z&XVd%EQnJwWrrNvK`m?gt4f!;PWCye@ZBGFlZ-0^J!b?S$eX%xK3JipJ}bJSKrWug zu$pa8@zma91Xt{Jb&1C+o;{_9o#WA@5;e6xLKrbqCvKqrK5Twd5_b}mOLREp;N7Uj z+*VHJHHB7f8TaQXRhdB@tf5(>+AeDBiLOdCgm~h%;^YC-qaTkgd4s|Xgp~!sugZF3 z&-Efb57|EB>swX09=U5BLWypQze9bzOgMH|0f7sL$yC)vW_p9H5~%VU`wS`-KP)WV zt}fJ*gO$)j3y4uB{(ABIzIGarw=RX(zurlxSRB6e`?U!nqAJ0AR{^4%d7nk{jR0n3 zd{6s@z&$IePZCa=d)998A$(Ia&Dh#M^0MD#F7ITVwHoq@oHnz>3$@3i^ev?UDapBC zM6?=)i7@ROA#E*5=6v>KPUczomMtGI0_?`kkukN@ThF-D*0XOlBYpWxeC}Y87L)-E zb!{W-GL{B66=K29soS|^Z7xuSWRC~-;MO0GUuktHGnG@xbEMAsdqPp0a?+-yWz4OE z6y>IE%;0DKtk~(Z-nM%prmY$7-oMFq3T^>igjAqIDV%5fb2HWGtAubNzWG%Q&Bi8` z?S9~-tsds{)jwkCq#m4J*;2d{G+Sz|o((^1*5>_>3KvY4s~))bD8n#fpSeIdm->B2 zM=hMI#u+mb1)ix;oxVCRiE&#DUx5!27_Df$UXD6~evZ4S*%PiR-tB<|G=^VrI#Ai8 z`|SeeW#s*B)YO0;C@*D33EaY%$;MT0d1 zLIec)56z$fG5a`+;g+qdp|Y2XXfPtKT0iCE(${Kh3VYY2k)6Jc)%x+ z!UYl+BeKTDl$O=xOwSTHwZvPJcTz0mNpurQ%6runVaz1x$GJpsKN>3qKLR;>PpVIJ zMRJbCJy6(K+PpFuA0t**zr!UD{sB)3xUnw8>lM~|PJVkyuhB=cfA8l!F65`^9it)9 zYZV%4#(NNfa(6{X71ICEUK3=0|EFHVclaajXKeiJoE>RPVAHRLk3m71)yyq$4HCPJ zy&bU_C)|HYho-E-o?^oW7>7cn{V`p{3!l0b#& zE(VMui-^GeDzaX!|IqC3LZ*-8Jc9gxdydl5o@2Ea9ln+EEKwE3y2ruZt|Zpi3Ls2B zySSDZn*L}f%^<7oEQ=DD7`g*Gd;AQ3`EU2F=R!`Z+H8GUGQ$Y)Gy2u#eUU|lf2sDa z!7;l+#!b{Uu=0(v@g=3PaLP)F8^{(Kycx}NU#maqqGbzd78-DqEy6@FqnAhDEOvit z_R;7o(!O}&oFl@bX0x{VzR>z5V|>Gw-QVWrt0MeZ!IS%1os+p0h3tgvHcwt>z}Tz3 zR?f zJCe*cOzm!Rpvtio@inL8Du%IS*^;BZqW$P|O~;}BO21u-bmOhs4vVkwsIsH$%`(0s z=Fo@x13!I2>qM;kpE(UO0<}+nx);iY!=us&4+GS@o+oW_l&guQ9nzZkBwz@_!h#pU z2E^i1G-KP%ClDF*hbN7gU+CmpWalL}AhpS;1HdkUYM=~-x zY@G`Me;nx!d={ls2{SVpU37ac&C+tVg0)V(Yn>@<6H;yn@_Gh9`u?FiAYDI3cE`4vtFWy<3X?{h$}~`UfIE%11;+ygqG_24Q$~yv^r;b0D+%2#sCgd1UOq zEjRbBZg;Kbtfy8o2~ZQ}y|}SyjGr&zRBfN&q=k0=KILYGo4!$LapvG3m1Pd!{21yH z)kX7k80o8oTZo_yNmS;>dBwe|B3aKmks6{}hCFP!ohN2ndsjb$>seJzTW@`H`MSUf z9n_!3bsIch@kvNAIV*sGr&6O}qA4^atmRg;L2!X3lTl%48h`Cp;7Lfc5$Ul{y2s}K zT8MIsI{l$HVP^XM;7~GAZMp8k)g7w^1Tf|N2vefr;8TA2c_Ue#@{MR+Qam{RJar;|62){X(X11X{Wq8G zv$7wPasPN&fvEv4f8R56<59%J7@6860-}6czb=ye5iMzPU0t;2Z#A2(@NJ?%>sxr! zGmUH=xIYU38gFpEF=&3K-0Prz=jhqEE3(t}uk%mR{?s#*hNrI}L)8W$2V?A3vSJai zdq*iR6b(~9jjxV6C3qTo{7XQ+GRUEl+CiWz3%%HHL2 zMP%nxsCCVC)#F{-IZUsQD)m!0RmN4dgK(s_3Ly(|_gICaYMG_jF^{ThX(0m(r>x&x zHY^G0Zy@QEkOhfG1gtrlYjzqVu7y`f92aH&LpwyKrbgQ@{B9w}z&Ku#%&08UqF$PW z)T`L{3V1v-%s2{|3X=H*y(z_BY3qrdh?-ALh?ncZ7Bf0ImqJoDS5#9GGZXV zzEZv;YJDrdC$%#Jld(h16o5%`0w-FDv#_-xu-i5^==Bt67nrD`~G{I;I z;?>j`iqGI`cvDRG^J}QzoE{<-@9io}8>?0exOJb|QkkWxyGsRf^}a_Vjnb22N>d(! z4JFZ(M*}#FSx}KUw=C%z=?Z^VbWk%T5aXZrAc~kcfEb(iv#((3;2vDj5&c>4qskQV5kL_a6+!@ z?+_&I8~O>6dC%;{ZcJ-R0Q-V5RwriXQJr=6`lBIgpco(?v*uhZ%$Ize548x{i;m7WgsmK3IxU>&`a-4ZVUX zxq-zq$T{BIl?|5Ed;yCEoK6*m$p%t$=n5xoD%;ebU(yPg5=TFJURy!-g9LD9~kK?H(FV?T?_dNekr8XJJ5YtCV$m&9!}|el-eMcA`oG zNx2_qy!nF)J%u-+BVa41=g*81<gS)fmDla39#mg-5MHt?}k@>zQ2#7w1&zI$lSvrxED zTz9LIvHH$^j}M#=b=_&5=qqq{^_sa(;&_6R;utv60AXi!3S%_dxsYuEgU3VbsBk5W zcI@s2h;bOR5=$GWTBHj9 zEQnpx;Qli;{+tATPbHD1yvO+-w#GE{#!$*P*0I-}&|5^Z1hhB3{lJ>4{qvaMm*@ZR zvu7uVse*pz5UE^cGYRCFQH1#-<1Lg)#Atfv^6eb(&#N{QjO_E14$SDZZ%jysA;{hV z=la}?Cm@|12D#U%ICLJNWho^=$+vo4wnhh-4gVD(Zu?LFr+(?9UJzB#dVP2`cY*qM z)&I%CqOjERX#dv$oU9U{nG^lN*@~AEJxfBiAFj9gOwcd+eSuPINXH)QYUY6HdN^b3 z$egv2jrl%5?Z!bd)jcX0qrDP%;+uL=rWe2e&Aq5%Bj%l-xvsi3_@eb!imeLUG{tMl zgC$(<{a$`_(%S{$_ya-KxNcBGnqGR>f<(Fd-#G~*q#J?c3HrMsNxVSM><`BtV6}?* z0R29uxTjeyI^d*f8aFF>nUF{&U@Nv`iu} z99}MGIxd@VvTIW>!En04TrTfwZ%?|9#<*Qo@w=t^CM`5#Lt0(lg*1my>AGv|Wh^>N zJDwW6VZsv~Ll}J~(6jGovqUbx2)bQM6!jUW@EAM3xLNG*ZJG3BQtASD7T(J1>eyOq zbZK1V=NBGlfIjLXBkFFqSHKOiO%~T*0U&#iQ&M>wj}o@cl^;dUkE^4viu4n?W_fpQ>>E>!}Y@2dN5Xv|

$3G)t=*g@OkT_|*qDlMGSjJIRp~TDT2&Fx~;M1+dOtB3?R3=B8N0?Q>h6 zzvgAuZcph#mV*UBRw>0R7;_~QeHd{8^WEw<)I}Ahd7`<=fOcrP?^-e2h>iDu-)9T$ zMh_y5xzrft;okY8tJFUKhPwrMC|lr*FKUI{&EJEKddKR^RsB4AvloTORldL~O%H{t z-M!hpS=()qNeTLhFqQOMt6I+FOf;G>MP=h0>PafSwr-;364sR{WmzG=w>T<=KlF}F z@#F$zHJPIQe|;0ZF@JRzzasua^eBUe>Lmrod10UUtmX-LRl`Fw1 z4pUb0s|<_FTY9~3Q9Non83&n{8UAg^j^**h5{$O#KFX2Zp0x9Y4|C}16lkMve{%Wj zxDUOY?j+v=#a0vo0}JEf+8ZWNr=NwvRu@aW6c{Z5l$VrtX$K<<__jM6ZR!2`oePjn zE(8~u+`d^DX?n{fK>G(Gm8&&!?rM(i`($9D6;>1=r32Ix}F9eFc`!iMa+1WJ`$Cy zdX_B>nZE`Io9OG)%ou&={gC%Xx#QjO(aUAW1wQpBie&E1M(tVUF_}frpYRa*d zD1BSjpw$-lHvO-tGv1wZk0gTz++;a5sx#D4talafnyiuRwaJf*a3w|NHxDqy^i5D_fRrrZPA`HjChhygvZ;Wl>w7c&Z)d8T4l2mC}1x0Ev?l#XeR4 zV0~e~ul6OAXaId9C&qjtd{Gi|e%2qu%?F|U-O04ShEEH0?yoZLbq{|bRzZ1aQP&TL zGz8+jF*dVyXUCINCoob{GNBc&&}+O{l1hK6exyOA4_dzawyxJ&IY>2<4@7}P*XB$P zm=U`QN&Rk z==&>)-&m+zIcrp#Dj?PojLzc58@y0W6c$F_0`J&xB`@1xDV=;6`9?Bm<#6bX!F|Ih z2L7fp=}4y@H=<%t4Y_1DEzD45e9QQrX?LaFR(@ACg7uEWB_7}~WYXfrL-lI1fF!2` zyVAHHJ94w}6GrX6X@ewPUD^Ovl&@KH6^}{(BlE<0A9(T@hqToFk)iR=)c1;^z1`l~ z(lVK1-@x`F!Sp~~oqNTTES+bH#`sh1R~#V{1YXaK(`1^{1`Kg))|4BMOjRwHvTGq( zdVrdaapKAbTT^O%c(=-bXvB(g8|{3^)sKWpn2gO;SPSgVtgrtRW#r86Jqv6qvPuA1 zyNrxTgsiP|G7A6^T0L&^2IF{99%mQl=d*fS(b~jTxrA}elmSTMB zc1v}*ry8wvQ<@eF6$OWHwK`~HIcGPNL&HPqWYzW~-8yVFJf#70RJj~yG4|mTJ%@rG zJdk>jovJIg%XlaV{X#E)n+VZwxR<4SW zMYOjzJTR58dKu&tWFZ^dU%nsr^W}J%6nr338Gv={aZKx|Sx8J9X!Z#0|C#exx-tgT zck7G?eK3xB31?CM%*X-O?lw|w7=?k~94VOeWvoGTC8!iA+h?Goeapd3j*_Xca_~3H z;01iOH&lA>#t2xYYQ7hEk52r)3$7*${&h(>^*%^HH3wYOF%4RF{=|AzC%E~;#0B%y zthi9Q-5%|~zda>+RsXnoO<)JUv-icPj&G9WxtKkuyZ*;ui5eKl-EesI(a(wT6bcCm zAYscmmSSPsmIHUp_&McJy$a#MPJ_0)o~>@Vk>Oo}y8c4G_ASmD@@0yet<6)qZNk!>@z$(td=v;yvKOg;*cfpbcjtE(uMv@GiS1eNmps=W!r^H~WB<*^h}qw+i}?>=ZBc0d^M> z`c~G{M%3n!R|bV)(HtkI(HAi9?mBw@-&sU$z-2m(OBRl`EF=qTeJ$P9Ddx2hK3Ii3 z_sdC`R;)vnA)RDWh|B>yu;Y9@?C9#6%u5}G^a=e{h1=c&D$t{z6nB`hQ8B5b!=Ex9 zs_W@BRhI!AeQ9wDbI%>GurAoN0VvZ*~}2av{s zNpAglX2T*r;?Nwpmm@UOz!#s=Fr3h$p>lOhw9yMc$X|W?;#$RfQ`)~#-ZuE?@Jy+g;38og)SP`w!v=S21XnfD#a8vnH zgxg=Y>}z9O{bKWb|Hg(^)f)`BCK%ZI0VHQ* z&d}mN*>+tL^vAf~k#vHP#qUO|dZo=MiZKIYII5>ZQ%G6}wji%42 zf5Bz~ndx1(sf|7tjwRzKa9QEgz@EFjq5JSN_3#|1b)57t&Q**qI|^^kt|~@p(h#8c z{x_`~Hl!%uF8?)o?0|;2zF$cny{_hCMfw{<<76tDW|SvldO7WiOVsXUB73=2zigg9 zdOibXvOG+gSU0rMW6V1?&Oxzj6(^aXg559sfI~>F_8U>KQ@-y>dx={0!IViazkAu|uo@4@k@d3iqR^3Ov)I~n&E#7c|^kZm7X+KnVBCLKFNkn0|} zK0TFfi5f00(C=RD@FAT2t9jYJQ5~Ly*p~FK63kwnn>ZLOVg9p;V%9G8oLHH;*v{)p z^gRv86+_mQ5#wmejUz&=uCJubRF%+w%EBrLh=W8i2eXM6nEyjFq2s#RsRI|k(KMol zGi{yx$%;YzE}pHx2s!qSjjrcA$&6zC>mry#0HF{AkHxvc@ixY$DIW2{~Cnw(_Eryr(o>SN6=c#ZJ6I~(Cfwo+?#pb zH~sDnhQ!R+OweBB30|KNHSUsu9D364DzLF73!CELLd=2D%m!#sCayNr%rSspjlbkp z4T4oV@-g!TM%(k#-1rxn5DkG$&H~h2m^dtLnkhsaW9Am}${14%@GBnWu!2M-_)w1) zX!gW3ctn6&EpJJM@B$t!A|z)A;srIBoUnn6_;nQ@XqL!Oq{DqGot^ULa$ zddo!4v+(W4Db8w!?vH}CXx2GkOt{AD!-dg~Jd$@@`JZ`nlul8mGl#T)M{WzNjy!a@ zRp5>(&h^?ErY3t$G74Nj+)HXx5jR!$sW_z}@yR_eUKJjJyV;=S`dparc;xW{-ssSK ze$*A}98_@^f2e#ZNAm4Q%3So^IZG_Jg}o-9^d?|9@_Winc4tM3tjgPMKmK87uGLT9 zBR=7flUJdD(A{;a6lCldRbJ<25zcfey!_mMXD&1x=Ia}#JmJ=cR$@MD$SYQ!8b`hN z=5IEhU!SY3hO|k;nxC?N;ztw%4+378)}Lejq}eWvKIqP(UYi^1kT}bR7nTg9xL-(D zR!8JWAq=4X>-5@kjW5eQXPdzK&TZJ*Ni>09Tv-|c+vuMyYNYb^o6GJFX&01o6~js6W=-dE)dM^mcBRgWpThuWC- zMKp2cGwLZjOWXVG4lCIuqovE7@5`93h4)R53#-PN#P040Zh^Lt4txK>z z@Mc5p46 z7@o@Iv1Y1D^;G5HtB}oMmw*a>rE~Q%D|}Z6O4Y^uSl{o|lq;>BjR6v;hCw=}^Hq>5 zMs;AiOPt}D{+<;W8ol3GT02m|)NhA9gQQJbYYm?b+e*jn?QjEUr2$c=)g8UeDJ1FB z>eq+4yoI`bNZ{hnkjR+Ocfyr<*i_0i#4(z+5$F7Rri?O5LUBiev7Xg5lPkuE*sp7( zU$>C`u{?A}Nb7j2@Em|}g7^%?I~+t;k7YMvor42d9LeIIjU_>WHp0?qXuM1sZ=7Na zP)o^fv?rcldi<1IQj?siQ^ER^*0j2}%_W(WGGiie@H9K;xJ#s~*FGYaL!wcnmOqQE z^bn?Ua-()RGPY7&{v{gN&eSNr%W{pX;%fiD%graY{2&Ru-w|0`5b#;1i{^0VQ3xv=cIP*!tlXK9bm_)x7RX9eEiu2hUQ6kA z0VCXheBH>AEK^o!w;Z`qday#1uG^4?CKs9Rof+R;yjQI8W}n}{ROQgy@kgvJML%SG zs?923ec7}@Ef5gClwH9~K5z^AZa;KfkT+xc@kvOfb>>mZ`zv>dxGye|zm6z%@tECa z+kIT^Fz%t##`MC4*ePL)QYXX_mTG0_Md<+vhxtzv)QuFP`5bF%LByci@l9f*IQF8C zX9o%fN-_G?=D9Mkj!9~;o}=X^PW@TmJjbDzAdBp($=;+16X(@sa&wiX z_vpq4NM%bZ{9WQ9-#3s4zH0yKzr7i`qCsl%8hhojY=v~{GxGZT^ohA8?G9X$!B9Z7 zb-9nX`O+jCzS-?cv)uXv%AXW!R>V54Y{0>-Z1OLny90CbO8D&nJhoB9l26TJki@m> z4Mu`AN)1orFBfXspei;9aQx{Cq`Y*w3=zQFy%?s zy}Dz#mAMwb?oA>@R|EfmQG2>k^*=PaH`(89n_;gu{{Dxi4!@lJybk0)U>{X}s7{3x;Uq)>sqF9z&pN3Uwef{w~d;Ny@8gTfJVl0)2uFFiV z!a%LafeSwmwV^**od1AH*3cz@=H;bP{fP5}j1X4m1-w{^Zyaa8lyJDaJd zGWzdh;Gj7|ag-KV0Wz1T{2v;9_>6yYZI*mG?R^l0Z$tcWR-(+YhS z(7(PeJX0qOkB~Cbpt{U!6rnj}{B4wfuP`nQoCakp6hzO}ehSSYQs4d-iTj}@CkU>8 zmXqRGKDM#T#I|qB6DME@QU&-(^H3~d`d(KqH0JVNyh~!*pF-5?Ab@2iVyt3OU$tq?=Xca z&zvy0G^5GB8RV`DI_#^he;|?bxa9=!&U}fK8PoeProyp>%7^nri(62?Q#CZMu{t%o zb#$hIwa8JflYGYA7Dr#NKH{Pj%=ogQ**-{fSo;XvcbRF=B9_#rioAgf{QXF{QRf-Q zs@~r4tx=BRqnm+VVI32jnH&m3h63a#^fV0bi|Pa+27F0&0iKRvqocl9D(`v5G4ZcR z^Tms<;#9i=W;a54`7>1OCu4kT1%Yy!rMjw+-Zkj&-CN}^vpsu(ht_i-W7(Lfc^uJH zohgZWT(77aLgC`b?2L?PXqLEJeh9j7pM#(A2en>x_k}j!g?~~#hwX{^wjraj8C2TP zb0U(zOFlo%-x>>UI;|!AY4__D3h7r&w@hJKLO(P!T+e;s6aM@RzNguUmo=b20elzfm(F@aGQQU!^GNz$A3tM1Jb}On2r4BYLI9vTR4dP zMIM2KPreRbbo;NW^nq-^@-*tv&MUSRcNGforUs$&#pF*e709wXsb0JaMXri>%8YA_ zo}gB%N`u1|wj9z+^(DYc9`fn0uS@qWW)S#?|M=Nhk+l`jS< z^mZaT<~LeKhB*~j-u+~vbvL0cC?0j#9(qG@8hyL0(M8Ff%lk%YM6nFTA?=93TM*}r zusDrg*3)4fN(;^Asgyl+@7)J$7)<4=A%G#4Ql7PjL0;7=d|LX7Nx~|`+i5#1aSrLP zMLzQ{3M|4HJcrqtcHvqwf@(dHStKcv5ab}yhP37yxcp)xr>WJgc4nxB zC;Mh_&@(0b={kz(HFivGwkeb3N%w?KN1t`ZND&+P7EWFbj!j++4DTBun!tHLLks+G--z2t=`eA>E4gY7JEo<-;934 zC>L|2^3zvaZ)X}B6i2lP?3;PLQ^C`%sNIt-^mY1b*C?r04ZVBmK84UAaWG$m6hhTX z4L?ZpCk<5X>TyGLCXHJvW19i%?%Yl5TzUaA1F2uEvMy&PWs^Hr&u-sqp+c|AO%VmH zoj7NeUnJY~q~fcM9rmBya)ER#kU9Gjhp$uho5w!`#%MmV`bA zK{5>BJ5eVqZM2h52WaqldQL8JTNA7I2A=qo+T}N?^ox1pCS&N?Fd_@eV6s08iEiCT zF$90b3o5YbyL*j#K2s@vS13{dCr#(cc=fntSJX^x>9)ksxW~TGsnn9N?Vb$Emo0#K z=u_pqO(j1b{cIN%nOJ&g`f`enK3LA-0Sb2S?_m8yh9VeH&~2Krnw_3Y;Mt{Qc|F6| zu~M4kL{LIzrNsPbHnjg^pbxxq!8Vzjki^ha<3sy~Fqi6mMo6syuB1Xq;y5s{#n@M6 zU}M{4q`Sdc`tLRX=zQHX(Fc9!U&*sR>jSF?Xc?5&zU1^pUQ*r1ylVPv*>dMag>`<5dZ%$2>xwiEB2_ zx~FJ^JT2FOV)f3_oj*U!t+ym5hJ{~<>Z2klo^w&hZ?&dQC`bsEq%+?wEk#X*ren4W@)F> znBf8mn~^W^vvr{;(-G02h_mHOCf@rB?^_%RFLdwBc=)?M|6wcT39SfDIBy{7qx?6*)t@vfT8`xF%2HnK zcpF4L6|X#!9E(k&D0S~6CE|R(wC1T{PK&Gd7ui&$mx}v+*Lp1p-&+7MdIDK|bDr42 zRPdQSIvw~&>)4rZ3&_m-fycwFk!$S>uXV`Y%$Wgvb1i$wRU@|LP_3VZ9d){WK?6fSXr* z7Z@O>_8(f5zsi(V$||VKC*{+OB|`AlUv+ign60l6p)Z#`ll9J*%(>0C!k2il{kO|u zQ~aOid;M)fWA1*If+cU;vO}d2gRMCa_C7jXVGd-fMkI=`)xe9YHwXQHW8rM2@}GKf zGW(xKKu2nf>VQCDRi%#g#!132Gm`eE2Bd)ZB9ky5$2Yp#iiOeml+E^6c?*mk@?)nP zdVsAJ!t+8X=OErC9`$>cc3y^@t&Z9}65^VX=>8DkaM`Or`iB$O@7Qz{Qw>)g0`ZWe z@PTja=BW}m*MLsg!5rCzjBha_F89TDRku#fl)r6Jo14tGgHp4l4}J8-;jo1EpL8(b zPv$%ARK_9)goDE~&&c>{QPI=tK<&4ZgMZ;LFxXW*KKV^^&g>B--(KJ=!S@3%!~HRy zugC5J$vqPY@lM2_eNQ-ca`TPvQa*)XTeUEzF_&po_;NTp#FeN_dcH;M>(5aFuItI{ zOkT25L-e`U)QnzPx548zYneZnqk#P8xcj1J87Jo_J+_?v3vDkx%q*(k1-sppZ}b+^ zCr!i1%gIcJ^RjWv9@3WnVHgU&ZJ_CZzsjJ%7T8lUH|#s7F5AgE`w;&MTezA+swIsb&ta>zYe&Q@z=ft=oPx`k#{mrhywe&0*xZM%dId#u?8kku0X? zMCE>a1IMS`xXevbq566CbiIoP_jd!m0WZ!@H%usJ*sUPz+!N@753_3vD{-#HCRQb- z<)rCvyIfcs-IB}3WU7^oRC8GI=YU(u6Au;?CQjOkzJ=^ z?k{8oJLx#S%xyPoh_A}~u|>Pl`28MP!nxnf|L4(W{azeU9cMI2J#Qjt{URTg7Az(6 z8up|Z8fdLRg4$^pji+rI5NciNG@|;keNN)RI9dnAC-V_LTT zq6cFiw7vZcu;A5|xWAJ5Tlfbors@nx94-VWbC$joRt3EXMMBktZHo_Z#sx^6?i76T zXWW5~wAX)q%AaQ&MBdoEtocn*%||t{O-KM#b=f3VMg$OeYWBA1EDir3pyYt9&Cyw0@}GNGGvOXr*f2>&Bp&wI-X5w=~7#*k(7NK$92Tx zNNV%hMiz%2SsYPnSe$~}5bB~X1KLLQ-6K3XE=aWx?8x3$0j%kXAbbBVZ{qzcv@^Gx z3d;~*9S_~RSfed-1fM3v=RCP9ECt*%Wu8_<=YEiV5wS`+RGC)awEuB@Rk^>(jJ4dv zQEX0`6y5Ph{XTnMgDQ$H+uMxz9N$c~%aeuLgJ}-4Y0r+r%t~0wXe^CEXLl?<`mddC z$-ub6d3>I|WZI-WJ+|VF__wsIv7rO?ed(H=w{GvdGNBMed z>FSuS*ULMxTxCr;A>*yl{j|-~+QpNfek>bXSAkf-r8eH;R^MU>*)(WTgEv_Zis;t{ zhNi9qX*(tBOAJ`dK70~&1BRXpMQiI4IHO~6UY?q?)r|`Mn^z4zK;69v(EzFb9tVo4 zc5(0TJw-LLTkEX$)&5x(3GiAC6s_?B{y=1Z?G(o0y~#EEp^!7c{`(n#-z#Vnjmkyc^fNH9kj-JHw8#{6RjuwR;t^kY!9Kk+yZ8O>=lO>emeeBUlv`^Th<-F8 z+4o|^aA_;8lsS7R@o}-xYVg#QHm;B=JnFqcPMeN$IT&i3{aU?ZJAOcoUp37>72w{w z(5}%9q53OZEYK5o8&x4md$uL&Pqi#*H8H3A`Y(winipsbS+!fcFDo8B;8;G0ahzzK zlUiYYl8MO=Eu~NP?jyv8)Vgy*0zz8_am&Q#n2LZYViKZRQgQITg4y+2C({gSeO0TC zJC9P>CZZupH_?CHXWxvf+@WcXRscmk;zOQm#6zr#0(goBiw+j#+qNw|c|!;Gdm(p5 zxTLAqwvsFb+pdi>kIjMk~TYq4C9SeNuQ84txjiu~a>+euMU(8QS1e z;;P$=va$7#mDpAWE1U>5It3pF&7HB<)W8P)@|k;!nR0=YxUEL<9jdzq;9^YKD@)~z zaKBk&RY&c9$_?>{wmKJ+7jSuvVjsyvO)JlGrHSqtI-_IGs;Ncw=fR10guCJCG(cP7 zX?>%CE?G}=yEIQG4P2wHBD?PtyHwVUSrBMJ#fp)SbN9(xY^_A=4oJ#inIUJJNbxAw zY**TSU!eN{Xq&R5OtJ`ap|TZw#;=ksBx2%ed$gg`-0b%=0xFm8p7Lo@-a7~EJsU=% zDgD2lRv7<-t}TkUjsNg@^div}8Xb+XTd;7KysB5L>Ki{5gz@UI*S)1N+x0~QvMz8h zw}pcUuM6y2`W4|;K&@|6A(&)={r_G59Js`>(AS8SzQ)eqk$$@yHmS^q0{GKvk^S8J=cSt<=#4y99pWg z@cVSZ_)nQbKYz2{!xxQi%oV1NaVpABA(!id{VSawt6VaNqy$N3_@m|ghOTLvwY0^C z28mPc#k=}ZK?Qd8;c_e@c5q+%hOZq3guZ44HM58Q(l)fD>Ixe#gH6DC^+_3&U8={+ zzG)0`r=(vJ-)qV)p=K~(vYnWJ9hn+c=x1-iWBrZ1qc_o?r#IUeCI#_D<2r!5pa!uD zo|HRsvt79Y(Wd-wWc(#)3Uf4Uf%7hZ(IoYYDM9N=U)45&uyVq;5;s`uIW%BI^6z+d z;yvNH0bwHAYC)Dvmg^Cx_HZX|6Tft@;z#o;{bABv18^s6;1}~kkx_h8rHsj`(t1MZ zfSVR~zO%?mukIPwy1fKpT0Nz0y6Y7p(7llLiv4ZsInjdmaSy4du|>!&g4+BEba#e- z@7h7)mobYQq3`stp9A-ndcwlMJIX_p)_^t1u7@{{BMoLVsRq%@)ZS5ba2wi*P(&bJ zbiPCrY+gMCyuX1N*FQ5(vqYkV>r}8p7?@KWEY_{xv468C!0* zPb>Iy+Wz#Jc$}-SXIU1_G^2`WQ3@xsh$Op+zi{!cCx%`=xkf_>nCjE1V>aquv-SK2 z(HWz-C_JS1VyxU9X6diu6)0Q<-ly@J=p|O1%L5Gr5$K+>RbJw z*FfmV7?~jrQ}&4`Pl;$%3_}gVj;x#7KaX+7C~Oy@B2hhrTOD%jzXnWSR4T=?sJJrx zs?_B7pS@scje6=%IDVulX;=H@s1uE#SAg183;|REjBh#~t^Na$(f;<@=R@2jjf!J> z`s<&vX!bjNB54Y`R!vJ>+AH+P<>|0|eQ5CJvsCH~pmgr9>cyv7nTn;63YR>1hIE{A zLyNQ=CC(Dn&TbbK!0xgwZeSv=#}VfISNg4pz>*1uUs4RL4o~)9E=fx6cW~?keMPH_ z@XT=~ZS#?|6~B~oC5&P%Q21we?Ef;Nb2W_ol9}#5uPl%t|Nb+A;pVyjs_n?eyW{w6 zV`Xneg{sf!4^IVVdSx>O`rV(ouQkN_{4Aq00@pPH&t{+Q8%nkEr1C87E{C}+o(B;@ zw9lp<`Wo1P<|xu{%dGjiHJ-QOR>Sb6u>eGXmk;uv%J3(C29dv$efUVsT=Wr!hWfuM zsn-t~618Vu{obmyEqhj9%3Z*oISLl-mDDR>ocVhB7Y%!nC01Y~ZcqBEM2;xSZ68z- zUA$y#E1+MSo1n#8_m-pkVn4ba&u(JRS9LeXNH0LvPqMDr>GWy$+VZjn$*7bECZtlR zTrzz-t{*_^YP2RU+3?5sYRS zj{?G;o>!iBg}>;oWYSKUy3pG7-HTnJt2$fu{`&*}v3;$5TGslX)tM%l&#FQpW4^=F z4+f^IKDen=ixwsqmE_Zx!7e3Ze9d%%H8Hc=^cyfDEIsjoz}6t4N?*qbmAMWn>R+)~mjGDsf zpRE+D5ZAn*oC6S)2NW3W&aNJCs~+oIu_q}yDFw>%1ViIvibBgNZ3az z^dw)P6bW-3|6MS5t7Va~k^j}Wc-YZ%&qs=xO9z@O6{;6^e6(xT5^QS&7jMn8BnJYX zyPw{yf5_UES@j<^oES*Ecw65b@t@ebghph$S zBSjZ4-$TMYy7-Bky92V7F%S)TpAzNp|fuo<29qkh+Y-v@FysQ+0n z+BH*@*T#UjY-FwpMld@;Is`_{yK(c4dfkdZ90~=5WmD$>!M|u!YA!`gtu1^$r##0f zLN%B582Y4P)y9wM`lw?;Tr@qSp@{(?m|(eqM*7I!G*O(2f!=D+-%a(T4^jd~U@t_s z{tO?~Iy73(McR0IMp8RGi_6DBf3{U5U?;kFBntS#a8yR!-ktplm2hesg7h!!F(eV3 z8-A^Kl@;DFMy_nSDrtxGGpmX;#6-KQXrnNn;>^z?#d-PUMFq7iTbJrkPf?oI5wLZ6 z-A>zcM;T2{PJB8g9_zYhUn7<2vkvn8S{~l9mTHQ~Uo`TY4^(q$zj;Q{>3wIT?Zhl* zi0TLA@RjZ~H9x%lzTtOO!P1jPO;(3cd|7c^=&4cu`=E}hcoZ&Sh8_7j!76M>Awrbo zC>u|TrQbWU*iMzy@>VDEetL~E3nB>{q$hdlt0PkAZAq9$?iF;E&12z z(>e6N87SS6#^K|5+8an(bIP4|GMe40@qN()aA@RkGP!1^Os($-IPAJ zJhiU{-s}xTk~IE4joR7{%X2Mo6$hqWcv{z}mfLak#9ZdT|rk*7E?3NJ6ub_&lBr>?4ai^Yx*H}auB`V4O zQRT2h^rE;IKkfd#qc2d=xG_pbTu>xsou;)6@$hbSfd_w{lXUky3$##1r+@L& zTbRqjyWy8L$N59p$#DcsLb9mW;h70v<*%i=nSw-4jhaR@^emeci*O+?lTu7gaQ7p# zzlJ6wqV|XDd{5$2M5~XA%u?Jor31x{5#>wBhD1KifQQ8}r*T^_s@etj#*lM-iln`o|-9r`q<8b}wBqVLkt@e6;9uap0*PM2lZ)I|fP-h@xxYVvlP zc~{gJoKzIAE{XDxo+!N_*OUZ@mns3;whLwJD!~O_%Os~QHN$wRE*SG-q^2>m)@y#E zJhjtxaf`xfylJ;tT-G|Z?Q!emDf|h+MyYuZs#C}LdOQ*ZXW>tRW3DdC3l%4Ua^q0S z>gw?-k5v$`E2YeU7|h5@DyQl2kxzCWFsQEIDS

rS z+t$A}1j;(C#d7p~V}J3*(OXnd@7dbHld{!~*KK&%DJ%IWaroA{r%|A^y2bPwUB@^)(yYHcdk3>~^R zV|6!wEU;LQD)cDSKjtl$d3G8M;;e&lEns`s$r;1|zR*%>uA1elx*E@l}KqQz=^K++!-r>&u|cU}GHK z^H=%o@jr?_E%++VQwmcS#z(K(A%^iby+mU}s|I4DeWMY`Q%Jb(APar%|)TM9$k zq875A)azx|>7q8?#{)v59X5{@au7Jxvetreb`-`~!hmu?5XZY*h{$Dk`f=pE z#-RmSC(D(A*wZd*5{^opUSAv`b8?px-Ij* zGZVW-9X_-`&E%CFC>h2cY$F8>TBjbXG1P1`m-4jLE3DgX+-JT=O;^lS>~LKntNYsa>|&i8SrpoMP<(ONZ@ z5BfphuSPp)#7$SNvzm+x=$!S0fMZ_#^aZHBNN-VejR7>#NKS`+=R6&WH`PuY8#I+t zHqQ|p^fvEqW64@LbDltPs{1NoA6L}iF~>X5k&0n=DGjGEuq{fl-2Q55)#W`%%JdK`5Z+AP{oN<#7$ z)a(u!i_u4Y1f?R1yDgn(00g&Qm&$^ALxF;%bQtGykfqHq~ z$|^Qr3LgevS-TtqomzGjYTJKM6wpXhzZ^^IZDNUlZPZ%-+10AYWIWY(eLPP%kX;Cy zy@Zzkoa+@^8X)c`qn>3sCF@mJHYeOo#OKweF181mI_C4w zecu%LZ@I#Tc*D1aQO58`PukZue!}`N9uy>nsaHrm6!N&}dU`mGrucY_$ zVKwbEwtX2EX$_m2F8(u;>GIX}n~S_j~uE-CEKhHZsl_r18D-_)UoOYoDK z7S`0#JcG<^XOB_#$}cTGZaFf7E%T{3s-wGyy5rD6Hn&U?=)D39?nrxd8g8xF`{smU zwGJ&EDdPQK?tjq^_GhI`Mwj~GW;gl0{5L;lWGD!7$zx{4hAX85o^NH_>{DB3u^!)3V3@69`xAF{8Z~vho&;s;UWMNmmF&up zm3(TWQ%m=G=U%#cefv6fhnLauFi@oCi{U!{7?dJOx$ByjtSii7Q1saN6Ije@LGo&yFSz5flJ7LOZG^FP=&NO) zh$Og4plm$wW0!kb-$)jHwIa^%e3Aq_Is8Juh;FsWVR{uFon%aEhN?brFAry?vT7p| z7Z3u_Bh_6jk8(=@9kOx2Mlp|+6AkE`df7ik?YI_yllL8(Z4oZM2_HN>TE%fWtM`za z-VDNiWbzqEvw|`+qK>w2gh=F!YnXNQpEi!PeVyO40%m}#S1#0HlvA1jRw z(?5gIUXFLtLe`XdUCj032#k=xy^GK$UjK@rv-FBf7& zZxya8-w}laW_1>ibEDK?1=?*B+i)v%^5BQ}04)x-PLgKR^?j@4UEk?Oc!u+@2S>p= zcQCFd*A-{EnkYug{OUF?T+&x}!B&I%ok-C)7lYYl@)r1@#lTg0Bv)a4$d^EpP|u0X zpHQr_UP#mAXNp~EIXZ6wl~J}uME6}*aznqJ@;`cu&nfpxDycJISYPc#PVdq8hQ_>p7Z5n0v+Gm3xf64LS`uWH8^UV@`#1iK*>?f!of7Mx#!Z9?-wn6mk;20$eyu^H zoBkkg(`NZmm`;tYBMn7_na>!ahEw1VxL?>{6CYd<`mLe!8=x|xjYd(j|0=GTLydu^ zi+`N*;yZm5k|~PuMbuj~lD$=D6&~+17xH1DyrUss)dcJOd01;XdsKb~o(d>YCyMW= z^ZiB=GUFG4GCvw7%)7FeCz5$lv0#_~>FehfrpQ(-=$Cy&6ZA4#zfAH2=^a> zRw9YAqLtWIiCZT{2Lpf6eBuo!kE{|#*Bdv@2L+|F9XPcj8D)}Ih!Zz}erfLlO1KD+ zHc}*du!&a@302M12}Y2Dn-5%4iwfLQrgwm#kc+97swK0%n)NNNeWTPu1L@+q@5co( z!&6DUK97AcAE|yfC?zjou_u-Y9TfUawCojmr|bl7_^qpZ0%oFB_rMi?K<}888D! zAhMJkp+thy1DR=-7aYn>^P$P*up%jSn$FFu?Mp zFEzK5X?pX(ZIsLnv6QcPag32^n4U>0B$#6y}Ne zsG`_MIR=m}Gwk*XVjP!a_~4DELy1eW41vT8HaCZnOF@crZQGG)1%U0Z=cx>c*6?lV z$^41l(gpd=gML*z7OmvvI7%IzD+%tXOW4!SJDpk;iPPvviqY{u#MW`_$C0Q>$WF{g z7nX>Xiq0q<7qAyl-PRY4DQkR2$*hjBbXf#y(6-fo3xnwP$G3o^)HC@fiUIud>1~9k;kmgBuv*^Z$Wg3Ojykb> zrMzWoiXzd0!z}O+R#=j|FuKd=bN%{R{lRBn68F6Ax=J%1N!=$i2&XPv%~cX=dlaYm zMV>NbfeD8@?~JbCslSN+tk=47V5;7HnJq7KlUk*d1__5t=$N~#DQ-79SA9PzY(N=& z+{<(8*+fq;&lF3-^j2A*EP+Xb_JB9#Jucl=w8dVq<Z6y&R`D? zL*`npml>wq6a1damF~Tuxh;pi7BnrxcF^3xD%_Gd;V3-m5wo~^<|25jC$w$-oez~; z`pAsfyf<6^1Ixb`X5CmT?0Tzs%}tcl0+ew@olBR91%EJOAs#v2C0Uqv4sQaA_D+oeq8!a>)sOU~Cof>; zIt(|TE1KLPf;M~i8Cs=#{*J>{*=4qTyTY{Eag|wND%RSwWc(mepNty6DZyE)1+7Ft zTQCZ7WBgaIJeisjZO5NioNiZj0NdL4zi8tHPm2Z5JqwyWI1>J%sZiaHe~hZtkbM^RpKh*yE581l z_oSQ{PAKffKS=Bj+|4O9f`0rHI3lKhG?hbl_!6~Cxo+s`V@I!4*vF+7CNpx$cCqsy z!cRI&x+cG8b3lt|@+dA`k>rHLqF1lv%f>C=9NTf%WcMaq zpsWWYR5W%dwNH`&5!NF>ymTRORJ3GY%zx}7|7+C#iFQ#VMh#~RoQj9i|UdUv8#U`;ky2jW9R0z z*6**IOq7D2JXV>ToFRwBK9?mVjhi(Nr-R-BXB?eoww>i}6(l)23T8a844XlUNu68S z7|kUMeG~nzs|J;V*;A#iuIH4T4&2tvKQ7v`{fdxCRg_%EFPT21JF2THZfHe0?M}HG zGfkKJYQ+C9+Mlw3#K1{rwN^nlcv$A@r$i!>Bj{|`+6%4mFn4c z?A6a@SCI;^(ajGB%d*=hQ$37dq%&aqd%tS#AY`*pmt8w!G6|%iB71(Jk4{q;l*W?PNwvvQ@-y?twz& zXmaCg;2+?9yenG^b&IZ0XI9Nl1G!i${qnMpL;le7k-njB+I^N_v77F$3Z+Xh zGLWroE`qwx%(B0j!{ZpK=a|ld9o5PUuNuFq5A)D^KEY$x|FnU`tM_g>xj?y~!+%#5 z)G@Ht=m1ul7Jyjs3Zyi*=_zdsPTbJ0^@HHqYR$1y2sxsu*-yJgtwn!E?Q|$pYvwP+ z-$gW`Df2`3?u{rg)KT&iOXhfEU!lNhs>eLxh5p<824mcb|IGudfr6Hhq&&}@yB zo5j!DrJ$ge`rHnBx$pQEYy zP;{Fl9Aj!H&dw{T_LSDTb&Qo4TS6e|Er%&TCGe;U|AKpK3ATtc2HGOIQnU5{t5|RFbINv0r zW$zy^yTDoGsmhrU>E}Lovz6YUb~v0WQ?c`G_*?d;>2{(U2f=6j&oJ6L4^8K$ zqn>XED=iK+gL{>``&9QotVU*#)XOIucz0|5H0Ck>3889+h`@#=l|RnXKOv0HYFulk z$`qP86&iclAeC2D;ZIcXNl{^$`7fHuTEo7Hon)f*Cl~ju-Qo>%#;htKOul9uE51 z6D1kuIoUBftUNX2ZP><@K(W3O%B7JCU-yrYx;?OP&&E7M8?Wp9q$vmYIFz}izX9P66goa+* zZA)p(ia9zBl%0yNef0I`DYT7mTJ6N!3q17ZZ>e`n)E(r{NRc?#DQ;@HRg1 z48@z{SGIT zcv~noW|D6kEqqwI9bD*74rY*vPA%e*D}#d5JBdg2yiLzI6`zg+aowI6>~S%^YO=`8 zpJ{z66KP(34;t?yYA4yYB1!%;*Xrm#5&FfjAlhrPGjC3E>a6sO=HhiFTXg?eiFcg% zK>muovF&m2zP1$oO`kE5v2wm`Wwl6y)z~uhkw9PHz>Y}QP9f%xmItR0jV!3u7k3Km z;MYEWH9+G_ zQjo|+X&P-RnC9o#-W1bFJl|Q>H_kCcPMWMx(Rq94N6c)EpTfeao^2kbgpCYY@T7HP zf<^DR-8b~oU`&~PkbDn~(d36J2j(HkYUMwZh=zFl8^E;zFl|eQm~CvVQ&4~~`*@?A zWQAQj?`Sl%Q_RoKwIu4W$K`BWvaEP&rq4l1+<}*HP4^K=PE?H!;Zc`6U?=L79_u!z3B`@=B>GXa(VQbj8z=y!5q;O}op3eM_^7HM}M zRo7v#rIY$5O@B!YImlu;0i*}^!pvW(#P;Uz1j0Ad>S8?##;QGrvYUQ8gS|(TWk$)} z^Ie}QhTvyjQzt1>42f!?_=L5`;;QRAc34cyQfrlnB)h}WqX1T0$#FJC`;jspl05*R z`hDW2p-Iw+gAE-a9E3+ixF$R1yJpKyL+NBdS+1gU75}OE=6gNhK#yw7(bt*%R}h(1 z+9)>o`%OFry+XMe*6B|@X?lEzf>JC zY0mPsGRzQkp&H?+SCoPC>s1MUu ztr6EI4PZL>&IKfuP<8yCU1xUri>9RbC-Z0NAxRDApwUb=UjHE%L%hUCd%L$@PkX(5 z9dSzlqX#<8Y2?fF-3ARWe3;wpXSIvq?l1a}@lj+xc=gMA777XAXdp8h_*(bz7HXkv z{su(~eE)wT1OC&n20b?~771@a;ni)4C-(&47YluzQMY5=(e3)5>Gbd9KcMOM_lrN5 z3Y8063y!JOFvO4Hr%9(yxHtrjoSKo*fX5y{Y>D&V9;GhaS=3`}I&EZsA)|w# z?`(haq+-Z8%!NvA(WpKPBhd6)&`VT&CX(DY38-)`Vv@3^L5*T%>=0YQ7rOM3Jr?&k z1?U;*Nk$2)*kC%mg%+}eo>`@SH-d~iz~xqreRiuJgTm<`a}5MIz8j9Ee{TI*Ou|uP zs_bdViw!Ms`Kf!IZ&3kL)u$nEll)`9X^wZ;i(_=@&il_8pc2_bbzvKIBGx1nf^JVL z=52-VB6|x9CVi7Xyss!)rXlUZ(6tYnS~;ItZNQwDV>ifj+|&K|MwaXP^>Wr z^UlVCilw4<9ZP(xmb+H`<-^|aEmO%f#v_oUF zWzE6O#=i;KmT6O53q;?b{ay)BUgGIX$I;(8MmMO`HOa^ZGs$_D_NsyLo z|Lc#n!IZyDiBENkTQ)@AZqGeCOqiCwI_hZ8GxiihUe1G7EeSt3Fa1HHh~7-O?=N|3 ze86gd1XCsnDAm3snNZCZI9Q4StByB`sSq4XhyfL*3swkj&AwgJCLbEj%q*YMa+pkH zr#qM9)9iVPJ;rdGMdoNzL8?aF96$i(7$3Squ4EkxF*{()n_u% zztBzD$SRO$SC5_`NdxucwMh(g;N?PXLqRkf3#UM@g8ZnE+HFZk^vKQ&J zOMOf7S_=HFC;4OC_Ooh>*QL_48(c1nFH3w-CsHUgZL@Erh2k3Le|pyuB!<}+%NxIF zshyK-KD8gyq-#B~w4`Gy{H@)6IP7uwitk0!Lax?&dE z*xng>Y22REeRs~Q)ZcZJ=j{|%Q#!ZS*YZc~dG@_3rqxY7gr$+A*re=28M$Qtgbc|m*gTHGh$-2 z5r&nKOBYLG>Lx}^79#72mf60QS7|l^oAslD}J!-;&#q*=?U znL@n}Qcf0HHEA=~nu&kW^coxvdXs6v_tCThoc4cSSFDk z{S;cNtqrGQA5U#(G`k(W8Eh+xMTU&_Me|h^jg%jTXVXwZS1^{`_tm4S+HGhhDu5{K zaZhpE_1CsUTB^1|ZIh`FkvvO2j#UW|#L8NBDY8H1a0VkpNmk)+Up5z1spwBf7{AqF zZ!QQy!b|AR$AiSwO#9vaHZvQt#yLs{8g~42BxOETg%8K3Kh>T&c%(FQ5@6Oh#&RP)@1(q{3TUmkk3SLOA~%3fvUmx(VZ-+O#X6Us-|=Vl{Gv!MpF>YS%Owr}LV_f5hkR*V}MgH#Vj3#v%wrxTkR|A`u+@tn%) zr4X(!{+G&cw&{#NyojP~O0t4ick__S)O=Th3%eE7y<1OveMUmdbrUQN%@aUd!9maJ z=74Bx;j?bac_}UECSq1??x&jer?}&P(d=g@ntFaDb07T9nm+H38TFTUc6pFX0_VSx ze}{VV|F4#L@Kwttm~7`dqAg;A{Lc?(Z){`n8!M0c>4~||l8TiI?Khz$U!EYRagrb% zSZd0+G+ULWy=Z6G8EY*oJ(PtFX7Ox{=TY0%>Ahlq1`|b!WFI=Ix}>%gE0EhN-6Yr9Z2hV4hl8=j%A(#jaet4kP%AC|rGSk`8`1d7QL za~gX&lTeCWp37Wz_x2uBeVBRP9}5RJe8R?0%_>)W5{5Ao;<4KaXDLA?ZH4DAzHF&cNF8(pV57K8;B+ZyjTbca#6|`Tan}*G^elkP;#zy)7&{;Ab1LNoXIY z{yz7`JrHrYJWn_FNg_BsQhco;{M%`0h8HYpEj4QTO*onbcE~U9^0KDV(U8Up1IRj$R~+SjFI$^^tp&KjksZQc?CO1sT2nrbf^EnzK~wiCw6W=%dbr-=0sk z(w)5?pG$mK$**SEX7r;^=Kh5sy{k{jV9)2QFC-7UZ&mvoh|7hlXu}f|{sURif25pp zxM;PbN-t@dOFrXx6Q!x7^s8jBcZqgCU_A)iaQ=0DtsWiC6ZdioDMnit?%v4{)$T*H zsS6;c?Q%Q_ugOM8oEL&ISe!~#OBtB4#lKjl;`H~7OmsX{A@{Fx{v%(`c@kvGdO%Qh3e)g&3X zFBAN+DK0}NAL0fVGbb_V0{H69eyPW%mAAUAFz8(75qsGWhKl;JJ?ai_#LR3*O7Q!+ zz1j06UU*}<><|pFb;#=I3#X6pKUdx08>N#8&&jxAPVi&Mj8%VRV~{~3t3j_eC5vBEXI>s zBR~ytl7abVU^5u;J8>2m>&Ue!$K=yl{^istv~<^%KtNXMVv*nIlfR|k??c*K-f$wiklk~uXMf{ks%mYGOYvLOsnB9YOro#%^~U4Od;ZQqB60n{pBJfZeg;P z-=K>`+YJdQ)p9b{(yfJ=$F)wXB-?Aql+nsMBag#1Pmx<=PfQ=4HdgAj)Fke`Dt+7E z>Z;!xHuA+{;hS-rtTvO#(5Y}VDhyLnO*FczjouRX>f<{tEjnE9FzRd(I3;(Dox7w$DbO1GT?E`4i^uW%aa7(FS;^d6H1Nb&gkbac zdc!4l{6cB)Do1hhQMOlZX*fqdyq8@g2y46DtkwQ;n=VWZFB-aNBLDLI5L8sSian7) z)3>4@KG}6w%X<`l?1nF$Ho4EfkS{2Au$|i3w7+i8YjQso+8Ov-Bb%LakAAY{Q$_a} zny0w!tkCkCNYTdaSqH0Ae&o*N+~m2a>`8c&Dvrj>Z{Fg@rf2;xdL88im$Pj||Dt&o zhS0rI9&BWnN)|gmia3DYW~`n#eSJZ$%fA?ur2Y4^1r(jJX!R!iV?)0%8mQ zOf6eJ^nHkpgCLOxJV~*QXddNuQ(LJ5OZ~*Tw7LhHb=9`y&l%AT1`smjae~7M!P+wd zHGg*fpCZd;{n|WuUz-JBDs}DqT@g9TB#lI3t0cXae6DO=Jj_|T`L^I1Mg^8;|6$Dp zl-wuk%@Ub>5VfsiCLG0lk&!kd(3nyzQQCagGdtyUP|JZL-)L)%yA2BV?%mN=t(TGU zeww+OzeS_-ZJC+uC3|E#WkCu)>3SA0OypkaS##=`ZoRrLZc0kLsSWhte6u}0H>$M} zwpKB}>SaXHaI!R2C%hsx0WlGqwfv1wg|xMs5yr&4aZ`;ptUb-V+#Jr!V`WEaT#PKC zJoE6Yy038t;TUzvl<89?BBP`>mn&4>!A>8a3lT3;eE6oKV(Y9LH>;nkQ!l$%x(Gz& zUpmgJ!J{>%o%HIsPq3XizMN*qTFo2>_L{TB4F+n2UYBKdc}MEsd9b?`%L)HP9c>xp zcs@`~3o5{TO0)lD*kH;`wC)*XW7Y04DFLBOibGIFlbe$df^fA*Qns6sZp@rIQj&3Nkk|R|wH#>A?A= z%TlBQNR&&c7X)AWIvxB7Fe?b0D6H6b1J@UA>0s~V!!0#r9rNT*g?M)$eaIt% zCAPyP-}ypt)gRDtCq!kzE5bwmq44T_9JK4vfx1HYnQYomC*+BUG))k z`f&21i9Fj3-0kog+j-bFR>KAZPOy4V7yeaAy|Sm&T=rg9d@3zu4y3mkXk>?$*ikw% zKrYk(9{yVy&b-;*gs==Io=+ z_l@9j1dv@_q30wp3srDz>7GRiK~0yNhUb1qVQ^4}GsN4~SX+F{i4%AkTJ&6VhWweM z@5?fZdh^Z)@6qMguKn*=dtEOlZ#L_}J=-_n@37UA{C*fr4#mON`2N3WzWuWa_g_Y> zbs2^*LyNmfHl3)qrc>(AcP{+qm7hg>P9%i_IcvG>l-=CGIO+0qJo>4Mr&M|ZS!NiE zyojbT{F|;(D$H<2LLu`9LoJ1Zgi*#U##*&Bim%sNKoswj)~-pNV3+_m3uUN&jAy6M zoeODbu5*%Z0IK`*tiqJ|p1m5<4cL4WJqPpUAG@feS+MJRI}+(KAh;#EqJ+oBiq;7w#R7S8+5@Lv$|Zb9 z>!$}~Wx@tcu_A`#Kj87PK4p$dQS=(ML}D0-(6C=__Ck6~$)^f_gM+wW!Q36pKsRlFzey_lk?u8}aTtY4^_q4&QDK1Y_Mf zK*vN*{uf72IEI`g^z%>wyTVX$on1geQ~z+OxEcvqeb2X1m&#qeVliI7lh18xTH$@< zi&0fq6*ha9vaCm(fTr@w*=eIXkX4h?nn8mtC`H1S^#i1Nn&SY7y zs~*(V`s7%bER8YoHm7%U(`P5>k4Gp*H1wu-H4Qm7sI6Y<{1?rmYFAIQ6zQl>{(Enj z#Djbi zinx`YD!T;-p1fEuXA8kTooJC|y%2dKh3y<8I7CS!v011Xpy4St@BjAtV8yh7uiU=` z_Zx>+_aoPI?ch4%@=e!*xH7?bfL%&bp?BgCvkl6mS59?P@cEq`S9YDA64k6NNk4Pp zkx-g=J{gq?bG+=71f7H1m4mg?0{^xft0+bVmsG;ymuXDtCO67Bt31~V_#o%sU%kAe)F_EJB zxJ9KI&+_9^v>HOwmG=fXelIh47=){u2S(ZNU2K(El~-Hh#C{2Zqyr-b3Lk0O*IW59;fs zw|u)2Q3LIxLS7m>A&IGL@KN>2_T^WxH_0=N&>T&BWUb#BtlC@LNazz~h3H2IJIt~E z`pyFLeTEvBkDp3L0s=+MmMtG>mpI#cwhC-TSLII$ov zwxB@SXmK&FM&u=+^si`7CJVDqol2^-6o4K($=iB$3)0pIrSwHw3Hcs0{QXs^!ygdf z)^7rWeP9dyi|IGQ2O6qcm?G!^rvPira^fsq>i|KO!4ILONX0_SFOEGeIA@2wn6lLC z_~4E&hLRHXlWtrahFtMqgUe4wGkw~19PB^^V(|i<{z=~(TlA-fm|6#1LXL}ks#Vcd z2xsj`!K0YtVioa}^UetJwroN;z3ij00WUJjq?skvVi4>>>EgKq6}v5Mk71NH=`+b_%vNAC7_>(kt)yLgQJt1{=<{arRJ~$ z2_436!dDyIWLc~ZcLV;#7~HiA72gH{-k4?0h)8a_Rz*1*6(aAHkscXZ=~7Cuj+n=d zDvFt_RC}^d3Fek|s7;k93fqyD$Mf-|=s}loG}N;J>l0*A1K3o0X)V{OND7rJKU! zVh;ot?zceNFg%}V%A#t^?nj)Io#I0BWe6CL5*$-nOo z!`o;4P*qhE)SIXL&*NVatIR0F2w{I=dl-jh6?yy7N9O$wx zuf`gpTklt>))Wzq1|R7Qj*!0g%G~(5V*J8&R*tTYujj3?MZ)*AbF`w9nCua@*VOjI zO~{kw%cgE>xR2~NS@r8NNozy4Q;W4*~0<9ymF z9_2ec4p;<6w56wNkMGfY!pgd?)y)BBF>aK|7WSJB;p!m)Tpa|p{XJH{ODl(~QYI?j z_@U?2G<)ek)pyxFB%z$=ZPPjkIycelYu59DnbW&viEOr{eFFz6a)G zosXz4&~LE*!qKL?W-4@*`P=BclERJIOxC_^^pPo;^~eviAJ-i7rr`Ba?q$E-FB-2s z!bL1__#|P?yQzSSym8HTd4fZpOc)Ndkq`wa1RB`Tvl-|A#23*+rS2PO-xchog}91W zpZ|je{h}&!X@f0}*S?*i&sOE5aH0Q(lWP8sGD>wK@n!@6du%82-hR&U%74dcD)qu#R+l~I z`3IV`^J;0}H!pd8I1>-RGSUzTwc?go^yG zw2kpc1kYjJWiq&L(&h)jfQ$Z=)E>z6Fej&^Z?TYg2g$9Gtf#I$&vfL9AprFqnDzXI z(5C+rX7DwY7T~Rll-TWI|0UwxU z+YyX{!t`##^-Tsdf(s@`Uc*`6$0}$5rENCxMr$}bGr8d>=+(D4Zy^L*Eg`8C_r6U{a@ZS+qH)5V&@jo>pu_=M6+Zs4tZ@y zr4KCh^M}a&!(WOFrM^w4F}boNUUOP>%XRe0{X;Z`n|+-am)j>mm2 zfz5sqQ#b?Dsk+0bY(n?~gplU+BOG+Y54{XsP9BuEgc~d@)d4j$ykW63-*4+>pc6g< zVdo4tV}X@ zZ#-N+TEVB5H<)iBW&VCzLF%&4bFE+VKt`mtn3H^^n?n0qXVksgaI*&YXAH`+p>~5& z>|d><&?l?rm*wVg$P6sRfQ{Ch!4x`?CL;7pS4dT#q4vRR4q--dlJ4vVfVCB;4bl6c z345FhhkRJmDT52GOAXbLVmH1Fc)!A>iGftyP?4)7MZ{%=?l*)mV3_OAFaGD($s!V1 zz_MP51mq^|B%i}vEX(a<-u%2VzfNycpT}%Qk;8aZBhxZ1r;`Wp*fw4Xbmq2+O!;o} z6Wv5XY`D1sZU99OfXN_GYM4PM%i!~4x8KDk6TzP3jLgFsW|4~@Zf67PQv`v&&^Tdx zOn7uWk>hiZ`V4wuN>2`&3bB>J?p_!)#?5BY@%}^JmG%-nRJZH0wL+Esk;m+7;dz7# zY=sIM;u1Y?$iD^<$PRgQf&P->Y(%Z`0xk;64+UGU<3fxjKV4auESkQ!15)}h&PkQO zlQ~=axJ0NDnO5iN*y8uGbcPFk@omuy#Ze)@YJu~{a4!B8xjH5DcqYpVKqGF zvxV!q_zNsc-bY~kGbMu_pK`tS0t32Bg)jAnqPgU^cq$-+%}If{ZS3>awwC)%G1W4c zJ?Sb89p|!%moOtApQoiJYs_jxH$F4XxVI4db2evR9O`2?lYQ_{I+Oy57=Dd9Du`E1vQld8kVR`eng<2pIP4n42uiY%e{O`;u8|o7FGfY5^ z+)eBiO_-+!M8#bzLU1R-O-GTg>_Mn$rn;v)@lNR@BlXqu!mS`n43n=JR?%F8F>X|d zmAORkl%}srQLydsnG8<_z>&r)$MoxP{DKrUK(fZs06ZBMFIde zUMd=xH^E zKH1KpIrm%#(wv;r+qS&muo(RLa+}v|Yo||qPJUxda!)zVzpjfBKQM-Tv|wlYpqS+> zC*R)FSZ2teso6FLIo@gG=h0F6v?8NRQQ!@vFmSlskf~jQL((}r!FTyU=aL-}sUY>h zBW9LxMMZH~0q*#^+)Ro|u}qF5^3h^>Lqk2lGS24}v81fj7~RRdp;!4zi-IZw&lsF- zLbNI{Tk|U(gBw+tqdu2NXQCFwSk2)E8lus0#HuCbRQfq02wc7l|2Lx%NdJk~OY5V@ z+QMmm^jF+vhC}()-n$x~9_&_M<{t}J)bFpLt+-awe)eAs^)Mw#cy^NWB7Y8DqR0L~ zOijsS5oTDevMW(cJaM`9`GF<4_;4!yq&Uc!f^I??3GBE^FatwJqF}?!iQ95`blwNU zu{SHm(UeHpCQQ#(7{Lbt6DH!~)kaLwylcYJd6Mp`;*wL$upZ*GZ=w-Va$s!>t5ov1 zsZBBwOa6)o;O)LeB!S>8PZAg6fJ7B|Iz~8wP3#^cN~MGGyGO=zLZN6PyGeD+FtD5B zBmh~sqw1~EIi&%pb%0e19ilf2^FnRyw?Rf0*mz<*M;$Hp-8U;oj)<5o5&FUdFdwWtZFz|kk8ozuq+G?+6g=7NdnEt>v;a-1y$@APfw;IbyCuNWO@<4#^+21=yn?a_ zr#Z_M^XD7-1?V{<5V3wJmCSe#Pn-c|u17oT#l}Km2Dd!T&d`3c=^xf@;7lupAE z^u!a@{rooSocU4>bDna?DRQ;sMn?=v01jya@D`p}{vc2(FFjMb`jv}0e+$Z=<`+v} zcNgN3giVpRYMH^;NDTztQ2Shbjk5wP(T1dXDXX%IgHs#Qu z-i>_Q~->baSBj76a_9PiV{0Uovz0-W16MmHJy@ z{UXs6b6fukQ&hP9xEX6{^fZ{-Emun0tZVpA2(7rpaljQ^LQQa^Da&D~IhQcRG@V+j zz|fCP0`RUS&VwA<8$_zW&1pTp;fo8qaLwKyqrC_WTW$V^H3Qr zMII}s7xx3aP2#>PCO5k7BFCIoE1t!Fd}|wNJM&hRWS?}7U}F~6c8<5O?T+#8rGzu= z{C$*9QmNdc%Os1qi&L-O^3H%WLx?S4ID5gbhn&E_r#*IbKxquEEpb=fORXh!)wItF zmoM}5fn~#zp#Ac_qbA)ukEZLYpuUDJwb@CuUtK>8hVYJ|CMCAyryd(7e(xuQ?_DT8t@(wK*3@ZZY9#;ePM#N0>Cp0Wh?sH-*Ed4)rrYEkbcMh+F30I~ww$ARR z?r9|vxk}Liz7^f}JC&hK*t|!7H%*=z=H+`5gz@@EXPAl+P&+(YK_yaAH~oUGq-hG4 zVHcX!M<)VD(G+h#Re<+-7C84C#bk2=nCm$uDG!Gjr<`!`68k*9Q*mgchj}~H(<@#_ z{#X;>t3WR>3?6^XhJDkyPRln_M{I4Qm}tRsDp|#zsI1-h}I zoKmq|X6=LptmZ=w$+J$ldWJvxokh`Vt72Key;=GPtJt&j6Px!lzjaSp`_4YmG-8ws zZeQSdkQdL^S1niG9B(&oNuX~SKfly;`&H1-vLsL`)$sC-!CJ?@FkpdPyZEfgQ?#!9 z@hJ0NF|ttD@yCadat`l9QQVKzbDv*O1??qSczC;QrnyWC{;(?_9$qoZ1()nMA2AD< z>rV1W5dMa+X_@v+oqFl`*q_rGI1Ouk)E_b*9m`qRtS^Jm$>J}NX|yY!_7w4Jo236~ zgcZQ(KEQzh)Z8$=7S%6KX*GA!p(DN1qRbr(Zr@pllm$fRL^qi=TzL@`2Md=V*7raAC41(?fUgxCd?YiYnksQf^Ky7p0aYv- zJb2W1Blp@KutmO@GGjO9m3f_upQ7j^k=kZzlAv5Z+%*UAz7g?wu^UXRZJHJKM*#Id z<=DW5@dAcrKem#hd87~4Z>V7?%X3x;=E>LGUF=@pV&^E-!9X!Y-oV;vnsn7T-a&Y% zfHfuq$i(Lahp%7%mh)pOlr%ohz)fu!US?2#HCP(I^H3RWQd;+ndc! zKNC|WtlYMapJlUd-5LcS+y+BloSQWYy>xcDm6zhFVq3``;u)eY0j1XyYA7E%auo1r z8z^RoG`GU1UeTG$LJN{($W;(nixKcO0)7_dAte)v>S#-oi-y1gN^M@z%?HBCA2N<* z4uQ!`FSpp-W$xg*FLA{qVzFWru^V4943u3nsJTP5n{0r6Ig6$Nm@1J*fxvqEidADJ z2yrR$t>!CpyWCA2DP!HY6?eNB#*BtM(2waXS8bGoos(3}Httd-8OaZsUP{E#-_U7rA>ACa6)oZ+jd zAp)}$-b2q@^EjaYU7y#I&&#NWTBYi@!B5bua?7)-DyZsod-zTjF||Qk3m*QVJ1@wO z17)A=bB~Xp6!Dd_9>yV~_K~i*ghPe=3I(S*n`s7hqkWmh(&8QOLTeOhm1Ylo{9>o9 z7FL%|C!T<(GFo^0orl6MQnA?i`?(34%K$m$BS(sC8`bs~(G(S36QlLHS2Nzf(o2Dp zy6g#X8ME5k_dUAm)~;c$iAIn%vvrDj{+cZxKyA)7uF7mQ-Pv5{GRH1;na7<=PonP)2mG&D|%73d(mz zbwGTD$*!=;x|d&H){F4qm9Q6MUI}Q|GwGL7Q7BZkZg5$4-@qMSKE^GGpwDLDb!D?J zW3j=Ix~dsH%u;^zj96?6#&PR;qc#14%Pb!XU63She7|NiROgiNZMHe~((BP~b@p0J zaF8UXkY5~vx<3QR{@8k+>lQ@{A?shq_pc@h*h$#g@x8AtF_`H1ebD^F`GwPjftZBz zrbZ*UzDO6!opfV3S<7`bS@D#Dk^Om8xiZk(0~!nUxe=Hwn*!D;3N@CkoW8C8(!i4UvZSwD=fY`^zL6T+Q-HKkbzgorrI@A_J&T4}Zm0@;T<|6WJ+AQ0<+F@1KrYczquHQtES;{Q&< zw(z8R`ZO=u=OX!1og-{y!7(i`g1{9^QZzx-pl(040YR`>a*pyC9bz=yrity54)uQ; zjh%tdTNsk)t|)&EAX}QAq>d`yg_5S9&?#+EBF8y1D_guu^IX$a`&+hYiH;+hH-_cW zRjK-PNDVgC4#jwk_~bmF#-Q7zDuqUe*kDl7XE3o2$#o*6Sg+v5Xlg_%yUs@5lE z0ejAd6P5=rWRK{rY4bbxIN16LB!V+ae2 zy@tFd3m&^wxBCy0;gRn6Tp1vnB32KqC#C)Qear+}6E1>W6{uJ5{gbzp6U{i_s_MNC zkP5;q2=hs!eO=ZE7wQt^;5gpWSKvtRt^C&DA}Kr){=F zjJS}Cd$2fmFCW?20~!vXG!%GWygaWq!04h!O^J)6r(6_&sx?1ega(9RLKCk$> zpKyP;=CAE3>!wtq^kr~`GsFp#`-SE<>$?@SX3-S&%B@OXyxV6bye(HK`dyW4jW{Aw zp=Bcvkl0_Qhpc8RGG*Sj=6$niUV*uzRgi7;A01Mdev6!ND+=m|iQ!+UX`WHLPv;K@ zLNvN2NSl4_-}oVZ4rn@vt*dE#$CCQXV~Ah!n2StJKlaYsNXZK@SV*0*TILm-tDGY& ziD3t`{aQgiy+WOOOS*0fcL#o#>E)QFeaR?~UV$kEtFDcQev?uGnl5GZGNz;lP2TkV z`o=pf==jj{AG?nQWC03gOm6)%Q*8ZYe$rGJdGXLRNlWnj{sSf;x0;qLYF zZLI)tEmeN+5o@iLlYhif1DHqLtPozDOw*HiROb}H6-{m*6b zisvEC#g-=hTyL#GpM<|xv(7Kyr?8~xho9YRi&3Ho$8uF}W(?N0YbQb&)NFc%r(R`kbsd7itt>DC7>$xf0?T$qh zp??IUHiGYm{Zn=RKkkw7=`syZWL3uAO{*XaqpTAd{5Y|OIdqD=9?FqNFeQhzn%;RC zS2RdoJM+1?CVg{ciR@9#WThIyvFFzMj>Sec7&DE6@2-HO4mjy%)e%tvpQst&2_7ha zKXf3V0&sS|LFW&g%pX7c1((?U#&is>W;;j8c zh?)AVv17xmi0K2HIcEdp;e*orv$Q#AY*<9oNz%`01w%t_MWl{1p+44x=_l5DnmWAB zyg-CR}Y}QKF%I)SZFIGH_1`uiJTLW z0R*9M5{%t3N>F_n1jt0dRC9i6dd6Uc12yX&;T(3s1L<*iS1vsK@6aUgAG7v9?t7^| zC3$#}0j(A@M_(kPZ&{6#`);x3ml6lSXz2#D*3|#gkOzGasbZFavAVH#s5^-y>_w*A@gnJG8Dr`duX9uBM)c$JWiz3I7`z>`)#CH0$LZpz zFT@Pu{2!7Yjz;3n3K`?o^fWZV!leWF{=j%}W2OIEw`haGmbOp)FOWM&8famg^Vm4v z(@R$gm%aH&W5C5~dv)7|w{)=p+%Er0dK~`XvHuwQ|LoW^`09Q(2bMj1*ZZ#4*!kw; zEKH!7c8Sm{y+<+>&MYPl{mZXL;I^m{z-MEx+2D3m|K`~tqg)sKbvy`BribaTnpQ|* ziRcisZ?(^(vCvoo_cBYq`3E~&nuq}B9=}VI{`?y&;-7sMmK7D#_-mmSAOff{v#!BH zuf0p(RkG4fjGyMk9dnJ{mkPgiI&ypW5_OYfUWx2WK71P5;?-+jpCi*2gJvIxaPDJt z7vFO8(DC+((y5=1X((tZ-Ae|PKtuU^k7$>&ckhS&LqqfsF%e9k)o=)=*0*((Go$^+jY^e`x=3t@V%Ej&h;{ z+$7Y1u-Gruc4Aus2Cr>~J6<54O7sKtt}2o+CKaR?uzIBRsHJ;U5kU{tspY4wX2A4^ zVT2Mj?u8;9xMVl$|8f!k7poUS1w2%tSH38wTTcPNAu4$XCwuy}r)J=5`-#5=Z2?^A z0sN*h)o+QAkLHcJ1?cRD!3=}35`J$`&4Ki!%hrMOG{dc>Jt=kYL!$8Z2U6s7NV`-Q zIdf*D&Z>yzZFbxPUS~93@`bCIZ2Js83>sA9?(8|Y@l@+BX@ncB-0|ov6PTL((u#Td($Dot}c$u!VshT z!}l?nW6IvJ`+FJFHUIgASHe-+`}Hq zFv_E&)MH-C0xmm6r!Qs^-^v7rOhcJy4ii~pxiflAfg}2y-=zm6Py;_kawd+hieRfTy}9_fFR@Rqup6^bmYg2?vJ;8TV;^wv#9Y2p zpxTV0qINp0$cXzbj)U;1TP_xiVU!kwdu4E8{A4x8jJL`XGb>l3)3LoJp0gJF%37(F z8OC@j{oPU{D&NldhWJoq52E)xmW7ox>jGilS!YhIDTz#$Vy*pEqG=|pP_3$@72wU~ zOL4yf79JJR0py+Rc1- zz;E^(!UhZ?l)uQMD9C7yjql6XlT_`@x#rv%5(5BEN&`KR?v{@ypcZd$u?W8&&&-^t z8<)BsGK4+G_`nmcZ1oMCevSxps<@QWVCx$Qs`eZnH%|D|G!XI!xxIt(6HC}>&|DhxdTtQzc*iEg1Jw@}1 z9j;FEd&>PjUGe|>tKfMp-={lU5tan3L5%t3ou{@q9^xE9t$<*(-XvrjS`oXJMx z7OM$)dmA-||BM8*e6P>-!E-h#odsKQ(ErVe0gQ}b>Bb4c1wn#6**Qb;*9!YlZ^``% zQrIPuG=D;LT8n%5EZ|qA79r1xQJtKjz{Fu%_`!LMb({nT`A(=Br9DhAbd)})O;;Ud zT`!EXl)13jI!Rf^;h$lh|MYW-&%~Qk>Q|v|`Nony@L)F3Ix-9 z(MFq=8R!6j)=xMOa;S}W;t5CnJf`DjU?_|vO=)Rfk-GDNn-zHzoYp*{%4DMmSC#52 zx4qlagE7L4R|3UBydCiwyuK$E^CO;fJe%qrQIetC(*vmqL($;uP&=`2qRnq?nZ6s+ zHkNOdLDWf9>}^svnf4M=W)|mrV=!s4=B4 z84u~QsE4jRJ|;JNFE@`_l@4f)T>G;*QU@Q=yK>2vpLE8n9HzTb1(erSYZ>J77fx#=+jz35*ftq{dMGSoF8QTPCSJLr zb!qBJ?`uXm>wT?a+*aec4D+6>p2+g4sqjx+RAnr}<41jBrWtJE5R4weX!aDf2kz3p zr^74=T`h$fH)doQZTgPo9huiA+UIQ9xvWX2yAQ^K1|ufhKK5DQ5-=PtdfJIUT?-t3 zy{$Hfai;^*OB5L@fMZ`{gdOJ56j8X3Kzb9{cXhyhar1n^=!%wtTzbnKzWB-J`v~rT zNN_|&ZIi$XWSp9%h#D{BaHx;;yJ(l+5Ioa|;pJOG_`krp&t98PV#lE~m%dFW2QCL6 zX0*$r)Ls(?_&Mebb4*+%tVB#$w&?J zIA#oJ!%y<6c}=|_arjt-TD{HdZl!j07^q{hpz8X*CM?nu@v3yG#>mbyz?U`{l`dtE z37412WJ#GkHT^x{p7M0RFCe{x9*&KjNGJpzi=HX43L2 zPJ1BAK>98yroHiw%gKqE6G# zUZ=;ayLX9QVPO#lp@Xeyw(}{ZJ4WH}RVv&oZ3ak3OqJ9?%2^k}Wu|BdeLr3f#RtBk z!-_!!E z>OY!HOh27``xPGgJDNmbKLy#`$5psGMK=x5q6GOpvZwgw;^Z3(+;}$I#3X2*O__!r zoVcYnoOP!!ke?tgW9wlLRwPZwuu^&bFevqCLfw%5@o)&Ot@H>L%X!~^8>pWgqk4r| z8u<0JP+~+%ZHm_*D1D4&r~gQ^<oi(%&Lyrsq>ORMGc2yasP| zA>>PdJI2gLa0+?c8gMlZe~*`(X?Bs+Lfy2AL-*bN-JJ43@pw$iC%Xg~!kOZrb zxk3h>bdBWLF+X#PE5O>d!FcS$76E+m)t}$<4Jb8Mv5FrIuKL!oS=3M-GhFOf5YzOGDH2%m!W1a;*6bqlsdY;fC-?<%`;4N{~S5CbR&vUjvamKZr zRrN#~;hNhE#syFxUYzzSG)Wa}K8FIJ?7u>WoQrY(r%or`*yX}wWSiylFAnr@iz`Cz zSF9Sb@x%g3Qa?zH`cxF2RTa6uFcu6c6yt@#e-+IA?yF8OfWaR_s5PINuoxKJb& z5r}77LSiGO^#Z=D&nS>z=I$nncLMWpKB{xmfimI&WF1FBBW(0RVPv*hToW zM>a(4kCXXMLAs=qx7%!LIKMzFD&De{_gb86$3bXpJV_+Ir% zL9N#n$G~M9QO4y;J6But`7I^2qxje5*{sUw0%TfClWLA?3wbla;}uke*}<;NK`oGz z(qQs4#Jgqzg#m z(4*rXLs*YNqD0X;zF`iI_jpQ1*Qdw{fw3=d(~K^u-Fw7>OJ z><^s)RMHWEKnt2nPI%}`N2TW%v}+~@SHjCiL}B6Rqz7iJxg+=XS)d-@`<_~04{241 z)>^Td!3VR;gE^NhY~&F^aiW-;3d0C5Bn$?afj(l`4H#+hU8UYnZSB36aWnK;6@;QW zI0_nwPu?BYLT^=~UL`u7s>?1<*N{5G9XY4%{|A=!e_K&>s!de1XG58&oNQTOvg~rg z{P;%k2L_uSS^7-tdr1a>Pv6nW%8^xgS`nUJ!DM{%@{tczv71ov2ET}ZxGv$7R2X(b z*PdG}hzXtH46WMkv+&YpSp^_y3S#@D2F@mh6XkZ+Z^My1zaSiQz4_nm7nq2qKdv-N zeK&^VUoJyQ^9Jc($%q-h|H9;nvf0$tj_?F#cEWNrS;yAjL&<%JD~}L_aA)IQ>P2aP zVEUX%aX2rbu=FcO@?>?+FO6{doqojlDd0ZWsXDB|t9dlg8=$_VlLr&}&;x19L)#h< zIneGE%n8ITY3BG1fjAngtg>K<^JMsA-5$(=D|@BeSzzS|V+}<9P)L2XH?8qDV?W*` zCYgkNCR^N4upcqiMmq=b&4_utZoL@+dv39p!PLQoa9n4{N`D&(r!Vr=fvNj~lb`C4 z$x^r|y-oaV0XyqK+oHhw^hN+0namBG@WAQy`rVg!DF$RSvsJQ|rO|qJp}H!m6rHAV zpCxzdHKR*(*CF7H21kCnB!VU2-XtUvl-*9;-$LU1Tx+Le0>pNVfAg&Nx zshUe58>dfYlr#L@nYyoQE)x_Kt8$?pcP!`N(d+EA2KT@o{qFYRzP~%O)61|HYRz2j z*?2wu;a#+YW`6ej-h#{{tsFvK&2TMrqdoNd*~$iAY6EN@uVu`>rl}L+#PW0MF$!u9 zHsu#lK)hC7#yKX~a$}}#C&5h8Zr?VW>(z?rwVYCwpvE$a2Q=5saqd}w!=~%*DZw1W zF7o}HSoy7sGZE}M+r9Q1c32R>p_sn>tKb;BUFh=B$#(3Ej1=5uoWa$0wigAq2`7xC zqbH@e$#XADoP~TZnNn*B;ygN)7iF%ls&Y1WMO-HhSwLZjuoV@HgWVYeQ2eIzfobQ| zrfdu%+dN?^}kk3Sc`PsnHY9zIqSZm|D0t7xX8_T~1J0^}SvM z>GwW>x-c5lCvqw9@gi7CW$Zg#85+-nw=-fVCd(>6$z6QwkX`#mX+-)M9Vz|O^0eHg zrnMRvvC=Vlw`5!g{+j%swqyHO;sfiEGTiM_(*oMzi7KMWk8i9nh7M69J{==1@0S(< z*cydhm7u1dA&gP~tOLcd3-|Af+)wu7N;t-Vg{;_FdrfJwHM(}TDW^pi7m8342^A&K zi}h0cm_7O{$izGyV*LvDV6=SurAoz=5$bt)|4>-|sfz?}Avi@4n0h?`zL<#S-dDW6 z-;*Yr{;vYeXZ874Zll5IBz37cd*YvEe&eGuum?W3scY=;rRh7C zs6gXl(V4>>2;gZ5ee9pUG*79@bmd>87Z+pSlWa4@&dn)>Xh;x_n~Hr2VpT+>(_RE0 zeid~Vj#%nEArtdMTszX1OnXtJ{X&1e;niT-POOvJK6&tNVoMMQr}Q>}GU7-VSMPDl zn5}7%St9Mz6^^E%p!SL1JSD#dDK%d_#5Yi4c+MSP!yI=`cRErR+@J?;&a5lkIc4fG9dzq?oq_z;`k$gA)HZ$?9ecbd-C8VQKDmo{mUaY-*crO2B% zrETip#Do`7&u>AvlIh%bQ_v>Z25tmGb`klpNA<8rPbse_zA7U0rgO zrWwKPrrhw2v}Xh<<*!-Cixk#NYDaJh%+=}KQo0dS#F*p<+ZDiIt$XS**>or+#fShj zrg99kTXDJjS&3zd-L+}Fc~QhYKtl+iQiS+uXV!~9s539_h-d;1WiR_<-7&~K;;x40 zAWmN~#>G=%z>8O9*4sn4n)#&F94!rrle?w?3e-ZThDgi?tx`0lvR)0vuR&1ATD*F?Rt@TtyKN?^`QI zkkZs5VmUG-SMPG{=uvdr=WbPa*5e>haDt6y*j)0N*ZA5(ZtGU5s#LeWX^Oew>t~}b zQS|Tko*b#!(oNc-xTmOrMRnbafXK@4MvA1i%OYgI8{d{e&{9Sjv>M{|*FrRDL9wT(EpM+7QBA;d6&-6-qY!+b;B6KHA#rbo}Yc0ReO}% z8*RqvF-S7FEOG^q)iBHQSgdN}H6>B=IMO5Co~~0!%5>$TD?+x&&~Sq;V#>gg#y@tE z{*>(hyM^HY@iip56HQ88l2f}JR=UJ0Vfs>oj`E`%7&|BwhntIm5;pTScDX(||;sJKmmrlwHp=!JbiWT#*u! zzIMhjwV4wuK5M=OFuFYou%cvdmvq!5O^bC^RyS)#WnX|p8l1q3^O_o$)mj!FJyCx| z^M5=>hB!KXNI!`aUNA0%I3B(~&Hu2E?B1^0$jD;x(BUS5U+p7}a8<8!wtW{E;TJ`4iWV-qg+#f!)5V(6-o072oDCN|zbA+53pM=(i=s+WuNSGnM_?eyFyE#H`i6-#uBte{Ud~=2J za+M8~`6YR;C@VEAH;6mk6sn@}#>bm>-&pEu=o z$`m4~AICFOJI~*|XN(7ZJ#}Vs&Q!H>q*nY5A=S9=-4uYCG~asN8+*_Jh3Nq&SKa86HL- zr^K8Me>W#`K4~$+@B%(k+BNW}C24j#%7sfq5o@xpLRVLd6M`qQq-Ll8j}Bw-$@Ao= z3$?9AfC=evB7zk8oL0WaxOFj-jsFlnqHtLt(Y#(F4-GG+2axM}#DiQ6fiK(REW9fgM_A25MM`hqLAC2ysXc20~^Qm#(UD}G1c8Bs3wMoBbrzUt!iF?ANp zj=_@_VJpk3sY0K~Df(hVZ+mO8+U(H&Wkk}1`m=evoD6r+A(2y}?bl(!p4093Vf-Jv zy;{XoUw)yWtbBBc1=5jo-iJyh`rjH*)Pskkrf*|Ax#iulYHUqa zk5@M3XpU?nZERNWbJ!q#e-Hi*0q3f);`R+)dTM5`bcV!q?o+$ZSK4^rFYf1gTsxrn z&L$>mgRuZ*jc0{p-WiCgsOYqYel|@qJpsL$D*SHk$c6>RN+wLSzLnd+g40+5Yd8%_ zF8jSo86@cnoJNK$J{8UHP`~>j{>aP?Twzu`GA!s+?T(9-bYkwj{?cS1SEVLr+iU>u z$NRQtWaU;s#b?*wD3(U{l;4uYOL8` zv+24c15UR_L0}CG2gY?DR%VN!ftHX`h3=e0a24AnKU~qLG#~Y7H-WAn*V1MojUpqj zUQ$M$7%9!=cO6zHY6`y^_o=GrxF~QK=dVJuvDMPFfQ=lUvo_h2qU=dt<*yI%`TL0p zrP=MID9ZNDT|%sDho%o7v;$}FDUUN&%eI5wm3%c#85%qdtS7QmxXsjkyK$o9NWn75 zo(H%1TX!zS^Yb0046Fwd!k4dzu|F?J(lgZ3i3KD-@ z!E{~`q@0iY@z=^QuKD<1NpQ%n($EDFt%g#BmY0*5~EsiK6?&(WY(q5b8U4@H;SW3JoyG|3|0e-?Cji{h-OD|TF3N$}0)mcJr;_Er~* zsFvo{a){69{0LLZ;9$|Q8!vpEsv@vdFMhb5VMOXEejqadC>iCK?cL;=cwzw+#bs}MGO+7q|?{| z;L@Y1oGWz3&4fsXowE6RfA2OxsN_a}i-yBJgCu_^2{s%CA|3sYs@BBYRa=f{* zVgO=wSbMC70^H|M#2|Mv6OHGG2D{^@OwQVl(yiS^FKxvbPbHBG_r6|BkeXGO3u(ZI zEJOXG+L|f*_Wcc&Ug@jN=AJAC54e>6a@?ulL@90{E}CPWA{AUp(1SsU>GRlZag7zT z03t=P9YeK@?qby(<2k(l_(D{mXLbjL>qW8M8>8(#wEvI0_l#;PYS%>*0t65#8k7<` z0-;F>m`DpCE%eY75UD~?kFr7(+TI1+*&y@s=C;F4^c~X^JHVQ581{STb^$}Q0x;U zkGW?sD;)nbC$dpfE3LL z^yl7vO6sB$0MXTSYi2)H&TSjr<~k|gUwZXAR@=FS&A*uknMHhQ9-shW5bvVl0P29( zKk3!mefo6cW&35^YgDO&*#V^SP8s? zbp!V=etlP2!!@@vaOt!f-@vLoXw;c6ul3~}^_fSougK~$<53^`R>rreIj782&Cm(^ zvKZ>pD#5|xk!yX&b+IJP_|ONr+V1;yKjV@evM;>Iv&x+1i<=goN1IunGQGi?D)Cd0 zf5X&CwjOS!bzqa0cP;%KZZMkPwfU!pt6ZhFpF&0S@&ASkFr$!X4M#yOAQY<$tw@~W5`GAx$FxU~1t8k%bV{@wWg z)7Fp7husA_|If8b`OH5Zyaej2Kt~S(Y$!F+A4hkN@~`cd-)&N>q``CUKZz99O*5VEzqLMmpWUGjce$&q zEctaj8=HvW-Xnao%c|_YO?h{#L*F^c`r*-f?Eb<2UjU4)Kuqn&1g_>{r%GP^fZ%Ky zw%hZI;t;rmIs4L#Cq~e#S z^A^H9iPX5{;0eiv@HE$8u|+yH%csRRXD08e1i#qk?eelq$MU3Q$a`*i9$*jo8HXS< z_KcKfxvenf(>4_CZvHb(Z5b>Px`pDe@4R582pzF9YbV`+!E^fl;{x#C@jY<7=5CLG z3(Y>4Vp)8V-(C?)7*n-EKNXIUX?oPsqjTt(uEE8w8j#gwZLm_jqACps!U~{EV%m-a zdU5Jb=BmsVuh+Ca0wZp-X9_*cIQM$h(kH9*+;uSYd*_vl^I0Ers?SXAd|e!k85#LH z5ll@q5D5*sX8v^py|}?ugS00bluE$XHdfqCfkF3XPITipe(Th`UUS~LHS6Z|6)3*B zS9fCPn%Ps6huU|9V%wl`kDp(w9+0q2 zV>*d?hkl4J*Tf=cXT54TWOiwug;~?){7XL3_!Pm~;hO$jkrbL*Y1(#8s+o${2dB-N z-=(kAFXgFf^iIdX?A_epE z$k}#Vu!eDQ*8EDCOJ>+5T}3bkdwEP&lz&tbjS4?LO;sK->W^2M$l z<1zLBbeYr1KEpI%3w{qc_93c8RT?7*`xZ&Kf(l@|LW)|_-ULDX>jc9DUqu`iDqlqz_VqR zJMAmYZE+XX^SEaGCMZ=dcSmlTD}W*e<+s;}Dkn>fBDgy1V&ex-JC z{%Ut}E4zNjUa)U!Of|IHP@AiZrK!=;v)rXhP|`q(Bs|$kt&&P!&5$}PTWe8KgW>qS z7v8OH3ZV0~hG&OVscpwQTzUC5h|NMbaq9P@`AO(Wdy^fBi+1=*Q8vwf^`fO6H2qcC z4VRv~%I^qA3Ug^qo+cK2;8nXRW5GIhsI{<&BOtKb``=-bK`7y=D@;KxVLp5CyG=FfV) z!4fi{mwequ<)K)}$_cv!D0D(Iwsm$n>%ymF&=T6Eydb+L>>TZ&%9@{>-IIMo1j?5f zdW+J~OP^8nYE?=6yf}5sOMkxSWyLJ z3hV;NDkj|wkeA67A?3R2KuL8R0t(O3tS)xjOMgOpnG{($dZy2!&;pr-Y|fY}=t1GK zIK|N*x`FFV}f_=!WO;p+nY}XYTD8PqUm8V@hn^JW~f& z^ZkxTzkKqgt%u^9pY(NnZ6IB(Ykp%;@#(j~V!=BWLCY2_llt!zx#)stLt;WCE_wqU z!I&7{!Peti`!X-`<4Yf^nY5J&_UaE-`NJ=MzjckRD?5y5I7R+qtPOk1(1}SlqGI#J zvUgIqK*fUgYwfq=@1Q?J66g(bjn}p~`!;$n@HBTUq`L92&l(c#3F24KS<8A}JqR6q z{Ua>><2@!JyDuSUkzg+X{H4uf;6aTIFzh`3>nQ=D<8=LlAWgxbGc1_#W`!Q7Z1~AF zE1XCyS-eyv60TI5dzZHoPGa@6a|W_C^~5Lqz}Q@Myw&rxCz`mB(U)UiD8wUVb~xe> zsNR47h{7GSQd!x{u$P8hC+>2R?uP`)OrPZI9--Hu{U*8N>-Po&j(KtJaC}i$G*nEM z0~PJ{JX$uiD)j9|4k)D9zX;(PCe6Z-LiS8$4sU&bqJfjRi_&SyOEiqNncl;)W`di_ zC6g^Xn>?Y6UL@%5S@0ctlwvXZUozf*Mi!4CjT>$PFNb6etW6DbB^Qu4sTYu2CL zU*0#|l5vJth3g7O_AA@mZg_dAjGF_w?sTi2=UNxMM7Td-+Ls%0P&b19o?QGlwWZ~pk_;gG__i0qQ)7mnZEhhpJ@3T-Lm=?K2-tD;iSvhVB zj`4#>P!bF}s%)-1$Tgx<@BrWnnH?J+;v{{!D}h3QiZwcYJ^lst1zBsH4uYo(2daFi zF1bAl@tvxZb(WU2j;a+}$yy#;(tb8x-S_J=Gib(FA-A{NDo^^1Ji5FZ{^im$^J88X{g3y2>ff}MjMsj4?>*CH3YwUDQnz$8DOV2`Pjkz|Njial z7(AR0L8*Ogv$o&)eNBCH$@0hbSh95XpXq1^d-r=~J|@})VP7BClY>{e%nS}pU)L() zm-}CoN`5LjIWav|sY5Qt9`majP$fC@of>kvpN32`v6A4UV-}L)VGuY|b^Xuq{Qo5r z|5rVimp!knB8kL0{E&QHJ;&Buz0sPWJ!Xu6AB_S-^@h0DoV1z;s z)>F^~%)3+}www~07E)F1aCB~S&RehH9TM89TIlf8^{GRs{krQ^@WYhcC8c+PJxt!& zx|58bhljOkMXeRv#N-bzOP<@9xcEbROWy(OD;mZHCx1tQzSot!**+?J6LIgeiic#> z0B%SBfuGdT*IiB8yzBf8?--G#D#NyzOjEluZ~Nz$P3J!qPZ8Rxe7tnei8c1S$R3?H z5&SGBvpPAxpBgUyCcmjJxW!AaF?;@7R<}BGi5*0k+aJ&?i+*~;lv!fizxU^gI;7Yk z)U$zIytL-mFUPx%B+Y@wsh^pd!(xlykhRM0VdulsgW3a8`*hr&kvL1H&x_EQ=Zj&g zP?&7Y)lDjQ=B|YJU-%V1u zu9z#3=j%@c>9S)}%PV`a2I8MYvF;j@e(qf*D;h>G!=U`@=(m>X{0}$yACH&$=MY`N zjcAmd^y>hkJCB<6yE_`{il$^!^vjg%=q2(sWSNq>-rkjT+4t-l^n{0FZSCH70~J>I z?@hj~K)3gTlkmAR%pQ2840wQ94X~$Y|B(%!7kwU0ZS^j?*`}?^Qup`tfW^8~ zG3JDGd<~ehY;GRo3TRhROoeYsU)PKFTf`FOsklqPm316^GBplrL(IpzUJ0FgDefh?%rQaWQ+_bHg{EM-)`I#Qyf)#dG6tZCXT1_(>G5LLN?q%F;SXFGtbZQ zxL*>imn{}TzSupbpjzG%og4qN)yLi^zN%*}02SU8!S%xs#LFpcD`~k#OY(aH8H>fYx z*t)kwW?z}9VyB-))@FF0)%qO$G>?tEe@5=s_0KB0j8+u?`aZ|U_ErG11$4H-UjV1< z!!tK(R7#5Ccx)hD7NMtYzU}4WPlzqDb%$jC5$L;OsiMQg<4<)=z~vv(87_xo+cWDP z-G3e_&4Jbsb|j8Hk<*FJ0Q#@x88CSlLq zSMJp{FKd$o!tV@Q!G~E+TG7gAAW%H)ZY~kqaQq(;`M*8D{&(OXk28h+Zx#5Gn+IXxCd0?GGg5_*Ib$}ZK6C>l)bC-JC zCHCwY4hWqCTCWWEj^(lOL2ux(+h45t+jwc0QnhY-{RKoinFXhSb2s&CU9ktQ^F6a* zur9OfL@D)hlvV`)s$}nAz^mL)kJfC_Y5iP>`jP-u-GK{PkivLf8yCPtYYR5?<8Wpx z^4%j{jRd>tPsRFf4KGb}1eA$-jn?LIk!!rO?g1~Kp5!*wS1%G)0f*H1SBFGY4*|F= zIiq-xi)o+7oA2Zg?1UT?B;hwhV~uIoSiW)%V3k{jZVTcUzTS8}QfCxgfA>$r{j)N>QXS5F?;0|qwJKid~a7X?)cK~`lW5& z65>k1u@8EUw^WoJK+`l^X%Xkq@zhGbO2}QjTrD!T>LT}Ug@Q0Uf}0pUA}+0SmF6h1 zji6rl_>K}NOqB;eM%OqHR>y04R6l+Va5STVB95;coDneWW~u4+x@7)o@-?VGx z&qAU$FnzqNxzN%#(hyO&er*;=jMO6$Ua>-xO6d|8`5 ztQ|f`Q25SGi1cKTzV6*Tj*WgVq*Ww+{Vn`Bu-~x7^|j^Li=p67Ggfcpm-dyWCGue3U@ZYcL_Z zz#^LDB_aU56s!?kV%I^Ow19hDI&ifO+3$9Kumj#bxL5 zJ$KEKz1lNM@UG0UE1@+8LVp1|_Q~h8tX%3VM2h0@zpX+RitQ?HQIEZ|^GgR?PZo+mL`(0)39GNc7|xaQmE8$T7hGIuWF$+^Ypa@sY6ZY zv{x<>MJ?@+_GdEV?|GEl<&Me{9G3H=+JUAAP7TtD9~ao}&jd6yajW;VoIu~+1H!W4<+=*9QORwlFPiMPl( zAc=N;5hTazl%?gRJLKjF%vaBqAr0j$r$>*qTwcmB6DXALkjhB?1HI5Z?)f}t{*05M zO(3U$=I^WHe*`&uEaG)r#{&z`X2wAakxXH-Rb>EgTxHpl4yVO+)PR-ON!pJjfVWS5 zAC)_dc~f1=CkSEJ3LVq8p!L76d3T?2Uq`oEu3Wote9soVTgMzisExZnd%s5iz`G=g zoNi6D(y<_3_l9*w$w6v%Z#Vjmj|5LA0V4W7%``KWn0N+!)>+ykBZU%L#&q1mL@k)S7}Dp$5A!r z^c>--KL)4fEkZE&l;ESNI8za!_HbuEO(H@Rw?7zwb3v?VAPf$l#pbhZ;LXCn4@3(S!D5+ljc1fCA zn#5ve=Y=mV3+3L40W7ZNYGxcV>BN90A)Dydx#zNF5@Pav?+Ob9lmcM7X*@KMAO>1c zeRn0{c(-_8G5@FFipTHS;6+fewns6D`?=7g&#HdY&|!^dKL}O0x{u0FaRPU=>nAq^ zcrq8O{{na>{{sA6jZSv)%(+1cI}Y=sl5MX?H|yj+{>mjvPT&OHoOLA!MyJ!IMfGxM z0%xvbg|qIoJRDs5?KGh^@AX{VQSNZe??>U|X5=sZiT;)5?;D+htF@+2Q@mzntACjq zolDm-Qd`*a`}IgsYw^ms%FgPSlTUOLy0;(6c+i0kr2X+l~gmg{eQxA!4G?AB@=&AxlUEVJ)e z@p~(DYW4`8KUj?jTYgi=thNaz3yNZ;3D4}E$yS!NN$nKx_cZolY@&&r5WJ>EsMNvE zbH+^rNX9aNh=VGA6NPsHkrf`Yi38rLF&+^DPR6XO+qEE-lioZEIZ4@;M&36nXjh1q z%+7j#L^TY%Xdfd_tqQ>?s;mFURwM3L{?sS zUX9jEyR`P`aoINy;i7^*JC;gL&0gWK^&*m60*Pqfx=0*7IXzCjP`GY*-O^mX=PS8^ zq9mtIXi`5#!3tkGc5*ykRH;@m|5gleIL6@;Es#y45uk$AFu_vk!YHdV5PSK5bwT?7 z?)Yr6?IZFOC3BM7h=w(L;&p_5LOIBS#HtOy`hcW?Izhbw(>5^^>cUboJLV3B{k^Ds z(GDwi@<*$);_tv$Y60hT>@~9dJ)D0VYc@+mfC^eN_DAzcmo*zFuW)FV|K4hzb%?0D zm<8dV9D5-bo}uzvzC1xBJTEVNIlQhhWymjRPYJM`i zm6r{}t5U_R5tTC8I0-x;x$EmFE95%91TqeI48Jdw_7++G?7Rw4v80W-8z?{d73;Gc zEDs#_ygi;=kPL=zGxiHZvhxP9*UNa-tF{sj$2P*>h$vs4OV!qCxK3V*_ZMpa^( zMRD@)9_l+BX(IT#!rSwupHNOZ4u0M1G%JJvVM0hN9u9a2Jd{il+ivKQB z=&F)KC~RibH^H+Zh*zjC>Eg(Zu42DfEGNHN!S#@yxj&8g%w&XgHl1pe>KR!Xho|AJ z9QMj3`&z29bnrgjq$dv(Gm?W~sGfD}N!$SPv>XkfbS)84jg!y~RzM8<(cCY*Nw;zF zH;&@9wQ}~%)7h`aJl4E)mnl6m(e|s=7td9DwTB2oZlSo0V5~UHK!e>*w6#6MP7Fxl z;y}Bd^(K8xdH?a%Y?Rc^g{kO20;y`pRHeroJ+Dr3AQtYhMrZeZTW~&+{X15;(s0%L z`Q_>ar?WnCLXSlDyV}L9Bu<=B`TDVaPg8b}kSV?r@HCs?hsPlxciv^Do@)ao3}{AO zr`J5KDAVz@B$a7mo-T}xAeOS^o8rVUl zQH&07ok%VI#vel_l?gNoXQ7ysU3Z}Fapme8w?q*i~zx6bSdolV_fM?Q0s@^gOF- z!qO@;FkiOJ>|Ra=fhaYJmp&uPS)Qr6EfRBqP_Y2aBMKGm`MO-a?8xz*I(F->nxFR467wk%ht8L1U(_NWSP5VL z^ZEIOW;`ayjpwtAkrUErgd0gc)>EC|dTaa~J7>#Z&8c0_Wgo_s193~E!d%h$mdKV^25N?;h@Jukr?xn%!wTWvN==P3bb_v}ecrvX*Hus{YB8#!6xzj2QV5Zii%WGt(DvjK*#}REH-6D5Q zuebpETL65<{x0UP_%u2MHA_PEBeCNIhsE)X`;U1g{Jz&52(mw(<;~%D5FuHziieZB z;nX>MQx_vB4?FD6ur(HzCZY2IRn`XLGD1f_j6Z>Qkq62@B>X3ShyP|3_y6vT_}}b2 zv@o;l$Zcbls7gjhr9g_f9h&(tXO~hXKdvwpMy8mvF03$J1^+YCqYKsK@}h?JCSYM# zw9{B2U!B9YqVVLrO)d6wnFJ4N*VA|kF6r10wVIBCvf5uZDJa-t-_y8=m-dJ%**L#$^ToyGN*t91;m z$O|TeaJIetX|2K!aH6k;aUP}vD87x$Emcaje&5#3PP~x+(Yk_|ow(5C@xcY)qd45M^ygBiM+FEn^OCN!4Sh+t z_y=B{;csRc!wsc_h`EwNF6n~htdAlod*v(!ojND{fo<|25)Oh#!T}Q?0pMpyHC)MV zIFVzoYMSIJS_PbDIG-Y`KKliI|Ky8&o6Yyd=DOg&0Gkram;Q1XF0u$zxX&qXYFXH&yL9q3rZg-PCF>LhRK-VqS${QpD`~y3O=+h7IXM%_eDf8U5rJ z)jpG}NOi#W{v5z4L&NpFZ*60*BJ8bLyl{U)FmWCmR;Zb`q^`fIqora~$|JaV`EXv(NVmdYmKhxeP&D=#76 zvHjtljpA{fDaP`I&k>7ZUI+|q?px#rijZ3W>#gmZpoN-Q-V$m2qnG?<$29aXRbc}* zYI9zcA=zwTauij%QD;PB=Qady7`kP!`kGLHcZDVhv818HWO^8a%Vmi;%INB>Wz_a@@l$$0HtTO@8dH+L>;%?VZp~*Yt$6F1Ln1Y8&+r*~4=f6H$(_J;1+fLJomULVm;8jDmXkUBMMAlJo4tQ*b5!&6_Vop} z?N93>ZetHFkM=+O+K{@GtPNxmink!^{jpNmaB*d-+%?rd>iRVAZoV|F^scn-raLil zTyFL~)M?912ZC3DiC{W96D#Pm3IWWPQ&FfDQkmx2KzU%=FkPa(^ll8jav_1>IRq(a zuPEX&K;41vmQ%AQXJ}*94Z)>73fLdLG`S?QywJzgYW>rU7h@P@6V0QAxR~^pdW{Lt zn~?JH>ot^bHql-8SLzE@HZGQx4heiLGhmwd*C~&AzmC6p0sOeRls7n+`pWXf72P1* zfIcSa8mCmRVK9#Z$1zK;=4tpX)xCq-Slx$h&##$!h-ck7JaoL6_3A-bgXf^Nnr(nx zce7a=LptT_!3iT-$DGlpw?61CzW7FES0v(oI{56dmj?lL#vnvV8yP==%YvM0 zPWz(xQ9#v%EX~@2*X5pHgd9E+=vvF7aT9og&)6P%wQe@{_ZUkQynt(?OF@nKFX8b! zW^Cn2*;7U5>CuE;3Kd-8K3e3lGnHS28-h?z#4$N~5i3bUwTL}L8JO%3tRSw|%Eqj= zAy(zUcoa~S8KznfXhKKWNr3lk#zA*2KrY%6bFgJ%(*}^`iTOY*q|k&nkg8-;$n&Ow-s5 z1b0iVoeJ*!hFO8c8iWn`LTR44RFG`W65d18d<`42VLqM@4WqEHlHvjakyRWE@siS{ zaDQo%3&3*62=h3aWe3&SevH^v*6^~#bo9jLOlHG&8*?W>ZIyIQ;c3QOB*oM__J$a( z!uT;ZNuuQQh4a-B;NQD7KgIotUyk>se-;mWF{aOXt-@f;s&V*J#iv_YrAr~hf1pnL zk3B+DC+?(;J0l`LR|+lh8DmDvTPZKIl&!lId%9GleqIH=e!X4uF5?N4@qN5)nlVAU zY?F7}3EZsv#r59r)VsF91Q1NLgJNu+A_uyG$&Qb#&Te9`$0|^hvN2K(7uA=>^FSB0 zy`J&K|JZVs1}Ej1EkxefUoox|<1Lr=`0tBoK^?;?8{Hi#)C5}pD^ z4x!oMp~#_SGt^rsDGGYdJqeFRvKA2*A%Gb%T&7j`Z5%0s%!!nP7X@46NB}?NqOSiM zaJ3e}b6%{0e@Lut4M-(h^ST6D>Fnw3kp0_bnD>UK4HkFaz6WqM9?=~QK& z+PFEoGWg>{savA4kNR7lJX0^(%i@-6^sT<0R)K4||1)*~&|`U1rs^iwimA2RERGUtBlaGK0b4-s@lWX|nlP@ZC_q3C%O?E57=EguTOGap!lA?CqIi3-U@xKU87g zj$!`dxWB2=$d_|Blm^{RkuqaTX&BY&sfUdz{g&j4HcRy&U6_m{YHqcP3Lwtd*PCJE z3l69fBdy`oEfjYzg-9s>ChY#21!8KkXbk&}PwdFb{l%mkI!U2joaQ$q>6u{W zte_IH+0Fj^h8moWWBnox*VNQ-mwW!=H-FUKd*S+0g`pj}DbEFJV{SYt@;P);nycpD zn5`mo=j<`1fQ~g0(KU8M**qb#9N?Qx0YdpJ`Yv~2+?eXej>t;>kV{W3lrjDbre+}p zNqbXKLcbLe2X#sBA{e)to>e)05(FP*v8!D?Zgvx=?eI23Rog-Lk^d8iu-%1tR5zQo zMyDXS1@cYr08BgQ$w(sD+IG2R86a5Fs34oEph#VL&#{(`PJk&(P;W2VP`y%_mzf$py=T+i}&(S+c_w~4?bsrSglo2jS~i>2kK&j z5MN*sJ_qXtObiVjW(gNZ2O5BgqhEbdqE!gNg$&N;>t?v382j(Pt5m7G8EkbfC82@8?k$+Xb)qj0lw()J2s?K;M3-RNDJkz{(SfHVWV~-Y_9MVZF^azw+xd8puPoc|NO= zzGr@5>cNJMj58Om{dIXL`Dw$?qFpd_QCCrAe5Oz8SxVum2j`H1>ut@~sc#RJU#HYO zIAC_ExrihF=)N*oO`*T4IJ@u{z^BzxRid=fn7Vu)jAHYeHWA$pNwBR{YUe3$SV-?D z$cwuS1^E<0=3wz$L63m%=vVQ`P5?`yhN)hTtwg6F4N}B>;NTCiyT!~E;wOe^EZ9RS zvIPz=5P2bGB>|Q@mR^hWtQ=vDPL`8|7&@>UUZI0CFB%3;!-?mrMDNiU2J4-Y58O|_ zxP)0WY?eO3@=Xo#mIpV))RSo667$mp>+D_&633; zmnm0@&ibv*gzc5Guk|u5Z~R5mq4-B%E`1C?lK1j1Ixohn>6;_Oom46$?&9+_sc8iN znl^CXv3_`!H2C1?s*Sx<4OA&Ul$sz)C~Qq8vZH=p%YV_Ay-vPX)zmuzI%hNjs;YCf zee?b!W9os3y6oTr$D{POj=k)`0cA&D@+2#CLOQ{@kT6J?AeDtDS4mKU7{`UDJzp-c zg`<&+V4CEho^7=K6N}J|brY?}`kyGJcbYgmmM{W?IC4^(wh*{h2pZ2ZXA9!7HUR6` z8ZdfS$PNPDyIW??*?W5w){b3il3{^ehWUUttKS6tj@aUBonep`aB_)+0WV>3zt_ZL z$c1pgd+$Itm|(X%_9J8apcDSs;K(0H{ocM|9`6vVi|sH6cxBATr8EL$Rhb^EfvWmk z8Fk2RJFE6p28wf3OdBJXvIhI@!GxnWT#;Ypfz3S(6OECd^h8yE#Md$g>bPRUj}Hk< zoB*YJ6hV0!05dUmcnpvELiRR=weyTNJu99I?ec#D*w6cCy^d+);f9-dO#^@kA=aWA zBpd5EbkHCeNwIVZg4)}-M9q06nq*qL+(btc<@t3Ki-gA$3{3==7M@7}azW#bAH$6j zKY}0|u(YV&Tlzz7`U9c{A=ds@CP1@)l=a#I@%Q~jCQ!KGH zPAEW1P)3etu@7705@u9Qt2Eyl+4ab>Q%-i9=ro{m2A_SQ34q&hE;F7L2d z2(|NP1qwoEL zUJ!KG^XDqfOED@03of3l9$1Q+ITcVvN>kSz(!l8aAlmRYNC?U~G0Nbh@K89y@G)YEG|8A0@ zdQBOHv2HzOZ7!M&1M5fVMAHuwwWCK^wA^2M;LJz)iMuO;LqF*Ji6$B)(I`0Kgc{lZ zFCcztb-aY&GZ(gr)ml!s^czp$CzF4c&qb`3xa3g&tgWwGwJ#EP0r61Y9BuAFi_mmC2ul3(pzBzgSu*FEKAw{n4uj^@t2Jt z^t-7X0f%?HMV<4800vry1j~?t@m^#TjUhLTIEBId4$Z^*$g zK@sM|RQ|+xJRjzOwpS)cjvR8cWdUx@7M&0KqW}<%TDKsA<%B2zXv7u+wK0G+1+d^s z&_)vsQ~@Q#8lGy*%85}n#hl^rT|d*miWJMOfc@kMv{p#2{Fwk22wt$sS>nkxyK1?W zosKrkSzWhcQiB9!!%ak{A$JhP+kXL18|jfz-cK4@$5kA@-ZwWGKDG90wJ5YMa7rsP zm)$($Pj3A}YNl_63qy^Mhc(+{HG5d>lN!B%j(kvEsIp088ir@4ip2(4Ahs>YiH{cL zSQ-}WpdQ|&94-B+ysUiko}uVe!XDFj*qt2r;mH|a z5|75%xTwv>YrRA&T*e+mUwDy9=;sZJHGN#Q@YFZ8|E#!SMLNwGvgIAa_GMMASq>~D zx_;CBx!I=l%hZ*`uoK+grBI^oanL6rkNXnqJ7>X~4rx!hNT;{2dYCxg)qhi#8?=%j zs)BE44f$ZB`JnFB1conCPL}#%Me!v2b&%~BBerQ|Z^bZo!DJwUVObnAT)~U6e38$< zNM%yO4nqvq&CNK=-?GJR@!W1aaWubI#RoRAXJp#`fdi=kgy4onLjob3hS7Qq+T5cK z8gy(}4x3=q)utXa%q{Vd8+LYZ!(JKRsNRhL)k>j14FaFC6GlVOzz!L|Mi^8K!5%N| zyP8S%pI!sIUu4O|`*^(c^j#?TDj6FdZ)4q1qrns}fCIrUh9U{uMZ9eWLwx|d z-ip-~AyBO(h_IOGFF1ZBS6`TAmoZEyrv9LZNz3ra^CjQ5YiD$7<`X}`mWuWnbW$)b z9svQ|!F6m{d3GFz*aZG2B$x%gBQp=aP1J3tByHd^Exm}+){~hi4jzr3b+Z}fbhd~O z*`o%w3;<+Vh{ z+hA&Y40OzPv@HvC$6`ok6&|aBHdwt&0hLGU=gJS-xWCF-B%A71G2K~J9>xpnBG;`! z(&Fc4XUdahdy!n}$#vLi7DE$X`k+cyrK4Jy&_&)mj=Mdsnb@3?PK{sQi8xYZud zu?cuU9YhG+62ALU>jIO~TcVHlH~4`eHp3H%kyZZWMB%K4!OK z^73E6)i1-BxvvjSfXjK=6H2qH&-O*i+B<5GhmYO3nOb!mnxx?HJyBgz z;S|!pU{Sx8)w=g)NyI7 zG;uU2WRKQm6s*w+l@BZ&U}oPo(tu%ujx}V3ShRPoEXSbP`fNNOF*qoye21(yfsYUx z3>gryY0@w=!tDT1zedGSjM3c+H2@E=AH(fy3PMu6Ps>y;w#h~cfe;lET}aV2(R+$L za1Q^ToAWwhWj`Xj%*sUWAw{+Ah#?v3^1wT{W-<^ABR<#4K{iAY3_#5B2^mR1ahQbA zNv3_#7;$mJSblzWDn?r zC!fE<@~G{?DbPP~{_G#apy)8ThgTw4ET`;?Jg{s$TaX5pn@Vbj?omO}!^aV!LyhN} z;qA;kpQ>XEG`BEi=KTN4O=9iLE&jm&4Sx`^F(hJgv)?l}}6F7fMW@tXRSq8tM zf6o0AVD?stRkt{*a6@V#;rts+>Qc(;_D?Nl>SO_y&{WrvxtNv&2K4CV5W)WP^c=_57(3}D^1 z@4;xT8CEsq^Eb-Xh z#i^?2%A}TSqrd1ZMT@+)VvWHJ+!_18hWPV%_tC^+65xgJ`|F6s@tQ{oB?YllRbDgA zGNB8#+vimo469R3g6E-&37rVf-1{8}NbLwKk(@*1zyL+r)Y7DZt)US5M>zy!%WZL+ zvL{E>tpW_HBSoVznLEKI8m5ABLC{0ns5!AeIe@hGMWIhtf6l%41GJamD1TejK&1DXW_bOh@yit_&O9PSPaslUl+TuE4J>Bp^Iq`WEH&SpF&&jY6 z>f$V)Bkuz4EN+(*wf8HaMSdf6g2(ND0GA(_xa6}7(U_CJD*>HnU<_ksTikz8U??B{ zf7>bj*Nb=mbJw^3)wRcrp@()qhk@N~f{Sfu1=q0pUGb;I-5UVfD>+PZ)jCS-(;C38 zCjaWwm79X#ot{5B+1D^xJm=Oh(G%{X9$w2%Ocd-T9C-jIl0ETt%E#!EYQy-2ss_^~560sKodY9ve~r03mpd$% zh*&38p{Q~JcIMVDoML69LJ?A$-kz@60GzG!X2x8IRb-#+@4dVg2 zjknw}pNbc64lN5aC0UJf;tD+jOfJCDK|X85Q{jVZCJs9khwVoLa|44 zc;@wfQySa16(1lptV2}aVD`y_6*uA6^AB<@TE3p?nogE9XO!YBAuE`e82SFyBk!zXcgs#D>a=`Z@EwDbt% zgPjD@yH^z1P92Qk-e1JyM1#6F5ACB94RqD$g+?|CQ9_H2(%#3j8B^qtyR3@|U9guP zzq`enDePJ!S+1h3_U4L3P<&3E!S&S=w5MS2st}chO>V}Ip7AZZXmhv9!V|@Mw1xKe zi@-#!;sNPM>5>Ho-bq(&M79V302Qs$N!GV9*6Ptr*l#_yDjH+UDJVD0jK*ALSy3BS ztVP`WyA<}!b{R4`=T2y^egvIt0_Ybb=Zy5u)&Go@=j$K`+t{Enk-xgPE4b%&rm|n( zHl;}tZa%(=2DD4dZBv#&qkkY2Z)xS9wvuUkTSJfv=+eP1bMHV-9PQ7@a(zq#NOpk9 ze{#kK&mUCre)lNyqZh=9y~+U;^bR40Xp$R=HoCAmG2qY@>xLQ6S8l6tz%b8d2+6u= z2YoK)!=Ty#;9OSJfE?=lN|8G5pi0$1Jd!zspeN0u4`^NIm$>Z`m<(LNxr8nroEcCz z7S=!HwhAd&ea&3>)y&teJ-ed~nQW^mm^CxkVLQW29+)7@mhjpS_xKDBCZNC1Db>5+H;GC~wj~ju zzzfa)0xpRk?ZV=oo&g_Szj8243BMX9Kn=YQ1P!L@Z`mECpZ&d;`sc91qyB+0_AkJj zd4cZ3zkpYj!a9UV^4VWsQdOU6Fq=z%^+O*0c8hC(rI=-e<>?#-K73POY>bWbIj8?R zmH1kff<4~7wXk|rYQ{aKX#WeE+}oV`_#j#SL-Wv-)cVo9aeUvy#x6WHuKRr)a}@r) z=|2wB{~SVqna06ELYHn(!YR`2IdHo0p51577yRWEpKT2`9?c01G;(W z6*92`xv1o2=lXgwdG)2 zRG>J8C1``FB{*Up#~jhD%u||{;yF1{4mg37b%<~-Gg3+st!y}HYG!JssO1otnNubz z4yie1rgrW=zx!Kv-9PSH_rB}>z4vu3mTT|th7Fs|_xn80XLzjF-uJCMyt=L-uw7WZ zJpC$nvKKWvYg#m~01jIIG7BE4KOgKQ>P!{-FuLA>*Yo`R+L}kh@ZF}f6mDYNO0gDx ze&%fgm%pGuN8vXEJizGD}S|YT*Z%a%QzlNpK`0j^D~*98crd z)tGOQ^4<5IZP^3Aa@yx(g-MANeG6u!n_oyKu91s%$3#-^l1SKRyJVm;jrBIdfydqG?i)H!Qm6-lu(&K_Xdq;}rrdBF*u9*BngHCEsz zB%P&ctHMX)%a({%B*@5U?)g$EGs>(n&R@P+&+Ot7QKie@w0ZXkp4S2Fqf$vC%7>Wi@$Y*zAH6voi3#X^RyO3^=dl#K7;ebaQw^Tc&+M1)aGR zd#U=p1~Hftskn{{e%V*gZFwETXh;}KEEumFeQsAX`8>J{c)aOuhOa-|z4=%1>9KTy zo3HJsS>b|3tpp;FLtW>6F4qI9T}95ttykI!V~gHkxi z1n0&nQ`s)hxo#Q1Yji#PLMJ=t(=i_Qg{gJ^Po|?Cjf3=)EqaQNRX6Sa^n9UN7x$Y7 zv$_JuGkd<#g56dQ2i2Dy9DOr(f6i zcSt;KaX5FP_m1JcKtfl5ci`W1L01!^xW!@vZppe8gS^jZ=$iI62KNBkL!4(9S{I_Zjvr*y0nHGnRuzX2)gIE$s|eHR)4s zVfs3a6126^xvCm=Y*W|C+^*TyTAi~U8nWqVJDio{hN|n4>KoDTB3qX$N+31FfwWmDYW1pLOABXRqp-C~XP(qTeOF zHFg-3!H#t%8wJi>`}nokrt)<~8~qoXvn0&RL^{ehNOEr{WBwq=g}%*MtgwIRCbwJ0 ztYM(5bLNcIcldFctgP?H**G|O%=(^B(01^cjM=%O$+s?>FU_U)I)WcNWb=AZ#*nS2 zYviZhBWp2!T^srUpPME}LKwXpr7 zibvs3Jg|jf_^4&6HZDiBh=)|uryOEo(COApIiGTf`QyA90heZjt-B`%c zakUSP&bGR-aX;vV16F_Yw&Cst{m)BHm~KDE<45zmZV`=SM2{K&q2sgkRt zb%g!48U(+}+S9s+RU9{r;%a>tqyKXZB4l;sbY{#?iOWy3zmxQF|=`F}~ z+I|qM`AA)PTx+}73R--I4yAVOzF#!3YL%4r%QrQ02yFg2?retHOb1V|Ve^^-el-OT z{{bR~LM^v0a%VoC8nXN0r2+rd^dBStd4f{|H;$p@yQ=hh=h&vz8R1mfOVRuxowX3M z@}*YLCf(hyr7qS&dZX;uaGhi>c#0V2yL|_^65R_2&%j!9_e*^?Z~LMl&unH#=#R?* z%Um;~TbfxJLkITvML2Ez9iBhE>b_yWSYlD}p0LN(kZAwisAz5MbDmTS7-_%Vy>ZWT z_VA;%sh?g4EMI`V!))`U5PKp#H~uzDPW$&`MD;vO$6>|&8MB%XBptSSQ1rp0&|k;R z#}EJK>sYQk&SVT1wcl8GoMa{N7vXYYkDuEvwv`>+p?77G?XZ7fu0itD$9<)Zu;fpJ z_YDH_tnNMgJaFozx85>X?SrJNGev9DJwP&3Tr^}3zx(Px-dD-`xfYz{fKEj1KR{sz zae3Q=eMcWSUV2;`=zK+({7mvI@c`y!t8ag75h9?&4M#tvB00PVh(u7$ppbt2|C*t>)b-Uz=axE zCxG?)xaw4Y;Ja2b8UGbjm&Qm4t1J9RLs}L;MevBm#u--|FB~I>s91OzT0ZYutAAdL z4AZ2BOmK?vsJIE*Hb-2mmG%mz=}4;}El-rLySG#1ZYw6$HE1DHhcWNEilVlX*`snZ}c(fq8zCgi<6q(&3`&2l(DIJW%eJ|JRnkj*gG!S!l%TW~x~D zK-hv-=mB2Ix$3mhmCHX%@c2&ospIFe4?!CzDlhE#!&*c0T#Y#_F;=|6gdk!F*?zHv zvxMEZ9JcM+5$Ltk|JF+fdx#8h05juzaW4|dv+{A7$d~g(lu8PGdNYNP%F5F$FI})p zB#rc*iGdJ1_L>*?%kHQ8u!I-51=7+>^R6o77a9ShG%c59P{w175+(Z7z;A+t56rA1 zSrz32AuA(CnIO|Smjv+OxuNiuy{1n(@FI7qomDk|7)?$Dq+RQ@HVQsgt!UU;yh_>k zC5!!{7bRBBI&s}Zb5niv&zpNQ-gc(=fQC)w_NI6Kd?>W=Qt9nDkQ!$^Td3?jx$U_~ z{~3QZ;lY}$*1hH5zD^y9*r8hF56QiiemC!@kOYY14m^!4H}W6oL;SdQHGxsyncDK? ziN{~hTq|;#?nqhel>fCGtaJr85h7r2KDk@&u*G>`e)h+fLk+G^y)TAFuRT`?-bcCJ z$nt$8fX402|H5{LRq1N9+M$)dk`ck^i9t_l;#kgVjett_8_V=`m3C~8q0}*)8z12# zDi_VSL=A0+>pdq2rYM(_!NyD`um7HsbbBSzwf^m5x7f3H}#A7|8kIlN*7a z4=p&No0Y3Owp2j=LSL&}Es@e)Ym}D2Dt22_IHVSN-AfVanPW1i`FO6}cV+$_3Nsqn zCCqct`wKXmJECpL1g?PPZ_OBY!VT8Q#4T1~JlEcp9YyttI+B4d$bpizD~EByOl9O8 zm6SBN+-x~Fh6*(cRLW}O$_6re(_C3>vH;mCs5j^Cq-ofP^31TKB|}SbC4z;-Zm6;4 zL_uRVzDuhEzcS`60NJ$>Dkj7hOAXwmsZQ&#Mg&42wpR)Uqj4JIZHpXSoxzG-;VVfl zq1f_fw>-H9Geo%r>EP-Lvy4(1>%So*U@R|M3I$Tm1H%(Js6wnH!__S&9?TR?QSW=c zRn{UWG3de`K#1W9q!(P+iT|p_;{Lmi@|0_Q zm+MKD1T732`7brX`iE9|-`F=x@1d9T;Q^F3EeAm9j>nvZP;7nKKNe3gS}l)>U#|ye z9IrzQtU3N}ywwFBvK4iK2Sp&kA{zr>G|e!*RFz_xaMZ(tt9<&au?0|r;2UF7YmMj6 z>an*uI#n2owFjPe^Z$_AY9%fJ!qo*tmu7(0sB){$r&M8&QNE8Gw zslv-32-d1_rE6SH40#d&vT+Qz;3L!v0Ne18u^eHx6YIM{o6|pMT^=9QX6*nRuxpxP zEfr)VL(&5$8&Me zap`i6v&l`Psk&m!I9n39G9|_EN;y}53(^W#ip}XZJI03 z96`tR>~C4+8`GQOt8Wz(hZVQ=|5!n#8os-%*)ADWIKAhM2d={@#wN(qeres8lJu!m z#z5!5!IJ0>(``E;E*rj;+IMLdijF4!j(r9!?Qna|kenK@>r&;7-LQbIjH9jkn?hsj z@b`B$+GFCaVN|5^-q-5%VNTy*jNF9IqJsm~Hx~S0J65Va%2}3i^ zsh8y(PxR8WzG^8H)z1*G;GohM=1G?tJDb|9r_E%Tv>M&Q%`kNh{?kC{aDn|fSaok{ zlcpWQiX(D;I9xv|WzGLXTwv_v-Cq(Y)gbCEmV}LndAe$aD*dd@d1?+GhF*Zbig5t+ zM$X=9nk0{%L5NID)W4mdamP2B6(TcP(qOW8yhDzuZ6zgc(CZS)@QpccMZx0_=S1KItJhMU;z&(KGxH!MJ?wd@)>f zZ3txl8^g78x8`({3U8O!cjzibQ7@DTtR1kZLa=0>4@>E}%-hcc>uI@_UQ*r;vb6u} zmgfGh{s2bD73=gxNK-ohlSFUowD1;aS&---`L2Lmi-(hI!ZbBhik_yK)mXgVr~3^9 zp!J@z77jNqPh<9`C7{Grcy=wij0-dG3AV}AsT-I-&yOxFE-U%9uUM}PfFTH|D#MXX zlZN>3#V_$q;W6KevpC?C4avIpdR#*HfKV<)lw5!_iPMzAnr)afpI4u>w#Wb&U8T9E zE+R&{Xw-@$hL_o;s5p9t$IAMq@2bbS1*?VjunMunP5@S6iENS}d0w{9qw822K;TIt zx|ee+rWgr!%l;sbhOr7O?;+G*jVy4wfnpS7pqJzQ{F|KsE-ei>xhNN>T`Vi&&!TZ<6C0B61a-~R~4;V5!fP9sc0~(yrMkM@G zy-%?wX-o{9R;VIgxNkInBp82T^}cI-Gf8dE8mqkD4wMq4Gfih4F@T@~0aw5Rj7L?n z%z%mfi&ED!y{CwRv+PQcYz?xAL0t(OoW#Ha@2-6mJH%a(O=>~vhA_|VODHh|NO8Pk zIB;7TULR%+zDNPdy4Leeg|vd z<<1H&92A4$J@wJ{OhY=!qr2}!PIYhXz;>fJZA;Ek=c9!7 zjJg@WtAD8lfYobjvI1+%4(fd$M2Us|vo$?0beL z?T7#|Q4nm4>#u1AIcUKTU0$tfO+C`lKQ})^8ab2R)9JL}b1X2YDrMEy+!eSQK`^vg zMG4OlU}5XSRr`i1m(X!2p$WZu{3O2~aCoXORb2kd(BaU*jbf`+{y#t*rgku4$oXFU zuA}8(0|MhgcovJQK}Ho416IrXLiV@YLfu0297p6!cAC_-0l}yxH(kKmrV5LcKBr}= zab-*iz(6klxP*m5c}9S|6=gUL)=q@eSgTE4@L|<)PqgL}rhQM8h1X7#m+6_CcIB{l zrj77m#v)5)8KSvgi;UNmu*sjqe5%i9-K_;gV==w|7A?LRPxI6&w+PfOtOSN}U$J2M zaO_KBHz6Rj0Em?SF2%mg((q|(ArmExj0ymk9xuk1%!8^Yp4IZSMQmwK%?cDMq-l8I z#aZrn*yGVcmzEd{D#;xE?|g<#_w?uynr6!Jw@Yg~?4~YaKt-cGZ1`3Y&ngq&YIgCP z6BH)QloKE)$@R793R}Fv9f>f99piM$HP9?p0u~fk_$jtLgEf#CEvPW)7J^oOvY2K# z$bCHge^cl{0@)0t)e;>Z2DJhQsH6c}3Hf zg`kBjZRe5G^zMzmwPdVhPN#^Y(gR#u+A69Jk46`Q{VramOE?N~$~iUPL1yPn>$D{) zz0)x&j!HtYI15FtF%p9bNffx4q&8kwp7OA3wM;ij^pX{)eQuuW%|)Ri@I#yZz)6BM zJ6VQcAJJ;vIOW*wB?}|Wke0G&M13EKapQ`0vW2Ke4Vh=5a&hHOOxzh&12)yW>8z$R zXctBIJ7X27;@lx;oPc=kC1=TgCF_{L9jk7~4_bEXeNs5auAOomwzG6phIk}+uYzc& zdE{)@PCAo!6bNJ|l#G-MJWOXDdxx5)GI7iN^NpPzpz$zZ8Vei78wj61LcNd_!cSLz zGmaTjaS!pJSQhyd$cLO}LU<+4+#Zfd##Xd=NA76n;MuBW?z8VV%-KY_i~Ev2rT&Ht z{#dT~-D@-Bw`b1+F6Ylj-~?;%c1I{KcF&MYTghF}6h?}fjbYuMguuUC`u$|y@!5g{ zLzzxh?&_Lw)(5>Ut5eRNRa(hqf_mevx+^6u;pSaPrVtepO1nW)HVAHFitBaNyeNmAsojvn z1)ejd(3aH9Sqh1cfA>}&X)lf)!R9~UF}e}?E4xSW^1Zu zjiw#CDm{#c*5Pi^1+2xCT;Rt*$mB4~sLa+B+c<7F698g3Gao&}Mh`m|0k`wVejfR) zCfhXd8-pKg7cjO_52Hr4!4zna8GB{!x0?dg2e>hf(l}`;G|5D4)YpkJEZH9r(0cRO zIOx+VT_n9`<9-jN;P>LJi!I4yK{+}Q35APT>Wje_0JFuE>CLYyWQE1r0oHPdWijRH zyjNpH+oyjsIE)TuUL*3|q0M)cZjr#S=F@71e;C@RJ=mZM4Cn)mcT8ilTy$qU@iQLY zxhJo-O%Y#7INeaycDaH6{KmiqQeZ?Ip|#{p3&MDeBxSM{2J$a?j*UdEaJP^U<0&^n z>)2_!IsgQNvlL`zpx%hH#Z(~7c#-l=HxWGEgwm{wE=XwXdZ%m(WLi>3~9iXwS>2!wVe?NE$`xFD=$j^a7PQL|MDLPw2&2xf$Z zH={6`iR-fR2DwBn(Mu}1TPK}7UA+t6r1;)M67XU+IFHqIMY`H&W!0QtE>}!?`D3|M zC*WVmc5BwaVu=A`;2U2(6O<`#Ewbc>Q01KZRIUlQlMdL*69u-F6 za4rC?gM!VNp))Pm*K;)0hF1dAk1Y&j4dzoTDoIAVt;z)yr9Ag6U2oA7I4{dE;~B=dL%x=J%${;f`aELR4& z@HX^<;WBhDXQ^3+j*3_x&K;3&+VyfZh^W9sp4byb;6EIc`u%S?ffgA@&mJ4o9_(Sh z4p-8)E~lx#ReI*>VbSduXyZHx>sT^Z{}-NV2w=cw!B%FEWO+>!U9M6 zJ7MZ3D%hjxR>J+R0a3}5>Y*OA!C@y2){SuY7;#t6P-?oBF0#wD`~6*<10^AAJ=ZeE zG~3bgsLgnx~c7#%N6jbyh{h` ztT-aCLzfFoFnwGK0`*GZrF3U$Ymm1X2Ep3hy-owj(Jo9G06XPA>pANthC`GZraSh& zETY7iJjr}n(R}qu<}W#K|l zLHgNMbmn*~T(^;bPV+Brri`9)aHS8d0yY5>OKZRo*C8V8oqlC}=jYtj2X_8cz0HmC zthD;63V~L~HSXLxgL%kUb#M|p+qm8ex~|S%Qf+ zx4_uax{Z!i%TOi|saoho5x_+p^~K~b{UxxRG=5fp)~#LXUsmg}5ajm|@MhkciK|8~ z^W$2ahL?HPPVZngxeyI)7Lv+YAnHBilp(D~3vrc6uhSBJ2$ptM6i_uHC#~H0Guk~^ zAj{DV10K)!=Rm`x7Zaw45TiKrJp33HZ zg$DxP&4ADT0js=5H2gN0%_ek0W0y;-PrZo$&X3*{3vYc+qk^KvE)*nf!WY9NGxl&f8H6l!02%|K( zm^{;)lO?hyI}0eiH`3%KU;z8qY+>baboUBhePzZai(LVVP*f@I)`_CF77Gz^GQkXp zruSy1n284%EFawH<(BNe=VqyqAB=Wyzi%}F7Ubxt^G?Ir8R$g6V0=D(Hhw6N;j4WE<^ExjR;2YU;Yc)Ycx%;i1zcdQDDzZyUBo&X8 zz&Ky7(*9wcgJjyyjZbKLp*%k#ScK3DQ$Q>ex=3L`RaX>@sT87$eiogT4klwT{gq|e z%=5c)r4SU4T`#{0PPqmnq-tVZ;P=@nJ}jzmqwloq*lM$7wBSf5t{G5Zj0XuK+z~C^ zJFKY%J0AQAat6lcfDE5z0Q}Q~wYvkM(rWR%rdKFP)5!aDz1X}l4d6~WPEjK;cs|wj zj;P4Lx~$e*&(p!nkayTL@k#=sbd|r*Ei}e=r7Q!>Rz8nCU}EpG(Te_8XKStj`cI_t z+3a2%nP;!F(0N$f(%1c*(k*-9WrkmbZ!!-q<#(V@CAqe&*Y-JaW#0E%c6V2@cE8&> z)}5yH((*R0OV+rvieo~>^3aJZf&w?aLeF%nH!k-dpD?c?5be zUfSmQ^z6}1mn}HZ|CW{XX~uz(Rcy*UWZK;gnGu+!r~Nk1w0!^BEUpCN0IM+2t_12a z9lL-D8oa=jBmm-m%eW2F92{qgH6d*V4!DqH^KAGbv5k;M}B zA0d`6JGxB@>u+DpDc>~NiC;>Oy z--@0YJ-l-5#~^LHz@zGIxO#HooNt%mX7as|->!JVwDaT%ofbZS?TLD!5CE7CfQIdK zSi4PS+Kmc5F0|^8FlUNc37yW)%PkWcACb=a{Xj)cRkg35B;o{nS!jsK%2zXX@_6fk zl3n`;xERaBTf6A&i}y!7mX7#}Tl{d{Jo|)-D)a|gCoLs<;0Jk~ePTj4Fje_LYunmW zRxDTc`q;+My27EsYV_ni5Sphq$j^CgP^t2jvh`vz{X{F8T^#tu4+oEf}>rqBdqsr6(AjHJo{=-(rO;GYu2wjIU0)TBuYqcGY&` zz-6d$EEG}7u6JqW2h&0Cf<|C-1cV}jR;HW>$CL_l`FVsr%E}*Eqcn}zS{Y*EPwgtSOhw|xRHJ( zhXETuMd<)?sP`6oaD^X-0)LFbGrhS|hs1;qi0At6br_`Ck)SNhWqTZX?!yd`kiR68 zo00U9&CmWo#PADqyW!i_^HHC~^)l3JFGk|{=Ua7`KuN2bV+_BnsRK(ooK#U4HE(wk zK`(TS8NxOu&)={${$&^lfV!*9;mU9FzqZ&!`b(v$_)1a(L}@gqc1m3@op$8y4*fXZ z9smk#Wu=^_o$2cMe22e9*4()3PlZ6su(62w-i%DWX~G@57-vH{Yj-PXnq7coi7Tt5 zH4=gf&GL9P#%OAsB@Wted2N$h7j;TSegve0A@(HWgjsTfa$_S49{P)MKs4)g7JO@_ zM^vJhl-6hLKS0PgGVBPTa=G4l>)zK|0|}#9UQ!QC*6_WKNTa2F9LH>W#@ zSV)Rjy!BM?q+Yc1I~6@3TkAw8%Il(JMmN8=|x^lS&fPmgI63RVsYJWu+Xa z>j0nO)SNo$CW9T0*MhPu{OD({kh2? zP|1o0r9wDoK^4~QbuSIld)xV6Ng)$Qguv^|vD5{ktO-^|7Nm06x#^E`s6GizTt$oE zRK7q|fBw?X#roQ2N`Z67&%%7^<&N_YhGI6suC`96zdY`3+PlOquG&o;T5sx!<=G0- zN3u4|F8`oKh<}iCU-?~=)_tpP((zbU9Et2WJ00hD)RVH#knC@jvx;s>bBpSA4wpwuM@h&25UcbiVnWt~Nzs&rOyUX!);@Ab2!^JIBz0tW!kf2%2|Lm zh#Q__`V4CQ10+@$op0K|S|)|#(3ik62zqN)+sIF!%ZR{J6?#qm#K;$9;2ny!wKp?F z1{txkbDy)7WcN*)^O$5eO@kmXz+02=HG3qqSomZ`ImURXemP9%j4gBJ+N4=yAwGyR z(S>V?a z>88D)t%mJwz>H=BF%trK!gsS@Z$eNa=aq0GI%JEit%75dK$;^Qir^|kuc@L{%j5wF42S%LDIqK z4nXI1ixd2n;`uJuXh%bv?>HA$D1$}qhB|kG%yLVe+=brT97&p^jFPb`>>H?s*I&dw zAattkEn?zvNAs&>1ZhpwiF?yvY70$CU63S7ltfaaQ8AW3MTeI0 z_^t|IHvxfK7ple7!e>jeTEYgp#dpAGpLwf-bGAd2<7#5CyeJ}ACenRR|LOB&8YXdu ziFn4Ec07TrE8vPH|BLE_JoNul@8bUl$0-uwpynOI2JQLcI}lFR z13LA&w-7L(gGf7c3m~bm2E@86rL#?4^@dCA8tHB*Z7diMIson;557f1M_GiV&%Enuhr%mbA*Irga2v|CO18iNDnT8waEk+mSi+gW` z5fDhQq$GVA8AQKUPjPnM(^w`3QIEc+RxKDcw%c@(Jv0ef3wVs?uvhdHwyrk2%C)h`eD>D z3QkLC#%R(=CJG}N;nE*E11|9H?|9a6F6!vL6iD+@3HZL#n4E5CSK39I#wjivSs8=dSE@jE!L}s2O2)oIyxs>GXk;xxi#nW=)Q7L&u$k%6GwW0pm^q ziK1(IwlB^inh5aFDN@LOI0HH36fOfOgSzqs4=a&_)Q$=%*82jY9&gqOu0p0&=3cIl zqO)fXb9~5+UQ!`cpTA+@9=Tv&D3zY{ToHvxB3+y{|PYb;r!@vR&s04cNGJBYK z5u{WV!<8IMy8QjltdzYi=PGN#%U+xwQ@z2Q>?AYt&yaZJiaCbX=-`al_{xhA{utsPq~A z+-c`bH9((p!R+l}=h;Z@V*Rg$HHxc;?Th#l0TM~@KV)osmCJ=az0ewwmPu0#sg}B#Q=8GjQt(4@ z7~NjL30)fEJU4j4((`1c&Zw<QwBOJ`tzFpnK=p znxkb)&mdb=a5L@$AIDp8%tN{ghcca9gRy&J>}+c8D`(`RCT`InpoJuKEU4OQr8NIr zRKjlF!0o;!NiSI16%)8}%q?R|fuyhcP|)~jOlgtY?F3a64bygFq8DeIO@kUinwW1> zJS-BmP!RFHNim}(Q~ghwqo{(62uSMzLO|d)$s^ff{IXcfl&I8+qj!_$y7VY{0)H>y z&uz0aZdQRZvop?zo;l*WA;alGo82}i2_=6*UV(-Yx1v;h&x354=&^H%#(9^b`~^X8 z!RtruxD0v^|3?;09I~Y3t zWK<3vbS8n*2cex*VFVr)1|lJHkh+-UF!4YIpjj)e^vmqlb9HtYG^-JQ6ESJg4gK2~ z*CUdjA7*pU!-q{E&43Y|E*dWIzzM?TKtbbqE>QFq9?H%QwYiNEh(Jk{)py?BE>~5p zW&W`KAuTeeyPkG4Re^(qS($DvOLNRC-hgx!lEQ1UYTmKZfF6g5Qz)Z!dw@}t3 zdgEl}xmLnS%8a9sAO`pEx4!HM>Sw;iuC8#Y!Kj}?+9zoUcsWonWGtgut6+MvmZ-hV zlsC%6jS|(0Ix}M$DFVEnKdPDyECTHBB?k=H$Uuv-vN! znDv*Hj4Y8ARrE~t79QMyAdp>9zUxijjX|ZrxDqe5Eyf;}JvR|ixK>#j0DBi@>3fBE z7QdDsTZ)y*4-9382ls3tbVc%vw*>%vvaMbzJ>5eER}ilx6>P<5D$%RfdmR<+c3L}@ z759}qxr3rq#6xS#E7-v*oa^HM43t3O0F%pj8+rh9;m^g_<1_5#_nFb@>w_98y@M`j zeG%e)8sA{4q+ju%wJd>eGx}x@A>i$k3ydlW1l@a-izCcK7zMtX=|deNrWcnbk!G}z zC*@i|F)2+=%_1kyUYO+(KbSIZ1?7yP7+Hr38{SI9Fly`V58Kq$yr6S;(3?*6r}}_nXo`k2-FRQOa~m!9MP9ibOv*WYo<#Togz5 zK1uL4+^#fGYfz9G7)(C(g3_$z7`K1_lxuwy(zCusdXPiNsap?{{ z-bBtL|3d)ou(M5Nv3@8V@CpM%#(W_w%2q?xY6X>IcB!%K@Do(Xy7%9Q>e4tWrFn0h z&vZDwD4d|n8`XTXEFf_qj9wbD1KF(HR9Zlk?wzCWv+Cw&06tX$LSZFN3ulTXS$aTUrJt#WD(#wjpEG*w2L4+kDHoH8H1xPlhKu2Fd9`Zc6LdHX-`}6s@67MEWkApvj=g_gfV{~>HCnFOrE2m9RttsX4Q)UIA55Ol)gYRQ z4ftQ&16AU!lgaC#1EVp7v&EuY3eyzCUY>vML2Tq^UDu^SO7u;)H#SSDznpADbm#?F zSHmB~YgvenVW4;}^je<|z4QX`Ey^9wY1kFb3~CvU0-@N3*HeWP1#fa#P=SBKdhI}_ zs#}kPllR6HKnr2C$l%pH4RP$+QqEH-O%~@&%b3f_@CQgWT08g^#zijHz#+O~TfEX1f?rbV+zDA`F+K^;x++r-MX zfTWUbt@_}MjD{3_nVE5=qGpS=Z-J?k9I_3~uxp9;N(rY(nXaa+?2g_X)}U#;F+Vd} zAcvo#dz>Bn*5DuMr3weM8Te8O2jz3c(rIVL7m%@_ifDs#RN{GGO__FGBpJX=X{m9P z?U9l@haV%U&x0sZ7;kqrO-V=+{#Uohs06&3s>{44Ga%NAG>gXnI$sxMq1t%^Rrs&$ z{@nLXvh&zC(;|OhjKlI_vKG=@xDY$ZzbmmZp=qr?4B2T6GaEli1G2;RdJ*xIH27%h2vb@jcBm{)@({=VqW#q2T z9H!GVO{>{tpb53#o`h+HjT?|hx}lG88@lJXSV+d(Rq|08{J?71EQ7w-Eyc);uC^pQ zHxULJ;c5*B6(r~O5e+OU&p=b3Vrbcv&zNeidD7UX;jW-;`G9EB*1BrYzyqR-4mT&P z+AR0#inDG0e4p+5C2b_`r*N(fRX&m5{QD(&>9NAx@fIf^cF&#E3yspDy}8oRsG*Pb zfUeQ_J=&+eSshmA4Pchf&(hOmEzg_9pqnhOMBL4nb)1m#tTJ3wakTMwK!(oz=q3*O z&V2!I48c3ct+aY_D-e{Gkt-g~uYn{OjF%sc*n$3$9EK43K51+mV5Uf=BfH+Rd>mC` za5M-X4@(#j0Vokj)Sgs*>uRh?(21+ig;Ym5y=8bnYrImta6}bpAht$;7)|TTA){q3 zhnv_0aRS*mkJiCPoUXCt98^|kR<%I{*Mq_EYxNlY!0b0))<*MUSwRC}nH)H3RGqb& z@l6N@2vO7V%@{%Z@r*r*UYdF-rwO*Ni9pgh%gZS}vIM`b;QmXthG{M{8)n?3w@awN z1twsc-C+I=6gmD0T?#c4!C_xgf>%Amqjc#@sM%%iY1vnco<>svoprm{H$pFY|8A+x zsUBK>=;TZ2IYThW`Z(piGxUikqGbNC6O~~M6BZ&0!nAw7xVZ(;$c@lGpV;j95)=yL zxka^njWIlOHP$%?@pU*4+4*%ucs~>sqMDKmG4t!f=^^ow9F`D^lyIWLN?+*_8ab(D3(dg{1FH7({h}Ey*aUax$n1WreK#7@K zQVad&>EF7nlEYsO$R45WeCT;7eEdzcl}gv1!iqEEDiVbGvj%nZQu=VVbRFZERcvWa2>c#HDvrPmKWnwMT$fjYB z5kc4X9Z5;piodX|yZX~_chP|dKF&E(Paf-<6#9ypdz6p01nqpOb#UG1QybC4+Su?o zQ=m6Fyw(~1RLytcv)M(3RWuoRi#>RBm{EB&)V{IK@_evePK7cQfVv{ycT2Cw2aaPnVsKAg>QxP0hMHTWN7;PCS8K zKoAVbW9(&6Xqd2lgfDZn;c+=4s&RLe(}G`m6U}t7sQ#lh#)y$!k~CboMKXN2d?nYI})>Q`(Q<| zkVCW2m_yC3aMPTO=+(U(d{dw5+Zp#J!btLGf9ZXL%Ls9AZbsgZ+0AK304p7W>zvO^j4LfvOe(IX0*6%bT5|LDkC29-{+^U4t@E$d zNfHoKoDO0vYVuZr+gD0++gjdV8?IdBZe1+5tPkry6dHz%^arl!p=Kq~0<)opeX%2q zzMe~R{ZS3^t0i*g!BIdr^f*L@xuRt%EQm_oqmiSyl=R6z6wtn0aA@ar1D2cKw)Uji zrJCLYXckwNu2413BPiNm76HO>ULD_$D1Guy>as>OS)n zX3EoFWWq(p#(0zj6eC&27#kk^z11e@V=2hF|BleVP&xt~u*t1?t?gh=Qw0i$<_1wZ zdktAi<5W^PG>7e=`8N!W2N_VU0x&>e#jV#gTc!D;9(laFm1zf-1h}#xK^a~C&jncN-CjE z0w1m+Y+@iVFLMP50*YQNerzku`B)^RX%=$2VGZF&#+WegA}Zjvp3$CTvssUkY%t{s zGXp}JT7jik3fOnG9ypkNoe5Zyzyi;Im8uC6?MRq$06_KPG9Lv{&zFMW9zPr)$%Bfe z#N}D1U=a_2_%o{z0~lp`z$6~peW>>g#wvLT(lT5L)DCQZu6OxZl$+EuVi);k#x*hH zMmks8yX%O~0ce?3s7)?_5Q9>t*ME;);6hxWu*j2nhRdhw*zu)hUnte(vmOE2CcYqQ z1Xz@{Z;B3Q)U_8~2f=kP>*07g0duG7h$_&?1xTKl8M8ibt#wPn1I_dS~9QE9BeY%C~O53dL+7s%*6 zgA``waw;bT>~5K)bgh0$lAgquR&B`_FlsrnFjjV=hR=0Nb_{5@dM9>%y}pd(zujQG zO?(cjRd@Zhxpb;BfqiVF`M!LbCsNC9!jIfhvVVlxTZ>$&Q(xPBppf!#R#1{;x&r1qqcYhtp z?i-j2xVCt6|E|EHnTpIy$hB|VYq2c`IZ?WERvI_eP0Fv8AI$16?W!{HXPE3`)Z?q$ zjSbyG%SsI8rkFr@k-y)`wTZNwbL7~}OLo$r#&9&EzN?-fg2thinveFsWqgD$$F&_% zzCO#Epy@kWx}wxlU5{FpK$0)1V}}RmB#Sgdtm->Cs}-)YCh2bMsr~wQkQ>-)<3~BD zER!3P?+X>yKm1mgCpQL~*m{ynvP{&r*jmIPPs?dI^PC22kpXN~tKNnyNd)bpV)9Rp ze_oZK@Qyz`ve0zSXv3;0e7}5Loo8P{7 z8Y(8lZZV}f9f;5@+t%Jy@4M=edRzWV$fq-g0k`WR)-p#f;AZ#0&<7$CTYtM#jBJ{A z^7bvwD~TD4a#vDztThTRPAgGa;e_m&8#3+Zh^{dqGFJp>9`Q}6g3C%5b)WRR2eRL6 zNb$Ri_co1Yjh_a(pxhFlOfnEcr#l~07Zx02Mtm(KaA9{gM@DplNm6JP6}bo#!(E3w zJf%ursie`+wkFF{L_aj|@XMPu-B7nsuwkjL7MTh{EsDb2jn_y?)h-O!#!=KqxZaF_ zGy>Vl)1pUv7gattAMZDfe0rfz&Znn3sWs-tTH~$gSw zAb~kDK=?~>HDMxC&PTERzp?iI@l5ak|NnC~8yYIBCc_+RmQxXj z&CjWKek?XNoqEYjq!V*WhWwal2pwn5FeS~b!~Ce@ROcupMGhnR(K!8(AJfJ{C_j=P zb-s_j-@m_~@6SKFxHOl`ls%u%`{RDU->$bCPx<0P)WJ}_L(5Ff$$nRz##&7;#6UzntE)HRy-$7v&MdK6&BbD+>& z+8xK6rhTIvPp^Hgt?5(|Mk?~9cjtjjz_-Md}og(XAFX z#!k#2Uif0UrZ*YIkAM_+y3l!(2SSh_wejB0)DTj;W8Z&c91Hkw1sS(C{wLZKC=fSe zW(~Xd#Hd0RW_;U^^ks~2u~_b~qf%-cA00!QZX=B!Zcw^Z86ROyIpt9BRzo}`k9kN_ z21;+G3jI@-7m)8N&;lAzx>1I33UoZ?8d~2ROs!fg33Z`9aKzXH;A-_l)>~^2_7Ky1 zx_6Rhk+J+Bm#RIL93~fB8pUykn}}Lrf}^PcZ|Py}E|vr<(N`fPC!f9Ef6vT(0{ z3-j`0F6=rRvwQMV%Z-Y;OYgPA_Kd=m_#*il_9WB4FeMj z8@{&!%qhH;nz=7Aa>tm+{n03wlDZcOC(1POqp4e{4P!S?}&Pxs0 zzCUHn52CxsFc*_QNu8GnQU_xo^8Y3^{Qu1ht2VC84%&YFsUThoT*frgWC-R1psv5R z%3OGvC+FgZwchftHyBspw-E56__FU|s(1iEm>ey=BIazyo-?j~s4>Jl8DA;VePv!X zL(y{m6oayRZ_p763}t?NT6Z90@=&L27YuwP(XCjZT^)Nlr&oE;iPSRWx-5m)JtG4{ zvW;JctewTbiMi~Kcgf6suECzn!9w~Fu;KB8w?|91;7ooKqLs7Cv(AjOBwA9-4Altc z9woa6!Ey~lms568C|~!EEmmcm{4m_KD>w$po?m6cUUxJ*&!}^fAT>f6G2?^w@K7#a zXk=`L2+3cb4;Ja+ahJcAwo*5qT~j%bW$tZe??!D+vZbW+<>#Nd-U7*2KjJrg^xU)F z6~8BFqiyl239{k4hl+q}WFskb=53L7q0Zy;s~_iRjO(3h@~pnpENJKXp4N=f;+VSb z{Y>q{(9I*MtcCLZ<`;1>Cn6lUo_qJ9;;y*H3$G zGju}gs7{pBMjf^Uey?cCKKI~8dSHym638`+eQSgrz!WuN`e|(4%Mq?IFFlqQOYazZ z9o1T6T|G_#(9jgK*T|iOX+FGiJibL1RBUjmC=4O`xKy#OxN;(qWNyrQo>7O(=53dim~T$Y-~MZX4FoiWyxx+4WyZ!Rk#Iib5&&>Y)F_P|C* z#{$jQ{#5_$b~zD+N887X`m~;Ku_BYj8=jC=Ol6Zzo>z=R* zA#ieU)Qh&mgWtU=6vAh8W1=Y@AHg=Nq2G$l6tAyg!?SY&?_k- zqc&_RLc_V#C!mk|(JP-9wO{LPaZIv~ZKj1E;^h(K+oUL)+J<+#ZdRWn8x#Lt=Y|&t z0E8F>^)|f_Y|h^GcR)Ea+Hzz2$j}wT)gy;k|3X^-wDS!7>&3g>y|gV3?ArF|U~K&s z?ThNFiju6$#vU6sBfx$HEr4L zHq1<(X%VBKeG3JMagOZqYMG&Y&pA~-|LC;AA#U0DG-cEJ;*w^JJ|_`w`h)qFAs~m_ z7T0zCXK!@D>m6Ns0ZtoX>r+OZkNV@5HT@k#Gw$47mK_8sdB6dqm0ewV#%???0VG{n z@7M!Lnyl~p!Zl(K?)1g=nI)Fl3F{}g4srH7wcBBasN$EQfPR%3PaEu4_Zk|*zWn@C z+DZfMpP$2yzTbOf?beIJdW*5tWuJape180r6j6wuSsQq&K^~r0+;b2qRqazq)VZ}` zVr)h;a{pge7>)^kZ&XH=ktOZ4I{q@va)SSIBB} zdGXXyl|7O*r8nF3m;YVmZ9@rC3^V$xQwn`th{K$(ccI&%3xJO+UIU1(h0i-XqS%u( zMsmRllY}*N&ZScJ&`41u!eH>YlWMF<>!oEBd z-?sA@Nj*tJEOs$A@l3DVK|^t}`SL~Q{;U`v+Oz=Yngim8ixnT=ahSQM4dlQ-1oXAh zUhi0ykL4md4!pv5>VtXb91o;7_d%AJ)VTknO8$Qp<(JhEWw;#orLvC_lz!Eu=UYcV z{l{mi{{7!nj93p?u3?rUW$B_n+iq^L{8~Ln?`wACn`&sAa#4>4NNv6 z=Z(mr+@?v2VY`v3&5SQ$qKD?d3AhyP`OFibb9h7e?x9bWgb!o&oije89Y)u0FSFkL zUD)LS1o}5!=enl{McHXr0CZt6(cBfSA9)xBzXB>O>?V0lRw+HyGh+B?E{GYOxN;qc z-qMuNqxe-n4KBIWZCIUCK(>gw@W3rA&mB@cCIV`}oD@FJDGqVN+5hJ+8Ma!dL{@FG zWQab1b93*eI)rQf=g;Imn)dmhm&4`tq6 zdiP5g7Bt4cLEkXl<^&ZE9SupSc;6ZV>zFJ~AO~YqCKW}^TPS+oPfMQwCkJMKPQre@ zrL`s-S}TlKjm5s@Vku0SxyuC|foMECSkvKqY*za|OR||z?@JY|_n%AOBQ-5|9nLU# zB9Cz(${-V8&A%GOzoq3oJ{KMB>vW|T!Q(B+FNQFArdRFu&<;18=4WG?_gfzwd-Qp( zn@o4ka&xQEv#HPP`M4gSsVpRoPE+&^SKOJ)xd$lzvZupzrZ2!@HmrMou2!TxuU&GP zL`ONCesX7eXO4JHQHHA#vnElD6OL%I z^eiZk0-Okw5Ia4{6$QC*NM}gN}nA^3g?#*kAP1g_@|m^r33!BR`Zi z92zUQ+=_eY<>7F?>kaR)43sc*;@NAQoPG9S{_~)Z<6ud~T)fk{10=(QVLl3UzRcE_ zFKT83VruH*R-nbtr5B~l7D++{M{$(^GC$vz`V$x}HW+1TO~1sP97{5$M?(JXqt}ub zF_#`zkCe~2RH2WBKPxkw*{>dlWl>x-t+(UEjUFiJo{yrapfiioA3Nc0t)1Ux>7Ygp z&4D*%BH>xEScr?g5~P$1K4t(ER3d)t;6LdVA}d(t=nb^5buZ{aGR%s$p;5Mb4{<}t zmLW&|F$uzAOUTi4*cFyhwOLw=ank*S zF|Ny+4J#3djLWjQZx3<~TuhH7>*yt|GA_Ti*z6iMSoHUJYS;iUl|gHeI-S+u6?=Fh zfDQRgI{cXnJrZV~PFGlN9HH`)SN-enz%9b=6IVK+0v_g`Vi8^u2fWQ^(X?6n#l;E%C!Gye%Gj+8Z&LG&KW5l$ zy3_dPkBHl;CB$+u2Q-x%d9nP9pUT@QlR#2@U!f|P_U%zmS9)e47=E zpQ^P_rzmr!JC(fz;IY0RHb|aIA!=+PyeKC~I;SUn5CsA-m~rn2rl$@(+cdOTZL>`8 z9=&I4p0SV`z&%Lvi+OD>FZ%nZXxWaGT|Nfw$FRzrToimRPW6@x=YJFZ`mQysT(6Xp zQjX8c_o+@^nFP^HQRV4>ul3VMT{!17uA)S<&J25iMi8|;I+^*Sh1PodBkRP!asNUl zmUxl!9AqYgkZ@<_vFf!^PS%eyrw4s0uAr>7&vc!WLg!W<^aZFN2Oh>w>4PPFNRQ{p z`a9L8l7VvgbGq7Vn}u%8N*^!;iE=(yaVz8;sdDer?f>6$ABYD@dX_NZ zBcbA{ym*WddowXUSE_L zOmUb4=gY4E8&C+bgj$q_32TEC12p7=DDi*g=#>JA@wz73un8VdF2uQP54)yC0;7sWT@ z>t^4~0>DI6y@0!wGL&0f)TV-Szo0m&?oj^vJ-A8uu2v9K0oQ>E^{>l|G-lB1y)Q9e z+RB|>_aRy38HP^(LOCS`S*<)(NFSVbDCXkjp_2om#`GhQ`g8R)_|NAbiso4>M(!{& zp~-r?WUINcr|-a6JJRh}-u4(+p#{OmUtkL3s-1O^WDlINxU9wjQiOz~9&;Fv<7Q89 zW<7<7bD&kB699#{|HiYhxCfv&6`o(Xi#5AC1di_I7jJgZ5{F5HNQeO2FX3Q=r-@uKwH z->lm6x2FFSTwj%Z?GW2$l7@PbpYuPz9ujT$dCu#D)pMSwbDr~I8E|6WncS@&db(RH1|ol6Q%(q#9g9UY{q9IjIZb%Up#gIl%9Do89%M z%-uUJVf$FI9+o|p8r{5fr;e_@srpUlqGWsXn;MVcF8-d=dUwk!l;%F8&jE{{ZvIk^ zED*f55`lTlQE_VMa&{!U3u(b)uiITzCiLdjvvd9|it3yKgA~}K(mqelyw~0gPwNOlNQ^vF5IPXd1LJe4F$nmuj2-xjwU}vP5;A z6un|DicUUP=4=H5d8`96Kn!X)D24je(Dr8wIf8rxi}l}%wBNh{Rx{P=JTF?Grwg%u zDnX=_3r|C(*DVqx&3 z%r;cq2Qg-}qnqj(_Q#@!JV6kG6U`Px{-Uu7KP}&T($~Ke3&K@nVnuxq@2trR zrVD9pTK<4F#w<9$~^JH1gPEg^L#n8{uNsCgl zdh!sLZZ6U}FggiRp62A~n#5j7Ik+qdW0XlM_;f%*iPYtPhcfrUPB((%0~($`aMR|?EC>=I4JCrrczi#dvB-cR>3dY}W@ z*SRH+2-6Cd8~w8wM<&#mx8x|Jd`sx3725FdK}*yP-7b!tyMMmsMb?Ib)^h-qRD6J| z6EHXR$hynb+i+jzn)tQAMNIeQA1v-(?AU7Y!5)vpxxtGcybnLg>je>bPu44)!RP1R zZ??899R%HeHcxID($z9IQhpdw`BawC*QfD)XtX`9aH)C|%q`*aqkF(zc3h46e3Md^ zd-d~$CLh#{v3(xMK;(!GZa<}1eVGa|`4tEZ=3UQz!$^=`H%B686lXd+yLYAQsH&&v zm~QFO(4;rhRGrJMVbQhF@0Fo`UDKZ#)%g4Ns<*|j;a~o;4~h+G!+*F#pQgPK4C`{6=H5uG=;Xe?(@_Eyrkn61aT#ogG9Q{jg&z% z7JEnRAgC#ooJa$SdmzY1qKVl~)CD7v zhdB|I;ov%EER~n-#RtbJ63UbI3N3EtjE&a=<9OSfGG2X_LOa2Km;t; z;%?8^BB{mJ&J(U@X8{Qn2KEPpFZaRt4vK3*$JT!;V-G&InF8&pDhA+la*@$&N%z1f z31YvV9J@T@?)N6&_scS@bZvU-vvTG5F(d11SrQQkE0ai)sa`ja}GXk=p+)UJdwe4X~sYfVU-PlwWn zmBpt}T>@D#fn51(bl+;4BoU%A@0B zyytw|3HC&{#UrL z7v=39@xZdrr(J&+%5+46EQAwSJt&TkUd-bflCo47?ZSGT=+M<{QhU*k@71z8QE=Ap zh}&iJgeSQ@!aa%2>+fJBD7}olRB@--ZLi@{Wqq@orDCDmjypwr4(ir+$~6CLZBffR z$JOU`gxTB218uU_Zh>zY!0!M%Z)mdF^fD1*XMO`cu*ZeK6!^tV^ZKNZ8-7@-(CW$iv83>&R68*3cM5dp~p4zheGCfbN6N;B%D)_T`aZh?@7sRyX7 z#(HeZTGJ{=K4XD~1IJ_x`(*-^Svoe{PXNJ%Vtslb8)s0v2!4cM7Ns-n5}`85mdv>y z7hW!BPt!K33P8>l7ZatLIgvO=b#KMl$2iH~lSnAjKxT7j?bkR6>gHtu&2qCUT}zcv z5t3F?xljogqz2797DI{6rg2M4l_8s5nU_cInP3a%0aae^=Lt*ft+C$X3D0O*lKNjo zv1Bg{D~PY<^3J?)1$edy!ooXWEmy$vkVsV6 zK!4t$jrO?@lAqT(S(gc7yGfQCM36OSI}~qSrsFzS<6XPOZI9k);IC00(crIcg+U>v zSHEd~-dhJKp}l@Aqlko72P@HA-X^{_BEH(TUQb_(?Gr{pqLr}B`!H%{lj>pHMj)G7I)8GB2rKy)+8g9&lb}(wyM4B{ zWHcVY3i3ADm^3ecW_lz=TepY7sa>l|*8R9njU zAZ-asyR$BvW2UXWvRH3LVg4`*C1INt2{+skDqw2KEjOs&s#vZN6^7NH!@OU_A~v%~ zuhTJTz9p9DHM&fH<|ZY<(qqchm_KYv8$nmHZTx#$>jHlOB6e}9>T_hHiue@epkU+g z+Jo9>sqPuPeP5J=X|pW3u!-Qxd%JL3R<7bcNhq#A1_vz7?7vUt z9=S2IpQltCuQ9gTRLpc`{s>f!;e_beSfb0Tfq+S)^SqJr;xi{ho?OWn`oS}I?AT)o zi}+?wvU?{{8+4?8Db)>jr?^{lKpuuNxlGo0_?pK{;8G_NCfl%lnu1J}PFcE8 z%2l5vnuL2A1#lNu40xBa8n9SDbwyAN*?N2V8cN_zm}^prMvl!vXm{Lp+Y5g;U1fJf z{nNYsP2bTK+o}^b88oLt-R>-ltnsx|c@OTKN-evmnM;_M87$-{s2=X*@>D{ zp0SSc?ot{P1C(d?=m9h0qnjQZ!@yyC-f=647%WeD|M5`PptUFbg^=4E+cDj(VNM{g z4YQ8sUY~=vTeicDX|bLoyH@}#XCCF{k|g1URGn~OsI#y$c?FcNT%QEsS*xyhOMW4& zBCtE~AG;A%h%gI4wFK*<03e1&K7X0O%Eq1{P9>WcbWU@9&n{M>3q{Fe$sV@PgydMEGy|HzdoZyNR97GeBJ9@FOX> zcDSK;!;4bUY|o$v7W)*Wu02C5;Wa3ej2cp!_|^-4!#Gee>12ML1C)m=z=5JMnI-cePpayNW62;l|B>_d<>kjDc^};uAq>$-Vs?P^c zW6R!t=^ksRI02y;^7qlGemjNUTOHm>FeNZ+ zpLyKY+f6?x`@jVtZlOxC$3btgQIDhds1`EyL5K5MVYF=c?N~1Zf@FG{b{Gza1axwR zhSJE@FUhTqTDHTZkL1?oONl{ZmgX40hh&a{$rrS%^y+wUc9UB=YD2%6b*ugPd|n0! z_aOOzk>XwMzOva*b{`z9aF$LftTKV^U*Ph}Z2k%>T&jR)ceg0VznUyIQa0qLPD=d_ zDR($j2PuUur^QxcisbgN0jQPRGORb6x%e^2TPVDA=^#|g++tM z{cp%-)Lp51-{vD#c8$l_t-SEkr$xF>3F$Bt3P`|YhCLn!ouaHAavSpw)n1*_1MUtU zUL7m?O!_A7k@$TPhG>pTC)&`Q?3`kW?JG^pR1ftF!%*RdWc6U#e_OH+H!M|-)jQF& z0Ty<1TUzb%oYxU&&2|@G^8G>-0l-QZm1+4 zi#m2k>6DdNuE;-tjxw%-t#tM*jXgL~guQb3X-1=%=ni}I0kj_vqfpVEVBPE6%0Dt^ zGH|OS2o&LJJs5(tCcCHe_e`e@Y+^O`+s}rMiUJ(=Z91*0Gt!T|no*}aJy_Q0_vv)z zgX$UY_o$tY{@#6Wul#AFqlsa{VyU`z%qmD)ZuiYiyOz`K_poZ=6us{< zz(_F)0~~{O5pst}$g$Xeu{>D28~g@!p!n2KN>uIhPH7KJ0@d9!W;OXbvne``e!e?d z0GFB&6bAIG&{#GS6T5kHZ>L_&04ptjh5^jf)3cH*3MLpE31~uccI+hyx?wwah>x~ zw*!694qY!tc(|~L$!YIe<@kn<*#RvEq#~RR3?aiUZN-x39dAIhawAPWJ{UVqfor%o z%Uo-D(5WXMW#n=zUhWNr+?vt7{HF~?zj@_O`Nx-bHR;d>nPEjk$>@O*ez8x)E)g4_ z&3q5DI3=?GnD%l+u(SUSI)$~L69TF(H)bE@ZTe%&5K^0w7Q}7RXZo5IX?SAabL22i zXbwo)q{PnWeSTgm+DJ?0_1_n5_M|n~0TZV8Vc2QkK^iA+`cUooB>->mtWg#~p7dfW zmXE5j&I&GaUaG$1P1f$O@$j2fvMd(5%`P3lZQ|F8X5Qqz5G69~`HyNF_2h}|fIisL zvUW3RsX`$_TY~l{Q)PM;jL)l_$H3XT=n_?tP8AtYaC@~!o7^CQb#{TgYzR^i&F#39 z#>267A|QrMorBwrkNXLlKIXU=03A7!``rdU2n%^fFgFh$t|a5pu#O547x*FfieS?? zg}kAoVpBf5mc24i!^@Km!=)a7SGy&UrYTuWJZ;jeyR#FkyobjI z%)_wVJGD>5*cK;;{&_&Jwoi*-aSkvQQP~x-liodaJbIP+ONaCFV)-8o$D(~`-IRfYlA9#9Y4LB!H}HEJMk zGKUcZ4R?Gh?*%ynP*e&%l1;_|yLJDU&Wk3(>vvZGUb#uc;|s+PLC$nRL5GUNq?0?1 zUBJ*$hU6u58%~3)j_ zAE5ElGG=?m9?FdjO4fnko5Yr`<&3a`yphsc@e{6j=We;UY#ek-MX!%zf!Ta+J!y`j zL!&5*dHMRZdjuTY_e~F&v%>w8-&P!KLEZad8-mTK0?y2Rx`S+xC(;`!gizo8M%dj9 z+Rls)BR0)Y=I((U`go{=^p(eDW@SZySgLEPH>z@^l8Poonhs6V#2>y*$5O)e*!gLZ`%M}|cs zDzxqHrvE{vfq7ZL9PAMd)=`-?B-zXj`BbsU)^;S;&Qnn@Hk zx0AZcO!&!t1=TglfNV;-fXdSciX#uE2C>^*v35=84%$WB+@*kcOh8{v|SnFDEi|^;! zE8Q+TWa%$h0#9%-jCY5N;IwxZz@`=HfQC_~Md^vs;6vw|&kwi3;09MAoB>6_)#|wL z%umB03l(jiupB`^W%YJ{8BA0tNt)G;8y!lUL?^hKhKJ7#X(anax22v9hRPD~NFW;U z{o8tl?+pmemS~N28SYbsm!~j#w0ZbZ?Z?srbTSygxt}8xo|tPCZITykdRjN)7D(0I zaq`6O-FDjpA%EJOuQBnUwXyfX9nc;Ys}^ATQ^f1l2mCaOslKm4=%*Qo%qg+lSmG$Z zJLzk;HxP*`@Pzz0^0aA+#s^z>X+?n-TnbHGFOIf@))^Dmx&fuyRyFKhR~+Q0;eROS z0GOa4GhPTbJjIh82b`M{I^9 zBotW|NV#=KK#T2mHrW_8?T&_{Gb?N|!z{qQVC4GcG@2Y(X*_i&TV1Zbo&^C&3|al7 zxsC^H&MOhVL6jSqxK_sUA1uJ{fo3Ss;pw!J+LrCyRKPhFZn>m_&G)%ujk16s#*7k7 z#yrV+?bai8!SHb6Sg(CZ?I18mix)-Va;aLX%R}fGP3Q25m9%xD>uK>30zwCYP~6@7 zM!HjRWDduszIMv)Dy8oC>UgTmfnss@6G}f++#hKeitR~$E)F_-{3*{kLo?A1x*jdE zY-jWO4{saCVh!Jlk8$^rqkQcIsKbh-aOGz!|uBbAw3gT`D@EE-Q9`#IXkrOmAUNi3f%L1qSt zk74Ji@k)NzNWq2P76QMh?D11?Wer{f`IhDwSAOOiuAo;+QWI_{{FDe>8t6VVyY!1T zb(9n0_jyZ67J{MG;s`{XIUZEoaQMwHvzmA(ZcocXz;!RDM+X*|-iY~t&72bgUEM6x zShC&4Eck5rP9Zy^u#<2P4~NDRdm|Cnq$g>0kBB*(&7KDsNNBIs=sjye{MbuDvLF0V z$lhm<<33fSM|%X6JQ9xmlls%zBVqR&`iN=5A4I7A#L@Y0@>M;3t3H!L^OITQ98SRL zL^o?7j>u9^gaSceF`Y1B>ccE;7_vD7&kIt`efC)_e5>=?F&@0dg6;^kFqsR z0{i9cTcxXTxcMY}R9+$~$QW2?FKNV3#!csY0mF6XpDcsP7LP=#d_%MadEvupow_$upDQUoi%}5V2Tz zu9qN%cG8_eOhr%ZHx44UbJDwS2~;@JL~aoC7c8k_{1=ecOH2Aq2$FuaR>8P$#{1A< z{{K2F$p2fYBGFRX2H~qQ8u~SMyIqce*6qE1lyK(Oq!$W!$)*sw8&On0P3$V8k_;-+ z3wQCZgShw`FDJe9KsnkXqX}=i(uP?^w7op*UoF4f9+VtYtlk}lPRrqZxM4>nvJQb&l_K)jL^f<})C_7}s zE|aMr;O5}Sa)jIo<>eK7$ez)033d}C(*)5{H{6L9U}A~A8ErfNp?2>sQZOpeqFqbJ z6J87j@ibC{z#2auN;sJ^7zLZ`$ILwZ(%;7jJrItFj)2A1-i4Ec7jMMqdHU&EBfE5g zKGiZrTgTfe0;bF|w&lCuxEgT+WcsC82Aoic6T5-s=Z{mEr5<0KOBz?OAP24(9VrPn zO@6`M7w|2=@2A+O>Dq5W9qikO>K-hlSd|Rhdgx;hn>#Z;NJ7b-65i`_%jsy$eXj-rySbZ%J`3O7qBg!o`0u|^ZlZ_s zvdpmj(~kF{r}Bk7dED>3^eg#SxEQpKpC#s0(FQQT*VrBi(h*1%d9>fSSWPI8*0ysH z@*QZb){%n+fC-D}i{bho5)f-|cuU~qiUpdp&Kbzk(7C_4O$WrM7lDE1K`1s{?wJ$+>ICvjy zops|t6h17Crhq7_kNwifNV6auyv2EecDUVxBBQbD+n?u&RvGuTnw`OQYRvs_N{Zz! z=ekuT8Kib=t@VnC2ggLsQII-{W6T=g*~ei9ECD4y8pr8}VPTnXaX~)X=;rDG;=trP8A5(ct+>mLb@$ z`n%WzF)2ieNq4VtHO1@_;0SV%csy>iw(G!+JM~Y~_1RwR!qufVtBu-*hCt_Scjxrj zRUqHwWIY8Q?xYQuTY)ee1*k05yB_aTl~R9QD+8%N_q_Spi7L@YAN9CXg9SKf zX<$iK3wu)saF-kZ!_HW@`xl~LN0p9M{qsxn5XTg#Nr$xH(WZMteU)piM#(i!p8*xvjb;Ds)k7!h8ucXE#}6T$G#MkX6Ui9 z5p#ebJk!a0$SFh}%mVXrCz_kYxAn6<*QLgOXo?~z@5)y1( zo{E5JkH^q7t(S1(!ksWk_4az&9FW%=iPBz53FZST2JcfZow|<`3tmf)Nt*OXt(|QE zaEfgU7yx@rar%>5%BNdMr6 zxco~#P`RDRY!BFs3O=|uEE&6|C%1%~?($hWTpyG2fML>@ z5M~z-|KYBLD@qfveOf~VL+T3R7af)Q!;OX>U9(@Vn6~w>E@T+0?%6Hj7R1D3<0l-` zC`coGIKeLFI15~}m{h^p90Tqf`QMQD z%qwP5$u%^P>EAp6IfRX3xZ|nIv{w?iMO2G1@Y%QE!(>w8)az*vIthe{z*^6j=PjnT3>y#e{F*elL;Qb_Z!lJ537 z>V9`~^}EsG%BYShcH>&@5bw*UvB!;G2=rs(XlGbE(K^wLACt^jf;7B`E@TyZLmJ)R zLOQQm>?Js>b9gv=xP##g!$iyJx*L@jgZdt+ABnJQz#D_Z;Sb(~EJBXJa!~vG7#HGo zuGR8azz^D~D~@HVb3$7z)?aU!1c;#e!f?5TfL(w2k&ye#-Jv>a7;GVC_~wCvFYiwj zH|xsOJs@oSZbn!9jL(T!QC(o(p9v|m!>r2#A4W6HbWf{hdwd=8y?12WRrXJ`CqK_< z{Co+tLMt&SATQ7#QPl+x%-epM3hjJnM3%tuA*v<2=jo`C?meE6`W#+{E1)1f4ToF~ z+KTwcu$7xS6vbqxL=qU*al)=ldeM*`v%-hybCaGHVPeomkld1`c%=)6A>N1N`15Uh zsgTFy=smOb2yzQy<+;om6te8cfd@u}Ux;q)pcPwx_rDIEmP(b_IZrHTfSN=22QBA1 z$^jjCf@|8y|4kT&{VuI9uJR9C5SjsV+awp_7V>2G_I7o4Oj7MZ#^7QFq)!0-Yvxo; z1a|l{6HFm==%`{J6SDMu(icm0E|C0d`gsCtWK?c@dh3hABsi)Ga`>G6q7dbBff|J4 zC--Nu|Ak%vEecj8-<0=4XyW1P1OneE5U6j!i+UJRFW$%nQ8wD5<+2Du%^sR9wJ8E3 zZFmW;plUgWb(KW!mA7lRv1|w=Ir(A z7!QvNRMxu~`Zz^z_kx}`2#R!}(cBlMzOS7{@JUZHnE+I_#swl3K=f1JdXBgy%ZqHp zHOmi96telhyPl?Tk)lVsAck+(>e%>WaEhlq-yvsGkKJVV&hyT67!u9H31XY`U`d zpTxI$IsAltFPvZR{L(6$iakp(&zl5Y<#mg(+^B(hhm**uRx-v4Cie7^pESs?MDT^s z$9kJ_IwtCjEE9j*ZeT+JIH`}XWQ2riaB=m!YGj;)D2vP%nLF(u+$by2#nLE15pLuq zLC&B`Dl2Q;BFy+&IA;9iGDElBI`OxDYLKxhvXDD2u{rvqc#R`xa0?3;biWg`3rgzS zpj~%PG!W7zX}Vb?mk3iLwnbSQ>HhL#djyQq{-Q%=@>`~ne^LihsLC|Nqd!#uFS;eg zTS42f2$qr2!)hkJxq>yE@kN*!o?a$nO1LOE8T1Z~(U6R@O+fO~o<6YAe#U%~;*=h; z8*;HXUTa(={hq?Y3~p7LW!VHy{q8Wok5k+`&}Uz=lPz%i!m``^YW$-INTxaQrw#0o zw>*uahpE*QW))@rCSdg05F;kNB4BiC6p=*)J_?2y=tMz|t#>h`Yb`IPBDuJL9QPkg zBM%!+a8E8~73&&h+URxsR@Q6xAp$laZ1^_nw*fl?Vg+zTq7rot%OPUKH~)m^eUFtu zANPRs;tVSy%rVP0SNM-7C}6y>2K9=c?fg9PwI+Eru6$MebC`QWi}5=$WFiq4m64)q zb%$^&gKo1NnwOhA-MWb6I?zM~vP@cx_N-+DGA*xfQz=a-!KI7=L=@?X7ln(7uUGD% z|`wrZ$8Mq~*qiOcG<5SpUl1&+9wkovD57c4GO5#M6H8 zI(uA*m{e|dX|5R)Qbj1%-|u>QBRJ2j77#FH}^I><yhq zY-&W0CSK4r>oVwql~N7w(D!Z1>th&nZQU1s4%!Qa{KZcpUe0?iS|1U|KyU}zSwNs$=Fzl*I{EsOfJ*WHukJ^y`W%5?( z)#&pfA1Rkw1@1c@;22iy_5TukvbDJ@t8!Rc&s<|SWc{VDHG%qbxbxOY+WUQ`m6(3< zo4;}cDGp?GYE=5Q^aEiDq&a#t(!&U7_d~fJ`oJ#cqKRj>*r}hoE^~V97z@8b3T>YA zO|8sYpVazU@0)OV{F`*2QI>z;+)_CSw=uRf5DM-1-}rj-uq5;T?f*hl)D*2~1}1O` zC!wZtp#%lR{N|S8+H#v?mX+d^p_mJ$qL$t{gs9kJh%|!aG`7sxsFj6TqLpCaZHr49 zDw(CVP25Z1 zb8HypW?0w7LWiuypc=TTC^*uQP6fYzu!THZEndKZJGt%bC_<{DlupaLn;q~XzSku(;_%wAur zmIx3IfuEMc+hm52%hFY}I=a7%zT(A{+>G4wVpfLe?S%DezZ~ROUXX3}M}WVMUe70} zkOZJ44-|d9OSun^hM#EP?$KEr%C(Ult3mGQd*3STVxy z3nF?(1_EN5C8G{$8SWeArb(uZaIKR$tAkqIMVdT5gPByUAVIN|b{g@Kz;b=5^`%h; z2HynXyr|mXMzTC*PI@%gt9`Ts@vbS`YPXxCG?MaX@eL94F9Y$WW5^gzQCC?qTBO@4mq zT>IBMC$Hlo^BVOieMMUj7(aH$%lO9WV}7p=lF`aq#E-#kNE1}PyZwZ_FaCuLI}VJT zPC=9()fYwsOrL&dn8dGxhUpo!Om=r!FlAchnzFu}Y%9o+Uwx9x7O?3G_+2s;7*xm9 zk=uRt|;Rt;-v8SLc*pgr8qXD%Nf_dKCNN&%=yV zYqs}lN&umiu=VUe!^{X01UP3sLNx%f&)fp@TOs{NI~hq653Mi#k4gcqHi>chU$M+j z?awmsyrEs}$k!&1G>te^wj1;3bC{ylz8ih|BHrdjMs^)YJS~Y$*N^?ZU+ zjU+x47!#9r`Vg$lTy~(VVdaF!z_(vW1h?>7)DHnrWI&`B-OMS{4fAj4Y{|^Og5q|K zuenvD%zJ^K{hhDdSK&(m_TU{3v}E>9MGRUKb3&iD?+3_%cY=g`<4Tc085rvGq6-i9 z2N)B2coWUWyjn_fvb zT*WyQ{!ds$KCrDdcsdPz1Xq;pA<O)KfcOn){Mf0P)5mr@ZX@MjQaVcLnR4;DflE6v#{a7bmgPYjTUB)u(`AuKi5^x z7M{OgV{EFKYj(PBU>7|B+`kRdr7rFo`{92oImMt(v_7T7qUG3Y)^6_VT&L0!^4rq5 zC+r7g`RX2wVn8n=tWc$aI80K|-ymRsQG|RF(m2V~9d+L##`OzRU!F2b#liIcwxu{vn$i-_Vjk^KyA95Us z5hz!T`FXXD7svxU6*ZV}EskljNdOxa*KvO_Uw!82KnkgKQlOM`G6wycL3-^U8Mej> z0JCxWbEmNdfO@nf4g${j>>8>-E}8V^ORk`m94Pj>DcZAQ{hLq0`y=-les`=%8SaZ!x&|tW@U~70U+Mn9fO|$ zGv`TbcaFC06C8e>iOzHjfoG;KGt--Ob5cA4jXKr_q+pXz65bDP)?Kk4La0*r&EcW< z2f=vv=e`_^|FP16oHE9U`q<%GpH(;DmK<-eL*)jfzY57nRdA~mF^pqMpqKC)!@Ag79!*lwDsc(x0N9o4Nh-l%13ReWbZXg3F z{kVh4P5dGl@+d?th!mG84E_*3fQ1@|NHfMh9LrnX$ZHa~3tc3{fs<=!PPq);tj} z9T7E#I!q1eVoTE6&}s25bYm7uZR(n`G^x7ScT(?#L}(YQd>1fuBG%=umuV zY4SvWU0tz!4hq=IKW%xT`0opoMguW=?8)$q=Jv!23&N>2`)T@@Jg`{dvtpvEI@~wCd$`uqd@xGj@X4X`&(_e z8u_piScw~B(D1qH`i;39q-0JCK{J*~A<>Rsu$SD~on22@FAZm}Zf5;7@?@e7VC3%? ztOppeBgOaU)_~!s0p7<#0?+`%4i+-v0UegCkx+2!`BIRC=?CU2!|Qoo zA?Lx&+qXtNyHy+iKAaH_NZ&=#x08CMu6ya-M-|EPttx)@5AP>YJAYZH+f-ljix3-X z=hVCd_pW1&0o*qL2$e>C{&z}AAqEq&zM#h-I^ZwuAXax8XhO>apr}qF;#+j!I@5D{ zb;0}WvaoA+g55-(;Y&5j6w--Pv-9AQlXqcIIV-QNnqm9C`Dt)Z4x>=FPANL+1X_y)RQi?x1Jm+oBqD5R7=V~ z9nEkgOX1aiYSKMxrEQ9FxzyaCnu|p`wJ+7c4g%@e=K~+M7@nUxjZS*aUFk!2--jL% znol{Wj}$xmW2fpG4W}9);3@n$33%%FM>2nwuVs+&<-23p(<{M^E7@c~Od&yz3X0m} z>#X#s`%OQ5cR3U@n4^xaJG6n9r{%|64ZsXVGd-kBeR(Q&VPLNRhUCmqF67nR91AKS zZj#Z77Pjqr1cr=4HtO@Q+@G`sW=}&0bC_MuJBkD*Fga?;3EMLv0Eo>N(KtmN2uPcl zqWmEdwi9Nkvlk7GlsD)$yd2Wen110o`B+Y;oy`RAn&kx?^E8Xz%XkxAw%(`HgytmD zI;o-9<6+>-G2>c6e##JLs<{;$fL)PD#aR41uRsyh&M}xLjjmm~(SkS6MFJv*O82Xq=M7d(gR{JaMGuZW?;={@2)jW=PzWFeFJL8=?V(U`TXV+QF@7Bwr;A=!DK zL2}?;MHLS@yA=D`%RMt;+xsj!1V7gWod(e)c+p#_?KWM#_H&J9%YI=T+;c059}K0u zXPU>>ESzQNKG*!w;4nNn#yGL8Gk}eDK7a4nlo@Z;cfW6$c{iBlwy0GbE>+#AbsWkz z85^UU@(beMpVT_#I#-px*y^wY0h)juYhJpsf;mupw8&B=jO8QrL80fGfC*j!K*=!l zYe$#Dz+)IF1w{Z%j{ZQjr(%|Q3tth(*F$a`r2|MGPLW2{9+sYMq~_vnn$-Z z2Wwr)ELQ82KCsMu-?)KN&5dUFmu1Y2GG6Zicmq0RiO8*zXO-D5mKSMjsVNdfR?0-* zXw`#g;fJ{HizeAW5NIwPq{>=bth@H|$LbqdbtCUVOGsAAxlZ$W84e^6K=Uf%MuO9H zCnbg=17>=~b{dhsa3|~pZ8YgC+s{Z3Xw`+#+6=YztXT|^nj*#9Ck22|-^ir4TVQ5+ zOU??m{hHDnM+UJuOC@-QNU~OXMk1h(;9}`aovY$mdq&0(hnD-l)=kSRFFdGqG%r;q z^;d{{P6Dm?Rvu{i8Vg?WwS8a*qejT$;(pzQEuahR5 z#-8iC*aC94m;tSD)%Qhj_3JOfv8%jKP)G6 z_&%5pI@yVmZ3D@^-4$q7FG%x}pt;PBM(I#eE@2kaHk3&`t*2~2PSg=&=uZaX!0C6w zD8kPKGg6*Tknm>O<;2&ZfrH8U;J{P>d$Zk-K7PQ!eSvnXA+``@PPC6hz){0b22;VQ zZ}Tvfo7}ag?dhRRBJ9i3%JXwq{!E6m5P_YJ@7QY8aI?zC1o0p;8lr)e$<5eeUdpW0 z$q(NEPUv2;O@Vz485&3e7=&4{Wq_`;?>1gd^!OzMa+oN;*KXiFHm`c9BS%;Ge%f>0OZTUW^eq0s@z0hhz9v7?TKsY|;)psl=$`$NxVSY6j6h z{v0jJ`RPE^O#RmO#TWY%OzL=8OF``PwH&{TlAc_*7j~@6n-A~z+4@442*;KW-~O@Z zY5<~J^vQUain(JwVe}s*Zsp9rnx=Qq<8SHFNJ8~32~-|H1dV?O1=_RK<uP!Ckn{=iO_DSRC=}GxnQW&nZ1g5`X7DQzwakcRpp86)+i_|UUcVE+Fn-yRS@WCcRc26 zNi}93yQquWp3dKk?{;6#X)xciKN7MsfFF&PH@U98HT|vF_#W4>f{B`jB8JKh9k2FQ zZkYeNAu##DhWSY!m2YpGnnN!z%VNxOeUHL!M|>O$+=VYe`fARinH!V`ITbx6E@iNjVRf z-$Mtmal-?eI39GbJbZL|?urkDiLQebfFdLX`2CsOhS$u-FL!ShLNU!s?`GNP%M^(< z9ZHcQhM`y)X>lq|cf(902}{D}uIAY_@P>pt$Zb7lC8^F3O~jB@S5KD5E<+ zhrz^5R?_Ei=zSCMJ-IqZALpT8R${YQX%-7#TBY?SJfUFyQAzL>mk4164M|U6!A-&8Ib&~h|a|mp< zs%C_1HF$OaNu-bSyNnQ|sVeFmyjx(}X@v;pZ16 zy%kkA0-xuo=u7Vqxp(*6h=kG&{|%WHUGC0l{!#@%>W!Ger1B8fo0j>ju%$Io8}w=>nI@WBsQhnR5m4V@K!e<1@Pk=NH&x*N{4(>TJA@+vtVrNpGcHJ5H0mMfz zG4oJfJ}-3^h8(dZa-}`?E_M^ ze^zu9YlOMZNpH%H*X|iv%gA)`UJv_`0fb6PQpL84q1iUnwt

i^%`LL{?bhM*E-+{GKO*KaTOsmr&!T6Whw`z*WsUBv3HoOiSx2glma+8xlM<(qOY&cb&3WnObOB!z+W zoKyygTd-=)Z38Yu?8*gL1NeVlff?a9u36`EpA+_r)(-;ITV+>Z%fB_w(h{l~o_AUA zwRVe09}GUF{W}*dKINcOEUEwRpSI1+`w)Q=IVUF!l#Xfmzs`F}V)y3mG1NEN*^7O{ zgYh(IX;fl9#9w05AKS(ux^vH5;5Mk&3-Zp`v-VwQ?%d{Q=IBydPA55d3&%9&bOkjH z8~#bpG@~)A@Cck51wFHJ_FD`)@&3`5gQ9!$ptPBQHYx<4;R0v&^s;@@ZAAw~(25rQ zl!Q0Yj#iA*&5s@$;OS%=9ik&<--;EhMVq1On`873W)-p@g`U!1wirMH#K{ZR?Re~t)9^Ew$Mda zr3SFO_=F;1n)DDEZ`y_%Rn`nPsM^MZ(Ji3sgJ+buQJ}f3;S}zl%zUol#Zu+= zA?uu3!7IcISBgRm;z;wmemqNTb&oi7>d{1W69EP};8$>&Qm$|B`}7(t*Tg9OctBi; z%U?HSwk9u%7UFV)d~^NqF&MI(1uY1YnJy(&L#;v3R>(dX=$vh@v%Fv2cDEhJ02l#j zo1)<`E&Z{bg~~+L?=>Fy+SG)}5X zXQ<&a2X7GjvELGt&LA?fHg`^tb_b`1-@9ixG;toa>d!J)iEH5k${i051?7H&)#{)TR* zWxbD$AJP}@YTJ0O=}QQ2+m$xJHYyl8)$$-AhO!seu))zXAw*ep_niFpRxIxBMdX8) z;8iI@nc$h43ojN#9L^*zwE7Htf>6L^KFl11P&`>*KRd+9NRIgP3rfaC8TG~+tYyH8 zWX9lpCZQeU-k|vU1BCf=?FeQd^Yzb)Yo}VSj(cM#T2gIV0#vN06Q228gLL)-+v3-i z6&)Dc`Jr(1^eg)m<(dudPT^uMpUJ%(>zvcTMKY0n^7CV2f_d8cHV|V>40T|LLLBj% zUmm2OHm41Et_5}Le>s`oTlu=^@&;J6zZzw-yRUfWZ!aaI>)!aO@01$Ba%WmBE!z!s zGx)xx!?fh!(7-um8EyxWb@wcF%40IpIpfEIjgPn9bl$mIi@JzBvVC9jBAY)u;9~+` zu9^&5yI0`<-r0f>eX-q8K#{`ti+`u+3!@J29c)x`e6fFi(8CcVYZC>Hl_5EJvz0|w z=t&bWwF58a2d2%&Ln~Eo8W;HVtaY6t>Ijufx1P-_h)KgDXA7+NC8xK#>{N|3zXN)> zPv7t9?>L1%@OWMQfq*|C2g9WX=}+5kJo8cph9K;LuR)xY+-Sp8N~ zNGFGB**m|nApcK7oxvx5V^@ey;xK#ttk=L9>X&Y0e&Zc>=jpMZ4GtBTH7jf5|0b@e z%ZL#!V!IRT(I>a2G{1bM%BbCDA3Fh&zupd8blUVs-Y_PLOXxi*YK@o5I*!l}7PW%R z8&XON0|eMmA7T>Vp`eZwLf-xX|9J^u$*1kSaN;v4NSgcOPIsrLZs9@i?*)%ssuT~W z+~%qrI>V5CZW{Br+ei0)RAAs_DL!YmdUb2MJWh)vps&!gd^~?}pK_mO{+^c9%5?ON zeNpaheyEsdkS>Q{X0XNb1LXZpDyG*@Rr=9!IuiY`#1Br};KXACpXy!HHZTHH`}^Wy zPtUw#7YEcT2An+Vp5b~|YKpT@A7Xb43T~l4cnB-}M?1sW&DgM0jd&8w?){sz3GF^X zGM$V@FVi-h|=u5WzlVSQ&u3Oj1 z9D9JhqUIC>*5kl6EHbO^qFxk-)qJn}YUI~@NOVGCNWGp;8gER@Ew@wwKVMDn(!HtR zzz~C2n2&pDvwW$Fsi1MRej7_GpWY?__+<1lFg2mkrfpR1-dP)mfL(1X9y|ZYN7&gM>ry98)0g7>CBo1 z;4L-T?U{b6spg!R{dYy>Mc#D}Aust+Via)EU;e8hkx_NqzTiA2D()%@Ytgdtg+{N>eQK0s_@BKdYMIerl@~K71L*5?e}jr=SLn>bh1`?jI@=R5 z$V}qhd`G|c95d+1`IlFmJHc2 zy8h^O?ci6`z*cAfF$6s68lQ=(8?u0o)1I!B*{eGb>C@c7pdXq&3O@D_>^*PDh%r!$ z^?l))&j>P@eyiQL>uk!_4G3IepMe<=lgJ(o>YG6VBcRDiFx*PfN)Xu_cr@Rxl$m+M zW>P^0lU)=|Bft%3da8-9T1{v*n1U#fk^9O2TObar7Yn6)M#f`T4<%@%fQvJ#=TuB3 z@Lje`3}rZ`r`vuN*NPWf?miFj3jagJ0PL_O&Asc|{2k-oz1ga{O&aYlKE>q4d&} z!y42F7)qQm(?CZp+%=%X%OJ!%Vm8bR?;yig20%Z*QX3RS8`A}dZ9*2)+4^`&lFIst%$Hjwam3xu})rZ4J;H68Xx5++x)4u+2l;N@eo97)z z{BnP?ga4o<5Fh~i)ZjOa-CsZ-EDM_j4OAAJ9>eCz_(>3qC- zDsYa?O6sG*uhjm}A#yE&MfqV++LI-q`lkh9NxO+R}JO461a)@5WQ` z)?&ash!Q~zw~(HCSi|fGBI22J@;!L!?D0vRME9tG{Sc6sRK5wUc3Q;JZ>8q1psVd8 zT#Kd^r(Pw5Y?mN#3n(Iy>-VLv#xaS@o=Gb9iv^_#A{gD=$Ak*3f2Lt8K;}k}dyBsh za8*|AB)<;EHzBTB+$(YF01N(P#>sF$>(i^^(Y1n|Rx~HykFL)spQQHVW9@p`I^5&C z;V`d(+(${>#YNa#&8W!%EN`Z}9MlC(!gA>R0QQ#!&;G4A5B8V&#qz^3&mI;Mq~L@I zFj^&r{~M|`ZwK_tyPY$PM{&cC6Ak1eKW*cg@Af%6y{}RK<1+19Fbjd_eVk8NMD1+N z=;70)hb=H6->>yUG` zWb*D~4+%S?ChH3KNAIw6cv1fHmWu+jOnfA=-YuAoT8j!qp$PY-(A7aS>-^y7z=52@7fk$rtpP~$03V*?3lR8sn z`Vzomi^o7iaRAidV5k#?gL*XFhV6|G8+)kX0bH_SlAm#z9mjm6;;)Jpc)1zuZenov zxrS)4$oVA>=es{3k)bUyRm*nfTCU9?UB2XniO#%wbERHC^i|l{_U#7iZip>G4AKM7 zv;zw}BJe=I6Cd*Q2#JqT=itifwV(RsJ2 zx0)C7xA8vuy=pmJ;PpD+4mSMf6CY!8b=lgp{f&#^^D)>p@7%-fhf4NcdG{R`p8H4I zX5ZVS%BJ&SNcV6RWntR0uX=S!R-(>xlkMu{fkCCksMk@?P;MrX0V7i4m`!4 zpOwI=di-@x7?IMPCE|XxE8F^`|~En%jo&Xm6EGehx8AZ$$)Qm`hwXXwugbM+fz-SN^%N9jQz!FJYcy2~qAcNNBl{kWmpwbr<6*xfHxUS>g zTg9fE(=6BLfm7Cj_A#%4G>LImN?5sfH`ff*4|E#*Pvel8OePIS6}6Tzd>L$d?1Q36 ziPe^8ZV|s5+1WV16;-&OsOE-qC7MlJ+jVFO*5)DCM(ma<^QqRemWWT#NTDtEO+f&o zuf|`M1vu-kFPfNslI@qR*SKiJ#)t&SJyo89d3&iic%ar1N>`&Q77L?a&P%B_UvERC z+4pW5xB|FFS-#OxQ> zae?if+`5)G_t^FLGmlfrC&TBDLB>>9nJWc<{@Vw+gBg_ecN=!tr2TmJQ+4Yb0w%Fq zbeHeAugMgW=&<#uw;6%o4F@MAl$IIRXLSu3igJ~Z78X%a#dhr@Iob3dzgzfAGQy&B zL|}cXPs6pgw3u-vjs}p!pQhky8rwXxC$uWdwASZzB$@DXQ)h@*OXft^z_nc1QBHW& z(lf79ET{6rX8*X^`yjMgZot_2-Pp@;o2p}PUgwCD_g$5kQYaK$Qk!f8>*e9L$BvLc zBAp_4w8a9sx)pMK-!fHor0>l=2V=n@7l0VQvKL!gcO=Jh?Tn4O9mj-IXkH67j$XSq zyR9M`LqB)$f&BsPf_pxY(`*yDzajWZ@H&Cc_P9QgsmKKtM(!aH`M~tRchZAvl>K7? zNhPjq0TT_G7!HAkwJlUi@Avj0inKn(*ScdA=&yrosTm8ww9CET8;UzSy-jueS|s%eDh~k{AS`g#SHT3KjxVR9L4=c7WkO z@=9DH=@Qt^Q}XJ}g!5wA(Sr~#M#P!feF2fkGWoD0A#}qng8zZ1F&~#|kPlIbZHDDc z%!+ypqMFF)bbbbF*JDEvv_D!`hy3J5Hol6yee1d%5moi#_}m|yjuORl{~dNKc3^%3 z^*!YQ1Q^?k6iPSs0tzhA0u|yo9{C!vl;wH+{GTUxUqADBhFx_2dL?Dr zE$qTT9M0YB7Z%d9?C9Q01If%svDB|ACu~>q37eW}Ny_e0LWuIqQzedw)^mLw_(WtU zSq1P+rF{P}v5>o4IrbQ>f>D16dc<|<7yO+JN>ib$0Wd9vR^ui*1foJ{qekGd%>)P; zC{%^?4cIkjpA@LZi}N&GOFHmezH}fTKNW_}E9vS_vw2c{R-rIlEH_Hto@}ZP4eHNW zxXTHBV3F&kyIDQaY!9GL);pm$AYl2+d0Sk61tG4POat!6Z3>eKrg_pq*8^7_@YgeD zs|(s4WaMqAWEXLr!gc5qf|?XW{WoN#)`+4Od3mFOst;Jc1w>4|)U4A8{_bmd8nSj| z?6LI$hpX_HA#kss-=~?c;?%-!`aW|9BV%prY6Q0UiDu(0Or4dgf$hrQ^V2<$ye77N z(V;zCf_5)NOi+ew&Wo=n%}+TgpVC*1&g3X%TZ^y_AQ_*Wbe^o7XPSC$v#DqTR_V*2 zpmGiFY8!#^6_nHMd3vaEIQy>8Tkiu!!4sYEoh+Zp;6YhG>XxPG-_Tz{F+MD%WZaLv z{7Lzx&Lrt*ev3`^`Aec}Mr;wkOICzu$9xH~S=&efbxf|C&OZf{?efLAk>gFHjGwY% z8_zmu%9FE>b<8)u7&WM6#pEG_vwpjD-t6ArCY~+9*y8iRQ~1YD*oAjYsJVfU zp1Lno(J4n%Zvu+bTRZ!R*u&YsPIRtweuXe9dg30Di1atQV@V z2TERT>I~l-v&~h%6`!Z>*Q3bxAYI2T`kc(}zFc`DvXm=5a7Sx4WZe{R;L@&3i)8pI zJGy0NfDX4E)(TU-r#pFpoLRnNa%VjhGhbcIZ>)xUOy|1vU%c8kpHj0u*BG`mn$#IH zD>clpMK;lxa)%c(GCwI#Q{Ia#kb@El9A{SQa3eA)>JyE|l%5y>ZVh9RzKQau4@K%( zq~9qLXIDV3=nqVjX1Qy>j@)54_PG-x16wImS?f9l9&ykA-vKF`uFddY$WMeGB9{N;cGMf;OCY>C)_=t-9 zP2A)j?x&lQ^MxRm_$5?LmN}5X+7XnLxL0R`$}@MZ%Bic98#*SQ?Y8fv&`vIfw=HnM zXe+WUY89c(v(FH)u*uP22l?l)o94RuX(^5(KeSl+grC$kGtN*zT;MZ(Bh*!DDtbg; zBhL7AW-Qe>{(95J1HC@V7t`{UMbTy3V3IEY zD^sh=pwWl{`WG_So4(e2 zHP{$&qYhs%`b$k3Qb7yNBbRKZa@RJI5Bs4LL1`Khbdj(q)pLfpJ_4LPc*lCy%J&GU zl291Kxi5RdV9jku50+or;Imd<>u21{b6UuZ_d{%RQXGv5vvTye5EJEaswW^_zk6gn zfULfHV8rrrfy$Q?{fwTROf0YvU5n+<*-~Afylk8FGm0K6c<pbkC^gFQBx3oq zy1OC+K-_b3jbVw72&}S)D{o4h@oR(o=j28MkbIJ(u`*veO&G8uhNtcs3QH{yJmdg6 z;y8FC9HS)<@DZRo06nS+n^`?v5O75Ic*KGt6%d@JWckEG^N;zyE^@8w+ao(VmEN-Lj1_^Ye#iH zWBuXdA)K-)ZbqvaUFO4EI%Pe1W4j(l|I25KrRM5xOZNIrKy=*xpY%LZ|Nblr`jX+_-#(VkaXfV1n{RA%uzS8hJuwXB6y7sB;3r3W5ZU`-g%yWus&We2RGcHWJiusuYf8-SaFv@ z=8pA~7bf0{(%}+&&8qDs=onX0D~xM6z|$uCDR7H#b6>T$<4F}LT^4(kUXX~<(=rs2 zHMA9RX@McMUqMn9Iry(iirygSY{M|CLsU5^+2*(QGQcKS*u}PD7C~bix6)-gIKhzY zw2(oKyKQ@PeBmW4{jr%cVImH9GJH=OT>p+~f!sqq^?|Gm2wTG5wYl)+Hpbql859Kj z)L{$Z~t~XT)s>GNo_62I+;L7l95T2WUz! zZUk=UMEm0isuq!G<6iTgY+o>@BqrdZ{0^tX*9TDv?io?AR(8C;?dpKjWK2%O5`WmygY?oXJYLZt^X+ zU|-Ses;8Tcy_k#$)>fUUX&cacR&4*=JTM0)k!~1`phyf@ERL3|n-JfxgXmbS9&{Et zN)HEn%y!$9ZPas0+(I&xPl0LBDhAFcGJMG6e#VYY%5Gs1=vYJdIs1|UTGi1*I1P(B z86LoQ1M4xC*T4zux*UH)-EB``;_R=Btx+aGXO+>!hIVSiGq_Wmbh>L*hQt^}?r_R? z>w0$2Jp3O3*|RcLIwI|qz6kydXglmXfz0trYweb z5HO&RJ$AA%)hZ@tQ5)f$2hY)*6FZCZ>vu!;bM@Y2|R5a1%J~-`0}dx-N`OGs%QG^ zZD*wO9-j4VCkzA62L;RD%bx^W*JYdN>T3E))H*Fl#899n`1p z!=t5klj{Cf8W6sdC^Vwr8FQmC;chmZ$G{`qMrj+?IHgu$oB z(eE=P(9H>FU?0ieJlnnTy#8+pHP~f71Z;rto3^}9GK&|sI8%{dinQ)p6vcPRKn-MrIYMaBj*G(O6U2V3ThT+kVDfAvLL>akFjLLY)fZ-88 zHjQTetjL8_-r?JRzjcto)`<5*=5XFj;qFx%`-vxsUAKPnmFuEvQw=ZS$Mf4pwYQVm z1phwYBkO+K9t&Pf$@Xuf_kX>CCsQzw>p~26_j46H*TrF?x;Im+F5sA$ryemw{Ns@J2Rt5bSRa@tNlF!$}SNlZ7zDSWz74Ua8(! z>g%4Qa@h|0&VrwD7DN7EFoK5}KQ-8|=4sg~fVw>%;`XKM63!zbfsa@(Lc7{Hj&OhQ zqq%~e5uyynPby5u&0}K=;}sy(17FPwaqL?rpzWtKP^r;=V|-#Tl%~50ipp&I@r9Of zyBV;NVi&&*S18wd>1wK-SM>-79|SO?N_z?~7_yu+Ri${S258#HMoYWG9UCaXrw6)1 zhev0?zKv&o0O*3@=9 zOWtHw#rp%aG8j_QtQ@frjp;M-;GXA}S|_i8!H-3eETaP7Z^Zy^27aZojVY8>tG@5f z;Wo5@q~8U@n!!5&UUpBA_0^GbD z&(c)Sli+Z}Zll)MLWaf=>Yyv=m3u&Vtu*{77Os5G%9@YTtZ_($y-#M-O_zH&9*U!C zp@kRutuEb_{O^ZpU$onc>rT@wIonaOtc86Ob6_aAgWPw6ZIKXs5#_%4A+>cA0cxF4 zo|%)zP74snBex@2#z@M@nQknmP2B*vexsZ{`9=3`>Ob>Y3`T1LBvB`rpr$^6}A{$mc^ zgthm%?Lo2(Kp%<09J+6vuAJG*N(f^)<%qfh<*7PgAtz%-`*W}5G2kQP-UeR4DHRnf zksfXXeY!|5v>XH;Oq{$>Qy#*CZ=4ljSaD&%e|*#5?g!uAp3|~_;LJ;M85v`AYH?xb zK|b+hp?ioMKOuh$ROqAf0x;l{ogh(e5;fCbv7k9*85uPg90{4WQpUY(HRjkj8bL?a z%f2@zC!4g(6s;5ql1u_e=9{tBb(#!t2v``&>#jM&wXEF?w(V)8JZm90${cAKnU>UD z%0s$qbF765CguavNt+iI2POHr=}{y*F0z4iHyy`hk3Zfr#V77H#ODNn3q!HMY;M{C zITB)LQB(I{$-0?6g)BT`)Z0;S zP`bjLy2>dSj*uq>*(du=GJ4%Uc7@paefqi~!mu7QHs&+olG80rYE$lyG`0;dZT8$s zfCvjG%U6C>ee2s$u}pAaLZT0C{(eqYWNXgPki&D42vJ;(=X^s$}AiyW<@ z)dU*)f&SnFU#Ljbc-f}T$&cYj4uGP%*=SS6m%3|tqMZ4p9R&v}{_!WdCpeLP29Xnt z#@KA>li~5a6x0Xzm30~BKJIC<72H>$TwhMoU{PRrQTE-g&;u)Z6AlT}nVj%T=>uCL zot|RN55~3qg-EoTP_2 z`s0{a_gUSE8Y|gBel-$|g}g`7suSK02$$ zF^$uk-vzDI4Yz-*e%~;6^kysbk5s+k$`8b$*d#32o}yWd4|(Gn!RaF){RHrOhYQeS z<#ZGDrpoX_FELEz@)+6@DyYh9C1Bj|ATuN^iSAt2Y3$$N|8#rNP8)APx$&*Lvc|QzqzHs>I)epb8{qyX~&>d@vzl}13 zP>>skn|(vPf)mz%-i-ek)ww?36}yM@xMLbLYxc&!!Ps9tpq6vWk_+N@0o!E25o0!A zC`;ed>_DA~EPJ;(bxD4HQ@03rs+o8AFU4q(`_5R}mqL`l)o7BNq#gC!&Rtu18Q*Iu5;+m&jzc?}0unLgt*o%)2Z=P1V3xBv|dTCbJDCsgGNIN!tdmRSf4xHZ@><2YJ zxa@!kHFn9RKCj+AtU`q|6sgTE1O*j11X*CY@g4HVTW6;^G5_Mri@NXZ+|JJ7Bbh!) zqPWV)=xzKB+n_C3_iy8^eQ8LSF)zFvUl|#b=om34d#DHINAolqB5dndP@3Wj#y)$D$)cQXB$Y{Cu_h2ItMxl-V=4d8nt2v=FD+NB5weGy@5#S#_Hj6;h zM%Pe=Rx0OfTFCI@f%8>IfEg1wqWkV9l~Hfh8u^^m9iOjB9ZIoDYpjO6@bP#M-ypvyh9XRTm3#Mq*3?T*V}K8`OA=WoCQS6^#-Pz$n6h^VU$JflL4Cxm9K@F`@W2bnKw9-S;|bVkzwida9jJ<}f ztQy9}CB{bU9o@pGlifmmQHq1J)oY%PRBE}t0g0xx9ky!<)?ez?_SB4A${VAb40|W< zu!GOZPUuOZc`2;F^wvn89~<;`Va;Q^d6umgUYFJaeY0AGg)=D2%s-5G0Q(EP{0Q+S z5dN_tu)=^40to_A%73nYCbqa+OuW`LH^xxDs7(l|B6q+={~uZJAJ6pu$A7F-*UX97j>iDH-x(q9Ho%teIb>HA}~4IUW5t=|obKY^vi&o1=~p zw$c)cNu)m4tIu`&UbowIU4QIWxk+z%@BMr|pU=nRac96w_-xe&lQ)%1UbR51oeJ+B z@8i`*PHW3P2S$c>*V}otdn|c&+W35u#YoVr4>H_$0S`=xIzJ3`6kr?LLFMzfC0E!{35s>Q={Z!}S zPK$sQoQa7@Hhoj2#xP<){;k|Cyt^88Wn0w20#TYB&fIicRh=IZNQ=JRFSnu)?dqcb zBNHThGnUJJA}p$?u6H2x*{#xK_(ef_JjGHE|&ZkE3(xtOwqJNC?E*9x}(`)}qVHMxQy=+R#=7uOsz)X@fk^)H3 zcBZA7Kj!T<03Gf;;993{J`8^EFjye4h<1gURZK!I+~jwsz~-M%Cm~2+W%9>8a}k%aR#OUdw$@P}4W}eqNMZSS zY+3u}ix5UMwpw6d#kPWxb&v7^cc&5zx03YL!}+0W^N_k#;9`UU-JK=vwx^$LqFb=H z&%I105fnsA*m#$bqsx~^=A$4@*GAgE3Fz^K1W= zNy1TAJ?`W=?nV*Od}h@cX-&@|1y8W$*>~nLcLSzzslT-Gu8L^#4W7) z_kHHnkIVjDQ3x%B{R@5UK#4G7w!RZ*!ysvUez2#rOFGxS$7{znJ1W?i!$Kn1#2MKOhh*q{8C2`c~# zdeTcdIqOZiMqtVJDhpb)yF1ef;fNh>UiwZ)ZJM%koc22$b48JpkZkJf3PvT5;4U)K zZ1tyyGtGTMzsHV&2=G{$&>ji^aGydbe_M1)YA2Mk$v8_9;t1y(r<~;G1YhxXn=%j9Gy zML)2{pv@ZPQvt-oCM&TS(wCyrt32~mYT!~qlN@AKbp!w4TL43catVJxOCM_BEd-9= z{H||+qFeQ1_r#EpGJiiD;3*F|UIa4HsgRgZ%7sqsN_dv{UI0-wX;S=y(RUKBZM7(~ zZDHo6c3Bry`^Oj}mdq7z4_Pu7`ZPbpo9J}v^;tr!>BBJ=7|U&on&>8&AB`$}n?T+R zV+(g{;+KH_CV=?mLAtO_=DIC1ot+E>{xVy{;FpF|j+xeQvdELU)N1&HM5e-bRWh0c{Ly0oUQUM;u1f`85vj%hg?5uRQ5{qQq(gXMY^STk z#_&s0_pEL}6LoaDW}>5d$9$5Nd}Gm`c1R5H9ZAeeO+78}2~&oG*%}}sDdcwnaKs|6 zrr>)*q_IP=+C)A+Bq3WoS6u{^3dC!wG=F73Tl_J^MmsaaS(W?NW-55x{h_khGgxBwWjteU& z?mAxwc(>jUR+Vq0S;2h0c27_tgw!2mOKN$Hyv7VkH=C_>PXzSYj&Zu3$3!68Pd^bD zhDmX4eM}@Yj-JP0dZ4UsRIClEIwe!;blpz(!Z>eH_Xxuprz~VhXMxE;j0jeZ=>mop z!}Fe_IFKxLVjj2=RlEquq0k-Z!7@h6zo9r$K(kfk4Wp7Y-0;?kU7TS3FvET+k$;*S z4T7ezPFN3JKbV1q!ilprc`tx~lhfi#T^|5IE)m>?-oi|>C-+#Cequ*{o54tvPFP-O z^gSx)jnixCWQvg~)MKjSH%xDfuBl&}IgD(martw$mgC3!u(IC!%3CiL)`qlXu1@|* zQ@_q6WjK~dh@aHh>CEzB9T=6-NS57O*XRHM7IlF>((Uf15#nX^qRKza0C{Z9JMS&XE9%w8z=A_Bl8KSY^S9K}O@byvP^xC%4)8rfP8>LGGb*%j zjzga6f#=8F1T42?(*-UgGq$9cfk%_Q0F1;SwP0}<$7-OYk0XyKe(ANINe+yJq8qix zCA*0jKA3^Vi;7z)PbO^aLJVR<&020S_uDv?gggZV$~P5xL&^AL9JDR9m8c|eCcCkz zUtN8QKz#YIL)rEEwGamLmF!AFJK3}&Gv(i=5)X8tW>f-B5;oaO#XE`xbCWRA$#{gfBUrtz(5od|Ks-lC$Nqs>e;&cF)zBKmA3;oc3KyF zRSU9JKT~c#;YM1R?^ye@)ec}$0N3?@rh|Ycz+{3yoCAZ4f1aU#J&%d0KW1H_T%Pxv z`M`rOx9qLe_? zsZtie3aDV?C7?@5ULVw<2=QEV(l^q7U`9>ugxj-ucs;>jIK_xy-FTv4D6eWjMWG@! zkKu4`@M29#xYiQsC5mW$d7FxE$0MnL8>sMXCDs}c6(*=bxdF5=%5nCtx+O=yx!9l) znVW&E+-HJaz*=L}{_=SVFM^(IJe(yR9)mO(b*2xCr-Hsn1dDIVd^ZjOX#kw#)gy=X zSHh9h$An7f*d=`>zs6}$r!aA`zL69k$P|sZ6lwAC1jxnkbRnY_W`|$~hIW5m zovWkG%Mu9Av zhRWb>$(8Lc&baxLZs{Y_V27UoYt&*kuI3{l+!K{(I5t6}Dq*SD0iO;m6@+^j)nFW& z=9tYC1~72Bc6lua`en!p@^PV=j6X_SCU}*>@n$Tmh^?6D=NNxmKhcf40Bs!@r**cO z*Je(H#30|l?@Cn(%o-gV=?z_{e*(P}GXFl&iwkmI_9R54VaTn5Qnqx!c zx{)GdvuWjR{1%!XKn)?s1BK8=cJNrdn}?Vg_(24ep07rKx{&=zf<5eVmyf$(L>T~u zhn`7_K7D)Z4V@?*#W9h;K0R1_4t)ef27@fvY&OH83BVy#9P|LF)&;k1w9`w8`tM+z z5U0}A1qM|1p;i<#^GYQin5(^GGxk6V;RTuYEi4+?(fb_E6KQd0?aVsK`gX zbJ@K5rBSnb|4T5fm4F33H$tQ_<}{H*u*>!$>di6 zF0_b*XkZ`#Kn%b1a$+~HnpQ{@MoCrQX1@L1yRcdoEqsp=N@p&jU{shzxGHS}`& zSr;UIpMV_(o3NY_AfZZ#IH%jmUrejzgbNy#iD9OMq5D<1*Xj9wcCOR}hQvdP{fost z<~+Rvhwk#&gvrX}D6XfD?ohY_f{=~Uj_z1$K-MUSG~tiFV1wXBr7CdArjTY@ z3dP@Eu>^`*U}$W#X*Iw}ea=U(3q~Z3&#@yi+u{>+l%yuOlz7FBox9tc!xCu^SBbQ7 z$7|!uL@a)9_3L7%>r#S&7?IR4Y>aTK$N7k>z|L0tXezij|7MX|)$ zVP0D_CCb(1Tc0!~-S6d67=FIlwXRD{Ae$&8&UVTQN3|{XNkZW@MuBz~w;j#pF6r6f zE(Zj;k;zbj+P-{Xj3@^?sZ5RV!Zl%l>$QyJtKy1| zj+c`3S?9jvrhNl+Xw(XL`;2-_!XJHoxo6J@k|8ARiVN{iVK4X(GaseXB@(l^kcduq?!I0^=Z)0ut!3 z2Z>au9d4mbgb0;Xd4dr3R_fItmw}0zAYDvD^x2LG) zNkw-mx0m?oTIKcQs8dJpeeeR&A@7-g10my5N5iswUpHfys1HVta3?UN{(#oT#NLLg zzTW*vH4598X84Bx!pJ!$^&oFi zW`9vNd4ra#p0d?Z^m%K1rwMQdrLk zr#>n?xlk~E!;B^mfT*u|sqUDC+Q@4QM2m7DY>3Jm;(;blO460|@n*G$D<2(hMI*!w zD?m6?uu3~I-dZ~`hCpE62$0yMZXPT{=D@z@a7XmY>yasCc~9mhLb{)9JLHrh-8i`JT!csu4zfhQ3MKi< zE)wQCExhZYNk>Q>Z~=gh zIGn$NVTU0}10Kb+CC9bELI%OS_rI}2B>s%Sz0OXTzU1WQw%s;Lv7Ia6rGuf?+ z2MK6_K_yc!Ek3UjTZCT3*!K=CF&7Vk+H-F#!xlWISOC2cdZ3^!z4n1$_Xm~ba|k!C zDuWw~0-t0}1aVDHCjV^)6ZP(e3o$`5S&L6{?lzBQMCu`~{i zWO!?PJosM|U>@pv?C>l>^Rj-c6PS5|Fp5ni=N^x|R+&Zf5XaI(Y16)Tlsdk%2iY+6 zRR_Hn**COwN0%3Tte%|;HoiGLTRSG!&&*|ZZgl`vZu?w_BLmOe6-#ET9{7Q2=fKVJ zkV6_!ioySqeFL%*^hq&0GcL2rT04m!vej&)o)ZlG)?F-N)K?}LblOM^wrYZjV82rV z7bcOWC`+eypckI~50bMg!C2EC6ap`0vg4lnKSQ~3cY{ptPcn#k`Pcrue%^_O%b8&N zMbQxMmlsSv0}&+<>`&a+z{G=K)_hZGEQFLqjntQy%1wBz*ZKKqxLaU!Qw^%Ii^ZjQ zh|6U>V9c<~LLD8uB1IIa_+PKwFf(THkjz12B4Ho}9~ilNIDLHRR36l-TzwwOK*979 z-!blbcv~6<7(f^XNRo8$g@dZ9==5>DL!m19rHqvp9vvPX`Y(_i5L z#$=n%2_-0c@H~>8ubY3)e<4~J(rZp%Ee2Xk50iV#x^mi32_%q0O@EK1=a!eE zNBx%Y<)3P=I*fn|k!uS|iFygcZ>r~0?Zby-W&xQmbyT)JIA4{O^aAlzvdgguU`E+@ z&{&S0gWvXHHNtiMmD#WP$=(s=qML__n%gD`{*-~#fpppNEo%m7H_L7AxhjkAIa6!HlY@S8h?^3Dq+#eQMi zwbLQRhw>*iE}N4hd5x6*)H~wQ!eC4(iyKls0dXyCIkDs19?Q71lE3_Fe>?j}P%C)* z?AdJya$@39J;~`G^D|Kj^BRuj#J1Czp5!KsR=ku5QYytcZ(}aG`^ojvm`%q|HrtTd zhyT2KY&Y*QbjZf+XN%3{bJi<(K(}}bL{ft65?bW{J<0rEIr~1#c$;>-{ftxd?rGA* z50RKtO6RP%DeLz8JdZlg#T~H?Fd=qsbvcRF(Z}xNBsW|DNZCVH8bFm{Vxckh>OAFj za)NC$2qAO)1#ut5N+iQt$Lz~59P;^=&q+ZWxWbSrq>28D+$fJDA8Bnwn^uFG0V)C{ zYxfB(S`TgtT+1O-HGjDdRz&grlr+Tp)c#7dEFH(A9c!;Cp=JSJfYjvV*R)ga&tBHL z_Crc^I|4h~jcWA13C>&xH)#c1#70hTvk#}ib2o{i#76ALreVN=iD!l1wA9Jf#mS&t zD`4e`%4qdUzafqR#m=+8@QC}Sns=_G-3Yq`4Vm>h%;Ut+|Bhbg=1-m`=_UR57B$!B zm@jWglE6FBVAG(V6Zoi8@jcH=G)C6r1J6|=^vP_1x3hfqJRhvj4aX7&eBFN3#wDS< zrP!0rMarK58*L9MYU{O3nYH?oss?Rnr}X(GLv%!x^?bSvab`+WKQqpCZ9AX_>*R-H zF21zGi)pa{LAEm8cq=dVOGCBR0yeXamuZ4DY; zzwtyW1wx&!1AT z5GTN7*S7~J-F5`*g`jQ%C)QrH%1?b+N}9y2lqFJJ`~uk)^j?YPMz!(6$B-%dhQ5oy zlLp>Qo7vnjztKE?f*j$ryk}pd%qqpm(W#%8&*mC}-ul@*fJQ%-9mTA=419;>U_a!Y zE`9G6VMsb@=Gi?$g^^99@ZyIYRNA8J@T%xng)ZHW|55mQwlMB7_a0Q9LQiQmBrJQu54>C={)upcUtml`c zu5W!Ga@Rw!@98b)G1g_OG9sc(b$;6^eJ#H0JAJLx6r(E34~p#<%(}4Nlv$&!7gUqu zvo)VXU9Qx~Y{-fG6xUJ}?B1r~<6X=;e(qdkIz4;#vI%7@uRAi`6kbL(RpvHS>&N|< z04RA~bv!?)=6ckulwLCx9%IyPvw=R|b7Qnnt}Pql&HQtdn~vset^3lDggv>EmfQhk zug`iNTILdNn4qmv_Dg;62j!Es%ll2{n;GIicKIe!*4fv*QKIg}e!k01dcHoTzF(XX z^WiJ=@-N1IHpg0a=JlNJn#7qLLYy8Sr!Bub@@=jj!XkXZj(Hb`lVOhO*$NkJ1&WdJ z{M~nj3DBpm$smbg8n|KuN<_o(;k=&Gd)7etI&6Nt`kEj5Wp`p~a!WB;x7E5iuQwNU zYHPKe?Da*R+M|6_Ra~hlCDn@V&?_a5eh3E-2O3MUpXD4|kLP790tszOro_NLmAhry z?=6#xFtk?R#*Hz@l@S!kp?PTq`6f*cQ+UqSniot9D4(R>bo%>T6l~|7oO8j(#_#uT zjTE-`2zQ^gc+&Foy3MBs|GhW>MsHW3-_C!}a$NIWF1W$KiW~WTS}4skEc7VqIMX12MSu0AR#!a55}l9BTz$0mva%sz`tQR_@67~(01)Gt9_RF?p^XlkJ_uT0|9eT8>YV%_A=i<8d zw6C3M>Res8S?m_{T%CpCab39eIbYge4Xzi{-j_U=)$? zSaJ|Yyn9^d7nK4IIZ2PZ-!Ub85;!r01^Gl2*TOWaU!m6|rPFa6+tlZTe>xNhu+uyU zY3lT~CILxAkkXeF0Kqv-Wt~$!@b!_I`yi`SCCbX;H8{-?;q$7J-etf#B{y}d^@>OV zOo21Q+BDSAJu$#j-#Ty?U?NhC)}`d!FNe)DZdi24$E0XyD}Y`z6ECVcux##Cb0j|N zD~>PM7JbEOk#x=cu8d1`)VENfiHfV~BWKo9FCXH2hha=5JT6sMSoaQ-?; zlsLnIucmS0+&ESK@p+v@zH97Qw7!SMTLBL30o#$0r<`4Fj%Zati&wM0S0!^Y)Yz9l z*!m|MNpYe530;Gprd1EUq62e#!7x?0N4%2$ksZyODInmnM=ym6Z2 z94>ij13x)b2iq3Qx1C~C$h#~eH@%wp`i}AlbKqkc#RQ-JVK9^y^KKg;O3j2!JzFXVRqcE_UqH;0@wUgE;e$ z0sEznzL60;dt0e}6GDnzb~oe7zjudMl~0@6w0krm_rHT{*~8$IatLHBoroF5Fk|xQA%cq-(Yn* zhFPGX+Om9{teyxfpc*fiOMSD8X^d@=Ooy!_6f^HPC54U{k;z^Fy<}Ke4Ain_ZDTq$_p6YNK)SRKIQ@%ZE>LLQdwDdVuGO|uy?gQ?qZUesn2a{ zJ=&YY2UN&xRY_4b9P^EEZ`pHU5_qJ`EI@icCw=Ytq|8oNpSkKdu_wjMD?`%JTC;|I znBzIVLmrkT9pxiZDT?eKgR)7m=F|+8$C7oNjoTW)W;sB?R#}$;pbhjsZ(Nn1^_CqG z=+`IOY27-pJqEqBG{se}K&JY2bU1&K#)<b)~87a`WUYGn58s35yY>hlR$nxd@7`^wFegF3g~7^ebZ zMrts^wQJiPd$%@2N40632RhAAv^jCM&c-FF9=I&r{(U^C?yF?SF9jLJJIBt-CKuGO z_arg^6{R-aY`((|9U&ozv(+*13UYJLC)nS%r90E7Xntc;$L(|yjSq%3bpH8k;2X7q zJzCKnUDLABo7FuuTZ>6a&iyduXVb9+2&JIKl5s%BphVoHx?J~Br^EQZzxGuP=0*;*Doz~@%R3sn$_~mwpe8urqVlRN#rXx3}%P< z<(#iWfom&r=dnm5Lk-9Zf)vWZ!KvLFr?x$XK}8uD^W57Be`TlYuVwEpZ<#4rbXqPN z6o`M3{QbkP_Okjd+;s*9-yzk2VYUp{$0QJN(e<=a6Z=V&K z1uRXwtnS4Wv$@8`=*I}}9YGzo7g*isYx#vP&@c#aYjd`uzmnoVwi?E_7`|5o5v7gD zp)E7Y#DK6(>7@bQJvISvUW)YH>{!v3&}ckNqHwA^C^GnChv$}EB^}(E&qpq``OL| zmz4{>l#l8_-V|AJBr+ zNUbvSBT8cJhZhsvbMW&VWC(VVMURR(F6r_V^?2YzPq2c3IoqW;au12rs&GA$W`hYc zNKS42BK1fq)!T@8nSy_S9W{A*0=PUUqlxJ5iCK%yk2kGk;nCfn6QaY}QFwy^3zynrMp#q4@}6C!WU#enm6ySTgf-2RZd_4JK}8 zuFh}%)=bVjOEZ?}#9+d^y;HHmfL0sZTX)YpVNVR|paL74r6Fkf2uEgqLiT2AsL+p- za?)!(Pxu7s^(}4I4i>FHd-g#_alJ*1(~B+FEXISo@Oa>M8Ci5n@?8}$PnA!MGYsP& zM2j0%pVQd2Fq17l0CJ0g7jQChlDs~!+0V+pr{$1yqkHYE0JG)BQm5BG5$NL!3%#f8 zP8au8`aPaM85!)KMrd@Jsh<7vpnl_bVoarpA8J69R%fCCV)V18?~HGUHTq~!a%3XK z6d14@Ava1e=p;Jbkt2Rxe$|EwmNGF3=|(rjq)QauVTS*~e3HR9OgX3(Q|&8Y7W&#{ z=NEBf3*r2rYLh2`89G-dio@(WLV6=uiUFcLB>`Dgm#wchDQ446+SkoqX7T@(z}fcY zF2QP%HkQvsv5INXfz~m?kS~p;LDz}_A)!1RjN`u-nSIkHgjM(-Qg9(@=Qi53gKR6c zUn?;9_CONY&JtlT$2gEAOGfkwVV-g+qJ;7#KX0zN$1#iiB^-Pm-m<>DE+c|R>J+`} zq+5b^cp-ux3NqTQorIYhf?63%2Mv9#iGP-hFq6ax*?KL(<;t5pzI+`xqE#&j!_VG3 z-bsijFZ5KNwosH?d2kaDSZ>P4fgX7jP%~;EjF3rksz>lQB|so0I{EReRC9Ez^Nn`M zTV3zk-GX&q=uwh4x!SddftjiDibVgqE>%ogOn}}IU8Htj^inuTWc33iWDnz<)SZg= zaXl&7_T_(=eOh}8e-P#a48?}C;O!wQn+CSH5uQUi5VdYUZ_|bJq+{-o6p47x5UZs6YgR5VYfsrFTk8ewwmA=O#v$_dZrVhB_1=YszY4O*$ z%`jMOL(JS1!)QJbad|2PUTB+>t>)NZfaihp)5w>56E%o0EiKG@x!|~Z^6h2{qL<%0 zQYw3`p9;Ej-L5VNv=5#q9v#(iGd}!fP8tk0kg(tbaRT#sj97Iqm!?Dz*Hpt zZ;wU)a`B6Aj{X$8&rv6a8tx5aGRV4`boR;RxGrM^8!~+eTI-E|&qaWs=UHn z+F-}TJ>xwo_qql|etG_;R4N6UDaGd`c;B(4Z_1uqU~sln3+HYE^ZwUK&wg3N7=`mE{lqFOq^E;$zC&jA2SpSRii~rr5s-~YE?@l^?`{%gN z^N{5=Zdh+sJgTA6hP1h!Yk7w}vs36%(s#bEh~xR)02rD`?`_0_qU~ko$7Aj6%0zKK z&SgnsmhXpiqlV+kLX>bWI`#S5A`XL9_JW&~S%Y?Q{_6TRvvVEv`Uo$Z5nEg>H7Mv6 zeUimkle$4h3ufVs7y(8F5TR+GX*02VT84wMrl_n1RS8{=dq>ZbU}NjuO(Up&N`p#M z4&sEWz}{ZY*|USTBndV@cW!M}!7Fgz1I{t4xvva@CLMsDA+Oc( z?t*$tnc6qIPlOsy4&BxuSh}RcXI2)&@en7xH6p~7DC@mV@F_&!Dm#x>9ldp+5mlu) z>9!s+M$mr~#KgturOv}A?nMNGs#j}Jb&u7B%L22;i>rq|C590mGbg`u`YF$EsWxbeIp!A-fQ#o_ob@D_JZ z9`YNwD#Xi{wZnK18kss3>}HBU-G(~WWY(TDIt_2}DQ+$t6m&F}4t(%aST+y2A-=ql!^kKBr$qdHaouU$J<#QX=T z={YVv2z%?LesWd@7Z%6=4VuTQw!EQ|TP5u{V3pN3&Ywm>|De=uNVgrsA8OO=psNf`^fpibj%ck8tMb;sl z&GrrWqfND9nT$rk>B4=7>b%lxE7sgKqu^syWjq)yya;bZtOL8aybi*%c4|kC5w@6I z2ebE9bL5?9tX}%P=;RnP>v;^(m@G00$!5SBibQfS0nKp*cXGZ*I|_DVlkrltaapD! znVAB0rj^^$LkTrlMmz}mhGXnO#vHl?j9Z#I*M1bFBn+0d68_jNj7N&=iR-|mur&2~9xe2# zNEnN;h4@T-5CPVhQJowb8RJ$2ECSGsAYsE}I3e%z6cthE%!BupF*|J5ma&NMjG6!H zct9~04Aeoa6uk=Gi$*SpNr9E%3AA2p;efP@EJuN6HO%Wm3bvsFhNo4=(M zq|Ds9Q@_qC2pk67;oOSuKEcK)v8+#Mid}5^%6zr)H~_D#dk4?A7O@x94;`ix$uByV z^em_g7Wm&9pwWMqOfMg#b&Q^bkqeFEgZ;%I={Ik(n1#ylqxo27K9 za$Qex=&Q7Oi@=mVUVX=@JX;#{#RXw*b8_F;yi|ac-aFo_uVWG0Ae0o?3ME> zl0von?hgE@Fm=Lc`b4pc%x+EiX{Tx%Wj$b0ELLmbeP1q*jR%{W zxm$zsk>DtOuM)^`I6l{2HX&LJEUEK+&+WHn=h=Mbwj-sqVKK4V)bPiEEN~QYWb7}y zR8R1nlsk=c6B~G0sB5c3*T?4vIz1qu&*!4XY({Im%j1&=gU)r3Y`&#M#rC6LK%wYs zv3t)Rd~j5jPSO2{-HV0xl4qh6SH4`=qT$Cddv9s1r+4y0eToPr8K70S60Bd;Cs^{MyM=~GVB_JPJtJhZi zIXs3CCk#S8?KexrELAzLOaud=4jCkpZv|Ser03w$47S)H4#Vue>`A}o`%E+DyH#KB zaM34EuK#_Nun@b#ba)E>p{Xan=_+~s3B%LS|@SPY(svl^+ zZLB}OZ}}75*ve-P?03H0yI06&zl>IEZWG=+FK|f`a@O)D|FZzfyS9uMFGVmr+Ffq6 zy~f73xUWPI20o3JhZV(PNJ5n?Q19TP_@y9zVb`K5uxm8k(njJz9C*F zwZEVXv<11s$v_ko9G5*>zCzzobLklS^4V=e<#7f5`71%%CH$r|+k`RjXivnDA4n2nn-Lq)>omhu^~nD74TEM zRe#QI`_}Lfus4eGEY=|L7Kpx!B_#@yvaBnQyE@65a%?sP* zo&qGyn0Ohfwg`UbBV*YNCbk~fa4cFVr#gmXn=S|7lf9ntzNM)^j_UY06_JF2ahX`Q zJ+9^S6JIZbv?J!>tR*6Tfz-AUgDmHO%_$6Msf=7+q)7&#w_c1A!2DJLX9&kQ$&2A> z&kk$jv8aJA(xse>ceTid!g(UR-YEmbSPRI%Q|PH8WP@x9Y2*@|!%Q|^OrXF{--%=< zWwvjmj|xOWkPJIP@i7Yqz9fT6!``sYY&J&~mAH2np7tfTu_Qfj~ORK^SHMys+$i*nUuzlGiJ z_^c^LRi5dXbHw&#`-ZC;Fxs;X{}4f>sc)_y@l8b!ti#TO*7O+NN%)J+i@1RgZSk%~ z#cdnNsLm@Z87d?BFh0m<$3qBERUqo#RfD}P*m>$U{pi2#D=F~nv>)4&JOYz0l}O;} zL#~ZTr`M>SZHdWfD1tQ;kBUN^LQC8biNPA)K@m{2cCO=G(6ZVgo`#Fzw;5bi7?Gi@ zaXETHT{L`Z!l5c)O9@w?N%ClSNgB3&nPkSxNf>hmAK(0o3non}w`-2rj;h0cKxil+ zho`2+5$^nv4x^BiYfm}Kk2?dziZu6NIya$skgY!RvZSQ1;ypai7OYp*toNW0QRZ(c zf@DV6vDW}L-C@2BFq1Tp<;oUpwBH(L`nA?4v%d|y-GKR-r49*VFcauj$u8%D3MJv5 z7#OH7#t?b2iPzWBUu@wSb>T`DPCQ62$HN^qt_J@|Q`BqJVuOLq1uos1|yxTute#44?31)D`UkFK#fbWIDNb^yTi;9l)8;q#cWQ zQT>{agvXoqi6|W-9>T44PNzm%Z~!9+zwQ$OOXuY^q>vL1JrsCwa+o%LZs{4b)mLmD zriy0kU=$Z(I1nEvZ?(Fo?_x@h21HYU+a1@nOOHlXK91s7juq1=a*)2q-nBoKw=p!L z6E`vK&{gq>C=MI~SwbRo0UiSOxU4h@0_e_TRs*<2k-jWIQ)EX!->B!aX4QoYuZlL%)a@@JCD!X5?4KZ{+T^iKE|?n?YY`~N#v33gcd>x&0N*JKY` zQvUks?a$7}P7CgxgjR*JHY;p4Sh0c?r5S8`$yD3{XPovr{SkBPiBEWKwF(J+EL zQvPv13?J&JR#ht4DPG|s@H28(i}Op)8^)^gOiW0i2l5z5tKR5PxK_<4yEZOO^5)W! zz!7hXGK^We;Xi~qU7tY6+5_pwbwNeWJ+PNo6iidh%+UGR@SA6Jz_61XU@GZAqegk& z84|}|4+@q#v(>5ccHh{8f_3JY+;UD5)ATWF_zyPM01uR7#7QRcJ=0$_?fN3cA`wc; z@}mnY;QrM1Pr7K)7c_bKE~1g}`x1-kRqV{5q=ev9!G_cGwY57P^V1TN^!Uf2U={leJCooWA`KZtE~0nShV zed){Hfiu++noxMDZoz6d*`li*!PB9)kb0!vK6-D49rGRB=U&Xo$dvXW(NpUR^N#?T7-fXAav99$0?Rd_ViK{H?iP(c6o_e8IvBjs*E{C~=H0(Az``deyzt?!hibXv) zF)+DrWck(A)38sqVlJMf+aW)2=Jd(F^G}j4bOxh%hx}v{3_PUk!~^RPrK8jLF$rD( z_7}tEqZi7Q!Q7)S3^Vs*k2+KtxGRlo9dDSl*1m5YMlz!eJq)rfq}X71r}yDE+B`0c zw|uyOYBJ>RnWqvv?PgF7N2})c8(#CEg|3G!4|-w3i>aFrco%_I5*O?K-iPz%Y-@-s zdFs9i7iDa6em!VjUC#u$klskmUmUwjA)HRP7#t7wH`(iu30X+peviu!@YIDGPPi<2 zuHw+~{`?%A`N2ohrUSf>?FJ~bO(5c$`lsH|Jyl1$mJz$;d1Gy*xyk8-0q;+eeZ8Db zUiK`86)bwM^M9fA@@0AI2Z`>EVw!g8Q=qgvQ4T)`l&2_+bjc(;vU}IofIRBy0#68Q7j2p3{3jbA~L-Gd_O8b2&v*Ny6V zk#2+5d`^Jnl`@-q&@&b{b1@UuLUtJq5=V>R;vYNj%-1pa> z?fqlRrTe*aQd;bqTNOl1`sHR`{ikGq&ZBPxwaLrUfuz~1Fk90e%%{sGy}l4nn%N?UUhL?q! zn;{F-T&ra!`D+xZ-GsB|+kHvUjPRD&t6+LYNjngn3fykUggpfQL3aoOgf%mh;$kQIxAkz1O8}j|%sBe(c1S7xe9{8`{0g_U#_H01^U!u>39> z`2W6^_v-fCY>T^i>wajZaBRbGo>2h>&1U0rx8FI?%DncYLB|b^hsccsRMe8=kJs-< zD^ZgfR-q%*_VPI!olez|{&r8BIaQ7k1ll)Dt@ilmjF$b8vC)L;Bzx_jUmY{K$PW9% zcGB}n@EsR^!gNiCZ86y`AokX=4Mr~~MK*{z(DvoGo)7 zU1`~79w=S2Go{d70z6_nqTUOB5(>f7cHUCq`H#hVR7!=oi5elG!T(yQ#?QK zDlE#1CH8x6U#&Gq?(imI1-RA>YZ2rX_J=w148DdN_RB624+Pqa5#T(zE{{(FbR3Vu zl?ie?rjN80yCJAS<7A_+&4yG|;ErmbIhtTQV56Ab5% zOQ)ZceOE=O^D{4}%!kntYx9;KvOQhv>Q*^oy4h*t$wcN`k8VI`u=H(g!70`%lAESF zUBgmKJ-O(kJ*o zvH+g1>b#PU@g1BkXl2CWuYzb>r!|33ax$x7yW(>C4?pAJC8?qEp>0#I zLwulsA?7}#uo~ukDmV98e(q1r!B0kVCs{yJ7^BBTpijkOmn(>)LV4WRE^&GDpLE&> zpzxDhKLL7NbfqX=`I@Tv?k3p1rfB3&wXCej=Xxk@|1ZVA-FPylqb8?5`7}X4{6)F$ zlGKwpkO`Jk>#@oqo9O%HN<-vT(Wm9E>28Bw!146?T?sL6fQ*1ZS!BKr?79zgA(|~yA$!^s} zzk@5gEPSOdr16D_7c4%rv5VcikFsecX-XH6d~%h=;8Y(6{CaSelHc;x^_xclt?hMp z3*qD+2q6&ohmv)0hw^xCx@iQz_nBfqNO4bL-s9X;cSn~(MWDOgm4*F46zt&(DF0@$aT6LjNVaW(m-#(%OlSnbP>nf zJ#*=_`pptUhjL`XjbGu0V@}s;j3-xL7VQ3y3|nZ&e2N=$^VsP(Aoy!C#p@%D0t-Z6 zxnqtrcD3Rl;M@>LYksKVVGey0z|h7W0iQtaPAY%zUFxcPqL(TWM!u_l!B6oH3cDLv zx{&HMMrYvM-2I2?p1VutyOLS>* zgY-j68{aqb^xv&$C#k{q#<3Rs+7z=H9^>14?IAzS?ZA6STa&hxKl~iyhj{0jZ7#C( zYyf%BrLvOw-G|O6`yfFcbudUY^pst~hT#`CQO?c0b@gjWwUf#a-Z)c7Y(plTqF#XURbl=1h! z9dCrrYk~$IP1;?td%C8g{Pith7PGs5Y^dYCThoR6Nz}*s>c3Mgs>eax`HwHt1u$)1 zCvL=I&}ZSFa2GNn`i=1Lk>mm!!nKv_JqD0lKFN}7>G-9)X5<}qAA1;oG>0ObbU_{~ zTEm(PV!paAtypr2VYGYGP3N+bi4e~ZfrLpTd63hFml(#_6vR^2hwo9IU`%dndL#V5 z*n0PPrvLx{|G5K0j#;m^3=?5G8PXOqdc~~FVU(9tPUY|_naN>A8&PRyPUW@Nnu$47 zidPEhFo~fY8ZQa;T8>F5hf1f{_woK*exJ+ld-(jJb=c*?4(<7P+#k2w_1b?#7Po_6 zFrkdy&R+COybFPn-OUjJwTn2iZP!KYr;iH@wn0bY@1@(ke3p0;vGQV@qVb&*HNOdL zU;p|q?BRcpR{#4}@!ZFA?o*qd$In6q3X2tlFSWA1TK@|LzHV4k)sUvWG4hyQ&{4+s z7DotX9+z=emowwP!h0e)0p!qN2TJn1Rs8@KIm_k&j;fA_8|Jc zG@!+t!)~iCX=?BI)~k^&`ZO&r++WH>nH2 zUbYI3dL}BSUBn`lCndZvJ|LLqhWB)8EIU>0$~Z*rPO-xC)JXE31=ntnEr7W!67gM; zpu9EVwk%t@>`c4hF@}0^2k&O(Y}8LAzK+4@?yr!N+SM%Lri=>JHLX(~5QZ^74;65d zb5lQV6tR8P5fPgj{)o|jQ9@;c`0USu(Z(^+ z4@=lA-k`?Cn8xcQPxaj~y-qMK?)^^ey7S(CP7X{kb@vq%Tq(jMy%h=j7i7Jjl(!7uIJ6x&x`ZKXa8pyj}8H1-L*`^F8om>lvHti1B)NU zH84%5qAs`S2}4+N^PJ7MMZU*EC83tDsC$bt2o`*VfRM#A*uS_3v^j{6ttF#k=cyhbGtVGjm15+@IACECXe5pEpBTmN{biU}G4zU{hhXl|?yome7 z+gtkvVJxb>C#J5$ew-71OQNHP_zq}}9j1s}Xlf zu7*^<3_b;!>Sv8H4Z<0DYOTxdP$pLSd-%rLE|ncQe&1+k!3~7t`KJ*HNfmZJe3o1x z`tyR4e}zalccmU<_GJaTe!N9VlIlJ2cxq&|Z@(4Y{RO&syJw^PG-FV;EOJ%Imx|e$ zbEZc9z36b!@>13L-PllBaZDZF#M0v(yKlhh2_(P8^xc~0Hc~U0dl%Jy)KO6rp!wU| zC#>ncCN6Ds){8YZh4}HitX7Ll8L&++yEN54r6{nEtm3095XtjH1^2Q`CwuHJ+@YF1 zq({O|->5X;$DZ;n`XD)Sn;*G|dipe0Cpu1d(CjYJjpJS)lSFEDYy*L`T>9yVw83GQ zV6_%|s5}oQh0IR3aOqan4Z5$Y-mi^1n$S()#Z_pwNH;>f+_V-gv1LFeeb>}_an%c! zebNlQGtVu6$6r#w?JvZ+8B|+lB#6c^OH~I{0!|OUqWG)P2e9S6@BJ1EFRbhsLyLV0 zkjV_AXr*Iw58JUe_s+q_8R4I5H-isKW}UAX?cG%inuJxq`2|DF>0-79N0t0E)h|Hp ziKWN*O}2(xrqgg7)imE%ukP;i{CA$v)E5e7RG?{Hf@$-*14mm);Bj<%)JV*?7UZ&2 ztq9?W5Jv9O+{jDMEe`x;!%RMX<5|G*sCwEX0n^B5)uGnHQvxrae_l6cz1w!ZDdbXHR{A(&9XxN-`n*6{Q1P&=Vz|khRttRtt*TpGg|l8@Z68A8@n6brqj89 zY1_T$P^KdXF1B+0GuB?Dmp@H0t6c6-8Y?bWHjg0$^r(`#^%b3+Nk(^!T8|=7q^gA* z+HBYiZ3kRuXQM=8l-O*QqP6uBg#VbIQ@!_*rLpx_Hshd28`ad#deCK^W_oe7+j%&x z;@HQbG7J`rh4ZdRtYvaDqdL{`&~366ZQ@XAI0owh-MaedX?VR&Sv_eEquMdGBg%l#j1Nj31VS8?8zU zvzG9f#nHc;1I{-o&ziMsyD?*$(GWx_o!5TwMJh83PqnN37cw#WF}-TBvv;i?$NAhH zhbt9Xu|urB_56z+->vdi<>ofGlT}4@%9wZw9=A)s%Nb#6D!il?8WH-#b0e|Y2T+89 zb;erBIjBk8M{1%97Exj7W@$x+_+?8{p5Q=!iT__(2>zHc&E>OKukQ)d z*#9WA#zloQyR`2Er=!s+@Cz@FDgIfY=`EQX*IcI|;@K>hHy}t(KPP3#(hd) z+YcJJ^BR~n9kZT6;YawkJ?Nrfzs$MFU27<2abOzf4h1ihD=aV$Ic&-!%aHjYj(#Mm zW9&GYnIkDh9v@#St(~fuBJ{XFtQ?K*kD(Vb&lNL6IfU@1@j7wfr1tXIJ!o#jxIOjDTH`=%i5Ua%4TnuDLb{eWt{ND z1}=FtO*vM*zEC+n$<48d?dYLJ9Ga4Dw8mf^vs8CzE?rgHig4g(DK46FF5DOncVjcC zW)Z?IPc|Bnx=fC$oY1XGzemeCIyQ#^v|${Pj>G^TqcF&Lrb{q@;u$+p|*Rm_)~r& zG2u@yv=AoaXuJDRU#csF{(nWO9}PO1`NYCmYObnahhB~Xjx(!4Cn8nl7>?mOE{>W| zgZkW&X#{8u(pSZx(XQ;9_cI`0OJCRQ*L+c@wkwFjwA)3 zu8`(3C&J)LEfst#NUVZBSC7gfz$E_ESs_|a#HC%lH2(2dK3t$Sde81%q{*zTqIQH; z2b=Lek#a4L{OP*hGc}Xitb)E!Q?2Y4-ns-unEU}gsCHaa{J7xZYt$_}JemMbH?Rfi zS;=E1n4SF@=MW|33YhoDJU34Z{3!)t43vu!uft|%Jp#fdL_lbGdVxHhwSCr9T&Ivd zdOhp+9)9Uq2vdk|#2i6iL!4QQzG0pEdD>Ty&TNBc#gZ6>*x9}Q9l?h`R;hz>Y-SEQXRoVNl0kMtR)9rzYu*C3Koxtcz z79}%XPQ<)Kqq;vjl=XoW(wg>e_D?_k#jPmdkPi5o^3I!#tH)l04Vocp@8J|TxUo_L z70>mzA&ma(!d(Ibr|gX!`6kt0?7!7nkUQ@Zy*9NrTFiaz{TXpAJtvUj_RKi0AoWUA zXA5flG|%_SuudnK6Rimz$!s>NmEGZ7qY8q4sX&n-&xS^)v3GGg=qL?bIuLH_X6X%(I<(8UBNS4@&k`yc( zUzC-%xunU>7PFFiAR~h9WeDyx{m##l?5N*HfJQ=6KhaYxt?W-?r52Mj%}!8-VTrcK z)Q|uf?I6c*sQ;EOP6ha zV^e%9el0{CYHXKSz67$3msXu9S3$7-#O%O*ia785LC5{R`<$PrJBbI>^JGLvg8}X`5u& z%TW{e7^y+qf&HzRbjT{18kWBoUY*d^Q2TX5jowW#l!CaAS)>XsxEYWfJ#(5g5N)wDcGKhMz`aQ2Sjek z^61Dz+E#9+;XF3OHMh&-MP*qqQE?H5RjFziUu5V))85Jxjw}xe>Ra6i$jXkLSf5?e zlC`LzZ`u*$5AmAiBT=2#89XG>LFf8z_(tCGXJ&s*%~#& zsFGEB!UNExl9yOE!|S`yT%Hz3ddK#+hpOa({xbAzwxH~z%kff3pYp_U%ub57cb7t< z1=1`{eJhWdbzD8IgUhLVQ@32fs)NXO#oML7)TkG9bdd_0mEb&Omr=TK=b2qAKC$qZ zAWrDpwqL;Naw=QV!F(rhFZfdDJ?7x=k081KWF@@MDOwa4Rj8{4#d#9{nu;|G6~Yql zT5D5)5Xs|k!3Q0tXNLwZ0=d|#>iPwX_l2SLyU)4Pi?q8@aki zQMZg~u}7fQTH6xW4ym^~@zDeeDOJ#__*&x5leW}N7Ikh zJ({wcoM{a+ePPG3S-d1AH-GC72s_B1QwOI>q8m!%qMv>$qz8m!&?99pwK+6K@omPB>rL^5f}~ z2*=OmYhal+XV%4)I3ziJdtRx*4Kyi?Imc5(L_=#poL57`GAIK7xrD8AS)jv)cKhTn z+)en}cXuJyz#t{cKe@xWWyYf}FZIytx&W17k(iYenD9>(o~<~4!K(qpNYCMO>+7o> zpmX!Yse1O_xWQ$^>4XpeSX8hn#pF(=oV^bw7q}(&RO8#!JAhtG(WaV8AA6vtpIeVK zssM#&CI`Y_sZ&a>y^#&Ccvr3pi_rmy&sU`dN+Wtw86>;wHmBp1B+*KTK6pj5R7O&L z1{Xa?R>A#r{;a2MF=?fck($e@0;8SpxT0{zMGjR%`&552AY_1w0*t=O=v=WPz3(mv zmqD?lF6=5JYgB3!6uc!hD0hA@9)SD-*>S(wl3~|2&bO8qOJ%BC9&nGS{RG2Yb?s{hWZrAjPH5OyqQmjrks8PK>U?ww*0H&muj|)EYVTE zO)I4X>*&pZ)LeMFDDwgpUK-5oTW`YZQgLzNwikvwKAGpZ92!zLH7qk~Y>P^$B|96Sf4HU=eV5s?v`)@tVTw8(+PoJTx zFH7Q>gT6fd&%Yn_ZhR6AD3o2jN(RNdLdH^9lhG zABOKmKDBrjn|j|bhJEEFIJt>HE$Q~`uor#iE=_`ibmOz`sqEp z4;lj3vRNvswu8~D!L*1lMJbB>jQ`Yc^(f4G>hV!$O_CNm80sFRt6@X@_$W3n>Wp!=* z*r7Eb$o!0}4`h7iJUv-D;pF1@xg=q(PjAu4fsx#E#~aozv(rt9Jc<%2vuxBlfCu7e7Zwk&e`ZGo0t85(4!pIc zqb6JJP7j)k8b(s0fb2aF&GHPp^6;{h#`iL*AEQSqE zsFYwFu6EQcpZ*(!U4I4BJdthxtIA61##1)(`3>jx`b=iD&$_lb*fXhbMYWE7pYx%W zgm1!to}Nq7)JIlnbkP3!&`G2{#>l4}e-K>1H!^zIsR+UvW2?cbYKvI2^JE|afWk$0 z%$N1XEbPXAJ&67*xaEv^|`qDy!mSEb&!9(Y+08vho)gvY6thgzPlkx$e}u)ld5# zXn%si1%m(r9mJ37$^%a?ip5u->4LTe9BW2FOVXW(P?`h`ZZ>GSO5K3W5Gf9F&IaQq zYP{CixjBYe*yS!66evt6th(j$IvD)g)MO|jD?K*!0A{&rFX!eOE$d*u&pp4F&@w?> zttil|J3hjs!Xb6{=^0=5no5HMC2ykz=RX`4M>@qO##g*?qxZD~asXj`HK-x^k1H<$ zd=h0yZUBlqG+vtjqaFF5Bk73n&s_#kL#XRTv2~f> z&Mv;#W@3Q}`1ZB$LCp@}fJAUmhXXgke}&=K0Pdz>%Lu@o7&dcWp z>V;`k3{DB7UUKUVb=W~$C`A6b2X^b>Ht8g%w;{}C3d%J5VVzYPRemBcf#TIb1iNwy z@;xpm5MpT)hfh>*4-Qp#I>+`mP2Qkt|5edmxih#FgFBinb9Y$`=D(Uw5U}JiOI*K8 zojKh?%DD0*+KiiLo!}S@6g7&=o21Pel+R@g%|Rb|eN>a;31paX6NJ$3Oc)dQ>$PN; z@#4FXt7RFEI)QE@$-}&lRMi5b1BMTaCWXm|8;C+*uXu14?tP19M zY_j>41yna(#V$;9e~z#)6Pcpz<$KDfDZzRz_y!u|i#ZH0C~hz)ObWS;Ybt}oq}wuq zs*o^#4#1qp?rdvIJ>F7bSA^N|^SLIT)ty?Hap2*4(;v79GIlz^zh^T5k6tYUSGfOQ zu=k(e-`lLm41GPTXdIf3PtV>)0h0hbo5CYL&KYc}(^a>J@!!MS{@%&7|UU3n#N z*4X}w#j08QV@gkLLe7H`*nr^S-&3UJ3j4n)&KDaIS*{oBw7ndazQfGd2h^SbTAQq0 z&Ni6igSNeN?9}P|V?0HUkl3PgY`IkOLcr_ft=YBOfnU(|1czH5Eql9*(6*yb%@mQy z%3&9nZag^JrF!O~E|5Ck_IoR(PQR&=ykIQX$ zYFyLtcK$Zw?6=&n`~AvLX6gLpzg^ziZBciO2RL?3GRP<_YC9Y)*s>oG!Uq0YhaA}x z--U9x#}#X>=P|okUa{!#sxo+Odyu_Ds34aYp3yT!m1*7om_tjpN=Qqx5L;ptB`#(f zK}zoIK;h#Rk$#o|0&(0-1~akcvs{Y{-Xz;ic%$e#5Abg(08S1g_5gvYOM3^Rm=;Sy zDwPb64nr?Cg(Rx?$(?nrJ@$*CLtiX?XPD5Rgq;yWXUrCJTWv{7CBLm$@C?0-xMOPB zgTNRPtabAYNZ2+Njh()UKz+h>k(0?K6n~PbFab6WdGve$M1^x#{-FudGlzN1VX58L zgM%}J-o3K1DCmwBSKb8t%Cw`NbY`zfSfN@Yoda{M@Nyx)!ZJ( zUd}%J>eQG6z%<~bb2xBchjaYc&brCz&T&8rHSO1`xV5)0OPL09T<)?o_%j3i+W?9* z1k_A$y*;b773|OD(8%c=LAUytU8 zIJnt<0qhsPtJu97k4#?>i|fiAHC4up>N2$Z`hXi9Gd8!F6X*`;Dw281(#zq2+}k~% zLdik2xiG+4_5oylrz|B(W+rkXV*!6d;AHDE&wo(44<@y32gnDFu}GjfrKuBokLIKB+Pi;&4*7A1{*+sw z7c|njLq$75CX}O}7+R+b=6J|(eXini{xlU4HKK~DcB4KGIo1D%$-S(|H-~cv{)E}U z5Rqm342qZo>7NVWmZlClXpZ_PgfQqPJxUu~w!Jp|;o-4|-v7d_kPQXmXJl`7(rj`sU374g9Uo5piT7D`aYHvybn6}v#M92MN;WXXNd(~KRB-q5w4FRhfe?qtJ@1@1+Pk!yMe4^D#Hd@Mi z^xazb`qf&@iRhYgTk!!djXK=aS#Iy)z`&vdAFu7oVIoKsllF(yE7LNNDoRDBzhenL z{_5oo$AJ3(9WU z3o%28$$@p@$$?nk9nAEynEsZxh9RRZV%O(mRz$v1ftG8v! zdzP=tNYJVy$*m?ld+paC_WtHxs(FE`8LYm-u@`QQop3)KNV(oLH&q;X>epv7t+JG; z(|3cEz(nvnNFA_?1F6q%Vo=y4c2cwH1pJTgxr=)!W0>T@V@TvLRea$|E?N+msGnA9~ zGy_t3j9<6QCyC0YR+q>8m(sl%SPgB$jTSRiM{QNYa6^z-ZGK3=;U^JZn62~uZNKS> z&ERmJ^)k+yg#J)gX@N!6v!SSi?E!i#%b>)Uft7VfVXm9b0jW`zdW(@ey{IbKl(#68 zwV~L*TMqgeKQ0Kb#YLb=Dt_Zf^DU~{EMf3yNU)=KFL#HqKsJw3^U+35;rUV8gQWYdSvT0{Q>xf(IeJ=M) zpUAjqO^+#QK9CJRj(5zHvII7$=$(xG`dqo87H$?F4lLaeZMFK$KUsGwtK+wwYo~W# zA*|pgl4VNoApN>2jNC3ubyR(r&-OXl5P1iFy$&Jzgx1SI=6Ny2fVaPbQ;rS#9>;x` zrSnepk8C#-UB6P%3wxa|LLLLe!aWT!2*e^;akk>uOOH_yqOa2K zCkel9B+T!7qZPY9k9fql5urGLFAP}eqkd~|V~8Jfbj{~}?Bfxj2ubs(6uH{Z+@`V1 z4b~o%P=!$p{=@?@iSHlQa&BK0`i3g8N0FOvxJw^!2lVqEz{p(=(VQeyqBAXqR*3-% zf^^kGI~Xuyyd^Sh_<<>*TdkX*?!)c29TErRb)|E5cyo?wEQ+sQKKi_g2iT&&j1=1l z58p4E7576%JHzFNQr^Z}Txl^E+YV$xennDS3xEr#bq0)~(r0|PTLCBYYoTrb zE$nhh8RNSn@WN}{$Kk-5;03;Z)jvQ4TYULZHHN{1iN5x()g{POSZSve!6C7aT!at_ zr#TK53pS+){z|#YJ1T^pn(^pBsP{dL+OB$JdZn*<$WQCh6*VL#bo&<4a>+}YE*hty z4U_@todyI?L#`U=912SC1b&-8@^Qe zvi>rN89?tPBWl>u{F`p5Lv;fXLiP?Y5db#eiOCRbN~8#9%E@JII?%~^8YQ$8V(pgw zuGavYr$Vpt>A~B%z_iEx$MD^uF{FvRca-fSLQXp2`tY)OO^Le0mQ*E@!aSsmrNv6jIY zNa)%U!|aY?bzsVCVea5v5{H$eHK8aQJ$agU~zoZQ^PK*gUxSprsx_?jVvC!KQhb1%;=7}esRH4*}X_> zJ&S`t&FNHf=-Sk@aJQa7wQmpiQHGu3;91jJ<1I<>f*Ml@uoTP{G+b||=z9fNnQpNm z*)_SHV#M0YKyP6RIksjWg`^p#Q8yLDeY{-xb!i&x^oXyy*+cHk0l~(YF(FgE2h&{s zQS`Y7&57OThC_Vm%iR;_$3SbfkfvUr@5X5I>+rbF;aD^SxmTi}{-yt%VH{b)bMM~S zSf2OJ93Lcz1>4{RlrMN^Q}8g^sHn1{m#yeU=~|aYPP=uD1jm=#@WM}m(5BL82m5*K zx@HFc7I1mRo-Jm5aXKF~4xJ5AwXD(FB>ADFM-!4|GFk-@2eKPXH31s{awYOo(VD8wj$wR4016$JTL>y9#SUtl8i%KS9zOdp zGA1Tqs*n*P-(aUOQuqFa3caVgZb@V4_Kr{wrw4>5oUY=S{$B0(ta`(1P8|OIxiQp` zn6fzopt*&q^9Hd8s&{g1qcmI`KatCcr0E=5?gu)Jmy$Alt(55_>tVZUy-Xk*G)R#G zS>X)zTbxc@(ijj-9BEeqpG*{_NFrOQ+bw`_+^yR3WBc}vtc`V2v0PQB#o^(EkZTt& zSYND$M%)=r0Lls9HHcYCc%Q_dx~K0z{hhL(6~ng@`) zp6S@Q6n9CPV_MuP?JJeH4VScL7qT(g{UuO$6hv1E( zkaLH?k$@SFm#|!xv|H`^I6Qb#vn>YXU! z=2ZWJd9e?L)QlJ?EMzTI&b$`wN15FU+~3A3oY{TWy*X^*R-mT@H+$%giIt84yZ;ny zb6Uk~ftqBt`)@EQbN!&^ybhz)U5)^hV`)tk5+59Cq$$lf6#)f$isVty}lT2(uB#s(twf7Z=nqHmaD@3hhjnV|8Ysr zvj>sE<^#UFjyo<(P(H;4i}YM>dnfQpV78FJM@3QlLZYe(iyG&!c?&vh6X?PH;;n6; zD_**C)_ee;GQtS3scCy%uyoSw$T_E^^2_J#SLt--*J$zP5Drm^tIL8@(MPS?Ht&1-VTi#wFWT?QMB_rpjoE}(SUwBSe|;G@S&*Wz?M@*;BS-En zRd&;il6q(bd4IvaGRRL8$R!eZsRnF%?^^(_{}T*L1+tA`3;a ziHf5j?933%1>nL2YY`?b`Ot&uCJ$3&FQNz<(fB=170_Nz=?h)8S@~H|5X&>9v)5%Z z4GIQfFUMcXx4$wDWMVh-3lytVAh8x+-{7WKjWnm!_o{In(M*{|Z9dpSp5#C~4@K!f zY(?CDgG#>^`vB1=Fjg#XY1IO!k8W#kjd`T!E_%c2wRRJb#Fa{kmVT$AP=(rMon9~W z*M!}?qG3b}!;~HI_X>1-)xg}dD84S->mqK+^52-B4_?1lA`RcheW8%u4H*UUb!m4Z zJQal{t{2k#A^ws?dz6Dq{b2f0s+Hb>WE?mzv3tTYc5XIiy)X9p`xPQa$Uu!cA zZPSGXH$UdN82uqN_z9~~CIQdhUhC6<)FU_c3WOYLi!aaY?}siyOOmhM?8kPWt-!&ib297zf5(;kv2V={}h_)vK1ECkWXM zZH*t$(4F>MC(5g-9ZA7nCnLR#PkVY^_mhA@GnJh+nn}e^V zX(iHcBINP@mD}&@ZJT z^Bz{FG9iPY`TAU$5%fNF9JU)Aj|Par?TiQjvE0| zdYGD}`ZWLn==4?3794XHtlTUcGRy=IfHy;qC?`&PnsG|Om3{oQBxY8QmW9OJ2aWdk zCk_`_20Ib1KTR104lXMQ>HwoAQQm} zsKI`c%<8eyJfq^XOzKgUq}lXY zdiaVqO$6HSMG(%Nzr^v~g^(HU5avWhghJo72)o00ES{q9@S9x?%Mv7HyO55LB}iSe zU4r`3afBga`)A1n2gywDyF=|h{Dk=afv-d54ffE-b+b+2)V)L8cq4G~()j;$RfHW( z`NMo*%bByGjT-Nk24020b{6%Ff#t!wLTK*YnYOY}Ym6>=#wT-Rgny1p_o+7vuXLQe zgM6<+a<*QU7*?i+&0H#-&(m=PS_%&g$&h2_6Wh>wlCZ*j*B3{huSxKB8J0QIChX2X z6Bd}_5XFjnO7%ns)UCj`(yK^wHK(r##>qC8Jj4@U<)j1t4+nEQP)^rP_hTNfiK9B5@|C=~_p#4za<9GZ;=16DF+&pIM|oEBG{@|Y zx4BFgw~3Q1EGpLLLpbqN+Nr*Vwrklkm~7t5ic@?kA3B}>!VcbafZU3QH?lHN+yRka zF1Kf)d<$kGlZ2@{$i;n^7!hArhP%-C>G;Jg--pqH@mqoNa+&~t2g`c z`ek{88r5Jom1pyExR@3eJY~ZEd{v2mnljtaMQ%%a{LwQA)_<~!g4CStvNl=qFhx{v zdER4+?E8aiQEp3R=1HpYVF}l`*$>#y3%pUC9u+3T+y#-*Qg-IaH)pJZGmp8VUX?wD z$nb)e49!>-JM)E$ZA4#q8}u|hzHI9w4rNsMrgs+8YW~^}H2p$x5;rnF&&oXb)&ds* zAk<;^_yyNsD-F|qAZ4CaQg5__;bKXru32BXFD7|?Q9GT@qS%lP$XLTBU(4n&xzAA9 z5e>DTifD}C4I2km)}=d8%7a~+QwkwS2&E9R4A|*2)AaBop)Q6gRsyXd42xgacN=51 zW4YA{FdA`ReP_d(+ia=(*9UFXDF~aUTb8ABSNzT;xbICN@MtyPBY*ccIoOCeVboAT zH25HipUGOwL}<+n3FaA@&gZLO_JTj0zIkIVHKU{+mj+>`Qgiyh&|}W}_G2I72#4zf z4U#PQbswnT`e6nZod+C!b+va!UuHIsWkq7ia&`jD%sUSK(Fg9*mfW@XS==C|@sW1X!1&}?u4jbYUarXJkidxQw-gyUKetcnaKi95^?(Ug z{=nM$pEOeIfk<_D2J)^JY0Xg!1m5>_NitdiALrQa?$tShGUYK6;8agEtKxLye>|odey3Y*p%o2s&AZxeFbRiZmbKf5vpt=fj4{(lNYI@inVQ8{pr+D? zib^Qo9v47BgE#Biph1msDd#XhXKe8f+p>f0)0h|*3)*;+h7W!OfQA$DD!b#H8l)qo z7Iq(eE@w)E8wSAmunYE(aR~#=A#lH357Vs_cb}n5BIC#JJf4Q&dqw5R9l}ooA;JX9uV@RFh)FVs<&f54sSuXtr6oN7V-f%4K!6`Na2I-oCdlCeQ?3XJk zmNnXsD;>P-%HTI_r^SRD`;AZCG5AZTXE!I;-}RBB-t^w%;zOwDSSy%G&U)b*x{Vu{ zhX38B?Z^AHT!xG&bksRix5P7YAwhk=spI3#;#j`25Pu6RZ4a#3Qk9x7Y*p9xU@TV2 z2I)lQ7O#dW=~~?uj?LI3TC%*0Lv1BlJbmFfnVYZ46uZYPwYmA}(lMtWL^Y>|q$P$` z6MQMiS{Q49b1rU?Z`&+U`=s@>L-{lQY;uIf9LK|`(1t_H&(HVYIsEJPSG&yY^HP0w(FD&8Uhcjb2j-mSAjKVxy$w$2KNhT zqDkj*=?K>~pW|h>QEQG43wE=iC?hJ%&J>h{YVXixGc?azyU3%xw%KSqWvXm1)t&vb z2ux4vB+jmDW1L}ox47B!*pq@nnA>RCDMhcHW=4mc1QxKyTSqjKDxX}R{N<@}wZ_}+ z$FWYVX*cc*m$^NRgthoiJc)Qm6Z9nc!RX$H! z1-YHl8!G*+04BIEzOz>3M$?3!38Y-$Y|bCi@D|}TjoH}|&=g=CV>1mJNn;9b-kN>w zCUK}K&m(bedd%S=q4{`*oKAQYIRbd!px_?orj0_MLn7*e^S%GznB65Wqpt>2J2_4n z)@VZLlQDi#)o=Wf5W1IHzfmgc>C8ZxEd(9G6?bwao79i2>OqegXuU4f-;1_KBj}4A zZA>0>FY1~uv5}sPty;^pfbH-<;cO*QX-8iTD}Dpq;DtAl6Th)Iz;*f8CF~npuFuMo zsR9K0A7V?ygrk4tTBAL`II|oRq$`b>KpyQM(n_+NRJIsk8(zJ!ME}?^vi+V@wde93WJ5g@Aaycd-eyJQw8>`sapKW;*`pWi^ z5C>nAL%uMQFdH(Ndsk<;#_8N8tZ6aQIYD#9d0zqrWrY}>=k`I>$!cw}C_=<$oK~C~ z%b{)Ih|&yeWFI}BhRvrI^lYDW?Av>85}Zb_CR&)T=heD;V2u9O%+pmfB<$?{GCVxZ z02GhE;to8LZnTX@<_mdLSMm3+j9<%~v2BqQGS$jsteA!%;`7^*%CzJ5S6bM;ngwjk z5PfITlrVVZ+-tjk?2M+ej{b!8pzVoIj{CP5MfyV_gT>YE_k)A;c3Y;5h?c5g?k&?%Pf|@`WteX0LYf$F93L&+e9uPB$$_!!7y}HEXhs~5 zhKwUxd+?+n;#WV_U>M2l2Vq~3a*?LoYpt%*wY#rsULg!xP4_f8#(zKvQb**)y0c?t zuF2gg(OOd$16QMPCD{30g`C^qe3`Y^t6=qhUZ2mP%1Y(Pvx>>9%8f=0cBxr!GDH0+-}|Y4p|mK9xXlgi&AW5vtT{mTQ-%yr%1iFsN19LU{j4JaU?_8=e8$Y z+#jTFjO0h1o#1qBYXtj?D?jUkP*5sM=<8R|g%7I}hTljuLElrme0Ebx)EfWLDMe{Q zr~$4o(Bt&~>h3+Gnuz{>--H%GiU=r3Xo7~KNFYc;2c<V+8X#x_YNR!_2jQ?kyyVi5oxo_{i=e|g?GkYd8voo`2m+$ZUsl2hx zWo2JUQ%-k~IwaSw zvK^aLqI0W z<`XuOMjUiVU@-UM?J4}Fw%q4>%h=+Nc)mm+Xlu!MKS+i>dcIa*pLOGcA8T zvr2tIdIyB{dyT4}p;TaD1;&W*;D8@)u&6-(7B-!YVx7+O!D^{R!#iVa|%P= zt(}-DT+D9HuXl>np_dVqhv(Xe1Es4#z=L-T%^4PxsMQH|o}Ho`mOJn!=K~YH$mb~3 zp=ks28k-LQvHpgFjSfBh%COd^6p~U3NS0gCG8mt1k`!<#$>1~Rm7KF({rS`u-jYD8 zi~2}u8@9*aEN#(U8l8hij%(d=ieezdtGEb3v_o-A$Swok+S9O<& zG}8KY5g>XtkT)??I9Zt`%E-nn3?+2EKc~1sc*vQY&^}Zsp)TsNs_)=*)tiFoxK;k~ zAe(L|u))q?g-Rg20y#1$ za_9>2@JH#|v7>P;H_vu#j1)}NvjLL(h_krOIlIy}S=M1cFXYGk`PB?BgHq->e-hu; zYRmw|c6Z`p`_BMg61 zrPt=jAFy4L!a^HHw$+HKItcz9CYk7d`Gc{7e-7+3=1X>wWoji8l_eD8ik)F`5L2dC zq;YXkTJ0C&+$n5+@>c0E?%NfHm9TKd+3*+>7!n zqPielF>jJ75yLcP#Bk=~!;F|WG(jvo^Jz(ts=N7@c?7%}D-6_YwFI6LR(|l_i|9gf zlQG#UdH~?*FG%5YDV??^*Bz1n%Srd2qizzo@sYp||J_Q5d!*^fS)xB}?lX6TZ!(oq z%ALAU)k{>G0!Dup>So275qB3O867fk;n1|SeA?3*ZSqmXYnjYd?&u|%itM*uMw^SW z_$=RHM6#yN7DzCfBmhK(If|5t8I>3@o&w&2Zry2a)o4XO)U0_9^wfra@ppg6f|4Lg zIIRH#FjZN?=X@Z~4xkLsl8xC#tmbfshmY~@m9U}KheerR_e?ftVQkU1eBL~?M7yVq zc&&mbMFJFmh|;!_8R>G8|ZoB9#7i)QTxrGbZ=sCn1ees75 zZ-Vc3UZ+JFXN(GbJc}mfBT`w`I0Y2^$19*a-x;(9Ju<|{|C*H1lE!rkCvY339+#R% zvw%GUSM@eixDUuT$J05gnA=JoWcm%2uAu<)JJRjS?eZJ5pwFmj3w&PECICzz0q5(^43tTL9^NoOoF*&plNpkrJ zo9|aisW>nB2r_>nc9l8)HSy3iN?LY{qo0Jz<1>v~U%E+iUL)VnCwa*LH7*Spa-uH; zCAQ{n7jOjsZ2HqlTXlYyj%9C|pvL3fJGonQAOvM%3rShUr!w3;Fibz2KQDd8`K>EP zR;qb-yXcv8`#FJ=VwZO_5EL+gg8If=L>qt8!i zZVG;mt~{$MQePNpb&iM9B`=gj-YhsvgjV$|b!x5Kh`>OAokbDogldv*B~TrHqX&R{ z@Sb5wtWHob;co<_Y1JP2D5{cC)u_V`5V1&@@txbmepJ_UtS>69e4yK{l2Y#hx{^fG zVD=eG%mQ4zZDeo;aJa8%Hc9cbI~B^}jRp15Y#O-0Xce1$4B+8U9jGJun{`hbJW?LC ziS=MwZeJpL8Q?G$DKKLf_(qmar*a#fW?VWr?OS>Gp7QJfiCX_(8f)3n_CR% zKblge9w4acG3~M*GZvHA2Fxd+Xm3YzxDyTgU{fb^8nheZPmCmsoCT+5s0#j zDX(YjXULd*hiA1MZGofBZnInZahQcZ3)902x_tUpiH};K$ruFlSP_GUu%nKC-srfK z{akCA#0cXQA?>&ZlIF=W@Mx=}VS+vah-@%F*4w}Mi-ZY~vH|F9NQhW%9#3)m zzB%u92@QArS}|=Qy*%iOQhV}LHt{;Ut_xhoo5@fg^~pZvxbo(Ix(*SXaq@DyuvqCT2Df153=%2QgAjnp8>X3Eeh8SMz)ZaLVpujilwB%SI=Er4JCXChQYCO_I8RWIr@!QvZURs8RJ2${O1ybbDl04aEq9767hx*_R|RK;D)D zK}1A1I$^~fLZOf2NuaR0`9qw+o$d!(k2kb7M?dl$8aYUVy;YuM9h+!B_~(YSPgM$< z&0KiIyR@pA@KiEEB`a0v0boEkD6#O?qng*0WexB=douQJQ}}FaE`07C+;YY-Yn>h3iHP5i9_jp3=wUq|sia%mS}%Kbj7# ztw(gqtkP){wK;v1W#=uUjS%UAT>SxbknI@15gl%52PLLz^SDtSURvY1wr^5bGWW8V zroNtsKVTdpepLnE<@Bniw~81SK)Z9-$#uZ zO8*f_t`Xy00TAn;40D>eblNbnmx`q?9~=`$nnU*uzGf)@QfR3Osta zycdb<;I`vAt4(LlujAdVVT7ia8oY}i7su)T7k%YF1QxT~l|~ezv48osbCgNzXpvv> z^QclnQzburfuK{Z4t}xu#MKL|6u(g-%>>Fe(H0c5T7a4x2kl<=&|(-U124yfAZ88F zcxm<^LmhrLh+wRox5XMSFJP=^*`lt)IL;Nt3ounny7_@PWl*R>AkvVDr@M>(Mp*^Z zwXVpkqYu)T&>%YBHeeXovp<;?0fz?9;P*IG{U@ zE}_HK?-aDr(m|<2(;e0zg4v5$JiQ3s#q85&1ZqCT5%e<-)q^Mkorm$h8FtDFMqM0jaSuxUD~z9CWiybZxKm_8q}j9QytBb3x`Jp}d4cM`eHKt@)-a72{#dyb; zwn)7xnE0RV!I}^5N2a76Kg{~GLSEU@(oxj0k;GlLYZoD`JtZdD-Riv<1IU)QFczi@ z!hwr8!{^Uz7`sd82Y9WoRS_DlF+}t*cS_eUXJ|cF6XB(WY!t}Man?k;YWa$Y__DP|maYj*zpAn*KQE;vw(Wy=8ct^xDTn!-JNk(f}gVZ zqqNpeQ=(9DrjM4Fv{qDP4^1rZkUd+Lbet8RuV#IUN+shy-xR0Y+b%k9SlZOkuV<>^ zZA$%~NU%*WG0Bqs&h;|S?bY+CqOVI?lozo%X$OI;M!XaFpsfgkk?<(I|1B5RHkrYs z+H(PlL-Wjpj{^1QIV*P*mA@1Dg9c?!#C0kw?t#~I$^AfJc5-O{UbK3j1gE5PiHIi& zsk@ojP50>1)2Dh`yr-cC;ragU*{@qka1x&m{d15#LH&w4Ot_xpK!;xw^-vBL|JtA? z6GWl0?f$|REdFr<6oouA%6U$+^1A~%BzwF9Mit*3{pX<4)Mv4V%D<%;-+i7pR0SM$%Y}5OU`zrB=5uZm&9mGl=H!6#h7)K1aZNS_a@q}HqTxO|4sKbj%rkl& z8(Z>289u|fLipuI4t*(8?~CbHX>@bvm9bhrc#P$Xz%DJSk>f`p&X^rivDG{3Bq9E? z1&FB-?wR5kEA1&IV-YN-8`%>#3zqP?-5f5{%rW$)DBWUnw4dwe{1sYH%mzOSVsj z+pV&u^o@~J|3{K$Rl=%)c>ebWP9Ys9vKm`AVF`W{dh}Wh;}Lsbb*Db%cl`bflI85Z za*TgB_$CRR?P?qA=l&5o?BDGm+5imgoq7b=$83`}Y4nTB7{t3l_HCW!iK18eU`slA zn5!|T2S* z8|zZbt9a?GvZg%8LbB1h-bw06Lhx)*e2^4fK9M27jsC3$JQn>P4kHMsJwnbJJtimTq1^3b7F>mnpCoMSmJRl$UIfVsfU? zC5QD25LTl1t0jQsV0$PtX?0dRW@`(^i9~VK%)qx0vxUM|`>118)o>BN*=m zuo6-nxTyrS)*FCs_zrE1i8(k%Z8k+d=WxiqXFjuTUsDmzlsI4$Z|OCk{mb1dakxg# z%fJVdI%#jkZ4Ocwcq`mJ9}U_@R1B~~=Byfy=@nz~>n9TYB=bmQ5xu|z=Bi%H z3hyC6AkTZud(&sE8mQDSo>Qp;?J6%tZbkgSij&1dTbNLE!`%?)bMg;T zhFpzzel%8wrobr#UOdbyOMM24CWmoFh-1KIA|mq#R{ft@iqPE7 zvX(N0mt@0%BhKvPgR)Mr`=4XPmvz+pZxT=z3^c^K!EMM^uDQQ`0#--hsi59Bi<`A; zhCEwZH;T!yWxck8(5@6QJB3qUWjC6z;}#bb1f{I(s)^_BR&H^H8LhYN6y|kFFwkw{ z;w=+UpoO;;uH1S=Jg#meB?oF>SFT?ccCNBV^vJ)!P8~5|%6W>IX-Xlo_m$BDhy{Pu zSo(OyG?MTqmW7>y-$698%UX4nRuCvaT`{07G;b}Jchdd=qzy+G6%Hw)7dY+a95L3G z{$GZ%$4jPgq15%|n+}2v->0JRfa~C81C{E0P<+}B$tvlB=du=x)&;PVu&}eW2$vn$ z%Xeuzcno-;fUBM479S5xuF}XYE$cD3(bvrBDG#$r+{6VBlf~g2})E1%|h6?B7~)ub14v8P-G#$%SQj4CV8at^AALJIJij&I1|_(W<$< zZD1I<4CbyT>()m7J`%QByHblg7!W)Y7Lj2kz!XYSmJ+R|uz&4%GB#sk;X&t(8&BPe zV)oLK4I4?vOii|{e3Pua))jS0i^{)Gn)I8QJ9+%g&xl` z?!C^xjb6NvSF3k8NSi(;5cD;fK8nu1I>{+xq{;1-vrCfpa-o%T-KE#w5;-dG5s1ps zME2SUrx%nk7AERQT?S%o!IbK8S+&Sc1i_al$B7)R#`$#tIY#&aa_|-5j?S69>M42| ze|=c4xk!pZeW&8HDl?BeU06hRISNhu_1d3wMQ2SGFo1h9Ek>7ZG}H(Rkmw$>V_Xyy z5t_@2AjIR8O=-kebmj|eN=-pbACU z@53@3C~mFl!+ycCKBR!kWYi`D5jjvICosu{i;2kauQ=liU_w9RHmjh6lirx02FCoD zspTd0-zblkl3cj7-v)6=*y{y;bY0Cu?Eue9f(RST_bZF>$V#7Se?%o{PA+dTEOjEQ ztQ^zlOuR#zPsFEpz@3#rH{g2qBE+A-NLyLKRznrxg7=!@WVkZc+K9jo5q=-0Z;!`( zh47a!Djfrc|Cg#}(#ZTvsB`pnw0qG%4Z00w9wfl%9NYTK*FwBH1fLxo-;+nQdb^*}jcuT|5 zHv`)R-K;SJnZe@1tnG@Al$Ei%LTA;FmquxEynV{pNdHZ5_`2TJD^>#00Onq!Gsv`n zh72ick@Zvp>QjF(*p?#Lm%izY2k7Z!oCv@2Qh~Tt;&f3tY_5au?9-W@odU-x8E{`P z76WsE24ev=J$QLp4f2(dM>SJux`$cqaJZjmL5Xsi=+!(oG)ns%U&=>{`O`c0I-NS!bP627Xrj!m*kw){ecjp zq(!y!rw8-BH@m)9Y9dCtm?c%dpE>rn2X*%Ue9D)Zk&fx zhk=hbuZYz{7 zLQ)i}Xm&3kd-}@_G%_Rnze>V=br_gnQ>i=hpr=iEdURy4g3EzcxYrc$wF?=}GWXYd=-@q|U8{r7*iTUhT#+ zuaG73NNO-;yUcTS5uQDDB5u&)p?zGlgqhGf3VWM1VM(NkpRt4Y+FyTmjS@V@R2a_z z(@(OC_Y|0UnDi#Gs+e94hF!8)=s3Hp%@yj|-n}_37>TKi$6Z{mc~!}VL&KYrd@S0n zpIapI4X0Vxfs*;=j+m9ptPO9VSzCa6UajeIR!yAJO-2Y@nRWJzH=FR2VKebP-dZl- zp2A2ae5hsTU(nNVXlTikVXYRnz&OmOv>JTuFLTP#-(bP0$6JlSjzx{^1#VN9b%D7^7UVob|H$idP*uZwQ7?Gh_Maw= z1@8X0&+Pkq&x9}#m9)=8)_6|kR7Sw|kK@>PN<^8exs8bC$R}D)x4`6)l>N9%@HzPf zv(R)FlFjmU1nOtf9~&6Q?QTmFt1nTshtUQPqH z&%&gg@CWP_y?wrwx!XR$5bp`jyaEPzmlDL11NDWHiifA?qRHn=c7Eu4a-uuzzNxO4 zDp!M+n%^3D_w3uc`*YtiTb2oE>KEd5&^WNgc5dfdVySg zXa#;Ln(aDQCdXNnX7l<5i|-rf4$Sb%a{ONGLtfDu?zq@zXZmqVI&JR-0@~WPE4=oj zG}JdQh@6!2b;#5mS6TPH85bVQ6teT?*`*RZWZU6~1fb{~RX0DXDnOo*1pj?L+4zu? ztgYG5M$3=!M4#~n&9F6_D+1R5*W~FhRRzqx&gCwpr{s}%?AiA@qD`W$&(TbU8dw9pmEnJ4mA+j>xVTe;g3@UCdZjV4Fo6WuNt}`!C32 ziprO8VrVEgF~c}j5ZWky6fo&*ABjTse=%T|{iyU7O=W8ej)*o&jBgZIt-Ia>>05*k zrPV#-&zo_+s!O)K+1;hviMwg}f&a>h_?L?53Y@&P>w_`y*vo^ehLWbOu6v?eR-U^* z{(_Eo0gTZ^E8uoEpH`#hhibRK7PDbYTTOhqY=znVcIvZ`VNyPwc-lO;n!x|AQgo(P zR?q2@8sOa}`TX?zhvkJDi1`OmPOsy~|5nK7i2T(!6q|4R#b&-fd-dM*!P;L?CTzES z4Za&qor-sT>VMZLCc2-5XbgMF-s`ns)&f&?d2-^SP`(76R9(=ny&vUw5VMR z1;TaWtjGQ9$>XBOi>JfQT`GgHlWpu{PoakK_r(xlJh@!{>kCwSA~|S|^G+iz+b?oM z;fEscz7Czwj9uUv-ml$F6S)zU3D;kp>CLHziKiZ|=-j1o(G^Kr4rD z1;v(qzd-lvRb^P@wA<&XQuQLYje=9kBl7YPkr(5JTeeDUO;ZlFw14pR2Q)r6_s%AF zTt_2K06<6wiiVEk@H_89FW|Dwx0T-qNJzeD9WNbu3MpRkHS~gD0A_S4n;s5hd{V z3ak>P|4&@-!=$YO;0lU881nS_7Qdifvw4OeH_;kie@Jort*ol>tC=j=HJvHomObs+ zm;A=7>YM2EYBt%Og23c}KIC47sm{r(Rt1f(LCFV$s}TDVdGWY>f%y!n*-3eOEWfN( z7pD9Q_OTY^j>je1(rv)EYH;;u+E&FqM1xB8Ne;iv*9iX~fbGWrvV?V8-M79d#;o%IwDjvf|lG3Ve z9UfhomH7(_3{?mn7wMzDzF_r(>*UQDexXF4z32t~DZF5Mh;KiGn+>PBT~M-i^%;I} zmI_EH1ksNQ{ZLrxLR#gqnBE$WXb1|TT=^vvXxWvQoNh6B@7eyp&VFWyo6N{%E}xi* zBMmOvAcA#uUUJ%lhinAYXqo#-sdD{PL!t6)S@wn33BBc1$>oybf3F7YmB< zyp7wL?i7%PvE25p9qvR;6~FUxk(2f!0rP~Oyfj(Bcu+#^f8vP#pZYZ8MV5qF>>DNj zc3q9jaE{SgF%?KJ2a_z{T8w=t;yO2m@9U)ug~$!ckuoY&KC7m9Ma@U2yi$hZno zy3Rq>$be9D{*9V@hU_{OL6GV@dv(yE$&Z(&U9wt2+73*Q#REx zgaymw;2nd^fEM?u47W9H3+v4S`XC8=ERq*4eJp-{>JD<$g_oOKbxew}`+4yZKlymm z6@=d|qzrw$s&l4s&c%DS2;=Uv*8EDK3XpI z=hg~(wKEJ|IIdvDbB2pSh#%Fz#lZFX?kWV3lUOH`fCkY^p)cE0!9iiiUKL zO#U=1-L4J-Ql`1!RjqGpVl)yG_e+)z2aJ5$au|zC{N5HV>u~*Nh1Xg^#?nAs8-tNC zJ@3-(N?@U>X{WieI5UeEu$lNMPmu(T7>$BN@Qb*VEXx5pglW@-p)|XHi6bs%r*FN$ zF`&5vXJg+isc{&Y(EsEAp00w-^NJh|WeGImyqzF*O+B+<@f$@Xkfx@7cPZx( zUUJYRRd1U#p1yQw2wX*d9or;kg10#Vw7AMJ;S#N0ZQ8M*ED5d&G^4qdtH#r-=F69# zJ;kw|^O|q-?BgMcc*}*@)pp;6f$J3zUCYc*WP{g0%fM!l0ac;wdz+-BdqAkU@JuTI zcPYh+t@Uk-9A5R4s>H(VhT31y%i6!7y`DGl{ixp4-R(ny6hr=7p3@rQ_bM)!4~y>m zw>L4TBkw|+ykyrT8TO-^6Mv=$I@Aizz?&%UJ=J4JFQ@6`%EH%O>7lmbnLI(|pq z91a_Bm`*TY=U zlrBzTigGCcltsWjf|tFms|iv>cogKNM|S>$q1)c8CCipk)r_Z^Qw6qp1uF^5K zs)_H*tJZ%G&0-&VI{bz6Ki_K@Ug{ookrd#Dx~_gU4tB}e=#h&S%SkHLMFwY%P0JW` z<`aNF-el$W$<~3%%QoMu>@AAmyb_^dE3ir>hZZREsy zD7cf(`%x?VTcEl2*{1g0y!ouzq(@a_T5U7eTp!ZLy^$= z%`0C*qPe`|$a1@ZtxY<%$#m|hiW2);`!3MBPS3;Yn3zGEs-D-1_|R_|*ji%f7uox^ zXT!d*lMrRB(_CfeQL=Qv_{|-cO`w1I^}FF6_H|Q_1m~zZBU_94<5!{eSk31BS=swc zIu?J%5HrHov&UkU77LN2k3endl(yn8xb^kB`cVSPUiX>yhH>SJO>i8w*}>FF-0azS zz-%Lp-RU0Jrv%*Az&JA}{RQEsNp$yuHP6#Rx)5kA4wn8bFH*DDcKKmS+G4` zS;Lm|!3-bfU7D`&(#Dy3jKdaCKtqNgzUkioMGr95^rG7?FiO6Y|3ZXc?D9K7vDplJ zTTA!^(H`JGy)2&;=VD&z0&1CBhj(2}h7ux9A1G||E*e3`>7#z-v>seexegOLR-H18 z1#j4zP3a5~9XUptwC7(Yh3RPw4Vgs3*`OxTmhQZr($IaH+)2FQf!KEiKEb2 z=N2W)IcNW+JF`iXXDw)gB{r?@f%h~OS;*&K2|k7_85{|!2GxETbnh95tJT+hJWwmL z9qrLq_1_T4N?BiN&u+$Tmp^T+9Y>agOn5FT7&)AmZhar1l%e__Q&QevGvV8Iym)lK zK^fG0on5R~Vx z+THhw{Y1R}lRU-cqA0?51bfSxVN!r^_3ha$k0}1crq0bvvG2QSlv5THzf$34wiNoV zThRB7$&n^JX>#y{BoMO4b(cP-og^S#uJ3+5yK(YP4&@RpZ?EE!meLI1@aJ$|%W_Fo z${=vn;G)JB@svCFB$vZ#&MAJCag#ywyqclAIEd;OyktLgHvY9-h`u`G-26+)eT>Oa zEKMh5Fj4ynNTk7L`)>ClW=Qh)M_-!Ku4zRwW4%Fthw8HGNckcpyJ%>|7s3S2x zdFPF_hskt?dNB*2)?mDV?m=NjQkDXE;K$RJX-8^oKD`jbAALZULCRyX{3EgK8PMT; z;pw3AGm^q{W&Mi`PX))Wj@scv5d5O?GbgS-#Z?b@aJnZsr_-50Cge8x&*`I3=Y#?} z>PXSilcEaH;V@espkNalKnS_3LfwJlI?is|H_KiHXHCay`9h%%f?w)Ujk~vC;?~@!5m55y8*#P!`Ik{p&aNAAb9-FhHl(yA_`IdRLz3v+3+{ zXb|x_MIJB{z`xzSat&WC_c|grujnXfty{@NV4sr+to|*mYY=^PvhD{@iOQ7h1Xt@r zJca~u&lDyhpDZ-Hcr=NNsHe`hTD@Nd(1_9Kvn@D_lXc9Y0Tkbrlc9G`f zeLIJ7D7I4Y$;f>!aLGyOeiv0|{540o?(stz0%$-XS#X`bEBQz*xz9C{%j(Ru2kFRl zE{tf#EJl0Fx|o0Q!g3Zc#N=$G*)}Lot;Oo^i=M-9zt#k~Run}Y^{P`ouI`VR0aGlP zRV5Z?wmkU9PDfzB)}Geu7p+$XgTCgy2rC$WCh`HB?XjMmk*f)MaP%0Q;Ira#p~GKJ z>)!$I{vZxV9>6Ljf&N~zXJ1f;$`k+-&ck0nX`b`O87RC64SHXk100!<@0Aqmqkq3) zedKv$9V_TA3e54@@N`a79bja8Ql&B)5@9Ex&(H4#-j+XBBJ2nJwm=QFTQYTWWN@ z(gXpHP#~1b4pz;qi?OT|iF$)mnNrVr8EZx>C{{vcUA^1mrzK0sUd{KMq6xe7B>|&c zg0BejF?^BYQ6?TMv+K&Be<{4#OqUDqGeq=|R(5T!ySpNGrNg_9`+l5Dp!`KAELTFc zu@ewDqN(sIL*C@fIb%Y+FJCWB)iro8q5(1P&ky)77DyLCYzKy2av-m)m9cujUlKUIm_E zt)REy3*@NwCRb2zFZ7&H0rw~Nt?VNTFBCxc6dI|qZSfs-rF{*xbmRiVGKEeh)FmT` z!(us*!1(3c^C}@5z&h+QsNY=)I_IqeA z_jKcUsoIdXu5K@uZ-Vmg6d<<8TCqT$ejc8`ka0Iaf2v@j(=C+z|-*p>W0^BfL`jMNxh^8QPBDogkm9`gUUmtn^v zM2aC@o)Z^uI>prRyuf8!1J2473E*b9li6MTg_8o5!T)y$SN_jDHs&DX|8!mds|NU2 N3;dt@3jI6wKLGYG%H;q6 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-15-NVIDIA Leads Open-Source AI.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-15-NVIDIA Leads Open-Source AI.jpg new file mode 100644 index 0000000000000000000000000000000000000000..cb47964ee48a5ad936ba1cf4369cafe5f3e4f731 GIT binary patch literal 60385 zcmbTe2RxPUA3uETO@wR-AuA!-Au>V;*_FyTAt56i4I_J3Dtnf_w``JZviHuOna_3K zhrZwM@A*Hk*YiBz@9Wfk&VAk2b$ve9XTCq5yW?NSV~8`#3Q7tH3=9MU1N@IT9z@6? z@NseRaB=YQ@bCx-@Ck`2Nr;Jvh-oR#l2J0!F*7mJF)*BGiHK=PNNBFGFtA+t|NP(aHv~C87Bw~s3*!O;lNn242i>MmCdQ9((G7e6MXRO6#*gjIT~6z_KO^xT$hA|uZmm~6_dLm zuW(aQNm=WTwvO&yJ$=)Mj~<(uTUgrJJ3M!Ea&~$B#?#B&=dEvOSa?L_yZ0ZW;u8{+ zl2cOCzU1cR7Zes1mz37j*3~yOHZ_0m?CS36?fca~Fg`IkH9a#sH@~p9zOlKry|epg z526bLfrTau{D~&)f9N6ybYWs+V`1Y#bYWmRf)5rsHV%sbE`^K+-hJD%7p}a-r<4te z%c&+{71Ug%dho1+keW?soP7-<4W{h>9bvEje^K_Auz%po80cYr46@z|)^RVX;!Dp(#O4_YDX26}`^kr4{4-b5w_=0xhBED6sU^v4S( z(D?24Vis7}t#lSJfevF9wHmM}y-5Q0RP{LMUz~sj#vLPgqr#T@q&NCg+sP!LNl)nj zlL}{`PWCBXI1JWJSn88IGJxic`R|+sStr~z_TFBEh<2-7JbF=ZGElT}8)6K+%UR}r z_>AHXD}OU_Sdtv}FN$l#GC6}&=`A=p74v`dOL)%2rW9i5&B4s7**q}z#s7}|%Lm~( z6g;*~X?s5bP&VRsUSj8nJOhaqjE;Xlh-|MMbpS^{60}5A^=Elg4`a4!H2`Vy52OAe z%5!@k3;{%~0HUa`JFft_==&?gr_=Yvq0Ehda5yF%wU>h(FR3-b9Q`X9|KaG_RA}t; zgnikSBX_!pU?#VJ3ns*Ze~~3Ua1OPxx_L5xN-RA1|A}AHHf7T5!*fUNWndF4^5~g+ zPeuO!JhQjPzX$)-4-3>deJ^T`9_WaBs|((?=jg$XrO6U~8(>(v$HvOPr0KtU)sD*N zm!2)tIE({4mL@sb#LT}p@s};abAUWI*KNRx)ni1<^OT2pF~^91X^Hy3mN9;L80QG4 zB)hpC05;@5pFdQR}$WP4#Bn^KFMYc=lTOT8!9VY-; zl!PSbe+U2FJ+ca(IC@2yi60)qYqB$QXobW4R_TWI0`ubDUY9LbjWSX!e^5IkK>U@|4MnVcf56%fv zo@?s6H-%pxBQ8U>5~k|zsgWBFP*{;(Tix{5DF5l+zNv@v^DKHQXiEf}e1VZ{o8>v!1jBZA1Tg%&2M`W-If6ZVXURzy{WTIQm~U6avcH z4qPSiG|M@Tq<}P*)SXfWxdQ2Nrk-QO9HT9yYR^+3(jYDXnQ@24)}v}9zXG@Mf8)aQ zm;n)J>3_(oIw1@74^`4D{h3|+%FtY4V9E-Z{I>rC`Ov^03otD)e@PNKB408AE|?j8 zvfEkE$p(Rw&HZ~=%A2T`+OAxpPc|of1QP-O4n4L59_uqbOaf+GnoQAw`^Pb&5ndnY z?acxd;Qv1xU53`L-B;ep8UPzr-hdv?a!`@`VtXj<-qDH|J?lV>X|3xCmwZHK&J%+<(bK9&_d(UeTeA)4$c6BnAhQR zVv^W_hyJ2FSw8XTTTj&-Gko+8Re#UZeqRoVsVM(%l9Ro49KQUDYL|i(VNor~IcOK! zCNRdkkO|^AeEkIo2z*yM+reZH4<`FQM0U9(2QVzhhT{q1*ZA43Zf_wIQ(YRJ#yuP&L?%@ocEDI1@nN&Yw#Df%=2eeV8Vl4 z_#C&XzerElKs#Kh%FhxzjGHG-a6(O&0Yc1p!48A|JcZ|b9lMZ_8H1*nLlZv4*acnh z3`jCWZGJP}LA7{7Cm8cfKLwr|w=`ynz82V}(Tx*F0V(FQQTV?4F~XHjI(thU-fbm` z1Mn!XkBpgsDMNGT_jJ|i-Kxj$_KE&dt&TR2B*2KwLhj_%@9ASicERaH%>PWkAf0a` zJv%;+{tF0D-ut849_iTz)c#l}>VYL{V^aEX6`6YCcmFeCje2|}p+;uKvKhTW(3o-CC=3-2#V!zPi+(#j^Jn(K`7Q|FAzcqgueG}x(PnFbD z{L(YvRyeu)sVIl`OCNRbJ73pdH>^Fh`aRt3I{ubhhfWeQ);SnPOTPDG+memFt{ ze1rc0KPe;hxj17neYrRh36vVwaVRx!V|FOXBPcb%=P|g=Wq0~bL+Lt^JUb;qL!KSM zOBsF}vr{7qJXJw?8w2_|UlV#9`oOoq;~(OE5rXoR;fVZj^voDHmBBh-(R*NO_yWNR z#&Cf_2s&lXSQ%gN2WDtws0Sfuya%-DlP`&2hn%elc9taQii;D&JU%!I!Nm!_g;nsf z!@FgNwhnJ8-uG5JhCH+_u(k%6I~*d$EAsVq0_ay?WqoCYGFSjC@)8F<Alx7 zcegU}pi93wWM!HhA!nk{H!dyQ*~oeiSE$1$uXA7C>z9?@&5nF2uRY=B7DCU`U$!Zo&rbn3OfCGUlIg6bbRa(<8Z+d-0JK^vk;sW zI0#sJPVPgKeu(!6G~ogffzR+{$1r_iLUhoCK-|#hCEkv(fH@J^efxa;Yo!7Bd%hEwjQDuC@PXvqfxt0>rMS|oY+=k5 zx@>|j6)}T5JGH=KXqkZo9fO1ty#;V_N)w<9&5VoYfWe8eIPsx?S$iFdE8`+#8Ld?kG>sk#|nL@3P;gXoPwfAaq)36Pu>?bUJq* zuN`n6T>y^n!c(o?R4lsCBH=6|!#Nk#zyzVWwGrYksrv<5EX0FvW>Z! zBND$xMTVggzK4ts=JQ`zzW%~D(E!-p@DGgULy|7Qbf|J5Vw5ZRf>LD?8^5SLX?es_ z^WB9GktEM@jaHIN1gk413BgM%75;&$eNKv?w_F}?fS2X&QYpd1G3F$MU4Mdb7cv^7 z&z@e(QNlvF^l;W^)~2__e>P+U*qe{VOx{A)jI9W<&@c(jvw``D1?5|>JN;vn5v1ms zudlD_m6jj`yv4H@ z8o@vjxddD3J%K;b#Kl{j2l5josAIoK^jTNKtD!?qUC>J7I#p^0ucK_$>P*5muRP{= z-KX45tczNAcTF_A?YTc&B4QpE{XF=@#F z?euCSJQ!L9nAJofgeIN79BiKP&0#yQ3E9ol_1I6m6R>fbny^K^jd|O`f&}4+A{?|X zaI#hlj|)u#SECd9k%=Zx8k zeIE~}41K@h_lsfGB97i&s?tsT?&{&x!`GBAf{G^+*0#ePtmKq>c)D{bkinc%` zNoCEOX=A^VcPsRKK;`&PF`YgShcfOUVrQ{8scc zl@?59JldPF`|B7{?>3&ZbVOA>vf!<=ree~B>K%F|*l%Z4<-eRUy*BEd80gf0cI8Up z_4(_R?iVdZ3M={gafUkh!|}DB5byQOl)kHzCg3xxl6qAC)isyAjmok9={Ju=R`;7z7TQ#f%NU81bJsq+&BJs1-u@OYN6nOSVomsj@AJi9xSlwZ@ zd-9+qCZa;#Mf;l##lfXFDQ$L-tJSj@bU!bU5H0AcJ`u3)45QgJ?ioJd>xuK4*ikJD zp{Bha;`oU;=on$dsPO!1@IZZh{kT_SRIh>W5oU?oB?4FB)q$g%Mn#UA@89)FehEKf zlUXyeuKQwj=P>5E(_qX|f%=+=-s>eXH_xzy#uP<9udunppq2iJxGv$Nq1y9g4Y`$E zLxbvk;k+qI8m9Nx3vIhH5GvyV7rN@+rtJplzRVm2(RvWT80gy8B&)nh13B_?O1iRM z-;=u(Og6)HgjE!6fc(HCT%TLfM!jx|4AL^0gKZ?2HLN#eQnVL_K+J9kotJ2l znXS7@Bfb7hS8EMG5%n5)DZ$q=)n%=smR^ezQr{0%bwT-;bCbO-a9X+FYsGRW=yC7) zV}$;hHp$h3$`IQx5k13om&zi}JS^GFoR$y;X&$+SNPuag;rxbdQbMRUtK$q;G(|R*pC4TSyi2-Bv+7wG%H<(y%`rH6`OrIn=o$9D%6Nd6YttxFJ9H_ z*>bo))#TcUm%XIkCB#gE6miu5AyDqZo_J5l?9~sk?%Rgns_?DHHWDa47!Z_&?cQO+ z=*gcFVKjkd+%%auUOj2NnXa zI5E9)#y@jtt=8>ThM7vm2$%escuyX(+wW}#yyzs8up9jCCnj9{1azThEX-?RETD1hRc z>F1q@b!{(+A^^QmSeHu_Ii49BR2-=GaejvNjqyuNKmj8kKh50EnQ~P&Vv??U|L4bu z#qJTw_6xmc@9K1-KIh412Mt?ee{!+A&cN@rq`6ZUe1r--sg;U;`fG9l3%+VR>1FXUbpuT^Lmv^QHQX0Coh8|I6$C6AC zR{4S^s5-295<_PmGtByIv};-p^hQ@+TpbBOJ_}{{JK8opNJw#+bv3XhuA-zipH}eu zq5g?UEP4~8z#?JJop#rwP(+}aRayIYp!>M!;-1FZ#g*Dud&>k?iDJ#ysb#t0fPcvBva9Jbvw2m$sR@)}c$A zDMDTIaHGYRY31X3B5QJQa=Vo)eSBM1%|>0SMDe@bK<=nDGA|)y23?~`dqhG(7*e9f+&3`UCAozpRmcw_XC*OwH$xf%EwKYB3sLq*1?b*RS0QXUt>qHBQJf_{JI;;Q4+oQe%!dWGymXGeD%aWx1@gUokICnqcaNk!}hvuY%C^MaObzP zP2{XlX0>eo(h*o!B-F=SFQCW_oPuf>nvtxYp{#?#Wf|sD1gqU=3@tBakpg$xMq%y4B3rb_vLHQTbN6Kc~Ilr_g8&WLaFL zv#CWAwd@Zo(^tr94>H&?&nvASQT9sUWHdT(`WXJ)Pn0}&!Gn9yeB{hpV1!dCRo;NH z0pHl&t<58Ct=dY;&fl&)!b_GvU(AMNM|Kc@a>KE7LMi0$IwxD|d&PZ`5pQMs6)M=5 zRaK!ro?)$utS0Zb(D%QppHKYkVgvr3hAoXd$7=EiX>XRkO+Tgda7AhZ=a+cQ0hUPp zn=cP87TQ@}eWti05RiivlXi#{Z#cWKEha?07v#gcoD%hj!az2GW@}MfiqBBy>7Gc) z6{SKwr5^Y8Ki^dzE{+-;B+YyS96io{^K% zKe|T~q9<1AIVnx*WWB&%rLrVkeI;b$GSg5uoyZ$|{xeR`&UHDc=PpcsaP4ax59E$7 z@RT4LCs8KR(o^Zrr1Xg=`NUvPX)Z|DQhP+dTI9v+U_w!|qcyx9!hau}d(D7wL%{tB zM#L38)k}#vXFCg$3-fp1jxCAN8|+$T{oZ{T95lVFtxhr0@1$EzZA@2G7P!3hh>XjB z3S-3YrT5x7+3a>pvETRIRO zSADiTd9$#C|H;ht8;;g*xXG90A35o$GFQ7xL{$q5M~_UL2@q`{|1&8aV4qhz!HxXr zmv*50i1ZVN`6tfdgSp~1nr>cZweHL(i8G1}aXXol<8HnNC<~{kISice8!S@gvwOOD zsfH>JDBGv&1W6%3A0Dl3W$D|Tu{gq9)%j>jd3#j-sfY^Ay_-*Fyl~W+Xv($6kj!Td zINFHFIT$>Xoz#z%Nb~Iuj}g=Q(e!K4pC=ZPhos9t8KpRUmSzE9^2nSQ_w#C79W}*J z{Y+x$rOfe2nn_Cg>&_3{I@FT6{y0y3BEh8g6;v=sfQ&dFWNfN0a{)(bQHYg}B0Hl;dqM!mUA{N?`s(;7hmc^|)fq>Q?wrKLWQ@>j&5g-U z8Eu4%&*_AJXQlokDt<(-{^gJxe*uwG z-yb&Fw9I2M-tv~beZ0jXl*SV>#TFR5h2fK!7>jtS)ZDxhZO#@Jsw9;NU{VZ-AD_Bj z7aNrg+9Xf)HMOmA5PL<+iU}HAi`0TuF7R^y@$J?@Hh!?m*+~sD^#JRS#06vXDz*GM|w6R1qHq<55Jw-_47Iq5v z6^4W>G2lTia|C-{neFgmw7CV7+y}<@$B2A8ndEj1<;F3bDeDF{tx9?p2KsWU1HUrW~z0#YMMFqi$(F@W3$(C{p1q;qt95e%GBG^JO`nh8zuRq`dhy@d>QD~AaJx*Wu z(b2pq#7hk{+2Q#JyJ!MtIp15K4rdNpVZ5c4Ln<7xz{Q0k1h6iQ;lIQI*e)7phOuTe z&I`lhG^c0`1d5yokSz^aId^n-6^y=c?|9k`ygNX(Y^4QvyZ|NRbp{Ak&%~=4MPE^# zpf9jKathtXP_!^Vsc--RmXn9}bAl?LEPsNZUH z80S0JH37Ol%JbfO#|kIl1=y=z$KRq?=XTVrz3{GAfV}t4%3RI5LBiWss=-kwg%%Bu zVe_u6)+WPgRokNHVZVliD}Fo{#d{!j$nA11%AxXHmUV*WEnna3cJlVbfb&rC2VJ6y z_vP}?kjH)b7Qkz0s|!pRWZ&Q-2T^~{{`%z6VlY2K^LY@3m2dO1r?E? zk`%N*IRH1qg&Sq??KQe;b{T+=ttS8=A0Yh&24i;D<~wvRT!wM1vwWMCYZ}?RU8-t9hrp8|q@UC)+bH97?+vp^xe>X{SW?urbTTn6wv}JN)x=6Ms|0c{oKb%FBt#YN)L?iB&6F{ zj%q%_pxL^Iz@mxwUsK&UyI;GJncJwP%JIHxg!wbmgJXn~yFEr+GV)xh#3FBEBIkm4 zFtYu8jOtotWYhJ)H9V;=sG9OBm#Kr=lnni@E!xemzgBn$iM`p#CYT~xd}=rLS36Z6 z-$-!t(*-q;zEJ%kE((@|#*FJ^mMgCENHQN+aq0~7*zbPDlWC0g&U?j@msWcY^Z9o4 zFZ-3UB_(62F;;LV#+fI!(r$xe6I6=A=B50wnIP1_%g(qq;(MJ^BMK5TaN-*Bm{5_> z6P%le16A8X%=($oQjBh0=`W}zS>;GgekTJoFdR|6dvX;XOa#Rapa_MAgb5%~RgwX` zhXaX3aHQZDOgcz(AVKj3hX~`(w=Klcw>05e-)UF@1RN3@^fX{kU~gO=a7E6SQ|JWB zJ}vVBnZpI5D05n+h35rDKv?cBf&c&z1H<9}tQvkIl|YVwsDa9uozhK9Ac{bCxWqd< z(B(it5SRnrQ8)y4gYv1IvPC`s)7dczKYf< z*T|mis(9vtEo*b8pJjdq6KOEP_G6$RU-XZLpznCr$^;8u;Z()LkVaA*?Dcb@>Il;0 zta&=&kvzw#L4K^2lDYUY+}A3rIf1(3(JzAPK~+Lkp> z_Uy@40R8V0X%tOkh4a=Uv9_*1CVy7euUIvIR^CM^I1m@!&X&Z7Izb&3K}!XG)um}` z({71NXK#0zAuHHR%-_$NZwvQX&=nNsJS@8U17G6Gj+UXYsSf$9|Hq)M;)0>K<&ud{ zIDV7cZ)P!;V(^=3BQ7;vXNp)(S&m6^GMYU=4ILx;5QE-PKelKF|#q z*Bo%Gdo+$xw@e!wHZ*Y6>dXJ=b&OD`4E9J|D@N_jxnN8O%d_{?fv71dr0_hWQl|IO z86(vGLknrMprwFwAM+eROJIl*^P6%Wt0V5=YFj4W{29SzR)40LrrwxW1z}mP4|;4I zA4_vQbOjEIJ`X!cy0e`7>!N-@kLMfC9qJT{v?up^LBlPvgn`Xj#V1tXXlt&jI#iLM zrkvL9V+X~oFB(cPU0uWdkR^1C7_e)mGd%OaE`mEc+>qx%b@K3#WyriA5${w`w!$Q~ zuu{6kITz~00r%LKeA;qnNEOJA5nn2_>I*M(fO5rx?fIK!w&fW9p8`?(Q|?^VynS`e zym8nqLK^X=-C@`7*5$7j=Aq(RG7JW-bNtw2B-Z*cbHZOY&6~)~zjLwnyc#j~;c8@Y zA$#?f@OULl=je?!lI73mS-(HimF(V(dqUg>EWFkkjS&$1pdCyHRDz1>k)C2LZ|y zhsymE;E(_5z5p4~EfA$Z-3hQ6LK_;`wxaDGY=Jbv#?&@JXau`d$OA&{92lxeLSA6U zf(lNtGKwx5pS0qARo>pX-OwreJ3*+hn}m#1@oe4m8=IrP9lKth4#kFnbQ0Pb^wm)> zqq)o8V$;)HHniLA8TEO7QT6J`n!+t!Z>b^Ok7Ao6d=i;^cl|OS+8LAfDukpjg}=11 z^?dMxsiD;9G5>GrvxAG{jsv$aZ@Db(WQj`n>l(%G7P93Zwe)JW=ChjH{d&>!z+B1( zN#!2kC>QP$CAJ-AYdgVB@}MUQdGpavQrQh7rlLd(7vEZ`_@=J}!^Oc@BuO6V&v)tv z2`ctz8VQ#;8NRyXSg$K0#i-N5yVOM*60F!ZZOE%^Se~(85m_ZD9#j#%SkF`XIM9nx z_*&?*%QP;86&KSIoUZh8)2y$X&E64}*1D6@U9coej!dw4{GBw~)Su2zeb?34Mq0P@ zO3rkIx=C8x7fXo(qlA9T(bUZahXL!a7GL{6`iaUVi*8G8O`fX)2{GusG$MKNC?ldU z205i)m(!3COEsaJa&+;W^j3TC2NQmYKUdqoyoib>_ivu~c#HH+RM{HE^^pq!FVbCm zs=S|HG}4^d@@}SB!W$_Wr8GV)t1Ei;I)WqzMYv$REux<5UH+qI-Z*o;!5zWoevfR% zZD_+-Gr~e&wRnReWkGkAn%^B+uP|$06M7T+Oiev)MZOecMT)O4oyV2RF5_1|%DOx5?n|wo zpzD`z?1-fF=A>TKO>NxK6@f;HK3u(~frrl$;v<`wJqdq$bN&ukNTI}O)7wR|p;kZc zezo_B8=dB{jV{pS_hiyZw{ehs*74*`C6bes%`Dl@_yJpl5;u=eW13F2kWkp=+BQM{ z1kqSB)m~S1v&;jvc9z_yy;rZi(G#o|Wlq5?eAu~EARUlrc~*tFKi#TRR)yHQ#q^J^ z-z(5OptW$VUi`<$tQbCZt@5cu^K8Q{&`{~8FN?flh`Q(8DZ8&Z>v)Z^UTZf8Y2}9d zVM0Yj;T%EkbyD@B4dEHX*(lBRGnC(_Z_4uqOA{hx47n^S(&Pn5X1oD`EYJ7UR#0H*Iet?CJR7jv`w-ZDJ5U#-a?}*sT7y6}@ z!^KrV#a65q+ltNqDV2g!EH2%pp2~<+qv48zio{%1X404HEnA5nRV#}s3`DKiRlN3d zB1EN?5gAIl{P}Jws`AJ|(R(^`KN87)rkO$Q1+e8nTe1^?1zLv^BssvQf$k#^k;$Au zJtxWKX(YpW2Hjr#>;P_N zgH0gR-Ixa`)ngq5!N2@UvbA+4h?~ox(#8=t#QxKdAbXzS=iPuNY{Ku#;vw)>My^Q3UL{nCJOqP zhF9@aTR32I&+511Z3;H6rY)w3q$fP=Dxpz%1|;L~=9+!s=(`r@4&YqyUW^W98!9_h(Wam~dXx@g{cs$or}+g1EIA&0j`eryyE|0T0U z_!&tRa$}>=pL&5R3}O05tbLFA3RnfMy}TA6iI?#(b*7+3!+v5~+E+iPUXLP6Lq{v= z`s=9d_%d?2GLY6^rmyD(`TGEutpb$0*Hhx(0(rW>7KVT_XFV?hBxX=N4CxK<$9Nid z=~52fuw|@uyCJb)FJ8ctXtY{%;C*3_ZUy=3^oonclTzcBEc76h)I5|E6MC_x@&<=+{EAitQLX;`Sj$*MwltN|Wy% z;z4J4e};_@KXxMUaxNf?l3&c5X%v1T`@C$mH90|gg@3Xz?E7&@ zsB6pj%d9$U&wrmC+(=fLSduTe#(#mlCY!HmOk*8qSzwpYkvt(`WN5*dpwIIl%fW!51V3~i*gJQW|zL_$q=P&NTUC{VvCT6+PE z2+|<9aTKkAP|k~Hz$<{V!6Wt6AmxKQ^}>WefB=2F)d3P^XquBW8A^Q=%z=S8X&Oc6 zRC3U4a8?A=81$pUMgdw!15V4);4$FrNw(}OO=vHmWOqf5@Lb4GNJ7qR{-TQ|dEL<|y}>=T zIBH@oBkuQNLmtd(l$V*#4lDlI`SztaDZaE*<;LwH08PmY1Y8KC{H@uJ6Ckcx+R5ix zpZ-yOjccxFp!VyWF{g*Y0(rVF%@H5j81hxU!gsNCveZn=-MiYZgDR!$^@3lzrI-xU zu2>XtvzY~F^=+h!x^idA{z6jQ>^LXp_Z1jPytH{fcyqPrLP5gr>n{xgY?n*68bJqA z_H=`syoEdFPDco{mvD(1b#;TAQ^pHs@6zz!BvWbFZZ+!`k9Oath3uC(1@BGdWR>e? zn7m>u<4no5RXNOcv0$!~+izH31e0>ry!WKcheJt_ zVnZewj#Kp^nLr~cr>aD>VDksR*`Q^|^WMVfI(aDYebJ{5Z~CeGDcnEJ2H-;r zE)TTJgK}K}-9y;kztV2lcR+pv7v%K>it=Ki9`y16Hbx{RLfQ`{+3Z|UK+Oeh1?v&bfV*JnJ8r_?T4KjH1?!d8FsaXozP<^#SPHj3#m?+5O8NOBm{23QS|3j zc%6U1Ca~yvTj{Ud zxYU)c9BK&3H3I9Cte_6$y8t2pjuzA}FbM@zKm;%3-<3v!P>{dTtb?B%a{YTfSD9tA zzG}EfR$(}=YHm0$HRk1r^M;xepMCG0rv&m2srY+zmcR>;1sHj`eR{O)kd3{K32_g@ zeRZ#w8oV~bLd)&&&-Eoe#S$n{y16Ml>^pYA548#(d6ku-6D0_7Inx1b0Muy?0t#qe z$QnUz9gcENoIUgfmMW0npbr*gJdhbm2HoZ02*TmUV)T=6=Q@OfP=|tM=o(6_%d zok1m42fa==_IlPH&rROB#?JU8#V)^3r37W%pUJ#=658`cg~emBGs{XA@a`MRA1*GO z>l{{AGhF*E$(^_snV;8GYE?#diFIQVG#jaW+Au6ue&(1VIsNA>PLGdyY$xFokG_Bc zsA|LqcebAgq6ivZAxdG@P_(cEh-YzPy|_>~@huHseJ+ppN%4`vAl1}@)Df0#xS##x z8}tk38CWx5F_%XWooZ&$yA{Ifo~kKUb``F35`Rx_+!fAt!6C*eK@ll5FT{BWK}Bm5 zbdVS@Yte?_?_D9ZH-=A4(-Bg*WH<-}JcsrHjXODTFp~`Ai(o(Uk6tQ6?=GM{2Rgol z-3a&r^Yuye$PRQ2#EIaPA=DN+F;E`*iva(f>7?QSv;%4vhJ70+fNT?lAJnRscF(T! zn(__*+GN&OP5qip(eT|$VO?C&uB6RGX3?lO%Nfselk>L~{yEZOysDCv`8|hsZ_=3- zmbH;y3}K_cbcxayt}Hlq8h&>+UC)Zmk~&5hq#ZUL@w832%gffk&tnK{HXZrcKxm%Wl&=xfJ2~CLsQ+xgo_eZEF3y8p#A19{;c$WRcK03Sfl8^Z= zj$0)|Jz``bH=~SfisEnH$QmDfU0`fZ-&1-=xAaC~wx3Vp*}>>g{%c=U#t$`uMGL=4 ziZI-Xvf640crj(f>P-5)`Rq1Re3A&Gg?iL5Q&N7?BmEfOZHv2EQTjz+`!gc8Sw($i zDx@M*g!>87>Xn}SqzjG^EmpC=MklDVIeckf*TU1Y?ACqNb*m%tMWsi{!9nyDDi-FN=tl~9v2kRM`%k%Fcu$#{jHhsoI59Am&ZxzspXR4n zJ638O-UnL-{MsKOuTtAIU9>;n1dRhWBMobqY916ATd{=QzyrO41Pl6rLO31b+^OE! zkR~kN6)QS;^Dynvg%wcf5Mgfbv*-2%k*3$=9^JCVc`Trppn~vkrv~SEP*?nh~+zz(=cgBLsr*dS914LGTOXrA?-^(9q8V1joaM#|jI);=$Fh+ojp zx&6%=f0#Nu!&u1lIhS&S)Z1~D4An@BYOldebseUj4&y~SGcT*kW>Y$$-MR9y{ivqi zgL^xKy#qGOK-ez=|HV)avI~|dze+TP!CGGp5 zVY$&oy`C>=UzbN(N;2s9(#zP(eru>QYd8?|n&9T=yHTu@=y1(?KSBOfO7C}g@h8zt zgg7>)v?A@vmdbnniAcTS1Y!})7rK&nCpc6ygpJL#X~$VKdAZCFd^HVP(-zlj%A3}s z-Ei#tlyG007jD}SJF-1W41PF~$M;YS@2&d~E18YRkHi-}D*F|emK{I&d54spZCuZE z=F+x3QgN3uyT!oCRoEvvvS%^HQrs=lLTajlJ zYVu4n=$ngM!`tzXDou5mZn96icz#m;VPzED5|-K2iXU?79lp6IQRv5Z!~T+EA5I!l zWT5)ZP|HhV_d7`1OPIa>m6v;Fvyq}~{tc^th@T(Si>~ZO&EMp0yHON4suH8)l!4um z8OGp6p+xHLUJ!yz(Ra#H=yWjOyc=J-ZxX>qKCLNFv$fa9Z!;NSIk2>>G0wNf!$W(8 zka)~J`R##SMyn&zmcG`SS8Z^JIH9Oey*8ORe|dww^J_Wo#wigX?Cvk6+N z${%5hWyJGL%r?n(3UPKttIIFVY#0MPmja{Hp4Kac&**u%1UemjecknB$;$oWVE<+2 zN4^K=NxEnmEmfp_=z#p4Y20>v;eH|NMoa46Ov#{MNoLkmWyqQ>>mSc|{vxRf2KaO3 zGR_%tYlOcCRiE)wKe_G~Q#OdzS-M28q>8GQVlL{DW8Q3LixjMWaMji@c-J!{U5?wb z?^2n3v)_(~8&Xt4<7>T*1+k;Je0s9^Aj9&tpD32=j1FJtD=+2s#J^MUs^9Ft)5vzu zKhRw(CW*&|{k7|z@|=B&kHWbd)Dny()iWh-x(zOc84s=4&d8ko-04h^U$P@?<$eMA z6(O^2bD1OZvB1H!8$m0U7-F00oZt0Vj7;=J-RwW|nh01wA&NIM?IuJHf{JQWnm#c% z#CILord#x?n8%oXV72NHN(-aiPIuGND(MyG;$I=mUq{fS)?+@>=Mq%81^a!pm%oh= z^R4bdfe6HgG#+|U7Qb2ZyL=nbF`~&@Ixcge*yQwbN>h$H^tHUe3BHaF@!%Ll`P4E7 zcbr(Oe^Lq;K=^8nhe|W$ZugbqATa`Ds92cT|Zpmm}Sr*L+@otQXSW-vP zn~7dQ59x9@%aym~x89K-Bh2P!&jd1hqwX8NqiT~4pOv=Jb@$5=P+h291r}e|_WlXG z&nH1q90Up^E-ENl#ZolJVuk5v&j|`$Fi7hhbR5>bb}=talk(^yQaM05BbEEZ-Q*Z+ z_lSH~aTaFwr83kv@7?a4LH#+4S;JgIqnKjfXE+bn?reKy<&;)yt8MJ)J`?B0&;NxR zmr9&N>wC64>PLB5ANZetGucp}xEiDXw9Ix!uBd(u58Zx86-p`DEx_+ZVI! zFbVodC@Yz#qKLG6-<4NY6y3Hju1OzvL?E6dFXFjOvz!~v?X!Q+zlZpCt_t&BiIMcM z@Q*oV-cq%`KM9%Fe7;{2PI%$a|BiLFFurjo?t;mR;L}Uh#o4RB)d~aoY<`*kVDL)xkrQ_d~@?RVGLek*}qYQ?1ZremlaYvGJ(sJhEDhh)#0MEQFA zMB54n4rMn17p;Z2glb&oAf|;fEx3tf07{3sKpRXVI+FrteiA8wzd_2B&S-EXdqrpc zaQttaFzb1YfCGhR++M3yCxHThq%iZMI&HU# z_8(_||0z14{=I7Qu1=k-?@rNg4z@s~%Ie`!`nP&o#j4qaJ{7!|>0^Jg_zAXX5!%2$ z#oeh+x!b%F8E#2+*!EY8IZ7xxKd z?nce(*iqi7ehKO1d_I+}x;dZ9Yas2DIzQxBlpRMr#)GKDuu85`*dB{LTB?YX0~B6uI?( z)?ZI8dG0$`NJ$|cvfHkvro^@%rOa)525kYs+t#WVVw4o|KiFx!U#MrH#K3!}M&ui2 zo43V1e?+;|iKySn4Y7lI$LNFR zkR_S)=*z<{qk1+Vc;J0uH~|Dw2n+UYU`YSMG$$8Ah&O#QDZ7?LWk|=Ey9Gtjvxm2M zWH=UE?PkiDi3TinuILY;?kD{gAtk)^%r)nb6HAH=Z`ir7%DN=*=Sb^8$wl`Mvnf*+ zOJglvC@Y-wHFcY(UN=fiN-0f64;E9`NaDITtsd8^1`iv`yRg}7Mc#9Ccls)J=_4a< z+l`V(ei~5&vObMC?kl+O8Ke@%QF#{6EOqqXi1e-&_WQft2Bq}7jL6)s{J{#mYL~h0 zgv+$sjTF*2O>HuU`m^S-czF6u|Mp-CX!_I4lFS|Rf8B3&kMi0Gfp1YRS)wxufi}iW__Fj%t zU`;+J9MfmBTAgFJeUls~V7R&NYMyk5s+zs(9p>G&ck7waHjIM`0v3-p+o%kk_7)}j z8{EBukiIYE)`(hi+5+gcc=PYJ#yAMOk5&&9RR2j(4Bs(r$`G9VvpuGFamBFLtt~(p z)EatnZ%88MZ%Xm$b0(hqD0Vll^MJ*4MpYkl;MwBc7}s>2^X_;1V!!B$ z=$;FswPv?q*pd9g+(wneaOL&v$MUj3fxI%}HSi*=-7^MisAg?~aD)S^B zmC<6)sP*@AR@|1;_palqT6hu{w3=#AN9V`jEx%2q9-TM9*0QT zll0idjQ+VDUB4OfsT#4eH7$OhtUZCZ6~0?v`l3GfY&_5d|1gs=4oRPjdHD`@c1Q2h zAN|p{@n)U&)w!INef-a@Bj}yHrL)*FswHK0FH1L)d&IAOi7qymFNTs|QF zzy-L&zDquj>1rJgEaJPQ>ss*(6{Kt+ZC0K1ItP8 z`LEA>b%Y!+K!h+rlP229!JSetLIt^GfB`|hexP)5>QYbrtA;$f(+@<5@1USA>NMno zG?@LA{(nCOJ*oN=Lo!W42}6c$6|sQKZ;^#AaD%h)M$+~A+fdRGm{U8Z9Y~|)C-x3!3~Js99wQh6x`h1~ z)+_IRt!&twv%49s*0FM=K+(wKOx0vtN^40M#%4ora7Q2SI6wCCErdLJ^6rxZq#U z(13$5G$e|Hc4|;v3$os5sR@UiCWt6>T&tq?nJ&Np*ot?e+<~ZHgq56Klzx%uQ zkCPW2=Q!uQ-p}!PJf4qNujARpdbUab!KCa!+l;}2D9(WFbKMVyAP0NzJTGAA%fE_M z%#M^Kro`fv`?%d&)a*4Lc+uAlnb)*Bjc?k~I%X}b;hestE=E@PAnJRbt+XoYUB1iV z$6odE+K~GZAMN#tMi&PsM}(7=8T{Xk1CFLZ6SyOOVje*7Jbvn!Gg|FtR##b|vqFYr z!)__I3ZKyu%3b^s=02ETjPfGiduhKUrD)p8>`00cMq?CBpg<-+lUctEWJ_bh!-AF= zq<3a^Al-ppvDF2y$S}2Lrn_29GJ9LzS^+$j+79ChbLdA4jX+_fU3+TSY(~Ai9~yT6 zkK@LEMA>3h{o&=P7=MAX`4)v=_RiHvLx*)cIbmg}YI;(xfA(O6^qXy6&lF!yiQAxj zKIa24l1lIm;En$>j$kCE*AdV|2A~qTu3{W{9uWV%n<_x@{r7smotA>}je-90XZgFU zFki1=Ok2RmfQ8wcc=b-|g#UL^1b#%&vdce;H+^QBk_qs1GeC++U}RorBnM`zlzN^g z`E$$el=%x0~?6LDwMcQx`n+~4i82xa5=|Ye?6~c%PAu(lG)nW&0C{th>A(p zGA6*T&mw1C(!~z)v0uA#;Z2rI0f#?o9Os9+V#y49zJz!N5eseXNg~3g=E;f}6n|ct z7miylmYjMBC1M#Sl@EGkD|^8IZ8@g4tHAbEg;mtyGka*t_pYXBKb~a;W^-u*N~`D< z?iuB>WIzbyhrag(7d*QuAo)P~3B+jj)*TPAeIM9Ons(lsjZY1AQQz&2za{TG!9coh z6+K!tsKy~U!ZT3f!WX{5UM*?)+^_MVm!NmmLsYj&o$Q2HW@EoPVRD=#jH~n8Y_zds zcUGp0eWrZ|WLjRa5)#uqLgy~k zx4mx+#8Z4zE~of|Hs?GFg?e~_9q}GxlXMQGKKmD^49{cDInD7$9`wIAlP3a1AP~j= zD*OIdY@__AGsX}$fK}Z)V^#hBEkS^)zXm@GN`n;ze^;Sb09$z$6)%JFkUv`k6Ob@` z{2AK`NUpv!K>GhgQvM5t#4nxI1Ai%&KMm#Ci~b8n{|kC&oAfWk@)u+IBQ0XXN>8}r zsuHG4=fy|2MwN&hBic0$!upLxk-PjnK(JCqVVDdBnaZ&*MfVIm$8MgJ`gLL1OdkCI zC^Y$lyoaLTr2mEIDD48bY>AAKvY)CxoMXottV>O{s!1#E+vRQ7Q{A*IPgJV=U)|T0 z)9-OrQ$+Q|$gTS-t4qC~KhrNs0-{(2bJd(hd-^XFh(LC3awfK}77vdm)|6_>YDcNa zPp?CD_{S!?)!Zh5xJLR1`G~I}o>cnnG!Ug#g ziJOXX((BWi+%fg~hZkaaC7JE1lVboBAbuW;!TbG3Mvlox7yjjd4}bsyaf#jk+@9)h z<9-&u|6H}_d+Kkx_;1?c`&*3t!K%PLF^_)nbZReG^bBd3#w@uoOH3yQtg$FTBUYNv z`DcX$3RukE{U2}v6Fex{alx~JxCVAHMn(U>gUn5g5tdgP2GGLngtK&X)}esE3=^Tw z>GyL+@DEM+2Pr!%wJ=YB$>K3`fH61yOSa14wm=hlyXMjQ+=DE^#~xo$FoDqOix*8V zM~+;$<8ooJs^jj(u{NKYctid50_dP7N_6VmczCGtMrDakXtRW=Yj>*;|Bw0I0^DUv zKc>)vkSor2?#b-uZQNM?u^3|ubHr9xsdxzY#F#}T@Ksk|c__KfAYxK$r?|lw6*3&R z_tH5lfD&nxq5Y%^I}Z8sR$QoE0<2V7x5}J^p#H83CFTtgEL}2eX);rkhqQrEpyC16cH&!u{?r1k`JBP=~ zeQn60vBfeW*Q|a&KN*-lDyhv?B-mBK;=3zT1}2|>h`oj^aWB&k$(l1AWIf0~P$6tY zFjb*^tGT#l%JrWK<-S#Z+}N}NaUDW=OwUJTAxP%XhG`-5A!#9g4;>5@(o)Lg+FR!= zYBft$RQj$f%L4zNv%U0ZJN_*&&hiFETm|-wA%-9ampumVq`+XI!~jmhDpih==mpki zQZU{%8SIo$|C2r-&q(FFdQQ$_wg=m}tOJ{75ma3%{=NUVdhGwR3i})6dMo~}`f8cL z77pY4a`p_E9odOVb26r94KpUs$v*v0bOOo(uqi<|QFDO2TF+G>OvuX``fQla|p({~1`;l4Y z-<9KYo%GSQhR_-qXYC1YdlL;=!tFfqy}{r zqQs}29R%)a-=fJBt>DsG*vU7YTF{=zM?c9d&*3kqkH9_tDkPn21`?t z+5J?nap4olWEo(&D*Q2g+f0@f*~4&eB!p{<9HvC?9`T8(SaEL7(NR>E-)li*su$w- z8;cRq98cso5%sJ7`#{AIUb8{pl(tiSZP4dI`RdlW*_W%XVX~4iJR6?|e>3%XjSa{) z@WW#^HR3v(fdRvJ*TXRrb?1c#W+VK?#glr^gPP=@qNTx?o&6oAwgRHOP>(C7fAy%? zi1eOEglQDUcu6I<8azT_M1xfu*G9EXtN2@d@21{#hBN0{pcGb6If=JEKBwSznG8mcl3t9W&QEyw|$S>`O9ZXL2A6wjaTMRo&q|D*+ zo24UVK6Y-GC+mC{wV9lzK5b}QAKsF#=;yYtK(-NPn3F~Gxr*s038s+>LhmaH3hw1M z6JakEHX70vekXcfXGwPzo<6}9HH#XG^d_qpRn57Be49_h4snRrtx>8< zm~3tJ%D?@p=SOk$`YIfnh2zP$q98U1zi!stu3i4hn@?wkrW<*?sI^BoJ>r+3<*FWf zdEc>wmps{ArXy-;hx$?rzsIx!p+2FxRfgQjrbo|~e^!?uirpgu1cb8!8$u96cp3%gsy+tl7b8GS;#9{$Tw;%Jn!Yo*+G?2xpN&!y1AOBpA&pZZH!Nb10PzqTOj zA?PsDv6u42pcgbm@2*XU?IsOLrY}+|bRu)~&Gi&Idq$~w{AW&=FWZ$y&x1bfu~%Bv zhvlKkS0yFEQbx+5Yu!q%ef3i~=6)TjhHpz?CC#r+Wa%8Fgim%P-_2I_;w{~+;eT}+ zV3Q4_45{%QfB!wV;|he=*>AO$F15DPc)A58dw5Gcr&CWP46^ZAUV+dlIPTO@j74-A}AOH17J zHZL@*zA;$Qq};C;E3p5O8sf zOjCT91RH$ipyx#z7)HK&9yBrK)>$^gfFK|X|Iv!n4BIT@-TTKFC_?gR(Q08TRjT0oL*y;V;R-YZN z=SwegysF*AIKW75&J8rWwB{-Aty@j?h1aR0R!s~0#^Y|?8?ou`i&;UL{>C!iboY2g zccVDUV53PN%=_i_`N46&<&3f^C``ASY>|0b?9r9n(Gc*5$~?EtX|Zak%u&nZC`G6w z-cg-cTF))xd#+=^>HZkSRQ^%i)xPbMyoClM@(q_8dHB-oSTyJ$a)(+tFfpr?kM}1R#Kl#1;A+@{1&QTz3?bBDXt^z|!QF}!LvW+2T zhHo{?d=(f5f3YkXzO}kkPGcT@Xpep!(-gLItt>5MRAlfVu*NWDa?uX9Z|B_G_9Ju3 zZAUACpl`lUggM7tF^#W>dH3aXd^0!1iy?g{m?Klfn|&=R|S3K<~y?B zHY(-f%b#gqbYEJBQ*J@z668;YUz~`Pht*s{Yeh|ajO~dsOc31~UX1ZhMJO;0Zoi~R zwka_cA>GaGzJ6HOFsVMU+5^=cg59>Cv#vGoo~o-Y_iw^p-Fo7ZB6cBD2z7k&gEQWq z_HMCX%YplCQzL=C$A(vNzT=V%nr@0un~M@X>$k5oik#3N>%0|E(F%cNEa}PmCUCBd zI(>hpr$lTqaAT7VKUVBPh3`U3mDjgBwDb(`p3mA0rfxab)X@)mjpry)UnEhwQ;8Ll zFyIpwH=C$)kV>A!?Gj)9&4O@3+dNl~K9G)WgyH2qHy1dxA?jNK)sX#)%v*6QdzKPM zp3gWp*;k#!BEHS!k;LY(h-+>u$cPUVoVS;HzrS zc7xj-&MjOXD_Y@%Z1{eF$6@~n{R<(2&Iel$$iH09fWsyXhIqCFO0Xes_$)8oNqRNC zjulwmPh5{#&D6ehE}vvSFC;-9{TbWxP+I%Lqk-A#3iVf}T305L;?zl4OfMCEXtwD{ z7EMTDj+tjit!NQh*3fk!_vk)S>W=4@q)s5lLRPo2jxM<@&<)m1#F$>{-_xEX8d*Zq zRt(;RaWi;|NHO;RvNq>XqJ}}o$o8Rgr745cRsy_aYRyiNcx#nTrffeZ|0zM;mWPt} z-#r8_@(QJqL$|Fs<~gY}MLyFIeHXTsp3<7&-4YF~wlc9YDu#q~C~-{1jwaLO%2wqlIBlu#gzJ-L&W2L*%CV4WwKG*|)p-TJCo< zRl8Wdo{pZZ1+J@Uom^PlfFS>GpZW)c!L$sRijz`K71K+ftBwB(JTR50_gNKiMg;(~ z;!88qF>1cdT#VFg1>$Aibhu8=h}bK0En;cMY7jirqMsuId2cVqS3Uyg5r~?{Z(S*2 zDRKXv`{b$(CGZ*il{vwXQ4-Wp4$9~U8xJX>0FlwP(?0x0$WWQKG-*~?C$B+pMw0-?> zjLuv6tXK8JxphTKsr4zw9k?3V8pB9@15sm=Dc<=i zuo7Kb$*OS}!#z0|kxuVW$nyy;Op;K&ee|NGz2|j^^wR>X zt_rbg;h@um#QPhrcOwq?_hD}pTc{Z$0!b-k_a!tec159!(=@_Uwe2@_@!yKqwXE-+ zFo}K)%+zt3(Tg*%j3g0FeZ+q9)Tqo;Bd31W%gnss*B!kpTf{Y|A6eivztldXY0Pu0 z<3?ErS`G)FKqJeu#?$aKb$BaRZ9LTX^3+XJgA~fEB0Tk1n8s`s_^+8y*yw%wjTK!A z>$A3dQ;gC|rE7Fk>+W5wLoStb94CK{uo0sF|Qu zzmX~*Nkijo&(d+-%PgaL+gLrSba;EwDdXbdB0)pXIuJ!=8?!9wE;De08%O{ ze48Ix*g-3L4v}4N%7<8@-U2rr&EHrDJ;^WFJG0a+t-7tM4V_-HO9{-P;J;`@ztJ*P zbtmTMi(Y=GCh$S2Z}Zq|HmVOC62fmQX>hY8r~Y&oh67y*5Vk&K2!Uv3&=?ZD9q)8T zN!U+vrAS*`UraWmFIWcW4B(t@a|-)b7t?Z-yqSNsKZJZrt@9geCZ+1ew0Qsx-{?NA zTW+h$G$ZO|#qw}pnOW7-m?vqt;^UX-^yPd@XVeMf2KpW}s>dsu*@jg^^(!-+f2}~S zXZ643d2u4}B_Z?kzINzuIojq%XMc_eVbqposZ2Yz)o8{r*Q zx{pMB9~{5AtBdNiUXxO^IW!!q59z!VDdsuwxqB!3)6eHK^tjY<)|`2?&J#{#4cDDw zINLT5rUohu((tD*eL=Di+@v;!& zp_1iW(v|taZ$rwHrz=h>#5&zyT(B&izqE{%~G^CNve=(Szf}zX&DWs*I1zriOe)r#nf@7PbRpAyuK$6 z?^~UGJhZJn%?V2hPTqX-@L5!TzD<_bt?q-D&5( z?CNo3RIaQ@UQfQdTfHv)IlE-#)i9y)o0IjHWmpBVV@laQmvwSl3{wx#Urf{S@Ae-( z%Q*s#i$T-C?V?IA_yLQZyn>+@i$}bG&y=6}DKpYS?l;zIncRsr4om)TEXzZENhYY^j>i&k#!1(zAc9_t>%_4sXN-Zb9LCn?jJF2Yd^mV$OyvdA|JDBtr-FTj z>Fxh@w`bp+LB|+c0=#jdn0EJ!>ICEosOrv8-oFDte<_II{<8fYy{Gi$pAsa7rIqqn zdNp1=v5O;2uTCT|RbDEa906xwOxb2nHn8=IuwRMEam#M^iVF5br4-jFP4i|!hAkAi zh__vUH|mN8wp_vW(+IAr5`m-m5czMyvoTFmjW=HVTurU7%yr>y;t#}jY`Ed+Y*>`i zgC7&w+vR#xlVi_xVA+tb2_v$h8{tlDMN&sv;=6HZcj_|wtX7#hKif4~ABf8?JAy6$ z#=30Z@ZNVJFJ)~2m_f9a|6G()7txtW$aTio-05ci-pTh}Ycz|WOcgln?Y_y*kd@HL zSyzB|nMZI++5?ek-WkgW!1W}|K^(NDf3$h5E|L zY1GEqLGMgmeXGsPbVIU}mbuBcS0`udq@_couWSk74A3{HKAtAxXLb+VG=aq8OiFPk zwhrUN(B{SfR(6B|!^TU+?6}im1tgXY1ICj!){di3?k^0DWM8+7;$Trx)-ZYo)zOWU zpx`{6>D$R+eNxfF`-h%%r}fzOGF4(mFPRqB6|L{v-ba4k2dw-SJ%N!{oAew7 zx&iok0`vWrkqp3QE97WE^c`h6EJt_vp4JCh+~rV0q#1gI5luYw8NgzIGO zjuR-#BE)K+4nTvs*{|>@;;lmL zI67~@^kc8J+adtV|3H%GU(a~D^N;n6ey)<3gSP&6Ck}ke*mHw z5Y3n?`p-fV0H0^H5oY(D-9WOX7cowIPht7$hB%eX>YW67*{xxP?w?m8LHQhvX^Kl=N00R#@<MF5Y{|oTIZ^RlB|; ztIyk4VIZ;xbxeF0$-YN+)Yba+$WZT{nvTi^d1x~SN6*g&yYRVAR!4FGyQ$s$GbxSf zDWB2*qdPH@53Vy!5-3*wtU#54ANY5ls-F{}`bRv8V_yW=XgJ&`Bm?8^vLOTkA#Uc^L%(ar z6$$Z6KdpZwE}e@!vIhc&58>xI{~f;l4)VD|_^8oOBx~6pKXe;lLF_OpRg$A%0i5MqhD*|SNH^9=X7?Z6 z4&QqQq0}!aTw!o`pkUfFJnDTJXzGZ1+_x@2OLw$bS|@$V5B2pGN=7AjaL=Y+8j(aT zcV)YGB#&xmftEy`hi3WHl5cf(R^pSAc*3_e$+TeX%Q}%`u-R5YPNZAjE5DOdnyi{+ zK77hn7_YAs-}NOwZr=5-=%2!t*=0+H*-NB7o%V&m^S=LpMCC+#H1pfw2h#oDSR4g; zmKl^Vfzwe=c=QMV^bvVo_>>>1hO|!Lbk{cDEjBJHwy!w7Ii^i%?~j`mJ>Fbcbk1X= zQhdL(XH4A4jN3nRuW zqdlmTZod<)qMUzSW5w+0f+u0Mz2p)3Na+@P3mt+?>_N;r7$`Qy12(S>g$>zGo&=L9 zHL;H_G|UB!R5z@!(>fjYD-CuC27fV1i5B@5xEW04-1Xjy^-Ofp<$L+OdFFNtMT=sD zLf!43W6(zCiA3vTPi1F)rZlMuPv6!)l(JC*rQ5CNeF|fGHZf$^#H9pRj+Cu`Eb&=k z!Cdk@4OC7|A{nOVn)?Zs-g9=*ysbp|$9*14MsfE1I*d0Y`9@|UA2;4a7EvhsacjkQ z$H#4_gVmY3PL%i2XE0mZ z<;J;&@2{lnuL;20XSLt98n1U$zj*G}WnQqU7)mMKzd*+TjPe(>G_teA+k6$<=c|!E zo_v`<8)&DzAtQ!5mU+|oYg*mWEUnd285iTH6`n6wr50L^H9IDuk--WE-i?_BFj1p6 zf0rWOCZMd-FnDR~%cC+1?Qpg(e2j=Y*d>mY)eT@}eN~y9tj(z-I$OVN(L9qdpQ$^B zoEwWG80enmR8@|ANFZuo{f6-9dzYviCs}fnfviwlznkMOKoabS-iAEaP|z!nf*Z~jmtAKi6vf@(NP`U3QhVq z-AU7>Lyc&;qZqF->$|$y;w*lYeaR~6MbI~khNLuDZM!66!#{Nj8Yvjxv)t0{d{0EJ ze#xxLI05mR>vAa35sb@T>|Q^&MU7A@G&hzhM5JZ31z|mNGki(a?G?Q80a|LyivxIY z9F!l$EpB!WPA1HcQ&mOL&B#oU*NAX6Rd%NaZRL80v~|eL)xJ@|Wm;nlYOXG295DdM z*y(&%^?eDNx@i21!TCA`cACRY^h~p%vg-<$?BduuscNs~sO^W@x!}j#XGhEE8NGGI zd+?dASr*bck?81ZBL=AzO@OvBFUl`D=g0$g$wZRrp zm~Bgpl?!!mFC~5?IJF@ytiM-T%KU!Anbzj7nMEWxb?*9!+9*|KkKW*{@q``~U4xrh zJfk7?ZnozbO%JbcXbI_WEa`+!zoeZP%lUD$?j-T8w=Ez~HI!axE0&Q(4PkS|u0M9Z zd#D=piEsNPO4^X_pe#o5InyMk`J5fwKvr=Oe^u(u4%v!2>x;xehiR5rSq^@ShUE(P zzt1zBY@E(Oh5bX#_PGS2tRG9=-~ZWoK_?2n5F?7@C!>T5v?x>s#pnvdqd6$cZ)C#; zGwv9A#W`GJD3KPoa{2JS5dwuzHMvN}$5FnnQV%ds8`ed3RwOuECc=Uf+HFNSA?WK5 zZ5&9Db2}r;WVy|Q!&hdX@q86L*vlpfq^c*TjW-7kiznXrD|Ba=ia}l}i^IFS>zfZv zXeUbZeraEwtt=7FZBZ@tR1(n^do1*e+Ld+}JCOBups>mU(j{P{(a%CjHF9z6G;p}A zh1lj7mTOpIW`RI`sQ_g}XRKiGS9*n&AAJ|`pyHCi^BioF^wVVH)948$T$3@Z#QEHI&q{ydTXM=*@3S@Hf2d7N3~oTCzdlq&yIl%0d6 z|LY$AG|k<4cWi35bCucO7-Bbou67>@wMUiY+*Mz|)1uWrjaM7HJ_R9~PY9N&CL$Bv;{;l{dtX_b@R;LR2)9 zkG3>okrJDRkXOQU8GYKrpXBi(O&N0#Ge&Px*Nnfd zi#b$%oUZ)jaxEz=zhsN6!vkK1}>++Msq_iY~k$nl)QKd?*1nchJW5Pgj=Bosb$6(9_>DyX z4!kYPK;vK#6|f-^C4)|tyg~HycmTYJqRPMRHg^2RV*ZWA;|7ECxzAiRqa;$!Xm8xg zfWKZcg}g}%IwUnWN8frb@v#Q-GG_zy&InpTx~7z$7eIbl=~?5cWrNSQid97R2B9yi z76*M3K4{C(^3!(TCKEbs{{Zhb+o+_Z;<}n(HxKkIvo*ig`94|BE=nE|lcqh$x&69CO zepYrlb=i!Uvs}mw2Gl68miCR^PnMO61<@y#V(&u`f(`FdN0P6J!fZvCgaZcyKRjFB zBg^A@z-LLmpULJx+&LXC;Bpa;cVVfb^chLF%;+{^m=(2S)t< zn|YdX-vM>)HubJFM3Qgo+CtjR#KQ}D^=kpj^#jABEqhRjY;*mIa=J&Hl1exCefAD0 zYE;(ic5Xw*X(RjKlUL&RAJ!l3MWGM*GftE904o*rF>1*dQi{9`98WTDA74=V`4S#+ z_j4U`xZqpo3CYxw`Q6h^GgnbCxbjm)TZ{Dmj141FTlm}Q#^k0ln)pQ5qG?P!e-8fH z`q9@@c43l{Vw;n;NQMl~QuvjJioHf#O1}9oTCa~3v}dH_m}wRu5}!r3cPq)NTta?9 zl=e?HxU)xY5)4L}2b^YqcTp?sXX?L$^zvr!TDah?;9Mv_I#U1jmHC@xm*;j5yc(PN z?5(s%0OgiwU)!^nX*vJyadG_^4Zza)QwF%nzESSg)|OZzk@K=j zN;^$2oV>-kt9jYme_hxuc$~(Njoz4&nf208$+kU=u{fQ(+NAmXS_R{6e@buR=ug)x zamb0g!t(l1&dwRbqY_+*_&)HonK7@|l`OAj86J9N`Pnl04#J8j)H;Y_?i}0csz)Tl z%!yU09;(bJ+?@Y4^#C=NAdS#3->m!OoF#c@(6NK0DSG0Ez)Pa;+q1v1gdHz=SkWwh zP&g6t9vgQyrgxFv5J(ueIo6P74YV)aTx}0D>7Og#{$%>`rStO7rA}$YAqN*kd}y6t zrm-t?#gwX0Ilgg+F2cFpHtC>=6u+eK!!&IUUG&1}=on{kdrM-c^v}_{P z?8{Eyz$jw8X)sK{ga4=a#$>Vo^Y;IT3J#&X`alxYwM_sFXfWtttIV@82@y;<@?f*T zfZ(U|43|24)c+~aF&bkG7LNh-V0HHH&!j?_to>(-QiW1YDx^Wmzsm)ig1;B0hhJMW$~8s0 z<-m9tV4{-Cp3{kOknA2A>2?)-Kf)3!u{64(yB>|uI(+%Hz$#jySvX;$!7pA_db75E zc#Tnu8!|kLW~sBxta%ogv!*mX?r@P_3~xl+x-Hu0#eZXl2C+nlU7(uHING{TLNI>QXaFSSTo&& z@!wdXcx;^{KWx3gbje>J<|0OV00y46(j)Qi+dLvnDS%aRr{>C#whwCH9JU8TgqQQ& zsm7ZxP0*~3U%H^xg%pNUhaScDHFmk=kwrH|UjAj2Xb@9G;5FAQk^{TCHzMhB(@Af7 zbY0TGoP33;_@d2GKkwKgrKS6LPHkV%k;!X8)R$bbdhw+zv#5wpbe+J-Z zdc&ygRJt`mz;Y%)2p-1QTYO6Fan385p(AQSr#^{P{nA9WNlb?BBec4IbT#I>-=T=!4{THZI5bbr4zIL@3AjiYdCDYw)N0lu#cZnc8rssZ`-H&7x&pQj?Pwm8CY z4hKAb-5`2UH5bm|R1~29H~2c#iM!QS=qxCec?(9LYz^ z!h>D)eh6=@c3;Zi${Bm`1HPASI33R_?f-SElSfbg=qS!k$0RNTt@&Y>o&_;KlyoK+4z69`EMXMewHqmePf=9SxvvjEei@=*@F^Pc zgFShhT?6H{sWJDFL_0E@JxDxNkS$zzSd7jG{l3us<%g->$z>#!@o{0s-p4mY7D|-? zp>8lC{<1ofTEobSa{SY^Mii_?yaWgyg5t$v`=u%~)^3gJhcQpXxNhBZq79G6@o3x6 z_b}`+IEgIhO|+I&B2Bis80GiW_p;u$tR-zP_)yMpNW*P@T7D-&wF7o=y7L4`useR%sg+%^8ezFXmnv#D4jx=O$3 zXJC~l-Mf5~OQ;yZ-P92cH2f?6vhuq9q)}h}!ZA^MOz)z4Dp9QWu5y!|6R|bhLheB< z3eswRi7+p=vj9pzN*G1jy`L>P#5XjpE67@WPr98UwV{R2U8!y!h5~AT;OaC@S)RIu@oC*& zquRBTUdw~*57vkBPG)zVq7<|2UD{$`{g#kgHD1|7m%h>dbvm_F6e874Xu0D^m0)%O z4SnH?_u)+~U&5~N0Nu!|JEF{`%A5*}4k1A^Psb)FAn{VK#8DD_xWosL2;&WyKrC?P1b5y{pSz_y~VG_))%%=9e`emdp^ZiJJ~g2Us*!3LM24m`)zb-%fH%! zId<7RDh6JvBa)FrJ(S#dxsI&1&wbRoTwS=zB9c_u^EZ}IKIb>+i=G+@2c!DsuVaZ4 zigG;ZsRSWTM&YBb#c{@YnS@?S?i}H2MkQC{cE9BIj5rLs3s)mYf~+irxhl$d>K~_y zIv*e^wwhDzPp$;81qKV*(64-G>P7!3||+q%F*M@{qCRl+M;acU53c1^C@RcdvgU7 z-tNQh3*acvf!fHZOcD5E2j8+r%1&$R#)qTlZByoV@wCi_4sb~`L$q-Xw1PO?UoVjF z5?|eUCr)wshP>|7=3xA^1<+>ay>t_@e1n_fCI4PZh{nwff6{00`1OzyJV!)bz2{E) zqv)4}PR*iOUV$7Qy77}9TzstwJF0`%gkc=?(W=Zl%REg==_mp8k3P~}vk$i{Pk68_ zUOY_f3f7!*?d`2PvTdUfVrx>GlKo#UK4)bCK<~~NP9U2WLw?GEw@S&4{*QwCPMY>6 zs0fBYMIa-h!8KKIgn#!Wy(*;FtB?l(PoLj|v5-IPGDzJ(LQRiBnlOnR%$Q(oCon^f zta%Oz*WNdvX{5T=o0vmhzzjc!MYPOp=tSpj;F*<;D2%U`E2ofGV!B~-xtu+l%3U)46QW2d~ z@PK%=E!I!8B z-@l{5uqBvT2W(=;-8z>AT8O-j}IPAHI^{YWoat~RmvIoTzJk`%^*SZ@l% zdslyLmVG>kaypT?R$fp@()}rKH8tbgK8s2bQ&8%m;_!t4y(u5+SFtG0r~wgec;?NJ z2gB3o=aT$tMSSB!c73- znwNLsGP}duafYYt2e~TNoR%}E_>IrBcSEPj%X=%gD4Qgziw3-U6;n?*Kws(2Zbxro znQIGU$Gqs#czX;&;{Aj)b79GFSFLU>81 z)55g&d+4>G!J-5Xe=&B3aHf4}&bZw)?$n$MheJkh?@o;2fz0n0y^E zZAu7Z9u1m~VkkZyiQ1f=+b%}E)+i}kZy~1=!8R5xNBHvSYQ&?@gDe z=IYk`Rt#(Faw7pT&#mDxlcRy3?lcOwg4RS;2C>&K`6mvRtcz%FN%aiN@)rH-;Iz}( zYZ^OHA*)P~HXJL;Slznoj99Gi3^WT6tz9^z2}hG|<|TJF*C-0-8qqB&Y8PBu!kzzm zt$ZiYqw=Z^biq<2p(Hy?8~N!m&c2?zy1g zs*xf?m1F6J+?>s<&Rbhn#-h}NRqp(u!+c!iiP}r>dfuHD(Yk6W%=X2#a*;VJ?gjvl)lSvSh6P9qZ zCCDx=-9ZniL?Oss(!&v)lo49+;74THHKrxIUUmkQsQ1A zFX>j!xSJv3+e&E`t{bnZa93|k%g#}@85;FUT4a_RhNQjeK_PsKd~M;@|%}&eDW|M zV@c&So!sd?BYNl7GkLzSGR=1n?;0pi$7UuY_;<*PBNWW)Lq~x**2GvPP?&ykxg@&} zh_Qv_Ow-K$lJ!m!Jo(jvijaI0t0PmG|N5l!r{fqUs9Vjg37tEnJtDTMd-HbGfbIn|& z!xZP#9;g{7N7m~XmaR@!e^{=|w2T--HhHDuKSUPn+^tx~d1s_(iHv2dePq+JBEEUx zwOT)OpyT{;x#$|8k7ql0-4-0hR$!)<0h?T$7ww!ct2zO3gM$av*povYsB#zab!W$~N*`@_aBt z&VvhI;A7R`XTh?+1iLm^)>_xP4>g#Zd1XU}e1$&zggc)uycCh2idDQ*BSswpu0E` z!?eQF1@E=rSeF2{m+#0$>iD#cBYBr^u#ZMjWEZt!WdI#2pM6D_Q~v5Wo)yIceJ&w@ zrY^@Bs6->hVrQC#k+~*MOW%_PbKWnwO*t1O6J7PxXN&l7H#!BJvUr`yEOTl_p!5e* z;XY_XKA9+&SwNbFeV;z|Ii0>WM%i2F#wZ_y|H)s*xcuR#g&429AQj)p?Bx#~JSd_m z)Vpi5IADrm8e|F~HY4qonso3ZlwOdAFgdH#r70!~&&8S|r&rYb=Qojv%4^$p+X;_a z=K&B#qr=?dx9&)tYgV+-@4L8eRJLg?;8PJ4+w1wgZ9lo4QiI9&3(G#{Q4l* z>wX2OZS`>9WpcvG!6|xJy1(2328V z{BUTqlRIl#bAI!&X_2n1=(in;eBG!E;w|lh$3rFvCoAbhrQYN)r6L16t14+{8y>5D zDQ8qpl;6}se1?)M&1#BfF{TudPFDt_4{CQ7y7GKi6jn!wa!fRX(o4mek5XixCxoo z=bC*>H!m%}^8rIjIadwUG=8o{-}r^~*}FS=Y-n12wsE&iD3moT7~l)IFBheq83u&J zd>B#Loq>sOvc|67$xDlF13QgQRTr`k@#JgnT+iW}3HtTX<`^#!9I*X8*eyuBG_YYb zeN)U4rC_r6sauZcsFv&~DCF*^FxGdGG#``)xqc-U1gG8F%^@2UBafI74$NYxkuPua zw?E~K|KZn}so-YA*fyIkU)jwc`K0_Hu;*K^%o^Y{jWP$~r7|95m~(Z-W~eM~7C&pP;Tk zLoPKS1<(lt>cuX_64|B`2G4todEePOXTg!Hzw~$A#KX}>aHR5-=l7V(}FYav$baXdAgIQ%Jj|+hgx{T zZ@E7=zA+pi9Rg0UE!~!Mg$T&k;>fhj$n@R3mvr~XbLTXasl%%tpOTN2s3pO@6ir4nW9U2VPN$9OzP4 z5>%_w_1=0$)D166zT(T`p=SM|%(=#FfkqHFkWs4+HoKxT8bA&iTh z`l=o$G8Cn-D}s)2d>&O@Z$}4PORR~HQbcDk$t-KLuE^HHYbuY2jH;qM$>xG2D#Pd| z#ye$CbEazGt)(Y!vO>87?DB@mgXXLCU#35B3o?Wk1LdHGkpmrWVtvvWYoEh|6O9Jk z1A9(znx3n}UE?38C1`b<>LGA+CZcsO|BHDYDdaAS(H48}D2rg|{kou=z&14~Px)y( z{YJ%#OhnU`$Xv!3k)@B}we%;MqMZ(XI-^d|)}N*xEcKQwHI^+SDRo1+_G?_P`7{G+ zu8BDl9k=9*nMbqO%3l}7HgBUe+nFp%tkqi1{bXETTN#m_pd=tAbZ4zodLQ>U7En0X zc}djPgA=M1)=p=R`gc@T(K{)5F4G5{GbCji# z`t2Vg4qk;>SUQsl9{Vblhy%+BeT4d#w6Xe$h2F$cpY6Nfg!FBHlzk9-AcSInDqx7# z<`HX13dCIy^u?t<;nG*Qr0mcZaBKLpad84mxENE^_rU~8q>{&jZO+%lG`dTUnm}&6 zJcME#@j1N@UYgHgFCCE?>!G1og+UYB3`{n&C2~IW%k=Fy1euYv70ldvR=Q>;=hi#+ zKJkpv1f5ptP1!B%R5(Q*1b=D%$m7oX3eZw>-&ZO zI)e&f7Wa9an+M%LUUn_Fz-obG(^_nZirz#G}UL$|c>#ylK z35=e)oy-TFf>v%CV4JbyH>6q-x<*iEO0+HmEm}<9fCtTVlvPKY?Z`AQDQZTmH@qn8 zp%_9>NzSkfkx&wHmw!s{wI*PJ=X$v{WF>w6{4?$)sBg{ z_uyAa#-Aqsa^f{y$!a5Awhuf#96sj{6qcB>(`K9qTnnq3XXZJOE8ejV0w_ANR zROV`8g91fE(stE5G5b7@Pm}em2Pb^5vc+3Vmw)Pfu|QDK{Y!UrM#QZl>z5_uNu~A1 zWuuoy^KJKE{{f*#eVSk0I9gtW6Pb(&bC^| zujW6HdNcX<=J`gAz})7RoSv?(shP8-`-2~O$E3@PgWBIue_l8CD9=`VEms$+#J^^K zA?j1p&f{&ds1>cHhFOSBIa7ZS-U3;q}W>>SJSbnJeJmh|`m36Hwzo7ZI z`|hHWcbTNO2v!_W`ATJ#K`&nh?;uJkkNb1cv_{fwg$9F$6;mVQH=I69&P-EI zZWM;|L#kW^O8l?`IFz1v?Ona8_i+QEb*Uf8-D-H7)Q*ZyOt% z7zS2ljJ>p$zmVQkQ0m-J2Oo;}YcNC*xV55jqn-D-N>J2>TbHxcMq39*Im7jo!>+G% zYEj{r6sR(kM2=gJqZdd~B<(PaQrbeX-Pf z$2Ol7X?|R{&G*jCi9EX~(@_0;rTnv^{yF#)N(%AkRy|t`buZuP=;|#+pnOz_R&=KL8lXwNq zZN&}fcl_e-oL{@bd0)8;%SPX@EzSABvG7(pA;)zugKe7sSmO@gts9?THx)8el!H}{ zv2bhPn}7$6hJs9tHX|hVk%g&L8pS*G5%aFYQBPbL^AshE-BhD_e2W4~4{qhOALxGQ ziB_s{J?=RpAjuFr*NW=bNMgo>CuZ z-U(q5j+IvoAkfpLVLS}5Aqa(SZ-l!%;iHig)>Ly32Velcs?ye64bR)AL zh|={!pjTQ6Y<&AjnoIbq9|+$Jl*@-qs|a=Z=1Xv(R#n{YsAX6Bj`p%tohRL8Y;UEXZS}}M{&%*2BwX+eo-H;y`ck=?LB(`|7sOIH`nGjLu&w7-YfS&Z@ zqnS2E>gKXD(sSU)^^eA;%&$Al<#EyDX(%!zSEt^gvTE%Ey1N~1!i6&_Vprg1xA|g! zTpwm9s}&rw1p%=wrAECC2{@W{)9hPB?n=rnFmL0MpzS-Z3dK-^q;?u6X=`u3j=B?6Y|qfEi*Eo| zMwU_^w4}vat3tQ8jtM%o8re1L@`ICp{3P3PZ%4N3EZe$>#uV(CcM3QsK<8TPHhhB+ zW~$xD*YPc?sFIAJqn9W=?Ln^Cp&8?%@32?v;`(ZgsE5gm;pB#rfoC1Aqn9#aWqi~} ztI&;Li6OpgTnkLcOqosv%CFp4J-YEHP%Y6n*-e_Sl2EVBqyc7G5K+ih$E9nRVbp%G z{+JI$?LZf*;|~V-)X9)5TfiX=EZN6Gl;;EDS3AwnQ68qD2p-)%za5gGeZQci4MxMJl) zb|4gfp^bncwl*_O`qTC{*KxE#_~e)Nb>PWM)qlQZ76x9|M^JRppo23R%MB>bSZIyo zML1K{5SpK-6Z!ZBeYU^ar)~j+5l`{KG&9i50*BZ4Mpez>wx6~&-mHG(N0*b3hi+v^Z_x2h{zafzJtK@TJ%gAq~I?LF%AQ=kfp zyp$=@V(>|iUz!cqT1AeqH&0d>e^a3;pnU3lOrQ)nKq99&w<$Vvk%II~)H`hQ`4eC7 z& zJA2n}*?DGL%QaFp-ck2b4OM1Dg zsj7%t#bp^gj*rcw=FMqVch;xeq@ik-TpN-O8L5Rr2D1mIHpGwI2%LypZ~Ah@6)yue zE-dP)+SOo`CHzA=vFBiE-Sz6rwuGrj5y=<$c5XcsL4H?#l&?GlLkFo=#ao`ir04dY zGSqqtj*pAI3fC$|_C6|g6>|*fk03iy%El3VgbuGqy*KgHq|RJLaT)lLy51NM>3HFw zR@u+hWtAxV0Pb5dA%WhwlLS%|W9(m0$a4``7BPVy$yZ#Nqi0xJM_kr1M(>}0T-1CB4}M87QJwDY!RPC$ zz`sg{vi7&%omHWT8vWpB>qJg*t~5|^blMhj_RH0=P%VyXdk2{HD?69;AH3RZHzXJj z6#XANXj4y@z=a-1s!b7zy4bwu8^0_nv!p8ENumYT8>QN7rHgbG?--9u?#+5g`QbFP z;@(ZUCe&YKRNy)dyEL&Zo08KjvHwb))z!+Spm~z9-*>w`BkAqB&ZC%aA9C+V@+tAF zz2Zy2VF|DfVk}F5$4MS|!DAyiPx$`8)@1fSQ?~Mdg-M=H-##GTS6a=B0B=ef^WTu? zL(0*l-+u6_f;7Rr?ZmnT0&KCMtN?8eGath`LxlqT3VPrh1qj11@j`#+dRQ+fjBhaU z2~AURSQw;cF$Jx$Xd2?$N3NO@-bJGEIGQTPv+reI=!4BKk)L1|%YkgW@Y&)I*3ce` z-m*U4{0k2wE2qhXd~W!5x;DcJ(if~IxdzS1RIB+}yCcrQRhKplcpBuAJUU0AooM5E zOLIc=1Iq9Kr25EFX@Pr;{5wO2Z=T_>KYya~f^ z>CK*)V0#AhyxOP>M@mxq##qxm;Jn`^s5h5VRS_2P%7;AO&9BR9X@r7LE2cQ4ZpyBF zCG>}zV)+DLy+9_XqV*a@y*L|tQt;5>m!Vsa88>96eE1SmErbFKZtDxXj7(~gs~$$v z+!v^kuZwx8m8qptoD#C_E0w!nlwo*@^*j0godUYj2(1aeJ|4G-yb$FJixJXoaFP3@ z3x^RRSDhi9%y&d@dj1mR z+qXq4BB9Ycx>s7zxqX#U#hiq$*0!TN99_e=(BbEMk<^xm);-reVi@jaZ}h!_8c&$7Rx%Hmnly7_aWopi-t zoklWkKPXEoppCc8_};i=Od0d9WLKn$t-X=IU-ycmeM{(#-87VlvcF?Aw+BsvNZ!sW zn-qelQfb8#^Iy5_M**Ck+C7B{oH%TMXqdffH&%NUAho|5>fplr?zlvY@?L5qd5T4d z{snm{rUX+}fcOuv7PBI4vQpwDwA7U=Y%_@63K{`^42+=SdiMF(wvBo$8~ZP*-Dd7i zMiGBdVq718sEim$AIRG0KAr$>;!#SUJ_%;mnZ4zBT8@tjpARz_yAO4+y@1~7UA^bQ z|H<7f=Fr}5)7VE#kl`~jxB$swyt)GRp4gff9!7??K?iN9e7RIz3t`>T-Gf2n?0hg! zSg~0920%JHFy86#a7&nQM7 z&iUXY>VpdPn$d+T&caMtDGPbZzogDt?1v5gAgcLLn9b=6d+{6cf#bqz(r<{it-*0Y z!ZH!^N>jXqCR6eX34dVRN3m?h6?S*sM^odW2V1+iL1hMD-lYa5$d^A;{{!aDPS$?&V_4GWwdYp4=nK z#XoQeDV;YG@UU@HArYUVA%u5-4y5U@CFr z&5*o(l*^0YPw^l8@wkoKtV`lpQ?#7HlUAcqMuT`%+GsaHy}--&kyAu16rPhT`JQ|~ zwQK}VCH-1pC|YxCKPedUReJvXI}^mFZ1;*m(B~R~`~{`sJzgCSr;T43O^KAK+7C8_ zKZR32Uyp>wN+ch4EJa+jv`_o-lIC;wvvnD+d(CxU^Yp!zn)UQ>7p$$f?krDsX$rlz zabeh)vK3sYxREt9DIBrg<)kP?&bfX?+TJl`4*t~t>G%+%BO&Sm9jQ(1{Ffc5vt2{PtC9;SPxiXJ2BHXF1rF(K7|%>Ir~2*b5W;UkpI&F*b;k zKbC(17w=K)H4Ff1Bn%xX>FZN{={Tfl&e`Vy+X zH_%74ei=slxle@0J^Ly{|0T+a55@_3VTZ~RZyu@}yt%!sCrWA=u7o_m2Ib`_y4O9}brvQ*?kw-od*e8~fVmVsD=DItPqZRi`vh~&Kn`qgZHY zUq5I@OKQyescrCOi1~c*Ygt#;ENQ%QXF)u*f+2;TOPPYWOZ_Wx?~^sD*9d3LIl(&FPuUn zxz)z)3s?F*F^dsSz8BMPXTOvT5>*3^d&@V?leM%%m#jY1cMVJieXcFE+EZ49=;{YFMAr@)sj)5#g$2uuA6ZQ>=d$N{ z-|)Sp=r&=nOxPO(BCkJ|P5Cjny{x{f z$hFs!s8-7F`vI}hin6|OMrQhsl(T3s@SPm{x*1rJ0cm|~s=U$R95k3}l7wb0jN|J! zTJ{$uO?y`9J)2*XXUv#K$Yq;PXA|7c`)*gG6`JZ~yy>%~r*>31Sa5?j=k66*WRX0h zU-240l<|I_Sr9r#fxCGpcR(=j}eN<|gKBXLs&T=r_Qeo+VasWtGqP#y=aX4>_XPW&~K@6R}@q@1^|&OEU&m z(v3_LK#bpadI&ld2L~ob zgij~G{KyUGrn9n(B^t9|LgRCV``))l9j$Wc4Tv$eRhJ>!1Kh3xBnn!pn&l!zTGWEs zVLw+BkF=G~z?Xw3pOTes(zx~Zh4^;B?=rvq6T$$XlL3bX7WQ=Jzd{ld&OI5;l6ND3 z$nj`JDplJCU)`HYp-^nzDz*?ER=XkPkI)2ApZe11%FU;}7Tdg2z+^G^8#0(cK&-jJ zs0ikjeP|7ve*7&37y%?PK9xpnFL~Y)SO?f1+hwLAY@*%U>5;MtmpXr>-e|OhSup`6 zYe|46hFOIkOd5@v__d@6`RSTU8<_G6&N2>9vGe!;MO&g0B>uIc*KE`O zI4wP4w87I=tJ%RePQj_k%`pc?aQN_LB4y`l#6rjLo9oY8_XGtq=eMd|^lWOvo4hUW z=%=-8>)0GLxeWwBXV{x!t_Bh^3|tX()M&w}6SK4uB@koiffn%E2)juy6bFU~?Y%b3 zUjQDF*0l_kL(^|rKbuKhEDES;Z9dfEQ+bzkO3a3Jyp;q*D-aN$4zwR2llYGoLgI0F zJjieYt7Y@gF@V)7ny$lm;J&Zao3F;GS`+kiUYH|@&jFlG>VQ9mCIKLzW0siF!102a z5ICv62mYvFxeDAt0a69#2&2LOM66<0va;g%xsJeP;s~Bj=5mI#;EN~Be^4tZtu{N8 zzVJ-hJ%>fs{4M@ms(|#ga8ukW4sPf>Z3c#%yiD~8^m!k#kKJ2#A0Os9j`5EXJYTIr zdA}ehzGmOrD!P<12+x!Xb2U}d(jxuzW=S@9MYgK{>^@aVMth}EEPrn11G8r-)gNx` zQk9G@YvT?FUAD-hDSiX%=ZMUvRzN%Pe@Jhh&L^xHUoWsCvxdi+8gJzZs||drf3|u- z$EPDCzJ?~BTm5BpX<#oRe`>C0Jza@6v0Ht8xoLSc1O&;ntd<15g#phTx@%%_VIIVN zKSpcLOWYU9n9J7Er0=A0h*o&h>{ervGGWDz$k=ZZYC+=Yne2AnoPEn#HnGblrD@@y zUF`H!B#UXW!)kXM7^@P!lDGrOA1B7Hjw3Q`(uj2QFIDyV6>U6Psns^d34N+ zN9_9;If-1XV3t z27g=H;02PNnOP}ybiIutBAf2me`7m*{QlNxHXl4r$(VphT(Hx?$FYL*GCcn2qYQn9 z(&7yw`tihsqG(gsiMveMM>DxI>5BSV2-tVQ1+R29rJPja{3=-50zF>wfnDi&!`6*J zs1Te&G56y&D{aTcI$zPmIP|?_2z2D%t(cFc`alN{fe+ak1E=;CdI1m(creRE47ZYMcFpX zW7X)2g^|FB*&qk$QFiFU*k1B*^U=@@rycZY|2KqahuOcN{;hI}USMnKM!5;GS$<|> z31RY;hXp|fj|UtUI$mf=A2bvHhBy&zK)+F!W}*fT>bBY1Uy_JqKi=*~Omu2;G#>^n z$ld-NzsC)mV#tw9Gic#%qovBT*X_zl+j|nWSWnPO!V?Pb)kc4}{USQEzJDVHL6VL{ z6SWmpWJ>wg?B7&ALQ#iPn}GRWeD9H}J+Vdh)ai*Dz?N>Bda=#3H{<7@C~1stqm8Ke zNMu<5X>_;==IW?(%#`#~wUgvkOMkp~x#KbLAmSWmm+m<5vP)QztZ#DVby(8FooSl- zn800@-HXcl_2#T`IOOT7CQvo)Jhy-M^JQIfCn8f)Fr z+A#sdQ-A#Wo3l~CyJX;9-+FvVmKQ477ew=PQ`sXS8Ah&;5!+ly3*3H70owJ?V~h1k zLhi#N_03Bv7bx9$h?-K!TrT!KMK0n>q#IEmJ+g!qaS|a5QC8QT?yc~k9pAgwEfk?0 zam;zMBkU$|C=N_5*BtPC$=AqUf5^#2v4kq}_}{qYHOAWp3D3*Bh3tnes75*r45KmT zNKBCC1z%%8%J5N+DHXGf3-P=b$qOPD@KcFZ&VDr%fn*YF#bx~Pbx=%SurIu^+uFR7 zwcAXb1Lm2tQ6j0hOdmFxvhJrk-Ah~Jqa|Ufr(z+|i^b!*oCLUL?t2gu=wSfjfU3d# z0r(Z5bL9Ug1tD(!1^%4;g=PBJS~2--+&N_h0)-*PX~IiWMi)mSQB>4basJjq zB_x9P3Ixm@Tx3f808RI+v78GNAYkBY9>M;;KBAT538Ui&k^S5$@)c>7TK-RE$0CgQ zl${PD1q3*Z35_P+X#>mMGqB*hhC2S=v;ram9A4I%#oxKSj`YyXPFc*q-2CcEc}*zF z0P>i#VmGe{aYvDguM%!ETUQGP6|G>-_5KxK-}1C6c6t|-4f?5DqEeV^c3tFA5v}5; zVa@n`Wts+H_%`HuhQHG&?~;C~{E%hGFr_G+=FU+ClI7wAZ4(u3#l#EAFZd}m^;tm) zyah<%60evw^@ZJ__`x)Cp`F2;(QV^-Mt|)RLC%nhHmlLbrfkRc^+_#{huspo6L>BH zMpt^+&BjM_QtK+Y&?H|*8N9CqM6V53d{Z6bTf85{W`z{)eJ#$R`6m(?C^Bk&ImSME0_n0-UEl|_L^g3<4TTdS!ig)C! zpIop>b0=N;jbjb;a+kVC^ylV_Lcq_rO4_*De1~h@giN}nV!-s><*(m8S#CI9=#?m| zc*~@`b?fO~?yDeDOVUW_q8l;CkL)s72VCuNe$>1nnwNZ40Ju=Q*=LdR*N@3|K3y0yCUM~UQt&an~@^9&ki40{X{IX$sT?1sgpRPMR+usZf z&is}c-$;Q@nY&Z|NO%o`qmk;T6Xhk#%7qW*@lsdD$J?Nr`gIhneK#M_fU<5PnwBcr z=To24%Cs@6ZER-%XHc5R`pH`=flBGlLf!~Zl;op!RJMqb;lw7;0xaTaO%0CNXZAZ? zQ=jqzWhG&pc?;L0V+N*LVNrFAL30Lvq6}>~!)qXz)_m%)N*7Mc)$Tgib6 zL0A#BPb%5wZjiE;*th6FnZ0PFoZM&@EI7W-zLh=({v}#~NN?DGT)u6oAnUk{Lm2tb)>o zA9nH^M8Uuvj6TO8oy$lWVDs6WdgpS`rX9NGZjKk)#P5(AT|?DH-BovGzXTeLT5L(? z%aRW|dJhv)c`C{~fDC{>RXzf4iByB>NCeD{+d;}fuXzpPkj^3M{je)}W2uAaEAJ~I zJInkAqrs%f7({REm)-faQt#PVNI>ZV-AwM$HqoBeZFdd$S9v$x_9hLyBW+ujrT+Ae zNIta08cSV~`LcF-aYgjpu$285+?Qzq{6_g~)~h<|dcq5H5oSklc02n?W*DtHOPIcG zC-VFK+VR%cKIq0P(|J7<#s%Nw?1(ju?-(D=Vn+MaD;>&L>Y%HA`(eK!p<8DAny^?@ zOR;TInOa19#0Jf+0e;iYh=gTiGj zcUsNHX8#_V;a;Okh;%r>IQo3Ri}Px2zt8e z4Xh3Rj^>^qobAzxp-<{0G7zIX0LUkR;k5_PhpF|MdJ^1chKJ$YWEU`3l=T&tB8_9V z$5K&Y&@BPJAXsGlg|p;f23)cyYT}Q&IQ`B1%ZjIO4Gs(VjE~OC0)o8!KYAgVlJ(2x z!iip>xj#&ffiZmY*&DcIs}a!R`Z`w;6) zMZw(L^d$U_8jka~&fA=el@BW$kLx`@e&B{C&(Ro9kA2EPqc9*`$*g1 z>tSVYQ8DddOjfzxW#kYH2I!gsHyX9CwC*ck8h(mTK+|;q+LFT&>%fYe~+f(Y^YV{M|8`5f2vq zqEj=ixSq*4w%;dy(e=%@D|x?mZq82eyg{x-_*DM9JRPa3p=bvppTl5XQZAc+Z@u z$OssA)oGA~!mDIpwai9Qf;X8OR7W@k$-9KzY5S!txa$))5zG-D1Xn8s8hUcS-n z3=h@Whq)5aBsUmxV`5dd7MjG~sZ-cULq$7`eTZ3{2PBZ+cWQ1$wvyO#%s<}FU7IAA zR$y+;s;JzJ+2+e!7-u@#prsaPLMX)VAma~OKqyhv{o>fEc@0g$Y+9z2wS(j^ltjvv zHW^%IZrBK4mTd~9T`2m9dRlr+>0S->TVR*Y67@eG+hCVtH9gz^t4xA@_l>m0P-f!Z z{W|ty=4jU9daI z*yle?`BZ_pgJaLM#2VBT$FYtt8^4q3F#^DQ#M-6D*nwNj2ux4(3p9Syqyuq@Y={=E z&qq^R0o@(}$R?p{5mP&>b+I(?>rnPeoZaLQpiOW(Iey?x^oYpQ%kW+N{u{zhjuKA; zJ*Mul3UneFA*H;Yj@pr?ht4CqpfynldLKp&Ct7%G+Ckq$2__(hwlh3*SQWFBd# z8y3%=Ya&Y^k?ntL%dtN3?kH}yZj=nRGg2p~nlTY{^bAI$Dd063@MG{RFGqmVDy&@r zjT^KkTn)>i{*f|xxJ-xRgou^VOnZh{9?Qna}~IQODV}9$uJG@G-Art z)`5*g?XhaP3hqoK@kD(K>(G)9OHen5hCXMWJ@0NY)0BTRk1BV3tpD9onFCQqOnL@Kl|%W zWG=?L;U7$4Fvg$6U;o+U`X9e!8w5WFptCgewwIH>?m-WMs~Z5-3AXzYzfD5DPO<3i zPzxO9KV@LN9_|CkS%B<3M8)U zx!(D}WIDi1y)tg};{9BlF;=ow;76?yf;>HbII5n7Rbe zYzc>U#an_UZv0k`2te=$BEO5OTcpHX!o1E3pu@Pk(MLHVYJFdC!Zt(7P-mX~22~B6 zrnfsj2fFtZ--k|i(ld7GYaqZQ!6_cK140o~K{9b5#NaMM8JmwuuOdzbNKPaQv=;Y# z)HbZiYeq}daYY$1|8+|eVgsxza#OpPi-6or;N4&#!c!o^AU&}TVcntZP4t^m*g+@t zW{3dlOb#ahh9o~606lVn7SZ9Kbp1cHvKDpCKp7hNY zRxdNPOIQf9vIM1o9iP($r6BR~1wLP%V<7YJy2YEA%rWH$$@621ukflj$c^pX0+KG^ zH$;34ikJ@tIi40+=S#0I+qtQK?ge2b#$;})Dt=mPSOJXx7R6*H!jr!aQ;GPP>_GtN zpYmH@^M#y>fvtnSle!_q)QvR}ad`r!LO|J0E9B1yX=o(IRtPJokbZ21{P`X_jw%@8 zfYuXV0fGbJPHX1Bgu{MPtTcQ+vL{UeR0UQBY*RR?=^Qzxe}8Q0lTu?6#(YRnxu6_a zSuOrmm4lST*zXU7!&E2uS{UJMKx5H)-79`G{=eGQi4Y(DBRR~sI;|`cpZ{zzr#by; zF@LL}8-0jr7k|D|*xzcve7}=wz*M5C>fi7FUxhsch5gg&G2hSk9~t=mw+z^Z1A1By zrm0|R{IscP10l>Y#f=Z3C&lza(7>?e2b~4Th&eNl#CDw=k=|%*L+b`_w{s#Ez?uqp zP+)r^wkfhO6`pZHU;C4yEau)xwE{4{!K;EK14JM6fGr_JW}ge+@TaWM`{sUL3p~z4 z*?9Du2yBN46~jl$Kf`pR`cEIzLKtor+q@rZ$-S{_UFCjmNZ&oJZ>f!8Qb|hKgBJ(L zipdJ|t^Oq|rYOQq~`Sskv2hW}Bd+s_pECojwxc`iGQ0z~ssR$%u;i`NXsQJ|Y zm8XM(6dv(qtc%_Mg=BWJlLGwh{)KV|5%R5vyVgan=qG37P)ehrbI0(O6vICY34#UW zJ+N~k0d`K_f|v{p-*&^vatE-D0KV9N`yK$&0^m0S);@BOkjcSUBMqdzyhQ>uWSM#W z^;&&p@2VxQ*SB@4-OVGItN0cD>sdYrFJgFo?J-y9jkd43Hu6nEU73*gx@Nw2zqfef zuA7Ya;3xlvm<@NP8!bgTN2s>qm2R*Q3;&|6uxgM;h%!JgjxYpR3hcjQ-S?q0C}^R) zmuF>1wNiGK*M{zP;WHPVrUk0`LWV)RClU3sba<;8>~++R1P?e~dMWEuQy48ByM>Q9 z6kZ*_^+j8ecjadVM=G3?V~^c=o|AnB$6wWJ<&E#mQ}}wGlk>e|l zGN0J-1(yvfO4;<2;GZar7e- zvpbQ*HOIbsXL-PGu`ieSr7LT%E&w)2uy6)Ex+e#i6*RL2?|=Bz&=Z5XnjDsAhyGtr$N;a&$aoiRpG#AW6*j;WCZs-m~lULOyL=}VRICv zYL_d1`SY$a#8NYXW9K@$ha@V)DqZ~0)lw`pEp^uTnUiD$_ZZpSngM&d|K1p-=(EIf zDW7x0qcT6ZiA*IK;(sZ+$d~9_zxp^zLSJ1$dx`tiHq#C38T#Ck_>Z?EE;^Nwk!G`f z&G;s@P-Zn=9=Piu`qo@YxHq2&jyp}dW#WOsjK2y#v-KbZaM#Izfu}AE1l&L;ktePN3Wh<>Fb2vsg=5qrF7fk?e zvDf$;MGLaKq3ivb$!FfkG0_2{!BeA4)F~o2#_j?dM%sUD@nW!o@DTN1ortjWZ~gWI z??=njTfZTkgK3g?HaFy%}q60#^eJlziQyLA%9DYJpORPo#(Nh4C z{DAtX`_v}##1o+ZU>Uwoe5wHZdkmI0DJpeC{2cT78H_Q93s_)2CxG_&r;(n+ME<|S zIK!RAOy@D99`Kd}%mx&gM=>ByV2A{8gh_9_vCa9P5%>0A#tCdWPlr#;u|L>_{x5Wa z(+-~qIvVx7$;#9nQ8HGA6imOVzLw)xfPG?uT!c=%xaEiTD=X)gq%>*XB*dB3&p*nJ$YGsxiG{Hbnj0-hqko^N<+BJ4?D>@X-Km{&j|RLQbgOP(S-8y)$gD{F~1NhU2`G zOdDrq8ZUfA5uYul$#87oF=Eu@&VHA|=|#*%iIgs;E=x~9V{>=7FpRSf@%Yb#=Wq;~fBZH!^@&w;?{n>d)pF+? zEtgYszMQ`Nn6+?O=ZX@Y+L^n83Ozg&^7ldCaSpLw~x3@thnCsqpSZmdG^Z8Lh98~Y~nCRT7Ox|XSI@ytq$@jfHCm&1J zD1?-~Wx7$D|JIl=-p~H7nCDG20QPw$B*A@j^;(;}U&|afdBlaAb6PhIsOGaeZ=n_Z zuHbaAd48aL0j>A;wb`W>zGsp)S-b`*gK@di5;qoxO-Wm_d`@v5dn=@E5Yazjx%8wl zIkMfbFB|uTu%@ECa)sEPw{y7FT;AhjO)s+o8P+@Qby(-J(|iU1t~Y%KO2f`evDEF` zER2eEv5&X={~8uV+%Xe!?^q%*x#5D@OU(PZCgSml$KWbNeivDoDGp6xs5z0Bhu z0iOV4VP@K-u#BJWT(Y2@RhERBhW}MK)xh|0BYR2q*ilATi0!p1Tmy$C1E(Xi3)K1E-0-1l`GAj z8GFfg_BO30rKDRI+q9a@lP`riit}He#c>(r_Rnmapa-0@&RPybgoZA@cwnUH6n+o+dsm<|Wc6Vdfx2CUZ!YRy=tl1x{Nb?|BPpq|Yh@Q>uG+~&foPJL`se&L`aV^))I;`&8-VJ~Z-Q81%5#j$aM zfx{w&xP%s?p)h|qaIMQ=(n`IJvUc^@%prVmI-xegOsJth&y{&UI&?takKa6_5C0b$ z3>Lb!lK~9kSZ(u84LD?jsTD@?r2A01$^2Suijfs3+5ywWxF8It8D7KkdEHqEi2tz+ zQ!B)oaSbE6E`zfy5N#L5h};9Adc31dg*YoBi!X{~h!9O)yeYHojHq2JAw9Nt!kwXL zHz;b?;-BO!5_%tzb(~zbP;cY$R6wdOcQZr?@HE zU1=BA4(m(iyP0Q&%0U#m(hfUDhp+morqe6yW#F$UGk4y!qGy($abcz_#bYDRFeTn2 zxy>eiHU!^H5jPH}k=6gA;LQi6Qwn`=#lO)JD1a%zuP-{xOzVzt960~=wKddYmHwKH8H6S~f%YX)WQizjf=l^|-`)4W@ zEh!Y@Nv;)YBkQZ6h7)7?O$}I>K**v}(&_)}#Gk)+!}OepA!ft`c|fGz!>c_$9g8=ya1fGjY}bI=0*;|~adfh7Nz(Ep!~ zF%agMjn0!~PdTTFP5nXBAy37u4^1Qf6p0sdxukpY%cLA^Gp4QJhdAac!|Or=jf1&c zEzoe`k9|o+?~82ELvzymo%sl3aHTh?2v%wX8%1WIM@oM~thZCnMOd;5Bc6<*^4o&e zV(j+oSUAoFki???yjKR^BReDI+jaUv^2xJ)ZxSVeJ?fc0tmH@y_6(SkN&x>;L9%Rg zBB}tBrR>R%oK>oIdPqyBw+C@S*y|CPDU7w4b-vVHhI&l zpa)C$>_B;5*=Zp&$4_3<2hU6Imy~7yO$c*E7L{wWJO#2RTg;*!9h}TPQa!zbiz0b| z(8DANEy)GxCrz84T!EuCU{+&p@oj!@JVG*lL%=a6q{;qHOAv@o?934Xw%1{dL) z)NA^7>LB548vx(~YRN4m6K16kdx=R1`(wann3M@p0+sX{Ed_R+j5p#Dv%Sv?UihY? zGObZNM*x|U9aamY1=Z5V62BjtkKBbGv?e2zhA>-RyapG;G@;q();20NTB4K70E7=E@st@^4IMQ?efAbM!^<@aSzDO$~PO8 zeDbjD>04X^0AqgNot~d!R|s_ z-QiL);{FiYEeKN@c2!Miu}$5=S74)7QU189bgBi7>N6Vmf2Tvr zt{R=Yk5I;71cb_>5-mX;Nu#zc%1JY_rtE-cO57BXer+G5zsrmBBqZCtwjccp{XVE{ z6xI#v=^-_bW+@w6gD#X~ibIWk;|Jzbu(k7s8?Oz$O%2+t~yKksbpU+l8Olq%qH9 z&So*yS~>#i;~Pd)vWwhqEY$y;Q$PEg6eQaB116}PQ}!Pb+=%ogvd)i*l*{==9RowYK)P|OI!ylpz+QwH~~eiTUZ42B%TF) zBt)1KP|$NhrKc$44OLY^6*i+iKA<)&5avNh6cOTQ z4|x84(w{R>QtG2)>VxkIY^|17)TKJ0P(460Lh5G)pir0;DR4V5@|-3rJ;nw*k*5e- zVJC_9q4L@>*=)n=J&Wj*k`RU`*8%d!NkYL`(Gu9?B_;plh?VoCyb+&%UJ8Aq3KA*7 vR*Si@A=!(9&I-}@OyxSxi7)9~4dVYN?f?H@KbaB1evkYg^F8a( literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-16-Collection of decorative enamel pins.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-16-Collection of decorative enamel pins.jpg new file mode 100644 index 0000000000000000000000000000000000000000..1d8f8621f00a8e2b187943b42f0ab364e4af2661 GIT binary patch literal 50861 zcmdqIby!qg8!tSBG=ea6my{qebPCcX4MPb?3IM~?OxVSiY1Ykk}e0%~b5>g^C9W?_z9W^a2BP%aEBlA5LT3QYf&U<|P z5D0{UT}(<;K$2GwB7o`y4Hp-e0H1)8kdRV3Vcxk# z!o>doi&WbZo7ojC5D=At!y;Gll}u;o538V+TOckTIRzyZHQQZw4o)tJkg$lTn7I5y z1w|!g6;)k5eFH-yW2p5L8(TYjn1j2AryJz4CoXY z!qt}(dR+PR4ep`T8<0U}gCiWjP0`r}FYSop3RI$d1!}vWdj%S}m^<#e0zuQb@ss1o z#(=Z@x$wVF#vt#62_oUkSD-I4XCuZTE9?rMX@5f3;QLn~k;~mH&{!7oPRafMw0P66 z8M1x_`b{%@1zJJ6oG$<)DP6GmSX^9z`mR7P;fML3@so4NM&WH@IlwSipzFbbab~#x z-3v5i9sZkcxaD7cOd9@IAJt?dz_UuBI=Ptu(8tz)^j|!hq8VDHBzceuIPXu6Gmp8Wcf2o51vxfe! zRq(5OLT1B%U$*?$-T${5%ImIYFw^;#=%Qi9`N)%T?BSpaMr@Qab6l{J0{9Vn{=L@Q zxs8o#Lgm%UV1j^;Qaz=DT4rZ&8dBeMcz*`${$=W+HYu}|A17yLk{%OhZi)s36dN2L z9Z3j5XhIOOxGJhJ4m3hG(cwqv)T&TI3>pOz(ogfeH@U1lZ4_0N8RAl1vx=~vQM>ZPpO#lD?kg;M2b`r_8pAKTf zawqV)DaS#|?e1x1S0EEIM=97IfwaUe(^qnz9OV>;b%q~dXs|;TG_uikHN+F`(W-)z zVYvbCRWu>N0l3OA7$J!5dU18Ap@bk2u+A_w2(_A^Nlq9CrBw{L>Jf(4aKZYc_fEK7 zKaztCH3anM@XOBGPjxWjGh;*f!Fr+9fFX73Od%WawpXq|ra+wd4QV;WGl7pkFq;O< zDQ!dljsOcFNUqmRxrwg1i6B)~s1qYGLPHdxfkBLVM#k)%>~zquPNFL6nkp7nbbNFG zE+A?Uy94S$Vq>dfLDZ_a+37f?SD=M{mt#VA{b4hBSN7aC_5Vr2)iJpOb^Wl|iCEDd ziN6Bv?o*sE0CLXgL)IeETe|fWv(qkNXZQEW$n2G4V{@bZi4icBY?N@X*YbKvL0HkT zfVlt?m6r_!7Sf+2=s~7GNeD+eW@ZQn8Y)O|azGyOWxLc}fwt3@)Rd=-h>o=OvRjuJ zcG60~Wyez+3d-N1h8J&~;2Y_s>+r*Y!%_;BD^O&0*`){B^Z@dUR*Bxj?mjjirb1$b zGE4(9h^7H38N?r!4atoh1Pl_!K}@X*Sm`&2wgNgykOCH+SF)2d6MdHd<s%kMwxB(-?h!O zj(8(3H?D_wVmKe41BS;>z6K@M_H(m0!&Dr&nCx7*eYn++!Xgz1aMV%6rr$s9o5>JY zpeMj{&7x+;f$Sb?b^&I4k3uYjI~UnKEQLCfsq_dvL|wNpH(jmU0OErm^;a zMEl|P3X6hg%TIBc;~hYdTAMuoEZyG{y#)PK^;94_{5U_QgNq*qC&Dz?Kup=0LCTnrNnK@oO4J#qzuhY#m>QFf zjfV|`Bm}C8368A_4tN9Dary(4c>*pMOrR=%Y;uXXD6mJ8Ux7TXKwRp#wUOt3_8F)@ z1ynA||2P6IWHYk5eT`v{qUs7%8rb;$3N+p_Rw{uTMJ=zUqAo`b%cFK;#L&3*301-I zk%Q_oR0Oc5h7oe+Cd!&3GKpAv+b@k4HVl0Z zH?-FZWj-%EFUxYNX5heblVMe$;3Q4Rq%2kU3uDu~Dn4mwhPrn%mU0VSRx{e)y zGgQT51M2`|qfXKL45=vJ_S$A5L*%cea^bZUfp|aC4N_<5@BuCo^uPIo;oWZVf$du@ zuK+Wx-N?9_@hhrF5ZF|G5<_~4pg|2Op1cAZKd#1KYgZNQPXh#M4m6ZgMdKvo!SW}$ z_Uh5`A;6ThWnZI$Cwg>z44422-5nq{jG!n*j&%G;l`oQM%{qGI3%F^}M1S(=5J29+ zA;3B?{U_Q2>Ht#ZxLGe$=!bEFt`|*Pfgcw@OMF@9b3lktz&L4{k?D=Wh=G5Z(^Vik zo?1s~f)bz!9eF@Vm>NK7!nf7AJt~wYWH!P8P3S=$>Uoi}06M6m$3)1+34|$>LRfpO z_%No9{wxJ_wh*o1CbMk+0H30xpjVk$AELu@yPV*(ygfD_e2Z09v) z9zda*Z~aQlE))Bu2Ex8YgS6LG=1U$-F6)$lHy>pQA$tA*fF+#jF2CZCQXo_eW04`F zI7Lam9>F^~wuqS^%jPqm46A?1yX-=NbD>fem6XV34dbAig!4Zt=Xmv4LA{_?B6Dq? zb7c6QD#zN6vIF632%v|bIukAP5-qc_z9e0N80&!`nwMz%pRa)aRsneCMtzyGAytSm ze-sR&#$p3jDIGI?=m84Y{)LDzQ8LDz1Y$bsIu9z;-i;lJG^tull?%^}B#!39CxT7{ za+C}ZXQdt07cn@<@*k$N@)I!(+nf4Bee8>x*0V7yjIUx-fsR0BCC zr(5_+>X3QR@`EdoGVft&ebbsp({Q6Mk=pB5eZ5(-$70df9*3kRtUCZ@h_PM&n z{%=Phq1EoC5QTEr|L6GV(@m3{$Mx;nW&@K}z!)^wV`#$%!hj(Z0#}F>yskhF5y+QL z@SPCpQHG0?N!0TKZ~$!Ra+D8H3Z1~1+2_!I?jTk0n)~|(t?Djds=~ljfk>=RHhEbh zbKO3l-QJnhabjv=R8b}lxD-ZK;+xBrrG`zpm{(hELtj{1qM>n6_KbpG(KV;#TD ze+h?;Z00ia{$B!?9-{b1p#9__+g}iWAE&`SAYbwTYiNL3 zSzO+7G?oO!HYrh1{Z=*rf>BK|R8Y`3NEMX_q09`Gv_9g1=%8`j_*9^kDwKmbw&=nF z5R`w&LR0~y6pB?KlSuWp(6B0pP{0ZM|=RC)65C_dhcndL}%E0lz=(5Tjm z$W$Mf6>&f(Y(|arO(@<$H+@^nD9$&~>lhMQvl-N9zcJ~!c1*r{d|yxluUCz9Z$(mj z^WKQ=w1cS7ZOKG!g{uAChU^Awsj1b@ zgyMViv@Ily8`aWp0dRsScY}8{Ln0s});LPODUz zT#>HMUZAih)%^2jX{Nj(C8NB0W5txuG{SOZ#EW`4<@b!$-cK#IaF&(N3>ZZPCq2*L-?DdFj(y!Qd*y=kb=wff#XkUou3BZ)37ckxOw#=i3{pH3B;2ZgQUb$;?d$=Ew z4T!?mfHP)F$4UgU8~|wDuoXZ`BaDJmY}eZKcVGsT7Qk8{c~nHlAc-7CXn-Jz69JM* zqK1uxofRe-X9gF^tPF?Ua47M+0VpgdI~Tx72zyk-=L8{C01AZiI#{6K8i1LoflThN z@BpBl>u?R6veH%yQAI^?4m6zZt`c_<&ZX(@)Zhl^!vP0*>eq7?YdT-a9JLF03Ekp_FU)Hl7iN_w z+(G5I2x0(NB4ly2s0#OLne(gz^m9Nx136|wUg|SZJk4=^2W>RU^e`fcM^8qtK;UT& z*OKkTNfWEcVk(W*g= zciYDNWu#MdTZYx~4m8@NgD7C~(T8PI^QQ!|-57L43;Msx8ya27N~&VSHbXRAo_~;Q zLA+$hejtfSmav%2Nw~DgD`K4|ul=sJo=7kTYsx+UoVJid9}kk%YCV*MiXK-7K1966X?yTO<7~4Z&3Kl_nr<@pKgJ->)&jm{8;@A{iKf*1OzR zE)&pov#V8o4a*sg&v+10Rn#FM`iJYWX!cVy`wnpwnSr7)_^4sQsj%Qk>ND?)KsF#= zRIO{sAZ7R9;|{tgj_@XpNRTc+S08~DKa3CI~@ z5gLq?mxLYye%&$~k;v)+8F@2;7B8Z+MVS%^Ge9q@K4ud}B^IcZ!5@~3C?Le3wud~j zN68wc+c)$SiZTN%92fw_N&T&=s8~Z-3fBb`CKRwWGqfEe>b#5Bwr3GbI$yyXzHJtJ zOPydg1fizv@C^d^9C^UxIMRpFauZ=ADE1R&U8p}O zwFZX4~w(ml-1g6Jmz4+=(Q zb$}5A;{l$4Qb&6>-k<(Lo?<}*E9E}F3stk*{$E@T?jGp6$iEK@%u!<9R)dv7DIl4-Ya{CE@@cE`M z0k^;4JQPy*xS?m=-R~9Q-q)v|!sKTb?|wP<-TCH?k?@pNM&P{ECuHG53mnzJX>d#B zmTPy3%WZp3-+Ar_w0lz*6Zi3*o*G1i8yZCISc)fluFZqm^Y3drDhQZqgHgAt zgA~J|mC%|d8Fh`KY+gb4{O5|PA8*o!l80nF6M-ZV1_V|hg@UCM`T?e2379_TQ<+6v z_DiqqLSkTp07&>bG9!o4fNg>>;edBAAnJi2v3~06GEgd{9hih?LX0RN5=qHxh5Tc)fHX6Q@Sa zF$U}dwJdV7Up|8`I-)la;h;Eki?d(Ae?xR%Q^kzb)Q2@s(8XkL zq*;;7*RJEleC%c6*AeZk$hcXfcf1FKtg(gJ4`TCXDG$n6=yV5WvDaFi^`*9=F1r1sg2xWbA8LppXjWh)3BiNRRtTJ%Ag9@w*r#h7m3_T#t7D7XM-_)rdHOV zaVk-vL`s9G=;#KvA_;tz5GdDcY52);&?-DdmQcgJqqzl0+ywQ%W(P!LXkukyrE&%^ z=HWettYeDDM{3J<0Ga_pNr&({WN0fe0^mko_SSW-feIo3I3Z(|*&sSExhNEmCOT~S zLylC8`RV8$_VE&6!JFVIU^QetK>o_~hsc@%V=ao=T!G9C8fjvXG3@7~U8@#*bd`Vx z7n!~3x_GIAJkNM?a{0PDK#F;eR>>Q6x|F1c;6pL0Z3#ms zPIqp>;L2yw(RYU|_Yco0Bjovzo_BLArK^3Pb^oTfB74UuGB}FeklOjYjECmwknj?# zU6k4Kn|VUP4)WZ#FSQuS4Yy%FF$!+INqHUbGE&sK-$|C${b@>UO?5AcwfG&OwJ?EBRuFU zhh{_A1dA9O@N;6Zbt` zkH7Ps((3los#17kAnqU=)kk>7s34^q2LE2?$ z$Qkx9;AcA;Qm`YLJv87O65N|{3o@K2kHQ4XFn~Y;@>UFzXaH_=65e0XV5j@6(&$WJ zevm-bX}`DR9OAmAlO$qv?8JJ8*E1IARWCX2_0TvQa))ch;|a0g@TW4wl1v;l+5E>B zgSSH(?AUpbzPUHjPr8>`vvchr+asM|gl29cK*$1gYi@Of~9M8qi&bxq$MmbxOh zJiMhI1FFXYI8R0hLe*DC7mc$D*lPqfgVtdHFb9}F*>5Op4}|WU{XBqm24KN8!F6pJ zz_n}Qxeaw%iW|cSfKSOSZ}&D9}Z;w{2VSP z$DhuLR@9U)pjrnE{V|yOy1JZ&g$|yQk1&6Y_>ZW!%@j!I*iK}9D1_fQdj&owa zFIkA3=ZL3_@xtFN@|h|Kq*_N=?1^O`VGQv=wqzcJ>>-!T`};0( zHk!7L#pniDPRA82z4O*9)p+6#bot|Q`4@}OS1{IhyI8TMM6nkwE6f*ok27; z6}6kv0zhXYfocJOwg#{$D#NS-LNCA;;i;p-K5C;D*oyYA-4mkcGI2fMIl+gssB8+= zcWNLTw2mY8GvBw2%ba(e0#J1FZzyVkf}$h<6b%Ow?i(mNeoYYVPy-B+k_tfkqvZ|* z8FcoJ0)RfmQ?p7GX~UHkboB(3`t|k=ag7uDIZKoTTxj*FoCus*-zTt|KhO=74le1q zJN6Z_hUHkOtqV;abiUo?@+*y^CWGatqw+A0CKM?Z?d4F}?jzz*-|=qrI0m624Wf z;pQ%#;|uBUl97msXwNYls+Q7f7o@0lksdmUo|TVtX|FiWdkB*Bka93;OAxM=H*+Dx zrlI##H(j{TQdAq>_07P^J6rfJ?RVRi6|KDdBCk}+R7&J{z7~0!V`u^|jq|Gy;|Ikq zglpvr_i9(^oUP*_@1uHuMf-^#6wi3k>=18^$C%B2zlYC+6S|mB-&P(5ZR(Hptko6& zq<){dOrrCCMoH=Op!Pr(>ujEmtTQ3Nk4FGLUS<#Ea_!lpzawP{fyfRNtYsDw2czhIS*Gl} z0MZ5gSY8hQPv&Ul4iI)gGD5Od89!dy|H-&^`g3fiTBpe5f2f`BU1hq});|ms?HYB&zUd=kAO=VJ{z&4849yN9Wbv z;6P&7){jyz;WWZ3H4mCuV=YZ1_fv!F+CT_SWJJaGN@FSPD%Tv19sRjH z4-_o-C4#CT8I9p~TDLM%9S6ON-#o)b9@-2LPQTjs#2O2oZunF7rY%kQ2b0UkTipZm zn%rrPf8ejlb_J(@l1mB_l&y4^#`IMu9^p9Kg@xqpWW->8@m&~SGksNEFl}Tu{bJI$ zC!=2&=ZORW0ILMt^3sOi3pH2-IQB)&y}YMZ7n-P9;3B{PPdX^2G5TX$lohxS?SAB8 zk9-E_+IcDF;w)5Tyy>Y&Zu9l|tXwvc95*i!?R3`y{)UFmic{)dmQ~<8cw35ic&AQg zPS)T)`ftjhuXGPKR9Uf`eVnVl+!{Z``V1SRDMqi=CJczk#U|_8n8h2ZvL%%zpKraK zm|tWlb2Te3OceA;_D~e4rdyX(eH|Pc>J53597x!`Zw{iF??whKF|$AzV?b?un>UL0NDt;L>gaAW&2V1V!L3`N6TfwSaR{!7fMY z*Z5F(j4NGeJUJs;QVFA~9z^jL0K+kkja01PaaeuH0<`1=N_>K@Kr9cvNks}cVO-yU z?L`Oz2(Q$-8UXy12O=?0z;#orb%O-{3CI8JHG=MPpS%I_sGg7|Cr&w+f)L5b2OtKoAR-Po2A(IV1}OEy6spVD5?(}xVGMi`7O zK6f$>{{Fo4Nu*s%y38B1k*fW9wVF|gpB!N$nZeA`UMeFNe~iE1^TQ(PyCZ4N!9t6Fx&nxl7l#(GRBLnQtAID1Q)*923;Nj&Fxgf4wPrVr&H z?q`g6N~zA?!4I|(I4y1U)DAsd(>8ssrLjv^y`A^_djh$#iATIQ26ozZ>J>U}-sAlXFEhZUMD%7UxX-(RgF_{Au_ml@%slUGMh_*1# zhF*qZ&3fUkbWke3OGR6tVt~HoYVM!3(2s^Ucx}6DQH9l%FGtsS&M@fI);-UvOjTYG zBgrJ8q*r2@te@i_I_eG&TU~!}>6jU&B)U<*vcRdEr(LMI`dQ=YE02&DkKV5B z)a3VCpGh2*ux=*Vf`6?Dk&Uc72^FQ?J~4LMW>Lwlb2|Fd)8BnY;ulUt#+z3%NkovF zW9h)u)QB}{)^yL*Jz$$qV;+BtTX}@q(SRbc;0?2Vw(L;5zhLNbhttxBlO#iWDxSW>qjGe!0m4;#-- zc^0Y+5d;G`Zh$+uF@wK1tZ;Rv;3#AvvL{5&<>+#F1^R<^lM;p`U3#SLFsvxto7({P zPM!nXC!A>BasXuBLQ&Kkr9@{B({`vAtL~rPo`8?*m)5{fPGc_Dud#(HEceqjWq(bI zfU-Myz-aj2D4%%d7Hv|xhkQbqVFcn(mN035>-QK5bIo)lhK=)eIS`0MzR0zjzGb>{ zIxd;z5K5ooSpBxM3@b)D)#-aXTWxfGr+$6UnVSrkZ9E%~wt9&k34XYUhSbazV4 zFlMyk4Z|On4j(SpXUkghouz42t~I&+p&PejFiXG6&la#I`%S~cY+zx_iret_J$c5k zV~WYeS9Bw}Mw|1tkyZDI`np^XO8V>`ylv0yq+H%t|5fu~0i&*znl#k0YPQT;{UzHy z))xjJ+{JmRwbXaj>lpY7n3s&s5`(Q1T*Ir^7^E29^^PtwV})rw=yP|TsC!C}P1aIK zx7sD+cjk+?qG!`{D6}%}lco^gHGN<1c|n(IAX9(=wrcl5D}qoGK?r&}0q_1YcQ|*! zMV&Z+R9LtISzw>(3CRJQKR{S#V}j6n=h~zygs=%Hui&LWiBa(PQQTLpco@1T)|MF> zV-}p21$)Jy?N_1#K7PNCZ@OfCFi>#OVOKncB&pm77f&`71KU#b0XECM2U5_rFw^Pbp+6!Zl* z4zEarx~oP6AJDCfI4|z1_wTP(4>_jdeOb}ZT5m`RUp=@2C951SaMy|aHX|z@gU+it zdBbieyv!#17-x8h=~di<;jn>~L7vJ|JLUG3`h_U>CXvI|3Ibw5{=h@AJ9SbXH-)uT z6v;Bs0$p@%-t^<86H)P0x;4+EQPw1PEL^#HF3l41WE&g)sv55BLZUq2DOEHFRcmK$ z)Dx-e=Vs9VTvJ&W&LgnuAP^|es{HN#TAwF(@zh#vw8_a^Lv-Vk3bta0htKjvF)$<= z2CKOelVX;AyaX>4ez|+w+|J7B{L^jpK0RDLwp0A^{)u7Ix~`JL^3TAXv`=X9>~lh+ z7B3&hcP{k)nHQ=qX6N{PvAH!<>E_2B@nuBgaPfHm_5QNck2}u;`Be>ZT) z&W55^UL0R6^#P<9L`N~)Z1;O7{vQbpT|f(#2ze0m4I3eP?P3OXjK7mKB(h(>y$idJIRO zo?$QOAKD*Ex#%PN;B2E{QWxj;?;n0HgkCz*J09|m@j4jNV;@1J6L|i zwhq^?TZPU2Fp%8QM>u<^9er*DiSJR;2GhNh>d?Qo)}*n`w)0u? zV}_?OrmP}Qp9w-c4IfrpKcXF822^b%Ghc2y8%-5u5lvLJjNoz!d}jDR}cVbMt#Pr#N-Cq(Dz*W zxtRW=MN}6_??Q4{NQyWP#c1`lp|dnKCTYYIMWJ0ZFG>z9lFsz>Lt>$}KYdDLPost$ zX34*lH$~h3hDASPy(PvSDauk}^2f{n=TBv^DXa;~Rkdr-pNlfmr)ABOI z8=4x2s?^jHW_5;D+(g{uwiyWy*SQBT(DL6~Mz=iZ4)93<(djj)$K|8!AzQPH>J z7enj5`A6rw!`uBy&)eoZAJ4Hi#Zev}6gN9+)HF6JKSFbKu5UXqJ(P*ltB8)1@dc`w zlo3RVT!kb19~>V)NQEcbvHeW*k$lu*qapmmwQ}wS{Ze9?3Z-w2a62(Ufxr!fF6v50B_T^i1GUNT3N9!sp z&lr{+`88mU?BoVA06MhcEReT}6B3{<&CQipe>k5_MlzE5i3t1)mYJN9?UY}gt%zNM<|?SB3!TCb?p!`@T{C1xI4wk0 zEvtYT;wa9>{%yKAXb<(F50hEZOu{l#lS;1 zzfeI{f`8-@+IF_KjuFZ!pgwvva~oHnUxXK73c5h;SXb8tT1!;+5%u(Z9)Nw=Q9Rhg ze}Put|A1C!RKe77m>U!z$)2M2YmIM6vL*+h)PgEIGy2mlHembh#`|2?4r{QY^?`(S z&=XAv6<0LtO_E?xG3-}eKf@34g_n4M<7D`q@|Xp4$Rpz=PhOkSS6BI|cM zT7%chAgo5e*WlQORH3JyTDEHk)-|yT^wm06#@*P7)hq#t{S7faVab>7`gflVe|Xw4 zYC9E&9n%)l`|joJ!FfXH{zsWN;+SRPZp%@J&mGKr6V23ILt;I0)ANQ!h!I*))N3>o zK&@?*2at|`MaG#IL*d~7==B#zjd)1QkPU7qm~N3Xx{MY z2M<;a6==G9^IZs-{IL7a^6BA9bttX+NJG<>fD;DJ%7Iwvx0G^*7jjSEbl=HHaG*l<2dZ9dCmQ z#o|`;68CWv%}WEtMQ(;9#KPf$jTU%W&x_QO)aqi8l5JH2k8Qv!EOR;Tub2^s!E z?2aXKwJ7N^()^)cP|v9e@_g=CzFnoHRhfOAlHYS zomm5Wu_RTV{T%orKDL!pA4V&gR}!k6_Y9qz?|>zcYucMacyP-kAqGCH?}p*$=jiv6 z(>jl<8QEeD257S)9JPpoR?!kVhW1_K8LgKCE`>P1Uhu>TXlH9DGvqO(*szm(>%4gm zsd-p#0dEuerS0ixZ#i;^OU-}JpIiZC=`SUo*4wmxnG@MmOIlS~F}>IA<|uD8va{n1 z^{t;&(Lz(g3mUu~SZTn?YK0(Hd=6c2A8cu6&e88XF9UcrX2H;Omdl!T*sci|zuUJo zpK^+Dm{;mT$N~74D+R4pcy6*Te$?+>Ux9p= zW_J#4dt_1`RStUkxhAh232LTa3`e~g7&*+ZhU2d%GQ@gP>XVypK5kv1A1w@Zu{{)Y zHfWL8?CMNfiP&n;Px(rKbLvHVsmMLl#D4d!3&B~o2oI<16UVy%Zq5!58gZR$a+}hM zUUnsG(qPTYOkXbkZU@i(QCT@X+1ujUue33nyjH+%3T{YEyPcQ5PMr}grE2o7l!ad< zkN;;#o}^{DV9#>Xey228nta51FL&RGSNge8xa@&Yl_$4J6hUH6?A^)VSnR=!5u^v~ zp>`F^FC5^7bsf88Ys!s8TIe+JPY#U{EhZJa1R5wM*Kf9G zY(_coXs|{m?g2QL=AE57`TL-J>Gb7hyz%*B1ao_lF{dma>l;O-FGN_MV-qLHG)+s_ zlD}tKnB4LN(I=P+B(AtTj_-V2>GpQJ#o>@p$)qN=dxj^jroeJ^b_UbEeOtB@PB2Ed z9AjAwBQaa6MW4qQRX&}Uo-Y)#l%fyaiOw-Lfzi^s|30diFAmZxby}>A%__w^%u^87 zLw8i^?802=-89Y2GVD_(w3XFXmk``}ElF->}`g{4sIhfWpbWAj1 zPbY>qx+z+UL2ihrenH{lgBm3q{2iji#8TR56bQay2ZqZ=j6T>5=RzY!?@VvY2TFx{$+JOIu@M4XY)>s2h6?dJr zexl=(XJUp44;aiHiX6Gqv;$)~we{A8de08ve>&8XK71$$>zMf#(-VrmQf;6RqdBrE zWK8z$mA#UcD2FEHtA}H~tHa5+>Z)v|z?${!h!Xe759>dI+qR!h$u~Y#kmXiQ+3_-G zpU}(MEHEzPO@+UxSeRinnOu!u7p!3TZaJeS7_nl?f%fDHBP}iUnxt{jv{3k)ryU? zVcPBRW8{C8KK%T<&qB4ivMyS!E%n*t8srIBLbglwS4q=@l^4v1OM&xbQ*>LkgtbFm ziVTk&q36a!+r`YJbMbXb)5FkK;)XLmMuPM4Gz+31CA}VqqTU_1?S`6EI)U{(%eOFc zSI(ZO?p9}Uke?rSPfpxYLQ20`kFE6LxK{o~%cNm?6<56WH|r46eo)WGx`unIwt%2h z>)i&vkrdI12#Sz6ujJH8l2*~85H89X`wk;*Gds$4jw~=57O(+8Z@{o!yDt=(TYWfi zvR95Ov`Z+q5n{246%yBfB4kMjVQxze65_)0#tHe+m6_io#G8_*m9I|1h<@xoh__yB z3j39tKv%^i8jajZPCb8c1;W$$$T3{k$IyEA=B&6@b+A(l2m43+nPECCl{Bz_f<^5~ z1UBoUvy9G*QfduXcVx``#N4Qp<)qBP9c4+4jP@}z=iSSf!-6NaZ^XQqA8C2{{Gc{< ztwDPu<+e_!OO9k!|isYkLKJ^;fVR&&)Ur|MJKeA zCAZ;64%J)c-F@ejjiv|SKX2cb1PZ92aSlR)us8Z#8^+%-H8n&b%!Yp%JjH*$Jee{XDRz6Y*P3l73!^VXTPGLW7;~_q#ge z;vUh4)}&$h`&alEJ#k!u-fkC~%UT4H{Fw1i1(R9VB`zH#(g@b?Q{$UF$BzmmY>SjS z8>hG|$$X~uc3bZXRN={f1tQbbFpMUqFVh_aX`qc(!JM_lj=oY)ba&RTuu(_xAiVj}Gs z(zdp$*=yxFxw$~F!4}zJ{nw-Z$*=S5zP&OD!{xiGLmwC2eZ~NtqXMHO@l^{%Gv(E_LsD3?+P-n86hKkqDf$}G;2 zoqPRaAc`qLG4tqpIx-GXNe&~*!Ac;4n8DkJIifRBV#b6lHC(<%1c)w5Cbb$gsiLgv zKkg=Syw%@-dQ7?Zk4t1*Zf41p=#Er3qxV_8RYWYuD=!j^3s!u$9>fTJMV4Kp0HZUG zE$AzH`hMqvzi{m+f@gEp$tn4sEKy5ioy=RGe%es|n1uu(8kfx(GgfL}jb-Q1m)%m> z)uw5x#F?g(>b1h*(#DStj#w)4+w@_gdPB(xVRVJ^dLcuE=|1`$zh6~D`Mlj*=?YaW zLV3KV|J*I@UXCP4`^`{L3OD*P`vpk6S}UN`VLwB#R`OJrzEe@gOpo(8q}ObD24FruG8*WJTq+biN0R(m6PB&A14u`Fbx=eR+E3UGkfavR;sr{ciOA89R-^ z%xYRK&;9zHpvg+5UJ#j!VZ*s*k>JW=#F9l@MQ+PP@a zdJnFceLjNY)9?PKWsQY*h}rQ-SpLwn-R%E#oU;Nw>DWCV-sYDp}4i&eg}9<|0MI(LT7QIrG!=kvad zTZ)FUN%g+^2$vGxMT$rjd}r1tHgM@UWyNQr1LdWu<6VYd)P#53THnrJnPVM%m7Vv6{mdi__@3w3*1f}e6OV2EXE#j;Sa z=XRJB<;Ep2%57?jocVFqOuphJ?svo8GO((-S3kAHD~{1qU>wc1hlt-Y9eyiZDYCp| z@E&2V;=&$KQd07BJJ@|W!aWquHK30i8X|Ce+-L92DW!VMkV7r(T*6y*)H*Bq4cI2r zGZ4H?(hC>90vU4r99^6o-@894*27-)aXyq2VY^_aaV#JUUqkWRD~D?C;GH|7liwI} zJj8<~f4U-ifTDs?^$=<^fsE{&9Tg}Y2zZYG^`04!+5m4>17#W<-f`ApREySc^r=_} z(QP9mxg3+!8ijE^9U<7Ibjms9KPSWTV`vC^X`~Z!JRJkZOrATR8x47Mt1)YlrcJBy zVSg7rtw@zX>Nn;;7w9!({;ukzL%IF3Z3pbe$*YNl^(s!E-J%Lta&^8s3QDycEm_X` z*s>z(qUgg|wi9=UoPD-&Ur~VpVdT%164&gNOhiFrm6j5M*U8g}MiWuhr>Z6apE<4Q zEoZqi&cHV?R=?4LzdFm*R9{s)A|%11Iv93z+~ouEq22FlHXeHUBG*Wkd4FL~LSE9T zUSfeuY-P4W(qjHWzR`ErY;2uh3a(zS^KJU&&|gStt;p$osFd9t-8QZKqdv>97}j$>fB-UZ!pP#$L6uuSjk zb>HUR3RC1EC{?h_XL|QlhGntpN9O{PsfOOl(~2dpA8|8t)60xsqaTx(jkg%6%?z%B`8j^~dbA5yelaE4b2n_IN0@Voa|FpZ94q??_qD2$-Oj9Z}=#W~Jb&se7RlaHRciwOEUh7s0-)sI#PZ)3P-Hc`tQG>~5 zDdO*GT^!-vt2ds(e|u9eM)|2o7`&dLqCA;WSjg;bZKJe@kD=TV{Y60oE6yWsR@xxn zB)PS%IvZ-^J#f2h8A9%!*gDMqTquLGSyxi|d%SuWdAZJov%lIn?v`tdE>(vq-v&w7(C{J7AkQ~X9%VC%g+HXq-Y))4Jv^2cv72L-<*t!+{Su%wxeLPC8}3O}1KFo! z@c^RL8D(34Lp)x!uEe>p-z%yx`J6QW90$_ZoOX^>y{Y(c(s3sf4zp?ysVm`?!Tv0E z^2io8wSjBIgCZkZr8nvD$@(}N27|SY&l!m->8a~Z;9(j_A#yg^Rh0<#QJ52u5}vEa z73YAR^!AxiK=VJni>+(6LUod+Py`_TuTB!0dI@xfRC*Z7Y~y?f-)qMXZ|t3jUNJQ$ zQbGDWcyBSzPYO>^21HESs1W;dTS2`mGo%jj^U_`NS*Z*hF6%{kbwbROjw|HEkqL%P za{NlnhY?JPNpVRpKB>@)T$Hh;2YYUCP(6;bG}fD#QlbnoliZ;ih#nPyLkp+40mjf( z_|TFb?(LUjhqf@=FyuspBk5e(N3s_NG*MUdG}N>FNu0XUjY*7y2n$F|sdenKlB<2b z$a~-U4?#*3Y-nAdvzC1{zdi7ZLHQEs3pdL>TT)Qs{UEF(_1V=c=0ux`7M!P$PiAM* z(wM=c_x6_T$hc%J;3~R1a`4Gct~s5A68wQT@3%Zei!%MJ_dxJ9S2KK|kRYpMOt0D~ zw6jUmY;~%eKKc>=A!_aV2=B9VESBp+Q$>~j+!Forl8Cr#s&0CWO^t8K5=Zi7gl>K| z3qg$w>R#tx6tZ4rmz0)DsDh3DpbZktMo|%C!Bx^(T!Qx^(T>v3`J&u}Xou3)v@Zmk zXgIK@ntAHcBlyu4ZxQZWtCKdPwzxA^+H%^ znn2PAs{EPVY;CbTkX=`8x7`g*-x7&MYSza4kva9oGPh5ldHc}jegZzsqzXw}t~-fRj5U!q(sv2?RX4M}GQ)8WI9K+Eopn!d1_yn^!sd8U`!Xb6 z-CdUV>#t?avj0UfUR-1??IXK0Eh)VgVqNAO+2nGIG6kv6)YPdE>+I1evj5V3u_iG9 zl2~XwQeXEdbW+gykUZhRU>Lxn@gU(-SsQEl(l(@aTHM z(39jL?>C&O3@9%Ee!v<&J|@!4o3egZ?mVV3L_S+NJCj|RUL*r@Q|wM!nor<*mGvz7 zSfZX}*!%WJ2K8X=@t2p-g+-n+deP-JTW|Ltc558ZT8R2wbsB_5X381D)I-OXl~L}} zAb#$AS2hb2t;Dg`@4FP^4_ocl;G5`TAKIEOm{4MT;Oj?{S69L~jkrl&tWa@VtuYYu!Cog&AP@8K+f?YOGaVPbcny-5&1|jmEJ$?< zGKnO`LZ;9AJo*NQo&PA893OJI?R{w(lK<6rYtvYi*)3F52Q*TF$b$gZu>bcX z)&F|1(^Ibdo!bd{(8Cu+o*=du1 z(dR`d00%2zar~brhA-pwI2{X?4hI;a?q6foaWPI-1Z7B@J~<06>)pv?aU5)j-zr#* zi|{OtXs#Q+ORjKcsBVeB8LAQ}>amNv8W7!!dxDVLx%n|&K%q5d-OCN8^5ASOd{dE2 zkz6&{qAUBGRakNH5B8wp5~=;-pI-p{s?b1Ua%l=ql0Z*2R|w89t9zRtk;$TK6HN|{ zaZ@7m_O~-pVb4e!vngbUkr8=x@-tq^H1JTj;IaDamT|ajdc&S%IVYEuO!*JUM--8& z5bbq+u-t!d?=JZBac-CO#iw{ox}e^2X)130qx$1951eR$;yx3(geTA^al{07ukd0g zqxr_sFyBs&@LlF$rW@S?NB0rX(kj5T>^n9Asd+=fL#S3M?(1)^Ne9}sUJIbQ<%Z|X zHz56K!=PH)TRx!8zcd3 z;*UC6!Y26DQ-2L(fel1qaTn?Ul+9t!pmdSKrF|WLtW7J8Ki0U?bph@oU^JTlGRQeN z>k>bPr-s7bB&iSyE9533=KWCcI5KZERA~LPV!8daY%%hN0c~@EAoo(=ln^JxlBrD~ zhpzIj!1KLMSS^2E`q-oe^5>ML6!ssonGfVuOC?u5ob}OvQSkQ6-MlU9@Pv!|1nKXz z__*F>y8wsv)^H{{_1Hpl8&{7I!OEv8`zsOgY>th=oyUZ|L$5MH1B{HTj%`PGEsB!4 z9gf4o)*n)iuN3LsG%02>vM?JVjIJC%5?T#yj*UK^M|4zpg^lnPWXsmxvtRdXLY2wh z3Kdhvh)O%Dec1ZW)30;p?>g2dxSmlL(NGf3ZY`|g=`V9{?KvT5a?en<64CT1tKmwK;wV6bNsz*P`?@t)8JYZjxSnw)f`{*#S=aUP5VD z;7c;F!qaljGnV>R)LB_W=cacXZp}x&=%qV3ltnwbTcF-IDOlFT@GG{%7tunihNrE3 z7~}H**RjQorEI*QWPyIBK%I;TXKJIzUrCQE4UIOyI9c3xYMejcdg&jTb@?2MKd-#l ze)sOTtg}YoTqEh`U?vb7a;NaNn)&2O?M+?I+P>0G_Sxut@exy3}u@QwR! zX5@PHD=;5=hOEn6_tmv`i{wvzC$Iz+b?&iR{fkl-OeqqItScn;%7lnin}eBw*5@l= zDC7G-x?=0g@`KK_&-L|6hQ88C&pK_M*}o{$x<%_pCS?WD z0k3bu72Ccpt_)x(y(5yd(1%f_r$9E)PRg-WgAZq~-{Qt_O0tRa=hMdP4s5;J!U&+qv&G!EWR&Gh`1FJD>1fqxQn_{lcsvd z_u&`$)NlL*o#e?OOSc*s0)@;1ap%kDEz!Jvv4uIo>XRG=#t*8Cy{J?V>X{eT81DoZ zheUbC#N!d7VBs@!OuVNw-K*W+XhGciw;}cm-qeH5kDlE=$~hj3yRpoR-9P5$?=ZGB z8Pd!na17$&5deEd&|P^|)KVcf<`i1nQlUcsJTU(zV5)%Ug&(oJ~1U&e!1kBa|(j;Z!?`g}$`E`VnfJF@xg|_JWAN(oD zOihpi<7>U3NVY<}E{d>r^5sPd$7j-&RbC&hsodd_J#|+PT1DL=K?b_4-sWiXuF5{t zDaWmQ(GWkgSQq4=@YIZtK~ahcf3zhX#Ws1sdTM)Cq77;t$*j)l#}P{?*MraA#L2O+ zlMuG+h?1Q`Op_uod*tFR>E)}{gZLS#pQ70FcD>6ssWfD06i}0efi6bO1y5GP3WRIQ z_k$}KB*QP3?`AA&)(_w^{8d%oboPdR>21@^vY4M~=vhW7b5q(rc zV+Z;k2TYE`+uLg9zGFBy!QjWa9}A0rQC<@b(tiTIR_A!2FEPMDYIY?&e84x5$%IK@ z**~zX&9)f-Y;C!Ru5JquI_&T0_qv0KtnWNY&px9hUENA^diAZz>>(PuT;mL=yQwHr zj^35Hq6K&pL_OLF){WsmRtwK_aekceLwEys_?XH}IlU9jDLiumu!3ki#lw%TZT&9h zE{EQ`)kdsb5_=whvw~6|#*N=QEAxhAw}#{^D=$)cob0U#d=KiO%C5+Z=RPYO6*kRr z986t%G{~TSeG@BKXE9Gu?RW1j>^e)@BVE4dGC#~!c23?q-~T&|(6;-uuguEF^E<70 z^~cxm1%N*sxq3!ddu1;#gA7s5DCsy*PoJ)By2w;mpZTPjXJ>iec|pHW_rSsnzJba% z?ZSB~ek-%?d^5V&`?(-f?_?2`E1l0lktbb7+ zdysC4d!S9i03?8H5K=LQkGlJ6{@5?|fT^ZQviGUO#gl}lQC&Y8+(lP!(DdiTl$6)V zv}>G7Z~fd9iD()#?{Q6g5;irSy*7lD;SV)@dha;XJcg^^yo)2Vv(uaae9WAE{Bvg6 z>}?sl2S=Wc;ey!c;(0a0!Wk)*F;3c<;wiB;Ft6H3IUVYlFWRJ?F_%7K4al0tVtJ-+oij@X@6HQmzof&QZhQ_o|v~@FSOU(z48#)27W0r~Oc6jbj~& z9ywM%yNFmfIE33OFXcbx#D8w;e~xM?7+qU?8i|>`X^f!?VKnqe;TFhU0;*x1)siyk3I+}l_r=kV8)w7aJG#NveU(K{M1-|4+(n|7rw9Spb zSBuAq6b%vl9n1{)ub$ZY&W-&c(x%EuP|bUO5Csgx>EAoL=L@Pwdd3TXyO@Kfnoo;J zm~$Jo;Ce1}e5IIlJcO6aeK=Lld}{Od$%k9&j~vcBS=Oih(6=;4)D{yCOIEJl z#e}v3E;C0G$zh+k*TBYSm)}+8WGDE@(IhJX-}``cxl(+N33*%Yq!%M`tKKbeRlJqr zst5!3{&P(r&qg4_AXmM~R*5e@d6HRQ@PMGuivh?%gBu|of- zPBaKw9}q1SwB)KaYN4+72@UI5)6>&zFtDpG{*@eE^8n!L7TYO3$ld#!-IzNASkJPn zdW9HzvKbLKjCs(Hvc;P$2|BC0fJV_GG?_$-q5p#mto+i;F#EUi`mag*0IbZm{{KChEU+gV z$^DB`l-CSAWB!2~UCBTXVH_ZU{+s4_`RB4!?}a&mnmPc+1mHKER)A?+b-{_d7k z$`NJrIaAkF%|~pTKu#9|3ekLgH>mr^So3@8m!C~ve`EX1&hlfw+$9S^dZv3ib$HC+ zhKg;TJ{%-WJs81ltLoaKxcBO1qhnulQytA-7JH>=iLI|TWjNtM#cUArT-|Lx2k+{9 zS#dN5oupmB=|ocuOegSVA)=rqws`q{`4LkaK$uum^}xDA1-8b1#m}x$xUz}!I<1YQ zNA0Vtc6_C%iG-qfApi{n%H*)GB>zE-zP|f^{hKBXSV5d|08X%wa$1!7V|Byu(Q`Fl}O#+WfAQUPAvBaDW z2(p&>dA&lPu+P$S*AY$@-wBO4EGdP}&+F-w>C4KoQdGD4N79rdLh?j>OTC%-OFTAQ zZaZe5x`*XNAVM)d+uc(d>#OQMF9+qZX3J?C^hD{1b-~@-$>zO?qo}}xDD2THMMsf% zbk~m>s4jg$^4EN!sp*vH#)k5m=3~M!43!vSyD*H{!Q2<5XocFUq1Q!t1FX&2&8{B@ zV7t-6@^2=$h;ar7YT|i+QvHA%;qDz>^)%K8CJT-C5mVN#73gj-DUMK=9KLD-ekBq1*%WXAnd z_cZGC0xIQkB&%gvkL7Vr0`4v+RB#IOmv$Av?&s1hoAadf*%GQ?8{%X`yV$W!xvp4N zr?-NFObvw}uI>i}eqX~%u>1y9QZ#IB{q6fg%N!iuSTm#!4EYll7CW=kBY$vRmK`am zyzc`0RrO8{Jg==2f8!24b$TZ$8hM+j*(E)t^$f#-9_B2$|B(Hqzw<*!k^+@V3m}71jjSqK56{4DbgRm$&e0Vq&4>sDgb^S`0oc>8!kxLRq?g z?XO7{KD!4_uBW1`X;%~;FudCfd(TGrU9YnLXYuyzQKF*pn`lSMqCZ)TCgk+(vLLwC z&Pd(|mqo47-t|7kQ>9jFyC71_UTW3JPH{LAkzv_jv*SJ%o|Y9;t>_E0?WaD@eoK#} z*|>mZQJzu-LpzfJ*ZIWlx|XgfWH17QE&{?KY zNb1rSO<^)UOiqnH;!Ud;t=2pD1a;_`CM!59YLcmO@&Uv!(`<*=c5+L#|J$YG7_zFb z(QK=@mPt|%g?4{v2p-YZ;a=>0@lzDbw(OL$A^O}Y-ntgsKui)>EJ)bYcgT{PmI%^V9Ky^{)QjJZ+CN9*TLEW={U<(r;J(4rtv zuM&r<<0RIq{@>{q0m{rhbWhixk@{3H;F|bWE8lPOqO1ci@UeR7FUk`Dww%mg6|>Sz z{UV_6Xy?%vbjClrK{#Yd4yFOUFO;%-L;?|kE+PFXq#!SL{4GC0MWvG<-__?l zSde)lhP6H-^Zh%a7q1KA-=gc!>gMiMf&oIR&FTm6Gl3u+G3(Vzv`K+t5y8A@6PJ-D zJDP1ZbS`B7HGj!Ehd&4DR5xzLFS{_>K`_^3(eZ5wT(@Gm^TLvOZDR=YKFQr)MwDmM zCan{4pvi`eHW7qOc`OOx-_GK)?YFbeyv7;C#(IQt@{r`{HK@HKA=O;;M1@8J)&S(#%r0qtw{fsZD0?@^-=@UAj<$uk=p+c- z9R~F_{anW+n*xU6{W<~uTfatcy<81u0uXf72}2Ga&nxwj?B1~VV(7EE-wQqj zZCDMl^RgJSuY09!=8W`j=(cB7C;vcl z*ABu-F$bdVaOY-Um(XyQy)b8D^`dL8z$i7e5~GRGmk$8GOZ)&W_7uk~xu(=$1-UIY)Eh?UT; zIaDOcHj*p3X$BXqHy1(Xdzx+TQ@?dSuueS})96s@P_KvI(2WgG*Csn402hv7K%sW~ zli4-%R&nEsib&M`=c^}hBP^c1G>`n^jKjKGZd%a!#80%_h#qUQXdjYEE&E$MEBwd` z5*wdDK{>iCVENi9M?Ki6x8uo}4&r9#Jyvoj)1!R<6Wm+S`JIab+_&#K6tJt=Ke|>A z@CBx76Ib)@gUZ?#acN+2Zl%S712AVSR`M2sd7-Jq9)eTx&_ZA*_8!=-a-A1IfyF9K zrqsWV?4Q_Y0qDZ2H4_Wy5E=)X8znQvn0`16JW2ty_Q^x>kq_P*)CzO=uiAcxy)RlU0(SSl;TVGXWZFf zeXnmU2BKq%&myxZ&w+hCiBFl64U#il#l(NWq1}++@pry6^>+qM=~IicH}S&MZx&W~ z)CA%Q1kfM_W@#dz)K(=_({NWB(&KI_5U-NTYg=xH*}RcyEp3AqZ-E`?h9*N^lWtJ_ zRmPY@{rj9#gEKE@BlQUK{)OE>I+7{M7K#4OeO^U(vx!?*Figo>=+F0YqK)y5+~+k! zqo=LB6bRk1mZRu1Z>7iS9g-5aei0Q*G_k~*0><`@z^MD@WvuE^wLVl?Shd9~g-q`A zNQ+QNCgb$ze3#xL^D;XEGnc^qU?mU0kv0GBDqXb3m423i5%iopjB~M4f76{G^D;d0 z($_C#Tir3E);ch|*sylaZ4Z8tDV<-FnJlnDf=f}gya=(OC;7SPVH;7Rj*o4vde)<- zm}qwD_>BX_Ho5lyg-PLQsDjr@pP-QiTq@!&U86S;}Lh-t{ zdj_&JYl?C#)?Wd1f!tGo82}8c{-wmYli>NXq0BATwYMp0M63uivPUgl!N}6(lAB#+ zA%NFHN!wDFR@?JM({^Rju`RxGF9BQY@8r#yRxP4)PBKQZjfx%pk*YM=H>BeF5A7ko zGp*R(R3Oim-MKPLB~b@(ix8UuKYff9d9TlVaGKFe>$`7^`Dd#&OoOsFGxqf*?nJy# zvy*L2u<>4eC32X?kGE zIJbHfWuT@VMtG69$FCxhBg)bzQk5^a1Sx*Y^DCC6$nc}R?`p^BI7LwmcXqT0h=Sg_MM1Gzu#>VA`;wo)(3)^T$?A)Z4_@30Xp~bU>7|zUe>V zIq{L;AP_@l;B z_#0{G7b8C4g6>(1zj^u*y=?Wwnf-pASxUBpFiMHb9^_PuTV!-_@p$ja=6tqB43acxjzrn>Nx28 z*Q6zylmqUm3ZNR=G7O&E)XVtNb$u4l4nW{Of&%_)+yJnp>_g}g_9H*xT54VW}zUQD<`|Hd@sw2TNE@7Ln!E1CCIX8ii?TGJ5j}s0`KK&Ah8@;dBv>n)!9I zsVQoKdDZUehSVs9b+5h>Qy6G365$vzlIk(x?DN}l=Bg;$QA4}LakysOzS!3#*-`9%l!xj>M&c5FllQqEpxKg zTZTWRKM}7t*dFHMa>5AY>jVj zed(HIio2csxqtz4U2bE2Dy+fj?+7+3DfqF#z)M9aOFde~&>}91y33k_>ZDhv7bbcW z7lv+wAhsKYvTM%sa0!1GxhhCDCg~J(R0z|nrZ%(9zAAn463_{1vcvJ2)m*+H2uC*> zH?tMe5NC=k-3NGsZ#CDAyDFSNepnLaR%{l~%TA3UAshf_i%^c3-MT||)7257wYlpq zryuj=RP3P*Gu5XD*WutjzFFUU3xu2zw@Rwrq4-Yo&Iw_*pD|wAE`Gv3QztUTBfNe+TcRU8`tafn$MYN}4v0#juIl}0 z=5vEn*`U{7J3mh2=A_czd4_7nx8X6rmf3c*j@v)Gc*tWNWKxi5%w}eYdznDrD?U+QZD-D@D}Aq^IJMb@q&Z_zW8|6 z@4roFl4WssL>h96jisDR7Ju~#FFk1wYI`{#t-r#(Z4R6cwM5A~(NzvKzB4QN$Yy+i zYodKOE4t2toP;SgB#Gnr6OP+}4uFwGIoqRcd`hR=*g7kM>}3IrfdydQ8vlpP6?(aX z*k=1SO7`wG2*aK~%&Xw}N{KtPr&!^6a=w6_cNxy64_buA%pXTL4w>O4K8&w&--!ov zJoT{)HKuZXGdU;KJ|?Md4$7Xy^^fi7N^a!^uB)LWkjKC+N=iN;j%+ks2Ad3DY}eKm z*Z5H!06PX8Sat>doum7g-en%>UCMoc#s+YQF8|lr@X5O977MA7p(`llhSKIt zskOglMvXVmCS5N?T--;ie1Z&4%cp5!yx2u|@hin?>GhpeU-NTRKDi9N9PPDN?VXdS zKt;NK*xaBbf@-|UN%YCYYARjuVv>%LpNweRGfHhfVuDczoK$dztz7F(C9mL^H>Nh3 z`1r*8u`1rDN8a})OCt0?uByo&vU$iYcKP`1)f=_VRD8wfaWPe&KnHFwvY)c^y^lth z*Q5=jw^?1O7`STzSZf8Y=|d~&ZQh6Z4*;%#TaLWzj(tU`z!WH#vtIf9zVpF=GFZws z7@&U?vncTrZ%$RA$^MN#IeYNQHYLt_#(Hh}Jtsl1E1i3q)Z(r{8N$WGAOHqjwbe_c zJnD8JCj8nPcK?=)!st$T$o{t_F{SQyzIj4#4i@N zbf{b(YS>|^K8}3_n|NT%=Zq|5uipA|ESoAaGDLATR*0WXSL^XG80GK)?BxKMR!rpO zl6;BHxNfiiQLUZ0#D!|afa4Rd6q&Z5{lO!LC;f$4q%UYJ<0X1+r_w&U6+aA93EXK> zEyLN%tPdwPSNJEHV!9^{80&|ue^vK4E{p_bWh*BcDaGy?;?Gp*deqHum|S1}MHz?9 zaJdcn9*}a(nF#7-TtvsnP-R?JFrK=dj+88TIs6RPn?I$mIBgQi9wMB_D0PY=dX$)# z8FE6z!7rN1v9F!DSKv3f?E`sMcdXQ_JgJ+iv#u1VcNhr7;aaks{id)zhxSNjj1p`V zM1iXd{upd<8okos9#f@@CO9Te@CA0WaS5*?^e%N3O9O7eiT6WWyx+~-nO8wsRBPS=0`;wRRBl^xKFK>15tYh7g`1;^{ba$C=+_m~? z=fm#Aap-YXiC34J(Sk?EnnV4k`#al--@|RwJjvEp@$Bwp{vh9(^TQAF5kBdD4FpQbv z&smI+5@g}`0BM0`6tu1mXp@Xpo1O*ouHZ8OY$RSKNg*p{utbZ zXUaT+A{%i1NyoLfhDvA!2s6SjU&@@GJMZ;Klp6hFf6qh@R|nw1Df9={Rx-r4OH z*kGh*kg|W)#aJ_N)Crs4u?OETUl(%pb=lRS`W-sy-Xq(jmP=ik3v{Ub$_%WWTT;|y z#^HDz+1g(~D*qIZ_orBrWw+2(1yze5g-^)t_c!Aab7+CR&t%t}u)iIdIo?{3NnE1Q z!Ix}&I7@scZG1FfF8T=^unm7~(x>YQ3OoyUSn95`wrMLQt+U43$?IqGPwPj;O*5hB z=xLTw91A-N4b8DBsp6yuS>CWKp`PsI?LOJYxT>1#LzG8M@Dx;W1jqY?LA_k2XRmX_ zySyNAazAylze#uN=csqd_;yJR^0EWq`$D5(3j(kBYR9dA{bQm300dG0MYUn1-Le@& zW92~aRZ2>8XB54t^jUY(Y5*b#`rcm@8Kgyx1!z~UJ*>uN3u<96ZA~6MA&53b)%BA4*hR+#WR)Ix3rBc2c*E-+tj&_Oj59MM7 znm3oi@q+b^UFf${L=Ep!UO(kwW4RDP%GBqC(RQY4wpcW>npjd!AWD8IN`9ZEBK%C- zsAU{>wh_ai^=XEkFE#xOxM+4OS!ZHH@QNW&{XkvO-1SEqTM+uP7HS8*&ZJlAAuh#? z4KWGr_`3k1g~WZOH%Y^#%zZK=b*BO0;RYlP9rL-{WqljeSTVPOfOEp(RQs857hB4A z$z(wqClG5q<3<)+`&%W7a&T6Vi0xV>Xwv|2`Z%1bWs~1{QlGJR2ONdC{|bGcf#cbu zT_d?|X|=+8a)2cwNcKnr$rEEgx!ubJK1}tv+O(CvpKz_N6)e zGJew{xALq!1~Bjb;V6?O><{_W85gIOye}72M$Jt16=T$`_F5t8K9gn&p8Y9C)ni(0 zl6nb?p0z`I^qF-Gj!414m2Q^QbQVdqZG1!H4+Sc%2<}-<*E@{E+FiS>={N6%gI~uB z?##JY@ZxK^)!FRiN+z>VRj7uSfM|C8hp*oKyxCL;qUI_YHSm_ z)fhn|IFuE@SvR`h{%-PfYWx;+;{zQTl~r{XSN) zFPj}wbHpp{#e+x{L%&61lR=lOrPX!81ix^!*42UPl$kZ&ATNVQ)d^{f^AcLHuF%6Z zSyEwi_hu?1d!J~#X`n%^l}h$?)FwPKhNFayEQ_^%(Xzk*`_|e4mFkv15a*8p=vz{{ zY*lo});HDL3P29hw5hM{PW|djoF!rk4>le-4P; z{&CEhQ^dafhc=@nx~RB)1vEiGcqlvc8VH_r1dLj-h5v#l5sJtD%xw1 zcQ>}C+|594D{j=8(3i_fP5I1Ra^r?&Dl>+cs9jBv-l>Gu8K?#=g2h$$;l``F&^a>luV^I7#ri+Z0}!a z14%3V-~Fw^i$#@O25Wa1-_*(sy(DwnBDl1Qbz&9?1``Vf924zNcVER^n;EY3jZfo} z?~s)54)SWtj!i*XUAEF{XiPL<{%+UOX>X#Nv`Anxchbwh7cpDo@gJv0tVh-;)%uie zVd<_EwvXGS4>f6m`8A)*XZh5}&*gJ{^^(80xWSy>(wNlpx4cVwS}f*v09mifi47ic z85OCnx*!1~7t&G;i}JEqrk~o!sRgUKVf-UKhXFYpm_QB(iY=gF=?78~9npbhz4{{c z4HQr%{svTu6-wV;7KX_*q`rv>Q+rWRP;jY5d6#@yOnelyCp#b%hQ8WnB5?bi(zc*Z zs$`F6-jcWLzQ3&xx~G57xfJx)cN>OPUUrACIlt^qhlUnmm}uHzgYv9t&*066lMfWz z$^>pBLb`KXnqEf{(VOgoYtrJgvd4L6zSHLnWl^^jnRA!&D#J1pe^fXqzU(kvmi{a! z-*O@FccHE@;7Yn*T#k(cTG4 ztN9&03?2OhZpCk#DJ#=bH%0Q1?prHQ>Ne-7V zHRf7^pD+Upsqc-_@6>b+Z&}5c%3Z9_T)PQh5bwS|D>mtykl0gObQp56{&W>R&>~I9 zUo4YQ*kjlA$hooYE~+YQd0B@2)^rmjNRjm`G7<*c1KXAu>gxRwrbsxtD6M8cl+#=Q z5}jrx)1l;T^UFh&coQ5!xS6Hh>U{@7v~0 zeOBUC3I3>l9@lQV@5d%IQlEF1Nq_+VPBKkztfu(?Agc{JlcE%M0-+MUb~3@;t!Ry|BNL_I0#d8c8N1j zlc#pr7+?~B`w@otY&G1=T0hu|4uSVE8T3V5j@5Tv)d{Wn!Y<`&JbS>5$w}tW5fkd` z2rYxn9j8BI%5|{9M&xi^6Xgf*2Wex2MNoI(wJ4kaW6$A$Psk#ucZQ(O##0GPOrWHC&9yAO1AdMgG?n!;an7xUq_+0 zDf_5PesvXRGJcDL!di>?9gEH{0ln*PPIe?g#%KeZmqI0{l5*S1z417z zOp6wI3Y6evkot(}{(GLAgl=50p1Cn6O{lE&#|_Z%EZgH|2b|bcd+UtT8u2bJU;kjz z)!PkBZm{fqKX!*qTqa^b95S2|o`t74>&Y!Xbn(vUM<79iSI1$h&MUD28y3tZozj zAwS8%`igPdfnj~+VxSmMTtCG2@ykjzRd&krHZ8>OF`U=pak2u(!nj8|8+hvSZmb#J zie?B-Wos0Ex(fatS2auvN2adk$gsH~hb=W*T+SAsmjDn470wc--zw@>GntYK>;$AWWwmRrNHIoUI+axq-5y>*1n83X@YuP^{&dpbU>o zio^>$|AA?iJGy! z*g`v#_qNOQw;%C!4R2U$Qb2?his|bn;`S`*o+r;~kEw-k%Y3fgYeS=K$wq++;6Oaf;_(;5MD(EIf=o$rp zftFT)w!fjvBa18b5B(>gVzB*>l_8jTq(xu4r$rAef81cNG!amRoOf<7 zThj-~+lQlkT@BzvN0XQG7Lt`p~A;J;ktutH|>?XCzo3-2* zDyQW(iTcoCdyMZgeG)nPw3DiV-KuJPWdQzdhd~n0ji_rCe`lccBw%X< zlv*C%>ra@4^y0GrY84Vu!+Ii_Fz{%ba>}>Ws_0DGDMv^bh1F#R3>q%qCMD@eL+f`K zbp+nK%!Y+OsF(o%xl6$Ao4iR4u^Jrn{&SMJn@P8q51IZwy<*mqZ$*C!@K0K;P@B(| zbD0|?#|{2s6xGSz8!>!SW-X&^@To+40Ol#IM{X5A!(tNs+=wY+>p@eD-cq$;$2rRM z#l0n$o0p+vVLt z+KBuiCW1Ay!3TK!a(tC5v#_*>LlcXfi0n@32Vy)I<1UpH@CDu z%OLAOc5YTVrIFFena9xV(Vz?0e!t6GJ3k{FWsix{aGDQdQgKTk?qs zz@kkYE3w7ZYuYTuI|Siia?O1arp7a}(MF#_jhQKa?c$RSHmEZk8y1Xa2>(?q7_J_0 zgBNq4j($=NVdsB-*N=`I(WSNi`AkfOTYr|=kW{rd!Nb2M8_Ak-)J{Rkrx$#g*>O|a zzX;^Ta{bVsy>|@9e)w6AjsWh|9yLE`m;GOqw=H#d9|LYm2N#$Zn77VfrL8!IVe5&0 z@S_Mg)d;z!75N<}hbo#<@hrtPk^J2b$5%35aHr zx(+d1d)O~7nq7uHR^PDw0DR4gi@G3gcXQB!-QrEvIa3XWjn+i4_Eau(;7z74PsLu zC&_9+FQcI2Ap3FSVK-0ZvO3^j$W4geBxsWfxO@iB++2WcO}J~C<3k91?2(UfVgOm! z4wU`mK&oH&Kc`U7zd#*kQ6MgiXmDLmI7SE$4I)U2zECjb&T{!iGyOAeOfsK%DG2qG zNo>xjFT(o$@HJkumb6)eR_t+Ko!kliYg9EtQSIVA2?0{%2|&kDv+nJC&Zx|@ts-f_ zrQ3RPv6V|Fm;Gpl#RCz`gmpB`@(Ek#X(007^L%*2<;sE!n6Azdty@#_FUM909d&B; zMkE<5Knxh5Ok=h59R-~D(jHtkXya7If z?BloleCiDMjRjvhxVCDh=&}GC zp=>d%c{eH>pT4_aFL?J;Yyy(xRS5%gCDu ztD*Ek3B1j;96Wsphm3pGw4-4cML0%HbcJ%V`s$~U42+g{bxh;mWQgG+Tw*~P=HS>H(cd~n`V zkl{{q*-;%i(WC?asYfEF)AunOJhUXg5vX5jbarN-RQOPw^!op8@L!HDW0f$!p3FsDK&G^RVni%Pld^EztDa7+XevH{mqMN-^(4G0_L1I(7-1$sNgxF_p~E-_x-?#wvb zlg0T(%I1Jj8*5TM{8%rkh>-{O!e$=#)myAV6{S45mp%Etf+d=;MgVxR?ZmEm6h=R1 zldR?-kvQa_4hfI&Ps*F$!&9EzcV1G44smy4p!+JW;Rre*B^0(?_!FlqLrKcN%23byQ?^W#e=k>=U%R3NTqg&|7QbHCP@o#TYofZ(gMI6K5mO-&tz&_ z6??8uP`(6 zUcXnrn+ceo!F;2;1>)X2wsknhY$u+8i}z*HE37<*pOsS4MwKV;!vcx*eP%_6(nOTiNc)f&R*^r-%J=>=+z{q6&A{Cglv2+TKgO`2VPz}blf3Ro{2Y-*rsfoyr!CVn6nuxxG(uT3qZZIdCd) zk_4?A6mY3nIEmhq72UdlI}VJ?43vTRoZUsKP&h%R*#B0-Qu_bO`VOe3mT2t&A_^jkK@o(g z2vUSA5(E(lz1L8b(4-1MM5HN-!GdV$9iq}hFVcIFA|N6i6e-d>NN*wjbAtE2xBj=5 zk>s42edbKr<=cDqIOtFn@tvB`YmbV>OSNyLRSJ)$xgvyCNuLKog1!z6W@q3YSKOa} zzMf;=Xm%mYyvIFL?E9}vN~zmH?xK!sHFp_(S$~?PXf96drX)!1x7w;2qegXw>^_+l zCS|)KL-Q{tsSn9Fsui&XVh!N_zVy0^f-pLw!2eS23qgLSV8eLSEO63LHrv?n*BQ8G z9#v6HPVXn)Qv3LU-#c8d@v5E=?B1gXfUB6~x;XNnXnD$)nTCrtqw`(AyCbq6R-Zm8 zl+!N~Hep=1e4gVoql1y%K+1=-Glk{F*8|5iv#j88hdlPOiJopf*P^nzs94>CLH zAtPsVIr|qHi*sGrE}YCHfQzM^RfpZ>1JRgs>|mU}?O| zuFvd*ltOn`kX`Te~>TOmwt<3L)SG`0$X-O`bW< z)P*-7&K^V|vHi3Et6;%|P1^4dO#TvT3wTVn2@g68waDDQ#Zd+j^ zcV`Pcf|?7e0)p$*jlgcWs~~J^`w#3hxbetylSe{2;AcZHZX^|yu}9@2gLYW~no1$j zgZhnkpdsG|ehlcTY}(t3MX`M=IH>w|26%`5Jh4rsFwg`z;@}+)>c{_EAMqPMZ@#vC zhI|M#yb%q$L;whx_CnnDNd!m{;9&*xO7`x0Eoe z8&2CQ@sE;et^FIeErPb}>FR5kOAZf+?v{d=P<@7Q2?eO z&k_=~?{kiQC-x>>$_jqJ$T^-y%D^3#yWFpl{UTUl_0+77D7ICmXT7ZNwrJ)NW62;f z{`!u<*Tr{BJJ!AU$+JR&pEKRgL|PTMXv8MpCEhjEtPgQ`7$30{^Tnj4Mw8&QGw<3o zZ=_Tx^fpK#U%ofyJvp(!!|?a_ZwfDne*1TS;UeVtZ@0H^BPI>G%UWwFM+ZYrh-cQg ztpJvbSHL7$Jy`;_$J|l|X@qwBZ-9}$z<}FobVYk@?)F|ds;Xp{0-~#IpZmaLpo5E( zkWdzEZ+T6DSdkvG{+yb#yT?Umy@2Qt`75j0l>#y67Io(_{TL~7+=Ys=ve)CRN~`Z~ zdQ}(2Io!mUHao$8c+ba)P zH=5+N3`lb)G_=r){Pk&~lK0q%6-yp=lIT65&vGaa(MhROW+F ztEo4;(>Ve+^*d9$t&b_cyEmBJ+I%s(mC)N&1d{qW7{6)k_1m8@xr@+N`_$Mt%#)s) zbYnw@me=Hczn=4`_{L4!>ffX~o4LNM2ea$#n5nyX^;^Q~>`mwQzSJKwTsJXGF*p@C z#>sz8Sc^vL5wUi`#W*HlR;ESr7$L0H#JZTjGK${!NSSouyW(-o0@ki?rS(yl=EIN5 z_p9%6aks~(&azocd`r66B3%3`wDwonIw#06mQJjccE+!23tnbde;yoX{;mAxL~MP% zCQDIc2FydG>BTpug!%1TF$UFM5;OQ$@Wd#^OkuYwOGjYlitCaO?>)Fg+8lqp@H;oVNt%;r8)yI6!5355C zNoio7-dzdbC1^S|c(7G}XQ8bM$IUH{|DswBEDBDKeNko{K7u5VdtJ;vndA&R9vgA7 zahP~J54`|0SW>0Z4&E0?>ZutYL+wr!y{ex?sSSc(lc@R7ZeuDmo)3yAI>@>R&RIN2 z0|Mi|eL{sRSQ#^)weeF*?#pm%gCWBs|75>tZ&e*8XMeJ2vq!eyrB~{9TePoGcfA^w zrFIPs+{;{tlb1&Yq%>PPMBmsBJ}NH#q9#>-hwPTX>6boWymO1qHE+?x34K<3>}bwY zGt3V0o=k27PWS@HZPbKl9WK`o_dWy(-1MJiR!pR|#9H>MP1xk2 z_ghj}cKyho4a1hj{Q+;kh?y|y-c}<#ZIQZ%3F{;b^L8}mV~j;NvgEF{U-EANbKxvF z>F^rIjRiOMBmY!)B13k(Y3p8;Nm|>+r0&1d4L33$5N$cWszkOp^n}yXz|Z)nv6SM1 zbzVu~YZ;5jN!$d)DYx(K#+Xe_xv!LTTERAPXYJ1;6fhjvgl};W~msxFO&nE zKSPk?LAt17^d&jU12`U*Ndcmz0mpm7lZf2H&(kJQe5tXo6T+4{o}Zd=ubVf|e-n7@ zi*^U6!(A<0(e{#Ytct0;S{ zymnXLVBwBw%M|H-z8c-C=X}TrcM=a7t77>KHc}{7-7`s2&}+Cb9h947yUXGm1E5gx#d8 z8(6QgCj4kAEazGR4sE9u;=9qMs>6myvX?F9^6Rht3TnVFh_;E;lh6B`4;xFG-T5-| zBj*=xShs`B@-@YWhzXl~G^dh;sFim(|>UMT-6 z%Qn5AjHW#^N+Cx<%+Y&HodmEe_zzm^Xc!G59DxaHUDfjF2+TP7(a4R zY2uG&76;C&RqejFJ>ey|_|W3|oDJ6XAr;LlFO63ioX^>bDc*l_$F+G^|5TR5m63sR zQY$X2ri~?KvE}4~amlDrZtnM)PO042i$7AHU8)YZ`Thr15F1C-@0JuzpO0d^(w2Ah zYk}Gz)tn~%^;@17(pnY_$|`!Es%D-$Ds-PHZeWI&qKnU|SLyuqdwr!5hTm&oKK6c% zBOXMv#c*J6+9!N;?AOizy5kgtuLc9CCBjZWGzZl}iakQ|a76e2%_Ea~^%H;FbbcbJkmz znYS`qB>TC-i=&Uz;d;U>hOxBNk>2zeHC|XMiu9t032eRUwg=%=7)p2*$isajIUOKx z5fbQ=_WZ7r`bYJ$b(rzboCl)$e*NmE!PU9XK1tWK{t7F572nd7V^`C+;4_iV6GRi; z9#z_0`~H{#ifMY!eei{>NeNwt;nr)5ZlUC|7n{8{5kqg>rKd`rS*#Sxid=Ghj`jvb z$s#M>6dk!&-D(^8<^HATy+^r(ys2F_SX#H=EcgGzV*Zfp(7BabSikyRWGwRvwVc(d z@52)5x};0RRGhDF31nn^${d!HxRQzpLiJ@baf^K`wrr5CAL8mVaxBW^dp~;GdeG~z zqDOgXN|?u^)!HcG#(SA-r;1|Uf1IAFt0WH#OAVep<-M8_mmu2pDA)AK&x91YYExpK z!IpdNS)PnSr;EHpGp~c-#)?Svbw6yA=eq59Y!Mg+TS%(j|D<#vGmR*h`KTTy5V~tO@A0~s~i|&5Zco)Uy(ABwayvNvl z>Bw8o27?MI32mh6gG0K?Ys^hKH|dkkf$ij9emxORXQ+%GUK&|G*O_K#4o>}8L~GJE zgiQRVaSkHhX8V?2VpknbGvjQ8PZgEmOa1aBq^oQmy9*o;z`Cogm&PNlUj; zl0Nf@j5W~ULvbTYQP*(oMksK-@h1HCk+sXD^)T`tvMdgA) zx~8;ufE#kFGj$AH90WX0%ACB5wAJ33zr4R`|2+u~YmXy)s&40{pD!&d3RZV=zVh(6 zf1qq~ZDK^QNv9L@d{(K%`;iYQl?pMBGp7j=6Fx%yc9DixSSn6!kLAynHqY>zThI>l-_|7X52R+_e-70-y9R9kWXP9=iGib##?pwi%QulK;TC}0z zDG!fx{0e&`&&m3qwDj76;(VQpaPKH$e_Zjn*5>$j0Ngvz3Y=;D8&@rmh}#I*$nk6< zqkvKJ^p*_3xvL91jw*6PpuHjuJ#QP29hE%*?Lll?@Cy(*06YA$4X#J!!^ezd&Xky- z_hL=l9Yo{9O^g!5avn&y1B}e*0`v2W9HD`}BnKq3v~Q94pjz^^SjBr8LaAYrhkB~& zr8kWTjs@|@s*b?F-^SNC=g)e+!pV2^Qk_Z=^)D%#v6x(1Ewc5)+GCMOa!=s%zncm^ zl#reW?YpD8l2U)-nh}U~z~0nsgwKQMW`J!Hc&Qjg1NqkLDk?f&HDOE#2Q+w!V@88H z>4guMA6gMji9cj^zPoF_WZ)vZHNwP7^xM^6OltlKNN_BPMS6>!JL2~nd(SL0cb_b| z;g2R9Sv3LZs9K3RIcvOt`<+Jj$8${mVJ55TTK@XG>5Ury+y^#Q&|BlVi7uCnh%kiU z^<@j8rx(UvU49nBMYaD~j}(A;IogD`Rt#A$qb`w@m_~e%U^;4tp7>HfR+C6E@!NJjrLYy}y(knooa&!wSC6J=&wWm}Vi14!{EPs#Hr!0|vyn!DFgmWr>le*}raaQ{+>IR`x_2*TjqU6BB#7t8OufCWEgSW%PbhXHERJ7@3Pp4-VT&cYg$RCq z#rY{u|Eo~b6{DcCHuGictVt(XGlGZ8!ZOL@u3f17Xq(`bdc?+(m^ewx0bg2LT2*SF z!~NLie%7?;#h)&18Xuqg>-jVH#AxQ{%c*8EaMG%sf-e}p@@K$PqLhTmv)S1Se_+X% z_a9;!a3CbTC=J(ChwJA~DL#YF^HQ=WCDj&~sS4Eb7`tA(Cv}uVWXC{hwy&ItzLljyZFT%q zNyH9rcecmf&+oIc^MymmuCDEXAk4^;YYWdP5F;1E0;1aWzU{|Waic1*vOnNqroqd4X97=J=U{~ zt9jmWMEBzo^X&+jCg)>fSCKc919eL|Q!f@|3K^^3Hx+GjeK=6ET=<4fmmK2A_tKes zsg?`mwibPJRY1=$Cb{HIDgA1JhQ2!;$Y63=l!$uuc1eR#T;I+`o5k7>+8zWxG{#GD zMB8oO54S;pv4CpzY}UvL#5N&rimG9Bv8S2iG}Z;=!r zzQm>A3P-!c^;kJ0;8py3I?WGcHd4I1e-}J{##MBADUWYw#ZVH(DLRDi*q?r3Lo-50 za^UZ>vVVN+4-9DDBN0v}`LoiJ#TDYGx45ktzh^$cg=F#H?|0i_?~e54c7FNk#vS3^ z-Vw&R&RK#V_m`#oGI^G|Kd}5q?V|+us?_2SSthmh-V?c{k}r4rWMQKAmls*SXQ-|X z&ik2d|I;9hJ^kf!LGaU@%H+=2rjcCCtkBaN4!Ib?O}}eJM6m93rVn3U?k{?X`R7f* zhj;b^KBZONaaoSNQazg_fx=FwdzXrD7v{G*(l62vDLuYHnx~K8)?70N>2QRWO894o z8G8$TtWqC^vQ9|ddGJ|ev75ywDm^*T=-(%W2Ffjug zV&7lfS)Ih7y7^$?tTCP4*-trc24atW`KwdoP<|V4Z#ib2F2;MZ-Z9`>X_P-4lIc*w zP81V6rp?{bG~bBD)!D=w=ds=O#n{s!s<_K`t$TcRR)XFnt&aJ)wsL)XaQXyYjhsV* zD2HYCoHW+K=AqNV{+ap{Z{3Vv5q=#Pw7B8au)dEKE9)U?Q`w(&4WJiN_Zq6-$s^6* zi0?oi^%&H0%a(Bgn_;xe#Ch^X9kE+R;p|Zj>UDB){O8wU&T{B?A|Gtman3m1{!w*# zBG_8{KBK-|+BbavSE=@c~|tab9mFZ#1O+*%PbYJ6kS z;2)a=`$FHMcBUg~J9D!!DLVf*8jSoYfY18Cg}@1h&Ob*{U7|5 zJFP2oo>rTuC*+3X3a+h^pK7k}_&Bqz#VzhF@6Lb|CQLD9u}cG&m-%71Uvj03_9YG; z06)`o-Z`c}xiT5xFOq>hg(%#lM;QIrni$Gj2xUA1$wPpCrx~Kws$a zahT`DmE7-fwWph-bu$HAcZOBw`wmh4Rh#g!H%83aM z7ln2KM&tp=QG34#)%8vt@d-hV@!2I) zTM}9q30`n2M7TsSW8_Ol!n!q}vkhP%MtRm) z3_zp5<2G6xSb$u4H@>$Hf^aMe-xaMCIcC6d=Ddqgk3)ha3<+8d)R|+wqg+f(;bep) zFqQS0OnN&|W6F_pcgNS$qU55`ex3&3&6|U?OnzGrKYR6jpWp92jUkxr9*kkfJ6`S* zxU+*poXprKOG=k!i>l+}W)$z*za;bY4Sy3*)<@0Q7-y)5te-bvymD6`K-wOHJNnAo zHmEmvtMW0EEh}R~X&ypXuSH~^Kp0zogZW2*n&3Wv{87M#Fhs^=I-?D(Q@T4sSEtll zqhe($-sP{;bQm~R_6*tm)apCY%bJQ8R@b0~VyPk2pIQt60B!vs1dCUDkGfD>-e_Q& z^s0UDMU>8rO>A^_?r~=~m!O?rBh7u+Z;0lNj%@D$8$3slV*YdX=t$^lSh88=j4=_V z@i6O`6BkR8d@5B(M;#FouVWdMZnhvENHvMRR=Nc@GSf~qS9ZQX$t@o&yU%aKo|)aL z6D3gAHCX%9XmmR#yPwQcd~~w?%!L`!TBd+nvoIkI>Njdv~*pMgWGcmvUq)2H{~E zJ(b6^lI16vx^S|MhEMjKZ{B>@$A?Tr;iN$a@zGy?Zq{vF{B!OLzD)8oWoa6rX~Od$xgi>ceyI1{3oTY7uubE$E0swF2t+2?17? z^PD}4g$Y=N^YCAMG9V^299rc-gN7B_hWd9d^F4EE#G;|6)@45+TA9xnt-y!OG~p!?Yz!b*wxeZ*=Pw^KLRb{^=xFOcj~buX^M zIH)t=!B}6ER2IQKlQma(GuqLtMVquvhtP8H?>WAxG#m8Tc&Goa9rgU_K&>N#%9ADr zEp1;LwnA!Rh%Yfrl1@K=Hmhj$2UM{l`pQ3`uMYFtEEoWH0|!fMD#};1-^`*?Cd>1k zD~f3UwhP7MVxzYGCTF1pEMUZM#1e3k2e&zwbj1j96!b=uQz`^(4MHtP>Ye!L$kYzL z@0R19IhcG`4ZprQq|$rFn5IP z+CT^EDssaF;V}Tnhd3*gi_|hqIM|R&{)4UwCIpK~VWd}CI^N``XGIe+n5H5ic`5&g&37!CP zMR+*ub#NEKJ{j=3jZ`*^ct&<$QZd(qP$=#7QJ-EtJxRU%4;IWHIUv`uA+HQ$W^HM=q%g4l%i#(@l zh-*WP_r4r|rnjEW5Ka2&$8I}!QvR83rcv$v+yw7qOBNNgh8DpsV@Fzn4@##nd4o;P zOvAP~qFw17&2Q?&E_O0c-h1cQ?cFTgWA{GF|7^$2>VlUnf4vGEV)XaS)0>mCn^A=Z zP8$J%G2)2?Mv21n4~WT-+_MF|>Mv>#f5@raLCkm>?X;z6k0^H#F+!~fU)~Q`x9(|3dfId*stqcI~g3$nN($5 z+~;;vJCePn2j}*VYgP0p4ZO*W4;eej6TlwF>~PEQA*Ka)IqdvE^KmN9h_{m=U0D&p z*bhg4*DNV*805E^RiTxL_DOCOgu|kfpyU*mkq3ij3uco>rwjumYiW>Bhlk;srI$fw zRoB9nnhAs(c?L>xqI6_YPf)JtTGvSnem4;9%A9soQBsdqT#u^vyRobdo$r+mFFEC_ ziP46v)qBO`^FImWZ)>N@T}0nYvwRFPcx_uBiHC~|Noh;cPcubN)Mo7fx|pZ!Vrl>>h}iC8RI1E(0QdrIaFn;5BHInTRx{$1X=2~LOS zrDtP)o;zd7G_EImX;Al3d058I)Y4vvLS@{RrpH6fK$DA5YMRQd!=aNYJ$qd@XrAxj z$IhE}U8Odqfqe{LW*XknjE0aTUjJFjAS=T7N@M_`R!#XT+G7h)$Oq z`#cTunZmJL`1j^!E-nOSZL5WpS@>dF{hdN@V3dnMhHbz=zdlejKSZqA?`SgxQbU8D@fvN%40zpLPuwC2Pg*q1$RPSXux~UA+VawvKAh>8h%0&>+hw3k)nWfKbr_LW$mIpmj_O zM_WQ%0V03-Ex7FT&bKj2{`gk^OtL=jsi{;j+0h2Tu!j zL^xSLxD6{}&^34XoHzpGF>qn=V)scQyR_;~Cek`a8vMTjrxN7Bq6z`%E_FIc`k$Q{ zl*D0GMdj&6aN*2aQoA47^Pg!MW?kLKE4X$=(#+Zzz^GoX#UHh6Q~ckB1YxQKev||t zQ1gNFsrgzbP68H+4?3u?{NW5~-Ta&Pr6f@V^c5pHQ!|~|K{G3&L80fTN+EUHo70~0 zY8!&I4u5UePAbI6h<7tyh(s;p?IW00@UYHe)QS*l1L&D*7@N$)-QU=`zJg>|g_KT; zNqp`{fS4U%Eaw5AIv55F&{wICb%Gl*45|b^fVH3r@-hIE(L&bC90Q9Pv%HVRl?rfI z&t@CPZM-G_N+~|cgcVW7LKZw=&^&f+a9CM#5)$xDptn*B?=a=?PbGXR;S3I%!iFKj zdlwa6^Z~{nE5AdCDrfuQm()Hn5|F`g{o2S7O|~peWUnP|_a4){Rvx?JSI7jFWJ+(l zRdRXp0VnwmDq8^CXW9Q9W*#|#Jk(mkKe0bO6 zF0VAIZR-qAULd8{5hS{O?#ZkGy}S(K$6)GFWOi8+GCUl{84;7^l7;{riZkL}Cazhw zW(Z#HD;I#BL9MKz_MM$390ZE6Gv+I6V$ct42+x~YRl-eOSy|v;2D48D>O&=am}MBC zPB|f-7654!6b%S~wzJ?8MY8Zgx;i8k&LANcY0~2Y;u(DU_mXzH*K`MQO+M`%I0+30 zmy=k7SzST3!dC$71~4}fz}#L}_0njR6${k&4aFzWo#(Gf1dTC(4b_%~oUOV&8+yynU;y+MT&eS<+%-*_wbs4X6y?i|QC>Pgo*$`&v%2MSk$4|IiRA0r}O2^cJ z&zOGgGiMZF-7Rcs;sG(68bu+$Ym}QL!MLNWw#ogLaqm~wPng4 zy_0CTy6Ux^liMIvAQ<3(Vxjo?mmFomz%r79f3ZhN8Qv8od5^=0|8D*My~h;(L^x;C=P|1{QDTBIAK2+XrsC7yDBsBB>ni_7EvpEEZvc~NgTZYn z*p!G2T=YH~@F;P^z_e>1PkFn6D6A=4BZ7V-T|aOt`t<1(vj|gDzSx73N#xy*o1tj7pl^PAV?j`f12X0# zaPYLJIs!pF3FTp}-n+4UoO72rT=XwTf2|GkmY}DeFm(nf)A?lTC^(u|O?Xj#058KM zmDW^rAR#UBu&csdVeqP%WtqctL>^i}c&!GyQ;8224YwMI3|xmrZvr{0cOn|(olvpB zUzOss7nbqH3optfajqW&f8egj!Ujis-0E?s-C0Ba;wWGsP63rE!Zw%#_FIaEbk-*UI_xoAU~` z#50!|jgngoWkmy)rY5Zoo5q7?(RL-LaSjdw`*nRYhlF18TwnOU+2%Nmeb{y>J?)XE zN9jP9@zV-lE1$6(yC$vMUk9Fr2xd<_V!y9btvaLRf}XakNJtyXJXc(hIBX}DGQ4=a z{;SJzgZ{k3X@}{89XgC}m&!PD#n$amEHC+bCqM zyFK7(fQ^EPe;DV+yNq)qVjYG-Rz~G&gDF9mefRC6X4(*Ej1xo-!>?}M!=5< zB$sj#;Eyj}Kwc&If4aPFXfUMP0+1sNsVh^Dq^vdnP5^f#A!C4af~*I{TED;RkHWCBpAji z0gOS02&sQQ_7yi$J5>%lHj2|@_`E3?w30n|t^dJoddd~oGgvcsgD(C<)JVmzNtUz^z)vhO_)Gz?{P@)wA#|-iC zQglxE9dxJ6QGbxXpN1I^?iT7Q@v9wQr^w;>ZrD_4BwmJ_BaN6s)fZO9F|$lKbhZ)h zkwk>SJrX0IzlD}+B3=W{T9!nF>s*lN!tGOMw6YA|``ifv|KY z4`EU+P<%Q-PSMbdfZ+nN>hHkW`cu#;FpdYN6tG2FT0YRULKLnB>@KV|HX>!BdIXPIn#{-9_hWW(qJJwIr&Ep3v ziYsb@Y!-r5Ufl@~PQfh`)^1PO#XUSHmTlsN-A`Dp?ETrjbgWM(VyF$ikmdL&Q>PH$ zjL8bAd_^_cAU(Trx$2>W+pdw>rDY~U{A$@}#cQ{}281@mP7wT>r?bUBDM==V=PY>? zOJC{Iq6*wSHN$g>sV`SO+rFRH*~MKtkZvPaLV-d1I}?`UQq#9~xi#169fEr9-fj{I zQ_3ijNC~rKP%Zl{&a7dK3K6Y~uJ=G`^wHl0@V)`32%c$E@j=9`c;1RNbHB^`!Q*tQcbXEa|F6*m^onrE$W=M8K|mD$9msy9V1UU~ zif5DCuAZitnP7&c<>1D@IK`9QxE0;NnZPbybZg1AEtm)_ce-Kn(a=;iU{0qTs*B|p z24I65uy|`Rkb@7BX26*KYnao*Z2@)x5rzR3=0HJ#B1~mg5M61#t!t6~PYkqr`Wo_h z6CMBdJn$338o^AUA;KXh4ndQ=psnC);&Vv9O!Q_~D6-naReIhKHsw+PC%s6O=?)L8{mpTE`xU29X9EM zixYO#jXYSUBH&bzqMNYWvjHw)L`eM6nY6zJCY1*k|G=i>NdZ(3J^7i6W&aj5ts!lI zfD{GeMo9)M% zMzU|+;8otDy~?g+&qqNw|9?aQj0epLNcsZK1d=X>L!$(XCN=^*R}l`eP{=Tc_JU`D zv;o_D*siSk9Hp)PYeS?0OrGt>lYzqO6Sf>ngCeAq=cA2Akjfay!F-AWYIlx2N|K?3t5o26YDO+i=t z41)E;EYzn!Vw^y;0l~ma1ewy*RTZgEhen1~0V^XAB0;AH9Hd|)KNty1$iRTZV%H0G zWvG&g$w-)`Kja*DhT`-X8hPv=CGYB-PXdpHO$BAcIl+`rCbBj{HXlszve`OVx^0({ zwvpbaLev1ETKTSWNfH3^O9kFZ2_zaap0mb10E4OQ30ZeMcC3X!{6fzKlPS4PkQ;WE zc!10Sp#oWw2kJ58zSZaeE`$RUh`AsaH$I!GR3P-ut^TCkptdDj6%l9#aT`E6;7_`R zEN*PwOQ4S7i3Cw@inK!Z2kIaQ+=b@SzX%^h@;}gkGNCE65O6VW_}}T4c0XwMN}TUn zI+mfBddN42k46MX&u5o;UJV?XeuNjU16~4j%L_a&{1`9Pde95h$L$jRb*{F-(Ri4a z9q4Wv0t(cjI?|4XR#778xPuRc4Aiy50qx0gg_uxBRssYRx94H>J|Ckg`+$DB}LW*19!v0 z)I%QejHFJ&S54L0ogRcIZZjIJ9@pbjK0H`o%6HqKMd3d%$u_EoWV<#UayMe_Yv+Cn zYC0xf$UI3@eoqD0E~2~{3~JeWmpwr`>*ET2Noz=)T`p-YUdU`LkUwY>xIbc*f)-C6 z(Ai*7{#y&W4r;mMiTjCwV$z->4x($CB&xorK%x7FJev6F;WFGdok9?h4ceKqV~BE5 zM3XxZgMWX2G(97tgpJ$uCXXiV$>K`wpbLt6loGG^0-612JF!NGoWr{E{Z3nI>uH55 zg--@`=s9a%G%kG|cc@zb%Sc>Tj4J-l{mjFrwG-&6ywS2yPB^Lhbyk4Fc=+`9p6{7a zEy-K)44gInP3WpEEV@01I5*{1)2*)c8Y{!7)QjYS*Wd>90()hoW3 zDf|0sUN!qK4}R`FB2OZmK!4m8@(Yj2h;xEQX>$cCJz3eQPaZ~<+CgPbGenXjlvZ^X zCv#={-f{n|mSI=C)nw+`t)l4zFHMYoMFg_-P`oZQKfz*x+`J&D1_M*$CB{=zrYMx} z0!-*b#J{s$jTb<>;UGQ_M$sSCFcLZ-7+4X;4OS_`zfr*le9sch)g-tps1L*xU`pVN zFr}alMTkICn=-xsPJ51-|5;F!g{8F)*5#QZZa5k?4-|-l$PmyLSxEBqpxr=mj65I$ zv|oy-|5`c&l{%se$42pdMJn0bkVZfThEdeoSCBW21m<`VM5kK?$f&UxZZwCf8w?_}u5; c;R0=_{aY@XN|vI9Q)CaN1a&89;h!J>4_AWW)c^nh literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-17-Nemoclaw Agentic AI Architecture.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-17-Nemoclaw Agentic AI Architecture.jpg new file mode 100644 index 0000000000000000000000000000000000000000..662b08d0b2c2491f3e258ff270ad506eb3ecbfa7 GIT binary patch literal 139575 zcmd?RcT`i)w>XLg8~X8s7*L9ZDv5v~0Rh1Rp|_BPKnU=WB9K7nAOY+sML@bliiVO# zf|LY8Q$f1akOV>#=^!9ovHke2`+je&x9(f(ckliE-aqfoS?ip$X3y-|duH~Wy=Tup ze+K`268hWR*vwdH&mJM6J-e6CpJ5?GA>n=d_wU~)yn7QC7CvzB=%ItV>9~l<;iIC* zPn-}vE-ES}E-fK;QW794dg}ZsNg0{5XU~dB$jZx}k(WMm_RPPK?Afh)@W8>p4juaI z%t_IcXa3Xa&pRQ}L;DUN{%7AFS)sk6d-jR$`SU?YW|zpld-wf|-v5~n9NZ(kcmHnD zU9iXBLVNb@+q-Yyq5X#r3yU1te^6-8-hKOpMGqW2ab8{Q2K3NLS&dtP4^jX(?Ouz^ zU3N zv)hW)Zu9?78vd=70|$43S44M#d-ndT^Y-oEC3Tm~e~}=%+dfoS?Boqujmv=mIZeA; z@;6f+ye{iA`fB+8z{KhwfzZ)?yER4ki3(j4T7CXMx%>a}c>sF3_vpS0T=ay*^Tuy` z?gts`pcuwfVlcvJE+|AyOWblUZ*V*vn^BjWRBdbU&s2;4vT+%8+9ftN*i_LWU&ClK zF+O(IJq>}E{h5CJxn8CC%iJhj>XUJVT}!D*?n(10NJ~JqPNzqm5jP~Wqeb>kBl$p` z%m5Renb2eRu*fIn@~}Mieeyj+Qp`S~sS>>69po9sOASK*NM|G&7i3TqT@pTfrnGGb zlqCgylil;~bYu3X@Z;h-dp?|=a24XQJ0C+3TYK(%#YFuT{2r{AlJ??3K_UAFecUXHo^?)Mp0 zz(n`urG2pmWgHv1JJ}*5c^4iYiI^pOUcJ;{=2EBP?vw@m_RsO>TFD$fPdjc4zeS3F zH}&hOad9S*VBJkTEPHP#GT>Fz_5brjtugMl=jveY@dFL~uL4yB?o}o1y@F?V89=OS zeiDcNQYzmuq0yiOi>Sm_p81U481{^YTvYW_p)hHpw^y`R?3?K}G`w`d6V~YXAdtOkx%@h%VnQzf*v z?JyRj?ct-JxF4-K{?L>Vc~n@( zddOfE!#^YIqdnb1tnVxsZzC#0!*j`XBP#^_N{KVN!)auTT5K+OQXY0uqpifqS^+bG zl;#SpH_N21I}GKT`IM4^r0aH7G714h-iA((vz2>#=EW88Qi(gbx|9~qxW?WAUa9? zcLeFurYW~hwYs0`af!Cdqr9=hV4tSV8??VrpKyzYKn($>xmAlQ8R@k(zJH zq@?zUOhMI!CGZN-AN7Qz?`E^;el{WmcQW19*Sf7I;3>1vgWFAZQ_^n9u=;GBcY47= zkXIGNJW#!J&pX_%2|?n1wD5eWR~JH$m&dpr?7tSna3R-aAoZ`WiA%iOfmlI7(7yMe zK9^?@EC9N@@{Khdm_-+L5gNW^(a>Qp^K;vnnt_Ic!_>{JHH!dY zhPg#$8dF^+HUI^ug3qCH<41^XA4y69tFmh0;`j$QNBp!c{0xD#C!(2u-{p+%;V6*0x3hY810lTF&8&(3#-=ixn{Q@~>>k&5aal zK9U%|=cW^VlMhTUs_KJtyO&emixmd_l#C}y5_D@&Rd(?Hu49JIjdO*Gs0~+P%zAs& z_)3;OW6=4q__LDCz96nIrP^5)PJ0iNbClJ%l4`U;7=X-iX9C${IMg4Zc2+5JK{taw zK(PK{C9J?RvbmCZ7^a_-yzpD%78D6;J95Ii@QI1ZMgPPlvsS!8H^2YaP281 z60h8OZb`#XZd^h5nvX{5Uk-Ht$lzzC5BzX>FYeA_yWVo`?+AE{)#snE9pYlKvhbuL zq;y(p;7>D`@db>P&2)NCfn@$o0U78Q*4C#l&UtS=BTK4>cK0xa)|w`QU&>7+nKZV# zCp_YIP-TJJzmNuBK2F7PmB12jgcq^eTG`SP76yJc`%fM#ZOz5T)wVGdJ5GPK%i*dUV#fM6#h z*;P~`@s`l~=rMq2G)&o>qD0C4U1Hl(8eHENm!X3gW}FP}5=I=eJn=5JyxbJcH%{Qh z0W@-96uje#_Oa@qE`#FoD;DUzoyC=S0dCD7CT>@@gbFu0q19E zMNFiP%@L}Fd`M5Q9DE@OZmFnG!a@dT2T*xKUI|P$djw=(k%#&3t}|2qCrA|vt!%wc z#+TvokV4LO$4dx-ACZ}t$G3=`cWLQ-8kRx~Xe-^w=$JxuW}TYCg~l|#z5e~5&Km`b zo)u6R(9$yPVYVne__pcTjJwE_>BUZgurhn+Cdn9^Fq3Cu0~2PHZM4JZ0EsD*)h2g> zw+SFW!cuUOqYAG^&|>)AGEcV=|0s9nTm*qzz_AhS>J$z$)Cp zzmxQbraz8ETwhE?RN>e3>wA&ZQ5_G?KriH^{q`0J3T#P^DyK!XF%YxO>8$~0!tAL! zC^DJS4tsaYqO?Qj`B-Jk{Sh7ISEO2ulLsx-LDqEWYH#QJ;Q*3+a2lnx{lI%j#0X(w zXVBX#G_?m7SnP>SIkz-8U^*&Wswz&M4(@yoQ1HuKftbj6dNOJV1aw$Xr}dWl_pd?4 zCl^|6uBMbcdt#zajs2H$3f!>FtLlX<#)ZD*1ciGUk8kAmr5c7+*^u1{nsHq&fhK)P zmrGJQc9u8&(Wn$NLzxQnR;x^LqUN=JUL!aLPbOW0&#%?BQ*Pj|w0-r3+OC{hb-h#PZG{ zMtl;nQ&IeBS0P^A&G*U#PO=(e1ewZdpW&e{64dTi-Q{mYAy^4kga+e&v##VshS>>l z>(vl{)lXOXMdYP*jEKA1qO))@@im>CNjRgUcXh{^5bYItpt!I~kwP!vF_B&pPr!d4Nv{98@OhdIWFLyGBsx8?NW)DKeIg z#heSlG)x64T8nr3HO}h2szg zT#_^^)313r;wz=_dP%U)(dXA$m9aR!MLl{u%$A2}o29DL z?~49<2~~d#x7=(y1E-M4MdWCG%M*s0xQVDBb<|AhL*hC_{pe-OX8R7Q>p&h`6g8)4-8IKvo{1NZIH}RGi@ko?2nhsJX^$6*Z#{9$p~ow=k>ja} zm{R$Lx@6cA;K^_$Y6b1g1YA}sc)0mjc9CyR%IZLS_L;mZPh+CHvK~@dzFrl(R)*f0D;QwDBmonTddeAi zfAO+$_5aQD&0l=flTBu=)hl1TWDQoDzla)BUb5a1h>$-%s z30U!Nvs>9SxZI8BqF;v~;nIS!iE(0m!j$QD@Pj&AKPn*HZx6G_G+GbYe^uV9t ztE~^*(V7`aXO%=;Se{oY5tM;oTQeo^IuQ{!6WM@}Zg`dUN(l3>M^DWJn*P}WEU1#c zze?(%UF${eF^6i$vnLbL>Y@@vS!yY4lBH|Fr$@cGXp7M|u^@e^6g2wKXo`5b3`Sz2 zv2g--IXFc$GL5jY?<2@Yx3Qwo#(_|$iU%Xe4M*xV(2-t+n$xA;jNufTW|2!F0aR!& z5OE@9<|!aAC-&F7TZ1iwEN^K<{VeoH;*UWT1+8AUvpM32uXw9GuZ>fJ%w%X>`;M&F zS5=tGZoHHBGHe#N|BP9N#NeiLbCs?i3~&^DVQSLz+RM_ey>q(ixKWy=^~R#HOE z$V#5|kJj)yM5mK065$uM-o9wM%%}cLE-rpMy4B=f3HU3%gwW+oq1>)<#%U&fqGa1y zA23Q)D&#TsQeZDmLaYGJU@V!c?DeFGF%hEnZ+cyEP*m2e$U^8Y~qUnsn zkB|WgE9}(TXz1OnZl-LGTct4i3 zShLFbA@HRS!8&LN|0uSTvZ0_6shQ03M<0Ag-ij{9U6B zyfe|3@Q4=E65uk-G^v9R`^cke7X7(9q0Ev<+$z5{C^Hp(OEOI^W0m=Rt}dMvG6_-` z&=hqdpO|!Dsu!M2PN*~+|M0@CYEicIV)%nK3x#RFe6ynBdT;3ES!K)Loo6{M2J56< zv8JZ9Y*V%4?x1LoSGUj9a!&X=E6xmqz?q-tY5EtrflW}7SH1C@5^FeyQzUhT(A@Zy z2=vmQ$?{%e)oV{XRkIdQhYDFvbGXr5`M6AwWDfdB-l?gB1o?{50v2mlp`4knHYLb{ z)yujDuO`g! zte!|!<<8R}@um_w_$!P~Wzfqy_#C#gG=?7JDT>fQ*jT@<+t^Mmo?bj=#$agaBO<78 zn^p2y#5|`*KYV;Vd@V8(mp2G}FakiXc0nhryK`KRlFwrO8XBfyGrS!jp?x*4`~RBm zIY7$pm1}{p(gYlJk$}i!_{=DP1VLhq^wyd6)-B|iY63mA%bDs!V((7%grR|t6j^Xe zA!Qy+j?W=bx(=z>e_GJ@Cx^rs6zAnl!h9J?TIj+A0IaN^dB-!&fS?oxyUN$u`_@y% zaD-~aoo?dqz$c(v69VHDB7pw zvVmmmYD;Taw`Fm3eEs#;@BYv4as4}tM_)g>AWWZHh2Acp4}>BGhiKk*1K(z_g7^|_ zP77`L6D4Qr$+F^$EVG4wNk23X<8Ya4pk39o(^Qewi~Vw4CVmq+Rxp(B`Sf zP1lE2YRih3w45@cLaGw4ft(bKN+*_7@|}*U;IU| zNH^brgz9UmLU4M0F4VWOH9UG^m=eUi1#f|b4@pzFBek6(Snx@F6`qN%j2j%zvgYpAs_?D!GV#>Nqf#B5rGNXAk9y3mce(e(={A6?@cT% zn&LD z-NvBcB@`C7$(r^!<14$Q=LzDHr8X<07b61WCK=^*nE`>tmorP3_hA(&4Lxs;vl=Er zlu>-7J=kc5>+byQ2iJV_V%1XY*c`0P#~2(k%UnenYO+5({zr@vOVEswOZ6s<7}3V~ zvb9<9=dEt|(aCitn0SJXepLHDH~Wdv;=sm+ku}9Ndwbg@nJ4pRa+73Iz4fG{!_DV!KY|(BPT#xA?&|p2##nN~FtmFbYkkEKPO1SAImN{Isl+9BhZ!9h zk;sg=Af=Dcf9vY)xS8PXL-yZd7bGnR!oHdyr1-}&q(+t%j9jx-_=KGwuP#cS;TI3z zu(DCje-bgcV1JDjnMGhD2^*{DcHB5Y<$gF)9olEy%py4Hh`H(zl}pB3CWjwUgBw9+ z{I^Lq<15?8&YO9Z^Z7M0dgjev&INO>AX;pX$8?uBu?Yt2K~4%XFYQt1Q0*Q>|NVeB zO#Q1B?D?e0q-Go6KGx1?d&BVS(kFShZF}GAX|#K!9R}5?-f>4p1wgkqpbD@-wsooX!uAPn|rbtGMw;E8K}XaQe4=(kX( zV$?yu8pRjjwl$7jqq5?=qBs&qYIAlF?f#72%?#fJ>se4i5a}V1>M;;u+Q&zSGGBh2 z2LeSyHkndGEsUO3@11{+L{+f%I{IM2kj7le3{Apas>7KQlM??=XdThWj?WEAjEAp% zXgbre=vdPoowxqe1;Shs73jZ<;<_vO1?O0p{ANw2! zU{B_@k0+ZPUeXF7wAx|)m~~Ef(@k;dXIl!B2#Sr&%28>4z+}Qx-uC&kewu=pJ}MiT zExJTjGg70+Mk;7C-WO;Zn+v@X(C}4WOW#@X3$MKh{Olwskc!RZ20h6wYd>bJqdvG< z5T@Qk2$!iZW;rmN-0>WVnmeDVH7=*EOzg?A-+)pQVp91B*`l;xvNGj6_!VR+eY1<0 z2Ly#Q1UlJ$8yWoM4g)fRivM1L+w1&5++o|X``9&kI_|*{0i~Ylu6aWD`A#736%E1d zn68qfyYKs`!IRzUk9wWric5H|itLM(8%v>rW>*`9M$#chNJUk`1Eh2{v~8@u)umbZ z5yH%(c2lFjc|+waU@fKF$dPtx>{Je~^qS3ux`y{9{Z;tH8*g5c99^bUiz8!HusO8N zT;S9(`|1Hyg3Y|uaTB+a)XdWt`#x^n5JSc-#(Z`)0oMJr{t#M|G+D(YG;U)2%4a>v zF}&KImxUTVvZY!;ApXu2d%1m(FZldNsCdI=9CAC4pcuWs8EKHIBhx%L`!dD+En?9R zN_9&DEiEw0PqAA?Mq`JTF+tZ={q*Z=mZJ*?LJ{HZiZ#$K5g$yh*yYA;SsjTigSvPN z#6rRtm8~^NRy|?Q)zXZU4@M<4=8C1BSLh4Dl?eQBTVQl$y9IDUSIPH-olree=%0W0 zG5+7qm_1UFs0Rslpgje_fN|rGSH9;6i6^ZrCoqK@dkhP_>KsFsGWo}ZW%P=_6b3Q% zbAdoPGXs6g~3p%9I6uWg(W~Sb~M_v(3er}M4elpgMc5?wIvT>CfxZn z!jNPgu>M<%b$iKq@7%*&8uzr6sN~4EPlq|1#iQ6!&_7i; zH4g9P=##_?=h=e!$vj=^?VZF_Z6Hv;UKAQoKjpT5xA&yvxQ)GcEqIK;#biowZJhL+ z5$Kt=q>vmHmqYx+k1?uOO@;Ql)(Y~TIxOQ}bH%@TJxrQM?ZgsuMddMiL>)xed>*6ayd=@BMf&!cONQXX3`kyc9b(Erc=OISCU!|wQZ#HV?PkNX#HOqm1d8@- z`_p|P)nhW%f@>Do#OPqwFplCqAWjv4LZn3tD;KX(lQ*jGIAvHnPQ|s6#VGfMv#Hv9 z+e3$=>p3dSfiA5V(dInXYv>!UmLv41ioIT(*x?u#i%+;=>acHuJCY z_xn#UOEJW+CbqNFr_-kx5ctg(LKaow4#jmNkz;goo6DXp1&Pl=)ftIK#!NsY|4_E8 zm=Bh>Rbv%bH>b@()J(hLA5Kco@wTiyeC=sy_UF&DBQHR`!f^$gUDU1$0>>B*vu}63 zo0n=HQDD3UEi|W2;i<&+ZXMa2utg`q#`0r+<`Ds&4Ft|41rzF9RY;3Y z5M&oqEs5D3KcE+k>y1nn{+!ElEK2PPXO^X!?URQin=bmpCvQg(iPL|v-ttlQ$Lb#8 zw6;>`I(2zP(H#>(##+RhFd{Rq@neA;Moqpu%DiMxu%F^uzc;0foa~X&e<%l#h^yDO5e;H=3K&)0t&m|exRuT zJ;FCexjOo=f>BfM2g=;q^6VE-!G&sD6?!c%)Ads~$!PgNZrw--4ckvmn#Kk@LUC04 z2`B>}0Mr)^;;U7bg)DK5FSlofP%5MIW+6VteJzVIg4?>qtk?a;r(^q04PXPvyy*xk z#bi#UjZ07j)?j6#%nTjKOie^c-vFnB($}42WIug1Fm0(++YpI$^3tm?G<#R0U9^xyQ=)K-Mf)1n(P8A{;XLI$Zr z$Gxd6X1Rk@Su9ParCXlbYeHWc$m;q~j{%w%CuKIws(AQjh$Yga;x|lWfyDv!oez<< zdTAz<@S>XE@6BK~(jen+w`f@CNPuw*;|>6Jq@)ME4*FZTtPWM`3(oLswIA|khSnm` zR4J&YSVQi{MLabup06~iPiY$-56$E*+hn_j?kpsz7C+3?#eb(?YQCzyKe0~b?iq4f z8{JY3t}G<{sE)9e;D`{Y@IWCfy7oUrb18&;&t%5woEkIfoJ^BaaH|}p0>Aimp95=E zWgkvcjAH=_AR+^{HrAM^E#kJSuJU0aCxnLYF=hr5c#4_3=ID;ML(!xFnzr@rj~$iV zGrEmul^b)oI%_a!@n$o1xLmiw(8j`kF=k=H3}JIwa;!z`0~;PhUG-`@yb5|J99JQj z;TBxj++CnC?W0kg^k zgG`AI5GfzTj`k7&=LFU{2TxO2%_yAiE?CA(dg&Q$)H zJUX_;Ji(}{@Q5Y`P^Yh)e#Vf zIicx(aao5eNBgiq=2E_{vuhavzZY;gm1H%U2j-RZ~atU`{Dc{ zoSik@-)(OEgt}zTx9ak|LwWsmLXJDN0VO+~7`k|_nVGBNn=&A**V)Y&1iVm7%tb&) zEO)4}4PzvGnNZB_ckrc!eqOYK>MPH_3Ziz|mwp8|CNd!*;p)ymjQirM&-nH<)t<#@ zDVT+$g5R%=q@s-2u?~agpVpI|)ohZb?8}!2s(N?Qa7!qqKw@3LLCbf8*g7YS|DLbwNZ6Z(3;UfpE0e0*Tv}SnmODXgD0mt}Yix$QOJ+BE zBJ+E0`Hcbb_ms@()u|7Q_!${rVaN$;+nh&As@Y$`AF_jWW!+EB!0wBrKDskciON}E zP0sDa)SF$VOjgTkKS)2Gq7bFiBaWD>D|@CIJOa^Lo2`m731VV9_50@SOF~t6d|4HU z)Jv|OHdXyg_`Q3oMyK`!jf5tR`8byi;I(fKU_kZO1rDc?$l!;%cWxyqt&;Zp1?}4J z`ku?@^xUag5Ztje1S1HVN&R7uBE+7+EqjNT~K&!riA)UEuPw4S}T0#r27H#I3^>bj5;wwHB8v zSqm1^3epiiGvGGHNPOIcw@ zb_tWDp^V?1|=#ucOGm!E$yv<;b$Q$Lcp>x%~ySl0% zHitbO{IT@DnDZRBQj#B1Dq!%1t@KqoXsmeM$3`iX#<-^e)Z&P6<>8SoKkUGc+D*p` zKD6^>4y&qK?(cb1>4Fm{V{Z|2-8@5eIdDb& zJ6tHjV3Kmo=nXAXN8XJMpwhztP#EFtxdV-z5$qvqJ_1Y6OZ94*%lVcYW6!=sZfG{x zak6s>Y5oji#y4PqZ52)OSgr(BEne*bS;%5@W^$p+Z%FG`pxLdbYXJ~R$O*M%kqF=q z%#NedTq#MB@P-?~y4m8+|3C4h35 zCm$QyL^h0++7yL$Q_V99NuHX(WqM++Bv$Bz4!HRB-K5(op_u5u+V$)?TYgi2grp~R z%}7=Ch=ZP$`bFhV=lh%%G7Dv~2L#<^ z8K@`oV7J!qSNO-0uFWxZ&W|R9fWK&--1r-lkghrKwNSmK{Zj0Qx!gnE``i_l7ZpD1 z)n#m+K(@-73gXT%@H)Dx$amk+G%5~UVbAMPfEe{4+vSrw^QKo2tGciLS_1GUASMVC zGe=F$P2W%C#H1!qn}bRb;DRx~mO|zY-{OH#Q4Z(Bpj(sE2Ojtmk2rb7$OzPXpz)6o zzQ%RghV23>X3+4mx<0vC%4@6;vGphG0mu_E>makw5bboI<>rs#XQgv!s zoH|5aa1R&rh#LX@=AeagFsOL&tV{vknY)n!z*)$y?^%hfcap)m`9{Lp2#LoHy?l%J*zX9HSCHjphm%~4h_q| zefo`=ZBsqX$AUI5{YPZvzgd~nMjekAISOTED7$6^wqY^awj`9%((=`4qti3HBj*BT zEN5pAx^EE2ea^yQ4o|uBkHRrKE-Pa0s*iY3*^301<$-Vr3VG}9m5alq^t&w6$=95X zOd3L3xk=2?JSpukC;tJJJiavjplgZg92MnfqJHM1qC`F0`o4Ye5bllXTu~Lmc#9#O z6bCz~iv@D>y}GpOC&j%6y%3&$V8K16Ch?<9SK{B_Ywv~M z`uORmYKieT*$E+@-IuYMfmVSbad~0f_egxy!q%gpd1$9tvT9V zSpxLn8JC{d%6H^PRk!p0Zj;o8XUx+KO|MfC>=YGtMf=w$! z-WAOyo(XXRTFF31xAs>UF6>;f$Ivf_Kh#{9SC$_5(VH?7UKBsX3-$~7zF9yHGHs3QyIzW+V3a++EZjPuz`j;-{f9)gM* zo@BuV-kA+~#fSP5K&QKnX|PgxShqzxLg?Lpv-$s@VfSbiZ}FjkQQ}$Jes|c!s>7md z%?yap@hmGyA-_G@Zb!0fYtUIB)p84p}~&ir+w_#E=a-P|kZsWI15GZwjR z*)2NjS2vMTvgD}1j#MV?FR6u`)4=a!~46s{5q~sYatAj>*`N0Zrzxl z+I732o#)5RLTN>5{3qdM(R0%`d283PSy6sLiopOCbllem)9zG-5JBeb?k>m<8-d(e zI8gbY>B;|QWe+$F)NJZEXn&XW0X@c9Us>T=rRAenKNX2l=B$Kmtf*IC_@izyt=1QS z`3A4!kc8TVewzfu_4S`ni;IczSH`y40f`t-I6K6?&u-`obTOJs3C*_wUzs3c{s>7i zTxQgB-Xwgc+7ncFmGkt00sfu&f=qL-)i8}R0071g3e!G_#^{GegmP?0JSb^}fvav; zUXJjpi07o1`bRo)k|HIuo1C*|+jXE6{P=TRLhO6#h~0DP z?``jm?qzyX=QcW3m`OG#A9Vis3s>PhSRG1bSFe)rcN;sCJETJ79+k(hLQdTEtP!95 zK^p(nFGa5YTnD`-bB8R8{`C%F!$c2`DGk#k2UZ-U<93p0gIoFzTRIGDG9~8baDK`Nwh06k7U~u7L8= zWD(|}Z+)W1ZWh++d(=o83~+rT@X9jptxG94Z8k2G zSW?BG6n@_o8f}=J2gv9c*=&IyzyIG%aQ-vw9(}FYix;dHs4BbOjxqyYOE^&xM_aL6 zmlG0687?E1245spVO78|(`980dB#B@XqzEQecDe842yeajoNM%dJml-sv7NmGuEOKN! z1f#^XP-;(&Z+0{(ksH{~sBp~1xxu;WZOM(M3GW70kg~wDGX0|+Ix5S3KYaXfxg2$3 zxk*cY3#iUnu4V%NY2QXmx5NA}guDsqV62F!!-swSLR9WWN`6^~yZnB_FdDk1<8}L& z&EPNJvK-fv5<~~B3*K_QI@4dtTQL5aAPA54FfktaH7iK{BQ&*cBDYFp>bnQG-!3)! zk#fmyMyTGgl%<%Tu4i*~F1C7!7O{Ii?b(m4ivBXkpCWx_>Pd=7w!}|`b3!c4X3IdB z#Y`=9el|;nJNQRv&oC3WoLK&~(@P+-4>*|jcoY0rHmc!5eE&-!huY-Ynq*Wz#1INl zXn3fA@B>Rn9i1I(04;&4RFPLd{%3~qzhzm2vL4X+D3z3UDK4N3GF3gFHbcv3zyIH| zFaE1c3+AzaCOcQkMHX2G>0qfTT!5h7kIGf!Mmv>MGu zM84y!8265Dt-qm$%Q&-)j17vU9(z!@v@bmnkeFc1?&VvO4<);pQRB$ftfIjYe)en) z-diAd*9edMJd_c?&5k8=(M$>U<=*2j{;PEV|D@jGq+TCgXDogGA~+++4%X3rR=%tv zuI*W34_YaiXex7iC`>+1kupF}iKh(xl1|8gthTMWhWL(2Au-E`c#`RnX{ATrDQ*R<>Qq7xM0*h`w@nO%@w%0BRNxR0!df}DqT0&$h6x3i(WIHQI(Aw;QscpjsrP~ z31(<7wzXprm*RS|>HHnzWb1i9xiAvJN1=Y0SA1JvEGEf>daM4|QB)qKVms{4*ttJK z-~8pBxJu&VelHEWlK%)TUpe+ymk-FEF|Y|o*3`tKBrT#LzKjRtQg~ea7p1oc@#V?h z;@aQotuXmmegN~U_L1tT+^g2JX(5%D7H9)iS8fwxG9|fW54DNN?7`X#++`Rktk&;h z<>m6xBIZf3HN?~-l}Swu52A)Y=&|(iQ}*w>_XOX9YNt>!XNAkhO*cF;fMdAe9CDqC zb?#qzr4@5EqnElY3KlRIcsi+)wFuvK`4R*K9w{B0V>Q*VY@#nwdSPiNP_u!?dG1<* z1$LeQjYx;_G5@}-xxhAodBv3av!TH& zH_Ho(2@Hm$xi`;oCZ(+9gEcOQA!-05A9Rqg-&~?UGVb+$%(UFerVS4|D@-E67|EGc zuwL4sDQgq=pj@V|9s=(3VR(opPPhB3DW7O%7cYXZ%g7x|XsPWz@w6V#}Ib($EEFPDL! zk`ua)8GM>wMBw5ou!&&n!E^YC3|@B9XLC&Enn|SSd}G7rUe^B4B+EVr)@By%OOQhU zN(bLMX~Nzp_jb@7HKXn40sr9{hrObIMid(w-c(`=^AuCEgv~cg1y2 z%-4PYdcPGV^zL%Z*HgBaj!qc=8|YvEEckaxBbHiy{gTIutrHN)y~M>v!N+ZVLV}1N zsSO{LeVL^Q6+EAMfyW-B0)IUJQ65hj{9rxX`(|k;C{K%Y)X^1S=mSXAY>|buzTvUT z%lb-nYZ~dEjIz!J_pvZ3H(HQqU~-_3MqBgz*=_5O_Hh3-VE#Vo#n)y$9zb_~=cDH( za=^}kw^v`}YfS8jsq`Ei3D^E8Q%2Cb*D;bp};ueS0CjcYkHlOD^LCn+F)z)w#HoB_f zN$);let6HS?aT&t{p6ZyEYS* zs$X#X{E-m*8GT)5`&aeGFOwo`7q@*nqSoUiJ_X_kG5QKI5}%y71DhfCXw|RbZBx%X z>Q8Nc6UY0sT_f0}Z1_FgZz*q8`?ns=U}9l$=uCNh#<~0&rKlt` z3wZtcdu^-}O-Jr2KGhm5aGiRd?C8>3LM%j?7cTDOe7faB6)REuF23ChmA%zRy8LkP zyi3{ZGYmaBULTZx_1EgPseoWaT@Dwr^%`g9_5ejr(VhxX;-oLO-?IE+TZvhy4^6i} zwH4D5p1FU=EgQCG8jl55QiC11w(ZW4i|y!*a)pdbIN%pV0FjC}&1*tl)G+>{AD_SP zqKi^tc{34jF(Y*G-~C!+U)tSk!uw9aPAO$X`UDqn`bIDemN)3E&8)ItDn$x5SZeb1 z=kEh+oN_KwMsdly%d5z}s)g>m66!zY^5s_2bbG`rz^>@r3VtT8|JKik#Og&L7e$;K zoYJN|MHJk=cD|d!-C2u`rBZqhWZC*{6_q(HGpm3O#3x+)4)gq$;7Pn^we<$va`xS! zOo`L2=VQLghoFqt9ShU!`dfy4GCvLD)g?F~3=B|KsWKHER@{&|YB%FrlIBeQyhL5H zr2LGsn(h=#^=K~dt!{iK-dy2HrB#w9u1(}+#*egW=@f;!-Bn(;ERgxM{P`HAvcAWr z&k00}Jjg*0e)o}&`)+oMYlPSizOH!+8uL|Y%c`u_o+o%sDJd*?GVL4Pt&?%5?|E4O z+sK8zd^vDfx9H)+n)eZ}8=SD9(}uH}zmsoW=3a4iL(XGyI;Hd?VZ-#=@%s4wi!zs%*# z4_8WtQOEU<8J!6iST6T5{3Ue)lPr2CprJ$Z2vf}lGJG~BG&1W(-BM|}8%zH(j~y1=T1K}_VWod7mg0fn>nrolw~87b zq*-91*km^Q-^7_~6yU1PqH@2{{D&ia@S*gof> z?(MJ)s1+ja-0eGea&p_BgCv4oTn?USdo`?!9Hyy;3OYCzxQA6AhwOPJDV<73RG~pR z^$kr0;lnCchKOKHtBt%A+qY`rNRwX4I%09bRz6R!y#%;pWZuqz$Fw) z$xBt2*a{6R=1PWTE6X_oU+0xRF%O$PjdxFx93aR}dRKHsCNZ8&_#`pCvyoOBI;AOM zpkyzDS4Ll}yAMdbd+9^3sdJex0R&pv0KjYY1DEu>XZTDFDrFf~6*251TY+@;l~@qj z(At*6PCmGzK3 zU+$3i!k#K~?%#c8cQX#yzE9u#X)T_X6sbCH%-t9!$s$Ql3(5IpQiAr~0#t-%qAnX< z!l_*BSc-u?lEh70Sm;poEI zX4_FKyH*|ibPqNw|HF`fSPh3oZz`=q_%WYM!;W53=LS<(+bALt41>d8fxGeXk3#gT z%nJXGC=zbx9~$3{(5#+n=r(GF;TX>c0N9}Fr9>8SRz8r0j{{EqbYt#XoU)|*l3le7 zQY}unpOrae<@?0R6U8D(e>Kj1$v!`p`tqCcu%y{MYJSro=Ti|utJoE0JZ+RXxIk@2 z%KiWx`YAa!1fj~@sq~50^HqDDnKAIDZ4>9x4Y-S@_2gNJCvdgt=l=&`?;Y0E((Vmo zS8Q9V(p0Jh0qKV7R(gjdMEaH@5PE152&h|BKw#55NDU+;NC~}2Z&CsYy-0@up(S+i z%ihm(&ikJ0Jm2@NKh~NxbI)8evu4dbcbVS=h3&s8GJ|zSn?<%88G-JCt+cOX%TgLL zhU$*2-B!3F^UKD`KQSi7Iilci^>&bbz*f9^UQa~|0s&FeMrz2V%M|uXM*f5xDlDl} zttx6((|y>!g}=xB?5L+~ROd`m!4T`@-J$WCihkcBjLQ=U2ny_hYAC$Pjz^=btS_mp z6kL{VgPB*c$rYRKBRU zu~tsS&Xiqk&;8f<&Ez3@7P>41jS(;ZFhTQa4p*NlZ`BJWj+iBp0sIJia*38|jSF8_)6c!!V^Eity7tt6Rd)h2JH3+2#`fY->Tk*uwF(*nK{8=|zBwez}=tAPPOf+@N3!2>VJDJ@H`O1xWBSgrD zr3;{r>-dekmq=ty`D8u`gKc&Gj3*MeDoO7jjf51z-tazTwL<%f#p_tY3+dK&wrq!P ztUg5uFuy7P{r}@vksLHCC8bn8t znqx8O3Rkqx5oSmEOSyKvSzgS9NKo^A)v~hnht{Ty`nZSYc?kxkUPpHhR`=P=UDl7U z2$acyNFwrWNEZ?Q(yk~d1z6EZ**k5Er78nGP@Tx%GVP*F5Tj7NQk8XNdr*kqn02(R zK%N9SDw}ydTD~07zi5GtAEgD&FI8wP6OY41MX&8?5V%gkcQ8se6*_#i6Z`koM@QCA z+rkn)Ksv7Vie8%}kO^y${sCAhxw!LZg`7hSvMr$^uKj7yd=%l@o+j;6kxAq43hVTi zPH>#YAF|1tftr+GtVH$llqh2@1)^u{Z&^QASbu0EJDA*YWfFeGx2sVE9Tv&VVqZ@> zx)ao8d&-pQK)J6fGsE|WuRyDtzih=~G{Z)~G@@#X}njA#FT4U{T(eFn$ zKBRKij_tEanw3oL-!(TwMw1sMilAa8#bZxTuWpv+MOHKf6;Cn^hSRiY4n@#wHt>Be z%U8kg?DO&(uG+5TAKW%Uv^iD?&6A#}Vr0I_ zK3smsPM0R{Z@RLehBTAmS zaqLBj(rRTOr<%xQbcJ$6L&mp+tl)(?_*sj-Q$gL%$D&!{wBHA}l8VOzwJ0d$zu{Ekzo`8UXB&lxD zYd*M$i=03QqB*m!wZS+*3UK7h8Fs9|a=O@fyZ1QpL`ZJ)MNlME{OQn~hxuy|pjD7K z@r$7ZBtDOoiMCf*-SHc%w=x)%8`P+rIUi5GcZ@kO$=Mh2onUUdA&S2Plwe7j@~ix# zD*j95NN*ID3!sr%U&~)6Vs2NDZ*<~3Bdp-BR(^!d=VcY_vSshuJkkktTZ*6q`21qH zVg6v*Q?!Aav6uXMYyG{0G}PzpE7kP)dE+N%g>sDBYkwK#`7rbOKGwRk%iD6{wxr-r zx?S@t{EWRq#vdy%)kr30Y89y58ut4bDjLy`{v@6~T1d{S)0-acI|}Gu;rhk!@n-1a ziP`RHGEc10n93(-zT9RYRmYO>(PWUQ@A6zfwe}lMwMX1#Umfndw$2%QU1054Qi&(0 zTxh^qA-l^-Z7F&|Aj`;W>@3^h1hCGFD4mWyHOrKB>ucqUXC0|~vu8U9Q+STNTvqdR zO<(?7k1Nj|+CEb|r7?D2Om2E4aTkU(uWc}JaW6#p9pQQC^XWj+)~Pk+gHeJ{N8wSV zmZNpaQRoY~Kf4YgMmF_ANBCP)zZirjcOv&sQ=(UVxW8uJBAuB$I<6&c zZ$CFW5kFK3mI*lP)pwZ|7mCxBzfQ+n!5IBbeUKL%-cMc5gJ(DxH6kkJayF>>y%-H>b)Jiag~Bhmi46^DgY&xDl>wU z^4kej$_6rIqHADWAKaePsC95U%7|wzh#MW55MnKwtlUH>EQ2o$?z$e^SqshKyl=aQ z=Y!?t^z2K)H@z}Nxv={6X<=xyn!v5m{BR>~S4Q6azN(JPL6_fepyuOQx!Kq|E*=7K2lFtD zqNEcmD;M1$s}zi@V5^CV)Y_0CSSYdjy4+vbqpcCTR zg|EOU{Zy~0sW23f|1+&1JR`?Vm_e-Ue9>URJF z(%nm2z7hKRZV;)APn;m}nt45nSXdoV%?QJWW;}8-lV?q2%Y+~-#tS^ zBDzMS`b$5fNVQPy_oF6{qW4(*UnB$W7ZBD`JULST<=V~#0$Zm!J~cSoQO4!Ej-_=W zUk%Av?@&{CzQu^*?iYF0t$19|RJpKrs;bbWi!TprKd1fiuy)qhB67Z8hSFgXZ$G$f z>5VqS#9Pq3=%ObSlgP7ss~RdF@hIKYxV*F4MMNY)+InUAz6TpyYAlbz5srt%TIiDt zc0=syzuDx+*Sub?RuIiad)G@&x+MSwDu<1l`qX78YPO_$&lQyNXJcW-P>$9pg!-bI zm(gVU<_o-2?If1DCcqL2;fmB1`5ONNV&KmBl_%NN{#N{Ctx0BjIf<@pMo2n}JTwK$ zJ0SEUAo2i*+GJLp-p2!u9jlBmXSbx3T^)h$h)Adox2@GoH_#sgQL^?_?J|c5sK(ze zNFUP47*vaXlbKO%b&_UkB<(sIY05Fp@5cxh;E#K}>2O&!QA*RaM(1I9_5MasgQ#O@zYCPW4z> zw#_K`GA#|M9FtO24Y4`ULjoyIyq3waJJ{^XZ zVr)9d1a#i!*UD|IG;pC$rYACj+ft4D1wdFR;aXC)M*E_3?A%)fPZ$C8l`8grsIqe~ zX4{J-VUeE`|Hhg{)C7-Spr|#c$&N3&lQ_BqT*n&#&TSDi_ymC8Pd=gyjC#Lf7{-sN6+0s+XASZJlrC$ts zZ9YTVtEM`>@3S~XOX}Pq8sc6?V^-|8g`DF_bGccQ+M2L?iY{f!vYMn0Sg449exPRz z{{6D`CF$~S!z48W@kL%coU`R7OpJXlkS{}%(8Utv!13xttb^%|n8Mrois{X~jnq3A z5fuYHc@;Uvyt?f_iUKvi80blmm|ND|14E?}&j*5gv)U_^o) zu@4ZZomUH13k4SWi^A0FDIK4(h-+@XQ)F)EqDIE@+E=)esaIhyF`QDy@^&t|&dGq~ zfmtwk`F%>Q6|T}g!p#rRO$x1(*xLo32(Nc%K6UdM05zmAb|-xe#&QaHnQbC?e_j`I zE<9=GZpm^5L$2M-&^&&bzS0v)-v#1dIbgC%tBj?MXq%eh%Bw3AeOh#6T<-7F z^sSI6e&Nrk7=E}Ke>8rzdBHjUX>5}x={B}u7V1{us&@%YEh{6%2w^NTSuPA!==l8v z*Xrcr$ep!Rk?OyXjE5NgOv(3YON2Xs)%9NO#bY(km2Xsl#{h^T@wo}xKt zvxSFr)0*XYb}*J2fo7_?=6Jt9B4#y@{XB94B{DL0ah1*~Ljoc=)^PoZpX&*J9v%lC z%XiO18src ztFM3jr*i*OT;C0@*sm9x(jIp3@!xv+`-i6sb}qnIH(uSaYvN1zt?##=4*wM$-E;NH zC$2Al0bgDIZ9Y0inyB6^ph?(yjY@Iu0GuWkYF(P{2B z-|c#@@%TYA*E6oIKd*Xrc5&W%`Qq-2drveMKS+Iedask~+dt7402x=9V}x7(FD>*H zezyR^6~=e1Tve>lYrPuh7oW}Dm{0hRUH5;Qcr?^nT7QR`*Z<-jWepEJ%d{C5KvM%l z6nxFis?F-2LMTEh?)s$R?rsD?3n1Vpc6S^Nb0Dhu%j_(N?R@+=;n=RkqxwFQ(51xG zhj-;cy)=fUyez!;Wyv8Y*eq3WMO#}&eNsjyBfg7t-?)D?KV-+QQqyr^7u)yHbZGp^ zXY?fZsH=0!5alsiM|r%J>e}y#y#w&kTp;ny^3el4+kle|6JJ< zvf!FIXh4EwqY-ioP1*q!N3vdaGLmfhpA;1(&@UE$Z}K&o=1x?KJprIYZW zV&A|t;cQb*&zu7AQAsP!y8bF;vs>J5^|iO}hJdvNafRw|0lJju&{DP!+NwJn`JhL= z3G+7a-`Qm=w%~LM%=evbjGm}YUpM{cjIJvvuZ14q##=tDb{L&CJI{{Q<|~TJ)u(Z~ z^-IIh0lytdPA(%#fWc^#sUoJ(*1B~AU&y?h53wd9?=5;ZkzGkyD#PLw(F0%W$;Mge zX@BWNM);~ab-%nLN$3|t?=6wVH<&fo-mKu^;BP&OMH`#e^PINMzKAU%AgJlU&3ZEY zjfufze^S0jME582-JVa3NWPtH!jNs>UOU?|x&w69(rB=FAI}7k$GES(+BVS)KMo7x z&aBA{dA(N#t{`nXMdbubCz_vR?a6DIjg?~uS*&pDRV6zhB+se8lE=0vtFid|!PtYK z;DGxf%lmaA0u^2>pRP=l^n?Vd6`rlA2(3)4n4|nozx>QFJifs9ZOvYXxuf@5)UDp* zgebB7lIzF5~^3@MGgqrRtyO2H|87`n`;Eb0I3P(_10fwrXN3T>^=RI7w6 zo_5!+ocO>_zKAx~bL2&X7CFCdi_cL{y&K5j(`Cyu<71xCl~tYa#`Vt3@*wWi{ri9C zp*c!Sg-A{yiouH%A*Ms!^+qD0@=7oo7Bn<}w%VJi0UsctGZcl7J7&ikJrAs`CqD_) zsz9Dkl+e7k9>eYnD}^UWZ&0UPZdnzS7c`b{h{x$SiJ#@6t&i`XD9BM%uJ<03f(ffX zKeA0G(R3Kgx)p5h^@&PYOCHE3Gt&x(P&;2nS5fM{xnfUBf*NhYqw0e0|6RAG(-kSK z;{r{;uHwrn_%Rww+X)C@?Q*PMZYux9uptczSS`5)(WLg${114829?;}tP0JJe_nNb zMlNz&UMcP*4!@l0Cp>o8%;-SLu5u2IjAP|VoTug5Bs+6|3u=de_e%ItIDMrk-zqtz zks&{3IJq7e&o$%3Tr*C0i}99d%)A?h0&uOI?ZxM$G_jk+jhxk6qO z=r&_KvJX?GS)`3pkM+O{1+Q_sNhs&C;r$i~_LXd*5Q56B7fZRNr%o?P-Cr!CO>Mp1 zO=_!FY#UkJrm=GaKYzQYpEv9)N-V|{y6Ebk*(c|VUhTV#Z^$sZ+86QFif3$HtvgYj zWX3d$e)J$filZ^Kv+$=E&EFyDt|J+~I~#%?Pi5=gxR}y_i)6%*wE4Y_G?V9|r+{yG z?SrmF;`b_vHRHRYLq?WDm=p~-taiZ0Po|J?=Xy?$pRs(7;h1-kcq`kHUGlnqn!)P6 z!;*hjPyEWqVJ;)>7Pq*tVZK*Tz|*+78^^fCR{GQKts0oI8g%)YoxLuDRQ&e6%M5+$ ze~0cpp$DLCy7w3`nRnv1pV(c7w*EtM&+v~Mi+__u$J0r`_PY1Z#oq%l(Br^&F97fD z|H|Pi3qth#d7E5u&D&Yq_8`X9@`Y)vm3zgM!`_3G$LOxXn zrY~NGjVd@S5*lk68&NU3vZfBd#GYjtCwk>e5+Q|4*>Ep1GpA#Au6cJ>y&Od0t>0HhTdqYkk;eOel6y|c`FRlJ-DvV? zQpQaf{P`0c90!3n0IuXe+%fxFgwk>~k9R7bTluEGoUh&DC2RZAF)7nrNx@CF0>@UG zIH*ovz1N#>KagF+_f07{d@U}_j}K3%5r&|{w?w(mEndJ;7d7m?F>cuvS}R0j3sAY% z-Fox*^V#60_vg8rn;#jd)SE@`9t5heo-e9U3n=ERuLN472{C#3Kt`G1C9S@?sp#P|5VM zfi5sC-AE@9=6BykIw3kDQSTfz;cLWNo8dm+e0=>_|1;rZ%Zyd;LRXoZ83x*0A|r8mhXpIr38`gHXgi2Wo}#E%dL}Da z#MPi0F=1rf5VIPDj!9921YV`7D=2@L=0J=l226>TI6utY4ZV|I-jE|TotJEi=dc+M zr$#QBR*j>>I$y3Y6dZi3*4MW0^Qyr;i-V<#WmXR4CZrBRwIf?f3quT8bnk$G*NE5s z0sw*YE)^s9z(=AdBcn8fQK;E-&d2-XZ#H-JSnDm)G1=fb)y;N)wp(i${;`OPXrPj? z!)vL@81aoRklpQPZP|4hV$PpsrpV*_K!4p~fY2w;Z#g?b6US83A%zylfZY%F#vZjQ z7hm*_S-~VXv(5}eOc`{w6jIyCF9(Y8cyRC=MUZ-Dbr+* zP|ntNU}v=fkA^y_5PuV`imcf8$c1~aqq%0r;UsyKAvq-phG(6|Eh|dw8Fd-)R-4T* z$QZTI;lv~&;d1C4h|vvTl7+B~iy*~l1FzH_Q>$wnz1{EoBQz^jDPzW$t)R zN}Ei1FO0wbxv(g}xaZm*hEluGgZx=ndEIuMXu?silOr=#y5@qi>0E(9Y|@!=%n4kL zOi}8~F+LYQ0|5bj5B|rG_#a+wy`<_3JwAHi`5)5$cNAT1^a_y-OZxAX{)IKY>trFj zJ2IC1k?Io6^$`DaP1vs5A$bjB7__=VM9Dt*^3y3681{7_G@A`P;h9?wHhJSiM>@yR zeAlOY>~Cks5jxlT@?xt4wmv%q`rwfXi-J?DvO0eZCE(R7?sB&XfJ!n3HH73G0QCXi z*H=a3d?I7YKyJmZjzjd7RP|q-q?o+5LBaVoaTU+O8rbPyShcJuS26G3{M7i?J znA2C3^-gWv&4P?z=c-a-f62{1X@Zk(A(wP$Y9T3RyDu1tqV@^*pXYrp2y2Ugf3Dk2 zV=)tB-%k(YAcp~Y4!AxS+mwvyF0x!#AFGgeH#5|)^(&l#_~b{}`Jm@H)TFWc7|w5> zDVCiOso(h}YlL6!DMgx8<8=6$yPxOV8%bjVF9AAGwX@t&U4;g&uDG|8R>ln zEZ`*=@qd5Of5%)lyk%5ku`c8mu!m!Ql5>88x=rpYh@j!eb0XnT) zm$vFugizit$m9@s;q*HC{+?Mo;H*_qt0s|;a=z8fjj?V=yYP_XURX#rTs(Ol==DwgH$Uk@-sc=gH@0q z!+munvjr109+b~Qjf!pf=2n;EI|l1ASJKIbnqv4*ax6j*1^f_ic$~W)PB;&)o+9n) zd8?UXDuCS0RTiFty8ce=xIyeH{Z{4^A3=rCq+{9(=%?m@jlU6DxKrcJCcTu@7O%VK zx#5wPymDnl@5)iUX@dC!_r7@h3^2)ua~h1vjE z0#N#z-kS8N0xTO+AUb=w0_XEMjESBpZP<=E!JM>Qxo-;Xf|)a2@h2-oIQctkg`m zw@a$CP+3=~HyEz{du09pJh(BSDmQS?tQ7lkXHck{`5USQ9o!%iYx%pr+Pd<_1#`k_*fk_W1gS4yOx(Rq+5Ox@OIE9#M?DU~=(X}`vN3dQ4J`~acgcy7{|qssc$D~y zPhK*DYRL%btfI4GuFd;hpO3Ecf;Jf*yiznQ-ZE$Un-imMQ{Gns(z_{dyW z5Uf_@D(nq0L23_=g z_Wd8Se{Mh-Ngdx^cNwwj!srT^GKY4M4`is=!QcYj`Nc}JigNw|eWzejHMxiAn+meo zm`>9^v{RRe*o1(bs!Uu(P%}4iU4cjN8o6LjbVwDdf6Qioca!(gGB2G)nk(C+~N+yysO#JAXzSz2a$$W+l z3kqwZ{2^wIeBTHpRh1jRmIw?4ul*xswBPu<<_~}p=w&`5U@`yszG;&@o~ViA*KllU zQO`QGdl9fh-6_GVxrRTJmla_>dVRURfe7heGY*HA`=O-u~UBf39<}X+|{~9ukpWiP^jW^vT=9 z_mI~Uy`KIi$a}e)tfGDzve73Z+f8z%{xXzPLhsMP(xDP4BRQ0lM``6N=zC-GLQZu= zY8l$KAwAF%06=#83haPL15#l*Mym(Vdo4>~L&crz`=>uf0<%>toSu>b(!rs}2AGp% zl$~M%|6mY)+&6wG{cPA%Y}JV?c$hhH`bx&R z6=B8HS869#OHR;Z<1G1;zLvmbjxWx+lf93^7TXnPRo?LUT*(pGv5GE&xGg#H@LOq; zKP>NOvK@*}^Fv&f>j#;YfEaUVFLOr$kN>4Zp;(@5dU3eHL3_{c66U>3jwnHHk9N-% zuWw)pguMKcSkTs^zbtR>nH&86yW!__)0h3zftCxT+7(Po-k_@f9=Igf_?`ay z-h$DHq4!qZdO{{6fdfA`!lqPp^ZzP_3nAnln4rYby0{6`Dz-c@ZqV;#JbmLTJKZxu z%P5b=t<2=i-IRO`rWlM~PU2n_m3w-qJOw80wHA1o`vf3i!b3-4R}8u1vx@Y~V}sr& znQ6x@E)52Sy&JhQ(Tl$0IjVI8aOg#=^e&vTUif*tG8G&wA-+B5df>Bj7wUO;1S9=$l+?b0Ez!%2n#!Dxwee`)96#V!3)9QTK=VfM4IcRN=`N)itf z&%z@6RM(gr)t((*NGUm7=8XO6@NVW+koJ`yTnF5H>+<2@W;LT3{P%EKI3?$o8{}e9J}$0!D6G%ETu<* z4Q*;Q+#<+9cM#UE_v&QSWV-2|gTM8qt_w!=9dm@eL?`*7tG4{q_rh8G$Dr&r|0R7feS;uG7s<$7(5g!ep8v-9)xArtyi1TWJ;a4L;vV16r0By|AJc?=a2rtatmY>fhXvO7lz6 zgH|O@y}0O&_na8hyI~ojC}AwY^ysWx+)#F0+-P7z*5T!tO^mDs1vG@pqy2PeXH%-v z9;`*ZjLDWsX0_2kaq#ZWZr&-?y5HQcwDyzO9iwzRTjnyZp~VG10psr>2*uliBQNh) zQ&xzo+ZTMW(#S+z^JEW9G4ch@lYg(qG5%?ckdRPk{nTpY$#0btXj2#aE&{-~i??+o z2reWokUHjBIkGmMbvs z?n?sgrh&E^+)Et~1PEDtWt}QT=Z_e6ANhLJS8L|d(dq+1ZHDHy?@O`M!PZM9IEz(C z7_GlnIG&#*K}LQ@Ypx^Qpo7K^?-Qo9>F!F_tg34pRxk~BUxl`Wp!N-yqO4%J0dsO@ z65FCurk0y)%G4iaWI$tcEs`%g&(&W1x^v28HJ^s3J?oRIK13Yej%kqCknv`UtAq}# zKnFJ}r~*HZ62!`Q5Y4eJF(D3NB6gwXAZbfmTlw9U^ILpUu|jEl?ypF$jX|0TyldF6z7uRZ2O%SNeE>H2Ly?g- z-Whq7}+#ywbgIh*aurd3jp11fy>W?+DvWB@zS50?mCBeyUg=Ll~9l7;x?UV#?=Jj5V znHac(DJJj+r)r(CDxPVbjMq`LW{h5*SxFELHpKCp+lA4=b@}!Bi>nfEuxh^;p2nVN zdlRVy0{Jcj)BkGwe?|12xA%W=_X_D-H-(k%ld;$W4e>y`W}?LOP+Bq4^uq@p)eY23 z&<)s+Ma_C^q~9JBLH!rD9~;=5!T&fEw}?+lA{Vu`Ziqd2^x$7b`npEUSAnE9v|9}Z zUr;$cRoMT?d&l~uoFKRGX&EarEdO~yU%P`Z_9v>&M=t=!p`%Z#hm$dZ#^Nh!Gb|OO z_NY0w*`YB*^5=RXs_-e2?G^MDl2t6-a7JrU3p0_*uPW zV7M|JlaP%zY16wAtJL$snFEb}X=Y&Wcipfv{N_h(fVaMofmz}8JBsl7XV`bhGDj*; z3QXmt^O_4@tCv_l=)ETfEdFsYiQ z@6kCbz0a3}a)sD##!vQiOKQ6cn{wScGs!Z)uF_DTg$pu5l<)>}xDDA14b;X6a5J-Z zZNq=NsZoWPG142KqCp0;Aixp;S9?&n$4u4r%lCz9)9J`TiKb5GB1V$fnEGNw04rxK z_TWm{UZKRBDYfDN&|uia^JT%nr*Q>>30i5TdVC9rQ-^8@`0$+f3;3wBY`}F<3UzU_ zAewGd>_5;Rph`VX&|}t74<#r_L&ZNr!1>?Vz0HrI`aeJ~AJ9P}Apstlx0o^3jqI>Q z;HpifXK}zpYo;%w%Q<_<=B?~Pg9FJ9rs4o$i|{kPxs91&MKKpiL(sM$s|~7oXCECu zdOm1sX5@=kd}6#id-3O?LFIUbj^brR?P{I_vk}hu<>w^*kT(qJxD5)^jvJ~dYV?QD zy4Y<2EEV%@W>bb^-Vj0p$%ooS2n8Vp_XQP*yARHPpRxQ=>lTDBsD~>3$hD1+1CLQ{ zqH5>zg6N~_K1>}$=CwssZ>Sz-}{+J zAJ%Txl4`6Pw4Ece{n|-Xm{aO_Gj&^|O!&Ppzk39Eq{M820&b4k5Ux&{(0#$r7I3N~ zygfJ ziHQ8U_7Ny_3YO=4x1)0z2TkC3i-2|uc;n?_*%Gn|I7qqwc&>rT?2w4Xf_GTPh`s;r zlK<~h24-s!jm6b2Vj+XEmPNzPA$(RxRTS2UP}9~$xS*nc{KaF|BD)Q4drZkpbTajk z*o_#qeYw9_pD9#g?2*~?0u9S*O57GAE&L$4kj{qr+=9Y?ltedBGEFw?7mxk)aI2_E zJG6NKM1iMEvNZ_Acy`8<2e+?(cB(~WcH-!*sv0pJ4%0b5A)qBxxmEqLO;uk{9}~s= zU2ibJ-r^U-9A~(IB+y@=In*=?QK?e!$yCN9r__)*FkTFaE5{pQ1mDG=B{r;kSDd)l zE#AGZb#V7_HI?3&YeT|zZJnH7j#MiRdZZ{MjIQ;#cqy0ZSY&G=eb(bA>p)!bSHCdz z+=z&|XU#lbm(~FQ1R9n~%kEQTqfI0ue=&%MmtwE zGShHS>f@V01rsWQg^=*>6{TU=f~x4fuPL%BZo*h;ld!B#^>$y3)8Iwm_Jpkz5<4kK zK5ME20b_1h?j+BlJm3^Q#f{n;uB&6)JRBo1$mSfrD7vJalyw7wQczT%E86x$B5l za_NAq7D6S~HG~Du?oq`)OEW8>YJy_9f5&|LBGaVE~^OCNIA7;3;l z4T)NxSKXR*V`i{eAFwj0MRHmf(E^WBmO1Q+CW}-^AW%?wH}>Rf#imZl!-?FV3ouu#eO`D0EY{=CM6d={abUDxV1O;)EMU8Y!EKkU?` zGeS4;H{5s5@hS5{4D_J)Y+0O&ws6k$u2z~fsDe}nnmCAW$p3M>76X{9HY+3{ONHBD zF!n?LU_FDsF03sx@16wK{bEqk_1&H4E>Aj($$WB!Ae4Eys-mTf)q76MupR;}FdU0Q zjA;eQ$W2hJGo8<3G3l%bv}WDeE{Qy3m1e3-0%LU9}VA$;;+_WaMY z->9i0BB;?%gw&86&sC>`kvWiK66iYddZ|(i;o2{T)s<4OSWoZgGM=cM{<@s+>@W72 zHN^E)?*Eue{ggxUUtMXimrmN0>b#z<)I`u(b2>P4a7*=|GP4g25#k*9KJ$D)O{Q1D z5D1u1pnebi7(r%SVn~os+FPE8GnNVx`?b7>9d(N*!mina)|q7W`EgqY%f3u!l~iRW;zpQw4kJtZjR))(6Mnqe@o&yIAvZOLidonk z??*1@8yKw*A2=dK%E~K6+Q4Ai6Fm`|3d6F&V3~FhszYh{6UWClGc8T6b;6UE8W)-C zE4jz2o_ClPitdr2OX;5{e>4sT`y2c-n_V_TTqeC>+jhyjlrp@Ok0Wer7Ue^hO*JuX zD?(uosBgQgRfMCk-8^yP$ONl*oDZ>Y&UUmORL`nF2xK^dPP;{~u2Au2S}ZI+2C1cR z4~$h^T3F`Arz;yO9wd&Kk1Z}C_UNK`KG++3f1FN?>yHV&H|-G0t@I<=^-2HMz3$-W z#ckchX8a73bvdHe5nDrH>rcDV_!4Nc z74#$FYM}ga!*9~cdlX_*JgX<88yGx>Qj6@GkGPD@(%nD0Dk z=S6veu5HN%hzI0EK@6Ry)C{{<_8MG-i2q`El-`o}ZGxhpkSqN|9zoDC z4u$ZFv*0y9Cv_Jqii7%Zbr;aJ>i_w&zeU1J8P504i#vJ~;GEzToTyRFMCtlyM zEx>x@scB1CjYb=BQT3Nlh~wK4(<5*^_^`hFeRhm>*<%>#`1{HQrS^a4zNep0$8>xK z+%tZx!zENjm>}v@bS$%WBf{2VW%Bbc1_%VxKQ-rTOoE%_#LPfL zbn_qhQ3N?kuzK7%B&@2{mF!|#im%@lLo$b}7C#HCkHq)R9Cdu?3hHY*qqNmSPZUmrp?o#5VR* zxqe#LN3u~H+ES^VZ)83k)l{+AAyOZ2u9 z7GWRh)=5udoOFf z^JIf#-kKG;;i^DEG~aunZraMQ3tp8?31-2>$?uL9|E9kBx`L$Mo^P?Nx}f^pw(_X`B;_N3V|eX`mL^E<)mH$E4Lo-;qQO z=f~tH8v%qDLoAZ@31O2bU9MriPR`QM_NE*A06aK%!#tt6b)mRNIgtA`1C$Tj_zp2CU*h(LNjC^Pz;vCBC=^KSFo-;=g!B7sk)(&MKjLwT7d- zQ-aol!Zi5`%n(i-dzzE@Uft{y{~c-hj4|%Vgkd9b?n}lZUxPn|L&wvv;q*o5cy`<7 zuQN+yOE;c!H>G*nCKx0GZOu<^YEoGG2SfXLnt#hH3=*X;vd5$w*?uv!uFu)upnIm# zUAsf;tPBwMRF>!~@HX{pbFK3c7R9F@FOwY3W70{Km z1pkhzj1?1`^zcZ#YdXxTsO>k0J_=@;(c{Fv(P5e7B)XA(ZgL{mp!um0H1ls25Fq^d zE6y!ZWVOC1R6q30?H9vV;j+6*Dl{cE7-U`B#_{f{JAU;S1MqJ(5X%_sLwimYDg+#Q zx+sH2gi#U03>V-2```bck8`~x_2#V+5Cdk3_T_A}poibf)jJUpwkA^9xa%JIe12Lb zXgp*xHmfYZ)CBm$bZiXYI}bx2b*(GJvpG3YVnKZiv*EY8QP|GZ;V}`M@er0@FET^^ zby2KlokWQuM6XvlF+4qadv+#`szvtBWA=4d8wQCOJAg@^)m^>N(5cm*8q4_MjznI* zu{qM?8nm=!&ERk?QWxS0l5~piZ3a@jNL+>M&lOj!#@V%(wrcU}y9;O(=#n$*f%%&` zUf_dxLs@lS$P-U|K*pdD8tYvi%DHM2%|&t3MLaT(Lb<+xM26VGd8uv?`4nx8Y9uVz znX3rAL=s8WvxuLVV((y&pnFJUAKVZYBFZNH#3bTFiIY3wk=p7dz50R{ph*96as693 zh6_x8cj7HweWyPscH5-QgL7*GG4wNufXCH50$*YBI5u$&O*&Zd{@K!CX+j$+E40ea z+X$~vnkFRG78L%XH3umZw}(=Ns!W29daLu|na9mJqEa)XqM>A%6Z)WV^;L|&W!K&I zZSl^bfxZcSior*7$B>r=)mKOCfnJoyIK*Lw%0)& zp;JK@C$^ZK%Cjr)js28{z3KN z4qdk15sM$K9eOqN=s)~7^8Xx#|GaUhy7Z%PsPxc84Jn?`xq^ZM#C^D*SbR;hr8sWA zx8A8sSXeMjaSLozPh5x7SG)Q!OiSF0K99z-=;BRS_DLL7cEJgzG)9o78<&1p6V;o3 zAI9OlSYNkrVt%Jk3WNg+J|loPa1&(lc@930`Or<>ipVvN_C4{jIo*X7y>Ezmp7eO1 zl-CNR0082IrEt-X4h+Q~v8_#7iqr+NPZ5u0=RIKgBCy!|JADS@mojBrVPSWVo$=F6 z?jiE>X07nEaCoEVW1@6I@7p2jwmI5KEQPXh#Sw3rVeS$##rnB$tr$LEuifad`kJVK z$#`D(yFUrnE zLjp}HI6>*>j&B!i(!`U!eJ4A|8L!%v4`7yE;x=q~3UDR>K`V4v^ph2j&U zZN!NHf1SuotQQB(;AwvFQ_zIuVJ97cnHu`=5gSnr>|g;h4h^Sa@u}T=?|vIB^p$Bn0$GBVvh9zyEj3 z>46LNzdO}HpC1-rNf3z=kL5O|Y7R`AJ!sNf2in!)Eii#~;21{t+7!bzJpQ~|7@6i6 zKK<4uFw1pBJlmT&uo)GS;wy{ZbMBgSTig`RQ`0X9i0WSUd0gRoN6-G7mY$Kd&%68D zOL%lSb&Ym>SnN5wTpJ4M>4j!)UU>U|RpcCC!qUNu;MYE4NHVx#43PD0T?b@~3+$oa zi1PyM*%_=;Q2I_AJ=qDHugyf13%yL-E(0`lymjrn`<8U(NaX5Art^VP%EHVc>u`M6 z&%!SK4S``?%6r(OeDFtzhDcCPY)))bK&iK7BG57(Vny3Dl=y!b`wpR%Ce=7Gp)jWV_rM^#m`_0oG@lSM^HqwP=77<>|+*F;|C3j(=WKEQN(vo>Pig|KnR z&T?uZP*PUv4#NbetpA665*u?OKpjW8#JlpOo8_?3&-uqi2>i*vjNE_UR2*V=!}9$5 zYyUVXM%#LgX>n<&yEsZ{v(QLf_%@q^U}#iAY9l9d;BKC*pT5v@eGTj3W{?l^(j{x};6z}5EH=3{slbDg z{je^+vfipvg+T3Tp#ce0&Y!2+R-OJNwO~r|DGpvlCnd_mrlNP+Y7TAr{Fe#M=bkod zxM%vShN$1kOJm%!;O%(Twf_MsI?(vz&ZURPRpUvXwV#}zSlGPlKvCY^S)1z#*+7-r z9Qd|MuEhuMCHbI}2J6RkTcvfhJKbFTk)|qeL7v;$Su=--x@#C&=s$9*v(!rT5saJT z&_0M|XLu+YCMJhmQ_JA;6ed4fS0k}yOCPa5@tuj@RNP&zu3AAA#o~1?i8(|EPWA-%PoqzHZgzJ(cdm6&cPg@%Zj|5 zOd{mCa1*%95&VidKuK?m<1Y0tHlx2Ybv0DrRN@k3dN4}xEQ?LZ?2o^{%(*)kug*mN zBtqC(@C~BJZkKEXjQeD)B&^{1FMrM=Cc}^9bHw76eyo(-M0#B77aaJ-N5QJ{^7saXgVl<(E2a@hHQ z^d5H_$eFA|l{EPV;*vCNgSS`#%IO!ZKo|68n+^y-xm#z7|5O|Ao|hM~HfPtSXm)gH z*^p8hDy#lzvF13H`FrDB*>l!K;Jzj`am^A(sDTc)rJ9?FUZ>u`e?HJAEIN4wrY5bg z6*Fb70q@M4Us4W_e0MZat&%)!@p)XU!fUl>2g%~$E_;M75awP>I+@P~#AMSx&2oA! zY^P#A#D1Dg;;DaF;pM$tTx-ucv7AS@($*Ol{`ULb=J4U-+3iLRaJ8-K>tEe0O!=(Y z*c9_k1%U63NS@7qWV-3v7u#Ps)LL{)$sBjKhN)SWDMP;+68oixm7OlEpoh%Of-=xO z5Fc;b3fkwc8`P!JzWX+V@f)u3qV^>VWfpMforTb`#GdSFx9230R1T@4rzoV$oa@^_ z`z)}dQWs-HW9YG+@_Cw4Rokd2Ql<8;GL#1V&Y1)$l~(4^uGFX-O?_PT+!a_EsMmdJ zMsBqolcrbEmo1!hx@L3X3$!wUH}Q@BS_3`#D>G5`4?xdS z%2^-=i)#|@F=|Ydd-iohxi5!o;VG?f;Za}@d;IH&foJ;V{KR0F+s2y0vebbFxgY?c z>>yAhCu5~Hab2%hWX+0&r`*eX%6b1OI$+z}tkAt}j%5|eP%-tAwe)02QyI``bwc&< z;u#r^Sl;l{k?bS$ndN=R@5wZP$2h@mbJb04Kn2-t*}HLF=Ee!{DiErYbVe*bsNU;G z-+fhMZ$*^ z9vg2>2fG|iHF4-Jz*(NI)tutu9kEx6+2t&Mu1skEPTx?;)OL9|?n6_#!mC|eRT~#> z%Zx;_kej7fO@NV}@i4pDgyRT*j_7wV^D-$$pOILq%PdkUxS|wNr7))Fnm{*)o08=+Bc@J+5@yUnlQvyLgR*uTeuJi^rG=_So1&zIjiuijv4 zY>un5V#3W1ElF@+cvkJyEi)LYriwwTl6Mspe>2obd)=r!A5-f%(by8?HuN{5 z#B4Zu-A=P%+<6kXCg`3KOJ48HQ|}R}I=MsTY1|%}8INIl<^J=oj}o?dQusWuRe#Su z49jUIkyD0CmGa4y+!~;j*@&=YALQxq2rb~N3$xf1ZwP0lMkT+VXu-&}6>Nhz5gEd; zk!J0zI67jnNr$%u z9A7>q5xLtZ?Rl$pw?hNe6%^}$#I9?S!>D@}g7V|$!|omtxfXv`kB5I)67F)Ys^~@w zJR~*UlMR*gv>SifC*+vrf)GdZ=>EWi9fw8^(N40>B~?8z-!nJ#6U&Z)%S3Z7Oz3U< z3@aAeB%VAGx(^}?pE(WLSQP>Jj&@{OqSoK#i~nN#5OZW)IhYtHSyx@Nn(?vF{CWXv zmS*C1pDr}-NFRn(St9V%o5HGc1cY*k@O826x)V6G6iUhAbS;{l8pJ{`tf; zpQQc}!EC3e?cCZLslp1ziy9;nQ1!S9^@i3>IJynf-^ogvmX_sF<_#DG*1?qfnNua% zpAv-$_1l)8y>3VL%`;m7bHah&4ELG}+&KDjD);Z_HtDl!J^2&7xP@8xkBnpsZJ)it)xe=`R*XWl^9 zQm$cZ>mU%4s5qh$6{&D~OEQTx{n3Ho$-X%3Uu_x!a8&g}>HP&XIm((0codMhV- zDyp@lBRrAb+l?Fs=C`70msZw;v& zu~tRH;1eUFW`wJ53gOM(JoRj`FJs#QQs|=`Z018kr77DkJS~H1BP2tvb>08 z&1O-_Et4x=t3|lEvRCoQOCA{@O27!i+c%isPQE|^SyGfDHIfmqtrJk=MJTjOx{(yr z2s9d2hBZvj#^nd)31;_qHkup$#g@T7`=&=-btP_keP=?#p!X+}8@{ce#yH2IHTE%K zGwvBw{&gx3c@Ya5>uw&t{XFGT4L~H#a0<}E8}*T zG)mWt=PnbZul9_i37Q92`IQQkzoKTyz#+ia3-Df_2C7iYTPRn-k+13aisSnlXNuF# z4+va@Xn)|)zxWT7xz|Bq5x%sMkLsp~qBrx)-BfM-_CO z=grleTc_}{zj`q&&p+0-p0WL;TB?UCE1fbp46JA?fKb7)uq4xy%VKezCcHucFVh6l z7EJd=B#W|EUO5qRNJyX|d`~GpYx2|9=cQ6V`hGCI(%jBMY!^_@Jw<_Nz}E6sV&omJx{Xg7rMI;xQcpsvRTgnv#)Gk#oY>x! z7cL#KlHSO{4CM69$r5TN0Y*=n$-I4qAr=v;bEPpQ*T{a!b@wcsSSqT@)w02Z!Ozo` zdd4Vx1_1e@gv5gQy&(<#Ri6nB@BFeZ{FsS}i57}`5`UzrrQdAB$$iP+%2Rt8Jx@;0 zoNiGu5oyxMjT)3xUZU~}wI!gLx*GoDQH(mto8=9%8PohyFu71ZQQP*_r9za$M$7ZE z9oUd}iK?Fv39c+_<)Jedfck+^NX&B!cZ1^UVmBPN>oc7kH%t3H>}!TG&htGW={evxmytm$c|GvPQHBo9tD8f?6+0no@{n;7k6o)iD_-ua3|dH~ z+|Tis^QI;Gbr482)$rk_Vrg$Hphi9Fr2mI+(^Y>Z)3E1)gC}w8*m9-A2=O2qBz?m` zVY}hhg>CbvF~j%F!}V-#fgWAui|YLFBEitYLZn_K*a{a&VMQ3qm>J-D>Y|fpJ!pbN zwj87nX^SOTPqG$Bdgm4cKM-r7y^rk9eJgpsdLzYfieLG29T*HM#mAnG@G*Ru1IZIY z=kInB)>D*gSA(Ldk8HDyGgO;#_ro7j6#h{A<39$0O`!WaTht0+1%M)gw5rsT^c5V} zw#XWibw~fwiPb!JC;Ruy3B^5?jGL_nPjdrRIVO?l_lRO%k)5dSprsC04sIVTYNz_X zkdE4in$Ych8SY=((2LglEus;H{)v+V+zbyy=!N>hbu$B>B9rL+>W=i`dWk((ul>nc z=E|kkkVBQzx={BX+xg-e5Q0bLC|u865U#7c-EsNhCxHGKFY+6FrmR!^9=rZ>PAaRs z#>uE_!s#(XX@N4JU?-b?chITCt1d$$rfWjfS$?i?D@Bu02Qi&td;jdo`h+L;12H4x zLqeA*98;KfWmDr{2jO3S>NUeAh2gozC{0gXd-9zh-`rr-b zdI#=n#d5cU5S(nH|6_1k!%B>b1Pvw+b`yESPuU%M(&wM5|2t#jKYxvXK022kK4%)- zB|9Vl1-K~^bhB4+jhSAi0T_&1VA?f*RHA%$#oC?x96+d)T7@zfPiI@4_4e3~uEc?l zxQ5*0-FnhNX2x)Tt6VcmNa5p&*A&0%Uu>_;0r-_ofJ_XVZLsFx!sZs9$-9D!(A$IU zF$~8!qQ9_lVE<&*RjX$y{-FlFB+0?RXdi}O z70w>-{kgar)nHkk16{ciOO>Z3+k zC4Ob}H`vyqFa-K%@_6XY?AjWeO`5cS3CcnwC3)6)F@yh4q@Ri0;nGa=_!6^Lk*}09 zyk*w-m1Cal_Tu5Y(B(-piK6!j8*R88X!LCu^i^_T_!d)h&58O@os%oN?{X0Zc!ofqh#9M1ymN_Guw-&(rs#Jkz4tn0n zgOQgfAfVda(LB0#Nq&3Q7&=RLg*e^+&1WqmoX|X7Gd1_I75En$kQlEvg{LR^5BK+E z4-Bj-T86oY#{EdML6~3NeVrmb9D669UAH`2-9v2#WVum1&b;D-sehn{_sLLO&Q_|d z{IFGZ4Q0&)4lL!RK!w}IaD|9t=ay8esGt?D#V3#y|{A`1aPx}mry%S z{l^X0%?(lMYb7rR=aSBDWT~wX?UnX#M)-Gs1rsymkCpXWDxyj({kNat)pg|flAl9( zBhpH&)cM=B9%n2u0JfxicpUm7EAdJS0*2D;%av3@7)Pb!{zi5jHE|d$qU-u0?NLYL zoxFF&RlfH)g?25+YO>!X9k2cIj1ai7eQnlsSBziQ_+&-ub}(j0usd%anS0GN6tkF` z%eW$3zgYXMNI}$Z=~yD^a~{(LI_OX3R5V@4={QKDU<1uuBJ@9B(uuPmU?E6v6t6l? zq-QV(@#09m0SwU{nIL#&AVMhWl`4ya=E&$ra<+4CvJVv(%GBaf?j-Qm%2V6AF3~}& z75 zcZ7g*&vepPV?~TyPT&(=RY#W9PTF??=edlNk7}pch`>z zL831_jRpekpX`hk`s$%Uc3Z15J7exk*jY61!EKw+jOe$a+n#~nUcaY*Pkux^@s^|a zRG9~Ou2-q;lYY^uxc^<;&fBRA@9sNsmQh~~-)GTUU*|T{U+4VAwpcznd*+x$=Qlr@ zZ=MB#i_!w;T zXE;X0<9yGobfTSz6-|Bm-K!t5LGz2zt>mW-T^51Z;#dibftk@b<<47}#tG`8 zF$&N3wuw@O;Ghj44|DAhPo#yG_%dRfG?j&1@HQwUA#*4? zT=K)n>kS=^vN$G!o{Y@6MC<C`}MRt^^068(c^ z?}?+;>cnOvq9hT(ZB+yum7Wta12Imqhlxg5p*Ynl(j6tGBQ3ZahwOZoZ_;g?i;1z7 z)EEySF#Bg2pep!8n^I0B6@MQe~#|b+O=9wW$J}X+!J`S;{Yvpepg<&W<2aHZ= zuLwB2o-7_nx0^;g{5m}Dg~c^IU?|)z+2F?!nv&yR7`@+-vULp2Is{!qOWopEHQ`dE z3M?y}VbymWu?<$lg;I1jZ3m;!bJP(8KQ)`+1*R6KSQ(6q7{5U}n)~19R;`b|k_3L& zovK<5FK3uPA4J3Z%F|@vq6|FJ8mwg70J`7ndsIR(K^Uf)$tzKd^mt%)UJfRap*!w6RVGv2GS4R6G&Q)30}s=~%vl zWHkK9DIco!jxy`()QXfTay9OP=MBIcQ!K(4C6koT*Nn;Y%2(-WsqCj%q17jl5q1*^ zc>3=ruq({*$M#H$qxXt5pg5F!TACj1cxmZ^{M><%nP$rn*Z1T;+CDrwez{RT?I5rJ zL^uvxquMHCQy6KLJY%vV*8iDEb}m2v##ZAwG=8cLaAuY5-KX;c-LF1_?ndKwc4Cao zQw~^Q#htVh<4AJ@;hSg;Hnw;FO+55(PhEWM`sX={K@i0vCS`;Uiuw9wA-ebxDtKzy zaY}GC$-&m+esX^Q84RcUkG!2vfBehma3>X3+DLESa|qT3KO2*^IBwU4HRV6V5xKbu z1ia8t$QA5|`c|K6OCtzG+J2#;i9?U&#RC@++104}1Tp|@lfPIQk_<$4pvvJh5r0;y z3fT@%db>5!JB5{f!NN&R$eR}9UOGhXPu*D$%b|c>2n4c<(m_E$bl~xdcw{9NzDVsR z5H~@yY~6A^fBy8~J=tMg9|kiy%W*?u^2e+Q%t}>i8SsD*$~F0|Vq6Zqclx?$IaNrO zeuT$*+*cs1ERB33(wkbUOb_|Z)A+Jz8ZGt9*xC0PIH3J|>oTsg2n}eIQB*==f9K^3 zGV?7v_F3}zeLL{FB@Th$rpf&aT0>vAp0!_UI(Bmz16xy<`=19oVZoaW5c- zJ61Po!--ecTMt2EZPlx9U#VCL$wqB7_sVV^BzM@a&6Jw+I?k)`pNYb&hh%L_dBI8c z8VE1WHQ14y2B&H%aOh6%u%ws4GmqDq;U6n(rRAY~kH=?%s_s+xFUPa^+Mf;QzXo4( znfuUSM$w+B{EDBfv=+FM9<(hu705CjhTq~rZnzk zS*X0pr*>8Au5Zw;3iU#99v_ViQ_XK|(?K<1^U18ykL2)CQU+O?KzX=yE(b8Qxzp|q ziRocng7juxq9!d~s#hc&7TP8d8dI(@7wC2#Wx>2Hym2Aq8ze}dV)nz(qyyk~Q{VF! zIubt{vetcZ(}rFrXXYhKK2~+Z#`Rs6Do?{jSdlh`Q$05+yN2N}l>fBGaUk&l$m%oG zK#tP_dO7j1=mGse8KJ>EQ z#hM8(X=Q~F2S0g;TQ%I|!|eFimGa}O=vD`|%ED1KfFrr9*!+QFQ=OtZCmTf=?dts0 zvYCOMb9n+HyvRiqVK{Z6rDNU^=pXx|>Y?k;9cwF^`b5fuu)^oG!=JAwk-Tl=d64F1 zG++XCdG-Na8#kQ$d|#mwDk>T{aAWiQrsQ~etyi&^??{A$3DIYNOJTbYKQL;#IC~-F zD*j(*g#R-+ToMBsaHqMVYEFOa z#9yFsY5~LdSv117%73%@FnvB&LpRa0;|0rEHK~fGSeRSe0=Q->RoD+)zLsadX%g^; z#MsQAuxlhRGC)K6%i@R`5)g;qOINg9B7FLW{wiEv8hO^SCPv@pgVC z#eTRoRF8pTU?=CWoclbk@kQnm83&1N!;-1ozBt_QfoGX4;_Hgw ztp~dXUEK$Jv)g*g&9C?)66~eQ_=jz=?}1(M=Rn(olHw4^wnjBw`R=MZwo=2+w`Psd z(XSPW>gt8yhb9cl1reI44f1d|YRx6{MRk6Zl<5z|rl?GvX{SZ~ zqs*JQPFI+qi{)fnB&Fbcmnf?{E0M~|Kd@Mp8rxS%ZzzUy@DIca+i_3qZiridOtDzj z>Y753i&#p@0!H)3NMq+Gb4B-$`WpI1Ij!U}vRO;6pNtZ?GQ`Vr1wAmjmp3&4x}Z-m zKLjWz8+m2S4YV6?$B-=9s6Z^B4VqSn6cJ5hJ16q(pZ^bzA1_Q-5K5a6xMUZwH)HN6 zFvO-zIzf=a%e(shZ8q58w9(8ltIoii3W9cb8h*}4c_Yd|my9t$jd(HLPHxnw;~F06 z)vh(VkO%739ZG< zC)lvkmjT`b#7c0KYl_%ou%zR)=!7GTM@nB$&>-1XN672|w#1D6Fj(;$3;+N`nAur+ zgjYw9Wo^(Q$ssH(VLg-mmI`(#2&@?3Q4{abRuEnhJEM#<@;ZH`F>6R*f)B9J%!W(0 z*t^;B)`0OTssKL|WgGDq$*2Tos9P&gf#edwc6 z-!$NgulfCRk!#nI_VV&WVpsR{Rh+oO2mkHm(=Yo~507nRwcc%!!rk<{ScfKglc6hOjcZ^jcEY;clH2S`LdZ*P3yC#E?K@Fen<8*<3r>+k-RcsTM z#iy3Vect+b&g)mi4VB{IZYKCSOf zK6hS_sh+gZsrDr)|5h%|u2wTq5c_x#@EFCrDb0oUfC$0FncpoNHKLHTki@gerQjqN`;84JBVtI;i0)d0lJJ(+b%%j!5Fj1pt9 zs2F7!Uj9AdZ0h&PTW2CpyFcFLe%JLvr25i;!W}dgf2L8KRS7Go?GSB^SX%mY5R0~< z3teUDLviRqj&JWe2QHcU@f%fB%VNiTy??{!oG-!W3nuJCte=np3`_!(Z z#bwubO{&D3Ec!47#4eWnJj|YMAPLW1In$G6pf8v4YH%NR1QL9*!Fv5^%v^~H26%7rkl@C6>@oOwh z)fA_6n2#`a@2cs?TPfe{Nwght$zf^;lZJcbRI+w^Yd&*l?mcexN%P)nTc;1h)g)AqU)!KFH3MS+!j$y7nO#yvGq0tKXRzy9jR~G9|@lmC~!@i#DY=SIOx!@yPun-mX;IR1JE9 z-edwoFd_oI76%V5d0hj5$fEEMiOt8$Fj4bZ)Za)x9@`rfyTKrOsN%1ZVqM3p9&{iFZGpJ8@ znpO~}Zha=c8M;MZ_<|x_=q_vq)_FDGDUag;_7Q$r}mLuMB=o@#dP)`pmErriB-R+kJ zTJ-|`^E#19{i>?p2qjjO(CUzaAL(UAxdvV=#uj~`-j|V3`60g^^vMjkynwPxhbiAn zj-RMp&iPr^kLpOwH}GHARI_+Hn;HzP(*qyp#s?JFwA>jQsGD{yi%Ix>3cZFOB#(@4 z=J+AI!bqv(Al=4E-vlBk7+#E_v3{CePI(0UO0?$A@% zZ~1S=L}6uW6aZ6EcPa}kcJkV6Voj+jm=^@N;{m*-hE0n{0_#HVTkyjdFLp>xQA0i2 zH9H+fHNg;nY`1DIlo$!br{A+0vY8UVge*-CQs{7HFO&;h&Iwuaa>US16798+M9r6Z zrY%;IN%sR0%EN#9NbfWO=NEbIAuw*x>oRsc`@GeM7Z19ZFhyXbgakK1lJFX2qTo4u z==+r65)EL;8WY8fJZ!%1DK{5?p1cfo%Qsc;Wk;{6I}tuhw@ICORmdl-e$h^1e!)4XMy8}a1 zsgh|Up@MVTd`#PXXmbjcMyn+S+t-|e{oOEM4^>QF1(gg3$d@wV7?FrH3FgkI@0P}< zOU)3B*skUKYPUC2&y-{upeVMmc>}L@v`}%6uAFNc_}bgd#yfj{`hksjeR0WnTluVY zUc#_ES3uA*>d^-i+caye@i58MEW7bUUE9x-Rl&t=;4v{7Y8GW%;z#GT>{;qEkmlT? z+*^4TqiG{pvI?I_|xw0U51%f<1; zC?X`hD(yw#co(8T_MWVR?-Y8LD!4yH zP)k$P>}i2fC67Wi;s!X9Yu5yXovCUHGgDB91b?$xgfzMlatfe9^t!J}x(L{xyTtLn zn*wbFWU;}4S@(YrYH#Q7+BMyq4aUDc?eCvo!jYAz-hZ(@Pm(e%vk%XSq~zQbjlRM4 ziO_se$nHpQpIx5oG@%tLe??4I_z~T>0je6GdV_j?^X~2sfiVz7&Lbs8-8}w~1Qg7+ z`!@Ftp|-R)@RjZSAODP${@=Tc7K(Xk*)N)%Q{$vbN+W!fy#eaY17&pzCnEB9vQitD zXYWrWT*MQa29kH(-O=QcU%#pcL3%qGj>;&abUJ`P<6R6(@VR531j8hUonZSaIYdjH zwJ|Xx)DZoCLU0G?TmL(==DoTi-Tzd=zi9EyNQtkO#*kR$?F3FN6sM$$PTt}=fV)t7 zmrC^CVpz`DOG#@xamnz5*i3Vpos(X;ag7e?i+!u8U}CT4P(i?A@Cqxj&bsi2tMlX= z*{1&U>voUMe{5U6Ni#71!6zXOBt)qUVsiGw-|yDH z`BwK>WXpKJ#f(ciiue0GOYAHbyXiqsSK=%pTT;th)Y1U)04ZF?F140fGrdFW@j9wy z4iwifopiGc5SSE(tG+~-ONm|}^|QZHeh^QNVk(+Cy&3HL=_{m6+e%vKb;JGtW!rV4wPi6pDZkg+)U107S9%=&C2XhCV33904V_uBXE6TA3ts&DF4 zD{as+5YoXiDK&HYi{!T{OH=d&$~>{>j!!8=Zqu8DT#rmj=Lt$kr4CX#p*IFdzDe=A0QJ2lM23sw=W>UVO65B>FGZv9X z*qr(S2WG?Q21q62N-0r^iyCzuI}3pkjsh2gGyRp%`DwrXXuzu>=Kx*?RqCIWF8*n` zU;4$VG`-Qvy>?Z;X$;bDSGodNaSclAGPsA{Um^ba5DT%ng^o_mka!n2a;+&9*(JK* z;^`VM2A>g2d!2);Fq5vnSfbx#bH|RGTe@8=$q!T>&eLVx*o*U>+sC}6kzK*xbmu!xq}M@$_bsf&C^{qJ2|3dyaR1z1fbsr}-csX;@PeipNypR8e*h(zCe> z*D&CeUH(;R8w{PSHnacYEgf}zzWbS2sXbaP-8UP|9Lv3pskWOX9kCx2tR}dDS za0x(QLIF=$bH#PHt+DvhxpP12cVCjz(YlA zbTW+mjPko`1FA?4aSJ-)HC|-g8w%@C{#U{lb9T8)DGGs#)OM&CXgidFh`*a%DC1 zdqt^|&A9?UhI$ZIANDd!Bz<~rv17_>M-i{}A;s<;6GhGxz!__1P6cZ7%~nWp<^#SYBc+pYu5_G}9RV&YgZg7uVNp&!Mjg z+$q<9jS=MDHrv@bJe*Qn_7VbYJtX9zr;`H+U-#3m(LoB^^$4EXNKTUhBAu68txKEg z6?K1)+jczgKHtI?g*Pm}z)LtJzeYtY4gl;5Y#S3~&caxpipYSizt{@hv%O%?S~Hka zLLgUHvm_C?G5z)HVLjc&4YlP=5DA%0TMxZ-&AvH*y11v_It~4%<@Z5(2h$y& zM`64J^5RY%@=8+ig+Xu))!J9REaHNczu)@QoDn7{N?dE>Ioekm#j2o}hpcV`E)gi* z1OfqPa@A3~zc7Nw&d(3<B^MwMR))4>$R{+A=)~k3HN`d3Q9fQ;*YDls? zP}@oE(AQPv8FB!Dlpinh@{Z*Z>gt^8o{MgX0U6->z*)ghlFA_g$=R@ZQoRI4+G~CV zlMPijB3;ZP-?JPq4Q0o38K^2JC$pze!V5u@QwG5t_yc0%`#GbhPR0UuE1&zA59}-M zwd644(R>xW7C_@gz*Fd*1Vg3XjGQ@_Cg_(#@*dB!m?3SRKt#GHWC#EN&ao=QqLk0W zWK9;LIve_fn?HW~D027h(Ruh$ns!8yRH+gVL@C$=Ovp!hybKiac)kBUDP(%oOZmdA zu?C|+C8>HNMQp=B()5&FAJ(f@SR^GtRGK|_832KnKml0Akh1DPyHz(#exqW`zV{F# zZs>uvaLK-~G6fndDbLft;mu9>ff=sW3!`S{wm}oe-aES6qT5s%m0f%81VYsyc~Dg1 z+iR+Q;!5uHhdtH@RcA6zy8M8!kfdT=tKW$e6d>%ChqAMX3wkXur$M(<`>WpS*H4Xf zK5?=M4{fNXEVfu}nQ%Q1jlS{?|IJIYXL!JKNbcf_Ck4cc5Y+dTZoiL7%H zpb;%&1`MlGh}M%2)wPZB3ieGn?o6Y;PRt=?FZYk7rq#>O<|R%LUgwTPg`&3+KZ4zu zO8ZUX`GrF~DG~2V?N%kJS2PXbxa1Q_f9Q_JNpp(xc(`)GPkrMO+kPC0Q_UsBb>#f? zb01$#O@}BCr%IMpC$H9iQ>oVC554HkTN~4|UuZ@w#Kw(<*zk&LR&Qv3p#g!XZ>MKP z&N>O@1|%fXbaiH3k~idVWmT3piex9=k;)TGD2V4=8Q%3C(gq21oqx1`erM=&a;!(& z-U-W(;;dcXs8$UPb4>FK&>Cc4*EofI*sq1ffgsW;Qus9X!@zE0Q@4=!Sx&iSt6oIa zqs6GAs|old@GMe)2Lj##o6f)NRU&hRq!}*@&mIEGdx1e@6gbb(8>Dn+1GqLt>TkZ_ zo{^pC-)mG{%i7B{0eke(g+n(ibYj$Y6N+6b-$Irq_WCqZIRc7_n}^+q92QM;xv`J0 z;}{@F+t*TL8EHkJ`s@k&5Kw`?W<~qGksE1p5UF`v35lv+bCC&bRj(Q;p|_BfW?%qE zG9yZyJIY_|fY%-C9QQR;D0!3B4Wf=DC+&F9l~)^GzrvXaCeTQ9P&w)5yz{k^7-V1+td3U zaUrL{#D?Kwh!0WhO-F_HoA{THB(>+N{@$3L#3+uAoJytHNUj=vTmokCe1oM|0+D1_ zj)+z&#gsZ~DRi}8$t2q6KYno-dJ6%o&$>>E!&nbYvNYTm=dI}ZhowZAY2Uul_ihf^ zATqeudW=|z6353W>lvnLrAg5t>ID&f$h!XEC`iqng<1%pm)BTZQ$->C#AN^3!wDu- zI*-W;FYu#o!FE!Urrr2d{@^4ruz$}7C%NrD@f-Ze0NITu3J40d9Q@jlfR4?tUWW%E za1e8MxZ%N0PNtPLzZul@a}g#R@HFLYU@ro+kE)c{DlnAHxD@b6v=ENvwOxaY965XO zzVeTgNU=7o<=PtWD9W)G{%(2RP%3xgt-qL4{65($Ycu0AwhAo9I?wPR!$V~P>7>jZ zqWBUL&Z33~#(F(TtBy0!-`n&?@t}WBXtP4dep0Yb=+`}~sV7MmZTL1BcQjFxkmy&g z2AWAX5*YxIvb)*GsJU&lN!lvNLi^FtVvdo+kUWAC4U4q56LzoY10k!d7i9#t?p~?X zJV1bnrQTvO=Z*Rdl|n~K4cp*S@;C+jwg8&h+c!LmZ1BI6@C2Mi4(y}qX%!$>!nKb$-ZYb%xF4| z(9xxtDJ}bGWWdh?L4jV}@s>jk22w_B@$&q2`EBVRF_o8{M&#f#4wFN7BFW-zee5a^ z{E=$1?D2n|08K+Q0ssgbwQ>B4T)MaTi4roN)apGC_IZQuCvTg_e@)n(La59(1OlrU zFz>0>R+WcdzToJ9g=KSoeiwF^VJk^QIRei+tim2~@x#f(R9GV9+~`PWPy7;f%} z5IklB!g@KWq@#xPqyuC_!cdWl?cJ*LSi)Esl|G`K6q%`<9V5ik)iDybzaM>FDw%io zyWP$jNy@Y4N8h?tVPUz?yMrr!k#F9t)TO9RvYf*Tt^ng{&`|9IhT+Q7jgocSl!l2i z$EMaNJ?`1YnbKlLEE&_YTZ|vF!svL(m_4JZyoOGY-UFUN3x>xBTo5ncGN*N(aZ>}Q zx+X5U?Nd38IW#ldYPJ~UPQ{z83U!P6(Ly9>= ze`U4V%S36IP@@3xbhEE;K*B~&8`9sb9FCGN92iqz!-DC~${$5Pri#E9Hi>yb3;JdI zwbz|@t^OPj1_T*~i&hLkCQU~%hn+UnzFtVLQnOigUn!RANfhCr!m#UGsQ=5^`GhI) z&X1=K4!_q@97NSzkv`M{EAVFjbJ9RP_859KNYz%i-p(2PL*ZVX7|!$zTJ8j2{01yh zTF9oQ2iF=)4>41mMDv#DoYhMcb4k?W&H!(QR9{#2IDwzw$9Ix&3|l|gs~vhK5~5|Z zkpCAOS3PZS6%h-L{3YXqAdyRXbaBE^KKK{oSI?gfD?mWkx;4@Fwol(s501vQJS)_1 zLH0SL%d7L!jE82I62Cm#{at$5qYxjZZORf&uvaHe1k+B3X}vZeaWmaSu}V4nML7ug zNNUmJXlbEueCG+*OvOl|h)zFCNeq}i(l)njT4eE)XgV>9x-`mwb zESu_1x?Kdj-e?w8sU`pCVi@jg2DW*RC2O_*Xh{1=TG(Qxu4n=I{32kYTbt0 zK@kuXk(tJQQU|$LTa!5zY$gsc=oAuWe9oVcUF@GKc>^DT-q&gD<%yKyfvZQ|DrGu) zZ!Mv}iI(kWUcA~s$oiK)Gd0P4mT%+1JzBQU3g#Ik-uNR5_|O7SoiM0T+^LKIDS^Hg?!Wgcb$CT3znfc-0K zU=VnSDxj@Zp^gw8B{?-FVJxb33<-(p5S<+>$mcd!)ID|l0G}RM=bCp8VMN&iL;zeD zdQh!iX;86rpxp}(0^0*_8PI9X!~fKQ`*v|a;_LjPdJhU-PBnFl1pg^6UwW_iYP^!A zcB%;T2D0VDr;Wfo0_#d)U&b<>5z?vbm3cEYXh0Rr0*Wu8FMa-(v0`J>RAJcWz6w%O z`eGxI{hhN`t-HK2a|1oP5=A+Y--^3#JC4#YzcPc+ZW<=w5b)dSX(nbMVO@DEBcYUD zxV#?lb*0?fA2%9NR>dTw-A2A{DzCq)Jp0qGZQb9sT(6nW*L7U`?LQsypAW7%v7`aa z!PnUSrf&a#+%DQaHLY&>#*&+4kNZf;sjrNHpK}fn#NynS^y-}v7 zb6Ig5dhWEW*XJFbeyJlJ3a%n5U?I?`h%w@p!(vk#q5yz4}gWGMb;h&*P&15{2u3J;;ElI3S%sY4P!PC zFoPkGulLLuo*lo`JUYMmNW#q2t?AUKEAxJ~@4;AMY!kwm-)l9O(VYVg&;vAZWXLTF z_Rzc1YYuz4bHys7<#@cSGUdaGv|64Ly0j4yV?B#~^~L$Y61nfxW;OY#V5Nq^N{}x> z3H64#mMaG#F=>;XxTLxTyR47Cj_PJ4p2y>|_V=jxw!A>??K}5f_ zf5udllAqwKStNT06`J%&7+#^4Yo?!=&EFc2xJ_KpllMI(i;7tuPH9h?89}x9gq5`_ z<rzEbK)!hERe zHVo-#biGaGh2dHr-N5Xza#eC2Ml`RlKO;HEWS@7_oKA62wzDeB?_0o2 zLviiAX25qYvr&`y^g!P#c{!%8=D}Gj6*Tc%NxO(LQZK^Jv~3o+*L{03VjW0_c9Hg< z)n^dKWlF*J&nQa?rHfOEpKCR?bvvb!4{e>j8%^N_$3elmzD>&vGlQQ*XzdBGJ>8S8 zmHV@=-9ICTPK*JsMJg+|9}ib$Dm^bVEpLypC>j_kA)zXVM#!$UR;m?v4}@}bWdsPn zPp_?$%ZZg7aI5Aaxq^E`#SIyh#r1BySMfq5^?y-0N0ck->5|Th7CG?^cX(B=rBE7fJq<8$sd+)pM_rCZ2 z*82Xlva&8N( zm!xy1=5fQn*s2|k`@q~OPsrPNZ`T*1`{`x-rM+~~C9#aAIxg@xM15)ZPVJGajmcaA z29-D3V#r-}?t6)kWf*tN&}#&QCApIm?5wzBA}&>y%hZ+Ey^I21rdENV6X308Eh4+W zolW7;N%w=E(X#QdwWgPGN;I>u)c9>AH+4BOFH(yL>wH^KP`@UGNthPbOU{@kIki0w_by%jVC#?EJmG^p@Rh z;dq!No$JKHiUqOv zKj6c<-dea*5-~fky^f6euVP)6eLZqzN^4%$)eC)VCyZfpbX>0%V#r>hiue#wa&g)w z3)e=h2&+5YVWrT0C5>uQKz-EBw0&`UM~I^i3ztqk#`rG$$TDZX<n4;_YR11Egku*<-7-D{6w|wXuV@$BY6ijD=H*$}SSkFURI0yt!Ss zQVbk}V)?GG&5O$fPL6~%+48%`DSz9q9(;n8Z*e6c^6>SEK1o{=t`7)`7qyj5jfk1+ z@_Q*Y2IebV^2u(AtGqYg{6N2t7G9a9v$#%kNcBQJy=J*RB~qRbuXxDR*3^c~n&i&x z%_T@(lC=Q+l_Yc5_O1U;#`}%Fx9jQdwFa&gWg%$`(E&RZ{woZw8Y<;yU*4Zi2l{** z?lpX=6WaE53@bp8z9H|rf z8Y5noT5F>CjKtW@P$lvu0LAydd^7agCX~Gy0$>UYeAZQ?mesw#?&f<>c~x~(sXK8h z=6ZQ;IgbhrKR-rlW@DQgd0lNSOJ6*SEPzZvDkxRk4nIqfxXNc=AwTLaZL$!;uN)G* zYr%8+Ru+3fhwr+y>Jw2mCBU(yG-tRl#A`pg!93LHVaA6ok6NXEaCZlE+Plj*ZL{E@ z^pF`Guf4cY{4W#mX%Fn#zSe%%?YgF%NM>LA-+HZdn$~intb7M0WDLHvcfxW=(#1Qc zw*nb0suWe)syA(iiMJhbUo+x$ewhuiv|bB@)$V3WF@tyD35TGODSJX8hfK45A92k+}Bb^3pLePOxHAU@#Gxqg42G){q%c)o~0pg$r|-Xqp_gJoj&mDLQVTz@B^Yk0vs zAw!m^)L;!!A!H0VyOy}2qv%^Yf_c#_=g$!N0P`_|U8u}u8RqH}&!yujDPjVsRh?E@ zEoI9Y!V}gs{+3k;mvv)Nah&XE?z!!A4}&Z*>G^S@~{a)86&MnA0!HuP623 zi1w*bYAmh>p=~^1%>av|R&=O&df^;M1yv7S%1bn-=7qfkr;P0_iL1xs%Pgd}k+a_i zmQcPZ2_O`BUT<50lRyxCSnS=q7soK7aPtZ`CJMpCXTrOT(^H0D3|qU?c}q_fq0SRn z*ssqF!?_%00Y0tj(eU!}7=NL9xw&dp;yh%2)|G6ZVN|@Gx6{LU{io^iLdWu|zSf~Z zUD+O4(LwK9&?TREe-mgV-KlrN6VXABgogb~Nvc%!YjK z_d}$36ngp0QicD<@Cs6ZMIR^2EGsL6c2_^}s){`COD!W<8KqVC?dE5p9y~Odm2lq| z?@Y|Xm)CgOX{#7=g~AnNi8IDiS>E4M-PLnx)!IP{`Vy&x8h@_#r$H`LW`bF0y@D*y z?~5zt zf$wA)M@$+p+2zF)20CpY@@6$MOz+PKkmTNyE320-xe0eZ@#%F=uGZP8q^n?P@r?um zxv;jxY%s#Gk|qQGo1~uf%-H{`BoPVl4A>5*VFs;(AUYk@uD1hI&J+tF`t?}cSy)i0 zrd>%2HW9x;f24nYwI*VOR!%*VSp~irK+#G(PTK zv4ZFKAT4Vk$_iCtUMipweMBZi?uoF#e$&>_K4P%|2B*;)W6T%mqL?>7`HIKw!>4V%fJgc0a4e{=`AEimOSbmZni=Mj97xV*6Yh`hH>^d;Q<<*8!ApixQGWbZC)w zy^0;S{`#OV)>t994lR_{o(!*3XZRUoCb}B%R#vKcG?y#-fZiMPkxDnhCbnHH)=868GBB%0XU_W=EFaqfc9p!(}S;PEi5S^ z-HPl`r11%x*r(pj4K5=9yNNW%bZG8;USS30`ets9Ho(i~b8S`7o1+O+SL=@OzF`;8 z)A(}tToOVM6|067cSSU=gy+u~a8l_M`an(=H^&2qZEL-GQL`QUSG0T8ffY2r321B%QWs*r2(AMJvFgwYphs$ zrH+zkb)0E;a*ZF}Y?ORD_bNBRYc%}xxApWmBF;BGaaYKOkBA27Fs21Bc4L@9lO6z7 zcOhb}#YSSh_#x5w1b(RZ@DB3cEu-sf9Rm2Q{s0#;j z4w`wdpCrjh%YFGVjs+eRll2ucw>)j@{*g{Makbw63oh0IecD_xd}n?%pLl)GrHW>W zuxd5bi1r|Xh)aqo9fxvwhu~{7q*R*3<`P$l9Ux-$h0V$wS4(++$Z@POrRw50HiB~sopxEUflr3$pnDvhVq--W)hoCK zt00O}1}0qN)|nFXy^4VXs7Q9e1n#cc52e`E_4w_Fl{+&fGWi1m5!;!HLQqU*s|}>;K3)nrs07eM(XevEm!$ zmIFDk<%K-W3cmhLf7M1C#28>S@5mnt^cSGt`>fai@JTJ!uf}(7EW|sHSShrYxl6S` zBm<+4xoFLTco<{XlsQnF4>q<$92=pUG!FJW&x+C|IH;5HmC7SC~OS`(} zSs>eOeK(Tk^?ocJ9v$(T$GnsuY=Ec!kVv<1%w(wd!HXps|f_0VmNdTx7Q=n2y zT--@@QHkN9T!uq1D{oi(7)@_awi~R`;l!QC$CA%IW>Xee``rAE&42s74VKx3bK|yx zxPFPfzSWZ}-k+@t3g%Fu1n&P46*c`wjON-)9B`U?8Y(`DQl$Iw5cC!57gQ(pkPPL* zRWay_Nb>XzvtHts`7Qfv;U_nlUYd6@gg?C4*+lVEi zY6w47e-I+@Q$41Mfzlu~=$a(KX#`{8Z~XVj=Aly-BgE-M3vT_+T6UEE90r1ox%mf_ ze>x*KEjl3e$7?5+Fg(zZ@MN%BZp?~TV9`y&Hp^h@&=GP!DpDYb9bx|`bXmg9CN|%T zEX|h*I5hY0ILdACPrZko(XC)Wl!?iDv!X*0&iUOe*7yZKUj^;K%FfZH&m7?1R&Ct& zV>X@tIkf$AoA1ZALfcFg2!zb}^i4PF3X-Qfnt8o9*B4bxW;dUnq6Xroa&u6K1^BFe z9;}4Nw`s~2zVZ1OrXc0)-AQEb;g_{xUmVg7-^rO&$C;SLr}Dncqn~0>Tp>SE19|9& zO^nNxrzkAU)hGNF_fuDuF`9(IlJ#-Y9+0l9kirNkA0W;sgfD6W-OK4!O{Cjnfe56h zKY?ctiGK_+A1cTng_uWMYoC+GXkBr@huh1V1Cq8%;H{j|3`L-~2eX-Yp%4KTyo^*`&6n>-VNV3`(XZT zD?MYl1mCeA;wz}CgNucx=4U1Zj)M%dPUXIaZ@MO=BFemA1pHL=s^Lw(WR8T|J9OK! zw3)2tn(eVx>m)GXDTe;SmZa4@tj1%bBQPwR7atI|8_ zcxl8K`hXg3VY=8^2&NHP&>ERh6g85eYEk*NZeJHK$w#(b686kX+HM!ZB?$A!Dx+eP ztl)#f>;;lM(X$WrmWv-gnyHwATX`0_wn%Xh(*Qk(IM*+=!8XdP$G`P3kY|0>3%*zE z4xU2qT8Q(??U~TZwf#)?FLOaz;9~0mnDS-x%bs$}O1R{JTXnjqK6@$O<^>c{B3nJ$ zB=PI+)8O7JhlFKl)i)Rs6-XdQy_pX7DmNdR(P*OsgS-nUZco>Z-C6ke zugW`xOl0Pw>0d+cA(KO>teZ_naQWp{b@AZX-%~DV5rRP6=UlR_&3B)o=yEE$gH3WM zCQ%>&u)U%3$;yf+dGDnWv#2)Fe4JDZE=M5Hi-Jr$Q&Xop@SDl|fk2W8Qv;VQkqO|V`Hk`@tt0sb|K>gr` zA-78p-p8p#p}O2ZUBu47h3VC6n+5b39w@#3sZ1yl5-8gC&b{_>3tm!ACc3$xglbT= z0+|5XR+8vmeu!l*RCg@hSN7Xnp3bBNKMDZ~6bMod_i1wN^W8s-;nb#%h?p@5x))eAt(WDs#`Wc2LbQLN9(}xy92#21z`QE6`^KwLU zLLKyvU^hz^@P;?W(kf!2nsCZSIL-pq#S%lqy;c;*;D$2QG%3mmvlks5nGif&T4is# zn9=;cJJ8zg@Qf3l7m3BE6NUq$0Q15q=C*)z5f5RT5_pN3o(6Zca7tKf!$){C0G0E5 zabT8R)tE!Rv~A~cwX{$isVMJ@~fXWh)RK~Ne zGstwmw!b7VE}C9iW)9%xwGeXC+=NCE8F)R1(fKSVf)9=J?AkF(#%xPRhrrqV4grBAgIY+dRl(E3xD(OP1Z9ZO#Ttp4c@-*?>5tkRHCU0a0AP!1pvt7Q zxmc)f*WJ{?u5-+kB~MIM`8*44j?3G2Krx4Gn!vwRI^hQJ-73Q))k8r332n6~x)J2Q zyuY4GXNV6i^$*7vSS5fiU({=b;oYq)@DdcPX8n)}cp6OAsWlE7MtSX&>u6svQHE`Lm!f9s*WdiJg#Y?K$s!H+RP7V8 zCtrv}o+2e(K9CHts{@_E=tFMfCG)jp(!BCO z>VG&-cO#6xFS=|%$f~}cxC-2vbIr}&3LRmSqDnuUiI_NLm%-$@4Fhp#^B^@(DE1~p zFk>Rx-$nIY*q1iKoBFd!(7@LZDi>S^9lW-&h>bg6N9*9A- zA^HNXU}Je{U}N5oqSmh{OC=@ifqZQfW>6d_Qx{OzAS`lKbrS3m)8$J_DYYH*7uB7a z=PN+pDPq>I2L)-D#+Qk0xK_KGG+Vx)}{Ye%I}R6sh-W#LudYT$A28j%DCYY!IBUE zw>|%R-f1vitoK`E4%Y)iEVvig9mtXlTT`bk%7GoCA5kEs9)kqcez}elEV@}L< zSN1PujVNg#He}UbyOqVNb&%x8ts?9pAigX} zt+^LASTDmvi5IefjI7@Egnqs?HJ=05=F}i$mbl7c<7%Bb%Ix$p*VMrHgb`Z!s z=YkfkQMrG$-JWF%ibNBix@}eO!7=pBcangw*@;2ii8Y_=;{tD*O~9rD-*-zKLwOOr znRP~gvZ?x2&lhbb(7`p4TFP8%^(|Zc_eGSZqGB0Gaigk;F@2+tWu3OJ1?i~dmpNq@NbDE()s&B=qU%^kD`&{PR&4`K2vSaCQeE`Wg&MicH& zf3ay#j6|c?wG|`NR1{?A7DD4Si#ww0nDr&fa-9Q_HgW{(vuIqVvU5aVf~^RFhm|%%E>9SN0>Tg-vuCYSYKNBw*RhZ=7;Z!%Sxv8-jlc zDx^4aP(h60jY9%fnJ8=#JTXHv#X^?tId7P*LcmW#5KnW0z!4oZW>n9}az5y5O0F3g z+_*c7%$0fTX1&-Lx{kOOSf_5|TBuyPK+Q1Q^Vhee_?C1+&6f2gNS*aJrlJ-fy5l7L zt^f9L<;O-_;<-F#73yDVrQy6NkXlHFR6842K0j6!S#sZw}qk%?YcGy;hn>!s%;gQT0y*cV+Uqns&>b7K?h5J!83p6%hza z)AcnCdv(RS$i$Js8&rdYe1r6O;I4*U&_hq(t`_*-Ny>@jdF^4+oJG;!PgGn7BZwqU zOC=tw2$emLHirF@2P}ZG1!86~>9bw22T};6w#Mrc$5D{h%5?X*Vx4Wn zyTyHCH<$ayq1D?JWlG#Wj$d>p-Mw$b0}IS((&rrQiL7hw&H;N z;z453_30J~LId{i<%K!1nN4ba&P=UM7{2gcewMzHRu#V?x3H0p889hlG+n23?6o^& zO%>UY@YJ%0J6x@W`mxKnNe|B@7cK9;Af~V1AX4AE_(RkaK2V?i4j#hvu+vcClem0n z7`k%BEhX%g{!v)>G|M@g*Ji@1wr5Kfhqp6|t?bASykav+Wl6M;|J6PIm&KH)+&h-m z;onI%OZ`kmS-0ei9=(Pc63f_z7Fbpz=ZWfNJ?~-0pugR?-f)J%ikIQsyF%|fLjPiO z(0HVLcgPG`n(XDMuXu#cULVov_;H zcBI?NQ2RIZWfjWy8pnnUNb3%Sp-pxdYK@ab!te|N5_?|;ZL%;W|c`YL!)>Gw;-e{7NE zO(8ZwSl#`9x7zQu&Oow3V$4Wv5yj=4q&66IrO8Sbc;2}_aqpg9{nVV69Yh<;DW1p} zbqu+g@qH`8@2;ec*A6vTB03Ie`^n!GB3x36^?Vbfc+Z8)ob_kns07#f|7JXF@;kZ z`Gg0pKDi*0Q>+L~u}rUtE@l3G(=t2o1Xab7vWfNvdlh6sLPY91BaA&oLQ;OQ@%vuv z`B;KK?g1L~BUI^eHqMv;Z}{4j zL9?EKpV6jUdMQloR%nlRHS*pni{1CYS23t~WYpzaiN#dq@OXux+?_1ITJT%}y76oO z{_#h>!V?bunKT6xk=Z}qu9~j2GiYUH3_}}KcZg<;=1hyl3DD)mR(bC*rF5Q=Fy%K} zzIjcqS6h=rw~YpG^Hk{A_a)iBvyY} zp1nO=r0<;Gj>MPK1Xde2HEqkXI*iV*CSIOE^Qwri%}big4L$zhb^mu?{a0f77DF^O zXNYa13$FaZJ)5Rw6&-$`B@H;xig~4T%Cebl`$5z?W6fP{s^WfW`U$J%H+V@Nuv*Yg zMly?!@!0HrM^+T#|4iPkxgN?+DQpuT{@?vUjST;P;8=OV9WMO2*z=uk_U*@=`RM}G z5m69c1)Z``F@C(Q`lz_H1?Tty%bZ`x`cYi83JZq{IC|*$geU;5tz72+ObJARc-s1# z39NVBk``oLwx?o5^{lwY9-+@ERnea|P{pNA%tC9v=&rL=)zjP$?VlTPh9Dsj3fIU6 za~gcpDB58F(N*$YRk>(6d6B?!wxS!XOpD+8RFJh0zm*~`zuZ69M_~|a$c*cIRI2w= z4=`o_Wub!V&N8DKCj(uRQL~R#|0Cqi<-4Gu1JAFMPs^ZruhAR0PV4DVTi+IN84%&K zDP%Sia#`Yg)n8;g3iniK=%_m*XvvB|=H^MC&tlH9-Lz z*1`}G2V4QfCLo@u^R14NY$<}xGh7<_s26Leh}dhOpmAcGT;TJ@cHUwzJX-k6IW3h| z9W$9em4{2el^v=~fFnzTP!%*X?Trc{;F|7ze0+oZ|LjQTx39~+Q&xm3#MuFIRS5!X zwkOU_L++-v4V)R#tA5?e=Z<2wl@wP*d5rppZ<4B^edUUMkS>NI(Ld@I7p7$022fiH2<-%jLalx_K(yZM3oT0cx!lL;->OhFnoTWt)-{k-CDW zh$P&Kg47ftbNAZUpd1H$v95dFd=yF6Oi{5GD00=l2!b+!C2$ zvA9j7Rj0j!pMpO6%p=2Z(2OjG&3fzwQxip$kKOn&{O2&Fu%F8s(EhhB`)q+XBT^zS zFIN>UgN{OLGlCGJz1%na zW{q`{Q&eou!sZy9W>p|Ap|s$@y}C=xXH!w3`89J3LkYssaHl?9h}_5ID{{LIw$YBN ztIq)k_<_>m#9L%~3-)zUr^=WknJ zEN;eXx_MZ0t8qEiEk#o5+n?dAhK2j$?gA2EnU~x7;P|&28?lki&WMy@ci*S)X;;8) zls#66)VM!q03Xdux)^zE2KheA_})jo0&;9e>l3du;Ig`S+XiaIXn=Vy4j}y()db@x zmE#>e8Lp}=CqB38dv9R*gVD1!WxvqXa}>|nRSosK2BE#@>r69UwptIajnKWOEi5bd z?@@2bp8wax&XMhtGLbJ0eSpR0{-Bh7vB zxcvWk$4-dRhFw@5OffGLAOiU~cqc@)&iqGKsM;u}Aups`i-jSS$QfdoC5O4Qv8eyq z%BGJF<5J0kW_prq63LWvgV;`J^;or}bk9}aMXh!qME}jLI15cpO?ea1j9!dvX|%G8 zkUMaG(d1%|(kaQ5?9cm4stb~LmhbF9SKmp;gg$NLH-;D0l|&CvE=07?)rab5pM040 zObNTM#^oDO3iQvyXi&K%9O4jJ?L+S}1~b1ELYY+JqG$tX-Fabd@Aajo_z{w#*XP<>gHNQbe%gD4n4+M?ec$i<9?sx9BF!13p zSe0iv_-uQ@_)Bj7Ed`6o3umh%=}O$zo2Np$`fZg9eBKbph@Hy{aM}9X7iI>%UcaSa zsd5th`y8U#)-L#<#3b$ zm2%hzSiZ^XdQe66snGEUEbol+Yder#HRJq1=8N_>haQ(bJDXa}Hb6f}hViC941eT~-c2 zh3@p`G3Q8zaK}Y+bUkf;U!SmY{BMI=9rRkp$l1G#KLrS0*{SVczPuPTcnEHtzahU} z8w6-N4aL8tsM)Hqo&0x4ykN+J(-sqPoNW7|nbk2(8FGk?-3Qd3%Puljy)9jaSZJwm zb4%w@e6Zl#==W{Y+iRvuWjv(vz5t7l>O2|z?+g8F?ROlHqsp=8zbAF)H?P~W{dN9l z#QneiVkM*6bZ!J$>-ulw%+FPM)XA_J#S6qNr_B(s( zW;=YlMu{gz7!$Q1>Ds6W2$ntd*ij8rb-?f1l|~B13v(h3s1rvKrInl#cZeUP7vDzSJbJS?A ze`Uf;;Bi6?QrtC>JQkNk^}C+Yq2MTu00a$f!pNV#GdTu_f)=i+Mix~T>K|lE>!hQb zwQ`T$D-TqKJUY^_Mu?xz#JQ8YRUAq~qou*s&1-HwBqN2|1(iOp4gu}sWD1e?VF5rO zOfLZl!dbuA;M&C06?J8p{hzk@I#5$>|8=5}&>k3lw| zt9UPJQ8-yyOSCfD+_N!rKg#tY0X-Rx4y;|y4J?Xzc(-hw!7srTHs(3DtL7in9Wvh* zqPReE>WS@vXVemd{er(#fmKhKnYO_3(I`#VN4I3=S%zqC=`)AAkU;Z6czus_v zq~XoSyV(#&=dsE?coYK?0Z`IvL7la2(TC13t8u-P~g&o9WOd8=~o`0eO5{=sK3|> zaIx>Uu61+{1giG$?oqsP+sU&#nPG-c%S-317OBkB(pOOB_@@@Pok#aO7^b;)4^wl| zac3gCfvT482`FyD-5vg#7vjWet=uZeP=adLH*^m&ajVwU3|}L(Ac61y5;d86Gfp&# z+N$~N+Ql)OjMDLeJSbJsbcTrty5>YK6@T?`R&w9HfXkjTuafZsr%U33rS3)Tx9hR< zxg*EldZVC;0swqn`CJ#&L#yaoX4Ti2o-VAPdI-mmRaQ`L_;-stMb`r8fs&HE0Z9{! z=^5x6`nDnw!9sl2BAe<$-ZbPrSMpH>7UE#ECif3^$Zte)wE8Wic!|HmfX!zfO7*PX zeIorzm5}M-l^S#j;aNTf+!nKy4>^SU(gVYtm0UM`E)z+-LZ8Bdj<>Bnv$w4&M+F5WWV##My9Ucd6|@!y zsATh3_LdJ0pBh*}rqt0)FW6gioDcS{>-M`G5PXRvbkw<(S>JCRm{ZomSKQbS->I|} zXK>g7e0EY4F+)6cwWXk?E`F}+b%hjdHH6#Mt~3EuXXgFrOcwRy1Y7~Ks#rY>$={%; z-kc%cf?8Oy+w5k&tvCB@+$y6lKc;Z+PzHkab2FvRX23$)hJ7 z8HIMb=NiZTsr|A&%u^zI<>qF`<39J6$c}y2Y>OXa1#^R*lP{*Vb*duM@$VflgJAY2KwZdN1gMjxW(8ho{kcsJ&2(Lf>)&|ka?mg0%gz-f zA>iRN|4N|D56*JHFWGsQNBB@xaz-~B5(k&oh`Op(1&h@p5H>VdjyD&L@PoiufPCi| zrKIa#=e^{&dG_Yyed|9?g=l9%)I&S&?o|E!G917g>2OdobEta^oQ!3nyV`(T(MpmRt0}$!i?=1*jN7`PGeE%Qp-&iiZ#IJ1kL8rAiJcMgz110y2}}Z#rbnO5uPc}cR@PmrK3^j zo>@Kick~{Ly`e2Ddbf|lS5Ut(&)0u|i>f9H`YaVv)H^;qhPqL)Tl!VA(-?xf@LK9=#x5cz& zwJFeuj-ef?3r$j7U2eQqk7z%2VcBKucH>ir_&Gw#874mfztuO-8D^P%wp?-Pc&Mtm z^Oow3ajP=h?P$yn+l8zIx2tO$d{*IKSFL=pe1 z)rpInTXc=>Y`9ubA)Y@y1mqV#Y%+cL*lLfny?U46*M}(WSIgb!f0)6g-~ew>gJHVZozr{IpyKWG4d6|#vl+N4~`}|a6qPJ z37cC2S4FDnjA56cl=ZM8%k)&TP@{m1YQkmJ=kfpbLAzm8@Z^Fn?X1Y)pkNjA{k?I{ zE*DE?`rE@u5$G1ru(5Z_YH(qd0kK7%FlBtQTWI$Rhtx`*&xH?FxuX0gB`*}Zy{j6Q zZ^uwcriI>FeQDFhU-zh8LYyVeLq$Q{ckn~F=4)I1m*qHXC9I^%aEXfAHuN}C1M&UT z0YeF7Ton0PaX0Vk7>0|JU*Km2VZ)e{qDMyz%^2+TWHPEihR@QG{k<&{P_no;lGN{) zQxs0trdW@>iXPd!s4x*aK46@_1YCcJBng{Wsl{XfZgGX0m>l?*JGSjbLR zCx+kp!4U2B6Cgtrjj*#ne&RENO=iM&g%@J9uTKPa6x#mqki}lS&tkgzM^{p(eb)!c{rN&kS0YZU zIM$%5qy7wLE4!Uq(d%3N+O))C;R>qPWN>_@Pm258wkR~zWoFRP7gAr&)1f?k@@Lg8 z-a$GcYnGsh=!g|~AudviixfiDWu~51dfoqJ(Z|;77*Ni6`x=R)WimHdqDAs&&puA6 z?3O~bt!UV70}JQ}+^8X2J2TbU2o)ZNCj}KUP@eI#cOLqnf8?Dw9QM}CF~E~r`4(8P zvZ@*_KHU|QISm-y|MWiWI4`D=ba0UjbE?t^EF=ybe#W4lWawBYw0HM}o6eDH`w%<| zzQEB4711n>?JR!;hskUn17rx>0nc0q)vA=v z@^VO}5WR`@*mC|3#>X#(G0@=wIy34Ku1o16Orc|onUSlEBp6ja5(!)O8ZL|K zZO4ws#u^Z_t=T=-Jqz&t)-91nCyQl$f3YPHq74o*fy+FtrVNUeMS)wA0G5m2WKp^y zjU2b+9oru_ZQf!O-9M6Dc2>VU&og}!?%dg#dUwNxZX`}oeH*8BblRq*VcjWoGUsoy zZx;Ii(-7<*Vbx=0x-#N6Hj+gPzbwFjzG^)vkzTm3;xS{L}w}Z^G({}b7!Vsx5tMAY)pN{ zu;%8Wj_+W_-ZU)M@ODTCiF4;|cIB{Pohn{wT7f=4x{$QZ@8s4u#R}rJw5EfP%rE4+ zlEwglqtmt?Sz%Ec+x*ei1w%-zqK=RC5{onp&|9VPWZQ6jXqn4>yk#~bs3vTSFD&A) zOS2IX)N7BM{4}Z9yKDfsoRNl*}>`GB*=(iqZPHjpTcu#~=3m)-{jYik_U_4yE%|Ox~ z1t)FuS=Coe%>&Q0Yj|AJ?BT&9E#M9JJDfwHN3RHn$B`N91)R53{}knDStq zbMjA(=X>PCvyi>hVaF(}d%8qcQojub9&j|Mb9==CvdAHF1BlUZn!63z`5%MsO#=Oe-U;jz4h=EbnlGx*n;^l zwhx3~Z1>Tn9k)zN6T#rgq3p`NiVh`+=z&=_t8T)!B0~$pwmav@`wFr(o`o!f@3G&r zK1^h{7{BRAa`G>>T-F()$>l^N8z%$&;nl;>=SFvw`@-e`lwWLSSZ+7p^e`q*!y>ID zwZjUs4Wf!}g17m=9Z<8dbnsDh=$N?UV3$MGx6c-b>6u`6O8nmO(tQPq=Cu@h$c9hn zq3Pil>uGTUm-o!oQC3l+&Sqhsw~BwU_3w2r(!$PEw;rB=9GL}blsK=h+GOnU9nLDKfI(?=J5Ux&mM zX%sy1+lS(g@`;DX14i$tnT;UMv#-EICl`+9LX6{lKC&A0n2pqP`tC2b=S;?4B&(v) zIB^c!xoe!kIh!hBEst5PuO7ShmNx&3?JAD7@dwoQAN9O%7W(3~>sGN*IiTUYMEtf6 zJbT?TtjBoB(?8j*!1NcJ(C?#i|vtFX!vVq7JG>CXXY$mqZy8gtbbIUr>qALOAMV%-Ou&42XVc-&kzMRhK}9# z8zwU7y@GaB-?v+JnRnbUKe|FriV1C$U7BTfS@FRe@^QWM^~iy-IK)D}ow8R!D=SEY zBbZi!C>zFDhv@H4Uv`9ix-z?n9w**jwQ zx~g=pfYDN_tYMO26jcSrXg*Nm6bpCU34Nmpu+(v|5 zo(k3~yAN(v9TYv{+UHL`ax^j=&T0ov7fFj{?>0WKaW$H>92mXVv+Xc3SBUdwb(liG z*L?=9E*P>dIxRPBS+&&7ZOow7gt_0u7<`25&G6a2a7#qTCRoQ#*Q?XA&5;~U_ zWZ9*AKapy4#e%1d%XM*ItAZ~{EZH{9B(Zi7QN(?Ng`H#J0)OIFt~2GvMR<*u4@=H< z9?reZXNsuv*En{gk|+sLd^h(Q$u0V>k`sd?hT@#m-a z?}N3WMT=#n3Zyu_Im-<%N3Ofngd8DWf51c1qDhr(EI9(Y@tvc52Mq3Uo55l7!{7bQ z->4aR-`oG99=%$$i5;{Saqy{t?>bcIi|UA* z6F)C&d%HLP(Vc|NE*n-D@((d&ccgClI&2nsIhHtjw5^V!4z)6Zi!PaOJAnqpT>6cT zxe$HV(|X?8I7j_7aU$HdZsjH(Jd{~l`0^x6+$nD5PA`i7J49*0J<%pfPeu32@sT$N{E!u z14=If0!io~9YQYwLC`Pm`_4V*-h1Bn-kX1{IakSCSu1m`%wHK}m;rmBgJRv9G?V+1 zAK5Uh3=8cOV8U-qGl0MN=`O=z2TtxOrMAmM0|z+{~9zzxiUADiy#8 zVV;hkeP(bM=;sRI)>U%rVMCQW>gacVa!^J-k(Bq{d@+mCG`U%si)&UoGdsd;i>5Jh zm$N37AVIEsP-YGU@+u|-XF=yxR4tr1#QleS1sBE_b}c^%+@xkD8xM%o6SBI1vRxyD zREUb6-Ysf(f+$CSIyO`tP4x6x9JIi&*4hdN-zNH zbv_68igxAx_@T%IoVI(V_a{eteg{MNF;JDuOFWfH*mIn9gtA@XIe(mhKoI}QhW1;1 zzf?xbs3+$S)XYzgSfgz)prf?#_N1SIZ@r+z)9cP5*0grHiuvf90Sc`u*|b*3$a|`W z2<-7z2tRpUV(&Zn2Uu=#`2{5lvCUbR_zL5Ftac^jFy`k7A zw)=m~=O5PJ|G7S4NBfklWzw~Yff}5)LaggTJT)uC$jx)cIYXkJwk0qpd z^!X?LK0Z-J6cUtW=M#HKmplVThN*_}$u=X@>L%p|%c((!klXNC6j|#T3lipGQ7l4B z7YFFl+j7Riw$zqRi?(E@(VN&_I*h#ghCvQoME`poHkdBkz6=Nwq+5iFna&=K&AO(8 zO$G5@o&|DiOMAxcsN$&1zarONPXyF${B`r}*9T{>d~C>Q_?GbEf*T02d0F7>*((j- z&fd8E>$aT$DSUDM^J?{AYEMe@$!FdkW@{2NFP+o8GR}gk4X)xCY#mP6x9iI$nG|~& zfiosRSV1zXKuQeJwB3kGgtUMwdp9mmn7P*#seJM;1w~C+NeWAnV@@gk7;N^Qsic=A zUOm`13$4oE+*6nN2v+)hgd*#y8mh_3 zuzGTl{bU2PFC;m!(D`);rCrQB}_N{Zn_NW11*+o3Lt+4|M?ZTxrFKVu&K^@xKiIQNM4dn?Q>e>Dyvi*%7-#=KJaP?u z?zmZPAnMJVfNZ1B$JC^^eWfaltsZIlQkF;>c7T|U_l6NY`+K1^PtTi4fCz3Vr5Q5wQ{~wFHus^!7#(UlrNMu6^V-xxVOYN^~JBSM#|xW^gzb zWV$EEOuU0T)F~$PY+3W~afM40X6ykU3kL{K{`%)wVYdXono96b@Bx+p0gQ;_A&vMDpxwB2IKHO!}WwQbSHA32*Mo~{V)A- zxSbVz{d>Y}ydI_86IDM8PwrmxjcL(s)Jxj+tqR(UqfpY zA=SE8_gW|svc34PZx?i4S`LHSWpYcFoJ7o#w_QDeFtV8y_4iqHdGirzI^T`!iJG|W zhRr_^VotIz$oa0e$^G4{-mC4oR6VPR#LJ6Q9v8#)LLUyaV~b2lqx{Qny&Er7ZkxEp z%JB|&B81x5-8 zLk@Sr+V>X<@>1o$jS?5gY<5kcbWzVutR`@=g0;A9a*mhllBjV680iS*~bKvj^HVDzHE5q$WM;?(2I`o-Z;I8N;B>J{=Lf~Xfo&2Fni*J8&|u> zP1QWfll-GGwr=_#16U#DR|{Ylnort2l))E4N(|e5%i2W+g>F;xubc_>8fR+c4e*-Z zx}+-6`>J6Q0s?E1IB+>ja^cTpS#Li6qo04&Y&iVP2|oNYDcVQDDmlny?bs~bm4uJd;nGU z!R(I+UvC2Zp(-hS**^5R{oA8Aod21hdBqg-zm8Q`W>$H6FD^ZRiGI)@y39uO#HSQ; zR1ho1_rM@Ar?1^_469O|WbTbv?@~MIolV){{2F4_1M$bKJcXd8qxd4Y2{8>_lqw|P z<`*zmk?YJ0K91n{prp)nB%|F2Y?m!=^;sx~jp+F&idhNI%Fh^24<$=Iu7fx#$+-Q1 z`#|83Gxuj>ZzjNs!owcW{!$LL8|z*P-M5@~&~ShfA^pay0J-yKv?vy)pc*=IP*Ky= zl~!Us*jhzua8A(0j>f$C$6fp%wRSi*i<%24>&@&^-?P}k8*N+uK2+QTt3$7X0bFR#8q*_!{p>&&wZIZFT08xIg#$=oG}NuSa{*I(sz= z^71lY3Y)nI`C7HHG5BB2R!p4q#-qDRGWHGkw@tmgQVt}+jI<4@`_&<`$Kc&8fpkI3 z`HHy#K}plSrE8tfFvD>8i>H3IY_gbXg`XTvScoZWUo>Xdj$1ZNQq=e-hmC0*OGBOZ z;on@*5vMYK-(hlq@1;jdaaRD|Sfl{@RBsgEL9t|op6=rJv4{Mw{=V}4jq-WkQo|)( z^0||cA$#!%4F>#$ZFLF zrq~Wi3a*X-hq4H98QcBSv}ejF4{%GBd-NIXHdLp$vm5JWQ4As>M zvxpMomx=wB-}>uIk!|$|JnJc|z}3RL?6cg4RX0IL@GUr2fK<{@K;K7X>|y&X->e?E zR85r4+7wDFkogzY4Xf1Fi78s)N1p!f-k{s63JG{?oH?dYAVn@Liah_&(Y-nNvQLf1vRY_=2B2xMi-YU`5vcM3Lm?A+Ua{!G(P8j7Jf zw@BKiR;)zA^A6CFGmgaMc}_*#44e+V_N_W>G`3-8 z%|0wN_0lkuUqhzw##$<)d^8UO33?Dzyy%)zCuH^~^RdK|?|x@s8N?b3FRvDjr5QE8 z?BxRYIg(3FMj?G7Ua&8`wNWFof|Xp@n7`h?u2>_BiIJpPmVJ)^pkP+9*0mi+AkWlU;myx zL4WLI(q>s*DmipU{Gk3-sO(fFZ843imKAlL%~*~ju&Irsmk6@s1A{Gp`+pEi6IjK?DU%=$wyIu>`4&BAeHIR2bNFZxrSEx#QXht(xM4PfY#YwSg)PI* zg*)CzhWbGEC;sfUS|Fa*W2<$eh)*SiEejKMLT<9-HAm_ynMB^>!zGD1-qJyhXV(%{ z>aNhcb*sRtdJ*TwFpIJUB@7%OZy1Jf@Gn(z7Mjngk%(?A<{8wdMw4K=C)tHYA8lu zSzCyCg3s|N`|xmtzkIpSE<>mrZ5V1Pqky-iuZURKyIH;CWxkbPHMQtA{x6-X0s@<6 z{xYG6WP&INKX+lWGw9Aar{hcC^;K0%l>8#V;kn*n1r?AI63?f^Mu+{E_mmV8-MrS$ z*wps%BkK0-Tl$O5#j9@TugheEL7Y#U!Cntx!ynEr^P~a<47=Av4xV!7XBRUI~_AKaDoi z9|rig-qi#1=b@O74Zy8J=T-Xe4!4TtSSkWx5xfER-!(!!zLt_wBzH{p0Gd-JdYg-j zOqDNQOZnd{Y~1WOm)cF-IHe@U;}{!(K^G14L9v9tZ-9BaX(VQ2j_- zwnjXtS)l&aY=vdQ$>)1bAc60Dz~S&bx4MX-t!?+uYwheFljgJhK-w< zEDPYb3zCXZbG71yRt-@{EhxObzLzfocn7{-mu<~VTQ3otqj%cJ4}1#9zh@(4cJYmO zLnWKk??OhpT)tRLkpXUW!Is4WfBZ&+^r)9g7ll$ytDiIqtX5<`x+=6MUnw5pIP+w8 zzg2i$+bdma1>y~6y)N-}_2~WX6hTHfvshf_I(0X)OhcJw8TKjb%lX%J*$7V2cN`b~ zIiBR7&zmhlr5pzww#mJHMm7QI5}?+Y;`qiltx3XXW6_)b#GUS+90l8A&#dexOgrr6 zx%+hz57@#!C~VUy?R2S>6ArYG|LpHSTI{Lo?AU8|^5}ox@o(2pisPf1{LX*=+nxV* zqX1R25?R-*7&C7sT)0rzm%dQ&1CsN=z8II^GSbffT-|Ix!O81ZMHKiMZb?oHzw032 z!IsP-*{>baK|hY)`3*2Dw5IWsqr5YrgmJagLGRne^`nQO5#87qhkX7<+t4Wvm0z-Q z?Bd4r$IV+SVi8dlD57I0|ttm!zq1YsP zV2@R!0w0k#z7qv&uQ5Ac+aKpv;C~_CDMlppEpO%8t*6uDzP0ZXOOnjJ_^KEY4b&fg zGo83OGe@P?r(`+jixmqy(xwLQh-@S)s#PFWV3)u5>GzdVj~>j(>a(v8#vHd-8pgCv zgH68;&>g!TZ7CxBs>3V|ct5~8L5CLD)To`I$MQA)H3mIoM45$nnuTeU4)bHOM(uP%}PfaxT(v)V7PAwY2dF zgx;2M5YKoj>jXJXxYkiR9!<`#7vp~9HN=Qa950=*w)D5jkabA05@T$A>KcDgxhG*V zTosUR`jH|=OFyB?cMhvdt^yxoPvkh(hvxMDC`vlYJRaIlIq-TB{yAfJL*k35U5akn zVN_(U&Mapn>BokUh7tXLUe$?+S6BX&krO8n;(R+O!mTL1&p0*(Y{<>5j@5-(INVRD z-Ge<#2{+hju^&Bjg1)y-`BpQNU8J-0yhjV^$k@UYLIS@Hz`o&x?zQyCm1h@Jk1nJu zQ7AG6$YoY;JOLAd42TXIH&8@l@x=weJFm+zPYWv1*<);Fg|Bse-{yqFDU@Jp@qysZ z+u3pMm)SDa$(`Yp5(DOZbg^*m{>{noS@nHn&GQXRIm(motO1{7w~gbEmr(?k#vO|@ zMjjQ_Nc%(}ax+ksYYcR%#7M-cWuWR>)+}E}WQNKtr-L|pLz%MFcxRtAF);`9`={D}3CPl7~+*{2rX8%vV!-=WKk!unJRZpeg%0SySxoX@sH z%Nd7Fwd!>Jw&D2nTk<%|fvLCb8E?6u3Ti83VZq41aHWs8AhRq4BJi=wLfj5sah@K` zGB*h@JPFp|-=St$wEvOfu*j*m~Q(uBenI8LgSh< zi9e>N-(PEl(>t0Vj~?onJwfWfu9^Ty8E1t5px;&trx8hnpzgRBc`%-GM6qA9(}P93qn5d?Fbtz zGn`j10(4$OE=>8*-9Q`Z9+{Eo(j{pOoI- zXN?SRh9TryGiF5Sjz-{+8Zq1hrx`yVi*Lo88jrt}o**uaEIV0kB<}K#=K*!=-#bsJ zcCcwN9GHH_kCkrkYkB0XuM*Ohvy?Wm^FrDc@ShyA>HB*3x8Z-Q5&z3HnSd~j$zc7z5eX$jyoTG2u=Tt}x#5w)sDAQi5utnzr zA*DUS!u#h@@2~D)OJzPWLgxCbcoI8B2AVxumydWHuh7+&)Qr-=^x|2Ua^H7u9Dtg4 zz8@opX7(;1+~pKkZnz_`t$)R+z&vktcWTvlKOPM7o4cRux`0SE-L^$L;m)Y&(;QZI zb$bW>zDJqafOxX`<-4~-&mah@cAdxUELWn~;*?S7`Cx{kmE_*oqS__>?VcdJg-z2L z?Kw68YzqAehjTI?wkS*+x+7=tSr$p`iRH=7eQ%*Q8&+Y}F35MfBito5%3zcZ&Hesy z*za6YrlDnDODV~i7Zh0H*k$uVc=EG_<2E=E{e{2&{+IAW^d}pLVO3Mx%LMxft5NIB zSBtDiBS4%}(;~i4t;nwjfIXJ6E5*e&ewHp3#tHFh;}3&@cZRtu62a6y@Z!rbigj)jsU8&%1c z12Uuq6!aBBr2W^||JnBoq$DWROx&Th|Gc9LNHzU3n~#z|K6MWbZ+G`OQB&brLuE>h zV2vEc=`t4&Q^Hbt!Mzq{VZ12T%_zy+>K;2b%iq$UX#}G6KL2aizr6u-{9fL}_;eK$ zg(0=QsX&DG>K84<&1~D(&EAs~ES@*S{{-iD&y>|WGiNm{>UTZl+`u%Ccmg|%mJ)XH|2WxKgjE)Jm<98?*$ z(W|f_zI&OZA}Jb|-0z2RIrIT=9|TxrrC`C~7^Ne_9xz&8Nj0Ciy6RPY67-UeQ4|Q1 zme7NV(XCcnypwX1vOBuF_!$yQyNeU7^H>n`~Cet!NtxJ5E^>(riOk)17eBa>`%fr7KB* z_3>wB951H^4q>0ft(?R#SBlLa^RspK%vDA7gt7xfA_S6zBNw2wjNWCqQ56%GfRTQvSSyOx*TyhNOj8RY&b4$Fj-76St@tR z^kwbFy831Lhh>q0rLAG;9FxRRTkkcVA6-yY5G)S&9%6<6%SZLk6MuHvHQ{1Y+Z6_? zC^kOT&U`V3)|YZgXdZ*;hVQnqyV)}J71~NIM%0Kv-*T01Nxl_1x(C$?iLD!Dvf1B> zsfM{-U}5S6cWHO0-i^71%N!-|{$;S8r=C5J$0wphKPOW!PaukTSm>9VUr32@GCX+? zU@mk|*O^~(Raxxhi+Q3699TYU7El0FMtIptL-AQ`7EiEOk&@e*jgteSo=$>zzkOxH zd#ha`iE$$K-@x~O>#Uq|JB5s9BhH@w`7ht%f7kqfRdeD{L4973@nq zPIX`0J^;gvNh@=->kQt`3mGGi2w6gAHsOb6)iWc{&AztUkfTc;vxC(q5#1Vx6^~*V zOCGyX0$S`D`ROeP7(e-wN`3pJ0F8#dmdB3pBJEY8O3>?lrlG zoTQ(sl~-MDP`Q!y=E;cLvbRfSsX>@f^GE!^@kJuj!EN|^)Qi&PRh9GtUtfs9M)Pka_4mzeV z4_ck3!=&3I489gT4kX8+0U>6(dEnEmg_u^|G5qRCI8kQlY@eR~0@Njy3&Pbg3p5S| z$WB{{3tE$ScxpVZ`c%2NxVkemzSO(@2y*ME7lz(3x;IXpZ^^T@lS0)+Ra84L*KBO~ zPLtbJxlUA(bu4SrE#{}$sarEJ`#@KGF~3ig^Ru2>R@x>sn@U1DK;FAyV5}`Qx_|5G z7cI|BRtl!@p6@d6-#3vHZ9aY!#_0Tf ze=~z^>iCM*$Hq>eAh&{2(VL7uiI>8_?>9o>k4313dzX0efuGO^d@oX%*hw)DxtSEP zYrar|N=9h7*1lRI6{OI3B5>@iMj7q+4*A?FAK ze1sS9s9-{UzNeWf>ksy&91*MveQSz!f0~G?4t|<^vP%*JdS#58@bk=6`XO+ zL%K%r7C|R%@m&{}l_xA0zNU2LFVBabHJzMct>3K`XbMV);L(+T;(MC1NXn?bm4|## zv6|6TTuDpvdh@wQgh|y(I`ic=3nCgS#V;dS)iLBbzAEPN3{`hRa&|mqIweos(tC1{x8Qj?VkN>#x`?5aY`l@y?1LQubhLF;{k| z+%}d>QdZrZ$@#U2m2^g27G)_T!W6+vGYNKk)jf2sD*=>I{VydhRzH}{5 zWJ}*0q7%2JNCa4m&0!%GoHOv|q0oK6key)f?Sk>OcB{`pEkc*>S>RB{z@ zIe;NAX~?glgJ0C23C&w~ga)r#K~?0f|Dg0rL2eGcO$2Js=dau!leI2He)-an1?R@n z*p8IXBV&pX<4{kL3VpWIK7U1+(38su-3XHm{+C z%`WNDdmXOOJpXf0<${mb=xkyG?tT}mQo0yo% zESW>}2MWKaiN>}qvvYTpW`gR8>Fb%%qnRdWc%y1xZN*XIVLv&>{UV{T*(|W~BAYR6 z;KdI*9baVk+!uLKC5DRW1Z8&4cxnifQ$amJsPT)C4!{eAzmA*#dShHI*fcTK zK+_k;I~LNvzYO=Uj5|A)~O%WwzAunZp7_de%DMN%}?6pox?Z*XJa2* z*yKE|pr4!cc_E;;y!GQMK{j&9?FW*suEN{(khs@sy$$NSnx@8a6WnjW)8rb^u){kV*j`ivDUAj33dXgl#Hu;-uVox(lT%e})+Y>07F;C|=Ed3% z=%y6S%2Xf5h;K`)SGDr!fc!~%)4b|sBKoRfR`o&mXm!7d6iA?=+_HuWa!T6xO|@bn ziqpXhQ<&ImKHqcHJ>q(G3pIxu<|oLO-dQuL;8*(Wdd-Kdr;c`V;FlY%{Ns1h_IED) zrD#eqNJH8{*ex>3HGQ)A$%s3X`V)M_;Kll4?yY2r7PKFK6aeo68t^gq*7JXNsjP$=f5>;$%3DwD~Jew?Wl@Bui{KghTBK)=VT@hI?-ce#jxN z4qbx4F}{2!`%=M>a?ji;jK$=X?xscSTrRejNK%Qqn@fh@?AK=+rb#w}&7{KKF3S;i z$YQN!srUV6TY4P~0*Z%x%P6!YD&w@R29TgjyKeHAdg9y4enRvIaLIk zwcs4bGX|*Ieu-n&Wv@Zp!@m)p0T}ggvyvL#dsnpi`T`wZd@{`?>fMGwFx=_wR3jGnG=#(;{ ze^)~#G;Jgp8?2yPn=Y7zzCm5c*ZFSft}z;X*h~w`mHN=irok&4JLHMxZGVMmE0Cv} zt{iW`AVIFM=}#t6x+n$h(n++OuiPiUB;676MpYXfR)u>DY`gF+-kx8~;ZLwT)60EjOnqO(73la__*;N`;qpki zgP>J-g=ovX|A#KHn45`^!c1i3IfzS;|C9TxT~kA9eKykD8`?%X1VW*4pawhI%&1F!kdLwwv85U3sOgKdwMn)&)NCGFcweS>Bj+cKsTsKTAuMZc1{Q z|NNF}8_z~S+VJ>RiHn4qSzCPatN`Y(l!_)i87m0xas-j13`02~$bk1oU zdevHDcRDr<5qdR$R7Kr4@^#sdG(X`grq>-NsoqA!qGz;HyVG>3>*k4ncKwgmIfqsr z6yjZ=jCY`!-9!2$M{z4Rzv`D0zY0f~Iu z>ogmok>aIPj)+~o&0kb>f}mC;b8|LtD8WN*cSQ92aM-cKqcLLsDL)D$z^Dy=#8xXYfybLJnrF5-Mxp1NfOiz5_!|kmQ60Ree07`kyO?| zTNcP>0V0Y+YH750DDL4A<^$P80b&FLNyR=ibKmY70DFIT;4OJ-iz(1{P|K5N=*BNf zrRR#P)3z-!6whIyP%whp`OPTg$-%`Cf*t?`R;tYyHt3Pn(V0P?$e_@>Dh~Oem}e$7 zskbc?W3>c?!#V*FC**qU>%Qif!${q%lI$NWCf3g6U77A~c2^#@A{!V79%C|1`CRpPnvLbv*B+hov~lyhwDm`M65ZET&r z9PhdJ{bfvCm7WFlX6&k(U~r@0EzZl$KBdhHWy-^u=H?@ZSF%SeYG|>9`8%7LXf!~Ytm7XP)MautmloKN)RYjEIUv=ZuRLLf zQzA{KeV0)(UQHcUb%o8Y&VEJrRZ-xrf=YJh5(m)INd=oMv4&zVFMje|-BUU7?~sTt zOcHv}6>pnq(oIU`o&;DAa8Y@3B|~#BJ$gBkpprkLSdF3w(zx%@4bw)VZg^$ytr>5q zkF|4l@oGM)f~cLj`kJ;?&sfgQk&Ip0R^wMJ93QocnO(a`=H%MLa7puy&XGuKAl?Thr`>;HX^J!78A(hyIuxp7=ih9k#G!)FbqaOONl- zv9*~(Gk+cV7MrAuxImo)k}o3adim~YNN0XIy@V7y)tXD0Ac`J|(2$urmX9-*<+YI; zvVWBOr5p_IhoV*_I}^kjB!_Jz9-^{g<4UIbXqx?-aSZKjnR%Cgf!TETXPBy*ZZ_W( zrpV4UVQx&onwT6yLJ;@cBFBrYZ=+VbnB$@gm1~;p@mDQg+#(*rndG45wc66w0Cb%3!KDRIY@u z{T_CJVn?-X*-@>&M-ZDY!d74*QC4!0e`OY$;w1^}wJ z*}1vIhm|zVrR|ggr`t9WrqWw_9;%9gV6k0N-fV14jG7amF>KcQZ3sn5>_DW*&2Zgl z{+u2$W!V2=?*+9^2rQSTs1$E>+DIBi2}AN#t~eI)+gXZQon~T(WpPbDOyowhJeDy# z9ywJ%xS}kTojkk;*doBanWNpj$RQw9I8@l#xZm5vMM}U}AXCd- zOV+*g_vy_!@0)vm2NGs>9$4UZZn;jP*>Q`Wl|SNq;Yw4O?;+@EiN#}z z7Q~YJ85_-|(T;v;qRgbzZ9^tw-%eCm$$*uF2v5?aOdi6eA5TjXL>$fKf`XRk+vt5v z0jg52osLqXuPCu;4&GX2@?3r)Z7U76Cj5Okr|i4(THfNcWh2%-b*vi|In&iW{$0t- zb26@5$_{-nrP>p06448>KrSF`JF1(p9=@D+_6)=mDlHo1q<4kv3Z{lEX}%h9=v2_H zw6^-nMX(vn=SGJWO@s>;do^(PGLa%v6Cvs(qdelv;6+Vn$Z+Dx2`gLP@e_bcT%byO zRb!nY2~sMXSy&};u02&q3IussG

s`w=me8_dwKq7IocqHM6Maq;u=cTKaGq}#vC zd$5Gw_vCO=s>Ub2)1|7qj_VAhmrv?27)CsU1IBT+G7!juBtfP`^IPkf#J437o}V&` z8}yB`(S3uc8|#F(Ip4Ru&}6 zV#i?RTikwJfu+3iJZiS>WpG8`qDvyGJ#U$JLSN54`Ul*698?!d=0V1O&2V(DDL7ro z!H!C@3ri(gAuJ@_Mr#G4lQXK_2_V$I0mOg1Lcc)aDKp%WqT)ZsU}_d@;HwI7)H*!tp{knfhpiui1hzfgG5Z5Kpw9b~W0q~C8<pcHACE2S>o44L*LK+fG1aY<~tMT^CxV!Aj?MbCf>0MirNXx|o>OCFqy7%rR*g%Zw zWn-!BCW?Mj#uNyz(AV0(;*YZPy8r8m1m6)V2X8l;Wf)uZ3D|lj>a-p4 zv5eW*M8};}mn+6xN>j$B?Tp$HyOu2fK1Dy@;ZDG9Z#-3edMl+$Uotm$H+kh`B`RL8 z{nK6ZlwPgPGWhSUJpH58-&?)P_787ktQM9&XWKJ2z#8LIHC+g-@4iG(si0u+8nQ%K zN{xcRyG(>QS@zw6`Yake%60TX#W>133<3YBIK#O*A(B@}vHUo<>3G{dc0b<#)UfTY z##*bReL|P~ScY5Xlg+zE+<|$OVUP|&?5hLQR0>NPA6pC-E%L$5Kr0B?K&fi85JEp4 z$p9&5zoc2_K{Wf$=_y|6AofW03_Ht*czJq5tuAh-bfvbw=dVn$4VkvG=TDI#7#hNt zG{DKqN*aW`xIczL$IP57`Vo(o9ubyhXdT=BDs4YGbZ_@l?Pq6R8eMbg<#*1qTI%VC zcg@dPTv^`Q>ywn=%_FcG5^AR)G^*K;aaqXqbC8~WC7WIVmQGP z+xcz=JU+Duwi7z*kgDMRn9sf2K4qw#^B(J*@!xo}f9d*H6LveR=J_i*FguX**MHmf zZ&&5}&(LNX&(O~P+r9sGV+ULX5w4*Xo2MO$9oWd>W~W%~OV=C`z9nzsTlv-aLi~|U zFC|KmP3On2OhhGr`g_GDPW)h$?tnOmQT1a^Vy|VpG4;;*KAB@|eT?5w(GI;bbgi#} zdsI19LHrcfPJ7#VwEyt7Gvcj0)mXsi(;H^YS?qlo6T$}z%uD)k^5v8syr)(ync>ab zmaKtaHWrV2Ugxphd`1a74NyrDwR;jNBE2iG>aWSUJO+U>1 zSH1}e4fSe0vkI{q8s8^r1RVI5<5l;idP-rt8dVLs-vr)JJ`Bt078wpGn#fkn7A6!2 zR<)6xcpD@x&=A$AWFYoQQBk3z$x~TRG)tp0hvmdkg=|9M@Se(WoW3}i^#~&ri3msBrWFNEIVGpe#{wJ zxZ$hV6@ymRZ8CQds=k6yKqX%MDJx$8qNCMGR?174vz<1Gob670=5$PAo(mAz6?1u}q@ z_4W5Q`+dhHqI0^6` zxH{5PqikI2n>Xa*8?;o`g-X}A7%1&7T2fp*c44Qe+SKo4MwSb8Q0@5 z*e*Anr0C>EfN0O6aB4T6QrC-1>tVnEMLxwGuf{P>d-0|4StgP~F=Dd?qYj*m) zo*u}~ic?HReIA$5);8@5D}sokimG$8?H=~+WL$v|q%s==NeMx7K29^PPL!#-MmVC< zd5T9^CQK z@6VjyD~_3Pc4HFjs)zTDo|`_kFifk0%{cP5CSBr`9ukR$)zvuq0ffTw5wwt-4VkASfFA!X8_$G=Zc77s0gZNh)`Jrt@ zu$YvD0KDRjpBxMyJzAztgu+0YL784@p4r*t&uwq%RvI9Z8L#7%_oE^c3BeF`BN0P} zl8}uh35cMAB^hiAI>pa4ZcBUIrQEcbOWOAbi0_{i6EI-H``yvXQ51vPzNl=&tt%Ny zX@6rOr9RwyX~A@-3|OtKxLIMj7`H=f<#+J?yfU3ps9T7Pa<37BsLed9Nlxja<`F4% zTcCh~v{PvVfZ59=y+9>UW&6<(UPx5lxlv0JLgeq|ZTvII2i=3eUiarr)^-JrE9yso zk3bA1eLZtma@V|`h3BZtmAZGgAv3`nO@}o;M4= zR$E@yKz8NZHcU=GD9GuP>5}Tc(a%;Twb`t(oj_`oV!}XC(T!y+&RhyKu8J#6n9?!t z+W5pUuIHMv{^CL#}Y*j2~_4upnHFi*$4w<5kj_a8zi6-AZzxn z$U&R!=U#6MAtmHzOsya@lKT`*ziq?-tGL4RHa5NW&jHDUE)#cb{TRTTYsEmYQn$b1 zl~anrbDy!6H*k>Dg5tfJGtpLoS~+ZQNVPfyqLb3Mh%kg&RoM;^Znk((xgZwJ=|I2K zmh^IDx)l6Oij-3Dk{Lgi`VFWVfmihPP1hUtyJat2-Y+$uECK3@GD??ol-_D^*5nwf z^oyXLa24_e5f`#I>w!6 zjl9(nAHC+dmEYw?QDy+%F2teQGg(WL3eCGP&0^ba()gGt1OX$AlOxc>FFk_hdJt7Y zbd8rlxtj1m>}Lk_Z4qq@R9qz>4|v*Am!>}?A@484DIm1%7+tFSlY=*TY_u(m4=8m(fZkP-lA#iQ3T)XKJ zt#|s#U;5J3QkIU+4;f=xf%xbjg@ssH@uSKoU=GITK@yY zv_z2U;ciX?8%g&lbitwPZEP0)oOKq&p}WXwQDQ)oHA-NcE9)8;gAw`*97q2Ad({7* z7gvA{aEhrVntZSu!Z?uYP>EVH|}fJ+taiaC_3T*vdUAUkU_l~c4DF{ zK}gL~g+)L!N5A6@N#Ar^BK)CgTrSEAC?fdQvWoArT6E?t7eNtSJfQWQ-yQAg1m>}< zkFE4c38iWr&OZF_iaczHI1*gXhN+IX1xZ+ExsQJ9ePdCfps)Lg5Mg(OC*IHW46CH%t$)8mD|1>toXBwBx_Goi( z0R_>KDgi{g zR0~aNgb*TKq$ku+LT7XYBnS*3AT3A@Nl1_qAXKFq5HN($i}c=moo~(F=UneT=iTQ! z@46sa>se1$Un@_Z`+xubWbYI)6HKleG)dRrDayC-1YQTwPe%XWq@=HrgsRAXf>W+ z8M93Ft(E{XBy@OWdegPGOC-&KLz0%>fI5W9)# z$Axlk%MU*`a2ZaIEz#`MoZ%Yv(41-P3L%zksb{|)bMWU;%x$ma0fi3NXf7rbtm=#O zvd$!(x+$$QDI8P|wFG}KNUsp-vNWnEE6rd^9jj)uL=Il4Zit~OFRcJXP^mhsR7<~hB4vZBXDE>ZlDZV!{aVgbB+Jl-=uRS@B6 zqBVmm*3fnmp4N11xGY?>;pOJ=%Bf)4u9hl-3wPteC>6C=d9c=T?A(#~h?#Ne?+hms281Nbx!jMKZU#I)~3kp-Y;u#^BS&f2IUHz$|uP2tykx&Vvg<7q<1oQPaq!Y zqlT?^P(?T?JY+v7Ct-ch;B7f`tln9or`F1`#=z{Hn?~0p3`e~u^&`eGf+mXATbGix zYJZKzj4Ef&DS+`tMgoh4M8;hopA{lRPu33Kmd!hx4PL4gLAikM7)d85zR+Q=j7v~ zh~s~;40YQw;-D5jH)XyL**UbC)_f1qW)(Min8sWSL}#+;S>1|_+2u7UN;mn&bPKUZ zVA_R?H}#KKgL`9#+?KOrK<4DsKx)9rNvOlp&EVYZhi5&cI@=R|y$zFStXnNCDh#aG zu0sLF;b=ZaYNXxWSQAb5iFW?tNICf&pbwbMg+fvpbT|4M&6xEQOr@m#yzyuj+=0Z$ zmI|%g)o}+)7y68lO^faOsAEh8bDnrgzZi+_2{o2PG2r3%LH{hqSX1W;OZ3yft@O+M z$_^(~Xwc407zFE@i9?s;#nR6p3Uqy8;kRm8SF2WOWxab&{jegbYAtBX{5p6OW=Q^8K0oF%3We)UYH5GoD4k=3h)&Jy^X0br2sDmcwegm9JFEwUGa5K@NU2S4XHMM zg>cK-ny*P;w%3YMw@E=S;Khz{mQ-G>nbQ2K7AIzkp1WP|ef%xjY%&wh! zdAOh*A25|(EF+bamB!;=n8R<1u1fG--!0QbYN9pD4S>dSFJa|gyIk=7=M_3Z5)J*5 zT~dyzVO6+`w-?3sJhw9_N&fD>-!>)ZB$w0>fEgHJpOC@112OqO&*py}ouwjQZ>5B~rYNO#Sy^%Ca!ciCOBMZ4;PWSt&N*SEr<Ka%GkNfLZun;ngEj>xvH^ZcRp?-w~lIKP>-U{RVYudQzYKYzJIsS`1Q6};%EfHY4Z67g6+1upLp7>UYZPT z$2}o`X&ptdKZ3cl;}V1-;8NQ?Zow*zFL&TS_d7X9f} zdATr5-?r-H&ANN6Y8q>FXhiJto>jvy?!I!;)K&Bumw`F(+}%RrA8u@$oj81!M6L(v z?tViGISG-cv7xbT#Qv_@?}E+{O-_BJ*HTS;>b8I(Qx{1TT@{7G3`8U!qkv>7dZT8g zA%7ot^s#tC@`;43grQIGl~Tc%M<@)gYsoXAKDv{lnHCEql!tJ2^BVKaGuonRlS0ov zsf56Kx`i5tjb^{O-(q@iG-FvK$Hoav7Z`Yc+LbgNbtd59kk-h%py=GLD$r(A(x%)dLs;UPoGd9*UxKho@RUx4HRae}s(c z*Wsy713wr{(pEf3oMEgA*MzwDR#)fn`t+qrCd0wfPJMDlcbtL}9GcP!GY+TXZzhkK zBU?Z95j|rnKe*p#K+7_GI=5An%954#q8*D{)WL3*JZU#FKCJMZ3U94?!&PEnr$Cy| z_v)bqx1Q<*^I=~Jz1*nH9D|!NhRe!#t6JGk&Mz%QrwRCyM~9ZMbqKfiMHl#R1VLey z7w4qAWFKd)4TWmE=jT<^ZJC$zJTA$JqXp5df#hyuauR_Cie@TG)?G|+8(_%moooL? zeIa0lDfjL9sH)|Hzn0l-oM_c|9|7sm=`}i7|di%k0RY+MEe&|ssSc; zLRe8@Mq1REi(LH&>H;6!lTqoM46YW<5L9)`zAHcRTVAC)P9!hk!74X+#FP1SjWDR3 z6s-K`Wl<0Lq0x#qXp3!4_oXeh%tW!t9n|s+$H-%o?Y)#0?|Xq=E%1x@@yls-QG&(- zNidKiS(&=2(p>lGGtiLsJpWVtn=>rkzn%Xn;}XZG&oZy#T+Xn3{e1nr=Y3$--(0ny zes_xh9jI#lq8(qxR14>B2$Ct3FI{ef z5s~6dL%<*pT^J zz4mWpN5EOU8LxQL@M-Pulfd}@KAcH>*0>|hy#J3A@V{68M-5=7y@6i+sAZ^brok8% z(u%A7hx$kcGGpH3;tpxa`Eyj2p4T^qjehn65mV zUEu(2!TrS9pU<%VkNGBQ!6EF$nIJ?DPuh})TS!HVCo$hU2`_%1nErmXo!52ciebUw zT<(r)X%d*s#hWC4@i_1Os)p?I&)2M|&SGi{3bwP1(q#M^E zbtO4d0*=6TBXb>8*mNrtq6Ruf3JDp^pl&WoI|lQn-vAl)w+**^5poVd`3g?k=-px} zlv5WBp|pHL6LKpSIacPe6C%c?4X6<_yCSqy)#gOA$(8-j2>A|dFqwv#J1CLvbB?2Z zy(SE~Z4iia9t*nMrJ|J3;kUo2*2m+KpY^QG_5otTo}{_HotWLgGt@1y=iG%k@f)PT zG-q0S_qP;tEem5{8f~&LfD_l9X1YGQ9g$V69XJqf3)>R|vFqT=*-!!^j1)cn=st7Z zymL{H zALp1>+G1m%H}~!(tjA7NRcV7?q$w>p#zZdGm4p;ZdRto>XYxfcQg6vXPKGcq!R!SR zo?P!HG;gn)&fWGTL>eLtg{|ku^CdH{Wq(o>b+TTS-_80og8RV>*@eF959qbO zV7@s7_(G?QyyJReuP4LMxk-lF!^Zth7Sk*KAM;}sGd=1b9oWR`Yg3Nn!w3!6O^X=s~lUWi+MM;DFER;PI0E^W(m>PMxMDK!$}JV@{J`-I}y# z%7u!4-3j#a4xFV)(b|<|OT?&|jjF|mJ|!2>Vqd{k5L_kp(|O@X|L*qg2m(Bp!m2KN zJ_8WO7as?S-ahOluSL(-j6L^j(4%cVt~vGK4-3%UJfe~Jut?}3QKhwPCt=Ir!hxkt zj|54Y5Z@NZuBoTlULbcoIlwSHHVLCED}CKZ`%y~P#V>+qIyKisN)#$4y34hf5vc&H4sO5@$WVp zpxaa=*f`53=hnCk5Q29zT{cvu1+p%};L)N?4!(VXoxsu~6yAQ%qQ%XAWyeOPE1r9x zkoYd>Z&;&3rlOZyNJ2eAk$#^X6QHO9QUV9j>{F~1iUwxKrZ8>6Xh3aiBsypF>x~`w zR=n5p+18Z^Y_DuT2!nNTz!A{7gHreVr|P(?{#3g@RG%R+GcHhS%JTc1 z$+8-2Av{(A=lx-Ox}j)rz!#p{HaNoXt4kf!$ykkAmAsbMZ@*J2MziT}_Qr!P0wWlxCj8*?-1 zYECTbLCaBL2X~kWYre^#5Ksw!^vbvCmqA}~6F0BOnYre?PQgCFjw#^BD;7qP_`g_g zXuFcq*bi{W`Ud;OUS>&EHfHWB9-Td#$NY1pxgOVwODqv<+*P+N+lj$RvbwG7hsIut ztrM~pnG{_zv${Ts$V;UUr6qo?Z?Vjg3w3(f(XSbguk9M;_EDi+t0(^*l-}16Pj2E*`HZ8 zWTG-b{Ilb)3YfzsWQq_seFu&gb^Dqjs;QCr=KA`CyYyrK_h7=#`ESG|=nbhv7NPg3 zk1*QrKV7z6Rtbqqlu+(JKxTAoT81{Y6$bG-0I2g0zI=|8I}i7Jdp0g-JF=Oqyk;B= z9EPLjl9G#T9?zrzNT(JjBJ7$1zTEaGBNn!%*|h2wsF3%$u*Cw2<7q%okxs%*aZV;0fZfdmrXC z><}|ja(e{XRbd<9>q$t*5n<9*>ghf+fQ|i=0k^1mY)){jjsx0TZ8EX=^!c3C6{2m= zmU5rho&k6o5HyDhW$|UxXcl`nPcdywQKhvR8w*v~A3mj$CN`;^vgIx#4bmNlVq%LY zZS;z)PJuqNSl^l;;_WZ~^nl7W#ZcC^IQW`)b9t!8h__q5*WyWY;NWVU)iLpG4}FfQ z7L^ZJBJgk_%nkw@ys~nncD&bbl?GvJ4q!_KwnrCQ-6VMm%rRNldO~npp;K}6n{`j@ zEL0ghgHWb9%Fp{H;fM{Uc}`y}rFEIxK{HEqk%v$tU>V)qdvIf^XtmqG@dWmBPGz5_ z`YU@&pAqR9WkF4dEjM+6_5M~tBb*p#kcWtfafGpd@?<1y5qm| z$mm;DK7@|o3u>Mb+B`nOMR=4ic3SkOO{M*0>3Yu6e-Su7!{uJHuip6EzBR>Fosy8={J3xornG~rV^gVTL82vvX^ zOy|5m^oD*`_UiY3v4UosW#1&(-4wW1RvZC_6wpB3gTXkOZXQDp~Rm3gjzXky>R3QL&^nK_@_E~L19DOU>ZH?ja1=glg7j~klv z(O8=etfZapISw2Q=kHio7F5A3^Jb7aCJC%&Afa< zgv|-jOnv(=0`|kBbo?7qdV;O$y|*Zb!^IIv?9;WtY36Hs0)4eO#kzAjFlU+Bk#Cy90qyou=53HMs0>zTa%oM3+@-Q;bFOb@_vPdtF9(@#05PRi+l5l$#`9ReknYjWFH;vZm?! z&h(7vjkW%$+~w)2Ud1e@7|-CTI;+Vj+t|71UzztwuB#etkx;H+dSb5dC1_m8sK{Zv zmo$Mug`ThZPto~*yVn2q@^rp6Rth-Wq`3W3SXdPS*$}m{Ndl4g64|()l?5owH(M9^ zs)s_1K|$7#xv>C}wkiy3XK}kvuhSEp_szu8`$->zHJTHdMF=Y39t{2Z{ z4*^)E*;2K0QI0k`A2yy04ufT>4aOEJB+azv_24$nlVnXDvp7!4%OcY&6yHuYrpV=P z5%!V;Am{i4^2EyF2Qp9ZTx$WhFqa~h3AqH$?z5f>s`Ba3AxoNc^eHV^oJ?upaDk5h zZI}LI@=^|2pGAnr?)gYjZ&zg}r-GP%^!^Y}1#xPxvW~c-9$M|*Lby&*5+;c*!D@}o zkoFp={auKq0AsM1%#K^<*fkQ-Ps-gp#5X>sELG|DpRH{_`|31K+%+{P{mu z^#AnE&Md7j9rb)&9g(9O_(4#IKPfs=ibj8yzIJ{F=B`eXOHsCxn&(4v zs)`fwrHtL^q{9Wa3WkmZ%%p3=-qo;vXY~wD+d$RI#HW|8<^sfc??@wd!pEsT@?6ITA%~*mQhR5*^iNo{A+9u{uBA zI)7$yP=3d&eC`YpG4x@ZhiWYRQXoJR=R7*fsL&C+m8i?qLJKP;^CdLBK?=l#jHsMW z1yodP<(&BEQ;gacp5l`LD+*s!Vy8uQjW0#6^tsQA;*iEs^2_TrH&u2R&aY1@FMPb(uM(LFU+Jx8t@;DJ!*XNAiKPpz!ZbrWGyvNtZA{P zJBai5eq)-RbnCWBW%a)1& zV3%cZaZ~}_Q`sJ^x{1+6R9TZusfx;3ru*3ilYUZI>kHXF^%XrqM7wC+RhgH$^{rj3 z7pntB1#03~S8b>v;a1fafYC5_iqWWOvM&74WGDX6NOznQh5DKEe|iLZLri`E|LkJD zUy>GI{7<$1wVYl5ex16$@SvEEN^ZY=C^jn6qeHiR;!qOYzh}Sq#p}*NIVI;TBO;Y1xB3w*|SM# za;ZbI7N*cc^m?$FODCLjzl-UyMQnr?=_*Tx)N;9LNg$Y<1 z`&ECWh^wK=-UG(?UW2jDnUzD`*zr6GLwzCDQcfiEp_1k-L8Fk279%-jjJShfC@PAa z21Co-GQ`;Aey?CvJ%fALz9v!K-aSpCZ^(53A(my_d_uKBYlyP3s`ATR%4VY;zj*1< zCh^@7<{`x4z5ywdWBiDnD~G{B`@*#UFm}fS$vPafN6UQzjB5Qrc9jhH&*{kMc4mzZ2!lC8!mr{ z_FjJJEM&8%w7}VYPbum2FkX3_uofUq-%z?OLb@KSFV>Si&MtSSvb@P^O=ZlZcW_oN zvA)#WsXM^l`Jk$BO8>;T@`*zZp^Qfgrg}>o4dtcJ%Yi0yn57Rliu89(?sba{)tOL{jzH(ZCW|W1({z57iOY5IT#5f*K9 z6&4_=@pEokS`H_o^K8k6utjr={&9EmMQlHHYw0M#r7ldz9begB>;y47C-ySo)zYFO zW{hNr!MK!%V)GfGy`$6oJWMs89CVi61I{TeAVwLH(9w!w?^%Zsh_~kb24_oh>tNm> znN`}`iDa!hdh4FNYzLEv5CZ|N-xKp&>>!al8Aq=kyB3zg74Jm#_P&eIy&x%EEg~YT z0n^XfQ5$x02uXE4ZY*|-c*0XW3XLuBqyRbgbITYR9X%R13 zmxFbiZhB`>VUs`q?tQag%#x|8*_eY9<7MRXw=Ju9(H* zTi^>pclca(?MyK>R5Zr$n19O{wLOqYe5qLMSKH9iS)157w1y_e)w1@xRYb#)4@PZ3 z$`Z-W)NT~ny6ZY)tWKxXw2H0_aY^o|d>6Y{Xw_bDAoGx$w;HS!-G57v*QYY_x%Aw7)>OB5n5z>ZOODoIj=xWklbY7)cz9q#MT4yDgpnMON5Up7Zu_@a4 z-u7%bWR!>r7^}i$hc;(91asr~ulU_vv@b}Bilzl9MlVuTKrG8P67Y+v$a!~@1cr%J zK*`F9-WZbR>qpQ%QAr+tBSC(e-lV%g?J!tI9uiffzsKZ*poPgLzXdXK_qp@~RjfQO zm|Z-hVNvto>Z*qVzvwR&pBD&IPOLk#7J+^#SLLK=S9+P#m>ydap@EHC?OR>4)fo~6 z&MH{WHCLK>GQTHsSu%ABlbWTn2@$Kxi*MhP#1Lk}%$6?(9kZ4)G_iiH&(m7erId_48%xt!0+yPp1(8^e4hJ8dLlYT6$VM7l2{hVL*r!px<7`XL-juQ? zL`Ro-L5wIe$&5bIsi@p9YiXpWRhb^J`P;}r=1uuYqBq?X za2a0QJknUKybi-Fk({Zl^!&z>B7SGA!A0lRiTjG_57RFQ3RDQEXHmz>8*ClznPv7h z6W)+&Qv8gMlOn(8zgnStmT{PaVT0nV=Bs1NdV<}6%=^SXk8gzEZ@nQ5xm~V$FIl@H zoD;#?7HIt3{qm0lxiu|nV74KWjTi5eX@{5G4J3$A+A|<8*OmF^3v$M-V2|N^Rv`Ue z%2l$Npd_H*ZM$nZq}MKH1RV~2)0bQmN-nRv8t33@V0N{>4eY!7>Zxro=adKq;Q{b1~-A`=Z|)_ka5R(vSa4J#%7t@z19^V5WAR1$BtyNb$V+`0s|2 z%qu`109<06mssutRn9y(!vfU#>GywD2ZaCs8ZAHb;5u-@DBf?}Q8umSsfmHQ+|354 zH6SC>(%1c7by=Bu(GRmvnN9JuaLqH*KW{DqO98&ZsSf5QHO3O?YdwD7R_Jx6YJ!tT zrRP`a*>d5N1C?U#nN=`jelNeGIJ(J(ezA-mTj_+#RukXNwS-KDs~q$nx{+2MtlN^xdZVlFP(q(TOWEl{J& zxGr8fYX3;@T_GoWQEkhJ`DA`jNVdp_q^*TvZpR0L{j)b3r_`Ey;5@&MR;9lvc|-#X zGB#)&{@y7$fhr}dXS!oc#w>IxhLk=*34LN8x2@WgRrsnpJt#!4ZfB^3(r@OGr4D_% z82v`h5Ek*2w{!Vve`0CIrzobbc{S6j(ANNZ>MPX(d!6N#INiD0d&$B6aHu+Q7Ugk3`AR|g}NP)s^R*>@uq4t(&KC$avFKstO>ma96TN{z*ir0jkKo3dA7F2c)?-bfhnKhn81 zIKPh@F<%P(7~Y|08=F&5Ep8X{o26$k>~iKXT8`jBp=M&cC6{p zW|MP(S>*Y6mW2~Ve%{h(6DK$U;fpn+gD3LjQfr2V{Mn$4Th?vP{CG=P_Jv8rl=VK;oUb!k8euESFDAae_*%D16hv%E;V+g&f%%U- z^JHKy{R^E2AZh)fG|>!N;UaHm*^`+L_;{=?_U=bt@gb>?9Dvd zD!V>67Y>}zM*9^bi+ggY3MF%%R4l-iiwDnyYG=O~tHYHlb4h%P&?*Vx#J>h<<0zw$ zOn8!N`!87#+_rp1xkf#qiR&hox$Sw*!C{CE?d{>b^u%J)W=72ZX<7qGM~;gopNW?^ zGBK86O?^Fhk&tIYEkwB2%#oLt#-A^yT_2m7X?ogXdhF>eS(8dGeo7+s?t!9c;TXk? z)=8Y>0dJ}uQW zojc6`NxV3|z4YlEfefIMsLMHbefgxsm#w;xWX5Qn5)NBespEh#iu6p)(xXwpr_CgdgnB6v|6FBSJ!kL!Dj{UsI zIojbbtJeC}l@Dj*0JGfNSx!hG<#)8_gy$snA4sM1!!Q2x$^O5;OJ~krs;s^2!Dw6z z3I8r`N^p$nIv^yqtf1odlIupVri*^$0U)_!t;HiS?y8>mC|dym?*7Cge7(XljI zm%B+dS+uN?^mdS+Fq;b+YRxMgDK-p_y%+3lp=c(xJ|=8(Z;4Tzg7RfkPGs$PSBO7( z2^ibVzzc{zyW3-ZIU@2G-*lK6?ifPs7*zz$6;pJPzS8&4EdN97>c1Wa=@&ma2T1hf zo9;aaxYe&=sqinn-692*y7(8vhq7H<6%}Sy>+{dXw0my`wr}}QbkleHb06Y_&9e%r z!Py8ibpyLnIa(V;rv>bd$&53#fCi3+um?SqfUt@KlMIlYhiN zISpQiUsJW-@um{U`a`;IRx<_?50EucKpSx2zi!%pnff^(&t=qzqSRuM!KhA#3o9p^ zs4w97(~c-;G0;V2^FyUs(~9%LHsY4x*hVzLLGh}X!|sfGg{51$uNa?1%R2|?{pcY9 ztLChL_3{y-8*BqUOmN7@hkM^FsbNPPeIA}akqfFt=qiV>RpWcAO<>kE+UH(NVv-k) zgU{sC8bbyOMGmW$80SwnrhBH=Db;FD8Kkg5L7^?YYZ2WbRgu^Ku$%57NYC>20zx zt@n#HXgqXtfB?MoW@jk0q0_N2t~6bWvJ#JFjn&jVW5Jd)y@Mcu^u4H^i1z)AJ_Ybw z4Dp>^s_l39Dt2Hz(XermdP7={rcq?JK=?4NsfhspJ$S3@;L$dgi^}%99i+KfOt9dm z7RXIE6#rZn>zzGDP+JvG=pd+mJlMLG%kxp$Bbp6ss9gpIzjY@0XlQ=LoJq>Blx7-1 z)!8(C�Qaj5V8G$Nq|4ovc$%QI4_dnow<=FcLDbQfF><^3Qamc3{u~medvZp~cSD z+WZDXi*cE&UlUCfr8w`_ts3)WMD@{ElkvkAZP@1srb`|wpp%ywB+B>#rH0_YT}<~? z`7qa$9)R#AJ6cNiz?8d%lO0=lWOC#RSMe+0!rin$Is3FYxJ!+!SZDhonZ9yLH1gh3 zI5iq>fl&R91UWB}ac)X#<{v~t&1sfzGbA=VBm?wv*E9aSn1C7obZA>g>H+kgAeYWN zc~PQ2VC!%&fe?H-SD|a%*1O>+`EkR6v5ZueZ{|@(D&8HbdSYdhUO(HA(IVzZon!;e zLNUbPj_nGlL`vmqVbzyw?O~t(sl05rtBqgiXmUy7kLrA_Epld9B>&RFkKN)W0mo+- zLCHL#j(QXh3%$m-WyWIF;Jwy5kZNZ0IOI{rFCXS!SJ5xZzF**-32cppX-Uc&W?p1? zOq4)85|ZaQqXn@=^qOnxtA-ftRy?VEtt_ReVynzMs6``{O zcl0f$!J*o04sx#2rj5I_%O-|&jvhmjjqN05(S1J0j#ELlq~h)i7B~4b5ALH~%Q;3? zrRbkY$Z@Q@CBKNjK)T0A+2*vIBfiQmt^`Gy{hb)g80+apbMCs#9v-ii*o~^I;_c00 zlX#vY%YASI*kjka?R+9#K^0;naFL>G{^QN1A84q}j3-s-Xo zoww!9BGS+-a~RfiwY=THpAlroC+tGuYYa>-0I3XKXF_ffLVRFj2@!NF2|`+aGQm+y z?Bj~eToGzwa8DSd+N`wa8UEhmb>{ru(^|DSjSZ1u^y8aKTK?0HW@<*XI5VlE&hg7F zinA}x(~jdK1e{vC+1oW*O%3E0ZY>kE1;1j(odo5djaFD^w$I>dO*YUHPJ*Wk@SaDz zR$$V*N}7?3Lzv^hV2Xf|;rO-YZA_qxXx7N!ji#(rT=D}Cx#TXzvGm3-#G2*zm{|9u zD#foN5Hr~)OLfYH$K)lsoEJ>3k4~G&JOxzNo--pwzL%5zY$0?Tyi*@xr_a=sqpxig( zMFWcW-U_MKR21pj6gUqMs&s-B{8-o6)i4E(I9&w7(JNn2&aL}iqaku@RjhW;BI2Pt zjV;pDS!vV&;##*da_u{lDu8OP%lLCHLA9z|w#yTKR9Rr#6F6Jl{d7SvH)h1Ar{>^k z3b-Mi|IAo_7pD{XT6SNK%(*Qyf^whb`DMiwS>wl_FH z{vl-Rx8XFBY|G*>5;NghX=2RiJ=N@yvS<#Lo!y622C5O7@|S%)tq^d@;vnf+NsLHX z^FWV<|6rC*`vc=T#mbtn(h5N)Z@#!f-J@N{SQ+rjW>EFw#iD+_P9G*EQ4zM;zFbUI;TS*~%r1dn0N;oJtn~K_ zf>2xMx=}Zfcp+tE>27|^AaWwL)8v5X`#F-^c#Lbj(7kMnP+V>sYl}~T4)?7#Uv&re zib1_$`1&7iV3PC>dsS#tJa@@vAcP;c=}9ng%33l9 zp@SyvhOlFW#u2D>HlTx7{ssNZ9U*R;`RHvoAb+Ob>-3^ufYtn3KC4Itupu*dk~WNhWuBWj z=5pNw=D&aV7mES$c^3jNrR!^8F&f{^~dcn*e_o)7(mjrY#@WRvh$GuZ7dH)^6YM%Qe*bi{-*n8f)$L zOjE)~ZGsO>ElQe~|w; zk-M;vYCx=<%Urz705_-*|6);0-|p?k!T<_U=ON-;*)4{JRo~Hth3s8}Y)TmJiStrR|-^4E00$_7)EnD z)W2B7vbK9KmhE$vhpq4_>=rtt3pvB)3KtqEPZG8c*S3Hf2KxEF=!5CMSPDK`mn>T) zjryue?iQj>e|Bv+g6N$PI3ZbmVVT`0IjLKRK`rz@N?fY2%8kNsau|6*Cu zl-zx5Wb@Q?57w2{54{y;a~kk)7jtgy`0T(2fTflNj>?+>BKK)8G|V}iGlg~gFP2!y zNzR9@!4F|2*=l&+m+`j2gR6`)fp z&xs%)oIzm=SLS<{cVY$G5f9*p^U%-`i%~eRC|e|$smRoC=zk`mDBt?EZT&=%2e;VLml{*9lLU6}t}QFaQvjKHvwAI7A81 z;yy^(?hRlkhCnn^0GJ5-a`-*AiL_6Yo3b5i+H0;z%azQjp`9JDkbn~l$ z@Q2NvwJC_C`E@{WrwzbI2r7%XgM%>Oa2r4lnj7*yntKepsg0%dg{IvM`y*2&(}Q#( z2xj67PeUO+&yqm5yJ?5I=ki9WjJmzn)3h6>>^^gcz6}N8@Hx4E%ya5OJj_bmet%KT zs*dq)ZhH!T_|-G+iOU=Gl%!=tUbw>?N0_q6!u6H>Q{TT>?n^k%tMcsSn(oiyVxK8! zt*pql4=tTMk*O(b#GCB?GYp;@9>27t*s*b@CkSLeobES2v?#g+7RzkL-*j&UdC{MJ z2(Y2Mo(SPLB_2BA?W1Ma>0;&pJS;>q5=IfAje#)#^6l!Jve`7Y(e@rg*j6q?^?#?f7z3GMO9us8V7`!MEJ6 zFh5f3MBw>QfwZ|f5zVo%DB#DP#cacf3h9xu3=cH4*;-ui##0_!7l#;ArLhJ6pad~5 z`qi$v4Y1g~YG$9g$oC-%Ur%iOLU;r?Qm*0bJa+S;cy}_~Kw#wH76E_(fC+U=D?c|? zxv^6s%{%SfFvJp_Bn$ZGdXE9{jL?db=Tw$)h+kjwQvLD5fcR$E)zabI@+Rqf4@SPm zlTmgpUzhzi!8(=P=A8w@IHBHt1@rg^!Q&N{VxjHDFv!dJLWIz3KggxH*yqffZ6%&0 zllj8B&*o_lpSss+6%>&34`23uUAzV>E=t|dJmX)77E38molOtiIj0)BXiYZd=@-1F zO7hw6HWlpF%26CCVSdeXoFnW=blqy53af@SH_$pw_9~}|f9Ml_H{30tnUUpIDn`9Y zX$V%pW7}>v-ZGxint}3$tsz7o^rba9uPKe?KBH+8RAv$@OSDV?d2I-8uqCbeUSiYm}jZ8;kt)Zz+lqJK_ps+JM zT|uKOt)EtD8W+|K?d6zPCs)V8sztu8X&P%xB8s0-DlL~t%-)?|kT(|zxdGI$2YL z&sh)ff}CnK9|t-$-XyOxq0@|i^R@m~RL3sgp6EF|6>g%G&FJfGr$|nv+VB@U*WcTK zk~rfqS~T7fK*HOJx(5O&-Q{Xa%78=$x<7l`i=2w!QoM?KTR3u$O?Y}Zi!1tj|G60R zp=k4xn|0g~+M(Nb;L71{U6h>^9z20?dr41>*<}-mPa`Jz!uaI)3@m#XJvn(-se`W0 zbw(;I$qOqSg^Sy{a_%XqJEe^7K@$d@SUZ~@CiO z08o3>59}}J%c?%J82VXwt)@E^*)~{AZM`jx!56^S^S&K{tEK_FLJyUX+q0i)+q4lR zUu<;=D3M{Ak(nST&?qQrcwFFFFY>C(^Xon~E~X1PMEyI2`0BSlbb9S!>=d8oY&D)2 z#xIa1^Qk^z&;GlPKvt{>wPSLI_NATtHy|9IM)+_h*^{b&{lD1c{_@nPN75O)sq!Y# z&@M^^1r^bh4kk)MeDpbjj2$enZ$u151Zj$tvR_7Sp`ka&Jm9LK9o7?I^=TT>YD$0l zbn_OA>Ns(%*V*+!^!px&4IiM3P=S@XtQ1slQu(z##L-g$Dz3RGwzI zBK()P80xbJ6Iz^A<+B~)@`#{%YF&N=xIpYpzB5|}+2Kk_?P~)PfPQ1yF4kWg63f(i zma*Pq?T}hPj%v9{N(_glDRSrRNq)mW8-7kM&dV%9ol2^nwASZdvM(0CG_XIhlsdnB zG?C4lzskI7jEMFIxh0jNrZf&?0mHbCv_1Eoh^XR_d)cM19!m6^^#6~w_l|06d)G$U z+p-l!1q773H&waE=E9B+PP!q}<#bnT95Mh= z7M%}Zb+G?bg!@R3+IvYGwG48jj$kkjEecg=_QrUSyU9o^#!KhQ$)ovx0EVlzHCXi| z_dRd_Bv0L#_yOFAseJ^29~C>{CdxEPTII5NB&o#^ZNzxK7yPh=@?GWN9YO%^El*)& zhU&YY+MYf5y2jWn94nyP+@YYI@G!%M(+`Y0yN%RpB=gwd8>a+4Yyq$DQGO;MZSJj$ z_2KUNrJ;AW?g4?fs2e9v#}S1|)Ji6g4Yio6i6!n|U3ZZeNNUABw>PFuIU z=Ph_K;kG}}sn2?0i)=LEz*qTM^hOdBQb zhWIEZE}G;VmLBq0_`8yKGUn=_BImgC2oN-;UL-p%tah&9W)R4s*+~^Mm29x+!*TJ+ z*VW(WnVlA7cvRg^#_FOa#GH3!Z|5kGH9REmH@uj9v)#vjx>6ie{ksn+pYu0+dv_k> zu$gpBfd(I`bxLpKd3-?1_9Gh5ruRk}@X&JYZ-x@Jp6~U{Ime8L&Fk+btkRKjCzKt%zT^q7a&IY1qap!NZ_2fUu^V>w?6jtRTmQd4U?M_2j@U z3mBcN>9$XQ=hVEQdr&_~NKy;Z`v!eq_+PXTbdmc*em`KQLDhxPu%8uZuP&9}H__>N zsP-<8V}&xO`dR#-uQ&A3!*LdFLYT$4?T2g|$yz<-Xq|e{LVq<>`O=dS+XE|Po}W*q zYT36MZ_T!DcoW5Du5@zBNciJ0wAzKcL8kU4b5I6G*F!;TGht_acV~A=);EPY^;)89pOhMn=2Pg(zq~kVG*VT?M_k5*6+OO zHdTz|`Y#GI|E=8%Sz%`%eY_U`EuUneo|oJ|dAVw!@??bOfSisPpgs}c5j3=zcPlSr z&CKMij{;Brg(w{G#8tI>qcb+>QWdS-afSkz84KWQ;^efrd@*5?3a8031rPgucw7~T zXERgly_c>DS(x3+g?Ma$X|I9H`qPuI4L>xP;_fsH>NBlc<3G?3ZM zGR_rA@m5w8qH#93dq7&Gz`R4|BgC}py>vnlW$c@$RXGNaK6RCoEg*4DjD#KRESOid z_Cvjdp3rnE+Y9JWWsOba7dH)mro^qY;fN-2DO0X#IBi=zvX zt$bGMje_!KO{KJ^DR!ETP3!HmXHIfCZU&Z?+h8jP^eeWpGE&kCPK*MlNo}{4^tkeT zGWf-9D0FK4=O#bgq>n_Vwo*p*>>WIa*?UT{hWz4bikg5QH?gxm!(T3m)&2tu_>V*v z7+>gf2fbvSV&W8%<=`=JZ)fLsP4KMQTnm4WorAoQ`|38Hj%N%i_pIFO{j$S_Y$!yN zcJ&tG35h!536);+>ieYg&S(xi&B$)Eo6HAUUN_gyIftl18jUeAD;lJ{>}?1Gmr<(8 zn9B-cO&XHCue6up=qk_|UBZO&F>4z)(8#Zc*cKuRfK4J|V*45S9lnw%FWx4HpMZ~Tn_Lb92c%Kn1QUu!zcI__)`Z{6ai{Y?Ksmipc34%>*pdR`unZfbvH@L{j0fq@a5#E@FSd5YYlbEjkR`noLqKFN_n z!w)B2h;l?F-%$yV-JXLS#9@l@j@CWZj7z($c^UYYPz4)7< ztxiqgtwC@Q=&3fK*Se0zB(yst;=E!#1{Jx*Mhft0(mu8GfPA+S^X(m&X&14dTRs~a zx}|xDEi__J{|vJJ=)R*F8HdSz2_*D6xW_dagg_;yg_Zb>_$5eNkd=Mql zM4al=B#x6=8y)4#dU|{ij*&6m&qPdalobYv7%d0Y7U1j6+dI9(6)B!ktGx4**%~oZkE)U%4Rhkb}3^n zP*I#q2_u^7Wgiu{*6ia4JOz)CjMgfLws+&?jM=p@OqwgE@D?e?3u%_QVWj4q8E%y6 zEC_blnI>zm@|Bd|WsyBV9H0ymj!(BQjXyCtcZ8L6K-}&+{yZ?%ly#J7aJ7u~} z@zp=4lmIf%Q5lGI8#*}l|1|w_T?%bj%R$vn!ao?2+)v60jF;{z1f@8-w5=GE^vKP} zn|~<#J$mEgy3+PHL#IaHTA_`T1rQud4zpiKE6O>{w+}msRh4N3d2Dgjs3vSSPWl;w zwkmdhJy3z)&b_TwZ~{2J@h`ymq|?6&CG(0|CPS)U$Vl>ubw~}OzJb@vz#*3BGGm4w zfoBb%>J@Hvw}DyBf{Pmo19n9Y8xx2Fx1ClR#X;3U9)%kRcjk8dy(!%?*tz?>lL|3o zvF9(ikKX%cIJO&$YcvOtQrESqYvvtmU4BqfQA!b>&u1RQ?ep^vdd)Q$z|jcw#p`WQ zK1qBk-oaf_o$B{^2QW-TK8Nu{7aFO%o7eG;%w3 zP373p3x~EW&u%Cl7mhS-ug*k>R}j|E3(Na`F%e~1GTNK?wMLZ`;&gOOlJ%P-eAUwf zBwJu$F{5s}kVS<{m*_9w0V-a?C;#rQI3MEj`8+VlacDfN3)DOr!U=M6`2&40h zG2GVQf1Q=i;J@O!CdwNZwHBG(Rz5)BtZ|_8z(s#}`=C%|o-~h6>*QFgD)QLTjf@h9 zbyATo?xw^fbgj?;1Pkt~ zKuJ1}LVo(8(TwS@2im1yeyn~xA%PO_q#^EQj%fbOVl{2el8F78#A%5eoEQht>};0K z;*8Mj``WqJ-Sgh6Cxgie%ntoR;A$&F7}qzIU-KuWoy-VEW{2KJ*35^h;55tVd70fqU19fe7a1%w5@K@?k>g zJAC2wnut1^iqJ;sSaym3>hzry$rLT}lG&Md3#HoE7&`#EyfkG1__yJ)#5 zuB^8auLDeKFERA?H@*Rn>3;C!BIg}e*E8fo@B+NWoL_Ax=f9CHWFIth#C`U&8L6iG zm_9FhRb4FS(eabpd-OY9l^omm-@p{OxgLC&U+$S}QM$&s_P|vALHbcv*mY6*RJTVD zN(EH5LL0ZMs*TJxnOhq<&lf*jd%*UHS{W)L*W7xx9_oNn@z=?#cA!y6s~r(6gkUosnd$J{Wf}3A<^*~SLr~QVvg+G zxz;6APmauV+*I1W`9@A%EuQktSKKR^9Y5J{J2aYmjZ<9 zmgzH6!E~vkjJ?55lTG=`t$vmg)Ut3yP`F{W&Sc68HO=s|7r*_jzCa#Lt9zF`X`oOz zF$FZ}$CSFv+4+=&!nsj4{grdBmj+lAIj=aBX1f$wl@!FsdNC3fq0Npy{Rt%udLAc7 zPi<)lmeQQji{*{eyLYrD8wkVTZS*t6X4|QixgkAk#wsYXp^*85`=d z4A@hhwr`{LS;qDTI4-1wBE08o=tEphaEJKtq4}nmNMiVus(ZrU3{kf^OTnLWPabZr zgoZzroqo1B^A=e}qmn2v&cdO=mQ~(G&BXy3D{2=8$W*;wXNNYl5=X}<;~@PC!AhJS z!fWb43fryX`17@ouPK$@_PE~&lc$i8_4S@KnuR4Q<=b&h7S(?lrz*wTD%_-(QPo>) z%4|JPa4J*wziw@fUyg}NzP_~YGFdK$G6n_khxk{u`&G-MU7TWx;V>=nJi4l;QjTOp zwcd+~df?5n#-zSZh~e;B^kTnkh>r1E6kBwbg#>CCk*odVg_8@;Ei`mJ0l{#Qln=9uD|K*P$S3Y02Pv1)R6JoXVnyxqomP0kv zhpON%d?K7Oh4&^>d;1yjwB@YR+orgT;b$b$2#@X69gtR1ck@ur-n9;5xMfixs)R+? zGA_8TRJ~^9Px3N8IuU(c3{hZn4;x$S)9j_1_UF`rP;XLxl*Wx*ZAZ(Fvt>2WHzB$E zl0k1snvO3bV5oq5l6Cnf* zQ+^a--&ntCzf(4d=Qz*+B&k^2lKAy)8PotMPDyH88Kq_$*Om{jWVuPot{tW6edToQ zt8G2$f9e5#=~vl09d-1_5}&Z3)=c&vm>gZ5lYTRgHzKpWt!d-uy56dmi=|VYJEggN zexi6wvOdmAAuo*}id=Ir6a}xxn}#f`)Zw1UlJ9`8LcWgpefg)+`ybM!M3KCtC^DO; z;6Iuw&tCl>NEzLK%=8c^xaRe*z{9KX4DjVN*NwX;IkpxX&?IF}L|tm9>;&aL=H>A( zAEPlAP?iVN2YnBR$@RGdoON~ChzI+W5$xTi(zuMS@C3X^wQE%aVW93x#o-N}I#??o zqrgl1SSXLgpnZJ2CLgnC~m*v)fd1l?#8UR}^pTxnsEV=T@jkf?)1w%BHOz&xAh{|XI19+oACH`qEjC*;$HO@<$Nk4k;iUJPbaG>WY@4ZFsbgNXPHGNq2ZwuY&$inxep zazQ62BcnzcTv+Xi50H=pRN7M)nmX8|Ey$U$bmk2}iD7+}yDK0v3dfj*)%(+p_tr{! zVN1RTafH&>F3TYRxwIQ^c}y)$P5K6Qy_BFu(T!ZKaoXsK65&FWq5GN!m)dt*vBqp2T^(y59W$Esib~ z!FzCk$NB1F7QRLUbqZ(1B#o{tWpEAkZb}&jKfD@OuZ8HkXO8Gq@_Qq`)iXu(QM)O*OaQLk|I<1khF>693$3?hGjMe2In3IzrDR;$Cmx4faE5?+&PP>2nN z=V`#jGbBun=xjoJF4TBCr9i9{>a}iV^a2zihWpe~UHTcnnrc1jA-7n_^SRt~6pVoT zk@Pbg41GUfrPx&ovz{(e0VsnQD*6T)@$~59omW1=?snU? zY~0zVsN65~8uHuVy(w|@H)?56S~$rVLEN~sbc2xhTv|Rhr^c0}A(B3_d%QwB@GjI~zQ>_n+z(oh|(@qLDCLKD4QvrB^1BfZ1cWO^J~>6Bh@CnE4=9`p*LH zWGhB}%YvISbJ5`n^L>aDgWh$Pko^RKUBe=J47#XI+vz~q^Ep3Y#v+r*?ZgI2t%FUH zMh7B*zKm3=LHGGM{v!@`%;^$$50?L6LD#o1gV1Qu@nKj}SsD-rK-G_>^N&P#>{hzA zs|dWxv67!#qNWze0gS?@f^_*(>4A-<9)6dhZqb-%ROEC+zMFANNV@<)g``Nj_U?B~ z%eVL!ek51>6aIc{t0OUi^yJMK9CL8a+hp&*87?PG`sEM%uf4!3ZKQ%=)7k4A{61y^ zA>Z0xWRgK{mVphS_tq4HR3yt({Jn6ZP^IOmsjaUs*TtuT5B$LtP%ovLa~X!$7ZrD^B}!__q;dH=jA1(3qf|nK!khqLPt=a`>9f67>%V z6hCVY*ul<=Z>~g9jLq}pru!neO`lG~yNyx{=bv1GlARneoGG&y1%8NCzVV?GR!X^4 z0A{IEEen%jJZJK({L00uSBCKrD=}##^dsD;FHlkdAB3T2cp-kFY@g9I4 z=-m(Hr#LBJ4_vHI#E1qzL5VokmO4;kJr>vpK zbZ5Pbzsk-9gpfjAU!i8lD|2{A(5i^$BL5>O8z-F}d~-{#?%mTF^K&2({+079NJK&W z6JZkdpuY#Qgb{w(kJ8=%M^|C)c8CWR6+L^?>J}Z*w>*b=6Tf3N{9LI?N0^P=t!QKL zcGkDTQgT1Gw3^a$vb-aXR~!{47n z?NByHrjEjSiD210woyIaoFs&_kVidEs>sEmvQ(wpmB%hCC3|ibFo@x_w!BorY2&D^ zM^&;0>rEy;ngZWA2t-w@N_{u| z%JGi`ELJ+m9C_y*L?Z|t7C~WGNph^Uv~8oJ(dM&H_sy5~^ds4~m& zEkn_v_Z8V#pX#hB)R}bdU&!E$cGRnX z^2iKbwVeK~s<_NQ)TmDAcp-{EcoJ6K@LP4Nz)^C=&E%4Y5=-RSe?;=5k8Q z*|`=;G47Xnbz6J?9}_nhzRfpZL-INZ?L#l&2T}AfderEl`^?0dV zx6x)RBTnJG;Aht3F>gKIY~HO5%pj<0E8`1OY*x+8;nW_t$Em@d;v0W%`cc}FV)kM3 z+||}J)~3O@f#XiDCZ{{{|8d1pFhLnOBl}9=d4FZv&(7eMZvKM=`Z3p&&p9r1ply%; zKR@2bkR6+qO6PvBU$T#y7JK$fckGNy=!+l=wTUJt26=fO3kTbht#q2;sr%0?hOzPf z$|Sn2p8Q?hZv4-jhuYuWP9*lEcT0b!(}6@@|Cs+%-seAZL;Gw9g};c>r|SL%k^JW& z{LhIFplleZv)dwid>`MaSO;jvqx=&Pkf#e+4aH0cdH^Rvrp9jgm>Yj z(_8k}#>TrVT8;?zUl_AL=pXLLE6jI>L)A^c_L`Tlv2lDg;0G>JObpQaJSreYZ#Rib ztuIlty|;#{{kdWk{divw9sz9Ox~mOS$48$Srw5Rw^{~z-$UJm8!PqdGotU-d{l_0A z6S00zv`?s0BK{F*d18+EM+k!7&nk&1yfoMjmb^9X921&dKWPsSQ`4J~okQ%|+>(aE zctFibT7&0B)zJ}CKJ_CbU^Pa}$o&GJ34THMg`&pC`PpRfq+=|smU9>;R>I6oXRGL* z;Oy>o^(D+W;B=mE(|;=v`cHNp#D&d@}MZay&)rRi*%dti{sHEaJ! zN(H+#&b5NpkxGq@T9$oGvRkI+d!f>pM$Hi<+X~L;<)Y0>KrgZ{I{~oSDqzFb1!!bV zWeh@ zxI0POzOM4@BzbdbRneEyg$eQlVJxIf3XJNAv@%p>m=clv_}#1j^pbg^Z|@fU&5-Zc z*WP?qF1^YOWCpS?C2E~nxLQo4|2$7542F@}dvkX6&RKib&!=JSYEB+60|P~;!j6l> z<}^Ci>CP%0B6uM-)qI@)&k5o%ZAC5I=B>uQ*w^VQ6C+EpJg${TiunQFe++C%&4?fs zoJK~`K>4+*?_HykM&PWBc~_1o@ItiP^b3I`R06B&L{RKquCNOEQyzA+1N<;} zuNED|%&gcX2zHZ{tByrb@OF}by1OYK8i`RbY;ZXtztd`63=!)EwE)>C;wuJ8K#BO- zjR`K(dy!F_8sW`0R==l3rJ~^hesd1U+HJ?oS8I;smiaEb+b)g8+M@aH2QaR;MgClea+F@K0fYcls#MoO zXjI%Z`u@_qt=?2(F3T``$P?G${W^3n5{)&-UC~l!#Og+!8p+k2%#~R6qB_gV z9^%Zb1HF^Pmqu~(jjoob-P(+Xo?WSp#r3{OY96?(nab`hm<|?)C3?ikQuMS+HOvS99Ou%pmWZBgQsB2& zmw6X&y~gIxE~_!jrO_JTYIAR;G(-LL_T#3vre!4? zZfp`et*lo|2=OL;XGTaOWZV4i2I>eR%uKwcsye56YZR4Y>iJ<=eVHY6ZdywjA{E)p;#Xl^&@yC!1qex0OnXxD zdq@ypW2<5_;b4xmwm}tJ+e83jd@6^XA}##}U+hk30$JY3F%V?ZvuxWB@{UUG0~WUV zEVMTD-lD_ONMeIjkIYUIWJcMfYHYQ@R7>TQ5Vk+_M`kKwZ$pFC$5}TYIvVl>Jo`@- z|G#~tbCfeIG(KtjrT_YmhnU|Oep3~B)$)>?fqNTsmF>z5WR_hatYZNGf$;G}^( z2A`If_pC&`+I~p^|3{%jra_+qJ7^Z**n97T6UEbNIXk5{eivM}6WXXY>%q%tmIlou zO0{5*=;T_2R{ome8`sbJG}s8VaysXu((E*1YADK*z*!m?V%%&{SUj z@g}c`I9SRG)kSDDHdhXK;y;p{RB0jM%jQh#9C58pC)5pD9Np>+PRtM&#ZTyf1wn>8 zSbniWHO8xJJPlJy8+-=x*j(o4w1$CEZb)&M>*qeN`1nsGgc#9Wr&CgMZNY{^08>l$ zvt2>D)<})+Q!ge@t&)U-4J~ZeBC3C;yik$~;rlgbIN++apHRmt_w~Dee%{7o7u{Bi z-Xi438LQ`4Jd7mu@PI-bGA?emEp^3$rB`(&E9FddAY%*cT7CDhTc{s4YAF?_I|Oza zA7PBU#a&B&FBTvzU|+PKslci?R!W8c&ES$qh?`(zGeTnvDt5oDO!QiG(8J_9!4~-v z>&^pp7b%F3MR?>DSwUYZ9YB_Q80*XW(^qTb$l=jOv&w2`7PtX0B?W|Y_mL?VGOzUN zXWVcabFCbD;K8XpP820UQt^d^)VQaj8dH*pxv^L>&vUc#PMLsy(R8O08$7Ef zDO@R+IDk&=LL1lXt-XK<-Q4xxhCH4Y(Ga22FYJyeNB7Vk=@j@ju*Ful%M4RM(7zH3 z{R{;@WivBFnHf$eTrzL&*Nq%#R>Dgnk%rNyr=Y#!?9EfpQAs4q-TMoreWJB;M zmCQThe5xgV14kg?Nj|KE<=0#nSAPGQx)tVeBkVrRBr^7}cI*pl?{H!jP~am;M&^Yl zmF0yj%|Fgcj6Wc&;$jA&9~28L1F%&Myd-D_%dZkl%T=OeCUJC97;jtF=Zm6dr|zrY zzYhZ0RV16@u9?CT#W~~75h1hn$f-S8dr2uZdR|M#T=Yeij$uj4;G?2+`vFB$;`>D? zM>=y%Q*Mpr8ls}3qHGY>t0#*Mn#_`ISy+r{Y)FyGUMV~>SU)Il8oZCC@DV9E(01X@7=KJOmnNVp{xDpGr+l10f3H#IBc(lff+R@rIDA&1tqU^vmEgg zu=bnEXu6|r;rnT7wpMeTAUPZb*0LF%Ey5*~16YTQ5$XO9KRy=2FBV|*QIR=!2WsJm zN;T#Lx?I&wTkl`dYe4f)~yLS?97|6SsZ zs?g{~wikpNZFIumHn`KK4JjAA9X%>76wt8eLXM{EP7;EAUa1?Y<{^;D6MP{>7;0LK zw`ONoj2J&qMf;>=^YGbrrmKJGs*DVwyU5KM{}YZXG#s{dg=U<7Ydt6-gl44~lf!D% zhGej0{huj4R=x}QelYP5l4zsC3WS6A3Xd0}4x*Cfe=;*yDTcWgXR8-_neVCdltI30 zYK-Z)tK=?DD;1z!PmLM68hzj4jW+pid@RkXvWGrFe_A6Cs=CIVOYVL) z8*Zx!_AY}jmDb$)`0xJVCe932klo81tczudm0bTLMxm=SZ4nOl!HOzWmSQRN)$~7a zs6j_8nelB4D8GSYs|4 z{aJzBKyUHtC4&Oz@@O{)HHqWjmKGm2h`&!J8$TFFOoX3Y{e(b$3qCTQdS3V4Qw--& z))lj96*4A%dV!@pg5o@cjKs&u++(K9VCp2V+XU@P(<=PiS>L*a5u@}|-Xi>#F+J*N z<@`$Hq2yOx-;RRvCs2Pa$QxX;R1rc%yOx6~tSnHpmz-C1O8MT2FBxFHS?}+jx{eqG zdDRT|QTm-nmK1_0C?4h59!rw%t&)uIonlysyk3y*A|m942*h^U9kCypLxHk_$4Yhz zG+p0KxT{CvV?+C3Nl9vqwF1bAK)l!sYz=(k*N%;10JxTvpf<#vt;QYuv>2R{Q8>}r zdPBIM^*|{10V$chf5pxdVjNE++9w7q(y&N@l z1+4TW?5-6%RnOUnv#rEt@G?e6x6SxMFG`J(@a}G&E1mpH0nfu(FEtM_lx97<4=Js$o9+#i}3;v^$R+1p|Em4WWTw;Y)xu>-_iH~nUTe#Awamt{Unye+0GJE}K zzl%pKBHE3_Wb+GuB~I!(6aRM^nnww`?(Hs{SK}kO>j%&a(f$#@M!utR9(+7&0iwHR zr{$BNl0g1G9z!F)jd&4YxCKu=HPYd^nwA%r+7lHyUl_*WMyCckSk7_MAo)yYBds++ zAX~#E)2}Ro4W{(LoR#r%+U)HVOs^jyC7Enb#HWnSg{XQMo(|+!B_T^G(M!VFq|R{t zZER>_?wg^0WX3v|4m8NXE1Ce%kH*1FniX+4yZZ9;m5^s4)^d1BLFcmOf%zmR9gzTR zI>b9XIyQ}ZGw-^pzq`4&gY_bY*tIy*iQoLqt;EuT@u((1hGBHp^pMQ3I6>H{>ot0 zR6s23wpO0oYCOv4cWPA#57=h4(s*TX6qI9*N!5eTtYC%j43dM!Rl;rz&Jx_?)|S3< zJ$DaPw7Ng9=aQcO{D5sD3n$|m%fqBHn?J-(eSf6I(y?&tc&!&1`k{X0uyTZvF<_oa zPtXT;@MFMQLDtPX;roi%&+&L0?A*S8D5DEeLB+T(`@7kSzCU0}MSq;{q-Q!);C4~5 z$j0Z@>nm7i_{sEan+=tl$A)fMAyZIXh z5QIFRsT2jzoi?PitK*M#i>0;u8ovo&45{3($8DtDSPXKeZTlxmD~OeFw(USyB0Wmc zSGSN=O><;XahJKX&-enJ!_fWpc$c!Ft_n*7e^P+q!4>0Pa^A5ULXY+|Po zR6;T%n#^3%0&HS@7Gram-H6_CogS+Gyb5+KebWmHNW81B{H`{k8EBPljPmZZ3?(F{ z&T#PwdMU#Mvg5VzDQ1rT;X@nYB`-w85TjL~b^G5s?d|_&=(-WBk(r)b-BoIPc^1r9 zlp}UTBkK)=Qd|$6D+&nHWy98`$N)MwXiKq&h>?g7>tSYmK}FTaoR!rL6S(O`xFqD3 zZF^||FA7)54Nn0pN{e)yo)JFCs1F-F-c9?N#sduk z^tZ(c`M)nH@>%#iDYl@4DhIVpHy&H1V<*jzu}*29IXKvw9?T?F(%IBFOxG-db@yH9 zs;_maWevz^&9$gG%-~}h%so?EZR=RcSwGeIgHm?%_F^CwCv%)4eLk02F9k2?&CjPbBuPm1 zc^S0sdKp2}fqA=d-0wPbKb{cOr5}uUEN* zgyP-(Z4xodq0s|A!`lT~!4(7kmA!=l#2Z}Q;oLDPR!3}eU^i0mg%evpcmiq_NhQxZ z6dw_p14Y_OJviX&?XE;AfOJtdJa*1|A0f=l`jz=sOo+p;u5~Bu)~wzIy5jS7O`2c={4^;y@5Xg9XJqydcX!ocY94`Q=^sfHA19W zEgeCm=JJkUs1f`(^Ht<5^5oP3J>!-iL98mu-q2R~4$-vLT?sTHX-6eJaf#+Cz0%U^ zs)ZU-xqg=p0r3_KAc*RXtXZDh>8HTzlUHR*of9FA-_DM)MLe#jnL>xN0#LRRe2OU; zd!i4&FpbA#f{+wabSyR=B;!0b1GPD-wb}|a=En)zEP%hG%s zIXmT+Cv;AZi?z`qfJ6bSY*X-OxSJQb>P-X&R$&Pzg#YfR)$k94dZhm<*ttkGj7{_Ad0roVA24R`?w zBlX;$yK_tMMF#JZ+Esl7a)e6ums*ZQwDpI(O+z+TVAYIW_2E`2&=gt7ux@Rlo>fUu z<65Hu|G5FwZa1vbY_z&O@xUj<;ZCWDpedGL30w;Bo}aw(Usyu@kqzs17YQRO^8LGG z&H0-9n){YiqetVZmD#3t9_j=L6*`$mb{^+j5mqBF`Rz~^&p^f1T;-G`viTW)Q02XT zbK$%9qXf*zNRQrG3LuPwuKhy+=Dc^AOIh>Z=liqd@ZJ7_KMi7L<$lMsZ3(akYuPdJlqUg2*sM-{HJW!z!FCa&n-2S7d>NI3g5SF&XDpV6c8gX zmN zT`E1b>8&$wq}jxL!N1D%s_EBq^(7&clusD*11*5xx?}33mMU45pvT-BpXAF=pJ;Bn z01e$bl;&f^^i~h+eF;!=Ng<1fT^qSsP{WLiHE{%GcWyucNM=`5tq1TJ#w<&EFOqET zLO_V~4!GGw*+REbEkDMLvq^fEv(}{sIy_JwhX!02C6wKZ+50n!P=qGr?hWg%3ZWMJ zpi4Lp(UMaJzNXU(J&4eYwy^00nnMPJpaLIJ@c3k z+a1Xbw)a?54Jzk5u^DWpBPDit+h>~KeZjjnZ8r9;K2zWH-JgyPQP9~vMu=TWPFvEx=XMcZfNGaNL9 zeZ8Ft8HAMe_*ioieaIXHW^47AHDe_orX&GLrXyb#e9hVx5x7%ec!E(KY8^c3vhqNZ z9YJg~7-%f^PtTFt+_9=gpFV!yj-B;W)}I)bU0xdklLeAa>55k4F`D3BlW#76GhD-Y zSBu4#*K(+HaF9wt9=%OTed3kw=pm%Jm#6O9%Sp^(B|cn2Jfaxy|1B_9dvoj~WnU%_ zS5jEyu4t~+@?7~u(b$F|P+zjH9xGAd4+M*&8(Cgkbk8I?6^aTC61{N4Xzz2q=e1v# z4tDEQ)r7jXzfB&P6H0q>3|@{<%!=jXA1Ldu{p5{eQ#O)#a3rlc2mNZYDNLLcqtzS%6#-@_+Efzxl?4O81{|1!9 zOirdpM#dJFgmIph<9qg&FO*p%n+K|Uy+RFu(J4Y1_saE;>LciLP*%$UzjLrSM-DW2 zBY6PYmhETopPc-s36P&ta(p3N1?GFJM42AlMJcH*eq|3Gksq}kQ}LF|_vzT5 zw|CtM-|iCi{i(Zp?nuP@77RU^G+*hmggt+nO5K>#oL!DM=zl0~(3Cc8;8wG)GJ?ho zar^dSKUw|}YuiE`@XT1qSZfjtaPVI?ovLQ+hr*ok`IU5yXF_>lK^Ai-w`<47a_2oc zn8C%;D(cA8suUhKWzuKBePCA2cg2d6(Cbnx{)aCc9D5)kjmXOj@g=ltttZcYVWl2h z)Nu6{ZTgx_NIf)y^0(`O!DDsizQ(=KhcVwq5!p=7a*LmeWxpJx+*@So%c%&5eHBK` zf`6D7E$Z-BjzO#rI%U7B>F><|s>;szhSNk#lo4?mdm~_T*=pmv#=6rI#E{ij;+&+4 z;iea9T!y}Bb!`4DsOtlC{mBxZBd*v-MQ&wd>X#H`D874)F6{b@(f*#pgGwPUuqq>n zS*9g?joShhP27ef_3{>@3wq*S3e)U-yzy<3Ge*FUIh+w9<{-mGqAm%vDa486a469p zeWXsxklp#Uev%1VxcIrfpaNKq5|LZWyivic-;- z3=mt?PFA=8_CD`oUT;| z=N^+!U@=!k)6J3%`mjI9##h&HG&D@J3>h)~}$2DX}~7h&q$DdDk&xpofHg%08;}$4~7cES0w9(Dfa820jw_0lf@OQHx~PEtTdHZ3V;&% z(HRPFy1BBKf;{Y9LrdL{zTuB@P0q9pDPQA&={U*dsa0g|G@~AIx+_*1x0-(B@As(v z*B!lIPsX&MU5k*!P4@DprzLJk?tMN@3eLj=ACVB5sNGP5lA;sZT_RVVkXI1%gn|@% zw1Q)a@%;A*W>J=MSl4;^GKFlk#DL#s#DZmA* z4OmmSqicPVYArS(TM8&yeX>tDqQIv}LXI>~ssRG8&2P0F)dn-&65sml&!6f3f!5fR zW4uZ|9~^ zmrFGeFC})M6H>`U7@|Z|Z%(TfjUVVe%ygQRW6JmpgU^y`M}g5F%JD@BZN%v&vG8Cg zwRp%QW6qwqab#`OD?i~N>f%8J6?;_R+muSGcB!kCVynI? zE|g81)qWIc&npCv>96S`1l=vDsHoU#8W=Zladla*@OO9PK%ideX?RENeK)Q}J@QB!Ga)>; z7v&VE>zP&SBW>!Mm zmSY;P9%OP4T^kr?^?C{=MfaG`HEYxaP9jfLI~ahM8vgyL|9{!tH!h;b11lFQV!ejk8A8yXeGz1GYK~qcxPG8n;S1pCAqD z0|hS>1XU_tu+Sxw{ra-QP9m@Fbv=ntPYeVndI`4Dr2V%8 zq1=)8Zq0oim+Y;jRbltAEocHyT6`KOA1u`S$fF-Sk>1* zb)c}B3b^oIJQ(1OCPd8+=wIH4DuZExti!|oX&{4_+Oj#CTAH8{T=wJUDj&@iv^A!5 ziV4QLmKay|hoee*oa@h%qL@*|4p=U9Gqj>~RJqW#*PiRL^Kdje1C!4zLKrsAVVNx- z7HD`UwFBUgXe?t3_9g^UP_JM8r$>G1Go1xpGUi0*>8Yto1c5G5D}oilP7a=B-gkO> zZa)8o=u04!g%>|u0^D(~Nd)OA)pEqWEmgUL==a4P%w08l z>Qvb3@M4ONBcZ5CxOb=hr0(8suSnv&``pa_xu7rPgi!L}4=sC9^o5|^OMdBqwm9p6 zw>A)cyg`wx=&XmZpnJ<{*Lj~$`LYreM!|W=}oZ$OQ$Z3m=Ky`Dp8GpOHLOUt5bbRIr>LP+u4GO!7P82id}>M=yrw?p~)d z4Yp5)_z6aquT8Wp29{>ki}1C^^kl%hQ%3O9QYArQrfiqp+|deN9wy?!L#M^=JmkDjZk3H09Ay*4n_Y~1>(HZgnS9#|h*Tnwrd5+~+5R`+6 zQuG|EfPex40RfLvq_>0s0iqOvgc^EiM?HwZ0i;U_C@mx*iIh-6Q-rc+Z?EEpA?@Tf?-^nxI`aJK?^AAC{Zh&^m5~{+AYE;!hGQz)d z94r}HWu0)@Z1o>~n5bRu_u!-y+ADIH`q>N3a*5mJ+x#R$9>aKAMvwS`y7+Ki{HjUn zTUxN-n!!RT8Na0ITwVDT4J*=V2Y)F3-NP#vvRi+!Ed=ZRVB5GK{xz6C+!ShU7(Zuq z5HIK!e)3KM3s4*)k910vwo=P-P9k!OX7(ET$m8uB;Xnt6w{8sY<^j=Y zS2;L=JKI?M@#9>@$mb*=XlO80mLQrrUuZKnxMP^m78QKi!5W`sCyFc`QXH=6@RAw0 z6woi%Z~o1^bHe&s=WbpKkBZZky|wYjZpkVE9uhN%(iVOC;5x|^F=r>vq>p{)R9q>k zX87AJQI$A!1-yTkf7kBn2-Q_hFjs?;^C5bxxGW)-=9&y)ZWQ)@gr8%si!oFIDH$M(cB>Ma#ZyS5CHV}eF(T;rs<1S2LHyz>(gnLudc9tGrh zuB-8yy7i8}pjb$DjBBQ@*e@#oR!~JK6Lgqtxg={;i-~go)hcwpRr-fdV zJUWiR5EN*n;I1Us3j8@MyeD`WUpD~mO1JLxR#XGR?%bBFpGfqSf ze-gmvn3}E&ATfNUB@D{JO(=9T$nZ4ffu=R6v)N=Dr<8f5Xtf|IvCXdEiC8OXR^mO; z)I^KW6uUt9t5^v2(%f&l_q(x>-{NlI@9WwaYl|b*X&QQeB^9#`F!j><2%ci zOy8^HIW9e=RjUezp(NPz0$N&dcWk^Vc>IG8Kiq_7(O0(=VS|?pU`aAbHiW;NtJtT; zsfT{yK16g5;ttrCCab+a&YqI^kmS#^--6<4o^!Q6ZWv@;8%BsMC-AHBfQ`Vq&jJ*y zuSN=GCzmZspDR)nYugKdcrY|t2=kGWs0Q~bohL%z#k6*o(d;Sw`|)ExLmeJb+ullR z7wCrRm@q&{Ufx!q?!p!gJ-wpICeS;_e}-Jq9GHy0ws)_t)yM-DP~m5$7|gj0%YA7S z8qs*xG2w71_b3+g%6PdV3S|cpEB2t^;`?5Xdc<0>XJdNMyFb;^egUh~@>#ZO_-VC- z9ntqtQ=#JRl#6d-#LuyO#$u+R^l1Z2m-0)zZ=`qmwcuGI;4ZS=jir$RuU@b}B`FD(vzAMh6 z8@?e4r}-+R4SBPg=frbI(lYxQ_-YBwMLX4ifHZ42qk>U-Z3pDJ{rHLzYPNMS%rt7< z;DwIFE1A`QVln`4T57-N*VgR*C$9$^mkqKv)8-2fq}^2j9MqYq!BxY3o*hgW*Oe@~ zZCmYO_s@mO9X*TF$)p&5vph#os0P^j0`DyvOsj5h6ekZ(Pn*gRI{<-}2+c&C`$r7?SwNClrL3`1pZ?{8O2YPFR`R)9= zsuCX`!OwRa^i9{;{Pdu*eL|IwKum;|O2@z{KgA4xmX&?YWA6_MR(FtFKz9%YdLI1s ztAV>j)Vs7IqUxF<4EsIF?Ayqi!Hc5SQSax^qeA3+Za=Kd36{xdZ{y6W%m-h4GbHme zL%l0*$D!_`=bL5&p8mnI8cvJ;%IVFVpJs~L2>A~6A~qfQ3fx%UBo{M+aL z-#&6$;;*f{fGU^o$o1rtC)Uvx)q!}*RStA8Uo%g|sP2ddXHK|*&3pJso^F$&IqoR+ zdp>eK1s!XHn~IInaQ*39`2Ly$Gk&*`Gkl5f%&Tz5h*G?H9ia`$`29lRUS4L6N&;yy zrRTS&Y;1NGdQl@YLFg#lyYVo4*)r)LY|pgQLQ^f2+D>h@s9vnx+gP#VkziBfcP{h| zd(3O~F(}vQ^FNJr{`~-5mG!Z1th=k?b#6VCPvyz@oxZy3 z%~*6&P*FKYI<}Bmzw^-ze#gxJpDSql>kDGU{@&4FE-r!Qf#UHKA=+e@C1Au2577lV>}B07@zT%ld@@~EySvAddc z`PzrM4-?0?S){nvdeNc!k0HV;kG`Um#|}#)?P?4y%of^JS7D+cvuZV-_U0n|J9wH| zZ(`}YHY=Bxw_KiCJ$E!%oYiDb>A7q>HbjQMO7XknV86{wuk9_#Gwp`g!bM#N%|lYK zU_l;K-#wsU6m$w4#JlGQ;3=8ZZNLtj>lM-}9^^b9p zu+WA~d6wXcmXu8mD>j+r&dhR)K3xXaTU2RVaDq2K7AhHDNS?u>D_W#M9Hm4@L`5CR zhx-es#g?q%o7fStUandS8^&f^x?x$^?z9s2R{P2PqSRRsj-)ig3M@5MXrKav3@XHh@j zT72M8*71dkc4F{eZ}=Ll-!4%_BL>8(hMTr;cmz(b&ILAn`@sfb2ZRt&Jc?ERHc>0m z?wZSaZ*hKQ4Sa9vMgGyC4Dr>z`Z?vUAI9r%=KX(`INLB&LQojY8KhZs*$EKZ=2S4Dxb3EME#NA6pbOHAH>r{Mb4qE906aTcVf1dnVp!i3rEy zj_6%Q$ocQ^nRGGOx>ozI+{X#MbZY4(Fpe9c*UbNG_zs_hgmw2yjB zp~JwINv|QYp|^3^Ec4I8`!J`*n$`4J|ytV4YloM+=p;)v=7YfN*>(KzJ2zFt=@rk8;&to zZI7ndFk&K_&4LQuZUzLFSxwEcLs)u3C36(5E%ID&3*o$ia^jly z;>O`0FGL*svHKpOMR%+Zq{dG+M(rdEqFQM_*>9SK>)TNf^-5L$SLG(UdabUo-wl|( zZ4gtst|{DX@IbG*Qrdy5{S8JA;Tw0dDj4Aq2m6!Pl5Y-QQU7!MvX~lF(&=hKN9Z@2 zQZy1nJ=Wr>s37!6tqfVKE8m_4Zy$<_+wA(DN-eFQIJ=mkpY@BeW?O=D6lu~YJ=-N_ z44bzSXI*FbJ3Qa*;IT0BRCi?H`zex^u0gIwG>?(rz%aB*fqE%3rK&%@VPuQPBa=dAy)_wSs3N z;*55`t@S`i)`ecuT+HN@cm3jc*|$MGygNCXTCtBR3{^AiE=x4~x;+_eZ_F!Zay~0F z#i!CmKl{(TLacr$nDBL1JIn106%?Y(LuV=L;mHt4I~mlzb)QY-zA1hm-GB_u> zmoCEDf|KV7gifaj|BN~(1Jbg&4s~XD{{!gr-O z!9$9(@M6-l+9hg}w}%oG7er&gVu`h_%)UW!4*m)8+qOB-9MlUFq&sm)1)WV@^XaE4 zRJ1-v-C2ses^!nUa|NV!WA;U@+aRyS{OBoyTWTA$G9l*mKo_IxP@D=LC^eEtp>2w4DSO2A4;ShEWYxF2E%~ETtRh_t3P&Q(fg7y(ss(zR+|a6YsMc z{-BqtSCShsx)Nou97qOc5)K=8@q@GmES zsw4kN{aH}@XBLmHk67UZPItJ1 zb#+{L{cSDlu9-^g-69;1TPM8#S7=+1YWk-E?fuy=wC02Kw48_MLdX7{EBtRCI_3AX zIsPZ)Vyo5VB#VS#XkW$}n2B0P{05&-%SSi2@A|qoo`tXQ8H0)e>3W~R$QRe^a&jOI zF*VA~=Sls;NS4^tFi~;-n(Hhv$_v9qL@7f;-)Wt8pXK5gW*BNGFe7NUw|kiP2#0z{ zNu=ojE9m8ac3d`Pm<9mf+V<|xSe~prz7vGl1GFrGdg18t!saK9|GP4Q}JvCmD zW=;PPyV;{REMtfaB3+-cC#mkMg=x8*%iCe_^Zq& z4$V~=c7b%PXB_9{H&EBKb6;J%EyPNUTH3h5t+h3X*A9o^}Mb`mlO4u1&c=N9EQko=7k#5A~p-#;Z zbW{=x#eY{3w>yV|7$cm;CToI2iN^QKw0N2mPLfegLO#IW+ePp=iJ;72V$0Rqt{irO zI6|*d^aY6OeYl|JlLp90<22*-)S*L_tuG^eOW`@c-VPy0NGa!L@L?QrZh)#|05%x{ zKUyjXPH#EmsdMLEF0&83W?n!*SRdcR{-TttqTufZq`miEXE?u>Sop!*Y&cn9`48+CE}*WIJp!@Zzg70GLI^*R*V?V6sGge$A%}Ff>@9fF!GTTD;wGm|ki2j^$oRGPi9Xm_b58ZRea>-2#7p?AW_yF`m!}_Z|_gywtuHcjiAom#9Ozey3^fP~H*AaUk*v zznu$N*AZU&K+)3vl;=Liw6o8Evqg zvZRtAV!@H>$8Y8|MO^CcW21u{ra-)wUD$bO3=yf?-1nw|A|5z0K%+nRY?S(wrNnP5 z50}p#@OvaxO&&3N?ic&O%SwUv0K_!_O3;lG?&jTyX^lrp)EJth(2|ltUpC@{O6!fU zrb>eZP?Qb?( zQkZ&J)&+LQ^di~{-)eOFytY&rylbWR`bwR;`>wOk#$JG5%T~Kv1<`x&_aAJH1;_4+ ze4|Ib6b<$%uoRj4>)N~@ERyM07Z*B$%a}`K| znSohhHlcKNasJ8CeG`?~JsqB#4`dj()f0{V`%(2)>)G^Gfv_5uIzWbR2t&YIu69Jk-y|-1FfVxUKNG+8&wId^( znqzOzKeN0hW{rd1h^NmQh?x|;27fv;Q!Srzm#zDe48Q^WTGevL@8T>3!2dzh|M=Ip z7K(eHTIm1yjBEy!#MeqEp~xjyF27NP28%Mfi}i_x)EABhC6c%aUf!B0ZjIA+MVL?C z%3tohZ6C1@&I{(@At5GBLr~B7v!2NXAc(vwmVk2fIFwy%mOnqDONaalVxIGfGii)j zZ8Hln{LL>`n;FX(6Hj7UZE&cy(*h_PkHDo^-DfZHQfbn6PLr7WTR zIi_CM4VL8(ODZx|9;irJ`XwB8{j}L7yw6==z^41`WC|}mq%(KxR8XNL_lqLqV%GVr z2kn_gG%sc4RKbjbyR^1Wnz`$E7i}8t;Cz$bVY^^mRFoQ(dEE5ihMy5X1{NZQ4i7R$ z?aQf2H7C`ug-p=3{>Q5xiX? zn~L}<4N#?&kQDhp>mN=Z;`v=^&@pGC9$Gy#ONWp1i1CYI61osA#)4H6tLkql} zN*hN*Pl<~Wgbk!NrkGB;9H~kx{LI(7l`WmI8QY6WGzy3J#RF9TP%{*@G~CMJU>hnqdC;P>Y2MCkghR?`%&ZYywA zyWr*#mitRtGmT0YvTwQ0R^W4UDG(kN8T2jd=#1#U{I@z}l*Rc|yna6P@l&9-*o;#| zh!)cBkFy8KKHN#>6OBmEB%g0$6F;8d@zl3kyU)<50PY22W`Ek^T~ zjtM=czs|83gE{G5gIe8sf-_q^=^`)uwb`cohZsMmnAc)_+FDNxR=j;@S-PBy1Uq9p zVz(=ce!KqnygIQZyP&ggV0a_tCi__$F`l5?Fj2jJXvuMVmDzZI$`7^}1T!(KD@<){ z++S_KPBakf&TKnvV(vvqy*_qgJ0`2+xXyKOx~xwTfPXZ$lG^u++oRSHxbGL=3`$J! zu^dtwL_bsVa@t-GrUo=|N(=$BN<$WAxfL1sKq=WZ6e)P zq6uqg7BD#P4ON~W!~=6@>Q7OMkuuXIXWbp|6`Rlu`FRPJEKPI#dAc}WObA^gTZ~6Z z*_z8IpmXbja>SL}-$XMd#Wi~#&z;eHWG|$awFgY{M_0@L14Xxglc|gb%+&au_=L|y zfFFg`8D$?=`k};kQkRE&#;><0c77X`my`+N&5+nU<1Uy`?vH{hcS8Z$Il#Ds(sk5= zW`hKNk_fnkBBO=WD?YSF{n;7wZLh=hK{PE9S|cboYjf+0&Ly4FTF~;U3e#>**w)Hq z4yy?ey|<$Axe~!Hg$aqFh1I5E&NkwzoidgL`D5Q-?%s)NtTKacG`Oiu5UTf5{-BZFZmQz3=ZL*`O;)`2f=yE(g zh`ofdl;Os}Er~u@Zh8HOIv5FxQ5MEPT(tG(pC*{)z3^Em;m*i)$Yx{M_|@YJ%WZwT zt8;PHr3KN+D`e+O$z|fY*Jpv48UUtUyO7aQ7x{=@oH~#}t;le-9jVOhwaZRQtLBpL z3XW3CEi+$qFCUN~&rg`URhmhy>`GPzipLf3ev2(d`Sf;V<<3kKDVQPs6;foyh{NmX z++dV32x*Ba8~qyhq<7r3q4@kD0pCa0F&c2y7Xc5&eC=*(T%Ccq8F_Ac`8M+ss<)Ns zg@-=1L+%l3jg&3TghqS|h)XU>$jO`L^-hRR?>7!F5vwWrUBFZ-QHsK^($|e!$_*&l zgY3u{#-a^5m>$7_q7nR8B{6xB`we_q{c4{OweR1x?xvZS#3HRz@NWR3Hd=aB$vO;Z zK8&>$6!?*k}PE5ahbs6D|##SbN$%gBn2efhGEVPu_3 zr%Of<06i|_mQ&jJ9TGoH^^cb;A&MEX!H&@fB?Wb~zgB9rzCh&L1nOD|ZQ#;fZgY>p zv>|RADTnODzLBxyfelUC`%o2GF)XJw(PBEs&dJ%Mw{oAP0H*oMk2(-A5+kNC?!2g* zbzC+DukzqE)()2PI*Ko;)29sH_ITi9Dhm=HY=Ng{Vq4TjnjbtY2DvxKtrabiz>0$~ z9Nx+IuR@2?;jOXk&J2`Qjjv*+#}uQe7%_1#8rQ~mZv+{%g^)!HZE6WUJ;inoH7eJn z8kAJSfl%g)uDduZ7n7`{hRgCJc}#x-5Ca&5Y-<5oyMx;}U{g@oF6u);AOMfe_P+e5aRn9Y@BtGM^u9_$Ol^`YBpNO_Q_uij?xOEME=8w&A zoYVs1@h{V}xAb|AGTqhq-VtYEukEA*Aa{cuEcd16rFFGLTOKYN-%Xz<#och!6@sky z6i4QO;y!!4&L|3Y?l-q+q}dp17kI~l2CH`c(? z5noo-FV9TG_&QDhU`sVSZ1IV{Vq)cA7dy`WT2)Qj^jdhg!0^VF+3&_Pz15dvZ^=t5 zEH*$`8%$P8R&rcv@l0_r?zta4;I%%|e)}wSTF{6(7+p?%)U($Xn8t1;X+eg*-#80!J8#X63ib(h%Yfps%ijIvXu^)`lh|27Vo z_EfPu=`#u$D{z(;E>21;83aFHS+ef(D_)c}2mOjg!Z>(Tl@-rhwZ&@gUfCRXPU@c3 z?dc@8=CJmDum$aUCWXx@o4-yR12oC9TMpZ{B-BK1zFuFQj0&C zY8C9t*c3HB5IJt)*_7xZkUqd!8q3m6c5~|+-_&AGrXs^vzTO%puc|84Yi`#bF>O?j zZ6pp&yrI=K&A;w697e5Z9C|{JlQaOob8oer2dp~yj=}Mlz*WrT)E~hf8bZOGp=$2% zW6SizIF?d^T|6R832K`UcW^KszMogM{!xwPxv4lx4Itpxpy>QV$+=1vzh!N*|I43G zRQUGm08Qk9e8Z)ibL&Uz=jPo`3=YLFdaKX?uz;UqPA#C@#@X_zQ!J~h6*p-mzILxW zPaqiAVz)sUS&Wxi@*FC^W?%f1yUbZkSIjHNLIKvixu=Y7rno9 zcSk(QsHot>fR8nD=6ini*LeX#f0TK`oFKE0LX}66{Uw`~2^{+4+$C)r)qxqfsdqq} zxGcWyx1yI$ZhwKnKF|+di`RXhZ(1%$P#}mA3P>b01=LHNawM+0MRg?P6}HV9^mQ31 z3*Y+U{MM_G9cvnN-$Ru5YU&%8wQURxbg6YVfQ9~WImcB=lDF8i7n%&VLA?OaYb^n2 z9K8q_>7ia!@iSp_?a4@sawL^aAKu9MfM3?ef83J{QVdcP37BZ*l|+z z%@n4xI49#DF}M#*zI*HVu3en(=KVww+6*3F;JuK$)UPi6xh#Lvo^kPYUXqdAgI!%Y zkol@`v}&{i`}e1{No@{+_kOUQHpLzjHM~EEqy>bkjJR^24CGW*tVg|z_Y;m-zize! z7n2@EOb%vTJ6!DFy#Wkx-dq$XY9T$h2^l#b$?10UQ*n#N7w_6GZ#s+>KZjaQ?+#&C z)t5J)4e2k9ZEqu2h+4-H50=BJoGQDf_ph5Bc_yxw9t^E0P8{0b(%zR|EPxxEyH)^t zI;F;#_}xe1qQb?sKiFPGsBxW3q_MgubF3DsGNueZ=0*P1FUz7S(I0Ng~1t652vssW;5D^|WXUA8WX;eb8B`U8(UPiKxMw z#9#=4t>&MSN%m;@c1G{W2D9xV%sfu5SK0OEyhh@+Zs>?qdNsQKY;H6rddcrWRyL_Q zSj%W;{kKeFxPD=AwFPLtIAp02w<(@UtOun5!ROL&|5iv#yL>;h#GzW;?)eLYy7pbEtZIlImb<`mq{fo2$aW_G z@8NOwmjp@CP=kDXtykQoq@N^OD`I+Zjlu3 zr(=!{>j2ON0=BHcWxK`$a<3E;4t4TaI9I z3QH{!FRRsv6){oZ_T-$?KlO0M*>{=#uJK7!$HZS5D;ibS-JjK*wDL@c`r2$(RZ@^! z^qM9VMz^4&?GW0pe8bScA3{F7J4PoQ#J#Gx7+3dx*F()u^@}sE5LsxkTHvx<@}YSi!?Jo9^rk5 zVizRj9ddDn@T$xZdPhb}Mp-hbQZh?m!-t#v3N!9lLVRx9Jbl$A1W4}QN`AOdzC&4k~5>@Ch(3F>TOJ}vP z5I1v}{Q$y!_9y)yt$Egdriy-X;*-{C6ji61E7pi)GkqR2NBKCjI#_?FY_+P8Mqx!0 zOucd8@zXq47z>9Egrf1FItyJL7iTM*Ssmm3VuLNXH&?~TN{04p{_gs8vkxoNd8N$_ zu}PmD4{zrOfRRht`D8bir<=kCmP1X%CiexV3lY|D=UXN)i(7EFVm!Cd&ZoKhL`z;2 z*;2I;c#lOeHL_;cAPV*9f%bgT{tvct@+cZzc>j2q9UU-w&nROs8#PpsDUHRXp6j!D z&=98{6p)eA2k`L= z9S`9_Ifb@4u(g}oMq4Mw!m_KdxS_O8hlPxWR|ET`>RSBl5T4vG>C8x5Ic7HPVnZ8h zgLrd-ez4u%-giDM@X#fZ)rQcNlj?z@!ODP244rWJh~T#XQ46o#S6N@H+)axG9k%xt zwlTunJN*e%WuI*#5H3tKS%IL|sv)umG)Y!1nSR(QJ?p(B3q)%<9>8{UHcMv)SaEZU z^oZTohT1(<6R;}LW}>j6mUL~s0l9Yn`0nfh&`ecNHJ@K$SUiFQ>SwB%+e5oqlLMeT z{xFz|__5H*X8D0_Sl>}+#VU@KHm7zcJ8payQ3ip6?xyi28y`3+7ZR(*^)VJ23t$3& zAe^$B)wfP)gU3hR%xK;AU(=XHv=n>1$yvuhckn*X%WIrZxCtjTy$>EsM@+^uS*dI@ zt_9`@TsLq*{={MtsahFICm%wT3(KfK*b49sLQ2}RZckE+-_dT}W^ zYGG_KO=B|}++a`x{1b|d@tu>61v0}}KWW!^-`aS9B4eO3ZBoB~b#m|F(G~#{K8Ujl zy^GIoV0gnCj~i;^b#HH1tG0I={w@sdCE<-(_*=NeK%Ryt5Sic_h~#W7Db%yJk-=j= zbokYU>6E9RR9hUtpjQL>tZ8%>obKmbTXwM)c>rILV%5s1dDiaf2QDof&*o`r$2pT0 z28Lcw55jW}jk0!MA5#F3%-2`?fp%dz3_tPOlb{T$BEqG@P}0-pI3NV5?hyDjPB6SV zg!aVGa8@G=oQr-1f(n~o| zzKi`-Q;~f_>?A4}mPk@BF%&V<(Vj&dUj34G7Nw(j=N&TOeK9qclwj&gSp2ocs2gJN zOS|;f0Rv6#i6!s27$MwTheuw~i=q+K%*G?F-|Xh17I|Mn@G`^Z0}jQsGCOadAvB<2fWKoD#3?zOJ`2_xKzU7tI!_RF8EI5&#O1z zYyDt*jR#PWPt3dPP^^BSJaJ~@f zKs(k2)=Ttfp>{EA!#Afbs>XA;tM;-IE0?GlXQMtV(qnV~Z@k8lbA~Bi_{HyVNc| zblA;VSGnI)o1>Sb>(Rlz;LMa#O*?2-PyWG{3{STe-5?yjhb@@{7q86EZrk=5d8WS7 zA&5Uopk>U(91$9x{6RhHJTX;w=BY2+|0<@CbhAS{&S69;b*AbFAU$%kfRCd2U+|Tg|5SkdA965vt(1;S& zGS{{8D#Agab*8-49AjwZCFy0fRUDDUG^q~EiyDFda_>P;hE_;49`DtSddeses^d>7 zwoqm})$xDD%>EZna!!l_@rp{Zn~uJglf;qQj~+*B+X?Bqy8fC$h*(J7kOP4sMn{rD z-#O~v^ksxUTqyF}nKI93C87Oc-1{m8=J|=@V0+mfk?gtS#<8yC6;~Ih>H!8hF2nBZ z7yCFxVJm)9G>Av#H}(N3EXBVF4={68NQ>kYt0s=JUZm;RyD=FqF4ecpYI089~!Zw{B>K~#3rWfX+aRfNV2Q`zn0*bNAKrBPbm07tqF`tAG<=fZPH?_`PD`)I<}E>7Wm(jqCDE5nVS zelGw#RJo|PcP~wP`AghJEY7(3mTwv`W%&L=kRI+)Alwp33j&)Zp}g0g#}AKths07q z3LojWH>3BL@}Kk)Hw-rCb@A?pQWYfQ2SsUcc<#~}+3z15l2#;rY-C@?{weHUZd5!N zACgLUEUfrc3sK^uEH6AWMvnj4^EHcxRO6}es!!5ZU5_-MehH&J_U4eM%GxskL^$ME z{wS(-k|ibCOkK(;TgC+DVm0#c>rSo@>X7zC3X`nK?4Uzma)F`@kQ8^n?s{Zwd#N=Z zhiMWa``gU!rpK6s`KycEnTZ3>S4rGQN*YP_fX6|+M;{^v)a9;ZD+Lx!2SlfAZ!gD- zVUoZNTRkq33Jry7zP}!MYzTO4n9e{;X8KD}mD^y}x0Zh5q3+gyJ91OYbjHFw2b;O` z_WRPI_mX($pBpt_dtBDa^IGu^9Ks{_LVvI|h_~%7?KC6B`Ep4|d{>3a!;7UTM!Vk8J zXR;nW#rfgF-ycx3t|ZFsG`P*AQ4SrQ#m;Xo2Knh-+iP%Zv`c{w3A5$MV9$t31j8%T zGi?rLEW5$+lGrpbcA=+-JtV;ATa4qUO+4_6Uw~80ZqtuTCC~pOt#7td#rih2L}v~O zwNBr#iI==AgHr*j)o-DSI96*j__I31=)CT&5W6fZbu+uHRjb1;Ht`PJXNxKQvXr}V z<^<7ZXaEkNguS#cVb|1D-vq0TFNS2X0w4e_T1e6H)Bl?e&A*foKPB5ah_T(W=#dJw zx|zeh+!5=*-J9O(_Y;7Ke*HNudWSrlVq6L*BrWm1)!_&heyyrV2qVGCyYX0L*boEp z$9J;B%eqXhdJDWm%-a1>d05W|#?VWEXF2=&^03N=O+!0w%deY)?!+2!e-VuRKtFgy z-e-n%N{2$Xy7w_p6<@i84w62Er-pa7V8G!jKiJYYPP-^Z_V(+oS*^mOK10Hc;^(E!WASw#sBDhJVqG=G|N`vrO~) zxiR$>)J9R&Sf@aUlrZrGB5h_W%bvnh+;Z(Mfq|sVQgahWbH~XNt9NCeDL|wV*et1E zjI|i4LEo8$M6_N{<#yWf@311%zA@(l3)s!rmQzXsR;Xt|828Td-f`FRF1z&N8ro+s zo!>7hNtEvBHudsAOu2L32;whPaK_>k2CVC6wrzGUrz#59rE5OMR#&>euW}vce(ai= zQn=8J!DbDpXy*m*Gk+RoryANzE)0U|rC^*IU$t4O6A0ohV;=4Il`Z;LNt5;!xj=O7 zT=m0L5B(g@rWKZa?swg5crawKK1X&C!ah*G8x{n0+nK{G%vVgPCiqNkOHvlkOg~ph zNwY?xkX4x zI5ugNe1M=zsM+x+@vS9do>aiCT4#{9&zX$M`w1FWwmW^hPiA(N4fX+{zaHHK#97R> zR>bw$rf#5xT_mL0UbkyG9{*B;74PQje$l_OZlKgloG5t7GyMl!fDj&FsRMS0#Crr+^W~S7t_0wh^@>K%AGR7k8lrEQX=ry{BZjy?NZI)6GQ5?*jYI59j6W-1i--n*g!4wrfDj+ zYQETNSA8D(*QHv5K-$HWn6eb!K5cw3&-uZ1u)g%!*Ltd;LYh0$QO~hrK*d72^#p;( z4U9nt=pKAJX0A(3i|mHJ=$Cp_{~q?Cs9GZ(hnsU5>?=DHzhlb_s69{~dM@ojQiB5N z$r%19S5h_O1&3eXUdiMTdt&8h=P(T`8eqKXrWA8S@yovPke1HM{0Tu%Ra6%U+ne5# z=Yqi2d^kIFkh)cbcL2z3g&)#x7&RvdP0gK|4!}(n4>(E^Iu;ft5H;QhDARioXzjFd9d!}29Kt;+cW!?F@6MfX?!9wo?wgt2hG(YHBJ9Dk_@$&<8ZM4ELz0=ve3&n3!NN z7|jD#Br7u#$_!)vuM-0NqhMk%IRrw^OiM+}{Qvv9Z30k12#-mOfCQ`nLMj3v6~S#A zfC&HqfC>IH_y4*G2!SBHl@Jnq^K(i70T7565QFZ53BZJecL@OhHB(Wu5cA1^X|#-K zS@|9Pq7ky%S&)18pX$7G!j&}-A?0-SurZ4QY=ZJ8{(Jim6r8hj*iB>0g<9wopS$4u zWyGr_0094Yz5l0Le4_{zUQS2=A^;MCKw$i9Bmxi+QUO8KEPOIVG+M02w0M=o2-)a+ z+F51Ii+e-&k!(-j?cYuTNP+kVsen`fX~6y$RQv0mY}F9X2Q+=u04xMfqSo>B00kgc zQbw8LF!ERkAp`+r%;e>(T@7A|^&CXKlp|uEhzn*i3$t@ZAFkVI&1B%)wzHM4V_X)HlyMgA?xvsD}DsEENzPmM$+e$ z{xZ(~N^KdYPKY+z)eOo=l!(oAd<0rN(g`G3cgBh-{kR@47!=s&mzKhTc7qZ6C$FZO zK&T3UUQrV+XslLXrux%zfHC)0G4=zGeK^5)tzHjf;*Q41at zdpVz@%+d5gUG3@dx(%TJ@>tz>jfoWOlhG$B*r)||%P@q7xoFBgE_7(um%uvly&RAF zBVy6?i1B@IVGy|`weS?3_7Um9BoREKY-a8jfRmO_x~v`c)ax={PfRyg@LhuL_&*Gt zm+|IT^*!Xw%AxMc7D}k={353RFnMH7HE|gl6k27&WxKZ}5HmhX9&dE%5n`Ygn+tdigs&}ilE86{Ve6Ix0G^|e3L9gXR+ETsMMN+8I*=2|a*@Xlw4 z?9g%=#tCv6KzcBf9dvLtN`@Z9Yct8kqV|1Bq2=2hT{ z^Ah4-17#*LcCz7&L|mQJnF95kzr5p^{C9%YSO)qRl9J1Da({GeUP-kI&jH>^`@>V! zbcXci_feDniQU!b-tBo9HIuu44pLlxGKJ|QT2=pOr9yDcP|@CK(${~z;%~rr(Xpy& z#d}jiKrXkAeljNbT~S&`*6pYGi(rMs=j)no3n`=&a2}?8Z zU~GA4heMP_or#at0!I+V^J^MdZLkhF@5$F*wGUHh4jwduQv3CYi%7BH)}%uZwK$*PA! zu;z_3>)E zOHXS^gC1fxr-;mf;{Ez=&JeTx)1s7JDOu1w}8Ssk!%qY%bR1miqfx= zG++E!jzX-)3UmFL$fRh)zzO~qvq~q!Vh(6FXEtsDqpCVG09O|O*k4`K!P4j#O1sw% z)Nf52gvAUymbbd^WQHQUvg@l&)c88$b zNgWw8`Tjf6LTecUT~X=5KV0+u7k4h@*}m5tifXlX2rRQ1tvX=k6csmQ8i@oZCq9l+ z*R^MSolnm4h){1~$c4QUGPMsyz}KDf?z%*F$i~b=7b_d272ZqIjv749__xL)RiGOsysO#h;w%aovaIm}_DeKn< z>R$;sLAarc(NNnVx|MmDfWEGb=9P!#Z@G$XuTErqHo~;25EFgI<}P!|0_^9GYnWNq z0dqT(6d`b_*7Whr5u1|hb*tA=2~0{BV5%HSm~ zJ1ugWsC17L%EG^lEWePTd$Q@>Et=a!rMUZQ)?M(`RG~4Bhe}U|PQ3H94&zBFH{9*o zOSyJD<6R!7VEy&$@T%-;#kRQdr=05P5>4WGDf+kJGi&~*kAwo9Op+o2vUl7W$+)E| zBR#(%oO|vF+GblThrez_3(3?XeEgW6$7n=z-j5Q6GFx)mRtNueDEMlL&FUb#d~yoA z=P|A+%&%t2 z+ym5*bE(FGMzqa>0fPs$&nA9Kybi{ zeQ;ITIP4;Gp#kImwm{^og1%Qhq)xn5ib25Tk(t6BAxz7mo*W{_VSY^f3ayDXDTyUt zQ=?ZC>$BysHm3-AHvLUnsoNt)-r0$o!We(BBKlAu0>?u7XEk*if7?mVig=_aV<INgPh$B5u`zB>|yxOcrK6k3%%GL|Hkq9*a5qTB% zPCVu6cacc4D6`kCafVZLGJk`%at6%56j*so$f5bk^`?^y-bX>Y;xm%M+oJa#c88Xx z|9n8#A9I}JQz^Mf%oTem`g7Sicr&cLsIh1EK;Pz5TB&1vRB+y&L}atxkH0vKAP=8v z57LPCUh+T2<3?MV!!gmH(3j}M7g%128^~oR3-)|N1Z(lv(0qJ`NQPTG`e3Iomlj%@Tl~)n=BQPA173v3q z#($glin4d%;2m(-&rZ07VFkJbIv0K5?%uY4nrWw$yc`{0*9J#=nvy7@(Vz+ex7ly7 z4fLG6BEL5aVM<=BrYJg~JrJ0fOD1Junk78utX`TpLrV@Nz-eMhtkVr>J~GA(y2Y^v z`W@E2R@Cn|GNFI2&oGX0?3m?sveX<3b$2c5e>b~g&$70s!n4PGkluufL=J{622`b~ zwHPGD&TW0v1Ff3o+KF+A5dBQ)V(FVi^;e9M3U;a6X{#!!f!2qe!p`lN{5lzPI}LjU zgmvb-{F4r)AT)@?@l8j9+Kwc(`w7+j;T4rWz2^lkA{PfnojxBLSIm+`XA0>*8gA9B za}qGZw3vR+L{(MFYZ!7o{m2Ezj`v%*wjT9EqW zvxc+D`r(WLRg$sgL1f_l-LSI^S~9;+Pta}px@=pao4vYG6u;$0H0xa46mgiU1z)oC%(&o4Hn zFclPW_@1q`X+QntKc`3(19CtaPBK;$~2qgAmH|GKRk{U$QKZ> z`6y98MESagPR~Vgj_a`$i8}+uSLaL4rsfGF_STn+uZ_cxRBo|Y_kLyLy6@9Cf-~#$kXo;P8F-T&145@u`z`Y zE)@BRdd-77V`2=6dH#)~40DZdl#7!mKxQRkhgJ8KsD3|sv*(fRNQW;xt_?csaNoG@N1%V7!J(V{s8 zrS4ga6bbp#ozfVB9ZOvs;71jk=YoHxDsBNGm?ltfg4>*=Rj$-2S6Sw-Y4LG)H7SFq z!gQ-NJt326T6h+Gr>Hzp+Hpk{AOt)WA|Nk`D?Q;f44 zUyzofsQqN*#kU+SZ#}=H$JjBq?mMm1&`OfvNF;J%?08!Hj!b7jJIW$v8$OlJV*at< z7I2s=9-J1Q8Rjul7Wg8j7Qdcl@@ro=2uV_R0GK5m60_=cwb1KZNu^<_m1&&^ueR0q zs0J@JY&N=rK;38hg*Tx3-A$#%PvmG%0gO)`M@uLHd;>8yYRe#(XLS1$Wv2R=H*?( zj+BY`Y1PKcoCV6~@X{p9NXCY>8S3BJ6hBzbGP7kY8LjhSni+1YR|aYqREILDGFB-H zKsWgT+zXc`AtM-wKB^;KjbkAmo_G+lcKt|^j$nmy*k5W&#lQOj+g-0$(n-?1-+Wu1 zcXH8N7h`zGFT1_sm6eb4jn2#rs-x5id9g+DlwQSId+!`x8vT<9TDX^^4i+%VEV(bn zx2?=|7O-rqOLZ|jiD~$)E-?&UAYSntm2Lu{yYnZVr^xR~;KDB0B{!S+Jwd|=GBYOSfty#TdDmM&)ov21 zwj(9vtmtj%ju$7}^g^m_`s#x8cO}VP`(uSE$e~Kp-W}I+!~Gvm_1(=;VK41EHb;P~ zOPW&vDNBJR3;Sc)%Ft$qfzmU;^>f%}ni^g%Z?2pESv%# zL`9VfY^IA2n4FciwZ+@cJa9>*V>;!ySei$t(#3P7sO$37=X#iLfRUE%s>z)jzc=L~ zDcwTO>(m>CFb(z2cfopuYYG*PHb6DHau6OG}z#6vxdzP)_$QSfwry1ga& zGfZU7mpU(re6G7MKgKmhhPu0xuormp%c?T-uFxltrXK7`SygqS$f76@yGmv<98_`X z*`XM%F@X?+I*n-y*=Ws{lOTB)mDXZ@H>}_5L*m`>`6*+F`!iUc5wqFbDRZxuC^bd` zhbftp*JG*zxH=0IMLme#UMct>DT6TfPvrr;gf>22!uUu_{F((F6J*7-Z7{uu^bw%_ z-F!dK_WY@eWT@4vov2J#ORrZCr*~YX+u*q#jY*A5VKpYWY9F$GC1>l0f&j$}72_Fk z7x4Nf9r|C{1M8XUt$?SE(_95X-;zWGN}1gtzo_0!3Fn900xH<0LmB197rS=ObEvyT zmGHjo;ftluO>1_}zswagR#Sv}GS_TVJ-%yBMoY}+y1$Iq>FdKZWU-Bog(0N^3w<_d z$5p|lU9Tc5Td1g|Lgr=#D5M+%y!xn^e-QEA0+hSMrhXbZY%GI&7e%@gpa+UW08 zhebS6^{gR0pW3yyMwO&Tb;9cOswYblVbc%eKn#%Gg!^e<%%n~^VBV$itX<|dhfBn= z{bpq7-$cLJJV)asuc$>M`VGP5w*c1{TKnOO8M%2uh|JFJ4jW7eswUkuvjMlZ@#2Zb zSCcn}AA^|^CE~6mUZ?|z!+Qa?LD=tU%k^3z>BN*SN@wYQb0bKLxeNsvsWeH&$}ZTHOev;jriqSMK!$OlXxT@hyLGRnpJ%r>r-Fq`ZR zODpW229ugz+|KZ#)-#ROp5~ENii1~AlpR;mUG#<1^fDtnV@E&sZ3^OYm)Yt+dw$RT zMP`$_#4uEg!Tz-^Vwp-@YVCm^s&_di?Vl=Itc$}8vQrqiW;rP57@=;oyih`lAFDqo zTPAYe+tL9Zd!StlbBJQg6BE4)0YxQCHoibT#cv+6Io7LZ$oN!lf7kpJGDjR=k z9qeQN!Bz^ib*do#3}SX)O6;*- zJnocf$`jaRJk(KAtkm~oqW+9gi(e$Dt=&@-O;i=b@}LX5QHCw7R52^TJLD3=Qg;Gs zjQJ4dyz>@cg7!!{aEHDQEyDXwD7)64C@9viLEP01vGMZ_m54Dap<3!_$i#1AQ6e}j zfHa9>(oda=e4wM9Rvn4k_t-NRN8izHk+Nq!u54fP^!jtRXmQ-Pq^0!ay1hh_OzY z@O@nx87{*rm3C29#6v>y@Y(={Mq^IfuKWB>lt}-j&n^wolzyD&kpcqQ;&z23oL${? zy3~+^KuUtf&QJ-s@FW2|EQ}RJ?F;xziQ4g{!THID;l0^bI01+^97Li>DG?2Sz78Uw zLx0xZ8(A!5o~1=P49`RgCxRyU(EBDn*hXB#v>?5D9K~{Kp(joZ9^d7$UZ^ zwTrmc{tP@=4ZFX!dJG%m6s0e#`txa!IW`hEqaOC zL9+9%k*}589)xKIFwdpIQo?P~R!qe;ih>3q8AfgGbzL2$3ONayNRWB!Kuo>q5rlAP ztJ;U!4=!>5V}7ZAzqWsjaf)NpUBXoq!Hy0L?{@>$-qjbcI$-{jgnCmkvl&w_|B$(Vhbiu3x3UOB+V_2N&V#YG z=5U3b0~z>kjR+n~;*mFw;QxMV(fXVBq6@-fM@NATTYJDXqfwt(9#Ir1@yi0KtlfGX z=MG^@4vfvxUkwxN2X+=J!4qwm-Ro}wMO4KcS?7hp@F;6TEry_cO4OlydWW zvJ7F{Hy~gUh!_y24WN8dj|t_m+gFw${6VW98H%K2*1#7iN*NG1RtbMp{|~(W|E>_t z^6|6)MGf65odho<%6A_MhyZ5H>IacHw@NBzr@B+Evbj5QOi`c7%IDQZf1*9HQQaym zw}7C>hmSc0#e^Z$T!rm(pJBbRSs=t@L9kEWqXZoP2|?YI6laAjRmG8(_X4V{fk8^? za^1ZLwhQvt(zr0*VJvjzjD$z6t|d!dAUf+W6E(>_DrQI2_g7l(L55?;^C163tFRWH zC{00}TB=W@gdv5=GK|Q4@@7d5c&@g#>T}|hRHI|dNS_8#yRgu*M22~-4zY?M zdD577T#BA@-FFGhW1&Acp0x1N^p*dvcGDBT1&js!7Hlb&5K=KZ@gCotN;iKt_>wV$ z5Fj5 zQDiYC++tgNH4^r9H~Bfm1WW*c5A_B)@IU7`O^(wJRp5BF|EuTa-*_BRAhK$APPO?G zxEdI9A`7y*#&{nP)p^)muQZ6vqs}uG=bBIn0EXeW-{&z-uYPv>zkY5iRu|m$9#tnQ zI@^t*){AN%l;=GsHg#AXzXfO)w=)P-8oih7yobaJB`hqset0-HplFr#L4Mc9JO%?@ zR&^JCm>mByATa;=X>^l!grq^|O$H!fcy{**5KG>|R{jA>;pcRqW+vzmCYyT>MK$g-#){;p zG$FHnduOaXdt#FF{E^po=dtzTTNKH3|4kLbepq}}v*{YqX!ZpBe+j`kvzNa~VZ+j= zxIOQt^Q`~Y4Iy9k8q*=39}xXZY568Q?mrpS_CjVeGMV_h_sB0-{9W^xzqw7J_bHMk zXpYf+gfegwgIjSTxI;`Eb-3)BlCdskQ#QadDX!cVFTn7kb zdK3qy${Nf8go`A6W860JKo>}M>0V{f;1&jZB3!t6; zs=W|D%P}lQH}%!}pNU_Ps>h$m;y*-hHj9gstuxQ)#sdilnbwxDRpm6#mYN++OZWJ; z4@sgn%Iag+R%uGY=NA(tcU1VK=>@icZM8s#-1h6CQ<2;UA_6fq-^Df|Yiax2_Z~6t zGI?-1|I9W?82Wy0GkuonFs^g_m|5J6W4K6zi{f`4dgtv8@sm@&tHGUA={-K5+FJl8 zo8VskMFl#8G>^_n$P;~SEWeYBu)gn!M|B_Cra!=Z8$;6_yQgMGoH31nXJpNW#JA$I z1bQ9jB3O3NY(Ig|Ixcw0AHx^h-uv^Oh&WrR)!5$x?kd{4tiZBerL<|^hMKIK!YDP( zl~!ZIk}U)%y>VyNN=V4e9Htv|x)i6Rb&h#uY=5utK?g2%hjyXw?NG@#S;fxZR75y8C zcW(a$x-u%Oq6OnNLG`SYhV-+Ysh1-fh3onQpJne<(snU39h;m9@-EDUqlO{<61*V* z84x9iQs)1qGR>ADSUJlX-KCACd}L6sbsDj+Btv-Fah6Atwv;Rb%(ejKk4k&sxyahj7>s zbvRHXP2Sci+0^SAX3z@4XlYoAJkihm5x-T;*v>a zY5?^(MP2m2cspm*7h!g(8ZhrO96{G zjlZFDjW}(H$>0@mGp&V-s8-$}J!E``vZ*C2l~OdoVl1qSiy(Ye>4ur>PhdNNE8ETq z0y=vOASqI&4ngv`l+cj@$08R;&kF;Ox}K~}XT$j?4(sY?uc;*SI*>SW1Ov~_!L3Tsn5e`a+0T!JWKvidk7lgzGyx#hRD$ya2B?dwPGK|Y2 ze^);1xQBUCx2R5+XI5hQ@2F0eD}$Vnth-2i|N3|tssN?7UXDL@F^6(w;y;=KNZLC? zwN`_d88c;opr{TwRTr{>I?}gVrzL)#e^9mGOTsF zaLGbkJMb@3;EqitS@Y0w%2=bZh_Z;tu{-Y`+XS}ATzkg6nF7=Sm(QHzf*W?G zwU`$omHM6ht90RHr%KV6M$WS64SJQ_Fwxn6hk5n&-7FIZv+_`{YOiJ_S@kSeht^|` zWMvVEmdF{ajjzSWlhbJC_tI@{nM)EncHUIf%!@fs#1>bO7rj#C2ZzrF7;{9BCRyv6 zA-gvI%l&9xH_B*1UE1EsK4(KrkSO;lSKg!+FRu-TUx-AVLtX6nA?VaF^*1FmL-l2x z=NOHO7X~LG_K=6Y*e{4YMr+SrVgvGmNXd34L-(@_6sHtK(? zcq8NzVtk)2_;LPk!vEWyd!CfkB(~veF;H)k_F>)n#n^+onHGE=eSKS4RPwWzPQy^# zV`loso~^0k)BUi_d>JdFg$P7!YQZ^25-4YH6N* z9h+)od67^3&yJgTYX4;T&##hkBWoLvBc;~A=P68vX9fNFQsmPFg8DJv+15PyU}RA_ z6?6;u>Z2e=PO`Y}`}=;m)JT*=U|IhPfBXa+hMl?rNI^DB=mV#GG`@Z-hv z%WR0m-jrB^Q^-6UE2E_6d`f;ao}skA$G4wABCqz(xr1~4!9`!%;z7om&nljcd; z^A`Wr>uKbDplSQckUG5heg6%Xd-M676X)L#z!nTFBd73H+36QGd_M~My4nq__xoA) zU-a4hQlmRiA>+7Qh37|1>x{ol*#Q-48SKsH6~#ywH-9qDYqdMi5Xk)Fos=O!Dg8iJ z<62MofR3bndZTioOWSK#N1*PgSA`6-p^v4j8a#gAG0&o6+vw}Fc(^_kAc{mf%{oqZKG%|3fQ;-0f_NvD+@eX^{`VH( zq-FCBIU zl|DFu9QZoGaui7MGBZ4%U|d(>yG80Rc!q<|S>mvZ$Sq*WL)vcKH^cL`PAej}?pG)wENU5 zvuf-|dk#@?oJ};E5g2?D&P@>L<)f)x#^d11e(AS>1mEYn1WU@w#U^EU_x*d!9^k2M z?Hz}lh-h@=?#+stz*kofnC_p*uY7UHDvkYp&u-=Szm}KIW2ransT6_cG!kB4NxxXT zhgt`M+tx&2JmW6qhCXvtp55&3@)q1RZL=5D{#t?F<-GI}8s`s_RpXcs45h@2MuEoOt#78)J_34&I}(^o@C?dCLz6%7CM| zECMf>|Bcc*+#TBtBUr)*ZN_b3Dy^s7f5!{QMi(R!SUw2AR;h};Fjmqjp zrlKD8S74lZ`MW-DbJf{S0n_C+w*pfKle?Wrf}Y=RUT$Bv5of;;{*^#3df}f5DV}-^ z&5Q0!E*$j>3+|#nDgyg>}%a(@bDCk^`A8%(Eew*MgZI^^vyBOcxISNixk7 z6ccu~Y%m%E8+_!E=SZE(b<8lqzjQB*y6^6YAV_~o?BPV+&*0YxcT=ZRSMF`Ez=xFQ z8-0dlCVy-kd(J<7VVTdjW@Q>^q?d^z==E51>&a;zTY{|mAOHOpw)9+FM0+ZVxu#9^;*$=04y&}^beekvt-lC-NUzu@Z3CYP z9AS?2?cW}4$^Wl6y0GG1y^$DHf;am~_OPgSp{mf_@P}()~AxE zE%U_7ALvLTR@dk`lDbp~pPB5(zI}R4$twvtya!6J)LTs_`oSsldU!pJI z!9>+t<38uwDNL2=dKv6RTUgs)+m;%5%tZdsB+=(oq|xF6{_VZc{>_tHKzIJsa?blw zz6m5f_Z$j;yY%W1HMoP`@LqAnZw!3mK|aCu zeWcP?sPiwG2r5fhs!H1%>MN|Fex}b#y?A)1Wd25lXMlpe$w`F z9QDcT`>2mnGTB#W>@e+33fMkhzHalsR%r|R&%eLKY$P3WKAg9x&KMD%(Rmxf`BPQv z;Lb9BnMj#Yk$0@&TRHQ}Td!w^eTs^qCtU`*R?`R)AVQo+)aKW-f#lD+u(RJ8Q!fe3 z`!+oXwX4QnMGd|~MiH<7t@`tuFSzB82o!oRn<4omMsvp0Q>U6e{!h2ZAn$^l%AE@- z+9ihIx^ex#&JuDd*P1DTyIQ7=lbVAP6;}_87Y3PZ5zb+DlHEs7n{zV{b1Uy@L6tS0 zN^U(AUvw4!(Yo*=x7R(WoLoXyZUc^~)}X?(WSMPTo^1U3{S>Kffjf$gw=vB5Q#C9; zPpM_LXahO$KGCznl)cuT_mI)d--?f2tNSU!oSc0Nh=qQj-W>U;@Z=O@iMiB@%GQ`< z8Y^KJyFhzt9Le-i8vIP}sR@Lu%JZ9S)xOYZl2E0%1=z;j0>1P6Kg8Cj1e^svu^V}D zGpWT^{yO{W@6V#fv(P&$v_V6971Mc1jVH9z+0l)o_RM>lLRbmr%q*;L6rPXtAJO{% zQ89=Fz7gsdmS4tcgcANgNJ+{%7TWm3M-f1d2slR^?35W+#|U+ zYp9){fd+?nxg$A+-}6NKZ95x|bpBvOyUuMLu8sv^vn^j2bG>Q3D7Z76d?}8iX`AFb z4(!)at@6wC4}+$M*oh%|+sC`K!lE>ao0oJMY{mptFRMhfYnR7y>KPTQxwU;eAO}*p zR2&oTa7o2-B6aUiP5Y^oTt&XbK-3;dynoLC+Y^Whgc`Mng8nVQJ*W266x#R4;tUqj zFkfe`jZgEN^^`;*;PfI>GY(l$0&s!K)Ph7G-M}`NA6=j`^U?B#ql-Xk8;GRVCk8oeI*yNv+_nW@`L404UkjAP>n5?*V6Jq!j!C^t$KeMU{##rWnI8j&{ma6hlB95f}&U4T2)Mc^qu7%*az&XhM?Q~^0RTs)M z%P)Xm4C^a8`I=CM9+LK54HT<%9-E2h+{=NE0~iFd+Ll-&?#aZZSx zcG*JkdptHG@5%;s&F<7iLTMKqIt2WG$*Tkr*kQzTcKJFzwfHi_0!pM*2%YgG7%$1zHu;3EinOByiZYc%1ceesJ zo>q;MUZu@9QF2ytX6h__Z$(Z{PYyFE_U1Pz+BWAqZzuB)3O>{`9;$7UxoKn=@*6`c zwKCTn<{l!Q4ipLrqGrxxr_#ccwIiZ?7kM|BJ)_7t(pA2u?J`gl&s+}fn!*=+C2%Mv zEs;&Q09RfRi(T~swS@NHj}GIJdi7D#@Q?0wn1jZC54GHARv~ACxDltfT-%+g$=Z-WN-bMS}1=-H-oP19#jE=6XJqRQi zJUo9$d(O9`X6@Y1Kq5sT7r%%{N#xh+otz`)*{H2IlZZ(_(SIQMif zY!pme*|LSsYBkJoLUJ5#hzvMbJq>#YL{)RYrZSAGOx_nBUg8{qH%9q==qqVq#*9k@ zC!>sN=&S)pg&Ieju}1D{J7iT1dc|WLd{LEb|Jfkk+>M$jLLvuEsy3aeNGS3g4e}`UuLX&mVEwV%aIQ*@a&zRrbb^tudG*wo!AW6;;`DvED5DPj?3C%^I@LV zPp0&mN>2HjUJh{-w4#OOJ8wPUBo$aYRCuI{zg;o)#&MoCmhWWtfRn(gW3^wmxIJ1M z6}BKiQH!PjTV|Dk?LB2Bx;Gf=e~t)@}%d{}x#Z_vI@Iup~I&HZ1QsdWA5 zMyU#4C};jfT8yytf*CrrQ=$Vpm)Cb@JBl>fbUAh16ykg^X8Cqa zbzXe&C{H4O+o|)!i_`dHDxu@9j{b>=S$P5lRm~0OeSQ3t3W(<_aPZ(^4<)nznd|2X zH^Enf!?%DTSh9NlP4qlcp%-yjI(-Z9=7bLKl+n)0O_Q1{BSe^iY&x$+0wdxE0_o?G zt+q0?meP`ArUG$-`;E&6Z?)m(E0p+Hzl44h+n>KCKD9onv3|YK=3dT}^CfsA9mPrD zDp!RQWmYtF&NEmt7D`8T6`3w_*#SN7>xdPL6kkX zeG{CkK+3Bn0}@MYuLi;{@t8Z_NOu-@*f48g&`m(@U;Hq-$kgy7QXp)lSPvzLxTaDX zoyRXv{QoXye6y^Ece4jdeL@||S}(IxjOiTk@R?ptABk|fp+K7@g0w<58Q5k{eC+dAJWM$9d$zzvK(m1L|W>RxQLZL z9V6CTpw!=yH~M7rzIG#|OpPSA!`E=O(_Sv}vBLB5-$?C7Zr_tj`2$2wPy;*Mm}MxY z&^6VR+31{Bf4usSk%tLVJbFErVW&*gBz@|$k=X|x)1~CsO~ZzaOMl!xIomX9zM!H@ z+E)Q{Y%%SK${ViZZC0|Ws8>MN=3I7oMqArSi>#2}uRBtBcwfBHeu&vvF#aZb!$Gv+ z%13GWAH|ZdXz8d@$o&NGfc#CfKqd{G7a3?rPxAE%N3|L^u5!&H-lzL+ks+&jb-Uw& zXq=XwQ^^ya{7hF1*{zjgOn29tw`{g@sp@OcnE#QIHf%F)#KhzESmaP#jvZKsK z0BYbFIFB84k+d#A%=i>X>I&9ZzUUuX%)JHZQk%Vx#xmMQcQl#Fr#@G?tCp?k-%F_b zqU`J1s~3CVi`)ml)BY6)H56QAx(ut`XmFmzi4ABxqBk6|jZQ9l+nYA_rpEae@LKP( z;Npuo*k!{yC{O)rxT}|`YON0hlS`34mCvZxPLUJBRCMh-su%_PUH{CzqBzzN{&@?y zr|Io8M!+BX_8h40k= zQodvHKj^E9OS&a`|DpVQ$Lelz|7cDTEoWvQNsgWy-NMRblBVdjajP!IOp7{vnA%#m zs&%RXJ99qd^s@wzygfh{{mP5hG>*Ilkb0ZJKH}B=y`S$LuYuz=i0tnybmE*XxJk{! zui88E3N=^D!w2)jl4^jLyMwg24sYN8oVUC3uvC3W=IiNSvff84TbP`YYuFfKMaTQw z22y?0WU#2e`chpJ&5c=CVeRk^$W+xhjX-T&BjQN7KS~Q|S$Ycx=N$Vetw8Y#MNw(w z*HrTO?qen24vSLZokkSf7G|A9>x)Uuh*f97;~u5X^;~gvX_4dmEBfrY&iQdUB2E9Q zbt!`9!-6TK+_X!#*vs1FQp-vQnmBzcg{QqevHu58UmXzD6SYmVAhncBcY_j3N!JG5 zBB3C)i*$DjNV6c_9fAl52r42-cZoEzq?8CE>399T@B94&_s-0n*|~S(Jm);;OqTET zD(t|OvBXWo`+*-$lJ7>r2h)RNbI(uBEO{)(IR1U4^L|dC{{(IA2KIJc@Kzr-NkI5${3l)v+>5FEp@iUD;{O_XPHlnoNN?w|5E z>h^nVRiHGMi=Xqzi-&shuO3{y0 zRJ9+o;1EF;9TY<;y{m7(KOLXb zj4lq_AGsSUyj#wl>5|C)tM?lBi-)6B4qHm&nz3v<{KeC;P7#M?^?=7|!u|0sqkLRE z#MyfycQrlLXXu3c6YbT4-5gu`rB_5hz`2F_=AP-jCuB4yg{M4DCJy?JzD3CQMV>~_ zX=&$tbt;$LdrJDT?72*!(T{Jtdhgm*K8eQM^~{TuL0=yvx$U<<#|s_*;~iB=8gn~R zVMXy^xhWV=bMLJ}!2*JpVeiqAtclleZ(LrXOk(Ao=4(=v7nXibcl??1@I5NI8wyf7 zJ%Ximw8D+hhojF?_aa-fVh;_rcjos)vg|c5->e|RHFo{!o{@|sZb|#U=eMrd09a^9c}jfZ`KBBDg>vqt)yXAR)0sLA;JTyI%Id;ONoS}F`y1B}L=77t)W z5;FyyJTea)$_9yP|K6zdHKqMpWp}Bqa6L?z~_aS8@A0_MKO*NZOXNWXxJ_Q0z+MSv4*Le%F-P~}d2 zCr-xwf`=G-rVOSnjiFZ!ERGa zv!R|O4%1cY^ejJl+PZ#>98xb4oHi(KG0Utg;(I%;A8}19&L~cGZ!uR**Q$D7o-{)F zu3T-POfogW8;Ngc@e=yKI(=0H>QP%!0-VP5d~$Pw4F2z7XqLbDiS zp>LAGEvMF>IcJ;v?h*cyVtse-7THnP>oI)uvgA{W8S@TbFC`tq61kLqgj5l)+%%GS zK`9nBna7{(mGAi-`H%el|20gAs?wSwmZf{_XILt@isiJKgI7eM>6?%Ep0CzhCe@;r zv(kPKk-Zny5$N@J*FPY{Xu)HgHyj5q+(fTOY!<`mmq8FxZDVShtWNxHLPE*ROoIT!`T6wMO9_n&z}&S zQC8%k!&kaoJ_WKX>d5c*IL?wmWU1i2QN1gLs(O%C)fXhWVU6*u zkEoJxYU%?y_j&<0az2`#s?jPpOwCkG2IEj*E_&c_%l5jXMu+K@J^^#GQwYykYfedh zs%+HWdwb(&kddnF5ijEwKY%2+5Vs7*YQfAK&`xM4k0ro>A1+rMP_OmOaACw1Amc?a zf@2N~$%jm%LC*|Z_!xT|d&y#Wi1=E#$WD|89Cl9mTHF9|h#^vwfxQ%YE) z|5(FI2tq(a_Q=BtG7#5}nrd3{ww+BJR}7c1jNUowl-(K@>uns~8?rzo*FxX~i-rcU zr!1EpPG8EW#r$_UQI0x+FJG`HG89I$+Y0_yB7Wz%{mj%#!#l(Feht@=RcS>@BjVZ< zt>pQA8qYtp>>U&TW3z(8H@e@P^i->xvxc7Ji) z*N^$H+LHC*pj(yuXwie-%he zj`J6X{`&KiDcihIYzYk@2_aATd-cX|ByKhEhy%W_z7rDnZ#&dHEDSpyzo2z_{ugJx z{DMw3B&=oiowQ77eS)G5;_z|U@l@!pjut`g`0q!QmzqS1S*+8FHg~6PDuX}!ci;ab ztK?mt8=%xZcvYcaOL1%}y4gw0YTFGP$sT+3i`0cs?uNZcBlW3th|yYEe1OMW=B^HX ziJVZN4zT$4qT>kP4eOI*!`oe?D6;61Ayh;8>(ota=rKa!cSvREvEC)LXpQHB7MAi$M-TCgY+N$SKVfY(F&Un34TQ*l$nw-(Ze8Dw{jZ`!;3xwDKzF>Fzh!BfLU zSFK@b>S%lUgXZgaL0KQYUCx`W(t9;G!cX4M{47tuZUVCRkV->a7qV1?l20n40a9*t zLM{Bfk@8GZZU+OPWxcBgpokuzXv|}4I@*coe^B;QfEVHE6gtu0lvDMpUHRCykX*_7 zy^R($J(LUor`_rqcMMv#uh8jE4M2nEtG;2R%%g%*zsjHs+i)AbBP3jehxe$ zHW;58H8}AMBypWd6poxWC_jk&%7!5)cX9R(LR&^O9$_v%2$XYa>-9|U9|R z+xkwm_>5~d(Q8z`K^;ZzSlYbA&8kVx#D0m`2CsdI*S*!HR1=Iuqo2+>%QV%Uy$TVZ z=KcQ=+S`?53c2JXN(21Z)Ao`h#N5g-t<`!vCrF(;jZ4t=qrH%BONjjz|K|9928FgI zXY8Eofs&bjB>*8>SF`o>$Uclf%G2lNbAydSW>(j^fse)hCCgfEG-rn#H z$GjJ;r}|hY68M4R*?Oiu9Mv)v8Hp4+iX^Y$S=aw2gG#62paE6d^GZY;KGz(JdG#0( z2n!oFS<<;B|BJ&})6gQ`Es!_4%6wmzyJh|L?Q?w7+SeNS{Q_U)dnK`ljDd0+gR*|j zT4LTUUg9J;*DDKpr~1q*HjAh1bruQ+X{E`q^K|K7cE5s@s&C_JJD3Un@{V;o@Qk6HM?}o97k<33Wa$N-W4rI6*ZJ|vJ`Z<`FW$1&zMlT1D54ukd+`_N-S*AbF!iU&0JvM=W83v4AU3`&WLt4k zOM>Z5t6h!iU2^}$*)$6MkqS*Ox}8-zd`Dok8?b;1bx{uy{gq3ZXPgdKx-kBW(}7hW z4Lf=9QCEc^{{3xW{^@x zkFiP@mR%j#%Bj)YN79Gy@E=v*2#37~C;!DceFLm6f!sGDUZFqI;*o#~%2(&AL3w=o zHuL@piHG0v+9g0&^ym4nTj{}!*y>%rIi>w?{g;*z;}zq;xH7akQoyVF>JVb-DsQL( z-SASmf0%Itl?dc9JC((q05R%b)y74A?B8bi1?ms^opa7}30+;YznMyz1ax79d;_)A z1YPBBBOIsi-?nX0f06H6E5D!(P5EziFXoT>dC$NfK!!SP(QIsIiySi_j(|0WL7FW_G@`C>|Jr(+ki zR)FHHNA5dmL@D=`%N0hq>t8dFP-DajP<&9&(m?gFp;DOR*g0e}=uYF5Oo(Lmt z$7|WE8H<)jm9bY`%dt^Ozcnki3`HN0J=U~Zd}J>`i5lU`Vdc>v^8gyz+-}P8uuT%V zlSps_CU9%rNlyCq>_vGZHHnz5Ti^jVUpvO#hX7djQv_?GAVUUpu?&XK@ z5-)$3SDsm%xW&87m}!YsIi{tq45m$Vx1|!|e{|jxaRk$|!U6%@$ z*2&8}^TqyfnL;`5gI^R;=$V>DucRThz=pH7{EMZ^uUbQ~aL7o?GkA`PWR?>uS6!F7 zh^UtFg+rl^=ZA5=FVYoKh%6KD>MV{Va0*vP^Fiz)Y!hgdw)uH^)HF%Ugo|z9MB;-qkzoEmN&r+sC2o4T++8*wHp(N9})U1lZ z#VVbj3*#D>no5yALALi%HuY3r(|OLh``pV#3g$}(Of7d1XuS4D^0`cD>!Pt0M4Lje zVG*2%FoN1ux1c^a`cJdiW9M5Tp!#j=I!7?n`PkS$-22v9;_wLB|{l z!~?O$E2Iee_lvW*-Ql$eN;FeDo}TCfc3UogtP(47$uc)+?YQP(l?j`fAY21~2QcZz|$ zJ!Y2q06M&xfbvyb!2rcNO;`Se54MB)oDLy&>#hM+qQvY8to%$2#7Jz!GsF_lyBt-&1h$RQNFRG_(8x-0i2z>bQ++x`R4y#;mbupAkZ-1u)5{=RDNqM&0Vt+(%|-T<`4 zxi?+nrAYT^=eZ8O&~6hMj3<(Zx;jfT_=pb)ZY##WsIC~}gjmo$Det_Bet8GDoQIWQO;s5WL|xH$`Dc)?@fU`S%b1qZ z)CuKJnm83i6-0;sLM9q@=ubt|^PlpmsWR%TzOJM*_A-)j;X;h(ZvGPO7b^4pQXbi{ z;}*G{k5(6O)ALf96g?tZ4y=8t{?@9=zw|7pdp=2S_Uxxwyi@Cc|L^XZsu8l^>>&kxr!<&zki$@PzEVa})$g6GF{q9pW0nZ&jL|186 z1xtu3GCX4Ea^VwL8~rNw;^gy<2VzU^ODyS0?2ALHm{te4{^UmUM;obJ7xS65ryH*9tJO|3nW}D?inDeZXfvI=Q1f;c)TUG=+QTslLMiWGY29 ztP>8v)(U=UyWim{szlI|3Ppwv!n|TfRYsci639~`eDRmd(Z8bu^(9Kk1&%+Zz%fMl z;N26w7>;6gTR28^oQGYMqpTaBQ9kY5@rxqE_Y_so%>tmH#Xw@BshmnFs=NYatgl`Y zW9#$rCJA!A^8I4(n8@39d`>irlD!acxB<*)OZ==f&6L;jMTd0dho4y%$|V0~5JdgO z(J#3WaNH&MmmyZN98fJ0hRIBs7`$EAWEMd6Z&d)ewEnJdtZ$Uug#hY?w(F2qiARL3 zH0x%nC1inRfox~}nZ7J|wCxtH?Ebc#@@d$)<|IG{LP>OCeiu!S-|niNn9u&^W4o+C z0qW<8pDG2{S#MM7l&&fu>j@6B#$g?%u^p_ZawXrZ$3wL(ly1syzK5?Tp#I{3Z+~UR zcEULI9lLlAKGLxrW+?F-gaR11!!)7ems*EP{&jY1)ZV4?;wey9klSCJmazAmrH9vP z=O~G&7?Bl8Hime%bo$AW8K9Abjv+DD0g0k?`t{){;@cYdg@o&s3ESbz(!&XGK=Hq} z-yE2Pbr^(dEqI+z7oPw4=4mNTpl##2uHlS%cjv)gVl zRX1Ns4^zd8xPgqIfXg$bLyij(w;LY4s|1Z@H;e($jUvn`TX`+eWZ89AN#E403!qT| zx`In_K;(DwI10gXU`l-Z^!}g|<K+aih-CrCJv>Y(%$i}g6+5Xgy zpM!6;3;?qe;Kb#F4r(!{awzyb6ZXAMX<6e!2Zh%48gSGTqpq9dQGXJq4y68gJK!oY*(Q=r4V#tC?PpXivuOebXfAMo6{#4<9Fj#rni?fC^c1Q`*D(K zuhYHNRrm#u&h3M?^SLtEKnri*lM~In>kzl2Xi5^pzc?xNH^*-osV1Z6heMJ7bHA2y zqM5~ZcGmTgz6_>6k;h>38M?Yf*YzHJi%l7E;m4Ndy$wKN)36KE zPE|%cXFEN@>5bc;uP!9-Voem4ju^+6oxbjX&^n1>SRJX=-${;#Z^3K1)gey;#{rH7opq#XP5hcQ&k zaE{qk+F*XX+;!O``8h!dD*|<5=*>|kxC5KSMmP-V`NlQcKsf=WYkILiM=@`-6$mH* zC6H^(1)mlXKm{1VC>I< zE7KWc(Z<84ihK&Xq`o6+-1I5cOgvJ+NyTB$G-}OIOKOZt!^@q<{)4L1hF`0G$^o4- z#T7gD)rX!Wiv7~^Ps~L^%|+oVH9^V@+&H!O{DxgfriB+)Ctop`brvkFKi(fVTvHrY z(62jvF4ciLWPZqfuC}>Tesn~{P-i`cLRL*YxRdsfeK_(KKS@6COS#k>==rIo4uy` z1HAG;4aqwJYLI*9FJBtHe#u+=!$3^#ZmPl%X|)mKcS@89w z)}fb<-dx{bK{(SbkLxJ!cQd=;}=DiYXGW&Cdd9r+8sr?rLR zP{f)%4L^HR{8re9L^i4Y^8SYeRkOa75yo-sU32(aFK`H19q(=!>4MH4ZETrX*admS zvUCWL)w^?li0Cau@t}L&%4}}Gjx@4(gN$z=CLc>LtT{625ph?+|K_!^nIj)%_t6gY5%D3VZvHO)2EFeWYtqKZ|Bi`8-sEN z_i&#nDk7_G@B3c#Kaqq|f6{tqulUVeioIQlk0&NR|;ETRTJOwnael3mq>@ zr8s3w4&Z#oGj<(VRuu0l^=zq@LW<=DY%4s#xd5Mu(^sO4d)`f1zKM$5{TRN_?8qU4 zZyqkgUV#UiuK;5)Ta>pn%8fDLgNaSB3llY+E>6{!;-vI5r`!=+A*c@XT?Q^ZlKZ7?1&dD}s1@!333U z_;%W^Oh;PSYGLlA4{q67>MDJr5r2lgWAU`DC`~h}4&G?1o~!Xzwov(s$j7G{a6iqL zjh1attmd1u*`th8p5KoCVNNVbHe+p5|EF?UIC<-PWKc!25-NK7hq*$G<4TF$nR<1| znUP$2OA7ZOH@0AK?|)-s|D$YM+mrK?Ks2Yyrxt$*BB-VZ)&h5BRA@lf^*r}UXF6Ha z^#`tVCF$JYJ#&{X!(?N?y1v9Y*vgxmSz584s=mt{h|?e-&j!oPc)}8=c=hE@^Spx=c@4aCvgI}$WL?uoKsR}b^=e~-pa74j(LNPkxvI8 zPePe;fbceo*6OC_t&Xde?!ow8(S><|K<=KwL%5{GO^%txa4-7?qPzQ>tkIp~YRuh8 z=6?G83*O9ZZzdCDU2~5Bd#5htduQVWp5%C{-J}$Gn6^E`OPwcvoMS{2=Dc7utDiidpS#&HW%E z@qmGHHm&{YtDML;dN6cfo7_t+FLHXumAWP1o}PyL>oRvAN$te`S>jnG!t&A$bTHY41m9j8siovW9*lwleh^O8#yH?P=Q z3~%&80QWg-G`I&2JOZq1M6~2YiJoejVVVjc2$?4NBF0ZsbU9U`+#jHvC^BX+G!a3a z!`(1x6*#?QJMeOUz7`vP32!o@04_38SVo|DB=ZPkfRDCcsAlj;5B!5T12$6H4;vF+ zo>@8>4Nj0xouDmHM4(h4jo$yZC~uCcL4ZffH9>P}5W}8Q477rnNYTT97KkGm@zP2F z{ajqPS|_lV@ep0GCoNw91&|^q%8~@`lWTzdm$rPSrU|-~S7n(W`mgj3=&G@3B1IE) zvx>QQG>8N~0Qjnl+d;G(eF85pPPRSKSiTV8h&#PtPXGe=hpdJhLQBev|JA&lLMWul zNTu-)s%YvzRZv85`XR2~Bx{-$ahpr4Fex9ab+VEotXZUCK(9m-DdL9ZGg4=C0VeJ8 z3h)e^0b7pV%@+(6{m|1hhkRCnhit9{d3CN%uYZqXS*ZUxI^B~&{r!gUjjtUJ*GG2j=BNr#-6`KaM*ZO}XD(B4g4b8b5!UOZ585 zH*#ip8W1uNS1KBjNP{Wg0SvVUOu^uDwH{L4X#0-{&U4#_U0#m0iXAn8gjBFVp6Ims zQm7L)f4rVSYpKNc+dQf3r(qbK<8s6#e0|Fyw~Fn1#jW5-e+Q^w{usu|N(w-0XxI^m z%P)xY@Mg~}#STgUCa~=m>R=^>Mcd>7A1>f!-xX?IYSK92hD@bpmg0-P8!sZXt-WEn zY!LQtv8Dk^GieIrj&z`x2Tz=xKes+H)zTG<8eZwV`&8aw*MpV0Azi$8X)=p^Y3$V> z?-R?_WSPLan4{)JM+&b;KQwry(qhwYYb(D!k|gw|^KOD{#VG%lVYW`WS=R{c>Vyn1 z@%Y(xn;c718U2)uWJ;8=Tj~Aj=qn#~ujKL0u{!K+@Ww6L_%_^f&s4 zk(rOM32-P_i#QS(gIbe?aegRH0DgyRJ^?U$B%dEPq2t8x)a^fVms}UOTo=#%&T;)H zGfMS;?E$yK%`8-tZQNAA(EXPQ#}H4xq?` z;X@Us4i+l16T_B&bIJt}5s!h-OAv6P!qoYb#9Tfj`2d(rJ0}{z44O!UMM^;AK2^n# z@jhZ}AkCk1u1+Z%ga8v8(20{)z$fIvXjb5HbFe`IaipGaAfZs_5TFpJcWM7Df{XN$ zRyWg!iLioNT+4UV%_0?3JUJ)V%6EQ(5@^Av1m^I_vkOfnpI2QSI)<{pvl3v7BDdRI?{|K+PfRs}1xRtKT)aAsd1U#7Yw4WTzMs41 z?X;KgMxX<guAxc*dN zm3P@xyrUk5F*qF%F^m7_s=NBM*!KU@B7kKpAvhfnJoUvbuyJh-z=}nGYV0Ue>nz1+ z9a2_ZO_4V#d`5#G$q;cQpZZ~R_ig}!`)y%aC~4=5t;2(v)8Zqpx1CE;Y=N)lob&Tb zRqoM0c=U(~X1R@{$e+}YW8mitgvnS^bhN2DQ=O#eYBQ9=#4tLXB!veXZPrQyuQd=J z8KlW*Oty0ZYPsM$rNoKJAVZ%hzfr0n7X;S8)Jb6dfmax?DXVoFLIJjhJ{%lfayLFq z6-m^yA%cbku8rWBaYDvO^lU3`hn`WE+Pc>JvmXd>c|E+4$>sp*=EDz5T92}CXCg!oY0o=AJXi#xp_L=W5ec8m1=7x zKtcigm^NMkAQT7J86oNpym8J->nvktzh)KOE3Yu|YF2rm*!N zc5KvjY1rbogs*=i1ak7CUg3cNSDA1UV%z9bA91MlWzQ%<-@A(9=u!1Ag>?ZL19@66 zP+SUN^=_eKt0R62{CS#9~(HZ(Yy&|2hM-}%R&+K?>`w-^- zfH&?COM#PF7>b-w!P#EkO65*G{loo$p5+t!1z#GSH)NVCB#Q=XOubk){m&e3}<9w5P`V2H9DSCDcU#7pJRK!wP`*Vsw5ch*ZS^@jE6pXDt zB``d}b4oPN)m%-F`?l9v1?eRjAV?ia#w~}01svwlEdFUgsHablYJX*x9PzEZ!mznk zW#M5y%Id+o>_%JtA=_$3Sw_f%MAubfO0n~@_C;)IKIK^7&jtae!Gj$WLWPcx;2%b^ zJzqMXQ&Nsm)`p%53f2x#WN*LB*$EIJe>b2jShH5`X4!)OOU0^;iFuRBT?$9&-tSTA zw*)f{KheX6d1xAP@0Y1sK&W1%zDavHwZ*%cP*M|}j;6aeB67zMB2u&GosE=_bAq4+ z>7R;WFglVK3m+cWgc#i?Y+&>BX_#3$W1b*pkrg0AE56x^@#*CI1{z5M(c%v2FM4Ks zM;Ejr?(yaDb2FH&Ql~m98RO)r;N_po5UrdFCsf}>JfeDwra@u@vdkJ3h@&30* zMrIr6OCY##UfW7c2C0bxq2(vp)J`N;w;NrvBT)7!o+vA*$>h00N2SC5r@7N{D0C>4 z--~OcDRAjDX@-G=EFJ!YqOPQw*IrhBh~SWP$YTVaAIfo z@m!&9SS{Tw8wnK3SQ}Nh-+yD;rk6Q@K8ENtWSsiSg$O(=t3tb9j>!s!zU0z$Rrw(t zEm>nmLck0JsUbA32aXTp*scpjs>047I`~tGk^&9SV!rU>VvChscU5Z4`b7ajQHI2t ztPH=B`VBTb%1>9aI(!x5Q_PNV(swcgeNtY@u&EhUMDzrwN&^Q7Eh{1bQ3fT+nz8dw z2t~;Hp}ZAegh)VbUXp+Px!3%>3&tw%{!sgXHQd-N3E}!g_vQ)y_{BZc*T7`Q4qXOW_Fs+_gwU@8FM{{!p7TbeD;kD&1E=Q;F{PtHm z4CgfuOyESC4yC#HH`7g%Sd%rT6N55y{iy;#xGdt4Vw_6Q4AR38end8oWds6NiD@ig zmO$(NcfJErOX*bMzjEAUz`+(ls+_Cbb!`;~8Q3^W z(f`hbAP;FvW@VkTM4NB<1Orm;TdCpA%C)t5A%+!JAd2P@20P{!r;bZ3!VMmXdTy`9 z(k>B9xOTb!BOBJMl#sbUqZ8NZ|PgHr*; zgTM_zNeMnc&$Ce||HCx_C_;Mhf>cF&$#yj=l*<0WkqKK(5kQOw#zX*b6ge;;aFxpJ zpuABm^Y9YHG&VsMB~}8|91SErz*SS$i(y6m(+k;?lfnQA&KXb^agluvmau~c)2INo zOD!t_%)W_aGpZ?B0Y+bJoJ6Q*41vF=f{b2_9c`c#Y1}0qZfDgg`!>!ABa2B!1Cp5|4(VE2d~)np<}|OZgG3RwoIE_UAetg-b<)(P-{+H_P1OK?~I)C zcwO@1%(XZZt!&Z>53=w`)UuM`I4{{{v4pdpbfh?3OI_)-ED3!}sCTC*I%7JsWj*@d z6p~}6aeL@G>k+%?ax{%^HS{QU?iaaLFf={=ydM2;WC5ZKeKgq)2t{(T$_?F*Na7$D z_}-I&Pq72}q(=Tz&(EICTXsXBgn3tivYQ^EScd$ZF&`T#{yCSn)!(ry2DOz)?nCK6 z<@H9e*ZNe6j;oVJUu?T@w}fq04Il=3iVfcNDSS~pf&YbHly)c1pKAiHK4B;A?N74S zlZ-%}=n#2t+Y#8-hACh1qZLU?Vh@a>ZNFRGS=JrDg1xLb?G16GXXAW7F0Y$(n6xh< z!Yuie+rI3cyeb2=*y{&R^fKuft}Rry1W|kYKETUm&>e1X(uH)jB3W)CtF(P| zxa7Ruf8Ch|)9cXi#whei0k(w$BemEiA_s0;*Q64b}#5rm zB1<3uivkewSV1UOk`gveM)7i#vKWCaWJtXGITsuQ@FB=RXhSIHG&WMnmuiAGe4P6Z zfSNj$B@u^1fXo_#OR+D&D7zmMwH7Jn7%4)Ebj0B1>_o~~BgK9hwp{!twE}W20CBp{EKyIluRdOP2QU=IL?#9;7_-Y!2%xi1=D-sqBQBBB5nm>3{>XJ9&@ z^>$?8RfG6su9mdN;}3DgXhJ-i8f~Sp&tVh6RKlOr%R&*u5|dU<@eoqo@bdGZ@=WU! zebGl51(0t@5t(oFk$#_1u&KCaK?6q&;QBC;XbWnw(VU(pc@Ugta0+-kG6tv7Z}-sCi^x)UnK_qCVjfN=UTMxM6m433k*K+s1D}ny3%(!(dt1Ch!<1WLtc5CgW% zOTWV3&2FT)`rH#~dy|pIT2V57@#?LQi;HvaUhQ`FZS&0AdLWQr(yN(KFEraOCRwF8 z=6cR>v;S^xfllRN^?U$|!Ee+d+V%_U^E}K>ZcbHg&eY7v#p8qFe9zpeA~FD$<{5|Z z_xKtD3&$ijj|?!xhUsB&Uo`e*Kx-J~Q^`(r;(%9HzI&okH7NBTN4IK(qe6-cm@UE0 zgXAP{AZ^}2NuZMlKz?SW&?fDvmv(3w$)B19DPv}|w4a+D$Go_b2bP+wUi5ypBTw(m zUS8au+*RHJNF)xlc>GW}!$?I$xqq^aOHoLOE1}RbARW^(%JbxyhpO(Q*b##r{(HsS zqQH_nWxu_YIhsG#_k67mRhptLN40F&L94v?f&GI~+&m^@mla_<2<#;PUA0{B_5|wV z@In6#6FgC>Pg+wcAf~ApJTF#+jdCJE$sjA04*n-)_DB>zOIj@N5Kd)OM?7P?d76_B zvDWWZAJW=oHGR)NIDQGfdRd&aTe-Anru43R0m;FvbXFQ2P0Z(MGA&+Q?r${)xi>H~ zHI+m`F3jVyDOxGt=Bw%GEuq;w6_q3LYI1Mvif}_M6eCa?gFyV~!=Y z(bRve%6*ugv)u_Ya;ZrO;n+xk-726tt#JCU$Hct1T-d*xz4rDFZfHsA~~g zEj^y3!BvHY)FrkfG?uyLM1ZVLre`tZg9kz^Rbhvfy?kJ84tt77){(Ux+jt$HA~ce+ z$n6axIrL7sCDk^FTM!$srRra9<90QQqc&#-ak)lFjl5Z)WS@dW&p#Ie!SwwV|QFJ^Y1pqQPwHr4Mu)fh*Ab*YPLp%L?tSn%<;d;CtTEglRZ z-P)4wjbZkswt5R0n^KoU-J(T?8|t=SFQw8|M{;rPLp2a_?2IWZ&^1>4pTv6mDE zvkQeA!$wWvsOKeRELS(LXW3@)%(=M9K74dt5@HH|m7Bd#>)__SQCsM|(Ghkk*d?0Q zX@}m8%7BTkhySMe$_1ihuZ~e;1|ohfw{&7Cf}S*&;s5p>YK$JW(B>MHoRkKqwZDwo zE24fNxWQ$q;0rDM08)nbt*7w_TX>k)CB@v`Yg1Oj<1SfQ+#$ zos}IE+(<}SanOixsU6ouO-FRkd#xJws1y=#7TjkPa%y)u8M>=tJ^qF6-te*{_-3mN zGj~U&qCT=LRC%qojlR7=+}dA;JFc9mc}}U;Gtxwd&RF>c+K$zKKtCC_7+GnwRJRL@ z+xBOrSp8E65wdusC;XMvtlUnAfId;CxMWbhfq9sD?c8%*dt*S(H#*AU<<*plkufFF z0PY1X9YVlF>T)6Yd}azt(LQidJ+F*lFO1AAb%vA##l|tQ+Y>$`%q;2k8*{ep$uSi* z=l|aAWySU0(pM!vS%6z6DcR$MG(@tn*61``b@i@@jtK3TD|?m9353*{{sr>#;oGuL zUK-O9diGU`v6|ZmxUR}0BSR>Qs1ez1oo9>wB6?#f^4IrGkyqq6P#pQlJX-JH3ooSo z!ITO!^De24u?l0UXX89J<>cooQkCECa6xPtSKYO^Mj4rSM?sXdqDh3bqHz^=oQeo4 zWgUcq-7Srqt5pVBhMDfJMtHRw2(3$0Hjzh;8eHbblcimAOoNvVRD)}JYP4wNoz?UY z$0HSwM)C6G;4M=MgLA_StA@6nzxWeF!#qin*wfkt=AK(}>Gs(PbpOtN!(IKjD{3C?3x7<)G|h zpW6Du+iSTx_PU~0i72=}>KIwk6u3qdX+k0NNy(is_~)G<0%?V z7=?Qrz82&7#|M4+uq$r^B$Eei+GRsxvSk+Adzr=i81qB3NW^Ls2Wvz6*@xlSltM@6 z3r|bvGfhU5+2Pn9sA{fud^#Q4&L)4W%v2M7lx6m&cp>#LIbbpd)CZ=gG2Wb0#z~+= z8+?wBDEg{Sz(g<37~vEIP=+MRODYs9z)nQoB_+V|r2I@1>E%_82G;(x#-j|z=2vYDzOrXgOuL64L^(qVJ+ePfn*u7lPKYe;4p5+*o+uC zF||WN<2!YZiJou-c%@;gjA9eTwyFvwH>?OiTqsEb0rJZUz&NrN zV~xHIkBJ6$X4Nnlc;`fgO%*pD7Pa2Q7#vUJU2G1Z2Y4AjR}8zECZNt1DDI1f@}0V7 zFl0<5L!xqqNp6@a7R(KY|KM!Ck|Qq@nfe0u6R9K53apZo)+4T&T_Y? zz|Iv@G#wGz$%&_?CSwpfe*YyFn?;AVZ!>*zuccLx-0BQ@`H1*j%qQTM^aZAs_pVKU zD!H4u9=oE@ghHx&V!pytANwGD*EXDQf)ekNx=aeqn5Q)Ad2+uZpS_)4+$o#X!^ekA zh^0O#l9<(*oKY06g}T5M+p=%k%0UsPo{s*)yXE?ds(5d z{D+`ku|bsT+kY?gvjVFIc^L{`qEyDBKbe+KE0t1$H**UBnE>XJX51WjiQ-%U$8xlk zqbecrbq>N4MQzB-cKA_-cY@>UgyoTkLI68_mGXj$ytECUJ3`hhw)W zmyH#L_1uqlzcVISX<2SH?nOQ%Be9jwr_-ox4J4hft=SYCy%RP6^9}lO=|M_lKb`wx z0TOcU2+t8n_VeExIU84g|Hk0X_m?AHTFhVRQv+I(pF4a>vG5J(ZMin-NqyMZV%$Px z+)LrD)W77>G>IBI`Dzzzd3*P}%x1v7vaFrGM~Ry`+ege_Yf-D@CQuXODJe61Yi~C< zKKy|?Qa9Lj8fqxLbVo{~1R=Z|&D=d9 zRKkYTB-bOlpbc4wC;+5+BA0fUts*E7+IP93C>ZQ=X6y^o>Kpnj#mv$CfZCg?eCI@c z%ef(qNmDB)U*Lg!8u~}4XL9Y+>8^6%W+DZ!hk1;ZgzH{eJe%Hc^MG7|qs7EUc+2s!{4^0m5 z)zud|=dX{+T=1LIbjO_vHzc883;>+BdC@oqOzi*hr~|k$fCt>}34ezuBt#w)7g6_uqLhCLy|vi2K)OY~d|#RBY$I3BDFR~Z&m{<` z*acbp)2RgpPp)}uefq3R#Y4tpgBKyk5$&%AqK>Oz1wt*b>#OU7H8NbW38i>E)Lf$k z1iAyM>iT%828F120vXS~c^xDZH{?G(c0UFt{m|j%uSIr`5xc(Ncl;8ImDUeD0c8F` zpv_CCGAA(Jbc1F(>PqL<>0{|6{*tz<7s>93NL`Gt@5N68`{Jrl1RDL5P%2aOC4EQs z;Ka&tfS3L%bD~TT^QynaI~R;sQum*ofa>(AAw~;fIGd!7Seyt!>(IlfaI?eb70S7@ za-Oy#VOgAU4JvXX@{^A5lScdFbk=d%QwRn5u`d^FZ3~ZZ#3MjL{LZt7u5EoB)?2&G zOewh*{M7(QsjgjewGK4C68S1UkyTRzlG+9Syi1vC>fN+JZfsyQrSPSnam(+7q;|Nz zJ6ZJY{_~IEHV|`sOvUId_fw)XaT$U(4c8Z4NQ;%O{XX%U#~0i@<6ilCh6ZFUSUJ!= zLT}+S^uYn&0m3s)!OE^^^;~WLOQXp_wf~k+5p>_j7Q)_g!EQWxxAAh!>)mF*fWZb1 zHKWBF1{iC*k6A$rZ zt~<`HmH??$# z%!@orTT|AsFH0U8^v>pu)FXjdDmu-PKK@y3G>o4cIvm;cJfl%U%Ql7eI4nNpk&1I2 zv8A(Z%3R`(l4wNyGnikd$Dy^+_mYa}z&Fq$nvMvT52uR;ODmJV?R)yibNObiYIMov zC0%~LxYQdcZikRh;q*tcc-EGEG{j0(ysiw``o=W8F{l?)OR|{FoAei_Y0W^7J#~AR zL5>odCP6vpHfkrEK@;*LbUGB}_v*^!qJ)A|faQncVkj!{)s@agY5ov{&UqqmRk9*| zAiWhaQbxe5ouu@r3yS~k_u?<*Lqkd^4E87G|FLw{VO2F>m+mg<3tW&!noEP^1?iL$ z1f;t=r6eyQA|>4-Tw1!jMY>Z$y74=_zwZwo9%g1wojIJD*=wzx0rm3-0u3YrN;hx` zYdc!-PKb@PAocSfmkEUDQXAY~@qxq?+LbEgjSw?hB9yrQD#Zu#;6#3%8xqzneUC*C z6A%RXuP2lu2^l=#=L$QNtucRqC8h&yRFc|l+WGMX$T8~&R^h7|JnYY%rnSoyh7{I& z#W!i(|ME`ApxJy*Wr|6F&(x1~$8NERKab7kwfqI=ZEm&QZ~1W$Y0{BqWIJ5OU;?{z zH8OBG^(LulN5WQM;iax2n~l+Wv^kyATpsO`OPE%-|6El&`tTb;$1$^t5`PWKL7c)9 zI$W`@O(wQ$s-)jHXz4>EUJA$YnoW@%BJ7?m`*uejk~HYEhObW-7z`?3Nv`@kjQIFWx-wAS=tz;mVHv+c*29Uht)Od|)ok z_8OTJd{}$RL{5dnY$06$agO@4Tj9aq%=S7f37ku-^@uLSCe;}kPO;;NJRv#ufTETR z`awVP`45}LyIa9CmZheaEJ6;mxM;chSyKs0X#9B`XKMQjg(wazpR{2epI&1CF4`Op zS$$xS@JZ`V)1A)h85yWSb`1C!q9ibZcv|5&>ws$n_y7$$?%AlO9RpuQ!0Dh>*k)~W zF6Ex85iZuAoLh_#@9vl@9EYwe+)pMD4=oZ{StQ781_<+I)2Z|dNz?D1&HR{ zsZSM*B?*9YV9K??nik_8P6#Jl(}?g6zlxjO69>*h?DB&X-C14g zW z_*&)zig~V7tzrlDOcvE)t$2ZpiX^7<%p7`}2!P5}P$o{Oe{ky$-cfgJTlF58?(x=L zW&aAUKA3}V;)>>S4{7mlsu(CJypXy=d*(Wy~6audRTpHnQmpYbGN3 zN8I>xE#yGIT=cJ|y?(#LpxRz@!F8i_fa7z93JIN0<5wmiv?7p0g}ZEft@CRZSRJ%+ zrd8f_^jUrB*E-oL+B-9x7FC$6$9jg(0=Ne%Zt_?@%p*iXTl(kw=zRbT3u2I{~baflL-JKAw+hnljy=fXbMRoPcd_ke8y*$=|#Mj(7{FBJ#RziUh|%N7E? zpa0~$bvP5Mev7{zDXd|J&3%_4}a#Jb;z|b_NkP{pmR>(Iwpi zK5_2)A>jnka~XvwO=aL<#sH3RYG9NOKL?OP2EYdakSt0P{%TJWM+;#gBjN#&EL5&? z`D{?DSDt%tA37^}yrob8yb;IZWeo@s5yD!xAyPtcFJwIhM@GlCo-3rjDoqc+K=y0dOP2AP6WNWh()`2ucF^*rogh~O%4<%Bzc83*= zw=?`IJBo}N1t-FCo>hvPng663@DkD^Hi1`)Wk*v?S5ZZ{-G#9BF*weklN%(#Q)*lR zB-p9H<{eRIXD74}oA8~-2a)d-V4R4i-&;!zZC2jbO}16XYql=D&Dr>&_awT3-Gbwo z1^*F{j>dPoqkU}q{?AACzecWDc@z7-_ax-Dvvj;?-sb*&(#=A4>D40oa7L|!Wya1nN4yh(0$Q{tZeYNu)a z`#tvvvy3L3rAUTwUk!~3_;*5%tAsl*MzNJb9umAf8DS|w&YMCGxP64TQYf-xq~3G) zq6D4{g1mr0B(JpyC5e1#A|)#*1@%N2BWzl<u2bvt5bsSLvFYBS z54)|PQZ~m=zvr-Fm+rfK?-%ZVB=VKm)n!9h_m)0pxmW^T3_>j#wc7ScI`(U7glQ5e znJX_*qCp{8HTyJa6U|gcm&fEDGVtolDdEZ}a&->v!(6zg<`y!wDadE1=sK20VbroF zhB1sC*g7z`f~@@~M52XO8jkQYXx8p!c0#Ppz&`=MvjxSDgf&@I(f#wbYbf+MGY^m^7Z!v{YyuUySY94bZ-M}j0x!j6SkN*Uqo|S0fv;YG3JdfYbrSeA! zp(j)u5;iZTe8aI=YG5XS9icZ+2^K43frxaDwg1$cFrayKd`1||(oy(vA1}2Pe` z_S$+#V&XRM1o5tWC!}8(tSS9$mP~hwK|Lf5<;5p}SL_cYXL_W5=zVgCNHHZ^fM5W~S$MQab$O({eIYe56g6c+^>Z@C2A`&ALnEYy`&s zTR@zy^Fil)Pv1updR;O#t<*WW*+arVtfryCGg1p>K7yo7l8dH(%Z9PagUHeQVUe7r zIhvYcQL=hk;y=cz#sc(^0&OjCKc*Ih{Lt9R&t?dCy>s(X@U~2`bf*(_5}x%lQ3pYd zFTocDamJqzv27PqM)h{^Zx<43BM`&v_Hh6?tXc{k#E2e>z*xv@RegA-5UQGqyM**g zuVWjz%hb3ao)|pL6-kE)nrxM zLK1yYj7_nx^XUTT4~_ne_-m1{b0s3A08fGLt+1B-Omv|RH>t#>O*~YN<`4FVo8!)( zS}q09;kQ<(%tS)j-%_xCtyc8qcn;w0#XH*lyvB5q7Aq#~xW3P*`1V8bDvUAi({P5p zjv%ip`A+hdbEdhig}15smRAUc0h_1*wW(d~&hva)Y!-ta>ipdG?H|sYHVl!?d~bejA)t4f zEZCS?)?|orLuC2F!Um1mxyPttz@=mE*|KZQa%C@Zusk<=a4NO-p;y!j?fwPinRth0 z*VOdv+?+E#Ykf=i)sY(|K6PQV%CcATN10vIccY5@6~bUA7j>)NrY{fO)FFAn)Z_-M z-g+Ohy8l>sL(G^?)6QdaxIfjs+PI+^dYR)&8$!NKtoD=LoWyYn?(ZpUdO(xVN~(EB zuB`kHSy}Ckg@1~SdHst&@i|S&N3zN^=h)KA5H~yH9YTT?RjU!Y@wRFG&c}>R_DFfv ziS_fHbbBud^x~_r*b#abUrnH#O)Q7R*vdxovKZu;X3O&}Mc)!9gKnoNjS62piff;= zF5$H1;GrP1Q(S|(_m`@SYnvixDrm(~R%$_-odryIgZALHTbWQA_03@jZ6x~Mmos_m z+@p4%2|?F=QC#(hC34E`{#lXN!e9cHO`+0!U+1hpgifj zmk&vI)WV(7?M!C*;shLn((m@11Ebp+QK^L=rUR9_0XM(8)Wv^h&wcxPri<7HARVMp zG^B!>od!X<-FIXUDM!qClu8)?D+yQ9nPt7jL7Nc$YxdspoXIZ$YRU2vO8qVJ;J#h6olt~A zv%9!kw);vQPGV*5e&uP0P?e{<0XJot$l=Tv@GaKIEa%#)Yqh=?Rv?)o+9Y)&VYpNA zit6yE&!1^oe}3)8&YH1rVOt-H+M~~DIj(wPy3rwc%qmg)^HcnA!<42B6rfCs^2<=k z=!=Dxtqt4d7Y2-#(bMRdV81P<8Ik!NinjC@Jb=l`pZpF~E*``Uu#xx>L;+V|F(Y4%(Y_QDg zWFzT>(Q%F@vdNf60;5}6cx*V#Q`FUF*za}yD5(R-os`=;t|vY(nf8I>*nqJioKk+! zHr})0CG*KKJJL;kC&g;XJH3lBwWjIu`K?QD&$D;SdQ>5pqbXll!OmAMJH&W}Ni9Ea zX;FPCJ){s)L)N~$lm{6InfEa9s{3xP8MQ zTREMw1|hIjMg#xk-z-tWa+4MKD6;q{`>14C^m*u^96rn#EClNO>(lNqUT!Zn(>!ez z_{#BBo;XsQHKi~=b;DM@pz8E7cO&l~DCwJkcLY;F6dR`TB`i*6*5-`(p{I!(TjN(J z#b%tB8GG@sr8ie7#0~&;BI_#Af{+>xlH;QMAK7B-Ij6uCB3uJcd_6Ddpb&H?=Q2Vf zq%y5BFZs&*-5;Idr%8p|fG*&U3-~8!UBxZUghR3{K$CxV|8a1o?A_JStKnlZAllBW zdpQ?ZnvE!c%w6-fC?B3Z0ntym8>KqS0AZyCmVXF$VO^2`5NPPGW%UR; zWl%+v>i|8MyO6H-%`IRwxuVkR?@q^`$3@>vRqVRIBncSgY~}!C{)aGwNAeGW_T(yN z{pR#P(*whTy2Sn{*)RE6(-@3Oi{(XA1&GDY(EiV0h;IQMD=5M1-qQI{ zw3i&^Qp^3w3m+Ks_nYe=IG|;IfR1HOn|*RMw1E-tOR_9#{PXI$*{r8Y86ako80(WT zps7yU{Z}LSYq4EJ`~*br!4DV@tI84bM_^zk&z3ClN{(4o8{YpAKJoxbL5t7x{O)uY z=#ewV0A{miO9CLc19U_3C`Izs4Nyb-g4czS0BsHdP-$b}yT1)LyFu`iy;v8Z*jkX2 zm*k}A9T!b;2Z}enkSY*$4o*vKZgk5fSAooUt@`_w@YVM|ty3=|sCt*7G$KcSLK?+Y z>$cpcXm-RWfAnk`F1(k44AB%(Jy1Ui*y;+yCs#YNPAJv}30L`A0y~ynoY=E3A`dpv~*;@o}%fb=xvS+dnpZ?DZ zD&_?=CG*$-Oh7dNXUbtbHu-%N`P}ROYKZ&k)z$w0`PfRFOu&n<6!Z#MSO4Y4y6j^8 zuO}~Xr?&_?Pl5KAC7-iL+mBlA=z*DR-%$2majPajtQzxTzr`3g^8 z+w<1t3i3Kt&+8hG>P?6?@yOj%bPgZfWT);Ftet-Wd@V;gpGF7-Jy=Q{UuLbj6P>E! z6~#g?jcH@oh}KLCQ83T);N;3ivFE6b&-@DtZRhUUZ>FJ~KXH}0 zI=&A7hcHTYAOG|zbnRQ%R^gzi@!shPW>7|T(6`k|{NXX8e+U9r$G}T~9ZEn~?6Pno z`T5y;{6FjSXX`KjZyov_$W=&ys68x50=aUxeqBL36_5Qva-G#M!z~FYuc+|}&|v#$ zK9DGDp3tc9up$Y})eGcKx#z{x9%%A6n#y?dITxMQ`Sb11pvIHEMm#JrU~W&k6Clb! zd7x&g(_Kufr|X1kn#UK2UF5)XS#9UK`TtB=?-QVf=Q`v%5*YPafkl6R>(%vcmoRX% zJ&(TMpHbL^q5lvV5Nc=yapPh9Bk&;K?VLUVxuDOPOixZPo=1x$jZv+AENc94x{I;< zbQ1}*Mey)*9Rvh=@>rd8M!<)Od54_oHKwZg&AnaFFl= z>&-CS>twi!Nb+$5wCr^OeP~$+H*7}zS3nHJ$gR~RN^{{djv16G2TWpApfND$9?9oH z)8fOppL=(G>jLzh`+vQoW_ohNf9@S-*j6k)sc8J1??dCd+GOsM;}x@XstTBb?C|sS~(5; z)1LQg*zv_b1i|N#yue1dWkS$u%c|Og%zXr2Vb4$3#*ZX+{WibYSUzzcrZt-icDvU~ zoHZ8V!!&I4+OjG?uEGNuzt8-BX^CQW(6RhvX>m}LU1l!uW=e&1XE#f>fS-Un1yQTH zjwdmAaxE@@5P~R=I5ajBCjpgiaaBer%Mjjzj~`OTPN?!YX)k3}PoA~5q0ZKC2Em3S;VNHw zcJmxY5rqWIsS@!5hS(3760>l?ql5W2ucqT|Ick0Y!753nKyQ0RX~`{}3#VncOnwQ7&NvLij$@$8VpQQbgcV_~pb>H|2X_2Fl|vk<>cP z4=njqZ`Fh#DatD4W0<)Q<~{rqP0jgcCyZyvPdb$Q;)#ntRNlJy1r%eXt3~pCWTpHr z9VYozlK-=M_hjTwr7|JeEz4B-1l>LQ&Zkp9J!#E90TMxif!2$TVW6*)0(Q&!gomN!rvZ|CPZ#p5YB{5i2a_Xmg{XQPfK9+& zTwBhC()4Re|3qz|HPAAH;B)f9Ohcd|Jh$w>KlWOYaxUKa!hszlr?lK|iNEa?x=|L*F7ph!R3ItUn4Z8>_S;O~@)-ses zyfY}t@Wp*ANf#0(gD!DMzsDUAa{i|N598EcVg0O{E@QyzIE*0&~MQ zD+{a(mgtAyHy_CguZBJdpN|q7Oh?*?RQn66)lVZjT@6Ja7|$6Ffaua&C#U*9WTi~h zh(mm)R=ioXUMd!TC}X!c+f6PIp?Gjj#TB33oeL143OX~NSwCg#i)uDdG~q^0=XPuM zYC&+^2!zN^6)D?mBC+je7Y{)NZ&94pHDt0kzQm5tJCc5}H&3*IPHnbj`VCP^d9Y(% zAqR>U_)^OhJz1*pv>ZKzLha<|LHydtY6PDATLsOTR@P6KVr$e0YMq33KM|NveF~i- z2bpG0;f~y5Gn)kc6;PF?IO+)X*RX5`g;H300I}*3YD{m(=7*#cs54~DS%;Kpa%j2l zX{)yn2!*)ufe6iDO-)^AI)kriDAL_7fHWS`&4e}1GCD4U7oZ^A?rV8LjF`T-f?7^O zGClMwz1wvS6KhO@wA4S84Z7(gPRm_1?-q#;Uu%F~GjMt$3ioF$F4@n{Byg1HCFLKH zS%6#%D(XA_Doe-BpFSzYa8u$P-!gE~pamS*r|lzfnTP$DPSPldQ5>%%{(2(y+ht2o zD$9@xa1;$^Y!CU&IShVBN`MR_(OL8iWy>}A>L7@RxE!s)fKlId8 zr-$33Le32qM47;f4gz#Z-D%`T+ocW84hJQMCV$ zsU+wiw@dSH<<*annP2l1#7*LxGCD{2Ry7!5f6C6De#i?xZd;F0Z2CDrb@~=p#{JBg zOMjx;VyW|&>lL&TJ?+iVlV7zNiPQ%j*oICmcI(EVV^TSc#66w%|72|wwCB}lw3Bh2 z6X-Cf(bA?4uqv^``7ub~J;3QKe2ufM)MiwvCjlzA7a;uX<%1j8f1-Kxo-^r(oK5sG zSF8;(-L{g4Hl1%B!qc=)KouIa`!o?xK{q8*hOUDM#~Z@+4qAwAP5Qci+bZ|c`NJ#E z*66>OBuhB_YlVATDWQL2cY^sU^j@E$loHY6yY>DI zQ+4ae9C5}~4&g@?Fu6h*&(2g8>$<=Dk2+X#19NOHN+Z4rgZN@mChN+2mPoVdS}C}V zdJ}yFwDr|3&!zjf^r@S3A%?qL3wqMMVNvEF`b&YMKVcui%~paIewnbhED(GYZ-Q`z zWO;GVrnbLd3HnY(lb>LPkfyd73%>@HN9@O3^mYsD?)*eoX9-!x{UBy}g%%I?X7XX` z_0s_;M+#1(z~;a@iV!a|HDv7}DP3_1@}LA$_L1O~ptzs((hyDS?;*|L6?nsUdO_%C zl7Qwc_A6<=A(?IP06NTLC0>hsi|9Yr69F+YFlZ@3bo7#BL4Ax}FmH_+P1I&8ecXW`7x)t(kxI~uR6Md8~ zP%5or!mBba(fwoQJRPW71Bn0lUj^^$BOheQ-qzH0&x+o7D9t7?QYNT;1F9Z}femGz z=s)p%AAR8+rnaB*Zi76sXfi13{qPsDDP!gg+_&FzfjB6jL17hfaVTe~?tp?Ahj~tN zY^5Z+++(&6w-+7T*S8UbBTRb`DptT-BiJy=Us4z;mP7_ZDk8tua3tlwAgL-TO%aIO z^GRMK5TLi$Xogag_r=0=dH*5w@aifaP+5yFwi+870|!IgoonNu$_pj#oetw#s)unM z2kW|ww;!3n6ylwu!v)U6^t6G2%8|**COqdx>1NA2Hxc$%5z|oelIU51^Gs>gF~{qw z>kcEqle{`@8!wllq8$TIJ~s=x8$3Jq`Nwz>ik)Whh_a%reEh&2XjU!4Ukc%V*wK=Z zEgb6LqdUb+klnIT)!I@S?aC>0{77m^=WB1t_~Ys%T5@!e_NIb&=RT41 zO}`d}=jRtZ-RaJqnpBallALp>OKm2Rwy4Kn>I5^=tDzi>escmaYI|R<8p9H6Pc%;> zK$Z?%ks)TTJn&BDf=gwj+9yY&Yt-T!kLej}gThv24%$PqW7uE-@`#QIUpzFL6(BRD zlh&?FU>0}ze&lD}%HzDEby-jO6%*n1K>^m}Zi1`Q3|`A1eN##?wHb-`QEYI*hoE3k zbn8RMzNW>%9BWW4JMu3GPG|(Ny--)G7eJu7C>o{2Rhs%7ZtC-Llxiw><;pbg#k&aj z&qT_}>WpeF%B1Q?hT%YS*41S4Ctn43Gy4D2eHOhRJjLg0V=hK{~0A%;Kll(sf9j_8KcDn!)etr#2t%f%){Ajl~ zu$*W9XfUlZCjlk!pa@O#tP^LrQ^Peev3Ursz0&2##MC=f(DZ~BV9Yodm=E?*sz{Vx z%3TNypOkcoLOEV`n&wZ}@DrQSmt5pa3t}swmQ1LGuj4S~u~ug?(7(cnl8W`tqhrdW zN8(OECCRv;K_H%Hji_J4Guh)=Lqh+nu4K)KvnlQ}$gYI;mw;0h5zv=O6ev_^)RZxV za&pRe$g#UOvZkNOD!{-gdK#mSjml0rOp}iNpc9ITLJideo1WO;cxDbaVUCVxs!7)64j$x*1SZ^|-JK$|%wt{RNG{cH<&JrN zHO}i>4I`Xx23IsH&I1x``UU&423g31NMb4cb{J;D&k8e&)+_%#^-r|WTKP|2a)n|F zForqPeZWI%{VS893mVfPd_s@ht>p1-A;8{h7)A+JA_pEm4t>72%=20r6a;F()=EE< z!6<@%D-;2o&Jv0MpNcr#hdzN3cO8crSW-ERxBxNC9fdojdWk+b6Ff-nxRfYJnV@8X zv5HyDTb)_2kL+t7GRlHL*vB{>jcw#_0IbfJQ5_1wX-HxWFeGjyUCg~$+(bqu1SM{w zV5vl+V2p16sc!AxiijKtVTKziXa@)g3c;zdU}Osv$`k-6fhpx!@-jG$GOiNSOezyl zMzY8yTHr2Gv^AtmjaOolZi;-L4ROC|+}7YzQnzL;d!>9lUgYK5Y^^Degi;s6E~PXx zV>wDgz*MmeK_hnj5@(wr;70C6$GWAW(k34=s?*kb=KJGE|9GCoo$`3a5x9dfRj@Xr zG?uE-OxB~fAaSJe_N93NmqhieaXnV&#pP-o1WxM(!tA2g@-pbzdms5>Id8|j67}F$ zKV#F_DgE@Ryvz~#W0wRsTq4$3T?~V?AJQ_|osoZI$+{y1sD}yBv?I^oCy##Xmx6~z z=el-D%G%;we+w)#wBeaFx#RvT&)R{(?V*mQ-I|}Dh74cGDJaM_Z>9j9ZZ)TFWMZdO zl2Zs9$$Tmomy9agL7%m|ulEaG?b1PAaom`?Mvhc+$M@1n(fg^##SBQ@TsRCze$iIu zNrDqO3s>ECUj1-bFJpTz9(ZDu13wx+HF68(EIR7B(DwD<9-#e)Fi(}bCnS&1EsyY< z$(Ek{niWlaW@?-`Bu|p@8|?u=49*>Al40O4s=i@f7-y`neI|zX%YIf!~R~UF#D_74delI%20C&%$MS~tcAdt z?dzE|<_L2hC)S4q#?Wyll-J8s%GzJ}g+wyZd!@>o)TiLhHot$x)$AfR`{39pkO^lf*46fmEm}ti`noEKjS(I_eUG&gNywHe{q>t7> z^v{UE3;jTGUHe>%$d+{Eah5C!uveyka*@!km#L3EY=_%TSmFZbJ)QIN<6a z=53Q=pC{I2AN{n%dI~O}vzlA5zMG3VU7nBLQ?Tbk)fiNWEcexSjL|0xCK2U|%RDHy zBhZo<$V%ZfV*>~U95CbB+~je1>G_ActoNMR1#HtblrnFq3tsi-=%4Hk(xt^^U|uuR zQZA2QLTA)!u+SJ&gl56$-&k7paF+t10&zZm*ineY=IGghEME!Zik-FnVAM+lqE zU$Y+BNo`4B6vaYS_{|Cq%$~5-Wee;RB3P-46Z>=zdeMChX^TbY$IzLge|#_?M{p(= z@$spOHZ8U;yR*u~c&e=?)2tM2H2b~x!DrY1VODN9l4Fj@A}#nj^6SQM(E3}wWsVty zX5aLB$fasciJkY^IL<2=+Mp=KNGJj=wsm2nrx$E@Vy3m#kh)m-rc9)YtW_hDq9lh= z=4<|C=$wq_MtQAl+OVD*a)itdl4P1jHuXlv+)mp)HHT<3WWgw4ek*sEZ^5RB@{K&- z#%QXsLmD~t(i~45vkSApZ}8eZ2`5^OZ_H=B_t0tyY3Yi!KiZK`LotegEM)qW;W>-& zyWGV*Gymk75p-R-kaLYuOGc}Eab#X$5rr2Z*j@u3*-tG#^^dSVk?Mo{G|#+Q#6UJy znRjaj24Pbdzdji`pe<9YBMx-Q4=LM}Z7{~Ef2bK&t-~SS%lTa$cFUaRYKFr)*gW-a|elh1so*V^Z*}ZSZ6msC7OsG+&xz0~> zK;JKAw*GSv8HCCbJOs(~kb_9PMaTLazpUY~j-91f&{&spjDT*6b2R(z=loV6PEy_O zHxZgMcH%{+D^+xV?qTp(Jq>E1 zJy}b(GrlA*7Mf#c(0j^DiS_+4G%IeL{9RI-F_1a1W{NQubhCS9V|h<5d$Zb_Ji@2FPR0+)V_ zocS*Iz^bR~CMN=J#V(We_Lw7xWQnVea|(fQ?XK~*(&&W<#R9e^!c+Z0PHp3=`f!$6 zbPZ;iqR3*RPkBrmsCMM(Oo>^BerIpR@R}`>zmq*8l4mhq&Zy=^AA&sqnnma6`Gp*^ zr$<5V>vc468#i4&eX2t{ddy~yd|!NiBdpJ_=3Xhv^ZhY7VI!(FYyl;3LK~FuXFuxV zkNqUv%_aKxZCpyT31gk=5ofZ@PhPx0ENo8hpNGGQH}J4zzPTBf*0d~nVKS6O9@(@^{sC2tbivAr=bBJc$Kf7`m z6n#z=%KbMzs6|Iv#I^wtaNoE9qh;L8e#n@RmM#6ARU1}xEWLmVtG!u!+>@aAa@6-%%%vtcT|f+9pn09# zeW4Ci@lYX0HxjZeZ7dXS4HC$BRTzqfcHo{o^!qok*VdwNV5~nNGYpv|%On*)|J#f0 zHyrnv%n#wOkKlW`TU22)`1Xl?la``*PQ|u%RwQ5fD|M>ocIWT;&I-tFRo%g9gZ{A5 zbj-tzO1S{x^qV^+C_7Vh^Dsq7$OiN`K5vu90(RY5l}PzVO93X1CeAx&WlB7ZDPG<9 z)vnYRqEyEYjlYa+nFh-W?W)bKX(z3BxHWIQA|Bbq(bA_32CjIbtt5MdPV{K=T$jyo z8Xbv!KBo059w@R_z7TO`nm^m)^g%v6_?=pr+n}}h?ZWy!&*s9wWv0Jcw3Ww!O8OWRqaSMqTvdi`L_UZ;0RqdVOx^*JqYNUVbpo zKcLPEhj6)qc$>8)V6WMmJdPc2=@iw)<9Kt5jqYFJ=Xj|lsMq(ag~BfCw?pcL?I+Wa zvrGguFC(u{?5N`PUY|Fh+c=IvZ-&W>rWnJP0G}YJ@b9 zM|tg7UV6TJec*o?a_=8gyJ<~D>=26wL{i~**qW$GO+`v0gN}{^vD&S^xCI+m`J{=> zeR_H_&SvM`{Bmdgg+@3w!!MauY97{ra%=g)X@bxtv8fVe?a-C1<U>44K>O<{ zDRS&rx+`Z^qI-^Ow}|J^9vG zce)E{6-HFv8~hMGwa7u!)d+}`)p-px$co|Np9M&M ziD@dHtES~mh13W_r*|^D$Z1hYCT(PBZMq0;q~>iImXp(^Lc(!vSm4DNEY%^cOt`Gp z_95i53K?#QWJn?85wHXp{3QGA&ACB3nP&~?to5+m8L>Izh#)|S= zk?KsTABLPEfY9HB!-PYZLV)o%6j6#hLM$maI8?%VG9(-|+?!nN34`sJu35qE8!yq^ z8$cX2pqPL!-9>1uF zZJF%Vl20*H^>#m|cI}gXlhLDu?hY(zrD#!^`2zc!P!4x`{iN1}hK$^sWggij{vaa% zp%jFf#y{#jI}w22<?cZq=!QTUNz{O07+xR@x` z^xsVLI!~2kC}d>UZt^=%6@m$5UUy+k$h?-O#IOz}m`aJp#Q;J+_R9iA%?F| z7zQ|*dw~Y9F`Tbs&-L*H>H@hMpZGq8Ys`ug1q(($Ns>+^FwwT9VI4qi*xpvL8W^lU z%w`j-8lj<|quSl@9xyuPzBz>H^yJPRgs z79gbhq}c{kqVMbXnO-7@ERt_JP%8gQPQK;PVL~YoRG&m=F{9WCr6ePe|Ab*J3nG#A zE@WUKdd=YjJT!F#y=hfW{DA-;ms^ZJ&pXyYVn^HA0QZ_)| z^OW4u)$#Frs5F=6yKWocDvx=Zo+~?WRW9uc8#BnJ9--q316O1^+L9mmkb8>3A^D~g z#D^h0_A^JaeAG4fu7M}oTiR;@#tNfvWuQr)5q`xxx^!SQZ+b_>YoIdm*)p1E((a0R zS#_(D6$qx1vM-^fCH>VZJsRVpExqNatW8)CZE5;B8D);HoI@^AdP-OI!=W%H55ty@ zL9xITkP?E>mT4#}+i$1kz-o)?*Zd@1TZ_MKH@rg?q2$1~REdGG@YE-tX}$l}u>^uO z$E4twRz}8cv83@Ts33pH zHd~cfe|WVfKZ#=zU+i|xRd!WoGIE4h=WL9l_TkG1n=iMgUJsne=>@L-3}l%FdrW}F zrQA1d1%P*vjiZDkHT||m|GiIzykST<;V_}i89=`QNJZ}S-z2{QatYXk)~ZT->lCR+ zrl&4lRh8}Y`;x-Cb0D;EVmM=3)@fWS$xmeg!yh(vGc#7nu~wi-!JEdmpRS?te3N6V zTLdPGd-)VF`Gyr856hj!H&e!16cQ8~^IfM+8C6~s+dfwr{b&t$rEoGsal}zhDx>qu z0WL{jtWu|wl=q>l;?k0qksvWM2FOhe{E?AIo%`e>ekM@Gz^mTqy&#%-&*m{k7BPGW zib3PqA)Ws4l`F{Soo88QU@Spjkd9j4)d8gvn=q@*?~cJvMx5OhrKNfZrE0`y$R&mA z_*Eq4=ZNhf&-LX-$V=9MiuDu5k5zWa22$Xlt>_7&03Rd1HEx0$yOP?ep?AT1837tw z+3wEB$t}&X+lnP06XA!v{tZy^U3x;~k#$v2Iq`TmBLCjDTf@}kmXP*q|5soyw5Kj7 zS(P1`gp4m7JiGg#@)hT#h`98AzYcX-Plf8PKyr-C{-hE7je=Gf$THU_IYl*U^Ow&J z4hq4D6i!pmEpd2qt{MH28LvRRQowX&RNA%vufaS9*rH3)sNDKFKFMy!piA-#EeB1Dwsae;gzj#ys@8H zT~I#Ej!pX+AII8uXvO6tab5#^??PKUekV_?8ZStl9cWGA66^3o0+E?Z>`Q{LciN-6 zS|`(J;00D*h zIX`-VMpPgmyY#!NdnEUiwxS487y3$w2d)c(Z#xP&F4wo`LEGlzf{}1V@yGTm55lu~ zkws_kr&(kA@mU*Or*5hOneW*dgrA4{?uTS`{zCc8qa;)mXv;A{V$7wS^#@)D5wT4x zT&SzPs*_yeiVK~)fJY-=^n;~Kwt<2EV#H_|IO{O;#?8>r34KOyVBM`vnaY4y7Fn&E zi(Rz48ggo>L+6=z9eLO4WvtmNBS)Pqz9zm}ahL=9Jg>u_$=*$3j+UaUDBPxQqIRb9 z;3Y$n$)txX0GFns1eI_LeQxn+gYmNcO)X00BR`)AIYUGT)=~Fo`Y*)$npiAo1rQ?=)R4Q;%n(rUN6s5$)@JIVd91E9* z%LYl;Zw>u?t(t$A0uuAJCp8A}3+sj3AB;#jFHSgnF1cuBa52M3aTIiU{h>_g(2HRe zLzz>Q`6ZA}|FO&@tI4wVIE?tNCaVqIvdiGQ|EH51{wG!ZrQ*fL&{L`Pr8%Ybb>=Dw zIjg!#QI^7{KG%hvu~kg-j@xka>5Ra%=!Aa#f~S@$JLm%Hf{<-?21V2&CsU4VdxK}H zGB{(Xnc_W-5x3N>V}^W$?^nO=#^?LGdA_xLd_1U)f*+#s7Wxp%0Ul&9K*&lD*%bA~XFuvTv4$E@&%w&8MG z8h$XfMy%FJ**lr}Nc_Am1GwbB#!mJ^oyQFeB z9Tas$OAwrj&$hE98qOchO>u|@*2%k-X`0^rRIG66jwzFeQaj)y=E8}@NcPz*&BhunuD1*RAuzQZNoXLK;slG4 z5l_Iw|Aw^9xH+szu9N6QC_t2c!pqYLRdIK6;B;m9~wfv^`YS|z_oDLJ#=$>KrRcg+V*}e!nfjX33tMaKB zqe|_6CWAZ(a{}6cJWQfrHe4PFc~Zt87WqNMDK;rignCm(+x5EB;+KSdOmx00-;-=> z@;3}vGbwMD`aRI?{r{FjY^usOhkIqHC2QfbG_l*Z+rdiIY%J@EwBKWiLiu%m-GwEd zmS28NXjQg<1K!ISFE>urc!j^n^x2R~OX)a5IN;r~y!|XL5FV$gcA3LhZM4M3X?bY5 zdjW_-i^Jw9@a5v+mC9|_Ikpuc-dIeRZUFJ+f_eNmC%}FD=*kLpE%2R1jf_!i7AaB zsq#+gy1WYQ6Cq0dj60;bJB;B$Rf#f@=y=w0;Vg%TP93PcKVU3nhVMpTI? zc1I8Q*Q}CiIc%65?6NOYcm*(5OiqOT!2a}=_DPgF^R4GDyK=Al(YFFvGjmR!GZEs; zZtD0P>PzhiYsXohk2NXDGsf%eVx# zjQ6GqL%Yt9eSOJ(siIBNc4yOKz&OWo`H3KVW7Dj%W&F3Rg8&hxl5o3zx?T6B8)h!J zPF}}>NE92DmROvdEC&d>9x`f`z`--ye_y1a`{QPQ@Oqd;sR5m)tUfZ0CS zA74O%ywMZ8km^Qb&bGN^%=V?5qIthYd>gyZJP_AYs61@{cviOwgb~SQ>G`%|(-fU2 zL7;X`&yapYSKqNd;_Wfvn|+hPd^?ZeC4T<6hNUXJcco@bhv()-0k1sHrmlSY;Zq9^ zZ~>Bdgofr5@Dh@kg4~oxU1tJS-6j~(C#?2UmH_8*S$0vGpq8=@zozeE&y;6S8s49J zPcuJ{g;^an{?RWNZfSN-%gi zC{d>L8FJp~kL~1T+x7B7HDcwZw+Ga|4a$M+A5ZJl{QT+`%{#^Ksn>Um1dWs9KlGxw z>{6~i2pJi4gEq5g_T(|`ta!}_e|`?yzaQp0I9^Y1BCdA1!kzlk96_bJ94R;ceHKZT zv3gOaKERIA`wY+c1zZxiZl=U`-$45~Xv5YUWT*S>@_m`PLL=$wlVknjEo9e8MKZtA zWpLIHvQLg^nb~EeG;{j-yMKvGe+jZYxPo9xk;<8?<^9j&sy_)hK^B20{_Ax8 zTSa6(jAXCgPRR0HbSb#|lPm+OEX6D(_~AD%mofk9%6-ACC3{F*sGE2yvzfSM*OS@$ z;%|b8YMA?%s>-47v+ZVYd|jvdjm-uQJ`UpxB3`9GcnN{;aH{MU0;&wCW;qYO+`K~l zj@m>QGEU;1^jcoWiIe`J|x3Ws8Gnr7BwPOPpIIp~{(2YJ;L=W=X z9N8gcX#Z6)#gdt}I#kv7Fx{WI^WC7TP*)t25R5H#7iB@A`p zaPFi^77#x|9W&@udB(7{w5Yr;3}_$GmOQL_V=h!KJTDqagBlD*WIU6S z&Et}5W|#^d?tt>b9aYgYpl0IQ*sZVp zht^bH40*)=1HC{(zkb=mzqZ?1LkMJ8C#lCZn!8x*2x|$T#|#9Z$0C=jD8rI;xd7K? zK)j;?$B+pjr>Z?AGz}O@0Oy)D0e0*vN5WRY zbhrzpKyamj*ny;bZu?7iTyxA1a+y_-v^wP=9HY;g!qT#g9Hax2grsX@Mw65?iAQ?sxA(RTrG>rM9$^~qKN*VJ&q%w&F2b2gQ% zv++wdlDCM7NDlKysb#&4ppsmRQA1_qBFaus)QPf@WhA)_`JxqsMGZ)52nz@lW!oed z1O`qKWfPp6E;#Ce-AHN)g5!>bOP+AhYz;3_MA!gZcGz zj-1(A;0Pf5k?RU16zz*Vil{ z6wp>s$|MyE${9(@8B;-79w_xD%aU~_gW8ZUOc(j0FWZ`-{eEcs&ovmo%_m=)30?1+ zBPbnIZA9!hK1s$yGmo>kiA34X_;{i-_#e4A$|LpPw}1*nbj)WUJ}T(iBVQxWst?IH zAesKtfbt4(X8Q!N&VyxiBC?f$XWCL35C>*}o@7~SksUd4#_IS)W<-9`^UWcaD{^j$El>FCMM%F3FL$psaC3;zHVRq1HYa_nv7u{KjmC@U)}n*(a( zF;d&%7|UEvUoH;W4bMKSvO!23($c_MMhM43H(Xnf_y+ZEyb)@N)wH!5of%=ljCE_j zlGZ6)-sr|x@dWTgES`?(Ty;O%AZ@wvUasIvW2rb!)d^_JX`8X;GrHnkn(!>M@jNhI zB9ewF4!1qT9azB)^2+q(EjwD)wl;vk02+4ybXd#VUkI?%fH!CgLm(XjnH)RvPqq=U zSwtp6d=kVb8N!?+9t9tF-s?XIkvCJ? z(bTOChTkRH{{U~c#5H%*HK+DL$yxXvq2b;OsP+SEHCO7pyW8qgMWR33cl;?y9gVRv zHVESwV?b+1fE<%=YL@2hobydNs>YFA-O*`Jxu^zhlnjO+1favDpg3 z(XbA#Z~+aCtl1^)B;YE_jv=FI(Ms7W-}Yi@ciyANRm4@tOjg#>#(uo#e~?xShcq{v znmlr$)?BD5a%G*P9WoS2Y&BH?`hDH_Jlg{_{pCABV6##?Q($#Oo-(+}{8MI>_bxWK z(>f^Of_a}To!oeC&Ub=-MsZDGI%A57hwZ;4e3fTvQTH&~>RAYdu~dKnXdl6CS{(SA z=gWq;1F*+W6xky+7Yj{0_1y0B%HFJI%|8>mpHF15Q#dv%7E1pB7lQNiD%p&InH#v{ zVtr(`x>5jSI!B9+2gy*&TIMt|C|MgvOo5MpMChJLbL7DA10Q;PvO|+3fur?CKZ?b_1Z7kCwL(;58S zWS^4Hc55pIl4wgZ_-Y(6r#me^sa^FnkJl?QEimyZGUHaHP4%KBQ!il1XH)`wj9Nl+cLirFS_4 zeZe;*QLuW)J~=8|>A}H~?_-Sh_XYzS~JiLri2$jPqiUFKw%-#MBUB-^W965CIVdfeq}|duArv`K4`%mGBTPO z%eT<+RRrxA5;Vu)uS{}j#u_*AS^g@mWX_|lfsZB`hBoJwY9zeXf8nF0Z;XdO!A*Hc zebF8DJfFESGvJ72&u-o+54TL_{{U*ZkPk@PkG$@iiq~(2*$k3*f94YG{{SUyRCzI5X9NyZEFVeJe0(Y9**XZyX*4ZV*>T z4J?0{q+mWJp()}=4UV67>Th)CbOYbIL!jrW+Yb$qRqu@B)_bwNDvY9hoXgC#N;GEpq$I(DB;u z92H5oT_PBQW5u!HpNK}9lFhOZ&^N&#tgoh1Lejwn#)yIa z)TPZI0;EvfDz*BTs3r%Lbx~+PFjbV>LK~a{0r)3io_aakJtU8YR`PQW2>7bE+^S+4 zpKlX^xiqr}=iqRcPkW+}{WE5yJ{1v`s*PO_81V%;*Xg2T{{Wj9a?sFrnQ>ujYN**G zdX_ln}9 zSyu8o2ecKJibBT+H)0gZd!m4dk^bpa=HB&!F{{V*lKXOrmR-aO(&hmQ( zK77#`C@M(S);slx2k%M{M1L_e*m(^yZSqYlC7zhy&OpbH{I_mC1vNYALCr)?kH@@j zKZ20RYgm8kBbB40*^j&^_o}JsZIx2Kn@DUg^_cBIZO?U^>1VcAvp8K++pTYS=9z09 zrw8igYs^(qpHfio+aJwldo!k`hlQdQCNm@6=j832owr)C2y=ddq3riRP#*%S8gr(` zXQk9UTe?A(kc}?X!{ByO7|OWL9VAluxM?V*mCQP3qpKzL0lwnWn0!zly9D<9e? zeJ#$%LnF!8KCLNtV{%YCD}oZs;bAJbd}9k}4bRwUYhCHeVia_+HOED^*m+Utn+nl=U(6 z?xi8$x^_yM@ak5NOM-YjOz~_!Eyz}8I3MdrBU*ejx?goT8DzO0XDSG^7dmRl^8>`A zuXgN@k=;Jm(!Kit*ZG3dn5dwooS#a!K`Z&|(S!2~J;o1`*-15^O$zr2^Rr8SIY}VrVyByY-15Rb>4_<3X+EL& zCWKdKs3`=3cNq64-4uZSZQmq^2Mvfn5Sk6(bSWnwWi7yXBs*r-!UGwN%t}le(tith!3My-->@HJuygd9IHyQ;oftd9L3N zPj}sY4~Q)lTtQ@xn$;L}O)TMqUcfYuSe*1vTCG;D8=5O~jFmGSI;Fw6bdq#qKMH9acXsCuL7U*?-n_1(shuwmZeO4O z0OWVOe(xuKKac*Q+2;@3s%>{^@Ai{R1x!tjmN^_8{U_@Y(5+7f_Ho0UI3u03o|#-X zHB~Mzetd@@tgknSpubyeqA6cdTua(O^;*Vm;(zzmUl$)}4!nZ-&ZDj70SBR1Z-|MB(5*ooo%M%Zr8!7m3ELIj#llgG~#_Fb=>cWp+JQ?h@uApXix_M=}&D+J?g!mB4`9Y&pBgd9R7S^zO`i^4k^K zrEhgkSPmT6h#vB>J%-WxM;cXCzuBjatOw1`{)$Ix_X%XVcTPGFwC01IK1$25c)Ocz zUCrP$R9qO{N5Q!MtJ%j1^(*yZG(r?xdctkjB2#Z^PFJGya?kz9+i!RcM6T23{S zQK;wQR^zu@n>1iP*hqJl=Vj5j4*PxLC?%4!hNgljU8VIfxEPx61CyNlmoJJ~Jt33F8#};# zTzKVo9?)aBS^GJ&*{t%2s|!T(HLud?cF7ILHxv97_}x#FrV)}cgW#KeK0K0WyGdZ~5p7_eMF&!wXd1=OZDALnpa#!8SJi=x4`2^uno|cuWB($Zu zIN50=ral2imY5QDQ%cT07Y>QC!a6J2X@`g0qDxFuLZ!^lMBAIUAgm;(a8chAhcKvy z=)vd|chSV-lFT0vtP4m*qsS?tR{AFo&A5+xWzDFIa>7T5R~k3ul3AXCO9ew}H^IS9 zvT%B-l0e4{Ie`2Wc_^AksN&M5fD(@su^{D@5%lx3aqEQGnawzJNIqdjSO@2-1;ImX z<7_7cprWL7M4J)>da{8<5ym+p$U`Uyq9b&qN16nXivz(a1PIa-a=tk5UP$;SiXwKx z1|ya?=Sk1OF>~62f7l^S$;K2L$Nk?on^hSLELo|S13yNri%vSB!j$Fp6*=DxX#Ohd zNx$|83_ypIG$ zCsG=e&@xdZmm#SFvyN%E%*Gwb_u!>!^Ggyt7}QV`U8%Z(*(e>7H4Nvf8jvp~R&gEE z@ol`lo7Bj3K;C(GkGOMFCY6yy`DA!&8Vj6BzyWjm+T(M$zPZ|iJ(yU2tAFJr+EbLx z(%iV0J?B&0dAY}`205-*80}DO9090!=-&kzvanK8x>JJhWsT}dd=0w-Lg*%uc)cwq^6FZ28Kr)Kmk3S(^&x?t(O~Ta)x}QpryKl_0^<`k>=B~2@(aj4bjp6KjL1`R_S*=U6`93KU%*e)E z?0hnaIAtg}gO;2X6qZOI?C0Vzmy&^z4lEYQf7>)?oJyrmoYeHLEjn0ldYmuz?l|csiB{hVc0HRBCrHJ4c$bVF9 zq0f1l+3s2*9z4|A>aN=xx#IwJJ0we6K~J9cc9K4;cPlYHT=NV`Gu%DVd>{{RIt z7y$aAO@9x{6_Y^k7jTGlbcnHz!RteY!R>6qA? z=)g!`aMuoA^%lNLcwsG?R+3ImPcpH*3tVqDDp?&fqiinjT*izzxw{?Kj;;zuHv%k- zW7qYHv%@t}SzSbG#4L19$r~KwDuatSp6f@%c37hs*-srCKLQ%qSC z1{?^(M%yfP6|uZg+I5!5;q(Swa)FREjor)1DIFSKHR20gUYSi(_{u2A?_sH6$!}Z{ z)pcGRxzS%N^!G{!7ePC6HFIg|51v;r zCtB)*+h@4R^iO`mIS(gb^gNbsJDZJ1anZLV12)QW{;PEtG!H$MRojUycF4=>u5|4n zZqV>g$z>KAM^u6USX#c+DBklGmdkOvOUb>|WRHHqRB;{Wv}bVTeeJYtwNlc+>2yeO zAp^hxl=;zaoi*(~36~AIE+^ULo3uVI;W}<1r|k}|m(pJR!0k>9Ao2xYJA*r0jC~Ji zvA@bl^ZY_BbeS8mT%+)py?!g~_)$e5Z1*XISnHr}%@hB-eKx`G&(*x?%W2eb@iY&;X^O=CQ5Zu|f|6H*qmXH!5N$v9a4ExBD|;)^{6 zEqke{YTR2X4F%l#6Uk_wWVavsmdjY-^`?-&KMqR#h9|yTB9z1`JK283Z7x62A@!G8 zhA6F?LFD?U=pOxwwWop@8>XHJ;scOtaGec#IWGWtAg7Lk>p0a}bP)cfGXNjFcSbB$ zmpD3jpIIR5Gt2(&O5MUSjP6G+r748tXz)DH(mYX!BN0--$XEu$8h?T_6OQha9_^Ek zY?`v^KyQE;NpOH}PP5_7SQ&7q)q)I;@Fyv5G7;)Uf%@Ag821r|LO}g(krqpwmxd1p zgc#1&Ddcww_pZToQilfP;IFLGQ%z~2k?s#Kfrm*s=%RLmO&q)_TF~I<1_ungI3T6= ze^O@Q8*yw;Bt@jS-~vJMR?kHFU1i{DiV{8ks7VpbcYDNsOc}~`?WlS7*%<4jYDjc( zbCu-&)4z~F%&mo^>7T2}yFYRbA*}isNFE?yeAA75Z8(MpA284AjqRD-gBW~C$w0^x zxaH1leq(Zy_?QM{d&u!83Et}s`dOS$kFuI^reO6Ot>3CO6bNX9er7zM7PoSN?}@-` z8a&SF-A{XVj5L0+$J~(S%1&P~$Aemd4~@fjO5#3BJffI~G~v!89wkU{gdbSs{;|qS zrfW_a8=OB?yCB8(e)mCeywNea(>AOgoff#ZXZueL@@UG*aQ?b}(5|pI)$gO#gYWy+ zw6aW|@?BqYeAW1b7={kAPCn5{YoP>)@t+0+ndGI1cgr5$F5N+>y;3x3h(4kGw8k|b zkVH>$knGs;`$Z(R_$AVTWyQ|^irvxFJ=;pwfy^{#??mWh z#$|bS-&xpuao*a9;fC7!evcV@0 z(mF>1@ZCG^b}91`CYY(UdjZZJd6al@7-J8-?4A;bZ5$6i@qm){R6CJme+;6S(IMvM zx81tWah!eB6m>w7VyI~+z%3z!0DIZo;o+PrE;NECSZD5suF@P1JUrFZN}kCn$OVlg zkPoWVQVE^x#2fo=OSewxZS>MIKWZkBehKGCG4`I(<>0eDDj9}j?aPlQgw&Bl53cg# z=2XW-%;oRN@=4Opke`ZZLJf_-n~w+FpQVBFhZ4cDPCPhl3~l#rlO0qP!NPgQ=OAkV z8Kl+HNGatyz&cMQnWb;6*Eo_vSG8Aqa&$!ul4q+}NcpI&*N>s0V>8IS@A6Eh;w(7E z2R7A1RSYCPDOmC5x@0_MWrB^Zp1HXfs$FmvKe&rRnS4jLrYpf&fN(62Bl`oZ(N1JY@&Z!t=cZ-fxhQh`VPz-qt1zTHN zSQ>w}Td2LyGQ0UI!EkP(C$6unaXg}Eh<(VdG=@0at?msN=1vuKmy2MK#@e3tKz*3n zUHp|nB4=iv%-u(m<|X9()QF(GztuG?fxl)p2IJ&|QG$yc@)MwpqIn6WMUjnm&TdD6 zQ0KPOOgD*up8Hepj$Y?lF=f3v(wKJC(`wNS~ORP5itD9*Bm?@LoWByz)!0;ZOq zj_qwFlhhh#eVLkAe)Tq;(Nr)#7q(WOMT8Zal9KGsRZ&!rvok}7-V-I$L;nC(X^TH@ zoyCXlI8B;s%Wo6 zV&fWjW3hwco`SNbCuOr2W23kJDy~b3h-Q0rQ(AtsIGBI=QNeMbnEJIWVepjbKO&nN z243n~;i@q2CX&#a;wzlj9v`frry3@jJp)%SnuNoxxnb7R)u$uwMn@n=RYm6u)Kbyd z=_p?KvkYOUeyZ4tIa{3#U_V{Hsr!~N?T|(iqjb@qR%f@s1F{XIH11%`6*3>%I>HWh zY<|p9$~-~RR$42aTwbV<9C(E;3;e8a(=fPxslp5MJ4-&Eo<^Ravrzf0Rir~(+GWN@ zJ48`L7}C#IapB;uB;wk5UI(H}H|n&2R%eH6-9>Y{TxYh)QflVIBWs7Mr^@^~LnD?p zX=O`~{{V>HNj>FIb~}bhJX-u8lF6PRu5IC&%VAja_Y3yq6Q}Hz_QNc+RLQDrx@!{25Dr@7QBlZ%k9#Ziw=&CyyO7dm;rJB08rf{jQtLpns!46b1PzL zaS~(_Mo0ki!mg6v8!ITIY_h%5&e7|?J7qcABvj9-te!nV?wB>Bd~LTSn-7zA*Tqe1 zq>a)>+(Zru-+k1E)Kfj}jgqy`aliocMC)}Mv>75Ucw<2%9CAlRM@re}0-R`QZUE?a z+$~x|u3*)S08%83<7FhZy8;NuEs|W#vK&5WA_kmp5f-=WWj~?$@d`a{R5Xzps$3e- zIdwkeZ-CsMWjXE}?MU;YCJ)we z*agSS?*1Iyd>Qh2lfxll;d|c^Gty~k z^ic-)-Xp1X zJ*ug|ik>m$wwC_@6)u9`!aP4Qb#r9n3;uA|IFU&kyfmL}dA&zAS$NLsekZ^0ejUyp zJaPV6yZOGq_Osz?P9o!}<+xO~OMQYomsZSlgXGe7VD~FZtEUNY{rA&&Vw>^x-Ruyw zxujD?)72kb=c=%>aE}{tvm&UyRts&cY|2(QyI^mFV{h|aV}y7ohWK(8dIi2W%m(wy zHAqj6{Fjx~eM1f@+mHKC{MUC5pDg_4=KlWx_;U+qXB=b3^+w^vRZ?52oHa5YfOrV= zx5aj@54AyHvQt_qUSAa~dK%|*JF6x;0FrWarek|r*E!{^Xl`6dzyrm2y!}>u@?&R@ z#CE3)o_are8+qRaNP z+QhO;Et7?&Eg+UuRks=2qYv|3W#pL)p`4iORa#x$n)Mt#&XY>C6PyM@CabuL7zYkALePZ%?ZdG$g zsF>9_I71#pEj@o&{{SGIsBoy9(+3e8f1e|%`y=_N*UFaD+A8M*K#nJ14Y9TY`00smmEN#Ih03)+>-+yj4SfnyV18~P5t)8hpgVUOh z(CGlkEsgIVsz6cuQjfXfsV+q0&`F#Qdk*puS4-G*J^x%PWW6w$s~iYFR*4vkw6XYRuH{+ni9Hahrt-E6a-?X`4J zG~v~ff4BlCGK>+L**-FrA-&o5X%EGLI?zr7XvZkNXJg%n*jm@; zB<=$0FBJ^r8zy=Yh0J>-?&>(#itDC`km`cbft^0F@LfAZSpq3be1pG{R(g#5t8P}S zRcvLow#cxY%zkFlR!9@0@LI_ zm_=f5&f-csLrGf&Ey|9l`ZzJaM2{}y15Z^S@wL8+I%{>%R?r$)9}H1c0va;hb60$4 z4}Vi;r=*#}Z%Z=71en9x(p&LCwVY*A>1a5R(bYj6Gp3NfHW!_ak`6PDY6v};KhfrM zWS$2`=&;u@;Dg|}hZ%zx5mw|DIn%t1jAzy>dBM|g>~wCdxw=nAYaC3lJhZXHmm|8u zRdMr~JdCNMb!9Md4K-A`I&-#3-An9;xC)u~4lT9Ny9d3ij2sS{vb`1BE1`c7+oX}r z41%K#C&{?xxTj`(L&KF^RZClIu}5rP@y7K_x9TZDU;1kX&~^z#=U2!jSrFRXHxTOG;_x!NgMP zIrAM5(lL|~d$bIDbW%v$l1Xop2yTQEz7sM)IV&9rC7w!(S!vEZ6lrPvmHXlZZkU#x z7i{d5{z@dYyXOIDrl=Bh6@Jrr(fJ9D&_Iy;O-Q8X(4$e#?X9}(1az(=B%s~ExKG;O8V+r z7<8A(JwwA-#tvu9`6~HnoHApR;CD?@+-dArdP@798LNUPWX2n)J7bNyrkdMZec|g= zR;ezWImDFrx>yT+D-rin!d!6IXgijl6`E_Zr138r+io>ZZIal^Mo$AGoz09hyaB=7 zhT9OUn;|DT3fbkE=4%Z&#WyP$-oW;x$r9p9LkyWgfEz2Zci%6$YnP zy8|G!e-(cEUm~IK1Od=jzyAO=qwZQ)L9(o~m2mfL2gInsTi+R?2Kl*DotA3CvWTL} zBBqq3Cf7q%bGX58qKL~;JHsPj>Pm<4Q!KMh1$$?wmGMg$E@Rm27Ln8~VUKoFCAyA> z{{YbLIT7o1H6TA^&N8{}>f2XnxlKV_z}caBYrr`f2L)dn-S$>CXWi?b*Bom#4&3$X zp>edZirRQ}z12O-Ws$_>knj0juA`;sfB4Bo}Whx;4F4{V=!T3I|=Tzl`Q@ z_BvyTf%@)S(C;M%-%AAzBviER430M758{35Mzx#esWg958j~8(*XD|UQW}#QywKKx z{YYv|Ye2mN`jFJ4>+?fbnv_WnK)pr}^GB^TAMX(v6k`3+YfUw{m);@XC}KpRB52%7 z)`qg8_lV1})Fzop){nf|M0A0pE+7v*lWnMUj?)7jA@Ll7NGF+g&TyH1WNmhxTF--b zQJl>u`&8LKVltB2qZ$6r$m1Nh86{mCONObE9XZnf0J3?fb*lGN84vb`Hc{h@5$-8W zjzjBGR0&!39n2Xm^S>v^Ao{oq0o2jjaIhHCJ&bS{l@SNC%;6 zMx-?+HKDB;K3{?6x!-6Oe?qJmN_Kx|nBq`5HU2*pzSa{J(7{;(Ku5e~-N< zw5~(fKh12rJH7XAHkyer*Bro4k5;U;;@vf_z4ViA0X^Hk&{0_D#VcPKEpte4Jwa8+ zy*p@j=);vC3thh&+gl@?JNW%sho6tQt6=T9g`H{1F3)O1DXm^;YV%c2&`q>93E2?ya;s==<%x@rXp-*oadqD!(j|rTrP>eCjX}B2GvKAi>bHJmlk-}sj?U!-oP+`M zQKGYyW12`ukeUisH1zq7Gl94TRxMSsov)&S=VVn6O(|84rL8VpzQM95t9lj?MP&(=3$P5 z-jm*vaL`@uRW^`h!cTaU92FAu4IU*)ih-x|k8_WgB+;}-N=GkuBvqF`#R50_H9WF8 zq0coW>27%dI+Cqprj5HbfF4V7&d+T8TfK=`d`aBymEMmPjnxfWw>crwXTdgPRB3xUkT;CzGxeNX^96Tk z;@^_<*96QX;_F+p{{V<+`Ru(S$o+0g;+N65;Zos<1i08aDG{^DRzfJ(KH|9`rMp}i zB=Ir9u6B~_9m45xH#~ymeYC#mPkV{Bs&MB{!{3nrc_UytY_2}itauWJH+E}JT?Ue4 zs+UuTP)Js_5tlO{>{V|GaRr*iz;CqNrxDdX&v1#j+c-J?DyEx^B)dWliq%tB6Aptj zo88Be`p(&0Z-u_gy2Xj+2XN+st2-XvBGiM-421O!oS^jxaOxQ7(1YcT|6B zyILp%h6@`bUjG1hK5A!Ww08Ty9>+s^mRprEf$tGILP2mF95)$LJL5(!p~StR@P%C* z_DD5dq@9L0)5gq`o>=F)R*0BeO&?%q1LQ1>rG7%aR!Yo5B6%|k^^ zPZ5#N?%N)~eAkkAmf>5+9D3H{dD}MJqejQ$!;LpuZ$@ zJ{;AV6KKRxHZXZ=aQ*>t#irqoNXK_gNd-J3kN`L20;%J=GgHC$c&Dj8y&RwhfY45o zt;=2+{i>va`iDR&9mmmSp{jwA8C>{Y-B>RAr>fpQ*{)sqG)qeU%=@Z|v_1;iSVpoi zGnLX@ZReGH+wPdITlg={BRGEjZ)^rZ1<0HifsE(J+`8lS<0gm?i%QSsrCnR-Y2cQm z+++=fj*-W{=mvk1@fQDU_1WQZ;YedRCO5u6hDyeEPiBIXvl_bDFxrae838A9 z#_Zp}C2Ko4c>G&e7OeMmM{M*U6+acZ_-8vpcxB4VR1Wag;ltt=LI*y}emPw0wHRFt z9sdB;Wm8@=(R4aO`6%I$&KPuUcg!n&rap?py10*UL*Q;BJ%yxw+w^$R?lMk7>o<9H49nL$`Lb`o|?H&TznvEdJKW z!^S;GkG$-LCUM9tXTvK%X_SnT)}JQsnsZ=l{dpYY;OAv6?O-0fq@On;W0{|$h?DL? zU&K+iYXx*Hh1`=s{s>agBgY*?>4tN_@$9fzz7X-{a;xa zbBOg}IPq}}`;`qDDqePoG#Lm9y=F7w=_InFJ^7~}66LR(!Pbi)gY@za6xN_V6?kYhdPSMWf=6$Q1tM+8o`#`vUUjnL+7UCb40HG;zM!1800-+k`66>= zdw!CABf+h~S@|lswlYiY)Xws2z)UvjeQpHSS4ev3dK+C#w}k-oFp z7J3Hc{L_vrn)M0z4k@_iV>govU$VeZ&c1?w{*!R0s)f36&U$@Cf zirr5c$YXQvQ0i_~!JnF@qy1`%faCk2RzC3HbW~nnA^S9NX~)E-#;3VsS)2d|mdXu9 z+Sg9~5?4YW(Sg}_$!A|x9M1YGD!PN6ol`4_KLN5hLTUFvJ*dGHRxek(O2a9qOk`p(LR zkgF@Hrf6iBTA5w;X`}9k%}0U65FoS~kNnD+42SOu{u(-VI}L?{;Gb+@9e~HV4Jg{% zyrV;Vnnw6U(A1yA@=1eHS2VfbcP$2t9*E3-DWz3YS?J>c_jx{hscff(yETn(xG7

MjN-XnhxY`vLex|nhj3NT<6OGZmB-bjP7B?+4AQwtG!0$vw`U zxrH5;*(u$slBQUXkymDy%&(z-Ln@&s!OqftM+!lET^?FPhln{@x#YX|Ws*9crH7c3 zF`u6`76!uvz2uhpAb;AXmB4(C$+3BW>cQo@n?!Wej2suZm7aCMfR{sXm9EAyRR|;I z(574NH5wYnvW}oMZQMAx+vk%@D_}1_RJ;d#iX%%6`hA3*=b3I3mm7?AHo2p1Oom6d8cL&P%?Ae@ z@;aq=nei#3Yu+4&iI%vH)7->7!~4w4Y$@;FK|)n0_h8MIk)cmLYH+)(?maP>7Tb?|8n^A9t%vhD|;= zRor1AsH&!BlG80#cRp&$l2*`hA-GTK#SJ@}*PgC$BRS=_iSk0Kve|Ei@2s7Y$lgqA z@W6cC*W4E$FFqVzE%W$q4m`8<{{W>u>Zi~x($zJuI(<+>Np4SxTKg5+%MTAnXp+@y zzEsN|d)lUkvLbiboDzC0JeOPF6g*N`Xq{5Xox964K@jmI?eQvAqSa)w&g#3>bMJRj zcBZCwX~&b@*WkR~r<(F%am&Z;J^uiI!?(r7G0St${f|#q#TRNGdV zb8_fBA;H#579^*GzMKF%*>@w{7m>@>W5qS=`hVuV3^+dJ^XvEj0K{S~yf4MPV1qv*9WaUX|!$|<;dHGjj$tKccc1yG1S2g@0#OHvc z^m*ijx~Z*mnCxY^IS0vI@y{OGxK5SArcV03&H5^4Zhh|X3yF4t+J(c4Eo`=WBy@DI z!>4N-lIIS9^b67GdQM#P$2mMlSDTj!E<0jfBPru7hkk{{K$2(oQ%DKeV?HFN>SxvJ zmN+zge}Y`4qT3ULjvn)z?a6(279XnG@iWWDjMKI4q;MPRI({8h?*-a_C@(SF81oGj zZSN~8K>XBA=kY0u3N}97B$B*FA*OuU@o_k4Tb5DN@Sg!t+G<0YUwUwPxWo%GTG0(OF}G$ka%c49wub zbLOS-O$>GRI)=UFz0(Y_xD1xLppaK7hUGNe188ka%cYiSp5dJA4g&GIZZElex94vE z04444R|eanym6g1JwOhIlsci!=;8ZGu; zo=TR{xRy?E$Bd1M=&h%|w$BWojxIFsxKAOduY7LHBf#}o4=ucxcbkhWxhB3o95`hr zH*C*sGS*gc{%yoOwvmpJt{V~)eTY=Hv$P9za=5|XIq^K7tFohYh{{S}*qn{Fa1@t{fPmd2tjmmk`$>ZkBlbYcwU*T02jwPq05_g7{ zBJqQ^mla$1Gu@v}#5Wr;k~Xrk5hDZhjcw>Ya-8ew~HGr^Z<6^_h2a+V$p&jQep9a}V`MpTCY5 zyw2ZMrQXkgOo;9B0Oq;R4jBYF#~|*ZU;P4}#c-HMAkj@Dvxj|^+qc857=FMRBk6C| zpfvZjr*q8nT`EEy`RZ~!@s)x6AE9G-DoKmtn04!Gni@&?WkKzBm9M|i)zv_1VWp&WfuVyk z4O_J}X?3RC!q*yxP_lws#v-GoXgc2Q9>YQ3IV(REY6ZWAXz#ogRe84B3*S7i?x_IO zKr6qy?6J6Zaslc55%BkCu-y1cs-m`2MOx_$jfk^hj+rWTiob}h932g+-$Lpuj14Yy zRnKE%n3JWw2Iw_bR z6Jm4Qt{26qoz0*`SMoPz%IU30$e}5&kdmz=R z?0jzV!5CVx8MyDas#lg=>=U(v``Ac(PoCSUw0ucusBmMdTMV69CR%>=bEW0Ef}lId z^81u(`Z{{ZOjXoT%nyJE{!|XbEHezpBOm71Wh}RDVP~7dwwOsY6&04}9sNh`pL3CV zCl^t_`z=)!eLsW7i4Vt;qd;4_OPi9Z9mEw>WB%2%R=@{0#UFUzn{uSkaTSuh=%u*A z=k<)c4j%ARz?&pl-(?b-G_~!^PpVT!)iOc9B^pRPmD{2O z?3j`bxgoU-ux_JCPx`9*NC$$AEIaZ=P)kkZqDxPMuqysYe3W@%{Zte$h=h$LTT1}& zQJ)S6BoH*BAxV6GNwUq<3_k-Jos6P$eaPlubJ)iWgOwu7YxK*MG;vD!1djEHIRJI% z)ifc;1d~c4h9}7*jvCgClkieFx3lgsHb+nSJzB=Jwe2Oj)DKOLS7xxs!xlXId|BJke<@r7;Gu|>mm`kU(r z!w(_)$7N}M;hUwJqq@1!)X+$IHb&qdapb*8YvbM&WQElHM{Jm6WCsDif{nzViSCY= zO>c&>DxACNi-{iX_V_DeOQ(|G4Mb;P920%(nsDGk>OH9Nt+vYz6>Y9J)3Tw?W8ri4 z0ggqv-wOpKK#e41dzN(G)y~m1>1L2TorcpM(?DdE7ZCku&{sC#jcEX|T^|`Xj?;3)%nsb$;XZWOP zr-#kw_$>O+@t3O&Gkjcsg3W@nRuz;oQDrWYC)CzOPR???FnmrF4Y=>RTi+$R7S~T3 zt`yzw7O?rt=6l39MsLknH}uS zji+EX&o!w$?Kb9goL_ZiQ%mhom{|OPWh4*}g1&|I*Qn*ZidkMIdDiwuOId;J=Ny1n zlJ;p8H8o3{aHXEEmNNOEl0eAY;&lU^zxi9P$Y|;9+!av`e2i>Dc{{?#hh$@YBy3I= z^Uo~j>a58n@Z^045=rx2H?wXV;oJ3wS!!sE?lls!H@W0IG#^(u@++QSX#W5gUMi@u zffR+U8R^k+X}M7~A}7TWD3eMB!@(AOkSQeeO#+riMH{4+R>d1ZE_=B&E^Be(0S7_rv=;i1f~h(ISmoB?LYb^*(7kB)39glJ9L-I_nf)J zpEWce41zP46HxCnfD&b>k=r*+ZTl?QGpL|a01kIM`ILD{B^Fgp7pN3{bpwJrj%mE( zbb5*!N{X$)rINHv)6x9?1rwrzPhJ^}F8dDLDA_0zvIm~&<2aHpSM3eLHps_&E(J9d z(nst#j~<|Iy!xj6F0T z)Q?hQQAenzgT=^dO=O{xf!#>;CP66jfzb4NlUXPO;(_gG^(JzUAzcT9k!syHp9Ff3 zB+#qkn_DiH=QM3I2Br`&Sapq?Kp5gE3O|Ui{qO?WQOBS8NvMEAH`O3ihdos zOKQ2YcJ9OKy{C}}pY>fv8Kase%M;rqZW(Kv$c4tq81-^<$gO_?-ZVe)Wki?+<`V9X z@c5}>V_T5zkPh-lY1tU@O*URff>u*#$=HF<383behU*}P_~5O_XI-7#xT)TkD~%nz zAI|B?lgXiDq2hNzXPT3{ZmeOgAae_{b_2w+RdBS$?i$48k;@qyg`Vk+XLhM5I|ZqO ziNuYX6GyK-;U8Ds;-L1E*_D3^vG)pD{WjbWto=G&7W|0nyA4kpdNtfpy3ux)Ktp@T zsmY6f*5s=DWW9*E7OJVSyCKZOanF85T&||mA8$J{DPIwhk-f}sH-WN9J7p@_#FeiP zhFCD^9DPFEWFELl7|FoLmx8gWc343m6nRA}J}Yfw-qB2dQ~9huR_BKOdnGyYBkxuv zNSVxynK{YW0J{$kc4gV;7Vva6FQbZT>7WggvNxXcCyxciWo1prjBMOHY@}sfb$k70 zJdoI-8@fY|TP+uCG!$F|TTK2Fud1b;z&5s&&RhnsW!UC2Iu)jB$A`EMOC;6Q=G`6Q zo+^&+TB6aTNa!=m&1r{aG_%Ds=xvllOi9%s_1Tf^gW$P84Dm(ErtL*U&iB2RKpX%% zPWzSKFLiY-ikL+#tU7&Q(~a;GMeyhK90uD|o`pqahI*3x0kgRc`tnxYlJ-|>lZdWW zcluENe(t{jt9)GlIGlP;awVJB}bR-igPx8&BMW zA2fRt)F{qzY28@P&B}v2Hd8?YM1A6n^5&F|_;{(M+Glu@WZ6SA4-^C5T1W(eh>SON(8=;FCGEPKn$GzB2!)&%d{M@g!xE$F@nH}@wfx87qb$z$X_G)84 z{$RaW(#E%2-qwOMmlfsHaN&wMWAq9dikX`6rb?*}0na=&cX~b;eQh*y-!HSwWng!8 zPlh)Ha_UNx*?lkTZ}wUS{{U#XKpZ*;ZdF^j%6fsO%VMmrcS0INPsHFW3m**FZBFkf zZF!S9FOx2DKKFkmpomnR^Bhd_89&i!oidd4oO5|$bwsdPB0QQ&d%jIv@3m_+WrD69 zd_h@qbkP^f6TMv>0m$fl7jzA#c*K&raymCB?hBCq5LAz?xlmKnz0d~IlID?s@u^&W zrC)YoL28S#hu0QEJF4N7zN&ULrLH3yK|7Vl_fX zowhNOx{ZRW_WXi zEL=~-bQXtGING3Nq;zzSBh?(T;nOS7bkIpjM-3GW5y1>B3}aq%hJk^}Oz)_Z!T6s! zBsQIpkJmrQCgx!{GgY~*>el+8r91{VO+{I5dAA{>`ljLZSe|_z7rdTiG@SfOh%fBT z1jP?{?wDyg_??nk>WNre+%h0R7;)5oK|Ld?oIs6{vAuPipSd$o(8hsLSAq6JgF*gk z&tr*x53G`(jkwy{+00$tR=fuD_1(S-?9XLOP6a?G#KyR{X3eIe?f(FaXk=@h9ML)K z0j1f&2l=kR_^p`K=pF#vlGhA<$$i+?{6_}-DQ^#!>29$1%zJ6$Gs@PEPIn-F8D4eb zNd+~a#4sIED9`MLzwIvX9p{a1mC(5NFQ_eq{-D{s*DvE5gC)#B532(Memz#~_-Es4 zL(M;Krk1j{8c14QB+m?B9Z5J**88yAKjS_*f@hpVL=5geV`lyT0KsDVF`KUO6(FJE zhFaN4>cLCh3=z;OKF(w3##gt1<65DMmLRY5IN1QL=6INC)h-5n};5q{;Irq=v>lr)u|< zl0fz$BzW>p2LN!^lkmVq;VX{KVE2+xAPHPKv}eO|hjttJ*SP!Lgk`weNipTg6MLo~ zRfPWlrz=2eiN3=!e4C%W6J0Z%j_*h96t@|O^^GYN9Lk=^Tre1=0+P6$nt0( z#UuMccdOmBdE}HCYlnLe5z#wS8t+Ndlaqkpv?duFypyz_0!Rxt#u2s}#?0rq>5@0Q zr&W9#Z(%ND$;Udr8*{a6TSZc0fd)IB5oNxE%DSce&j$egt&&$;~jKtQgR-5fJQt>Djj9w3Ui0n zM8l3))1R8EpNXKr3~7<$k!xIk;Z@o#^|8apR+$OQWn-gr#&%crqL3EH;X*BW4IsH}<_mvdguG9Mz z<}CfU{{Td}W~HSEzM2St{UgJ1^HKi*8qt&VCR&640Fw)zK44)`nOSq6){@}(!91e~ z>Z#&!m`KfMLr$l4Ox0Z;8J1C)k3&4IX!&OWKT9#<;?loh=M`#})kw)u;KlF4ol%Vy~t=~6gLv1m+ z(Y1%U?`j>;M-e&r;~#p6>&080gFwnweY%O-2gq)#C8xSo3{_UVwDS$y_$5PSr*7{L zckLB+9)SLYFzKBk%E3@I4CQo+IxiVA7m00!&m}(HV`%i4Jl%?QwMtOG9MT2rbSB(W7LI% z$OU59$vm)~WH{d+avDR8j?QqOV+lNhkQjUr4LQazu9VP?C`O*swuiMENCYBkJJ?H>2ZA zjCk*#9YD_E?F0BLif%TicqUp6J|;Kk?kk)l8R)Xht3k!Ki3kpK(Rk};^j7qDOMOTN zDyZF_nz8&>6>Hu-m$-aWjH;ERau2;0JEL6Ja^KvQfO;+|Pk5_^z_Mvsj#^|Y=8KJP zaDWVrem>iC_ol~D&<{ilFtAkIe=v8`QN;0&s`AoRwHF)PW+Ah~G$8djKfzgcyJ_1Q zDYwVZs*6p;7HQf9cL+i7al!l4$Yq9a%}X1>`z{3yn&jZ!8>Irnu?fP)^FgsR9D(<$ zY|)Q9gM{1Lg*Y%4?~#Oq8%f`vL=c?Bkir4kOLRM>7>+pw1P4YrfQT{MkfypnR~m4e zb1;gpG2|R%rkxvs$3(tNg8{mJSEe9&|m3TS4{4P-l0S4`31TuA%U9?O9L0H}hFug%eb zeqb#N;lgafJP=WOTOk+}wN!5Lf}ps3j>+n4)XI7?rk_v$0GX=BNLWuL8d^x6 zPFN0k?5Lkjbg7ZLy%L&c_|w!3{GEia$MjLj-PFfb1HbbjBz(f8lF?2{AUTK2WQgf2 zDgOZJtL6jjxataTNZ3=<(bO<^7|Q$%pkxk&;rd)JpbCHZhksA%nmL zrR=`yX*jRAt9EHFQ2IRVz|!3ED=%ML1n_~TSY$hfOBreYi8eVTs+LE!%^kti(8qId zbUW|SBiZ90QTIRH!Giead_d@Lp6N`9>2y^LB-V40N(MO!C~ZfJeMzm+ge=H ze5hj}4{VXrGzULpwxh^pYmXD$c%J_NNd?L?6LmHkm(w>(+QG3Tl>>_?xO$T1BuuMy zY*Cjs7%AL!ydJN2MbYBwd8F**%l@8!-P+5C59X0v>*d1lR4ukTH$1%Q(HX-V`%mhd z7E5g%Sbq~&o2^TEP*NYF2f@!6`<3mkg7tf)j*gz5J596bi;l04_iETh#Dzc%Eyk(@PAmb7Pt}=l~rL2OdWp zl=ukfa=4G+`(52%F7~H8P$;cNWlO;D--2gY6r%%4;KMwOQP1 zE;!QpLG>a2rDVH8?H+2@PiV2?I#cS&Ri9o@*~cKcl%(X6?4KO3OQia5b?@bLx{jyg z5V}@XTq`83tf+kR$a3i(a!Rd})%wKcNsyfGQ!eJc{3fQO&CAW%g2B%URct}_dU9AW z9_F8Vj(MxT9-)%&MOA8q2UG@`;BI$$Z#It;jO@D{H+zmQPl}hrHu!3{7?FS?bzt??!wQ?&#`Ud#IwNv{X92qB?4G8=E8E;L(CHvGH72jjHU{>$N?O*K?yV z(NZ@+)?KW-d}M4L@j|PGuic;lc;mX+ZPY_Aiq}n+mb49vkBWg)Z?#d=#Rav_bO0R< zceK&wdX?++d>lMC^<~|c?d$#B9;1_!llf+%j{QNzFj2PcM^M^WCn6@#PgDz=t8K$h zRzI~K2II~XhLTPvs}7>LJcuNYaQ^U({a`owqvJ^hJ#^2fk?~3hEOcI}4hO|AhZx!aYhKb$m4wZn)*XbGc`So8>(sA+jcVmmj>bWxR)lA_z;*Q-cJ>vloU}t3Y z+U;=RZU(2g&`%RxaF$5e8=XzZHy@fh1~rxny-jPqTyEU0P1a5WzHkkt>EO{<7Cl3yH4cPO<6*5O!%A(?A5(k%c&Uaes?JlVkMm468-cpf__uQYbI#@G9g6PgMbqKe%V<_P00)L+n-5y&3}qosXa?}|87irQLs z+JJso?Je&|L93ZQDu;-BJE?CTSI7A2kw-j^kOna80o&u=w)J*;3Qi=l+2U&_rnTFP zpCpY1(mXd~o`))9Y@0r7OK$CM<#@WyXqwwYStDLpd&pSl5u9osuf=5hLgNXln(BLV zXO5l;G(R$JCr(}pCt$Fez9+q0#BIz`h{1N_E+p;Y$!z$GgYD0Ty0yZhEXMh99+J@D z3HSuda(E*-z+6kg)|)7$z3};M!P?&)Tn{XM2-tWX)^f#fbxjDl!%)^Z)Jn#++qe=k z*W<{pYi6jpO2yDb$x5!SyO`dJ)jaBu*fS|8z>kpYxCf6sJI4>qKk!SeJ)~ zaIYBHbrQ*(4}+bb;*3n17Bxl9qX`&q&l;`11b%O^*6#>c@nQ1MdM|(sq>#@Y>5;m^!)_=k6<=W61?a zT7!-_cBF>7X<%{bBxfHmtzfC4tY*4u$Rr&w@DIsx4oHD5Ibkh6AmImZk^_(UQKf)o za%wBZ(v9KJ*G>TF=a&zXx`U1{00vDRJk1?+X+JQbw$<=+QKg4&N}8LE=_GI!R5Hin zSnxuTKNi?Q0ksl4#|-3}Hj?;Hf?`N~lz+;bBfr0Y?xx?w_FR5>TH7#( ziE3jyWOKb%HN9{-?7ZgZze8CN0j(NFx8&h#tz2JICF;jrcA1VA+GkU_%*iet51!am z)1&saYnm<}q`A~wYAdFWqq=scX=_?W!Jv7rX2-GPEpYcO*B5q$!`F@#sJF!=-jHNt zAGZVzmbX^58+clRl;BURb+fBjt!@tNoES|+t+ctYtzZ?Q#7 zQ#Of1oZ>N_XJvBUAhiDgiXXa^RB`-)o`~NaMP{2YD));kXl^pVVS8k20nXUMpmj7{ z*&%PgmEY4x+$ z!AWkcmD0k=LxT>$5Z%wsQ}G=eIq)G4CaZ12>FE(1v8BPjc|6p6>)TOi zw!sIjBs#mn%NZ)9J72{2i;R8L^+mAh4H3rB){)5Psn43;@ax@8**vo|89nJ3`7Tt- zd4cwR_0VarR@QcCEhcm6rLOugJYD1NSK@^BT~{>)qN;|Hcf!VpxBz~(0m{#Dd2{b_ z$oH2y;5d`ar}#p?J9TrtAcn;&NGHJK3O^A^4P9JwQ@FSmVf`?r^hH5LQ>G{Q1YOI| zPrYYN8u3*eOLlFl-en%N(q*wX&&S}cYVl1^B}C*h7{PO#PB6Jw5M1TETwPM^wS(1( zzh%GG66A*I9`~6fadX^DoKG@JnP;Gq=Wt~_do*t$@=#dUK1RmDsm|C8bzL)stTJ0@ z2Bw27fH}nU@=Y^5rqe?mGz@|^hP92r!gQRceHMSS_oDPk`ZMoMG-Q+sPMTVb`yYBq zrl0JOnrThtB#@o-Y5vdNi_vHMA9`q}OOa|%ni`|YgC6uf0qx`G@Q!kKfrPdNwnN^`H}NW7MlM6(GT8DP{(_Nh{tw3k~6+?rT!|f{Sf(~_@=+~ zL*|+hHF=?_OZ-uA=?M9w7K-k%0sf@Wh#Zenm-wQ)q;2`3_@=+~ZTY5+qG~{uEZS@T z075@`55+eR(EkAWm6+3Lc|ee+ekrd%-4EVH_@?3d5%W_*9;A7~m-wdP`WF1q{8ex~ z;U6>*^&zXxO#D+_))D*2{{R(M{)xZ5nh=ACISwHBgtIGOw9F44Jc^KQHT@wUyochO zhv+~2$Rdb!9KkL>QU3tdBuVROI}oM|J$e4?9`b>+*N^U-^H)MHw4m+OkHt0r0HPnf zj9RO|L_c{uLjlP0G0jMAvbQ_L0QaT3*;kFDA2l>YJxIfpkHxj0qWjPm>aMZ<>7c-J zP$R`X_`akcWPH*31y9*;%`_PfN0jHGR1c~?XDrb09fXbLX;?~e$U>I(5imN-kw#;Z?@Ok?hk;pcWyajx+qb-%#pG+ z#jXcpJXcNPYXsLS2RA_yvvy%~2^i?4d~$--`YNRE!zf49R+3~|Eyd~(d)Bi4BtB=uX~!Y2|%abP*sqFjNFdu$dy=LNdqb&jsB z?t7`>5iqmaeg~?qdo8;`WV}>KaFj()ELhZj&?;PZlK8IiOxc`0T$PILwdO7gY((!R zQwPJEySH|O!cezM?yTSS3c>cr#MZ0jwwjXFRLS(tUh#}zV`RAPBa&Zbl8KZxL9UKD z1De?~cW^u79I!B-rV=&=2~J(45`1#5n;ovA#Z5K6dI<}^Ii7K3Be3Lx6`)t#?^_!{4EZR*=n_xH~% zbQ8Bx)vTUJ*2laqb;a{i)_MS;le`f%&3*VDX<7+jZ2*psJP+$Z#x!f&{yS582e{hCpIto{{vBZ0_Uf@TpcI2v5 zz9Xb%%FR3u83YnCxXo3$v{g=S2dvYad)DTgv}-%5CXS>?$C+QJLvjl<>YtVt68m+x z*Iaw{NdE7@?fejkmZ|FOJXQHgZJDCJHip3CE}bp|2llcy(~Z=6C*9k2s+zvr6)P?> znc)$zwk=%T2Hf|~6VIumk)g5A4LpY-vc4@sGIrIE5tX2x`(l@Xtvn(4&aR?cb$_J` zrAs%#7;{>|z6kN~L0Th~z!y!i<{B`R*dhdBfS(s`uIeP5;+I^QX~4}&9C`Fh?KIh0 zyIQ5lXY(j|MG0X=yt+usmaYOJ+%?5_@%Mwc`O9o?Ohdkpq`*0!>~nT-u*TK0kF zw((Ps1VWqqg>;k;c*CS?Gq;LF7Fc6HODpB_BLP~Vu7qgl=_5X*3d-QGN$mbe}L!DQroOsbMH=etY2k+51m9e%zR zqZsxoCuV)^1!pSVGcI#Ak+H2i>L6uismR|WhGK|DGtB=0QoVu04-(jI^D+SEI+ikA zKp#{u9LUz_`bvW#!S!3Y`}SVB?CN(^a3z*gD2=9(gO=w59M(Qo`e?&uaLJL84-hl= zl&w%S;zH?P@jt_4)P2KUt1g+xq1{S=%;x&BHX-N#01HTZCrIDTKo~qu+4u#``+v7r z&1j>!Q^^~BA5$e9u-s>#K8vF`J0DFrY4C5W{{SkJj$Ft%c+CYAmF77bT+%a!lY&V- z0=WB1#qh+nBZaNd&e9ypcAo8omW+(&!z-1tK}}!8)VTHJBLwpvw(GFacd^n>IGQKL zDf&DOgM5dA;}=7xjV?@a&wYcQo2_fYPiu|>xX2@$hh>^ZQPsRm*0kdi0gUISNAOc` z)QqK_&4!*9(niyzj|6HQ$=_n)t-Ctlndh}vaSc;m%6Ov1Dbwu`K`!5>*!ZgsI&6~K zOpk%To7ty#a`8+S{3HGkQ_jxPC1fNrDw-Jph@Vw}5JuiB(4~YE=@n#w^+p;8%XP+o z2i_EGD2EqZW^@jnj*glT?I%WeaX)hG0o2WAVTIBjfHx=RR}alxhZL00aWFvn4=k4d z0R2u+YpQU=SsMo|JeME4WZ0IXM(4T_fB25i+!0FCjL%eB57@(h@)hEujC6Rx^lF_RfSqkrc= zCBms7ma48QXj)w)uMA)vNjMAH?XkaxxPHrMiLq&diH&i|-Q3UKv+n6}@#R@0)UJ?W znEwDl3m?P5ka4_>7}Dm-!R5YI-bYRPHj(dGmpRn@e-SLfbi`x!nyA>%q*8=4r^&H^sjMmb69ZSdS6hD z%b3Q3Um-Z~C`Mx=F~mvtWA7B&$lMt9FThf z+F3E>oFF@pJzTi?44`W;wfaK~&q0i*fs^X64|zxqklhBFin_uw^i>!!MH z*r+80MrtC_k-spyAJZOv+kW!7-)QahZZfb{y~O4~#~fn=DScG=Z@ay|nrArpXG&~7 zoIgt$-Vxx^e)PZW?HFTAi1Nunua(&ihd}Xh=kifTT`3-{VZXHR=#b`2?sEotG$%C5 zb}})(IBruDG6o$}Pmd?N0Z>vl7H=bqkB+b8lGjFmGYKR1bp7dx?zXC7lT{9UhJN); zS;ltQoLbB{_?K<>Aij5ePBb;6%;g|P@=mE;j{qnlHf(8yZ8?2PH_0=O47Y701 z8+85XqO;O810$rEp09BzEclZ3Zm+k^Y-nqrD@X)yf<4PI8>EU`e9WhNmWJjiizO^bbF&MIr> ze~4+*{1a6d$zvg~+Ug~4KRD+$HvQ^6x7Uc=_tsmW1DG16aQSS4!YfS}KD0tZ;_d$c zihsGMgnxAqA83aE02P(1y82T?MNPt^Y>wfgCZFb_!!+#LJ5+8zYOTgo_w0j9S9`V6 zzq@qqmHFiqW&Z%d5V=V9Ylho4oO7N=#t*=pss{_K!v?58-?0M&0Qn|`rLH-eS*1O7 zxRnK!UG8ROuBxJ@{{Zk>>6|{`6?wU~PtH`_t8>$6TmJy%-AbpoK_9alK_!sI03xX3v_#{oXFgN< z$2n6Xx6`og0|&ubGM&8gooqPd(t^2=G-q#;d67441uwV*);x++VZl^K5%0+?MoGew z*th76Fy~+pM{GOqvURYu9n;N>ka9wLC_x*Y&_yiLazBDw8=NVE43B7hk+uR)Jo}VH zQ5hW(fJap`xjr~UxE@Nu3=$3#v9ZV2!jaqnbAg1g0B=C4v~bTPecE~NjxbBy8a(z@ z^*5VcK6~288IKtO`;gjMXRFO5j0K;s<4Y_6_fz^o9V2h^Dw?abCT<|MyWb?E`6xF! zcg|540l5br<;?21^6e(kGMW&48L0fqrLDf*D%?!9vu=4c+DF`@F2PY%PYX+>nkE7E zo&u+<;>#sFi%nj)4w1CEa)~E<&R*i<;^UNp;zx(M2S=*oss`uL>8d0?F2h0oYNEG} zs~#~=O%y(5#mgK?Jjp1KSQ~V2b}8#2f9(>|iT0}Y=S(`si%21AO?`%Tj{1f(;#lr! zK7(YH*{7CTbI+L=Mzt8(B9=E6$s3w{!a~~{H08h9F6=qz{nS}(*A|oBQ&Uv_wwx33 z2$nKK3KbP#g=L#QMv z@Np%4818E7W5+B%Bkxyc-I3ZlqA@h{7z>$Eai!E_y_K^xbbD}qYU-PgS&%29Bh8Wm zn>N?U6lJziN5!_e@El(okJ@emwu0wxow<&tBkp)q6PV+Tf?kh$dA-A% zmb^Io!aVxeMMrZ+ITDcO9V$!OPIQDqGC7`V2tlV-f3yN5ME!n^rxes8*am z1X4apWoXeHWS@CZs>Y4B##M#6?WVLzQ1O8BPrxp>QFn~T^IW_?&9!C1mYvtqd#1ER z)_?%&8mXhP$1Ok<(Ydoc?V;HCtGXElb;fy}uAPz+L*{M(^3RHyS3M;SC#x`~kCUIV zmS1q_7J0gf$0jq{@$p_yHBna;OcbrQ)6~sXKCHRVI)hJu3et{R3l$3;TG;;p7(JuB zWK0^L@5dzqo%PolL^g_ZMFBpvgk{IflgVgo+%rc=@hwmlQ|CI}xF0@i%;)CD&1GHg zE-haXQA_yl@hklb2HWXnKD(nDOTFis(b{-!%V(rDl|`;|J!i=)u+it{t>TTX!&(Dc zIbdKc*K7w*0>79hp}!^4$2-l*FrFoLZZnS?o%5w<7IvWpI5fOC(KCR@q)6XXe#q*% zwa)2vz122Lg;dpYlj`c<1oQxMRiJ~-cF)1m)oq4?w~6a%Sy4Q745p_p0QY*ixMh78 zPtoylO-S*%dah1e$cl1&Eox0aqKq#00dDw zl-i^e0JYdI3~&VbWnSA|NpG&cTs=OXmYvtO7fJ?N>^B9wb^Z#Ai0NXd;4aZ(r+Xg@ z>qPOnjs!Wb0oepD&R0^$4l~TIN7)U>@qZuN-JL9*&V}CR5B?XXX`}om?2;l|P2%jrb)P}_}6 zV^$p$+=v8q1bygQ_(zD_X`*#Q?EFJ-FtyP`TM%rq1HO>_9~I{HnQ-#lxcPj4-}_(u zS8J!kJT;ykcmDuZJQB;t)7dxf+vom{<>j+kb8i0H)O9{9xKdEk8X)5(ePm^j#5X5x&yXG~pXgza ze0rfsvQ}Ad)fIQ@R@`B^#S_}^Nsvg%1BNowwg~VEij2cd_vz27y8Jv}YxMlJ$;sd5 zIcFC_$!kyRN-Ji}v@0Q7&G5bd|hB{?D;VW~fwf-FwGB|yBU5_D< zqfE7L3RF+T&b%6yXH#6(mghL=vsZCEoKM8n#xO-a1F@a-?ag#Mja4lTzM{5C3uA&e zO6IigK;Mz-vDTYq!u?~ayihg?*?(YbBJT#MMBv9#P9Yl+mw&F%p@Kq}%i8YXB4 zIU4M81-oP35j#`HCWBwQ4oGLCeMdSo;j15VTN^Mr6SeK|aL3*i32CP;(8rkEI%f&T zHg=Cy?0=JnKJbb1Ij!~JE+4DwB59^1;|x9-EkK8~J9%lrBdbQNd{AkL-mG)Er^7$h zD7JCCCe+^oGxwn*osV9hSp8pF1~ay)9}j1h?R?1oU$K%)s6N@>VNEEH)7!Px@xmF`FhSwOMkj1+)Uhb3TG zDP!K61f>KHD1Da9= zWf)2Ktp5NTUWKaIV}aSLsA+6?JG6Cj4CiQf?@E~3X=Kb1pQR(~JHy--J2=jIBP5p+ zPD$LK5~pzGvRk#vI%azr%iWEF$F^EN&a8fsKjTW4W;P$A2dP~YX9GPHdMbFSDOpVm zLs-BD$^5O!rlCaal1>mOWQV&*-2ij;O=`~LWS0_30Rq83Dagiorr0uY6ow2020Rkm z$s}`2Ky4UGL^-2ZK*uCu5af+eWTBKuAT)jnBV=I#j%d*iXwVFlH$emj5fl(WVC;tI z1ZW8YNawmV5D-BcN&rwn1OW_`5flo^@kNwb2+#x&K>;2}x+6T1bU>1l?n!-8UAhL` z5_64`-OVA~%7G5#kO|K;=P;Wi1wgFqmlT;8DX?;jI5|X0=!R2qygMm=-bqY!TJd$xkx_*ztl0`_j#$gXhCwwq2cH%^gs+)Vzvmjv^=O zKf~gvxT4b~qUR(IA4cPr5H`Z84mybhf(~*D;`LGAvV2i2#{plB*4yay57k!y^@%&Z zy$aIc9Ii=P9#Im3NhJZ~q?6*0DFmFMp_8&KqRJwQJf9Rqp_Gz$K&+IUq9?@`-4Fsw zETB;o11T;j6S_;eAO!4?PRPkA1r32Hi8$M2NQwYbNkEiH0!jrX0!i~kTu^aHC=8!^ zK}J$fBnfxHNTHWQv^bL09RNt&b406__@xCBmVHePCn`ug;RA>y@=8DuK^Q!Vu~ZOItaTc5pUuY_jrksJZXrmfeidmEBD`#Q9j#OKJ^%TC$B4lB$?hNQPI zs9+ud<$6c3KKS-?PYiQ32&a*{i*kAAlFgd1_^yQzsCK|IHjfjMoNj2}7MtGfSjv4(@A4;XC+3TI%BCNh|4Ps&ypv?SFR162N&JkG)rRKFZ1- zIpI0vnZVUHPVMg-={XyMe~Q*u^pbgcj!IXXQ>Y=Pf(lUyiIE-jw;KQ#nE11_XlJgL z*2-xg-D+VPNN~p7`7cj2wWAZ&xySajkGXkov|BX{yj^ydos~4~HAUhl>>G|?4?Wg# z{#u_eM0Q~vEIbEg5kUE#X@kkrk_(P{tvQyh`vylFe_7c-AgTCfp|w0QZIXsK;HGsn zeb^^a10Dn7tXqbI=@muZ0Q;$W>3tgyUE@#$lA{hPTYJJkhe=v;2RuFV|*lc@(2j!Ai7F|2T;%gWHLtQ zzpEs54fSx6=TU){w9M|91?4E?9k{v*UevfY;2*7`cEK^Kvm()(;=e@>B9(tZ+ zt=|)D`)0VJTuTijPSG=g~~qzyBMajUAT%`H^}Ol1vD-U&Klj_kBaLY zb0klth2DcWA^6l{*=HYhzf(-4@xL2R?~p4M#9`F@TS+FZ`{RBVlzP*-6i`;rv!Fvo{!cqloD=>db-Q4|?0z9d_rqaxD({ zEH4OZhqUXQEq0#%n^!J)!n*w{{-;9q`sr`pQ*XfC^Bj0*N^!xUdh%!Egwb?OIb#dY zi(8Zg-$&J!Fg*DTrH4v(V*mr~93iSxyIKG~B@D?eFrzgIgl++4=}mgUMTqIheJvq>p%!PI=w9S*OKEYz_yW>>wf85qNJ^HjAR zp*m814v;u>aA|S=i!bQ#GItRZTb=Wd>YOzcaBnS45&0K@q%WGYsN#+!sdv#46HnSY zPu!r%_}Z=7s;ka?fG6fuH(u$4nqa4_sy;Eqbo|OWZnL$SG0{?Z8j6vt_dxz=CVno8 zL5b#;Unjid>^%b(!5F2c4xa8`&ogp6#;Vu_l+S1snBcx~9 zr;IcBD_H43`ngZGn>`aonfdLHs5+{ro_8L4-c^l)-AeXEQ-VBf zIjSOe00HU*lMZ>j{1ovR$0V1>j%ZXP1Af^`5db!6Qj9>j5)zpUTFn<_PRKR*_h>ZMsEKXc4*5h}Jl1=vC z_BbKJajd}&BMTo_$F$AQqtZ(3a@0drCDV>8r*F_TwC)_eeOOv?x{Fva5lb`>vs2`2 zd~!ewj8DZ^iJC*}rr*KX3aX;AdEcs89^!mnN*i`M&M2`{97%D5Q2MRQkK7eyUmx0I z$ft&;Vn0_%`L0vmKcV}PImO$YlkY=q)o*B3jwjLRppQI@oJsjAik>vS%bZlQKMfIq z_pIJgA}lKJH#)FIMQri|%opsS>hcLFz|SCnWcZXx)>F>^0Ad*(?rM`2vZ3Qs%yBox z?*RlXZJvl>Sc*6s32c?IldB*`(&C+`rKZLkMOjdMbm<6bQ-p16iNFM(a?-(Ss(VLs zY>FUy%j*!O#bmG3^tT8hZ&tOSBI%fl-CHAdGM)N-PRY;3vNL(7f-fAqBz%^g*6UP2 zdR@Xvh~@+KNp(*RQGuq5ymyPalvx<@k+?r1lqH)v5nhSFtTo_oU}B^{0}s!>S|aEn={A9 zDw}HEX{{~9bu{%rat~-KGwhYp2MMil06w&PDroODv5lPHlABA!^^u1MmymwTknv2- zZl_@)Sn8m`!6mwajV%>F_Z88&o{t(;$>R$0XV*E^gD|A^xaoh;BzL zCF?ol7P$Ox|L3pi;1-;`j)shkSD{4M3tA)5VE^+?=%x+dgIXjY2^+!UY?a^sC;rY_|;C_0C zV5^EwE}@WSc-bjA5U$O;p<{8e1tiCu3@$}W#kV+5t5q=luUT8s@#T9k zC+?*mv4Wd}qB=b|BXF8`C7Gk+x*0Ghf$csc1zOT^9ip&?M>ZaI0nNge#W!AcON`^e zO1a=$;({_gR)d0Zk(`wVsz$zNNu4pXa_=8{Y1G&amM=oE}dtJOZ{LQ3NwXwGWt!sJ!Fs3wQj406HcoL*W?2(1i%ii6&H4<2G zKq`~8YDQc*W6@sX!07=F9S=ot&E16)i(r;ZD_t84ySE3p4q$kx$Cqy=HO;|>Ip;CY z)p`25mTD@e)6CM>F#N}eU6--?U&cL?Uno20z-FzDo$C**+~8z+te+8DS#+b<7LiiV zVctGVS=eR9-EQD+A#KuT!pdpH&S7v4Bn_}Rox=LQo0}d?E5`b#)AH!~ynikATkBP# zj?87U&FdGt`*bH8!Z zq=MH_`&EUG)md97s`;JG*#4ao>MmC6MG>^{v<8|EQ&kc4v!4>~cwQ!aoS5|)a!dNJ z$He}wTyV!OIfuvOR_a?%1H$H8pB>8h_UW3b?FE$nA)bC}j(Me?ch5Ypl1H4m&vr>k z$;l+-=Q&9v6(qLzWp`_mNsQJrP&2wiozWAB9FuN#O4<%h!eRRaw>xCtto@p`CBB); z&3QK9Pau8@ZH(IFk)~SOcg_wwlKf#psHUCH8YOcG>eGc9R#k)Rx(U4Hh!gV*(&6dm zK5}$H`tBs}TniyvB=a^Jbx3SBMgpwvYSb@GjG4K1pJ zUo+wtZrw*(Zq|;H>oo->UAm;m*lOc@WjM>29CpC!wj4FuKMzn?Y3ObI0GYtmkRg*G0tfcK|fY6k)vvtN^I$Z1-GH|4m>9V>K_f;SS)t3*0dR9bzG-B z+TDOk;%Re2aBxY==*)5M)-K1LZm=~mhu6g;zO1=#z!j9>)2`)JA^OkH-d5J(4{=2m zQ@N8aGvFJ)@mYQhr)+RB+I?r`$L7A5qm%8&jnw(gw5snLNkG_IA0eo2hLg}3}f5d>&uV5j*551Aj5-;9q$D*$g2zXzs7RR_i8e`yeGt-Y7B!8 z@;k0a3r|Tmq*`g_C)QvwJK@XjTRsW6&vC7=Sl^Q!Eq1uc-vQ(jKIPB2YC`y`X!+`B zdp+T7`!cI#FBj4|!M?dIVIy|Cp;sSky*+YFxBv#b8*{=_9Nc{*XFhGp=k&Z$b%z2; z8!JcKXeC?Gag;&0siuBURvnVhB=pfdLS6K85Prya@XOM?K#2Q8eqJk%TB9|0z6+++mKyrH$20?%0k(eR_jI+_ zEz#Dpg`#;~Bl?L3;CN&#S86j&4Aj*XgSw_{XD`+2jn?xpNb-GVA3s%Ke+Wq%7iIOO z&yq-t@w9+OcXvXwd?fP9ac5`~aSsc#hrJ?(sgO+F5B<755b@~Vz7{XV4HfbjtR=`WIFp7sFt__f*l zk_m@a9UG%QIKnd0e>YK+;OrC)Naz9ei{U*!s&+E)+{Y+z^BMcnPkoF%tdrfcZ1(`+ z>~kIBG7xi5!YX&38@ItGnDtc>2!A{`s^^sXWwH?B4v^!Ya~9L^jFo~LEG9=v9O)c= zZd`)IO^EG78308g-`oPiqwx45*3;F`C0OLBHesgI2bd@CQvum>M5W~nq~#2tS45IP z=2MjL%TVyh+E0n-7Q=!#b}H(p>VgT~5pJP2bZ@pWtQe;p45(FdlZAiBFw@aqCZ&!B zK=wGavK;7c4?X&omYMpVv#^49B&VZ3piDb(o@w<5xl#jWiOEz8U2iM2ffAkA7 z_hXvIRhvwPnIvl%(&xD0xLsF+Z0)t$+o)z~23$`;;IloR*;`RJ<0_VAj6S1E=v(ca zXFqbd!a}l!Q6etF(?Cx@B*c#{K1k+vHU%^a1tV{VRGg>aVaB3?akKyvjp~R5p|C)p z4gy%@k~t*~N`oBIf=K3#P0D=f02q1z0 zf(Rfmi3AV?5J3Pz1Vo@DVG;-xloCIj048hio_L&2!$->}_NJ02xy=0shz@dYMFIthX8M zcNnW_&#f8GlgY!VE3;cGVXvTyuASPWk-7ICp>VwK{1&&f%ju!57W$Xsnq%tZ?8o%@ z6`4)Cc|?HDH`t=e(?WMNMHxvb0p^lU$kJW10xm-+iX;Y6A(TY`D2YIvB$9v=xp%bz!gbE}Cd8Cqzr1i=Gd88BIm6U`8NJ&7V$^tynUAj_92%8c} zJkpX6s!%-(^p#;0nQE&Cu2iaL@)<#3)0DiX;@_%s2n_@g6ayg;<~`^VlaPq0;j^Ny z_Yad_@{oU38uY*~F%u@1hn#%V_P?XzJFI_d2cLd0Q(Qn;+k`IAT>X6W?pgU*nT8J+ zF5B1~u)Bq-BbHLq%Cm+C9r6cGtBbTd2LJ)$y9eO=ET4;Ru(g5Zv$4RJ00u&2uByiX zK8Bko%wq~vLZQtC&SM9V)S@x7=P#9wZ|aN`?rGV(!llLecJ;;+2R6Xwd8o!WE zwc(=>)3}~nw@65cpK2;|j}k+FaO7gl@$=O zW=Rl^Mu&pd3XA9uI+fL2vV7PE4qONJ%7!dbi zj5=C0k_H0QG4SIsz~>(j&Aw$|I}nr}4tyOUZkc6zy}s&iZA>V$z14G*57;CPdzs+RA?d?jcO^xM4?TsJLjf=3^R zHL0{*#_COsardW|pwjQXJn1S(^&t&rJ*PW=kX+igwq75N#yeHE=U)DVMcuRJR(|Q) zhZ$D_Eq2NabZ6^MJJ&0u**}G)WN~n{riw|LY9^~LYg*1tXx!)Gyvui=xV6t9*t`ha z@mdbo_`>;i;&B!3xK_zW7>%W@($)^@_-LfY~1fJdR@q8usKnK5HP(HR4wCugLvq!A#sZs_S11Vy%2q zHJL5McsW;IM_Z+Np2Zx5Zu(}sZ*s$33{=jn;u}--+-c&_4>pW*1SmHC&@K>C!9xuM zFMMna5;@xdJV#YBp6g$Ew%k7$#SCu^!#lPHf-#)!%_rckXAj@$D{R!Y6;ejZr^FpA zSTUvDR9l}KT4>x^9MQ=qEjqXP1&6ZHQ_e?v9_)@<%9~Y9ZliH=Wjr8tkcR&N6;o>p z-Z`w1xT2tOKe7DEqo}-bJ!7>y5Vyh76ZZtgRqELXu_NX^1w3X?;G)fYbJ|@E31fcL z^0qIy?tn{4T~z$k*UAAqhI*iS zUhkR8b+py;Z6$2le#=gg_a?=yb@Hwwi0C1SpK7LX!}3t&y!Wjk*Cql-vrSE_{{YNV zVzj__k-FXb$0}S__#2HeG$5vQ&UD4m)7DjFf8r_#Z|;DkGFyv=A%c#N&L~EHXLWMg zHvNRLO^#sU1qq<2oc{o5tf`ds4t~k#Z7?%%)}MG+O@IXJAP1s?t|~S(Y|-!vODV_M zRxGFTQW*g7S4yvrn>`cErB)1tV;D}dA0*JPZ7%$jhEtES6y`(Zon$^rrqEkXJOXjF z=Ny8bA@WNRbU^{Ic_lsrpzN$$3Uj&Jmp=p1<^XNZvmbIN~UkG!NB z#zY9@LNgWt$iN;eomKXYGr?^&D3A8s5AamAylZ-&A+?4edPdBD6xurMkA@Z(%J3Si zpubVJt-b9uuYfyp$7v;D_;#wUdvoh+YNn{xNez3lNF1^8R*<%%NOaC`t-F{8G^uxE zSlwOgZrj6lz8c~h13k1o@-_>13+5Z3!RmL95>ga>HVZfI89Iivzjp}P( z7*a2H+mnr6o@sORJWJy3f4*UWuGKBK1PxZ|DE?7pAVbp3Zv zGq_1sG+ml-%@s51=51YAiv+r9Fx}O~HM-SowMSD09V}B%DUK#Z5X)Qkj>@WQtDOmk zX^WcYF}fQ6029=BCOD?^__#@q@zO;DrVVJrNF6nD1$p^$^K#$SZFl(k4n7ZBiwRhJ z*Qjqqa`XMVv`{vPR=ZUR_ZD z{^S#}AFKQqEH?V$#>y#XEO5(Q8_X|KRdHpyvKb(qvCPuUJT!x`@daPo_$z|A&cw+r z*664!3}vD4lEMhvags2-zL)BG@nbOOpWXAlf6?&e&hm8k$Km6-t<})P*=3}ijBv|b z<`4>vq^{pR1>c5v>Mj+ltC9ykp|Q5Aao$3}c@hZOzAL71dw6R59bE;|#b|xXibEAa zjt1$F1C4DXqV&1?yTgav?cDswk<;`zbMjs}pKf1@@+(;PJ4;$zd;rRj+kIi@1DZTt zuwJ;k_G7_U1~@X-D4b5uq;ZU_{9TH0qiWqG)i)X7=^38~;#XG>tng(#?c2}%Kh)!O zEV1UHLRx;Fc%uWKt&*RPij?~@PdEPnhadA@!?YgGX)Q7~ZW*x2QAyqU!(O+w-tJ{{ zD~)9K?6PZA)Z~7Woz4C$(BSHL`LXWCyzZx><>~Y7$||XB(z%9MXdNLN;zRtD*ey2# zH93^jqsfCIM1K)hNDOc_+MTe+^a(S=#P6kbT$EM*j^AoWLhU2th0Sc9-D!qb`DWHX ziYeN2S$2~?3^+#0dj&i8?HllR-Z` zCM37<0QK8%FJR~0JUD&Ct6KgQgC9k6r#teP59F$(uvOD~@m0|rb!AqQ{8Ur}S~rTO z7Ka0eHSyyK4>VE_%*H(9b}{o!*3!ou?rhSY@a{;JuNn7*#BpeOXqqouZ&)Nmu|Z1U z52Q&LCz$~M00fCGZ}y{|k8g}8pPzUN7fUXc!= z^9ZUVLsz~bFCHy9Og<@Dkoex;9hIGA!Jufl)6|CY0ynguxi<~ew*m2wfhh%M$Ya4> zn))eQrby0D+F&WB#cYu7THo~$h^a&GR`0h`IfTVI$o*hKVA`j~StM>B5U7A2_K}A; zW0vW`SyCJvCVoenb@UZ&&(UWeWx%Fi1gsvsGB@7Rf_)X@qB0vK(gx=wvzoS-iNL^X z<&YkJODlvllp6@I^m8{8cWyopYI*fms+e0Dri6Ita9Era0m3fY<8Gjncxoho`4ro& zQ|kIE=1e>!Xe%z+9^Kma4~mf$#;!4eF(b@n4X!FJ)h!;ix+#Z?4hn%K0|EM4dZ>JW z)Doh;)ijMwX&zY$d|XHBv^Y3;f&mCBO$;RAm2A(0+Cpp4sx#`eIC!jULM6A(vdLReuE2r9LX~eeI9)E%e9EjgO<+tGbSb~@F zjd3B)A4b=7Z;1D*ZW*%rb$eu%Mn0Mn9Vr>kJQqQwr}U}YEUgWJpPObsis*HHM8Tt#BlhJ60|fm5X%c1$G6pJ+m;l0sv;SOxte2)md6=+3LR__ z#S1GbvKmMvV+Wst@~`XhUWCU>_j^3>F}XbzExU?suaCP1OfkiVO*sbzS8>k~$3bp& z40_F7L_NYsS^O++9*z+DrrY;V)O;F3__5P1e z(US)wQQ?)cy zMb61M*>Xs1qECt>lJIuUPm-~rw#f0`RD#y@Z;IvP;>(iiIJ4xKd8jn-K6i2R7( zHpzOTj|0}7fbHh29|n1aGll5`_@dk}Pt&xYhU*;HTzOyGwZ}hPZaY-BI(t<$${^xk znap-MhMuYwi?l;UZDlRajnyt=aM@VwIOE4Fpwq(zZMsT&hUuJ``{M_-RVPE4NSB+I9|e-OJ>&a#u@v zwNup?dy@S6xyZmg)`y6#=J9H*rF_m~C$18b-hQCK{v}t@P>83bwZu!C7;IpXC+ z@qMY={NF6)Gw-(gXZt~c^fqcpp4~2O4AMK^aI#z=&+PQ$^#a*;(Bi{89kGss8~MKr zE5dQ%u?%}H0ejsA-vqz??d5e^`;}+K(?;d%1a$Gk~E(8k;)b4Mn~!14F)v^K8Q>)9auV?#|d zITIKycgvES9kG`x(G#O}L)^y;FJe=Hp>Z*IY;69`LxhB0wXT7H^ zJ|t)OBBpgK_kwZmYxDe5vkHcRq0DgV9x#D|8Jx#EoBFxi?@P>~yJ&duMIj_%(C5bh z6w@062`qD>J#vWSe}pX0mi1_bt?r88t zSxn3kjk0pGclTo2@HuyEA2$-QJvUNq;Ii9J-~6ytkW(;?H1RjVgZL$aS0g3m9DB;v z<6=NaB`PEIjq)J|i!sE(9BmB)!EC!GI<6JKA9ky)cs(Lv{{Zu; zJ*8v2OSe}`#djHL;C!qSknJR7d^44Zk&k4BvN6nX=GP1mx({lPmXD39%r79l#E=Q{ z@K|f4(oHj8Da*@woaFq{Ekt)D?r|UbQ!XJb=jfgLD$*j@hxWN0;0l#YePVxDsJ2{z zJO2Rk%6fx#4fnxNU0ER$SF{+}D_Yk*&3_^Q&I;<&+HBEN!41OIM=6vE2rfcIdHWkHKvuMx&IJ;wob?hv%L)op8>-=>$^&patSq+C){$4N}cB4`a@18?G< zLnvi!pm!wPuhQHjE&vi#X&tHFNDo$AN#=k^1ocadYIXt#Az^XJbeT*PgbUqqAvpW93P5}8-!Cb zSF=+X;9Lk(R+~Kqs^wDMP*6`JdSSfN6}GziVjopWQ%u9Ld!i%{xlV_Vq=u_d(|~XW zF|ycMyGf2%JFo*>MmHG);C5Wsj~|B&_WDD zyg6G?WNT~f4R-i`nImVA`q)b29ucqNXPVo6T+-%D_c7oCN$>$KF3s<>c9T2s-{uLH!4>K_%so?5zdV~n$Qt1olft#QU;yhB`H^@=32&(DbsgEBc;H+4altm7Rl~ELzB_3%mK$JxpNhl2wM21iTl2MeC z%%CDtQG$|6AQX~;L{UU}qA1Z6K@^u1XPQCgiXw>eK=VmJ3Q6XIB7m}j2bxJlA5_$y z&`xDC{oLPj1I`n%KO3fzj%Y0@rPwJ=epOTwV1+S~d7zh+M5crCHYoLZga%Sc81PGt zhDUB%(T&{WfDoE78E_H;25{guC@CX4c*qJ)xz8a_xsj)V~M!v{hYUGQJ>K$AS+%*D$5Y zh9|YS%SScQ`v{b?i7R4|YA>e94aY(i$z7qIjt0$ij`tEumjmR+8BQYuk#YWtGhI+| zp{#3hzD`fvq(2eFO+`Ut0N*4ueAYtV9$lBu-Tubkc|_AsoE*yU9o_skR5jdL!WOrl z`te3&A67T_uK5hAny%Ar1QBeu-78(th|r ziSp>(o!@O|f~aM=++c~(MjM&t#@N{5QFf!moGo*=!371<3W}(t&LN4;%Sh#r79y91 z?Ql4~uot`SJ=(vDvxb^D+B>L#$AD5}qTQFlJaKd3%WXxz-^1>z?`Q7>^S?tz*w}MY zZk%1m7eHDrP}Kk3jLtR!2u9usIetGw^9DIUQ7ODt$j+P*NbPW90KD>6lO3+(rwMzGR&g>76S$gi& zIHsYcu}e)z-~70NKYmqByOxGT%)3h(I-P>AtG-%ki-YbohE{?@1cA;!0KDmRz1=QS z7uMV=CUNJNkUl|Zcvg|L^ai#@W=VMy&m976hbkf9ZGv-;l7)A$ zS61BJJWh+ck-!HNjRTO%(8n|CA)>~YaR?DQ-wW2ZfZ(~}?4>B5g=ff(;@OY<^ditxB zl0A%~VPvg#&3lh5j}EJ@K6lSCsGA+*_d0!%j#vxjKB4tKql2IAY)Ds@hRTi{ma%2c0HF7<_bM5 zCAyf$0noG%6A$qof9k-$O`jFymRzmwc)R#UPuw7}) z71o+bW5FMr0Pp5Hsva2Z(#>JJx~|VlEb}~RjLJiWILFl;PJCBy5Ahcj#@mjsrpw?Qgu;=wx+O9YryC!<=ayCI`H1yTfdE z;ZA04bz>xG&I1}oPBXW_{L}nBdY`@&7gj>bq6rNf$16uYAntH6ow=@;{{YCjce#YI z*uPA}TJG}lJEy8MJ**dL+Rtf$!Zr*a0lH#iR(+dwJOurD8e~6tUbedSW5QMSbJoRA zEgY3#i5&~}1oPOgE83?3T=+_%v$%?zbk8k{p`Jbt@GI8nzO~bJc}y|7%KTSHpQ7a9 zo4%Z}&|D-KmcC%a^<%>U`6k7>EQ2OwJ}%s=m%wszkRi#y3)s`BgB>hwBTEB)!Agdm zx!5U@(~N)`cT!-caNkN%4X|UPb~ravn+fnwG7;1S(jmtH@Crn5sV$5SiN=QbA{cBt zrFRNLag?yodWA$T=YB;x$VY;dBLt@z5C?n=BA8@D!93VgxrKu+a0A6gsJz-Ke?40Z zf7iMjNn{>L=0dZTTybEhIBf2d_9G|nRaM=kz}Di9jqY&d=im>yM6L5-@=Maj^KkjD zL08%x;%%B>g}0Rss2{TF%mLGY2**cwioG zB(8peL})8Gvhh&zO$t1bN?n-%- zbx$KrJf8h2!9g%TNhC4;)PwmTi;{rORQPCPbFgM~o>4|m-lj!sYri)|RVV)dh;*Xq zrr{SdOD#Jyw(U9O4pqrG1m}{?SR$;UCOKe=I(K@+YC40wMpfloYoAY=7B*7tyV7#j z`QUX@NrH+Q0IfIKE-aYNeJMtU& zES08>w%Bz}9UDuvxXJCv?{bytf_aScyf-}ZSBcc?pObH?=&`*o2W7r%Yh~gZwmNCw z1P_xy^{{gM9=_nM?_5s{sTkSX={v!sfz)^vi?!9p(U#0MI{}Oos);o)vUwVPY{cSx zdM=IO$Gz$Q02P(P+H}`DT&5%Fq9YJ3j{_P;Hydwz0+19TqzH8H7zYqbrjA8#5aAs{0i%Rh}Bh6aQh_ggQRs) zWRkVm4FS6@RUrk~&sEX8JbQ-=wajxF9925yFB_kCn&azow~JNJ^zJ-y<+tNX{{X^F zO+559bh3!&Eg|PSJb9Iq?5gc8zl-hR^EN{}d#Olv#JGP{=WlmdE;l$R%~K)~L>zK5 zao3K4eNQaM>Rs4=W7AE)ddDqk_@@sLL;waq zy>muhgPu=#21<5Ivq*2Q7lC?6`l;2rFXapX=S#EXR9gNcp)nU!(m^8!GdKi&>jn7G zeWhJ=qpH1CLmYK&WMliMYzgto*B?)gH_g$PQF5NP(bRE=4^!3|C^&M8Nk{jQ!qJbp zIa}HTE;jis_4b1;ETc@3_nQ#1(^@L1V_*?7U0j4xHO@Z-^Q|`_6P{?E0LM^I+_~=` z7rp-g)x|T*)WnBy?`M}LkjrDJq&*uc1boygp3691XA@6zxWS%>Lx=E3Q5&Lvbx|9b z;mxlYDcUOficB;)(m~ayay~1XXU&J4bFQX+H%*`OVL$u*nSLwm#yeHATZJ9&I+jI^ zalRtgVD-wzKGhUM*O~4<4Q?^>U4Mz|>FPLUxr~{G?u6+X+?C5>tHT-`a7M%sy?%$2 zlRQ7neBaeLIJmiI+M0^o(7<$2D?1*vZ#GE zD@eIkj;;{=9ly zb?kFW1 zT99^V{s@efQ0H>_m(Jsv$**@|7R7TTNJb=lI zj&X+b`_V~7D_!b8UO5cq5iHsxe(_Xo&|n|nnEX>Aevv#`@FxW@=2S2qoAfv+>8mDy zxV53g^Dl0tHogK|eN1N^-oG~PqdqAZaP~Q*c!fQeX3+lyHEhf5lQ=Lmxz0EM9m!0%@XPt3 zk0x07k4EqnqJq1HqP;;(A$08~OuHmFY*sIfDcMauL@WR`?r8vYC2e>b;T^k#DD6T< zmq!*{N&3em{{R(<<2r#YL+2PE$_Cy6e5M=w_Iy3tf6AYfI_q7wwl;cgEK-)>7&j!9(B#VJ zcJ1Sg#%VS^fp1JIi{BD5kUE}_Yp4TBke!+C^HkS+{IyLm(&i|$?E~vLe??9Ymdlfi zr}@dpmoMPCC6aj^EYiA1xa5~}NhLO4=8IpN?a1NE(Y)lOYS{-lH1`JcI9k$Uc2RdE z)kEWoGbE1K&w17CXVsA#%vA(dQ<$jXp`y1vFmO%!A>K zuTKnS6H({ooyzXI^m5Zo-BC)%gW%wu5rxR&o}If3i$ER?ApOcz7l~<}9R=ao=pT|6 z_Rif*#gVc}hsj`I<65Y_x{7&#{ZbW`Cmk{0oTg)ki(Z>+ZBwbvIV&C#r*GvIGqHlq zyFUW6lzdrhnZHO(NA>Xk+-_HXEQu7d|lCZxN*0+)ynEd ziZDeKJ|(6@@i|fK(Y8yY<}4?>2j*3qz;0WCrH#@DxG1*IQjOAbbgn1h7dADEHs)Qk zORigkNjYqAMm{Ghzl7K_Hslh!Se(gOT0Y%eEcrD%!VUFYNaPpR^myxsAL#l1qVJa^ zPBa?jEIaHLGO`h*?aQLq@msZ5xE~dpsAhBKx!g5LG;CND+xWU=GsLYiG6X=phFj|`=?(g$q(`JK|uEktJpt{!E%A988I!($$N z+p=jEbT`|*ypH1GR}kr@R{%a7mX_DrKN3=w_tkByV}7i$js75IVZkjwS--S06R{9? z3@6-E$m;g)*I^_wy3ksDV3$90u(T9Crf{`EYb2?trjYXz!f-w#Wx!;{Pc)uiUCm98 zRNL-!74ZovD{1Fx+-Z=11v6JuoG&IvH1Rtz$-js=KXSOW@qZCog z+U@D~O!geL#iz+kg6Ufs8fqjX!N)7lw*J@ekziHQQB;A<=`25T(c8OXw#xRly3y4*p4{`>A#+=A9Pp)ZZ+Ccsjz3mRw;wfoO<7G+ z+|yLUF#iC{cz5tkn_@d#qpzytU&b`lj(r1|#behMkJZFA?cDzWWXSk9CO*=u`)j0p9Af#L_J&Rg z0~>(3)o3I&<0YeFS558GI;L@Ol31Q1EIrGd-JCp^Ccm1Ua7#|f9L5~Wp_v+dPC6{g z)$&u$;VN^89(snUv{oz4bt|grZPgOf#zwgs(Se>_P`}&isAzFhNi+|VAo%sO8pd&- zcFR{$6_$VcTDRk1%qJ*F49FU7rBsI-%H^cl zpm0*dv9f3sW0D8j{nKL1v`(Uz({e=w(s~^aP7A#B=aPzP8+qp=U&*uPK=D4 zY_aaMz1E9>EQV=67*w=qXRY)FpjSn@4V=@Ii_IKk64Dg=gnIle3p6qElnlg@r0AfG08pL-m`aEw)`xT`&T!QYH8w4Nlys))&^pw=?#5B-Edgx#Y!U;M4YQ)rx zWsb#geE$H8rm2CF<23#8E!E|#l4AQkWs+CXSBYuoVArXGQ%||WVzyNhDRGz9O4~yHdE>9?f)f)j(bC2?Jk`a_yWe+f|*h zGD?`LX<04<`^YlW{{U6vBOzkrziFqS zxSrEw!sXU85X zyGI+{t*jCEOnX`}vBt|;L)sS(UYi|17g8Et>7+$i96pc4d{(8uD9PO^#kP&KvV9|- zJwoT|(g#paFM_8Wlb2hBRo6MDf!e8y$b|lHbOfb9;&rT+kosiOcN zYz~EQlQnel)j=c@jJ4whN>>ay2Yzc++4cAFeP^LmIo^#>;0J>p;;T5K()(S(?9zL( zeNt6GI3SK;XVQ9ie2iceMP@h-`}od)QPezoO*jK{?Kh$Ktv9`DUA3_jx+BR0%@I@) zNOD<8Jk<%xB2afUOCa)yqRJ!$QcpCZl7NaL5tNr5kORqD@O6f#iTIX=!$&kQ)K*E| zvGS9oZJeAc`Z{UrDx#&PiStb>gCk=GT<0Bt0ed^J?#M0N6Gu;Utz#e=Njk(8AUbK=0*k(ApzczSCk z;w<-;s`mTopZye9P*)9&4Pf*1cmDuI>?yl>vEAtAptVre(_Ly9W1=(0*EGy_XgKpJ zi+5?)`!#JueW!5q?37YK$f*Nj4s_CkhqyR+j2ta&+sE#voO@7A-+WFV$u{;mnby+W zbs&%8zJqWV5%7l?TV<-X!A|&~m%4X<6GMr{bHBY_R9N_Wt%Z}>ty8r0eKQa8Q){=) zk%P#C?kPC%UqCLMl5hprz4X2wqr`_Gz|QQncxZWu@uxDyHMh{cw$H zJ4fJ)w*&DTtrr%FHG=wSS;}bhfZ#dr;Hm0tms(OqWgF?-BJYD`I!R`E;RGhTbo_40 zWwAtoZe|i*FUnVz4ptQ_;9WBrew_OTpdyas=`Tvr37+maZyjVJ(e=Q-(df9%s!@ znvBZ^xa99fol^CYW(bZf73m8Tm*v~gR!)@?X-WlSV_%Dl% zWuA(fx$wG5cQ&d$ZIOiIAPvFC^H6f?&{=)9%I!xl$Gdel>rGQVJZz_7F2Msh9GrYt zS)%NsuZ;Mc)16;KLjiN@9TOsu2*Zv;TYCwPsU<9)JMr0M$};aOtW zM_FH^mRqME!71I0*-O1!rNxK#8dS4qbBI>nxX!o#0J}kLVfvKNho763bM3QM5%|XE zaIb!OKB2%9`ykY?CsjwbN9(LFzt60iB$@@>!bx z*Q`*IC~ft%P8^HCV^8y0?%(@4cLLwzsG$Z+J#(DucQ|hwfyrU_e+A4#o%XOj%3l=Q zr(HW&zEw1e3My(H_I7*2_lzv1+l%;`w2@bQs>j)g=pX)7pN}r)B!8B05otP)svGyL z(7K9;&hq)BmS-Mrdy9$rru7_b)`VfNK0gHOXykYrTI1t@fim*(j6m-4Qav1`_UQL))xl@;0IL}^HkbPbEcDL5wOpXMOOHA55d~y z5t{u(=K|>iTmkyn+Y7HT#{+JB!t!@z_qwY!{u;Oek@Cr<@i&=i>+oKcwB0mD#@EDX z+VPm5DaU!&9`*5`TIsUKp6B^{eHSdtj!(lt!_%TzJEmCd?hbb5IVH8(=Da6*Q>T0k zr=gQXt(0UrguPW%TW|QL9jricEfTCiad!w-+#QM*id&H2E=7xLahIUQA-G#{m*DPh zE&UDunfXrUk{!{7}$NUJ_JM={uC&n-n1KmT1Si1=FvL&y^W5NtKYO?u^t-9W z1&GL#z@@4tW@hJu7U%wjuRFVHUf~p_l$gGSd)fMNtjNdGaMd#af)~PyhF50vk4P`q z!YQMpYYozA@1N^JB9t=LZ0Uk5Z+yj3%5coqSfNJ-=%pJ+Ra#EckR7$K$@=BnE^u&n z%z88a2k}apa5sCP_%{J-Ug4y3DNayR(}d0f#SfP%>7U0(hfZu7lj51@gz0Jh^2Q}o zG|T;dgku4;MVs6_Y`q682&z9-86ND1ZZrh7GNF1DN=F?ZP@~wKjf=yd99c&E8%!`iw@cydO z7}^~Dbsfdy`fW3DeL6|h|MrW<1Gv4-Wmm}mv`GFM+YOiG>6~eQ;V_c2_|?O`=13U$ z!lvlf9PM8e!WtKAiV@FoZSf|kA%6ScYaM^h&Xp>+m_J`)BeDWPHv*GND|QqO{2FI< z({Y**oUTo#{byNm$26`rn7=wz1RvCFe!k+Lqmvw5Q4So`cbfX}s&ogvP9D%%cl$_R zr&U*WUpz_$fUP{&4M3K7zTaG zW5iJ>-*>AJ$u!$mtRsvCVtioO*&%M9OyQ0HCpZzR)X+rd(i};S!z>dEAx{BO=ZARA%0{zvP0N znV+WgzZyay{-x6Gm@c1PWr2|7rVJ_8(_I`+>?NoigjDaN9CWN8timiYF$Q zoyd1PHM(`CH!-moW>ezs3{J}<@jG&fIZVllIvi_P(FX7j0GCfD=9~u0!^;g z+zQ(Ox{ph52G6UZ9pTo!s`fnE_-fj>#z3ogL375^i+t7{UidR#H1E0`4M}Y6b5OZ! zIAjK^%BoXm#9z<$Y5yZe(-6Kfmc>l+14S5sZ#aQrJ;xGs@$k*%F@`?rkACX#!k6wy z_F3S#?lDm4Ly-JaIb>;3^=w>&wxfyG5_$6BhyEqg=HR)__~A3oNdxDQAD=g*>*##R z=Z|0Tv^GefQelKeUuj4?xsOzzFtNiOyt`5V?fs_~4&O-}J!D5{alDlzdB@1u?%wg$ z-Xm~b^eYF>%_dgCj-sA8Yfiki>}5h;QJdTsIyYsPnJES;`Vjwt&=Ve{9#aOUOC5DZ zx=3E2xnFizzYmCSd$s?>`|@55YeFiI7M-2dhLp{TyNTB4Bo6^Hn2(O-}?4O>=ou{aBS%fDQba$9-x4py#WiY+H zae;2+*+MX0qX*!sn~miN4TmW{o{`Kh0&f| zXec(FL2I0=|9orvNZ8?7NbK?B7uz9zRjwLl@e*FFkrg2ddFK%k9Fzm^C*S2=HkEMv zi?Xe-FTUh(DJBBqbPIpkduHS%fngTgV{>lRf8KjDJAP?-A&!VO`|_*dzaL<#_qSM? zQl8Y$>|#nQBK00#F2kU9hNcJnFBAqNNotj0i&=Hi6>F zM+97v465rr;=){AzF2!HZ+6=!r3D$?f0P-CXy(xj+)_C|hQB5qU+K@^oQglV-s@Y= zUOVADnq3;XK-!fbCM2bh!Q>r7z=J%1cWPc{J^Lm1{K&zfDRzPmYigCPyYv)$OzSb$ z$)-;f2Y!#}mvULLxj5Ai3)gOc{k%?rUvcK$vpd>M&d!x#>tRNkGVi`DQW+(-ccgOA z#8`eLi(Xxgq5$G#6DO8b?}4r|9Hnj!er5xOQI<3BDOg+12K9aE#*HLI$|qeI7+iCp z+(Ii5d89&1(*M&2{lBHq|Mw_uZQKJ=(WT(ZJ6cM~BG(M(zQ?XiTuU3yEfzqLlSCex z|GaybkQQT_mVI~-`GRLFu1K*R97cq1qiBIAXgy1Uu!jyOKh70RZ|aq`Jql1*6e^HV zwTL~vQR12Beb+Wg?WYxD+#(zi+V;_1RW*MIhm~5%ZqIgk>u-06kML6Wf)~dl;Ycq2 zTj8N>-8qKLI-6er(E~vE;jfTm^r^}8k>PqDT$4iZiEYgosLV}JP(|)Ef`SxrL~7e0 zM_?FLf^g{?HJBP#TsKj&>=q=uNwE|^k#o2|e|Ilf_s7V$YejCDXk**hn75#-j@5Qe zFZ_f8<5O{kqqYn0k;iJwaAsZ0z_-HKP@1<5tdeQ7>|woJT}jYo;)LgOTNzb^R($l0 zaqZK$eccoFzAKA>L-H-(`SFg=b`0CgGhM~eg63s5(TqQTyJC^N$GZ5{Dx49w|J)~> zhq8TQnY%mx&&XZJa&}&rzx9L`;fJ|(u_FaZJoxJ3_5kYT^MiIa7yisco1Xcr%@{?@ zG$>+x{?2ZRtt}v1Rn+{TOmC@QiWur|y~}C}dOc-(lR4J}SbJ>hgqlugWlW{+-w3u1 zLlt`Tu!Z8&Ad`Cz4xVzAg~<4t@+|2NN0vR<3ds3de2{)MjFOD)Y9x`|fPIV2qNRp5 zQ9{Lx)vE&;5t4uUv^1JJUztv%3K4$C*VI5;zsi+ias~Olzr1o=?OcKR<*5GsCgYyD z4eC5Bu7@1#8;h3i5?_|OP|~`)K8Ck&p^7}16m=aNF>AN&3seXC!uyZ-uFP4+>|@%q|d^8Z!Jg zOLY-nRen633j9k1y1av0?~=n({|{RmKEZdQ7jq#06K&5sb^0aOxRYIH?a1o+I&Ll* zzU762zm1!e(5y_HgpTY!mM~Yx;5bj-zF})Jgalm^*y!Rwf7w*YzA7qcRlT5STj*S| zr%w_L+5@aTmv@MQ{ZU}d-kDMdmDnz#2Jc*RErE?jd3`sav zmD9=Se6ky@ZWXXqU2=v%&e!8d^H~D$oF3#vD0EOM6gG^Z+>na;g8u*kh;z=BZmZcl zLbc6XnY+mkep76l^xf^E116mo-|t*Yaw33lfRozNvT8SXY%g{Olp^M~1?_rXV(o_Y z^R(rPm(ue~?fWA9O4x7W{qG`_D4M#LG$YJB8i|c*t%49#8vOT#?(?W!%-I1xoN65o zI_?X1z+~d6@!AH&4$}lQ==NCw9W~RH) z6`XvwWQGbk84beQmLJb&^@J1RFt`^|Ila9PH(z%yZHTbRQ`^LU$e~~J?ad~-cOF8}Qa3OxPEr3)GF;v* zw1OwGJJW(+ofBy|DII8dquPwGL9|~_>4J161@k^6rQEk#4J6zuBRwjI&Tx`R2>TJX zyX#B_Z{KN7APqCf?#w;1AYdSVF{MBq<*hnjVvxI zs4Xe8H)Isu^G43|q)+t|&Y62Z)=OaMR%S2vQEk17*8Gj1C9Nu4yk|x;X`q?CH@696 z5OEt{Pf<00Kpk{5>4nEJR#5VS|It8XskD95WzPstc22^v8#VoxmHkWf190!%@xLA& zZ(J9mMVexRqKt+z+wS819LwD_rXu&}?$;jF;+~8RuVO9KU^7Y-jOIJwjAEP708iw+ zHdnIDM8(BKaw^*9nB+6_k*X<2ej_eP9k)5Wwt{h7zO`(?yVIR7w$R_8)0F|xAA9MW~icr<}Y#&;C2=A{X6GjjNQJRtzyYjfiUIyhQ9o0{EDarufA|O zFnY_iJXjMj32JVu4<$%rf?bv*r>@n>N4#sKIRu&Xzmbnv#EbsZ-b!y$Ada3KJ!@P;-^w>67 zO!SoMBnR>3(n2L~PaU(f61|FygrVwatFj)FWc&J9D1=Rr)_1bpep3={6dPT>seY;M z*yFh#Hx!6n=kIRpPEn9KLR4pL*UqS|wkV@c4|ZTi0Kf0tw{vXDR^c_X57hU4YO{6Jd;siRF3qJDH$v8p&jSs&yxXK!UZLcV+WBl*^GxW=v zL&6|ufh+PHJPwB{@jLEyZf_niokLv zQA69Rzw-3QIWSiV5N8EBAl$6wl-9*Kv;X)VkHaUUhr|qpWAsqpi++%4bg9L5!J*JJ zX~d8*-%cXW2KQ%t=_j?9^})+S3C#%+5K*f~rwSM+YTJXea`hsan@b?|Te73>ndU08 z$Sb`~(7m$B=%TnsiRAYJO!|^4gC1$dY!_q@UdM4Yol!eeKLXf?c>OYrHghp5qVMO3 z!|~_yzX25MfYSxVG_J+W=kum(ot`tHi1f&l52np}Qm*;o^?4rhLjK9~>`Wuo0*EHD zunx^B8p643U{2;mn1ZITvjS))w#|Gi%9R&|u7ofNmu7udQLC*uH9_7IF$%$T1)paE zANgKd<6e3 zN$(HL_oh{DE5p4sJ(&5)KVd!^YG;`+N&8zOh%4Qzs%2YkXrU=(@gL@o&fvK8&Z_Zd z7gU6@8Xj1ZV@%D(&wulO>~_FR(2R?k%r$G@>{TCTNPcIx?P11L$bs8m8O38XJhWwT z!GnuK-_f`HKm%Xj&Z?rOY|Eny<{{Fc$*eOLuXbJ1BTNpbn10{`5r3U;|GM3(2)xE0 zcVx^i4TMB8zcrVb+IXJTj5GGkrfgBCpoEY(#6v6^pl<}L4W&>$SeO~vIcmWY5`m_R zkvA=>F0;89p$N|2yyTNyy^1wBi`973>N;fB@K%O*@JH9!$wtbOtKO1`%FFQ{+%bzmQ;dQ=G<&%C7H9kGYhMMQ_Ki!^>pW5i5(AumI8{;VkZK+O z(cgUso)qYc_wD_s1pGH;H#E^YaShh?bS7G8n}LXvXwIo7;y41mdQYyK1~)XNmr&UA zT1;k4O&$C+kTQkPnaK1$f$;onM_`f{TnT~EX6sx7SYp*Wpisd*?UgbW*hAhx3;248 z^sEU2@~IzHTU_oJcvm{2(X4RT$mtP>n-q^Q zK2lx7%iq+&$R%WB8QdPMV5KHcR} zR$$bvUN-G&PM5)>)F6<`v`q{2;X}Xta&bH*Qlb%YrcXJXIxWrWW$$7(onp$UAjqe! zcZyS!c2PHu@vO(lth`{`23nzdEk9zggc+BIQoA+z;cWKOHSX4N`{mt|RtG4(KZ@ic zg4Bc;+PT=;IJ>CYw~mWb^$zqwerr;6$Aq$Z-iRKtzWVAYPBxY@KBq}AHtGsPGo7km zWtGBlz+*6A=L;W6^=zgY*B-hR*JZC!Z8+4P-A|fI0iJ&Kl=cS*_^Jys@0G z>2g9&@W)rx%_SPHcH@887n3}PeO*ye2wBm699le0!8W=guCF*j<`8ZOk`pszo+y1X z!Tt!@tL#!DM*yewt9&X&TFU*0Aw9jt?e5LY{FyE|-pUh@Bzy)nc7uU*`QvD4kAEJM z3{y?oz7+L?>RW|i*@CY9z|MVlvf;RWwmdAhKC-c*(_naL1LU#XlR4l5uya9#ghHiR~-d`BGNi+}gY= z%IknK%TJBt1rpFQZC_*l`J$-UJOz?kvX^VPWNwIJE)5U9)|wG3AHz_v55lBm7$rLB z?n?^E-0e?`tU%5*{IUhasS-7@|LrM8nVI(OD4RZPqknX`=ebtvAMQO~@3&^sU3?^M z*Y%rwJ*+kb6`{bw9(~dnEV{!*;8=zLEvbIHTvg$?!Gmb~hXQAyc z&3U2(Ri&cjkJR}ccxvFpU27_BKP@9BH(V)tQKiPdUTCejo~Y&goi3a7|cXa5s( zh;P=HqvyQ29_Gn4nDv}IX5)68%|q@?HcDQbmT@hgk6FYYUW+5T?XKI@>@sLB=A)Jv z=yvwZ(zaDEN=3+@Z3HI2Clnpv6t*E|=6tiEY;}FJLD76lKm2}etDZnhPm7>3=R~S_ zPZlR4!w{pz_OTOp1$@V=sl}DA_4^^of~1G;@LT;7(^@6pHLN-rnE6w-R76xQ#Qj$k z(<|w$q;nDNy#3_)qe$jWo8Z;nczfC#?WOzAT6ee}=~~DvHv0~lv-Ag_Z~@03^`K_m zl+L1SWhd&Tfq(i-Bx6J4l2RBLR7NKxSE|NhvT#%##j#Tr-jd&LQ!en~)p@@$mphPA zH`SdoEaum<(XkMk#o0v9z>BaXCfHI_j^j{D&y18D-D*%_kVbg4Ou@xYEVB$IDl)2(3 zAHOHU-hRIi>9t|@(t!wV4S{Tr4Lcvg%7b#2|H)3jKh&Ob6}quoIggFBK#7DA^5a|^ z)u#p1EA6`@` zvlmqN`MBa6N2#yTW|dsL4r4v&1VW+;X_p=Rl$j(*@~Ab(Pb_%qjU>+iW%8>ZN$ z!{_yVAi-!6OJ3ZMSpR7p_}Cx%)P?o%C|sPR`VsOC%fp-ZjJqBrqqho_yl9e=KB|j5 z;d&zqU@0lc87mmdkBu>xpD^u_IwZCd=S7EUxtM=q#U6Px>kPx=3|cQ=+FTcINrVH0 zJLZH9u(^KYO^`bHK!a%gD2;~r_5aPZGj78+rcS(mgdH;?L`$x|AdFD*%l-GZ!qPGsV zvs+I&ICAEcm$pP}Y5SQXU>p_0eu_sFbJ!bmfl2hyx(s zp+6nj)00}T({RkLqRml?Nun-T3Gj+3Cy$Gu@MBCfOFDamr)A?B!#spZpTLr>EFq3hb;N7t1EG?{hWbbGW;%dSUkWh#CO@@E|OEca>oZyxRr4ch(g!> z04o~w_fGb}C);bP)uGfkMMT@meE=<^n4Q1i^o`%6WU59B$q&jiMkMR%gqN`02gC20 zbGjdpIBCUgN3S)F_&qkz+HYo!=0aoeg-ZL#!C9p2R-kBFCDpkIZC)anbPf)Mb)g`$V58Fv%W0BHUK9i?9h-a23=G8kF zHTxKAxeyHrk$q)M!S;DWLh;SAHm|T$5w?)3Mlp<<1EwnHSmGNkYF;g0$VKTiZY;3o zwnsQbYR~(WXyGriM%itaaL@8_?U{CyC=zF!&!rl@pEgPX&n(O%^5sG{z3YT~SXjWD zj!B$=viDYX`>PV{2cH+6tEIJ5dCNMX=KJ=mhVez>Hw`zYfX}H!7P&DW@8(p1 z;H4M~qTUI1i0YKu=)>(=Q2@WW1(ril2)};U-1%lIh+abp_0@uUT-v32+R3Bh zdh1qBeG{P?tT`jf@}PYj?tx+|3G|3b=nYPX5c) z?eZ}7K1V<+LX3Rys?%C>boS<9|MbRx-d%V0DA%gQKB5-7j6-=n!K3?ahCCnq(QD-K zUhAahl2EObpiA>$yNE262Kp|q-C4NlWQw|NKNk*cIVuB&Y=r5xnhkg{;bi3dv`t0F zsjBK&$Ww>pQ-`1-qEQFE9hi9Ev3ct0{@+dY|7IHge=kW68ja8$h?ZP9daMpKdU!7j9lV_->6}#i;_m&hX297*P;m)g#e~UaKPH zir1t~_f9Q?>7W&IMVKa@QUQo5H!l9cO6!JcMet!cIF#pOe?|Ot1IwDcZHxKb3;AQX z@-5?HthIzSdc^g5mMeX$V3s7W8dc35ZO#M}3#1`HH&31@G}^x~~s z_HFY`wIt%HD8m%t$%1bYR<@33avIY%1H9<$xwJfSfkK-(o>5muU%ZBgn{RI2vJ6eg zHf2j!T6>*A9U+x-Wni1-xg`Yhj?^KW84vU36KeK`1I3Lpbgymu(FuAqxm6;(5eMR|~*RdV&4n>%nGXFal9 zi1D5HPy4zH-D*pAXH)y;D;f}T-4F_BKi?H$D?SZ(7CKZY+Z0wy)6uLtG#=Dz&hTe` zl){<8Cq(qO#yZPnKz8fDgct`hB4hK_rOrueUd-gJ zT4Tygeol((q`GWR^%EuO#1Nzt95>P7y!BeO`1MIOxKP8{&=Ss03#!S{M}|cQS9!Vc zvef!t0y5uQSv`s8Uc^bcxkb$shea3?1!)}1$>CRX*s@!x59_){ATfK<$7k0X-dypa zRb%49cdm+4qCM?2txuH~EM|!$hazvtv*r(F85YuW=}VN=GH}p3cKgzUqdZt&z>lvYoiVX zn$8bq(KNw9J{qZAzw<$U-uY*$g=VC-BS*5w5z|5Zq(-^X*O0>h043GA9FWvbap2dY zo-~^Zu9q*p40}JC4-^Sda|3eG#JZkffEH4NjgYb{LduKa%Ip0+wopmCXrOm()k?0vQ9b%A6zXPz6`X$j- z>Z#Sq^;Q^rBS8r?aafxcSMTQ1oe)S673dq>*WG%+=sN@%B>~Dcp~$Z z`!gcQZ#lY}J+GOGzlaCdAbze&vpBoA>_JTcuoEMd(D-z9So)B7KVQ)4Qqx#tr5nxJ z_V{-dYeHL{y?w^7^}nyD!A2Vl(?-)vyocY~o0g)_fch(rJ7DS%!OQzhtUM`f-iCqL ztS;{wB$g-_QdCq@6ovBuOqH|Nn7uIOnIY^3W5(8dow0UaS)%#o+!LKA;qZIOodHa1SPQwcYa+Nk-gSM|ID#oQ-nJKs-_V zFn8=mW8K3aY<9?V|9T<*o0)a*)==`#G;kLERxbN2q&?bV&OEO`5Wy_hLyW}WAGeER zQ_NHo_1d5X)fts*aN-18?y|h?NjmWjiVhpp@gU1qRUiq4&(Gen+A>Zfww&232D2`(c&@*4)2bgx`+@@CDlwO&AA_N^Kv9M9)ss6|B2nu_0! zWv;-I^p%}^{Q?K-?UCf%<0H_kq&fcqe@8)x?pC*h$02F=D>>QvFJN+*u^K^HYF&e; z;1)Zy?~NbkczzE^lkAQp(UjGlPpHq#Nswvy!gYD8mLh>^iN~T6lZ{w7^+`2xl)X^0 z3@JmSI~oONE4E0ZN5L*#f%u*A?|fI1egn=t8wDGsZ%8&PjMN+QJ#d6OS!_i1F#6#_ zBuFp~c6r?b;~UhtB@@YQ*YGX+oEGVT_x(*KmNkldEtm2+guT2zT%8R$MT{|}7~xtP zTO*%W{(oo`xY9H9_y(*WlS0I7dL(iq39G0-<9^Q%k|e|(k))&Mjx6d!KtQ%JLq?+o z6mJW9(`_Sa5WZPg@Z?2kQ3#$3(FncYGS6sC*rTe6^|HA9fl1#A(MqGv5@Ts5sP|lN zHRa(nyOt=HT`Njb$S(E#oxka#fbXiH9Av?^0ETP<^3n~Va_J8OQdb^Vve?>DKPe{- z!#;Rq;|9Met)aAJy4ImP!lET`E)Sdam7eI&DTO(HnNX?1QG({qpLLy zYj9A?q~av|O3B&V2l<=V&o96TX1gH%Bn?1+ zpnmsN6UGmNQ_K9C=nABWA;{&8vB%6;tf5UTZStmFmwCzpeO>LwHyi`K4*f9->y5l~ zA0j2XX|`JP_(W*RSDM3k?E&KHAXVd6&Sg!FOoWL5kqM zzQ9fSd@W#g2wYsPIoKiJG|c7o&mfiyVks)=tf!TD8D1c+tG2HTd0*Fk?T<+`JF8}% z`=M#S4Zp=ab$Bxd26owZ*+^@+CPXILRD2-V@>Hb8k<#=>wCTfgL%(e;{Q|arFsHGze6lupm@)LqoS+xMpigKZK{;QC=Ea}4-YZSL z=#uT6N*Vc;}#PFw! zpMK%;9A*r-6Q>Y4=|dd^6 zcYGeOB&$|H6)&wCZF5=+opCsa|{)EfW zv5E%1&=H=!)hD2^BRsS&l4Vb&g3K=k$aa8h=T!($mG1q8z!S=TysTtJj5tHFZ_hWoMkUC|1_X|!t!Eyr_{ z`%12s@O=ujEJ@no%zY586>RfNAb_PB@q?zP_8Hc)V3NXGVozUH>&mWQoO7<%=%lmU zoyu6fh=qUUS3W1lI&rvROzIoo^ebCme$_6cMM)?akm)*o{U$uj8@i&dY%fu>~cjmy+XH}@7nXHiNkfgWp1VU}?^uQ+36>I+#(D(s4j#Gg4`m9Ps$me`s# z?a3GD{M}8>`vdTs6H9j?R!ToR>+>fnp2~AW>G#@bM@WzJa-}J zh~2tgO@^C!lo8493tTlajreet0+D^`cMInms%BBMB?0-V3dNJkxY!$4YqkgB4_W+1 zh27JeK3*HR0@kvjB{PtcTaQT`)Dg=V;8O$hMjTmz-i=ZG4rZZoV$3AJJ_&2Jue68X zbKm%n@z!$7ALfGTQa)=Id7C#O!4%QxwU%_|tY=OvBPnBDkoI!YPMuRTzv_I#0Rb7Q zEzVL)HZ7oQ98DKp!BgyX=>&QBDWfFsu2Fr*XbHl(4L}WV)4pD7>_<84d@~*VZ~kI2 zKS^)(7R8wPAGmYvJf>9VelaZCm4z>f1ij#cHvR+b5&o<;ps3L|jlroG_%am_Jg9fV zycv1|)9%`1YQ)PF2l}bW*Zoa)2db#);k|i5$lKn%qew-FTW0J@z4VNva&;_Ctac8M zOp>RWWT_U$rUD1)#PUan5kC3J2G~Wp+S$$sY$8M%@7p#%?g=o}2E&7Wgs$5C2u<%`R z6xW@;KQWWaWfiL{mrdN-&BCpifHTVVT`~J_zN6S>fF(*FekzpVox9L>lkI;c^S+|d z=UV+W6_t@>U%tYJS#sJtI+k)4ugt?mpV~V)_{JU6Z;I&|lgV!-aT9lP?WL7xaqJ-R-0ZubGFxBKniip*O_=mh_zd8$|P zd@pBLkldcRbu*mSj#7s;Df$OFmwbmTI!isxThc{Xu9e+!E!MUYW;i+?YUVsOz#c#A zLwV7H{U1PMeDU@)P110Ui)WCAjQiImBng=5ev-?O$IUt}S~Aky_z$;-2*qz0^m~G;H(QL>Ns9aqnnRVi&%rvqa~<3+I3LP;yA^0v zPEB>!o@h9J%3KZNaE3iPeqwCmsu7!cXlMuut!6KV;B07T zrIYA$+)+jh0C`zV$&WcuQ3qUEj)W;I%a6|cCeH#Zf1mjHd3Q_^?Qg#m7rdHI8Tqo~ zCfJU&-bwCoc3OWyeHT%mbnM_uTE*2wV!JsIf;#w3MEPu^zkjjJs7i07LYn=>$9kT( zgEt#-9rDQG6mil4TJcb@eBGVgJ8FCTuvwYboXaKz`pc6-!HVZVi|uMv=%`8)DkY~{ z!ZqQe`sQb&^Q0d76C#m1@Nx->2-h+zPRXA}%|w~fpesTdtOjOckABAsP5~4uisU0Y z;I3fi5xybF@c++h~PRti#f@Brmcljjo}JfLdU3k&|}I9D?;9@m*ey9)*ANH z{C5dC@;*&x4dlxA{cB^3wsA8pV+6y(cKdUM)MJe<3LSy`-jq&7Tj+@+jJ zdL|+%{fTcw_W7Dy>8MYJo(~i2_imToxzEahY6_McLkdTkSV;&d*eN890~#J|nC0EUX)D*%-9G>BeYWta+vOYok`0tg?D{NAZIkRTGXT6X z&io_IY{tX;LB6;3!VA(wO$!-|ex$*auiKisq-Hw(-{!o4ujEwt+ zefys~_W$oeB`x2F@$X$z0C(qbbw)T7);kh-93SaU8Jfy&90 zwtIVk@=->4$`niP8%ce=BOpMned3Q?S#N|D`Ex}sUHOIM4cNVV0d;fY&8j}-siqch z@cBLP^M3#Zn3O!cU%*+nLkKNx@7Y<228`P2$@OM+hyY`}Ji9GQFAidS1dAFWgb9O! zX67#{k6*3h>6vXiRWQZwcdhr2cDJHO$v14Rn_}Txg*I#cbK6vbO>xPrfLWr-E6Y|o!4?W zLHnkME}=`lf88fBsVa*zqfoak4*Nuiz7ad0!#pC~WeR@adSjyWhbc!!RqdchY%krg zfYt%&LNC)Gpj=8yZ2==^XbQI6f7VLnCGY+#9_h+}7SlRQQyI?<1&fE^%@Rx8>fI#A zU(ls6b;={&Yu`yG&Z}DDv_Ox=Rc~ksK1-*R*ycS^EZq{C!W8*nJ9T9!poz%2m3V}= zix4;5Nuca}7nyY?H1S7@I)62$6`k&$@(*UQl0E=fMk=)x8 z=W!1=@w5Z?=*P0E^Xy#wLG@mP^lk^+;dH{!v|&{KW~=rALKT)gHH=P}_eX{yp0J~U zF;l`fAGj6dXZCikZ)%SUtOn6oO@kKgdhG5ia?{M!eZ4?ydVM0SW7I@ovG$oe;8O0m z{uN%kPn#JnHu-ode#Q&;>~S~~CAoumbWsu$i->BoI=8|a5h16n8&ZTBEWV%qtKrU! zML2Et?=N$Q=@zf!X4Ylq6)}DMv1~#f?>ujfH`vS0%+q5p`)l)X;0;qH?U_qlv(|vK zHcvaPdf^uY+Pgt=%fyW>)Ei^&ZOxJVX@NDD1y@y2am5Zx$HO?Sp`ktza!}uJtVHrb z*{s)vvGl*guPMXf$KgL$o+6@|Irs#mFzB(&KSpz_0+576+zc{kQVibYYe>+ZdwR$( z#Y4B04}bcx6}O5ghUjsCKi5#Mqt`$C(_cg8-;?JK>&N!fPLo_f z_YZ4DLp%hXNuDUW|AaY6GLx84GaGiHYqUl2C^d^l2C`3mtGSN$iX zbSi?_2+dkA?q}=zr405C-nC8&KU|;{Ps305W;3&ba_VaP21MH6Bjz{NcbrK-ZVdzrrZ;s<=HNuR>!^Pt)T%+ zVDa*WPQkft8Gx_s)Phop%JL0rTfc;;x)nntrcv7&%fDtXuEiW5o~NQTz^Bus>MKv5 zvR&pW{{9VE_4A9>F`0paa?8exuBVM{BOLqT(pZ01y?_WoHnZhLWD~zU zp@9OFi!`YS1$cYUMVk&?tR;u5k+PFVO9VyKwIPfg>twg@+PM3J3<`U?9Km>FmtS@| z-aLw@*tOV4;c&+Jzt=Y+MQ|w&$d6JR+a?h+` z&R9Pd4nW^se9k(40lcez;75D~5OP(~nERhil`H@YeXDsfVup{l-@s3ELrPXLg5|!Q zu~3LXL!{tJ0e!@<`spZm%%+J#bq$y@jV5o7Buh|DPOXlF)+3E*=pkWB`-Xn?9{P+3 zY51jpTB0FnPZWeW}AJ1(*rCe8PE;pzya68ch%?vlL(QfN1k zSV_M;q@>VhOylI87O-#%q|y60BMb7!NTtyF$dK`h+cU{n5b!9``p|5?h%x3&FcITU z2T5StN?Ax%kkPqIZ`hi+Y10InyA*iK|sEnriWJHhaT?vt0?O!9rEcoc@tjMLgTCicA)O~rTc-yfM& z6&TTxMv>(*0ZDts>EMvz2&OQV(K<4l!!?@vt;H3ZHiBsHgH-DRf+^3cBeIz=$5e7mtC7y!Or>FwyNaZQcAhKvbt2u z#ccChSfe|PRLcdaqke>`3L$_p&K@~x}_QZVnAH;1XH>EPi&Mm>tolD z?U&YK({lSv183e`uP;Yd7n{#p2A-7lN9&UI`7L~x6{WbK5R%24PNk=9q!#cu*BZ)Us2HjC z*7z^TzVgZS&upCSlSD2E}S$>Sr^aPRW4U!J_&tqzvX_Wp2f zdhQ*!pIbvB&6k(VWJpL10E@ES;*326@}O%PK7u=Wsvu`?!nPB-j1}KFRB9n)0$ZkI zNM>d*F*a}P=o6F=f0f2%j^L1j+e6!iy7_~FS2veaRF~yPr3tOq99BxjI^+*2IvS(6 zem82gw;??qnrn=7Y3h7I*iIfg^jK!4=9y&Js-FYrD(Re{xWfrD>>?%7sr1E7u`a_E z8;XN9QK8zmqtvgSW}5tQMHIXJ=3xIaL1b`LJ1KKv@<=+~<3~rAV_r-l%t+r!O^f!v z-V@VjUPe5vnQ*|r$Bq+p0R3q}j)Iw7>5!n-zIWk# zldoyWp`sQMM@;mzgFX3nx{E?jNqW3q#bjQ+^M_q}l)>MjpIM5!s$6h@K>>GEl3GkJ ziUn7UH2ZZR(U>#=w>`XJ5Y106&cuab@!l_TcBO)sbg9GRN&d|zW@o*$-8TOLq}NRZS^&1g<;!Q9q8y3t{vi)M(s~~cXu#0k424moY(|XEIoAvRvmDPGMQ+R&d(tnfae1{;Xm3F<^xM0I5f+h>ZdfPnsd~{7HhrDX(Si zp3MI^w*JV>M%Q}-pOmZKER#Mp;tO7)9P@`Cbl@r1Elydbm%H;`e?+X&EWB6(nCf)D zusao3MfK(<}|Nd5Yu+mJ`TfO$Bag5mySk+}zfrO~PGj+6&Lg zo@;%ouEo6WsLl6}eihj(O)cfu0iAGJxYCw?X{{_j`~kbfh}Q@~*IwBagU<@ zI`^ZB#~dp^`vi+=orc!*sEB!)TOyA$(9`2pFa4CQU=hmH6lF2uYx1x^@rH^yor71a z+3rxHR~7%#>owGNQzTh=ry7rPh%ZlHW8NR`3a1m<>&(l4c>Mvt3>4e2o~O^w73Fj& zNqwdds|JpeC|+d3BPLX!sk6#A=)EqqOv&BcAG%x8Ok-y)NvyPU*!JJ<2Eo;{i+>jI zlEO!Rr55MYgw8PXy=y#75^$W5$fg;^^9=duT2A+W<11d$At>1PH6HBlUn(*ln8hQF zV%Xc3LzSxT+XHt{uZ*e6Rj%}^@9b+T0u|TSix$9Hqxr8VnMpGWMx#_ZfPkE=*tC#F z0!dA-vmwYW28I5{xXQX`Jzw$Rf}jaI{;6hHEek_yY5NzN%v%sIpKJX)^G^byAwK_$ zsjm!cE9klnR;)O*NN_0b4#A2;aVSoqK=B~KrMMOi*5a-OihHp_aF^ol5L^rNrtkaR z@7^DIa*{bSv&W9?*=sG!@9Ox`ZMu^)lXF9Re9mW^d@0S)R6}KJEN>Mmp>fc6JfcRI z4^>5Qzj$!|W(OF0%Aq&*q(T->(eJINq{$!pjnQ{dM8x%UEkt~^xOLLV{wsH%?CwuG z^vWa3?h8{Nlkqcc)EQIQp~9nb&ij7TG_I4s01wID$8TzFWLyZ2&xJMGYfIam2)xaf zPM|#P=HiVR((2wWZ>t817^mYRQC;6`v$#>tXXN{;_>r0$f2Rjs-8bp6KgbU0akE-O zqz6&Pfrvu!)bnOvz2qL@KN+!dJoUKInHDXfYL{V1iU|wc^h#esy_-GZhb#3TE$r7M ziFn+fmRUJbrr((>GguMcuF?IEO6vc9#oW{U-=X>c0*t7zP^nQ-gE3Wi1Apko<7$X* zc@-Dhj5WoYZnS8~_*7O~U&R&YRO z8Av7c+$*RG{B_~T_KcFZ6n=IqBti;T6a8twa^Mrt5O>;?Gy^e)m*g3N-QBoeu<(w+ zdyiAT*RL^G82?_mBBD}t`TZvvHa(@HaS24x?sd04)F$LrTq`<(7>g2wExO+v?6kK4 z?}z~+{jYd?f6b=$M_Ssm(){5fbG{{Mt68bMwOQO}46CDZZu$A+xwQ7%grCu+LV=%E zr-1K1{11=C7{53Ly+>cpZ#m6<`U(jfPbP|2&u4n3k)n-~%d#x!p#jaonX=~_mzby) ztJUIogMpuWY#Vpe9Cnl7(mF`}9?A}2FHtGh*(P;P15-Q^5$Z=A&VHI#k}5J*xLcL;Hd6!_#kC4{V9Awsc^g;urOgX~NM^1@XL5 z3u)&u@>a$-c|`q7V$`l@KO0hac@F(~KbIGr2&T5~^%80<<(?Kn<B8@Y|8gjpKie8*xK6n@uQ+rk!-rDKNvETCbyxhwA3xMqM z6MtvCxf|H|rrzC5?QKNiL7T_vOe4ItUH7W7;Wlnh#7#q>S@OGUBZ)?;R6kz=_^4#m zUqSM9hFx*3f;ik)O+ju1{udA{jC*3#-?+e1@u}1+esT>LmvRIha?pC<9>ep4(EkHt z`WAE9&PMqlD|S~L9}R~G27T&L+r%}Qv#ZLNkg`o!8aA(MgbUhmG#D1e77O~+*pci5 zyxEDP2m6M{ll(}R)+S;u`%1}ODR?$b@9!<6v=m?&bmD648$n+JQdk*4!cCjC56|~B zf={gH#;Kj!uv(a*6%`8wqi5sl{Z?tW^4tC1sy#osKwFnl#)$(A*EJ8SjpW8cB`#(pNGR@A zask!Kw?oiqjx}@)hUU^<3ni`yPLQwr;TL)|qd?jMf=%(2Y#2}*OZ+Qc|5Yo&DQE3enw{|+L%l;LR%urc?sgP%&^K>3>^C!0)UP-6 z1_$*Q!U)R>G_9?BEY9~b4bu#Sko4=Kj)jvU*@0a<4B7c&stU+5l-U7c>E$0 z>L%xthFHq$6DUuwwIGwFAMR0=@m+Z^n7pXbs8BdKHu?NVlj4kc!MU-=tFOzjhK;dL zZcv_>oDQX{ma9NxK@n15$`l8erjjZ#k5VD1WsF{=r&{6%4nt7pNsW?Q^|w(!sHlSN za$EXK#WI-$W$;1GAV>QwF0Rd_5L83Tk%AFr5?_i&Lu`W?^&hfQ@g+n>2@z11ZhXKr zDml;6YJO#$^bO@(*SXYaPn9US2y3yh1^Q}>JG{y_*5dFJ6Y2v3f~wc9aFv#zo&n-YkrwoY++(pW~Fk-OT1JU>=(v6GtKX3hTXB1Y)(xl&#=Y2=D;^HMq-M9 zV2U{FT%%0F?@?I51SSl(cQZ;sMJ4WrLYg#8fNNDeSNc$N0pzdP*?{->@}%Mi>H0qe zQkbo=LVAo|3%+o332olewYGJ1CL#80JXUmKgG1+>j{bHROK~@FY?Qerv^HrNib^m^Y9qCM zB_Bvf&pa`n{vvxOD3ggn5!6B>PktZ`x z+O~l6Y|m($t27{2otl1k$sk>nxgWEem%vX~M>VHEQq~q4?_&s-e>r z#mot==n0S*yK{@FY>6Pm3#$yO2#hNyXM1BgNg+NBD{r?#Ez&2PO-|2E0^Hd;>)y$H z$SL>J39ZW3G8x2??!sSJkl9)f+OCjI9Avmw*-zNU1SZn&Mk3B0Z0H>|XI)PgRRZ4}3UqcV>$9N(F0`Y_Bo z6B=ltzsyrC{2TWaH$QLVMj%UDBOt6YrPKCmMZsRXZ zFQ;Z$%Wl%p4FDXxvC+AIGD&a)omF(>u|U)62b|6h-<`KTsaUgD z7+&NHA8w_b33=LlPKu|iwAo1CFx;;cd}z-5TwJS!fg{`v=uB`OdKA4Tu~g3`@mcI| zw(9$ZGY9!nJqMczw@bMuhq==XSh$QA@8~!`&L%8>beGw*Efx(?JT>}GS-UD{h|oXr z#Gb{$X3bH%`@LyMCjXlplM+w2I$UwXRMbE9TQMt^;Wq(F-cB*=E;a*Sy$ZjG&L%A} zzi02>dro^r1?OebZfeWpuVuqUkJFY*Uq9o%_cXM*i+*B!*j#LyYQ;#OnCoYpL;pG+ z{qa(hvbg8r$N2#f3Xq1fUw&`BoCdl@lz!kSY7~=}hRg$3H8x;0-jx@^=|?pnSLhzB zfjXayDAC%;6(`<1Oo?Xg<%fm?-?BX`as5s4S{>&2dyVcFC8@|WJDCzzDC7@X({A11H1Lx|d~0v09_W&hmP`sF#h=Wr-Q8wJ zBTugLk|wRhHC_RlS8o9c?GSTS2@9GEG|X+>@DVjkQ)2?8S^&cGiHTztLL_9YG<)a- zb!T`+I631%k8@dxf_fNRkr9yH3N2>%-0+tx-sm{0BdVoz%<*5$oEuBuOAg!slY<4+ zU7R7RN^errf={o_n05>NV*8=u<1%Xg=r2K8TDR504=^kpKXGjFbPwwhXd`FiV5!eG z9*!K_xeXJUXQv_OH(EOaYCFN_5i=$<0KGQ@{c1Jq0Y6bbD6T|ADL6)?#m(~1N+=0V zQ|}b0gzRni5TnyRJG#xoQ+d+s=3{P@=-)S!Hsz4#Q^m+j2qdxFkUNJcq!AQnx4nQ@>bb4t-X5j{fN;;+?Nx z`QhigdlB%XC?!}s{GIcc=%eFT3lg!L2C!N1y6;d?Ms!t@&h3=qe2f`)lUW(*{J;~3 zedg9Wul6jlYu9eAc^;f>$~){z=nJ-IN!l&su@aAHzEz0lIPBL2V`h@Z2Ua8IQ~KDY zxUW4Cb|t599=X_%TK{gF!;`D&;6473i8wp|Oo2-F?Q!2FU+V^A%ZO=a@(l|8mvI(n zQlO5*h$sZ@pLLu~$<#e|5M+})3Km6iY2s8{ZswSsa7IlKC){g{gifi;v>^wGkEopG zj(sEA?o)t*Rb54HG76F+ZBQz=76Jf-XAvJ*#_9`Jj zMQ?=HaUpT}-l)^rY71RwnlBFcFKCLtM2QtFX9XL5WPq32GUE}SM_`01fTDw-+)CTGyw;^UZ%|Sexs^UKY8&N10(R_=$ULrE z0<6*RJgMMq)QfsHRCCBXezr`$eePp_0VA*h&NAGKc%N-OJ@Rb2O0Ofa&S-?+U``j| zH^}~7u_@wDHKvb&_cc5o=&n!o=2HcZ)EVIE$f&O_3XBuZ`%4LTm{9Ay6AT7aeOg?D-=RNKJgCzY!CeZ_<8H(5O)uT~!)9Cr3psD}CB6&$E z()Z8>sd5|=Htr`#O95$2O*RR?h1AHxxR0fVTt=>4xpVD4t#7+3S=VPt)LXg(r&TF7 z^w}$4Ex?5^Z1TV#Xx9dCt=!UEZBDL;=N8|u%P%L+OE6L+FBsW32$mOxOxt~{c4>FH zqvkD|?>hNZA%eTGs?hDVqNll*<8V!tquRR^@2;)G{kDtZgmO+La`gbNFtlkS7yy5M`)(R#@>&&afTQkP_+?r zq0>$xwIMsIO6Gk^<^LEw*AP@W9fLIJ);eYiSFkAFI_dqCfkN@nN7Y+Qlxqu2OyG$D z>#2apVh#>N6uwKVB?d<2T5D?C*dCK1WA2ImB zIwTWfb8bBKx>;LQW9w+ZnQzpm15dqkKB-{#(*A0*wvMmeZ+tD-!QE^Z1q=Hufn&4F+$J- z@7bxbJtAJlS_!~;VM+0dW(eqH{#rK!->-$-lqC6$!};0+3>d z`l{G<${*Ux1fb4ntCGaNO)_lKt@YN-x;C6=;Dv|PlTwF`*)l+Z`xR4+k+xfdicKEO zR}f`c4TUUg?0rU*UB05jc(qv@8&$XX=if*meS6QdUHcPt)l3H6htV-1KOL~NCfZ(MH zCeCguGTIdeOg7o?gyqExkvXiZ*+<{%nJv8yc zWXwn&iOJJZE-OT9rMZGqd7~^0jQvxa+(9mNpy^RkQYn5^bEHsn{Jl_Xv%m={S81dT zR2zs6X^$2SK^UM1V=}Q8H6O*VU_S=Uj8n0D2$Zjb>LrD0ZHehIAb=7aTi}q(|IY`F z&)g;DC7YZyinK*bsHF_hG8Qeu>0$Ds0TssC^dmU^H7Ht$gGSl^u+V>#3eYLMXQW1r zdDtfzU?6BUSRuf`Uyo!=fNJp3doD0hpk5|f;s@muYOM?u(rBxR6tL7L&$}@rDY3{g zgdwZ5GYB)bS*DM0(u`^U`svzDcF9lJRo;Fn?1v zr-`A-NFLSUnd#cM0_$(YP-WXXc;vxl=zo=q2oV`EH0gDO;!k;upX*|mpZiyz$JW4O zP(T(O11r7Vu1(1EhzLnH{hAC)?7(XI+MIJol}nHePttY9)7#lhH{*}d#dgZ2hcw=n ziJ8H#4QiJ^wk#Et$D`2WI%)WXSs7?6D=D)ez&Wd!Xk3^~>8N>G6NuY&ATcy0K}#96 zg`r4(`iad`vZ8 z1S^wC6%x9}K&0Kf8p(t+VmMM2Gk1NAyW=@_DX!zYI9nffZ;Z5^H!{Ln(s0S3g$Y)l z3@XYHv11X8ug~5#U{!&>w+&(>CKDI(CuWMnsC(HamUU(9TjcolSoPU#r5Hz30z>BA znbOmG#~QBNSs;~x`d*6xEUR47z`QQo>GU7=^>!!8 zAEInHez=`7+5=CW$ygcL>){W|6@km{Gc7f1*e^T3$=BOU;!9mkf(0U2rJO`Lor8h=7_z1?Jg`&cKNp=Q%TB3jQz* zs_cVQ7mW^Bp|apqW|ShG+5d<`s%Bw1;m}W%?0{U(H<2tS^16badi|}9y|7_A zSN~2?;9<;_Q8YPe`ZSZ^8(lUJI^=orn+;@vQn8u3Ksl})Go%3}imBy}nA7S(=?l>jP{ z66Ql7qYnf9AxmXU0RiDLu5BJ6t770->}0zBz(TVGtM9i-CN&}%-34N|3ElTeY_hfZ z{rp)26qy`pM0zqRM810mHh<>%^9w7B<3vlB=z;zoEiD5>9<}AQHH(LPsRl0=yhX4# zTH!3CvsQzhczGMcb>S98S*j0yN$NoJ0%v0()}^{Q21`9_ql$Mf(FeS+JIP)NdwQzT zF$Ml|a_>*g+6pKkpWIGfiuB`p=x=Xs4dbq`4|ANX*zLf}J~7cW6j_whpKeW9`ih$! z*I&}wI$aDbviyPyT)09a=zLPXp`MIO99PrFlf_*f8`awlcQit)M3u`2^0G~S!-$fe zh+E#2iO;A)B^8!i+T+wUd`bxy9j{F`O7o-j2K+rmT0A>JJxS^+j5vHOrcY{SUE|!M zoB`A4CpnU=*%FofKoakAbX+Tjpr04qlF=8cRo!}qM(n-Kfv2UH66+mGi|wQF!G~Yf zWUDQBwmKgGrfxDnz> z3u9R{fv%pH>{f~ymk@e5>Cw~UOrd-U@lJSq*Y0FJBiXh*Zg91>u*=%U=jkpsAZfLn z0$I?N;k>tKmtL^3%Brz!IYANE)phSu#-=x$@=3XILhp_OYYnRHX9HYz6?6%H_Zs5y zt52ydC;_1A!8DgaL9N>KkT|t&pt`+C(`HU&!ymh~QK{7a$>ZSEfiYKUt%4B8AvZo? z(p#dhkY*W=iv_g_CB<)*mS9K(O}Ytdk%({9K`paUDb$C~Wq?sqZCnFoBp*W%1YXX| z7r%Namx^$J0yK+#DeBbJbxKL%)v%a6)BHDF#TzrS$>|a`}J>ZIMk8}M#Ir!bQbpm3g=`QM93Ys)tQ@`>0 zwz>IvM_a_fZJlyBElH@1{y<=FSn~bq^`BUM5mEzFmilF|kZ=c9ob7WwRe|5UPI&$j zcAmp5|M!Ve1x?_L&nidVWTpcBCr_y<^0oBivuog2t9G6hx7Moo_zWED?98E+g?ZvR zk%9?S(MXgIFR)&I8_fpm814qdsoIaT6l%d2qc-zGs>`@FDGq$?DG#&ABO&5*MsaWW zX_>Ng_db6|_hU(E8-|?y;=5ojOQ+hSC%w|$@i>d@!YbHgD^8wM+*~GrjTIsB#pdH2cwc;s{Uar;lm$VcTtc~ zKiqxcNOHf5E<0bu?vT}O>|QV)cscS&%=TTCA3;}OhZ_^$!ZM21>^$bZ!~gp$3`#zk z?aQTySv|XFtih!h4&`U5wBiKpk zqN9~t_<7FjM&Uger7R7v79;81;&?jUSqExUuhM+Jqo?)J=xB&t)-#s;1#rHj2?`F!$~Lj{=uY9eKSUEWFJPMa4m*+8kWt2R z^MH*v2*~obkxk*^t>%eWS$v_waQ+<}2vz3Qq%jg2U=daZ_p$W)CkTZWi;o`FN(R#j zY6!#Zm{78Z5o$qD^70nWCV-Js;dPYcs${`?4;Z+#)i^J&7;KsH`xdvz?MCLEeeT8s z!C%0qCGLlk (n-tg8eJua|#^Pj(f-iB!xPt&6Vj{`=(j|Le$Sr$O{ZeFnrTgL zN=Q2qT-}BKiEa;#rW@xvvmS(`aG=+6IPgOBlZ{h2!bR0{WA(?Fa&~v;*@-=|p+U1Y zHL)hn4E3CJmuM$f$h_9E%WdcktjE?=$ul!as_4Stkowe5r0(30tWMJn2;9LvbqQXR zV|-Ua@WUx7CI|J;Weog>pSxHUZ#1j7i~;6w(DokZylS^MWWF~fBh_U?M~JQd_Qi}U zH~F^kQb`~tw_JHucJd#B2lzen%-PqSIIf(1=9v@YC^s-vIEC9z?He5S!sQd zwvQeB`xJkyRlQeE9hKAA4fMAe9Hu-G#ZMWR7?Xc|5wAOBpZt~aoKS+(V15=0{5tpr(W(yNiRH=H!8^a@wid&^WhwGcHirP6P^T~_Uk@-B*^Lxdk!EP3ph|#!Wfj2Ena?W z)bb`%3!UzUI2+%_|AHF#mX+NUHmbh<{;=b%A12F115*@b#;?_DIT|{xDz3O$x3AB& zr5TrDmcdfF2#$#e_cyq9Y$9bE9Ox1J^`W`i=KYDGg+arZ=15n{=$6@$Fvhn_gDmaQ zisFR9D46eP#9RTpt_h`HKX+|6uP`XDYU7RLVxP-3u_LQK#K1-MkEgXW`KS$6lq=Ka zVFbQSC;>hnnZjSd0T{XA_6&lq9h~cQmiGG``evASG0puSGWM zhZ6He@^LEHN=VWZY9 z7D_pZ!l|mri@|bRY#L0lsDDZq$Z~I#7y~Y_3<()Usp(XyEeysO*v@0hOZv)8_4~fc z7k2YOuKdwll?-krsR&r8dJc1Yv%&L0mBf)DxdCTHOGrssfAjE92pAz+$Vtv9Db0g) zFEO~U>%JRo@7QCg6n-L6w!bk^YHl83i)$aqK@-wUOJ<;k)czogyri+U+RsNOZnSY~ zSldKFsLbn3KqI{yd5LfV0AvwB3mPSa&Q7$_ZV=~o)WA4D!eK(3#z>6#5=&Q-_b-$~ zB32|8p{e$mYFkYfWZ^yG>)5}Dd|xP=gDQ>s0`p8pgg&Pm_33BUb;_C6TTG}e1%=C| z8UaK2f^mI)YhB6%}6mF}KF4MAuY@Gj2Iv|y}>6(>d=6i-Ge{iP90fR$6y8T3KMVQ|EA%^0_d&!G6)fg>zIKJBdPm zjPh+q{&=30qGY<0I!$!B>EY`EZ+=4MM;LsMuP%VAY?!Q?(_q!=yIH(UME%CB{@Jg$ zpJ@xNxfy0YV-tC8MmJZvCrmG}eu6OcQpW4-xS5V{czTr?xc!8^%wl)~dJvks*a&h(Sha=) z*XDWVD^PNLQ=jHgDkP5Q-Si02}CYt^8^$E>{x@i>OlqK*bGjEsCXrcF& ztSQf^*Gvx(6deI^#*R^z2A*!7+^h-*b859dj*bZ9%QYULQYM07j>4=)3E**v^ z(fC%JA#3LCPPUTwF_C%Or3w-iJ`={$NvN0nN)Q7L6R;COGQemH{vu9~u4wRxd}(Pk zuqK4J*}?Ue2OJo%gV09Yz#e?4d=mZ{zfXPCIZ1Uu^a+ZQ`6N*$=~K~WG+Y)lIiOMi zcDq0g&=q{VGz)kAHo0%diHF@=O1^D9(;+C5XeisA<*=I|0)YEaTj@YajN3bFQ>B2;!E{D^vt-7gT%I&LL zvwyod?^ETRCy+~B8R$MU6C1H$w|(tORyf}RKTkyRkPS4JfUyR!w61|P{ zXzhobpPx4=rS!{j343{8)VOxD@8T~&`{;aYcgVLgbpe3y4`WF?Ryu&|yg%}DBW(RN zcHX#<#F`Q5-6<0C1H%|2quKyT}P)Hy<3LDt?;5-^?vZ~KU7j#F{ zaI!67uCHTc>9~-(Sl((^TDD1Bi*eSUAVMgTLN&_cCd7K{PKr*2#Bt(WYjjX|q{^nW zwpsL)o~})vDvNC*jX&166jN;aAOyGFwm$BBsO&T59#=EZ83*lQuQo}OTbFDUeDkkyP>~(ukiq3z(brPE*(6jg;`1+^Ks(=~5lR2IL zP;znx!72c$K>*>1!2%AaQUp!&-lznDMcVqTcFbF4arzxD3Db?l>2@p}-9N_WJX7$d zn-)-*qm>6o<#dt=V$}wyQ1TrIb#bQo7DdpB6Wh3pP=oogP-5Y+8^;VHh4^n9H(y0< z7B>HQsZK@xM<%@#Yj;l71X>l6u_bWVy>8#1_3WiN7#W&9CU4YekdK)mnEpuO0Sh~R(FG06S;>*jMb$*hXLWJj;O?H`)EK64w1qxD)M1z&vnVBhD z?jlxPj<O6CMrL>)lS)1iIS|RKq?_&eUd3B$3Wo*l=N?ook%w3kcU0WA# zS7P$LbD-`uUjcSZ)u|-=cM`iM^D}BCX(wCN>AKA6KG9!7Ha&ot0C20GHMbldZvSpa z6;$E%V9?+yc2*-MZOKFzq(Vgv19dB>#f=jHO*XJ7EJ^sOzf|QAGP-Kta!vMmJU&9N zpmP`8pl0uzFReL%c|ch#WG)Ly3%>4N2dC3`(Q_^rv%bMWpX*sp`EsJ!cl*iEQ9a?+ z?fA6?*mLl(Rh{iW9 zgAZ$(A+OXZV#j-#9R;U*uV=@}m-f$dzMqdTziOTCZ9JsG-XbBqoTu4=8_XL!CCIF{ zz8BB+$o)PoOnCR&Jn=i9 zXl+trur(pzLi|;KYjoV+J2GwjlsGc`aDA=`GPF3w)E97}7n;<_k~Woh&5#4uf+kd57&egmRfBkF55 zRma%M{COf7!#$BMTorC4dUE-<^ON$sJ-; z6R}=eZ?FJvGh6P;mrDN}-q-blYDc(aYfP@M#9}DLVR0O_r?4tgVEEt|PbTl0;BRHB*% zCLzpU5Zx(3<{#L#D&3$Khaaz%{OPtd**p=Cc<7rQr&I!W-lUHwu`O-7J&=bx&1K%1 z6U|v&wg|oC#T@o&^m3JqgscR!ZHB~J+6dHzz0f~FLyo=vGw1{lKUqlVowt#u&fa>T z%o5yJpz=G28Usa%u^8osF(Bj$^sPQb5oWHelqlCyFZ`iJb>u;EQPRSI9vL+$6<|s< ze=Z_gS){z+FqzN?!6Z2usw4wN3XMpvP50yt6MRv&<#ZIkS-A?K?>MYY{?v-}a0ol1 z_K_v0i$!~@(Sx3j1qdcRj7&zothGG=Z>PHq<&`Yk(zrw$V1M|-`WH}l0+00!wG4}H zo4&0&H550K_nX@bG>UGayfJ5$Mt;%;l4z^))&G!8FL&QuT$jj!j2tlEEvzjtTz<27 zukB;E0%>~f(*VF*VD2Q>K#9Af29cyl<1}=}+MqwPbhhe5RtxUN9vb~A`j{Z&IMA^q z`KxJL^UF&0qxNSxa}62Cy{)AWcX@wG9vHE&s~x#a|iUqge^=2PD7BIk(sj$fy2y5Q{kfHAp@~aG%6j!9FhQG}?u2aZNXua_Ey9rt?X ze*rx4w8FSu?l${<67~&GWMpzC0RWA-O0v|@)yw3 z_owPP^7H$%zkvJ9facTv=1r^<$^;irJiO;UYIxfN6Gh*nsK^T4(|mwj-{f(INt?zN zClL1ZpG4Yw>)xE*cmGM#d^6gnx08d5DAZdBVGF^t!P(w@lBE0Eq>C(3i{ME?%$@UJ zz^yQ>15x8&0DCj!aZYj>VUQduT<~Nq)lcZz`Zc0P#yGx>>Py8=L`LUrqE9QZi|O!y zQHv*7fNb%3$xDmpoA_JGr`Io~1;t8sN?rz}!~Vn}nqcS*yRZLG3k)&q!24d~;dp-m z?*C532io=m0#<9Dw>tATY?xe{+(Gueq=;C@f`4UxNX_JVcBF=jJ#fAX_N4IC?9b$i z|Fg=4uuNWfI_xiCbQ5u!T#ts@#;Q?&CMRM~-Qlg^A~Nac+Us_#+j2Y*pt0#;ohP*U^nsK*r?p~hOZ!^vbX*-pET6Az(Zbb5eV|-`riQ) zr~8pEA8>P?4-f;V(wV=NLC3P}xS&qre5+C@iJ0I?_u9Q%DO1${X1EI~T(ymR^CwmR zt^U7mB|sdGKONxxtCSxx#vbZJ*;JBGbJU5?QJkA`)QGg#@SOeTL(FBz`ITRflFzpP zerJGG&He)T;{UDCL{vDX$)n1@g@*VjJx7C<6)ZhYs=)yW*=xBGeC)m)gdgC zyH`MfzJcQ_l9LdAqUNQdXHA3izX0QZe=uEyk3Wn2GjewTv0M~E($?co_W!9WX9!#2 zZtnBJ^Owxrto{|^bF_j-`+$E&Bn8x4Ki703GVMhoGQAc0qkh-_e1Mqm)Q0NMJpV}; z=o&&?USIq{NKcdgtLXLClHOfE;(DzArs-E%>fz^y!k2_|a4kzg$}5QQAW;L)QgE`6 zW-h%H3c#BMuw$E_Ad8DXgu>{S)7{#|wNhb3dUF^alU~Nl%f0%++ju4i?O?5ItjD}@ z9RQ83HB}jY(yzZ?_`%epN987WYRArB z6B{_*+Mcmjf@CarQdcXFosjqv5P1-P-k5Wsg2eoy63K zpl#KdUd%)Cvzs54$3WU!gG^6i5IhJ=5F3zr^~Bq zey+=wW5y~`gJMvP_i?#tx%BJlj~@i{om);B#O#nOe=g@Y7!_4~jS*TvFEeD)P=4W- z=~b?BiZx?=lE&KFUQ2plb@d6)76v?P5r*{aiFNP{$9^>ew))iQ(wAq^Bn+rDpd`5F z>F9hEYL5{pV53Hpv-+t;RxX6|#hCd1lp{NAa-e4+Q(lq6+`k=W%5ZkSlnOrto{sP- zd$+Iq=^VXF)SQSII2PH{u-ScER_1Hm&l5po`KV>WB5aSImd;yS*)mA)uSfdTqf0Es zsPa@)ANL7emo;u)Td10br{6niq(Yx2IHftPEgMYl8j|~rdjiGEh|y|Na*kam`q}5H zn6auo*Ef0AX4T#H>Z?$codpAq+a8>)!^qz*P3S}x5&QFjYPBL@^3*NOkBiCJaR2C2 zgwpUQU-ho?rNEajW7_6Cjx>KIcgzzkf9&Vcdi0x9K6q1J((bPr__GdASo!Qi#_ZHC zeX@CIQcEtMHbvbk@~48*I0f4ce|TUE15MWodXw9cVWLpEfN*#`VfpcJ0UDn+F-N>k z91cpKLS<_o8GE3;*P%C>CMZd#(`ZoSoRwEsWpn<@!ii{>uS9c@Twyh}AK>>mg^(#? zmN5kM^X$g|AIjb;D2@mG+JxW^!C{aD2*Di&hu|K9%a9P<2X_dN;0!PXx8UyX&fxCu zGPpax<-cFm)=F*F?q%O}^+n6u=RM~+^5=5D0)JKFck|*e#QfcS3nWY#I-Sw;D|^dN z?~&$}joXy{g#p>aa(x1?zQX#vutFlQCYIo$M!U76bTFXN_gRa>{`QCi(i;o$>=~Kx zN#PV~T~wqdrkWYeu|2q>;{cAOu)szRQoE`~OjDrt)n2Qx`|kDC`N3C3rm~-3z~toa zD_PpJUZi+~JR^|Uoz$hJUlx*#w4%etz*VhJ6g2Ym$90l5x2_4z1O-?YW z|LUPxS}f0>ppB~Lq6%ofRh{D8V3ioo5P(ldT}OfUMshUvsxd~5g8^jPC~H@H$^Y~6 zBM!T$diyq&2sb|?YUKN-PbACYZi0g5;>CW{{wEoS4AappWH8cdax4nO^F^JBBUjdM zDq3o)T$Rq%vP4EKIVlEze22f%&HzPlQ;!QH--qfknY1)ODNcXibJTA@9S3QWZamEr z`Sr}MbadpT7gkeC_R(?>`n6OHMj*B=2nLg=lbS70)#>5O{DF7B=4wsPCY_q|nX;kT z<~@{7`)SB0OJOrw_Mw|S2!TTR+sTs(Nk#^_$68TZ2Ic&B$tK2yp^^BUN@{$mPRV{_ zwiZqC4@3#4N*Fr`=K3|E>CoFk zb7C(@c1yIZ&2yupU!zG)eZL&FFu(KZG>%+ZJ#dRFeLTFbbhC5dJ7qTD$0W2N+4T5k}K0gIhpR@+-q+?s|OkV0OE;k5F z{D)9^w))xK;R)|$s#z87mn<68_Qd&Z7Td*6dVv-!g7~Pq1 zWn@HRfE)5tqH7o{H6+aKi(16ClQ7jgPLV@4N@H3BmVtr1xSe|KK@cC2*|KPR&a*30 zU~X&&Z-bm|sg#`n8s4K^W~icB%6?k7$pzRz0BM@3f%}V_n;6E5Tfcck$SI#|l?FeR z=ys2jA^(-W5s&TU#Ps=1KI@c519zkFJGq;;<`Z!pw)RP<_SwcpX%G$%dOv8Z#Dm@X zsc)q{Z}i%u%~Rff=NhXd!*_9|qSXKN*f`ppTg^3Z)7z>}?_)qGW>`DaA)2;Sur=~Z zRsO`D14w?j5wqGr`4+p%a2F}pI^6!_6Yh$HoY6y`i&q4l0&I*y; zIvg@Ed(az>!7cI+#_LqFaC?;Y@CRI=kr<(a>}niuvm{bVL92GC)gbOo)u4P?QqcI^ zL1B5+SsO^&wERw`hi6Rj=myZ?fTgF@rnHGe&>AALlUDrb@kcsf6uA{KMy=NC>dO_Q zCx{kp)_CguVNnfry&{Y{%e!W}FRZq9(vNdg2`R^3!ivT#QzuvXegR3JvXbguY0Z>a z@!r5LjqJE#i4sI+aPuGP;o+cGd*lxniOAn=pEskX#HC~pHtzY!ci!T#8AGH_J{<@| z`oZXXtoGSxVPDK@WRg}HMsRO`iJcepNK^MW4#>&{3cM}5z}pu(a&CWbT)5;{se6WIgEBC2Xag)TO(}L1SG+7s* zx=uR>hYuMat(Jrnsl%v*QPOsn(fgZAEmi7P*(Fvb*LGwHds@lRT$hd%A5X5s6lXw( zQ3Rs2OWOs&l&ZScBC+YKPT#@2IO8eNqQNU=F)>L}MEa2&T$n+>ZV^6xft9gLz1KDw zLSx|ATsiYWGIFtx`qQLs~6VtDXZ@T!)&QzgQO=Nm6$TN!+9C zIXN1Zx(i>#b&JX;WJ1INYLrwegs9vnt5M@!KP|nLo7e77x*#!9g(*K^KWZC%rSE^R zfmX;rGyaM0$qgGFU=;o{;PGV{T)w4fMITnl9Y@eKY9F(~zuvp)4I%;PZ8~sdEEZk~ zmDVIHgp6L^wY6Mmy)CcUX5AG%cD*-DR*Qb&AA3Y~hj|@O|8d~0?!M2Y@fu~A9C3~g zGrXPc!+}5&mQ~NUkW4Ib_sa07m6vms&0JUu!tw>P*jCRjaL0!C9ASyk!X9O?N1Z${ zj9r{vr$6oeEAbZd7)_Q7doWqbUW>eykNvG$tCQGn>So5+(LMD3$&AWOlusgbVC)`B zjm8x_E4oZd#lH`$bNZ$@4!(!!ZQ!~Bd&n|xk+{2UFr5NgUe{neH{Edrt#4KA4zObz)=iooI zv?*Gc2UZIt48pI-=rW-aF)J^#@};XCxr!J z%~D45CfXwNfcQp7QPTNyCnF=l^pxe!y>8?5eoB!zRls!Khn|$3>Ci#Sf}g^O8>w_C zOd(2ba8w2gQ@Z?bgtJm}xEIQ07;gJpKc@$~^mwu~D_f+?+yiu!*O8rsvwIsGtievo z)9Kd*DW%^+=BsBB zoo{HGKlJaK0)A<>B#MO0UjkZQ0@?uFg-=EripUGT!#NZGGku+TWmo+9H$qsgnQKjOqXD5hAiJ zIhgu0+r33_g^z;fqYYcTbradD+62Gll}MsYTvh7TJ!(oZbdmM0$MAwz#clK#ua($~ z9*FMOy_Y>v%CJdzxtE@`Nw5_!zA#K4soF>O9*gN-;x^CUpL`bllF**LDrFNt!8d2Fp4ZYE4>X1b_~4{cJGl3PdO)8FXYlle$tj%wl1#@A6xgu|4gwml(UK>XSH9E{S$>PCNJ0$~SX z(ZW0LI=gj-UulLnGQ*85hg&o&*)~e}M@G{AUgb;0MZ5*txo%nhDUE@iU~p(qQJE$6 z;=umypbN!d&RK1U@wj~sM0YpPM~C_Og9pV!@0_r&9TZ8Op|L3+@*(iH<%rj5z%a7~ zTilqc>OuwvS?mGGaF!4UtVL@PBHKrGS|Ib1&8b@o-vljI5x5tEDHPSSax>PVkjMI)5tXe{4{Sez8q{706PjJztdc|85p;|I;|( z_l}=3g0D(x=rJw^GXkTbsh|(FjD-EcdGWXfKHOoc@T6hfe?Gid1{pZVaZ$hob9esa zgTwU5nOPnwgo_KN8m;kj=X%AMf%F`pIAR)y%K7-lrfIITS;v|+O9rlYv-+FNTkWKH z#9vOZSz^{Q@wrr!?N@_R&_A1qkYwTb@+GT{L~Lr4VV7md(Zp)Qt)P+&_T3nz$#RwV z;S%@s7UhOch9SDA6*7>j>mb33R7A)Td-TO@p^;z3M^MqLmf)^jqtoqfrJ(o^1c%iI z)C^x?>{+p6!~Sb1Yz{g7VUzoz{KNfG!0aa{ZzqLK@N_eO)Vz4(Mc(bY78@HVMaXJf z=sTdy*|29r@@urEdvu9{?0Ud!NnAUF%hhm_G}lbIq4$^e#uH*n-^sY=8SC=NGfv(k zXl%i8Xu~VhI>9{Z6N+B1h>X?dk|Wz#F4x#QGt@%Y&GUNcsOu$gL2 zuE~x~+Y8H3YVvToq7*{z_!_|pFU|&An~U8MoBgOY;;8eBk~IS(am0>)h@_@3O3>XF z(1%fW;AWZ8e{E1-Rhh5SN+2`K{dX;xXFC4-fWv{&vuNCk4#(4>$tRH^4*KRwf7MO3 z_tpc(!pQXXzfb~^pA2lP>z^Dvv;Qg34TA>97A?025Me(_^8=qFPSQebkN}9BU<~*rri*K$3;^htnw7?!-c`os&yd_ zI}RavJyIbf^bjEFQ{1w;Qw83MO=9|VT7qtpvd=fV*Ik5$w{%>k(QnOIUQrM%zs!@F z$P3)5>%9^s^$#h5R}On#f?lb7?twKaC+AoG?{7ig^D+?UP?=k?vK0eLb&^z ze~a12J&J=uag=37@v~d#M^K8K#5?qn+A!T#_3AV+#P7T_2ial<(b5+7qi@Y@L;FQM$LE<5d_F5x=;_ ztmx@N7`Rs(Bc_zvtk4`co|`h9>#~oyw7wzq8QkO4{J6WiBDlBxtD@|gQA3SaRuOY> z&$EZ`%Q4|JLe#v(=Gw4F`@R;ym4A_~c7M+01_?1|g%piLr>{>SL47a;?UVaA;6v3M zznoQ6xr0=p!(IX#pK(T=Shmm+)&%(0pr>}CUH_Cmr9=|uszzqTI1sO;w|JA+o8q;! zusk8cca$eqHMM*)`iw_^51zIwM4Jn3c#))^PLBPAeDY}3Q>asM61()yQ!##;KS7J2Y98Tf3DbPY0F0JHC#^!9zHlDGl!VR5<1|u;3d5@hT7&Yod)NDxbEt?4TO{ zuExjnsl?A{<#7~vTfdCfWlG~3n;U`;c8qajX-iZqG4;GEKI_;2*RlN36=YyYjSU$u zb`v`$z)Vy${L%@eJ&Lq6$BEb|Xq2fon!vAg zxasQ4Pd@zjwj2sE=XuS4ThQ4J71t|#NDkCb+I`V(NMcH6o26)v->}#UvU^ya0zQ-G zIY+;D+&z-l5Re{o2vB%0`1_5wE#KAEU2E6OQ3|?J9{+MekwbpjJ?Fd`xj(mAN*a-3Pq|{#ks3}M9H9MXkVub z%f_v=7wqOf3D!a4rxYo3`wGT0w55J07w^&sf&1+)Kp7%?&W>G!QT^~bkQg)7^-m$S zH&4QU2=DT%ns;p^scV8{eKgB?t2GU>8rl`Z6{O?vk-wR(4Ib3RLkBBofugFF1RU&$F4HSv&` zY@=;1Ih-F>s2LKVhFM`^el~V{sQ_x`^&vGhG%?*90G}|Wt{ue)V8yeA z#1iV=rGkLlfd16J=#DKX(baj)clOz?^tf-1(Kt#h>jyKo>D?shL9K>5V$ey4DlM%E z10@iCd`~OdGSUzK@T2;tll>53;)1+p*nzwLyclpBNohMYeYo|!@i`jTJB-nJzRShy zxq!OS@$L0wgF(Qt~Fxr=jDWDs9UZho9{SQ3RV2$ zO*L!q5x2|qM`n0Bx}}TnQ6*Mj-d6L!!6RlwMz!MUI}U^F{yV!}*Tqw}e*QB$#>Sz8 zB?R{V)o)M(R-S#FUELL!&le0wj$iYu1bO}PNO$i9^Pz9)nE%19%+(sr_sGeO=$X*qT(dX|37cJ%V8=lCq1cbh9ndT?THB(p?UAwlNLu z@qMF?l)=|nJWIrVbs{>%!fPi4yiH?BwDyfakBioI2UUCJb`rD5wyx2JI zgLA)J*j~%xCh($uyw}y^LHIIuN|-LU$kk;Yb5Ut0KVaNan6F%<{zB77QVRSABC_XR z*$FovnCm8+kjl6kDBDNvgxeFRfKiK!{GYQ-8xsQedr53uNER4k%bOGq*-c#02Rg0X z1W~rnch_eOwvDKRoyMrf?NvY)C8klfP8dd)3hPc@Whanj&rM7_wg|%;R$D09SXot# zS};ldVi(T8MvWek$Wj*z%^K8nI-gC;ip$=E2on>hJL8r&zrm)P;A@sAA4~*>hbPC? zw*L7&=*X@W<$kC>D^^Q$>_cedz64w@gP81uKc&y|bTeijO`=~Z3|=UXUVMSx zKr}(Vi|}IMw7k;z6s%U(M9tuI{av}Ia4RkO`Qdepy`+3NLO56)2aUp!SnmDms9~86 zorz0tCs)AC(@>VT6aG(UtOIbjab&+&G$m`fjw=x)jGl4Vp)4aTXBcsMiICby&g@3xUJn)o%%yGIH!V!py#Em74+i(RMGs_3>EB^O zMKopK6XiQ;^Xh13jnaE7NTn3>=I7=BCRjdyl*K0({$%J`phiFZK*X$7=s(+d9f_B$ zR(qVYtV&xX#%`(B0F`E8c&Esk6)*8QW|idzNC}CH@)wZ9s_In z>M_`jlP8U8-s(8<++;I?fK<3wtz`!Fh0+40%orwuTij3!>9$FBEiSM6&U{=s%#zln zcvqLzxR`4}$?6O?OQ?qRMHy%CICH4#z<6)R23p99HVSk!W*uE$Mt!|$-Rh^i1qJ9F z9*kVwc&ZPn3U0sm847@iro_B=`1B)~jZJpkfaN2S9OYW8D?lFLG z+;|OZODw)5O6^-}9zk`=V#)-zgptQSg3&z1%PEqa#GBDXOMAQ;Iz8_wD%JP}`gK}g z>AD56J^wL1BPgZY4W>LK##xeMrp0FPBmN?+sAg3msL3HgyCM`w8QLgxNJncs)w@p! zsI;|kzYy3Gny1+sF^cn z1|}0{7fW#eH20f64{Sm)7A?-F3qDA^rHUuGzT#O5mprqQm_i?bDS!NboacWpS~N*H ze2JT_PY71n^j_r!4`Is$96Jh|bzUuGLP$BW)H>Ufbwm>2!0)`}R$vp^9aou*H!oOa z$shU^E7c(;j6ldL>p#Jzny(VL9#rVKurBb}as-}rnq&?cpIjmY)$h4{eUSrq{oa1r zd5c*HgR!wXy`PN|l@mK5Q1+B%RG=AN62DZwX}B+xpRJ?7LpXB$Qk0KGi`sJ=#l;;X;Q@n<@Z`T$yq2da8`DMWHok%JwjKI8c@@4qR(aud ztJl;qThnr!HmN1@w{1EEfyJL2%Mv0?PV5pO`DlD2g1LBpiTH-27Fo1Ja^Pd=^5UC) z_15c2oMZ{jKgU*6rI4tRs&xt${lO5%q<-K)2R zXoy|AR_p|j$Jy3;^mvJ9S#b)@WPm;z4T69N{6b;nZG&j4?>Q}o$%G6L)K2004Sb%C zweZ1zf#~o4j|S3T|K~@T4&hA<)`tE?)G+=~c%4^A{mz!3vR!=}eL|veY|tvJ+7%HY zTEP*=a*zQ}9ubBH-%Wu;N^8^TP7DnqrX-CuNtV7y~xD zv%b;XrdBy2;*`(&-S)*U5K?Zg!WtLC*O^dvZ&Ni)*XvON0*2P58aHBeuTI#2kK=PS z1*8iv)vUY5SuSsJZiz$yecM`*qT6B9t)(OtEXDRT4J_X! zEmsyNje9|dsF6urQhDBY6>;XMXUog2rP*jri<2Z`=!Md{mDvS@*!l?EFNi@lh2_sa z7hz?xNQ(3?h}m%QZoyqyO}~I3K8fZVnOzkT6bkH*&@`Vt@ z^K1NO0|WMW^?>B!Ffij_YoPVyjuG~T>gTQIC=y{&6-&U{+-B4jeT9RYcXzE1M-P7r zGr9~4NQCVgrm>mt$-8kRZtzM1$-wl|H9!#|cq7VD3?xw?Tq3fa*^}f`v=r45HfjWS z(1<9X#}(PXR8Xm5AX(Su*bM`&jqIE6*Y+4(u*z%$5Shiq16_u zOp%o^(I5v|IEkTnk~UD|1`dKX4bKl|ElsD%FWcH9okJT^8=k!JKz%$JW z4{eF2%L7(dKJF1xW00AB!N3_#u5fLg02C7Jgs(MS>YsjmrYk;isa8Laqj&wfpskE( zq86d=`82czu;_bM>KEW~+%g!|8Y_S|TR4M)W%v@C(R*(N~k?Qt> znD>14P6EiWMwZwR3**pb(n-~|kkMspfwGwpUC^4{19pNT-*kF7yFL`0YRGb;f9Nb7 zRbfzJX){mr&A(LDFl7AGNows7_`c3tDLZ*rv-}Lx*r!<>bO4v4faVYcjqE*d#m~3!X!h# zK~*muM%8EWe(W*4Sr-7%T{k>NpMBA2)o}^#>qZt*$maR^{yw$ngk$-w)Uvd;w1vM> ziOVKEaDG5FPZ$zWV$b_-ReMRuqX0L7u|~&ZVs(jU;GrlQS9IZFC4&3h?YB@@x1B{U z`B5?DKh0L$E3*CyuVMa-qW17f1G?YeGv+3?`QR4h7dPI z2}YlXGRISubyuF*b^>eNrSNU!>#oV^gi%1>BFgI-{eJG3!*x859u^9~9^?mA02BVK zlsapb^l>xePf3V$;hljC@pT=(I4e&j*CprX0-(WgdUVbR;68=3*vtag)1gwn*21r6 z|c)iR8%mtXtZ#Z_EPq6X%r$iXV%reuG%1uDw5zXXn_e9?OyKH)S8-`LY*F& zkBv)o7_e<4{n3??-kTP(`D3|;TNPgy+@4)RKPoDD>~7Uy8&p*VB8J-}JxTqnE>gin`ER zM}QOJ!vcj`t8G<^f-rl?(u`hFRw7_w@8t+%)sv&-fO}mn0tV-rXONxxR(t>ZKLnQ@ z<2-jm3b_f93AUYF5*lQ4p5M|xX6ljmf;KcFLm!_1HYhrvp+o*jB8NDmbryXg%}`<7 zNLbR#9AeCL#!%_AO*|-L@U**+qnlhFgbaCF>JGp!QesR+d@m$gBI@pi$V$#d?rm`F`INg#uGt!( zKn#a4b5o;QxP6pu@~%`)zA+ zp|6r5tnt({lgLq9dEmEeWg=_~6}DA$V8a|5TSxL(&>F2FEti?n((r9g0m(EWAwNwx zCV_k9sb*AFRrvRnh@gk(>!}gPU!kxfs$&%%P^PH>E%KbTkxR8g8XoI0@y_2*oKqVJ z?ySEXH<9eSDMNV@>^6)NHCZ425E#YLG9DuRyR(57|<1fh09Z2 z^kyY&MS=+Xf@w*gMBqb@wx6QZC4uZE9(N*ZDc4rF56@i*o%E@T3Y~ByvkrfDBne5Q zMMe0*DA-$qmL@R3zyb1A31PpH&TJGmyW*o`Q26LGJVq=-lw!%QC6$`77aiOM0zad* zWCxLr1zc#-lbr7myjbm!UkxWTge0x^$Akb;`*}K)QPy^b8?cO22|eW`xq=%Ivbmy! zLioPfU(_vFIzKCdjehoH(d-ZUtedqP?NoF5R?isMKYm||yGKdc*z*Nk#*X~TyDb5m zIpE9ASxA4aYZS%Z#U(LMRWg)4b7WaNq* zLdH-!OSHAE=s&zzD-N+;Wi|?f%IYF->%9-uHDnhZ$A5Z89oE*jbQrPcNjmr@tso(X zfL~L2Hefd-h$x(p6JW$Y*g39w=o}z8ePu+|;I~!}+`d=hRvh~lgO?)prs9h$!JlSy zMnJ=q$i2~GHQH5sJFq$B-9DIaZ;?G;K4G@s@y1zXzEyeMXh8b$>Sq7W@9A?BHpZU< zdQWxzvld`T{e?Q=^=?%paC_z5UQcSOaFIbF6>2YWf;jM4IAVtk0zuj5S0M7+Gi5o+5Pr+>g z)ata?6U`F;EVQ@{bB3u;^wwXU=pTTh$-Y!s{5gxy z61`fq+HQEJrS}^{H*IKI);zJPt;53?l!<2U$fy@nKNO}&($Y3wji2iT74P&sk5pU> zdiij_aJ8K9gQ=momjJ^uHlzZbuu$#{@78g?eh~;{7pUmv-pCTyLmgZcjznC z(sO!v$01d=bG+HnQZTi-Iw){H6^VK!_hF!0veO=%Gc&qc+2lbsI$za+PU;zvMu%oR zszVUB1Iase`5n3QH&DR}EkOWO%6U=nT{y2yt9juFl${Hrwy|b5Tr9#y<|hkb4Dq!+ ztFAa%4)JgZK0KhSas;=lXe#YTPf}roy6m~HWJ0mir*$Nh!XTf1O>R{BSU3sTf6Vx? zwa0sBD<UtEUG-Jt%9h}6oZ1448`E@3g%qa6-OTr4 zU4F0pUtYrJBy8?q6oUQ!*`+yQ)ALeTvPx?Rpf8!Vr=qphLKG#fkU;)mPBHW}UWf#h zAW-dKv--QeKZvB+8M;~E+g+ZGJS)?{XyyX>ynR$Q9>RQRO(n%*ZV&Y(i1HXbkkIwR zO|G4Pv<~?;tr;5T@BDRCy~AFWeB~4Wx*61q{RcO$yAs+gS}$AO)}73En~WrI0j|a% z9lGtz^_j_nkL>s8rk}KI6G*`)?xkWA+TW>lhZK}&<&glL+>Xu4Yh`&mE|KxxItD*d zgohVRIh*A_+I=?N#}_{W?k|l8<|k*(utjSl1?l@#OQbaNbCv)(_{csgr z6W3_nlUe0jAxQehAbu?D?=K514ckAnc)^xQhab3kNfBcN;Q8*lOLNptfi0NzE8gLr zhIBUiobKxwL;3=yLdtFX$;@+>Nj7^PZ@aHIg(m$R)qRFuOjHuCnHM09a zo|XtPtTY7;6f=;D^y(2qk$Ji@?OW9WTPBU#!8Lm@H(1y6IguYyG{IHwfor+2aoXgU z8GM%-T{1k!utuBb+Z;SE^Yww=rx^=GjVshxwws^Xm=l;BHAbi++1I* z{r=E-Rak6Tj&HhAHE!Br2E5H?bJH?_;+HeXHIqrLcTo9I9;Kl8pIwsbv!}pkD+cx4Mh*-M7+8scDNuK_}k=;3r{$|LF zIUzPAoxvAI3fS&87L^-x+IJq?_4^hVDg4QxeIt;tY?<6J66a)CgWg^|yescBU8prwU7=v(THX_O?`$}%;*cpQFxl1yN zfs8MW@@Y}|ap<1O#MnD8wiLlxP?X55=(wx~B3w7B^04sfedRWgZ5>*(aH2h&%620SMsF$xcv9I`dx)xaip9~5QOHT1{ zGUUnhL6CnI2pJ`Ei;k0>X?*;@RyC)1TVQy?K8f}gEd?aHFsGUev$c#rG4;XFuj}Pz zV8*qa?-e?VcYK#qQ{4!U)xcegvYV(=)J)eFpc&W+2n zN^D7)mt@sIYD}T>hTxS;;DsS6hXBRio3ABm4r1fmogbX_Cr)GNU~PIL=nsJlM>omR zle_hNl$3Qcio|~iI^rEH8RccSJHG-kl;2#V!k_t`5f=+69fvP^+%St9C3}jw@;^PT zxf!J#=bV4fsnk~gexHB3H1ueJVIYYm=Q1B%dsY^W(5!|aM}r4ONtl%HmqZZ-nAYOj z)6r*|Aa-JRIz|A(OI;i`W^?#H|tVhAVKzF%aG6&7+lLD^fI6UJy0+t>2kG#9H? zqsvQ(+E&_T%*d>5v+^$U75UoCFMBu4qRZzHdDRIRax1RYc(<47HfRy#kRo6M>Yvqp z6iz3Wp_0O-0!$_VX&k@K*N9`LU{cc^ua)nz@v4^A3swaB7@t_V0ak9%So;5%zDK7+ z5^4HJPuL#K&$AjMqR55H`JL5{`Z*Jds=jg0LGvG25d3i(w$X zP4uU@bG1G>EygOf4_E!~TpO4Z>_|dG;GD=KH?BOdZWbMDUqwFCqaE&q6FiP%$Y!jQ z36)S^ZS;TEZ;iD+mo&!9;S6M|*}JniRTV4UR(6Du2-PKNow&WzE{l6B< zpNSLo=gPp*(yk`{Ho`wyFd3l*yQ#a<;i_Qasm!b>5q-nsw@>!W@vc5F9OtLm$+4DI zL<6DFbqIXh@BxhP#V}Ft{}kxI+Br{hT*Xlg$;~b{-$UE(#s1xs^42?)>)q`{zWaNR z3DC2nQC?9P^Yh!HT+kA}jHIGz?b!2)cps#ju{yqoYhce)#0=S#@FaisScTv}1Z{$0 z8I3n&d3rQKGx`A?9eND1@p}z)GJ4au$s>Ak-$cZ*|7+KG1sh+!aveK8@!}~FecaP~ zj<2_(3-L8U6o8H zZp%7x=NcVk6f1o#fiy0vYNsIn#N$7N21xHw3%{ujrczAHl9|7o#p$&WV}|7D;Qm$< zY548gCI?o*vES(>R**jIIgK=QoV$tZdV9zk8i>w3(7g_(F{S!GiP#%2T9^ujf++-a$Ax;Gq)`@b(kJD+jidKv*MrFGqW&*Fgz5oLR4OdL}EpGf_fi`Hviz7SnsLhNK z8^obSbompSppZ%;YJpmt#*sejlnggb?nI)C7!w)$%KLwFZ2zyXi|A*9-^dB0xAL;n zL4R?BA9ALG1Xh7QBwtYRDf5i6jgcJC;~?b7`QyCE8l_CN`DwI%{N(gD2D6|qT>#^M zPG3)AHruxQ_n{diGa&W9oMBq>zaNYC@;~fkT6Lo^)$5bnyX)(tlYM7U9u52Fys{}d z1oscTiN2V(sqfyi4h|J6dexU*_cA+(wmx627nTWJxA~1)osb@jAoA7KxF#RF@9TYDRD!peA4bjEZUg@BL=g1N+Op*sZ!uZvs|4{K?h@TGy3uJ7bNX zslza4x!MDSH@^i8s@TywFa=A>MzV_o%52LU8v7R|CmD{2|*=(&+s=Ho~T|EG{E|Mm3BK^aIqJK7d zP&KQ4B>F_^k})6OzIcF=hK74{AM3UGm!Z#fR<>+|Jyo-P^cS`3nYcL%0=QrfWd3|C ze-eY!21S^FVZezM*<%^tym?egDh(7UM6g8~X?>rU$YRG=U7Znmr`$tt`k`8O!tEx6 zTSxAAxz4MdX3}m;F>%HA)Y44UfU;Q8d9oZXq5Zw{g!iW+#r%}Jtr_)d@1Li}wB1H+ zhR(=9Zf(Ef_CEiA;n@l0LP<)gd~f9+)Rk9Hm8FOIXGO`Mo%N@L%05a%e|!Sm5msC4 zC+VdRkFR8n8i6W>zF?I5#U25XHgc4{P99%k>i`z=*L>QhBFV>^V&{ONfR_@7g z@|hto+5S_YK*`*^yezpmV3m{56dQ?#_(w~}#TX@TTg=Rir6LCnmpQ!t(olXz-L-+{ znReR&@AePzR*c|TY2=R5Q-JTViQOHaXS(ib$g0gZbFkc zCAZRxy1>3rj>77sr7bCk&H{A!ct=$_zz)&n8VCvv~;`JYL6m2 zPxLAJ`0{Bk_oOea!cAe`2+#z+x8w6=S)kympf0*E)HV83lCkKT>&oX$Nk=sSQGz;F z;A-f@srcG|Ji5PkA#BkA+@07?v*ZX?DC36uyoO-NixJS~r>w(^( z646Ee2ZWQiA>%ob{Wv=ddh2r^W^ZMao_39k@Mo53$F{P2?~--k%1G@y>)x<`%Gvd+ zM-@wPd1-kWq*g=2;JbRIbH@BHgbQuJ9UWHKZ4J}eLy0lnsK$i z$TT_#yjXP$87Gu0V(Ye+C*atNyl{ zIOefB8&R7g-wfUOB~Lp#StO8;hiH`Sj$?55*G)OrMnM&Ra#pr_jbbyZQ)!03esGB| zBT2cftLmRu;*)^+Y&YF_Ar9O7%ySeJmXr-!=A2Or+!&;JPuJb>T}KBCC9M+WP;Fzq zktvGrI25MG3CSDK)MgCa(_P`Jyjcl<;*==+WJNt-tPeCOh92FRKy%*u1` zkW=0&p!aYk;LnaLh#w$wKjP>A;#U4`)8TjzpSRb5NT8rvW3a4{cE9tqwv|Qnl;Xb5 zkognExgOo+6e)j!?fY7Dt|NPeg{OJt<6{+d2flB@df}=^J1xHMYPXHIM=sB84^|77 z0357e{jfyo-Kzf(RBto%HI-~$gJn&iAi@hx_dU`PHF8nz`}9#rIFsSVBp|CdBdAYX zrI#f1`_X*gHazV(W81pe!sam7wr4nxva>&7gZDJ3 z#B-dV2cto&1-94b7Z=1%&Gg6R!716S(fQIKn{JQ(xEr}2&}E2!d$uTOjG*WyZt6;oZPVBHFhjhoNQyiRclkwkdXDMxDkqW*8;j#6*=v}8bp4G1*&Q~kZp`nEq@pFh{{Ij- z0I+;bMIEmBH8W-Jp(MI8PDmaK?(rM#_Gu)7d1Eat z#vCv*cUcF|Jwb7vLnFkFX)f@}?qmLE2XvBa=jG>TFAq^M%PV*B@SjBuJwvp)qje$H zx?YVV;+vp_paKa8#aRG8(Oo;9uA|V$9BlItMGS2otDHUMA=IFKZ@F0M2dPa>9^BQI z-;~Q7b>cbwOlQ5d()DY#AA+=<)lDDSDm#asXHIfF6bBIgAZLV7=i*deC#jm9)cwoc z!-JeJpt4!z@*cdeDQ_VqLODrEvkZIF_O~YsI4`1l?-#_o%qh! zXhn0hX@}CEBi%FIfYJsDTsd^SS$oCFIkTRsy8i%h{AqNWySZ#V=>##SA)~=+EiCS> zl__zyLe91Uufj^p@Boc^^tV2${qAr(zWl z5MZyZjE-2CpW?c>{W{@yFPCQZz0+1U)`nl(Qr}LU;xX;(k7Mpva#d3_=`+L&x)*@C z$-Q4;josbacVWDq){*=cTI*Y1aAb2;O%%=`hL#;JZ@dMJx*YS%_3D2`on|=C5!r7S z3VIuC^`>|abi(Whf?Rq9ne6_~%aD@&cZsz#)ku4$%G`!DbqkfqWCXFr-e8qF&n(`V z6Ae`Gw4JoLu55ZQx5s8VdvWE}_vbEH`ga?QwATI~xJeW+#AAiv7O)ItW2e1Ua1VDu z!nGCk1O~OvbB;PSs*3$`tbK|ac_wia5F9a#jP3jt9VPPXS7n;bPu@P7M;0Gf4Q@_% zcfob|x_q+YvwrKJK5TL1mtPIFvYKn7~Zxqe=6S)THko~UzOJaM?;q<~L`>#tC{ zHqt^NBfDz=Y4Tj>gRQN(+$v;|x}%OV#*@oY8!d%`q27ifkoxS|-b8FyA6=~HkB0hP z;pY0QhsXiJDhognRmspZl$(5NNn1fHwT#~oLC-v%Wh9($CZR>bj%&zZGhjwnCK6=Fh`Er@?y_C<;(d|4Od^6&FmU@Yr++OX} z?j$|ViZJG#2Yo!0dS-~}i92O_vT5gg3RwI5iyC<}>OSCw^#1^E#yK#26uT6Dq?HbC z?l@|Nbv@H;~u6*gi@HmQL=pJM~JH#8b8Gh7U>a+|o^c_J0tJoWkawsy*gcQ_+c7eP%cj?^gPAab3mlr3O{YB$5c)BOAEm zs)2Cgx>{lcMbo>-I0O2uomBO8t!$FW=Q-us7|-0NTX=eDaOpvfhnaUZjXAbv zUnl%t{{T)?PotbOTa0z?()#LyE8YWfjU;3&9m3Gc&LM`|4|j;ce1)R}9tCwfoQ;B( zSG$`*SgSR%xAEiayOP>@pvz@;z?|}3?w6OgT({~v-28j8Pn234ODHM{?%3D4yqb0k zd)cKqHMJE*r`2eUr_A$APs6tgIymkX^uR+f%VcYVnny*d;Yv|WK+G@#_iH}&)9QMT z>}}hhhl{qxl$MAvIpH7>ePTcu*&|Ut%W^TgDW9Se=Qohb@#E0Tahc+ZkR_4s+%#zm z4cd)UD0?!cp+B9uRF^+LVXH@ek<}8m25w&P1{XW+&al6T>1}O(rclzgzULj!@LuQB zczgMIxuvJgXq_V?8l3C}U0+)Sj4_65Ip@)R zg|C{b_@#Vj)jtI`8mQlZ;97p=0|e&-mj2{e-wGn_m?I#9jz;PpbpCuKAsIAGxtcA>gt_wJfhT3O`7 zT@mH&1!6r{d{q?;o!v_hLU2{)FwyTti48dkK{p>PrH*S#QZR`a$G(P$&~2*+-C^A98v` z)6}7gedfoK-CTWM6rgX2{%a3TGdc3Y+g`{VJ#UO`Ljtn+O)<4cGr!)NOQ`(w{pqvG zX02A;ASK=7Rc|AOp_RI#Ej?8YOlCP;7IcJlOTs9|+oc{#Ah3oIK>+5Okne=K9f9VV zS3n>TN@yyDG+-k@AfO2(Zj^!s$igJDIwNdl-pZjZ=DkI3o!o!e%@~#!#c+Z$DP;%N z?-{lM>6|Vb!5y0MuNc)b4j!;o))xXpv}tBKold~Q^)CVNto%R0oIAp=wbwf7SS1-O zERFK9-n%kDAb?2+k;1xJ;=J{C;v3A?IOs1^Pdqlpx+r0l*bxk4NFM%v<;$E&aVM!= zyYQW{-0b(}Q?ibrQfV%zo4Xiu^MR<2T{*5KT5?qRDJUR1qtH)3C)GLvYqK za`&ERDzmez%uL|+K~P-QQ^@SNbn3T_{aZ^Nl8)xl3w>-NuDmly8j0C+dJZ(AsFs?N zlFeCG>XJk+4+RRNiY8D^SY$1fG9XD9*yU&YTf50v>Plm<1=10ZfLg!6>xrP@j@21c zQ#^6QG!BYZy_&T&k;wEQ;c)MHIWLHR@UpdC+j_R&YwZuAf~qqdZg4)fHyr)taK-G; z!&dWR;?5GInu=#lQ$wh;4#Qe9Gq<^T(5z+bmvBnYg(%%{;%j9KT3lzUjywV4;Hwsw zvikU~Tz_Y|PYb4wnxsfr-e(F$d%p1_XzcT3hPEgO@Y$_VI3}i9cy5NSNehgT2WP|{ zOC#F{5DG5NZa04P#M-)dGMu;(sm{O{`K?E1{=WiQ2pRs#9{&L8S{GH7)vS%LiXie1 z!;v0puKxgZnmP&Ocxy`=k-su|1#c@!W=gU+1fF&`ykKR&JnC&_97(hqK5(pp(ZI(uiHb(*A3PK1RGP&N-&%ri=2m(}X zFK}Z{&am zPum^FtB3BKJ6}r8PyA%a@r>jb`{D zB%WJML=&)vIwwX+>yd@?_UjjMbE<bKhrahFPPIKSFP)3(3wv+~sMR zHsLjt6_gKar-{+;>b!C!co4l^*;{HaU6|V|jePE^Fc!Vfmr!$>ak}HZtWQB#!L+q> z*9TS4CCrISq7mZ}05AYu3-D~Zcb%U)l*yA%RQ8dcK<*HYM{g8$cV5M4L^8TLDXL?Z zM@9(%hPY=Q87ocrE=5yI*(?yvG}41tE8~2CfBOLSKx@BHTT>kfSj>mk4<2b`TCMa|JZqxkHTiewfRhGJ%Cp0;e!1#zg za@;SR7aDic*VI!;-7ktqF~h)vh4h{7!|+Q^gP(pEfIDcm{weKKi0DAl*4Iepk0%{N z?kLS;&_M(N1QoXnSt~e(`%P(~4ezITXfEVD$_kVhY2psWq<}g%JCz2u?#(1YPR4_<&oz>D zP14^aV&z`i$l3&TFJxL<8;>V7-&lB+aZbskfyOoh_Z&Aoyq2qssp^~>$t9RGAiEgk z6>ZAp==A1xxzi0gn$`{6d;*nXrM^pSjNf3TioLFIErHsY8Ak+#JwZo?8-? zRcE;p94TB#^9rlUKZLcR_WscgaA6T#Dcc}8*tmM6E@a+o)?b7d0baP1g&~sPYM!Fu zLoj?yU>cmmvxo3rdz-U9J%p0}5xi&hX`Jui6wBLE{4#37#bmo*Ynwx8x762EyOeM< zC;k#DZPk<)4@0@p)KN;;0V5`ivFrr%Ag_1!Ey9*Q8sN%NVWndxm4VeWmuhp|Ksg7% ztbWVs=iUnZ8!wj#nN`I1IPB4$bnRxjvHSwPpv0r133H1=3)>YM)niL7XQv*HIJ zx*XKxKOhv$j3K8Zb!I=7Zbm?4QHbMf8pezUwV|P-$OJ{m%O!Bb7~}9rd@_t`Eotxv zC2QBDV|!_jBch7GC}vIULxX=z_wl)sNf2p z#=Y6PR!C%mc;ha0HZ}v##b>K2C$~yO?C)u0IM<};llm3bxO0Ozechc( z-09liROhZd)zz1TDDK<=46F^7S3Tju+qJ_xW5Ht@)$rriJFVk|rfq$auhQrV)g%Sk zZ@I&mK4En0A8E98w)ROBG6~%652V$XF#W9c3!d>e4OCcaO!O^#TOM`BC2PG`jE&G`KzYe>7Q_%4+8^zN?DVyG=E&u8uK{LvJ-*S$hB+AH8I%Zf|kHa;~Yb=1avyrCr1UyZ-e_U3rc$Nqa{=Zd}Ue z$F7~_l&mai`qBkeJ$#j}*DS@^9S+MjtgSB@Up+peo|-u_$mG`I4nZo_ju-@ZECzIp zw*wrO%9dSJ@w>&Lap&c%Rw3ze$-gxKNI~YCuB;GI%`tbd!a~fORy&Jp4MozbZCMTV z8HIn@TmlLzn&)h0c82?hHHo-9yGsGdbF#V}k)n#xPfjokSPA!)tf!4JkG-``uQ(%L z-l^{sci@K-ed;l&BYf?R%BZ>DDdE)E+$E55Xyx%LNv5svZl<0n^gYBiyX2jdUyW&? zsAN#a*|!1CI|ZJ&-DIdgO7QJI%Uj4OF*2^UQ5+0=&x1!Lre$?I2NH{Y?v`;J5JqGT zl(pNI@>9>|SVCKDw+?dL``#i1<^vE?wtMrWN@?GwVn>GfR zoCKmQvBYve{I4@{1r1*faU*N#T3Z}*i6cG)9z}NRHxWZwX`;N*&iScyp8lPm-Mowv z2UDJfbbhVFc`iJ-dalPu$=fa&dyMS0+(6kJ+M4;A;7*`60aV;z^tTt)4*C6D@shUp zHd5VU6Vuds*~peYW5f_uJ;L21C|xPTrZN!Lx0atDdg0>c*O@fs%Nbb7XV=Vyvp~*4$u30CIjx7v3Kg24!wu=>XcLMS>N*vss_B}o8v6Y;- z+&*iLPTRL@CmX7Iii1yZnkQrlJ$8BcsdSZ*)zQUQ3&DFEmc;Ek4KxiZ+8hGhrT@QUE#9tVcp;NB;HQGQg+NQrY7reWxBRJ z89f=m7Y6P=D_Khoj;n_*ZZXl*HdYB~K{#_U$-DUDDyH$?J@M1a^l^_^Ij2{7VLObYzC8UfTbX{Y!yE|;$E=|ee zJyMyRU$6rm6(zfmWQolaS)NOI5?bX4iTI+X^5GE7G|d|@4fE!^oSc1cRjw{csrhhY z$-%=eSbJ$dEu*KaoJB+?iO1D(lL!9*g$jN*ttH*XJTViH4oM+m-8yW_NnVa0O6jow z08ZV*ecN!C}PjyU5rjMR0mR^8OcRA+3I*_e={ z>O@>sHgw2JNq>rOB@rbJmR->eNxYOYQFe4W9;DDE43sU=(CyLdOyr@cqHOq+era9j znvIlTqRNrpA~NmGHIva1a0dk83v_qAD0hjbka}f3Pf1r&;M#ghiH>oI;&ZwCf}1Cj zUoc|}+;#&c3M%OJGNr}B78bfkV`c-83Gz}dwmZxeQVNSzWEo+{c@buQYd2QO*=LSA zX`Ej>1anihj>ytZ?mV!yF`0N>+-%fqTYt7d0R+s)9Cv{1HdSxqdJ_6)3TL&l5+%vr zEOqnB%+GDF{9M{Frh_n60&OT^x?nEE2@rXCr=`f}JN5 zQL-}d;UUmm(9m7=HSvWJF@>UXTEch2i47T{5_e`s>-#y~SW-Er9oz@J;T-Tl z%dNDw81Rd>yMd9a9+(qdQyO7NSdt#(FM1vCESkPYq+N zYSV}*A4%vEIJ^f$=Zv_XK>q+Tq*|$)NlP6(c_r8`^Em3LSZj?#;jNYKbb4cGbF~KK ztJl>UYN~Pw1e|2@3$4qJSmQ9;vB#9lE0b$SNk=CL`e2XgH4QE!2M3{6JYg6d4{Lb< zkXk(V3p@DcnAoA9jO&%MeIahm9t%sv@jvk$nw{_ZBZu?lhOZ|)GQ%lypDzCZ9&;(< z@TIj=cNDGfF_&&qDm1!MM<8&Vs+2XGVOLxy^lNn04mi;nJFRNvk!;5oKjRAFU3z<2 zp6(N%e=?)wLrb{k6IS_K-?+W;0{3M1jMP8*PE^cgb7E(#N%lZrMaPW887ITU&pzvI zYb{kjh&U|P4F;yA*=WvpJhScyP|t9+#Vtj_&7MhGDq0)NgUox@Vg8yS%yh1NQ9Kid z64sxBxsL56#-=)&si~|av4A?2Lmh8*c+|#IX_>%jh501o+QPT1pSqx)rP#!5Bm%m8ygiv}e7hWcmppC1 z3dpYT7tIqG;#-lc#EyzJ9EJ4`mFF#qhF2dxpL(ja2l_Qjn-P`oepxTo$Q+VeQ&H%W z!KcyWn2eW#HXa`p>G5%6hV=9KbGg-ZWQL~X_va%x3EIbKg|YQ0>FC2H%cqR(kdy;m zagV%s($kVh+_{)q(nZ|U#UF!4ed?bb?c9$UfcwI}Coa!bNoa3*cZ)x!%FOo8KHCeQ z{VRa#aH=F>Ye77fol{;M_dab+bG=sM8dNIH;`psI$?U~Dv{gd?00)JymvdMLSlW5{ zEN1x!eTu(*?wC3FidSohFp+_}Uo1{)1ck3|c5_a0p5af8f+ zE`AA0 zzS{jZ#Zv}7ZtS_f@w$q{@a)7@Ivqu=sED&Z2OBLnXA#|PlS@MtWlS(WLg|Ea7RM+9 zksVeV-8-Tzf7WC9dzVJ1xJPb|I{xX>Lm(c{Hs`9~>i*|v=yG)3o!_arwA?4RTw<)J zacN-$fyu+bt#1`nvF(QMK_ls^4g<>$G7pXyGm8HJYmy%FI+kk3v78^0v!=W{zAX&| zMmNg&j!qo6yB_}l9&^;DxYLXO02jpoxQ5PZU!&Dv2OV5-;JW_+XZBjIAG^nGzp|dD zhT}u&RK?mKq;1Fobo1oA>)H*tTxX+`TiV!(1UH$_g4J*>(rX+NI%=tAne5Wu_1P?M zfL*?B9vr$oPv)+)yq|yn0QHXV!Z6ZZtvCK3a~Swxkkn9^#zg?i!8u*1B0V zJPZx`4VGW9i=&;!Qbj4it%3U=5x7{{1%M*$5D#k83r;w26#68Z! zH93-I#6jdUtmODD zuNUzhy5mIoqM8>-*xepj;7RcWtp{h7%%8H!-yzlLam>d9_Ha-0UG5&Emx9h*ea{ZB zFBTk|!n_?TrkVJ2obq}pG}hO+{V~uy2QsbaY%Qvo!p60(k;Gth%0o^z!uzRZn|!8I zR>N$kg^{_9U>b%q)l||(&q;JLx1K=vtIkTAy`~!RCN?BF87f<3noHq2hk@9I0l>82 zj1aVx?$4f~lqxAL$5v{yD5X&Ws>ZaD88{D9Gs%n{LXz95;k7h0K`z~1rNu--u9 zbp$Fc>YP2xW5j-~x{00syB55gVyO5JMP2KoCI@JD^iWUxn+x!Jo430B-b)(%54y`2kB|`G3yL&dadI; zZ24XA4a^9x^jvr%1QjkSVY!to?;j#ozJyTlfd ze3y~?6RzSajxc@px(OW}sq#ZjFx%2K9DG0@di6yI5yv$PB$hoc(lRo&V6Ea#1>yVs zbLYC!wq0>|b1;muX8yB$e8S;fp?(b=pWI(1J*l$Ts*YV9L}k)OdQxOZ!T*J5}KYy3g_=2)}ctP)G zbAuyp-vAbyhWG~IHKo_voDxAt8-NkHxO9N^_bmLYsCGE}ZXx_)$cyQK5V#DuH1hy; zJEO`09n7-E*0WZv@B!II;g)dCvU*y&4@flu+IzC>Z<8Pl~X!LG+<|-QaEm|qKfrPH7Ufp!$$n` zvY*C0DRkkh#MBi~O7^pM%p*EnJXB5*xtlK(TQ0$kf1*`KHkS;8q=UFwp3&&*?`6DI zd+g#0_!#*r>M7=Ykjpz5?6+~Hr0Q@Z$S*HV8={fXy``^dEpR;uUg_YQIJj^%e`xlu zABuXmiK-!}xK_GXy`$PJ9XjE7mx*5-x7yjFIb@SL<7{&|S3hF4YkMuSUGJP#9VL>F z(yE%>KoS6G$lR=w@1KE#SUA@3=MWw!fNpx_+o?T`Z{bl@Uuf*k#bBzCP9&ABGIRxi z6ZfvK*|V5<0?DFAKNN9|ot%2Bi+&o_zemH?@N!F~n2!!YaXts*ZY1G~_-C}%PE2kG zx^v5zM}coFD=(4nc@85d$#K8Js#70j^0a3UtckzHB9pe-hgum|Tf)+ffN_ptq@N&H z3hj@y`;Q-2M-YI~TPevgQwL`!?K{M!9nXU6WUc%hYHBF$;JP|lKss{ghU`zvE6ES$ zz025q&{g(JX%R-o$0ZqIVCBeyyi>%`HkXL5RIC6t8o5{i@DCwME094206{4GAteAo z1P}qTQ96eG5>bFpAW#wtL^O3Ww^J$QaPbXOC@fbR)@q6n2aAqAsH9`AkYwF(c^WvgN5j$-LkpI^_mKM@fdHhP{#iN=(Pp}f;oZ_ zE%)MA9iLPk?}e#}xN?$X^pfEIqX0pOL2Vtu)Ve`!bfBLW0+^SBAN|#H5ACp^$sn(5 z_+INwa>zUO{iy!{qOP9VaEJ4>Q@8yE1T+%g9E|qkbA96p2m%NofFObh0tjIQ0Zg<~ z&slV(rKV|g^Gw!82UijRQ3B}SgMGFupKJD5W^SfL_e@8dT1YBYp5CFmQf8LXYj6&m zF}7TC^%^yO;cFQdSUu_mZ95snXX^-~$Un5KV}GfDed`W&jk(Qk8th?nPr@>aPsEnX zjVyZPqPkPdr^{FidmP=_%BJIKW2m&~m92Ij*I@O*2+c_1dg#roGB-<_H8?K>aV+Ae zh%OVgp`uEe90&S}_AeS$);_~mQ3WkTkuo>9y|K8rk?Oo+Vw$3P9>}D4E+pw`Jyx-C zFPRBi%Jdah>EH{WC!&wYayn3hA z4Ive9>&-mO?;xkEnVj`92V?JCkG1-Pf2y}|&ALfxE;Y9}LqyT&Ees{)z;(bMHQAea zW8ptY04Jg5SD*W4x=&ZzF9pYDgw)*o(-@6}lH`RDPU}GLj(YNjpwD8rR@q{&14y>lc zz4cW~oWNQ=Z0C^1SC~Hte%EZgVZ*R+?UD(OnvpF`(qK2$1OU$w=1_1`%>)lL5dm~= zz*^Sv^&IXX2FTXr^4)W+70`PN^yhJjf|Az_&N*Q%o(W}aey($=rNj+D?iH6;X1C$q zdsnBL>u|N(X&fCrL)^x_+To}!c~=V@I^f{Dw^w6Eh8XwIo!?&!@byIJxGHrzvtqTK6<>f<{Lok%AWmY_t&sYak$k2m(Wf>A;D@Ay^Q>mw*Q3;4Y8D zTpw!mi*-HdRX5Y2GDjslM6V}s0A`>$j-_*`Fxi-0$T($hZTC4Kt*5Gu#jk=g8sK*v z@IhLNs=wIj;HQ$Jjp46^xJ?e^w+9(N(PtXA_X@~or@1^t)`vljzSf)*@mPWU0)WRn zBf=r=B@j!12!&HZUoLtpc8oL*NvbGABlAz9f>C&C4V z=}*A&Q8lXlZ!KfH)39*PHs-w62wmgwMz8c+&C(*rKT}#mTg;_C9s7iZ&Z26*8@NR3 zmcR%V4nD4f$*0f2tp)o30BYcCJP}vU`5x8=#>V>6N&f)96y@sc2M+M%MaE_~TU(t2 zo8EahJy#C7MRU4bq`q99Ws-BA?>G_o0pJ#==CiH%UhxlU{7rBU{8JQmxH0PFYkRyp zY_rnwXBo`)J^VpetEtYxX*hp_ZWVPf$5Bf7B9P}%9D>)?_G4c}U#lLR{+1yW!55|%1CWT2jB z%+}?tdFHvTWxk=4cdTpTsA1axjsF0znr@yir=!hOncRc$N`|sVKTs~u?JO5&{7~z) zE3LKUdUZ24IsD^%edDUyGag9Inyy>;N^lK77I4i){(_!1-R<$2OM`uB;vUmU>PF-% z55ZH5S%2WjDIW6bsyO?J!6PRF)pyGaqqGNlu4s5ZscGsukCNv-lXh=Y!Cj%=s%T}+ zs-@;uHQ}2>TGu@+FU3TD` zy}l;5T$Il&9&H&P!FY(&%prlVB(j#I27}NOf)~H1(mCTxh7Y-TCylAa9}@8m1ZBD) zuYNK6dHxBPrlY%dFHcE&>~5ybSmy{~u6{##=?-bYU4F|=XSG`+rMKH5Wg{KxwazX% z@!fGB!Kz_54_v9WV6wD4jP!Ft?6l4mlI3vV^%GSXTc?=^(&o-s|@HFl72 z{_ae8?f(D^F~ye)w|^ZqJxGR*SQ#X5VD*kh<390)Z;PUOZ|^)&MMB3uqM$|q<(!|A z%3V0$R0Y(qN^{7(=>=lrg^{?q#nL?ecbvGSRXdRI8>zJm_YmD{sym}r_Au?ChXa3d zPou)#w+ylO-I=%|ysqk4vT}8Fv>V?{e**w%R>0o{`T1Q2HoN`Uf z#z@^SXm{~J1`B}3C28f1t%ah#3Yum%IRPzh2w2*SelqA?pVcZhDk+>E1d;EtLR~sf zdFHBknxKj{zz*^|apV;x2NBI11a&O$0Q$q4JI#DmQVAh=a~j>Z>ZEGqg>Q?}uW?4u zOGCjv;mvF7Zj|(-wZ~KXG%Q8Fa|7QXkhGTLZf7W@sFCD`0y2H%s?54u-EULPmh0}= zc`==SDxuQNJHsOfeobiIO}5!*srBb37I}@_v@{en^l{=QZh5%werU37OG99l+GMPF zjkyO%+@I*QR8Z5?Gei-F4!Gu1l}s$#!~zpdcH?!OwI-5k8R;+>v!tG&E``9nTWxK+ zn#o;O8(U1?@7HbvhfE)G!O}+O9YLg#$)l>xS5Q~Fd1Q8&9$ZM(AOwButJQS4^*HwB zbmija%Zn|?VB*VNB~xl3nUKj8t<={XV0d7ze`RIggf0cqwbc|7AEk#o9|8#ERh^pY zZv3BZuDZH6)IGmZ**d%v^s&hWGi8o)Vm_j*aorI>S;hDaXAG2%DL1u)40s)xGCfV)shlePrsr# z{WL@e<4PBKNx$%2U$f`e`?TM9k6B~J_`ZfT0`q&swI^k{38*+jF05#*~ z#g)r)S88RBy{T?=Z>D*3TbN#bQ{m)QMNK=8txodS9%CC7tCp^gN_kl7s2y4JzW|EE}OKp6vrxCkA%!e+{IqtBN zNflMfcw%p`FiV~54gS#Q?kjO}sdR8j=*Af(Eg`vgst=0F+31XNNc#80*<5nxvu#V( z>HFllWjwl>$!v!#R=FmqXL%`}Ooqk)!&&Yf7hShd{x7yN2-w%XjK>F6+9+@cmV~ zmf-`;@c`Ei+`0udMHNGbbyW)ZuW? z_N6)LI+JC{wJ%B5I9w5D%{ETxYF9zb`Jt&aT!yWaLET6a4^+@{8k0d}?ae16#U#6A zoTh@w9w=wcBPeIdG#u6DfOh=TdGbW(v5YcFBUu}Lr#qraveV_`2mvFAA~%uZ0?FMt zoW}BK$Y5#N&TdP*@zFxBy4>CvTU%E0=7Z|HOGe&HSmTe&H!s7WGL}b=Wz8p{+xVo( z98~92(=dD9Ka%FJtFH_KdF9FFu0nE{FOp2=N4Rz0Wxvrr8QxBjCDzplI^~T20H~E2 z{6}nz`jUqq)r(TVMhSna*w5{|D#i*uOlg+7I2v~)xW~a}@?kz4m~rQy%Uk(V0Ns(v z@QeiGINH9Ax~ZR45OCG%%cmh^WscDcceu$h{Xj~NYpro?BdV9*q%2--&(n|n9FvY? ze^yid+v;vFyWXV{=rx49Nbl-1&!$ti>9~T?6q43jsbLM@OJ)uu;IK@*GL~qcNk+_M zJZ0pQk1mC2?_Hu=E^@~uU4TmH#sM-bj$D@hBa<#%F~?KLsknjjjLqG^;&2rki;o_*7$2NvL=(u8={YNQ__Zm-liInc7Z7n++hO#Z1Jn*fa zcefw$8EfHc`{j5)M&>lsJrnSu-BV{{W?KTRk|gTlUxY%s$n59^#fCOh)$^ zK7aDDW}tqB-y~j>;TGOuE_r@g@YPx4(QBt@A6Gv$ovDAM-sB^3uLIz;(!0^<+ez{tv{;T9IrNhUF=_zK*B z?T;mLx|W#D*%RlUj!NjK63F2cPG*qadFz#b!M6G?CcQyus$`9)IC(674o2Z*SjM(j zY1?JJ>_v?{WY@Ut*oiC3=I76{*UbF( za9t`0jis&B#`4ZrRdaxZIi9n64>W8g#67ub>I(8djOsDJ&t8x4+*a!1;`Kd{*7ws% z1;w&DSGCxCKm$m|6bqMSoJAz`@Gvs)i=t+CS@b~G6~xiuCKFue{Oy8Q6(#_l@2*mRg>+k96UpR+D3Y#r2e z^v`f<9<0Nft{tE8bvvU5(IbmvYfgu>?m6{cn%%^FO*NohqTxAbp&+=;8;&qJl>I*w zS3V|GOTo90*S9T_D2u@duO-WQdi;IkD~56OnSZJ0`o0a~m~IkJEsC%)(+6ZU9Dv_? zt-o7twA79=TWUe%a>J447pbgQyLH}qVW*0sDu@TtQ_mikx8SEsO&KJXH^>%@E}nl; z<-?MF%sj_y4vUX0EyE|0tdvkPDBbAN2?MnkzaW$DT?4e6oHqMaqB&*JmbhmJpbNGd z8aIqSng?f|V+)({t&)=SQ#XGT9Li$g){*=d73*>4l7X?RBJ+e|g#(}v8@;?^|e z1Ic%Jv&S5NRo2};9Bs?Q+U=Jpp@R_t8yMU%jT;4=rxEC@t*nX(4ug#Emy_4rioV}f zYwo0~o_9v`q_lzw@xo@ll3G{89HH!|sCGpJ|S-49mpPOiq=mpG$OWO z*(G` zCLYZ#&PudxEHu!=GpZOn*tHE&YTvrIPONIjQIuI1kg}p7lVFDjs~kU?S(2YQp2;Zl z5rVAR>1@1=y?M4!; z?$bTOVZNA}KII&k%|?pU^4b`P*qpoOR4cWH-*GOJPjNB&>^#q!eudg-;T=N)kRDFJ z^I6l8im9i&*$=MX`ky!s8;pM?X=bmatbT^LHdFNu-zBo4rfj~fjSguTAdG^e-TOPF zxJBREp4tkaayL31{XOb;Ccgwx!T{VW7D6;&jgK&~Zmz#@t$(54=_Qf|?AbA~9wFUU z!p(D`xzn~bf*8x2BlSc8&UJV$gTdAdh0=xAhI%aZ z)UC{r88C2n0G|{uP*Q$ZQO_h#bpDRu$7uCN`+P#Z;SAgOjykrG9{B(+Xu%*2!SP;y zSJmgo`f>b!)!ym)Y;yfsFqh5+yVG2yeu*h6SV?4W10LX2&jQ>ZQ%5yDO9Lw)IwWuN zTmCn%hN>R%r|BEMuA}0)eS4;&zB1!X0594=KC>0&Yrp#AI-^wmbjqM9KnHw!7|Gks_g1%x$MrUQE`QuV(bq1 zG2Y@Zx_@h3np1Iv4?&QWBDCpNNH*Ap8PfbuT3{PYJMS48lTQGuZX|(r{ zf?J&A?7Y6J<%U@2kGD4z-sO32v$G!4=qGa@Y!b%PypYZw54l&|xPOgp(7H!aR}DPc zVqUkJYOqVY<{XXNq-1;5b(@H(?o-vk);Cowz1#DHow=>wsdDRg7`LUz-ric_dp{CX zdY_A`4yXFDHY{YGWUoYU6_wV@ZC!?|B-Bv0M}jfyI1umTw@r@`_pOPAy>QSD>pxjt z6dY}9h52jhspX819D2j}E*^fXE^Do8<*az|<9Js807&b7>avWCcQq0){2ZX_83k@T zdAEKm?N0dvS{ldC-n0%_cIyj_w;Afb>!#%UoH*zCk@CFwc5-Ep?r^&yK)kb5ERb_) z!S^YEU>v6E8QxRK&ZKO)+(cU>)dL;WY*hA~A9|&=JT=dZ_2jMF?1W97>;C|~lNB_O z)Etmn@yWp8r$I-RcBUHn z=;|6-V{8IlXC#x1r-vEWOv1x#hyXEzCJFZ<&keoXRl?RONj>Z*`JUhKhGe;~tTLgf;?00 zJ1FoxOAB5)gzQHx$C@;B9n!j_1faAO5$C}q6sCj;keme8GDp2xS*>+Wr~Dr;aJN_GZx5HTy5(cbni=9XZS6gP~CDW1-xCqy3uy#9c zWqmcS{{USy)pbz`L+IUxF{hV3&)l=Uxpr?$+4aVfI!m0Dk<{dAk%6U{8S&jstD6)_ zT*6AK2xS*EIwAsH6v8^KF8ev%S#akK-*}3twANc)Rcpz?F+@v8yz&ZV$A#}!3#_zP zD|_kdBK2G)k2nXY&owH-<*H?L$jc#noab{Sm993*y1G+NVuTU{ zTH?|)pZQfBOT#z21w&!HTPmoW_J=Ub5o=x#R-A4W*;Te-s;uMmfoeFk;&|KP82$@8 z!|eXn!Q&`b++6SNjvuI>10>_?6q0sB#Stn4K>(2VG)0h3qqYD*#cVxpp z5|H9#)(dfh-bFy331W=B?gRscHkBGfhUcne^G5TNl$B6mWT^%&=z#O<<0N=_B~ZcM zHd(Q$u(vqz@Kc#hJvAq#MCjeCo3!QM1ujww@_OLo+=0Ov3VSXNd}a!O9t(j7sP~@) z!>O@V-N}YCR__B~IjrnODJPnhSpyp8$qsv`x7EqrMs@*L*d&K@O3qSGHDKBHBWwI? z+MR-t!z>kTk^biHSGhY@;+Qydf@`XrbTT?exNX#o550LK@J)`G_Ng63TU=wOr+d9t zY9O2s=DY9VqfP0Zl*3XrO0(Q4S!~7LNO-D^pC1ec|J)jPM0FOX$zsPy5H6I;u)R(F& zoppk?9vqRn=c#==i(13U-QgNQ%GJA9q3UX4jh>N!WdjL5ZrE@0~B{! zuS}_Y+M6tZ7d5^aR(cNhKU)>Ldiujl47DY#Y%h_yEj;!;GP)gS5!KdkrE@NFvtKB1 z-8mlR$v6UBRY@GPwmNN1bKBkOB>8SIx-S*daQ)7}$mr{(qNr&kp6HVABfllQf0Zje zOLU4B$1`dg$295$JE$vD!Va&k<63FRnyyhWd!@#GL%vDyTuQzpO;;s9d)|gPjYE-L zhq9j18)xC@ZPHMgDi|Eb$oGI8I&|OMv9_D*e-Kqy)>P8osi`X<-HdLHLpdW)>kkmT z)p?{M{aX|fhB1{74PfNald|rx-Kjqk&sz;>rMpcbE_b5`_j~6J-(|{p7TkE3w3;hb zRPm2Zt;CW$)fmoPej_V+dDAaa_IVW-2X-ZAJV(UW`pPSudYfsZXk=~3&X+r$FubGM zoi!CV8(cpY(!Q34X*=3@p8aKvcS3ru=lE3dtvt1H+AMVrhTUhUj+T|NkO6vGHvGPg_hC{QDkz4on8^LYi??#RH zU&aC)Hn7z@8u78w%NgjdmBZ#`G|?JJZuJmS9s91~u5(9|xv5uP8|>b)-3&JAt}xXy zFxNhzt@WMn9F{|bI5y?P99MPW*lQZwr8~)l}hv=tPV z%j8}3PiUJSq_Mk8ntZ@buA?(eYqm2QpbkDSaUlFu@5OW(CK#Iwh`|77KK}qD-Ax_B z(Ct-?H4u}5JWb9&XeVjqjo>;bwd_2|IUm$2Y8fy(khfsgk71G%-85f3FpEMePS^ z8Zbxg4xa@&hSx(iF__55oaMXo9{&L3sv??jP6MTk3<%^nc=tQ&dx~}v$r|Y>8OiNh7+Xv%{j|`>A?jS2B?~TB~$uid)Bt0o!k0291Q-~PjONq ztRQKa7w_!>xE^D#ye7>pMI{4==10t@ptoDx_U4&{d&-LpjdMeo;Nj$Qq%xtb#!d(- z!E7w{&g|xuna4bp8P@B5tu2ASGLUXQsc4WU#cOkt?5%hboocqVo70T=>MAt!6l`uA z7C+QcKH$e^1gO(hha76%JSmE+WwvIP%SRN0!TYMZo*l7Ex1cgV*d!|{CI_iZx?F?0 z#nDuC93^OQJ=vZ!)7Q`ORkd6hwrQ)|p9vd};I!_Ljq;mx$Hhd{VkhBiqQ7lmf*gphnX(#P^2eG=ls!5!Dw!-7f(BFXVD5&GhquCdEP$=}UotnT-EvcsBPl=&F&NOt`eLyqfP1p~E6Km=rtgCib(v?$$rimv0fOVslPC?p_gi+$3I(| zC3mg9>z8sz4z~I?!f~?X_EJ;7=s@)WIKx2O+^Bm+zul}AlG0giF1U9LXL%!fkRr3$CqE#fei6JcHAf#&K1$Oaz8Y7)XPwMJuHQ)d z&OHL22Cspx*Ip%#oKeDNeC%uD5sa4u&*ZutkAf*|{84V=%F8`8x0*IOIG*~f7~1WL z&^9NY=aSEM3qJ_jEf7-J-&Wb=i@$R#o&@iYJleo=){?b+LsuIxf~E(U<&DDyt_K7oH5Dz5B;&;=3EFIT ztRKJYXUWRbo&>7I+KQ~nHG4N-51+Z}SLtBK|+ zu{}nP?cO5;b<{X^xzrts*;!_LX`f9AC*q=SxI3c&PFnjSPUinGg zx_ zBra@Hbz_L%eqfc=S;qV%_s88^?l08grT&WSeAAQGo1L=nLwSatu9kUAVug&r?uf_^ z^H#)i>n~zGka1LV)whQsp{0vpDWr2=bJr`+--Q=qHm=Xu`=<@n)JsIk8YY5w2NpHn z%Yo~5SjchZID;?d0S9nY?4Pr*$kgq zrLG70B$BOoB%p?JP*_8d$oQSAG65qTldgrHhRTM?mzS2gz8(rbLX~Y0KCQ6S0$C@k z`zN7LGYcH|wl?D~a0B9wMi}IDZsx}jehOVBGvl~AHxa5VpW4CO34u>jHAIbMPkW1w zbhjxylgd1Us8=-CD18}%PVh(umuYT7-0*K=*ZcPe%f&{ajiZR1(?r7hhhXpOj!B95 z8l26dnx3Jp*g29y@J2jVqR7YI>tb~aV}+%rLJ7}2c&d2Lrh(LKyXOSj&tdQ2+U8SJU7@I{d$jivg7)Y3x%|RS(#<&6MJ9$FP6gfz zvBUT|@tnLTJ^Bf)^JB0P~qhH!!i77*xBH02thadS@tx?(fBd8&s3?jM33Gv?ze`G_6V zP=bB=TsRfpKLmP~MRdC`;5qGXK1wWi-Z*Kx-K}=QNrj>HGPj{4VU+l&ocEgkNj;w|~1V3qCmnnqVO;p^i&hEhHWpnJ8wj(bKN8gM{9JdPJr z;hrY7TxdIrnwaToS^$tbK-}%j6~oDe&l@Gy+TO0&hr=yRq0`9D?xDHJ100ts@ddiy#PoLE((jjII&8 z*1#FqE@gY*P8(y{QpjM1fZ)Ur-=2vV3kL{#8{(&JQ{;R_xTlrG0!Lleq%NyZ*vD)( z?$Ef7<78#BY8skWcd+U@gt6xtJ_pHl_RURC>OOWhP}AxYxGTs#l+jJu&l*d_HIq}- z2g2xG10-xk!=WUSwb$<0DC~)#wmM2M$29p6ozhJ;F{z(}9bgwbr4Tzg ztdq%l4vRh8f&C+wo_HIFIFE&E9bayxrENrv40)16hcq1WTj)5`ifLMV`%MIdyHl|? zGFE0rQ%%1VjWWY6K%$a86?SU) z0dOVE!uO`1 zV~x>D$G1y{`3KE&TkRcOER?>6VA%j9=su|&_fso2v&2vx%7bx}OfPWf$SH0hXUSDv zUy{hjK4oAo&df3L_?!hlR>_KH_f*Vb*x@_%P#ACrD2_|B(G~_bMAOf~0*T#Efz;;T zrl>H-o=*OXvGM>jkTK$*)Ya-ZW==VjHge?!E~5Hu#?J>8UFG7QY*iEb;jeJ}N0PL% zaLwBBR@!`|4 zjq$qgbc4Ni2Xzi)0QJdRS{hEwtVDN=uDd;>2A-9Iow@UQ$v-74SjR1filS05%x9~P z6slWkGO$z<7+vdkdy+htNR(9GzKYSply=HSJ5wDiFz+n9?&U7u!qc_-t`*Ui{)!0k z_okNE##KjWdsxUgepz&kYo4jZpY)BB6ud88TJt3xoauJ>1J!TMbi+Rn*xn5CQbO=O zR?8zx-Q+XOs;cfXo~6!h8=<6+%!2RuORQ$L3R>8E%7~_zj0Q4X;QQAX z-RfoOt4Rw2-$m-uuPs>DY!9NwtjkbQL9oNHQS5<0;T{RW#YrHi9kCJV* z)>nHqtv?_5T~GH+OVTlZiiaAm%t_ zvCZ#pRMb1GsYg!(#<(g2hl@%MfX34^-f%f6J)YWLD8!`5T~a{}aXza@im~GghEk75 zkD$R;NY|fC^>tj2oJ9j%_c-CLZ!%PKL2@9zrdBWysS7*DH{YI1O>g1(=`GOK$YXs? zOEtBi?#B3g(=`)G8+6GVng-mkxNzmilPe|k?Q(SaaN{mm+qQj;r}jU#z5Q|;Gw@F~ zS~^)6M3+igeyQ$(2jo=JTRXrE4?X~7rzbzMoxhZa_L>D&Gi@?kQ^BhgC7HzCipIhq|C+Trg`5*H+K6_iOi2xS4G zlvzOml8FQm1X&2u>EM#0X2lri%=8}gA*Z-DH_GQ|WNj{ck3g~4iI+=;DYjv5^IEorgR1D@@q z_+b~~_A}LFZD{u+I%cD$k)^M2WP9hZl$%=TKI%?TQp2cur$T7ShvgK1o#lm7TAbfg6mF<(Ou&PHGJm4C8ycFgF;_t)H0CK>9O2+{n~~wfZ^@u8vBbTc zQL-uO*;g*MjPJ>EnuzU8#JmNq)2CCYZQNX9eOShD3S&RrWhDhP-q0fjm@HzB0p zFAn8)_2#dY8|saBZG1fdZ>1mzcPV#fH z!O-2k>+Cu#u$?{_?s~7B>ha4iY_7izt%cEB=`Fk~6x8m9J6q$ZXL;-QgmTBtXLuiq zF0|d4@dd)FrLLCME8wIqZon{L0y?Lp;{MI$wzh?rx>{)CcDRll*D?JZ3EwLZ#MjOr zw{}r=vCwe^6~?-%u*ys?-bTIoC*V1*-gkS~;n&7r2;G`qDlZ&icD&WTKK^)tlR&`9 zBa-NI%Nv1fPX6`FI~AwR1$|}-You#4?hBy?hcsgVuaf$m-%*tOK7*nCJW(f83|qJi zE`MHIEk$Nz#$_vERQANj7wiBfVEf^!N|C#D7dzeJx}1J|$39m0a&FRUs@XhsRFxAz z*2w6b);xjaRuUaj&R+KyImaZGv+YK;vPldvM&^kLk>?vb2W69@^j!y3<&N+pA$?~@ zX8?B+Bmwgi$%v^7m54 za84Z814qk$ zoVXl)T>k)4yk4IixVx?XpWeMrOr~;Uzu)^tH;JiatB$7RX2fvmV_T0Cj4bU7GV<>Q zYkseDsl~c8sk-~emV-9I9a2U^;v9Y%i%Iss1Sz3Rl zo+)I>k@6BxkXq|V(@f`hxRt25_$r~_O63H-q@BW*Q15J&TepI)EWMrur>ZG8e%=YA zryX*X!LUYRLw%Jl+}zbfJYEO$Q5ro}V{2|IDe^Zuo{djiXxq%C9!Z$q4tK&{q13FB zsBW*`>Rq_NCGM69!K9s(Ev>q$A{=wq)8_<1I<~WgOW0Xd^mU*7yu5 z^f791B}1oSd;P^-+u{QXR@PS700SXj&|EF7d5QJ3%y>8<&jDwy6@reOM-d(LgALo+ z^I7xA=bxVWE1}eNJ(+gm?C|h+P1k6S*2|T(zL7%8$~jxMrU$XZ$#VWN?1}LQ-1tJO zSm()iqPg0B4R`V@sD2OEmegA6>~lPqQ=ak`ll|G}_pX#J7U`Np>D#Cs&x-Oh`sdi> zxN>9u&iwjs+{Rooc>3E}brsT4@fA6vs-|X;WM-1=_4D{EXrgN_)b+Ad5!?fYw`RL- zm!-Iqv;G;sQnm}7v6HcfQ$2DEPIIy5vDe$1K6!HTWuCpQ+td_T0OS=C z3Z^*YR+k=~6s^GlG@BX590 z*H++~SiJ@cf(Ueu&PPr9Cp%%NE#XE+*WgJ|&_~+9%3lRA#l!aCbDrx~{%Rj-E^M~? zMm?e5Ok4r!vMZ6Dj_8!@%z7%d0f6uh-YN<(y0Xiv0@8P3Q<>_#g-K|WMNk>QcFO6; z4i580KHfnnB#&!uNo^aIr5NUx(VlrpXx(Z1G~p_~HsdNg4Im9;&wE!Xb_6u--bbGe zw)c>OOTe~zs|O1=iz^!HU&gW;Yiy1f4r@=Y?DWpSc;RVo9AkL5+N84gRBav6Z=}e5 zbDi~P#b5DtZ4Lhbhy@+G0Nxz4z}h-Eaj|1|#yXO?zi3iWHuLzV+DL{55iVXx5+ABF z$OSG|_+wLhdLM;r?-wp9uuu4!T8fw3U?;ngp|3mSjGjje7x-UQR8PdVxT^8=EDm!_np)Y)W^NASA9F+%ZoM%a3;MwbE zn8am!B5t9~o}VD4{1V$|=4`6|xJ9SUsZJ0Kiq8R&`69o&RS8sW23Ot@A>3!n^XkH-pZ*~0u1tM0V{*oA}; zHzX&Sbe;-XB6M)c-U86jD6A7la3V7YG>ZSr`pHrxLI1*_%G0@_X`vDHZs$!ts_mAm{i>mr81XP}0l zyS}J&g`xX7p}f~V{2ommFA4F*!mZ?0F;1_VML#ZjR24HHP zBvkkeb69z|erMdIo_w2*Yiy$8ZVR@#)pai6_fGw1T~J|jgGZ`2-;$&3s#?p}7F6Be za(iZFXMA#U9^=@IyiAvgX%#Vc{ybveN}THFA*WIlHqO zv9R$;?R+S|2|DN1@x%~0+JqFX13qUfuYL`^obXg~_k2MW4RrIw3+Jh0C5{t}frGH; z)p^^rn%V9=L~3eg?`07+uOMJ-v|ItlKZ$KFprf`(>X|omq1kD7Lyn5cHqKX1U)OLa z0J6F$E!LhCtBLL|mGe=;(8i9TbFad4v)&4WvpT!Qo+=7jx;-ywq+?hP_{NZVVRKu5 z8F4hWxodROvGKbd!<&~o&0~1~0JLkJ?$m5%pV5Pe^plWF$ml!*)w5ocRwnj4!%qy3 z`Cki7B=TrrdvXpBHPyJDn$+7vogD=sbB>mg7|2}1fUgt&O3+a2%VB%8gVS}=_#V|+ z#+L?JsVbm};DJ?ZjMPHKxUFsIID(a@Iu)H6DL(UWR|*_gBI_ znn8ZNNgZW$Zr52vS~S29Q17fqRou9K!@zxw*=i()2;!`nhDw=S?8_qo!CG@CfDs=(#MB z82&1%W!B@Rv6jaZj!EBJx}PN)mYOEx4Z-&%#LlHiji4?ocU+QqOQUN+>C3@XQ&zT% zk9*f+ONe17r!VNOU^pjrjp7F%$8|0TeSFk1dMC0fUen>~dTW*9IvR=0Y%!Odb9N81J}VjW z&MYo@dWrL=mfKp(H{lt5s-2joY|%?AzQEI*;d(mJ+BavH3EfoQBDK{)By0_*Ftyqo zJ*R5wPx4t#H|%!NFK#piwbObB5!1L<)9b{{?%GCL>}=Dr(M$L&@C0=ZsIWsaWXWu}?!#Bx5Fq%`=E)p=|1%i=y9z40B2 zI|W@OTd3&EDkP_9YfFjn>F!^sq+}g|L9Ut! zM6~icCym3Har5y=q@F38hM!x*s(iOAO3+!jt5f@nn;OQEj`gE0C(J3AD;*_w_S;=U zY2`g)-H!K#kJ3j>j=QOL>dKpY%c$isgY7Wx*L%@+U2o+TGxW(x5NO+1L-(O2Mb)%X zzqyP{lj`*6?^;?OAfc^{LszC3IADoAlkPnc*V+uxlShVz9JQqL&w{&6GS02PcTPRk zJ4cr46t%Z`nBel#axtd>k8xL%!7V(RCXBKDa+vO;aW4!2p9I-rvTK#xL84O4#x-fl zKJu4Ab-P2aM|R*wbUbGV#14v7!O?0AK^I!`b7O`== zU?1YAGULQO!(7TpUpBHItH1${zV%NHOW}ScZXWHjcGRr<>#Uf3Z>M8O6L7|R01AK;(I{2vOOQ|28r2WFBTP3(QS3Y-B$s1a77}6c=oF>f}r@7xHW8T_0 zrgKK^VcZ`v%@lWQw9gSXDrlt7=H_dfec3xFVw#?&X1S!cc5)+UwL2|qr1dN5n>2ic zv5njB3T)AdYIu#O2AYujouB5L^sXZR00~Y10KbZm$!1|DmPS-SKBw0KI}KB0A6c&Z zE*L7BKs6jdInz>){rl5WRp0ZEMuXx10Lr&}&JN`=_nlnI8dW;3`_`5p{rpr~mfUCF zS_*d`AN`ehNj*{tQ)Zcm+iOWAQwEO!S77`YuN8I+j}Xzt8zZc%waV&pPps!KXGq6h zzXgTVRBy)>mMgsl)`CMFMLdLpI*y~|l=gaeZ^8SGpTsxIRRvs=%W!B%=qG*fIWH4g<_)&QGIq|(c{t~{RP6J$JO2O;J+RC_7|@3X_fov- zV@KI{McODkc;P%c^kCCPCT->=Lz;nec}!gj{R$_q?&11W{!_`$sipXd1~@l zb4=(e-xTpk*&}gl*x*5F8xVdf%(R+z&OGhHmS~>`4pJFpcp=kF3tHdw^(%sL*B4b@ zEvDUA=hV*WjIvDrZO10J$!vHRfiC=EJxi!VDQKzB(^X6khP$JDNj?h++BXhO!(3HY zYM!Dxmq&*Q7zhU=%`Mz4)2@C4wQr~63#@H^ol!>aegv!U!roI$b+pu(`=PE@<4&zz zd39Ky!189n+C^g=;4?&BgXG}0{{V(eY^bHO&lx4At?kG^tCc{{R?VsqNJjbP&S?FMLf061Q+S@>(ClY8v{U%5Qa&+UlZ(ln#p; zL2xHHI2{%S#oAclF4lOm-1#Dnmk|W9>lU}dCk zq?)VeTX+`*o?s zP&!i>s&g~C}3qZ{ia;CvRh+Rpo(qq5G? zB&?OMjK|$o#@7uSXWkb9{3qe3cNmJ;BWoWt^R?xp83EZD_hniB3U?>mW`>q>=f0At zG?Vrp%kL?#!m^r59xQY$Vfre`hqInxq%J9ZR@`Gtn~b!8M%n6>kfO;}`RbS^bIEUj zz{i4mNg^_i<>a#)%~n<)E8Wdq`gX-WxR1t1ix?Su|9a-M2pE@|eRsumXWu|6Oc z9x;i}ej}d^c1}915GmSc$jX*J4EL>YEWql6-Zo0;WUQr_Ij_+mdvJZGyJ9|DuA#+N zxbGG(M~Y_0wy58qFBRgOA8In&Y!OEKd1bDM0K>MHd&;Wfe%AQ!iQsJoRqAM|2?3PU ztr4{Oc$I&8&uzY%{wcQ9+IX&cqpXrG5*JAMCJiibJb^o7Wn*8*{4JwP>*W&$jIuAK z-^s&ZxsAVsFIQ*!b>c}NgdDiN#J~PNDrB#>*xFfUp^5cv*v~t%8hrl%;a=_C#nzh9 z#T6AV?=74+Jk3AfRm-86@yAq(&M2yeJk?z-miHcU(!r_2XP4hLpx)P6x% z%f#_Q@*r@2I>_z5N}{gPYF@)e$fIG{;n>MhYWRM(lxmg-G-EkC{{RIQrKjVoge}od z>Q;N8%4qSG;pDUiyt|$t0c71xGk0rT?{HV#8Dg%zTo}f%vYn3lN9U)4ww`N2!v6q` zuah_!9~*T5TwBlFy#d+hWE0d|*>t|R+wGudzMKH(`+-_c#(Oul*(_}0ekX;n+aPYL zXkJcbBauE!lKv7Mxm>t{;WdYcs(luQ9LFq;7z`YPTA0oGzKpKDHXLu-H)h-j*(faV z$9TBLa2U)N6XxJB7c;r`q1s;)#^zcrFj{F{uykxX!2P1Rw+3D8R?7bRx{6AINFrS1 zx8QlLRqKwcom5+F4T7KQ<-V~VX}`ixGs(yB=N{Z7F1c|ndv$d5qx=C-?mQ86r=@*F z{7E;f)=AvapA3?t)$!L5)6p4XmN^JGI@k_8)@J+07rV6~rL3%x(Kv$W@B?x=bjok? zAA*|T+K05>YW|G+!IPc+$0|J?%KmGgM$6@VQnkgc26u0dii==W=b0lLr2@*8EJhM>OofHa_)t!yGSJa+$1-3^OBP%=_0};m*&i>7Z$b znmS>P=9^>i9Mk$NpOps*@P7?5+WOulxYR)`PWmf#JRrV7j(X0``4zOG;whlF8(#-- zwLKe1$E22bwI2TPfCs=UY7P*z(gw*|=T^!$I^BjJg0il;TdkDs@1}~HJbJA=eZgvC zw=csRL%}sQj`WMoW1(}MsnN%Ra(qhFTdelGWMQ^jPKFOas;e{xs<8!P5mw5$WRmE(Y zDWKu1c;bdHQM0^zU$Q@%#9ZaAtg3UWCw%jr%d)z^XP27qE2!oBm~r^Vc`D*PF6UW2 zbmi=lLSxH`Egb+=lf^*`hz?_9amM42PrjO@Lzf(M9a9lfRh@n44%RfBj-Fl%y(E)S z*0$ef+R8SN=S&=YPU*^*(i(RN-KpekG^S>uoFAfq?dGFCny~F51K#C)vPl>tBoB%= z28SX{Gt@^1M|_fDU{{M9r!Nr!iXB2;ejA$>ab()M!^_%U70^^*;43j)N)m_ zD%+bWL%F%bV6-&$t_Y85_=L#s27&?mmmYkx3f??-DZR5zlc3PgATXy{Y2ROShLkvv z=X%)a2Q{jpv~Wz)cXaizOnPQV6Z{qJmdk98v5mFB!7PME9ME&Q<+{x0?z)k`wW+q8!2x9BrmML~+BWE5 zf3!g8wcIts{5x+g6;Z}PBUF#w$HWf>(dX%~%I~5AVH=0YeM%xN^27sx!mbhN`pgTPkRkMk1)=D|xqLHDmIdF49-^49*aA%$q^Am@vDg;mBEYMI2f61H>pT-*=2UeVkv zm&zwVZC4594a||X?I3xMt6W@H4|T`S)MLwkOlN6ksiuIoISZMb;PW39aYb}&fHHE{ zo`iK(vQkS;G>vpKvYfl zrbRWLn)~OID7MDODREpWt1fX;8i=XsUi=wD4=)djs&!~>`g#D@ zwlOQ3%#cX%$sEoVmvhl%Sj5N6BUr(g6ShjH;v3W3BkZ=ffE+PVP~Ldz2a=n9xW#Ir zk-G6wAoX#%1Hde8Gh{4w-n_@@AN-cJ+Z=bz(A!58&ty?G?qE0>KGE?jK}kg&4LA*J zHNV=xOxoKNM_b;(frgB!(zrR<)CgGBYDjF5GWVP{gC$hlYF!<*^Ao0O!dl-Zt63?e ztBJI55M-6(SZ+xz>Z0B#39LS+dEK`|8hR~EcDXW)ib|@MMH@*Rf%01#JAZLZ$jSCb z6iQl3i0Yw`MTWHH+?CFcv0fIy`5*f9@2QUFh4*&p+u~OMeOQ-9he| zbZ~LxB?M(1$Ftcf0eo;rG=3~Q{EFlIxbIiR{oLQda_hQH&)Tst-Kiy+u?=H9vfKct zWESKcf~@4>_~IEb5sk7O6nQOSyET&Fd4g1}anh`V9q zN~=2|s;Q=$I2r>3X~4i6Z-p|-WSGv?w-(*ub)u(BW@#-x(+gJxfOpp>=Y4X2HKo0A z{qFBZ>Pog#X>i=Jm@7YGXL)vnbO$RrUA2{$(_Tj)y-D~h@dpmuJ3oCCvrhYjwQ@-# zBY?XDY;G5pk45O8!8(o1rK+ec+%%p1&OeggA2FQ%8-68`@BC?XtEg<0ZJ?Y!o>=p! z;NzHFbAq^jdK(lbJH1Ox?eBDkGzRBRn9465K=`G zMg(!SpnHH`2i3FHn7vP@AtlYBwi4-3=q3l2arO zBVQmFhj(7Kq(_db>?8#G8^Habj!Et=$#oqg{{Z!Ws877GpHw7g2zx`DzI(lEEDt?! z<{#vkk}4p6n}*nbW)`2|t|_#&$0Ks2!W*kD*;&bP`DWoNnx5}ZMNiy8YN~=TbEZ0c z54k@V5>kin6;n^^LhtCU+Tc!+oFaThj;dn{+~x7s&as@~VQ$RT!??g=>A zb4vwnZGAf}jB}uVhh|a^;0)ld%3_S}<_~!Tb)Dmym~H$&Q!`k|-GI>G(Vfl?7dHna zo+duo&kOBETf?^7hX&kzCYlFmqZ2X5a~rhhZ-`!dEfK-9K__lYwDE1`YhP#eGQ1W# zwW2729Dx}A3z6{U1x42BPXl9di%`_=m#>(9B=YJra9xg%fT%7n zqY_m20Lt^R1;aVp!E$Qq!1*-yQU*8K5@SC+E~nbh5Klp9qO?+%i9u_g0NYDO!28x2 z+;ww!XqJZF^23ps23J2xpDuiT-u<}w@xvUo>%yX%s@~aQ^sh~lCVIv{RBz?vp1xGr zrHPKVQBO}ktmMBV{{WR1=XH|hSZAqY0}0P}7!3g(eRUkv45p{I#|xTrWYkNSZ?gAG z{{WX;b5&_?Wp&<)A?dNOHTnc(o_>ByM|O4cvYvv^R@!Q6c*p`MCDdeaosC{{Z+|D0o)$Bg1_ik>~A57cUP!8S&$uS5~;WI()It7%5amGHI9?%${up1D@B%Ep#KpIb=V3$=z* zIlO$h?l&8&5$fHDo$^)V+VaAkQxK_holf%d*I#*znsL*Z?d&3#j#qm{&s+kx>_@?NR2hDRg4>J@ENwd4&t7~6w<T@U7k5da`OR zI7uC#4rh9)>Uk0dqV?SqBr&d+#&aYqf)UJOj9l2YcrHk^jI73E5|mck+% z7w&MoXW-v+{>^Z-Im0Y0osB+wJNYj^>O2_d#dG)(uOkWewh^6#~H`RIE4;7fY)x}3eEY1W5gYU}ecgitQK_gf);583; zT&IfZC7zgo2|Ykv&YLNflO{PYjmJNZ=^RF)2&L5}WO^LdpFU+>6SNrV7poLGqdz%y*O~{x0bTUPMLqG?rD1Tk#ZX3~7)YnZhz$LBl3A$=`&f*wA z1D;B?q@scs2WMNyGJ1RvS5nZ=P;v&A^>dY#H3TpWk*)m+3Rye*(X?gl1H>Y_U7!Y0 zQ#Zmku`G_aG9;F>LBJ4azZnfMg>END2O6gqS%?tp$bwoQqq>up}!ccd~EeOqyeG7{u zYsWU6ts8!l7P-8L1zFYbJT&x^&*|gWlU*oj`lko^A}2(0&{M%P4$dH_ z!NR2}r(#b)y#c|t)7kH3t(>ojeyVs0YcO^s^D2s;aO3bu9ckSt zoFs%q*NZ#;1 z2R8)nbGNy9!+<>|mh9`qGf4v|17&=TX(u_^yLBtBT{~>z3$3c>MNL^s+KQ@YcdnWa zOdi%XZJcfKQz^JZe}dI4Op@<#WNi{i155gmor3f&Eba~STrcopXa4}J)$`+bNYVU% zs_k^-?zGKX3W7!nA#y*$exftNmUtz5TO7AW>MrH#w63PtVR660cB*)y;aKS-f@#fr z1LL?u1Vh4%;}t7=KCeY?~HU(&n-_(`;O>4A*USC zWTSb&@KF(Gdm@utW7I=-$=@YJu5kwhZ0+;c+^s0&r)5}uOaf*Ab|(XVTdGRAb-PXm z$EQ_3P}9y&kmQnrHjJDq7YGIZvq>IGV0L12!&*aIqz?}@8b{(Xz%ULDd{fk}jOCAM zE!oFO+XI@EN~WfB{URqgIo;TB!N;tu>=o^|Tx}J;j>%`Dm`Kp$>bSJ(4Jxjnn6XPc9PgwT+(r!kGXcc@5?tCeY-6;7ee=dN!08#@30G%{u&n+Rd!otb6v^>QbbB27EBJIbt7`WNi`y~Y|ky&V?bdVWG(#ZF0T<(Kz zL2>eBG`Qp1wSkk)jS=>7KmdHo>-xsgCi~sc1}D>uA$)C-Y?^7nCjuJw?No+7Z^@mjBMPD zby{!161tY(#13<4-q_lT5iXi;<|6FJ%sI_HbIoynJmOoO?~dy2b6ewp zTfk*2Uk;ba86z$^=j4Wyc#n@HeRVx75=~tc;5t%uVlYoKM>ijkS(@8~H9TiqEkm0$ zFvP|}#|=$k=aJ&3aE0EYl3KfcUARO{k8_>f`%LMDS;e&PsjPiaWRzoL1dWdY!yh6P zw@TFTdpK6>e1?km!*<%P{zit*<`D2f@;F__<-+z#t{S-6toCbl6RIi8Vv(@5rH7%& z3&|WGd8?-5%VjNGkxB(iT3*e$mF&w5n&Z{t!n<4&WsQ9Q0M+@k!nZ5m=T_1YB|U5-Nh9^V4@Fg9#B$!auI%bU`g~6F&sLxu zMtr#|Do*Ae8?BORVI@}jCUGIdZfmb*>)her+sC`!l|88Tfy5kLbf}7|s-h}V^R5#i zI^E~x!Q#Mw(V@S&hAHjL|jOUW0imO{88exBS&<@)!@n2PJcWlsHDJkQU z*Z8fjAzU^m*U7iu<%sr5OvGWs;JkFMA4kDeE> z__p#a)*8CH$Y!T(P`rnbFb)nnuOI#o7V~iJUd{_OGM+$B5OS#+kQrC@yWO4i2H)!EoDe0~f(BXW2U;*BV-i`nS3m1m(IT z15WM8+>m|?HcK^rSYHrcKD10tt?k{lqT^XfBP~tR*dK<-HJs{QE~aoPq-LGmK_3OmT5D=Q2s&D7DPIh1vN}oM z99qr;_8ao%v6eXJHJ7B8`;R_;9v1%qM&tWER;RnZp)D?X$vR$ts`H)~t^WWHo)N!G zY_suTbyZajGoJ-Pd_kfD;xvcfY%XYSUE%?NX;qGelP@`-;q)#Yz6|FOgHNHz}SabC}J#57}~P?oE<7<{r4x z@HTk%f;LS(HapD9!OiU+u=sUUX7NAhHw zofvo$F~Q9vIR_*D6&)qUnUC*_T~AgyIh)4scu8f_x|El-jL_qdK+ApK{HkqzsU-DP zQQkY)i*Xz?=cqayKaysm=U)yx$YTvY%dwDtDjQuYek8|t7wr`s-l}o=xc*8tZFIG5 zYh-IilaYXly|z7qNcj6#vFOfSF=_qMy_b0h@m$5{aT(v>SB(1vos|8oSfrDYoVVgy zSF|qzt!yQCL&3sPO2v_|^KAKL}cyh(YJS5uoM&+9r=$CygKT0Z>p%b_ku$N@b)il)7OA8wO z2YRIA%;k6%nrF*Xt>MM;PD{LUT)8r0dwWl@nB=$c4W1o(W{N!Jy$?KvXT+Qh+3yiu zWV>9gFQ=}T26IhN7}+1y#1XMQRu0w1v2dM=*9F3kM>1$NMEHh50O04(Dp>7bhabF@ zG8cKry1Iv#K!)FEwAv%JSZJ9?K^EbSc1zxQy~n95%0GlL=K919o%OT7J|@8Io$cKjc4Jy3L1&7aKSuNHI8BL zWey;bzA>`HK_EMf<1Stz?aE&JwZ0;F-v}dbG4EL(n*0^$;No5txlvuIUsBo^Xr^#+8bKMypEc+n4T_d~9CoU> zp5{u;Cl4*<<~Ll&@Ikt8#iO!m4Z>IQ9&mhN@zDzUN{6~g zfoOB$Yn%WBsph#c^tHDIF1o;hqaseHAZnu^FpwQhY=A@*N&`lr?4s;HPNYSy(k`Qukb zRR*~wfac>VYFw4dpvg*H->92HMO-yo)hAFL zc;|7sWvp!8+>*D0II_6q+|>3m$Ab1s`ZmEN_lKiB#(9i3ap4tA5n=I1nVdQfDT&*t^}A;+QX z@m>4yijj@%rU@JENa--<_-~EZpZJ=1uHB)^<0C_1l(H;yA)}Jk-nfe8ckJG)hwEM) z1-_ZmQ@Q@zK{(QUM}qY_e6CnbviD=m>Tt&nOtHh%y8i%g2YUdXH}LIfX&Rz?ng>?3 z!zbptm7-RPc*qzF&b^M_q_f>8cJ8O0nq_BFGtb@^bhF+qkECOT#+VL($^l%M{KplN z;~hTDaBD3sgpHjuN%ISU_Q`jox50d&hMXTrbD97+(Cd|%@}A=fPgfm84tx~tGDbB%Gt7>pCSjhL#FoJ3f(Rg-;GUsXQPfFG zP|4yrb54E!O2y4*f8Irb9D`mtmgksQn4)7aKs`z@lM(eoTe+|^?>o0VNV+8r^OuOwhJf_=nz zEZw#V<+w#{veV05aHxKmV`#}g4;86(WwW#RMMTvQUHHMTBmm(pB@gnFPn$x!+G%7c ze-~RJxZkQ}qNSVj6b#3IAp1@qQK!WG7E6uqQC8Yyq@ku1Y>?+?a%kuAS*w>2TrO7$ zYbz=wo}y=$InH)ViSho3?^@~X&zfguN3=D$^HR~2eME)FyOHexbJfgwu93p^l=iL` zY;T~3T8S9vI%n?eyO+sxc;axCjBo^;U;~w;eSC(T##}Uj20h9mVb1#iovEpNsMyyy zof{0Dlt^C3olbt3z{$boUi+&;2q?9c~5wm*b#Y8Ji- z;Ktaj5;29X?x|w~c^oz_K0ziu)h=DEzu_a>9Taj+V6{b_>L*7}3mu5z>8;r6ItAnU z>W577H8_mqmu_6{u0Cn%tCd}CO!HJoWR`cC9^=i<{)$8vJDm2qyTmjuo{EsLR7stp zF7`p+IV)LZzY&_CB2gl zal9XchJT9G+&G51@xrnPO3z)oXKsvOg`cZon%3y~>WA-c;Z#fP^0e05D>!~Y*S*8N zr!PAL)F?Gif-0$LBq6gz^2cQJ!l1F%$!)qpSrf46-<0nVI;KmtMMYCs*=2co)60HX z_$=7m<+5oqxanpZa!)KCX~wdcLf-Z{VR;`^XFoMxVyAU&#weWk5Zt`)&@JRI#C_NU zAF>beS-9VZvCA9nl%lrQ+-oDJwKerJyp1el4zG`r>|8(iIp3~PLtS>bQCh>db6V3O z`HT#y9hGragN7R{m6o}utCZo>##%7Z$PWeBsrz2xi;apn=5*A~q+_(xnDsrxo_QTg z);QbkjI5rVw~PBexb}Hy@8aS&-h&|%dJSU$%z4ME@N(C90<(7R9HGB8m%EHDwo$X% z$kjBi`!Tn#y=dy;yzynmdYXu5WeAsBBXISI>8sDnlDxOtZnO-|qQ(rwW72CR1h@8E zoCR(e%i!+1Ro-s$+AQ!xV6wq$WSEW~@B>vZ>uBlkDeFwN7Hr(_x5=lel080+-fJU0 zNF7ye&cnpi_bIAt12oQkdAq@={B!D$AfXE+l0Oku(>6z(7ZOfBYga8Zq^>99$r%Hi zM$bkBesRYy_yx@Gx2uiTQpQOj&o{l+lcm&=eoN;Hw%>B03L;min{N?SCnk)&fD`1_|3owh;LZUZfW z-?~S}`n+}dgNHTeA5 z+JR>j{)%0y)p4w;Z8TK!#^>{U z<1GyxPH?*wFm}AWuztgdR`hI^Sy<3#3_T-@E)fO|>|s`EqDz?Ur&_Y;~8K zp0Z6549}2KAh^#f^bQGxGsC(eb$f+GR0NqTWr?8jmNsYPH&oXSCxboRJdGLV;{Z%~ zv%l$Iw~zeHJxz|=a)@_UiS1P**?dFQC(U&p$t?l*ST(rF+Bo$+&K}XnHIO$Nsa@VT zQ= zWPFuDq~jtMdv5+i&2PuUi8$Xza_MRx>t5OW;kptw631z$^Hus7ChHxrN62`;C%Gac1E z!MN%`*!-7MK3?nMS96P}#|iqZPq|yFDnX`n&ylC5pm2PaYUyHWidf_hG68_sU|{(9 zsa4Ic5(960;r(fvtLj^QmB3GPjNtBD7|QHpmQKE3ZyoqkQqffRZl&kZ{r_E&ap zAbqwoXuU@HsGs?VpKcT4%<~>xlRl4daR67pZH^cVin+INQ|=6gs{2=PhLNrfYndBM zuzBDW$(DG@<<*#FOX5p()YNcI1hLdpPlv@bH*1H1@NnuCSK^C}UY62DBOE9p znaPdSWre2VJ&Gw@?+)$X(M+@Oy~)>6S5rkOh+ytR2tM~Mb6Gp11lCp-V;eYx!=izt zlioew`H~a-)}|bpc-g!s!l#NjPSaOc1yolF9YR-RdL=~Kb9&a;rI#65XDa3xj7|06^cGimB+{h zWvXAR7;3Px(N9}e1Ia97OP!6j3!fexPYyh@-?d#GwULL2qj2b5)h)JaaCJGX{{T~! z!|2SCk{f~3d?(FwZP zvIxP)UDm$cW@SEW4%#^Z$96oCx{hq;mzZ4^J6w-HnTz1m1)_QC4K*|@e1Lr+4ai{r zso8#GUHE)gI%GACcw}anhl5`AMJ{{Tu2r-Avyk*slJeWDsj9`n&LSuAu6ET?Q* zkF@8JvQ7*qjACA=lRr6RW}eqmH8H83@xd3F^|S6(b+2?F#~fxu;zG8!)VgQtuDH_2 z4~vQ0;1z`T4iknl9a$vr{{YK~{{S^67aGQV^C{tisAWdv`k_Z`E-u`XR;IG6hiWDy zdR{!h373u>(nyl9yP(F)nt1qi@1K@FvRPxOh88v1cIPcTleP5XXiPohqb?hk?Vly- z8jj0tc>}ntW_bYf{TDam9t)+Xxk(K#ei+k?mt^POyzRR(ZNu)=Y@?mAvE|trIT*r| z!pzlhd;ZLSdUK2@hMnRqcI%vc#sui<$Aze_Ox{%FIj#Lw#jHXN?GM* zgE44tm;rfS#TTSrDIv$TD_6ES8k^9#UW%VN#6jEbjPHfZh1NE+m$yIsDz9v`@I?S~ zTtv4Bi2(Bq^@_|`?$-O&3o9-4ESe{`?F{GKsXq=Yi865gtUCmEv$%YKUTB7UufU{;le}ygfzW@gvtWwnNb4$q4o57(XTE{vEBkNk(Ln9$_rKhjI_Od4`{lIIaw0 z@jLPM_f}ALrNz+uzF!tee zVD+rDj`v95r8Bn~TtSfZg2v{#oLW0F)fBs`x6Bo>DzO&5wzKZMh|)_Tn>HMPCuu0Gb#E@*6>;Fv*S(~ zH>+I@8w78TnVPhoMc7%s2;rke#WB%K6B=2U=^T|jeYe~>%VDHx!nW7ZyS=wj8jR!~ z=x!D=4%exrsg2WFYh#RiL5>G1-Y(X7Zbs>riscu9F^}ZBpTYHfexsZCo(%q1cK!;m zhQ3O>Eeo9E$ifPZbl^T1nlU6+N^MMbOIw!e9vc+ZH)(t-?fo)Xyq_6QH;;Hx{I8U; z=j~*FC63OnUHMs`O=rVBKLXLxN%-E|ZEH+UIfRl<2;@mA&P|2NJZZ`btZ57V5e< zu9UBBKssC`15cMVzv6SeUvA@P_HoWGKe^wwHl5F%j-FCoB!?&YCkF2%W6ugjMJ#kh zv9h8$fp4_KvWrmXMMpI8gY<0#k1VZEK8vW`@a3w4{{U$c-l(ChF}u2n&Tj2!>5!)W zKyb6(Nww+l(=&(jP`fImx|xz zbEj}E(2^srrgl77LS_69e9lLqNIz(g~lPp!V{^_lA))#-#mRB6bB{TmGbk*QJeF4TYa(|Y2Sd(UKw6yR+>@O^4JaRlU+ z^Io1Eb;}7da9#<$#}TCAdE{vNIJEeax>a`xy|m7g4LBJKqTc9TM^x+`o}O#H z%|&OrmR3+z)57l@fE+yh)r>qK%BB`q;mvUM9KyN(0OB_f=5bD%Wmp~s;mHL}BwfHb z_1rmhU8jer-vueu3r??OsMb3TH3a5h_c_1--1PEV5Hj3=ysodS%aZClyc|8g$GFD% zOO`UwHpo1adCxJpQt4!Dv6!SD<Ny zs~;C!?9v?`i%{kOI=4H0#4mhtT>+@JO->z$dr>>V!*IOM_(t&CDg_T3(?R+0DXV~P z%#pl__wwC+H~4?5#|}=+zI=awhl}-Y4*WZ(!+}LhEQ4M!-x7+hpSXB#qH;DkR@m_y zsJA8;z}NL=%k3XDgyu^ImjjIQAgNUKSs^r)Jzfv zvBbRXoPe#XDLXfzj-pX+l4l6rByl=SjX-l$>}#6aZ+7hwwilN)4r3cFbo|$nTXD1c zuP+qi*Xr80i6NW6ps8o(;UiES^5fAtX0K&zvxwgEVZ&XP<8-^t&6fAZ>1k({JuL;y zZg5XbD10?j4N$GArvx?AV|L)|7h`g`Pw^CjI~~qPbbyba{{SnOS)~n_u%Bpu1<|<1 zVBQqCJMFoxA27LPJccRRAn5FHJE|C~JHy|E;a=g25zuYKc zjs{3Ow)(jvBs88`UZwmRZO8HLWs>1tJcL3E3~te=VH z^^9j2#@RUes$ao{)REI)*>t6J*2=h}r) z*AX1I9E#}nZYASuMHMs4s%oKga$=Q^dzrp4TygJO+)Y=c8tiS6y5au-gmucEJ}Tiz zUdB}hHg{+)7|9q1K1cX0-w<(M8`awNPf;{+xy}YCTIfgv#DI4BE@^+OxZW=|%9`ou z=7>F!#~e}8BDJ)G6Pip6Se)WqCMLvmbqB)gLHKWL60loh~t z12qR|eSz-6_C=)ddN74UmDWIM0$(BAR+vRJ67@o0odWcL15HgTJW{ke=)T$rXK0sv#IU z1dh&;^;$Oj)-C)qYIW*sYdCs1qhR#SczdyrHJ3}4M@Zsca+1N9dfbOKU1jNL0@rKf2xoHhc&ez`aY+*iD~H}HGx{{VzI6xQ)Q18%oQ zz$Kb@xy@@GjE1qG?Ss{Ld$N|x{{Yk*TOwAKIAxo}l>8l4XQi!?l=Q89Kr%BP;Pg7Wtgj9BZ`o%L zU#GrP&m@)dz0dDuna(XF?}7I!=)tMsE*6HeQ4yLKNh_M&fjoc+>ZmLXGCDfCpd#3U z1B>_R(RXp_anJQ_=5qZ*E*!5Ox2H{M;9I>ZvQ-)?*lNCw)bc~9Kg)HE;%~v5vs&m& z=pd`K%my1QbGhJr9B`|60t#7p>e$Lo`ORaT_xFQ+hS>65tBERwHI|}x90;Ow+<9yQ z>h(Mmk1a=YqUx_vh0D{xKZFlpcka&%?+uxkJ0!2eYgpYbVLY1l^!M^yG3+cPJ8Und z7H%$~<7P+0bu`da+hmV5a=#8^+I9>b7nHvVzQd~fKDY~|oK0K9wIsSJgEDE(c8%nH z#c^eaX146(mvyo@&j#M>ELZp|DC#7GFAR{d`%OH-UXr7qzP9l6z2rWVL0sks;0Y#V z+u$Cc5WKG=3nbhDz80^5IEv*55!q>`seH7Q)m`IbbF++$9sS7~dk&(eNiS1RsIl#( zYg_@w`RDRpa_tV{;oFgOovSc5EsvI0c!saOsIHtlP%VrHc#SK|e0f>kcD-3Jb~H`^ zJ{W3WYb&W<=7+hWdkYB;-T2(>R*Q$Pwp+E8)P|k1I*JU)CX5#21ngHeH#R=U1wnI0 z9_;7G7*2Hs!a?6D#4!WUpkrZ}`Fn2;()u?G4xgtYLevgPbLx z5_B+XA0WPdQxoYikayNUiudmVWxO>L&p$?B`GC6}H_O5L)boW(t>9R=hIuODyGd_s z?v2&0WO|IDoPasoBVw1w4`n|OTwMUt%6J?jA7!m!0bSav%rLpWA!fMc#!eExc$!HL z-?CRcn%#N1^IlFoFs~ADJxoo#RLN1OhwgOq&Z1WGw&6u3uA&fUE_)6O@qxFJ<{UL9 zn~sB=zhG9|S8-1?j_Ah$?>jDxz8_&3<};xjmFEaPN^7X8V@VN=zaAM?EpL*w;fj$@ za?s`i+1Tja@f$49ZDi^!;wyA?;0fn;2F(!TBhSTht|!v&BCC;yERp)b$3il#swI*t zC!J&#Mt4*f65~-DoN`?RWS@fHCRV01L^oiaX$gj(Gv_5qS5GixFIPR!r{~luVs#@J?e-Xq7UNlN}4l_>dQH?YZHdmT?A8I#NDmaUa1v z?7(MMHh0o;LRsk$OaKZgS%6 zk_qL}X34FypQj<=UeKzzlkkmjtc}#JprVq9$0LSIgN|{JGFPD}ZURP^#j7nPuJua! z-pl<+?M3`S8|o=tC3~iMhYouhTyu=T#zPC9{I{SuTq-Wne-i!>RqcwNdWY1}$8CYJH#w75aJ4kt z586v@RhJvr7c$vTO$3q<@!mUTk~ZW5^5+V1i)lF0lIL9K3F+%Z=0_c*A>9e&R&&Bx zt*;#R^jG1J#9UEn?3zkTwX0P%+_ne3vNVVvs5l1!%=)`|uL<)wUhDiRxQ>gBe+BQ# z8W@2SRm9@mSutVQKl}4u6y%Y~kDl`B%6jLQtdSR{0E z4o+#oS}QjbNe~UV0`ths&_3{}bmz?^H*9$GS!;T0uCz9L%q^YJGcGv~kJTz8ct+0h z3*?xLZI|*{_Lq&C=KBl)bHy7BEl_ll0=BU~ILkP*o0 zD(bV{`y}3YVy=%|wC?nX2KfUH1Kg-_rQT~_Z{lfU4T`ot@uP0>yB~;Ct1ULW++tWJ zV`S?1{C`+0wATGFWV@eKjCiY?y^6Nl;$)1=4@p=y_^PtdMn--r?JP7(dI_bEIvg~f zZ99YTP3g(-MB#!$sFNq7&UG|*4(i{keaE62=cPT&jmt+iJ8*r$427U0to0+qB!JfF+WR()N$PVk|lhaL~qXAvV&9Dq&FV!bA6Y67Ce}yyoPRa$w@3vG&Icm1lamimEOuQ^ zDZdEXhzgc4SYvOcm&#xJK4o*2^`K~4SKwB*JRhDiivIoZ3N+_!1amf!3%U$sc zlKOG_u9tAKTP~J60I#ZK4{agt^T>kpl^&~sZC{J2n#+|q(&OuC-P`r?QF!!9%O458 zGo(I+$NvD%Rama6d=bVDTG98ZacgZAZw6S!RE~9ZJ3x#A8ZbF7YxqmGl-HUnNcUik zg@kgzU|On*nu!=>c5|l=S+ByCju%^3tGT6*8*iNfB=C*5;EPQikAvjhGoeT8@4SAK_U^Eynq2WiX7)8zFFfp`?O##(AyRW5Tl0*foZFdT__R<4S)! zOHy|R2=xFwJ#1xVrk1Jfr-AdxLs=V3+)scR%K7oc^^CajCCa`m=%}iaR{NadLiYz3 zc2~w1?iH5A{x*DY$S~(eN2Shh*(4m`t}k@_6!@AJm(btnCOeSXB6|t@ku>wT-ML2I zi=c5=V|}4;>3gbpex{}dZ{O05ZTRDLf{~Aj?v6;^8=7$Dxug($2r6iLi|{+2gs0() z%vb6)aC2eqyl4x=Z2e1TfVtHr;B4NHZ}h6y8imQZRXmEeNs za^zKK;QMl@r0q)A3@i#b~P_^+LsSJbgp3e z_%HXo26r( zX)5J=DghbHZzOI$E7!av%Z(on*xO>a(#dRvg|a!*Xbfg~w0N%w0&=}M_&s*ZW}>mj z!~9IgM*4G@XV|sdV{lhPpQo8}{mcIV--`75PNylwcyagf`TgU!@iVSd&^j9pb-mS( za4vlB*4X{yeby7TY2>2xd1{PJtTxhQF$^~&KDJf6`&8hoW9nZV;;6~Kih4Fh^Euz% zxIb%ss7J-u`DiWF;wpI>AurS&rZK*def-MmbsQLS;W+sG`Tqc;zroGP(`T12Ch?lo z{;p|Zp|85}eKkB!aH*brx#aZD=iIv|YnFQzvxkE%RPTzedOqe?8C*oCP&Hs-I{_rRCuH-=%|%kndtBg5k-BF%hK<>s_fw*|Q%MV6 z*_r|I*>XmH7w|R)zf0wgHn?Drc_>#_P}H_aG`Z18F!vXGXQ9Hap5si$Ip#)?-Odi` zEiKvAtR^@d#sh#|%&j6l9TaxnDxiokPX#=!bGyI@dr27tcrM}FUjV_yTmfm~4j!h4 zI_gIEGeH|eJKXZX<;Z=@j@&Lxx7s9)(9_E#o3;M{SRR8b(|i?fps(ygqU_nKomWi@ z9_D0e);*v+w^+k_)k*5SO!k&Iw2i^RQz#w@(i`i46!q%eQw8RxltlR)*$*b3yZ-<* z{lABAvq;y$ZKxx~&A?Wn_`Q&vcO{~t;JdVp1>upC(*&zXI0j-fElKwPjWGEGpH({F zZKqqEAzjn(RPJsAYF<18b^!)T$!PC$6%mdmunzX3 zM_`fUt!im4^t8R*D?9mdhdeu(@meUjI57I!+vG}<2M*DKOoipG*oKUK>Wt?NP+4i* z+S}E1a!JjU;f^355_uHw6kRSChl*=`Z=le(XXqZhoQCR)t51Zf=9oix;9O(vaTbyP_Wr;-PL=6 z@2iXN#?x~$@+qDgc1oJxOGVPC+o(#+!HOvlE3Y?60Ln!yz943&(W1Cf?wed}HrdNh0j zTOoVTA&$9Hnq(s3h*(3TqNZW#^i2uD!jZ~-dU{cl_CKQETWt_d$C>My2@N=Bc3fTH zr`xRB%SjH$m`gM;*3JitOkt2|BxDHrmEJoi>{f!ZHcxT^kh27_gBrYwV}|UI!z^++ zj&_5P2-qJL(=9hTidX|;1P?8fTP+tJ%|XLlJNS;lL@8k&*^FTP*UtXec!sNpc-FRA zrhB2Ld4dMG`z>nv8FQ=jp^4&av&@l|%THuoolA46t)9dssghFpb;wIeJ_DlFwqxOs zcXi=rk63TNy=lA3N=iQJc2t&1)|q1yxHZIb$Xy$YZ5GDy1;Qp;sObYHbZm||fD8cb z?_B=-V7N;@xZfPD8#SY1vu2tqg5zc|80q*-RXBZpPIf*@O%*L%gvm+6Qb#EDqamD+ zyspQ0qK^^;|w`Pb;Sus<3fA#5vWBt#)@2f&d>f zl}o~xSRgTrbK9tbOQmj4-n$@W85@9*=&7=H0C}sS$Qi>Iwh5z#pF~oTsxv6SK0NuQ zKFM{a5wX!hk+VUQcYGI2sbrOpk}6hrP_%mUbw2Ovq+4wVx@^(9H=N*?oS&M_x*Xev z`w`;n>pR+58Y4L5K`To~*saGzqQ2o4DPAFzJ)r4&=yXuW6GIK!D1bS0S`IwNy-alv z4o4fDZ*qG4{+DC$S=}7=PRb*;uvGpqgf}6&p@ptH^w{Mh4gr#3A#K+{Yqb3(Gsnz~ ztz7h!a@NNc9o{!L1cLF59)W0V_WRun#F4#~QIK;On1FJds=!coafur(T~!rSK!brZ z<38?7pxPs~+NqmQPiwP48_CUcp6`j%l7$5KNZj#R!!ukZFV4AUb6N+62QZ$v@pVO) z=q2HMByN&3hB`9$HzS^N&1<_x;=btvG2WY*paAU3b{6X=rstZZq-EmiS?JdKs-V{F zWR_lk*Wjnvtdibib;jdV*drT4$B(}X+B3|7S54j1 z(^E!IAjrnU-k#M-Z;_JAQR#BGaxsQy!_8-PcCVnaI$h`@Yk=z{40rKXGEj;c0J~Mt zfZfC}(losL$Qvf7;s?yr7Pr$zdPVJg(=bx{OLIc%xnw7EjlJb#Ecs|ZDu5B@Fep1DvO2I*Bgr{ zXytHjz{=nneeE39(`xz}rdBw|RZkljIXR%^$IA)YiE3Gv!dp(k)yE}4th%_I8KQ(9 zLAcph%~h;{95oe3^%!yUI8QgwlP99Nz2kSD#*ul+F89 z69)amDT=$a6s`5{_TYRY4te|VlaGGZRl^xHoZNBpR+hUreDDb7dAKfrQN;A%yEj>Y z{{YKl8T`pm-+qo@bky7=^KD~?+ydWYJFh|7cy`rCLi>fvc;Karc_bqz+y_(SvOIOi zR`$qfcut;5I##z1o?tuay#D}Hdy2zbDOqubAvvlnk-@vvNtTjv@pr=uFx_;uw+b^e z=Z>DG(MIQ-k^;T68MtfCmo--jswQ-jInH_J;&M;LNAC-_%7ocG(~XdLs4cA(COVcf zN4Uh}>ogn%Z8;5Wj-#FO5t119V1PUpdX^W+-0IWem18V&ig#;}0gcWuRx;`Gwq+fj zKBrjr)HyFRFpy)1BlUL*1l2B(>Txb{<+@Cabd#2~%|00FqtrFgKS>0w$uYWdWpc~O z>Y8szqHNMLyn*jA=6 zzOwUBIkqw48->M}1-~xNJytmS#nGv%?d{rX*0g@33YO~oD*)3oo$p++w6)GZOIaz; z*(F#(ZqfmkqDOn)v3gm4T$uesDK#;The5#^1k`c|>c!E=}i0C}91XG@0tvzJ}VQm3q`m}zmtPWT*? z5i&}N1aY@hk01tA3Qca4OL3E&mvq9k;l3oMhU6{Anej^IYINb?^v=NYTbzFp$9b*u zlS^sgc_WAbvp`J0&v22D>KS86^}XaN=F?0_W9sFeHVnKvLe{s>bzk_})M0&HM9!yU zPqi$BxaZx;CVIErAq}Pn(ciwe9Yx-GUA~{CxnDKc<>r&lj+@;jEOJE!D_q^fM0+>l zd6X*+B{ef9rF)Nh1CCpaXP8i(YJQzVJ00>6sxnj1dNjVfLC7F-UY8$Djvk#nxvpNw zmbLBxYH=PJOqm|z%z&8JgSqF`AAZ)KSdF|&PCO!aWnx@RImlBWVfxM(@WQHXd@aUi zj05_?2i@r-8lP7SN1Ky){!_~Ro!dwy-*Ud;o28l`8#s1fmuBjEhKeRx+~sXTKI$9!q5iCmJU-? zTxo2A3ZG1|%N*>BG{aMz*U6qwKUKKuGwxjP@~(x=)X7O;I&*`~B5o>UB<|WJdB&!Z zf(TrBa%1t6Qw#WZo*&{`hEi8abF)}tk$@Up)w$ZdMwYabpFAx@d{0dU9X!y}+GLf@ z%+WH8kkjBC>}RJ%=P7u})RD6+mnEUkU>Gb1#3w2E_&l-33H@P_*JURrGWp4FABgoG zr;e?EisOR1#X{=Bev)Wm4QLtU2a4O+c=LnlDFhT2I4MCp<)O|esVAvkE2HfxlGS=R zqLdO&T=_s7EfYILwNg>TH1gC*EUnaHq-m48@^SI@()eRexuoGJCH6k_+J)xkfmw4m2(oG0=hWK22jcczC zyU%a9)rmzj%UU9kx-vRBzV*kKE*VWm=N4GsE|c2_YWCZ`M3!z3w?}EQu;UTN*2z#u z)^}fmi9Y+~b$|P8N=PM0I3xaT&bcmPi};_88~BklcB^eoZQco=$ojf=QfZw_NYfo0 z(3Uvo50iYTCh^DBYLn^{`i{?R~L! zcG@^#H>a?*fO^n4a$Y&MQ?lhJbwR`RRS(WI!wGr&m4v)Eb7orp6TdL^#T1{(YrDp( zS+`B?;`zf|WyR4}(#sWfEHv$nl5XAkmCNrGTscoeGus?-&7k*SNOwIjsit+RDIXNj z)$lpdX7U|zvBMOnpWrMpH=PkV+R zS4WD0NZ#(6CU5Gn4V9$`1`R>vWaN6pEkYT81zYivEn`QHkW)|Ksy=P4sBgYCX2;4l z@0pckks!wHfL7IiMcqyDrAsn}tPmu-s@GWRf<%F}ZsVK=-Zhe^dvn zRO9doM&WU5L}H^;PeTxJM6t^VAq*~}YNd0&k}7BW=~|rv30hihw&!PkFnZClOYqk#YDzdJw?@fU$rP=F@sBFb;x$wB@_l^1}F4*9! zO8WT=Au&xOoLQMeL+#xu>agZ14$f;tRkI0^vR!FyWqQ zihFl`P9`;Uu{7sr$?JTPx{{5XKH|cEQG~30ygli@{J$hp+^cD71w}Pfu*^9}b zA?*WG6}Cu@MWRN3=mMRN^J#ChODli$ z2FhNOhW`MjHTU^{=32wrrPRm|;(B*}%iOJLyH2@0g8m$F@zV0MvE8k&1U1@y?80}y zf?=5DX&y^AqTs*jYIQub<*tXc2_|UoEHzBe>kKOzZY-sf1=O`spLSLv65W6;QsQtK zw`->7u}hht<4EZrCYUY^Ft{GB z11(;OgIJG!@v`z&mZ{WtdOEg}3^TF&bLJPV=xS+ZfXkiSL$^G^8D1aNI)3Xjj_*m! zt>QzE>ZIv@Mh4m4Nqe9_Ty4!IrXu0E9XYQmaMz)aCgHaMoMAUCF-z^n97l?f@X^V^ z6GKjYVY0{C%X@BQrmwa}@a+H=f$X?tW^WXczK&{c;D#1BUU9|7pA)Brxbix@Lst1xBOx}{27!#=a!!Vf z^X9xfqtM)iMq*(RpEfI4(g_@056TU3n$6Etg#=Neo$1#@)#nQ`ahxO3y&vb+bx zR;s%t?yl2Z(#h$ibZ!T(IUk5sk~do%&lTK%3F|Cwp3TKjL!?(iVW=Q~dfunZ?iUd_ z4%@FK`d3BN{{Ruek0;O1^?Zj{!}hsy#aTk)$;%GeS9Z!Oi;Zj)RFU_yOzk1R$39Au zt-aM)e?#jo$=KbVC@u2eDp)nj2#Y+}!OV|(*H_d|4m+36$Mt8X;`_XlABL98Y`5@L zlQA?)T`1D|549t_KV@a>?$cM?u_N0fot;OPk02A=QF4~9XUH5zPMEmxC==WIBRis9I6;9wXGrseNd zmp}SY-?bx0-mK=qtau+*CN+1RLvy=LL2h{r1zBJN(I#-)RPQQmgw0zk7;r(xN{?)e z)6f{)azorn_@JcY*OqXg0RDEMe79WOE&l*$&(mU}kE^96bM+G@iq$83O3ho5>XkYJ zu?L}4Rw;1z#7{lKyJ5t>^;tVEOl||H!u5Y*moDRrxK8FBrMDYoLFdK5X+ObvNxSo1 z-?EF5XAE$SS9yogRY^|c;N4jR?@K%fzYWVrqVdf=a~MFEQ%Uj22l*}p&7fzPE7qTf z6}(jN-Ma3^hhw{bq4Gb)c{GnwG5Wiyij&g_Bf=%61VeL!$pUbPT!bCJvisGyhd5QO zp4MtxRzuw9muMt=hd(7BjV$$jr16Xtd_PMpa8pnTi%<7)v~SH@_Bq8~fx;28=GV&K zS_nf}93YG|oXPe>k0P}Fhd$xNbvLW=mO5!JPVOB1hBc74VhQk8$V)Vl*&pEl03_DP zuM%9JNXN5WP6QoC`gyI%v6g{st@XQ{hdGBJtbNyvZ`=UdjvaMpq+6yP?Htg*RAYZx zt#`kCDpq!>S&W#Pe8>cqB$vsm>XkeVRmGMWXsGI$NX7!v0rj1>!++wuz1kOKH(n69 z%W#60XVKC9zff0Ruy(aw4b6rK->t(8iZ<@|A0@Bjnj6JO1YWAHHI(&J z+$qba5y&5*z-nVi=tdN{{Kc2zv3AZW;TR;hiQ}50|OjT>M7)>;^5Xd1-%y^ON-MfyP3?EEaQZQw zuu8bK(p21QVxpvo%TrNL-NzUsE_Y69mBiqMGT%`nYv6laT0@%MyfcKd{1McD1)DV0 zy|vlknj#$p?tG00kzoFzeF*I;LOV3Qy}-Zw3-<9Oe^utMz{b-B9UbcM$|#L-xEQ0T zG3D%LGC(Jg-=g(2>}749q6+phcwZUs81azdjIK$Kh?}0a#qg&ITCSWkY5Zdh&V4;I zSPNt=#=w<%{{Yk87;As&<~~18QurQt<+HJ@^>iLY<7Hz}WvDV4^m$pag*fqeM^jvqlFE?&(JaFTh$69cH&wDuIuFh=r zKaDEw_0$kMyo^&kx=5N%`5i*vPsFy%_YJ>=?h!*@_{<7^;o~8|^FH9XeP?S-ZA|o) zSNNimH@T0FNkdLSBLpV+7mF3y#kBl6X?+FaD#_=KfzFyZ^#aX1 z6YySj;+w2@4lAv@QWv&4XD^RX-yY@PuC($&`#>x zwRG+f44JKEM22`KI^zeEaxwGDliLOU3Vt4-;YvI?k+rf){QYc`kG)cGEUt9~CKkEf z%z>|H1f9=6a_b$fb_sQlfrlEyH8bjDrxgyTjJd?L0g>UTB3GXHVw2POih`4DOH&(N zlZ~|u{mb6gH&exDWH3z|zzuJz;C2hbd=YJ^x^WaV)U4MlTImCN`lBnQ+wM;8@Y&^z zgX?`E1Rpia#mU+I-43^*{wrUrT6)46NNE9A-eYx5s=AvQGgG~Qc#_f=oEv(rYjE3z zNdExUKh-f?eX3+Ge3uH1AZK}x$ExSOMAxUCK8K#?>vD^2G_;&k=#wzANN0N<)6jxO z7Q*P$=vf@KjNxW@VUpZ9u*neANs>ZN`)~7G%k8qFI~?=K*0;J5z0Gd>JI};&!sP4^h4G0g|ois3j0XHAHL_rbdUCZI<_dWapQP?^IQMHEp<5 zU0qQ{9aMA?K>54o>5vbq&^q{)$=Ug1ejL^PicrAGpC$3`juz^Bordz}n6ma&Ww=#E zP3d*9zG(HpUUarW;tJ^ZJ~7}SMLQ}kQaSTRM4j79v6YYE9ysA!J;oPQUM_bALF&^m zQfbf*+md&}vPn~Rt}VTUTHxcG@=z>&iS~POd5}?DOfS@9mG1Rq zBc2xCv$j4Krf6eetEmodPK`Tx9oBZy+OG#$>K{G7hOcuwd@<7k#)HoQ>=v@^wI!_O zw}tFn2gd%HV1?M0Ap>KULIZ=3wdUmu@UW}ycou?`0@G1g2?xQg%G0y7s|OHqH5`xO ziEVMywZp%tb4Ud8AOX!_e+inQ#B`Q=OBwXm#Q7wYu=^x{09<)b!%@n66FBWphtK~2 zwGZG|xvRM1*}yb>FKDQWK>jDHbLwhzmk5}0-G@`iW6gMnv#T}!*Ty#clPM#K)UWTe zQtd1{Msc4KyZ-=aJWowuXtLb3Hsoyetgfb_$I_5TtrmJWJl1<=Q&~?5ckt=F5%@Q7 zd^6o{cN)6dih*4yrgI(|A!c&ZkF0mfthRBZdnvr#xEk+ugj3T^=@c@`aOonGa!;OD zTkJZ%y6f4D7h92WtE`dJ%1FlUBOrUZFCBgwmV33ww>qh)sU?~rMGHY_+mYE%UL1CJ zS9GZ3w|1pj7XVl;5WHLg{9!z2J#INJmpQV_Z!gAbFh9of<|? zbp*eJ{zeXlqlCC}-$U44`}<`L&eGqe48xAsa-U$E%uz_l73}LjyU3XmfS2TprAtCd z8>_}}#prHmWXHK8P8_^y+KK3jY#>1j-|}|x{tf8I4&D#`3I6~z@+v`S8?!$YVeJPx z56xOXhi@4F0HS&KCch9~Kfq1$Kq9)l5ZA^FM^VjL!8B8|TxTG2@>QRRF8=`FCHW^I zvfko$sOe>C_k}dFtquYM9tAVpUi+}$KzV;xc%+J6An9*+$xS!7k&&Dx-em{WiL#6l z>pM<9NdlLMI$PbJa!{qek^cZegw;0z?F4!CQKe#hQ(pkVw-S(PF9q3kx!CX=(^U|p zZv2y!6;(7c_tesu-N=k&IC-hU({PcRy%p^~1a6D*Ex?R=Gm}ZB^{1?St7(2SlbK@j>&6SzYxT-6+4mIS-K;G>A zW*p%Am6n$lLD9H$NzhfvvE2aw0Hkssh^~kK01D-;v!yOKOMH%VLtNr<%gI^3!FA{8 zZ^B8iUBAIdRYXfn-i>vnsLng7nEd!*z^{KsJMJEirK2s97QA`nWq3l9hi;NpMH}f~ z8(i?$Ioz5F=2y6Ac&CJ;;p$6N7urc?WHG`3el-vP3pq=DH;vQ8+(TI;e0^P7!dmv+ zUhsS{_3jFSDmZEigMwv+uG`E4yiLWn3(R~^a&xWIf$35>9Xk=IoR1~zyGIxB(|8`+ zMLp6cRkDg8_MTzEqaiH55@n~KFIUQ|e{NMVQpPHwsCy}=r|&cg8;`a1Nahcc-gcpW z?B(1K4-45~mGz9Rj;30nmjf$K2`@XlZLqkvX%w}UTzPh`qGRQS^(E2B&O~5t6nikM zE-$SQE++?G!WVv&^Qg3E$zrrHoIMScpBO$g*#*Rol)J5Ao&gQ zufXS5UOnQ7Z?tcQt$;-za?atn^>Mg9AoT@wUJ2v-Lnxzl$T|r}Hb%!d41ZI);MK3L zmCO{l4sLZh`(x&>I7*7)aJW(1p7`B8RPBkShdUF`9_2QVYVf?GrivP#naOH^5)hLX!oP~EWJg<1Dt`5xmR{t(-o?D3H77{2Rye# zr|xvsFul);_g^DgcW{(#_b)1b7Mv{=4c3E+X}xbvE{dt)pchCt4mj#U^u+akrZdcA zIb1{Vq^XWh&zT$C45Hx81K@Ip!_D?rHYv={!QHI`#N;ocmiPCZBTu$C+BgGV~$Ee=|-z?=}e+FI$pN_j$w@^hW zhB{o;J?Efi;CxOKDwEsU++fcC0EH`J*pm}~it+bt@lnIa(ot5)-7(a0Ij_efxL(A% z6D^LaDFJJn5f1l8WhAGh7`OOS_im z131a>SJYja@uaebKFuvKCk693HKXrcAF=D*C6|Dq5r#5JrF*7+03_#NE54O$3dPRr6hE)%p0jGQ$i!Pn!Cki|AZDyjbU+S9jtu)~*F*x{uN2G<6a_FxSfH z2Bsdmcij1vj=T0h#DcZ&vKZ>AM(lKw20!Gx18QXsA3=b5<1Du(m$H!))8Fp(eAXtnmg)xh@oq@7(BGEcj?`4 zFEz^^S*ssE^-rg9bhu-9?Xs$e3ta_#!YJp^=4HdKNAX!}h@O^r$1koL@<4dVAM&~^ z*J@TbR!GXmy}ccXrtugB;G? z!x-IYsA$BE@1vHMhB5u*rJ$d1QR=0$I;q00_$pnDWzw~+&&_gn=`B-J+H5k(-oDHY zQ@NHnlbjX8rezQv&L6oDWuWi=N#z~a*6FgH= zLg^lM%SK3A3x5VyM-O{ZM=X-_5uKZ#5_+xkdtu<)HI_St4+CBxp_c*~pGL@-{{V?9 zhVS@H+oEH=3x?@qBa^DG5x4FENj7auaN*mX^BX;w*f7!XRg>`bOaT1@hrw6z z7vVX6dq|>x4AJf6k) z)xKHu(!J~rlrf{$Bo1vP?0l1J6X>k{^ntBoo_v;?-NEyghsguvgyQ3IYc%^^+-5`aQR_Izf$e09YD=wI zZzmW8V3l!m?XR>>D6osjgp7vv@#iIb96gu->nv?Phq6Y@YjY)m`2Qo?vQC(0u)E~KO_$G?gaue5aMc>dZG|kM%aOnpAt4JBn zaIij=8+L0zR{TIJBjI{#7_qKTrgr>^<;rSUmrl#w6d+ShVvVDgd3+Xb zsi#%DMOjAH8ro@T;c)dN11xv?g0B&mvs_@gy$Bl>&!FNyDWDEC*DBVx89U^Bj>_kU zc-AWU*=@NpI4k6TsZ$d)M)US$XykubEL>;6(7GpU8iR-+#el}d z@!dNOJhIcrE1BIn?GDgJ;ViIM77nF0V*@&f1J9DQbt}g`C*YR^)pX1(gvJ3o<}gp* zqN@!{Vy7_E0MhNGlj}GrrH*s6q>^?S>aX8%x-q5hsxYu_U4Yh)xn^I)Ka4J>l9s9$ z&Go8kNbcF@KGM|kG?Z3aS~_1$y;MgYeTWP2PnpN?T?>MEbB}IT?R+U`Z45F38&^;S zP11hOcU<#`>bPN(RGf;dw8L;@Y%HSYwIVLo!FT>k)Xw_Z|T1|QfT5LL)(xQ6`qR6BAYj6Z*Sm370hDoSR#uZpg(9c8G% z`32Cb{{RU-3u6mmw_WBw1}3-om6^Q$5cWyKy&r}vq|YsLwZF-7Vc9>Ko!2_9?D3ZQ z38ti{rf@qrG-~;klHy(*xVRQiTT?3l_Cxkto9}E6uPzX9(_<%*eA)g>9dqLwwPcQW zqftURdd<|gmRw&IcWa*9_+SROlO}QYim8gzN)Mw+L#^6QWOK8R1-z-cS51>F^sQ<8 zF$GsVl%SVQHmvM)um!`GSl>-b=Jy+0V2HjjOr78we6Xd(9FfyUFo3hw!?C%_*UMXJ zgkiK!mEJCLWtDJNzMeTCrGql+1f2n`@N==YdM+}1Uf8GZ?J!l`CXmSlMxsXOZh#sx z+>m^6(O}@#VDDu7nIu*n%~;(cs_AlDX`2${QxaOTo*7CGlrjt0Y! zR?STp#W)V!z)~pbDr+(+%Nvevd$1HCw=jAy96&YT6pN z(!)?3<{Ul1>KX7=?!H}K{Z6iwoMUscRRtK3ce8Juz(2)EmKt|#V{g0#yOn1}+b=G3 zt!(bl{Q9qNkD9p^Y^iu_-L1g--v0mL!M{pg$uj72CXowrXi!*0}UODk!V1kYV+63>e(FLpy6HZd|yjf5`VnH0nF9w zKjc+wYeqQ?pli0<1K^rTTH7V@loaL;18OetRLe;tOa6uWTr+L2$xv@-a% z7Q7Fd#aZ}{$3zP+)_RMcz(iuH?#ylKIc}{#Df>B)j9IU=o;qrp{CvXY$Fp8sS#H+0 zO9y8&(z*z^^0BeJVa%zZ#z($ z@;im@07)2iD<6-Ol3icJ{onnQbE~VE9B7u8x&V0RQC$32j%T@R8O*HJbgR|U(G?Ji zmX1eC&`FfYa9nw9(+d|}cVMQgiU!FobAFRa$9C9c>D2Q_(Kp zPj38PD=u}hMrL!p9-*GJ)bls$$sTuM@#MLBuA@?8ig9uL)opRPLowHMjmzB5t_~8mK_es>2kP?JtJ}@e-9b-E%YAiKRIhh=UFfui z3Ta)Pn&#w#fz)AEO-WH46&D$6Xsqm-R_Ui>wTT)|a2vh=>+fED&nxoh)QEA~n^fi* z`^tFQ(Tc~NG~+(Yh~^r6Qw*V}naz@-M^5J-&d@hsUQP#*Q=(uMZ)+&$6$d4?Q)khf zGJSG60ggwKZdOy$`Wx%(dxyA;KGwiPS-$SU!FTmphLMH3C$>kvDrc}CScy-526sAq ze3PDs4M7j2?<-?y^}RFbw4W?Dr#6*V{_}99ww#VX$a?&)L}Fxrh6dK zGq{1^`q0eN8I0@6Yj^d7hg1>WLq`P08D*A^pp%GVXL%7lJ!9geL2qq16_plxXd#K? zT+Eco@GSHAC*$uMeXEbt&kYR^b10^1aBhFPBM2%g)4rBAP=>l zt?^w3!^1Y|xK0b@WcKbfu4GMY6iuvbPNrx+%i}(?{s`#nsNKaz9yt7Q!xoHc7Wx`v z3}=uH#@l1-Zl}5FsV4oo<1MeR$ZXd8gtEzo(o?s(bKQ*Y?uzr7lRbD%Xv)f5zM2Dz|->XtaUuXKmWP+fR~h3Ki{ zvsf*>Puc9kXFb)m?tM>s-hh{2uz3uts(6Y%0*;zXz-X?tM9|ks99JRTx_lglowg-* zsBOmQYIR-e#>#73&SaL(LxXnPQ+bn~n=Ch^+OL!r*=-jt8sa;gu4t{Vy~^5ohzE9v zzNr0dlCP(twA*SUwcD&5MMYIJOJ%IKx#kfj#5yqJ>$Go^u)@@md8YT`G5kYq;EIa& zRN5{U*(Zem0BMtoDP0iHEe&gSbf5gaf~uh)WunQy*8$rT(;-M7G^#gd~O2(#twNM*CpdC zWuJ=Tb;`q0WriqZ?`}8QwNsg#d9*`<2bMFRB<dt7?=9MKg1P zBstjmV=100HSo)Csk`1S7nx)M@Y`7!(;GT#0{|2Dd6gxY`aSz|inv$5oBGR7nj~ z^sV(fx4DNL&JGu2kC$=9mO7^(wBIhmNhzu3sAzT^L+}Clq|3qghUCu5I6wSahJ{g5 zA1rPi^pTwBEpZ<;Q6)7=Co`8165_b>%I&6^_69(Hp7msl9LIi6K0!MbuEJX9%f$4@ zNXvvVx!4~ilV!d~IDI2`QmLtCr!Fj&gHuagk&r^gw3X8PJEV0l3_A_=WRUB~=9%Mn zvI{i010o~K&1$%MAw$5W!~ppNM*eF7+9gQ*SoSr?4s%*Rn%-U}nvQUJt4eBZ5Hewq z>}30uSQ&N)Qnd3%eHDOX0k|XJxUZ&NPw;eW#e3@*32~E1&yfEBmFYR^q=uGB9_^UI z;&b2^pL+wQd=}=_2JYOspz_@R0K)71WpRBMXZKr#xM^h+gptT(5U(ZHD}>#6n~Ex( zTf{dCw;Y*jq<($wDit$z&l^6|fJu<$#GZXrn?p6+A{_8?vppBvR5rswz~rRaE|oLI zBuutTJW#l>J~v>Kl@3Q~jn42HRlF%y`0FHZ<3~{f?M>Su#Hy`hD;yoBOGSGz%$9-^ z;<}8CK9ZrH_;QK0bt4*ho}J&cKjACVE1Y(|(7)f0=&F$|WxNlgE@|tomBlvNt7#-j zGa7uMa8lbf+>che*&yl3H#xjHnS56HbG91a`AQ?nJ)NUdYS>SP2t+F=^$zYxNa%b0 zKglv})tU)5XtB(>rQg*#PHKP=GQ|_X>$-x3_-3SIo*WQ5+F(XmQXTPXrs9he>ln!x-dS^=&;j_XGfitF*Wt?M z=dQ{P#)8u|ER1CZbt{0#cxm-fS?hhiqMnMXj!JrY1o*_x8GA9{tERJ8M=6gLH<=*fRs`iuIN#DAD@C-0uX(X;;2Rfq za7R;>Irn?KBQiCZ@hl0key!@r3H^jaRW{bg7PxC^rqd96z;*~e03K)Dp+_{apQDxS z*o~Y<-U`TGrzb9@r$6YzA#-D$nJtv=_#&!~=wSF2a%=7GZXalFe6~sSyH{INhx(Bf z7$}v;rCGY$rT94^)7g(IImh~f3b3YI?h?|x90#`}W98@z0tptpSbzN3A=2eO#yM!yPDOa=p#) zb1;lhNaL111Lu&WQ&XC9?O~eF**)>N8JCv1;crk#!lJJ4@K$T9H@p%TpjGLYrBZ^@(avO^`6j2Hv#vq?QE!XDI;dA zVT~Vh^SbSJ$z6_`Uh3S*B|ZzImDv6%Oms&8N_SZwPIzb<(&MjqS-Iy}uG8)E$5tiL>P;gE$B$67+z)J% z(^z1OTcd}kJe3=aBcA7JjwqvSlSM0JYfH510Pni8;A+Z>7-}S~ru2CnLt5Qo#{=BE z7R<4eZak)4+tXIsTFepX@5tx3B)>c(1FbxFZ2b6OJ{4(JatCP^$2Tp_swwT zb#q;B;rRam=5l@P=fOSP?I-A8w>`NdcY5DyCl7d1-F~E*+MQ7*3v}dwgRhj50TGqX z17Ygzg^SYR{{S;{pDXZ!Uf?+%XyTOQxPpJ0MMc6nTr})8V}AtTtEX#>dYPvltCt_} z@_e~F_T-<01-2@#%&%~;de-Wu)J1{pha8{gyptnINmK?xt<2Yk2wORDiy!P8UIbI{F1d)ITS$L{qQar{p%zfJl$bII$ZQA#k9d8H(|X+%;^-sGg0PFQoj|jP9utE#O)3wY7RSOE3H%)ZMP_?I97V- zX=9!z=@5qvi6a{E(RtncLr}((q1w6;AzqB_?uy$>!*c#Ej-{c{7d@|eaL2`YUZCV* zWnHJRyQ8jEv&i~k-_yCJuP1UkjIPDTw`yu`AE~aQ1U*ly<9U4OCRy=dLobgg6&vq0Sp;Wvy&*N-LGf+FaU5svl1q43ah+ z=!|%u0J+~3J1;LFa@EayCiOk5hr1k(#BD6M+-BkV9O<6slQ19>G7ncA|tJm|9%dIy?sj ztfAm5OQ-FAPl;d=pJwRdQ}O1}!9cXFqyW{5w~0vPUHd zV;UHH!0m{}U+0nr;ebyNe(#%^ZmQYnVQe0mDn<49f?Z|dqjjQ#1pY6l z&yI|&XJ~!Zn};qJXW4;*k{8Nf8%RE=@>+=avxj8>w(n1a;&7X(y;`mX)U3ErKAGUQ zQ0W2h7(oTN=(vB)pB2kFI(z=9{_mFDX>)bwi0kg}rG}Z+lNr0H$EwWq3YnyUakhL8 z7h3!!dpvgm@x<`c7RzX@E}q>hU8oil>=WYNOOl>v-neh(`*Y1%LQ(U-; zmX@+tMFDoH92=`)za>?bcflb3ICP^UyM^UQuRvbz44<-UnyK9orVNr6X<=&%bEILt z7nfg+L=IgSf8)xJLHr->kpf!kmkj5wNhNR>4-R~lzQ;rC-a$*;(y`(Gme)A#g2D%n zRqH;&INFbjxF)NI?lmxjPG1mot{$t}F@v6t3&}hX%@>ws;0k&-JhXGn-;h z>UioM9W{Htm96^35?ZyjuEE27o=?O5sMcRBQ&ZR5;+4i2*zxSV?lp26PD0G~^%W&2 z5XV<%4T#mWnua~PJGnt2!)=P_{g6o|?!{ddPMTP1Ueix2vNgS<>f_CGPr@P_+xUh` zy6aT*^pVs!2}FNHN!Vcd<0*VWf~BaOIlbhNzkgw*=jOWQtFyTHld}lThMumr{Z{a0 zi~{D-2iCwCru`K*mJ+V%~?)V56;I(-?30w*7Pla%!hhDOT-UEGc2 z@azzoo-)y>l^Dw5umKufRNgD6rVvdtrHFE4ossfU+g~eK8rHF(a`IKiXPvePYH{Z5 zr*dgxYiA>yLH8>j8>WJ;^A$wo8CvazGlAr@#l>0hy+f(*P`^3OS^@J!ZFtVjH3bYc zU)o&UjZcsbv(q7D-8059D~FB>?A%9B1@E$oqcejlqNndcaX!-z^9z7R)6bUKDSQiq zQ6c{Q-@KG3c;!2X`jtjD2|yyPjZBy$zkdG!td;3?ARp~={2#ih-^DGUxxuaWov_1mu)MZ}$@*W7It(h4BPMH{1e4Stci3%_vp zW4spL2%{7M*6NDtIGJ&Hd2yY)ImdkA$H{q@sA$5-^6gqEek1}^2SM{F6p@60d=0+yCkvGa9!(&1$(8a zw$er|;k#&!qzpUl6P|ub=ax85?2NIOJv`jojdqd$0BF=40sZObh}YI0b=3#|092~n zXQw^FNofm_%ChvzgU+f`?|DE~P11M%`mDhzw*%yY;MGl!`iFk{FG!H{o!`Exu#(~b z0KBQ8b|-#6KJVT~-@Ek3_mv44{{ZC?1oO(Ozwggkv;L$z;rdbYS^N>=hs_gB2YNox z`Jt`9Y<$*z1R4<|bL|X!(%UmW*!isA%@#A%pw>k@2Kw9cPsdds=^791EM8IL#R}ZE zskr>(e|aoau>Syf-`)!bHbu9IMB9Do5&o6Gyb+3s{{TwQ-U|V@i9_a!-mF_rl@ZCk-rB84fCAsF(2ww8~*_8p}9eHW_~u9{_;qb<8JxDe$uG;edv3? zih||!GB{+^Jp0~CjWtQf=Uji(tF>og{8KWMxkX5tXE^O6zj;}3bp<`!i0Q4>vdGBi z$OVqU&N2p8b-_ z7Us(EXhHp6UJ5P2qB<)GA6kH%czVS~rno^tZZlLkhU7Ffb90X-fz*{miBHqbf`2ww zAMNIlQPoq&4CYSqt{G%}=Wq*hMtlcF)o9&T@jR7*tdZ<45s&WM(0lw>IQ%OiZ6y^H z)ZWi*c%05!`BHcRq@wRG7&s^(D$=Z8Rw4;mKF=e zV+bEbf%>X!cZtum*va=OI}I;tJW2p=iWiSKh|kEUrR_h6H0AdC2LaCxaQZ1)IlL4#aLNpj;|u7oWZYlEmj2Id z)g|r~!r2pEd5v0k``3)WA6ez+UrI-s>fN7J)7S9LJdjAq8p31L!?xU#O;scCpSiY9 z%ioMR)z^Y|<-S&X@D$=3eTRFQBK;Qb3wzt z;<(RX)t-ymB_u`IZA5R2-bl#)3e)B6-qkpkp0=7OEEIl@=vwI{V|}SN$R=2%SYw!7PU}H!n%i%?RI?RSugr21LBozT@xYs zOqXmtF*AVTn@D%kmwHV&9;+RmW5}$}3R?}r)M{#}<#BYGYZ=U*LvIb2ZK`xqaKzBe z1DN&@;72gIeg6Q7X}D6Ew#x)={Ba~lzmj2=#NLf)@goc9EWR}PY;`8`bRHkH@dPx` z#K?Ufx<+j3+Q8<<-g>l*2JPajc$b7}nJY}3ddk_H+|Av_bg|!LIJ#Szg=bT zDz0}5qVfPZ(9(0Z-UI@Qy|7eUSu^06w|yn@fJUQ+=iSw7Y;Dn};d?-FH9S@_Mh;%z z>Z6Vn;W7EnDsJ5KD+1=mI%|DpGg(T<&oOAzBek`IIL_Sj%q<)hwVY2zcv{WFR`_m= zbLE$~W3^O>*mnTCX9j|H=Y*qZI7fh)YxKD|;sYkFJ;10M@#D!0ZHt2?lM!G2BYbH! zjpkQ1h9>~_R7Xn;gz0p9<&I}h^(tvhO>usj90T7L;oF|E{1!3uby~~f$8Kj1@G+_o zNpoW)W*X+Usifp{D_L#eZU>f}Iy>Bu5fl;2r304Sz7_4MF{L;MHSl zyWFhqh+1pIM!xD;-ym}y@3aR-pB1U5cI;ej*^EZ!+N%!|!Cx3?`XEg<{un36&0y}F zF~ro@Dhex=Eqqq9H+elwV7;53q!Y`Myy1>BrMwYMF%Z^`jBfT`N7*Y;O>nBFlUx$M zH+3NY09CCFFCTE-%C4=gu=-VPzBT~C48{BP+dC}9tAcoz;au7{taYL8CwXHbpm>sc zF6D2d{Ci&zmW~iTEH?>k^&zum^26$4p_=yU4l$4b4ngt0uV%$pWFLcD zYu?MR5k6Sy7=Df&0C$7L`4YM>Xcit8;VUHcR@;396&ePR7|`O1zqT8!^gf9LBBp3y zVD2Ifz#bVVA0>sJx_gajs+p~PpKaT@b7OJBc%E6iw_gq$QDMAs1^%i?Y_(2db{i;* z4cPQ2W!*a;{2w^0imXvoU-*Ki+icDm0FcYx0DLomw7dgilZg0|lH02cdDST=lLqB_ z8;SUi$H3e_6jEttcOHsbX8TXT+a6wPt#@+p*L06Jc8l3J2k;$z4LuZ-T&iCYe8#Yc zY2|_%1_Gj}uV{gg2;pvnfZ#Al@bJQOjrgLf^>?1-T`sBRJclL@jUGW`YOBkfNF{Gt zH*dAg()cMO(;k;h;jrjD*EZswFM@`5FgVuNJDUTuk9kPr3$;WI6qLAieNn?eIndr& zDvF-hOKOZhq<7Ua4NoNJB};T5Trb>DTs6keMU3(hjX%Xe7oH`fczdd<90B^rP+Ewo z?iJMLrOu8TM(y8B@W_XjN<~egcCrH6x*7c-6A)zMGJBS4X%5xLaymxEbEjs5v6ei7 zJjm*{6t%SVFbQd-WRaduY5xG;7dgGzWR6$2xO5c^-SN9+4~Op4+A9lPlLL+=XY)bS zC7(QL#5J?4#iOpO@}hgx{bOKo#%CW0Ek*0?%32Qon!_l&By@kGyYFW2M)f}OkaJugeZ zmM{l-eV}K)W~t+VxHvadoL_HY>I9A3?_N`X5lG{mQvUEvOMa;%X`DUfJ0m_`m49cP z2|~jZd}t?Y<}*r6tR3M$Ro*Qa%5J8?!R-9s5K}we@Vw4%Rfn9}O*-vVf4)C8Y}x1X z^i0;S87E6G6wV~zdcjyXfZ}xo)oiV|daczHUsl1a_-8vWFuLbcpU#v8)}U?P5&}$T z$)`}`P8V>j2IFv`m4soiQ`I&`H{KelAmOHxBc!8ruH*wEC1owVNndZRimsM(9ML;d z9i%%B@!e=|ovLx2>V(kTuhkRN9D14<(YNe?weIcR$NLVH1_ylA zj2)Zt8{%xW2G2uummWINpW z_lGFR*ofutPyYawSKO?dvavUb_JvG2mVs;Mp3{JfCC zyPU7>6lIgOv54ZXnXPp`vJ5qGpQ!H)b4#{UHlBjrc8RWVGrB66&msu$_Y$UFyY3&R zlGfQ+>Zg}UMBHWYMDgoLd@|F^$CBPSl}E*uuvAuzQ8A8i{wihd2?Hkzp5+$1s~{wh22YZ@n@lYabNZ^vOix)=R8B^8f~@p~ zRX@96YX_`$0Sam;rfD))X_c=Ha~RBqlbj6YGcbGV8#j2!+Z2u6$C=8L0~-}3bj6^# zwE@`&!Qb+s?G5d4X<=}ZHyj5JMJA3Q<~|P%!A=-nYHQ(%rwn&`H^9I@(4vo2RRdPe z1a(9@**nS;6TPU;J_@>Qi3lfUGO%}5WRGT6sFUv=M&uokHY6b9I7VN(j$A>?1?52b zdM@!Xte{*~59=svQSwj@AqCw9BsFjj_#A_=2X!RasZaNk{K994-IfOR=oYHqOeVl{ zPES?b;p3TO<;9*?7H!sOTM=VjA;||Pm`_ntwwayV8IKZH_Jf7yu6rYZH&f0Vsn{PA z(MkUR>E+&{Qw=pCmTg)?T63uRc&=~ZWb=zZb&s;L-!)UAx>UZF7ew$SY{NS|s!Ay#ETIk{<{HxcvOrGS$1D&2Ro@bDjsC$PitlNvqH*>_(U#>zwpPzY8>#1u zy`_Fy_|H#r;N;=A(~}&e_G$;~b!7$S^4my?KK3eFR_lhLBM=&-%!SNiE^^x6c zcp0(P@ZBJzx79d0??%kYsPCdjjf;FTk!;Kt+C#$Bt-VoHv^A7f)cd*49=iwx7kL(x z-JX&uD(ZhwsC}2bK0cghEmMMSe& z1v{%K@AjD^hJ=|PS!1rw*IifnCjhi|gGW;6oN6jczg34 zc9-cgPSf7fTwEI)KmfIlbBSZKoMWQT+MhG+a>v6QNfiV#IsN1nnxirRbpT{Op**%5 zM_f>YKy9)l&_-2~x9ZLh5RTG0`EE{@`jOoX`BJH|u=Zmg1La(rDtAo?{Uoh9X&`3@by2`~i(_jpcFP6D*nEl#M)^RGa^|W;MC~S0= zI%>Q1KrE}OWJS(_!0aM+NO8wEe3M4|Nl$H_md!;4gj^yKSGJYTaG8MTNY#?ie2TAa z(Ufs|4VD8;BxZ+6w|L>^ zyIY)I=Qy~WocOJf;cg`~BDuDAK}PTIhKq1?MXEpUDo?l zcX=2+QHV73QlzW5L~v?i+rZ^BCED8G?{#~cQ?nU7({zCe`zbGo_#{btqxzxe{VFt< zsB^R~KJ~|!Wc6B`b-L@tIkLfTr!H|})aP;o$#cFq6%`Rf95B7(S#i?b^ifRrcJ5-Q z{ql#VT^dL*)91=ue>K5w+>_FZmWHI&MAuVH0dUS(aI}&#LzWm^K-)S~O|QEH{{TqW zd!`@NDYf@p??&u%{NPa@h3tls3T_*S8DVK~Xu)^{E1$w1wki(C?lH$B9Qnj>4Gsr6 zE6i=WTt@znD}B=c0Lojh74I5f6-qi3ho4&B{Df|(GONd!NOTbF_`C+XD$+N)(WXDnq~M~j|Rdb(KYX`e84 zW3T}YW6n-aJ_?#o#Eg)#Nb1-@#yu(A)QUQbdtBoKE;0;YkuIPI#H*ugt5pq6P1o^g zn&BMgKJl(>x|&CTJ}UZ77`w56itC94>~owf<7b+oXgzY+Vp3hVf-=WNE+L$L-Ps{FBe) zD%qQ%*SSu1^{Bwx;1a=*s*X=clf5i24k69}ZbCti$S3NHT~Q%x`@mdItYay%Pr*Y* zOkb)KZin?kkdu@&sl3F&M15WIsckgkdf?JF0Ftk7j>H(|pm5A&)XeZXG!L3HXOz=z zk3PvL{{R<9_DZyCNaPU1x`2&oTFZQR_EAYg8TzWp!vRN-p`|rd6Iu=%kBY0Pk3qoM zU)2)eTG5hKRYU{==?*c-BQ`6rQ70ZiDYA$N!X#s2gH6IT5Y0(ZElhE_k2*%LF9i>Y zEp)UmjnuN2_n9T6gQp(lP}jUO0qz699h8b7Texp3Tc)HAZ4{R!nHm}Nll8RSZ_(PtYBjX8r^{MQ7yKL%Vbe9yw4p+bUVnMCO~&3H1aFX>p7<; zTE_Ty`ffXf-c!jEJ7PSxQ5bUK^T|0oh#GuTAZ?C|sf>0l(dfr`@qx0IVqpO_Owh)I z;-mwS@k&5r8r{b40si zE;^yG)CLxCRqujQ#alu4mVV0KR*+R=aNtq0uf@cbn`LPZAQZn$JjvXw{g$xhpA9&y z!5f?1$tp3mP*$NJUPq2Bg`pX9vq_@3Tj zeUH%i`#3GXHS?cT&%2gJ=d!}u&MA$}Y5FMK^W?X8`8D1q1Csz}$SiMR&}+NaOh(O( zt@)q$T8haG)wL2HZjs0DU95cdmi{9j%hONZk(4^5pu5h#k;#AKI9zd-w2((ksglND z1LKLM&vRVj2U2jMu!mf9gS%rNc}s?=;^iwC>}iY-nOlxK?De^ih8D*5=`9B!rY^Js zt?ZEVVCRw)!2HhePWnPJKnjYBuxl*!F$00!5rOVpo<^2DwDdS#hJ7xPz~dY93zc4D z(On~e!(8G*%4CACxHz^;r8PUiEi=Y&cL&xRDiv|Tk83!h-1nR?PSWGrs%|T3xc05)v6RtDOX?k^pz|aQkDB8>t=#GETxTok z!waB+kjm!rbH?X>xXwNWQ$fTRcLqQz%x!SNEp88<>Z>^?%;hA~-!$S$M0q5VaH@=? z^V`h~l$R5fS3$46-0b{!Zl1#$1EtN}pL*`yjqvTx<3S}=8zr61mEH5U z((Lm-Im+_IuCiG#wAD3Ix75&6jOEQZ*sl>;?BctM>Fmk%p0ugLRSZ(~qU%&J(AeN!_lU?q_rc8UYx| z9P+MG&q*tvQeCfopxj?cbF72XKNq%|l3AW3iU1rvryZKQMZ2BFQU{uDyfs^&6cPP3H5WKU*F2 zL=TDBEPd0mUd(G}jnzC&au0=?Mw;qjkG)FmlJR+x~H=a>Z?~t+d~4*NH8xFRp@-=e&uz!En?Dun#g8brZAqb!odT?8c_l>J!8_ zJuY)dc&esx1a%wpRpanz;k%4l?YvKVjv1YnvDDE&C&w*2u5EANYO0%M6?M|_X0tZD zYNBOqj%()`#;i0QEhmw70_t6Wjv%vMxQ~MFua*|lF_Qovp6%r0EO*%>1Y=^7d1_5L zFK(TUP1(od`Kjw`ZjsdonH*@#L?O7`xMX!)s~CC5Lxt#1!dr{Dmdkx-3e#Sy9d4<& z>Zpz2IJ`N;V3Xv${B(yApktW9Sy#0`2di}yy_{NVt044;IvfD^5CG+{TgP@|OLm3R zTPv{f?XplcH{Ig09fUW!?7GC?8gPE{x>dTV&^@kWi%8@`=kL$+Tz_i|?lO_~EW=Z- zONCi>a&$z@ed=fb088kpB|YuZq0R?8HJKj}wz6JpY9lOtLyJk@qDx=u5>GUl22nag+&9Nczjr^&7pLOC z>5n6R85H8(U1Fn(d7j5dESQnm#^5hHle+Xr;ju2hEV0H9vqtxP-n{O(GPf+ZT)5+o zGg1pu(QsdI*z4tdjcdeU08d4nSKJe1sJ`7|p{I@RboCKMmX1KXE+^m;MRdO19T;K^jg6*G$O*~N6SRwN?CrQ*Wd=mI8(wqx+r@Y%?vr-7@Dk&mp?#r+b>x2AN-)MYSW4rb_ z#5^@Vn#+{|Yon?yIdfV-$v!za_%3O{)~VpQO*`ifk4sowJ)(J>cm-wJrA5Bg+1*_o zBus_%voVgxH(=^J{1+}b-EO*FM^%q>;-2>A-IPHEUmjZHmk!e|yb;W_WqYRA#eJMs zQZm_gwo}bhDS)!=Ntj6IrjyuwD4CM*VMra?Ix4HXL#QAt_}?QT6yF6r~3h=qpj^m)Y`$` zM+tP{r<0t~k)JNBwDFzBrfbQTS!IdU4{IMYK?g`UYcf6jma&`mvN3i($nEmedR@ey-fLA>(qmT^4U{xosQjA#1)WL%R424 z9dFiKhkcSsJl9O@wz4eC-FtgJ$f0u{P?xuWfb8~=G4X>_YYlUQ>f(SW9Tpc}B zebYNi(ue63(0klbyYFD7w+ z6C5B9d>4vy4LJoFO00VOqC<$KHz%bw%Q z%&cLCI7yB^oOs>cx+h@VQ+%}XofU>E4L?;qPnpE0OFGBTSsa-v|yJ5KH+Hoyklx-&=u-X)t`0wJ(%^Nv{6@ z6(`LPNv{6@6(W`RzPJ1|pCmsQ*8c#8ljNq&BATl2@lvAnn&bZfikIf4{x7Zn01E#A zk{^rff5S=gK@-xe{{Y2F^Fz|Bzr{)OQpLWv{4}2=e~asX!%6Z(NPd-F{whRXm0kWS zUz(V0^`GIN@?&RvTC@hMeAmOXYyzsfDJcX4oN(M}FwfG(Y zuD`@wJ8I(R(^U)GZS_$BrQ`zFG~o5>rMjx>+(CZf`*#dbUhbC}S#FLN!%rk^)LtFP z06Dw(CSkD&ZggBjPj>X%Oq9RbY5Tlrkn?dqINQgX%2)6o2;FTk&`Eo(Z1KOnZ0x86 z2|Bn&eNmqUvb$NhmQK&8=r|Ig^%ZndnI)C*>N2p8TVOG$o{LHBm&^pt4m{dI=U;{a z(`UCmyPoc!_!W)5@ega%lGiq-nx-#DEo)lp4_M^y$SltvaYtyJUqej~5Yx+18zXL! z4GC#zX&?iDrs1bMH7*Bu)3Mg+JI3#UPD?{q7w2he8T(PVe-NYS>&$!Bc8Q;j7zR6+ z=iZX*nImB38NeP`$|khul59Nc)}Ls&d=WmXCk@NKAxE9K8OG{5q4&xhb~s9C8gsEs zuwpKVTTmHjnV;l!b3!?^k%(9laZ=eH*d$wH8mR3pH{S&ea8&ep_ zB!jn_@iOSI7bbkSEz2#_-oxrMmxB%&HD$(a*k$3`wnlSsi2negbFSDe^|W?KUMmcc znY7BxV}4`cy5($}xRN&|U^`ujlW>hjton{?$G_;i@aNr+ zcO#2i--ZxkI(40aU7L#Ljl4NLk^&glG3DY7g&RnwAz8@pf{Z}rZro1@ez5KJ+o2aC; z)Kt5>(?=J4<~-m$lj^t6n&GuF%T-rBT}?|}G|%rO%+2OYcK9uw^U`ltFn!NBSgwBhX(t=ZlfUY0XcPcxSHW#ESQZ7tSm&g0sh$D#1^Q=0fc z6hh+KP#XCDtDHFz@!Y80?70PkeCh?HQh~m6{N3_WJNN!8?%HZ~>>?&0GVJC&`_*7q z8&zE`($R6Ph{Z)y3o0X%I0iPm4gj8@f-rYaOI;jXR?Z^fDvF9gCm3U$^wDQylhhDO zW?GSRrHa{cxXn*Q$sX!yXKq^*r+EPbCrLTvs*LeLBdQxLE|rb$;t+xx`$Uwevb0Y- z;Hj2LWOKbBVCdD8%b$vf1do}3I;#d*m10K;mj$s*$cfqjcOMYt37;lb2e&e16G)tDYh3(!YkRO}SOnMCO))SipT#=2x7Z#W!dzuA`)+rlQNW388!2Ao%4oSK6l@ z+br*)it{AY&-IKoA4nVrhE}_=`DNMdH>&G3Yin(_FQ%wyAU&MGKJ`(>JV#q_sC`Ys zwpyv(&C)+5bN&+TB98e^+FH=sONSues{a6S@$yn`HE)sUeq~_xRc%}4cFrWIZgKUT z_h=2=>Iyrg-+~MA4X+dZiB>d4uEp;p`qHPQ6Epb8K z+?*M!L4PM7aKe;RU2A0GUJa6@v9ZfRIX%pKG)8$0c@N;CaRUsz8(^L?b+=StjgCnx zR?9`6(@{q2TjYHe%AL^FG4IuhM&lkuxx$SdneMe(t?!JwI%?AQo}Jh(IrnZ9Lp(`v zq@iuH&2W!b_EwkFo#ctfsqq~4Sw1JW%SlxCI1Oz+zMe!K!-tj@p0e66d}mc(S%%P2 zkjFTj2RWl9zTv)9Tb0fZBNc`=0BH^*z`=YcY@aZG<#ZzOEy2bqE%FoU`tj$(s9Hy4 zoUVUod850+{E?zf4)+zjaoHnKAHGsK4buEP(g^0LBjk|bvUQ^&2Q*;?&j30h#4K#? z(G;-0Hyq;`80XPy>*-8CLsd+BoSeO_I!I{AAo9l`J!hnonfASI;Xt#AalX&32QJK_G+7 zAlipxuP+s-t*o(CG0vW9cDTGf-OwbG204{mcdfEi9Z^Dc0fF7ONXMGZf}qWkGD=91 zx`}Z*e3LW1nK}lR`|_GD*-11fikKV95>J{#>d6EKF`+;*547K!L7p-F>pyrZC7BX+ zsP2;t5&%mH#-Yn852=Nuweq!ua}PL82}R)+wGY!Nf5gN4;R9JskNTK@d?tjl{3n>r zahC^fIw%dEBkan@cwxa!q^_Erg3%kc`pwEHM+Ho>21g8zCzcvhv0LjTsSq+ZJ>=un z8nArurj8`4Edoeka039hBMNo8_hpgA*BEPJlRW9MjCrdhqAWZ+USgVAL5?`KI}%Cg z08*R7lQcmS#D;0Ga2QKtY5PaTKT*UM7$hQ+mf<+Y&V1T`f`dO3xSUg4gGY<%9^?F1 zB$&f%n|*>e*Xs0V7%AJzLL1*W9}a;300kx3MjE(YCSKnkT)FO`_Ym1YYi!T+damqEsMNu%4)p9g4 zwXSG_Ry4#~d5!i>$-|M++U^xJRS|kaboX$+B74RjdQTuh=vGEs)kwywb_|S+g!L~J zT&YfBEdca6ApF-?F3*P~wRv4tOJ6mV(o!>rZgB>~B+Qw>OyG8ugX|v?HdgmadFVLG zy4}X`RlZ4~9qoO-MoRM2%24lR?LBe2&PBzs)iKfCZII0#Fy}G5g>_`TaEGMCX4dL_ zcXCqHmj=Cu>W2fC{0igN?}p`x7NK&d;TxoO=E;Ay_wJ5bcT=`T#KJvUcFWl9x@@5O zB`lU|Yo)sDTWqdop6-^a4KYiDjZsK8a0i~Ot4Qc=min5@hXU13%7Z9%A$4c583$Z$ z1KaY-iA!&xx^Vq;JW)e&pt#Y(T|-q&<0NvppJD7d0B@YP3tBDMaV>mCs@G4%d@}la zxV1K_x}Ffn)@1sx0~#Dhh*RzKmGn30ox%K3Eo8MZIJ%my0Tj4z_iv6&wa3FB1rqu= zvV(%?g&icd{4&YgJ1Psi@Y#B`&Kt&d_D&M3Xt*MHC~D0m&e?u^&GecnDog2H(my!a zhF;zA&p1C61CE_itXA$MeTKAfjru!<7<;01 z9Mgu*=5#HE{RsO!lnJgpMk}PHyVKF$XrhSI;-YAUd<)d8RHe;ah!2du#7$ z+E@rLkiJ;tBMxg^KCgl1swUs=TrpEc4JFHlprE?3j4rCTxx_!j$s4|~^6H(oQe9~8 z6qE3u6iIP2GXa8}vGI?v=--y(?pcc4%be&P4LgHGebZ)~A7?&2ed!V0E>b!|CYm`k z4KAh{hCK1O2ukW4U0}EROcQa9%89kDYk#xaVq@Iq_&m1^YrXkdo1C<7dtH7f3NaAQ{8u;e80%VN?bNHaJKa>7-@6J z$2^`(DKnWJrkRgvcDEPoapbM`x6y&0K{FEVTJ*H!t4B+qQMH1&PVjXFpT2V>)u#}m8h;b((MPTFA&qiGM8GbglI{ACZ8oW*zyZ}!;S zPb|PUPeldb5M5JFanrb2;>nBJ8(UU%{{ZG?J}N62f$6n>7iAh>v|SoWXr&|#{ah$h zRa6R5QXr+CK;!CyT3hi|6<2!dXEBYed2^4~01PMTE_VYX^H2+Ou?z*S^F$l#3&0NH z>Sqrg^&Pt96qQEb4W2ffW#39b{t0I1UPc`?Kl?xOnfz~F*~~FEmG2G7CLy19G#Au! zRXvTH(3>FkLwoY)413hp*G}96Xm&hAfoAQOJBuT1%`h;yolG(Bm0<25xD&J#$NeQ* zyIGr?c{gh0ehQDj{;M9I76cF4e>P@3{@OBi;9+&yZOr z-fM@~Y5xF9TW__)Y{q@03)t0d-<0ViNn^;J%voNRBnBBQ91nnNUEq~jj2 z$|0P4(UFGd&k4z=bgd0@S};k-O%hJX6gCQF0K_so$FI3U^@hY1rQw@ctBw~qp6C+! zJp2OAc}y|xtz(aM)LHOdsKXcI8ioTbqv(O?-ye#+o?l0Bd3Ki#!3P88D?uK+}~0|yF-U8D3vnj4F_Y?bzMt^$CBe+iJx{7mgo15)zK-V zF2Zf3002@TXy#V+q`qn3cE=>+9moX9IOokeFxKJO%en=Ip+xoB89#Duvp72fIIuZn zXDY@CE%h{y64K2S%TK%&oUWV4StVHkjuR&$^O5dWMsdqIRPfb+&|2SA3?M5D zoIr2hy7^<>mfObsF4<5)*r%IN>~2D$ZD%~Ue|l4{4QSb`7mc}=mqMw->n9mfe0EeV zuW13=PU+^>vmB4jX7Mk4T=(RSw@YPFXvRoXzMR~8x{}#e267KBs$MF+%b{BQXM7{{ zw-L;RN-G*SF4(0uwU|D!S^Iij?ROnv&$OI1*!GE3jjLues04nN!g+;0(>v1aSMtW` z(_RC4?z8QvZgLzFomAY9Tzm4T+Le3PqP^P=X`7A9MKk&2j1PHK{+Pl~tPT`9%BJdk z#jbN8VSbkGNduCzhsocl#CE@P^qajkv4r<@k#0I_cR#sc-wRwhZlv9A9-DHWu6ddt zBg>r9cOFVVe*ymWzcuyUHx%OHn9qmLbLaOnxjGkQ5c)N~;Zk?=|E2Brsq%ag2B_4xVqDlP}%a=DOp<6U~vux`1vy>3KNZV5R`n@wn%| znpt+|-n*{0%~OH+%6UAR)e@Q0el?%Ff%4?E64T4xq|&=HR3F#=NtRJLy{BGS_YbmL=G;yMfKeSoa?l#g{C` z>C2B5%59&hovn1TIlwWjJLPlRs1iUO%!d>3U2C|f0R04R4)(P-AtTT&u3>BK)Px`W zuogc2TftmY1I2U>#B7pQaVu{UN1982tXxJ$vNj7&+f|{X)V5NO+Kd%xX{h{J1L2}% zH1WLU?2VzJ-t_zP@Xg?^<=ORZhLZVgWql>O<4IYOP+Y)!f_Wc@UwWU~-Kxi1XSh*P z+UV(n;d>#TIGO{A9;q8_a=1HgwUNE9rkq2K_HK7lrQ&Ps?9o$nmR13P(gr>YU&QU1 z(^%wivGG-y zv6uIdlbn^Pwtgv|mP1<=J8A$pWIVI+DUm}&>w_Sgo*3XUhdNI1f(hw`$#e6qG21n^ zeiBEBZ<0m|4HXqduPoC54%Bw{51Pl~ut_b?y*FJ@Uh>z@=^DaAcg}p(T<>=&pQ4 zSo=GD3rINVG0$~V+ckyUE!tJi)g{_Hgj&SWOdXNy`ow%WD}BYDcB`wyXldKzwEQXLrtv!i=WBM3PKpn^KCQQ5srH5OXiT|sM@$u((r zR=fT}=}}ZRIwmv_KxhMXU9KGa_~uR<+D{PMto^)-+mj}eiW(-- z9amyumR_TeK5Q_$$+&`UnNyi4Ea%Z-sraIly{s_I97c8u^3ivSy|VXly_zrxcbZ;2 zr?{B&;g!^@*Uj`!-=a*DfeS`{s;{1-5WgM?xhMJ{2Y^^|ys7wO&##ZI%m*cw@wbLt zR|xm)yI0{_`gwapY?Hn~W1?d~Z@iVowbGtqZ-EORD@ggSx%fWYrIqz}AzA3AIa?eK z&>Rbh-|k$&k=y&Q=QgC1`oJv>g7+o1f=W6UO&m^;Ns@hF?+{gp^v7v^j{aS6?L9@3 z;~Oq?SolG(HaB$m}Ro%O^PAaF3 zt%B)0a5l@cgsZ0?aXrFnd1|gwNa}YNM1ILU)~+*Twbp4kQ@Wv!!FRYaJ9Bc%c^Wd{ z50S@AtWOwz4p#`OTUo)C@m;9PgS{G4-An-H0pxSRu69dCPKb-G)8*!pNBexk<>w=0)+yLL5KNUf}Sb17S? z!#is~@Vc8#Q|;`is*-A{n-6&_+;uw~&Pu@6J{p za(pzrD`umz+3i(U_}s_W()ZH4R&sd}{M9dM*V$>Urq@jOnd*nHq+|wl%FtQfA*6s$ zmN&TqEQ5kMoGeY&;`*A|r2)7u^-qvfddE?z#rJxfm7XI4VWG`7*N^(qx z1I;@}UHle!M%?Zt0dc#hh~NvXl22RlyNf-<+; zxJ#!R?Q4YimxuVi;X@tnrYeTGweMu47q{yC!s*|GT-A=TTBNP0bk$H)cXB!W&Kw99GyQy}k^VHz%t0ue^ z>`o|cEHN^sb2dwZ295BmEUuuczDE=$qr8XL8nKUdekked*~Q!&MRlSeJ-|Kjj$zot znxk;V1#LGGck{qm7K7^z{5Dzb+}j&?E4EG_yhqws5y4AF=SJI<#k+yAaVr$g7UEe+ z5!+tx^*8#h=WEy3UAs);D}^kQLk!jO7B#Os0C2L+t*jo+Y~$S_Q%+0l^^3EvHJYfs zmKmN4b~>B=R=UsG2OU)0DMe2W)wYrHv`*%|q=skKJp2|?E6v^^^jB);e$4J4%}8ex zaT`V(#5YF!s^9%nUNy9P#*)uuqv3&Lw^Y_r(KesCjw4RzxuZ@F<#_t5uiI9-Rgz6Ts_!E)14OIKHOqxyOn z*-ls=CoMTXBls^N{#Wr={{ZaQkN*IDllr6luHuvb0QoK7{{Y`#{Z+DA?QS;Rw4ePE z?LcEZC0DlJoM-h-)OI)8RXa6J!j(-Qw%_Wr{{WSoP)7VmcRu>Z{{U2%@i!7YVf;gM zaPzBq{%NWCYjo{@1aVaXBC^diIOmrS>Z=`SxUi7wSG10q3CXVaSx3`c>g4^YoPU~R z)>WQ$QlEQ@nuf$})uWsT<`R7hs2|L@d{X@?r~E{pyoaRJAMfAuO+|!y9Y^S$^G0bW z{TWLYGvo8l`JgIi{WHF5>IS_QZ0A8#x0_AIdch3aXME5VGY{`w^H;*MBGGi#*BWDt@9PDsWKsrzT0zGo=cQCQ;o%Fho)436=sQo!jLuO2JcobdKo&`VYbswj4FH5NXIZZ=m3{oCI zqV6U81xWO|kJG#6h}BMizjw_|LumewH|WpJ5vQ9!n?E%9sh{-D`J|ee`pW+Rnra&} z=+pj*-!&5NZ>*@Gk|2Fhocqf=2&N%-7sV$Sp?s7ZgRPhH(v9p40)X~$?h>V8=FN~Ks z`^wQi*KIO1%SQ~3ZorYratkK8uXGWa9fMHJjP%Ovbuwa7eAgEyQdmXZ1)|$>J=~Bs z7nbO3kc}n34!;%C=y?8vgQzKNwbj&=a;VlvkXXTk@rUz7G!JuW9^pOV?WzGn|wIFI^LJuTm% zhxZduWm}{~SWm6Y?3)h3<`%frgU+54{*aB6K<{pno!{vS_NyfA+zvwZci?eyrLyo> z5LAKGx6Kk`~Zs~Y5MR??c5q1MV*_c9lE zEY-K_g@^5e8AlDr^iBb@jb|K&7m3i}CsmC7-$(Tfx6tR5`HpqM7E6Wt^98!w5OnT3 zi=6E(jiYks9)Wk-do{V-JlZOEu;H$2-q!5*W1qVvavK%$hlltIs8H0hOTESBIa!Q8 z77kcHKe=mrF5-A!Vr2x+d#Y*mIpY~FZ`#WCexu9OYVvXTnB9L@&wVe4rj~K?`O7K4 z+JEk|TmyvtRaBeb7d4DR(H1Q(L4&(PaEf~=?xkpOcW8kYD+CB0+}#}t#ogWAi@X26 z{NL|fyccJjbG66VS!2tSJbP!YHP@Vco^VnDtkwM_p<~96+T3mR=b?W`NQWD%Hv7l? zN(&(ZS|gZ++@c(8=Lox}fBtA(@?G=1!a!kFM zjoY*}5BCr2%|iL-zM32lu{06X?GplVQf-xPn_!mIJCEKM%J?u4M0n{1XhUwSmb11GqB<6{K_(G9DwwRlz&88UrpZGx9=cf6Tu|pQ2G#qO&N}Ufl7o*n$iH z)l*Xij!amjn~5lEYpOE(sEWq^@h}h9bzbxR^y`Hmg?Gu=u@sh5Aq-dwjLL#D+&U(e zW$eo}Dm}TVt>ylzqKTrE#z|-_ zGiVL1Ex&3mJUM>+vfmgn>)L*RAnw3U1z>Ip91_uJSb61s0jcsck@84s(UCl0mH^m& zlgNtm3yb}E`IpXqSIddER>>?1ovBgR4oExwUGX;AokwlICHST@>Mu1o&}Y{a=;^T4 zw(9k808KKm>f#bwdr9O=xX%k6e#j(2M_hY%>txgZ%Z1(naIg0;Kj0j%5!=Y(I)E3j zwB)mV-UBdoDuptWG1vj5;KU(R`*oI?nk#6c9t80bbva#2V6$Ca*)7&<$D~($sBI3( z_kTEq39SZp85kLCQu$7!G0mmFjVAT&KW6raW?)apt!F4J2bRKC*LrsjPs z-%PFhG?aSMp3}3cO-zZm5+KspnNjrJMd!&hb55FcT0^^zA=cT1lF_^wW#8Ql&0hZB6} z`2g#eCud;-)A9BZKxTj>*nYJ$6KcC3;~EL1dgBLF;g#xA8ia+p(kjC_Ej?6FzH2mO zOs3JC6g}2M^We&c)9?-u3w?AU0_t8f-OdPfe#OzD{X!O>CB{#Y9y7W;UfR7@?j6td z)Z$Ej8N=BxM&i#pL-sRDgAS8Ac=ySF>7y8h;MZolmk^*)TcFM3cUUPhVIq&Jq0abG zb`uFU$(@vlHQ>TOAyQ`!f}7{=xjgJs=y;F5K18{-$>X}#iXtbZRy7SQgKUw*#oA|> z+${D%Hxqr91Wo8`Sbe8Jmzw+FM(3eYhUP1FWVUm{Px=96PGUSQ#u|7wIUC~HI&j~E zi$_D)W^Q9b9 zlX&*7kH0yA?5VSmfV~| z;(kssygYqQEEH^uBFeU|3KY7!FI7&o@e5%tOdK1iS3A>PIQEuh-3)+Vvlhg>I-694 zvWgD8Exj#0XvM3K&xteqCOaaj#Y4cbc94{^wyxeFbFO=Be4X2vX(Vc6>Io&uxWxT0 z-cTBQVxfHmYx))>KzCWesl)z1c`Jg@8FDJ1d+@UNII5^m$bQ|W0f*oRkVI{=!exwp z59q$a*p6&J*yraOC8DledTS}vKz{>8nonT>?0mtTI|}d+O^S(N9@u$AN7xNHxKK@V zSGN&%<&MHOU(W?=%4$8ix~|zTe!bBZc!C*3DAPb~OvqywMCO>} z)zf))!iH8SKI#ZC@u#G?)*-_|Q_7hw9+=v#2bXlZ`4YEF6l_9LY%{mstAAH>{EK9X zJ_`51X~toq#h9YhYc|6GnRDoxM1f~e&M#!^6vd9cV$)u2wbSWRkwA@f2soiG;S~&a zFZDY>!}@{0zdHrXLxy?ox-X0w!9F4M2Mna#KNLoDukOZh%}y1VL9@TRt|GR~jhrbG zayp6}0=l-b16jH*FfnnQ2ROe+T3$4v&CyW?g7d3qQ`oUdsL#`{RsNQKc5+q5&K;oq ztMwm(OoSRXL7jrb9A+r9l#q*Aw>F|*`Nt3jX+MBY z%4{q%T%2#GLci6S5iUmPAEM}BDI#q^$+P;5eQ>3tbw%OS^pul`&5jLR5@#HuT_8uH zS2kH3bx8c7BsVbMoeOGOWho^iS7)}b9HBIDPrs%gmgT$b(*7~{6q0A;aP=9@D7ERy4np5qdA@p&FVDT-v!F<}(C3BS5>g%c>m~hXwMYUR! z&FXx9Xc37vbV6*xQsy|&4oNIH({fDq5sU`INs#JOHE@tebTg;c@ z=9?_uYX$zVZZ40dtf}VswXWc!F#B?y@ zYuhVP**>l%1wnC)9yt8HD=F>7rk|`T`Df-qmz`9r)t1Y0fy_itUaCxXh-_>lY4~;N z-Vd}*(?fg(adMCEm&VL*nh_^h^(i7h3_nOPjHzxzJ~$suxH9klq_P1m^{;Bhel}1>`ZR(NIF;RyQJ;AsLv>n@dM%AMuggQhtP3eT+SGR^~WEtuYLN> z{BfQkU+m`uN@$=gw&h-@!K&=B5ZRv6ytHtOug(QOs^f~ZAhRdlEKd>oLhj?COuL+K z|K8)|+YBTC1$q%laEpuDb*}!LBD%JOyZwZT#e9G`zEmcN1(L z>V0fl*IGivfG9rc5K?iUeOzgLwXbI)OPJ>INbROj*G&DdDI~jY08d&T1Eo5M0b0sI zNt7Ka9O5f%GR@sGK74e)VssNf6{tSrg<#7wdM{jY;P?gcM<9AJAAeWAJf|1xfO1<7 z>(Ko}#Q?UR9cKc>pZiNtzLi1h?eO_h`0dqj$)-44$S>9mv8rF&75+c>niTp}9Ka$J z2v`G=E6(x&X@{Kwlih}uqXPF9`p)?HxgRlF-k9Y}2ASKC*QMLH8Oec;;4Nk3&Eu3T zdHXhpV-{8{i+Wz36+;bj@EYSQGsRY{H|Q0u&+n5sU-q^FxUvZ&K&&x=D?kEoa9{=( z=ia40UFbEp1;;_aFTkv%`68O4@?@cAT1!i#kw%r=Z77wdYkhAdRG^^tOB-Kpb8GGd z0Y8C#v;2mpaP-9b^BRfTA*LqOLWc1KaeyYu6~Qh|{zcDRwHV(s)HgA%=~hO%5=F z-r~4+kL?OMhhEeGl5sGwh<+~}P^`kD+4Pc2PQO8uXx zTVSa+KZRn2gB;|Le`^dc!wBMaX&j(I(^(%Yg~l9FVxB-zeU4kn$?sH|#A6HOCFm?KT6$jytn6IKU&XM$6s69xsghb<%7{x`Cy!)M=_jVy7;OtJl1s z4T}ukJX{I;fksmWZQMGe+-7MSes({Jsz+^KyCZrl`T+(rwOx9;|H}5w74HDdOR)%3l@wXf@OWTCKW0_c5I#)6eN<81UaX=3uaDDnHRW?u z)2Vz7D^Q}=12vYDCNe=s`Xc8pqw891?&it0%^K|yw%PciQ=u3FnAO4IV=jnw9F zh8=3PNPC~^s6n&p+O`%p;tkn5v!U{Be+c>QwxNaco!x`0+Xgd}w3_JD!PL~j=}fBfIg z;1YspEMA=2?>26XDp=$YV8q|qYS#dJJ^TEl}W(+4e!i0+1}7&eEgDm~Eoy@TFc zqo)~XXk^)63l+bbpkh~pqIH%$bCYE^)^XG%jx6mk1~<)_x~?RW$BwDymZ<@zI4G!2 z$lh>gGv9F3(T>?N)4D^>;uu8o?-yC==79>9Y@wXe+N)!htL^y0liNPQ75}&hXnZN8 zbinB`(7K4>K52VD6P@OXL~mXOyU4smvjE~52h8!d5)k&((BowVQXOCGu=^nV*hZ85 zaJ06;zdT@)U2>}!#lQMOFo}ddFU=*01`DSQoRUo6#jA5!9))JGif2!x1xvcnHACI@ zQ|5NlDqwmgDmA<&AV`j4@akV?9wi`3?@pKABv3uSaT48eab1 z6f&ngBl~>UO6T#1?7(8os0%LpT_o~`X%j_HZz5Z5r&g>oW|V)F)7-w|aIvLm6~~nR7iVW@A3`UuvS2%7x{t#L zn3xTDWX_`6YP4v}>F)xkjMd`Fxf`Dw+cMNNUDd+{W|W4=DD*Y|0T7JG^adEJv2j9Q zW)blL+iR{CZe&-V;wh8(_**(JLB<1GT9v{Kz*M&+%?ObvAKKDjUAgV+HZ`K8)fox~ zsG4%b#(MEXBd!hYv%#=oj;`IdZ!5_zPL!4+@D_CLGr?wV$^8fH?DrO#DS#TRfFn{~ z22&os`Ema`j6$-4b1I8VA=D4YmL#v12&;BYZ*!dI%u6wwe1wh%Z$UtVh-?h*OX z_)&cCi;NKlndAlA#HIeF%m0cB;imJ)Eqm)qsfw=!e%voqZk)rjmlC*81>TxgXR`TM zCxTNdybb5AFDw+9vK5MEUU`Gt_xQ1hjD7Tw!1cBc@ZIGLXP{|fg1@e3QJ&Ksl1NlO z+TiqhYX305Ix$>MG;;+1<{_EEU_fQR=4y~^H7@LG>5Go|9&#^Ps`C3zn&E)*?`W$N zS2Ym$*NW!um{`I1^GA zQ=R^-FkjH-m7CYc@~lFXt-G}elf_Z~a!+@F%?izPqgQN3n!4^WIlLbMs2J=15yq^w zY7F0T3^`&v=A#kxD|jnrKMwr!`660qV23wA0Bp>TLgHE1gKNp(?SrC5MX;O__hXEriSF*J7W$qv`R#>po7{SUDjxOxHfCa zE*Ye&$YH)h9FOV9yezr3s8M_Hg2#Z;hdra{_i-^B-+XAJFQ^z}dCf1Ze+SCDlbF^p zrio(7YZ-S=23s;Qjl!;LTXr>PV&XB~2AUXh=lT|u1Eg8O-&}l9 zD$U429VkxC=8b+Ba3o;qFV}5}((5=3`d+((=XyQfnWm^g4bre*AqA z+l|iCa|1nkleV07YE4-C0oKrdxzYSY&&in)$q`dSbx5-_MWC`^PesR?6W^@P(?DaP zycN$QUp+2w%V8$rnX)lfD24@J?k%&uOQ=PsFed%XwBA>9GsyP7T-Pz83fJ8J{O7xP z`&?M-k#A!ygu${`3p1ugN4D3CYyefZce3VuFSV1@@N6N}k>^FsTvgQ!n+MKdR)p~W zM&nhxfzDGWSaxYibix%?_czwR@Izax$hzMp83pTAUSB9`S1$Pc&3bR}v<*0CYX|V_ z+l(c#AL#~DCc9K}TlAa>l{~Mzo_$B|s# zrOE;2F%YoaBQJ^);>AC3752(#Z}a}i_P!by2!1F=?*c+|ZP8$tsH5+{6C&*WW;SPb5GrxSC5KmaM5NJNO%ND^6t!P z_nd;4sR5isM%#AMQYCJe5Ch0}Sfa%vowl0K@NQt7;1Qq{ODq58k>*I#I|dqA&yR0? z4f*2Ws?0u|M7#Z}$@TS0Bsx)5Oe%iPuiBS!378vKE4RH+zgN^*#{U;MMLfGTCPJf(A*q#IS>k*(oBgx}4-gLlo@Jwh|Co7>n#AYmS|-7kj9gnU>q5JS2}2LqpF zuZQNdT>aeIgHCez*BG%OxBTB}u$~u}IWPEYNJ+|7*}==2(eT+DqI~O~=*3#_xg!0Q zt>CTYA+B}Kd2ATEfZ^b2PGi>~n%czSv#(2$n=ETjNwL{E0Q+Y|A^GT($TwkQw-eQ{ zdEzQ3N33R=aw6DBIoCY#i{0b%q!Ddsa@$RESSJ8YUF+bRPM}k_X?+mp#k3Lgj(r9D z`~%NcdQ=R`i1+z#%hEz3Zy=OWmYEQxhwQg{DH^m7HuSc!+QK`bsSyz~X2Zf#l&Pys zSLwF8eSklkF{FSubeCTIDv);?1EoyUxEmO9J{=-%|Mxc2VdGWn-J;>^p}s;zCnj73 zH?UuRP-I;88l&=K{GEtXH2Yd}B|UQ-xcJEQ`|K7jQn|s0T=i_dUmPI*KWmA8LQ{vo ztsiyQ{&lSUZ?u3W|2dqU4_WmaYv_SoCC{p_la*stZ9~HC8+ZO6)Y zcbbR(veTyfPY%SdCs{TxQxSBpzz#c!hwidd!TZm|WI02puajxU^|bm3cN1e&0gNjW z{FyK2+IK6@IXM0N|C^UUV3O4IZamB_tgI|iUKktuY6Pln%G|N2m3M)VFZb^N| zoqA)S&3n@)dH)$Q2rl-e!BRb3;bK~3Kg^Huh~iqsmdTa{?RiN;;T%SBlRDrXXbTE_ zqon`TtkWR?+5O?x0t9JV&9|TY&6nU4SrT7*EI;27o@o5v26@Y$9pGyenEKzwU#7gK z1De?G;U7-ptfBuUoqMf$E~#^XuN*#&bVvLRL?5me|J^?DQ2H{_^ayQ8z#;EE{}mW} zZh3s~KkxScc2eyqO*BflGmO<6pJe5GjnR!Tu6T{X5fJ}pyd2^kmcQ`1-`6>y`EefM zjD1Wpg>R);f>isp@$FEKga7}GZ})bm7UMOBedW{`XlJ;njxlxlYj}_SSdk&GY34o@ zn)<|$(Yp!0qR=(@UHkTqpG+Oo<^AnWlx?{)sq?torf?N&>C*(7>WFvkHGFFSqp=uN zeX}(&(S!AoEQg-$@Db{#vw?QsZt-3}VRpZ06i@OI{;IWaLRFz6G+V!8bK9ncGe2R# zF^Liyjy^lNN6v8OdT&r|%2ue(#?gNW|8xt7{VJKiRYGBxvR_QLX^YJv>MCb#d#YMr zC^XuDESWP-C&D|16zqHkdp?NeH%o{vV(&JFy5@zB+@x%qwQNYtQ4H&X_ONf znljJ((|U?7PladmD#QWR9I^U-Mfjy%(aVlrmCb(pG3wW?E(5 z6^7=&SAO`|rYBiF7T1?h)5e`k*$oBpwhp^VcZw8GJJ&)7*~Y@s9LTCoC0$S!wryL4 z2+L+;=aXH)6RWW^b(qdv7^ZbndT>+Ng0E5S>9s_yNtV!7VXnro^tHlG3moB~)xq7A zhK8$UkXKdMECBeHi?#R?b=q>z;sS@_BeeVTrta_z%sDTW<)14T4>WE-h5nUFuGYf? zB9iB8n;(6psI=G}b0>{#B{c6f#81CW{Qy3)tiQa*#vbub3Cf25(Teho%L>+Xa)UVc z2ilhDLdpKbd%$Z!mcrgB!nVJx1SCWsElSL4b5zm!h6Vcw9IvwX*0=z07fbE&b>;=w z3D$-f8G9f6=G#Ks#EP357EBoIod_PivxA-*J`NIdm@aXWiG&En>MA1HG96fZXj8@- z>8A%z`{4Q$`!%>^wEl;ncQ-XbzPw2}=7e=C`qKS)Ui^BN{M{-r?etbc`Hz?*@A<|; zOx4lls|tjz%zq4f3l>Kt$#IT~aL;O67QlWLcT+drFr~zGxXPQ989~TjEl*UVt#ppK zA?T4+cNb@48K9=n>*z*uby^FG-`o77;$3?%Zo($waV>D)cCpz#ZuH#rIFshtw&Y=r zCI7Iz1gKpxu3J|(q%JVoe`lEvX1vBI*1T4%dYgRu$)mO zI6d2SMr!L?@qFN*@+qmzX_?iK*B6RpM*oTBIO zY;PH<7LFm*=kx^_yqc5ay0&9-rzAf11)S!|MR{8K6(0o2o`A3 zbAhj)ahyfq+3Piw#Qq_D$5?p-^Ir$C9B2ph`!iypGTU)=}#zHTt- z>QUMo)aH^VbY)GOEb^ z;HA9&9x1f&D_g1yBRytu8+a@2Z+4Lnb%wCo+NYYT@)j-eOh{a?lS-6W%TqoU$42&& zb@@!@)&(G|d@jFXwUoJnKX!&2Qf%iph|x(GkSzfrAkwiVcYrDRz9-!n(}!H%Rl_KR>ULlh$nqmUS(O%9am}|pJO=^hh~KLmnVR-`5w~l_4fOaaNxbKFC|?!N}hJ(A&o+Ga=AG)y;K ze&O!q-r)cPn#6dckYxjIUvJNEo=m zACoE_7=qtX*WO`#^yJ29@h%f;7OtM|sG6eVmXCJD>XF)2+6O`$+36atV_vo9bSYiv z;)aj~l~QP8%}(rIb2hc+1dg@`0Is&~twuQEiw}7o9Z-QJ*y0Y!;&B&PbfVow|7?MD zh-ZFQG}B^GOIMw>W$?;>Rcr1g%>(THJ=EaUgLuIXjV?9x?X^jk5C?5WKHD^})3aPy zCB0J3WM^~_OMb}Mbg0P?8vhH6yh!kgSWXQ?WTs5~gvmja3NXgrrfCiIZ*yvKeU)c| zbMQFKpAo(=BaL!z?K8DFW!QcN*8FTlFdqyJR{1H4KbO{PAt2t3e@>GMi#|z+_-a3_*}Em z1Y;j&!l|G!t-N#j38sfC5GIJw97SkNG5(`_k+wzi(5ZIvqJ=?;l*MMHCbCJDnytco zEeZicNu#L3nUOBoC-%cmup;81U(kKRjrMH~u6bET%0+D1!Ii@Tg}l)wYOc#5SfwMh zeiu%b@M^?iFBEw~LT3B>&)^!(!bp6}ELFRyM7(1dJ%zOM^iJdR7S~~c8HeA~SFB_S zLEmjx%V9=i+rcikK*iUwrcln$txBOaKhB}p#~rmL*)25y^MB1}dk_!%agb*WR51-W zZ#`I&=RlXp?$p3Ve(P#E$A#<0DsZj;s*#x@4hC60Gej4y?nC6U@c1^}HSNEAw{%nd zH>8mvv;xbpiRGUd>MpL&i%UrMRD1GNOZwPH840ZVHU~p^`01 zD}{`Rj4O#Nh5R9UF9rwPg+`+L%okl4@0QmaX9Sr*Ij`G4JcF{4t?;q6QZ zV!foVKfCCzWzz84`5%C-Kan%ztmP>=xb{ZZtt0Qcf4X}d@RiWqb=5|Y&H(;r+&U=7O5G;WJkS(NL~5L1IevIGwy93b%-4X{9%7wLz5yh%I(9VM~hySnbg%Q4|>&-Q9-IUn=u{%B-l0+@$XYrzbZk zwV0UvlcDduL3^rruxn{()wKP=h4fE^St?hkK5EPD7&2{pfb=#Jg)L3GB%4+*ydQ1) zI|vE1nm?=BbQ(8SoH34l)=S^jw}h7YX%|2$Mj=Q#ZNy4>k%0A3GB>w*Lq-xe+BA3JnI~s_pswj>g%c z6mFRvVf4amsIp18LZJ8?+*hrs=Froif9Qa#<~Hz|9kom}z_8ZSgIoe3Ow!y$s%)JIN@2$8jv7NkOSzVAb93lIU`k&>XBg?i|t8;0CU=ryS zlVw)PS(nW6R6ViP+0x2I3N>{**QvnwTlrgU%bu<*+VEKMVT?*RdtUX>>qYY6%crd< zstqc53XZXG*w8W*mM#WkgzA9u0sFx1pp^pE@ziMzt+A*W;|kDlsamxcL^vNyXxvEC z!VK~MPJjI`8OFu)BlJjx)EUhJ;a>uIUC==*nzFp>HmX`aO_zNb1y@2(3FIr9TWmbJ z(e!>&8)NK%7DfqTWz^1kJSh&vh%>*)vPOd_RQ+@w*H%B3^ANEBz3>==Y)%9IV$)SqNepa#JteJm<(%#s?5q57I?I>u)+WnQhBTR+U$1 zlF`T@i~Y|Wm=Cohdt+gAS?{M{8~ST!xd+6KZcUN}pS8vS52bN&eK9uJWgyY8&+Pv4 zK=B+P<&CpCxmUvAPDcIUyT5jUI4$C2#T>IAhp&zc55zw##Wzfa@zn?O2<}&pm*lt^ zzosrp=neAZq#k%N10*kfciP!p)$-JWjxGhVKW$47j^nWTOP-i@6p0 zkY(=_) z$-2BD7i{%$LIaJ3IeM`_J_}H(*c0{bYERcBi*bdjBg%+kUMgl;8+W>WYusge{#r;0 zlF|D5&1ix>@!2JHNN0N!;co3(4I}=?-D)69+|4+4kKVS}VIP)Cs=ITWn&*VfvrAQr z?)RwOXSK)N7ptAQ4SKy1ZK=B~@<677rj#9I(e|F@?tpUj7M-jp3(D`vIWFE55+2rT zkA8d-94OTj(JwVyLgyiYv9pXl%!%hDB1*VPBSoXgPy$^!~*a z_tDteR=gz>BYx|-OTBT!3jXN$6^6rfJCmsnLACw?u9jttP%cjcx#go=w@jpwtkJ@; z_jcpSPenrx163@Fhe*qSu?W)cAU2-v(=N0=yELgXP~Z@+p)`a?{SQr{fpFd$Nx>~l zUOY!mbUyJ#bO3s4nMd)?#u?6(_Sxl3TICHVIME?6^+LOI>+T;(#S9BtbsP|8X{4e2e04?!~DgP)kgLk zPJ0(@>sgHl^zrl8G6sHWQbq^VW=Mr#$!$89ORJL_z&_|II!vk1xp%~rjE<8>?PiH9 zgYcy@f^oV`g-dP2EvNhWQurFmS4c>fPzBBkw4|rT4>{;h(5&7`ie71oDt})1w0d?x6t^3X z4pmCEiYrJJ5iCZEWmX6KpT#|NAt1;(SIz^|e$oY%{F(Z2*h=I-bH)ARr2i#REP!|M zU|0<5`GHp3!CaMhd4UhgT^n6VV`qop?Ce?blt@XUoEGJHrmqYGAx9TYy9#_ZT`6>9 zrinq07P{ZlGi0T*uMDkBjJpqdl4PIY*Hx`HcNs*;D9M_7ZgI)L<6nFzZFSF_+%V+C zP*^S<%)qD*!T-iS5`OI3d@9!VrxeqF6e3o8;p4*|L+mmt_)o;4Mq>=)XVBibGtluj zizRAuo7s1teNP#Xp>5trp=qs_>5wQ;=f8V1LuYF|qX);Fx)l}`PCNG`D$AYBovIg> z;dvfLP2qIaf6_OQ#AY(MUe`4j-k+OYd|wLv_KHiDu=xPuXN9%pB0OBzV_tX}UyG3k z0!5_1HWp|$W7fcBg2dQ6sy6d)Gke5xP;~Q}MaG!;#gQppNPy@J_I@@UuGDAv|3pNP z`s2Re40up}MsPi34uJ1|^G0GG)#7#Q`spJv$()ci9)HZnTFh?aTKqZ|zgSYn_$?K^ENCLd?#CM|>+%ZD9y$H9 z2KW(P-lIj*i&vk(C(}@ZtP1sqH^3WfOf8JLSv;-*KT0 z6y3{$(BvqG@@84yZ!I!w?6H66oXL;GDQ`K0j*r>XMF5^Y0uV+V|(BUga;dX#ZlqFhuL%?+^rH75Ah0qZi#t*A5;2~%Z? zP@tw{9cA4_IqfdB=PpiRou>ZY{WS5RxNuyq?l$d}U?gV1-Q;ymmvs8`%(;O%dEnq9 zJ^IR>{E&-ap{e@c-Ow!Hl|gL(Yivv#B_bA2D_qT#j*klI%3%D^8wX0U5;!QWk_oqw z&wikS&U-mL(-3=ZAoC5|h@hHM5Z>#j9*0?!71NUJ2l&7{@i9@(8d#XF_joR-kR z@ss{ILqz{GGt9=i!0mhn?uMpvZ2+%i^Zz&8WW?qT*nlEMy3Tt*B6OsNQoG}n0%L_inP{%$ntwU~1vqV>h4T^}zb=mI9->$I@} z#>!`Vqp0%GauQ`h#diae{gayD&w(=N?h~2CbKF$XZcxar`jDNB&+cU6YFnp&>x8b$t584pPuXQB&u0E=W4?O}2c22UOi|7=a;tF2a|7W*lmPw^Yh7?%H z*qMczp>pev#|^V|AsT$9?%Lc?!ft4YmcZB z1p*UyCgxZNpTaa}FD<#Z23#h3AY@H6cc$I50DNO<*8rPew>s;`^&Kjxv$lga4*^^5 zjV|<6GpZylXlj!@qZu)&M*Vr=CxnRQnS~$&001dPVQ^Pd>qQg8>PFWXEJ4I${<%LK zz;KSAN<4mRsy1>AXmNAfv0T{M7a_S)6SWd*-Ko536PzBTM9x^CIb{u^X`j>(0CyH(yTtUugM$WDN8(n3WOPe^WhZdJI4M&kDrn&!45v*G< zicOM_{6{C?Hz3lHuV#QelY?}cHV0tR?YFBsvS!d~MQKP1IL@IDE@x)CIht{P@ zg5Rgpvy}16kS-nfkO&0sC~4WJ^QWUu=xMA;t~^h@68x5(9-|nRTVCOFxBMa4ao6Be zX#}eDGH(2C)l^s&VneR{-9wwy-Xv$XlM5i6yL~p!?ynI*%rblovlhVZq@4+Y6w&Ce zJiwNu0>tC>24RcJib&MWTgH?kTYs{izliSSeRv#iiz(SYKJzDktMthfq_EhbDD(`_ z{nn!NPnF-nxFI%zuEb4_%y&0B38Td;$RuXFl6&jvjI4P|+Qn`chBsf&6_wckiU&Z@ zYd8%gfGy&UDRK`oP?k%K*FR$U1~3 z%)!BHjCU(-Y50cgnBL~4hp&t=r=^``jxB}`05EETuv=?bHx3z#Gs zTl<6_uWNf+nr=Lw2JjD}7)wikRu77d9s}xX9j~TVC$FfaW7>Un&Y7-i-jv!-3iFz` zzpP2+Tr^vx)7q<8#LMjm%TzG}K||UCEG$hD?YM0%BE;4o`c6tiYx&8-M-C)v~T2 zjR_rE>z_XGVW+Q*u8-RR#r4cH!6mDR=HXQ5uJ2Y1zF(Xfr2FJ`bEKN~C2nl3=1pM6 z4&}-b!N2k((OHoKeB158mmN971NdobPq2%t%v;|-CQsV0***kj2O-#AS}HLu=%Smc>ZGis)zFau({DeCNMiO-J}cNQlJ#hDh$g5U zCtms3WL@2xa_|a8zUoM=a!98B{?$L8nhJNd@X)6rvrG=s%AIJ@|Wcd;v#TV3*D>xo+lP5zgn3bsI)q zz$7&EBA=xa;x=Lb3j_XtqJjVa8vHng!?mObhDj+<1LDc_$?jv0`sc$HI{IE&>gi3K z_3Q#$DYHVMk_@#agG=7JvOLMX7mW#CvwVU$#ft1>mrHTVJ=1l7xx-_wTZE($=lAa!*Mdi&56Z{lDpL=3-^LdHeLl|uOc(V!^^EaQ4+nsVpU{k9|((NYpe zq#sQ?&2OngJcX8-3k)0k%E}apOMxCG@WaZ07XiZ(GkhT>qRT^zTu5r7+8>Wksx_M^ zt{XBXw`3eL;|`9j-Oyr(Q#D0ra20HTUGi>w{XBKjZc7do(c=zA>l`89G7|+$^@0hHy+|VS zxm%*|`{Um(Rzp$TbJG2wI(z3;rHvF-GCUrUsoH>X^HwuD9*&mJd`w~bpD2GyDB&;D zNXz(YVEU4OSHis!cuN!|#}x-l#bpt|yCf76nDpci@J}twOXPDh-b+zUl0L~;g-T2_ zBgWk_-Nh=MaBqyn6ZV zFVz!xL6ne5uZ)$GD9s;o)-K(8cD$o`ldI+|Kc9p=$BSL=a|uqnD4yTIYnYdE^EK_C zslU`mk>-!%(1sKn7*R6DRU(nTpFg@(LN+IrBEGb@e!kE0M#nI+c=OqyF0d@NrFb{| zApfFCb!2>n-d8Fy^YP+?+Fv^U1X0>I=HH-m^&zV5;_pi5&;-iR+T2`t-%Z$l7`#hC ztHS2@pV$lOip!P?t3QZabEqpRf_OixH*sh#nH}On(j6oSt}@YYP(5N9b3@v!VlS^G zwRm4d#4V}vd2cYEXKwkQB;mXfFgz7@(34D+g7qQkea;cq5fVYhz>%B>vLonRiL)>E zEpCIJb6Wn8Bxc(13H4>>mZ|{Rkoh{EwL<}Z`e|B=?Q|%a6@%aQ{_-LB+&HN21}3QY zO}hJaDD-s@{f7WgSV5DPe@L`ik1=Tr-h510lTaeb4xvi-8luYn1QHvzDRq=M+L4}4 zdmK@7RotI@cs98zel3rbm|)=p9ysJ%`AdLu11=GDWNRW)%cDI}D99)iN_@Q};fC(e7J5PrfJ# z(IGkl)MfJQ=iRW6Gv&QQ^_l`2de?Dr7-3;zm~i^ee%6g=oz>u`DC*FuLZ~S2g$v7?JK1j784T5 zBsRc@j$~O5Bhn;+FIzGqq(pv=>y}yXy($}E5fB6~f^~KtAD(7ekVhryHwRRg-8O+f zHn5GOcSSVc=dvCbmmCyBt+6Me z;Ex^`pWw&N&Fe^4(L+SaMg3|RL>>sP-ak(n7$b-LohcoF0I>6kT!1-^hr`+3(riZ7J zLJEd;`tBK5fdLgoJH%YBU7mj2@ld>`io_jL=$biClLT}|`rLQu)ef`X$Q5No6WCn-G%Z;Q&nI%lc+!A2EQHN7aC_Wcwna7Zv@4n7Hwn1_3 zJOeLP6+L|R@FnjPkx7}hbA3?GoxP6<1iy3@Kb=1^*QwEa7F8eIMS}NvtHI#Q*#~ ziPd=MZ;gC*v6K4)|YQ=#0f$$38NWbB&*(Z3dUZhSUM0eztVj;6WkX@=$r9ZzU_| zgA8!an|oDtp5o}6CScNV*thbk9QU}Hh4KEj6dMxtK7AKP@hs|nTTwMm|%F{eTWC7J=XRmP@hl z-A+Z|>OV^{;0=1m;qV~gY_Jkpl+#kM zpN(0j-r0B1O3?!yfYa)+y*PV;XX%apO3R6bpIF$$Z;x*sNj``IH^k*be^$>uB9glueFuV6oxUX5X9-V-scwz7J={Udu&iOjA#`JaN7-?AXD-WbaLer8Khot_T2^Kmy@(a%pK^Z{ zRUS_oZ;KTNJu6RDyV*Id6`2k=^5M>Dwzl;N5$Hip;e`W~%Aay-K#3%zyxb=rbR2I} z@JndG6^d!j9>rd1!;X}=(svpjXIE#iawwpFe0XW|P0sL&VoJRFoMCc!6ThM@u*Z>l zKu^zRvdZ93i={Stsat^INsS|3>V}yt8{v1gkdT6{j$Fz<9qf3sIM|3r7X5fabxEj$Qthkrp{8@q6@k# z3R{u!>ql2NF2Qg^#N$v_Gd|6-#=X#l8`@<iZEXG^bgfZk3v!sRy`fw2Eq*AWIVCx zJAy5p^6K_hrV*81oB$-!l#4Qul|;A6Os7L00woPLMSk4mN$ZK$bBCmK=^T(;rb-uX z*E%l7-LUga?WAXZb~DY!l^v^TqX{{^l6+KEBGuL z+zZYf8l&xE_l3CRI1hK4Kf>xhG_IZ{+9_n ziA=$ZkB%c3c45sGECY@$<?xm7@;#?J!|u zCDz#)+(`g2y0XF4Fr|#x$8E^u}$6(~G; zkeoY(d-9tBuB2vlfx`HH5x6dGCkfL+MXvoymzSv9@tN!KnTkCxA26PfMFgRX-vk_~ zW6Jo--tlm?kIQV*33=C9+Hrsbh&Q5ABFY{U=i!;u=e2`4g zx!+!DeX<|9_BVaUZ&tM#WOyS*8KD2-Nza>Sqw3p>j9$6+ui2PX3p|%tQzuEl_ac^= zQ=++5F@P#AZ=%Zm214I2g+;lbV$3aoel&2&{p$_pdZ}=8)otxXJJps5Sq5`ChKD)U z{2&$KU~PKN+;8P-rhaE=p%_{R;)?4t6@RJ0`1ll2)S<>FS>GMYS<3l~@L6W2EZ^wi zuq=I^N|7$*P`@FM#IYS*Jc@LfSsn%^_Y~D|gc-Mb&Q5;MnRVU)izL=$@a5!FmTMXY zOVDQLPQ{Khx~R3$o&k#(7WG#2LUY;UhQU&_S>Ho6nu9ijUD--lTp*m$Hl`9|Xxw77 zndr87MfXF+d2+zZH+^1@A)RkkJa3n+z!o-s47C{d?b_{rW(%Xrl~{>)U9k2y4%;@K zEN=T;u>!$#kL3SIre1Tv#2PGEgoUEV8^ztXtsryu-*%e_;v_0q8^hLZuU{ndy!FMg zwI7fdbh)4ca*N|c*xWAndZTZF>YtIZmkRLBC9hqe=z;j+fq34h7S-FurhF-mFZGj3 z(rK@Dx!>hd)ujUade#O2@=cUnxi^1R%gJ0Bq?z~X3?AE~T!`cyCd+`JF+F*3@rQnrk=CaO?O-(O@)`qdxGrgzj3PDhp z^9j9YuMe0EAPYw1$0Iirn>H}`-&8i%Tt*`!H}T!!B^h74;I*1h+7ntc;jqtc#J8}F z3(zJb&ARgdaw|jIz~62e!7{2xCGcN|o3!8u;t)C2@Nzu@O<^aw3&foMML0FWHmuT^ z6mjfUa@Xff2BL2ZHtppca@LDdKF1fPi8zP-?Zg^nfu1Y%3ZadO=AyPpuD@oGFjKVN zF~OFb@G|u=f$Bmp4Pxsr?y z$B?YO4T{%)Q5y)yFcN&GQp+};2@i4bZ&WZ}xR$$bt^|y}qz81j> zNo$=(v8GXUF>5I8M3hghhm-`5=#~LT z-^nEreXX+jjsIjTm-rMh$K*7ly{ z4U6l9*QrRO;|-sz?>KoKn{_vY5;RPOd z^S3leV9~YNX&yEcboPEkVYv0*aI3mIr^NwRY;eb+KjrN5%w)7TiJs5*R*%X`s`M@X z^3>2DRN?EAIV77I>>u!fc*q}q-6nf7c63g=8%}Gvl^%5M@lZc`F({aE312nOggXmklcgW*=Q07X_B#lbld7|`aAiycGI^myLj4)?aMvd z2>4O8>K}@4)WWK6C2dwO*YF}=!W~Z{7vuID?YjB>{c5lKrqUT?w%mUk%@x!*6S zuZBIi)ufLdAuS(&>eciBiCaEMpw0dbFg>c#Hr<_p9cj7Mh zj^FwR@N!3+MA=Gh=+f%@zybEl+GNU#d83upU}LNZ*AOdGDHY03lBPTRk=9GE=5q8? zLaYj90(I%mK%Iv1C9KZfxeP&&tdPL>*t^J7dUtb>Ev2M+?% zO)AOPS)Y*&)6K55did;1w53hr-G8dGHq4B$4!Q&#E=##YJd^!YeB0(iF*EA9&x84u z$#NmOy-YupL8;D)Mi?c3FPv#aCsaD{$!{&|JuUjZH;g~)ZS?mPy+kLJpAo{I2#rt$ zM%<{1cL`7985|3jI0mO=`p6PqauKLrX}BCD{Ceid34r1oH(2X>EWaUT^|{TD@h1na z%TKye&A-kVZw)*W8#a=Df9Pgi;oT8e>UI%OI69H}mbS~l$6;Fsj2Xwnc5K>*qq<{9 z(sBjv>UtPwwKC;}@(f-}=T6Q3X)R4fVcNu1Ll^V9Q76o*e_O76S7a2hd_9lnA{d~e zvzU2TDrTV~SX}!)PV3OHNrpgA!aD;e?)N^yhO{cnyQ_GhRAg+&86s9;7G@+Qz2rnF zWFgl{E>^TS{eajYUz*XOFw+k6P5~%aAn!EA9E&fFMZJ1lyn9r_lp#j)6LTPm%dc#c zmwWjY2mKs1;=Q-;x-?ceQ9|ogIg`VpP zFU}pgq=XgE7qc#bl!1-KW>C0?K9c!6CanjwRSU&j6kDrjpDX1DsNW_2e9u9AranYZ zii1;&_{BLRSto0u%`=c9G938;2MG z!B_t~#JvXi5+)u81fVr6I@H{N26W-B zgRTBYDUD9EGAIcMjbp;sgAs}xe=uS*yC@de*^5EY6@hU1dOIHyR-qNBy`h;3yq-q5 zFQ$XFZ>Ub43-JIWT-XjItd6SF`K=2GI#PlC5n$ZoVfUMoz-MCRy6{!mX&|Uxyou!< zXROkpo#Z^t7ALF{wwBFjLotw6`!0ZL4hP=QM;w*Ef3mrfkdfDuh^B?^%K@(YiMDL{aLV{nzzb+#Wt+A5I&GhN2z{3Jgq7DV~mE2YmE zRq+=6g~0(M?DzeYyXQ*XRXrAIDxrhf#s0vVtBZTkPMrgE@P$R(b0aCxyX+2zFb#|* zX_QaWI4uDNZu48xcx1LOFc5o5niCcTG5hqdz1H%2SkM~ha=nh^H`Y!Ez)z80mBwU-6n+!OyxvDbQt7DzY&QKRV(=o;}`e#aqA z_!=lyA!MD8K@t6f-2D^ON;?Ey7;u8}(Ubs2T#OtdvpmzGSPGD6xdJX@0a{n^69ZNn z?2z*SZ;}_6D@4&%6!k4Q``{D=^W4+z<+VlhASYo!wK)jt)P5H*uG8xwV8-YS5L$le z3{?Kx$iMEXttjZ8^YkkqEl3mrV7G{V1AEzzRoWF5`_UAD9Y>-#7C>9}2qi7Nh`4xk qIadSFWmhFgl!^tZ6%y~!!M}vDf`Fib|0cQf|0RC@Q=0P6zW)MEz8bv% literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-19-Signature.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-19-Signature.jpg new file mode 100644 index 0000000000000000000000000000000000000000..13676602391fac0e2707958c9cf84544243946b2 GIT binary patch literal 11853 zcmc(F2UJsCm-dAKp$DY51OWl1NfU%X!~!DHe5IEtCb#wMP`|R>On>uLj3 zQ~*Fl`2gez;2J<_0$*J({)kk{Kt3{1mcYfl;6z#l185p^Fc=`CxiAzYHzo4Xi zNk#Rt+O_L9v~_Ok-umONiK&^ng{A$2hYpU9oSeO%`1tzy2Ly&ae-R!L85JF$@H+8L z(%W~*8J|99W@UfL`TDK6q_nKOqOz)~xuvzOy`!_Me_(KEcw}^Jd~SYWacOyFb!{EL zv%9x{a7Z{h{!1?^0Q664QNI64v45u*J4G*Q8X6D{_%FSvsQoAcVyB@!D^GVu%NTs$ zlS4$|89nFKS09TS8AKKD;JEC(ell{4Db1b3|E1dBH2cRC3;lo6?C*;Gt6t*(Gl+_^ zcp!EF0UVK{CDW<@OF3_1F?u1#K?0zP*9XpV{qT=ygo@%a^E1DfXX!d|jcOV@ns$Oz zh0olb-BU%pDhLi?l+5`5Bm1ytP#*q643AOj#o?-3K_9;1xA!bQtLuqs+ zE@&`O&ehRKguB+XK2@7Uf8_|l^2v|L7^2!Yu4fZ%)eF;#dKS<1G^aglI}yCdOA@K! zFN2$4`{vU)o%x5_u1(#xxtmt;feTFw1ao1o&~h68CU?FTS6si-)Bd-~@=ar|p;E!r zXuDcW3#pA=Px6*rJP*G>21+IueozOrT|h*U_y$m-3n{gtw~}>bKEk9rEGA3rZs-Uv8D6CqgGTJ+Rc+LX`S%jN=a2R; zX3B^nMSenz*~CJFI>kt;Th_4vr}2${mOENB@b8s1mff7 zuMaym3?p&Y6@8gDJguukb``)gHXC>w+Ewoa^w|Y^+R9y_)pv374D#X1|9r;KQ;H@| zaYUT|if< zKs1)~K`}=Ncma85y4xLxoE_I# zjXwteGU}5X-{C;&TMSp6JnBcLqz^% zI{qi;OcUwGyJsjn_Of_v=5{hg!6t6)0@(KxN zI4xmJCSS@5&ZrntwuiJw0Y!JT?O7lln-Eq@j^ayz-(&$v|9p zP;oNM9cfW~ znq5II-|b_UUfIW5Om&!;@QI~{m<>F}5jhB#a(2*e(@z8iCsh|uzRJydl7;{eBn<9) zT0XCY+j?JFG{Ae*Y6$WCK?Y_-myX|HVcEK(-BO*R#T|>qih}Wmcl0m)=6bMZgtIs@ zKstkFL%XiC{pv4Sp^;)0Px*rMP?f+5t%CYBdHtOS=8~FE`Zzb5J8LQ;`cgcOGF}oy zVOwXkb#g$ZsO@<)+q8dfi(geQ{&c@5CDc(wdG)JCNCVcGvYGL@>z|s&pHzM;3L0_C zjuMA8?;%&xwEptAK>|3PMqiQuj!04&xzw=J)CL@xUgXv=21y-(=tbDZt#Ir|u?8D0WN@H0}Y8AgEA0s-h21i!Wa zN~`y}VWkYmq3NG#M~-Z8$H{$!wJ%rKzw=@L8bAIej{IY1u03p{{-udVGQhkq$GtA< zI38x8yRmqk-bka{?Rd#$r4O*POxR9( z_ct{f7_8pe(_OTf39=pea9*H8<4s7JK*6Ec-XvI%x8RgjGGUw!wDy{ zviRD#u)pfQ{Ts>*7F>-k&vGiaY|C-U8oU0msxr|o3hXNZOR5C|(;D1qV_jO@=cQKL z4}xEN^SM{;E{KwjU@9Z7-f7%$Hs*(PB%ZKW@QwD+eYdVR*(ZEc9d$_&nLIf5sTi+fCZkBE~m^$Ahy}ZKM~yL)L}1D z!PR)4NXBNgw5IC-dyhP2J+GZ+c)ps76ncTeFBo#$5(ZjE)LQ2*2yTW7E?#3h`HFbA zPxd$P5GI0D3-n^; z#nq}O318pqDx!*X_U+q(s!mLkVo7`(b=0B#L8eyc?W*j)4EGn`59y>)2s@S9SbH!4 zzx+v8N}5~E=kvqJSZXn64S;mDhQ|um7Za@H|4wePBtZwa$)~4Xdnb==|K10S6ySgt zaE!YG7mq&SX+kohB<9#Tu02Rf4&gg78D%}%H(9L@>8s50r$kbN*o7kIO)@Zc{9Grc z!j7g2?GvsYgmDH42hqoNBL{A{eyW37dy}_hU_epOe*X_k9Eu93ga{TmA_E(lKVir8 zoK!n6Bx(KxIyC@nUKW9UmIH(-oNR4BZmmXP>%}lVW*>+!f)m}x`H9*R$c~|RC((j#mu08^l;KZXGahSOe)Ajk{Q`-KN><$k(S^*9_S0+=NYPwtM8G}VP zl2mNP>IwxPaQsrNBQK8uUW?U7Ej;G zS7Pg<_quwzY#rm-zO3Oa$L5c{f=kTM&v!Pm(4N$eIF|yOxb_9 z{=@Y35mU)&{g2V#wT&6ypXU`iawo83H7d%J`{(tZzI_n>==(_AjSF2t^zbmG9l>wD z{W;t?w;(CoS?-{`?{&h}pHcFij5jy7hVuzwb6CpFFJ$16^Rbe*%`9EwIsfiw1HNqn zwF-|w^Mg8xL!#P#_90&tbPLLF5v&%19OTbsN61k_uUWf>u&=|LDv45SYZ3%BZkp8~ z;qyggRW$q%rj_JATq&?UF4t6c2(`%!&-M-36~2B0am5PpmXm6>YNUCFo8Z54=ZBoA z;n5m@Kn@uQ64%+VFj}>2ZER69*^BroGf4){B28DxfB>8&_v^1A{-!_3fb_QAc?^NA zi&R?ViI;1<^pj5vul@F}Z=B~@Vv=}>rd&l?YQJyjGT+(6;61OirZ6Lyi8>mh%86^2 zf3<1EDL*$ExDwTV+%ELs`25;-pPP@5x&~qagUMSO_?U=7d_JpJeQ-6_OSdCd@w@7F zsDc5QuNt2+&r<)Q!~_={(#f(|7`e3L&s}IwrL86p`*WotonmG~qJ&mF3yQ}+N5B>M z*Zb{V{#)BuU`j*{oc>5`jTznhqK-1p!lz>)^Q!H}sXc)Z^hfNcy7P}xMF!zq?sN6@ zPF?rkMrKQ0xL$N6T2I2jJanhl|kCKi2TL(8?W$frImPA48o`b2WSrrIY zlm`wQNxYH_*9&kF>-#os_1Z+&s<6XOOQV=bNQWd5ug%q3HC25a@JYE;TE!@M{$Atj zJ>gI-L1=1XBGLw>Pxu4e!c$rkO&ne4?yzJ1xvk4_xsi|NUP8w{+>GksiITxhm1*%< zuutfZGW40|BO`DqrMxdRxziH}C?fBUqR!yJ;j<^Ed3{ef5@bJmow*qNvLqO+obQeY z%_T*ENArAa#D9VX9)IkDwJ8gqdS8JFS1Gv}59YFd45!Z#DXILRcg2SOh8t$Gso(JW7W^_9AY>*cMt}RWVK($K z)(=;{xjvF?tr9FUBwm_&m*v3f0Ut)=U0lFE;JINw!v_zTGvu^!>YF3LPfjfSI=}GV zc3xEzXXUGGT=bmC!Rk^H)JG6{J zWDVW=YH6d{EGm4>E$G3Dfahc&RLwCD@7~bIx;g`qb9a@f5 zPDUO;D>vT>QjO$ngitNs{VPi}qB8(n4wm4CYtfjZ)#SwQfdQ8a$Uuh?oekIMIO;rp z=ocKzVq1V-DKxL&Q;-Vmo8^fsxVf93m#;YOlI}?cP;Tjviy5{3?FY_-)}Sx`al;&w zRud)lIZ{@*;NIjjv?m7DUpg_hOBv$mi=J0tmyD5ikI-Ng`^Sm5Z5Sb9Y_mQ(>Me=J z6&AuZo5qh)pQ~r}|Hg|_!>7MFb9m`h-Tt%PXEQOFxh>39Nl@>X-?GEsGeyb#TBb(Q zS-jG-C$1vvTi#(J;-cWG@0`Knwd%SAD&_q038FoAY|YfAJI{@YYSpIjD|-7YfWN2U zp*G?)xBWs>0WaQO`g19Cs(xvH_wYQ%xH?`lWF%8(0rLXM+SbRs#*({!zI{CUkPH+Q zC?IB;q8Era@DpwP(}cX`m%8O}*AnEk{nUhn=0-HIC9TOsKr~(I3|hdlwVrK5J(Eb_ zJAk#3fpE>l)p|&c&YY|LR+%QGzlY!T&5y(LZfE0kG=M$|ycK_QZy$?Usi+<-=qcwg z^zoLuR}*71d1@#&GLZ0`pn+5GGG)j$Nbyoj$EFgVWl zMe?`U3h^K0p1!6#X&AY~yAbJfq}wQw4y|h#`GzvXyJk1n(@glgXfe45ctaAXwF5$_ znh^zt!q7^weXg8YSH|W!!{*CAEN1N1Z|g>aDF{Y2Oxk|!&vdkZn&nYMtF2fq-aT?# zIpyGVtC{$P_4*^tb;vzLdo5sZx)N5pjBAfks(rvc;o^}mY@4OnjA_-)QlR3h zmhqQ-yLvX9&eLnOspX2p^v0z zSb@)uk|kNFFi^JtJhEC;e|P0c>mSw zxg)6p8k+YLS^-~MiIFGnVKB(n-l?pTN=%lzs@1lh1pXE&-3JlEyS0wSNMmcU zPowA8gEy4qy**yvh0xxbL{h{HwPAxtWqq6JFPq!Le87Z5vDE#rE3{ABH-CCk&k*9K z-ZDjrls~aux?5p?^o`^*yGK}#=a%8b2(aBvv||8&NryNPw(*QYZpR4bH5Cy(0i78? zF~_C&`wd1?6gL`wpMQDR}~toiHx+9)li*Tmp;jKrsTN%FYAZU;7qhnfXd;ob54+D{XhQZ zW6G2ej4mXxBnQmrscG*}e4Fq2L+>3-%MUt@N2~lYBax*iLL$B18j~5y3%|mKFok>) zqh@)lu;Q#@uUwT#Wj$9l*?dmX-UR4#+^2O0tThhN>`w1*()KAMr>1KB=V_U4Oa5wu z{gRM;L!UDMe@d76iKrMP&@{}$!z0J&gcD;lZ--80#;M)c;@cOgf!!i7EyP~QsVFNM z`17!`f@Wbg1+!xNYQBD)$Dm(XARnRAZ%fh(;1{oWbMF;5n5KYAT5%7nJglisn7KPy zcIZ3)*pc9RmaqM!uYzQr&bnf7`u@p=dCe)hC7yTiVQY!Y87z^dlXMk}wnwqzvWY=9 z=C-##=uK*j5ls~8Gf{^*8-WF-aw%V(&HS+M!#ovrF=IXCsn@pd>Y&I%9iD?Y|X%wf%qk>5 zkQZ0pgndCkv)RVFm>QtMjn9Hf6_`Zy?1NL6XL!f9dT##|6!#(MR*@S#Md8{j(qNiD zi5EJwBKF2kS1eUmV-+-xx^Z%C-CNrgE}x~9W%Xq|zK15BYJ%y|Y;D&p!G)iMooB+9 ztRKhrZq~;1b2Dv%uV}n*{EH9yTcSi$qzbU=rb7$~d~*=O&6>wmi7ij_p1e006I80- z=|x|(B7wI5AO_(54i00&F|OhpGGe@O5El896Edwxqa>9TXuBQ>3#XRdlt)7O~zAOL$CSKiRa z?H{6}v8nrMtX%!qvt9aw7>X4Ixngxv%n6P-FY)q$bQ?n(tX7K9dSzi(M z0^Yq%j*F1n6SjpO8)j)A%j>kpXrpUpD2xw8TMg1A7j91yKa|Q}FWP^BSESq9Y451V zDH4XEm6Qc^=NW@5ezshDvwrDwJVw4@>jkgLF=}YO#0YFCUOrzBb1|)YIGrExRAaH5 z(-Zz&n?dz^<4y=A2_M>X#FSdOhr<{?4LoTd^SIA_>4C>Ycn68dg67X_#_q`m78B2-z-IQN5K`gd585Dz)*w+g_ga?IC5*Sv z`8l(UxXt!-(BR0AaY%DLJ?cXLA1zSX>d*}%1EY$`=u_X84>q+Oheyu6%OB_5LrC0u zQa6VWZ7TI!Jj6cK*R2rj!K$z~2Kp2v)sWMfBs27Ru=HhFssrW4*XyTlY$7^4(^*g2 z(SknpXWmJdAQg27ShXZ)>R+Qvntx#)5_PfX#S#)af7{+tgPL>%1nG<+f59hj=mv`f zN0|hXK95cVW#2GfzoJywG-^09sv7c)6Lvq@gH<;IY-QkMA+i zMa1g-#!LRrQT}JwMpP2OgCA5@Bu<<|<$}hw&1B_wNpA45nWgBn>%9|eg9luzX^=Wv zLhk7mF0Lz;Sw{x;NJ5?QBokH za$zv_24FdWVwpugw?qZDN4tyE56M>vuT?J#WxTAU6d`*XJ0tH2+Y7d}^I7qI`i}i2 zpI$AsP9fPVndS#94vq>hs`i7y(;#`t8#XV~h{+5Z`aYjpn_4IHo-JPKwf z^0GB`$qARUwKS6~ZFJw4*D*L8%b;HOSiF=!N1<1C^q<1E80PD@3^}a`!PB=VoL*Kv zhVnb|QLq;?(7I)f+kXl6ohvI#-dUG&d6aCi>?!D(-uV}!zL9)vIV~eMtJ91rx%JB0 zKbin&AvPYiPYDrA`ZViJxjKjd?)9DNY5(>+jAhCGkG94QICWzv%uK`GeKO!OUPA`v zDLD@XB9<)`>OBzCG`xk&rLaHF7{FkGVm|51C!&t*O@vpTi!SuY43WofQk&Hj<{XcocpT| z%S7cW^;;W??W*fa$Jk88>B(Of`vLklHPUa}=Rf{EtF*MDt~z>nm?wGOkwMV2z~b#5 zU<4A0*3G<`p&hSt?@P3Td1?sja$O?92iL8O<9HrgAB!8^Pna2b zwr({$L;VQJK-5H;&5zl+8k=2Ey-y_?xd3ZDyMMd@5+kg%Kq8OJF+L&#Mdbr_cCLAe zbtYuMslI0VwAwC2%`ghdw-4be8Mlq@H(#yEW%HEjpjh$&@<3-{75;nNz~;gE9Nlrn zvjgjX_!pHs`uk0(Ot)rfFdK(F8_78Ia}p;7^DA#3Hr;8j>l!@$x*-}Gtxfr@W)ybS z9N``_Sp3H}Q767|9{` z=YPeXSWPnw44C4-zpxjc5*+!F7dGTaJ(KkaG)eU8#%;`=!$f>kmvevR`nlb@yifkv z2T5SVrsPt`O|^vPC%XK#pfR>x!#3k?Z!GBx{R0$t>z6NT zBx!8){&$;Cj`ik{&-TawmA@TQfROfB#Q}7IL5CRF5_k_C^iAUkl%u?Yrm10ro6OIY zPhW@K?(garzTXm$rhXd7fVl0K?~c*bNH7}_XgW|>FdUdU6{etuRtS%vq$~~ZRY}lG zLe#>^Hgu!cM|Sn`*>kQk-YqWcLFR{u2rcWN-zMz_*ML8$so%f7Eim{`Pyj3GVK-zh zD!@G}5aJ~Xnh9oJH6<|-Ts`OYZwK75ZF3j*x~%tvCY`8lMWT^fft+zkf}X<*k&1mt zWxwpjCs*wDOpXD7NMWbO!L96TP~Ku(X@azu?cCD!g$MNhyqio^g!5dxHotXl{Kq!K zHH1>C)H93<7D~jI%;>;0^L#ObBT{!Jth*h3SO!_|JhFKZ+ir5t3$iQ;nm{<|@o^x3 z>@&k;t<1tyf5lRql_QoFn@2Rxv?3x2`mTQ2Z%N}n^M}ge(hc_)BuK}|0*Ld`Re$l) ziumbyiRedfKfkJImI8&+Z$6?{`}=ZYpAatNN*oZp{;6Ey-)VRAX467OS9(!<#)FPn z+mHvK&xD6_F0gP-hH8976gVLWu0Cdt`q2`+Nm<7@1q}~%ejI{_khpV%8cwsN4j^q#dhKS)$n+EN8iiR zhGNOb-Unf?oyY*oWj>`-8gYW}qc0qV{@XBf>^N)qVjedW_KRqivlXtl z+MQbuHTm{nWkQVmm=?>713zEJr^QH@yk?f^4oJEqhr3$ue`*LkePug?Uui8|h4Pf) z;*w3zV?~kjq6=MRw$vwXqOcq8mH-YbKuC4x^{*lW7D*nuX^-jAp=X=qmPRI%S17sL zKhol*p*{y`*(O}|;Tb3u>CKo{mDLfSI(7$Di1I-j;@)aD;+f_Q67Y;;KDGSIOhI>U zHlOPf`Y|P|I)ibGLDLMBNi2xrR51$Lt|Ag|!nMLN)mX!-Q?c9_XQ}GNaOd#RFqNlSL zdg>Sm*XC7J*u-PLPi{Xy zgR_5lX(|yg^Dsec*p@(=Y^Rm{m{wv`5Cg0wsBbQM@yp*kUw#iU|F`SxR?%w;w<4Hh zc!O>zL<7g!3pQT~33H6jZ_?l%owUcUE0PTb!EX}~FHreNBNXSzW1bF z$VTu->E_3x??kAXrU}dEvP!)50#{*aNTA^m4V5|faG78MqGlDIu7ZZgk3A@BweTn- zlKC^*)*EBm?tIkTYAafecZ|Z8<~B#kKIqEw64h$@a?H%QUq23NN4wM*(mU-eGkJ;C zJ%lTLFhR^xI=Y~Vb5H(VZ^`d{CjaZ#R6hx@*27>1{Ef1r$IOxS)fEw{XIg@HPW@?~ zK0|6#h{?udzt%K5OBXfHBIRS9WsBSXyPT)1X#?-~SDVZ9Q6Fu{K(k!bNFDVc>RnCu z2X5bM>W8P#_676f5Zp>?>@tl#!ASw!et(g@7Y|n zvS3yD`u=l>??X7O#jIo%R`NRduBU&*b@WAMr2|=dcLK;Bb~uDOxd8;}OKd@AOCtgs zbJgvCG5LV^QNg7xgp`ArLAUv*XjZ2vI|$6F^1fKuq}&vEqpC(T8Yo78v2{-RHa#&MZLqiNuT zmYSv-Kt%-rRNxPAGzO>u$LZ)GbjOZEAP{=`;|xsf%uI}oOx$cIq3nD-Fn&HBUfwf; z;=*UnT@>Ku6_LGg@v@|}v@}duPVt(Qg7_6_soxKwqNk^4Vr1fCX6BMQ%X?Pp|Mlmn z0bo5&%MYQUp*japvr^HpQXRDbFtDG;sQ&r@|Idetnuhim9ppGY10#4r=?Q?EiiU=o zmgd+oT3YaGKk)AWE$cD1vywOHPU<~`oO5QE3XJ)DT;OI=BgftE8-iCJxxAofIK|1u z&2wHzSmc7Jw9Hl6YjW}`w^Y^CH8i#K?->{x-8VL|vVQ!;#@5c>)y>_*)63f@=w)z7 zXjpheY}}jpw+ZjwC#IyPrDtSjW#@b?E-5W5uc)kQYHn$5!+vY;=pPsy8Xo!aa}+m& zpPiduSX^2rY;JAu?Cufw4}Qx<1Cbip72|=za;y=6YRzRC&~V& zVE;?5QGkht3Y7D$ks~or zJ8->UuhCOr;R-x?lhHrQ*vQ1PIajq~#U$!GUuroGoVWu#mo~C4+#p#?`xG4k4z?82 z9IxJHmrT@dgVHxRF$ch96^t4XJ-Ev>b7RK#9FBli!>Ek(jkX_BN(tE=&drq&OolB@ zk4WKNM@w8)V~m|7eBw~l+W&e^Y4={!STNVX&V3B;J3=H`Ru___XAx!Puu=6yXlBbW z(nFu?2$;!DE;|B(wH*Lfjm@jxvd6jbTggzKJPY+<18IQ<9^K)f8QTf~SMcX~f)}6CFOc7AVvXBltli!yY5o zAU-O*G2u{7a_4YJv1Rgf*}F7<@$ws(TD~s|{%%xw)RT%M;1k;sFhz$hMjQbOf=56d zvoBLsLWtE(0&)146}g#Z9EBhHG%3<|=F$;x(qXCX+Nj6Yejg;kFA=qm6DDuiULA=3 z(tZRW6So`j-3(Xd(7LaE_U99o&Mcb!DjPh$ZRZ022zY14$dWeWDErc%?giyJ?i*!F zM3dxR_c9-{5T>p6hsuaaxW{3)6r{dp;eGMW$xuPj&LNG12(29&Je0=lnnUfWC;IJ( zOLvq)rGFxCY}90r;X3;femG}kek)ZUZUU@%nyWDQiEx<>e*fO98K=<7X0JQubt(R+ zc)+`rV^M0wk|oVdBGgG2t+_0*1`oaUVKFg)f1nQnQS}j&FK^a}HoZA$xgv1{SfneR zCE7zfaj(8A&L=~!rqojH^oC|gBs3@ZwlD0FA*I?6JvZv?j)4AE84dP5Xa#gQKw&vz zSK>O3?bZ+N*r_HDYV{T4O$SS6ot|?8zx3$j4y7>U!~lh|h~0Ifh7#28!>RA{c9KD1 z_fRGgd^Xct{BF5#_#*?3026><|6E;>1Knp-@zHg<>p^_~6PH@5C~va;$(i+#JFa{m zyT4ex;`_RF1fX>x;qea#S$Ejue>u+72Vbl{Kkzt7;H`so?^Z^;aHHSNi(h-~r&got zzx6>v^(@-l>VtfbfZ6lD=t5Xex;&yW?R0&~uOwI#{Sm+e5PZZ&1>&YVv`(n|csF_i z=mjG-*vnwP;oW1vbDk4AfR{PP>Spi8M^t1Ps`^=te<_VQ>5?(s#GE?$`i0&VWq!D- z&(;*_vc#rX))7#63w|hbPsa&YTe3F)L&qKVpe(EYPzLaR%&~%Pfzu0HVw(42FuTgI z;hcEl4HsD(!*Znnnj@fC7ZU&7B5@c)|0=4yssgstg(BSBd4)R{o1tSKNV&)YCoBV- z2V(vGSe*Vh)QnY5%0grdxMV*p9RUpCs-`P_nd@>2=w@lExe}vHrO=JRYqsvM!tY_K zrY^wpwILxC-f>>-d~JIO;ba92oV%|PI}013O^dr<2=@xHq9+Sibm1@Sc*5C*Bf`yG zD>dLf0SYM*+tnBbFZKq1?2N{w^LMk?P`utK(Xvtb?4S%02gFkJIR$-f*r^uSjnaR= z^JDf5sDMH>SS>jA1MzD-6^ zm(U}?hg#2hdtKtCL+6xz_u3cezch5`B&6wNL(+ZzL-iwo!|&8KpJ*ZDd<647j@} z{y;@JjxHYokgGE~=ZIF-fa^gYBqb#cYB2Hiob0_w z*uO@4JI-OFK#FtCr{?{&aRC|_amN&`L)5^3)Yn&`^g}740SW~XJKQDE;PiUW8SNnm zc`$sEE?&r( zJr{6|L98~ep^DmXira3D2-`?g0tTKSwh^c7Vy`Xpy+@@3&)tli!R3Q%-$$yX!VaHJ zk#tOpiF)pyAFC))flG3r!?LC66)C)j2(pcRw9?9BW)C@U-A*x(Ac$RQ3h9RV)K_h9p)g}V~X=g)Im)Ouo>`T{CJKa)k%ZCj{a^j(3* z`GRI?YZ$u9_?k>EGz1CcAA+97m>qliVS6$ZmQwp@6N_st6660W_;BQL%4-y1Iivx3 zJ*LPtOX@J`JUEj#lcASVYLiwm40BDOD?ewwb9fn49@BLwQtYGEp-@Q-27f+`6byK^ za;(0k949(R9|5uvdRE)xt|nMqXtS(YW<)?QA_+8F%KVI~zEz})8?4vW!&EL%;8!wn ziPrT@3X-Sei^JC&7Ms?e+0ryE7CwiulytExaa2EEUGk#O2bYlr)6a!vBYop6ZrjO+ zSbQR|zZr3GZl!Swk`ksonj=J??uH%>TxHDMxc5v zDT#i`<2CsJzPnxCE?ziV@7vexEvK^T^AYiwIj$D;eBccL$+ zufMY(lCWa-o|cz2NckAM-aooQL%79-G#=gDlxxab*74!>Fr+ZpK0Ws}8jSR=1o%JT zzT`Rp>^OB4*YS*$0Br>l^IwJ5D`%KTiEKUNT&V(B%~x>=s&S^cfNcO3@D zw*m$OCE2F_)rHHE6l3z*VApM1R8;^Ozi$W)r<|X%5KMgbN_u$I^X6Q=w`?_R$qY0^ z3Z}j@#&{!x@}i)*@{(6CBmEQ|Fv-WB*gWdev}zZwJ7V&o7qjuwr?(na7U+ z(M2T)eAnXRjLtLjl+(8Cvj+0A0vu#bTI6nlPMFTT28xhs%2rjPWZqGKEWYB{t; z9@(L`|MZ*+X6G13=RE40&z}JQ^?LB!p?Hj1f zJHftm0xR};IW5BRAER`>(Mg#{z+?X&qrKZno2<#ZHG~%5g`X#7Md18gg|432Q()sSypg;3%@@Nl9QJ*-tyI8E=u6OOgZi{ zB5bW_osjm3!kp+8=ki#26wKEJg!=&#J~UZ-hbm*lD~B0zz`KYdByVAm!{Fi&fE zqqVV&pq3!A=FSDzX$rWGh}>hH^os3w;}T*{^W#DOp^MbVl@-kHa}nqiz!rcs`D=a-ux%W zO~p7tBZ9~4JSn9SPT0oT`$N^!-?~0Y^|PaT2W^f3es-#If{TRMj8cYVqlh|rqqKP> z1!*_xe`Z{M+uCyWnD@jVaRNsnE-uP{@DMX+%#n{vHnl9nK0+)hbGTdXGfY@fe0VYq zONE%T$72S%oE8lJ)Wm<|g-ugIbo+b+l=pqG(Mi}hIx#MHFk-i{7g#8pktDHMaH)=P z&`15B?COm4Hr7{gbj z??4r0`rzACPx3z_VgfhjEL*tz>N|C3A1h=e32(|Y!Ka+^eIU);k*6qvuTg!9q4Uho zp`GP}eKfZJ5Ci*nI07;KW|w^>q5j$Js#jprkt4yVM8Di(AgO=;3V0U704&#uVvF0P zVC%{DxRn}WiR27!ojP%P@eds;h-*~m8DMD>>Ya5+DtOfxNms#s1h7Aa%-YhhZNVHd zd()ekfq40{vT)vY_h&^55CDl@NVQJu1Sy9j{s_ zdKF04?f(vwE|i^|TzPtj+OFR}rG|poDWnjR>AwHndpXk+MAQ2f8{1W@hR;UbbV}#8 zWzfpUIEs_KObrNEdaA6cZ&3e=c{eR`K;g&+ll%*o7Yz)5jveMofDx;ls2Jkj#%^xP z7oOnAFEroZJxB3KOF);E1P7IxGj7cHmK1R}0G`rR?^WHqH}IB5KU52jfM5H+^J^VL zNU?m|+`5CN`}nmp-1@{mIvI}C&$i_jXNNCsS6{-|ek?243e>eXK^_>UJ3n*-K0$s= z^G5a^Gr6EV-rzIh=Fh(3@Hy(?y7W--&uZ@v6id&1;Q5<-F^z&F)W=`93tSz#w;()P z-6;_?DO6di`AezYn)fl=ykj1UNf(0oZq6A-Vl5z=GH07SI(1mA=6ROn@DbF+*yHD` zDx#0;)_lEepf?z2!m}P)3TK|o&&Yn-cl#SJ%{ULxnNp2rLP`uP7^EHncE)Bu1${dl z-S@fHvh2J#hua zGY<6Pt2RaStk;J%!yeOAaxE$II&!$)8<^4_eDXEc_OVwqW7F6hGP8r1X9ML7;b!%c z$0b4SVM82B9$2u=&qp;Sj)Xj6U4DyTS+61+)P)o&3r@R(QVb5)C}!KTHS?!Fdvrxj zFmm^?OaeHsnW>9B(KXN1FjNVv-?<-yNhxJvc)D)Hq}{JmlOjLoo{VOt3SzW?=!os> zsF!=+Z}2AeJLc1?&5E`U-|=KiN-dwqnP-cutAUug9F4ty5AqViFDQ9hha_ajF`icc&b zAOrjj7WyM^;d`=&EHf-||`PUg}VZOt7Rhq;jB_B*DS zS-+*{9_TIxU|Y=mQ4_O{5|?nouJG5(nnH4iDm0XHH7Dz2b2*y0TO!T5P#(ZV2=!FP&K-i%?5YXYf#rb>#}~-f zlrkF;+Ts(E@1bxho3nmEf(qRkH-9;<;qi8>3DNxNC8ScX-wds?aU#)JLP6c@`NE#vQz`Ewz`xWHX+907 zm%!jA&du!U-#E-o8V915A<|Tkmu`zCnCQ#*n~q0_eJg!ZxVf0IXC`S`tzY5 zoR&lbt;&VN8z87EARTZ-%14YfAF$-&wXAto9<3LEP%a*2OW4+~cw|s{V%t$}mc7VCiNz6L4>Qh_A zVbDplZx3j1-G2kg;@|_=lxYO}AjwGaFE99RZnc!JP(ssgJ>lk+es^7c{+J_inOx&H zeI3CbtndB!is^)CbcBW|$|QBBM2Vf~n(pOJH&(s1+u7xgRfQ4PR5ucr#d0Uqn>jO2 zjh0rcB~{U1_P}&S09%Vr;*Nc3SH4Bu(J*ZFar{)TIMU?MdH6of@n#R=tIE}1$g;K@ zq;BuB?1$nBIc7h)(yd?Ie95!)l)JZS&46nE*tj*iMHeEmdsrLnKOjCZ_-w@-)x2cv zh*al#a(a774)WckB4VY6W%;$k&ucvxA^7^+WDq|?+jGX!k6Q8>7p>?KnY-j=BQnw9hFJSg(|iph(82itMaCwJ zYq6_^p-6;)rR|v1`UOhtJk&J14@@pq`qdi64zcOn)@Y-{TWJ*y-&L zxT?e$**l|*TI4wb4j2tAUw*kcDN|Er`?CDhYaN*y8{!9Q_2pwGL=c?@;}*o#);yZqY8`lrE{|sF+zh|MaQ6Hg!Zj)zAn|&~48~|GeK|dB>6%3|LwG|}}eVY>l z2Hjl7%9YDO)iaFLJ?&mW z2cLT_(=_5rkSlk1eQ>-Y^3 zVEo+9p1{atC5qAZi-khM-u@{=JP3TTjdnKL!G!swHzg@b85Vy<#NV=EQXdXbq&}SP z4K-<>Lt8}jn2|nDdG&tYE_jA``TPvv^|tk6m7{{cP)&zsT6}oS&+zNv!r40+p${%| z%>*c(LcH7p@hD2A53Gbk)Vo=cLY38Voh3u7v&iiyV;0B+{p2RCXl>pUGRz#b5AGQ|^eRnw9szJRXUdT9*t23xi6&+* zi>A7QsGWO*{v;REr^moYbo*f9*}pofL*c6VxAtikiK#P7=d zfx^v#p`Pz&%+F8_>}6v5Q!GOd{O6`-x<$mj&$eTi%+9k@Ejc6QHoDNIHLj1~(=f8n zk5Zz+^v{zEgzjotKQP1}DdbHFPz-Sju{icZ=d0H;>oP5h>=OYCp&WZ3Q5gNX$E(Yd z?6WB^Li1y4`UnVx(ZP5|<9}&E30Wwd9?=`Rj&y$V*ONCdf?XPdHc3FLtcjy^AOgAV zQ$3_cusaa1amNQ4L~>}!>IQt#qziBs+IGTA%RLZAG?SAyb<8b@lG=7E6Fs9hvQ@~_ zzLVCzlMHST*=ZAx%&L`W_PX-35XjJ=H-sIN(a*@PWVB z%hQnGjfJn3gP_%#dg0*&SLh6g^z}(XvRBbi)R@KKJ0nLxJ_rmv_AK6|?o59)L@ojR z!otiPZ3sShom%Iats3}2h!n$C9>zq3W+0WQ^mqn-((bmO!W5-c$}-3# z_UcLFe$#=ySrn)h7!&jEWm}s5r*(`D<+3-<*9}+&VJ5@3$f6_+_0BQJ>;k)&2i(qN zv5w(+Yc69msFlVup-UAV`#ar$PY3eYkY!-r(pW8R$vlxJH?XZRGwuj57wa(6tx+2< zmkL_u8mKRo=q1(ZfD1Vh*v?j>Ccd_A^~_XK6gmP{3#<71x8on@PY3#)MY2w4v%3s zroOb(ISBdkW?+sOIMZHSZ&|lcAS6vy;m_pNNA2IpjeRH#PDpw=Q5aAEZR%Kw03}fy zy<}EJ>}|=>G zVd)o$2)C?ma&e}XQI3i@{vgmyaWWU>@9OC-jU5_K%#OVdj%X6%rO}?=`*T$x%J7$$ z+jC<`JoTa`cVhy{OX@On-_3L&2ifLV$O93eL5&)o*FOTNrKuJZESYe{wC$eu&p`;I z=B$7GBP`DEkd?W(U;hN7j@plZ>AU_6QP*&Y4;wT5nP(&&5Wd3+1%&}CJj}dP2 z3GwsLQZIh+qA6t*wUgWRXhh~dYRI0G9<9lIpI>wYtMh0BHKwo3A5c&2o+0R0HzL-F z#0DA&5%nGIVmu3IA2yEBt*@9x5XV{;jh$EJ&LP1Uq5Jg&xN5}VHZD0#{AXW&aIFO>rBR+6j22Nx!1q!xdMO&2CS6q+b;hRR=Dm{5m@O>rIm@CQl)Xr_9^(CrFIZqjYb)-VS}a63K!x z+2pMJq&B`;Et@S}?@$5X{CNb(10%yEd1+7ULu=O)sE{w&#;aPMzSU{U?6 z5~YE_6bmPNwtXL#KPy)~q`^^AdP9Pt)oD65XvVSmr-B zs6$DYO_I~CYy)VL*{ObD!ryY^8+|VS1!tET7+rad)Y@Rl{*28S`!Q+PxtnzV0T&ZQ zkP2;2FzNeVTC1X5&spWzeB;{P>V&=>cB(Ni*7ph=cn#ARWkCW{^M!KxVmiKhUtdtT zghdlp%BE3BSBg-vEl{XVy!K;cM~1SCk)^zF=n+%V7+&8NdoM1%A9+{v^sN@pyO(XL z(|bvN-IPO=d_?@a>Qh#+cXZAJ9+nIko0cfvS*9+?vS>nN-meToLH^uW$J7!eg{Q6G z{rbn#_sRzyxh}OY=4ykxT8DjcfAB&{u{u;3}Te=-hJz-;9eI-?%0plR!au#Ddts#$| zyH_LWv|AXWBy48WX*){-v&1Hq7u?BOjx3jWwBI^#efSo0vMCO|k;`KgURV+(pf_;l zhvRym;O9_*C1tMCAust0{8Ofnqu#4VzCYu(uj5iWLbROWPsb@LIf#jA3r9cQj}lZa zhTimz0D>BMYRi&x5f1)i7YlI?c1efTfmHQOL(EpqCZFQmmd@uz(aS~O<2JHr;pwWb zlud9MmpaGIRdkPwW>G9}qrPS81kr@b!AWY|gyr=eSy8g02n|eSBco{MG;!_e)AUDO zR4!?v=(~PmMl&))HFW~o*NxcJA5xET@^MARp<9OAN(q@Qqxr9XHaskfP&VkU!i6dA zEL?1TSInj53zph#fvD%L~tXVOy`yfCo6BY4l+(Q86d5L^hV>u3c3ruaRS=ua<9OYBmD zXT7zz%07!a-JR!9d_-ZgSMU9eiKw0Umxqq)R+1f2-)5}@Ccnhm3B0=FAb99PSIljj_*Kja$eWsq*!C$h{*j+p7d*(rf{!2u1GK)-I9Otzs!{O*<`2APukE`S#J-bRx0kSnl9*Iw4oYKY~GKz?E4uFBO znW8;0VsT<6LE=M}W9mj;&OZXC@%t2U8yfOG#+&wI)U5VFB3>rw=}H|L7pNnLk5~;znihch26w`?ITHw7g!;cfk6Z03$`yI_8TCrkNZo3(y5+ z9!pkwo^5B$^i4Qn6P546F+x&3Lkvc^gu?f}7u@_a`MgBpMOHe^@!g#WgMW187yygI z65k&InJ#cGV1Dk=W;MU&BfB>Xhif|-|D40o43fk`XRkv%~|# zBIHo6&DkrYW?ffq@Yx;L!!}ThO#a*R&b}Po}ZOmbOPojKzFciKPU3%F?(TS^Bzx<$0l^M;BGaHzMz}& zUP7fB4P`%>S+(P~+!&VX@>gF^ndQuh^<%wd> zY$-a+dwv@Nzl&2EqQbE&=toX~+(-f1a{eLz!dixQ0&u z0F$mV*b%GGkzV!$pl$Izy50o9|C0uJy^gzSME1(9(0SRb3rn4@W_yuo#x+(s-!N`H z&gPg#jko4L8<&Oc#UgzoM?S5H_vTGr@xzPRY$Y>D|0m-RM@SO!Da7zJsU0c&R>93` ziC|mUj8H-~(Kv|!Bm(FeFPa&p(jd$fluT zIFRgn2G-MxLuMT!3s#zI7H|2q3U5kNHGQII*T;mPUbg$kVAH9VrfB932yecW1k%1; z^#>2Xc1Ec3*C^w)V|&PX?1H&xlP+c>{ti~@(a#R6gPDqM4UWNID(+j%fWhCmZ~wy^ z{5!s&yBWS-$apu$Q8%5h3qS9S01R_-n&9f-`1qnDg=9#^)58iLT zijc)lw#9_i76V-j=H{ieS?#@$H=9}52(VAPYl3iT&6f4ejuXP zJVn@yx~!qQzLXNCX)pqyW&aO#j~wOGn~+!AdE2sjXp)!VdAB0~`m}BepD!+PrE^Rq z&I4ZE18&uwbWeE^K{Twz?Hdg*I{=`sJ2?!mjrA-#ti0OXrxQ}$82Z*%gF;gP%LM>YLsQ3 z+;m(W>V)}a!V1KqfFgh?u7P7!cdYP59iz@Jk>@(uCF2}Lb{5&LsslTIhPM2b}Qkwn&-oEi5@@0cjC}w zx&$3!BhGifY-N(YTveECt*iuRVgLw=X!0;Ai?#CF@u_&fldveIQ)8^MZ@yYDJSgMl zpK#zbAlBo=44-^g5TlLVX448p-3G~zB#OHu_ob}J1DbOh@HvnQWJ;*k*k8_q<+SoZ z1j-XUOgAZ8qev4id>>SW2A^YdzkXQ_B?=Ir#fSjp`^!Gaj1Of-cbvUGAOFs?4uk!a&n*QH%p_e%9&(6S^b4l-P>nLozCwbl~r(e8)Wu+)>bl1ZRfI zge%Q|eD441jwQ3h?dQ^+v8eGvm$z_6X^cZCHkV!8Q*^Bpn%xcCi1oU9jZ}%W^A=^f zoLUGHJc-sZ6Bsja>Ah&@JN%3UD%vC9NjKI)-l#w`8>v*?Gj+UNYtNPiN;Nnba6(t&(JyPF&I(?Y$Gewd< zR-#}^v_EtGiNz7H*R4ENBDfwk$zLYu*7XteSesMCTgV$9n1z<=!tb_DV@}@2S$%h+ zhi~`Po(JJVQTN4vxcT149{!}nV`Q;tR;~LvY2K@u>43`@`C%ms~!3U zMFtGM!hM(y>Jj}J33zB&=HY1YC=Iz?BZrv)xf}B{$!zFNx!G+Uz_z?$isyKIYiXr0 z@m8k5c2?3fCh8u3RT7qlA=~HFDGfE8MH=LYC`K%RBcIMMJ#r$fA!?u?0pIB?Xf1O#TA6U4)K z%%Z^>!g}0ouYqU&jj(~Ph5dU@8=h_1k#>)_$iv;nmJxCEG)dgASl(B|K~^O+US^x& z*41M-LB(uIQ}Xv*LrsqC*dm}eb}c;gu~PB3X~SB ztdz?zAW;O9peiwN=*!%w8ej}7jV^9Ud8iS2G=;U>h7yze)O-$J{7Y7({WO$)Ze#b$?6HAonu?c0Y8CwZcruOv z$r4k6vs@DtRZ&WxPYY5t%{X}*MghsNphR_<4>9B9Bb$oL>zWpM!?I{*laRvR;p-C)y#&{`xL8z~AevH1&_y-%6Fy0jtKI*C75%O&Ic!Z*K|Qp1a0CX$?^o6vS2PBp9Y^m3BSJ`pqaXO;utAq*Jk!Gt3nebaFk$uFBl~pz{y+RC@nB zbuj-5g>}M;o6=L!2wI0S(;0~{bC1#{)<~C;Cc^Sz@umSOmuwAhN`1iSm%nsYDMp{yi zv#<@GUkNiuKp2a{V%{tk;3_gqm?1Y8$;$IZR`P|w(*f_P9|c92bGK2|%Egt!)SSZH zOBQeZ#q5f&K2x~DO6&U7um13oC}vL_Kr+FcW%EmceVG8?zPKVjNLv;$1a=*iI zq8njm7fn*oh^q+{HaNpK&x;3nj_c*^=7s1D3tZ~*HXhz5N%JON=nlOKnGKrnrTmg z*mzbA@3ea0uLFPa?R6sGJ7X@@iz>WFkcyo;YGY07LnA8@Y|E9~58!Pd5*M0>0$PPK zvllXqXn!*`+H}NGyvye_=j!)PZx)(4Iy9JmmchZXNc%B00gv}dqrf*lWN+G*v&IcVU^wH@HhymA%_Yuj{Ka5m zHjAdWtb7kKr^Sj^g~xm95wXo1J&!3(v4~v@fA}^}{Sm+@tg6Zg37!sfhu8F=2&2H3 z`R{(*T}AyWpR>rcBM3AtBo#6=3G#-shma)=AuYX3;ZjJ&r}uaR&H|XW0EM_tR}%BS zesE2T^5QD|tv+9b~+f+I$SM(1P(bG7%w; z_$51Mk?M%=X5&`TarKj5c-87@0ZOq$NRmuoxFlbQ344vJIezo6=qKN987*DI!> zaH=|~fS2kBNFKJXlh`oC&v^+ccywcr04rsFUsX>wP)sn1no5-FawfSv0|(8nkGPlT=YW`CX@^R_A0@K4t{SPcf1dzHb-vwW|+_-Ww z+xUCKTNfp2+W10sCPZjF)%pay;3YB_6q2m>UwkTz=p5QPHCQx6yxzws&Ue|cKtA1SM0Sun0!yT#}d3x0R0iLZJoPh zZIk=Gm-)2veGW((MCWHyixf84g^nU+7W)rV+^cuh4*5g+)O^*ib{zru5~>Bp=_BAo zwIh$}Fls!r@C8Y$!`RK?E5t8;th|+|v7nx)GS*eyOX<|pA-vGEq9-Su6Im_44?AE6 zoTH|DS+mb9j(GF=YpG06u?7XDPArf+R&d0;tg+FiB~$XNE}A1Cy6*_!EYLjfw!3gR zumGCsGYY3?t_-OC-d|kEY~nvlBdlKC=1eK58X&wfrlEIucG*Shitaelmhhp)v0mei z8*2tUA8-u?BX=73kMehTit*4JqTftNK-cJPLh>&>M_N56$-dz9p?S}B{x={W=*r%0 zp8b}t>)E5WG$)8xOwFLHV;S&OEOr4*-Z$~h2vkJztkV>>_g9Ib!hQ*Lw|It}yKEr- z92I{*ZpQ!7@wX!tkPnmYu)S`;#R}}9FpWml@7~6GS(?9mq9Z=1US(v0VVkz;8s_u@ zH3ra)y9)4Jskf(2PJ52sL9jSLKUN_Z zAdtDj`>>Np&hkf#ENYh(u+#qWdNp>_?TPGAZQ zRQdjC?ynf%xg{NWp$z$-I=!VnM!O*i9$Nc8PrpqlR8L)`z%OJfX2rnWQ*u8q_AMKO zZbgKqYwXtG$mzTcZf@SYSJ7FYb%$MEOk7*BcsnYukz0?bGIJZtvH{ENWJmA8JAzLSA zH}8xXHu{kp*2Cf0}g^0ebw+|2nh$miaj z45e3TFSzT@wwUt89RXjHS(r;4RXJUaJ2(gPrR`+b`up{#(me+5F9mC1_&NC`=Z=8g zd#Y9`Vi!<++9mFcPbSCzVj{3IpUI!MqGC)oO2UBtup9EbNSjx9_wldk`p0pCHe6|r@&unT4LHMgyYv)u zTV)C1)p)rwQxd!_`IL_Zb<`cXcOJgv3Xnk?Wb&^sc5+1*Yhp{!eyKJZb)02hirD_~ z^m9yfXYOSL{A!vvc$ui3G+F+tvABf(^Xs7pPrP^c!2m4jcHLbSlaAxuZKF|nzjR;s zRW!8YOMS4T+sA*D&S%i@vxvKBkg{VPFwjTA!6ReOcq5@_1Mn-#y_YCGH4X45agfo} ztDcOJO?bs?g*tJq{>L5$2bqjDU@>Ex*(brG7?u|z4t@@EyinP6YK5AgVP)|C{&Df(cRWTr%N}V zS&jn3RxC)X53_0U7!8tJ7JjYY8>PUXYHfIeiwyf6wW9q#IqlDquk^>YCvb0$!L)X7 z5d%|m)=U*;QU#p4ew;;#e!-j|U3>lJ>Zvm?jz2zwnFGVP+1sc&P`##v(C41obf6Ar zM~XT@GJnD^G8rGTcurUHVAvXS^5W&?7aIjG??Fa)%B`ub_9I|DPCZ6~wE6pnG)(w+ z>4@7d!89M-;Zc%x=bH=+tUf;X`IQX421>U1*VWpy+Q0WadpbV1P;3`_kc+=8DaRVq z&@(#aZ|>S=#tAM6gI4a^CYam%g!mncDumZB{ME5q9(Z_eZFyRt2Tp^xzr4j%lWKD0 z{aB+RR+?%-+u=Ew0OE#{X!~VKR`x!Fm4s9>i+aQb9f$ZUh@6Wmk?P{2_eq|qsh6+V zXTWiEW_^$(Z>b}o7tGPs>G0pfY;}XeByQUfqa0`1Z#*|GdR9WD2>g6Z`=z53{XVR{ zHy`jWkCPy6fuer@Yd1DU*_T;Xp9zk%T)u=6+%#6z@)!Hbar0utmuXT;fI@0}0>v9n z-n3J#!KVEs%I-a|ZDF{oZ#` z|LT~pEJ<8PqDM{sz9dLd_J#va63?g7L(L=vjtR>(P~K3@Sp4pKy(mKTuaJG#UxsMi z$OY=q9kyB0N*h4R0!zrSIO_UMQFZR#^Ued{=rV6k9Y*cz6&(RpQ}Mo$C{6|T-otf< zNR~)Ic;DbO8b$qSN`xI*t^y>{J zymCQD`2kR}e8pE2mIC#BoX=9XsjV^sYhSI!1&jMj$hjwoQq?#}_;+^RJ1Q+a*hmQG zKhDewJ1IRftTGjT$+}#Pt%(2pGK;xo0|)E!Tj1GASi=|o zBh9k1X)P@1!&l{dHta z=_SAv~8r&NhJ2X#`XG3%so3!#xKAXpnA`k4m4Cpf>tkzuKN zo^3H*cCuEYFsIHBunA}DIXGkof0|f-F8e^<637{^PtJz_&LtlhFKM&O?#ih0jk5Zg z0r372&tpCQ6q(xsgx&8<$DoMQ08=vs3s1$%j;e(DEy>JUzdpnsRr!H4Z65iS@xvH< z=AWPHd!YCiK{CSRL($;SuR;X&9HiV>pR#sWg7EBdy8gA1h70Y%dBz-p$Tc=u!WP7b z7Mh*cvWK#IiwO6Q{@lqNFn?|BSWh(oGaxnlmQT0sI%z88_xXQ(biGg0rSFt)m7}Y< zKQY{C!(=`#IMMyq^79-y7sz#KTJ}UAAF~|zrCGe{{{FSBTW|0F>nnkkNNxlzETnY$ z`R2L8B_MXrg7SeQQ^YoJs$$wzWM&@zN_36;5{$(Art!?Rrs+I9nMxwiUkE-ka}S|6 zEEA3lWV*M2{v_l?yhq1JobK;{>G39MB}JiNAXipU^b-h_z?rJr$#7N%H4_o_T7-AN zD9RE67t zRgC&_#ne&BHWOX@lm^7$@3gOK8}F)Z(sF@v_K!gS#4s?uIyQcw5s=6D7`gGpsd%r< z`z0-u^uOuQ{vBMm$YMkCddfDce64t)ZhX;K+`n#^9~L26;h2-VZ`CO!PQxi2yHvMj zO%J;Y8hsG0>XazFJ>yK3+M7FEe;^(;Egv&wMpMmH`<(G$g^ za9@nEf^F(WbLuzy^r`)e_ZXu7Ck>W#cAnsscY#ehTO+ywA+6ChLU2#~$M_~}DB1p# zq5GM^yGQ`|D--X?V8XiHe}Pxzf`rq6VLzt%>Dhn+KH;(#Htry4?E(dc5-G0Ty|N^A=L>KlX-IA4Z%vJ!M7IfF zSfwL-WBjTvDwCdA1b)vJG|%`aZZ|s5KEclLtITuH723uW17zg$tFbB`)K6YJg>BLX zjo&c~e5X{tz@Nr@`Q6ltrl|+u!ADYXyz%|_@vJ2oZlbDdLD?tMbR|9ALO$g0b6#;s zJBdi!kqJfiH44wCQ;HXqI!ju7`UHP1wrptkYxY6-m-yoW(*Iv`msjD`k9w}5ttXc> zlB*B(V|@03sF%uVa;kH%lVZx1bWl8CFnm9Y_WHP0bW>=gqwz>WJw$rnkG@5Jc; zoeE!vAXUrNdq*atewn%KDX^cr8E#Sa)P)k*&&ksZC-k8IK>_q_EV=i1Oek0Oekj3z zl;z)u5Ezn7;}}r#>1pm}ICr2}vjY$97f)@qqBryGqvLq#Hys7I;5UK$rDLLU?`KRs zi;-I<=}UL|0m1EG>FV{L&h2&F)z)H%nR|d$OpoJNqpW=p(2`Ef9|Ud||FB~da{>N_ zLmd5dLxfibrodTv4#ZF*IYE??Y!1T}&JbG`^LWOBB}eTN{QeA0y2XatF$gGG_y;SU ziR@T1--{Q_H>#p?gBVl$v?H_weACA>M3SDC+*BrI0sZ(!fl`+k6Be=1$g2a)nmpgG zQ#M+L$t>-EChChy zyJcz3ats&(D%*jj2`%`o<#E-h9BqFEU?#uRZC^U|o@A1=F> zd@dI2k^%v(>%ME7ZLPej_|bIH825^y<79?;;!DdFPrSoRFaoPGYzuyIOx`aA#hEUC zkm$L7);U$H-@-x z-88AcPHRl(7qa%xCg=Z3K-LpURRu6Tgw6mmssg2ZfEpH1{0!>t;VT^52 zIHwW+yxz^Nu%!@Q*hH3b`e#^|#){FU{PBQiStmooUZ%f-i4d?DZ4ycSY-Ef{U-aL(<9(*7ZQrFGXfyL~U3G~Iei zfMSZ-(a*T{#Y(-ZyAU)(*uVgX`>f9hdv83*F-L_Y(U^B3yb=By)*qB;C#=|?5Y}xH zKqhOl5z=mqh^a{P>O($xe6=!~oorwNp5f9&cXvTJ{1_F+8Fg2)u$8IX`CiVHL7nWX^)a&rF1|E*`3MgOX&ggw5+w-MEKe)JCm( zL$qsVupg0`=1aH_Q{;2qQ};2Q=j`4$D22L#=878{v2*DhHt)s~ERvyr#wRk(NRi)~rLi59s!)`EWl zO5WKP|6IcdxN*I_dw|l#HT-f%Q{R2-`6H}rB=vV4WFHD$Pw?((mhO!Bm>`pC1+%Nj zsI_Y5E(?$>Y@l)UaKmhLxEQzYsx699g>e72uMX!-XhYNPH$MAiEIyKGY#h?!iofso zq*!6$0OgmBm&J+^khj8qo%b{KTn>^%;O@&P0+ibieRAL+ck-Zd*pz3hRKcD?9aFZ4Og)d z{GIqEY-c9+biV6Ii_9kCheV{9NCC$*dbcFCv7|QO#c7=1U+b9m-=hu?sMBbW7zV(! zjcRXI_i(T$jhn6-5UwhS)pfEpS=EGR%znSaGGAJnUtgobd2dLnnYccUO+pEi*5y%R z_Eq$!wd2AP`D~DgQv-R)J<3;f^CV#x--7m{rFQeEi*u@1P3$B@m~oCpL-nk4G13*@ zQOZ?e6ZDRA9Mj+7YPr_fw2rEabqZoN=e=MvUVtY+fg|bQzZFs8kTq{}?v73u=CH%< z!lnClPan$t@}R`db>6sXRcX2Y^o)-*$A*8Z(q2`(2k0v(`70AR$3@I{^bbQHb2N{r zzckjBn}%vP?tuTPb2LHxeFR~MGPApeP-#XG@*eF&R9RoK43VctFBs#R5#OnH2)GZz zZAhlu7ex8L#~s}0!VAJKZ*?8)kDngcX?AF`&`+J#yTu(A5`Xc(gSGEkW~175ay{fj z03KWn!_>X{wih>(d|^@MVy`}n6(J}3?3HooB^zeKk2i?=4u}odeyXvEw>*T_`bor( z!aPu#>i$MHfFi@jpVZaN(c_=>F>h2Gu}9l`2MXcFh(MnQ{K^WDn#KkB>(SDN1B^Uy z-4o}$xWAEYw%0#hs*LMhGW-KcNr8c>v@vQFDBwM&+gI&^-cHvdb@3O%Cbb;7tG`;3 zdJ_35@ZuTY4!{J`T{YPLi0w!3J?cCAPesKn2y~4IX2Liqv!TO+q&0E6v$v;^o4@)$ z5D-bgl?AqRy;<;Fg~DkqD0{hDVB#I*p8bJr@n_67Jt+#0l9A6PF{~#c)UiPJE@DX9 zy;)HLA?&l(;ugVmuRam47*r|4TXVhkQE~vkEHH!bGJ*;jIt8fRfhr(Dl?oXk#I>n*6w`I^hRgEadFiDl&M*He#pl1+KuvVCFPmN^VP@ zX&HxDl+^Wm^j0?JmOuHTxP05g1oHg5;P}}3=zu28zIgv9DAdfT5M+<017c)B!1 zQ&4aDsQ}eBcj;#T0bsf5lgG%Y$ZeN<%EjsM#>%LL!nUG5bZPXbLZ;$C%`>W?LjCpg zC66ZqaQM3$6-X+78|MB4fnLIDuOd_X9Aj&?u0bNvfYoq+2aNg3l2Ha?hFgn76Yyi1 zW9xW8ToE)7$n4z!7!-7N{R6u2g=gMfQiRzC5ODNL$?@X8LRt1Fj%%~d!OqJ(HuO3W zo%f4W##t`zX{;!vN87wAcEOutOJv1-SWgLvQuB}{MD*hG#u2(`OBxv9*ab-~f%y@6 z`!w7u+9tG6Ik-XV5mAmgAfFCPZ{_0l7q#O=W(T`YQi|c2CgERZU)FB=eU^j%&Y~qD zTtHmf1(&7XCr#KcM!)h{?S|&f5C`Zqp7*zDxR1D=7xP+@xff9QF%|enJkJP2HkZ3; zZ_6`77kZ!F4j*2Xj_P_}tla)UEC){+cEsEuO)* zlY&i!EDqlrwGyD~w5+gN@YSKA)^mDy8-}Ct>{>L)5TQ?QJ+NghzVCW^qZ)rrTEb#; zlzqyU2h8D_8d`vZ?!JXsUxw=*&;dLLwHFR>IpFG;+=yg#2!oPTn9V7 zi}8{wI--B}_cjiW+k&`m$lo;7DFK?e!7Uy>SLSvpN!O=xdouO1s)mzCZ& z(PzH+6)r35YFxVHOs%3y=&F5itce6>%eMx^v{X(=C(O~pYVc2$rZ$_mEl(%6`lInz zU%O_T)gu=r9)a!6J}*;rKq7%XSQ^X;O~h#oMEwl&tm=(7f_1LApTqzTVpRkM$;|=F27OCCx?ft&IZC!$&2ts_8Y>y+C#9u^% z_O$A!QAI?^a*Yr|udh|p4V-Iw$CIL$cW#rUx_K`(3tuoC+)<}pXkGzD3L$9?T%RM) z5{e!<;Wf9r!CRoVhrm7Xr+rA9;1@dw92?FGbj0x#gs&qCz9l$N3t2%-NE~%d@Zbws z%&?9*W2VPX>+dc)Q?&8f?#-x4;yGgXOQHj59Z)Y z<$_8j??J#U<%vuRAPK53?NG)YC0#jnLZ7fyGPU9!@7vS{=k*T-uI?mgHQU9xGh~!u}Lf7&R)dqe@jkD@qmyVpRwtAiSS&=!7iw_4%o-zLTvixpv{!P^bauGA&W5?%^TDNy@e;9 zJvAoX%4L(ATTNHHo^@mm2n6%;diU=Y+bsUVHk%3c(!1-KQbBgHb}lBiFYoy-IST(M zL$AzgTGTlVKgM<6ziTy6N+%B=Q@W|SrfCuVyWI<6-F9o0#DyG4ex9XT?}d_9B(4NoS5Y5G#}uH&&$XVhUS&?A`vw z>5O6GeMIU(_O`Cqt5bZkr(QXG()b+!&=G`>zbJbyMcY(Dc>W`^WFv&Z2n&jnpZS zk}|OlNK9PBHB3`I+jE9vWYhJ2Z5R zAu3-qXOxuWp1$a;l%rWcFg7SIPMcT?lX{*)vvQe({hUueBpwL-n$k)&bG)NY)P_m( zN<1@T^L|6i8OEn@Fb!;asn^1c@#N*aqrA_xjwRtuP(YwdAAYMGQ_aO7mjygm&8HR^ zn!;L7I#Opc8l#onl&QDvAe+*L)JtqWGItj#ow78AhHhT1*4)loJ9S_*dB5@pV#1F; z=@Mgr_#vViB%UlL4VGD9d15ju#3_PJe|<^FylrGq2WxYtNvav~kP`DTL|%vJ!Har* z;3+?e8b3*&A9hQHtT#9>UH3OERJs~!Vpr8>UHW-XasPN+A2X~j$BA)%U13}HJLd_sKYn}WNgH=}(4u3GoO9^8)0Sl&XiP3Jm;^kPcl#Bj)x~4cs-55MC`iKlA&iE{TLbw=+|zJBCe-7T zHY)a>>;W9+m|tm0`8~sjD(jl+E8%hJP=s0^7dhF#1vdFF!A~~`EaB`oW5y-becm@^ zniyV<>|2i%`R+nqgtEZ&hdKGfvwO-^BHtBT-%R@PBc`@7i^A_+UE?W12V0gjgfoifxQgh1w?;noN-wMBy zekwQOvVQU-ax)(b535v&+9KZ!uXq<19mrf{*@TJb)YTfv(VkRki(BR+c?DP?j&fm* z48k%Tq5SKIbeHfJmH=|~PJWN}*LDS%m`wm*pSzoSPj^P<0g zXqp&qtWpUQ(H_p39P^v9eP^V;*vNw)4RCJ_h8 zJdNCc|GWu0ztuD<;xkWsBRYKHzVa`YhLq#;1LUdN=4tHtn2$n;&y`XXtCzdr`G7k_ z$=kZaPf(1?n9Y1l@Xmtr=QN5nsA5Kls#}35?{*yFT7Gyt^8BqjP3q||!cOWL{|V40 zZBdUc=dJz@4u*!IcpXUJ7$CUgbf3#lQ>erJheKIQ80dyrl9tO{13#& zKyVvRw^g-anlw&m0MN=rA<4H-0oVD5%l7+;Oc)yM=OIv1R}(H& zI=e~s6GG(KeH6WZW9lV|TQKDdYm}tH0oV#BHL2nSt^4Mok^s}rC66h|q!12@|wa(N=%zbmlbEEF)3*`icE93uXyN%VOr|Yj?;4nFhvPsd9dwfyRI$d6ar>ikj!%A9t@{UfFTC@-9TF0_6HJ~OSrfbzCw5w z6^AE-;q@Ut`wXiJZD=T<^V=Z-e9-j=!f+N^@3ie4vw-Vw23S~!7I~(&dFbpF4ADsw ze@cL3I5-I0c#63o^r{2EVTsQE0O)0kzEC_@ATcO?mO1}*dKe}KM@3k7 zU%mt4Ln8#uM%jk6qsh0dXzOU&LlJ>D9hBg&kyKfCaxjC_(t-0@0Q!xLHWcR;)q1D^H@SqE)cRO%k(pOFpFHLpQO-|^MXTQZ1T6q+7J#U6# z|99VwZ{A$yuJ0;-gZVCrdX#(+UY!aT(xm0$%}xfA=#pY&42vjI+FN|&>-x}{bUldu z+-Q07+d=qTGHfWx9~x*v<5v)O;`|3PR`wllTMct-^`yqF9P1($JilI?ab9SYlh|}U zgFLe^xBoyGOP@kd_5~2ovS%EMg~qkKBxH34IBeNAtLTiBazT=IFLtT@yK=m#fuwR{Yb91Zj=L0+k`tgx37lP}+eYk80md`-3pm8N)aY@pnJ!6{gx zH%4rhs-sUd?BQEryv&2dmrgS)RcXnJnU6cp1lHA?JB0o&s6Mv_%w_{Q$ z4BVm1j)|>_7g9C>|3?TX#LR&qI@lNiw-l zj-^JeTa{Ti*nn~jLS<(E>{#vU$cloKV`7Q1GbG%d&Y!K8`FDdtUd_(?@<q5b-FURD|A-o)Pd9+bsx4X09Gah;&@abP-JlN@gUA#?8&qSxc^yHFienl zE+Lo+=%dblO-Fpcia1%W?26mN{(;;CC6Qp*X^bkY-GSM}KaedH?JQ!}_PYb(_5ygA zxJm)r>)k&P(={~aH7=YGbci_I9Cst{oKc#*b;VfmZPqIB$kI~Zr=XOI9v**}DPLd~ z09Lhd;EO?XRw?TxX_@7c6?Nzb1?KV^+{zN=n~&-a?0zMtpa#~9Tm%Lt*O&>TAS7~m z8gvt}a30HhO5_aRnaCwvn4?1sFzymgQ^#Qp4Yu%Q3D=_`u=M@Ff(0-IOydlQC3wlY zsCi60i=ByUE zp)&^7F~n#0s)Va)O&ay3dekI06WDNGUTnZ?rSj2^Q|{5$=3|m&r;b?PXXXB8Sn(C(fhpDSS4oFz zr#G{H@*S+1Biy`P{qAVG-R|MP^3Lt8{ojddHdTu!A2rayvLRKdfMf6Sdun+$;!X4|4|HK3NrgpTkKjdxK`|Ssd#vd5oXO>pL|us9N$K8@{ec zSt4N5fYu8{Yxy=9bldl~tw5>a#EXLv_!z1Klh8(s`04$b1DQ+(;KQy$lCDD3vxEi4 z(8x#jhYyQK@>r__h<0dvi4o4 zy=@b8X8PjK4PU?FfMIk|7E~;!Q4JQHy7&EK^%!@2I^`X`@Og>Xt!>|19`D$XasA6GphueQSE&2QP7Qh6(~rMmCW9A1KvPatQjW0VyZiC z-RJfbrSKIeK-csU3&f-JizaCl<>UUiE^tgG!w8bt9^T?pAi1g7WibSzUJCIXTAJl_ zsEd|5bOmSnfA(Q?nB~z#{%NB2E+w$Xan#zYl-_*5En)ev_7B8(^ItiEi@AH%eL&3|7~mt zyC|rI%nt*Gry;Hjq;XHCgJcbYq^S5mCNK0yE%;PB1iY@Ea5*Z!Wo2eLn&21!Zt zH&)_gm#9BzAMh=bH@vkx)<=U4<66K5wQ)nl=z|n|AxZCQ+3d;!Z**y5LcVAgqMxho zBiNiiV_SkH*3l?SMwg>ex%bqSy) zX3(4GE&-HT{xaCxl6u1#KlZ4BuzAE#86M=X|6KEZ_Uw7!J|*_ZTfH4v$uNNpd4Noi z;j=%_BGI@?J?!B75Tf-u3IGsScfj5W>B^~!3zfN_bt%{UJlOnhYapmL22W3~Z3bA1 z62&Rn!)P|tMBIDg|3Dz3t1`FCOq;sh>kFI}L5o{a!-r_7zn;8s?sR8CwnuX@C>v7Z zE#2#LhtiyjfYaI~MK2q$ea4W%7xgp;hSTN`qZXr8Wz$`ir@!o-@{zpEX`~E<;zD!k znX9U2Cf!n#IZ3@BcaAb0S&KrJMhQ@;@x(sZ+$J0SKtWarBJ{xu^e7*s~rkyCld zh8yAjEJ!Mn--~depUwvMV;p(7{RSt|9#R%{7F(>6&_=E%H#hOzieOc(QFMAWe7Qh4 zxN_FN$2y!)S2cash}gZLd9r!q00dc%APXO!&(QPm)BeW`qCRai7@pG@b-0Dt*#Xlh z13Zs%sOaS>_9hTH8wW{!>&&0;>Vx6m@zELQpMoF-3PYCd3|?CIWy8jXt%vz}*##GqE^1ISb2Z zpGB@|2Jb%^-69{eAXG{aE;K3=0-qzmh5Yr|-}rOE`S^8Zlm+u*CUl-B9P5?i?_l+c z>txv%!XLa6CF&V8dkNS@33~(E%ALiTjZ=h*^*m zd}fL zg+8BhMJV1O&TvHe59FdeyR4!g7hd#LnpK@U1!RL^9^gPAFny&_VCnkS`VYiPQWCT- z>as;yRpXo&+<22^hq=6H+==kzhCCRFwnjOXv0Z}*qMsq|lmtDZg+0SMc&OZkdO8{T zp3GU7k}U^SoO=GfQkt3GHLm#c&1aHZDMc@PIaYPy?~@JJkVGqTl~PvP6`e;CLMQ}{iz&I&ed?eW>wKO1;8cUk%biDeg$|z!sKHBl-4i~mN3(8<7blN>-cdI{#<&ACXdv5B>zoqeE zjj{w@<8CPa;XXtE;bRh5sZt=<#OHH_xiKbK$0w@+^7k1`x|c!JjEp;=(O?SOFUl2B z{9sx~o2TTBQjc16%;O>`)DvauB3}scL;2&&BzM&T`kKw=Y9+-`_ z;LVLb>ijTkAyiH3b$v`d;KqGnNrzsv*gZXY5PSv=hvEwCn0xzbs_o}mC%rWBfk86* z&z)uz1RL_$L$5EFbY-m%y}6`W^VIhVjh{EJHtSp3)p~CTGki4!7T(+OtaoYZ`sLT- zBb<8=u7JnkXYWFWBbkYM9f{;G8sqR5Ii5^e>}vJ%rY19A73^~SWh#%JsyES2)Jd>X zZDhMI!*F<-dBbx=bqG%Db_6>S763UZf&<6-_h%`wBDZvNW~~sz(%aBz873Q6B6d^& z->P{)6h6iV-CpnE@-EfC{2r~mT{S|D)OFX}#H>nV*UnI->BDU!=kw*|hB1k;S&{qH z&Jp6zG!E&^+a~~eC*1YBEOL0!Im9e{YN4i%^vS!$3lm`4v=*YCH)3I}f_Fb3bC=&* zALKLsuC5tcUGOVGr}}W37HEbGQ98-3Xd0)xCU8b|r`U_T)L=O}Rvb$Jtligf z(_v(1QW^A(?By#fHo|!=3et7Y_S^DQkh8nfG%AnPn*ZG{R$I}uatI~NW%H8JMOJ#? zob>e}C9ISxeKBp;u)lFB&#IB!DPK{~*?)N42drlCKLvgs2 zT^{dD%+3001541Y1!Ms*-J`+3l!R)0#Jx?qKK2XqweQ5)s48tyQ678DW(GeI&Bct7 zkU0w$T<~4kq)3%tf(H9-?1i;r)8JK>PlZ`DM>39!Mr5o;^<5|XY2v8DX(>Br8&6qP zH@(6vYV_psYcsSXl0B`L%f2q;JnPpu{P|I7KH|mo$(KuxKbK%zsTuwXB^+9HN!=7L z&{B(E053Urapyv2UB}!hUd(9!)Yt-DXeuG~tKYaUCrFg6{$nvC%d2K?%R9YB)!1if zhd3^Sx;VC*`!R|TPlGuIprKCFHOUA+Q5Ky5M9V52_4Utme&aAjh==8(a-`d?{R?Rz z)U0LT#Ko6gP)@c^7;VVZ+%SDbPc#YwsJ3wc-^ZE5*OD#9;&6|QXUPagoN#+W-CI>2tP(# z$fs7^<4UOta22>xpYmu$s zugP%%LbUU3(03JhD!46LJw{YHrO>+c$D4u%i6atm9`6xL^CqT#J(qcphJ@oPyTgw~ z%S~X9w2BBG-@%BSZ<5B|Z;lY_X@xrOUojWOatqhVKHJj#GR6QpG@?3Sj6P4LJh-H3 zhhj!W5sWfM=lwVvRl-jwqXR43!T;X}TeiQ6OM6qTM}O|qTum76qr|jW&*y;Ivmwn% z>k8yU-q|nCfoj*~#C2wJN1d8~@kA?^+~6b0O6X$eeShz3soCe)j&&GLy+Hx&1wdMG z^r%9M>=Rb0hSg=%=}Y>`chNr(oVvff?>az}ewTxX7J2p(zAWqk^curFSfyGSYo1Z3 z_nJ@l6BHq)gs~C$$^qhJb2zNLh&CAjD|62Z2jy9~EJnjM?jcUL^EK%m;>$*WCz&sK zNh-LKxK09h^f|(Ktu-5oxO|Q?3+|-{(f(!cZ5p>dEwTZuQyqMn=N8^x51ZM@+>CbH zFQMWi$p-P(ltmXFA$H5(29>+b5{ka(UBbdglBdardqryi?j{IZLKYt$hp;U3WzmLd zUh$69O&+nf+_uZ@OaIHP*B+Lrfy=UB+2{#! zHVWFDUIs8BWmVDaoqGWfOp#-?K!2XgE-^Y1!dx94yT%UP&2wh=&Q6Ihnc>(oY%cSR zgiIxLvR_SE)_A(r^x44wIDfzAUFx=tKmUg6vZ0(#b z<5zcKCtF9uwprvpai<6W{Us&oE?;v}*I*3H7D?~}n^yo!Y$POk3p{rHcECVadC~?{ zkA@I)N;yEy<__jj7{bPha$De4%;}g_aH;V!Jz+EQ zHpp%K2gozD)s&b+Ms)uB1sDhlmp!+pGx^rtHZbbTM^aMNRF36joO~-FVfBRfp-8|6 z)1yHrcb|g?w^;+a*l29{86JOgQRW!;$(>bPbY$0%-i6u51RCt`i4wJF>&_Z~we}Kh z$fb=5(faQn@;@>H{|ApD&-TBpYK;rAjW6P$*>sC0GOdmhT4}->HjdSkB)G zuBB_^Lu=!M)9o&LyW4k0!J3hSeG!4;Xp)zye;qyT0QVVT5h-odW?UG)n%C@k;y z{QEpoSjHc^t&4_=Rm9FWb(@`U8QFv&%2~o zE`k*=fAY+El$G_Q;CB{jPI`QO_eJ_-UI@|G;@<4_*V(D9^Vhev2FL>F*SNgXOJ3`T zYbjl?07SU!fBm}}U$MBc$8xSjlC`72Z-l4V*DR_LgRt!E^;DfO^ry^e_OSb0V&+4w z+XMx+{DuVS=r>Vs&p`}xP)*?NmqAC-UH} z(pdl?{BFQeht5+`_nN&7}1v_*cnOHTTDfJo8yLrAeI-M7?$Hm>u@*29B-V; z?Web`hXq|9)cNWtqabpBT~;OUE#=euK3vvDWLoiu;gt|&Mf#5=T~w4yGFrsiYOnhi zE~?*>gm@KiWv@pDPl7Sp`SjNqT|FD4Z*7HK^aJ3^AUT;&-*_m&UR~>iwSsN~j{l0A z?Tcth8Qd4NPN!q+=)2W2Kb~S(;?Hb-3l~iFva?eo&x8C4kVP~~p=-DZzSX;8=hWTe zJ;Z%?oL98})AS;@d+ddyo2PY6CwO#>>*uS-k=a8sNrK>Nzy58qACe&g?f0i69$9%# zC)Dv9Z=hFjo4DLVv)>h82Zhjm4?u!>@Lq4WVp2@YQ#@dvr%hFue3>g7J@YZt+Nm~(T zvZ?S^E+ynb;(>9)jdP>e2%-Acb|K~ytMLp67LC5Bj~bTlX;#q~znhzi&dr+m8z=F| zqd?&LXjc7X2}w&W&RV%8E%L*@uy-Y6S3wgB@|EGgF>d3%TP;jeBzzBMgG=t2l1;0zpe#Nd(lX&=c^@S?B@p^t`fYw!}LTihC7aJ+JOrW z;^La;fZ255_U!S5XQEqk_x5db$Kti2HI~XTq{MMIH=n~`gY0~bKf_1mGViAC*_r44 zX$+~V?|EFinYVgnAP zmC?9cFmCQ`J*1K7Fy_8`hc*9Z4GF7}CgQsuT^E2^5v^7*H|_ECgmq5lCG1fVf!K8s zLAnu0m%RM`!puYnXb=Y)nF4Z06-KQ0y9@7Z{L8o^)FZ6TYXIC^7@K%owsAUc8}Rt` ze3J9BoD%yv|JBapjc3m#!hM!~sY>B<{`bNdM+*{^#_vC2`8nUg$ws**e>sx{m*gY6 zxR%MEcA}tEdx!VZ`HIKb8ulM0QNour=bh=->2OYqm$RN~b9+o*zT4pv18YBZ>u_}aU?!VxSRN-W)lhbs>7CB)RsJ_4fui0? z9(RI?AHCN9_Oaq}r{sPaV&VIi^P-WFTY}BarPDUk(w5%uc_1zYWpKz|f7oPcXEm1W z#Gw9SK=G!V3e`1+N1QFhA5onaOcT@HVav1bhFo+VnYZ`zae^df-kpB#95Pk9R}+SL zH12v~5<18Z@>PMk}a=L*O)#NDkW!X9=MJ(+{%HxtgxN*Kbd(@KpBg2L2P@&vc+8Nvi zkEJBVY+9wA5*93|V9X9<#U7&Hbu}xUqLs0*=g6Tl?}PD8G(`h@>`k0of1@_8v~E}v z)0x5C(O~Rb7d%idm-#x&)R{s-#= zKikP?$uoU={ZGx3BMN+EE-s{ey`#fRvMUO4Kwk2{yU2D!Lwe8XNI0fb(zALcbpjVh zr#wGL{#*+_Nc&J4-)nAXp0P9xy4W|-G>~xI$6nf|gaf~_&PM9_(*_Ci>vb<}-^m}` zhA`XUXneMO!<&2Dgrm!>x_a31($EZ|L{XS+in+y@(1*cq%wUPJlJ*m$a)P*Oweg+O zocb`8ocBovDxV?b1;*cT0m0Z8{cLhS=8hR`aS02C5!bLZdsu2^Ro^P5a#wL^o5J`nTTI=}z7Tm>Ru({v8U1y~ zMd#?(84cbYyHtmn#fMdYB_U~YK-S=j%FH?h+>Gr)RbEjt{@{Rh*2U5J->h!L(wt6b z@((SDs6ER4^zDS*A!UrfK555dI$n;;eNcBR`gN{rZDY77QQK{%7oa~?-ai>X{qN9D z@Q*f8tMB5%AX#75es4+cRE#H%7y4g}XKGJg-8$U5a&CVfkbg~I3Vk3i^z2{J*6%fv zX?(gLz>8S6&-f=H4ZyC9MSL!SBGL-!Tg6iD#J$$yXN|Alh3|5Yg* z?aqQmuI8d&M9`hbU*h}ZlG*{j0i#1SAToTzTb})1!IFN=&UoHr8ht#TJniNttl8un zU<)G2e}QMpdgU^2%Or%CgS;7EoD$x%JKWbD9Wr2+568V56r?%|ArL>|t!rAoYO_z+oWoF;V;c&B9VKP(OOl=W$m(NG< z=ZRwBYOPsNb!k18zWh8~wW=1UU*XvCjBndDWw-)COmm5A8yDfR#R45F$5St^`E#PW zyuS2RykK0sG&qj8#e!ZKr49OvCyfR};<1!27fn#~M3erC+|??ca>8A-UH8<*S(_sCwSxfYFQvV zO*D|eYj(A;*)p#V30*|1VUdUiM@O6Gd3$G4?^rmM_lGINx0gg(q20718kXx4VSgYx zHTC=zy`28Opz2~%T7)2HeyzJ&~eejF& zY=^P1W|rCFD_+K@?e{u4oj!Fi-whjIgbk)*Q*@gQE#6vRtRdVkr{uh z&ZWWMy&|yr2b5#zq*q*ypb^Q-=;d{cNx zDx?$VgK}VGzg}MP+2>l=vURM3<+tv<`ImA@@0Zj~9E6$I-yK9UlBWIA9&n+AW%!=Y zl)@;7?j5nVq+0cUR7o>fUv*=uS!{T_=p$6do%u|VJK8Aa@LG`c&MT|{K`e}Mjax)` zx$>I@Z!*Fq)be==+;|qlUT?R6W|~;{`nctMk+b!DI!oth+12{dvN8@*MedSsP*cF1 z4aTUv!a7c|kBI=1SF?tT?wr}(Vs%joI{v)x-&*2GS$<%YBNJMElSY^1pV>KPGHAKJ zYz{&gFaJvDegBi|+>a``3Zg0R))|i(y8e!;-1i7q|DYACT=XJEi8HBv+kNc2nTTgJ~qEiZ0t)44!`4CSFdQpkN1_$>R z&51IsE4AgzQox1meF0?a_7WJQscfHTE|FPQZy)jx%fLaD?BIKa1s3aQmE;&U<@clM zI8Nc=h>YQUKjoA(?pmKwB@<~t-3nawK(ChINU5j(k0AahlRG zpVdilZo9iFk*L@3?t(MLocBkEc&8qE#zE09n!GbwWQcq(ue$zLv-?pK?u|;+$w!$K zE|=g*PrTz6=`XEz-Np2!*O|YY+qmh*jn?gexV+#^=KpYavfaw7Y_8pN9xsX+{TA!h z9gVQqFctHZ4v(lf0pXWtDBaxv2baD_afwAF=mq&J9=`c`Htl~Pq&!vOEA@kA=hN;S zS)E4?j3aj>TEDW?sb#&hw$56I)TP zjomWP@A^f~fzatckoM6n=iKy|UaOi(li>ojdA!*YWrM1yC+1$2uRfY_EdF$qyen^$SM-Br;aSif!ACGAZ+13o$^i8T?GhnH-T=G?3kRI42M5>f*jZ2-Lb*bdyZ%iW~UA>>g0egrynk_tP zUyFHZi)Qx!q3*5YqHgM|FPe9op=A^$ARIOcV6EsiaG^XkxRyPIlxs_ zW@5hFLV1ug;lQ0b zk5{+t!u=h|aPu1F%Fm)uK)~n?HnW`LxHE&BBIo~}mShsJZo4TmUKaEG^2*jC?Z7(+LT(R-L~Q$JYt6&ypVZcuX4M3P%qxp| z{73vTOU5P+j6t(Xqfc`Ig!=YhcpdzIx{m(e{(V(ppp&A_Y+!|JYk!tZCu`mCcl}*7 zxuu?K$?|=h3wu(f@3_|%*h!PTwR_3gXP3RuS5_U4)R9YFro#u3>Rwj4uhkV79t-~m zX*D3hlGJ=w(t9|h;T8fPF0x7KjdzXAc_*>mhKO|9{m?Mx7jyGv_)$8y0 zI+WOn8F;p46yFLbLs~gp`Ic_LZW+xv-CN*Vyoy{gzHgj4Ef9`QF<=!i_;&1?yjB$L zp$R-sp2)v8@6N^s8^lJGJsOpA)W1shLd=&ymiI;&NuN%GsbcMA>;Ym^K!#BQ)1XM+ z;XEZ$bsH6ABJHGi0}s#Pulq=qZ+6E`+Ya*h)7MrYWNAb9i0C4N zCBql;>|Yj`u^RjwQv}sXW_g#)nZ?}l`RV~<PYI)%oX z$woK;AA?*o^^fsaTYVx`5{!OhT(1Lut$J9sv$vcmS$gJjDac2;Pc-}jDz=tqB4L&* zBjLkifEY=V4)QvHxWNjwl>o_?R~atJoD6Ln)Vg?kHQk2}7-lx?o^Jor$oqXS-hAbK z>fsJ~|7gREz_p#7_Oo<&%l)P+_`NHP9GKMl%8CY%Y zMI8GrRPkmVLzj>VxqH}UiUj#l5RjPwX-X}lZLHi4zL0KIydN=y!QZXcKrNDa1G;A< z#A#4>i*U_{6R|^D3n-NedE;I(IP>3r*Lw>cy-lzCN+j@B`%f=LX7&(N^rcZf_sfrT zGYaeN8Pq?gKeY5t0Fk#xydzgnFw&aWPByyI%}xj)&lG|WgUK50#S{@~@F`G^)EcI> zX-xRWps+61@F*m`1E%f8ipvHJuZK|ZFCcc3%e$7FTp(Ck3ytw@%9xSmT*H>PTEKDF zs+LatP}w-3-a z)ZQ@LUXITs{$c_*U&(=mhkX)@ym$cLYadi>@Zli;M0HF&VS+TKRw~uP@dfr3pN%oc z&#kw;_;9c9ad`76Y47ehR-{mghOPB9nOx3p%J>RbjAuN#iI_3KhZEhKfQ|%lG)U6k zXXHo_A1Nof`YrHvb<9&}f6^o*nl!P-O(_sjv;WSL6rEZ{JDrapwGZ3gVSSn&WObvzc_ASUcli(}n2EJo%`IyQGB%YDqvUtwjdq zUu(2hMQx1!1;Ob!_)ypgjenf2`9{5ehxOwr#l7mI4;}uEkORLzTIV}>pgIb`FAyfa zo!_xb1NvGPX;f+z9IwlE6Es7P6-*6a%YLbALeeN6Y^1{X2Y|*#jwKLwN)D>TT@@F# zmO9xSx{5cS{i30#&IbiS*Yn&3HADt{8Q(|YY?vbMvV$fxcx(!H5{Ixzg!lVO==0`+ zLEv>hyn-X*uR27QIj^ zAAKm?i_i(oE?1wUMSZ3B!$909>WVV^jVru@##PKB)c7ZJmko$FcD!ob0jC-aDkWaR zIHSs-M3p!ZnM-GQ-q)VT-FR4ce0?G2xIF;0uoIY__buXRlt_BtT*3VHT)JmN$OSpd z)c8M%7C<8pQ2`w@QioK@3LfrWk&w3dJ=?;`e zSNrC*&c$Z_P>jj>BG{A2HU&aacXcPqg9HSA1x7>}QvWiD(KW+D!#KpziWScyHqXDZ zBuz}M#NhRedB?vmk6e3QeBV#49DAiocGk8L0===v{te9E1R2x7l`7@2t_FY`t zns!2ydO4XLMwOn?>&f*a~*h@}ZFcpb3 zs;SWoOnB6^GFGzuclb>bMX6w-H1vJ$hsaMogBh(zdfgr*ACR@p_h`vo_$O5D{!goZ z|I5EWr2<9mf(4HyP#p<`3K~&X_5?ex_2|lTPbI584sMfW3|bfVq+Hi{>&meV zR0~tEC1DM6aEiw9#*lTIemc^Nw`>oJ*fydtKT0-bZ%i4Q+c|Gs+?K7E_b06VgdLEf+#aOy<_Bi_U3!2hMP%&Vyr>(W|Y ztYE-lE%cn$!1D$j{!fF}&H zK8JG)o{M2`yrP;M@e7g6tKZaCd)co<%rDD}j0n4lU*Tpp!)}CKor9qEa@)a; z>Bi6T=}~?p@Mj_hfjOq#V{Q)VYnf?(LEK$Wcss!t9>mYS>3An>c_z8#DB!`FVHU5yTdbkUfR{XvHuN%6wmv>Bl7sM8VAjhdcAl<4v!k zex0fjkc^hRJp}gW@O?YL8T&c{m4~=>&gE017LDVvm2iR@H*OOR3=s$zJPkk(rok0} zM;GuHM46nuX)9_#j1(v!v*~Bea~!}~_DZ;ySxZiMyO&d-im1@a=TmFd@Gpqk-`e~* z@&{g<*cTt){&p^u1&@Nv=jw7XdIEn(?d)EAm%Ko8$SbnCAmCaV2=)(k1>NLF}pZ;oV5`-IbFd%<+I`q)R@{SEau zKIb8myW~JMQQZn^P~klh`mLZK3%?SVMyx0p>$DGXbF}CxjjWLXzJEiXTGeWT?-x~5}}ubNZ7ORneA zbD+Q*47|=W@lmrY7q|>6mU;(Gutm~OB2acf2X;AM1Tr0{-r#U04&&E z8&o6q`-=dfcMn${`u>ebRlVY|3ky*IN7j4YTX$h+F#eRKs6dI*{TAQlukl@xh;`Dv zfwK?cV2IJHw9g$Dp(E*EskH1k+C`d5Cit z4%(9UOLAlMNQ88jKAfD;26QM}Ig6GlRG)p=-75qm9oNwNISu&Fvmyu-3z(?j6HAnsGNY0q$Is)shXV!BWHK0&jrU)>!0`)M+TKp_$LE4$ICiose^l;k2 zvsb<}QJQ@b8wQvhPNxGrFE;|8njyD#>A*YaCTO{PD;R9iUaw=&FT_MRe7FKUmpnK% zK`HIb3S{>}3&^`C3g}#{{0MQqWT-<&G64-I3TZ$FgHLdpNSr~-0XxmAeEe~h`)Nns z$dl!=iQ9>U4KD`Np$df@Hs{}`1!5C_;!;vI#7v??9GB4(%`^K)VKN1o= z8sfsml^%BWEZYD^>9S(PH&KY;c{}O48;oRFvOh)42w(om!Y>rBHqMdWe&3 zn$O?APMl^zN#;n~=g?2x(2XuX5sm;2dxDN+pldEt#q8a?hBiWxQQhFsO=)CKJM_rA z@z^d;pPTgVUe_blt-7S8lgfgT5!-_Ta|HH|&T`5Hp8*76KZbZDyiilAGgdC-w^BVw zr$N?~Ykl_VWRI`lK^ONha2CAutDaT%lw(f$f$*fT+0SgKAXZV1Ew2zQ{iw>Jz}Txs z)^iEG%49@{&&_*o);m9-QqLt|t^#4T{DJq{vdA=h$)>$|!$JGGkJD>;JCk0k*!BW_ zbuewVF62_SyYnJxarxTO2qhTuY|C4D{n$FYNL7>Ec&`kcSi2K>EL9uqs#&dg*pvm@ zj~tBkz}NZ->e%Qqy$0P?@~0;#l~CeM%fL`{BQACK;tdACE-2&y1A;yxq?YEX{8C_G zyZzSpL&b8Kt_i)Dw4(~1KUk&!NU?s|`~<_w12Fp3)lZtCT4<^S+m7VA!S?7m7~1c& zY~w}ZkTy}k#-{ox46-?c!vygU1^8V0X6pVk+K%{s3AY%ebaT2(M2TT3VH&)ueF0X> zi$4IVGO59nq(yY8rS@_nMKZ+$yJHJLap|=Q0FU5Nc)s76K_m3J0^%mLr-2~a0Q%Ttl~8ZE2kDaQuj;$S(|(pK3srk zX^&Da#g@g>VrFk9+x!HtZ4(vl!|BF3kP&qH{Ma}QBCDE>jxKKZfj&(Jwp z*ebxF$f#6!JyN-291e!U4Eu1;nlXX>_tb_;NxYdIhkIQf8+0Vgyf_oBvRA`#MWFe< zw97Q~GQ+`!rPpjl+8NHCcD}XKDYA)p#VvGr;l_+L~eCF`gnYZ-?ygJ!$1A9~Mb`u>8@C1>No`e<(X zKkmQ({a*ZE?*GmX)%|=9*%oHLY1tsUg7>MV0&V!(_xy39$H@}d$J-S2$wJqd2tF=( zXcmU{Xm=G-;4KFXZ1!~Utl6PD9D_4*sPIle)4&j6V!?SDky@>Bui8_hKPw`SS5IxT z_U!R`K9|gY8jF^pRJs3x%w|8wDmIPZmK~TQd(rB^Kz`Iq{$A5KX%dEalA49mOJ8-_ zF(XIhTTHkk8?p@V&xwzstB7dKNrJC96K{2JV=~oft$jp{o0kEw|Iv}ay1iLcD5Mt-Q2cI(8$tMr20n$eosGshCKFq$X>hM;M+3XCb5)cVOK9( zppo&?h8R6$Pr?^>)GJG^Nb7U_*f8qd8);f_S#f>FF&v9(ZD=pF4XrlQfhq`Z&C1&S&4#-K3lH~NEt8YnpjFN zkBz=)y_QC8;IldNh;@lSNcR=}q|Xw72HOP;IEvdZ3BZwOvWfXrtVRO68<0h{s7 z7kI!@=2owL4^c)}t3Iw=aI3to>Y@kPdI%T{YO_d#Q)@251NF3g%?8cJJPy;^_OfGU zEETRB944VWrOePh!X%5kdYPGxxgI{6*$=6~qB*)D5dO!k>g_LPa${x1gSl|_!(Q^( zK4VbuyO^df|1)8_=RnmR(VZ;Ig!fi3(K35GUf%`r5P_`%mvKM54wXJ0H$*vx9-;GZ z@PSDm`dhdX`%H+Oz-BCLFN7dAEpEuTIS28a%I+abX(b|qs9DjHF~9Tt>hvMo7kP&} zy(I3xztVylV?=a+=$fXHdr&m~I~fY}zFes_&mUs5p<8p`Uu@0K(FJYI++5hOe{;l1 z0>qv3;5DbC&OezFT3v+)cFQ9B)1sSZgzi1I)-Nef0WLv}VH->K!I~W*W*F_9ZLA#i>B&9=!%kCzM2Tzdqz0w`>u`dt=%i!m8w)UdCtd&6!TY%?E+ zn5h;j_HX0?+XiZeo)sNrb78RRFMBdB(D8FMtwYh$ipuqO(Pu;zU+Pus6IAhTuUrkS=tX}aJmg;b#1;ca1lsq-=+mJTKXPmb zq-WMi4leoSlgIVVhgOe012aGIYLIevKSfN4?7|5gaHtx5czmD0N*?in~=4rjNkjjNq zS35Y1vod1DT~o-L11!`b?uEZ@Nt|@zh+VgvxX*qVM_ccg<2N!J0>5JOu1RgZ0z9#| zgHYVqgdb^R*#{ZNCIUBUt84jZLbk4$S>Pk(bu!O2ua~73nj|eeER# zA@IjG&{az6qdNN=g(H5YN}XdC{`BUfvmzUcbFE4IvCpta_qiO?9gcLI@#|HL2I?2v z(h7tS08n%%+{ynzf>x<|>dh`4oj86x(b#~)KXq#?iQYp=4Jreh6qK>7P z&!aHe&g*5rX{6xB?@1QTX38U|98xC1E=ILWy8?>{-C&;*4Rz05H5FOyz%4U~R(9W5 zT4d$<4KF2&R>&23-;e9|1}tf;hx1xH6-QYt^jO7;y{PJASerdB^(enL3A}BJO10j| zlHV69&4%KMhN6Awwc>h{g*;;$i0?^uc7Z+keE};Ww2O##F6kR8o8=kYTw^nFNA+XF zx`xA%wmr^%rh4;6{C(4-8dWq%$I8Me%npRTXYp2N18~MVzjuS<$2r zx7$G0><`|_mf!|kUH+&Ek;LS{q-)^z1I@7Aj?x%9vWZ?s0AsxAGhP8i#riBdF5nSR zTt7LzSYjtji&qmzvi!Q)<^>0nhO}$go+h@lF)F$rzunZiuL&I3l6=_kSETK66NS@Y zKN<#NLsM%!k-<~YS7-3EkkKvw6H7Ri7kQ~5uB%MeAFvbKEeo~2MQ*H$K;<{9mTLy!aw3!p<@VB6AtJ|g+fd(Yq{cDP^Vb4Z;!XP_g zyuEreRAqVcUHc!&j}J%_?qhS-lQ^r=N~qE{*ufW$czZE3R|-O_S3ULOCu;izACfon ztIRZhDNaw?ZK($l&F}RJ(7p+U@4vF@PQ(U3j&a+|rrk#k-FSJc*uK7Ff=v)ikZ4Wg zhQA;bhXo5EgZUKq@oo4QC0oQxCM#Bz2l_<;V!wYoRDldNfz+xQ?AL9J3s!kn1IhUJ z6k2sNTGkV&QsI-dH(oO!N(_-&0|PD&VX7~ao3oc6Ojx7vnzelO?CSw+m>Td%EaW}e z2{gEzG_g^3+su@hm?IqvV4m^Z5J>$c_Q|<`b;1RCw@cXYRxlmiwo)8H%shhnCyVl8 z*5i&W{-C?8uqm{^?47X-#gR^<9qlJP&+!7YPQxn(v>nrE_l^CV%Y?5<0h==aA4Po8sVfjDUE8Bd;xI=48yn}V4S`=6v4j|7*2(-I)b?-kM=82J_?l>YA7I6i#I|1z2s67{>msj!dU1 z|1bG7N%2YG!ZTP>T(3Epu(x%|{oJ_1w>N9eG#$1g1GW)Y58F5C2eLW}^k#GuVPHyF zPo9RBjKd{KtoOvxPSrc)(>(fIyuj-^J!inVs9_$UEr zAuHHF3kSpu|IY#OxB{VOH1AdB}@*b0=w< znsPMxKbITgtVmc7nBH59g=NV9{2-%Xoli&N;9|3PBmZ}r?4G21u-@IkK`uC!2gtgG z$D8%5$`%JXxxOf;6Q_n0AFk9%cBT4?#Tms7d@*p2a(?_h6!_RFH4C?LwplIA2A9XhSBS&Rq*>fc< z{|T4{;Gj|4VE^-@K6wbEY-xq;smAn<$_3U7e`0lmaq#Z=KNk$tt?xBRtJ{djKF+l- z{nH~42e?y!l9ejD9WkwtY4qqs;mYM~_0}l5f-JNo%00X1iKXuJc`d z&$=nqS8Tjhztfj(S_0nOlweQbNbo!EuxlIq+_UpX#0=IQP1>!$<>-&As1ToI;mp>< z8O0LIbM3W0KtG@ZOdL~9fmV$s4(N(4(N^0;A~bkhe4o+5PR!|bie?O z#?}2P4fqwC_1k0lofKwqUEozSbt$(VqCn;FXnF%D! z1bgo}A{g**uLTU>8d=^82EZ5`n}kn6{dC+R9%q@f5U)iYh!(mUjCh;&$uYyRu{NV^ z``?BUQ<D+D8TYxutMA_ITs=@j>z$%R zKv+F_HS*qH5X&D;Cus>mCUmP>qARNEZQ15)N5lMdBn4iIS+^T3e;mPPQ;EOg6ix_& zO0og)r6;8z@h;QGajfh_Of#+#xhZ9ioEs}kY)h8@itbi>q!)r>azsc@^R(peb-nkS zj(tZA@|7A#g~ikVWXG`vNCxnURzYL2o;y7v1_oI3*uC&IS=vvV`H(68T&ElCb%W3h zr_>$QK{@(&%vL>wvf#~_RXxCPGmvGit;CvyzRXBEjH5qm{+{qvLHz<8l(bl$F^J1p z7R0nAxYQ=j9xrc@T={xC?v|90KcS<*?(bYM+wumwJWPj z|EYxkKS!B=HE4l!SoUl^dY1VxU2W#N4(SLmF1R>Hyl&LvF*jkf6!V&I;ckzdl-_5b z9_kA&pR}oyZh+bNUZE&%&{TkB?R@4{Lr^pu{1=^n{wwB1ScTnIw!5RvqlXV)Ouj^I zvZWm+H04uXSy~@Ue~@$oQ-w-<1!QtA$udD)z0&+%`IS#zhP6+oOF`$!Ub*~s|9Ux@ zvbV<6<2w7m;tTj^SQF4%{zpX}`Q4e%iALH~brvE$Bl0u%;|JL{hr7`2BhnzfqUQez z&>{P0Xe9rE$7EYL7AFCjP;A6UsP&-*ACV5+m%`*aE9+gTUjlmnAs%pR(7(rdDRz6h z49O!i#bj>}etAkNoe{#yP5Qs3Q~l$qWRMs?IWibuTuGzfeBcoBo}&K$Ag4C8Iy~UKEhq?YHUk4YLE+MI2wLYIdN_YMKHSDD_gb@EJ?*kn z17Y7{O4++J(MAe+P>1ijJuSQyw`#9iKjd_+ z@1!C9B}Ye6VvCE>jhlN>n7MLL#`g;AA?3bHF0HE=vAYFDp4Hz9+Jx~JP;nHSiq!*+ zWtYSI=W~o8ZtCKIY7u5Se?MiobHClbF>iE<+#a^{ZJmv>-(Fa8 zDSk&SFOi{_Bd(ucXX;bUKJ-0L05(RE)kv%>(TwKf6#@i5|Egw95s23c^O!Y@<%5(S z^*`ymp*l09Ouh><^_HaMovzKGV~VW7+3fx9Wyv|2JS?!7S%VFUoRN%_MV&x;Nvp-2 z1VVi{R{E{O%Fo%9Ql-CA*!+->puH6J_;ead9i^lkLjXd4aw-`^&4ko@!YOKcml#ZneQ4wixr4hPNiT z37ZZA4g1MYg(lLx83Qr_+6S^WG3e`l7p(|Dyat6wA(YhGU2+J@Rk!dmf|dM2<& z9aKZvB$`MW}{ik0)RmmrtB&W_ElCbszy2~(wwH{JzVn9izR^&`5hB9Oz(_eTIg8#JiHSq14p|Wy z+`mY$+|N{YU^@B5em00L^_0t1hzydwU%Z$ZaSQzTEE@UiuUPiKoqkqNGC|+ zXo_`9H54@rGWblOE{BHDYmn|cRbc;Fs%PYY9cUg-5iYZay@i#(vM$DQR_A^n=Vol1 zOrf$4TLvA4e8kD`W&ZBV{z5;OrL*~8%OrjO&3JNc8hTk&(7HO}WHppPo8Kzbo5EJL zO!_Ic95F*XL_!h=<0_7!OKa56rO0_OrIeuPmh+=Vw{$%Pv$THC~H@g4Qm zNseU!HmHa_XpXk9^=mU+3X3Pg3d7vHlbU2CPRVHNF~H$=0rDk=owPS%2Z^{S{WRUK z|C3XFk0)d{tDu*_gY$iwsp8$1vX=pHIkh?CG1d-{Wq`q^Q1OIqpk}n($Ik*MRDjS- zkrQsF%^)L{+x96(k3OeR35Bg2%veOi-h4W@bz|=2NlqaMIu(>q7JmKy>fGWzZqg5_ zRZ-aB)}J{|3h#p$Sy7E>yf&L?17h?mQ4v7BLI3^N4kzKakNk;r4cM60jN+3}#K|_Q z(N{cyH$QPEGwf7}WQB%{AL4W7>OBul$+Fs)w*EM1ay96Wii@s7j|ao2Ypf~_3QzV< z`EP)E(Fn^w`}?V!%viRlpHGBNED5RfH;8$kUMt#J&#hws;rY%=>;aaZc$9KcSl73& z@bPm6!MzM`nmq_#ZhOyU3dJf9zORtcSNkmHzRDQ?(bDZ?DUEa9u+&Pn)So&0&zGWF zgqJIQAt#?I#7{YHxl0PQtWVh^cGQ@R#wkK}{W*{uw@0!e%9`c z!X%OMFHwzxIXGg3M>>kqIx3=?S+OF8B8*p$NZqBh8;rPr-`(wz;R~wTU6NS&mrx4p zpBaI1_m`i#WPw4^N|}rSJEEb*P+|Ruan87@iIsQ7d`VQ4stW z5sD-J(X(^1PrlUo5;Q(cq)s{w94ID0mMrt@@DD_=RL;4{l955_$g}vkyYKt#8D2lI z1k0;s)C?I;!Qmv>djPQ-DM*0LYHReQPIAi@GeH_s%a8%pa&U`%nZ+$6cfVMm$gWXW zx{+Vf^JdT9b26~3NtR*lch2MQ6i@C5qZ%dhHrw`XyOH1nNxZ!E=?`!q%(k-twesXbYAZwxyrwd)`b*d@Dt{&!;qxayot^aI4WZkHW% zr;QR8>C|Oo8O2DXS*v>dg7-rb+CHI};OLY$&O(wX-c7PH42*047xn0WN@5l`*R2)$ zHq~Litzgbc^BB_6#VU}STqzjys0^p}Zbx`Sox@&Gk)-V`if2c17kbu9&R@AW{w@wy zIKF$cvejo-aR^BJ*E~BL^<$TU1f=E8v^{&-;8xMPzYF`_?E@X}+0Ux`X0ge?eda_` zPh?JjwPTk1*?{n|_Q0+IYELx#_fawMs7E!$L!tE>s2@?$cioe+8`@xt{KAC?^R7q5 zB12Y_H%hTaue+l&DQx&>HjEw8yk>~0OuJ|3gH{6$WeflIc! z_Z+IJ>;jHnk&Ea9_`>&KT#&xBg;vmTRWZn@~-8_)@GIOx-bICWgou4z* z-pbQDp3#%D1_mLU*Bql;8=`_W6jdMzN3C1Rb#dLBTyGX>gB7Ozjh32ZP>-sIW17le z9#rbg3OhD%{k&|p1y$9Vo1G%V|6q2*d1nk2WY7fIRNWKgW(6LizzGHG7PxD*UQKJk zJ7w5v7JY=O&T$t9zPokWK~lE5s|@O9%1^Y9Np)^M39on|@j`+#`fsSkKdY(z+i^1$ zs?v32E}Zd|;cayoZ@_LT9!r1pNS_iq675C2gDtu&%L>GIW4+VRd#3X|@N^qvZ@Itk zU3Dk@tn80B6!0&`#y{0f`ZOJCoQw0HA#r#qBBQ^yE_$hZuIcuE!3QtH_@@0}b!=Dd zsFt5h$7I*mST!)&8W9lja1su`BInq zrPmK>IjXPBU2NIz+O$cM+|N;?&Z66iR6XC0wE&L$IFNg<(XXoHanN=9Unf~{>Ge6vaG_W29q*$N=%x|v%z5_9_vo4msNbs5t59*}X< z{$i`Z=j0hco$Ig(?(vo;8@V+pOU%au!EdG2LBxg!VB8Lq(UJ6`S=~Pyup3tQ9Pehf zCEw)YGN?r8o*EIZfJKifz6ye}H-3uxhD6UIBiOGN)VQ^&escsU-)r(#Br~Wq3BEub zv|791C+4301rfl(^DGE9rvyzoRK(<+cb^XWSoQ^XPpnCCPAzRj6Lz@LUy!4{CIGOe zd1>PeWam6FYO%rydaozYhaZ?vmP<}-0s&nM0;od9w~c?EfP=0LG5T=g-%XPO0M=u`+LMNhz1+LhyC*gYfHdgnl}J6{@llBMb|f zk+s>U1UcZE?o@!TdKJE_5`gb9?v_9HUnhGCn0aUL-NuM_$~QV@+<(!YeWXi*W| z{{S5pAlAIv)gX7SyFz#rS@rXqY%3e$le(+`-tT>Ae(B+zsdfpX%hDitmVZGm>FIGL zzJ4kL6AaVFXXj-91<`~qbiESRAWto9*)TdsR*eYBzKZY2tN(Rncw=2JY{#0LbQs;1 zZUR+cej?Yr!~!ZCyYOCvJR9?tse|E|_4~n@l=wOIn(Ob*WbMoF0+JiB z3(y#1?7VG?MuJ$MSMUXO+@M;CK?ck$KHOUvOt^x4KLEzNTs zjgW#TvF)Xb;|nd>P|97-Mz3iIG(#`*+R^j2nbw_nE`USV)NFjo528 zIq?keQovo7pr!RDW@<#li6-lsHHfNxNY(4U@} zppA?8HIb)6vI`?6DB)B?9vLampEX;A0q-VZ!Q7U6WfAv=lOT^kpe4Eu{r!KMc8f#C1e1%i7u8X_J0Zr=DXueFI%f z+J81p*?PEm0!2L*F=Obij*NE>dpw7O_gBC9+Z+hft_swQR)^%D*df*@gnCrTm)N`F zSvyvaOmpydy+&)Vcqvj@H}Fm;Jp7K>DI0oPi-n1A-oCTGce4I35kjw3UnsM3c`Yi9 z%{%tdlF+vkg7!5URkhzO$G-q)#ds$TNzQ+YqY?d-m7+^GOP{YL? zQ4@tpy8f$Ct^b(fLihi(K-H+<`EXC|(Pa0;r4IJDGNa z&&PXQA~nxjq~Shy8?4@=cd?SIHQE*)_k?Aw)FcziF6;i+IB9VvAmZ!FXQ z66`)E6x0|#YNj=cP9*&<4v%%0_oRO#do7YRfBoCL?O%;3CmwHnO-RzD`c77 zYj2(r4R(uJVf@mDU2CC{(9HOe?(+M53YC7?n#oLgXzIP5=X6vR*1^Qf(91MA{CZB# z%s?M%8n%&7q{FAaLjIbGRD85On;tV0yO4G6KzAU8a^N6k2?!gngsqJ?h_=7-y8A-i zVViPlIJTPlGK)1pjedXQ0yL@=D%r3#^qW^zwZ9z7N7ikL!t>ZWN+b`%8=?32f#WR` zWNx`&6%pjtf6`&cyz*jdPWL)qIk%_T>G|&AXLdkCO71iEcfne;iq6M=eXG;z#=e*Z z4NMlefnLuhVFen~Qa#J53k1vRNSc}8oNPpYwla3BrB9GDW$L--JxhQ#Ra(}vxtD2N zT&5r8VrjJQA@KYKVO*a&@$-lP@DwO%p2|N5h%iKC5pg{0xBhozqJ9~4WdXIXJ#v!dlJZf__*ddUXB)9g zj*VB|bD?d0Z}vlw>G#rjA{lz^fl0|!*poXPE$J4$W#A|ttgv-(l-g_c?3$%LK8T|| zl)3aQt zUWTWhlNgMDbG{})0>Dv8r3k!Htb@9O$ha^Vdq=Jo~D*ZsLF zQS$9SbS_B&U!qWyQ1AHXT_K$i9}@SZvRk+Krdh*fo&F?Fwd2WVIr**}wGZ4tk7jH+;NO`43*xGhE!Kc!Zejz8S3{`hUZ-g+ujO60u;|wI zjJjej@%0?!Zr(ULn8^1{JpOJiTuJ#9NFB33?;c9UEMH4LNvqK@(`8K?DqK}6i=&of zbk-S2x4y0x^y3vVk;;RY_*W0>l=|)@{&tglGcHUkF zi4;7zJ{OT2%$Ix>kI$tep*A$K%>%bS)|`Fed&jaMNZ_n{m^S}OqL5iuluFd2Oy9j< z%+^he;@zaZY2%m2_XP-OsDr*5v$j_EnjymsCN?Q1z*q+sHJi5ZXi`o zSH6na>Es8mEhM#)6U!B*1DIaVOlUVQd?t|3zQ5OYzvik-GpDb_7_W;o{j9}tCF~GL znJD$<@bGl(=!<$kY#wfmCaz=SiSY0El^|sLdNZiZ1 zjrJ>mmdS_@o1CsU^-B~dd)_Wg($;76gTRTq?UxzH%^sDzXt#oHNpF-&z>E0?77qEM%~Wj_DeqYw;C>zJMrlI(p?1Vf`$&YWMc9tecmJ z6P-y~`-}{UROKNQHgWZsN=_32#K$a1@hXq<*UXwtjzC9zK6L`)m;qU4dpb`Om>C@O3cvV~28T$#)pf4g_;e=Uxb zHM8VkGdqSOf8@Cr&T<5%o|hKf{(b)EnR#d&cU>jMFbU%csmFhIdntk!eoa4NlhMdO zl0=yDs5mMvVES|e%@+ww*+v&W>~SXk4LxEdO9lDIjGjoszR8N=iSw0!a(@B24Pcg~ zvh=q_*}obg=igrssev-?R^jI&NtfNCr%M;z^T(&cYH!yC$luDTr&_*k69IO!pK-L! z_I@s|<(B6}4LKLG_1f%|Xf4NVxDIqh;m0D5bQv-c^8l-fDsChZ@eQ~yrq;mYJ$?r18FK3+|b6~k8;GIT{>B~o8#+bI+UILkf6 zTOyxU{6l)gsbVrxRaOR{m^A*F_ASC0)mr@p?8e0KxLH@ta9e3Hr}<=*ewtWo5B(3v z=&=Wg(O~%SsS)eu_#Ame2iLJG#Sw=Ns5@oO@!ckleXv`L&|?L(wk;fOOTlfI&cgaI zcHne8TDw`{0>Gz>i$MF;C=ezq6vP?n()otDLV742UpJ?%k*4SLcHYSlu*`mU;9j?W zrE+nD%YP|vDP$;qyB?3I2N+H)$ET%-q>$2W&%91Tg6DtwsR7hqTtK z=Ca)D$2@+9H2@d`l_#g!mrKW{bGu+ZxIc%YiAL_X^eiG8E5Oq=MZ#>pF8O4?avE98 z)pvVa{V=1Wc3daPGnd8Ra06U=8_ulRnhIxijkwbau5RVHeE|k$ zCt8B}E0A#eosqcGa@rI?+2ay<%g3Jmjir;l`(Af(W)ZBeJH(>`Yk+@3-+E~%P?fu( z4^}yGs1QT1v6{ZEgEboPD|(RPB6nuZ?J0DB1wDtPQG$5CSf4woYy^m6V$0-Jx8LBq z_C|YGahaj_&oOmP-ffp!I|pz<`Zgl3aRNBo0TNCvZAgVG|2F|Ic%LZ%D07e?PJoI; zZH}c6mwPMyWaUDK>|zBIH)+Xfx^In0#nEEvhL0G2Y@D$+>?vX?mG;D@+cE{JgGATf zVo~CoBL2eQvg!BzaF4r9OJp%sh&f|5dB|%DM5$QFP;jEfi?wAs^qOwnOsdEgL)j>j zBSF|K0K7`M+KywNE-4x{_5aD)qhDx4;BExGae8yOSp5RDJz*ySbSY>n zNcxS_wyj&2g)XW1@J-f=i5&TzbOr(qolRF3P->clm*XIpjU>n=;>;1p{Ex%u;>urU zJpESw^mTVt&u$?f=ya#y*`Z!tQ|1Gxtv1jizaY`ZS1$z(><-J;BEjcsfJW@~;7ux* z;?e6hQf^KQs8gMC?YC}!)d9KvU@(iFJ`s@EE`Ygg-!m$_Q5ys&)+p7%tL5^L`#zw0 z<^yuI=fOkCdx+ai z<#x4dajPvil5QzQ=E+BKqUf4umSz6Bhl*-<3e{srt5#qGt{%-ufz z3-V1_b~X^-FsH-7R07ZW(+6#F!L2U?5v7f91m7hMqOfJv=3r$-d%wPI;!h7ow_q4p z%(TXXCK~%O;cR>kq4w@~cf;t_s4Q@cPn z46>Sm4=nt7-s;&>zqB@V&Q{;hBa-&1!}h7FGzqj|O!LTBZmQy#WTZ5)G`QJm;yL-6mha^QzkPfBlJIcnP;DQC{*jYB+?D2w)dNz=wG;Be|fYC?lfZU z`PUc=a&oRMr=wvtI+AakwQo}>Mf)5EeSZ^6heouoXAf*BktajhcR*4UsnxgTk3r5= zYQ9S(Nc^D1W8X4V34d|&2gKYHVU2TDYgg9KGnp6dPJ^LV`K0{xanRTxL2%|U2OUX~ zOV(XZ8OgI;U+%Q6>FTvS;^E1!Ea;1PW0MT#gbZBGEKfkhnY5AoC+JzxY}%VTh63W( zzyIJS?d6x{v)-=?EiZc?)Ar$Gi=;{T*ze%**Gg@vieWor+@w{gFwemdoe=V3bwNGb zT0tApFh6=J&rOLg4=~UzZ{dRVqej2qxYFLRvY}p%?<{EGoX%>JXma{vJdsOq9$sMY zDk*z<4l<5j)2W+rTz#-9Y5vjFLtL8pZP8hSJiQZVH)}z*`&vlM0xpIhFqw_+K}#vh zx4Azb$MlFS(SfgELgVE9Q7??5W$XZ7H{=UE)b;ZdJ(WZXM<;QR6lv@P8Z;w5JX0$7 zVVe4aC+#cX2azBU>S<{A5N|Cp2qZ22=ZJE8iJEe3ks!khs(Re><|S!zro&=R=Ny7U z{VwX9eguWrSFDNB7(DgeU>96PML3L~W9*2-*`l%e)QUSJjdtsXS-EdnQzMWQDlb?miA4MN2M zkG8ccT`^`J*qCsqd9OgNTQj&3B0s(IcLVyt@`4p4+LKIyQ^ciEsPH;?G0!R-SF)b&0{OL?rUmGF zJ|(bju@yy~@`47s06xfm4e}`U z8#2hH=#_JBk@u`{D|WxA?zL(q2pH}?XA4VlmeF<@*d#wG5zbAqUKgr#={QobB$W*7 z^Cs1k|2^i?&Q|**$8j8ap*cztz8?hk72K_q|0gi71iRbS>v3GY8l;wg%;lN{M}H&w zF66?o;cK-fA5Wq@ics|L9YAccu3zsro$xM6mv;1*Exy^DhqlsOG|kJXF@}`$e<1yg zi^}!n!a_nxLQ#7l(Z^Tl6;I>i30cwccBkdjhSzHZ)p0zta&RO0L&T02oWdrJvEN<- zFiaGDg)H+OW^CP)D6j8;wef&KljLkfS&Tb*HTYFlz7DqgJFwzYu0|+Dsz}3UMH!hm zaKLrsPeL=17Dr+IWY=bhv?8FtFu>~p01c+#>EK0J7UV#KCI!ChJnyxCHF1X;-e1Vo z2xpc>%s#x`lf@3=$f2i6%LkDSZTr%gLcF&9DWGzJ*qk)irwmk`&6-Ar7)mcQn&?S* zZT;GD99Wqaa`}4wx9Yw#tf{SAHwYLDMNyS=U8Km z_Z{yr;0$p%@CyQ^z5$R@ynYK=tMzd5n8f_}$kh~q?B<<$4w)O%KxwY`pa!d>#uX}0 z_+?(RJkF?s6+M>TTaZmdNMMBaVJX@~=N6v8yX?V>0sa0QH#92^G;3c?D%YpkT?@xc zxeEHWfxhb|h3_8pACE5ow?w|?tx1(}+BmIFsYN%%? z(i8ZtZZ0}qP7>9veWI|N;Qh@1laR~&IvP5I3a&(VS9eU&6$c+c`To?W|Cg+d`=iMvKLS;V)Kv7 zUf}e7F7(OL^yscv1?0(9BwKThD5AWcrP^Zb0zaZ^0X|EoQ+H#yfgrG5)tTu#fj2DMZazsTCCJ&chpJynz2fi%b zahM$=1u}&vGapr>wOItJ&{h$x1Rlj$vNz%`mlzFvSK@vsZh49JaGK4u*CA%|zJ3jN z!$VS)pr=PYtZa_FJa`V4Vkg@WIXoh*Wb48i_cOzd&R4;3O%3?U7f2JyosJHOdtm?- zx#6_;vfu_u^eD|6t3k(asGj>LK~g7rq z)rp;w2~#8B7g(a%9kX9_UWu{1Aa5u>j-Jyo^)Vmj3%hX;ozQyz`1=!KLI1T2R!!wN-(qqqfw51c1p5l&_Mj^%H{Q*n5w$TJL#@_u0R z?RjPMwN0J+?6Q@^w)f4y2*g!ORzSGuixTwCIC>?O^%zhxnZde+D{epx_yB$$iZhE; z?X+7%=fx0nEXl#Rw+ESuuW~n{f?dS5p{i^#l)Fc*B=X$%)|Ibv5;kcehoBYvE?DM$iXyS&YoQDFu|GWU0es7Fo`;+?&HMbYoM6ZHXW6DE z5SH4vfmqYDikyVSP0gopmq45uKAhOIqamqi98u~O&xniL4FahqgLvR*W`S^JX^q2` z5f+WEf>Z9ARFFfblYrsJJ;j!@FrlqZBNiBufMKv~(K>nL6aC{yWu+>VYwp2e6ptfO z*~FH>gCEkIuWF3~lEcy_9z0@Yn>|)QIj-2ozt{oXXEbl<1M#Fo<+@3U)G7-zsrXR3 zze2zi@soC*9mq9!Xx~K1*x7=%J)4j_^5#uh?FT{*G2=9l^!yjj1VtQ}tGO2Vn0@wi zlr)Y@%sm+{sEcKk2w~XK?1j%Q+qd{^BPbQ9-T8h`;BE9Ez>kljm0#awYkbwaTr*r= zHN$@CT4)n2O65u#F};6cctNk`5&JL5O`ynDWZ%?zVlZy=eJY}?A+r9YQEOitc09as z-YgKM)>Tc8nf^>JB}!DvJi0HoC)5EV-@BnWt!KOrBa;U!FUhd~z-;8Bh-ivbtOAM- z;tPPKMS~cy3!ABuWHWK~hGrHaeyO{0#e}$mdV*QBy%waWt{oDRiRO)!*B1KhX0`09tYwyk^!(hd$RAzaW!q z%;W(?dz8+HIh&o^&s=$FsS^7^6sYT*t%pA8E%%?S-yi%HJUS7F`g-m5)*eP{xULk; zN)TzaRn8ov30_hbe9wywKNn7GIZF89<%ysGa5|OQ2;{ZE>J;AIhcX%v6dO3KP8l_W z0a*obWdDUfmD$)VhI)?f7bMko4?YJ1P|mx1sXo^acSQwor`!jJYl8f=Baik{<&@4g zmXtuEm4L387OebKZ9;dlRXLL)D38T&XhRDpe7r{-SpyB5j>9KY^ z+!rzR1pyeEShjg~R9A|t2^fMxP4Hl8IRP!kF}n^xNu>znkyb}=DS4-7Z6l(afqtm8 zAOBSQc)$Ac_No-m+W7u0yzD5}!*^dSJF`_{dkC5_c!i=O)pbw#Z!m)CLy!=Xe~sZd zcKoR|E!rT`%_b(g#)kbh;$u8_t@*eq5E}H(C68cBJorK^3^p6_8`&D9Vx={`vHAR? z=*`HIV^f((psc+D#cQUR(>wT7!q&RX*{OHUfz9WBb#ANh;_aYZ5n6PA9d`uerxV%O zl$*bmKnjP4?!ly3&yq3N^O18lPdoP5x23v}AXmBSL~?1Qajs`&n0$lTD5$0iBDp;} z$=tj@*Wo)u`+L9K2&lVwm9C)Z(w{yL?b*oZ!CANDh*}U^U%vX09~-b^5DIM0w89ng z6{x5YS@9rkK%E5;iOjh`p382abm(KwFq9Cs4I;r=cqoSJ@C)DU=Bpq}VHHZ&3&LwX z>zO15qNY@UWMc6Vu(CINoQQR?OHRzHhq&k*fme&MllGH9YFZr?4zBB!tpz#Hd5P&k zZD=(dBz5SwEro+D_tLac4Rw`hDPW~PZRC? z-9{gfMuPZ;7;8{Aqv;K3_NIsExA*+4#c6&F{(DH|fBo&Ma|Q6tWx4Ztq5aikgv;#L z0$29#XoJ=|uK(ub{>`;uf08vrGmnD0PF0W>+MNSl@j+u>$S*Xs3A0ZjliEC<)M1t+ z`%ea*8{lp*oUdmNspyr{$;~}9JRE9?p11%b7GHcjruxHO_U)UOT3!&V|H1;%-;;p< zFXfE?3gdctG%e0YZKb{ce!#psIE=R83^id^Vw55>c;Yt6TytD??HfH$@j-v-F6Yi& zS>sRT%V*v*r~U8?DO*e09U+HtLJm#7QE_hE3q7dM>PSGw zwT!*H|HAL=^O1hvmpxzXML> zy4N2{pz!~bQQv>}^Mvf1#nhpZ;+POIp3~3n_`A{jOGFxSo~ZMpzU%wIxb=pJ*lqTg z@H;D@#buWk_i!EBNmz@R&3{i?FHgBPI2x_$1^{0F%fGikG-c`U?k)JG+@iWg85weO zj@;4hZ9)%!u=)2Wtg`7q=kMY3DZ(CnNC_~ziGhTZ?ia-yhja`RU+B3la z^a-F!{P@o#k^c0xG=nxU<~q>rU3zt+Q{-Z4ee=?{IZ9$5{1D~AK~&iY(QB}@K4ofJ zqBlYxd^v3c;-%zS>VEDTXoGAZ2q9OosYj`2iCG`8A;E4&7Kh)kyxYui(_&iilE_RR zS}sG-W@m~6#;^41x;?8e2@LggAV(eGZ@)TP5v?`c%~!!-~3 z#na&3GSh7gL&Kb?ZCQ2;fCZcm7d-tJ6xwQ400{#P-w`nIz%rt7C9KCZG z5PixAuB=wv+89ShPN?$5=Pr!43zeXNQhiiqWs3ifXM%dmV{>w*r9ovnd_%?zwlIRh zc-4=83Ech$k#3yUrW^3#C%GQ4i^*op$$S&i4JlxRwtSpCKWw00s<+;)dgaMUruT_C zZt81?uEkA1hfQ)WQCVDY1P~!BxS-~F{V!vP2CZU64n-S;o^)ecvc~fgu7-)KLeJ5 zFqYkdoWwMvODnOW>-|yI#CH;3=j@nUhtP~WG~$rj*6IK{xVD0aWwW zQBI?vNIXh*c5G^GOzE={84o&&nceAIbq#RZ`c{{fA|fcTSRNOM!0HKUwQ>$&!X)ye zl&9BueDY)jqy6=*{e`(DXS_LRd%kYFBsxL-Z=q{ zs<_}-J4m(3s;*g9;NRB3OItp-@DUO7P7bW9`Q0aAtdHiDOcD*|C3Nv#MRle~jMCr~ zmvZrf5YzA2+<$FF(;q`G|IC>35x*dvAeuRyZL#M*0x}BLv=aD=m5;*^zn)5=*tutkLm>f)?Xwa*Q`tS>X{xaB}eodCEXmnb@;@Cm1b|wuMRHQds1f1Q?ZM4%v9dA z5RNMfw57^XJdD~u_`l32%z-3?$1lkG+_impV>R4S+$dS^OX^scgLOINWj!AKS$`@j zoYNrqGl*p$6;netKX=;8(TX+`sKhpRQzkYYSZ}_NBc;WGc4{rww(La z%7bo=Myd6Y&m5sBjom!-@x^g6=lY(IUQJ&Gzthr>QP99s%VNnl|2vh7AX)P|f`{Z7 zRpF5NF;@62^v%U_O7sngXK08XyiNTqYW_U1Rm&8Oe2REzs{z3~eM;TE6nFE!A{~Bj z3`^miBDp?Rg|8on?Qp5!O%&F&IPD5vm5ILXKBO|wPitm0MCI1X)Z%`fZrXy>3ro(T zqh0VbOqwnifJaxVqxr|x&P1m=NEa~>KIVxI$I+~W00u+gOgbsKi6^1#br6-1dhk2J zjEQ&XLavbl!xeB)-OK%aba9dRqUIw-CK2gb1Mh&$K{nU zc92<$>~ub(iw2tuIYYDL@pOyy=}@YipfCrSR|{J4R_58Yk_pT#;t~(!`JPrLHy2Sd zKi_1vCeY}z3@l}K!AdYEqnNK=D3@I8Y3-YhbT*>1EI-wDo^jCsRu8@lqC%`+n=%bt zZQ?9LmrflR?iuNbXUbS(HNv?U(#Q9bWmUONle2tEy*h>OfcO_4dcqTStX#ia%r$bm zBm;+_#LL@L>*1;U^kx3vxyFB)7)$|SS~A=2blO`QN}GPZ*PYejOCSfs+6n?|?vpfq z6g5>IQf5cJ$V^)qcZ%T%k4aBk*fRz3msj6AgeYjC;td#o&qZU3hU1ck4#R{yd+Kkv z3=PDCgd_`_L)VD0}mvio1B1Ov6g!ov-Fo{3kN?;a|&3^7=Tqc|Q*iYzW0vQVr^ zmZ9eg2KK%I0`(k75vb$fT=u2{6%pfikhh;!VrMYrajW;)cG^}loot<4u8@~`*3#K` z^IxJ0b%4d6B=$sZq7L`S)fhL)ZSXI&2qFfCOMrsfE5qcr=;Bk^hha7mI-)P7Cf<#p zTu3c#V(6D#3JNk|((pph%#-_2Z8M(qkWX~%m^VAU^RU8&!Hs3LIqeE`XLXf6;)(3l zrh-I%n){I-9!X#Lk=Wn6Io@?(MLkkE`)gEs>=`SQTTOKORwp5I4l7;$*I1qi8L5#p ztJAM^$=*q!>sAx^Oo_4p1Qf60$ckk<7%ynznRu_Ex)GN7$k@!pM1?`Ddkt}9IqDuU zqxY^C|AE z3kW?kcJ41x?v_qpg7x!HK{=y7$jc2;&Hd$>@!CD%1sB`pI@!k+DrOT1XpT8mx2{XP z2QQuA87TDpgZA4Br$1ifJe3zA+U5)$NSyh6_Jvo(BIQE6{As35G)y>G`-AJaqJe?izTL^~b(cI%zvhv%_`HqT8!G-a8KqFBh+Aq4EtJPF!RG7kO^ z7FsRueqaEo8cb!%R(*R}jw|QZgr#rL)L8=|6Q;UO>#TbOJ3mV^QDZMOwek;gI)B-R z>p%ICWcrB}4rdC+d!Kw8I%@sm=6C~8MlJSvOdOxDmwdT>ucAIxZpLEqI}nMuYhJ*i zO6Pk16IzMHn~ZNbWQ4RgR*~4ORK5itcEB@aS98bH;5H1bal#>qDcsus%@YQUdgfzhyNdLsa@sucaQ;x#P(KP%TMH&e}yP-cOH)i}hUi>s%f3 zfIF(ZAsHh1+XQNh5y-uI5&v83H|K~ZW2fU6z__I$wP!v>B7~HIUOI_0R3-mvq zUaXQ~=K))Id8eN#d4{?>4g`I$Hs6T@=AL+|!~Uh*4kz$>0v{!K53 z2kxML8?Fv8H%>T~aqynsg1vxNy0L&bO^=QV8|54)fevGL0b?`%G$*Mkg-2U!O9H!d zeP3n85YxP~w%NCe*!y4?zQJWE7CPVGlprx? z=Z6jLm5BNs-h7T0!cRe8y!cSxCa;vy=|JBMaT>LVeg*DfWhWV{$tg2_T#o9=7uBlA zV?!UYgSYp8N~bGwNO9R~wzQ1ZXYD@T*bTqwX{8ddDNgg&WVI!khSQd!o>LO3F>v>t z7^+iMu_K4_)IjbUvAf^?b)4C-Lm@WccAN@r=#z{TEvJq=X-g%pYWiaZZ$t6rh09S9p87gO2vrdBxuMp3kx*ooUx{KG0)Fma zVg>(-eU^K9g|{M^H}Gb(p0vJhpr8yt?Z-aq#f4!e`W6m@T7ee7YFMT#8@J?uTRaW6 zKN6VQvBJ;hzTd9GGCcu~Mwk^oVj4BVX_a2R;2@lU7@^~|rkqbyDFsgaVPAvWA{OU1qeL{&&D@9laT*Utf^=M107BTp&>O$SSv+_C0qNXghSfSAPS^O4@CGW~*3j~5?N zQ*O%ns=7Lxiy^4_2xF1(Vbkww^7?m`c7lu{kGR-l(lRIK$T4$@-2OWTK^*#`R^bhw zwg&gHINQ+cMCG-y=OC>1=+WI1^6xrHGsy*SwEJeiTLK?_Fx8*05I6lZ?kY0k-ax;) zxz>7-=sv~eHqq70yHyunI(}T~BQXh;J2*$@?nVpGd+4P9r-%-{^_({p0^&%NC^0+SXTo~=XSRpy@%gQ4d<%Z`B z;ERY(P{dyNRLYd2Q~|mimIwG1OnNWPhrW4v=7Tyb0|5Ve37Ywc+9-3u??m7B-Fk7A zqG))oitKF3?e*Q$xAp0UZCk$n$9 zG!^X|AdHjOb<=t}EW;|{m-Irli2}}tUq6!ZO)fL_03%t>VXZoZ6ZSHWTFml9`<7Nf zN9);wJ16W$H;iF>ul8$XHmu6Y_xYVR_Dsdb1p%`;XbH^#L}be)Y$0;~Z$NhyV|uWO zhD#|IqZHq|d9u67F@P0V;iF1?oaJu!U(+O4y*OYnyb{glyw&Xq;H|O;s)Yu zmD1sE98XGiG+c1~2%v+%(mGHOp;{|MF{=Z;Tnf0<;2ddbZFh|%JRK1YKvUeG4O6)jE^Zio zY72mTc>&Ey#5HPoYoMOEy9kz-q=*j$2e!C`!*uA8ng`Y+ypl}T1b(d_mTG9+y#;} zI6DEiVu*NzZp+6e@0jvpK>RVj2u{*LhGOSlfjV`Qx!oyS4zSGaTpP$nS9vwA={>{y zU!+6LOUw=ek(Tew!7nkp*~90Z8uxkQZ)(f}R`|xbt33{IuQh6s4S-~g*X56abzDz0 z+38FRGc3#c=rBZ$8l2hPESAV*YRQ0g`^%q29&N#T4o9!H97qPpGPzAGEOED;e{`7L5cKx0a)1ToKIcm}{N{F$djp^7Y&1{5I`3%3lM_E?! zRmC$C?uNFiYpcqrpV<-}p#io=*GI9^bu3Z2IBi|iu7uQjc5as&3a01{U?SmjQkfb- zof0*8Sq{Aw1Ui>wC`iXTpX(Ek*B$=?GDu~=ar*c$h9MB6%kobNzzhPHOJQic)!pA) z#kk@m&5&>_qm==TgI~Rdyx)0y74l`~VNj~$tfvh%oH{O%ewC%y_Z*-V0a2pXk&q`u zh8=`)@=2Ou42h@pskK9^ugg!tB{McTlcQF`x+FEMO)9ezjHB*>%tW)_2OUuP^}20h zt`dIl#&1jdgO4rA*x{9=(`FDz<`|0jxQ~R$Evh4EOK#Je>Vmmi^l@u&o++e$~nW>Fq9p>OTf0;2LdpljK%$ePD>6UawGLC<}b;Hkecs?M+C z_(E6Q6Nv9bN`nG{2fV1qOrpJ!VDy!k5Nj?{%O|m~XOwm=#z+2AYl5}o)n01K}vdC)Mi){T-$s6$MLhknW0W?txf z4a?fsD`M7)yh~hvSzjRk!t iPB~T_t?eO`lfS#u5d|y$!4Lgg^KO4O#|Qm2`ab}@I22z1 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-20-Jensen Huang Presentation.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-20-Jensen Huang Presentation.jpg new file mode 100644 index 0000000000000000000000000000000000000000..2511021bdfa9dabb65ea735c776095287d478118 GIT binary patch literal 168019 zcmeFZWpGW{jDcnJr^tW@e5VW0Gx|VrFKB#LUbw#mtN`X8gULnyH$a zntJ`c?s>1jnWNhJbI(1cb)~!RI(x6Jzsr9&0T}Yqa?$`OC;$NJ{RQ~D3XlLGz`?=8 z!6LxJ!y_UhAfaHRq97xq5MX^o$0jBuB_SpR0?DXYXvrv;D1ksa9(pELHcn1XQd&MC zUUoqi4o>!eECPjyh=_uWf{%)d&rS{`Xa9GPzuf>#1h`q47icI7zz0kyXiTWT{Q%PU z&j}0l-v;1+A5b5lVPN6l5fG7(-zPL-06suLLw|sQhJ}TJd7th7ejEUU35!L}CIVI(OiG5Nq^6~36^$!dV4Ude@&do0@E-n9B+1lRO-P=DnJUYI*zPY`-fB5tG^bcH60O)^% z_5S{E!2U0|FyC=~fPsOAf&T|Cs1Kg+0gVX*OU?#|C8h>%?EH~}JqQ6?9Gq9zgGkAt zeu-n^GJ}Lm#koa&^$%$O4cUJmu;70R+5Z9T|HQQdK!JvO|MH+Q0V04)B)a^dFl;qZ z*ni1LFW>|;s;h}2OJS=*{r~ZQcMXyOjT*BAW@J=${kPqdwk^}bIDaDa8};jKOv2*F zx&mo0bFI0Z8ntVNE>$$a;$4-_F|Su_wAV*!Yksgx*lfY~%YOlGd}KMZfenI**|np8 z0UkV#xinoheE~3g&1q9r{CtKTNZE)FT&q2sUHreq6x@)f8N>_U8aE=bejJR~9eLhh zP}aAa_?U2fCi#B3+V2xwL8RsoJ5y{VqTF;_on9d32#H%C9^|Rp{EAaW66Vm))jLFg zDx;^M_FQ_&h#qvKfcxC#@JpqTJpKj;_!j`*Zxq&+6vr-q^RzTmpj@wcy$B0Et;ufP ztdH+DlbvLn7-4{5$D#^_^Xg?`;9v7|t$Zfwqqos8}T^L@~lx+M3 z&KbQ^^q^tXO!TYjEeJVTZ=R-elNa=55zhdc z91ICNEMtdG#XaNdA6tn!q~<4~jO0T<$kcU3jiVi}7gFilR(JVP>7Z^%BGE=6kFcF@ zfnB`vZ6%)B7|BD0I4boBF?zo8d(;=8y|t2CZ2 z+l5m`85RdO#mfCebLn^O=F*aO!6zQI(b*r{+@Zz4@>WiHf-XFS*u?Vu#(hX!-pDM8_tN3H9c!fgFg^XaK@nQ~e zoT{=#Fpa&YVKkO zRO=HZQ)~0*h(6hmR20^xKeYM&i@0Rm{-Bhxr%>2pinQU6VcuBjm|(F<(KjmiFz6G| z#T)67IId1^Lx|U7p@&J|A%LPw_cYg5J64;zy$u*WszkV=MYnifj<9B!!uyaH9x<`O zcv*ALwI)6iJFgBsZZl*}+=dg`E?ubd;hr$cn;P+VH zi8W+ryvEuR{Qsd zilsI6*{B2w`%slX{N+hEuAY6y7lYiJU4k+Htv8$jKkiGXj9*RqjgOgMH!(APml}0` zaa!erJT`=^)DNTgj>k@j3;8glv%E~>=o0Tj>A3FH;6Zc&IYBL=(6n&_emcQSX9-g#DGoMshE*a6 z)vktdd&KCx$K%A&x(9^QW6w`;v_I{acZuB zng|XI7);Pvxxq0Rp9nhSj}BT3*_W0;d%3Ip*&HE2#k~IsbXR8D3N?sF+a4srucFJP%k&8pBlK3qB1^hrZjiZ?;J@Oo+$ zE``U_r8@1Fafr)^zwFU;DXJTUCUHOHHp6WW%Taz&t$|14;;Ts&P*l{$mK8i;PVk)dcqVDI&pbUf4I99kw9K%>* zHdDvrx7OJBzNtfKj?)>Ful%7e&(pCMk4zUdtQ)+AASg!?oQqQ73wt&OA|KMq=- zX-MAQqC`U{ijcS~pVMuS)I)@a4#}J&6t=li39Eh)D(n38G0mIq&koKMhLk_YHY)G2 zo?^<9TDYD0=VTJwgELn}U;6t}VIA5%N>~-=3(+Ey(xLolIbR|tl4l+MRWpEF&qDn4(!W4Gvvpx!%tw^>W@iRKuv%AYhJZSa0gpT-7 z=3n6(3g)#7^hX6{$jwTl<)U2rA2-%>Bn;8LA|2e7L5t0^`^5(H4XCw}2ae)+wc(%B z=UA`bT&(n!-f<>Pw2QE*FwUdhhRaaxW$EysP6)BQjYEkAqeC~Dk2qoW8(&AHtwelx z6IYiL_3kcmE_D1>+e55VD%7x%$k)3D{<-!cK(D8z9g9``m&UL##GM!z?A)aS{F?lo zUUCHwB^tp284ZHk=a|__M(M6GNuR*ZE6uz&I0TN1nEmdKMBpJo`7s!Z5{E^KE@da1 z0I4_O?cvYoj>SYJd!vpt?+5rYXrvzt5g*c~E7NtqHZepnzBnW?czo^mJIR;w_59ThaOsbn;oCbs^+}D8&5eyW<=oP=;S~KAE1U z=VV!{fWA-x`;Ego!RKN3^F_A@QM?i;Nmo!Y=_0zH3rv>|FB|`f6U#l`{WZMTT(n&cYU2(@^*`G1pZ4qqf?ll&Qu`{x}cl>ITJdD+4Nd50*!X1kpQ% z7W{M}eJaK7pa-8NinXn?4_#F6b>QH`Cbg=OC+?OcS748t+-W5~k zc!9LRR4%7O5a*Zjvt(gqnG9}ERTrCy(Vh-w5>os%-pWT4r=C2Z+mw$dW*(>fr2_%m z`)_Q(>!9Bs(QTOUwn07-Sdh~hd+0CoN4xsy^caOhczYdPB6|A=(~7(wJ!7TOR*!IB zwiNF<{8Vi0n!2jbjM)EI5+CqlTlTz$Q)KB7W^FiWp}&A%e*wafqY)?53?eU+*WRq$ z?6EiJFOIp&MzU4Qe)X~hKPlr6zCavt9qq>;Y;@3Z2OD5HU7{oId1i`jVQ?-mqmfEj z8SP~!3?9E!-ROa(JP#!7uHYGavMDFrlE z5%tleF=QbgRE0Wh-7m1tntvG;-Lb|-#r)iQJ;E>%bmRGAOMPJe%@i0nF2hBY+Ja$F zlY@(KClt%Q8yn+7$T9CpDj&}N4RX^g;PMi3p-#1-k8q7%rJm(cJ7yj+o_)vF)$w_l zL})<26dJ7l%Z7(>GDn~v{(NM3=#tdYCS!yeh}?J zv-=(F8xJys3}4}xQ3r{D?@Cl$1!~ooVVpSTmn6+9T}M$?!Z&cJqfGoW=N5Ti=JuM= zJ3xeMd?s7^HEOFXm4(%DVq|8P88dPOtsFA!ls9b+^C23DtZ+PxTCnZreRwc#Cj(0N z2P^>ch}IGpJfy`|025$n11=Vb&5Ppucf`kVjSz#Q8{-3V}|&y zai+Xe!Y8gGTZ(`2a=dkD;?P&?_09M(eCZB8m?5)yn?4F3B!ur~lR&GBIHC~_4?dCe zwkKW_1D+W_=+BlthGu4l!xh>lc2~rq7ePxCuVMY#HL9LV(9SE7aVNf|J7=Tviy#ho zx+^ssv9*oCh2ArL5wB(0Oj77TiQZ6*;DhHaTdtan+c-GG{bhn5hadNLrz;x1m*2C} zN2)gw?Ci)2F?dv(2X&3a_Pg|dl^A}&evmvhXJZsz@6b+gePKRQ2Dw2yqM`DrZm|3q{9<5fH}dx~W(+=j&r+FW9Rrb>Ge{V*?qrN$HGkL*hlUqA)}b;I zKlrI}0!n9U6a4CCjtnf@eOma$)I;vtn)lpIl6~i7{ypU}@a4FovX+BaAQ^}=frs(X zhtZ!|&FUNd`O7aZW)@j$MhY+(axBDv2kQ-@awk=hhme}s!z6kS87m zX+J0{4ElG8Aa&Q6{i!NyOaa?)U%_kI@eV2e1w2&@MYF*|l;I6LD#&BrH5i7P6gJ>r zIW%^l3RRWiylK*vzp(q4LKn|u4+=?BoG!rtPY-!(r0}9|MAfvQQIDH;GuR#GO%g6k zUH>fQ>Bv85kH#Qs#h@2av=f7JTyhNr)0rs~(XVleabm1uiulG*w1_!I!xyEcP==j! z;!0MUVAtsF5|t;Y;sX(cj)Do;y&J0MokC(bj4X;+8OqzQ#SZohGmiAVy~bdAXo;1y zTU+t53eBNSYUYz#T`g=*sR8s;u5AI*e<%z^w*{)}>ajy%-Lr*4#dRsNs)avUHI zma)mELGhhJCwv`v2mPfLTG#g9w0&kLU}kX4RT;Ov8@S_ER(2g;FgR{ zv~23Z+RJ_A=L^?zdTKM%&mSfw0N4uzMj(8End@xhf#}3-X^hdE&`Gq4Ml&Xe-dloz zG`{NtR(#L!6pptsk(28HStRb1qkOEwbh)wpR!C5swRxms=q+XIT?JDh7Q*mdF`H^S zw20nE9B?++te(v-XEvGAH7NnUNLW=jqyXdBnF8U*n+v{ME_p@uXfoN$pMu}DBGj$V zWh2cc*q^j~TX&hSv(eQSi(8bw?aL1|UqNu@l-Bmjh_GIZQi!`tMb-w`1%miHW1Q(W z%Z~nN)(u%x;u(z(trT;6tP<27%N}&umxYHJ^V|O-UqL@mRw&bogaw>|)tQrBOFx)x z@MH1MYaHF!J$J^z@e|@q*Wiws>~CxazG<+X3w>?8vWS!KYv&Qn_7aib-(%O82g$DU zZJOfvth{Y=J@VL}B2H*px|k!|#Fn0OA+zFKvB#}QB5>ySkjol*xqKQzKify4XIkop)tvToBP;~ zJ6UmTw2WP1*fEzJ;BWF~Fd)$6Z^o))j%EI|6p#F~DOddzQ zU_2CDX~EV$)4B>*ZsynNoy%g1LbD6jmJMRzWEmQ>9(>NlgvTghR&@Azjq2du)l_(R zCzAY(j~tycHZ2)A^GF?O(y}DmKU{u<`^uSa1G} zSqaYfRFfn9HPC1$?8el8slj>1PZx?zQI~2wxqoz58Nf$_?$qHjmO5m5q`9P-!f2I$ zBbWvbB17YdOYT0YCXws@MSrd1NG`CSmY#ADrIQE1nev&KbBAef!OVaBqL3Nv@Bhl! zbQuwzaW{MHb)TQTbKyHbvRrzsuyujb7=Qo0pi5>=TqMhV46FM?IXGyesh}lpgE*3v z?MQXFT=jftv0=}*Yl2eEu9@A^XeJXq-S=rxvYySzIo_wMj3;GWHO?1AG-;6SIFG6fPQG>e z2&b<1X1fN?8t)VUtUFQ6T(?djfH<0YfKw%vuonQ=B z=tpuIt|NSgWNl)-DMMf0)0}auZ)>k|hh_YE?Zs{R;)C?|dNA?%PgeY6hN_|OxGYAd z4YA$shHp&*Nv0+nAiIIymdnhy&W&JtvuYb>w56a5XA< zR%E)hl6ft;%gIq9z(f9&%UY#xb2-+}Fb)l4?9mwh-FP1K+4(Qs*2EgHBO`4@o`S?JB7C%eb;eF4;12l2By+{%svmygYOks-X_G~W+a`?=*`$F1lG}R)MqIW&GxOO9A9!^4s&JdSBvn2M(kYttwH430eah6yHoOP zY7!rCCpcNU0THqMnK@KA0LEwDLZa64&{@6@YcsZVogJ{Lg@?ZHh&t#CDsxYq z8aq}!e~PRs@HD2gN4z zlo}Edpq?<}nig@FTez%W%rrLA zF|}Jn6OlPzq%4*5M^DG_5dH$ZNd5xCw3`xrg{o*&LxOEVR|Or7hFoICGpWH07j6~{ z=_x4{j?ViMO`4enXq)$3nw`xp=JMe&nF#^`cNOAml-(w4njziB8wDNB+N|T9d5?yR z2?Q5N%%cX~@)hVEdZp3(0%>2L;-uIG+*M6JR}sAo`!f_J7YhtECMVYCB*f+VXiHwl^G}M@m_t23kahg4@fN ztu-9BIVLy^DNnB^Td8Hnh3K1GkaYTAT5SABqpKB4SQ@@%=I}bgS>SOcL(^KpQm&6yT&0}`9>MHAT#BMC) z7no0S^r24{a-T5SRk97T$x6}dILM*P!B3pZ`~_6k-aR`uSvL$#U$;H|V?7C%)FLF_RoZjmOTY{6{-6w^z@T@q@Rq*%oXys zwMs5=K$h>M8bDKxkZz@%A4*3vRiZvj;D?1Wbuw{Z(jfyVeHNHeNwZQ&X)0vX*n;&z z?76t!W%@+51Izoj2zxWSkPY1H);awj%Wgi?QyyH)^8xQkkVNET^t0UO5}4;XWQb#L_5qV9zJ?Ox_013vq=cJB z8LTok>64*b52r%;vk3rD!8c)(cW$vB5h*RiBRlprx-IG@KT>B;8HA^t*CLjpVf(tf zmwcBeeX0%te@d@+Q8jjsuM%@b9@HowJSp_syCuNcG1XJaRC8WWRH2iZ1e=T<~^{-oc^GG1kBt`U7=r<`MoeD?obNo%HUV|;5V5Fcw>U`@;yo& z*!E%EQ5*fY7W1P7N92b+?B0&v`TQ@S5UZ+(7OpP2op4Y{jz;{FR^4Y8iPHR;0A64w zaD8oIK>8Y*ZdVOxOC>`G&0|!5Ae>2C|-?bGLEg6h4L@j37aU5nVq zY(~7pX)xvuLfe)nCsHwhKGEydo}MhUuQQ-jfcO=}?X+P(8}*2}k6ZEyhZo*P;r+BochcJmV@^v{(gUr|wO;m4a3Xq1lagES#<3eEYqPQxD3G+iG&vqz zm;EQ#>cElQ*7h;aLReQ-dwQXiis^D#dIwiHP2{b7df`kg#N&4c_2yH^O1X6{X%knQ zh`*SgtUBk`3KdGFOI5Z%!-?sctTOA@FFw&zR?!L#d3U$>W$IgIPQ=Q)cDNN}HM`y) zHxJxYVqW>R)vIZ1wY%3;#u+pHcw0w=jrT+^^axRBe0jnqW%*v{bx<+BYC3s;kBzHbfrh9{X zE*g6@rEB!IQjmZ}NwFzELKDG@?cy}O6@dlA)*(UHkj2iJ%YHP-tw;>>*Yz#kEa`1L zY89Wyy+Oiqid5%DH7T|KvUZC9UK@yp_;Q?<&|+Ri#^-|H$EqvR?6b87s+`rQzLveK zaLt!%p04p5iM$Wu;4$V`B^} z3xXH2mcIf&M3|~f&W~JdG?-)7Qunz=Ls%!WGJR)RHedxDWLLVO7K^)=!=!bCL?pBh zRE;+x$UG{-q$dlU&yFhgpXw^0RZj6iXf`GsjbtnoOlVdo1YkA!6n5ItXqvF+Y z_hex8&WpmElo<217?q1`H-i&;Ns&Z3{oAZ=2mok?d7)XoWqqbP)~yjNWb>U`dVjsD z*Wj0YM?_i)ng4rI6ZFLD&BqwDQN|PxGs4y3MD%`(+P6_`#`rzgy&M_v+3{}?YTNe* zkqrsOvU4&d$&k_7gsTfYEk=FN|Ga}jAWNzjJkYa@?>?F+vrnHitK(r{42l(VIy zAJPgfhSDefk9YPQDZqG8ZFbqNvS^9x51a{q0m!ECE4g4r9#Xz3yJz*s!&P0cb-7aB zjSg=rGYu>(Pb~45T_>3>El+3y=**Y23TK8fOlhDgML8LixhlYEH4LC>AXIr@Z*(ai zGTRB?e~)?ew00#Kroo5g>#O~9s_`>ZG=EZpKjZKZvQObX~k_T2#A zE|x`a1)sYT9m&NTMJOVmL-$d}3bp4;v1lCX{`{qV^qbL)dI^H z<5V2VwLGt&!nIt6aDkpuJHO!rH~3xB?0j=cG1&`70Fr+a1P&+lbg`=3%=}nPfI2` zUBxE{_27KLqZ!rF!Fqhu?RvL*(6VG36dK$C@rK5{!Zln{)-Q zRO&0If_|tCZVi_=NOp{d#7itk$4J|+)rx#_f469akvBT%DRd|?Hp0maYaFTEsPps; zA`W*SX`WH(w!tSl$=nqg8C&O?5EShCGn$R-0k_B(21YiM2hqhD#>;RYpRHvan;^5w{Xh81C+dRpX3eCG`v{)@<_fMM5 z)q?ZPM=kjIy$nz`CsRDeK zqbnZt-f#`;)oU9!JqPpyQlby^5H@p9wvrS*J{1X@oF2yi`poe^SR?>ySNYLeTSUKY zTd;hzVev&Ff%RtgN53qPR3Jr=ac9^&*AEcR`k7^}y;V08_T}}rM1Jt3HO8pnn8a#Z zg~vS>2Q=<=6?*8<;MR4t&#cHAX@p z7@+KzB??D1HvMW+9+LS{1x0KG|Go?#<)Sg(fp19C-sY$u~k`3w`!&SUqoaQQ1bX9-Fpa7`Ws@M525_<%;KeBLB!EGRizsY@>Q=#kt0Pq_}TRNi9o+hdfzpPUspkP{;GXB%l_hx5zIiWltT zZ|)Rnqin)@Y`_C7-?z5BUeUu&zN|-o7Ko*xf!~;TcXyxEMSHog(=cRv&MrXaS=n1G zc)9BFV9h*;Sv-6kAC3*@xMye&R`iMJi3?@u1CcNaDr}xfY$UJUMQBU7~SIg@EE;0m%|8_ zK3-Jdiu59MT{*jWl}%L31Ni+}xYMLgp8g<9$|U?30Q0Oh9G}W-0woOy$qv=j2(@Ey zUk!F%NgPkYk8NG)cU4TWm7#l7$oAaK(Q6!s4g+F&kCZzHj^o#i63un#rEu9M&BeGX zN?o!?9Icdiz>NzFn{PgA=?H5%KUy?`INi*Sy66b$S2jP(XGz4w8yjfHocf(n?6V`g zK5E+|;a3+3v@L7C;ee~930DqcaK*g};m0Q{tF?kq(8*>&Ly>1GjDG=qfM0ZzGBrN$ zSx%IHb(8A!4CnrWdxZk_zmn6weYCyw!JjW|Bq@+oLdhitGeiD^+ zw61uT)p@qT?nsS^AcqLr$fHoK8}-lP=ZP2Gqvu~c@2(W@y6D07tNmE>YoY>y2TYXd z0(!e(jZjZZBk@34O5#$cK(mKUZFt&#eG~X)w~mqJ+cj2^EVb`L?I&$)hgCp-xUEc$ z`%hIru7x^%49|t|hZ02fV)LOX2`CpkH~r)m<{#~C2}M-0oO5Q|g+BJDrZx2TGI z7t-}P>y*3r67kQ(_d7Zc9{}AaBBtm{?H_|E8y7skMka^Jmv^S( zvP0ZDBLXxcPShR3uYYqWePl;ZMPt)rT$vz{V_auTVADY{gWD0lT^;x+6(>Vx3EWlv zuH&6;V}xRcsXo-KAGw8scJ4A5+uukpNoD8or~|~Xq1W`@rB$1h@u{QAVV?vuk!x~d`VquQ32!CivX6!Y@T!@i<Wh{iQVesO(6UzQ=E@ zI=GTJgwZ*Dl#y(q`ncXX@bCZI}mn|u5jsI!pvgF49nuHO{7$Nfn5hBNkG ze^suCx`rp1MME(*98RHkXIO^aKFcl{>tMIEk(n+0u~<(;(=QgX!B2+~7knuVk}|W^ zdrb+kdiXfZ4nG=FLAbNfuxwBl6`(e6G`od|r-85qOjJ2CVEzk;Yht$9I%`C>K+LwZ z88AFkA^yy@_sgpBFn%;7@4h;EUVpilESL@JlQ|AZ^^R>HqSPqu_7y+n?8i*yG1i)_ z_cH!=i@n=A$p}2e8sE>|d_p7TjtBPlQns%~Ub`GAa;lM+s*k3;Jp_TeZgxdhr!0;W zImW4}=zzUcgA{*vyS6IhB`CJ;AQNpN59sOQdY?&uhZ8$-s&U1gYFm0Z7ZK_m*#QXz z1T%(RdVII9SXaX7uMQ?rQ{6fF63^+063VKL!ZzWe2a3RpaQ7;XX&B*1PNI$TKKcVn z-hDL3HU%MPu>0a#UNR}MvD$Em|A^S1_&oPQ?Q4osL6ToN>q4-USLNyr!*>%Urbu{} zkxdIC)tlt9@#KOKI_4?41z@G!zt_RLWsJlIMHo!2zffzMJ^KML=YRnWoo^Ij%GkRk z+hqa0a-7lU6@gz59qt^OzU=zzrJ8hpC0*2AS{=(?Ybm4Ww7%x zbim!o!uBO|M$Ls~LW^2UjMi!Z?9qGocqfy{%9_n0jeks0NYQ#W#nWURjNSL@0(IBG zp_&8rl|Owt%~rd%-@?ruT!z43R%t_SJs|};>B+7ml_P%SMJZNcwMcw4@s&SHC$>yy zWgzQ+{sMGTj`rmA$4s+g_*K7>9Vg2DkRak_WiKf5k<3Pz*yZ3=&%iJW z8ld#&)@K{sCB}m({K!l!xsmrWjoU_$&55hR8&=x61vJU0337WxL|h3}G#>o&3y1#UF8T3qarj+V2YH}TT>8=-k zedNXQAVpbbrS_DRJ4T0Vt%xUPbKWb1UTkWd0uEl5v&TH3uGBUrb?|@YvZc&zpK&-f zU8^TZY7wfS@bdggN)_nW_C|i8?~yU8;s8jcx))shT7l6TR-}=tklCkrVS`vxWLMFBu9g zq_Q=QT;te%7EVHM%pmV~3YwIr)p!#8BqC6=RxoLs-l|FD(;p;WIXB-oCZFZL$9oB8_x*Ut_7`+H2WRfQPVJfah#&hg80+6yXF zVa=4zVs! zP-{9D7CLTx6zRMMcBkQFw%v2%#C|>v;yR2El-Gq1MFtHSNp0#VaH%j82Y=3 zQ4ECjzRtxh;H=I3mL4xhCY@|u6+*D9bh%-JY2L|Eu{r7#vHXTlEj7krzMP%AL|ZyS z#iV7t!m-$5cL$lPEV!|t#}|vuPZdv+JN9%EtKD^hw@Dp0%z=gnhRn7}U3sLHcAWDZ zv~EV%X1slOjc2%%d*ChHettK?&(D5WeK-9*YD^y#ncCoOAIhIx*BghR4W_DhqC&&; zeR_bD6SL2*Y%1WTk4H_~93C;tR$d?{7tTAd6wqnb9wkkH-o_FSY?$m2kHrQVOIjL& zlv&fiy!Vvxnt5QXzFZvhgB(B}+~3++?O#MZ34>GZKH{$|)YumZS_$5=6cqR_CZ4t^ z&9Qn5Kuv2FpKFuHBOmFuu}`OXq=+*Yv;8a?XE@dot;DmrnzW%L6vih_i_6bO7GGX- z;n)1#s`CRZ6@!8I8@nSf#_VH`e9-}cu%!lgX(MYN9=7?u`@Y24rM@TKg|(GnXfbJA zx*IWA;Az2-^`e_pzCL%;XqV(wgx@*@M8p0;!VNw~Gicdd*9)59?x8r+%-w>az%lPg zo#GVrd1ACL(FUAUJD9Hd0eTK8WBHe0r@EK`JRhgMu3wGkiL!&nKZybuoZ>VOtV!->1A3XMGu|%u zhD=wgmV1ay_%yMEq^%&D&4jVzUC6pA^oaxpXM&7A+U~aAUED2(dP55@l-N(&_u-k` z!zBlio)0(@;`4+$85)K89J+^f=K<0q+b2xkB|Kt>)5m zXGN)7=ljP6FKEG6RL7$|?a7#IeCO=Du@%0zHP8+gR z`62~ANc7U#^%h&oxV9;zc_(PLf4x6KQsv`)wStzAt>U%i0D z%5WZ7tG^~Y^W$zA?$k}_Xyt`%z0wl}#fW?HYugKh&gmqm<7WgWK~u68Y}|%Fpqjx8 zMUhp4{!pgKC!lv&lF)r`#|X|SuOmo9-Jq_`&D6WlHJiNFN>Hs7e@%u;-BF!)Qps-J zYNdKPTLKvA^haYtXVI;Lu}l5!V&Vqb=+e2l`p?&We+f|IoaALA?@)StCe&&PO@pTm zGcnuVx5Sjerp51OoXo=KPb=m$WHn3yeKGFktg+mZ1F_3Vq*y@Mt{{l#BAM)ljNE;#4_NP`htPw$Dxx5!TEv)_( zvhlp*mOvkEZDVK0Dys=Wqw)m#GGM$-?@Xn68;7KeM>b@lY00$s*lPDq0c*Okj^rZ& zj(G2^wS}ZKS@vUfUdsL|C<7r6jaZ(&c6e3qE9qt7g-2EGHSw$B8} zg|=e&t5IgMgZkFnEq+319gd~_j?C#sWH00JI7cbC+#bz)9xuxWv;>?

FntSa7IR zr!ytg+cQ1K7$>%+9V>c|i*3Vzoy)SFpNk;oZ-tz+g7x6fkZ3lU5r&TPKjG|Xo5;_L zT85nyr9*2nj3q}IS>3ucW-+FDyU~YabCjWe*POA(Lyub1TGHGveRrkvsSa}8hm1s& zc^~xifsxNzzP-rgOmK{14RIW|LdUqWyJVe~HHzw19~oe+(KZp7A2mX;xCv6KrL|gr z1@aDc$piXOu4L$3akg8u`mzgRG@rG{9;K(f_aQk!L-;33YuIK7woa&Ad7soaGcks2 zf`+jdGx6mI*|G=IJpDS9p)ORp=Pf|pz}o|w@R7YXy=}+LK-Szhce$KTnP%_QOLdJ0 zPIf{Vrb)*ZgFDShv*b!@D~gb=O6WK4LX~oQ{zxFuFt1t9y@{qm(q&WBU%j>yaw2ug zXkccudJ|_W%^Nki6J22YlRCc0nnludX*L5+X!m-~fd}nv0pdjn+BwgdbF%;fP}N!u zEcu|1BoLCL0|C%Qj@C=}^{_Hzf~E-qi_i^NA6^lwJ$2*W79Uy8coOgqkms)Q(e-iY z)=5IjWNcj%e|p`YGA48cO8JNbNo~fPuG}J!osg1xxSTK4{p2eVCSB;|Ssu*Abal?% z4A%#M%3>gVt}(B8DJ20dyp)p$^rm^xeZsbaS0oJZ1K)}IpUB50)HZ2is2A%q%ptkV zIjKg7AR`Kf4Qwr8 zOkk3i_xY}8tWM?d&LEXN9E(ykv8;CmzUzZhlA;A`~No5#Ox_YjDI727R%186>Epe^kCrU`EH=V z37lNH0I??egs(=>k9)=K}d1Q3ZShs2-eHj(dv+loO%x02ckZ zJ*?0-C?=Q zeSOLMXJu|a^3~w9Zx0xkDP$>5cLJTPKERm)cRMit9*TTl?8$3RH`J$M_ydA^oN-}H z8H1#+7bMQN>c+{+xt~#aD^tj&Z?trNH3+jt@-^&oHl-hbfAMiXlF|Mdw`1dq4q*6a zyR!9ff%D$J>Qdvyq))X(1R0-!0w-@HAGN{smgx;YKC>t}i5e@Y0HVHp($q|8Z%>}r ze4iNcxgtk=&Ne(H6-C!a2D=Ox;tRkU4aNXFU~Z#$Xm;|L_U(M!#4jXp=FHAW&& zoz|V<+um~&rR`dX3|WwfB8%N(XYumFvTco5>0I|{$2K>RrErS>z>zuK0Z`+9El_P! zEMwEYHLkjKr0Y065udp?)S{0HjK5lJ48YnlygkYy2|O2*`KFN-kBtS_W`RRFbOCL4`cAP` z3r&wFA88o1)EMsmw2=LX7xyrWg;1bP5>?0eDhEXA>`)bq&Ga6-6^?8{DUD9EudhT> zs4mOs<4HTcLyn~1k%AGfbA>pGcVRrxV?bvoZebqk_N7DU_RT)KEr! z2P7Ka;OU{QSCn0dS+f%w&vCmj$>)LqS2f=J0n&_^KBS0OJXzDz1?@SV%LU=R1FgSf zT?f{OfDRB)wnVEm!!;}Q#DKfV|L2BXo3R!af>VxO`$Z1zjXHBYzZVWY`n#CD_ysN7 zaMB~cCr*;36wnTAJ47w1y@n(;2J&fOl-jNWsXWY+%E(uG9N6DUu7V)w)HWjv6EftFBecccnaj#SV`x1{H6SmrpoSz4R)jrKqBu z=OqpB$cSo2BC{^pw6+5{((GvnA3m@&@e#U<_+%xGDMpg&U^D;%@ubOLsmO0;X4t>m zexm$|pyQy<)j)kezJvRDWt`b@392*ZeOp-<$`@WiLrqx!05DEVg5PCax0dJA}(*-f(xnd8EWBg^025umRl^k+=ysWK2^(utYc zF^9u)y@H*0T`$pkB-ew}CihfgywFBQ@PWze1vB07>nyAy1$zy-#)h|cUOW`bho1d6 z<$w708dfNf1V7u^1(qY@$EMy&I}Tf4%oWyf;wIw*DEFVx;F`uU`n2jU4>`-12w0zJ z(7#mCwSOa~KvjOgg?`X8?tY0oB@8t1w-~2R@zkGebiAPmyfS?Tsc#GZphN)dW$=NG zI+GU->Pw9g^O+Q*UwEFK+{Ns{FBkj#o_IOzC}ndb_2WEEuY1p!7sVd&Uvc+1HNkvY z|J64G*RiPe#AITyP$|=W<1cpG)a387rtHaHA1j>;6TMjUS#PL=H~b%fqD6~*#89gh zL*>4K>}hMcRI%;;xaeq5W7SR3q5CLh&IQ25FvwKyH~qO5?X!s#-5;R=b*IxV+g5$y z837#0LvTYI)f`W6I^Zu0G2zv2Bkm{sX*&i43J^fF?+3rLg4sZbnK ztuF&Yl0UID~V> zpK+7`%x$e$SvaWOoDX8R{-U_B+(gJp${`&!{5b8ZY_fWFM#_Nss^OfZ@1EYMT2}w zp?9u>)UwMNQF8V2Or<0!cE`wci*34ZafX7o;to&PnPC$jGIV*$>Xbaw^30k#u*;P_ zr{o(mu6s?D1&W&mvbSVZ+op>=>1icNMeL@}u2(sjMB(^$=pTRzRWd8UQ;Q|8u36q$36Z;v*z;Mp9J z<&ssOXu}?ec^g+x=F!oNjx=4Z6||xb*4d9+_hp0?n0J1&&{9QtDLxoeMKUA)&wmZO|Q@goXPxZF_ zR(4omqWjKq6V+d`pJm-Z5c4AC2Wa@c|D{so=lnf7pB1+XKMt6su$NIk>nbb!k8ONd zU-_G9#_bLZd)P~RF_q1qX`67eCO5!?4t2(`pD46I915V#R$QHAj?0fDrtau|x}w&x zC9o>}{H5QZ{Db$rPGw=V7>-+!f*8oDh?b%)PdH(>qDIG}8EKT1W%7eCUmVc3G z@eDFYiv9yAk2(6Q%&(L*mLhYWcSI{;=}PNx5|esI=Ytnyf)+UV=q~q7k_tczK*@K* zf6PV9+}4O=J#7;s7iCXCxLH0X?<-5)|xpT^Jn_ zFNJp?NkTn<6Vp{7q(07iq#i#Y!v#4~;xfc>yUB^$6Jizf_R%Ny^DgrA>XOR~NKE9aR)^81il4~zrjM}$!(Bwb*|?~&>_W24CvL1=d%iS)-$Qp_cz z;sIxq^sk0bmA56wE;#iY_9uhvz%8hz!3?*p2!?cnCGZ%7SUz{D`g) z6dH*f7vRdbs~-Q(>b__)x-O4=J*d^OQkI2xU7A9+v|_ZyJAff*mVV&X5s@G|yfC)oh))2=jHxOf(??dk zVRkd$+2fG+zHIgJI@#5SZn*Fq7~h6_UP)Jv^aj6S6+>2oZF(SSBbQ64m|D%7ib=uPjbzoQ5O6D(gShRPi5pNC#-ezJK+!Ahwk$@O%gsg~$jeXTPSO zz?=FG_6iT@zfEgYnkZ|&VbZy&UmI9e5WPid%C?f^9vo+|2Z^ish1wSmp|g0-j9lCu zewtHkM4Q8|;B;NZasx-f_5^Pz1kd&~pUA4p(SG@mhla`ZCzkW1SH5mi^QJ4biB|C6 zbag`iXcLXRJ4=fIJg1*^0 zEADtPmI&A)0Q8nKvx2fgyTIy_g|od1QPPg|2-KEm=kX(}QPMt9qAc$ZDPxkjF3#LA znOw12L=wJo`(xVO+`uB^0WEgm%nXdDLA}oE^MNJ`aQDN&(+DNcI91fmXrBhcJ{hRCtVxp0+55OlVJ%=Q!rmW@r%Y z1x+JS6r;!L&p$&(nqxEkuF@UOrK^)RCpS2czr@hLwcB=M=5|^m*1PC+o5~zhoT$zvTJDnJjcJ8j7JgoOXJX_j z;X?(#H9s(NM8lD8lLF65U#xJdAy_`q$saU2 z75Z7$7>C`4@GH%cjBEudDXczWM5IDWzBZ=S2hm1rl4JEtprO)LcoNxmqlt2OD8uw5 zkUY*bKuty1d%8Beb$%_Dr;?Nr3;FdjFeh`C(TW{AHCKz{uUvrz_9!bo7J~gZ&yqa< zu5v#Oo^yk?9Bw3s8X-S@tp*zU_FT?PBD{LaYtTTx-)NmrQ_B;lXD{ zog>1j#cyz;6Srf_#%u-ol|TFtzwCz$R1BEBp%**`|3taWNyVzjQy;0cwZt#+9v1Y; zNsco_40R=XrB%C?7Pizn0dzr?zlNwRB--{SXzhXev!70XKGtLEmtHtCN=(UgMJiyx_F7Z8)1z)8SL|%k#n=| z^B4$MBO>;k6EnB9&c`Bb(yTwf#8bRZeLDDsYjPf-{X}^B_?#mYkT9c@xFV!JhRYdM zkdZl>Pfer77ytnW9a`(R>jcAR&QmYM+<>9z}B!8bc;wn65> zKDs=W2CpL7sE3G&Mdou?ddfR8U@P+yNZU1+p&oO|-$w>+uSQBWTZCS7>^h308NXOB z$qCPm9_5uWp?uTC5AaUY$I^k1mQnI}K^m)0-<0Dek|It{ObU!=$9e5Wwo&LyRDlY2 z1%l(bE^|i0%#==%`G3LF9M-@7O_EGt1O|)qTBgKEX7cK$qD+EI91Yx>3GDgL0rk;|DZoHxHHz+gTEH&ZqxQW66q8ml*5`V=Vu4QDAq9F~px;;^A zQeNlYpsj0K(qn;~sgd4^{tgc7s9}0Mom)u$?5`+W3D6k;_OpF39A={Z?J7-*eyOvx z5_Y8U5<-Zp{+V7wTyIoi_krD!=C{*0I;h#u*`l$2v1e)y}@64_Ow92ZRdl|Mu zc3t!mo|}k{`+Z2@=pl!(n^ctz@bUx-9m*>2JZX20n@V5#Mys$j@V&c~U0UvX)TY~& zVor&dQkoa*)IRT5n&YikIf`4b%GyahX|Kp8B0F{aGw4WdF5M<6LZ@WtDr00(tynZj zeX2FZWyFh>CBB8&IWHq(N^nM_e?dFHa0Q7iS-p+k<;PNqsud`XFXof$aW|SZ_>ZON z{eOU_{=>+9|N5Q>f9V$e8v6t#gLQbV9K1>X0T{d79Q^?L3f&`+UL}pgVD_C;@!uE$ zJk7!QO|;QzCB2tox&3c_0Y`d{G6QdTvamx8*v8evuE$S@X&CHwV1!_8r4~d*vzcoR zcJ8$+J^M)W2Lq*anfD~>0lQV;v<6#0G7cb2PEBKkYDNF*H&@k{seF}}mlTW1`9)Q{ z2z8BZyC<4rYGenVo=arP$zc#$KNKxT!^ch7WrDIbmkJZZxd3;~z1(}Ezz~%cC{<^! z_dfvAaZ@{`T(8S}w_+SMDaIg$wotJ{=m5{X&2<5bTnEp<2jEhI`6_&71XmK^2UBv! zTRnWj61@HHNj~Cw${BX>E*zXNl{(^nioxzh>~T(0K}z0Jh_qy7XzrRfCIowy#`7#9 zGOXhCYeM6Ov4whNTB!-Qa=D~4N1Xi+iaCvv1jDfaFLFPtJ zTZKAHQ(%LPD3TvEY>!jI=?=$1FXPfAMTxXMO)a7q@LHloj}!W7kWuJudc?DA#ZYUi zO)tC{15^H!yKw$>zIZ)LL|&|sD<97QQw7i0EP0w*<>>C09SeNx?d&4{YR=yiA9bcO zWRB_GV8_w8jN;u(2Ep6Y z_ZJykL2+<2s}PW47m&-!O2fVaPj>Joj3~oaW%8b2cy-krSj;l#+9`UW7@0gs_qJXC z-HM3!q%#dd5G~$>jPu0%#9G1O+~)j|IqJI5S?Da{K47_x?^SMqms4l|R2wAHGG0O8 zx&dj&b$xfD=UemA#C~vjvvQ$=*3VG4zESAU3Qzo)f18=V+o73_XH%Aw;BzDxpaI3C zfC*>oCNl?oiY_g$#q2FrtHLI)h@?M&sSN|glWIR%gDYEXaS$)!(J@`EXVj=@?g<6TUP(#|O5wpjn=F)2?>$p| z`b&~kjFUZ#0Mw2&ywj`+JD8wT>`pWv)gWDB%S-)F_n;};+rxaElW!;5rkcA-i4xT2 zC)RP>_&ycv2Q2H7=9^$N!Lkq3Y+36j99G9h^t;Die^bsad*&y~=Q%;?++&M+r@sS{ zpZp1b1+TNL+1p+$@Qhzp`Zj<6FzAGJpf$P8)Fk@6bIfBhGcNbBT}(>7DWSaqCz((2 z(ANwnApIQAzy`wa=xCb5NY1<|lE#~~n|@xbsaDu3x>PJ zF@gbb{ofUnwO4kgA58v+*xO){l}AtyUBuh>@)>7FKhRL($s|xM`BkfnA=+LTjqTZ% z{R$!H<|8M8h?QjrJk79l_c}5<>HVF8$K_)-%K{5Qh#d(~cqq2+yM>spa;w2kl}#Ei zL`yb9VACNc{;Zmv4wm)wZEP@7i`8n?_}w?;-dmnkzDUNM7~c!s^7?ICcdy*m3ljyB z+wyMDI(M3;3O~VdQ^V9Xk++Z6J(P5~(v6it^2~WQRh`JfE;!(c9>N&?8h(v38DdY; zYV)LjvaG7O){w-ky{~H5vmaT2z`ov|O{At~GMoLFK9!1`HbWD}zu(jrHT$TnMOG06H8Vb}OLHUlJ0|^rtgK93Sc*5peRD<$7l| z7rHGhd)R=$HbOX8QPP?jVD2-GUgy}54ih|Yd(w5crXosfa2AR0O-WCn8=7Vx)Stvt z1sSFrD)^L0oHc8_bLSeb%=2MoK_V0T)w7PlezfA~da(Bl(FRdgSn3oomP7@Yt;#5g zcte5alI_ExoQv0PS`*o^h^7oXeZb9yN>|et%E3L_S<(;gM&x+jiIjjKh3B}53Gmw- z-PcR`e5C$obkVkV(XTQ9fDJ+<_KAuLL7Do7k7bbB!X(#@39pny;UferJ2#(m7%CWd z@J_OMO?L-OG!uA(2h8$GcX1Q~!`&&qfD;fRGjV(u$@$)ks=ByV`-@zR+IpJFMCSwMzNIe8cQ9g3f z7Dm~e`o=l%(mr|K*LhzwlUMLXlRh!8f6_o*ig3)m>~=-?UB=UdU-ORO7s<5~TNQBl z=dhwRVA-!aR>8K+dO9X` z^mp(~OQGH3{cqG-t1c>_Lwi-%tojli%}w$`i=Sap@fl-Dofe{z!amcv*}k8L2w{z_*%!~|@rVX(To3D6b?()(_UO?PE z#=DSa;1H%E+S;Nu8u>1+@P(H|sH-^JxyiS!uODnJ$yU%iY4QvWxVzt&e&cRe5b3#Lpde?M2W+@9!vBdg*creWFwLGm^^umu}YlY|B>RpSp%Ac+a5>&{WVfTkbhaHFJyL z{CZ4UwnHq6>9~*wj7`dW#}TSaUPGtff5W* z9h~XB)E)ev zj}qjvtTJ87X3{gN7AO#yjxjkwB$J1D-RSbAA-#_~(oBVdSWb&K{?}}sc|nd zlw6-J1rc+VcE-kU?+NYZ_!I>x-~@Y$U;pW4dTA_ZbD|nLAyx)Eh4D{+?BXW@+op?w;*@%tikWISJSxcd!|Q&wz{Xa`d5a= zp#(6Fk)>a2*fNL#Es#(~;Z${Hw+B}@R%tx{2Vgoe%s&?f4gX#DHO{D&kr8CgA#cGk zli0wU$V?YWpn{N#o}G*^pFqqkaYTyHJnhqz)_BptTIGHHnzP52e(YQefnP~fHfK;V z8`+~l^J7(v6QZfA<)Pj;B&>8#NuzG1-R*vCcWB2RU0CJE+535&`UIl}t4I)@T#vfH z=s-}>!HlN9B=6;_lykhMUh8$P;od`Z)iEAq?9DYVL$*4W7MWn_^N0;IKmf-mefsgo zFbNNeb9Xo<#}=iuhj&$6`*b})7Jdvh92v57$!e~IT=aJvD(gVr@bAw(?%-9I1J37$ z)SZCu(;do5oy73GLDMaGpx4hQ?znFI(5fvy_Nhw{wT)nob{|Z}{5PZh)?)~lnUYD* z4^EP;TY-&6%FCs(^N~y7S%9NlkX|jznfmHNpdY9++CMLQtoDzAt;j$ z-v9KxcKqODUMmUzzc-<9%E&@ZwS#O3HN~i${p+)3fhuErWxpk=8fjm(G>0)5Xr<6cDjKR{ z7gGtAP;DFcN6mOMT7~j2Qt*mF&)>Y5!IrnwOd1CC!(AoAM7`9sf)3ZLH;IiBhX%3GeOD>IpaFHW@EHcEBfpV5>S%)j`C7q?7N&dP8vj!p=rx}Vlh zM0sb;%84zmPz52R1SHbG5qsfX7h5>-$}f2-VNJygnECdZ7AX_H&0KX=jcX*$&+t#| zNTlh&7KUFl1?_e?))1L@kWY%()FMR*}ou?PRR1#~+=tm}YaRkCA$nWoYOY_>xb>vAU5mvMe$L1qx28`2M0!4Q}1E)nJe)m z9NyWL74DN=rurEI)seP9P3?K7dVT%14{m#(sDS&N0;EoFo?9gydZ*)aF%so(v+jZ{ zxVbA@?dV#wHty&1sY=QiF_og33wdh-B7qjXctw63#-#w-JIK*{a<&YPG9FR5o5UYG zAE!agW-I`nC9BTRwM|LaonalLs4(=qd^iM=pFunOv|(Z5(GWW9retI%VHofAy&AA> z>ggvu_|3tYl7|=ZX>J&|NVb7_qzAbwO(Jn(&5Aj`HWSu$FF@|8!`C{0hVs0Ism^l+ zs?Ohs@nZ@=v36md+k~R`&bfcp+Gw`Xz`3Bju>|XuO%K-7CLl0F_dT)?S;Re{3D53t zR}WF2cBqhAx*S{haCwis24$9*+BL#JhLpy~_`H_IQw0Ss_o3f+nW~x0GUg@t^*Z`o z*IK+bzW>=8nB)?EI@f%c=M3d4cb%U5@}7?w{sq!!*KUFdBW_z3aePDdb80aOVkC-x zz3==#JCuLjrTj0C+!&VlzmQ7)mGp%3{|=igGP>`>{euI^qq0O^q~Q|)t8YOruYu|* ze~mHNg?35m#1D1b6DW&adAPrTA{3jP=xf}8>=zkXnt%J(0ElrOn11~*nMw3$8jJ#X?%{WB+N_%^x?+KXcc^+-BMW>s-!3L^{z{ME;_xf98)6GZPB*%@VN)3SWU$#wlfN|CtU&vNC~)&0Ux!}fMV!CJ7|p?yES=u4)AyLooB1Yp@7 zCN^6_wvLsBM{;5;&{jei5iZtVCaVt{9PE|nQ`y&yimw&dYp-saKFwO%<&Dqn#4Vej zpyfZ8t&FHx@{~l};0+Cl|E#VgoT*!`;J7$(+WNSo8T`aUqqL}E=F)!R^Xgo$BD=`c zT))t-XTBH6eIt^&@2H7M$GZ3Uq`pXOk_p`*Zoxsc>+e@j{G>J?c3LNoKj`%R()yc} zsr<^rEP3z!Z$^H;N8p(cN^C_bM4ark6%6cMEe(M63tOTWYaO(FcoY*rDMW#*=gXxlkL8%EY5HV?+#45=hI_ zfm`y?mH9Ts(Yf9I81{DpmE1L2U`}D_;=x*t(iaion=5>jFIQiXC&}#s8xZgZ@p4gZ zcx%48?rqc@941ScKQ9~nv@`tE1s^T&RoImR4kU(jm}F>aV6qluO5c(S+;lc!=_Qw3TYA6iYBnS3ug|oY zV|g=su9j*Os9#OL)P%cgt)?W4CM*jWcrf`!N4J+2ly78KHqyvfDEBP#CH6_(mNvT5 z-)Uvs3IJD;rfs-F=(*aWwm-L+3Zp$`x$WixgV=ZiA}QLNXK7!kAr6)T%EgX9IM7+m zr6(>jmatKAoe$f_@jIGfM@L8MOm`YuQDa_j6|o@W2O7AGmth_Jclf*fU)F4VN!hh& z`rbQBa1;`V$Y8|^Ylqs>biKic2)w3DbK=K;OO(8ib)uUm$LQWc;;NndIwKcR0>@WhQoNOSxW5+Q46B(KBtjOlgLuaRhw<5q}=ps=$k`c``qC+I{vej(B~& zaPea-r6wTS*&yqOSLt`w*aEAJM-OGcv3$-#5`y^>ZdluOVqlVAn61qs2-&)qu4+NIpL0d>%0dmr#|gy9fi zwd-k2Nm;Rjo#jjCc6<{B%n8Uje8i<^doL)ft^~X9*UJVXV%v!5=TgmInw)w1sYl7issA7v`N&;BLs}@1mMp_qQwxV zPe6na|9r*>cJ_~D{y>{lb?9NwrO(XA$55!m8y8#Lx4_IP4~lgTh0H{2Etj6pTS3Y2 z*hh=cPt9CZ?B)6lE9A67BRS+m>qVqMUvX}C)sPek35#-A6(7UONrmN~w7@wU-af~o zK_oY%`_=xw*mmir@#FYjR@%vnhmc!s%glI^*IYa$AQEq2tf*07V<$MNRWL!3ushD5 zcH^pcEK=hqx5`ypE;m~szjnXA62gJfLfxIr0PoZMqdko3#Eg!b3K|DjR+o2@`Wq5noj(f{l* zdm;vuX}ch<#?7`Si&uCNw=a@@J8Jg^|CecZS&^|mqg_=_w_Fan->9d5Ou5Ov$B~Dn zhzvK?N^$z3ZV%;sX8Gmex8l~sZN+{jZCTbb=EfS~OuW8(nxxG2>^uS%H*v#Q;-y`> zo}Zc{o3AIyBKw}@D0wug%i_TP`|0?9JAGs&9c2+- zxER^SEsQ;UhEFVHiUJh@S`_;iS z*F#}*-(SkmQqQ(O(tlp_9m@ecqi8^x_qov*4JFSu+V1E3bH7_SN=; z^`lL6eiAWZ)wZtV#ySO6*TH;r|L%AI43)Qq}cbT>KOJ@9@BMir=Vfu0Qqh zl1w$PKl79u?n#b`tJ%Rj-2wi{t-ylFQ=jRYCFz~KUKZduSS6pb8@cd>a}aZorkk*J`(*Hh)kX0&|wPdSki!{PB^dN4QYP+-o^$PXY1?>? znJx2Oy@#@Jajv4h*dQKvza zeqs#uuS`PG+j)*3JUEmzy87JN;Ww?RTT#?p+gn(03yU|v}!87J2 zGr=?3`ZRu&k#*!=y++gb(*m-kQn6b(3~o2-8~OHE7h^|e=!S=Yj{Qpf@XAf8iy#(r zc$0wg9{vK3nZMSI(#kNeqQan=pWrOD6ToqC^%_J`Q+A8K(|RP+AVp6+n3}6iQ9ksq=TGma z;}Dx2MmN1I8cb?Kf+gq%yOWx8*FNv`B07w>0pmqp7>a)(@y6k%e6c{^pRFNx;pj}4 zF*?=Sa}Pe}BOI*j3~Ux~XZOT{pwG(<1Z3-!JDx5V+07`-yLwPywbFRzInOdt$?L|FZlno8#vMBs4rvYz8U9iJDVpC zx{E<}4{EJ5yB}<+I4()f(U=~h=*{!q+Thhu69jgT+*4<*0ap0K#KoP{o{O{RwZK|l zaz8M^>9~xhY?L17_cwnXF}p4U4w{jUK}2{|8WnNz4&H9(+0nb)LZeeOlhdi~!ngQ- z(r8eI4W_KtZnfz#IG$AT0YTx%QQ(3T?c0O*X*gk&Z}jC);XScuIYkT#Tqd?z(mj+9 zO<*mqZSz1`GuS1Z&=t=0*L@pAHD>pcimOC!_s~U%AH~xfwnhBbT}=5Mjd#Neaa{r+ zej89swN=*8tC`*AO^7QnKH6rP<$0hk3O@yDg8f-Cs`=9%kNF=)>BhYtP|mXM*vDHj ztpE^Ngo+-hba)fF-D^WmPfcNk2PkT8)|q_KK|QPv901enqvr-q--i{U|jA0{qY45UUY#tMm*#_7cD zMW&J+EtPW>I>*O|DQJkF^Y?u)S{!#;NPEl(&^ z?fYH*L0rB#Q=ald?;n8kl1G5;1wo1}tTx$L-) z!jw4#YWxErph(Pb2a^Ar>VWrrcjXsL)7s~#rc0@hUQwWjDH6?n&MB*39-CCTm%7k! znzyv7d-_Ag>~(AqIxel)ai%A!E6CeQ(+k$FRqhl&7WXLCyWN$j%Rjo(GyGm7x@x@W zeB!W(8F;ubh}}-V_5yrP%JL#*eyZh#n7c@9E(_$C07aOA8t`DJzn|n#l%4J!6FKyG zR8-7YbywvVW3d04ORwrxKSNu%+yz-9Cuow{wfiS%W#RB*BuYoEyMfp~-5IuWZ95k& z``LqROsI4FrYvqH5)_Q@*UN#}lwPQ>mG+n|2H!YyqR4w|%BV zZ8iPn1ga*p+Erhl2Yk(Mg6pj}EP69$-K;Q0KSgTxw|iB>eX%??2fxo+x7dqrb07XN zNa*(GTcRA-bltDH$K26mm0cyZ?3*NPbkD>-tZfU!|5K;cHuduuJ}^>f`mGX~dJ z(4DD0&B!*`v$2mr(mL;@r<{^^*i2=VVJreF@0+#v4m3uF8!pLlq4YN_nLfn^hdQI5 zya9`EHL~@r*=fZei)r1uM3v)D#`Wv`j^YMR zViX$&>@&{W(X0D&2~a|)C061M9%wQQul?g#q&-pxHjbB$q#sxRHJ2bqFe-Tgzy8gc ztZt12XMg=%c_Kl)+S%wzF*HA>AvXz~tmwa>K=aJwt2U5Crq!Bhfz6IzH_eg24ckm?KM*BrP&DSH zEw^k(Sn734Rr(L$lh7-#W6tdOYmtqojS`vA+A_1+EE2_frNLQ2lZ_GB_*AeKSyon3 zI#(Pl>|k&lH_^EN<*wjuXMQ4Nn9J%YZrsjAQ2}6;ZBR7LpGGxCqIm$dVZXFF$3>{S zGN5bZJf0@=pB%PTZ(I$QcfulheVnn*)W3!K219qV>6<4jZxff~Wzub%towF_Zp#M- znXY1mwSqFg*_w##BGh(2oYdY|GhLtQChM#`1I|QsHTp$Kgs;^*WNr_3KtPFcb<{bG z(r>FbyfHS0Pc!~kJ7U}%^=-$gqUh%<( z4W}3k9H$r4x3L5c??7+DZ~7Mk-;)R;sRmu-KzLXDmXtb(+?PLW`0{8l) znw!9$?mZ%ROVVb2N16@`Yr`MJMER^`d(h8=;?E22nZl^tgmw<6eF_702NYj2a zog1n)d+JqRO^(c)q}Dm4Pee_ySUB&Ma9m5i4F;xq0IDmHM-6_lJxCj(JvW z>Yt3l)}K>9C~eZh`%RkA$(e!_(H=<1M%OX&H^t;&@|aYoeQV9dYK@kx!_n_wDbcG1 zGhIj3)JtJVu`xDYBMycpJOBwAFWwQ2KANGmSA5nlRC3eMCcr{8PaAjXoO?_vaJBUv zAXWRwuPno(Oteoq=D12|4(10f1~TH_AGhG zmvXGYx<{{V_no+R@&sF%2A};{?h#>3XElNYBg+JyACY{+`;`yuSDYGT{{e`h^)5W; zqKfKM$o_hZFSJ5&_@qChNN2b_WNVLaygJyS`_>#mhcPX30uJ?(swsmSa;P{x1M@2$ z19YdYSU+38{{*)IJ2W&sJ;heSbXHnEuh}sKSwv#-m~rk>M_yW*J;?uRhG!5ITr^3w zC?cDXD`$}9xXKEpB*Zkt%Og?Et*mYTS8U^5mn!x8?=lD9Nd4+J46jskti3xC)U3Pt z#h)@;lq@N^MD(*T%H)}<)lid5VY=>-rx_Cb$spe3!m*t)r$Aj7_Qq=nS{VO;OZ8GYomt?&9#Kt5WVZG1iFGab1t3Q zny~W2YLI-zOToN#>}5(^p*sc?I$={blw z6!ylHy{}9=YoXKOO_^Df9KF+Rwd;0JXT;vDHJ3Qgev7mjxZKVkngo>haXOCM8AADy zmWnJ2%JQlJ+fSMFPn+~6%1%{6W`8$^Uw}l}g=0?T0*CaR`7kq?v9TiLOC5f*gSK6g z{K=GA$rj{D0a(?32aQ|UuJU8#pT+T>-N8d$&SZ6|klh@rY5}iD;-*^Sd~4;n^_GXw ztwnIDBc4*l@b>pMlup^{T}w)(T+A`Kg&6eEoNk&LKyt|JVNMknsC>b>2~`d?>~~iHcKb8YgeLibQ)G!;~#)5B>ChDT6k=h zM6eHa*aHXt6xh#j&n&r!&Z5^e%n`bp)M;;7{0Gp$LAcC-gzUOS*OgMGbVD4fdHhE$ z#lBL!BsJlsM0UjrF|Yb%n)^x(7;LSr#d@o4QmdU~XCQD_D(&DyL`wR~wqi#MS|s=( z6|wcIOVOSo95snSuVRb%G))lOfmm9tc9qX8I%7`ff!Yk`YWg<|Q)PoSMf0-~7JoRf z7SkWRl2smpAA&zAO{h39oYVskJ`gh`K_i=X0R1?AHNl3PR3TJ=YHaZs<|Z$1WqIryn29HisU3 z2B(SeAA`F8TqFI~$9D81IC0YqMd{MKlpXDH$3%h3ZlJ++Vv5H8az7c9l^r?9QO?5G zgHztcjvvkMO=rXQuTz<1JQ{HB!`Y{y5`k;i;W2>^dX3#Kbn!gzAY?r$>kOi9Lwp&r zvBg?{GVvW}kntjgC{JPs>w!TVFRHc_^EXXx2UDHgCI~858+!qXIJOJxj|f|MMuDKqs#8tFGutH_t(HLAo`ZsjfPku)tmkqT>`>o zu*o&XGH}2__mDp6W%IQQyb{!5wjNZoyn~bX@j=$Pa$N&2b2_%i1MymCeH0o1{`$j6T;crj8pHhFNz#2+r zF>L+py1(Lqr3iBPAiW4Tc)`edopxT5X+HlsL!S_8ie4za@*q^v*H}Q34g&W;Riw8 z?pD5-R-xeWWU(aB z32Cq9_8|ZU6m=wpr1DU=%gdG(ppqABNLtEdFV11SN0OzSIMAxV5j@Udq9e&s2~6+4 z|ELt$g1|mI=52mRM{%RxrHLW!H?y7f`|PJuDiMQi>baI-2tfk=tf{6%+2d|!yfL#XM68oYHr+)y# zC`yp)9CsyRhOIrHg5A1sJ9^Z(V)D3+L9kl)lEl_@&(^=hK3DKqVG@^qF9K-rct zT`>G*TdVBw*#Q{!Jclg7QA#r_( z(ZDhk?ER^Pq-_`z(n=yGNwD3N`Xrx+0?$0dY;$|5*TTe%CU~m@@&BFoUfavATf+@k z-T#x&gfFuG#UaehksQXT#A-22^SfeJn)@6Zljz8pl7%wLEfMs-?e1Jdk!qd>yZbjE z2S1QI-ziSX+$na{j{qUb%cWwVkc4_&tfE#8+gTB#0s6sxqp{>`D?z_r0W?|TUn~K@ z|L`J#1DgO%S#|(p0CkT|3}(R7=lv3ik4o1oC(r0R)Mu(DmHbP#GjBqti{@MN&AAWSW!N@|J zE8+9UZ^rUJW~GMw9@4qsUBrw3Ot#A>Z~5$RJ+}c@Yu7PsigzjUCCSBIsrL!KCK#=wic$)H-Ef~j@)<@ilE6I4M8Vg6h^U3<}%FE z_>?Cbzp{Am`<`l~I)Bk=cQmcuA?NwK+r3k~>@XVRyiCOBjCLGfnXYWL6~+}Vmt8dA zQ>y5BLs5Dab6j$D{#xi#@aH*ckz`zdCQy*b{@kh~9pUv5@Zwh`UaxvKel2L!Y8RkX z-ym2{j2q+ZLpB+<)zV^L?LwM`75{P?Wz9CAuHkMZRglpxc&rzpeAD$W()9R$+w4NK z;kP4(*=xqJ?9{C9VP4kDB_?0&1txcv$xLf4L0;H$LQ!R!!a^*9oFXQ-CF zKE19?Y(2)abhpX}bCdZSDLC#T>lxK&|6YHgb4cYn(4!2?ZzvcxW$eh-P5D^0Vt1wm zfr%9TMDS6tCp!|pl0(0*QE^b&SNzBb2N78H*JRAW28zuNRi2tDHqqdujcUf$q*CdX z`l{l!)#T?QbF!5%^DYul7b8Lca&)CHx$Jy{i_=@7lcNugQS(1AQIPK{*Zk!_yml^D zXNE5A`Q$0*$eLKQoc*#^<9STD9XRa6C0I(_nNlDwlv^&Zx_|}o=~0^XTF3g{V%*i{ z^CXOnOM|VE&(yK%EAvg>uCrzBGuoE(9?NBRn#Peu+8w}Y!ldZu`gC(yf&X?Ihc&UJ(bTs}54F0%ly8vm7t&!gb`VXK^tT4;OZl&!@ag+}h_9CLp} zX?*(>gX6I@Vc#pKCvN(1SumL(0i91wNT*oze zaS&!TM6ffV-48`w$)QyHAgfjIosF7@^NN*gkxxVEbZz!^=H*&BB`&Rg(7tLOcZGaP z%kN3FnxaZ5f#Yzq|cxq__C{C%_o%Cy{0fJOygp72K zA@eh8{_pfmGLl9r1iy_%qz#=2zY?UDVOsLpq(cz~rTjChkt-Fv2f5nnyV@CD4SL+s z>=yT1++pkL*wD^JPnyE|0al@t0a&AA(Ng1BYG6GgDOKAUu<&Z(OGacquXj(zRP(%# z=-5AiCHVgytl-flNN;V#_mU_r1bMh`9GXs*GvBryE4MOhz`&V;VXeg}fo*()FZ?YI zjXgVNxH|RPj*7F`jRS_>ZGznlOvw@CxfsDZI}mU)tl57^tH=YY5yDBkKyaE z$J4vYH$pH)OcS@txD?-KRdql1cH$N<#xV;333XHncNpA|A+6P-3A>0&p`yhHU%rVg zTIE~sJ?Xvp-B?=|k2GQ`oOHg73gaS>Gk)*4m*_uNCu5HmMM|AcVg=)i0(Db@Cyf}A z#u;uKc3*LNmRrUIbO^8R&> zK~3+3?vfTvb#ifk^S0r@;1WYEVPR-wY^XDID-P`;ICgs8cV&+;-&V@`8KDNReBSyE zOxo-Iy6!#=jF8^d#Sr~4Roe7%vGvCIeW)+$l(xnEK6`!ipc;Z0+uVWTpTU0sTEcps z3l)-9g7oE=D%hW3gFPTFR?E`7A-gji}{Zsf8H^y;2{V3YKH6G6Ht-Mx}iX4>KX{+>`*i7l8 z_aO1C{^l5_;eie2^^eaUA2U>R!6p1@OKXdaRvxDOpP*rXG z+|@#{z55loOE_~wrRMREiOG|Kh8syRk?sITcomah=P%An^w$i0P};M}?i~C(wFZNA;nR$) zqjs0W`nZ01>PXMjL~jw2CPgpWrtI<;AgaNmENx5bp;{X!3o^uoE4SF^kK-4BEx@NH zQs~|2dXBn(j7P@8AtgmSl!{;J{9-c6rMkGOH4#0&D9XDbBeIKrr{qiO!>)GAr&OE% zwEO#7fm*e!r(JMNkr2a_rYxb=e|sDm*$Ds@0|sy1WbIcReTeo@M84lwwX^bh6bi|M zrkdgiBxx{NNscEqO-;xk0(ccw%YduGF5h@TA;MzY>J7s>Agj5UDo9vh>9q8lqaxL| zy<0c-Cfh=c126}Dc&D>N<+?$sA}Jdp=2+Y`PqcR+KytcJ2cM#y#VSndOl$yqhiK-H zYB0?`nSWM;`M-s&>di^nS^%lr_mqaUp01l)}(vNp+he!Fm7eWuS;7D5r65M%qLIb0OlGgQmBq;75R zjhmcLT_dufzogs-7Ez+)E^V_Y~Ojx>w95h)Y7cfBn;2Aau_H4yD>1nZ}^jpMBxXkt4v@BM0 znFO7^@^PmXOG);;deOIWTLXG_VQLc?Sh*RM&NRqEz(w)oVs*2jvb@xQJ~>J4 zo>L9t$#v>%uC}Lk?ba%L%#hPoteEXrMg#WC$YN`xPI=#&6h<}=bi}7Q;eO13B2{A3 zVf#V7L`!#7-oP=rgh-d(cVt|@H z<*R{`Ov{XxXo`V%*i(j3hBW`b7=V%HtBYc6{+sAOC*9SZO9`KUN2x|~37GU-^YHA2L z>WL+_)FQ-WIX($G?P!GSZ?Lej-e18d(5>(Us;V(;Ub*HQGeScK`Y{M~yGSm0Hskj? zog*5r6{2}B=vZ|#U(;bIQ=KWV!XhQGyOXVL_|GEk`|ouQQ4?&gFm<>g^&?OiO0L%h_)pcWhv~`j9Y$qqq?amaFHs~M8|o7Sy-K_G zkL{YuR`;d*zvHeV&Q`m8S~FNR#&+!280y5mQwHR5F+ZmH>t{G-tZ)4M{5vxd1qGj% zM29Lw+aG@!c_W|wWz|G^^8(uKuSo#W+1uCkJ*a@fT1xD9c9`SoIo#`Tci~1Re4c1w z%Zl{vH}a|Bm+vNzjo)@Z%segH=~;`UA;zPYJ(TmlaGw3Xp1cVvAZhR&mffXGFW;19 zY;A?zIy(=I(Z5^!t$93P?~EbSA`sssMQ)AGE7Lf<_V&k@hbraD)F0`c(N;awUFmV} zgUXY}*pF5tBZq^QhV&mCAPHm~YX}g4h*oRb77~k8IUwh-TA!iua5IKV&SsJzb)+Vz^tZOc(XUUMhUzBbl#r`f-d(5v#h9 z1-J#V{8-7OA9*W-o8amrSwgIcoQHT55rPN^CjGt#0pQ{^FAWp|T#~Ub~^tT%%I#&IY(C8qQ8rVPnpxbb5v?#B*Is7ajXOYXB8j>)(uwc}XkAC>q8do+akcSXw1L+F%iR-m zVk_YAsz#MAh2C3~W{kSG=&#q@2b3_&6Sq=kYW3dI6x{U8#m`s^zP(0jNNJPIG)d2u zF$X5+p0tZTpA&uLgmLIKn-}FA z-2{Bomr(01@-KnGaNl0I6_j%s94<0=jPzB@#`RO|DnBmTD(;9MtHOi>d3XogA$g=e3@n>sz#9Axr?X-a){tU5b>~h;e4g zF>5EC*R+^}A{n<`Rw1xw$B}o-+6-6itKb;lU*Cmqr?b?-8*#l1)oJWsV1x zW|sEFiH<7sk2rbDv~IjYE+BvQzrE6m@acq-j=Y-f)BUrX^pK@ASK*7?R=;L1&PP*W zYK`_vYdVnHf&I%OQ9@2|z&FSU_Oq?QZE941Ra8%_D5vbCG{#ZUA)w!=iYT6UMgfy5 ze5R4QU&BO&G@3>~YxyySk(QI-c8v;H5VkY_J&@eMyXVqil-u4PvB6YgO^$$i z{gA0NW-E19%)`ww5&Eje{h=_gn#FY&0A8p|(1_@O0O~Gc%UNr zi;-*+l6P_9SH^L$*09|=40_f{FctkW#B#Y%F3Z+%J5r6d3-ug**`At8*Js9K2&Zpx zu#XMLE~FQ;Pr#4;DoEdDa!XKxN^`5WV<*cP=@Z1JR>+`@v71aCp_l5T5+am;gyuO6 z4sk>Em(4Eh=1>re^Ur7Bt)xPyN}*oq1JS#DRMKZVc=ln89NfwYpUKbQPCwkIt6#W! zit*#CC;a8XGz$$nu`Y`l4WXb)7AASMb^>oywL5p8>+Y|utE<7%(7tQ2>>cJrRt+1+GO zE8Fmu{LGZ$J0pOj1| zF?f38Bvlr?Dqs9xF1;`#Gi)vba3#+T@A{WCH)2@Pa@Iwcr$yd9>FYWr_Qr8Tq~yQ9 z{X0Y$TDUyvT29VM<2R7=dT}%L9#u3U#I&D`n-v8r6<&!_jHk+?h$-!&U}5kp#o5dG;s$5&@fghJHzp5ZRUfng2|m;_)V(J2 zPAevzP|=4Vh<3-8da#H#>d|(_DZ!C){Wq#cdj}iSZA&6tWrt*+BSC-V3aCy^*V_e2 z8{%Jr;Epy9MzIUEWvW4bw26L-HX!Zhyrpw}y{?7Ym5W!ORJ~m=#g;!t34$O%D+2Bo z!8c{!jAu0X{T*{Uc-~x$>%`tg@{`Ie;m%*x9t?yhdNE_T<&G&dyHv$_;5ot1yB^{D zS@@K8lQFsnoNCmLDIKD=cDU?tkng?X8D*J!4rKctGaNKv_G>p!^zNXX_8#cqb4{|{ z3Dq;UagKIuz5KpNKU-}6>tLw@kZg@y- zj$h-UmG*6sX7_m=sV7LnkfV9^I=nzuqs@ik|){wV%f5RG4iJXyK>vXNL)90+?p zgb2ap3=sGGvf{M*p3@wMf!L}=*QgZe_Jq9Bd#2G!#1YoLzo0JX;}wFWcI2pr`rgB~ z>K75ouo=dTz?Sh4jE6sZ@jr>Yr~BS?h|eZalglqIq@x516KF`yn0(NHYqFI;M;L)^ zt;%$oZ`w#G4%@0SGgmvx4DxTLBh^U-Da(JTBq>ZQPrW#at*53a$pjjrv5ql~#Ban^7(~v-=uSC35LF;_F zsPUm|TM-b+LHHA(1HUeCgy_jsLdMA)w^;Q6m<=&4vbaeV#riC4)!-7Pz+JIqI{E?O zkJg(QdA0E`0;=F$jrRcnW&iKj+WMO1QsW~T z{I#zM!@Yx@baAH!?Snb>si_2X0w9>WmWYMKzul6)ftears4AS<7c6roCI7 zd!gdq*y4!G*CWSKFLtWG3T<|r&Ku0Aty_ySY;yA7wZ0sm{Kf|AWGc5CT7p_Wa{r0B zXXnxWrqQIBPn&$xCmL?{(WsrOo+H7RQjzN;PIgp5kr0d%H|lrr>C$~u)}AXRoQ>Fu zg_$~r6V4az=!5-?>F!M9JA9lI*<6LU2dHH}ooTD`AGzm6Udyo?-fadtOdQKtwP5RP zH9v`aO;Xx^5B^O3V9jqG~b%j(BtrD><4Os>DNEzO<;rF z9l+iERbOy+`3j5-MxE)KTCefTZI}uB9E?`Y%o18dbHFRj>|0;;R2U+x5e1pUO@j8= zT|d8<4=b9V2~#_y;`Z2AbW#oC7D0yW%ljAMupDNX;0rHG9^h!9?y6XPr5a4ek!1*v zw|jHdqw-hZeoK;ixp?@{%dKLQg=D}p7;g@V7+*%+Y#uT{y((Gb=^sp7)&oz>^W9AZ z5Prw*aYK2=na@kK=2oYkXq!}N+XvOreOy#W z%Dh6Ifzfzr5aa~03Bkp<_+c51;ubmR&c${mu?cfql~{P1WcQoVKd8e_=*w3DOmQz3 zh?f{QJf_Jftd%*s4r*Cdj0BoSr2l$m;oaT8)%E`(|11r#9AayqwcjoAiR@~q9}o^-DcpPR%MpGInI zdN*&)p{ylk=@?A&sH|D-`<9RrZmxDoap3FiB8pzC_wJsm$kaNMQoIY-VQP-1tRFXc zm&Q#(;je4|ZCIIh5~dJyffRT?=>QceiM~QX33B+9NkVh0n!5A>)f?@6B{WlSk zEKg2C`ji5QWz3!WhyMxF`hT6%)jA`>-q42=R8QJu^Yp!q0=?k&b zU!9i3mmhyFPMduZ>q?j`xKjjqQy6l2vfKDdOOG|$NiS_~rp%)jQ~jlQ&Z{ms8Sz&8 zXSnVNe>xW|9S>N|Tisd>22bL>+LSu$RME| zqH+sZaH-xQY4QfS>>TPQ2j6)3eqDBabNeZki$O0DuhXIkmGqAuj)#XIQ&thCl;UCP zflr35ZNh|M+Fp{XuC0iU)m?XSEEWV$CsjOU=I9T>%m{fu!tW4dZfU}=X;3!#HO}}vPZlP zi?~K?&mh}3fwZ$Hnk22IGnIa&C&~bu3=a~x1X9w8W_7ZQ=P`OO&OcS-My#!nx$JZy z1D)bFRNdsp{bmL*;+mOzU7EM|VVEj;Uj=txu_GzR_fi7!gE=?`9&8^fi8+UZjM*Ez zG{Otml91p@I&tW%Qe^M|ujZFid_{E(+ThZwdT-U%`ZhGtuluefJ`p7u9e*zi z>o9JxgP`S`Ld%CL&;%li#+vQVg=}kQM{CD7^6mB}mu})Zn)ekm!rL4dQVjqZ_wqr1 zkDxNbA+hJh;XV-S6hoq8Qx`Mqh#W8CekG>jup41r=ZFTCRjJkvGT%< z{~BTpz?Ht#ZY!ATowJBGK?d*Rs}Vm&SLs4{8ijiWCpOZkx`Y=x6HM1-nCKY%Na(VX zU_q_r46`GDrAR!+V0aQkW0Op2279(pNhUo`MucJKr}f8qLnaWn2zLX(L(@HL(bT<| zm&!zug3TcNA}4pDk)!Egf0r#>#1hH5`<_N4ot@(Ba~GZ*7|RZ`%=#keI8;iJ?7f8n z<*)H^xagm;S8z4HIGjr2neEGU=dV@W;+3b>+~mWJGCi7r@`C7;qNR!@tZxtHNDX6{ zZF*4A{bKLQH&sed&4gXur$_Ri#Wo}R{fa53Bg`(_H*vk+LX8~lnOp3W_gL1zJ@g9= zx0mD2${c&xsqdfLM6vw!bVd=0OhCL-c;P8RWl_7lzk8f!eR)HmsbNC@UzIwq=eo_I zs{4L3=%S<}h-U!ZGyI^>N88L*J;t8a?Ydr%azHu8R+U2K_;~8~UrJpD4FV5GkF~~s z06DNdGR_!&?&^`1zCv%lBO)q>=#VjJz+uc`@_YZedajJAvjV(57r}u~wk#+(!0h@? zSvhQKNtlbke5+0Ylo5O&(*^187v)z#Uf$o(BJEf)V%|K zgjEiT>|R0+7PMlZLQzDSg~J&RpAsYbhgngw`&d-d=ARLaEpD3|CO5#;Yvw+YN}5Gs zJ)5uP(hm{tsXi_KVh2ZKawl%XzLWLUV#}wLzqWrU|GrkcO&8ku*6`hO; zm)XMgW0lQGza)HYI>Bl`6H!P$M);7aHB$mOSK1CSdFF}@6)E>SoGJ7m7DZ2?Lv1xi zt6In?)cMsNjn@^O0>n!>l}7r>*a zLu*=@0QvG7;~}$mamEdf9U7`@z#(d1f9*o;M7;o#9hSce>n?G41-b8C31Z2^s!EZl z2Gh+#_mK&-rr!LENvKn2tO}-qqUFeoy}_Ebj4e(8X{LG*;QIe@?KUHxeP-W&G%=<> z;mPx}u}C$6oKuArcaUip9o&iH;3E1aZVND_h;M3fsebPOXXCcd>qY7a`ep}LIG$R(^%|pVeIV2d zB3J1cZk?^(n>3w{37)@~Q%7m%*{ATI{-k2|NZt1?o?Ib>S_kl1sYsmsQ6u*WB|b=C z_V-vlwf1-Lp|Ho``GaoUl>V%DE^Z(-BC9hi#UBB_7M$?y4%+5LnMTReeJuE9BzP*f zV)*HO4;v&cD@bP+c#3(>qng7_t=7!C=4aXFtlS+;k#^xM^j5(4hf$nM>e1|pYBlNY zAeXY)*-y-8R<0i7sL4G49O^GWNTFUqjiUsIppPikx+-xMYKNFt9eq3}-yet%c=o!# z0I4G4p7@qckCIl}oI{hTQjRiAzZhAUH8(UW`w|z>ZE1p?v6k^Uu&uQ?)DFFn%7@}_ zL3R?z+HulLbLCDorD6e#AdXmoc94j%L@Oe{eiNq)qOFc0OR3DuDDx{EMzgRC9^^VvMvfXVc(l2M&0V_lf?NIQT5V z=VQz~bJjiXUxL znFR5yEmFIg`H6qpKpZ@v(3f;AGBr!d@M0wrH<&CfFK4bBq>V+85JrH&jm_jRrbU<4 zw0)99Z?&cvXp2Lz*Og)p*U9+VP9!NMYgKdRXG~ax%nAwP%8sT#vC>NY5`HT@Uv4ZX z9b=0pYtYeTg1X*Riq%di^6_-5Gn{md7!aU`RF^=D-WxiX@f)~ovDj3$fwy1e_$AsOw9&b!(@50taiq?h zV9kmDyEHE=rJsoZ-Qi48bWH2!&EA%676xP#1Roa-DRo$ss+Z-rdC;|JMkY!R27Lu9 zpvmEl#x_{F$s5hXW+yiJf2%WqB2LcFkkuLt?A=gc5IOnh&R^o*{{XxKg1y2k9cSfR zdy-St0*4NyaBuQRi^U7=OC_d%&}e2klcv1>$j$a3ZclFU)8LoQIzxT>GcWI*GS)$4 z>7jUH|1kv^1sGSPuwJ7pw{;2eDZAlv*$AjQo}&#g~a)*Kyri-sg>TIlM(>V&K^p5ta6s~>OpI{#TY!0 z`!m)hWD%){dz7?K?&K1~VzPp%6Z0mC?s|zm@o!G@qlSljO3hox0zkmP@^&frA^#KS z_m;C&Gd?uR%)^K{>*Lxp?6S?SOcLH#;92IM$J!f;EXj%t9WwUqV!K&EEX?i?K8y!M z(*sdoQT>JxTW0DxIOY-11lrwOb3S5NPZWygB?(QaGq>_(?5|5O*CN?`VdA#8PTp&w zxow?FXr%7gC^C@XY_-DM$m+*U4&jIACu@fshxp(e7SlduB~z-?xW~Nh;Fk+_-Jb9* z%}qlR1?Rf40{S?nZxC*-aH|zDKiCK&hjdRWL0I2ol%EcmHp4_1OuMP7)HUyldWg@m zA$U=P2dED9ubjnRMrxU)ovIZkv`>UX+VHQ)#$HbmZIiV(41-$T|M>iJJA#KG9r0lB z!dT2G!EU)B7Q!AjPD;LFNR0eaZnVrK6AA7j9TSZ8Wv;6L?idt$>*#@6Q-C9X5LHcm zBk*m&AmUk7>1vj^%#7{IajYSQj=XU^mBL~ z=MDID;+#~xMF2lppW@2LRm1j;8>~IW4e&-HCxkSYba-geGebIxJ|))*9`ZTY(nJkr zCd;7@)7OYOWU7(oEp6vWjps@a8G+b79bb=ia!5Ue>Vl`hA&$9iGcqd2E}04F$jRrE zbNc$cbxkVUx_XK#p=t1*&jIp1h5$Du9>qsOaFSpODzikAWhtlCeWbg;L1ux8H#@AUhD<|LGd|7J_s1x4hs~YW{tsi<2z` zSGVe>hypJTBsRNU@j~COhMOF;M%-6NYM%BkN1y!OKBOQi=0L zU%l|Ld?!KcxK7xWdZ8g(d?;U@bo}Di1B**a)QSh%sXo3gB49DszJ19YLcyK$MPJde zET262aG}m&NI$*pfRqt)rM615-5hjkurOPhbysa{i#=&E6BQljsJWV@PrFn5^^bRF zEn8ko$IuuZ4*EmQh7ih*B=3VpA8-5m+x7YMo8R?$4zEN%s60fis1tRLK9jA?imGLV zv3LDNJ4D2!IodROVAocF&9hUjoG>O5osf9`(T{?AoIn3zsP;0)TJW2s)Ph0gni3|q z+05?6^mRCj?UqQ){156gX0&>wwgPn8aiiV{FFT-m%13}lj4qolaRve}K!N4dtoH>o z-|Cx$9EQ@V0(lP@u&x}G*G0|>ohlH331(k5$e#Rtzb!lI>ohugPgKUXlB;J{TFreaHxn) zeYLOL?*rI*O@BNKGgcS2HBSZft8eS*x9bABOFCRn?Lm$WvzIE< zkeqBFDv{5izOaEw%HAuXCI7Ai7OqmU(|$5OQ1(sMTfZ7VdeGP_oi9sj&?mJxp#v+# z``Y(3VIChGPi&>8Q`kuG!m$0Q;{wqZitkOi=`5MK2Y>cm081hq0~!SD1q1$Gi_XMo z2^}k%e$D#V^v>=`OP;6$#PHMITFddcWN!CEJ`%D;!Y8(JHfh8#9LL#Cv0cXY+0>Qe z?i@Z0o@cZB6rFQV@2!KBok^*Vx#wjPSBl`KF~hHEh6+M2z?gvO;EN5v)roWs zsK8js%jtN``MI=ZX+Na}o!!^iV4k&!e5JZCbPY!M$h9|TVylqRbX{h;_sU1GSPgVx zSP<$`3${=*Zt@ybCjgJNvJ`5%B!_*u5t-Z9*t!B60_om`ZBZL_yyKCNhbz#qc`sYf zmprcr*WLTI zDh+ttP60u+3#>uOnebjYJZlR0-8bD$LHk%Y2cQgZLR+9-!P^6na=A66zE z-xd-w7Wc_4d7y}iEcC}$mASS4uE4h+?oL!nm`UAoS~hW!@a?(OI3>2XrOvt>q2}x| z7tiNN^$|yV#WNtQsbxRQJVK>gDe5LTslA-73ZB&09zvNWL(-53f9686BUiS`x8edV{=8C zTT`8UQ*Y#zjkP)4LOev(QD_lYei5-?$GWgkd~vT8)mOsNlVD0mPkZOgUy>6W_iLLI z47nqX+DyUMXFqp!cAFB!?a}G$&*gpzh)#v!^6Fcc`+9HeHPtB)$sMk=DR&sYKj2b8 zk$2yy7D=4{7g@sce{EQ&rp|vXRvGsPU!1i}>3Mll5r5(^?r3N;*rxg1=^o?gj(1{x zaZjGE|L~=mpw?k5MlW^{MW}k2H?vY1%@qsPDsR8b5kvFq7M8!i2K`JR^$rjDdt#Xe z&%L|q+IL6C>+0h?-qcD9CkjL{ zLjv>>pSe;3>mQ^aDt{9b3nVd35fWH+KPddLJiq^Y}Ms;w|P`k)U@?5val`aoe+a%uySY}KR#D`<}hbtgxxM~yz@6J*> zPJ6Cn`~%QvmMlvRoB6=MLc5FjcjG>EtD4O*^EyzAKWyzTIM~%pOs(zH(;Gebkbwh`fnYH z=o9u^ca&XhZ_%CD4OY~2uyU&vhRXYJ1kYpmlj7~u2i5G?E|Dm5tC;|X1fdi5edYI* ztlm`Uz*Bjw^-y}!Jk&jBp%XP#1hsDW%@|odA$W||T>ao*Hy_Xs5$(F+-g7q}TZ-F- zbb5>WmiOw_hVfCvzswORK$NVh{MBCI5Fc<-&K+7~>GFN5W{0cLkWv>~*vYxsfn|)& z{~_$Jg5vn2e_eQRhd>}f0|a+>cXxLSu7fkUySuwH_%LXKySoP+f+uJKB)>EN^VWW= zcGcc>F1q@1F1n{{eb@S|=Q*YkdLX>~&SjIp_>(BNxTffnp!bsdscZ1gY?JV-&I3Np zu071Nk47z$92*|$o9bJo@AX`s|6+g4rdsk4%fx~N@r_1;LbU17$4&g46w$a&8j4Do z3KKnf`h?1C0T)Ua!{IqY+Q1!NbR%wB)At+5FDq}Y++;<#2-Wf2eWhU%%7rm@Rx|Ij zdGm1a)5BuR&(@1O5F$8)4oXk8C*B7|+Tn#7>xM^#`yquX!qSkn2v7xep2Nj1m8X;h znG5sa@)|*I!zVjZd2=Op1{Ng8j>anxln98QRV#+^O~KLL%g*}6>m8jy-98#>nzMm9p$p3 zsbYLZGmYFiG0E`SpB zlaj|v;E5aX&pR@v3Q6fG6C(6Y*%gk-!uj9h(2p&VUs2jN=rax1D(=DH9sFU1>FZnb zbD-C$EZcR3qNP@}EHS{KAc%6!$^s(*-Q(iS<4w(vz&wUz&c@HrYo%Pth&Lhk`+#

tu`gM*qhur>3i%OILnD|8cY`pbJfz! zN`KFG!WANK^0;zAhP96qzX2_8X{UbsA^M4D?p)EMO*5QsU? z_Z!z@D*xS&86ddv^kbJ8@8%#1xkFtb7S)IL5ao6+NR-~%f2)>T02UPd@Z?&r9RC<@ zOfwEo$2lrsl|1;(Gi^>6X;MdKsw9z06|Wwl6DG+1Y}HgE$DcEAQ#^$T3g(LHD>0BX z7WiI&=WvQ=!#S4-pixGk=>{CeT7bQcedg90aZZl;=I8I^K7W%rH^WWjg({~D*PD+* zY}rya9p+cm*K*W);F%NeoLuIS3?9=UkmyuAXO&GIxQ}%lV1I7NSH7ITGOin$uAV`d z!5u9_gUwxX)KT8QlRTHMf7ES+XR7atJCW^wc?iv-)LDAh846Mayna3E2WM3hSxTWH zlfH~L!z(L!^cLdZ*$0oEc^6$=M*6GSwwKD4NbS163*)Y5d!yPOr_9>gzAR41v-aX9!wZLb~VU=W6Nb*BX zy(YfKhrv*Kq6F_8ZW^vZ6D>HU<(^{yZ3-O_XKJAqWK<7aKE?gwHswK^iH5XTkp7-n ze88CpBYnA01q)o zBp2yfNmF}1*goCHzpaqPQS@lRgU9E_N~L*y?iM4R!c($~%Eo}No zsRt=7Dpa@(&@cwCwP!+m@Id$cc%CW*qKh*lK(K6=r%J^en0ibEoq> zMt=GPDZnDY?Ab;We@La79RdDPu-Yz;_9|=bbr|wR!v=Qyj2VB#fibGKPS8F!gP1nc ziIKxZp1>*rIok{c2F$f^ZlqG-Q^$n!d(%D2Yz5}MMuKPkuNy@>+o{a683JXm!|m7` zAyhpG`4wI2Tw9mr5dVipeTOxDT1sX8))vn>Xe{`ubImcK;+QslxyRM*nyQN zt#p84(YKK_Hy&r4w#SBx;V}HmIQVAO7ZTZNlCc>$x}k^E-~Mbg=4NhuD<^y$)MHZMpNI#%QQzW7J4iiLqa3)aEdnhppV&8 z$P^ayiosH%BgjFf|BWX<7&YZKHQZpEEU> z(=F}2{zVkf*=je8GF{=YDl08WN*Dvu0a`G{fPe4g_!Tm|1{M|#bQ}+YO*C6gR?t%f zWInsO`KJ7Slk@7>;pNVIMx{v+Srp+8X68$FV@R{CvbUN5m9_ORGc(d{{<<~Tmxs1- z;qMmL%$437Luq%)j#$j#FuBqZmuJXJSFq^0&Dy+8H$K}8!mam1_==$qFNJ^hl>Ipc2}fY@-xKti;6zRjESRCB=a(gEK? zT1+F6-bBBf*3*!AsI%9T=^`P+nej@1`0(ts#iww87btIP?2EjQGL1cyBC=-gx@^oO zI7uI8sqBpSt`TkFOri(u#!Yf85YJ;{8`~S^*N`fS+%FCAHT#Z?XdxQB8>*~NqB<83 zRgd=2K+90mEO{WaBH*I((uNk+nO;FP_~qg7S1_YDnTSN8PF_61WO6VQI*i2htDe$A zixa4Rav`LmH-N*tF8@#wx$WurmJ)0^PXu$%Me9P9e9=Tk(CeP(1SMqSBcN()2ocvR zppsF2!JA#3!_JrBptUanYE$Q9CRrp5LxGYDZj;`qw_nB@KQyNyo}4lj7ccz_ zikYy<=zcHjH(5DDDflN~POkrknHIl|WGAJf7ps_CFC8Ac!T<1w@IkQQz2hAQT@kBi z>5jTTVo~E>)%v!I?6ux+rOkafn(ScP)en z=6cL?S;!nRyQRjlm{*fmcP;z0xTb4c%S!LPZB7}e+FmOma^!)Utcg+NIU4LvY3J~k&1sBBXG3G(n z5A}vb4E1%tN<&?|Yq-LE=5Q-Bm8JBiZl#27*VD$26I>P1CgG3cqQ>ti>|KH~kZ?ci ze_JCwfBSe8`}3yqKJzIc=p;fQD7%*}C2wX4Ym5RNic#}njRVC-0r zcDTu}Ys+@eg_F-*LtHEYiVqF^&&=H1c*n*d#}p&OoN=^jDeXp2$bO<#{jkk9ND6Q> z{T^?;hfU{@4m`};tM@(UudMdN+vUAr@`l& zxhLF|NM@&!(k64qb_BAn!GNTPmg-YmhgsJcd17XHIpFs95j7S=gN)tajFY5Ov}uiQ zu%5{|t!F<=>VR5NNbZrd^8rHcROu^tA6l^rOZqvQYJ7hkIwZUhOGkgS+fVh@J~3J2 zD9+H`VVm$Hzx;ga)WLf-`^c1wP29NdG0@B}jo1^kp=RZciOLW;7w}Z_^pd5**EEz{ z|46ju@444Z=Rlqb9k8yr@eSY$qSt2$qh>*p-kk^-6^ z1lbgkyKN6FpD3K;etm{bOM3)v_%zqdz$S#6Sxm?dvOX1R5LFrK8N|cpw^B>OR)oFh zxzvx(p6Jss_de0!%Y(F!0V`gcIdYr~H3j0C`bH#T+JYEoWI0+(CU7XO ziTtyo-4XYm+Ls>W^n!D57B2C8nlPmD>4JC{SGmA5m#-gIZX+p1zgPz9qufM>mYJ3^ zsji=;&1^&7L>dmRyjC{AzQ8&1%Lu{~FVnl;#fOT_)nne$_lkYiJ;=EEeCyj-BCFVB zB`2gY!KmXl_tHsZ+Zh~q3s(1AN<9j19X3k)V^zsAc6H)1={tP9PSe1SgfhDRA3w9L zb|A`79=xwj%p?`+*fBN&Ee?ALyjEMHCOUd8~IEL)f|r zd|5_L9=IWdvYOUx~P9vCmiK24y6#`wy3n zOG>Q14DgXi7+Ym0{s(L6@QoU$b`|=1#^mr}}!NxqUKOm;@Jg3yT zxTOU{Ujiu6%Lvt!&q=NQmc`LxtL4QZ9l$JtV1^9U84af`48Xc~bwRi`%v`{hr)05l zB~S8eSh@IW^1&BSqKKG05lFgtK3A9R+!&E=FQVd2sSG*$?LM;a9N^3wc8cW9H+&Jo zkUymBfqsBLK{N}GP+l4Bwk_x=Fn{FwVcaqr7+x)E{M~#f&wj`6HMKYA zg(@#JCjcC0`oYpsnLCYp_I{2kg)2u4#ofu*H=RD8 zsIzn+gOi|A99k=_r6HU0=^)vYU6VW z8x?O`zf_K#d61VOI__0ugvi=or-yxuqCW0XWGAJ0;C@}BYf|z+;36R4q$pES<5b>5 zuxeb5`u*0nMz9|RxKoXFnfx<{wsVlf%8!oJYcc_)^mb|>Tzb11MErAN2NLh&2S}8P zMthz3b{J_8j2yw?qmCLtGTyxUqAX!{yxTc*7oRQt~+ z!)yExG3SAwp^>nSW`#V_Mp=n|uW>c0Lp;2$mJn*)QLfbdww+7!VVlmShwLus-(5u^ z;?rc66{{*M|4hwDnV%2qzRLCPm^L(#<|HCYK$aO6+oFt+n%-M-XA9?k6U8{rT2?eL z@aEVNUd~}U;%by*)ze7(1+lrNU18HXokb z$;MVJr7HrbN=iO9=bRtKphjTDq$j#+%cF`2th_lcvg&(GIlEzG)&W=Ru7Vp0Zhqa^ zrJp31q1E8It?nm`aJp|L?Y0e?drcC3$abf@zMGylWTHhlSM7DA@4K_1y%o)zceV-g z{h-)dKTKJmY6*oP46xVuQ-OQdLsJR%Kq$3a`idXs0sojhLsx0!7_xIDqZOoad35O#oqMzOB}NB- z{aML5lHKc(BJ2GGJl85*;*FZ?%i!V_o$p;TB~nc1lKoV{2{o7;}6I|L3RdHJIG{G1LK3_3zt=VmgzCLk;* zGk1x<^y9MsR=#XS_2j$*5!l*<413b{`D`$^d zZ}7I#TGtt246t1~IgyL($=ymu$hm;`YSKfxTxNGvG+ zNC=d?SV|0wY?&_ge_g>MFMaY?Qj+u8nbem^tAj=OA0f@eW&fiSGlXw^Ua|9z=8=Rf zX0ZRAo#p(W6+pev_Aq9B-qBE|ZB7YH11$2f_K9fxxAqov@b0sX0Zm){vDtmNTWvI3 zB)v$%58Pb0QBE1sG)XehKR#C953+G%B?q8f;`2?=vWp|RY!`~zHt{%2NRXUSJlhyH zSAmUq^W|9RN8mrcr9J6JPkPI?Y=-H01;A6o$Fz2liVl33ea?k~j&68mzP~RQ5?x0` z@Oh#9aW?@PF@1{}fR@^GG#Ey~qx)3C$&JJH?kk+#B(+^Jk3RLrJd!*DI5`7L${VIc zgxO@FE{Fx{x(>bQPGyuTN53p06#e5}C;Sg^NN!yKA_ds8(Xsu%sKoy@x}$|>gvn@8kV*D_CS9_IX=+GMh=(QjCc zM@9woUAV6dF6Zo-gHsy(n={12BYu$x0{`gNUA$$ev&`_ee5eXTO)GV|Hxv67hrjr4 zPXTPh2r1vdY_qolh|EOZm1Ku(ORGywm*9Bu3paTlU~nr=EcZGqXc2@bWF?!ZeSk%? zTRIdI9a}|Ic{ncwY5c5J`=(#2HdZ}h78X>sQOSaovYL8dT=|RlagU_C-eIAjWWw~P zehv0wZ*#x|^U=V!!H-K9>lIIYG#(>gy!jx3wL4~`Rz1-SGHXVj%%@Qv+tO)6h%%%} zC{p3ZOlB91Bk6n}%WH+moH8F^RTEplaMSEoK=R80uSlZJGlw3UN7VY*2*}S;jVnXlI^=J|fZNGJ08MK_ znudb0zT%~&VpX0GQhiWRAIeDmt`a;AfLq2n_^H@gi6&c;`EckBhsO-d?gZLaZFH1o zl|~8Hzel3nr^aSBJWx0x4}q}ZNzuljibMp<@k-ZEXGU%}v62Nwo~yf-dl-Cr%&PbZ z=fGP1{@h=AWFI5Toou|!i4Sipe_tQ?61_maCZ?gDs=We(WVsm;Cd4TXn@%RQ>QNmn znuLizl<<#RU@HqQ^|Dezgwwrvhg4AJU&Z)?8HxbLj^o%h0W&xQ6y zO0Ql^rAR8(muwoe)+)=9n>NIDHlO7G04#`P#IR25EM-o0ZeatFL7;XkV-R=THszYM-$Z?_<}LMD>9LLmdP< z?Y!;pck>+_N14|+gt(Yzv~@?7-&Wk3pk^jQXSWq8=-^vS)9u^1IZ`D%Axn894KIVm zNcJ_+&vV%p^18C2U2BzIz>kuK72K1%_ZDXyKl(ks}Tc4)Z-glI( z!@h5w85g7Dj=!+B=JwVRQ}FR5S*!{jSQ}p<7rGA(I?+klrpzgvfqD|xa?!Jj6E@i7 zT79Ix9BRP?I9C&k?^oAxo|>22Ha&!WE_L0S3mh?UeNzBp(YhzUBugqwmtFRHe=^aM22F8jr**rYQG$)-Q9Mi_!3j=P9r z9SCGa@8Q%J!k=RO(c~fG5L#e6=Ppm?a_2r@w-!DUtKd_`L6R`D(LrDUfAg$c7GRaT>EBgK%ITUFB_z$^iaNMMO1}v;H?h3t5EsY z=8fv?Qr|m~Dd;gns<_f|#D?M9vChRz9pA%;f|z1k?40->^Hn6+S=l_i1~Q*s(w`O- z*ZcRARb%y-NcoA07<%hw8a4;vHZ2kS1erFJqVH(vzo-oYw{xP&ffwErO*?j}-f3|C zJx7YxTGZ)h1L79V!amUX>m+?*ioXHD4_qUM14N%qNdPm@YR|7szj&nFX-H(mxrljl z|8W|@2mTof#1tWr!Qg`b0H)v2NT}Wh!U$tDbqh{Ui^F{^3KaO{rOBbHzR%52(+)EM z8D~RoPbn`rv^c|Xl^p+jieotz0XeGZr@ZN-7jp00m>OoFAhK!fL9d$Jfr{cwUYjq08&!@=?CDj7QNyN8WI;pE z;TICw*8VyQ1}s2k{>cfB4NBvZSDk^cy>UD@T;oI@{$pHD8UlG^RdFneH&S9fmqy0b z+ls_OT$o}#2vV?p62J@Mmend98CccT!V8(^vA#YV(f$G z#%_9;sY0n_J1-kVJpFgkuk*VIf475A?siQ@lX-+})_XZHf7@JZJoQ!3bV>DWy{1a( zbeB{(h<|c!?@%na(YjZhxTSQ;)g6epn1t%-rq=AAT^fU)`pA7%P>PfYiK&Uus6ZQ} zcON_vM9|JOtQ=IIsfx&R7-(%0T05+z@t=Ut3Zdimadf~mH1U4Q`I4wImbQ2AGv=l~ zXAidCW&CqVI;La!P+}hq_jqBkdtPTl1{2em5&jy$u_;aov;f zhg!0U6kSfmXsLN+xS~Msvwa@2oTa>!%~Xw=n+k!+Ol}?R4Sg8`!~P$D)*kDWA2xZL!o>b=A|~cO1aJ0>MM2tJ z?ke(Po~f)rmxcsQkmftD#6;WG=oMSgk&8GBa!2+Ia4ou#@O~sT+h?J{8GMH^CAI)G#e2G{>sK6? z8a#uBo-1zOPY(8ZQvfa;PAR!I_Drb=+*}ARRP28>hg_G-d>%)z$N~QnDkhY-Mg&)KiYLTJpF&aew_FhaTCA`V*Le^40^H!^~ZEM6nS-# zPm$%n)3pzc6N%s(fX#1wVH^EWDd@rox?YS5253qVR4LqYWlQ ztt6wO9tT;2!IQRs$kNTFs(u&e8=UQz{gm_N7k>GQAm;q>L#niByMcluVUcZ$nhO5z zX!r!!#$+EOm*i}~MG{D`B1Nk>0=008|4kmj?6Rtj2zscQl&9|{xuSFw{HJ$DS|u}W zDGVxjB>SDTzKr}@=IK29n7!THxE)vEE&5I0)CsTTQ4M6>$#;tQN>jrk#EIgTirdh+ z+1Z%1?4@o_S~rV^D|l-FO!=Oz?fw{9!u${vCcN6A_yL@jgH}7>D~_e-Ukb;)+4E0XzTB z`zVWyueYw7jVd2}UL%c1!g8yWG6uB<4^Gx#sNX0Uw> zmNN>c(ev$rxq9L)YNYdm%i;OU?2-`f;!@!YO}MhG>3>u2{I4_C|NpM*n>@lq9V??G z2XNoInBiY7#9%xQZCQs}u9kd2KP#Ew8(P~jc}m2ex@Wjx5nAc#WQO#dYdVEzA*2B> z4YK|ukWoc$tapp3z~bUR2cKVv=wH`#qw6uW=$PKse(@OxKl7JtDNg51SUb_p`iljJ zWc_ZDDI+2C8W&kjeqDsmw~UMmYLtjJ?*@^NS0LswKRM{`fQFts+!@nKNXv~F45_QQ zqgqz?6jgrs_(c)UJZ39LtE^i&T>ajN9Zf9>t+*$45R?O33CSK z_>^57MECPw2W|R9``&`AsLgV(na1JiO@0MO+?U?uL8^X?DXbgkjMW?SQ@u)SOWc5> zD=J}bOI`}GrJ;xNj^a3y2{DHq;OR)>H)m1-;|%#dB`Hwc9Jig^k;5iKTo|X|T`-;~ z2bp%>A_D(~)zzhZ-q#U#cdASzy4-4&)haW+7%U%eZbOYw=%{!_)cnkIn7_BrKtyxBgB{@BsecW}P6FV_vYc&Y$kOn(46YJ;tOQ0M=I2D&L< zX~t~H#?#1;{E1%Gzx?*Q|McXTnBp5I;zaJQqq#}()^QmjmI1COIaE`OrLAvyPllo& zlrVLNsffk2EkgRfyGOOL5v98W=d-J|2dBt}fS)~IQH@YqpoFR|QU zZg{>OMK4rk{YS)8SVzic&H=2ZA>F%j!e9l@!5JWd?cPa^GmpaA9REL}Gst4ZmHlFR z`Pn>G-;CQVh7qady?qAH`Czm2kT^m2FMTLQw{};Zp9(cO!nry07vD9n1g6Q4xf-{2 zT||;J_$=B5IB_PqEb3_?V)BchUiZGb@QhhdY^nkDFvSn?v?h)>tqAQE)*B$>pYo~W zv6)PYTXCjMcWt*uDzQ<61C!@`L3M@6k1t59lEj_V z+0MOv{B^kR-N~m&@nPuW6l|$;tB`uzTAE3SVG(n48Jw%KwC=d{8NIxU>hY+9_tj_J zhkyO%qiqvMqQij3bFu_iRuEGeI6|fVqJi|^iJLdg%`0YKy}#%9i+pzkUCiNc5kZUT z&KsD!pdv7`==D{HI{%&Prm_6L_w!7+6J(UpZk9!-viX9@l)1mas4LVdezd z#mpS!9i^F;O`4r3;WBtIp)WZ^&}S>btJYMU?T9)kspZ@MZKIxo{%?&-dtj;WFH;#W z2gARK4ii2=H};Lxpcs5qY|^O0JxOLAwR;-)ei7cg>u8$aFwoA=yH1T(-wD1LDB1${ z2c--)AdGN)YjKj?vX-J~A4k}`i)9!8C3tHJeya#<*F>fe|7QICTPRh)IAYw%6s73F zI#F@Qu2~~NwF2wJ0Koa2|EJ+5Q8DI3Cusfz z+UOp)UYa2~Kyl*UY6#wUawvZdl0fMV_oI#?R}us7E0s zS%6v@O(QovK4W&fPQTjlrHy{AmW(WilQGpARd8mU&R8PC zja3&ezkL_new_5s2rW=1dkPrWWrp!z6Hk|@*+F(1(yEe@Y*v>^?L#>dc^fg1j2#zQ z00wkYV|Gp!?;ct5khzrF>3)1nfsB4++b`Ttr0M~kX%`$ODxT&&PBrX1Q-(0JE)^h) zPJ+&=QKd&RwU#6&+&*UsfNTi?cUzSKHCmFm^N!P-u*zDbF(i7NyJLxL-s2XLz5?sE zDOd?v0>HY9OmL2IcVUM&mgK1ToeqN-BMG~N?+9`VsuLwStk(_pw|Oc*NYu> zc5Hr#;a;1|$+?aU_@>VyhM~7${p3iHhA{fRL~)2tVg3Q zApbjGa)|e6T-EIWq#tB})Qn=1Vy$ZHly@Yt|@ zkXmV7v-!KAF)D1IIgmlP7W*~3Uq#E$B3ww)lW;xDWS&_-Ix03QDlp|CLN7NKQ1UM; zw=31ke^Tef)Ay+H0lw)p-X`MWursYS965Mxki8VgF8mYy+VIEN&(UK(PkM&%QWqU| z_VK;CNfAT*A5OvZu|l*Duy$}>{hv5Kpk1gbP_;2fUyec&`9){CD5(oYi~Fk|C&^>I z_7%4$A+*_vHr7jxUXBVWQ#;;&a?BPDqci7n0kvpDH2q~=4f8ZHKj663_!mc9&JoAE zfJn-I(~TxmQ$*o#bE{vt+dp=9_BvUqEcI-~FdECiny2tJFYLqDhx5yr;0e?G~aBr!y5P``>`v7uIW3k z?YIE$c#FB=NT556u8&c)J~jQJ(uRekIVsvmGvhCoH*foPgQ(H=-`_u&y6(vwaUciO zp;0B3+bx5Z%omzMA%rQ2e;x3N!<*rj435AvLGpuOCkN|dL}*U!lI}wz*Pm7tC^qSb z*V&Tn{xsv=iFZ)7g|LJ$pHDT}(`^edm9UCKj{!+@h?jZ+V{`Q`9#;5*@pAXihc@*O z5^NS_wt^E~&LXcQ;O5!P@+`_+l#V$u#PMi2li>sTc-={>9r;mppq&5P5}AB?tD%pL zYKn+|9OJU+F=lxz(qki!y6fA(9IVUZBk|b)K#pj2y_PAhvf-@SS>Z_5*@9U*?;tXD zz18g$Rb1-+{#k=sVZD9Zmpf}8Ze$Dp<6frukj0P0HD>l7AS%JA#;_aH-`6HRp^#`csqhual^%WiBnfFuhf4+D zk?Qz|bLBUm9<{bU@}eYcY<;up{#11TV>{+CQ-q!MdW*S1<}dSfLwqEVQxRhk0m1#9H#yYVt)O@$Io7Sfmw>5^6 z{c(LrK7s#6khMlH@ISz1O9AAy9&$_A=KLk*wDe0NrM%dj$OD3RGMSF}H^8vr1m8DN1u9n6Tg#43jf$%m<2w zOxdv^2AHPsi{2K^Ob%vKG0(*ULr?M;Jv73U3|M98i3MN&wP?V=(!NootZY65%vfV zCTKST$s@#R5*fQs5gIN}q~DUFhgeau?3zr5TP2ry^2w-;5-r(5c#EqyIGfWO96@U; zU*)t%wqn1>?1x6lbq{YW{Jjt7q<-kV7dI&Nzp-l z!v>=ax8meXQNZTVi(QKp4OE;iLh)$~S&GocK@jCzzZ2Ro<&JAVBlJNfLfbHk#%Lcq zafev0vO(&$Qc%<+1{oIcIgA_=gs^4m4_PZNuEISNuPQw^EGUgT=}9}B;V>1N{~wQN z^Q*=T^;uG>a$MIg3RS9>WU>(-KRZP(3<(Lf!PG*9X?GSwOst^>b$Np15e_o*;2OvK z!NiYzD=q_=5StI+jb}n1%+XZ96e$HYD;^ViDvm2%&rzW+XoiwX_m_Bz1v~k#5mR#B ziC-X$2cJ`*?_{7~p#b!0HWHZaQoB6&7OLfS!!sqZE3y)}Br)U&JF%7FM(h8#oU9z+ za8d@K8Jj<34LA_@@SXr*Iow2vh=LTai-|cWcaBN==VP*!?4TQ)t+iy~# z;uvzjGpLF)Lqv{b4h<>m;(?MU%;pdkpruIl7<%H#n7CN!V^+B=;GIbvE5n_x3^BZ| zr_l|^-yob0!p`iZYXFL08hJ-ARedE+4L6d2Of$BwR2CUBHCkYX$okk#G`W*n9>y-V zfe?^ld0N~0>{Cs3`=LT+u-dB_`xyDhf#kUBga-?(^I0T2d&h!N5pEL=AF&f*ZXi`g0kJ}n3W}?S;xlsCa<+3NHxwJl- zS_;j{(9TCj7N{D~*?&RHd>E7680VTi{-fu}acX)atzU1tTdy`|Xb+Z=6~gQLW^eiK z20w z`E?T0y9pX5;-Q+g*KXOA+469T1yGrPbLIaP%KyJ6s{eCmkx}V{^f;>RO=QLOSUjax z+ppDC#Rx_yeYm>s6HP$&#`m|Lzu5`;IN6Tytns<4#lGujZPa{z?ZnyMK9ztU3MjN) z?Avg+!@i=)Y*$v2?*d*JjtUW$=7A6`CYg>kgMOA#8iOgS+NoRs@J%}4K`8bQ((L}bOD~lk z6Q?G7;^(SmHl?C|(x&*L@CA_1e3d;W4oJTC=gJ6&!^-*g%J4|TbPr{Tb4>To;Dc}7 zU7a~B;g|vWILR~DTGLRL0(mT@`!7H=8Wt6tMfV?oz=-!Fx6`&Di!;rK0)#=+SN$O2C{S@x zT6;qy`!1zvLWYY)N3o4|uDNMjdY5>}*>V$cGcUMZh6$f*XF%S|(BGB)%2+kR*<0e0 z>sh0A8OrAV1(GIscNW`-c4GOW`4;17Y^LVZE%$j+5}|)&bv4%`7U~f6KzoK=iA?D< z<4|BlcbX@w{hgKS4sE&0KoJs$P%2t>ubP7Mj|%!#s)ca*kQ_Of#_mi%pEx#Rk}LTL+qFMSmo zcy5+lePKpuYpXc?#>ZD|Qp;0cp8Gk?<{% zG)?>|IJmKGV5Q7Kw(A$5XQuol^NNN0Oex=*okmt_x|008`cYdV8A(xXUhn__EE4EW ztram|DZzx5;rPp8nv*^YmEpI%^egFd^jvL$VN+Z`Dy4S_VID18fob~`O<#iqCFlI1 z3T;b(phM zDO=?nYv)h->!5p0*{r;5SpC8?>|d2w1iK}}3dp!lJ*H9sf1RWH_gw99>d91sZs&L( z3G<~k`#QL$7--k}hz0EAua_HT@N*<9e}QGX_;e+G6CBZ%FxaLR>Y68pUQm{zE^%tM zzVC#$Q5XMN@cf9zH*IJ%SMP+sBZZ`Vv-`Q|z4_AN0*#KaQ9*8 zC~v8{*2K^O&ZF4aogoP+0%GNv(hxXpxl0bT^2RmPHtm( zP6eBF#tBJqQI!<`kRw%Ly&1-bRGL>ysIISN3J>qwwC#J8`88m<4b!=EIDbM+;WAcX z*aSl-Pol_Q60B*ycBiZ%=8{MZi(sBY$i}m}&Sj`+8*PZ6YA;@zEnwEfpESUh=DXP} z_dPqsyC)%pP1uzo`f}T~um_!Srzj^SmVR4#6qNw4v#w|NEKdYjLDD@xKN2(>lao2h zD1m`a0GJX>S?x$m*%w?H(HDmm{zzoLiK0&zqlsII^Bc6@WCQ<>)mY0bqES}MkSsWT zxF#nlUQ?QaE4trUidx&oj)O*SUEv<4^^G|9s9Y>GF|Lxqm;Z;bw~T5lY{PVeI}|Mv zJh(f-9Rd`$A}#LjQi@Zc!JXpnQZ!hRLU4C0?yiN>ekb$e%$ix}n=}8CogZ0QYwx`8 ze)fG|m;NuDlN!+Mqz8?PJYu${T3NWca*3jWR8T6X=vtLB-qJz5BWUm%y$`|=wE{! zsaGgJ17~Yr1&T2`*SQ`h0({mdKCjktRBTPcva0s2_v2QAzWnGwzqH(coxnQ&-11|! zSZPfO)V8CLupvdP&TBjJp=TnfGb*t2a#O&!AejP z?Rz{-l@t5+4&rK|^6`+R#*eue)^2owUx;WQA|!cbvk3}-*X(}Mh4n3EZ!iY!0Ecib ze?=zgKockFpGqL*L|5B;;SxF>B_x*6V5bj<>sE*SO86>Iq-6UYpG-erj11&@mu8XA z$5=zTMw9;L`9@-Vp>8^P-#|KVPIjA?k91;&HdHgne4Jp`tKDJ8P-SW?lKCSbxt|f^ zw<02mITlm-*%QuKc^cePEgW_&T#cJ+5ty@>f@y@M48kRZn&%G`IeEfj5HjRmu*FvT zPwVe*0&#X?Z^V(LzHm4D0)-!A8<33Z%)d z@$A!Y+vA$YSjy7g`y>^z-gj&-BX9{kAGx?v%cgotOwyz%pLpxBGO=_CLR}!2gIABcRh`AynR90YH&Y!7UkuLDV6B(QX0Z#y}v6Dhy`hC@)D^6G_7)1zV z<=zjps<%hZ0^aZShyd^FP-6nrS4;^e=0PnsX&mmFWdLospxz-muGcVIJmGp59X7h6 zA{)72C?!CAZWl^v>W+u<2HTHHrGMH{BadzNVIz4LntTIeMsmD*lY(9 zU~m%XdbB2Ih$_H}))<`)-(e&y40SgwPFa1Pz}VyI?46|j{~+9>{$IE2)FnRNApYEN zj+s}XT@SxEaeIF~3I3dIb)S}n2XGy}x3sLSpqq0%Y;o-L?39X_Wp=>G`o((Ap>ydR zH3}~aSdMztaWlvM$x)BO9b$ZDd=0@Dm|kjjgjZLy)fBZuD{>E;-;d=>dcjEr=Uh#R z`%d&GvYm1FT6r0hvtuome+UsPCj|a=?3dHKpz~wS*H)JxgsY-z^M%*t|0^${ln^NZ zRbok!`;a>HkpsdxoaCEC{IDUkZ!ll0}Y@Kf6_Hp z%3CY((%^I7Jx+~tR;n5y6uXtv#P|f-#&Li>I?8B&()rX}zbbo&Q?I8sIrsONzcvDw z6dq3#W77}<(V~v#48VKZz`!R`B|}bt4{k*hDT6JC428xp;bLw|KIWmfgA;Tf8bUSN>1yk1HsLl=0K3psijK1PtaX`m#j6MYH z3p#owq_ekakjhZhU~Xs7MHo|%Z;H(ta$ihmr+p|9tqx{`{;xaf|MpsZK5PFk+uNn2 z8T!5?udRyrDLSZU2(u=&V-E>t8*W=e@0O0?qUK#)7Y{L8yNT0C7&T5pIjkkQ1gsku zV3V0aW%NYNit^co(^2Q>-8X;f{&|vhqgEFpgH>{m4cn&Tpl@svBi7b#SM*i?0kkzV zP5ev=m_!k7XuZ&}e>>4?U;K|je1XikvuR&8Vp@_i*Q#(Qvww8LxPeC`C^OY{J?0f}Qx_|4} z{v32MW67-XO6fjSzZGt$^sWmkLpe}!zv111d$K)d3J0dc4BdXY4$3KE|7ClQ@sA(G ztDwh%gZI`_r_d6=4f|sc)CuK2UZa0Kilia;737+1dH$oymP&iZH}Y5>O2@nv!Y^#T zmz#jjzM0>^Qo#5wk!$MxO@@$$lUkOQPSz^xc0yJ;oB}&Jec+K}D?QSCe1F{aiPwxB z`G;R`Foy)@Vt;Nx{0eOP>96B;8}}dIP%(TbjcN5;qkt&S?(KUa!zPfng9>w0DFr@kx3f}#>~S#N z6Jv82&3y9%851KEuLOxevu}~n%bg1J>@W|W3K>XU;PxiogwSh+$tT3mOD!6|FUwmY z{`2$f&bK;Al_b;}cY2{+Fu`|Bj%V!0uvp?i$r1z6Y*+Z!e=4y}3 z$&h<2o$UtvDKhuZvdK=JERoR;BIxUVEj}}x-FG^y=H3_14+oP@eN;*v{@y@z&$!g( zMrl&38{g3n3gw9L@zgsfe<&n3k>8i?m5p{yd$gV2szNt21TzKhFXnOo{CKKulD9+A z>#M)n63iaylaD;hd6aAF$hTD^@{;*AO5)|P>lf*13Ilh#E0!&tN>JE|MvHJKS(23i zMcwXLALt0Grnyk!IgGYe8KZzLjJTkkwmPY$B*T!N`Nj&R zht3@(X*E5_#eET-6_*&a4t`a1y_*7UNIX7L@umcf>s8{&#oRIH+Z8*|8l{84xVhTV zmiMGPdLQab3g0x%-ayQ?Gy;7tdE z1aD)PaE|{V?}MX`5&mmqg!*9xG|qsoxROGJbl$yMMT{gY-n7hGJ{`1z(o8miDYPGS@!N;U-C1se&Wt~5j{Tp3&V>;?Kzuq{eZ_k za>-4{==%0Y`}2LZyvfFH>z~nj?)?Ns(-LPeHQ$VZuG&7#UiE^OpYS*^F2wI zM8JoLts?fh^$Yv?3EBAmM*pJkK0uH#pIRT4*6VsB9|AVB$C9$A|6?uYgM&#F>CEqF z{eNSzjPlt9?C=`==URHtPE#g2t4BC_0LFEqtkw!8!zBS;pJf{dclmf5q%)=`qiXFm zC1%vlHmMy{aAY(w6l)7gT^6i7@H#nxh-2lc2X-kPbn*d_=4#;$8xl5tfS$mA>0_Ws z!aEoA@&sFQ_ScBv(_Rb48NnZK2HsnBqU}l01cj(Be^dCrY~a9|Oxgs~^+P!gcABK? zN}}N_kP(%Rce7tRJ6V1+mO^Iz57+LtTD>OeG*DrQIO!+ZKRPtAf$b9Z6kbib<@X^~ zLv`y_IcxG^DeoVnrCJ9kqABuZP%wZri$rp;Unir~5B0)T8-s_gy{~$HFJx-p*(`F| z9%>74wn)0-H*Z6|&pw1ncEl|SufeQYJcPa})@dK5uA8nk85*z>eAum=sJ6zTVC0Cl zfFGHqni;FUKk9cCNFGSMseShDa_alucXk)QB4IU8QjZ7xW`Y7%>euZ4+^RPi2IH5< zWs_9&CO%0uAzN?lKSk3QZYp6VjNdn}@h#qOv4P-pLnnCOV_60*N}zw_ddnTd%b{_| zYrUZ&j)5Df$3@A>iLo7&V<~|DGbD>hH=^LfVU7%YDefgX@cv22R|QAO#!?2`ohp8{ zk`f@2^$UmwK8aYX#CDYnuT+vf4b1r(s#?zFLSlE34LVJT^~C!Dx7Du!z9qb(gm&Z zP#>XI@2~}H^LqD?I=gWy8eRaLraa83R+>^D_<|iE9|j+vv5YoVE=co2P_d!mCMZ=G z8kzflSK#9dFLPD5%K^Z~XUGMF2*WKq8UvIANQxkBG_AMZPuSLD>eZ5Dv-Z3tlYG%7 zFh*z9QQ50gFRacVe9oa`4a*)jq2*~6+84#i4eKjJ=Q}!iSNKpIN`z?m{~yao|4S(e z?JLKqv#XB3Q8Z27*_?O0j}VRp1CVzov!j8##^CLp!xOUoH}X?S1P#DRwIwJkTi9D>N`aElxc{RN{G2Zf_y(LNH5eA{HUAr0kz}d2Qc9kR@LXl>r0LhuYFh z7ZMjZ-N<1DXUCb*KPwJ=(U*lYX-NlD4YXK&e3>)-yz5fiyu8jULJ z)m6#8G=C*E7*EF~h~oD9G#I9RX@`=6QN{^1sscZ1tAe+oNDIr5qXb3WSMvTN>A7+_ zwXpT?yEUImb6xw69TlGS_EThDt8kOQQhYod#mTAhaJ7vw9qy*>mi+5z4^V?xj|>8l zwi=EbV>2x%@YfXqUM&U>%U@Qydljm!4e5UR4_T@a57H~6@`@JkQ7Dq={?}LzBC3Nd z*JVi4GNHR9Gl6qYlP=sh$N|-2Xk)Fywf3D9K-HpeX-U&TIz3RiFB>z@wa&3&;w3|z za~PZnk))PV>7J6R@49S|mS+~<*zbM~TRHwrPZl4;!i(qCrfNo3d=*Q>VE4CPc6OYF@;ld z%RbZ-_kCvkcFy@79KeOHWtZ2RuJQ68fWb|E@d;&CZLw9y5e|9LrDIO(q}rsD!BPYC z19j}&5<1mBkYKoUxm8kX>lbIQYdPQ!6tg2aq%D?n(pI7{^NW7^%J$e=s=W03ESC>s zfLp~RVMjFb+$4CrodQRm={psd?baj6e?(kXobSmoqonM)fqD^kYd5@Y*kdtXVQN3O zPUUCUnF?N$(vjq_(^#0h{aQPsF(}OQzsL7(I4D28Ez-Dim&%1 z=&0ieInVRu7y0^rJtfi{#c1tg%w+Mf>|2Av#I%Qx1MT&-Eq{H{dgqNrT}j8yaS0!a z2RNo}o%hlp#AC6c_2}TIYF*4Qx`A8O(c6^ISFrM-`d{xi)bJY$t9V z-*fKq4($2P1O?5SgICYn&EWzg9#e}u!BJ9bR9S?3t1MH6jTURIM+yH{DM<(=JDwvd z`zHl1@cbw_FBlq>UB3l?ZZWcDTm9s;V{7&r(m^QfyqcCA zMb{C7781!VwVA}9DsXO-AO5A!!%ONmJo!!NFqSi9dZ?5l#B+s(;*) zkt*M>ZgSB_bWi72%4@8uw#b75UAe=o2VE@UuM@5I zBJPM@8x;j0q2>H7UC)9|_I^?u68Qt2Z|%j7l|B>XGBBr#YdXiT<5>eqbXr@j}&sp}~0p(@O9TAuRnTq{L}?1GE-AVBH?pb{<| zuKI*U5n9gghQ z`Y&CNx@)O`{S4Lb!N=?&xH1;s^Dn|9SI206MR|&C44Wx-4--N6dp(9k8ln3Gs=Byk zzKtV$G-=Ino-1_+^4njai2POuioMa;>=cRNKH97Xw*$R3-XiN?BWCp&K;M@IJc^v3 z$>A)Knf~*RF9d-Hddp^=R5ZOWQCWQ^N zow`}#gYE2Ze0FU@a_0u={+#x*7~@s7YZ3EMca@riw*6)+H03dwsn4ylR_ub}ZJ{fd z#VvG+qo#=OwW_S3|KjePVlw0Du*~QkMlR;MZ@5mcS2{{# z-A9s=_nXA|5PI~$Wj^;~t@;4@@OP>!6y-i&@PMKdto)x{jTV!GNM^|L1&P3Y>aPPZ z`kxf%D? zA%l(#6wv|sU4wt6-Vsafn}Mt3f6ShWLq4Db*uWUi4Z@Bu<=&|5qW~>dxkPGsP!Sho zSpl$XXJJ&?Doj1Bgyb~l%<@q{A`I2|@5qf)m-~y8Ved!GLfyp+>J+rXT5w^PA?W}^ zMvNp$2e}mR$x0AH?P}-R<=G?R`!cWYtb%jaFIk0~=nFV11FN}?u!x?V$S4I!SuIyx z=yJgkWZ;4}-`A3Qm2I!3b=7^ab5y-(G?ldy;{t@ zkqcNZXB71P>_AVYu`=jN??yyl9fes~SaMM5C7#OkSwtz|#xAP0RnwAUS^!LHj3afc z9IWygDC^@Uq?xG`;9>Qy3SnZ(X(K=Bd;!wz8lbvyjSb!Z6cYiLn zQM8uz@<|wSbf){a$q>blBo5Y?i2;@_PStBW)x$aXf5oBTaEY@zo%I?4I1?A zWHDR+G3;f@Y^wVjmu%b$xwS!^Xy8_*UtiJv?J_c?U>taa7z(tCe)LLdHAEGAd{i5c z{`C5HQrGsm4Ium=b(qukUmkg^$oTX|?^dAWFGoZ4fzD(ZCdrR6jK(S}CEq5G$6!UMa=pA& z4@j8MF(ZnZJS%Z?^f(yL?!)-u2BPa?n7-fMe(KxK9F}`=k-PL((@pHcN3{B3ePfj+ zMlj{N_-oc%4~xYi4~=-LG>>TNstR{|m;VD&6osUEqJnNWB@WP-8okpY)pA4THFH?C zkUTbel-a_T93j0i2wwFIkLvzD7$)V5x*dw&& zE`0KYFA*F^Bj@vfvk?>$`8^zDTc?*}c3-VijTJ)?8S6-0i(E$$|2ETJElIhUCV4)y)1WW$~i=vKA>3X#`Mzs2>Wl$ zLpg>?*_rR!M6}!hmR5kj1d&ua>0?sHa<~bp;WX~WYR_LM$|Lo*MKZRQl6|wr&D?8v zuS46*-^IF_`3D9@qhWdcFC-?7Wbk!!hjd>EBhM8@C!JJUrm+2^On0$hEN|I79!sl9 z&>t=(+%Zy(kb~^~0^@|kHJd1g3~e9J2`T;4LO-b%r0GfP8>N)q1>LdYTbZXy714DP zdxW|>Aq0D?x1-IuNvASb&?bQ8>^;$N@`fU)!ohY2FsiMB0wzaAcIU?+>D5V|z5RTB zDLJGjVedvHB2z89{q!=w8?{dLmXko;y*P>Fpv}(wttcF+hpl*XFTOEA!r-SzKr^of z4(&JhQRI|IX<-w8d7B^ce4%K3>0GY3vs8Yb=me8cgbMC>Dd$I)C(^Xgs_A+ba1^_l zPQu!J7k1&eFv_!*WCJ+`m$swKyz*4GJSIgj3YMvC@3}3t(RinDy2F7M|E`w0e+{ms zmIR287!K?CF?6n6nLR4VJ*xyv0}whW_0q{07t1sVRV8%a2R!7}S)W$JW{!eekgh3h zywcifkeww~oC)ImWVKb#Aa})0Wp0)C^LF*VisGnQ(Zdl*JdJp+n(9Qopj$jf!nTJ% zch)k8B^4*vo#0Mr@YLAKq{H(E8_--=2R)h-hVXt!i}Zl4GMz|uVEZ&H_m%Vv(wXFp z`RQ{GB44q~x(ms3$+R>#-EJM{Va)tRAM?JTdc;{bGmfUH!6g3#TK0UN9}pIKa7C7aYl2F?j6?BTS!x#m5KhzMWk~ojIXq@7EYJwjtdCCaKX{f{Ww^JwI0l)c% zx+~K)P>?G9{mNcWb+9LOkED<4Iq}($-Q*A<`Q*^0zJ{p$lp(R9ps1};(cfgm4G_<& znwEO_fjrK&w4dtPAs#FFC&ai^puR1~$%R=ZJNcS7*NV;~^*Qb?sJZtbPqm2!H90Y- z%S?5jnp~LxcSY1{^d-=cP~0_1Ffko0%_A|hl5LHZkI`Xgw0?l=wN;dX>1u6_3TJjN z=F1IvsNO7$&+{#`@y3HCC}Dg&GI>FdVp{ykKmH~LbtD(}%gcf(VN}U)W|z+&!JudD zP+MA7@c)>$8vX}>DH7?u^Z2#0NzghPI@-l`4_7_?buMhc-Y()Hi>=&pPAg#9SbCEg zbw4*S_j7Y1fPXUFOLduMn7dd|04F$6F<#1L&^m+-kv}xV3venA*rp=-`l^=yWyY}a z_#c1UB3+JJofM|`1DaYSb&$DX=GijYo#S2W_sFjS03Nq84kHm|DHq0=e+MSx)AHIX z^$kg$JU`9Ck#F zXW&9foycqwiTbI>VFT-YmAp2&5iFhV?2=&wA-5jp>!PTo5-n-%R2dXrl0aAyJBWCFiayTY!<&$uV5J?!^q@`p}j5y8V2Jui2_})`BkdvYKgM7aP`Q5oekZICOR@Y8-lzNV%lx`SfijjJ^k*0UaAYTOU} z8XD01w2^ZeY)#dpuQ8kZALR0?y%y#e8=qq(aPxX5$xtd`CMMBbt$Jg>JOb9(zdO9a z!ZFwAd9IcB;q8dh|jb$$5Q25 zppjd{ae^1K80StAKoJ~!M|@%iYAiw*Oh-;`X7AYe+^TU%mvP`1U?uHNgYlzBDhim< z#g)Sx8Gyo>o*+(y2hF5;rPgZ=J;RAV5 z?Z~~*@M)`h_G?`V0?-5?eN4(99lh()*GgQ`*l1MK_xY|A_?|;YnN>b^5_rjw)fU>Y zcF>oe2BSHdX~NKF0NSBwwiPduuoT%?(8V#^ab2rMAT-BJNk2Kqq!a~OVEn5mHad^O z0_sN^njLh2veZenY+p;p4>6K5 zo*df0AE;|zh#}zWD6?Uik_Uyxp)#|v*_YCl3hK-tDUzfCw9xH~L*II=x3|6{ZB`?+ zAK?_Mlht8tai-9~<7H~?%o#JAZ9?_Yomr|e?MDbhx%S<KdPy8o@ zyL2nF;?@@-Lv<1)XsD&gOK995z5aKXqp)M4_(rJg@8bEu1b~cG_UT{i(E6z~OR8_U zQBPp9AwEGWNQc5D>BtsLBB>O{&^=ebSQwSTTa@q*D(O@@kErJ;0rwR}>2eVdSf(gT zjjJ~_s{{fZztPK{tu@!TjMo@la3j|ec3$d;&AN+q?T{o3s`$Yqu}(R-?On~|Ej#dp zibH+v(rFfTHJ8aCeC|J0L=DatXN1Pmp+$OHC87^QfC;C7OY%gYLl1 z{0dvTu1Miak-bCOJ-rW;xGwAy=J~Rz?Ii|gL!C)6G4vg}DPDt#j(#0&g5?3;FsryN zw-ytW+^GFvCncUYj4YS?=`wqs^#ps9UIs6%i)qv392Cd0hxs~e^mkk~$@Rd4l3|Ry za>mLO-%hsJ1O6sdX5Nm6GhIawha)xv&oh=u4uY&@CbU75i7KfFw@ZYqcI}8f)MG4k zb&^@f4#IR*;FZN_iw);bA3lG{34ABWC!t}9;uU~+) zZ;=u#b5APqL{?^IKF%iY(N6wUi1QdN-eQx9I-2yp6T$5DYYWH9p@7^Pr4-)5=jv-y zj7J(ve=7n}k3(J_B-iR5$`7pHdnElnOqB=Ia@AI#@slR#<%`x&EA~RTaFMksqwrTG znhiz7n%9&0LXvJ$m$EB1M_rh~O2_WtAr)%#+_oc4=@0|ZwPu|fei6v|eFf8E=oqwm z0q0=LM3F?M#^t&6cp?b9%(ILXN8=}_Yz zSN)4#{s)FH-x>^*VZT6Lrl|hliuKB2z@@P-w>LR;K+!+`Ik>XXG~rC0@O_COlOOK> zt;~g<)t92XH~3-XcR5SZ53b32%B}rt9ip$Ab^1TrjTExdlu0{n-*vxrb6bkJZ9QrZofSfBO+bisb{?hbu?0L&XeSP?Q zM>`np4kWM_V}W?xV`1MbEF>-@BF?lwejjCcbGd<%A<6FKPs5Sdj7u+ZwADi}pl#n4 zF|u&~n+amVCeA72p?^I6|%BilhoHV7FfBlFF?ga^|O_#nWw3Mn*J- zR^z$MDQbfE2?yX78*2dTn`=5qg7C4gaQc}Wur2O; zLO_Il;3>B7-O!x?sg|7rM|3Nw_lvd# z*$ABdvuo}`Z-Abb)j$-s$FO4uEFSKhx>brtF4cV}LnP>-F%+`_nerMDELsy@!NP%V z(}YlZ@-s_mL1=m$BxY9sZs~dNrM}}Makb7w)A0SXG#%xVvwLHkM5?_Irf1od^UwSP zUrREABif7O-xALap$^@4F&EHfDV*c$*7uF)jMF$cS`Vlvf~Y5Ex@8q~ieb+$UF8Ri zyG<}g?om9|dOX3%QT+s&%$-|O(uZ&ILj(*`WS{(X0-QtqW)?ZqZ%3aq8wKw(sj)DA z$Ma$D)SFJyPieS|(hOu6#Z{Nj(V_sQ&r+?t=MDk^Mm%-sHN~F&0Yg)x z1+cEd5p^X)^~TY;#D23ycvsUv$VVtLMgB0s{q3n!CJz)1FG=ivh`o!)^zuQnY8C&; zih9l6FI2?pK)UW@`@#28zq-vyyr`Ugbgq_+n_;2g@m2nSXaJMpmQsA70pBqN+O*Ao|J*#m2tYG2YMF~mbzDCF^9zf`oS-{?^L%@QK~K#m1V;J3qK$E z|9FgD#Uka=d#Y=L|I> z_vY#b{qo7T7b#CZaqQH%_;QT8@lZ~&>K>%pZkx>9vKw>bZH<(2@9SVnU^jMll>*x}S8D_lrbIB;U5}P%|k{ z(8t6=6doJAs=ft%qJ5}nFH>%I`KHvVz0g(bIQnK_FT|W9(^f#$*6NN}W$N-!+2!Y* z`UZLG2(9rdC>_xagrkDj0ja%VOBc^8Vuk2VO<06dVK6TvW07Un$W zbKbqWP_rVv@2BF`EO)dY4Z|Y09Q>iJ!|8--sY}nFA3@FD#2gTQO`cyK3XFNepx8#_ z0u9n@vBdrs_A*3TAf3Kr>WS${cG+=BM)@-e*v@IK?1&{J@o1IO(ldXno|M`AYhl0J#`E^Uy~#61zF?<=_1!>zX{gFI0W1owu+$djVUq7 zK6|Z|T%Xo@3En|DkTE3xI*eikyx&JWa5Gsd#p0}!5#J}Ln1qB>7uOFxt6uZz@jTFl zmU055@@nwaqSF!r5eYo|@esx)Ts6CfqH2M4ZKyo)U8_io)N6kOGz|dHf@6j zb3zsD`zw9>|F1aue<87@6>i@k-d^>LoI&1I?9*T#Ernqzq7p(>jocE!IKx|7LxB}^K>aTMoa-hwY zIP>;bS$5z=CAp#$-51)HZRW2VS^J?(SSoNVV7ug3NH($mfdaM>KT#3yJZU4iM#>2( z3UptVZJ9D7k?`TtaB6l&i!tPaaPuOC%?}NT{&8QD-7cCcT_HO=4w|}L;CN7!f20Ak zHPEv@EpD&#QRs}xNP4at=Kx}x>-V(q(QlQ_CXRDUSv*vifm(dfWsKtc)0VoM1*0Gz zc$JXgq&g;Cp`+Lb1^mDyUdg7Dq3`3&)}|ln9vqQGToD1 zxh;O7GQ(C2fIIr@?nDJgd#s+nYaMtSAmq9BU@5MTqV&s8QnubYn``s~scb*_=Cc-y zK86#CyB`J#(%c8*?!Jw|))nFmmWCMvUgY;-V7ntPgXnM76uQ>XpH_3Kx0|x=SF!?yt$^ zk>tQVL=a(($;3WDpI0;`a28=inNW3D^iR6QT`l)Jm$dF8h1oV^Ai9*!qp8m>Y&}E3 zwEH08P^j1p0y|IK6F0-@C(H0v2$;mwQm*HSZi%h8PWgAypPiN4?qx&FWa<#yzXZKc4W581k57-MJ^|`nI81^_f}ASb*GJ=@|M^Thv|pQ; z+Jt$phkLy&>mMoxxF%1>GT8tTQV98;g!tEwLTOYbTs~yd-p>aZPnypcrnV(s!mz5z zU4= z-@3m#{#G4dEjX)-7!A&qjmNL+Ix_4KYfcOqiYV$|$KED!2m_Jqc9X?MesQMOvFSpy zhx4QM)L#(`lW@nyf&pqz7hx{BQxku#GPCh$&Y!`2M#6q&q=r>^8@>&u*X{8xEj{E8 zIfcI*UxzqoVHbMoL&2ItKd#@UX8^iW4EjLMaOpgoU;d8dg*>Qr$*WO*AGn22yt!R$ zh=yOQKFHL2X@7Hx>x809dO(Wv5-1h?O-tMrnfLNFRX5ehANbmRnHbrf_^(vEz9-c- z0SR`g@fe)dN)+*it5DzKTSfiCAI!ym5HA3%Y_4N@2P@j(^czUl!1eP*P>E{s6`JNO0UM#PxjIUFc?Va2B8bf6d#>F7jwq}3k^z(f&)U5D76yK1qq7@S% zcYhU@YMInkZ0*);J4vC39(kgS{dLTtNOvK~*P;Wx1TIasRu6Y=(P^3Ay~ft!;FPy^ z>L-1wmKhFckXfj;EKpTDL)D{_vi#w9L;8)yjyjK7phi6l5Sdl0pyLM^-vZ@Gs?v6%5O2cXt<`*XR>II@Jt}1Y?u4 zz6Q-DqlyWdji{*5;m&N|IbMYl=xax*r?;(OjC;u_ zRo@8_KDq{g>#_83VVqF*KqaajCp?txeB`*k5WmYEMjH0R2LBP5R)nou0?hwAB(x0d4S1ad$P6}?u(*93$kolof5)4cwO59M zdlRRg!&-C*0~hW#y27aVI|~Ppo|I>_5Fl5;TGfjAo*aU=6M)sw51k%qTEU2dQ7mEEjN&zEjZxzn_=t1IXenrYl zKT&|8McoE-RLQ!7L4rb*yym}~>U@$Wm8D=?cLjsNecCOrfA6dvl`e00E)U}5>VHP} z_aA<1T=p_l?haNH)-{Qp2S&U~?adW`Q;QpWmdnVWl8**_8vC3+^V>cKc?laLD>FInw}B%g{{ZsJ%N&U^*KILM4p_{>w{q)$WpR>9STP5GW z-`GSLDZF?B>Kgr>#1CG^9_&>@X1 z-WsgVSDR}u^4Pc&*2?O_$%}=nEb$=JKP{d837A{&703BjOx#F#737H&io|uy!K1g3 z4vE`%?U}BV_p5DIvP@*>iyPsbquHNB|7pv)TERfXAE+PWZ2Q?4Xi(RdkAl?FDB54B z(qL8EGf+buWi{+n)-Qyh^$p*O3GE)SR8Ei*(z#F!pRz1LQzl5WnrIe>I))Yr4L`)P(06=fcZjr_b*ekrrQn-h%$odkRYT@0|`8~eSYK;B>{og$)~ z$Xxus?5BuzY8zOxld|%r#Z{J?75PPlY<182j=S8oqP<61kVbKMW&e)akgL5=Djn4i zgAP3MaLFD#jZF&OR}Dz*AVKmR*FtfeLf;L{*}x#enA&*-;cG6zs<`GH$T)aS3!nET z@tErVtbZ!KGTqIhx?AJ=2|a85_%D#{=p=3=W6U7aahT{s$GLKyJ=aPez|bo{P7KA6 zXki+Q=Z^05ZG;T)-cLZ5K0a?51Gn(nv0vpfPom7|SNXIt%*zQ^c1e%O4G`PO|A%rk zgQdwEY4CENpJoB6$R&uYC6UjgnlD=39potN?EUUSa(9if;wB5kd$pC@o+*1e;RP)nn15m}3whoHzh5+T_mW2>fTfpOi1ehpp?3-BfpPC}7v~$~ zN`|XH4JKjMWR~wP{oIwC-4@H;UJ^vhn;%caFZ6q;8;_y5UtasHh3P8#SQSXp1{P1c zHN`Tr@&x|L2fr+OVEKtsU{GtSb}@j>_TLQF_zh_~AxGJP{t$0f@herCa9FDCbRliQ z#{V4VSW@8z-OQgD3e{X~V`Bhcmc^{`=)k*SR7?2h61@~F;qo+#s{=UeV8h!jCBQn5 zOvzTq1(m0%!c1XM^ijbM>6(ey$Q-YYlaCCgj9Al^y7kLF7gL<-O}TG)?ta+GYt7fL zZ*TZm3%A3<=6c9Zi31$IGx3Y6Q(&M_k2Z)ey&d7_5sy|zFSYK(Vo@)U2<68TMXvk^ zc)!ROb>I(2uXji;IboMK&v@=lTQ%oPQReC+r`Qc6j|v9>o?Z;i%1q+DWI9?t*d-n+ zK2p>xJu4s-!yE~%RXdVqEp?7;gBD1G_`_Kl(ay#_n~gz`yjA7#XldV0DztZ+vK@&Y z`~xCLr|boZ_sJDbhliy90VsD)G&SB!|Ld&U;ZiVha_5}Qg(R-$C-1ISaa>A#ds@eN^_poVVNK;zBThMTl*V@7xN3 z5QyIx{A~L4DA>36#80>pWbhcjtM-*bqTSElBx0IQw|t(yH)5h4$$FZ@=!s0%Slrr* z6q1Apkh+Cnb~cYk{aFPrT}3<`D2SOIZ(lMg9;<75D=;5Za_yYJngF=y9)gP!+@Gx#lvB7Xfx zs|UilJQOkM#-7aV4!P29ixOG{HV^ySBzxlXi+v_F#RcOm&ln)uZ|nl0a`^lqWqQ2N zs-D>Ne@Z>y@flT*AMEf`+%0%lF}#Db{5WPCC$xlLMv>bJadna=@z)<5mv!2m$i!Mg z$!;I3)IY(t{|QN$^ad!%;O;>F(Uy^~6LZ!Jvry3f-0UkCfQb*}W383`Yl#N^L2@|2BMB*>= zq-rYKUb))^j(Nkrual_-To@0&yh+11zKTc`I8}(qolM=P)`^TNd;3DkNys1<2G^lrOhi--}X{I1ha;{@U7YG zFzjF&^q1{Bn(3#Bx6)nlPkW>mo!K~msxfP!g@%Hido{nN zqy1q+DKfdp82ZN~z6|70MoR`cGQQMJ*pNO;PL7UbbP$pjB|4&a{r^V#!Vo^2|NG#p z(+z9%plo#w|Dd0x)k|pQw+>*7X3CYdpN%M9;QkXj0w~<+$zkHHrSIR{e*Y$Yk)VKiC z_zIVjZqKC_kd1OVhvPL``EpE-ihl)H5v3?i5Q?$Y zM@{a#FPDJ&tKY4Kl0n3{k;?jpdAA&nCE4Prd`Tf8nyMLsgG5JOkA*mI+A}GP#?hty zBPJb;yFJ&)xmTX02Qcca2kGK$ZJrYwtJDw38^s}K31}!VtLbl$u_5N)pws_{wYLmv ztBvE^IzOG&f3xw4jhHPcReqNbk@#hF3uwRtyv;-wN)U9nE} zt~5An+xN^w{Yd8akwRchN{=bN)U8A~3}1VupV0JMOQZ3c#P2isQDc?&CJX=8pcj&X z(DNuOG5u2E+#?1$!Qd2F8&W1lv-?PNR9KfI$%uA;aNXZAg{e_daa(OvnY3x8sZWO? zVL070%z`sXFE2hyQAk|7(g@iP0CP2tV;%yRg7nURclgEomcI<|MTR{R*J`;5#tQxj z3zgQ=T4=g(?_Dpof5cXQ*J@1Qpc@lQ4V@wMP(7m7Outs4Vjh#`u3JX+K-=C&{)}wb zD(cx80tiYe*1=J0_GS|k@x`~I29>s^2tylEKv_2lH{lQ+e&^ywFK(B_mAswter1^* z*AV{zU2B>=a>w%TuI0Zzl5Eke9_iX>Sy5T+fN0)@vY=rR<2sc{tF+S#aO|F~E#s&N zskIP7^+`W~dth3%(`VpBLJuWUPz)mTR!jA@R~HTJRCQuLcY8M+Sdi)_eyog1wt<=M z=kUu}l-7TM*3po|UtIX-#jLDY9f02xf-jYLl+?N)c=r4y2iXdN&YE$oTU*u-dCd4; zS=r%@W~+O~eyAP>CQPOYzq&4IXk&p%DN#yl=_n_YgJ^@D-=>=XdQJAAKv)lq2}|VJ5$;P7|QIwM8SVd(uBp?M#^ck~f+n&8TNp*IHL-OC0mA z*0xv7H4a&Pug9C01FR*pm10^+W=TFoh@;&Jqx81qh8;wY?FDQvJ6T_MXrNZ%d0r)r z1lLGVj{qlFv%t@zZBCU=uWgC+32?n+xh(Blnehc_X#N3IpL1Gg4&gN8k5*clioI~X z#@3ZPM3f6Gp6avu1W=!WOj13KN~riyAA^{b!VP_;^=RUfxtl_9MX$a_IcKY|@X&E(j15ker38+c$5X+q9*U2(>=cP#YTP%0o6G7>ck-?!v+3t~*-ipGGVwBUXOxl?6iEn5r!GVShZew$e;I7C%r?gfdw>hcA2X_we!?bLoUvyJDo zCxZR_eWJmbn-WEFl0d7ml+9a|sIAp~uA2_cVdnQ-vNzlt|+B9^SCY6d?L zBdf11&CLGKp6vLH&|PJ(Rw1{=*aGC(6MK{52VqAWw@HGxhhmN-XhqMaoFzU?Oc=;1b?vaYCQ0g&VpmRz$noE*}!!1nfs%vak67lo-gu4 z5jTlJ`Cl^0r8?m{#e{qS!YP@X=ee*KQVc$Dco%ck?X+|8@Ukg5LKZT$A#Y%YZ==F)nC9Re7%uZSd!(WsIqI$m{n*NWVzkn>8Kz=AYDB=W}*V=RpQmWCJlA`Zhc& zGeX}!0DVi|jU`pe^}N>%uIl+c{Aa)REhTI8CcWvqIP$oL3j(3uhbMi~o>rFX(1;5m z!rtJjFtqD)`KBV}Xkh-bM(*AxE(b~N+*~>gB|QU`qx@C?X|jgcr}q-N0m!iFQMy|| z{DV@RXv8!bpw&K36<`XO1{lOx@r4{SS26e&+MJ!`#8Tsg{@F9+qfmM&nKdKUBpqC> zZNOxiMBpNF{)yTjsD^q)_#OC5NiGaM*G z=`)zl7;}XXv#E2Z4|@mjlnXsbb?AQHkTzP=>!irUE}p=ngq31H${nz9(F}h8y=i1Y zGA8?0C{26@Yb?ELE~4xBwz_hvR$1D4;?u6&_&r-FVQ()qF7i8O(eapwq@_1~XmxV( zTb94xfWhyy+zP5KE?JZvKjlKT>KyFbg_ogt{y*Nbg-RH<7)bvcO-W9l;Gb`w;&Cj9S~!0zCP!X~(NXKob^^0LJ*GKI_S~+fqyg`^O%2E7bhbRR@<~nu$Xn1tMxs2_sw= z!l@ROF{!f|@%pH(Ly)X@v@6nDU#G@P`IqrBxNYgJbDJcayRS_{U%>Ko?MezoxqPko zVV~7S>IRSR{$s5u+xriG(?(ZXk=TpFV{|q)=%Z3|d0{0Jj#c%%Ti^7k;Y7|JI`v;4ETM5Xlc z@Xyz=zxrP)C>4@alZgU@9f`8WP@+Tnjs@UJV|P_Et8ewa|rQIPsOAH8q&uXlPIHGk;u zkLKT5%pu`IPU#-159e{3+i`#`Vzck-78Ys@?Q5u`RD^x{_vT+EfTWNz*z6UdX*lkO z0OC>#J1Lt!_qq)+CYN{vDxD6yZs0JlcD-TLF{{SdAe3T&)`>M?MwbSUG?&mc_&xz@ znXW4}kCr7!PN4e7xm653tpjQ6V=vPlk&@Ak%xp~}7_6*56jT!iWQNX4SKk9)Jx%d} zzHya6<{}P6_5bW-10uY}1_M|K0Hu>fvzLf8=FD?|qGEisInD>#3ilB00HkaUy1DN) z6xX=$7zgkLTpPBuCyObCO2QXQGL{n5wb9FA$z`2q>tw^Q&yh&GE+UP(aa3`yp0ixO zX&*G(RETvDJd_&|SFh=;BvYR01Yp?cw{AVdQr=9}f(KwVfG;Fowo zSs~QE$(wZ}8+5ftC@vLoUJTY{HV6pHb>LpTQW-QGuW6*A7lY+?*t_v0#!B+{+TNNr zi6HhKY7`T#$&PbO-z#8&%nxB79ZIg1gwEMy?yd#Zh5?ywJN2Wj1One`dN0?~^lN#Ywa zM^>+++Ckrcsd&8Nt0S@YRbr5o+g`2O*cycvCq<%naS6z zE6;nstBDF1Zd)URy&CI8$cs)1yp1{{V1R2wHaV)gT$X zf)zP0FrllD%QD+NG^}lVUrzM*Fqz3W%|3ciII*p$<3m`dY~NGdHb|w5Mj1jVxIA5y zAWLP77dGi)ToYdS)UjmsZuU)Gj&{1qoCD7}-ZpEcP|~Vgm~HVd*sKcs_bSWHC_I zLxWtt@e`qfI(9kA+)G2Qk8$x7g9pv+XqHL&hq-lWG$}pO$DMrO8Hzo?(*ml9ty>bw z@I?I3KLC6#h!`KsOYD?sF&g`pJ!(@U!FW zyt93St;9&UD8~KsXb0J|i)S^k+XAHAZT)&T$lwoV zne9f=w_aeIAG&*t>_~%!`iQ7bt0&kx=t1P8_jdZg#DjZS^#{Yw6c~MG$FqjPcm-%y z`uIaOfYW%}27Wg!tgDT>KxL{5;20G)h&fv;$TBNe>I1L{2q&X!^}2Xb`^2qtKOf6} zWX5VWHlc<+?;RxPB}_l+2c~_^n&?bo2A&@#xmd?h)}-st9;M~ZiIRfI$PQNId{UIMsY+rT|}sy#WiB{hJNNJ zN3%@FU0Of=jquu9b{ZJ9GI)ET4Tzap*SB4s3>qZ0$~YY1Dz&X~ z@a6i`E$7$HnW@Nl%y|530cfPqshn?O`_9(;!mL}dX<5W2$dY0>@0$E4H{8Esv0|ti zjJ7kXfQ+$!r|4Gk(vD@C>CA1gqfIbH(y29qe2hrhfEknWM9E4ceVHIf;JQ86wm=~t zYX1AFZmY&OWNM~Kag5;(8P2i0r*2HJ{d1=iG`sQg5qiUP{I{r|^VMymi9Qb2kNII`nyD#dGchfy}j{g9h6$Mq? zphI|p^v8N@wSMy(T0BUr3X0y?Z|Enu;osp2|CYTwEv&``T==K+Bt4{jv27cOeagXeXX0ijGd= z*cxaWmCeQeAl2rx2%Q3;wvhfpuO#sSbkA01XY@9e7_o$6lHuYz+Q!vEtxU6!Hy<8- zKHtJDa4mTS#u_8D(sb&il~_!hxF#g~Q*_lK*>f1mti*lJX`)|mRL8bZfmz3<<<5a( zkfqgmSCMX{jl{!C(W(Dfvi1-;MVf$QBLVo7Z!797T9)v{5jvlEZl(K9u-x&iQY$S7 zDTKdTF(1l}LN~NAK=a!1XXTsXbHwR`0)@CV6HUQyScmCi#~AlBR=>Q z?>6ZJsn1R@O<{~4S!5_%NsrX9+wz*Qf!~$HTmp8?QhHvJdIlK*{ikb-EuDL`jwtXD z*SpGDQ|K5AffI;58_+j=tChk{n1*pX2NS;^9q0sUNE*q0 zwlzXa##q6fi(6d95zY3!?zLGmR4kZe9)Ed$fGI9L?D%$K+WOo#0wMY!%&#P3#mwDk z_)XHfgE3LSfX%BU<%jZpz1P=vbrtDu!|B?jWQ&ydlSxq{i>J!ie5`zf{)s{~x%u2U z$9|He1LM=)VtxV|7`Q6%YZwyM!#GYG0c`f<`-2Lxy<n=Hcra3{AvHW)bq zju3%{>+MJ%+bDG;1NW9V6fxLEll_{p_*?%DM5|5Eo+a$MkbaY#>~-e%L++Fr$vHHM zR7*D>xA~5ZS#55uf$pruwoVov)FT&%xn!CvdrOh=&jZI6ShDt=*(q=AN z+HKiyK|AfP;B=X{sFmYhSK+n*jG2@CYa$&7dcMJj-0Q6`P3IKJ;4$C@y`Bo=Mz;qM zCvHT_z%5(e(V3*J>)CeFUl9aN4gQob!F#Iwmm~D84pIfZj^xuZdHSei{IhS>4T`v$ zRQ$)R!NA+v%nDfim{)((Lme)J@0Ydse&s`{@Jq}oI86&_@~Ajkwyhj~Cv|vNKY->+ zGo67AY<-Nevqec5W}rG?ZUAP?@AFea2A`qRlJet2Pzh_7FchHWA}94!y|rrL7)1SG`GN z7N?zd=t)Kvxg=FIm@hgx9?y~i@|RNoeRAS>0;T5A8edb!aw*eBI-4-p_9QaeQD;3X zA^(|bq5$U;b5eS}xb&cvtk=*g<*?_E3t<_2J{i`AY?3OIAX%ZmB8k^8t$(V4K!@27 znFowaN80{*A_HIr#2Y@Y3)fC|6 zQ-un@N{Nftl%NXC$0h?DSZ_3femW~2D^K&KKS>7L$#8bW6pz)xwV0m{AxX362~}%|(_pY$3bbq57-n4GQ=2M>SfDk)!R|1&PfMUrHbL`^;&lGascz>ujQY;<`zY^pha&3`^XjIjlXUlWOc)_%_4QyOioaa7wL1-PY;&?ux~Y z)M}~f9IML~>}n8(=7W)={x8ypT$pMj zO@sgGff_y->TK|y-zsL-#_SEvC8BBvI_|<2 zkgskutKikD^@a74^CL_U*8I zouN%iLf@k{8U=wT#_C13{IDh#;MpLSL5Wf?`VSXrK64`AeIJ-Gem=trCHNRoq`8Dv zVCupN8IhxnS56_qL7<$1iES{+ibA(TmHL$Rnn^ZbXX@O?zYa6;t$r&|+?F9AM#$>( z21@ZV>_ZX^6q?;c5-hZ)muQ%?3R^X4cGg(?x=~yKHF68K>+EkY9>@U3Tjs`_tjnKC zAYRzyOF#ayUH`mbh6^lvlme&!p4=>Sv6;A;9a!pLWBqVe3&1t!N&K9~Aw?euxI4v_ zWtwL4yP<)xBrGQc?)ExB$sVelJfghE6Q2x_1*pzCrov)Ovi$WgEH*+!rcMY+#^lvm z(>3j&i^6qv6Q81B@793X3W)*FTT3>WReNPj`|T>_%2pAuIDc6&vM>CbFjc`oU^W1< zmwWA0xR)|5-oRc}inJedpiTig(ei%*2wbcw)I^tz3=)1(+?+!Vc$hVc*1Env`2GV3 zY%>mkVx$UI?Hh*i4A)l|Ildk)MV1dlxyQt>d_*&>|B1B^bwI zPjK$?VBgR~5aI+aQ);pdjFxF&+gT+Ndekw9WMRDy?op`s3bb$L!)5{W1cU2L4i!ji z!C0M-LJQ63su(<$oeWQUgBbU$$ zhr`C|{q3j7TY~t zGOPFA+&A$(W+g#XYpd^stfvgzWk&fy{GG@eZm&<2a`!`?$^$eJV|8yyRAloFO(wT5 z`;X$N(4I&KU=_b?G;(ae9R8JSQ~k3{dz#jf**Bi}dp|dkD|colfMvJV7;~1oYcCFU z@0GQ{#yNL~)Q!b#WP|C2YW5IdfkYHKgW-~}#1#Bl@RO3oXqodbzN^Az9Wg-NAn95l z4iV{8S($O^b7-8yVEUr(TG2B&7p%|htoCwmJ` zN#WNjcoHaE)<+^$0!s*r+6l z7>IEcDh)jYpF2~hTQ_qIfcC0H@`6E5q9uSq_NbZfUQQ`9J@R;WvK+|B*DRLx76jQH z>PPeubCtfBnccu-Sax;blRjb)L?7LTEaoC2=$+GYU($)gU?uBBa_xEAOtzIQRG~P{ z_3zRjVT;)SD0KfiwD|!T_hIVS$S!6XZVGky8m;nO<@oI}xIcH{-~D}Ya;gu~!jI0% zDf|tCJMMqOOm4h~x6l>5OM`Wuci%^`Iw+`m?K0?VBP&Vcvzl`wPluAz%@ zR#Mu43Op`(1&6}QXjAB35|}_QT)KM7%X=HAqQE(eqF!hrl(%-Omco7Fnu0E^I^(`O z*Nv``avy+o{_^70>Pn*@Y_&o4tw5s~2eTCEO6aRM$9~zmv2(F*G&^jX3VJV&0#P~% zZPc8qHEcy|0+PhGVjuN>>0em6RdSm%$) zPlhSTt(KK*=DXVkYmJ}uP{sGL`}Gn`mm=3pW0m{e(d|^>b_Og5@h+M$JneRjfjOM? zsQ8b#RZRZtQwyd&Ka4k{o#mzRG)ajQ8!*FjqL1Ue-xwbPXl1FUnQbD!8GVw} zV{Q4=)Ps0XdX|vNS+cD67a2IYR;h3%&~`JVuyhHM@AkK!h@S|#)YF$Tazh5mP~}x1 zg66Hxyvg-Bg9=f^c>-v;F=ia1E;v`qxohWkoO~2F=?tRfz!awK z>FlgaDD2jfs{+V@i_E4X|IwYRL4FfXz3+;$%GA~ORwH0d4fP@kAzHaV--z&v%idiH zGU7xIJ%XYrE~@|C;g1wq{m0^PdG5D#m4DQooA;@G+S)a=|0x?VQGc{^6l!<%rZjo8 zDOyt@-PP-8&Zx-WWjj5`K74QJd@0;;{5>aQL-8_c^$X+ljP~o8WX@`Tll^?Y?rIGnW|m>KI7P!BX#nunLQqzMQOmyu>9JsLi%*Uc5%hW@i2NyXN2Om zUcqF}t}!VK;P|IUzm=P^ar}k=TvZlp58xQ2`@W+u6wWmSOQZFApT>wkIFJbb7PM%h zN6(xgtJ<3!6?32yc0%Y}$07*!8_|-EF5_oZ;`+X$7qjw?hl}iCyfVbd9R8S}xxH}Y zVq=ewN*-!nxu5VQrv--)bX&u5C1pm4zZRz9_*Ld`SGHF|Td+woRBwdyv*6MQQvg{3 z|CB9#83Sv3xA9*Uaudm_A5Pcn>&hw}zUICnCM8PD6YT7j!Wh{v8EmNldcTd~R#mod$sPy2H~H6bEt&d~#TEXoQiB0SpZIC#OeBMPQ8=Qd*h zXdzpCe}P2e<_@{2&&AWf8Yu@8D9^sVtSmv^e>y95!`I0mu0X#+dEY%0?_hc@t|T#ZX>h|BN~);qPp zm4lr=uF_V67du=>sy(FzC6>7(2OGb={FAg*(YiLFPzBm1Uen{#PQ@QMJ4eQ7X%iOE zau;^IJ0^%TlOFM}4-UCRcFGYza{mEvNWH7eR|LQ(?b!Al0Z%JU_6)CZKvFAq2&r9T zFHDREA$Qj_AY2=r6J;|ud30$(UL{LYGibMf-#{u^Xzih_Lrc&WkV_f|_^$lB289hz ziDmfjT)pW|3*+p*OKFa`Z&JjAP8PxYJeXd=78PLgvW9F)zlf4{ye5P#chYwwl5&ss zBqPegribt90!Jv}ROb|i?ZINHxe)lY2&7X;XJ%`=tsEaX4I3C$Fg^(h$u*KIn@JMp z>PJUwdXTO}b+ew57tK<;dPsm7x5KDi3V#HO$=rQ)X&hxMz?)CENyq^L=$&`zaTyqH zHmFeYGb%}|{oQfzxNy4Aj(EPCZi`@A{@P6WPFvG-2^;k*v-mnDCn@;x#3gCPz^2m& z#6n|3%jrX}AWSZ0bMVJ$N6s9PClqH5zhz0$Lqhgk5N}TPJKErW>mQa7YLYkmDoRK>5BB%C(z%v``=5D2ew6s*{<~#W08N{AO9;HSLTB ze?1}191yBDPUc-IHHVrIsGus2nmC6_fG-Y5Xga0-gbo83cXMmOyV-&BU{VZ2n(JUk zDZk%b{G()DcZZ?0NvGM~Uc&CRwFY5h3=idd%nTu95rxNKs@WjA;xLc*d)s(Cd!L_| zF=w5(RmEj}q80R0LkCZdm@{o`;3Mxp%wg0QkdaOv&N89|&>iMLb3PW2G8!!|G>`nk zO2%M)-yIm&=Mb1c{rCYRq;#&o#Y0GVtmb>Dv`c+kZs}!glCriKIY?wE7EhO(omC6p zw84AJtJl)@6=P6AU=CT<=c=L{L*3fb&UZb3xCM>3D5v+sSaa0SGH;y$#0xJ|E>FL?6I>vyv>N*oGwaF*TlZ_BELYstZxL?2EO%Fx=>>%t<@?S;Eiv8!(n*`^hJ8HX|Z2wp; z>Y@{81)A$O3j4rAl%xxF$X0CBS5N2*?n;fX5zdvI^lEb+RqCZ(pA&`RrTVyAn^ z?pcfHchLo?iKxWKjjd`lZQ7=YMcOL7yE;4Xx75Gn+z;D6SbX@i1F#i)zta2SXb}5P z;U({v;i%E+g+9~?r&bM97vHqM(@Y-MR|Ee5K=nbU{`tSAvA1f2CG-|n<=t{e8g{aT zn^Db2#YgpPNwCD3(TsegWNdi`(@>j!djIkmkF%BhxXn*97kcDa|J~>p^t}UiSM&Q; zg-}f+G^3Y1f^y7COlfxF42!tNcK-`-CvcgU0$zUBzLJ{X+=N&y4$K%^j^CDUhOdG? zr4vNzV1)AM9e<-p%rCFC9!Rw77)F5hBa}V^9*5<$JudnS) zN>#G7CVu87)LEZEnFhFolTkIdr}DIphV74slCSX!b{Otu%IhSPXNuKkg)x#@@MKPv zMx4)TU9Zm+x=^TNX;y3{Ls?^30TvVDB6U_Wo|Oi^?T~Z7l^DpBX+ZLxw~|AlP3Yi4 znT9wwI=VG)GQc%b_>Zh!hc~x?eX6Y+_n55ta!yA$Z5b``C^nNA`-x+dDK37gPHns& zYT{(*;|RsW?2&o?C+6^DD{W+lf}ev(jZbYPeAf^Fms0nt4R%TIN^^H#&T+H`NyoFX z&8NLD!tM%FRr#~{`^{a1kApP4Try^Y|EEhw{Mmk1$TCen*i~Xk##*{YXhccwB{3hI zZ)85&EM<1bB<9d25pn0vIYHoiHv(Nf)fWwyc{`DtaP#haoxscdEiv&>+jwUMiQ%k&@rG()cwkl#Xc{5{a04L}Y?k~RN_ega( zR)0yt*l%S@Y+ijRh9f?rJfp$wRC(zE!Ot)CEyeK68&&s&X*j7hu~$h~8?`2Bist$< z67guw`AZ8hjk0OLD;>?z zMY8$DIP)}$VFn%!ZlZC@``ZeF&4E(zXO0tr+Y;O!hMP456tb-ieelv^SneghJU#=T zS(l`Kh;^Q%2PW;`OXehv_V<~VbeN_-1NN{EnNH+TOQaW|yf6iyHCfGS&9q?cCp+I? zyFxqvaK<2p)g=Fj@3z@|pq-9LJ5k%^(ahHPDf z$%YzFXEV9>9qi5P-=lP?DJ4R-m2}b4oNna3$l?F4>=#HH9pJRlKUu-(nvr@=0ir;H zz0*@6)~MU|-RoT`&&<=rkfIpvz8T3?R8Ak#kNdmW>%4WN` zam+Gdp~fh=;GHZa9*hN=Awbq4s!{DU6Z``RxvMF1#wDAkwjscJj5kWs@$r>Mrln8Q zzIOS=bzrr;!Uk6_C8}AMj@Q02_E{=7A;ZA)!n9E-9a=FRa!H{kjhHsRijni5Dy2z6 zzzPDjI%m9S_vHyQn=irM1e0j|zl7iwmO=>yknT(5}=kbn`yYRFTsw zggc^yn7@x_2TF5D-5hJM0EYo+aHvLc-2XeBQ(IF3U4;aH&{5Mv@#eJ1QmHU#yXWPQ zdtc<{6JSHu_jv@4%E>{2B(DFb^Wwjjp#OLB^#6Gc<(n7sC2ej#=dL}Fa<`b0>?(@p zatX^bao>Ucs_a$Dx02%(_#BhQ5Z%zCm)v%?`i-(9%KlN;>`KnB1$($Q>J6eb6$Lz8$7{)a3%%VOT$uL|$Q@!{Y!2 z@Tw!Gw0RcRo2FTH@88ZWZoOWZeVWPwHxbkPSnhCL6i>mI<%FgSW_eRR{KaxNfpHVo zWibro59K2)p|Q-Yq0&Or43cha$DuF!xs|Bh$DfxQpw&a#BS%QjEH-A`PV1)Dq^ppp zrWpdJtKxgpMcKbO+Q&_BFTp+8!iWc8r{w87NfmlLWBdEkU70qdEroT^N?Dva3M0?M zhcSZ~(Muw+n`z!hyY+=Qm4C#`#Tno0%NLP!3i5i?Grm#KZ|;mv&qdk_uD)seO>+Tj zFyMBWp{uC+H090I@b?3DfNzPGJwbGb%^#N+98Yx2D+JhkEk{&?OviIxXawQ@*&cdz1oqvH+Rwq6)a>D_B52_jHh1WRR#d)ILk9#gY7C6@h;tL zH-r+6s-IVhcg1_x$y6@b1+fd|xo3G~gplfio=%YX^W(4vbZ?zk{?&K&mvhK| z?{`)4LJ?OM$^GXmtK+aJ?kMp@nrSftZh_V9j_z!oE#y!j(n zo&IHMykqJ|CmMV~S-mo;tp^*(RCTdL|D`U&ikd`}wsXfzrZ}xOKI^Mj^w0D4cMov+3@+@0RHijEktf=~M~BMdIwz6kBNfBC9CO7WX3 zw}Gn>B{ja!<~WWI1TFrXuzCIH znH8z9xuy0OqQ6Dl>SU@}7Bq@ADE4R8O!s=7D&Ky0Y33{2&xwgeFM6&(HfenLhWXVU z$y|K8NwuOQAPaU0`(*3#wYS0{MgYrrf3Hn>eJ)i0lG%K$vHD-}o0QC|yvDg+V0hKu%nfaUE#^uNW;whAdM3?FE2mh|VGb=j}nDMqkP z_0f`#PpJ-!4o4a26!p`;F4-A$(7%alE$MUn%lrlVWvdMre)dKJHI7ziInY z$P@Bb?o@FA*>lk1%O2;-CDka0c2uV1sYWCxep`q$8e#Y`t!@3(09p3y%kTwdMp7H? zDWv}`%<`VR+{ARrRuEA^UuN^I%SfiYjjR&>($d%l?Y5BkjU3hZLp$7iR79jW1x8w3 zLj-Nj@4C?LvOj3llky#}s@-JnR$#F#bO%diTRK^+pDo%eL7a3xnbW5tC+$L4Rg^o{ zz^Mga;Tf6Hi$fu5uFJwB=9R9s5+`>uj>pU^Wqjqa2qjo&veqV3BTe#FT35}hB$6Us zv#MM2C@r>7mS~il(2zE7%SzpyfNhRbeJk`_H%j)xyP99$-~~f6zeZJwh+iLas^HJ< zZ5UMKh&9gWBO}n}mTByo!OTm#l99}~N$LbEC-e9}dK7VJFo3H|ymr6J`iVo%AWH1C zgk|OnK0zj~C2Hw_(FX`@*v3CcJMR8AfJfa~AkL67HH-KkK+D{M5IN1Y7Q1k0A_jHX zu`<6g=`bpd|3Pe1|`P~EZ&>0k@4YQ@q~%XWL^4B%iE)=5wWC@4!TEcg{-QYol%2#2cv z`vs<|oS{3qz>A189pG)15ydn>8JiOANT&5*8z74W8XXq$6+9K}Lutbh)M&ni+ctP! z9Oqzvm@Fhn`2TrX7#eGJp&ymPr)y5 z=RP~)Gq4J**ExSzmC7r*;5dN@n6yiAZf{@R)?_cfzMqrm+!D78jKeTrwAZtO69476 z<`z6$s_WJDl%t#tY%3S_exyb~gyP-_x5^<8+IKW;>>-Fw}SPIF9K=dZV$H9-LS_M-vm#$ z-v*8c2&pN&yHVB84fyHZUgzT1$gJ97*h;h8$~Jo;bedo6E~jGdUC0A}a-k~hR-~v> zi>`>SNP>~qmM-B}LCg8fHonOv;(3x5g6d_+yOAKz@q<>~pdBy1@!jO4Se@^C zeAbTnyB?RXjbf9n9c|E#mRgb+{iKxiVbjHne*kl~|8}omW4`~AjTZQQv`=tLzjet* zTRnCV=gP5FTGfrQ4gEt~V{q3+_MK3J;+_^7GU0Ml8Fj}XSy7^YNeD+$bex7!0D6jvpW$^-dv6BBq z8bXh^zF90?*FWh=KI`EuyF<6sdO6I@6XyxuC%vl|95wGU%V=_7NyO z@f0>d+JxIWJ$1j=*S*7r*x(LkNkKu(N)dG>*K+m#lfOx#7$Ny-HKH9>RRcn;3$<}_ zl++RQb=tkx=h|YK=9kPABj^BY4Y*g2Uq1ONTctSzJab$Za|>YIR-S@;Fiy8y>-&k{ zh3^QhQSc1y$cb|NX{m{HZTO?F)y7p%W6k|U@sh+gN<5(o`r3I2c<+xMAW1uA`~a$7A0}~(Emx$?F9>k()hDr+lC6V^ z@VI)r^nD3S*5t0?yjR021Qs6U-+e4ze_S2Z}{w0)aPfv?5pndk)${-T_4XTD3Ou&EH4cT}+!4%j#mVtx4C(k^!`_%J@Zkit=Y z|DwBW{6l!}?>fYCY_$_%#ik0CSY`|~*SdR?&^x=3r4%djF(y#vqs>Arvw`)zT?s7# z7w-|Zma`afzmzswv0mps?1?G|_mw;;u&}xmV`M(TJX>Sq=1CiE*N|fN)R3_gaLeF_ zS{;xRUUN&Ax`6eVQPBr*{+PBUioks1sbNe~8iN{%{oQjvL?kMi9EpBQw9)+9gVL+| z5tkhb=O(L$17sq;6)+90V{wfxQiy=nMdADucQlJzI*o!82Q5(>;==&*&Uh2ZR|2ie zW~64LPC`f-*Bu4If5f6IJdI#QS5hpm{cvW_0bA`8h`;t6&P5eif+hrpTl4!S2Cv;1 z+ODy2^vLfO?H1;h`quV4c#f5e6ma`-L zMM!;SocP&eYzZRhT90W}cqa*=+I(0*N)amrGpq5DIBJEJrBRHBz$ux`rey^ zBJm$U?P)^b=NkZj08N7FOu|LjsPrj>hp3cDVC3&%C{g%(3ZylWC<9WK*)tykL>H7#j|OL~Sd%^>NFT=0!}5evd;Wk`6qC zVWJVVISM49CWmSHR_nltU~yedy-8gmdi^Z7g;>#ezM;Wkm8Hzt?L>IZcvoxN2B4Vn zN3bvo5sP;TNNOH@3MQB(>*}9z>LVnxeVPM?QI40}!x#RSz~nztZzShXlma=GOfm2m zEA}A#n9@$h-bbKP@iQ>CB3UcIXEy^X6Iowy(_X9HbMnnX3V9x;=RfC(l!ZM!kbnvYt$>|ZpTh93MTt#{i|t3&XSCCoq{?Q8NjP4DsVD?lq+6e zFhmkC?!e=!f8r_@rTH^iiAnfYH9WS%5g_}57c~=t{KYIT$xW>wa=G|b&O0$*C*odO5 zQ?rt1$Smkvsg-jg+`*3}%Est6k4-~fW0Z+iD(O_$*;Kt<5Xf{u$QA5jCcgw2VxNeO z1~O9?@q!UI3O+f`sKZLmWBKzt!6l&cQp;5z!6ChuT$Euac>=}{N%B~7QPQdPT1N(! zp#83VCal>BSzCH|5FYFjJH-^_*2uW5pvP-?G%4DqVlQ23?uc>!9BGcj-9km2?DuF9vAWyvb_ zjBs#R=gfv@@X2{H58EfUe*hYtpJF&A8P8PGmWuY)ry&;d;vRPtuO<@GmAmnL>!l3y zKeMKGx0ZkTdRByq%6`%YS$QRKB77&@K`Q3D__Nz|``vZk(6fOF&Dh#3^FrMl^}<@6 zf;4|1`Rl9PYeLmRc@ja0#DmC2twdVG2ZpFUwT`$I!%dnKA!f01ml%07Kq5u}bDK|g z*y7%zn_}9Z;dQ~km+CUc>+1z|0Lj6LtG=Onco-81v(MH{<o5YNgvu;%YPmB&q)(I#icTw6?KxoTD=52G9{0|U?4lyJ1(fPQmn)H3S> z>h*1XSRSf@T+a}TDYpn)2QDN)@kT6WCpGT$4MeY1bprMPMJVca#j?msMC9h?FJ%+Y zL}=m(bc9_PRy5jz5)SE>)x)vTVA=v33D8ru-ffhHOl)9;YQXU@D*bf_NEXal;guh_ z7*OH9;k7Y`_EF$sTAB56W&teSVsHt@ln&B$M^aU@LeHbw+IxQN3GR1hA?LprVksUd z$WUCSzlG!%XWLSur4-C9!wU2Y0>Mk?Xw!%anNd(>*pZWL8XE*Qdh|bdd+&HS`+aS6 z#%R&W2%<#q#OQ-z5RA^~B?(4Ph+acx2ojkLGfI>pdMA2`ND#gEh!#;JLK5{slIKj; zTF>6=^Paue+3z{)yno!ETSNx&@x8C#^}Pzh6go!b>j~@um=@nyj<7iVY9x=gym*`S zeA79dDzr?2OmP}NS1bf4DZ$7BOcg{oBmG$ZeM5}Irqv#7Ulia06s#)}tqRX1`_egh0(eCzWaBy4^RF~$fC z%cI}M3pGoxel>3Z?Z?&Jv23?FW21;_E%yzk?vbphZifopjZm?u`U%#lvRe(9|r)WhkjMW z7MdlUke*B-?BykfCguMD{&CU$*5Ht|IwTHPbXVV?!xLy5qYLB;%(|xfNO|BW|FP8J zwQ7NYe&Pg#LM7Y8ymeh3u*trCP-2nU;xU&E?<)**0p1GWKu^hV3NI_8G8h=b7L8)rrPgh2)}B&Y0u^;`|i9B2)T3h&jHqe$?AvYKqUC+a4n z!G{N2TXC(Idvi07izwqZhfRwe0%))EtktsW&!jrV;{ULIcp1_J5thA%v2s}!uFor9 zij15fyFryd1zp-WfnUw|6#CCgb1r}NFAV0zUM z?vjLZyU*!wO=iJ?-3H-{b}PK!I_92=WnGb7OgR|pkf2oZwBxum-opMui$sS$BKW9z zougS6TFlmkOuZ{VXth8kxZM#Np=woFs-b1g^XbtZf`qd85NY^Y(!Y{=|=p8Kc3#nPdDaaq#b#uLO<*XT{}(y?daE4o6sV&9G>ci zi#Q!TQ_TofZ}KoS#%#k{SuE@?A8O|TJ{|A5iO4>hj;fS zkymm2@Ubzn@?Y1hgQF$V&!GiPGk5o@g`WZrR@Vi%^{eQ`m9L&(ZyFTz3zmEy79~E` za1z{h4Mnh972HdAj(5%#!{m#;{}`l|e}9e>{`ISEm3}6fllT#IOhV60{bpS9ioQhq z)}Gz>_}~>;-Tl@bBZ7^tox`T&5ar<~Z#yAQj#qsoYzm5jeU^Cqd@C8)ie(Z9JMnQ0 zF@>*X^s}jw#?Z#qMBQcSKF;j=>G@xKj{;v-L~T*%FX`$Rb*4d19-ch{9QJG4%YI5f zoov4$+w+n7SFpht^VW}T4pJTB+(iMlLW>ELcGfEu!5%ZCr%QNi%$0p@o?H)FM3i? zD)2+1KfUo+Lpb$}5Z^9I)A1GA_XkuXN5@o4b%J8Gmn_Yh7{s5oKNRj1xHNOidyiuk zGjm#IxRLU13Vn%FB=J!eQh(#PWCCci_gk>|Jw{(#*5CTl8N7QRG*^(c(Iv8ywb$y7 zzPc?^ESpM9G=Tp@&mlwNwvtH(H6!Q0rYZ^9djj%M9el-~8UYL)a{7AfIv~0!959b= zB8(I|LBXEq@_VQwLyP}~z3o4?@c)hd?f>ELX{u~?EM)8`H;>B1tP;5r$%Uj0IuY43 zQQqQOH0JsSD$~FpE*TNim?C~OhhY}I5HHw83>bixRTOz&E83wi4;9Hm6HrO zEn$FpX+n3;Cdd<`26~MzZ>)>pVk$R&s;BWQQN5c(OOYwBeqTB&7A)Jyjf|bkvpLfQ zuy~E8Om>MZ76Y{ThW&s)A}nhayFPqVpebbQe-M#Vuu#aYzby|ebmkYm2f_7mzf5fL zr*?-_70=EFsN%#qMNxPet;*|s*>9R0b+&Zz(ZL(cW1`vuIDP`UXoD{3v$*Ouz&WL~ ztY5*4O(fhpL_=nqy=wRDu_PA}Z;VL~sCVU3jIaA+MWU?fwaSMmcyEo;Qe=YzCzCG&t!6L~Urof0kIREgre$8b#~SH(Iy zlPxnOfeU1RHSH*X&uAlAC8T)T=VY;}baQZZ?IC$Y;kEj^^!S(Aq_uA!=4Ua(P1d2- zYj3R;!u|wI&elSAlkC;5SJTUW->YHRh5oVw=c+gMdDv=^myS~5(ST7$!i3KsZJ1Wr--A^y|n%})nqCXQb7+PRL-1HM9jjW*ss&S*W8 zKLI(^H}q=}S9tF?Ls!-*ThxpcXiaJLn7-2~rH|zO%%NHyU$<@k=C)>|*D6%EDY}uQ ziTq>1SXt*D{p&dCXDY`GBx+ke6S4l$n3UlzXOG+1xSZr&?4Bn9|i|P{=kfW zyRf_?h9nVnxc;Seuf$ky&+FjV-YSf9V$SNN&`=nI$e3;z>vm4-Jn%}i7<$@E(p`0JUBS>&{@Lzv)eKV?wyl{XM~ z-fvKO>X`z`{ICVU_G>_{*bX*;*ktPBq*5BmmQMW>C6*-k(> z?xvFZ6R=e$-yB&Is08S2gR1WA@phIV1|i2FE{uaky`C_0i1f!(W7*qcuLfxnpl6!9 ziSZ8uT$$&vbBlIW-`m7rJ1cAE`rFCDC)MA+o665;2I@x9Q}IqvlZsKG>;j2qrDa@5 z)WmnV^6tP<9$*5tgP>1m=b2fWAe+?B5(P$ZWjX$6%qSk{1t=!ca})s$7{o+LQ@@u_ zy)pGSd4dEKgHtVl?>kqgO`ACWmQGScdbb5hJS_GbA$O9{@jnE8Ishqf%rkiascu4} z7?fhr&rA2<_!F>$%0LxK+~)SO3h(n|kL6Fr{4qtm+*x=}yMMwAq(_OJB8WBq1-Acx zz5g~X!_dxcdNUE0L8L!;GNpEijQ}~(yT)MZ5Pi-_{+KVBILL z)k0*+41%!+^n9xmV(72;Cv%Td<7NSR4Wi z=V=7Wc1M3B=gH`|&jeUg^m8j-%=+`KZjknDPc?md$mGT%r$M?bN?0Y56cA9L)q`ga-XV&Ey)Kl+Az0N&q{evdNY*7Y?qZMG{` zX(*6XRh7ztcTg4~23KF4Nm3D#e^$p}+HU^wLo7dC!9lgL!6tnfj~KWc%sQf|?Yl2g z6mjLPDt_7fA%WbQdpGRf9AFtF zfK7@xoG{2$No@Z;B*6?JxrI&0sJ=TZuMms3Q?OjtooIJSP4n)HED`8*>%mPJt47Dy-*%cT}URH0B=I1te{n-)GY~0cxDtwcarzB=#>h98r@-H zwNv)pv`jw;a^K9-XW)yR5xo=(D6pdM$`3mT^zg$U#undaPVZGnhN@J2=5?*Afc(5; zqlw7)7+PMN_wAAb-Gk3F z(u}O)((7Z8m5eLZ>r{jY>*Yol3BT3*J?C@5M5q|iIt7!=goBA;-^B$xf6OkI8f~z< z(o^>1%?wS?6pG~z^J7|(QIafZZ_l3LrJm0PC*r^+tji^?tTMy*5!y86T~O) zJH~|ohF2&BYh5c^oQZ_CF?)vL0iNJL^-#cjl$-J!)PhhtJ@t~EN2|PTlC_W3pD5+A zaOPKEpQO~zh5E+RRCuq|h%Nxdn+pFh)}*a-^VSW|O}DgpvNlv_O|(9<@WCwn!=N<` zwrl@M85L7yRX-2qsDj$LxqFPgQ$|MAmAhOWTJwXYcS7)f2C~sZ3+JJ9pJ4z&m%Ah=wJ#phppujaEK}P3h4uJ#d-}(}F zX(#JCeyCSKR~EkXew;E5T`cuVwwqg)YyE}4UA-m*z>&|x2|x^iozk@2)jSe&i%%|sMg@SS~(~C#K-K~ax(SQy~WqZ zO0Vh31A75t;8YJvoG4J&D{|VCY7PxPE;%g?{ymH#q{IE+_N>1pPb8k=AATXHe_-U_ zgz@hW$mk|{F3A$bJb^0aE9nF@S79|3^leBnDeHQDA&#O52vqwH?<4xIemBvP?gSDd zX|xIY@(fh=0DR2eSil3gxU6m%DHGz6XF=lsH~sbBq~HHDfG!B!T?4;!!2} zMnGH9Z@jL|6;k5iNYhSS^npPvHBT%3o99n(CSn*f z`l+(h>h`3SxU<>jG7zpQUf)huWvhAbw5e=d=ctI%Ho>>3?)lfy%4gZ3Edn@*r8zeA zj6R0_(Bd`|aP0-N$>%DJzMh1G==}*5{yEs z&jmuo&R#Md0};|9cv17N;yi)kuISX2C8;a>%!2}I{-zob=)n1eFdv`Ed_9nRT)0r& zCU(@|2^ulVg(5FgZ|r;m;@;^%a5o^CbP?=7y!eHRtaiuAlDR2bpZd(m#^fZYuP) za=3Nh?UQMBNNo=PHHIbF{u8ysZL+=S&&Mw`l$ZQW#BE1KsUbB0F3V4{Z#{Rds!!Qt znGdao{d&7GE9JmyFLvA0%M&4$%oUV8MZsmwDmUt0IC|_DQjLn*P$Zm^;*%Acb`(U9 zhHbC=`V_Uy8>yQQGrB80FQmM3DwncZg^0d&bVtMG=@Z)pW3oM3@sR-ilQOoU1kF~c z>O6+D|Er?w24u?NBDNZ01v72C)`iZS<#<6W!oq2<(M0q`?Iw zA((}RP2F%mD)r@F+stK6awgNjl}b{FA1i>D?wUQhrkHu$Hg2osl?HjBlR-$iP-t-f zM_XC+B@h=+Y)4Qu_q>&P;MV?Ik9@G6e+~uK7Lzw$Ka^LZ`dmvxyp1M4t^fyHTFqo5 z&;4L>j9gGk%dCsD691^n$W5&v$Re&wYLX5dRlZfYGmj{$pGibJfSH0qAHU&n*D zM#%GOrE0&?2Zyrgq5=@ox&h1Z8vDaxp-h9Pqcf!N$2#UbHWmXiao@9y7vGXJEvUMC z`Dk9T(h`#?%eX#X0cf88m;B+4LDxD%+=bJQIkd2I8}lc(2?dhalOFIs#Fz^i;Q3en0?_OCkp4e(vHqTyMX2Pkh z?jUjaGM9m@h_!htz*VC zNhh85h;L2f`o9`~-tnC>m7Y9Yjrvg!o;H%p53;|Q1XQ~8AuBZ9P6?o_-Xv8Omi?NPfRCV$d)Fi;xGEjv0+?aAzfpiX%aR@I7PMXw&|sA!AcOIuo%a?AHXHImvIs!R3QQ%>hh z!zxN4BKA&lJ)t%YpJiz;2KH;LnllA7#6XzD{3yAa9PkaF4>O@Zyq+1A=pa7I=!RR3 zA%1u_muh6HnM=lPSdqdE;5r=EKsX@wwBl28xE<;+H}kZ@Z=L>NkJlB5l%Z}f65NN9 zTHaU(f3ALp>97pM1-5Uo1AsWOs?{NGa|9KUcxWUP%VNoe`&q^`QMG|OgzXkgAa6qp zRQ+BBZS%~I>xRCbOCbi=RNjb3q`CKJwxUQp^;Q2R=EgdCCC!2?Eg+PAHS%hb%UPvJ zIyK0D*ohE9XSv6ZZypj5Q2OA0P3g6F(Nq~hOZG#jsd3>{c~ZHydCpMtCznUAbifb0 zwl>O-IVcU){op%C)x+fmXcoSPKY&}cHQ9vsyeMct$4L=gf&NmbY>XvCL{=S5;XuA! zP7)yE!b+*zh?orD-LbBkufhhvxkSe{voJ#OXlnEKMOg15T{&o9 zv?vW!NoKRiU4Aa$$e9px3CZ+U7RrT&iQ8V*qv#k<7;J9LH5|b0Y^fDoceeDIzn&37 zPzy!zInC)hS_P{1(xoc9_|y{C>E^B@M;tg(sr~ zCZw$8^ViG9vRle{IQc#~$YM$D8APzBLYnM47tu@~`FV_Y_{3#|`cov&u`wS1$T@bZ zMe40O@qTOH{Ni>>se0OT4PAGxlwM&BSjMxGu@N6T9%y`n94E_I%!hX=vK90V$!ac; zH>>O#AMp;R$fwW@Q1#g{=9POr)^$mcjM3l|B@N{%xEoO{Py$XX_QpSI0~`O2V*jfNiisn|iI8mWBm)5QG!Gdp57&v-I_@a{GVaYf5OebvGc#>0tpl&6bEF&ym15!qnn$ud}gU6(YXyw zDpDvPDB{aJEiReb$^aH%Cx52~{bkFRgVUzy-6D}NWUDkJPydt|AYx|n5chT=Uqt>` zkOm<1C&1P(V(!)DKLL3U`}|ZHa*uImQlp+lQ$o3NXd3{k34p{TE*;Ci|8T$PVv_BA zWGmxO!2Jm_oQ|Y%Q+u>L+9Of3tIreUzErt5NRx8N04+Oe9f@C~WAShQ#>$GV!nNC$ zmX|^m?X!jt$wgp(h?36IP?n<6hW%U}2-n_%35ECPkS8gE;zoCs#MINC4&|oSkQZ%W zTJgw!UfOug}=b-)LIJ3MEgUUTftX!~*VhR{#$Iu`S?7Lcpv9JNS^W);!{gp) z_v)hC>Va6A*I$r;Co?Iz5wo@#6`Q6ONXk&ahM#{{BXylGH6UR)NLk@csqUI;rpxTw z^yX7p+T#ai5?2?v0tu!O6Zq=&Dc8tqY5n(O^ZUR~?K6w3GQJs{oy8*-C1RNZct)^%OKxE3=1cjw5Vvm|Lq?pcu~9~)h7hrhXF z-K`HYYU>>Uc=lB-eO?cO>hu;hZn8{7jj;R_A`?J!vSV<@oD4r};~phrWgFRuZH%z7 zDIBWwp0`3mVj>=#@1`~L){+x!4>W69^qZcd>~a;8>-yHm*hv{I>S+k(*Uv;p>Ti79 za$;Z&w=i>jzr;RLyq*xgT!+&yYQr0>$}wj9N%T!GCcY5b=v(x+`neEOMajSz0P{}N zh4l~m4IRk%z1j#O1)uk{4*Lqm6as1$T7o-Xz0X9uZ`RFLiU-572KO_7Od$%ADJp9w zHxV1Tmv)Wd8$UYlsl~&}JyJcqjFM##W$DKC$u@=VMfA#-ATb_Rl9zZx&u6~-xVem9 zVKmQuzHXUs?K4swru;UzvP4O)qM}0OM^Vb#C@~C06@j)*a@HMt9&j)kc?^;((B^+e*zveh{580o+W#qT*bb;A=^!|Z{XgQ{0JH-`(fo1 z=~MOUG+oT)^!6*;%LSl?Ou0qF;!E2dgmv>2dM4d9UX>VFghdln{^F_Zob5g7n&2yd z5x3iE>hYM&p8Z7l+X>8$;oYHc2Dl5WFdgB_$A)#%{aYFeOa`?}RfO^&s=Rf17Q1F* zC1;Q-|AWl>Fu|Z-!Qgh=RXXCySA7-z?Vn4Oz>^EhcQ_n*ZJ)rCf(l$CncO0Il=3Jv z+WeK$eY4Mtj~A16w368=zbJ#z8BQZ)(IPXmy{V1jYJKcuA6^t&J*M2;7R-%K3lOsz zQB9Wlte7;E#bIssKI+4DwftYfH^HIwN#5;2`(vD;A|(;Of^Eww3~OLzOnnjrR`}_t zO>{|hLpu3A_i^v4UCgL3x#hnY%ujT<%Sc*-QE(~%iEA`dxtIWW`ZB;o$=Qa3{+@E} ztixV>nny-Rv?t}N{0~}t{>EYa<}vB(VE7 zlB)6o^KV6Yr9PAbhZL2NP0{=x2hRw^$F~S>QiR_)`N@>Cw7QBd;OZ=7Jm}; z&3kW3c2Pimvegv`2Y*~m={Ii73vwPi5gN>jjYP5lwibvf3-rrB@F zp%L$Ft2ENlD?zu3=_ey{s=v`Rl++dTZ^SHQQE@-YqpAr^hW@gOo@h7FT<21hP>5J( zjUF7}0C?odPh4FO@@ixzdhpCvehk~pwm9UQwJ>(oS_z8PKy$DX5O?Lcl9d%+eBAND z@0Yxs6r5S3)*EVLPO|sGZvHS;7+KisYJcfau8kVMg`?88sl2b(+BEp2 zXtLmaxd2~^lq(~QQQ?*yb@%BGW+k)YjF$Bk$|Qa6qkvEe>Q{S-j96X9H)4AOOQDw_ zGu{%l?Kdz8`zij`aAUz})omP*JF816sR?SaKnmFTH!|l^nNL-i68w~5UFGWn&ONcv z?IueXy6V>LcicOVGWj67?0BLpqq+HT*8-SA1Y?n4oLEPfJJx?x(Hi1R4Dck26kHC# zpcg0b3on-y773vgx6=}zuW>UT)>yyekB;5qx)$+Sisy!o-+RWKRX@EBQ1oX~Zr_ZRO;FOd2l%>#uKwqS_yy?Iinhx^zhi z$Q7$eG8DdY7N+PWNaqfIEw$>kalUllaOV8Q%7E)yEVs0b2XVd341dcC1UQ%&``%hXT$&_kT5B3_lRpXMH|` zIurn5&X6y@f@Pdzhk@^Qk^JKc+c|c7Rn_nIq~=zix8@w~O*IG2H`!IaIC9XhqG*&f z9OGlRer2*S;Dro5|%3y&UW%^W^V z$ODzDH!d<6J>?ns%Bd?hw3O%iTZZxK?qm(-hj=Q=Q5r1H#uT^IQ@sFLkM%lsn&NJ=2X~jPRso(IYU^#P1vi z@mCDd;E6bu2&+sYAy1UV`*~oF%5GvGQaFB3>PC|SM5&yVN3E0nF8z|$u@kHnN~H^h zRj_?F-NeIL$L$TQxuz`9|IniSGg0yTZ8}jfa^kmo4B)wbElwoDqVpt=g0s0H_w%b?=` zl>?Y4Q>w*K{H9k%*H&`O??6e>0s{I9siO4vcSJ^dXrvY-GyLMS@x7#Y_DM$7N@(NI zdP}goqAKKKnk?k14L`8SWj;Em8m#f3xt@ zCzofkI!aemc^@%PdyRT5jwYjAWIh!0I4q7nynC(y@ce8Y>BJJ(57*!Oaz|bCN6%X) zr--@Ds=;9L$MRpQ8vyv^=q9)wq6YpM?gU>=WCH()oglLdAIhEeUIt*9(NR&}_ZsLE z)ZO(g77Cl`LeXvg*W!?5oqlBu`?K%D?%#3oN84Y>GLE$g4<@-^>VIK0 z_vQP}w?7nCw&6LY(U;Axg@%^+PsPjo+0faCpk|0!9am;sZcjBl7aD}!(c;UP;W{dZ zd%ov&JJU6N5Hag0KzGdZQ$l5utYY7Vs_EvCFqaTEBx}PJQ1n&i8JXjv9=F@=in*l%bU&Wk$1gv0 zHl@mKyxn>sD+(aIR|_iX>KZslGT#{i8d3mK#!JL0U8acAfNg-~RGl)X9+#!_>$?&7 z;}WucDYoHb-XpN^iFs!~Tx`EHry? zkSc>))+^y%CXlnA68y89PkUMeg=MKfMcujeF33ME?gPM=H-xf^UMeY73tyLfg^rVI zAzM6ybTbmBF6IXx^<5rP%E_?1sgsKsIXMsFYpJ5+ZsCjn4TTiB%+b zLe)~{J>1I(OtMF-P}^%rMt^&D&}SYtSUwpuV6SX$ut5m#uXcamx{&|Kns>8q>P1GZ zZvZKL?K&$WPp+|o?kOQ4;Tlu+$y$gQVU;YBgQl5k1oukYOC4}S{S7k@KZ*rCa&nT&m4^_Y}?>56z~mHo5Gm1 z%S&_KrRZI{({f6oH5s#>sA;TqPWzxbK{$8Mr{IzMioD}j|K0Ix_NC(GPm?|92L&x7 zEGN4BUn1xlK#$Ve94`{_oGRsa_7q7nPCIF7LGueV^!Bae@%@!053-jX>V03Dw$ihN zI(oOIr$L0+>EZyAr)|1X5+G_ooaPk3*~B}2GEi3ISu1Kkn7Yd+q-T*^k};$hdtAED zVqZa_S~!T>MQ9g~)P+A6rJrEf21<-zs*;b3nb4gB=%$r zS7o{m2Qis~z^S4C@K%zr>t8RR`Ii>$-+Y%TAZ_hP1oFR5Whc`dDA%+DjMIsWi!!Yh zAuZA_l(r2lKmvGpJn0$0dnFN~D+&~qCxtuwo~|9Ft9+}4quJ1(7s{gvO+E7_E`^@ zbX;W^*bcyo!E}r(KiA!?wJoH6Zz^TnhlZepH_PdMRmO_6CXwBwAJXQ>#SM)MaJ42p@UX^k$(R zcaBZlKAVz!hg4l|8cuid@C`kjYuQ`RS#xF}w)C21g0XQtO{4mpf=?bY*P5AbYO@Zh zJm%HB)*L`(A0(;-2%u0h2Q2qs>d9~k%9V%`>d=(NWJv5>kruf}GJgDGhI@67oNS#O zW8V2bMMKmna8&J$q(eNo)9Xg3Pmf}hc%-GaABX_vqQQ!aa{pm*is1h2p!{}7#+IPT zk5)8Dy;QGFyg!`OzMGh?lxB0+AC2mCo+p{gz_;bat2}(Q*C?s2hh*sWt+nTz+N|xY2gJ+6YT@JgyNE&AqM|gTa+mql z0A(=P^`>Mbt?&KZr?{px)xC=fMT)+tU6;^>6%P>s8Y&L;rli2ph)f@_@s z+5{3iZXZM*{)7NoNidVqlO+A>bR9f|E&*2~vXT|5nqL*@A}b2XbHh(q{Y%5v52i;J z-bnjD?JKuO3}v@`oZseYawm08nKi3*e^6KEIATRm6}Ao4<;c7dR&A0|J0ROTgbo1P zX)^7@_XQ8^VwjY~pq)pQsfO~k=`CxK4mAh{wKgg zt$rc)?ci*Ax`iPu*1o&0|Eh~V$0BfG=Z}Ls_iUmJZ6`j?_}&_ma03YX6QLRgRCkC| zln@WEP66WN9koPPjDNl98*{y3fNoTv(Y@4~X$&SyOE4~7HO|YjK6b)COTP=J`vtM1 z^Y#*#%C_ga@>i3TpyaA4wvBheoyniUrhfElKvaAc1g1>KRGHY zKnp5ITLu~lfQcuim9cLb`!ICHrQrUlOoXNmsr(|f!%K`r4X5tUEC9s*ZlYI@$%dbl zfRf}V4UB~Vl!VkNB#D$x1Xy0Ib7jpnpNuO!V}6Sbt~Lx7#L}ffE8us$@Qvp=^^z;v zS4jVT;j~|oVf$M@kF8@)Et1R-Da=N(98@h9`h|@k3w@#b$N@{1i5LF1=SOpaAC?AD z-gC8V0ImKs`S<#tP}WcJ!hEqG*sDHIPQTt~tQKm0s{HnfKnqE}ORvHdUV#{?_fVs0 zcgelGit`Is$Q#uZ^zZWPM@H(S&n)hpkDC+cz~$L(WG@Hu)xzPBE)Puc6%CL-KY zf$Gt2ooX~)333DM44^SXrOd(Fqwh^P6P1~{$t&c^YC}SS_-S46A z!D*IDrPDsWHgx!?O(QI&lPBY=1;xwA2STjt0ItJ{$gJ{$!6!RXbXxE~0T7PPUcO9b z!ZS_qx8JRAO_E48UAexGOW(v4tPZ4$6GeN>x5GHbO(Qa|>Ln@yq7Mmy2wk1*e(tF! zXe0QKA=P(!MI?W|e5jvk@v`HsYgsEoYT`d3M>WyEGz83j@)!bO5_e2G+8XDa(xr@f z(>IL*Mq4DV{t6bQb9f}&)CMsqQMy&RWGP5R5fALWvwL#V{BG0d7!;wYTC#}!)s%$K zE#*m_XOWCso4473n?yIx)9kA|E%$>~GC6vPxs3M4^Up%>ZiCU0`_@nmMgK(a@}Y+n-rjn!$xin5BB9n(jIw_CidtC(Ex#{2j#aZ3N(CVysv;)baC z;vY}Vn4AA_6QCXE!p&qHnrM~r2B}>(At>H|Gdljj!Ynp9Tp6}Qo#@=jzCLrT;CNgq zTee92#cCUyWkfO&u=0YDlfON~Gyu`!f(+6dW@M)%!|-oS0Re-_L#615t(uWB7=vHM zjaLi(-Xnqj5AFZ|p7-&;!KYGKOPGEKNH(L+?qZ%8t)~ubC$as4B`YQpdS)ZI8sE_nFI-0YL z`)S34EWSEsKTp%4yX)p@%U!uUZZKX=EA(?E!u=t|NB+tHDwu?WiBGv8RtC}QObAn6 zwO*pS8c=qYGr&8G%{6!V4xjiiARWI&5WdEL@~bKHY9SY*7eT&3D(+F}>#&NEnCC6OMogW zR-ZbDM&f{0SS3g}>(d8MEiiD`Fkb9%9}vo2PL`xh=d4#M50ptdO`N}3bRg$o?VBeG z`H|VZ&qdS!#HrN{N(zxYt&Y+dwN3R1xUT%gJDERx!zK+4g=r??^qO=_gzmGNczcsT`w!|VQd2RtwB2954w220%l0HWr$zbDx7=K&t=wUFG1k)Zw>G#X)}_0 zDn-&H2?l3p1?DM#%)vg_dyArOOEkzpt91x6Y@;3o{+3v@7q6PP^lxq}DoS zXf!YJtGcPx!TL9mbA?L5kG?kDq**zq=UI7c-$(E4I{sKc&5mKE!e<$AFz=cMw#L((&h8PajSKyuJJM48IUg_CrGB279j5e~W$dW3-7GH(sFNiQP(&WsXTuvyJXNs5 ztj-n=ZHS1V<+7eQ3r3ut6m`CUvqB!>mLjG@){7BzeWLPfGLc9g9T`qiL;~8;JTU;9 zEEbaN7k+LA5)u`Gle?vMzfT$V|gV&JkrLPMqA|d`$Xr@U`sNm#(9FWQW zUCWz)@j8D&(Z8SK{}cQDg#rJ|dog+WfPAVMd0g7-Q_2)KJyRREb2Glr*`^9=XnCg~ z1)7o8&G6$=4zo23Q|*G6xD6?<*XP{aiBm^fhZDU6zhs!Gb(BexABbH76)R zfJ5^xIbBmHwM9@w+wkO2dFT65uJ77I@}vE+(M<^5(WTScFarqpi0+ET#eo07X$^c` zEvP1`M|eS%7NJvxcO*wT9{J;%Mvsg7So1;Lc_F7atxjl*o;Bz8RSXXcDsj&uWwRu-0t{t$J7)n?;<*<^`MoCHN_Qc$)t}O zT0gtG&~TkA4<-39o=#bPX0B7Jhrq!mT9j+h;96cr&VH61xi{GTij}i)DD8yD^vJmLPYxBVDcLr4{W(sTG4CSR7SYe|xO1e?`~P&DQ^H z_9KS;3D^KQ$g2)N6#!FUpQInMH*5gfHln`Fu`b3Ze`9mm6F8sS+(>*9&-aAPPX*+7 z^e{Y+6i=~L?gXhN%QtUVo9oC>*XN4q@%X@+!ajW7?2mV_wi}Z0f2q}p8Nc@HC4!kD zFV0^6{@Ei+K-KZsHi)J)ULpGIN&0d-Af3+C@wlop^S-eDXIMF=@}8jBtiE)1*zvD4 zJ|R|1u9_v_bXJULY04e43hqGSYH_ki${annfL<#?Y?;RsK!R<0305^O9!zKqwu;f11nk zYRGQX%>qNqTpJ1${ST2D3m?S{bntK~SC-EK_vF&PfCxFRAcdo$X!*8RuGBU6NH1w$ z%X|+FQK`Yf9}-Hp8`~ORhrJMfybtVR=E9h~iMb?&{M>zMVTFxWs*8OP20UGz?QrK* zec~xCKPtP)AM|!qY|NpJij_Bz7kg+>_#o zGSc{kmyHWr)B^^tJ(Xdd#-1M<1DMBm3?h=Lz^z=tqX6BoJ^iU-C+gLsvQdDLi2bcp z6m|l#^u;WU3yXN24z>cvRNK<3I)-=sL=yoYd~M$F|1 zbs1NxH%P7DZlS-!)akccp!Gd^2veI#{eVd~ZK7A=4 zz+Uk(-Fih(kf~q!sO01_Uy?Fs0D!%G?HaFY1CO7MZsPqs$aov6yGnm^k55wo<|+8h zZ<8z@8Ewt&HT5I;3*h_A^KWH*|wX78(A#9eOzSC6QNuA)Z!>hVogY4_7zvYY*E8O%;f(E7`cje&e(E%Be_K1X@K zyLd}I-Or>p&Xk7NWY~VCYd%YTWW6zcTwGF$K<{QZ(Xtr8^yv#x23^FF?fn zdY51AP$H-DmE!Sg2$z)z_Fi;cRGc9*81hBj(8o<9h#{?x#eK<;y<7a}LD=^BoaJ5f znU@K`*WPcqiuuLg1DF=|{Plt0A~n857FMW8OR?o*X3V|L9ikJ>tJOiCkk_H;3Kf^E zkJ#-ItFgetV3P6PL03SBJUOnG!P_>uf)y&nz$Vb7`yG+&5U-`gC=NdT!8)QFH$r8q{LAJUlwPbnsJ1q&{@$p{bmVIyLvn%ER4&G| z(cGW4-aRy} zZ0zPGnzx%)fu}lr8R^?>boDY6eI`-fOLdt(xw54XR;XZ6pSiV`tD0Cw#fJ|)B&JzC zrTN(@%zo~#B5$G-`}EqTg<4u)GA;DUg6Yzum!j{_#b?WBCB)ZA`8M^o;3nb|XP*gm zjrj%I&h0m=)Ym;31^h^ixMnn0CDfR7IiX{EZ+~PL4}h9El&&K&8Eil$ce%8kl$Cte zh8nMT%}RvcP){j@2s7iyD(8h1reh8&sYT3<2a%<)I9^`nzNqqEgc`#u44h?s7?1B` z2)s zf%E6y&2{}e+JubNLxQmk-y=ein{L>7VuZn<=&)-p^-0Nhe61+^Mp}MA6+jhnvJ1bw zeE;mA6UybYomB1daLlySdk%klSTLuwa-cM+*nNoRi2D!icVKkQP+oHBW=UX5SJzjQ zXhx2c61vsXUf#>x?G|uH%UD57BKF0YCK&DY@u`?viovSK%ygg;g-Eihnf_PoRPc`e z3R-9tyFqiu1ExV^0_cT>Uo3l)!Cpm9IfYcx5LuR~02X#&?75#P``H!KhCX=rm+=`Z zKhyNis44H=N*M8;z=o(eRjEN)A{|mX+Br56keC>vZtwGdoC^4BaL4=$)cl{wT^a z-DJ7fs2A|AxIf+OBS`~{jCV8dr%|BWaJM&UBL%ry16QOo@{b#nZW{8X6MQKPEhd{N8QJPerEI_@00BQ+jp@gniJH(Ig`9BwX z9gOUa@)S+yept()2MC54do6O9KVfyoBp25rj!vuQLj`!aWR8z2>%a>*{e%8*40$9i zI1*$u+96w0!R|`d`hPKZ-ce1p?V3*r(hNleOe8>%(2H~kHT2#?FVZ_mM^H3W1w#l` zdXZiQ0YNE(Kmv$>6hXSu1O%jupzq{6XUaD-XU??moSDCpm1HG5d#!yx&%W>Lx_*wz z^~m=vtyd{q%0}pfu1C zXkJRwbX^56D4ssJe^P5U1nUdT6&*YKV;4N%g)SPgifK&F*u+P0B=aElJcecTy0VXl zXynW-6B8uT=tp!&M|3X;@|ePt3dJ?$1#74&#>Z+SzE(vgAG zz)qNbU9e`Qi@ygr+qXG=a&}EL#Fh0Z%imV935_XzKpy!(JT(EvxA-xt>y4$n!jjK3 zr2S@PSIC>stf*@*@#VIKrESdb<(x1ph3#d%zBbK%@9{n97|Ykp4@njIpp7&jFKo^6 zUvgl=*I(*C%CyTCbS#bVfSBMiz^$-@C#r*M+L6}=$)2|!3WowFRYh`D;_X%`KRtzv z$GnikhLXGfL0>G3$h6%ftLs}Iy+`a4vs2KZeUou7$lW)=Sd?<2)>bxcSh>?E7#D17 z**G{Co=KHtE%rKxF7W3J5~vdJk%HT3MA17k89OzdhAyCnomA+9YzO3Y5B2G`IF!RD z8XVLo=B2558|dq0wA;F-g!Y7mPlc=&xnv%M{At0o)x3HZ=;sl9HzG7V@*qV)TWRNv zc}zPV0Z*=Y^MclHJ6z6wDb+`+OssPM>2N?F4E8Yean(H4dO#a>kY!e?T(Hjel`CJO ztIZ7~9_5i@ml?>X2FcQ_V8?`r51fpe7>;^rwkJg$CGBUklN>*m^a^FHQp6%Y2nRd3hasH3ssp(3Kh@`%P%JvmAZDWEo zR@x}ISLwK?Vr1*~hXnkC{^-fMVtMNxSr1BiodFvl_U#TCnRsX5aRri^U!Zq9uuJ|X zAakC-ApYmMmiI`Oct@hbDo!rl=E5y&q1iRXIMPDDhd>QM`oAu_a$|7t>}#%(6Txd@ zE&q^2r{9SYK#?$wf{ewEQJ!WC*xW!nk*^WORJJdemM01qeo(Tv#JRj!M_o(Tigb9K zYj|l4>XI-L0{jZE!g*;L0Bix-Z7f?KGy%7dUa4$*bsQGc#pav(L$*oMdG~^JY9W6( z`pQ%yQxPc^W1{(^uASrq(RJ@nX}vF8R8dP{*WZtRn_v0F^@c|K+>lPulk7FeZ3T3w zP(L+D1zZ2w;`J{!75?MU>3{b<$lvXq|DkfNLZOHU zzh+ckdMfZu3`DR}-`Q;A!Yc;bU0NN!JL73N(QLC9nc!|q7-44iAQ9VACFQ$o3;*Cs(~!h(OB&2)JEajwrP zPt(M|DV1A=yUP3paA+=WrvYjSb8?l`J6iQX9obakX^$>^pI}*MG+Q-+yU_ceg1%Nir!6Y zF+qVtgF9x7#!htMm({NuwdyULKfY%DF zbFSo!bq!IN;ssI!Xra1WFmJsC%2HR@O1T85azC|>)Q|6{H^|*?@Szz2$x(2zm$v%N zI|v0z!UI2QZorY@flKTZ44~JDlx;rQ$8gAjc}igARj`Gt6K|_94Yr;0nEp z-SM{}zJ;z|;Y`^{+0Im;m|#oOFX6EW$LBGfG+6Zr{um>KnQ_Kye%9!O7nm!8Gh4<& zG1zkBLjCl`dOYR6=uSG?J^4&189%DSrAXQ3IoElg+Keo zb8O3+?x{~i_Bqw;;z_+c6*lMt)16N0p+{Tht=jJyoS)QOOWQ1wOwfbgj_x6MTnnv{ zd-T~)3e#4L!1J$PuZ^T>0)N$hCLYrtWE5#|#xgh1HBc%rxE#7%Tt-k#ZYcPgnYHoG z{AKv^sdIoXdu+Ma_UVmqiw)bTD|JGgh~~^*$H?pFray;?zS7|nK6r1LI-23eqxDtD zrj+;as_s|nAV&rUl%T*>B z`m40Bt(kucEWI5qCbJ%i4|k$C`2J|4EV+jNj8f zD~Hezw@fFt_41Xp!`Tg-7ezf*u))M~J*a)7VQ{R${-nZ5xw2jK@=~ri1hkxsXD?oJ zne0-03{}YWM0ikwmI}@SXX!s-IL*dJ?(OjjFzv13mZr+i{cXGN6$b^#A0(kPAG#00 z&M5M50-AP=GWX34mqa~*1;3-JdkRaF^gSe9CR;^eAokAm^GV#40~K$pOt>jEPYlUn z07aPkW?F&$kRPn3>^kda&#$J**0{9s)Y&sEWU&F~_{chA@u=NTrMpJyVN8V-vK8zV z(TV?!cH!@J4F9<2!XzQYKih!+$4ah$waM~7Zo*ujul8a9#b}eYDyaLL-t*hcMaIDA zhAwO3lL#}$iOe}NLA6s8+w~xC#@_|jT$I2&Rh8punM@>1&Yey{RR$0R+#+BlsZ%1! z)9pf;`vrq`nw-Z5dONCSv|RXY4zjuR{yinMApg)J&OoNG$v~-$Tt?JZJbB5N7;QN1 z+Mav5GummJlk4&Zeeu${41Z28lbxT#Gxwlj`--t1(dbPLxF{Q)TxdNd>$N9b!h}b^ zrZw%>1g{MZkv_&CF8w8?YoTzO0cd*lqp22EbB!;dARfpRNfK_dgX~?4@oOoVAxY^ zm)*C1CYH!nAjke_k+~}C)9N~v9><%8^PbC^`t_vZVaJ++IxVQAD4)Pxlyur+_#+z^ zRpvAP_ku3)=_h_=ZgXyq_hiR(gXb0^isY+VBRF(X;#*e{xJ{AB!r9{r+rbKil5^wL zOSN)MMlTh$w`n?{a)f8^OS@{`4aiBwmCt24Zpo{2&()5^k0s|f#NU-{nkGHmmHP@m zt>P-cEtNcjNFq$e0AVAkIBKFUpv13n-->Vpd8gbc5K=hCX)DG+TnMiUt26UA7kIo* zxF1&gite2z0T9A}bQ3lDiSV&?8MH9JRmiM&-~1}#*`YEobc@&(taN=!ik8))NUu5V z6Lj4->UrS1YYPuUTUXAHx16jVsJ*Z&e-nH2v&!=1xutMJo$>ZS{a0>g5?vKirTBY` zKG?@vd&WGp{nMAJhK7VK2ktMhu$?_08VMnN<2}C*Vxp?W! z|Dg-6U20%S|FsUuZz-LR#hyu0r}xUtt~2(_7%&xaDm{L?W&f%THrJL=(ycwEk4#N` zFSbW&NM31nHx7?U6Y1?JmYV;mY32X@jzOUpiyD3mi&3mudI46ai;{=0VXl2wWYOI7b0K58t$lKN> zNOmK2Ii*}m70aSv`s!=M?s#Z1h0O9+ii*xK-iWIU+3QP6;w$zC(4b_1Df-U2eO_>p zh%F;7-0>3iR?5)?p1LtQ4aN1gTv8BV1&UrO7pM5vN$)WAZUKvr0#t~_tsNpy_Ppp& ze==CMgC|caHZV7k12*}*8J(^frmPb+$6FRVxp7RK-h|6#U9<%5GY8rVC%3=J05L)tE}|%sAt{7TZL}nOiSu3x2d+ zX|g_|U>Voc8fJp67G~;DFSuZu+!DUukD43At{vf8Us)%ofA0Y`MCSHk1Lz-&_uokz z30M;e52D&Jp*+W}NNjBxsS7=R8%)%VyR9gZj=J$Z-P8SZmlpfVr38&x zTwVLbLv{)SO*DUM&UU0D=Vv4-)+#uK9MAg?k~=!g%9Bl{=^2O|(%%nEh@!H6z@j(p zLu6&4cGTJcb}YTw@vb)2gYlaFTgSbd6g4bPd?FE&XXKdq9;7VMCi%>fWn1xnF1ZZb z(_q~1ZO@XbvO#AbHNc{FP-o+n|=pc&r zCvlrk9WBC$@Dm%r3pXal(W@ktsa;sTu_oh8 zimb zH>ZC{dCvUy-W&JwFW-bEe~R30&7SR?0j0K0q~Gi&Jc{>kAqsodE!cMzL$%?){z}=E z+ovD%60?W~ubU@}Hrj5mjm?nN{q*Vm^*y}QEa{pL+uMC`Ivt>)&*aBm{nyLpp9!B- zXznJ>v)(&~O)TQ(*iEk9po(J~q+5ub(fab+&_EfDMI{jeN7qNMI71!|)RE_lGb}6D z{7Py``nX=o{h&Iq#V}= zbk)ILkG{x$N!khdSd-N=`kjQ8{2~$a&i2ym(~uCiXU3 z8UCeF{=&lckYbz}wYBYyipM9AV2J^=puMUwz+}D$xDRo{JCS!8K@JY$2YhGjkN!#G zEQY{n+zpiqyk^cx<8Q1H$l>k_#@CLXn$Y z3NdALMr>8)@38s7KgY@aV)Cc}VW3ahA20K8l--e2hssejSb?%moluZTM@ z^|?Js0J0E{4&Q|3Lpqy^2HP%wmNvsl~99i75K|;)!|s70%nk_DHFBL?kaI8Bb2;( z*K2*MGRV$Sb!SkAQ@35nP_grOzdE}$TsyQf=yfmo#YBEUamvHUTun_=^P);};0H3{ z-nVzT&93%vL96`=K+4%))B*te=BD}}_Hg!#$GZ8ViBlzW4}Tzc^snY|=JHylJ~0If z$SNI2gEjgPFPT@Q#fs{ML`I`Ai=!KV0d`CT&p!&4t*5+?X@{JA0P!!p`}qF#bym%< zIvZ^vtOSi~*$bDLg07~?A9Z?|x+(?%TqKOd_wh+c*kc&Bj_zUW&d6rHaTK5H66hfW z9R72pz`3a$T^D5`CEg7*2ct=HhH{_nL6!~bnJa8Y*)`%4$NM-R0mrs$LIEq3gQiBp zxSuE0puG0PIdkL6u%*DUzW_BkjNr;#ESG=h#*1yL`B&L_f`_nBa&vky3`n@`O|CO* z+M#&-h-ZMP0QAdwC*cZEAxhx5sk6itA>^>Y=-txTthFIodi9e0D2ZIhefV^SrA{2; zf5$H6FMxfdbnI7FMV5Dc7c3tf@7|UXS00&orEGYzt|OBgkWPOm+tR3Hk$bd2jy}V- z+nBvwN;vcSl&}K&u7Z5^?72qwB=o!^){gyY3mc-lzC3LvlF~5sapXp?L!i(L*Rp1kmTI{OAv^BdV>kN7PTG2Dg9W@a9`58mfVDJpvSiPB5Oi8*t z8b%hGzH7OWgnayKS!$jn_W~}}yl+x6)-t>?j($?j<;YB&id`z@4P1<&ukb=R5@m_S zWgOy{i}9Rk^x1q|Nf1o`BO7m=3Unp(xHOSYREbo3+ZTYOo%%>}Pa&XEnu09$eA4JD zX)=3TaFF0hevslIp^dZ)?Kv)^RC=p#Og2cvNa>VPkBZnbLkfxYe2~OyQo##jnSfWr z&Z@>7GC4Ody~-R9Jt0MpCAWYkQ$Yr|k|+4~yj%rl0l}oKVk9pB7n5h0;fHSs#_?x> z1CYfNAYD&ji|KzMJpJREH2<}U<=^OgOIZ^ljVabwI%E4@ZQas14O`be+5FU4d&ffh z*2J_>d3FJ2@Fly9ORMWYxe>|kQ4TKKw}=`Vp=M$D+w@h2^vAJvaGcS@Wnaht!1t#&Bos^`2WSd{G| zIDTNtC#0GQXO|1BWMX}X*9uVd3Xs6n1<8#A4le846X1@H6D5c}^^4A<-trWPf0$!% zDxUuZJ6T#?C6vy{+tErhJlUER^H!Ai#YCNoJfY-a2zihZ@)yl04MyZEc6Mi+=2ep% z{^yc5>#HMTJ%}gwbAgVVEJ(g-3e9pz)xZ=~DxYv4-P`ZFHk=w8Zz_Jp65&sd^80)9L?HS{u&U;~S(>K~D3O?=aV(IGjs49C@oP)_1&Bav{q;yDN>m*5f+TC0`U9 z4&|SQh7RqK{9s^FTJ%sWfeAq~N&L$9j@ z@U-NZS`g^MR+}2izqI%<3f6``YAFFeWiLBzxNkzP)0OchY;SowO0cZ?LhF*Nx)nbl z0l19n-LZ%>fA+xeR%*EH6&;erp*>Aiq0*!!-xo%5@N0g-wneY@COv zFx&TU(um>S4vzN_Qy;V%$v6p#)^w4jL=JVjLrBd-8_Q?WtJhrdq~#dJ7Vlz6^(ZH7 zGd(x;n%Mj61Gg7f;YF3Y5L_`wFbWRd-DG=X3fb%2T;Ge@T5yqX+S=Dnk}Jw%tF}eB zD4Y~ih?(jjH_|!y$^S5Qo+;omat4u`xqwJfP*ErU$66LHg|8$Ax+s@INHqz-m}X={ z0hT3aFYp#36E{+tmW%2CAyGIrbEl}}m+_T{PxFKLNr7>KC4+X@6vK=G5aK$D9FP~+ zGxR~~rK|n>`QkkR+fes*l-x#63=ax>^*#wbcipiG?r7%qp@68=veW`5ag}uW0FoJH zrvH__@ZYF1|KoqdDABq?LnRC0XeMx9DtZ5wb#|Rw!dJPk7|BbG)AB;dUTHG)I>MkBbNgov|ixQ+8LnWz+BWm)l6O!nGJVl|FqH4Bj#fqR6y+orZ`v>uE9 z1(6!xkP*k31S4wu<;xV;NM{be*87Z7~?=&kJLTvd~GBg{#Q7la{gvG%3<&r*2G z+Jdwo<{3d8vo}jv*2Cb~M=WyYRQ5hsD?V5G?m27kCiTNo-car4&VG88(V~thR@J0)lMN#LAD_`|o#b1`fNOM>_^QlBv z?l#6h*;{`BW@QnlPc46!ehMhwf0}lNz&3S1?cL;-4`NT}Rcu5$z;Z<3a$(CbNT1junOZETw9DpuP!E-+|{(w|iyX%5RZMCD0NU=#a_ zeZg2$2COPLHUG!bUjSVu3KUr?u4hu4_OP+XA?SI&dPFY_r={o#uqOs;0$lJ$(8@-d zL{?V@BoQ$g&I6>OCQx2TCa9Kr5dk4PNdP?J{u51DO2<^vy=1lD1UqmR2yx+@CbOO( zf6OfTVneB)oFC@)=KdrZBe~-x4*7>2$e|nf^gg~uhoiqz@`{Lc#d%YY%g}(hkhHO{ z&}M95kuIQcNsUpO)khEzY)?CcHRQEBTnRbsbCgXFL&4R(F1P)j=!!6CBR6KHdzJZ;x-JeC#Zk~&r{ zeyWknppZRHq{Wp@s3Fgu(5G5e5jlyAmtF^28}s5iat(ARrgGDXyK9fVM?eChiU%fh zi^7@1>lVhgsW|?`oG;;`N7PhAjDR$a>kwu70*O6bkJ&-~H*8AK0T)Y<*4R$~GhNbw)>18TqnSQIi zKj4j@L_awOR+Q!7!gCCUhE}kLQV;S$ysQ3h&i4L&Pr&7pNttw$=ZjOx9=orz0I-0cziyxL{TRb3kx*bx%4_5yx0&)A?%C;+geZH5{KD+N*B z%EO{3Jw7;FdMQ5kenJW_|7QoabXM9w85iUJrmXqzMCtz@zb2@sqElquCVfB>pTd=m z2+clv>KCGKg+m1gAg1BGcH_W?_mXc~J}Brh6@ST$bZ_>)PoNLyw+26L-*w;CP(j6z znmp7i7xk&R^W)E#Bg1DER|MY)JN2wF3ReCY(H%o1*NNl-$Io@Raow#=N;P+9%NMxIXw}$HW2zCpPBpP{_%vQdjz|I9G zk0h9Mv1qQJ)i9s|M^e*0*IOUB9~LZNl*pbrX; z=En%i{nA$i-d?3NHFrm(Nl*TvRyHQ#E?bZi$*dmoXd4nwCxI#rPzd<_R z1HO@cGgXUn&ho)u02zY<36h)pp>6Vu7Kxhc$(5Hr0^){5;UgJxz7=|#M!LmSu*x5< z;dB%W&t2i1n94E2#Aj&DOAsT|yv|THiOAZqqQ~MN%W+Vl{3YyyhQtjyD(5;QM@*z z__{R?wd$v!ijfd_`(axwmjXrNkGLUDH@g7)se$bYe>imU>efAur8`Den7M|^%B98L zVl`)~??&XY_}7G}nmO_#-Ittq_ckFF&z}r&I#WX64S|M-db~?u&}#guQU)pN?dcJP zsg?$t*FlwEReJL9=lLo6JLROP>oZsU?W_R3d{PJ-byDIP5F*=)_U}IKu`L$>&dtI( zimP@v4-+5YR%mqWf(D<)5_I;>OMOKx;rT~6embJsn;GLau^RX@)ums8Cu3jkPUU##%lNSLkJ|o%N*4x?qz>O_!b*P_099D4UvVOG^pMW zhUxCY-Tgn03f4`$fPZ{(GXnocJI^|>w`xy8qimM()lFTGCj2Ioq(tFl%hN7Z6lZIe zhT>F6!whP|W8!$H!`@UqiJx#yCWyYe!TTPZ z&;7v9>oM4nKpMOJDKuS@K4JoG(;L03d3fptlHc^4Pf!kPnPiyfJ7Q8cyn!s%uze;n z5FH9uI?89MF-zftj9P^%ohS^D#ywy4gEv3}7MXVVJFsJ93ASb-rBlxADkC~TK6=Mg zydfxfCm;R51B2YcO<~OK#9UHMZ_|j=D=Kc0t?;=cez}}fkBd@*gea6^vcdCbfnyIl zbG^OML;5zF4=SY-3$_hbw;(@%m)w0}X0Hz35|aW5}&%T(JE9;5RswCXvBGVh@jq725Y z=n0veQJA=vu#5NjN_VyV|O&dhV%mS~W=O{Mlp^3R^PR4VW#!NGQkGXA zEAT!h^J$M6JhIjT(b3mlzL<5sE6W9IG&1z|G%tTgA%1HmpH4$XZU83AMPFdX#nlj1 zvy^W;hHXNXIi9X^)LV^o1z)Y!r{umj;8F`?N7nvI-*3;!Sd&Pf$t_KOx?E%vqh*T= z#syk3WV(eBU;dhwO{AX8KeBXLJj0w*lR9wXU*#Kb-@?}g^cz3X!!*PdkLe#6gO&Ep zk-gs@wu;M(d{UISJ0SZ$+INmYHt|Cw2L?00)sckGAU4?+aCu_g? z$2>tgB#AkF255}bPy`w`qRib?>8R)A%uY*Fq&ChIC%wD@{#nB>C7xYWC*c ze9LeR!6Ehv%uRD+vXTY5j`?3^rKt(iNa`7C(%a=PZR;GO08i1;KKydJ_O6| zu0?U0u_(lsL)1i-Ow9MD;upQfwLJy2mybM@`FSwM5>OMsyFBq?-WeI7R@)}d;upI$ zQ7%r+tpD7dv^@W`3)^HOxVAobNzeLLrDPZ7P0^+x`xKt>hoS6^0TDZ9 zukz?*v;!{r04~S1H0nv#881OGIs-@qYtNAkg}+q*zm-~j-Kuu3d9o^Kh!+rvmPxHk zWf@MOeRnc=^2@v#PDAcen60o^H~%?>kLfQ!3lSH^2US%!9jR~#rzKt<*QB}Kp#jRi z!-kyIGHU;^B9G&^?|t#%BOs{rrlti=1Tiv|0p>J>I4%Q1 zz9eCEzLsAdWlK6EwklGpQ;O3u;bNzz5?S0uu=kWnoJn0rG(I-A1^b7mB-N_0=$Uti zqzF9?+V%K?P@7#q*w4m&H!{=Uu z%P{g{Sj{@^RZyxvqG~f;U1%v_E^uBFa2U@`6BIndWPqPVg4&oij1f?z|FNCN$R-mw z@eYSgI3x>PxRIU(YAqY?DNrVJ$IyyR*!TvKj>cBzI`SXJrtNnDR6$c0omK=Mp; zx`WEQcS4yi>>`4PZZ=x2^72$SzK#tEA6~#_DY%zpGpM#XEywEWEJ4!KN29h7?qTjh zI*MbFs;mq{=Z37WrOIiZ_Y8HwHo3|@`ePeZdM+wk?bH``ngh5BGsYPSv0na=O?)LR zir?f;IBje2&~f!AS5KVF`LCFY=cx%RH9+6ZfDH_NT=|>|Auva@2*BtK`UH%WtrSy@?!i_83IlUfoaF&! zuCO%L;4Zn|#d}3dMj*z6)@Zi5nvYXW*p#hl!C(}srw`En?rmr_14$eH% zB@5fURpM!vX>Q-|A-iS4{aylF>nA(NFwZO4R*ynbgq%D*;`=L zVPDDszr2vh7*&M0FoRzlV(D7Bo_%KZl@;tAvvIn!fXz>xH6!aF6T=;U#79)(!`9HQ zD@iEr$HLgJbvFJNc2ln(y6h?~Lj&6!RA(S7crB$K zJ;sF5+|e=oHE|Q5`?v6>Ok3!H^s`nF!=V=`o|&W_`4>CiqR;<1LFHdf&igyV(!a;X z9mxGg^4YcL;3h{vLlHsq*NxWJx2-We>gP9=R+k!Eo_u{~7N< zjFp2*&PvdxKRL_M89E&(sX0h^n~k zI9G_A%;Vr%T_@cU0)@m@+|oCAVde!jnjqRmvB}&xB$5k5r+MEz&W%bvsebg~+hkE5 z)N?6IR_*Wdt4kRf9G#oIc!&o(A-nESv)!yEpKkS_l35!;p>Y(RUWw4}SgE)j;>3J^X8*^t}`YrZ2 zmy~kBEYBRZj1!^AT#sL$p&7~`URZ;dOa4g7V#dqMlKZQ(#^ff1@^_tyT$}Mb#8x}) z!3B`OZBwRSzfAmYrX`Kk*dj2mWl5KD*>_7CH}a6#b`XhSa?7ut&K?prbX@bqP&^z7 zT4(ESIpPupR&Z*Ct!z$Yud{H^(M8DvQ9w6=aI*eFX)H)zKIsQFTkp!a8$4eJ3vT9E z6JWS-9pmDarNAJBu12@2$r$C-Pm+19EW@pJ_eD7kMw)J6bNy@TOU{s9-db4bFVn{N?COGHeFHpj_kyz}? zDRkegzC(?kZlvcy!^a<5L6qxir8+>j>?cZlF z93l(+{-RL^DqQWaJr-L@b7j-7XLVEkI)nv)MBXI-DG3 zLD9HY|6eYW*OEV*U8<9=@W#3bgc;WrR`bX2aTU>NV-iA6#U6r;Z0ydj!^-$8;TS+<1mT?M(z}+-!v&MwEh1 zSs;x+e9M{bOW064u5-$?C{vr*NP%Z41cA{A@KnhU?E!G{u~U(^qlzh^U#?A-GB*XX zGk|>f1Lm4bVb|?=R$f5EvzlU4vea_>X;dBAd>*gh7FxEL3_mf!o&m`CC&HEH|J1|! zH$Un>)Q9?i+u&dPft4}y@j8`vYtKQyEIqDS;>^-%?MTS|?6)?0Q|cRi7 zX_T=@WksJ0BCLn1)V@2Lr+t|gL0|Kz><+8ce18&B0Vi1T1_!EI{Ehp)kB6J5nZ%rE zyL-~~T^&vx6t<8OKG^c?ph@n3Fip)Nkq9)U0}#n4Pl8xR7hDPq+~D4##7%TJ5~x2=cIET3PcUyEmIv zfTN#NU;QBxw<#fG9M8PpSs(CNr565yvp zgy?3WY=xw|1tO(mJXA7Hb7FAT@x~?pLFR+*z~~-Osy5o6#;$QY3`f~2kY?+NXk1Sz zzrpamS$lhUnSV8q$mHL%8-4i7FwjJv?2=*OC@HcuyUow39N_YM&Pc%AeX-RJo5h>@(+mw2l_@33m7)_7U+{Zh9(7Mmir3a8n6?v`LYR3| zJMUh_K6=ztahs!KxnMIL`ZYVAJNJ4^L2PvADo+7t3)JEQI%U-;4 z7y*?+E1(4ZcmWMskg`66it9q)Bx%!8&6XdoebejnEJwb^VHM?at_Dsp0w~6r;`NSr z);#;QSq2J7ULB>-(+mRa66+QBAC6#~MD31|1TXMpOIVuyo|~1HaehAgvQ6gRN+OrG zsv9KD%4I9vuK(HSV9iL!V@()Me8YR-@LKT6nPmTH{K?~X8iX4>hK%sqidb%zcKn=|OIw2Pi86pgh{bhBK??6iiZks1ir(7CySgi1 z{1(FBgJ~X}N)8u)2=u!l%UP8A(|DFC_6gxra3M*|T3=?cimx-IxNlxD@%>u1Q7ymX zE3Lvua#>WJiq~)F2`|)@Zd^UwQ<_S^e{Lu-btETuz{=%Y_@42tzXF4Q7@67Z2N$c> znHdy3y?(YqdAChBvGu9cqXo}XpTg)~Stcv+T^U4LaB5PpKF69YJU~$8B>DWan`gAv z`#4#SdegMaH~3FKf`Uo>v6C(Dn=C@7S9K0Gfo3z%9rN`}t5eFUn|TI@yq#oURC{Wy zV-9j_#z5M(@7?T-JW09j0K%udIAq!0xM#(*kdP>KM!%l z3t1nfLr9=p7oo+=8P3P#>9I)_$XXfG9RH{7D`^4p<|ZgHFItjW^>2!KRnw3EGZ3Q6 zf0DdQ{(J2H|5ttkxO`7Q*ie}6bnm@E%$_I}{{8HIOKQQpnR%Xe`72=!-2*KbkElAN zE2Ub`$;)&D>J)cyik^)!<#D7{fpn|GM_$NxZTzRPZkxCsd&cdacPqzce>8mTK5X8x zKY#UcPIux_tqG>!9<$cFYF|bpWox;}W~T&ktH`Ofcfy-oA%l|z&vFtnAV&uW9~-S1NStRewHLZ?@f>kI z2B2VW)w9qMP1&V?SZqoOLJn|VA1u*?_)%5}RA*<*PqGE;deEj6h2JG5x$hkDpWMQ* z>tU}rbgp~Lh){AT(A#juSO)+$yR2*<2{`k}C*_ML0bayV+T~9r{al4A57$#Y7jK&H zg1GwmXB>D&XFk1!2{8NF-2$*F$;RukucRE*3N)*4Q;ZuS9Rhz&x;z2g3r;7pJ#2Mx zQEN?g#ftWBrta}$L#`{NWmwyPP9!V^ThKiG&a+rY4KMVr%o@<`+@^SFzaVxgZ=R8+ zk?f3~Gcw`$IZhUp&1Z)2w>nBu+4iP>-peW)B~5CpH~s|(4$~oD144=dCtOMAi8tK34 z$1l=|j~|Y13u>b^ax)$Tu?(#hZ5gJowa-Er>4o!SyrPY(Nv)2+BgQUenWBqJXg#0TXW^UDIm; z&x?IzBxmB6tX&}=soDYOZs80GUahGLrp57a>UyDj4>!I|DnA=`G-&MSj6b4 z*7!%S2$Pc0ASv&+L7VLFnyp825U#OXD4NN3a6o1F9dAtcx$=`Xpf=mLBbnRss<^i{ zXjv#hMftug6M#I^t`u){rXe7zqi<~&Js$j%#&Vbp2vzL)CeWF@Ar{pqJQY%Q%S`E= zIsS#S^ik5CmJHJxE0G{-SZ;M6I~7uK)LR97AgdI3`OhJKk9-S-F31+Ha1!P0VX3|Eo!QM?wN0m9!KxMjDQ zzP2iVq5d=>4*r3^eP}t1%Y*^?)9zxt#jA9 z=bXFuS!eGb^T{Mk;$!BRW4xoiFGp{_!YGULyo-e8P!d_`yf<`gz^rjk|rQ$-Vn};2#xYRQ`CxC6J=kO*bt<= zd+BY^$ib)t)w6akp`4jrvu=3#GB7u1hH`r3ki!YG&EQbb&3}Z?c8uxZlp&DFP{Q{5Y5Nfwi+!Mw6Qv=f%UUPcY5sih+=HJVm)x?EC3$l?3Zx?RCt^W z=`Yq|QmwkztFB9AJ5cVzZX8dF$D6A)EZxTGwZ$2k7MylU)1E_}J@<2gvrHoYp(0Bc z>n&DDJ|F<}>A*D8&cnR2QTm|<;!9X%ToC7Tl6h*L#$^M~SBPGPaht)ehFV06Q0dQi zrMLL4I1wI(iAqebmasWL*)OK|Y?ksLNC4LxB3kI5^{tb3bf z)gPH`kI#8*nZwvv=9Dd$dRYhAC5Nr`yvmm>66_`!?S~}2c*qv`Y^$}!?_u%O8lSY8 zKn`znJ+XXv>ScVch&7(~Q4ak#IgrkmP2(lOM%B{nEtFpWc&8F0mAT{+I+eNYSxip$ zaQ&b>f;L2hmGNTQ5jHxw9l!h#-_g>lGDFyBo-#Ol6kgR`^GW!e_l5nlsmaKgh!c4( zA5QmD2~RiLp0?glmLCE7F^^!UFEV|eSXK_{aXPG0oh9El;mXI4Bu_YWKEDL_RGjj>%IKJt9{LqW=^6fo zY^O67Y4x0Y;a;=A!;Beevz<5_mGv;0>dshJlG{#Hk`lL$ZR?$II+OmQl2P+hnf*zn z;a6o#SsG?G0et}7H!h#%4etF&O#hJ*HL^NZ7}+WPcucGypf;J$B54wn$W{*?TrLC3 zR-Ecz2`htfOPBPKl&)|;c@^2Ph>2x^4z~&aS64l+e;8{#n4Zb{ka#j_|MaRYUpy8i)_hE5_F@~h#49Osii00J5YZ|9ic5ybQ_mkRs3^t zKk|%sJzle-Q|(m^_~N^6S>aJrHEn<{xzJT+vNFcTYd&DherHm=Ox>B+TO@?`fGs-H zq{q(f=POrW3#CwkPxNRdXz{w3Y|J2Ugx)o>7>k;>u*&YBbw)^1R-#NwwCCg?-6N;m zQF`opQtE|Iyz_eWW!D0KCQihGPPYhW5>?#0vYGmIo$VEwX*}pGu{q~GpNx(&3YkJc zwf{Ujy}8}n6sOgnY*b`e*%ozRY-7EJ#~4hXmZmkYJ6h@-ASPB4()|>Mp08tBAU4@VC`-uM-k^&R>k{rjm%^ zda#|j?zo^g=?vl0Q&H5i#I1NsOK;NA&M(@iR^*LB9u+k>*AZ033*4jy1*FDXc0cUM z_fOZYck%5(w#!}I*J)m^%cf%)VF1}iky1EXtMg~ff+u_F-p{Ofh2$5Xj>Qe$5zq>; z9nXxXi~a(tGNj>bPb+N_61rK5R5z)TF*vI`3HQ~+r%u-i*6lJLlU+q1IZ=oNhUd(7 z8GM}q1h1#It+^ar0qo0^r^m?kDV>cy&<53xOAQH#Q`At@>;S5KS_|B578`i5u6N#x zA8A+ETe<=j{0r#4jmxVfHp53*j)4a)hb~XE(AB9;Joy-cK?mO#r;`$tB$t0Mc%dxziXrD!eJf4Z@EXp?FXDU(x_nujEFW-NyE@CW4`BOFgxzrXC0 zW+3vGf4~tu`ra%+EWue;L5C5FX;pfJ_19fA0`4KJ^ybO}|B^Yv6+_@Q$n7(LR$XT7@2O)_tC(){qwLq=J;-^9== zv>~1U?m?8^y1h5aJ8NosEhhixQ~Hy}GNG44PIt{en|Q+m-7|Od+hFuPNz0kIcJ8-9 zSXukJ7t|F<)~MWqd&tihdJ(U`rWiM#Dd2+-haPpslY-HwB_ro6JYMPvK*-85x}UqiokXkVMHGG1d5$#B+1QvF!4#;bAFBv{Kj z*5Gw0I-6IMI0DzrnO_+ObT+@C^E@)#BADdNX=S<&j!lJ3+#oWOUnKC#KIfh~w0ySq zti6E&FUf)i5axR{j`;^>V?!isb8~u}f+(mc1P7Y#ZC(>b56yRL+lsY#Uy#0aSU<8> z)Q-OOLg5%)?&py?K0{AHqdLH&&F8=Q-#tTeGbjOIxH48s^tA)kkeIj1RgOtkUm`?>c>%59Q1ENwtGTZ>OJ^>Q!0@B~hB`^fq#I zb4LUm`y4Ts1-y{z?OH7@hLSvjF{YI z)R7oPd39#SgQAu0*Wt&<@dof1UX=m}=uF$`y*+Zcy^eHc#mhRqpZ#6Osd_Xj?ZOgG zg)VC5YOV&VFS-?bU)ipm(P``T(fe4Gtu3xN@Lo1S@IwikVkSlJ>y^N)$%dC{&d(40 z5iy|pM$63p^|0-PgFxN7Nb^p?_%m`I1ZVu~Y+Aup93WB6mpA)dzV zZePTP$JgthFQEq(e<5O%B!R#?^kqGsv${{luq@Yxb(CLzMBbQT`6A}jZ8m@L=Cchn z-0dV|>bwbEPrUR3rA4>kSc(8CL3L*!g_KsF|A}YcH0K%=jt53==xWbT2NC-_c*7Id zqBHF^c*YmhrG0NF|126E;V^vt$-jdTxyNRI(WQmzK(8Wg(SV+aiVE)_v?I!K`plW3O_d;AOrrW73bOOh#GOwAJej8gjUxsjdaX-omG%x?O^I{RET>ve z$Kn3uiUB8wfW(<>$uq-C*X0_DvXxCK%(NwbSHJ#QTl~Mm^4TA)`Ct5^e{0X%Sj=!; zZG5Tys{f7h7*1vhrR0iRtSI^FV;@zxx_LvP+C4Gxqp{4bH(11v>>0|T&&)H3vI~EyCQcVb!#i~y%O#3s8{Pn>r@uo9#ixETY0VgQ|hqJW9-v|X4MtQ zQ02 z`eSx{(lSSJ8!;8knhD?WAd1K`q%FMz zamPTippB|xPbPXS;Q2uf-0_6fEd*dFqW9M3{A_gJw86N_{|7$*Qo~8@{i}snK7$3c zg|PT6P)gBs{k5BHGNNfGb6jq&oyI1MM*N3t?%RYap=)p+qa2~Hb*nW|s6N-1_4R4b z^|r+S0%8$vD&6B{u$9{Nh)f~o9z%E?Exk*=c z=^CsqO5*I}t>fqB^!?-#aeqA+I{QfvdT?zQpM z96=y13PCwpQ0!}yUaYq4(tSEE`p+%8*X@Pz)O>;!Oc`#=YU33SydGOZg;% zYme0q6Z|Ol)lXh%@)=n0IWH0+vG4q$^|j!$ZZ1hVFKCY*m!e8qh0zZphJsIOKj!iGm5QkPf%OQK4`hzh8}Hr=B^3*9xTn- z)r#Lmq(GtvrpxRxRcfcS#qkg3rjrI!x_v7SqBfsu5gHnCNuG|QRp){XS@Q3G-tZx@k2RlGTq z^Mj7wLHCPk;(|_hD@H9r0z@VUt-?n&kNRL6yy}s~w{rQ@&nAAkl{F#1>{DkOL1%eO zUZtM(%+E>M*6U^ErO#A5-}3~&4in`SJsi!myD|PV^z^Q;ax!l(_Wbz->(hv-)TC6& z?GF?2^*BXavaGDr!97m5JBKI1D>JFz8l@@WLPmT8qd3!54g=>gbui=|xd}>1~tO zdK&S)_i4YN&Pb~=wl)RJGCz=Dl~p>j75-fgn}~3HP-P(}D{effVid!8Gz?rsyeKjzQFNh(1A5bPTEsowHqYqi zN0N(i(DZX(A?eGVA6pqxt~wZ{o8+x|j4Gs*JNliR4oO)D5uI3-kfc-0j-9HX$*Js%uJAeH~4tXKfj6HfHehYdzv)#s%B>(a< zeS=qRasA7wplkV=_m8|dQ5<@{%u=*MOc)#~z3>> zS}kO-aKuFh+W2~bK5vaRpmf$t#}&Wg53N%}&sA{xxcn3pKYHdGSiQcfmif*R-WSk4 zxfFECFy$|x_SYFxI!Ly+{vXbQW>}Xg9#NRKxy&ZAub|jYH|SD1s04GD3kN+VXnYZ) zR({Qk(Ujoo1U=pqgXTUABNIL?aK+B5nqVI2UZb1SkvzFKd@sh#TtD1O=sTa)gUVUR zjrJzp5_9j)P(e37zU!FhS4obJG2(nXy9o^1*vIt&Zx$Rh8&up6N-Aw^Zl^ZB*@t9L z-qES6wYJ+9pN~lKr?S7gQl*u6MxIdt=!WA+;vC;Dv*+{=^@l8mB06tj#&Mqwo{bD~ zQm1VjtF*Or6|g;&EC}8df~R)6s>E1%INVE+Q817mcetI1cV2qXb7VS~-n_O6ERbkP?NpW*$ylooadUMm zRWxL=b(MY$R=YhSgx<6>dO1d0K*J9y-4IMDm1xYdKo0S;Jh@am{}a(72<143fHsL|*8^*T* z&Vf_1`;0|oowK4%m+r^4?w6&W-Mr%z&~kb2wmt{D?TD{ zgzp_%GAw2yxtV;>*1{#YKtV6;Y+^C{hf*Q8kxL;cj#cy-YvCRCtjoyG@X|{(DvV+R zeN)3*UxLsSIY^OO+xoh+^~eigU6Dg8zk{-kO+#9-!dN%1z(_jdW9r0PbeK>d?_wzu ztS0tX;q9MCt^WcmDR2J7)c3cH@zSf3<&QtkygAj(?Lsig7B6`Jd_cINv_*#K8h>6U ziYg>p!pnL1i}OfrLi2Oc#-zg{*ZNvez%Gbx<`zw9xkKFU1;{227ejR|%zW|MO?_{{ zy-&D~vMQmS>jLRND@7rz)pIfBVbM_%z>L}zj;&bn{IUHNlXOZ$LBgnb7rUx-t9q5Y zSEwV1q_DaCw8QDL?I)i+Vdf6CR#vvK?d~q>XJuSH3Rlm%yVh<;ms~3k6Yvm5#aMZf zxD%g9Wr*`Bha44ovpp&OWY;#^82sa6Ou>A5O4L<~kM0>F?-ZW7!6U4z1$3GMoI>gt zq4SOv_~}b*PD~FXKIrq1VG1i%OxcQ2TUsY<113uH?VGUco+A4dZkc!)y^ha>$iXG{ zMv!!esLZ=r+7|i155`m@{UYH)GT@06ocHo&jG(#kY)eJ8H!EIWR5~8%|LK<4GfL~# zS~gsTn_|hlS>H`}=a=xk9@VmOT5Mg-y+v>?S75zA!fjNT?i~>kt*3h2;dw-=o0$yi z=>zGs#rplQ85$cgPU{(*E~rBCF&r8)X-O6V$*+CIqrn7DYeCiy9$DKdoU zYFE`u??)MI)_vq-a7!Xc?t1Mbacq;nI%n3I16+a=F|JX^KRQ$SWx2Jdj3|n-3=%qm z^E!3LEWG-frK>RGECuV8schVh1vZ=^k2L=RQ2k*%V54=O&L$paoxL-udkQ#g@>Z&JT;tcYeHoP`Uipp^MT4<|CZq z*U8lQZ2(hGd0jC4nMy!`vc5%q=u;n1`L}1ur^KF~if`$l z5Dfn~!bi>f2Fflt_WTo$8k%B!im|vP^<|eMicz-@SN_G}oHF2R>iFT$6b8WqYQ;qX zgpg`?UIkTeq14P!YDdq`Su2VFQ^;2RA93k z64wS$yU>)}vFUA^b=+)OE2-rZ@k$J&U~N$DDrr_?n>K|q5}m#c07MJ#TJOHYvV4E~ zk)wpP{-CNUbP!ndt{-g5m>=E!xh+be}N+i{)u8UA*3}Xk(gaq_pLXipp#n zUdl+FFpm?xTn)I+c)SRQBw#mw|Xclmb<-80T8%5Of2w8|KfN8V2?xai*FGKB|XS4-n|U5Szmzom6Q zny2=mWOJEO(uWH_Orc>fM8jfUj?P_5N_-QHT8kJBt&32i*qQDTYF?Oj;^Q#hp88@!y*<% zWMx1oC#wP+BD#KiE-_;>KzRQ#T0{?Ozp5I_p*W;>MqW+M zA&cFG#RHm?Kp_hF2?SwUt8t@$3{_=7)B0+(_6ED1>w(GRaAYMzgmSaim4>w5X=}Xy zCDWAm&ED{Dm(QdM+p*Q6U9)tN}z$&PZi)Jw)e`QzNOAJkkVJQ(kgD9MH2P@s8kqYK+kkPSl3 zARTkPKNudQcXt;Q#`{UF7HO{EJKL}@Iz;n6a9JGKAmNRm6|3=i9I&2z6#ZX7k$k*l z^CT^)YYUL$&-S@zC3aDr(VNOqq7=i@Z^|i6Ze6+M6V$r^*JID0T_35v_*^b*;_XgS zWdyS-Vx{rtQ<~9R?35!MLcY}(+m{1Y+LH3)`6%hBSoOLBoF`Y52P-+v?jEd!0kR!| zp~WBFqjITr`IziMdVWG8&YveE1fpw$WifAGOxw1u6zGAY`Ybmm1GkwPC$BeOs&!pj zOq06pPda_r2V{g1R+iCT5)Kd*$A1uS@5CUeqya zfcR0XHkK}oIh{Ux))13(f&DxprdhlL#$%bFX@%`UT4@|dSYOP};K(?@V%r6E{M#cu zfp6X}{mQ4lpA8BAMdy)7K=^L`E?)5M{ku#Y(9L8+i^k9k1Er*44?E@>*BtoffKSiW zc$et@%5v1HG4ds)6R$iy-yCiHLt9t2)2q7-uYzak7%2`d(=2u!qUR<<0|v;_Q`9#b zxFTjeVLvPSB!${|WzZk#9qNalyL}M6=oTvl;ql>OxpAr2PwyVRQxxKrB^guq-jG%G zs_-SB*;YCj#LBI?A$oHG=u!x>?rGJQYfl|W9z#*!(`JHK*EmPNA zrnuiSA&YsL4T*_GR-nbG3Qn9Pb;gUuFJ!{krWYG*cKfFqsd+*L*R4uP(w(}lY42`v z`sx3g{rSB?@q0Y>kI^CktCHkEHRgY>)N(4@bX;yi6uc0ppNzpk?Wtj=5kvnILEXRq zgg?8Af7SUPu=a0|7T6Jl+RPv~&ssDz-P1oA#EF_bAnVi6!p9^_gjm;xRRIdbvq-HY z`H?VX%$U&7g{{3Fg`+nz;5Agg?yTecFRTnkLyjhVC=dJ%)m2AM(J5)~sl3M0vYjA7 zPl9(tF~9i8P3Ex(ri7CB7p(1s8vLKOG9Tn~9L0xMqB&AnK~pICRc4o%|@2Vw3E17-1Lh}lvkfbUB5v^Zqw;|noz|`jSdO0~+qO%~`D#!GOj%rPWMC!w;z1R$OjT0C1lzh&dK79gG%CE*!;6&h zMIw2YP0om4=Zd7JdZH)0S`p6Y09W$er)pFddDH=R0i;n4fXFw>W1Xj@Kp zb{Grc@LhuG0`zxz64hHlN%dLvYA!Wd-xC=t$k-jBxxSrl12Od}7eN zAe_YU5XPuTA!|m8CyBiK)DR3oeJiF;y&oXKm%7_6e#Cwfyu17&e5E`!Pvrx0V2o&c z!4*$SF&PtU*RRnhzi&kM)}>G7KH>Gv3!l_6l2l~(v#0z*`joMjWm9T7p%r7(oMV1u zy-!Zhw1$dRT=yOaE_VrwzEirC5i;bN^jVtiUew&BD`x7eXP>bQfM!e=uzMU|_AXtx znVt7oBn3`W=v1)#S!K*<3%|GSwOBv$rI0`Itt(2f;W0S((U!Ejg;8nXm$OP-1MF+{ zzFmDZ>`quETxa&TrP+H1vejzE6+@T}?2SSk;@2GeKU=H%^XLf(mXMY9Z?h+VhA{S` zL}%c6gmj-%wDcg~xjzxXVw4)v2m?+0`KtdG=S9t+h+A>N;{$Pz{T9M4vG1A~siW2h z%pN!zh@e0@xNj0_Evv6-vqL~~DR&M%Ub;`7Sy;}J7ZtJ=5fR3TE|oHc7@$~aZ2-){X@2ta{Ec1)4MKXF|XxxDP41c!w!tSF? zI7Jjf+qO~unSuoUYagytEQ87Rgzz&)_A^iCL!`H?gWHdP8v3)rUSrBw0u48k`a7O~ z&i=@emgOs@y?z*JnW`-=1+aIrqKdw!QFIk-M!tW0gDcI51dhvgIaeQ(*losta^XtV zm3nhQV9_>Qh>(|YDugmz$JltTJ-E*Sj|||Fd>;Y{0T@JN6zapQ*un#~D#Z2*(T^*u zI1!NYqmq1AlNUCzdh?AbBtAeGMJC^;@I!zJ3v-u5d&>?Jm+sBOL8_42nX+?^idztR z071o{ZsTlA+HP<6<5*xSAUELL&K2h=dwFn}a|tn3x%HH$Jkblr~Wp^(D8qyyKqwpGWeS7MEH=3_!JO2Ivq4{J{~#Xa;Bx^9@A z=O$R31`w`A4eO0`(@w>0*3VilZ(HJ$xO1f2a~oD2XBU>`%mXRj_ZQp<=_q+Po}S)k zRftRd|A?!99abxXH4rHa@qZthe}zfP!hc(n|C`n)5r78Da5uBD6jnxE*w>lk*&jG+ zIXovOg%3p+WJJ@R0$oN^+U?rF;Uk+m=a%wbIr2#>e;(wa(h_{g6(IND;W7T5$Nb7$ zD1IIJXJz^Joej$*y3a~`#=P|0B{E&zj{50vq*PZ6HdWDkogm)fp?)I(&zK1?yfW{L z+r3mSo}4sJOOMs83+M>(Jik`Its71fyJll` zOHyx`4|}am&p&KajNDm5@2fpVB41ewPEOCBKi8ONe)FRpKx*z7N`5C>mEL|B%p6}= zlV%4VcqD$a^V#E<7G;ilU!Kv_?}X5Ewil+SQ`q?l=hhpDqrS_t!7LO1x@h|=n*22% z?tp>J40OD}(gp;tz6)}o2jrGM5I%CajH8uE$>&)m`Tpb^g0>`{eF@7EllMAFWUOl3 zPJVyx%^lE=?KigJQ$~^_ERWo75;(QebgP8F$UP2Z|5ily^Ep&%<9JbP%^V9spt@cU zSL;%`nch~wg*pzi;GE8}7vj*r!E%7MgcKd-ijO(9zbH6?FiK+4-$1t);_N{{XYx-$ zj%Z3Tqz$A0>wnt;qo~`*os)KNW-s3%GqK9rQfN%ZE8_?&dpqDP|#ST z_Nfe4H5(;1e0AvwrR%zRmgicaNtISmE-q}oSwS%5XFIcc!uh?1q3@I3eR*BNPp1#2 zH}Rv{a3X5e@4v9gFT`vO-2X9U=x%oPCU{kGZ!6HvXgX- z5j)Gp(@-C3*)nqaQ?f>PpV0UxVB|_yo8mr>^4#Cb?dI_kfV_^G#d)>jJQiKUXgRPW zg?9)q=4lvmE}c+}6=0Eu6$IB^^<0qm z54Xa#R(NtQ9X_uF_c2#KO5GyQfJUXG0srRXrB>a5LmS~DZ^opWvTlJFr>?WFg$z=; z#|A3NmQAVKYiS?r$hyyl5IndZYBGg9KL38h#HM*n^5$fH&t*1CXPXtVJHCS{@j~B< z_sObW#lSEaMar+(K<%wt1>?zRQNJfwsZWuuG-v`h1^0=nxWa}E(CtpL9k7}|w(9@7 zGFfAWe)ETa{QvPv*&*@otSnL_;HQ5E82^twAy5&Ddf~gl8^Xwe!;d=OPBJq+%>1kt zIJ6f=?|*-Gc}k8Dk%z!_Xa;%};=CJW1@e?eC1t>6lS@BeC!rTSVVyE*IK=NS1cB~TL*Cn4wYWXm`MDFz&vLjE zX<1G)v%jG;T$M8z#v#{l=$e<_)l?!Tv=uwGO(R=558&OcHxb2sCj%918;7plj z4Jn0|N6k0nzA%t>f0)O2?*W#1+b#Aij1(IYpvPzf<3)!NDV>yoBJ*&pG3^!_)di!W zCAGOu>#9qAVaDeb!8qquP-w zaOI!n!T(owVgJE36n|HN{N-*a{zV7X-a7(S|-nzL<4h1y>?^1VP!sMyl~ z3{780W8vS37gYAhRtL%InUYd6XSa#2x~ifO;XK_Dx!Zx#Ig!I)a!{Y7#y7c;$0zoW$Vr3&=&;NGO#mCDd{-boPa(G zm}1|C!E3=1xhXgygX*9<8x%cb9OaEC2Ib(wF?fOwl@xsWHf#!ua7Y2L#Mvxf2c=k)aN&6}b2zFI|>Y32F)X}N!pgy0}0i*!`ib$c{| z!~KfJMFRLhzYq2K`a3kL)ZnzClfHV`2;GcD%RX$0UF-93N+Q(0o^$bR-Y1`zSF_kc z)SCu)VsFcdvTI72aIZGil>+p9^bZcRUE%Nrj|c6(EFaP*Qu3r;MhtQZJ^`vF8BF$+ za(I^MjA|EZFLVD$&daiSdstIOe?ew)qsPc)IKSSK-HCQb&PA^R31V{X^=mZ zF~bcYHl4Un5S>>Uq;Tc-KIP<*G0(AXBqCA*OAuC!G!HFVxUs)sG|xfQys@DNW*Ez@ z2@#pjed;N)L!{)^r%nyXj|h3=P>ev&AeDD*=#K!5-I_#0PKY^zvCPqQ1k4NI+Navm`ksMKybf)CvonV;hANNf_|k-Z+^L# zU$$m-^Ttwcjru2~T1GFpl}P38juc2MA)wzu_Q<)eoeRYCHid+}zjF`&yVGa?aDwa? z<=3ANkN?yDE+-DP^w6Atew(Pde7*Vn)&6-cmWFK&B;1t0cYOWB1j*XG)n>aib~`P= zLhOs-As~;dj|J|tM_9k8x~!W;%RwNpTJI`)D^t34abXm_n+t-r&vv{ez+W`@%8c z6;-o+*TBKhX^pUKda}`un%J-x)6*(eWDhlIi~!Y`>jja zLEQs#TlXpVVL2=*qF@w+0F5xDR-xy$59mj(PCwdHPlB9Qacx*PKH%_oH}Ql=>z5kz%g2!#o>@X`jozbHZZGi zFA5eV7r{UoFM-;7M+e6|EW?x!2$z4$$p8J^cn?(b@0I(1wp{$DivDl^_J3^xh1Oj3 z0Q|jl%~Y0okEr!#%a!1rf2B8v%L}<>2wfsx{P?q>%r@&vX2b}1%9JOTl$s$r+44=U zQ-^u4UwR@(3fb?C*FTX(t73_?mdqcK0uL%)Xlj&UHt^dxuLFxkl|ys(Cn+yo3Wb;h zlF5RKWhCSpBNYY6U$t3OWN~IB`kSh@azTeQ!^T*~t=fTFkW;#`;4qdePGbep;`Y77 zcOU+9n1ctr7WSxiX4yklIkg>clN+8rXEX$REb+1H#VhGWVC2FCrwNM7BF9@Z^WjwIvQJjbP5{1b4^~_`K zF$}O4(T-n8`kVxW4uS9jivmstumK%W5YbECMgS9nDSoNdMz>tR6s{_`iyqewAv$2D%u)NQ-KJyQ2SV5fPIMlFHo5P9WWB zrhwrCKEp;id@X=+G@JnCJ8ASimdA`={Gs;dn)t6l6WLN#aVbnb@GRAaVj0<6H;QfEPRuo5RuWrGh!sajZ8MFyEa>SQKjAog7p#j>h*~S ztZAj6Md#UQ0|SkOxBmhfE0%TFSk;60!*QAisN&7;hq=+;h zA_U+ZO(D1FF8`1t@?y+lBTOOKmVR$NbCk+R_};31W6TDciIOlioW(H^fo}?3-O%ru zp`$j+6ikDXsbU<1vayH(L(U*q5B-HhQwmSX_U#TnK@Ld0+#;m6UuQ0Y(|a9Ye%E3+ zhX9;{@(|#k7+^dNg)s@Dg#gpN#JLMT)FSqd!)EAHx<0>Q^%k0nqYytRtvo#FMWlj^ z;RDb`6j)Xq1B^b5gQGx)(Hm-L(a`#hNHe~8m-MTtXdX)bU$(3N(f)qh*)LoB>+w%K z;6L(;|MB;JC8&RkvKDj507J4x0(2UC?&h-Git*Ta?-6kdTK4XKF_oj27DA@)hUs#@ z0^O&ohQ}kH<&_t5J7W^#1QWK*dqWtAUlC8|vpNLp&9G5k6bcIq+nQw_pgQ%^rep@-&MnApOFMDU(F`W=5%XJSg)Jwz~uN zPQLYHK#49*$4lLwso-dp5l}!+2^h)t>VflkPq`#(=uvj55WFC`l(LrZBnZH?MGcJ2 ziaCaKLTgbrOZ_NvC)Y$wQl}n#cpfT{5r-cpJSgS|Ex6*$v5L*y5i?9x^cud0!6YfZ zYct0PmBR4ulGD3RJ4)O7O(Me&=q%{;HuQ>Avah0Axy580K9Kp&op1bBnwybFrI_AT z1&;Rhz`~9FU*(LL*fZH14pMO1WBw>J|5ZLAqsAp=kYZOKfHJ@O%cr#Y$emv7p~T@~ z_#oigfOn&7OaatwpdW{(Lu>}V+UFKm>;=&L->QFq!s!|YdZNt0Z~n(9_oo`%->&{6 zTb%ggm;WRb{tILtEWf23?3RCnXFqY^bkypWjPgP&95c>L=U74Fk*eaC_pZlam%V!` z6D})l8x=_Po5F+3+^+RPVktw#ZWzWfA2U00*QS+P+mZj4?@x)iXuT!(wiG zzVhObebL6#YL#cWosHAKR~O}@re8Q4IB~{_QAamI=iGynnxB@!!pPF`jiCH@rVtq= zPH(l*jgO812fU#_lN=~MawLGz9JLBixR2pa_ksBK$|$b;x`DTh;7qHUF?qMY!9a0@ zBlCqknECM#p>azC5b*&s2Bhk!dT|I2nrBh@1NxCny-}r5R2Q;1VM=y5HH@QQ6daV- z0DIU1bLHS5&wM9QxYsj!%XZ7X3=zO5V?YWEk>;!@ir9ee?$TsoEZotRy_m6)MBV@f z4CsCpR*3}Q+5R`E{u}I^li(u`QVg3-09_4oA(r*NT1IXu(X}Z^U&kK>+;ynEG!Pl( z3e&P;;)T5fJ`1Gm6CqXfK{`kh@z>h=f27#-_npmuWSRdvS>}J>TE-t$D;f!XHrItb zv%70<_aV9Or`s$xNOT%Q{YD4xJY+|hN9}l)~`PzkfLQn<)dA;!u66i)xNhBwTTjOvT1bqS_ z^RYv+X=XU36FP&ksU>xzYi_4Vf3Af-_BtG4KC>}6%=3CUV}7KHOXElsEZZrI19!0# z8M=luav<4G!h_oU+%a=G3cGOwwUYWI2yL5dEdq)U-XqSV1X7bx1B+;yh~TAke_j)x ztBPd)r4*WhY{RA8`c%DfXcY*chlZOC%%Z79$t#HF6>!)u4BbZ&rwuv>;P2nanh4~2 zuwP)hCZmQPf$$aud;LOM2C4em5FjIjg%YVcgr&R+u@HTX_rI;Fnwt;1#GC>G)bC8K zLl}Tjkl!EiKgHzw)7UeAO5FYkcK^~K_vp`(kxM<$M=X9g==5UvAF1X`mCe(+r z>p5$nv~ePSLzO|h9}%e0N5S3yg%ixTjWOPO^g81D<;c96PWgWOM;hyB4nD>jY}Uk? zyc-t>RImEYvQB{Eo_Qzx47b&^8>7?e>wOG5>BoD3@MV#{1(FGxe$b{G5jmWW)r}s_ zgi5SxeKLARxew$H< z@zxW|&Yzz&%D{HbMi-5}->Ek2lR7B4;&d@#`>>I`8qic40$jqFkPZXg_<{&oZY0C` z&}1V^gvbXceeMRP#pbt2=m8> z{~vq>2eR?Hi99y4asIGwoqnRq3$I?K z{c*I+9lCBQEaERx;*4f-6)8Xx2{QiPHZea2yC|@yKX5CJ=t$9aMwL2&55E=hWYB0b zMwKcx981{KC>19JOBf;*XyIK$+tz;*FR?(2t0M}`3gwt>qe7jG0^J51ej8Fs+DfMws;Fli<7^0)fN~Z72 z-5z#Q6qLKOZahF14t#?>cPrC!8MR@yUxH~<2`TL%Qt^8u0kgmW124bOF3@Q=LqmOl zs7AG+C*vInh{EC!UQ0mK8va#Ro(3*)F(=e;Qh^Yp-<5-3v5dG!rpN{cg}wfB_DJ3U z|A@w45%wQQ`(OMDP7|d#uvpO1{jQB}ag9s2p|?x&|CFF@lL0714L^8zMec#zoQ4Pt zs2^%{h~;@p1La`s1#e>bm$u=WN8j<`bI;DZo$`n60dB|e{1pUzL`&P-GGmp1DraSS zF!O%d^|{JC**wL)o08s?()p=Lt%!2?)-a(Fij-p^D8-;@V@6PNG2tf}aVpDS$&@oo zF(9EdX^mEB9b!F>Fv?v*hHy55JhN$9RC`FMjtg37I@Dz&5$vawX^f^IxLIlj&bz>>Gl5Yd|GH0UmmrlFI550* zE)E~%4aG2hBD(nNb-@B<2Y(KIz)62N9OUDs?N=?G3CdHeiQ_gBBmAJW0HZ0*T}9ZJ zYbj?U%?2U;s$T4Hp19Oi{eIb4pn7E-;ks_*B0HK5GTfg_$$xX7LkLIP(*c-@Y8iS> zTbZ5L-V@cbsz(XMX?pAL6@MpcnrV=T5VFcPJFuQj7j(P^aw@zJGo_$&1jLkq`?`*9 zgPDrTfy_BCkD7w@r?ap!ULd&u4fuW^J5Z$lWfq5K#@wRd4B{N{gPw2-KEMDi2cjQZ z&t?GTB2G77Fs{i(0*J;4@GqJ}f2u6!t^z>kcN6qK08Ia{{e50f=%r3&PkAaMuSH21 zp#~hWvrtpj51K2z|JH3}=`SE#9UqCD5$548xrcIK_FdQgnvGNoBwD3I--Y&yeA}IBJd~ zC|S~YSO-OdaEJFnO8qk0MT7=#WkNwtfvVodV89+!;t1M0LE|I+Y8h>vqE77_`V;AL z=FAy|z&*9t3cyTNleKCMEcA&koT#F&BqG%k4g<)w{?4v{0nwacU_cvE#lD4UW&rVb zhQUpZxVJU7(6oTxrx+gbj;;aC|33q^UEMgoEm63@cx|%C-U^Yq8VtbEgsMxWz{K(1 zaXAyO$tTcQBg#T?T*J+SPdq3IIO&X@kW1W!R(@Z0fe*MFYG!@)ZrRGC1|?sgO+76V zW_w^+32-yP%Z6*Rp3Ax>eA*M@(&qqd*(dc}DqX_hnt5ZwwN&foIbo^sm;1y*&s&@| z;VxA9{bc^RMX&F76;G0FkIP)rQ+V*b&qBKyOC%Zk=6zYBFClTh%7H&9lY_5-Q*!As zzvB5z)oksx3<~Fbe(A${LdVf!asROiPkQDA&1*>bo547f_pG(D%&&`me^>nC(_pN4 zeSRy8-14Pr*1$U@mPnjunfTI!mBHNR!kxk{$7Bz@( z)#{&ciLY&H-NOYeHd6Ujyoi+=0hu=*&Ac~9dMks3!e@`n<9(nz?c)zkxOfRT2F<5o zvTRlcs6)YMed!W#;BSDN74U&WikdT?x4JZ#Zv1EHFI~pJ*Hz>g`+^n6gMsCwl=Vg6 zLbTn>YhTviS=7b0U1{=K6W~(lWPvY#GTrY2*NDl!Tz%!DtAYHj*?&R;9a_EE@BIO8 z{cZoizgGU!Dd1v{?An9A?|@~jKv1Bof)KE*W>{GF0a(#T+WlMxZIi*L|CWLpjlkq9 zqao`HTqj)(EU3X_?$)5@d=OAI>stq)BNUgs2l8L#LdHy)%7NN|eH2(EW2fu?-vj_u CmJyu* literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-21-Poster Gallery Event.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-21-Poster Gallery Event.jpg new file mode 100644 index 0000000000000000000000000000000000000000..4e9eb75831538d4dd6030c74fffa2f0840b7f5b4 GIT binary patch literal 103761 zcmbTd1z1$w_clBTQW7Eq(x7xHU4n>6cT0B+2uPPODuOa}hteHGcS?76cQ?{8ya#-q z-~0Rjuj{+6@B0S#IA_<{d!N1Uwbs4Xxt+LO06mr#ml6jdA%Q?hz#rYBurM@2KgAgwTw+n&CD$zmJW_i&MvNQ?%v;ge82zjgN21hL`Fr&#Kxthrln_OW@YCT zmz0*3S5#J2H#N7kwzYS3b`1@WjE;>@OinE>Ew8Mut#52@9UlETJ~=%*zqq`c7ZM2h z@3esLzZ3gk^TGq>^#BD083pZbUPupIfP{>P^6(iu>Jt$qG($W5=Nwi|7&Ldnb`lG*Axf~83}lJ$ao-O&?#J|B~Cur zzWnUWOE&LI(}m3>O6af89^Ptehf!Z_CZcXZkT&Q@2lP~H$%_`+(QSPT((}6oCD%Dr z&p$<`oF&*$X?m+9H?IQApi>H$N!~6`U>GbYIkg&CaiY-nzQEIZtgo%}ON`BU^DFMQ z9;LxfhJBt=ZGF-avxdy!y8%AfJ`0+ZM#1vC1DTy~b_%-`t)Vh;Ysb5v1g+KzL(#9c zM?9^P#c{qw9Jmp*^6}Mh=5s1fc!DET!K~m2mB&?SGYo0ZKR^9=lKSx^!(dSC>+%_{ z7B8-rDfs;h;lH^`|H(8=OKG&1YaR=19!pK}f7l7s1O5fUSf0|%2wBVsSqd-|=1Wz& z57v2F3P=OFbM>XR`dY!@Zz_d4(e9Ub`D8$ME7Rj+f!o`P!0lG%j4lV)UcyDi2p@%A z_-$V*U-)k;Dl3^hWpVT0mWZh@dkB87)05bvpg$CT_m!NM8i~e9|3_u~IJFnJ&K*<$ zu36l&Z}x|RxrOU4)!Z3YMS8j^ETJC9seKCjUXfa1L8@duW$O|^mGHjctNC>Gv{te# z9U!ayuGMSdeJ|e}BN)HpMXYTKPwo`TbqoMASXfneUx{}up-jei(4joa|;?6Jk7f6r~iIL16QiT^EcR(mVYa5 zu}eJ1akjY0p#%mKT-JD3f6STjzWK|)mxHI3!a(!hR-)9&q;<3CsfD{6TU_Z-0{xgN zDcofy{>_BE9!I+yt7c_dD@8BbEAD@;CH@v$y@(1n^Nx>$T6y1%GVPx(j9@}>IL?rp zV%)nil#ASt!4&MlMYc11WpZ~H$`AJq@_LjcN`QF@|E)84+Hp6q?v(pZ6aUsp-3@{E zD-4Y3-nSnC+W#2_7Q1*J;qK*IoXv1*!9SUMIt{(p1Xg53P$d;2L#j9D_EvW1X1Lt$ z7&Vu7pRD~Ch+}tyx$Bcgu0-y({DGwGW5gbtP>!40&TAK|gly~OeyrIUmd9HL9qk}4 zj+a~N<%oIU3=0V#4(oIEztWx)?@Bui*GpWa1k+wi8QU+&qwP&3)(X8Y^Xj9y5_cD# zhTnqLGZ6w7SJn;J$G4z%Ipd+hoq?NFZp5_h-;QMN2YCw$it9*E-x+Wy(cUe;1^pSm z1(9=kmj9iO#r+ehmcg6_iHxx zpD|HdSKNXU63;jDekm*64Z;6B@)ktRT5;n}kMKR|+cv+U!95jouIKS&zZSK%{S=UI zJsM>yRKbgObYm4Uemq?o)4OWpT~a_zfr#8tj8hCe=F1g7a%R-BI}N^ER~5OKDwu+H zx8FGAkl9*pisKq9Kk|bYP`R2gha?3Yht)F_=Gn|KSI zAZQGf?u@V0$YmLPlTwLOm;z$}0SW2V5T1Nm66&C|7%lrlu+p2mgdz5V@xJt~#$9@> zBnAfg#<^w=8GR-N3fW~Jz!;K5q+nVMNTsEvF-Z?#(pm=Kh*HG|%F2UpzO~XZLM)K# z7#Z8b!JwG5R)GMOM_54SdD`*aq;5{r2v7;tFd!H5M^wIPv}ax*+_?Y+37W>ZiQ@}GTY%HH3YuO z&Y8g`VHaZsft81piIjllMh3^Er+mf-0}Fh=U?4lnC(sqMvRhEWI{<~|ipp5uZZ%yC zJ~>de5nq+F{v_FdO;bL9vej@&I?i7}d0Z$wNiZ)$FVA?&w^g1=7wD?ju3i>vq*_uv z{io;({yy5>v^OAdy~VlkLdQPyFmp0EY8z2Cn6b^v^TOD8UHc3sFR`ee%Fy7LU`o4> z8#(#FV0jfZo(0jPE;Q4|putlhLdN~g{aeA*G^_7ndn0);)xM)WCv>RN zTwKYsuX4nnyF|=1@$q|SXW;`9!$X2->vaUO%Fg;kzUvwHA2DOfCz{L8OoV+DAn>W# z@&Rdz>6#(OwX#CS_Nq)VDk_-HNW}g6k%%)gj(k&m0PN_NyPX1TtosYl-4_2Rp|jub z?8>yi`<)RS@tYw9hC;`%DI@g;Lne-0>dhl57$f9!1n?|;2W5a>fPD;Xzo-C3WC*)_ zD!7Fc9CP0oCaY2;B@2{3C^EEyc<&!N@DsSleO05ojW6xQAwI!JZ2X&|^fyd{%=-*%dJAfb z3*B?_`#jN58*@tGeKCYz7(dU^`iCF!$6Z|*R-i9`aJ<>@;V7U6r{y$Xsyjhld{CO< z&pf`RRL*t&@CYOGk+)X(r2Q4ffCgwC~b!wzq z7-h?}+^tM}$!)I{^wzjiDQ;EJI9n$3Jd9VViei={^DyraXHa(Rn@YIsiPhu@1J0VN zYscbLCg$fshjA*9@K1TjZ1jmsgQ&UJn!@}Im{^JtB>_=0*S zFyW~c4YB1xBCFA8cv-Ks`8U$2X1oOE@|s%D*PXPgI12RK>v129cX>90zF}#276d^2 zj?>=J?b%FEP%CLM zL?+`yFouzlMQK!GP*?FSt+{fm8GMpfLbP=>MiK51#>N92q?&0a6JZ zKrt{nBQ`4-GzWoLv;xN;4>-(XU=^f$iGv}Ma(}RB?@n~eXHq~gNy&;r7~9k77?3Jc z;-vubawp}KWbS^&07572fe}oL@$(dL^S8=I(O^J8!gTH+S`4^HOj@Q<*yksAViGv> zPPFWIzzqmpb}4zF^ZTRCb-)>yiQRV+cNh^9vGW5b&(jjpEdA-5UKZA`IiH@=0T%H6DrsDSX@4btSZ zg0};#hl?G{PR){4wr!O^Tdaavc>}LL>uVkRTU_U=24Ceol#NYOTKR(vvE#z2(rxj? zcBa5Md>JAh?|2c+Cr-aN&b##fb#MNRc=Jlcp30^slZp6TEKc2K<3|l*Li<@*m z#lK3e{Gl-n8_m_(|7OFx*^43AHs4pxB%_M0IRR_~#b=CmVqh#;<)r|nH)IkmIskae zSZ$uF-U-vHSr3G81Cnlgz>CtA@uZ6SiW#};;x2*USYKi3HK%x-Ew`zqnY<15CeMY} zk<)LT18nR^eO7kNv~XwjpyTXnr8n~X3b6rC;^POaR5VyaqnLZUBs%-VOkU30FsIY( z8RirX>+ytU^GP|Ff84^bJ0lv8y_Wr^eeMfZ$0D44V>l0YcK^;tgh0|!Xn99>r9d5K ztzKo-G%&E4{Hk9iSPHX{N%D0{1=lSIum6LZS@u6f zJdWhEoFdTTjA!ULp6-??8~ZSr!d_(Kx~)t9LlV&!Okei#@@3Gleoe*jKI@b^+L^<# zaG;A;V4g-4TXyw(ode`!%Plo9oH}sWu2x=N1o7khwn7Ev#<1N4R4Y-jD4~rTWAHV|JXS(-Laa|pRD6~%!PwFF{M*(%<%X9 z?^nJ_75!7=HJ%qGKKO=_}oGgs};3rs$a)3YVJ9q@~9G55oLKWa!=Gi_;c!c)TQSnbK5iGf{`P(!8UTc3eJ0JN~9TO zL1h30`kj*d-$_IqKoxwz49~^}@xYY-Rf8q(j{1y}d|E3AtOV?(heRnbgJjp6}q;BiDurV*ef0M_aAa4?n)!&)@BGUHjP@)DpSumd_s%)L@8doTR& z^)M@Z{W^;Y3s#eL?eKD!OQtn@jt7Vgw z6K(Y*eokv4(S2Q?FyjVG&g@c#6=y@DtK&h%Q0<;3;rn@uWr7Vos^0@Po#c;~YCHOt zo@D(hUI`OWA1wOS=}DQL%lzpN~^?c4AHl&Tl|&Z zr6}a%LwDs0g0*_e1MXvu!TJ7|VXCY4`~^~aXo6$*k&`*<5@^dGgR|bnw)k-1!W|@c z2K{)hH=B`WG?HEYn^WH_aVeB9vHSFk>N&U=x}^7Y3JCDH5oHo%w8Sy?)^2mXP@!k~ zeat<12Oc*KY17QNp!^?^;WLN5c1&3xD)_c0dM!3&i?s27&)8e0zY=fy=D8g|?~^YW zuiB#Uq<~UuH@Ei~W3X&nk&H5~j_DU=HB;^!paAAj0V*fxE-fOSLpyq{0Lm--j`A|x z8t0s{Owo4J;sx1b+g!h+JXD@aymXcqUa-~KYq$kHBlEyKW?*gt)ckD-KgsDOyguEW zODn(w6D!t;i2fkwiK|EQ#x9-H1ko zuh*JeWKB14#>Zdq>>xmVTn7e;w!jqk+#cg57G6@z%=a?-yfAKx9onsL7qWFmc&a4E zly|!RD%I}Ba%)n{k}!;%Eh1$3lQQLs6`>Nsn{rusdPP&mAVhplVp$jGky9$0iieP& zHETwB&Cyz5=qQdTmfEXNq^-icV`I_TwHww2-{bK_eG%;E!_S9|DxsyDfBcl>wRoKF_qC&>oXc^i;QTeH`T@=@cvclk~-cE z0WCR-Xc*aK3rSKyd?Ya-$=;@B{>3WS4yOWy;J z@160%fW*i+k=Ba*1kh(x?im0;-?_6>fc7AsFhCbF0*U}Ix-;0Hs5(&F?~aV2mU4PQX9+NH*04|GandcV*H|8(OxfvN&!PH6_< z8G(N3HJ^2DmOHY;>LGK-DZQvEvOj%C=!H7GtB7F|rvG6RuS1yY z(yWGMBX#IZLfPBpb`{E#kI!#GwU%znGx?h2lLkm@v9*|gVn0DV>E6~_AvSXCUHw%O zg9Y@@F^5h=DND?={ZMh_ncx${1!&>+hV~v$IQ5KdPKc{x`$v{iOlwtYQ;SZ$EIl(r zMsT*)8wsCceu~v&lYi4ErmldTCsSrCk9Nh~Ee6-0GVnA10WIZVR-SItsBtc7XSlJ9 ze0PQ;F|b{#k25uCsVGw|MWYfTwN&GEvg%F=qbjNUW23Btzx_LZZoG9$Xz+KH>g(WE zTX`usj(ce2DX-59rqZX*XUOadqyW$Kdo;vK^qHMFqZm=Tou3#h9pjh2^Xk(kR`^*1QXDtz1HJl`$%lS+d5Nvs z<_F0tkxJLi#UEqDAyV&3^*;;!`8DiJwoY_E>lS39I((iXhl9;CZ(=XTQEu#Y-Y$AR z>||m3b@@Y4@ng`1FQ<-e3n8`45E0p99TVrPGJ^TNJ^z#jH#3(mN8!iDJ3}`?JGY>2 z-Ua;k)eYZmc1Pf=Sy7bxUqUa&W8b^N)+T(S&$5K%(XXu<4sJogX<~S8-FCR5DDxf% z+~3%7&9esuIow;;$`mj=$m-)?;OevHMpx>-ZL@u`sKtX;S>P&FD)jMTMTn+uczp0& z@ks-&G*yJ2HcMgD#e1wn_06Wt#^gl7#k3Q_`6iY#f!w?vO?5g0q4NSr7~+*RM-!>P zNdjR2Dq7a1MwG>r@_Sd`9qlN#3pDSE@zkwm>$7r1~@@S3_ zaj_aQwj4_HsHveoHg43a5I}R;QFM4xmwPF-a2x zQ2Wg&=8m_dN&}2aOj#uv05k!}{wGnT{KLhvONt5H!TpY+7#rzb^}B+!RzSP_H3Z-> z=#JjJR|y#Z;Y+s|9#8$HKU)FA6YI|YR6!~Q7^i<=9w3flGL_!=D#1Me+OIHa16tZU zzFd!e8y{C zy<0Co81(c|XghjJUy?d=a$2+Gk>a_(CphRwh`305!AZWp;-PP*2YdJ~8hvo6Rry#g zr!m19W2a7tfEw4}2mCZluQF*)kHWR!cOT9EsMlUZ%*Y%i|MBKC(BOc1wIz1dPM)}fo@;p8yzeBxg_Vo^fFy=t% z5KFvZ`xhBXEm+xy?|X#tUzg|-t%xoRM9x(M4Xv0A7evohkB>Lyo2o*EzdK%tw3y;v z(G}a&KDDKj{ZJ)5N4cm=j!3r66ff2xL~jRuiGyJq%T?ck)LaNXAAv00l%z`he>RRQ znRn0Lf#H=l`_B+goAb*c3ZLvr7(=j`1v>d^%6FvJ<{q~4GT zx8U7^2)HwEJpTG8nh|t*69hA^CCa7c`sp6>5VVW%in6-*rH&A4rq`Y=`qU1WxgiO(~vWh*B%*d{k#>}LSI zH5ixAyIpPNsY{1&Z%_;2D9Mi2(zXfhakk)-Xb1`qdY1GuOu15LJl?n=X0Gs%wl#k= z@>@|*%aJh2o}9oxG4UU$4{+tFB13=(z;NWHbV$@VC}N{no_|gP%m)xW|6R zMj9d6%UGq! zvl~&8o9U?%hfNwgJsam&8H<{LChK@goWOZUt@9WBujHrg486I`v6#dM8w;c4UMrMQ z8-z5td8L$|Mh%JUM32$gg!PkMg~BSL+z8y;)n&5PvER}<_l?Gj6U(U%tETnjvAr{= zIw8;+fM#UgIOag4ccdyOpJohdnT1&qN3C-HK{|`xV)FOp>>;q9lfm*hd^;m!{pslK zP5`dubXkMU_E>pkMeyzuM%8%N_7T;3i;LuZO*hV-hX`th!|N_L0u#Nap4#2lCqvt> zpD2Yo*I;Zi9qPol``WO+uikn@BRU9ml&t2u0CltW*Mu~J?cB>5}#Avv~FoC_>~*9b$rjjwUXDm>l`AbA{B zR&GHdNtZB(Z)SO|#FcfpI;A7Ym7pgC^|d5*dSd2TGn!jASu1VY5%jR_QeiNBIFIGy zWTUun$M{1X&Y+@Kj7#`~@jkSm6|<-+`A|Ta-c0+~0U!b(;a~Lq*VpwQe-}Uzfrc2_ z6;eJE11>N0j61mg7pg(6v;Zmt489RS)&q`!ySmDFGJsw!BmN)PnHI=~0l@ov{}tVx zzAgfjy=VL5rQTq3OYf_em10NDb&oJ^o(ScfGLu-Lzi^`URYk#1uwojD(5SYSW)D9L z*etV}m0`C0-nALQr{7vL>aT#^>+rqSNa9T55rVGT$^)Hs1VyA(hUHK@V^dejQ>xW) zryy^#B-@$_c3N#(m-9J`L#SHqH)vD!;mVnyR@2*CP`?x^+QO^R)efb~uO3Tffz`y) zH}d8!lt+&#_h4P*%nJIziU(h#sGA=Qi*6Y80|FKIB!jZj(@x=M?DV#rZKYQv3tm5K z>9A>9b5pKZ=SU$3(rDmF6udORH1JN~{0o z?PzS2hsyZgh~wcIjoP!rXn67mqq>1cTgyVCL%^%zD*PKUSmifX5p14G_-AzUQBX zm(oiCwZ!C(H7qO}8r*sX9B49c?fJ)O*Q({;5%J@IgWtVzy!~1h%5==|2@i1Teo{W+ z;<~YPE~}I&eJVvnq`n2ohNN1i-9#Bf_yybP6`D$RzF!KRU3p)Q*{{4>F3LTHR9K^* z2GV-U$r;PmmWO`|VGA%oBlA8V=UnRL@Dy!2oi_)SI+Tm-|I$!-QRHdNhu2+)P8=i> z3FkKqZi(Z2|Kt@!39)a+{iK04v&gQ3d%5b5=dAy}x!2MR zxW0HUalTDrUB_qNhdQ$2Jx?2A*)fm)yGOf<#kjVNs4pg&=Mc8ANzyXJ@EFcqqrJ=B zHKWCbd|uSPt~RXfDEO`+(6HN@xY%Ws-1akSdyP>##uP~Cay(;uowqYs`VSPPM6csd zCAx9wQ{;LF1i`j@%#eA_QRckuYrY8k%xzuAL?#JtHRnEOL&W!|%Nne-QKXtv^d<%x zD3ovoXv$3M2l#`F1EiMFBW(3=MqSHW!f@N3v;a39&FoA|KX*9xwJ|NmpHw1C7&)0> zw}Y|3%)3knq_gjfs-=Th&9@+(Ku-B6N!!p=qfDq}^dp`n<@ljBX6d{aX1F%o| zMpdEXS%E*`rlSUxu_+@9n&MuXiiY3ElB$Q)MTJ`7Jr3`(CW$Y;FN*$@9(qV_iys0g znJvUTthGZdodeCh$J}G>gu(DX#%$$PHiaGTH_Ypk7*apc(Cpy4uUdGW<&!5miFOpI zH>`XKere7MD8+6G^n0_?zOs47)+3JstF!w4Q>1dJjDdG?kpo0-m4Da$GrA6QL%XNf zN74Ji$#5aPVY#uTuhtwN!`i8wD;&IoEjI%t`#t;IRDKCz=SG;J z5(9xBs>0oqr$ZiVrczEvokif6lsWyuXLXRHugA@%sXSA-GW=#(&(mY+>5w2Vc3Fwh znobzBnmH~l!pU5WkKZEb7G&=`h2F8lq;=qr?R<)@%-`DLa~802QsC(0-xTB?g3YAn zwkJ?p#hTr-eG57QG_%jJZhCL6xLeTmH|WCi;{I9g>_kx#{geKajKu1BX~)Fv#BJG- zC3;{Thm0j6t;P#n+_&CeSbPNK2|uHd-Qg$SI_dd5BdOrbjrObRx8T@QZeuKMF?WeC zvx(xK?R=jw|9lV8NOgVeox*<-GAdywW1R69Z^xvzwZ;;+o^EOpM`81MPFcS?xBUEP zA9nPkR@g>CL7=w!Ar=xjw!22_H9&j{e?W7y>OCINnqoO=l@Sw*r~htZQHCBu89DKQ zd2&9ffMj48K^K*Zh{HTJ2QT^1*;)V-#HHLPX;5`EZTr=rqD7P+;ULrkIJfWUEruis z;iY(WQX%`fsijHNz+`poS@|`}p+YBex-P_qzsIU}5N?C{N>Gh~l!=2sj7w!$Jq?+fLA7IQ zLsAQmQ>_{};Sg(KHD?zrri_T2l(w_kT9aoFQ+H7@G*<+Cy0Q;%6RsMklwKBufc+P00ynzUS6!(o%DCy$72a;PwxxV1O8z7oZP`jw zG}|6Pw7R&RA|=QRFEnv5c82=&A#wki2uU@SLUAF0n=ZZT{=i%j=Uw7-y$$0wB@dcZ z4yX{yIct7OxRo{y??Ew*7q|L0TKml0K1IfnXPE3jfFx1${5yA%Dr^&o;bN@P+ND0E z@Xr&^&DOq25FIiRW_4s5Ol%8sQ~bs4&DZ|`GBjCx&?O>lArd+dcM&)fNEb=77@X1z zF7AtdEW_p)LWUr)(yvGmlB^BoVBC;g7#mpimm=yrk}IxtiRSs(%!1DmBH#>n8eu6T z<-Kwgf2xcWhS#QG&x`}0H04m5KSI#_GI(2!fj%2AT#IDvQdEL;W?DX)ZHLOu)9p0R zpPTtp&TsFA5X-SpuJR*xzpg}vxGLkERyT$q#H?)W?|4)q)vLh<<+^bjU$W&JT106c zdYwU9z*n4CV>KHMRxVLZi!Zmk^qu)ZdxN967$cxxsS3@p+81M z$jXtpi*hw}H1#D*es}*#klfZ>NaTDbNw6lRnef%+on#O^q~)cptIU5>kA z(KJ)?&HRGN!b2$y1|L4Od6;F%3XcM|rtTKTmK{G=cf~3Gv}A^4Uyq@VqitJK1aeNW zZc|=+w$7B!L)>}}RHX|7hpyqhGVG#olm+jyAW=%=sUUOmrp_I6ZJlo0TuI{(_V&1< z>egnZ4dE{ymbHBZ;&Kjhtt_eHxT9FZW2N}X#msY%vZX=Da1t3+l6Hr&XGA_A{@-LH zUsJ!|NJFccq>ls=V(zrlyW|@Ja{`097$ws({!$Bf+G$kSUrym)cf&vbque3!eZF?#kV(V5q9$OV3ckIu7S{Iyd)d;O~!xO5mqsX z%}(~Tb*#*`e^)TC92K>oh!N8<}(LsPi%DJ;X6FVR>;m>P$MUUwv@fj zD0uSlgF!R+JB8&D2@|P{-u%} zGY}is2@jMC3eA3QI@M=RW*(8X=V;|GLyRN1Vvp|HCeBP}hzKM;g_AO0<;w8Oa)kPl zQAoJgYf3DdIM5$HXXX3Uzl)c};%gHfyaeAC67NIMYiFV0gS+J@p-s7sDMjJCgbwLF zZ`N0~gfWHE3wo36_4)*2F<3%oOWIb6D3ZqN>I;YTX?R20U&s^68b%dO@ycZ4q|3YG zDdSftz7?`eCwQ*d2F`H?L#vk!45A8$(*)ZRG65Db?E`BoE@G;)TOz9Ud!Vc|1e$L zMKu!lxtDKsAN2^8rUo)jE!!=K zBm@%L?R+%dQ^H2IKt@0*f9x>5grj{;t8M+(q9EO;|0f3hitVQIlo-U5=St%!gC3Pp zJNWlVi1r(kptlXRZm`1Fk;)9C+{6UppiP*%g5fNjcqP+h=@qecq#*g< zSjaB-9PoE>z?>P`?^s)v$M+6;AZi4N?Fzvy$kYz0fDRlf$-`sFi7m* zNWyz;{TE;VMs$i1rBaOi6M#sZfppIl_zo{!E&Y@a{+@P9GNYa94^gbQ!Y1_@QpB)nwNd`n5(yl5U7yF_vLL8^Ny}=$)1c;twQt6}zb> zCWHb(u2m%!R@oXHoVGS-WU#^SZ*dY?UMQdu_d3txUX#W3{F$G9VLjx$d^NeNmsdVz z7FF0p6~s;*Ghn@Cxh&NYtEx0`QM{~h#kMvpfmTrFl0Z4|ji^?Icz}6vcGPL}? z>dHz~XT}b?c|Z!jE5~_g{7)hervi>)YjXkX8aH|Ixyl0BQ5)-%U|c=OX?SkW)klY; z!(V^2NNoL%SxF_ziX+t_DRKp?`Ql!s@=`Q>?HUe!jz2m_t-~SoP+5%(natlBj$OI^ z@+Yx5uF)cOfqJ9Z=^Y2CHLGFr0iyyVM z-jO)d_;w}9sa6U}wdP-haYm))5)Wky_URQ9J4vEjH8iaV018&nQ z9bxZ+r;3M!1&)b!Kae+jT?ja5>?VNy`?Zh0o$`niAYR5%0Q`+BycTOq5yqT9EVY@V zDS0L@7zx1g7jA>|ALh!GKJHwUHI$;h#bbKuaAw7-isBmfF8SY>zAHGT=VjB6mP=HM zbT4i}nv|63YDJVAPv=E$tcbm?JgpSWiFp(R!_);{%|0OQ^ee6pOk22+cgx$pdFcB+ zt1-K&`;kv0c|SKre(Otc?&FXAXat#txj7bR1v!KXQ>>|w?LJGA2xI5H?}Wd0qB4W@ z&RJauizC=}3dX9g0$*71m{yjBz71HnrHuDzf@tBC%RpRGh@tD@2~eg~k43AicOh~% zbUy-YltPtuHWe0~xz@9B{d5jSqI#U(LNSZTN52NgJ%?f*1Ma4Wn<}Nrc#Dx91qxkN zx(bca&H!n((I@>Hf9Y5|NRva%oc)z6 zq<-@~LGlHEmn*BO!8LWSQVA#f#chJ;m$)qxe0(55WHaKyRTlK6-}6LJL+`ne{jp<9 zud2ZaZJFmlSC?n5Op6`T66u2bMX%;FFn(mO+t6UbH`CIrY%GOmqiYkja?(9lDNc$-j_o)B17?lmF^bz{y@VYkJs^`P+ zs()TmkXAM?#1|J}F|y_%hCiB>#OT;8WQtQNNj(g(W^q^vrwK@NZYR(hMG*+kGs)76 z;pWjdft(4cp;9aD%BLxuSRJPq&>VmL8!UEDKK$$1dq4t&Q{2Uh0TH5i6vX3uszRFa zKJc{iF2GDn`re`UzlF!WfG`3i3_9@g#GOkpBP`}lzy9YYf1}cX_aXqz8_>bCRw9i} zv9HzWn5lJtO>_V9P6|&V2zJxpUDZb7 z4K6F3RKw#uaiDMwPWng}u9fj;_~a$+g%$L~d|5PO21d<2q3v&+)t+T;rtWX01K!?W zZiq6bsg17J=C#oqm52xjwpH=@@1OK-lPtQ8$x}NTyQAred!?KXF{RQ25v_zqM%`5< z-tYMbmeB}R8m^AM+%RS;4H;g=eW1DpRlKmg$*T$7S)3S8T;j>P1#R_UoSR>c3)$uA z>FG$ZGwF#&F?>%P%Y!zFU_x`siWSb0gvTOWgHFY#ae7Yi0FH4y+rc(ln(+h7?AMU!q_Q@F? zKM7LisIVz3%~&(Pko9w4W^?&C*AMwqjGCCA#c{_Ht(V-oKh*iY`i4eKjasI2Xq)|W z#d*VYoBD>0(onUGP2p?V48>c<{&iikYc2AXX%bD=Og^fL*}Omx84pU-?f0>CnnoAW z9%dOT@`PjtBYUoOiGF)Mo$`5!b_yv^q`yrhO{F#;#M?8__=`9H3ec3m=#^(fKG}ga^uGKt>NltzU_39q5 zuADg&j%z$|3ooVhr#NxfRZys`adjeP50QM*+L)i!J9o-&N?@haTuHfc=*iS#%!3cD zWmaDn^*wyNId1VgAkXIGsJsX1Pf|wpHM+Mw>v z4a&8J^1&`g-(^zCQKG0}Gkh9?j#Dc_YUkS#Cr@B5#sP7~Tz!7n!!?^{kQcZN^RDVT zr^DLwM(-yvsQ0l{lwx1Thr;4-awV;zKhAO4?<7#U^R0Md6UI#|)>2$ja4z-Z%(v$2 z&BS!K9EA~!Tg4K33w8%<(gIQxl5wA~KaXQF=O4%g@KjeuVT`FWsAdVF!6mXj3587vMuoX8IO+v3 zt$*Ip>er4Ues1Cv9MbfQk*}tdUscYcUu~4DyGn~c^+zX>ytD(XJ%S)qWZ_GHnWd!$ z=hPO%Ku5=KXRgESX`POzN@eODw;*jwEM|);nuIkDCEKD{hTfqYrk$r1gqPXxLUzoM zB95@%hf8Vu!%sEvdV?4Gh`b0MX;xdJsvEE^uL@bL1w{IE(`bD??ep*zS19QUkS^qZ z>y>!p?#E&-llqaT8~=dfwHODsuv1fyzMpP1MLc7igABy06DtoyH8M56%5*k!0%(fr zsZ;TC@`~p9jH9a*A%E%h`|y^(VT1qDPxr#l8oVV9#2_jU%Dz>)dllkN^2uA=5mEm; z*!sUBWRN5+k_H|qNyRbVd0Nr`k(U2R&VYe?aodNnyZv5kCd$Q%%*?)xr}Mq<60|8? z)%VCM9Bf^QA3|Q3q^`{qYZXyl?U5~UUHti@aJ++RF7qUH1~}DZnuRBSn?&?M&hAyx z_5KaJNNGd03TykGBD7n^d-z}}m-c4md-w2%)UtrHwzXV#SFh7k1&%!_N<=ludQG(U zPY8J_b6H!YZL`daWCjFL{VMxDH|fL_sibk)m&z&sss@2g)>Gz2q`rmF-0HT5UR#Z~M!2PUfk64X?Jn@VxtRo#vw4TQi7~xJO*Kr*sqN zA6>C?wp_(OPkBRIaUWS8Ia%Nz(qE2rE^Plf+c(xfQvqC&x-#2wmO#kbZ((^$v_tZ* z$A3sSN-vCxcV zqt_2#z0)wJK5!1JGjh2Fx#i5fCK?7WvVzb$*W5oB9 z{OdbtTd*Uy!24rs0o)6h;MXuFCtsQ+24(i0+HuO8NGp=Heb?fB#lm9D_l1hApc6D5c9v%Q5ljAnEE*{mIgl{>dG& zW+b=J0bTW($v#=j>Rc3(=ZvvgwnGfGpJb1~aY~zl_|*)Lb4noUoVTF7GCzR3)ZgiU zJe3&v2=|mC`bt8tl(H7K9IQ(B8Nc-sC*CxbySJ_~NO8f`3V4muzaCPi`E*sO2;*e` zeRT$$r@;7>IW(Y)>KzzBq)JkUGmyiF>xDV&58k3o89iV6Hm+WbD$yKREJ7PA&wGa$8ty z!DV7s?g2dFJ9{^c#Go&vX;S7>I6&V9Ve^03Jq%~K-s({&l^3rX@5_cr4jU~Qd$BaU`a<0x5A+1^W zCX=A7C$Xs3H{Rm6s<3DtVpUI5HB_q&rKBY^Qq6887~V((KYHl9CRP~$oNoR~;rNe~ zkDpt7GR?|b@q4AgkpMyQ)?qE}J25A4*d#g;#@*Wd7QVwNe1_P(wt6juYDEd=9yATZ zFm4jCXKuE3(tX=$N7ineyrPlziDSg3py0*Bmi^}ekJDCg%E|=b^(vSVOJx#naO9ih z4(gz#Y{}Pv#B;7FIB^L&Xtu+0XxwNHJs){#ojx!5y8q+Y5#AXs@V4(9Cxkf_4Ln`K zF&X=EB%3N|sVsRk9WlVB!w~m;f|x8gIp_Z{^_4+w{9U+EpcIM}cS3O~PH+hBQZzUe z4-|KYwzw6i6n6{mS}5-BQWB)NyY%M&&Ye5=+s-5(cJ}Q4jy!U1eN}flrxN~HP>cXhulNN>bWHQGC4BJ(3>1X* z%C+F1P*_=OXzqJ~`I}AT)9GdDc3cuPtdnX^euBjq#9_}Y5R#MlB6c5}!vFPBU{>SJ^K0q)e5g79SIxI$*hAm#d39+Xzmsz^XE4+90J41Su6T^(*j3RK`-;#x5bw?o59F4qslC2XyVjux7ag zITRhUccbH%y7s2D>>soNfiLd@1TYJjV)>YtSUOR2)a-ldGOSED6)%}56!#Gg4qvDM z%laIpv(IF1(O8tWx?LI!6b={E&zj)jkISzB^^L#($Q7lDdGf?InBaNBninY|hrZz? zd91GeojCzqb|17#_kd}1L!KM2*?L#>IZA~;G;aMD@w0fNl2eq{s~!ib z$;if*dy{C6&e`~OTjoE+V7aX0j{`lThpH5%-rhOg{99q1S}^{!d8*?Q57qo8Ih=VE zh|@4Y4cP{nzBC%Dqwgc&Dbi;{#-?vE$S5Jlio*}H8dB}d(z*NmeWah0oL?AsOZ+In z4&#^vC#-^s!+3W9h&qk~#Yh617S`yWA3RYM{lw1YaXS`&<7 z4sn%K76Ky!I0{OjeM5Mocp(NK5Ck%X*>N#J!m)F8>5CaNmf{ST;tXm03}-(~mtq15 zQWHk6=BCEP6+)kHhm8sc_S2OemsS|9t9!{5XI#7{h`UIuuloYyUpXHkQ@k!;Am5Aj zkP;tYN(R7JL!b6brql%Ht~1!;rjPuMIO$SfP8fm#*}RwdY&AsIHWYBY9Q}`{tttx1-mF{V6Vexz$ z#@7i_E{{XwMzwWyFXJ8LpI@9ZT6#pIc7e`;^Y^+l6O|5G z5J=_72pBRRPprEh|73!DWf*kJY8%Ek9RFyJh-ZiK*BoQzGl+n%_mATE?Ta%~S8S{O zf@iVUA{o=u*KbQxW${xrsHK`|p?U|YY4vFB4La%D7#6Uw-D(f#24iU6*XC#T?u}w2 z&IkKvdv^$bvsPMdr>;nyMId>p$^PPZ<9cPsExj|Be$6uCVP^=n$T`|^+JM*j2)*w5 zZMd{idC%LN^|k6JEQe~fyO1sW_HO)jw)eZY8Of_Z-6E6L(N}0{I zi)7G*%+DOP$*&||dd7X3dP@tHN2_^w!kL9gCk@~hEr|qu825{X{=zBMj>bP;eX}B^ zy3Av&P+=e&KPM(s;#R-&fDmS4D*Ycg@9g-}nDDR0J;P#d5nL0}y-kI@L_M^b67sRFGMsXiI! z$teiO(8u+JB020P0ml{LooE;uu6ps3I+l0wd4o@ywO6aM;np7a1;AeAb7@tL$98?a zXP0L-arG$GV2fJ78Dq((i=nGFw&9(-oYGoqqj#l;<61c|-8xCksrY<^PUCNlEAKLk zP&Ibiz{*i>lK(G{d+%~}SggrMARD873 zc;u~EUyR?=JdZs4Nt#&iFxxQhhmJp-Wqi2H)NrPVhKM+TMl<^Jx%SR^)mI!|e`wCh zr3rC88#8KVPT*Cx-S!vPg1IT6!i`TLovnfbHrsbUPJ=vwekvhFJt%|QDihf=+c3G> z;n1;SA^;y{QbofZTqj=&>K7ZtgX{@zBYesZb;w?r50GoAD(R@)Onei?FE7p9i-(!S z6mRplj$odP0KDsE`Jt)7%^j^5pzMfSgr1&elh8lAOod(|THcT?Y5FadO=zpfB;+aE2iA!L)xUxwRyK7J4s!4R`DA(4A7Q>haYKO zzS*h-dwc8cP2tjv%%3oZ6a^X^`m0;t@c&?HJdf zhk3q?#<4P5S*!OST=l$e3ReMqei?_K$6nu!nM-EB=BFuPa`g!+~QuL`rtpbrzI zXqfCBTlJn!l}31v>Q@_ zB5Gn4A@~~#;(p7M(-j@9puctTH>%!Et(7K5Mp*(X4n;u=3L@IWLVO7>b}UUE9OZ`i zMzt2p*#)UHVn2{_23mB^o)TGnsu>_6ne$n~h{2&;XWYuJ7O*fJDQ64l`djCbLXfXDnIKBSUjwux)Mc0Y>fvJ*=BPm=~$bhAKM7rorPsD=mO zx94nh8cuVFt8lC=ZkD7F4@-wDR`^jWE-mhV1W2fE!zro3##qy3-@-*Rip!rl3{;n@ zd7Y0UI@oFx7|`VqofIG4sxmVNijuNPN{1bLTR)O8@9D*7xWVa6YY_6-rMJ@b+D@FS zeW_EsQR#XtdQ^h=u?(xsV*?lA47rwqogCjRw%{x_7sq}oaq}&zu~gl+`(!RSaHdcZ zKU79XV+l)_B>W!|d{_O}K99ES-2g14pPhk6&8ttiaXhZu0o6)H9Y0l5B-QDx9Q+WL z?5HBp*h4fjgs{jn_50aVle^ zrcER{PGVa(^_u`BBkG~#ZoZ=WUHaZd71TUKT*PcHb#cXFbu6{i%;z;`_P4na&Fy+w z25P@z|8{pu6Qy_XUD4g5uln6}X8G2?UWm6fO*$04c(_Eqg_S7`4dzwJReF^1R?c=f z0o%XGByhb5uC)27N8MJA+8%Ihxi#DuWLy*qPP}=t+sKw*&4uT1X%L^_&)Ws4ZcUSY z@nQ+3vBX)o%e*0%U7vkZHAZ0hrS1x7@muDn{q$dn%C|jPk}t(=aEkFmi$uq11M=nv z?9&9m_b--`vBMK~B`Ng6pGR4F@Zrc|ywWB{E7Fchs9ysCi zReRJ_pqyH|AOa}WHn4dLD4AP0&fgc%7;zXiY*q8RKq_@`CxfM{qt;Ho8y=yyy@@vb zOBt_hXI?D3zoH=0uG2SP4Xq~AG@sr(>88!K_sQ*~)mz3ng^`vP33mGXE@UW7oG@lr zt)MNG(31$fUb&5_eJsvP`#r-)_D$9YPYj&RHX6uq057rR% zYHPZU3_q7!3)C4Z9}%XCi?J$Ow-?8+Ytt-VKek@)K_NU1U$I}ZBjF-w`7nawdtEsD z!nUf814Ksa3%I;!bo5p`RhkFB?wmnQ`2C^4x7*^_puAjhzbvVQxo)8b1hek_v8RmI z+u%6jdHuSP^U8Hl@ATS=MHKRTcG*G64$;Wz-CBv?7@jJO4QDvF2j7&KllW zrdh%l)+PRC5>+fP8Y!0uJ(a>0Fq-)vk~%;`D`L$attqR05XkY?T{G#xHX8DY%cu5T z#F%oNyx^&RYPy5Wv8bhXQ{6_7`!4OOQC98+wn1G-7;K;TWbQ*=n29(@I7{atR=>fE zEvsN(8DZVghoqzP9GGZr zaT2*|c9KuovR=qdMLKL!bv$a)iP+Z6=240gfid)R0PTO4VYIj~`iy4IJ2k$Cj|@5` z&+8KnO*)G6P5l~bpPbTR5vR2+U(%urAnph-68U^dC33;5YP(vl6hEn!Tp2zTua^ch z$qvVGU<*(K6=^s{z0G6~89qU%W1uFs`>vz__K3q@rR2IGPMX;|0Ie9YY2;#iE#K z;d5I_b!X(v`=I@KOZyMjO^Nf4Tt~+ERc7aQXHcgO^~J6U_Gr>3W7$@eC1rwgJ6%%J zzU)9EgMFFnyH=)dzF9#t-xnf`@3m~ftOCP=y6~lGmyB3@mRam0Q{f5+50sO*zKmk} zwN|7|-P5p7euwN*qoPW|xYoNh^ic|ymPk7|vm1GeB6*IeTfgJbi-h&P_~F@p+E#B+ z>#p`QAPA|zcp*SbZAUMGf+p?H&2IK6sf14CHF~G^qfX0bG7_=Y(&El$InosM5r!e697j!3sN%Dp_&ND3NnQ_U(K-)JTkj=$FMQow-v3|4(#`JamLnR`( zzKdrIjvgZEN-&{3v){EMuP`x$%FKY-3>i<9+y?n zMCITg!?f+f67dMT?I7GLQmh`KCvJIR9P3ZdUDGJmfAG;I$K3E+vZgA#`S#ulUHR-AJR`gJ0-xD|4BE{0PE|>qKjsC*|Wd-!oX=J@(@GA9Z zwl2_qSWwxE)n|XW$S_NBLU(%jNY(c$@NKh6J)p-$H>p=wd43xtg(fmu|b9T;gBPwzpq@P%~ywS1PQthHPdx z(wxn2r*C@1_<~!C)2zSlG{poG@ht!9d@p&Y*?{;fbv90h%FF#k0GhZdno)@7BWPrw2RdoJEc9NdsM4~@q0zw&~1 zvRa;ajTfe)xL5-+01Y;DBwt@$dxQ4@2t1)mc1uv(e>P~7Eo%SOUdTcwA#R7uRnbvC z2ttPs!q%_efZ)PFy->yw^8#^0z-=tV5hZu96ZmtZXg2=Ss1VFo^} zfp9(=MH}W&${jQo_eG5KA5tS&m4W?D>1t#ti*0_J%B$~JMY&hGIE??0UO>D54Yl!e zf)LWX*`N+PKG-)@FKE+*+uHkT>{vZt?M$h93bkb@eHhZYS-n7o4n4)S1d%AdD?jJ^5WrBePjPT!fW zslcdOD=x$TPr-y_02=Qgwlp_AJPR51!h^2qmX zO(9TgfUR?76XREPP{fsKc=rXfL4f*IRnY^O{p`x3mpBtaVo*9T&8?=tevO$5YS+NI z{t)Duggp06E21=8TQb#Iyf{I^j#t6aOpHN&BEu@?%H@6*uYT(v(%oI!zA^H?(L$&f zDmaYExtjSN`P7)xE}t|xxgmL2D4tN?XtHW$obbbRrQKWPUm`!B?6VhM9MtKGWjzdT zirgU3)!ziT>~PHy`A1dl>g#B|la8P53<*IU!i!cciVQ_pLgjMoohpW=_peO?0SmP- zKnLg)451X3T`$Z>x5evv@vek#*)WYYw8dF50W0nLO&-WbetOm>l!x4GWT!A@bnk2U z4WVCF7!?2XZZ@$n?3(;2Sn9qOT`KwKe)QQcGb!wghMjLV0j!hU&%xE%{Ngudu=lw6 zTa18;=??uZ2hCgUY+H_R-Q39Pq*bhk(hEly*3Ie2Pf9#&sfARh7Wo9Kj#|d7ZbNM2 zwvZL}5fe!kwnig-R%Hssz52Z|UdI96mFgu)m)DK(I!7S<#^q2na?Y1SR#n|@LR*W;;DVdtq z#t6y3+}G!6E%|>)G_Bp6L=^%z3~Mewuu{hk2XjlP-gQPZB%e=IF8_h$H0Ub+<&Y1+ z8W;3W3R+}!O#6^v!kK#1ksVegY2jv_&j zg2xmucpnbR;#Wj4+`d#(rq1k@3carX1+WF3YHQI5KnGovmj@nfDNk0iA?F?zrevy# zuI5G-=hY&}o;TwNxzeLt(tu%F%&(k<7W$bYw;rN_SsNqc1Z{^QIMHzq`s4%tpKYJ{ z9X#(qm=rfHF4T{tx1F&zmVFg#b<|S#do=iSV!h=4rC8ZffxtbVuPQFZ!n;#aryUSy(|MV35 zt$i0=yu2&&(UB6j1BbZ9L2FPryQI~Bi&FkhD`pYxx!xgLe|2ABr4RjCn+M7WWiDzyzr7cI%|u95 zE~qIsuKUQ*aNxNba;NzZ$=be-T8B`(Ex}kKQ)znZJ6`q`8Q!4K&|y@hs92@q?TL0E zkS`FYE2!Ih?=Op=9(w1J-U+3Zu0+YqqloeB{c&G%@7F}z{YX!|VA?Zn?te&x_E!jY6!rAY z+xzT~4>5-q0g0ymQ5MD|ga{J0_EIq~l8s&b{P z7kz+=8rW-cpf$~A7kXCPnBwT+cyd2vq`0%8A$&`m0~@NjcI8pF-m1z`_!S7fm%DmiC8Uz9jU z{W4?-KglueZTGQS6Pgr?-Z}~g{Jlk|;^77t+_^jRe%?GoB%ecYxx=d%H|3OD94ET* zf;+E}mtyW~yJkQzp@*LQR>tWN>^o=Ue~#-g_!+(OJRG3#59#5$0#YDbV8{~NOlGAq zOT}w=M+DYDeiX2>U^x0Y_*M}YHL*$Qo{Mv&hju{^*`sD zwqxS<7QV~g@(XLl{N`) zZELm=4X!36>|j~^YDj&dBLh)VyfbP5dlf1Pf0>u;P(;}EiY+~}3au`a`vK>(hgPym zNd>r(U#?}Jp`SW9R|-C;BaxrOVQ+)B2n43>og}k34Me_S$KD~gP5WF++Fsd|B@G1F z@qgCj#V*L6ClI_QAkQ%Pyt(rM7v@^F%sb=T)1SdO49_UkBld!#5L-HpPG_5&6SsX*BBSnXLDcP-Nx_s^K^%eMyxCO${9QZR^iDSx$w?qRqegk(G z!Ke2wNru7>vqlD|-{18|(k!M<|6Q9pR!~O%ZJtC&Sdv@FgB~5U;RTA7y2!$tQ+7@%}=EdM}{P_}IiRb>I!SvTGkq`sgz zx6rfauIky#h*F0&`28RaS^{p$3VQS)FG9t%C z!!KL2$NcuYPMBnV1PI@8jhD2$C`)ulMD?wjhVcE8PrYGRGg|nt4iPrV^Wa>%W?nGf z|AOC* zljrzIBVA06XSfOAK;jz@nE@?hEv_}Ui z(?wO|9}@l#&?b?WDjS~lPd>hJ-7{I*j%@RCOZ461hPk6^hPN)3Wh_IRR5ttYTQ@|x z6Rrl>x4#DB;CZp~_|hpr9z5pm8#-5(9}_*s!8y|1hB;iaofwpP;~yh}iU7Ce(q3*( zZtufvjND?*)l(J!{?6cZH}jW2OX+HcVsuTNP3|zGx<6G+THg6p7>ZvzHPC>5`8-uD zIr;0C$nx>8Kyx$XRr?o9pO_<`K&5y$V)8?7PxzU-8tyl#yHM@*&)z$?;f`C1^i~&l zL%Q&&tL|?@YB?0tYB$?2MBP5QM#GLfnj+~p?=UYj%Fb@cpatAU{X0rk_fp((DTQlB zhB9)uiW!RrMLSiwe-j9Cipo~Bw8jD#S;)< z(!r4lZlij~@&gB3$y?>7c$;@OiCjs)iIo?9o@CX!05zufmg;j0Ee1v!!_+;KbVBZi z2z)ust)O}UlG1rn9QWXny}V!Eh9vCpC~xb&*{p>-k}wcjb(9Gxh~Vd zu9+n4H`{c8ix>hcrJm>#LqC|M#$zFTk9kA}bJQE#TC66=l-)^T_3q;)s>-^oNsdkH zvAO3a9z>ujC1vXfWw~}L4GdPc;5VNV)MSuc`lYR*l;X! zt3y%01%WdLcf68%7EJ1~>7~PBhiX&+^_tMGaR{Lv*}bf|vc0R%gWi&8DUdo}&%p5g zFRZ?J-y#Pu4?fKPkJd$V8P@HMF?bhpr02hYag53d^7jpoV})%5BcV`Zk}?w(3x{(y z->47+=RZG4s6!xi@*k{{m84?+ARA+{u)R-1Ba#`|cH_%>AQw%@HlBp5{W4*U>)Q*_ z4z1(f(vM%L%l8WJJ4SBNAdm}k$5ADI`*T-^x`8hb?B-wg4DT24tzI(;P}T1G9x=bR z%`(nnTcz#c$q_yJ(I+WrCL8!>sCVx#n$3z)zw6He8aNqsii4V%h?Y$4KO}oi9DM12 zNXV6{p616SC#$4lvY*RdQX;EB*n_KVoyEv=mfd5%6I;`s9)$x$(2ISkp4DE)vGwk4qwEq+KbB z7zxsEse}rW>=i~N>|n?KR8YA*)j7=Ey@Z@Y<1y8A`I&4=3n-8=J9x%kD^?xNPM+eL zc~C79K?o@l$4Kzxdq@)R#d-A*2we90ik=Y$A^CJ4D()j*c+v09-uq~wb46Nepu;|! zIg2_(a*7#Jr6DnF(#V{DZZQehbJz-?)z5~o6BSw{Dnf}nzS(eeXGwXm>7cuhfXH7< zWmC@3JvMO%>d;JIZ9g_S$KPQC8{3HvopZ7z5FFEipum#9wr$v|ZxU!jG>~*Ckwmmp zgGnbJ2U@pa3U7~OKA`JDjagqrT7H_Djra`pGfvTudz9>je5`O$AMU52^wYG-P??$H z;#)CLLZv$Vef={MzUUa~$$asysk*DkZk7X;wp2Gl-aaKh_l@T>?`D|<6HFbNCn*(%JmM}dQZalj5G_HN> z(e&kTD0P`ZU=*`R23YBHAmWvIK#1-tXZ3O>{Zk z8N0eGn%=STTgKpE=yrU6YBejNEb301XAxRCh8V8DHG7_6bA=8v{r7bO!kBQxupwL{ zPrph|v7Ga(a}fD)xK?Taw%j6TZrnS4OU;7!18%wlc~%~_;~ths=;)s+mGd53K1;FiG^dT2VZVm`T`Zm4yDQiI{;r%2tp z-Z5Q8Uiof@Z@!QDLd>`%U)e1R$Hh_?xb~b-;lN{1 zX1ZMCMo#1?XujuIRp)k5OPzTeyxGx&?ofW=2CGAdOHVBd1a3%h#OEzh0C4GX#T#sK znCAPo-wC}6!n(>VIIW1Tl>X~2eBZFNx^7G;N88RA0)DkaA4i#s;-ZyN--x1EW5JX$ z76;RVx#7Fcb0VL=&`&CQmB#YMqVBJl2MU-T{65vWl1zsBu- zl-FvX?95N9gxEw#aFB}AH0FcLyOgN&;~7CGQF+8z=Qwqod#Z?yW0zBbL?o$S^Wz$> zk?;^c2GL^Qgp+R>etb!aV`K0~DSACR>W*;CcxiO^gg?{n0X8dW@Ld$?g%mSwqAi5I z!p!R$9YVBh>9=8Z;?xz=znKm>5Di@BIuaNXC$OPU5ocDTAXrsT0?zpZG&Hc`%L*Z(XGs6B<9oT7fv3N;=&Z z3MPk6&)Vbmo2&+fhe8(fwzcb3s^729;n?M*(ndRFR;-GrP*o1wCEnlGwo`k#!`ftQ z-z>#65xNZ*vEx&p1ijZkE{8lD+<+br@*KS6r7un8?`PZaswq^RVS?dIb&DJcih%fS zMQ^K!^L_Aaf16lETHNLC01q4Q3%)w8Y*|+>xqPcjLnjlx2yfL}Z4YPJBr`X?arpWN zcYTDJL-F}b=7?mIC-pUUaiwOP7qA2Lwh+q@iGR)Q8cJ5Bi}6VgOD3EmT=o_?7}OgC zdDq7;EC0CpY!{{;5&XQ$eb*OQ?Kjt3+8!70fOwS0j-> zKWGBT>?d#EKNax(hM5v5-dd#PD2zLhKh^HkCYK3_+h^2z%l*as>8>%T1$GBvo9LnH zWBF-Zez(tXFzn~?N)JrIYdwEY-!Q>@a95<+p8dB}b9*_vK`IWGZWR+m`gAz7bf|uN zsuNf@XHBtykgu0XfsTvqIn3|X4N_rfYF3hWbL^!CoQ4sDaFdS?h9L0@l0RpoNekceM{pp;vqyS|F85iwyy;>p=|*}*xzVty20xTZitEYXY-N!A6Q~`iMAR$0~292$%VFw)c(>K6OL8tvr(bf zf7Q>|xezN4|6|H%NvogC70AQ{p~>@wTUvFsQwM6dfb>BRbGxZ>@e`IR7Viy=GG>i& z#FkfBb*ACyC(~oUrE^}UtoJRu;~?V8E>HXkxOW&;xe&$p_9*AscR0zNTYp`{9-ac0 zEC8~l<2R*)RnxOOL<>!sIqO7(O_;S`o3nb$$G~zQo7)o!AH~LchQ>nMY*yMjX_ui; zEte15R*@%mg2xMjl)(tKzhkgWjuL)=w2Z!#Gpk0Xwi2u3eEIumwE<`f0X>?GxOJ1=ANlSfT@Nbn^%e3?= z{enWl80bAA`=OtH<;A=LZ>Z2VKnuDw>QZSq_VxWcn@bQ}&v#_VF*#k2`O~F&o0#D^ zdGd!iZB#OpZiJ1m$IlA;1Oi&&>^`s+nc~Viu5hJpGmBxu^G~;HqAb$;_{aI_qUXl( zhInA^O3Ci4@`x+_IE}*4`q|uZ2eL-L0w&aBYim0c0ns3DZJsLjoog&M>pew+(KfcB zQ5zA3e%&1Tj(2U}z#rVcLoNqX-n)uhPky+I^}Rf}Sm`J9%XjsmZ1mY&aYX&bhlICE zv~MC_(?-5NHcH?yso7~CiDrqaigzCr=}nujv0R|gwLK;4KHX2RNI^o@L&fISTc_#; zdvkPxr-jr!daNOj6tALhZf1(aLRVz(EhiA)H`xWGv!J}CVzW225Vt?^og?8kfDnJt zBi3ORUigsBc$S?2&igK6_NM&&@pJx+G2voXOr8 zt5Itc>WU_;6eTHk#Pugw;Nvgf<9JuooPUld%=jeF_7ata!{#_^Ja=^(UgqQELp#Rd z88+#pWHfUkuFw^3N1r_~aiQo?^jfe5&82;O(7HHr*11Ft!3n6e-64MT@;lcA*YRcC9~E3KO6lkKT8ZhfOS3=L?C7d7Oq`Z!7sjF9l8R*+E6I*B@DBV z(i7_4;1hpJr9W3n2bl!MR#O`FipiRURM94o3mLsa_nrCbKf~B&RQthbWsnN?$?!XG zj=>c1_dJKPxq0~oCtB-`NS-(@S-qz2OZb`@-sO~zP8mYCl_lmADVhxEWzo_0B%Ec0 z$CxLR?+M^*d`FX}d7gb{?b#zD?3YpDL8LSBF9*)e=DI&*)JUpac{SJ@c5cb;2Ly>K z_+z~lEWM?J8rI8lmc62HtB$p4MGuvgFolW6(9sZUNZgH#K})3+cOmg`Rj ztKmeyOX99ZJ=oAHLavKJeSDKA#~2U)<)P%yDOEzo?MtT^{fzdLlkq1MCwhQMzTa>M zotK@Er?SiUuJu!w14!{$~xUDqSfx_u%xWK646^y6tbdo z+xDB!ZK-B=u<5H5O6Le#>6ciE>p&c>A2t3VIoSC`?TGw1S~=M1Zm3-FNJvMU5Wtz2 zmYiakabobq>&HlPrTP3_j4jTmQ3$kPR|-0YO7vQT#4E%9y6J1iK-^rfy>PS^E!%=o z+O!8mGJ@@hr%HL&`dW|MI&y}kZyRX?5%wI4H};;Q8!DUs!(EyMTXA(9+*FNrFE}3$ z^2O<9NKQ@VbD%!qOHhT!Th%Kg?Z^^|P0xkve@K);N@4{#+P>P^+%qy#{9UiPm$Y%= z4n(*QyC6v;{Cy>FV~*o4gqCulEKNT7Y8oM}`yUCV;75E4mBV1{m3{$t+t7iZc|~`e zdrG{WG`rhjK4NBv6Z`HTl5deQuEr#}RsORq_PTy(0zxr^@f>6E*6#Qkh}!**{fBow z)6#zdsfc~6`$A#Qhj6%@9SEt&xwm)BIYBh{#1ZlE-A_#-48KUpfOzMfkJz=#OwFSi zb)o|0K|}}f`Ey@?3k;sxvEQ*HWO_w_JLT)We@I~a_(vNBe_5E15T5Y!yQmpz!j!J( zr|`clhg)ydH`~%*`22cK(kCnK@+B`74)0*O_M7jTH#!CvjKDJfbmsJ!liZU5^UP;~ z8R&|*^W`3F(n*l(u*ZiOi$kE?xbJ3=PV~!*b-Th7ThU&G7nT>0`)ngMdc2K!(!k`C zTZrv?_+$msAKqsOF>~_MLz3|AEAeiqFQG3-7CFMSuESK5RnnqF&RaoNgsT-n>#(c` zaJOfE{U5_5vuP#hS*&}?ypvd=;mFmD^f*Jqr6fOOw-qb1oc9&8}#4^vo8!=ZF`h}~S!hk&WnUn^q37~s(f@HL;;I&20zB)PZ z#JNp|?x46W`TTVXc?$1dcJ4=i)nZE`FQ@$3rZ24%*Z^%5I`u}=J8Y#k#ws;b-O($e zrWfYA6{$fsKoQVil^6t{(G~BFwj-Gue(76~L>kApnbKPP{ZP;S`VKc2%5;u0#MNfG z9h76Iv)~j#ZIZ(M5cT1Ngc%ph4*T|+@rgLI3qKj5knR5djAr~Yzd&&UQ4`NkJMN-! zBvI42CNe4oTUnib1NFOcKiNrcy#>G^hA(KEKG}>;_nx6a;XrpCY$3T~th@2GL zeg{gL54MCjy)}sYPB62a8>SjSta@?h>fh+@#A`f6K?yh zrxhj*1BJD|ksgDaWrW*a+?~Q75ASDc2>Vb#p2>!BEfps2fK5W5vm2oY4Kp?$>bWjS zx+1+x_z~#~VYagv-FJk2A!G>?YNzl=hXTrI7)d8G66&z#1r1>ldml8CkgXC_>NJrR zJc$tFMW`@DeeI#G?mFLp)hN8UZg-JLsLq$)t1M6dGRtytG^vBAwjusy?UVPIsOqxd3(D&(1o>WkOk|V=@x6aUoe%^mol&Wc=+2)k z1&>&Y5}2GhSPU}6&n*x@bhGUN?-iMM! zJ{Lp~T$F`olEN6C${k0h4n-A9DU4k=XntekC-P2iI0h+|1O&WY^O0f0>k;eMIG@6X z*=l1oFBATucBVQ^B#Ghp7CKTZ`g*t4gNZ!0fIrC~YfjvBi35+4d%}1Atv5ONJ^K4L zU6T|qOwn=PQWP@Gb$ivSk98gF!N-GQ6;nmg1_SC^8Sb3ck(>h@5m$UY9xm7ov6K@n zbd+;xg}~{a;Yr}zkQ;r8Q{zRW@?8vZYqy8VxRu|=NopBytTBV1)(A4^Jr_lJOJ8xi zI~TRpVJ&7qsZxkqv_@i1l?^@FCz8xcClU2U`MDW&j8iGfmx~RE0MOT}M(r1ym;h|= z&`fI1o`}i6+7UOqtPLFe3cc<3=2~o|3Kf0&x)x{Oboa@N%`VM>KBnJ%lCg@2!Z$%s zZiRZw#oomQ;UQkS|57ns3T9e39*VV)Bb>}eQ~CyY*>FE8`C(=HnMSC zZHefVA)iDJLkNxApzjZ^DB{b(V}~7RVKf((S0uQZbns9M)Q3%@O|flbI@9m6W(2Hn-s|l zih*=*YIi^1;0u}>;&8i9sdB8X& zMW@<$I>Q@FS?sFFXMesnumz0$(H${`guU7`2&T%qvM?0R0<38@%F!F_i zyi{62V8olD9i49^po);W{`!e%D~;8k==n=d497M@nQm@QTR({e65b;gt&GRxZct}9 zv^>m3ZXHWpG^Ix0fh9bmQ%Q|E-0rmG^2*?N`#kWCG-l)U{jGJCpvm$8;u2F_yPV5= zMEYz)gktv5q-E1f73BcpKl!d(Sr2>wxd`XZ?xLK--I&u0APzh0SQ-cpkwzPg+{BBPhF1-(B9N?koc4_Dnw zzDYSb+D`vnx01TOi+{8K%N0Q&(j451<#Il=jN86O)G}MtElhj!3?PKY!zAqm?;9dw ze_r@pxpqlKMGDIhjD;7z%OVJOSsxds^I2yV&+MAO6HaCqNzLh=z&s*<&$*%v!NL_o zy)V!g(vF3d1n_@}rjI0l^Cs{Ffs2J{(Z!ZW_NH}!N4MxQc@X8omb)fSuP#YYf%qen+wkLF?{T4`s zzqa(8D?=xtCh^%>l|%B9^bqFfD%hM7UxPqodnLw4AhXR|hFuOVl!Qd68BIBFiTu)2 z!ZkAdn~2727CHVQg)8NpBQR85_Lt_$4TWg~Lq1yWAKdfuo$j~fC+Y7(+`m2Q;)Zi3 z!VTv>zB+JrMK@`5{^Gzml$imRWpl7Gs0-z=rZwaj(fQd5{TZOH-x+DuTp$1GuVV3C z^Ji)bCr11-3X~XUHj&@LEBX+-#6h<06AF3KBEiy;eyub9_Nupy5@KKeLx26A zmHZ&mPO>^Q4lmb`im_kshLCSchQ7DPeA;@v5Slj%AVK)0$Uo8X3OtM}Y26`2YsTt2 z9krioiDuiWx%Js$KYwTPA#4C&mDw>vnjp9s1A)&+$5(t?TO!=QRYih+9a43u`@}pK z<6Qv>yfgk@w?{?y`DVR2E6x_8wYUy?X{65v%zTQ>n*B!hJ_qU+-UnuSF(o(yJo^S8 z{9LOadrRG-)(xece#=`KE)iI7jik76cAZ5eDg;`gnYu#G^k&O<|0Ju%@EZj)Z@JtC389C zJ(DXpL;7Efr?Ose?g88>Esm=E`tUK2d$%~&$aCx_!PstAumXfv`55m;>mQQ;^iAkd z{2jZMKtvwZd#8yXh%%nsCIaIL;`aWc-b{U;Fz0rh+RY}yq}H%6)4wLLk-$67A{xfY`JOQx-4Zu<>s)M*MTwoWULwqVcmF#0e#`?8aF)B|jehE_os$WzA2k z_Bo%AjsWus1rPh&D#Yxl)Br30&84V#*P>nTPj1<0qdA7%BJe8dH0#x%WjHmNH_mM5 zY_Uy>uc^p#as8+BQJi5SK$X0}1CJi9P<=9($A)Seo6UIhH$T&%&KbxZXVoyCD%2vB zs8SBWbm!$ww6vdCyJZOJqJtoI9KG$cEf>FI!p={a(P`bFAoZK*Ij(Xo4V~(iHGCe) z+poPIV;mc6(ay$}FlTQ+7+{=ae(0S94==>(Le>*P<-K zO3vk^fw_V_r*&9$BhFYkqDOPZ6f00g62p-V1ns^Gd(KcCI_*WGR(}=m4*!GiSgI@R zRt-1Sy~=a;)b4XZG^wJg8r*QTm>1gHRwIWqv~7n*RxXs_A>a4AsA6TDRzlDJjY_>Y zM>mU=0aS9{>{bFIG}1U_sz-tkt#kWR+4%Ro$G(^k{0He}(?29}{exm9HobAdE16Q0 z%6Yr~ytyEgW+Mgl#U8)2E8bHb1^d4uutZFOart%jHsr!X0&Y38iav7b;a6Y5hOy2a zA#7Vn**3p1GM?i;_SYLX35yX(PN(NTv-tFwVAhLj!FL+d@Go+8{l~bYuK@(<&+BxD zmw$`dTIHA(_3|Y%q6J5#hN}}dKFNf9m-v$riYA`Py@4ooU#zS4tyGUd`E9h)@7wDrA<3m+;V^DzE)J-8$zSDIEB$tt+ioR?(8z`!CgAX7*cSENK0sof3`lB7I_$W3MDC z_he%bboe17E|0lmAbEZrW{%t>*ea!2w?t>u#{&cLvyZ=BYf#Oz*w1j;Gawlzjd8r}@_(7BZkY2B*h+WG2Rq5+s z+1QZZyO-RSIUPQ@MiU6TO#I$Vj0H;rTYq%*-r+2*H-0p91RFtLH%mkL9vQ8;oly(z zB-~xnMYJxHN(KLH3L2e~!PX|8D6=VCdMqOD;|9(ji{={g6+P%6X$7F6-7p;;R$>s% zyV2;PME{20x*RLP;(x&sY$39L{#*E-yw%dmUq_6DlD|j1RrrtVCCLF;tY?Dr(j?+) z%>;-xBar2Nc*=Q()EBJ~L z7MPCY$6LGOaCpW1LlTMEQ%j|V17PWnG7NQ(K;?MZB^%7(W_}pE&a(wmvQ;H zaWuQB^LCb`_)2B0$z!hiwrU3Q=aAqWqv>%m9J1!x+<&~Vy6d~Z8j#Et??jEz+MBTX zXgInf^}R-L58Pl=diIzUY#6V6a>ZmwY8zHn%sgh1ysh@*$7<0u`{Kag8;la^x>th=uimmV9E)qw9ofsC+i-I3H0G%T*b9&WytaEI z8ea|8jgr?%z7Kw;^(_ty7u7c7Zf$A--Y{E*ai>9g?VYBTS_@!GctVKx68co-TXTOm zJNg6_KnS?-C>C@C?QyXq5s7>jKE+?`i|xxD*ZwneH8h7=OR^MpqPc)**pW8F6Kz_c z@93NDXMjt2@?h+ex2{Io%#!k#Xh|)qrCS56dO`2!pFc#jFtVP;7rSeNp;YqPG8(GEGco#YNVKVkaFp<>2lMRzggsXK4uD!*MY1+)#s zUInCzX&js60cJ5+ofup7!^Jc?EZ9n&*pTMZcj-ZmM~C9( zLglX(S}C>8H`>8k!$<07^MEasVZq?e*d(b(CW7BH=08k@8lqi!K%={bEQ91^sGLb6 z-7qj>l#En;V$+K9L;g*7>Im2!4q;)`6T3GSCf(=F`3|U@({^tk z@vCWWw#C@=#fI!$a9@?<6T>n~M<(?Z$!5$WwG*?0uH@KgV)eS?1qVtjr)}$=KZ)0T zCiRIuSz0Un&@IqAv>D8#EYByzf%zk>8g9Uz{_B~AqC0RW{S11HQIsc4tl&bRJWWnO zz$LaR&3RzJGBC=ulIV@pqn`r%AG?~29JKAIN0fi3%GtHLgu|~FxXp3T z{s!BOf7kg@M1LagiSeb8H@X+^oUfYyAro)l0We(4Sh8`B8it6CSfvTq?M0Prx}l&+ z>6}$cWM7-CbMmreD6!;6H)EHq>tb44m$F*t`iRLxKjtRnyy*`VpS4*z+|En8F$?WT zKMrOEW=!*p<1T$RBVHcr(jv5d6HjKprUip)6Ie20I*a`h=ufu$#LFzW0L7dJaAmO} zPcP>bCFz?p-$uzGUf7fjT;pNSFZv`}{Ge_1ZwZWy%kF~O{Vg#HS=@Y}IM6u3Q~br4 z-%fvOCMo7Wfh07QX3a1aAP0Y#Kq&uE7YEY8h&&s(r4jRb z4bp62UUGQ-V@pm0fi1kur_4w!axtut$a80=%k^SQ`23kUtJHA zw2WUVrl~XYBVB)U!6LKBm=7Dn>G(P26XOd*jzV^{!p{v3_xqiZ+Wng9U&a?P4O&T7H9p2Bc9d~h?gXK|(n3@WCbmyyuxN^Gxenz%!Qd|K{=4p9m<An&3yTd_^ONtJ+4q6@!M%2nHNp8l=I-JL8Zj=zt6n!B zYr5v~V`foXZ~OAT$wQXEZZC8+J1m?F&9`2)(_i*^5YT?G%_qMetPe`;xfnR(v#@O( z?Y}6wnW$F4P~Xm?hB;h49%0+~@(4D3`qW5vlpE38(p&WdevQ|RW5~1fR+hUTFU$wE zp^(RU{7r2?VU*V~^KI&fgU?FwD;O;eJ8DOILX*4?^`9Z#s{(DaJn&CG!@4HNB_wY+ zTdVH0qDXqYrJ=6-ivj^AtVwQrG))JFv(ZZiKA}0`0$Z69ZfY|cKnR&>*r1EXxk+p@ zcQ)#Z?c0u>a-eekndU>dk|p`u0*7P~7k!nDXQ4xDK+svobsYuPoT5b?OPKq`cKbKk zku=j_Xe=@f>En#hj;~fG?Zdk$>+=eyfr>vtaKpy1C-v{V9u8nw-WT&h)g>qRR8iL9 zC1pfFSXO$=9Li=C7~~d~lg4PN8$HY3OsQl4M!xM{f08t1NcpE=!fe_@e~+k)`c#j$ z6I})a^It51hZ*g(0t8WKLzpE2-z{fFH16}1&+H1rpR>C_sQ^z4AGhzB`Ejx$$fbsD zE^p^=rA6wa=FqGUR_o)7txRbIS6PaZg?)T-S-<^QkDT>ihhZj-<4QAy-^XQ!phd!J zWJPVFcM(Aq6S}ChHzk2F`T+kkqB4#CW*rBf30Qx8p&&?~ z=1Jj+7t$NFI1{w@w%5{Add!o}GK(iN$rh)7s#?ERPE6AWE#y>jf`{FCaN<$>?~!3h z2;H~uIa%@-sBV_3CiWO0vaqCGhj5iA^@q5BNJSW`o&q1wI7-@G^E=(|^rNCtNNc!3Hv0glr;zik+II^BeH0P zNrz-G;)Z4Vx-=g#oV<&-)4}F@Ba7WUYQ?ox-T3Y7RwjOkh%U00D?s^SEToHA^&Dfg z{8KO;iZ$D~cNd74(YCE`rucP421I_ZvfiM;E!E0&%*%8c*5b~opSGSymHr@oB3BS;_^G{xL#^Cn?X8j*tnGNw zup30G?ZWeeBgmm{3F8eRdnUfc*oK;CpjvHfQ;m&U>rjHrd!aXbu095MQ*B4-Qn6dY zDVxnHRjP*{lw+HNl)qTT=^GF94Gvs4wJ8)sTfEX6x6s~dLu;K}a3@PImIJsCjOB{g zc*pXvq2c9c=*GxZ%2Q7}(erjFgoT1em}zo#9`U7*!skysznLyRJTI2=z;8TA@(1;) zny6pC(tL>tHq2iZjDvtGpiT;UNSwE2{foikbN%<)mbFZ2BdjhgWGhGodvEP|>3qjN z^kT~f!=$uKt75rf`7P(X+aLXI2B|bw)_^Fhx@T|~V(a+S0Xs3SHqy1yMRQt2r*uqV z$-muS;6)y9PRA7P-m6=wUob15OCLi2hS8L@UPBe^Lx?kP{od?RHaW%|@%Xm}iJjVE zHHsf@6Ox>g{g0Z6S|#5Cg-o&it7ie>x0W7C{V&(&T=S8fS;mi9w*y$5s01bWzs{*W z=Yw&5FIB5?J5%wPf1iL1&MdZlFEz})!@GhJi=y}1!YOI7hwl9aUnZ~1#7ATrLYxsf z!#6}V>)(sowt7@>_7f0Ku2+>^yAMS0Ro)1aU~a>X;yjE`Lkiy*m7dS1YLnPwE^(`I zr(l^g${*#y|6)HZ$!O`alYM1d%N&*D}EF%j4Bu^)S0mRa$LW71K#jZH_SRM z>+aOjVttkLB@CnQd%=9MgTT`PTOA|YDMo+F@s9h9uU`yX`ixjLDho6^rhB>v->&3Y z2E4@WA9Cw(ufuD_t$U(-zjCB8SZ4SmGLizE4S&PrMq2%Gd7(ZdU3Va?MRV0GRweOa zPMAF3F%(btKL%X5LY~;&Qz`3E#QxyJh=Tr3PboN(ZSvR0vE3@sbEzI^aSRvX?Q?cRQ{z$2?u^8CQ(Od}6+|;+>kGkU!Q*X5?Wh6;hcCI}JMBaM@aFTYL1S zsg`mZHM@$%l%c0XZ%knwU7q2W>-vB(_Z{!0 zl6w_3BLOQi3m{mATa&!3 zl2Rf+^tylJWN$M7!*~m|#$EC+E&-Njipv5{tkcdtB|UC+lp`bQtuM?f+D`}HQwICR zGeth)HecaVQu9q({TijP3WW@Fs$0&+ahXY5u;lwrEs75+(C5p^dU7`AEnUS~kOEYBMst z6P)HaO(iw9!wZALS(8QnJj3?lJ1ON_#yz&b{k{M&Vj%t6Z-M z)r^z31pFp9(o=r)9lx7PmD#vYm3XxIF-=%e)}1w=9+);^Srv19z>nR!*zO>S>PmwB z4@ptGE%->nHD%ENr(s2f|MAV zwjiAEcRD@xMXd>rX(62Hf?M_4s$=n=a1p1mAmdj}UlK&N(0J)#G96Ef*fJIUA|;*Y z`(+gN2B)2PfAyCNhdVr2nQqhCKWs`};Xf3p4Pt{ZY7#}gYj2*!^)K&E!&cauFAG)W zEcSiNgB@1mZA#riWyFZOlpKq&xCUv*JrZTYqZFQ}Mj@z8#$^5z3EObnZ4YYW z98beMRwnlbO;Ua1eN6gpftH2iC(`JdN`HYai`FC_!&Xep9}>%Q*Trl-w(Zet`jH%j zoRQwMUT?qtM8Z;3Cvj&0g&<4pH5GUJ)OQ6h3u%7rdi`$Ww=c7`U7qHl?rw;w1HJL8 zW4qakwjFk3VysKduiX-wfrDBi38w_@EGHaz3ZINpalHbD+IPbmu3WAD=F+dJafBtG zqh%{Behq)rvi0{ICoHd~L3&G$wGAe6(kFg>p_DDK;j`hbgT6H?rhp;opq4(^M8vA& zznZutt(!qG_6nfkV3yrEtDlcU!AM1l4sLPe2HW?G38DXWAvY^LYP110EIw$(VIh-{ z`m!h}pYt^}(pz-KcJS@o6IR$!uZlOW%fzetBb^!*rEeINzkZn(PYIH+o*|7DJ*ufq zWGQl)bUl<5wsUd$WRK{m<4-X^x5m#%ATWiSWRM*1xLNLYm4sHTZAkLc4RAURL(~0 z8wzgiF+(o{f2d&?%8{VFCpE1-eRb6EmCHpDL`{o2l(l|ag9v5gABspbrg?16%7k*T*}1#L{q|Pe;3J zRm+7S$JVkW8+G5djCdx|*uUNtwZrhMT6VIXLXQ>%lmMq&&Na)^1s$T+2_x-`vS_0l zuekTP2g34RUKOI1%{+o*;GAXgC$f!=%spN|NiR>;eH+$+T(WivUjvs;ngb0~rM~oO zL~v(0wX)bYkmfzOwwENT!as06+>4=|r)G=9pC@B)HX_StK&s+R>lazHB1Zh*n6Wtz zYzobCV={XRea^3+4!sH<65NHRhJ+)22B~RQ_c`fvS^)Ss=u^LDop6MT8~__DPR8 zpc`S@$sz0mHw&dk<~mgee^dd}U0dWSMfJKg39{O(Tvgy7<)lI**spd7)cP&p=b(2uENmt|(=jejTYq>@RME zx?xWbD&bNLEH)kH+qdHq>HA<6ClJ}k{)_6cwTY)tJ!0zPgtig3ADP&cfo#4hj^6wC z9?mLLDJ`t#0+;DQLjfPb*$ZpTeFyvuD|>~4TOPFOCn_%=CgIkd;i%uE2(MC1DHGad z5md@{Tyw5D-*fNoJ3|iijOD(f7-wKf%1=-c@+$#*_kRh{j%q0uMl9tnu=~WKaw$5k zS;j1OKYeAoHTtmf9s&g@{K{D}$g1glzBEiE{*^92T%8_hTckvF4Blv{2MYA#{WRbD zsD4xo&M}NEUXAA0I;EeU`g}LJv-l5bTo8y~Xj9NIX4M@CsuUVA*~MIQ0Q^fpzojw@ z{{0JK|3hC!pfRMTRB(o?-U;@CAbEkkd3#5a*5Z3Gp)8Y++n3yf_oYUSl$6c@pf{b6 z5Py=QD}x^WXczLK2;|+%}9tUvmtkr8gf4K)SWO=R5%m!E8aW2Cbn8|US&h&QL#Up zXPUM(^_5~(4Ajsvk0+jpuD)ArJ!+I;VehfS;P(w4M>Y-nmkJ263fE_t^wXhXf`{Z# zXm94g%X^NjZrqg@Lx*n(^i0jom;`v~cTf7FRa}g4IhDgfY@?v@1T%esOfLc8Q%>>w z;>6uNDWeY0h2lFLM_ARZzfe;1fqTrm&i*gHIjc94O?I!f0kVzl9IKf7Cg(Hwz^qJJ z8$>q)Zi^y%_pw^1Ig-LmEfwywLT853mS{F}eOoe=x#Tl5%vsu|{yWplRp{l%Q4&H+ zOr29CA#Ih=Jg>SiY)RUUB{=@k|7!hJRm!rrqSu)SN}0CFE5Jp4rc!TE*Qzk^a-bq#p!?l>YivHjf+giq z+&}h=)W4=IW8WKcoB&W3*SYcAIOVx%-(90ULQzy-6z8`F18C~Scst$Q1Q){miLVDw zz!oe7xK4hW^M9fctq;$hMfCvuw<}tF9r6r5P3YpEe{9$~`wwG5=IH~#hu3)d%|r<{ zdEwBkwtzhsmW2}L{Opc9n(Hb-bkxB6ScT9v_MB^qAU6uaQ_IG4qc%&5f-K<*{EQ%O*?Lb$8|VEJjWVmT`-EYN=$ZT)I4D&q(-l6-4t;|02Qet1Xhh@eG z;_vx4lR}D|f`bfGrNhju`IMDd-UFX2aNZi3W%*B>8>2ifIAi(kMynfX-tcuyQhx3E ziuu#^3EWl7BdqYPC4l9%q%d_%-AA%M8%a|LQ^6(rYXMwc!~KTK8bPDBt8p`Z{O*yE z(*XG0x(fdhnUXuVlVkD92R>D^ap|hX8zM)eC|3@f-ZR!Ig7ohxJ_WaHz~8Ij#fD2cCzVv7sd^n5HLXRQ2>f2dTGbCMDL=m| zdw-4REf&zM6gHzCD^Y}^OI+FPggX2e|2>CxZtDvj{(@*h4cqXFj^NzjDcDM*6b{oh z-rj8oQ)Adb(~Pt)zZUWq22TU(5tg@M_l~GJpY=HJ9*mqC6nvyGNqH2x+p5I3b5dcJ zrqbrUSt=Y>PLw5J>&Xr9H#t|_nQwIqL7tuT`|n~{7=ALmacZBTDCPyGUkt{<79s`8 zrNnQ`rkQ%4R`UeainT1o8FSy@m1&OciF?Vgv+R|lchrlCv(mq(uY-;MsyZ@R|2v!& zQ&Qx;O`5K^N9&g|M_+WDZbQwLbXT;wdMSpb&GA$p^PiAeZvT}B_P}MYq$LXc&pGz|JAov9gEwG|7eQ{C(QcgZL9$|*5j(E^fS<$ z**D_A!-{FI2idPY=16<9I^fzrrsvJEHaI)Rt8)Mm^92- zR+#;OgL(5ixeZ26OVP6;o@p`89z%dpsX#B@#Qh8=DdFHH6a0C}Lc{K85NFttS>%;4 zfCZ+_YY0;-#E&{~6J=7yKRi+C##6f%#*x!KF*v+7WyAAjUbGH-UevyzpjFH*U5ZYc zHhTu>0qYrO?S2f_r^GxQ09bu!Zoag5K=LEC_=!gC+b<(AI#VGJ!z9^VfVhNVgwEMC z1lrK5fw%!nPj2-uQzYsn$Vxtm7pm`KV4nK6HQ;kt>Kdd-N_bBT-SQap?tr|#F_C_% z3(j+gpV>fgWbjPFLZ$dgryR^CLHy!{W-e6o`F0-(t_qhCzh1iw=Ho6m7u>g7H(~X@ z)DVN{0}plra>f#bgZ{_WIzx2EEj)7Ab-oSx6%yjAP1Ay;nK}W4xJHs^+|2)bUEJ?L zTkuoc)~%N(eUkEBI68S8h$ zjq?0S3Dr;a?Pv3?b4*W6^!clx^t1Wki5JNx`sP8#T zCx1wzQc9Sk)ib0j056gF{e0AReKblxq8~_n(El>0gN#JwU44bedt3zio&E)MQ8Fq= z^@x`e>BN&!eMO##3E!s>xdNZ;M1DJNIRP|+s;H3)GVD#piZ4QvckxL=?nk-fx8l*f zZIn{EpF}y~dXKR>z7n-IE*MeJ*)$%lf7y!FQq44X`m!xOUxsT+oO; ziu7P4`A_x5`QE}s9zLh<=@Bo-KiN$GVbSn*6J~ruBfE)+Fd{HWT8`~I4yGwF-fgU^ zxLUiDHkNw*UaHxG%;hMpx$r>23MAqJ_h@pyz+Ap5Fu?+q2fa@q6vR%r2)wGCHbbVK zIk{@*d*GL2Uyn=b`Ep z$m_YbGEMQsuMkl`22!~DH|f37a*O>!;UQ+C(%1Y&UG&~nHX1-->H}@t1wI?A-1?Uj zs16#nPDz*eeR{j!d!{DVNoZ-o8bCrA^N%^29M&d z7?{N?(6dgWD9F#tC}I^B^=M@=2^HRB?_Ukh1`F+N)>X8hWGKDoB0(vSdRG_YS5@kd zqTxT;UVY=<=(5^5lfZ8fNnAMwFQa1NC^5`Kem4vS|?Q>B?j*FK(UTH2m)n#_E2( z8l6ol_PoE4<`Fo*-T?GXRr$Qv$&YfTP0|^TnQDFhd3ppvpT9V50(zSzIu@_p5b|tW zlC7auaO2!SlSMD~qkQm0L~zrf^sSr&&4-ng-MpACF>2{|4aZn!O_T5g2p2(Jh|7H1 ztgaQ_%N9RGK+`kwlTqPNsSDa7TWfhEr4Ol2Zokvpz6w8ena_*aFecy=BEl&+)((_Z z<`xec$F5j4Npy|e9?)|qBm}cbo#gXjhGL<-k^oMQoJoO5)SM zLMuH5saxMK)st^bMU?_J|h>3q4f@8 zuIl{KEuHEf!?S+TjdpTWW7D!P(+<4ol;t{xMPKm97IlnuV_ROu1NuQB>lXJrRgfn_ zG&=k3J(ojVu+UufYe|g|E<>ozG8iUpsGTsTl1yalc2%+)pi5#~qWK8x-RWL9_??2e zvLaiwtZ)-W#E_nhRt0q)Mpa_KUE|WZpz9?%Ke$!TaSM(0wQi_QGKeFQcElPJW%)Wz zHr4m7^&*tQy=0JXwxJ|g0;am*F6_G^{x+x?)wCH2fLg$ql>Ynsa6Ha&*d*?q4p?rJybfR z9u@T=Z~|%Jukw34mpWkOcI`|DvzzKxVu3Vftz}=8^f+0u6)vYCUHz@nlhPjXESWzD0+5 zynySAd9WJ&YL2FRnvKy?s@9n#z<~ijwPTwcc|RN=xS6byfS_|}IUl4h5-=1gf-bDk zlU6qyT{Uerh0}<0WIPf>oy;eUBOF~?bIU;7tX+zkfA)B%L$_MuSC8F^Dy%09!<9bq zfvTE*{c7O6CXqc(#Wt+PPRN>-;pNM9+FKpC-x9zx@WA^UwtZ8xt^H;2DCcQ1Eyj(c z{B44>;(xE$4_|CuN`ul%`PCeo=|S9zE0SP#0*bWeZ||4wX^Eo6oDRXiWT%FzbA~y~ z=ROoz*iIne(uC1)BVFtAM>*SuOAGenD&b)fC{19(7lX-rjk+^18f&CxLY#8w78mEe zPR=!NTH>M6>Sjm5?d$VwHk~RB)zG!ay2xEOR%LTzzq~w#sZ2eop##ta2@ljhZE9>& zJG4pikFnfM`J!g$V#Dyq7b~s(On>#q!c}VK73H7yh$?PVI=v5qxew2`cbeP4P13zl z;JM!Nl9~es6&qDC_0usMYFR(3!2TPL^;zrcYaa85G{|DrJ>>nv(y?ocQT^?WIfGg} zYHR=Li%z0)`j71v5r~IM8HSmu4=Ru9CPNhZzF{IpPF0^8{vpk?ec)sm85c_dB*ma} zJw)88b_wJ;b274ENIk<--otV$&Ykj>-uq=hklY3HxqJ98+z7p`f#PCCTxhox1OaOv z2VIxC3*pbO$voA+%4?e0kk!lkdK!(UnATO8|EOGL&HW3*NpC3~)fX!4 zEPX1+49|uOtx8^}rb8BEu}=LoL$m@X%kjfa7PZ;}kGT1ip;)*2v4_=WC%k4;Q;5l` zWfQ&BqK7oD_wO&5!$%0}Wu!j$Yx{ma;uX%YFJ0HY7FG9wwKDBCz*}8nVC7gL_*)1Q z1H>J$*U!225*A(q!Nn1UDvjpya~Z`)hcHtJxBU^WqFM1cgE-Vl$SNdn8akptWt))@ zegz4ITu=le+5w>EXYcKDTV>`7@DS6!je6b8qT*ot#8&%{JkV>j4rapd00UhzNN?<; zz&@(|ls1K7)l;9}YO!;`0pGf55*eQW-5LqkLv&x2n!=G=XNqXM@+RYp%sVwH3Ipg< z*UhzX%sX&hlG7+hD!uN_;JVfiDGn$MVaZ`9w*6_5)$D5lzG``h7ISw~X)hZGR2yMQ z4upRPBA9WVVnGovK@2GbyVs>mAQ35tUDKP|6P~=NXJ@VZ& zRs-c^8i05`>Gnfnf=XRfBVF6qEut5dL`DregNrY!eqR--s~&0k3_fq}+LSUTNi<

z_AEhI_@Bh{)rL~xtCTmO0`&x?QBF=7d9`qR>(Lhs(g!+Ue=&_(qV>ws|2fh$v@&zi zNN{zJIwG@Z9^hrY1&KmyN`>PbXJ4&*XEC+wRTO%-d|FeLbvJV=g=7W7{b`X zrB!#)NPRGoSivFF+qBf^kj=%trVNgLE2SZO1vMhb#i`Rd{|z!Y=!D64X1oT`cEr*0o2p4|A^fpxS|&)(mN?{`Vn8 z-rDZcAUUF~LM#%aMZ+aDk%)%Hge7zdiSJ`|OQO;?{H7_)E1I)hQZJ+01uM3~%Ag>- z-_5C|V)d-mHMHO#Epb{KGZq1}>#BeTpJwA5t#8XtN}ESRhGY82r11D|qBQeJQhb5k z#>~Fd!W4bgDRNYo^fPBJELb3wOVXfN!R)tl=77ZHHelyq!J<##2@3gkU9bW-tXclC zkk9iYSV+UeN8K82H*)&3kLqZalT;n@QpqT{M?K?rryrx}hG>=vqNPEhaBbGj%RS_# zC@-^p=3F#EM(8}7f-0bmq|e(8fhL6ZH$tuANSykeycMW$7B;$Y}6I+ni&yEqqb7z-<~ zV*&BHe-KZ=lu~Ob_Y{-&-l5*ho3iO(bQ*rg1uwLGMmg!-+T>G zpGK(!jsmvGWcK8{$>J2!VjQpY9s69AM?MecrM5(M&!}0VJgR#Xosj1N2kOXInZ5Ko z*SOQvDDgvRuuT87bj8a}AZrwHKSg9FN(K>P?dqx1K*3VGpDta9?h4npbW(^r6c=Ls z<5Ym_mQQ#jm7=T9#K`a}g7d8uy+pxk=PE^}@JBL*G|Owm+0*`NHgHT@^;_4^dQ{pR zz`&_d1F9J6JH7!Ue%3_}CjSG)UH{j>mmMs6^n7TvHKR7vHDWjgK;|T$Z0;xbr2gl! z6a#O@P7%k9N7O~!PZRHA-5+P)&WdScaDhL8DeC?LjS>(K1^lXzar38xiyD3N!fG{< zpB_cLBed)CKcB%2sYs9Amt|^HMIZyxF06J%Pr|1|=+E0)am@rHmh_ix+1F7K;Rhpws`t|_X!_w zJ@0)k=+GKV`xg5S?lQ`>erSAvK9OzKLV7MiQRsDW7vlKJZ=f^Nr1spnxWoW`BVhD> zhTnx<@x5jRp3a2)&$Z7rr*hFwhQ&8Gr4tP)odz+8dX6Vro$}jCp5Kt8{}?8+XKdW0 ze2t6E${4a3MLny;+Rov#h22s+l8hAnB<2+TcqRyZrK$6qFHifr%;ULYT~-BGg!v_- zUX4$zHnsnd*ayek8Sf%Ixe-wmdn$)`t!Tef4^Vz2NbH4Qnw{eS`~7g6Yxddi(|<^N zKfUt*UjqpjxLlTtG1!k9P&U!+2QryKfh83878E-wk7A7y4{hU$G+E+rut;@)!o;qw z8UIZIApUCg#PE=PYxY!v7>90C=%gMzOZ$YQp@@#lUm|ZkD^j#l>`0y?ihN}5HU58# zt?7DpOebUg?>=Ra>U-7ikg~X#go|8@YwvlnD()G%-N1oSFNY_FmfVxTT+A7U^${iDON8Aymk=`S;>XRag}6~HXAJ37~QZNxp8JsuxE&EKV;ch?iz2K zpLK~^t5&E*6F3C|u6`o0EyrP2FETNGwLhBbZE16k4{Z-wlSugZ(i@#~gI!pXM|wTm@yc?i zhPNTS!_O`%{5mzS65J=F{k7*(R%eI9r;LY$WgeP?WUPkg8$5p}A2K6Pg)uCc_6{I5 z9jUzybq64f0Jmd?odnzN zm_^xk^4Hp9JyDRbmPW}!bR)WwP+=&=`6Zc{+zkP1D*;GK6)`v6Z_or3c&WglJdojS zlT7k4-;^1P`H2FcTq}tiYh^`x?X!}d@m~vUlBVwLjd7k?rkihr?7fF9yw`Ace&)IA zS7g%O8sZx?Q)o|}z4w#HWe^rjId@*`itIZ%SIGT8q=?G`F^0*I|ISX|Xo)9=(6$m^ z)kpp3-G__sAjHQK5IE2ehyZMas}FJZQ2!u%-z-su$(;g^%EN0JiF4ynWz;9H{P`J# zHfj8^ojQ#BE?YM0v*>17*3XF!#>s`xY@4w@lgB_B{J)-sIl_!78z!Gj!0{_7nH8GH z7yJSI9hlCd7*3a-m87dO%U6A{qhaki(J{-}dvdo5Dke3rZ4@J^M&f8RHc@U;V+m$1(> zz3+`1-?b;cL=1lZ090bDorJp1@tprdLi*1aAZqfSY^mEZ3bL3#Cd+37Xv7~WMe~6( zGv)>mf-p)7S(krEwK}xM=`GQk!tDK4?^?`Fj(D#V5G((X9+?}`#zJ1~)Z;-O-6)>4 z9EnPb#ansOB=hz?wtBo-U4$cUsQkO0e|?2)%_Ml1e*u|6>c`45wb=YIuI~XoHgz0F zuM)hc=nDtzeT^a6wi#(i&2%3MA z65hhcw?@5*kjh{9=n!n$IP)B&p!C=cz2N^yM&XsD{n#xCq z5Y26{e&Y_%T`_y(#`Zx8I6M5myC%v^Gc|fkgUV#!im^>H=TiY2qR4BUDf>tx2H(1b zi^Jg`5+Tt2?gCGwGP8Ym%2m^oMLwJcP5?h`q=3bVXyoM!qw`W5Za-b`o;B*%L3{N- z?+Wjp8%IL+=^^k+@xo6Cx{2j)c{59hRq=<}9?Qtdue(BhFcF~a>EMdBq}ErDCMQqo!-J;3{v}%)|9`8r#<~F zv!$*1z5B@6^}2b`r}oDz^cFmw;lZ0oUtX%)80X-_qZa*%<*Sa8&wD$4htf(JU8Ce} zA022=;^LJkvtsN~TD~v*6MzHzvu|VyNNH7Bik~r$y^77UsVwfYT_|uf=>CVq0c3^O zwB)_AtX6379W>N-W7GbAa26ChC}=?XdpFg&%(LVY8hQ%_MWOJ)c&2$}guh?oZz=Vw zlAk81HtIXkr+J54qvL0xa{6cnRG6AZs!xW0t;@Y%ikMovprHrx>BI4_Aq-HynR`f~1VUH!DdY_;F>OHLVJn;RR z^ZuTt__yoCxCxq}O%eG@gr5oxq0ez$o+~I+Jy&udcbOmOKz# zauu#JYm;F1*ar}sn8@Mf?q70Okc-XvjAB3Z%ZvCxJ^1#H2!Qpp*>*B5_mUeh+J6FE zWv67R55O?a6#$2~8lrX;STHO^yK{3vj)#@Lh|hPSc~CzBsJ+|=&Y$cF+Oa4y%nej3 z@L2lqS{VU1%}9S`OkwupH<>2}F31D{a36KPXr{DYkqR0aYeakqIn4u^+^(wuE&D5j z3L$Gh43j|IQX%ecYiH?9M>)XB?f*{JL9RGJ;XLnFG2EjP(+y?t`z67j6nN8p8h(l zt>^m!g=uL^1u4bd-3!5`Kyh~n?h@RcwiI_S!QF!f2-e~b#R3Em?k>eXH{aj;-uqwj zWS*Qe$;_F(*IsLFUNO0w&{iJbn|`TKR9G=I`7D#(k-dudC{Yt|I>7b58{*qGy<>pj zH++?V=C45W4BhSBI8HqDVA|GKB2&ak>(rQJO;jCQk6ZE&Zkq9Y{tqq3XLsW|b00NA z0p(0}G5;K#hcbJ3DvWo1*?vjvUZdD!dC#3xXAo3YkvD)6K8M_2 zXZ3F^bq*fe-wN7u(s}(Em)`xHmk2&h=#6|Z$$QvEol(jXSdG&UC@a1HJ;n4pgb|Ye z8*SP&`n36P>LxR!BSiQ_V#83}w~Olk*ao2}%gOno|9j}c^Nb4~dvEGaTnBanpH7$` zl)FmXL+*4@JcLQqAC9{IRY*@u)^kY4|AArC_b5Lv91!*{?oDQp${VBy{*D#^$$ zU?9_308_T>ao+MK%4{b8eJcaTxJk1XrNMC5z-+UXTE4y0yjqLI*$bVuo_pS>WB#S* zI|95zA`M3RB+>CLOfC&wzBt){U0U@L?_Fw(&OHScW71Ns5DYu6dF9aOe+0vKx4r)2 zX&Z(y?IQHiUSFwP6P7YY&zm4Tx@!995po&gV2`aM+szH%_xE>Bs zhntG{3wXpCDqju?xvprNo!1uky%k!6O4M*6*5GX~);X;HLwjX_z#U8iJH;x``M=&i5=gT_xBVH)u@Sn{eLts8+cK^xPF%VpnTZZ4j(ZNFY79X6IjQl5;{QZ(sZC3!K(U zPnTK+K|5vCTm%z8_33(V7x~}MmoZdw^240U_<`N>YH9OKok4n9gBm=eQwC?H z0s|P)I-CKO=TXLo+F6_{C8-pC!$+n>D>A=W)||1IK@*8qWiI@Nhi?(8@Z~iF2Ll?) zz6A|Gm6REU63L*k6c;9cBSeiR*sW2TASHk9*tLW)EMkijL8z6Y5+3cR1kPgN4w4iBF5M1W$`$2^=|{$~*nSOt#y? zHb;e`l&L|*{@NyL6e>)q5*>EiR6N&IeJ=lqolpyE>VJ5fRJ&=UcP(d&6Kh_^(Lu!0 zj>EQW9!+aH#U>8~3F_%HuN z@t-54uLx2q>(?*bv=z;S5@YYZ+3D7r0g$Kj8k;CfRlXLz{8O3x2DCBv&)->GRy|)Q zsmus~SbF_>QaYDM=TVJpkLOR4ML>n>t00VBgTZ$?SFHGDVu`JV1$BlGF|N1#J$Tt> z{0_SdyO)nSzdHupwAcaq?e5W@+e9{zJPNuqs<5;Vm9DIUM<!b?8ms_1l#3Os_BMfc%tA>&qa9-dFKYcttDv>+0+#Fn0$-{a7>?Q|tCYKV{14SWndJTzI=sLC5cBxsK6BkEq%Y|^O8jsBgYsOC}V5}Y$6 z9-^o$yJ`Ei<?9rLDfS%|tO!ZqI?vlF9_3MY~Kyzi{3_@GJ$mpEo zS#--)&GYjhes;b~AKQ_ubpYB~#%d*Q%?~eWrK+Z}Qlgr9gcrFDNRXCy>GFvUY@BL< z)%(crwUlnJ3ssb~v7!>P_;Nx>jUO#xr4hSO!mlxZ1EyxtC4-K*F+_Fw5aWBZ){PA_ zCL1Z?)Mn_{C>Mt`*;#Md)V&c@NWHsrlm7@SUr$z`(eqVyno5li&+-d;U2V9VM5nPl z%iO+H!O0;aHY_>v5N&ik*aFa;)M)9@;^>H?#cr*#?%qYm&-U|K&M4r2EEk8q6=D*@ z5+WvN`LnXx;AK0X-|5*flf+?3KJIaGa!b%*uiDZBdhcFQU-j3vlR`EC9-LOL65_0U z_;MuYt4ktCh8Vr|d-OmFld4fNt%jSo1zzk6d0Dmk{6w<+vX}gpE{w@~RQD@FnDh7aBx~V4;7E=eUE=D5vBsRph%Ln z&n)~#f$|Xx_O$s%#AlNwYll7?i1!~F?fw6qLImT!)F^~X{P!t%pVw3(>X|)(w&k37 zf9Qc-apbK%&p56m8Mg=#@m1+^%1uDX{Sm8g_`ixUr$R)$NrLDn1we&*%W)!HPLG^U0b%7o2mV1a;52#egTGEryQ-S%>Q#@z zp^8ogWx?2#`o1@Ms5QpZxni4YPIeYCx5PqKh%frG`f^}Vi3CUkG&VTSxh*ht1^*Wo zOQn+9TWB9q6Dx5$ZpYHyy=VVrY4~_%HNi>aptu)oJi=CJ|HX;_Bi{42A|D46)dfx{ zDn1q(7CP0E{dW&fJ?C{Ek5jp2!NJ+5zbyMB77UCxMpvNvynp6&wFDghgODI~=IuuD zS<(y;!wuPa9>jaXlK-Lo@}#;Q%J=n=rQ;~=oqPb4HDXHcv5-UwL0qcg@fTK?!(Ji@ zlp~1;xasEU-_fYf7^`uKACy}}=B<*}J@kSZzX3-1DS%5DXO_=^0Sre=CiK-#ht^43 zt#pmf8kfTzPwli~Po4qJX}ESOS4Qu9!|*@9NxD()@76iTO*}8?$#PVsv9C=6(pz)3 zn8*?&6@A+rz#s#$=fjUX$$Tz`!d?YTOay=qF0Ud#w@vTIOa1m?v`?bb-`c+hL0UINh;?AlBr@%kdp%iPP(k!!v+71OfAMJq;!Rfra ze8)zkrScQ*%9-@=HmOOE-?u*iM)ji9l-Gt*&GJn!&Ql^!hI(?_yK$9jP`$w(k&j;E zKks_TD`xTYBOc>l_6k#4!KU9knFT<(L7PFGB#*Wn<7YJZuVo^)a2Mus_(w*T#sUPv zfMD#CI|wN%Zq!#+m#Tb!0s*e%Kc{^ic%{foj3`A{LPoD78({)HIBS*Z|F* z4nyo6WYEgzeGPfvM{dRJ`OZ|CLGWSyjN&J~HiqU@6{=sviE3`4nxE-#G9@2b>IUix z)D?D__#Wq*?5N|=t(YevY)pN=59F3lmGk4~zB% zgj^j)5_#~D=PE!J_UrTy&x_@pSm6=x3PLNG4dNXa9>k)$NR07VI5H^bPZ`>5K^n5?n1uWsZ zf47bGJIKTg2qu{O{llpV%iWylR4F!RC92wm=l~A(&R}Jh?XpH-3!xusB0N+$!Yg`~ z$5*n;BAlEZ0rQb8JPMTV!g}^(pLlZ9w$jCEW;r0T!&&LwB*!NGwq-$$UzGMfGoZ6a zv{s)J>j@m=?pm-^C}mRGRmLz^; z`2;ak(O00VDJLiQv8@ds&KD9nGMfP(vN{R>F1-^Unv8P$3w0Hsq-ukv0$MHnD<~>i zHduuHPX;krU15io5+N6m8!SZ%-aqF`PGR06#+`=)fgF`<##IQDh)-oI%cNIqh5_xh zn5Kr~E|EGu#SO0bvznnLy4~T?aV7?YX0w|*T=auGx5I#el~Mc_EkE}6;bN>T&N)px z;k;2&iQED6G(}`jbh=>pKO^=sngpr28BOtN;VWUc*}@u2cpyoBe^pb1QN2Fx)@Buz zVg*tHw>a8!oIs^mu}^^x6Yq$%EZj@4uFK$SyOspMzs`q6>L869!eoK1##nMxIfdP& zwAgIF2ZeV{;;))48OdxJ)hFF~o4fO$72rEvt#FRmIL9|bYv0$(A6aLQSZ3x^MTIs3 zq>DS{<@n!$?7^;o;eOgYM5su#N#9?+1PrkLz=Y(n2Oh1$c@x6Nxy12CW zlxrn^ala!xr|{=%7!_&Ea-yndAt&Q-ii<@`H#aYu#QcA*4ve{a2_PQYOqvxsaxXE~ z)ts`f0*Ul8G%rAhxsR2**vBVYxJcHVfu!9aLjZXC94oD+fezSbfhmO1sqe(e@fnx3 z{?jfA9r$;#ua^2b^zq}8{5x_g8H#bdQsgx&UQAd#|0=&%PX*K8_rY-8?pNfuwL!C@ zbQqymt?-_=pL8SH8Zlu^U7*O>K-v}!!f>g{xIuB~`!Q$C%X=(CXgw%SQH0+tEYF@P zzsB^V!G4?7z{o;|7ZqQpyM409)i9MUys1g$v$3JN{>M+D>2%ZWejQvTi@lu%%S(Q% z!lbGScA7D^;DKr36Q0E1tAhmo-9;qyK3njk`w?L84Ks&1_pacQ4#5$pp~d1qhDi|| zpqgfKqj<#BaSg!bh5fjt8@UFp`LDP>ffB!nKBt%aCb%~t;>Y)+r0bTeZ82&e7w>Gl z`Gte?&s~2f)70b z2#(vx?Xz|02NzHo)i2gF)c|Cak`gwz9;`N~76>Bkye?{<%bXYx<29m+<$(8iHT}ee z`Ut=PSx!|)t>)Y)yIA+x`{pX9$lAVGT}(6>dM&*2qtAONUQ{+%Nx>1Dhsk)cCgHWi z0Q^}+ctcu-nz~|NYiYqq$+@yx;Y8b&m4mCdmSW>~uEqPVFX*RwGNfKYCrP8ap8tH< ze1A*DD9k231G^;z9f>2Y$SE?AL+?6$$lcFxtivust`N28oVy` zC_ZuqHj(h^KQx;G6D3*8gto(R57KdH@=L<*^^|#N5U6~dT=$a#A6ZYW^|&8)CqZ+O z3k%^NTft@MdvEJX2PNr0dE1UQ)%!OgjNk%(*lDnLG(ict5n(F2u{{u+JpB`|Tc#(JYVAhu5SEg zsU{Tq29>`!XE=_ncT6l^1G=x}M!m=A!c(GVi&7rt@GohCcjMFE4{lSddnkU!{sNKB zHf0&7M;;BHx@f6>BKXJKeW*hHy%ifrtwzb-(Z+xwg;sDt;kxQ4<4XDw9f^R|S1VUP zOT~tUxTOTB8ztkpg+3>y#RPb_~TcvUt^lFs-gKOY^9~2Y8WO1z!S0g5YO#} zxIfp$%YEylaSD5q-DTwYsKnNZsp310d#_8GVr>0&sk4~T36Ga+$Tc4$|4b+6ZA`o3}!yc@W8*}i;Z z&SzVUVr2cSc)ET(?4!%ISo)eU*(thg#OCy=D;Sz^6LK4Q-;#NJ{(L=$hdJho1_%K* z$Mgcf{GMH=BYz!J0ocvM6AP%+0YMbk&TLbPDh22%)Mq$r~Xq@m|dSc+^ukl=S9?%=d;ei`yE8p;wccR z^=`GA9%n2!58uc|~9?1d=h)5bn?0UaA zHWJUPGwt398sC&qbnAeZ=tRdfh$(%YH8=FU+)@|g9rsFBD;l*RonyR=e5UJnHKBnu z%{2XNow59>{Pr2wXqi$FPEy}Sj&sA748fBvOmW=toB?vEX-sw+rCH2$2zXR2TYB>w zR-CMscJwR_fSYEjO^n#bT)8_z)1pe&H=QDKe^0HfTJ2hf>jKMFh36SW=?ef8!o;@Y z!=d7BOSRTEnX+_3|J?SYZ!DfxO~OxCd<=6QZF^B(hXvD390DH0Oee!= zuPLwhr4O;$Nr=mIj$jd=E6)V{m(s2VK1TT^Szi|qRj>?AXSGvt}%g6x#O8m;qm zSD5Vd4{(AskH_j(RZ|u`b{NmGFGDpbe@geL<1jwwQ$wD|oBa*c|b(x+j?t0dil=7Hj zZ1tV9D1ViW8T<+Bhjuw2Vp<&cz`u$eks40m?9IA0k4>1+Sa#Yu?mCq!wq@rlZUD_X zbZL~ei8B&1UkI2}iwv>L=+{b@WB$#WJGbXA5MCWjX+Hr^pd02#$8d}up0g4z?$J|s z)2;x)_Jrg88^fkkBVoihfC7BY6;;|W0TI{N_Cay9smZ^&N(?SJI>d^MKliJVA}@U@<0-1$dR(hdP1M#IdA`(G0;x!@&tPUIEh1$^qVB~IB@FNOfU4n!@a zHCMOu@yhXwgXShxXW+kU1_#z}aMJ3iAfV=K%@~n$7OkJ=uM2jG{5{*CV>O;*AB48% z^Y2h4i47vC{P7RzE!b^8iSJNzsuE3aCY*Z8 zYKs3n_{i2=l8X||$RnGS%tI>sKj2+*@0)(vzc<_?)V1RT%Cwrr#&@y}h?4M~gA!wm z`Ak%D@v(ESPki9Oow8CyhxafF6?9p6dadl=EIp^cElnHTM+HiOGewa5kw+ty7xUad8opE9x=}egJkm!G=>zaeI5c)WM zOMJ0O+Ze|O;M8BlKZfQVE#<(^N?$yzLT7-!H-M<3-0>f+&qSLS{eEX2dB9A1)3O?; zGPVR+<*fAGN<-f&Ej@sS>1KN`+sOS zHdFcMG9CBr-|?xN%)=#v7i*67fGN6gE3nN_QkSYPWlyX)A<+e`ml4aw*9jHp{o2id zU$J`9Xo8T9N2&X=Fit+0>fw)#zSuQi?19#*2buP$Or9MkHUA8$WS7P=uq;pLHDuEI zgS(-0I&W&wmJGp2$sQxu%2xrN$yfI}E=>wzz4C@aLNt{Di`3BUV$1YhOhJIug*kpm zTavz2%6Fr)GGhLTOjR8!UIBKimjVpo?}i`=8dER#n>{#nZ8QxgzX(F4r5jU#UG>fH zhw)nNwQmXbL-09wradE?f6U2bhJrrFX3z5$5i1PW441tr87Hz(!?A1Vc6JWmM>F!v zg|;Tic%d&X=O+sgTbn0>{%~|L+{I&wITp=Y>FaohQKl^g@}CFoREgfqUo?fe>SWb; z&e(HmM=-Zv3b?PpGx@WSAxA?TF)}=3A_4OOMfNPk{Q`)0H9x3JI%z;*|T3!H$&S?oWFG@A}oi_NP%YL zhj1OrkK}AM70?9k@wq^zz4cZt0!-`&^UQK`|({PauaY0 zeK&Yk7#IW#UL3&k5+b}K5xtKdv@EacZr{OfHIop&QlZ;$j@ih1#}r_dHswCfB-HN( zHoI^pRlY&dOUN9i4fizPE}fo%6q{par0kd1Yl~WD4gMO+kO#f2L}qSq@KcENq3Rt2 zvnn56j?M#QS5LP`xk+j|86TGx)?Kd%Fw8?{^bAQ}?PzLQ zet%?VL!*jkSuj&4=H`blaC6 zsyI|XOOp?N&tOgQmxb4*HC_j z&oBZ81A8QWtNfO*Ye;lI8DSePpM3gh^QRX=#asg@JG%vL6m+7KNjFU47!xwO*VC}y zI2Kk=-C9`Ra_d(q8B$Df4*yEgZ_qx=xgkqsZ6b#$qokGcch6+vgI4g|)n)*o$ng`s z!;+EPT8YQA=rj_JyUUA&%FeKQNmCb4}>MNGx%W@H1caiDbmv>lC})A`!5ZbEo@ z*Eh1gis*UuT`^+prr*KulQ%;zQruWlI%%w(5wP2xWtB)g0uD+nh1w1{S*a@_~6sP>0sS^8It9NXWc z95)(Qi>5gSmpB;*(Ka0J4MQA2&FOB|b&DYm+bJ3`DZ+;Jr3i+b(gI}oSF<<60l3Du zy$SCQZoq##U_o_q7bobi4}*JEveX*lH)@olL$}p=qfu#UtJ4lBz;`H96(=9^Rv22# zSp6don1{D;Ehgu_e}SKp*kM`YI~WOA-0k4r3MaQEuy7$Vb_f6)MnrPSnY?yy>m?Vh zS7q>+SofFeDYa1McWx=Z37^c;H$d#*BuN!f!NfEJVpur zi7G}cUhe~=@_VKU)8QwCO!VJ~gZ)ZOYQ(*Z3K`?8R{gVc(~6o@$$Xvt#zFn>lzw1! z9i4X{s;@m|sdae%uvj>F+f4wzy=dCsmY2rh3@@m+g}1s zlYkMKO*RiVa#>oEUpT*z?cZ1*l{l{<7db6q@^??fqnf8RZFU6sN9{kudQZgsXa*C*;@+aeJm@G#p=IDy5FMzPwx4-8b_tApGu(A01 zUHpe{8#>vz6Y|wF)kYxT{KX6`vfk7~0-^JD3W*xuh_nC6*OvQ5q`4zKFG}vofw@j0BMl@5Vtis%oNj!Bo z=sz^>m-!suTP_O>Df}s+;Uj&?;Wq@dtr|n*v*@9dlJZ9N;~bXdA0I=cQG7%|<7TM& z$H}Lrj7Lykt>NcmjQm>fy9X90=V`2>4>k?92PYCGzVs0uJlC})N?LOyxU@QAms;%> zd!K{=m>75kw-qOm4|anNZ<>WeB0R`UsrFBnIZ=*s`#O)7p~wI@I@*vcoAjTX5W91- z$rf>Svf$DxUpQRJ$@+0B+?$Ewz$95V$^>)Au)cyiagdsNeV4X&QU61hjmMmdt!nt{ zb`fFj+^4h!|^t4ysqZ*{1cubD~=^W#_;<|K7R$dOwL1 zs*PMZ<@&QzrqF9rre%#(vT5?&`PuLFqB|~vdfsxKbcbJxg(0|gv97jpVG-g};IRahSGp1fkzmorn zRhE4Zf7c>zO@YUMKX?LE(^_$KqMLD~ZHh^aa%CIt`rLK6BIGfpk77yGZI0>!NIlr(yrKiC*H*m|`1vB>K0mtU`4DgX4jQF2`M{;e*#{ z`Wn0aUma%WLUgoH{Z;DkG;_w9wl0^W0R1q>K|j_$6$YC7}9 z=@+XnT+GV2L|iMyP`? zp0yv(b>p}qycPf}9Nu3BZ$|`C9O8N7x{{M+WS?%D0l&`=W<2TNO(_@*zZ$7BL;Ztw zVz8R{r(~!DA!E#R(?vf9HVwLB|DDiTkJL%@?gKCPu-q!`!4}CDJ`8J5xx8O1+(&ab zUJ}3xY+-%A2}75mw{hQ30*|z0kAmZ87Ox0|Q1g$%nf_<>o@-V{|uZjE@BTA|HdK*D= zcD?-C1*3Aqo(o6g@oxty)@fdF9&s@ya;gA_r%LnulwnvUf`K)A^Ca?;faUSnf$IQ7!x{u`l$5&P?ZklP`;$l1EYjSwootD6m+xups`V@ z+ZcFb^HZvVa?3iJgjoXGbDx|z!Qj}oSbmmQZq0r4*m?_kB_^KEVCFIPiQMAM0{1tb*{7Qvr6BU%%GDn#`@j=^|8_Ij)J!^gBRli zxa-Z7qZb*f5u8RY4Q62E@b=_%=WZ`BIK|<&ePdmO{W^jAXV9NVqV@epl?q0JX*lo0 zBE|*5l@m0qHlnF~Bg!qThl54CVOuIrkO!;W8i5pSo_X zA8ZR(hkLtc#av~^N|@(44f1mD)xHSr7Su}oKGUrPND;2w@=UyitSO(>Y}b5?-%(Hc zK_Zo%7_ifnzOP$| z|4UHv+gk`*-`Yn2ha@JE?zs}7N34gI9s=5elqxU0l3}w=mMRg}E4T|$^s+k!tE~#I z%zVRfxQH>umEDL~3@k<8-mUR3pP#!D>7|(Dtr>K8_PG%f)O9y_-MIlL)?s;kmipU z-R4$5vf(slN))J=$vHCY^k*GE5cgg$$AOY7_kNx5lQ2#@>>oi2^<;TNuFguNEJJmo*!0d_eRLAm;0fUmnI^qp{kp9(!7kO^%9EEib^$L!YmXrHtvc zqc0YlM@fbthrK2YdTlvdzx*;VRhKQnmo|>DD#8>$J=|8NZ8I4QWNvA+3LBXTPohEv z>LTMF-%?*urZplBej*ahS3LA<W0Moz7gF7BH{62C zsxLD97QT`|x6kl6&{&D2_6p`{$COaKuo^=9o=tjP8;%HTb0DneO||g1jo!@YsJ`Lf zHsoD)wo>9?xn(BP>h!hA%U7xPEkr)0SVx9r-!M&g_8Q2q$)dE>CeQ(8^ZK)%5*U?( z%}D>jN)>B^Yj}hVGmDbegjc8sVF^bEopp*WkB`F^6iYS09=_;{dl?FBySjx7S5ML} zon*fME&9U9GXsb3*>XEOH|W5uwn7VJ<;mYg^GhD{ue0fmFQj3qFT+j0ra$fIoa^?3 zXSpb@b=kp3;uH0ElIxpLiGDDt{d_ye1x=E>c+*_zKdY!55xHNZuyd`|jvssfP+aE# zDmbblxYTd^x`#)=@UmsWe_q7B%8BMJ3;^x5y@K%ZAUI-gQQ% z++Ehz3)6YrG|%L75x>)FGToNyYPceS0BAxCyG4tQ1}oawu3&|hY>WOm!2G%YBnH|F~SHZEB+ zh@DU-iuXh#c@d9QPOFN z8OGWTAlH6;m8}y1BV1whKxb37y}qZA~o-CCly#XwN#>5wB7X?COg|2t=8O ztNfc&{4kpn0onEKjDJaj∓{!8>i2`on2a@((@)`zTog)=qN!gjLM8II`B)c6~m_ zCIkFfgGuIlPI&>vZ__Zrk4MJ3I$L~P7r&FD;IMR1THBW~3c#WD& z=wb9OQHr+N-UNS_j0WA_g6R@h#HvPXq&!zv2lF&z3t*02dd?VBHFE6>+5tmbnkl(* z{{5@6Qh|YOVS$z)Cpt&2(gLji(3}`*@%6bUAa3yy@XpE2RQ^U^#5m){ZH}-gA^Ge-Sp(zM)bm zf9%DCHr%@VSC97``^?0*a%1*~06D354t7Dwa&h{+X0gK}PiS}L{Z6JUqS{AGcBp6UtfNO+UW*o$Dr7VJ1?N22+CE_q$LTZukd(Y`xOOV#h9KqqPD*Yh96C48A5P5 z3WrAV-8St+Np}O;kY!UVPPRhL9No{y#)PT0G`Y)Fz9(6c^0cXL%+GHjI7plH|IqB4XK6K;zRL|lU4R0SM}AB0Dn@lI z?CwiLTWp-#*vE+UB~|cxET2N&X)~K?w=k_kFxw9VQay2CY0J6D?$9Wsu6jHV-8c9~ znz4)_JK?3}DE+z!_8=lNq!Q>Ge5lTkhgX3MEs_5AZbIq-Q_m0SLWEe{(^M!!RI4|e zHKN>@^ah>hHd2PWWWBc+wxdq$R5w7R-ocD*KWstyoD(aRHvgh=Za~7%NR|_Xyet81 zuv5$9avFWyuB0TYxrH&Rt`~ghnSef8ji~wC@#9rSWZ_CW(<TRpEB(;yY&xtkt0R4SL)-T}Z$MbwO@J^`T4Z#>kkLZ9O_sWq80QS7|@W5M;Tmeb^ zUeM-US1D|1@}qt)M7Z09T*(E!XsFZ5GX8g(2fV{#RzvHjHd&0ItW$%TtM;W~7t8E_ zXiZL@6(7>ATM&z4ha`GqQ6muJtl0O<(~o2|AWB`1F<124EkWX%8SnbceF@CsI8hD~ z4o08sZ?0JW_@t?X^(02ciWuzv^ErzMLtk5k2bxa$c*Vg#lW>@rOMkMs#ka?z)kar< zmi#FoPLpg!QK2w}RZVNPAT0EeiZizPyF@JK`?j6-$$N}hbjZI<9>X4`d#;jd2{g=ry6@# zhp&DDB63;pL@OW%E=fxTgB*%rluE{{L5hqu6b(Ex#dztp<1Tkn$lzhLeLUMJ7h{^4 zAbF_QL~KinASv?g7;?oi3u4hrNd?MN*U%@h-=$TVtlUNWmZmV5q1!`f@au;k59{wD ztN|I3)_Aq-?l|R;uWX+m%fMR+J)WaTPDv;FqY5gmeZyv6qdU*6$P#PwxNbGz<@Tb!nQ7GJcH9_QBm-40Kl9X!X*a4T>a7%_Nf z^~GJ-W8VONrdF=D&Bk-0u*bR^DJt!O{4nL``BBig$Zn7zLmV2JqKQL z&JY)t-by6%`6hsglZS^!u)8bX_PY%8s47rxZZ)HQ`HYmI5apz&Iw5!}86jUgF zkn*QCi;pX^MOZbrISwa_+AMIuIqi@e)O-(5Pr$5z&C{c=mh$CX^QFlzNjI}Nk%u0& zByu&>GM(3q`py-QEeKFTH+Gy92aY3?MDx5MDT?P5W21G;VVfl5pHO@VDZNw|k0QA4 z+-b2D87a;D!vHK!^VG_hvD#REK;1-nYd8LS97M!Wh6=$yZR>t!T~48<2XA}n#EUk5 z|2_<87q0-ldYvnor?`H-A{(&7_MSV^M{NIlMvy0|lwQStd{2`n74Br|@&LtfBqb=9 zK7H#_ku{~Bsbk+nC?JNAm6-J`lbm#;!?t!f;9;RM#M0T=CowYW7;}xl{{yZL^7IlU|z@1DdV66PP0($ia zKHj?iM#-|nx2kXXjO@t)?ECs*>jX6C0X;8vER59N~>W*7S9Tx|E~LUAWHF5 zzFz*_dx_8bWz>%lZZ{K@drRPa+1-Z`VP=wEjICMxfLQ;UYT;ncyi;t2T8?v^nB!`% z&+s$8Xfm2#UgC3m{G_x8U{B>+jItHF=&eA@b^9%yw(n`0ILN(sSJek4^dLoD+Za=$ z6$eD|j6#~98nwT_Zk4apg~;o+_%A22$X|PGvreMupz>jYaj(vHoa8^p17`h)c17G=%7cPxJy&E^Xm8m1 zqIAFRE7Y3G-K1c;9oyY8!Rz{tCexV!)rR%sY(`gDN}8d4^}RmsFJO^ooF}Y!)rjhk z-3)E-QXh{ZM5^NO+RRu6f(|`of2_J(Pldm!7_KqkO-moNH0NWrYzymAq8e+xhS2rj z&$i{3(D<+`42xe+uSnaM7$^WbtPL_Mgdp$XD^Oz#L4nFlQ?ji5q8tSldS&CO8DGvd z(%=Gj-+$f_SBE;l9;)Q+67`a<8+TA`g3`Sc+&Oti^`D6|oHuFV~-U+jZT{MAM-89;4Q?94**+%ZJh1+s(O)Bjb2<*nN5mi9!r7FiLpGTzxGhNZoem{>>T`zZ!Mp zblc$$Z5;HMK1$`&u>R{cRkc+DwN6GE3=u^Vj;-Y)Jz!7GTJ!;$$Lz zkj9OqCDB^nY}(1OUo+Wm2e1GB@9y#Y;41lNLAZremzCV}3xi9;5pw5jJ?TXWJASs!2kej?BC6-ObXeZoNw@Mbb^=m8I%W*uzJ#62KdK{%qyPJ` zl?uyYo>aWn#m}B5-frRJ96Ya0>Hh)wKnB0u^q27NqNC=qn18@(7SV3U-nEc_z-!y| z>y1|TO@i?*ZKO$%H{HM#*>ClG1Zghqt*423CO5$Y>59qD#)>!C@aEEIUa&a-0D#v; zu2_vv!CHNbK&EBO@5eEq#dLHelKW7!^9160Vv~XVmpfS0x)V;!#eK;_=Zbq#gxu= zAGk^VE77$JcsxO)OJwpCXCEsL#=I$AqLEz*45uL1QKfj|>d6XC5Rm@>do|Hcs<$kL zV(Pke&XK6gWpKlC9vE~Z*1Io-x_{fXn~yvnB_Sb*@1EbKay}u}f3~ld=V4M82DUsK z<7xaU_QMQH(p*QdG5!PVSWef_Nb7zh+2#XzZyQBr8w-1MJO;;B1M;sdA+pSRXRU1N zJ|omMtAN+@nG_5n{{RY)VcN3ikd-hf0dtalO-cq*f4!POd8bOda1T*T$IDa%p!!ne zHzafGQK=ZgsbdPx6r5+hUCBD11^9jSjYBd<%@_yE9g9~?zX~R4CR>}N63BCu=VLg68BFRAye}^s6Afd9Gbuu`)xRd~|@RJVH0QnIn4eM&zOyfJmE*`~2-rbzbl^sgz0O}LYV zQ;}a+T+gIi=&x?m!l|_Uq}PqfHj{64A>SAqk3n2EaaKoCYHsb=#qjoxBHOm%hdnb> zc*}9M@Vn87jskIo=8MG^Gk05y7Um(e?Edfy;5bBB_XM@gF6IGwD@T z_9h2<(IJcs52a>xH4J2nb0QCw*nTwgaV(LQcVZ7*)p9u$^aN8)>rM8Tb?OOM+`BoOhfDX!E52kM{QS##M+{h{2w+eL&Oop@WU#< z?-we|>FH8w_UjI#b2d-Qzyhi2U%Zi+3UIWcayh-?rM{gk*5Xw>6O2~;ZFOg2pi0~h zfa0rY4dx@jdV`T&0JZaAXx|yf6zya?-IP2rsIAwDAcFod@G-+@n)SVQT_(@NQJXYW zA^u&fiqr7_0ET=o;q3+uLr}7S&QKP2P*~MZ8fkI(hTh6e2t@YSIYcRZ#iI6{k>}&#M;C*f@lAd|`6~Gn z2u2j2E1q-jPE*AsqN7H|09#IYrb@V^jky)AZE0yZ0(N!J_c=9NqD%f5o*@y9;XivO zMHy<-@fE<1)-!2!F&V=niZnes_d(R7p4xPl8Adjq4Rm_U^J*G3#1q;to41Aw$gJ;X zX>AWNJIE(BPDW*unzkL6gWIiDloI(Ot&&E~kaJNh4wUu>H5M?`fRNKpdkR%F?$iMh z>rO+>BMxW-1W*Lyr7}UDY07w|KX()`L>%U&h%+_-r^gzzsOg&DgF;cH$$TX_RVQg4 zl;alm6r`5tJ1AfYBNWKa0H?^s^O|A-wuQJ1BQ4E(7l9&qtRzU0noax#?_2;}wS5)v zrpr&zGzi18{{U@XVUbvJih5$FsYhwyiwl^nR9quPv}A)`fAKF-xt$^X;afP^kr?X7 z)YqGT~6Wq4PKPb*{ zD%P9hYnQrIcP$w>=BR3VdOnjBOs*!3edg$Xm5-;`OtXe%APQ>M9;u$uWis7F8F>U! zU+GuZkwXTSRY`JLSm0;YvNRdmJ9JF)E2NZtqS=%jZd_zm4tm(^QRY`TKNfga%TJ9X z@)G9iR{izD^i$|7&!f=pB)DbT0wMdxx;;Z$ys@~r)25NQf17uDpGdiaNebK?#N0I` zPcGIQEW+8@o{A&|PxA}57 z!Kzo}yE!p2TO+EL#Wiaa13NBy5!$e0@I}?N*pyDK8XWP|k4otD_TY2PZ6MytqX4%R z$m>pDmwd$Xs~8RV@myw; zr?tP2?a5(fevc z6re2F@(J(EN)|1O7mjhwL}vLpnNp^c(DVNQ5O_;d@Z3?$d?NDa&chpL z`d2qzv++ttpGeb{RI%LL0ramYDnY^PS59tmyC!kvb{0HuUEg;+*0j*UDM1W-SB$k| z)!R8_W>`afxy~y{tFl|NEIyng87_Yc^`C~GD>l}4dPbLOBa${UT}*ngkl7XJ_mf9+Wh{305$|QdJB4NGNezvJ zmluqPIud;X>bYpl7S{}`<;@{JxvH|QoyhsXY|${b)F-*}VJc&Cx!a8WDwxye zQ}UY9HN1&%FToU1HW%*pBeiuJU4gi>-nTPEcMpl&#Q5V0pshir#yOBN=&7eyl4l@5I_2u0_tJZn)|T zR0#7Ai&pbr_;O3bZh#gWSx$K8+lu+#Ktm8~>VFe>sueKr7*8a<514nF3kKM7rsjO`6?SW}9amgpOE0mFqZQ>-tLS@#97?<~stNzOJYMk;$ z!)$Q{UsKf7Wyj-Dvu2xUlr;6Clf5lLLerFPU^9%GW@$m@fI11GirUxg;F1I6WOW9m z(=_k4U9Qo>a(Jp+ZZ71viZm|nFo1P6?miCiis-sDM)DLFa{SIn$sW~B=rohu;8@7m<044sxcIc&hNn1=xmB(LmQ4%S3{z@7OGow1(@gNAB_m# zjaa_75ge%Et0y(i(@w^z+_M;#Slv{P2YS!sH#MORznk)aNj{ZQcv=aJZZJncX)%$X zyef1|u4)208o#H-=SYd$aLqXGCTZ$WteSibyRf4NBB^Ns()cQ8k%9$uI>qC@ndIUj z5F-Is2C+0ok}XMOMhe3`ntZ-ZK0~M2*(4cbuDb_Vho6jcdQUqfHFnCc|w%U%NYmp7qv8n16h^m)2%{9Y2 zu`EtWG)Hi(ahgqy8&c}iEvR{?W@E|EB-b5zc=tM&pC=GKb2sMwhsc3Q&53}FW&jw+s~D%{%c<8qp%AJ1@v5*MKYuUw+8ohAX#(yW6C##XDW zZ1ROs$*Tn9Roy1pXSx~nsbh)K_-DaUYC&G|c2z&Zm9?&TTTjw0gWK-hy~(XzA|YjN zA2sDdL|bqG@ruW|n955W4?sIt zdE$4|bsJU@O)AR0GJ~JixCOQ?C{{Z`INWM1$DsqBYCQ8>_g)&3D?@UZF}eYsc&r_y zeQ76gwASHH6qq!$!~+|_BQ>M_nRk0@E$lG<=0Ci93Z)7(lPag)CyMOt#-VB9>#I39 zhDFG5cr-6DoI)CYeQG>l)5QWMMLRbR)!3_vd=)2_L1GR-u09vr^gU~((G$(nF5t<= zI~vxW=2bR1HViTi3ZyPETJNGu<`Ni`?Vh#0;r{@KdcTPzD`=rre)N)XpnYmK)X?X2 zJt|#SRkyW;cRY>81QU1&_4Mmhk0-BvFdAL-%f_k*(TT)eSt}=>ZX;^ zr8cjrv+PI1I>x+BU@*I`w1M(}?{9ECPn^?fOfuke%myl}O>ywo8VZD!F{{WU`b5UE$7!`B{NaR)% zX;%|!?+Qoc!v6ro1sV~m*lx9Vr0T=WzWFELgq3=c?^zRC3m8~H<<#;Ct2)+&1-ji? z2}zX>RFU_GV_6o`T{e_afoDQix@~(Ligs;1%J)*o%#3hx$oHq|SB6O$M+;S=Sm%;Y zw4uOLQh0U`I_bd90HUOMV{<9gx@hdI9~iCZneYYz7s|-uRyWL$aC@UD(*Z_3OB*aaTJau3*LE?Q$E7dAGvG0fYuVKN^&H(^qv2?5t#R z$K|LzX{1eLifJ^PM7C!uA;D04(qGsvk80MB z0=&rFM#O*6fH z8rxRq4-3zAQ6oc@)i;q7Bg)Gakl7AZ=!vaFkk;DMa~06NBM%R_2!nc198 zzK?waD$5J*#&L|&-7Unb=$oU0KlRd%zd=v&2aDv>FJ`gPE}zREERvt%_8!%dAByFj z77OTuTkZ_vJ_}>koFmnar1~Rl+SuAl_6rLr=aq)$cj&)UinkTT;zzXKEI=FuiGER8 zzwnS)-UAdaEuueoV+E@pSk%9N8bI4)Rl?k zeLDRd*QIr^>i*jH?n&Vgf)Z1>cds5^wd(%>1vKlsyH2ayjG6=2WWY3zfyjiQ=F%#dqg^>wlw>P{{WUuKD8q0 zY^O5K5`9lJ`B<)u{l=6NS|Zlbe)P=8(Nq5buT7BJqI{P5KCIMOE?0$$uprf&h5I~; zSm$!?;;Ex^T3Vs`ZO|zDoVGcplCxsIs9XC>h^=jIq)bC_xsAPX*ZJ0eYV}xbtuF1f zJr3q2Xri)fnRw?XAK_kMGOEoYuiXql`cf#{@OF`LtZQoxu?kdWox2M58!rG@7gKw4 zEz5Tq!q*k>whe;n=Nyt0AI`mE2~N>~Nv!23dmBOu*JeM7?eyIf!~XzgvbRXLoMlHf z@xc4~s`_ z6UnPc$&JCJ#W>Pph5}Pbry7j)qypxg4Kp~YAx4%}jYe6H1ps>g0K*Rr8QKkQ&S`4sm#jQ>r^!DMqMw$$!4c)aYxX1HRigF)6a13$2*Q|rV-^=W-^yH zh|$#iM}4WW^JP0v@aD1iR3XhfbyqdPM#O~ikyYMC2Ba;jnkOly8 z!8OVFPfnXy)yz@C?4fc>4DGJV#oDilZnRTpcGAd>*ky6J0{ zy+uA}VQ5LN>BC5LidMm4pO|8*cxu+(_WnUI-gfpho+a^w{vp(U*K@RuyH!`xo#0JU z*2~0gadU1RLLZfm0s7N&ylfFxo!-Y=Bz6`ts`}cO z=%TBYUgilRhf!NzA5{}>VgLu4#m!9|j)MrqeJQnaqU`Q^Hm9n28eZ5ATOF&6`%IEV zUCvw{^?$?b{iAgatg?j?I8x4d6|3WK3%7-}Q3wj~&byGQ;;BY4wSrTW_aM@4-J=DM zPfF04)^JF6+&YTq^sUJ3qmFuJtxr4K%3OA>;{DT}TaCH{aTt8oeCVYK+sN-)4S2Ue zk-@6+AY~x;^r>Ydt?!W;TIn}U_9;t7I)W;F8XvOW0vv$H zy>8v<65O#8#9l7B2C`7Gw5oF}9%a&$b6sA!p+jkCBv{fjpQ$y<;|CO)u_@fr(%@O6 zm~9430EG7Utwy}Ni%yaioRkLxo+^fmV?FXo3|wy@Zau4J^3E7+Jh-IuA3rezicF+& z*ev8=mL{cGWjVxc$F*P6V^gZAWZln9S8B#6qHz8qk;WRr{`&1nW>CD80VJEommUbeJo&z9|%KkWCfSJPcC z^ed&Yk)VlYD5!&k9-_HFht>~!E$ze@E+g`hy$g5!E7My|w$Xe$ExbWh7BiPWsWqIQ zr6r-{-Y@Zs#b$KdOS0)2{`OEXWIy3vZ+7l9cV_$V^B=8Esyf|jZjhhdZyRr~rE%;~G6r|~z6 ztY1NzJ9(ANg8@Z(NVrmPO>Es+!>K{$1`*79FzK4*FCx0LpJdT8uZrp*631J(lTtF>$>tV2ShuA>+S}ONNgQA@t8m`6W={{iF!}N^S3K>k zhO&=&CC(4EcDOw(dHcJTZRdvaQ|GqRGvJZfS52nr)>?2f#VAnRZami_Eg?~nA+Cc+ z`wh*Zf_Y!ePLxSr7HA_B-lt=3`CM>u-Fg~j@ZW8>ILQ@(r`}^wTWIAaF^q%Owl2?> z$+|`>yP@x1bHcS#nQUb$JhIT|yi+%mr^36`mVB7=+upYR9z;%sIboH1g&)qji=7TF zOV7O;g0$@1L64YweJj!YAE(2mcvi$*y2m8585=&FV!A7)ppJ@GxrKS5Y4iBSF7NIY zEz&?5G1mgQjZ*My7uPol#2c{~J#kz9E75N+d_jG6VQ%*je7od9kiBb!D|K$$UI`Qi z$x^1Kl-i0&s*%;UY1&?0$7?;kzUH0=-5$Zntc#o7MrmEHT1R#{V&IZZcE1fvf5K+2 zQ1cUH5%jNEytC7$x0u~p!;{7t0ng)IP^R2&nT(}psq@W_uW@WdXsQ*?7!LI|uMSad*^eOG_R* zGnr*)BdMk-BDxvgzMcsk;FaTM$U9Cr zq3`Zq-^epLh;k5~1y)~E)~WdkXej5C$){(EOwpQR5YeA{GCR;Q>r%##VEv&_{8#oL z_6v{y07lDRLPySP*MDeYN001(>=z&Yjh4KO{p{5(gvi`s%^<0+5U$7U!l0>Y%%nYf z*S>r;xBkoUJX5dBb8{(m$;ZoH7pL0Edwm70V=pU(1M6Q$t+H4_ZDSTTTSOzIGQ{sQ&=<)rhW|WdNMkTBmeqo%Amx z#kf_+)}Ykx7S~Iwj-;BdW+soEZ#4(G zx|UXsEVC2BQ5K+xyIRf%_bbgocV|7E#wpOFj-d4w1yXFz#;WQ7FYfIiQ|eS2Yic)4 zG3J#Xi^Wk5@`^WAb1_*HJg(>Gc5kVsWK?^NA2B$^A(G~r{Hdh!(yQ)t^(|;}LnZ-2bq)47a zh|sb2txJCoUCNu~0yOvQROFWCCOt1|vw3W`(|Kyez0FtcM&%4@j~vv7m`^l-wQXp# zADQu8r-OVsX?3l@T(boEtBtdh3@h3G8R+b7TT+!iRy$OmN^w_Zwv=@v@oX_zX>tTO z2a58E<=n>}wd&syq=z0tum@>4uOW~S)zc@UIii)>GPO9Gt_?~L6$Osn#YFvS`KC1t z5QD8svQ#3J9%&{rM|vV+O+@CO3w+f2h`fN}y{qBIpdQsAXZeTln(FAYcGtF02%Cb+wb;ZQ5#K^gq&ZQP?>Z>{I( zZ2pJj{y^0m{Vz?_7Z+ABN(Vy~!6bf_t#9EyJ5YU=TC0=j@#Ff|-H*cIxHZM#k$n_m zgpggHdH(eIvD~5b7=P6+*g~vfR-FtUuB( z9wOtFjAI{-c-O>3qd{wG~3ml_qwjTyF%LY}nLdWSbdiPPjiYFu&%t2&d6 zkUG<}Z943!xbj=djA7dx1L;M|26iW{bU}Nj6MBsB=7V{zYDOz%V~$SInzyKFa_N_{ z+T0@~v78WVQ^DeETb7;~UN#-n;0_CS61yZ`Ks?t^ z75&AriA*M2Ns|mf&eL42yt`XyJR0pZIkxGo8-mH#8?jYkCUL$eyPd5-N7|tF7_K8F zk35rImyAh}k^xAEHOIuhWT?KTv!3O|?IW5W5BPr3fIn!Q@A+#IJDZD$9TE1ZJBaqD zYf^cdd>f?9kjT-HeFia7X~{fV9o^z!CPEzd>(aAjl6YI*_}fH)_fgz z7OQ;-y=hwKCBiBF(EKV?+L2Bxmgd*PjWF71Y|}&KNYawPj2Z?lZi`gD& zr^HNQ*nH&ouX^}%pp6$oI)um;=HzFe;a=6|eh|4cnjD7a(qP^nROx+SS z=gdI)Vjpz}rYY2yL=atB&+6LQVIs#mlD$US@$G8zX)mOjOiG)M0OlCbWE!i%K5M#y5p+X zmP`$VpJ{`b82sz$?_Q1JJ$5^*18qBbg?a=1+Qt6neCd0yb^8ARnUjw$sI95!kd^xj zFQ01Ta*_9=KBBQ;)851w=Y9RFqY$ zZAIO?lr`3kVctIxTov8VnK8xy9+kD?4RSdkFhJXwo=ELnn=@jjGj`N<8oq^-Aryjg zjgZ*-RYJt~qbH7NbGe#3zYW^`nFZXjJG1hndR5(9Q;s>-Nw*>ZaJb}}+VDhv6u$7? zu+%kkYk0XcgM2ZlX>!X3vX>02@?m+;w=|Pm4p(}fS0vDC7q;ta9c~L@SPV96 zz0^ErwiR0kSJ~hI=?9xS0Ul$L1_5B^II0 z*r>b8$0C{*n5q9SY#7t2&p3EN(nFhfcGL?M|hG0nKz8)bQ&5CR@1e-rfoS z0M-CHw|c*6Xx8YPkfbPbSo8vj`^w{zv$5kB-W|NsVFysRXk=lyD|Z!%n)DBdTKU$m zW}kLYe(o`ziaJ-G^q>c3noktcX)C2{`%=-F$? z^2G2f*MDe#`5){*>=z&Yjh4KfYSAo#+D;MA-R(%VHQ3)U*oUCb2U9}YPk?;0f_SL8 zRy!Yux-{p-mbVka!YEYC3FvW(_Cc2=&m{9$o($4j!@}=#JicVIl||$3j&cX-Tet!` zn9A|E>}O3oTj~)=hmpl?>FvEYdYZzxjJ(0I(zi6R`=Lic^)^so2KTIQ70Ygt>Ed|= zY9-&ejP$M8mvl@$D-&IbCcC!t!g*>szhhZcsqEJ=SGK2=YnlXJE7#<`M_A#Lk1<9E z*A?gbmY;D9^F)x74Y?q&AE~ct*RJ7*QU1=-?Di~d4pea_eXEt!E>`vl?W2TRvC2v1 zxBAiVL@LTo>{7GT^JSLTXx$rwjz~Re>phy0@;}s!)Hn9V(&eK)k56ibH4h+Cahm8- zGqI=}O_B1?HUR8{{Ag>7IM2y!&*)w1J6k=o4eT-8RLx|J&O?0wrR_z^Ml~rA<`|-j z+%G?!M=kZ#-|Hulho)(g+muj1Cj;g+2nCJ+C+Str-%zp?oYwyUgTjq%rEKCT9mYLI zD<&N_@_+Sq?nlzMG|eji07>(2pK67PgD!fC5@WdW5x4LZvxW+9F0euVAiclw>d)}Z z0)2YfPe9qJb!(zDT{^^l*!M_BxySfaz6(2iOKM5R8k5*-bKf+mRs~jXy|kLCJ<|Q2 zOnMrP?}SlG{k9R3YngpTvD3K+9 zHM{f`&)i+DwA-SELG?APb}*^A&{;sR#J)^Qo}|)VY8IC2R!fb~=)Bi2KB;#&+R~7| zf~jXTC`D_zjf+>eDRD7mLIj(G1>k5bm%p<-2d=D5k6hOKQz*6ur*T3B2hp4AnUFgG^{Nv&k>v6_ooGcq>K zPKJ{3D&DzWp-JEl)F(&}pmD%`vLYqv~5_gm52aqXrd!QpgxE{i{!v@2SwTrDSj;PR$ zu5>!=(_HE5utZ;saa;tN8swR5hEMwy-q_pP>QIHA_w4REA6mWt00_0w8MM|>hoNj$ zsiO%(%<{L;tj}9j&-=AKzM87*6et%b00CZ&FNSsG8B=dK_vbY}k>Sh#0JNEJH29Zq z-NFG?%!uVC@ZOv(Et3q}k6hPtYUD;TYc}4?YbY@70eA&OlR4*Rv!J$ zecM>`Q)*(lD8Q*xj2^Yv=vLxI5$;i*&BbHtw-8?HhVCa~T;;O3t=%U17U~%Ajg#C~ zw=+vqiui|bbeBRl3L{_c^fkumkhHh)0oQeU2gJ*p6uS|sZk>lr@U@Zf-on#R@fyW= zjLD8M*w#~acQuKRt?9B@T+8;UV>8S12*B;w)}xaL!^KBUtn!kl*1ey^{{Rnkj}_co zi$Ubg5`|{}0A{?S$F^_b%NVpfhFM)sV|DDclxjuXxkaOgv&8eq6AoE`uHRMBQ|zi0 zafRU56{JNa#;a`<9LUlDwePoH2i2stQlc3S(U2Se0IgJ->PxxgdR_e1Gx<`sKt1c$ zdx2@%{Bq8c%as#&S={(~>SeXKmnwwAB1iVAB}3kE9GMANgx3 z{v-bYeAlPXFojF8;8&6S1i66dSJ7wp=4_Ge&3cBZ^FmWNC#5n^A^3-`Y7)sUyF14b zKX)G~73N{)g@m#C^1Y3Er;KKX`oiK1k>y@n9swVCSD3?%*<28Mt!)*d65P8C@LfW~ z7~R9EsPrvDO*2zyf%)P&Uv;LmDYU#qtM>-5^*N@3@owQHI<83d6+K4vN3%(>^s9YR z)R*#GcvI94;+Nw|i|RAV+e!PC`m>&S(BPr}_vw4V-5 za-n(L8$EIDLSkf-{4CTXy1CZxZefZjUw@UmmL2QMwRI1uO&}bToaVlRYs-CePnjT@ zW%HMA3CAY9e_!yN)|XC%Fzvgj{OPst7}TtxWXgIy+uOC1<)c;V$ax~QbX`*G%4pS; z#!mx1DAPPKd2tj$Lj*y~6{&RJ4PQd<8IK`PUuu~pEiG~3t3q;VGaJO$R~I&-76Tq< z092j_)4tmB2i!Jt3h~V){%mjAUNkJZAy{XM>%JUVMA`#fmQg2|Mmw6)j1!Vw4Qo%9 z_B$;i5fZ2h0q>6W=l=i@8&-XKbhZ!V4wy{-wduNi2XMt}dCLsmT&EF^yjGPJs$Ch= zYnM~y)|TQakDBe1)Ef0a2YA;+h8Uo`wpb@#1d0W7zB%yDsXvLXZfvbg&`RGYmMe_% z?ksOacQG_=8)21)N|iSftfjH^aJbX1Q_QxSJRW#8PWepsvp{6r9SG}QC;hNeUDoq7 zXZ@3kY4L`y1cWWoU^-;h(8R~8CRCc$7`Ll5H&VeNIgQA{sH3)?XxZb5omcM=I#kzp zhW6%5iE-vR$s^XTX}a9nVmd+wACvD`d#(8wszy0h2Lv4Y8c_^~wO_x~4g9;zXouBD zTB@>b1b(!dR|~QhiPUnYpB0_V%epvSho}`Y&T)gyU9#3DvW=sRoD-T!AiF&;_J{uf zl7V6W0ARTP0Q79NM13KGl3t^<&%%C*Rxpgda z_)=+B=6y*)1gJO&-n95`B>v5g1Z7rL$IV(E4vp@-U14nTua+bXNc0uTp04LZS6Uv& zcN|t$78+b&4<mHZ7PJboQ=K#`2}QLnEMK0V6)u(Ai6y_Frn|ej}TEcr7lj zSIkI@A1gdRPPLV2^FqW)T*=VyqkCJs*%y7vrr1I1O?mz0)bP&2RoN>sY#y|?o-LB% z_D?PpiY9H#6k`E>t20y77WQv2wj6fHdd`#_TfW4_JJ^Fzjtxhb0VnL(th1tApxz{p zXz#kNBDPG7z~+eZ7%3#@pgk+8J0jc&t>Kwb?#YC{<%{$qpb#%d#K zLxIt;-?c@9mR$`EG81zkMndX%aoCFL?X-JH6d2(}`~`8=Li2e@V{_Z&HNRpIU{EmSjOaMAjcWvg`O%TV-gkqwx)9&}|xV1m$u-r@ zbYp8CJ%78(wcXT5inHO%hPKyjq*0T*7^|Ku)0hLPipsaspN;uZ>08$kN~179SmQfKYMKBhw~pswBTA}NC7^R9bdwzvBbB&Y{7ZPYnExFx|GvK*^_V`DWXYG{{Vc}l;)Du=d7;uIU5~IQ~N?i9n|9|vzqlSD(c``N9MV9$5UB)PlBPd zy7QW2COQHuc1sB1UKoHX+l$!ACbumMoUT}6u=Q)z1mN>dp5-G=t2RiiOQtskn3||X zTuqtt-YyAdthuf?pO&S*fP;{DtnAlhZDR&Z7}JAPIiqpSStXG5BFR+)^HM8PY-qA1 z=BAELYd_2arHW2^)HYHt$~U0TBfWAO?$~6N!tss)uD&cf?KREoQ07Mka;Bo?Sn+*H zNWXKAr!_u_VA__e4X|Zh$EodCJWXo0_oD|n{VG2X!tJP|MsP)9)455Wu^)kNE$xsb zUUK8CZJ*Ms%i)P*`|(_-)-jLgPPVz$L}0)jly_RXo+H=oMhTpIF(g!b#pr5k$qk>v z)rs5|4YhN|PAkkTp=qxcHRaoYJ!{Z4yG>Wc8bz(!-oTK?2xO4$0P&Mt&W~$lrfbV3 z^QGJfK4B5a&*NJyB4qWDK8t-9-nD<{ROPmn2@MmE_zikAdY^^lPu<*IT9fG5AI_tY zSx=v{YvlSYqj&Q)k2_nVPFtRL6ePbwKDDQ!p#Ip5u*gxG>8H}}JhD<-3y*!E8t0v5 z(}mT{ktBOp8AFj;MoE(6V}kMJ(YWyhhdEKuRX+_)J-y+U-dBm#edRc;?Na33>Jmy@ zIW^dL1535IFLQSxRaW6jWH;8blULNVj;71S`t7V>t?Ljww>7|c>sYwfwRO0-kz9qx z$}4xoNvolf3rjfK5^^3f+PI1AH4Wdrwvc`m7LMr^c53)0TWfuJB)C^NQ@3Hk2ECfs z#I}qT2Bm)q^kKMvI`GXIB)8QeW=2pCJR0;H8Eu1?mB+BAl4vf`+h1z-w%T0UrTzSu zb`YJWBOcQ1dJewSnh(SaEmK7DEu$8&SgDd-gN$|q^{PJ*?clWdf1wyrN=q*G2cr@N za2f%U-@@AL$`nR+q4m!-osoKx>wYV|)OGo8Qdd|J%!wI4D*hFfuIpEG>Wg)dBxL

)tm>3RNSFYX->a05A zzEt>kuKxgG_>vo!Vk1Wz@%`TQ^sbk#S!+?dTFN9;G^Bw$oG-*nSsD*LU=kJo0G62P z?Osj~&yg8c&tqRf_=4ST>}Q5WA~sXdXPWRk=$g*{SRWvxV+2-|-PMSi*5!mHmkbFW z;<2WW&A*xgI1$#}yY4~FQ@n^J*PaJe!89}pk$7%*x3^uShl)-J?Oxnx7n*gw$r%?B zfal)4HpKZibEx~QLkjiJ3QzX$41JuOvJwZkrBqioi*jbPJm<#xTczBW8*;Ly$3N|l z@y&VGF<#^1Yi~B%M3iAe0|O?$i}*_si@}P`z+;m_20`Nc7q(#9eX6K( zLxfys-m^5H3*PHG2)~m4coaJ>(n|r1S7qW0k$DWU*vP_s@Wc65CXUyzq`EevsI81j ztty?l+M+X?Y_FJ@HI$S1vfgQAXf3B{4p`*#`d3W_)~N^w zYzIh2W9L$gBtW z)%4ZxZ`*VlWy6>IyRb9**B|0-Sa^d_w6%*ZJbWom575`9XmGu-v5G~*MgZ$ena#Gs z$|}g`BG#|}0JIIfRst)Cn+@hRHkymX-V?RfZlbunp4u-rnO%AL7X0z~?qdGR%5fxMJYMzuzX^|;Bh31;rPjYKVKib4sDe-wc-9Ev0g*)4W z-=%bREL&5}LF11ON#YnSRybNy89CeAII5SzDEvo&{f*Y==a7OwiG3-L60;mJ#|nE_v%%mydww8|W8p%JtLfA``wYAd-M@1i)pFX0WG(WBm9a(OEl z$gYYngq|CWW7zw5(fFjX;&l`e4y(Y{kue%&7hy=+^%L zEv$_RJutNt-w&H7%De~t<*uzS@x{bUgm=YH_Q?9KSRUf3RFWaWEZ`;Z9;Sr(azfvU z{{RZ7FNAf-hx(Oa{>xXUCG0OI!64u0MNL)hq%UU^$2Fe_#e1Yey0l9L(3I?u{q3T( zyeHw^J4cDq84w8v%>Bs!01EUCTSTzdqgB0&e%^!}5nSZo4r~y!+Fo43#B|(9{&j`r z&2yzn9I0DUo}o3hn&R1p@!hM4)-`LXW0TL6NO}w&1#}vxfwefpymzKxIl*I-T%?)} z;+FHGl|bu}l4`ul=88ubcGAfy3JE8sDe-uI>gv)-+!YEl(z>7QRfO-`x7dzrLOn&T zty&Xqj0ndaYSqkTB+n(jvujy7A1UdYsB5V41)GVCvLR61tzxx@m0;IWoM(hutCyS01iR%zE&xur6(qgQxN5(ZnO$gGAV%*3TlivsR*W! zcX3D!i#c8x_bINkM~X5&(LKh`Y1Dh?!FLS$ zS41Y{VK^v_y^+}tKmxL=<=nD16pRX`U2)jwV}Yf~%`h5TtB8JZX+v{OBcY@x6%CY- ztm8EJo1V2&NBOB^ljapSxXrtFBvUmA(e2xC%|~d&_oUR_;A^w2v(Db*ifw z6?q1!cve7{4hsClb*)bmJ4mmRY2E?w<=(R;#l^a*Shy-q4QICC_1NgGHCvl01Hmhy z9)`O8JI9gi38+WrdYz!xJ3oc>C^Aj7pMcG0oi^=c2x4GK#~{?T84Wgz{{RwvTWHo+ zw(;1XvP=m_Q@5To$K_n7g3X4Zt*xg0st#B)V|erio$)Zm{i$oGz&=|@0v~qoTs5e; zw3taB%EGheYeZF*^*)lHz`9K63{y)P^xd3(wT*S~{{RG-Z^<9+EqSJq;%#f~T;5I- zZzFldbTN3JR%tMAc4_%7mOz^$Pxh152oaX|k zpTzpo0z~1*t}|KQFVyZP@VpXCgM5H)#atY^ktDhvW1O6t^v?%)p65Z)O3YcK3f`8|#15+Ib!WYY! ze1LsW?W;0;Jkn(Ql1iVBYLwp&Ax>evV13vB0IyHo=)+#c4~tMkZQyMhI8I`YP!#pz zyyDAhd^y#w2LfA_QJ==U?~2y%Vc;zq(VuxqvAfr$aL*i*=^h{O>`Ve&A3kLRl1?%F zjY>sdp^f4Qg@%WvX~1Gb7n>)pTmJynST37mVeMO9Ew~q6D!Eku09O-iPhdJ$UBcoo z!nKaZa@?h+KFvsT!5B5`Uk?*m^W~D(bb`%))=1|VuN{H>tDye%-=7`LZQ<0?Z0#dY zml90FCBF(!YXaQ%y-NB`L&INgwU9^$3{=;Yc(%lT=TMDsB^Y8(Ni}QXuZ>dKCD1S2 zc^6IlrYG*X?N?UW^#EIIi}2E6)MrSfBy$IO249jkM~m(2#E z6;BK~6jRbja@g&4yKAS`qYzD-X?d0==vTf!I`glHS`D6$C9L{*KWCH?^dleRUY)6s z=?tyd^YlGylJWFWUHD{(mmgm+r|I)t z-A0~M>@D*)I2H8AfGw4LJrrR^%mI&T^F0e)w$wZy919$4c&)Zl5DbCP{{TAme*)wC z7BwA4&+AWN7H_n>NTy~)+D|+l_2)X}%!%TGadGo`XK7NrZLdt!C1;$HxE-sI*K|nr z9YPZC<%y7{{{Tkq{b|W*aiYb%(j*~bw`Pra8&68R4xo|qBoit1#Zd6g(UVU|-M@Jm zDE|OaT^u8E+l=O}GP<~$)P1$}yQpL4l^u5um0tEsJBZyNK=P1C1CQ3aSS+t$w&9e< zN8R5K=XuY}a%)NwR^f6v%~TUhQd>l~Uu(3|ukP&Z1?2Nx6~;SOQ(t(G zNYbXgn?a6E)Lar#$5d8;myD26Fd&Ucm0D!#9(+v$1(>WewJR=_d;02WPh%5%M- zXya{J=i0x7wSN#<+q^b0#U!t}gK@VtxZVYq!@6ahvt5Z;uttqX1M#c2ej&QmG|6PP zjntLIZ_tBAnR(+|i*`$$F531qJQsWu@6xZ^R$lbup^Ahq?YR3*4#n@|xQ$gC9G4Zq zd`0lhw6{WdV+*t$}ql@$KdHo2=10VaLq9Mk{E_ zGhD;FS{`b2s57)W=hQVd^fR||zaO1fmQvC)FUm>I1y3B3vS3K#+!{?HlVU!Hdu<9% z(6Np~glv84?leD!Z+WLB)~hpHG`z-`?ONUj)ikTS-6r2sU|P|>8<0p`4t0g1a?2AD9uN7YZ$9toY(CPGw}k${=sqo0O;9kfY5B#=2ur%IT%t0 zLtcaXLjM5B{{UhC0ARTP0Q79N2g0kpGSVC4s4n&;OlqmQv(HKzylg~WrE-$#cUJ!Z zXDb}8yRaaYeq8oG_1XAN&smOZhq<;3Zwzb;V`v%lABA*ju3z0uBC^q>le)Fig#+Kc zbk4e)cvfe>5&fahD;VKLbNe`>e`zP7f)p}?-k%g@r1z#n^B}8Mqx~Qb)P`!!VfLK{ zSd3`ud0_o3mw~oGK(5m9r?=A&lN!8b4)RVhTsGn&Gzn@+rE_}i>BV-sYKCUW2a4l# zQJG{u_1S4(>lk}gkjt{$M=Wg{DcVL34P;zeUs?rAR#%90-Z@&`Bi@r7aA*FIUmxt<+XNf?M4~TG^)Zi3p#>*cGD4&1?gwYa+u(r ze>$-DG03CJXv3bHg=iaVc@*qWp!Z%YK3xq960)-%+q-@=fI)q7HmuOZ%bb1h{M+ToD{{UL5NbUw|S#0MjJgLfJfsjWeVnt)PXS1(AonJ}gbtPDP)@96E zglr5%$UEfIqPUhd{p?}Aa5Ej54eZAR>Em{-9KU zVL~ueAFW&8$6AfxR}d?TjYv6&dAJGUTKj=AGa zKTkrdAa@;?n)B=L5#RWNVv6PqVbzJp6>8#kF}z(A6Tt9Vkp5pSC3%(8&v2_7u_&AY zUG~3u65IWX%l%mU@7;}YZlDudBuSEO$n8uq5n82GiiM+6*aDq|XKXZWK9!^7EjP(p zEHZ#oo|U=bdvudrfXv$v9OtD~NLMv-(L&CpbDZ|0lrDWQ;rT5lwvIJukt;^a za%ne_j%`UEtAkB|n@QG3jgYC&BNasI(^;R~PatnaY*aFMgH)M;mSG?r6~^IHW>RR? z)b+2lN<68Xk3)*dxVU*nUCSSHN;Mc|Kj{__`fcz%w!uD?KxL1OzISd-OjT5)kkbDEI4 zq;N#6I5o~THym`W4N2LjAeztbT~Ss?I{{7<+|q+hh;ZqOYYr*mk;fIAWurb*^BPFs z1ac|2TpFY#!&Yv5rN$h`OJy!Uh*h0tMSBJLhHDbeQIS{YB1>|o2NlmOTfWCcaH*}% z2VO;GVZT30?7j$u`yQkR-wreD-norpHRplFX?!}?Z9G3E?4}l#(-PWDADJ@OSKv7)ebFm3#Mmk{sYi8 zd5qUKL(M&1P6@0@J{0JO{Y%Lo;hOGJ_R?-+og3)YD#;N)D+5z@hoR>3{4G=;==a$6 zWW{D&{5sX)&z579f3j=WNncS(#xtDNbIA*p9vgGvyZ?Ii$p$0E7k9?d*fdUdhF zA&)GW!OeXiCe|1{*M$6Njlba_w2Z^FBs^oiTxWd=aXh};XWA6V4$yhUcF;?2J*bv4 z;Cc~UXbTz1MiqYBRhkIeC}biw>IkY#OzI$z_N`l6+nA1Ho<;gGsXQ~G+v?g(cN1PL zQnCAnxBmd)2x2o0;1EH(DlnYmJbonsoapTX?CH$ zZZnR6ezaMQrh6&H?msN>b|?8NR8U=~F|$pf~1D=nRor%SWU+DR_2 zrk@0sVo9kV=8dz|3b7^mNSX^+)(D&L0oA^ysdZrD&vqtplEw zMsdpuSzl}spgdzHlF$Z;-EJ00*oIS;BeAcc{0*rcW5aU@)e?4$wafni7Vs;;ENslG z(zo3mJ?paYwy&t@xA3-~Ay=2=Wjux+;-$s4a+9+l)?vAZ@#eUUp^%mt@9$Khx<7P* z^{#`)k9BjY%5E-%v%>yglotdKQCxv+AL&$`(KgiA`#@kpInNa)oh)`YsX0Hp?~c`u zCF9bbF81&aYLl=t)TGqmys-@wfyUoj$JQnndV9iz5Jn^w#@k;QlZgtpdm5HqRpec) z*aMTdjmO@)y{c%++MLDQ3k!Ut<30OTa58=CX8H}ttl$mY@!G2CR;_WW!4d%9I6pAs z9`(mK*_%b(9mj^Og|3ShczJ$vUgzN0VvoZyxNI>iyW6dC7WbYWzKJ5bu{N>*2bl*? z%Zl_pF4h6z({E>SxDpDM`Eyuv_eJuS+Iu^_9F&m@r*e5U$!o6>vuOV8-+4LcM|$4y z=8CFXE8BJ0qjxyLuRHja;(Ke0{{Za0Art|WmPH)5x6-tXt|qKS;H_LwsNLJi9!$-- zKE3PGY^{uwq6A<`uNu;kwAj?6v{UsZPa?Xl7hS)K`V;o)S$7PA#;s_H-1d9NHr6cY zxd^$%c#Zw7uBYP&F7Ia936~1l;Mb|#>J2Pu4311Pt{95qJVoI4Xw9{@)WHt2H~7@X z%yLZFx`;yIlc4E}&1=w?11FMdwdmBWw3P{VuPg%u512i2U+|FX5?!EljQre!wX|fK zGilt|A8hd*FI=-%M0cun6wwPhC8?|CYJMmC4j(R zZfitqFKU5qV|b5DaaBfi){U6Xe9uG9HI!&?3O%8>kab?hslz#l@H1YZnoYNjtgpyp z_KQ^w8TIS_73SC4RmF-4SxJboP<;sPM3+^uq+Q7t=@^1e*7d1w^*hUn70j6+V<6)- zmmkh?-k!?Qu~IptQR~07C;X50ANC86{{TkITmJw8?GWj@g~KA2-k=gW9cu&jg#Q4N zfnomuV7UJP^lY`%{0vY$JACXw0#V87-lkg>9nV+@r>+fH7Uag!upV14O3S;}H2W0Z z^Q+0(7o65rw~r#2kYWptF`h+cifOWpwykM*Bup{1@kPefiTz{snR-rGz*NtF9$mY`bVkdP?` z2;k@GNw+7HK)~XoV5fosr#{%hW^4gm&bQ)!u|(qCH!$yt0Ns{YE!QqWk4neA@h+KU z4rD-m&3Mmhc~x0YikXa$LI}&8~b@)2xw% zM9BS>dRLcgpAt1~U6}-?-W>3ewm&-J=D3n(EbL81LWgl(v(asDM7Bue?j&TAX)*&4x2ybM&=Gs9|xf~)CTBnDooBO_c^#1n5ONi z%8vfELLq0xIwD!Qs7VE?SlbxI2nk!IGM?1RU_C1CoHM9GF`t-KDErpuhHQ=Iu@a7e z=9AcH_MZ)D8f2Eq_KR5E$o}ZAAL8w;y_L%fA@n0Z*16k% z9_sLMAxVe(Lz>vN_=>H8b7n`n9M(Lw7RHy0DO`LzyUI|+_g7IcJ-1QqCbt(XUdM|;8%HMBQlxewyL}CoJvfqa zT}|ePrfLdV;EcKcCB|zG=R{jb40iG`Uqf2PqLzfXv|-*jdDw%+Q$1?L@tB;eb*ksB zbghhnfk{i8icApYG%+|Lg{Ke*tl4PHx>9pljtFYrnk~TAWcI`k1!XmKX%}WT)=V(w zuSs$eGIw^(WF65Q)^@R~rKOWQWVc~llBFx$>y232#Cq-2AnuMqgUKG1mEmntTWw-h zP>!;Wj4~^nz12!TnVC3&D=MoZA~cdbJdqN$EnYBvvcIejsD2RLAtqk zJX;)+D#>ocamGQ-bYmq!!NKRHZkl@-w8_$UjQozueT`j*TGVavY4~Tao0(&@|dmMjC;lmnS)cZKTg>$reL_ZKT=fIlMR6eWC zVrzCj5%E>u+O99|pr05F$0PaIBc5q@Qua%r{i8Nd!yOLm{yp{7e`ltjgmTC8TFq-P z?el;2t0N#)h;<9AW0-CwPxn}gJ)7vo&Wlob3tzJzC7578xM~Hr*L2qe!*dy4z?@c% zzl?ReYz4h$9*@mx!|@+c0155kly~j@DcW6v@g%bGylRS<)^f(cV6u$X&A*CtdxN!Z zb}{V36_~y@(k{5Z(a`6q3Joc_@J09EwlJ~u82(k8kCAF#Wj~1X#V>^8x{gJVTiqj_ zr?)l7d_;l?=Csvq#$&mRF>Lj&yI;}tDSRWVCXsT_BSXK-Cp|O%HNr)E7OCJ!U~~KH za8*arqh?j^jyg^OK9vmLAV$Y-YQ)jElqA}AXBo#CsO3=6jQUonhEkd#Btd}ylao+= zo>p`i$UFinP)i+|<27>Guw1I{2si?juES$vLDLse)$Me_Czh;nyVAWr!@-vDc#hV8 zIDMWH3nt!puQl+C%ND#Cun(>Kf!4eY4EcI^bj}ertF}EqlgowXw~3 zpTb(b*NASlJ5MZzkh)>8Yt6MO%vywq@OeZWI)j`ydhV|NBV7Q(2!TDB&r0*_=X=YG zi~X-I(d8Lu3-@hzIjJjHd#wrjAn@urMz zmOr)ZIOxKm!GvUv%HLFL8wO^Yc4>(LM?gP1%kZax?639N^$Fk1Sov6+sVBW++jxS~ z;#7oR#v;!dJab)4HrkDhV^Y(nicq6DjmB|VwBpO%8U7^jgIbrku?9%T-i(e`xr;;P zT&{AZO?KBFF&7qwcTJZnaKTRp@T_U$)3uns(`xG-(B?!uX0=R<8|PCmYah+fS5Kv0 z#jELt2f>~p-?*PaNHA*WYMkv>vJ}tGn=R--<79=C@3FN(Mms*O%&AO!{@(!78${oN_BJ zrJZh(;{}8SxDpft18ZQ9rF7OFEuX|!mewL4GS<^AhdC!W{{RZ|B#H7dn(DkeboOFN z;w2_E~~+ld=Kqd!d6T6nthCFVeX z3eL3gh28z_#CNf8hq3{XD<_uWW2_sCer8H@>XFfSd+Z(=YpaWGRw)GFZCOU#b^7|# zQo`F>v@+eSb1br%2#vl_eXEMsqQH0?O}5L-uj9!rD1ZKnkfrEwp$ zI0xblfbad6AO4M&usjRmYh4=NST0MUcF1)#>$vkHWbAuX{u9*G{yna(4Im z>N8I_y7ihBXc-4{x0>$%0JmpBm$YI%ROXa;j?MmAQJ2&*(&h3WG3NgO@RjOi{?B`F zG?4?4+0XN)-FTAiHvN(YW$qu0SE5V*00{m60Q$(I$?g{hvhO@4rOpdkT?B9Lw@S47 z65BazFA+=$X`m&KQJyMC)~@7`IFQ5v$Ov&yyYSWRu1G-WJE6^IHN1#$%0U&ZVvO9G z(OP)NP`3axiC6uuwYz8IYiWoVF|$z{RNQAXb*=^_ZlIPQsN%7&E^OwE`BKKlyBnCqidB>__pV=4 z&@86~jk>FUc}5K_0P$ZO*)fpK*A>9GTJ{eV_;Jz~@@8_s-EM2iZlP!=bZ(&4Au2rGuih2!haj%fGRI$ zI~c$_gnL>hs5)x=Dl@riLqpI!D|hWdRF&lO%Ms6zZB(>o1qz@lB=-T z+l3s=EQ=eSoHN|iX(sNqAfoInhmT`cVmZZC6(lkNCJg{@X^=~EaH|Yu+pYz5OKW)q zE9Toszfn+pJB?dXmPY;nE81a^CbwPcti3QsYNb}zM8=)lXN;HI)G+#rQ6TlNTe#6Q ztGxM^ac9+33dp+fC4|76d0r;$4xx zxY6$I2@_jLxbKS5Nj-^iWY)U9?XezR+^Rh{n$}MgT$S4`&~Ik0au_$t7m-^LXpmg5 z%sHz{pP!M{Fa_ z64*&1Wg}+#3glt&F0mw_$7BlnZLYsb@jcpw6Y0)_xdNfjv1%tfsaf1ar!ctuD>3xl z3sJIIwOOZz<9AgYV!dH}Nv24cxL^;Xtyi9J3|#}}$86t40+&BVh0Er7BlsV}qvjga zyL~ou{OY@SU&67U`2PUAka{YA_3P8FG>-<{9whQf^fgKR6QW#mXRAg%15&(u4$R># z?KCYohq16}=N%-uIRo(Kwcyugz!Tfs%k@AjdHx;T^|!mXkbk{UKg0cDdLlQ`s*`~& zNUs&{k8y!>V$mP(MsrsaN6_yF`ZlNKG5#3^Rc{6BaSknueGO2#(e)cXC$yD)@tUn3 z%v$wE%$^IrE4tQLZWsNQtjYX8s@vy%oR}TJ%~iM5F6>wsq?R&0$gZbM_8ufTQRpI?XK%2`_dWTU`wwvLL@sPIXKi$Bo_MOoA zJ2S}7r5}ZQgx(3zrvCs{DMEj`)kc4Z2;iGr&!AyX+QeSXo_J}wHP+nt16PEfEJq;r zHIs4StzO?d>}1EN=C7JBsXVP2IrpF(RFc|ULQk0suhdjSKDeqvH8e>OYCc?_c4*oo z_}7`++)H6~BteMU=jH91?vdIl71aF7ax0s@@cT`7Ez&t=b}h7VSxzmxmlllz81Y(<2B!^?vJE=glaml`3HKF1*-xA-5BXMFqf6Y+ zknQAGRiJoZRk_k0_A5s*ADFT3D8HslODP6zqz9j& zuQ2gpk+r#=LCHORYSV3!(n#l7N#xTujg3+<-SnjX165*7nMTLYr9Rr}V3&DPO*-Lu zHAoyaX7nv;YD;x*a!Q!KW+Rhckw3w!sUo(YNxMlSnL%LZ8#7)`4qRYhX1;>(^J!CP zc2YyFT-wTF8-dO<`OuZaW6d?M3`gO6*>wFy1e0tIBwmG2TzxAZ#>QLWHL}XAkbqT> zaBI?jC1_7|p^a8ON9~f^apr^kKA)v|_k=Q9@PD#DtxdHxmXTT&o5RJ`hU4pWM+v^W)*VZSiiJVw)axTr>I9GHI<4ey9*;&YZsuB zXO}0h&UvdA{vy4%aLYS_J$bAkSac?rdk*yZn5>TLPxzCo*lybKIBubqrGJQbR}+aP zf)kz?1Hm7aa2{dwU&@zw?t+#4g3r0$&EnlYW4=2B44w8K4PDdxM`Pw{TyBe^ zuCf?Kbre$r#BOj+Yo7w^Q5>V8`Z=oR(?NCA_TPotx?C=&9BOy5Rv%7l&%Pqq#WuS! z*f(zEf!3?&TE?~T?it!!2DwXsGcq^b_N-k?#_(!)O*GL+tlW`P6|LD#-Dp8)qsGiK zqP8t`h)u#pj4Pd|iph(_+H`+05`I-8c*4i+F>FZ*+&fjamZC|OwaqPK)u)iFEPKcS zTLX%@rRypJfqEmHV@BAl1~cQN{ZfGU=DBvdDSH470jKPy`bHh zlAbiNl%ct^KPv-(0j~E*y0Y;Fliye&fm4mLal7eUHQ$D`q{De<55U@bS5@KbD{E~f z5;i+wIRKpIv5p#*Nhr6=xt&W!4pCPs`0^K%!ZW-@mS|WoDsXEn!WzB2nkpsLz&8oQ zhCGVwrn$1cNzLP^grUlLV-@q{y0hC~-F=xCojrdhf#B{HF_`%`mbzd4Vyrz+p=uQiOij<=?ruFT3kd8<+QsU#b2?G?T8Q_C)c=3PR637H0Mt`Vy$ z=ffJLB;o+$)lF{4tLZWvsd!u3t*)kn{Y~&;=*F6q(5)75Gg;k0NRkI%sHk&;&3Z%X zk+N;LjU9ckYK)dT75+esMCZ8TtyJ!baO`lv=eel?HPl>qH*)T?xkvN_)-~RNso4y@ zvoiXpBCSp=mllX^@2+i_7-n=o?3`BRpNub|08Fbgk6yLQvpNzTN|EX*7HV!WV&-V{ zNPJzV+$i&dMBe=>SG%*jLSdaBJ%O({ELdYB6;@L5o$F#E=D9t8QW}&@B!DaWt!2p3e94?S;L>9>`Bbam z(xUXLNK8bItj8T#vap`=K^ze@j(Y)D?{v4*l~!gRY5oz~v|18nvhh%`C9o?>*2Y`w zffPfN)YqnXW5C*cL1ep#Pa*#BTBwiuXST+MgobcKTO_>7|nLA5Fr(jPPs= znY#OUp@?X$xh3=-_@; znMcd_uFrKLW1pDSM(bOCCes^8)~B{*Av3uN?kg3mbY?`ttGezxYRRZb%*s9JTa1Z1 z;MAbvnJW2n5l&M@tRhg^?Mt*04K#7hOB&{Zop*w#owZN1#nFaO;a=ya34X>m4wd8I z4Xh+Oo8A$Kg6+xfYuxNgECERCQ8rR55bM~|tjxVZ&0LhGa4Gii#{EWWE;6!Q6WG(; zgP{)=@Ou$jD+FM0qMhWgr2ul5dL5nHWo;HV{_>1f7t%$T;!KZ9>#jDU;0ovOpDl{s zn`To;q3uCXdWv-!ty?oliLIi4y&9jwnf;k$=jOGH{Zv()&=~quwlV>t>7?^#5B=~e zWYM(c85VGl_yDOD&!tMPNaupBr*@Gj= z^bK7`o#Sb3z!QnEJLfgYV+VuHN|zERqk9pWmlVdN(djn+KhxU*-GTQNr5}rQ+l9xP zjqGd6lIGp`n$b_X@m3ZcB47FG5&Bks$rsMgPIIB(58m5bK7ZL6tjoU%cy{wXUSMzf zW^0B1&joATGHYP*V(zOd!G#I_JsMC zQW5H-HA>r6y|g)o;zY-zS3_gt{ZiWkRuZWG^%j~Bb)4hdTt&me>yK5aNoDJiUX^F# zjW+)P343Q4C#fJB)k}X2>H{L$-=AP9I<0Plkv*B`-VE^6`j?FK>-krGnq~9oR~#SC zy5AM}p2-s8AvP6UqjUU1x}7o&FGGUit;)reECzV5o^?+*R`Ctoagt1uvjLtfFN#f? zY^-khk6E`mZLO*RSRj1&2Dsf*Qj<}1EP+)9&;?r5V4O%^caMBTL{&{vzO#E%9_)PT z7-F5>1WeEq%P8qs!S^gQYxU}&7zWAL>jr^~zdJdf?pJ(9 zl$2v}MP62W9&N4amoQkhgx2XSEK5GgFd2`0Ru6_EXs&O8K%|8r4uh>z*YDEy@_*eG z1pKEQinHMbAl@EH!v6qT!5tMQq;%suaz{1B__O{(N&e+_dF@uKL)6PzxjcBBvjE+{dM+K>AH~M6TE4W<`uVYnhf%`&#yer;khR0a33dMP65nxmnd9Wf#@rv5BudKJSH(%a-j=q2Lut2c&*jp@W?V$ zocmWCm8ihd_nR@bOSdvf2;eUx6%~f4@oIW}B>v`37<%-lYD4`dU`J8KVBA~_eN#^M zAm+H}r#g*4GitVtKe+X)s~BgpWmHfI!N9Ik)4_folt^Z@jbC>>b*_%W@rB`moA+(V zsV-~*n7nPbcTrw+T;Za2A+*PQNdx{8OL9hk(HH*!Lwe_}EiEpHPqcY59tQEj{AdSltF*Wi80J`l42}rNKi0TQ-wf(jMCJn`58^|b`l!iT*D-1CHtWM$ov_{H`G|Q9 z8OSx5@W~Owqbc_lscEiix^VK=X%$ECKO(dTiDJ7Obqgrp{iOpnSCaD;k;d?rQ79pz14_=Mg#V+*cW%GAT3OrPEyoruv-)*N52-=w(?` z*wtjw-b00xYkC^WTbZK!j_dR_VRbv80ckV!qn1H*UVS4`fMo+P^;1<(+K}Q<%`PWNptZ90Jg_s~?>MC23-4$f^IZJDsxY!u3nNOiLwQsI!(2O#y%Ud#d zdrP}q#eh$BHAZg>+(FH)$@&veOgyKYMF{ep>cTO3L-OeWJ!G$EY0F zF+Yc4yO8;ovs(`242pNStz*Y7tYngWMH@4E0b1H`h24Ob^}MzI;OkzqYv8CLhs?Hg z$?oHtqpWF3VQS4QZ}bg(M!U!L3o_3kjfMHeopaE|r_7CD5fb z&!4<|Xx6?gy98r-+$)&D&2@f0y^8zAR}(_}IA^5S+~mMV6vXa@~<>4kCgQl?cOuG(xAT7 zVYrZsiB~vb#(Bkfz5B;;aL$p4!R);&XzXUn#+ZXMFV>fDy(~i$3aIWUu`TofsE~Cc zs4HzaHL-HJwa0qOWd86XuBx*lo4O@d1XYO00;w)Q#ZMVH=CdsJ?*u^{dL^^)GPyO| zrPxL5UTxr;vUFG@EsSwp&7>w=5_qg+^)yD5Q@pZq&TD$sB9V;ss#<-br_6C$V2@Eq zV}&njK*Q8iv!6-?XU~n}Ju^_>SlY=Da6+1~b~rVGe{||L=Yd+phE%zUV`H_lDh=FI zT3y{uLSGGx)pHZdSj{;~I`Ryd*f6YX zjVAWopE4H5sIO~iQHLU_PhlH#f%w*ntLjUFJYQf&#E8T2sgmB-Q-MCi=)%1!??Jbp zJE337x$EBuMsm>trma;>mj-h`*)DmGXxx2j88q9Z{{WUm{S9bOpkCWC^2z)vsclBw z^Sl99d+be$BHnakx;Gz%O&!J57%|H#{RtIiHJhlL4xxv*t4XdxPw#Dz{{X;68iKJ5 zdiJXj!raP7xhAYf;;mf{OreMSLay8Dx?^K>5s$*O;<(W!7-%9tfT(SCMI`z&Iv*Tf zC;IDHB_B+GO1~G5tQ-A)nxoxtYGb8n7eBxg{nJr5gY=od)m*auO-pd>Ris^u#U35F z`|+5`^-)&R z%sz&*>pA%T|w zmM5BnPSI^{tz&yw8|{#sJ1cU--n|cAlTWzNrjjd&$Q;=pd$}eg0kj`VxuT@MZ;D9KkCu0C1EHvszKCgCQ|p~N z#t1a4Yi5lZV~j5&G_z=0TOaP~8*Li@0K?+;4MNSJol4<|;fFQqmY)?ZSP!*X zOBwY{5n28>@lKiIFAt!!B+VRyJC_6xtxcqznA=Zd&wkX$u&ztQ(iwFXW7lA=m@Z9n z-XdMPf{*qEX=I7Hkwp}=2udi;8K4F2NktR@QQC?q0L2thKpJ{Zy{Kr?hP0i6cZ{Eu z{uSuj7mjSTFd?}D+Ti-1mA;0&nu=+(CX+q&mI=zViX09@3etgNoUDa`AY&EcIxma0 z4J4{FnXUf-xaX1dHR$?BiR?9a!MB^sn;gd68YdM|S(2|4OI)-_zI6M}4oz@+yJJ?h z+HwL{e>fK@`|Rhl3gb1+F6IqFE0Gy{c;mr!t~yjDPPa4I+N9K}*zD{s%g-4Ija}P6A$-luqU#g z#=ZBr(N^zDvAr0c-5C1gn&+=P5pMZ5lN3Fbx>sF*dY{6Lpbj{yQfpy2rg??uhIK1E zlL>Vn#GKZ8&v$Hg?sdmwUcZA>uXLL$;fshApHMJsRad$t)9iV)TC~V{1MuRnL#XMu zUoG9Z1pfdS7&Wzd;FudNX)#asUb*Yv4C+>dpC`;w-8rjH4`MAOUTp@#>~6BW3V*uw znRlRH+5+BJn4jK5nxV4h*}h=$k@!{XZFc_H{F9OgW6f2hwqm;KXF58ZGLU`6CNep# zxA6p5JYLyFC;Lh;X$*Qb-}vPhR`V7GsRuF(dLX7ibS42R5e7d z+~7@;ptR)#c(sMmtoI%F&QfijnARH$`+URDDkV>%$iEWuqHUsH~|p zJNs7M49~~CW?A^3QMJJHx2Wvg*6pv1;*$$?Z@2xWspXpUE1E2VJDpzG{Ojfc>sFIq zR>ECbk@!}vwxgzB0kU=6dmbuzH25SNhST({+g(vfc4lOg>5mTVi2Vg{-X()7@(GZs z_3QXoNv>K)l2LG1r=YJq)U@qd?q*v!-bO#f-33&9g#6A2S!l(<133r3db#1PA$2J? zNc#t4TGMXI+0GPKLj}a!gS_c+=np~oR4S51awc$>g67yAw$c;p zjw=%3rkFRH%zcka_9K67Alq?n2jNvNG+i3~8{|Mp>xzEa^%t}3eABCOe;}$I>vzT$ zvDtW*?%pQKBVbkr?rGlPV%@uKWj=>B@17a)hMWD5Zyf$)ECvbfUq8ffE5Fl}jXEsx zkxP|JakkOzHu}z=a6ff?huGEVBt}e;*B}16@OhJDuDIkJQ59_l z`<2dV@h{DqYj7$_qm~%qXp!G|%VBsGQL=YrvgD1J!0lZhjPy%=JHwh?)G7g$M&<*p zW_W%SX-&?lYaxLL?$a0!-t_+f5ntbGkt`1(XoF=U?1RuzIdi=NG38pEplY5i%)xQi znDDeswT2B|(~(W=JBCK_#TE;mo8hf4?%za>%wA?aI*Q+iRg%*Ix;CFacSgDP6;1plZz0Sx zGJOqF!|EyWr?6b>Z=QcF zkl(Fya+{ZP<~B#AIrUqokLP7SN@NA4`HpKyt0E-GSseRSIo4Rta%*SH=A)Y5a`dTb z4K`)PBb;Q_$e8h(gUi@RJ*s5wP|8K6IH{EJP$wds2_0)TGRlw*NDh0|Wp8?rA9^k` zLe5REjGEklu6f;muX=OT`@49A+l zHgdH=^mLP|56Ze$P-&tuPV81W3!e+yOfq8``c@s+g8|8mkzST#1XOKotxZj#IHq|0 z$AzzAW3+A+pDvqw0r{d~k4pBZu~XO@r!JKofaL{fs`n+p@v$YuXB*@njZYq@c@g4f z{Oi(gJTYq~a}fNiD(AtL?e_swRH{Xj$kEL~;%nHkmM0$dsc++H*Y5%!&M80=F#KxiyV5lQ``BS@^*9ygl4|#HwHyC zdpPuA&ibBKNhO|8N>^fk>#t=~cSW^cx?LE{VPx}D#RE?KRZ zUN3Q%9u?J~C;c3-9n{uzx^47F4{ad-0D#v|ZR5EC9%^E>A@P-**b)W&MI6&z2WXrM z+f6E%B{Cu~=cpBX!S=sn*Wx?KRt0hEUDVp9lXqn=%NFgY2LinR085y~;+s;Zk2f41 zDkS-p+9@j?HN@+r*lPNWryEyl=hvF>OZ(Tpx@lOP6TtMZL-?Jq6lKw1f8sJtgWs_J zb>?7HMF4iDo|NNFH9#VaoL9emDDd5|@R_=~w?=!H$Sk?}dw)9cdsy08krZW_fED$p zhVim#x}ECTTL9%l{<%D#=ZV~c;r)&To1%OIx9^g31d(h| z8_|&9k>-mZ&0MF77jCGd{efL_^{z|BL;fb9f3PXsMkN$c-hdK{T1o&~PP9=#1r$+0 z1r&FpfC^eDpaiGKBShijVy-$6a78wb^a0suo;Q+RB(Q^+FM>T!W!1(G)Is@$gX+eDAadkzGQoq;R0dDJ8BG*PSkE7 zhT%k#1#O_<{Y_K*eXsoXTGN%jhPOqmcclcL)l>UI(0|Kyqx(|SbeU-7mtf{y%b!{@ zDu~rBJIpBl*0az0;fgsW*g2PS_CcoLXQfg)#gP91mb8ahvd&Dur9Me^4ruo$xUjXk z1WyESpHW!1eh<^9$&FeQ{>?>o;>LpIk!?wfI1*>AXD^MW-Mj3rdFhTS)Fz1Liy*u3 zMbvpV(yW8gxHW}!qu*JR=Rm*?-Tu%!^7oqE#yC$!vIKySWqOLNiw?wrb z=a&YOCd#ox`BF1(IV6+nYr4OKMY{y~WHIie70%q~(E>~o6YSNrUH3+9F6KAVj0(JT znfn#|>A%?%_htO*>`ls>fTsdT&T7!orXR$#zuA+2d1$a)st_{F!jt9L0maO98fiq`X?wu~fURC+B-it196 z=t~=1OCIa;tmpVvnG&5QaH#~kyM_8xu~O_8QGzceq+~nIj%6w^Ud?%`_`Ouv{KtP?BrsyoTV*pUn-7= zOK0HtFQa#a`EISyRC4HgrkqKMos@bS(6;dkXoX%_%cR4ccCFh_6=?Us0XRlK!t^IA zTr@cN@9vcBo!B4tg%di`OC#p zGkA*hzuxl*tWn9WFA*0y{pgV8gr_F6SA)f9iDbdBRD0Lb`X+;>U-)vw$pU4e2Ox7_ zGb%GWp4qRYJUeZ9{{RT>gtoKpQxL$-XDgKuz4(1@;I*vmhp=u%b2t7E)ok)aUTS)! zcVkl2ZHN-WSPtT*kHmK8iGf4zC{vOL6ml92GeMx-H1V=|a%T(yf(L#0J9n=X)wRubHn~b=;(MENY_fsG88MMRocrUX3~_Ah`iYL}kr! z?0Bx*!;D)|DE+?(Gq7TsF>_7!}Rx(!H1tm=Bx-U136Nmhn1;F|Tq- zs!>RrwsZ1K^S0?ZH9fO=awCqQ=Bj~H$H5f9`U}B+CDZRTm~9!nsZ?{nfK7Fw5&r8d z4`Z77`$;*q&2HN3T8)<$_mZIW+;LetT~U1f&tYpsCzKibQw5?9LgVzWKC<6-#JZEw4VV8h*N8($IW7nt(y&=c9X ztt2r@0ggps>}qV$MlH^ij^k=5`f)@K>Bi?!n8UHLV2Ze*u$GiM(Ts|toi&s z1ewfh`Bzz-_M|AMsHUXFW6$n8m4@DJHO(h#YAT@r~lgN)4p4ROVlomp3A|rdH+#C7i9$X(j zbICo0dvy8)Z`L9CR~_PC4qM(Q*tR)Ub!h~BNc5?cl(!mACyY-e=C`W6%&#rQ!{-2J zB-XyYqT1*d$!jBdnt=ewLE$~C)V{-|cw*7-*4j8)B-sEeGrQKjXIh%#<5QLHNir76 zC%q)@D;d4>NNG(cYNiIihQb|o%(CTDK_K*7~^P-C0GioMMO5z<$AHZo2rhmCb6njZ`5{eW1O^ExStta-I{{VK1 zDSJtGE0R5f`$|vl(tl{n=qRF-sV?K>k6}NwBqNY9N86?J6j4*{CER?7U$#r_O*fNc zsG^Eq(p|^N5EpY1_k|y6pZmg!DLRtw3M;Ulbm#c!^Cmy?(uyj5q`Q^L9>Q;!{{YKF zKr;UT&qGBNy`;N`DjoaJ`RHjz@zz5{6uqRohbktHCXv;A!_6kH-`S9+&zeX6Bif27 z-Mz|!S1fw~MHQ_-gvMKzt#w=5Tr?LFC$M8(7N6t0*oaG{6E95nqKa~gYfy`4 zjrh1nABis7FxZv6&zrVu3`w7AD6IMvO$)NfA35|D>7E|(-Svi<1*}WNLm4KDDQ-T8 zeWUoy>JcxU0o~B_t@KMv<@un&J@6=^vXZmVq|P(q6zQPpMJpxjCq8^yHb$l3?>i z6oy&7u!< z(f(C-JEfDc6j4kG!3UaBJJCfZ0os!^Q9uT1gwaI+FG@mlMHM3xGd9A zqJR=YsT(m|;AXj9I#iY^ZzN2#tYrERN+_g+&L>qy@lS|^G7%lExGl4}uO;y%(p_t} zEgX9h8)A>86jj8QWcQ^5EGVL=M@gcK+e;I1YdXec>+4UG`$dm4r`INmDy)L+g)DC6 za9-eJJ+oaD4$6Qr&lFKt1VouCxim~m+c5tCXi-H}wi4W^sa(H^M%_yv>S?sHk9RXYf9NN`OX4 z%_@j~PF5S^vN;i(n_ny@&F!4`#Byy*wCoQpo}R%ve}ROQjP43OgyAX&C)bUe+&n^e zghfQf#3kev?kg%OtElQc($&*9Ff_8XvbM3cvv+v*{Dr&6%U5v!fWS9F!6Bhxaq$U> zNy#axX}Nj%1%*Y$C8Zy#YijH28$Nz&@96C6?)mbycVu*Id}4BHdS-TcWp!bJ z2hlDh0Ohz@;NRnh{njo5uwBTgs3@oyh;|_%+kyuQ0V*0bD>|W|EQYrExyx*Bm_)bz zVsqY~p<$O>B7SJmhIO8nV}x!Q(X^wM{kw)e{cpAGN5g)$s~5mQK>`O4g#ds88~bKz z9p&5!(rL3rRQfs3%tBe}oXbKWBZBn)7!u>q5t}{ULm<}U5TMP1ZVgb8!^VB*sx}r6 zf%vg(UkJ+FA+V)Ug%6$9J_J58`W)O4-D4T@h+eyX2#mj}KLp6ZLNWN+hrsVIS}Jp) zgtb;50{%)=d#05FajyJThro-zrPTv>sy+YWLtxAb_F-Ym{tyttKM+@kEthU1!-1Vc z;NXIOb;}M3Y(--15b$a{1hDf)ruK25+jQoKK;7IpnZ%z<75!2Q>;F>S&3>!<|EtyC zkMsY*z6zmwa0q;uKP_&*%f6GF@TDUe{rann{_@Z^y8p2t|5oUCLDq7Hfr5K_@TY&1 ze(M!9BBb?0;J5h{e{kV+xAZg+$^CH>w)}Y#I{h-6|9g}0f3fZ|7(Iu;ueI>`Ri8Y# zBrEX8vH7JXe>XP2cKiSNvH5fRqEG3|f89Rde@|lmT&vSs{70>R>A*kCh<{fr*CVxk zT5b_)`?U1Wjbo_-JG;LM_a6!|xczZToVEzyl=$92{! zSAU-Kzcnx)X?^oQDu#d43*RT#|DRPqJ?uRVBnU&l)d%7U4~;{h2)0i};pFzu$!q?6mVm^d5QQYhxe~%vAxNwu3e7O|E`ed zNyB82B*ldRW}?8Ok@^){F@ng-t7=8edT7`75NKNlrTWkHJ?@hK(mlC0u*(AT1OL@S z$A#ruk;2#!m8D~*16T}C`dL*1f@BEx5Nccy9FRD<3IRJDF=VwwdO@y)o^qp}tw(*I zABY$PNpS%*60jT35y`cKJ;6*JgA8XdKw<#j1H}1*55RGMG!-!!>V~#yc^3D}4uR<{ zP)&fUWXx>OpThQF;1F1I0?p*f@c+-w`GGRXZ6bzXdvJ;5fJD$~0ILc|j3HvYn2)Oz zf@mi~@)#HlY5_W;;)EFFU^v(b2Jsv+NaB)+el1Q&K!!u&Le5BvK$^hsvI5vAvMi2h zJCUpa8Y%dSBMJHbM;*}7BrI+qhD+NGgeZU0* zZ|WY{!q6cgTL0jZ5@=9=l_bQ}`5zuZ{@}Dqlnnp}E^Fxa(qJ@fxNPCJa^#{cyiUj$ zBbk^3hRvD$tjYK_r%CK{$B*>hq4h+{c=$xbqZMZFqwpme$zPZ4N~$c9EWD!QR;bTT z)YaI++Ewi`%?gu(-gB+2A@G;=hPuB-nj(oO?(5y->@qcD$@JW4<0}Is&$(k1rv{98 z4MjqGIVHuI=VsvoNtrH0@y-z6gl*|w$ z-pg5(d>9J7JGCQ&5!vC|q#7+9HMl5aNfw&G4{?4$>aoFAyLXCE&zXL=3P`sXlOP*ba5EX$yi z2{OS|XlzhYvF-F$ycE z`cim)i%{e_Vg2sci#js`lxokUz0X|n(|JWYP>dB%+Y@`^=^jt--gV)O3pxJh7_fw( zty+o$MbE4T=#|F+8iG|_)x*={K$HK;V%|;l_@G}d#UN>zR*7t`mQpezl&u3kJamD`(A^mRF zCjqFCjGyL*Ff!vo@}+?{U7b;a+aUnD00p~>%eAE-k7NTm4YjWtU)4o+Wms8q9Rpgy z&h0e}4&JS7H0J1Y*fzSyu&K%tuS*qbt%M`MbH3pLCt8vb92RwBQz$)ih=*`J_}9Yu<_6IH zMnbG_a`?pwRnAheyYL|Gblvj87wF1=J*Q^E${f`!NY+s0$X^oo&aQCKTv{wl6f;ed z^&McH?dVfYMwcs;dU7q}LUS_BmI@%9C%_-IBrZeY_aP8}|g@-Bl_VjGwVIE>wN`MToSd zfN7bCq@|-fo~$+@ioCQNmTQ6=SXJlJ8m&ju7x0v0jTMUe`8CC9Dz5XA;quQ;sKgY8$yD4m5Tkk{#6ipQe{{S7rC{2aXr@iv6{ghRj=x}BJiKc2D0LBt@J`aU7fm5=>Hu#C_!_JqGdSNR~yb&QnE zzOpUY2!Gqt!AqX1(d>jy6}L9BNpZ|QBOLU+79zfJuQve>YYtu^xywT2#c$gwp20M4 z-H*mKQVjEXF;C;JhQwS%SZ}*W3p=}+P0~VP&@Xyhw6zr@Z@}(pow0^>xJ(gC&b*@M z!pC~k$iEm6jZ%jo2uyS))4u+7p~zJ zJb}7{ull)lPddpkin!zMCghQ1Kq>Fur;<7NijjINM!IqLAgTMx;sX=o7XOF2!Y0_( zG@Z6t9tVV(S0xSBkK6BQ#9dk=DvX!JLev>pfM*eb| zF*ASbjSlpsDtE@hXrUg%M#~suAFq;sn*e2gY0C?i#|6w+res)Ec?S*b>Y0b&<4Mbd z1q6(MPI02HVrr)^5V<%&pDViMmnDRzH5DK{N7*EVO4RF7ZwxOD@4&X z&=Y*qHt5b2s2V(O#$R#ev%r5D##}oXi?e8JX2Sl>-Ol}@!#<<$a?#8A^~MyMw>NN6 z&lEW|i(AgY$8!y0@I?xj>wNIS<5U~Xc&l`v6*OMS6oSND6=VBsQiPNBcW(q#wBVR} z-^MpuINSq`?TV?Uq1<{6WdE~ z$rCYD$7`ReZ~r*@WxdhS1fFvCLsMjR6{n@*)-xS*VWZFI*EF=07nJx_1Ge(L?~KM4~X@AiBq#?q-^~?`LFo z-{$qRzL`20R#O{4?X0R{v05=rg||MNi3?ZG+D6tDIIs{tdM~ZF&q#zs`rBu5g62+) zoo=AAs4TQ0nO zL7^*i^6jfQ?=Mw54Lvxs_*4=j(;s8-MycSMm!7smh*Tq8ahoV5b$Y^@JIXCHXB=KM zji0-tIZ@Q$9jd(&BKulSv4fj_+ zpF~U0!^pJLsS9JGq!PPk&6w^OQ#(G!O*{T#EV;lu>c%pMCT&+u_jIEf;r#6Zwf-Jf z5Xgdl6&>?;MJ5X(TRH@AC6WW7{1J+k`Hm?;Y2z@Gv2 z=3iYvx=VP51__g{SgH~XuHSFFo^mh|a9``l!PEn_t3g7XFSalpGw5n2#GoDNThNs` z`V!NA1nk`qa6j~CMS_m{Mgpt{!W%g9GC-@~3|r1S5XUPy1WXTs9e$o|^vcSuufKT3 z&yG8WG=T%*7k~qDVp)!zM#Ny89CBa@P^$+Xxx%0~!g}nlkp8U2k(FRT)W3-c$d?2C zGOW9@k^*NyANNNg)Wn}|H2?wa130cfd~QU~y#u>K;HWE(eh48WBH#NA6bZ1|fvyaO zkR;@G&>$izhkm*nCBP2&?un=#xlliDa>xptkpvwd(1!jG9pQXA&}~wMfK5aUBxob= z9=TP=<`;C~QWC#FW)0cpdOiD@?>K}xgDz`4k*g{s@jlCY1wv^6NdPx7-6TmPrLnh} zKPSv9Q%^HTj6corO2~WzBo&8==P7?&c-~-$d)vJ)hFoEQ1cOZ&;v`|yAts1$f{$GL zP=sJ01{@T;rXOPjHc}XLEJ2Yub*O4xfgCdm3;>86;D8D7bS3su;hS^=6MK7Yb4S*Q zJLkCzA3xA6psTMYE{>X1W4?XUq~`f;|5Q^9Ic2LPjBvT07{R5=NpTX%6k$R}zp_U9 ztI(I0a@il2yyeDJ-7l}$KY2(vs7*t|yDaRwnkMlrc%*Yu8>3HW^)sbS8x?6YYs5i; zbj=_(8G1BD`vc~NYZDD+T!Ch$_C!b@kFd>YA^C|njs@q)(LeEue+=m2ift#*0$>Nb0Bh!vxH8CX?g<*dOz?iXyOlZ2asrL2wb8Zo6O>%5KnUu`l7uInDJ#TU;sSM!Pdt=2qL&|xN z&sf1DWF7m>j+oINKie{HC6`eV)zX!g)wm8^oR3y>4Y2OsCQ4`pDY1GyUObiRT-=u5 zC@LVybPj*V5R*H#zJ@FC;i^!FDWA9xQ8rzxqUgE&RzZq%a24-)`iCN+3zg^ZY!(F& z>5nW-?W1NW9|G^*#fO(tT??(YoDpFxr6gXe{U+rBAC>=VDsRFj?H(4NHHSUGzYd!T z9OJO8QBNiiwz^S?wM^{ov~YFUZBN!~Y%N1PQ?JM*Pl#5(tu=c!O^WZuLo$Zq8l#$w>iO#(IHi3(zHJ#r+zTf8HM`Qx+Lu`hu#xA?2^CPvV4Eg&5?bwbCH{i@1O z&3l9{U=;x=!E;0p5#-EiLG18r6DjdPom<&$X&KP=W9O%_llE9<&_jh6c3=JBQ%l65=^P+XhmV zpd|wGf7mEtfSA_uBf)129)ba7xHlZ=&0jXH(s4siw<3aJb8Byt^}ZgxmocWhUhrnt z>aw!C^eE%sq-4+K*=Y@thjcJeYP3xgCZ7uOy2-o+Rk5O?A|=2T(+vL@~!F^ZH* z1XmCG>Jwhlf6euqvRPrV`-GzjZ|BG1xj+(s!laS~D&^uFUy@`$*6DtwxEac;7YNdt z-xXyti&&nsQ&d2G&Kb%7g2R3re?qeibtCvHx2ECB%=wONBq=^|UJ)tojHNbB#a0SjP^JK#_ZBo-PcftjL-~9_D^9f=;wrGNtRU?AB!4g zEjh}gaAa~kQ=&W<6lEciB*e7 zsZyVrgu2nnwT4KWXu*9cB`=Ocz@yhTWY?54HP~&x#Y|fhmNI)mHF0b}qd`-Id#dPd z>Q)7I?yI7dt*QH69O>cYNqBi5%d4!Z1}M8e>qoSQNhHQzD3<$ya5b zP4^>A_6g;G+E#*}15TjO_p$wsqJEkc3}O&o{)2!@upgNhP((lx2c_oNhMhhG?K0pE zH(^K&-q;9i&r9+UP**MSvh_0EJSeML%?ciQRS#D>C@!#6o#Gi-+_!~I36OSz^+6b| zW1s7r--&cX2m&tsfZhm2pOY8A9StR+(6ZFP$Kqi)4x7<}z~6 z1TKt44`vFRd$0Rv!b53e^wut9JU{y-vy8$_aqA|Lw@->3g?GaF$USS61z z@x8TQboO`1_y-S3{bb$nk9O{qTl8}n(TG5*jrAS~hrlJq4`U+5OI`DulvIULj+_{u z{gw7d=!-Gnz_1P3-HTS)ik(iR@kYVs;kUcluHI>XPv}&pR!lp^dDjj@#9yAw=!SOE zn^D&gHy~T1bN&MF;cWe@19a$JC4MZ^ z#kpSKi0LqhIrAn*jwxzgjUzS%Mes^JC|d8=Gh!wa#hs@~tIQl4pOTg0-gw}T4ipjB zh;rTPYRJM|6rgNO*z%&o${1VC3gUiPBKQ=~_`~CMQfjUHUN#OlUcuie!9KU6yzR_4 z_;%K}yayv=%Q+qtwKgVeu`FMw4+&*C-jnR_xZ~bizR!M4#&sz*!+w~iz@fWjq~BCm zV%VC3JJtfL?MY4}k5;F3EU$74v3M|gi2G?d`)=zIA_dAA=pr*DAbfR%O86(2`8b)j$7c|HU{x5V4}Mi?F+?q+j;4G#mUY#O2&p{kT6cVCl4QoM ztEsm{g=g*+EqfQ4;4F*U-TqwLcl8YF`lXF57>AxKPdN0VvAMX(#UgYAYAvT+)?&Px zuI#f;0+n4|4-2Jp)fTRP!jshR?4d^5UrPDN@oo;o$NEt_f9=H=SLi1rtQkF#RECz` z%)!1c=xl8BH&+|V_fVZerw3F`+T~mk11|u zx?@enU;wU-ihz1M89hmE()S#rpFC~CeF4MwDJb_C8Wxp?%dQHdzYV&3|9LpYG->~g z>R!<2m?Gk7GU=Bz#R8=6#x*)6^$y)RT?s``gvbOOBVG}Ast->VSDl@{l+KnZ8fTYs z{k~rn&zx}0g(aO!<1!d8W)-DSJux{w*tB|5?D8a(C9iAZ+E?_VSwIBZB82JqQUhyF zt9BO9f*g}JCpsQnMg?R+_8kQz8Ge}Q-#;VF_FrY-gqit|dgnV0^RwgVK|G46z{$Ed zVc#K;MJ3_6o(z7-r1BDRXgvu5$p00AgIe^*Yd~rQ-QX9aqLEA#@<;T7?2k~W>g}>( z7^cY{Dbq37*eqAAit8O>!(?h?&^s?Ik|SrrXV!eqOBxC2(JGdLSGfr%KZ-18c#xqBhHD0WhXcCQj+fn0c9K(xZ z2dY;gC@3?k4n8$}s#D62W%j%&T!CV@OxF3=8^mcOX~d?O>S^1>z4F=MkY9WEbWejC z{>UVt8(M;GWs&0NCcB{|a`lr1CNJDO$d&Khz0&;Ux)6>`!5}_Gx*UqZj8Y)_Vej>p z4@X>LX<=CNq@?{A_}GH3b^|A|3OK{^-`zb59XXkR|8ORW`-K!cb`Srp?K-J01f6agPK)ms#Q=%tMg9FmSz5FdR>&jt`VrVJz z8MkWPs#3!*_|W%pWyD*K={$EoZH(7FA%ZeFy4{0$s1~nuQTB@J-M3QV_MxG%W<;u! z)wo-U(W%`--L5KW$&p^7&aZ&)P#lfNNnfJeJ=+xnQCqt;pVbZHOX92F5}W1E&%g*Dg)h=>>BETp zJSvKmB?6cd&R(v6kSR=6ot{eEvf&lOB>lBceDxh^L)<%aFj``us`_;fp}kV~N=x7> zHSmnYbi9-Q%2GkltC2>v5L3n3k;zB?${+jcAASizVwd0KRy7X29Q>-5*-osCsH<&=V9Fd~&G=(+_9gV9DAG0ZT zCMaOGXN4*FZG;dC9$$EWXVD{$tR`n{!H&IaTBTDjfg2u8ZR$!$8oUmZfTUkKJ#T*h4Ek(8`EP_G zCw};8#SuH@n@Iv|Be||JS+?+X(S$)YH7B9A=2(N&^A$$XRWHq_@MRY@Et#aTb9vtA z-n{AnFH4y1;o*uWUPyRBE3`#0ZJfJI8EQQ0d}%9KOdt<(2;42wtl#az9$e^JEu=W_ZKtM8lV>#T!;r@1iz&UM&nEhxk)vQJ zijt`l2mC%>vQC>BD~YWtEhV$#2cr{iSz%rm-06(MELWq#ncb8&XD+S# zkPfInr>q_#@>1*VBi*wlkC?=phox6iIeKz^MsF76=R9p@G^u6{ST*mQEV9b)*c0n$ zw4m#J=q`k{pYx2veiroZDggV9vQb7a7jnDABIRBN()U6UQ!(}J`I%0(=Gu?b(ZkJ% zds94H{Mj&e`3pLeLX!+FR3%LkRm6=hKHQ$z-uza_Iz(*he6`N8FLkm#gWiz-@<@Uf zj``Rko*!xvZL)m8a)JZM6*7cXcL)No5o$1ah{}*x_~0=LTyAa zAZ~t| zz-8H#DB;@dZddzs>t>xxRDmq5T|cj>Y>)4!C&Jdfw%(J*x;Z@y@m)-%E^98!7O3JB|jeQGxUStQ!dlq=kqrRT!GxjcjTl%)xfF;JCR4J`=xD`zmu}DB93rZdz zL4sCFVn_slE}qm~@2+{1S2VeiF-;_ZK~KjO4Um=MU+BX&zIC?mdQQNqgR}7iVgJ?Y z-qHt!(X0kY{r)s8{W}rnn&C;ts-*t&)pQj}yLGgod6O5N4*{d7F{L&ZX1$!!&V{Fq zB;L|PrhKq+hYpli`O6BK$xZsHhQT)`C(U>WKUZVCeW31m-($I{0bP0Rky^>paTm(;A_jgPu((U3G+(q)7YE!T*i&kT$GlB6#Gy!eq$&ezuK|a`1aX{T}eX?(->+Hf+6wM>TV@hTiV&m8;D;sp>IO12!+)pa^>raTZ^C( zQJzx847b|bZI2dLe^Yixz~E2UPKfH3lUP~ADl>ux{1G+^^3*>OG`J3nKqC*q_1y0( zrF4K;o>)#u^;^#s1xox8=?n7DF|YVuj!1LAi7_K1D?qrtg26)TJe6+B4fSHOWfzWS zF+6)iBSXZc0G9~qMz8LdvrW_j78lIlaa0+F)3b}&T5tPsf(aan$;P&&_p7D#Sw5;@ znbOdr%*-gz`XyL=s}1!Q=WAjEBS}Gu@cRVq_Nj~iW+VnZIv-sGZ-5Jl;KhFsb!^fN=lXK?wc_Y`Zx4o`k(wgXngjdjrs*L$NE$$TNwM7!+FzXNDNy@pG{m)e1o|=IJU!T3- zVQ%&#GRDdr4|;w5NxjS15C!Y~M_aZ!rW50xiDx0?FFnuW9uWI*ZhB@sJos#uA9tH8 zRSfBtQd>?#lKO+K=5fZRjixLCnM(lcLLEfFiYp4W!=$o}crBZw1v_6@9@Ew8@y@QC z+8%eFyO`?a`lKBC)~5xWnnzmR=IR68v16EzuXs+?KcvVsCFHVXS?yCLyc zfX#&0{oqZ28HU%j?AQ+8zAk?*yf{r%*upvG(!sne;go*0K_j_Fj$1K$7@oE$-tN8^ zXA)wf41QH7%A%&l zC&4=IIF?cWJaS>8`o|ClK=kw|>>|qU=^n#a0)x~P! zUguNT+7w+4d|v8X+^csL)z{Ms)tY}YHoX3Ap71zbGflA~oeztv)>MDz0eLG^!cw43 zFt(|gneS(}{xg^ut_9jo+;ORaY6c(d7seI*gugXaY!cpsAd|yha-DzSD{N0I@^JXm zJKqEgD1YbLmd``x)`HCHdumhFxaC&0#**4`EomA<&OvBlC=(^C7Fk@6Ky=mXzI5P` zBGGfsuobwZ5vA_>P2HkTaEvm^wdAv=-7KRIp1N|lVP(fv4#C$>;o{BG70UTw*p^9B zpBgslW<6QFwtHJX4A)qBy7zpvqfzUZ_g8N+;C2QB+ZlqN7E9x(Jhrpch!*IID|_i5TJ!2_ zaE0ks-)&dSf_;c-M>fA6_94JDRGbA-G8;CrK_i?diycTFbc(lK;=GgZ)KEMZfAd*4 zymy(!4eGIO&rF9}cW*j6FIz{sIPKo}W3Ue2lT^M(4}qk}f20`se~%18#vqeZGB)43(BubzyJQ&MkW_ z_m(_=Ze;RKbzDG~q1)%}c2mVws;k$N%^P-Pr^HuXmMhIFRw-QyD2R~g!|9>7EPEz7 z(#_dca!VsjMqZ9}%*cXf*gpHkb|3bp{frcow5H{vRGcrCOJLy7x&E(6@IR5|HwiH> z4SNUn$`}yLoF}Xui^k@Pd?LD0;6HhRPu)Q&QVRP$W9`*9R<~57brkb6?>W=Rxo-@- zwClUsB0Hnu;c3<9tbf;|mPgggA7F{sOc^Wyy*Un-m6CUDf^)(MKIEU3bT9ngTFDWm zhk6pmhX1<;?CBXIhl1(X(I@HGClKLpanYZ!64KHkkg-+aw4Bjlj17e@ylCm>!d4Hi{MUg-9ed(V97W|2HosX25=-WIesR#CE+f-j zS+{j!T${lgg&uN+xy#I9Zp5Q#5%?p4)>)8NWI1qY6%CQ%^DE!{pJ%||ly^h3uEbnM zHVj@&9@1W$Yn%9#*{WY*p%ByVUz#Ipqc;W5$7tnzb&1qie|y;Fb8@1#QoJF%=$JG% zzmJ9JMr7ExBFiq<1x;sOLbc|lPl=Ci-*=lg6kgNuFHC{MZ?!3i{or8#V0sV>T3~vP zst_2s1e#Wadikq^{Ri&&DAw)2DyaCc_w|Pj1GlQWUC=}CBEunnivciae8-%x&*>nDyLD(HCc)HG2z2E7aw%SJ(1izpnHgm!dPX%dTFr zb5!vkT2!lQ0rj!HlvWqLvzXEQzYQEX#V^(#P5(eWL>l5zz~xU`|G@*NW3>u9&O#LY zOG8ncPKNrQtO@_Msx2~Iek#P@Be}Q;Qq6xs9SGTqA1w2tI-oy5tLz}3EP2cC#GmL0 zA4zw7zf-g!`f?%3a`zSf8_4|q;uxGO!yV>ogUs!)oIO$7$MKuO5NjcQF4hy|=g@{KLmKUb3qb?eHwV1?sQx}y{+d9Ip9^zs{P+UtB?8f^uJ zYbJ8oeKyqbkNx&X`jzCzyjZaN?06-UuI+HP?yqDoK`U-=YP<$D;fFQ}VY>gU2|-PX zNESTG{QJd)`%~s0X`J`k_t;mxK*V%JM4p0+oPX@>(~}vrci=>l_BfhV|5#e?nfF#L z9TC!i-7~U7z-_I`#J)K15D+d|uYYL#g0puE*{bZIZE0_6Q)wyEb;H*xi+^@t!Bn@% zS9|PX27XEmX3D>^$DXbK(MX&!BFOLsUfj&vWEvQ}Sc;CYm}^?LUvczP4xFoGLtI z{?Rx_R+#wCiy!9kZV{Il%%-`98vn<*TW>0GF9cgY&}g)L(l_y@h+PXO?CEik z$=^x($#n+MXTSnOfk8oFfgpi#sK5R2MNq1Kbjwe)fi6b!G47OVId;~MksTOv{<9$B z;WNSv34bTme+pEQBLhR|!Jr`!&K)g0{k|IY*Oob5{J2G??_gy%QT$k+BK6=gZ1p{i zU4TGV-ah9LXhGX+qO>}`bhb_29!#$PaS6&UaB>NCuZcb)dbJkD1Hj?IP14(=f%fT# z0I!_u<%HDo&O;!IL3}cKZ$$#U9p3Z+yj^JlYe9{{v=o-m69TcZoDU~-3NC}Hpn$kR z6yN+{dyHxVtc9F?Blw>9y_8jty%nF87|iKZr^Ow8stM7^gAcGF4kTh?NMu!U^k$P7 zt#9gO&P%pu+FcoC!SBHQT9B}sqiyH;Bf%ipwAFWC6j?17D zDR?eWqq=S3$ls1=MNe{zmCl}JG(<_ecwXNH9-o?NT!mu9z>^duFDs(H_*}=Vlk$!9 zhu#;J0qM+59qN}&F7cIY2kwkj*j}LI>rm|$)7h2$_@ZXsP4%9AGehjfyHU73vHr9H zcSAUmjqtcRa<;B%`rAG?3_!n);(!B2IDXSwKJcY4W@qm76X%-vu-{e3-qXW-U&FFq z)p|0T_B^9a_8TmTg$9f$OJUbBYj7(C%`JyXNETpz`Kfp0nYB#4EqiB*!ALW@&ypK8 z%w=Xfx#%`wIcav!L^}%>pfU!FSa%ef$X!K0K|Pi;Ln<#D<~7m`Dt~t8mVDO+i0|yk zuz$X0(v;ZWJ?_rReOuH?f!ItYctI(zAFY?px0o`NQwfy}C^1QVFeo~yrXtN|)&Ei~ zoodO)2+YXFDLn+LL<%^}DCbo1A_}?msOZFLoa}kK7PnNRu0U7Xzrgya#qewo2nKWx z+>~IO)9qkL4reCS_VYp5P`n1l9y9bDD4WKx-8FD&i@?&;kpxXTn-9 zuO94^b+qoCOPMY8@9kIXR}oPN;P8z=Q3UrHGTZ`JGaDVjrx~ShZ^gSppa^%m%x`lzQCfE9BN&4bq$|jPr9sA$G=a~XTESAlcm25 ze@30OG$dfRdv6+Y!B>;xXPC=v2Im z6)@m~xad@LFz2FB`8XPJOPWi1dwM7mpbt6(2FyMMH^ROZO^AlmBdXxTCA4R<-HJ3( z1^$j`F#RU}glNbiFl5#o@ENvPG$nu}@}NAyS(uK&hEF*0=%fUkEXV(bhQR2OziKmtZU$3` z)LRi|5k1?(YOQm^WDw(On;?PohPS2_GJufx{l=y++Q*nVB^oe>UT*%4G@^ywT#c z#RD3RzVgEEoV%2F=pqH1dxsf~%)%H#6%+#mg^3iO|E~YZT&Rk?r!;ci@8gqyiVfT( zp5^gyU#pOp4Y()9F=wZ}KUUyR#V0U7;BMSt!jhO<3ZF#pV|q*~zMkm_y}rP*pSbSH(Z))S%gR@fOi=9-P zSMNUo7QCGeiYY3psu$2l{3lFVni5i1MCOT$^r;-F5{2xSrX&j5lMPw22p420L}^Xl zO=H&2g^eya>12InGDe%bT9Bkj$HyK>u0CNmthJ%o`l%0)#K& zw1pk}EGU_Ou;L(IB>4gKe{5uN-v21EEhwM@3>Nur$*B>kI!8Ha-~u^f>xEwkO0s|2 znIiuvHi!-pc#{8*aPu@s{nKYeBo@B+h2ZBRI@9qSlWuK=xopw56Ub%{C}xWkDQVpY z&k#Q>zNL#M*g*BEEZmE$L)FqgrIxN^qNTScToE_snqJa8sfdpe&G>t9BtNu%~yXeqb)VQeG$XJi@%$@C%FGn&Xf{g9+eovBj&b;~*)n|ZBLx!EaUcj(oEMh(x4gr|KAJh4SZU!UPen7i#SO3x{G z_sB@w=vYyg-r#e2qci`ZL+P!o?we4X{%Lz#r zpp0&Bqu;G(8OyXc#i$TOj_SB-aQPZnmFI>{$!&*}G^=<3(Pl zZ#Lsk^`_x{yiAvs95&A%7>Dr?IhA$^AZ>Cax3d&@dZV2LH-W_>-nY07Njk53wx^n81R_ggE*DLPl`W3z4CP*dpw2QaOGs zkzs+kmq)waNC1#pW>UnSIVFhiN6S-pTaPzKIgUd^Bof9UqHhtx{3EvN*XI1LF!A9k z)V|{6xi_c5^-YO^&=k7Cr1g1?X-fP?R8Tna@xu>5n$1z(Nbb{vX z(rg_R3vb+wHYqK#M^60kt&e(M#6F#j{ zABWL>@QVuJb187sA1!9))#p#W6lBa$6sx}Xp1hSzI9RkiG;?8*&K>WLi1uoanNSbk@5YQ@Hhzkn_7(fcU!?yEqidM1zLk^5-tpZpak@6&3H zv`Ca>XANlT%G|~=k+aO#Nqn@Kd5-!UBdZOqzxb>a9g+OmKWUbKb?xuSpMVO`=|0Ku zIB{f=Hxm#N^$grY2?>$c`xlk@XGRlIwiaBTMT9b{-UX!!v0D!NiSv7$DS8c*^4~EF z=)qNh&L)_7@fQc~r$Yzef$5=vKuIEw5b_y}?r< z_HyY&q4=v01o1;@2-pH=9SIpc zQE(+!!-4!YE1_EBr?_+Vos@xd=tULF)@k=(R*@DCeodC+2ivrADsB`;w^AReJ>B~9 z+19)_nk9%nTrJYpN2|TR7G_9mY)M^R7rDQm?!I`>o>HVSN9UVifvln;(ZuI(VB2J= zPx|4%c}BJc`DMfcc-VK=_#|lZ)ZjflCTPLEiH=vz5kaNDFVNxsB~BI@p?eTs>JgI> z26^cOM%(8uw*F>hAGknoOEPnxi5#!rCf)zd ziu$ls0-LAw>Qh^))Z|%OQW@`Qh^wv#CvG2l@y+R*Ntelvoc#Yb+5DZf{<-4@+EK#y z5g+(#ib_7H6$3#r{K1sEeW$bj^F~(rATJ4~w5o#pOC4v)9VfLTum~0;=9uX^YR3_t z{z*N2-#zO$%HekveDeMLJnJNv`ZbE#rfR*moX#iG1xjyy>;qh%-0BRDGK!Hu$cO?i z%8@jRQ=EDCX>%-5-m+ALd)8?DNqV2yuy5nE(Hqv2SN%`V=#W~F6hfp*9qkY%i`c6T zkxh(`kf$?33)Sa< zFP3=l!6y^(t1ckrBLdk2f5O`tGSv>=QKyo)?u>g03-W$59-|!$Sk~H+eKrGz!+_zA zWzB!{(c_+Fb`UYs2F+8{%-Oo{_AF!Jg@^U(r_Z6(z;H(7fc{?)C zRq4wQzJ)!=YuvBCitgMo49`BIR1*riv(`5-IN z?XIMf_rQ}6!#)J@T+v#5BjP#cTi}Veh%W^xFn^JGAF?vk?u(m{yln>lVA0x9Mf%!( zm@AGRDg+-glhC@}CcV`_PJ=OF@(1f(X$QZ9iL>RRx@$Gx<3j1d!Grm{I+^zAlnw`uq1Ozd6+E4{UwyplF5 zda&!F7tBVm8NM3#-r}8S^zdV6hC%6F=I#3r26cjq5?RQzIy80lm z5Y`2qP(byEIal#K@dp_^X7Ym_4rLaNU3NU2yZA}HIOGnoRI>MPh{3$TkOsTXHsyhS zssS}=`rQ<%OIOa~l}SNdb-s;eR*#tI-N)UyAj~(R75%x(Z-l$!fZr^GdY#gAY#R$* zcUwH_Y4As(R;(`%@Md0CRdnh4RZu_k`P|VV-u7^}ZSb@6xB^uok`9qT)x1w##9QiV zg#*q~>JBgfQZ}K_d{Bk7WtI5qfO<>~a(6$h142agxlQ{^{WFnLFh@+*n9)b3+{GH@ zYoLvYI$H#Rnp~qj_x{5CH)vV8+csAAhVxL30WMBeo{w~U%?S5djnxGyl+ zS_}uf%con+*S-oKoZD86_Qep{XShSZ`!xYw|0}u(Veq_;sSUhi#V4D{{?X>X0XAt| z{k|(-)3q-mE>oO|Ib{Uf?ZZ=xdagO)8u` z-_qeomC9j+o)j>W&`D=~?mcoHTNA@OV%Ys9Sno?aYg~vsxlrsEBA>^r>Wj2mem)ZH z^|IS?dANAbB$qV$&y6hRe-thIRzIScRZj2-HMcVRvDZfe&x^ZCOc|mBHxv8cjZ-?C zH!=5267nG70QM zz`2S{rxlu~b1ktS)bv4BKhG!z-o$B=Ry34{&wNy+ z-`2ON@7!O-y7(x%IiSH4f3AsDxrr+^!YV2J(fpYm#n3__aDi1`DnT?0h9aQ+d6#F` zeBthj&cb$g_Loum>^ap^vlpuSWHyXwaTco$PhuP$+KV$RuYV0qQjQa`)p1Q@lWtG$ z;_zsj^d?WGR*AGFsh$YfUldZ~34XPw3LMmT5^dg^5j4tcD6S*r_2+K_dM5{7gV6ku8Y2VJ- zfhP)|#}3BH+41T2;p0n-=6JNzrd49ut`XWIvZ)_ZNg6{qS@Fx4+716c0Z&)YHPZM)`%{Qa?34YmB@2kFQ$nUeA=oD%C=>pLcuC zzdZLn^L}~e`H)E_lgVCt@2r(|{n}L|HSvf6vf%{GJ!|ui5GVg7L9PFf1l3kNbN}Q2 z_M+RTpe}|%C_r|XSaa(=*p;Szh8#R#6G+MdQP#xNC=4bJf|AFdmL@UaXOR2W29d0~~_7#hifu zW)>>1uWPQ)y>}`}hH!R9%69kQ&+w|oH=ZwM(*1I>g|@t8306l)0O6PPWS{l)NAsXb zO)j0&ABB6#4<(u_rk5$vH4yRKq$8DmKYeN19|dgTXtz`DJqPPA8BKBkaWfg={SqMc zNIm3-?2MOHhVz4eF-ZzztMqlcviN@msl)-dH}~31!M)d|0D;4}_#>z`&8O%$o0m92 z;grgY#_NOM0h~QTSi2FhEaiKBBlQ6XZdDm*`BiZ3$A?ja=5zPPU9oydzE^y{Ow1A$M2cuEt+j0Fg*BQObwM<$5Xa*lizE@-V_1< zdyH$_v=>dpBKz!T$a^Pj)&HDb)vs3B&^T>nqhHPEe}2*Z`^;2cG^}d5T=_O~&->~n zgy7Fjc@pI55z>nptly8Ob-r4^x2`|SfuqI+nEk-tT|d2I{HtEuEoHx=1ya2O5GCHg z>1!_jOoB|xeFPbPQpo^LpY*!&DkbBn$yR{bC4c}Otk9;rq7kX~6Tj!OwKtr9CdnND zVKTb|+Wk*U$-U=1#oRySv+#H1waoQ0c_;n`XB&6`L%46(qQ&O=iE?Rf^2g~+_dj1S zDm*?&x$&FT^!IG@@8xTUjO*?Mg;@OK;I$uV{9k#`A8Zd?Ue*TQ#$f**oX=d~kz;=` zMULF_aX1hFK5=~*5$h?2-#XOY!*SGyXg|>YNMZmt^7Wd{GdaLhK;xr_ncn$0Qu=3> zeJ*$8|K{}_8($xC>FJZQi6Br#^lHk%`1>U&z_|^nvw?HFEZs0 z+TI=ZtXTY+ckAPoz)r*N8;^E<Rea*&>ZM)PrHEG{? z;9Dct|03a|&7L{yqgs>y?fOLj%Uhp@yU_GDYhR9#oI1kdm3c(5TX7q0Zlo0okhOj5 z05ART=;(hr6#oll{fp^FRQ_8N0ubFQK;*9Ef0e024gaZ7{wEv!-^%6x$@_T^n-(dG z5K(b2{{UNgduOE>>iUGjl{Oywb_jNE=2i7DufiV|? zaKpkMh~@7G07aj6{OJWK5&v8s{y$OPF@Q7?FZ9*0lK=bre|m%-426X7*gP)xLhg@q5aBM=%Om*TF-X~;(u z%$x~})CEiUK)D7{<1g)W?XQE{pz`vHasLgZ`!m*ky$BjksqHHr44olOGu6cHG{fJ5 zg6s*ty`aQdtAY7}4yj@uy8;Sr@w!+@g{Q8m)lD~V5y-N@)$--?ZbMm`Ks3TDp@8n*qItqD7lVZyUHIC$W@^QWVZZ$;73(aNWftukic%6&{a&1`Q((?&yMWA2# zd$TTi1xvumo&`(D2o_V*ar+*b+WMolf~vF=u5~Z@)YQP24Q+fVC;$+ z4%GzTOp$-%Hcc%FAGg9J>I$mlt((e|P?InQFF6~|^F|MCYS~9?C@b-m#|PSv-eze7 zQRy-+?a&YL20U#hkc7-WYUdr=6+Uu9oLtagEo<hkg_A}8W^UGGiaB)N)(Kg48Gc=+pz?|_$|4Fg-L+;<7LVP< z3j6z`ci3|sV5_T&5|$SGt`mhVwdqqyuOtq^Lyq=6AUr{5rpwjQp`!E@J!~pbRyQt$+AS+?J`mkm34C6!xZTsd(cf?SyWz&~ekDKeMNm31!T3 zEcC7~5TUNX_ntJ=7i%5wq~nu)-(sLW?-d zD8z5|Ue1#F9@M(4ZM0&fkV?;Ml@>u_kz@v_OvzLFqfY^-8Kkr@iq9pAdmN5#1}0q4tt46KmDt>`!OHzV#bH7ga{W(s}U!A$j7}su_DyZ_Am3Wd` zfZg)d^bgmcD&-?!iD?W}no#)XpBo1CJdc)7hkYrE&%HB@JK^76$IY2TpHaDbX6^}b zt-odVx91LyEkdw zr&yk7XTont+?c3MxUR-HKiD|kdJ@_Mpq6LIl8m^u`3f#cR<`9|c=psv1iUxEp15Mb zv+eLL1${lmd0`pTQP5z9^7+-inyM1Xz?8UQ<# z{W39Y!{}|BOaYU(lP@M())(B`R`(sW7anU{WuF~T4L@+3=I)s&)wp$YEZe-T{?#jq zS7%<9dZh>?(9reDkKdIpC%(92(chuiLHkaKRAeRix`Yp-d0;# z+nL3Z;+@!JLxdQ3OK7TY?%X29Vp0gV-H_?UR-&`?1F`uOhdrw>-Tvrd9%sn|f z3aN$D&r6S6396_)^eufHOxox)Fni~L)EC-N##C=dp=n+0gLHivu2<#)+4(RpFT*Bv zQOE}V=xQBZ#@U03HO4ZY%2OY)9hKG=Kf1YJR{Ls2T;9OYFJo#V=+#M0(t6iu5!Yp# zw-4-r%u4w_$bZn!f1{xPJf|LSUsS&$y7U}o8FHAn95Yg2aaLEae$uoUiLmvMjdk5U zjq-+qGT(gkivv+aZ-9su?JyfIJ)3B#%o$fMhaI&dXx5BJPJKU6fYly2{Dy=YDWqrkb6{0SgOY(0pke5->fF8A!5|pLfXKz_7 zPm3W~-b9-=N@sp&f0)j#rKE*#b{$&G9(2mpEq!X@=~!p?%tq3J%Ty5C=_!(K#&)~x zn>s_YFTV&*8&9go6}YaX*)wU}E{y~wMf*dQZB47--zZFDD+}VDOmbqob`y6`wl>`l zO^3XVx3u|kuY+Kow>*pXdzjUIayK8Ud$uQRD!1c11!dDNWO!&J%&;K|C8VLA?sL+; zfD=p4dXT<2v8=-v8+BEJ^(Dht9=bkJP*K)*mMyNYcoWITYiAdf#wr&UNJG&_T&3co zFWmWZ$?`C3$%(kkx^6*E^LCe6IxaAD5tX_TaNlfA z2(rnXS{Hw?Y|1L)s75l3X+KEbz;Syg=Yf8_tVkG7V0OOC0^woUVxivv;yUKp(rLdB zET?l;Gc9V`DaC9tX%kIWPoKrx*l=lb$W1&@Dk7cSz%b4FT}`Vc?TON0w^H@5KlMaU z1*)75bC|V0>i#i2u@Ow2wy5@1oB_r6M(f;MEKI(g@VLStY;19Pu@(Bu(eE&;P2W0Q zN=RZ97AmTZtD0|??^euit?nu|rviE=c}V}DIXIFfLk&bFvBbP1`FsguA`&T1!)Tuc zo5-?rX1Lq|p-qcJVUzAwXM!I}Q77<{4^1n*rQ39ZZWF}Ol%jw{_o-Lv$(O`NjNhUo zzbEeG`m=mzcG&y8w*(GAS`x!j#!GDXx!oLN$B$wko?IS@mha1Jqs+d;h_p6Eg=|3U ze!Ax-TA-$FuEcA3>U~rz?Ujr~mSG*NBGpr?aJsrK;UfV?{hIST?HWu&f;7r1<%?*( z;>v-C; z&&a1wKkPiyR~`a&(VbzQk<&4)eP0#jNwJC(XMRe$veZqG?>a zKF^{jpts8>d}=<;u91r7^D3ic6~%c6OA7m1=|*l2xqq zXid4dpd3By*sE?OOjfD2ye2|0@sq}=qN$6KY_9zYzZWeR$jQ(US~YJmD;#ZHSs+Lz z@^&fw>RRUBDa}3V`>+_aO9p8|McB&Yh>|@t#m(c?SbjP*`d0y@g+ou^b-y?_< zYGO!mSl0H1vfOK?D?qhZ?QeX-m;zusPUr@EZB^X7$7}$z>Qh&hkh_-fwd(Pf>`Yg}l@%ra!-@LIPHXd$w-+v`LDVWOdf#-e zWnkbH&Qqp{FvV_!>_~kS%Gll0F(mrCd&fY3e&l`yO*T}tRZ$@<9j|QUcyn`rQYFjV z*hE=PCtL{Ud2J;@b9q!W!0$LNgt-)hDsPV5FE3dWk`rqed#Gv_=56=m#DTMk7gT4E z^E3T=Tvd2el)WF)1f`3R_=^cfs3C^ljD#TxI{<9NlT3gzI#5s+CpRq? z?9w*)c-?K;qFvlL`~BkO$E-+Q#oZU{U*6N{Zo6|9%!5CBUO6nRGG4aGThvXExb!sz zKR|zc4Q-rS-OiL0!HMhX3RRnWHT~S=St8ZzgN(*)F(?pf+n=kS+)JAfkQE7Cx7--7 z4i>RA`fhC{lI6dW-gKhtR-;^~`hKaqRf1^Tfn){5mwWn^__$uEaj$;}cqYM@YKABk zI6;DtpM^(Bh@|IY^e>|vib8sqG$$=+%?QOH)x8Mi?UN=_EhT3CfhQ3r$m$=V)`UrO67a`Nhf=`C$;-1 z5;CyyBQ{elWiQ+v>mP!LVuYkWM{5jowJ71u3|t{URBt%UddiX4f%LL{j{ z5GW0A03$1&4%pl)lx^4DHsxWovpvBJ+`i!Wh@Nm9vLxb{ereJNIDFM@H(8fdi-k{~ zu<&a3S1tvWsOsh16z7zrE20+klf&qa!WB!O_{wKI4^yYUPr3_dYpHI@no9&zXgJMA zj;!zizROMk-g`-T!pH+qwnHj7BEtWqY@Y-+T>nbh=6n9^E5MaYr~Jp`MCiG#YYN-+ z6+sp3gZ88KTBAR~LcD*QGZOR1hYk7FjXa$s&+Rn1ONR+3RKj*8mn2-uQEgp;E)aO- zT&FJLJl0@z;@SHL#;CMx-y&N1CN-%bOWWRR&vzX56T$yGPp@`;%O)3+^B&vTx7fzsD?wP{gsP;1lt9L0-yyZ&d>CO?8P{N9o2;bOf58f}8 zq;v5YcO^H(aTt<5Cw1f7pm?Uc$2d|kEky^P%UhwNog}-o>2QoLcqOgJalu$Va8Neu z*i%%IS}C_2UF?z-{>2N1H`A0Vqq{FTd>PwCc}3Z{TW*V9B-ltm9lS)XiNy zn`6_5d>vdYkngXkLtGMrUVU;-V=@DbYDd&%v=?#0zYa|Ht}$%g=pt;ED?iEf#;MGgN9sqd_^%lsg@mF9Nw}SY!+?6j)gv&kno`x zo;m1kpl+zv?b&4;MnCtfz`F~VkXw~0DbwCpIc-rW-~Bzx;_^IG4_)nbbHw`7Umw@e zYho}NE40tWAq7*}h$e<|kYuVZgE`)hK)op4oD*L|VIVEUGlc{zylFD#Atq-%{adx)4yFXowT2s-Z?h;O_`8uY+u~YvXX6kv(si{BAuxYLS z3}{cS5@&9D_tPooWUVHZ>)z=5oZvi>`HEM**72IbRqmNc4cQ0hyM4}PU+6{t-j*1n zW>#;f&=|^t-@KG=C^oNEsnp}<`y9{feIvakJvAOM>N~91OlM8k7!#OP`n8m1B%M3p zmEHhTfvz{@HtlCG!|=3CW75@KM+aOb*=Uc69I52d<&tgw??vi=Kc7Cam3qS3QD;g0 zwJWPizB*k?omo1STfD7&xV<0q>z#qD#k&}!Te-Ez#ZT({WeDz%mUj94nfNn8wMY>V zSbidr)gH)(SrW^{d-kePq(DAo?CNk_EPzXJEBJeQYD}0O%sLy_*9?u9>LZ(daA}Vs zylBoCYhs)~O>omR>CH_p>Sf*w6UXh-4{ct{qBYKqF@N`)uR@3z3iA0!$jjK!*n-Q+ zn^Mw~MKKkRXB$t*&ly-@=8|-=W*12jWx~6kMJtQ^grud`=6=n44k{~3lag#A{pLyd zHmU;1{QgI(6o-cWC%nQ)%mCii%KoLxD_!AAe|*gXDX$kb29z^|<(6eDHek&Hip2k8 z%`zB)pne}pJVAYbwS*k(aJYGI^6!?AChbSrt$3i0-q#A0w%0O_3MRTr)@mLM;%b9C z3phtgt)`~GSe_L-Ch2-cif-66;vO~8v|@z`Pi&*$3`1$J*a`{i-QegaZzeq>-oIua zKIkdQBuOhAvMikV7Fa1~CLl6(XYseyr_)Z~YPKC?g8a&D-DKI z4cJYnp8$--ng#kQErqI+AI60?jt`EXERii?v^E!d)IGl zl>N0qt2On`Vc8p_b@cM2=I8WQ5AKpD_l~shl@9mk*Y(S%TT-FBs%+hQrU8lOZgOq< zp%g0Yi?j2aw*^qCC=-TbouxpQ&)$d8AgX(N@q8Y^Xo-#s5V*N*kL^6@S9_?9n^yCX z`)DE|{Xp>5VN3+9;)__ZBE~2TIb8tR13Tm)sajs-dc2Qj-Bd*a9!U<(gB>MZ_p~_; z@m1<>@*Rt)*gAmdmNW@T7G?ET`b}4=%s^Mm`uqpX3ad;5OJ%AxK(?rSPnhjxXS8*i z70-E-SP+XYU=c;zmJFS!pWV4&AGM9<<{G@#or}P40{GD5cJm|EOm3vf#?eI!4#ezQ zJ!W)Fq?7lnRcFS{b{_7X2>Idm;fD`tTCWNEXPs|a#E6ehN{pX4`IgI2#*l7H`hq>w z5ldz_Ct2Ndoo}GS&|lF?UWF%`1~~9aM(X7BXQ5-neDW#vO&uY;Ie)RSsNC1FU(k-c z)jb{gw{O}`NmU~BT(2)AT6W*=zM!+m(DOG_8OQRn*LN(lJSK%3b|Va&=GQ5K9 zf(6(HQs&76W=?Z6GgJ_*S z3CZtPdR;m~i{xXOM*cm1l$ocnGPi*Ca;x26kDjA(7fDWyk>Kr3o6dzV^25F%H%`^= z95+4pCSvB>sF2>$-I3$!<}+q`uSjJUVn`_Bd*UFRcxdvBaNT#{+_^WL zkk`ysq>g}7q%;k=Qt(yZfOYF9*ybSp#C!;Ot{EY+QUlnXZUc9W8@`~{QD zmaaJx<(SVye2)$|PvFqd3l)lW2(k=XRFGCCF>cv{GmCt_5Wg}C)INgcFVSB~~%VBZJ2;t786BK`S zmB`yexH}+kDP)EgSj~X{bRi-13x+z288Tq*>XQUzLtI!j-(GzFu^5tpBYsvmsmqU8 z=7$U^urhi(>-vD4UQE#H2?zvJT%oOJPCMRl6Z(p#|Cdpe!1KO)CrFG7D)TqjB`TrxWD%x?HI8Oi zuMK5`iCH`QR|+3S^tfD(Ej;ONsUL$g4Ev71eXT$+Mva)-u!(_Aj8rS^6t$%EfB5>| z#NLMYaqTUp+B-dB_lv`aUtxnYFNTnGwZ4NYy6=RnV;pLcONPx*q&H3yAISvx1=IG&3tTH1c<7tg#tb%XBlY(N8 zM+UaY?}WCJ`G;0++Ci;Ry)+fxD+AZghKluphmUGO(Lms_{3(x_02qOr+|+vltaSZT z!D1o2Ak_;MuzXj?STgo>xGb3=1jpnNACEi*_*GPHWIe4!;Eea2PDNSm4fW`o2b#j` zFUQTxs%nML^*>PuV=UAUXl|)la#|Y~vlE5)@L!BH9;7aDRM_c)-{fN-SKM7MTiM0n zRw`Y4>?cL#PN|5EI*WB|bl6$FI$CvjA4Rvj^su%WW2z;uB)EG{&==DA-c`5^e&8DX zJl5;Pi#aT|$5}ofiQcrj8u>}>px73*uMQqI<3!vsU7pP!8OP|${Qg*P?Ay=HO1BIR z&aXe)B1Ejclz1rHnvodMOCEh5XFcemXgQU}@aSWI=KA2JiXC2=IghYX&M_A|p76Z? zsP17>Prdb*PgAmWxbqjb0p%%)U@R-~^YxR%a#)jTO>&1+WPxU*HY(jHny=o1h8@Hw zP1UT{et%6pVrMSnIW?+(Uvtcs+$=8XJy(4;rrVR-YJm`EzgK*guyL@NwOIXGZ~%cq zl?6@O@DO96ejt5uNu-Hu@5Xrt*78R z5;=VNui!e?dLN)g%aYqCi?c^kF4krBw)mqn-?o*`M6oL3mwiqg4f$w&+Fa|R;&)Q9 zW`|?=q=70ORy~|tF*Y|+Bb<_33A3%F1aUD^)fm4?a1yv@(iy46jOrCFZfo;ar3)%apRgTQxhZ;3CHpRs&!=sp) zevCSx)UjjO@lPH3>n~P0GbU18iurH*rJ?}=>D20@ks6|m7>-%lz|*6Y`E@Tm9>9C6 z!$N@p}{`g4PGHTBe-xOC+z`NN22YK?)9 zMVNd$1FKwOGa;9h9JJw9)5?kic&UlJ9@*5%81(==lDO=WC>R8qzH2ZXMKB+=+hQB( zyn5F|`pF_7S1)r`h(gZ#8t}In5+#e z-qAU~K8T98DR#V{7r0lk-^X==eZp2t3P`k>>erW6L z1XA8k&KcxlI&p-j1J}j@5uNz;Jc-rdDN;oG2741_6yE!@m=yF{Tt}ql4IRiyvd z9tyy5Y6Y_2$GL+7!@X<+i5m;GHp}dc$GX&f=oknn?{bC zfHJ$6r^gC)7$Tllc_DNQY(lKXYyS{x>*x9~ER0ci>&!PPpkt0M4)ZX(nXe!s=(9iK z;rqK!w-0shYaf1-Uw*KYIh)oM;qdQG=08ltX&gcA5tiE9I_viF^7}r9j;RNwqRcYW z$Mf!U7R$hvq_m)28ysW7HG`g+ z-cOnH!z=_mv2WrY->KPHJ~~<$6w=+H1C~Fwc0TJeW%Bt4Y#Fhbc00xO@@luS>wr*% z#WL1A*RB6;H}5qFS4S;BzT)5wpP|a2f}Hf&!h>5iT1zYoG1H|x%(bYM`@E#s{n9Gj zmsOb^^(UMJU1*6g>V2l?JKglqiq4>N9o3f%4A@PBja6kdzVw}PDo?-P;3hh>)h-c( zTx(vNCzsN%Wll@VUQ`Ghxj$#Q-a8{GKAxv|-2ZTD%Vt)CQ1LEgi~(^TDU>3GJfNLr z=0>Pn)O4XFdMlh!*eS02IPSu1&n_9h%{eC|b)}@iQ^}(bKAU01Or@WLA}JM=iguI+ zCMu&#Q@?Yu>;kbtN6}OSzPLOVFZM7UFWd}$>0f;mK(T6@02B`aY}Edwfa$QMvU964 z6`i%R`I%E1ybUk>+7*n!ll+`6`W6TvU$rq!oHP9%xfRfLmCE{VqI@-ffAEut!ESvpJZzNQFI)+ThXup) zF^*1vX5UcRdn5P1JT(Zl zcy3OOI$uhK-2HJ6RxkMCDJvU1^z!%xxlvGY6J1@BO&m%25>a+1c~4Pr5U(!D&RzYC zub$n=_hqqK>1J1_>8|HZ41iR1IVt)pj-(S{jm(fPP+bL!a60=O>^u$B0sSoYv~W+d zl5{xUzHznHX{i*{6Akqei!{U4h9O@sfN}UzARTuxopFe8Fpf>&!#~>?YA-#N zyjr%SUZrdGy|E>We7WUsu5Q-S0RVe-mDoY4^Tx03;MYDYWxGH1(FlqVSI{e zmNiQj1jJ8@gnMG^l;ue{IsFe@=>A8)b%9JU=iWAvJrK~Ui!-{0t6vSD2w@W)d#EVp zHS}sRRI>2Qsn{HfvgI{pE#5Qt!ZyD%GDnYO30vM=!Jk>VKWlw?TztdeCZRj%TfLLi zLV5stshUhlyj~tXKN4L1I=G><^W^e(%Y*ud&fg3;af4QFA3RszA1wKGg@YVmek;}X z<+r&E`t!A5jbV+-K3IljmE1{knWL5tFojkDt{+516KEPy(%uclyZCw`l{!T}*h}Yu zxA195#r1PHf?3=N_Gk?2xWBT&FiGmB-!!FN<*lheUf>A=};sHw?B`HBf@Er8H{r%s#yLxRlD|@38TL_iv)O{7uP!88>T`a=XxLF%V zzte1ad9U0V+L1xDB(aqEE8&%2Qcq`*ycjh3k4;^iQlM#09PgDo!+@I25N|!iq{xw( z3n%6jnoPGtfHmb*w+`dN(eb#9BNL5xr*L2;+ zbe2}#5KHBQ3F#CE9pU4N`M7@7P^6ibSa4&7x9190+z8S7jjMP*yZ+23(1Iqk#$?NU z8j4CY7{&ITYnNe44WZtIhhD`K<%XFh6@UM_DwwjU{aB|aHI&Na^)!zPu2HC#IV@rv zmMR*^7tJ6Q%vMLf&F?!VVh=_H6m!Ma<5wm#CCRQ)q}X&Xo~B2J#vAac2kz29&!kX# zgE@wPLYw^YW@4E=09;L#CAmHfBE=(^NTS3jz($^P$iHA0{E5-KPLr`re1h+H&61m_ z=>MjtZ0YG*wreDz16a{2k0fTXST>cb8UE;pQLf#s$O5mwzqYe46V;H^B5Ixn_kI-# z&36{#m0i7U&S8o#S(haxKi6&c6NGeOTc!KC>Bt51n7cW-p>RKS+^UppGoKWDp;c4M{$lZft&iwR=5lsq2xs(%{{1OAwG5sm*uU< z>yUEUzJoQ-)3}Fy*OxytZf(hNma$wE1Wl*C9CY-IuPJL=I8I;({|V4b_Du9o_WV(U zjxp||7hMDw+&__a6;K$TGim1WGm$tHq++%m3XUCzzGW!oTV`GfR=MCa9|?uO0S=4x zXI|oX^0-Ka-;}uc7VJm!)_F?F6r$i0jG_o;N-Ys}H|vj8;9-91@ad(z=9~@!6c3n^ z{EPtFqv%ghKO%+{Jj5}KI0wWx@8g|8ZDmDrVv*VERG&pMy2~h%J@{o`c|r(2E<4>s ztpOUlJW>uHkMyKag|AwRBy^UaH1zKOF(ydu2!3;azVhKpX^x1U(QgGTk|)%4uqY1;R-$|O9OWXtmEhJ?hGJQtO5zB6wJxn%kq{qj9O zclo3rlfiE7j~->6t9J)}#0n!@m+Spkq-@Qybb%ImTvWR8D@g0{-dOJq=u3Y3RWQ36 zGtMA2!Pldw{;D;}B)aNtl(<^S7{$rLdsO9k5^!DM@d&Z?r1J$8_KS_q&h`hT3Dl2* z>n56$Dvq`qg-PUW0;=N`584RDycuI6%d!Y$n^-dC0F8*?Y(rH-@Kv?am{i5-cxgX} zva&>lSe0ZG)vWX;%D2Z;9RvuC*|Hk%^tKfGvOLUE)G~JH3ye|b^ieC^(qY_1sH228 z$ebu}B8gmd$ONUSwqgLPZpB5;IRj*UiAX5SqRKJE*`t+mqp2+F0F5 zb*CDu`lWh=t7G%I4l3jBS7dHU@{w~1Y3K)P95F}MiuDVj@)Wi}=8UKbs)2!%mB(ui+$Ys^9YY^^pdH%iu}q3g zBZalHaA`rXEh|pa4gRIGmv7lypa`=kV%Bl;ewp9Oo_(ggP+CxdHQAgG%YUw;dvG&f z+00HM-{cZ^zu42GQS)^tT$$sjSNQcx%6-R5;a}f|NT0@)YgMEMj7K26_SV3n$zrHf z-w3a;CKx8gT^YpZWp{Nn@yj`^pH|sJ_M;arS^dtwy<9pywO^^p&b9m8dpK~+K0T|u zU}#9sloiPlT>3#?{L6qKxev|o?4t?3RBXl=%Bf`2<|?W`N9t5rkrQ1$4sDcgZGAJn zYr(T72F*)L9y`!Tg0sl4E?3FDu2qYryrl6@B#s<-`n0Ej3NRxUwoFYpE-@OwVPOU? zvUeo=htSWMTAo%g2h%JfFBsszc>2FEK(AZwoV{Kj@${Tl&lQ|8zg7C0mrf10d@{Q# zaoGOLZKmRv1=wmoYr#3DS@V!>-Chbq&W-sA{?Z5<9ibCEYF)8M+wtvLHv1?HZ}rps z@owk*E!>CKv4U%?1>lW^PXgtqJ+&SZ?4Cx4+-17#DSdgtrLcJ@OXVzRHu()~`EV<8 zAr)4%p#gNU-hZc{3YPaz>X+5%9#%IHlzc=2(nCnMZS)hoo> zAw?vZ0f;r8bDn@YUsf#=BV!S%Uc{TPX!0N`N+}$Zofzc&mGs@Reu{%@qa#;xRaGp6 za`9No5-#bZC-dm=$3Nd}MW4)z9cP?TC zwys{rMX1B)WGDT0c~0*2E?Da&qC-_S-uvbaW1q}=qex-B?azluT(3J;+PA#;^K~nq zA8cb}XLYviD+tEQ3rkvF*%w}6OGZnc>y?q8v><>yjMZn1H!++e**p71(maC8tX36m zIv07k8L|+{>U1^tNFFzch)EYi^AR*C6ql5z;YZ`=_Q%nM)Pk@WE0%q8~hw^ zN#JG{CpLq*!%W87%5+Sg$H?fmq}I#Xa~rP~rhV;V8}m;063)!E(z!?nwOtw}tu*?@ zog4AsXbk+NI_-5mC;i(7cV&gD(Bz#OCnu-X$ATO>B8BvsHc7s#p$5O?#TFIStwaey zzABJDb$sO0le3~Qw`9ub4XGO}~Dn$Ihv-lh7iuhQ1008qqG z&HvEqJ5%g-jkUWc?j#F@hmpK1WHDyNHXb?r&JJI1!u;0Kn{uVRzzTZW;`>LYk&$_B zmR){d$Bvt}FBQ-?%nOR%jJL4|V?H7->v8v%OlHX3DAYB@b}12sOZx{FG)X8Q&(N|} z|6{QjY!Z9|W97=j3A#;V@gtS_`rp=TV(mJ&;Ves-d`v$Em<&7yf5IGYiQV)2+fGN!I^4h7T-Mc5fAR6CJ|LLH9bw<-Wn)g-=WrgpS;@eDgvk+)AT?lO5W`DBl6BCnb z#=7m{)svV~e@^fuZg0JHU*w^_hUzYgyW8*TE`QjONtV1*9g?%R+n6SHU1;o4cgiDGMM1US zme<<+EPZmT`Oo#x;BF(VT~Zf8s_D;iPM%$fy8aiF(SzcYChKW0%y!`Q8iI+c-OE6` zwPY97((lRjF(hBtO*GF$<`fBzc)j(qMCIM<#DGmpUD?#t)|H2k+lE+-ZjSdpO7qu# zjODWVjA?1%$M?q&vAI`HpM9OTYngB5QgDW? z1esuo>13WK`pl)iR`rwbYcXrl2qVp;>wn2oVI40Qvg_}N8ljxh=MOu)Fd8TLC1``k zYG@+a&8EGbVd(#9P6wSZdnw$?&h#1;EWVUqPpKiyHni}0^L#mz^wtdTZPP6D>TVEA zzDY<>K@qyRI?~Tz$;cR!YsVBj!h=QVM?%Qy6!+)psO~kU$N+`|=|cmj=pFz>j35Q7 zwHV0GBXg07u6fqO?@N~SxV3sBxK`ai7?qh$;adX*tYaANFK}yJCY@2zH}^__j8N+g z^+x$b`(Y9a7MjChB1BG4c4s#5AX8i$8x)(CHjzefwArHsWmAq^6pXyjJae!I$O=YH zm=<$O=YKJaoqcoxM>Bfjqa4#qWrWu&O|jN+5_)Uf-X1ZSyAMi!=W#73H711~KGaDy z@%Ql@EKUQ@&0Jl}NI_LTy&&Qr{2goKZxEmZUPM)_p!?d8Vac<~rhIkn%{Om=<^vu+b5nU z1=K7j?}2%-j(EF~^^mN>i!ajD)B)ubd7Cm#*6jd^8h2347zcyN%v zsGr~&yD%1?^GapSH%NP`dbm#AFrEaIum6Y%bnNiv4CXOo&T!%q_#vQJzqFX3WGUSOfvcmwWes zFM$XZdKEB92ZCBjd1@4u7L&7`vPRoR1T5l;;Iimgte<%_0x2Rz`cR}++;4n^i!uOx zqllhRfx8;My$AY>X*qY}-9gV^OolRdUfuo;kYQopZv3cgySALPz6H3?vQqXc)SvH0 zc#xk0AL7$ilGVQiAaNnRk;SHUXdL!y>T$Xh8W9*#AcCqxn>w%soxJyX?es{Ab_f~~ zp>q0t5m`eJ_vHIQMdcp)nnupOV?CuiW1|Rg*X3v)c8~?V8`TCKLpy9bAMYCdOxwA< zP9|*K{}fp~UvwA10KIiYbjgIZdtAU~u#|~L|BGpY4zl`sYf(b_-hL7qP#kXNZ)G>} z)xDn}K=nBpzst&!>(I6?zu|C}0;Dh!^N-;!l`BkG_YmMo^OpoYHP8sBcAVt~gq*Mwm2nYw;^QzP4E`IA zyf_bkg#hRMmSuBKMI-!-{p0@pSRrt+mW_^_oLPW{jf}e;duZ+1-BCg z;hzKW-+_RLgp7iUhK_-W1;3%?B>)it2?-Gy2?Yfi8Gd&#{5k*`4+WowTMCsx(*ljw zoscIqsQ{f$x}g`SHG4tNYv~b&fl2g=n1qypk%^gwl@H7>ASfg(BP%Dbps1v*tpm~3 z(>E}*vbM3cvv+Xx^z!!c_45x1e;*MU6&(XjPDxEm&&bSz6&4kjl$MoOR5ms>x3spk zcYNyW9~c}O9vK~*n_pO5T3-3Oy1lczw|{W>{pk4e>iXvP?&tl(ufMnu07(DBg8%&& z*#ChG4~`2F85s!~?Jq6_L|^zX5*{)N4L2&jlqQ;mI{__EC_15ZQb9v61|6@~1<=xC z7L$mcZ=2!rFSLJ<{l5bi_Wujn{{s8JajgQdkr3d)L&5_{0PZO4h62!{e$0*QeEBWb zKg6JOWgS#OkCdh*Tv1uH!x^E7?D#q4Xw&S8v1@sk^%*c|W!vL^(PQoYVw3$0h_1s5)lX^Hazx~{p|+VqRiI1_ z)C$$7a7I-6`YxY4L-4^rHrjj6W^WVpdo`{CSkcS9rcBB@9(_JMl#~y~J z1ZkT$qrBIxxnBSIVL0|V@|-1T!e)=G>Euj3D`hO9JbPz3bY|-$=(piS$eprE!iQY6 z78`<-KrHNp8>Oib_!x%lj{4E!uVw4XMHdnk_)Ffn)_RPHe@CjTFD?-|){YhO3dqx@ z`tTMNQ)yX{B4n0v_ZdK5VSZnqX;)(4$tjdxp}LWhC*23h>VfH<^@J zrHGxIu!ua3sQ{ZbWsdoxCw_h~IlDg}#m{92&Ii)4!k0|ahU4++;*y`Pwec)T^Sl&c0P#Z!?hj`p4?YONBZIn ze)47(vxREklv*A2Xt!;W=Rc1gkvYf zopU9niT9y1K2^%H-P$K4Z&~hH$$u)hX@zlvyWVedrL|MZLW z*h}U6Eho{f1S}iKrehQ zDXxtk*qO#-wOxXo0!w8rsCe#bfU;jzSAKHdE8gU_1%EiRIwPK%X8)4Qt-W!)#7}e( zm^I)Yn(UME7!mY?Du&}3(1LRBJQ5PwNmf;3W;E9?OsFLUEObaN4M_i3cLZxTwMm;v z6Z7GD25_;m%M6n8r}nzJ$y(n~8ns=acBv~b{qS;%3Vku2D^=cW9!RYtd+o(rXhff=D_w9fNe$XOA^w||k zDAy$KVt*#w8;HYi%Wr#9QGu(}oWQ^|I;V*l)!i99Sa_zCGpq&GMhE>{T5+};p{h;a zN1o{Y)U4hwHin%1)zIn$azF1V(u60;lxILpU3T-)h{cJC&4*V=d}W53E48+PboVFy z4nJGSL)I`P?qMIEh^Qh$zO(#~mylisrk*FbQ@s@Bke9E3t0)89 zUWXS05?Guj+Rp%nefX)<0CjB2B*O|sCcdz0O=KnY>si7jsE_4mi*v|Y*Oy0CIk;vk z^nSS^eg=#he7T1{9QX_1v(optIzFrj<;f?+Y`(A>stvgya(kjJGlToeBB|2-8DM(# z*Hzc`Nw1J02`!u%y$2~3G-aOOjCR!Dr*0j8d*J+=yPA1WN?g-V!%r%za9`<7d#J=e zRBAOaK?(P8@)W3`-B=Xfb&qL(CAGv1`WD+WVhOGmu1z7Az-xuKj%T&&I~9|K{#iFh zL7Eqi!Dr?PzyFhO_@iRryni39%PWpmvq4-+1-HGXEf4oeXae+EQ!Zn^!!c?RqiKLZ}+&~^?(qfFp-Ffjd% z{~@n6fC-rAQ=op;(=d93TtF61u!Wb7$nt-)^*@xT248aa2|J;5E_OA_OaA67 zdteXh-=6YbbN^xVR%qu;;WxM#(x0{R)9l{(Eg~A#+P4QL33#?>mkoxXC{bm?mh_fS z|AM|!2|nIv5u7LaT+w1t=6MuyZ}b)oo@|Hz0?|r<*C8?Rm;Yd=!1U~Ps(V2w2Th%? zc*Wrv5K*_4_X{Uv@2^f$U9>zBJOiZu^KV_|mdin@GiI(2`NS_9aTY)D(|JSQzwRt? zgNoy*f3CKA+OYixGRWa22aXc|vKk7f{ovGeLCCh#J=0UT*#k%bZnhc~P?Tunf4!fr zOKnE@fd1Z&(DSeJ%`Td{sNnM^Jf{CsX`caoa~Jl4PhSt8KL10g1D`L`gbEzH$qv_w zK@-2MgETJOkE%oVn&C!^!YfB82`QXh|0K*Lh-Dp6W3Q7rB*(-bo|PvFA8hO*;Sb7} zD?>P$VBZLWs6Ts${5w+}fP-*>lUk@UHXA=%JlqVPdRC#;@Hm>z!(B^rpR&zBB3wWG zC-9%{VlsulzujGvY?uyt(OBr~P5J%Dt?Oq1_kU)WNDy>U3LifM9@zgvHsc?rp{o() zC93eShubXhM3fL7@|_#bO8D4>+m|A}houewo&m4Jac^!XIczy>gGt31?qzQlp8@}v z`w>nxs=kmLAe?IcA@{I{172N7Ba0B+YuzDzqZb+fgX{m$nC0oKm!zqJTfemKVvP@H zPIjkpgwIVBlm25NU;iaq-?5@_Df%;@^dAM=@mWnpWaCFbY&v(H-DeY-JtKbj`on7H zRYZ0vJWKv*p7;a?II*b59W$Q+ZGVlqU_R%aEB42Nn$AhXWlsJ+Ya8XH3eJhYVWGay z-nyRGgWK%@mxv3!4|FB`f`p3|aLd1N4%A_A?dtx$1|P;L==DWd@Y({dVC5M? zrh5*=6F3q0a9_67yKL~E*1UrAlz7WY2yVhZ(sO(B#GC8b{;uyUpC=mkjUPN@j7L== zdw+EbZAb6Df{Ag+mMHur=o`bVhGgZ?f&3J&Tz}>-qbH(}@TO+~&Hp3mW-nab|G+={ z8zJDev(2XzWvPwHSGcyp%;JoGlCg1S3BT?#;L1}hLGVZ}d_whKq(xZj;zk|AEg)UW&v#`0VHf-dnhV(67n7aRIflv9D%dA1F z7vX<4%p~s4{>dONx+{?3m}aOZixU39bm8pWrw*SO>kMu6ktGXUW)qgkacU-iWbE{0s0!fDg>!rJ{7%(O})cgzxK%V|Fi z?`I++ww>*MYbX5r33o|MHLT3^l0y@1FHkO z`LZ2>2tw=QFaOpk=|_c&>^~cN5_doTmMk+q@%%w978fw8!b2SST)j}!QNq3cTi}#L zi^M+`{ADpxKk@Wk9@Xu?Pl$v#RM5dbqx|~xP92z1#WC<(iShtGR1ghG|M9)V?LYCC zeuSrk*CJeET>mEImY#ibw^Xs>dw1Kq)y@~Iyx;LIsru|n;f(v2o;80b$FL?)j}VK&Fn6T$Tf zru;YKjh_KqeDG*WaSkRR_qj8+G{eJN{5QP#6P->RcQm}VskUni#;3o;-g5Ynd~(Wy zGgIjqa1Tukf?Q}m12UKYEkt_bHlpd1CJNYTYBF%iKZMuUC=;ju=4Of7`4&2C;s!ru z{98>93a-hu_nO!#<;b_Eri-Qrp8<%RCcf~?f9op?+7J~&F6_sXvijc^VO*={_?~(W z88pROD{gh%a^JEQc~k?Jp}!?f5ni^qZk9r>4B=Vdt!u!qbK>9j8Pwp}bZmBS(ES2l zDSMs)J%4q`>QVE;*^0B;JY>0g_!(ebmx){GxF^8EW6_|oPmk+DB_{x%x1!&vUE=oe zZ!MO*AKc=dH-QuVAHWac8qpX$Ee5ZEj(Tz|#=pvgOfOiDYD4~~MR*2;qCF0WTxo^C zEiqPh!YtOjWqRazf$$8N`5$P^a9Z_l$-s-sx5v~xxYE?6w?c#5!+HW|=#IE57_0X)S1? z&h*p{w{l?N%LVK{@F|Sy0Tup&-9^f8p9e0&InF_}htLm|emys8*L^dhquaeP-; zYslH^U&-KjQtiR4Lt-u?X{|N(7CTUn>3$h)bL$~-_E&)l{%Jkn1&RBAx((C?=z{_9 z1EDIjSjBZrC+`u`rDsonZ3jHPQ{bH=VffCE(9A8TD7ZlV%>vLNVS`A$gDQN*A&;vg zp+sSj3bpZ^beQU2^1OC~ZemKMFE#v6bC|*v&xTpzH9Sf`uNvl(rws24HZM5b|NQUx zA@Fm~KWGyF21w_6=+v^!+f}hj|^e4dG?JAYNovHmQBi0NtTa1~0$*n9Pn}})?ID~P@EDV9{K!GV$0I`+2D+-$KUs>C ztawcmm!}gK*V|kpj`SpB;naTFT`J4$Hsb;L6H!2tQZrpsQ83+8)?iDK^6`uYZ;u5e z;xv+%g~`|WAK*8Tj?nVTrD^3)qY5|*w`|BJXOJ&thSj`J+way6Fw+* zS}X2~#EQ4jp&m^50!erB>aZE8I2a*JGbK^jzO7Nxbn$MtK91zkTuXw@j6|xu|CB4Z z6`Mp%#laoJZ}uO2a(H7BQ%)-V28Hi}#Y6aKWe06|I?0*h@WCL-P5HQ`nmwZV z%4^EfX0Pmd%elP^NxGFKgNl67c8LLQX)MYZs~vPophGq_LPzKk>^q*8em@H`*9hHb zf6&x87O-bsI>&d==Lvtw^q8pC%VNJ)hQN5Sg1)0rZH>&A>s=NSaM~2v8PN@>PuX$G@1k?!FnaG_6@8yj}W zzK_1ORF-3%Q1ki+%^=Lo;Tn}!XWxc%QqO?j8;m@U@4+0a%IRg8ylwe9e9~@QD3#Lj z3Br)GgGG5t=5E#)cXk}8$V{i&Zfub`d z2eJ}K#;r(C;W!$H?Ek|c5%slgwGt?E1hpi;L;Ap_@g>XMvYfv4qENVh2j49*No8X_ z>6ss~72o$fU&9eShRRc1ot-B8jEW+%h=ArRoHQw8MAkh?Md6t!BArj!&UU~lV`JAu z=(h!ctmA?0>ux8-Vnc8jpJ$!XA1B{oChwhRK&SrOS7!Hl@DJ>U(ViFqAx%a->Q*rl z0c|d*Uai%<3rn}r?4>REEB2E*8lK8}Oi+4xEeF~Y{bc1>tFkO#+!pLI;L8-m57?iv z-?x+0jA-x~fX5ZwTa{PB;I(p4b_euIIl|Hrt01+MEMJv-f3pQW)t1kheBtsOfT>_k z2FoRSjlC3?=CKI1wb?LXjtNRg?-;SxU!!fI{yb*Bq}9pFqf5SYs)|w8M81>DkpX^7 zR{V7}gD(lsWI|+3I-K~R$?$8v8mOe*CSAF9280We)4&$-kl9)a>uYVk2aN0il?0IyQMpz_8 zB}MKII`&l(zEDoO*(Ef!yeUs4wMSE|GWf~8XRZ4|Ua@kHS##15iNuBCTZ#VD6b+A@ zecwn;F6%(@(C8pPDn>NTAYYj%o}u526|@4p zH9ankMd<3Hj18Tl2fcJMsFnQ?zSLyf%E>4-sLi^UlpkbS+mA(O*J5^YmzhFw&>yvFZ#gy<8 zPO@J2$C+Jl7X6h#E$i(KbTVnQ^0-nDHp^g9$O*pNg#~wIn;jgqvGLJ(<6f!SZR-C? z-gWydK!Gad&aaaQ0@oTCK^Kz2YX(gNLOq52$g<0IIk44vZ!QfM@HFwLV3qDHAqn9= z;mp^HikHXwqyvC+!74o-{&fP)zRp1m14Qz;h?3O_M-~~Hos!7cb&N1cQ4Zm`H>63K zP408Be)4&%1_?D!UCY$+6N9(z?9Gr&B+|(Drbd%>lV(PF!W`WCbJAaTs@H4?NGcq6 z>JQ$z(ZiAfNLI8J>uTHgd@HWvZkh8<_4;$4RvS|EBe(qTQR|G2QDd)-ZyJHwZ^VxC zLn-*EM~Vp5323bsNFmc%?*jCs*fU|+3Cc!Xb_^`o{_s^C!%t1z9lz#jTW!HE`5mMU zjLrCO8m`q6)CD)7)~&UI!$6!;WVR~aJxsku6k*7RK$kLD7cDjyi;%u*PRnFRLVo2t z>zz*>%@@ahwQmD$!>oo$YUt>PvdqS*HR9-9dMiwN_@Gji%myUIRA9L<)*e%I7JShbBCL;*VaFFsUD=W+`leK5axUw#1MbFP zy)G;EvE&inLD{8XcvoPaZHeq|d~0IKM&t53cgg`=gIVH6A+mS=pwSAUe05ENXME zMV<*lPlL8)`c{PEBv%&N0hZqri$mk)n3GLD&bBQ@uK`LGiE=j}b&Rhm`fZp~%I{RY zd|#J}LYRwGzwW%>h+$nv#!vG&`DCxLF= zTAgSrfQH4oKD0VV(g#>J+b{B*wMX0M@g(r2kUNLd=b0^4sUtFrK!!)Je&=axtUbhe zn8e7@?^2*hCwTl2GHA@V4qwNN+OvmkeDEf%nNFWW8AqY31?i%;nlK8M#OXvZPp=E^ z*6T-juD7>v1 zw?g-M27F#^YyQJS!NCL-{egz9-lBDHRZa}f+{I~5#&s%QFz0ss>69zludf@R?_!^2 zBBhz4Q)Y7NI1wwIq(h&F6RRhx{oP7z5dEdL{?<;W?MUU7YW=mItvU-^6fG z@`)xI44%2SlQB`Ybs*a_JgBziKwMTYI&O5ZPR!D+icK{w`r;m>SpW_2-~dSDC0leR*=#^V_YnqOKG>)ZXXTM-yJ*1oUl%THb+Bv5Gs-hWEDHy z^@>ticPcqI`Ko)>JT=GYTZj&r`7C}|g4bT(F~DZ0zdLm^GCET< z6?M69Dy3~gns&sUguO+gfj&-+Qdk(dqg0V!G=gzh<&V1IR*U7Bt!_!8uR;^ooIg!U(fpL_(F|Z?-VugS%f^|?}T_AkJrQXsNe0vmOh8^F%eJ7A%@F*oXLTfe!U!Ir6DJs zrNh!Oemp4Q2A;Ezk^r?2W#S!AuYu*u_}H-G-?7nZ;|mKHn_R9f5g!6c_ z!6k~vApWb}l-8z*^;kzuYWGRQ_E8!i(G=Yxk$4n4zAyZX;3^wDyfBAGl?M6B4lySY z&31Hj3P+U>pQFsHRla^`5ZVb#k~iwWZbx1m30Le|&|NyAa|P!EEr|&`5{<)0=@_#{ z-{f1vx{5q*sscFT*v|_S>K`0E)!mI(XZR`xSnugX7bnfsJp|4rn2$_^i+yzc0ob@8 zN~Os;;iS>p%>E66fE>{_w)Yi+6*V^aoJ@SM&5vyOXb=%q-DVUG(w(Z{OrD#iV;$Jx z`QepKkErgza)Cw#9eBd=MYpT4ZBuE>g3_zqHGZKA(MWpMYgnCS&AVVw+04!Grw#*R zio)iN72BUtH^1{55cSBtz|;4UV}7*;9|M;Qji!Nd^?Kx?gH}#>!{szKz(v-#`-#Ta z{7ePq40OPna)V5II^t?jp$=?MGO8CXi%AWnD`E}6)@3hQ&RA+L`i!Hjhbnx$$WJ;g zN#_h1J+vc}(+)J??u5|=I-;k(ch}^p%uQX?Lq>w>ja@m(a(l31WD{O&!K^=YmT)H( z0sO0cuH){3rXrTN9-9-z;b4xQ@d8r0E$yAK*Sjuet?1Fby2uSktBr>Z0u8xdedEc( ziz;zKMdX-A$RpbE@P)!dKsn(q<4Wz6>9)0&ydO|Q+9ol;w&ftUgC6Sq^4%_U18iCA zctq-_Q3L&c*@(}*+Sq(IgyQ6#kf(f*vJg3Cr!flqpqQ2EuFJVAa!*p#`ZAPKc`4aP zrUS$MWpBB48jgNvM6m>O^@F+JQVBg0OoLe^M*y|4mdR#h8tkTmYDYEp86>7`Nlp5w#=o)U!SHfH+!8PNAM(M_j4z%{u zbkL5mK64(o!7-VH#O^#i(!anzk+>F{G*_wmwh~TMcp6z2Cpy|I`s8t{og|+!6|fj+y_;IZ}#>mS*4SeeOXiVC=f^!r-*RM zYHT<0hr2rWeh2S``zd%}yje zcfQ$0%m0M=AkOWLQKp$)I`rF0ZWy%IvLnInzB6fu%u@JIT*!?;9Z&pLO#NodF%=P4 zkg9XpQxJR=Unbz1#3d%N9++BGWf=&=d4p}9GN>DiaWHIH#FSYEEVDq6!*%4BcJ5B} zH-E`0pBj3kEq7g_+kOrr+jy^7mYGIdslbc6H=l{`U(O&Bpj*ipZ>XQ^l#o$lYt}Tv zUW{G&a*OhT97gdqp@mKoJrNCJxH-U2!n3t^M#Og$DijD}vDDQR4b2 z_BVTTV25bz`Q4z>orBR&XG}C|Lo!emSA^T$UtLN32Fy4D4+Z&3S5&b?6f!t8MG)1IVA&%nQ%MwQckS8j6Ij!0I50j3JJC{v%Wc{72~9*A zQRXwlr9i7z#n81wass)74|L)d$M(CFYl7t96V}{Ipc$JM=QQosuO%nIWRlj13T}(p z`7Ic4X5V#h^cDG{|ITO?7T1~vUr$4UfP1O@E6W7}LDbT;57?L$qAvA0x-mp?I6@18 zLc}%S#FR0&3p)Mw0Ho_}9d>FhKw1^~rh0eHmUlv54=Kw<3;&$k=-I)U2H+X(sA2gOuo+Zg=Rf$@Jm36r-7%U^lfd`(CT7I zZ@i#^L6jCKn%B+E%5Y5V!wL*|O+T(%txD*$22MyWf|@5w_oryga(a62jqT~G4on!w zUTRmDjqfdztg$%^pbfWlFc!j8>mn*W)JERCAdKNS7_F+G6g2)>)d#pdXdD>hhIRmt zqw7^|d)ei^0tbDU$Q*3N==2qIZAX`+kG76}2m zZ9giD&<@109I%Ya@v~GdvlTqPzsGmSWq!FI%e`8!Lao zw^E(EvulYo8NeCY#g1ja@h|dq>EaKGu0D)Y4)my#Ww!P?h94@I0WM*QEh@LC#@yb(MkT;XjlDO z$x3l^oJq3|Y5g1sTJ-It7G9OL><1NZx5#xAxsnQ9@}IRyHysR76OfUUSjf69HvWU< zr!!8IRwIT;&*t)4M<*iVEvwanm+fZ;N(c+%JZ(n&kT$@3wff(zES` zD4)=F{iY{N3rs*|-=c@h*_KB&jSxP`Hw}k9Dmf3SH_G>TNRQzJ?D6)yzf5KU z&Rn4jRUn{!_O6@@hZUf_H5?Yp5bbxQ2>U>e;(pR7I zX*X?CtAa7L3T&(UY~T%AB%i%SCsb*H8!~pdRJJCclN_u{r!>m%;p;BMRzd$(%(5&vLCmjPU3Th zXQD9XZCxzU%vYz`2LMw}b`Iv1&@RJhHDPMn(5|h5Yoz?3)2hA**O0poTaO-a&hRy~ zu+Zg(H;Fz#1HcWWlcMBLnpAd^GHLGd<#%FuTU?d8-(;Gv>#|O~(`M6SIMPLJz0_bO zFCzPWT{q%m8%zj@o}_Nr=_H)LwWxeY#)buny(#0;Ofalr|1sHHZh8XBSuh90!`nIJn^w*436DkD~{YY4G`} z9=FT_9Zl{{b)PgQ4edp~R99*7;GIi$K}fc+Z?_p#$0-zCcbv+<`~=8>O}#3`YhL8v z?C&Vpr^QClP)GMZ(lOH>N6>B=<=WF~l4jm;=uK>~L{4T=*Wxv5`i6i9OtjUtRCOR4 zCK8znIiFQJr~&FS|Ez<$lZdJ~D?C2bJw@IQ1ZTWCg6{(uXe=7=c@X`R1aE+ALDRCL zI}KG#fk?^+VjH>3XyiJxj}1FD>Z~d~N02PsO;%1X6_Y6Tt!T?40jVoT0tD)0M^R+K zNicqTI77}kd=Hd>=uOpYsD~5dS+^xo4Jh0uGK*X1Fe}Rd(t21>iX9hW3g|grl5KY9e`v$^f=h@{DVY>Z^u3s1~pIa-p^7;2mB~&KszIE zeAmsBH;%!|54Fn!S=ZlKgdzx%UDuPwzx7sKw%c^5L1)-6$UXl!agO`eZ}7V6GDuSR z0omM}Es1?83OXc(kS}4iTK-vH51JE=i4z6&o@bC$e!FDwGeDoK?nw8C^W$&k!U!k{ z{aT({R=Lep%^8zC1M-`;RL`IBtaG&Lnllsga&Bb8(T1ue>Z;ux<(syh$vgHzBSG~) z(!}2FQl&AcJq@+G%VWOFWD`P8(c*$_F3yBzWsp#C1!~aDZK(5nU1f71Zd2_IJr9J1 zEmVz7UG4A*dH$IRfW;sBjNY@3lZO6S%JU_6<9`XfmX7Q0UdU41m}g%y%GN)WTv|p$ z4j&m7lOQdqaQo7}z=2(q96KWv2f7F7$V4IZ1u zB>~(qh4ft%RY!lQK4NecXd~N5Cja8Gz`VQ7JhN-3h{shaBpW;^xKk>MYQ9hVbT#r@ zA3?Ahl~cKFAECf^79Olc>pV+SC2lcFhz+IhW9P$XK0+Hes0=GzKN_*_1+&;_VGvg~ z%72{#c|qAh_NqYdS{|zGDWqKQf7P$E&RQoI*PW~FPdEE_T|#UFRfyF4~JO};l<(v|DNKF z_gZO+B!r6{%&O6CcVW| zr^!95m5k;aOmL+?n{C0jCkL;Gd;*Pumb!6kUC+@x3Y(4I`PE8x({+sgFu>v1j;INV z77pG>?_iiOxT2Q;VIH_}Akc|{BCA~(SL7!+@y1g0T}||_I#7CN;7ve1~PTM5s&TnVTkgrf)A@Z35dTRcN4ET(e_`7n`#vtvgzEhuOpp8-B(f_w}Za*6{o##-7ZZ;9OBXrWw4Vk!Z>aTy68>(*^0D+9#AEK|PzQyXl2K4I(nY!4 z;+j(8Dd_DGulI`|YDSx->J=*`s{Ls-@{i(|?iV$`AGY7~c;D52gD*q^>AuXvmU@zn z&0B@_V8;Og0|T{Oj%gnY9Jr@IurVzjZTFdyw+tOQBj0?guzaSKzq{51wGW}lI5{#T zvYb0iZ_JF2?Fqky*_`#%ysE!?Mcc}L)9XfJGp#()-pGVpV-=K#uXDt5a z3ziW*ib-3)szG1j_m-O^k}}I~0Ur%2-5U~oMb-Ht+?#V*yLHGng;bTKk{S66Ay{1HPZ9Qr*-45v82*vwXEXm+HA#V2~y90ZnXzb^Ia zFw_+d-`N;}1(G@cYVj{w%07@w1Fi)rw;LtAWL!eAd~L1A1Y9M)gn4ZgX2t3K(C;C2 zewP!U3hpp%>6)Y4W1gOCIslCL4L8)J*4j#y@E(C#HIwJSU$S!J)b=+*dEo_{MpiABKeqnu2M$EN}ER5@Cha`12 zvz+vuvu9;#KSc(rhrbaf`h_T?l>8=Sr@NXM&snTU7O_|B!IS(hqFc!utCPD~_{pWH z?}=ND{*oLod~A67$kd7pfbUd$d6)0~;?;WN<5-kAc+5oJhaemxXNW#j)b$lvEkE3l z|MxyMO|7dEr2sfYWig_wlK#j9SD0YTYiG<)JJ5PjuggBq#mgrk;`2s65*qaI7Nljv z7XI11t%;%PW%f>A<@)4jHOIx8*yIv5QG|K zawdicFnwI=PCmE!rS?zPE9)5{l3T= z5(ob1kC?rJGiwR-z@bf-d1oKLKeh#pSHjFmf==ry31&tD7T#l^*pal=4;n$G3oDmU z1qG59UCD|qQM57+OM8ly#aQ@Y3+b+1SbRa46S*5jR?WHt^ypE1@P$}%&!u{;*7s#r zh>XnNzy|4;KzUY`c1_a8IxfPk#Uq-{U5?r(-QH^5PPW4nwf3~uB-q3?`dcUaK~j3B z>f-r{Q#91_L4JobF2ZU?<$0nKh1KE(%->Uhs@6Mo>Zuoex$m?DJTc@kYV!B|wQ_G506K=N&_Tsh2G4feQhS>yq`aXvXB~vGIn$nMPxyBJUl^GvBcco6 z?+uY?l#@UwX-YIVHCR0Q$&rUu>f)!_;nou#MCUfwTp&ib6SN+rkt(lDZg!6(} z=<)$x<<8X;mpl?1-k^uYQB9(;`7@fUFZ6No@V&#K6&u-F0s7PR?7+KG$vUEx<5}p(;@*5} zY0oHwW20%?&~Uj zhwr`x{}x^IJsM(j5EivK~N7 zw2hm}@*xqGu*nAG_ruLMOl?Pc-WYT$_Utu9i1cd*l&1qvEd&d56l2x4FFa}Yjm|^9 z8SDOxs6ORrA@lpO7vaJ##lzb=GvI+x{h{&hcxTssqa}mfs~Ys#$o^_qcPN-8Pyk%u z5g%0s-{aP=OOVXCS(FbImk^$&dfCj9=LS1E>cH`rozEgWoSrMXPEx?T)}LlC5u6Smyxr)129REP^5%X>C{aN<+hhMkvL~~K zV^H0y@kF@FYGOuva%^KjSQwJOUwvYTGnd2JQ9m_*B+ z0^R6PT|vqxu0(xVlcMbz=`yAVs@0TtWU0l(#bF=FC&w)PChkzIIPcf{Y4OCU@V+|B zlKP-}HMzaYs$*7;@dDLxzl)mxc(M^WY0xpQSFqZS{{{sV*Xd16_}xuep&23s*{HNX zi%Ce*Nc4Z}n&8 zEeXI&=zaNP$64ZhuFplVx_3X_>jX+yXfrP zeCb4c@I^=`I~$C*zp2?9ohND4MBa`9zf-czpIP8#zC+G+H=gRr-z~A@pPw;#jlt&Q zfg-+2TQ^dJ6=h6XoPlv*3H#iFR62h;r`RaRbJYhnmjt5Z7UM~x*70=*STFNeWCwPY z8m;DQhw4zyAB@ywaUoQGKCxFGQkP-WkKW0m=ss_GYu`r6V(~tVhqa?tb?g~ni4veg zAWMq|T{B1Yst#P*98@rTzh-IQxsAeeGWKe&>V99@2_F?)vX!l*ZS*3(%E0U;F#Zd_ zs<(CkDZ_bAVeHR3$i4igTJhr{$W-IqSr7PF0+X<|C^;muHGX^MM>nHOMt?GXexe6x z;IW^$&h7Ho`NiYA<=APvAvw?G`xq8gk4X}eO7wy5kTj!Yo-)+v0xeJd8pIvGDLw1~ zXdT0-W~i?92cW5<6i}%!siFXhwJC6EG1V5wl`JkyGOD0boFWv8KC8k@&P;K#(3}t% zP>AR2Nw2Ip64n0bX~L?hqTw)Z#jgVCluxqS?-&J+IWEZ>^P?elPC8z(6QM!nFpqlC zmb!igza@*wOcnw=vi7;GPsC$OOt5~-wb;jHk)2h0t9C1}BEWGl-6CgD-q-INP$a&+ z>Z&xUYJp3Yq#Y-|#j%Z)dg^X69rc#Xku(jqp(JbsiY>Ppt!7bT^2eOasZN*PL)N?| zjZ{T-vN~Yd_peh^k9^fSJgF+rLJ`0c8~fCfARb;d$=5n=-cSB=D7rb+K{Ln$eM+2p>qw%2mvY zGgCoHy)|##Q98P`lS}o!2Nue@r zVm&cl)uw4yivR)@$I`UMoo|c?*B?r1GM1>-M%$k@iz(xuWQg1z;<_j_<(AwKpb#t8 zZ~Q%HE3+6^pK9cd~)n%>AkUN^4G+7;Y0aqIWMP!sQ=~`BI zO&o(~3~HdZ5Rw~=Qml&{szoQb_GfEl;x7_d!~TJG$iA4(OgwSoiI}C`=_q>vipaIr zqKe;Z3V`0F%jIfp_G?3-@}+diwR^UjHfCbZ-OXV_7@E|%Q0v#FRE{EK3cX3GO(b(k z-RgP=gLEh^EDX{{8Qk}*zZq!JS==kcp-Jimb-oj_mfufZ&x4A@@uV-j$X_GUu!DT} zIvV}n$DGR(ZmyHU+U4ezx)?zyPIlHLk|C>_9rQN(f|)V2`qxEGV)VI{CuuEC{k|ml zgH23}B$htg!Odme{7=?(UC=ynz!+oY2CV8Drk>UWZ$e1mR}U+sla`CuwPiwXq*|10 zbk~m~yjiB2e0L0Ff0@|ilisA4K#oX>+fZ4_BCs8+poUDw1#5PWR?j=uJT_BvB9UCY zwvs^KFt22~M3o3(T#m7(##E5CtSWS1q;obpgW7?IAXinTPM7R5cLUbAyQn;;z^mF7 z@w2()@J&?Unz|j2P=id3A}pX_isU>?2G$@7>a6vcrM3XD&MPNZM7sB`c-=^ra7u?b zt5LvlSEaHD$BI;iv5Z#kLMc3ZMM4QDinnw&Da})X^2UV5HKb%=nw>9#V0Wfm$Gi** zTTMCSa7f~v)r8C5U)m!ab*p-HkWxXZqkuyg1XGfAjOL>EMBe4~v{B7hxzJKdxPJl5JsuAqa~vTze6)GiAyRqx123#7a zBuE(19SJfek9TU%ngR|*Z%r5jBC0$=`&OxxjL4O$VvK6yTY*fQY;{-EWOmVokQT0r z)?`$`70Q{2Y>L#gxypRi6OF;6eBBRPrFU>-7zV0IHhNTPA!g>XQ4%MCUw=c2zEkDR zM-fF(O*!rrhSAoYhB_;8CVi^C&=H-ur(4~aoP*HSJwog=BKWu|#o7?BEkxuuJ7KlT=J9r*NmNnHZ-lF`a#->~9b`!KU0@0h}C~ z=V6)QT;`cQ^VU3?lWUdrQeNY)dP`iO1k_VXWN=MMoG8U@Ya*>;)5xqvCz6tKF;xL> z;;cg^%1lgju!Ks99A^fZcW_C-=~Vpkt(p=9ZTSfCbmp_>i_6IEPm29Yg)K`wfk`ESyq zyP3A&RP#xfIPXmnE3R6t4`VOpW6fHyh<%`L=|ifABBhTAK^1RcZ0oHePbEsODvqOa zJn3zC4GIZBJs5MJaGt#uOLTs*1{)rhMR_~Nl!>Rc6>o-UT)Udo^@^hHp3 zCzH)gw^B8`K?0IlfOZ`!DB_M(wt^xz$U>jauUJcOVPPb)x||b)JQLfB(#F?jExhd( zqV99$BM&eSO1*nMz2)4~#_aoJ`EmC_rWS~`%%3Wz!MT6E$){=i{{Xb)f;Clz&~eyN zaU^syU0cm1rNS0u3JDnCkwmadIkfrC=a8-&qV%d3W;n&Ivrit_oVZY=lKB!*JJ&1^T4rxM_I5n{~ym0Y~%(zdVipKXjUDMQ0EIQMmh%A^jV$v+9H676M zTOyV2>?3Z(DJzQJNRY*oj+M#jlQCdxY3`k4z%`Vksc^C;bd6ggn$n_-X0xQYk-O9k z?!&34$QoZ!*VigDgIKcN7SA;6;_6Q(s4iIK*0G96O^%XNRHPmn}ALYT4{|LGtA7yRJJ;N*Gx=mTj^2DY~oIayUzw2J9|vbR5?BCMAIvC zkeb>{SHZ|5)||S7#x~^FLE_f8I)eE-a7Q?5tL zG+`n=!o4TqIX~KmKi#i3^G?mR-~2YX`!(=<;9|C0lJ-3(;xwLLhg^?QUPP1miTGEc zd`8xwlSlh9xH2*5E6lCs0WfQy5xFaJ3&KP3Rg!Y21Y)$zYb*NIOWSpgHx9LmlGt_- zIFNBeMtap-SurOkiqMu+*~Jd&*l8QS6}c7{*7rp>2xJULsIQ<0j(k1#AGH4fX+wZM z=Ha{Y>t8(RDAOj?>c;}TTSoX#WpMzy)TfutkY{QT0IjC&Zsox<&Ae;k6RYaCGaxGz zu1YENt|x15;=OOjpAJ_>w>P$Ov%tMk`ik>KjZ1(9YffrQPK3%WF>GgOWKd0a8Usgr zbsA1Gig~UX;|nORw?N#bSq~?Q<;7LyMMfvFu90aiqT)3PjsO+V=@Uk6))l}gF^c50 zpR`G6k-L+&x=kY8(&3~$3}Y3?O20JX(l_PzS-DN}(>dBQY1Fx|PR1h)XhsT*eJcmV zUKFs`EM$$-WSIF~dm89-NN`0BCI^yu3&Oglm8{R_$NPxbuHriTiuH2}vTbgJVw@FM zJ3th<#Y|&u#FdqfBUA7Voz45+NbevcgeUI(E6qGx;X$NoF?lC3OVd58+qI2SD;rj3 z8EkMX&rRfBC%jiHnA3ng>lrl{pfrq+3wU}P-A>&dcOsBTuV)sJ2z;>`g~l<8=JX#6 z{{UxBZiDSEKJ9c@7fEX#=}lY}mE<&zYf$ixlPk*KV~=z7IIkwvd^>lh_>e)TMEhLs zTo1cm;b(N$)4tWh4)ta9%Xfb98)=60ri*`F><+FVXLlLj?Ckl7BoSC z1vWycgqjj@MVZ(7hmZ(i_|{Z633D%FpIWu3C!P*ERt`<1{2XGlZMHLV^G=Ric3cie zO5wFj0cc%9;A6FQ$5oA^1jsn+T(5{V1eIH9$j4ewawQ~rM!hz59C{j|G>TLaS~{ha zw^uC4JJlhjPoJ0^R@xY{)|Cp%GQd`=#*lI_4Q50>#Go@%p7YNH16GyLSakY$!^Vc=GrrYk?NaChy$VhBjM2tf!tSmb)#%oUM@9k2N)}^tu4%uPewOYFv$@L#+V)8^cbv0+~ zZ4V;s} zR~HQ!kgQ#mdI4O-mkj4~I}`P(jV08+Q_`gIO^^!B44W8dIjq@r?K*i~Tu7?N02<+1 zY~u{r?lVYRCxS(4sl{{`Dmq*EY_4Sq1i!p)&C@?x!=Bz-S#ZFq2CFQR2V6D?=xe#q z?Cg9`r~QZ)T~U;>=c5tWQk9^tTw{F^&Dd&Etbt1$S7ByN#36~#6~+r0t!~;mV`X$~ zFe4ZvsIH?oVlYRm_J-lE_pFe;XSP*o0l8_pkEvN%ESsmbhD81H!xv6<>iauOcbh$;%PHyr6YOgHo zB9T&)>I|bjqz?4e-v9djhG8083RAY^}t7KaQQ0>pZE!) zMM)VC0W;wAt(YyQ%2%0QGHFs$Mg@9Lg*+c=sl;C4LxWh}G4OI|mmY1a<|h6RO42e; zT&426a~w>(Hfa)1oCM?ESF!x3AlqQS&C$UO-sBl%Fb4u)KzBZ;I{0iUU?iRSXHS;yXy@IIBT zd;p6Bve!MRz9fz>?hbynw#dt=YwZmp{pw{_jkpy!IiTsi$xu|Ffl(>ql(!TNNguE^ zHMQJ!meR7O>xcPJo{BrwBH2Flg!77M!lbXEYAr%g!^{AWYPEN8BDO8vx>io%-=!qC z4RIg`?ypLG#!5G1UA8a#Mx$O2s~Sc*eELuhKBl19b4j=5UQHk>zk8PSrWVKD9Qx9t zcRZQ^nPzT-rE2NY6?Ja)k^vnBTeW5mFlaQ3)a~z_d2mUQhe1@;Jl>qsX5BuIZtKe< z5PqChjXn}ETDht&Q<}20nQq?9qGZ(@xVK5S710(|inAU7^6vGlVJ&XRy0IOcA|=i$ zcep{+(Ce~116cQtz#eEy?3HsxWk+tcDP+eL&P8AhPg-t?GlbXFQ3C%%}4O-X;#jf#CPieGttHN-Cpq}L!Pc=7S zW(a_6V!AyRA~0*84h+0kU!z0{1KziY%U3K);+bg^vt;zE#>~j$ioX?_qKeL?jUx+{ zOiIhrp9~-#2jf=!q+YnBXGItll184*4f6d3DYo7^R84Xhy-OA9~ zD@8I45w|(73DIU~Bl%5BIcY8n6U}Vmq>~bMc0CKn{{RqeHG3AZ)0K+1Sr}rxzTOk} zjZZipyw!^+8eOQPEO-L3f^n9ElfLD>6GQ&eoVyIDJu7ok)8x@rCWnAKS3%*80$V72 z$vg9mRNf@D6I);r^sWl^6REDo$_c{d-v?{Sdk)zK*8``mbu;*i9VRm@$jun*wzyvd z>Aq#o(bQLGuV@BcGDeRig~=YYttG4$gN?Rkw}-q-XW~eulF3{oOca+G=C>=0_-|m= zdK9kX9(0?4$4cyF3}Rrdf=36^y*h2U-I%@3bHzS3()=wJE+5UfBP$9Xey7sBPsD!~ zwar#?wsU=s+;m9CC+cdC#5kVQUA%^YO_QXo}^ap zo#CxFP9dOxDHZdFg8Wx;;k|C!XxROom@KGI1#9X^CXV71mL?4wjmk%HSt@BTAMI7I zY-7HNB!qwtc&l_(w4Q)NZ*LvEK@mU#oh*Vlna@Be{p%B2zYw~?y9K#FN|`kn=mhq1 z$#7YSz|R#m%x$poMxN-_<8CTznH3;*V}J)!Stzbsm2(({0E1l4u);YGJJYo<5=Cz8 zpys)GwZngLjPY2db{y%FNh=5HaB7v>Ji}=3T55sR8VaBXBBoDM8E#?4rpAg)iZfI$ z#1WEC1}j?f5`PD6R!n!`Mr^^)fFoU%1bWo*F;awMTiCm}5%@@oFt-U%x zv4Ngyo#mmky_(UEOioumxT_X0HfuSxDBO!!+P9g!&@;tX*X57wtT&F7$u~r#j03P&78b=x8vSeltn9ezp3+Vq>NRkvIe z0~L$niBV8Km!h}5s6U8_5R>}dYfyO3>pS%ejI%M3CeG)1^QsA^Kvi z!y8UAYQ5c~N0u3@on;5gj1s=2@LmORlU-Mb^eDVnEV10=T0)s(KfTyj2DXz(-!BI| z3ihuF>4(E!5e;$0#KX*xpMHb&tsL(iOQ##lL&LmR;2WE5Mo8jt(FQw6?kky-!S|}Y z4<6O)y2a6+_DKt5XNt}}i|aY8ZakS+2enrq_4KaG#g^h(Ji?tn zi_*Ck0Y-D0=!?2JrmXBoy93{~Xz5yvwpY;HMCLb$f(Sh;EVTgh#WqSkqs01r7f{*i z=&Bm_-6ec~e*-7;&2Tp&Q7KsSfn8nI;r{@HXGMd4WJo??oZxYf(yHn9X*=bHb6ioE zE7U3_%wpL|EKU2&)~}i{4r^9@9yUy%GL)hB(h! z$bH3XvD~dzh_TU0KIPq-p(V>RZ3d~ErxD{c`#d}4s9J727U6?=$);Ta64dfeu&))9 zd3YT4qgKMiZl+VkJ_~F}H&jsEs`aIo_i6cRwJ6m{BT=3y(n1|D1!Ti~?!XgT(cCYV z2{kriDM@C)WLA98KKx+T_u5eUVyaIhd%IE0?jtM7V5f?)W4R;%nscx;)NC*!s`o*; z&_?PLIIQdIl#G%()K7L*Fb_3CaPuiSsfWyYe?Apr4Np^G+2rpijO28h@ZTBz8& z78X*1o8}j8wFABmDo8L!O%@9i5sCu?#YXH&z^B{9&fJ4YRxZIG%bHgQJt}yRs`4t_ z7J^fnhukbjr9vc_0xM>FBN)gWQK#6L=LZ~DLoKqr9XnIK$770xOTh=vb*Qr=ks zCP1wDrFe!ZxQMNJ9tptBQI6D?&1%NWG9BG&yGpUS0)^ORJBSDXW}1-UI#lm3nE1s` zkQQ9_rk#Mybft`#&gz=!N;Oqy$fDykBe?TB)K@!4Ip(UvCeuxxQpSXV2xQ&Tm<}4B z9AvFQag4dJaRy4gkfQ(MB?OD@x87Cx1=MLUNoGVOF53o(m$6$iCJ zec^30QU3ssn`M8kZP81u$89We?eN>fY)|}4sU9M)}>v7YHsN!;#PmmQQ0 zez~mro12#L8{JMt+@jcR=dk+vn#|RTu7y6u3%=x>55liG5W#pYm~2h5XV?6ZSz5FI z0L3(ndY0Try*0U|J*DVo+yL1m@CIv=yOYeicJ(#VYF8pV0!QAEV!38NJ!sBiXw6Qo zimaxkR-qcESPxBRC#6TaoT8ZJDhe6r(-qmMAqmPlM8-095fZt8%9l%13G8i)Gcd9Y0NJ ze6_ik%g>=W#d?~2CYcWxzomH}hWEC$qlZ>t3-Pv z65H)yIAPYjW5ymIm%`eRh_Ps2?vr0sIpEjN-y5VKZM0F>ll-a6-a>IpW6o0HU%e!b z1!uzC)a8p-HWh+NG6)8$9mLg>AIqAJP_e)i2E$^FjDrHZ-3}d)25z;@>DqHh>cm%P zXAhSU6#1-a7?xGB%{A_VOU}nNADG$e zO2u5$d%h1f0KY2TYMEk3HC{uCp&c!*rhmO!3#nvvt|mFk9`&UK$s?@*bmI7)^c&oa zX0cLJj=hT z(CLwX#ENt1D-D}0MRfXdf5d`saYbEX%NGvI2x5+r78+QybShb*7(Nr4& zrk1CoKF~2qn!iA;V-lOp8R*Ra_mLRj1 zTIquKOyum!lG_lu$2D_AkuCKYUP1DMQUwmXPg;fT_OAz;&yrUTWY1Dx5L#Q=Gj0R{ z)7H7md)to@-~E;)F~`-L&{lVbbPJ73Ej-Lqve#*Eq{pGKxQt|Dn)BxfCez&MT+MkJ zx(1CFiFmRrk{DsPS8h~1L}P+0%e7A##&K_H36gyC$4cuhd_q66(piB*yvqLoOGCM%?b^Lx#W2l(;xO{%;a;`R>-scLZkKUM98Ahi z#pzS+EbMNnHm-Rc%nXckRl(YLt-JWJSg^sYhf!LlV`9y`I)~2#&OQe=I zA<*SW>zd}#c5$?jzySjc{Yd=lO8(~R zJ7@C(sK+~NBg399(eHHV&)Z;)z~x=QBBAk5i*&nPK1Gi1DP(0J1Ym29Qj?`_XMV?W zvX?;2vPe ztk#NWkaZx|)s!V6!5r4Kz1b4=7|`aT5#g~`UBZ)^gUki^uIW7ttZeD=g-%5m?+8=U zs{Nbe{pLZg7Vdd*Nd*x)STy~R=)8@w-1Kx z(i_y87|z)|;maRm-ly;eqjPDm`9TDh^F~DBx+?zwd-{Gf-w|9}KA|U-0t+by#6UP+ ze)sA6*IZ?Iq|S;ld!v*dEWA=g_VGCHgHlK0Yhu_DPEYqjlUF3rVvgy`q$e-v&SCeQ#^Wh4(!F;sjh;rF?eU(TPpU$_tP8nrt{?8>whC1h?*8w+?)54wR* z;3``g3|w@rCbr1rR0h$NxExm{(>dF_HSML|ocFAaX4ErfM>T#I^3-RwRK2)Tq?(xS zH?e^#d1=j6j4X{&>UY>BM^jX-Cw=_aS1y+#+8Xa{2{y3<-x+)}D;tyYDPR;#ww##*VcAkAK~yhI^*r7s~e$M$(coKuWt zMl)ALp_esISwlA!36cnh1PZMr5^v(KBPyn9NK{DNRC8H_GUScXa78iz+m1zaYhono zDX*s@bL&*9A#)g#MomS}-K}YDuz2R19f>WHwOq)2#w^-#lx`J4?Ieu!uE~`nk8TA~ zXybP|#R{m@naRkGy+NnK_!u>-szQKaMk_uDKshz1HPy;bLfy5wmAY2cI&nZ76$YB| z`G6c(RIsDMG3i)Jb{WsKU9nfJlLT>6NdOH*2V|8?P&Q&fiViAoG;P5(6jBC`tKB(8 zBva*R23#;^I2Db39zi&(u}!^>dWx%dz~ogYsU{=HV$Lck8_kgJs}jOpkEIC!kD6l< z)nhruScc^a5l}ks9chaTDXFk%Ex;17#Zi(1y{kuV-nA6CVN41LhMOuJ)cM0yo>+1! z#)#nrM^$b{<67x%HsY$@TJ2nlo}gw1z{NO9MJq+dX{#Ps3ZOM@j$;OrX$as{w!v00 znt6>2b6Hh^pX*urCAfRg=VbC-{wK_ckfHa`!f|6THcmUfWMW{Qm$ti&qTV8BBc2NXNLV z4$l?Y`AKOcQ^4i0dQ&cQKA$OYLEhO1w*rLqHKeYZ@;GbjgCub#e(2`2$mXZIQnyVX zp{5av)tspaDe7u5PR#`pXOT{K+r>qijf*qTDQZ8xMovv3d~~23(h%`TO&zHTpkq%g z1zbhI=bENI@T+^UyqpSXsEhq*6j4A0F^xRbPfZDoeuP)ge;aMxbqi^~ zgOOiBn(|+a+HjviGlk|enxQl2g7n%2c3QNZG-nnj5_Qzf_&En;748O|D-M$16HyBR#zQ_Z-Vsd02! zy5^x-i8SmxngO#*#%iibQu&oEqI`^EqLp|w&GZ+@pa=;0s(56<9MlfnRB1L1FgFr! zmZ6evde(8f0ClL^>pkc|X{5P5YTRykBC+6?C#`8j!vdYfV`A2P@J|(K?Zm6dtZ3u_ zj8?^^(z@+EXt^#X9;>CRM%-Z6B+)&z+fP7i)2?AzqdZq3t5_9c!D;S8M@8`Agqoy- z{hF7@D6!qh0OXpZ@YR_$iL=>i&y6k?Yso!wYicUU%0~)#YU%7BZh<$_xsou|(pcmE z5&6E9s%&XbbC_FzSb}Q(n=87Y8DelN6pxtwYSop=wYrW|$WxP9y+HI>yfHDc6BQW7 zaacDHDzhjZE2oRb)?e9y0n3%fE0UV_Utyb{d_7)~mqXVsTlblR>uC1a&T9=knxE$4kFe*8$ zQ#)h?tB@8W=BR|5c4W#a9gcnnL8e;UN3J7C7c4lVS>IZyf&&%qrO*P-5B!Zg;F(m=}NBL~*JTGvHeh(zL29P`C` zMw6D%?T&MZ%x*o$PwP`D^eU8W%sa<3-K<-JnGY;`3gk7pBWwQvvf4;wf+BL=vs*f} zyQgK(IjppbB#hxe^)(i`+k0H%wQE9?l1IvVRxGwes}L)#)-Rx9Vo3Lq{8fgB%y6V& z8tRNB(dW*j6WqkK;%}XJT5?<~+?L&eo-12g(|*UK3#d5^*kkEj!Zr?RwQa#P^+$2x z-F96YTD7{imucKL4{G|h*5NMgVz`LqBv8tIjeJ*NUR$`MT$W}455~U8&@H61&@3&c z7?LOz7d{bM`AKe*U!z=Acu4)!m`c1=6EPG63V3EyRYuR4qM@eXW!+Cvk zb9`6LeVOP9#ct`!Ry78&8KZ4ZU6|Xv6)jpeGN|t(drz^N{!wph3rjP0 z9RT;ODLx)}f!iY555`eizYOm!;kdejeXBd*qcQ3|E7R+04?%LH!|pskbEV26zJtqt z#BvYN)NcZW!iw}?9E7>Fh6u3N0rzp#3>xJvJTt6cM$Zkn+B>^rKT7A-)MYs6aknr? z9Pw24y0;>=^>J~g-o35lO%%)29^=$iD9E@t73j&g4Gb-2QJ!PdH2_`JwLX%bUhH#N zvcK8gf$d&=p(yTAdryv7<{n#P%o`XE;F_VSMLmwQacMQfeU)3~4M5eSi8yREXzO2Q z*EH)(>q{rVj7K!U=Pn01{6VJ+^*XTcp*o0O-5DE#qa4*)CkRTR=diAO`$=qv+4r&9 zcqW*2R&$a0)$*Ml(Hi%YmDtQibK0~m?_@{HqMgfwn!&S<>ixGZyw4h*Sx}K%F-FLC zy0Wn97+}!VhfUp-wQCc550uPw#0K!8$i6=16w=~&4mtaGidR=H3!gIHIrRa+*sq`TeLsN5)(PAG!0H#pnVnfa-0#sD;G zv3YZ5qPCqv;F{{Ux;L@AZ#h`_HI=9_l0(H>icm)tQtig-#y=4$8STefk~iY3&f~Q* zIS4hqv?|59pXK1yINgMxsF$d%T73{lDZMKE!suWsJ-!#xU9)0c^?_8_BLh4O;0LF zhKmu(t=6uiAx_?6BLW<@#kaB9=s35Pt^eZUAeEolX% znMo4rO01!Qs?kCo5>~UPy>@M+*1nwEpv7w9;*Bk0lu|qwBcAoySlyPlA4=rzt;v)X z9qXdf$ND%2(zBhK1X`9}aY+!#%~zWuO0jOh0h$(oX+*_TRCiJaLOat8m|hK6XUkTx z)Cl5WjB+bJ{^JeBU6|uFlW;dVp~p}NpLnLq=cPIY+HgfnBs~r(2&S9!R^wLgD$Ew^ zjzvx7qSYh?DS#`E^_@Fx1XcT!hT^hr*zUoiz(8B(=B=Y)L8=|<(wwF#jL?o?R-n0< zu*D=p3yjkxT=7Yv6^5tuGkHmi{!?B24dPW>rzsE7q%cahSslrE;P0K;pBO=E%W9 z*BXn(xaJk!2=xU20EHouQ%ox(W<$=^Qb;-HHCt5G;n1$G9aNi%6ac=MC;V%gx$$13 zblLe4J1UB(#csr^IBJM5?N8fYP!1c^VyHgZdCR$ za(B6bwZFP)L~c1-4;$_n`zs$|I?y*u$ZzSYxTpHw4ov!#-eY z&9tblJjKHIH3=%c`K65(R>K-&S{u^B*BWU^m^fdV2>w;p{4wzm(zWD~rPvUD^;rGo z?~i)p^s6}#zm+yrkGM~I)zbtIF7}Kp>T4<8GghvBV3*eKVlM=$AlturJYt!rYPNc$ zMq4@7RR;$LkzNaD;v11@)`oXj@q&c#YrXhr`MbNspodlYLHf|7nT)B~T=bxFDQN{Q zQJ9CNM8>3J#YD!M5YD96AMp;u?Y6MxAD4loBg8ByQ+dyJG~7)wu2>T;`c+0>J?KRXgdhqzay; zrDCo1_a_ZlR~IFZT7A^SlbXzuPHEVo?AbY{hQ&f;rQgLSG+9}3PZt$Iq#~QwKn{Uf>Dt`G6$Uc(^hkyO)^C}sQOh9%}aZXFy^%( zxESkP)N&OydMlHPmWG+Uw^%)D)xG8%b6nrrkF8j@x+TXnin-s~OXox?uBB=kGI3Gp zv+htUQu#rI(Q;ZdJP#0yTeukh3c~SK!;M;X&l`<&-Vt1D*B|xHxi1u)>eHY0jdVfi zVq5iZP@w#TaAE)JkW? zT==5W_eyBvY?c-0T7=|X6&-U^XsUdSR2Pxr>6q5(XhroWwkn`*>T9OA(V9zPCgGfB zxOuJ_cuv*SX`UjHtpq`SU@<~=dJU`YU5jdxy~SR!Aj-??D;^!qCgOQDr(~?RAm@TA z?q$n%W%!>>GsGD|$Tg3qX%}VwRwK}2y+2B^g8Ftzz+gKX>@>{-OonA)0*C2b)n<{r zX|}F;m95q0nF^)MFOJQM^p6Wnl4uVk#C*&cdv&h8w0j9+*(J1WJ;ieO+AZys$%9U} z4{IRhbB(64X{}jN(duybyJXfU9g7@RR;LxjtSzmB5u7mtHMyy&(=R0=HG8=73c2}E z``0HgiM&T^Gf54_p#=TSpZ$92p-KEqoJDxQbuyjqiFquI4YUs|k$vz5Wxk(#Zu=y+ zWN$_ou8zmzM7Eb%)wK0p+%&NY?yatLpAp#!V)EvUj-#sAP5Yd_%{6q-F}62TS;=EP zn36dL0nbBR75#&Z3I-S~D`*l&!C`iik@pax%}73>mxp1rQI_^M!SWbi21T`abA?M zD%<&Cj@)CddG?+5?L|UbM~OQVP4O15<3A1QjWo9Q(p!!U$_Xj;=9`aHXsuz|jpe45 ztW9qu*erK^&0U{|_1!m7vP8GJk9=bcn}$Vv>2s)E>bGrkb0a*Cc{R79#Uyzs%nztN zYN@YgXDXc99k;}}tnYOTWr>mGiAZw1_3d03nJs}Lrq#^bhu*BYB4&q@az%PD_)d8{ z$hO`g5`O4o(x_f1of|msR4sI?neNDjH<0I@qczm&UK7?eiQ3xU+@MdtB)lrtHKTHr zdY*f842vN-;=6ARTwUDwKI2NXwYYnL#lc9)3_i8#`Uk`9A4@^@*kiuCCoLm`wSI!E z_=8)v)BL?E&O3M#U{R*wcH^-3r5aq?7Z%{|eA#yd>vBY>>cI23cd2cmcRL49TIqaI z4aL5xZF=!y-^|(NulzeZB8g;1TWjRtOO70R8uG6aS!y~Z>9xOv#>eiz zRex}M*O@vJjIYpa&YtY4I5fq+`FS-=A(66bV*-RU`B5fTr8k=9tJv7jdgWnKaZ=vL zl51nam-hCB(o(9&xvrSK4H7fn-diKGP*nTYZPcZnw)|JA!9Bu=NF;S4xz%vvnmJYO z79u2?a;_>*E1Zh49l37Q-h&@uY3Ir}bgqpw_~M5+=CW**!sCk5dzUMY)z3EDC6%1* zd{m0xE@>T04cMqW%p3|=5*sgh`mns2*CyhB3Lh4dM z8Li2RnD8qO_AsEDaCcy1$@lYHnnPs(tS04+ZD`6^a4TuAbhd6R3g?==ZxD=NR%uxy zaCxmOS1g$|%GHF1&N$V4NoOS540^;ir}#3l1Xz}2$^LI#%UTtM_Q%0mZ^NetIaSY zl;w{#pK_s&aaQGgo+^y2O;Xrpb|6zkg=xq)KQ%K#xfB64v+f+!uDPg_YAq+5lS2^{ z>)N97{#vs5YBBQE>;`khdeb9jjkv3fH#R61ZJgqr${F^TqNV#WqNhPv`(V*aMQm7< zQqe#OMJ*Hnlv2@(07@xwNs0s%l(duq;J!Q4ZY@3}Tru2Epo7XFWC6&oT7ME962@By zRX)(~xer~3r}VEx_|dCd+cujOw1fLX7H9c-s5r)dBVG_m`(K|k24dMH^(Lu8`mwp_ zk~4KHnQ!$eC6&v$Jd;SU;CUWcHs)|H5G^oZl3P|LZ+~r9t=}{Ir;F^&e z3zAJSBSzZcB?q2;4NNa3v&n{1ykoCVrDLMVtYnOEI^(rTmYs)jO)${dwvq`Oaz&Or zmUEJ5xpAIU0qs=dK3D)AMGqlk%>ua;y!|US;WnI(m7wPzN~dY8#~A0}pDRk%m-?V0^*jtux5+ZO0f5%|l~uu>pZyKg8HPzZGPTtD5Gc zkRYt(xk(zPKqLGs9{NCYM7O!u9^tjSk~<}G9w<50Xa4|r*H7S9->Tie z`xVQ0l;>2N{o`FR*uq8`>0K_EPxchQO6NL?=`^*@i7%y0_9`2k{>{d%T84sTm3U=2 z6>m(_+UbeM6@K=^?1%|aE1k(+#WiJf=${36v-@kqw$t39S%QUPIx+RAyldilG|f8} z(FQVQLpN&jzXo{L+eowfCCe00atTF}S52Vk@@jg#7K+<gK z^n2MrX=U==Pu&&N3%Cp>Z#g`%)w9Ji##OmiojHxq%6&(zb55F+lDVv8(i>#ANSi0Y z1EHpC7xC$OO~vE%i8GH=S~bIbPSl2<;q(NIG=SsR8REBuEuoy{Wpm5*>${oJrJb=P zr+enO`EE&*XgMCWcTtk&TXT1HAZ18>}wxUwwp)3 z2^l6Phaw|#9O(}oG&Zql4hUkG@< zI2t`Z=!J#OBWz%)^flgH-3a14SZ5qp&v4zDBas{AxPZ0X?i3B<>+btBk^onm4Ua9lHF^OF*kZ+L~~lG#AoXUKuWed@m&|f zO%lrEOrH8^mPr-%Cho(U-|^0qX%3tAsN+;(+eibi`NeTmr2aHqdlh^WtXYW`H<~qO zTnwJ|yW^YNJ6{Q33#ULNk7_F(Sm(8T?|*qUy~_xhmj@#U6`6Ue-rRsExspX*ftZ|D z)S)QFpp=u;*MO=bj^es47V0wPS(Qht906Q*nk2f($s)GxwD}#L0IvRG;Ds4AJgb{i zD?QDty+X$KaDrrVrx-kE>FZdtzU!Fa#y2n-^sZk1-t$grVCFl7Y%e`=?eEs0(zU75 z_Afn)M&|^cs%yE+-O!Z{PpI?_8&A_TC`3@9CNNou<&G-kQF&>EP`kMyYnSi_rFp98 zPHsUF#Qt1o`@nzttEJU1qtY~~uOu1TFpK@>>?;^2*K=t#(_=eV)Q!B$rdpHssTXvP z27ILd029S=H+~nN#eOMh<&o0L#43eR{9oh8HPrZO)_*2lO7O9{oS7eYW7vHKYUq&p zR&Oa*^DZQ3W9gHc-qw|ld3(&~HSY~tNU3Gy`G!dFBypaD^sX~Q@TRGGaV8=1;em=Y zk_-cmI34TMwF{R;5?hkyR@jWZ?XGs~NxQweo$qyffdpfS;sHkQPL-sosoLg|s^OwK zy&lK=8^aM=mstoTF=anPEizz z=R!tLrhD^IYM0R8-v^3ak)rKP_Ns3MsGG@ajj#R1j*=T|D~TtL8|7wz10X1Pu4>i(9t|rfXJa(^7>djE~|Rsr3f2A-9KP$dide43Jce zQnbZ79-Zs1jYm;hqO&wi<9tc*?(0VqUg?b2mH_7pGVzaaI*)qdXOM)zuWz)}ZV~dY zh3Iq6YmE4X;pn`9u2@77OKrK02LPYqALqR()9Qz7o?&syl4~uPs~jAA)}_$G6Im_P z(vHT25)#kPCY3TVOS9&umF-j_#0PU?x9u%5fr#R=ETO}KYpS-}_I!I*H79KeWi___ z=gKM12HM=v(g|dLGX;lZNv>POlLHVPIRd3vR?_AZ@fpokg;nJs)z*x-BNb6%F-xC% zt6O0c%VeyToi&0IYff8@vDR#tSv5S&Z&JV5g7l^`I`LNHn{_?5k&{K(ijg5bYTQ@` zplOji)oCw6o(&dkb~YM72U^I~CIlQBtk>+vlU3!E$;qpXEXI)rGS$&&gh$q}qD}G& z=q+OM;|8vyRtZfO%z~OjFyVSAXNtiTNR>}#7+nU0-lOmqk$Q@ zHD~~|SP>J(6lSWs^NP7QD^#{~PQ_SG2&m+XAmXOjeJOTvKr=8zM4c*jjj%YW_Knpk zT&+nab7UGpQ%xD|T3#IR)}!K0L|EzZZT6{ZoQvCq8;7k@$@Qag^{2`% zV(#JVM&VSqA6hrJtsJA~FK!f@s-E`sqkDSM${fYr!i?1|?dwauy=iiXF>^IjQNB$( zO}V^~e2~V@N3I5GJjV5|C*q#1_74a&Ww34>DCe;2{OPFb9LI%tn^T)p)b1~?9JI4B zIX#DAT(QZME^6K7-_SEkqwOoMb=Hs<)T=-Y*lgzpzsYn45Gq~F9I)y)#V(FUnT3(SX7S`Ku zWym)K^j|?+pA08=@iSV?#o>`llw-#4TDaS&naDFTO3}C>)ZmI-=C63ZKeKqQ`2YZU zodsqi@k-z2A-rc)mxM$ffV=& z0|uqBo=@?5zzeG@wHuP)HfyCakre z%hIgJ7!OwpsY=Pn?WacQFI2D5=Wf+R>FXIFpS2cGK+)Z3$XI7d7IxzUiql9&Dmjjrz{PtZ z=~dpqdRIX44Ae#u=|#%qImtBq=ZdQ)o!5%$S^#@gO>C#N7Y}A~Po|&l)F)0P^sc?F z6!xSRDtlA84w22;Xw4&XD|XLA#y(oz)2tUGfm@N;qAqYMlQNtvbArf3oiSK?p&?jT zU#S(1O>r8e*(@rN%=b0^9~|9YH@EoLFXBFRIltaD()a}Jn(y3KFX94!w59(5c-HaQ z$(aJWjV*xDC0_WhU2CV(({__5(xzoYa^Ht!!r(4w@eRqqQVwfx!gi2(9DrJn6s|-(7g1?^U8%?vDUE~! zAH+sQbkcmz=QV8=j!#{`H?uiK+YD{B7O)BDHLrOIhfatdta43cFs`GNa5Y*~Lyz>s4=8<{p(SrMaVZ0?8T2-Eom!f5R^lG)~$@)1E)ol6zMlV?UW^wT4P@RU&z< zHG6sOn{1oFsgk-hrz<-im*b05r0X!>#0ZAuZIA)ReQTodRr_h$G=lpp8Ddm>R&=%( znuV|$l##ju8Adu*qi}kqPPtdZeVh%T`Ax@aN&ynSY@ssi>)yrq0CZ0IqxDKZT+2 z)~g7!C+ycMSrnX=Bd_zVzXJS7vCuT1?2TPrO}h}XB9_m!b6z{~oL()~Ab28Z?QMZr z_jPRbG?PtTZViWnd}pY5Qg(_s*5cYy3&u_`eMNM-{{W0`JV6)@7=qn75$7bIdf;Fw z_oNg7_%=0kMgfi)GoCcFD-?$D#+DPX|nE3MESvB!;tf6KY!uxSsT6`a7)9s@F0Czt1jpG~GPNQZr z5m|~L_2#;ND93LGdXl2N8teG3);)}6OW^6o_kBpUNbT&k13bj6hvoA>x};~H#0rip zYp=71h}#rp^JEIcb`K-HSi8E0=K5O~hvyk>>^L2|^zYKKF6?5M4BlMFDq=Z2p8Rm# zI&u2fsY$@b)bD$9-SEq+i# zA8Oap=2>NA9n{xEjS=BDHpV^iH{L>2{YMp%G$0z^*5uo)Y6?1iqCx)vj~c+9Op{qe zY=91FQ8OSkv7vLotI%4Wq)^?7Zp5)+6cN_7tmMo!lFNgNw{IattYofIA7NTnx>MXnT(&)`{);?1bn?Y90g2$6+B%e?&RXIrZ69lb54}{9?2apA#kcTH zcB=^lj-*#OK66`gq1Z>6owJ%@$?sNH+?tszOB#A0C-K&JW&m*19B!2!|c3Eg})gn!b}|K&;!l0tu9D4ld6AL&)#LYSgu|0sM`Bwp!8m2XHCk{y-YYe(j52kr33eZo zM=hPl(AA6ui#1rtOrF&XU4WPdk|*8GIEb+Cfr^STxv3qAH7rC5rpI!?i~-W3l`twm z`J|JTrEs}ZXEfrt6yRx+Ns15+*biFt--7z3g@&}lqPC?t|sRgF}ZEO-LGOz}nJ7ak$KlGu+V%-cxld8nn=^@*)M z%O;^aLVtv}TFPkInv_#+@-5odLml#I0U4x`7!Cn70+tLx1RqMP647roR9lS{`9;co zp>zJQeJUJPnSd@nl^)Srr=XhDii(?PqiCYyu?-bA(MHil!o*Z_QtcFN6kIGrMN70% zv{7)e9e8iXon>d&t+c@=BDIUmkOTZjJbsn#!9cH{el|6=mF2puNVk`cWp_V#FG2i` z7afl(xRsfSRaE+#rshN3AEk6>OSKUzE@UKq;%gnPVEb`La*>XtRW!tzM?RvO-78;L z(WTR4j%Le4D9SgrW15DFT7Wd<8DUKCS~`74n&utJCavfq#;2)It7;P7 zTw~3Uue007_)f~46-+Ik{2WMl7CgNlOUL2iZ5L0q->jjZjQuKmP% zR*sUH+Tcc$pX87LNU3hNhGCA9k^{)qbwz2lFpgco7 z66q5z^*K}CydLRx%;T^ov8b9gru86=G0i0Jv{Dzr#aEq)Q&=9XAG%Yyec2U}a|vwL zxP_e4rL#LaRVHIKbCi-jrig&tD?;+y1mIOD(1+<-Uqg|tO&eHuV~X6fw#GQD>i{?g zx=U1DqO+4XrxMk&Z5=A-YKAZX6)tNtsVfa?jMRrT!%pL{!%ZTbX(@$-z^AAll*)3N zWV24hg=*1x!p2XvWm;z$qW0j&8RxZHIjXZOSiT(dTxI3QnIY!1HG9ml8LoYtvu3ri z#LmmW=WLe;+PR&0=TVpa<6S3#^2X!oT;9F^03iu}rnQd5NXuI3bfQL+COs>e71U`b zIy9e3QV5RIFXw~(Taihr-%l`OBNA$zgB;ZNku;0Iq1kFexsw#H=5?zM<%LABI=>** zxb2vX;<9XFLfb&iNYk%MYsa^WNG#{OxJYf{d1YgTR2e7msk$BdBTq@Tmipo=YjiIx zt+f<)HS1bO!VH+@)h~-KGSNgagZ1XP4HDN<@P4ynscL#`@kJbf{OCY6?u+A}4lbUz zXHM>jlI&Ivls~i$Ab_g43 z#(uT*wZ^MyB0&^##_e)HbO|3Y_N7Z3J3BT87?pnW5~HnW8rG9h)fVZiYI&81!LKgB zNj{*$@UBXz`9bupDCPeEgm*%M8{|9f)GL(r0=*t7R!Bh|aniig#-Z*sYvz2cpz;Sw zxz)5Uanh!pSjDcV3*rqzDevv%hC;|9RV0iYjAbsp*7d%hcW?l*vWX)d9mWM*XgS47Y%9(%t0=={%N`5WbEC*^Lq=m8nVe&a z@xKsAb-xnb>Kk_3ysXi$u1{L^O?Ev;Owwf+4F)CoaTx9S*N$J_GRRd90649wQsnA! z)}J-L%BpZW*2W^^4OFim4}(@b;MG$90GCobiceyr zxiyTq59mc+@cXR(BDGclf}D_RGB1)N^sBxWKk+?RKf)?+V&bmPTD`t8TRRiNscM76 zHzDkW_1hU`l(wHfNn=yr88r=>#b{N#(X?BXVJ?>t zC-{>fisx@U7aJo>rOJL@Oi~_y3hRd`{?$cfF-%BoRJf^b4I|5M;F1a28C0x`!3L^E z4r{yA^as@ZM{ba*Z?!N|M`K>27sI~~N2iNJVubC%l#{rhTI7}QWUeIx#PlBt>fR(W zeS#6PKfNQI0DUSASHilciL2Y#-^a10U=h^iNIr(Xj?uJBEe}phD=7BHz0X!Yrnx_e zIu56<+Q7PPvb?s^d7IgQJbF}5nijuPNW~hOcYrhtEiEl1vPqwA9#_uhKDAfDz7~HC z>e{j*{{W&$>S1R4*yoDa@Qv;L_lIrd)Z#H-E>cBdgOSEfUDWl94Iv|Y`-FicZHmrt zGg|V+r7eS7*u8;OH9bPxO|Y8c=46Q#Ny{8_TqlZtD_LobE%bnwlXJR2entNP8u9H* zTff#W<+#1LcWD^0UB_tp3T}VOOVe{ec;i~N@dx&Vchboc%3We}@|7p0b~?Ytw3|kF zEcEw|@V4HdbbQSTodF&B%}Uam*T>(>-BZ3cV%(;=I zn^h2)|MLeq}Ixa^Tx(iUYQuduP=7XR`lo34q-`KLq^tYEi2rWlwzq{7A>?^ zq;`TrJ!-B)JHLkaX6sl;9DJ%XUh5sQd1K`j;vWxUiqBb1!~#cJ_c)|hh!AtdX=^P4 zExj4Xc;e9$Olg#+Pa^L zP`K5iWl+(_7z|ga`qu{x%0@SjYT;?7dQp;U=WS(lt$=!u*19b}QhRvbC6{8#N6b2d zUU_MFzy-?Z*EPLka>6kzncR3m>OE^(7w00qxf$r`FPkZuStO1)@U4zXHBVT$`!1Ff zXjx$S?5`we_ffb=$Eo$M3&%~rRMC?^H<)u>y^5@M%AEqRuRk2^LPbyEHQnb~4*M9{ z&n197{c0$;Y)-n)RZ&EpIV!M^Y=wmLgDB_Ke2660Pb!J}!T)W(iBgk@c zea&39xAm>NeIsH5;;b%^Sp2*SB`fHKV<0Blnl{_ov>#6qk{X%zQPQ*Ku`Lo=te2CB zp=_Nsqxj2=!t zwW{bC@+!2J5XsdAXqxIvPoY#_Mt_J?zSy7M6&(635+U1_J+N!ewM*?+Ouv>ESv<8r zD-dvM)tL@bTb`CK(Ek7sr57rD*OUJM!cBU-6HEU930>f)%8B-j^^deK_o;mK^%dnS z;_IQ*REgsIl5vVTPU!)f=W zWQvTgLpnB?X0VmFNf$=-6@=JwPf|}cXHe9nztkqWWC`XR?H#J1!jq1NwKgv2`YFHC++m42TrE}DN4(kI=@l~*9{ng_6Ezd*P{$*Sz`cN>Kp$F*^7;#tVZh-2K-Q^a>N zzR-(Ll*^duHE9&txa+~M7M>TEQkFDAPU`fT?{21yh6{sPP})Y9zGlS?$CF(&EBhq$ zIqOpPtsIEbrAY$3e46WY-8o~o8@TqZJL`LNc*=lS(`TMEK%k7%sfu$}u~hKUrz@>U zjih<%DNV3d;MGw$-PWGhA1(o`9NIT_C~9J33oiB`_RVpZRuT|fp!crE>O%Zxs+%(~ z=Dhm&sLEWWQ~0zvb7tW3DmS>$~CWTb^rpY+}o5SfmH)C z9P>k01@4_xcLA}@Sy=%U%fos>!8Lv>;p0M5R8IwzyN2PM(rFLEtQ1;TP=tUCPQ7Bp4GCw1V>U3=qO=|ipJ9P$0JD!8y zsWrHa*Gb^bRxKZ1hURhmtQ2AgKzsg`6K|H`6|Ih${tVljT#NSe@zgKNn)A!e3hKj9 zp2WiPO98`qYER=|K~3U)9xGumy^2l4ltw##b_ z(HNdIWFrMg;=HfK9}Y&H9mb`jT}vF%lOm>i4?#+GEu@H0*68iLbheG){XpFW+9Nv) ze|xq)>*mrha@DVUuWFjIDqh~pG7gHzflyoN7h02;tnKBVTxV=*O?E;&vAl?ryU?t4 zd$efwT}ljJaPIrRO6s&-e_inor)h0-sLdiH1`EI(eJho|&~*(p!rxj=Zd<3Agl4*b zhtaHFB!?$}Kf;|jMpC&llSuaYF79o#){!FyK)h$BYlrbhp(FSz=H^x1FP00gI-gqf zc8zWz5rM($Tpy1iw~tEHqLsu@`9No;Ijpe}gkZGlTxv~8Nj;7457@&7fv3xM!)Sp> zQ}n5>yh9vP1d9O&7!~GvmyM-OF3t!FyxvNxsvLpEW!aw)-#U{v(8nhtyCRZSGn|`k zzNca1U1hE}=bErs1 zP_>;ae5SN~BtO%nJ-SknRL+mY@@>);LHq72X4GQF74$(|{{V{4Ep1cQQC%b>7V-Z8 zx+pB??k0cuOYNcuz|n^F}1O_GmK~Pq*h!U zFgW1*S1Y0Ti(A%wHTLnMn64Rf8oYr+`{4Gk7`oLwK{ttKvYO(1TdS$ak|1yaC%3gr zscljV9+Tn^i7Vmn6VGcD3wLT1Z`jx&`}1CN;}05N_}@^swzsupi4G%$K4xR;YtYl- zhL&{llFKxgL~}82t?OLh!`(j9NcfL&VR})fi5Vd$yJU3t>s<8eMp`1}7PmP4N5Y!! zk!CKhwE1I%^=xNv*V4Cs9e7IH#Mc)WH>8_uau>^9H!=16tKBuhY}y^*ybI)cO}lqa zmGj4fd|@Yq@23VMHI(P(KyXGn5m#ADm8g>&UsJL1_rtwD(?*k1($mO$ak`<&Zc`PQKP@BPvBsZ_5DnFyl`pFvi9Wq52Z($oNqMp+c_2cfQc zHBrDP0Sjh4Gu}iLK zBoE^uo4s|&yA^;tl$?WJN8?7i8vf?<0{;NZBmNbu;HTA}R*cAQJggry$J%&|l=$F*C$)7Hw_3rw#g9*f$(dNgM!xuv2x?bM(?@Tpmz z-b}IG6CeBLq8&%ptw|XqhUi6l`8MDMnb@xdS7b_zbGz2{H(J)}*eAk-zD?<7E3)PNL`&~aZqSjg7* z5zL%CY#1MLUqfhr7xZV*V>jYh?<0&WAX0abI_9&Llhl1QJcq=d61ec@zdoAoB=Yc8 zhjl!U(zAZZP>hP~wcT`hgIb?ZkwS0b6!uZTu5SKS4TY(UxrmY=`>DyOnP>bD@bAQSdT-mMvOZon6b+UG>DHoAwxUsN=APAdas^>uqQ{&X-o3c; zkdB0Qu3Glv%wh#qIJ#4Y)VzxZ^e4 zSZT1B2g)g2r4DV+Jw6_oChfHfSeRS)gE+5Kdlr>yrA;+<({>1Cni;bnw&W_FGBW+(zB1{aa=C`yxEYB$1 zyjNs2H1dWPmW-tvo@qCSnd8cJs>`4}V{zoyt1gg=nA=t7&>Bd|jAIo@e2DWmjHG2b zthsHsX1m+13P8u@`qq1B>W2)yY1}zkz)5D?%_3Mx$BON4w1h>-TAD_5XF!MPdX`6XP%w#zDVy~w}&+twy0t~^HE)Rf6R@rwas1W_j-JVp<)2- zRq-HKgFS}ZQMB9l*Qc#vc%)j-Xw10IJ6D(uZvHpgS-7iGYu8glvCkN(RN8kb)J<5S zIt7)n>rb<{GRiU9sV%0c>7>Clb0D87#x#>4@_lV3hv9rIl0s?5T^k_{srq;T9<&f0~s zx776*oM4HM6*Dxa)W>ROYDXB<U{LvKdHPy&8FI!O8Aw2p5vn_kFYg$y%Y!BOID; zX*MxQr8vbhot%n5GDXH3lQ=Zv%8}_z9Ew0o?@NkB(D62@7D;%zo;TYk#Uok#Y*WS93ro{pTk0Dop zD_IgX1Ey&tRW)SE+{uwZO`@ud0j2ghBUu5I6g&F+;dW+QDG)5v4<6zJ+3o~)D29O0<`WDS2?-2 zMrt&?Wi@{4ER$K1L_y$HEleHJZEwixYE{%d>miYaAXm?Nc0%W7>gos9rc3_-dgr$? zifOrFMT%7!)ur(DtG3sGxWTR}JGUeX-_h8-%2*OJ>xyb1>K$zNEgS&)*E6cjTrjU+ z&@|m6RMSPg(1OgkCwC&U^)H9;!0e!hjQz*;_BCyCoYI#u;Zj6H6HLYpdUf}LVTq0j z=hxP`EoZ~iSOUOhdevL0mGw71AKeRaah~;t-N3vHuX8}URr~U*QSyO6EK{;7vzb)0X$mOL!z1k8sIh>s?2Rym@K;opE7l z9ERpa*u~UKS{H`d4YM&TO5>8PKpVJ=n=tnJQgHiXYj6q?olPIpJ;XR z_MO|hRy8UmqgT14AlkZT6{Yy<{9YixWqWckb}u8|wYAR|Y00EaEoBnS=v7zKJ=P zv91b6zHpP~d~=R*UNv-?m=9B3&x3BYT^ind>qva-WB^Q_zMs~+>RCkPhOG2Yh#GWP zz7x9Ksk%R%SdK<3#q7kE8jLbQ_sakPuCK&CIk?xY{>2;~L?q?GIW6s2`X`eXyKeVb z*u3ES(Mi4bCU!liOPWz;hC#Fh1n1VcPaWG0bHm!IhuFjdqh34eLf+?d#sF#4vMJ!6nH(>cYPj zjqc8lTPqXI$QZ4U4no%RA2CX?VObWTrn-tT-K$0m=}xk-Cw0$m>drT4=kFE^OhwwR>xr)<~RT)busy$_`U(()Q41 zk*HZ}I!(RZlcBbB1$LekW2JaU#GO_Nej%;qn&|Cp6f-YE!?*LVQ1I7+A@KgAs7kU( zZRLO5MY!#aeKGk|x4s0?uQl5}N9^d03liu$N%gCBZ>6*sFH@aQiT0_biS8^eqnxh_ zP=zPh;=HcrU0+_;=hQ6i=6jcOCLmHuepU3ItEJgp=~oa>Z4*XgJZ&6tT>k*WT|Uy^ z!f?p~D?7VxlN`4+p+!O24pZhwjZLX|r(2pm9`54qSxML}*kJx66@K^Ow}^D>aXgk@ zYMwU2+sOO{drkGWmuKQ@8_6P7yqGj7!FuPP>08V|jOM8=*481dj~UW_6<=QIt3A!q z+Q{pbh^9#T(s)ndpqIp#vFnSzD>62s@FmYf^y0mh2y_Hjmup@zzRqYdv%JMgLKZklF-C5k*Lbl5kpvZvyuD+k1E3!7W<^u_aKP-%G+A)gb zylVQcsp07%(#V2Oyu_YxIlwgs!<}1P*R%;G)Z|cz#w2WE+Z-I!Q>_Pg4M!+7zK4`} zrdT{XaPcwbpt5vW{3+hiNQWAh(! zDjhFAcTR@AlV6_ReI@6HL}*zurB7<{+a}$qNDy^=4}NR62U2wt=T;{XPB6DC-1MIV zSSmgCu_)pQ{J8$7zv)unT$_IjO({8h$;tc%MP0j({t^EG20>|`-%P$^{ZDV!xs7`3 z7J7Y}oP>p0y@>#t+D`H!D$zXm#PPNL?WK|&xXhR*{QB2j;O#dys3(d{uNYIw@BJ%- z)GwKQaTI_P92iFW5zTt#giyl}nj<>K%jPZyPp7B7bx^gNF{yi)8up2)U&$_`9jhC6 zVyH`GfH)bheKd>EN=*!rFh3}CIB&+iV*5*m{{Tw2fZR(L%0C~ds3g$*TVBYk=-P^3c&p?-rE<`;ywh;*Re-|t?lit%%4K4_fv zv}LV7f7;h%{!?~hD@R#l_Gp$@jpPS{qX!jJM%pz1Tl=R68NsW1zD0~nB&Jz6{oK-wnJi+e|7v}YBe z3~pm>!!mB#PxlD0FhO2P2y3Ak0yLPcp(VIHw1VyV4p~27J{E>s6X2;(E0?)|s~p#|Uum(-q?OMV=KdwJ`!s~Acb*hyey3qma*(JewCAXi4cF;r~hTX?W( z?WjP91;q&(BwNTGvrcR1V~B%*MOFJ$PZ=u0x!qRxYY+}IK-qHY_9v9%9Mv0}Lo)9` zDxHnAmof*>$_;eU7+{D7#VeGxIP04!$`Q7!L!p_$^B1jlEj6@_!3Lt0;*A-X98%`6 za-7wVha-mYZI3OUmD}4v~T8s z`7)UR$I_};T`cRc*50aNpDJ)`I@3=JAY@)CbStT1NL3;y*17FU+8Ggv12xiLsYr2L z)!=4B&sx@=##J{s`Sjr>&eUE>r(Ae);yAp`yjM@A=$9`l+?bD-o}^b{d8bGHm%9TX zRZ2}VBO9JAaivQXMe)F@F-T&N5OIq1Yx_8w%1m*^dED{snTNencP3G>bM5kanktVp zRajSN+s29w6v+xGqJRo0qJRuiQqoWWR34)>T5AX<`_YX0)tX*tO`x)7JLu}H40!8b zF!<&nYvWx)Ffse?N8l^#U@PTsi4lLrajCHD=YK(3DK@nyl_#OY&bwqHt2bJ^d^XL` zaa0`A0$8L9rjg$@32mwk1+z-yu(_n9%_eDFW9c!2NC=>kcoYC`%V1Ma3`IQ?y*f}1 z)*b2CLr~zE>f86%9Ls^DRTS!M$xHCXk8H@ zmr#{aM;*<2Z-x9OsH#eSVuXit3qbIDIahSa^)=~QPluA-7v&kP%|A#;Pym2q+O>oV zi@>cGggs2y?bameP)~AU&1p)$m78;Kyc0l*qfFNE*0L@wijqcZ1eK802(IHEm@Cb)w)lRXwXV;9|7h!K}--z$T@Q8J7qMHIHyiXCT&v#m)&8lXW)i z^I5XROQXA!SE0D~u6)WvRfsMtDxU})JTv>%=6w?1h_Th!%BtFm%U?A?lVxl zpo*Z^6vnvXy$GH6A9E)Lv1e?WyF9rSpCRU~S0e)^o(3os%&2`0Ef_bapTmoPOGk20|7yE~m9i7w)~ z)ufUSxr}Glv+ZV(5*3+#hZNh$`7~^hg0{{Y@8ih5StoKDG(W#>AH@Z0@+^LPaqjab1TFus~nYdsP ze(EnlS>WQCa^JXWIU{~+YF{(8j&9|Fxfu1YM(`(yEp*5ww7wtc^W)~maDD#(I`c_? zFTQuBeUP9J(Cxq;W%CX~WCXF6QaDga`Bg01CC?6wa*gkGizCWu>8~ z;xp!JmNE6fC;3-HY4=oDJE?y6#ZY(KANW^EY%m8D(2TGH_{E3)G5-Jx&}}5XrS{P0 zDy4SS7=}DzeSEL*u9^N>m>!gwOJW;hO6+d6ddn$s#Zyof@+$q2Q*ehPtxG8$&!Zy$ z0AkvFqnMPDk7}WP;y85w02A5iGX+(;jhS4Yj0ofPtmw7rwEqAMtD2^*F9u?J^Q%C!^=qx$4Gq*;>Z2lvd}_zA?U7 zd?|iJ!5GJYLBZ-O--aQ#weZ%T6|f#pkfagcjmvf`z5X4?m+of#&-QGsj9oraY|9y9^E#p3Gw85Y|S1;v63))?#TZD8r#%oTWvaN zUBU%YcOKR9^~Se*KB9FAuB2~4c+{+Mi1R$G5S>~Jy&saqv+iV z^>1{TB;FcLfIrv0bG|9Eg6G6~Rp?OiS{B?#p{lwQx7d`2BO|qG?NV}i&#iPSO5Cqv zdEV%^d39-TVf5dAo67Ks=)gySIlHuls)MswuNd$g1*y^aCV{z?X z1>zjXuDJt@)^WnZ3D~DvQg%HD;!lpPd_!jyhL{+^9HLg0l!8dk0Q~86zlv8Yqb0PK ziq}ZHM7iMm;=IypuiXLd0PR2$zfjvu!z6J^R4K`O!&MsC=v%~^mYJ!`sM}q#T*}3y zBco^9qm#uNw~2L&ajD%IB17`VI8j{u&eA_7q_$HW%InolIK~l@m5G#_Ry|lijixdO zINQL_6^Y>OC~6)qwu2dsT)Uq@Kgg=q+OxHu!cQW)&j3QUmRhd0CO}p4WA`VM{HwNd zPB%s|P0Gxd;uflBzA?nWq-Sm`pVVbCNgFUDF(>k@?Ox^Qxt2}&1murep89;+MCYit zT>k*yHO!8NaCVBfy56Mcab=q;fzSh9#iX&1PE_;QM*BewI&qrv9|nUaxhlMy1;*o& zI`f=Yvg!MbcJZvSd7H-MBhV3DlvSg1o_n1$K=MN#7_izoQU+_7_=P!#!%@Hq4K$%o z(+0amU9liI7*W!=4~Y-HTMKY{D>TVRrK|)tfgr$k)<6jV>QcPMDoaT?~-ed z)g$t!Cmrjq($jsv;aEN*x7ta?dfiRaIpH8D6iFFnCaR%6ZD@bXI;18+XF0UGS01(L zdN#Ebw!s*KkzQQ=O-*a4PilTfIHAViosRcik{!F38Lmd-Rd|U2)KcnGTrLSULTK3X z)F{{)ESiL~{pyWl+%jr7W6NU|(CL~L&HE?_Leq9YlV)v}nJw(WV^T-0cls}bSz+Gk zoR3Q1@ZW`!)fJs9v`ch}FrvBVCiF(Jg6eRV`ed4Q(uGru*JlFx&;Y=6rIOtwe3hxd z5XNgXT2?aee7jhirfZM5*7n?$LyQ{S*5+vu!q*{fWYS(``N^%}9W^qkJ8WtH0BT%2 zt_i^)SFq^%1(NA-G|~C8NgAs^h9j19J)kWFk}XxD0$I0qaG z92M@-nyA{bpQ%WiZnG_x;zHqueF)AecZXcp8?0&~+3%qH8& zs$L+~mrS-Gox%N996QT-uoH^rgl{Vw$?~h2mipz*_3_)BX1l#LoZG2- z;Ji6&{*lNXYrfts;R7A2sz%R3X}+whs@|&nxvZ98$+&c?nyltqjt>;EG4d(FXeqn1 zsjox&^Xi(*qKI?KvQeDe47Iogo9SO5tvF z*%=y9T}w?YsA4B8o|T~Xkwv5yC^L@rbjX?JoG3_A!wH3yhb!u|)?q1BCigybYNC8RosF|{JPjJJ6 z2>ObMsXa|50vDziH6V;rgnojM$U$?^RME#HcK}q$5%j03nAkQmj^4EFE0RM3-~+gt ztiue*dm7Z6R2L3~qeKWWHr8#^tIl0I5l!qaq&dY%r2ha7F=Hp3U>@~q9f&e{q%jf# z!!Bu7*v)VwkCUFX=9Q%3Mg?^~7SX)nmCyf{3xET4^fmY=7j5s#M@Peap| z2ovS5fMHh@?jRUZ#YFOAl37BEqbCotgD!Dk_BEe zgHXyt149{*WGKe0Lv}|Qs@E&Dnyj*IZfZc<^YS{?JC`oB$mIlMG_Qg&O5-puNep?c zwwXGmJGuD5}bHOCJWNWdu}vaBC^-4QOjvoQk(0BB^Qqm3m?FSB*U;X9sDO zUR-ym9&$U>Q$7i*l1N2*P}ua7(6rO%y+&LbrRQ@_S##dBu&!DUig%e+JkNSO&q|Wh zEp8OX!Kg%;>rUUr1zZ}SkdP`ykW>;l^`?U3H*;1X0f?$4@l)H*15*TUK??E3S`iAN zZ*UFO9NOKyQ!S52l&5DsgdL5^zFH)bZqc0xKHU1{B(p`%^y}JrcTa4(IH$Q;NKh`ZLtLJ8f&I4a^medK%T$d_QX(Mo5q=dshwNi<^k9_QX$K zmFRK{yT}5SJt}b&WU5^kN~|0;M++(P*r_~@DUs#QMQCdpCEb}avKB&f!0A=2qR6b~ zoOd&nh~yvldUqISuc9yWJ zEKA3=V|bH6ww+Q>EPSSvQrQHryFBkowDRq@S4C{1PY>Pla;j@W`$>-9P2U=U#b)U) z8%exYJaP>x>eQJl;ymY9)o$*!m3!G3qTnKSRqQJV#P<*O!4eg)waH%U_SaCSQ6lgG^sTO}V`kekUM!M38mqCD_TZYj9692*G<`Kno2!K2 zeZ@5mT+%JA40&w?7I8BSNXME)x5P!r;0oTF&2A<_$Bo3}*0VJmg#g7E8$rPo0hB=N zRi*=WYSRtO z=WkxQZbsA$+LQ%nX1Eu zsceof5cbVUN3!X5kxQgPSwSRbxqlaUZVQW7XyY4zJt}J-5DR2WQgMM>!Nu*i;xZUi zDf2rRw|8UayR9l{ZrvCWgG!c82i>nt@lKNr7mCWbEm#kyt4482=*E*c$*~_=s$_ON zbgq)?N)Fh?J`D(C?$o&k$2GCL)r%O~JEV{fmC-hZ&lJYf{?FQyD#uTz__1v5Lc4So zV0=M?emB<>JVVMnS2pV3YS=w%Xvvb|dfn%YBf7RKt(=PU%}{-!<)uC9dpma#4Cbbi z*s#DAl&U+}sH2^n05MLrT;`O4ZT)HQ`$=Dm)o>Zj^eljhCxij>ieXqO-Z|_3b6%YT#~e;1v~QXKREFu$S9L#x<(}n@ zQePE+YRGIgO*Je_e(f4$4UfdvE8j4gs#knMP(&fPtu?(SDqw~DvAfLDi%RphQ=u$ z@m1+#+}EmCm@;F=I$-*m^!o8LZ{kPLMZ$fGOMl|sxQ|g%YO=>2gCNNBFaWJ#3SU&em*Ow7Vp{We5dQ#ZR!_t;g+E%? zvi|^&*hl+yi}4$Pf8MIfehC#vW1Z0F%5A22f)MeN2Wo>v4Y+L?`d016k$HZ*BZ;tk zkzTV}o80qcwLWpk6fP_+Hp-r z#%$=dw281m8L2d8K29@Sw7P}W*}}ym4`G_!(dCjj{{VQ5r`%Lkj)t@zrtGpZaw}6s z@k?oOv@*hEjGV3x2&^e4^9BW1$2qQRYIL(-Th-*&ZUedmjRqBcDuujYwMGsFO>W;k zT2>^VZr*~od@CGtYKP6i+PP4DYb{hTsi5C)GR+GWb?8MVuCy9DZyMTL&*5vsA}-`m zRk6_ISI(10%PSqd>({(Rs$O5)`7=rTzwWO9*DG(LB>5$4YFE2v6*Ng4Ah+Erv0nA) zej1&U1p~D;yf%uvg4V8$Vz7<4r&Zkr7Orng;Tzp&PxF4#ke6JlvgbaP(zRDZEhKGhJ=&#nX0<;<mFu%}P6Kyqk1Q z?!Vvmrrm=}9x>-Xtrhh5{42i+{{YY$56!>EiqoE7%3)>P>D|Jd6HN1$)KyFNB6Y|2 zf2pdQkF9Q@b#o^>Gh5GJT0HditewZ!jGtP3g6B29(4MrOdivI0<4OsmkbLH+p1!o6 za{AV8O(A1PAak02a{AKw%jzpHaiF_V$Q{}$0O$LyY=xIK0Lr4kKj^?IW6-NU#P$Zp!rwHWkaX=PK=e7kx(v#HYo8^w| zX;ot#NvCiO=9itzo|)-R3mYCA6s1a$o<@4m2Bx*P7K#LoRYLB_>T5MQHG5f}JBygw zHVgs)aarSv5Et!XGtRN7>t4a(9V{eD#d}wq_)^8LZiA8ZuXE7sgf{3g+L#+Q(TL*| zvdxNDV2O&&xt|&5ngk^GJ$bD8ZaAoCpRG`mb5tVR+|(^`%~5%(KQ^7fzvuhX8Tji| zJi|wuMZlhHFwJF73bkGHGCI{unoMMIPD2iBK_v{eh|E>0<+RzrsTpGB11aLLZe!Zj z(KpPslY162MJ6=N%u`6I2U;0;tlAkhl)EIstr=wnI@Sn~$0b=e1g~21>(lq{VI^tY zj^Xmgsz})s>7SEP6OmnX{u9xIx+_R)qoO;Dy*3M&3Niu> za_!*w%%^9{E3Go&MrqNI$0E79S({2KM>}cY+eBc}D0A;yQRvpOL$M-J_*MPEMK$M8 zq|#0F1sJ2pz93mers|fHqTsXOfnJB~W;!`d_!&kMUj zX_*3b8OJph({!wb_K%+Lnkid&Y_aQ_qa;ZnI3F;nFCbX%83cissA1ez--0|mo+0rh zc9R$TI)ctu+OK^H<+CO5=Yn-zPTn0yYORreS&uc`Po-N!9@5+nFtyp=M`xsIOKlKX zZK4<%I0Cr4?+|GsRf%p9aM&g92T1*k7*o(T*T2&X(35e{c0{~i%^7*kOuEs zzin{gwu73+SqU5(mMNVi3;{n%tA&delDg~H5=-)hUA5LN>~caCTRd@Gjv3{5BO8uQ zS&DXy@OTuPxYmaIH~|kQ-+bJ#C+k=A{WDb7UuA`?vii7Fw0<-N&Yl?! z#o|gEJ-Dw*(pA2~pGx!1J4w0H(IdNzhaE`dSDshBO9$o7S}!IKuD-YNU1-bI3fX9lh~g9fyUl?htvu zAo^Brn{3Wny$oC+&P8Y4BD9zs*Q7z+KT^Flv0ZLR+1D7SW4g#R~$1(xua&xV(uLD9O!r z7Jdo6ln0$%N7EIDv$>}-x#qg$O0k@EHHBlM>Nn4edubaduHGx?OJ4!lQzDmP>t2YkMw|r#oxsmP zD6$=na(i1}59t#KRT4=$oL7%{iuAmtvI!?7iu6Aj>h{vRMr8+Y8$qup)n)Ufas~nD zDjB3v>7NOq4W(XO9;RXX16*Im83djvYjQGVR8T+QHRvAj z{{R_o1b#Asi0QE7{sT@sibfuSt#5RJ<00caNC%qqt0`fTU@j*pLq>Q%QT(gVY<#Pr zE}#V&N#BOaZhF^cr^;rwZz6ejpbij@aysLVd)I6MxD$*BlF%4ME&wc0!d$;WQq z)s1_Fx_zba-2CI7YfbJ9a=WmT5&#&hCp-eTbQK9GEN!%9c-lj?0!`}wt*eH0*ki028 zS~l8jHtGOCea%O*Db+Z1(XMU?2?+z#)be;LX`?xp2OTTC5vuf2PQ_Y;Q4VdMKdE>o z>e(2W{KfVa!0OstK_=fVv!0l*rrsrSh5(AAE{9_??t5s-^&+)R7|Jc5Irx9VI+m-b z0pvVzgZGH9Thl%s*~1$7pt<+1)*Uw9-2U!0U!bc&c&c+<1fkr;w(#AJx&=rOeiX9! ze$GWdx>5L6%@sV&l%kf$7k%Ma)E0{)*jGC~p71$VQZembsK^))MNzrZ;hY`FGg9K2 zYE2$w@qo0;f4^OY{{V)Af@9*aTSvOKmm4<^rDovKYG_!qinOC5q_@-NlLKxmQVlBr zZlt2v+>$v>QAxm1N%pBU%SlRdgjZU0xQQ82_|(zaq+@_9XvW)^P4hDi>=?vyDz&w% zMJeLCmW_7f6-dInj}=mfG87*xxzFkfe8I(7wNd2?(w@y_>M8?hQEX47Thx5fDS9tr||XFrNg?Z zWi`d>9vHjU_0thk6lu#b6{3sgGL=Ze9L%kt-9MJ6oQxXuzxY^lMWmZ!Yz*Q(f=uz# z8LmY;i=~T09Z~BYKc;AMN_RA#^~mtb8)Hfa!WH7C*L*H-&wr& zAyElGO4YFNq*n3=DbKxhQR&DODO#^$SGG94GenT87Z}HC>NM+FblbGKgk^xi2cZ?K z9ikxNMNj1bI#qiZ^&{Fb{{Y)}KlAE8@2V-A549Binio4Xbr1OXU-y5hs;d8I9ij_04X`vc38X}{VT>fe~E_YstDP-+Uw=7ouG3k(3il~*$pOloB- zjFZhLy&-aGl(c|c&{~p=P%#OqqLHz+xfLBQXb~$7t)E(Qg3J_mrW2LTUeX{pH^{ii z2Q<(gm*9OkDn`Utt6G(gAS;US<+s`GkmI#2k(#Q^MJs_6 zoEjpRYtl$^GgZrA&<5->hFXg4Ln*2@*Erysw$2?X0g~ekX~1Ms&M?H(X!xla46z;D zQkdIjaaKSMYS)lFt}D;PO3cQqvCqy(G{tPyjY2)7)fnHJ(w4D1C`$1pahj~_R*A@{ zS$VBZnOkiksmk+AWu{9a4b4u3Rc1Ie^d^8(ERr!%6O&CZic~o?i@7vWmNjTZRcZFI zgI1nDmbHX#V^U^uI5i|q%{E2(snCvVbgXkXa_Tn~ENA8w05}yaC+4d}n$U0sYw7m} zIgzu{u_A5-UAL1CwL~7bq}?pG%5&6KnlLR~PJ^gV6A1 zOnOuro!qm65Pb4BKMK~gvKBU0%)_DU?@FEbHm;(yCe@ZZ_FGw(dPv|E^{bP36G1wy zv=LY;G7r2em4O#tCbPF3{Ke1EfBNgMjzRwb2|}-`pXpl9W0nOk5a{}Dli@r0Beq4^ z2>|WpyvxJz+clI%P&Xbi?_Rm$wKtj_)so|M$Ux0_zLVxz+@un&8aBZd&W)MYG)RSZ zt_3z&*tf|imVHGkT}LS@c=RTg<_C4$jm1$VIg;e52ID`4U|-moP9)mc_O0b|PR2hq zK`I7)tT1X>NaN+vM(oDKdsO!N9MVXUf-=Vgis;tVZW$G+q1k<>T!{eT4u2|?*;Xvr zd?KEBl1rUUBLrlqOq%L6?}dIIv?})=ZPSjs*kkentN8a`zwrK%Z>ZUYF|3|ks(a_B z@~$!;i+5!{c!qwpw<(i$J#{=S;rNsU((S+I?4uRpE%9dPoCw8R zw)lssU0{SzDX4jio5=O6%UgL+?nOU^aJtrPr`v2cliIU?;WN}&V1=yT?aA)-XiVS* z&o!iC*5Yj)HJ^trwPX@pxdYO=3m*sB+hDSWKJ{}-zqk7`o$J!CjZ)XNtDmhyz4Sn> z)r$_7Zw@xbqwuR~3aH=;sx>=v{{Srq=T836fd2r?LHX3RB9leD3S0`PFW@|HADuW| z#CkCN=~<5GE}~b3fgP*D^-TeFKZs8lA0mvLSGQf>#T*R9hB01Kr0RC(;uX^t1TetO zX$7XHbKM>0gWrm5dy%A4P@ot@` zUdwMR$gGD4CaJfNH0C%-)9^Lo-Zr~QuO3BmSe%odwSlKf*%@<>ttl%Yh2H1Uu=uvp zbAc-Ub$3s|-(+C82nw|>tOoc#c=pl;;?GV9TaBD^2t z)()Yh9_AMQl+iC^%ItLutv-0>x0#iY?#E+YwYQILrdNC0NuszU5Z29w_uzwGe4yr} zR|BnRrzq@4EMDiLYF=TZd;0n^xZ(mT;BD$1epE(u4q^onIq)^+& zT^R0g4&W=X@O*+@4{7Aa{|N0ERjky=ut3VKskDI}2xh`dK{ zqm0Vn9QxNk55yZe58uUk--~s#tLu{d=gUKYD=s1ED@Qv!At>KgdmsEKHbryh56+=J zIJIXABpD|FX1u750L^W9dqTXv)Z%ruk)j7Bh&ZIGH_)z6GuX7*p7J3lI3pso8WsYd zrrDS+kx0e|O3?DfTagMTVnFPA8kuaf_p0zvJ5#E8nOvnrv58M3Cafwdi~t2bidSQ> zl(bPuVMP>B0Ywy00ZEElC;_;n#Xgf10NM>MDF~(^r9%uFf$dR`-4$SvDKkwmXjK-d zzr%`S$Fy)N;l)EdsSIV@@F}JR24kLSGghH{S@6O@#~IoULX+FvPSGIRC(P2NdsNTINV6(w89h6|Qa5&{S`1q;#ju9IVMO#K$ZvR?I~j z05B>Cv*U_ncFwPynjWC8YXT2SQUU8!8QLM568I7GR9LH)B({yg1XUrVLFTjU89nOi zT}J&4NEjaUU5ZgsTA6oRWNv{%;~dm*Y2giQ;2zWk9+d7132ag}pbiBmk&3)W6%oLq z;CWcHCn9Sf3IKnf2pd} zndednDv{F*$fC^en(V9aRib^v;i4fOs;H@mlD3OM9Ty2yMadeTW z&P{t)gY=;cC`#g~0n%IBh+t3$Y6;6#+0-O;K~7qz3HmK(T)!r-$k4GGAQN>cH5O?@Pz1o!!K&fu;phSExvI0L85L<-nIOnC0hq_NSC@*nDs5_k z#zg>C5*0OW80-jRMJeP|F`g=H@m)u9 zV_4##cNDWR6;!z=6-iu)I}cUf)qWdNS2(J+iH)8vwNd_?C)U0D#xt~b5|Pk@Up?wmhP;2RWVba>-5P(?;iAImuVv! z8pQ3i{Z{Jk%|oBzu2WpsH7zj7BrE_1w(X|-HJ~7LTJzr!YOi?m#0OUGQr(d3SWk$W z*+1(=Xu;!MX7LIc1`nli0>Wb+;0Uen4;78EE;2o9CnwO>G}ZMVbv?wFUTmyPnd?_q z#2S=Iu_qt_#Y3w+M&uO0^{Qpn_y&`1+7Fdnxh}JRc^OG&C*G?0^^x=)2X`5xl(MvbF9s-Cr{sXkOhgyfx~uAE)9BDB+b7kmi|UTe4d zkr|2=^3?mE{{U5PT1a2TIVZ6Kt@vwBO)o(p0HywB_pF;aw)lly9t#oltCv$Kt61nY zpuSXYK{zOBg`ZKJ5z1$ znkVw71P@bFTRB+@F~({0#H5hon2Ohy;En2VdV^J46I+I1hf|FE)$}1qiTP90H3@;> zcR1e*bv~6WXn;zl%)5WA2Voxd)%ZV7F~cR(kD0g?$y&e!&QIOlo@>(dWsS6`8a!<) zk&01Tjo!x>@qXiHg{`fJS!V=(0RI4uc-q3)?_R0#Kg`uN?LSL|6XhXx$G$5fKL$!s zK?Hi%x3H94k3Bg@8Lp>CxQ}Q6im|PDNv&;Jz#xwG$=J;`rOwjaW}If5CUo3R?c`AsdK%B#G>i5ryBRCuZE7+zaQ^^}bUJpw zsml;&oc?vz_>01}Nv1QbDy#sVrkwsH)L>zW+M5rH^_i|x z2ppGVfGalk0~x_NCc6Is252#(E$qveBaHT>s&@xU`XqFEZi}nx7U17punaO(RyT-z z7}``%s98G+kZ$L_b&~kAPtzq^d8S2>_RUt;yhCksrbr`T<(GgfeOZc0T^>mP01@?g z<1ln9^5(Fv{8OvmIWKhJdSD|mK=Rza|?z7PREvGiCeI2mkG4o32qOW?_@phwWdw6Ui zAlusyE;Jng9A>h#OQ|hvDa?u~_*h*J9k;h)yGufBP+3Hutcul#xbAk((|LY;>+@T}=9<3hG!pV4q6qS4mraPTd*L zl=6VzOxBc_R$*PGqCIBKrXC#Agv7Ls3X}N?@ehsKj8{G*vayVYYgj*ceF?$EdY6y< zZDHZPJ?(V&N$+EiY5;Pvj-30~%(s`98kOX?cT%*I%Eg_Jb4@O!Hmzb2#TcavfldOg zA_k4aqY?Bq+4u(QYu#D5alL=f<6MXUAttQpI*Vy~oHt9+P8fQM8U)WeO9i#r0)R2s%j4|{btCx{?{6max+xx!1 zg>`RbB#o0KME?MIaDSaJ9d0uCRW;IEqKiYL8xDJHw8v&y^r^^y&R+y%vI6v&yEHPqc11 z$vm3Zw38?@b5LAFf%p{2$zBCyozZPso|oV)Hp=pIHNq^64gfXmHny-s7!k&c$6;SM z==Yb_ZX~pkk-e+3)BJU-S~81=5)W}&rD&68dobWpTrIc6YXcB7uneD-9IRaE^!(D-1!wU?N4)<4R$HG@TUOpI zvAP7Ko+#vHCUkkCx%vE6V{W;UObV9M#Fkf!n4|`dMlN)1=900mJV9$@h2?w-qif?$ zHtovAFeFm2?sW9Ekaf#on~U*G{{V!J$B*AFK1MENZBbc~YZlXe$pNb(_98f8OO=l3 zwMA4n(63hloR^Tv_go*wi!ixnkc`w#aEdZWtm*tkr`yUSmf|&Y&=#d;Gfkt-Rh~Ii zW;6Izch{`ZHyKwST8Vr`r^>k#rarYw=vKK-?^U?7{{UGT$EdDj#hxOXJu)YZ;Ij@r zt6N;ut*%pKGAkD2gI;&zJHMVASk6j-KNC%|ZDXynzPO9UdNg724b(Yt@3%FrJk4#b z7g9i0Al4(Hx77ST?TyPC1OEVg*2b-mv|Knn8#SZ|nI!%o3qC#S^g8M?4r_(fZWbaO zu;#19tRO-2kIu2?lQ&UEp~q*W?Y7u=; zOFZb* z8uSsVUo}H|W~wHnlIhvr4;ALx!Wp%7VIu*%v8%k14I|TG)$ccMMclti)Lnj9;@t7I zWN34uY3Lk|1!%q047kXuQIleBD6J=t?Vy)ris1bXSsxHI{hLoDun$j zj@DAxPa6EYLk!cjku9e5J$Fv=5B92qYYVa-j79}*MQtMu(jUNAE8)FsO0?6VM!SJd zI&T%zF2AObKhojB@4QlPu$P&ju_}(0CNWgv)a@qwlg8ghs{(Y#-W3fpv?ZdV^Ox3& zRtuFc{{Yvo`@huHYO{a-y?@>QrmEKQ*veatewBr)kCCf&x>hErKX%nfgU@wAiH&EW z=Cn0lCN-1~HH%uB+e1nx#sygiIIP%TCl#XyIHrKq(*`6J+1MZ5isG&0@<=$XNW4iM zumQpRt3>oOY256=;+jqm70fS;Y_ZGZ>rFm1u+RF`AB|E#^gMgxCYc3?irRPv-?Sii zQR~fln;P{$icfQ@>WitvZf-0B1uml@Pd#y7XClB)%8ZB8HH7yyhN!A(wIPepns1iX z7c6ZiY1=5-Kn%?so|N%`dQra}C;p{_2dOm+@_o;<~1}-ySnu{qlxpHP3~-U5KHhGa}%M znUBj+0kqSZhAYsq=|$*4Dd5x!xivd;O$%B@8BLa~ht4adfPn}V%-bENwd1+Oc+GL* zRpfBiS7lvG4E$9%Q;|~K%f3@f3iqw>>d4ZXvLug(6{ir&R%qd?QrnfTv{NhGE-Jn` ztr_CTsw}u_Y!J@$^f;#ClpIh6lxn+f*u`d{=C4BCDWNlR&AwoLE4b0-2g)7!t_sn( z71-#+`LP;u*pfdQw(T-O+-AOfy*avnt$Y6fjLvk)kFl?xuDQB@tzx${UkVg|tFEs? z8>|?2u2?P-y1fMLv^8+}kUxc*dPG>eu8Q&^V^9Zr!_!?><9BN5A)YCn6C*jR#Xosm zcseqiU5n=2{ezV`+gydNh$Xp}Srd)iX1zdKDYE=4m%2f2t`t;CRzxKEta4hfkz~6_ z0=cKXRgjp@DgGzDk*^tC5C=-Yj$*`fN(oxyb8T4N^WIaqu`~(79!)_9G1L=7EQIaO zD?~}uT1ym`{E&F8>2SAOSEg$+54FO(LmjoDV-w49x${$PM(tx}&B>Mpn3E83xD&t? z*!*RZ2^Yg@Zb5`hHf`ySE0TDE!CT)E08JmkQIW_aIs7TqS_+%g($#epxA7*KrbQtV zNg-)adVAKsodj2!tkKB1WN(yLpUHCfe;A9Ni78R+1$sw@Ksw#)<18AIDV>bt%F}j^ zFt`=x_xC?yb8(y$VO`#-bRx9ma@_@a-n*BzBdKPj-hj6~6w0?2Fw8j;AYoibiCv75 zs30=tx}7#n{u;4v>lg>}u3N-P7WHI32&_#GeI}AQRKHQ**0%62A{$rLPTspX1 z0awy5Hql*4487^05ZolY3b5HK@-yC_4A1svx=p1|%Oq#krjBV73v(ZvANPk*Q(bBC zEu5DYvI4STZ1<$bjFlU)9@i@&*||qTYtnQJSuAvWdud0{674d0dl<>5Vjzg<6hI$8(@SZY=j? z=BeK$)OWj8+Zpw(j{w`ta|syYx{J>Zqud7D&2vskO2+Eaim_it(;DkeQy|GAwe$}U z{{UxQys9yinzIOxOtHfpU{+Fi=E51wZr~BcLQSIOD8;Lt2Z$tVi(v>SkzA0`zRPU# zjOVR&RyVfVs)Z0zjGUUOMYeRSa`PjQ)UDq7)UjUF--o&c z7y9cZ$jYZhJ?f(Pk`F1P$mX;>BYWiQsXQ%!Pc)U|xiV3b({r)-wEAy{JQ|QfsJ9mm z6(byHgUxx^vX4sehMyBp3rjHbol(I%o00tL?~Ct`js7GUA1|lrkg<>soD38B;7-xrAt=`OO*5ZFGiPn{+kmf2y8IHt@yQ9)R)X4j^eVu zB7r}TF0BX6`=9_Fjx$<54-wqxG06DdnMV97sl{HzRc>0dCsT>81;BigSCV`4TwJz> zHy{oLdLF+dP(aG345VHy4GSBx*mqQNIoOZxby7eY?@wrY2sZv{z+cV+CdtYB>bTJRd{?;bAM%Z6uNiW;gUV8=6&Ym8sfY( z$q4e>jJ9jJTNj=!@u6x`xlyy!=VeAM>q`__bzE{ybozIN?(bzA(5h#iE2xjbu-hWI zOp(q_5Tz+1Csr|JULVk-`x>*SCz_|?9X2Rbl;aiBqcyZ1T8@IVFSPmZqGoJ`?^!-; z*zD$mj)$H~{1pV6Wv-5O1}x)>v#8rGox>sH29rq5X>6s!Fs&S#5=pBY`m@655zl{G*YNs793ZYM$aS9)PFONGO6hz#;z{PU864uP zn&{7#%sV97IlT`?zE$jCq|Zjt zbvT|S`NnIjwvb4<2Q}kbZ`#$tS0O>KNYQlwU<%Ea0=pw7_dJQgMcErSn8ya0ryhFM zEkf`r6I8VASb5t?s(pz_>#j%oJ9FB+XT+LS_0Ne?WOvw&o;p{oo6j;VXQ-^%EMt|+ zmC315xaB^_C3x~jrx3*FZBWw%%&}ks1$I|jT1vRvPO{UYfMXSLq@LwMw=3Ajh7Yi( z+GS13qny_hG@shOD32VdeaGl4x$(uqTIk+Nw$lEU!+2uc#*$vjNp@|inwFL9iIlmu zXQ1k%{Vz*zWve=r>*if4JmR3#Mt`!L_1d)#v^MIO1JyzO1zb#|ay}t;7WoW%R~*vL z217P0w!XA`(x{-~qm#o-AmDcNsDw=<&OYAaV%*5pG}F4Scbe9R!{uecY-H8FN5hJi zEXvt!wJ&k4jt1{pnm5OsY6sPw=Bb+)LU{ zp?f914DYwoHDXzbY?G2Z)zs3)ooE%82Njy*l6`APJE2KfvbweGQH4u}RrIDEXHw*g zm#pilJgE0mEN<3iVd+j%utqm7P2vqncsDJI!MxY*Oy4_}8RoPvB4Z0M9jlL>%S_qr zNy_MUvpStWQfte|qe63@Ij=*W%JHlt13QT9Ysqv1@M`w+F!=*>UjG24OBKz$eGf{w zMlK=D)70T-)3u8<6UT8X|BjRMnw@tQkNV+g?Y1Xy4ieRZiMd zf=@tNjQ;?Huf!kp5t@o$u^;XlDs(H^pGx2V0Iy&7f2pd~YX1QJdj9~s{Y_OBQP|2` zh}|m}R39T&+jOjLPwq8RB=fyhNto7Z#ajBP#Kx#54P^H==tmrQYSOn{VymL_R^p~6 zTNv9=%!3>b#24}>;S4C@QpAMZn(igk~Y^-a)-%SV<#+g zpHWS|Xx>jKFEOCpLFCswq}Imqj9%s|cuLkAlN`5lFcI#^7|HagHBS#(*jXd2j%DiL zo`bb@cM{s$n_KePSfn9==OY5B>r%%(rlR2+Y@rzo4o4=HImyXfsnd#!vF8&YNea|; zG^2Bz)J!p5EgXVY_l7j)yS-N{kPTs?7jV7JdcT4+byP+x??OO#It7q|**M1)t#dZT z#Vj!hq5yG$S-11x9%_VV+|Ep-X0BTDwTLC(KjU2$gjqmofgw+pv#$6Cwxvy}SeLEI zpbSf{rn6-0T5}AORAov`WkTH3cT=5tsEr^b-AWE>F-i_8fhK8^deVl-rnLY^=3|PD z#Z;DEt%_Oh0{5)hCoJ{cmWNW2 zry1Q+ZQ`oiHgQ+6@>YnMFonE}FTWKDFt3G7lv(J5zfJE1dU@ zM%^Y)rF{K&{{SCP^{=3Oc-wT@MSS^ob8!CvTF9lbAYxEgQKWIFlkHrpebHS`mT{+% z?Ntwr4g4&iN_wzz=>i6C?p;l3=qmruWEiHJL^DJ0`zDf{;~JYA<+O>AUL zZuYM?5&3fM?V7bOh%W9U*z8EE%EjMtse*kMKGK)mT|u&uU$6_CBs{FZV78O?GUo|WaqQ#-KS8p@-xv8^R{b7Mn>MDVq_m#ci= zg>#zk?^#Lhz^>j%{EZeURdKQ~Gx%2*sV?*_=xR_ip14h@IK^n`N6fZ$^yZ+tk9Dv< zlA&PxOUqw;~VZ|}7V$*`m=wwA~?FOcZZJTKv16B2OLuTj^gAj1o;2KhP zDJz#3$ChdL?WFuGq0)8u^=&?BqKL*2frhUuire=wBXao3uS4)=mg}d^Sm9XosZwmF zlyxrOUgKZ`sjYh?5vC$>Lf$j-p7o339}oSW)6Y^%=QW?D-56$9 z2as!|)Gx*Dfd?HiLZGy1id5XSxsBl6V&)s=wh}Qz*w?DQjV7lWyA$hPHQ`v#{3UV6 z?_W)6mdduN!0AyqD-^29K2&nric6*GE>j>XSI-)@p)J;>JhC}vE_&D0{vgxV?@yK? z{6LENv&CK)mq?yTIU^+1-EyK*la<-cG_9Vs*7!~<2;}p%822IQ2(D{PwKtb)nB%>B zm&3mbT1u^PcMyHQbxlm9Rd7YW7 zprH%g&G6OoSom{Qy^}3-eYj`aHR^u{mv4ZP$Qzz%7!O`edG^1jCXe8hQxJ*UY5?3f zPI#$wPZ|FJVE8`QNrFw10L9#O=bk8>RCX0Ncd4JM0e|A%L034r`T){b0)p)OK_*;MHX`xT> z@l|V)bI`@47R%5Ir*U-=`2z~8;z{mO*EYkN^DEtRRkokzjgK|GI3~9_Yt?afwuh() zVRPFw>xhwUSyvq^j?uMC>$vxrpBVJ5t9$8Om6aG{)K*RwlNBoNO(T=L@cUls4)V4a zJu2JyM$}sniZhP2p{(C6sZxWJN#V=K)U^yq-?~nGi_T^OYJB;Van5jDp-&@g|41!Z!*Af z&MTR`MusT|pf!d$K2*<3Dw0yM!e2fvAD;AE+)24Z1GbBJMb@Bv!^Y!VclHrMX}vpF zJqi+QnI7&p8%{sFk}N@g3^u*DVx6IDa!skcLm@?TF{F|{0XVHaHrh)$%%JtH{jUtxD?}EN_rShlTbZIE7TGQ z!wlBUOOyt@hr`|>Yu#icHyQP>T!GZvsRy-oCv|g@&1lBV%A%-=j^mEiJm9u!gz^hp zIi_#nQ&oE*9(Us{WNI3U+bPBl&?>?-R+iAfqm)tk)*?bR4Q=jNf2$>_(cC@>1_KkfnHs8YUfaq z5lQ*0-y^h+eI$X7SEX>{MS>go9n=i9I7gY-na_q85U1u9gQ)88TNe3; z=Tj|*7iK~IlAeN!q5l90EI-~KofV#jw`aEBf3IKnf2pdqZu|Xk{{VOTnx@PE=Bhgx z%W)lmTEx{Q0UEB{-Gm34{MyF9PX_kGMb?q^HJU1d9ogA6n=QDuXbqF zv^64ZnzbUC26Bn>H+6SOp+GZpETJby~9Zv?j&6I@D0S@tmO zqx_n+h?4$G$Z`=N-^uINs|$%Gw({2ldloD@^H?)zy+ydUGU<;XI4_nXwLy7sA4rNu zK%0P8=Z-4t{_?^oqTm(-*RN`@x>O4d?03vfym14Lm}8pFNk(eswsMQH=GU-(^ps@_ z^{n^Wy*aBFs*&6}M&zhGW~yepS;CDy4%M!;1Q^H$z2n0+A`4XGwRv}fG~Y5+kgjrT z+N>fFz@r_hpk%jF1dh3#M&mi=we=bB4h?deg~ukCg3Y|koO{=Ar^W0W=QTm+jY zyBl3iRw2zGH15a>d9Hf%1k|^hs^E;)Jeq)G)}6;;WJ;+jRE=^Qy)eqGDFuph$5mKj zv|j{OIaithuSEo6o0Lri4W=dd#_O6Iya~HII0CGyzP?OJbx1;;Rq6 zRu`oqmn}Z{u9nbzmCV}lAg-FsZ!HOW(UQ0!&seBdK-3UBv!UC zmCsnu^MPBUKu}hZ2^KBWajlrGPnDX?)0cdLy6COPk<+z#n5vwzIIBv`xZFdhczn#k>(?*0@bShFa_`fO}9+ zP-(r1D0t-yX)fozeEoO-03R*&uc3T%C^{Po`R?=nK3nTp#`e^x>hCAky=_c%UkM52SRIXN5o@G;XR6RzQp6w zyf?=-QW+#|y$);LEdKy~U-=OV90xQhmhTV6-qS0h0#c?-f2D%F+RJ>^yjJ|7?(H#E(Z-@3Z(OFI!QZvp?XFbb_-$UcgCcwpK zG{eqz2o!V8YzPI*x-jKmIIo%SZXJAU8x|*R$j9ejOvcVNCRG0bmW!H-y4bo}CQW;= zw1B=}XA7Q{##{`nk(%q87^0tU?b0wgC#kMl{A|2&js;2?mQ!>^D=H}smHIL4e@fSn zCR3jEPf@hq3jxTc$xPeAo3(ZOugI#tBbHl4F(4RoRkWQ#4=G@|o;c)>k-dP({A$L# zd2fAl1W~%UP;%;W593oUi;0`1T1N7jCSWnp8uafDBCU(49dK)av$R%`6lbk^4vN7M zlyC+pi01S)QXr?L7Y7kFIb;XDLw4XqYV2l@G4b5(1O?RMyl&>{?q+Padi2kUT7X!5 zr9E?ALX#gb*3r6+f@f2q_+sApd6PLO&{b`JON&jAEJWkpulQQ|*KjTev8tXcm_5il z!9IejDW!IEEAwh%!!b_vrDu9%B?;?TfwC#qQ*65fJ?K(8Vs>5^(jELw70Av9TKX2u zm9<}CUPa*DjVx`9n)W+zKF&QcS;^gsRMqZ^mp2&%9+l(Y6eLEsV7TdCm#favISOmY ze0?Ly5oV<=8ZH{an4SsJMa0qn0C|ml2cT)+VAFnHTNUF!4s6~-9NFL+_Smk{2t2Ua z7^tf*h0}$jey1aUCbOt(jTC9J-Fwv-?saP&abPG~6Txpv*1quovXWLs1*;MKS7~iv z`ktX8#EI8ESo#XU=}O3(U$=35v$q2oW9Tc0@pEYyx6~Fr_a#0_Jw-01@fZFQJw9!g zo=vMswLw2K4r`FrZe-Nv@?s<;5d|FjR9oI5&g(-zQt=Jmp$pqwZjuaxk3v1_oC0W_ zB~!owI#jV+kC?ey}_*;cF@gQ`j_uzc6X5@+6c)2bgFnU&kOqC)}Px} z>doQ8s}Yfo{?(m&8a9CdoYgAW)}zGP{{UoGIj+eo9LNnJ z#VsM?q+?EMGuEZ1HA@+85%Omgl6C1)N5m^ovB|A^5K8|5aQ^_hDhjaQalhRa0I5S9 zE1>Yjfz&mtOJUHqTk@<0>MPLx6WM*M#Intdh+!@KhxMXKxU{z>TNbr12!_0QKqM2= zy3YtG(Gy|ede!|zh}pRBR4`rw!-~$AG$UzlXHE{v=T$$5Z08#qMbFl3B z6;1_9X^J)h2fb^cSIy-s{YBMG$wGV zab68=ArgXnS8t+tjM~Hy6UBOwlv1(hNjC0!+p^=Ru1m(cf?HTD5jZ9JII1>YC41}W z?UbH*Z>3>tv%aaWrL0*ae$n*rM5T40mzdo)5e1gx_vYvDsv3l9jLfZu71dnn2@E3) zepLdv`|t#etY{Kq^!+s!@&Zbsi=0l#^Ss1JqZWOCgp@ik#Yk1Nk_~24fvj?%!csgsF%{fHi zXBev9Bhw^V3nv+&0ID! zdz`+ts(B&4YT)$^RpyY5)v4l{op)f@AsZPau4=M;);$VIC>PO-oRe0g9Mo__z^f{_ zHBc~_qh!2{*2HUpSvGa7t@75U42I?QgIUdYwD4->(aIXel?i&rq_3f+8S7(K*5iA1 ztt@Uz#j_htYexS7E-E5+JyEIQ<|%t1QUTgX&3TROlF6w}7W}guvkrh{*QMxyoA$?2 zq*t6=TKSr#;+aHjhbJf8@mD>O&uv)*(1WF1-V^(|WDt))2mb)BsH+x9WRCzK;|HM? z3(q8xBr7X49%ya2<2mM_ylCNPQJBuvpiC0Z3&kJVkaHJ$4blXrfi?O(d9Cd(lI`?_#!gQ zhzC4hq{uPO#T8M2^jB!`*X4-^i zs76MmKVEB=yqhHRTH1v9&o$0m$&zzG436NeoEp}G_D!8lbI8EWT@{Q<$g+Q^l zrAm>p?Np>orbg7vRn@mAsHc69p4D8)hXzp#&S~2-S3bx`YEQF~+Nm@{izqy26_0$F zCc0*XQzO zd3jo=aww9}mNL=5dsAMV!hV%bCfg{f?wc_mrFivYYKJV{3iN!&v!A7EUXD#=NnWlMe(fDJ_A5UoogW68e)T;?bea;E!qppza&c44+g7t= ze6<54UQJzzYLp{2elbi%h}N&bF^Z)hYP}C?XiV3#IAL9$ns-}(eQS%fJc{eIwjma& z>>`(phIEMqcpc~be81Mc_r>QrOrJ{d8`J)j{{ULcpyE0%a4V_P2ktK2eXZF3-hwgLeQ8nZjP9MCEQvA45{IWid{LR@`nKYQX9)waB5dC|l;nDjx`H3H_|i zDFI)lad(rv(~fu*rDk3v41|UrlqMEE?pv!Twi}9Ide@eGK%NP;EgVAXK$WwD^saV& zTTzg#ZE)d!wFaql(cJ0DwOp1hmd<-oC!tB*omPs={iIKDrnGdDTI9BUYA*xZM?Z?Z zxmR)++mFh(ZM4<4yqtmZf^aG)xJ2?@VnY{;$@H8b*1o2;B1_9@Kfe$)^2dq-$Kq)U zpZRDH>t9o7EGDtjCp{X~NhjzkCR_+_u2nUQXono~YX?#$Tgi5hlriVlwY8g@3r$5P zxtKGvZHW~AwNF_q4k3YgJhN1xob2+5BvM!doM4``nx?N|W2H%VG>RAu1kWV>D~j=i z7n-+;biE4JR*bkZJE`OjIX|6I@eSR?c9uRK&9u?1r58|B^GKt&TzXZLVV4PTUM)*f zv1gL-X5D4nBaIL7HZj+Z^;_+c={GAqjnYXXxD&vHv~H)kIXENh-n7?J)GS}hy-zU! z+qPBd^sIeQSV+En5iD~vfI;i{RW_|_6Rqxg;Q6VrUMwL?*qq(qb2q_Ng5 zH4Q#1sAN)(pn?r#>spLHLn^o=j+w0#UsEL4L&bg|-5GBP&#iC)a5q!Ilv~OzmncdGA5TUM?BTlr0!)(qUEu*;0YER z9lTd_e)llxt#ueAV>quY@TRkGZK+$??gTSn^Iq9_@J*r>j0_d&Toojg`FfpEsI-ba zqFr*@W|U6*AGl*)RpPyL>bL$>E`G}o%77jT^{(f}`f#_ixw=K$v~?Bbw)XN|O1n^% z&m;_E`6#yD}@UqmI0XcH5K*9L};S~`Dhb?R4wEzsGsd~ zQ){Ldv#^?+=_TPR004HXqEI_jmrADL&O6j$K&o33?{V=;`^La?TQkH!gMZ&36Il`b ztf$_#?tkTr6UHiA)fW|`DeCSXQb$gej(XMWw=Q#3Pg?Ae%MmFtOGr&aC_;Fojd#8% z#cTL)Pi;?GhS?V*Ydie_G?KdpReUOe=~aozt5e{+9@#jmxf$ZEl30JZ-|mX8D#L%O zf1s-3r3{g4+I|UGXHV5Hi2KAuKcDok6&*8QLVO>LTlgd~BQr|P`4THg?o75N@dxhZ zBi^aSA}yn5(zQHC7nvy|*17A2cj<3TO?ofHe-J!2%@kzw z#}(pi@=WEnvl4T_t)a0JX`bES4<46Qv~^;`10uZ_SiB;^=jhD172=-_tyOQKS%F4p z$=oqs&vcghwcyiYoj0J#0P|e+)&1C_@bFM~Z1R0ZZzldY*z(ym$?Do|%y*w@jZOn+ zBEEr#;pUz8!*_N}(tij&1z>p7Lbb5do#KoDc&Sv}V{=|~+2oIM*RsPqkD4{XKDE~P zb697F1Ve{~`5vOVJ7p2-DzY!`_jB~GR?>Voq)i%KLfaUwP7N<+=CmzIMe0=XP3wq~ zdv?Q-$Q6~Lc#7fW@}%57&0W=}(}mhw&oBpm)gbk))$qZzg=R4DMneHw#mPjDNXmD( z$#ZqHFYjBnHe2HYxtpe%+^I##&2`=$n9Xh(g>u8*sj2K@rEEIo*sp!zXz!E<&OIuy z@HLvI0@r8=1EnEq?&XhT!&Ro0lQ^XCR8gOlkyej|zD#a4)$SgYo4_23&QvMA(O+1p z+~RJ0EMvD~7?Xc^90e*HG}S1>;X?MV zTa+}9hc`F2qE{sg41rsAUkxRLcIh*k^o=fSb+kdp-5gecniI&yYU3WpS7((=;q>sz zPkAHd%57ZM2Z!LEYpL}GkAA~}^)=D_NYw<_X=w@jy>aVZ7l*W1w66+6B~j+czGD;b z%_Nn@b0QaewJOoz`d16%9Xfk?JjhoB_O7;lQ7!KrD*yqkO=noPxeOhFIO4Ny-$QCp zjkP;Z324o!_$N-el3j)sVn@)L^A8i3xz*(=M$sW3TJJ42S#<9PShc?9j!?=#`kLYO z3wJDJAntRVRbNd^+FZqX_O+;7^jhYX5MO8Bx6QQc3l;~Za?Z1jL!riN8*XWRj>0S$ z)~-Z&suqC#opaX}c0QHH_d3(iZlNGsV_Y@dn`YndU52XPxg*-R>!jT>gIiNGn>S%` zHl35{T*45#s^_3;>Gbp;VcY3k-lHFvfAWs?(+cJPM7Vo%R!QO ztG(FMvN)d>BmKW@1HWpi4aBzg_-y0~>~A#1*!MFIK+iSCJn>Jdc}hkBqIX*t3p8ER z^%+^6Hl9UP!u`4Y#o*UZZDDU5JHQztxs7>VIZMaW+NH_!4O+#WmfnJj=DyK!??q|O zBkT+B^~3$&>T4eTa~b+p#rOK<{{VOTn##QUhv{3xV;OFFPm81b)ydh%m#){vtB}>m zS}q>B>wHIjKY-YVFceVVdX*dnsJDXkQ81CEXIdeug`nV7w_pW1Jq#qon1U-k!se!$ zTnf#e;~B0;s|qjPE+l{+4QN9!VeeefyTf&>uwCPl)uKHuEOpu=;`|>#O?kGsvNhe| zQ@eW*#CFYhx*WiKSW0=cIug5%lJk{{R-6CDW^OGi7%ULY^v;+8JlPP{u@rkiLMC)|;x_O>?Hj zB$(RZC_OWd^hP|>MLBZCoNVj+(gsa-{t43Aq*)nyn#$93v3Ych8g3`4uUzo%ge@)9 zmFE@M)wMX*#_5T$bW^W>UCF)!sgHHC36HLG^n&0<_0O#no) z;!#~*l_umB#5|j1A6n|PNwNnu6WG?9)aa#M$yxG4lhU;S1u7Shn!&TJBQxyu;cw4M zz9Sv!!ZB0KXrNPDVd+z)u=J||aZ)Iy%sxShY$|272Arm*Lq){khC><<`_!_>hN|;@-Yd?p z4Q@o^bClJ`TEHvKbef{%wrieof@^wT38Xe30KpZ0GmHE2s1G6?nL)w=>VEuOf}+Ijo6WlUi4AkZUSVTD^Qav@CSf*oXs~lRH+C zMMZHSRU*0#3fD9Bs$usvL324R!K8tSBBY~M9AH)0`c)WDO1}usDv2#bI#sy%#YGxq z3b2WaX@f?cpK9v#!@`Q>?9MS=Hk}~K(rjrB@nFuB_;YR_v ztSHW;YFVuhTBQs#!t+J@>0IHI7aR&~kg+%%w_2R8TtZP@(1X;TEI~RfKc&b|Npk&l^^Azsy0-xUy zIc_-~)a5InjBL+R@LrjBYvQId+RgxtzHcn!wQs$x(en!0TpZ`5X1t#J##-aDn6-(f z1Z9rUq#w$wKZ^C&LALf}J;-jfr6j>3*NS`_uC?s9(P@`&5j%+xw;!c@zlCNuz6R1* z5(&0cVcZZ074zr(B$~tVHRRF(*_h(Gp9=g-zR}oP-!GeBTObd;AJV1jR*3Y=j}3-! z-gE9##_n-cd{wX8X$-PlTtpXO!gY5(z;jssF!-IZBDV1SvA5b{5B8EW+>VMnAFX)( zRG9B%b-{(E4SZ0@qJW~3Wu{GtXTXJIZ_J-$Z?V7=MK&VB~Sr@6!K_}9s zXy$??4`~PPm;%}5k3D&)t!0qhNbuXlI6I3eBAAVvUl`t7BE@vj-$lseaz7fgrj_wW zh+$Z5)*EOCnBX2rAO5=M?#0Bh@0DU*oP4N2$?ckEpQ>Hx`l8&yC=RCtj*cj8u_Ti| zh0}y`CjIIUahm7;D#tr`Qr-wr-DR8t!=T{SuAG+Ig{(40-b}I}ZMownx1!ReyVB;< zF7`_k{K!vLVcww{y>~oI`**~?7}DamX;?(V<@P56hYS+xy*9`0FTRu3`JkJdh z>-*Ra4<40=@bb%a26NQ(s#P4}s7i{BM3LC|g3cSOU=>G7@_kYXi6&=82 zxf##YiWb6qgv7DkO~zKGkbHpC5O$#s!p1ja9gS_@k1tPw{YEP$5&O-x&|O;_y(%b^ z?~au%+DWpOshrf!+uExdmT<`Ij4;Y@eX0{(jojoyVu}qRsAYwEpTQj+uXTM^?cm)6 zc20rF;4#KO9<}AsuJ=*uK(DBLF>hnxPXa*=ykQz>M3XV<%)|cxuT==P^d?OusqyD=lV_o@JKP=-8mroNB(cc)Dcf@~XU8<4l>&3x-?5uSMA1D9~W zt$k6Z-`eTA6`him{mcl5>J$T2&B{crZA?vFj8hQxjeRU_YQS~HTGuX(+{DMyvIH_q zg+0x3)vI|C8i{CipAYT)yJmcS<5}MnyiYEjHJ+7ca$uTMzG)Y#CO`dE*LZT#Y;@H) z;a73|vtAc}e#I8l4>I8ojktC$0_uS1I3Bq2KryQLw($wGRzk+%?R?Aka%RUR;RS`I8tNgP+Q{Ul7Bs zcqhdhnt4NR-4vwoy?& zilwM%LeYp)RmUejwOTD`O{Lse&fKpbTIE_Zq3v@!!_r>(gT*#5JD1F%kKxUITM{qX zk|qFvpkUX}-v{*A=kXoI?%b>^h5FaO2|&ekxY;C(6V~T3t!eQ^bX6DResSqmbXYuu z+uE((?6{QoD@C3ZU{^AYs$A@qR_OAYL!}hZc&5CYMRXiDAKtM%L93>hWhAWMbY`Z$ z)grWsncNHt@^2DrS3WG%29>8VE1WwDX`&pK#$D%*Z?$XYh8EiyIL&C=+kA^Fu`+ed zBtHxQp^ValR=c)PsN|1&*wZyj+l87*Bnuo$Nv=f9!mZrS zZ1fk9u@5o89@Luzinb9+;y)Nhr!5$hd!Iln&^%SBi(5Y?BZG~g3g$cm5?R^ZYEy9X ze5;>N@+sHaj9P^9JGd&m0a?_OjC31kA*tOMVrFBKde;wie)3F6V~%TU#Ttq&!$OUk zel7L#$OBUdD)sh>M4{gfhrDa#20r*;LYRMvJc+sLJJXXh?4 zSl2o(<_8tjQwrvVp)RzUiEXEtQd=^&^~Dp6KZ9 ze)7l;b6tFzf0!`DnvFE1hm`=FzsRlmWAf(kDF4 zD_K5KRwb0}st~p+$KD-kX3klWyB4Zf2uQ6l7aY}j;;fWHo{}*cNHsKUa5<=Mm?L%R zPz#KAt)UaK&qKP=h>OEorNf*71CP?Y#^T}RnO-t;B)dpn2^?3U!y*3wg?)NgmXL(& z>t1K5Nr&AS{m}cT)}37)jOn|pni8Td;$#Deh$rb!wv0ycO+U*SVEObJ{{ZV%P9%mh z8B2)RkPdJ$-kmO`{*bZEp-QL)M?E>sU~Gt%^d+_49kU~yyGP|%*LPA)BvwnD2JPCm zB4}F50Ko2PMo1mAR&;#@uJv;>DaLr>u8+hjzX{g(D@0i1M|I}CDkzdTv4A+NO(x6i zkSGHf6@F{ItBUD4ih7#v2L`zN=-4r?mhSCOQ(XPz$a6>za_Amw7UzuC?c*sY6@hkh zngGg~zG~>SDZcJab5nil*U@F#;8e?FN_(D#18$5O9P?G}B?ci+k*k(Rp%F_`hNw4F zgH#%#aMdE!b!wSS1(OPqTNMauc%Upr0ku=h1N0(6%_`xoJ2RDxPXJ=}Ha_QX!ibGI9lNugsOqTFJGC zHPBm2xhayy+$hCn+&BtrO6jnt70+Hzf_bbat_@uo8j}N>=j8jwwJ+t)Yax%!X+dZ> z>^V<0p=xr=S*m%h3qD_)eHT07lz4V~<+Nzj6r`(8(Le0m#4|s^IX{V?z6F!kx|wWcC%; z>?LhXop__onafB;S%ZQpmr9Ft`#$b?rH!$Q=IV8HHKA?=Xpf;jB zRN=X-_@wF0P9&iNG~keA3g|THer4vlIHbc0-qYpfd8Wc9_O61ir^DStqJ?wfalx=4t=id$> zYtW23*6)d|R6vZQr&>!rC9E}(83%^Nc2`~=eO45SaEf}l=mjai-X@L(JuHqVLzeB- zL1`3|NsbW|k}EsLx<;)MX-jiyB4#&y<2;?bSG{PO7N6mrERh&gQJu~Ao7|eDqdbN!FI0ls?Y_@2Qh;y|L;nEktc$2+j%#@2Mpq!< z5zlJs?|e(COOy7NA3#6Cx$R2v?yl`;n&cT25xC=yn4(I2qKZ&QxgPEImehQiNt65>bvs=~s=x);(_U5U+fvhMwywiqAZZ1^v zI(PS~K2+zk=iZOAPJ18Dj#MsYT74fbMI4cwlD~nc%7dAZ4G&R&m8OZ5wn#23)chmx%TJMw{P06_Dy~vE?iG+9 zpyxl0d9%uQXL67D6>VsKFS+pi>m*l4VCubdOO_y|Bh!}IC330^6QRLEFhTULlS8{? zkU@CAF#EE4*9!jt6Eq!c#3i<#8}dSl)O@Uc4;9#H?9M=t{K|V)HnF!g(Z_sHzJ|$e zEdveGoPqVOXTmo>W3dWG-YU<=D>wfDgs)4K5@Vh`5$-WvZjIqMWzm&@UYTRy>(PBDm`|7dHO2fx(Hl%;kmr+E1kp8|UdNuf z$;CMo;;d=bO?z>)bBeitVr-;TJ;i8|fl3u6#dTUV%K-)JCrQ)Z@>8@D1!-NtY}m%! z5n4tmqFmfv(b9|m0JQ@e9@WNpmT3~~LP1cY^Qob}duW)ebDY*5r++o{`{wF8Rt-42 zBS$ATnMK4X&{wkfV%}x&gIZh4)R5!_FC;ah9?@)lJahrfE-o3^HNJo{L6n(3uq;_F$8aNzsZ zyN+U0?^MSQJ&h?lv7*Rvna`-G0TnoX-)Jq4MryT8b{jW*B9P7EixilEoBse@8upf7 znB<iZe!HvTT`ddYc-&oNtH3S?_7-RLvKdp0O6S`({rsuJXhdkL*r?O(=l=(=pGmgZsQ72crT_mj}Ivw}hgm@o3WFNe1&a{iQvy$?8kg7?-fIaKEMJrg@HtvsC zZAU=V3rVTnH=q16LZlzYnq}~;FgcFuS*HFUHBJZRUU1soksV+1rT*Ep3J=eN?N`pp z#18IzZd)^_AAk=S;vHs{_qc9TFLlvuiNRj z^M7c`ux!GXIIh3qX0-Z-kr=j;HjW|^;d`s&^%dfmQf07DoHbI$Ppw$D+N?8KD5ooy<)x2m_(?6jrk3HRPdY})1oK|C_KZMW z`ewdRx4xdnRF(-IG7fk(?4BI)sMP#4;%5TlF|_m6x6z^_bF;C%^Dbm0@K}nyY)V}^ z9*2=!wv#sGQ=YunRS4b@*ODuhYb%{H_-twh;;8CV`4-IeWf_U&Rb5&=){pe9taC|d zc?XI0ySwRL*4fJB<2AYP+rx=rGhEDf$lQ`^mepJA`mC5-fZ49k!WZ)1%43XRAQ`FZ zPqVVQ=(?1UrHhk}7QBw_gqIQ!I4p5pmx<@NwzJ7%0UXzze`o?udgiO3-mJS623e9Y zIVPxhmgLK*MPi%;!sFhyygeYe)RB)xH5R?2#5Kco*~o0vv7;kV)UK?og{&VcD&vaf zS5&)&9B(9^^`+unW;mf>2RX%Y53ESU{Iz-$Vq2?euBK@oT(LD0X)-wqAChVx?J0Ae zqPt&*`fF>tgEPnbw4rJ}k3sO&qgi+g$_`E>JpLxL^#shv>s_VYn6l7fg+Ks5O68}v zdF5?`I5m}BAjKziWu!SZH~|mBHNe@8pJOn`^HcU?tz9~O#&fX# zRM)bf{&|`%Cvuq{lx#I1tKL+A;KtvTF`<0@!}(EgX_a;yQRVwpuPOfkxPK}PPQ6tAR9qFx z!aUPjF`QMMr=|@8E_xaVa52Y1*k4h-G9--o!(QSJg zXBmhe)YlQH>Y%Cls`td}*0$Ge;|CeFQ~v;uy?N}{%;WD@RZ5)^?9-yTREFwzUTdCe zoxHeiwFL6+H445CL99`2Xm6yzuUZR&de%bR5IWR?=Co2Ul{=d1;8m#P80%RU?$Wl@ z=C^FLTZ^20`1~rOoVGBXG`AD&3Zk@hOG{l^;h<(dY3MYU%VZ4O;29 z9&BST(PLn%wEGV&Iznbl$3qy>n^5?ywy1C0WVGa*9ymmOJxi~yfO)wtRq}32j6XAz7wP`=fIjj^s)yO3x zsFrG7Pn0#zT{&`jtvTf;Gg(A#Vk;V&Aijo8${uSi*9>c-xQFFDRu#hSkk)kNaY5XO z4r@lle(`>_oTr-8u^+rz+o5dkv~~Nnxj6mZV(9%Awyt=D*Na#9r=LHF!n)+(b*-%r zO9CT2%k&kPjS5Y^W5DfN(D;hhHbH~2dRNh4;}srh4y>igXrfJ5O?C2exH3OaL0t~J zXy?P8LI4k!fr{ld9}}eaDI$!7sc!silG+IHNdpz#DzTpD6lqjW?H*lu6Tu-@9D~h5 zkI1bZTIDWfcMFV~&5d$Jd9q184`wa3c4*sm=Cq?dD>mPZRp{JODAA4b1z1+95w|>5 zg(T1g)Nb{)r%TH7STS>2))@>bgm*gm96p>yd3TB}{%7L7DhAzawz$s`+HT*)BvVJ7 zAnjg(@Z@J&Pxotwnm;a52(Ll-R|uP50QOo^jn$0as>iTw6l@i;ER)iMVIQS)dnRY> z*^FZp_#+D_YSsG!dQuxDMI?GsblO)gl`G9AJvAsH0ZR;ue7EpV5dne7u3qOWaN|Ad z)E_!6as6(0U z4wW9IawNJ#7+s4Rh2o^M(-wPXzPWEI(HmgLSsTz)GtXr67T8BYfzuTC7KpVTrEjr8 ziOXZ04)rL%2kzR*{(lP2k~p@Bt>5MDzLn7Ucf+?@_?A_RDk zC8Ug`zEe|dJ$b$!wFF|;-jr)d`cO<{#>BeayXpQU)GkkGaHrgnTw##>`2H>{)O=~+?OM-DpHePvfpBFi zKH~F(k4p0RM(NQ5Y=xtqQX zFaUDR_eE&wx;6fxl6koJ^^61~3jB3Z)A~hdapSsX6UVbs?#_sVtT!Dt)WFvDQ*= z0^C_hcRWwIeSxkE00lV*KAiydKnlOBTSb_8?{4S!b173=Q+n8mqA8h8h~wrBnxLoW zA6maHyP`~o_m;wj+Ot>%K_XGD%Y_>AG031Hkv2#kwS;N4JByk1I4ySCDD>FObKFF+;O&!;YmK^*W!_cG zWahmmRPfZAmAJI9^JJBd(;jj_9@Xa8_P^QMg!YGW$qSS_hdfrzGNSH5Zs8BkGwn~F zRb>GN(dkuLikzNl1Mc}nP0BA?g^oEH+iPPh?_HmSe#3IpU0+^Y#(_43jN?CED~gQ| zlnZ>yJ9<}Fr}%F2D^SmGaSBX<<1fH``ct#9Esng@ETe}~u(DZN8GvLmwpcJD{{XLC z=+T~F3A}-1Jeb(vkJh^jABMgrzmzmKEgWy0Je&$f_&MWP<|SZ`LQgHcQ`e|yav#_? zAmB}I-$?~Bdl_t9M7v30kX3;d+~0gF@oaIcL#8;w{NP9%arVV>_xfI|r^c2y_H(4+ z{#1C#rh>+0#K&Uvt(fOYgq-%Sa_&uL-JRo?aubj}E0fnXeOh=5NN_qFRVk$GVI*Q+{`z@j9ORnq zJ{sDX;hI?nH#Zf>X?By{f~y-Ic{r}qO1XU=4UD83q|-BO(($~iZ3+Y0ytF9Xc&@i! z)f!}GknxJ-ZLZ>;X=nnNUmqY+Zj z)n(Z0oo$aml^KfLA9X8~^X2~lo`#dpoBnz=PbmGm>~!rFw`E#y*{zT8t#dwnqpXIH zwahv>{OILj?W5S~zRPW&xU|=_wIlAWb7Jmq@3kLlnEwD7rO3zaquA&Mooad5w5LwC z{^+a`dn&K`!%Y6toBn!I_Hqwv4#xig_HDO0QB68+wm;RaV*dbWN=N$2YhTC)45pv6 zmvQ?@b~b;rZT|qcv{oP6-~M`Bh7qo|BPm5Xq0EC{`AM@nD`Wx4szUuW)iI$&e z@eR9UrJ|4ged~vPEaRnm6Wr~mFL5K9aA~+TyC#@N@YNaYie~`Sm2!t_bZXAwMrd^) z;}yT4=?^r4nNM1{LCKlUYAWp0@cxj`wV8U?py~Q*#-rt^w4F6!jYit&Vu&clE50d3 zSsrxe=ElbWP&liW%lpcY093aB0P86|iCdG*Dv|3~u0H6hN3BdY9M3uUtspjb5u*P0 z%m;U|HRV>e`?VR?b;f5Qg5 zO&d)}U{Vwe_4lO+l_p%%;TE>Raf*G#{>9+b@+Q>9W!yp@B5g`$^{naUo=;lnTE;%O6$9JEorNGdc`YT7+Fv!VJ|&ja!V=GkCIP98^k!!i^Zr+Dyqc2R>z0nH^&1Go=30TvY;+zurj&D)Cjsi(F6@)jMrVX@Bv)2`SM)17)_Mjx4NVikd*1dB|(5*E4b&VoW z2eo>tvy5zwdD77Gw(zc^Au=E$4dGb!3k2O1#_CGhHa*uZdC^L z=7`syQvIY)6VkJxO$}V9H1X z80MmF^5{*HdY?pL{b_LZ%0=TC-=S0Z)wJseO!38-9>v#B1LqndJ0Luy{lNlN4s-6^HII^chIAf87-o_vzUf> zBqV&pIAQsp&bi>b*q7jYRLd2*@OK}=iv`a?lj3wZV4fYa!*X{sWd60AuKYu~)->rX zVzRZeNw}4v&gUMSlf`m4x9uCBz|*!wdRxi)8g>DG*)H9`Xpjz=-gu^)rpGh6{{VrS zQyfG7U8MYtCfM=+03OnQMvDc=Wsx2>1+sp%LwQ~?n%5DMIJ0y3QRI>T04{C$Qn2o2 z#_n4iPtvWFWn8k<#zcvLYkjA-YDdJ7lZvg{_@7<~`wvkxtRu zuX^$5`tx3+@S@-g+q+_MxkHb`{{YuXl#Z+7MWDLSEmVx{xnJaKt?Z9RZmH-hqVyt5-W<_UmUWwza|esUFU%91b!Sx{_xFtr;41vh;D{+hQO;4+WazhVCoMU zslFt|wW2-qPxx7<%&^HbDFF1V9}#Tcd7xiINl7Pgcqc!K< z>sG9DfEDfk00lJj{hf1iN8W5^yh6T~_)BKCei4Ya0ems7Dd~1+H?libvqz1RL=21w zWp0F5&z}lddJ*1_m(XUpB)N1qPUn5#+x2TSjStR;4T|afOQ$ZQ zVHB=#SDe=o;U`F~n1S;b!8Or%w_6r|8xbpiq)+m%uwt3U7A_7mX`!ze8qP+)Yn!X9P_*AapEdjU#lw0 zWtSz1=e<*S7^@*UR_ZH8Iiq26}l^(7)DEY~-S_BfZ$1O?9#^PV}aPX>4&RPlzSE$t&KZin}?H$@*` zTIc0<5@#nQ*V11Lt{~C;7zl;flrk|y&QI2}y^&o@Hkww0@bgXkHlGx7-<8AMN!ZsN z;tv{LI4d5ZBglG{I0CtU6ZpGPywxsRSaO?=)xj7Q$VsbByCxtr^{o?DA*XYpn^x2& zkh5Rhd;I3BKB=fFAYWV&{`suv?>yDX0cp#95}%$h$LmKjLP@e#y1X(8Qsz{UbY*NB z%DA$gXz>T4>WI9F(q%UfG_Z zN8?IvtxrZz&X>9xoba-^2LVlLi;YN#{M)Hp%NRggg%xDnTlSGZI$^qme;R&uO2T(0 z8s4y@0{-9+K<1wwyQ!;lsNDYm-@{c6%m=Hl=Sk+;la_V-4LgfdtkS$|uGug#>N1h& z$2qS?(fnI&;tO@Q*6pF0TaPp^%%|5i;XiIf&y>nB*!QU{yhm|sa2W_+p{C^>q$^8w zbzc&E6RYSq2KqUqywYSjj#d5udmLArTY}d~&r@F0pnOvgXmq_-PD`sfS8ELOUTxyN zHVrGp(xsvlwtP(*9;&P|2kTjB(AUkIT4XY5!B{XM`Qp6;!}`wsvAM$v^W$|KIy5uN z;191grQnT4HQy4YoL_V3dJV%EuXCpTIfxgm8;l&y2R`!k=nT3H%Dzg@=4@lh{48vYu6+2^r;ri zlUyf?qWcBk*?)-gYOiog#a|C*O?n%L zD80mrGas1NW~CyfvE5haZnvO+RcWYjx2>JMU7m^XTpa{ALhvM)#Q8eiG*$OqjtxmfN6F;LGk!wGJtq;2E)Pvu#c?&^e$>9BfC@piM&T{pi?zYGu7#kWl_rO78rF`?%Qt72OpJY;7uW?rhrJ z{JW)Z!n5@~W6Nws*Xvxp{-vqDRw&AkrB|_A>88xe$LmC+8(eX4Ox^z4yP~7}Gk?z` z`B7NtY3)~I-TY&0Ce`9%IzR{hJk`&CX9L>0&l8{cnaBSCTu1Y2%8V(ZUlF_?x{Nt= zwIKYpiKuB?K_FLl(W?%Xn|*Eyk&0;v$n*PM9LaIY_pbi{L(*0yQ?{zvSudInE3UAF zDh?~Ql1Utu=G(Di-T;a!$l{bJ0+7|10TnH<{q04Xp4ac1m$264w>@BM1^$K9;i zXGs3>#V8JEiFo(To5i>G`*6j#k4#`~{c~7$dUTHdXU@V$!2`ZAUB8bW&Go3R(*k22 z*7xgN?S<@OJS zuWO*O#^tSReFzMEl{}2qRY>P!iO)qfWi4bs!(CKbI*fd^Z8W7H!$-0k#3PVf84qfe zY$fh1uU#=m@YI7(QT#PZk(|4RX>%@i`1D-L*XlWDWAy5vzD-0f5h% ze(}yVx`ydj>?}*j3F%E}m1Rh0d zxwm8jTzySa)AZ~3*E7KCeZ@Zi08GBPjI0q7ezoFFIaN5_?rT}=RvL7BRg@oUnY?9c zKjonO>(6d{BjQ_vS~$QS?NG&a59~#<^`DiLRJZEa^K#+cQ z#4p2N6E|0f-!4DKtVQs*#7)V$w;%8^{xx33kJ%z!b6ke$7%3dr4>=7nVO?GKfi-Is zCTU?|?Bcn(>?Vd?i~&B?v?F6^MMY>phnlq+qALko_-fKv%Dq(9g=#u1nzs~W%<7T(_{p{;vcLfGPqOL6I2)^_Tk5-O}s*$ftY6cb5#C`iaBrD)9} zA|BPsYgg)u-!%-RU~5-Pmgc#r^#K&5gZWc6ZA8l{RW*=avTUnd>@B=Ns5zCvtI+tI z=N@Ut*0>PoJ!$JBXM;wBI3v}q_3yRBA`)&Uv2C^cjW+g1Nn0C_8LV9j=GxW#$)s)A z>??W*v?p&iUp`I8kOn`cOmt&$T@6+Ek$8#*Nu@s$?}+0yJ(M07^Oi5{$z)^Qn~J@E zrTBQpQ~NW@JAlf!HCIBjFk!ycyjgZ*wYQRX9dI}m*lGGEtEc#KTUl6IT$#j&Jp7#T zOK;+6d_Ner()8%#MBuk`TW3<6R?v>4GKDXc+B;%~)rx}mk094WOL?l4LR4F^Y;%E; zSUL@)S4u>RzjzFu)zRy^jL|_Qgh3K1)!t?U=8%kF6X-I3I_ESge7Fk^er{<*bKLZ} zej69@G%KYE~-=U?S$MenM+f6b! zgmOX^T6I4&DD77DABi_tnx&28HqpeE=W|8WNZzOX_p7*QXP<`}q%Uk{vsFo)Au7R& zXV=oDzVHOGU)-cE1a`5=V~$&y6+FT^<2cP-yZDWBCG2zC+iX>G?2K|4pXe(%_|>Mc zOI!Uq-t)?4F}$ULT>S{m7Z4zao#M`B+|7BEa%(i!7E1FdSq5ySSUV~cVD@h z<-F;bySwF>^(Lr5a-33PWpXyYJq>n05AW8;#IZBDRdp&!KH|9<#wE|KSkYOm{7YwY zz_cb&xE`3OWhQ-94a6*p@wzx8s<1Vo9HEzrt51+>90BQt&|hWtu8n`W~wg4^0{b=z84K`;fTQ)#CiQI%C1-K z-YAq2ozfrYUdOc|&`6AMBjfU~G4PFn)O=GTFeEFV=U2KV=_A^_A*RJ=rN*H^BvuEH z?VRj*(6QaxySocx1+u6e1xI_O$9s=4LFTfzLq{!cjt(a=czscU1OZqOp}N%|ht;dN z@fMyfEY8Kiu3N*fdEO(0V}<0_DmImlmn*{OE%9ql`!)5VsQwOrO7jO?99OS=WRM+R z-Z9Au)AFw<`#BkBt(w&3OssM%(7ZSCO5;V-MuRfFSQ*lC*TLX%g=ek*Z zW7A$X69-1eab8X0O~r$x(y-~#oJc9Tx1gg0nSKVtq4%dqyJcbMO-9*3 z7*&a|YNH^w=Jo=;2V1rBUBFx~YKP&5lof9EM^H%p)Or0+ z=~^BkxmLI_ai3cBuxmzn*r{r8mr#&PuN7EM#BC><+q*;&whuMUI=aqy<2AKUbjqHn z(}hHlvJ=AsF@8I86^-Ic6K$yZn@-78Fh2e3p1fO)&FIR_qZ1SY zROB2|_oVi&Q>3>I#G4rjB}lHrz_u3G5o(vWE4W(QWm$>)N#Uw{pnSDlX%| zUQKh&BeCLDBmLUHsDRQ+NaJy=wOlp_N^4>zxiYkwA2Of9rV>n7ghB_kL;)1XyP0ya zg;2fCHVB!ebjJAO=cfjjdv}UI$kk}>t|yJzod&^>2rLgq<%(zVs>e{6F;-MKJ*vsMBHYOs8^4u301QqCFs(3@-M7k2)Nou*(cU`YLHNd`X}Vh(wz zV@BqXZQ+;Am@$n;Pqjhf2yTqgUr9SpZ*V_#AsqnzAH(`nV>q;v{o0*%$u0apCYbQY zG*aNM@yut{f6sa&u}sgnmR8c$7?Qc-wmuqaAKAV$vy(Uub{eawECWomL_axSn0^(L zrA}=%ds&AYB?tAT5Rvvcxypm<%`cukDz2JE&Y5j6Io}$e(yoX-t4ks^8RrULI8(fj zYCMWG0vYb2w}Iu5;ZPE5#P6<;hu6Q z>Wy&Am4s*UqQZKft!9ziSwjMh1VF!~a=O9BtrMI|^TxwH+{{TG4^rd3Doqv@Vprgu*`=X)4 zQMJA3u*pmipWzhWvcdh)Q4({E(&V1>Spfe4vO)dPP5Trd-4xTi*Yl>kGK`b?QOblk z*g^f#P5TAE_XR4JVVvY1lrH0%IZ$Xc>9*(iMIYIg=lBIbFGyMaXyrkn8PaW!@CuGi z4$?t}jY#@cjnYxNN;wc{Q63r7{{ZKT{#1v;+Ii%{zm;AOC#^fpao(ROXk@p9G{)q> zpOrOyAEzP8!lUr5k2&c)_oc{!$jko#gi_1?c`xNfYJH*TquH*(p-;vW`C7aqKk~qT z=uKGg-Kpd{^%i=uBA*rE$%p{$$gc< zY7++%{jBlFC(OWoO>}nJg{q*;$Uhq39#H44JRRd1Kb>z1uVbF2HM!`Y+SVhFv?u3K z&#!5)b0x|T#<)`+Ps=BmfiD`2U^#eocy|;p4#@Fn9ReGkZKn3hL>xf-ZLLz zUUd4jQCTK>g##wJO;^Ne86iz6NG-01s_H)yEMo^`Zp3@yxQ%DxYcX#r6-T+RGQHO( zxMHYHSA$QwzhKRrQc`PToNuEvb&nNKsZL)%H$hu|6SkV-MV3tST>Y+!-)TX&E1%B2 zM?uoXq)C;=1tjk7K3gMx-1#s8(*mp9-vuX{bsdEf=1zvJOQ%fT6x3^@Oy{KoGTFyBgDj!z^)? zG^`gnApv8+spGJBJSnc~-@^mCR;(TzPzT5;uw3)Vba`C;%D)$frF?K}(c#chaq^12 z7L1GhG+71DIfKIN#AR!0&%g>nn2Y>1X)INH3U>u_n6}VSaq?EI8cHrcTH8x? z6)=nr)g^N?Hai6Xe4?zSs*crc04T^5g5)9v9jW3m=}s+3Ujl%*n+K&KQRz)G(+so$ zCeck~Q&AadzIVLIcomC(rX-4fTHW(^CpcolMMCJUTYUK@|h!98oP7HcA7Xs-&jeoEFj+xLz^F0O3`P=)^h zfQr+y)UT!?jLo*JUumi>rDYi)bgGtQMGC|@V_vjp%?q7L-5XF`OD_!M;-q;bjwfK` zU{p3~wR!7SWKGeqJq>tV&+8{Q+Bxf;M)0_eW?(&ySd!*+wg=2`z;(?lTPV9+E=Fr; zS)`Exup0N_syh0wImoKo zmA955xK4$JO&}$lt1%>3Nn+bPEQ3EO$;NAyn6Oqg(OQLz!jUqMyrqfvpr~$nmcH4z zg!-(5`Bo;5Gj9IGb6XnGC9MeV(vRg<{u$~TUY)B6wR>fr*f0lo@Sets|eygSVSaapvQZku~8#pl~qLAm=HZ;o@{oDqu6$+k_9wjT~WAF11k ztfqP6?ov>3^{U<|_;;Xbeiy&Iwu(VCF&0KG-l8}5x<`p6)+BKp%%n%9*tYh5dcHCT zvl@Ybk9Vpg*~n?+%q$x*d__{uA)_lY8M^OIp#I4bcShs%Yd@nXK>krl5zM{9x3pSizbV4aigpWEu;MElLs!o zzSXg3;ms33@YvO^1;p`@<|-Ko3J*bw!N2iOg+9e`4Yr9QOCk^$IiX#PL&tT*0tClR z)s<}Am91@T?}-jNHBz(F+Dkd_gz+lGz= zH5ALf1_RQeu)LTjmL(`x&~s46tbHNjn=2{cvc8oHUU@_0JP>i*S6VB@{u*k=^4D_6 z>|>NEEOXa2?HI`7syhl>jGgwxNUtgJ3>uEB;sk|35Gl#8S9IJ6?Npmih{?ap8@+0* zP2WS2)R0){tQ)f8xsL~>y`H4fu){gdYtwJEguaiO(tp3r}6?%3&}G%OhV*&T_?xCb`)hu_mpv znJ$EMhMh(#MmH3lIj3}@fGUQQi2P)sf z;I~#`ou*aDqV^W2pQy9J`?S7QpZg!3dJB9P@lp)^n%%yWEkpkR3WvoV^!5J$zpGB@ zR%Z$1S@p_(bzPBGMhUdX_3K`Qei3+|AMy_$_w-b4@SnuX{{S%7KSvbz3p9E9Dg*KW zr1vJJ@}ci0Pu9Cf_+8>fC;lZNR{*}#Y8dcVpde)kSoR-t5$w?Hc$Kya0q`@6W=S+{@p2&lu(^Wt`RIs^x zf5)SN=~7(jSF&4Myf$;VE0R_+MMDkW*}iR{#8AA4qJn4{<)0Mw4fnRXayQ-i+Dz|OFP{+bHzjGrm4A68sz2;cP3Q$ms1 zq#wIqprNgSsZv3)euAC6QY%8Y*%#1Ki$FRGvE~Xq!9ZwX_=4R30A%ES>il;W7e*J z!bLQDw@DDjcs1re0=Z)a&Ku2dzG4D<07wi?mrA3vf^2(x*=n=&o=tV+Y*VlIiFYEI1&! z&ow-e!e>CZ8%;iDD{fGYg6jzyfG5-L0Xrq~ihpBkNJ^GYq{+JK_39FKp z0nZiJ__^__L_g_({{YaMxwgYsEVnyg{nTMj3LYw($lkTLaR6?WMoU6RN}C!cQ;o?z zYfjjVb5w=12I*Jefws{XW+}$#mRXfX1#?==jWl~lV_hp*y9%c^le~1Uh|1?2;Cbe! zadOGANyloy8fDewyzU~sK3@-ckM}YwLrw79aX4cbtt@A%I2|v-fMCd+`d4LRqC;nF z^7&&brMc<)+1uk()XD+DR>mk)SJ1W z2{%%1sLRs}Pz0r^rRpNo#(F?ZUZZI$7Lu4=fs(V-Z2VITR807w2^va_%TTlCnokq~ zU`{o0|l3c&udFIwF%tj?Llb2Zf}IKycvguTp&_3H!8zNKAvampJ{BIo%4x!ayUM z>mrPW=|naQ5!@FBtpy--uRgvNRG%@;EH1Z3UB$ncj--9zisx+XXO8W|p(Iyk@Rl3{ znsP)(EQ2S$E3%%Ev7~6&%Yq@bPAcl#ZX6CO+QKvUMOcn7y{kWA4sKxjy1mTR($KIw zYVz5HP!_F>dQ$Ec(4CQ@rdAffjzt39ROYM~6${D$>q)an#=}kpH_qB>7nk0vw4C*& za?&&f*S#;>03B;3rN?@O=RUNqS1LAM^a0kY$*8CYCb1{EZ(7K@y^wb_Ty7~HG@6`a z_=Q!IR*ZM9GHpe{BPOX!sXwWwVjD>8=G9_s^I6m0#MwC&$VsWcBXvYCQr!XSXVZ_Ya9&J@ksJ_HK7f-d}gU=X4T4?ZkO|ex0>#ByATj%aBCvQG>wL9sTC* zbbD_K#T1dr6ULia9AJZ5BjL@Ic*%SNk;k^HD@LZa2q zPVHv6)lv}u04kOs9@*w549NXx+QPQvBEpL)l*Qn24E2YzXP2qhNURxHp z#ttGscktrAd0!KS`?lxtYLosF*9(DeKRVk(NWO=pUHFnK&l`Vi7}_|zz{N9?0MGc= zmyf(zXRo!PvfjQ>VB3eTKf=6#_@+$f?8!}k6U^W8?w_qy&|LQI3*sHRXsM=Z_J3)K z6d)kuBerTiL&Sa$YbjC<00~)0Q@rqN#DC!>xg8=w`O+U0T;m>8f2}PC%xP;_Fl2vv zx&HvOw32S~b|~$dTdi(usQk8#Kt7d0rDi$irLh~O(Ni0TPg_tvqpox>e#=g)?EOW$~kTbz*< zq&rJ?-o$bG8uwdEzakZkumBFinz%Dt$v%{ZBk5GXv{OIpX~O+K=cuQ4DD3q24SH{c?4+}kGD6Yu++w=FIC>Atn$ma3 zAB9b|xu>+!#fcC6 zBMH+}-0@Ed_+tM6`(+K)l+lIfE`(R1_@_yEbXSTbEZcZh8Lp%457P^uTv8b%VfUBv zr#iBey@eW4im~%M_+wgG+>J8l8R{c7=>Gr=G>t~a`d7QMniX&`T}ZD@9%{CCAs&Tl z!b-{seX5{j;8Jy3PeJ=6+B~1)%=7ElpV)TL5G>S7ET@skbZWvNJUePl*l=% zM9A#^9@1`(!~l{u`XMY1rz(5#UfkXZ@Sb?mqWq0jhHLyk zy-i2KntaH>;vlE$IIg2$u5(RWO&-;4tYQ&DIwj%%kXN3mOd==D3Yuf5h1b{WKNS3&}Yk1|M3XHKY^FC6WSI9OpFA zEnM74{{VKsK~D1OkiQ@W5w{eg{j>h1scKPd$5rlSOK6L^&w(wE6- zoSveMK+V=QONbyMG-r^Jn8#dKHDRvYKZ$nS+Op*^syd3;xzHoKnXV?D>&;`FgOBM| zQ^Hnab1=HOSqR4M!#~Q8CK}%7RcU(-<+k|PdJs=FmEwhG(~>oPg`8yjQt8@MI#sEP z26rct&ei2l~P~eQFIeLzhxB&*l*l z;R7E2^$x3|_=`)plzYrtD#s%kH{|er1prXg?qRpy4sxfAbu_wVqufdM zo0$B#v6M`5ezgU(Eq`+NGk~$E#A@A1&sx+Mq7}oh6adSxO74hY{{Tv(H04Bj;F3A5 zNL?8+_``IiHn#FG%7O<8{CTS!Sk`>EI-@%yC`4G)!qX$AYbBN{Z?b?Xau3=VLc=GeTZYF1w3gfh zR-%M@Qcwp_%obkMzhU(PwuB=cDDsD;XqAlRE0~|z>z}$RXza1jR=1QXLIGEEI~1d{ zCZziz=xX4=rO2glE5~fB+NKc?N_OE!>IMQa?M~REwKIJt`T&yvQsh)y={Ha!ZNigq zsCQCspk({$H`F_+H_!z8#U}cPc%LB5lH0wym>zMwKwOVSxRdSQ~GFHABI zS^${IQM1$yFTF6ylRy&}q~B16puQ*p?)a!#@@hAblSnzB2ZNJJdQvE-H%b7nBVg2i zWfQh4ubIsPF00juRx_Q#o@t_C>M>E> z2z0Z0>v9@kN;8T9n&18M6J?Q&Ky#Zuwi>i3K ztTAwD;^*F&jjy$F)|zg;>m%(Pb*3AG>p&XKb?;T)>)x{(;QG`rJ0A1_qW5DRY6iO= zwVv{vRd7#yf_1jy&G*f#SzDa=}!?F z2Fnn+&2;)C zmW;9Hbor?_o`r=(%&I!`?M*&o9F7G9kC2}1K9%IZ7Q8JT+_u`4w1I9;Rg-Ra4*d47 zR%l(@CHVBkbN(lnO@~69U(Fmb$lF6RfJq*ir1k`St#cm92qYh+LI6E0(r^43;kfMN zy%*A|bx^Dhe@fwol-E~MEt0W?I>_e$RPwY{%=`*nqZOz$h}C&{Bw^H!1tV!okCc#q z-xT>pb1dXhzOGHC$rnZ^>r_IKAyxz*dRqY?aqCOw3Ouj{CeS~G{#3v>`_dPuZ5@>U zRJc$%6c0i&{pcL`;+8-$^3uU;B&Q=AeNRdNGNZVodXv_+R>~HH2DL>$-UN*Pbd60T>qxhcn)n!yjzGpcZ$FT2N(P>b|2OIJIUljmN!J z*mHxx`cNZcd0~vvbQUdrr;c9~$tIAH6}a`NnE)JsMMonLF_55lrUbjE&-%ebOL>r0 z*Kc~H`h+p$HOi7H>!_oy5V7~i6s`rNo@Q*B6yl%e%A}pjN7l1N&n1#4C)Z)A1osY~ zEO^iRCV;VV=bNw@t1`|l{5^Li+vCh!DIKdZjQQQQzbsZ~j&((`w6w4!?|8$o`ijo0 zvNod|$n)((QkHZvO2yfD4eBc$2N?kSgH|qXWAhIqp0$yCstAtAvhH{Qde<6Sozjx( zV`?jRCCr3O!FL=0dQxgTg^h}_$qIR`-A_opv(w;v$ih!FeqfC7+kJ8P)+VhRM{i^p z5Oxz@gSE6g+dH<*_zqnCD|bSVHa8FNSa##*Z>4pb1;6}AE+qc|YO{M41*ywlscV%| za0_6X>3m7x*fc*6*=mtp?`gLDxg6wYip0E$S=n2FXpR{rL`kQTL{GS^*$Y!BN>YnX z%c!}Mx!8OoZQl@dob?Rf`U$VAE9B1uYF2s=jdc6zoBbXqJJ@!}BEFP|#=aYz0-D9A zKC-C(wQ@`$s?cwUyc?wWnd6fA97`Vo<2X3y+ z_}0MJG?p8NqiRR5nzVmzX+QK5wEqCVtNt}w+QQRwLlyHs#%V9yJcBzslOI{yGt(WZH{yS9PdcD6f;qQ*v<<(o&19of zow+}Epq;EbVg8R$HyHX=nMe9Cs`88#`c{aPb$<=yXuM}?$h(g`eiifw+b3xVF%=cy z9}TiRWo$sn0P*-&&_UhE+?#7fu-N8uz!D87oBr4sE&C_K>pPq%cvuWe8wuwbq$$a;MFO4t)nTKbMcV#ZRS2 zEy0f-U@Urkv@kKG@STxh!Ue~C(l>>)zq;1Rk9yokNrBrT$n8qe;AF@%?^9$wPI&k) zNpcx&&|;Ci1*fqkK~Q}SbPnMVFmc5(FQZi@-#Frgf@J65k%z)RW8EzLPkZty+8^RcF zha=ZCa-h3B%y>gs01ZL%zKjeoK7$nuz8KbXlBbVK^?xv9B0tE}5Tt{1E)IGcIZ&>7 zSNtQo*S!_%TOt1M=|voMDSC~h?NPjdRAI7c z0$_4#R9w_?tylvT0Am80$BI@p8R<+4cP+(HEzMrK99Cl-H3BUo@lqr6RS~$ShQX#I zS_vxAP|I0RNs82g_bxF-hMAjr20`MvO-3d{NUr*LcM9h*{u(5YjM`$zTU4_bB=%>wk&;;4PE-jnTF?@Gf=y6!PmWO|I;RBa|` zv0TGl&#gur=9P|V1u8Mu5Q- z0VU7+*a*GCeO%Q5;2#nNDms;Lc}eJgzs00{kSW0FqnP*HKkIJ#iu z((UyEnCp>+YJ5Zq0NmNe+TgYeCBnJvvE?1An+OQ~cz-azye)ps0*KU!`* zsc}G9u@&Q~!+gu{Qb`mS8(5v7^*)tP6!@H}$7*6CWD&^8*9L$`_i9O)B!J_BIr`Kr z+nXbbMIq&Y{t-M z3NYvKpbw_>h@?6WoYxJqe<2Gha5{9YGjM>7q>)^=!>x2`z7~Nd%S~*2reZk2>CfR^ zX8fP^OV50a8UUu(jAa3l#yVheQYFI3caRkN6I1^5AC)3S_h4uMob522b-<-yxQxKy z41HMmuS`<9U?d*C}0#sHVToLrAe50RAu44xqN!_03r9aD6 z``9Pz)|i02CAlk)b5BVjCu*|f^c52{Dx-3uliM89I^sZak3)`>0Qn?^8w`GLK}}iX zEN~m@MJQGK$DdPC2LUn^fK32Bu2*R#4hMRHW_B4!GN+*oF;i|(q~USg8f!_F#7Iy3 zq*4MICAdezz=eB~{V4lFh}ikim+ga`QjnYS6o%+`8gPOv{Hq>+*`Q*^;!%&ZGiSQ4 zDc@y5(s>)$ky0xz+yrhtDBzK|bqE;49DUTW6zJlJsRA_sbAv>TcCZ7ddaB=O+*&CJ zL)Ae3wC*Bv6d>fng#9XICJfx7XJ@TN-Yi+Hod z7k4_li?UW&z)g%WR~%QcrOXVt+*I`IS$FzAgMT15mat7A{wB^(G>ZAN=D8>rqIZC6knXN+t-QCY}iudxlRorv3G0*|qj8~HQ%y;i|F6O&QGI`H6=l(O4N#Yp*ISQZy_*Qi8=G!%`WlcpH7SWLlOM{M; z+vstD;Tao{3mpD+!wdfaF3el(=yRIyv;bSc?pbk?arxHLIpfsjH9@AO$w=jxfCnen z@~r&Zo)&op7t6MCPpwdQD%=h#u4dKEI6y;`xIc)i@!ig&0(AVV4&H6gwA7QchZ{-9 zwL6QH-1}w7ITEIRl%r6(2Z@k=6@rq>xp_5zXe1NVQ&J9F8Xwv>o#ssbG=FJcbn{UD zC;V#?;#XNn1b`{A%^{Es=hlv9Pa(Vgr+L%NPt5-SjVJb%-*!XzpYg2ypf>LHQfcH4 zKsW-QJ0B`F+fceAs{a7rNBnA(&pfA=U(Tu|u2>A_iKbZiG@Py+wj@4hLn6~c5_7y7 zswdvyRfOTH80=b z*&@Dt@DqjcZJtg6Bz~3esRVgAG7rYIh?qRM1R#DYRAPfAj~vhc&fj57Yn{ZQjmY%^ zfGmfXe^IuHXOJEcx6+YaN@Qn`UMZ!Kjuhu3wIPptsN)NjA&BqVnzv}HmnW}UYQ~Ge z+6H;a%}8TkL5|c|2|m>du1g+iBa#-$Q;(_Up>3$AuO8W`Jl)QrP?CLA&;-+>$&#`n zA5eLwshuOo%QBAO^GdBZnTLrLJ%vh-&qY-ftUHWJ5@jlcZXb9aqLr+od=6S#%xyj$Kk&4#BlsviGY=b#gJdsVA_k(2%*{F&ODkI?ryu=$9OW z#BvOe&W%uL&F{82IBRdosRhbDLtDQ+u~_S^B08nwNp8e-##H__q=(H#cy2=}?!41h z6VRC4MbJXv{VBg_jFK7?gWnY%Cq_AukNgIk^T?!c!2GG$S0v4^qkgnhA7>x)+ndA05J8SE_wF1rR7+K z=C~`{c_EaIq+-3g`oK&N70PNJ9$6Tjy=q)8X&Kn_HS-TZYSdQqFz9QukHdjm?tYb_ z7Jvf%%4&%HvB4OnLC7@Z4t;C4n?WDls1Aq`4i>A}qFrN*nnsKor~RSxV>Pv{*?B13 zSDszon66oBB~HUtE_KO#4|+8*?WQ2=$K!{#+5oG9yAlivVOHCE$L1RiS^>i2OtG}K9y&4#|o z`_v1Jp0&;klb*F^E3TZ;$|W>ynF$?fHj`DLoG={Lgcxd;i&D7MiHN5N9GcOJAaTH} zNp865LK{UpXd8;|JUe39R^v6zM+sAqK9y}XtzOXWwV0vD18!>7mgZ8uv^{#p+0r$` z4xKASV`WfT!6Ljsc=yF>L}@t;I2ewjKM*`=<*_r$4mx>U=klxPM>~r>)nmA9k_J0q zQk9ewy+ZNV74v=Ox8i+3tnYVz@;6lj82p7Ku+2 z&SeRoo6OA~)pQzzF+54ANvUhM7g7KkR^Npj4E7b6FYC!A|OYtO9qXq-bS*)r`<~%eb-2d2l2&s{s{PUZQ_j` zwH-ohh@vmHC5V0Obv-_n?0OE94uPZF>GtK`C|3wwK6CZrivjZ#o*LHUjih}h^-)Gy zj4gmaD$Cr!?IhROFPRj8R4D`yR2fZRF4=0S#Oa}}efY5k7DU>r{PzSVp z8IDP`T~is_CPCMy4PKxl^3(!LW7fEjg!ThU@M6g_vden@aTw=sZ|7a&N~AM<*d0bQ zOht8!1tkXjo=Z|j6dk2DwlZqNURtA~Ic4)c>lMKIQ_4jl-I6iTia=+y@+|uS$OAYh z8L5_OISRbgliKfK$i$P|AW*S<{4gAJ%>Y7OcVK{6b;0RLAYG?1hWu*4w?yDZpr1q5 zkXW$;D*S%542fL=1BE!^faxJQ&PQT6sZE=5IRN$@MKPgh&}=6l;O-O%VycE1QJ-T& zl?}9lIHh>puM%u;K~5!7H4;dGo-%mmfFdfc!0o{w#omWp9INBnrZ`iO%1?SsBd+4R zeP|1m*tl`GV>hWa8>G?<2|Tv{0A*_MSl^L?y{a!SYw$9s_)lNXiw@;&j~d~mc0T-K zr4sp(j$47-rAP1bT{U#~D^p0|Dx3e2gNh?obX-J!xEH+dnZ-0H$qDNy4vl zQV7S)MnL*hA~PF-iAeS1AIgD<%DaZ#bjYYAP0mT{^s7cKiWnCy++=f2ZTmPspK1h9 zm6klCI-$YkxUU)UMZT$LsLd_hi!^M+tD^!j+P$@{)>3@L4WCh2^XPhg!9IPeKh~31 z8fSpTta);&4#S>#qgK0Hy+Nf7oT%jeE8FDomW2GalNtOCQMm91gL2`Gn?`@wJt-lc z9ON8l7|nVFI_;Ex8H&*f4~|83pYW{oG%X})F<=f9oPvKU@~vJ-?60S@WMpWZt_bv| z&?jKMk_QLS_w=m)04%WMYXRP(^FtixBfT4x7RGt$O)^;`uGZl>snSU!7!l{|RD-vz zJ9+gq0e8=kpYG~Gk?mOXG7K&{Hg(PeQ6EDde8*MRL2pp{3^VmWXE1TsZ7jz(@dwso?ybHJUb1eNQH?Xm0A`4$ zGnnu@Z1HW0{{V0LSJ4}vaViUQ#zrg8JTvgx4F_0`^5*nSB65k0SD>&7C(Ky!>BVT3 zgvo~0QXKMWhiD%*@I7%&1V2U@_NQ$Hvdlj^Vk1_F5GeaG$mlxL zA{!KMTpW5(1fEhEvJU`pxY9cA7~9XO?M*I)f(8dPqD561Vfgo;M2j9U4p@5Pq$m#? z7oS>$QJFGlP&=4ufe1fdr{_kXHGy%!|KX=8))ZU!bt(@3ge0Zfi+t>BxDNX2a~YWqfxfsgc1^ZVGSs^V!00L zdVJU!{VETzU2e+H&$$$8Jq42X7ncoZa@pf>LO-Q7uBL?viq_%ObV4~lr8TWJP34yB z4gUZ<#yAww2CWkzXaR~qPMV#*^orm#60+{wAiy7uN+pI#N-WWmqbHiFHND%W zB)UfWfmQs5hkJs$A^Z{qgq%u`+GMZ#G0c6QZ=G9XrH_V_5Mp|i>s+lRKNH0cgwbO)~q{QuJ=8VO+E3Q%jRjgHkO3JQ7Vvpwkpo z=mh{d6E!-fDlouMW#g>?Pmr}6$&U41ruV3tUTClqyXKevB(eq$6>z{pefJJJu7HEzltIR7u77rXV7G4%On?qV2wZE86@>+v0lhooHh9 zswc5gGr_^B7G@plEYK?=i?QA??^1oay=sPOb56tNWz&5Aw8MPlRa$AKVCH1cbnV`t zW+ddA4I!i#44+zhxfH=p6wsM|+HwzC+qAhD9M?A)#a)JCxTLkY>8;E60UK7|Xleqi%AO-XS zp8{H4F^XduK)gtdGxWuAtqXeWW?9`bNg~D?Rdw?kMi>c>->qm{>b9>i#khd6=R6Ou zHCppXw6_v8S4}Jfg999H_XoXQ5=U~8eTy&Q+@q;hQ-T+_#pH&?Wevf)z62uJ8257ypktC>W2oST_aFo%te3R z;5K$I`a&q&XiaQRbx+@r^v9(C;muSXG1Co8| zrh><}X*W7vk*C{f)@*#J6Kag|*zfhJX7Lm&hDaV>#4C;~#eDnV+r*5g*P4)C>ICjt zTsNriRm_R8?2D`Z&GPwz0{dXWyn2n zXt;KdY1QKejE1d`0ZR8tC-T}_j zRu*_0?ywtwhj&`%zSqhD7&-1vF;0fzL%O?(TpZ+Z0Q924bE6kf?*6%_3E(`3O#OP+ z8r&qQ$W>F^8bv6tys$e*QRzj3=IN0ceqyIRHi~0Hu0HgG9WZMqE0XAOCc)c1XpZH> zF*3f`qomLrUjnhLq#vLCAEu+8$%^E!|Ru>{o8%qB%f1dQi7pe|@T#v7MTJ*kH6 ze~Smx0;zqmH*EPsbsY5+8;(H$W4=WoE&0WcL{pK2ro61<x7D4wt`bf2s+^MD*R>m)aM-yW^HED{Z6gBl0>ii_ zpA=BVfTtaRs=0!LdktwFTL}{}Jx^g#nQmE*Etn|oTvdjJSncze0qAi^EyFN7+W_&C zOPH@^F2z{n46J;YSN`i9W`pM(4=Zkc^NP6k=*^QN7RMPh!vvc^SB%kcbfmf!_FNN! zM;%5f>KH2mKBlB}oE5-a4!qPe&9*5wj1^O!bB-z4jkIP(--@)GBK~Bo84eE}D~j>A zgB!$=j}YjNl7~QEHR{;!^{g)t*v&FeIGuJ9NL<%#@S9d^pAXGu%nRGM-S3JUu~12~ z#MURbGNiGxIN*jDBB9<5dQZfN?lpfG{f6EJjU$XPAu*GITs6*xbpsF-upX2W(!PZa z;*H{w;FUOJAk%!a+!_Qwc%|MdEvAj8fF%wock$Mt+GxR`EK9ghSX5Xvhcp2`{{VVT z)dw`AC)R)^3@IeZDh5e3)s5SrHR@jrG~E+YGp+rLOMPwgp**P>?mrr9P%8X8@X)im zU3*#CZ)+d=uQ=`AeY0N0XMbm@GkuEk7HnjNIj&1d=q&- zldD5>Cown5K|ZxfrJ$atv-v?>dClfGP*rh?lpr|Vz-N#FrbhQN0V#zM2nBH9Rvo)jxU4(^x&7Lx&r;oK7TmccxE|S{33>;&89mgK{OMX5w(VKt2R)Sk0F45$ zfk`-5#yXH{88?O>c!5)S66_(TbLvz9P@8N-Q_i1`bH-2UK&)4iNDyZYxSl|%aIU+8 zvnd>LQpXf12bj_lMnG(Tg+FXq`={3#rZT8cE;2;hJMqpb3?f`Px?p>AkJ7C?Hoqun z#^e3`iejawm@>(n;B!b~`NA~{Tpv^3o9Bj%lQB-?pfzEji)q3uk?X;x2;(E7t7Czj z;(-x0)GCVdN&r1Fd7#~ePn?~(;DMTV*;qDNl;_(yr!-qeK1Na3J*j~axCG@EMm>R3 znqVsNu`o>ih#l&5hA`#E21OD^9Y6{}^;19$+WSY9Z0K>&w?B;vXZyQxk5=ZI_R{|V z3$q_U)Z(m;xe}5GIO#wVmY4>>$@lf9?_}U?4tOPj#(k-RQs~4Z{HN~iBQ!nCv2ET8 zj=N0)Ch}i+%&DAnflIbmW>lS!FK{Zq+ChAI$_eMF&MCp=49CoEr?(tXJC<3xE6Xt4 zFfoo#Np9S9r=`TW3;aXYffJbGIRzUeV>`3OMfMoezG#iT3l8-IUCu|!5s2-IX}XZM z)$-dMlj-gCpk>x#7v)9#s*|cbo4lfg3~npMNNx$}0QJo$nzE)H2j@T$8Qw9CvC079 zm#qd{SpZpNMPvLgmCx%^s;C2OXij;C|+MM?I1 zPxG`r<$ICaePd*58XxHCTuJFdzC8S3Fv?EReNE%}p&)vu2nQZ>8T* zZliSoOkSE^qAg7=KoXaxm#B+VOHu+-)Y8=vYE9GuCjMwGQRa}fFdSZ*UYJ^VD&N4kyFz>qaO8iX|)>4>lxREpJ8%%tNYA&^Cg7j5-6Vj+e zYV0SaZxb0Cnsb2TJu9bzX7b9XQ(UEw-WAZmizatFD8~#tRMuJw4$OV@DR&A6B*@ty z=aY(>!FI!mo;8s0xI8iX)mD|`iequ*IN>b1~(J~S8<8TKnp7rEO+1Tk)FET*T z4>}N7AZIbRQ~l9XLt+i7yG=k_C_MnrUey7C7t45E+Xo|OBz=8p5-rRnWPc%vBNcW{~!Cx2Q|p?t^8BOmUH=m8#4beVRa?~at^ zc$@~w$M&QXIPTsr@E2LMEn4z+hBfn5SwBn^%Pub=Jr^+~| zsI)n+_(qVw?$t5sY2(9?vH2B#mD08qKxL6s@IF&exs^!^8w39DCY^x5Zx2Q@@)~P+ zc}N`WA5OK>Mvo+FSzD8a+$zV}vC+D8;+hf44-f4ioM)zKKe2X(B^8M1GuFD7M^Fl~ z0iH{)Mkw-kF(Yo%)83dIr}iv?LkL2*QoVm#k#yjtN{KVqXzN{3h_(W4$8*+^VMJ~+ zao37=5zm_l*o=qEdghW18DnNSJ9Cgrb^ieCS4H-0^T;30g@H_;>drCLQwy15)G1xZ zfG{#?!r535xH#bBgIb2yf?RI=DL%>-Nl;roGtg5e#%tRWperyQ{8Ig&J9kLj9suuG z=CN6nhj~!rA18l$WD%lb66|CH0KZZwxLB$bqmo!0kK(GbIb-G`?%T%YIjvD?F*zy^ zJm;-T9kWKeQBaS1R~kskib*Glwue>+jEYqgE(D9Y`tAyUm29L(XxtPxT!BoQIHYZ_ zoRN+V08oo99^I@x*+>-Nu5+0%I&NNut|h9*p-~$EU~@>L%Q;kI*o}GwH)i932#5}>q@U+aFx;`?;ZS$jpW{m5VT__Orz{Vkr5jO+#tU=! z(^}1lK4s<0=l$RCrVS=J3cE=#TyVR&&%GB53}_+ySh!<}2SUdfpxmiBjn+aqJ6pX- zG`N;S`;EztIXj7@`w|EFEQ*2rK+?7P7aFpT%y7Jru1tLQCxAaXMx2GiE_ld2f0anq zr*Ha(KIG=3F-T-NM+=N?VSp&*NN%j(Q5IKVFm~s-6=+-qUzsv}eQFf8jwtaN3On<~ z3L;#OJqiB+cpOr2i`;RqQaUC+Ny`iX7c|8z!wvOB208DO+5_tCuNcGkUq54iOJ;i z{n~@=ZWsno4mx13^QlqgU>oMxdhKzVEQ3+cZzHJ+#Yyk!RdsUKx;?|ne}e;Pj`Z>55!Hn|kz+xU+@6A)X?51K3h|5PMZT zNB;mlT9XO?06k3?3l}_yeW)5PH2iuAvS9w5=|3~-xC1@*&uMquA5&q34Lys+&9E!MAS(jV;$Z{IX= zte}!Rd(yqN8r0bFz;*G>>d366WFP`_R_&!JZ5uCQ4>g_Q%h<1cOwSQ4Q{-nSJMrGF z#^2~e9dXSx?h;mZJ+DB$n^4hh*Ymo2^vOk7cBjLB>k(%m2 znl%N~Ke#wGT?u5{l!hKxBi5r=X&Wk%FKlL$%$0HkOh@;(4k}2QnMQMw+|vLPlo(>o zo|F>g;Z?AE=cO^OBT`g`-Tv=NY#CjM@j)1=Kfzfac6K44x0(b=}I*iH|9v^O!sJ5Rlr!_g5Ql9B=OI$d{LsB!-Iq+&@eu}IJNM-;CFPN9rToOQ(-hJp=+!wZ;?9o$p5 zAH44w$4ng4Qr;Bg#<=u7C?WyCffV~^6b6Bq*~!A6%A^jvaI3rh-fANhCk$iD4nYSL z;S4MZ34`bwi-7VNzF7$L27rfgEskn2YKn3hv-PQ)YHTkc1EyHh5fjNGDB6L(qopyj z;aGrM)O4q;kxeryHu9ZFJXD@uF~2zNiU5d6akLyCr87GmkO@8UR)k8;fV_67+7JjQ zo^hH0giG@gjCzAZ<*zeep%PM3R5A(9i;TX~Sa=(>bS!SvIIf zBh+@P&3a>E4Ya<%3VFV?ZAQ+|vo52kxVFN18K04=qhCP0b|$IGS!~q@V!N znkj{`MSyWM(%e%^RBZGB=58t?^wxTbnwW_TQ%lrk>7}RwP4u2+LRyn`0w?n+43!S* zGfYIA%_j5fQPR)?}VnqVMY)Ar(|RTMTT0Tk`2(ri^Xd{iwyJJMq{{?Wx( zn&*zSm~QdjnAaH{YDO~hT<4m^xNd7k;!r`Xi-+q;jAxsJPfNXrtx$J4sL5h!0jDMQ zM_SIfx-Fh*yB(>*)t|KhR18aIuG>gefu|*ddQ|IV%i4@ClqOmAiT+&h#Kn|q_m?_zyL$g7qgG*Eb zP$)DGlz@8mt1%u$Ra&thYS9qO@g9|W4x+Q-Ju3VMrEL=_8&+LFu8PS6L_uoi?2pVV zegzDe8TQ9obr-s3w z75l$Y*A-@6I}N?%iI}cNW(wa;v?RM5(^{3L)gwIl1o@5rt8d-^01Do=)a{j4BREnQ zAQCGGX*W{4I;^Q2hb2ho@U6(KH2ZW7CAMGq*@4J4a9xdM*_x2gVJmI>R9ng5ZyC?E zIpbBh8(-w^2T#tUvQ0Kqx)|Zyqk-vG;ZhPbT*4i^SpTr3-J#c70 zX#@?6pYH?3DlG0!;&?n_tdWk?Zfxv9)VFe-2{$TaP)%anYO zF~@U7g5@)6cY@hJL4ipnjBg@1FO}_{4OuZrpce-@=W*#y^4yVtq|>lms9-?cs{jUg z=}lP@LBwZgX2vSnNip|F(wnkCerSLI@y}W;E0z4UMFVjQ(;ev=*W}_*uiO*GT4>8g zM(wB7YBOr>F@8^AD6m|BMvma5gzf4VrA}=FrVj0=oEk^8bNnc$)}FSij8B?!PJZy#LhYs#waPYpa!+9PC&c+YjuRVy-Jk_lw&)zOKGcq5Bjkow9^6n|I`VLPaf)dJ z2XSscyFd>8=mTw`$E!Hvll>!p!yH>ruO^`p#^8}GtDjn9Kr#y%4nNwXP|%&#uoOBW z_ELYXAW_~zev~SN8BlZSN=^ny=h}cGD4EH5G45%8&jRFx>&eM961c{HDg%;mRQ*Vx zLqsrPytGlC;+wOOil7194z%)eImSJy#lBpF*Yu*pD!LbNp{7lzCjeo;N_oQZwEZez zJq~HCg9K)P7<29VRAoaQ#yzBUH7P}0Ve%=(bH+)a3X&L^mn)oiJ?Y>k@3^DC2A?DZ zu2hnJDFh%kdC45m0*6Dv!w>6Rw~j6ZS{$2lLHTfdS4?ft;BIlwaa@nYcJg#Mg6HOZ zWd1ZUqspEC0MUx)gIvs;#I9F8YoL(g5&`MnvLd)h1i{p{{4<_89de98%&>A z)Z($XbxEZArH%zunS_nP=v( zU6#M&h4B}Ltn4nW++JIrW!=MZj+Mq^nr^}ueH*DZ=d~CB(ipqw-${y1v;gh&Z}+C{ zrQ1Lb3{L{K;_$Y;C9t?nHfD^j@WD?bu4@f&8FSM$>Du4K8+d#rWu@sd7r635OCcE8 zG03Kb&oFY^m^MKN2Avxb%91{mil$jY=Pbsp%L3U-VvvSm*A#|F8xuHYB$9g9POY0D zJAquvl{139D||ot9y87iXRQd#Mn(u;FlZ5BP^MVm=ew*8uV;X&5g#CV&G1ILk36nG6dUZGZqydXq}5!~={PUp4pbA#vEy1cDY}fh!MM zQEWLaE97^mviq2m#xMXBv6MhSCp}J006H*Hu!Og+;Zqi2lCdK;I2p!GMUdR7Iqyne z10bKmfFuRVg6O5a6!fIt*(1qpdelX+fUI(TX~I&xGxOSjAGj~@kUCRCqLMJU?lVB0 zgJU3lXfn85mQTb`F~e&P7z$*0+;Bl(r!@ZnFcllrF+YVdsG+rerSte@q^NoylQgl^71?_ZXP@}0i^H7gtPi9Q>BIDT77@Jf8qd$!s zV`VCQv-QO&Wqg3kKvluX8K4OM$q)A;{V1p9J-<3Bh_T~yjcOnI_h0=9tC=lbAC8+E z)Ias?zxoqbFTS%zguAJv(>KxG>W!L^ znu(r(6`rFnO=CH@q$ZOzz*1?sq$ZG>1UoclnVJcpWAjc1q1{a~ zQUcInP7;AsOs1Mq_IXeo68=)mbJUwNdU4I@L*T4i7Z0 zI~Jw63*MkNgVwXAoby$D);;N54R^WRDx{29)KL~SZa7KnKovZa#-@(Yn$e2a)K$o> zhow6KpADC_Uy8tErCy2*cB_#@Jt~qVvW!}AJ*sh~f;v{TFpo;V93#@I!nw@0k^aq9 zmrW{o72FF@M?p}_V6C1hGzpw1kjQ#e@iAK8xzl>!R$P{i8+0`#4~}frsO9&p+!Zw} za^s3=h}eo=m3BF69vLd^Z&V{zDR6q#xaRk)nBJ_Un5H9QIlb!IN*ldmB(-5D#RDx~ zL?lC8Cx~=$FmIPN+^oUZnyYtevnk@5b`w5(*EAP}+0}7e!zo7EMh$z;xuOXXu$AR{ z{)W&yI;rB6qtt2iIgHUk1x^oIOx9%!MrkOZ2MT>KQ^hnOq!hFYU@Fw{)FZ7;9a15e zBCF6Im7f`_kv1_~MVZLlvTvAIS*NlE0gy304RN+lh85On<8qF1+a|7B#K~@N$tAQ_ zqDX@w&NhxnG`>_M&``B!PZp+)Fbk{&lSkOA_s3Huk10kz;Dhmi5S?Wl1r)n>pgPNf6s) zuLJ>0vVgtm1R*2Zh}h~m=ApTV?R!^q{Jeq=Iiee_>9Ndgt-22%c8mynO``*iziGRnG-fr><$k3&0oe(Mj$kCDHKeg>Eb`}Xu&L%6O1T5(VZT>dn`qj4agr2sX605eDqJt%mP zs3nO#h&0EE8z+%ai*??_Zvgwy8WZ_rIp^w44C;rVH3~!`QlZ_xr=?4`a*T4ry&(tvCa+!3wgas5taJ!PE-oG6omjIKa~R-E=l08J?Y+SZpZ*GKN<<=APS7gxZ|xc z5(y4^b4n#_R6bxQVaRNEqvc_aXc;41o#1T*deW=&*N;k!tZE0!NcN_&h_4`kDFG=g zc;I?uR1#bmy0VZzgbr!Mt^ow&-juQx+qWmwiU3HaM_v0!�woA_uMoF;uR8PynEi zNaBDKD9_&mrbRvD4U!2Ig}LiaR@?_C@t_8IdB#b|pvL^2#EMdKbDoqT0f+~l;(-u_ zIKb+CMIl*9D#dpC0akKH(wJR$AC!tTiVG09Jt>9Q;|a&E zX$(Tr5O7!68fq2}bC0L30gGoV)1@()cYN8+FeEGq>(6mqH^k>7M2DU{&**EjR+n+y zRH^I*aGw*TCq;w|<9~7Wr1k`POF8n0&lR(wcz;mwCx`AWY#@k~oH>z>4}L#d=iX1< z5niM4oCvggsc=Iymd6nLLt^6XmPlcHt8j$+S{x#c- zlZ@cji(@yq0IbAddm4_;*hrAya_)_0AjTK<}}R@G@EnfWd8uN zYHW?bu5d?jON9`9$0OR5%1Hn)9`rJS&T2D& zH#bTEPyor@$vlkH%A*W3ei*1oT!J(HH1?QuQOy86DF=nf9)g=Of0t{3Gs(v^&G{U@ zDO2v_cOIv$04qw$cI{*C4Kbyan`s1r)0&$gaxiKV0Bn*ledrk=c43DZrU(_AJ2vC3 zMj}R#Fp-C~NYMo#g2&vRC;<#j=HxiX_j#uk00F)K01An>MsbXW?@Fg=z+x!46Dd^~ z+;Q}!4$RyL@H^tFmwffjA(BJW1L;M?Tivtns$})S#VWoJCtc@Tv7lO5lhmeW}pVY2B2fjJ=&lqn3xkxCW2}*X^2v3 zyQvAL=71fVOw)5ongCL0CYzcm2we7~G};NEViQa>;%SC}7}Ud0G{#^mN*HPh8wQ;6 z3{^>G$)o`9Rov2+CXgv6fFNWnY0=vg+O9_dJ*wnTP=K!uueD!>-`b^)996jD#Zp3S zjyQ*u_$MgB zfHZEcR%AR^6|ZPZDlrRP<-DR*_oiLgiBR!E3m-V^niJc|&ZS^LLLG&o$zJ;Nf$gxk21dsHcXYXrez zQXf2w@M^>n#rBCM1Q?GQV~kTlw~9vcupLg(k;W?{qANo5Hv&+h;sEt099E6J(a+2i zZg};oQ94aG%2RGR1J;{w00fM8?NQxZOl+Hbg)Ttij)UrK z*kN?ba-a;Za4K;p86yY1WW<-M;wJw9QlJ1q=A@JNU}SxF8f#)+=8Tes!TE9ZsLjk{ z8?n!&XSSnj2q5C07Oiaj6gcd0PT^wf%#NdL;10hoYD9F~fQRy`(cfFha6c-QBu9~e zR$O$?dSFTjJnf|bdJ2p?0)Ad8jVoj3QZeaBM8pBcDhVN0BMevQXi2=@K_CxNPLa@H zw|W3)l*+#>ug^+R=Ym0XBiwUHWB_rTQ&v@fx=-Uk4IWk*u$x(oq9hBv|6^LrWqnj+gr6m3$+ zlb@#*2Hl5^!2M~)PzDBZ-hjC;Xr@aigInGa|I7x%^FVmWiN5cXJK9m5VSks9ra542X zETbF)O0q6FrUD5G>Hwo}eAA1@Pf<}BlsWm6XV!rV7p_RAF~H!AQZ%u+`DEmJQk4hAk%I!3b0Z9vF;+lB=BS2gpKn@!skhVUR7;L@@lju!0StJ<&ay=*l zT=iY2r(hDh&jns;7J4fPapbe--!;qK_*++kQij=?M{qid z=(PK*%L^Fomd%tX!6bF4S`_ssrmnci`6qB4GBa2|+7pm>!bL>aFSz0} z%|2mR>CcM7$Twq-*r{8^X~P^Eq#C z#+NYcdIWwevsERgW*(WTlg4&xr^{w1(AS**0A}1TOw)yx(sP4ClMYkTSH;%uK2S;Y zsbcXcZ0?Mz{duo7jvJWIda@dB<+D}HE>qFZiEVtdobMes3XA>{R5&C9iu3a{gOj^7 zpK6d}f^kPMxkmg#yoXS{hUpV)k+r?}u0>pw#cu&K3C#9u1+Oi3gkr$taad4U+{+@y zA#s!GQuQqL&xco^X7J%cycax*?4#9T_h$C4OT+d;$4Z72ZZVeQy=_1MZU=3NT&ifa4A?WTO>#`lk}#7pz;pttS&Z_)}`~Z zp}@@+9mg_llsRnVepK=pIDCv#1c(N9l6~n`WnuI2=|zI#vo?7E^G|6Gc?1fCI4{7d z1L1Q;fNz)@Uzih)=9gq{#H-`jQ%kSsQMUl?C1@EV5}cMHedfpIrKCDsRWV4s$^#? z>(-tXv$=^L)d`wF+d?|~;*YdoZIqvGD7aX$obW{<&Uxe7qDg^lWMFh8)7^(>O#Nxx zEI6qLB;u6qVaY#AjK3Tl`-*CU{{VY9rE!vnUi{*a1uRHB)Fvj}NPH;lX;hqiywC&u zmUH*O6uSVxJQ3cTxP;F%V45x~2w8_9=M>dDrDtNGo}f_Mx?uhl7UT{$2IJb5$70@p z&M2t7)bygA#bWQoF`p1Y{{UXA{{W%Aa)4J^@d#&MgZ{l&{{TXIO)}ANCoCz9{L}_tdX1&2xXD^duJY7xB|=*?Tw~fw zi1SVhnw28YYCuGh7By-+dXhM~=~j_FsiYy1!|hh1i`tbItRmG|R}mwrm3W~=N{dxt zTor0`N<>LDMd7O2(+F{)g75p6E|-5_Jm8&N^nYw0p_mWZJ60^v`P0p6254ktMpS zA^X5`Ys#y+*GXtc_HF4Src(zX9+X+xy~O5ecNJds5(|ro6SX7x6DEd^M~U|<-*>soN5la`Q^=qiHBkh?}fP)}O7Ze)^7C?S1FuBe#G<ZKWM-z05r7zt*zZnVqtF`AfwC(QFn+w$S5f9mIF4}dpUSMR3-B{h%)|Rq z-AE^D6OYP=u-4_X!=IQ_9$nA1l}b|iy#D~JKa~stJdVE9$s#t^&b&guT4}U|kTw7X zS`4f51<$=6QjtK&R@xN%tAkGRH$Ft9Q`#Qku}=}>2RSqVsy5&j+-Y4w?@eUx4h@@OF!gOf^%gl;WP-hlIiK*)fNTLEwl zJOvpTq1g{Ko??-KK#1KG{MjcIBIl54+mtEYz^1MbDtMq{TXt}AMK09?0lx}y1d5Et z7p6%x00J|LF|tj>oO)CeM*G3tw2E*nuah5jeJM^sqiMx3BY43-eT5?^0~?RlmCoL#qj?lAcTxen zNK$ua=qa0tT=k(_a+0TtX;O2H&@neRJWyDk)QW`X1m>D!kPK#kBgq8(rlENh@N=59 z5aW|gC`kLXfXtEw!0-|jmK5MBqz>FT0!)Ce777LuQ(i0#CXHF#j+Ug8k^BMf9 zrZL(-(t(`0(~tb~%>zt7+^(nW+sCKjO16vMfH_m9409bP_G$tCub-uLjc9|MW|eKw z{{RRSwF{$`O(`4xS&z!BMW)1<#3MgS=~mqV#?edmWgP9M7bsfn3ACXY?NA)3X zU>sm#o+1GH3Z%)g5};A*^`@eMAvnf;I#rPs20Y*aPU5h^fO>=NQU_3QP*N(CBo(F} zN#Kfi3lnYww-lwL$R)t22JB$d8ZuPv>qWxFqGOS`U@BEztAIfjL8 zD7frhF=y*d^6Z^-ZeF|jphTq^2pHm*c6j_MFF0o&VI$t7GKC}zpP--% zBmg+21d;2CtLCu&0%>=-#@N_r-heG3JwHl$K~!-tV%RzQ)S_9qkSPMyBnI3`Zd&y(_AxP$_rqUzJ4` z3ypGs3_g8EDIL{UDvo}Fnr=WiEBa8=vv4|4JC9ApMk8!{)0=vN+Xg9vo{*NSScX4(y96s1AW+<03beFn3KWV{yaacG7Pl5C7Tn$4rFik1 z?>XoFz2}_weeZMc{r9dsD^K>$tl2Yr&suwCKC_-?pH=}cR6!~r0165KfP(x0o)!S| zfM=MPSeO{ku&}VOv7h1Kk>KOu;^I*f5fhNmPyuOasHmywn7CQ#7&sZJsb2}Raq{pA z2nYaKMI}Y}B)Itn`2Jynf{l%hhl@vnk59ozPfgGF|GGW30|=j?F=E)Dp)deY2~p4p zQJ%U0K%|@)DF4;~|ItuT(agei0^bOt`T3OrJ+SxledcObQ_3@LpPwDl?k<<<4yo7;c% zLII%tyIRQae^>0k=|zat3l$w54IS$ry--j;B0Cx(ItD#2CXuW*mW4Yp17F}X61ljX zs^8d*{NO*NmL6j`WK05U%%}gT_HUZ~KT|B||C46_QtW@~H4nf;LqR?~G(vzh;6jt? z-eA-ZyR5TdKN*%dG2q&OGhLO(JwDhdo}uCKwV}Wh`e$tU#Sq~TsA~@x zTRGDzlRhK`K1L`14EQXS^#llw`SbMw?Ze33DfttCChw6-xC3~BKO=q8Wp?k?YF?17 zG_mq$to0vGUjJ}nKys>ub}D_6e++EidHDPF2@o?gD3FM9@dUUR`1nP7_hkiiB@Fua z2>8DUaE?LJehYo168|5UkiPed{cFGS3_2=o`~+x2UgUKCCMfx{ROWvcjNsn|``4&{ z1hdG`CMbFWtmXM3Rg+ore7skD0wC=F`>@LI_DH(6{|#C6e^&cnWIz9p-2S;c|7WuQ zJ-7cww*TLf73a45i$4MIU#=0{Vjg5&ufKj2SNm_rX1tb(I0Y}7E_Ut(?EZ%xwf&nN zW&X>M|1rP1b2^e7$0YHHoe89o2N3=@(^UMc^cXmgG|k9=GtG5AU)?^}q50Zq-07&Rjt;&iE(1W8BE<7Ihu z>LRCuHgY`0Gv&0v`oB%NFcVtfgh7dzugvm2?Xcdto_a#W~5kukiU26)D3S|8EKe8iHjQ*B%6!TVT$XT+sy`+12luPq%b z7fx(huh&oxgKwwG7#;5@n@=?cbjpR5#e$^SgIr992`=NU&MGt@4wy3z6xXkrb=e~- z=44~d_6!jisVf(37C*)!nLsd-< zmrT(^>S|5?cgJ10e7#)ndG))^>(K zLrqc)1c1^`9l%7WG?{d1ahWnkBmODATT|uPB>)-?{1dnLFo&1{ZWzPD$wL$Qg5$GY zhm^SP@lfuV=wh}w6oU${SP%G&GQv)2f?trd5YM_GFPH9km2Z6vG=@gM)AapbL+wOG zi~GEeE2@(H(0(d!gnPB>9)qY|k>iWIdJ9eZGTp=~p!UnoU8D4L?i1ikW~b{aGvmY9 zqhL(4|IfR3(PtMlSX+?Q2|A^vt=IRsuO8}YG{ltZi)bgN&h9Ld2s01crGzH- z$B3YA)f1@0L!-Ed}6uUtH0V0L3gWD5VNkkKs`X-WDE=F`-rALqUj^+PPwac z3qiH7uCeu(^P;>f@d`GqlKa*}NDcs})75`O&(bwIpBWwZ&yn13VLASLtHmHWq{L1W z{)n8D^2Q}mU%akR^JXznBk8Kxhs*f%QfAjvhRx))#;G`ix*k35#&(K+>3ZkIp74wm zDAbho2S9W(qu7c+;4<7a%Fh177auYY=*z$HJWyf>vbl|Dg7S{AcO6?7V;b|#v3!h+ zMYXMt*dAu?i|^!osLBW=JT+%E>+tEe6drwUsU$8h1hEpaJm1XE<@SndJB^shE6o#+ zl{DgsWxkS{e2c}-d}ry(Xr2DC`}KUnUHPq^OVQ+5T&C>pb{0D>!)TNIJ=cv+SBWoG z{vn|XDx4)3?@_IfO1SYrU8l7e0$M&j0iEh81XHhPQUnM-r?>RqSdQLUE)6|MkHDrl zPz>ru9wiJJ<}r7<#~6to=CPCrF7;^4j_FTD8y3#mM`sE8@8R|4u?8A%9B6kaH#V6p zB@32ap9&mqDm|N9aPJeWeV4>sO+SnapN;&!kRu>CsmeDLvAH_sTGYqE4y`7>olg2S zX$`mzFTK1hF!%b38cj$eq3Q^ zeJEu=zqP{SJK;AqB5!a~Tt9sihR7~EXt|<$4O0WIfx6$z)>yq zJsJ`WH-*709_y*$PXOlxLu{s80HbEs=<7>?B3UIk@U2@;$;{Qk)DwVSZkdDnS;6fp zG+G<^Mx=U6-E@Q?1wF1QWEzvB6aLm7b`p`*Fs`C5Rb@Y%bJ18bE+s?v)paSGD>bjN z`%~c|9@)zzrOH9vWAHvdW`0`*=1_>T00&SV7#LKBr*bMRBV;}tXx%AzR$c0@7pbeV z8sC5`@7f6t41RAkm>Smv=w*B;&CU8*<_nn1u}ECXjh3uYJ+m`tIBj_Wr$_kVLB)KQ z@~EBNFHsvWr(8E>ievI7xuj?;&t}(xl_j6SlX0NdW^X4C=C=hXodfqBN`29&IjryW zxj&qBrU`J(*62zv2y;FG{7JIb6w4Tcx*E#kYP!bM;B(#wQRmuqn;esmdPCjr0>Rf` z)+?0gv~W7GF9nuqwf|-eX0qHdzIEHM>QK2{P<6{O`$^%is7VCWnmcIRzM3ZyS^gtmy(WiAO-EJ zzM>epzWvZl@{Dlq6hlf`QsHgpZA+Zvc7M)!hCe zycK!CkF2yoRfz?!*LpUk)klCE0&57l#qlvfNwg zpgB0Bw!2BG`z_X3C;Cu+=Az8@P9YDU4R`ma&PE+;6t&NcT%KEmf~e5LqNN}tQMR^oNod^A_Q?2UB3n59%56u%Y4b>JYP{uhV}EjbndtM+Ag`pJv!&W2gkYEaS|&WtO?+TiBN&M zqsIUto3qC>jC-5_?f${Zb_OJ2Ax#Z@{PltCR<^xYI>}MC0H8E*OYo<{=ld+eQq+!X z*l|JW_wesq=w+$lips|EBlV0TCJf7QG~&0+UEPizqTUGylY~l(;znr;>B*K;ydRbR zdjI&Lil*3f%7z}T&ij%VY^PvPS!ED7Uyg9ZlGP`p6W8gO;Vh|t-xH47f#vuOgyonJ zsLuDQ?0BvuNG=35B~V+SEd(@vw;cmP3?_GN%Si4;vmJHI7>v8p)~PaVTZe@2J4E?I z4XubTO18CzW&K_q!0L!z{g(d*Z<)(-UA;O! zkB_6C%7fiCSr%I79GPv=JWZ^Y8J!YwFkSx`ZJ7~xIk+nN_>x7g3J83Y=%-k^Kb7>iPZ`{L4n-K#OIi zp=FD7PCh*W-s|Emc*_7TBprE^MTf`sp%1z@z8C}5&U!uVFPm6l@uEW9&HM(4JUKLUDJ9AZ^)4)Gu!+%A!U(7 z5mq6l=2_1D7Eqvt{GqU$S7F?4TbE+ZisHe+qCTnGPCK(RV1GhE?u|VzQ=s--&AqG& zFqLGzRUblmYj>mV&dK{qKWW3_{*{t+f60K)aJCEu$|N!!F$`zr`f?b#tY&OJ@~&H4 zsR1~i^Yat^0!5N;;I+Knp94K2h}WDz6iCbyY00Id3#eaR+FZ-T9&JOLXv+#tI2fAu zL1y0=6nsgItKb_kZvhfS;^KZQBM~v0EEs_5oZ)w2{5NBGZyFm7Om#=teIIL=%0FZ$ zo!A;0aWolSg9K^%J)j*q_AX^MT?oI7P$fe@Sl_2NP#YN26QIzKV1zX5Y7a(Nwa^We z_77FLA9};6eUi(qfor*DP{BR5b(k$gKCaoWXQWH=ZNLk?-%@84Tv8VqD^Optjb?OTy%= zelso6LIoy5{BiF?pL!m4(be-Oz*cRT(w|DyD%BBB{rYZ)_l?K-8``PgpOXk-ZK>JW z_}i}XMCZ{UgY?T1^6F2$@ea$P{AU?+g`G6x!=3`dY?<0hgB*oNJ&QA@p@ytgWhq1g zCsYGONBemR`c`-{%u6*#x@3Y4kW{XC`WO{znAOU0EXTXJsP_~@9*v%UEd7whsd-6) z9rZ!krgo{i;gRAIqAiBy0TfV#LR1idJG6j8Spg0~HRZ zG@<5*hd`KQVrDoNpAjrA1fyk|bhtUr!W@$D?nd?eE%|&!ZcQ*so@h%%@HcJJ{~+mpIxv z8oZwUZJFs^zNSm(R1mk9D?c2VWIG^+t~@aX&zD zG@_OFaxp}XjZ@TM5KOf_OrMIQCiohxXN0|)7Xf(<+hP9MY816-5{S^FM2y4L+)GIk~*Vs|4zrVHp$N3xJOl_z?&pTe|_7uvpRl!w9_u!T~ zXq|8J8Z-3voEPh0qOgwS%=Xvxkt%;Cj#Zl=eDO=qP362^PR&C9BbSj;q7|oPQ-#n- zXSCSOLhqN){ru;`e<2UeT)%!8gnfw_r|Al)8!*j~a&w1)H0D!mf7J&Dsjz>QByZc& zJrOMRRnmF_@K}t}%HHZZ-DH0wSZ0Wqvc*0J7FY__m41VWVvj)lwlg>-OsiRRoUAmU zgzW`yjD`dLEJW$Co(c}%m(b~q3L6Ib8URr!da$%>1TJbCtowV88YZm-C*flevngy+ z`Cq%qm1W;Yx@8|gxcn5jDF-9QvMHCMf{*k9`s3KXEpYo{ZkMckKf|OeCRoz>3^fes zPfQN`s$Ef-n90D~3w=LCN$q;0;i%>UnLpqXrwH3ir=B7|U-GI|L8y0%lpH<7@+*L% z$mQXBF)j$W^%Zx9@D3kJ&ja)#{cV@!=hWmHgk8TGgj^`yoRd5z3b&~l_`u6au1DU3C9?&J?PR~Ov%smw~t4L7g z|I!ShOPJDWT>z&$=C#)}uF>yUIKf*>`&S4()dtzOZqU z_}!@dA+#nXa+FvfKPX3OA92;;AP6rJ~9SZB&D4jHjNHW^QR z>x%dCALFo_tW$fr#AxR%`^#W+jL>Bv&vNA}sW)AdXIz8ipV}g(7L2r61Ez1(1GY?B zQfIU(_l_ZtrSCs7y^A#WjWMhoau(WNn~l9L5r`o{v-b0edn+G*KV%H#AHNc{BPSxR zv;KBFLAIi;)8vx14{5<*C)o({aiJ{Us{KgGH`qDhMmkv)?qJQV@=>%=U&|KfXlrMx zRAYGvF|>2@wav$8R*j8CB+|Ey)Pn46S+ej|qAzW&Ms-T3<;F`_dA@u)GDW7_F*Fb} zMo{*#%n(~crnj(=$dUfhkuE(St)@8#n>t9WYJX0$MWFOMD(U0h6CmWx*!lTzQ=w83 z!+mf0-#YP+DZgVe8=1>~_2_doP-(H^Ha$ja%xW%!ZmB#5Jbd^9cNLYXJb08@XLUt~ z#b?JjoV8)-0i-h0s)e_n(x-ujWnXxgpRweffdcfLZ(;i(kmM*)?`BGZe6(p}uzUf! zbMnex$oK-;=40&>peB2D;4g~qsk;4~YS5uhL!BaZ2&p^ai{KWiIXKxokC@#Oe1OC$ zs*$u?hZOV3)6C8_NFe`=!4^?K=bJDtsD12c7WW*t##d`lc};V}fUVyi)ZHhOt8a5e>dMY+ByTBZpII zR0mBpF7z3u&*o%AdHSIq?fryp-#mYUCIQ)2MQ*9SI_~t(1^QcLa+A%aZDUDv#B4|L zgJd^N+_~`4%O?OcPUquqWcqh^&Ixfa^8~Ozxq0OSeF(~X0*qD9d>cj8RwMy{#}ol1 z+W)hA8SZ?k^_T7-gAFCur|=Iod}!5Yl>CInNZK<7W6z)P;Q(kWlG6R>Atd1OF>|6Q zaG+X5!HX9-#^68=T*Fs58}4cmsq|d)MQUK5>C@afZbiK)SSr_Cv{}TYd$)1S2oug> zL9p{z?rmqU%qdrKX~(Ir8k~wwhPzu|ktt5G3kUNztfss?7C$USjHka)8GYl+gY6x? z=lyH|bUrnLG>dGYogrW5GuHMKMNNGYI=Yt&fuBsFAOP!Mbn4sk-}BH zgaoaDcZ`S9RQZ)Z(&{${+wY5{LwGk|iL}#R>Qyi&giz20gbuI`!Mg`A@S-YMlgBf$*q4*HH?yLLv3a++3Xq zi`jB-$TEg~WmH6s?D*U`eC89YT3o-if}wo73Tc}I3(=~6?reEqXl!Jx(%7TxK8*DQ zFs%@DnM_f+Vt0ul!CCn!a=G_c&(nq2QiX=`q)t=N>5w|@%7am zbG4srRX~$mD24O*5+Pf16NkTX$Y&^#NsAZ|z0C8s;z2y@vvqp-tV8UXe7Y3`>+CJ^ z)biKktrNY3F&qYcl#Nxn?>{`#w|Klw5BzJJADC}^qn+v---&jkXUFcob3}IXYwOM91V*c# zgQPcmZXWE2Irib5P9|fN)n&<6iuq69h~CjxL>iy=bE@v6d6t=6nkgRLAjo+MrsQ(3 zTTg-3*d`2w6%#VK@BA&0K%yV@JN!N@@h6Hrv~{K z;HBP`h&;Z(i;RyVnh_wb5}OQjUxf0J+ODO%EUjzlf-fH;fnf^WJ~>*>1tteSiHBFGCy_H;3vFxG&O#S zcnWSB#+M#a&KqY#>ZDUm>NDipfe)kEzrc6QwXu&eXT4uVxhNk*a)tKGE!ybM7E4?L zk7tWF*Z~xebp?e_06TeC+I<7>yD+14pUXl6`|4WG+)-4-X1wCAR8qOyT5tKrHqY5d zCfgt7hVjM?LPZ({Cl<6f-E2w<<&}f`5y^*1OyN47t$>4H-MTH)oLq&<$u zoJ7VGChNz>l4h6s9O^=?Gs7I9Kv~lR{`-$LCd~B~f%A~v5~YEgDZb|N0j2F{>lPy^ zEhkXbanfG{Z|Qv{-QX|xlgC~s4tT!rEENV!zyFx*By?Ft>yCtaNC?Uen#RkurN@EZ%O zeF`P2-GasPjYTf~O66>&St2+-mah-c`(_R`%ucdG!eZv)+BqCR*I3Q{OP9%fVgii_FiRfSZqCra(Uq(Lg8i)=5 znE(9ZA!!ZAGv(k05|AeND}DLHAjo|blM>P9vtQV4F6uUS5~ue&QHKYAXAC;dRFsmWW)a3khTE6E12ukQ z2&wg-e!&GM>=pS}3d8Ii9_jItrC+?)J(LDXJtN>;ixWI{&T*qhl#hA>J|8xZ;wTUL zteyZfrB$E2wq?(MI^53aONuzLRM?chGD=GNan>ODnWa|ZM=zP31;t%<`g#MGGsdrm zZ`QF66fW?-d;2^5_$R;s4dS7~DcK-#Ii!@kh>_E=LjRqcdYx-`#mqI-+9}L4jyiY? zYr6Se99;*)>rwE1a0Qed+_hxb{xhnPSPvtJ;%LJvBVrNf2h4YaPq|6beaMjS)!zN? zV!j=nzY{s@m_e&$sl)!yP|>S=HLd4D3AdBusz|WgG;dGObe#`6-Rr)~o;NtS7`Ypf z_h9g-yN-RB4WVvW3t01fJS`r-QlUUSRQe%k+M%iKs~erHjz~5P&tW`{SJx#+e{Xfs zEn4|jYindf?#s8P^)49ph^UrS5@;(3wM^Fe{SeXc8Sp3KPsyFN((py!<#w|LW&>|x zid@ZzC*rjB=;Ow~7OO=f!~ zNZk28@0c&PoN%EZ=&8E$hnC&;8UoFuzHf~g0;LI;a(pbCUz-U_~F{Nv3 ztX=l@R>oogf#Gnz4XeinQ+URAyw`mfq-boL$)1;j5P$XXE>(&VPIvA38!83%yl>dWwv*gUE>%Od1jjfP1N=ebi)sxsOgKY|#}`08fgL+ba*tRU z_Qd!bp!X1S(VYHk=82s#kmEyn^sm8?Z{pOKaaG_&hm#*>so&BW)pEt$5ZbBMx-daL0&tUSaAz^-<-AT8clay+Um8!gyJ( zQ3y)*N4gIT#VTd_bS^O>BRQK>lx~;4+Z+_U-TRpQCbT_eQ<;&IDjnQ+HQsf&Mpofb zUaL{jf6*t&sp_hgwmHOO7W66a4*Gl%Zi>UE&3pb_R zbgZM@jt(S0*MftOF)SUM|1>5ti6`@jq4WvBvllf(uO=T$@kfas@aybMZNssm;kdKojedRcAY=Gd>bzv&u9B*bJD)bJUb%incR%C zPzga2wgAFgDE@heBs^$c$-nqZY=64>@<*vdWL}tR{KeKIJyD9#H-3{HFA-?k45gs$ zyiHHX3F9?99fYl&bc>TrNX0Gtz~MMW6dKm^B1I@lJ37k~Xc&@O6a8DMPKBXLG+w>L z2SpiAEl5*iQ~x_$xA@pcJDsj0S!K<3Q_^^(ZT~Rz% zfN^r{sawt_iRB#+JkGu!qNlb=O}i)x>!1xj-wrkQaF5l|Iim6^BFfF3+U#xkEZuU5 zsV-{qvcdcb&>pR<`i-ZZz|aB4+p~%o^)!4n>C+dXQ`r2R(fvjPa6$3{Wdv(J?(JN_ z@z{y-A<3Caw9b|A(F&0Z=vM zRYT%FiP_-wbPU6=3Sp9LouKUk<+r8E6RjQ2U$`W#SnG^pMlrdXt2FNv1O8jjxE`B~f6ySxWc z^gcq@Op_#-teOUN$dbdhU1}RVq`frPjD(tn<@%*jOe8Df8UU|$_Kdvd)!fgrrfZW_ zb%#|TaLE|c36Nj|Mr`r2m>{)K<~OZYUf9R?Z8iMDVain zl3)8xwvwL;-TKyHpVRM%YBw*6s|WohhcG^049Ro0pbVs6q1HzEQZ`H_kZB*=6J)Mr zfL9nQ&ImT%5c@6Ue?!@2e@K-YBd7Wx8?Pul%-uvy;FEv$nwigk*b&V`m+&msu}*bU ztT4*iq*{}t`b||#V_EfJiOM3ubS^nN0{u13eC6z7mW#EC?=)QrDq;_b0hAn0yzqXL zb=4eDr&keqU%V0&rAz;4)gpQe4J+SJ?a?T>W{um@Ozb5X9p{8o{poanJTY{lNb>ef zkjv(1^o!iT^n$Qzo#SG*TZ$*VS$;OFJW{Br8btSL2M0?3}sS zu&}XKKf7wAM}1C6(#j)t22$4~%@$oQ$s0rBHiiW+jJm$M8~aW+fwqdpNsy zz$`N3p7!8}t~Pz0N1m=!Gb_1@mCzf=3i6M(_LC7AJfr}TqQhU+pN z`QfnUL-ndjIbMpAqltXU!$X+pugmJ~CkXEAZVT?F2|-wKng zpOVq7-~4_|qfhcTb1PQ9rVXGgQV|-Luy1%{a0t$h>GFK@Nd-M#_i9A&1yOLD#YUSB zxG0UAQiitW3DA){y>I$}lL`JJZtne3;sUs)%k8F^hWbM`eFH7wz7*Pr)AsXQ)5Wf| zYPows!Ro>7Lb>$HA2TzmWlkkZg zH0d+c+SbKLh#7KzmF}7?%P`Ul(eE`9jt7eZ-nAurnvm#=??2j-nz0{hn^-R`Ru6sL z>d_#_jos^pq{B zR928|fy?n+&l5ImFkK6&a_?Ym`n_mqT(oSGw*XwzlM_>ATrAPNGY;^s=uj8e^rj)| zi$se}YP-E0T6PdlKExa20}Xmb9k)L?Rm3pxkL^A9q+h-}rYcuX(T-nw6JTuf^<^0A ztu6;ttjXfF)GC!v0W2qGw^l>yFsx~0(p^uZixq!6c(ej~dpSnYy&})DwOI3zBP5T- ze7E(fK)8V)=}0OKJd(=%Kzk{3-9B=bEy`paFTRRgIrbmYeF;anw}J)xzfmb|#3jT% zID~o|Z}zG@0i;!9bSN#hN>O4ydX8+#y)eLZe*Og55t!7gG;d+ma|krodnevtHaRJ? zcGKgz@&uRyks8-8?MnaZfvRSb`}$8XxxbM96&yPW_900cqLJ#^UTcuwI-*gdeoU=M z)A)M$7NUHerHO<&J*Jz}deFY#&!G&fbi}sfj6Tjar{cwWx;T^P=Gvs4SBevAngEFOX(whJL>G#_;kFtboRNeoX&@@?HXZdidWR?&Ah%~&e3_a&At`fzipk#z*^Gk26QH2gJDkvWs2==(b+fY!i;BAi zis9JQ>9p^lNrY2TKjM0%Yyeq?awPqfI$Tk%7llSSwe`5a=S4g1F3ZvuiZZ5FlU%$$ zyr$dvLz-{C#f83Qxm;@HB3RnTLMFpUolgDW6xR?h1JW)uei%##_SZ8{9dCj{a_bu9 ze_LDi-cnUMQ*nN&OKOvmraH+k(7;#ThJ(o$tDaZZb45FC1@=sqobRu8b-eTnZ~M7P zgp8XwKK1zZ((um^trvz)&#ADXSo-#MS7}~5EcdwnG!0LE0xZmgzlcxG*z@)@=O<#SAkQ6?5#8Z9S63-4 z7QG@O_nu%%L8mXBOwTfs5Xf+#Zi%On+MgZB3Do&JSpq+BYtt5{Pky0-lU3K7z_tfW z=3(D!BqT^=c%4mAvcS7Qf*7fQQuP(EdI^d$TdKH51=Cs}vP(YqM1Lmn$55^cl}Zvr zP+gGvC@CoY;_BTn8Hyr?BpJWUj_pX@_EuFbROeZK^J(FBQoO3MqU;BqRUh?j0_>ek zn@~d{oE|vA*IL#B!pE!ELC&!cg_cp>xAov8-bgS#a z+v^*x8jHr@aNojw&>n!>%~dJbDOPsBp@6v0tQyN%Jyc09B**&d3I8M6U%cM-($v zq-Q7+DhFWUs%$0$+HX_1eo9ba_16;&;#wB@QZD<^8$S;RGJJchh-b5v?EYf^XuNdM z@%H^Q@X+zws(V&@LBtTmK+U~p)d?c z!EuS#eQ(u=?kb((!t_UBjSTy_PmM0Iu>_yDhXyt)kc}jTpb51r=hIfK&RdO3lTBUg zyj)K8j}ByNOy+lWcE9!$zMuJrx@(umWBbjR@&3YIuRu7CQ!7K5CGHF1pbSBehddkp zkH6aIICu`(7l~DP0BTGIyXesAg7`U-=y=zSzlx$dQKdWxo#tH6FB`e!Qq9Sb(WHmN z31$Cb4Bk;CmPsY)<8mffElj*a=)5Rgv8(G33Qb43yJ0@fiKkPC2@W!K*BI9r#r_1S z^)4!}4GSa}DFbng{T?)Q51*M}k-o6%uRT^sak?qO*MxhN*p2e;^RKCjyzh`I5 z&s=*)(j%#|ou}+eW_ydQn$y?M+!Q~oCsGSnUZ9n^&V6O=A*_s;rAq75;(JGsBT&VZ zR?|*9RTE#A%S@yA!5)FeI0=iFY`3z%cThzOIvGz=u=zuhaPaoHPAJN`+hACvz37JV zZYax26E#EI(*|9R0j+4;QMkcY)O&$8$$s2i2W$^6BGd6TAi+Og$-t%U>QdE(eu{A% z00B>M$T(4wnVwHBrt!WP93xF}mm41Zd-TRBUTRm5Hvc^`bfG9Voum2fyy|tNE^?{R zA2ojc789f2f{D{Nqmk57L^T`Yw=rs2pjE;HAZf~9kvGtem^u~!mCk>QaA){F)XnI` zWEA|M`X~T4f+Ct9rH3Pijv2Sj=EWaXNx%LJlhUtz{4cwULIbfCxkvzswc5yvwtw{q zz^j5hw9H1>**vFcC)auRH4e&m$8~=F4Tia>S=+UJ~~gzf&HPv)(tA zy&yFn#x;L|RcUr$*h9{rIL@iYLAus>CHQ=gnO^G!78oR5Ab8yICo;=}y{Rcch0boG z+<`*e*^aIB*c62rXPjucC$Et+*qx5RcykJ?<^8Vv%_Kh4IL#sdUUO|Ut8cZ9o?&ms z@dO>QX1L#|^_u=v4D`*kgXKu(<1Z9z==6rN z)fg=^7@53OwF`(UIvui4kfY6q@6Ji$s}sJOW8qo}0WpRlnzo`<4!dmgvcQO zPZYa2NU^y(=sIp%qLhkXB8KFH2;!nh(H)oNZ)Y#Q2fTP?UQEdwO||S3(Q1I+#Qb7e z^8W86CH2x_$x^Xe6G8tWw&@)&Cf`j$SDt?F9wB1gnA+mQ6rTPFg8u$S;BakcUX;~o z$&66)OpbgL=hx5J2Y;#hpJU(kj;p$@S%XcM7Qc{bg&Z6D1x7g3O)jQHA zb`yZha9~X&j=WV1_DAzs5~wzvNAsehg(~kDW7*C(icQr$L`@7?;z=TNafgmV zCYwQWJ2T}9DWj@QF*3YUJ`G!X6((C+7IiSlQBwzwd+@K!AWsui+nR|20(u;?=*>&P z4A0Wv0Vj6PU~=t2de6{KyT#Onu8zr2=clo;Z!*63cJ3;d3L(p#%Ao%9(De^=#5ekH z{L}R|j|Z)bxX8jbfJ#lFcKWrcGZM~p?w7?W8Zr83b9SL}@6Za@{+^l!o%YI@anM0Y zqID|-9tK*m`>D@2i#efU2QiYrh8--3X4kkmU7a=T8W``2B!JUtf?d^YZ_UvppMiPr z53*ZQFXotECWAq5xixlxGGY+p`OSodx)uZ_UI3Up>DF5RgIGdw!dBCOn0sjS>=1q2 z$9KP~;XQEYY7FHY{K-@5|d>ULth6C$C!iJghv(?K13r96pog-oXO6ueSf9d&Fm%CHpRq0?^~7H_eh za*75GIB0X0`7>@@mSH?!iU~-(jFICpAaF*;#s{eumKh5#yZR0zqdjfQ{7Uf$5ebZT z(!+6EcmR8@MCgo|wNQ+*C}nB5ed#%W)B}4K)0?zXuKr?j(bzM+^|EJb0F$h}iCdl1 zhU4Vh30`_pU@F{o8*=>SM^7G8>@dtaHZFENmoR?L(%=^7Q*5UGAA#~ESkDu{#=xVu z=y*M`N>ZI3@SYSeHHQPX-#~5PGai^3TDlsORiK<7(^T?j5b0t6}>!zxj>!BfVpd;NfP_Tk;u^asVdE z!*ggyYNPn-XYvCLel=^`eeSaUB#D|-nl#!=$Kf81ZxlbDNiQ%8I`0LoZh0}>W;&Kt z85xRm5^_ehbw&P`+J;YqjjMbTe4SkI!vuQcgs}+`m5$bb77(2hjI>j*8RYf~#uXB_ zhe?cRjpWlF00-08CfxcaU8am3MmKxd7#eFBtd}^ ziF5x5Bkj(w5Bx^zJP(T~QjMg5swAJ1mVggsyoO$EO3k9L4q1~;!oEB|{DfKwwf;@< z1W-xwe%wdyBH;QN9BQ7qsI%D+RgAVJ*|%L`&?lUun%e);$>8~BtotD~RF(J$3H_CX z3ZONYy9`)%hesic9w+1WK3i<3$BW}Mghkg0Mw8jj&ksLyn0qG{&tb?|i}JEn)0zSv zPw^Wkk9E21ODcbDg`{W17ThRx-yWW?>+AiU%=QDt{*WHA#up)l{PW?72gf@ZVe+JT z%}P}GNUV`ft72dG4aG6n5t zh;vK6b0h15;1r&)Pz`T}!j;hTa&+uuP|`09y_7ED^VZq=_FtG7T4cxfeXl4iYN?zuy^c|d#qJOU6p9O z7;7O&`W0uwePk(`_2~jzN2&N_xI~NE2ib?6o9|OofAWW0@^MEj9WRr^TpAg8{P_(g zD`h#B4^8{^Cu4dfwO@RHxYClE_{b4;fd%ZZTMMU)BGUV`uYn;iR8XXt zas3DL;yHFD_ooy#->Q53bcVbpWrr#`P_Izm%=_r9@j^9Gy97R|^FtAmuIbt@ave8n z#7imy;pN?I3yw8yt%GQa2d%N&wt1v*&%*^d*0;*Q4NmD@-(gxJGf^2B&X0q3x4hutD0e{N1c8L`#k(&gW>xj6D1|rwx9a$-*ej|bE@GUQ<=&aN+8)q1zsHkeR%R?uT#B?|L+5CM$L2*-cgdBJ*f{lbk zTTKRAktFmpns$v&IO=x;VuYXyQ9Bjbp$RoeT^ukzwD*Py23`^3*ErCaMm=h{_pRKx z-z`8^q=VaH&vUDM1J7Uox5B+%j(-aGR&HKFfLA!k6g+6aR8}VF|KjSc;-dQAxY40R zLXhqpx}-yd0i>IOp%v+l0cl}CKmqAy=+2?LL;CBnkwZ7q$)_;59Rayr35mJezh+3sc!Uwhks8ZdNE^mcT~Z;VWE!{3 zP*q6nuNr5SWL94#v~)4%pyHYhM|M96gJ2+X0z^HYoY?b3OE(HjzYc&cxESB_y3MPs z$nzj*R~o-ly*Gl?Ue-c$@dZt@Q-u;u!N}IB%ve`b2TeIP_(e_OwDN$g0`4Y2^$2gC z#7{M+ni`~iWUDf$s2^E!SoVpny1`#*`96+2?04~~LnDY2WT|s#TIU!l_y2fyYG@Zf zS}K2X*UxR8@~2_i{Y_EpUJ^PPtkB9_ht>QlZU$qjtD|jhJVRIb=X3~~X}q@IrfLCp z5kfS}rD=qH*aGl8rqUkI@pG)On$@YuNvSHT&9k0Uh=$`u`+&+vd&17kmn`&=hAvel z^+Fyp`yVXS!`o%JKkFM5a&o9njnPZX$x&MIBffsDex)nc!SRZIlD&4y-!N2;Q zX?<5d3ZEk5({Lf`QDyeH+r5`NRWyQGKFJpxMJ$ckAYPc((=Bw`S|39Z!x)PVJq^L% zXF9ujAyYYRFh(F-<8U_gO4Y#sV_+cSwsP&g`c|;VYi!6T8vrkjeVR$&Z*uh?h@;e1 zPxm;QLho<@k)wDmL3$hdXY+P4rPDaTm!Gmd*5A1B<^KE)Z2@NhvzX38w)3;PpDJe5h1A*cGoVDq$Y_XetdZ4Dag zl)YfzA0nmh?$pd0Ro!n)5C6 ze#J=o(}7!Y2cd29#52K1Lf3F4F-pr$Ji|c`~`XRlw#w?%|{Xt+a$$S?Cr?cTHd61ZXrTe^awJ6 zO@nj={kQX^YievC3qYXHsgBCkmQ*x}(BzbqLUUtF7Mkq4u8=aC+aFV{0;Gagr>P7o z+pwL6?gUzCFt++qn{6A>HJr|V%@FE4Jpg>zsCreU5=T!PL1oG zpQ=-Awq$&e#0eb;PPI~FR1one=UQ>4o;Wn=;FnQ{SzSGw@L-K0nT^BYky?@vWP6~< zBU|K6mB2JHs<~QGxu{4wq15`#g^=TZ?qjIxfBV`Nk zTMotAxhOFGxxpL~8`rG`UmEMFFI(f^5^m~ic`Sd9ank*`szhh0$A~|G1p0dH6}XgG(BR)CHX;n|^~*M%y3TDO&4NV9S?+x1)3oOckbPYe%boR#AVr5@Yfzz^v_0 zC5iDMn3f@k3=rZNQb3 z5Sf1P{vaE4P55tas~NvUhXu_eroa*{qKQAtmOoHd}C~fvrG3kDdqt*gL(3o7H5-4+2=~ zO-;Yb8i&f3A-xhO1t7!^&2qMOZrZq0*|L`CKZppW+~d?SIE{{VJV4FVDkz|{?N??h zv+pMIEOpo}2>t+RLofliFK$~LFNpH>#bb6aAzRa1mTQ`MKz7gNWTHrmrlkRb41i=R z*0tp0lnnQX|G7e|^h%!-&t-(l_XeStstE&VhoNFIYEJ_ZdBfSp@tsXUPCzJ(yk1i; zqnd1pp?ND`^A4%38v+ZM^`5j}SD#81OFag1 z6(FXmYzEtffQYvVuJt8xVxi4`xi=9({2%C*-1sg;s{bYbLHSSJK>R3yUH41F{Yy*W zHcBhHkLc#&_1){J&xKkqQ&oPkxaw}7|LRTx00#^bJapC-#0^hRWV z6xHRwHhTow>!0|m>ExFmo6U{9q+20b)9pU_D+}e6EYDG@_3Y`1sQj(kGN=wy5OG@ zPt^nfxvzqi4(j>a>2g0_l6VnoGEWtmxr|jWvtYzeE1Vryp*=ju#P$hHh8cFyHyeT7 z-(Dkk`^Jze%a<+2x}{fU0<$bkCn5GsIOeKW+kmgxM{1*Zt$#?XQ;^hLBykUD1KBFW zNCu8ir?TsXF&Z5h+sd=HT&EJ_o#5|J;T__$37IGK4+mls6}#lKX1mm$LKpoeRQSP% zUVGddQ1s6J-XpcD?iR(@+_AB2AY1<`q237^JI&4c==UZjk5#ju(eGrmOmME{I^q`Y zQV^Y*EiCG2up)%1XENF2K+^WfCfBBuVUbr{N5?~u2sJ*T;OmyApftID57nDgVcSzzhE@TCG-D^r%VM zEoem6OkE`B(_yq$qO$VbN|0Kayt^;rYlXoCe6HoFiR~k)db?~5qC`u*{=*Tud5du2 z^y~BZj9M>K!w#_~`ude`DCbD(-}Hg$>!F4T)EgW=5#&Df7_Ys}-?k_XJl~7PqZJS$ zIz?lm`T)-YNj~HIDjkG1-EoZAxBoy59-nhwUVW;HPoQ}JOcf2uv6@w@T3RcOlHEQG z!{*IAG20^+iwdfxX0fz&s94Z=W-**6_9;nSBoFE@2D?BrH%eeV1Q5<$luw(Etw5oB zUah1fzL~W@8`TP9YpEY|UnqEY9((yC=^AzBm8cPGTN?aa&)j zHedF5u{JTrq2)q|Z{y>*@Y*T0B0r5X?LUx2TG!jEIvm@TdQcCl zA3od1iHa@4VW1PToYJWIn`wj8fok(yP-245I5&-UI^JW1@wh!4#LQwmBtAWZ?i3o~cB={* zFVuT!s3mG7s#;A!x1hQhn$|k~f)z6R`8JwZ9*zn>EIn}O47^1zF>AJp6~P22j>{r5 zGvWBwg5>5t^)$LqgrW+zF{_%2bDQB>QA1#>`3ITg-w$u0CiF}HGcfVyn$^iV+!rDK zuSf)(fd4@M|8|J`F;@|JTRwJ4Nk4UI0-i>+E_W}i{ILUhTUj_qY1~S(>A;gCjw3lL zjuKUxPT1hF3Qbkw(k=dBNq9<>CqVgJRwbJLc#EQTF{uwV^;wd)l7vLoLO)oneDyA1 zSd678l2Ag?1&OVGpzP3)AwEeEiPr%(WVfZgVvN+ye}XLFyWpNS&4vlZ?vr~_BEd<* zAAgOxnK~3`zo^d(*hjtf#yyD(if~xA84$I38yIZ3O*}6+Btsfm3a1G@zAl@ zG9?tB5?_Nry%oPT2HC3oBPdW@E7PX$~q~iis*zH`kVP_yL z1!#A6X8NTVJb-YAK`#1WZ>N7fPg8xU4uN@6`Qj|CIID3;1tx~NiGin*!nIt`Qmjc? zC&SP$9OY@&>!C^=&rvwiJ~JNbG%GRl(X8HMbnJ*bqWAXitmiI`6*&}7p6wY#y#&L{ zm-E@FsfWW*NMqnChuO*37$V|t)?`y+*^iEYZeBC7zT)dyQTaxO>? z*bO`Laqnf7o@%CAlW4|w;YFWI*2+#1Pa1;&EPd`ZjYgjG{k&g~s#d^>ZCLS!8!gc@ z17;5Dm=zdo6BS!iI&8K+<=C@=+sYEe-^is0_3;&ID^5*4opi**rnElN)yc#PAz6K; znE%mi)Qe3!a@4*60$P38mNta0*u0&q?5(UDZQxKU7tWhHbPVXjXr}pEYyk1p06N=gVj|n0S;IPx`h1TTOMaOPmIf z4fet?%Aa(^!tU#7&iTV2bF!u*1rrG~lR4Sl0dphu5GX|E_r1f1N137T!tc}ySIvkO zwQrzoS!ZKBD}9UXFTOta6y#ABpA?{!%)-j)v&$#4gQB^a3K6#lhUKWp1s!IAX!W?8 z7f?lp;Gp8%U79Kaxcg(6@tQK{pWiV1GhCf-{dT4*ZPS;slVt zyfl2@^e)l{bjfB7DS?^YfR@-`;6E*!xB712b5z%Q6A&D+(Tj0cA$e$@KRtB_J<^vv z6cviO_|dGjJCOIh^wcnA^P-x-acQ+x*;B`edGmHU+-}1#F;BJ(q@RitQQ*HKC5Xibe)#7-A_(SK_r&DJ9kR1NE{@` zc3~lND8WUjU}T7Mm8s;p!CSAYYwUx=5zW1~jgDJErD|$};~vV8=NhpZI#JiSiD>ha zjVwBoVvTgN?04m1wTEg~3zNq-s?#-nQk(~H6VElRMI@)g($>{2YIZbo((G9RBy!6J zmRak@{wA$Qof{%@DVFQ@DfQgMDjve-_A-0a*cy3BvHXss$^Xbp( z9q*zJG-q=ARhz1I?1*3Jg1mZ!+FF8(q6fR`@^reOZia7tKzN{C&ay`x*CIEFv%bKm z&IGVX)y%f(Ia~Su6_EqYBK+4@T5E_SW?W_b*tRa#3}}X_lfO6g|&lyBv6f) ze%QQnKkKHOve5Kgu}kZ(WCe6VCMRPp^&Ja=b6>F znbv!|QJ6p~ScYt5>&ebLKZPy0{oRc}KQ>jW7b^|tCbj2GJ^}VYOLC0kO91DXv7GfY z$G2&rgg9%uuj%h3r|2)4hAt|CZF5UG;9 z!2S;u`&J>Wc?&@Qu9S}?SS>Dv(q$GdDuz#(M>ip>6i^&r^VGBC>AUEL6=WsEBra0w)DL zcq_wvIFla`w3Bkg=7_w3)V??LG51bI0)|f z0lhvrXajwYnkfNO*=VyWZkJSElMY}k#GzMyz`T*l5H>|%k-nzp{$R9%0%-lDHK*~wr<`v!x+@6m|Qc>hSW$_eqm;Bo>e<>TPVP^ zxAxtpY_t5X-BM=T-&5Kc`g+~>Q{NptirQiqU{749S0q9})V-@cvHc|@RTeNznow1^ z92bfkzWC!-?*aiY&KLpxsBJ=SWS5tk%w^-iyT2P$9dVcR&ts96C|BD=K$lt^4XV`j z>5jUIUmHAkhM>!zpr^=sM0ut{a?hoJJ zwM`hoh9I%}rQ1wFi!8;p4sPya)v83;g`(Hg)@l?}D?hjVvnGjBTGr^IDb~H%;?Na* zh=I~TNq&RDdgXH5^?>c&Z+C(#`&ZtiUHuX6yZzV4|?(9FxMf(9J{>gChNK<8dlBTBZ)exR9Wt$7H=d>Nt4P^678eJeq>VjIOA zn4Gt3z+B(b^42aeYTRG94fn=yegYd{$B zJYu&yF~rR9W8Q|8STS)rdWiylCsZp&F+*TojNq}qV`GGF;244btiRBs7kjmFqFHIN@!U)U z!Y3NW*Y`$*C@+Fki4Y>1)F|{oVxH1*E+HW?mAFWTS@}7#4C|hBF}>xQMoh1FKDpD% z+Yv?7H^^#UsT?+ETHsY($*u1>8U&_hD)D-B`>_8x5KMo0RZ%2&L=l>Di(!Vwr9PBV zy^1~3f|uWfM;KzwznC&x=$;=1b8@z$2>g;*9xBM=^a!50*w|gdkl`8@nC+nA=d3GV~ubd&oYz-2BRm94k)PpE}ye_J2NyrkLjF7)Q z!kS2ywUJIoqaQV*Svuq-FUy8y&Cwx6?G(}pI)jzC*BLvRtW}n#%GP_$g%#Y+V5#xp z?~~)XFQ4vg0k$#c+|JG!i3ByF-H>l6?*Z(@U#vryFQzS2qI&!v>+%Np4I8X?#qdKB zc*9@krxXpjW*ItT2w_!$fy4@g_1;7>K*?BVsd5~xXy zj^*TE%#8Yd-f2Avht-PFyotxBeL0EoibOes!eFVJ@MHch3|1FaDr+K8R7IkPKl!=q z!~UT|24$U-Y0B_;Dc3zN#gu#*8@SZ8<{Nda2af8pfo(iRj`_>vhVKk1d*%G9)}gvv zGbmQ(t>_A|AvzxeqXVEfL^HX(J;FSo){M#h9Ol!B__b-nD$aX zv|Ju-nm{{NuoCQ0dP37L;0T(Iwv^n5A}z_MNk|eqXGqmf4QafV^8MqWkWzes{ zDT5Q%BG^t*>s-`m%^mt@UdOR8vFS(#G2yh@{2k52btm-7s6>rkq+c6>H)^sMl_0)* z7`@i?U4@C|F8)Gn%5)!_D4f`3uDd7Yk>tcY(@*o%M(ieKr|gm{slTt781&D9;~5B~ zoC+Br55{bDz4%oWYP8V^BH|rJ#kuT%pdSj;LR;U?g&Xon<4>xBj(EId@XvQ6T@E8K zt9<}jAHh=!E_Em@0J-bsvS%r>t+!mjkMNCP!L3W`jzWsd9F)aB03%(Lv7{>9g*W+NPBQ`;sh{HCRrNXl zY2nFkCF(nWKs-u?DqzT_L4YwMkj&>J#*Uaj3q4M6%stLSUpVcK{8VOb8ps4Xm6iX=`EI&iZdb-c{7UYRPp+b-`0-zU-_#)M~~4?jGPr7m-3> zL4v&V$ysR?404B8IR)5HIh6Q=+li z%WuOsjDR=c(4Z27SqVP^pw|v*vxceJG1Fwoi66RUc>}-R@YUA>mJ+Dfv}cMm>$3(m zMn)?I?VN$O5|AD$=9#TD>jjfqSJhwaVW)++ZQ;wK`j9?+%?jv7PGU_z_ zJtELuRgw_WS4BUCc7}FoNld6N7SY-CuP-*{=l8X947MCJl#X~Qdb*XNU@fBx^^k3= z^pG20woU-~@sS%2^P^tHm?s5*-%QwjB=xr;3iZH|b7K?6aChH{D^MZ0nSn31W(3O_ z5Y^^7@mA!wBuFehmAEF$o{G`;F-_Sz_RYU)XzxhqI{QLZF5ebEUfA`1fGft~e2B9O zg7u5SjZ5Ge|7e`y8rI;D6!^T{*HMtyKe0IGIXr1SqY~=d>K3WZv%aH0ZMa1BOwxpo zp7VyH$lF)+HwigL)mYMdfu|xoq+88fzf4eVyb%#?!$ga4-KYv>Qv5pIz+7hwy;un!;qA zW@7($*2~wdp8&yGIcleUW*dXm@6E|nJwg0shZM>s{!w{I@F@9rhA z+U|w5++H#kaV)Ycs_D23CVc)NWDH+5UQl6ORt-(B3hlk?ciz8^YHKA`h!+S>{q|b8 zRoKFRUa9zKv78@{XJYTe=ILvF*6$V2l4!*lTjZRQkqzc6PD0Sn+~!vJ(nlB0l+!2K z058l8W#NRO=e%%NLjYfv!e4lFv|tWoO`CC-QmwhwHRpt4bbq^&suGBd zNmS)Rc$%W+fQWhZak7T{m1jV9@sN9$+C+^LJ>6>RXj=ci+ooQV8o=mI5hRdfTTD>E zFl*&-)F!$GRP&)OYBa?jqiAsHbNtwm_U*W!{XSicW5eq3OYSkc$}~I9rOIQz0D!e= zS`&LwH!+vMMZd|UJ}1&3S!sEbyW{gMSL6@kc7HHZ+x!Qy1;7ET4jC*WgWBHG<)@x+dkm+UGdKi}mwLu0}7v-t0Du-Fv6>rBD~ZJkWys z$Iu|1OIwBQTC%_noOlN?SeNVcGKm9+t69#xBkObF>Ti)I1HezH?N5BIr%=iDbE8tNf1m4{hg@o}Xuh<(I%NYucFaw<#Uzn6(GL zce>T#3){hDzx({Pv`k1B{;XIp|0dFUUwvJ&cR1et%!ztno6a+60vP` z4m=qD^Z^{>b&Jg9HvYkyt1D$f#9jM zT?xMa*Z-BO%9`&&Q0gB&+oRZ0u7F%1gPC`X#>B%F=`_7}f6AABTmN7{1C-e1p{-6K z%NYOIOAnzh90h@qMMJ?5{QUWo#57x7ng2lduv~97H#%xSawdBT&tJ?wD5-D0s+9~r z)|Sx{n#SU1zEkFj=#rRc6HX&oCIFTOg;9KC?cG1Vc86pu!gWq+-){T#7Oa))_WcMj zst``A&_>~pff+La)NHU$WhiGG{6~)8kS%=W$en+X$y*&GeyR+~z1G)UpgFCAJbHx)D6Wp>w&H%NTGZAvJG??| z3H`GP-+N-aVj)jI$dOo@*-4U^XK74azM?8#D5xVRa`Rh{pv8#Mz|%kEew%*oqK>5W zz4zAm+7xF_OS?k?VdC`dY}sFiVt+fm6KyRI(v`J=lHFWY78!#clg~Ng3~p=K($)g^ z$d||{+w-iB^cBes)>Ndll)xqpoNiE>VSz%~;_2gc7sG=o#N$KrZ_bV82_a+;!-f$t z5?+=_5`4xJsRvl?EEp#s{4eZTcsu(;neW+{M}jiqy&%TjL7z_|#faim?$OHh&!$#r zNmk(Bi>l97Nw5*M7Rbo0p>Vc!@mqf99X*jR1fU+%#6=Y`oPpxAJb^VDn|DC&)Hw<^ zeO#!pPXk|#DlfKG5K7RI<5q^dE5X}Z-jsgw1^c)1h2(mm#aQY-orr$i`(y@R=>9U2 zQ*~;@Qr8+^gc;;z#HVl8s5%)N7r;*{bWY9^WV|&O6Y#<%e={B`$aB* z6rhfum+FUTaIvqHs<#cRn`vXY_Uu-d7Zxkd?X2Te<<0*(0g(h|Dx5Ef^>2+{lDA4| zZuljw(FfSZ0#Du6q~>(hf3=;iD!^-dO0C7q;JSx_M2YV-m%ft^e|=qw^H&AS)MQv1 z(r{u8A4I;`%(WP$pe>_A@p#6KgF5g?<$iFU31x9gdt^_oaGP$Vqq@@a^)ybRRh{qM zbB;F0nuRt_78T5#bs01Grgn6%(?u{kJhn0DKWzxvt zYe$dk2q?pKhd4M{j8iTx^|fPIPvT~2#2oP`s#JSFV@<-+Fa@Ea6Z@J}LgjxKMc;2} z`Y00x1@RhZ?XXp^g5CTUs%hgb@Dq`zY|YF z7qToUpgsmG?Fh$a-C3B_I^1g#YKrC!xyHuI&NDrXB@l3=KSzz7^tCB{?Wf(Uk{Dh8 zsfu9%4NTTbyOdvO=Lok0^9dOwYe223%L;@$xk}S_`JA4K@P{}_*`NcQo+G7L?K+ku zB}@5xt%oWJe00y^m(mYB+)>1;7#pGO=G$xLQ#RN%7{8XvS(g>85moN(L#C zZAnr@3!f)@h;~=Jiuq)jUxV|at~71JiR_IC{86#tglSYm0KO*4T;)88gJ8*+{)*RJ zvC(w}oFL5$>D*Cf`f(f+X~&>?x6f7Qk}gGatYxWtU&!pKC2|}&(JWuJWwit3Hz`Y1 zj`zH;{`h1}s0SXxlU$b*OULx*JpT8CU9_3dCj;E|oA2gRe=_FKV~8eWU-6F6sQ1pb$w{?@Q(c3wHAG&M4%nY);9)`{10f9!SKLM~{h zR{qUIe0sd;CD-=Yi9L5or$|xDe!uyL(e@vP+of^+A5=CUwA@V{G7oa@d1p()!YQ`k zumR=gio5(xKFN+;h^NkO(~2}Snt)F$@D;gju*E+?32ONj^Oerzze7vSs7%cPK8#ZD z`6%XCL?6R)hk@#RMNcw(ehQp^6yKNknXP#s&mNkcmiHGJ{-vgSzw3n;1dS5BVo_m% z==?%`3KAvsK$tE>4-(CKN_$8fr5aIhd4D6amw!`Kch7U=QxFL2k0hF9mg9}~&U*e+ zi4`ZDXqIOH*9igg(v$gGrtQqz;yOZoqW*{*#ls2Z$xgSq#`;}3CMDYpFJ9a6tWN7+ z6!;I+*$+$lky~2P1*id|N>Q6i;9Qs)U(@0qB2{5JFmJA^DMkk|mukRh9=a!qN+A{A zaq&+U{j=e_MM!L#L~$dJcQaku9xZg0{cHY^!7 z=UBGEb@i-9=-x?{1!kSy=qn~5W8>7)r1;1zNoBNXUG%pzy)~b3(7{wlQC+Dt#`Jml zZXj4mi{Ip_!*@T6;VKPp(gEM&mnZ(kIjtw0+gN`x0QVQ;Px9^Xh;E<^FkQ}96QzglIs!#E#lKlP^Db?HDttUujAq{Bj zVo7X0sd5qZ^~2krjx0Nla%lNz+u4l~Mn!Ga-CW$W>dUU~j-dK7til@H%c>ZA_Mj*? z?kv`I52o(n&~*xuh|FJkzAuh}w{q4(`~=kVmXz#(?v9&trsV85p(MCQTZjN`AQXGUmJZA{N8bY*ZepZ!{9^?42dG2 zrJ_2;o86gWcLXChUXye+_}g=qm^gD~lt(XuLEbWlTCSQxIFCO={VVjSNxqT%nNAJ< zLMtNBi?YC72 zB!upi#2#H*_}wH;#+}`OSIM<)2fz4Y#h0mTaHEck&RJtMQRvKPN_Q9)k5XLUI7gKl zpjqUGnBtr%HyqvQ?9^y}{DYHT35IjDTes@t4eR;ZRq?MLRTI$V2Rv0Tn=aAduP`84 z)gp9xq$u8t^#?f+TO8%no~Up?i%Sj3C|_yOzW*EVm4rZ9H25~y1S+}ewYpcDE;Gci zCtB}C5<~QLlKUD*$k?l&E@KU-t?cdz8B{l+J3|SRBKhNg<#nx>I4Kdx_s2H*^m?!( zG6ydYjSXVSFO3i%FR#<_lUuG7#eJq#ICJ3mFGhtqI#S>l=Cj2~CR_JY_dR+8LXpYm z6odhaP;K@$e*v~}grDEUG)1#3ekrBslFTE}JuiMU`19p`(7;U11z6k|W|BOf7`c-8 zQxA8#+YQZX)=NSLA_{@SzB7R^0eHW%Q>s?hbN74BZ}ZB%GWPE5bE?FbLPt6jqI%zO zZ`;1({smLe0*a%%hbBtCmAL9-R%gZhq4#dSxW36y& z4^`=8gl?g!*g(@$Y49=*qo64vTn=68WDb%l{(_aQ?|ZNHa;7Bz?iR`v;HK?g%oZ4& zVvJ)?UB0?dJEA)35BOS?k~@i~P>LV=bYI$L0kNSR|@>Qyi;^r$@de)?*WKWFrIhU*#$Q8V&@1Z-@FoKdxmpUlP2xRIvcd! z1A-^CCQcQ0=S0IHxfy@{2PVayk|Yd21{Bfe+$&MqeT}O_<9>|)fgUOguaz2`du2aw zP1`8k_sRl46&Mspd#;G%{$0NhIE;|YZp_Zm@i^*Z@saJ?s8o$v8N|4fL~#}AW+dKR zG2J`43}&gu37gFWs)scZ4aFE+x&xbI_PVINH;CH{QBTSdP}?zp$i1$n79geupO>lF zvp@uQXEs`PH7e4ny zf)gbo1a{s%_mNtn`URj_u!ZA7O|tjFfaudR`#%sB)>N6&>H*^UPL7()*w4oWBqHBE za8YV82Ezr9{3d?F+ys1l@2c2VazCJqy~H8oPQ^)a3%Ed3FMeq4(>#6POgt!GI}qEe zHon3IB%WrtmBq>889iG8qK0bbezC(9XER0Qo*a&wLA%Ikhg!zEGQqoUf)epza7?i3BU zM?%>(*(noRES@|}D`wl;@MLVLtbgMrT^h^oyEh;~-1F)j8nYE6eglmSu^Q9;=zg~o zp>N_(*n5y!tt*=3K=qpBEOZ&Q6*I(B(s9jy9X}l%EfegkkXBb1dgnTSGg&^f-7q=s z&AK=jLGrZNp4WJ8j|{>}2xZ$JQq9JZ7CivmoHvlQT`mPqzNoo1fQfMH zWCV&je~sXrze6IW(_@~5Ao9lZmFPYLxW#=y`;+WxkHp4c;6zPLsO-N zE9VYFz9va`4lU&8Bodp`Bfv(e3q0yaIqv~Eo*5liU3v9x4{sXGQRSzKLYdyd;@E`M znOZnMi;Nmg1*)lxkf2DSuBLL0I`qa8AGg_zgz|^%_dVbZdtIL4{l5iW zZca{d`Oq+j>UNK6Bt%FA&Z$2~%6!oBl6qDCNQPKv*T7PZqx;;p&5rlGt_Sxu61lR9H;w#Eu{iFr%1{@j;E3}g8d4@j!(m!CJLBy^e;_s8#( zD=T=l3Myr%jf>)Y6s<_b5|WouP7bmRpOYN6(;yrmPhViYEAv59WBxF141&xLJ2B!v4(Ujw3ISml0mNDLVz)>p+g7 z7q22=IYZ?t1;gqBv@Ga#xX#daW_(4tNuYq&WmWF$ARg5p=36nd=jBBRTdGNvyKNdF z8e%%5dd1KBQw1}pQ!IREYs~WTH!GaujZmO}UMqN9FsBPJ1UJF%~X`wVrsydlJ4DdRP48{CKqWGuP*UWj_ z0-`mN*+&>od;+jwJ@Z#C=%=^Iui&2ox=lVMEjq+L-cxp|E0K*g5dHMY_?UE?nY$8vE!`slW@!ezc?1O9}-_ENcvc9i0$|ao%Az_!+5*V8#$um z8RG+=0Z;~|1j5Fzduk*yxhNfrE)i?BD?^hHm``ciV03FKUq1%MF4Hmx!h9Odcg2eR z0}__^o-N7mBb_JErpm3%@ZNG77%{&oJ7c)`ZD`J?V;JYYR?=kwGr835lxWsG$yNYz zqP_ib|BFeB!!c*-S4#t)If;G;(jRRhrha*2AY3W2AqySJOiGcw5~syitBPUoqyX>9OCk?Uei#UsLunAk)Ww*Z*-6R(`5c-v4G()C{Si(C5;73_@m{x`hN& zyk^~Q7-?gi6(Y-h%IADL|KcR`Q64a0snhvGCCbdd!Zusy6Prq2GUI#ixPB4G z)zQ53wvqb}l;L7sb}yRNb-S)@FUfq*yAaH~{~yTtcTjc4ASIU+po2&XR-mWVI2 z-18cLKe6UHEpXCELZ?+8OLQAK2Jzsg%N+VmOKn59)3W9vD7mTK&u8U?nfsd0F7KUf zu_SntGKsAmi<9T?9#Srh6D2y!CXv?W6>lgc8FO^D4Z0G{+FuB|y+Vf`Lh%Yhj+KNZ zHetj$f|7%bc@G`+0A%rUun!PeL`Sh-Wmc0R@Sk>$dhta7#~|iETzr2Uj2#EbxE8Y% zCyamw?Z2Pn%VfyJJJI)DQtQao0U z*OqZxGi}$;vg*0=9%BI*5l_zS-O~h%UCTQ$`uE>$u~j>lCu3%WwcH(9mdUiZ_Y-?j zELs*;>=Kr>GFQdo{qEcShtm)g&qBY;OY>2a&uz_5`dY^o8yfn;3f=w{Qz|)3sic&g z6_O9I@6_VeO|LC>thQPMK8q-oLe&SNt1f6tDV3N6c_2A@taiP^)aZ7x*Zz-#LR)WJ zA61Ih7P8N@l$`sLj?{y({ztz?sBsK(5*)8zRvJ!r+M;y;&O;A{6P|H(!L;TT0#Wx= zm&P_PX=l8M5Uv-OG0t_$UJ8k`9kvY&`39M)`D9GI=CL~;Kb1qMtDwDbwss@Rt>jiC zK@)&2K}98pEl+5%%7CD)`~H$N!Yf@iJ8!Pg@z<_22Kr^R6V3aidRDT#SBS*wrQ-sp zk6d?0sg_NBjP=&ZpoqKh23luCtB;w4U)TRU*?#g4wLFlpkB znJ_6;GyfBne?&5>4=q%1qXR9vdt#WF#2p#CP^Ni4CcAc7U`;xp?J=u_c-6x1|>P2&pW9-13tQ*UDQ&z}8?WV>OLpLT!wXVy7K*|XF5aVWjZp~=}R z(p>e%wsFuF#;qZ$C6n$8eJtt8!!yCN^`Fp9YG>h;m8PgXTZXFPvvtjpIJv>0dmcyk zg6KV&%BL!H+iaWrW9l8?F=AO8eUFLGiVoq3V!L zx_$zQ=o#VN6iEb=eDstVFu%n%Tb+r1f*U(NABDvXJKNL+9ko#~t}akp{WEP@^Pu*c zUVfT8ns~*fx}&!1Z+wf1V;;9goza@sSs#=8xx6CW4LGCSeC?qg59(y# zcrupkPeu$;*!-EOy2;>R??uI3-71mn=ZDIO_cfmK=&45SBdJ#rF)7So*Sji(+W2hhqy>C35qc9I^zog}G+&nXHGZZNTRVRPP5I$#K9JYPrsMAwIrE>%@&KVs%r} z@do-7otz*uW|Xvt{wjdGX^?wQ@N*0M*2b{DbgfScEI)3riL78%UB$qf7keQ>^%{>n znF2j_9t5ymn{SUek`?uymhs8|Yx^pj8h(Gdiw7#x#!o#uDPb}lKqQGK`xob^K?|iZ zhy9fZY1&wLoXCyWJr74}@G*sJO5M z^XA=O*x<&?{ToB}a1{Kb4?%8-^&KyqfdY37sJbr?c>4PnIy^f7?(+~(N@D*Wz;(HY z+|H=MBgQ@c*M;ZE9Gp>vLi8TQo`z@f=&OF%a~=Oa9t3cOc?rfgC!7p74C#f?6EFt+FB{BGpjOdc@@PALxLQEB(q4mabnAfKeIztoD~B>kb7_HWn5{I6X1iG zRTC=|fzLEz<9|wj^N&QG)qfWz$s7!HvLntn7?L`G%e@wWiPh-Xm_~sOO#i6((t1}z zF}7aR8j!jJ9zCq~8>4tA4*`KUK0~T+Fhw~A)>8Y<%&WxjkJ|wyRjd2PgSR7zH4QLb zMV+h`Xf_WGps#AiRsSbItwyHAlIpSNZo@q2&Z>6^MDDggz;l5?cP>OC6r^6{e%>_v zDV|u)LaQd-3N$8*`ddaKJ5V2T6S*>Y3iojgB34|(*??t9LG{E8KmxRp_HTKyiOlBf zs$o;l+;;o=3ISsdKsK9$gKT+8xXpeVQR#n{^BGv~^bCARS=OK%$>@twP!4SGy=#Py z&pmhGCK}>oSk6T}A8K#w>T($o5eA~Q2{CSs%vo1lpHTUDX&d?`wlkDOGzOp9wYq)Fs&4d$@a!~G5Iz+N_278#WExJTi&5*OApD6o8VU*rH6}bggWQ*r?1ZyVw{Hl zmU{ZPLYrF2tXDf`>pc^eso$vh)5Wg*z~AS)V<$wg_fLs_YygDDrt9A*1sMh;Qa@gQ zr1)YanBVlAS|0=vvaH0$Y*?%-}a`mXMqA}$4SXp44^9i!~oSN9Jc-Bqg^?vc9QnI4eUb?e+@g92fC@wfzNlQs#!&6s(t(QUocQ` zjS3Z3xL!k>PbQnyfR6mmnS8j)gzacu8u6&Ih}${HV=Wn66i;3?~1Fqq7cc^6la{ z2nf<3laP{-R_SI4(#;r+(nxnphjfQ@Z@?HZ8VQja-CZInARsOIdmrC_xGvZ(wx{m< zob&yh?@44BYurwm{itmBf%N>DuCd$`LO)B(85DYlZX53=CZZ(5>WmQ8fDUP}p}6LG zwT3I@PriyF(#Ip7-ZX$H*RXXeD@%D3AeOz=NF=YceE)1oi|T=|BS`r-kAyh2Ibd^f zAps!gA0EaU6ETuB;C5^IOzCW^U&hk*xjKj+_4nv>4wveGI3Dl|z|o-mOz1{k^Gayz zf*I-EEt3A|#T^c(z7a3EYD}*OkGes4V_j{=GXih7&Qxrnxjh=+xQ;sB?3t9Gq@Tu^ zwHDmem_-=5c{8tm;qV@gSojrZN}gn&R_er+9ZRZ}38fd+Ci*j1FIaGXTQh?;g`^nq ztfamhKTR6z>%d`ce;&Dqc}a~l!|OAl?sjrJk~jY`#>ak2k^kiex*AM~DN_h~Ke4Yf zN#VL#rb@AqPoe1ozK7_2ZR>)K6zDJZ3 ze5A_9VP_LHz|M|o1v*!zdh>OZHS_OJ4DxAW{8yE#>CgjWsH#y7EugPAmR#JUI<@;! ztH*qv+I-4myv(0Jv_w|UdR&aplk>P37>kI0qgAH?FEqG?%T>mHzVe1Rfj-wqXoc4^r#T|)Rs#^sR7XZ1!J33p)z zUP;7jNMrUQzhpO4NSCBNm+H(rZ*ZnbP_5Pw5j>8dOjr7aB_DiioUdTn9VFI(b}I1doqnJdsHtgZJX*<;#YRQ`U7rJghNCYx~}jN z>$DkorTmK&_XpTQ@~g!(w#} zSeb32Qe9pxBI{!OUhipP2mEk#p`%12jicStu9&&gRS_F;B^X~3xp0~Z?b>7(EMtFpL9ugK zL~+w=AzsG$O*x{GT_WXYT_>znQbC{K<^0AC-1!_t!^kG3r8myfWS@3x;e^R1;RPzQKn#-ni1oSHY09AP^M@%{Iz!5v zBY7juGA;y*@UKf9Rbz&0=NNN?C&I6onvutjKF@<|#45A>N}6(F4f#fNU4x}G)l_+A z{V~;4qKu9@1JjvCsYM8hT$`Ewk)OGkuI@Qa+RzY1IP+0VvPLzI71QSP3}mJQzSOs| z7!UA%M$Gr4kFNB*wYRz*1na4nR@DDtga~RkO*L&Y&YM&&Mb{+%5dn0WX|sBQ2|Wa% zQ9L^uwLMefhs+YX`G{dAeTQS)W7XXFk+p@c(ld&`h|2*EC6 z1-cd8MVmb2f%#~VTAa76OXb%9K%2>$$o2hIeD8s%;JGq^4VKhGc?cU@Gq7aW zsM_L5xLyOYVT2J>7iGot2%ReQtZ zWfC_WFtwD}GmpV}551-viLag+lTX|HjTS-T@sucbX=0pX=sNblM8zg&KeNq)TAY%4 z3hWidyR_JTs1@n6RknyHR_daw3{`Z({J3e#KXY~WD{gt#ff!$mCA#c zVoy`RC;x-ZO9a7Qg^1*A{mns7ct5v5qASq4dPV$BJ7ZO%iJdaP?CYFM9luH~ueBua zRAV)BlU$EeOKKhQ`FEo8+PZJ)AHVE>^C9V)n<$@J8S^;0eCoGo-1RlCtDCeH=q zNm6*%r*}WfI+7J!j@f3485Z&lznY~}?9=a<`48h0Z^tqVY0IvHKS>u=PVUuYgci{tpLnrd3mDZ zP7{$J)e7b014yrmqL3}cqoba$7xDZd*E*0NoTE%~%;gNPm?3xjvFi91?9V3>H1@Ph zIK+P&xhCXfrZBrQ_t=FLjS&nMOf#idEe&SnZ2*jf4cql>ES7kKQ|xsrlz@dhiJ+_)Ckf(7Ia)Vord63A&pOg0yVz3$@tZMM;Z{NGyc4` zHYUAfiw$Ap8F?8rw_DJ7QRRjvF(sfHo&YuuWZ;CLa`>QG`Lk#8T3uiyTfQGg$?2#( zYupvqDO|p6d3%t)eahBTKxZTR@s8vmg5U`=PXQW8IyBzt&~Acvm-3MrROstFdukO_ zWoabUFh=xGQVFORi0dgC31;4!Jyo(B#ZQ>qDH<` ztH|iHfaXzb$a0Yq+f|O9Z8E`H1BpdIN~HDPaBO6`(m|2u7rv`|yEY7BJI_R{!h6mq z%O6EWC^i?dn1^}02Je7Cm8oCH{qEb%tQyH8!DYr}@mQFYfH5>y~Fr%VrXIL*P$J+$1TW-d6k_@=_!g_3)C7Vtx9hZSk|@-e0>Y!2_#0;uycRH=s;3!qTjd3)j0-&j zrb=|icd7Jro^FH%I1?Z-=F$-22FzkfF{|mnk!5iK_nqs&CWHKVWMRzc6kuF%pkKF! z5E?yFaQSxcsxAHJ8Ge*310?mHV~p5vK(jm}YE7;=hjm96Z7QeS4HcIBB!{UJrvgX)fR9N$XWwTw?M_lP3&SdGLIWHPmC{=eHVZm6+j$FKY z5KyawisqTCW~1BNS(_>V-?Q>?+AZVK;t2l~{RpfA_}>|R@PJARZmXb)Y(v(XH+>kO zXVvJLPhO?+j6?)KJnQGGIPG!IK21)4s$Uu}uv|d*Ri~ua>G85XG+u++BE{BO`8$W1 zLylfZKJG7jp&M&-4qHZ*!ca-0GnG7Se}^x(s9tdNch?i=mmfDJAM+Lz_Gxv@#r6h{ zx^BZ;x8|^~iFN8kbFoz zI1-Q#1v|6xeN(>{-;nbKDJw4O$}l_e_XEWMbuEALRHIBDZdlN!W>~dR$j#Ksdu~@> zS4VEYPH?Ywl+#PK==R-%#BDvEPf}BxBObbU2N|9(D1Ga#o}EX>(v2ipq8oT0Uh&`U z`#8Q9q{NPuNCxh zW4dP~w-fi&$l{2qxXa@j0Ar4dhMF6WTC&=T_H8Zm2>-cTqV0wE2d43 zizW22zF8B&{=(n`)Xp}|N@noIIW~J$Y@Q#1+J4IGtch~0#Qr*~6S|q;;C;$T`H%XB z=?sF=o*DCEN}jUTc#2vig6pLB=&yVTsKj8|4W*Ob@_ffB=X?F3sYi6~y@XQb1g zG{zc;YrlI&GPhK#yPHe!J?8kSz%5iPU!`%%d)|( zK3Y+JU{UjEL@>tNF=`}e6|g`+#wUe4rEMoVseOE*Db!!EdsbnQnEM;+pNy{ysa`@g zu$R6uC4P+CQbE=M7E7g;v=habFKr9PabMfHtle;N4-UfD9FbQs^}`=?R9|B>I^KXy zHBt*cp$iu`f6u>fmnX`8rk~S)*=qCYlw|eY!owLY1eA*+)MKvZ^i}&FxB-^-1()fj zvzorMV`hVi&Ge4B=N(e;Z4R9d@^LZ9+YQ?5QhT<tcp*iHadlZSps24eIh$|AzujDz4ix&@QZp~LW^35ES12zdi1T+egSTs}Ny3LTPT;HQ*a$aZI9 z2~ikB*X~l(yfFKmvwd!8r?Guk%Q$!(Go-F9TBgnC6(az#)Oz%T{i>Yx>nr@;+#sJB z{1o0*{v)ZpO4D4AYM+Tv?~JOW(A4;*clcipN0l-Alq-~Vn|1^>QP0@dAyKets%IO; zqSpN!zq<&t^{E+-rWcKkQhHk|vN8k*%eT{HpZ&JH(R*6sld`Tpfey!7%cCK$;bO9b z#iGKQk0HjRlV4{k{f$k7JUptMj9Ovl;~e92HMaAj9%pP`a{HyO?;GC4 z#I$=$Tt#0djGNEqha+BRc{29J`}0|U+ z>_>PH)nU(&{z84Uz|aFFn(U_1f+g2JiQYnqRNdVZtBe9-p64T}RGWp>k2bvXtyAP# zqxbFJxVf4ja2}^*%BFAZPK9Uxcmqv-9`qImO&sKTCH7 zmyGO4(KJ;S@J6%f3iemZ_OgZ)n{1=P<-dwj=0D=*e?cJI`yvO~kP38NEw1-m+9F9c z%z|4pIqJX9%~Kb)YN`xvrSwM?91BO8x2hle&s$u2tagFKq%q(ZgY1h>FFaNkb1dN3 zcoeUd7ovq?$R%`w&08vU-2L)+Lu0VIDgg$=*1aCXW3ZysQ){@q5~tLg%@=K_WunS8 zr!_GFWpR&@830OBvjPf*XV9HLX|C@^9=jgC@8$I%uM=6t&umkms%ruak()dKbJ=jcuB)Pku%|BJ z+I4x75))fsX)fC_e%fm<(4$8o!rE3q(Z40P1t2}f%5M9Y_{T+n#WK8f6RRI zO#Au$Oc=q;Jdq+)L6uHOjrC%zKqOqzb@j6 zA+8b|l)Ad6+J*cK>L`i9LA%OSr_^1meIP7V%j^7UjQVdMV{QIo!c?8DcFC5?L{YX1 zL4c~=Qm!+iEN!QwJgZDG=G|Ai+eeH7SX4+EFzxK?3pwU&Pa|$6DZk_!9irDTyT?hJ z-ww2>ahvyy`myAOZh!)*9J!OCKKfimg(IUUr|kh# z`!na{R-jc`3!>Q18^Kaws>a!#dX^XEM|{4U@?24ct`IXdS68j+LO_YNNQ*Jw*IVth zU+gtC*sy8aG%=3Elry5Ug6AZNM_Ha2%k)g|XM(`15ksFNCzmzIRgBSx~YcdO) zwRfX7#(zUM9_cJ{Ht|UeJ|?oI5oEc&8%S4_d(OK3tUx$~)D zCR=nB%_2rpY$RXh<$0)MINI?~knoY^rPGd}b`CNr~;4 zt9m14)9;V9jeY&@qP7#JuxO~YiO@UdZij!T1eVl_Yj58vc^hMz+bTAharpT!+5a%b7Z(s!0Mt)l%%JM63)XRMn^qqJ9XbywWv zoD}fiwP`*2=ar=o8+>q$Ik>!G@B5gbr+hfrF%orNo~*3I>u~WTM9=Z?5D8>r-TeE$ zg0J`wZf2anAO77)%`Gdj(27}n8ZGbt&zhrjfH^KOU01Mc48Z156|;%zL9}y@Ua#)^ zUL>l{Sjn9JsQH&~oZ)^76e-UOkq87b6;cz)NME>w1;c0H*j7GmZJxkM@do#f0llW2 z)$%(Q?kPfvkJX=;((BjaX>;;oRG*O19_wu_nhIbm{YrOa0-^sGq~1kr1?40gR=Tc! zDHY5crll=7+vga@J{9I^2r7^@bXh)JCj>00`m16+2i8-Mm7ysMliIEC$2D-N*LG*4 zk#dbXA0U4r$Qk0tU_vPRg(griYO#Z++E|I%GKI+~0F@;C6m~cFP}{QM&Sy1mpUn4Z-Bj}v=!OGe z)}J0i?GPxK*0D#!Lxx>#d^UY|3n^xYOF+FnaTx>kE2maK)VmkRwAVmS9kaZDEh0i< zj`eWDA#~9hSQLshfCV{N{b_@+N)^=W1Fu=Lt4-Uh=#0Ub+(k#ng{hzU| z7FQ4TWqBmT4AgE?O+4$kwQyEbSF(3{yLl&{PXM|~-ISxm`8lw{=CTZ&gP;po$pDS9 zLID7*+nSyIM_$~j`|W2pa}HOJWP^v#v7tse{i&h*k|CgcR?kn>KL5nfgX(nRs@)1B zR57=u_lgq6eNaErSBC0A=Xv5MAPGjMn5>A{crA$kKkyEYfM8hs06@l6+r`>&h@L#_ z$aD|cQV&l*2~bqQJ6}fRJ{x3f0`~&GOTSJZCpk&h=LR;y?P=tLO@XFcuY$RN+<@5% zsXnDO)ot)knpU!-G5!TWTmLY9C*fFoHng5UMC8HWIRi+Qg;;{Ls=sp*Zu<4V)+MZ$ zDDzE{O4fa8Dt-P(giy}?CVsoGYU4GBOzI@&_^#2Fw{n!ivHoU)#(A|H!53up@UX+n ztnneLxYs+JfA-p%s&)21$Q7Y15E*f=7cHf6&2k^o4sy1wSWgiTlfWgMCR$%mjZ5DB ztBF6(|M*)wygyvO! zbNO*LYc;`z$OWDecCxqQx8R4b&@w#4yC`)d)HePFc2EIQ&szKM!#_SAVlsA8J)Mpl z*9<(h^{g$YWLYv6GaD#a?&rhCnMPhY85IlQn&J;-GrT$C!&ct{EJbG zNIA^6Ota4D_Z6D$477-^ov`CpS%z|Tb~t=YQQ`a^NHab3n)UiJFi@|r7ItrDp807k z8F_fsnG!2G*jyks9;uBN1Y>LRCjBYil)fz4G66w1v+)7&cIBz2 zfv(>gFB_{b+|9?ayQs*VP6kkTAhEjbFEPT)eMahJai>l5r8`?MwT_|VC8YI) zbhU11_UL)5D1p9V#7?8%XyzMqoUiN2&Bj+R6z7u>McgJ%)(BxUJ>Lsj{PQ=bB}8V= zlSPybm2%%qIZ=Oo^Uv0HqT$a>9jOTkps?;&cQ`_+s4-UjF|+ffnWPP-Sg-o+VK6Cw zCg#tfvspsfEV{v7{^|15y1X@=VCK?hBCeiVR$hxKAapFqHq%=QSXE)r3H+}XvWKaka-BD)ApT^N8SO6b5U*6@;Hk8Cn z0<+_w%aw55_eYudT!=lEmB#BC@tb5Z$HQd)S}u3E!Q`VmqV zzN!-0>XVFXHTcl5Rh&^ZsqAd@w7>SvDmB{=is~M0ncM_}L@E&HcYH#e=SD;g<~@zR4r5ZQulVND)4U*1uWH2inL=_^NU;&zLV2(l}YCqM?z} z*1^#f5B_BpZCv1&EyBA7Co4bHANtrV!{=?K(@1pHC1_@@b6ZGKrZg4v`Q>l~yR0Rn zvWwQ+E^64jqu)wY?`Op_J@TS?)tE|41*)|$*LC=mnyPAFRt^)qk%0EEDp`AuuH(yT zsu^sgv9me(<4{PMvnEnwKvO4oG^_U9QLrE0}Vy3AI|+X z*!_~{&#NPsmE;_ce($a|hI7!G;!xdaXK-b{6O_5B4^QpkR|WgJ6$Mb}O8$tnHo%Lr z^vgca3ZTT=gU+ouugC{BA|)IZ_iZG&S$x0x=w6Snf{JW^D#Z7w3j5=qgW_G|Fb_3U z9L}lW-2EjKE^kD0z(njXa<(l`2Z)P`z3N^PJFg3RP4F;%DFtduTxSc$t>%C?Me zpnWDIW_o8W!3MZ?_R=j;HjkX$Pdr9(c%L9kT zGc+CD0lO^z+4axL3CItPA#iBh57}b!P+gRrKULZD;?s?{_ecB&l&>Fg#9QEeKEX2V zewfnwjFSlCpyyJOH{{9RsU~;*Tdi6Ykf`w&2w0hf01n16=-hr%R{})q>LcIBB>Hy8 z6c|_wq?j5M$M2m-A8Z;G{<@ya=WXo>6XZVPH#inlsT^$jsc?&Roju}PzoOevvBr^` zZgVW4ewx1psFKPg*PthXuwq-@e2oJW-jJ~^Wq?Ew5lKpz=L6NW9H!<^aRz@jr(K75k`eyWv8UviGn0u376 zYh2&S$$!9&vjNJ3G-mG0V4(gy>{kE)SWyTq);jZDIzaS{Yq`AVXrWI3z@R)yfo-AHwl_&vzZ zO60tntxnUCLLld^H!@v)|vgp z`b3!_7BAUtg7$NsIpkZfqaQA0D+!_X-g1wqcT#tmliaWIYhISUkqO0{HU0z-0;em~ zx${7`1*$LF6tCV@O$|>lVa?V5Zpg^rS}_sg0}S@jn=y$h(dNlf(x<$Gp9t~4xmNW( zxfQI;iA_AJU?i#d#hGgYcFX*6?|`oycp-DxH`|nhv9drXLYqC-858ghC%V1dsO;oWqrTALlW3O$kg3tA+-p%lg{{2vnYY@U`+Qr^VF!{WFHFBa2@4N6eV|LALsX zimc*g42nW_#t;OmBHPBJ`Tb zWwT~lu%JHkwP3^Ry!Z10W@k3B9>2_S|GwSH*}#{KnT_ zFhJ+Vx4i1P(VcDVeI|53lvfMbm6<`kcxfM6;gy&Rf$60`#2WI$Wmji;6jWRVK65Wz z(gyKD7mg%y3%X_!7WQs*SW*ujg7BHz@YOm7wS10mrlWzuaQ(&7L1UciK)awUHud;F z!f;;ARKxh8p(sRBVN=b0?;*&LbRc13RbfxV1bGZ>c^5NgEEX993=j?N z4_wS_7{KFvQ~rcJq5al8XD5jdXo=ewO3DWW5O9RDT9_*?C&{M(H9Z^S%R5(faizl> z9m}U(;|GR)WPv>f!ydp4(5@N%1Q6Egy9ZDlY_T{ZoHS(q|9VsZWU$D;pbCQi|3N7- ziw^60w-%ek*0R%oL-IN27HD+m;k%Au+;JRF6HK8xN*(QsMVRs3$LuC0pjHREWcrE9 zSyxtaAZzUWpz#B(;@ox1Ku5VTf8a@He1Oher-vHf! zrR{|e?So*Rs@b+Z$lIUljGh+Bwn8jmKH8e465;r zaxH(yCX^+W{C4lz_kLBfw>Q3*eitYb7}8W-W#Zzz=`7gN4O(QC!)Xc)P^_*JWS`da z&s~T~a2hrTNn?yrcZk59J_p-B)dCY6Qz(gmQ4&^T{q`;V?RvS!=aqaM*_l~jd z`d5<@q7VSqJE+#3693F7kEaFvGLO_h%tL*HBZ4maT;}1FTuq)+xk$^{v-jU=j(nfo z0h*RlO@`U&E8H4}sT|36vrY$E;=(wf(x=ZA9?ASdQiFwLR|`*m3B(~)Hc+gGfpYxz zhw~H7r+gouqSd2y)4+7e(w{EVc$--Bw7R1mE8e`I(05B>D1d(LE;CTDw5RGrB>7rX zb=v+}?M?v8YT~3=oXQHc94%r5n%DF=sb*oQ+#zJmjD72XhCZg2T@&d>U#guSPSlyN z?`Qk;6}K4qId0qyV`;Wpah)qb%4)h4xoD#L0R>3cgJ(<)XSSR%%@JCFd~7%zKWwfl zikewGCuG5evt-p|g{;As3vj|vzFSc)e!;=$InFQ@HC^#C)fTG(I^fLO8)s0Yh&~Yp zPt`F%P%`h7r@-Z14-L&iPN&zFM^`}J?YGCvPP;n;aEkLa+GKDLLUcM)R>*&ndDot# zcV=!oo9;EM`uH3$N=5{xSMTU}F(ZmSvX+!|q!222(~CJPs+n}+{E5{lpPfY99}a4m>1*{)!(Y4W z+G?kjb{r<~NC-OH5fg~hubc*nWsr~<53QqOSa83wUF*rQj>Z=bE1G=1>gS@r?R&N# z!Lu%n4((0+Wj_`1N$x`gZMDE=s~-=)dOg<27jf^whxCtjmqJ5A@(Y&`<6`@`$In(v zE$-f>J_sUUM{O==%OblG%po_rp0(90vrZ2d@oE9no`M9i-PojcVgDx$KK}ON^?meM zA_Z5FreVVr1)D`vTs9@;j=DT~IP^^rni6|{_8~xp(;V`h+aKQ~MRZIiaSkE(^g6h< zJ0Y)LBjd!JRa*3B?p3VQgg1{GI4W~E0T`Y<4n%c`{bV^ZzM-~pDlC!8sgdav+PrV8 zhj$ex#ler5*zdSt`5;!m?2PIkZN*H&DJx{#FI*RD|8dc1+K@nNQOPu&{VKeb6@Q(| zgFE03McA-{EW>}|v{B=amEb2AUP!8O6+d3(iZol$R^AT0>!FYBQ@Bif`yJ~SpB=fg zjzUA|iaVm=h1UnA!;W^Rx|qvovz?=sEVzMcNY*Mba}%71LS6A_O}g#8`oHattm z^v$ST`ZFEH-iu&`ev4cmeZ0vnplur*914Mj}(Um%>`d1+nY;F?Ci841+UO z>9Qp2?Da-j5i|%ILphVoaU+*dV)ceH83lG(wu#`%a25{C6YJ~lPCac{hF z?YEU!NZa%ohuIYdTjI$U=rTV7#?8G)UTeFSw3!N%OTd*V#b{C!k0nRTWW@(TR5 zB&Pm81DBO5bs>9ph|8VcKr#;9>4lSnMc*jns9Ny#TfD=#1$pQ@)8-OE_odLfVHBA> zxXpyZm*tYLVw5aZ!l7-&Nh&?n>vl}$kbc(XyF{)zV%U-Ce$%93O2a+huSBGLlnSj} ze`zCGE|hpbeXV;pu%W^S4>;T;D5(nK-;>+`QL+ zu&#^y{Nn;iLMHS&$tICKh+iMcHho?r6Yi`=@9BNpuMFU!W5qV-_rRJ!?Yhq_-wC;z3!v8R^RBc3nH%z|E7$ji$6e#!{Tmv)=?UyfJ9~N}y zqypJMq}_~S%VAG5kZv86atECJow2&t*a_-|DD7edN(U)7lG5uxjZ2(LLLC4MUON#f_kBDDO>Mvjkd^p zN#q794v+tAUp)$n;LY{V9$4Lr`5@ZWuTLVaA^0lhqa8wZK`bkcBEvYz=Y`Kb#|Oin z650%{xZ2IrK6BsEBw05xkBu+-sB!Io!EOb&gkSn$sr`6>tvjPN_mh1WehslI-sNjj z=6Z$=^oL#joT!UCcQqFJFZY}eHFRGImWJIuYowb$G#M+xMaf{e=ZG%W`|NfA5#0x2 z=t|8FE3jp_F%t$TN(466IE&X2dbDCcDiZ)>~cu^9#`aZn^G{ zzl$Jl6gXX3BX3G(Y>L$p%R$?_7=wA3i7_szt78Z48B^4V`Zn3OGvfk5_)4_AR6o^9 z|HGgL^WN=~kEpAl#t{HqD#)@gSruLM)qbh0wHE87u{FvoUNA(f`WB+S~?UUIul}DTUL!5(Z_pQXTU}nsTgM zz~DZI+0^3B?z>4zN|0fN==+6x4h3CXR!eE+dQ(71jsYn=D&z?ZNxhzI;f!DT*TYmV zub7w1J_s%VVr)cvVJ=}{+FZJZYG68bn*T$?CaFV~rKuaAR2KX?_-gkD5LN3KdI0WR z=nIPRhG!y>>!6GH5A_ZS?usBZc3|3PDfpjJ-MD&UlH8Q3t0 z?}8$9o9g{P6EE!kD?(H!j13RRVGx5EPRev$VO7nXxvaw8wMkvE6Mzq|UC*OuFe8@| zHlQ_lzWeIlK4Ls8FGfg?Qqw5c=1;|>B7H#~hjDIjkRgRth)AWgv{vn(7zBvC10OK^QxuRBtD>tdtR(Yzj}_kxqj;7C7?5vN0>YEc(5Ql-$O@pm@wNGIJ9vG4l;tFN|^+sl+3CJa(ODx-5q z5KBGMz%sod(FPvmjE?3w)1<@0j#IyP(#-SPSsnu8h*5jEVv2-6tgf5K-? zQ?<_}y1;D!toXDnz2&)M&*&0w!CX4|#zmZ~HPz>z+CA_p)x{R9*(RVFkMlE^)6jG= zf3@0ta;GhGh=$?}Mv#^P!tK)vwLb9r6JVh$d$r4DPvHA@{pZV?50u zf51(7qNl8%!budYjJmXfE>$XY_59(k za8Wby+0;)s#iLWvp13$|@ht`YM{^id#(#M(T;BC08#X z&l0w_vd0Sy=IU>AK28Q0>xmU<0|!@kog>W9=;c5pB$8kur?uaks=~Tmmc3zLK669P zZ`{$Fy*iYUe^)5U94-COT6y`Np~dFR%GzArZ7UTp3JBZLLqd+&@_t;oiizIIQ9f~? z+Kt<&dwqPgt%mA5qpvzl+|v!JCnt;0YMa@-m~}x3 zUhSH8qFLO7XDcCZcUH<d<7GkAR#@}BH?2pmUKShq zphw22vRQzhchZ5?#WStDQ(DknOJz*mw5$kxwDP5XA#Da4P|@b$6%ocRBZ07zXWuFK z-x&W36ZVB#bR1V*e|@{R6JE>|YmYVWZH~^1iT}2zb)0kxY$2(}!^{u+Or5(7o3cau zg%Vj(!MRXQ8*YTMhW4bK73&I#!%pEDv%Zpkv*`L$Xx!oQFhjaBvv063n`ucg5bYcQ z;{aXDhvaTmIPSi^se9#zRd9m%n1K0og(>l<%u3;@Qg{E~3wK~mMRu1mK;ZhOs(+Z$}6l2nq!=Ov&ZF5c0x zCBB^ZfwxyaHV-EtUd<7HMv|F~f3*>nMW*x+48B|!1qP(jWfrTGoBKtL%7Sgt2>fVa z?eU_IFEppM(jB}+TS@?Q{Q8_gOCpu195pg$HnV3#892O^17o;w@us~y7PZPozt(Ym zhl>%d*~|k)3-?d$gw3scy8YMvROyMgg`V($Y;t*N46iDy==zzaYNUp&B(t@{p~xl3 zY@6&f`pgeCsv&;F#10e!y&IVp8L1rrEWz&6{waU_6QLQ^-4!%XiU!iSsR>gSw1Y%r zyN1e3!Agn0jQvbRit0=a?3I0vV-WZ<0sqA*O8l6)N04u0Tj)@CurlXM?h$$dxT1UV zJp0{EtZc({YVQbvg-dFtG??F@e1U<#QSt~wtrGlHq28OO@0C%V!7CZm_wjA$duy*T zjcMRaSt}jiOqGj{aX~2(IMd-;3#zGe@aAUnQ+13A<`62&twuvRycSv=WiMU%^EacN zw!fMB%ik1da{R?-n-g}b z1j5REowgm|F0Y@VQULe%#3H_5c?oh=MUGU%Ms4FT%+xhx|in=U}Q43)(G zji>OaARNh=KOXN-MWIZsYsBB8z>yq6pee zdzw91aYj#$bA{3P3#Bs_@LRP=gVizf7yeK6vZ)*=BoZ5fyed^*FFhv{nx+4XglJgG z>aO@LjSFPkU_T$?~#!b^PdsKM~ zCCK{Jp2!;rt64F6GEbkUQ=hs$$S~5F#t02w@PJf>?ZZ%hp5KG+p6agdVh&j@{vF*jrF=7=m zV726o{T&9gN6zxJH{j^d&hg1qZMp^9SA$kc-19wrP2!@@f7UZv&N(#g2=AG6_c_I8 zGMAazz3}E|@p<3hSZ2~}Av&c3ky^5f`qUNUuO4?>e>3}p3dM$$r$}AaHSUyn-Q5{=>|J z&7B-SqX%GZ(El|}^_57=xYmuf%|u|qx@}p4Si#&E{in<42u0apvYCRBX@W<9Rbcq$ z|HUzZVDUdiiM#i5;uvjFv^EzHk;Sipgk|@H;xg8m8weSF!mYgD?3QcWrAK>PN{bVg z572c1ls-2^*{hg8025(nu}1zR>+L>ZR(ny|)!;+N%TKDVzJ~kD&P0U>(wZ@$n6=Wv zT)rugoUVI48+A?%s-6`7VwRyA`71g9o3{O{OZLO00|SXq;RmRO^fVN=iX zJa+Jpmjh(=ls&Bkgy7_LTcq&)4Sw-JQR3c&!E)pN0Gjyz^Dk+bK-}(xDgJci+}esc zyH-HgPAfb%bQ$9Fv-E0Whvz&N5tbLqyZ;xdC_v!Rqdv-&*cF&wd5YOJ7DVIp`-1tr z+SlAp#gV8)p6rZVq$EO!DO_to2T^;!PSQ;AJ?;chUD7GzavvJUUw(LPMob9{GK`LW z`ViDz6_B;yEUx2+0$B?*He=?s?@3Y?mbd&-5T)`Zq|-*dbg&? zMqpA-=-4n?xG*RiPx)t3c|PZ?;U2+l20h-wO|Y%!#rh{_{Aw#mz23az(nm}CXTRC* z2r{hiqRoz^jHkds@2R%7;lUZ^P_~1M&tWM~b_u$+<~-^%Ean*!|3ZDMvz6ZP0YREn z8fShf;RDd2u^%r5tq6w)L<@7A-*XBl?FK9Tf3}D5`UUuHwtc5O#`v~wGEuSs61Y!l z{^}6B`5x>6N{Au+G+LA;06YlCTR6u(7zb=V8q}C*91yi387w3Y|6xSK6~>)`KUZ^Z zEMZ-B0oaC!{$~slIDpJej^&38nksN+VX$9pRGPd`fk)a-S?!s#j{luDRP`tMH0$-)dr8cD&#-JI z!XrzuSXZDn56q$h!6_&y$s=ziBHWXAh0m#~Z6~$Tw}v5Zx^i3#% zKd~h_46%{CFEJsvXN;M zIuJuE7r#s}Zk`xH>Qnq4cleCKebk}xu!`(D;UotAZGH!{y_KA5T*lYpuX$^Qa{KmC zO?$;D(`@GFji?l(9X4*?L4>ead}PKu)iuk%*XD0ifzYw)E#kc~CUG7$c4EG-SOHx5 z6OB>EtLeSQ7 zEHug% zt!%z{N-wVS^y%+-mn*Z9^1un^PZKX^a%m|RHQ(|&d&hUVT$&~{e|R3|H6#XI9=Ls5 zJbt?CwQg0%-2tp*ti5HFBPJbME=PQ4E8Gy@cFT16R#YjuF;KL8(*^M0sHXcQ8U*rR z6W@S25928tdK+YJd8G{vHGKMq=Y~8#B^QisBdw|t8DAUV7%}zcC#(1`#nrmXe)#Wj z9(J<*ki`6~N`6+l+c8 z*j)FaH))WSr@30u+^j$RDH7%)fDiV=A`?O)IN&g#NeiNRq zy#uzPW%~*gIhBR|9|-*_i^S!(_g(-MY(~rIeS2ITGtFg=yk1ybqz-1n(N`;arv=`6 zZwFaK=C0{&6*0OMB&1AOhiqt9L!bnNxXlTeHl}B~iP%XvMM5nj1SO0+I;0bOHI9Sx z4`)5kjW5|(_ly;T%adcS8{oY;rK#TEc%~*81LvB@2j2KIWdB1atPU1xVv`S@n1NN` z!~DT|gCPDm*J;Y5r+5ZNOJ2AdZ={|b=4*>G>OZgEIpP}GYRkXkX&)*T-vFAVD@OJ zKWI9HIXvb?JwhfOo+U#fvEzdc^VAN%R09SMCoY!xLtts0>*t?r zV)&^H0;?Uo2u_4oX?wkn>%$C(zQ?a`Ii+TT!akQ*p3o+e?_Y+bqCy)p{-juF=PLv6 z9w1<+ZG0@}`Z6HQ98`o;1a6=?Gn1#x1H5Y_4sff>;&`<;|9Y##T`kqH#x$D0tdP;w z{wi9%sY^YCc50f3e%egZaU((n{!7ahoEKA%(OZi@@e()gmE)rucigN)nE(=c`T4zF zq-;1b!MW_l9 zUU6tQQaU9C-fkn+EtSG&grT@q*Cj1>&l^D!R%@VyH)UTENt*6PcC*BoI;T8K@l_De zXXk2j64D1Q@ZN%#mG1Ung(k-6y$(;bzUT3Sk_U}ny;4s|*o@)C0kmIfUGPz%zewUz znSbNX+VUo%rRf)!Q1o4o$hXDLAVJLq2d7}QfR!b(gJP4h!T}?JjBlQ5D{0+&=GIBe zkYIYPGXp{FfGSx!|4eWip_n|O`?STu z>be1r*Hs)#DbFQXIt6xq<52`$t=A*E{LZ_8O^=kcS`akX7!w~Tzc4bLn>>zfWf2(m zjgo<*sR)eBaxH1Lie$@w7+sXG8^{_>D6h9Ym6M{X1;axk%Dk1%&afG;Y067lCJS41 zGMugI6m)~7-K8!{{H80$L^SG(y~RQ#Pqp?f<8R=H#T%BJ zE-CIj;RW9Kz6Txbc4VoGn2n690HBET6IOQPo;|d4%o@Xb3CUNrrFm5B1(*(OIUcBv zSTo7h&p{o6w!dgC*7=x9V>69HUSLefD3vHY5@C*84pL?AYuh!!QP~~>3lP$RmB;2qyhI3(GSs|hten5Z}?;Kv+ zQi3$yJrUNy4`n(89Z@Wa8vdPC7%h8@j%ub24y+D1zdCUIHoP(a$5`FBo}HN5di+ay zMMH+k;sl+yO9nD#8HM#7Beq-w`|G$pe8_GU9^6)?7vQ{Z@O@vfG+rh*-d$^Ozv2AC ziE9Hw`r%|#^YP-YMV*91%-DWNX?<4U)kQxMx{5l>hDd^g##PmOqilo`_ zTwn$7_WwWX>RSEbTs#GoMf7!yjpk!lkyFqW_Z@~L1Yl#> zS`&EpGxFiUw{SXO@B_>_#j=6HH~*XQJ(jvFBRY0J^yQq98ihuHx@TBgcJ}|19zR!v zecsT_fWEv{GBHoIA5>zvpVD65$Iz=QJF5A-lEa2lkMb1n*n;po!qmCR`)RfTn|>q% zL_Ds6*|9o~XVP?;Yct+o$QNn|!^TF@KVAq~Ltf2X`n%*tc?#Z>?vuIR?H$<3GT9=O z7$u-s9c+llvL!t;13!sw@RAkAl<>C@WeDhR>G-d7V2#XhtBHc93xjgnY6%^8sqUWYL&Wjja2(9g4;Wo?BIdx=nXF#(U_bIq+}kvz+ntT@YkgOW$av* zMfSWS&eK8eKBIHHb|3Kvx6E&cmuym{6zHK*d zdpm_(x6lpq!Ts_9VRG)eDIlZIBF+(DvI8()$~VPGE9Q=CMrMM7VasQ{eiecx^WA!`O$wF=yw?Y z4K}IS$Hg{SK+TY-=l37!tg86q_iO<7c$cu4RM1^P&Xg>X_M#6uTB;#c4Fkq3D?69dlqlrO;#-j2#bMR8O%}eEAk44RY?TUPsQn8`>#m=jikonM9MmL%N}s7 z$6wlnAi@oeKeV(M=6|WNv{)lVgv)xn-=$^F0)H%eBE2&o9$CSvv^;e=lO*&KaAeV# zP#%@AIxb%+dv2Tj@0F6_c$kT&$i(VdX@wV<_(CAZF)V%fh{wa9R_9PQu+(b_2|R$< zKnI{^%jfP&xOe;V5L&~5x3KK8@B*mkoQH{V192a#fVA3*e7)8oJXAy(Bgrk9|y(^Cy4UR9Fo$ z!^E?}!eIK(s-yjzlM3H0nDm-IPYD8c!-)Q z?}76T$N+LPKkYV9BC&wjsJK8+-aFx;|A9&>r1vQ~&uBBej}YQRoUCi{1lEw`G%1HP zHYbK^tr@JY1(S+^-hM-wo1vu$XAEr!Ulml)m8k*PRn}0Yu}o>GBmzUx84mtiRM01C zc`;G&D}GN8F3~KmX`ZouxH9@gJivVJrg+5RQyI%Y>IaR9Z;C$?{oXE1=^MV;AK;p# z$bR2-MfQ^L`=f`N%&+#y$?&k(t=A^KjHLL+SYVz_~j0((v6_N zzi$`5yyi^*ino%#ZC*h0aVr|M$B+DOm8Pl+BU{wCsC)KSgFcJFFAUU?f(%}{$S&S1M_9m> z4mls86bL!M07X|fS@8WT%&c$VYMJt~jk^#N)LQ-K*Q4)Joc!H}K!(fZ+5=9m=7+vV zzMNYNt+FOuHt=axJ!rbi(|Prwv{7e8vcEV=I^zCPbr(bt)A%i7w56`IPUw#+lYdT^|oIJ zTGC~rTrJ`R*L)fS)hL6sXqUu2&F zh6%1QFI;6L`Ic}`#21FY`Qk~vdOu7^t_JIV{fEE9AIISm!q-ixxBA!M7q2x*6X9AvT%RgR-{^J@QmluXqeZx zU{)ZKbiKh;CP3{REFa6EpRU9@eN$rkp0>~*$=AcdLnHtbzVD9RN^Vwc#T{~H1oQ^f zY#2!8w<46U)|=uEzs;fb++P0=lp;Cy1NG*gtV*ERVjfV==HQfU-^YVIvsm5%Csu5U z<7Y2g%n8Nu!6Jai@4@GHJkk=f_}lQwW+NUwW!^mj zvn4f6Tn2}wx-!DjwAv|w73}FvqK-sOv@|zG`-1Csb&Us>vftT_D9Sn8jfKs`m#*xW z%e;85;|$yTgTEp$%req%qdU3TCPKYxusIXxg$4FF9j5P>u6Fiv*{LB>C{3#CPE5)V zwDAnCdAyl`oy^0dPV3XHP=V^s#H#!w_ zW}oJO>(LWIUT7A3#uMm6wZKwj&qkbA$|}H)ybM^{vWxw7a-G!HzBq=s? zm`&M-Cf`{ahCbAqBU7V#lLU`$eAPuO&_J1fkJcOc&%%pFzj31ms`V--4@Ej#aVb*y zR75~k<)ga`R}@5GWjIZf^CdK0!62_VKUmdBotX@+#8MGeWj4N#CI7{Ke1kKEoR+;& z^Rr4UKJ@R+39S9d&oBN!*ePV+qqeb^caClBOE`_?p!Li9cUS2H53&>wN1u|fB- zT%>hlolz4;<|Zn>t*GCkn@yUdp<43iA&+>!rp;GK=`dKc8(&vv)%cYx*!d%wgqO9e zc3v|&ONeT5Q(Ag!RUwq+^{oMHEkp+iug<#Ad!rW@nzmzU#~@g*aC^ zI(Ir-)FnTv1-lA=EM-XZ?(?%iFLP*V%doSG>0=I8V9tV^V!I>3A`f=QH;q)cXEbeN zyH!oI4bb;YLWh@6Fy9Ou)2`>iH9jSpT&%v{a%5#%R5MXIJGK2nd8Q{K($5BXT7*g) zqw%V(Iaj7yZkdSq{4|RBUvNf~%4BQOJUX$O@ZiA;qqq)9?8|f6MNRV4p8`YHBuof_ zyUg0G4!Xs=x>A#Ma)arogq`o`TW~-59$INoGsdbX5Cmt5 zF`D1`*W1}PZ8fWL%1M_|=25uwk9Adq`Fd6K=!8+=2}O86;o>dH)Xj$K>l>eS?U}qZ zMOvhdWZ(HWnZE0g8oQhAYS+AA>c5OXJH7oXf%0baYxSnI+A%rgei-=F9P%&rij|f7 z_WW5Ka^M=cxz{loz@E|E$T2bEwsA5v-1wHGeNv`xhv@S!U1!qxH-AyyYOer&3>s(O zl821laK6}~l@m6f{PZnce>^<0Sg~ZbgpsNF{v^pcJx^o7lyJe8!@!v*ruS8>x|gnl zickOOR8{(Iz`H;;5g6la+;D6Z#ocp=C=osFT12qR?`r;;oj?~){^Kg*Ae$Fw3Fuu6 z73B}2--LO|o$m^X{@^+biZML{!qoV<;QEHgN9fC`_2`GAP_1MUeXZ0oD^%v=@51U- zu|9IpROMmsCqmO(ha;SqWpM{WW5x`(s0NQgy`O1dD z4)yiNFr#|gb(%!2eQcA(V{>OE2vROU+4t33+VC1z{=Uw)oM!sV<08~fHq?^86N>Z1EeZWNZ;L1cZ%y}*BZA09)bCm^{FsL3mwq+&{@C%H8=mO*n##|H4{cr)q zhhkb&)Y1;`<$q#E6k+upsFo=gh4l3v#}62=Kd@-tG^!^J{q(s?5o(NzymPTUw~K0M zO47M))EZLKTRZ*}gKzjY#J&?4c4pgKaZ#(Qrbp>>@^^woibv~Tv@*<_-R~D_WV$8_ zNB-Z?F=txs{kl;;f3{6m{Eth16*6D2KRTk~)i+ zCmTb2+c{&tmN@%F{7UH)`pqpww}!XVD-i2lx6jd-K`{kMKA=cZ&p`j-(36p{Z?mLMGA#tI{chc?7f?LVz)%n_;kjt z=B3vGdEM<(-lQG4=Urm@;%_9h=yu!PSdxl%u00H1#M#o^Rbn}r zJ1Zl%y+d5`F9+t=O{=5zF5pnlBL!)~vV|*3;;3_8aI4-NMiB--37>7Wx?4=T^*$se zsB+4;U)gnZVxH)hwn_{T)ha58Lwnn>x&!B`J=M`COMP7T>okVHImm$He9DxW|7`&r zIlXOO$;`JQ?*r5{48J2yy8t(2Kq4bE|DU z_ZOuOaG3X3PUv7Xc738RqXmXR?`V0YHui^gShUZxUv)P`8xTOV9d*zOo%R( zmT<>&%$L(W)QXi3tzQRAxYk_Mx1yv>)*Y}g>X{c6ivApb1vAS`6vr_Do;s1{THZ#D zVDtoSQPT{1W#qnLPF>(n(7<-NTf0-^-8r+zrWO?XS@Bcf{rMf7c~%xg#mz+6=Oc`|b9KP*)d-PLy1ZGdgq$U~7KyVl^u{Lxa>#8tohuwkVbi3ynkfBh)> z_#o2C+s8FmYOkyJ(P5-PCsN8cAnD}DxT+idOzng55(cGibC9~LCA`EEe<_49ervz- z9UG@+61w%}pYRJ)qJqB5tEmer33z+Mcj^I_`F<~lcw0h!+=!kMQ999H2Of(%@b^OM{sRMjo01&8d3 zW7~#5KW(uy?FFq7oK_Gnx&sJ3G#8dG>T5nRT+*$cPUCHSkuwvUA|E>O7d4u=+3T5K|xJX92v;@@GCrV)Sl5@ zRl}tgh(SX7=JTzpBB{IId$$)p8|!5-y(1dpWGg$Vn6Rv;%#t|Nr4g$3?G$F6vL@7; z5ET*TMfaoMhJw~(U;0BO@uB#%>>atd;qK;}?)jZKY9PM)`oG8vpJ z4f8oF3Ia-CAdZNjQ~Ht?rn-$xO*I_6!f1m@e($-0D7D_M%d43;lbCQQiCa;6PHD>} zWgROvrui?8nm5i=Y7-dw!ssec(Fl0M#tH4F-_f)FVPEr^!Ke0bIa6edrl)#4eMexU zmrT;vBxEt|CKdi-xx@3 z9YDd`SnER*I8}V;oUdREFZ+5$!thGt8Yyw^O2vqR!4MBvs-_%|+LB3Q&Q0EoQey#? zG1=a+?m7p1)0>d8%{@qv84+zUDiBPQjh@g;bK94U)N$C=>Pu{T=!nr|jGLZHv;QDn#*eq> zzLUcrm1`j8=EmhKVuSkREmILxz^Gtb@pSoYYx^wRJv zxavJeQheQZ*iiezldwMOcjt^TubNy6rOxy~+a*Sp%>q%y zf{uX5sy3CDK6s^+acEf_i|V*&RE-arrb%DW9HWH{mJdj?o@s>?^6@j)@y2hUIu+@> z_8LX1UNyO<(mCr6cbnCgV_q7Bh0dRqJ3b7ugQZ4}1#kOjFE(m&z@+6}!29Ec!lW*x@@0>zO(X@h7~RR!3!%hW0p?d5)0NdC*jVq4zdV2w0Wnt-i}H^-a95Athjcx z(MrjjnVtO$@{eky&y3~mC6Wa54OKh85Pdn;m#M_hrmGfWW&Pd=0q&+nb@!2nBzhSJ z?3)dg&rI1rh`(t553JM*w;TUCjJ&*{lsq&mGKFk%o2Nq|!5=)HJ&Wo< zh~=$ZCu)0L&9uI3b`5M}IE94hOdg8R4+x6BQ1P%XikgLHDPMa8r|(}Hrpubm{%X)vE!2r^2eY)xEmAJN)%b^@Y#tJEoSk|I>R`?xSa%WMqI(DCJ^r=pGj%vLX+ zQuhEBRWpL7KP5ygzCKYp^N`4!irdq9w7k<9-<%ruZzzx=-!FJ2ry#Ue>Q z{wE7Fv)o|f_ipOfB52v)y9YgjICmHK<_ubl74(CT?A*W5a)$MxZE~=trea1<)9V=N z6uTHdQ=onSin9p=hIo@4&qeXH;rWKjPFFYlLl1d>|CAorMl>>-G^z{9f4%))(T{&a z24al}j}etQKwnOxJ8Gcm$U&pbZzCJD&0uUd;9Q`EvaC79Cf;X`xe-`@?sp5xDaq(X z$2n><5Eb6IyZ%^0%y9IiPF=#*JXu#09Qd>OVyR)iq-z45m6~1sYzabhWW^w04l6jO z8~88e_ZQ>iJ7>=Rf@-pK8PE0B^1q;S^OHrPV^~#M?{BP1*Y5rCfqlEM{VdVWGe>EM zUrE8QR1FtT%Q`qHfbhT4)+b+5lZ*jXU1^PQTtMvZ2NxpfTX#Yjg$6JVelr3UPbAgS z#`j4xjBWIIrpK@PDUb1j!oWf<|L{s@+eI2`fig>>e~io7kD-sXOqsn!$bDrL&v-MN zf;O+nUMnvae7QQ&=Hv?1({HQHX{syJP9GGLk0RYL;j`tNHbz(+G@FgFx0CzSb-DH6C557Sry}3r;UDG;aKY1o;G_@H@))`w@O$(zg zOUp@t$~ugO*xtgHw~A=oh>uzGXDc3o&c}W{>sQ*`%-5MB0C^)XXX_HsX=T$*Om%(4 zly;`DHH3$x&bL1iP_mq%CTU)XctoM0_;Zp`oAHY7x}NJO&zU< zI-(R_;S&T!oMm7g^^k*SPzBPD)?L7%l@~HS8aMG6+EnF~QOp8{ukJD#3r3!z_q;jx zBlBm=wS7f{*m8|RN(1*uYy19A@nkt9A!XQdUw%o5w>rYv+=fv;JMH^BC+06V@VHm$ zbO*q+;DPXHA-$8xvtO5}-d2oE{)IbXl8#x`yV8wz`KuFsdkrO0?svPlV2NSNEuH)_ zU@3}!>?^nNr$VY5ZovE_1Ylj6SCzDX_=XXb;byP%3`SZuzi?@YPbYJPs;D31<*2BJ zSUFi+2X5T2AIiK~UffvPZpF87`5S8qRmtBu_ARu2FQ1~pHd=G%pFmK^j1mA7nJ{e4 zZyu72+2sE)$u!b-;a;bKeuDO9a+Oq?N!lOwOjmP#xD*xa4>|2Stt@M4k55`o{%__X zf+tmv38X(iKKZY?A2P4u@sY9;`zfx@_MvKROHZ~Wo+oW3aa%vx@Iw1zQX%OT3iWC@ zAsM9@;dfnulae7Bw$1Seg(nzF)kqey2}Iva@dop`+^agKCfqN6JD?GjjkF4vM{^B#cg!eBw+Y{ z_?S2IzTeKYJ+`$&jGbf#v}cIW-_i!xqy#$ZuW1H{@Wk= z`<9viXqKlAO;6CZTxgLGb3FTQ2D~2nA+%8sKdAvkE!{^XuQ0o?D?*89)(d%@NUppO zW5?CuVcX3~&rRRn1E0_v$fb~oUukZC6VpMi+CnF5zIyrloyeQNP=k2vdokt~l8%#k z7wH>q=+1XHO;oLiOC77F{}0p-=)gFSQ45y%)K#Ku_XYE%2EefEsn=dxJG?^1qpDu@}{Gi*3M*ej!ThJ=L$a{;Tev4ju zE$+Q}KzvGw&6wG-lCQ)qQ3B@wQe(Qu<^vr_*6qdsKIS#dAX4*RDf%tu!>V&*2G5n9N0F8OsznPwG(020n)q4c3 zX&Si?8F3W1DhpwR0HnF0L67R;p06hEx~nd#;Ql#QM&^MaC5PIkohqy~h5p^qJ>=0d zP02FX|F8J4@$W^~lb{i!Dwe)aJ?2}OjTUc8gWDOm= z?*k5gkUwEjdzOQhHwe?+B4vYqE-puOLK7NM@?}FV2xoh|33|KamZ_qm(${t9J~jL8 zw9UxE+R~k$_YclXewKLn%{*Ba4HfjA!U$Zq!?zfP$jhir^Zz(FCP|h2+`{2z21loE zmmD3!h%R0Uk}ea{$bS($(x_;LfadERcy4Xj?Dy9)yj z3}WF=iH|2RM_T7r!DVSFcIoTb2^MMNsE`pG*S8Q)OLhiCSOn&!7rj?qIk*h!5M!5} z`DrkLrmsJM_htcc==bI#Ng4G?NkOT!kIGbdrEv6@v01d)c**Q!V!^}Pk;g6RYUa!o zGi8bVl}e|DFsBLp17VGyu%hRKEegMIEyP4@kCCS0GQZU3B}o+lq| zNJ2h-|7vn$I)AjmsYv(hET%U*66Gzk`A0RvK+UZOm({10u>8J9&Wa>ppohN1Ko%1! z%9QA~o{a+a5V1|>JW$8@@+$0k5=Tto#q>tlSuj>pNL6+gO^nVYUDpUm%K4iPa-rc{ z%>*peJtRw;s&)NC3BepVa7#Vgrc81_@0+sdcfqmAvIJFuh|oPp9i&xy#nXcFiH_DS z+BH5bvl;pHX;CGuAzB0BW*$h`1=AZ2S%XE}e=w-~Wkvxz6UPnDpsmyL-bs!LPy6>+ zQ^)M8jc&IWzt?__kmn4p$MExJIv5MNltXj1q?rjcEZb6j@s^I zk?L~ST8Q!CmTGK@5=R5a%x}LI_F%K(ckx?<_LKSOLdgUWSsbQ&`T(={bZJpL5M0Ar z4E`DoI#is#w2j^f6D-ZieTtu6anlT9&aF;mHjDs&*uhmyx&`VKTz=u+2UW)u8q*OA zw=7f)crj5quI}H1RP@Nm<)mP}I0g+ZL-UrE5q{Gcu8yzK`q@h+xW=Cu9Gee+-m7#; z-10FIUgw6!CZ_!uQt;;1^9{iqD|2NmzOl_q&%=bg8QsWhtLiWb0BJ}QVKsi=PfaFq ztdMR_Y$4w>YL0VwU5wS{&bHEQ8w0zOsFJ!n-nbhmdgc#7((rQ2;39bX<7PU%mSgHK zi>sI+KA&XF0-9Ct*#S{4*v8fVRjiA3UlUb%TMg$^xu1!;A}-l-zQxlsaT-PzGMCwH z3If$lUII&f%%C(DKQqbx_XDY0L=l1$sa84ty|RP7dCRf-^rzqv4|sGO13BHy*a~EF zjEj1aS#&uueJtEXJ8TQ<*ZA^*#kiZI+anXHs)wonEK=+~`xylFSZl{5$0&VOENw2T zk7D=&Rw+34GaAkC5a4?sd_!G>9=uNe(50#QShQ;j6HrvrS&|mM)MlIXvzRP)8 zwEmMvO$ofM=Z%rQgdx*)@$;R#$LW!oMM6xaJ=$letZg zFGH%*MoD+?5S><#y}@&lIGg;*Gi`mTBN_dsn^g;>?inR|y3(pF( zZt${JPk**Cnr3OU!g@%9x-L*g=EmK;YM}e|f9jbt{Hq`L271HEJR2&t0LlHyi|16u zKEaM1GmowXIx~5THbapxr?w!?ZeTlx2KQfv)9qZl&j`8?ozDfl=SnQuO4Kh^RkI|| zxHKF0ho`=jj}qv9bJ!b6 zA2%kVGF#{kEZfr+{ox4_*VTz0ie(3R^8a3KMa^u%It05t;_Ih;->INgE}zhXUKT=g z_h=62c^ARTxma(2p7hjX@M{}%XR@jCI}7jsD*5&EVr$;-sYUjnJ*A;4`Q?@%-bK34Xda^IErdE>-F3y|`^7&ZH3_HfPaQMs7y3#v>3LnGq`JAG*lBnk`)+t7Nl8%T_?$^Utt~_E2QtzEhw$cFQ6f`JpJHSlyrGbp z*$$93YfqB4(GPP^Siv!~Z)_{R0b~3hk(o;NwT=CqaVq zm0IZQ33z0vTJ5-*oH|bw2xG{tuo)0XcgCC!?3`ALJ!T68PrU2y5o{1>eHX%6`SeeI zR`ySzhr>5Kflv$DN_XwXo7(ql?CD=#O?xk0^nLOurGM3vb5i-;E^E^>#qR(UY#W|` z+XSN_6oq0nqS0N@bkEthW_$2K%KxnWs!WBe(dI#Zn~6IB#KIA!IWLep)>h>|Y2O$N z8TsZHB@v|-o^EGkm>56FJUJsw`tj&PVHr9R8mWvE?)>+iu=jQA|3LQvznBVgbIEz7 z!{TuVXqCHmg}RsLa}BBuBAN9zN6aMO&$sOZ=%uQUsOw`n*1x=%o|MheZ(rTt@J99V zp3xq0t&V*B&>?n&`q}uA_VLOkH2?36j9R=8ZvdHi`6-?=$mS6#|F#I|JU=Gg8ALr#h2%y=V;ekV zFGDadX0sJbv!ng$g!DaUA%~@7XwQy!%|MGis;gp>#xjvYsY+NX?d=nkF0ktwkUR_~ z2)NqL7AdA6d!GiRI9DMJ-TOz0z6!nj)3>Yt!Rl9_rZ<(cx8iT2ZM#c9i%v-HR-IPU zOp%9o?T70*k)2i``YM1p&gU>W=7PR#(Jf=PPC)&>D{TPt^uC?RWxneLmzn$?Z;7`D zJ$D7CxTFL50A75+A929=mVAm6&y7zc*ZiTr&6^p-_F$8H#Q)LaU@jA za)3x%FpZ_%WJ){8Q>SLU@`=cyvvoNbN~7zcOd!~9IHI`HOKvL_0nm3X)=tZe@WHCY z4qg`4+bBX(m0$uRrySqAS5xq%M=QZoU5l%bj2LQpca%2rta7MI*7A(Ns)seG7oqA= z%Tw$|_y*IeTh+p2sE>%Pi<$)|n3Yv&GN`WEhKgw_d|_*U`qb0JAXEG)RA3e-^>N#I zLI0VOc!TXjQv*KcuqOE3zq`Quf;OFjQOv|I%ujN1r@H@{TWIhuMt=%J$WpA1Hd*20 zQzNam+t%rYB2XNx#^Yq|)M;4PB0=f$6`i-%M(Z9A(mjq)$B+%AG8w1u(9JRgcHCNTE3FujA|*FpKMu&vosSvQKNi?S3vAYM{hM7G$~0A1IOk4OIZ6qJRS5>*{~lj4 ziksWNqm@h1a>r`I2nqSre6hs*%84rYT0>bLelbev^g(ZdGrVr z`1d|F5|U!&5da&)eAsAQE&K*H)dPZZ+z8ooYpR+>di(L|j|BvX;VFXbXI@tz;7wVU+`Cati0e^O@gjRqReA4FkU z6)Y)wjp^S=f8lv)-zb=o#H3Ys24PEQ@GN+9PsqS4e1DhV&5z)~d@$W0|!VO26efi%)QmluO2NwL2H|sm}O>OA1Wp{+tkt6c% zE%R*AQc#1JuSuUOF3&MWgY4BQSd|^)HvC3_62afX@X5n`z)<`7;ft3NznzF!V)P%h zmpoH&r!S_$Dd<<3%D$10;k>VLUK!*hOxOE6Nu+Y>(hTJ$TDN$daf<)b7!w{eJt-S| zySJ8i+NR4CO3rEjTa<5+6@kQ5*H?y%FudPP;ZI#^Z@pV`KiJ67D{C+SUi6(BpF+x0 z7*rVV)O|%@!jS^o1{j#7OFGAIYX6*8z?!D@1hmfiR&7)YKTv2MnO9ZxRyC4k8?|6; z7hyGRc&d3JQ3-7hAR^r+U0PY)IL~U}JDsI*!Z=9LU1Pg78EpF4LveO+d)xU~Dn8I# zFknh255nr%kv(rRmO{Yl#&F;xf|&S3+2r~iu9W=tB~wxj#CxsMU49m1>e-&#EMNXi z;*_2l=tnX2A`^+*saERrWN&y%rNk6I=gQ;nF?2(s&`Y)*!{rJi2B}6VG66(84lVBB zJGXQc^)F~3KE-uTUk0ft_GD+7TR;!cjq$~8<=)lR+gY0LO+rbm5`Cl|(;FywREzq; z_w4v_3X*_$f{gjE`_wJz|3HOJv_N-Mu1~p8{_`mS*Epjt0orm;W`zP{IN>SZt7x%l z&V_q7>Do7Sxy~{ zJT12)ikz)-P|U~bY~c*xcgyv8a%d^_Cjkl<(WzuFJ7T2ADI{YC&D@Y?1qz3k96y3e z*oI7Prw&Wokcjfgz5!yaQ5x&{5Oq5P21OWipfuQv@XIG4f7kTHuaCAD0c=C;p6Oz~ zkg*n*whC#G68(bZUnKlG1u7YAq-#K$EkQlu_5)ygP|{5M@t`+`D1-N>gf?gOYEFzz zmwh%@9DP3<)%SO)v1NRdUO>Vvcv|zfX%{R0DzZn*|Y5 z`qb5z&?+!Dy|4;(+IRTMtad!zkFcmlLRHVU-ODL&k&Q@SdP>h(?IGd)4tt}oi@-$} zWrVAkU~+pvSBW{i-1b7}P@e1w^dx3jk*2cgkTsR<5>jzxa*xfS=AVO1O+3o0;m*lL zn@8fEbZ)G7ZSvuH>UTSHjOSw*ZqqZ7FBxJ778E<@Bw=_3T>(AGqG$eo+jZn)WjZF0 z$>kgX#SGV=FNo-u>A^vI5YeCNPfo)BjIkUnUY!^)l} zs~V8qkmDc_|EtrjD5Rp&RE35gEsDGY zU*pxFWCj5VeMtZ*tj3_=I8{XBXw-wrWR6p(=k9%V;IGjc^|h zI1UOIvsI9}ZE7iVO%XNCcUS%CH=>PSf-jb#ec~zg{s&4&%4;DuquSaFT~D;1<&%TL zb9NR{S*+q>Z2h_Yz#u(i`s!=SQA&8UN!kmsJL`BkzFrSEVThQ%(fUd^nH_;*N#5rV z_@|f#V%~vgyRBrHhdvbA`Pp;Kl(PlNeKd{lGZE54_ipVQ;_dT#6OEd8pz727ddwf z8Lr^1N?3;0o5k&)&$f{<8ZW*vw%DC=r>PKs@F~LYigB`O>h%SQ8Tq12b>ifJtuo*| zSPgH*k}o1l)5(wOY=(3aXJoAMa(}5vXUcTl@SBLw6A@N%*{O%mcDAQ8&E)=$vZRh6 zRn9Lc7kM7v`eF{^8<{9(f{o9UNWwn1!D_lXYEw!xX`Xcz+LCo110EFqMGi35+skr# zm#Gu*Wa;;JkMo6Gh!IfPQej4m2(l8&Bl{+6LOW%0Y@yJr5H@Y$5CDerQ!8 zihVRJQvI*d805QA?PD^zoTP1l*X4c82;QyoL2{Ig*jJy7qg_w{2)0P zj-T``#(BbG3f~Ij5z7*9EUUO!p1RDN@U~cHwI!H}omPP@4=ZC|9@N1Xp5JS9T1xy{ zwyX6wX(yAc7$*%(&+WgwT&QhT)&P0z*&GDt{=%E}IrmatY{BMj`pYStuoMr%;Q8Dq zuH@RNp~&$L#`3FS1A_CY7DS*83aT}ft=ZA?@}&4{+of@$VpG7}SENqY08h{D9-(eP zh%9yD4zGfR!h5|Ag?%+a4~-(j>GF+Jh)wU1!T-_p)^AaM?-%b-(kU%H$SeM^RIrJ$xAPn z{biAwOYfKdOo4aEYTe?prf~K}uxjt?`ax-O6A?{AE^dA0KYgR_7mkieg)O;}AMSHC z>L{woaoq=|Hl(#wZ}5s@G7G?KuTo7-?+EZNvdi=~RBU6ctD@sYPtlY98NRe_#T4$A z?cf6EB-rX>1Yk6aom8{tTH=@jsIh&C)atAPIIYFETn%$!%O6ZRa0FYbNBhg;X*4Uq7%H4G;PX~n8tx2+^DPj?QI|Naki0ss8sUu%cE$1a&ry8jl=hvP$@YO9N_Gj-+{9pQe-9Fp+4l4O0?;l zop&WpV!DWJKdi&}Z*4O(2G^vFbSSW(3 z1N=yzn_@5?cd!pz$QzBMync?b zNVUg}5!hZS3fzj2x}>nuOd|Xtc_aMY&UjB7+OGdE15ew6g-b*jd}@D>vRW9A9T}AE z0L+$JbD>_U({hZJ1`fjSXgWKv3Pil31jl@;1RM8o&x1>Yvcm(QtC;ZZN2nYa41CAf z0|f&yOyg=KD>3w$H};-YUDbNWeo| zkO#MAZpVa4q5-UVy;BSXnNVY=Bj8opa12`u%>!1t5ez>=?8R@WLX-z?;11iPgm{&e z`RSJG=wQrfupRIuH84}+U5hNWsS6O_>(GF4q7#-A1!qo73YP>_3hVh8n=^QSwpx03 zdpTf<>my+l$MRhc)|sUh8-`DPx0OfuNp~~Z@@~^e%GWKJlupQ8)Vsx4$R&=?i~zy# zk$ULBimV0=XL}qE z_xJH&ys3`)RO)vM>tPkB)P^^0sanf+xzLFx#2bFFP43PXB8F+!wTw%|R}{s`m-t5W zi!@a!Cphx)gy9DBO|>zSw?kl6f8{l8@0 zQBunExh+0Z>qaX)nw1Q=@%7;t%E})Tk^SSm|U6S zMVL3}vZ|!Z#gi%l`j~Y*k{CWEe_-?`J6U*bD^oi_m{Nqi$`kUk>1BC&adPJ-*FYfr z)9U<+(J!3)0-@OAQ}oIrEE?j>_id{&_6zTc(j{sYlF4G@MT=Kc#$r7W^Z8cXk83A# zevVc`>RiJ=>gcf_*RvW0i}QXyQn!fb1Ff|4o@xDL!v0skWKev*_I}MapY;1W%}}!; zOLe&)L&=UuHeea}m+(}4;tPGQr@W!-kzukZ4aKdInxy#3P}ro`IlUniD}I}xG=X5k}L;mkie@C+0)H8nC-3 zaB9$8u{Hyc1H&6sNRbS3{byrbdI!dReT*9;G7HID9kfb~T0jVdUePDBny~l#y7CaR zyoRq>^B1tT2>(>$yi(7t+(_0|zRtiz@rYX^L;W4M~xAik8Ha4tZ2P%UU@x;U* zDsbJmjCyG9^xs8|m~j&?EAt&scFk?IG4TY;of)#vocz)^I5QsU`bv;ARif85uP?HL zW~A*W`dXWRZO_U))OLxQ2gVUV)7+|Opt>!a2Bi5wp zYew8*hd&cVi4p(*l3ok%FK)8X)&D@#EAZPj<&^(3#O6x=yi{oY3a|$Sa-|9$CX8BJ z1ijZ~QYtGgZda0BgK)lF2tz6eLL>bKOOVAe-8nm}EUI12J%D|;Gz2J7DVgb(_fO4| z4aW&pzf}I=l`&`9jjVjjUWn-Xep$|z<}AKdgELv!hQcK~57pD95o5sqU-^P?`WirP zHsrjLjr16E8?Rk@w-%HuQ<^zhIb`t9&|9lDem66`gah&JZh^CfgPW`?=RO#qK_6kz z0N-OWjojc56h?a^z5C}~cVRmqKOuT`SyB}3@%2$F9jN3-*CD}Wt7=J0CvqO_M zYbLThc4s<1_N%jO&b%BQNr)xs+Zy=j0B-^TdI;2+7eN8QBJdhyr_8sjzxvF$`so)I zRIbT=n=h&3!ViIzjQ`S&JTg`0)11}UdmWGliY*9ZjC86vbEk^8)K8clxeypaRTr|9 z?t|k?{JG`L=#2f?{CP+M+_AUS{DSFltKq2Ia9($hGq?J1qae1@2L$ z=9J4ypm4Beg>>)p^-jyyoh_`XicI5gYb1Ofy=VgKn(Olyc}xp@c8cM_7vD$}8TuFMLM+v^AC7E&Z&-08|Mjk^r9-lYaCu zV08mieh7thdf;I55Xnw;``4IzHAt-QoDQE<%_kxx?u@Uh$0eb^SP1&Pz4DHf^IXW! zI$SWlURL@&9Na~2q0GLo!^{pHoNqSOuX2!my6>!bZCMypoRED z=-oVb2bZDvl%;Jt^(jbsG6W=|)xuhZ!`hL6x!S}~MqSe0hTsw{9vzqc>JQu_Yq6i) zhW^0wao6*HLy|ozvALaNU#YomCluR)Yn5+E(X7vi-`ymaF$@akJy0-Y2bhAGH`}+hY6XdheAfS>GFCRcDlZ#XLRN}Plbsdbf zB9|%h;U9&!UKz9(vl<5o*L|`Mc!V}ZC!wmI>+y4xT~&U~Y&RSPU?GF)V_`v0&Gu!r z7v*dYdV{j+%dZj^;0QOPk)t#CL|eQ4utB^&aB8Qs#%(&W8DpJBW5t)*DP)TB#!ZY~ z`SE8RI}Xv^Af66Y1_PxQ6$xxt=9SOGULlmVL`T;t6p(mD`R0v0qCapUW>fwg0`J0Re$cwDWcCX}t`!{-%r9Z2fsfpj7bYzlHdq+~agl#~loNQ( zy;%L{EQci;HDVhZ?~Yqlt#E#2 z!&cVRtS=6t4Ba=nd}r#AEZiW*?gEH%=0E22V>bp9%jX)zzBeT>5kykW zgbo%3j}*N?EyRR`i*x*`@EHp6{T4D7t%|nAN@I6bWhl}~UdFCn*VBrZsMOZFkr)Tv zNwQe_7)HKbBZ&en@?MVkkGOeTwORFC+|3W())o^?`~Cj@RBzs)?_m`Q_{KW=7O=6b z=+;z~{?J~D>J;XrRK^^=XzkuFp5?(HYzpqGqkc%)V<~%M`8LOCrb;Ha{%k`n2G7v6 zL5qZR54|Up+q6ut!%SW#i~tWIQDn@?xGx=81C6cyqU!leqn9+M4et-S zoAC-@PA^0*mDP$En@w9!@Pm9Jlwj)5o0o}Px(UUtZ9C&Z*T_L~JNT%TBjpQbPp>s# zRnSx;;pk_cxp$dN=?sz~^2#-r)0)^E&X++qH@02v;HCw_nv&NK6FAXL1p`c%gi2Yq zYd9@oG##4mgligp6o~skSG|{QvorC))Z`6&+t&samuC6F18M-YZ=C#l(qoOq2uyMI zV#EQf#hi07poTkPp&7%g@+(_nD$dj^@(bgksGV~h=Nk$7dbk!FOGfO#v^+RF+$#BC zG-cPo0Cz%rCQ94Xv4hFwTAPJ2NChpx7?nq`Ni0fji&46v<*H0IhW8O;rQV~ogQ_Ay zLDLQiL_*c$p@+ep6vy+@`^Zk}7-dT2`atRHl2dTJbSp>=$&>3CEf}e|gr?71m@!^^ z(dv3#eyKZy3YMA^l1(2wPkl+r`i#x8uZ z8mdsA{p>}2x1LF=wx)Wm)OcmTyDK@(L^@A1JkkCGNgvgxnv5(x zlSxs^HwzHX54K;_k~~)*-`6#nF2TsikZO#)-0OH@?H?1D(zpIrSn?zFD&z}@z_>ZL z>wwZ86b$^vVqr6l+8g$SOdSK{phdMT&5`M~o&!^pG-7U6ax^S=tkv3yi2XHpSVW0QYGjf=VAcjnE66yBSO5VEe7%*s1XsG<3vGPw-h7r zzYd|A^brj^8urhl_C{EZQ_U8Q%C_yUcm-ZuE`t3b;*~32{$5QED&K*;eD*golYT!@ z(n3C;soR_CuSBW?_GYX#;Xt}?sr8MdwMo)jpXp)?Of=P5V)NLq7$Qb{%=Jl@qhey- z^)@3S81++%I?6oVT&A1=d$izHVzi%6d&hi%0)UW~+I@p%oVyRog(8wattibf(H|PU zhYz&o~6=s?DTJ>)s(;MCBDCJvz`Cq{Mwx&2Zs^b}YS?kAf z=4`SMFB6L6D;8LuoGK+74aQ&LbE9Ye;b%`uny3mL{FoqkRACR;6S=2ZhQ@54(n|2x zKsYjFsJxV=7?ujYDDCcd>kmPSBc)>?dRAA&LfL$3GKCq;=XK-C+e?Ook{+v_nXnE9 z4UB&a<k@_W+B!GB4f{+AZhJa12pTH{eu!$ z8k&xx)NZLTP#x5TIvhR#GSf%BP&fEFEmXB2fG%nE_<6G9p)N_Nq}%5dI=rHn-1`!M zP_d1}7*=*(+0wR#zDZmrLRLQS8VgpSTJn6}aDGp9poeel@-f45d-|_A!PE@P%d>0) zF^Uf#ZpzH}3_^%<4PU#p{?Rame3}Hu%j`A%2++*NoAP2&pN#!?r$-}}ug*o(@oX7e z+6A-71Av>{CU>1v4nZBdcPkp{a;W}&O~5${ne*qU+8X!Fdm#;fiy1;1 zs>|#S%sX|m8C3`DJ26kqcBUb%>at6WmX7G2jg;b=0A@I_!!)B>xo8t!~}mg$)J>slFyeu||`pqTWb-|5IgSt_@b z@Pg3DG->Wq+OTm2dp`mfA_LuHJ=ELx1I z(RNu|E$vYT$u3}3r78Va78ozt{LAY0ntw&v=vMH{(vqX~Jf$#E(+nM0Qa~i!{d)0O zu1cH2v6Q4o2Z?2?9Y*{e#j)YK#XAyOEIT*vNeG=z1=SWg{**;J(RoS-mb16RmsO21 zA7K?F8n1n{ei~bBByjSC=GY(7BAFs>u{|!}1;OQw0n`uNWB^JMlo0q3T2QkhZqeD8 zSsl#HVlh^lBr860`D4I-%#(^aP^OtYO;BBj1AC*mwa!J`ltg+UsQ#06v$z8XM*$g6 zD;Y>IG@-J7waRRSAQaeMXSS*!-p(|-UmDy#g*xcpyRwn$;SQqhFZ~BnG$|5T;k#9( zKGyjMj}uc>RzG!L!z=OutKahU-}wyKIElG`hR@7v+z;j|$WsH8I;~L4-qevX<=XR$ zgiumrThiBp{V($|GFJ{E!)gAf!ziZ8DMoYe)=3a(-re4DSOufSM~hTd+be!BI)IGF zQaIde?@wr?kKsd0VmsS++V9V@F9;{)%;y^CcJR};tc(7olbWelS;rdE#ZP->}x|naMc-M1V2h$!_H~lU(*b@*x%FV0h$$V$TKK*)!;T^T!DJ*5{O62a_7jE|4 z8I=wf1*du`piTm;pU_~@fRwh}YVG&595xviOu3(Jq$AF+3D(aQ>*OR_DiZGjR88ag zL&>=DNnPFfiS`!K)bXswf19HW93WY303~DLc?XnLK8M`kCCmggeeFzsi=E$4$^RA<((jG2M_Z#~=!ULQDi{$9oa@F!(y+pFHlrh3KW#7k zO)^slWJSr;uG>r!W*0oPn>ac0BYf<{@dl0W^v^bx6#k5&?rTc*$a29kMl^mJ-QNr` zf8bDWONdMs?qo*Q3Yrfupw5aTC|aU={5MaIW$}v2mW9U&V)}`6%qLn+auxbhrI+JE zb9^&hP3grM@}4@p-;7dPeDcsYaokPz`axTc-fXL@A_wpz9b>^h3D~ISM1>me!D?O| z{bXNK-Tm)J21Ee zsH8lL`s9DIQ{}7DK-rAa(VpDzqZM7lv8`S_!32x1ne4bytrk8{WB-Et8YiZ%%g9~s zAupKwSx-+l^3su8tc6oWR(6d-|VAlZW0R=e`(fmQ<%-wx|~I$Qb-X5tXoIYnyw|sjidP)ui6h zlfeIW^dp9bFwh-BRDFe^uymtonKbu9E@1hj zRb`vVr^zp88qDDHnut~TmM^I?HdofNy-l;I4bB_tB#ctIYN+itOo^tC=*)l8fyb5C za`?P^j7Wom`VnS~+PZ$?F@Ft?@#XjrR0XZ642r)mPm4JZL~m+>9k|bpL}GgO%Q#Z+ zHI(m;j4y6jx|WQ_z}AUOS;fRml+7#(FTzb+CrWD88+aD63nRoU13I#ZMYIxiyKvJz zxl?m3OnmAZ8Ft9954q;{D_^Kidh32Tn48#=ZQu^p4={&HTBw2byELf>%^=AiP zR(-T#+n}KMdKSiQyQtZrim9;H$9U5?@(^oKq=0YaubyLe#l7-aW;CxgF-Btd!z-4;cq6y-#NNV7$+>+i5dVf7(QUs_RKK8{$F~zrxiJ2P zy~v5McbhR%jmwEzA9Sjvn|s3abxX4)gory1KO<|jLVMtInls)I{)v%8l9vnr^M?Iq zj@2Ry?{hca)kM<^&BZTa{x<`@b43BjNHvb~%*ekltXs2HcP<(!G?z3uvDTNrl`&2> zom5Fs>&vuPs~&&y@*iPX_}&~V;+#aQpIg-f7e7&6+q@Y8j;&Ffd$c%?(B*(>?x66P z1o$Ac)yGj|Q<= z2I0#Ajy|g*LnCKzR7^xODN(-!GU2cXl)KXcEiIC?AMm$p5V;J|p=tlEtS>^-Xj= z>$6zw0-!;#TKXZehEJ*|Cjwoqw7i9=B(bFxqK!rY9Fg#B}8fw=}}*?0%MTLS%P zN7}{RnY*uNwYATg5F&6sEIZ9_?++tWykS8{&e~pP*Pg7cyv&xIzeAudlOruK@%5$Y zmT)o=JaP04KN@=ny-CPQSP^>qrk_YeQuhQ;wOw3itA?8DJ6bbhaU*KRI0(L~J6qdE zT{%WRleLLXpdSp=t*s{1d_1b8&3ZaEF=;*mx8}D|QtXO^17=3h#>}!xM#RQCWIDn$ zIT=r;S)>gwA8p#Nq|I)(g~sr1kvq%vx1JiEdu90u)?A(xS+`dlH_tyH*C!quUm6IE zP2Ve4e-LDt2o4W%FlCmOnA(tV2LYEOs@6E$X0%Bh;^X5yG=w|~u!Bwh3PIn#GfwvA zq3pCz%vD@uN6I^~3lfDwnu&6R$mC>e|IOyT%-zlKGML{yo;LD#C(vAAREWwxCT0e`6 zy4|6f9XPLsN8!~3?G>PmL~PtmMlFnTokg*5aDMF7UlEe6YumSxCJ2QI!tz_KiUS5D zU065`%QyO{2er9#jFQ>%57jITCOtA1;40uIjUEmR5hca?NgTa-0IX%D^EH<>)(*qw z@)VXAA>`l7GEM?4TCj9O4Z0-a*w460#=ID&>uCffud3-^8JT8)ef#Zd0w*k@`wnML zF=K9yj82<|dD;bcr*5y1eq3YgBD&drqKwAP?Q+!RdUMo_s7k9tPkA){$I~UC9I!nj zV>xbd8)&Iu>k3dO$g%_9&-y;P@JJaK`W27lnI;IV;kN_<8@Ck&RP4XT_y>#5{~lv%m^*)`Xt zpK>RYBo-)RdT`+-*{I4;Uii%oxw{#&ivgN%lwJG><-ldtK2%OHp;igm{bralRe-%= zO6upF>hal_hA`O`5!tlPG%#bNjroRHdU-z4qsl+jiux>Z#%F0@`KEYl>ym}&cP2Bb z6z2oh8&d-ec%?grs=yqoiN8Wn%X{L`0~687LZVFc+B);Y)+;nQ&u#p(l zRD)0K3A<(`+aCl%ecR7CG{L(+51Q-|`|Qf*OjC?>8$hGgn=rDp&nM=I?qMS2ZmN!= zFwNNh1HA;?byqC{Cyp_FooYuN=Z29<@*upUGGT9V(ebt`f?!hi`-e`Ya>yGnK&cE$A9oXANa zoLrzXfW2VSD>$`)YC3LmV`Bn6J5^eu8k>>hbl5wH&4J{XS~B5AQHhr0cxOgX?$xKc zptS|zrk5-!jNf9x*Vw|lnK*H}WE;gO*Kw&{tC695BZYmVxP^#;i+bsJbqXr!@OeGS z(5BHDM8S8wIL3@9TNwYF5gw2`cj9bub34gHc4LT;{(*3YPX9aOA63_FpV#(W zz7x!_SgMM-Lx`J=Va_^&vgsiQ3IZ0NipP|HkA57%Z@OeP)WJ>oaGT^&rfx?=Vnc3M zhhC{h*QhEcfH7OGQdi?VXeMVmy!x9HfpYo5qbz->VY+1Px=GaRoV;sM?YK8L-Dw5& zTq%v}P@^wZk||z*F%ow5P28)yV*gVyKN8^;TrOZdap(gUW?W;HPHFX}u}&+Gpq^#Au#hw%!C zF#*i}3+lY>sf8z}T8)2Rm#;H_K7?{e1=`(a4k314Vlbuw8dmMURAY;O+i(72{G4Bu z@Dd!9V5&b!fX(z7>_tex44X1U1`Fyv+o!%kn{Y*TilpNE;%AFmnqf$>R@^+Dr*79M z@Mr$0^YE)0qzIPtsm+Tile#}-C?2A3y+}3IG$Kz`y|pAYpi?EtZV_H2a;}V5Nfz=9 z>N4l?rYJ|LRbefSe>I}=bEv+2MK$-h_d7VR^qlhVKv8c%Y+11m@x zr2R`%XML-KeT~@Y&;JoFW}o3GgE3?`#H-Yrzw%7KoMO$*2I>W&Vwus&gu8Jw@o{ zV}Y#8_zB;-7W4<=1aWVwnH4Kpj9%gM)C!&J92uz47M>==m%1dD%O{OoS1bl}%3EeX ziA*O2fvLEmLsmaY2)qbFfU3Ouni5WY)yI$t(XIW_XDSYNWQ0C9Ci<4D3OoO6MgOTt ze&DxW0N1mLl2)-RPwwrFJCZ7_{4w*kiCkoBcp$jn4X%P{>lGrNMTGcB- zV&)zBB(di9*j-lo`x~flCVRLOQ1JneZGPIA`_7 zn>{yQrr!PuCLkK3uB?|5jA#H>K8|#8_)2Jk!^AIzFI!t%^=GyATIWqLb8doi>W!D5 zcaaFXD%}QWCt&Gybu_`(R)?5|{<@WOv>}YGRaHZ6aovHU1CCOmr23Oe#^P^>_f`* z(sFX2xWof2i`0C`2u~>s=J{285NbDcE4<~&l|E9-+u7OJ&5ifn`~wzCd^H!+4o%hT z=x$aB1ffj6v;;`8u3B-4Jphlci2IfAv+A#{-Y5U`r6BSvaS#;UPZSCk7KuK|YvVd}*?8Z)oSy9`2HZGG;cZxO3vqV)DWwwC|02~30aZL7 z5K|Io++WF$*F)@ik+WGa22S!rg0r@9m_1F2+Qdt!)%p5-q7{`I`tT=&YUHkK4DPkx zy|hPA&o?qqN4!JU^6j*Hv*r}D{;7|X)q71FXxBzU{01~!bLgq7HH93~e?K`bDJ>e% z9wRdRn)-fxiS-fkkXoOzCIML0Ul4&z9oR&_UeTSyT4t5{Q(@8vreh)V^pW1S%*T&A zmtsYFI_nO0q!;My@((T-%S8iUlHzVfKKPrOaN|nZXQNr>BCAJ4-Kw`)>sON}VUm^$ zZ#$l6GZ%p^u;%%8G+djYo~Cvba~7+y+}bPkyJ3@5NZ|>{g71kWbsoCQs~~`h5M61Q zbiAcISNF@LS_?X*8shbUb7>-ISMP<~L(f21cpw)RI zZkZuvw%4rePW}d+QsN^4(DH=9xA7*t>0?lGMiv^;-T8Y z6!1Ut)W)H)qKvvz#?y$U=eek#Fei1k2V6es!uQ%!)Wo=j<((Ybqt^sHlty~0aSEG^ zH+_+qS~0BZ(i1xL1sFaIA&nN3B{z=qMayLUaIa=L^I$q|a)U^kH6M5_lczvtEt6^5 z`?=x1`_8t9N`^OfW+T<4W|=NaO`V`i@(SkrqF3EmO5ia82J;4$qY2a2CR;Idxhs+B zOY@Au<4n#H{Pb<=D?50t1m!smIaO9G816T4*YoYo87eeo2%&DeY~b)A=ujo$QygTU?71@0bi&`+R(r-1EX} zO1>aq7SxiXtCq_xt9E{SA}dneAZ(6X6EhF4V9$PpmUo9}99DzoZl>Zews1-d zMT5tbO%KJy8;JLAT-8c)w$XC!ZrvDN-ei#)YXbf7aBf+*e#^p#ncfr=`XD-^RL6HC zF8Cqh6qB2%m~}Pj7%lau*InmDVCTaiU=v`nW5FHF^2!<4Zeh|>HpewfoI&w4kR!7t z6~&nja=&6SF-PJQSNr7j>U&twbnxhJ$Y$`Ua4)R3KUC(69wewosMM7-k@iuNe9XPF zF%_7EA*xNdtXzE!BH&zWugA0GM6%eEd<1SwA>YD`*AS4M@FWd_jQhSM1KI}oG*Vu} zn^0zeXA6HJULX1&NPd8mnmpEYR9pMF2?3NSMkbC7slz#U;$uM4J|?(7_OiHciNq`Q zXM17gQv!B0E?6QpdY_+J$fSHMe-$0}bzl`>BUc*3e7AZ0i8b+7mx|l71g~Q|$;Nr+ z_e=-3f@AV_jhS`$2@@J>UBkFonq#!~WGDL;o1JmVt)G9?lkin4SySD%x3tEaTA&nbAZ2Lb`~iaDZKE#gNEj>>k~>J=rh+j3Y~931ClE&tw*AaNoea zREs_#zu3^I|_QFheAgbt6ektm+5PTW{Z-#)~Vuk#0zokBUP*7Lx9E&G=8B= zFtT#fnx3#GemfYN6KVNGV8ruoq!MLJW;xbGg6$ppXypG>ZJmU5GZ)bHkBlQz)Agc7 zK@YNn^++?eA`!!7I!aqybS5Glg8^UTzr_$DTn}g&fdd;g9q^@Brum|Naj3noC2pF< zvmNHFOR|&ej8Q+phjs=?5m!eT1T`R^E7x6#M~6)x2ZUoH6* zh3(yc$b5gQ7|H>;3ITMD)__;f!*CzCS;x<(dZuLeyM5=h2ZdaySfbKW-Yp&mKKZ#* zXV*O*fLFyU*v#sIxxMnbo`R`SKOOA2Zp@E6q2lGkn2Cn%{vr7X%gN4yOTh4&6Mz&N z&(E4SvVaaR!?&c1r~$CW9p`+S=g0eTFEb+cIP@QX!>8f<4~oHV{GT~-12A~xhbgl8 z)r-g3^gUxUX1r9bz^UxaoHr-z+b@`(+&uQn$oBfN!cCOl#5#adZ^_@8{jybljyI8rcl8j zq66p=MfD9H;=}(E0O72R{%@2#zv{u|=0f>Z+X48{Qk@)<_phlRz-q2@V z;{2-R{04o^NJWM~z%BIg{wd+Gvi>TwyR14f{{1j(+GC-3)nisfr(@T{3ae5erq}AL z0a_{>s>V5Qyu>w|vr7WS~r_1vGsxpsUg_{9k zOLO%STez}J(}kCd2N7GMovSC2j}gqp_T`6?@~!4JhPPQ@hJWHk_qg_M(}O)kJ9`f((DeKsYBInW~RlY&!n2omvDBGNtQT; zM*}|URr>`UpX|!jk6H;lxdkm&4aDQ27xP;Mske@{#f}*;*JyA1kbHlNVy^Pa;fcx9 z;qO@vOh_k=*~I1N`%)pnMlm&0>%bNp+@Hs4zV^mNiG4H(UwSP@;CXhi5srw&PUpG8 zX8aa=Q7TuiL!2Ob{f_}RmnX*NRu+A;9Wb)`XJZ+?CL((y@>L|l=As~IXujZWqa<6y z=K=qC_wr)S0Y{O{rpP?(mC_V@cnfr2Yh6@0;allbvV&iH`Qvf(A{FtC-GkIA*(! zLdug<;yvw)I?M7ZXzv&#rbfw3HE|b&*3yKcUJFxU=xt1%{r!w#FY$-9TIC@7QvSR`Y@@_@Ub!mY9Wk1h2Bn{8P+5LGK^rU+J)XNYwVvj(k& z_QX=0FZ>iSlNAjnU&v3Kde@-#+#rVO2#B0?1Y+6w9nJ8 z;`ZEyhF5QEjr?}vDsJX|`_`J=xCdy@;PnBA75O}$Hsm>;m{0nLLgihDh(7D+aJV2| zS5StPl&DK`HxfcM^rn9cc75<0uk^s#P2U$WzBg%Vidk`qYWojFJ~jUci7Tpn18pS7 z>gxLQvFt#v&vsb(sjO9aTkXLpZ`fVUMeBVp4sE~%yTr)IO-+{8+(S>NVMWk6q$Q&? zu!cba=UBX^0yAzI1TtZN`TW|}SGQyVv9h9q7lZb1q2c$()?KqXs0PW=!;xr++^|m< z+EuNHe)jtE>IoRj9%cnqkg5nd8*BtLXpIS`JA2wO#9!SVAkY};PP=;@vkffKplaG| zlJ%w}Z$v2V1a7fMrLoLoC@e!}PPGh`O~F8O0C)5aYfR4fuSJO?%NWSf=9lYM6|v>= zGVK%}1O1UniNHHm&+_dFHi6Io_%M?Wf$5zj*ggaPdih0_5>UfEmP{9RZw*o6_EMw~ z{g++!A#7JWj+J5yr`7hsnV%dKtGh(F_z}Qimy&|1^$^R z+Sm~`*<|m=TYvv$x=^0qEF5FmB4pc;^Y?r+7=SFAMQZ}#Nlf{LM_9l|n%-nfZwr>X zGrs4+Ox|+9=$|R#q_^yBSwIaPl&&smbjS*w&xLj`{7Ja=N+2hS4%cmb#t31o6=Ca; ze~>Q*ZbTeF6h9P2nWwkERm#8T#(l{MMcy!7KEiCpU8}wagbT0Z(lF^OVfKRHBY3TU zYH&rMD}%2-m<`_%!BVR&PA`vr=gQna2`u$|?-p?PSq?+eN~pVSn8^0gZsK*(@;rRX z_Gi6N-3r9%<$L8s1N6kM&i_pGg{Z&?tK^1fJ6r!SOURl}M+n0)cnWMUt(IluT>hH3 z?LZ=ew?()ua(*Z>1hsEd#2mgw7TefKx2O|D7EApBjvJnFr+vI_zfp)b)S_|>2Ci2{ zC|*_p3me&_O^`hI&>R=)c94BoFYoOI+_#mRj^)L>9**!ZNJB!`#n1;R zZfzLrGZ3-PQ(Ltm-*|gBZ^ZSg5aKW38XRF}DY`5k2ls-ei7T*C&i-eA&v- zt5}B#&-3Tas`MR;Mi4QS@^j=2KF`JaNQW`chmY`9+a1vm)luIYx5AkyGTmvlon?rm)e=N*2mLdI(s<8qY^I_PG@}Th&&~I zH#~*P|C*JdV^2R|{>4}$%~mBnb<*b~^EELzJ`Ll*c$WN)T&ad7+@?;__7+8lKY*8xS-g*8>*-Y_GO19A7GhU{Ajx$9#oSiaL6`mO(X1V`t%=&Mi!N7 zT2ph*k)aUk%`b3^TG&>idtbM{hH}cG+Z@8{`tnVjvDyRo=Z0#pUSspHKEA36C{lWO z>aw(}IG%60XR2U~_RDcr4f=2=l~j0ymWzCVgZ&ouHiOMwlDU{$Gjrf?T6e_Q{W9>{ zIrYS?zRC#zRahB@oGw?j`*Bi)hA?5wRkNB8bV?2;4uk1ZS0rjA>vF8jMf77Rl7wHL zVR~VL+}-;ZxPi(~7FM@df&E?diuE;}5PwP(dc`}ULnWB$a+d1Y@W2kZ#4*3TFIYgn zee#4L^H^Ei#tHXE4m-*8hBhXe$>}p^MhDvblWhlnaR}VR*>0Xxcc8T2fovFKY&96D!re}eukl;u#9Z2o4-`n|T+pe~fpy&G<)0n?+^pVEJw+;of+c-52<;y9GXBRItj4e9o2;Qf&J`jp~qA`gYrMNP3774J|(yv<6LCjIB(>%X^nzoX&moM6lne?npF z^Qwtsh_GNMvVB_PO(S{6j#VLk`H$G@K z*AeaX7>4NQYIL7|O!?>!;3@B!8TjNL3t35Zb!NQHvSH7qQi@yp;xiYAr9qSy!v7yg zwFAK!+R2DHvsY|@AGN1y`(_9}jX0OE-e@T86$kogvrE{DwZP;z+_7Ssb&Z3q+IvIJ%4)-(wSNCt-<9CANhOaei zzhF7@oAbGUYcM~UCq>fiGv>hX4vk*KZ#hA<1sWx_C`a-?(U@tVg<(URLtu3E^QZ%{PJi_3@3U8>Tq%vR87{6n^9i>6jo$ zL*yrGGA+EBc0A?n#U!(_w{m+87nFOy!tjWWQG}am<2>pWcOdT2<3mB96vr$xf=4$g zq5X8Bnuhf_4l-T7zC5v}Lq5mj0IvxIF!D#Jy5LYQa9Ao!JNE%Y@Q|wZCh^8z>WW5| zg8w(%!_eVcZNnrDeXv+_+|UWXlIy6RSOD)zyPK`PB^v9Sw+;$IMtsFi@w@_#(n|Uu zH*xHTC*cDB(sjguy|iKd6| z$@nq+#5lmlQ)s|R;;n6Am3RfP0kBwsP}d;o=9<;iuw%uGj`Ko&t=V>U-y-JD1IV#Q z`or2YJ>+nN5HL~6cu1(MgF>KD?t>z~0@7^hFn~e4s;hZzw(d5#()r-kE{ks&wA8tH=-|T)>^ZUDt_VMI%m#hG> znGtHep>7Q)`&O<_uj}%Ys_mpxg%=wwb{+6p&==^Y&y$bkTpFZ_sBI|AjHJ&&R;ANl zd$_b-DF=(SQ9vr})`F{0<3FXZHL(*Be zHNm)Fe}JS&gLHREcX#(jE7FVxX+=U1Mvs!w%^0IQB&55$q#G#_P@Z?+-}?{j+O@pz z`<(MRdXbZ@$!Noy z2y{y!c4{xl8id_@rZDF25aQ6Z%q;c#NKMb*h%z2lycB=dYw&S7on~fNVYY__-Y&!qvzZbKSPqT6nZK#|+??e3f z#fG@A#dKNTS}I5f$<}Y6+r;8XpN=Ts%Gz>n#sTcrRi$`&*b~guO*!l|?!p3lh>rIN z&L*rynURi70$-ZiP+7`OPIBu)QbAhB%7^Pl%4LZ>5k*dMP$q9`1zWapKDnL+tEz~n|)|> zrv;Xlrb=`~=zO_ZL>%TUmMr$E^R%gV0W-!~0JH*k$1h56#C^oq4u2yPyq=#$J7Z`|Y@{gdQBV|4M#K!$jz$W%B zSaVV}gr3yo-QNbX4@cxLskz-5IcI<3D89>1dA=&@46eROy^(%jVmSV%2cuFI`gxu} z;`zo@x{J}M-##(;b9i;+pb`xXe_wjzG7@2T1c$gWJUe^HI zb`%|!J~^w`g{jP-dTCD(@iuwXi0lI#&6LI3UWx@upqWXU#d_x1W2K`}m0lue~v_dHNK0W2T~!vBn6 z72opgIf{XHp(`c6Ehc#nH{(Q_dlGXLiNAd`58?}7Cg5&_#U0`(qDl|_&GLiWTs<7T z3P?T^TS$39I)pnM;#^&-_py}o`4JzFu}i%ZrG>`R=TEk?yl7O@Z9$NzYb5ctbzuco z1ZWDoJFR}tDpBEI8Isp(q06b4P*raI&EDvrxYN}8u_8?S!`Nriyt2I7#wxM~X$GXT zaw%IUgv0Ba;inZXC{;8UhN;3-H;K*O=3xEitFfx?-E_I9O~#&Zea2}c@z4^am4B8c zX2JSDzzS8xj$;mVROz>24y=C0svU=kSgAaM6tkPuYA$`N(;c%c+Tpp}%K$!_dPLyL z5i;ymsaf6jAzNF7K9p_>sp#vAk7u;Gf$=%;pgK1Y$wb1TWjT@aSM0jkmuhVbGGCVf z>>=hHS`vA3UOmuRZ{u3By@D+_#z^wKYLB9emN(XF@`(Qcw_eww+<8OG5pcgzi+yO0 z;K=796#MH(T)}dKY$&a|_UOnAPE6sFAZKAKF-(JqzO&&;07BM`;d7qT!z~e-?7M;m zXTlaUJ%Mbs#Y-}?iX={GUS0ZH(fc0mg5YVTlIE}2F#X>`)ikbAz?q%8 zYLCv`Dc-ZQU{sC>wXo{$a2;;{&+(IAYH%}NGcu{)^uefa??+<8QKqdz8 zf^e|Bu{Hs)%}|Z9{1!)<04n7HJd;Rk@xleY)DuN_XHvjtsig>kVpNn1&)#siPIHgLb%eS0cv_e*qb)sJigsg-=n4oZTpQ3ejn34XY zX?eq~9)@Z@=uh*!_pb?Fn34r3Hv=n&1dVY!%#j}YI|8sYF_%!Pr1e#tBu}Nme~`|>t9Uxo=UA{pOR0PY>x+Xj%3mf@!Od5J|x>Wmt`g{b~gv9H94-2 z<$O57G)XmYi5fTHe#mPbCKm}->9}GyPkN_lYOEd(!1oHtf#rb>H`YIyV^0jGtRiV-tf5<;NYl_>0^I#du$==izvOa-j8mu1GmJfIZ zEqs1rc2q;}A>U)3gc+H?5*SHZ#TKs6hGa$S0}PqC#Yr@ufd(2Wayx4=N3vW8K&b4X zV(V|poq7|uhp2NtdK2IqUnap=F)T@5shw(zA@W8CX6c+e$q{*r2ucHSe~$TurUw-C z=-gs*jmfa=-Y(1=3W+<>6~)=J&o>GzGPfq(Q_$~H&Vn)9+YlB(Vus*dPzUX?>4pdJ zIzcC-(AVciAbL|^sO>)hd*#Wqt$;@C=1}Ssxqr`tVa2L=fL$c)+`te5mCW`@|ElZ~ zLLJGPQgCbdPuGl>*yjxQPNF1Z@33ayWG_Sbx$)`}2J%_REb-?)dA*&RGRW##rcmZW zF;Me;Nk}GM3SAHTr^WBOHN8?h#S4^Y)XJo;=W3*$^Sj=;?t&IS7kb&+fxT&qF>Q`J zMy>omu*lk<9=Z4bjwJv-;JrA!cDsYXnUwwyDjOSFpw+8K*th5=op;;w$QgU8uk zLt`XK2Vl+v=mqe`)bmMl>K4gvaEP5hn)|B6yUwdK@3yN*+xkaJvo0B!44-f|pQ(?C zD+nE-&ptCk9?_XqJv`opMxeO|(z$8C=O^3;Rhvz*5NQ1V-)Oy4?~ug${xkJ0lQUA! ze4`6bY<$PCE$|c-7dc(hb{@a-^hp2K9KH+?_>8wivBnoc|82nGtxLKVA7%Sw0USe} zh~UgSe&L)(-}#(qB5Wv!V1BNwB$da#2fNg*zmN=FRaGliQH@RMO?Fn~U;&z8I-YPV zUadcIDrDk&M}&N`mN*I;j>ExWXD*W74^_C~zM!1);Zc)u?~#?|AsFo$@mV63ru{dq zIpCU+$kxLMoY%J3p$V>orU!{7>vyTm@JsZR;#2Z*s@JxnUgd74vkesxV>xJ8(~Mh-8xSX*n}X=5puM5;HYhP?Tecu|CnS{)n#ME5MvaBMZ%rNOmh?mYNu1j`puEWVkLLjQ<_RSsOrGBmoWND zUP3ye;@eGFAO>oSE>b9}tk)IXl?gFcd9N?YvDs+AHgU`MFjO~msGHKOG!QFyJKvXR zLXBYUDCwcalk}s`i{iS_7=CFma1=0%a-rzB(+q@r_(|($Nu_P^@%|hWq7cN8Ml4`|%nIXeqY|xM?lZWZ5 zm&s+6K9=$H8jo;X*R>aq5T@CUza)~_TVK}bUJVqQ9hq_{Di!<9&c{+}zH9ceHBwN< zi$_^4rYC-_?*iqc{SRe^O-(NILz=cml{ zf=I2}=}>TKc|4I0THp!zFEsf}6np1`dIm@x)ijoKH&3)SX{BVHSwL2+!5{0IaD_JG z4f>nKk=%iWA?nD?^zboE9Zo)b=-^SNC6zYc$@pN!(H-oW$v3lp8AXTC&Erzr=uWQE zRW^W0>WKaNF5iib9wW|S--Nr$V{~n;iiX=3YSBdZ>swe@UU0b8MpLSl*m0cSk?M!v zDRF`1EO-FA6IV6C8@qD$F`>aC+PWb+*W^Wqu(zt{!wfPlA3+!8p}z)wdY(+>5HD_m z#zzk5`w*{CaH>=b&7*_~{#k5B%pS}b_ z1{{op(;7wMi^(k8I-&{F(NoR;1FS-+BEGx`0r@qu2d#{+-Ta#yuy$;2Q409`;|dAn z9MWCMZ%k`iUW_*Q!LLAWqe(A>!`#MjUb7kWp4vzs&DBR@-ZE%*(r#DUceWm`I^?r! z^hC|PAB2+n`7etTu~@?g9EPeaRXc>Z(A#pIteV?$Geol{1yg3LGss6RiKUL!mGWvT zFH?FjJc41vuG#*|?P{ALc+uOJ@?s_-3lsoa4A?!_ECi&U%_Vj*w!JV$Lj+MjjD1(LYwn@bsNV&Hpg z!p#9n`iw$eH_OaTA>Yk*3bA9BogQ0a^!>j2INpX{3y&$uO(@p$Z!?;`SXl#0q7-7b z!9p_UO^d(FCt&F<{_*e^T0H0{K_!y{m@15A7bK|`^i4mhsNdn6chIJGb39|so5*1e z|MH6te!lttKIlJD;JLS+H?mTw^e=wR>$L&)8nRUFCKUiCo#y(&$c9AnZhHI&toAo_ zTFIvZ4fLUtou7MPt=VzwdGAuE9lw7?(``2lRZm4-G{+KrQmW^zoKOUS%1TZDj#^xf zW!Mzdc3e}&M9nx6MQk`8gsStclzGgGj!@n10l7RjK@XZQfw8hvrvy9yW=KhV9L zgX2nsf?w@&6ov+_!&k|Rl)KX{g)Raof z8#E2rw-|iKERbAAoA{%Y57~heSxc-uUpJ^fC-Rj2%`Pfl0U#akkl;P}@w%=kA$@!; z^K+Llh3||nP$eP4YCG7rTqeCgE2z2Fesx7fhSI!7D|DF_ z#_?q7dly{#=_}0Z4Q7d~IEJrl9v>o!+c`ON2F4PlV`*-rCA*^j6r0dG*gF^$q- z0byTudSgTz*E@4((ZVdEH+YuSXb5x`T4|~oSp`wkbJ~Y63XdMVz?+iz&J*VN|KX%T zdjE7utSCqQu76LTJyWO`MuBJyy+M0Cw|B z%_C+EqFM5NpGBQHkFc+6q>ur8iLFN3KB7hvsHd1^Ku{K?!|=J3kTKrJ`_GbefD}m` z=Wgxa3YX@q{6pZ!IgU_yJRdhbvm5Qq)`t!Zy|Xo?!_9Q$q8Hhy#^o%+%F2UACET=) zsuEH8>KWMhe)#0<2YJyK*!09Ck8ep+gv1ifj< zM3kHH$M3j`ud7Z##lOT?W+ug)9nEjRhZh6*6avVZl;&aYu{^M5HAr*?=WwYO&XHh2 zmr`qZ3LDN4NE)*zEVX(q-ap`kX>S2NEEsXRV{dY|W(g0lX?p~zBHJ~3vjgbb$g9|+ z$wwFok=$^%xTt%Fgr3~J9Ej4;70#i!i*DN-tr0z04Qoq!{FdO|2`xIs6 zvJ)*+XM!qcAFr;hNth)*69loeHsw?47lwj46z^~r&s?(q>>JG_n?5b5?HVR1Y$cc6 zzSaB-Zy|M$;ngsE#KS1{6naSAz5?89U zk!sSol4+UpeO;Bh>^xoe4P&gyYmw7Vd=^@{#r|ipU-CsW_D?&O;?HAW)*X>KHpmTB zywjoIZdd;|t@|O))e|gM!23xU>QZ>%qYzV5V1=TGyRvr816BWy^YUdZGF&7+z{^?_ zuBdaI;FYHQS86hgwy1y!4ja9*#NR2@=e&5OEo8jha4ntr#**meq4W7-#pshI+JqI< zsg()PQv#cmm}8Mn3ikH@^JTjo*ZHK)Eol}R_iczBK3D<4AT?Yk6p4yeczcivN#*A& zEtF}`7Cws-zKk@O5=f&>M-59BCk}Kzr={Biox?F@ik?KKjQN z#K-lZ(X0J4}ir4+Ynok%{7!lpz4x_OE_2}H~Txhj1S zAx6^2g$lkLu|u~fZGHu2~I6?9%Gw*GgJJ z8^0joLZ+Omj*+Ntb8Aa4g&ia-n#;boN@X*i7!NC&rN}XGE4&9z39%bb;nyS=cTZC4 zJgt|{yacIa3M(_-ymLNQO!XoB;++LK87kazVG_d9&xy^tI;`tku&3m%sl~kXT z>OU#gytWVdB{7LZHZZr7(0SlKu}nIP%!t5wz!SW!EDC7%s@+TSMA<}f{{!&HP+zwB zlk5jFguy?;&Hk>S061TYUTuJl#`2A=uJhWdOUp{|veknFt?uCDk|X8Yc94nT4q`$u zO3jlj-+nYU*ONB%BC?02dtrmal@rr;H0X|;i3M2}Q)}!Du?*xK0dsGEz6J>VJvOtC zAEQJmV_!EzjLvSb&TfDnWfVWM034f@+>aR3^+%TRHYTsdy)4)*Up#{B?^BqApgR4* zjKcnT9R8#ZAm(=mU(HoKdUp(w0LTu}9H~6JpD2&do z?Q-%kEn63iBs-J!Ev!9rz8MBTmI|t#j#ky* z9FDNid-dA@{alN_TWBdhqnhR(Po0gl*oE`TmF0@v-Q6_&go5!wyzOb3*oPyirJudw zAo^m9|KYE$*rMdc0S@Ig3Sx{fGl~a}q?^mWdTGuN-TgpZeQET{L*+h!;&^YU9%4x+ z$73TimvKvp$ecNE@Z=+8fgHe%_(2-0L@rjt0VHA@18qk1@{kb|c}{ zLiv5}jnS0VqH1=#DGx5T+GtV!u++8u=Pt{R3=FCqF^O#u7)&xrNK1HgY@rUbNZ|(- z5(}|}XE5k2rniy&E%+r3EB=j~Xc;jGNkw6v5Xt061ir^8HYeD(9YUyt%1YofjK&L0 z?gK)MD&z4nj0^g5SE^`g;qz%FCYk@tWfnukGjsfqLB+hS=nd8GD5o3Br3S=GSQGM4 zQ``9=A7#J3Tk!D-&7%GXZ*>S>_?gnRkN#r&&U}Di@JW1W?A}KoOe>9A!TD!VcRsQ^ zN>mC*t2yI}+1?#{e$Efqz|oQ^Ry^hL^Ihead+AmTDx~!O3wQPPPY78gq*{|9%2jWz zo^82o90lJh(S-aefn*iD=|Q2R8;Z;V>Lec**GD|;=Z=hp)w*o)KWU;aH75F!9^^pv zS~$a)ao@ZxNAlkoZwnrh^eo~R^L#XteL28b@}Cl(vrt~)TZBH`C0^rWbxEo|Zd;38 z;+@l|@yI`J=~$(*g@0Uc&j3?R=_dCzO0#@@W%fGg5!DkvB5uD;0`O8sp|D9+Rk~%MZshc0Sr1~dQ!0U@SeDWn(P(l# z9vL1Eq(+rAK{pr5riRD?s|=w+&hg742AwdoG;L`Xq|{9%%3>WT7O(g{ZCX3HUnvf;M6+!OFA2UVDV1076y2eZi8b z%DQUrMzu#Z1%@q(HpSrY-HXaJWW6TdmAfM_GMfGlbk}(q{@ZVj-Xs{WzkA*jv@ye{ zU_5y31755P`f&f<2;u&=RGl1e-G|&j=dkXG=R4krCL$*_qgl;AlkKk}L=5-$LG|Y5 z34>FAEdS-<`B)}xs^BPkP>%?@&8 zA#MO^fJH~S$UA3xA83<~Em9#;MEiaAm5j%71PL}0Y<3*Jk8cT^nzIjMR0_JSh7D@` zv|jUl>+T|PlmH(Pha;UxEPIb!rOg(orF~+yNe4{k{q7RQRIM10zOifrf6E!`roep} zbS+aXnhW)fnsKrtYzG)E7IrSZwe-qK?=8FogMHD`6DLtN=UB#+RQmU{wjC_2w~>P0 z&GqUPfqLXEzb10D2GRqrKW~A#z7wRE{X(Y23+QL1{0G1Q^BkORt&aZNItWgC#ys-Q zC!wtF^1yyqU-KNj@I%_>8ugT?c3fEcDW3D~-~D@a8g(dx@yFbhkc}g*aI*x;`z_s3@C(3~!`s?FdWA)pl}rl9f=H{i4856tgVjI;X?U zaEZN|#`UF=&fFnGUkRxd5X^B==8n^~JhJ~}zM8#a@_}lnj=zLYP_c*BP7(z0bQ%D=o|VH5}HVE5?|`LMyN*IOoSa2 zDt={w$Ww0j7b0(qT9@B5?oh{@y}^%7#;8PYF!a8Lluq7B z2hed1wJT@M!(GM zFvt@^)yv?qc8sO&59PH|iKJSWFtRltgk~qY0w-FG#mCx;Hmxfo{F8hzf$JVCabl$N z=-jX#n)^&63OSSmG6B>%|AJ-gOIQdd#-ky%X|RX5AzMIQ!|!ZM$`KPP47kj5)5HmN z+}qLRL1_Z+PU!20GN>BIA4O5yz7_j4nNqvh;QjVzzd1a^@=ICsv`Wb*#QWq`p&{t< z)O%RD|M*u=#Z%aFS4ZQEDy>5GP{}G(DCzLtPR6entEFV0!#^I(+Ngs<>Pig#RG0m_ z$2o$1z#(^#d;irflze*JsI?N+&B9*ts{3Ml&H&vD?QeGiWxT$p zSC{KCskIc=5Q}_VXlz~&MD?h-@hT-N-Q51qsw7-hl<;1+nJ*!6S_sXvueqohM@V66 zHiNPmZRrmFEy?h$wN|3eWt1oxhaNy4$sLami*GOYoGpTFf4PvC`jX9mOoDQzCUs)( zn+LpJapUAT&A+N}YI9*-E@yp&O|^l{1qfxYP+upLWBziPlkZ<{P8y2ck}_q(j}2LG z5J8zL-9=Xhee{w66(jc(7O27e=Yjjw@oG!GB}?u|OY4KC^S$_39;CQbZdOw4>iUh5 zp0n9}&KdS;qygZ6-En?jnb4E!}}R1I(~BM zoQs#+lw!TDVRs}^<+7Jy3)ED-1I)Ly_{u4;9@dsN)Mofld!r1LJkI>86YMJa!u4!( zBYwpFCE*JRZ&U9Dp&C=A=82mUc6eXGNm}I}W;msNCh|%aWk!9b=Q1$%Cx2Ws(XRIL92 zXm$`fK!*AJn^VvMgOl`&(y66V#b&;>94ya|)U8cz+26;juUmUmd;bU0#va8_$oMPY zmyenG?J3Kj&qq^OO9Os$jKa2E879Gp6DNZPuJQM& zTrh`Kx%}197cF&hpi9#NqS?q@owyP>yn1`E-nq&LvG^Q47ufoJ^_pH|(%T|IsKlMT zh&yW6UjS2;ZBA>`%mcM_AvezvG!m=Q_#l0l=ay~iJ}81C+GQ@Bx$Z(i)WKp$F*>d2 z32!WFIk(Q8!GOnNlj^O2z17o=}AJe*I|dSEvkQ#COKq_-3A0!_|=jy zk>O@hUiaUU(}m{@)|LbUe7sQP>zC`EnihlWjv{V*PO}B=ZF<^7pSj~+;Su+E8UxdV z_;@RyH!4{$L%x)3Ms^N<3#ZNr!7v4UFenTyfg#;))<;1yszlGc1mKw!3_IW9kW?7^ zU7dGgCbip-h}mJlEoJ3pxr+M8TwD2DD@lbqIDJ|De}E=R`Q&1W)7fg`Y*U@ARz6%` zuX@UJCgsTcHPPVBNf&@?7Djgg270!mRSoxR(PFCi_lg`>C|%zfQvTwOHF}+qYQXF# zsAyfC*JUj-F@5;rUWy{hGPWcDmvW@+rMle>F|0EM;Df32n(X1ch`bsES>{AA0ZNC- z7C+)mG>>B06vN^IWlwMqdw~b6;u|_>n8^8bRc+$3I~@bnT92Yp0wp!ewgv*IW56R9*4qQFS0^ZlX8s_V`Xq zqM6>j^O>-Nh!GR=OAIqgNhylL5a^)U|ba|Lmbp_HUOKR&^Ps7yrf=NVy*HCJSJd$0K!d+z(0d<3P%(*@dBr z6wRzVEOXvgLgM}GwJudeCQ*#5$z#EhgDkRX1^JM%$DA4F0cQcg;j#H^y;Mm$>`+&Y z0gWup{)KE4H5$hE`R}w4GFr^S8Ufbp?xYi%FI{*TnWU6h(- zp~tEkUBuS2At~&rHW8`1 z0GqF8UHZc@u~8R4nz0Hy{j8;=;GQoUTRcj9yLWW1uT3>!c61Gr@%Rl+HK*p>+`gNG z)y46ic3RS>t{aBM_yjy!E{kj6!B&oGQ@XOZ%1%5~I4`_&XctZq*$vOh_PA6}2$c7^ z^C)ET$bq`u>WDRSe@kSDJ{Qw=bGXGQp{g>n4W*1vhQz_=gTk9bAXSs)`zN33O!dBU zks^JKETf$J4V!PKQy?D{cjIITM-zl5MkxkoLr~-<;|&((7Y|ABgWQD&Lm|)1ZwNXr_}lt>J^$YQj)elErUMJpsUd)yYWKbYxETgh+zHhY6a1W-RO` ziR3U0sLQV!=_K<0E#pc(uIs|Fm%C8zg!Ar;ZZC^v`O8)c2lB%kd zNmeG@jYU`rF6$BD5_zau>Xb*Pf4MrvR(n}T&_%VArZv^vV1?Ibt)io+T1ph|zJKxA z4yw5Wr}3NG7(p_j;%hW%^t5*)_m!qN@SeSVi;>}luPd`Lw@N!m#IcTU_+^TzOw4x|r?$_0?7h7fp z=P62RH>3u&Dh}H!76*5(5~tp3Rww>YbnK&Sz0m!T0iy+7t;J#>F8I74iqzFC+y4>ITvZed#0G%V~XAL1zDGJhWx>@Ya#*xOmDfjUzOU(uE~{t6jbFmH748$(tq1&toEw#@-=^v zk!AzXJhvu;*5$0Vm3t@-anK*ct)!6g`{+usUGKht>IxZl=^4vgWb?ZwoyG8H4&zDe zRA@78T*Fi4H!5GhJEdW(mBy=~+kG?2O8jPts`{HzbH!o0uA6nB1y>GZhnDxRTKDeJQJstT=0ffv!!YZ&mRX&RY(l}Xju+GB z<&Ub9X3wHYZse3AG1e!t3Rw)&=;6zt>Kw~`ODcF1LZpeNIgqmNwDG=Y!f2)n6{+j7 zce(@x^ERZ=euSNVKrT`+i?c|atLks`&8)Lrj-CyP)escMvir%Zi?Wvo@};>u`lw?j zf|09J{6(_}VcK7q>Y_Q<$&R4i#V@H1sidAJDY@Jel0t3UHzq5;4_%Nh$Z9h;6Y0=& zz$^#LKVwrJ$BYAi$5*IA_|%a~HYwQg+i50FW7tGjlPqB1gHqA*MAi$m>A6(Kg&5wq znqK{!cd|U7Emh1xO3mA2Q|^?+pRVauEcsr3SWRr*tzR%V6_HeKyLI(9IC})?lOqy& zdlV;x08=Sw#icveTx`Sp(h0a0KEyu{X+4?x^HiI&m&#FAvnrn;Ym=1*uUcZbx?XH8 zIS^o)6^E3*E-zeDx0t7l|8LV@u?mtyN+LWebq>dOoCT) zfetNLilJ*zw%#;~Y=N~m?{k_Q-fI7bcl&p+&|!fmCU!S+Q`CrM0hQJG@A!S5vqX-E z(l;)2_I>2}Sm@9=mdIr&rdR;Q`S0~xnyuFk zmT={ zt0#HR*#=hB7>_*3c637NO9(LehE{4{^=F|vEkXH9WUurlR~h(mKJ(5u?=`n+&4i?5 zrXe|oyFokERf7E`HXZjeR|c#tubp_l=p8bb2F+rw5N#g92S|K^Non{rF;^8+Xqif7 z*Gg52EVI*G%MGb-S|G8rj=dcC!JHFj5z--k>)V=K&o}sZx~$bLT-MzzE;ZZQM#BUP zyWwWDXMQeQ7#xhKwiBAQUgonPQ;Z@;%1T_WjeAYm+E?5rywxVB*2_(-^*RDmQMbe8 zJ>hb0p(uR{55=r6c{88?tD-DTEk9835we!?$inYY(y~;bUMu+GDaB+53Xpt=JsqAp z1$GBX|BW&G4^S`vc=vwPTt%i&=Eq;g9LS2IsyS7gq;##ErPvPo8ympX5x8SR`unh9 z(qTSoull7qIW50!UalC#Dkz>tj(ZuVYZ0ln@NfvB#Os$qPr#J_*I$`H5_rhg!(QKH<-p9i8Y`Ra7&HMOXlkKqV?@BHBv(qw%96&~|e zo_sP&K5TQv5L~VB3Ar&OO>CViMD4&PzqI8P5`v$MQ{73)Qs4A?QtQ>F9BAn4Ypq+F z#46&o{U%Q>FKM2%&phxwpKUVIU_T7#z%j8<(uD3jPH#UIWFmVsnloB9_f5}^kt(UHASR#p*jq*TH zA8Vp+IAt!SBe}}bVO`Of8voOWB}&|lz-MyQ3%~ejeK^HUPx#Mgtr#{A@5w6= zN&%2o+hA?;o#w1K?C8k3MC;4TA^yi*+Unw=kU+ML5}t0VphQp&h36yG11u>U#tG`S zrdX~|k$xP<_8n$#*OfZZ{&wFzT3i;gC?ocY<7IL&{GiF5sW~Q!-})G`TSCQsrxO!Op}>m)|@T9j&@`GVWGF9>+e| zn$7X?<*s`H<*8FXT)w%l|8DX*aQ^r^5Q_-?5w?}ZW*N%o)S$>1BXFm5fD@WpiqXu4?c->#&Irn7bDqM^`B0=cI6CFqux ziuvQV(Gb|){hDjY^9Htpve<{3^gNoHw&^*O)ZJK%u^6 zlR3vwzG}F0H3k=6`D$HY9Cf9*Fp_su1Oq^N0zPhb^8wG8ggHS~ovgIIrGF}#Fd*$F z4<>?43>M=vS4&dQO&PxoX4$&%ZgYy;rEVhMf7?bgmb~Q_I8i5y*B@%fpnIenbZj*@UTFR~kR zQ=w>^(nMiXvc3L|I3pc~i(gz3MiLf46G1=$+6lO134MJV1)$f}T6aA|auwV^xwhbDn^EE$%eF?RkXd`fBtEP}+j;?31s~ zLI$YLMy7S;D-sl>0~)`7ZTRrT?}fM3z{Utj+A+xNmh~5Bxi-vHDfg|3sq1U=U#*TD z@xe=~#V!Q(K`p%+LwI_w_`JO;Rl6JoSWUd5=I`HS#R=X?R=2NTh{aCm4rratE&x^B z7~}A;Erw^$L9WqnQWJwgZP1Xkhu$9qpC`FeEMNMu^2@s7$eFdmv^UeoN>#jns>x5y zB6#I|>4xOfX*##k(byZS&UL(P;1iDj2H>R^M$zk8N}Ri>w!tOKjN`p9|9P+ex}3Zw z9=gNC^eD~suRvAj7b4eA;ha$vcu230-2}{~2XWy0o?FJzx~;1SpPHKLImXZ-Cj3Gi=%cx_Q$~K2>v}V46xj}~ zcDQq>TUy#U88Ebx%VmKxdg^pwdDkb~YJHTFXSzO%{aVHQmNaw4DB9wPjA8SCpx>T5{T*5?Os6P97K|z9YM&>2vZKL*!60}`-eA>1k$X>1J{@`k z2m$hx!a)7@Q3z6Pzc=ggocEP*(dkSQJCyN&=~JJ6p@pSp7@CP)2a3)67}b~*ToJ&~ z*N%uQ(wmP*dC&vd9EIF^<(lTSiY*$b24ug?okQ92HvmUqS!E+tUt0g;yzB(6=DGUR zR!;i#IDZvAuabO=RyHo{tdZ8i;o2gc5r5XYGR@a{FFoO+#Nr!xdgJX>v>t{nC9Iv1Y{{^2rn4`|a`A+rYf%FYJBfP8N;7l=8h|;Vc0o8<|=Gb56uVH28K9IOcp( zE&?h&Z;Ic zBVEtL=i2&Por|^rS~$jPBl6s6!Ta^gON+jVZ$2G>-<2dd`)yAGj*{Ew>=+`sq6OA#nuXXfoZDt71ie}JwVr{_lz50HjjN?uCPUe{TdKRBDh|UlD;vXfpNF+1#b;HeA?G2CYs3;QT}8R<6e;!O{QMCf#{^hp1u!yAK+xLe8KMAQy=5Y?8`Tn5} z#j&!#IeKLx-lh}Hjdg1%ld0tr zI$S2zY8ZnAdm6Gx|7C2RC8llaoqjfo$_BD);hLPbT*fd`suCd94ncQrZOtGe?n_ck z23w)L&4!xRDH1GD8u*9J;$dA}?V@oQxQqBW)shxKfE&aIH;B?{|{9*HK z)X4hf;J%^%pnpxl5URoJRdR|$(iM^ZT}9Z&!q**D38A5Tz{bBb{KZi&V*ioQ5$1t| z(KU5LRFjva)h{%;R9s3&L z{R)Nuo8Z!@sNV08_8XqMAxBb2P7_x12ysn%{{!FhPECL2-a!N)q+BFmI0IfNkpFym?OKOl$si?1{qQ;SutlDy;cTH%Z?`;)Vq ze~^sOg>MrrMO>A-@ zr}|Hzv93|5P7A`e5t3yhQw5a8t#F)YPVgU8#Hi8Qk3X!J%CYO<{+;?MYuBrD18ZR{Ht*S3N7wt}}lCA|_ozBl%u?=jr zq5AH@5+mg8MGG}W5+}Oz!T4R>vDt250Vz^g3EhDNMpDJlbp_wf8>UbBZ|p3bt6u{c z7TeFM!Na4(J~a5a6f_Lt)4&MYYK7+y%e#k%jA*yzW8V07P*l7_w|Vm00IN~Pvbd1b z4g9PqDF8QF)?2rUFvzbXH?nrZo!6Bthy^`eD+iIDNYv35_ldXqkBUw4Q(L*YusX35 zn-xmn43e=wkE1_?z=c~C;2j+B-b#zXzF@7rDYTRkC90&i>6CNh@rqXYR=~9Yij3?% z8Z{yF1c$e-5q5j4`B|;Q1H^|OGh}UiLDc@g>&)SH@&9y<+FDhmzi~e9e5)7nj$~r% zSO{_`#P4Jbr4hB`wxnG}$YVNTTo7 z4uHOleP+Tld(7hsVZRGzYKJ>FIsO}LaQ8s}ioY2FeWjtdyN~k2)_xz!E{zfzYU|yp z0vphT_yLpQzeWDVi)j15ahIJq_O#7V`I77tdW7`*!mEIm1m2Yq^kD54k{>7X(vw>9 z?ttv}4;W{1OF55B%8ZA(E2G@^rH&gz*vE<>-*_V7#Ue}=TeOj?0BRqPjh8qAtz{a2 zB*?N`4+_$Tf8faH5+|&b4Tnq(lG)D=P9&zQW=;Rh$0~)W4hA@FL1um^nS8Y#6XZtu z8D;Xd#C`%;ow|&nx3L;jIMUlU`kLck+_0|@doHBo*$lk+KNv&Pb(AG+y%g>z+{$(e zY>eMB2g3|27o#NDZ5w?(o));*r1{OB{k#Xa+HwD$G>y-)s%-y101rX%zKYdJy^P#a z+^KMj8j!7qJ?bbDC}Sk~lDY0dt5()xb@OH*XR*ah@>^Pf7X`WuRXP2XN{3cajt7&H zxE}ui{;IN(%%cbf-#{wt6Gp%sBT9L>8@i|A;~_%td^2&jo=SK zS+{a6v+W$8VNRTG>?zAaXTL;TBzQjcTuNM-B{)&! z;(c*ZwWO1P#R4y&sXpBr97v^)qSMRX&5tV3s&*{Zs!b5=%FS(U2<8zf_dI_($G@}t zM7z>3o{QSL7LCz>lN*m*R8nctpP$Sw=lDiH!nB0cUCiYRs~mjrT|t9x;Te==;w2eh8(=C){^a`UP7w&{{T6vkxeDt z#@v#=)oR~Gf>rZwB$3ajBDAe-VvOP_zyh&zgjSoIC`wz8w6+j{NX1+wc9Xb$Dh7$z z^x~e0wAU=1^|-4~m02NHGbc4>I5UAyi3#n2Q}BYkyiP60L@YiRo8FKr2*m}!sNO&Y zK!x487!~2x#Xf69^W$7-^jg>oucQdwUMFjXCKP7W2gbl4TRi9z~R zgtI^}pi`rpcU2_RqHb_gZDlD;xXuq^291@1p#W6N5XUDpff`vekC)b`RSD=T3YFs> zI+UAz2(lwm5r%y#b;$#80Gg}hlPQ#zC$iO$2wqRjYmXTysNHT=CgjI9HXd=uTBzw9 zaqiq()YYOgm^GPmA>=+l_NRg7lND(wgwlTOlsMwK^0OJqRNQLx*Ge+E;-iVrP;e{U zrnyngDJc*&)R2rg3VkZwjl;8KuoaJVjV;c1lUHroU{YL~(otNbsXLtt-6&idV3H{f zb293?CddVs)7G{0>%^4|lf)oV({*#oi;B6tYV$i2+)Er_kTLFRlT0S#J4dZi)-Pds z$CEbY9-S*bJy}%#@Dc~5Z3->M=5wg8V{xtpyi9YP3g)h_TIN!qahiS3q|+;Ggzfb; zR?=igCI0}GbW^9xtK8?D@1Z13JZA`4A*&B^@uoKO>S`G;H$TWImQstk#%k?-(KoPy zLnDI5wQV9ZU<$)oRKDimR?Uu~Zde4}Q&BXuE)8E(Ry#nQ6cbkMt$f$Ssy5bl_NfiE zB4h7b_7b}QS~3Z)DxB$EoiTB}k!CvxlkO%5dM+YgB&w*ZB1suY^6}|PB4VH#Q;#XJF6LtgXBDP(FgR3`(byPbvVd9ahAmjBA7|!35QK?IT0x4ZF{5R=moD z5^A&&rt`&X8R%yyd&FrB?UE2<-ms>#j%(K36gWJZ>xT=p3d_FJt=Kcl<#XH`Db=&o zbBeJZpQj=zV!+i6UU^`*l*NSrb?;YgirYoD7Rq{)O}c3AVljxz6V|kgeD`{g+HtZt zSaliXmlC_P6UQ~zM4F4+Cg~zr@%KRI6x7k6QMkqfcBEJ}2wb#LZcf=X(Nd&m@aSa~ zDDHLoGVRh}7{}iK0MLyTx;c>PP?K4_BHMP z+t(`vqjhdQ1$Ms>&>s;%$CLI&ANc#%J!fqt%%M?Fxy5}}4yP^_9*2vFSFIv6v-v@G z#(F5^nzsrqvBE+(_br2p+KWt%$|F3|e(?R#nxxSAl4NpEVTzw_ru>1y$8n1a2=~h& zEP24s6wnhtDEvwP0P9t>vMMs;9)lE(77jD~#ZNj15*+6M4E|L#QJm$t_Q9#5fHA>g z`GZe*mNUW09>Sr`Y=>fGc;b&IuVIoY1g{~^0*_pSP>0#hKRk+-E%P=zeQOr2t1U~R zDr_0s8Dr~;k~kNBT>R&@PZXtMhl7m$1w!+&;t~*dY#i2ly^R%?eJg-bE&+f6)!AGf3IV29laR7Q`Q%p|H zm&%W!6yVC8F`r%E<5cWMsg{fW=&VbO@J=chXziiFW4!m~pf@|dRXO|ts2&MseD?Yp zQHOE4k-UOO>&M~!MM{pV_f%uoAN_i%Jc8ei*>6q8slCaQ@3+h8#dIihEeyG=&BK0U zTqAAzWL9nFp}656EPLnw0IgZ_PFw^;ZA`bL&B5v6&z6)Ab(oewzp*f3^f8K+e?tc6M0 z-H+5>8*#CIXFYhVd>LQ=0Iq7&DV{>|DFd*_{{UX67S4;i%i6yA6yVeqx{gWB`4HQ} zsCgM4oK?@ZG(!=pui^gy*Qjt*a(`Nq2ogcL*a7c@Q!71LQPV_A8_#vbk~TdCMO^zT zss&M#-zJ}70_2=wN3~vZsyO1gXHw3{>4Yw>%E=s1Al$%X&{j^Tsr=qmxF8O#gHq}j zYG2Ajli#IKTWGVoQ}cRawWS5|NXo3=!x7dkDRSFMW89NlDGu?6BB)!2bG9_Xj{Me? zP>?|+`;uyFOH!#V3F5X$*X08q!kH}lTVOkJ)YMm3URel->S};9&cpy!JF)6(DAR9Q z6IP@6AWtpjwl*`i&*Iz3YQ?#^5=!1wWG`-&dV5v~N#RF&rS>;-@~_M5#dT7QBdN_Q zQtCku$(Aa2el;YL%P0ZM6&K18OSln&_j~<*> zf+=7Mjic#RA%ZYQLpxT2M8_X4KPt+kk+nFCb|4%A$MLEoPoDO2WFceE<6Eg}0t}vM zepr_Z3H<97R^JJ?vcBVOB8J8tiHfOzdn1jJ4Ckrj0a8g6lVzBNLDwd@ZAVwSYxYYc zltX|Ra8IuWu8l8gEAc+~lJ#5b#8Rt}=S zx?i)t=46^}y`{O9clo;tj?(BPL`H`mrySLb=!}!3OvTAPFn`9hBeGb~1p(jCQ0A0< z?L(69Lbb1+?YTlA>yo_x06KZFlJarBMqoaot)wl{z#iOIm84O|Gsb|s0qV!CZFKcW!t6|y3<|-)PxZra`sU~d*qP$i~)TsPwbhiq47^&1@d80Wso113J zDGlKTSC!cPJ?hZ_eQAXQFeZ{?Xo@(C4B$~TF>U$zt5^Va#Z{6oF^|p1wKXfSwuWTV zGd{>c0D5Mi{?7B58A<;DKZR?oDG2G!Chgo~6|$>ugogz8D??`$xc%74{3%&kj!8dC zbgUy74bRrI=D86ZmL&cZ<9TXJYC0N`L{B}bC4Vdv*Vt8iONnL!f}*X-32*>44O!}1 zcUl$JcV)>0RSUbRSQLR=x2OiSS|milBQ+9Rs0674@~VuYwXrTOl`NqlL7k5O077eB z1bD!~tZ71qW9G3wgEZB@^02G8nkN@MP(`TdZabuENX2K%eYYejHC3+Jzt$-O&~sGo zXKkeY#rMT1QtlF;VQ#fajzQXs{(Y*VT_l8-+TUDOZNvH6MoGfakZ>No(N)jm~v zuqLu?+@-MrjUorjLQmGSE?{FRnob9yt!I)}KOt2WQgG?=q%n5318k%Hsp zTn}7S5=3KTBRd>+t48acr2uZisOm4}fYLYs_TsmUTV_(Vi_kRAxk9R^x$9RC-2gUC zV%X{t$l^YBiq_sFz<~}snu)=usZf$=W82(IAOVq6y<3XPIEZcAnwSKIcBVIalTn@U zU`aCWH4g7Y%F$|EZJ=Y5R3@~JOpV(!)QXDpP_wgyv(LUx$_-@2tO+tqZb|6~cs{`%_KeCnPpcYPW9!%z@obdjng1I>^TS+C+%~ z5qzg7)YO)c$_XtmCanu!FyYr~(FN>}wMK?#j|djyx78 z+O1#S&ol-&0o-`{R%{v!Y(v}`B$Kat+MH6>Y{sOd-Jp%@X)W3nCg2i#X0)u+EAsLy zCd*HX2HZ{uU`=#(>Z=d}xob4GH>8`~PcI9TRkn@zUQTMo%islGric^_R~<@@5V@UH zs!c$d2E+rsM=9fqd!3^-U7sh~zHb#*TQkz93oQmJvp+p5U~$r#9Xrw~4EP-@hCc4b zj_k4y22EF%Offri#}#8AEkQFXq0d_SEEUt(;KbhAGb-{@7lW2HHI26Dx5zS0RJwts zmm!E1sc!Md3a0FkE8LT`ERK9U9q%Ds6ea?Shsxt504iB0jy_n5sWs>*O6L`vRGHM; z5oF&XR6=pwRcoicGKk=fg7+q@Sl(e+w-qYfM|pANEQ+M^PI;?x^ZhPJj{yOn<)wL$KMtlMSVzC(~jaZ#)7YBq>;oDxj3>LZLi zgKyHRwbC|7#TODRstl8GJ5~f&CT(CxljSMuGtGC?r_7sUsMJ=6VFUi{bHJsXJcI-| z9nEI6@<$>^0pRrw(x!&y;hzRCsqJ2MOgy7LbSGJI@-p%+0NhD5{=KQOwnsS6r73U^ zHRn`Plho^jjAAL|;<^2H^g%Mhzun2jXkTAV7%6WFE#2Fou2T6Xo<^Ec$vm3%a{MHe zx#n`>Dye(1B56Y%#p=Wg@}Xc((cZH13C1uz>qgzvV#FUA>x%Z|th6~|^&^dx+-(51 zD7s%cw(3CB7ITBx3euEZrOhH{(pE&0YADcoNfLwIk6N!7SphAa`&11dATKo} zXJ-bwIZ|2~UA@Pbk(x#5baLYd0WO6FDMhP{328r>%twh^f4J61?&_D~4NEHRd zqE0&s)t*K2_Nw))i-ltRE(J^--k~Ix!~~3B3eeM+X*pV|ihvAqYoXFCJjle3!;{jA zvr&lPuFV?@25_!M0rsuvU771~am zEjXron&PR5P}6biCGL=&go5aIQnHf2dZpT+bNbY6woW@yOA#uL$q?e>)X$lY4mrn4 zNMW39Wl%}&R$QR@LTBEx^=(p15{`?XY~WX;9%*S4noZo>RCSN%z&-0HT$U?qUCcpt zI8mNWXB|pQn?hP<-oraVta;$oZzV=%`B$cDso~#0H05BaQcYSY*|D>GtUCgU*A=lO z5L-jL&{P(2ymo3F{J?WfxRyJB0bQ!2oSwAhCn;&Mq*o%fjEkFzB)JV6HqdJHH=bz` zhTMCPO2*YMq*gIo#sKNN6-LP4Td`<}+FB2LU4;2NmvNrzuj5c1N97F^t^0jd`B_ZN&L-r`MXYvaP&i zK^bR1#NYm^WwpFtBaKw679!q{{Z!<+@rB4>HZZW%N$5qfoANdBBD*RNWmhr0L{q< z--<#{L{rDHr{Sey%s?KTRMFY)UBGecfk!f1mCD}5LK%n$aXy(fF3pR{A54mzsz46o zf_ngJtf8cAm%{pP{{R{jxLU&C$av%M{{RD3E>)EL(sAest8m)9=LmMF>fBVbM{8pV zC-;*coE|Cjr0j|1ZC#lxOfl|d#zz?=pVFvHb2IQ83+z9_r@Of_oV>q5x~dPG?ygU+ z1$Dt`c3~&2kqa|tzv4|dX=8$OR?^Mb^gn<#b{jyvo-yzKHBzdO<4wKFklMK%Ml<+T zsBGImF#Rg>g+#!|FW)~_+EO6BCckbK!VuKxhQo+h<|LuGIRyrcw^ z(E1wlM1IvZw8k0L)Zl}~b3&_xxsjylsm90AWNvH>XYi*Oly(e$pZ$8z@V(@k#))rn zC@Ch;8$ZSG_}4=ns)5MK{0)0bs&P#oQ-h0Xk=enP1-M^)iqdCN!6WE8o+>1WuK*GF z8kifA}CsL`#Nj~PI zS9SSEU%=JeWZ`=LG=?bS8In$Y0Tgo;$=Ir}wtnf)(yR!i1E>^>DwFal`r@vQu|v42 zEPI^P%1o%FEKtP~hHsg_3Wj;it&VUy3{zTZAL;d^-{tbd^c4y^-IXbHL}gu%BLnfP z5TM3z2=ykYwC?hf0rnWHOKm*yW!->1vPD$1T9Frhi4LT~Pbw@Nbt4o#<0_KubJzfC zl#VQ)%7qo&^ag5@bG|L*{FilstneHXQ2LsR= z;(XHgnb*x2`>2)eLmZw>G=Ph>Q*(VP%&xG+WRb9=-m)7~50M@{x>JIbf=)3bylspG zq~!kq2+c9B{FDWdBIAHhCZUxKa{2Qdzh$WQHPnHhAa)rw)S|7a%;fK(46h~p#Qy*f zU@=zh^!@}V?6*Ljapl{w&bwSpfyrx zqlYYb9<`Y*>1Np%PhM##DTALOuJ_=`5;525QPbkl6k*;#GRw1Cbw-ZoxzQiZ*EU~ zRysi=HhO+F>`XJ&~ov=LyYIk{o*%K;zo}=`viS-A@*pTG+00`=HZpV^nfNXCAn$o~YGbGfvjtNsd^MJ?lyv zRgC1EnyYIhjzXf~dg87;Qn0{pt#H(ny{>ev71FsgN{PWaK7yk!GlSAG^vS3|@{U-8 z>?tLppFI8jr3 z{pcW*J6mOSQ_-J$*x)^ z`z3X2iJ0#BR;@!DgMbD-YFl{UW?uu^l53QNa6vxQ+?#et=!|()*)}jaG@O%FlHsHu zbdPgVN4%aoR&F-j(L3x6&sB5d8rv6C9%mBA$3tv4aD>SR8np&wLz=3 z#d0~MF61plj?tVRNvTd@1hK_cc?_R*2`9O&qL#!m&e(~|b@w$^M>^Q5fe8N4<0*aO^M1n3xTByFB_bzua3-zkJ zB*keUOZg=~BxS!klu3CYAca9v!xT27%0$a|I2AIfZdq|sRz2;(l~_5*J-Mjo)sjaj1_pYZ ze~nao$afG4KTOu}rq-r%n%ELWkA#UwzZB-UX&W1NA7CoKm>lxhU-Ru!K{S4FKyow9 zZDJEOty$3l*ax`=uL)hq$-w&6XTG*bg09i;#VmI*g!wsRT+KJFjT1=I1yOxl-xTD5 zr1_daqOjz>5g*=4bLmr?$MS{<9{H(sq*t^?tkR{j6S8KBls-CnsiI zG?Ka|Hy_+w`_?`5cegUO$4L0V&nBbPul%F3Ilw;0nu|;mOL68(r*C52O(;|4T*93B zpD9(eHv}}-4D5M4Rm*)eeb6}ft@O7Njm!Y2rJ^Heno`8KqhnN2Lf*nCc_^j5aaNHA z?mGVfDtyu`d8%J-c0ap^Aa^FR_ou0T>~t?VNYu2BD>6k}i4MWIdh=9n&CT7?`K4AR z>Y&!0wav6j0d!n+=CtH+SKX1#%bB85T4LYq|JQ{G*q3={OM=O;EjAg)DpL*X{ztxvLOPAJ|ttDJaV9)}p%5y5|eUR+25uq>P%hFhI>I#_IZvBhu%a zQWurkpDHSFMN+%DXrX3E`Ne8VsLhj{)^+?3Bzq$uA5&hXHk(%_TWE8#J?!pX@&N74 zSDNx+9?UTV&{s_!oE9Kl=RNUNZ|tPo}Z!Jc?p(N4> z?ZIF=oK~fqED9xf5FdJjNrnWqjBd=Qk^rs5SPKLYmuz?GO043vNT@!ojfPpL`Qs6R z>xzn2xs8jN`gRqgxgw0@B9B2?b2N`Ge}H>qHOmI1_9{(uC5up*$NHbd)aLS3UDA|A z{{UoWv7|RL++Z+N^c`u}&m>mhZQOn8{;vkrkx^|TJzG-q6KqVtcRec*O)E$N<$Vn{ z;^*eW_4cMibg{PLxjyE*;ZKq_Nad81cP+K6Pv`*lr$!-w;NY;NT}bR9V;S|TWnNZ2 za%rU}qYW0Q(w9;IVHAE9n{Z^585j^MXzbZWTO3puvJWR{I27Wdo`ESoq%hAESuVSe zy-}USvX%n^qIc&2)XM{G+}%AZp(M32a=wEWO_dwDbHV1FAmjisL`XM$#;eRchr<~> zm&H!9ZEg)Q-ZeO0YBxFiqM_@!_AdRmXvQ*W@gcZwP673*FheYl#AMc_5Xue;;QCf_ zw00bB$D0IPd2N&JT?UbA%E&pROS3=$c=BtZf;3EFK&}e)^yND!(dx<3c8Z0=w4F}T zRvFcPzLdlaK`b$oUR_$ei_GYhIrJ4*0+i;8jipe<3t-om%I5Sox`Nbk+T(6BQIsLR zQN<`Khwo!FzGxi*B=@e2JgUKa$cZ>NqZ+|kc7e~*vNcN~Ay8K}w#0$=vG`L#_uNOV zc2dRr-st57)}}?Knvsueq#swkN5{`ikf^==Wd7cVqWM*VNS{R0E8ka%>rMV{cjXggCDhIgjoc&Mu)ng;~{5U_-q>2}q^4R*5 zTJoiY zXP`AVoUY~A0KJ9v%A*-0f$dPqDajox8@6k7koG7#o_#Y=JnA?DpP;6(EO1ZjO18!Nezi#@ zxs#K79xd@66Ay?dZMYK!arlq^v91``y8KA+vft=uQi;EPW`wfv11?5?E^EUbaM-Ue z7QOqek3xgIY%o-?sCNL=c&Q>gPHU9h^zQ=r%JKEvJs8{Wv~&@Czj;_>@6V^TdkkP> zsORZkMevR}P4=ZILFEY5OKtuldIR~_u-ig#Sc>%MU)d(lKDC`oAxa+0pTeIMQcBzd zf%U0)#?g;{m1-go0nvVyQkzH=6J=+OqFlEF>rnaIjsYK`r(e#}w$wOL^{UZ8_}l>} z-kesKWXbDjmlEPQe)dZqz#582GB)GD^r*~US=}KSKD0f;B5|Dk4K)?eP4s1}ZKh1@ z7#~WtG{M*}Kr0Ym5s(k-OXkS^`9DfdlfBp~sVgH^OMs(l44=lUm3dp{VV_Pa2vPS8 zpU~FLv~jAAk0;i%lTg^w3D0pYtPv0~Otn5@=rN3bb$TqSlK83`jnW9pAO(+qO6QEF zIUPS;etjt#B2GX#KU%kK=ESFJfIG1@*!|U^$d@zJAk*WD zP*?_j=N&6@Ie<~}vw99{t(-o1ECvU!t!hPXk%7lG#agA_r(GynXi|jV3uQASE&(wCO=3K#z!B?pntXS0N$ zH7QvqbB;;;g*`!w$@Vm z$n93vGqDDJ6b@>g^|X@40+Ex~1EqCC7jJcJa@C5CuXCPNKqrNX0!^*yC82fGH+_m+#})QZ$md-~?Y&)|fn)`8amk`Wwz zK6dpr6WyYJtUH)=RTT`EkXxp1Cq?$iu5VHCC9d{JIU|piYg%+AQ%f2)*1H?8JKRb# z!~y&QrZW{I_hpa0WaxK5TQu?=(lQjYeBrY@Lku~S3Hu8tc}yU)TZq{#sp-d=dLSK*$PG% zt~;7U3xM6Id+T_C-6{-w)JjfTBs8p8ViaW6SteCd%-cm}-0ICRWr7)8o`(Pmf*m^Q z<-E&=Aa+uERVp@Dld+qq+q6-w46Q2?szQHwny^BF@{F2f5S37+jsWRWx5}p#4eOz- zQ?{mbu=$q*%qM8=Q#nOlv8`9Q-lVWQnD(L-lZJ9WgB7K;xjjk5auM?Y2iBYlMlh-~ zNacvSZ2uZD!dywB^we7@*9itdKBIHKlU}#HwU)5eKO?lO(w4YqpwpIi%9m znWi!02Z~s(+>e)!ahlJOiR#qYoU3tLaw{1+hSJD(Z)rMvnuzKWu6CX(jH@hh?`|r$ zp1>UB{c}R5n7yH@Z#?D!$S2p1D%dfg9$cl{^cBsCn>o!{fw!DAgY==(4s%qbmlpsu zyc2*1YF$RwOL&nKa^CdIWRfOXAs{zmYQx$~bte0Zvd!5=X5E=@VtWfjir9IMpq@`6 ztVs!&c~D>s^1`BPLXZKR40km(yq{)+83Ydbted{5t6OtI7+|%x34%^OW;v=?4zAl` zP!x4Fjd!bD+(=`KZU>;E4Kq-hKQc+&C!nUOTUjwWoZ3YiQR*)gfl;sl>Hx($^HI3D zI5IxZt}8kXuuJo9C#Fqm$E8hr1g$AZ{**~VF6?Q_O31Bc0=>rP1+nX0WthT@CQ0sU zj-O*_x!EU}SFLKH!2k>j_|4527RiWm%)&#F#4M2 zjW+c+brR5zQM`s2$lOc%3a4v$t8QhL_K!-+)NkU`^51BVY03M&aaa@TmsgT?0B53# z^l9Pm6N+`MBc`~&xxABQyk%ql_IWimmY+8N0GsAH>JD*PI#sR3#DId$iu4N`Q4PAF zP#YZ9b?Pcg(JqV~M|M`XhDgB1aw^!3m}SQ`Wnv04Yco^TVzr6FuxA5+4|?HJbCr?P zB^WPxk7?t#Z<;_)rfR}F&Y4gEJ?om$tgde)lIG)uV~`DXFvI1bUc#!Xj5NNaLi1ik zZk|}c88rCR(!SMfA;~>zR2UsA&cxK{O*%`TKs_o6js^!>or%X43u8SpYs1AoEP4~}{(!~JJ915FbG3isHZ572%gPYq!S5c%aYqWJHlY^2#_NU{Di_6TJ$-wPi?ixyx zIO_8~hBiYCE@&w2)j$<7UAZgPtC{y5p#0g!E7GS|nm03cj*Of{Hh@P1r7~EeIV6Ff zYI$ZLgu!omgh@AD77 zlETofzR9I?jlDvO!?uQNloCp(EPGeKg>@=Y=Bc6PRj8p-(bSR|7UJc+!hNwP1QshPNe(TXbx+fSPr~t;(H%;CHTaDIRIK2cDI; z15B|H6OU6{PM<5eDy8%=uB=*1qEwuo)dc!%sf>fobe2}du1l~K`}i{`#t+lAXsfjO zNOY4|GA#7B^20s=_p35T92^BweQJ3bV~ggH@H$moG{-79)-1wB%IJmAR@p;ds~PObX~?jF!OQ3Ut<@ zRv#g1pKY$CdmS`5)`CeN0ZuAUvN{ay$*$TxC^CPm3rQxJr~o3L*4D)=Ge;PjY>qcF zMLs3M*)Me z&$U|9tb)S0UbUj%yJb<-*DYG@yCb3$)FXKn$&S@|z^ay7u)Du#Q1HIgx{dX`JCLWD zF&F}~G`l%0G*%)3NVVpqe#|$s~%bdW`g`WxI|T z%e!q}&0ZyFRh34p&Zx>ZyJXTxBNW9k$k_mH9qNbnm}bSgnLSw3?ll7p1^lH14y4kC z0}$nXyoz;cv|Dny+rtt1QZ(?wftip4Sys26ax{!a=Je@V+JZYVWb0m>a0*t8nabwZ zypgFlh;8kGHvxzDKpDuU$EcgNEdw%!=OlElW*Zlk{K@1jLFjp{%VBeKh!qth=t|dL z)oD^!h@mQy_=^x-O=|91LvmZF$Qx-IN$yE=xaY7u3b$>h+gdiojhWk?rmX#y5N-!& zbIo(kr6|hjYE+c%YnbLqE!YCA38FdPXC&YNdz#R`A&@E}zo|W|Fx^_)NUJ1@t)IJH z6y)U_xawy4qV;+kR*`@=f(={*_2RRlzm9M`g)9fQ1zwCuS94%{SDR5*l&>uj)fp>k zsbLOK8YoY=D5_f)VQW#DH5l>#06d33iKwH11>}=j8j2^{)<--901SJY$h*2?Hlu7) z^{)DNlu||#N-efB@?yWcj~kDn6!D@2K@_H5Fw`L%*8TwWD zt<;paY9@wUrMfPx~#~JF! z(0ZEk9}@gXzSUKrvkhx(MhI5pz$Q`%bvJgfG zMdRO-^cCcU z^U}J12l%@4!&kETB_7^_)g=)Hgl!T!p$`7 z3oLospno&XcV&V`2O~YXu4BU7^Sb9gYod z+_x+$XOe#kd4<^JxKF~JXslOpJ$sW&c_v2C-(G4PeAY#zt#(wnl@s@ICQo)X6mrN; zWMTm9C=r6)Gx!RW$*~Hme8;)1G)l&Lw{nE{OC+3h9XeAM7oETZub`>?ji)_Z>sKux zWhO&`@5Ni1iqMWnC#gy8zR!TiE%l%<1K0AabIBVq+KkWYbHzgxZtegF^`xZQ)H!VQ zB(+HvhQSb^7(@zshNae;|e`%u|_dgiO**zc?!siFbF@DL*<3!<0sOmfn~TEKmz)cR^YY6<-Tlp zIH_{wHBsg%#ct&8Acg#DJGNv5A+4Ddk-(}lN3fQDPo*eI?(B1om5hrUcUO62UQg1r z(q;#Q{CTJ)oDMkpRdh1lT}K+ZQJz4>Y}Yod$!VfDBZ$bEVa+O7BxJjQPvKUh0w}_- zT20d&<{^&U^IV+op{$Zfp&%u*uzfLBV?JD4f*;1EL{am3e>v=Flf=b7cO;(Os%2U? zRg1tCBOcEvGw50Gc_HDPdz`^U$y>sjcX?q?Y}?na6r`?2nC zT6msTI2c-)6+-T8fPHBIMw2SJ`U-U_sH?MXS!kl{F=LV2mpH2)Ul=5scof-XSdgk9-vg{s|z5&j ziB(6i6r=#QkWG!grYt;TJvK`hJzv6*hTs<;D*}#AdktcU7L^TgH(Jtov7w^1_=p-43l)Z@5tkp^ITU{@Zt_Y=sg$z#&G{TD`P*;-T2 zT6HPHn&y+c3Uid*lF-JWU8jl;E>;&ABoavf0C?ljQOK9u8m8V(IYHH?IW-ogsy24VSaZoUf=M|w-AhO#JmM4y-+r9WoGEu4K# zPjM>7NB|n4ER5&o{Hm4NoRUJ}r6iJ1)|gYAQ-q8Y0FWv@{VAeGl_jtS1!&oXTr76a zB%`h~RPq!CBvq?DQSEnmOcb8@r8~8t;?jDW5=!>?`$9Br*-bGoLqEva9<@q#xW1Fm zmCI+jH7&G5c32VzrAI1DQn{OY7Nom&aHdkz-0=^D|hhD+y_&uKcoz4tY4O$AuQ(%W?AKrD2Pk^|}y= z0{Y-ow~uiejqJpqTFP;hn!7YjMQDkd?%u&jiZG+SV(Pvtn80DS3aohCIi|}bGG$$S z$M+)@Mra_1l2du?4@%yhC?|O?E{pKBObfXsJ%wIH>kk;WYU}()YVBn z0VDvBM>QZhsNF#%d{diTc$!E+UFm2)pF#f!eLf zFiuLd1I7TT);Rvtkr+}&PZjE6;@bC?ha~x2teHG+vL@m-kTCK{*9LUTD2vgFhfNfL3{ZPh zQLU}v#^TM2r+KVJa=~HAUvY}3<}d}1KGn2o$**{qDBnWhXysyd@$XVb6foEYQjTU} z&cwL6GiFEnDKxGQf1hBXX#dM< zJ7{N=j6}*k>slz11^K$wsG|xxQb`ys&2!49GdGMZLL}Y7iayL_U|F;6Pbd?Uz^03o zQ-V+7O3$E5QI6!24YA`Dp4Mx1AS?$4tXw&fepVnKT1$1>2hE>)sHT<4a#CfLRfPZ; z;;CL;2!3tAdz!e+tM_ngBT=<`x5AtZaw$@2&e9^PZD?gR$Z5RxjfxZZy=zY1WRPwu zc8*)qX+H3dx^V2MZ?y zu+2+%s7oTWFprm}YHc?|H&J<#$DH>yvQdO@GC6sjqD2V2$F2-*NTDP}B4?ytT@(=Ydp}Cb@&(*y-$}65Dx$^35AGwgA1{`(47QEsQ4)NxH5gri6dpTe!cuJ;!d<#x_~X5^yP zi*bmMpDk5Ku%}5l0{{|gl23i}B+(%L=}u|iW|7WbY#Q@w;wsJZ(~0Otlwt8SX~1_v zrx@$U<3%O12<_EF7RPhmitDDDf=_!GyI(^iS}pdaRmb|UbNHIUkz5j36Yt2a9}?di zYxG4xSi}xU_chI#?k;5kosN0{E84?TZklHWYKxk~r*~tVW9n&;!RDWvr?x9eES=W` z;Qlps-qUXwz#ihZb(B?+lBDGB%;@YCVC*^ft1bJKz+gWbzSiT{^r)T%1ADgLsWp$a z=G@s+w#I7*VbtLJibX)DC!fNncT%LHmL8^{IYlny4`3@q?q=1C5zG~_kJh8RoeG7F zf_-u-jJJ|H?NPBkfz37)E^*E(l{;$0=g|l)*@cH2&0G~>4WvBkO1bHB*A-*J?f0NJn(Zui;y;=T9RNmADvcd3-{yeRg1{y8NjEZ z!3sG(=Cx3n)>MFBOfY`tlfaCm@IO5AB}n6#arPWwS|(Al3G^h(}DcWX;Mm@8x$kY zG><&jwd?JE{jRSiCzG-;eptt#6_uJ((vrLx$;sWGs!1y$dW_OzB9(iOY8IqoRFN*{ z+I#e*DhSA=$pjJBm>ZE?O>D_Mw2d6G{K(31MSK4MgY|aT{5@rTFEX$zCI?mr2EKZ; zm`H=Bb6(N#rOIe`R={OqpK|*KAL0#lP?o1Pb*0ZpGlW1782Z+HvowxzoPBBT<2;O? zY}1DeoB)2c)fASd4csACDn{j8^(PfHcJZ$y?))mTP~4tzQ%2#J2s(NSQn^vFD;VSB z`qXCRHdi=N+lrV%(kWB}cd4j8!pi%10)DkCm#E)2LMNGW6;m}lPRZ2mC)%yUYaAnU z?JZR1PdZ{$6VG8!nLPr{YF~-W=NpMp)aI1kgg#tqHup6QQA-2_mI%J*nuNyJ1S*e{ zJetduH{@%1CFVG&o^Y{5F`Tm+yLE7u@T^g=3`bE>OCXNl?Dnb)k%Hcp9iEz1)us8H zuS!nav8(Kf0KPy$zafTVyNm5p&QDYM)rjK*rzWh)5qvJ-qtcYbk*^$|TE*14s)dC6 zJ}_O0_8lsm@~L7YKu1i}lBse=Pta4Kv^#Ry1GidhO<6AQL{rNXfP|Cur$eWQefJBy z(Q5MrkN*JIP2Jxgr9OAiYCXx(5flYbpmym`8x(RyR^0gDQ(Au zoyJO*VgRdc8n3ymO=4Y%fWQv*r6}Fm8P%1Q*#xI?J34nH)r~mBk{2vE@0!6E6Uy5n zTn^-uSK>y6H@oEa72PU+?U~6Mmv=fvxk7R}dg84ltjqG_+cnI?4dvmBV^1m1$hM6Xy3= z-!CVC0Hev{1CM%;(iYp-@~(A=o$QUB7;eUxOMUkwdQ^%6j0$0WqbzAdI}ECTJ+m13#rwx*O1zxK&Zt0)(@Sw(GfMk%vVY3rf~YE4LU^{3E4zD<`8Pnxk5ynpV8M zozl^qMVxO7jMn9iobf8FF>G~a>sk;bR^&#D`3kb*<-i#F3c9$SmMWb!ZCPK-agITu z{Gxd(LweVi+rHDo@v6r zuI`ncKCa7=3}1<=(7RqqiDP*Q=oFD%l<0EmV=41!#$3EJOP7`9_a`+yve^aWYWoqA zYew6{@gLsZZdLx^tDa7oZUMv*sQxD)nyFTf(YWJD?#zhnrIvlr5{uceYoxHXo-jUb z=esp#UF&v=1XgSvyOmnEXRHm1rz7*tV^`WrI~!Bvi`?iDmNogXO$O)_wQDw8T8eL^=D5%V^??_=O(2R0J$e0jb`symxalyGA-51W9C%PsidjjsNAe- z-0BfW^2Sx1_6DH|G{bW+8R!6|w6Tue0~Fl;^*bAv;X)w%>0K>@UD;isRUbPNe;SR0 z<7kke_Nz{;h)DsOvm{+YcgbktPe3wiwc(5!BOe?@%F*IRm-lBDrhu>P{PPJ$u&FX!CqV zbt+q9YUf^!fdPl%Sa+I&GV${J)=c*=5cv*Iy;FIlm9r$D%DoD76m~eORc~{44MPKD zWbQrjPPn<2H3ZAOy)t;LSZ*RLHy)mXv^?8ni_9sz&=c0F!p~wr{%I{{TwT+GO@$T2!=-SJ}N$d$Ss9 zRx^^jGQc47(vn2U2dA}d$*1Y@ugh>`e?_W^Y$cc?A^d5nr>MDI>`>aOIS2AI-b)NF zIjx(WJM8Eg9&UDm&_m_6{9V#*TgW5i{ApN+TKi5 z1c^~cT;{JN*LQ!qMk75=2*qdyY;9llyGxvH?(UYR_je(0SEO4q!Bk-kQ0 z8`C@rYgN^B+}%w%0}OeqwiZO;n>qHb=Sd}Y&NxlCXrr^AT#nrlE$!hhdUwTa*+V_V zOC`Zo$j;H)u(X?dc+@$!X&iJURjaF`gNd7WWK&*!NGADWcg9U>&FO9}ZIkAaBgJJM zSl!*Ny_e22^EWuGRMV!rkTj%Y{id##!7Z)OLlB5ReYnkL;TmhZ8kDMOvKH_qyb99< zVmo!sMQN+4ArdIidsUmOZ?xe^$n~n@OpYI&SLUdTgN>vU&g0*hqx+C{-sP}5dTWKY^C}l)`?VgpQ->UA_?_U?Vq>~YogQ94YDjGw;fFtVHU0wOw$O5FP;xjNR%CX<-O1ibHKQA4K&ivxBm4MT}d6C?xkmGx;A3cW1y~H_Vp(O zC^+w$$hMwYXUnlH$0VL>X)J`R$j|9rR4}o_t3AS1WmTD9Yk1Ds)qfh*(~+**GhmNF zRAsO&Hr2}3?u}#ib;)ej^<R%};8xjWSH>|wyJC0bF` z>U5&cddl7HHpbmau9h){9+gTnc{e^Y)T9 zXPj1$hcGu`m$#FsViW}xj}py$bTju!t?6yP(wj*jRQ8q;TaCk#E2cFg2K1OkhpKv+AuRg6;e|ek3m8-A*s}vG0K0pO0B~4i|wB@aaTU)l=7h(rTqvGHQxE`iXl+TeY{G0MfqE^)+#1 zP%sbUnxwZ9Oi2l~2X^A36TGtRQ}T~`hZil)6q4A|$YbsFs4gSkL1Bd*0jI{?cF(D( zZf7IqUQI+(zJ<1|K`8`aa0jWVb|&1g1bgCxyqvRh^`u}~d*ZiR9PO3NYF~2!k4#hi z)y42KN@6%w%^jD zsaMdeEPi3-zbdZrM4W;N{3?5iINAo$RS6wQ{cEBL*v3po_7yZp6oSP1;;NaDd!MBt zozFp9IJ*#y312Ic*d2yyi2cS^+i4AQJ&n+mj|k}2vWQ5WblUP1A0U-+r5 z;28bgs^tAytK1=sZ3qDTtHORPNE^oUMhE(+V?*!8@BNy&F;y=Hj*L8QE1WlKZ9BCg zuO#jP#T1mZ0N`lqw3MUn(*nhmg=X8?xUXOMW2msamcSqN@EJc8Ab;$}yk$<;1a>v( zeh9xdUMbQf&krGpk5KAI^TlgMR*{`Wc(dFRGrs_I>&00S$lNG6{Au>ld4qcp0jo^g zM#bSpcFKC#<%~4BS(D@n*u_eM7|?N?_86#~kGw%YohpLF5<-uyEy_%~hh!!8ZC}=^ z3)0L=Mn}1;FEExd9^O5w31f*PKQParsES=mlYI#zlkHz3=iRucx<_jnWm0kN){9n1 zq4f7Al6&|*?$0ofrYUOOlF+4OuHL0lGferDf!K3bM21*=n|x&ONUWLeR#y2}=Ij)4 zTRL+zHo$BEo=K-tHjToZuIx($326xj1&>owM=FEIQ&CF(ZWJHZriRmL+)huWbG@%4 zWZmqxCOa3LF(#sw!zsf6dsKH9YZL4romN>yuYr^AN;KQlnMq#Aw`~%EmykPqRjCft zz!Z^06$S+ou0h=3W3@wfW0U2HO{Gvz%aT25qT?q8N8w7*Zu`6sdTNGK^7ZwtIih*4 zy+p^CkPP`?eP~;z91-}_N=qTVs&GhS=qODwJViyV#Gc@=`ChOt;aMi#BbGjy z70~%?$;ed|N%Z8on+8p&I}Y{HMxu)))SXAEl@umJwk6N76@EQBa90Vry@BGkqPMoP zgakz%{0U$<74tVA zY}ByYnI9zKYd&tWGHz8DED_`I)&jx}xH*B3!QdVoM#=lTE-Vo}?OE-9|W5Z$bU{ zeYESF>sur^UCoYwX1L8)P>%iCiDKFLKSj?YdKX*MWQM&Z+%REl53!%HHrXvDh5cc+h(+f0^ZoU{_|p{ zi$iiSNv&&5COAOI=DDiZg||8(N=T^%w6_uma1?q9){4+7{KxRAg2xcY0~CzKm*wCd z=D4LdBcUXS%y5H%GH45fg+Wm>!)$W${KvU93Dn@_QbF!Xp+Zk$RQs1*glV|db9Mrj z2Ls-+u5~+b`+V;)y$3Y5##@KoLzs~1@AF$Bc9xOv)n3YqK-eUWXY+> z?Xe_gA5&R3H*IuxyOwWyjyNTl{J9nA!i|@_amuaVx@k#wBttDYT=vanTwk$cwYPhk zNo|OiZc8xw)SubTKs_^C#xaS6r6-|Jd64Et#VL)S?gD-5b{!foAU4uZa(Jn2^naUy z`C^l;6duY<$F_jRGRmcotyzM_=94Z41$41#F)VlrcOK+)qDyG)23f;5psXEq-iGRr zbDp{j5AN16oK+M!lXHVx)4{(oNC01Y%(_KaWwZL#T`k3{Gh$gKaG>?=#ce^Q#To-}nf9q{G;${lNKifN zO5)mfV30?+cWTO|RNa#tY1v(vvs*%>`AD&?KF(`fDLlr@w`>Di@1!cm9&8r$;}qzu ztq^X1EqyqsbrVH$>T{P@7aFbj5yJa>16wwF1ChFDLa*yq;SmYRxBK7T?NDkLS3vAp zna_L!f%;JADOxemYAI_iO#pQLI^6BGjJWDZ2D6t|gu511Pe7|!@-s-~caq)trMFOr zYZPWD*QE-QTC!bc-Im3dlIG_1Wb=%0sRsZW*q_7-fJTm3^y0JUhzmloFlQqjsyhua zf^U_9W9iKeLF*!vq_jr2_Of{${K6{+`b&$FKrDSK{OFd^ZgwN{sN;2L$r&W{q6tS- zf>LtPvp8F_AoBpFTPW_7kCXxEDa~a)%%qiKD^kl&bWn@6b6U!Ls`oRDF7_~H)8mgK zHTkOckz|XRE+hyCVsTv&(Jk%Q&7R;f>x#$Jq_9tsT)TP<)@rx3_fey#$*U!sUAKV@ zP&%A(@`}ugNp58iJ!k0Hs9o5SPi-y{HCc74RtzuL}01gFRzzMVfJ%tHr z8=cI#KK0925b8U~-V{_7qCVA0j5BaeSasZ(4Uc*icKs?TWNhsMxf{E=q~xO=M@CN| zoYR=%krxY-){)x0JY7hp+BTw}D!2xxXNr$F>q;AOz^*7&X;{|7B-%MEnyCh*zU;zc zDhEdFYWXNhkG?w9DBzgN*;O8t=ZC2lnlz^=2 zv8H&hUjrHMBa*Xy_;s+cn>Y zsV$j|=__2X;r%i0C-Ws7Ep?+#Kr27P(QJj8xXG@7G25IBS2ilNx%4_{!YU>avI0_B zNFB3Q;fHGFcolrAx#V$1(VSz7%i77UO``;6Zf2JoPBYe~Yv`?p9hH8yE#WB%^1#h; z`h@H+0hcESrEN-@s%KJu%T1GX#f{4#BiA*B zN-}Ap)Z)!D>J%|JryMGi+Oik6Uh>QXJM;WP}{>V03>30v|2Q|-&w3KD(@w_C07e3~uwgwel=W{4k za8`-5Zv6KjX5jZ`9Ac~_p|~R~I-b>r&dhNcAa})R!*WCdJ)m)dYF-&aDMbWoNy_&u z$fdA6!T1`^yHy=qnzCJ0ep;a*c`iX?^7>ZP<6+GvP`gPCMNpn;mU>diyQLW?*0;jI z2RmvBZlH;R?onGRvbCa!RACo(Lv5wRl4L56psh5tjCM6%3zcTVa($~tB|CRX!R#xO z6J66+lQx1)+9ZgIiZ}wFybh#Q5N;&6Ae=@$y49smTJj-}l~?XW>u^!sl5QBz>SpFA z+uoF^Jm!TprApNj>}0iMsrN+`tM^3})eT~SoR^9tFRI0s3Z6*s{?k@$ZyF7%xyP`r zy>{GBtHxw~v8*GE^*frYEwq2aK3~GUdQ(cAQNuVlHprh=y=C%VLr!>Hbsg`=`jWo4qG^#hECpT9Tua<;Hz3@3T zV%+&DfHF@~d(}APRX-|c+|f1ZQIF-tHw);Hl=`zv?Lkzt@%YtAwIzJ2W9wOMG?Qm7 z$JVR6;NTKZtz2j;>L*c`q?`#VmI7K z?HPa8V?W)_YIT9h9BmaW7P3ma%WVhqsdH~=b0ZdR!n2gwYB~{!+zg%<+?sl1@PK31 zl_h(La@;3j(yiejUGMk|)>4&=Ne+rHDm~4~ubuuUN#=O}08u`kSpNXXYhOmg6t_#} zIM{==eADp(%$_~gNjb&F(ewZT`U>L2O7XGNLvxw!L86M_I>L%5phAw64u*k8tpF`_ zx^2m~mcr{Kc~eCB`Z)gp>(?{ZwxVzNlY{72pZFDA9hoxN`h{kG5kHu%duvc)1(z&& z)~V0b^ROAK#285r#z zrks%%pOf;A^e1h^*D)+Hx0SRElho8RTgY4mjQgBciq5eq#wstg79`^}3R1o7YT+AL zqa+iM2R!zvqtl5^toYBRS!MuakPS%;iaKSFsix{m+5&`csa7p11fVGC)ZkU?gjkR& z@l!_Ne9gsDoxVpG`3PKLi)QW~nlN)0hKaE$q z)Fct@3>14-k*3?Zl%}mTppP5>X>6{4J8fa2u$DnZ1g?ZD4mRqyQ;i{(^i zH5b|4bCG~OsS+n+(+lcrNpj0`wwvUQlPpHET^;faFQ#h4DueQ^HZoE^2kTWXugn<= zrJAsDl%9poYAWnYsh%V!pdCr9Ro5kwUoCJ3dginBeMb7>m~1H*Obl~W?No8NoSxV< z-%f?0c^sbE%32)*+(|LOe2>DaUdF2z3QHdNs|M!Y)({3W^{jjCLF6Tuh3(MQQ;Tmx z=QS0nG)3*=-Uid2rlT74(77M$SnvjGAbqOf;hDDPwXD|C)dMswB%`^lr3kI{F^r`x zv@MaT$!jIV^Tlr)j#rJuRXsY!DPMaWu6kmhbv?!L3R7fbu{EP_;yy=BzW@$HqZFg;@U3wU3hMKM`D%(#*pMt&>S@r^x>RJd88w zO*e;&)v)%`y^=c!^=o%HL%912wk38Paa`4{wZum;p!$(l_g;BtEK7D(7_28oDd@}g zjjc^URalohnDhd(wcTXtHtNm*XZwS#RB3gX)NE*?Us5t_D*H>l)T9#K&eE_LPzEvj zRa2uLbmrBJXE~_a3sSoHjyV;K*4}idqcOl06|6Tm@)d+I$>$lstxXrhFv8|*#SVXk z3E)>pHQkPvY)JCc1BG7o(Ner3`DrV%DbSN#)frmDN`p_g4K53wq|`dUi0pKgRFipj zA1bI7%WD29ySS4OF+dL+MPhk(T7=$Y&eBMGvF4mGa+Bw(!g!>;+DNyl_>Oz&a}sP` z!>O#rSC#iLEXS}FdGwj0#^5vPIjvnP(o1I&stw(KWzBZ;!km4>o04;vyil>xuV-M* zZtEZSm|%XjwwAG5jjMv)`c{cc%fLL@im2=d6+AX5gk+QTCb;Kaz0Ro8wusBM(w13w zY!$H{nF=Q2tDkXJm1B&2pbD=ng^w89T+*p2Gj#~X5ha>T*kX2r&{l*|L2j4`KN_tA zs&U3mKJIAZgp$W6u+2oAr0i0%S1Ib6lu$-hNFKPwa<>=HaO8$K^{aZ6ue%EaF+I&> z$YoKP08#11chRJhIO|QfVsAB=A2LS3_o!L+@_8Mqv(LYUT-3JKw>K^Vr~d$2x8=61 z$x8Ym=_|3oPz7ItQb+~Dk8D$uPmuxK-1n-N5Sw6SEZEIO)U+Fuj;BdrAppC|Sk|J- zihza)o~E)in@68W`&^e1!y5uYY;>#}l)BVpXsx9(C_i`{E-RT+jO5k!1&EYZ+nuDB zmTWex>f{c9=Cg4dv6a&+o|VQAKTf*xh+^Ze+UYEs^6fzKwnjRWoYguCnz~^Lxjj*N zq>f>LiyDr@9Mv1YF(V^()IOc+(UJ&U#N5bzO;f(LCOy%&nCGH{kSKpBGad;8- zAS2Y)rk@p?geIvra4PV=qq}Yg0yd`jW|u-O=v}U z(MK4}v7jBda%(z0PWg_=QC2bhezkTR3r1!8RBa@_nK%`+))37IjyS|%dR8uRj*VQ` zl2Yqq9^Iz6KfG3y{Wuk%x_KWvDN*TL7B&|4u;6^n?^1{fa?-Nx>?@vC)$})w7i~-l zQbr;;w}I0*tMTdL3vncpu_Kz&+X(5mKGj20c?2gXtdg+9h!j5?-Zf0H6CU|q_0HpK)R{R#t zBMBY2^sYw2?n}8=NbQMGdbM>1^_DH54cM-F@psVHjFVcLjPXRgu&Qlvv&lI4grmm^GzBtz>0;4(nX)~zld?wo?SHw)(yXG8Fa<>f0x^MBkXJO{W1=bb5n=%3@@YvVoDKy!nNV^@ zdkSDDZcke0t2Y&NCvAq7Z$K#m7C9z@!1Xlb$v)M`Ql#3{n8q$c@y04PN|Kt$8>uzH zS{%z`sc1*R<253#4r!|rCQd&SQB621RTzF1cuXYb?$H%0D3s-$;E|s>P{iV>&#zA@ zDGWk9@w+)ScI5-bw0=0O$t{8%!-6^*`f7Bkw7HSRX+>^N2$CqnIVANZj&f<&5<6oj zJanx2B%0b- z0MmEIyb>#&$v2`q;|^73$Vy9`v;4l546-h9k(#?820jf(4ZLckw0SEeH(-0z_i%1d6|+wEvL@HNE5jtAP6;bnvP&D>;oh|*cDJ)-9+k@p6>b@u zo@;thz1f`Q%Vc%dH;^C*Gg}{MK!6TO?OdI!r<3Lu0<2r#2bAr8o$Cse6L&=_jV%qS zwE+yF2;|kcoVxCAE1jBngv#to8nC+JZP<}9?ODo;vgJmUsx|R5R_5w72Oyf}?=3~$ z$(a1AJ5~0$HvpIUMP=OF%+eU5JbL1^qX!+g3X#? ziMnH!HBROd)0R=iZlt~w2}h}(tt2nKK?b&LuV-jUBlD>(wCA}4a2rKxMWzUswgqQa ztTua^Q>7)X%!IkQGUWZ*=qz6`Zf>;*wu$&S7^kYo5eTiC=BXt$WR8eM#J=r>&0^{e zEZ9`&Pc^9SHgamCD06_njb}^UcPe`5Lk+q|5E#p2tvJNsaJ*H8g<%D^X0!DTN=N_^ zfOA?VSLv6TxuuIAOHmx%jO|99GZNxF~I9uMlnxPTa#K1`(w+(8dz-R zo?y+=tyyVCE?kV()vcPY7@Ufk)r>X>Q|9=F*;yFy*})Zc!pMGY*&T;^mtfn4ZfYR$ zhS^F}M(EvMWRQbms~ang$Pp`=h4 zn$i}&Om>WCpL1CfT{&|cT#kp4TUyeVxLj^IKrlY0s@qzsC@em2O!TitoRWgM%{aF$ z3NhHb5YN-;)~=)qqmDm1v``+u%9CzAsC68HK_Zl+Xg}jv*J{#}{D94FPV2}gKaFHh za}u$@$sa>q6mDZxB8S+S6dvccDmf$a*K_ew%YJ_FPzSiGGfej%Gru00uGpzP&RKHo zhDkBb5}+;Jsp>a$>FY@C8t`%WR?5iE!;{RGJ{Jl)(x_A;02BGtOEj*#BMbrb%}S9J z>T*wdPeH9m7HLk^QcV%SB0HpDJ*q_;J783!5)4HNBz0!ZLziN#INm&q<-?FYPAL+| zJBCzgB>wRn3bPt~pvc=)b`@bf)?Kk73EYv^v2*3v(MJ1`#RCS~NK@R^#eyYUZ6wvuM-MjzVQY8Uv4N`G?2Qn0#@pDFlC{o)2Dt*VM8~ zurJGiYv(VEwoh;4X)ah2mysl@eZbmB>H5|cF5u0fWh)#Bq@dAUXFyR!6bM==DQE#j zYQWjiWFC$EMP$~Esm_^jJsE$Prh%VLcwbX5;Qbw^W^eeKBf$rU~2nwd7E5Jz70+f5ywKz{apb6v^7E4Fd_&Dk4T zI|jGk__hZ05e%Iy1bZ-x%b6iPqYz$qw3Yk^+?6tiZ$gW%Wt@FjzV(M5`6)q%aMsWIQmrR zj5?f}aaJP*@I9$dO%bQDOsKZwoyQ*a6oDf+R%86C%fg3^#F}rAr;W$(tBmcrjHo+W zC~RR!&ftU8o=r$1M*jd;l7GBus|ZX2ndh!*aUy|@r}C_7&R=A@bk@@%@>@J~_oZbK zu33Nrr7lSMRQ@$yWm36+nY&hW{iL6~X)1AF30-kj>ySvK1o*_g8S4%Mo@YAkEyN9Qeago(nRv3AvAQ(PZc`p*71DAXX{s4fCm_? ztwQOQaHsOCjHfqiidKpyp5ofsNp?h0>P28(c(%?t86jf$?xM0aAwATmB<=JjvYOB< zV!aADSZ^nDj-`2C$s4kGpu2FNok{(%aUjHJ8T1vKX<-tsOks)c)qKEJaTo;q3er%f zv`EcLZ&t}hqZO<~-CVJA{56HCU%_*gwS~){;%d=4Q^{OWKQX|o zycWczC%FW2p!s9@kEyLIYz5rP9lw<_9Xc_~%Yo@mommvDBr5uv%2HC%7L!_(?uedZ zxaZroV(PYNHjLnRHLOZNa?BX_=~*|IOLcZ4Rq0(6uAYWfTDGNBwy=pfXOa8X=9z66 zwl1lR_o!pi;baA*5sY^0MAOY{Zeq7TyB*J3bC*7elNj>q$nWP%Y_psKde=Xx>H_Bx z$C948rrfl*j(qv>qNXvlgp1P zxChg{Vn$_=!mN1)yKOU1NbM2kL&(V`kvyi6k{1vN>H1#u;Mzk!o z^T@t`j@!PH<5^-ZhYY$&O6>$!f+1%*@% zpjDv}mxKrFPqVOK0A|f=M+im2isPwKC!y6&n`=`wY(#0f$tJV!uOqfx9Fy1z=^iYP znASz+oNi?WMpRO#Q8(`uI&R+S%t9_T3)TJERn2ymcD_=m^c4-3o*``FwPQe{m3qor zM_{8zmshyu019v-nJL16JJVz1abiA|%cZW(B~sF`NCEk=pL&W`W6UEvR?M%CHcT5 z_OJJ91_%cgFUZmslc?w^#ZiV3V!oo1<%&n03rV(7VT_ZERb%tF-NJgB(UR3j`IyuL z4=0wSkhRqrBO39Z#4+jVaB@&=9r4z$Lu?`-0fiJs$2eSoDsuk-CR>WmbCb~0gd(G$ z%ZCLviq97N2;%cBs>YN@<@dggIG-w zl5ha;ThY;KVOLk7b4iMO$(A-akG*v#O^OuxKwo;PrrUj*HEeTMr;=G*C=HC)Em^rs zO^%9Cv~?;SFf@glQ;Lwm3GXL8q4g&iv*;?#+aZRj>EWZc{e z-cgJpa|Go%*yyceF`%a2T$rO{&IMT;VoptSdY+}`O3?%I^sX9lbD7yWazv{(y(Z!! zAB{vcxlr$HlU35xO5Y&wS_ewo4&p1ds&UbyjXIL*ShDc~%bzohdYaH0q&`=fy>a!d z6w=J?jl2rcZ8{{hF@yl)t!Fq=zJ}19OKL-L;y8gTl1H^bnw%G8cFD&zk$0w)ejhaJ zO*PraTHD!EQMxd7YBy$dJ7R%6z*jwM7gM^H}KiA@yrehHP+iq-d0I-xZ6a?NE!moyc#nk@ayw-NaZ|3>k9qlX(=-C3_{DJ4 z$3Z(csiLRVAdV&E;2L2L4nV?G>zr^#4Iu;%QR!Uk60del>0J#%t+?Z!^-AfaiT5iF zwQ^Xz)Z4tE?}~zZWESnWB!R~@Fg6~4TIQ@b!1=lZ zac%&vmNE=r_O5&k+n&0cV(F+oHW?Ps#X-r%Nt|*>1Ky(Zl}Xs6>s0P0;IiSLwtZ{g z`mE};DLWo?W}vK%Xn)kNIqOA3r%5!n5xu}sDb7Ol>T&u}T?umN%TDU#cc6KokUs5G z%CCRcmsYqj-XvfWN(1_utLB9nX2;`QP=n=6&Qn(;d3z4Og+n8@-~P2hE&ND$89#+q zmhj3jI3G&rjW^iFQ(eYgT;m`fO=p-9mvgpB`cve%j5iCD_=<)}u$$yKUc)_WvN36B zbGDx3Q(B=J&V7L5t1J+qCPpOoP)MlmubJ0uYk}VYRYsMSa{Ldeu9(JF)aR7adKM;h zTD9e(Gc5NwaMKkLiQD}6{WlhmjMuaa%hfqkL8U= zAH;D%l}qk_ox5?IS1jcz9TALU>P+y%fgnzu0jNyUOCU|z_86$9xpz1Kf$PmA69grP zC+kT?+ftmB^e@XNlB&B)A3!MkV|ja2Z8bp2G}0XKJq9Tp%7f(}fCiVdT|=Fg9u|L| zM*w@8@{f$xNpWMRUPN~^60}Zn*@j2zzpZ*Z#djlR%VXWEBg9$()_gyAV2TW~CSxRz zzBwPLt8~+gO_Le5B+@=h>M}p6rjRr0e*?sa!J0Y@WZQyI z=Uo{w+XI8o*0{d`Tr&74N(imJqQNKq<6TIB$loD3_chp~qPsbHyA63+-#yPxYV>Lr z9zJOeQQe0DPt?_glyn@`Nok;=xjYcIb0G{*VOM1R)I*MH3v7eAhJC6VyQY?L8jt|2 z8giA6gw$e1FpgFvjl^^Un;dcp01<(oQCA_3G~^tbmCB&p04-0R+9Fjxtg#r|7*Xlg zv$Y9I6Z3^p?@1-Z#~?r8qNvSnEX#)Z0pyZ-G*hQ%c`;O-)si5M(njhr>PW3fq*YJ{ z=qijFd`jTEmd9?@cp-KR6ksUhHLXV$L?a}M*F=_PS1Oq0jZwBC=Kb7EPcG9Bs&1ZJGdtT0Y#82MmfGCiw~ zQo6ajT)u;gZQu;k3$}Xo>?uhY;2KEOf=N8rFEf*{H+_b(lg9(zoUwp7H2tHwr)5V$ zTPBf8HW~&y(e$QCC|=yu5>CNB@`30NHA_-uIIRh!eXHtft4K4GnvD6fakwa{qJb21 z$@Mi}&f2yL@*!RF9EHI5sl-frwMo;UsTGw+CaKEX4J3q)AP+S!(yGmS9BN5m0Qamb z?-6-NNCa7xql=`H=$acoR zbr+>&>@36Dy*xGzL0)^FeQNGcL!`1JQvv0a#|Z7~T?{i@z=0qc$DpnaE?y-HhPQO9 zy|MoBzGKjub*Q^vp;W4+p}#HKftgu+SCcA*w5G+ttP)bjfjvc z_5!)pIW_L1quM3AI&^5_>UpWs<}>_4xrh?#{{S9qGN*iTTh>-Ez<^^X+}1TI#yh=D zp*Jp?7uC!3=~D&uG}TZC0-inVnVO0{sno*YR4pk7gGnoiyA}z>F`H)lb4U=%WnN8T89V~FSlv%M@l5hJN0CWD_eVY z$WSrd*B3H_eNMG`tL$WtFed3Rpj8RAIGK!LvPT?MeScPsS zc2c9vxuv1aR<|sV1Ufp(}a{~rGaTO2OgE97M}RaFBIio!i^jBGA(T&fk0vN{RLc-YlmI2D?eZd zO0yld`HG0h_pOUtF%s}U6IfKLmaNtkn=`Cz<}rtz0{6hJ6>y)xPCco{?i7&5NaLv| zileD&s}9)h$~=wORvz@;#@37$k^7=H4O!F5S&#wD)&1HqE0w&0t5W-Om165ERvkW}X}c zB$L@|bwic;m^$3c$gyxISKGMB>yDLr-^{x~3PuJ8CacFLplv=_LMAyq{{Wp+OWELK zha)|4PnF4NO0sum`siQ`vnQ=yv~@BF6bvsJtgB67q(z92sm*U`u|&M8lE=Mr)pAST zI;p6)?vooRpO#$I@!H0t00H!?0z8hC_@WAOYmL-rT7p9sFqI&1qk&Rj4%7fMS+{cS zVzi4S!vSz9QayuiDssS51Gx03eFDO&VxB1g#W;+N1B#Fs%aSqHkVb*Y$26bn$yMah z+yZ*nJzZZZmDTTI+;^y0?vN)Uq-6yBxu$t%In8GdT5{EsQ<_>NSlYM%ewCZ5eVroR zBtC0)0A$th4nFlYlxJ{V&D%AWBK6lK5z|V1kC~V{G-Q>0{-09Qx*-Tm|BtJUYXxLvk?{y}a9&i?K+l^ERHfIDaRBjQW~p8KevbG3!uH z(l&B12YU5k?GWfM6}Tzk=0#`!tem7k0 zGlYok8ETH&CuQ9s=91dwc@6UY@l|d>L?yF<-lmdA7X)JpJJoCU^AI37&wAPk-5JTN z8CMqsf+7q@t}8#vc_cg<>7lm?ZG})t?^v3O&uAK5pegpQs&h@*7*0A}lwG{H3Ec5o zEpH4q!HVUv)Yf&3$vnWiWwZHK#rDOu`@=ZkRnF<8Wfq!j%QFr@z!g!|q;tu|Ot$hr zAX8f2SivNKD`jM5CfJeMZW*mRJvSsa+RwH}mStVX8Rol4);O)zQyYM;di7j%HKR>S zQoMHC0LZANJ6nz@y197A&OrKBOxF@hPBH~!2sf)Xl+<)ajGBpzVGd0?JxR75I8-ZCA?&#g5T(3v;X zNUR}itjRDw*{zt~-q~=hE0I=Xa!gD~^{tyY+?DfANvo?3HDuGDWNFCKTxVtin!?rX zR?ghWd17m)P>C4M1z>B;lcL5Bd90|*-m=)zs$92wn7V!AnE)%^y9hvoCij%KJZ;KgrE$2AbD%Ii{Y5tQPCW1z0gSz+ROdlNDUOX$`(!c&=)b zac4v#nkj|Z*R4Jv(y`A9JJimpjIxY+(xC)rqKfkBVPMBSPHURHoi)snP+ZzvNiQv7aFR^Bk44Q&L6JUO0BV)KlN4vnbyk$}Sd7*9 zq9{o$O>srlt1C%p&7~S~_!&*~wUNU#hlA->B-vlX9&0P6T+b(RUHKs z=+00RMPYmDbS;ccVC}y{jQ;F@Gg%YJfgKEB4Kvax9!&77ZMQsR=kDK_1aV`-5vGOP5V zcA(n7sjCGL9jY<-RM1(pq+w1!8p+Kq%TaGcRfI_+gp6eR)uRx%z-(?cZQ}bf@WD;SQ}L9~r82TCS_Kpe0I_B>bsspxXEcUf&dvt~_hPJR zR#M+xG*KyLSY(jnuv9EhRrD2)r@k2=B9#IQh5-8z5BS$lCFRY;gevizWY=XCDXW~bY8n&W+!Z1?@woM?R-2R#=A*T= z;ACabQfkR27YGm3nz=P@R7_(2CQ8%I_hSb>!jcPFHg!@xU3yU*|LIK6&f+XRN5+dBIlFGr8{XYlTxn&rc!$pm4V1n z(>}tOf!O7jYWh=}WRLG2eFY(xE1Y0dYD?9R_7v6+yM`F1i4HSOGLuV{idzF6gbe*^ zBQDp%=bqH96P%S`3GIq>QEkI8`Kc;VPUkhJDO+ScOoU{Nel#(_<*CoSugk%z#@w;_ z+||WJn>N;oTx+T`aHWkZ+n%_om1B%>KROyiw2(OZRk?IRPBGR)Jdh4^Lr96#img4v zD&)2grBD}3B$&=kTwvs`C0;GsDAIY!;eVAzT;nX&QKTU8Ni>X+Zt}~wU@FuKxfY>v zk;HSy)Ym6-E6pL=I5oWXcem$!;bDw)8Rn|pTH0BzUVNWHU38pdrtIdc7cE}K6tIL` z^)(nP8kRW4Gf=%!_IZ3?{QAn_%6i{4xiqDqpjsU1W*x9I5o78S;8d5$5LFbWGJm--6RMCY8 zo@)6mxVf&$gBI&f2b@$*AkF}#5(ODFtUHo9z~-pWFkJIYyS)ql0PCzvokNq(Yid-R zF{x1^Ys;`R&1Fw|rGEWCXWGVj=hBpD?0M($uE?mzbCyn0_)KRqOCcklPHRg^S&ia( zibP|ntvIx(V2Hep$gI1)S{rsKnK<;T=k}iLQJ=Hq_oU9M?^v|3i{;$yj(}Gqs_WNx ztjeND;`TVrS6xy`4t$}v)S8Ow?IlgPBzp=}u<8!&U5>4HHKVZGHnc3wB0IOPc&kFs z&wnr}Nc(29U}+h^NcOM2Z8g(~g0~|b1#L9b(8?BTQk~pr09dofb5<;DZdn1t?axZP zXKXH6Nm6~QeA+xvjsW(qdX>3-)1ox5sRo~Hune%~t;hp&n#a4iksE?Q_M+cV-1}Rg zu6IsP-ZfK8-84y*RO6LG0N(n-Cipmea^&$V57OB+VBZNxx#0h3h0g=5m&X$C5V%D zy^e+n=ero03NN;DYALkYo?Y=uQAa!q+0x;TSOnXCUfC5xQ`Zt5(LjKnnXYv?Ia&zW z#VFoPqbhM})Y*ww<&9=E%;zLvS9Lm8)tr?o-$ZF!Nikuu zOZ!%k(`24ABZ=&}bImy+g(C9SHQkTAYo1kkmD#$7E{MW2T;57!k9l6o zD`wBb@OeIE>0sxginVPerk;LxP|aGgT-#0t(YX57bt_8Nin2AQN;lV5VOnT=t_hDQ zpzuv~HddnAAPa6jwCG__Gk`voHryKIt69%Rbi$Nju=-up3R1Zo3c%EYNw9D_(^JfT zbBdlg0+L&rd2pv{j%u#D3dLm-89E$#)s<2y!#(NU1xt4Ut-{j}?)lt#^`$OY_Qf&s z%1(MzyX<8-UttJf3xoB|A|B*^bs$W}H~t_~Bz=sa@(oO>^DPY8arc=0!ktW2?@#Gm z5?Y-xZL6^u5p3NqXlgYr9PcT$V!}&SkifTHuA2HnRH!(vGg2+~qw^94cVU&?&P+qH zu;fp(MU~{^sH!tuxqPudoi^t6k|4F8=1D7Ok}J`KIo9S2WuP$B4X2oj6mX8=g5YMTorn8Cx zm0&a4xSdqpnmZLHmD4MfX*pn8qixUf1;uDavb29Hj8r8y6UQ|wLvb5!MY0|JX^VH} z-0cs?rDy6dAPmDe70=pUT#K^X2=_I-DD$TyIacLq9)Kkgj-s>gV{4=H46RF~MAO@e zCmAQ8t)T>EHtYabB&Mmz=uvZt%-h;Vr{qEnXe4c*at!eo;s*P0`*uR8m>U?oQ#!t*bz>TkY?P(o#`c8OA)%QjBULAZMj$K;^l{ zY6iD5NKHzy$r|BE0<)7$xTcw7G4sBv-pOnuz)YHr)!!jolT7;)V=^%&lwy;+W0P%J zBD!7_ZK^hkwl38MS*sr8l$?d=Dbb{p@!Gl^6IVHuDox$nHaaVX4GJtxIzf00rCaP=dD#!Zste;Z4{rw=+w-j)$#yH7PZSydrc(#lu!y zOl=g&M&<&r?saLcfM5na4O!FmHN8U2Ky2d`?Np;tQ|3n$t3%;5f+~)c8MsW~8j-)% zp4?Uy!$$3t?&szCxZF-S<9+n=IWUP(lIT@ua6l9u)-I>r5Ffmgy1{fnXwK+LM znmdx!otYCxk~T@H6%|Jn8B@ChfIHP`9f&a}m`inhbQQiHb*jEsv5eI^e)m$_JZcGM z10MBK-%?wPqY@?;v8)^G*zGTV%*=O?3EkSP!z@%@zQ-G{ChoMc z)W&ldqN{0=Ji2^OA~#v%=Q2D-D=2oa@i}3d>x<=D9P(G(r#6}+AHI+l?i5s(kOw=@KaO_vy?KHA-JxToSp zTWgT1!vz3#tnatUaHkv~^c4mrl=Z@XHJ`I@;xzV&D@0l3p7og#u%4vnH8d8ULSyqJ zQ4+H89ZlrOdgT5UZQDdbi`|e*a>?=_{Aw_+$1R`6v?8%mcR2kjR<{hpJq0d!t%u0o ztntr`kViR#5WRe@Hh_qOiD@nY$B*`tb9+*2o=ZeRR z;kLefK?a*WgqEo&^Gh?Vo>Y6W^D^*T+w`i8Ti6>(T=f3{g!);v@G8n!8E<820*~QA z@{ImO{x#SjNZ$p4BBIjt*tE?q4L;qNM-*Z;8RrAmv@NVrTo%DT{cFB9gl%($r#W2G zuz{LA+2qpQS^&Q;OCI%l4WU2-pzTXDDye4R3gdMjHO|!q%VRz$jA@SbDq6esI26RZ zi89V{^{Q8%AxM;n)SsZM<*6s4Ch&^X(vI2)TR0i@HEm{$3sBxNcOKx)EW zvi0NfqEsaJV~tq!H~i3gQjeOl;J#9D22XREwk{dEU<$%eXw^+J;7}=9f{c?;sU%ae z+|*d1jNtMVeN95{d(uyIiN;2LwN0*IR^0Mr`U=(%Y|c_h&`@!cQh*0qqZQ1O{{X9q z?e%Jxb;04vf#@kp49(QO$gpGpe4{lE>D-;T9{8kKna|ARf!I{TmvI9ZPaP4<|KB?^J@-et8t1LTj4U zyh2tMF-ey_05!>6YSY{~XMStbp@5XtEY5Y1;=R8oleK7zRWok+VJHc!^0yOhf@ z8?s63QmwX|Zo8(HrGM_IuVRfD(uw9&t167E6mM|bi*f5$4AEO5P=KEKrn9rhm?30N z)g`urd#H3xi|W;^no9C^L`65GRoRnuc{S5)94PE6-kV_NMRKI`#dZ3Bf+2=6B;}pD z`&OKKWxc)vr~p?y@vuvl2GPRH;+dOcX?Q@y#x?e>$Rm!~HyN5qJxy1*yn@;-p@w}0 zW{pLO`Bb;m*Bq3pYIP~7Xly>ED!5cue{eXf@J}PC>saK?FfZ-K;9| z$3tpu%TejQy42oLBds|=1EoB@ayc8=gvYm~Ln*-HtxoJsR*;3{)kI8cgN&ZE*<|v^ zTCXkFW&TxM?snY7*2*1Olb2%G+6Pi;jdXntS0^1Qhnlq%OOqve3sRuxCYYcTik(v& z;+xP&k#o>gi^o$?`KuTt)k}-~XKCi8MZJe~jVE+*Wnl zH015iK9#LHTIy#ynrJP>`^dxt#Xi!-?#|<#x$9Nz^vRil+Rd8ZF4nfYPXrDC6}3qy z=>r(W$=*cE4W*^FHjTh_HB- zfMlB|(2B>k*8a%3NJF{qXlwZ{WbNuYSWrkA1c^}qfK)?$gZ!_=J?yW3O|SnhVnZ*V7^>miSJp# zA5C+0bPUorU^~Z0yV(tLn)9+lFm{{Sw>GpW_?bjQRpC6AE=O1~~MR5blQOQ$l; z(DW5wPr9<+HzP<#YUyCrt=1kEIET6Bu$}4ML2hjePR>l{^(_|Ncg*rIJ(LQiW2H%X zotr|hQpXjqsOoxq?giRLK7z4!T;2_~aYz%?R@B_wb$0b8N*vlDo@91sHx7%`h_=AH+Im_zPn{4 z#tBe6RC;CYtg^AalgmNS4yLLJCze06#Dgb-O<99P^3#0yX7tTNgj8(HQj?X@r6t~w za?rxgf4tqQ3v1|Ln91e^9hWtmJnI{IN}*JDsAJRG>A{Tgo=KuKqL!?uR!&~(J4;PQ z;^jleSIhOqYsD!bZfu&xw25u45+RNUClzi5WAh5(sR;BtVJB@(Q6SDUQf>fM2$jo> z)v%=3G@_BJVo50{3sNLtbB<|NAUUZx6)pA{0Z3U+S2;eE)+GB=N`O?T!OvR5+OBR< zJB=zgU~oNY=(thqP9SGE1k_ukInD)bxe6&elJQV5Ss$DLDf_Y1)Kko?n+GKNQ%Up+ zmbRq9KpY&>o_kbC3#h;-xTnV7Dd#oGQj(42!VMwx15q<2bL zr3y3CnyyVQ#eGc!Om3AMI}Y_$dzFbxJAge+5rboLv0GI6B0E7WYmU5k^L7z}YoOF7 zy14`FY`aHuT$KKI+L+4uVd-9`7M+%-or#^^fhu_@aeH#AKVg8y=cXy%?jWgeJX1Ub7{dGn%{yT${5!jTFrVL zROu(`VRL%94{TX40**sp-DNdJC(!@ zk4nOpZzrk2G^yT0ouXCL#_Xz!!%JBVm|o_#qz!U*q2$%u8y_GN${DMzPuaxgR-CI0hdY7CrBHBYki{QPmC+epYK&zYY&sjYlOagWJi{!< zCQ_^DDXFIUV}pb3R+1J~!O5s?bQ({n>v+`hxI7cZV_dAp?1I6@cuLc>#1>!|UNU;sVKDOyfj}m(q0j9l*v6{niY>s> z2*Q}QaoVId@dPIsW7L|imqd2e7AZ80$#xFnO#t6E)0?5t=Y~=ZfvZ zVAR&G);OzTIb@PFlqnsDb5!Qm?d;>VbxrfQxb)jzbY5mMZ!Sj(Jl1HLPE*G8TNUp=A$ zM<+QZxK9sHs@>f~0KK?a0sEnd7_Uu&1&u~>=hU3m^_1wPPU_2JIy9AA$(j}l63R6g zz|T*qqMvGWZlsKL}6f0eOGBv(tS+OdkJlerXsx*l9H^NzTqW)D9zpRGnF%${6oOB0R-X+;9UnLN3y zr1^I(E^P{F8>?<78TYA6C|{Hoe_U0W6z$+{Z$N4#wS47}DwIr~Th!*N zv$ILGzDYd+t8*dyqbDB5qJr9aSotAW{kAhsm%E}x$gX6Rkql~y{o_(RrZxWadhtXyt&&ar)xEP-@PyT3MG*VJUJ>Gf6A5#q^Bt6m>&1= zKHcs7GktQa&Be~$#CBC5)SC2Zu0Vac=hnF|hW4jV@D`g2jLUGQb_4yQNc}NgU)W41 zbfYeRO75nVVy<#c$-M?FnftlP?^d9WV0S(heN9zqqPIL{r1TXRpLHiY2p+6>tr~A( zEjKq(@F3cpnsJo4%#IJB70fj48(_#8y{lJEQ*!%Num@9FPLhh|HiceiV%+xI*K=Td zX0k4R$s6o{oxK1Uu8Mf$iZ7Uxk5gQ=(u<3-Q}Ps1e8R1vZYZ6|DgpNNp`##@@^tJ4 zM;)}Pn@rnNxD^%CL2faIW<9G#tFklMU61>E#&*U@KGkjuw^ANPIP|P&3rRjyVQXC^ ziZBb6C*09eDQFOFyO+kJD-Yeo+kHr==DxUOKQf@I(p}4PJh+D+?24gr2apTQ?8mMv z7}JZ~)>C~6CAekFNg*TH)Ry{!DwbG70o9weV>SA(EW;gzQBZv4 z4qtXV3PE`+iez_4V;|mOn&afw=YAsEtMsbchP2W_~mO`I$ooBDc-K*)by?hsL9?;?UmxK z6js0SByq7zbWjI!Yn{E+=eqtZFST?}pj+A|`M@kW`Hf-T***-RHm7beUWE^2z7vYC zw0{wm98D|W4CcBU9U?aiy^t~O)YW|tLz_`^CiX~y#z3y~PSEVEfFVo2_nVqntW06< zr>J4*IqKGh8$AsIsVR^>YhF8bwoHWp3Vr6IX<;5+&>u?UwQmy7aLpWTvM$)_Yl0E0 zUHKielqoMWq`cOlh9JC>dSexXHm5Yw9f^Thv0U$Lvl4yIYhvCLC>s%ud*ZrP7}7BI zlBh-@X-_>(Y1wJ)SY>#twpwsN3-zt2VMy1Yu4=f8XpO1R>d0;F{GbA*TU8))NYkSC zsEy(QlEW3o`J{BKB4SAGO)`b+RA*(@`FMGcH{YY}g@ z9Fj9pORBRRWL0}966zy$W<@=DqFd0b!2-F{sjwV!YP^@yImxP28+wV>v9EEcVM*Ot z@!dVdgaU9o=Cb6LM?8~O?PL)k&pEBtH56d#IWt9GFu_hwty`A*-qtROoK{Q7@~~Kt zl6?hNxs9N~oE~}uOOZFJZhWff)lE+F;ed&PNB+pI19JODS0~ig4W6ozFdkR{3}%y2 z)!I+>3+*odoev#im|}Q z<|~2qs}Q+~!Ppz?Q^jj02JGY46*!7D2OOH2O$kAp8f!{n^9q6KQ^~BfO|{8C%Cj`s z=enFk(=>3eTvn~kj+Ynu*_nv!4RK9Jn#}EoHMBM~)LY0z(h@=F2TGp(($dkHBLf4V ztPMX?Gu;bIRx!jo`&Qk=7NG~)t_dEQu1UqldYw>`anR)3RlAl#sSq2F8Kr2=lxFJi zhW;&p^{cnDY4JH0>6YjYdRADmx4MwM<>A|(xD9qj*1fEbSEJtS%+z(^e!@Q?VTg4- zO+FYjo0z<-c;u2|KJgvvNqjkNBt^RJzS!RRYe>sYt3w;L3bkEKmk=IvtXS_=0zktP2C zmXiThWRfP3$GkUUYnOiztFHN^*^|QOCX&kTT}DDJb$;K;PN{2L=&#i>llO4;0>09<1 z8zP2gAbJYeu&_&X%K%u_#+Ec3kSm(5Do@>s(@um|tfw80034dR6bg%sR8A(i1IUy{ zKfR3AxM4A%4CT*qYnF0$)te-c<&1h$!qb4FvAI_gn%ta&PA3%;5={vRK*WzyY2i-o z#OKgZiU({^?x?vYl8U~FNp~HPDmfSxz%|e6ULaVR;)F9W?~1jfNj255noz*EV_hD? zQI9hvRg0G@FPPbv1M{Vh+XYTBkJha!;8HxBPCA;&&4i;J*&8V>jNxel>>nV`dghwX zAvoGFPbb=kxIu+nlk6)eX>RUHX}gHTsml(11v$QW{`NCYRm$X!ed*ywIjt3~p-wIC zK&gSy)Nd$KK+bC61GPr+s}9+s(Qi{3(sr~NM6)sE=C0ceLJ99oOLQE#$*G3}1EpM= zwUDs%Wf#pMuJv8ZV~Vqk(iEhfj;2tH%o38=#a~RY>rF9%Pf>-ZDO^!jA(46OS+|y- zX&YOW^sQo}y)l(fYN$_?nMat&`xy}VY4Y({cRn84cL!&3YqIjerkF>iZ45-B^f8qM zEzB)9PLA3T9A&FP%yF)ERkH{KII5RLm3~;sqo1BvW8~yo42mIe`?la#q;V>s6;9Mi zuw`&+%#7-tYdVd$V@PPOGHDe!QOWeFq8}(Ys`18H?m0CwOiSzor9_sNMWHVnWLY(U z&^6t_IQc=XJzUA;a6lQS*lZD~0B}0iuzZl_Msi%zu^EA3h+#2`wQ50+IK?IOvIw@t z9^z^C>Jr=+A2u>-9Gi{VNm=MAVp{^AyJ-21bN9NzNm*hCuUaigi{?$6GPXwH7LmUfE|&()1wX5#{~l&t;I(z?{jNIZnuik{jO3{MqX zP-o>-5^EJT3#~brh|IgaPfFcFQ(7|@H3zBHxmBh#fG-t;XMLz&&YO{ALV6EM=~fj~ z85NY_?DS@_lAfr}Ad}5Y8n*x;t9TujoC(iPDqO~(Bxn?KP|W`TKGiil0^{X0-zYC9 z8K_TNvY5b#2SHr_0Ej05q)9qQp1(FTTDq8Khhn(}x^3%TbFIfUrTRH$AU1w#?ZQ&$ zmF{uks@gV3Z>PY|0tN>d?^jKs$ID#qlj0?~Ws4($T~(~fEJgcttno5*ed*{`2N|nJ zQ$5Q?G;t7e#}%?v81anYR&~T?OSEB?RUn*-(OD3X1F81Sd048ln~b5ctth1Aq(HXZ z?mTC;PqY!YZxrK$%_bW?Ym#`X-wQU4C?%lbfG`IY3r>9BEJo-3tZ`EcY?34~w30AX zQIbz;#i`3g%`TSYQ~5EWd87!wtZ`A>i&+zVGI@!fpyXFI;vW%cvc{20^U49p@|1!3 zn#yk)+S}L*T`t;HAYe3@lSh>w!xI-ZegFHBl=NP{67mpp<71hECUjLtWWi#yt%7Mqt>T>MRJzA zDy8BnJebEHiL8l2sLMBA*sYCkKW$B@3mC`DdkmV-ib!E#7%ND7k&|8YouI5^J!I}y zjw6iVs{MM@7jl;y6rawu<#Z@X>FR4K_OT=`N3CrMUsh!`Wn?=}v)sk;fIFJH7j(r` zoDTIA7l=?p3N(+09 zLvBkP3dZfTTO?=ST*nF&j19-oR%|!3wjqsko~%h2s9xvoSDr<*V`x7!ky{X01&=u( zck9h{S~TPD9Jh0o(A#(bIgo~LO0hl2vrt)bNx)DlHI_J7u^m0?q=HMk)tSf1dbQdttrO?1M#fUa?8hG!kH!dTao5D+ww!wwMvbhhbdiLq_@hmY%B+cDl(sNcTz53CFABaRn{m#1*4?VMo`fzp9;DWH_S&X= zwN?Igp7$5}gok1=Q`3qZXFrQX5tL)_>dmXm){qqrUC&|ARzH~0NO|YKH6(gv+Ad_1 zVDtu|xCsD1Hcz>y1r?$cn|iWIE~F{J#br-=i20H7)XC*CXXigkTRT|Y!xY?e)Yn8} zqOQy&+?|ncPBPrL$`}J(92!Ce4DI-nP+HAvY@>{jI||L#HR(ByAD8j2NpsYdqoLB2 zAsfq6Pf*qEw7C4JHnhHjwQ`f*+}*ina&y|CP4^P-a!Dq%9w7y@NZ1$^(F#$g=FFv0 z&0nEFTD#=Cst;B*6p9z+Mi1^Pq?ZpghmV%7!E+Q*f#l`CyiIgC$sG4gY9kTSo=V4( zN9j)cPUFD9sBUIK9tJXLB-;z5h|Svr^56=oB!%3UZfT{ra4E?HNOO#P*6Jx3w%+1K z8b%6(B>GoHVd0r&Rg!JQo^jf$>E;V*2GXr3=UrW$#nsy=RBgV)70+I*qwt&2rx`s< zRvJshfmOl}YTLK5jyw^TZ&GP(;aFlD`q3@w$HI@sym`)ZdLEkOr5#D!#SqJtA6m`6 zz7G>54j0wWL0S5yw_|Liw)v+cgk07m#i?s57+X9s9?WrC(!+hJ>FRUO66E<{GWvUX z$#=Q)pF9E0W+m0d+}mUgC>;6Obn7U#V{XK7z*ia?R$F5LQ4Ovr=eTvuvIY;s+_&1(dXDFe%f$75Y}h1?GKMxbs8Z0x-24osnGLW{;hk zRhR>lSj~7x@gKsWmh^vmnxxr0$3rFZqV}xYokS3M{c8?O;>6>nR+94^V3S(Hf<(%; zhLhfIJ%3t_?v+MzD=Ds7hfGv^!5=dk=v2{%v}n5}@h!y(J0F#!IhuYCcOq#FfyT3z1rYAOQMZRO4;;wwLuO~G^)g=p- zBo9ilv7dpTtwffk#p!ZSx6KoN=pC>cR;8`DU=K=%7)MgMr7Z+d z?ZD|;mYX3@nFAksp3>SROdq^z+;-OE=5zU1H0mdzyeTpn=X-?m!m6*Zs-L@szHkrL zv}K88;v6XTHIZ|tFPK^7W;ChU=sCr$OGjBmOaiDq0IH_qTj#`K$UfCwEPTn85eXUV zU6z|^6|76Z2OgAElqsvFsZ{x@JxW?;p7yK1VFdRzYU1|N1tH@zNsr349trMOW)m)s0ZpE^T6$G03dM z^MP6xI_xdej(say)Us`kOa>K{rEOC*jC+}pXp%`bSgQNhe~9hiwuNI7?t18TtI=gthflYLf_;{&Bf9Q)5&r3Y(@=(O?y)}!;6I29X4WIuS( z9lF9x79{(cEUX%q1Cvq8yTKIiH5Ey!@Lc)xkfVV@PjohC%cKR%sNTf)6{~3piWO1- zz^TILtveLuR#H}EC?xKHoC<_|(iZLOOCzG6ovKBtvQj%8a2`$QZPvs_S zA?|LbVv5*9=qp<7(o7&y)Un#4fI^DW5>Z!ZnMKLy%&2tFIOUeMEvzDi00O0q1r;Mm zs-%;}MC!^~4ls;G+Msl(WB~BAZxa32#W_=I^~tR5U{dDTr*W|(CCCHR*Bz{D7Ev)& zol9q&bBgM9D|v1pbWk}M?>(Ro-58`JoCCc9FIz1Ct)9xK2iT6p4 zO?1sFh3j5Xq+i8ht8NRC(}B{v!FX>Ge7OqeuoSU4drdX5l?+o#&vRJLPu*P86wK{| zz^N@^+Y0XF`qhx2R|}F_BhZc7yCX#^yqcwZHa<|pj%wq8&UYSzrDsWOC34;K(w#P( zwK1yX*DOgQ-*#2l02FbG%GRtc3Ny~2kSluKL{Z?@7QLIJ761 zaFE@>W2id_!yr8~itMbl+dG&SG5JHFK-J4#=t_=QNLh_br)c*gSlFX%2Q}zoXi&sM z-#q#?D^Z`gdJmj#5GxRopH8(ba6lvsiZVOZduWJN^Em);DtzV0L4#gBT52$HhKHjT z{weBO00GBJ0pOfdW6Gcma(!v-j8};$K5N|TPeHr5&2!!&zr~wChwpXwt-Fb#w}qsY zwxAr7T=cirTAaW_nN>WN7|nWE8Y#=2tJvYiRF}Y+=lY$c)yoa7@{ZM8PKZN(i6$@) zIj>L__Tf*+gC5n%-RSX8dh#(Ug!UEmbnz-nmFRIx4C80HHlVWKTdWq4hREcvYNP)E z2%7m{y+$%)xvkwcYb`Z=wU;FN*8EDB5pNh}J?j@*RP>I<%9LjA_dJ_X(QYikWf@w7 zPqVq2WdWFE*Q?y=G03F2TpwDF&%*Z6$g}x}gPovpT77va>Py02S{u3)EpMXRMC@6W zPD%QjD&@EW1p}a>yjebJHhPg(Nf?^wX1qlWwS2LXKBlsSQAF5{NpsYmwY9BCsdAYa zI|uVMjGV|{9GdQ;-y%7uZer!NsAIZPHjdR=?S%6fqri=NnzmUYhbNO&;)$j?-~-z< z*7soh5?RKUvEn{YbAd^5a>0V9E`6$5BUq3D$*SU5B5m@Jpn3`wo47p{ia<}e$Q@7BK_t7){#zb?9NR^Sr?|*qjPe7Gf^wM_5gMqREcel zOBN&4)HgG=umu1u+cj%l5idJ0=XI6X{Xft<2=M?SS^3s7-SF#+Aj?QbpxKz&%?umiCLVC|sWAps}$a0_XLuh1m*Z;<22RG>Nn#<*bh( z_~#T6_-4TxbXnH#a>Rb~jielXGAqE(TJ*n-x{$o_q*mzrmfG1t9*o43`faZ?u3E10 zHjctl(M4uZ3ic0zHct!cdaleIExI`$@FxEN*=x!)`#E(B`+GP*^JkDox#^N|Ur1HL**%0}hj2Rg9V_i6UT+Zymtoc=u)HbBcgCv2^VVZ23lgY<#L0H~grMDcM z*4~|`k1KnwSo-#^sm2gjW-*lJ?2NmZ$YbNNxe5RNJq)y1eh2X58Q z=Q-?dp$KkNzl>U@GtN3!I?*-6#h8ItFYe~Nl)G`B!lH&I{A;F~agMu~&8a^kXrd7h zldr8+)EZYC2_#m8Q9AO+9@PcR4DFLdUZQc*#z2f9=~1@h+XA${$j6XBI&avSx{BRN z88{RR5V#c&lPrgt-nY_PL!Q;3G-OoyhG{zSh$=2eD3ERKlT$};G|~vr5ua-5S4>sn zIQcbx%S;x{=Sl!0)}~mbk!uXu29M#ChzFEqk3831Zm?TK5pV#mV*A7q+A9d~&Dyzp zy=PIpW4hsyGxt>Dxg8p~hj}|Ap;o6wStGXBd`oYquay|{B|V2~p(WkwPa`zhV~zl-=!ruAl8t<;an z#aa6-fM!BTt3G5=9Ove!-hVLUn!am_!&2OXR#ONnDBI@epQTPAl0))F?erBb_bSYh zW-7c?i!-sP#iXEo;~!d+A|=2XHJi6VtCI#eq-%|Xtx04l3i6ficEvtZA0(PBe<4HMh60_ zEN9ZRo)eBvDvYueVxZN<_99Vt)D$TD=Z`^8Rt#{2DD|p^i%<}fIjXRY2b@HhPnnN; zg<+CzU8ss2dz!lqyp7o6p}tY5xd|PzB6H1HgsE&U0QIGKNF17gmdTiXDo8qZK4peqqVxr?S)Dciw*OYMUVx zWRr|m5~!Al+BDnMl%YuBTr6Oo=BG=DbBuy(7gN?FwnmC60dvP4t1{l^QUhagka36&>|2+PQfR zlG^e}{QJw{8ui*Mp4PQv_J1ZRi}ucGy0n*fu3KXWf#W?Z4tq&o0a@eQnT0D>sM7#OY!TST=?$sjI!3Wga7JXb|pF{+Gd)s-{TB)q@9 zLnKV85FKlu`*e$TSIjQJ@?$~*YlqB)sI6h{n9JT|8hwf!yY#IcIBnE{!7+})wJmMz z(haE_yJM|r=-MkyGf2o$Gu&1+YpXk()1xNUjjI_%hyq3c$6B)+uiAiY;CfTyh8s{2 zRmjpj{Hg)Mf^s3SpfNmzUQfZc!DFTDlyPCFFoROM}lTS$O2|d`-e~A79os+kvHPd%|QY)}HW5r<` zuuY~^xr$jN0dbD`s}RD<2IkE?!_RH{KvGprwu^h;RU*qyY7L_wR>in%;xt1@_7l2qcir5N=yn|%rF?vdQ?A$_Yy%mp&0ur2K5 zk+&y7>sqLcA=ppNPOKw+84;YVu>GvW0^NHGSs)5HsTD*>D!tYG7V(oTih~`^X0G0a zMk*@o=Wgzr()-LOAR5!MzLQY|p}-w$27A~qbqp?DJ%x0-Vx({YiL`wyvZ_&ny41!C znYDcju%^L`Wct+S3b`}}0CQ2zBJAfi!+W!*NTWXcQSw+Ka!oQDqjbS=NF8du<;|>5 zxsxWDW2%Xz*pL+-wbY^R-6IZeXJ*aNrH4Q$l0||NN#F`0Ieu z>TP6}g?RN8k=t`{D}?d3tRCfIfxhVZt7BeSU{BmRQ{NT9-$yKy7Gg^uTJ><0BhAiv zxT@0SFlpD4w-hO&1HJugQVZ6P{3ii!R`h-z%WI%qk;W}>n9>bQKlDw`_jsOCod6qc7 zLzbatfrd<(2e34)9CNNB826zoZbV}(?p6C^ju?_prA0nujN&ts+=|kM%pl}s)RM*) zGI7$c+C@DJuNAxz!I91>(8__ZL2qGJ_D8pfk2I@dIdqYfV4Kj- z+a%`*#Z-5zw$PYk#&hmzc7Q%nk7{^t-JhjrDXFbCH04QMk@=^tF;Pi>6nPAOZ$V3j zX!>zfniZDhs9scaS;mDqJx*0prR{SET$1{tN}IY7nZR{Xzbf~ z8Rop!;@n)^F$W8i&{wg9aB=0fhB%6DEj#LT8hymKp>o8N+P9?A=4cLAj8{9MPA%@D ziz5TpyR?~POjbCFI7{7H9XNEQ9ZqR1Z0&(*;|0Cy+*gRO0USqfttPBf?FhVmD%`$W zSmAc$uK-lJT_*aEkzLrMeW@AK$%Gxpjw^b~Bo>IJ$vEb@TT4lxlrck$@=bKG$!;S5 z09l9hsIL@~Q4|}#rlccmN;8~MNo>Kb)Ps@F^P;#~vDqDwm8+vkcB-q^F_ZY3s}`Ku z$sh{Wy&y_Iy|?^pK4)ZBIpfsVR3$D{k({Fx?#$gu(Is1Wa?jlJ#agqB<2$oiFE;7U z&Rm>ln%sg&;g>zD#*?00XC`XC=|Pxn?I31IImT-Sa_dZe*29p2MDN zn44Ih=3x{Og!TtGr$(PW+Cr;QLq>Fayyplo7KaeVe7Jzg>Oib$W0^=S#hCp#sjlx+ zPlF>3A@=W7>1$+4sioPYdw7P)c4A*y~Y7iJO&-^!@}o{cLsC@#nE6O{@&X+A5@qAppB2-6fIV22yRB=S zH~r1U{z6XT?!h#=zgE}+{$mOZ!b*H%*bQj%d6|7x4HtDXA}2FksqSI%JldP%{CD<4 zE@V_;DjG3u15u3MTZ;30F@)JfsbV`+)>3olgq9*_XU-pHmgOHgy{1dPmn!eMVIo-) zpSBY_B(HE1NNcu*AfN55u2q&OXCZs8#Q1(%)+)OP7wD0~2Af>_o=xb)HWQ(OjvZ9$ z{Kcm7CnM$rr4tf2Ie(<|51Rwl#)8ZTI28m{VB($NDxYA6k#v)9JY>Z8A@(-%P57Q( zCeW!jx=7=qZh6o?v7r9@R0b5veVpVcvLqwyg>Ga@G!rguuJ+hu`_-^~K2OujDw4=L zBJ3s|=V1j=eN*%<2%%v?oQZJa1)5Gd15{hw5L`=KlEpx0^>}gThoaq_5FLF^FmdSH z8X)xDOLYu7E;kM$1zmtg*cqKl91-!J8Q5W(!sZX}Q5BUlig)kla2@8@-DS1x?c92x z$8OF|RQ9Wx!+XKw__{Io@icBD5#g5CdxmKVZ;(44e_Oh^ z&{2E6iq|Xg4WM%_-V@@G*ZVJ01BRl$z&QDnt%bo9- zQd*%k)S+Ip4UUwourIdEax>TYUaF~sGx@Oy{7Q{;55CUyKCp9t2zH!HY7n;bMzkU0 z5zu%OC&eMffBE5JT&Z!!jJ8C@z1Y&>ksr-bW6TtF(pT;JeR_c`Fr5 z)-x0LF>DNgowka zo?MjnDp%D-IZzbmz^q-?UT}#RX(sMN7Vw9$G7P4r8S=kdtFn%S< zJuSyoezA*2$@qN)2>;EAOzXRX z2*0{Ur*)0NX3@(**1XS=2+vCK(l_@hA|b5+_OW2K8Cy=X{RtHs^<*kXa+!e);!Aa@ z+X9A$2liX~HB^J{JhC7RE0Bo`D!jLdNL_Ab3D>lgq^013uq6z-}Sn>N#$KMqpDfDE@v!DS_O=v zj8<)?GF0;3DCHp|e6;xIW%L`OAJonrMRu-qmHZ=fYHY1W{)sc2y)8+P1fk zD$Z?j2Yl4V$iLDxg?vl~<dikt++z1+m40S3W4vc9(*QrZIhc_3>`{|64Y^%@!*1vycXM6298`I{g0GY%;0SLWUa^<# z7*SkS{hIN`P3FETjUaMH*xqBli%;~A=cLNYQNBmk%R*A^9tgaTMn3~KACyFL&CS#@ zL5Rz#D(j_urn_qsk3LMBn{I)WpK91}J$5o91kPe5G?SwzETDgFj8Wg77 zmDsq-Cze5lft`q|qZICf5rA2oG4DejDc`FDbDr>sNfvX$0x=rP#Aj~#MGS){QnIJT zP-}~I?h#>*n4xQpPWdgxee-IUYlwE6HuoF5vu~1JTynwtTo9xxA}$)wZDDV4U($M| z!h4YO8q|?%J?SOU#vhse2q{3BXMjcX^BU0zc zb9x2VR(3KPZ+*8^W9&lM|Tv{Q0IF{7?U9-o;KJ}{#2n^tXlWl zF5|8xs%6~W+#y-w+<<39V=hSOAA!^DqDxLpjeT>r`4(Ik;A26&NDiRzWXB;5;oUDq(< zf+;beXN3||b9KsdIC^;aXr^#nU-#~J$2?2cc8Ee?NI1hII)XHnR}c8ol{LfYU#^@a zWQ)Sy;m&%}Qj`;_yykh%W22H%ZqO`2tEaRbCKhfZFb~bGQ;8%lqe4HA+|eJ8SAopb zsZzTRd}=k9eq=t9K;aY;G@_Iq)s*uz#Mu70F^LJ3bqCBtXEh{GH z$H5fe7VN1+0#W+-dcfOtTlP&s8)Sn!~MHFs;KX!CDeA(v$}KIV9uXrX0FRSwKebK zSBTAPS@?1I24q)GpKWTohMbuz`##&J?{D6(njtYKkCp84qJ7)*lzzZgnA~_TbAco~aaF!rQDJ!i=S}})7iSKCxtm&;hU26dIHn?y zYQLR#xWe9$_9Hv%Vdp}P6~3dAr1VA*j6H0DK60G zDhr`3pic)VtIns0TZn475 z<0L=2D!w^d&R=tcyG(q|mkD7njFK4HvvDxpYIBk14K6URrOiFz;Z6OX62vo5nF_W$Sz!{!T^N)*7N#8UK8T($E;$nc#4<2EdlSr(n%T~ z<94wkUt*>E8ndazcRa^p-adnHi5Yye?a#lYszM?GUF#~-n23EOO~^F9h}2Lg2FWl% zadMgY47?Pjp@Abh-Ep6Pgu60S6N2ec0+3;&c%9Dbi@l z49@$HL$f~bc=PrHQ%jw9i&KY&;YJk0pV>rJ_?xLh5y_ttKn(n{S7+2P!xX;;{0_Tu%g7tgQQQ34JLL#W%jR%_& zy3DY7JQQ`?BoOJCWP?MR9l$a_a9!)FNN1oPG7G%=@UVMAPy4ucL=029bIeuyL(L-F zR({xn6Bej(uL(xfh|Ef3g51X;x%wEIy8){cKW3~I9*z{pxLJFjUV-=+d*x!($3(qPPdZWu6Q|;k+{s&w3=ZKbB=Jfiy=6xK~j^jqAYdt z!>2gxKx~UT73ecs-hhM>BelB|B`=VZ$cC(+b{{x!u3Chs%!b-M@=Sjg^3qCkSmoM< zlfB3el191Rg8M-UcY&*MxnpVcz{@0(K-7$;l38e`hp%r`VBb~Qn@^TiU zs~-Yv(QesqK*3O&$3ak1&3yM2Q>L~8TC!Ap)(>F^fwq;(^3)dn99WrLoK^Ch&*Zn^ zyp`on$_7poB!OI$F9YJX$(n-dbH0t4uRJ4#c7zdVYwZ&!yA06B@@usl?TVFXp-i%k z%(Ktya%-=0bPS8$b;|ZmA)lnJoe7d?%*QcB2#3{o`?`8tgWXWog2_Du2(v6!x~Ak` zIzX~;Q1NWy+RPXqkQbz;KdshlSu&TamLbHx zno^Xl6M;BqdO6uU20^3zju59b+`Ygh^w!}t(Fc}2)@WVyxwzp{DSZ?RUtD7GBp}VD zi=Y!~objz6N^F|bKoMu)MCDT+^%K_^Jvp}&cpx}V5xS|>Y5Qg|d_&lp+_H(qYKjjQ zXN+ji>0FojWQdjcz_{D$u?Es3Dg|!6$@1Eu&z(?c#8Vj)SG-WMqD_?{&BX7+%`h+< zYc0_TnYY&tP_AmNYMP0VDk^tm?7^hMgC?(ZgGv^4GIfC_QX=1kN8Lh`IKwZmM%G(D0ug@cH%8oRuX4;XDT>R1 zLNcWNVmbQ=(Hr zu2sqKs`bia$onO@3(f9eBE?fj@KYAk)r=SRt@@vznm@b?!(D~Pj+-~Z!!QgZmI}ef z6EGgl+|V6vG6ct$Cr02@h9o1Nd-8JAE!hqBS(8-lCiD>~4>O~<5|@uf*^Dt2xPMq} zDu@wf;|u$$>(xG14#1kuflhAb^z4DSho+StX)O0eELSYx=D&4?I( zGpr`)8_cYz(DTTH_v2necl0!e@^k8F#}P%MA4BMj(o`)$*%X$&+cN6K9tkFV>Z;xQ_A?6G zyjKq~UVxMAtc$~vhv3p{v_|DL)7^9*%NBB+9#6kZr@mW8CYr>6XPGtat7cEP??#^? zIZ2N8bVFbJ2J~KCxEU|vxWHP%b(P{FyT%xyZOKDDiYq$Q!B`nJ>{C?{?lyWAo9~aq z2MB~nEqKxE-y5u?Q2bCVOlEzBoP!e;XrGJxss(6&XkBOiS#H^0w=d%rTtC#W4;D1K z_@-oLMQEkcbT~S!YfaeFS%u@X5U)IbH9aRS1TU2|TN*t4Nkq;A*zA#uamT3-qW}8d zN__5TF#_aDC_On|UM+qR=UgP#vU~&|7IZjVK4R)~njaX=tuF5SbN?q^66s$yIK#$W zale|N;*+dYZrjphf2yjczKUEZ+SW}pekJsj-5Gt|ltWAYEb>*MPJ3KeS^hD?Sn`F7 zW`hMuk`b!~0rl|6QB{1wk|@SxR?*MBz>k`e4MG|VZOESXtKenj!aEmt574R<+?)czIX+6TP3wdSsxeFG20TQ6sMCRoATnohY122KH*VYXe z|9~O98(5_?+Ef2PK1>y~x>ezwi*x|qFtHZ4Be-v#n==!rT3ugRBuoy)`;bvF+$5V6 zySc z?W24nNvSwzg28ybbjZs_J##qLyl^g32XeNQ1F;thh5u<>pdC&XIVSgE0kh$W2C?R* zZpMd3Owcz zK_|K2KS6x5L0y4+pCTK5yndGUog z|IvVS#taPaK?$2Gg!&kH>J?P^%wrXD$PrJQr@2ShWJlcPjJY>4bDtD4f%S;oqnNxB z+)_%$iOmY19Y~+$)OpZOfY`%XWOhsGk54^&!fAVDsNui&RIJ>l4>BM@kN?v*uFK^f zVw=#CUx!WF5bEfGQamueT2#odG$O52!L#^6R-(Qe&#Diy|B2#Yj0k;=5tpmU<01yV z*QG7&K{qx5@w#o`Qf<{Ys?=0-!gbyC38~>Ze6o;dGx3T~s52z3^BOw~s8dGCe{$y{ z`lG+;KoO7nPJD3~XtJJ*8fU#T=Rj>2lIg%HY(%tIhZcrWk}})K&~`j6hOAGau<$Mv zGcMD$utibT&?@uRvGgiolXIbas2r|Dn7-%Inr29Q@77~UulQUPX!ztKFG>@F-Q^j# z8ebjjj*h=?^D$KZO>&!HBY)Y6szI}klF38a9*Hbs2o;m8HIJPwTEN9f=&o{sxx)D4 z77#rfDK@m1N3_J6?u?R8;c2{Hnqm1MeTxvm-D*DFAL_x3ayG&ey0Lb1b2oN8%aA5M+E!#i5NR zp&&s|{sgI@M@b{gWPG`~R_6HSVY#x~)rMO&SesidKrhSLbj%MqYzx(iGu)n!WAHUz zR3N+6U=c+2VLNeI*jBLB?ku8x0ox{*u}m;T2FP%a{2KlCf-);t6R5(uzho&K^X{}w zZCU3SboL2rInVYISJ^zoyM2HcK(62eE|5dL7QvF_{a~ zuw~60<6h0z+c^I|^~~&x`40PRW`CwJ!*ECI_(7PBW?4zB2wjQu7dLD6=u&w9IGmS> zZ@m}4Jp*Zjd&#na20R=63K2%Y8bW(-`h1L$Ka%78qcp>p5|=S?b7fJY19@DD{haoC zTlsoE1=bueU(aQ$F)m0O)ylbisnD?vznN%Hi>X*YS8R7`g^yjjVvrI0b&v#PN>biy z#W}v5oYo}^&$=dm(lm^LIytvtBASjhfE0F!k!Pfa`dO)XN|=Z!b*J5VP3;RKcJRo~ z3D*pUINMAxdvU=G;+fxFO&V}Q8kBvGXPb^48V9s(C&7aR!U_~6R}ORbn{_lc#c{A8 zj)*G_MfRyP`@RE`%%Z@uq)B9wG{}eZxvgSae1gNwoL6iz3#$>~pSP z{la&+FUz1=E_D`3?oYs%BN&>6UNF`mk zWU`JjA!;p1V>Bcx7VW$o@)rozi-9VR4vAh@-NgrFLerwqOoxRVasvykfD= zW#5F0FR+Qf08ac)xg-AWK~UXc);mG2%njSeK1XyCkM(Ju1`lS?=?Ma7i+fsHgRbb5 zE2m$uHl8bSpD+sMFBfaI7rs+E;$witQCxS@UlvUj=-770S9hrT$Wnx96^br00*78 ze>^C0q%(+(Nv?JCNJQ(Z7a-itUzAv$p5dbpJas}%`RU_PIYameIJlnb&E?#cg{i7K zL*Biq`xNJ6Su&i!L(Yi$m6l=g$EHEW1$vEVe!XgmORdq=jOvQAl->KXgSE6mY3SjM zt&SDzB^0RjYgI0JO}x$fHy~TvEOp~8OYx<{{KAq}le??H{v>WmR%dhUfVknVvXUhN zG;pm6`d*oiim+4^g(clc!n`^w&?FIzlU|@n_-4~oS9$$x#Q&4QDhe!J2IyB@}Xb?BVdnpw0#GqCLTO?{zq$SPxTke54nKbYg zr;lN0b`F2qU{T-;Y8zuEc@;*#ZdN7DBq%6oii#gNTw7G?i~u>+o1_UL#7XvzfWJKN zqj;}0TjXZl(6Zk%=V4~-NXr&fuv?m_t}HaeiNfkrHwd&2ZQTlu7%4Y2m>v`%Q0u=p zmQ!H8v7G4!hs|sckD1D=ZS@JMFu3x#a6QJa$jYPnaL&>1E_6v55oadw45Gr9Esl!KGU|=<@(Dk;w8S_m0O@vBlLkS+yW2$g$P% zKgJKpb&i`Z`nkl}s@_OppTVGMoPR`QSzyu8G{nn4_6)C!da%Y{n(%x7-SPXxlli3# z*a>j10%7Gnaxq)InL65wgmMZ72?H+@*vxfG_{B6HUzZXS3E z!Jsy&N9ReA5^>InszaDa2|KmG49<1t5HzHiNh1xmueKHG7ojOTfrOY~F0W?vk4-5L zaKA@xgq<=|7+jF~I6nQbAKScexmkQabhx-z^Bn}i zgOPg!DyfGdq6OaT5!Yhx&_4g9{`GjrIb7xF@bOzG^(CD%qB!#v0rnC9B&xfUv5d3H z3WG1mk#ijGdevip1ERAMMFag%?&WGJMIpDO)88s1RcK2}>12JV80~O5h7L|Q7YFndYF0G-C zjTeBlt4uvilu*35sR*Vlmg%#O>}?2R;R1`KOQx(j#>y`6D|V!Tjzy* zEB^4V7y4=Vbu~GakI!;6NJnu66`5-ha~uUyXa;({EP{KSbFcFo#17|)nvBEv_90SXs0e01+;A!gGaC7X2Gy5C=q2~sDdu$rWXQ&D2){F1Zp^7f&M!|#N$`lX zoL0q96d^!a;f4fcPF1JECEHx1-hJIbQpz0qyi{P4+72^$ot2#mN%vle?dZ#N@_w8a zG@`vRxx!;pj9fsUQ**&fr#_90hFo1++nH%i$rW4z-^@@@V9CKx3#-N?T8%4ECEr;o z-w7oY_5kI66*BcHlw|8a$awI^@6oEK0`SbxAc$@&@<}V@*Tj<-bU5$WM{D1inCa;vMc|?YJ1-Z0UY4Y{$&JLQtq)Kh zVmvf*fQB59EGbzL%Q(z1cn5WT1ed?p9edWXh%rt$0>|H8Y{O9{%FEu(dzmJBE#6uo zHNtIRYTv7^QN5^1gz{#jTy5(|huyR@dRs}Yi=T=*QB8OgJQ7dd29hEp$sEb9OS4i} zJwp*#WKV=XBBkcBWSCYKQgZ{!rQutvA2!?4qk0cIChoP7=#q_mN>6Feww&sXtAkuf zRn4c^QZyra+@Gw>iwlv`7-h0(ZZ!>y)Vm~000I)duK%fjRjhgqy_MGRT2qA6Mk(Ab zOF=VYfFx#8)4E`xkJ5I`IfBz(C-q5aBhK34Qy0g5a>3cET)cjbrs}w5(U<57@pK*8 zO-c*aHG$hRSL2hug`hFn`Fj36%EwyZaWs%Jj}*3zJqdZ_mX(%UeZ#~eD1Jnjp7NnM z+=ptpr@doJ6h~<@GC+S%9y_X+Q!0BA-=@{6;02GHM4FvN?O-PH%Moxa)iM_t#y3u6 z^*Zfcrj)0+8Z*r9ISB3<8p~!$cRyePNew-Osg1$7_G4AMkV+Z00n$A-r*sE&ZjNe@ zjryc_7LwdxfK!4GwSmZuPi4^F1AQ{SSTU>>O-M zFS7XFR$9I(blo4975_5#lHLM;3KN(zt44eOu+coCKZKXO$Pz;&OeAr<;*!|{p+OWT zb$UMo$B>N9&NJf2qe9+_k#C)~b1c+p=2SiCsSt3(Wo~HCJ-^3vwpCTv*M${a!B{m3xiZy3)Q>8HOF1CwCA* zQY~!`WHeI*o0_KQZo}#Z)_hjlfgk$sf7PF!$$P2BJKw^EHd#*xt|p5NA{@q9c8RY@ z$evYidMBdF;9SX-vod61{)|Bh^FU*C@=2$CP^UKu$3F0+nwqJgO?!q$SQ~`g95-Lo^Ko^auIBe#Tbz4_?=aQZ;f=NR<%vSr zrnQ!+&^^kzaGeW$3tm;MJNy2TAEZ#?hOngKc-!nY_PZlC%KETAn_nJVjgb?ZK(|K+ z8Ft#7!(dUqya(pS!g>`VA`9dz9^OCDqkHM1!tbQvYt!U1=YJg~>bieEGF97J>n)6d zEMB0XA^#X}6`YpRQ|tG^tAbK)^qJ+(Q>%9WKxc@iwM?u%~q2F)zqq5CoFb}|vhO%cV(j`VF?LqR%7 zVrlelDN&&r{U?xaJ-TZ6@M<1vcS^H#wk%&;(^A0>sC#cJvkJWb zYg}xEIeDf@EdjU#q31m)y5tbbyYlQu$q_;7#s=3X^*p?-CsSPq8U+{*RPiC2T^O20 zeHl*IT`%gt-++vnn#0a{{FD9rAoiOqf=~!$J*vUPdj3_)%KK&T*S&zn*sW!&^?F8xj^Q?4>DbUxVa&$?4^iK7eN*b6WJ{t-rHq%l zSg5x7Zt_CTXGdS|AqbL3l4nf*`O)JWkfi>8`dfCI3<3#0Ip7U*Dz9-dy#QP2vIU;8 zA(O*<&$zWO>T?8vdW~)DYGalnELRiVqY>s`R>*RtU9#TI8#2##i}HPNzj~0J!^=$? znLI_ExqEeQK-<7#L24vn_$%rf|Jj0+&jE$cy)~fuo6rwxkNY(k(Y13NJ|KuCh7g&Zn6o>lUlGrr?BzNpzPI z48qaudIN2-D&wO}#(OWuo$g*(DD&Cz%-sTK0KkIn#|b+Eo50}L z7G1voP;o1JxPdxrG}EfuK0vjNbZWBU3YFRY=X2oy5hmwV>$!98KcsFzfl>)w{Egw= zf#*xWgv%r_@yW1|*;jq-w_{c)r1`E*cJ*?JcJzI>RI)RMcUJ<^E63QLZ{*WIa)%Bm zOQkR|hRsc!J*jBK)W1wuRyrdxCvpH2gihrZpT|kJkCvD{c)p(^x@75uHmoQ@wcs2s zYK*yl$-HPL)y!t*=RJzB{wXoE!S}#fxv-TPYo_U|RpO}4@Jn;~L-I@d<^!oPp3&jW z?@tLNBlO+C=^K-xb%n=T3imTWq4&7lJARC2MrS#N#S&X4@5P-3QiZ+1dRoG=)@3Q8 zPFRqf=D(S87~9ZRqr$IZoW`6-pQL@yjA8pkitcWEzTWJHN*4WNf``sy-M-4hvKkj@ zM*!@Xs|@$}tD4ncZj_4uIm?NvE0UHE`_Q$JB0nKTwI97$-zNjuFtTs55}FG`xqiFT zuzTPh$OStOSb5WWE+CtaElC@Z)SwlR`#bS{N{W3&7k znSd$FyX5HuK0q-!Z+6>OIlVG}dufcPITEp#s~!A{xKS}JJ%#>S#R-2NXISU6*9vYj zbeD-*@gUZ-yd}>CVgqN`*y6hN4QNr?%|Ru67Z%-n1PGgBw_C%gP|-dTXzRskZ^>7x zxqQ?u*ⅅ4L|F<&{H5MoqXYba&>EnylnTrqp_=Ni@ki~g)}r5g6|)r71y@;l!+dH zCfnpw902jeMJ>iK4R+ zIY(>rQDb^3D)1t4`-25WgSMxt4dk=SKp^EwXU|Ss@y)cPT~(_$#3B~E5|)(W6WfO- zEB&Pya;e1^WvYb($* zkEIJVjCABqCVNfgzYEuXicvCo)v@#Xp+JV?1A>gBoY`a2BEN0=J&fjn4a+239%5!h=`mqQ|o?QcbN14(iJ>}qh=O+uK-uk)&8|mjrcRrpgn^=7Q01n(g z_l`OC1Qc6G7ci(+J=t_#$859KdRjk^uX``r0gWVT`^wmNVb?q9TH72+V73z2p3DQt zr9G+ZHL3RXal@oK9bgX+XnoQzh%r~G9q?h!omxIb~y28gs=V;fs?mofN zA~&F2z@2?~UFCl!6KyEC{{Y%<__&^kFk3tWpq^gi27A9uc%^s)GT*(%**%47V+kBQ z3UQw*y#?D9bz65a#tJ2U9SrAxK$P%G_VRA{C7t6321ho5v(G?|^cZyaah;1}7pn-d zCSMOe)#S%)U z7$^HPuX8W0@w31ZrNA|E=QVgg)2R1N=7f~+hDk#(31+#k?E^Rat1uq7hCJ-srVwRz z*}9B%Jmk!JjQxsdsE~tEX#zyM&1}PbT2|ksf2iK|*S;GNwgUAw3D)e-@k^!~&;|NW zqJcQ%)&9K?0~M&nFML_69Lq!Hulu!e!+1Lla@uASp`=%FUk!T+l9}AdncL?u z!uQo--p*&nH=tIfB3GxM;iR)u8^AYHr5g}cVH~?CtU4w*_sL?R7tnA(WMCreEj0rY*YJSRi~xNwXQ{T zdShL0ujx`{;Wa=x(@hUZdDw8N?D*ZN7uq&LJ^uRm@HB5eN_MtA`~4*N2DFg(s(Ycq zi}CxauN>npPKVrvXzBFPWZ;aExJv%#YKlt=HJh@AygLUOF#wDzq#5o5chf5YP?!R z%4TltAd#5_U`Pul)fvGDw9M6h-S?JrB>-qFA!Q{>$3rTlb@^eC z;*Ln>2I)`6kLRN}Hz2&jFB2nKC0~_^U-<}()|qe^=HmM`&r>O-G4`c|FIIjsH6>Cv z#@n^bxhRj6SN<~ni2|!}M8jf|^^LyPsXNwqt5t|Qq!6W-05Sc=!63sx-|YeEHdbat zPYoqdPVXS{g8N7ZV_$5eWk?YSC2JPAoOQWpxL(AuOU3R098Fz*7)X&%AUyhf1IqMQ zi9YrpxL9*Yc9G2dYTZka(pmq_xcJcTDz%K)D(&s+O!%d7($4`X^A(as!){q&yp5>w z!_@(W5B-xo?mdJjd$wQ(_cDI(Y!su)9~W$-n{T-c0|xBL*HJEB8&MYo14o3+NnArD z-Ggf=Nksb5yVqaB zFHKfm2MvP^arr;GPhyC-cc;!K` z*DWK`WbE*9qYFlTI&V7tfPa&8m!=S>7o0UpC069Qd+y$zk30FObp7Q=t_|go8xXW= zu8h++I9z<#W=9bq`DP%l_q+5=98EeR*ro3DAbs|o{$OE6P-E8VaMT=08NDTrQRNcW zF~D9UyQuMnI<#p?wfa!0p-ePHuf>`2*lkHjUijjbF~)W2>-xS6w&#V32cj_USh0cj zWB35_WerKPY;E#&Io2hIcJvKM@M+)0W%4?NdV52H}FSK)RMCzXAE>`5yZY#mv|7SfiCU z-KOvxP#n9`$^xrIJZuJ4Zampq>FFEeUgk8 zAE%!JWd8!Ffc&xqwx1FN!%odgcYy>xuIjlGFk-{1cI@8}G@!NU?#s2=J9h)pd*i$4 zll}(llKk1?;N}gexR9V13@h6<3Ef(B89%9NFftJzLuppPbk@SUrYbGG=4o}FxNfus zhKXeNW9Xjd!0BM1gvbnQ*Zn;9QFDHFivAd&sW4w}JAhm1$wJSb7x_@v6>yHgrfS?u zz-S(xvQg}p<{>vM<%-h*zS6tb?PBkyzq>b;eL430+V(^=@Iz_%6@43!pDDGiX9DHD z#*bO|b`C4_2L`Ug7ta+f_<1#FG&lst<&w?WkGKFNl4_ zzyCzVjGtwL)B0|(bon!T-S+dbxPiDz|9r~#WVj@84&ujHr$)ps_}4u}#~tq1ba{Py zt8XJ&;@rF~u6_^&ciTDf5g@9UOuB+)zbqE|Ye_bl>U6Xz%u#&M9}&NXoX&D%5X*2Q zUE@^*lO}^#ozfsL-s#oWoYpQ}RHJmd;}J*AfUcF?o2K@-eY?Y?B+&9S3f$kQqQ%hx z`SxIgM0>>~3{uig8G{g0>4z9azKzqt71gTk*eJ;Th?It9_W_#gU6){~EP$#{O z;M;+3%$@LPv(BPuXc%k9_jc6iDHRbgKtu1|>pR)9}O#KGr|MA@oNC!w(hos23*!^y>$Zya*+T z5l3DSL)4ho2+63I5~opfKtcrcG7L_GWR#wfj2tj1FX(f78!Apj8Au-ZA0tkLp$9Hp zl1Gl!MnVJ!WC(zWqcKCD=93+z)C?ifq^E|U0yuwNl#DpP5uwD9LlQMg4QYbY^n!41 z+4{io31E|!CK)w76%L|Fg$w@rSpneY7H!!-KK!Cd$#~0XgdFlMcfbdL6#zzFLJA;b z7zVIIlZ*_Pgx>vyqAWrVDUK%9^Hx$hB#~cNga?!+JsCAMEXo6&CiS zE zAPRpVdsGHrky7IP#kQb*=1KNz$p^kyUTHU=fs(P1v4ZP?qMu;DdK>l>;bF1R3z|TV4Kv+0*~lBY%nZ+h3%k zlWziN6hMt!=e~AEllCL&GK?7OdRE}jH9`(8_BZ4F z$tMAIgJ^C8?|0=2{Zj_T-y#xla}dq1fCPLM2=d!lyuBwC_(1zNg?Qj7kd&cgK|zwV z#{EK;SLzCrC5$So0pCvU44x$AVa~n#(8zJK# z?)dA++kgD+;`n<&{vM=oO&2+21;rw_2l)J-JoxK>#iZ}6f5s&KsYuG|b>eTfQ_XR- zcs0s|yQnl^ySYAs%n+@T1Drd+h3d)GNlmhn>)wPhk97{V6}ZG}1t!e{A|TSAUZ=v> zK3Z5eAk;ngJ&yl|_;>Un_Md9(uMnhv(fUWne}#B3|KAXQ_QL;H`tUmaSG!>R7ue-L z$MJWF|D2D1$4~z|vj6|fbkvv6zCOBL;sgq-Z?%FC)Y)dw(JOzlQjIl=+*?%zu{W zRtbNXXW(y20IHY25dV*4{u`}-N58;o)?YLZ5Nr6;*!UO8{}0RcAIkn$c9QHcle!U2TxFPHq?R7!sY$KUiQ<4L1G;gL3_77t_7JI^F#T=C91|d#ZQs(@TB> z0_g$^E=H*yPXGuYz4uoFbLM}*IKI91sR!0v{s9AFWZ1U7P%1B}_%u1gUP7aH&p$le QgNrE4oe-Fb3fxTm4>0bR`v3p{ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-25-Virtual me.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-25-Virtual me.jpg new file mode 100644 index 0000000000000000000000000000000000000000..cbe445fc27c2aafb70667ee0980df9617f8e2661 GIT binary patch literal 5141 zcmbW52T)YomiJH7Irl86?vv(4>nb$&x`pBuR#Dl9!xCq6$csc*#L>PD&#g zBsM`npj%P{4GiA*eQ&0w=1tAaI<@wx9cr)htF!mN_FmVs*DC;(2233WfIt8Mx_N-> z1waWPCL|;xBp@atA|fFnCMAbZkdu*-Gt%6qgs?ENv9d5RGvDP9EO$_LQ?7*P=03J06 zObxp31=w!BlK}L$0sqfHcwl@2LLy=kQnH%?^;7^J2n@!<2NMwB>{|*0mcp`?_wjFPen zOjS)?L;tCPp^>qPsjZ#;a|cJK7oJ|;KE8hb0k8iIkBE$lj(L;vHZ?6hBQq;6zo4+F zxa3o5O>JF$Lt|5OOLtFiU;n`1(D1j3$*Jj?*}3`EHPrgX=GKqx9n7!8qvMlP?AiHm zE)W3z2i8sb53v8lMSa7ChmQ}&C;H6=!t=dxFf~5GJrP2hN4iAT9=Gp{h7v;_Cx5K& zB4HQPL(|&48YiXW5MSlQ{6_nm?0*Ip_J1P#C)mGp%>(3M(9O>SQv*=o^qeOzg!G@D zFAnIcC&N?#LB}s#DvY@Vc`5EC8_f{}iw{a-cLpFJ0oqVslpDEmwOM?iwknJM9;tB< zV_AS?)X@h72SV^fwlFGWQ!I%Bc=3ClRYfwGlPF)?=0}I($IshZB9??jEsRK^Vv2NP zU+y$lQW|1Eud)c&6n#8t23B3(HbK$%26mqbmSIkw4{w_)>Lzx$;}SdVbm-(4#KXhl zvWeQ@wI%Rp7{t{EQ&fZgJi)9g%6QoYqa77ZCRiNgIhg2qh}Ds0Gu1BJZkW4(eRQUi zIZNR~9ZIO+Y*!1UTmrRlf=Sl^yQPFFTozNA_c2|qRcHASlW`3kJuqym>Jp-V0?Oid z_WW_kAAH}DjP;N^0WL%*={(JR4fuFNhjUZbxc|)S-}9NserKGsVIh+P;FM<hX*xUE@E-Gq`{s;GJ|(0`Y?kirZQoc zzSbO^mV5q5Q1f?5XIx;aSjX|5t*d4vk)whLmg7B(?m)*}!8NevtgnuZUNwKU zCm?=lA?@H94vR-tpL{3qpAQ=Q%jM4TkYu2;h`HqJE2?b^y2%N>H=CLZC5IdIIv>TB zS-t=P!qRz3thVV^FKLX>diBIL?k_Ls?bM8MwG`2 z6!_9tb}Zg`dc_qY4;8mF3R%X%r?_0QTvPC^PT!`x^$^jPOgKxDIJ7Bnnc(q*V zY3Le6vFP-W&2>g%M3&5p-Q#ZK%B=t-QOPpg79IYM)@+X3?cAR6?v*_MhgR%GVgG1D zmey#lm#-BFCo>rU?cBrrHGQY2{QF|p{Rn+JWTkw7k&!(*vTIta<3f0JQB1tN=NC7G z{P0~>qUq#l_k>1$d?1Y`(--Y%D7AN*wGF4ee8nD26%}(@#zy*s$>~dWyfxV4kb{(D z;E9L+j%j=5$N96FQ+Lzu{P>?l>pave>iKZv>91!TQ&b{f=Qc$`v8-X`U1;sTfY3v? zW1f4-)t}+vW83DHpD+VGK7R~3Q1tEc4h%p1bbk9;-@(8|1KSUo2YkWXpHpzvicBrO zogW0p+sDgZy!jd>`Xk>ac|TzKXE0SOcvaS5;$@|+aUKVLeO%KRJlR5JO(a$>PG(|n+Wz`JNuIV_J#Hucbw&^{)?0%cSGj> zoG1neXrHVAOVs^()UU1#1}0S3*gv49P7T41T30`uM74}H5vPQk2^G3P1WOwQY^G~a%5;H%TNL=x4qs^Nxm zU(V;H#o$A9RRD;d?r(wQeWL|Y61p#RF-idZ7GL|ZH43f;1)uzv8S9&m3_4bY(5Hy! zV^>$Cz~UF$sO$$wmTti+NA>NCF2l8Huer$mgT00AySWskd=lB*7d80^1}49|NW5{F zb}aX^7l1H-6)=p zI`>~y85Qs`R=BOIw4=p=Ke<=~hcr6|NbC<$MYi9bfF-Y*i*|ZPEJAYh6YA@=3`lct zh3J=niTatY0UWymW80LnjcT}piaZ%>W>p4stpFQd_1=GG##Pf6%AvV8OL|^xIbPBF zDBs*E-Rexg=$ooeAk$fg-ZtfE!Q@7WVz?+eBCDf${A2Rxy1s(3%OkZ)+wL5uqYqj%3ucx-FDkRG&hl(= z1819UqGLL>llN)n7M0|Js+-(46ln!;vU#SSZuzsy2)#KF4r6c(CBg=qM8COEt4ne3 z$1oUO1MLwMF4L}!RZdH+e+4dTY*B5z=0yG*xEtpFVlj?cD9zo`#Q!8Puvwx z+%UM^6!cafqldWR?9(##O@36n8`b{QAR zxNv7b?UnfbVpqZA-G~|(qZ$>Mdoug=ajpYzDR!i?Wx7R0az$VAF85%7y zg$-OWEc9C5&AcAlws?(B z8Aw;){p+3fWQP1@E?Zl_T<=fqHo)bFB{@zYx0JzavMEz>bOzeWalV3vg3RF_1&-7Z zz63xwgtyZJ#zRH@EV5Udwkljsz~fb%!VJ%^y_w7WW%RDOXHq`L%B}ie2ffz-ZE&~2 z+DQa&$i$x6a8yo<|3(SkK{p%e(3%h=_=N?(x|1Wk-*@PcV}~&CJnQ_acO|s@%W5JC z_SJ%R!_-fWynTPUjnTx#L>JI${4&^~qt zHf`ifL; z_+9a~rIxPZK79BZz;CsPWHnE*y{OF&)+qEq$My^JvOl3&aiDIJIF@I~6@=%>37{^h zf0^u{QHgK#peawC%dn^F?g}#%_t>Rg?TcVH+iME=%7qo#kukuTwE`#hWE5|{aJXc5 z)9fDZmSMfil{B^#bTL@&FmSrYenxGntX(ww43Dd(A&K?#n&H?c|HEKnKOfZP^H`gp zr0}b!#J>We^LPgHLS9nePlZRH2IbDE+G56fByWB4q}ZsoA=HQf?>GmcR^46;GSvP^ z=4UPB@Hls%)f4)Y^;QH%kl@c^>%sU!PRf~%T}76{!o{}B0wcq3Bvz?P>eCYwG|WylRFo;88hm<^5}6^cp~V8sBUq zIcTL{Vvmz^Ug~ob$&p*=B1IWMn-(=~NeH*~S=V%+D7^-POePbD)Wyk$b{IM@bj!`c+2mHCGE6La=|7ves0IzU3}qNOLGEk!cPq zrh4(m$maKv)G#WKrt_3pqp@s<=yqPe^pWoqBGL=*Rq!0D&ems2;V;mYg%#T;zUN*v z&+hyYeLv2=A$!HQigLn!rmien{|SA=p)j6GG=n4@Pvkh4GQX93g$3L>ZL_(iX4eWn zP4~s;u{`xe>i<6dX44{jIDBDOE5ReOx=q2$s$jOq&Rwurl$MWGg>~82QIYic?kf(A zxnMk6wRdL2^conPKyl}Y^oX!_d7dK)ESccOlQ|CbP|Fsm8Sr<+YguZ;rTP zJ@hBR&eQvAj@Fs@?qzZGEDNluQ-02?l#yOEDcX!3Ef(;h$YIJ7(}Qmk%A}>Vik*NJ zDd4`XJ0j;HO2VnUv|!}XkNP3M^UTVobr&TN-Zv1|w^#6nW15~sLFX{R=sX5rsITKj zSh^)wed`-j=ubQQSSk5BdY9AL^!T^zk>2X!lGAC^uxgWs;<*k)GusEI^fD3kr~@iv zv+wF#i|P$aa$ew>^PED2R_>370bHTZ1$|6)!u^naU|)VHH;|3fDTsNxzWC`x-N?JR zBfCF@E|2^wuK^V-{TkVni3I<LphgPdC$~Ku!j)y9Ptx;WXbYXRw`7@eJxl=)TcUM4an3tA5bX1j zzATO`7xX<6@`N_s$j49_%;)70nbN_ao-&hGTj;J5e^5va-WIp#$`^6kbX81HraSTA zp>Ov!aI@%52OBYxQR14>)^*c@^tMSFn$1t72EY1Mj$h$pKRlsZ>$~iYJde47z?U_4W(96C1j09!+u(2McriR6VV=VOkTXbdqGRAl{Ynj~WmDwS7{*A{PaQ;f-*Fbg zV$%qpZZUG#1MrKF{a13>VOP$xT$9nCVoB5r`%vZ}@1qhO030G?6py1t`xNHwFs0>} z&Z#?db-7U`yQ^_OP;DIMpDO`nkdMW2qpXR|NvRb6og4*xx0u$4yH{F1g^eX1|=c8JyD} UOzQnF@7Mq8-u>UU6J5{!8`>*_P5=M^ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-26-Server Rack Hardware.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-26-Server Rack Hardware.jpg new file mode 100644 index 0000000000000000000000000000000000000000..3fd03fa7fa0af662f77ae0e6a580214c15b275ae GIT binary patch literal 16078 zcmeIZRa9Kj^DfwU2o@|jjWlk-g3~~P1qkjMAZT#ch7f|gg~r_>1a}DT!QI{68fb3+ z?wvJv&Hp~!nR%PFX7@V%();Y*bxzf;s_(0MnSWUay!j*}F9SeA0sxQ@AHd5J-~-?l zIyweA+A9nU3{1>dud#`6u(7bP$p{GXh^WYc)Kuh@lr$htCK@_+dP+)W0T%Xm+~UOt}xEP{lIiHVJcO^Sm<%0o*@%k%$uymSKaUjastEK!i?0Lb`ADELS( zJpdqLpJ+(`eE|NCfrN~LiiVEy3iCA;05TE^3Nk7R8X77pVzwXRcK|9r8UZc0 zBs!t$7YsTlBA&pwoLBTx)m_AD6XzgaQ|IrPuiui8l94koGBLBT^6|eH5EK&r@KIVu zR!;ttx`w8fwvMiznfX@>ODk&|7gslT4^J=epdZ0MLqfyC;}a5-l2d-ALUZ%-3kr*h zOG^LL)YjEEG&VK=?e6LA>mL{#nw*-RnVp+oSlrm$+TPjS+dnwGxV*Z)xxKr8fc-}< zBmm0)5DW4BKLqMl%rUbS;#Q|F1-Z$W$;3>W_q?Y~R*e^0RQ|F0zbKLz`raxDU|QIHTX4+S3p2Hg2P z9{KTpZt7@F7_O@e+fBVqs3RiiM@b04`?T=_*l>RmeF5ZzyZ}Iz;QJLC}X$k!P0=PNY`Vi@s8uL%p0%N-4|EkdcWxK|loDqP>aoQ)@ zd-XZ_I;sx>ThhOPD*r{n!T-hHe7m31Y!|`P(+&mt0usa5HtZO)dQ81s==U)?LGqJs zS+&N>EyEGVHH;^vI+p{0b;7~iR+RPA^jKe=33FZfrM;(o$P^}XKE?$6zgRBY$qRrA zVX~qZT+iPtJFeBWdx!w4(uhl~p!V_wQ1(;y1>h!M#=YNMUUcVex8+CUaGw2|PixXs zwuc8hitZRsTd8BLynh``ApZh*_uO%&7zCTDK}pYT-fyQky7b$ri+O$sxgOO~{;5|Y z7=Jp7QEUO8jMgF%gtGl3*>GLA&r4`1V3f_(^jll(9eWr2~ zb`<w{tZ+Que@`t(p87Uh#pi7*0sBuOCDFH{LEhfNNQEz&tldZ8;lFLp4qDH zqyUgvu8mQt#YB|*uqPZg8lj|feZ-5njLG+c*U_ARqdzkeg3$nH>$7y~jJNtX!F^@z z)%cG^;Lnhxx1lyF#+bTXt5EkR33%egMum`fdDM+}^>eZDDLr9EBOoD5(X_0-`s#SG zr6wvr0755Cj-K}d`0YA#VE0fBo=vpYdL>vNb73$eF#F;CKMkNMIESydBVHcc?6Kev zFP!??5Gj}a#O$&D0`DGBz(O9=obAW0OJJ-R6_XZ=>mCa>(BN77L|Wo>{~I_x;aFX( z%e_Rj#=||4=${nI;G)@IBSEOt?LQ!r((@!snl7EM!9aMM>~jt; z%1;#Pq>lrZjw%I-l`qiCg%w#Ne z$hn>@Btj40f(dMO2s!wXsOC!){pG^;0|#L+&}xuo>-l4%3dRgI?RT~;9Gl;t(>9aG z%-xtM5_3@fHP?qXfOs;_P^HFO6a5G1ode;rU&3GS(LuvirN~P&!Lc{chk_?B+Q(!G zKM&wO>*ky)Z&{4=BrSk3!KCq&V-7r2-j7ymAdSibKSX*e3xrS^X1*mQImEm9qPf&3 z{3C%~GEY{h^HoU0G)&WlR_D5laq6-fV}43cF}-^5j(_L!zDj)%{X_=$T_eV|9^xTm z;udlDFIEQJAWMM@DEHaD>*3v^J{k^=#XJkDhn`goS(^xH zano>-voKP;wM%q=g?NEghxac2O_Jg!5#>qMKgaWkjSp=wE?g~7^|`5gvbO7nMxE`7 zK%Zo|{|;zeU%ysTYamVW{4iD8&+;svml_=1Z2b23z#Ft%zwU?4vW9Rb>seY#tCN3j zCs}8N24#hmO%e9WTHpDfb;C;HKa!jy1rDwg0ss1Nfh#yIL?|XblaJZ5gc+8mzk2us zkd?IIYwby7No)J13*h_x@|KA2Bl|TtOXJKVp`Er+?= z`Zbo_g#{JaxvS{~5WoU=O;C7$mf%GS$EWx2U@nbUlv`=7_uAE^xF*drPq^eT9NhPf zUxN{-N1v&w zvV^=gfO1z)2Md8($YGXDSbf82SiSie$t5zGW}3Gvm)S&D zIUjr8HfDrme}k~Hv;8WRUC9{f7K46x`psgNh?Q*m>UrT1ub~8?6ASRaf2dT`o_873 z`O5f&-)`h;$vtO>wp3z)4hZejgft$HmJ2XHi*?+1gLT>Te zP~Pe>ygcgF>lo})`Af=ppME(1w4Lg|*GhSOI=EIkMsYn95MtnDI^WV}XDqsyKK}&y zCw2^R*N8$OU)Xn-XeBXM9{=lm7}E>DsQz(Ef%&l0pYJ$eh@A36`6O^aJw>J-=-J}W zUo@9l8zPpNRq$3q)sS^Y8CL!*UxJqO=lVPNaa%K^Qg+#)FCd<}7yz5de#4GW`(UPe zP1HgnhOwiiwrGuuj|IGyu}*paX93Np;vbve8n^o?%(8+lXZ~26Am+Yl5&T^L0+`^E z=G`Frt#I){{<08Pdu#xB)Axzxop{hi$5W^{{TO(g#AObAHQ2_s#s6v|mKaYgUgjCc zFLmp?Y=EeojLUP1%M`qc-&c)st&8Saca(-XNR1IhiKx>EO-0Otp6AR>T&YnE@Fy>KhS;GY+%zjOR zhAaLBu7HU!uCgJaZ8{UkzEojMrye5S*7tB@c>(w$Jw<7=5s#5xjEzB#7Gw|m(ysza zF-YkJnVX)_yovlV|7|_aj=ccRAE(z|v1sS&oP zV8rnr`=(U6cy;e;)VtK$T@|>pppWbQEAa*J>3209UNUnctQ6@fSX`xP2g|`p09Hmu znz=bm2usi+k(>RhLa5Zd9vxZ+LozBz_y@J_qGH~`%7icmRmgKHm6cmQPJMf30dKsS z`I-RIE27d<+k2RU1Rg?!zi3_;4F$V|8zWMiVTED07l1qE3xHdT?%ZM!KTRd}M5r?q z>FKEKoA+yLd#}xC4UzQw^Mc%DE;skb=74$NY@8@dfsDMBd1BTXB9YN+co0M4>jP0z zTu)T-Q`MZ0S${a>~Zj4b+(puf5$xq_>)b-{K@8CDn6do)dbwxA1ExA8h zSX`mkNoG&nt?|G?wNz1Rs#?>UqnZ}nTncphD+^-r3B06VI{}4nZ~4`=W==K-r|6UA z=8}`wk@_D?KZAwK@@(!%8l2e=gYiA=R3k50uF#?+ND(e4yYD&p+q`G;;^{6a`;$OX zacwLejeVe zjSmT0PYQ?}ln@-UJA2b5>hyt*u2+yP*ikLIE(&r7c zMrCwc6AJ|)dg_{_h!=nqT`Jel$k$4k{-S~(nzzC)kJNlif8#6j&<lbH1; z@FizI`CH`sTCF0-XD8J&;+#F1JD(i6bt-Db zA0m%yCXl%{u;h?%?NC6wfrpFId*zJpnbq5emi&EF#`g!g`E z`bxQ~%ggu5n`nJqc#E4VdG3!peUNn;(QPui&gB93d4<10+X{2=CV6Sz zbuXsPpIJ94FMyAO(z;BO{(@SN5I9H!wR-Y20Hs76k-5Y|%DZSRXF}ncw`zQb^?!v^dh;~A`77+%2B?8?}l=3(o&5fDCllMn$nyfn8lpwO_nFxS?Ff?DBltDo{`$)}c z)|x)%x>vSd+L>3&cIlN2q?TilDrmP1VsAZk?-Lnwi^SjMnQC8KRu^XKW8xBQN}API zbZ%o{18pm4jDHW+&yT_$Y(fGdC7p_uJSvpsIA=t*(@gcls)X80!LaG+nbq~p#lO4i z@S(j2i#7r-G=qa7|GCXJ&K|QY##vIQ`vk@0{_b~doH-ZdyXw(e6uq5KZ?yau4e(gIXv4O`!P{+;BctgqYsjsT zJ`b4YX<^p4&7qV(*YFh7T{>xb#=(fddsELZf71Y~+dm zCSsiI@Xq`qP&5*JF@xOb=}%@WCAhkKI!i3&LC~KwUQAtQ;>+#=aG#l#DF19G8B^%& zXDia#BDW1^I3lOcpHs9I5 zG1R2yf)@0aeritzqC8fyR`fSZr~ta9-{#-p`2u*~RmGaI^5t9H3&8KSYfSpCCUrk} zLpf1ae)9$J=3eFA-jAspsx*x*ubQ8^)3;gn&xCO8z`*0Dxkg7)DoQ{qt2cgJuUf0= z$85@{zeVQ{y(0(+W*iny@cX^DR@w#6l|)^#7>U9Zd(wQUjXgF zg4QWl%-lN5jRQNTW@~fAe!gs#ikODNDq=dNF96<82e|z#j=euT$I)Vrv}-f@23@jl zvNn|F`1GGimpHD`x%hZ#2WWr#m8o~ZL~NCAi4#8>rZy0dOW=LttZ=L^pz0pmcdEU+ zBPJd;=OTRwxvKgUu~7$204=7y(_pASF+iHW>*a5y-MHWR8fOoZtFMk74Cs6TxCV{N z)`b~FUJ6wwA7%7EK$n|MffS(+ezrWY6`E;IGbu#{wo=0BfF2a1s*S?>hJW-KC$sW; z76wW5F97N;V6aBPP$j6+tLJSBZ%?LX>mO{qCNl6)>Dq$f!mGv2RjUszcAUW;j=Yi2 zwhMJWM_2>Zn}N}WGWpQlRL1@ zN_E_xN5GJbzur$#~i0Hq`q0BRYXj=GyHRXh+=;1>;;e=e(1{*!@7+t^jX{ia;C}AW1m!PVI2F6 zcUkt)NktL2JG`1`W9D0}@*BN1o43Z_8g%CQgixGl0;j1%FwQGx@B6>l5vBpLq|!cP zq`GANW3IIEVdkeY?OWZs{lMJ){@CohECRULiTU0NX0Jg0Rp=~`r2qZNk+bhi2kz+|?;wJ~XE%lz%*%vm+^dX%lj7omGioBumxhSV}}DXlGLBqa4g^CwdYdO&~L{CEmhhE zT;L{ACwc71vLvAX95Pc8SO z+~Zvr=%%cQ@s3WO-m>%Gg2BpCcQH>zncV6Z->uV&sm#Wzum4(kHx!N>g!@)!t+woP zTC4V#@2ZH-HL80(uoUd=Rpb_aP?9`&Yd7D^IL;Cr|4vYrc>3rF{^8LHE?^za&9;)= zxHoa8ZKw&SS?eYPML)koWCPzXe%3hI&$-t_Kc=L)0XqRaM1G|%JrD-TwkHuA*opaZ9LCBl(Jq%*bZ#e zG4;#$0|ex_%*U`Tp(@cXo_d!}tAn zg*_=pGKujbx)vb0+~t9MivujF6aP=p6`oOU;~5@R!>O`~u%EOo=KIP4n~|GPadK-J zkbrQ0wp^3pSV^oO60!4MC!BJM#v+N_Ql~rW@*OsB;JD)B>o%gbi;b{+dXvAc!8JGR z@4&A;LeuV?k=-rkojdV|y@5#`87@~tkI)5bmn}m9nfi(GdAr2n>C#}y0cD5Bgov%-6gCB? zY%=`(9;f!{>2&Eyyb*Dvd#f1YdXGMVb{Mh-D_|P zx{tPH)eSc^co&E6?#c@0orY@$=4{W6wx}iE@glrzf7!IF;qbU!G8ed1Xq^_XtG<3uTVbqw;#i^w)Cgx$3&p8rClC{_wQ*KKJ$({ zgX4fvPVK@X$y2C6sFc42_-|(2tj(!nh$V&%r>uCZxidJ)DN~5Oa0*Z|5)v!>^@g#g z=z^}Gy82uI{^@h-3&5`9T4CbzJw?I=y%K19(^UDn+J-HE04Cd)=Mg9>CFVLDHwC^N zZ1Jdyx+N0q0?KX0JX{%@-Os^4g7KZQ?vo8JA`$L}u`25OMj$@%3!t94qm0}NlU$n8 z>(~#vjv!QtC-0R4ALUS8)w{V*eFW&5TpYcor20`PDk8?SSl)%KQe(<#ry=S3F`4qZ1B6TGg}dxUoG z=ab-O=NABuLi=yZ!T5DgdKPvhC%g`npGX2@zG*)A?y8txMtCg;Fbyl=jMh?Z#d4Dl?*MZf;HS>#h0Q z2tvo0qdJU^@BQjaN&Sj{hAPr%uX!KF(3etDH{G1DY_FEjYgS@xfIwlI$?9IMFSNyd zpae*Qp4s9iqw}Gyb{-|=>mfbNofmOS^rV!QtIZ*eUjZQ8heMCR@a^Q?fx)?*DbACP z_y-NvYHqT2$*er>i2F(f7EW%{VtTE;;eC1H!}fR3k%guAIp0nmyQ@uouWD}&miV_| zBgvO<-_P#8`qQWd%KTYZ7@AtLo;E#o!{$>WX|<8&gGL5IIx{-TIh?S$P+)(1DMcKx z`;IpWqnPw;iho_4&}$M#Xj6KJ?6gzn$}tb@YwM8oSyq0|uP#fWcwZuFM?HU(d3}V7 zemwEClL?bnarrHb^i%_z%eb$r>-Xu^rHvlop6@4PiT2uCZRvZPGaM>me^M=3iP5u& zR?#~%bDhGIanb>ZU4Yx1-ajtMWQe^j>)?qg6=4 zPSEo%TK^UCm}j&6)aMAlNbS+x^L8WTr=m=k>nB^a4UVb*hJwT7Iu}3t!2Iimk_$&Js)$n;7s!1+)-l@COLw9WjbjpUk^q3K*Wb2^>XbH* zmdchqR}TrXFrBO81`t@~k*33I{M!K1Vv2G0&*End;3PNM4Q`*z(Z*gc``4s#xnP#i zT(9A5c+-3Qsz~|m7*atbUOdza6FX96-d?wG{G=Q zOBcj03lpT6B0HLL-1d7_X zdUJ~HTd=n$wCzFTpER`pG2(Wd3FRUDpy)hR!DVr)QTr{DacmyzoVdxe=aVckW8=oh zfMCU?nARqY%+9Bxqps3077o*&YVK7l*1nzMtpG6gG<1X29DvHW;$4K1_2DwBN^kej z`rn4H8Mhs`=KLE}UwdQ!zfBIf09f=pefB&gmrOsAIi)T17eJVm;i1IkWy|zdmB~8< z_I03I{Ph>wY7>oRBurZg{9{Z%a~K_5XNJtbp^;8k0(;7T0c7cwl+}oJ+ff7=1Sqrn zP$D;i1IvSl&k*n&d=0O!&LjNdWfs!#eWG|M%L3y>^N_DH9bvE`UbTPO@Vf7wC4 zZ7dtufOM7Ab#%k;vhe`6N99GS9@F+m0AkRBEluu2Zq{H7+iR7M-UBUB+{(G)h>l&< zHz8F0Im8n$0M3#2iG8h#md|Y4}|>CT^id* z1LP${e`gD4EVgNUyPzgF0DqwJ56~9iu86^j{n@r2`4AG**Y=_NU)ySeEqVO}MY3I| z%N6x6wS_8<$e&FZC?R7I@|!I5c-QPw%)MBw2A6pgys>81mXAXeP;IDHv-6liaUV|+ z3Q!Hl>n>V0#mkzU^b&u3hzK{^eXGv<3oY3nV!TGp7r?(T{#|M@J6J)7&;vy~Bx0Y9 zRe6=uhYGLr9I|0w~LHa@id{%>-KN@iwC)i!MC+L9Gr zQq0j0&wpl_jFIKBag#d#`#fB8m}4|9wZf*v)`Fo`Cd%VYQ#a)nxci!aJCkvmewu-U zlvXnYWF|y>D9)I?D<sxiyXt<*p90n@Y|*9i z>vMnhXRoZsU2Y>Mrb@u@_QWGpoG`Vnd<(1Ae)@!4cZWIBDzOW1-3+OX z!kGf~^aw}Heqc(+mfZ8YRA9O?4mXn;Ptt!%tbDymp+;`^KRVzHlsRg&jM z*>+!&X)W&1wL`yZ)4IeZbXDw}lU0msvn!g24aP+Olqf2C6f8G>qra?dPB3(wcQ(vz zx@;1SdDm-XMKbh|e~DoXmBTiRhJ}<&XW|!3C^0t5Z_X~kUhWObyPO9^+tfI+8}*v$ByQ4p z!#t-bF`sYdfIoUqebCAYmY2I3N>Rvj@IX$r5|*UHLJuJiH)Y;Qi8Uj~46hZXc2#amrP^IAJ=fIwQ)(qneq%2PefQxItwbk27k?wwy?5H(A|AW0Mo1=ER z!REIQ4Za6201g3&&%8Em_ST8*+Mdza2pQqESpkkE|j!5u{* zxZO{%j2b7#@7nhFJ2F95ijEw*QqPqJn~U_I$C?3}7XaDyX1S+)tg&=!_8<)j)|^;_ z6ZFWqoBL6=GWB~O@mJMxN!FkTx&+v)MMLAFJ*zT(e9X~dKTLD5a#F;%cAe(vahJyb zRknX!T-hj5@-eaFz&xmjB&!4j-PTXz?W^dRjVXq^9Z?gY9$+=0FpK*Bh<=_Au3~H$ zh;xaH7k;*#a+M#=OtYc`JmfK~GCU|}GgY{0&i7Gd@C=lW5eM9AslsT$W39veN*P7~ zd8T2ebObdCyIO1!$&gw%_%&(4vKx#fZz3ZEOjE)hSq~D(HO^gV{Yu`UI10nu6c}p#9NSuH=>9# zDU<(D4Ko);JTAY15caHH+Y@DqjCQ083hE$*CNZZM0DD@Subz2Wv~Oeo6IwdFu;i7% z37n%tMiZ`Cu`$S9E`4Rm+I>d2wEE}AqwH`jctJz?qHK|)`+bc=DrsaUc5ufJIMM8; zNX4^Pp@^dm$Mcn>-bdo9a5e~xD?4DgA7{`t;F1)LHq`?X#3MAzMt5RTrQR~@l@)5gL# zFx%64-tRh5h@8lgQ8dSt!~aj7VL~}xVszh}6?`#afE(jflvYL*J_B7p(ej9F92;{) zae2_rE?wcF{A7keHCgx0Y3BH@EvmWQ%7)7*P`JYi?H;9MuoC*!D07+V&y7{l-v_D|X+zp3kCA9Z?{*gyuj7@w?6EkUGTMGwLeP{ZQB%VM!?) zK;fMHWp1Xut+ez9&fYKl9@MPc?V|@21B`I$rN>X}I2^la)(n5-9)l2g@?winmBaXu zFj5y90RC}~VN=X@0IJ{-E&mm>F^ST@V?$R`B#C~c4tW1~5EJKly{_%H@ydShr>cXn z>>I1d(utI1p@RTOfg>z0)d4D%qreQpUDp1xAEWai8UkB1OTFm zLFD??LY0hdTW-~zz90Q2aF4v@2ylCc#@f&DT`obqn{Kmm*7c`wyG}Cp^4n zamyDE+EjT}Fh?1K?-KNLafpMIknGgQhhhy=4x-1^2L|b|R%$2$vjuZ0PRLguU zc%`AqAUhWadsJtyd0*^IZdm*-pD!Cxd{|dI>w5fF>Xl}1H?2Z$<1I~>PEwMPR;EmMhyg#y#W2T+Xp%(yzbi85s8LT(d!dyQm%W@aCXCfwE zblq8jw)b(R7Qe~gBtN=ibqxGzq&HRW{=d;|9uFPzo< zGnBVV+32E{*CmA+{Y*J)Pp`Y~gNx!xL>T%c2A-F$5j1Fc$hzNP#HLmPIwm1zpwL06 zxp<=1K+pjmf*BtKUobs?KmAYPY&XJ@&n;663Eul{sw0T!va0`hgEy*hXX{XJO6C^; zfEAv=$gl6xa({6bR({dnu#Tqt+t+w0W9@RnCt($e(tIU6MRthC12rm-$GRl%gA`p8 zd(ZmG$4_~8qXJpz#~0#~PFfnrZH8MMu609-fs?h@&uTCQeb@JOOe;>oFTjFMtwat%r(&+IRx%QU?i>q^s-Tek#U6&Oq zh6{fITJ06eK!g!Y5!Rq-&8KAimZym1q``5HF@Uju-Oi%ZlDL1?SeuzrTYXb%UQH9T zR8ajp75uSx0O}h}cOu-3-it8j(2|AZclrKal}tSyizjZZ-F))6T~}`hnL3ouE5nt9 zcO@!P%tfCD;~@mfmzBmnEXCbyy~sUuuW%1QWaIuk;G{38l~%W+XKGpse=D84B$&=C zTmHJgG0KluELTJ#b9|KjFc*XXw5D~TN1RJbR>PJ;6(?xre=W{NL1@g?^~>qht>gy~ z%=qP-{I=#maRoUU^!LA)l{ZJ&*mP#T)>rcyo@y)uq_NT;?Op)n3pKVO#@9ZAa{U)5 zH1vmZ#KEqQgBYe{Ts8{yvs!boxT$NmR#Jw)VHU z>3Ws%-%8JTwjQawcn9`%mOU!G3e|tOT$>^a{x_Z&sBr%O9*KL$dVyMMmyM((9w%Ee zSAL;Al`Q@}X>nAyID$w=h{hCyKa6|-Q2ev1$LV2$7dtX3uSF&ImB@#jrQrV`I>wtiw%TfQmYjymIz68P09YHtm>76(&LLv!$r4a~hH{tDxh zxI8ZhUGrm|UC|_lMOsJoMx>if4@k{$sB2G6?pi*6lOscpqbL@_5ThReN4eym+Pb7) z*q;6rZK2kVJza@$W+Uw0j#6XiWw}yEb)%f=cyo9=H;{{b`oX!U{vAfs;E^uWuv>3@ zFB>h-q?15axrl!Xb^4q6YEjuDx1COi+6#b78|@J$AE}86qEn9j6UduT_Zwp!2^hDx z*m%&YUOdoQ+z^?`LyHZ0KUn^lp$6Lq3SW9q|NIa$pk@n8QaLr`-qu_$AGRlpOdn+Be z_k3Jkp0-A?tc)GoPtX&OK0c;%H4U>oFaI*rUfX{(cdzss|HH1iW4xu*XC-~8H6@xC zRR9e@CtXtxGRN_GJ#*zS3?0l;dg*u8sreJby7bgp3E<7tW&FU}j5i5+#%Cb%Mc~`< zW!cY&I?4$GRSBS?=Mh}~EDUl??Dhi4g|>rUk9VSV@ESXK{tH}Jev{m{c08QE_n%@# zM~4b^xSr*LWMROlg8WjxqImY}I)^X9qZIZ3hy!=p4rOL?j9Vxg=E5T$zYGJanaNIF zlqH0@c%jzbArQXF8z~CMrJPtF&0!JS;fPn%@>4^LEkqx6)~VCErrInJehHLyif9rt z{F9|BY|e&XA5HTL`!qWaQCMDJrc#^{Dd0ZAjK=Ww6o4Ny0IlG^X&pEsVm;K@!)x3^ z^lFiD+?KerzuFs$jCOhUXh_gqkgEl7p+V_Q--TrsgN9GDrHl(DFOP2|C};kOAE_cQtVSJdd&6qeZy( zK`zlvha>*4T?(CeY7$Yp%kvq$&8x5QHye<3SANNXeQWK|wqv?3gE%{$#j=5QZN(xM z(NL8KtZ;<(?qBwxkBmxk+YXn(smSvmO${95_U6-( zos872oLFGmig4EHG10>Bc(5I0l!sX5BW=MZLK3s7+g^~UvFHMr&l&--7tu9w5m?VU z%UvpYnkr?Zdc76T?0;g5ZJN2E_$}RgGm?F4Ee}kxQ1s;IF0d~UTDHj9!K&9W5N7WD z@^9|L#){F}r)U}IN+D-BiM48^PZEMuN))q|A$&}PY1J#~L>#%)Wc4>hL)^Xvt?K3@ zBzgz_G}v^ZRmDteul2^~13}RtIz+E;lzW^gk**mpHR;U0s5!;;YLLyHv$rEd|6c}u zdszP0j|40hW&voH3U{UY>X-7GrpnTd7w;2)zPFuqdGlBbN<#pAK33(3);2PId6sxA zAxxGUA)P?#bxUwqC3nn#e`am|tWCPA!AMJ~ZHjO>>4Fq9eI9|+L4wA&U*(aKG=W{j zC@%nF1?#)4=f&EJYS7fk<@ZUsF{hbd33B}v+)W?)tatP$e-Y=40xOhk??A=o1M9Hq z*&x=f*#M#&;|iQRC8cW5t%CPb$d?k?uN|iSSf_29QE!P=$8IGfeyAx#Zs zL@GUp$y^4+xFNN?ugdKuZy4RXM3nU;6Oc2b?2rGKULxXKMC~M^1u)Yy1>U7?3}Y$V zd<|gE4F?>cY@?h=+5`fbJ0w5pl}ns-`;=6OiOBmz(Xni2IuV}z{i&%UWK;KIKri}xY5!Lb^ zVsNRERiBMN!FJOB!}X|Ud5v7X^#ai8(4(lpyqi1Hiq|Jt8P%9}Fss*hZONdeqSt+8 zDhdop6ks35OABRosDZZ7}u_1 zV{28w+J)ah9Yj1`kW~92R0Yo;lEI_DB&dz<*(bs>ewmHUH^ue!Fdxq)X0n{|d;ha} zwx3WL&(Uny-CJ4lZsd~TO)_OzF}~mYZUAS0?h>bfy+ehLzKE6Z$3Q>*tUZHkXx(v} z1*;g5X{qQRIp@<6zcDM6l6cq{evq)!f)`vHfl#-YlJvcs(}eJfiam{e)-|mAGek0WkPTA%NkSXQudYaZJdd(Z7xp#M_hK zMzn`!U93GNj1#;7=2;P@Db4RIY3y)+2AALapPq&$8#$br$_hn~@p%%`%(7ETWHxXu zNVIA51KW8mu++^Y3tC35Xu2EMtSYy{f<)++Z`>+gTU6+GfT1e!HZY$vFJ4;aS^G%o zuu6h}H(ItpV#UTZud29MKS%;EB*DmpUKS%e^wZ^dkX|HH9;y*em0Ua}&K5o8K)NE?<*J0FCZ=9|4-LnUUq0 zlGRf|9xni|8r2DD0FmnddU<1XaXqjtl03IU4nz}G&lTXGZi8O{4nUYP0};~8!hZwk Cr^d{0pKnO)eX`x6J2|e^8y-AfOReDEy zFQJE&$#=hdznL{_e%zULf6cll*(YnAb8^;x_PgKmJbUA&am&Chb!9bWfPer12=E^O zHwQcg$Vo}bNJ+@a$jB%t$SJAlsHtw;pkls#hlY-w1;oM5!p3%&TZrc_rvMil+e0Z{ zfkz_Z;^H748L+gdypWi<=s#XUKtVx4b%TnDnwm-U9@{<9|LX(S2GEiNeFSzy1e^dN zEddcN0j>)G;rmHK@NWn3?*{=P5itoV894>z4g3Q&w*W!{A|gU!A`%i}V*JyA`1=4c zEy?YBB8sGUbgjrZJ?KP3;aEe>x~|9i&=YyQcyB5GBLBA_709t&R*U=Z++kS`G`Jcf4H(a!MT!h5LM8stOz(qjli*H1<#3c7bNN+3Zl397&;S>!a zr+X5YRoPC#C8l>oZ|yls$-piC_x|xep#2-N{~WN;|0iVs0`@;~%>q0lb6`_2zBsn884 z98jl&DOCczLDjzKgh2?lqY?*f6}oItqM+w7I3V5+fdi(EAQ*bV_wA2K=yd-N&G;$e zo!s-LsFxP5WBVRW1L;;N{zX0z{gA`QY-g?}@2pV=5*hBbZ@X+Y6cnv?ipMej0THum z)#fzOl1ux>Pq7w=&Rc;7-j)`k(Ako?LA{-GWxsWM0Z+fo8zaARx1R+j5%@^sfTRmb zR;0E8bhT$>N`tHm>?(d@tyOGB8yXEj%1xK$H_q}HZ{PedLK;)-Kz}XXNA494Pg1oWthlFAAW_GV#Yvmt+<7BlE8>;yNyX`K!KpaGfUdEd__p8| zyWh&P>7K2Y$3+H!t`gmUwh`b1XO%E|5OR(;YE41Tf0^mu9)VjwX~Gzt=*6}S?9Ir`jnH(oMD_$xRZMu7(p%eAPuv}|QcA(} z0gol?(b869OLdWpFU*~$;$JMbjl5L3h--a9^MuA)T0C}!oGvOqFUlTq>wvDyA%mK4 zqEOJRqG*Fr|4%R#rMGS2N-USZK_TsFCN_h|2$cPKLMS3LH>z_970$Dj6@P6wUVJk% z0#JXL9ZdP3XkZ-hzYc29Ke>eZHiMfv@4<|3g;Tfyum_#vStfoU3rU_r{)Si8E6`(WhgBTylIGT zAkaTOC=ZkR3G(;hzoHbR`uT9@HO+qn&(yb5WAJL};a~yC9xn(Rk@VAX?lG-vj;mU2 zbTt9P!(adQ%YnDBE2|@+ylzCoKP+vXK>vVOa)d~cuPs!OEmY1}EpKeWV30xH@{=n zF{_P0JSfaJ6KszH)hM#kB|>K@a|t0-lh%&@&&r;C+h5`TnrLs65Z7~e`3zmmuRdAk z-8AaqAg3bWkj}EdoYwP8b?gOssgItRU@?2;%j@zB(_oo|zk@QhSH4J3g7uMAuB>Hf zoT1%F)~6@%+2FX{H_#Lt(_{uMR#igTYKPm;R;ig!xlj{&U%T|MzmkhAO)7AJ(|4UU zqjE=-;kfLt++o3@PP&49z5ynH9IS8*`6y=hpdUy(|J8i89_9TQcaVObW}%x9!J1U?RERLsm9F`$LseMFm6)yvhnpXQ`^u+719LEV zh66x2z})&bRcx=G5?^PHf7P}E9UP-v-5Utvy2`}*2(|`S3+i0BtrRmES9~o~89B&a z_G9?tw8H7qIu*p|A}dio=ZX(;C88c4Ru5x?xXRbl){NLV64MdZX1t{h;3B))^WsAC z!ViW~#kbdpJghsltMf}e>tN0~<=JGqu?PU>%IefLrj|*J zzwkusffgV%t5yAmGx`*8&7IR;los9F8R*{IM*cDHc%fM1MBhK=R>OSSekXs^^I$R9X$g>*?aB30aqR0Yq7S*?B& zm0yeAT!}x?PGSust|TuLslO}pg6typf`0ROz~Ju}Vm1eaLip0v)Qg*=XIIaIk!Nc4TpSz9h~^j+NqL)(`2wrp!v z38ls{!uDpb>F2_%_2{Ddjqee_Smk8do~^Kx7q`H?b>6D?V;Xxc4V6<)rgLksST5&*8GwL*t+Nhvkw=!Wpz#A=s1r4QGpJUiGGh3>>g8RpZI`VSd1|@s$Huq#N31AP8E`AcaIu2N-?p zlB~L|zDd&5aLO@3)G?Q;O+;{#(F-PVWeL+9^j^-$k(S1CZGnp@7+LE)8Ivx+A04Y5 z*7;h+-d74|o)ry+cy^|&Y(blsHm{71Q&qEftG_Q8Z$Q1Kp>Y4;DEeWm4bV0QBHP6M zGFL0TZkT6`8EZLI*sm6$rt<>i*i*-I1?z9*JTT?dg$YkNL3`jwbPqQ~b`YqDc%?bP zw!_=5!y_MaqrC((u}3&yCffY8o;vCcvIPm>JVB_07U*ajTtm~bUng(?%*P5XjK){H zrU?Gz?3j7K}M=L?j-I*aB2aG8_aH7a0jLspph7v6%e! z$FF;ANRoRT-=LtGQoVx*x0Pe!b0a*-ji@pF5v^f+vw7?F+w@&=u=dZ`#Hq`mmESMp zN-g$Z>#gt-SBkDvAm^xlhD-E1yU9!mPwt4YcWMvq+*F+RiVXVfOZee-dHTT!%qyt?H z4Jo=@g>@9GHy^9sL@iYBTg>ee~f56y(oYBha3B+LQM6W6?HOX2^@ZPQO*N;BEGtBla;8NSH%0x3#nR(?pY6o zC}Ivz&I&mC+k`&$90)y$$K$o(u960L2*2%IDl(5<#o<9#ml*_AVV&UjX(Tk z-$Jj;425ki)oh3{(ooZ8;z2!IeeE)4?SJiYMOkvr+vH~EVwJ>$HHF{+%PYz~Go9Eb zVY5_~p`Kh9EVOf%Xgf|8FlKPiP~7Fm6C%t0+89CMUDQ?Js-Et zv~>=|9yY+)Xs{JqV6}^~#&^?uIH3H})-<*Khp5wr5uU~T`5V{~1J#3Y>5=z48;#P! z1n#-Lk&+BbpE4ARWFI_rUEd!oF$O@Z;N&H+4i0!}r2E6WYFxsvom%oV46Oam9&2-Q zTGqQ_BfMKYcQjwapW+A^xvvn9Ksx`R%soHA0Y0m&KZ6q?nT0}o89k~s(c@prEC=~x zo~%+TG$&j}_)O%68P$#@b2w*rL6>-VkOC@y&Ahrz%OMRnH|w!Ol%sIEXFXqCUa7Pm zDq;=vutT0DQ$fN$*ZF3iVq2QyMn{*1{VGbwrw(nkkn=nLtgK~LenwwjiY=Q7iQ#}d z_>%1(L^fa4V`&V%Q9!zqz!so?JluMP`oke39|#?CB>0H~Hqlt|)B}jR2MP)=%)gVc z0~LErt__Ajj&iwSx+uN$>(yP@pVW{!Q1Bpd4+0yp&w0ff4>`=yyQT9SdMCmPfO-d2 zza&?8_!;=sz+RHQ0=zRs!S=>#kCzf!b(LmJn7TU&>nN#y|4KAukB>od)K}y@ds*yE zJkUw~#~Bl9Oq)zU?G<~_*AFKlihU^mKex#UgBd-@Vyg3oB*Xev#fO`wO7AY;xugUt z65<|=4!m3ASk_AaafZ7Z8X>_J^W7gXGyeskRAzV|f>2QaZm`^p-B$eOGs^00AdITzK8ts+38 zV8<8e6W96g+Cq*SRq2Y}Sudk|cRb_Etr7M4KQ?OYlxt~hpae;CZ#8ZAZQTyg42))f~W8#B0u3nm*g1_tPIhTTqP3v4M9 zIivJz{h@GEk;mlF;ff+;++u!QXcQdjVyr%5>R#vDZzOi2Lc6oLIfuR;AMyy@i2bH! zp6G2(>8iIP>2Ab1ZF$~`n%H?;duw)1iB1H&0ZNkSS#q;HIo&s`rk3M#ct!8=v8{XJ zKm|!ZQo?h$D7edt%+}KK1mm)HA|Ftyp!7q=LAQ69q18`q?L-W=#=A0qph9tCbE^k4 zx{D2}$-Zlzvwol8%D4$`3XS#bgveYUVn(<4`_H3>HjHOTCm#n{{OH@=<7JsqHFm@* ztvn3gJ;eIsi#NCXZ$phQYz`0?u)@ra_LJzbI>i|!j?s?;M%8v+!7Llvh#xd3OXw7q zHhPzL+KeSX;-=8&AdLUgHN<399)1>YU_Y%YUw}0+K9nlQ0TzXZjzV5~+oSX4Um1T_ zMV86d)s0_3FwR*Xc|Gs3K3&cNf7?_mJL5)dMxbZKWdT+mNP-^=va45t&LxxZ*H8iK zq~o(5L}uyVFD~V0SUX-eIj2vn&t-r04!Aq?>Z5fqhnZODQZpk;ossD8LA<#89O}z*PBe}KM1%5MSeUvo_-Zp0 z|8-*iF1$BNT*>ifQ5Ec+R((Pcvd4ieTI1 z*eFTC=xi+gRlf%g!2v%#P-)4!lb}-df?OP; z#v2QSo==ORA%!`=-?X<9|8flcUOBV}{YfIVx1)1(2YZ~iAKMT?DjIO-%t{N1`5LHJ z3i)vdO?LD-O`*555_AOAD9n(FBpJ}GD4x>M2`HS@RhUj(Yh{6-!b^%( zY83};Szo7L>A}l&)^#PSQ~SG$z|+uZS5}>Km6cVLzE)aXzX}crgdb-aD-OJ>wgshz z{~c86z$|LLsfk!!S0%XU!@inOsfbflcufgrg#0z-8@1UGljJDxImt)%gq+gbtvn8L z^4gOw_$-L*3TN-7FG}oI4j>rO%V45W_t4=#wDGmt*3unX9noi&ztd7&Rn8;bQ|gUU zsFNs(l&CI&#TbKOhs^hxz;!nG{39)Ou3?nZmigq5#Dcw#Vf^dG5$0XV3zsymyH0_eK$Q`ky zIT7_df!Dr?|HOOyG7hzdn$%2)glp;FS#WYjJ`VGq{IH>_;@P^WGwypPp7eN!*@Sl2 z{7p@!`DM^PPpF=CpQ8{bw*zq$Ogk}JUeYu&RnMgQdiD3Z&di2Fz2%wfDJyiLI9fZN zCE4py!j=O3PmXWf1z2Q!GW?vOZ2o=?UPL=H>rE3mdo_*s=VCY2imK7Jh9pO~8pvPX zPoHSq{pA8hbH+0gW#c~hrOd}0+sYb;Ck@=+COByS)O}1ZpHH1` zuRrouS8eU=nLPaGnVW;b)4uCLEk2rq)xQC0K!G)td_C3x)X ze03YWBcu9kx|DHiQfMW*MD*RxRN_rQ^B_%QE>ph)3fZn-g!WbnvnYva)Tw-fOC~&8MNc#Xr7f* zb*VGhT@T$2$Bt)bUaENha1Odx%Wkb<1Ti{!hk=;#@BcWRWMRbC$nl~dLVFAsmLu=t ziR(SO9sF5fy&OKt6+_TdG0VT@79H8tUTcFFa=(PW9Y$BhH}krFS<6E8Lh&k{X%c6- zKnbs$XlQ3bx7b#WN1=Soe#pT-mOeY_x|{TZ=m!;nUTMM8z8DB%U}^KXOvajgLriZz z4*99)nvX8X`>;yL{Vma!n$4AWsm~9=zG&FBUM<t|KSq7dX;gxhp71PVIGs}QxLXCoWZVb(O7 zplc^~k?KDxR+n5*(}S4w!I4TF5R!mbuAO=Qq@3vml|YJF@?tEoE? z=C1q3Btki+5a#s|8Fq=s2f^nA6{jpx+7Hy=_^kBIcCaC<@KfSbTJ2#Lcgoq z{Um>2LQ4#8P1cLxl85W7carGW?|^sqYWJDC7_N0LW{rI1*9bBbC2Faf>Wix%?@gvx$?{uxOIek40pG+q3QeBN&#zG=z( zC39;yv5G=@s5`VL>eZ_=;QT`2bicR`-h`?wpt^p?Q@*Ql9tG&UTsD^B-sJQ2^BVEj zS0m3QN&;&crESxG2xvX81ur2XCQu&J|P(_Qh%;3fl%){`-k z`(*l)1vO~*nU;3<(IQ5wn~EG0k=d6YeUwFd`|H%zX5h^Y%P@p`p0>7-nq#<4c9NUm zjhn51ikthn9BhI~)yf!uaR|#Q)W#!^etph(6m$s*INdxdW%qsY<^6q$vyz{$1%f|W zWP(hWb`k~2T<=$0S$x`A5E$IxB8^dOp(&7n6}r?;Hnx-{auG;OE4&z1W+z+i<=Ci) zg=6^!%JZCWvxza*d7-(}6HyNEp$hbSH{)NPI()J!9~qEuU&)1?p_TmUJnPO7$~k`q z%N4pKl3VQkwOfjy_Ku9k!RW0NO&sv<2%`UTGfcv!aI1sVddldjOQ@e?Uw_?Ch3#{~ zoQxeEwo9dZ5hknSHg--#?)UG~4(EGKO_jf#qR%t8xOC|vKY$S*OHQdUEu*fDjaZ2o z!@o-%Bd(yD1)U2ng@+K}AbO)^^~IN7(b)nNjWo@Ot|^pwuX=0!m;{@_%y783p$h%T z(^*oFMO<+0;?=2=PRDnSuK0>^?8Weo9QH|b zs>j4tH|x~ zH;rEv2Q&sVK>k!D+T*$IF7m;Kv&mQ1=9DbZV#WADAeT4oV~FM%*6>?F$2u(#Ry;QspxgvM!2!EzUtgpu zxN$m>nNQJF1if@PD?V=Le(I@2pN0ecuAs;0K?*j}O4=TY-I$Uda!4lHOMl=WeL17m zmJY4$U4OnbUmeMf#h2`{Gsmf%hM>j96nvn24?8LzzRsNk_dkK3bkEal;-&q(7_N)I zRR*Aw*|jx{A2u|xt&sYsvegL_bgY_y40Gya8)BRqr4HS$-rK(trguuU3oMtv@PvBa z`-lV1f($UxZ$yKJZ6PzPYiJyRI(*t-p`%}1(T0Xc$!sMZAGlp_;XAyI!ikA-99zt6ED9wOq`s*Oi4_CvHc@(P~NykDBDG zOfP-jh2)Ke_Chw}*s%8X{DTS`-ESjSi0Hx!yxYrjQqSE7A_58Kp_-J%P&+p$O>%+R zQ_i<9#zoMiqWlw`4idcEy5>O-DoZ9e2TNH=hnT-@YjsZoNUsEoU8xXx8R;cV8W-8+ zhq6p8GbGP$_~`eORI@%Q_cA}b*jGUU)c&!50dPqOY_Yl{}5%WJpm*=)+d6eZUc$q^TP$`TlOmniY^GBT@n94A}{KDdAUH>INW5_|tb2IcN1xM;H#0tf(f2R7xwzkJQWDS&+AETgW>2{p*mkxPSzxvRB->bRRTZp5 zx&nd>CiuBai)PL&H{_@*#(k(|HKN4~!z4W`k*R7AceiM4me^Q^$GtVKUG=~=O6fzN zAGw~K>>n^*FKN?gQ#%JXYP_(R+}BeKiS;wq?^Rfiy5+=e;{cj1W$JU(W~b=;oY6DR z(Q~A5t*jT-ke$?278M|RV@^x%*Lmc5{K(?n6@%-U3$nvN8ikcYC(tYhvyDwqwhIq8 zcX9tCunL%_0Nhe+^?@bxHg_cT(ow%QQ2<_Pq@QbLXC=*?TdQjMEReGnFGEHh(*=b) z)2;Vz&mQkj_>-L0Jp~yYMYnX8Wo0!gO3QuVUXen9{wgdeBxbn;a%4Hr+E>205jd$f z9tM9p|LBZK{3<6w1LRaKct1gsLsCX00ndb*+=4`)W8$kX)xgrlk}>O#!8nDoMkKw}Q}+L;v#D4Kn6G4KVFviIj+A$$_x-I0baj z?ha%IlC|;K?ZA63$HlLKJ8D_MQw0HTXU9-8o>Oj3q27`AO8utvCc56n4j1fCA+*>w zd3ku8yaRiDy~ct_oa#&+r?Ttsjbq@yQx*1t{hDD<$L=B_e0(|~7p1-&^=$9^)*RSx zz8^lyZW-WD$W6x+?NC?tn7S2FfecWc(Z@U-pz)F~)NlH1Q{h<%*>^rOWk|>)olD9y z8a8O}O$h4J5<^nl?*?5(LJG>$1{aRsB2~bapKIK1K_tG{Lbsa{DEPLXynC_rhJ6Np z_nJUXFK7j-KH7YZz$|SWRMT9d;Fx}-aG>TFBYv1LQ_K)Z^Ziot<)3F2$v8l{E}&55 z1@{_F*AZ`8`GRGs(=5L9=A7Hs4-_I%(+5eMEn%uYQD_q)vtq#Bkf{^xLe>AJ)l^8z8XKMq9fh$uKnyCBYy`=7)^!+3;oP4 z*nu7&Up~Jk@Fc<~pWD|K-yIUrI;Xi`6@0d5R<8d72O!aqUHsZ=Y3V7%6eyx@{9Vhmi1b2h zibD9}?e3Dc*7xMnf{GpixZ&Rob8MZrTsJ(mZ1uuR*R7>eEuSP`(H;IwY5t!5bZp0;rX$v` zs>^J>w&om9F}n+fu`Sq|yb{bb1phZxCGP~`b(3yi&?@8|4hYU~_%xj!Cywb`H%2?V ziebUR&;`rg7o*T}os003%p001y5QB@W{H~UK`dbo>%nv-r?kzE2eROlA`#h8_8X*0{=L@2<-6LG2c#D;K>~4X>ca6kbugGao z8~DUcQSuwMf|`soowl<=^iZ&*|NfSwtnnTP5N0s9a)ce1J=Ok`w%UWVz*8}mswi{S z00uTrtRSq7<*){JWrbG_N1}EZ40C8g=o%&)J?`x&Ua}6 zwj>!&?!dv|*ww_+?csA_rlmQ4dv3dD3Rwoir+JZB^_WEbBE|Uu*_lUSr#*EK&D&*d_L^nwOTR&7NokD%*(+Q+>0G4$HaO<;i!RzkL&N z_XF`KTeVfiQNwX*%g(nD%1!aZ%QMS7mmT;q2EVP%;ugLa>IT^eE0a;R=mXe8o7BBW zI{_&#Dpb%UGalHh(}KMk+6O`aT!wXOsWR<|Uzp<7Mwi!R+N0JJlzh!&k z9dV-Hz(WlX$*4?^jjq+0^qQUFBk-Op~Ew&##;6=wcfB>>L0XD`M z;ho5W0taNZ$aO9unV_iCM*L<~n&&H}gA*i?EN?5<>XO}w<0_{67uayh9PHbmAL3I!enLn>O9i5%rJ|vE%E-;~^cg1u4GpUh8|Moietv!si-?3U zuQ)d!Kkwg8&~R~a@gL$-5D-xC($moM{*NzYD?p3`{6YJGj`j?AK#Yb?jE3w0Kqxvf z(f+xBe_v=1&@nKvuyJtl9-=B#69Es<(9s`YpkrcUV4!Myquv7;#F&rid0t{YQTu@X z%$|hTH!1^%;Z^x>QuUD|Mm}Q)KU_RAatcZ+CT12^Hg9$;XgV_^Tqh4#P|^+hMfz@+EFdi+ui`-A z8as^OkumZwGadhh_7AfEKVW|UD`fux_J48B0Ql%=DCVIP17P6L%XM{Lx@~4oP&*PL zDpAq@wzp%g@=?F=wT$~j(eYtNoYb|Ts5IzeB@zj+v8#mF-l=bqph>dur|Je#gtTs?nDWS0+#dceDK{cy-yQdfP?i z*N95BB}pPzS`bU;fEi6h#WoVS+(!ae^>%k$`z8o)r^DL|ByeZoPK)@Vz7#sHR*{ngT}Ys6s01_wm)k0w$w3XJpVxm9c|{^nbhH+`8DIF?on)q1Ue^;$V2QHr53bW?oSq1 zjRY-bns>k2EcqLjn-m1rNHzK#M3?2Wf89p`mK zm4l4owdXJugowtX5l2UB(kYjc--05xQpX6^!ZLqcx5pxl*IV#@n7V{`|@?jTs4~(=s6L{vd&Fw(aBgj(+HBw>+vPFRJCq zkLx0rwa4ty%x}*_>_UAQ5`RuWbZ=x>kN`OZB~?$V|M;AjreDeaBVbx-e1`^@-|5(| z)zbMSsGfhLDvC&2=#%VOywg$^cP&W2{$Pm&yo8Vd={r*iObCQCsDCip2(%Lw&V>?3 z5<3KYCG%B!iLAW70r)uVuMBtAiBq-L-FC%B9t@^7bMLMgw3v+ABY}6>f2B^lZ$-P( z0qfQp?d)&u!joj3}Zq)t}y+r$z#>;^1bsUHfKC6tyT}y)!T~ zfEI!kH$JHxT3BM}i!EM3APaN%@hT|ff7vv~>!mZPYHykMZ}|!beb%z9r{eV6+u@ev zyMmG9{eq}b9+ITr892`ncF;;|20+|w#v@c=?bqdH51WtxM6!C+&ExzBp2AbjDy^YU zNZ@NU^q~D3be!3^dHVsnnL|nyBr0+kQPj}w!h*;^ac67Vy30<7biai*Ykt*}0STy% zA%T(T$xtiH8UdGKSNc)tKUz-~bWh+DyK!T8=k1`9DiFX7}ZFX3E^@#=O&KLI@KwYG%Tk$Dw{pwz|~HnaEsE?fB`Y0-u3>8NUrL6BDNrPp+3wmxD;LQBTY4x)qg$sA7Z5&cD5) z%}ZM6PMIHSs&^svxS{!2INM!!ot9UJ%&pcM|5QfptWBDOO)Zw;-IBHX!PR? z`ctaznVY2TI0t_lsv@oaGS^_}@|m&3wyYpcj(2ZPaPB98ie}m|YJ@JLnx>w1PQ1~} zQ_G>kJu0C@+jgy8GsJcZW*2GV!yari@hhHUqn;W!xYJ?Sl-PhQWu9#=nO3X37q7PZ z)p+wu9nwx!*rlAmt`5tFK()g(RynuiaW)@Q%j!zK#Pc7e_AQf>GC66OjF0(ChYnQW zb)MgE7rW{`D{sG_l|nY}U00<^-bml;7LGazyoVN%LVN;!6LeDgv74MlWzx3D%M{m3 zTUOV>3uW@dHq#}Add`14PBp;u59(}-gxMtWb{Mg8j!yaknNI+bdD z+&2|p*xe<^{?IQS>s)9ss>BRqbLDr3j=vE;eOjWV5Qczl`x298{(nlm)gSQBIwdv2&l5ps=j7i>u2>6qlnm0OV2UgNdw7f zUt_Nueq&H=;$YAYdyQ|e$Y(npuBu=_M_stxHnI1m7ONuX%O^lbYekmOFDROcf+)pg=pciYvkrn3eE%;uk5rzA7vbSNFwDAd>|a; zx5i)y>(lO2ixlr;*xB%^T zkLhGu=H1i!#BOT;bCm8s_HXCk)PJFc zsj@W`NULJ#)L5EWjn-G9eIfNI7YEpxyxp4*gYv+vcIQRIq5jue-cB2XXuFQ&hyVk; zP~Y?%YvrWW1M+fM<*qq~#p9IH<>1M@GI(Gzgq`n)Jeoo4_3r58w!wprplV^+y4&i! z@zptQ`%B2D>dz#a4d(78-7!*9>AiErN?D0ZW=v< z_(Uky#z#K$YlAONVote#LN{<0uO7!v6%(f~8};VuyCH$vmRN96cf-R<-#7x}5B^!g zNT4RBnBz7NUiwuij#rM&VA*S2+;K~yXo_qGeDHi->AU>s!8`s7?USDrSn};eU}6nU z(O;n?4Ou-Oj$f;m$|Mtkw|^i3yzMg(k(E#+|Ai=hnt=D$ z)Ihlp5u`q|idDZNOv!$BBo2;AWq|`&muEbFI-Ky@G+M7blzE5*a1!ipeO^MM?i416 z?Uzt)i$4w$Xe(p-YL3u*zOJpIj8gVKrn`cABF~>nC9sFbppNi5BaAwlljox~EY<2w z$2GwRSx>J7hKU|2oj6GlI9cb(AOZ3ClY)4qujc_^cEwP#e(LXA?n-Nh3{8o-lGKUb z_O5i8>WJRSHD`wQ(92(H)lda>r`0)uJ-0az=rfEz}-JwgB%&L${t4b7;SMhBDrrwc<$=wc=DgDJK@_ zOsJ((F6Vya-6~4HOM~8SvC&t)IznvSb0nr9ZjVc&|5Z*vl~IZ$eSlX2me!V=BUN8I zvczwzM*@Z@o4Qs0*q^j-9&{FPGx;Io1u!!f-b8%($JbsJ4Q)8p8lISmwSi7XVzf_G9vryzyB8t>Te_+PnXoc(5BT-!M7iYb;hT$*U75D7;S`V2 zMkL_TNuGAL?0SfL(vzU^wXHm0vPo>;@a!F@-ni*wIjiOxHRzVsj2uC^Mk1w(4eWH9 zl|y|F*@SlvjV194JzbHK zKK_J52$@3W!T#BwCvzqU2u|HyW534Oz*z7e=x8_&bjAioV3Z*NV!1nhreF$`OQie~ z3D|KUfexMEJ~PWtno9N#4RxTc9q5Hez>i-%{`jYSn~@`!ptRMy{IP2DLWy4R%!B@# zJJ_ko!aDQ{mW-f!(1!>xM*^)V2dYEoYheHdNmVJ|e9Q6=&Xz6h??~W*9BM%M7SKLP z`c*EI$a$S1__+6|(96i@i^Q^z39V3rXlX126QA~TE;MX@U*rOMwzF} z;20QH&t9kU2->#ABVX&F^s6(+ z8};uxVw&VkgT_P5@Hr$fikjFnV(l*i+O7T{th{Ye>>kF{$I$+E3X7-ZVD`zaJ`knq zNpC{uO-I=?D)V{LA@I%3meWu@8$Iv-=1taFsg>RNVwh8W(52k=yRnwt7AhNJyDLo1 z+AXWaBfJ9pVK}9)GosPkIuFjoK~!qIdGB_Ea*9rg@}p}Px`t+yNFlJHs3>E?8LC%l z;GdhIV~z-v>FP5S5l@4A6Z1zc5(#A7c(zH zKjNv;W`JH=1BPoXz=^-|7i zC21s8F_DSBiHlI6Zsl5tE34?*=AqZz%_HGb-6~ji$guq|7WAli$4hFy>)or!pIT9g zEx~$tv3EPn`)3)6sUdu1ECnq5d4>1vKnqv3Ut4ifXv9+n2MQeYNo;XMdcdpgkAh$8 z6MgNHouuA-+^-HebCG4j4@H|U4S0NoKw8DW&~)?Px6)wI5;g(e0!QbWbQo|UR+b*x`vHa3 zYoa^wWUQL3P4=2M2X^ae?2$`CR@JlIihxj{JoV4Jh|&{YGh551C!g4XUa0rgq@0IbjC-k-^zC6+ zGRd2kR`#6VOXJ0Neg$(s@!Yfdu+|WF-quj1?>4XMcKz$^VUv|iAF^Q|yJs*-<2M9I z&PYRAsvv<^K_Aj@Fi}fPmRT9dd8PyJ0z5v>mb)eP&Bx@QMB=TYB+^iMX~U=W2Y8*@=(A1 zd2Uk|p}I=-8`j}av^5ii4+Eka9w?bx$05?#^^It}Vb z9r*I>Fy)j^@%3}GFtEpxMjsybeK85a|NTx5z9z>-}h#wl-w+c8ymdg4PUME;!_Mls;+eRzNQa0sa zFD$QWB+%Rm=+sB!=4#ksM<;a6PyZQ$K)5+KDpO5TrHqB}Ux@pF6j{;aq|Zv+%s6a` zP7k_!g~{aGkC^I;eTVPawHoSPE7-jm}EYoKP+f!hc<^19GZ`L1g3gnm< z73Ns( z^GA}^1sEx4deVOEci7w#JCTV_U66ziB1OmD3Beb|7Kgcdb5A01v%B7r^?RUSYXrmu zW%AsyM(D?i%I4-75i7Tx{9!XdQ(RT5Ww(FmqlF>LGg~6yuePeid^|hfCw5mFl`4Vx3%+!1ICU zl!>TRPxHCV(A%3y@m*P7h05VPznEn1lv8ci zXv513C6U;xYs>d08`7ML3bPw*7%S0VNSwfmifY66Y^*>gj1?VzC+1u!U1jK|nEbAB z@rTvJv|n(-JEuvp90QWk_o^OT3Izt_@Y1ogpSAY{#uIs2l%DAfY2(hAU!_tzi$OPR z)+tFYiGL&{I5Nhaj(KPMJ!eo3a%NoJ9}^xC4Sl-Ly{vi9_D92 zv%1J4w%_|BfoBSaTHaZlU)FTWz4VblROuw5Q3DA`M>nplV(9Q^57u_-u4%OIHYAN# zAML!San$)vU8Fq~mKkH(wG!I07a8>vz54UpWV?0g>2LSe(m~(bi)vcpQ#36fB)Rp?&m62Z6Pv)HvM5@1hm7fR~-?uk`o+#8@Q z&(&V`SuV#Loe;*)Wv=wrDsE!5=SxgftLdlJq}rl$`UVhRby3%nXhq7|2iWEAyqb*{ zDaOlh?p$rhV`df&Bca)E1UduOOPA|n{OqrEcvezh(xK7F5q}Z%ox*Ettdr=t6gx`r zv2<*TGSN(YUfsu~-L4nTVX-V9mieWhh{z=$LFb<)79Q(EVAwm;Gp)S5#u34a|9ubJ z<5b^U@ug(X_!@X#RiSiKu_kF_-n1bqOPi7y-9N(P^qB9v$DSwu&+* zc(TvrbLW4W_{oWfH(|cVCGbF*m#_S@dAtdbwP&hNI8Qsh-HRG{ z!%=O%EFN#&hOxuWIJ-#O;Qga#=J$K1T{QT2*-^7IIYK=9=zIA*2voKwjB)GNWwfk8 z0t4@XA@07snXEdMLXU*3b*Vy)le>~b15ZVN1?xk+?(l3r8pOveiRM^lR1_gMJxA8k zq5R#J4Q7_^YdSZbXu+yh8q*62)-+v)i09&B@0J4MUBGDH_3)ie+D&z&T>D|(;OQGU z`iBLvbi0M;S8Bc84!^11Ghh(Up^Q|UAAd9O^d>#6XrmkZ(p6;twi6YB$ya7WjWO1h zK<&qqR_Bu*O94Ng1?!;R^fTJmNM4iRadtbLt<3Qz8{-iV@=YG+7r>@ec~oddK0XK! z+E!40IJpqtCt0QILMfy8l%+{*uLG5@*p?U3D>vchT&43%_vE(W#`5S`?9FfM5+0Yp zHjH^!C-~j*4*ObjYFqtM`BZ6`zGb*-JN4yrb3()yOc^0xE;E~i_>)rqc1F-cOQZ8` z5W!ba1zIk4>I5k0l1UQosH~a>HLEqEyi@X}#3Z5NlK5{UjEYqZVl%8^}gao{kQu(b( zX=E2Kr!!9o2wCHQ3ob*)VX(!vDE*e+A39}%VWz8?)QB4a&>ce@##(mz2SGhfiy5lk4c~w>DNsH=@(CMUrpD}Jra^ZYzI|`gcnU-O$E5i^26Ft<7sqznufN#!9@xO}rDy*2TB5@o@K}D6>~* z0Ggf;eRGl}xFYr??+3jmzVMhj^{En%Z=*rV1-Gd+_w+w|MN3QMbQ=%t8)>Mfq%D;Tj-SQjkAgGZoCyKw}UY zdr(q!9%h2i6Wxu6{ch6JJlnF~JADYVX;+B*1V6o90-Wl8Hal32VWIewA!_yxHW#Hg zR+juMLTY?Shd{gHk~Yfe`05>JiGd*EG%N04W@dovY}YixRVEA&UiI@x@2czjGlQx3 z3CC?hIaLm#^FKh$byCKdb*elqW$k51pZO@jHM6H%8vOXcJ=zF$?Wt4gtbb11fU^}*G?Fj=iYwqM6lI3R+l?d^+;g{aVG|hx9~p@o>qjr5g?UD@%#a&NH5!Uk6+a zatl#)+MG1LqY<_4lGVSHlUK5-L;W0NkNvzY^pn@;f2~qd*a)wA!Z+WQIAf&qdW+H4 zUfH?9EiK|Na;hmHnsc+SFoBfs=f(Oe1bG~|f0Q>#KkjD5! zw@Rwp4+YnTXEH(sf$NqP14Ib6KVBXk{M4PV&c1?78Y<1Xmow) zvJmpzmu2Gl!sLGWD{aQ_E~u@Bx1Vpo_Mc4&j+k4eYc52zj}P^w|HjSffWXH_>DUyu z?G6~b#4eXo=2Fae9XneWB*0e@w7lQ{a~sZeUn>N6w|XSoURA?F7osueB~G$6>y72Z0XQwt`ev_Q+_ShmeZxDX?E*yMy}~Lj)vU>*#Hg*8Y8^Q8#^+ssfJe z;e61Y=t^xzZ9{0FLR+K*jx$AG)BIlX5Yvdur(!r8%kF@{uiVNQePz5E>)k~J8eH$n z!8%otx}(njvho|5pv?=$Rj}ebYLjM{pDkL{=WZ2|*HzksYscph2fX*xNeeD5kK#q; z-K~|$g&LzAM0UJ7H_!9L#2kuTPe~Zzmb+zh4YZm*l=a@E0(PSGHp&#H2O!D8gl;2` z-U(Zoua{(hgcG$?4j-(FYf=0dXAb3ecaj(@AlV$P)^lva?fKn@r>-y?tE(Cld!0H2 zIf@@xmf+k(7>xPE+Pzb|j}ThE+bIjcd%0Xh!T6pL_O!Te%j%#c`0S6#w?GFpbjjEO z?%a`Sn!|uM(kpJX3KW4F-CnZY*DLi66599YVuXU;7n2UxFExpJUwy)sOfjJ-)o&}N z?igMQr%y%#GSzkvR2Kf=3v_!c{qF1d*JSzTJcYvkVcn2q{Y2jdlcyrB8|FQDXU{7{-uB_Za%wy@hD!MXSv8{O}iSji^J|K$vEFt%4Vkw|xYn=6WNWe*CA;$rt+r zk@HP)&=9JiWVR8v?-`n8+(sz_0nl5YG1$&)tR;y{6eZ zxwq{lIwf`u;DJb3sRv>HY0<5{ZG;e4*@yCuxG+~S9(&&*K%~ZHnDDBID0Y*?qg^c)k=xhz(1lE{fNw{vdHZx#TCh%| zkSN9lrWzfCUK>qRU;^Faq=-~-Ti_QCQGdH7Vp|0JL}1s5)MRnR(f}fj6XG$FJ{h2x`Ca=cuN)EUoOw#GZK3COTP$~$A*AF);$JP&dS1eU zbFum~)B|R3n@d1*IK`FKr0*BhPkFej{g&vlGWBUeb{N$w5%JF!C#?lx3S5b)iZ7BI zVCF)lukfQ1{s5*#?mfj%@cjg7PooT)1~La`1r(~L#_%KH-z^&|Gar5S3w4nbp$Vtg zic>8a%baU#scnh(!O+P+-FG-T8DVs*w^^ro z1poZy90ca25|G`M;-e#ONEx;uFxKaIAL=Nh9IsBJ3}H#)t!qEPVXtp!=fZU}*KS8I zJ%bwEmq+Y^?kWiOo1k40d)5f>XF`hg$vf-eKwszk&b#e)sRqOf!pXC#U1JallndY! z&V`HF86D>!+Q)EK012#fBtUhJ+sNc?hw+5RH_mMCL9_hL7(@GH`n<~RPC7g3^v2>8 zBzU!_^}>1g8XktNE@9h>T}lh>YGQTsxEui=7TP4y&Pkb(tgZUXM3xQ-(}Bh+5kK&c zQ+G#Cb5MS@r#daYp{UocVKo24#e=dSaYEy2q%X2ubZ<132P`zX)AFLPpdFUkyDb*u7=C;tMNLF0aj&pJ8U~~_5ml^* zHXBLqjYZnrQ8^8!&^vEYoEcp07tRJHmlj%Bf#w)!h+ zz6DZ|{-Uhj`w7V#OQBAe171Osxdx{O4!!(Uq@;3Udp6kwP_E8$sG0zZ&I9LKfR zy;%E&&f{7OO-zG%*7`?=>qA6W>GH=~O-P_zpilX{LGB`^UMVQm0;9aqNG`0BtVCP zl+onAfGaD!(}AeoG9sntbxWG|?Y^OkxbxoqYHIpUJ7R^zhqwTA&`#yuFOAnOLXn#Y zbs@`{-*_>gIrGc^)l?VR@_ft}?JAWRTRp|KlrdF6Ov^W{wo@@Sa?h;f8@KM-qqa1s zVL6P6Zx2$0xAnv-b>GDF*n;~g*TzMPr@O{5bQB-d^tQvw>~TVm*qBFL)#VttQuniy z_oXZewLZlxR=N2jyq8g-j4-ENO3M%$t(+Qccu``aFb#A@TpD%Q4t3IQQ8s>`jvnRD zyFm}L3=CRzNuI!pF0zLdMH0DLHxZ*N&##vW@6X>{L=Z!VC|X3f0^m@W)QcFzzSdN6 zxZ^$@11hil0^!tRdCYQm5^`D^`Htn%;#Lvr67sV`#kdF5_;% z32K8%wi5^&%G}RzxAkmdUZS{&8YD4me}NTMT@AV@TW{}5j-uJrXg+g~Gu*<+Eeqi; zgYx01jB2j_tGFGoW;CDY{fqvNRSWd$_DLK*(~&O8%B zR%YQ{|F`PmJvYt;5OIYR7Xhk_M|tqda#!ZM_ru}o9)-rm3cYBD@(e>(oaFSK-0UWF zVj3>DYVT_Ys`cZo?7U22e*5I+r+{5(&(6?_QVRO`R(hwdfC zF?1_$B(Y?0t5lQJU1TN%`>@csMK2;6CsBbMkEOg%9F14$NPvN>=H9&w5qRD@bZM1$ z{35W^wyn6W9hS3D@>{-vil>J3g^#F~+l^*hBq5>lq8g0#fzsKvJ*T)cDW(EQ_D0>p z-o*sM7(9-9{cS*zfywpJ*gI(V478SA6Lq|8T!1vp^4CPJ#5TRMw>3{TWeLaOIXKLu z29lrH=K$~JQGN_7tlAmwuik*WH;z3>5go?pQf6!ZMDSu%1tL$j?hhvMTlpcu=3nw` zfb?%G4Era-=+laeQ`2k1S(U(eL`Bw3k0|hySMhM-EV9E`f5+tJJnU))zi=Eb6$C#7c+(V4ct>zR8a&#AOHZ}9sp(* zcmm+vxr2KL2M-q)7at#wfQW*ah>(zomW-T)f`Jag$UsL=&%`Fc$;86PN>9%v&dv8w zP((xo!YL^$AtWOpEF$!`hk)?$@rekDXo!hvgqZ1>h5nxhrUf9y1A0LAU=Ry{MG69w zf-vm>sp;e;pt!FgDH|Ts(XN!rL2A_W&#q7>tDt#=*hHzP&r>_Bwz~ibKXM z_~;I~jx{ce7llwrOcoyND?X=v%#IXJnvc|=6T#3dx9o;+1h zR8m$^)zddHG%_|ZwY9T%aCCBZ@p<9v=N}OGGW6a1un!-@;jwY?35iKxl2fvCa`W;F z5QRk*l~t(fn%cVh*0%PJ&aUpB-jUI<@rlW)>6yhP^zzE;+WN-k!JosUq!o5&*j!QY_$3$p(ju($s$Wd8>2-*L?VL}1YE#{-iBP+;rk;cKM{ zIS2-@h1ty-_n+)Re_onO+d3vixY^&FZ>5i~LiUFj9%6t@_YWB04F?85qna-@U_X%< zzyo%v*oYM@8+t=0-3d7&9>M^7?Z2-4qR*W6XkbI)PcT3;IqYEO;DQT01zpT(w8jAP z^CUM{8W^C<2?K;h9elV3zkpsIh+_bTGD!CSKI(tx>i>(yunm2;AUf6+QRz_ z12~tRCp7QuVF0JeU~P~B#s9eBj$(ky=&QK?CpWDXuyq6?=fn#G#0DZ@Q-)gCl#~>p zkiYEp@368==wIGx@CeXRp!kbPNh}!Q!A+v&xj-m3Z8gDGX3^mdIr=Cy6L*19rjG?z z$L<;Xq&?6uS(o__)6YxE0)CrVXg;cy4h#D1bw)q_$JapT#JfH~d+9SxiN)9OGim4EbB(g!mZ#(pX4H$_(y9d4B+K+RcT_grkyBHu2ZYXNnh8s zHd7@^?5eXrIQ@FZ(8OJS!N$<=Blkuu{+#kvA>vVSwuR|%Yy`5*X6cG?IR56)(&uIq zc3cEI(|W`_@ zi{^cV3|mi~@5TvQ@{*hNe8S?iW!=s%PV;~``NP@k!T1bBPtl|tG-WZ!lG$w z?!Gbk=i0@oF~D$HeZmjrc~r97CxlSec%HN65tpQvgayT*dQEH#Mg=j=IZKK$zr9H zN(;6_M*adJ_F2yNx;acl@mp9xJ=(9D3o35j9CMB`59cX|=2mj`<~}EA4^mk-JAV^$ z(lOljF(E^84(&CFHsyVyR5vK#QtkQj4Q!uazkQFsBi!LRms~FWXwJv|+C8ImZuX%5 z#3MAyHcD6g*;SW>sEm!gqK@(@ z&5iyjd9<#GmVj>VzDsF=1N*y*Cap3ELQXv*J|fpzn~S z0;T$>vA0$gVdwDR#~II5JYZTno~LnjgBA!?ep#3h|AV%Z;q_RT`n=wDpuedOD$;c& zt6^|YwG{zj#y0R=N<<2PtnV68ZwT)-xZFClA7D9|`So4Koonfa^ z4i;2U<3C)(|CRNdah>11YKQ@9ny+&#L6#WcFaG~F*CXJ{4kj=#v;nY34AY~jGkmKe zf^h(+fLE)_cER#zla7*iNu&2E*Qe{lZ^hgr_WY&-<6wBeNTN1(eBmz3<8? zB3-VS^ufUd#AT)Vgaxp^P{>={)I^%;o=MLZpLd~*8eLi}tzVRYG&;M+whEWxdl zCfEe4d=u17Y5nM^P~_p8?z2T|=pN}+2d`=WGUdG*PF(tfUj;RXrBTv>?ARsjZzEEP zmQt)~ArD-zecmoOJd63k+S^4)qe}>RKBAECa%#l*)bq`eZxm@`bm2O0c6xs)%b(al z=D4Cpw=g}!Dhrl#H}~gO#>R`uf%Cw5(im+?JS5>%6SADW4d|19BV|-AN#sO`Kl;) zB$`J!H%}^kM6F=e;i3Ij>Ps{{#sv3KZIJ65Syt8hFG`RZ@dyE#j`Qd;re}u7&A~gW zxL-=1`2|+M4t^b!L?H*nsI`071J=7=MXQEwKHzx{s>7T6zSL0HD4AvQ=TzHkI@cwV z|FtwLpQT$ltF#Yu%T>Q&M5K+U)_t5gG4%Cf$|tX5(OoJS$x&E!l)qRWgFmD8 z2?ls4-S#;;g&nG(bIoHaqVfI1+ufX0ztyGzE8+kZMaB-RO$rzjuhd%Gcj6zj&kkx% zb84T#(zz)D2{W}r{d7;dpL~^zX zuY1vK`zkXjO1i~hX^q9iX$lN1k44J_`pLWvg6LM(+Pg}}18DyH08AC-8t6-oC-W{K z`l3BjFk=+?%JeB3>O7X{;YBrig#q&HHWLFC11z}*ZOXlR%%<`QTu+$Kn*eRhXrMGphmY>c*7-M>ZkQP<#*6sN5 zkD80O9SAJaoTVKgFw0qBo8VV}*Og~dQ0Wf+IXY?7veeoQo8>vbF!G!_q>Vr7DUEBZ zUyZS~XEQn>;rp=L9a}+qu~fugQ+U_%8rAf;`qg~3 zelf+i4(jnZ2x9OkyS1(&U9!=m@c;w-qClf1zxc33XTyJ*NfX9hRt1xsn0Z}uk2oc2 zK5Q7e6h9k&wkjn}{CG{TT>SC+Q!*_+3C)*|2ztZxq)ZpO){vn>u4>i!jW6@axQfKk8KfJasin==M zKN%kPZnmzPIWJe6u(BK9e}@6+C!TIRtZ_C}o_>PEY)*a=#4UiuO>VayPgK_AxrdG zaZ`BISNvjbHjWUR#E<_q8~-25Ug$IFcO_boI@tsDBBt5(Rs5_^d!=$fsgTwRef!{J z(W#C4?EA0!rH)Oq(~+T?OV&$RF0z{_TI%mH+N(`IIa7wTp zTHqWCnMgMVfMcRw-m-57Hs%#`W)&r$W}CrVPvnz5o`kA|Eu5_Bd~WC>y~k6y`eanD zE$;00%QA^{4*7oB9qJQNS!*qb{(>dE7qPZqO0|*&AK`V%Bm2JeM=CT*9w54~Jua?y z&rhVr8*IRJr6&r*rnHx9XGE2-#VM~K6$Z#$E^gi#dQN}OG+a#W^a(kYhni1Bh4Oi| zF0w7FYSJ7=80J;UG1Q=9>|)UaNe?J{TOou~pxW09xkuNo35g@*yyqmsU8z{MU<}}p z9~YK;SX86AzB_6XM$FX4)*UfPG@IO|&eh=~Cfs$MEY>q@-0-$x-&6JGLrNz%g}tG{ zmr|eL+xn@@;&tM8Dhq=9pJ>;DMy2M?e;@i3yyp4aXQHupQibjvb=(*1bX{s`V{YtQ|MUZz#F+3k zcU~>mRqXKIW_dS05s6lwGu=dukYGCHz*XG%i<{)JfW!8SoU@|erY^eYE(k&LXW%Nk zX9gC$#s_S>Z=aggHh)yd`Yaobdp;>4O2VAc7wO%_9&J9}Agk5ymf<_NPjzm#ksemZ z$+kSbIT{*4VK%kPq}$VEAMTftE*4LgVT$64S|mS>!OEd!ONWAvO9fz6Q|EN=%N?rS z2{3>tbZuzjh$cbTNqD(-u5lSXbUBoJmztw?*z(2K?gaIB_NuXxSH-tw<-1qL6TWJ% zEQ;BGenNqMSL4%MTpHO^c1_=Sxb)Ob*wu{luPH=UOGw+%>V>MJSNI<7tF<5AW2<)c zgDxBZXVy)8=u&qj{3r*8iux-ux}jTHOzg3_3*Rg4wsZImcHHDFyArktip zBec{cwTHF}PZ>~y&2jtAHc{|KXVEZCW~h`}p|)%8?(izBFi(G6OWk+=*f*7XRdx<{ zmqZ>N5DzVzbi($UH(eb@&0>`ePVi>ymk%^sya=MY2y!#5O~ocIdcfNU7T=EW+Krs4 zKRRg2jaR`Y9UzyYCR(I-%U=IFJ&!&XF&2n2QyY1$I+ND;slbJ}>NhF)8F&(<1POhu zx?H&-jwmQy?_E1Ao%q#m=;1>qy0Gd1*c19>;RZ3aq3_Id? zRh@jkOZ>JorZ+b!4VU)R<*f&?tGGA`hmCqU`=wnwzXeXU&Z%oNglP>_JmMPqW z(YZIv8Wm`qY9Eh)gi~0l5yr|~&1+gTG9yaW%%9?p58m{d)PMNkj+D2<(p6NB)FfGi z@9@30d$91-Oe8%1cbFswAW$nV4p&)ahir46#VN!wYzMw#WqtNm`cT6<^td$Iyqpw~ z?5+GfqqAIKh?OXa$CO(LMWLf?eajL4=bYZ)&A(F;z<)^Toz=!@J!`EI*hNmbe#yCB z%b?3QI#a&o%~XFdZ0QcJq)SeauW)zIch5&v0n^KMa4xqz*gOXV>)ZJckf*E(XsW;+r7e6h7}3f03aSDd|r~DLqU6*lzBP6M|D_ z_qF=3M^ndTcTgcmnaZA$1zyHg69-C+CpERRsrI$`jfrTF3qg-~TLBlVowTPLThH7X zkovo%g1w`dubAn6Jmk&cA9&O41Jg7Ssz!By{cf z4m^5K;>_w1sUO+G=&58W^@h^3oNSNX{y>=**IC{ z1PkuQ(Z6yp6xAYaD|Fs)y_@Xe{>0OUIBus$6z-NAs4V?vH$c=;89t2)h&__1CJy)8 zmzNxTEJzveQ7cBG;tx-Z*Ux8O~=Z5{8!65varJCp=5LaLPLA;hGJMq#IZKafY*cSa8nV-@Plfc&bj*Qj9uglDZ`jG zmP*g>ZKFt6VkPC_oD^Dtomh)p9M09jAs>C#_EMB<18?|FR$8(Fyf_`7cFTl6na%B| z&m`M%=|rErYhQ%`Uv+Uv90lj}cdEXc(5=pD$ypt-?Q}f47Kr z0JM>=DC@Lh|4MoLXPMxwC)OGpGA8YfxHvR)f3bmF&U^>;RCXyLD>(GQXFzcfTs&$66b~e2}aj6+)8wOk|N-3S5#rRq= zvN}J96)0nd{#JViK`$jRK#$BAbeos+1`dZq|B;H7r^f(0q8LD^=DO+jjBcOcR!8|u z3H$r+LPb%Brt1^UF@@@>6}GloD!?(~53dv)7#;V06=~_}BZ$MFWfq_)_|7sTTF)T= z+z&0c1%Jy;q}2$D8I>(0B=dH{dh_^CPKKAxdj>U7)O5d?3Z0QaSKPXw25mksRj`)c z#gI_yHXjfB=eu9B~?{y3#{c`%b)w@`<0LRbvN<0V>cEkR%{MEaj&qKGHiV*^M1!3YQ0^dSo~Z% z1f%RaEO9Q`15*dUPX-imlqljlh1i}umiNm!C{=zkG!*5aN}4Gd>7sorkVRQpb9lSy z9nzk2)HV^Q5T=q(B6ZZJN#~Od4kuY5z11>;Y}~#tIDDTv_4F~OVu0gKyIogJ0Zx-e z2C%`nh+3rgfIaz!$?N6=MciA*DkPWj-lwnNL3pDP#hukBC!G>*jh$1ABs5!{N{Ee- z`V7K=i@`QI$i#~Px0#JSe;(Dlle=P1$!L0}DCCrQw9v+;$j*^X#0Ss27RJl33@yUS_-X`;%h^+5 zw=`M4G-4dSA`4d?Zw$DcktLQWQlG&a6I+b5K zyvU2o_P;URpw?>`@kf7rx}-ZQ{I%_BAaGBTJE=DQVVvA)@R478uY(fLZ<1|DRemkBvwyF-67$_$D5_bt_XV2P zB#q*L(q1ds#MQQZ{21x_coZ$VG#vl{};NNNT%gY|#my2-4 z-kNFGEhbjoTb2a6EnqCJPFLC!tWgq1r0B2&>R7qd=#k3nSuyW+WUQ@yT!%l^s@+Jc zq9mLA_+XlqeoW0neaH?D?bW7*ZF`c3x9+ON2A?a(`VcpJunfA)w18)R<2`3XWvWSE z3r}wW+slw2d!bJvUfv}~Og6juiL*2$$iyyZ@&3^Jl|Elo^C`hf{qkvvdX$avAJ{L> z3xjKJeztEhXx~G$qX8LtVyAph7v#Mcb$wIEH3IH^w?~}&2u*`@smd%HTWeF)MQm3K zBeLOWS@~Oa3A}*;rn_!bR&Rs6wF&e{Uti~+94FB77mnWwI@$c!bnx%OL7uY7q$lOQ z;%~8~eX*AfW<18lD0_?bc<`8Wky^`|&Hf-CYA4OI|*P~~8Pf}mE(Amwb2BRY0Dn8p!G24OpYMvuf4WL?oiZ4qg10U4QR8RrWQxx6g#OKcM zt3}mf0NLWgNx81V4NDfWi5{}2`;A`hVu+a%;oDjV>uT$P3Z3jptT0~KqNaa=fyX`) zhSx#9sW)q3(dFSGL(G{zU+oS1s-GB`qPiP?pFS8(ThPz#59a)~(gx!xFP`i0D(Yk7 z3MWe|{)RJHScLfcMfK{f3M*rVD$tbwgV&;ir8z< z>JI{(_1{uvB<}EM2NE)^)W>ABhh)5LYkt2}Z}6GYgGUYz1K1hq#=g@tojbIrH9#!Q zBM)Uswg)z`B2o{c#)Eu8`7`o;sM336tg@1vaXY^ts(;Qn>foGFl4-WE#-jE+7o$t* zOLM!C^IDf2AFMR)DdChg@#a+39mwVQBU*>MN6c{0vT{>2k?HUrM4|}oX zapw&9h0Hceb*(Cu^&`m2Z|X3>*4Si?xLY81;;`}n*aJ_Q-Z7>BO1!S|w$MvV6wOKz zqzF%bNX|;sMjx%}s>EyB)Kcx71?EqZ1W$&3mc20G#m|!S418TJNJ+SQD^U=eE-jXb zI1?4P_NE)nF;Pgag-lXvtL!S6sPc0mr>^f`iYE4%1&~Wje{+yDpocQhlh&5|Me13qoF%ShFg;**Qv`&jD!J;?&ykOo$9M z&CSU^?na6+XGP=_HO5t$wDxjHZ>4V@r=R<`7wG%^p*s&Eq~dmY8Y?Sb7vT+}q|9o6 z=4FRd$?Ox@-PGl)cnOmkKBgWHxG{(sx4AwpIZYBU9A6^>sdBq$Q{5`bE+u4VYyyT$Vgkt%jOz3fLLu1OtVnJ=fa>DmR!5(O&X`$WD z*pH#c?1k>x`i9E49lh!TR7<)`ln#0{1D;)wzJ4+=!qacUO3YoU^=1`3YWci&I~PcPdb@m0Sa;!3pyGF1Zm2c3!7e{4EVGkeWn(HL z&O{?1!7}NAMf>yp?os8E+Gye^^vIW7!R5m}dAbzF@>kgd?$1{Q8?c(BLMJ{&K0efG z3}o|5G%;hdl@v8Gi3har$rjnw5~gC+(MX`-4GqS9u`+WLWAW5h^L=h%3;{DPR!1^B z3&S?r!dMhJpNH#>+eC zqoNzrTfOW8j_*L<04Xd8!s$b{zhAD5NoKXT9eY?@xpk2?Mrx+=;x*#?lCWp5S*b~ZOClTnEyE?AC@?zuu{RIP@AUXcja-9p z@q;Mtyf3Hu!^$=~Y6<{5#Wp?NV=fR^HXQhyUBLdSTj$LH^pBP($t6J#2FT9N4!$K4 z{+a<(-gXlGwZ-V~od|`x>H@c&3s}!y>NXr`hh?S?VXeO3(#WF#SyO=2v0|(u=sZPs zHG)G_)RZX9{V2q6DKbaBlvD+HN$w;wujA^;%Q{mh`)tS$v&=@;xuYmzc9vfh^9pDV zE2CxZW`FMe(w$qtNcVXNr$NQ%kGSQyQ&CE4m?p83LT>J=u3~cyZbhnT5?C-#!ItbG@61aJW5HV zKR=gGP=j9?8_(I#w&{L9hf1bxb%>(AU;gh=_<{P9P^D)L>nfPGcT|@^zAnhw1fN0PiV$nGI yx)JZ6iR&Rk#in-~W1~9bPz;g4?W8+p+J4fhzx@|(m&Z=I|0ePOSBP=U^#1|UWbAYR literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-3-EINFRASTRU.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-3-EINFRASTRU.jpg new file mode 100644 index 0000000000000000000000000000000000000000..c5066b45f49a83c79afd51066d03535b12057b7c GIT binary patch literal 2512 zcmbW!c{CL49tZGe%rItbktsBctwm!u#YjeC%y`ojWjBfYDWbMO5<=llHg{LXWJpYs#E61)TJtcheIfIt9% ztQ{bD4fX<25fQkEuqYf3M<7Iz;<6IrVq)U*GV7&e6%{Z_iVA47vby$GWmQczGHqo!fa|s>+X;H8bq(LF7Ku8(_ zm4*n~0cLGZVaV?S|0swM6ecVJ7eye&)*5Qp0U-z!Dg=WH3&UV*-BD}r046OgqoTJ* zWc|UTa8;VDer)=6Q8n}G$0*x=uDXFoP#gleVWXV9!sabow`pw0?bvB(gg4%6VM(+i zk*)0x9kzFHbaM7QcHGPRgpY49Fv#@d@V=lg=|PUb>udB{M5K=SJbpqT-U$ zvhut4YHII4sH<;iZEJth!RqXK`eI=4<3Ixc^J`PuRa)BS0JqSz8`d8W6xYJ_Ebga)3}9ihAH< zQhG*5Gpn~<(FS<;Q=1DmY_E$#Q zy(4FZnaby(e@?a#e(Z3)GaN1N^7&$v%)Gr?g4T0>2;}{^{5kVWUu6?W9MLLDvKtpR zto7s84-(T%%^W`Gyp)sQMi)QcWD*npN+!k{@ED-adP$KOllaz4%3Iw*W<9^G5SzXz zRNB!ZDev_8<7Zpf$Xotu>tYP7C#$E$)?=(`v5B_Wp`!r+n~=Qk;RR2XEJ=cM#^)y3 zT`2;oQGFFAW;K@FS<@i+>ExY|qKd9NKO|1xFj-l`cRGS^BBCQaPqNZ<`H9y)5+xdEG z%phY%wP~zbY&^bW3?pqViNY;w^;H*0x<4`g`8KJR{YBgL)98I87Ls8Mb{!n=Q5bd^ z^r*>M^iPQ|^+H2mb1M?NNmG(jgI9RZ98E0B;BgYbrhL{ReCKtoj42y*^|jzP;fx|` zOX2R!c7GI3!teOQ2Pf<9Jv0e>$Y8B5R&M#yYB%dJ&wkrsd-^I&i9eCS^2`KzXL9ne z_+wLXHySJkHt5ZF5VqgRaZT+7vh{H+qx)iS6w);_tgPp>zCty?)X-vn{3qgoscwXQ z&@EKmpo__p7()F>b*p2(W#xbwSGZ@~sVX*3>AKa}4z`oi?2?>rYqnzNZOayS=jpPf zOE0z!>ALN^nvY<+j7VuL)uFYH_|g%wX5gl09`;VoxVlMRbTc-0usHsWzq;;z8pl`T zshgF0ns&sY`KPh5*M9ot)u-Hy?|4Eo(dRKc8g82<*b*UI=PRSXCZrGx{O9tf8$F&m z4fx*ciQ=9dJCzw;#IsE-zor#d6t4C=vXB7f&iR!AE`8SWs~J#p%Z6}l3pff$jzP28Dtm@U1~lguxr zj_mQ#Xl$rucIPa}$)tb4Ug2TdcjXcxMIU%EULX2b%U}eo(V_>QkQ>D12&p`aV3y%3 zHX6V3_DO;>%Wh-o$Y?LQrE(?A^Tn8Tj~?JVeSdc zpLD9LE;ilNIgn7+Tf~Cr1Vx(gh=CghV_t^{fWqAt36noMsIR>1vtm-cKQgAbDUxO~ zpOBSqMg{KnWv}>6p!it(@B3ONZA<}ApR1gZSkukNMJP{iCzpy$>A~zRM~Bey1ikmy z7Gj;3qs4h9ehGJNDKU6LN4e!Y9qCTe1s5XWbsdFQqCkhMWjA`|CU1b4b@q6=cIr}! z!j*1uhY!{C&$L{!9q5b|tMSwATB_2^+m~9;p}B1Kkg^zl`KKw;XUJt|*XI5!l&Wq% z>iuC(o*qr-40-hF*d&H4TZO%4j^vnPI9kOkl?c!6n~$BC)Ec18x;Ag$MI7&E+`+@p z(q@Y8Ni_krob^=aqg1=*f`H3s-=ccojJh@X`P0-sQN{!2+!QmJU;IDjj7&f6{X&?3 z;n*Pnee!+HuI@<}=slvAl+Bqb1qX!Dkl%cQHEA=u?9D}@<7!l1Y8So_6>AFD zta5&c=Ax#n{~SqO?JD*m=?^=gMV1Z@=C@XmOHXUJ2Kx)={(MFX^cIK21R9W#h+AWe zEn%Jy!7U7GI0H7Rh@8~<1LK(q*qF_eV zgTL*-O3q+(&R1nb|1nn_5O#IdS7(JBS_}l>gq!kGc1uU&Y}0Y>5DVkMtyq>202@b7 z0WdGO;bGhQJE?CU(@k6V)Am0K@~iL-27Tq!IsvHs#K(1oZesHfUF7{-##goylI;Vp zvEGfMmDnkm*~68lI-LC!c8I^~Af0_85tGw@1n-V@t~Wm)P}*5i+IGRuu8k@-TpLA} z59w;E%i6Hu;K9eUP;^Q%RV`2b{8TK8;&}8Bp#Wtcpg1Yrao>sxN5HoI1(Xoqlr^N9o^Jja;?CI^*QY&HA`6bpxN(4gjS3x z^_yaR2;7Utqngm(`_-v;-TG|D-bmI?;FhnDY15>>njAe?01ZheWK_DWBE+W#TtB7=v9%>Mjrhp7ew4-ifpTA$jpX4bC}toOk~>?iueN*UgxWwKl8Fw$`5OTNC0J;yA#hudSyIfIt8M zBK-iwAwUzLrlh2zq@bpvqN1UpK1|0(Pe)5j$IZgZ$i~kjD8SFd$0u}DModUpN`#M3 zTRITYykvGd{Th^^#H#-phINj6qHodG>2(P4UkO0ArKkaA#ySb3UYE%>oC&a z0di&vmLswnl&r>Asf4}Rqi)}cj){$nPkxw^nwI|P zF)SxHFaH^$pz!&t*X77J6_r)hP0cN>?@;eQeC+P&?du=-@^ui6nV6iKo|&DS$9`K| z-`L#3ZSQ=S3j~n;EEeha&w~9yE@qNkhsepv$f>@|1v(T+A~I%jiX*a=EE>jCSG`$< z_QjIK}%w%CrTjjxX=@rUH?8s4{OY=mzhj!Lsr(hUE1#{P1#3fpE~4*83x;o zp6c?zJ96GRn&!1+=^*!J;R6cu^WHm)tl%EbmBzfxyj6$xGaW=Ruy8Tf*Bd}FrDU(7<7b@(yu;=?@zqPp*Ukt5OzB=oZ6*UBF!g)Nq zJc{>#E&Nnl)2A@RtG%n8>u10{={Tq-)>{SyAK&RDC-7hyY8=DH6RsJtjn2zKqNjCV zEIOqq7#D;(vf^Z)5(HBF=6I${u$-8mUe8(GX-C{%H~-Q>mcrS+@lMB#0ADeuN>boI zjA}_EaIBgqc~HeEy5D{BsEyf$EDS?=^D*}e4|HW zBgR`gRVB15WDpCJI=8klMwi#aWy9Td)oBj%kQr>>&0p5OWc>MpW#;qlD%b@L5z@bFg4U8zXE+^k@D-zRTf znGBIHzi$|*=Nx#Q8A${ha+16Zw$n$W9tZLk;h;JL7+qwTT>l2s6gie^Xk9E{FMoFM zMx5KGtxVJ4N4iNA`BjiQg5OZ-ZwBM1yqiDd>g#&bWwUJTv#j*TmK+_vo7UWFR8wY1 zds=6esa^|}(j>V={GY;xiyjqk7-pQZRG$T$2n5qRK71O^-l$ZiXz ziKJ~#qrCw81)HFl-4Y}1p9ma5`E17al-_@0#}lGS(7tPg6C^~qF;u7?Ap#;=2s|SZ zIIGwbNd$rr1kvqO#yA0l7MsOb##=46i=cmkY{nA7leXZEuh0^|)OOqiL92%d*rx9c z2wWvBm~=e=-^p|#arE}7#vi${&SWC#BBzHyMexBO=B%K;GdvGL6U4sXFnaq; z<~=RAj8x-h#aT|L7;A-cxBQVC_9`^vHjgxT9J&K4RvBK6!6$IQSHXM0zy;d9X*e!@ z+Eo2$a0ba`!vp{1vj0;n>VQpttfC*Q=*KGhH@;v0VXH_)?s?iZfG^odKasX5YL)Qa5zmuiW6=ftkKJ+fA1>w!j)DPJe|0pQ#(NV1;W;8u z+P3GGSL%fl{3Zk6S#0R9Tw~m_%Dfkcms-;o -w{OM%3Z{zIe-~S~?P!QIP@CIuV zf-Gtm;>BMyoQaGV&pPz~NQ_*#J?ZXd^!rk=%9!Tvk(6F5hRTl<>J2jAs0wy0_z^nj zA;x$D*IjgGcA&~9T*IY~*KO}dG8&zxhK-(4Dwozw+)4N{5Gr3hJokFuiX!FqsC5f1 zXzfrS<4xIzD6SQojH%7k38(<%YAqSMaD7f{nLOR%xcf_S+mAj&G+cm^SgChA5J-^4 zxuBnvxEW)?_d+%O+%wL`A9rLf7F;ak4@;rjm^bu=)A-ftw2XqVr|vS86j-^O@Db@* zMva+GzT{ri9=?_MtP0MyU|t>5T4xUBa;YxO&%dm{YWa0Al6A`cP!JyGH(K8r1lyHOw@*;?HOF zy7{B1#5~4o0|(k+*8Ufe#kpii+dG|ijKZTR#7}-FEgKnMLW@v>(1D~nGd#rvBQ->> zn>S8Xti*P~(N}G)sWecaMs7?TFBdZK0lmviF9fj`ndU#`|a;O0pu ziGY9M?0a}qhXN^59{()#$G!*SwyXUIBB1J<_A3gc6Lf=IHddr}HYa{}tx`NE9{Bg^ zu~Z?QkyH$Iyk$}ro03Ok82eo!K;s!h1i)=XAXEuwe6VT(9w3RZh7Fs|glwHEJanrQ zv|j9mGw=g%6YO`<4NXb=%cWO|zzk-$(-&2R)Vf9lnyj;lK*u*CFh);cgC3}b`I2I9 zFz0Agrds=RNlpUG2qOWSPQZDs=I$>qJc4fHNQW`(vd0x30JuVDegZFAEr=%qC-zpy z&^iYyzp|=4F+7-421QGElGxdXu0#`o!&G+_&Yk6K7Tfz?QLiEi48Oag(IGJ1$7XCx zaUk~k&)C_`6pgmYN1+3x@W0d0`Z)}NkMsgBM3DmLV>vGx_sBO@JxLql=fppGD?*2W zRL^;GA!*MQT6?^tadS{Sr$I;UK)e4KkCu2=NCbqQbnJ{1fgPG6e-l$8aNlrd?e6^f8d!nPBEdrP^mDptcEdw5GNo4Xao@Lu7s zx4E}9^V9*M_{Cf-F0Z_@=V-TRe~*4mFH^IZ_@krQbtd=qb}mw7;sQ|uvfb*!mN?CK z_rHAF!g5P5OH-a&ZE@-iy{AHkGljh=!FpZnR|GHGN;DhxEIy6Z8s%O#^%I(mdazD* zy4{BZW;#lNTgYnmE*4}q3=>SbQ7!tcfuXi?^fyZ&`D!wPFX}S`fZ{hgokeR-1h_UlEE=sm`cgBBt(G8h2Jza|#j> zMK)Gj@4BS7TSur9w9lSyL_JeSPE{wxjQ57>6gyj)O>O2PE|Wnq{Q4N%Zd+A-{U)ve zE}@V#)g?C20+b%FgAwE$vU96WIcYV=s&8Q!q5abx(C*^bnVvDY>+-@|yex(Yu)BO1 zZ@;i$jD+hBEL*kvRhF15mY*HsbYQrkMtHjM0`k|0_j6bHH=0@!-cCZ=vS1Yl03$8? zTRSw1kP*_&V+P-!Zn}TLhRT0A{=kMAJxJ)~x^L$JWc~os`OgFTRWCRpjAyas%sRz` zuRkz^7UuXQ9}A8MNpgVAIcG5lHokoi(2lI-KwJ)6bg)@(=Btty&pI~S~66-oQdDL|fOzV+Ockwb(8 zlpbc8J#~}-kanS z0W0{*VGkAXMj>JKSq0gUb58so-}<5b_a9Xae9-sR(1ll|#&eQSuqgap1fk}{MyYD^ zUOysN7)qq>Ff@Kr72xKtuzA`XhP-;P7-4i50#-QWM zxxW{)7)&npBXu_n-uGOrGPN9*h+_bTU0-Z2mkaVCuPHOz|mlhTUluW!!~AVx$Wyj>uO z2xPL7@aoql8=+8_@FQ(tdF$Ok(!PdEzqQ)6S5pM-)LtMEcs{Ik z;8)>o((H#QU=qnY$`sSfp4S1d%FrjzX~R3f!a*R z@V#DBd#I77w=Svd7!brGWcg_Fp_O8bGeNOOs{F3COLpCw$URAqhzGD5(lr1fi6o#` z!z$mv*NcXi;?p*DYywoCOQsTm+eAQzq$s5^9E0WAbwD)4-}VnnMw&TColXpO*(mQTYfd`@uf-7XOTwF|DoNgaQw zgrNuhO)Vt%YlO6jepgQ?mD#^~|K8(3ci`_j@5^b?+&^wVIIh+jF<-6m=1uh4r#1F^ zXKF6?wuIY=n5WlD2ama8okagaIiXHJOOGOn_4n%eA2rUwjH0Ts-&Se~m~p<$rFx`p zq(w>nlIQ0yajJ!a{naNQNu76Vx)k-w4e{9cc2!zT*?;Y^_O1c^f8*_!tBvBN6pIiQ z$iyjI)*!hSQwJe`ymJ=)^S` z>^+bj3?adY41%H>w#8l&fiK#Od$5|iaQYh)8fhmOM}`LHb(~Tan>@96qZjW9d}RX} zphQ||(FQe1bi>;!`IT$ACM*pws?qZ&RUh;WAD;FMmmp;bHx(qca-c6{tK4-i(Ggp6 zN_eykY70p!@q}DxMsZa1RK)uyoZ9(O_BS^0b=g!o+f|->#v+^781%Bkr)GO4Hcq zlc;h5p9|rL(b6w-6$GYwYm2@Q;dVvlrk)zM`c0?~!kQHAiZvS;UXb-A-|-H5u+74r z1Xa(vFL$Ps>cwpJ@)L}-vOoQ?C7BZLc$$k$H2lHwT?JB$B?`OVUXE_`_O>pePNwCxIUe7dyqH2@pKWM&Fz zawo5WeZAz+LQ7!qy9Ca!*WBX#%G?j%w$tCv&_gv$*u|m4oFPd~JDz2e+sD39q@bQI z9rcl)gRQN^)>FU6GDH>Ok_9_98p;;WG&fw;B^@G+*q-$f{di!Ou5s!TL=u=5LX35dCzHQ&t?;xN!vCz;i$P!eEE`K|e#MSx05 zI^Fy54uKX~b}|<(zP**twF~3$7!MCppulQVn>=ioKcfEFx#v`9fu^ib@O2_!Z*rD@ z6M-dbB5=Z_<*l)FwBfNi=d*8!@n04?;yb|SU-D*P{Z1G7O=7OQLVKpuVC~zfrMahj zkez|JP~|sR#R&)u(`>jpHT9`F?*nA8=)(*dsx2bVeR5(12vH{wk2gyO6}I#X@67fSE$9%_JI3Nf`kYm4I5@DFgbe(8H( zV}THYsp()QCCD-@ZAnPRBjATo{cxK6<%hs+hsm&frE5Rfx$?x^K9pK5Vx7 zxNTnQV79A>5pg&N@Fh(0aO7E!(D8#V%Daqu6Mw$kSo#8hKRI`b^q- zVQpZx-gCTYeUx3Y@&@l25kJo(wpm6z9K-by_jNj96YCyo>tkQr%=)ASEP9b=EImt? zxu%9+Gqq&a%0NvvvApsy`j%x}zue+nH^3if(8){jII0k+w2PNYGG+okBw z$_%VQqEC0lKY$}%@V{K0iSXL_>T>!`0v81_iT^sN?nzmz1|LVAC-nF5#jeMcQ*RgiK<9)pz4bJqJ_;q%r zWybVIyam#@G;91IpO*t2xYak58jTp{Sa6pnEZ{~JF#J42k!o4AbCZVz8@%R&SmHWp zvG;GXdapy0nysAT%UgRkoUPm(QymY#Vmp(5T3OJlpA0x`B@PrrYUT5rF@?4I_h0rM ztJsxno95&C#&Bwe-_+Nd?IM*t#tY;uAXOE)V*KWTmk_T*bwz4UszuMud6kU~1(;4V z5ul%~Hr#yTfryE@c&Q@c%wqxXd!e5$gk)w&B61>B2`8e|=@LhE1wPM1EcJ5?14XDc zR2R9ojT)?G$C`M~%JDxWm-ffSqlsm)GqFpkFAD%Sv2+gxwp@E z%wrIGEMjH@_egvGrkUDVzZKVf6m(?e*HgI=P5$~ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-31-Financial Returns by Period.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-31-Financial Returns by Period.jpg new file mode 100644 index 0000000000000000000000000000000000000000..79eff1cda895c535c3ea0b0e7031ad63c8c9fc23 GIT binary patch literal 25606 zcmeHv2Ut{1n(jf61{K;sj!F3|VT_Gd6a{2NV>gzYaWb`zQ4D>W~bT?Ue*lsd$G1Jkp z3*F(mdrv?>fRRl^LYPmShhKp2w};>n6BA##e1(dHgo^JL-7UU<_=9Z#t`S{uyda2= z#{^ushKGL*58DDT;@B@8IawFz|jeNfki8ZH@HbnN1i)a?QvKPQ3I{Gg5F3 zMve+t;H8E(O$>Q(&(~Ty?AMgYsIk21VC1_l&!HksuKOIx+X8zpbYF9Zy|#FCUwiqQ zLVt=zNRrE~=YHODy${;WSNSA*gZ7};HZB-`z)-vq*E_tcClglnGViN_Z;Rh9x+Y9y zzo7a7@^#>XX^j%ozMe$WYUw(zz(@%o$)J9_0WhV&OX773F65cvN`R*rk&1fmnR_dn($;%}HXVI$iFgCas z)1w7&5W^J)YIp&oQrxSs4Wh3uBjH~L`WMCBf`B7Y&MadML!%8mGa2ro>QOf0EHEDqCKj|P}o~6bDi+yrf zpdbDlD8VTjdu^Xr9kMozl?J;C#Ky($(muZOo9B^*h9$ddjL+ zRfooK$`gE(>eH`WiUjDDE&6NZ5tB_b+{19SQHR5dDy?@9TDFtfS$zC&ItxtW&xiY= z1wG2EUWZ=SXS>5yF!vFMhy`};+{6MLsf+Xq z*8a2CY7}kX=9SId8ZwZ(A4V~U1t{kx!_h>VW#$Ig1{|%gj#-q)yzmZmG|H3+^CJUvQ3x^@hh?MWXEr<(VVYG$v|G8Nsd0AbhLFO zA8B*j@kn5_sg|9W$tmOW&_QfY?5}8gsa`rWO?E9kL7ZPtv*PoSWEpSH0wg`;|=V}bdbZ)W8! zX~xPYz}wwh5O;CM@r2XYB_77$fO*u;+Xm3xwUTZ>K*x|g-4+9>FJ zt^0%T1EnO8A0Bc zMk?y%fVTq2P>wpx(xPIYA~L?sCu;L#nDK1!eqoSPyxvM=1#~j+fW4xp(bPzK*ar`a zwn@2?!HLoN)EL^wlQ+sn!~%w|B}5J7>%hIhDE=1FuT{}2aqEu!R?#V6(n+T~fknx) z%}fLq2ylR1UYE~UI}Efm&(nRw5@S3Z>B{EQxt~I!z@FxP=|Anro7UAW8Hl-EwRi8% zEhio~=*b4h^HAYRrSV($WAX;mDw9#l4&jQ)r-zWXvi+@x^iMmj@{e4f$kDzBgUF+~ z)M>siXR?faiW-kugRvvSTfjFU;hK$JB;QNi_dh5~w8EvkwiiDFK$z=hh(l-Qb(bAvYlw@_)+|em7134192|_ykeZgZUs|r zGBCo?x2(S!2-t5)co^L4Y=fm7hU-B(kJwHUSnO1$!l&Sy(G0tpgn0 zn@uw<;ykDOAXV?@wZlZ`@Oyx_1KP1(%mZAnE8g#D75eI$fW__f(n=I;eH#+I?L<$4 zykZO1MW?I#zHMlf7u{;Ext{AUi4s>O^e0#7(;hsamFM~C$SIclCh)ov!N%HWY7rxw zyCPZ8d6xeA;`ixirJ0`Z>SK7khLbd~!1i`8|CSVugF3CJg<5z#PTs?|0118t6{ygxnAqBv5vM)@yK6E?hHSoAR;I1rV=}84XHYaqo$EWMF#5wS`^?Yl z=Yb(Cu)49iXNbr@<5UWlFI1G%pS~J-c~Ly!5XQ^dojac^GOVLBc|tvy$zEODQ)4j@)T+P4#T45!rM|cxH<86*+vAJzN@>s=0!2u zR&wNAmGm{`!%B^fM@@(cAi5Y5X|YDqC(mqvO=Kx|E~^TCM@-;+xQoY`6GAd&=aztQ zO#w@kmjyRq0lHFpq2^03i(?fcwJp_yIfMffT_9T=5tyUY4D@XT+j|e=IZs6mB!SEZ z#ywj@H}i8qFhJ**fZzExZh!V7ntz~%($?Ocpj>Dov}-G7d#hP;YRSi^L6~~Ob2DjJ z5`T?S=!joyBQ7J}!eKnAcS4of+K6cVr+pBxgyA<^pC3Vx#{|%JaV^`L(S@1tG^TU# zDJv^OKDO$jq;6Ir4{I=-@{DbT9dbe}SBv(oxf}@Gp~O^+g0tp~mjyUap`?>}J zN@*pLU$oWna>LJeR(L)HAxA5PH6n8oY|UQRlAF~8j@&#}RPO3Ub2YXR&EcJ>g5IG5 zlvJ)@`cBLS4+ty6b?fC}x7A7*BkG;QU;0UIn2&y&gI({Vh~V)->`GkU?J|iJ8*`Lf zGLxcSjfiRxIXQvnnPCCqeNJA`jEjy%{N^;@f}x1UJlp}iNv%$GuroV6Y88#pRUSIv z}EJHw-*u> zZk{YX^cH>R^x^c{3DMp4BN}(=t&w03MqVe8BjD63=cMml8q##@TE5nFr>iC9ICBhV z{OhrfO){;Uiwvdf2E-`CbQH(#s_OfN4J@EmsUPm>)KK`;PO4$YClY6$zz`I#fSLyl z@{xE@Doe#{mVJhY2L*yyU}+p@KnqtfT^1v+P~ds^NC6M?u%IxPmB}9K(86afrCz7~ z-l@Yg3;DXMZ!G7e5#f`ijD`8L6D`$4Kc+*B#cacHF>3%)#oqx75O`2j440UEapYE* znGxG;iw>46P>&CpyPzygPvT%ei&C2l?ju4u9V)(zRU)HgG!I&iSpJyxk`J7{c2`Ci z_NWSRm<*jyPcdjc45sk09M{`@4&ZgHxqz5!UAK}O^a&?+3FEa!w%>9u?Y%6DG-3vc z`~i@pYKKUFAbqUnlVBfak9uB5>f7Fk1?X#ZGsaio zvVO{DGdEk_;k~r;beS4E6x^Vg!5^~pfZ8vTRaBjNAdb0-U+wyCsn1#V+N9S?GdB0I zdRXMRq%K#+tUfKfUp!gO8(o-_-GDD~)p`33ep4x}KCcEsqc|JoKEHVTOO-rfv3nMn zj}NT=taCNQ?rqQZ`}Y|K+V$KY@9RO^Y!>7*WSh{y2CoN;-80rKr$PvXV+Q_$6s@TSO{a@t$$Q z6$X_ImrW@;&^tkHA!(E!Nm=@w3}U-lB0X64z4JXtSNpcZDuF56N6FGo7e>WlaoO|@ zc>dfQw4%ME%_cW`!nXx^Pt}HOmsD5e=n!U(m~B$xe~2X0Ihe`M-Z*)8NrRa8+Z1yL z^M*3LSP&DJsf}KXX|oRZy*W&MnzX(Fx3gYs@?mcHMH7 z{?a4wGhPMt>V9hJC#RJ%8aw?lMosS(4e{p{=(({V8E##G0#NSHD+p;7gDnDf$HVC| zZ0r+E*e95X7Ex;qN@aw^0#D1y8=cXNWqV480k1HSL;Mp?h+w zI(MGzP*Jl!0Erui9hlCqwsa>N$n7h+%hVLsmVcMWjNH~|4dL-SeU&knQJ!3NdDmMi{br` zZ?`SqQ$2nuWnHR@zk-(r@@%7d=i(C;ec7xHT?Q8(PUzly@lEx02 z>ooLPLs=a&qCKF%(QK*mRQ!-#8AbF|p7r^X=Zr1-QKrJ82X`h~1qC&Ap06e$Y&-19 zaJ@Cm@|nwa3D80}EQZX59BgXVn?mlonCiLG-Z{iD7Kstcov~l2%SH~hxUqbc&2hQl zE|d^)z0YRNq>bO(cr^>XIVxCxl?2pOyy>i=d7bLTb-@xM{UbV$`Y%g|)PhS%r2cj6 zXw7=0$(p8HUCOeh* zR)5v3P-mkG?o+@mAeNyT%k)wg^wCw7RQ4v_>)ObP1yK4Ac zYF;4g#B2_W*1WFWASNH%j`6IdW~e>Ac(PF2)jP^b%g%k1_JyxtASf-X;p`%yBu%D@ z_y6z!ee#e(DNlSG&0XNee_mTOyv?m|S{AUPTOUg+8e-J|%GwDkMx4$Y_4DS7%jTIo zyM+~UYY30p`j}q=RPF42qBp~8G&9wt7?iR-Xom}UVPVOOJqAN6b#_A&m1fSSS<^Kp z#$mF`l?98}#=e%0Vy8kG5ewAsXvlA7MCoXciwDu;2RCa#V=Ou#B-!?L|n`?v-8{32OTqdz0V`x0jdoWXY~~bb^Q|kHz-^i^06R~?0Igh zygsx+sF{@`n?DQ-_6pjS~3uJ;$Ah~K7=_6+N zp#T<$_H5Vx-6iQ53e#Z78cRJ2%!mbU=!KGb>V)s|R)UH$(Yf2AJ1FfWE-a9O!_(%( zp=c#WM6tLw7I?sllamV;K*cS+N#7WTMtF8eghG}LZQ;<%c8mZD4_lFPy_=6bGwp;{iKv}$PjNbwEPvb)QyD+(5-~q)44tc@$*OMa5v2}2kSGuTn>~Q$EP^o z@qMOz8gr7Y{P33u=m*}?cf&sqTrB5C9J-6GZDLN#$ZztAZm(X+tWPb7sR-Udee_T- ztYzfRsDJx5WB>HAwoB&`yFT>N*5-2G)whu>at5-`rI(87@U#u+kn{mT0`G=Pb{8m| zmA)GZbF?Z;u~BnIY#BU2HrdSZBSvDXZJj~>Ec$Q~LK{Qd=>$_2lG;EZLBq7kkQ8OU z^K8(Oo4M6)+#D(1wyGjaUGd=7*xl(MV`dQCPm1w@p25TfIi=9u=IYEZH}x=%cW-&$ z^abAxLaNZBIOkWQ3iQpr1mLAY6~o<~R^R#WNs0Q81^9df#PNb^PbC$zI_EDV#iQhL zNRIA}J4-y0MiPX!TM7Lj*ma;&_(F>B;o$lg%k9)Ju{%UdS7hPjOHevT+S6PtaK+_e z@A_8SsMR}X(g0c?2WikGmpnsLW&MX)Fqaqgn-k|iG^eY(rMsm=L}jG@WFx_9W>zCT zkpSMl1o<4cdbu<4*dDqg6YzX6UPB)ED(#qVH*D#PqcD@$QPF*_1_h>J<2U3U4XtDQ5ZwO z&>^FSJB()pqhlU$deQ!nT=@+6glAikV0>}ZX>a)g!xzc)xzjIs0q7?JD@qUhkM>@Q zc{8eB!U*ytF78~~ptxUtd4YsSs3KJF5$n>6;X;rZb+2n1=eIT)utK*X$s3Nw*vCVU z@IOlDn0Q4WvOQ??kVSr~qZxi{G>qmv%bu?sj?Yc>)er5Zd_ntwjzL94c-`R6e6EYf zqXLuy@8z|ld*=5wVS;BHvZdnnVs<8BlFt_t9NPsC0)Y^kz5xM6Ua*<%`h**wZn{|3 z>nXb*T6+h#y~V3rX34~bNQ+!ET67 zfzM>Z@UrEkSHCeWG#LvRRIRavtq`sHH@7mq*PthoAd!A6sB`}?qO@gUbm^1E4VPkN z24*D^g711o6Z!yqCc#|7s<@lRR9J+}px8xef?@`5(t{h{gmu5@oTQlLx2x%yBwu&< z8eLPZJ8q;gVXDSJskYM%WcU~-RwPb3BDfEs1*`e5ZxfFC&%zl6%*Nri! z&|3yx7MN8?>`KT6`pR5ZS`f-UWH1p1V<8n|JOz%KAU%i^) zPQPP5->i}@iME{UO8|WLeP?IGDl^?lP=J2F$D>SQlCrNnc`Man((>PXVHK7VVEmB1 zeuw-X{HCKSdvnV2IN!2Nx3APM^*;3JdCQ6~mYbi@l&+tWKYgk!pDXA&Qf`<$XEjlJ zlF)5YZZTKsAlq*hsbzGBB`n@pRkWyM^UQoCHjq{h1z!$mp5MzSslb{es?-vYMkj#Slq^Z$%iAX0wof^mG-25)@faCYm<2v_BHXVqxi6T z%L`+&OjB2TN!_|6r_Fw4wGErYvN9T>ko~){2&L0v)}Dq>6DyY(yDQdp?A9ix;*zpf)l>@gbj39gl-AQls(5X8 z#r6;K?nR{s+C(O0TR;tYs?M6?pO$4XWWzFYFSZAu?`05=KFKR)x)^Hb%3Qu+i-K#< z_uMJbCmd~@5|7B{Y0i*mS==j7e1gaf-nV#NK6Kzt^z!wLOwTyv>AS;%A=v;mKcz4U zjRmpb2IXUBDgx86<}$J8i~7PthYltQ)}E7=M{ZsZ6Z1_B^20A-fsUS67$k05QptGq z7sah8Oy0SAVEE-8Dq>?_P1?)b`9IW~{xwf+TYnY0IM{x6h1AU_|u{>%+U`C$Qu7|2ffF8S(8{kHwF*PC8gL+_;^c36OL z)9)drGI35Uz$_Shic1k62!`u2kYj;}xb8DtewGrq)&#A2G~)G2c0o3GK$`pnZ|eUh zUBhLy6-KgY=h7{O)JRN+TIp!r?)(W%$zRbGbM3VsqqC$g(46o4P{YWRnreo*i5Su~ z@3iF|vtCk(NX%e;upXK3PvGY(rkUxo-Q7O0N#m72f4Rf7y!`3y!IfJSPwn?iL#G(t zgh8k@p~Pcj8>6B%8Jc0P7rtL_EwtsXc?zAzp z@)LUj-TgCmI~!m4b&C7>x{nql&%tO=wDYwYU*$j>NTDRYGeVCcH@a^y}PS$^MCEH2QwElICJYc9tHFT(PZ` zP5h0Fqt_+{3lPy0E;jVqiNRA`G7JODL_-r-18xmj|7^iC{r;H%%1;;B`d-G(C=AU| z8ZZwgL{&dSMdPL@+Jomyi{hxBsOn4>zFk(Z_q73xhohpT7;67oto4t?L+oim!1;so zTx6$pP&99NOr2Lh134is2w^*Ikp$&JaQ>+O>kkXhdI^c$1?P-no(;W$@!?>aZd~ds6`XvEDuFgF<7Qk_r2Ed$I+Cd$ zFoqqHxw^+HN@+FS7)tJdvkN@XorRHL-e3)Q`2{L*NRFNe!Ld`&_xiKFi1x*2T;wj& zU1|fDwA_j`-|XvfvNhM;DnZ_l6ud0F8u%D?<>S;%Z6=48d3kukA`GSZdKq}qw{AUP z@diCLr&a8>IMv$D+h(dL|M-g4zd0SODog0i_NVaSAIxHYXG+P>DM6mwx%?L{bK`7* zA$N}ha$UjzXGHr+=l)dV|KVT!E=jI+IORk^QonywL&MS{Dz^_>$Xm;dOT=uy5Z z&yvsVAiWhQNTj$@Xp_3bo!#Ab3mA&|>`x8T>6 z!J(1hmL3Z^2zNoz^^KV3!cQ-5<4>VOQE^+FfvBMTOx<-&KGx zfzgE3(vA+KU&&v5lzqrKaSi9Ou6t-M+^KAoMNZA$U-0CZNCB~$FrOZ8#*pgcG9soeL?uk$CXRQ z5+s0|6>J_01nIQp=jfLQaXTmyQN;_V97pt<__hOK_afa_Lo|{blc~*~YD;g$)vWDnC$HNMAJ?#w~VWqabLijQ{AMD1w|=W*5C%bGNcv+z62 ztyxjGa=$j@6PhU^KBvh#H+#(`_aO?#=5sLpO0a?fpQ*0|6j)|WE#YSKpik8-naqpD zAR{vk1#Ldb6MZZCtxB&uM{)OOVCSU=;xHi6ZNH6Eb3P+#PU5l}WW0{0Y_vw&<*U5i z-g}&02Kp8#gQ&%$+I4^fo%2b6(=Tw_-k#x+2)bUtRCCVsqBNOibHa7N0dg5XBT?!C-2d+4&>mV*H?F4hW# zY~pBy9GeZ$+GhvSj}~Op496xn+#GxS2d&F11Jj~C)!!4LL+vSW*(w)nnTsRr>Q?!y zJ2%WD$(j}~X{OQ{GenQ68M--Bc`+m2`!7^bNSHqH{h7ifWb^2@C)HKiA;E|_>i`;F zKQ8Vc8cu^7+0@$BzVyBFTaix4cJlE7DLRMbOEn|{Fs>xJ5>qFxmYAiaS{!Uoie{yV zy8kJT{`;5`M-K~7Mm=oAP&$OWzAYWG)1*^8S&Ip<6w_^pK~iw<%G5i`N)VbkBnQ;p z=r3h2rRRTKURE6}#`kum^ydff$b)F3KnRknDG%CDpt{PyVEZ2EjrJ2yy|%} zVxL}&p@{M#lsXXZRhEoa)U2qEmqs*MfJMh7pZw&H#5cOJQ_T3~>u`M$w_`plnvJ=1 zHV&VF>%n?>?PjlMZcZpU6^Eq(8vb1m<~ggvG&neunc@;}Kyy1u{M(*&0XWu@h?=Xt zP6C9|plqJZf9QJPQcC;fbqqDRTtxRp$bH*mb=_`WvT%yN+}qJ4V=<5*%<7E@-yS2mwFH;fZJ*~6;X>lvAw?B2v)r4pr90;;Z^kh`e zohfhk{n&kXJ+rq$1UJ;+IkTkoOR52ws!sG~bhhk8b|njHUi~kSmp?W(&$%g0F?~v3 z=#sI@$qE+eD*VI1eB8Vsn}JJ#jxuf3q{)Ck9h$UA5JrLvI%~be`TdvpBN;byUfH*N zoGx0AFU-N7R!6t^uZ+^UAY)YGt&1i`#)9`K9OG+|{zt4anlX?8hO>ujpkFHeQ|12~ zfAJbTq;Rtwr;w8+Xrk@G*e@l-apvg~jvgOCt`hp8i|T5_RhOGbdCw6ApW>R$FZ9k5 z-t}cSwE4VOrtOBOc+SXkcdP>t)xexw*Jg z?9pm1c<7BSat=GiXo-t9Imof`=2cgXLbgOq`CB8`PhvUh3CM&%eMwcutDH0765@aoWp_-V85jK3{?b^gw3W&K!huJl^7xo4t-- zG;FQr1}?Lwfx?9_>b-F3x#VN2AxvfcZb1#`bOiGJxg?95B@W1T`xP5J*CqcfxE&pi zFpRZKJ%%&t2mDer1jj+^R8H1xpQFZba$Ws+9>O3WaACv9;JkQ&H99x(oOjlD z0UzafWo3ED?N8ta@BXh5#eexwhO_VHFWK_>;q{AktU@sEJU>%YMMbRRuqn^=UjU+i zN@U80@hW5y(}$^KT-Eb*I$!XRG{YQ2Hd9-lE#l%+%_OPw214`P0CTG1#ZBFIv?U+( z;Aj<&bC8+rleAPM^U#~Lpwpp>UwZaDTsMB+o%HoUr&f2-1Br${3FNqON^0x>du;c- z%V78w$QyWvo5KpoL3f3>glpnZ_I>u-j>|D8N6PMQX;c<1w+^HE3T#Ch`#_r?NlnHxa?E#XQyVeTmZfS!=z*Lz6~eZJRED^B`vIMaCIsoXpoDox`v~zP z^emh1<-*}Snb7#ljs}qqf=JOr=3!M3OwHgyo;d@59qK_kDoOz!GwALqQLr8|}EzpVbt2h5ycOq8nKnNO ziMQN>3p2V-euq_xYesvpBin<-RggC{OX06wwpt@9gF4cDNLNBKg!`tgFeP|f4!;|& ze~MoI+qx;vRanqH4>9;ElJy7WbzPTu#Gl2c&mHA3?wmz!dG zC_*zD-sV2ea53^W3n6u%wGBId$3BVjk=j|p zf!6kZCs%sO;F2%MiEWFBx$mwb>$7bYGkgY%ylJDFa5q&$10Vo zij>75q0hGhpR;$|SIxZS%TFbhXL+e66IKuM)@wlXfoKS!A5X+(GpAkZmt|a&>bK$^fn5a zV<7b2>Se_+vz%4FcJJG$BkhQ=8QCr|a1H0$hZkWpMDu}8?KJ-C^xjk#iY;q|zG`nC z>1@>o$!1SybZ+B`CA!RoETS}bQ%H3F#;UJ>2v_^B7L4MC{o_I*Nmgyi5nIX=(a@qg z&@;W*!v|O(`Lyr!Sg-g;A=%RW({v^4?V>%k)2HzMlS@IzN2fRrMc>)#89M#XXW5vM zZs_VJj&5ss93TN+MT1kaV9C($U>tpL20GgX`O8_U-7Ie32>?NfdC~Xi7z^O9ad58A z{>04{LynSm0>tCFLZ@&ACioa^i8Yt z#B@OR{;<3b%6|Uzue>NdBG#XIgm*Pa-%!h&OV>}wpoGhnUg3aYWE?G7S7nKtcX~W* z^{Qf6z@n4eB*v6$w@n$95Y^drcXgub&geHC@qPWp^%n-b^Ukt{(&PaPP|CF=@X+X< zPOYxXbXi-fqIkf-b317C8^UGs6-h!=XDTXYx#V=SVyl$c%0%(`n=ce{M_xZoB89Gn z@rkVKRKZ$d#A`FRVjtF2Cm2d|v~(5nv%hX5baN)dn;MX|!R^0LInQ|6+&a_HpqqQj z^t!C&H7M7~z4hd`E8b!`oHv>!xduE8GikRdkXB3@lT07g7p-NUb>vq8|NVkyyp5g@d`RsCiF<_E|04q8r< zQ)>nsGt@$nF=(h#A{`r&Z08S0kUwi++$FEV>d;np~VTT~#w&wd^*a9x+YUs*L&3IT0o(~cMMt?yke z>f2WFBE2g|-$n#`KD-+aLKQb#WKU?Ls+&j%`f~K-Ej&41F-0sb6WuhqbJMwT^+FXm z+^Gz)A$dKHok{*x9rWG^V7eliS!<7BL|BK2dlWv%W(|!AHqF!1Q(ha7$l&urKEL2b z=hN@JQcJtg?t2hUsWJ~AG!AU_?^cB17~gm93`uycr-0=ng4%u%$`QZ=(0D^zf@08X z7DA&)J;NUNqHodoSBa)@H*bXw^tv%>=#H< zu*q@0MMYm!>wN+*hsUA$5ES_-=5DY+aYcKwiIprb#}CWKPo1wNePN>`1=I|pacjPQ z0c!pm;Wpxv?S{K|dCzPNS1i61#Gfm`{C+@Za@-QsC2#Y|Kc$+G*H=f#k82i_#`|!L!YiyC!^?N{>$>0v!&%Y-?pN%-4saET)>!f7 zW!?-|drn%NLGr>Ox%5sQGth;vu(Nr^iP()U2-y;d!J12l;tJHYFgpm@AD^Kr`|mcBt>)@ zIpvWBZDCi+#n0h-mz<(u>v0p6L)$S=lU;4p1Lj-{AabRCny84pPS5MT+>Kg;4=S!) zO_@7|it<>d4oY&YZ@!{wrvE`5arK4rBfkbL;J}!#+o)EY>D-&%vt0U9Qo8(Ee=bBF zE-+0YJgZuz^*DfdH--%X#1*-_6DGgnp6-74o^`4P3nY~;VOWFq)CYB2$|uvcGqv(> zLYgV4R*~&{hj})_(jSs{=+nz{e+IsDP0EDj5~?;VU5ijuYsB%akkAT9w7ed#9clSF zO9XoellY7M;_T1;`Q7kHdZV`rLWF}LH{wDc%QX_EJ`lNi#a{UwG39@p#NTuhd|^c- zxV7%$ZkA1rPO{t6?`4N3Hw~QaJTzUiJB=mdFB^1c#yh{<^rN;iEa1m+$<^JW`Mbh< z_MCXXgjOSaCHU86D5-zm(96aJLAnI5y89PzqMR-6$G;#ynC_a=ZF?T?glDH!P7Q)H z`oHc*91OLL=90x)XSlk?%x5TV%nIR$B|Def8$?K((e>S+n7x05el#IBf51(tJ1DMl zzdZE;d6!cdZ(~lIcW2&8oghOSZmG99%QYlN_HU> z$xuqQab1(h9azg$xtFi3=^we(PcD<>PeS`7Ty8M+d+3s5$c?H;F)RrQ$!)L!Xl0okJ{e9Xgg!-Hll@S1#0eYFn2ZJCe{KeNFC^u%|sePt; z4!FbrZwl6c;$1c|GPo$X;4-?$UU2A_62dk7bVTJ@P-aV05PJdNF8~z|tonD1!R{}{ z;5@ARKQ8g#bcs0d!}gc?o8Qn7$KUC)>y9^MiPI5B*?=6rFz`1tjl=>~yc&)JsE0UE z`uXsAy!(GF?%%Yy|87{0b8CGbcv#?;F6?9;H~n^LpW{?K_H4kdxFEa@Z=SOz^cpv z27%rjs-MS4%%0%~oSi|D>6oh!65Q$&VdyLz3(&Nm8HS_LXvkdDPw~EcGxal!iL~(3 z<~^7)&Iy8>q5GtaM+aD-EgU6<7Q(SfaXJGvM8ClTnmFbj1Oi6g`scZ?*S^8NnIYTG z;JNCHjUOQ7r5((q1+Eo$aa_zsEyF)b;ji{@8<^z4Tp`l={ySIo(7r#*UhQzw@c1|Y zgSmwB(0{wPEV9z=|0H}e&p+!Bj*$Dep|3T@;Nc)Mfo{Ev$N zNa-I}vp4wv`C6t24KFZ?9PMBkD5Ebf_4 >{n$|!Rf0Y&TQc1i8C7($>fhV$zCZO zR`P}O6|Lz0z0!B7|Fihp$6S+)YoRp|QK8=qoI&F673I7VVPNeTFE2*Kqkop0;E0-{ z%s{R$U)5FHSrVYc1+)LUJAbpnf9y{4brXd|xxVDrNwWxxJ_To)^ox3q+T52~#jI5Eo8>|}oGg)QWVbky8$fq#+kB1@+ z9m32A0lbj`z2tzJw@Iu><>GH+TKrmOZ<#zFckg!cI;gh`A6NES+v>lmE^MfE~1j%n+aCE7w+ zmj@?W2?#)34G|#+>)~zQqCuR#gOaz*Pm*93uY`OAC)qP>LhVIgFf-)0E9~Gps*;|_a5T>dR3#8N3 z67J$V8rxb?-HEv`rSdjJQvz|?RP4gvcv!`9ByZ2c%S^4N;8z*!0MP;3OHiSCdK-n> z_1@wS2<)B|(FQUdU%e699_E*C2c>v>zzah8GE$(dOng1{NxwJ+c=vWW$w{<{o@5;e zZbf&XA0Qks{6hk#ay*M=_sa*SDIB>g*qP}(-{AWoqf2m`Le|l&8PE>$u=#pEM82r4 zDSi6dPR8U;nBom@vALH;&8_oi7BaO6{p1h|>(Z~zh}p@mmq|C8DLUGq?Gm|g5Bs?_ zv}u~G>0xUse5sILo=sepAo+?otI(lU)IyJn-Y^c*jAvzH zi$5Z!WxteaJIo`}rx&;1)r}babVB)Fpy#%{hGubD7-57AKDjC%XkO>jpXs^h30i8> zvf?rpu1wknOdHSTK2KHB%eV&ta~@t^w`$AMK>^$jfdcR{jw>OvW+I01Z%!*WgSrta zfkypx;5BC&!ocj!OVp_`mmPj&-_}kH zIiAOzM}5osv3?$$7IS@9Cf3R#1#(HXmu>x+qatl5*Vlu9-hl&I-Y&FT*Td<#=0bg?lE2)OO6o8J1!Z z(JxLf8M;`14duVR7%8FKHGV0ZP{NNcX&-NSqAuTZ0s3L)Ryp!2lBDg+w=V8{uKmwT zCQtdOXHD&7WZ@HKXmF0f#2SWYXy<^$RJq@hOfoUsT=lyWffd!&%V7#1??QOes^}FP zCNRkhlUU#?N=&n=ZU`OXUf-j* zE%7&kDpx%;5Oq8#-zh@5?{>-dKG-Ica-2xR8(8e_QQ)gg)%vnyMutTEmXkCgP6yQK4 zj@x17+oayKp7gbsELBlIJ}yOdHXP*Z{Eh9_Dy#z~sYlT=ab$5fw1X4H)D)W!%Z!rQ zh1d9G+=8|(5|CbgxfU_F9bd<)_D`g@O46P?(56S7eW9$t0>T1#turpk?UF>eU8ztC zuLcje8J||_KH~mrN!zeBTaC9_ICf&lY*C8t2IN{KTPaX@$DsL!4{qhoGL{>$_sboZ?LQ`uett&Z~#kn)Ei2#kdSBn0?54PG{wxb99ku< zyOC^H(mAN2N|w%`m+}z-%|c!rgO&%=hAlL?8OdJK-}0OZZmao4-L|i0S!D3r-4Yi; z;1bax4^N?>14}!Jhbt=`lRt97G33E_XbD|Z$-qzT+($E;lRPfzjM(x21$Wp(P5=M^ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-32-TENCH COXE.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-32-TENCH COXE.jpg new file mode 100644 index 0000000000000000000000000000000000000000..977c45692fd63ec7b7576b2533ec2304846e37d4 GIT binary patch literal 1151 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrYj)3ddX{~B)8*veCH-I)|zij9K zF)ht}oBpT&KZELj2E)7GOtaMwH^{1Z)L%Pk(y#e0y7K&|wOPk2)@_-+bNSZBbve&- zKRLDJ_nr@n`O3cI>!S6$Ev{_qdi{Re{`N=w)w4yXFRDM|{%`-%e^YqtuK!rDH~5aU zr;)waxl5(9GxpuxWxaUcJ9p{aIUfqIGdI7DI4|gZ%k{I~qM~qjZ`X9^brHTsho5Z; z+RnAv?c(Gwci6Jsb}bAN4)5OSxqL&Qt7tOU>5KNyl>altU;5ABb!E+7;eSc?U;pZV zS)=|ZqUOkV*VnTn^c6Dnlp>WAPiEXC;7hKH#|7Ue)JZ5QAx~S{fJh}N>y0#a^gkD@UOLXPLxM$Jbo0eijlGt7mIOVk)LiizCYvdO5%IkkpOv9YpDw6YZIvKT~i zYDrBlJ1*m{QZ7xc%j$>*qdgKM8;WEalhf>|y_~au?e6zE&->5wocFw+^D4)b&j2l7 zPajVJ1OWh46`-5|_5esY906BDA`l1^3aO64X<*Q3jDfb!CLF=g#F$`+#}iF$%!y=c z3La10VPU;(JDpB9F=seC(stR{(PfuzO`Ejc z6|Qq27(tH4(USA7At`RcE?xhjB~v@j=@gXuW<7laL$fXBR0~VGy~7SihSMH*4^J;| zAK!xktiVHukFY~RzdRmx;$(QtH?eWtGw})MFQle@pPrGKUvROo=u+|JlJe^}_%|ym ztEw9so0?l%+uCn+_uLmfc=&T~-{awtC*skuamm#4=^5$li@D!kF27lMyDEDpUsG(j zKmhuQr8+;s{^7!^To4!x3PWtTKnPdWP%KQ1v>mSPdH@j|twW|IBXMqd*Mwatik<(G zF6Z=+`esx5l-cqI?W649!BYO0>@V2gE(w5vf~wDhVgV-bPC*p}CAG|hwSls|LH?p( zZx(kO!l;;+FkL2iaj@`4;e^idC%TtIqGNQ|OA=>N!mw+?&-6~l4kC%syy#&u|IWx` zNo;G5L1g6NNUS@2_p0dU$Jm=)0DcUm^E%V>*!n-$=tBDHc#g5$XoDbdKyN;8u&~0}8&H05T+9>2rQ; zNf9eeuW%5H6S{N@o34@v6CJFw`l)7ZrFJ`g^h1hv3JX?6VMlxqW~M3Tr4ew)_#BZ<}~!nT`i~;v=qH$$MHV^~^gTU#+JEj>*c6oZ}Xan3yDX_SUlDU9QO1 zG{@lEsXReAE@}_&jBM)~D`F!3m2GIrAi>opI0Wq%rQOitl!EHv)X&cNbPRJ>hLhSq zq>K%VFPOV`tSn2{wvLYC&oM+EF{kXsxnKQ47IBd|Ebg2s^qP!8Sr#-}=kz=9HLFhN z5uP&}Es2zNU?n%vOTX`%l^rS;bDSj~$&33cOon zhl{ho&4di(Phlcncl9bbO-=6u8tT)=jm~&Hm|pj)VrEqC&){o$$ztnVcy;oC1azy6 zJ<&c;R4ZOo0-&SGO)GYIbY;iQnu@5N-YA8}QANgcC}~KVNtN>Y0v)g<4z9HL+1cZ% z;$W6~i&HxA#U|lBc_3d`P}P5R&}PzhzHQ-kOo=#_6>*}L+;CT$-PZ__4PJ+)^-^*# zmOL3EMkE^l8ma`?^Rqr5RtOJQ?GfpdhMaCz-#$w(lDoBAnDy7r$tKy%_vbpfZAzf? z2vc72BI^gM_JI8H_@p3mT#;aysO4$^Zzotz305sAKT2MD2tAwAMQCpZC#G{+Iv!Ly zAv11e%F*UZK!?>d(Kb8!Y5;9pu*-)Omcc_DyO^_kvGom~{dtwYruLzGb*t687bK0= zO5l`83AnIxKR}rB0K^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyFAr@;>sdF3u5Y+kSh;y!L|gqF#k^ zy(-_6Jm>03ux^!kH|0M=`DMGgT4sB%=p4Qm^D?IA^69?}0C&ye`f9le- zY4_L7`gVJ#&9SSZZs#uYUYXT?ORm_eUuW{CCAALBU(d#^eWay6O4ZnW?&WHm=AlGg{&EMN)n8+u)$Up4xv4ZoP6`8SR;#pIj(E zw`g0?OclOzYfTp6cW%W$LTA7Alh|@~p8oyW4Pq>Tla&PztQHHO7=LTcM>Z}`)|kX8 z%}pF%*>>JlcE6IlRok&Z<+ttfnOnAWohqKNJEX|M_VuK^!0emaE5E+owEyl;p)a>r zuFhX#aB0ojlwCVIAM|KTsNA01e{HL1*@BqKjXjf3t_r`^e^a%KV?lNI%%UI_g#xQd z{tU%_8?LavpDpe5A;!sJn)@d1(nVbTeui&jd7mD-73p8}pP@`eubwf~#G~%?;@c`cHrL{f{OUXV z9{-v(>*cvD!LrwNcdgfc<;;4M=3Xtjylvf*jk`WCcR#afOKREf4X1N$+lqe#Z{=3K zq+@qKx4h`O;8c~9hKXh#5(*#BRR?7W`DU;0c)Rw|4ym4KNtGPEU`zGpKPp^Zx=P>f zxlS$lo&ITG#^dUnk^%o2HZ3jC?g}q&-5Qni<#Y7AS+CqDCP(aUx*yz?k#|Y^#$v6D zd;DIVxbxtxj7-nvFV-{E_w1Uu?ppTELz`E*oLJR;fVtrL`uM8yb~W>|?^07%33=}< zXH|cT>tVP^PVZ*Qo!GDH%g8v!r zGi_Y%w$$W`)YjX#|8Y#pPM&>9$E5o9T`zs5ThCMOC3zU!43|7s+vl3Su5rm~wpcp` k#}kTGp^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyFAJna=xd+c!I4*3)G=+w&H5-gO9I zPjGl2*4Mn8_2lORUw?)i%b=Kw%JhqC zxBO?=GXIWDTMCc#`p`cWUkt*H{{)tkfSh=yj|yc_#x&GdB(m!1U%@?K$wHMv?9YxDAR zo|pDLxMQ=x@c4{1MG{pXk3WpI)t>8}eYWaqUdpudC(4sWif&c5=y|ZK?rzYYwR_T( z7}0zC%i}Lh-`BcWr+lAsihWkopsYVX8g-3i&LJvE<)+9OOvE@(q3sn!z`I)s`wST!xoyo~*@56TQX|l|nc8Oi^?X>%6@7-Ye zE^DMEzHQHiwNcMBbN)<8H99W3pl8z)rXuMLYCD%|7M)am7yEj7ZPh>fM*E2u1Kp#v zFPdm%2ixxJEm!e6t3EB%smafu;p5j;+jbqBot??MOS*TXg5$q#^~C;n7hNkW&iK3t z`(9nQ=3G(c)Y9I#OV9n4?!=fL$=h_q?ySOTO?LV6SN2jbY_5K)xmWh~?&Udwt~PQ? zp-U~jH~n=I+MRB{_|>;*w<>9((bbOHLb8GdmFM0%R=d12we1|ol1)1@g-pZpb$2QhRRy|=YD>pz L7jw^Z|9=wzk9KF+ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-36-JOHN O. DA.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-36-JOHN O. DA.jpg new file mode 100644 index 0000000000000000000000000000000000000000..2b19b63483421082c7d824de71768939d6d7f1ae GIT binary patch literal 1389 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nipuio7|C9Heqq!8bms=v0G+pkZ&5qYvu z&BO6f-x_}g{-TOiWmd%|i~chN9+|%O)XJxc@^=IOoUb}nTytl_{ZGQR0=%h9p56D^ zH`CgFnt=SWN4)a4H@Gc&RQLM)Zg2>>h(Xz@|~-7JZ%P?RH13|FTwdsfS|a66-*b&GHV~*3R3N-M43kv`>)q_;Tmy{RmT2Zw3+;vuctEvUx@tgV`516PL&FjutVd4H_bK~9Gzphzfb=#Tz7vGsCZ&Nrk zIq_{w-=pRS=9Xc)Am=3KQB7a`L)*T!~97` z+fMi=EY|(C{bi*8%Dc92dmnD@m9ywNA@xdCZw;f2@NosHCd+s7-p&4#7{neu4SSUF zPVn>IlgeC0#wXT^c660_zPKG_Ki%K_yWp8@Z_VpV#Xh9}wv;yZy_IC&SnIy>k9Pi2 zQ^90cmw;7^85kQ_t@cWl+{xFp-u}1tSSM@P73Ip3+bbT&p8K$Ch3rYTH5bEG&&-lM zdg7AJOXjG%DyO-G+Ags_;N5vzW}d5Oit3JO77w@WNR)k&=AK=#IOO)pbBE6d#+qK6 zn_Bhide}yvQwH1%EJRhet`VD98Fpbw;B)JMgtR>`9;BWx*R@|aTk@0j literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-37-John O. Da.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-37-John O. Da.jpg new file mode 100644 index 0000000000000000000000000000000000000000..5507a00c63ea5d76215a7896638af232878bf84a GIT binary patch literal 1483 zcmbW!dom*`+d&){_{TPInU?3>R$CLpy%W1 z?Fk?V07z3ntpJ+<28BYSkQg)?jm2WLv<-E%aX4*)zJacxDbZ}DDUn1nCp%EgE$ysG zB%8IX?Hs9eI^B%2-hCa-&B2LITOI;ou~==Kwy}z$5i7s7#cB zeJ-vPXgG0lwM@GlqjQ=+MVP^fK)a*}sFO|1a5Ju)kgXKpPHeJ`auu46vZGDStwIYS6T3`%2g#LuJ4# z>T`x7`eaGOn#wUqI>lPbFLtn&=h}NPoVH89x!8HGEb3i&J$mbcAVYN^_=lF!M!TZ} zH}yRmh6hq)%;O*5UV4iwjPPBO?hb#TpORB#d#fa7-Xy`UWro$t_9lI3JzSDd7Fso= z@D}e4&=Mv`FZgClB#vBT1rF%;n0HB=6Z90qIAs0P&^6oxS4!&8>nOK)#mdHgp?ta6 zb5)N&nHM$0aN(VfA7Wk0tC8k?DA_u8wv5=BKDwLhq8B@e%`5m#$>@dR5k%S3J z-JZF&cxBp|xU210-DKN`)HkwGHIOl@j3g%mZ88BIaE)?eeMF0#iI~ONk9hm=!ku}d z>)WbpYN?MBZyr@PQ#dyU{e-`UZr&xTXLt98xA7AAB<^Jijh6j-<-LilmeZ`EW2W@Q z@*{*ZUi2h8t}(}FPqC6dPN{9bR^VUPF*$P1p~0;OGvf3d)T4M|#R~SpBC)u;MNw7N zt_G}2bJhhrwrx<&yi1nyGuGgKsl+#B*p3VPJnCH|B3cDS zT?Jo>6x_{T^Cy!=^qT`b19W{K2Sq5o-nqv=n6mX-vK7Y}qXIC^ZuszkQXjt}|7|Lx zmTZQO0N+0lRyoacr6ruJ?){?c3B~=6vOA;9aBV-gQFxw_LI9Jbm9m*euTN=$)aNdW zrB7Lm+w`+4lp^(Ib;!p25czy?@La0L3Bd#>Bfl)<<@muK^1yq(i5kcSgQw8SdS(@l zyP`!t9?#jewL&W+bAPPSQq^oS}ve$5Ld< zZKUmRC5sS)k%?SG$&o~u9ktKy^Xy-{`~E!N??2zq^Lak6=PMWxya6N*SyQY55C{N3 z!U6uD;h;?)!swD#@7 z@6k6jG*r?wF*7C*^$ZLNTSGudBoZx(mK76|C8#K<5dL!nPXG)8Fad~QkSZX80f8|f zK@*@P+!G4=8sN`>M8FUz430peM1>925`YK@3>JZap->1!*v$~W0}u>UQbnHxlXCQi ztAvYL?|su}IObx=aH!GGO{>1jh&iWyR;4W8tpSS*>81_Y;9vpIppl} zJ=OK7o4c={|8d%n0fDE&BO>Whr=x$l5Ep;(QbJ;S#?{QM>}%JV`2~f)78T#QTT)S3 zRb5kC$F6_Y*woz8`n>H$Pw%U~e$K$)(CFCP@rlW)Y5vT@;?l?EPb;5S*S1_B0Q`+5 zyuZQz;lc=AA`l1|0^f3hMCif;#z3Gd`Y=h7BiuVwN|g|ckhVNBPsN<1ISVZ*ROC zPYVh58`LeV#QA2NJ4+?9NCH4^q`V<{v(xli1-8|y&%si>j_6$A<YB6 zv1}4q`UuSaeo77BQ~>lbN^h;y81ecGIO0Q^)fI`{&br{9Qe={~ZXFH{T5k603w~4A zd)y-dZCen-+~4Aw+Q?6w+8%WIV*J`hgbsxXocC}+C$l?kflSSwhIe*sg}n9tn8*}{ znI28o(8zEnhEdX1){u~Qvg=E)3Zv2VvIHTjb=2YIf&h?Gd_1$s-O1~X^w ztFdyt&QwnKd~J>OJ#cp3Q>4j{<5iio*xRM;21eVI2C;WrJ?D8+YvkbO1$GXk}=8# z*P=aFvSh>JcLbn1EVA{+cUSIOTsAga@C>(zPzvbh@1<4Ney7KpX@7o+e>cdvA#;b0 z{6U-gI=K9*3#l)M%YAgrC`VnALJJJh`>j>3DrqHH0@sF0Y}jTNY3+yqAnSdw?li@eh5jRg9rSug&mlr86~f zR^bD$uHqpskCE?jgW&HpLMMX3nP!&$-XxquCX2HQk55 z`_=d33$v`##w6hKDV}?r5Ulz{?a`|f9Wu0y$e<;-d~S2rtx$Rh{?o8c(Q1*K+&DDd d$38MnQHGeyv4_F5@~a9+PE_|RI!c1UKLB`*tu+7u literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-39-Mr. Huang .jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-39-Mr. Huang .jpg new file mode 100644 index 0000000000000000000000000000000000000000..9aca0e9ffb6d3b4c04adf5ce84409d6dab4da576 GIT binary patch literal 1353 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrYdV+A4%@bHdvy10K%Zzhn6} z>-mQ^uMD18o}Xx*boSil<+n~h^(l5K9JxK!~C zi+of09w>{n^TJ||koVWPK@2gIW^Cx+o|NMpJm*K45 zlbbK+<+(>JN|z6?Se99~=)BMN<{wQ=&#(5Y{7bo%Ens@Cx0;J*v(rKMmB&R*=e#aW zWj!RbX5wo8BSHTerr-K@`Tg}Td`rGc`HM_CDdf?sQgLs=lXX>T8&fwnJ&)S6We4xJ zj7yJ{);PM!D+|6qH2+xlTeH*Nm!9nElD!_N$f+$qckT|Bc~2%s?z+5r>eQ+4*XK_^ z6Q1@p`R0jfR?8M=`Q4Z~sYGt}!Mnd!+V^JMyxkqGp?mQBRH0e`d?3fJAYcp zRm}bSzBKz^0?R{pA6&ZjKZ6_p!tD_oJWRY7XIkD)HlDV9%9ozWliDxOn?E%-ShdUY zQN~i&iW`Yj+Z;ugdRRsUUR?j`aK_my{}~EbWNvR>Fy+9126vhI#|JY@v|E1Bz1aV9&c4HG z%IWuw4JSp~?|VKeYgseX)~5Iq%(1cSXREZYw`r9-kdU->^`=S1TO&(vmc>8+TXpGr zz@jegqCV%KX%g@C7tPU{zAMb%Ofxc~GRs2qzG>XO=;-PlH(MAUOc1*FyzuchKkMtx zkMw`CyZ&d0H~7!+B58sDPf4Bs46pCSKb!ja>Y}T8v)9eNzusAGtA0k;$%PTCSB42F YDD(&(jbGIDb)lTcif<-S%KvWy07WiD-2eap literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-4-ENERGY --_.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-4-ENERGY --_.jpg new file mode 100644 index 0000000000000000000000000000000000000000..19b02a9c776babc8724fae99bd17b6982659ba33 GIT binary patch literal 3155 zcmbW3XEfZ~8pi))Mjax0Ct5^jbcT_lB#4$X7%fVo1ra5>B!WbA(S-=Z2u6<*$vL{j zQHCKhNSKKpWulFK<+Quj{dDhr_Imc0_g#Cvzy0pD_dc08nE}pT)4!?@fIt8MIvv2t zBybs^qoJXtp{ApyrKP8*V_;-wVuV5&d05$)*#&ro1O<5c`Ok^TNt_dr5#{Hz3Ie94hCoiUBTnN0goT<_Sn(na zo4FIMh#x!bVQLYb=%w0r4h!6dn3D6oNAwJwT--do;^!qUNJ=TIsH&-JXkNaer*B|* z^_u04UyxSTHYgX@TescZJv{vb0)y@chlEByj){$X@-#jz{dq=aR(4MAtKt%LX&I*c zbzOZ!?C_TPa$`oGBj0sA)>0bm4!PMZg20T966ar}jsRDW1fSLQcR zS||g+VZ;wgH$6RrSMC^Kr9U6CY!C@wJId_G?-1r~6r3Na_H-R@-i`WDHh_qscc1KR zf^s8PSZCw~n~2Q^_@2r88Yc#I zlcLT2csWk^-9WT6lvUFT?dbr<1sVq_wqvW+hvFf6 zV-plEmWnj8o9j~lWm=RpzM-EF0Sbm`-`*!kCwf^J6M8Rv>=fF4vOc+44Qpu*tH_Z^ z=lkZ|3kIKmAcBo+*VOWQFE{%AhAX`I{;k_r=ziH=)BInxwNUnTS%jO$XjH6p0`UZ%HF z7hR1f=G?oUiM6a^p`94NNQsSDbeU28_aQOszEm!$j(oQX6G4=ZoO07wrr+bu=0esT zVoJrcx+8?0-n!3_nO>cB0}FcY3{`v1%eG0DQ)0=)N8>l2?dJ8XgQv#lD=;2W%;A~= zRlFBfo9;*ZYiKAuoCp=n7=EJBO|d6_!hh)wXi+=eZPag%ML($+*lY+rX}EB{V5hz7 z?3+)iI=38qF!sS&sCPGgRJ3yFxOqa?jyWB^PEUFIP1l>C^9&AbQs!<1A?%14gDjlI zbx|%>x+IsWphC|JRGQ(YbHAs#Xj$`z5_9x`(5$Ch2g!WygrKG}E2*L;9fv182fGv)lz_O!8ZmvV}yxxm{B!rN#u7-M-=R6HlrY$fyl4zy0-(r3Bw zI{%s)Nb>jt^#HznhOKH4rGwN_uF|GZS?-^#Xj`gvoH0LKq5K+y&d#S6SgFUswXt?- z{|G@ZIvYl!7(41@QYqdwT@DAU9gU+it?3Io3ES!PID-jh$F2L?G+7OoYhcGgJNKu-9u7)!u#0SdC zvZ^Ll)^8dUT$K`ghtHTUN?Z17c7^`>DRJGEVrmiZ|!5LNy`n%iIEO$I6x z+3W7=&e?@3Y+-zDIh+u0RFq$suls-wy4F`v(nM{&2baSX9cCIm617ap-Zg2x{qHJZ&MLY8?pW%q=u6eZ)d2npIpv|fo$?rvAaw5_wpmF@N32TCZW?=z>%wdEa zd4>|uAGuL5Hq`IOH9b_!JgYKL$Q&F?x}Q<}1s472ib4EJxadqZi|rQ!ET}zNWpAkXVP3sgm)e#xT`s%uOjXKToFjWdyBWKSjJvWi zJy`5tqA$2};{eVoJca4a+z(KhY27JOEq?t$<_8ZCkR0Tn(uRC$ZCq9Q)uhx@!9lm% zHKvnFx}1{ay%rdzeQlZwI%z$3i&ova7%|wqxbu`NGRQ4I6dBw2I?B_&_*KZ=PTp=- zH+3@wM1~?ygSj5Sfv3=3iSX)H#kkmrBxK$4i33s)a4T7d1fR8Vw%G`(6TsNaWY$9> zODOCp;fQ%jm!8B-=cU#khW!vCFxmAvlVgR~4GbQ>D?Q{Sep{awqM(62uprMrS3cOr z1?1eRR)Me#uBy&KQw+@?Ch1~9RvAN=!pvx2FSlwB*3af?`x*z}t4MYZ250w(I`39; zj2v}1GLjZ;pCoS%cn52B9xK&k3K$tTq|p^p>{99^iUd}0Yll2SPa8;qF0KIwV|=BJ z$27h-RGyD6+Jz4ZJiiM0IIN|Q`@v&+*&drmIssNqv;)>zjil<>uu0$%6$CmGICscvHus){ACrE9gA;Qo%wt7a>#(@R-!y?brN2SS%u z?MT4~_92i&m{QW?@)Lk}Vq^X3w-NLK)1jjmZf@`V!xj=gY$SN}HzYwWC3h_+z-<(Z zKI-eeP$t2&riok}$j73I+Lp*W)AmYd-WpKVVP0iF`ujnCn|y{+b}g&(>qY`4X{@An zlQCtgXCjz;p&(8)-AU7}p*LNO?`bS%_~PS>0j5M%LLUfV;Z9Tc9)*bjOl%`+>TB zSP$X!9G%I&CG=@L40PUJbslMj(S|%TGKu~U{tl}>rr9iRNgM$xa_JVbyykU zw`A>IIxydqF2hEe6Sgh2>aELjJHHZxq3~sjo3NyA=;v+gsu(r%t>;y zOv=>1`mWyBhx#5VE*7LiKPs#LW2sh5Y)jn#ND7@5Tl$8KOdH1%GSfp8+{= zXOfpJ4xL&I*RSVOIcq#C%uC=WZK;?e+2GHMez|&PgzqIyahX$i+nUzqt{))}X=tv; z4*YG);#6X)tG?9oJ#N8^r8r%I!Bn!;5ANe3EJ3frHl|~*bHUMA4Q<^-K2E8-?F(J{ zU5omX;{K;Xb6vqiiDzwUeiHq&m}2CzXft%rmA969=z%;-?xq3WyF_3nTXp>~sI8~; zNuk{30^OI)zPR@y6x+@qq0js06C>GYvNGzdkL|3of}5<^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrYhLdSI9mM8+PiGm?qkm!ZigAZI(z%& zwME4}qEFP`WL(XbK3L1_wIV>LUyk$q-6Z2*p|-z%K0kTcedEOQU)H>p+`4|F=ab_J z;peAY-~BxIxh=cTb?YPlK5AP>e)%G9w%pe>HShX=hUe2}|7WO||D*MG<{m@&ZELvI z-q}p@IR5zl+FY(Z=dRTkT-K4BI{D#M-le*ii%lHyoRo7n7 zPI?!|`Ejc6aaHNdlM1r5wsK9Jv+eqmyZaw>zIc}x7T5TnLE&fp>u33YC9bAq&8n0X zzjj3aSI+*&k-Ov;rpop-zIzb-t+P+c!9uU_-m*}|x5=w+{%b5PF6-)(?#L4_FP$Z-WV0FX(|ksD&N+uW%;@7d$>cA_3Y`>>`Nao z|9E@;vAa%Us77Y>{XhR1{AWa+caGZhaee)HU-N%ICjZF}>bh0%?A7%Jw{}kNt^Ze* zx&Ql-`Hv%S8Cb8|cSGh#KDGJ1)AX#C{NDaO z-)d_)w=FHdod5pVS+9?F=l=vmD?0o;b7XRVV%Jrdh>L+DS}P7PBry6d+NJ1aa5ew? Kl&SsyZvp_D)cbk> literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-41-Industry &.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-41-Industry &.jpg new file mode 100644 index 0000000000000000000000000000000000000000..1d3480773345f97b75c5a8bd561b8d06a025a8a3 GIT binary patch literal 1485 zcmbW!do|eY4ea`#-^FHS}&*!{QC-ep=Z*}u<127l> zFnIx}2e<$%8jV4tuow&mhr{9tYKjB}1%jrE>IyX-EsCy=7MZMPWND&jKs6+jSKFFW z*ILo(bc%_+gB@*yr8S+lJOqZr;Rp%@lAPKLdj!kSH_;i^D6(8!D6m9EL!^kq8tDiIjJ92_Z7|| zE9@UGqTB^XA`wW;vI_?1$Pa>uMCn_hRh)KW7?G+5G#*yXIkT|50cU9KHK87SC zh~B?ya+&r;_U~ZH|4a54>~B{$ARu7!&qEM_Blx)B?40ZYl!gT5dA$z1_td6JV+C_O z_@=~BmL}Cb$>~#jjPU?7rw0Nls;ov~b#sQ(n4(4Sj!>nnB(@D7>lOR+d4{LD5Qou$ zkv??a>I>RVp7iT#5(YLaBn%Ixx%(W9pFBH@ugDtqIWRoR2&yS`W!C=gbB(3i)4NZQ z(mW-mNH1JW6fL%FOKNnqEDDLrt51(+$)#)5`opKW%>VS~u?eqyEidJI=d>bxc(cV^u_tx<4MD!L1hTsDOkPak{Dyt zVM|>Z)Y!+#xKegfTeE$@n>RjG9iN`N>2}!jrV{digQs=d%bH^UHAHG_#+_I7DE)4 zYPzGBS_Y##E~ax63LfW+=w)SZ?pDQ)8R4|f@RLgJu|p@Rx;jx@-N&z)0^d&V%V2-Z zgpH?;(payzV14+SxFt!WThEO6ZGKFwg)GCuc$fd_GMQL!+rX*ch#&5-CriwPb=d=K z_Wr&@pR|;gn#+3wU9qRS#}=iXjn<=2r_2`6BZfZu zq1!wjPa+gSV210zQzU&-Wjez+4uN#j&+q#^Bo`vax$8zdO1Z!4ZG9ri-ol8~=(7&( zsk<7_Dwl@OmMW0Ky(}C~97t_PtrnFJ zc$kHty;e)j!-v1USz=yiB7p#Zc2QiB-;y_EX|YE2VcjhxfxznbVwu@0D`)-?TTjUk zX>uaUxih2Xk88X`}YEFE}r($u@Z5BOl^1Mjmvz_c87v_WQ9pz z83djwu9rc;5dx~-Oi_Drn~K!pOhh^9Y*)ZLl@%Z3_sU}GT*J-CJz@w1=}#3vU4H<7 CjE)xo literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-42-Financial_.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-42-Financial_.jpg new file mode 100644 index 0000000000000000000000000000000000000000..55d523bca1cf1b88918a0a30296f139bcb70afde GIT binary patch literal 1471 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrODq*0Jot5i>9W!+%L{w_WSvD-s+ybZ ztZ{)H(zv0V@FS}o}oL$0e&)xBh@61b%>sq;?{O8|mm!D|guV0+z|CWE7{*>Ih z)5qQ(-2VIV*63B!tCsdQuZoRz7dBUn_}m}AG}1j%Fjm6I;6y?Z$GQ_cKAX(F|0GH_ z=HAUW5B6`YHOii;r!sMpOF)pr@!sP$$;<7gu3sH1t*iE|^!Kaw^xwxDe@zT|yT^E$ zd$y2o)_Sc4d{6t&6rOw?^P}TbabVlax~j62Q*U2bd9EutYwo|PLuLCW)>iMy+ZBr< zz8_$?Fx{%P_w63@OU^dtI)N&^Tp~5XO-4n0yeX2pZ;$K>wmnjAR?e%t+L_xckY8bf zRsW2xs#jim$$PIZo3nQRudUIV%DoXEcXpn)dHiu*{2G;O+qUjnt(`aj@#SqtK8xQu zy*%ss=Dj`NJ&QyZA6nRP=)uD0jrT9l+PNs|bo$ccBBunu8&@VgQPA;-KWwq)ufEIX z-Io{b-n%VQPtVwRv+y6630A)ZgI`WqHg`>2gv)NliDHjb&lUdB{nhyM+Le&S7U#U( ztFyO$z3%$6H}1RMt1$UnW~ce*e0V!C%6sQ7kjO|f~8gF3qz^JA8r4jE#kj&kFEqJ z&W%&$j&J<)YSxdci^iL>kFa(NOxI)cJo!MX{2TB7KYxLyKZyR%&|4Jztk*v6(cJg7 z@$nP2vqiqFuD*R&?_S}W7_}z{9YyLjWdy%h3uHZ!^4a)GFwS>{ul2Dt4MrLgwpENT rwC?uySk>-q;{DAn_VdQZldn&%Hr*;?Q>lEu?CP$)#}+SJ{{JQb?j?z$ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-43-Community .jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-43-Community .jpg new file mode 100644 index 0000000000000000000000000000000000000000..269eb9e74b315b057cc7f1d339f365f52d100eb1 GIT binary patch literal 1494 zcmbW#c~DbF7y$5nc{vCKF+fx>G!PO&ydVSwfku==Y_&x(TIi@ixICbNAXha?IW1Ts zaugIaQ~{|X3J8ecfyQDL6*NNBs3@04K_HPp^7_!W&h)Rg-_Gvr?0&Pezi$_Lh&%-| zS9|$*0SE#BLK{G2U?tGO;|X}24uL=*5_NP*#`+{ZJ(7u`k%6(9DeW^eQ!3S*vDnI- zzR;3NwO%@Zp{*U8&8As7yE(C37u&O0(}zGrB9Wv=nxn5jhh;&vVEyMnE&;L*a0X5= zLctgPqTupJziIyo;}>EY?+ zy~=0xdcO_ajhnvl59Wo0hJ{CLixI@`ii=N3{2?VZEj=T1|H0ftdHDr}KNpG1emQ=k zyrS~t*>kmZ_2(NJo331KyLMf2g|z~IoxyV0?6rRx2J zdfEj6_#+D)Kf?auBBL%077Jqu(=G@Thjy5Z#aY=ni%}!iFiiro2%y9VP$0%#YQwHJX+LM z-qfp245wAUzx&tZL923YF#?uPel_{7_1V{t2DxI)PJs+xoJF*Rgtq!yhsOGt@2XP{ZM+)d(PaD|gB@(AX9!0y2J02dc!PXj+*fT zU*2%p+t4ecw47>$gH25O8hM@7#^RRZghkCfOWRF2-R3BN$C3-B&74~UVmVqD1bi%v;PcsN*&UsV@@{+Y4BY;_ zo)?ngNffEtzvJa7u_Kq<+_slh+mFOeWfgEd5OBvcIy5%VlwWx>$w9LR>$hv8+ib;y z($R5(W{R>e_{5;fzrNfkKi4obN8$pdo4C%GJi#}sn>ru_NHlUyH0^eSbss07S~n(H zE-ka);M}vuj5B4~6E!u8{tpZf0naNCkdm|p{R1(=r(c#%ZP@0!S$ZE`q)&>@+q_R5-G|;A^^$XD`@%cWw}G)Rfo7WDO*obc^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}ni(m5;VHKAU;Sty$;z0+DYCU6FkZlBTDZ zt}syBSFN>QecP#f<8pU=SH-l5oKjBZ#Hp#< zRRsP2-ID*pccuP=%%v9`A77YxygGGAfosD-(WSw@p;JWWY>#b>U2JBXrM9C@e9NAe zQ?~_|wr8lbJhpFKt99n;{>RU~{xhsU_@80Xy)Ca^TE+!V-n;)qM%VLywrT$v3inS9 zyIB8(b@_jW>wD^dajZ7C|M=ScckiFoig_tpK7@n|Oi<}N8R;o`@RnAVr_*xd7xQ1J zy*Zv8wdLAJf3|rG-!1AmsXTu-SL2-d#W#W+~0L5 z+x5!An+ExYz1t#Bm#Vyba_x)fQ>Tf~twZ!`3nPPDy(Sl}n)PY_Py3^tYWJ4;+_L?8 z*y&QXkTCbp&fhH$H@GifXxjTI);;{f*$w@{Mn7+F-o+wu%y#eVP3x2&@!z|=Z&Ll6 zt!>Ht$!&AJa+fa;zpMA;N?3Qsn|1Y%ujPN2{uw6opW%~2|F4hne_6twp8r#I_TS@r zt*2?%Og=w2zlnRv=YPjmZl3e*mc`nS#Xb3F-oD#^VWrW(#mj#F`nvc0t+$VFKTR@E zl7E_17dY?d(TqEppU+x*%$j>g>#m?#Laot~YiUcf6V>+Tt6YC6di>qCL&+@LuBjw^ zJNanKI+NYeowp*@cZTIH)Qfe0!pG{PSm}51^`HAse=VD{*uI@YOrIGU#|1(Ui`p@uh$JWfNzJ3?iZ#g|XBw?+VD9abg z^xSrtDpLEo>Q-E5Mp4L-GMRqZbEUZs@v JF;n^fO#s>_kaqw8 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-45-Marketing,.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-45-Marketing,.jpg new file mode 100644 index 0000000000000000000000000000000000000000..781841b00e47da431e0165bb0a309fd218b2dd4e GIT binary patch literal 1490 zcmbW!X;70_6b9h$%bpOdiD4006NDhhl#oCmhy`T{3`!J&qT(nZMN2UW0*c7uLW$5e zh^Qc8i!8Epr>P?ucxD{YiMj~YDlyo z>FSbg%`CpOqEe{_rgmHC6bDOdDrI#Dh(@EaYFKS`b#01~t`X%wM|lsxp#TP84F!n+ z1P4NKpt2b-Q0)l^KLz+RAOs47BakRGMoraFvlf7WP$&cjg~MSmRd=lF9f0BBcq1z( z#5&d?B#}d)Bna|QB8%<2fW}B!q8(X@a{kH8cu5Rv3 zkKI1LY`;Bw{R4x7kA!?3dXyJ&A~Gua^9w{b3X6(M zZr;9AU30g#uD;>H&&>~|Ex$Z!ebLd`)h&DZs%L2U&B)t#qhs&Kr>19S=jIm{KP;`f zKmhugrMf@E{^7!@To4!x3PY~CKuDD8Kyfg*kre{(#6lk8tRqqqPy}Z|Ue$dx$$HnM z#$j$dM$?!&v~g;c_DS~dV2S@r_807LR}X-Nf~wDh;()Ed!V)<@9%GI>RM*&5Ls)5V zNTeUoz!P)1{GHx0nK{eE@Fna~gM}AyQ?$f#vw-&4n23&g+i4oPtm(JprZKDHxB;dT zXzsA+jtS#>`bmVtTkoKe*ZmA9NVoLU6=jz^STzbHFMv-jzd&I5$Z3pYrtTw_GxXMn zcrEUh;|!lx7GIM6V}LYHKzkw;t4kVvDyGh#7W~jxS>(ZfF$c}Nkt zcx1RR=qJp!JrguBaap)DD|=S?_?go}5Br8=RtU_m*$Y4HX4zKv2E1)YcNmj$iz@3>N$G9jjTu#>-9gG`OqZv$W9G3+2N1+>PF}hD+IT zYQ89XQ;Bg$Ltn+kPHO#gmW#$%COqiJJ;(w04o|)3;BHo{>h*rex@*Ov!$Vpn6wd7j@Sz zS07M7Oqk?GGyO!OM)R(PRPN2}ce`c&;ru51wp&%-RCLU*J1v@7#w7m}Ou2f_UtVFvgfzuerws0lj zKvV*a@DLr}z%`zg^-ADsT4dEu7m1cXO;k3%uMN|(#eHxkpD(F$Ex_=RxdirPN`Z#v L#PjYJ<*Ppc-ZPB+ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-46-Financial .jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-46-Financial .jpg new file mode 100644 index 0000000000000000000000000000000000000000..844a272ffa0291999ba22e00df382282dd6e9318 GIT binary patch literal 1194 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrYhLdSI9mM8+PiGm?qkm!ZigAZI(z%& zwME4}qEFP`WL(XbK3L1_wIV>LUyk$q-6Z2*p|-z%K0kTcedEOQU)H>p+`4|F=ab_J z;peAY-~BxIxh=cTb?YPlK5AP>e)%G9w%pe>HShX=hUe2}|7WO||D*MG<{m@&ZEKi( z?^LQx>_5Ig>XvBb+%^9mE{m|$RzJK-+ADh5v+U)WOY(y}5BynkF{p^d6_R)d2l5(OH|o1cm2uT^#_G7-np}8FY`agPyZRNpSl0baMg*cCZCM- z*LK){xmkZ~bBV3fDQijQy9buP6_zn+D2REy*JqXUt>dAa|1!OL_EuEJT;Pu0i&RgB zcRl={R%$Ob`RdB5$k@oj#2=8Ev5Yl%*$v^zsX6|Ci{d}5Im_?AY0@#LZxalEyByY) zuwl1&8@+Y@~-BOG&i-h1uGz*IKjwGwj{$x_MXpP7~#Py9WjC z8&`hWeSP01el3={6331D8m~O~C;Oj4YKKRtjt4`L=iA6A{`a}>)-@+ zmH+pn`j2a6J*M-4F|)TJev9|q7eCK^p4YxCdnrmdt)3QAUoCp(TWu}pw#D+k`}a$m wY9IM+|Isn0mHD@^ke_XvcBo*ot4qMD#SDxMtX5qwTQv@c?!OllX8->t05^Q{xc~qF literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-47-Ms. Hudson.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-47-Ms. Hudson.jpg new file mode 100644 index 0000000000000000000000000000000000000000..c973bef18fa40f732314a57f50cfc0bd08140a26 GIT binary patch literal 1502 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrYx<482k)M%yRPnZHuH?`Rcw_qrLPY# zFZViEYN@;Qws4v0i#59=-@Ppr-MD9Jy3oRkbnY#4lC=&L&T~AmT71_OGgZN9o@tZc zISaij`(tZ+RyE3IoxgWE_jHXlN3Mmcdc4GGEziW| zGqpDz7ZUD0cbxgd;@i$sPyMcsX1$o#b;7Lbko$AJWt;pgKShZO%ICl5Uzhc@_166x zM~hYkS3LC(Rc^9p;uTnZ=4Vvq&;JYucBn)8Uz~wKeZ9`(R#LHeA`&-XAUFP~T{6B-^`NxshpXF)Ei+`=_3HqcPt$Oy`lIhn4 zO!pWp)-l{EeKIF?wycE&Q#XTj>{Z#HQ++aazLmS+X1BcjX@InLu6pFAN!L`nv>QHn zh%dCbv;P;Te6Mlvj%`-WwtKb*#ygmqOt@{b%2D#x0Uo)XUq#!^mG*oOx*q%G<#zuQ zbGMt8N<5tIyVJX;QqAoCyu-3t9ou*8|L}C;J+Zeznp5X5``CQ``mbx*cWv*92TgnD zU217?c=?I0AOAjDs#J&dtu!hBW@g}2?pmu=^4PF{&c9~)UcNP1#oni`?OXqD|LH5` zJDy$DpJlIQt5&+prsmDQp!u;kpVUdN-c`E%hRKiI?eA0Mjq58rPukBk4gV*2SE2vM z$L_C7FU0QObnWg*qiAb~oBtV79@NcPo4NAXqHnWr-j4TI-T%V0`j!9cn61kums&B! zCR-QScQ5#x)!MglY2=OQ+uY3^i$nf1g#39>tDV<%E#%qr^J3e!UH=^ zzbXxU&f4frcC%S7spWXw->3iB`WFoQ=H$%~xL&m9xn0Yj{Xbra>{HnJTd%$7_95f- z)7JPq9nY)hdH!M9!L=C|qK|WBnH^LPSp2aYw^a- zZO;8q+@HEuIu$jyTz&ug*LM$<%UNf({%7FioBp4{p7%dP-TXgV+a_1C+`6`4+nXp& zCm+5;3(U;62fkVUaqTJX#W$}M`2|H)*6i*(qp_&#&O#2BgoaEr_3z=I7IiUz>7`r? To@mIoiUcNzb!jZ%|9=wz9G#9< literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-48-Financial_.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-48-Financial_.jpg new file mode 100644 index 0000000000000000000000000000000000000000..6a664709a578a6ecb24150ba4c34ad1b7728458e GIT binary patch literal 1587 zcmbW!dpOit7zgn4o2$V*8pfoO%NV(fF>a+Kv8hF(-13AXmog|R?d&9_qTMn>mQW`5 zlC6x9eY!|UZFvm2=9XM*XNe&ib9U4|yU*@l?Y`$Z=bv+)?>V3Mz)m;-mThx(aRvwi z03sUz_5ep9hr{7b(ky~wPZ8uP%hgt_)Lo;eZ(zujk;s_)f2u6t77(D=0JnXt9(ReMKgSNGtXp|`^$qhsR}Qy->h#ItiB=NFb-2ta>f z$<{Bhf4N997Yc(xWAIBZ1QjovXc7jiZH`lR@W2Pfs8H#Na^%fv`L}BcIu<)8RfA(& z@V2gu5O@+Mr5CdCINde7uoaa&_-R(@H3;T`o|QSxB+a; z@RTZ>SW6RJAvh^yPg$DY*gtpaNpSTC+9`KS|G=nCI{6FjUMl^$)czec9TZH4S5A)s z$a_`q$M5FBZDPS0FBn2JOi$21M=y`V;^H5NHV zHc3k%=(Xe)1~`Tc`maCVb1)n=dVxEJzi)VXHZ`luR~n`aoYF zI4n^V7qCuM4>Tb=+~0^9>z+oQhY6xnr?*JSbAT!NV+3uxI)BPSgO|Y5q!H(&#%s2=?(IxEjX( zy|bixfo8T__exhG2 zhM-c&H#%rvJI%ggxSq&e^n{=$64lr4UqJ}(+vLgEPJVhbTCEh1f9Ee!aDEi0l`_oe z!QZbDw2F>#c9(SrM`e>lVR>OtssQ= z{V4jQi=?a}_mD|%^K9XSYFe_%Aag$WECit&ATab~m8B)|-m*J{>@%8CuO5yOA>g{W z*`;jvwJFau7KjDC5D*VstNBNREVd?(Ys^NLU`*_!!9|mutGwbgx3%8N*|3=RV5xD~ zVBNgiOx6iI&7hEKZZv)QYcc=>2}>xxaMIk*Jz+~`!fq2WFU0|Tqxq;!2=N@tK0c?TYL{Q+R4!<+yB literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-49-Company Bo.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-49-Company Bo.jpg new file mode 100644 index 0000000000000000000000000000000000000000..8c741f0eebc00e105ccda66539d8aecdee4836d4 GIT binary patch literal 1479 zcmbW!X;70_6b9h?WnTad5Vm3i1e-u810hJFU`rGNSj4a>i^xFOQB-Pz!J!H&RhDrQ zBnVcmK@13LsU(OxjEEa3D2reMB19Ge*%buTG-#)t>94l$nS1}-GiS~-SMyxc2Uhxc zdV2y00szt$&~$?h0E0rIQAi9LjmBazI2}V>9XwvgRNr8Q;VLuI*Q?CT%`L3$DHc|C zWOH+>i;bOwBZI*pQCMzFx~shtgT6cj!eX&HcpVd6T@$*cxh4HSM{^qxFn|S|;E)x7 z5g<4L(%c0k?Vd>JbHJa0U~mKyg~niUcx{7xC4fP2I1B+tA`u8}ce?f+AP7i(OGkH< z0XGtDwbziIAv%vCv#aWf{;gBiPEjeDSe%iuiK*G@Zzxn7TZZ!*7ba`%hK(Mc94~L5 zfWV-w+k!(vdC@Vkaq$U>KknlTQuhnf4j(z1_4Bdg+2Z_@1(L#3Ma9yR3#AujW#yNC zzg8_*)LgHAT1qH`tDqGh7 zTLawG8c=tS$4Q8^uc-7(N8YSV;S?4@Dp&bLTHsK7gZgRs>2zk;S@A+b&Fqz!(Bvil zn6kUdCALs?Sp$Y|Ma(~!S+_p)@+KVnGzJe5E8qWwDUpQjVv+M9QxVwtw*F$Ms#~>R zvQd%A+@dVEa>ikMh3a$iV)5pe6b%6O&Bavd4IGYI*?orHf7Ptz4nh(Qnr7B?oN$9~ zTj!5(S!J{;m#A9;QN)fUL3`XIu~)rn4`-sQbo0J-JlVJBiX8BiVr$*m$wpR-W6#_r zSld^5Ty@9g9r)U=lt$#EH-+)=TtX&_dC$o=Yb+>x-GK(%u#h2&WmENx4Vh)$x8m`+ z#J-#y!JeC`A1322o_;w)9*FOWH%eV2n7hy<>Mq|%?I|E5NM%xIygGJ^15%u!0qMmh z=`v-#t3aZdH+bPynt8gG$mL8sKJgpg#511rSWLgt%5yO}o=m@>LZ=^ ze->WxLDH!w+u&iiO&U+kwl8UO-tV73{yw~N$x?NcS^t!k*hz8Gwt=6{mEAvjJ9nk857mTg_K zds^vhVLGGqCa*s^ih{b}XBb_x4d|KmVb$3qWg@`N>#y^Oe{9rOot=M&zQ+<4pEk23tm}3qER&#Ni z=_$#5k{Z@W`ow69@~+Bv*4OSTtpJ-_MOo@SMLHUqA0P=?N+%Sl5(Fs%2~Bz~-h1cVnfvNK&v)muZ}vC8+0X3mEOmnV72q(qhPnoTKmY(b z8vu0*&;c0f=^5zh7#SECn3x!uS-II*Sy)*4Il0)mg;8bfS8z=SXo&4*x2}#MFmBb|IeZR1wa@9SKv=DNDQEX zfWQzCwG$9N+b139p8@_I5Dl1?j-G*$iJ9fBfWQIJfWTlHS}+|QE$vzLgR^yj7DC4< zs`NWO7utbA%$r;JQA#eO_|-RUP?I6DgvxE7a3*FRUOs-9q?ELb?0Hp$n)*eJOFDn( zUenV@8JL=xTUc6I+c-MiadvTabNBW0zaM}P42pOh85JE98<(2)^jUhw^B0+U`2~eV z#U-yxYii#T>gpRBo7y`%KXi3}?D;f2LL40%pZGF4Pg?l4xb%H_WtFnEy|cUbbN}G* zmlp^C|4Zv^{+HN)c|p#+XlQA{v<$zzKr{hofFZPWqDu6fzoQu(yt%}bA2D)YP04-J z#w4y{LWbV<8Di#K-@4g5? zwz&Zp;g`RE&E{DRm7sWDGA{EOa0_ZhTk^MQvH^iOD1lS~$8mLzk8z5caz_jLLlYyG zT|qK=aln6b9upi`xxvYTpG0BT&CsbvMn!HJnO*s8kqHWj&I$tLny+22D0#D;jh7b% z&n3l8 z>vqjy_ta7fEOM(vS{4|}{Y)JQSZjSXtMFEz3JiN4tAs|EKHuY_0?9B6vkZ6|qeHHb z!#z$cz+f()e0e*SGncu3O8DL?M`qiMe)0WL;%MCN6GeXQS)sBtYo{BH;ui)<%@^2@ z4afW}^f!3qj&7+v-{94-zdpzS)eSPv%~=lB2shGccg6W5Y)9NYeVb>s?Th64i^4I1 zulOCEQqJ*P`LF?cy4*KGzrf7G$e{#MbMCne=Pkv1!zo+R#4)y@@hC*%4V`_xU_^xH zwu=nEp1AlcSFnA14L;znjGC4E^-l(>lp%;C)^5=0>S1io7HLmVy`gwgpv!utL@L4n+LP`H?Q(`^s2dsjQOr zG9Plan1;r|qrEElZnKjr$EoowmS`ahNRMi3v3T$DNQ-Po=u6QxWc#WlnoGOJM2uN~<-aI$ORSH*%i+N4>vQ;2nF8_3|%INYdo-|lGo5T4#V z;^lxQwm4Zh(E8`F>l2ZPukSRWli$r!rH*R4;by0Zq`gzxg*zJsT5l}ngou=Tsnzfo zr0l(zZ+GXeHtZU+0e;SNZ5K&LH-+~;J@cERH{kWE58&6Ht6Mo7mpRfADtNY@i39GZ zxJ^o>1VBfl<0PZ}I=V&bwU5n%gx+_uvm5%mD$185BUW& zHP&@qSE>AlD8`lGRXVo`b%n)3JWD!Twiv8C(k^JQuZ|8YEi11_8d{QHDF>#3+A?;3h3TiJ2Xm4 z!{gQo5Wgv(Hd6uipQr(w%%7c;FSiaFv zFWyu@Gkd$hqdAIC`ArUsS`e;xq>G7Xxv#$O$vS!$u>uQCbNJe)rfI{bKsf4Q=()JW zn6(m-wm<#?wa$@4U&ECMP&Q{bcpg1fbtWR}Y!hW3<@q-1b(MuoL~*ye)(3g##E+{B>+zkag?XD z3%*{cq5`GmR<5S_sLSgH0wta#Js`GVl1byQfq8uw6ni`ZcV1Fmt#2`zv|3E5oN8;8 ztQjXfwrWV>XziGTE~es5KQ!H`$`*RrSyw>?6h!F;PZ8STRh3j=Ufi2g-jbdd?7om4 z6m5Jt>Blb0B@k&>JB3$oj#ut)qvsr6=d)4)=l&Wk266)x_~9NX6#g9~-yi`kpNh$QLIwPeOw%j~P1u4k zb=#txXDJ1&#+D76C^^v8U};TjtKhi4#+obBYOp5hEaL^G-WK24($9)f7&h_zfRh>5 zO7WLGjZ`2#KGQS35B6PTi~?h%0vQP>x6hs=(b))O%yj=yFL}*OJUg$6F1uq{BT7&E zx|s>q)oD?|S(tp`l<=F9ta;pT4n*HR*BesLKgZ-fyD?4$D$InPL$5E)hFz+nF${RH zZ~}3ZsBXGWCqy4uzRA}qN{K3i?O>vygCXZ5*IOKLW%`e8tWWnWf0SO%Ts+n(VYBR1 z|G7}Gtr7?HR(_tJi@iPQAM4Qa@@q}Y4Mmh`xl3&*KcZ{s3j9NP$c0u4yjUs^XJA@g1H9KQp|@KerByIU z-HYu z%XB~D<_PTQkGgsfArldaruY-b!`Ir0f!I*4amm)WA-|R0nrfCULEkIz2;qohagq?s zo?~KlJk6d#70nKJ4?5d1$DW_FeB?sr*Vz zyJ>o&rKH|#jVhT3_UN?D+5U%Loty2?{w@lFxT|@dwExFl96zKmjX%`;cp04Au>XL3 zWcVVFh`b@LysMNF*PpsD?-eVnmR^6@x%SgSkA+|RncHj#m`HkTd^imfcysk3mWU2f z8$oh+n2+z4Y0S#2s)a>P&(sbW0L>H?o8&*p#30?oKY6a6Kzq^W%mR9jJvz%Ce#j(` z`&*|O66C($1pwrtE7 zrF19fk}J4xGQ6_4n|4>PnVB&d9+2}2aO^?Y{ih93Xz6t8;|KkXb252hstXi&Ew`PG gyCq0deUxdVxXEX`cS~v2x>lDP1m#9w5vG3m2a8euvH$=8 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-50-Other Curr.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-50-Other Curr.jpg new file mode 100644 index 0000000000000000000000000000000000000000..4f3337ff80f33574392ced20fdbf8e511299a187 GIT binary patch literal 1544 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyF8RXQ|fqDzVvi^IVI4GWnHMcjmQA zTMS(mu1ockp4YK_qw=I|wrA(=nOxCR+O|<3(sSdYS+k zXa4v4!!hmJvTH?KM88-jRNQmBoR-il)8pCL@5K2$uR?ii-gQ^2Ip2LF*S~(JY27Lk z7}cY(XyV*})EOFI)kPSmE}bxI-CE)LHJ`8fubRS8uN+zASh}+E*Ostf)s=bw{AY5D z7qm2Zp0h4I8gupgw$&*m*W%x@h28P|$)mjPPuV%c_aVYQ5dkhG(iaXlDD73Xw>>OB zwbC^2KSO!;lKZvyR;PU1Ir&Y-YjN$$hrvp(;d-2uU+%dmTk*T=Uw;=DuXPV}6}=Yd z9kRDtJ0Q_f^~gGz&a?bsFjqS4JG>xKiR<{ zO19x1Q~xtu?VQW`CCy1};WaVu-#50LSh#ia_q1|zjtD&tW(j5X^-tsK6g58=XQxey zFaA@u->>fy2^})jZ8j{xw&5ZB0Gbu20!G?PBh-9Wxhy7VH*? z=xO3J40ipkDi>OP?`3)X)%gWcId&mZPlDs_vgKMaon_%DNY$P{W8S_UqVuztAGnlx z)^zW@(+|9Ar}`8*mAzvzu*`RUA9cq2-A=JziD{UDkBA8V-MwjCGWt$p`& z>*a~*Ql$`YXyH{R?n{Bz>Z>ZrX^;xQNBy#1m6Yr5Rg>f09U z-^$&~d7Ibj$2IXne?j*A`Tjejn^nI~J^HwnLoZj4ODynmQA)yJ?G=lr=GbWM-us_n z-GfIn&qTj0H#(`XZQHCvnp0FR`&&*5i+7b&tp4O*_2NO3b%t`2D7Q?OXh<2;l0|c6 zY%+8I)*e_iv2n$V&;0gpE_QcExAB`UoN%l>-FjlXU$eqhwaZ5is8pFPN@0=z(tSuk z@WS)L^G{<_MW=iSdGh|3`p;$REa@+A)lB>3U0SX&>r)PUszhku#gM9cs|$;|Ubcvc LYiVlB|Gx literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-51-Financial .jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-51-Financial .jpg new file mode 100644 index 0000000000000000000000000000000000000000..0215b4ed33f5aafdec3456e86d8c49b1241e378f GIT binary patch literal 1353 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrYi{5Ex$lxrj$qu2-gEE2E5?U?wQE{b z91xzAcWkPL*_Rf*n3vQqO9Dh>x&E7L-%K8`av0Co2T`CuJRw?g1r^exr5^H|0 zwhalt|DS=eZP{GWo={w=<@O?2uNqY02iDe`tDaSS{P%@bwe^ zGn`s?=(URAuG(t>?w3WauU%jC=Rd>4{Q|G-x{gotb;@=6c8Pmwo5wb9z8431mIyY# z-62=?InYLYSMHu$Q;#ir^N?qb?>owVOSU)s8A>9_g3f3jpYd9;@lEj+CFm2JyP$yh(>^+$Qvtyo+AuYcW*m@kpdw=IA5 zp8mD+&x>pS8EU3YbDqEDjz;3YLx0Zx(zBO3w|j4jK` zsxxQy>F3s8UIdi9b5sdCw))|3U)LztE|Fb2YoBe(JIC>$??GeN^4y0Ju8&Ta%g#Qw z!l6%ti^t&*Y!#jSzYTU?%Vt2{>I7OJC`r4Kf&z(pJ9E;Kc_946I~{_OmtbO Y>>wIa!Zbf_sl}&*?wd@al>grZ0MGkO^8f$< literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-52-Mr. Jones .jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-52-Mr. Jones .jpg new file mode 100644 index 0000000000000000000000000000000000000000..69cbd6f904fdb8a419f6ea929c7d7baf9c508011 GIT binary patch literal 1184 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrYj*cKJ2h-Klm5RYm!7AjU+(k# zwqmX;Ynj@ctWaCCkLz>ig$FE6on*uBnqHz;?Hl-JiDA~n zWBY0rH0MaqeXhQ~IOfxO`%KXj-)i^lIk|h^abMTe+V05z3=59b|KPCyv1)g^=$Sy# zT|1-O?@l;mJ40^9$+ZiEg31cB-~E-=v{yR#GSW_l;l+g0*F0w~nebgF;<^3l4{NiI z%GI?O>M2aBwF)=l*T~C=_~dr)^}LFgfuWzy8CPqs_r4u}z~8+%YttA0pX{gqGsGL$ zzc{w!KSN8u*ZUIpjh)B8tobV)dpTj#M3tSljy(!vPL|`IWE~cCB=FnDn%C)9E4gRJ z7cX73GAiqu__^e5yLz9ypQ^K9;*KFId#2)u65K@SkBq&Hk^S&Hwde zU0Ji2`Jdyb{|wj9-2Y{`nzLTYuK$bhosIR&Zod-Dom!~$S!VU&g>M{Je5qcUm(z7e zAWvc9!YO^N)Av6zsF&KUv+*GJY3-I?w#d#0T<0!Ky>%&jdC>gqxywUso#wEff57-B z(`0Ju>D_*hUS);Ff2&*hpTTjb{Rb)g4{N=&BQFQv`_G_u{fXJS)c*|WujYS|p8TJI zaq)i!u5%aHY-Qgx;rMEs>~EcJ7nx*$(K7M=+LEoTYuA(&N<2Du>CJO>n~f^}mTjnr z*ts(HcKnt5{Rpcy*TeaD)O9g~lk#3o14PFc5-Za0r2b!{yai=Am>Q${SU^IzQ-9j$cSbW7M{&<282Z6Aeg)RN5|c3(MVm_c=N_yU<;| z5Bm5X^79V}J`oan@>EzjBP#l0Ozfq&tKTOjU;E+5l#I+Q7W>A{Y)*bbVbSe7ckdP7 z=RK&Xtm0SKG(Bp5{G{dSv(~Qep5DIxfkDCO>o;%5MB@|hCTHh9&Mz!VmX=r6T_6B` zVaew&uz$F4au*B^hv3L{7YK`y4+sZGXqzc0*?S_7ombW&C!kaun0e)mXkCie2i4;j zIx%W`)X^QY>$K0Ze+NtaU$Vbof4c+#76Rp;2jKuaK(b;`YGveI+PFAXJw8a%Hwb1i zJsAlSmuEAX{_PPo^^!{cwpJsP#@iJ=q1{E<j-byZ?*7zWM#GPY#neyH z9KTu7GhJ~SISzJ?ZrfDko4!+6RM#Y8sj_QQhf8N$?gh4JxHKxo2HlUefnXqh;Z8Wu zy`isl#NE*Cj$sIt<$<=4h?hFWZ(4|8Z}jImUE{<2f5L~i7(mJ>0-SueYT)$rO5@V)ZuugqKT}hNL|aa zTyX<)WVJuNJh^4oc0^Td!$_%1nBlnA;wFxUT0wMo_AW-w}bmPVE6aRLpFAvAu8c z*6GEayGDlU&L`Q96V@u+5BnD;4(e30qdqxUmu#a{T$*-1M-mFUWk83P$Ey5>*Cjb5 zZ!U>mmH$riz8)PwaqK*i%1ri2fXAhbMG<>;=DfXS>z0XEE1ohmQM0~&fgz;vq+B~G zxUi|!Zq2ZuYyY_C!x&!92tLInKj_N)V>QYX1|`qEN(q6!58c0_NrSd=mEuo@+sz|# z*JQwBdG|4L>g67pdfqkQY%KEeWs$2Po2~uC*|Nlx!H$UAqWO6lpgp&oD;xX+qzjLf literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-54-Financial_.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-54-Financial_.jpg new file mode 100644 index 0000000000000000000000000000000000000000..041d96a534b3ea596658646b7b5a2bc20eeff2d6 GIT binary patch literal 1454 zcmbW!do&bS90%|_=FMmg2y%&q1Gs`>a-|HSy3}YDI&$p zoYjLSSrR6;*NkT)&oX9W2B}elogKAj_niG}cfY@L??3l-q6s> z(b{(HIueOwZ0YRgM0B-xAQ4xFK+tG3Mh&Bze6kS_s$2BHFm!4XIlT2)Qia8CoMK%h_+7!(eN!Ia&{mG1zCg=?Cw^F(M- zqL5~ZIATUYISTJp_egsiZgej}8<<;IT3OqW)^Biha^C3e1yM;+9w-(DH(iI&^rRr861B{T87Q1rL3!OHG~Qv`tadbw zr>a9BjhWA_(7wq29W3*I$^L@FSb1-VtFiH4^r9z z9Y36i5X!{lq`f2TUmLZa9^g?)i`W=7Yr2r(H$q^iC4b~bNkXSm>w5)vtZY~l=O3cf zpS8Ip{az8sk@wa-DT>^aR-%TUtk1~(ji?Q<{or!FTg{pvY+iN0Tx8KtX~1O>cKu`F3A{s(@o*i zG@?!RshH#16Yag9E;{%XRF|4HCKjeqx-B@dRWGJ1nqGaBjLeAN1q=^8bGG#cDru*6*v=55XcpU1LMCZm~d zp6ueF(hKKOtk=_`Jz!;k?Z4=2Mjox}sn99HJlB_0eP<*MdW|uvJl7{}a`C?Y9Ph-~ zYMQ?+bn&0BWr(%!-Q9`@1=oBhf6-&s`u1KW>|{Qaf8eo43hZ_cw3noo2J$7g_ThGx z=RM*gvsG(S^+n{!ki3JWyrA}k*kUeC=xkEx_4!-!Ylm*V$*GVnk(7}x4qG0uB*dn zC3jUGZ5`9i;Bq>b_6NP8@}vn>nSp*UD|0B#=}jmuYu6&lc)r7Z@hwFb?>>3*#6kJQnX|}Nv z!boM6N3u1_E7|cVg;623s~CnN&Fi)aX1`-fLA3^)JQ}kX{Cnd3JOh^u20j|(=%9Q#V|0kG}hBI z-C$<9*2d1xj?UP)$)4$CWoyS=E&>q<1QL;?qo$_AG}JR>{^vki02vSLfjtT`0xDz( zMTU?zKv&)q1APeaXFw__GzN>q6I6-HgnD(L0-;bUXcPv6Mk}+wRqg{c8KYrnV3;6dJeef>h#_6GcX23~jI)&bx4dk(x%#6g*{1Zb|JOg0Zcq zn7TizUsY?B-MGozGVO!x-@#J}NRP-MUY?=rh@90}(>kF(Gn)1c&c zwTvZl%jNt7@@~EilaGK?)qAsY1}x7H-Rqw*cj2K^@y0a|YQz1$&g%%P&I59> z8RKL@?R|RHVIlNxd1+m}B(1`%DKLY$YZ>(h;1mtTwD){3950TV*W*-8^EQ@73=p%~(Pnr?X;>%j% zb}RZlBrC00ViG(@fkP_gBo2KE%Ua5SKb80_UJkc%Gi!$GYCDqqD%|kpqNmP`<4jko z^O0Gr`tm+EfpJ5!_eB9BiD&p|xi|~a_U7Oph5zrUN605^=yQ#k> zocIYFmK9|>q0_yN*CG#5jBxGzGKK$IbBSNtSudvDkAI_Baf>_}8&*iw@EpMBn5M2Xo(`*5tUW>X1)gk+6NLD{0n=HMiUD z#kR|PN`i$+cYDYN!)rQ)xtxfo>Osb{`dRC}&x9t|GDVs@=37sDy%r3qa-`NC2q-27 zNay+EK9Zo&9AY7H+qBTacUZe8POVPX-KW1c%+dRzXhZxOtNMhbTdtR$aHjW2_DEv9 zx{rzkPZ&$`zBuEDH>xfNay*;&?G&jRA)xazY4eyux?+~z=Uuok`+lMl9vwu$^N`UR zxR>UrQ1F$1ZKq9Y=zX^65Fehp+MNAG<8n_B@UbaY#bHS|vji_E8@((;GJdXQ#Web- zOu5GN@mF6G@&yI7EQhY*uWp5VhK+`xx zLyFWnc+5a~c9k%>C^{@LGB)z)^}5$3sqNPbCKJwgWSFz{2~}a0g)lFwY-fh)m2}E3 JR{$f!e*oXBuv`ED literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-56-Emerging T.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-56-Emerging T.jpg new file mode 100644 index 0000000000000000000000000000000000000000..529bce7e39033e1ecd5aa3e8f2265c65393da480 GIT binary patch literal 1441 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyD-EhwlILz5Cy9k=j-F|5YFT&rsX- z{L5-zFKWSdyB)umIhp)(7OWbh5^}<%OX-p5>9*R$!-SI1M!LfY!j5n6`YT%c9wyso} zTg~iE$c9U~U*maSuD15rt@Pm74HdyAyOyxtd)?WU#VnU3MUF7Fh8;-YbFlANFxm&(nEsuL8S5!IwRYGK>&6c-wm((0z`Db<0 zq;4;n&fWUHCfEJ7qmJlZ)7{sf^J~#7mXpt#`0n1%M;8Rw>*Rh#qe*sapPFXIhPY+|ZRI6nD& z%(W>`_O2@PuKTrrr{dbP;jte#J(pxj2~(T9D(dAl!-y;)-8J7%Kb@$~c;JE2ya(5o zEiBR5^16HSe+He!X>!kH>bA~jn^^MgqsgzW-)Du0Gc03Ho>Q}RZMuHO-gz(U_@C|O zz4wd%vT=ER<%-Ij30^Bj9Ivw;dam1ir8z=jFTJ_!g*9-cNsV;pZWwTvu`a1<)W_&3%}0WJ!X@8Mne$Z1?tkV`I$?s$Bwa>yV)K8u=>}9xm&WAZ0ntu?$ovV`aI4*y&{1Ual*T+{xeK%)IIev zE6AD4WtQ0jK@kn<;06B~8bfn~-o5&CLB-1V)0s&s6Q^-Dgf{;BekIT~UE@^X)1tDU L^%LeD|9=wz!0&Qf literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-57-### A. BRO.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-57-### A. BRO.jpg new file mode 100644 index 0000000000000000000000000000000000000000..accd6859e6d435e81953b7a3029eaabfa9b2ee16 GIT binary patch literal 1561 zcmbW!c{tR090%~*FL$LMQntTc9J%VmL{#8XbE(J~25p&7WCVT>7~DY31|k+Lj9hfWNVX z_cz!-To|EC1Ofp=;9D+`2t#S1DFFVYg&2dP}=O=C>ArjO z2{0>saKH3}2Up`&=iZ4PiA=tJg6tEVu}+&eXV&FYlm=t-40D`wlt0Zr-mEkI_Qt#M z)ZifBLEVB1oKO0>vosQ$ECA$2%IcFgJ4~*XV_PhH?JdM>Nly8m&Rr$uvx_q*7rX-n z%OO*wkHGBiCs*@L1VAsdF>7Z-PM#P+LqHhmtqV)Y-=Gg1q&Xu+`LA#ZM z8%j<_;?{L;_uSVsX!DAa_Ot-Q$gvsXR^XakY`UK)zb-a=nEye}X2cf166dguY=pAL zHEYimFIlts?f$5C^DMpb-4(kQmyJvpJi^SwmHhkod+8N5-|4Yu+MZwH-wkxC&)A_O zf6%(F7B0W)OzzF*avvQt%vP7A(gT9^eru7dOk4?)z_p?h>bIE&I@zV2CQ%~#Q}9wu zmboq@d-HJn&66f$#k|I~q@pn06Q1k$Eb;M?UQ0$D-M|#C=!agHO6Jg^*QWTjl9_5a zyWoLmXVDOs$INrTLGW`NVGzgWAl0NCB^8*!&lCwkOVY|J`9X;L=^p3ZQl zk|<%7w5mJB#hq&c;G7{llu+)m8juy$K746rR%Y0wQUIlc0g>!*QwY3qv;hdHQk55 z`_}d03o@-z$0XqL$sT(g5$w7I?a`|f?K1R@h`=Sdd`?s5tq?{q{?o8^;cB6)+&DDN d+b$wjQHGS$ZwG^E^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}njD3@LL0BMKsqu8EC3U$d{I(C5NYZuyB$ zA3SAG+U>hK*K@~~MeAvm9hON8Y?K~|7H$8Y-d_kZp?VOmSEOse3duEX8l;SOhzYl+S>3@5v*saWUjm7TD7AMZ?rfhFFd9N*a?BY2~ zajxfQcwg_{8*%a7l0BCe%u3>E5lylZTGCW6lh$!%y1CBPUGM8Jmb2VDvp44H(>q~O za}P(xX?Y!yd@zZ9qIp6QS0A6@tQVbJOC~p1s%2-^Q!K%ruXMpKYbYBba74Q#JLN1eeRjR)IaTY{`&ZXnzz2Q zs&&kN9I((s-pM|yZ29kAZ{Iy{R1NGe9+&^|A!>D5`qig9nyk4a7i^!NFe!4TbOFO8 z13tq!rEByazWcl6&)3V(e!qU=cWmDBQqg5|O(s2PS{hot(}tCKr>W=L1EAAM9&KH-Y5CpE^(hn9UoQF>e74LqVoM44_P0WuOX3}x zRqF3YyIO`6?Vfb)ns?kz=AzU0(60{iiZojMV1HD)Dz~mv6GoO|Llqd*_|~i@7e%QT1CCRK9=x-PQG%q{32r!>l{E zte&wpMDm_d_vF4kp&ogr77BeL7q6&fiPm45zChy3+A8hLpglXM%=*1QV4$DW#x^KckIizW@H4~RN3$={C#tC=6{BJ#k+r9Tz9?z;9)|Ef?Up;xNJ!|Tz$m6w{ z>+h`*>n@a$jJST5bAw0uy#3Q3M*U~FRlffV>+b&y`zQWq_!ug>aL@F(?CF=EZEy3g z?Tfwebjs5z;TyBGEv`<{Elt1qZ`JJ^a@*8?=dJ$F;86OX;lR@W3|wjRWxs#hdiBh= lrhA%;eHu0D=N1_s-zYydJ23jUc*&jF+n4_o%{Tsk696Y^f0+OP literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-59-Financial_.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-59-Financial_.jpg new file mode 100644 index 0000000000000000000000000000000000000000..0611900d4ef770cb26eb265099e6ca8df94995c5 GIT binary patch literal 1462 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyFAvq*pATR`FP6i~hzRFADfSDXrMD zeo5`3=-rETH}5c5$#O#CKLf+pZ~qxyhZtrRM)`b-$-HisS@2-XrG#~F7|avq?qyh4 z8Sc5{%lp42KPR8b4Gekn_FeM4N7=Lc4NlzfxZdI_Yxw?E*3(?ETf4(FSslwoK09x_ z_vH4z;{6GEwgy>M=XYPzU8~2#nm)}{dTwIJzJsrBybKNs=&Ajl`knLk%HSa0OuOl& z<%-iJR1!^=7BHqB+Q_`%ynJuub@et|xsnO_#tXYFj;r2mV-U?QvFe-WDso9;6i9V8wU6n`$T`WQ1+y7@`ls@n2xr=KrdvezWmcm2lu z(%B(>e-2zcvRXk<`<(*MoLLuFmG7D(W~Ab`>7;Q}m`1YR5x(WjTRqb&S-;4Y@I)4c zG#P~SEoi=Q-AC3`o#pFS&8TlW;g-7Zzojj&`1UC>M)xqA)^~4dw+E~W&$r8;+n)DM z?9#dH_TL*7i+}K3tDCsYVOHqQ9osvfmwr}BbMQ?2+5dUX>y>_90%h6J-(LQ>+ZDB2 zX-n{2^TQpw%bO~11vFmVEVX<8_QYND3%2iaXnVPpQ*W`6lCR9F#nv5?F^oOJmsZ~G zSo76;t??V980WgxI)%=yw;UAvRT#gATsACcjPAbe8}v*&|DKY!N<>V#wO>Tnqf)8w zp?~k1?47!6->=B8N>A@PJ&DLxd#9P3Qr6z%VoKRY-%FS^?u2)?E9 z4FVhY)+N*{UEO23Znb^u-re!u{~2PB?Ve>mZDM?sXRbxENHdSfgU7E!_HW4j&+z=! ze}?}IO!xm}2X(!fD^*-RX}9i@bM?Mky6gq+p4fEiWv3ufbsEDr7hDPw?;mia=KjE<>BQsJJJ~Xld|qEZC%<> dx^=sU^QVJs#$g$YTU!#QELzv-s=D(3O#su)Z}|WK literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-6-CHATBOTS E.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-6-CHATBOTS E.jpg new file mode 100644 index 0000000000000000000000000000000000000000..7c82da17afa70bd550be5babdf4de4d96b4cd6d7 GIT binary patch literal 4146 zcmbW4XHe7Yw#NS<6d@oYy-P0w0)iANA_7tcq$Yy&A|lPut4LQ6=|Or4y|@ikr38rJ z1~HT%H3?l2g0vv*viG@j&&)mN&fI&SHP8EDy=&fCzgb__dp3232QC@t>gxg^5CDMA z12~%jw1ErM)HKvo7ief`XlXCd(X%knLm>2=mzfz^c)9rbc)56Zt_aJBT@jQL;^7fj zl8};>Q&3Rg7gJGJmRFOxt|0%{BOqE@T6zdQ2Ll6#ya11Y{Qp^J-2l@Cz#pgrg9HHz zCJ>kjbk+y(pZ7@x`dh$%3Pb^>q@t#|KuZTXFF;)aC_rE^1tpk@ijwlYI^z5upk$)D zEFh;v&1~jKBk0E>ACq2kLFjgK53BiClJIpW|7WyxZ0sDITq2@k;u4YyiZ_&$RaCX_ z=;-R*)i=0rVQFP;V+(b5admU|@O%{T6do8991I5bML|gkrlk3c3q%ogZZH!im4F=eWi2xrM?Ypk`IrkVx6@0S zduWBOo0C|b{J+w%2`dmpw*Nx=o9urF_U!*f_FrKC#?!Zsr$c7|6d*=R`GQ32dMY2<-4%L<_2k*N6vp|4 z5xW@I>!jaHSNznP3>k0<*!yX*K#WS(6?(R!e1k-9h99M-ITqqt&BjK#+HKsA*Rbub zx$k+9z5ILh!*%zqbjb?dgO%nOYjUtwaxDh;gG@XW;=?yB-Bm<7Ovy+E5B6# zTLLn4=fv4KM{iPQ_b_2gdP%Mo??LgB#q1W0g`BQz!{ zN(V05!9K@Xgx#jm?vB_6ANEdov z=H#=kMU-mtqHgPWIEav+Cqmea9F|;trChBcPiEF7g^a=U^)H!l2fKYS3JV*h91Xui z+L`4PA&nHgi27m_Ww7>C^AZ`>*V+WrE789EX&$ELAHb>3Xt4N^F_>L{yI`f{qz)Sr z_^H#9o9E51D9P#*L?JslJ)&afZsZr1xH6URAYx2LaM3*k#;ODL6y)%#YVV2_FDh$z zqhGrCwL4Fi&T!Bfz`W_8`U#=f{XWhGfPZQHVPY)n8OY_`H6nB*-jU~|>^avQeXn?R zr53N4s)IR4cx?cqN`?OdsUT3#;CbAh`;8Gw1<%%;4k8($&@S(i^`=Rf^NT$ zAH&aWR(n-UrPPL%YYV4u&Y=TvjVz$$$qvhtgX7R|!%9bq#ySD+e(t`L2Xboijq`dO zf*tVf)nugOl08R`GVNtTWY#)DX-3iTF3Q-jCwpO9ivsm?=5bQ=ny=o&fvoQBdT8OWJw{Z8TS58qd9|DZ|HaSKV{V-B~wq^+a#&d#^mW3Q`{QMum7fvdJRb zR8;DQ1JE562MN2MBiw#-+ZIb<2jN#{^)XM>C1Jd2Twz+V{ZB{3HB#?Epk6rWT>Tqs zDNA#+)WXAtmYJRoL5}Ov166+%6uPz!rMKMNs|;@FhK7rM{aLQd$h#DyP4CyQ(2EhO zAhRt%3+i`=pFnD#3m{mHE*E@SedBkCgGabGdKU7lipB04rUb+CQ;5&6r=kNjP%8-P zc0YpgBXz40;gGVUP7Fl>k&%}i+h=`WA6{kD^r=f)8WXW?HTy6y){(p5lE1=j%mG)15yz8zIkV#ni7NlcdzZE{b7m zpMMOMLt#qC%_g$LTkGUe%FxQ308hfO$%uk$Hrhg1RAcS&tElbiB{k2*pN9P;7;Iq% z!wvJj%d^r+AWn?$mJu!9rS7YgT=Kf%fZyVWW02d{3e#z^Pb#i0+fCQ{n>Oa1P|3<` z@CR=nGs6Ro7!=s=vi7!2a;(RaCt&p<(QB6( z%1==!#-#TuNWHh8(N6A=_^<>*7riMsJW9~6GAWRn&+t~A7JiXKXs6+i-q!ul<XlfMkm{WafijAULtXWDFLApe8YEM#@CH9C9e z!8(@Q!ZPmc#4mTc6;%0iVnna(yFL5F#@07B4<>ZoXu&7+k6WqUU(3?dd+3prTvuzz zDU}!`xr2BYU^{Slpp%;W#7_tChku9q43fa+K3faLCw%W)K0pEH{ou@)&)>B*GfZii z2(WKb@J?IDBsU+$0PY5@Mr(40I_6m9t)AfLrFV~dSnx7o)`(!vE8ezHLUK(9KDB)~ z^h=*%7N0oWM#)KJaMMfTZ8{@5S18A(EzS^`_jpPjV=&l;@>rx{w*+d63+Fd?0 zVwU86+E~8EJYU4D?Nw3r0Id+|%+S5d9j|KX?L}lvcrH3;bp}|!B4LMR`i|N8Cksm@ z$j)f%Q0n>{b56aUY}6qeW(S9fa#?EIrr%Ymv+8XbtkF+~67sxrH)V^oe{PwrA#&p0 zLE86(BK%Zm1!y0o>=dh8=AQxeRtk))Ylz5ZFE7F{i;fzW^2ldE%x6 z;yU6=0zQeA@RjCs-@E#F7Bt{*a&<~AVEZyJzd0&-M+fyG+2Upv?qad?u~)h#ivq;o zeylg06Ewu>^64R(rMqwRn7@`BkJmxo!Y=(finKfs?PTp7J>{{3dS+*c>~~2O+s&qR z((OuWYGT{G9}<0%;I=V_&F*v=E!?hCrR|D^+E4%J%#P?uVIC~jT}YL>n762IK9x$j zVLj`f{E$AlVGE5LGLexgT1HHpSmHjwHCZ;!fOpcn4_RRi4qYhA^aQOd;MhYPBGp73 zJf`~6}14BJ!nq?uC|QyNbD9?F*N z1wgugs9E3Va0aKWxVZNtG~z>iBgtwKyK4Ka$MK`IK&Hz&_&Ki-iPLME0WV#Hr9d)R$ABx(6D9*meK@yDU0xx!NBphR)b zE7D8VgWv^h8@HXoqIu`*n^!$$cyo`NWxOPAua%E1K4{0Fe`v3OKQ0k72K+cQm_2B~ zBAKfir8VT7KgS|@*Y&HQxfNmY-f81>l^Rkyd~vQ7gOxL?FmcmgFK1Knmi;zH^`s%M zs&G^Kbp7RA-W;J=?E9_z4KXTc=R>X?qcnpplQj+2`8#hVUT!%Ffg8X$ zir|{t>}lH2d9Eb)ZJGQ_P8CSAB7;x$X>_-qgj-hj(o!MK(v2?WFY+GDXb3Kd6C5b6 z&^&f)<64irGPpuMu=Sovn0^2z;}rki&-mibmrAy2mzHgJ$m%WbEJxo`oYMyCiUCcZmhh1klW@f#O7_Hiv*fCd#B39=czce4t+L&?ws8>b2Y!-RB&%_{QR(hzX@%e4`^&s}j1)G+jyi8$nzv z?GAfx&uZ*N7)c6hc(iB|0~z`?65`-DCfekqri1r^T+1T-RQ{e*RzHWkTI6^J;2Ids zw)x6x)&p;_n{UPs8tCjQ<4-L!47$U_Vo)dd`a+|UkRRS$YU#8R;kZLH8p*n(VL1HM zInL+>jE?;cP9NXM5!EZEzLn>-pDTlxQ>C7cY_Ag8c4_Jrz&^&0jdc#R2mUTPMae3; zzC$p^?m!alRhZ4p;yHxk8#Nn^Mwjhi-Su9zc6YL0+l9O~7LKuDDlE#YQOeP{Ksp09 zH|iFwJ9&~^GDEaU;G&Ue6}xNb>4vuun{S3v&Rcg`em<;j$hjMBz6g!*69OG#gghS`}=!POqRj-XAT=8edgty+m)tF3f8}by zwd-Z&w<>N|-l?j7@N3<}M~|P>|Msj!B5iGZ-rn)5=k=T3x9|G;2Zld=9#JSq$HrAt zE(D;yvb5({*gsrEtxE@uMxk+2E<|UicB6=BjI|5a(2tGdZZooRO~#X!X5^OE5@x$` zhKwV&H|x%@rM|Koo}zt`{X5u!|0Vki_P478=%Wzr=b?yz21ZrPD27v`v2VR?H$ty| zbo4`QFq1_Ay1fc-q5Oi!r*JO@mo%@rc#t}X)cX8*@!);#hO)d{Ql+MvaMv7yjwJHu zi7gNaLMJzAT5duh?A$t`s(H=s=vOkI8E%%>&$%~FZ|Ubi5EBN0_(F|Z6+V=^s40AP zox7!RZ_&px^RtngV|W1s^<4hr>=$|hVytF`%j0U(@={;1Y~)7z^&MHo`PG6;f&%;K zu@nvCJp`&A$?&EA_5PE~w=5EfjUmvDNfo>auYsW0d6l%pzO1FPlUd?j#xU8KT$PHV zU5E+oriJ_Xe4x+jP;XqkF#h4rK?iB!%{!v9b9#URJmLe7gOx6JL^Z?EcH= za$M%|^ZTCd?y6J^l{~+~u7`6 z&6G2GnNrvm|Kg(W2z4TSIE3J7Lp3l-xy%@Kze)(~XyA<*2Z@Gp9DEN+vr zM0ndzM3bi{Rtm>L2Mh<66&48|EDJkFD<2j1th}4V-#hrBkYf^~;m#8G_~p)wyy%r{Gg!9s+kfxoM_wW`v*yrm|v)c$x`fLQrn zA!9v|EkN}-QZz;g1U?jv1_DI^1cyWdaT)|QHELdO@KB&~&s~M=yhx&XxLSqh(Z)zp zI@_epjyxitqA(yJ?AMO2k#>#+DyhEb)es4#ikP}To!YT}iJovZ|6FUXBlA?NX75?g ISt+pn4+o`&fdBvi literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-61-Company Bo.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-61-Company Bo.jpg new file mode 100644 index 0000000000000000000000000000000000000000..192b617ec30d742a08e2f937a4a6eda10d1e8654 GIT binary patch literal 1474 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrtD<&lFN&Y4BABnpQufo==jmy2y#uGW zgy~4>O-pDv(Ie-=lJ@Xma{u>L!3#4T4~gVHNzuP0Z}BT{&6HPLtM&K)V>C}&D>}tQ zS7WC`Pef__7CZNUwtgRe>MpoBO=a#}&y%k*k1g5xGi7@0+;7__y|dnZb}=f#=BWVB~`vKZA{y@>q^d3fy{i}IVZxaeU!_C45tR!>pa@H z-eSd$ubJQLC(M>L>lRydP268wvCHHELvfQt!n}6n6>o!1E;_2VZPv`G&3b%i_!bCW zUm;%NreBzJ+qLU<`0Rklh%K}C#W42XIOiXHxuGiPNBKVo zvHExQ!I5zx--YbIRcD0@DvKKJeLroDd%zxbkH(9;9;+$N>%BhFWL>(NbbN~boxJN}X9EBjjCWjpp<)%(iPD0Z$>B*)F>?9$G}x95!C1{_@) zf9KX=)oUvirz}Vd&KKnJc*3*RH|u&>)q%}BcUN~TYuJc7^1r|A{}URrYxlfY v-ws7=xqYP~{pZZb3^@s*yMH(T2>U9s)Ga8i=#N0;zJ+H+0%KX_|K9`v3Eg}u literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-62-AARTI SHAH.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-62-AARTI SHAH.jpg new file mode 100644 index 0000000000000000000000000000000000000000..6e5cc1860f5eb3a31d69dcbd928815a40bb44158 GIT binary patch literal 1601 zcmbW!c{tR090%~JL;6bgmGxOw10d~jY~xR@Z4Ur0h+@{oi$8Z9MzL|#fp8G}YEXelZm zRaIA4mz3ARYvc5esHx))hJau&7@QX_%Eu>)lSWJ9{&QIM0D>FP05riM8Gr);0wX}I zMnICiCnxA@fIkD`07E#rpxiJXUUoyd0KfqPgE=5zPEH7f-5tb!2OtPeL1|TeE+o+o zDnk*%g*?dM#+)pEjWX?Al~uF%3Wf0qi-?Me%N>?iP{gWhXliNe=$<-lU}$v4*yOUA zImyD(%G$xv$@v=D#nsz~`jfApf57cKcf-OXB5AR4591ROA0<7`WMpOMEb8 z-(df6A=oYs2m}m)9=JdpzU%`=KscpUxdiozP&*1z1{cCDbn-z)@oN}H&2$xI@72j8 zEUW%eZtZ~fRrc>-q5n(v7wm7>AOHu0*q;YR00dxXPa*S`m~wFg(pafLhF+>#k4k0% zFjV9xh6v7x1>gloJ_PHTxeN4`=f_fNx>^b%-WxrnPwtH)m@l1NK~250aZ7cB$FJ@2 z$JaU!^Hnk1?p3#YS~T|1rtcO81`O?$tW2KH+LOR%PnB;TcQ1R3^=W+)-mz!46`9|# zpRzW;2ERVKqCc_Jr+jbaX zn)hRhIw(Zl2ca)g?+n`9x6&@{TUl~EcC*9a-DwseFo+5tH(IbHj?V2QSco7Okf*3h zXF4;7OCp*tbcQg`nJ?5P2L+p!7l#>3ua~N|m@%Wy?051DI%CxZ?(oOp1n5;R)N)Y)XA>--Pp~L zb2}nbBQ=MT4!tB(+jcj1ErS^uv82VUafQ8!!!#<^K0^zw=YiVS*om%| zColS3h*Wk*)5JvW9@0aVcmjE&vG_gFC7a|nztAne!V?agZ2|^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}niJ?>=!$@LU&Sjd@V6Q_WyKYsLgdl?PlW zE+_xHxWj78W33`xfiEuqzF*z-*Dx|+WMNK{D-f zs}8rVze1XmgoIS zTD1MzymY;5;+z==wuDPaXjx5@4XLWSwEdTPXUG@t(?xdc+U-yFKlQslqgV6og@^AV zbFa(4m65+}+Os_9(tPv7wYuNy#C}dL-&u95bl29dhs-QfnO7t>HyNKz@wAS7eSJ5Z z%SNXs)qfuUWQ_LcmdWZeSY^PzV9j0w?%TUxe`4$k7j&r&H8TFOF>BY#?Hu>xfAIhL z&mg$3ZENb?TQO00w=G=0qO5@B+mi=g>`we?@@umKqeGT0TXXE{8SCFW)|LFW4U=ii zFIaQ1P14a?M({?@{8MXU|1&JfzGn6}jHS9=?mt6m{p`BX-cr%ir&k{onbcjGe&eHS z?z1^tvOHfIy6$?V^zvzR_t9w{*Njzm)iH3-Dc;?;Y|r&5zp%nr>$ko9^>vf+&Xp#Y zUkiC=XZVC=9aYF)ZMN{{boEb6+mc*eDo*DHF@o;sde$g+7!n|bTq$*oRqmyCPTRz^no>i$w) zx^|I)Te9&jpGGU5nzJV@|7tASQ*~v*M#HYx>mQw2GTpRqz5h0|sM4*g4U&1XS{q(B Tc5QYQeSJ(z)3wCZ{{Kw?i=RX^ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-64-Dr. Shah b.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-64-Dr. Shah b.jpg new file mode 100644 index 0000000000000000000000000000000000000000..9b8987d5bb025db90bee55b90d7eeed620a53e91 GIT binary patch literal 1457 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}njD3@LL0BMKsqu8EC3U$d{I(C5NYZuyB$ zA3SAG+U>hK*K@~~MeAvm9hON8Y?K~|7H$8Y-d_kZp?VOmSEOse3duEX8l;SOhzYl+S>3@5v*saWUjm7TD7AMZ?rfhFFd9N*a?BY2~ zajxfQcwg_{8*%a7l0BCe%u3>E5lylZTGCW6lh$!%y1CBPUGM8Jmb2VDvp44H(>q~O za}P(xX?Y!yd@zZ9qIp6QS0A6@tQVbJOC~p1s%2-^Q!K%ruXMpKYbYBba74Q#JLN1eeRjR)IaTY{`&ZXnzz2Q zs&&kN9I((s-pM|yZ29kAZ{Iy{R1NGe9+&^|A!>D5`qig9nyk4a7i^!NFe!4TbOFO8 z13tq!rEByazWcl6&)3V(e!qU=cWmDBQqg5|O(s2PS{hot(}tCKr>W=L1EAAM9&KH-Y5CpE^(hn9UoQF>e74LqVoM44_P0WuOX3}x zRqF3YyIO`6?Vfb)ns?kz=AzU0(60{iiZojMV1HD)Dz~mv6GoO|Llqd*_|~i@7e%QT1CCRK9=x-PQG%q{32r!>l{E zte&wpMDm_d_vF4kp&ogr77BeL7q6&fiPm45zChy3+A8hLpglXM%=*1QV4$DW#x^KckIizW@H4~RN3$={C#tC=6{BJ#k+r9trxk|ozH(qcsE~| z+3o8r4fQ_Pyng)l59QK%=fKQRv;T2SUdf9`(ks=g>(*~ldg&cj`t?-g@!HJw_tuDY z7s^ORTtCaX!J~ZM{^<{W{xh^TCjMuzn{NN({g2CEYU}QYeKa=R6Ld^v+pUKmPXPxnt%%*Gq8LqwF+j!Efw}0EK{a@JQ|8T{BSeE(metmXq%=rx6N3%>$2(|yb cu542+_%m#)=*>5=my7c5&3_WR=luVh0IeN<0{{R3 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-65-Industry &.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-65-Industry &.jpg new file mode 100644 index 0000000000000000000000000000000000000000..f3a246d36597a67fb0fe46e3ebf118b9e49fc4ec GIT binary patch literal 1432 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrE1u1hF1c^AQYc+|Zd1=cli8lJ4Y#XT z=rzw!3IB9w@#7y6rEH6~%zETK`{+!r<1!1AtP6LRHP5f}Tot})N%4`AtL<;6KhK|k zwQPmT#^W*kFMW142VKs+_sQ1Dd%Z^fB*}>m1y*krc;wIRjLp=&`m=o7{i|CqJ-KtW z?-gfW>h?a@%p`_g{SVLkT)FCWFW2mnXv!oGvB%#J7`N(0&-r|9{dHH%*sQC6+L~tS z82Zgt?Ny0+&bOeeHhIsl-mibJm=zz+m0ii}+fCfbSnCo%5?%@xffC>T_fGme=REzs$CI zwZiY|1m;gqJ=xhWcOAQ+V{$v!y>#2s++vl&85fmZgL|4MIsV*Ly3VS1r~A!6#`mMk zZePt@$bWj}KAqf}6^k>J3x!&JUvV!C%=Xo+PPJB_6 z6|42Gx^+j_^Tc_#f`vEnXn){PD0}c_-Hfa4_F8e_Uo=gtr9@AAY?oTIzB)6}NTp@r zH0k@==R#GO+-z4Dht2cOReKdNb(W(fhe*mh8~!U-xn<-1H*MIyEq==*jjQ&LulW_& z-(6FvCp2rhWn|P}yXZehwd2mLHlDdfyIW;b2J4)SYp$L6ecE-xarS*~^-;B7@6E5< z7VxrPWV^SKNu0+sRhB$;SvDhsEMe`eI5lSfT|2h-d6>rt3cCIDF&Gx zB%|(!?G_eQE!y~Co5lIlF&l$t)vS7*{c_W-TkHC_J$dy_J6>|D_D|oXQ-{li$Rn}~K z=UApP&*6_f*R)(O&aZo|s#AC9PHXcy@cPq+2T%IWus;v=TVi|bRMxxeQKdiQ{o@*D z-FxorDiXdbD`Cm^uv=%p9)3AJnEkeluiTxSz;}mNs$5|K z+|{(bpSNGhOme}RkH^bpwp}_XKJPr&$)2})9uMZVo33{iz3gPO<%(C7^8cFvff;2F literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-66-Governance.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-66-Governance.jpg new file mode 100644 index 0000000000000000000000000000000000000000..0b949b8d1b36b5087adada9af3d7950124d6beaa GIT binary patch literal 1580 zcmbV}doksN-krF!TfgAo^8+mwY%Th`+lG2d_U)bd*NZA;Nnbk z1`q@QBv}9+0FD4HBZHEWMx#(D3ze3k zXc%tZV6w@aLZJ{0tZl4F+cuIZq}3)627{58lUKoFRY*D-I;8(BxE|opzyeqzAYCAZ zgAh0fZUzKNPHE_iz@LJo5J+hm6dEHdCoxni04WH8kU}D)rIAR9`<&z-AaT-)I_3^C zYh90`bi?qZ_>3I19<{7N$*psmNcIcAh>=~VtfH#6Uf;lQgAv8Ta?U8x#jVb*597~-o@j0_w@GlkBq*4^LFgr`1=ntf?46*{O5&5(JB`N z5MQw*=U1?QaN#6eQb;5MiCX1?q@pDofkR5`n9C?SxT21Rt<@#Pqw&;?oU#Ut9@%YL z$uGQ9b{&y2vVLY2?F-qz1H1UY$o>NR#?=qx5s+kf2pq5jpX2T=?J}(ku1(Qg92nc= zGkOmOA@A*kij^>^%PsjhJ?LCqJa7}kO`*E^o2V$*tf}1oZXb)tjuA2(Q`1TMKH^h8 z9>%=XLATBrW`l2tV`eRLDVhd@JJ%nqtl*j7nsIIUVK8vOqQr}K%N~T_8i|LM0E4sH zVk54I|Gn8G7*r{fDKMx$oh`!9VKCIWV}3GIQe%90Nn}j^+zz*%!QDBTG#~J7BFD3fhOgQ%e)(BV{ZU&}v5&gH zEZaLVNmh8ikO+gQl<^$5NS$_~yEo%S5AzLF#O#rz(4$2nXZyXAwYk@lRF$=_T79|@ z6gjjAf(dhzFlgnmUo{*pY@tk0M?9#h9O4f^bI)#`FD+uLu~~b<;B#>Y&u*NME6HyP zgXS)+Zh=V*Csr}=?nL@~y5+43WiLV?qp)Y`x0NrhQOfbPVN5QX2E^CcSH8e9Sa^==rQl!HphIU+4kD4qNvgPOuwMRLTrU( zR>ord<-mf2zKwcSm8=ng?tmI=TSR)}HO5}jo`ESCT+k@RF$e-|sAwSz2F@1vzh9KP zSYr8eIM+VK6LkW>v-|iG1{pC3@OiDf#HtLpnUL#k9QOVau5C2@Tw%(mcUKlTf=^? zpEgD@Q^%+*H)Zga+&TTFpPF%lPPwgZMl5;#)Z@p}<~BUXHpXr)5Pb7%4=|Oco^(jX9?TBYH zFt}nipH=_^ds^MpGV`^EvXZHH(8*eVF?|txfU;&$rh4V<9vI}QrJTQvoEJs4C8xSk z%B`vO!(_HKfo`1l-K_%rl5So(c90t@N UT5WKMtCN!OlD(@F9)kP+0E{2LdjJ3c literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-67-Emerging T.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-67-Emerging T.jpg new file mode 100644 index 0000000000000000000000000000000000000000..c444bb104c22d23882d2471d7dda9f9dc031819c GIT binary patch literal 1449 zcmbW!X;70#7zglu$sIyVxWPaGBe&Rugi{Sl6Nq%`1QZl&DF_0BwJ3tdVv1G;wF=@% z1r$L=IpmNdU?WOVEEpj~NW!5YN`%NEhk(i3pgm^#sqKH~*)RLd?(>_4yWnR)cMY4v z1`q@Qq*(x#10R4zqcLa{7K6dya9BLSK!>2MO)%Bd|J1tAI{t@;M7g6KVLLw1J%z_KjO41a9h(uXY(R!=6n6P+#Yg!7{V0FgjvIg8@mw++D z@SSaVBMSZblJN!F2id=a9r$0ezhHm6WPpHxG@pkc0w$Q5V@I-_S`59LDbFAuLDTU! z4eQw)TYw*&^2$|ayH13t)!gneC}Un!YZ3vhLdU?58*zn$L> z1Ag%PXtnqj407c==I4Y1T$y5;-L4m>Y__YP^${xqU=Xzd2BK`CYA$r_vS(}P+D3-C z;b75d`J(d?aZ%g+a4HAhgR{Nbd}56H3+e-bX+@d0sC%;b=#5>cO9}-1JpOf?t?!Pg zS;H`x>+cxPQ*82m@4wxXFEWGyJ}Q;}B2);2lBHitN^Q!;)pB;JS2@dMOG<4jf|(r| zn6d8k)zXOi`83&F`fEW8Z-bxJk;gS&-p5JE0B_XEuu~|e(ya6SyNxh-#)Uz@qkV$P zcVuZ?!uZQp2M*)6Zbuj0yP3!LediU-4dis(w2{Ze6cqOfIxn4(9&+M!Wc_rfuSip!?R$HwRk_pK2G2}&R8Pn@%USyQ{+857U+Hoz4z^du(wMWRZfTuVp$&yI*qOlhmI>^U=W(l9}3UzC(jy}!eIE=I$lfa)2fusocRb6 zLfutZTn~eLLau+IUP&1YaK!KkwWyb$=Bkt`W-0{5evDFTebp^#X7$cr;^oUFEb*vs znvS$~ZN=!)!P(J^aZRqtz4ek>r`tvu7Z_VvlyqT~Pl_VzynRTBf0BfM_la(Itgu%s z96bjEGrtPW6O1OF>V#%+#s2$+FnAIv95c1IhCxQwfGu+<9tJ}@)<^B>VUz~dt!z!X z;=~o{vm6|o?h}e=*rFAIYsAVem1lb&Ve^gCL9+>F^k`p^DjfH!^5n>P8s)^EL1~Ti z>RYy%lg04{19on)WzEq+n=uxap4W0N)+Xl!x6Zf@B=V9JL6o+q98-1A{<<{xO@~1; z%1tS|rh-IjWC8hG33KQZ{@4;n%5_KCvf$(bVfV5GW?7*6fZ0hkib3(Qu&E(Wz9Op+ zPYVC+wXdS~a&C(D?uwn>9j(W(_~!fcmVL?5Gah-Da#XodJ?Lzm5WBoOZM^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyD-EhwlILz5Cy9k=j-F|5YFT&rsX- z{L5-zFKWSdyB)umIhp)(7OWbh5^}<%OX-p5>9*R$!-SI1M!LfY!j5n6`YT%c9wyso} zTg~iE$c9U~U*maSuD15rt@Pm74HdyAyOyxtd)?WU#VnU3MUF7Fh8;-YbFlANFxm&(nEsuL8S5!IwRYGK>&6c-wm((0z`Db<0 zq;4;n&fWUHCfEJ7qmJlZ)7{sf^J~#7mXpt#`0n1%M;8Rw>*Rh#qe*sapPFXIhPY+|ZRI6nD& z%(W>`_O2@PuKTrrr{dbP;jte#J(pxj2~(T9D(dAl!-y;)-8J7%Kb@$~c;JE2ya(5o zEiBR5^16HSe+He!X>!kH>bA~jn^^MgqsgzW-)Du0Gc03Ho>Q}RZMuHO-gz(U_@C|O zz4wd%vT=ER<%-Ij30^Bj9Ivw;dam1ir8z=jFTJ_!g*9-cNsV;pZWwTvu`a1<)W_&3%}0WJ!X@8Mne$Z1?tk0z^udw|K%dS1+J$m-+)qnGK zs#2fi^4vG}5SNw^xGV8n{?Pi@PGuL@O}b@PIhRXoeO!h5C)2tvum0UCPyW9Dv|&w( z>q<`^O`(ii4K4u-a{ZeBv4(E--o5&?!-SXSr;R)&xkackhOz&?9~jFjveYT4yL^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}ni(j~+=!$Ynl>ch@qRA!Kukia^Z9#>+bL5rTy;I7Qa*FO}h3_}*o#o8=Qun69_8pGrpSfBbU$E?sE%*P(H$bY`JGSm3XTD$p0PoHm77LwcKR`D>R43+D;qCJ(c0Qj@OhALc_dX`$DqnENde~RDg zbMO8$Xg+7Yeno21e}=#E5trBP(w?@h_}_<9|5pFH+UxiHO;qgJLw}$2Og;X6cioOP zOEvHLo;qb0Zo_HU6^1O6i|Dn?oQJ3 zeOBMkdagLN{jkq==P5gts@&#Tt&w;W#=w{^|BhTBI{T(*DqezN1`#2ca7 z9{LNW-9A0-heYzHWc}qIR(bqq$i7h=m-FiM%%iLCyKi@YeKqFn=Aw6>9=&?x&1?O% z?Gf>#0ciYK|4It~tFy8~@(~0Bj;uga7~l literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-7-ROBOTAXI S.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-7-ROBOTAXI S.jpg new file mode 100644 index 0000000000000000000000000000000000000000..a90a72c202511c532bdcc974ea3573879c6e806b GIT binary patch literal 3244 zcmbW%XH=6*9tQ9?gb+#)5KyT>dX0z_;nJ0ULsdF*1r#X?p-2@hfOG_eNXLTIfYPNz zq!*9Vbzy@R8>$F=L8H@v(9gMx2|gocGbh>4AR7@v@sk@+|)8~r3F_u2CoB`-_M z$}4K?>KhuHn%}f^c6Imk_Vo`8j^pqXgvpPeh_mFm`GrNw(((%R+xM;Qon6|GyX1&pzt5Lcd+p#Gx$Df{nW_x~^1KVbiK5djz&baZ)OE*1^3IOj6k&hh>_@8p4Ynwh5FN!H7i(CEw(qk{nh4WdImg}puYl` zbgb$^c%x=6P9QoI?JY#Km@n9Tm@T+94OY$bzq^M4uJ!7sn=I$)sikcWAaLqjH4{frz-bs@Lu07uYg zO%*y2uVTMHg;W5?$M>g?=the<8-z!9v3X2x2}N}IE>oOEhZqV=&^mLvJPjks!V-9d zkT1d@wn$FUlD!#jOs`_#ZsyXQMn>DVwzChncpg%dVv&R8>srcCVBw6b_ zLvC-;wTj;GgJ+6PhkJ|d0@g`g0sE4;xkp0B6l_sy8+}iT?}l&9E!M#5;w>Uz$gyeg zkI9OCMh639~)jEFU{Bo^= z-Jn3-ijm*&Bf%2|qzHDt<8+{BFT+a*l<^TBdd9;^krCCF74m7{s-F5Q%@>4!1~ch)7EJshGZl{bJ3V zy-GQ}BPZ3%e-oQYPMR{iJjL)v8v9^8XBH`Y)4e@4^oIh9#iN}1s=#I_n*HR=bLf63A02fq11L!-Vs5sDQ>>`w=KT@Q@WD_m^PB$p;W-|TJj zmiK*@XB}#R4wN2wTnZ1_s!^e;zCq*!6bj%J#Bp0wUU3VkhXwb3e%Lz%@M0l{L_;3;8!X;1GfSUF0tdhmG&rMTc=k0M3!o(Y3ZQ{?$zL9Jke?{Ycq zLaFZE+I$n5XQO<#e+v}YwiT@8u=LR&5F)0Ytgk=!l6W+G+0aB~18&{wC(z(UFhDKx9y%QQHm31=I^_~+5ds!pW?e$31K$L`uffbgAI^3k_JBYp^aD;TwLps27dAxH;0Qk-Z5 zvpyd)4e`EW5v|M$j23~X`bjW;>L31kHK;b~;?Wagp&9Odl-~;?L?L_Xin99I_O@R@ z(M!aG;j{qBCSNswpVC$u8l_M@Qiq_yM*8aF_Wa9Cp(9~If`_n+52+8V7JRjZ}|@@e-{>-d`OgsFBnFl z)}njcu#L)I1R2Z1-iNQm#!G3}<5v7?%^@}RL$E_$3ikAv#Q4S)8J=5LVuD25YYeHv z`FqY@JC9s%sqTIVLw5_jldKwaI~#ii^!2MVzEn;1g_k%c$Y14Cq;iJu2 zjxMUUNDMl~@uiOj7GcabR~?(FX|bBf;*})^=Y~@RYR{j3 zOnN}fb37j5GTAl46x#WIMVn^=J0KQt^QEB7)@qLjZKwk)I^HJq+~FXBViqVixEVQk zt_ixOLTQ2J#uF6C;|uACYyL^ENV(fcSMp-}tRI(lc8btM`!i-2r{rc*{&!trHF2ej z#u=*C8uM=P^EakseTZ=K+XvhIrmKf1^*`(;d!M3Ap09BArD+9Kef2Esh>)GzzX49( z3{oTRCMq9j;||9)A(KdNnHy1c&cRi@`NUxHf$ zpZXA>Y>l!&ed$>}&^H@!$Vob(ndyjSoJ@<`zv^Db9sghr{h7m&BK=aVkycVHnKZ{j zX?XVvO9zabP^0A(cv-b%dtg_<@(iaYlnyX2J`E5f1kL81&~njf-LK53rG-@uRD{_* z67Z3pI_F~2DVzE>1@MkM(^QY!I1_16&Y8Uz={Yd1y;b^BN6Z=VJVLUhZ2HR3VQm;& z6MoWa4jI)MEoCr@9vZLe9Sslr@HqvA2doTu{v))P87#K;1sMkFwl??KZ@AKu>?cII XT)ow}dv)TB#H|c~>;JM0ME~?Rhpi6y literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-70-#### Indep.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-70-#### Indep.jpg new file mode 100644 index 0000000000000000000000000000000000000000..46db6f4f3393f956d662c5a5e6432c578b2c6366 GIT binary patch literal 1166 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrYhvzXha0Ua7m9xQVP}7dJHIjCe)ClJ zKC5e`T1AnnWk>Gzt~z%yZBn(ST)5t$N1-gg*cJFsE&k8YQtb0yWXZ4Do%TO~3byFl z0abV`(NUjTG4nX%30dCaQ>A~`)Lz~E>d&6e-8(M_M#g{WeOdqcnE&tp4C?ulNPQ`y8l;3UtymG~yv{kCrv9;g+GbmKwxw>XA^FPN={~4~I zx&O;>rQO}tPUZSS_{%A`|{a#Nq_-%?fiyU^YGocqIwQ+sOH#Lhdv_|qr# znK3Uj*35sqe0!->Z_l=j<4i#ZAD`dNH@`QVdE&&_fR&Pe?603#a+ynG(VE1_D)v9k zPyaKlpSk}_a@f+;7xJI@r~liR_MhQmZ-n=ci(l#&%zHLv;*yIdMH(fAZ9zsHYtC}c z)K%Y-rlYH#A=)3bn6=)C{b88?SL<9m`?=fC+;QhBdRNP^?4{VGTQ`LZ`?vTi`)bHC z8@~L(!+&UUR_JuSm)6}Y>o)yocqntp{`oQcGyfTSMZz{O<@}RrAGl@S!~YB%_a^>l zurvIZ{wLeo?ZTFmZ`KU7`HTqOPaY^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nipkKWz7$b9F#h*p)Pj9$Iftc_c4DDB?6 zZ}$$@NhLcn!Yj^C>+WHEId6Y&YEZ7J#JeLmSMz*~aZ_mvnaJRH+b9J zs((&9SGp$q%Pn2zay{^#XSGOj$%HejB-R%#VLZQa`K#|txmwI3OA~iZQ^^#0(2`_S zyUF5DM(JCzTbpB31eb*PuGx_`C(rr8lK76%v|j_s;RchvP}zj1}wUGc_(p zmHyuU@ThKy)=PegE3xT)-``$x-+gvhsAPaUm&gO2c~f-MJwdCmyhu8J}6au_vMb_nZDx zbC#Rem8M;t^m4WL^IhEgmWKUKKK{hZbotFvlHg#rt@b+o%eRphY zys9#li!HC&M*gyCn8nu}k0UO;To}T@cKzkgYimWftkm?L_117p@vkjWdv=AMy5pI! zk>m7cjhdd~iSLhQWK1yK*PpYgeog9UmH!Nt`;vdnicP!r=)3o{Tldp`aS$}TJ!O$7z zTu(M0v%huzY1F*3o&Oo^%Itp#-2dZhbnSkS)%n|cN!xmPiAo-PnG-6FjsVK zSiq(#yR*#Kj?B1Tzjt~4ef@tX)6B|0W)XG!`&HG5`OY08=xO+5i9m literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-72-Financial_.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-72-Financial_.jpg new file mode 100644 index 0000000000000000000000000000000000000000..df281edd82fe1208dc01e5922a33fe9f10edf0d9 GIT binary patch literal 1496 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyFAvq-QLjmhm_#O8my35)1a9f+0PZ zGVL>`eS6`V5g}1ztf2Du4Bx)1EK*CiZh7}P;_l7vnT$r~_G{)a7&n!Dku8k*v~IHJ z%X|MBt`}AIuF{%2M?BN^$o1g1dpG)2ey>=({89DQQ@4zY!n2MlEb7u~(Bj_lF11B> z?V*Ee#XA?@Gkn?8HE+eeY%w$A%HT6DGwy^2&RFr)f}QzwU{F9$?f2BziPoi>S=MXU zmCSsfBIGL4qdGb7DqlF)kuu50A6G}c+2Fo7X3YVEi98u|y*ElQR6kpDYQy~-Rx3S| zW*ITQk`RnJ*K&RPtk$kLwV+dX+YOE0)`_IdlXxKKS{OQAWsdTCyZcRbyRyGe`f_>a z?r*s+x1-jCujE$oQsw3Qd$w_2k+$2axL0Mf>JAnv^89Bg74+Ys+tn=mJax+JW5RVyHN6bmUdLbF zeDBHWYQ@CLE%|)M7A99uo1CR!ue5tVvt4ufOK5AK1E>JExR@4+PCLtxW5TYs1$L} z`P{d6b?4Qm--Ea7&3-ZK{njs=Z|5C}xxCltW_EmjAODuc2NYA>0$w<(D+n3!)qBl$ zZFBKn`7E^eUAoavuY=oo7*EQcGs>DbefiC-i+8fZ!#4WlrKR;Q@tM&4H2y;R$CXK2 zo4r2rc;<3Bvp!KRTX5iByTU)0f1z6}cTSDpbZ^%8OTSCMzB0XY8yIjGO|;{0u3MO_ zyLk5^FSE^R#{I`u_R5tq@Fx9Xy!JrQwY~1p(XZlF%=q%AU&3K)N^aL$pZ#QW=9IH#CHJJNQq!WjGA5xWOLu*)HDi@E N<(SAksD6 zlum;5l1K|Jx!L>dbI#uHyXSrPop=AbcdvXSYmK$m9BYm_#~5?Y-D^fU~dY#=U9HVzJMK4E@t zo}0WJ90F1|Zr&1=kdOfJ%P2^T$qS21i2W*rh@71KGUa6!Dk>JSs~lIw{+|zg8$f#r zh$1c`CgK4u&=L{T65+c55TTwIiGDkP-w&b-#3UC<$u5yoP!e9Kz5-kzA|}2-LVWQe z2?^ot0K#*Cg!UrcRZ%5Udc7xPJgy94!SNq1@hX?MGd>tW^NBxodqGaY#LU9VcI`U9 zzzsnO$=g!WGO{XnRqv^(YiK^yH!w6Zeq>^0ZS&05&fdY@!_&(f=HnakGW1ngctm7E z;+wZg$?sB9vp!~j%E`^kFQ}-js;;T6t8eJ&?CS36?fcw6IyOErIrZ)P^b!)a{9|Qx zZG8jtb8r9P@CbW+@~d7%0P)|{BK-eNvHwFaT7q5|NJxlD$bQv}=z=$ah-pbKUKJ&! zQ_>@Q;!4jW7JP|8IsQX=J2|iT12p4Pw-E{^K8dAkm|s=UkEn0;o4U9VaK9+$Ic|1CVO6{1{UTFq@e0*p~+W*yv1&Czi8Ig z3vvmNga!&uYoH^VEcufU>rg)}h{m?fFQA4K*w4rAaK);+E>L;GX<;E9_cw4nrm!E4 za@(j^Jny2)W%v_6pFFKEjpBk}m|uDh((4TKR>dD&8K^;ueGlj_Aet4O3{o(OM9;KF zAZVMIT^(Ify42tCfXxtpl64KQi(;4DWfVV&drWmXDmEf@Jdl5^NocWAUD%WTn;~2= zP#=7u6Npn>o7FhWBc9AF1tn31@@9I+MEiX2{3sHJdJ%mim`Y)>!}~Pyohg_26hHw}hGkKA0#y@jW#}&J?&2 z$y{zjamTlJ1Ab7(Wkq3Z1TqZ)(pVfK`EIJPe2L93Dw@OpLyXuR;DU_I=Rylcj8vDO zC9}#0Sf!Fyl+PHfwNf|f>c$u9S!*i}uCrmX0zC`+#a%8(w%u+B9o7v44UI`#IVP3v zgD2hj$S;o$5WuFP2CX_2uc9YnQfqoRHJ>IPk0eRYheXrHund!Z8hLb*)R=8I1F4qO z0t?{*_igbazm^pFi8Q$xbLH<&&r6=v2jjNN>O3LJ(wXFIS&@LiuT*sI_^5i=JbEzquY zAft)`b>!DE&vVctTXS)Pie~E{p;en#Ela2r%^fvEb02wvB^(27sboW#PiHaM;fovK zAPiG3lDx@wsdm`2SyDnSgx#!Xue)JMYwySEwTdSHoJJ}wvr!KlmHZ7aFD2aWw5{<` zfS6Av<-XzVk3~vSp^}_hn$53VIg2V(<*6%#e<&;U^~q0>qYIHVhp{hA?_-`d+a+uJ z@aZ*wiSMje&Tr!&9{;r~&Aht;0yovM(ELqVA!Ift8%l!Nc73rSCLRa7o`a^YRs4kB z)R>g?tG54CE2b--c0T|q=1VjhB-;R%(!nrA6!wqUH#EOqjuiXs?2>RrF=#SA^Fky> zt~4T00uvh{QZ*FT;92>|xaPYEKd`$TOJxei;z&b?n?Tx$oSEhP9C|aH}`ngv@K&+$crbD zH$K8xy~!DaA1JtwF%StcCORTxX%^-KV#dtIoP;%c6B$RvSJ$W=hpO=at0S9tFO>I= z`CXN)LC2LLdcljQ6?9s(@keTb<3t<#hJk4%>ul}u9)l11_j`}Jn>Md6l#qAZ-6v9I z_?>KPDjkaSk5kCVeteLOGV)YiNRkE#5ki(Kj5>ngpRtkEgur#DY}ARV$2`IL&dUqN zBJ-|@20P+~yi#E-9K-);Qr^gCNrchf<-6nUJ~I}@O)HWYVm_T0E=lt=4&VWLI}`K1 znsLc=#ZmXZ7{537&LRc@hY>xYE*d9cCcEhija|XP4_fNiBfnN+6od2T&JzYq$4ZvfkZ)(>S)BXzK z05K}SG^@Up3tIyr!D?=r*vTJXDRNbPPwFMjs+c}B+SEcre`Z@hLf=(pKm9z_KR)VZ zwR-!r=;#;BUZ`H_uXe<9c160>o8D}Q(Cun-i}wA9X|b_(H?L#ZhR+&}g+lei$t#EmqW9^c^kW=cJR zX=Q7cJD4t7G(S z{eYc4K=+r1vIg_XBT)&lXp~@E$O#(;3<{l>LoYwT10D1zE$^{M-%+o>1?j5vOqyX{ zz}79M3f5~*dJ;Wz=3d<LVNef@@UvB?Q#t{=7`gk-|_?_KJ*0i zfVW~QotHo{lDaIWj%;e-SrG?Rl3i|nc|g3WTU?uKoc{jeN>h&I1tQElEUgHKU4nVk z7&Yp3#v^}tbJd$czu3;wFs~M(vIc`jznRk>8h5wNS}*Y+QXMkB1HP3^=P+75s|G7jiHm`lsQE2C%{`&mE%gb9#6L22Lt`b0vrIiT@j+K&>DSqRiQ}Hv&*_k z4<7I`Tp7n7d9<XzRHI?j_4pd_&Yd2SXsGIz_Imi^6gr z)J#+jtl#Dgq7SXdFKk|m~EsCD_OW9dytfi zf-q_o_F69?Yl+I#xfc^)ti;kBCzGqOI3|%ONk^!+<^Xdd8+Ki92 z3`7Hgqt9u=_66G6r?>koR+sU>X6)C>76Gig5grH)u_-l&Ic;{C&`t9_og{ND3*`=} zyyhcU8O^pPfe4X(I~5LC43xC)VcQ}Qy% z&0;ZE<~Nf+ZIHzEdhahUT0hh#@cQB?smlah6%WJ}LC_-04!dW(3GktQJg^}7v{%(ntLpui zB4+uEy?T+yzX+Q~hQc)Ok4R2bpZ4=eC5_(U1E)8>DREsWb~$nG$gaj3c`r6L zEAflV^(8BFHgd%Ed##oew@M!LtR}%4R53fHC}##zx^-Wn)M?|gACIDf>wIGm!f+(u zsX(UhGaC@neLryCq^Zxgb)>#i^ruDxvF>6-K>#9{} z&_iOb+~uL+sqx2;pVleRFU>aSc2k-5Q04RG!`ZoOt}Nzg!kK(dThce1GHEgtg)J=Y z@c@TE#5?EFdBq94u?c%195!$Ljzj;XCJaM{Y$^ZLUTUXjmum1nblJ^CU3xA^gKzP% z3z_`X2Uy6`mxhI|W>EJ)KHrVpA_r*nwW#f95I^~VU7d-A`(z2&p{0PGeAF`xbEyB_ zRKprfh{P{l)F1n3?GO{+%lDYBG41iy7x_!& zacY`tO-AUX6FkuU0Grmy#alOWz}j;iBx<<+{)!t>+v4)G59p%~Dph$SdnAMjn_&KI z=gX_wSTct)nFksQM5_R`9WnD#JuPln@%HPlQ%__J6x8?P@PM9LuNbHYM;f7HeX1m< zW?>x*>Tl3xJG+)>h7rI>g?Z+CKpzKP^B{Zv^0F=W3uabY?u@}_|{~(OUw{Hkr?Bq!JoJ# zok@KXDL|NR31m`3klttEQ}BaP$oXZXkg1V*!Qq826Ow5Moid5&;^t%*7ViC0MM9*I z@x%j4;qcvdJizwT^_y?Cpu3W}@p7rogSVsXXm6adZ|=FP?)fD`&v)L@rN?SL9GSg5 z+cv-o=|H`KgaD4b?G~5PI+BgtHGGpVco+~_fulK2rfkgXG~IGF?rA-A5Oobr`I!al zSV1k0E0Sq1Ob_KpiSV88g1yj?-O5^Y={*cCeGw1Pz*IVLY?H|G(rdMXLeHyKI?Zmi zJ?+e4$!PUkk=`|_!2@JMkFh!PBFz2JraiYBMO_W`%|-<&x;d~4Qlv92mc`QEwR8;e!TYQ|F&P&?e$p&p*u1eZ<2+Ns}@Y|&{JaAdnRq)>Ak(j9c+joVxX}GVr z#yX5dbr@zyV$YH0V7&~Fa^)$pUdnB;x8DlLPg)$SYq6@uW-vlkg=>*L!H zZ=Okw-F%xgJ=GVb;uB|VqIdq4E?wkVn* zOm@zU(1(_Eu-#4^pSdcx`L&T~QqCUj3g^xESe;j82fCpq`-+m|@+wP~41@(fr+T7W zEcH)v8R;#bWslxsz@?QkJ82Syy#6j}%y{}!^w*m<0V|~JK@a@msL6LDlL@Wg38KO_ zwkM~?R5Db~N_MuhK1z{~W7N4rmY6go!$Gm@4@M_AQ<-C5J)2_N-#COiSs1r?KHJ#x z_Mc5~L{wt0Aq*Ii<4DMTNsp|_@^)Zp1Kprxo-mZusrX8ZEq zU+-knfLlgNw-1V$_0VCRLs#sQ4IjQrQlnG9W)@R_rgT6uH99rGmkRCgMu>iRT(|k2 z!|1(TE6tfx>W|kuji{NZG{@_+Wu`d`Qr|1v$y$^kI@KEtL`Zn02O*qvh!?B8J=L6l z<+;OB4iOm1(zvR0dXm7P<+hwwZ>tDb}G}x*f27OMDH?FRBqVYDmiP zX&n5vT%nC<69EOt2z=~1^vg<+eBdz5Ry9Pc7GoJK*?is4=ZaRoihQ`ulkvBYfrDcm z7_uRjbDfFnCiXPbeynfziH{Y%_`NRXJA6*fKE$j?#ha}!xkRUgiuar0z4tcNLnoTb z$%S3pmXCVfbW!Ga7cTE%cvUf~Et%z}_DT8AlV0&?34Y8wuEY&*PY1~#Zgw7WT{zG5 zm$F|vNSK`bd2R2Bm^g7=L)mA(_%>;U%{AF2c_s9rxieBH3fVpWfMHBee}<*Px>NM@ zJ8IY0t7IpLLLloml0I&o`i#TiJT{%NCt>H9vrWByDIeCR16IECY2zxj7=5^c&LkPT z9r3f4M_AWv?3n)KnHCX?X|5mj z_Y_7P)>HNgEqcs-=H`G%f6ly&ASS>XFL5$v#;{r&TypL~d_zA+w;A|GhxXyJ2tDMRl4(ei?uXQip_{2*0 zD40m>cH#h?4moj#_-J|6ZiwR5k=HxPpRl7^x-|3ZCU$h7`M10zwwcm^wNdjyG53qEM3Lr&ja$m^Dm zv~iJnT0}}3==^5cD)OM2X6af_Ec)~zx&s!8ydk{vy5D^ILv(I`VfCxscm7R3#Z!A5 zu}a9@E7yv{BK%#ZQuXP`Xm3K~9$V$_zVMORyu*?F+8MsYjh%Cc27I){hUk_B-(?F( zyQ2ASqSp1R*mpipWKAIZdTL8D#(YO&m(t+`6ju7!CY3VVExg^08VPBV9y$sy>Fq{P ztd7eoIU+?UH$}2GhVPbDKA)`+EEz!r)N!^0hKLvbAfyn@gU0bDCFDT5iQh}|<%=^E zs1NPC`^5R1ci85o&~sstXZ${&HjF^eg4qf4@_h%=o(`L0HvU9&}5Qfq&lAGj&F#d$6KXN)X|2k$pLaIpEs zbnz&Lt8Hx;Luxomar}&@^b34$rZGp(mH`;l>NR1GR=hkjk8RzKm#?+7y|rFky^Kv!unG{oqpPpWp2kUsCimm*>25nf4;`Gv%>M+3rk-J>xl>;0_S3{43VuCtJ9Sc8>5@;3}&SE>TMcD}CYi=$qRE^*4$TW9`pC zi-;yI`XB8O7gwOcP4xMC5K|<^6FcTz;BQcY#=WNc zxK=S+Txr%9ZSnBzmzTEyUzY&CiPkG;cQBqq5&o(+OE8J9O7t?9=ZpoZkW7!dna3Jj zxM`TIWW>VL8)G;+w4|LI^&|Ab%1^PR!_dkW7tJ`x_lp`snFl4Y?NS+6$Nj~j0Z8NG z%+@-t$1mn@E75@(*lz|pt&T}=-{E9r(7SW-e=@=FK+i5mx*(+`g{j$jjJdW_Lz4}k zij#4E52=QA=2xhEKP>Bm7d;YXZ7`uNKM~nw}nh0jelR^4;|__OrHYSsyBnK0TUR) z6>3{_{jI|ns*X=Bhi^%5g{-jO3v4Ds678){9~#eKW8`kvp?5+Azu6^C$qRLDX|Zz9 zxR?bmIP8-)BRVp~Fk-MBT{3LLg6^r+G-Vxs`CKjMMDUr<;cn_AhO83`Hb#>u($_qz zcroK>(*M!*^%O*u!&c_`x%NXeqVrr=8>8RCj}fkugzC>k89(Bt<+I=HeTd^Pu_GS- zE%lQ1p{}W>io!{lC(L_pk3$$PtDRlamKyo#ygUa!tUQIom10N!a6Rtl0RPg0F=w_b zu$*DDJtq-@?`{xqPtYG3u4%-cNRx0>m4V3iYjNBQrq-hacu(vfuc5lf$@9#r(}Kx| zJt>5BE2ZV>Nh?b2PfShj*}yJk);6QosTBck%M2|&Kexnu{1Y?t5NC4~I^y*&D?GAI zKySrPJJuRiAK<=?LDsH1;sLnPncbh5!*M*260qc9o|O-OZjac%LdYt#7UCrS#Qdf@ zxtjtqr>DgH~9ten3oFP~=L=?X8H1=;^b$4GM z1SJ?S;$7=*x!OO3{M`{UB15)oTKD#t&kqIuDdl&wr60K;P&0)Kd`if>cmF}k@0Icu zq^X(414CadkG_5Uhm^m2kFKiHKWh3w)byWHe)nL7ztsKT(YyLmY&|J< z46_bQ<)rH^24=Wtv5wm3pMDGCgf!{PWEDcF?Vz7~AQ+J@8SR{TIbWE5R$-`(Nb@E^p`M1x`WnfC7@> zYoQ=V0xFt;Aq*+XLDzrHQiiQkneAKQxAUilK_z_&l^10C--swjZpAKdGZ10Qv zP>N0UEVI_;n)=~GVckF9yKqe7yXiW|?r^*=q;&;URSJeCen;NKS zaPkZ1duO5U2g7!7%nBaZMJ(UQ`dFW<%Gh~!7t;wd>6bHyJWnk`x!o}Dkl-|nZ9`!c z3VkJPO&s1!J&Q2^wj)sD*LKo8-?&7{sd@ai<~!8V5+QSUItk3e33)jITfzkTYB z*QHm;g{`gU%WtgJ9ms%ki|uvE4L{i`%at}N<<)Ev_DZczuSOocB<#tu$!QYZ@SSx{ zs*4vZ|Dm72oNIV4*c&Qu6tKEE4G+%^8u)li%)F=3U@SxZ099Ko6_S^~hvW3F#-V?B z)z6NYslD)5D#jL^oa(>1*7Ay@*Pb~^s!z>re&Nfhk@F+NPl6AZlaLPWstkt8e}~Ba zTh1s&Zg_qf9`GTg@Jhbc4TXa?oNDw<@t40%HJ)Bt)fA_3glOJa;wi4cHEBx?Y@?_eWli563DoBT0jZPEETN6+)IHYZyda~ zS?0R`-Nq@;nEQZ`fEhofICx6LA|F{Y9gudzvZ75)6jx-_|3s8wE&oP$&3=Pd#pcM4 z`#lXOpQ#|vQ*9CD@h5N8$5dN39ZD}=$lE+9s*{|PU70Z+mz1&4Hu=dP$v~S>euD)7 zjjbO@uC;KYyKBR?a%$)TatGbI(*bGj%K_4bwI`CF4iT=7%iYZBtlFC_{mWK7&E!_- z{-))u?d#R8x{Xg~QD44_bkrPjT}N9U^2PEDKA(T$X<66(E}yIES-vA%z+g^^#7uy) zxr#|kvo`%;|lHM--+57DSf~L zy@Whd3p}nQ@Sj0NLf;!y;!wWk%ej&(DE_I+=1{~?A#Bxx-Wh4mv#Et4{Zu%sB^53y zkRkBquBP(K#{vqiZ2^vJ8oV0!v%^cm3d;H<6AQ6!od%=ATO?Q755vC%Iw&_LDs=lw z-_#bok>9;BnZJMfI%`yZrc+U_dKkx0R-=}1GH_)2DDeJIxwkSqx5&@HaSapu;)|w@ zuC9_p;UcRVJ2|j$j@w=#YmkRnqXw8?CzCH}Flpk;diBrOHF#LvjN{|C-|G|Xz%wko z%66=y*S3D5>DveMPO@Bwt=yQ-o~rcp_oaTgq?zJZ*(L6&PAR$AqxR{WhepXv-VN&N zSq=I>`1G|q$@12ZjX2@il0A(9O9PzBexrkj#qN5Ki5{oO0f(Pa^5tA_w~|hj%F+A2 z=5KH+y-rt6L!+&)eDFr)zy@uxrtOo8EMvk#dea-GIYO@TA6F6;Vr%*LBwx$B<$f6( z#cdqVAJuA)?ob%Y?`dP;;T~}6^#){dBrISjurJk@|7*CIcw;`~#&Slh$)UquUm#<) zy9!je)?GN_V(&EXSgL5yeCS@^nS$f+CWogcz5VFky0g~t(bGiT$61l!prEXOb2rd` zR<1-GNp_M(bA&ji*xHQ72IF!%&sTBEHJx~XmiFP2?PbTvv27<$?R$qS{irRi*0}8D zl}3{`L??xlg`d`EsMh<6Qr#8se5ZaiL!y`Simj%mRK3dG;D`F-5(A&;n!>&N$v<-v zW(RiT&tJvF$(et4<)IV03O>>;ZIM0)J8F(x9i96@Cs8TgsXqT^cPli&@>JQ!#85AM zIGUt5%+$$q2PsU{o26EbY3kaDa?1j zG@&>qg^>yuJtc*3tZDSKPq{5)@O8;Z;*@27Q=%q3q3|<*zG8K{Z`Fn}9&psfLeS*- z(*05!gdJAk@GkJ5B5%*1*!@WC?k1vIdm|l4i8YQ+ZDMo@LHth zD!lu|c!j%N2d}G z$S)CA@h@Vr%ASbhRhsiN#`P+_^U%;893H?DZ9V;$($xMgjI2Cj#S;B;6SDUebcz)q z%ue|q5&FS}FiUH#REZHJ^U%~4*DX(%19c%yZ8}zCEXV1L+F3_MPDcb z{K*r))>a(wSIz!vvp-qw|6DsOhKh8uy)AC!%V&A|eQ9sH(N_T1*>MDipAi*^s)5f# z1I`JETT33UqXoACUoe=^(>`)zv}MQ30;CNaU|$wvJ!o%!_KkaNAK&A8ntt_aeYlmyZ;rOMRuZD} Thr8yhzquvnZy8o1{P+I=pQ3%U literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-74-Compensation Structure Breakdown.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-74-Compensation Structure Breakdown.jpg new file mode 100644 index 0000000000000000000000000000000000000000..5327928c14deb997592d56944f9d42dc78b27eee GIT binary patch literal 14293 zcmeHu2Ut_vw(f!;y@>RVf&$W%-h-kP5fG5xL_nH|NDDO-r56Q(ElQEzd+#-fNbemI zIsyqbKp;2nbM`ss?)QE7op;~+-hJh)d?Pb!t+~dU|Cn=*Im!%n3cCbceWa|W4B+7b z03PlGU}pg(fS8bwi0}$A5fKpy2{9=-Ed@Cl89CE68Y)^gW)5~XW>!{CZXq7d8v*xw!X2sh1x$j zJUTu>pPv1~g$IEC0t@&37hr$FMUBISPe1@7Ao_(158o3~FySj%x-W2jSs54@3=s zfs-@di~v$T5qbMJ@^(PRhfmt-ckvW^xsg}AQZrGAc6x?Ehgt?t9~S$>8H(Pt=c-?iEN2a-DLTnSr{29tEeBqTj5Ku5 zw^gavhhXv|SMc#RkJ2eiB9^M0qf$vG#fg799Vv%+845gKqmQ~PPl$l|=TtYDp+KDr zBr(D}krkDXMcJBE&6U~fe##HvH^^WCLT&oD;=K%iAYJDi2*-1W`57OP|$!QeLR(@Q0JT{Du&E zL+E30zlaS)`PTXg{_sOuK>H6;woAv4$Srj6m=XHmnZF<^SY4zckZr)t%Ou6N&C;PD zJJjS%H+bV!!K-rqWqiz@i%*Oyd?P6C7e9+ z;1v;z%l4NjKl>I|6h%^mm|eA#l*X}IICa(VN5SAAMfu9K?*Za@CBofc+9uJZ3q$+l zrW2Ak&y5sg@DcXhlqy3b5LbBZTRGi4p|sqsht3=qrlMWLNm#%wmE?H94pf~{w@8oC zLiPmh>W@Bc{IJKQKCSbdjHk>NVoDDhgDR;|Y^5t=IAQn(goF@-N4?uQ;71=cC^s&H` zV2zPocb2%0g$4TL`6!@O=MHg2gCW$yw0Jkv%X?QEkMhkhw z@R65|uE*+i)l8uu%A##&u>%dXP8%!)*0w}>=|tKOuS?U2nB-l&{}KHA3VhI;-z zjwwUm`{l@&iAjP!UL3#03oQyNGad{;1CC2-a1aV+LNaZ$rZoJZE^N(3WDRQ8Qlz~2 z{dB862rLlTy%@i4@!LVHoN-E;c_U7inf46{iRHQX^6XFU&KYYqW+sZh@5CIZleuX> z1Zee~?xVzZB(}~TzEHO_f1$T^KS-4$!4k~{^DvhgO4Bc|NnM~zN#|3Lv<6(;<$OB= zWQ_k1;vXeHO}E1B5e!>aw3dO_wW;srFT{jAuc|KjQMe#65F~3wf%wJPy%@>n!X>}g z6o9uG-Iau$noJvt)i0<)Cy~$?!`a(yCxp4UDb4#=TE&>5h&o$H3`5F2`s6a*Jdio3 zjNXW{6K;Kr5Cvhhdbf9+{e-8k$%3j2t%3H-1FDSos)IGIifpU2PZ=idtt?$3TSlDW z+0U9)pJ;I;09p@S6^Ky9Nim0N;_57pk&w)kbbf{TX@C`}#UbuC_IK6thl=@AscVA; zo?-zd&75zdGe6JS->}YBJwGkb0gnya0*UO4CQS6y};psqGfu43XdgM6ZKR zCkNTDq6{$nPPf*TP?7;_2|s>}m{utXQ8CIUkrPUBB9hO*dY40&y$v!3A1=~wI9GX9mCsYK@^irL2x_pYKTFV8NFLet1!=5ul{S!#cbnI=j;AFbPvW3XA`U_vwK zk2cBXMyi`MWv>XeHb)OUD({u2;t!5ds77R*bW2Ro_+B^ct&7(ouD~rs1p{#K91m66 z8a*%SJ*%I9X6METqK7_){-jyG?h%&6E{o!NE-lt|sH_Geyq z&O$v4L6`|w9n5i&zuR9F;;B99>$uG2lr?W*l(|!Hmu~w|_lB(tC@>1QY^lT7YM11F z9*q{uaXC8*6}<{!{{ExO(q^2+$)pjvU93i=X_Sy zcAPx|UD_2)K2Pl^+tn25l27w{iR+ZWtOJ3?cDvnw1|<-Fiz zP_`L}f%-%d2J9dhr_$P>sC2D<>mJ||xR>h4Vg{whyqIuHk;mM1&$O3_JB{yu!_p$j zT2Z1Y1yEwn-48C` zsB)9w(N$DC&pp-CDx1dwgH?+s?Zp(#6L7(Y?5f-kXS{4U-Abd|b@R%vI>`^e)ooxx zaa^jEhGBu99Fmtoj%^B8%?Fn`=r;^JsiGyjwAGyAhg2)*KwFH$OGk%1vBkN~XlUuI zaVPg^pVe&@S_h+)h%W$PS^xZ+5^h%bIxA|SMeBCLL0OQW<>{#)SfB|4bv$>+-}=H5qQt? z!aGM#g7>Bm?FzYd^7VB$?yWo_8kBKb55u_9hazFC-0$4Ba~a||n##YnQCu#~5#`TX z=u8>TZYKLRM`zvV_km3C4!P-p0D})`h6m#>FXhmcI`yNkmdjK5-b#wx?_+)2$HZb7 z#g2z5oaP0smxfDv+r;h=Wz20W7#zEKlrv#API49(+)Y^zV#X6tMB2#Oa5>B6FkcV> z&l8i_1vT53mIS*xDIuzztcbvchE^7|&Yl_iD5KvcW6s zhh;F;#)Vn2Sa)T5^EYJl9ojQ)mW;1I%#D>a`^0tHM zlI&TJ#fKoGFbgVQ14P~{qR@d)aQQ|~ulHA;<84+U{6%{KMAUI%%S}{T7ZnNQ`{pVV z;mFSZ!R4iZ)yur27kFW>x~PL^>3$d)qhsMJ-6dMW>ivQP5m;2R%j>u$rrrdB^tA7Z-=0I?J3j_1FV&DM}bZyYs=Ww zQ|4QBoJ4m-qR6#bt>08$#<+G1eOfA|-fFEJot>S1Jj!OUgEV0cO#}k{wU7Gn7Fpft zh9x`kihaq-!rcPx;j)Iw?(;-RVo^*j$_O6*QF)Y!e)35d%i!CgzMM_hdvMuXS4~xX zUiNVh=V1XO#+UeAR`Oo+J7$?uh0bA@Otr}V(h!Wn64TCd3~h+MYEazMXLaB7iMZrl z_jG-2k{*0s#sa}BNI0xPUE_MKAq-$-Vw78#9`G7)SPb*OfiCP;U}!-8Orxx(Yi^9} z^Q;81 zh;+ZsjKM?z*-IBzJxRt3zo8(Loii83?G)|~Hy^~^glPg#rniTzxIUOqTky|qh&aw1 z$H}oQ$y;@vl$#PI!(YVrX)041TDtUZDI^@>Y{*}BB#s~kUDMVUEOYNXKz*?R&}q7! z(-EAiHoyWc(Vnf;Zt}+GFiDyj`=s~|t0YJab6Spe ztIRP6Wj8lF^9VeFS^C<9xV2?a7Vozkb$d~lbf3E}36g$3k(+rJRp|QaL%z`bCq>p& zN|&uGK8KeC-gl>2OKLeVJm4zvkWbNP(9jH;$1>4NA>hx9W3k5?CMdnhF1?8(KB`DD z4k|-MwaO5+%pcoRaa)>xu7Rn+&1UGE%PCr#qvB3J=0T9^X#JHWhPw&xZydx?aS0e+ z0Hxw1Zjn2ZrW5HX;m!+{E}7N-P8P>pN=xqbIR8!4gEUR4LZ%tJMVg~|i=~pr;-(Kh zMVqh6$=*i4KDF+2lCvdOd91e>ayHlnrC-=I-u>$0NELD{qE0_{%``#T3_ofxB-c%Op3?6BAUUo{DNI?!htZJ@FoLQ6l zBBy)ZK({%TbDtL!i&le|QlsNL6}ZwQT(>0l-&PrfTgJNRM$gh(zo?J)=>}6b3Fh9a zh9cyzIRRsUO(+7?M{U+l$7l&KK6Cu`cui|sB-BS?Mzd|1BQseA=_<5HWirDb0=F3v|o{N z8^MX)E3UT(2W89U_h|a?k1046zhNaQx_( zaSIl-YnP`#E!vUtQVM5r8ewSMaG*fQ5s^4A(f!b(B`)@a05AZcrd7_?KiDwr;Jar?F1CjG9%zkspPQ?)qCZE&aE#dn3 zW~xmtcA%c6l{D=Nde+#&P2|{+(Iqn!quw% zXiSr5*0o1g0JB*un-1%!D3>vJ#&0*uTJ8F#C+`o7eVZ$i^n5w8bTmVVcYIojCf!t^ zGi(zqt)@e&V7P?&f5y7joMsLv>(w(NYUK;Gq7;Z5QMvD|%)|)CUfLX_-*;Ifw@i`` zUy98JoOuF#MOs*x8II+@R_^;|_+~FaU^`O-yQka91NF$A#UNjhOD=?Rfv(}cng6Ub zC(X%cK^SDSV23xV` zlzQcpl?$#`XITD_)f{S88X^c?lt*l7y9=+$XMSB4F{`MH zaRSfw%?~5B{#=mruB*QHYLGPvMbl{iv9^llixGUr_gXr{peVQPRL4P#%2$ozO=DMx z?Q%_awvw|*WOy=^t}XN zJ$O=j&;e*(o*kq&MXPshmO3&kS6Z=s?3g)&>Ai4W{StH)FSdkI#iidXx6=n2R?7}* z{@l@!k{yy_FBkWIBS&pzDm&nN+0vE6m5I{MJ&LWqB1oyuXcdKu>QIST{nwCzZ5J=r zyct4e!>)Q(B5I!E$76tIimm8I`9^J%p zJpWL4r)IReX&|ynRgs*9tdHuuOv3!!d85Z?!7|xDG+5~Nhph2s ztU10nKmSY2V{ig8Isk6szn`}?+?06JO8Lc{ly)+(b|RUN*|xgJU3Ds-bT(l-wF_Ne zusC0us0k-87hAA?^$ElhB8s!gu)y&<+#2POc7z2OE`yHNl4D~rvM$JBG3}+CbGxz_ zs#-bX0#;sjyJw=4Y2HW9G0Em%i+sWAP7*NqzKajbsXmI~<&~FTDCZ&&I~aF_8}g+B zbL?_SM)ynlF0I?DPnRWL@vp{~h@yuGaZwBd)6q+U87$Bt?4G-@1(nTDe9SHV{Luky ziApIjesl8U7>tn7)tu$-SJF|vI&`_;Xdosdn$7A)M43C)p zHuhGUh?Y{c3?_V^ymWKi0d<~*e7;%qWFc|j805SF)DT;xwjsoTnieV4UYM9;>$OgV zAeDNI$R1@p*8{e3AD}JOn>^L8a7uv{Qjc+5>N7L8OXN@KC*Ml*^giO63Zg=oQ)h1W zE*DFfaaf=Yyb2spvcx4HxcP*%m~InuddH0KU0xYQj5hH&?WD377?3H?6Wq?8Hqd0-fP zbNokmi}Eg`6u0XpjK~yam_`T2bKOYVb89Qy0q>{)*eEDvGApQk;j|5zWZO&gA)!YJ~kO+*r?1@yzW# zh7hPc6O7|hin@ynwE^bnySK_2+WUlk8(F?9dQ17WN1hadiKs+o%-HrF#xM9`wtKU!aJ9M`dD3b22(~trX>P7Y(-c=jJQEQ0Uahz1F)4u4Y8+QF z>KTx@K%(=YlY4#zk&AwM+=Q*SI z?T6%RIl-L|QL!$Q$aE?Z)MA&DhMeHCB(iCnKDV@&DVDdiS`{cE=8~OPS#liT8ByKI zP|G^7d5X8v#u6hf#U^SYj^P}xrZ9ku(I>}yB`(m$SU=n^-7G(q2RLEs69K9oe6%5f5KNa4luJKI&&GGK8RGaRxbi((?h%#2xpstzC zpJVj@ub*3?l}pdB^m43oNtoxyx7w{SqCp z&~uTg>?_QAHBZ} zpa;LlNS*hce7-;8OXBBuY{sBuA#2tUwfiIF{=y@~-c=Jjko3#c$mA1nSVL}@K$E}_ z#PM-Mm^$9Jy-4CLwit}E_kId?oY2{oJ)h5At9w>N8g&Hn z1hr7%(hBQ=zBup6`+BAwzj)zPN8`)&x3G@u0gRN04~wnM9tkVvG&K{x8Yp{=g-iN^ z-Hcz$7s31HLOpR>mK!2o$8_6O|J@A3|0@0UU&sgVM`y8lYgbj)VO-d6#d&e;#6nQ_OA**j`?fx4I0%#SIppLOWE7phL>g^R0^Qm=<(o zj`4i&mFDXHsj%ghFpAdDV&4>o$dZ|y+1gJ(>Qj3%aJJ? z=N96Ayur1Kd(<;qeFuderryc0Son+IMoK^1c+rx-zG*|?=aRN++>xl@xwE51Y^Qp&M1O#( z={FYplmO!O*)Qi4{SP12CBIqylI&lk(fi8b?~?cjVESfs97e^My}jpae*Id12ds?+ zNZ8A3LH2|5GKsND8Qa+!?pe=_j=n`XZU?94zqdKF3LGHu!g8 zl;B%wU*?Mrmv#xTSvWLO+o=#FMJBN8|MF_1_#7OT5bLqfX0tlC6 zmV#$TBfhnuyEA6z`{=TxkA7;i%At5kj+o>X>WUKtTt5e`H5{;Id58X$&+s3Oa))&y z<|(11eQ>%@m%Y?88wMv47|Xj`Z)D+_$di~I@p;{I2@|q;1ysgCM$D$fY5rCO)qs_}?JViBPCCIbB@n_*wf0s(%pmAH%Xy5DWYv|39Rv$D6q=-kNgpQ|N#H8wcV> z)@-Yzla;w+G&I)xE)MhrS`h=nCl20+UyXBbpQJ4~q`pn{ZCw=e+Q$OElWYjBuw=== z?uspZ``o%-Rlzm^S}1idBNHxjKW@xC80l6F>a;I$5UJ8HiTD;^GgtQQiEJ9OxYWFN ze!J{QhC{((FInys+T!@#Z7(MI$~@+(nf*-kN70(8-J1{jj-=!jJ{4mD8q7mn&Zwl} zv5fm=zW?=u)sXMZ#o$WY6SSeu7r&p12elwHXsAdbP8ouf(kOx}e3L?a@*y(dnj9y@6WILhAFe z%_dxv?0#{~+sHc}2Y@@UM9E(?w-mb4^4+4N zs{`RTziHx!LHDH`phuHuqz=_uYf!Iu9%F9&RMgx21rj;pOx@cZoc6W8usd%UHsg~w zngrp^-UnU!Wu^|bA`fOOism>zCc_Ia?ye{7#a1v{WS| zmyps-d9Lx+Dj_Vej!VPHfdBRjQSHobFY_YOGGCde>2jFUsQuNaINi2zbqk3znXE?;BO6f|!!5tiO|t~A7dW%G^Fc%E+M;@GoY!DK_m85^d@0!w z(NxE^*R=UspT->3jU*khKqo6wZDCzy$(c$kXOjUu^CZuwhy2+yVm(kDMvE>GA#b1` zulZRk>A|@~8f&}G5+9dCS`u4`V^&QPOww3mBb7znWVrU{IuOFTyZhSPN<1>JG)BBx zZqq=V4kf~R$2A9LSm0VF%519Q%w~=%-O*b>NjqX< zNYQHUYFYbfh<|R(REfX6I2muLSX356!^eJC54s9L_{w|L<6ecSqf(`s%KD)uLALCV z77h(Tf_e3TXe#EoxBvOZC3T${jgH^F_VH@odFu}`9=7S@}VmS+bxIo@=(*F+(9_ zw&$`&cnx@ShiIgkzRs0+R9BO7XW?wIB<9jl=T^cYZsS)S@!xiT{W+5G>-_JlzFwiR zGv@7eEZ}qYgv>R+#H1S+N({jWi^-&BlWXPe+`waf_M`J_`%n4alANYkm;w_@Z?Ox! z(=4eTa!gD|fWJJJOE#|#>b&M(y%?CQnQihkA)#^R(m*moV@rMMoWh{1^(qI<)7xwG zXAm{%o%VRXGCzp);3!=yyxGJMc{oTJxNO}N1c|EDV6|l#xHngzU=?s%9v=(9SwLW*p1TKh46BD|`N%abN z`{+4>DI69!QM21EH>&pqit0vpn8qsy$zQ9UoFU4Jvm=9xb#|oq&lu`_lehm|u36UmNlXsn5j2ZtaVesC8{)xGXDtD0EHDyG@ zhY#jP`DmZ68K;v7fvO8`i<#|RGoc6HTQuQaAw;7tn8M*E$?TTM<$kcnht6JS{LHcq zlNR|((@&ul$$HO^{QU%Hvr`jq>CcU2MDYLM{FV0pk7%(13&cvb&KIzAqpl>qJLFF> zzCh9PcLe{4o>r+B_wT!ugJS`Ya*WwBg(NKbsj)}i2^oQI_*mPr$!mS{EcQFFW(U@m zJq6e|q%fmBLqlxvEflN?3!p^D_P9*!BwenwFb2l@QgzNHZ}{JxTUSQP$v#HX3Lw8? z0mJ)UcaPe+P=)3qXf`LO41TncaLEys^t~9Lx3iT*Jny9kWG7&a$dtu4W4)MAE|miN zii-kpm6U8sfKX#V+p4mxZ)>y8#}bP-7-;6KiXX+?`|Q*f0!F&aJ4MWV9{; zgG8*rY|I4A$H&FEp(PDYCd|@@08?zL7fo-SIdrd{c1VOJm=(Fsl2XCF^%d%Kd^OR5 z&W>MkTL%-9tm6s?-xS^Ha{1!?+&dN6UTEg(y0a|<4ocEOR)02KR3SBPQ!hAu#K6+~ zTtty0m7nR+-PK2v-tum>3tS!9j-TZTg+UsgD^pXA=G(PbQ{ZxCU5YV`u^+9zy2b>H zZ>N%=*1|pIlR|c?qLozo%X}Dj;l!0)M0(QeoHVJJ10z0UfeG{-!tS91(RTSn@i$WXeX-Np569~H4T zTI}t;s`d6JTrbgnNVgjH9}S8B>^=UDCY*1`6_Seu%%JI|?xm!tG)xS0+J*aN&=*Z# zm5!*}3rJ^7v;-6hz95*OaE3@;Dt`u}IdHp}GN~|zXPK#Ku$W-@+#o||v62;aGhDgn ztBa{bQW2z219`LPn4P7}iI5niwqA1lo4Y;fE)l`X;ih>xJcUF~eYtV|=3<{{?za_N z$KV9;<3Mve!IoPlkk@E!F&~xM{phl$YeK#mcPBk_nP4un;)gwX7GvBksFL{QGzL2f zbwSuCS*Dldfv!RIq*@%EryL#ijpy?|YGX&xLv8y~NW%E_Dm5kXCM9VP%c0cZQF2II zsTMLP5FG4FpLjQS*4fG+Ax?sn`~5)`O%!D?&!~9hG8Lf%m1lwDB@f`pund1l&S(Nt zmVzm}v>k%xh6uqyO%}_G!O8$&NX<92>kZ z8h^Ci)t-+nivi{_*46e^(_7dS({l)KuC3-NV@sjcevDt&8g{#%E@Yf%%|G~JW{0=L z*PLUU*5=W1>uEgi0Y5v>su~uMTf96K^U{G1O5$924WAc8*^KN7aA@K7rN2K1^08m& zC%FI9w1a+}zZ|dfYiA7fd}XrcnYVV}rpj4M>-sMng+Tsawf@hLgudIx3iDeqm*1Y2 z!?7IIqdW3AKgeJbgTa88Ldy=MG>R`qPqnLC8=$v$H*zlm{@UtRo3+=UP0n-oGaaFB zmwNNVD#y6xHw|aNUOhIqCHcL+MD2;zs@Xg(^qKJKDH_^N^{;1s$-vSGbSzaTf7_Qs z$#~2_=ev&nI!@L}->sL^pk44}EXpt-8en zT3^Y(+${ZCCVz#IJ*XtrtXEvfsVNE(QmMJl_7C$bfo?NJh2%oOjc{2NlQS RW-V@~IQg&t4NzgH{|!Q>iWdL? literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-75-Executive Compensation Timeline.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-75-Executive Compensation Timeline.jpg new file mode 100644 index 0000000000000000000000000000000000000000..dc054b86ec55ef6b9ec1c3a42c1d367c68888d02 GIT binary patch literal 64047 zcmeFZbyVBUzBif>AW*EhOMwDKi@TL#rD$6yP@u)7xNC5C3KX|e9E!U;E$$A%Jwa3S z&9l#W&fd?y_v~}dyY3(NuJ``3CYh|vnwih{JM-05pjERZy2%8888w(4YoPZFIh?;_yhMIzkijI+!nU0>Ffr^Sn@EQAa zZazLfT4rH!As#VKUOt{bIssy0Vq#-qli}cy@jRh=!t+0V9y$Q{7(i0AR1lCJfPxPM z;R7Ff0JO+?q5=Q*0RH^}qJU7*(7_m(kFbypY99kofFKYGDhLe?6&2aq8~Hl`6(5b@ z3AZ#lq53BDzzIyZaou<}%Z5<Fw(u7#y0IoSL4Qots}+hi+_cZSU;v?ZZ#c&Mz*n5Wlbg zzy$<={tXuL{ojE76E1utE)-N$5Gwc&TtF08WC7u$qCMe8Cy-VLf3hc}=kdcJdJ&gX z*@?-(s{tc6bQpg`!pOJ&6#fUazajhY0rUUgLiQiP{*G$_fDHm7nFqoLNCK{|S#o_J z{XbSJ$PhsrG30YzuWMlK!KSruoKg5J_+B~excC@N<^w#%iPw|w#SnRGg8o;%P0ESe z@y7&;9GCu$=d5m$>H+%wCS`k93q%xCuH5;7V_|_4Ki<7#5KHLD#(Bg>`5JB;;#~5* zc9MTs;&jb%OUABr?gBs~t2@n~!TYlAMy04wXa0o>Xz5iU<~jMkIm!ib*0+OHm1djC zJ(0w+!&i30wk+NSV;lMJro29_x6+>-dPKJt=qz1@TYb@_Ra2#n0$8iRW1i-_v6ei# zBApr6{zzM+$W!VjjtjLPHdEe^#9CaIwH6F@s!9Dw!6&Gg1kO2FtuF)EsK1}hTwLtugi)3x@Yz_Y1?7tkNK4P*?D7ysj_qc zW4YPYt+i7e8v8@rlxeOv@!fPzHcR-O<>3B%2KaDhtx2AnjPzMl*Rj>_1Sd=P-;S}Q z0B4)lAbM+w+tkzk?a0DtL$ zP5iy^i1%==UT>|Q)S%?xbBxl@rHfckp)9<(b~4iBI72RQ6D>wu+H|YfWGTUS}c&gRG%&xY}~cwHCduk1nuP?!Dx%m0)HT7ih_dllR6_KP8|&5Psd zZ5m;Oygl>*K++GzQYBQ%SEXVjz$lC0iOZA;BZ{s8I&N4G*S2>Jt8-Gzyf#w!a$&ZD zN9)Vf;@Uf_HReSRLyx|9t8Y3pNn*J__f=W9^&c*}J)I7^ed({ks zbdM65Uk9bMi~F3dl%e~IsQcImH7*+XOWUT2|5*Ln+hH|HR6EOeA)h%OHf=qIk&Xb)0?X4v3B6sNbxFACps zWj$sTqQ*mgY~Lrnx8VD<2H}~5Kyw~_Y>wrOy*;FywCym?G(8G9btm7)!&s$vA zB^2sgt~`Q+Lr*wOKXQG(?W#;SE57)gp1xD-S>Sh1x$5Ss(K^jPN}z8gm>l<8gP|H@ z==`^zQNxB`X5#QLOzHh37iUP2n1y@3#jJ|l3;$Uu+YuuNEyHeAq-lq7wW_NZZ*S* zFg*3*PVTILVx{QFY2KLh_?Z5B{+ZQ=Okc$2M6>*!#=B(Q*a!1|YO#tj;wZ~(#HHc$ z;qp;Ts}zl-GwcEI!ob%GnjEY(lR@I>Fv%@8SZ}kmL3TMTzj2UK(T46uq8o-aU~J6U z&-{p7eizM?5y?=olOT(=tfSY~L54SG^LMl^E;hFk7e)uP;AAZrQ^0!2EVfm?!iWu8 z`Rk6IsOEe9w(q_|Sz(gcUfk;y!D?0%A2ehOr9KV!XZ|u>j@%^l+kXyz)sF}%GdQS& z;2l;>93NvvkQJn^tfp{MWfXv#9}7JIJetYxQTLN4b?5eJ1X>%%th2av21>QHZt^&4 zsO-~yi09m9*Im+}tqdx)b&UzHnxa1UD}?HIjds0 zfqGJ()MdzyW?+|>H4;PWTes-l$XTX}pHIr0(6c9$D%T8<@I7(dOZR!V zgm7H94L$TyGAp%e6Q1G+nPssOoBj6NC#$Bxvyb++vYA|;o%eaR?6!EL<(fqva|ki( zTm8c6dc5=9=Lq0TK@x)0Y!s`ClMLoZbf|KrTcLxDr|$b!VTc&GfC1*pg1HVy$YpS= z2QZu=zS7_Bo8KVgIhD87_`y#vE11tlZzlHxATWBimtfB^bo{ZbrFLv#A9Cx)4EfT* z1iPc$TQtS2HmtVNv=(pD%m?VHA|2$ZQ0&0eXYV1PSG^k*6w=R7c1A`fXHn!w_hp^h zC=uFF9K0R@kHpm;4pnvk`nZY8U<8hpPr zOfmK`baoSu0ziE#j&qtjeiB5?K^KiJ(j}H>Ruog&eq!0aDQlUEj2n>&leLZkYn=Jk zMJomtD`L$*Jd4Oz4=*DF{S`aeo{#WKQG)|+pMtv}mZYijSNbUwH)<@Hg zD}cO?Vz~#!OPX`T0WANA6xcMPcSBs>V480jugm2+WX{xW@h9hd7Q^gJPiVau7d%{d zl%Riolz8`zxyP3t_NRk$haMk zcKJS+ASOv2z?GSE6jn2u`c#7NGlG|ipuYWm51#Y@1A*JOvmCWC;YYb{5$dhAunT|A z-6K**ULu@y2Oh;f1tF9)*y@KLV-l?L-x2x`0JSFA#kq1DTknEQAV>N~Y=4n8J9E&U zuxa>X?5b}c(@}n*D@(Ll!wdpRH<*gb4>F%lbA==bCZgI2QS!zEmM|Wn3lUEvswOoN z?_idmlzJ}Pr|vc+80F+HWcbyIDvDH~)poLMJY&L_6JG-hjj6t1^GH7O_Ocgod-bzh z^Ex`Q?7#q)^hELf*UR3PtgoQRNROR#&sj4#v(^;<9pU=)?*=!n@(t^WoQ0QZajy0! z!vJsk>X)Z{R`k3TD{38a7da+To8~hw%?w&{{22DxwrN8pP#*w6wq?6K2#EF+!GZcJ zQS~4lHYF=2n$+6yuH!|C1P{!?iv=zi%&$!Nwly9K{j4%6^m~AGft(cHM{$ z$nST4n$uJl&C6$o?OVM=gBaF;%LgT|T}62b7a}~>Tg+{ro^FS`qW-|n@mm1U1@FqE z`%1*dzeH-N{cw?LD&VWq@|?tgXPb{5aJ*VKtgV#c%)L<>qrKOeImmr{Jl2VmdMug@ z^7BVmQ53o?YY*_45}?Jmy_L*0(ARTwfr68k^f+ha)Nt*`XtFoRp}rC!xovs~`?ZJ| zTD=3!WRL0P@5yFP#^xWz1{nx|!W+JAcUF75^pAD>AH~14BqE$zGxQZp)Tr>dgx8WnSH#l^)Q*_@? zb3_bNnbGqPG7<gG zXcne{m6DYhuf@Ldaw;D8lwVvuXd4WsdXxuyMsE=0g8rOm^0fi=qWb~Duwgaetn(9` zu-}24gbhP``D;3B@zByeh!6PbUeK*v=^1}B1ZBX9$S5t?^(K2m)PK$A<)e$NM~L{* z-ZrY5nbLO?ltIGsHv=7OMc4TH(iMm^eMEY)4;hRz`~fguwau3C!oJoeVYCW_sH6^Z z))Gu=N0T7ZT@=gR|Lw}iKt?b!Z*;DH8`28k>WPUY$wZs7KL4rvbjtRtZK|Z`#~{_$ z=`^YZ`5xlmRxPUi9h$x7rs3%u4E(2NVP)seFd~mH3(J`W(2Q4iHe6zj0WAve+VsT- zM%%!a*D>8X8PR3yH=-F5$83eDmca^Xc1r+Mvjt!N!ha!e&sGudwc25gTAh-H^*2vK z+ZT9NOucgNM4ZXWueH~GL3~AI%z$QK)<}y?#gTs2#WCdk>#XF4k|f%Go9J|w+KiWS z`O=Q1u0_)R1HjZ~ud#tu6abzY15bS1GE*6KuJnn{=CU4QapDcxA<5fu-7 zDOy!^1K~Z9;md*>-UBOFd&V*gA%AAmr~WpPep#PWcYoI3^7I|`yUsQ3WL-=FkNBxv zAOHGQsYIE0)A=50e{@5WRVAx^VrYt>SWz^>GKKMIE0eU`_sV(kc$L(`(3ZZt^BC-X z4}p@k2LM8mV^dr1CQ9yUw6KSFSF7|6-6bFJl9>2xwU^Rc<}G3}=e@<%sFC|c@Sm_utO)&AU(+m#Ohv-2~v_vBFERSq3slU*;i)D86 zH`~J&)hF&T^bo4b#?4b(Jr!r4wH)?;5oOMik@wk4k!!Y-iuJYt`4Qd=^AO9_1mw+%eMv!*|d`sMC+mQjw04X%^U###6;B4T1!(E1*> z)sRaH=ibD`)=9mg-7dopPJo!vf$lvz{Iw4cEGnt0uyT-v4H^9+kg}7%xV8HA05I>x z`?OyBDbcDP)dx$v4QF<8-L!rK4|lbVBRj03RED;_|B?6r+XKKW6|e$V?JLQkh0-Ie zR3?vX=oF#;+MA|J)S=6f_I>sj#D}kk!nNmEqZOF}TKy+^OS(VTIeZY6tuq(QjfTr( z3SVyrjS%6Ncu`jCizQ`DvmfBN`8L@}B6@)@p6TqXHjNnD;%r4KYomBle;yv3_f@dr)hG#WxB8b9wY^b( z4e?eLjJFeLUaa@lQnX6(fI%UY6r%Ch&AD-mA2wJz!eLyYnpzreX?yzVHzVtWgs?jH67(v9B=8p9czf0Nn){iYGzF9w?cAeckIs9X<3M@nZ^nuS_|Le4#qg{*D#GMMh{wVy`M`=_X;z6gkw12gA4wS(h#x*>!uO6wREPqUQkw9IE@#uI&UqJ3kRHp&2= za`h!LIp)c!-{8vZn&T-HLTc4@qCkCtRmv94I?2V5jw&3G>-Fi$m1E?|VoX2Tu&2q< zQCkL5)-09a)LCOdl!Pn z`BlS3IMZO)i@l3a<0*fjwbn_wCZy+__u;D6+VyYPhL>`zP1iy|O6XytIMpRWqN*>* zB57?>O1UC@s8t`iNp@0Td>zJVnZPt=04^o6+at2NT8*HAYP5k13p^D`b-vk$7z{R& zRBEU`Y=|_YY&Vha9ukr_`&NZVBt>Mbm<)(Tu*`V9@bhzK#4IF z#uWFW=Qkq>>M3`0N@0L#wk=&IO&ljUR^GxAxx$)&NsrB~Zs(sTj5J z(-|z?MU7I1urM6*lhGX7!tb~55IM}<@PbYgE41aUnH}TAP>g^`_&y}A;t5Pf(9byw z?3VnL@5q$7wwXEG_IVNW1Axl5W75lEc*FAnK!zxdU+_uZ(w@)k`R!}$Lm6|6p#~nm z$DhfElLpod8@k0mv!x@bF;|&p{xNLzw1=me(NXTO-Tj>(!-2ls>3inGlYFMU3e`NB zr{64@{akr7eg06Ob{N+i{2L+-<6dHXt-4(}(DycZUu;|2f{emtB5#A-Z*kqS@hS@| zGVj?lWYiJK>lmLo7e3SsCX(g6AJg1R$?E56k2 z^R?fH;|Lhdw9^r-Ru!Bk&#VNWx@WR{hAJ{P&mT`*QSwzcA|OXH)JM`{fx9(z>4T13 zyefJ3(ZXWoA|r~pa5i2pL+Ud>ZiQ%I=R(KYWY{B2aWMZ`4Gk?VwuG>`j%k>;)1^pw zb|UjfC3N<0Q~*2%hG@3kVgNj1G&1vP+Z5Mqw*+QmYh|TF{4i_VvG`lOFB|ntO{17z z>Ju%BYl|n3mtM{=sDO`Rt8M0_1{T~#Je_vJx4quL4KE)6T1~KiU#H}$TCuNN%~)SI z#4%|a?#~_zoqV`2Thkpy@xqujYRxd7vi%;c8%I8ry^_u8c{O;IjRS<&_lg7(?l(#L z)EHY93=wZ1Aww+Y4< zx6AWxK?&mEETllGN8b~9T)33L*o8YV6+4GQhM#G0kqOa&~| zAi6gQE|crC@Q&N7%9q$x0deU8leCc3FUe@kY;E6{I0H_qOAh>0U@qPnr4jB&DjpXWIn7jV7pAQYExCfbB*CO9d6_Ru{2 z!Oq2}tRn139ouYtk4<`!jaPW<3T?P0U#A>4iDOI>W8~)UaAgKScoCY+T@|z|ZVv#3 z_LkOa|LL*b-t9hrRcRj~A?8n}rv^ko+zpW&NvES%LuMFU#P1=qj_K&7p%)6Y>N z%cd#?V~Uvh7rNfZ<%&x}Cz-XoG*7DA;gG}Mc?YLI8JjXwQm`9EGoSZTGB1TNXd?eZco zsZtMa@!>M6a34R2&rE{O^4^fn{+Ov82zASjXTFoH&+$Dn0&xAZ9eX+!=g9q;gGfTZ zK{xTR7jHb9YE+vb%TMZD%#;zkV$|%-`x>C=?PvJA;hO8Vj5Q16u{A>qSF;h|W)cXr z8W4U?EeMr}g1|0r4WSC`P?_x>JG~E7pJ~3ub6*heWIjPemEKBOtzscSa_(jMIK(4) zEDO!L{_-Z^OjV>rmfmQ`l=Qm?$mh93E!q~Tl6jaPX@@c5ee#uj1gNF$%Ocg&eE?W3 zxIMCPQbrJz2s4^dJ^$eIb=Z?B(Z^Ywbq2Apq?59N`?29gyqu~jIHv-oE(Y3`47>i> zBS29bRi0>O8ZY~a;u(YyuSNO=CPTuR>n+=y1dUtKt=U>LhoN~ATgWoiP?oDlI)QkM zL7%Tk>c849&oDcfC-Tl_$Hu_Hi??iuxLmgw&1o@FC=Shl!LQ?a7gLt^KTydOz?svv z>#KntJu9pMJ(Vn(Eq zUgSlcW+fOsj`EVrUf9NV(gSP$qGtSgYuQ>lZfS)wB7}mqK#Ia5i4ckZh7K19)xfQPzn&sqso7Hk0kpalLKziFeB`tX`i0Y1} z{My=KD0F*e4V%C|-2|DZrLhmt!}F~%pnkt(pzqeLAAEL1=k5~6!bW}p25?dP_WRT# zV=@0uuPiTi7<$Q?yK~*tF?&Ag8GD>}%KQ)Xlh#K0!oAdU%ZO0}8Trh%U|%t&{&m5g z3@l_|KpCrtg8y4`m9QE6NUst}=Q3vhy54G#7ypK7{q*M+*D`I(B*CQn*Y=Cp#3M3< zH#LXe2hB+CjPF$BWaHshBkbXlod?eO2-O}EdBu#M&ys!mrw}e(rJ)WkG4qg1`AJ*!Gg@ty+*`{v){Iz9H;la=apJ^MLo zXo{NyP#VcWkuDYL-mi$_VoOsrq&sLg6z___SB$qIh%s=wmjuG9%tD{F8SWb4OlHfU zGF8kj7C6w!xQ4y<5%3)CqmCCZB5Xa%^R2<7gQUk6O<(d?ETlL~5+eNrgUQ8lJOwYm z(Jm#N(r(EAS&^dAUB?T&UJVhGgdPb%@B5(2`h>(a#V514yi8g#Ey%Q=9;sgoei70Z zC>ssoGOLL{RTt6Id2)vt6rU`rP%^$vqU8 z$&54ccuY%MZ@J*y&uCZHTyAAOZU035clcB^&9E!YX$jU{iD&-M!Ou>JxTjIeJ&|wg zTztPl9`DP#TwiP_nB^@|+Z-<}biI8(a7BQsL53fhxoH-j{3qhltjBxTwBa@Fhs&lT zx@pQ#Rd!+3uTERdse#(FLBgHK71 zushY#;#xLjp5tvigY;D9WQdPopN8Ur;!DV-XjORVi!~ zJn(n`Jfbrvh%`ePnLU|D%F(M6C2DlmEWHv1;D&P35=~qYNg| zVf)?lJN(nJs?viy(?ykxPbp7L*_v5^WP{YBwE+7p;uf@GxBXRxnAZ)}f%7U8gO|xB z19VOcWur9HDRB99j?rj-W!49PYGX}(649y6uijup=Kx!fFayUdbx8{{SNpyH-u(k) zmF=Dn()GsZ0U%Z`d6I}EKVU=588&ib&<4vH&#&?ymHkN0?4qGY)S*}GU{_GiTHnJJ z(#AB^V)Y5Sin+Qd3$cA1ncq)opD8B|eiBMI>;3@1RF1hlZFm5*Y2BLtWhCHss$@4F zc~)quRz!<*nZ;H0OGk%d?N?FX(4-ze12IgtPn{rCJ1EKn;G#VePn#O5JUNsg@o}Q$ zq>jOqerjM)CI8G{rz^Ss3>l8h)S|r_dH_TL91IBVsSMn{KLAu_op4@=Qah?b1rwfp z0L2Y}QK>HJwsT=IA@@wQg!0916+N*LLa{m3b&oV{3m*|Vj(A)`o4F9LK0|gA`|nQM zH{@@h3P_#}ZQTDhfM?^Xc4Pvky|_2B2&`P|Ni=jVZH6DcIMkDNSbfe4Si5A3!9i;G z@iw-Dwr33uRaVr}A7rU)`riF~hMc%4P2jBdzaoroe-qkCxEOJf`GVJ2ucK3V@fr79 zvX3eXsOL0^TMNwHSrKhilMItiF+{s{b#d`ODK<}46bF41^lSfBw*vlv+{WK>a%(f=0XB!>=z>@Iv^r9?#)W@Dz`fHKu44yu_Y6 zXY`yce*pYtdHfT--FISq@C^?1$zaDp$atZU^?CMs+uULEAZYvdyDCN0tTJhEU>ko& zWnUZjcoHM?Vccw_V<=OGGtkS82>8-P&)i1xW%kkxnLEc)nhsY9rL=Z zqON?ky0S)8{$fJr~wAZ#S+5LPyRDL z;xw@l{;R=l!h8O*X8*d}0E@Ymep2wd= z78`ir&#Rh^63+7X(uj=tm}}`X!Ssvrv=y;kN$pYPzQ+0xX@@8#0E3hE*=Z((ruOB1ik<#@E zvo?psh9YVbL@SH8Y^e@hdG3yrB31fYk!XoByK!b^5CX@gw}`p#!6j1CZTwz zxFu2I&l%azz;OB|16nr$JSv2aHN9d>m@DSTH1fHP@ACG8-~&bE0y|EA07zVAU3EPG z;BO&ap7}_oweH-(+Hcr zkK4E{+PHL^1%A}}8O7;+eci+G?8%FxeQ#MOT3F^Du0ft&#~ZCfxNp@yF2(>YlkcXOiP|_aoYhd;!o*sLHY5|4+@DN8 zVyJk9T&n+yhhkl`>#M!qAmy~~Ca%!WxKU3&mJin^>Q=mEZ#+l6(Y(54#mbOe=RXOm z`MKyabz-7e&S}aXSBZ@)gt`Xw>)+Cg`3rdUeDZM^dsY`=d4i$0;B;C>5kB&?4kI=RUzdy?n$sE)}bQQ1$(;&V>7!aT;{s>K9M{L zdH_HR{|OA>jB+&uLX12Gg&t6W2=2b2VA4AdB(ZB1p<{=vFs!r|$W4@;i&Km$)u8Gw zyr^XIh(+h|J|O(po$i~L>;Av8MTwQd^wC9{Tw|T$JK|r)3IJgMP0#fh+F#FR7joIe z_w_-K)%t|!E#7l!5E*Z*h6A!{-Ccd}$qdX%?w{4nY*%G2Y74!qcQGV7895qy?(=Dq zAE{TcsAl&(TfYA>WXyqJTv5=6VUyAN&4k% z(*h-w$8YGzHS}?rg9Fg-fo_vefkZI+Y9TqSlV3Ozrp(Db4d%aJ4btH*s z8iM0Rcu_;kxNZcsCoU-eW?O%f_%xXqi?pqGvi=L(x@GkxydY3a`1|*)Zx*CUNyUoz z-z)~j2f2{uHPX!K8XuS3nN@oLv|AuW?HD*KCWsc_X_R!o=`x^v_cDHEn8DV?)H`zV zhwDozaS<)+bv#_esc^*_ysz<7vBj)FA93)K16eu&#HGcEp-kt$qK*)oLGUVp7peGG zRsjD2Z~liy6K!~_sBy3cu;0$T*9Kv>&VL-Rw)%MS=N~6vsV%|LO8&nXUI+9Y+K9y~ zB0X&IUl>XYt8r5OeK93@vH*cQ0e3QK5k_#Mt{EJYTtwQ-ntC%Ho@X3L8+9w2`A7LB z9=cWPRd@3$p4;ME|I_>Rxw7dW(C5$9HY`su#M>#-{>sXjzpw#tHj8kF8}DYp!XaHU z1=M)9iR~_wL-)rnw*7V9)!`STU^m0X&K1(?aqDk9rGVrwk)#d|{nU-627l>Ca@~A4 z>+GNqvRZ=_2D_m8+=jmn^#9QYjmw}u57*F&%!vugZ*lZI2|R^lsJOq}H??G?a3Bq< zGVfRy@A|GOiloqp)TR9Y(MI_1)bj&Qvhrm$I5(Unm6(P(p0pf7<6`lW0H~F6c)RU2s2J+ z@NnVJ?YEmwapq2yGDiMr!Y}0_j9F*J#gNEWL%$DBK*%ECBD1V~` z_pa@>u;0_FVys&2ijKjO)2COlqn49bkp;NC)zr^}a&WzB(DiH%CagNamn9s!@m7#h zpxLgzSS8hd?`Ej1*#~`=vSDQX#Z>Fl+(yAhIAD*ZG=sT^grY3sgRGsTGeyf*T-+N5 z<|I&bEp=7|pVCd?(OS#XEuqjB&WI-u0F$qPRljVUVm6wqUJ<4Yu~@x*ZdrC424lM8 zB{``+;kD>5^_ff)?40@wp4FFueY-|s%fHFV-l1lSi>RTu5(nP%jFGGe_dWo=#y|VC z_41S$u$J`bTZKeJku$Z5K;@67>Z!G0%81ZulV{tQ@7e=B2oDx~1z1YcRtYUAyniLt zle#s;Q06FPtj5I!sVQzul)gU@amo8QJ5#T4IT(BF(2TK9Mn&A;3eeB3{CumOYp^VD zi?+`7tVed;NgMBmy#xIgN4XFYp)^shxp2G{)XfPeQWvYJp9)0vJ2nStgm1U-mey62 z90YL&^`>j|a18VjzcSu{Uz8{PlVhCgD-Ke50K}RU+}W$%p}tEkD}W5}Ox`yX-21Z* z|0Pi7up*9ZNJM*a+J23tl`)&^c6BeM2AM^AjJq8A+GUnT3KQR3o^A60@FiP%*6@BI zbtqe!fc~=^!3DVr8yoKC!z>d6F3IKPjMXTm!u_W3@(d$uY`cBe>?8u}0|1EK?I3yz z*e?5#qdxSI^JBuz3vm^6yXu!P#@FGHFa_~e9NRIVtT7(p?ly!)AFsXo*XiQV<-K-F zh)yM>bk3rRQHvL!Ewa?UZ-SeqK}|(`7v+vv);U8MDcI`ECMfkfy?)VP3UyLT4{J2S zRg#iO9T)Zb**YHG7XNC7DNy02tD!Fw2Yq+*r`D3?$?xiSs$2P7Fl1es?z~e7UYe%1 z?V_g2nw>-qFi-Z4wN9tg>JgzU=^Q*uPz-W?S3+>p zW4ENgr?z*c@?EJ`+sG}2JT_q7SLlieI63~G+>+7Cb;G)sLYn@wKOwtp`mXw6Yh1s9 zst*7Iq|nGN132^jW5hEJ%JrMX+2Hyt?ExUgpRe6U{{VZN!~Na<6o&-kxt?|dx0); z!q&!#ZHJ%dSQQU(xl`v`;3Ub$mjhS!gHL}&P^ovhRuis*(miJaFJa44un#g+<3inx z;pFc<1KzTWiRM{2rj$B0E4iyuZiJ%W?t*ZKj&W%F7kWr1zJb&gzSp_qPJ5Tsej!}) z(A>9LaBAwZuv=oERVI%uHFz#dyTVXVXW4t7VpJ@>xjNtJNG6QMiVnDrJpXQSEtm1F zZ+!1k;Ieqtb10JYq+DbSoi_SaesBu=Dtw0~59>q17?VfVXK0Drf~R033~C!yPN@+F z?L<8~_Q_x?YEK%W#QZizL0t35Yf?bK`f8LoA=l!btf9^-gTAY$=PB4j*YB)?E;4!^ z>ihhNep)r^_C-=^P)6yX^?{f=%A{c?A}G&QOl095m*cOTFf5)z8pg&^|xwV=Jqf zs5t;Vss2%0cLtepe_m{Rz2-PmvVd@ut!%l`L+1W)`VF|BsL4-G!m_dTP!>qwvth?N z{Jd;Z;hPS?7{*6F2^6%UKJ;G-SLo{0mOEG-B&GG(UB+bis^w8F*n7yS)zBT#(N(?h zbO;tOju{&i7_{>NiuBpoYfgjv_EJ>+0cB97RYIP=?7AX4A!<1=ZPD#PI?{iLqmyF0l4_p26599PfAW2f9lwLUERehXYg}luWt(FIww>dM zIVfwwPzFxe!;?s3*!el9asW+5ewQtC#^}pD;#ky9>MYyk5pUl#r~NsYe|OLt)s^4_ zBYx?GFJqh6B-O!LP`Sa(3G`AjOU~c;KxE7B)_mdL8)`Ku9n=zy>wF6MX%&JzNYyzNvCPPTV}q^Jl2#7q9MdHKT&zV@bppz3#0A{E8F&P_{%w1QW)3R)NB^O86(2HCU5wPM>8sbkGS*=@>1~8a#hv>wQsGkoe&2_) zy0MoZzpZU8ohaNliu^&F2#}U#V`CsTVeR(E=p>fB^gfB(xr^Tz>hl_9KkA<)0_y*F z9G`z^gS)zJdmA|`d9l5EO`%mb8}Ic`>K|bBHxEm5^;mFh`qdv!3+g|nUjI9FVs=`u zJ66^tyf&XNKttrA~mc0=nP>j*V;UWQ93dlP5Y;zo7IU zC;!qY#DYN%-j}DjK)ruHR(8n2MRVh%<(u}2O67!w+9lV2H6MA8Ui(_5zQ8J0Y{7XT z(a1XfnWntlxyeTiNp-ohvb_cBCe^~FQr4qN^od_byT#c_lG0KXM4I9;;e6Jw7WO@L zhku$2szecfi;Fb-@{#hqT(tY1sTL{6ZMGU_oiT2kPni0KFy`sqrm#tW3lEMs$O1f+ z|6_mrxljJ>lK^Qy_Wv?CiP_P;@Uum7mK)M0nem7D|99pCKg5JXHkoPf>}jtFwe6qc zDJ@mO#dS;_g}qn7>IlU+#J?BaN>2_heBiAUoR0Y6wS^d%S1EgY~Jv6GG%Mq9}` zoo@EM_+jXPi$`})j27KmVc9f%k)Ie9zo@NGW|HapsZmZyjW?;!W6dc)Ra~h@r!DDp z&vEZklh^yZT%*$}gYS#tAftZa!IC!0$?{T8{_l+uKS@Ob6JEN48ZpSkXhdAIACX*- zhSpjQYf_6)ixB;2bbFcn=xBQ*yMuvdXEk4|4$d4}D~_UUGcnOcMkmLGdNaTE?CSCd z9m@cnF->B{`mo4xm$yZ^;c4$k4&%62$MFkvJaBZz{`atN2HFUc4dqOnWSeB5(Mo}6 z1o0-7W88`_*I=QgM|4WTviaB|OH-u6=`w-lQ?{(8=kM?7i}o_v`AntEkz&1v~(!#IbkwNIh!#BhI+scV5Ccl7$= z<^DXG@AFG@q1pbtWTJkOZqNEQW+zvCNfvu(m|I{&2!By)evTdq`#QUAMIyYf!6T$X zLS+1T&QQ^`BBy3$3Nq!H9Ponc(vC6rM+d3R^mHToWR;1Ej|g{~WJXb2*?<>@OL!x1 zBh7^FtR%#l-%db&I17D_`Zbe8KXE+GsD_}UDq^;vi%!}0<=1^Tb?(ip);#;9&h5zP zj3ITB^(pdZiWR}z9qNJYY7;naV}Z7Gu_c3&m7wui*&4@oSE&@(UdWJ#=Q-=vp-4zL zpSVV!w{Xp_X@p5*gEJ4W^H0`zzyS4`wt$Chvv<<+Et)m;L?YoReb*C!v&ly#+D~=9 zqVC@nI5smz7^1K$FG2i6(q)MF137CKN^3D~B3^m(&fD!s2i)5=0^vSMiv4oyiR}q1 z)Ags3efGO@Y?0s|5&6LF2}-4?-rDM(dgRp(Y0@NMH}TDU|7YhDDN^}xF6QxbUMp*m zrAO5k@gTk&;3xNqujsopYu&e1v(`0#+xzCjw)U2#zE!9162p!LgiG}#RoWz|VBgD( zzzfVj0(R=zi3TalaVOjc)*@%(I|~om=&Pofek7~Z{!qj=T#p@Oj21Z8kOFv~NnsZ6 zw=qtBK6ZXas2ILN@T;!;{UANwkHqQ!O04(t(xL@oWLDK5{JCxGkr6 zug5^_wZop;q*@}jmm2-pqXhHAEx#63BusIFn5n$JH_iRzQy593^3`@sr+qz(Rf_y) z6OIWzx95XJb?Z1sC4EF$>ki24vO@c8KPGY->nA?1wRY8`<>xeVTM9#E4f9DDeH@dn&sCz?F;l4Pm-3k<+O@p6)z4_mC}@7oV*z;->jE? zPEN7L=o@;)1pYucny=((WX~0L-rN%-0F!4OxCt40k8MBW%gf>1X|{fqi6x|A!Iinf zlCY)euxXst%~aN3yf5GpmEV?o$t&;D$LFawWaMRxf+-B9b}QY69c`{qoif%%9n-ga z=KIT`ndogEF~<#v?{B9WeO`#OtVE_Y1G7yRM3UE$@3;wH9f-gJ#1Uea+ zbVT*?P@9pF2IGS-Z}fS6s@`E6iXQ)y;kCLbvDWSrY-2N_qXQ}21kLF~lgi9%Hpwcz zp-M6+&h|^)f(-p_HR%slNClHTyjL09Qj--n2}|i6PLP40fOED>*PHl&x7%YCAfsq6yJq+R1m33sQ{lt}Tgg zvCZcB|H6C4%9;Y`ZKUpfU{&0xjY+-|IJ+)cw>+J}I{un4!}1q^!*@9EL=e22Y1;`e ztH?SfYb~*{v>yXO38@tdJK3?Bod8d6)k|Msq1N0=<0m*+W<-ZbJU0{WrsAPelcZL! zBFq$-D9~WW^x?{PUFfQ);<9ACSWjKi~2Kd>U5rLiLjl75L zFqQsWE6-V8e=of%Vexv(@(^*6L$^hijqdwj>;|uC9B=gw*QvI%ccqB+_L8dhCT&u( z>Uh{0tfQ?_w9QAb8RLT_YBlIoL2&3CKb!TJ#NlAdEWahYtWnXL0M*aIx-66BFHemT zL;gGY-C1wz{nqAsL@tWS84!1$;vb{=8dwtftydfLpT8d*p{jP~&iKWL;?cfvqx@0w z0_oJ_1tlb`HirGbF={c9>IVqYu4cJFnX>i{wYo#^Ku4I-ec0~9bV_Y)j`)FlP5bi@*#rP$s zvRIeoBEZpKWB@c99qNM)sWDV`gy8?#3>Z;Frba1(3ucEMJbWS=II`W|M+y8O@TuBB z>DKnV=xGccOj#3kXJ{eFP8JUl;%7-5h+vj4YKa<=oA5yYyrZ=7lu+?lU~x6b?BPZ? z+-1WE*?F%NBI#0yQFW1p_ZQ&co~A`&Z>`^c-Oi1`_l(NOB{&Gs{;(@qwGKJ{sF3S| zvG92?&BE#WF95st7Ccdu=B9!_*o%GaFiO))%2`~h{ozenbcq8&Fdt9+RsA(b8&*!F zpv3ZY040J?eNS*)EauA2I`X%4WF;f`;f#&Cj`I@a#KI? ze|HG}y#o42*MP3AWBHb<+<$+#?SiiYQ~|s}_63;vM<2cFo&%qQ5XCgWo-)% z*~LHqu_t5y&*uGq`}jsr;`UBU`yWO2E7We}|Lpi%eUFD5b90;Yc>b_I%x^^IhQu0g zGF2C=btgGes6=zaN3=u70zW1FtMJwWDVsMz4#8`nWK!m#=U5hWDE7H{V_4}sjcH55 zQ>6x5EYTbU#Mfb+fTgo6|36Y3H{m_Fk`HV#7=yH@(s-V@@+h9aI}0XY_yx2v`3d={ib?o#2R66Zi_1ov>)ApRhkJm+E11FwEX;Dr1xI(Uh=^G zFM#P>&B2n!t&g7oGKQ(J917;VMK#);jxUVtLXR!^EOaX;j~3WlnMI(h@!9EAX3F&n z--#eWsNvPuBZae0pKYM7yK8hrl@Kj0`|*GwRhqp`W9+LfYgrJOpp zuT3{ohG$m^!sK{g4u>(l#3Mk#0&?|!NGU35Mt(u0e6K60bQTW^!WT2YlR$W|?q%%? zCSB7-z@8SFrZnN`NihNvx&@AiJ9^uOM%Xax$Z{Kqpr8QS}3HlRiN*EA*HupKad0Eb**SRTg%=w?60^HxLVdQaAsyZo_B= zSfK`77_f~@motYA*xe{}4Ayr~8)-$~QUZI%@|8CxzBT+dG33GkK zT$c;l%?;O%8^ify82P(ugMO>@#-a1h9_p4{tMGMJS0h7cLI?Ps_kp=ecz-%T6u~xR zzg_QuwlJ64;AB%;maCaG_>CCPWtr-c4b&yV5gKy%<MV?MP|6zHw%% z@UawKHyM&+G?OmBxd(;ZDIz8)%jOSu*i-Z-gzEr>c;VWsu9>m5=HXO#O_`Pot%J3mNJK z7u_l}ym_ajz{0_q#olTItWmCeUgs>Zoqe&Zuj~CXI%d6%0GU_Du@*nNh{_PpBysYo zcZPlnzMn|m7#-t;zaXPGMb!b3XQZWng>uJ#3Cqhd_@v)64pM$@JsyCeTlok0tR-dj zfwf?l+Bgt=X+t_MI=kib&2oXFfj)FsFXg8@!Y?m6C)d;aseQcx$%j)A6rN$4V1I6| zvB#>uYm%ld4U!u~f#lK>couu2BVIEizR!a%;FC$v z?P9Z9CR>~QLO*9pVx0Uc_Pj~L(mnHo>!B3=50?^YyOKGZatKugxm?sw{2UYX?X>HQ zlFL7P2(qZ{$Yu$LFG(X!A(o9sy<3hEJ3b6bN8qw-C@gvj9f#DemeidTg#)=m(g*$> zkAx@!e^syaJw^EVnJeY1`Gk_`P)JlBMNX;e*(OlJ<;8DEL6<$WY zJ`gGv(=3Jl#Der0vIW;D7d^1LkN+U|uNkwLEm;L@f53f*C zMrt^d#nW3_B0j0Q4i!ecC|INt%Xt084PRr;Imw$gG44C-kRW%fYCh>!jhO0H+u<%gK&t};=08-_p0Z;!xsqiJC|+hyh%BrW6A4Sk(PjJ1D}MW7VEhrgjI?M9znI=}e- zv@VG<>MZ8rrE?A$`t|v&wTSKF1^^Y^P8xY z`ubJMRP>4;|FpGnvfy(OXpf79s_{6|st_F&Ya&2+ug1#a<@rDr4;rb;>jhQmnbY#< z3FUeHL&gSp8CY|}xT-jjozPMZyZ@Rmkz~-U)Q&vyxEgysc?v*FgnWG=Q?&NaevvM9 z+*ge0+hbb%P5BGMK95BlgSQY|7Ai*Xa8PQ)MY`a}=$>~tDGVZt&wWR_GH@;^kq0(D zU48qw!HS(36>6-8P~T!qlrLN(;-}gdEpqGXaEq!2lWF9R@%)fR6A%Y=?|Frm`a3Oi zpJETW-6JAP3Jqu9SM@r?)xW`O_#@0(1~4^xM@#Z+E!|2AWhn=6s{U#aWzWe-Wq7zY z^GA7PIiqErb~rfoo4vMIwNvz*X8nU-*^k=DIzE;nCd%=X@ z5@G1yJJfcZxoy~*khcW#T<@ZetUQD^h0lGCcN86Xjo*W8oLr#9<#~p7H|{M<+u7wA zEbP|sw+xN|T~pK%T$E&#y}%5lt3~w`v!dcgqgQQ$y-eq9b;t!bnJ@`O;~gylDDDsC zzQ&ENQp==$v(&ZYt(`|vzZjOI=-79_iRX@|*vt#Dx22#PtM{-LB;8jFYp#dfe++TO zf5XfD?m7s@o8EvtM+wBN(J4O>C;Lj4z%s>p<`20R^g^^51)f#caARc}j7<2-BJ-(b z;&rNE6C7(la@9%qnfW$xu+X_?g?fdzCvwKb#GSCBnWoB`1qq_^a47=hS~dC%D;XQJ z#WH*{(ozs7_VIpsp4@!*vtlJ#laSP@*7GXbAbFWrsLFmWS~FrC1-9OFdG#45uN`WI z=K$MIjLx&nl$6DdxD@s+b#V(7Zu$NOoiV-RtsL}%|r;+vlnGNPrql^O|l zLiRE+PQL!;>vNMfiMFUpayZWM4bt_X2}QRI5P9S^hW|dLwod8PQ+?TmFJ3zwW}yj< z8WRm^%!W;($ZFh0nfM)*q>^s+>H97S28hHB8uZTp5dbC{;OVLJl^)uxK}s+MMQ1 zU6;D2_ggX=adB7oIc5WabH^8^n1{FX87@(6Q8s&@Ucl|e-n$w-pOrb8FYE4}ILj+|cPThS6w2Mys#x5FQ*Ff9fw!42k=H5SQiYt3KctTURU!9KWrn zHYsnF4{x$WA$XFpxv)5U8lEnTFf?Y9$wK=JJY6nRp=w+EsQd1WPFQK8mFn1cZ4nZlpwX3vo34jz%OP#Drk^) zdhkuphDcxnyVOc_a(T*xjTJOp|1blgYY62W=qAj5qaYcWMPR4fc4|@B(iC6~n~o(v zGi~G?6+#nUphO@29IXA;7wmZ_8B*a|f}NLS5G(g0xer^31n!RR(krbOcjLaJSR`9` zBY&@}m*Mq6Bz>?ONJE((x&>iL_RQY$l3_xk1Z5!CX&6><#;?G z{Fo2pd-TzPp0Zh|oK(B}=rjgx4O5X8r}0!*2{8+gST|O}YYEA39CYlkf@nPLF$S8$ zKCthh1_!xJ!ffuMqdYv;2Wd_TzG!~TSkm_(Oh`^bqZ*a+QN#Qo@#d5~J7=(BHa|MW zI}}1AJa~K7lY8DQKpKmKW>Y?>41X~NEm4sJyw~MsFD6vO9$_24IL7VG+09XEyzzG^ zHh!=Fq7?@#L1g35%2M>_dc}aZZoWP!uPpMpyaQRIT;O|dzpnM)CqfKLfQEnZ6Vn)N z55C<`H&1SD|K4(P^Eri~HBvcq&KvLvkFz`-aTcCgq*>9vE`K^bzHL8v3!;|{)Hh9hf!4D>7P+Bg{)rcwzEKzY z5uOWYHU(BauSf}w`LzjKj|9HmCQ&y>EHYi{R_gnKpH}IpaS4|`z*)S z-5Xia`1p8PcXgeSY6F6>5ye>k6=*5lC)c6n8&Akt7V>J7z*O}#p(tHq7EcqpkA(L! zu=wuj-*!$;p9>N53qlI^1>e~1RX=W&``DwF+Dg0#>Uq$6XXD}LRWoD_RwIs>tuh6S z&hVjxt~n64+pO&HmAqm}d<&B?rn2(w!{sF!@rKx`i?P!mA(Rhw?(S%G2{D>C1xg6M zhZnNTe@XH2toxus4AxPV5GvasQl-fU{y4*=a_LpS(PTaaml%QM#Y3{~@F%|oZF$P{ zGSdeaF^gAY0IcV2xD58L8&kwW?7E}Ob(pQOu_@QZA8pICS1%jk{!sASa@YHb4A3Cv z1W5KD!1hnsM8pkNPa~zbIr}@l{iHj*1iiF!c@&FXbUT&P>4}~{)sE@p*GM& z!p3*qe)*FG^2{R;mjl{4{lE;-FGzprKVkuovUx+eBmx}Ya4>YjMCiINrcG65Ah$8i zK@>%ZW=9^y-j;VZYo7L;`87$ZZMDtOBs}w@)W-=}u81y%e7m=upI$e^-SB0WwtCgw zD28c2n5b2zNQS4z~-$~4n)co4Po`)u>Y zsv%sNhk9ufI8^w@Of=dh<91*BfZvDBn!lvs_=p}?xlZ3;1@FBS!yB{W^WLx4q1~`i z@oFQRyQoMHc4C~P>}S!vS$(Ej8S?>I7)MKp#*5^-=DmxKXy)mPt){`u zQ!JBnzZCJ?5HEdwx&tI~pY8lp&_o;K zJHEv<@ShY!)c4%1POBzkmbb`-OzWAg- z{IFr!;+y$7Zk98Y=hGbDBrs8Kf3B?8m!{4hLvhnZw5H3;H|p#4%Pllq$uvVqHpnbYi{ja*fo^Q-;fS$rbYrVeLNoa z2(foL@@+m$bd9UHF#YW5{b{^v{71_h+>-|`T;8cM!c8{&ref#Bb0*!_xn54_ol^qZ z^QKRF%)fgastX(ZW%xG!HS1CFI`erJzU~R4D4BzwZBSW%7MgfUHXlD@^G0cV7JhAj zk#^@+n5^TsQ< zpBH4ejBxj))PDwzt82n@LJk8EuM3^v_{gEOcD1*6#&1r@q$ZFtF9rBYU&}PMW~8Bf z!+CVLq@$B4ldv`&>c-qq=*OC%`y7fYFr(dF?bSv3?pyy zvyssyP#lh~;%$&uZE>GbMAzH_b9W{_r#9!NG=#a{9NduI*Adotja88j;fg$yuX>p( zB^>N8W}6jp4MDw5W`5R5_|;TakrH(*EktcFT=L!LhN{rr`GRTqWoz_Y4LqiXYAOBLIx7{ zXKIlyFddchHqz_Uxlt=GE52fbw@^mZ4{q(~=Lb#(@PGC*{(m--1tCvoL{GrN5iygP%fjYx69`&7%nxEq zbq7BgCewi6&!tJL?MPatN(X7x^NKJjwdf*7{G45uQ^4g(FS~0Dn@>Q0Xe@7kMb>2j zNR;>@LRvw3OH{a=r!V@RfiFitOX1bQ^I`CmTXJ4Da^Q3Khq+OIZ;QNxL_0@jEeT&a zmIXq~7sOp0;s*3JSrN5=QfyH>PvI_o}p z%x$C~*mBZg$E+jEAGt@rq!iiYkGflm=7b3OPxrs>sHmGk1aPE#A++l4*vvcwP#oV` zMTb)L_?>D`>eR<*B|hp%@@UMDm8%E4QC{RWOl;Pj>05+YTY`(&KPbiT_MtO$lnI7% zYHT1{rhXe);o>TI2<8j*idV}5bf|N37Wf3FPkyX&RWp78_(aAJvs0H3YQY+yA$za- zxo=YgKU`rbtRu>Hso(3W8`C5+>V~r-qBji$(0#iIrlg@~5@R}hb6w9U6XK7ps!#i~ zaO7&-9*fAHW*_=OX4mJxsYe0;;%{)-cE8v^O%iAOa&@9b%MqRzL)DOF8)2TRpV!OB zgCvdOl0IoBOso&u-mlOUYm;+ZA$xip8|P}wJp)Ry0Qv^3=&VV5ewC{>i#lisf_Qb7 z$&y|ELn2GfP8L2|>7u_~x}rP#kh2yVG|~*+D*w&Ta$v&3i!9TR@DZ%aVV-|YdO3U| zM=C_aE=Upy{1+fQ&wJ5}_1~C}EX;qx9SPv4bw7Od1m%vs7Nyu%Xko}C$Tz3_-S8!H z=)~uV!%b`AeGMC&1`8q6@LwI*>1bI%-4RQ&lRc=eLqq4oUXBX^*d`DXtpxA_ZZ*xmS~%JDcxfI}rhlaJOM* z1XF%)PS^wl1QC>J^nU@Ar+E z&rC!X3pXya7vJiEKhjv49I`q(mfbtn`k&2M^LXjA-hn5_p{ zroSt2Y}3NHXO;p)#orlBK!2H~s>2Ts*5Z%EyY?v^vRPAp+6u6A8RHJRSW10%d}+)c zoxGT)o5o{1BS~6&1)Xqhx~{FOi^gcHw|@AQtSAbQTS?nEohaP*<}Av2;6F{8H><7< z*>g=SoB>%O4Jo5YoGVk8fH+ zie6LyZ2WoWs|GUSb86NMA`|NUrMl>wk*Ty*yH8`pyG9WVtb&El<=UlsoT+i{bm>2H znk6!X74xpV{ecGb?^^pPe?P89$*n7bm8g2xxff3C+YPTNYA~&9I&GiGD+##nmXK!0=PPGOAGJE$7dP@J-UG^ zBD|C4T!%%~d{iVDYVsaC_fwsksnT%%t)+4wXtWyGav zby?RkY4!DYco%j=%2U0(@J@P8dwB+-eBlL5ALR`;_=j2b$9-R~>Az*JuBw7Nfa^ed zA4g(KcuIR;>e{acc^-opNcoJ&C)*R9rLFAjudWjZZO0ikP#m8{x>1hO!%=p%6C{5? zaV>w?^A8KKq(16F4k`SF)>$*;*t9Mx_-~AV@6>%oO+I2jET>*uTeVNo!uv=UGNCGk zDRiAhD4Vi{YiHSz;kgz_;iAm^T}@n3%(TR&aU`jfxiz^+f2o+38-$89w?w0!1OP#ml~il$`v%57kfA274Td_oYn zhMMF+m$tTp$sW)1g5KL6J-};UiarzU=HP`oYA2@IvFxsp#*>*%t5u7UmtOY07^FY{ z4s4(4{7UUb#TWsGsuYGY*0Q8h_&)9G92!$T_g8YkWGmDlo1 zVbt(D5KFqfe)bOr&xHlLKWhPk7n0BSTiLE5t@DERpUH7B_2ijH@b7fjQ@UdK9xtC#pBb@s=f1Eu8H} zXKia_>Fj--yCYSrlqY)pAN~nj2bQ?iTY_JSPW+8;I+6%CJF=d0seTTR=#V^P*=f7n zq<9Ko6I?TXh#xk7YAARLfufA57R}DFI5>d81HD_4$vlO)lOC-3p2Hl`d1l3++dIF_ zmIdM7Wc1?w%n!ka;H6BvLSOBP?uFMpojFwMv=$y-zp&^|Cq=9K%)AL5bP~4N<3&NdX;@soyV|xy@s#LTCF<7RUv@k z`Yb@yzjod){?Ok#;$Yt6cD`r0wW-l;DcPkES@A!X@0~5-enHz-epBo+xy4G>fs>;e z&@=~sdnLE9-MV0Qz81C`e&5|nRNE?>I4w>)isy`1^b3QkeI*u((wMVC+@hA-m24q7wCUe0MY za`U-f*R9tuz(v`=%*H_T#hyrddaB|zM%O--L%zbWGx7tKLOWMeBWM)BdBNI2lnCH6 z83E}}qC|-F)d&#inB246?#49bO0@gC?-ZT}ey|DE4k2>x+<{4BUE}hVPgwq}lmka6 z7-*H|^W28>80>oQUfs7Rqe^TDmEwp=+~H(N5$l_>LlvHtSk+ zS@^s4Ci%Di^6!4@L`_vZ4ZNHW*#G#hT96*pYWM*^n~Rf~57-T`4Yhj6hDi>vYtHuf7WqQ)UjK4z!0wAg;Z?=# zIP-8>yy#G-z#ZwM*O%uN!Oa$>56Fw*Bs7w2UXkNIA~2B%)zejcUA$M#Wts62AN|y@ zo%Tcl()l7(UT&aLEC_q=6{^yPFB0ft68a0UABJqor8MsS#}ff!>!1vAww^}I65kFs z%v#9o$g=ZmTvS2&IW(#QySl3-mXnbhvz!l{n+i^i49S$22ujC~SW-SS8Uy8ddZ-13 z%#0H+fYwB0+VWC4r~yJ?moAY{6)U}yUpSdFo{nAyvGHwbf0a19H1y6?YL26PdNG@` z(rCRwa=e9Nq!!K}+eh>+qZEtvv|tR2Bli_D=qz}RIWUuDKJx~r+*n)G_wCv`w37Od zsJwrQ1lxF~1}sK9;)A86Z6ZrDg-yGfl-Lu0>N>79XEU#JlCF&nD%Ad5B=H$p`;Nq~ zEId_`{S^DAtlqrcB<=;{#X;g;2dcrIcc{v5wf|rVck47o>Uqv<`W$-u(Tmpq_GH22 zS9CAOX5W*OK8%~LJ9w~IwQH)PgOylY#jaRwu>0{ymE4Wg_%f!OO_^Mn1C>DCkllm;L_HE3}_e^9rF|x*CMf zRkLk5sJ$XL$zT7WeGWty5OLCy4|kIrW6w*xd-Wpa@i0}J3%^xuWKJOSr@ff@sZp`v z>>dRLdGYu#M+rv4v+d8!%oVi;<~lNwSw)lxc~X>ZglpySDT#_Te`=JzZ0Tw_ZFZy} z!y8gHi5Bnh@sq1Ot9-k=rBol0`Nf4j4j_?Ydx9Pw*=2dRy1LBo@9JYH0k5smhRZpu z8poP77=Pqh*H#GUs6+PhZ~z`FI!Qm=S_HccRA8yOM%^A>ZcZdXC%WP1I(Yh_Co2Ru zj~KF>T>r@~$ve(E zH+;ECrD~tS-Dw)n(}r1EyflJ@WQ%oG0$ah?V7roeu@T2YMUCDubG#HlFw$laH;P1K zj~9^HPVt5#sT;c<)Me#4YZ2-6uFZ+Yz+VaGa>4{nV~Kqita%>PbzyvEc=C@qp0RwH zPF(t_JHkyq(Ej;`iJMS)^E=9)1I*G)E3om*kxVIi@M9TfoSF3#!>RI8 z43E(IM7e;%rXrvKAhP0p(e16m^^SPJf`~Ep(lO^B>+c_~bgV3!9>aF14eWFKV0&7v! z__|8wJ0kwAUQ6`L{523SGvDP?-M{G$UU`cPtN-V_WCvZPuAC0;(329Wi;wJhAH^sd= zaCMak5aooWNIgPkxr?Hm z8^&R^;k;xrZe$5SPJUgA@`($@m)RHa2Rn=K=6SJqbkUpr7d?LZuBqBEePWrSmZ&{CRF`WduEY%6928LR*omV<9U~+U_)i_%dTgvSK9j<*L81nVe>j93-EtvDB#x zaPmW>w7%v2hyZ~qTE%3Zt*yayrpGFIWstVdT-NiK$gCVa27pP0dW^wTO!pmLJiBsX zZz4E5oJ$6o70Xa&Lz5~CCtj$=ZD8iip-rIsXpM%$6z8i=ivDzB-Hmg-^t=Cb@~>uhQ0z5Pp~+iL7d@!1Qq*Wm(=o22zSUCysi zTT{Vw8OdNUuEKaPb#XE#Wu#N4_cmX;#$;#w(r!xWZ}?;OR*)hJ&KvGgtbF-AH|R{% zOQY%+^%tL00NeJyAMN?lm5!C_yfo&pk{k!zg^vT_hx>XWc|*Fm`PDZ;MjtgjaCzjI ziLw^q->I-F`zxf8R~7(eRs(25$wZ{f%4*#WaGp8L?<`#s^WJEbT}4bE*T441|7#Two_`he{5=6pXcAvdXCK) z8H@-Jst7eEf*v#2h)#lAi%?@~4M_U_0@(OzJ*5-2*?qz|rC4pvS$Q;mBJoF~2<|iE z7%pXg)V)sDZKU6y-<=%)O7oyF`QVr5RK2VdD31X1A_Fc9iZds(4-~MR{v$xnK>ObU z=J-%n+)s1sxaFA-WiJ+O;Ajw zc5rs3Xkf-*yVnp=>{?u7v3PP^5y6T9XI@S5+(d)2qRA`;dy4>QV!XtPNKo^6-{Eva zt^2#5rUh_XHAPk+1j(BJ&KQwA*eyST<`gL7=65Ede)6IL&1by1Jjnoq4h2%!DY*OI z$1VEaFh?um2gd9gwj5B9Pn3E96JM3{%&kO*D_e7N>X?5^uj-koUkrr0W#LEIT;k@p z#j%$iGef|WAUqgNX0nuUZ++2AsGV|)z}m=y-qL_xs+VzKbf!;Sh8(`hFV+z3a;0i{ zCMH_?`h9g?a$prOV=F*Jxb*#%E@TNAcsqCYJ~h!|p?Em6DFjG+uWjU-S!=U!Q4>;A z5VR#~=amoa=3)_pdDAlF>0}W=F5=g_wx2#RKgkpXcE#FWjT^{&2sQCwGYDJsQvsfT zWc!(iUW87JjXeH%AAt%F4>|HDv#Wn>6TiW4@252n0S2nHvObYm3I7F1>V7hg!>ADZDGbn8v!a{=0=*@G7`%p?;07&2SuCJ*-L)n+lIcN}fg zRqzMfNyoE{mkM=|5eF`TQqTVv_f`$3Vll0fJL5uM)rt>+|;J5qA;7oZc-n!}RA(R_6cIc1;r;E}TU3*hq^@aR3A zAGi9*CNoKpoH+9LT&o9o0m@P%^LrM0e;&LmXo2l9b$AIL%&k0IK9BOf%$=rzW|=Z!rcH@ zk%C}o6i8hpvc%27vyT~KCtAdT&|>Jgr?h7K^|jra=?nxq=6p41oQS@HfJ$joF2HnhTuRmj5U(CiS=DRIIdDsjVD z;J#IKvvC9R96as1F;?DfC5SO)=LbssxOy=yh~MgFW7u34tIWIYB#f3eZP?0#q)z3F zcDyA595`LiGR&(Nu}u*Jl)m$dqe6eMvcV9;H+6=H8N zNTyHpxil6j$~02Th_1?0&Npt~)*s0JB!18U2ogr8&+^EH^ZDTG&{#W|oSVUtS;HJG z5WCl^=vMTh;him~qO+|Dxx3V6#(RP~9j<}fM7in!%xz~vXDmRxdOT-zY=!<^J;&tV$9#{-5xVPLkt(A1# zvJzfuUOf!UWm*CnE|ak1t` zC4omc-L4lX@|xoZ{86F@6enKOO$jgui%kW#KE@NN;*`06{JuvGmJfF9?*3;D5BdK)1Ac%u_XIA%w0-#BfC9vu)Wym;`p5|;?`0(iH zcMSap55Y-=N|PY5O3>zxh@XhAt}Fm6&#UcQD=mTnPoBI6>}BT$ggoX`zBT`0rO8Zo zPDaV*o9nhstDCDXMWvfTGx7F9@51ZA>>geEy*i8L;QX~xTu`oea7wV^Nk&r6GXI6K zU46^z&4`HkB2GHZ`(nd1!aRF*DcY9a&c6VJ3#y{O%GZKchvE_MeAwO3s`d={O$x+s zwqOkJCo`TpZV-j_m2%Nj3E^Q=0qDqihk!kyvSpPUEIB*T9G`-en4JJ@0z|k|JO)_&{Ky; zg2UTEWPG`9o0hI1m;Su6i_4pL=t`dyfnVMfZa{pV0Z!Or0-soG>mY0Ey+N-`HIA8^ zv$f@c5K!PJ{oET}%~Su#W#w`E4Y;1B*rx(aK%;P&bc5Wv>s@7#4-i~3y zV$1|mbM)s7mj?a#h~9_~%`U%2E9$+yWGyp7c542WT-}fkYNw~4;U_7yCD%}N?PKP7 z39s6em;PoVI-~M0&$pAeZ}e9-JM|>fV3j&1mp`JZwE#QoR+6y?abWdsA}(Haq*&=8 zbLGWCSDKGy(cp?C;UM3)`H@OR*5sGJjTB%BKK7&g)E%BrzqW#(G~U6;s0X7wNHc2Zkb@V9jJ zi&2p8X}msU^`z6fG|Z;;(frcP`rBsT9X2hivhhh}kjf}Ud%K8L_%=q|YhSCD@cI7q zu@^@{2(^!>jE5rs?2k==`L9WvzKedUyvEb?`U$?&7Lf_iemaQZ|4??8QEhhnwhu*% zLvb$zXmNKf#hv2r5Q4i~3oUL1ibHV;6b&xLN+}MZxI=*=!Hd23+54Px&%4jL`YFj6 z8Ts%edDdF5tg=$~2;>JbVWyosKk{&Zzc zmy&!lhBIYfdftyWe?nn&WzVI#M5FuMMy>f?jHF$?^U5X6Ch)Wx_Je%JRi6#t-7vAM zUN_C+&Mh3aJI|PUu(_tI`cjSWQ{1Al^pvGC8n6jW+T;8(h{FTL_bXx9J_R_K!b|E>ppng;uF(nDFf^_xU zQ%zFPJjiwo#0@i;a&3@>IAQ?qJcR4P2a6&7B$AK1@)(?b*(ij~HinGnj08@TDm3j>;->L8{Uu(4+`gH6+X4LqE}p&*r?+ zB;<}RxT78=R({#>#4_)vB0O32_$ke1bF(H%6M2hRXo{r|``x$D$q#l!jPk?dKV*CN z!coRRCes46+^_R(9BFI!P|C75eKR1S{S0djX_dlNl+*JOp+m^y1Uq?7nLpl{*0najKHsK>NmT>$CKe$hZ7*5F^B;vYd;>AQJUkt?`?nWxg7q0V%HiRolrq^otC!2Z*Zdh2{O{@nFrOo-zZxgScS^gJ?`K!e*eC8g zK%$Rz*}2v}zMYG>Xa6C+A$J3ReD8a|1E3zN*D4ALPypSSU~}!3cIiA52>cg6l1Rse z{ql1g2UtX1x1A=NnS&-#ol*p+9zhAlzGn`tZ+CgqhBN7;lV*X{cD@iZ{-(bl9vBu= zbb2qgo&S86w6h`ky%o*7Ez1#6H&gH1J$f~OWtm^O{uc;SzWiYR#6ujOwQWn%f_X!7 zeJR;EGrI#!TMgXSJG3hgWVr~ojjbRE#HpAyIyOz zlCt7reph1p0^@`#_M{EyXAkDVN>5`k!k5gH-auHzXOEoehGXlX8%IX9mrETBE)ruZ zQ&tI#_De+A{b(uQw!JRbwb1FC{WBkl(2&$nEM^lAcjv+O0JTEfNRJM%A6f{z%AG;a ztPd@_+6*h1x*sqvY(495V5QK*j86^BZdqg=Ipw~hyXl^}qHlB?S?B=VFb?!ldTo=< z)DZ6kMGa^Gx3f$47zbayZWa$8Do6=9zsG`#-7tVZ3(I%v!#@9F&5wh;@k}Xgo)p|P zf#G6^qyf|K@4f%);d5X=`dmaw7FnpSqkSo7=+usP$CnJZ$(_RxZ3%M3iH3@ocwetnYRC5>+atXzFRWg>JJiInPF64BZnDHL%{|*lD2^Xly|^C=K0PULC{Zk5g`BV}W3{Ff`}@(P zi=YvuT^2s=6bAZnEMj{Ocj+I<4Lao{PC3w^8SV15nmGay!u7w&!$vRHr#8l@wp5~z zgZ8IGUR~Kp_77pnc!XEyyQAwNnu7jF$U|z378fnB3LWv9XQPBty}C3IeZTz+aaZUE z=E({DBUc1Pn(4pD?wPE9ezA<57LB?CCi^kYk;X*An9yi~~eoxSJl{1CbYB z*QNM9%Ha(el4BniIiLznU`%M+u%_kJRltDLG08f**0O`-H^Da=(T{7J+R?Ya%+hBF zGKDjUvV!}r!?zgr%GkEs>B?<<$_|X(7E0S&8ckF`Fnh+a#nV)S++}G8r8TAnqmeey z>@;JA1VIxa6!d#p(X=q^i|nc)<}B(?Ga6JiZuTwHjNdE4wlv3P%Y-3psY5 zABZv2)HGAafm#m_ilGz!@GP$eB#J6oZP6dd(JuLM%_+~Jt39ih%T_yzE9k`qYAk30 zu#x;narnQw_hoB;s#o0v>CC|P&RE);Gq85|uARTjxQedl&ffmr*N+(v9M%4ISkV3p zgiSx#b0H4j;|9*A<`429e}elLVSqt!@^~#50IYe?6G@PVtXIF-n7y}lJ9g#K8B}|h zj(H+v0`eFN%+JG@+}gAl7^O#nPGfkl+mHqLQ~C-Ci&zIO+VU$3p(;)}9N;4iWR>JL{P+%lThV&@ti zaTYDJ%=uqcp&byJzd(wu*3pd8?K0c+fdVk0dnY(mOW7pnKt0DxTuQbN_yMeyV<5{9 zb@yAq+FS5#rRPX=?|0A2?cUCdv6DPk?*?IaavQ(H4sny|U6V?nmUne33w8CZnZE)S!%Bq z-4+BV@R#KsD~J{2oa;g2r=FM_gBr#x$l6bnq&#_|KY{{Xz74(PVnpg4OHdl!`Z_#CVf-g$TFM}B;wVU=A$ciEi_NHxa zH^k`*v1h8z@T$);p;%3&b$oMNbnnZcF_mw9)d5+W(N3di?9-+_(+#HyhrhS3+N$K(h_}f*)A|tTS|qy;BI%y9PCl+VOK((u!pr*sI%Z z^9sCe(<1WwMPO|If@u0j9UJ?U1Uk^63XBI2cX}>J>%{+y#t+!xk+gp6@R&;rLMP*> zo_Jis6*E-{_g#yXYF^7uArVZ{qgk#0>utku^#xKPa|4gq#XzrMG)m+(O($$Y*N+vL zO!^lRpdy98t~h)R1TWUYwzgTOY)J6!tWiE(5qN(vHS|>PWt)>$O7GFbyDJtJQ zhZ4x7A><0Vv?vq76w4c4e1Bbt^RGjJ|4yBu0~K=>Ua+*|$$KY=>jiQDKId?f@q>Kg z+YZCW7r7p@WW61{jotT|_?=ca#yq-QT+DT|CG%ASQ!XXxdpO%@ai84)UpvNv$cg zere|ae%3CX`-_CecSz&KldEPAEA*Y#?BOFWmrX8X1yj7@sNu^oA@&!)eP|U#jN-Y= zmBB&nQFfRZgRqGOyD<<=Lc8!9Ht8>biX0 zyPbT(nb7zXg2(h}aA3Tl%@u@;IjMV_{?WHutoywHFV|~>7sm@1;M_{gB~tmiqmr4X zJ0r%FsE!U5K8*DNIy{q>?wKZ}_|o#K3OFIfWtFFjkN2E*Smag<4gSnncH@D$y!lQo z$`-q3M;C87Uz8=;v1LB6G(5swJg6dgH=g!K>$r^Mt%ngjvwj)ULiVkeZpBoT!E&d8 z*MJGJ740H8Eo|k2600Xd9lXN zqnJNwrsT&{7j9?D4gfR;*K0m$~oPOY43WW96@A5>g7PBK-5#zGF2hK zaWZ#uRHr}2X_G>{O3zJf!vB zM|@x459i$WOU~*>uSCzMduVTJy{p+@FwatUE-7a|;0aqc(w~HNtTyh7Ip(7&bCtws zo|M>u;CI#e($5U+=AU<%#F>2hWJLO8h46_L({ai~=mjtXkbSS-oaLS6^OTD`NXV28 z_Y`CGFOadazzfE?Pbv+=PAzLrjxm3FhnFYj}3 zr%E)o*LF(UfO|o{ag)(I#Fnd^8&#D_OAejw)LGG?({!@AzuhDJU zc(9yLrLjR%UlPar2Qm>c&{nXgi_P;5Ur%Xzx*R(DBJk$bj`xT?W&%8X5x3>K6*MF`1` z=Je;8%jf80{brc{wJE7O?{~NNnAQYPihN2=rfXeIRMIq4*Pyr=?2JI$tm|gmL66Tv ze#VZ=F;{ukhkh|MJ$kj=g6|bt|FmDY{Y#C@;fU0K4OV~z{}G{ZSRTBJGqM_&7e-@F)phmyHKS#W@qbdJL=$ctfgu8z`?P6{Qi)CRIVS+V5-vD37EI z1EdZs|67;G{EV|MprPCNTaBTc?noadCFqR3<8uU^J45IEr8J3uBsET}3-i?C*D;5v zRFk*S)_qvZ7)q#2K#?jW^DG!UNh9(5PMt!YYxIJ(4a@o)CRhHHw}XM4Bq&2So8OO5 z6eo2S5s-XN{b^+~kx>#oiTYs~@$8}Zrs@QnJ+VM)sv~t=HAv=s$)t1j4VuVu1&-zU zxVS+nLNEU$4lF$Jg3-+C&xO3JW|^r3R?90d=;Y z0Z=iX(TeO=5{0R{i-7yGy0cZvWK=TN7vHqLzOC5ZlZ!n0dub#K^UQI!>S#37jGt>XAoRWlM2~@1KrRvsd8ZqrRvuc;RDkJygkZ4tRJMlf5-Vu|= zl++;XK+NsEPxVfa2g>7Qs?L1Me{VnqB17221MReklu8A2#Ex(4n@G`b47ZU0 zW#t8G!D-lA?*g=6)n)r0bfJ*);D9vYYWi(4>|g&r*5u*(56`xE=Pc`FDEPqO?APMp zO7mbmQAvirCf~8nK%^q=;naV;uQtdOsTUpPfnJibVz#D%Bky4MG(%r|uKs6GD$u8d zlpEaytn&R{&=M$7n8VtjM8?e5s~MHI;+@^HteEk1v)S@V@41_o$Qr&N@dh$}+n?8g z+lfqCX>8ScpN;dN)W*ppeG(hPSn^%!Y6Q_V`G3Pt19bSp0s4 zjsHB_Ew|JWAKUIngCd4?c+(*XbsUx2~;^+Uk!u3fJkv~ko4^&rbcZ!V9LkQdy1RbMq~Ii+BT>v+f28xf!HF#sQc_YZ)0W~q>*5g+S3uSKf&-iJR_j?_6 zKZ>Ot_%+3xmrsT=IL&8L`4&l*6{ELCLsjjw*U{ka!NeBu8uUx{5uRSs^~T zYHQnEH;Yg;U0x7=wm>7U!Ly|%YlU-^U%f=X@t+PsErxC-SL-+KAsfcG)BeXtQ;DeU zG)EJy-)a+ML7N4eILc1e0FA6=vEG!kB*VFi{-OM?$ZJjVW&3aA+PqrFM>aDt)w0+* zm=BJL=Bik=6`bLbLwNrmS9`?d=6i^RR#Wqox7n-L5X?naZq7xljieq9Q8yWO7;u~U zKUtl_zspjhS;lU4vY(Fj=Mw^##Y}7^VQlmn<;UYYF}jNUx1xKsBDLTD?Pm3V!@BO( z)^XNY@~gjh7>nPRHep|ym9|nY_Ni=TBl3Em$ecvn0zAv;Vrz2&@{FBr>xyot zoe%2~6D*$#Or~MAgC)0lxdyBeBxzMW+Zxb}Q?7*aW*DMeutsO)q1`?8TTwO-&H77e zzqTzz<%;S<`NoRQ7SJJ;N0##L8cykO6I1w1i`F z2HgDa$fBmXrlp(h)CsNTR;9l{2!S{3?{3$TrrfF7CFW05r}mhJer)`~`w&5S`HS8D z7O}dtNtU!NM`OJ^y(t=5@kAa(hy@@(ntI6g7c+`OAXjmQW1X86IY(^4_u-z!2OdLE zg+t@%8dtK_6P-qQR=Xm{K~2~bXx?defvtqxrreck$dm@LHnC4Ma`$JESjqI1d5dFTiu|_ zuhx}8&T0(W6H1$su5}>kPj8<%@fPpDa9(jfN`Z~LERei>llb%M$WV&7vAHMCsI@Ed zLKp|qW`@xD1E;iNv_Cc!tg(+?bLcr*g@^Qx> znCy6y8za(f+*_^#)AKsVrbQj}Y;K+@`}la!6fvaRu1)yNcfw~k**;xIrEB*=W^%&c>90L8{%Fb#JMA1<$6jyJo-Cz+ z{hh6Zsu{jv51g-1N`l76ft33$-f%)G-v4I$W2fhfJ z{0k)bpPQ*IAMR*6>bXuAf><}sY}FF&ZZgqWMGEZS-G%sh1f&c9{z3l*8_bWc^FszO zaV>U{|D6hh?4d||`0!!i1Gp`QRdo1x6nAeiT=4na#ZtnmMiGQ%8fAem_gOCuRkq2f zAx?~|RXk9W(@HJ74@5|VO&qMy%RWu@ljy*Hvy%;CE3JIng5T)m3C(CDclJA@Se0Om zc8RiWo~oX*fEIT73$>_iw#Ipi?%-+_Ev$W7rHXn#ZA^&^Huj~BoQMMJTc-a~Qk}}z zwshNh7Whn4P*6^_zcx5K;0f3%Ko1q|@m1SNUNG&Zq2cRD@rEZd1D|PXU+vM@hl`iY z{$_3$s$Vma>^&s@UOK#yJ4anWv|xPbzCXbMl5jm?*TBqBkkXOV2!ZksWRSI(Q^8Xg zv9?kH`xdJn%Ex0=(d#kgFbAhck(vIbS6@|);xJ_VXzEyJCo%4?uY3`Q9+JyMKpsai znXRBm5KK;J_KF3bE&ACti5deFFeh6u&)m$)yE#DI85-W;`Vo`pjuNVbm*;~jxH$zS z#g&nGONuVejbBUTq0cQ7rgDzsZM85E^_w^*E^N1|)Jw{4oLd}B#O#EM6_|qcs9Bs_ zz6al)-dCq1>GH}3*OhVIU&Ya|O^Jo$`~M07T94LE$JSQ7WyHUbxABxqo=bQPse5=Rx9`Kzjs$#wV- zuV7Z)%`EB9q#pi;g*m|~hdQ-3IVOLEe(Ud@sYq-tLOEUETlV_zC_2PY?kB;PxMW}0Vz zbC)mv!kogI(&_nv9@~qh9(2<4-2};Q=ov=`lpfM?)p$1 zT1Rf>o8(X}-s$FGX&|$dL>M*?{K_ZEfI{iNSuL^u4wHobqm26RYGfgP**|LT+bpPV zP!*6XT_wg9S;h!;6&R4Wk_F{m3sHj;ts8wIH)l4pge+&vGpa0??z4bl&wc~P$ zQ+aq-I2{l~`o!tVm0Q|Ltk$1bUu1KuBiT7MvE9kX(Kht8L?|@9jw4le18RfH7cft- zN4f>J0nLs2LN-qC&r@d}N($`{w0MROvq8O3>cQ}FMB@iVBU}f!J!Q`HS;Cs>C3PD0 zYqKo|`k{7Op{q>`3Jp4aTN}2ItnA_V7-BXNdIB9;wHtGY0CV{u{p4uTKM@oHC~0`< zD}CB{HPd92)J&{Q^RbCaC9!~T)u4or{iTmS%4KcBgFM>%W5NQH)pM6}%IB?Pke^4b zty8S+%(b}0c>{JiuD%6fPN1E#`p%YkWyu=k^IsE`T4B$W|LZ3&#y)-pcdiA5dF>5+ zAK_qlGF&9h`^!oEF6;B@aTp9X*QmH(-{bpq*v$sJv#1L15~rx;*413&vlbgE(bwY@(xYoaBb6}eIg%i%EKk!K!U*7 z)$xN9zXe!Xj++H1j?EcNlFo=FNiV=GuY2IWe&pb#kEk7dx~bx5C6(;mH(wgpkYVs0 z0B-TXEwgx`u&cIZKa=~wW>DXu>fwIXSi@Qp%}0J-oBhPXe6l<5=_Akn;cKxlZ5p#PsKXS`B5$ z08IjeA;@W%MGYh|dRef51+ znEipdR%ck9*5&Gd{jps-7WV%ASV!>T=oYU>aq)4ix zOPI#0Ec5U}24d&4(wCzV?fz6xjo-HhW+nb%`bh>iqDF#D@Z9;zq$lK*s@uXPW$+g?n-KweL42P<#LWxEi~OL?NliYM~PjLHJ?8JL5LT8mId5$IV@%CPpxj%0p(w1KMx<1@J1U--iz^$d&TBhlA_4r!vLPt=A zG_AsUB&+e4`sf{|B&5p#ilFP`w+@p6{*mFF$YEFE;GUpbO_jO%f;ajz7+K&C#<$Pi z$p|-!CdMfW+Ig3~a#0ZBU?fY~tuP=qs4(L(H+OsszX0P)8=vdXaWPCipRIAbH}I)L zS2nV3ZWTJ`?X&%9U};K>LUB7^ZLuI*3ArW{W^c#&uqcXI!GzH^Dv3FONH&3N4d`!a z!r9358vX()+nRqctk-x;xgTU?xHv-Ef^C9wi6uc9&f(MK0yA+^H2AyM+0- z3m+q~E`gD;GEyFqg&C*wQjj{%MQni!hEb8oV)or%{L#9NlJZC&u=DGDIVBzHApdY_ zQ;j3tB$41b#KP>Ut9Y&J=@E{#+44L1`ISkR(_bKiS&gEHO4oYR!nc+va;QyDG*Cj_SheoHX z{gUFjr>gBbXEM?R>|5dc9m0nKhG$v&)+E%kK8Cj3@3=AfG za>l51ZA0AxQ2AC~ppMsF!gQpuZdP~2_Ypzr>Ra%TJ>D0*>jbhR_Xmg`yRZ!y)w~7Z zG1}`+^beg(GbM|*^Ti(Bd6=U?hVvTFOJUt9#BAaT-^saMbuF3x$=3+3Nf-C%*uCrW z6hjnzXd*azK-&+4oJskw{W!Z08N!=ks+Dv^b^04?=PA&Lw6)89J$w$k?S%t*qfb3+ z^yBbf|KYp%=i}eqVE@${7LAnb^dr04+GDSt4Zs+9Mj($mwGoV6qUM zvjn=;YBpOOb?CXIEeQ<&>zP)o#sA^9&zjnc{;BBb8apQ6{{cYTVq4|k5Y`yMhLC@rNt^y!J$5Qi9Lc3b1gZ=iX@}1F;`2Kd^T}OtvsO) z514^t4K7xl&`m?{BFFGfJn+2A56t5mttti0Zbwvqe_t%01g>4gi$eI!Mk-jIGnK-I zVn6ZEVfK70JR0b#4f>E_Xsc73$VWjp_d|YjA9ewhGCs~pt(7l_q;Ezgj@XJUHRa2Q zzc>GQm}g-zrL%Z$kff3NU0Fvos;a${IuT`*nBs|oKM9Kcu*@XP$1d?DOZLIk>y$pp z+0;09PZT|CnUgW)${RPue3OJ%TjMQUS#ADY=AdldJMs3l3t(_9xF6jqT!4H2n9 zR@8)riF(lll0lc%27x}X`-}-Mva6Pl>Q9#y=4nv-_UZz6Tm-ADU0C}UHF8_loLlRS zj7?I5TZ{ETsVHcv8lZ159NNH3x4Nd4*T%{|=Vk$#c{}n)|I@ z9eJO(7nfCuRp+TA0sS22A1qyJ0zjVYjTAV_?IF)a*_01wkvHBoUe!*%}!Gcg>If2o3Dc3A3WBxpfBOTpxsuqV=P`ZHa$6rm>OG!idU@x)^0| zt<^Oh1B;%8&Uk@8bFICwmZ`o1F5uhc<|cEawvS^kmo@ezHCYq$s`J(T4Ch?AEtFmO za7>J{(eKQ60~SKBc={bDD0h-|DMYiupOKwpial93p71FqQ?uu7@!`wO z9^C#a98r?x4lttX8~;vjr9SKV7)$Ey=uOVRWuCK!+WYdJUi*2J) zPq)*S)AYR`ymWhJ>NZu`!;a!!cE&KCV|~cSat}P}8yCB(C_-GC5D+dvva9!JV87gx z94V=UwR8CzX(GjuGPLrtK*KLNp~)y8cqWzcRe&dkhBWqG(OmP2ebD?=re^dw!al-y z-)zyrqAQ`qg>p+`PTlje5OL2QP+8;a9KErJA)_z7a(?bdg0e7Wm`w)@fOTe8J10Zb zc%8=ya)j^ve^vhy%U&uZkoi|OQuV(kLqEk5WVp-`H7;>zSg)hGWkz5Gwn&+(ywmzg zIwRJrS>Z||*SoiNaHGDlBTeh$Fz)o{p|j3UX!vi_5_j8FM_p)}U8H9E5ntjCgQ6gI z&1zW1^#0=b_<@$&bB96HE^dQjHQDhmy=b$PEMv9X$k|if{YMJ%OcKa7XI(9!t=D;MTMHs}v$JVM#PpCIqJhy^E z6kslp;>|^8)q*{hh&_nJ!Ld5n`8Re9kIfVQk=~gDzGUPXjTVZu^`DFf+K5MT?HY$b zXy*n3m^04|TVpxd%-9}2qr-lpD{6(FMTu)7!UbKM#Hm}?4UF`T=ACth`D0YPxa)}7RglDtp2KL*$mx3|(RNHguKEfi~O z(<=mXVH^rdJ1yw$vO7k~Qr%6Uzd$#yFWyqdY-B6(yz0FsrkIglF!=ecHDu|H z>!9cv>2ByBsh=mM#|njN0{;|sFLR7+RaN6R1E`XuxE_bdB@icj_j(qg?f#qsTV~8edwd;Jt8h=Y%VjuEC$x zDXu<Nb#apF?5Tl~pPd{57-|&@hLUgiNPmaWLvmaL|RIUq8ajIY+O{)O?L_NAtch z>E_KTZOE@DE^joaOO>dQh=tF8tkUgy8iFv;LiI>n(u$T1H65NN?!GjX7~9LD&1obX z_qv%QAkUHYE6Vu`#QM7jD&>%i@xxitYf!t*1zX)s8L$BQn}mEZtxR5cfjaUix#T~L z2p6)Q@y6rZd|KzKYpT$J1^kW1LR0%!qaH!1LzV8U&W!4rp|Ac8#rFTAdi}?y^S^L`o2vhg4R%D` zyPY-*e9%_7`4QWpZ$q2~y1-8h?kl(Keu@V25YzrzD1{fPOU3 z);iBk3+X4yWhUz4*MztSO^;Tts{4`7jNbjM-`SAS9ah7&Zep0UxpOUc^YgSZ5m?)O z@r{WoHv~@}FS@{kdF-pICt%XT$<#ZijOj(C@sc57>~4!ZT^VLs7_l~Slng}9bP3^9 z!rpx*Uj)37r}(;;G;{mT&c_1#ZXCFW+k7$BpQ1%1u^8|l=L83%v#i*F&6@W-vuw&1 zdjp?>M&@ddY-ec>HZ6n%}6=$-rxsd1-vr)5eB-6iy35 zjW8~jfr}j>@?9xyqMtE*d_dV-$}XJRXF=+*wv+WK``S`}x_v>*eD7D@&98%n{3UX?LT^r~3@8eNnttx_5-=>Ouvj zM%40csr!qY;N(AcikSr!d;2n!m6fYZOYItZuzB0pC}&BUpV3Ff>`i7luQDa82;v;@C1 z*h~7MMWVjLofWa<%dg#qD1uak)GutdCkN!jMzjhCd-qlA{M<(n$-rTr*UuEdjKoDY znC03d_tkPbGpXlBk=!D2{xv#^$=u?g-;(MetCmXNGnGecj11|QR|8epx0VF+FwT~0 zA1G#Wndoy&cjeCp*dYpvq%KW&>^b5wHHc$jILd+p!+X>9WEl#nIp5R$u+zl-KPF>2 zrDvxGqTj~-0S;HsqF=${le-C)ZEsJMVT%3fEQsVRIR;c`)v4OH>bRE)^Uqw4;YA~f zby$nK_`}XyY{F26ZJ^pi@+C|VdZ*i#qAzZc%y(+$j)i%?3mnlXbtraht!T1-4!Hk3 zj7@}7tsMpGN4a&f=5SQDUX60f1-Xj^Rerp9mAHBEqrv=YgWJK%C2)hurt$--=!%ylY-+f5HEdJAubiW zeh~Z1rZ(;xwf*gHm0&Nz<$rwJcKn%R?o(YXJvcmMC6XApFU zrj!1&6cSVqNOnAb_wqm4dX%yQ=5c?4#P%+^$sM!R4~hhh{UYh1&UM>yDO5v9WXJ*s>K` zY0aBP-7X-1Rsa1ORpMFh73`OUa)~MAjZ;iuwH*qha;ZW*m-H&=n)K_T*u>RvzY=#z zOXSkGIY#Bqw|8}^?vz06%KH*lN3V&tsFCiqJIs|wR_S`}TtkT#aPadewF#xU?50C* zu8wSc8reO0V8)dF^f@1Q)rod%2){!4=Fu~yPN;at(A}#FzeEb~r=ju==Qm9u9)_zz z+0=0-IQ>vY$vY3!7`-+jJ(RiwZJj?COr?cA`T6?(G49;%X9x`Nt zuz&*X(~@knUqPI&POF=2Nz)}-MD7P&x*M$*-`(08u+2}&?{+HTkkG~v@<>Iu_E3DB zrT;3|5BHsn@xdfCeI1b^?Yr>_rB5=9s(Jl$w&VMt~w|Bd9WC-b;(Wl$-Zc5}2SCF8r z?MpFF`?PIG4*hCEf+le5C3HYRNACENNW(``B#IPS=j>(m4iR9mOb{pmRWuM5pY>KW zXC{WlDZp9jA%={3 z#mn>JNQ~XV<9l}ctY^zDFr2}rNMApurSg5+Z^1YT-7}Qu-Zw=u)EU7bT>|a51!_Ro+O`r`Cy~T>{nyMYDg_a}A8HuD z##oF}!l-F#hVnUd7~=8{aYRp)@1WI;rlc7$Stoio^1K9bTF zAQ3TP4BEe_BZi30jM1h~@qI;ACaIUuD7;`cMNJB}t|(8u7rIt{av&Y5`O=<8>AAx= z5A)KON1ymBcuOdY*ibcW|2or`D)@&pj8jcVnP@M!!y{#w_An2r{s;>>9jiNbyMRhc z?Si2})F#&B3k*^C7NEp(oFP0=pn2UU9=9A3-mAmWfEc3JAEuRV4x35Hq|+_^a|X30 z)@cp&Cf$&nt$M2@)dNc(3BUI4IP8IoNOQY?S z`@{6KY5&D%))g5sifHJtfBCBMTb|Jfh;&#B9k|Jbp$(USnK2@UDnP3a4k z+V+2W2{^?Dv^(ptH8*Z9gK%jNgaqp0)}-IiLhzm5M>hq<`OaM#j+Q~@W%U=wjZjI% zmF#=LT%IBmPKOo1G(irqCT|pPbSbI_d8-=0?~H#WCjBHtSr!r9)^lHFprZnI%VBrK za(>$(nMa{cAqB2NC*ywJd#=m#PX}w&4hH@UWWH(1E}Nv zW9n5278a*1^==QT2$%pfr*($jIepS*-(pW5$Xt}O$(sIv@#ia85pquJ}$Br$seqc@p& z*-I|B)Pt$K2!by!DTig0h6RPn1jC*!_LYtlYg$oe3wPuPFtI&J9$-N*Hz=B7|9(-3 zZykcYdp;1joLV+2z9Pw)YSw$dCgeZv|E&tW^Hnh0EogG*&S{C-3RsM``Hq%aX_p!l zoqcK-V?pc3_xyZ=g{#ZIkO!}(zZtI>2@P5G{>k)MRARV3G0xV=AqdSge@XlpdO@-uf4Y3)JQ*cw+sZX72e-;_U@ z&tB;ZEMsWSB29p7n&_rFzKhLQq|q>qrU$A#+aQ;asG1!?{k9%0zj{-M%Fq6Fe`kr7 z70!V}>_4{N3tv))C2wknVnOE@r@zJYcJBQ`iPf#_TLWOnHcFp15`Cr4u=-BMAm+c| zDTa>lcHc}FbVMF+R?`g)`tu0fs-LJ5d(-U){Mn$b<1|CWz1(MWJVvN0`^$yxyVof( zPhTkbR?}f>zRCZX0j>EJcntSml$iH5fIm@6=uF8(W1wXFAtL>0Smn24AwI^Q45ECR z`*`(dT+ZD^YyOaExP64XMAsiwFmAqws#}rLBjZ$^_X%!Ux*e1~E9<^Az2Tj0lua^0 zo5V%!RV--R8KZI|i@RnBcjr~dBO7xCHPETDHLeyLtsz@L>lDBc^eO}_Zo zkmLV5q6jDyEbMNdbY+-H$qx{!opT2;G{1=fq+FNV-|au=?Vvq@r38f`5g5a1v583} z3cC*(=vWDpCDq@DmfEzVyQv-^)=ZYYPM9+n){bAlL^DtZ{$?X~HKt)ur!lvyp#7uH zI8_e@j&R#;xLRr$cATPES&|KHCmo6Vl42G^$9q0uUsh4UdGF8RJc8oT!w~BwqsGQN zRV>_ROYgCJpC#+&KgGzh!zilskCfCu{`Rxq4M(2Dbt7+-DAl8v#z0w+L0dZu;)OL6 z$I5cUVcZXj-)Pd}$$)zjqJ8lNDt&*3un?8)#}hnD3ib zzHl$qaAkX-iu~!b|2CnP4E>DoK`q@UHyEWzDOB%!I^?v>z1vUoX=n#83AkrDT{vYA z6_+Yyo3M*`qP;T=l90?&=7O!`5L6gKg69VEDqdVrt#tx!>|p}9sXMB#QJhez4@pw!^5 zrY7)iPCNwvMEk8t#uaIUi@Kf;VPC8Z1@3@HDzvKRNc;`-a7AHxfY4Zt5sl{fkL<*u z3?=V@)br>*=frmtzLh&M1N%GwqrK~jifY-m4G4%N2~AE47)X+H6bTI?LCGK>8EkS6 zHi96MgMbnxb^|tu(17HeB-1n*L2{N15`5cp&$$=RyXoV8yfNPWVD#RDu}9UeT2;07 zT64|G#8+M9za19q)>&3hGBygcZY@yE*xsdwcJ@Lz=z}`9!j^%RV=h0GPV8b>Xe?|j zpl%~v#?ODgbN_6nK$!+SG@$VF69oibPR9o16@jB_f9K2hDl{|8?O-sqV*e{qN$KObh<3|J#0~~@sc|j zNRn4n+;Nt`i|y5W9&HC2jn}QHOwmtx6}`cCDHxDmC5uXLDzNnwUW0m@yI%{FNH9_@ zi-~}QV}f+4qtniK3``g1W4p zG?*{2Mmc{02ODVGCu!!GwnJH4Kv57D9d{hKG#aZIVLUN@l=P6luGLp3((C5pEnd<0 zND@GtAm6~2irVI_+1n(i$P)$~w~w8YeOyJ4l_cIqYuUcW8kk(YeHQ5pbuB<`bmv}~ z^OnV0CcrM;w+BQ)xW=g`S<53CW+Hf3#ol-M@d@s>Ux}epyh5-N=PXNisCzKzRpJdTVjE|QdiF_VY6#<@Lz|32q-AN%05{UInrhduI#u#%W3y`` z2qQ&_HORihC_ca~WM;|E&BpaG@M z!@jSgK(>a)@=U?7r+u$-y}EcvSt!!&g)daRR~!M9SFA#PXWkX(Aqi{l^BIJ&d8mfKh6_U6SkM1PLG@(y8BISb zKM|@5DXmISycE|&Xg(z6Y^eLY(l+`LmcHQ0t##Zm1KDyj5C2p}$=bRUpC{R+Z3xp_###jWR3yv( zA=`L*g0&|H5ogb3E1V(e^x=F!zv#U8iRjxf7v~5#^Bo3fRqn`~5lv%N z`%6h=xAhji8r|>__z1lBnzC2rEbXCI+$`lGoTgVo%_BHKK9SiGE_Nb=Wyfw?>q{l& z_E)QMOJ4bXkRYyXhan1R8&>#}n&9u;EkrMh9RNoUK zYbRyrxduW&-y`9=;}pB>@)E#jGBort=3tyN#q(r9<@xJ#Ja|@sKv^x0@_>0PK#_qNusAutFX!p6 z?t>(KAt~a>HQ1VI+qH|$CS#@khAA5^Z}><9mr+S^NXS>a9Q~BLDD$!QjR;?e}s2|%zW~MU)Ep$CkVwrdHnn9yA z{~;C6&uG?P|BRx;VDmSb&zzaRQ|KCBywku>_ev|v-zSuin4q6}FlVluC_KlZq_#;w z7~n-K6r59&i2T|k#kDZBkzGomV6(${tPooone#jqC%_(w-ox@oG$STVCP$}RX~TP7 za5~&|4SauD5-prgQHt_cly~tZ2pM`ItztzBgRK%|ES8TXfM$R?G>KM;v)j-#@b)(|mA(B1?OGLnxTxt@^v z)mxp&j8)7#PDEIX0y&Ff-`PT61lJ{95vmH|%jDCm*F_Kwmly&si_VkD(&&6w%ZuIS z=;Z|Oyz`gVxr*(PIm!RXqwn1wk~EZSZDuW-A(GxZ)(kyUyhWXiGM^2F3Iz`@)zwtj zlL@M^waQwG&e3)XShN+{!$-h5iwkfnXb*8g5oCM&Nb zPtg?3of}!#&#;xkg@Z44T$;YG1)yQ=kBW7vzY}AMiL;f6$T-_rUw|*Ay>DJ*+i$m9 ze)$5U55*j6*VmMmJAE7PerU>?_mxs$u3Sy8_S<1vJ_IqkDq`S{E+DrGVC8z}Hd&No z;nRWvu$5@e-LB7W!geAV{ur>Ynwz>Rh}XpIqTcQn=|WR zVVgCC6I-XtYh>0xbGE(wxM0ww2^clG~O)=CPbR)KIMu98K)@qvQqx8(go7he^l3`!1 zLDONTPPjMaxDj`Y?`2xERGb(9!X?(XKB0!wE`HCz==wPQ=neg^cRfC7Tq$9y+ke>n zjCsgC`+ZM={6SOI>+QC#d1lsei(dE-W*bYh;n818e zC-w5#RNwIa3P=a@Lr}EMp~KGCtwXTTfW?*F!lsw_to;pz$td{j>wHN(eoQuokG~iE zv1H!GYi76qq2PkCMB8^Qz2*32%!so}+`Y{=DK;JK;Hb;{(x^emEIsMEFI1YqfxD+{ z+E4fLOb7cC59AWLi)*Z^IkPlaGo_*^om@!Qf%w+$P6 z0lS#9_mL`IbNYI*SEo){OO}ca;EwErUc&9V1?)Xrq*AaKgIN$J?8wnbL<%YigoATk`sqR-v1z&Z63 z(n5)=BHXrl0AQ5m+hB(X#SstwoUxRtmw z@VpexFI6Q-IIM$z=)4ks9>~%2o8$ZkHul#Vl4!4w-cJU}MRRQG2Amm4ucv}_o-!~^ zXGHpvUH}m+BzNoV+ENp^L%F*pTn6s)FHQNk1M5dpK119fp5%ndEYe#%typ9;sIreX8JJaj8i?Cq2D?QY@x4I*cp z{mwQe9(rv5u8l5z+}v+1bsA zKHT;UXQ(sYT{VXWUVOy-Ri#;3hD}74BY9rxxWx!Yr#!6Fo*rQoAK6VC;)ExG_T6V> zRvuvw?D8d)BS*b=sVa$#Ca$v7&FyNUg%f!~Qmvi0GJMv#VROBt z-xuqf%=-LE=K}07S`1me6u%>HBaBM<7 ziUtoq_M{a_Y!4|NFe|D55Yg(9q{-S%v-z5@8BeA)!v_N+(XUH$e`1l*^{G;XM4q@Y z9Is(6Pr(RXHtMeG7UB`B%?^>fi`A7H^=@013djs;{Qzwy&(#1dE(D|pxJhI6AJ02` zCHn-r+vd_!t&tv2GhgQ~D{>UQh0EvK_K(T-4&1w}HBZsmCyIj!;NKjppA{K;%?BFm zX?G~pQ*7dvrp+)MC|Y=<+&i9D{qm{xwU0p5)T7;QoqfnGN&BUyReW#AjaQGz%0k*) z+g#g`R-LmIUKS|hCudJBI_&+joqP~3xdEp(*zZf^io8+{CgY`E2tx#&$vrx{pKUW9 z0VuOXajWpGDaFT&^pH2C9Moc$W}p%7w0nnzA_heKosL59cl81B*##voQJI^E%BFH) z`Xi6qd&kHkN1mlPeX55NqLW49V33(}RDcG)B)J5!uTosQV&&jmvMXU(M0T$hJTUWB zA8jYHm=p-cnzu0TAIa?D<}1C_F|0PMo>JW0^tJv%<{y?Fm7*a&e09$WbU5YVdA zD;sReafYyhW=KR=T0uq^5^k{C)>1T>>|ke{*8!g%at5GPn+phj&Wnk9dzC-s z@dG4rWiWTpx~fQf>O_On4->p$l$PBp+4l29n~b|ggRu7|h@xvD^&-RU5AEYTgI(q%X-BTvgo7Qk##3S(?8M1|4vvUK*Y!ugpd5U)jjsEVLg z#Wbxbt8_vpcFuRZwU36IHG%w@ynMH5u7ar<#udq5b)&0z#WAF?*V{@Y{aFGf+ z8go_dz*H`4>tWmi@<54!Y$AF}iOH#ENsKMdRhG^mvghnDbb3HJ#qzVaW+lCs)LG$O z{KQR(HQy<%X`m8yQjV_VvJy*B8CJAii@%3eP1H6NY5qxR+TtF|r#6z=RM&M}G52Qt zlmk_ILzoyX+eBsDud862ZZ~RFmytZym6odDbyEX1$6CZ`L9+>H^7Qa*^2r4i12HC7NbDV!@z^N zM+5t8wJ|;OS)CNT3nc5#%Tk46?vOace5zZz8+>oxYV(30h-2*`ki{0Lici7dE0Zir zTqLsi016Y3?A`HtGN@2_ZQS#&lQ@W1TvrG^d;--{^%bx?I5@nzIbFyfU$)o5686qp z@HaK--#^%~CQi8lTzc}-c@03=Bd@YwwWA(fzawA-1i(^Y%eIdzWQk#wU*xqpCjP?5 z@-tKr_(zWC-~Ihf;8EeX`kDYPpk<_$bK!_+of{GcNYTp3L`v|D?5hrtUF?R zhZ7It^3)YPih+r@ZY%89r2u+O#>RdhIE?W9EN(m^ZtwRUO8-7$yp2h?ktFG<322lX zsS`RVOoOy1McMXSK_(x!qcRX*xO3BL_XfQq86pzrbJ)nGJ(dN+cxsGsODdgnS6vMz z`1Fgi1HA2{;E6=Co3cVC4a)hM#{3A};!Q}q>mu11XKDsOTCG?gud&TgJ_4(abPInJ zX>rK{mnGk$k5=*uXk+VC-rU^QR&>`vZBe=hT;=HfQD7$;E+&dF+rF;G*a5wx6}C8~ zojg3tWPV!|*Qr(c)IOmkL2%TMzHpi`k!-Y#{Qk4``(46`QqS82jB`V(^HwVa@#=Ri!U_91KK{k^ra~qh2BIJ|iMV^AI4qRwsozaY zfA-;p@%tWkE!2j(TSE!U+$EpgEj^{q*PmG4$Wjw;&u?WrL1w%>>2{5pV^}QNn>$b^ zJd)9Vt2Z{*cdgQtpOyHNe1@s-Nk2XmzP3F*h04KDV+3C8Voxc8EH~wEvqs-vL}bv7 zd8@gVi-)gtv>Ke5F)24TI#h?91h?w@JE~!_tB2gGM>pZV))F%k1N^-F%>mvksUzUL zy7W4#sCJEQ%l-GSsK=IuWW&239;$FW5L&e>sGy7PKvw0Zuy(kcPCdo^)fD0n)BvZ{4D-kbPc7-1VnIGZ^$ zCpA4}PIP8?i@>hJAbeG$$0s?mwyZ10gw5<;Ly_|Fvxh-#3LrN$5TE|T&Hj78|GEzR zk7c${;|riEZ^^7J!1kz$`O)N?sGsXHoG+uk}2cn1Ut6M!J0K}~$i zcuulnkh3R&=B@G2Zhp!*JoO>5cSI?pTV;f4;*T&zMe~)FdKMLxCnp-+nh1Xfb2|Awml`e6J+9eV!SvjOSFr|DgM!OSY9jA!*P zPu21xRpc&F=iUgkW@f!wY1wBkOR-Wxmr$M!%j$9Cv5X=ULbiPF#AhXqs+3xdLOKG< z5he#IYlr;PtUP8;b@y{JstO~bdtabi#-mEUa%%G%80vfTDa&E&Tq$4X=3+RcwVynJi2j0u~WR2guggF z?@^{TxW=#gFXIH&4vcTLh&#ZSa18|$?pwOnd*Kv6eNr|k_{yU#EPOc|8zu8WNah$Q z59|m(7lfJG!FD$Qzla*B5)yv9pvcDWZNEqsIl8oqG+8HBKliWSx&8wb$vjGzae4xq zSNH*<19h|TUJGE^$BSiEj$yMwWPgmffMt~TvCCLVY_=RvW5MVHvsje^ zM5gv##4@M>48PE5_6ZJ<*UWG{4XvM(r7hD*I>3Yf07-v2CWEy_UwA8!)qsJ?KWZjL z!S>?lTzrFmJSaH#|4Ms|9>0*J;J$`}BWoOZSI7%V@BUOCzmVnnrNCv$ePxS zzlGY@?*f{UvB1oQy(3cLbUY^`_B{84?Qp=h9|6&nBEY1g&lS|qIRf}@A?uIs34J-< z2CNt`wN}kP_c-hSeFy$@oPKirWx#Z;XBow}?hob$?%d%}Os%93az-gV@h1j34k-L> z*ZC#wk4ezPKo_d}QG_2ebAJFk;*a~bG3!pQ{k=c=iooX|G3pmO+ zf1Yv7Bm>NrSOI4)FxmJmKR|m#z=Q#l_H#mUzQVra!REby`kVsLBBChi{3+lIH-CWMS)VIK|Fjzt zyZ?6O=W6?pkILYmvh7RHu)npHWUZ9~<-^bJb3Yvj+132m`g#tO-Y}B&x3uHM6y$%e M|KsC=|6}Zb0KF#!^Z)<= literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-76-NVIDIA Salary Percentiles.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-76-NVIDIA Salary Percentiles.jpg new file mode 100644 index 0000000000000000000000000000000000000000..eb9fa7f4275228c2559ec2b499942ac1b953a27c GIT binary patch literal 20189 zcmeIa2Urx{mM&aPlcVHJqoAM!MUs+(BoPTJIfy9Gw17xzVj~Dh4hkqyBui3>l94Qu zljNLpZlGyy>6!10_|5ar%>BiGgY6==wCKwwl6AKGFCqEbaIbIGH z7H-jtyjKK-goI#R*CnrAmEac?y85FN2oVtxDG4b985zUXvn*$?{-<9jH2@_cWCZ60 z6mky0p@cvwAt&_!47^V~$j=w>FE0oV6c-PlfRKoo1U#VV41fcHLUC}RczC$D;L#r7 zegKyekLs*|EIzfSDZx2Anyb$uk_kEFKGxD|b!~78K6vU)L`+A|z{qs|0vGqiOG3gT zqSvmA$=^^=R8m&CdFSpuZ5>@beY1y;%q=Y8R`w2#PR>XdSD)v;e*P~40wZ6(ii(bT z{U-KZN@`kqMrPLgPX&cV#U-Vm%f8grH#9aiw|wpH>Fw(u_&zu^IW;{qJ2$_uxU{*o zy|cTwk3Km3Ar}OI{!J`!`!~UUlZz6R3kMe$ic9cAE(ne@*r1fScxMIhsbnX<^P4|J`zhIfOfc{NEy?~a*uTm(1du`@;N(Fm z0cl|WkULMkfoT?V#|Z9frT8iBdCVPkjx(A7$4t|v27#W!{l$Ty?>%dt!|5p_@(foB zuZEN`GB&j}OMQN?mWvpA)c34!HOI5f?P?9(Xq@}4QNcutDsE`@yU!2%DilztOojB? zbQii3eA4!IhH%L0Znju@Q6_xv(D9u?Eq}gi11*%2D*S3&Vkr%zQ`2xZwBWtLa`>YD z2wn1;X5tt^Ygjk`yt0jb;KBm%E|B}bUCZNX?0Bba?ynK5oFCb)R9rfHubrYL+-3gb z?tWf>_Sm3oroJk|o&lMvmFmpM(7YoAse|i}#2Qr<@a2vSg(kpl!l=JB6n{m2XD*#f z?aCK&8y;3YHZfEkReP~@e@2N!^pTV$zsk~W;#|jN ze@o7#=~J8Qm{m0T?ZX&vPCJ{9qE6FY;}akP^S)3=)TwxkroYHUwAH^Y)J2!-o}npY zHL1N3Cc1cAtm4CswMxcts!NIrj+eeNs^e0ITSCL5vBM^t#i0&^3iaAVLNC9@zhBb; zI`V11_dcm5D_iyB3~`;LfgIj_1@TF?qCaeiA0Xoucw8FEo=qunu_+g!#QL?FG=^|_ zgmCk4?gNJU1Yk!b9UjM>0CIVeUc}t*yofL5Ww^(hM>(S|6)0JgeEV9dtZ|!Q)>u6Q zS=V@&UO<=oC~e44R`|8U2bH)Bmxc1yO2@-pnr`<(R_^L>6LkA3Wrh1W_{;wYb%ZBd|5ylPBC4ZFTP0-;*OfkWPhpAy>rEe zlTv|2VRW?83$+~~!qN_<;~e+ySz1RVJsoj* zEs!W?J)e)(>2{14QjCi3i70^eOL(}b(@+$!s=|GGbia#xVL994iB*=P?Fs;|z z``jZJJ;+0b8FIUWdC5z&9YGH2r0$|o6=&+F7OlT^H#!qBs_;8gyq@$-kQ6Zfw71bb zoX9X(dD9y%)rLP>b<52@-kPBeaMAI~mta5Kc|6q6|8YvZ_o3j(mkCH=^u6X927>&> zcpZV96X1(@`uAlq<2O@$a7e^-n8KA$e&HL}n?^g#yzRy+cdaKBNy8g+#Lm2Bl1JF| z1bm}p^j9G!tJ{1_E?&n}RPDEO0+1nfG*n*R`CK^io_h9Y{PDnc|aKnhy*0Pq``IeSV}sI;zYP3v4Z~-X5>5reEJx zGyCR$yLYAUTq$L=#@91vo>1&l%w1~on!}EveWui@^_$px%x{!$YPjl@(DnqfB6ts`uGHZ zVbq=`UIX1>tL6#dg4)5$bM=xG#L^X->=2@ku+r#Q=}Cmb2_WNt0#r|*00H@e@BS+e zkn-v*(#&1xKMee0N^NsMV$;ZwZ}O!O5w}5)3D8F!7oGrJAx7r`8cl_Ndb9T=Isw9k zJB~fuPJm<`6~l3F)C`&Z31CuyTDaG(0a2jQAn->|W0m-nj%UH?qWiqG;BI#Ua73qI zw@!cs62sj;3E^hzRXuaF=LA^Izy@0_^8EJ@Tz9A`V5f~vcdE2BtldJxZ1iiE#K%2o zEHL;kE5SgW$$9Tth5yRyj)V$p%o3ldardsX!-vX1XW8g`6VH}?hP;Fj}!paOt?;eR_eg8!zkf>1UnWw#|F_U8mA~Dau z3AT0JN_zb=)~l%F=Y0H_lRp|kF`5K>BUc(bgj#6Ki~L*-K{ZEZoB+e>#|(Xco4!8| z{V#hrBri%|(l^esb0JHmFW!&&piJcP37_;z{2o`}dUjO5gn$cQk0Ctv)yBbqKr1gV zG?P6n!P%ivFcjSYTN+NugMxCU#INz?1;h^Bz|EQ1gJ#XWL6EIwil0Z5s)%wx85#t}ykV@Q)@RRnrNqde=Bdvk624HBHD*4PQJ zR4BjQT;hgud@1ut=TtIRNL$Cf@_p^GuECQUx_12`Vb&F{a=9hxs{*>lKWtU<@3ty~N24jj z2Ad8$U@3y_$AFGbe?B{Tvk4QbA9mDw0=(t3?*%j!Xnt*w)hZm&WAdn<0L4j%0zaH@ zjmg$|_zAEF+N+|l!>eD9{%3uLoHmQ!oz(vb{y#JRf3~VCv#QdPib5gxQ)wx?x?0;`P%tNMe&Dz-<1nyo;3K&?=Idw*Db8Zb!%hwNWaBiD*hq$aMSiRrJY@N zr?+I{E=7FQ^~L(uV{iBON*1sNU%e!3e7GWdAC$eHF?CBigU(j3V)E!qi%yJW9nf5X zvo>BKc)Ocd=Jltz`A%(rmPpjom2K$WV5!O*rLi6SGUZi{K$)d&DsiFqx~&UdT<-$8 zp>>$k0h@!=Z&CB_S%GB9=|uqyQ+|yy{%KhR!`TpXxEyyxjrRTP>=r;$J=WpyZ^rXO zWB*?I|_Sq;Dua5_=tgC$phfb)oWeup9e zOS#EgLJ2R+bh#gpj+_bT3gm8Kr)($45Q2QyH1+_K&hp9qV*+(7j@k)Ouj6RpjGt^~ z>-P9)b&W&s!DGtLx(`{{T$zBx-{$6zIr>Mw|Fh@2qI$~z`L2nUrpt4XXUZC}8sCM?YaP?q%) zpHgf@)rLwu*)T#y8-cP}V}^!f2y64`Eft~iZal#J8k#{v5#NoxuYB?ZAkanH8w@xa zsH{Cq(o3*Nc=YH}FbNAk!0tYuC7B8Bs3O$WUxHB-l_RoM-7-T91Sr)R95|Ket_ery z=F4xEe;JUv+OzWbA@Y%5^J08Tarc#(NaW~YnCm)U6_^E%ovGXy8naUDU)mUZ^St(v z3ZZ80Jt&NO!UvO;2t(gVUl?QeF|6#{wKm5N-W;wzG!cU}5>h z{EeP24u+IeU^%pob=i1%V$R=XvNeRtNcEn`_7ZG}6l+CKQJB~1WM)Hg%VI2N0Y;MR zE}c7U6@h+PuVZba`+|QRT7RHbeRY*_IVqC{@)cR9UcA{F_p&Ca}+1!w4ALh2~}9i<;L}7Oh-M2C+(am0+FUxtnl0z z513@d?*Bwg6wGwm`a3S&$>QebV^$K+=_v|(fc@Shy>J$B0-$0rmcM>MG+BaRWXTfh z9DRw$|MzhzN1hp5?0f4US4@VPERJ#cQ0pKd69W773v!hPW!sCr!8W=l8FXZJ3rq@q z^N+D6n<>A3$!C2RwtAS<>@!@|R4gx~LrG*B!5LC^mBk0$vrPR3w#B1|syP96L~_dx z#b{Y?YGxj$Vz=M%FJ5^U$fM~CW}!ba(SJzO>JwJAW^2e>d*h8mgU4uY)(IKu8*17- z?^~Rbp@h311m{W&vAxI9)eq?wm-|g|UZoatljnZ?&_0sPUf(9={lZ~jlWYHg_JDqzf#7?Z@SAqqkRIA( zZnI6|6|QkHrPkeX8S1ua1BNj(+JxhV98ztA%=@EM5d-mixg?-fF77co(C4fRIbS=& z$@6)Z>#e398)f9EC+vpd+Z%zJ{BEpm@mN})cYa~mGts3}(mUbhWe?|VcNzdEF^d4c zBA)(_ZisU64{`OPnk(ienk{XQ2l#T&jEQedHH8UiPo_3$lp{Ne?Tyrc-_u{i44etoQadTq?n>s2i1%+Oc#SLr2wGDv!Uya$hs0W?^Gv;7b21H#cb^8jYst*|R ze*jjv+QOcwpc4R@wR8dq`k7FJk>%%>U&BkuJE!!ie#0LV*l)5SeN2{ZQw2J0?5aA3 zXG=VJ<1MILso>@xR~md@78(bCKO~&H!o$ngPQf7cHqj<;1b{tqMdsg+wcl z1RqC%%l#wjfw7{-I4fc8-Y>8sAt@E>CT78Svkd(b1;BovjkL?g=HZ)(h#R?8%F|I% zEQe9}g|>x8(waTpjdO!)WoE7S4xG%M6@my-E#tRiCk2W%uu8)6`J+F{M3tM>yF^S7@(+WcYT@!`MB1GG=#>gX6?}jNO`OR;xZbB_3JA!LHV2jP8nF*1J+KohNc2-H@G>4%ya%gL;h&HzbWA973&BCJ1Uai$ZuZr z|Kc6UofipiegQsG(wFx1%SPfZy4PAi+3Vg=WF;ruKkRE;kPbkJ0gf~l92{xv>@U5^ zxgnORoRG*Tc<02F%Om40|?XL9mToWE z>wmz|bdKNdny#qsts2PC3wjjWkK$%25I=K8xy5O{hgPW$Mj~uWY~{(-n{#>FMNAA% z##d|Cd=380kdDAsVpj;Lrj=NAZXBc9upeo|f0L=gSrSBTebsm7qEx)-eBzgaRR;r< zzX_9r8i!2u6f2K@QMkI*R*c3HgCFC16f8g22I*gez0_aWLAFeiImi{XV{QQ_l)5n} zkq-?8R!5F4J1dx{D%G1bkR^@uOb*U;+N*d`xi8G`@)Gh6f(!rt^}PnbrKG-%`pX!# z=o7t$DMpD~7eXzI;ImTt(f7Y!4!_38Dww#tqaE5K6>xumI(3^F`R%3R(@!E@(FCPj zv0JOCIuvb%eA+15>2SNQdvQSBzlv~~E2jJjvrLTy`FNP?&M~#B8ns-{PU#Gss(tJl z?Qk}phTx$F-r@gFbvuyxwGGpwfPDs&)^M}6Vo1}B(!4H~pMG_6urX=w&C~LFY4>zb z>ne1fL7kbL*l>{QayC!01+b7=k;ZymLMgbIOeyhJ4J z(1YP=^(s~fvi?fv85oYJ<)J!__pxAdh3@(*daxTOx6RunBcC&9ePa<}bJH)X^^8b9 z%YfPG#ONwKrNmDqJ;E$dn0KuSA~O@|M(kmZ3{zFJtBuS<)0#s52a_mJ>4X{5JMl`AFVZJcq?L9fiYs%Z0r5A zNc+XWgA%6oXXlZQAtdqxJ3)LR!{4^zF7?kj3p5MXK3H+7oo1UWOJYVY%WkrL&Oqu< z-!T%b%Ocf@4Gt!L)BC>S3AbiVtrqHV#LG4M<+LQ!U}I-!H|j9qD$W;!0L>JVLLB>2 z=!o`(h)uUzv9Wx%Hx}8b)SE}=woQ@!iYpKfM#+NUffvJ8{T)Y=2r=t~8o`G5yf5qD zVm=Dyb-UeoVW!y^>cXBV>e@+Ah>V-E z^_Af8lwB}@h9%FH6iT|?+Pn}&mytS})|dK*csasb%$tJ!h-t3K z4MoWxkTH%%sCCWvKPtXa9-|@0klya)O>|IB9XOB0VR*OUlxBIa`;hgvN%2{zRf{i! zpRD6B$bP@7d zxxJrU;^LmZBg<-%f7Vf5QNXzm75@eUP(u1wzJiYYsp@p* z?=(;7usa@@OK2yZ0M^nwWG;Rl0WEIcO8&;BE0^yxEHmzXl`bcI96pX2??F84!}uZ^aUTJeMNtey-=zGuod(bPMFLSQ|L* zqowz)P?IJRqZGa%kYRKl-260LHlLzJYOMCr|XZc3!H56}mc9&|Xr>osL0iFll0K2RV!{S~>cM`l&*tpPiWcRML zZx>#OR}!K%n0LU-8rOQwWY>&6*gg``)8W7K;5~WN!{NK^d^V|G_&^J*7Y@kQB%%xW z;XYv)LJ$bTQt?q^z4xTiXR4=V7+j_6Oeid|bX6Nmb?5s+wgqq75U)R;hCY42LD;+_ z9F&fs?R@M=A|8~WuG?$9Vkns#HQS7Qz>)El#$ji1Id-*r@?4kn!eujXWJ$0>s4oid zK)2g;0^rbNETLi)c<@eS`Q1(S#Lc`S9y@CxK6c7WUMaHxKEVvj{O7y}!xPcZXjdrg zvzA5sMYayQ;QKlT6WZ;rW=YHA!Xu}<`+Y!!r%e3BP%j)y`Z@ZA6ZrRZ>$Qq}`x_#Z zWMm0;YW8yTiIDO)H;hA$-=uLS8eET5F}LpEiO2bdTp^ITG>+la=Q_q)L2aBxHHM*; z3)K%vzMlXSav-Dzz`O3QI$DeE-Jwl8F5H5YKh(-*ah+iLeA>1Df`PG|rcvpfPV|3r zEFhEE(^a@K*s0p?pZaDSp16EK1(L^(K(^qtHVef!4n}aO?p)#zU0m!6uzCK?5ecJE$Gnys1+O z)wLwBb3Ox*I5h-XET6H8aS3Rc3fZW`Ba>si?3PoW;)k1o7N9{T-{o`tGwJXTHsznt($l8+)BXOXjK=?Z zbQIeBV>~+uj$JQf{dOC45T}-`Q)cu7ev`B~WoD%)-|?->MyLTWEbi1U|5se)hh0y3 z{x(yt;|&xCbh#(M@+xYFE}!#HMor!^8HNeD3L0uQXsD{jmB#WLO<2V@sO3Jegm6CJ z`1doophAGAp&&X*uXimcPpF)sA!S5`MNZ{|jYx_G2aJ1KT7$=YyT0R5vaFj_oP=`> z72$&dGe0gch`*5VM@TSfT0JL;&>R<;;2jECoqP6fX@52?uvP1LW>TQ)3ew#(YIqEZx>a4_z2v5> zulxGr1!3_m!n%h~uXpqkcW#+&Rpj%zS6B~*3M>W!(-NjxgI0TV`ZfuUw0(UIy^pl$ z$-Kl~hI*GsYuM24j(Js^KEdX>yC`QeN179N)fy6fg0 zTUk{tQ{#t3i`8iI@o6Jg8c4gx%ZBP5>K@bvYjtDTCu9lM+tn5eEkO?`DF7~wX&dYJ z`W%V!^Bj3h#I#7PL%b#+$ColNm&(@fl{Grl$RzYe(4paiHlvXY`F0Av`0NO|XVz2g_P z>-y7W-cm^QSCHi*2C|*qvIY-=7=j-Mc-3&80C)7pH{P#TFPsOLXm?@PC>qFY^)Elb ze);?t?rgBEV%Q{qX0_L^pAZ(kKF7N8@p=tgRbSel1tGUTr9MKxnqHY0H#Gi~=E8o_ zyS&gq;NKa=Yn6yPsdEE4-TnJ_K5q5PvJa9#!wGlyg~1XHao!QAA1AW$x`w4J!?3C#L5u**u7)E=Ge;D%JeN~LU zuAF>naJAhs_s@k`oa=+Yj%a=NK(Rhpssv>{X0P?`vydaz`l{pPnZRahnFRtCEASU`rgvLkeL4(;eT<#A=W#g5j07%#h-A(WZ%4rsn2p|TI*V~SokS0?C2MsKSuSVF&7$i8Q# zsEdt$f0mkA9L_}Dhd11|W? zqv{kZ)7V(QTY^^mn0P9cj75UkHchZMlrpjCsp_z-@y z3%`WRRQYuw@R_&K-FQ32ySS(a67Md-@Zg|f7SS7NwK5|Qw|m)vx7Gp*a0j9I=Ct|Y6J=_3!E zo-ZQW#*|~y z&atw@UDWWQ^w8N7Z2!mYd7uF{MD5^n91)-NrK0+FCx5PlxowIqF}wzDpVr_P@?f&fmoCVyE5g4brOwzvd%pj(2U`)Z_U8B2bpF=^@N1>`is5YdHBU7{ops`u4{!TO z*I$N=F70XKOWyqITp2w{!@EVO1AmI-y4HwL+!D9+yxd9~+gO}6v9GJm`kA zdpNCHZSkSf_Q2Mv^*OP_9iD%8xzG6KcUbG>Wp zxl9WLt*f%MUy-eK^L4NSkH$DIjQe}p^Q~;TJng{;k_Vigv}rkGL*L!4HVXG+ zt_*(>g3Ct2ak#l8J7^%~@6c;|>QwTR>r@L=mSrK`f<$MDTeuiZe=+2#SEi)@jG$j{+g;o61O8^hX*^*i0=u?LMvca_O;!UZuOv?6wFUA{srq&XRx7bQt|J#}MNFxn?!%%;r`h ztoiuayP15!EXOOmtUP7;yy&phVMH~R1aq2EaILd#W<7S$KXacIr1Gv|)TfwjNs$kZ z3`!x>#V^X7Ibg}z4yz7C^=M$Xyb-TAt3y}a#V3t)G`jP4j@lG!E-c@< z^5AoX#@fB>FxOJ0t|Ft$nwxAX>8+FHB+n)fqIOovo%q|l&0@@JAS}UQ%erhAdMYSj zH9XAaoIQRYI)|g8uU1>S1egd3zyzi}WhcxQW!R?S@2v#UZ*mezWHXKfEcED}m|vxW z;0cD8uIupaww+ts&B-lNy_nY{{x)=yae(#d+qsIyb&Znls*8w$eNBn}JuD@a3KNT9 z<`ZRtVh==Uua>x~3GR}%zJ*`mtGGKwZg+cT;g^X8W|Yy)euQN4TMB^N@mNXfcLLGa zgs4(Q!a0PmT8^U3bqN-y>T)I8l-hl*UN2C68Pagj-V#3&@wA*{tR%N~G->sjK zSVL&S8IQI>%mJWQ+7w1j5)TNMhp_2h-qE{7@9OsLTKJhOoPf8KP3bDSZ~eLl z^}Ff~9{8SC(wGse_C@!P!9E{I6YDl}lr}n^=Xx?B^c||s>+#v1r?%2BNsJV~rFVOd zgZ-hG!1{9ABV%#2dV?{&7%X^~)#B*(0bkx@n|1A3eY)jK&-QV4r^D)mAR~jwZKcXh z&_}R>z{f_foHk*D~3J(OkxDo>jargu2dEq;fN?DRODZv1mV> zw_mt4SS)YB<(h=WEX&?o3T;2jyt69F5u5>yvkv#=DKwPyCmdm{$QS5=dmC#$lSs!4 zJt=`I=iPSw&)p!s#O*0;fxb)Q7G3z2twsT#CRbq1ooCvzLKZ(g*7qfx$uEjM^l-@K-g7-hQkExoyu)c1FB7rX-2ZskCoYEn(8q4nHcOd&be;eDLWwU(9%i{OFw zA-WK$l8t?h*ga_?C1;^+=1SE&pB*Z_J?5ju$-A%B5a4>KIoVm#4OKT_^mmDKG6%Twu*yl36`q_jVOD&Z!s zz4)ws2KL26p!?1H!38jvF9@kWyd^^9B9-va)_8<}|H+q#Z;ZolJIQpd0J!n67nF;6f`{b?U~ zFH_f&cRoHEu5&NlZ|8~W#)yj~Pe_s@=r+s~o;>tZ7FH~Gtk~Yik!OwncpdMlQiiMy zOrcAr{rd8spOUB58o%|iu&S)8q|lG2yt3Nz%^Ca{%P(t^8)Em$xS?G78;!nGXBqm1 zvQ>LvBk&YbodwsxqKrj-&zBF1#XREainHV8NlkSUD~LZU01ph1;&o1N7uGy~Q)1)v z)>ol;_o6%)Al{c|8-vn~zRQ~0ncDKDeU_)W%)sp{wMOvRDSJl)S4ZT7@lY$mo=Oe3 zSxb53k&$vOt}{(^@dw=@1q6av-j*0Q^jae=%@cstR}H+%G8^T_{tJGN(!L#pcuNX{ z-h~%*&>l&?0H{c(hjqL&*u%^ z(gch=h?c{wJ_xa%cQ9Eo&Cm}Kv>E^8V%FMdA5tl{tuLg3Z=N!_%B=e-uvN2rm47w4 zDT<~G|87l9qpMvL%?C6}sU-TkX^*v*@~T3tSR+K2K)DvXUIf>aVH8A~$t5YMF&SNG z+vz=uPN$sB=bRyL;@sKI60!yFovp-DC`4rFE0=h}{b2s-C3RF|VdlRg!0%PlbvrVL zm#Bv1_M?|tNmsloOsh5simgUzAh(xq_N?cv+N|YF#7=bu8ML0m>7wq^{)PcjrqP z{%5cdUM3l}3=(tTlRB5Gd?d4t1A&h2LRJp>fN3yGQl;_KHTaTMzWl6Bp36_@VoKt5 zp*T_1pl9}13oqS>3=JL2UDxjgC!el!bg0$Zo=O$B*(_3&QAg{FENd;zu(-IjR6CFA zH4g8g^TRBfd9(y?S$)<24Gq@zxY{^}?Z>()KP)dAow?Oa(okwzoH)p8)kg!-BD>l+ zw70ZHa*d;`e0VP)0)RV#qc@I|O+jv%as%7IbT4VoLV6+b=7D?s@mPY(VSgG8P2$KPg>zQF|3X2~L*7JEy1|&duu@Pd*2dtp6yd z<2?Wi@f&kRE_;=CZhWZtJVf7?i04$Za8~$=Z_r(YLYpp zVOJU{Xz^c8bdXukf?@8P~!|b#&Z3yRNd6@pt)-xJ^8{1nQV}-K~$^Q;|F*%GN27 zic$TX#Eiq)EY4De;fv8ChartAWT_Y3p1v=*oe6EGF_mXS)7k_A`5vb5ticuw4{O@C z!W0UcW=35%AH5-+=M|F44HB;AVHQ>;B*nJ;?4&^p>ROW7=7B&iJVQ?xzGwZ<~9!s;#eu)@d8n2LED*J+8V+%6fK)tzh-gP_l=D-)rJ zMzW_%UW_jo@G6nJIp5uBCxhJF%8?FaV%Ka%f?o8G%C#2G*f7-A&Xr`oC5ETLH9h@8QAQU~gKX3k{*MhSfO1`VRwyVkxWK!Z2!}YB=-pogJok#u0KkEXJlfnN5 DxWw|h literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-77-NVIDIA vs Percentile Data.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-77-NVIDIA vs Percentile Data.jpg new file mode 100644 index 0000000000000000000000000000000000000000..05cd0a1494f1dbd6981d74b1e95d9058efcaced6 GIT binary patch literal 21654 zcmeIa1zc2HyFb2RgrSjckWi48?hq*f5tS|#ka9q}9YR3q5Kxrv9EqVzK}wMh>FyeG zsF~k*-uK*B&-umu-_L!|eLw$u2iELav-j-1*R$4o;`@BpMvtLqfy=j5)Kvg1EC9g5 z`~m0*KncLZ#RcQy;DNzle0)3tB1&Q+LP8=satcyPCVCcTCVEE3D;)fsSJ-&j85z06 zuk#At5D^h!;gpn>5R%~+77_Y!5-fatd?G?3T4G{aAy!6Kq5sb>bPGU+hqZ+L4ur)9 zV3T2i$gt3D01M_maj<@S0e^pCVS_H=;DYh+2?#L_*)5|;PMQ}*y%U5A>@vjpSlis{d z&dkcr$<50zC@in2gjZG9)Yg4!ZTsBb(b?5KFgP?kGCKBsd~SYWacOyFb!{E7w|{VW zbc{SX{n0Kg0Q9$JVc!4Nus^hm4AU;`OP4^Gz(3lBh3$bUAhJt1tT%AU6?MRmoG93Y z0`Mqr#%7d##Ag@2w@dZdxu1ZVLuBqM;z!edYuUfAVS)dlmi=49{?V>+fCz+z={yh_ zAP*d$a%K7x{GSR>chl8#jmG{`$8*`a3pv|EEknXvAN-o2RET?NxFO;!a8iGV7D`Dy(2- z(>Xj=@UgTYVDaH_(a9`c0f%Y@;rpPh0hvY<^psDouUSgN`|79cml0nxoc1jMir8;{ zJUnK|y!mKSB@YjuU6kj+nFr5NUbC(q!-u?WkA-M&;LwPt8=e-PTopUV#M95eb;juPo4h12sq^){tf{1CST`pY5$g@`pwR0A^ypOk&i zTd&5}uT-w7B7u5blXN4JkUUs8DDJt#C!o;`uO7&4)UtAj6RSy9B{)By2xrP>u>%eY zd3f52*XAPCI_ja{XBfO!sJd2LTl>vy=N}Bs);~@)Ny1f*QkR&uoZVMo4pMefD?TuXZ?};Qf@KJe(=<5pV1q9Qe*t)iQ)~fw=dT3;61#l`pQ`%6wO0>^ zyPd9Gv^suCnF7O}t9|xB@$KFTUtp5c@KYI}kkpDbJyG~VBYC2Z(tSSI>U}jT?o_ub z-iql(+6G@D)flQiL=Z|3?AFSC$vZdL4ep5!*AAcd8W(09^lI-n20dea$w>YTT7Tj# z5OHqh2A$)(BLL#9OD&-47<+Oi^3-T3X5!r?XO#ggrcTizTRD>5>0sAUBfVD^w$Buq zpWgK!lMPtM(c6UuhlV=1y{n4DeRsb>oTO^^L?<0Y0CcS|#!tp8Ff-+tYGs_O>{>O# zUWgj2R#qp%#@!S^Nk=1D z;NCCl^wQ+=aC1Zx8j?oV^%z1T=-9#edh5dRYOawY>#cfXVrXETO`glt%i`WE6 zEd^#T#Ii{EoMwM}hK|ESU0qrL)TeLET%DM%E2a}%JqLs9IG1~sr)x!C5| z(HL1ePtMFdtIAhw`MBWG?Hx2`kt+M>oMO&P^nRgSK%OYVZ`s13J{;6S1@Nz zx{<`=Qn*@0Q&ITTke_dOj9K(QcgrK9DuKk&0yY`BaE1;uH`Ii?xvFZyr z!)LiSAG8UIct-}9aw_^ljt)waxj>=2(+P#$5K%+6sw%;?BFNM1L?K3OIz*q3TLD~J zp4>C=vuk~i+`zb)o6r>RSq)fHxn1#e)Jhq6z17pL9Rw*ssFJgOVK3pAi$E7vg1sq&BQR<))KES#6QHjm?@ZofV zJ5I)I5k>qEeSj~ghuhqR3Qt#->FrZ!WuoTm$bfFx;5kOU`%9eFvyj=Z zK^_$t85D2l?}-lF#Zsoc3zjWK7Px9GR*8i``a>r(R{CW-gL_d~Jd=ab`JaMeXIkt{CB^WJLnnMHUK@MnO;{)nu z-RWsDj^S`?yNk!u^&p4qA2P#uiqkX0FxvLNwt(LK+1tH zQiz@NP*yYm!w94=xz=dC!3Pbnc0xC%pa<7V^_pme(11_Dy3+&X8y#h;PYORxN~~2D zQZ%s0;DH8AL-(jgM_G5#07|hMdN>AUHTCoRrjJ@9)gIc)+r5GjNCQSH|09t9Qh}7} zx20aKgZtDo2@Pl01d{O%(hz(I^qf5u9o8qFUll1>Y4(KUL)_F#hScfB`h~3h6+(0dF1dOM9S@#t>RcI&l{ilY#4|}G8I3rcVZMOo>yQz(Cg3~cBGw} zm|PPZy{lQFWR>4GiYr|IWIz$T7ORDTY?Wp(M_XGvQSN_%Qe?KxMfJb0W~xmmJWge6 zfav+js}1xVB(`{M58@5uFLH4ff^~tfIfljdk&rp|Iqz^Susp*4|-HG@n(BAqDioP6K8PM{!SQBK%Ml6INngz z3VT9)`lX6k^^iw?aV4LC*6T0Uj~t0!Sa@Fnvl#Vd`U|^mG4pTOkH{kbjKpKh$ z_J}cKqqOyeU}GZ&2}J`}CYnz-|gS=2x~Du+$H=#cfPKm1Sno&et`+! zbX46OMIssRw{17-nvioF0m}!yCoG~fr*n?yk**2949VXNFX-E^w!IFvb>*&6>L@>p!Mp*#7jJMY8aTn7!uSr4&V$_!JALIb!xu7h;m*>j zZ*K0AfClQ61`nle&+#zBc8B-(vHMqTdEyX$ulHwckF_oMGxAzE_DeDsx7n8p*M@*CeaLogFk965G0RB) zx{-U-GBHnUVmn0ddP4HHOd2OR(PP^iVI(CG3OD3qYiLO3a-h&WyIQPPoGYjwxVT^A zft@o@38l0-_0`iWx5?z#zNhuoPkq}4ly&*6)o!#X0AvyE7j~%YHp| z3_aZ99=D~Ybxg!F=6JK<(jA0k`BrCm0D0VMP~X(IQt{m@B$?sEm_CT%b%~HXo>wYM zQ$<7FjE|2%8s8^=?5-FUaC*mnIuRNmU-gEDR=$I1X!sjX8EG?f+;pCOe|fsLbkVi+ zjwz}6a3s8{!p-7{mcx)`bA4i&r%%k2CD`pX))zw)MaPB4r#&K0K}59EE!IIBrj#Bd zaW!h&=^nCFmY0%04MH)&hWy`xjfelEIRD3x{5LlwPN-|6a5B-Olw1xQNV=hQsw%h2 zLW`!yaXQJTu(lwbktEG}pA-(!P4H7&qAW|)y7m=6yJUN#sgR;akYZy)Fa2*8DsXo` z1`COYQ&0k6ZMC?^$AJFYn^7~JWIeYiu7LJqhrn`ZwP>TzUXNS7#tn!ESFk1A{Tin% z$Qm}Mz@$ahV`6XAxDE5L$e1t^`aY{AIw5$k&Dt(FS9Ko29N_2UWoJ=e{`w`s>*-L8VxeZ3uxk7OTFGmuRE`0 zRoKNled}I}j+L_eI9At5VDuZOPKhEY#knN-WwA(tsoliwVLsSBQ`v323692iFib)Q zG??@3IS3(EpN@Q|HX!;%OK5%KenV3jVGRw1D%abFZ_S^~tE9-*!hi!u;&aWubDjC? zmD0413k@OH5}Nt}W$tXED3?k15=Sb}YV}?B_=~KfuU!lhQ6iBZGNsU9RhQ4t+;0hN z2^YjEfNs%oZZO%L-|6E&*UGaw?Fij2ep27Rqul1y!+mm}F*?cJ>3#6J>ir(T7U?qY z9z3t{ebXWg)qYBg=?Sam>b-1NCPNi9`Bd8`CS=~+4!K-r)rJq`BE`YML2Hsh?8^Ex zeEA%&lO}{%24doqHf+pr%rvKA3=V(SV;OC+SWhA;*>Y4egO^|)-6X}Upaa3@H~`t* z#Z8EA?LhYGVQCMlj+6bVb!}wPJUk8tq?~Bb>7JS z2So8^3|Vs_GI0 zq>oDNu-o+)dWl_)6SW6`2gfeB0LBil9F^nm%AV4;hGZb6x@WZ^cvpIyDy}>vp>$*N z?`0?w@0b?rck5-CvN_%QT6M}F#$xp$cxEf<;blSyEneOzY@r2QfwXROm*ef@J4}(d z&1cPfSu?|U>2V%AXN@S#%ygGVud)&8A-90_MY;}b&n?|%%Rb@59$9c2>8!( zq-rws6@ov ztG@Q&RNfHwn}11}R>}p69~5QKB5d>Cs>+j^pj)B2+>l4Gu4=JmP<>e|f0Z6p1fggk zpD9KVJ*;D0@;_KP5M3P{pO^yp`Q~2>Tu}=sI~gi0ETnpX0|KyG(n;9cPN&mgYt-ZJ zZ=I{f^<6ruL>BYG93Qr#y|=-)19_wkIBZ#~Lc|>#I{hgQ+49m!jJlpUGHpE9%ulNg zl8$%xl$)=kj9OJAN+K)wz-Gfg_t0bhL!AFxno(w>qj`@p6xBG225#Cd*%$r+lS7(( zx!BGt_VYM{oLF#s3}f&vB1!A6jlSf2t^fSGr>JC*>3VFJK)g&`#moM;RkMt~6KY+D z-=s#Z3O2@{MD&+txPPF-;vy*fCUwAksDz@3CdNRBI`TUc7!(#DN2+R|fftE8ZQ(?b z1(QY$s&cjjT^+QWo{XcSb2A}goO6Abj>C>OS0?;^IPY83od1 z!HRkvfg%E1DHadnzUQPFyioKkm*iUWzY#Twb?5(n&gIP}d{6!_Iw6}zQ<@)|kR>a% z*01ik9Q$10%N>eu&i9#&y*MwNO6ipBsmtVai@{p^ADT>UiNPhRdrz0xx6&iRx zbI$eaWfT~T&9NKEE)}pDn}w(lHLqz6>1Uc4E4w)Il9%3lz%{lJC}>aftMpiYIUn{+ zsElkL()d=8+927$0QqN7t;PNgi>~8qaC>RZQP(NUZjrN`_n#w?_insj$UhRz`pts= zu%iFW@@LGdQ5dD# zPaaJf&HSa>_p=82@fn=;W7HcK`LUsa?;8U}J(1d3cmSb?r&9&**~QQo^TnNQi<`}$brxYLbZYg_u9Lz@tXd*kzaczN**>|Vf= zu9lY5(f2tY`ctL(>{~w>dp@1AikmFjF`YeWeA%&LjN;OebVUQCQ{(NyMyF${{4?*q ze7s`09|mkJB}%Y*S-Gk8Q9rTp2e!3z!p-|v64*lTSH`JI>rz{Bq$voi&l7r;p3CMN zPq2FB1jJG@TE?36z&nn3Y!~kDHo}W}y9lPTR5qg{u*c{ak9z<&kGiIm-13sGP3p}V zSG8FRr0&aY*5pYOWf4bwh8JNi1fod+5Z(nDlPiekj=RW9M`4hQE9)zOivSzpS2Nlz%R)n=Rc&9CYPYwF$WiS z#-Z34B2N96-`O}aU*@aN{cO{=)z3#7wi_vauKkI!Bg-WMuX}#OE9zld9MHG z^G)80Jhm}o`xC^G>42D-XQ0E!Bmy4C)r)1!21&4Ot@#28ez#%h8 zcj*#w&9XoXB0Y-F+)EzRcz7K;qa#(T=|rrK`BUqs4cE4pdBk5))96ktHA#;797!H` z*cDDQ3=8B`X&D>Rh$ZgvW^z=My`}SZkSt9N16CZ)W&N0oobqi>-&FXd%OI!8*vc4j znL`p;@Qos|ceH9#)feksc#khVO4eL?cJL7twrZ13V&G|UuRd->*4*&SG^OGd^;Lh? zn5cJAUZuHVT>p|8M9|kJ9%#qNThwA&kNvleez~7&tAEV7g8EVaP)54T%>?ozb~9?t zN8Sg8|LNZK|ISPQ7erYk!i-0W_FpL)0IjZ&$*{oBp{j}H1bx2Zx#sZb7+ z$Xq9D=kR<0xj)F^FTC~5{!ab|MY!L!`TLf~g*4063iwVlL8D>7(!pC96*v301MA&~ zbaVTC#6|EJ#*8DD6x*Ks^~5IO@(*z~%-vg-R?5*t8TvA5B~Hqf?L*5@;+(G@%bw4B z7;n$wv`jNH-*}G(PK(NihNGiJsK?bV_r^zozbMe0s}^--**>;mGjy|RG($H%uEdCPWe86#rA z54oujke=$$Lps=%h;=dHi?^}D<*J+ns@2E`dZ!7^!3~PV*YY_+2)lyYDDb`~>`9t| zbqvdGtLxH2;f=Q$43$jH_`Y5)pPRaT*iqBFq(Fac6#cT2vZ`sApN}8H5o?huk;(mN zYLXdwT-sEli>Ow0Uq|Y=T`@1__DR~*Y6jq(kB`m z=X>F8`p3p^86EDUfp$O819Qq5`%xp2@zR)6^LR%N2eq>J`UIxBH0^gocfqxvi`E8h zo5yd=jKdR!94>WRfG#%YGlJT@nZB;!G*#qj2*2lWnqn7bQNUVmFdAneU2i68u(p_; z;(Smw%yGl7=ysZu=QqSL1!n~A@dESfUJSPK(9^o@IZL=H(nT)Z=^3KKwm^1JZfzIW z(3@X;Y*|akbc)YXaxv8@Xx}1l&SbjL{Wz5#X_3{OQfjD0 z(ZyF%pCEoG8z&+@0_A-4+-n+Z zL&kB=2DGjyv)uE9aUz|(vTd8BoBDHi?LDncQpeQD93slJF3UE*U_SR|wy#F?hPtM> zD4(emWcLL(5CsH^XCPRHRV3_s6i=FUfG zhrR;r#|=5FC)m*dJ%R;P$nvs@J6yYe1ruhx%aUr%{7NeW3kU@wywSjniAS+1TdT${ zEVObWn&$jQ!5lvoFcl$};lp6rR&h#B%NAN64v;Bx-~ zpQpIQ(Y&m$#7Ltc=C)46-Ie8-fI)w!wPav8vwX8bDc5|h+`QF|E`(}IO!y75tFztD z=K(cJpwD63qB=-NHBpZsjI5S3eNBPERZlN(UW;8d;c-)=TF#@GH@r^k3ScS+;@jQJ zty_JQZ6oy>UiD3P!6v77aSD0pzkK$-mgUI@al@N_GWQ^`zUf5-{rCTkIr{&-LWJ=n zQ|4Hx%sUF~j7X}5CweU(rm4je7Db;v0jtPso-KqO(sO}&TuDRedm__b3czG;ug&w> z-WA;{X4L~4v4ckYQc0v9?(05AI?eak8xJ_hJev3MIb-5}BqL}%v1x1S6u1KFvxY%U zMhh{_EbCIF_&n8Cdk(~6!9czo_65Y$Q}~uug_bLu7+C!)sG^!+G*hm#{aWwt|^lW{W`5tbh$>avL*{!0Bw5tvMU)8wZ z$94Kad%ZV#(h#Jf4EOg-i?>&*8^pFq$U?dv1nqs0yI2LWP2VrF9j|&Ll3hN+QDKtO z^6GUdw0f0q$I186zA}ntN^Nt28gUcy};7gLz@-3L|Nb zi&W^gNE-RL(w-62w!L28AB+9j(JD`Q({BH}dr5ac43)Z`y&B@3zKdhoaZ_<;N4t8r zDUBpl?lsD?f0BG(o!x4xSNhfAEQZ~fvhapAsP&mTkxiVj$t_5C1f^sLh8Pi)<8l{y z>+Mz~s>I9B(eC41pb5?;SH9@Rur(V{V)+}Tg+|94ohKorHNo}ts}}X-5;{a1satf| zAAY9$y#9f*VXgeiDl^sp1$_OqY`)_rdxtxX0u!LssQnzJ$%206GcW!^4}{frqk$>b zwbPVJ)Flih$RqK0UXWMPXGi4ep!{U~Nr~5#H|Ni_eL`?yUTq@B4=Kl8F!U>ddcC`mPHN|C^O^C6bRTG~CSK5&hWiFunjoyJ_`HEAt= z`AYlav#fG|q6H8iJLz`*O4?F#=WcHO0x5+rQ_zFKZN6&=qGe^n(g$n+ht)Sq3szM| zixGcW{xHOaa2vDW?o8Vjd#lvUOFq%DL$0?uYUu{kW5FCOuSq0!hO7fjSHdgN0F`;G zT@%g7)L! z0qcWc@RblW(B$`73AYv@t+T6%k_$2^5A3Jbcrnn_oukRs##lzTiofyRgA!}|wVr&i z+b}{PxN^$MVh=Tn-PuXvUai6sfLrye_%Q*{&tB|*AMbM>#R=_%(Omd(q5_8(bvK$R zniuuMQBCn^K+$fG`%ehX#&bq!=M>9@6LiU}%S?QC0j08vS(t!HYH%O@4I=p+m*V-_ z#A`k=kppiknLTlfygC>2ZZ-a^kKr?gS?8<_PIfsYOeYLk^PiVhsTZ~<{U*+ZkwaL*=+jz^B*gKNGkkPs*^Vu`8JB=)T{#dB3OrASV+GS9( zv~K#AgYr;#r&qbG47Ych0c}p zzhtk{cTy0FlNQe^MgzSi(HFZj=ekaqyqV(zcR zUv@aBlbeFdgCQrF0K zZs)C>vuoPqdwaNDq0iF z+5hhDhiAS6FF9-`%DzXX9(H?L-}ad=v9^{?$*wbxb>RrpwSFDppfj^!7NepT^9at3GEPnK_^uk>9;hl$Gf( z*^5*vXI)pV=yM{w)zx$oJx46tBobzyp4)mLpz+&6blU^yZfT|C z$qr^=1m^#AcvR=^+Q<6JU`uEDUv>;ZKh^a&uQZ~83p9W+k@}{}0#@25`jgbE7}$Vd zLj?__9-{%wKN=`eJ?AN5K>5sIbOdHu8{BP3XR`ofQqKZA7ol^eHZ%Lvn6G~vdH7>y zyJbOpebB(!k1s|_>%wFcR`8kp*j0;?f(K`W=s=tWs4dco`>X@5) zAvJ*ttm{DGin_Yq1xx-c$6wB3%0!9PZGr@8G;lkV9$xRh802f;^b_J{LF##v9>e5! zFzy7wCMHuTjiO2oy4dJFJH@#2f6-38i57a(w-)vu1&Z|y4Mi!n9i=$hKR4yakZ?g= z-f%o6*kD1@V*24|?xf@L?<7}}?AfCgIHOQSG3(_bXp@t^S{${_*?w z`#a&x*G7GBK%$Z3RJ$@2yBh$kSaNj-KE~xtDh$?*I4S=?2UPj30@%^Jftdk1Y389Qo&hB z>^!iIdFn+P2KMPoTk@X1iovo-_3Fj0Aq_L57&bjfJhTdn(iAZ&&qtTcotJUs~m zE3`@;pYv28^XxsW9J7$snpHmG7bKgKVS*YOm(n=GJLV3S{n_deci>(y+3 zSG|cx5_T;f7-v5j7Uf|mnkcJYw#`v_YB5Ay6m7TY#v-k?%kpWXc60kXuK;WE9D6`Y z{{zbcFWh zYiTKJ_Y#4IH}Q;cz*Dwn`xe}I=;SQ;Nro5VCn&MZ+tSEHO@-%g-5oE-_hu*u50xis9Bfeli%F)vYIvbmg+mC9hTm)k~A0_r)dgS9| zmaw~GP(R{Jn%9%xDk8qi`Mp%nihE*ksaGx?gzN+B{=RC`%xOUk5fxxjxRKQ^rk8Aq zl^r%!`Pm(Wyzk_oEqkLJ?wTPm#8wRGQ=G0Yfl5P|+zxn1V>!LuX z-^7kc_`&cb?Dff?Fx)U|%3>7iy;_$Lavk(xvVq8-GBq`!&(nLy5Uw$GUl$P^(wIA8 zTrXnem%OdVxWd&I@2@g_Qj{tWp_aWiVCth#GrVb0tM`aV&Zb&|{i_xDsu#Pz_bA<* z!9k`h&`z2`OO=gPWQtO79Tn!lI~K0;2O!Y=95ZBiJ+d#rpC z^0mX7EkED^)jsVm>|>|I)@KXk{rQsIcNRr?FZtaVx=Ic;{0Eeu?v+$x`I~OyQa*E< zu9U$kR89BHYD;AvDuG>TY#!z3FEN;q3{#uL#(O3fMR_Pn6R>$Td~PO^J#l!z5O#=7 zm35_);Gxru2-ez&e&T85YnrOK;*r7ja%HvyEA}hVpBi9_HHejP#|Cx2qzJYTL3f6? zc{8g0gsv+-kmolJ>I;}en0mC*<^;{<22dgzoOAMtqVU+`KjPNNgONCH8)$&W#C+Z8 zD76G;%Kg#SwNI6soMgV{q~g7G-(6fxFxn~j@r#>PRDRJSjw8~Qa zCa7AyW{T6?rcG+%3B5bwdomxS#p=kKjH{l`5rKPC_!~4W=b0SeZew%MxyD>kk0hK} zK%o+z{C&F;>3duq8@2V7&LHOdM~Pl4ZJj}gC@D5_!EliOaO>Xu5$zS*ovmAs7cBUR z?DDf`8Oc>K50z1S&WY6Xlhzr+uolZcaF?5ikdUi9Rv<&{zYmlsI1qSA&)n?dWKUsg zIQfxuoSEUwJs`q-zt3S7*61Q9F;J@CGEGreKDbm9q|#W?P*V{g)I%=t75kW7Ru1lb z+%}Rb-v`GaM@L3MJ^XLE3A-SgH$opOguBdca~#obD9{u?zm9x@LQPLg3Rr}!n_nA0 zvRBpcnvozVCyPk)eUJ=mn;}_~%2|H1e7cP4_fH-l>0ffgQF55Iu({;2V`h<~hOtx!}E|>PK%&+aab~DOn7rx@~XT2ZTI1TsHLK=5z+@$xjlzEQ-w6 zZpf7p7YNyI)y=)^O@RBe8nZ8Nh0Wog`J^9FSyg{45<#DXk4SDMW~B-R??ITPBoWA!(U>o|S|3WyCkSgwVJG|3P9 z>dc#ycNN}Y9e^xogi9<&-tP_$e-v)P$&Ix*q90bN<^H8jAjc*~jaSCv(hbw@FeiJq zhqqBdu3o(by*Y+ZVw0|U?U&vvNp=WT{#Out>yNrml`6SpF;5qOeZZ_SNx6xd9v5$n z;wkWbXWv!A#JA!*%Uw&`b^KYY_*`xGD7^Sfj^=0*jUbm#TDeifm#29#mfR8xsfWw- z52T(~XRSVGXdY=W8g;59>&oFz4XY}Bvsm`>5oJ{UtBwUzlK9frSpizls!@{xv9S8D?67G;s8jDP`iy^*!@M?GEb#Fy=$DW&( znWk1FI2HTgJn!(6?$##uE1AM`-gPIoLY{%VB&n5aE`-#ot>$hKLS6pIj zuaDM=3L4FU*pF}LVN7k3#;_ZcbNem#aLqy2FJ$4JN9V;>W8ZsyV{ia(<^lw`)krxt z$;<7zR#Se&k+U*|#B1&#f8g641yJdo>nM6ed-VLfq9SPlxT238NWq{LmYI-%`| zsxI+*9*}3sp#4Ha^!?20o*A7|8`R@sYSO9CLiBqFLzVE(;v|~fBJ!o@V{OJXRJEd+EyBWd$n``- zi+==ibKujdA@Rj>8lk#~6Bi5jEK^Rn+tFzss6l?-)VsIsImhlctiaf=H`kD`W+&7C z)`Qq%CVhWQfRrk9e!bAMqs{2~TX&Rve`e~o=vxfRRKBp z>|E_PBmA{AxKRDb@9B0iA9;-A0=MHTZ{hGRMlAZ~fiqT;f%Wqn^IM#stB%Rn;%;+P z#dp^PdPX>ZnvTfJ!^GwTwT^V_saNCYnScD|XvO9$NA zmgT$IwUKjqhwEFzG9Z(rGY4wWbzF`v>5u6rsem&Ro{@uRA`=v%*i?R`V%GE4_X@WOCQLVVUphU2~ zZ={il@Db|9YrZy{n!XvR@p2Pl=EzEVbIH?8@-WZ*{MK%JN{(&v`_rhWaK0IN^nrC|A{)Zd`Pu*q+%*Urc7B+5emK7PxC$1_!{D`Q&L?XHDPc2VwE{Jq~ zdUzIjb4I5=g`eMug=32An>goFW_g0OL+w{6#sN4_(RKPGXj#=n*A|4?pd>o0hL(0` zq+zG^h~`!1`BlBj6eGg=c&;0+&#E^aY}J_2<{HSZ+gyr zbN=l)GvAprGxwf#?}W9pvPtq=d6T{0{XEb6?ENtPunfSFlaiGJz`+3kaIinX!yMox z00jvN83_>u85tQB6$K50022cp9fK4PADe)ZjGBs)jDmuOo{f=)_8A=o1ry)XXB?c| z+}zZR0wVle!femExqe;*4iyy@1090|6O)ANF~wu9|M2ah4SdjyYwh=h!SiiQq*K{XEG5ga`HBLsLvL<9ubtG!^)0}yZ# z@g8%&K*Cpfi%e@zz~vjAfkG!y{*6#|b=Ozh{l_=GQC6Em~2 zb8_?Y3koYLtEy{i>*^cYJ370%dwTo6kB*H`OioSD%&tII*VZ>Sx3+hVPfpLyFD{{1 z*FWQe1Hk_|EZF^@1N$v5Tv%L>5D?%IkblMn_s9j-U|$d7V@@Qz7b?ha?eS^3d{GD_ zqBF|Bq0&89Jt8!A7(pYV=U!nr{u$bzBK!LS^ZSP)``5t!64xvM10D|c6?Ak$;JmC_v@{$S zt)Zt#>&Q8(WXk!*{&26lOit#C&vmr_SmFS*X2|&%Br%#dRl*%E3hDCPUadQs5Owo5 z>8149@DVu(Ica?a<&zn!J}%r|6m6Ii?G4}1y91kwlgc3amG?M^#`U?nC`^j-?lhE? z7VOqel2KOOD}Zyd_V=UrqtIex=wqjqk%zgp0xea6M(CnXExpOowCg9hDc>M|)UVNf zLZ9E0ff=^yN1h9u<&lS;fuEeJe8wj%#uoA&4^Zv7QbufG**rvnDrrL-EuA4CTxg_n zji2nSke;6OY(8?jFyMK_SAP1$GozqX(qp3lvvF(Z6e|%W8O(dXQ6cW{&zD5tC@bDe zlcZyvgj`iZf#V=mo9{3_4^x?MCWGtU{<_ZC2$=F@?3LoKp%K8yniMIHZW}=N(f6JU}?h3 zyfPvb@8P)NW=9Mysk@uunx5K2P@g%Yq|0Xj5!lJDV(Sgs@&`F1u8IuFpaod<5ELi* z2LRU2wd8s`vH!j@WeKw3dz9-{sx!DPZY}l|c8cwLj^;^LYjDY1FJ^m$?)T}>O#`(M zj!&rSiorsVNng;4mr*A+=6Z8#0!N@mG(ls1_amPk0eUQ%bgYYzQ%t&59`JqMTN{s+ z6rfV=I0Zw_co|*{{_yeKy!pvSS!%53P6a#M_+FZql?U!6=H zeN6o~Lk`zfI;57MfFEVjk5zL%^I}^M!*6AV0GR$SZ7iKm9_R*Y7a_+v0edVw)!kp$ z`^lY5SpYoy6FS^COr`IdBL)}P*2oQ=gI|`d`c3z3i7^AN^o)k9?T}qhUmi7&m-5CU z7YYKz&OjZ668%N%FNBloi$|k8nJ)WSGyf$ zFHS`#T14Ds6fTH47(h^ADM&Sw+vxhefO+obgiFE)TSQ6jV!%boMz3QwKCzgvWZ%?> zb5BQZ^Sofs>4ueSUT<1gZ?+xl)q5#i5DoJHz?Hgv=u@JYZxu@^x@&pv7<+Mve)y44 zAab&2dh(`dC$7JNX+pQWftuto8`vgCzbDU<6CZIe8==n8CpaQKo-V?*+df$fln(_- zO|0qgus-QN9)GI~S)ivorp*!5OrCtywtbO1U?1v;c4gAK(9HzDDTWM{hA=z;{LWFi zx118nk8v#7sE#9(m)FealFP0)YFo&GYM}Pwd}vkQDKqMP=31~T8&*8 z?9u2Lii*g>l za=8?xnP7bzV@c;B@A8NX?o8P6tkmVGHQ=lj#UgrJ@bwcRCx+?rewm&k&I4EKC@_n> zkRA#G(TCx=c2rt^6O{UEv{WDAyw!*VZg;i>$Q%8y#1E3AjWBll)Bm6geJbC;uVFxDDw_7iEW$kPb zvr!SNI6KfVp4sPX7^PF70j|tnHV>OZJ>vySJ6Scm9#uJ79M2KMZ!n@y`%i(LQiVsX zt!bbt*&RYl+T=kRF&5vynz+#j7i@Q!5FV(yn!iYV0PK%iG&n9CENoU65)YN^ne#Yo zDpfSakg^r;Z^pnkJc(y8Azh85fA2~uf(UpoM#_52!}oEY_Ly*BA=$`EsZw88Lci+? zN*^Wfaw@;D{eatM;H_5SwiqrqC5!BF=i;-m!))An`m#FC$HK`iO7N`T?x}!VdPx@S ztc!~~D4PM@WCF@uWtYhQ>Q#r>77ee!Y)RSLI=vkj7Z=;4bUTOuS34h1(<~@_SE^w3 zF~z8^0_%1R!^zjMBwa-$WK~j`A!MGUWzNi@qtNb`oyG?&$gQdyi(I4Kw}froNaU37 zMSxqt6m*&9mWbqCv$zL4%}$9jy&zT9m)0=E|M z8*d>VL#Rqw)Tp=O^~V@?T(27RCkJ`+#Othm@8&V6TubwVlR(-gr46S)EU&n0g+IETlAdNVrIu2J2_6Mo0I&7#sBw zb(n0z&b;M-rlc<{iC->~)xUvX;C;=7#WfAULDl?55q8|=#hlIMhxK9&SJ_Zo$BGlB z6iibeuJ>Mc6$|eC8>Jf$7p*i|CUYSZoQ4#D%Fl;Vl0X^m|M^CW+)Myak^FCNQd0Jb zOn}Ug-&$DMs_=+7B7w^P{^ph!;`z3C+ncVTvbs((P^9riup(FbIo)D2TOBp7E8dOH zl&#%LL#@pET!7aMiLwy36`*q22-iw?y=%wlN5wOl9Nu*OcgbhhTAW3?3abx*(1x@i z_H&k98*=H&2&KwdIpfpp_j!&~(1mLZXP#+SvG!U+iq!1`#y}OWd@4iNZ=zX?mBO;i z;g0wC@Q_xNotjDA&%@{*b#`>@7(1MP_Aw}J)IQsT;sQgYu4Xsi#L$FQD5z*!8>?J= z_f}IPZd5kXnW+V6Hs{U-a*obDpH00~A1e`7V#~_YZXPf2qI1Lg@YbhyGak|)G!%5t zIO$1PYMD3{ta_?4VfaL)EY#yo1o$JR8~j`?rEK%cK73CNpXP__?&chWtz+qRc{{=a zEhiLp%;{@oWa{N5gw{?a)JVcz_xlna;ALDo&?6?;rg}!5`P$U+>~^;jDhVNf0N}c! zk)F%Go{Asq1Jd^tCt+iF9f=QlYEtHPJOC(MZ6`M+k6MHG-VU$3SOknC3|W(`eWby0 zKj+^}WvM^13$XDdUkNoJnw0u186RUt!X-peX;M2jj0QNR#&GqxCjxgjm+UsQFvMznI=Mz~MCge0XXwSiFIl_h8=sZVUuc8{)4Sx1~uz>UXfxZpTW ztDCoZvmYIR3uTv^KeW#&j^_ctvX*ABuE{L9W&k#i&)}0)4Q^RlCx?ReZlz25i!2-7 z$IEB1aP`}>oP9%ntQF>%+EN8r3gDZXB!Vqh{>~opX(}UT4D@F7eHi`@(W&lhur)Kl zjVR75Tkn?N#=!G!ksh?3UR0r2E5aoTs;owda`^y2y9no7F=!3+w6Q6Mn!Tzj1M0Ar z+L*@3V$`p3JX7*&0#zRBQe%5l>$!?`)LgifX&YE^^d4Sx`q+N64&x++_y~HT0sGUi zw<~v^t}pfh$>*PN)<~q6_1HS<{+v!4%_++$%D%~)v|tTjP`KT-+S6RpzP09-pOSA$J;5i=p!A(ZP z>qE)g7912B%DPcDXN4!T@;b)jHeofDSwj4)`84$p01OHE+c*F6Z{i zE2y5>M;_o8v#+^sEA_!j2id|!7^giTWtl>*+ozLSZKCAE`#gI5hiAZxCj16YDWR_k z@KnWu*UCr+;rF2{LyugOx~K^1rWQ9A5o@Vue{`}4F<}%+)4wB=UkW<0gt2C&(EVFk zFsNOU@YA^NdP$#_P2K9e)xJ(JNCFK zRYBs)CSJ=KAzNoT`Q`z@d=6$i0PZtS?m*`D#uV@xJ2svrp;Gd(W1bUkiav$f7cGIC zJ-qWg}SV|6(gQ2=oiAHFI6WFNh z16fM)`xLSPI~$#MjQc9lJ+@eoqNP}E&Rh7Sts~!xIC)^;yuas@So4%|-XE)^t9eT9 zN%~L)4+v8e*VJ{gm_?Z{D-WOe(smSG5#0He>}LH%I*J*zzDT62CJmUq9c@M1E!6)F zl#fCe`2#M7hc$el`g7Xr@mgwdEMIp0(@kx=*@aFRC1+t0e-v(c*3MC>%9! zm5!E_8;?KU(~cko2raZl&Fk*+23m~g0|ZSn+Tf{{;;Nxu9(P&GjWzXSBUMjQ4&rCW z#bddBrBsN0FWt7&Y(K+ce`<1sfU#z0mc&QFOM7V+b$p4%_9genQqSY*rZNUJ?Lvv& zQ&U#5k2V5p-D~P8VpyBT_GE~-UdBvs##uXULsebVR!T~XTxTBux)lv0YqjI7F$>)} zR%RaJ1Z-QA`V_$ z06~;{j(J^e*Hi;L@f*s)k8eNC^*FW;#!E;&4@2x7bX%S{MDvtynh@5oO9v9^I;kz{ zRb7m3PxU*b=l4nXNVPE@l@Nbx!Arx%2cE0J(%$h(dU+Q6y9?0&4tv4f=njaV^HPGS2ws;0AV zt9FQnyR#7qB^K^6+p`B>^m-FtEm3^xKqPT@KHqWO3`>K)@&ti^Mg2vtg>(BZAIt{P zf$F9;pE$3Qk7K4;{liWK+PVPtx_qHLN>Cl%#pkIzN>FKMu#~|Qrv8*qAq20WTU;Gt zM?~jEsUkMCtE+OeuA%ykW8CH0&q9votBjf#?Z-qa=c<8=w+Kh4PzCEfjr7S;`Z>?B zUabL=`8mEf@Ix=Ufgjj|P8VHwraE~w&$!wqc{Cy>}-AJ(E`2mZZw{)X|{1YgcKYvZO}RpWC`Rr@A1ML()b>#E^KQO_x9Oa;s**a+MuF(Ia`hX65Cr`!qc`cKFU-rcRc0GB2Q*gV(*+0PmWjx&PkX@W63u!o1$onY@{hc0Oqt=n ziL46z98DiU`6bouD!ALLD1AzlIJf**w4>|MHs1H=1hpSNQVcv3zp~lx`7W@B4YmeS z!9HILohzub{O9M38lZ?O6H0D~;+$&+xV8G-5)+3V+-kYFfp;})d1nVYaYVo?;vrbl z6qh_-^Px$)4E3p0bEd8JZkMm zPf^lo8oedW&==O4LS;Q0AdjQvAE|jtSIH|qj|w~67@4cwX6JQK5F6`(aexz8 zlARv!k$hovTaWkCMb?ofEjJHy&`P*B3!|^jyUK=8Cw%jI8gse_z(j-t+y=+eR`5}O zw||E>?c@{f6-^@Iy;t#`5$dUK0$cXbllN6%)%Hm{)LK8-3Tg-qfw&SYm%Guseu$&8 zUH*;)cgasGQfcTTN(^)l$O~b21{)3<*y-2EThDd&Bgo}|@Gqrp7fyB6Zu#qLlG=6S zBb|Z3B8##f8;5li`8q$J9s42S<+MX;E5MoV&zed`AyDu`D4Ih40z;GSoJN{(Y8|9> zoR6sH(uxRzlB?{1@QDN-Iu>4fXw3uMakyB*YvQzb|zlB@&CMghpk;7ZMVxuPG z8$;As5kJdIw=S`zPJz|IlcOd30KgRc>bQ18EKJw2$UVN_LcFs`Ib|Nlwm$lda3Vb@y*NIad ztSlX|^X(0}UrC5AL-H)dI%c= z#5YU&PK424dZ432%L8H>L2|x!CDlk{3uX=IcIf#C^|9102-HKV0>y zD7mCd5=DDaw#7`2!~^3r=U+#-u})(ppF@0j4!xfO2WRgyGNHyS1L%_RGK}>xTXc=- zS0N>WSlDAH1;QpSd?DT#jKT3cm#^z1hW2~owHS~foYkL5O*d!Dn^cw~@u7Ad_nh~= zlhz3xQ@i#cVZ3%-X9kC@yj&DG2agW}P7OrCe4kq2wJnCfNA)?SB7L0ZSfxCGDNZxG^M)*G+$>U!pKR;tBfMLxV(ktx2nf|hqJmhn%FXbCz z`^Fhho9eT|zv`RMvr`*cySNW^^i|sdEClJ@{pPG1nDv^$kSFT$c!+}{SYipImBvtD zF*HawR{gje40PQ#VU4Te^zuA5ouJlqYrQ}U!3QX%U>cV_E8sH8)OnPrd1QF zT{)SiQTZi<^Y)um4<+cMqrW(*pi8hbrqC)m~ZCA~;Gy{;6vH4$}PT{Maqa(V@Wz4<^51UckG`u_@GNWN(O{M{xwqN5=Vno9FVb z0`#`9SdH$!?4Ul1WIE<=y%C!Gs@JE@dvJQWnk&#P=&dmaO>2+$epVhhdA)Ig5TeAv zlf3bS5#ddmBJiV9!0fH=nm3EV2Yg6mEoqJe`QE;*)Dj)#dqB6>oo}f&Y1@!4;0P4Z zLKv%hbdUidai&yMF#Eo?a@y0c@sT32gmya+%*BI)h+V)-(DPkk#sbq0b; z^P~}2CXfj@pR#%XSh8}~>>aW`5*UJ5z<)V5T?(38eNvAdBK)$za@Ji}d~(2&+n>u? zu{WvA$_2iM6hKRCet&)jO+NyB2B(cbr;Xu=s_| zY2c+|M3>EitTS3jc8u5O6BZcrqa43|x-ycvLwvm0WygXptB%u-=RjCF?>JUVnGav^ zS3=(J!rnhoKevxeE`VOrSwti7MV^W{N@0G6ibVc0fCG{BzefKBh|?jwQwG2T-l#|d z2>z%26MqN#e4na4`l{=l*2mwDt=7g*cT>VyhH@}eC`q1CB@ZJzO1|m8tC$sLWh*eT z*y$Yl5L_<@LWQeCI8Hv+Zas?H;jkpzWw_O9earcYJ%$Huy*x_%N>QZQ0=_P@et_se zFSM4i^P(X{D!U0+JVd(X1tVb+tl;*Ey^6HyB~k`~3gD#c-5b?h9>zf1dK#+FFc*SG z)lE4lPtTxmAHTB#@vL_cR^0q&8AOEL(0+)f|8gH)byFRlQGM8596S+jrH_6T76&Q~ zRbAVzL7ZzsCnaTUw{9SK1*ZO5_7})rkQ`R+M%RB1&cx-Frg`_6Zfu7Me%Mx&!Bugo zqYT(O+s6>snW`>DZ{`x)@DculcUB#Yc@iJz2Cbr$kjpe!QqA}cVMJFsNGaOK(Y0Xn z^g#JKdkl_DZ;eIJWhTMrs5XzCHHb#%!>BotRTU3rL~f>@X6{}`erz&YomqcTwN1gIY=+_k+)eRBVRn2-n>;eVWEEZYr zf@%r)Tl8cPjKoL?dob;>%+$fTo`@oy1!7IO?PM`+m|JLzxoy&yN66``i$0?9dS7(M z4cvSr)P5|^1L_79-R<>l(~j-FHKdkfCD|3Ueicqt$7{kWm%HkoaNwM*a#A9+n4%&t zOO%iE&8~Qk4h*C3=0K(T+$;W;yaVT3CUDb@-p6GQ{Usa7qQ(y07jQpKn zToELCO9r_3)iw1;2i1S%ac*xHof(?8kBrJ9tZ?SGkx;mEU6VsNPgWhXAWWac_nP_s zluAcnA=xptdPINGinoPh46HC)9S4!~^uad|8A!FGFe=}ii~1FA&(2BaTDb)21Zjp$($PX56NHE}Vj!DDU@^fLX4dtUR^L6B zs+}rvM#&3ks!jYMURW3*jke&xB{(s|=4@bM&{|p6s)z1@6QVCs!5z^d)&(maZ08q) zFn6@b!aw%A;_qwLV%xCLv{ydeL9U~%`K=)Jhjpt&tjp$uDenAMxuLGxk=yepqf1^c zbV%phJrt(~0SNHpWQEoUtfGmD z>ZXHNi|_foEO$s(^5B&iCslRR17K%v3yA*B=!9{5>45M4qXoPzlO~Nsg24T0{JmNB zKZym0CHmN5lN=@8p=1ce9 z$6#cB{K@bbcH*N&5sGl=AkKo4z7ktMI-##i2?0zGh7@-ix|vA8vkw4%GO zzR@kKp3*hq&;tO33{`mmplgoZ`#L@VmcHE+z>OO~)n2wVhn#FGMaX{w>kiOAo53lg z$us=OdAmzQ{qY{DHdO#}bl%WUjx2>IhCtzeSs3~6T@YqMO9TB9fD13N#Enu-qXttY zY(-(rm9iHn`p=lHn4hwaO3e?@i0ysl6N zC?%^!uQ_rvK9jJsfj2u3fD?XK$8A{LnO#l+$Po-7c;nHfh8txE@0p@E*n{#pp2d;}3Q`y~XD;XKQX%b5lrD z1DG33%TWtlSlBDFcd;HNR)TSB7v_o)khYo3-nPbafyHlLf?E2YqSIe#Xiz>1Lvse$ zYvBOhbA#;C6>@m$sq`HeNC|q%QP$r($hNH#I%kbLDS(t4Ysa!%s#{Q2zdK`+7p@}U z@rzcEtB8}dkdBoUKVzeB){swX{L1z&Tv38Z%tZuBSp9bv&8Hp!noBRW4bdI|(PD}Z z0Q=T6I=#|+a7$oW6AT>CdZ*m4g0`f5`2Ja3aaWt_4}hrZ!#fwy9m<4UZ6T;v_wv56 z_5Pz(&M(-6@bI1r7L?uV2SBNI>lv;goZ$mt(G1Ax!Cy{s|g(sP7B(w*jS~-NpyN zmQ>-d1M;`oK)C$-gb^oG|9knO8hx~Jc(%i%eEQgJLm+$PA!OXdC7vKeYmTxJz~|B? z;^?dITI~Gnc}Rhm2kz7S19uyO>7^gx`j7Yk?=2^$wSV*YzzzMHq6vTN|3?f#4wtE1 zewgBth&c;gRkqv0BQK-SB0iiB3JoKxNSZ9T2J*i~2by@en|*rSY98tnl)+FYhE+8J z2^YE{zEQr)6uH;>tah1dW?yM{nGbIj(ld7Zl5qV2Fl|=VP|@)26i?_gyn948j?C(4 zq3kL-_o=D8(6+U)LyVy`4yPX$>TV2jJ^|eDUr*w>QzAVOCkYqVy0#NdNaz2|Z$tt$ zgGp|-Jwo+&G4Bxpf|brT4*k-h&YfSQBI6m>lWs+;0}5O#^a}33rFWXA)SdeDQop!U zBIJ!-A&juA*%|dJ+>U@0JEY%Sr;RpkLpC)PQwfcl8rViSYIZKFj9f-&($S<3>0E0JM=1R6CUsK(2N?Mq$ZJu|rv} zS!WOm(rP4DwGf3M%OyEJ{UI5 zh`lbL;5wrj6Ba3{DI%DxEekc!{qgCLwK669%voe4h#7>HK4ot<@k{DcSJlf7uTNA4 z%|w6p4)o#V0bNlD3!K3W93#fH**4LY%PHx}Sh@8Ja)hDXqLj0P4s6~x@ehD_lkurn z>wcW#=q5gjaALZE6{iKD(M0KLb5(p-N1M@}#}`23v7ckd4W!nVhf->;tA>bea^w5) zJLU|m^J$+GNB_RF5^2>M;N6Q+F)V%EodM+UZ{ajd7scF~aH9_&0J|j>2ZQ^!3h9a_ zmUt++_;Sc_A%N%S+;Fun&Y0oZwOV*6E5Y(YRPFU(p)O)oJ)n!rMJ7SHMhug?=%i{J zU;XNKqVqU$Dobs^O2Qtxr4vNPEc6E+S9Dc$bejM_%zKoDsG8%Kbu`#}umwmSW?i9QlR`jYw9D&!3juzYxD%wdLI8ROBsV?Rf z(pZb<+K{H>ho=K*n$9OiRA=gLNy{|VCVz>6SgNB$EBndSc-tnUHPAN>^X#jdjg=hi z+kIsS-&b?fME5Ix-ZSaLBP>$bU${NM&(Y_5;FUYDVXGsBVu+4_i5;@(r^AxJZZb#dd zIX&=L%Wuf=$CSyzcP@}S2#;l$of|BsjKcm5ygAc;6&*NtHXu{;q2KaCqxO^Ds@-QC z>z-UeWn{iWCCmpv7&rgkx@;=R0ew_=+n%O4lFT_zTCNq5<`O0wYhEgMDidR%jJ8j^ zgtiG)9&IDK2v?8d!i5k&>!sHrGe`Y3BV;+rQ%j zu3XMLd38eSzZo{R+_V*5=0Y9@sdYI*IJvi226iqCj5437s?5(%RL`+e zu}QFo&R`7HWa#sr>KyclpH|ab#{L_eyl4f{ZWJA#um&tTk_?AxrZ6M zwcw5=9{}gW!4f+?*#$IO2s!~T72Tm$CUli6DU4d5C4rbHX<8Q@`}xKvW|>N*aKdP_ z2;YUu%^{lW0~?&HPjuy-@p@+#5=bP3sO3gV#p=qEv#5xw>OWIk$T{ zo%3WNR7jKQYZ0YFNLgj(Xdn#c5Am&K|iwQp`i zNk~vgZw@&9BfheS3%=^J0AT$5Yx=f)*0%lJ`*&i(#X{w~_tXKFQ(3YD-vmR33No9k z1@4UQrLnQR*g$96C4Tk4^r}dfScQk4TP%gy_VAzO!62kG2CiNwT_ATjl8ibD#fjB_ za9$(*HFR3|D_G7A*ZQ9%bN^Ltd*mU(KZ0(J=06goUzdnd(9z{>sJr={JIwW&9&&^R zGv8u^dTVn*Q^iKtDCwBY|GJ^{k8)^kW449CG;@9*{W6?5R`Gz5LB$AU!T!lRy7gc; zvj+g&p>|?I$VNjQFByiE+nhfy7xLIG3?@ad5JVXLgh?g)KVed#*Xj6u+@=z-V1?6B z;obTODCxPQoy^yMyJBc_X99-x+5EaX2~9PdSGrC9R!^YY|y0k)NbIh9D|{el>mn5UM39X{s3@ z*vA}lfgj@WnT^{+KFLFy6s^a4=a`w|O{rX9zdWZ^L-_coMm97XRTpy$;Tz}eecdWm zr>FH#^)&7^Z@)NXMvsEAvM7(jP_raLBalQ>tBfpp-C?#=1ss&=K5@9^O|U^`!AS0y zrlvfvHLjl)^Duy%fpu5P zDw2d{gCH%K*wE3DCZBdk#FAd}Yxf@_hyOF~|FWU_X~{O(^F~BT(q}yYw6$t>!T}fi z3rwUZiDt_KVZLV%0Q|qPFrM$Sl0BOj-IlpBHq>ct^O)j_za9#mmwn$oq%f!DUp>ZXTd@$QF7#la%ECYEa(Byz2y&O?zg${(JiU z<2@l5@toXl&`6E5Tbq4F0K%uziL5V52XHWf%b@(Ko+qATUPjw9SRC?(8IlBAI&c5Z z<>_}#QRI&jV=LXKRdbps82PIL?63rtO*OqJSnTq!L`;r95jU_wC?+?%by(ng0Laay zsgK*-YMAEr#xI#P>X4!=Y_N49Y$iL4#Q|(b)OloictW^GA-*{-9l~q-;a=|r#9n9v zA)!Sm&Yve}{bbK3q!w(VD(r5ukm)I48#)?T5(I1NbNfx$z=YA&jd5?^+L8!ATXGTL zr<+W5Hn3A`X;q*SqOYy^+^)OE$<>QY^Ck>y9By<(XO<|RCxr@q=~?$s)NyP|F)D($ zah1hiZ@2o_>Q>;!0s1SgRPm6(azBqs9Kh`WKtYm@t1#XSPFhuoAe2_nPD@j5o+1R` zOyK+Z>5=dP<`q>Q`mRWSR?ewBu?!xHMhZs<0psE{K+9A zPfA)!&luzSA=7zzmYRtgb%CQa{?Mc%&hsZFDw6ITAKwy^6wzD6`GqhQ{ySsfSDcie z{jt#&Uncdtlvtd+B*0Pkvy>PK)P<{YP5%zUe>Z(m65fwsTvYF~Y);DK|Hlb4tc;wn z)P>od!8lzdbVI3={yiJ*JCVWa#ym%k{ZfweRQ!|@Bn|uZvCyZ?WChyN@5Rc_A{9W; z_sGH18uqWXwzrNEbwfF#KKyqk`pqrnzScI*By_;M!0n76r~INRlI`U4{I#;$abcDM z%S6Sm9yU)kV)^U|EWdDwxt-zDxr}#T-}(w z7swGi0wsp7a?f4}&@mRAb%~Ci*kLYN4eMK8R)V|6^ym8QP@_bbMGXaUq}_M((=m+D zwDm3Tsr<;ZXM~=A-@Y6Dp(sYDr~dD3QgnmWw4JWBbQ2G56yxtlbyz2n!fV~cIk1u( zCtAVoh>q@y{ssE;N#SB@@tpK!mEUNCaplT{5*Wiw_zyz>`TiY_uEwZrPttA#7ac#@ zQr05(-OaJ9MRX(%-1jvkNlq(?b z@=`LjP8n>$RYxomA}s1ay-&8zGuf5|C~FZ2;h8k&B)j4yv&v&t17=J{9pIdW0rDg_ z^dLL{m(&Y>{(W`pDxk>Aw#r>%&V=!t5WTc&JeQ&ZV-62E-!3Q3S;UzM*28lJ<;CW* zw~wAZVo>csNXCy1DO|*>*Ryom^)l0TD$ohlLs73NqKMNP0-VnJ@N$O+v)ziB+CTjf zGe=i{&L9P3`IeyEvqCANb0onlg?$-rm(4k+Q;f?ty&6Vanhp=pe(5Y)1T>{m;krYg~E9>Vjgh(L0Zv&{G@SoF` zLT@0V3CrXjYR&`+T3W21_A}p*TkNc*`twt*-C4j_WESky##&x$3@kK%uC06(5Hm}G z9CeK`$>{58N*#U%Opy>mwGMBdn{869Gn+5WP8D11ffcbuevv+Z9mmwdJ836Guq;;s z(;hR(3N+7uDR2A=tA9;~ENv`HPEYaJ(YH%ssc=?Gp0f?!A{|q%f8tNDaPTGv#+H66 zQ9s!dEJZfg-5~j+E3mvbn`^rU1s9w}qV!Wwj@c0-zjQS|_0buuiitjo#x)hxOcV6(;gs(c*x z?A_amGU8KP$1u%JALmx2>4lxywU(!KQ`Hw$&y5VoWu53kVsJn*TbXE*Fvd9Bnbtb>_ zQBL3p*OF#x zN!Mp%iNN*%aLWpNYFC2HBOEQwY#mKFH4+G0g8n!17= z!`yrTiojSXB@7A-!p8jD-LyL-!LuP|F}{J5sUO{ZtI4oh79@h^m3`qO2Z^wE*Z=;J zHh*V9T@H>Oz@_v;s!hAxmoYhL4KCSz>g`HnXe2S0rQfwiRn>eXsXoE9_&4hK=hGYY z-rP6d-7a@5&9&u5=sMc2741YE3P6&{$NJ^2N7%_(H@&j-ScOXOSvoJC_h>q^-Nnxx z+MMm(e3=hk7dW76z!n&?mFZTD;*7nfjGBejXDi?Rfp3Q3`CRVjxb6r21aoWA-%YK3l9 z*%{|@ie1Od%t9-L5Aj?)v9F$}M42MM4pEBpX`ySj!`4=e?u0GYBvCjL{&ldOIn1*q z%mg|8Ec>9b!wJkj*1^J%O&ezgxqfX0F-;B`hT!Z7?H81A^tfk8`Qbo=G=*5502|2| zPFf^nd-3;*K*osN_`&N1Jq;kN`sz+BD1^qGY1DnKJkIgTj9Q5DxRz(ga&|Gt%iivB z)0m2x^<8ul7XI$8u6mc6f5_shgUl{diJ5_tX7f1v>e%8eNpZyFsV~FO83ozYNHz7o z!4vfMMqP0iro9*6?$x)A->Q0s4sDv7+kJWZWX~uwge5)wFLdcwi1mm6!>3I=0M_0g zZH{g}03<`USbppOn{WIRo-0A`Ny9m~<; zR`ge9lO1h`C!;W>wsJXW@h=;~`%bjqh%P>F!@@ik%-*-+yG;(*1?Q{81kH zyD|@<_asB@r}-ZOJ>QvGgx<1{(nEN5Q_toa{p}g4?Y+0|t2ZF@$2_87g{>CGYxhb4 zkD)kRc@~AoA9g@;m(2jAqscEQ2;?%~kncte;eR(hY)LWyS-jEik8Eprg+fC==v_6_3h798J_3GT5K5Y=JqK=FVmbm2Gr?`ek zPvAm`!HTgzsqvCl)783mzS+{{#wPE#$LzFW6PN{vDAf{L+2y^zMiAyVE$6SA2#>Wp&Bmm?GZ!`8Y40}QuZHm?NL zdP~{FDgUgzyYM8+z-Rpv326aPc15J5nxUlV(6Y3l*OqvGJu~D|Zz`iHu^T7^bTvRu z=J@dgWgB`QC&e=Xb|Q6x%Xxi*H&VQNUtC17DHdF^7!u^8c$l)Vc&K1!ck-fd!}uBT zQ>#SAnP8t8Yd)Q;mnOEHk=^yJdyNH7yEd%c(t#%eFMtE-?8t|zKr|Or&fsp2o|6w)}xUHgd9> z3H)^Bv0_nSt+#w>`Z%5o>6Kn-u=Z?rtLSo39u_509YC&_cN_+XLX4^cK;UkCZ} z*ac2qj~}}VqVu1_+3KnxCUts(`%~p2_y{3BB51!khyG6szdEV(FfH3e17@6)1)ag` zI{%h|mbrf;J^G_!Ydo;rVT21qq!KLwrKwesSD3I|G1UpW#wpGCH`SiM$3CWBbHvk` zfec?D>X-G({Cx;lPmWRCYdZvbUG411 zly`-31e(*H!WL@Ekpq`rJtQYZzj@BIe10>f*{b~KJb{D+(Lyzv5SgCLfSw&pfU(sd zou+m|F!j!^Aw5?AActIfl^W~d>soL9%{SM(>9a=-j0`HDR6DU7gl?3P$$=09tBVVk zMjOs|6)||*kA4WZH(nfWhuxUfm`xng9E;?BWhj=I-8RVjEF*tgh}zqY%t#3gYK-Y( zH#l>3rW8}*h>0&}@q4n!PnSFIO*pkge+U`ElBG9iA-I35(jMdAI_bl{o7lK*ub0zA zZPD29`6Ys6jh{!12b_itPh&67%RC_O(aWbFhX~39-Z>jJ2>|?9#5Bi=9soCdI2JCn@n!RHf9#d#_~_&3XRFmsxdN2(6VX%p;yvS{!ls^A!J>95Pj6l=J9 zK^2aTc8Q)30CxVRFq0+HqABLcdc(Ns~R=%D@p29Y^ zpz7PSio2vf+h#u7Zr;`8jZ&DqK=kqH;p7WTkuVk1QC?>)MHGGn*a0$_1K3(#3T(!8 zwM3Wk6<_kgcvn)>{kJ5orUX`;V?3IT)H(VH_-X;|^ajP>i^8Wx;7TI?Dpd&6bs5r+ zvNh=yShkSCQUrv&Uq@-KWcG*Hp{!8DQ^w>IC37L0^KyBcyq$L%j zFO9m{r|8rV{$cyQqUopl^De2(P$fty&L?43d+@d{c; zemfWvOm_BD)>_YcD!rXu$JN6v#B-b$Z(_xe8k^70BrK3Z79M0R)>w<)*(T||K%9_{==t<$w|@nakRT`>GF<&1cC zJ-0wp5)Dige47{|k652nbS_oZHg%|HFD?idmOXnZ;52LFq<~@<)>+U=RzlUZZTUp! z>lWWFVdeR!2VNV)u;sew@FUzuBl#9Ud0hPE8olsW8QFiHrmm;5KTTT{2ahg9C)lw$ z^tz`{RynhP^Zu~3*t>^<1hqs0pNlfQX(?Qqa%kvBkmUDsX{E_J$Sa)Lo5t>$_x!Wy z&E=gDQM%Q~O3MT&edo4$rPaL)h=jQSn9ch|zPG26FAUu}Ol3+BelqK!a7$v!PzCjj zxNz*VY^v3k5Axb)Oe#z8)Kp_!wi3QokZyCrz$z)5)jHdV{n$6jCkTc%5Z>mM1v1(5S)T^Mp$g{07_Re-%J41mqlV{(_QQk)NMbn*V zr(DZf2P4ID7V#XaO3&4xj5jk+tvIRqm<36N$42uW44Wnnv=N+2ZDFKqU1W0-YZ{zeMx09x}Nbnj-iMi}qU>kj@Ga9`V&6vId=})UnD5s^z zF5h@olsAQxIBl(RvYl->Lm1bXGyo=J3X(G5k;vY8 zkCadZUtzf15dWtI`S->7zj~kUSyR#hy;0rqIupX(SPJlt1MrC!0E^cI2D#U6yYt@u z?!vwS+((ZUz4Xq7Y%U1E?^3PM9%cHCEVK^SO5W_D#J6uxM4E2J|@lZK{o(BJhIUKV?I6WMY~7f|;_j*PwbY=hOys(6(8A{rBJbWN z7s3ZbDb}U_9xeZG%4AQa@*P0!9N4XwcZQ~v)mA;9AtK*yPB3Y?ug-_`a@JgUsQqGR`ZkUjWqia?zcAj0iP3Mc?sduj{|jbZWTrLL z#wH|eZCNcQ8`77;MJBRSvmSrWXNH|w(u2`=a5!imS9)`_+O~6E4#}Vhr(803PE*YkwOtJR1}np%8_+RzR!xFA1>);YAG}f zg7CaloZ$%E?(N7`@ZQ2Y9&iBHYf|sc6!FcJEfPbG#<#m9%$cq5I-krQG82BK*iXhn zT#J(bm-K>p*3I0ThI`P+MAdUQTqY`{MFIJV;u5GPDY3M%-_st4pl7FXmvq<-=PyR@ zmsVxH%X`d|f_X`P1;U3pD__m@_2|UEH5pno=gpog0Frh z`$ribU11Q7tVhPJ6b8!3P(Z$w3`G22tI$7^NI!D!?yQ-AHkX>|)pdIv`l8JY?`F`0 z`SK@}w_87v2VM|G{h`nJ>xCh7Rol0x6)lzWLIycAj~{UL!{^B2CE`6W?0s;2XG8bh zY?<3koww_z)Tu@lQ88dS_!*vOpri7b>8amq4xu(KR-`InYtQB|_JCU(DjlQGdQWtV zr#J|$gorD3f=l?KYAhy_PVBPYoX~mucHUEY;k&_o_}&AkJgoeeA@dAX#*c?7`qNw= zDkH8yJxCp%9`nqD`*|r(a16qg$5^<%<#;>?el;vsZw-M~b6C#PMeo1e88GeWn&Wvw z&nd$mDn_|Tc04#5bqOi1tc2<9n&r0NT9WI*zxepB{iKrtWF0%|!;fWP{G@&rh#ox~j4*xsf?NAW-F zx4>{`i9eq!WV%F|aOeB{w=V%U%RY506NZDmTKRxWz{O^O-IHgkE$MhZ3Y|~eN%vLB z?!fZXV7n@Duf7Z(qZ_^{a3hpZ2^quj%*)X(WU)qIzBTw#%lL7SJ>*tP@>bcM z_quX~h1JLE5uRb2!45~eH!kG{FD;?c8LEc(rB=db`Yh_V&;;M9TR2?;6|vLrHT~au zxc}`rwCm+u@#gpCo&oY#gKJksMH-~ds1;EQ7>f0IxuR?F8G;Xp0K_^UI73GdwSMaR zgWOhDeX7~el&Wb5ho)skXCi%Foj=Q#!~ZyPAU8gMKwG)A1%6oNBH0dq)B zTAx*>XXK!gC9Na*V>`)H6c15Nz)js8dxM5o45OBOAL4t&;F9*%FdxsKkax7}-OH~b zW>fm^60URuyh4~Gk^7wu-M90zs)@1^R*Rj(wTT~nMe_t072lAC{Kak;?1n9+BvDuYHy8q-bt4=y@wY{wR65A49QjBpFA${Y;z) zkDCqBBKpFMk=5Z6xYc2cP)ePy*pUmgH0ch70oNjR-JDbGQ9eoQ{AM@Lt5LQjA|u7v zC&jnXomHw!tKK5vfKutl0G6#)%35l3zAnI&!4``0(}+!ZHBU}2Wg+@U#Suq$g8Nw>3BPMERT3W7@8LiaXP7G8Z~#(u;wm}%NXy0SfCR86!4XxSN(n~ zO1SIo*;g=kldT7vO+@AsIzu<3B{54MsHrpRfU+9(?I}n_**ew%Q`k$}Pn+j0Icb}! zMBQ2H)x8YOMPlS~Oy3pmbUqeFJ%UoTLB;P)59#@q9n(KT_wc zx3AC@JAB*{i3q@H=;I3LzWMH;F^@P<$^3+qfBdA5-tmGL$Z%T#wxG&t=_50COqh(# zRXl6_$q>K$ejGZfSX*@G!&u}{V9~_VNcE|VLqIdT>F1Z!-XkLr_ID`Sj8(7oM{Dc| z2wIJVg@t~WN$tasK|y*@Vu{#C zBqAh@HwVb(p-ayx%L#7IEzKzma^mK_P-^vpo{9h+F z|EIwL*LN@2o=ER;0|9M zxKFXgD^)mUee=eAx_PV;;t=Ac2UC0Q{BlyZ_Evr#zfM&)GF;ke-8ckY+?M({wo|Ea ztFHU_(UY=YzyZO&KT94zAQ%5m?htm&Z+`nCRuPp%;tEu#6u)b}3+{2}4^62xpP8Kz zNFbiPXgy5EPNT_>J1sj6Srs`!>|d@SucOhez_;OxXZN0X-Q8M)y__~rA+il%cq7Xn|%)W zI>K(*_16DXHki=fI`y@nb-I-R%$%QWYRkue@|-mNxOnIGLcyy% z!Q-#2NM_$^xg4#H2NUn&rl;E|ytw1_Bl)Vs?ws$fuEJ4J^Bb`nweWW$px4I z4czUpg&aPq`Ry30vhpc}x>ZYuuEJ5ZDMxft-OH`E`}ARW?|P-M*}PI_ zH;CMdFe9f*%Pgm-)yc!;P5e8<)d*V#{Y8rX&o0$W*@o=0sG0=3si!rOR>RgqTHLvL zFx@6)D((uKIuqZOT8GBqA3n#*>5*r5Aq7>yy5>G z`?FH}WBhjIiAOo1A~UTSY!c#$69Qflxq1G8DKHmar5~ypb}g{3YpAfA>VrqR)twzW z?7y4p7_In*fW3>4Lho)G5@>Y~9o2B-o4YnUtzXEFI}F(G zZ=S~)SNbEiICd!bIM-p=Lqk&U_7f|XKj{u^2aBy&0V+d?kVdsEB{QXL=4(N`*I1x@ zM(}otUt<~ny7dK_@xB&ZFXEf@LpPr@ zUlf*)C=xsWH5y#AbMV>AK=2<6oUeip*;XTgD;=`9D-XOU%PY_`L9&E@Fn(}-)ZkyS zFHv6&tj0@+t-kV<9!^|-A9vv!uq8G|ci7EK=67vKKBPSBy;Zpg6s9!D{on9iTDo7CE3M#0 z;S=auhJG^;q7d5~iAj~9EG_NN&Fgx;z5F_<5<6Xk+cW|_bfn;EC09INTSTDsrXrxm z-WCuXD9%zy42@gMY*|A(h^ z@P452*)p!t{DNf2*S;gSx_{8|oNZns^8Q}kF1moA$>4=90fS6VHMH44$9*{7I}PrHIf!CRwozVQWkfO=>BBq`clQJ=c& zO!iay`WHjRdlO$dHC3vL>>ua#3p?GtJ~;57z5MWe79bDS5cdj$H!kc=Q^LfdW4`GTr6B+YHj`rDUj+O@G| z&-+kxge_vFfhkZ_W2c0?q<5AMhtn@90&G*W!=0=|ZrQ4?RAo3=zf5{PY^FFfaF>JY z=79qUsh#p98Iy=Ea<8?C0lPVRU~CMYM_j&m-;qOc`#!5<%Y5RUyzJl467T3j-FY`P zz|XX6T55)TA6R~5U~91Fr7vAKAjdrdE5^NaeF)quQ0q*_6-a#%q36!W{To*H-SOXH zF^bsW5XrrsApV@}sXGi0NlID7_;9h7&J`0JvA1s#C!wzmb`rN|G-jo0&|ZP;jggXu zT!D1gu-Z6t4;J=G)hOCY9iLgFp-`AhBZ7wF^D@^>n-2HKlkL{0Njb!NBPbp2V^Y{4V0WO$8y1YE%-rj8Uyy9{+BNgP?cbE!ivwvZ$Ofc0 zX?5pFQ0}gM>dOG4e7F;kGm?owdqmf?1dP{knbBU=?`gT>U*??@-HLLar>^f>GcgRX z?sTs!?l3+oE(>>@qO=U^YWB# zjviT+23W0n9mVA3sc8(=8ZPssC~gVVAjpPC<$#627)X6`BV#6jQfEnvWBQatrz01@ zUCe;V+FzW0{K6+#nW&O6D#y+lGd3U+RUlPAZ3GVEz49& zZO;;#uYwhipBR*$%4ke5QYyR46o@;2!4PQAmmTHhNrgpqsTy;;;D=^S<k9uWH7QpzHtn9hzsV(dueDhfx}02%+$7(M zSJ6Tln(J^ge$GP5HAg~X0NiBHJ?iTsx?p;b&3XEfF#*(MRV29Mfqay86}^ZU@0@k- zD<}HIJ9}Ldx|JreY&j|tR%S~-XH5*=^H`n5Ba4}1LNhpJu*_Ifv=Hwve$|m6TTW{| zLxraq>cCjG$*JH2!fo>d5v2p z$LK=oYQ(N0&)KG4#i+!`M`=5+#)1yv$f~0@A^vuWB$fNTniNGt^fU8IcT?DkXeeEK zT1k-MyS*cH=z@FtKDju0SJ(0%DL8 z@k9S3h(BOU!7t+A>dLgAE!H6YD*aqw(qs9;X9D7C;AYSFl zmNHdH>#<}$SC*B(qfn z%SfsQE3f#y^BZ1#Q=uUxM3xLpXk?@38DeetWve^`JV-2ljyM}W|BvW2{wJUp^?d)5 z82h}SjCXmfXX^P1C}l$kk1tJ1L1<=HfTy3U--izh9o!{TGfzhF#R{zG7fiZo)+~qM ze0-92>~^7J6QZVO?{xyTu{U4akUn6*msVE7eC9GxcNzji8_u=(Kt()~v)>7!8b`ag zaSEKxDv94t>9-@lv^Ser#QxgZ(nkRIw7kkqM>!fq;{juCtPlQo*{w!R>Es zgag1v+!G|On@Bopy#kQ|kwg{_%;Ua-?C&B<^koH!1pbqc5bq1b;M6jg>*JpLPa^mL z8$_Gezr?lw$(RO!%zr*F_8QSugy_4l4^1){FLLW26WoLzFlCWX5QIO#O9A&gWZF%m)&pK;N`vlB3pJ2tH-gfAy&!mkrLC=zqHye!ZQZf`$^-1W zzAMmEOe?p$Fzs`Z^4OWOoEamfk)o+M;jq>y4^fkLcm50Hh#glikj2v!nBL&Rkt&O8 zFpv2oPQ^!}QP3JGK29BATUbLShnE|HgHqE3!5FEb$5wSr#@M{PP)^qSN;~8 z^bNyPC(^3t@2HX=^pO?JnWDCDiEy9fYTb3=EFUQ$UZ2}HE7LJ{yD_cO$_N|NxyiU& zerJnwbLir1&*_ihQiJ`n`KcuED$aH@JZ01@ z()UaB)K9k)FBQi&A~J+KuzJrYyR1zau;N|~r4DURDEJ$w5TvH%6ldW{Vc(n7P@icu z#w{X#O)m~|>3w&cOH)FscOhC;R>I^Q5JIUzj`qr7DFm{gv4%}aMiX^&P;;8UC$mi* zOpi2l4kTUfCFS+@!9b^-?ak87QLUjwyDdX(JesPq(m07)uT=}b6>`E=@JNF7!pDX3 z@?do0xR2YxkE7VmUI&+8g(poUQVy{ZRy@cvP1f33%sRgU1<4E?^GH4(RW7a#6&0AE zv_>~!fBC`7q#8jnF382TbBfK<}{I+1m#^?l*&B+v{wj+F)^+~L&=ze5JU zq}?L{FM<72AS;dn{){pCy!p>z0ImpjN`hO5oW+8UHmKpVV89wg?s@s-!4Y{VwyYp2?tM|~o7N5U3t`Vwv3|Z6Y9zkk z0>+ATOfQMI$9GOE3D|xR&O?J@&X}+#ZDBc(soKfBSDK(5(MhDS0X3z_Ntn@GX+z03 zkwHo6SN9wjn(xImv|RtRP*x)rW$-GnL!5DLD#BA`B=(hRwCCs19jf?^eB3g03o30L zS~VmQ3lxH8kbUnT+_H;Vb7Gimh$Bt=HbCYgsOo{^b0%QwR+CiZ`NMPn< ztf-1@rIuKcAh#1G^QSQ&N_t0m~`_E9B(^wrDX5?6&6< zr?~S)ARdVN)UBqs!p~SC4Q+t&*ircvsP1vKnLp?RSiMLF6hDd4;M1jkU^I}TarSUv z5pfy2UrB?!#z3K5JhJ6BuBYByp3zgzh35u5)4T5A$rWg_5h%U4{(K_zoBs2^y8Uag zMMg751pC-1KiuopDuHOMR2y-DzSvEp((t=>qAbsn1PPS%^BrPI&71Bii$voyXE3Ww z;%Gk?Xafy}rGjNKOadtUSt@<(_q7 zWI{XMh~srljzXVx{WpnEly~!2j8BJ!zOgAmcTOZ#ajt@P8l`Y0^Z0=4i$ka#_HV2aMjVa%D(38zOJT zt`N-TjfNth4}5qXnfO(L#1Q9Z5FW0L0>O>f~6P5ND8_M!l{*>_kUOd zw|t70DfC!o_*`8U?8v_)O^z`uC?)woP%2E6Y}+@dd}*7Y{Wx>Es5DlHvULCDVk-9y z(M5UL=l(|Bpca32TRUO;I-9Vz^Vg>Dj1C=pP0=ObhXf2~Lq~F!0n2@wNk zNZ3N)&Onu(><_BARs`JMF1vJUMyzES(V4@J1K|B~ztj~dv6o6$AeA&YFsL$4*F)Ej zpt}O)4!8k|^sUI%U!bMm{d|-@T>{_dkQ!-3MeAj^FI=Ib(Wkf|t+-?`IC1V*k5@AkYO&&9>H`x8ti2$f!VWzyFyXb2xI$?_D3vzO0Og8>)G(=cVWGp8}j=rpvK zs**(r_a3hBbR%;+WjJTH)?UDd#aMUR#704Dy125bfi*1!$p;k9R_bt`Q6*uJ5N5Sq z)DzRYO_n0}P?C{v6kGy$%WsPS{-KxBXu#!jVA!K$6W#{&Q%IWqnoRrH2vNHH({J*x zTOh0JNQ*Oeqt+RKDsKa8dD89hS})mvr%NMdz~@SDe(c|_Lk554=f6rb4lSg6%!u{2 zaqhGx=Qx%X5P+-W&LCcvqb(xwHSzeY7fhoCI0?n7g`6gCS}b#0LX$Q0WuNdnOJ%YJ z0`OD~ml6uMTeZ2-f(yN|6m&EE?OpN5wN?*tJe}IA(!y=4aHPb_Ln07+`Jd>njYJ>W zv2y17SCSPhfNxAG8zi?|yb}#adfdpXec7JmhCQ!oy0@T(X_)vdh|z{my79a?T$8H8 zFl2m5;q5CE>qA^*vUuwry0J@z`SMHo6!!SjQM*@^DmvvQp{c1^ZTyR4-?|vfM0}VE zUtS@bIgT+tcl^lURGhLpZL6Oc=0jxIN0P8HI1O@h>@A4TsXZ`K9;_j0V~exc>^y76 z+kd#Y>uj^(>X0wAujOrSt3|3|bO2+~!ZYmtJn2h`pQNYw&`SBPRny_YVvUXHTq7n| zG3TXLHnsPJm0<} z<09$L=cT3?@!u+lZak6J(e~<7<$3V5_CUi*j5=J(^};3}eK7Kv53If~l0O%b|AGyb zvU-TB?rF6_noC$FS3_``(_(VI$$%TlQ2_k)EN16uEKUV`ZxLvXaIDNZAfK97P3=+aB&cf#b)!`d~9wlxh{7eUmDs96Lq z3t@+4XR?SZ5V{w^SR~`()MFzP=b5MElx`q zr_DAA20tW?fACIzdmQO6dcysGEsiOB42YdwWWg6zhM{P(tC?FrPl#{5e_aGh4tGYY zB8`1f`)&{2Jx)xs+$5l(L2)!0B;8anL*cLB2U zC$tk<(8Psp+2s)&6Kn=_9f?VwasfKscza~_aPWxL_17mA047ZNWonRK>AwDxP4lz$ zbE4)HclO$8pgZ9ccrI8|Mu|jCJvPF?EpVKL*h|Lp_}fMC|F}Kh6ZRPT{td%FIJ3|ocg1e-5x|RH zjN+&%LnVZ7HF&oj3Jlw6LsW-{SBDsP)5|OS^OlOr+~eN;RP{V^MWmf1aL&ofI34kI z;0jbCP|V+ql2k@`FiO~CZ*wtooYs+-u6!d++k^9wJMT%)W@qu9?edVMxhax^s@R7ai{4 zJlb0WHp|r)s$**C@vGhU7`M(>@vMGqO#h^X{HxHvLtT@=8A)E<6cB&~@wV|6u0VkM z%#u~*705~e4w6A(xK?SReEGgsQHOkA$NZr{{JpsR2gTz*d`wZ=FFjZQ(@1x(b|daAGOXOq}wyigipBqSqtr1ZZx=0$1Q zhiz6;FHh{~%*(wJJy)Z(_B^pPI3o#K%$`lQ?8)&Cyc+Aha3WT#aCX1XK6|TcOT~}! zl=Z{9cFnG_#-YTXrY|mAbqQcrs$}__W9hAFFygr#`1tUfoy%BGs}1A~a||-t_08Jh z?eB^3TfVxH06qsQW(RpgpoxS^pL&68zCjk>ZhtnkdCl_mgpV2_I3jSo3r2X*wbG3p zC7mHF1L#9F1hB+L|M0gL{OyI!CLyi4T08zxDZc|9zx+W!I{9}%qXzt)^x?+>^M^Wr zH1Wqa0OVwgD~AJWUl&9d7B#oT9Im#1aFF(;~M0~;DcLXcjU=9le04-C%|i*?QZlo>=d%SYrVQwciBzQ(gG)&r@ttT{QFwv7t9|yksZAq6 zZ!j0srrSw!J8lSv-#K2%bkLWrh?NK+O!uFE=tZa(B-kQ4&AG$vhTDs+3DJi=lBm{g z($(gZa)2BAC(@>U&|(h~KZnq1>OSEZs=*JSe=L!4Pd#Hi!sY!U=zS=k)1myBgZHdV z;)%IG(v!$b0hKAM<`C;f$=Az-x!3XSPi$V6!lHSOw7>`7U zsWanD)_j?!xBfgL6u62Pm)mX*Aq~!WmpLg+i4%7Jl|`_$?78AA>s|&#g)Pr>`_fk3 zB>mp#h+w46uPDda(7-ITQbo)4b>ms*-Q_P_#oZSWJQa0m3_bMB)P&Mz)`B-A7bGLw zj%`eMQNFv`!cGA!cttWu(oI&DgRw_%21>rZrB#$$T{mxbf2X?~6950cG4 z^jm@TxPT`wA>~P)((8SPXRZMvzSHVu#9uo?GV%{{=@tz8f&hoA?(_RkXm-puxS@%W zC~@{rg)Zra+L2855R1CnTL$gsmV9|vObR2mD zlDtsaw)iqkj_tKD0a_{uZSll*AJTOCa5CrEl@GNzz1Fv*znV)Ym7F;vhM(Us-WukF z7SROxlL_>ybvZAZ10j_M;o@!IWH7}9-(6!!_SZ_8{U3r}=5LkcFUJf~M2Wx*?t~3w zQyvaDSpN2*{cCQOKSCP&9p8sNnj6d4K6v8NzjXka(7MGDrc+^wU&7dbP9LiIZr}S< zskpRXal%z<@QmzNyH=%Yw4)$ zj`#A*7J~QXI*Ez34eq=-_2(aCU6}Wg4Q80?Y75&Y z140=AeHmQu59}<91~*+@y);7V6WKbd@5qxSJoscWQ1>Ah;$Yn(Ed^S0%^BVZ_R9#O{%P=N`OYn9vBhNqzXEbdNB%Fp zz9oyjcc-9TFeXYozz+XaD&(WJsY>Q$)9@FUw+gG7mz_eSHh*K{vDMWOKSm!dT zb%(Q2GbFn2IW&W8|JZ<)Z13pQ45m~D@T>Qu;`9Hkt0?++^HCtT{9Bco0+igfs^uJh zw#H~i{#*VRdh=h7LKwg{|$nT z?L!+;d@%B00o_3s7lbtoDywLfmr^4ZE@AI9z2koDOFjX?isZwijh8F@sgUB}?E!BH z12447&Xm}4Jw)7N^i@HjowM=FH+FM{M;Z%Wua`a^-TFL$O~G|=Sy0Ab zo-sK-sruq=nFnVop8_t($mt2Kh15D)TRK))hFb4xqmEg4@N;_Ydq+n(#j;XQ0~Jzo zS-(l?WPXcLe>}=Rj-_*mRY_A`lnw8)xb=8INt?fj&D5z?G=u|3PmW(XB1&M6Z}av= zGlyzqeL)|&A<3RTMMinA(2mha#QdG)=xwJ|__viJ1j|B~PDxP5gXW&1rG}i-h;c|v z5|7i1?t8{=Oxgnx?>$KhMcAc?B0tZJ2D~*@>{YK{cH8FrHI_J$nhLnNiD6a9CV@P@X;M)(vneLDhtu2Y?=3KyuBU>{^pf( zBRI}zgqsSQ8mdWb8{~V8?x{kC;VrTkmDsTI`WErBOgWhOSaogo5Bpg`rp?46SnHa= zqFDzl0dl&Qf3*llkt`1f{~Wz9*lfo1m|Z4ph_|cY3HYb8OS*w==wfk+_XOS~E)9h+ zsXM`q>bi!y^{n}tr&c+xmYL!L<9;{t5dySe7^V}T$e?Z>W4pg_rZ)mDi3`DYWM60- zTWTgg%UUr@;vQnwvN5e$hITah+Hn|yP=aieSR~GT6)4ge% z5MiwBMqFE5Ho|+)IxWavdYFkZRYT-dCd{j5BHkQL>|s$4e`+1u2Ky{J6WincTf6>j zCDlVE0YZAZC7zux@7@(bv5u6V%&1ZFnUD6%5B7VrfPH^b!(f7Szx(D>Rg>`N1P!-M z_FG8-b5sLliA8#)A_tZU9s;J%44HydYIu6lovU{22;>c6ol9m2)&wjMVBcn&V%@$l z6q>SSq0ZjZwx>C611Gr|XO-{G22d6b%YpH$IquKEQ#zyzbM8F4nMCz|gWI}(agcm? zZ`Ok?4niQ{Uo8!No@pdueDac_)UzX)<}hMMO6;;%^C{BIcNORNZR6kX%V^9@DytQr z%G1jRZ$YA=R`^6GSqp>{gD<@1JyWMRP{3OA2-{BI=E2mR8r>5KyH}sX%AgL3oiA_m zo}Fr>JRjU}RwUsT$zO0(Tvd#O<{p0V5JRftsUg>e;r z`OhkT@A)>8@FU{_a=^S+z2g6X8RU}!@W2ws8^AD<`I}*6jsK6h&aZW~ zYX6c7{0Gz^pX3^&=+1cUOdwVLE{^z@lzDQxSV!J3NHrkbh8$nE#x`i5@jgeWQp1m) zV0KX;RGC&4vtBz3!<%WXTu}`&F{h9bs&3w)GxhKV2wI7EE~4?mJ97x~4&Lm)<3QYtwb6*GN)$o|Q@te9j3kh`V#_n%?+QUCev3!7BcQHDtih?dRE7h`xg>kn0Iw z-Tvt^d-r6)DA;}`e-;+i)36S$29}WP2p*styh>C}^zvO-w{$$Z%UC=uK5_+;_U@`S zuq4_UD zSnWmcQodAK;(zb7l9aJIqUu|YzEaQdrK?~4VNjnA7x~F)p7R#_1Y^y7fNr#Hv8B35 zcm)B=U2M%5YBsEzxCDsW1yhgw3*88PJ^Gg@v$#lQC;><9D}Htwq8Pn}bTQ=b{6)5q z72%fL%gx2JSPXI~B91y`FBPPz>NQcy>+^VMJrR_IAz~Qy)H3m?!^W(?EiWTXRz9Zwz}%GY#jYt<+1K+AP6B>r zGsmQ+iqj19AA@74rsiyKZh`lRj@~a!V_bnc4?=QH5qM1wYOl(MOcc+T?QyzN1m*$e z4&}w5^kt>!`Dl!go;d$uNDlzx$GfJNu>1J4J|C0`*}n-9ci4X%C`uFv_RB?1UiU$0 zC?U=T7A5(~26`_6kaTey!B%$xITg5ffe?Mk1}8rng4Pj&GBC}>DT2g%0mGPl*eJaV z*JtpVb|Yc}yr0HTGzeLZ_e9(_SZRgOyPYD09M4D%hK~jZWr_t(#vv{7Y&^&V=KzB_ zsc;1%T9o!WIvnf2}XexHGmave$9J^YS!gAlbi9Fm@WI#jXMK%eT6!J$SZe4*uO#H0| zfS){(>SC`uL!CFm66g|eAq}V@71=!*W$gg&n7T<&cLtb-L@*s-8QSrn(E(WT&yVI@ zr>!gmrUWh(zmw_2*B4>MIMbk2ey6c;(g2*!o5oc(px$W4Z>Iw+o~iVEyz` z5%==%0*W4&i?$=I-y%DXR4K@?KAYS}cSzkurnV6pr)jKvwJOov#I$KPJB zsHO7?#6WHW8RdtFbHK(S%O5adtq4|E2qHlmc$Zfo4-&+y%>%ml5nc^b-Ia-Uw^2K( zzO0U{(^g=FzWdm@#{5FKRIQ|WkxLA3g*lSn)ALLTP~IC)yb$k)$#w7CIZJm{S~h*zYzyF95X3MZFYK7lPc&?YfY4O}TRlU)YyZ$~SMy`V< zfaMh?5)l}P%P2*bIoBzw`ukVXj=_>7>uQ{pS7j#M1LH%U1&@$f znGV*{g&(GrrzU^}SL(*$R$^^tt55@n_mE;+9zGoMyOI~1KFA54WeY;^C-j;R2TjMt z)FkXDx7`HXtvhD)4m2lgLaT8y%bwJA@nY6i4-+%t7%H7UnweQ6l%B>&w|8ZsjrLci z+N&9^DWcwSDbf9jU#Q$d)F>w}PRBesK#QsS;)R=Hm+K6mz!#sWrA?!HXNqIqvom4* zc}C|$tX?jHrhhw+(!;o>u%%hWrIeFm4{ETa@>cBbGRbZE4NNQi2NS^`E7Rd6%K=T) zhOz7#O{*}mE!UjP!5z2p3^nzw@s*bRvO$V;L+^Y!Ym2PZ^$dyADmV%u`&K90pU;+R zq$mDixbz2r?Vqy>PeNE_;b-?CGukvrM?fPtl5x?HDVA#(3MhGoOz*?jhajpzDnnKD z#PHWrEcr^Y(?)!zkS*{0SnlVr1Co(4}X%{{`edeVMFC#T3CgvgUiB zejr0F&Qy;Tb-jQrhHfg)rMTBs=kIW{dX$RcuMk0OYYjy>)!xtFHscVwP&KJ{% zoW2nl$yamGX$)^989BF2E8zAPMRw43yh*kI&G0!gK!(P{SO5M*muqn3JoyvYvV^msxbE3cll^A2J;vhWI2g%h$apDUZ(&=PPY zLv!-+-3b_25eRZQf<*6gYQa5|&1scyrC*$RShz7#nlD_Z(# zAbO#B`w4`4p$8p-)B8XSk<-Xnx7$n&H2TpO)~#~DM0P*9boRZ=8KD0=?|Lvax~Z^%m`iAF0i&O5>(8*( zMxxT-5y%Ah(&+Y6*Yxq8F4Nu_>O9e9^|~?C&1c|-8u{;-E78%MBM9(WK7{K5RsfHZ zPMqxR(urK2!{QNaX8zK|z7PW~m7L4wH#*p<;Fb?6_3+-IVSidO+NullU8>Pc#NfjR zU$D>go;1wdy8_WiA7s&F*=F-CAr3UeYIC*ynFrV|bMzZ$&)|C#;#-1sjUbG}v;LN_Q*}x|~F(03uE}M`R6AGZ z&23T2Dipdc&6I4-jVL0!*-}j*B=bs2i?m^1J^ux~SVEkaWA8Kw?ur)R#;YoozSlPn zNy)CyQl3`ltEt(pE_y5>N|N1G&dCMZ?|#8nLqFphYV=^%pl6n;o3J-ybRy}Omp5OB zIB6wAX}bpyM|DD~3XlF&=*%oIkJDdQ!asXLz@f8aLTsGeJw)nr+LwVXn{rh;V&ypzE^W_Uu=NbI zUy#tk!CJJm=0ca-wo(dfZr%wnH^y!#xkm2&K~DYMnb>?H+9zk2h=DGVM#70`leL!BsUsSyslXbb{O;k)?1)WMV_ zU*S3MSj`T~8@|bIo@kX2w(eXE!Qzk<6CL3I52R}mz+5YhLJ?fxfui#og%ho*wi5+` zNO;RKUTE!;J~e`2sN-=cT9}J)3Wua>>n|1PP^AL?IK-kR{({Ix0dJ%@9nn!54DTdan3r>g%kXLvO#Asi6tfWQ_c>HDg}9 zOL&5&8`u3ejPUzoU_I&w@wd;(*DgdL+~NLDD5JjOk?aXuqc^h9wJc)fvfRm{`II*N zk)V_xe~{i>bA=Phv_1tx>;Oi};Ll>l51Z(;p6=X*sP@LSAP7BoHx+&0t+SU!s`B+|P&79u#&lZgG)aX)`u?4G z!HUY(g!oXnljkv;VsB%tf>&F4Sur}zT62Kux$eGlk&EPBt_B+lkbeu10qc(cr@XI> zifh}_EdmK7SRl9)G-yI_D-|@y9uV23(jFP?Cwf9b+1d?;-s%au;;60AROWAs(Wl>3Y>*i{pzEsTsib#)yKy=%P`S~5!`P~ZO{uhU-!@Pgucgw z!WUIU-}uKx8{F>@m2XWH=K#b9Bnl}ZchIl{%+!c0?(2b{Cn7l=2N!WQ=E6VEszjFO z8o4pMGFh5OzyI7^*V1^+1E4>56x)Wk4Nl2EnK>lAHj$?j${-C|X83fc@yeI%AhRs) zhjF7_^5|`-wbr~%`T+$&BXw(@+xK(_l)Hawlzrv)u6g4) zk>{zJnT^bDH*#JznkQ4h=DQbrl4XSDP?+~z-wW`r-NWHd%qP@$(GYcj)y3-?(c0*I z<_X_$Y{-RIT7lCJ$tNAogwN5`c1mZdyT3qlu`p?Uh6C&{vS;{K!#!7H;A4SN?Y#dL zmk*1qKu=%wdTh1UbagJMN1?{^3a5KPj8 z$hl|X$OVj~UzlnZ`uZB6vjJ?;vw8A|p zBn~Rx*g}|c*w3SU^=&G|KDT8pCMv;h)2>2D9mRtv>BsME61AFAWh83nTD7&orfwFa zn{d=X3fK!=9GA0FKb{=9YZJd4jvq@vPH#j<7Gs{Ot46+!whpnb=Ozk!GT2~=yv`r`h~xa4PGCp_^@-|gUR&w(E3o>K?V7Q#`# zWt3;vI=Rm~$A-Y?mIBP-s8l$2#Mu&V@O{y42})vW+Ljx z@;YrX@p;2#h{<4O5QMSyi0w;5<|JP<9-nILz4hX5>1kBOjDVdCan$_Dwp(GeVi&hk z?4TnFnAqCbZj>yk{bvd}rK|CpWZ{>ze#4siyEBHSc09?kyxM;(bcp=oOea{zNH0=5Ulia~Q%;g3J7Il2{3u zVozIi{e?oE`oP4E^4H1K(2Ex3w0^N9DL6}6BW}j*&0w1^h9AC)!1CCGcMf~w7t(6I>w&jNe-c-aMVo)uw^`)cc1*B`rh5dNQ=WWAH zAxGlZ_*YE>9^1PSd&%+$QzujX^{4&Vv3-g^>llu^5!P^wO&Jq!L@Q>+!2p~m4J$t02FmnuxP)eM^`*TOCIco2Bal2z<; zRl@K;s&Z=TAR zFx-CjIaY!8Flrr*Y4&!G?7MMAg5$aTZA+-HGJ5o2FMX3Yk-Ol$UST>j1+O)tny{`aPN*>v-O;DJ}zbVV}ZV%&{`-~l*{y5d9Yn+ou$!%SU$#MYe1flb?v}hSwp1Ko+6^qleEx&+6Nh$F$iLc@670jUYy8He- z9)IOxQ!~95033T!DmZ_ai{~MafN~AP@?J?r3!-ND{1m)Tzo~n&qi{sHEdcwl0AAD# z{RPrrqAvIas)gMo(jSxxVEy(*e@%GE_LJRL+H0Ol~`W!Jc&R_U}1TKbHkE(Qcg)vn}=j z=6(q%hl{>egzuflPc-87bYFCvCCky9Wa+f5&>JMbIHR+_Nkqwj8_T9zfQa58`xUrT z_!kI}Pf_ue8%B$Aj&9^_JC`w?VNcQ@R@NV>KV3?P9d-dnJBPQ}(WB>i^roy6D1aa7 z;ySaLCqlBeb$upa|C|wUY4t73gWyGm^F;c&gOgJ0%m?6JAc*m%Zs``R?4ZoHwg15{ z5K8{Hf2t;^K`x^$ck`~pFB&ybXW`|D*4p7>U{z#)8b#B_fFV>QmQ^r>pa9c!LVErw&4yAX^;_mDs zU50+Gf~v#vqrRCMQkjZt9B%#d(Yv~tnq4_ipS zAf=jEJwVGT1`tD?#GM^~!sb2@uG}^Upo*c4-Pg)^VH4}X;&bOmX=+{6u|M6CnD>i# z`+RA3WxIhlj2qpwrP?YJTqn_zoZFOdY7Yq$1G~qGeoVTRN35>B&qyIIF+xIq4287^ zP=)N-)!N>ie}jonoV!DK=5N%}Z0WCJ+0imNV@ zl?Uypl({%}3u~7Iub&7!C-B|arnz44g%+u%Xf<7MQ{A7yZO3p@|B$OY&b#ckNl4ed zLwQ4y#pcgOTFm zV$UcDGq{)1F{30Z!x*9qWcf$wlURg>_mp-x6JmTh;S}~lRC;+HC5=5@ihF!tu_P3G zs9h2+bk*-mmsg~HG;>JR3j1QC?RFX}sbcwdT$i?5=TptB;;f-$nn#`Ka~+1Y%A%xM zzv1UCUy>%fl*73**W^HD?5~Ye*LSbYT`vbKQrvP)j8$JlXBmXecB9-ECJT~ks^(;i zOl2&_sR$)KE@BU+y6w8S zH&*YBQ}6mg7-D>T+`U)F7!vy9Hx#oETw zI7=aNdz3M$f#H$1IVNrwr(=*g}j`%++PZnmUhOdJiWx1 zzWxA*GeQVe#<>cNiw8a~1}l(_G#j%ag-@?Ji|apS!sv$3D#3HAk2}{PVqOo%BoX6W znt6@698GV?EkvyDh+N;6Y}m4z$r#}>bSpA_T>s@o6l4vjmx8yOCc%&j8FsBeko-}ZvAR*nDVq0v%-CZ*!H=jY zLUk2nzNRLPan#JXUZr`<3dgQ;3RB?t@h0S?pIg~zEKMV*T)afC?8O8#)rRje80s5k9Ok+GMl(szmmd={JTS`!N#tP;}G+9I5z z#5~+p5{j&`2C;~JEJAoVvg#J;*s=kB&SuhS~teoXJC;Hiq z*bR`~`yLH){--Old~AckD<cbv>j<-yF){GW}%GmEpn{=VZaK$JGl z1-uQcFGO3*Hj|Bl+fIGn*J`(|CcbylJAJWltdYec$7=uFq*`TFoRT08AZ(0~IMy0; z$n`Ze`1n}DRO~3r?8I1Qtb9+LdWj;RKyP%*oAILO8#9b|vVjl}>aeVqBq6Ku#onp2 zs_pYG=vjs*YW!*<4UB)2pfphyFwuA=z3PAUq}pe>(7V%fnCP~Mk*WUmgF;=a{GBVB zq)OWd&TYwRYTq2CV&!6%!RvBculv!Ewxx$pjxaO7P;MHkm9Z{+RXeY`+dk^j zYy8-(i2XC1p3y9Gwj8trgCtoHfym2ThWRMc5~5Z_#w zH4V1J)66F^ANDxSz<162PBk4wN4`{5c|davZV~GjsyRRuzBM!-nY0O<8B@o5b$sXK zE7{7aUuvLxNH?LR3_081ge7$i)p(DsU0{IJ=kBh7MMN3!_3crg zq2K3qZ<`%#nuSakWI6PRKa)mU5?;F5keNdAKvd3!VpaUd8RN;TU8c=WUpY5d=!0*E z+qlJo8FxHuH)^k=I)W);QYa)Ek$Pk`_VZcRJ_1a*M>NM-Q-X6$ROkf3>US?%#|t!FAa^Zkpr%9F0m7T}qs^Q;06k{8xO4I5 z!4N9?V>d(4W?YO!sm$agE~njYd5)rGlqJ!fTM=X18k)*U`|97Unwz{Kj=w-{;{9!B zmrfZ=yJpnB#*M~LA!Jlb5}5|i`uX2vC5xw@4Zeb2F5xzU2~3Qi$A7&nV_}xRZQ%Aj zVmnQ7Y|gPab@D08j+NNkZa#L3_G{tdX_+y)(aDLj+{Z@3&<8Z~-*V_P9$7JLeN2Ll zei9vV(?-{dv{Ea(c@;X`Ep+tpBkFY8_Qu?`b&}lkDd|#JHm1W(5^16?5ZB~aI`z;f zbA>LXUoC^9MSp?5Qr#)?apvW(XQ>yjPLG~9p@UKVh+toM_wXGllN2dXc8i&Pt182w z;dwf&^62UN<(}NIf4EX|OZnG?wtvbwhWD&B{dw&KR4`ULzHolYOorm{jxqf_lbU6R z6?;crTUkzUdLp_pO3^AjfyOShAyMP08pGpu5ZuS;5sqq}NZW8ZhRiX@|B+ob&HxdA zuq+ihEk*J0y7vEzl>&dr@sz&(L#SmOWBRYCshonF&4A#VQX%E5(VN@$8vh|Y-7V`N z71;kwnZLY}I&^{8Kp<(=QMMG_EiU>9LZZznNKb955}kkJ!0V1tQgiK~wrqSOj^@PO z+B|#OiTqM7V!YBypU?j}k7a!xKQwoxY+&1@EPoUSY>DBI>EBzNTExMH>lL5j&)+G#;U6h|Os)ZYqaahPD!Oq9KBMa!m83JSo56Q0(# z0#Vt1*n+!92NeDl&4fghc-$U+_E1jq(@fWoHp~j3RKPBq5MWUVg)Nz19GG9Bvz8UE zIe&qOe}RU9EaChyfS)-92wXbBi{|IaXg_iw8B%mU@(XkiNK>{REwrB9fV-M2B0wlW zRvgU&sKMU`7-2Tiz#gy_#e03udjp`hP?e5=UbZ%x^cUzIY`q<}e+i4GKd8J`1`ss3 zzd*3DYsV>oeMaJY%`lPOrPY++yS3QPI^5!J^RAjYXb3uV`4=cC4rwU|a=B`f1|A96 zC%5%%%q|gLlEcw=6ih>b=HEkM-{1m^8BC@;uW;~&8Tg!%){{g7o4&Dw{+O5k8ZRb# z4$%6_f&Kz5*m(TVl%fgPzYqJ~1Uujpt#gD`PIqXWzp7PpJN>a;kQ?|IhCuuRU6L6C z7tG7MG6wjaa$d9^h>Fhrt7U9pfdAojKBn4!>-FFqZ)%Q#@T!} z0XK}8Y9Rdk4@ps)AJhPkkS(5db`3)TJw66U2Oh%af!^tuti0y*$DQ0qqu~d6XwsI@ zZGa#Grf{x-MdbzO@vx11Rl^qBg<*M`NHILF7k14oI(&oc40Q85G^z8mgO;ppWv(Q7 z$Y`bN`EL%!R($5VICK9pWd6}``q+>$A1;8;u{QvHos&r( z{-il-cLrw&Otl3>xBrXSDRtEW_#HQI`JP%jg)B#J-LP~z&2qVC+TTrc`%`Y9@Xg=N z&E4s5##!+dpdLK?n^BMa%{m+Jv17vjX43M1v&|j$Yy$A#ESjS=ob(&)t2k5yu$P|g z%73D1|I+N@4%=-l%wL)^U5Yp^s-7+n%SNSNj)VX8wDfJpaNV-Wp2(Lp z?z%)RST#EQXBvq5f4>(pBejvs@3hfGg<|qgPk(FzzIN&t=mwV?Fwc?66Mz}$rX1J| zey^H8r=_L;%dvGco>4!!CC{2ixi?umi$`h2)1EuN@aJo=KROU3Vnm7#;7a$hp~?|#e-DF=cAG2cHsv$=z{kR|ThPO^4r zuXO!_0mmrY`MGAn|E)p!pKtTkU7l;{R|Z)GWV3)@{2YP!LrXdXE~(zsS`)>S;N_N= zCd`dBJu{P^GWUUozz06rU>Hncq9`R(2!C&l?S2kczt*nF24aT6_QeWM^I1mH;011LJp z#{DD;9ICmHNkOZ+Q?!P$iN`F8J=u8scOZg3c?xhwgvbE1`=huhUyq9T90Ve`?MEp-7o@~~*`5$AJZN_|+=nAnR~6@185!J>`l8>dEc8#^W>6*M#ErwO!@KCC^6Wur#i~r( z6D^l?eat1o`{#UMJcqLY>d~6UjgI83vJr0Obq zvDk`NTS=CL)}7gfe!^tsvyQs&bzk&&tQs0Fu%AcUlc(0%r6K?r~zMMB$fDnHxfVFm<_gb%EKjOu^cv75R(uTNX zg9H!(Mn9dJ#0KWXA3;CAz2~4VZS;Hm_VY7plgs=Bs@1ibp`~ z#^CO{-Eb3(2lLN$M<}@oQb5RCJ|995jB1+R;$gDglqcfRrtL;pn>t zc8-B#GM|I!d#d+>da;JGcv0KXiatXjVUJ}C>>LRN0WTGR9sPw3gjc(|*#D>2LHUVTZ&}$KzL5XGYyK89a)v}&eFocU2~J@LeSOq<3;zn80Mv11WFI#BGtr3hMQDggNkz{B|=YJ0}Y_(ok@OFT9axNkhoi?JWZ&vzKH z%;ir(S}H9C4s8T#yl_VGJT9b%N!!k?4bS~1FQ0ED365>+)e?-zIy?BSaMBM|$Q-CM zagS&DEF|cLXiDPOSEb3+#X1TOxzBQR#v9AoGQO!wVynlpaAE@lUn{@PF{gEq`&{!) z=gX>$uT;sr$=J$KTizekMEV&n(Mn!k@DJmt`oFLWT5Cw!wIKFu3=hBWwcxd!$&vN# zwS3WX0C>$Z3*hd0zUHU8OwH}s+0Yt>XeFk_mHEtx`|TEy1d22H_JL>iW>9Z7!N=vL z#|I*P2{-5p%{5WdfaEyp^ce>^3VxRk(GhBEvjZ;o2kO(#3O*rM?)o-wSN* zGM{;9l{h1DSD@%-V!@6)%-0rFGUp3h7;k9KxE+jKAsRD zk)Wgh#CfTNi-+kmO5)>YGQl)cgWjyY$ZoCrBHLAE$Y$#1Wz?UU$vNRD+sG4c!VXcM zcLjDM4*EX113k)9t@5B$m(mncLZpFk2(6frMv*=1aDXxs z4Lupl$tHLwF1@5^fqcyxI(V%M|?h0t9ITW zf!s!`SZ^{zoZM0u{hSMW^EWdIY7#V>$0_|(9lDlK-xOWVOF(#p1ujz)Ybk)@s;e^f zF0KKeL%Dc^U{TShCVQ0-PHGR!Or6~>2|LVb55oI)Q>q<$moeqm_WZ3|%k}loUS8hv z&eK93`&0L$$0lBtHO-&E?CTBA{Q0~qxs>9Y z7mh_J%-oLiQFMqx*jG)~%s4wAi6$eRTe#)wYWC}2-UU4rj!nF37EDh^MSza+%{vJ# zm$LBeJyCKJEmgJy?MF2V6tDGZXx1vAaQ$)qNMU1@5!;E1_4m(WyOsph?#}GFlFExN zk2Zz1zttxHD$R4#%UT%rdQ}P!blUaA^sHl;!USrpK4w|-C3pxTzuhBL-DS$Hfmgjt z4~=;&s(}euw!E5F%d9`}_ z*CFtrYy01vdV~YVP+Rq?7+l*VA&BAg>L^QtL=5TmHHjz2eCKIOV>>(flL8WwUmZUrPOvme-DR+~ zXLh38<4i!Qhpl}_Q=L~1A_Yu!1})C7{63=1NN#37imqp6S19XHL85P>0p!;y+&;`}_=x#}4zyKBAmgxmvmLMt*eiC#uS@P)BcAh4{5SE0RCW|A~hP zLQHsvp4utXgQ(WWvO2^}%=+V|dIeP37^Rl%T)5$C=S-UaR7SOiT+&$>%hF381M5^E zr(%1KD)`$s`|KZflT)g*tNq=@TYAo?j<%Y;x?ZnW({f%o1jq zK`(XtD~oq$qQskYY8Y)2-V_qsGk%aI)`OxD4{j&nXQM`4crKZtDZv#k(_BxbD5<=n zPjpc%ZzGq$1TdT45(TE8J;{-;%dAZY$^!*sMfN!PuhqwGM%PgIMWJk2-|jbJ`teM> zfKJrbL=sVQ;7DBQnZMb)7OQa7G@U>HEYM17GOa|~fHNfw31Q^<%=%aun@dfRG_mNs zT6q+WJ9DnKI|n9`bCgVXALfo4^J>O-MeX;6`Qz(nts z8Ef)$K!u$MRtYCC{ILnFeISj|xawSKQisd0?tj1(Ynbf31{1;-Z;ucO;MViw8t+Wc z{;Y|tHF?ac1nLXUphp&!4lT3i-l!*Gxi*tDL(GOJR^aV&#tN;2PnDZAja&=AifbH!%Ki zKx##H1G@5MoFh`Cx-L-{oFMEz<^Q zxOLi|D|944<2>eB1+&&F5ZSl!FYAe%4F)eCwL={q0Aln+>jesawW~ zz@A%s*2MR{Kw>L4QQA&uMuCo0xuP)scLJue+EwJ6xC0CZ7Qz=9;ld)qZP;-p0j_(nUIvPC64K3>ZW{H%2+i=_l;W)FM{`}6k^(eJW9|_R?YzcVo^a2i z+TvX@IHo7EV^VC${J962<(XUpNYf9PF=`?fS5Ut8uIESHMk)_|#Y(t#jxP%Ihz2$q zDQ}+Nnr&D@h;y^OJ@J=0)Ld%a-6x$hRJt|L(-WIR$w_Q++XUEKXXBSY?QWQ>VH|fK z#vXRb9t0=S{%Dd*8S|Wf{7hkXxYU*g>1!@1O%`OifUCH(?}3A6G-_zaZy4mwOt|}G zHTs@lR#`T@@4W(5_v#ys4_)7dyGJwRK=PlS#d#|ebNLJ21_kJpkhm1QnYm!uq9KVZ(o;;YhW_Jee^u#p#q0%!Fw8d2e*fHa~c7;Nf98Y4p4Mt$hcJy z6=rM%;in?azvqWYL)pS_Ia0mBdic|Ra!ENQMrqFb9(aLCAcZi(NdPyc+^Aclx~^VF z+4sY;9yBrp@xcwQ8LL1usaEp_ z&Sx5)Cu|_EhgGdrz@q>9a%He^v%4_GmUo|-{&)%iGG-TImkMxg=A7fI#UJ&>JC;ln zt1@{f3vtg>XDVW5N&mRFF}D!;z=cMg@$jt7F?z5|indyRm>Mxb;Y(Rug2)@W`@-73 za7w}c1%$M@N%k&@BB_oqh8|~fLZ{Ou-c*gC+BPXgiCB@~WoZK)av$_G@Qyv6o!aw9;O8DBpza6} zrY8-Yj9d9O0Xn(-Iygi!1h<0PfDXXNiEt&AJ|!8!dT9MP;>n|hmXXGz5`ZBR0D9w8 zaA!Bw#m}u#tOd(Wp?Y!>!*Z}gOnSe2m=mCnzUklcW$Yd0WflV+zg1I#?WI>1PL0)1 z()!5}B;U)Ct8+|Cp$0#+y@&k;Two)&{4({%l)rIu%t)f2qYIZl&FLn4-FY#!PdA7B z7W@W05k1h+szelhdnUZ8f#7={<)=z*Kg_G$SS>sEpLAMK5pMwx8KqHXTwk( zB-g=6XL{N_sqSMt<-KB-JP)z=QOi;K(l4AQ1&O2uW{|8b?U}BPKYVIIlxLkLHU+-! zk24;9o$2@R*f7lfG7WoBQEvUQKVNQ5$@&a0Z}t5`=3}>z7|C2jO}UJOzH)6#xFuVZ zFkNQefw)$*V*g816#*lqFQtal`}*cV00VUC<9jN%q=6Kfyj_2$c@p3f9T8b0j0nG@!2s(;`PK{xw$bBdA9)V;=ePb{Na5d#&?Y4Jcpyh7$4LmFb~*hblysgV2)NA zB9bCue)4Q!q$}KDHKLeO^7xV+;~k#nx{YFE%yK9u+hjV37DNf;ov5s3Ts4(0wZt`h zDWjM5WUsmSz#RT0pb;u8U@*trS`LzWrXBn7e@-fRhgL^T`;WwDtOG{U{q2wcjmG-D z(f$v-(m;}bS~sEmMBltZc?02Itfc6?WsuvLKp3Q$C!siO1Tp6Qpt(lL@sS~wjR=+v zUUc5!zFEjkvXg*YNySOMT~l)Vg{gE;&k0>k(xqHz`7ttfT0Wr<^b+UJlgxA%Wa^4L z^RBI<&}C-F(|ek0YpS!{{D@KIkN$P$nML?Z0uyui4yV6B3b7^ca_W+YRX@pP=D4M2 z=HJ}*6n$CxpwrgYwbiyIp7618(u`-!)3|Pjad)Nr`4MGB!2lkfErI+4^Wm6JbWUwl z&}Ql=i3#&c=DNTu8%HgjRa|37^D0y4Y4dZyQl7!v={Ri}Fn!(gB=ZBoTAbCROwbxf zik0CHNvKlWarQyJD}|bI21QmG#s^#j??Q1m9^>mcg~4iuz}@cQ}*G&*?fQ#gDF+GzVoX|-D^GlY0n#L zaq>At3Kn-j1!|FR66cWDrhOe&%X(=_hIRDUN!%X$u(tRJ_KJ|0ZX=vrZHF*S_#YaDR2WT3-zD9OtYM{-2c%;BL! z6L>E5mxS6Kh1s*g>08a)@cW%-qt z<~`j99jOl6qg+NY1@^8;Y%-)=Ye0+9ir}7*EgO){2v0TQE)Eu|yH+ZzDhg#BfZpX^e9DQDWdmUdeR zb-qTQhM1(3;{^ucdrL?Cas? zRD0OzUZ(r?X0q=nZnSuA!|Sw#rUjuVWRz6?8f&!c&BHqk^tqwfI`@traqy#sDUjRxCH^ex`+cshugN{Bjv vr>kEh0N7QD{U?%ne@*WFxA~mcO!B|vHx%u11Id7O=WlDz|6l(IfBpO)RXo-2 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-79-NEO CAP vs Income.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-79-NEO CAP vs Income.jpg new file mode 100644 index 0000000000000000000000000000000000000000..21eb14b43b21a3adecaa0db0a4eaac255bb27218 GIT binary patch literal 76940 zcmeFZ1yEaW*e@8Q1b6o$g<{1uSSeNt6iR`jZGj>oxCE!PXmJX(xOfP(`7;M_d` zw+Mh5fRKOyNN|r32m}%l5fYQql97^-kTOxzP|&h5bFi~9v$AsXi12Z8330Qs^2^nvk1!%)dIm-&X5I&U{0{{rrKDwK z<>b|#sB36yX+M2o^wQYG^p%;7t=&6&2S+DQFK-`TsGt9bkKqxKQPDBUDW6l*(!YGo zC@3r{E-5W5uc)hUXl!b3X>IH49~c}O9{K)bdS-TRe&N?59En>0y|KBqy|ars{&RAA zc8>jf@eeK>0N#Itb@%+=fc*ztRCl;=@$vESf&bvb!S%fpJSzNq_r(aPRSke|+-bPP z-xJb4PRg(CCE}JaMANfg4x#2>074zU8?sC>^#vz_|8EwgO#!~GHBn;M2LnEIh^!i*#*9gb9eRbGOQc;LL z4&{p!^>h^E((Y{qvmNq}1K?Vl_I@a=j)h{e}je zR@9~&@in!*lO+8e5SY^)5K16Vt)B;MFtB3sg+UQW<*0^Q6r1OzZ~yu6UwxtI$S6$8 z1ZSvcY@&*y;L_#0P-4M;$Z0q8)k};9j3UyzH-B5+Imd8}d1FJ4lQmvjmX8TQ>Mb<~ zl_CagDQ*Tae5)>{mmN%}y>C zcceE%2=SwOWXClD_dG&~PrC16!$B|Ct|eYQma%wDnioR)A(G>KQWB?AJ@9C4QHdLh zIrVb}Ih}!R+&A!3ah9{z^UkWbkeacl%i!k=AUTvq3`9pWeN;vafFCOH5C|YAMyoIp zgeoe<^vwspDS(C;&oqqb?c}x1THBS0iJ3Tu53PyHahD(I&0d)oIx?T-7Fb!-?aIHu z1!#tI*$a=0Ycb;0@J(JkCs?N2x~LgSz~5o6$_RwV){WBar>=!(Q^pet$=@)+4s?c< z*rppuggv$ltoIQMrfWHp_T)c_@Bq|GJ9{&%lR$NK4`XW@5O3LoO5U!jID*glho$_z zwA3oS!($^oL!JoOJ4X&?5M9-7t>dKi7;U|*RtAdTi4O=TXYqUx;D3)U+6J}EVu+DC zkA%5=JyoDxb*G9BH5kTFeK?O)Gk66>sC3VI2M|7r%q=32S+RBgM2R1!Ly3b2ECzVd znq>83Mtf3raVMQ#aMeQ?C|pLVlr3y2JOrm+rp}6gv*YIckf;t2ru%~md^H!SfJt~+ zH~+{svmbSO5F;LIPPQsYO8)D&zx{MO1v@tVIT$PY88X4qNt^S%|rpYaT{>+i7{kmi9~0FfosYJI?9Hw5#j$vfV2 zLmK-v3LuV?qrEpBsz59_c*7Rx@WHsd?Hh(YuiLSu!_;dwW-j8?=6!Nuyw;voYvoh4 z!@|syLEHEh4P&+_L)AQz&n78Xf|!9Lbk{9lz)E1Y_^Yt1rIoMl;0vm?HyZlFJ1Y&F zSmc_P>;2}N+0=oUDmxut{>%uC_1LW(=z1r>Uzjes2HLKMo{!CFDlgNd%ldMh?XhVQ z8<%IB6%Vb-B@Wa_H^cn32mPABtJ??vBkNnQU~FL}Y7d z+ebG&g&^%se|h7_6R|JC5V@>SpM|~;0|#Z5+gmOo+#10;5IhGaAc!iSA1}Y4u=~6vAq^eGSa_*mv=5(9>}+zjW^VEJX)~g2P48YwSAoYz~7dA zWt}TjUS1Gy=%Vp3d@bABXr_?b5@B9HM0^Oi1q866#Y2eE3L#9`(x-Lk!)}VW@tT10 zrTt8G0mi)LSgIU}ST}}bu?)b4ON`)9uq6^kvN* z4^_;F`NKhpz1rizckOtQXDJ1P7rox+Y@qgwV$)yKpQQ|#pZ|bDSjr%M5>OP>PF!NV zMXr*^9^=1#0JbUH_W(YhO~{a|pSmb8(q$Qt)T^ZB5qD~fCpLd53Ua=lf#zTVDBq7) zQstIA@RFi89gT6e(2jJT2ta5NDb2_l$A$`9(A^c%L;)i^ocWlg5TI`W+Xy56T?(}i zvq6GodAKph6VY9g3$PCDBJ88SI99W+BD~Jg(7+;|_pxEX<+pm|fOXjOTfj$a)NsU+ z(2?XVfTlExt{8W7dXF&QKP633!S?mBtR&99k+D@LOdWX(AX^19yvA>6c6=o<`@ZkP zp_?)@<`(cFmoiYb99t2@J#cRL10MdZtya8@jfh7C__$RWO$<1HDv$#!H{ZTc-z@e* zKn~tG7AgIZu)2>fL@7r}DYFJe4y21qHPzxxysiV#J*{`^2wD z)N%i;RA+!n&sGNDY*%mqcBQ(`5z!?(?d3}Av~D)lXn&mFX4gNoNRoBBR`;7)gMxJ- zsOE0%%ACRwXbb|4!_oqZ7oK$#r6SdyU8@h*p| zO^f$&JVVJod*RMZ%=dXE38>(cIop}iS>7}Ub6c1ziM&RC8ISjHlLKCNLF<^>a1P51 zNYSol$?2)r=psEoUmR1v82NM@_v z>+D3RS5+d!`12>(isGTG=bry!mSqw;%!{CrJ@d>>sq2Dc#A7vA`2qJjkj1a#KKC6s z`%dTyep~3`go67_FgtB4rDp+!kh4%yJwytRJttMN5oSJAK0`1Oojb5f3LQVw5q@y2 z->6Z1^|g@yXJI*8!k>za^T8t^)bmx4z~%F#Y8yzspJ;uR=2XIi*K*31W2y{kzYsG6Oi}h7^8h52o33alj zgzvhCyn9TMEw~X_ARhrt^kelM8S+$NyGOUx<~r1kx9eYEs$Q;5GlJkwgEDcy^$#2_ zU5{D9FeF?_A$1=MvVjdQg^v5|Ro?pCBM1o92sQ-lI^dNQbR18tyS{hXKde@6WT&W) z82k=l~19e)! ze}&a@$!Imo)$pojL- zrwGO&_o>Wgf56`mlyqtZmRzA5SnJq8QdvAJ`Iq&2iNG-L;;M&gpThs3OHSPt#6f{< z`#cDFaGIvl?*W6HH1-|#vkLaSuFfTjgh%ls|6P=kNb8a5W(A!<+8?h|{ zf~6vnidGXn-7Fw$dlb?14@Kj5EzK=o4q_vcvOSgo_a_v^TWX%xVJ+8S@)slsUECGH zOh0H23^T%Og;C%ORon8L6ns*%$zN3m6It(Z#?PBCmvz zh!JnWMOT@6vz8X6FWjtyAkbfnW~$JoHKo`6?rbMKcWU!zf{wjoD_3 z<_r=n%%i}^%G5Y^k<5X#k-3yBYNeLNns={@FV4LCBnq?r9V#igQK#59dK@d>C zE#T{$k3Y}FME2bB?7G|n+$Z$hv*wN{u@rqb0$=Uty`UoY2JfX`aj=Jf4%3{^ErxRR zn?3mIUFrybLVvd4{^bxirR^m*zC=n(EX`^s04n{V5}R&`DUG_aF2-t~BFt|A;)62Z zYi&LeOJDok1MUWaORz-cD~|FJjcYJthNiY=*G6^x$W4ni`J1v3YVfZUONwXc!>qR8 z6nnPfg4#FC_DrCZ2`TFvDodIhYHO@kL@wjp*@2Ae7cmx}wb(Vx@KeCwJ!E<$$fUYC zilx7cPab5jeDD>Q+Lq6UqKNN|8^CW{q z)X+n5$XXA{0K$WNqkmJJd5&DIv`=A9fIs^2>>NSnJ+5W&ut#3e>PyeMq5Mb;_D%Pi z)OV=B<`GIJuCazGYEd^Lttm|)B*G_Zi{r8FEud>Kpa`w*?JC@XsY?72&c^8WLxWH` z+DIQb!cdiT5TmskVQ=BqjO{I9dH_kHj1E1kj-+Lwkx47p!OO5VkUZ=fb z(Bn&$=76s(o^29CtQUzzxIvgP$5pICe_KwpYoX>`TK(thR)e5K@{Wx5w(J8 zp6TxlUXoCg4boCpm#wUKU9p`r&W&vHx=LsBf}gY*Z2AQbjXBgTx^mH_z06jHY}YlB zYvxf1e+J*ghirt#ZaG|> zQcxT;R|d%eCm*ys<5gUVaEzsWasFQI{pwI%qaeYS94n03Yej5siqh(|CPlGD8;UD6 zDrK*&k~@l9c19$Cfe0?jC4rjnTsWb{`i#TQ&w&$lR(ndM=(wu198Pp@REHvd1$t43 z8B~-u?n6gPuC1ue5gx={Xf0BnzjG1R_r|=^VT+1mZ7IJNYfm>ijE?MaU-{IC-st|q z;xq62F!!7UGTbC)Sy53C{v7vxC{R^}{oX$=mMOLwr8xj&z7FwL$gT#8wIm+6A5vg_ zs~`74JVx()LXcREi2dX#e%M2zON>Ohg@3G<(ns>Slc8KBj6-Lx#@k$w5!uxx*ZNia zXQbQ~A6wAF^c%X#MqQ^F)6SVUnFjToy>1>%q8{?>Zb1%seIm+@ud(MS3G({l9hsl4 z5kZEcBU6cAjQB!**3krQw}1#G`j61`0kgRA_WV6Z-;eE0bt*Qm)LckS7>8GS+5TAG z0-`#|kTvum53v6;8N(szs#aS!MVW^{_BpT$x+@wPpwT$xwe5`ZOh4`w+PW^0AGRJfJ-eBqc;9Jdm%uAuiUsS8^#I&y9JEk8QcPDUWJ^{n4T+> zU-9}@S2v49yIvR66Af(t{5V3cMp=aGAu+Pqrd<$^BD$Zdrpfi$PO+DfD^$rHU2~O5 z5q%5T!*pj!HI_Rv2U^!Hc@$@nJX%EN{v?#~i?AEDuXfYAaJQ<3tksa4LH*ZNnx_kN z1vbvdoSB+h*R!^i`xMS9#j!T(nf*5X)2145E%jJ)yQTwtN{OsfyiP#iEdcc(MnUPZk^%Gd=~y#(`79&)J@bsOi^{r2z?aV@d&9C*)ru71u1oJ7;U}N+9cE} z7teu7SB$eAd?i>J%O(sDJdk14DdoUm<$xMwPHRIp>cR1T>2g81SktY$J~x`JpG0%+ zt@LI4)l>vGBJbjZ)n(hwU!AHWsRu=!(y0o|BNB=Kfd!4fZ{2ATMgNIPg!Ko3Ng`HT3w~gnbD2T!SR^7DUzT`03#cSt#bnh{^ z3hw*IN6OWdG+J`&EeNp>8`Vb~&fT0CR%--s*c@4m=c03a;C2ge$d!VVEgNXl&nTqF zBI)8ezA|BrH=7P}Iqs&iyeZ$??N}};OH8#LNBb0DkEzx{p~)~3$PrFT*Kew~fa+S4 zQU@jasX#H>DTgLOd)A-KPc+{a#bbFwgFj!t=r>cUFEh2?l@?pFG*$iz=B6gqP4o~A zGV(sjqsUR>caXUSkeAFINDlDzD{;TPUPbqcw**qC^$u4gD@VYfi&|JmbYDQRaM5u6 zK_GR<$DxolF8%qguz4?Ci2-@=&P8LEuTQ)h`YwG<=#NuspR~cL(R<54jgc4 zxI$9K{YCw4U7q&{DVGYlqMy`u219=WA=c#y`JEhYTDL>uzE;ZmE}ZF@ECR9WrSDnm`f;{DtKGmZfw_FXkv z_nju0bq9;rO;mM?;AN>Gnr939W~0WnVaJQ{fvqW=90uD#gPzjXL?EI!ULyC=2lZN| zoX?H~60js<9K`^?i9ubLv&$YUN_2MQSI=^%v$NwLp&QGOMQ|q0?^M%1^Crl|4+Y~e zygMD2EA|T^dB&JIHal`2b!7n#J(&b7T@W2#xR+dqs%!Q4(kJljHb1vxV9E1?>S2hIIfAy;k$JG;1$Q9@vfx>Lz~o~!(s9% z40uHpLowd*XE1UQZWkGG`s(giLBT9k!)%kEW!;DmC{AC892M?aKki?1B~$e)wHz*1 z-`c5-+FxpKPR>gP;ICIQ&&^4;vw?Lp^=2W-o73*mT*0vtIV)?=fl*hUvu_syW!A)MUNN<{d%cJbov6AQ9zLvDIRJ67*y z?tEPL5>v-|M=HgfoQ;FGBG{;ulz7kiHhF-vdyyEAdzq06osk#7i*D@}K-$<&fr~vK zgaf0h-_C5bv{9={^*d|gWF>q-i&quduXOHy1F5a$l3}X~!Di1GJUIJ)cvJ&RHaOQ2 zk~qIZXrwYJ*s!;NK4yx@>cZ}1`0wwa`GZbM`p3ao>x!TucD6))WfB-%VIR;p|e{{{?s0B0xK9}->i*!-Z_Q00M+K?9mJhiINWjZ%@9XvGZdsV2xpvB0yxf3 z?Vhz2ceJf#FdanyqAN(**h*9-OFFX3M>S0;u?C7HZ}#`Y1Cgkhr9mFSCd1_o*-`S+ zoRYS^BLb+W!6fWN2gbGBFzQRK?wnV^4bae-R_+LqpPZG~in_Ty{K3axg0$|5PaF2dmGYQR)X-afnvWBn$P4L~49c}=F>7*aX^`0a_y^2q zPee2XOi;r27U;#kSC|cN^#2pa{&6F4ztTl?0{o0&kbAiYns?q{{gWi3DyH9rVl96? zW3vMIuPSMf?`n~&9bCQue9__c{v?_;{URNmmZ+!6WmdPA(T2c zEW1sZFQ`_Y!x`#w+-D6tAF{VHf&x(C zRSaFySbMN1^C*5Dmo^J&gdxV3!>Jy_JPn^%Hwm?C3X`?ce#ttN?~B6;U=5%aoO8Vr zE^zXNdmn7(vMw+?k~!Xg_wxy*#=n@fQYUuisZ&eiVWY0vcrb#d4gQXYaolr=Tt5B; zXFZsxV_I?|j$KCSLC*Nf!ImeOcTXTes8!Z*$s=5uI0cop!rqP5_PQatq^kR5q!x5W zORW4U4M2J|DC+MuOt-6aAI%Vc^-|sfEDvIkD1-Bka<5O;C2C#FT`TvSiKG@ZCD@gi ztCz~&KRCm!$!V`YuRqEFU>-%cm40x!q1Z8Qdvf8N9@=~606F_=v%WRLTI1%qx>AI> z_sQbQ7z?(q?Bh+p6vuIAEtE_0O-b~>>g!b2`zI?P_`*_SqW|o>M_eb<*DP-T@V6?; zs)G=#J1f<_AS_i`$+$7=yx>fe?8&%QDtj0r4>(qQp*^#4T_)L*Gmb~|>hI4?f_d-w zXJYJy4p-s4srBYJQ(X@ub=4y@r%elnq~kVaWw{WDrC zt{$PS8q1aPgYpoavvIAN;`%s%qeKd(Gy;cgjdIWOo zUF^HQO58zA)k^$U*@pIic%}pMd?CCsG}|vj%npPPy6%I|+P2O%x^hd+EF7m1W;;`| zNk6NKfQ~@barp~3`7Dlm+g1wb!Rqwaf!xUBrk_+t9UFhn=oR8M1F89O;5yJkluK!~ zH5^;9r8d3qYV~89M7CGNgvnQqXeB%?e!E(kNx>ihxE7NXK}{Hyiv*=fhr3PK{5j`wlnm?HwK30aMB!2*|qm^ zjISbD7c@(k_PR*W?_hqYF~e0udR%SDA(c-@();7eSQ~d)R{NS4rp+tt7~gP^d*$8@ zZEgLW&}m|vrz7a*#V}Wx3;~GhyK?k;s8dNRG9-#!7(LdNR6SuE9aJvV95x$%DO{1Y z55yus@!ejwS(7;EPlIOHhlOPhIS;dccPA4zDrf|kFwxplync$=)i3r;yU2}dd>Cj~ zmK!C$qp3YBV$;<4MPua0$5cj}8LU6YP@4wUrUL6bBGo#Sn(f0x*N_%)gdu->hv!WP+o;Qwdf25;K? z6LwCdnoH@mLFm(3%O0Qcr5mXkBR5sLDym+5ZCbH1qc~cDGByBSUO7LY%%w#137ZO^ z>}E;Dc@k(cTs5U#66cjS!#ljU$KYISHmdzDX~wl@SZ5?ty>qYCJ+XTDh-wu;gEdAi z>al4Xga&~nP{5D30Hh>3Xg)qgh=J7Dl-ISb?%~Kp?eAbooL*>r|1diMjmDTmf2-&J8l7!E0PJ4D|f@ME^av0}TM@s!13-y32~KV9h1fOyfO@z-!%RhHG( ze)j!J^Y9>y*_y-=-U&-R7#uHjgFsHUC;i66uQxzzIKxV)(&;OOieb##yUKR6s3u$X zcnaG$1Sii^CT+%(wLew?w*cZ4>$@VQagVMY*wgP#rG+I6u$8}=a)gqS$GUBm7vM{Q z0qZwSHfACY9^8zdYeJ2W^uO>YJ*mPKmFs2Qx&_QjOhZiu`UlMdo2Pbs&!kbh-!ex! z>g8WkaU~pPEqqzJl9@f~0HOrV^+`;(;c0nid!!G`LHi@gf*fMMQrCfu7d~5yu(zU% zy;K0&IHUIn+)$8p|0=b&t_&)ty#$={T_jw(lsC2G*alLSA3r!@$7z%M>`dy#OH5?c z8*I?}r|na4YFizK24Vh`@g4t%GPvS5aK%h8i%!rTVPGz&lN9U-Th08WP4z+CW6=6n znC5Zy+f}4*Oab;~4K~;=EJ*Kq+rzgZrDSJcAA)9ZWI7x7L34g z5t489UQzJG=7g@VgbHnkFpeEXHc)zjJ9R8)xsr0=KLrw&YYi7|R+)NgG$y@0tGCg+ z1bt_{MI+#Vk&af%5Am9%`A1d;6@ic1XF8enq$o^t`Xn ziIhSFN=-I)lR=e`xK%B8wizm06P2U5AUVPIaB3svlB0hQ zKckIL8?=Ducnb(kO7H+rM#ed7#=b`H^>a0jB_Lqc-`S1Im&pGP=}(&E?sNmC$jtTT zYC!7MPR8;%Fi`Rme9@vNv7aZ;z2R^u$lv|>t};d7n-KILmFAZb9MleD!WCM4i~%N( zBR5Zpje>!9HBHm5N559gOq;R`oJ^WCKGwmqi&T9_S?$Rl%Jo&N;oR($PGCN0G@y5# z7U&tKXlFGteVs-?2uC;VC5aILdo~z2s7xZlo!;*&Z=6n0>soAj9TB9YQ|N5Rnz*~p zsPHv#oW>iXlKdrpZu3 zB?hRMR+yyWnQn2+<-7Y(i@PkPu`+{~4N@Vdn)kQuUEozph-iLG5zBm@4zjQ?}W<@Y^`oejhY=U+ zx__!cN|0x5UJ<7E-`VmP-=&+;+QTge=2m1_zcN#>yzJ~1L#c`+idk3x((Dy!eBlOk z9Z2kf!xADW zL8&{^^1&gQ!gGUjW^KGf&zmFtiUhM?*z~g9{$VS+DQA$H(kE~k&rKjE>3st((FFT) zz`Ss2FUwvHf#q=d-bI#w=h0VBcSpLu>(oOcULCTI6N{{Cfzyfp1J{9OE1=SYZ(#vJ z;wz?7k}^!oLzH3UdOqc{M=H#B(^<8OgLSWqu4{pN;lMdh8#zlyr^Uzg*d)yF^oHhy zJ@=J7Lxh(#DMGR9Q%SEKOQQ7;D~tHAd@0zB`8!Ste;)CIe0X+7yM@Iddv7vRM2JWO z!qRZ)lnQoOaHq_?2!7wNId$ddIkshj(xumath2x75j`1{&_~awmrA;lb-j!I{d7w> zW5bo^Kqz_R)c2~p`uF9+v8;9M3g^NJO9**$qzjMEXq?IuE)~M8HOOOyFla@(HKpCc z(GACnBZG&zTA8)>z3Yk`wK9vL)yH~@s=!XKU}f}pF^bHj@5K(LP_txr%5o{qttHNs zTC;q;*t4y>Z?C0Ga7T%E%7;vHnWS>hCG9mpj3m@(X*RVRLGdRVJGgAg_0P#a>EM)83D$1LbEWM>ylWS=#Na z`A4KHH?&zgzxh>>3HHJ}OXE5oR_M?^UeKMS?puq-;d)KTN&b&ASI2m&E77#fZ`F5& zIXqJ8x$sl5mtjjJUBl##oLADXQp48oVr(^w$x{fO*Q`mBOFl|viC=@2o)3@J`(iz| z4Qqqae!U<98LY6Tpo+8C*XH^0tiqi5*JaV+wP1+p6v%%42naSaZZF+|%ca^#YqM?| zi*pkw{BmztXZ4UcP+^v57}yJXVcsz9O-5ogwbd0@TnnDXaiJwmB;0l%SH4*6|FmBH z<-)tG;)ZSBX4*FWd)moRF?TR@?P|$$c-Kis7;h-}0me7-RGG$c-psE0W3qUZ2tLe{ zo8KZfXl{*_V{b^|N*cq2<$xgHxB|~Atp5JYaYHowDF>83=EwUs70mZ*m$`8>vwTNB zB3iNYYqvr0d*2@UW;U^%{UCv^ub6k^H-B_5HNq=QcQT4@0rs`CB2-U2S}*fLJtX1< z9?@V~3>I#m$AeZhcl2#I!^yZM*Z-dqUg)}xRtpwG_UaQ*$>dn?jCPXri z-Q>ERoj)Hak$Zkz)E2Ld>nNnejD0sSqMKsD0O_CbKqy9t4eDbko^)auLjNgVJPBz3 zGG}A*Ixlh8KlioMZiV*ELAzh-i{zV)39)q4J6HhP)%cH#T`fj0t?33FEBKwI4e~ia zyjj`eoNn2vo2qcy6(R{1<}jE)wRx8=F=!k5S4(9h`p|P`$vo;cUD{L6o$Uui1Ui4wEV+F~wlAw9!t7?2*nmmB0 zO)m%kHJFQ}nS)VM>qMCQWpizJo9&QG^vK@^-&v8%vA!hYQ$;=lH2%Thh} z<0Jh+Wood@H;oI8($Zn4?v4#~86%|}-Uk&Hb{i%(g=6FoqG$q z{Tq~tt7DmF;Ov4R6{V|#3S{;vEa$Qqbo8eZe#%yq$A+$wCog3`T^d7o_lsgfOs7vi z;m?ehI=@afdgtmb%X0!+z$wS-3}2LfJhf%RBrEVZ-LI@x= zGTzjs__-dUt8pIb9Ar}2o`cL_C;8Oz>{5p|3*#meX73(tD@dLKX_yhaW;wt}KIDDJ zE95;)aBjZV!n}VG&icJ3=KdTa0h@wKXnXK*Dlc zc%mLGa2&BKuW3$KV{`tznhh4!&Fe+&#SsiC3cGv|nJzxZi50`kx=FpuHLryqG=C80 zo<6S1jvWG=Edg)C2s8d^=?YX zjF70($&+ld>D6m0oMcBm5ABnC`a8pBC!Ks6GhQtBYNxC6>zfR<;_2Yx{wcdbV}1v# zksWW4kC^wEx59;-8_n5li_+;mCpJKaWy+#9Lw9BEogoL@l27YnFZx2zX*){%*{6;m z5Lbr6mr0(K!3Uy*J4vnjD1TH_Lru)Q@5M@{jd#9Nn|D{`P_Cahqv@9H$gmvK$ndQH z0t_F=H9zY-~BRu zwIhE&;=l!&8(>NI@?rRHCxPnu87)rhqak5==^TutrJ*G*Np8JP71yI=SfV(AhY648 z#mmKDJ~#!F1@!HL#_W+RX9ka-a3KYAVd%41Rip+oYa!>?)q~lS?>y5F(DB)oh|GfT z6}lCp)PH&3_dh9Bs9hrOYTWK(L^s|fVekes8frD!((;-{-}UyhYGmqZ8jLG7;Fz7G ziu~W@RhB{R%^IfRNBcGgIh4`evOq_LCd|Ujan)TMWfCQ##J}or?7&=TO1S={@t9Jv zRVeT0R&{^Y={7j28~bWqS6?pxy!-&ZZWMns9`rWIAiJ8zcUfzKXrX*qFLm4Zf+5B$ zHzu8(cdFr1_wlenQw-cYdHn|zHz{qa22VK-U~dY&>F<|j=bdxeoPV?yX$^j$y=?uR zCx@E}34Al(?3EBhZ_?H|pINcP93Ksmh6EP;#XWlO-^@k`!0ZPJ-kpUjN-z#+)u6>G zzJkrXd`kI_QPlQ0h@#QUN(ilc)`=mLk`@~*YKm0+IFrtjl%*xnJV`(hDv%K8qXTvg zQjYL$@!I?T(b(6OdH+jFeCvAa(d7KrAk6p{@Z=wB&kUJOYj$T0Y_*ZK`#D0ft#t(+ zJpHun`18b%1do2hy#++;6znqle3XRfyqpv9`c*>(F_OOe&H86}nJ2>t+gi~4 z(ZoX9x6h|TnxvGlH)|=xjr8(E?%tqqLb#Nc65kGdKS}e9Q4;@~&x#Duhhz;!v6WUj ztUU@UxtG~)+sK?<6K!o;AEE7$GCV|^G@11rXxmM5Ane7aZ!~|RYpx|9z}3JY>M5wa zOx3KRw7A|)+xBQ+M2hTH158J4_5^&^1g?bEP*Nz;Qv2ju;0@PSWe2EYY7#*AL%@}$ zJ8kN-o`deCJt<$cZFhOz3{f}-OOA$+WBUhg#2OZ^G)rIYIF`K9z1KmxL}cd2V^|!` z#*p=A;x0+y9Kw1sW8+vpid=RQet^>FJ1F7h=h@m|y;@cXxG)}1!RY9>x6cVnK}7Cu zLdy({>?CO=?_M6`D5>vn_alpjT=I^`t3Z$fG9RAcOTI!XsU`Q?i!D>K zY{@A`lVbfh+b=K)1c8dPLQ*h}OZ{Gd@0OWw(@xg#=@DCR4#8nY*%ygK_%=J!ojL!e z8Ry?N=KL3XUs<-3Cf}L4xGU{uiBMjNf%cbQk(m{QN}ObqXivU&kj$2zAL5gL^0t47Ayl2*wdQxFuz@UKO4MbmMs6Gs)|sB8mH zN~cur(Hi_W!MA;VQV;_etVc&ioE&KFN}JJVyZR*uVqvMKG_zqqj&R7GPe%Ly%YU(J zU{Mp(IEpaI)ti}@{PIe1W{jJvkwAln%`8;DLXb|R@lWYf7X3zs@0?GR+>f*M@46KN z@Md5}A1r2JjVztYFXG2Gn^3TU^p_o&CmAE!`9#Li~+o6Uy?Dr(6!fU?ta40YxO+q z;I*6PCYklW#LWN1&Hw3Ad!>yE66|X;f=5((vuY)^7pc#J1~q*0cwB5IDg1xcuwqA% zkjKY=JJcM~7{0-`&&71@y|51HU>cB3y9g5dey;~uI*78Q4q|GIIt4$NNen#a@;+Yp z)#OIO9pmx|6+|)cSbsmrc=}}Ue4a^-?jpXRI9r{5AJx8T{ZRH@AV;1 zgtl}F%foFTu$1+D?DckYGSm}wJ)Vz(2+XJm=6~n$xo{(+EsOM&pt}(gbWcFCF9Oag z$ISI+4k3`5N3esf_7IU3Ia}iCW>$VheeP12uD7?V4On z<5A$NlD)1+=<`pfif?&a2?g8Z2QXO}wd;7hIQ{jK-_KNByHf!@^NGH#fy{<<}(xzp<6 zi#>RtFQSQ&*_)K*vNVUH*=z`_QwF!blAJQ0_x_Af`KFt&mI!!y?Z=VrcbCDET$UN8 zaSCQYYla`tDIv1etWt4{#0l<5M&0-ER z+AijEGU^;5Zuj<%YauSs2@9LjPJnemCQmdigeR7cminh2fW1)#+-Bpb49^9t+D#m3 zak8>{4f#(J#L;Dz?7>}Nf`g9kZAL5^G8!7g)r{l$B`z##%f~({pp7$Lzv+}Zmxf^mx3#+W{yoBma7(n5D4Sws6_Zn_TLMc zyVe|mRBzngMwsvIvobU_{|%85%s2qOL@G)Q$vZO#SJ!n`nUKiE%OE<+IB_*)|F*xr zT#%i{a`fdqoB;_!^361T{bNH%V$dN#E>V2ZYJTpo=gQ?*r{oBtZ{u0X-ubkkj7A z#yQeAB;R}R1oCQlyxwYPs0VOFbST;(Epb;hW`34TFSRC&L4M*XC`s&_CKLm}|FJe~ z!nU5`G+$S|nIhCEBb;j}n$NFAb+RkBwIi|WNtd+(c_cftw%QtXl5zU3l$e=J8pnLBwZF`@V` z-d{%5He*_jHn8yC`Y~?to$2QPFCc=^nAQF`H~^n&7==T%1B#m$mN3a z`Q9DTS%nT3OpEnFsB}z^u8y476)iB-c&$44F({ZrPre_-JB3hONlkgN8A5aqz1hqz zOrLQ$%!z=`OebEl;w1F{o$3C!qiJ3V!qX#8;*6(vSGM$>lhvbrkr6~>@g1jC*4S4{95<&E*B5tG z>Zw}*3Ey2@y3%?J=(fK02LvipNGd?DloCIEv?LmZAGpcX4n4~XT=nudrVgY=FX;=a zm4OrgR@-{VTsjE(XW^Jzqw4~dulI50!}$_eqk+O=<`_rR(FvOS1WQ;+$abw3S}*tfgdqm}SiXjLu0XusNuU z@*KN>etIu9*eER)eJQ+pQZJ%vy}uLmhM^AxY1 z_2aeYS7dOP%_9$ZM!$u8hB&j?uVb$u0jVX zOb;%}KA;%Cu^Hld!y;4uZ0+V{0{{>2y$Xps?QJ+)f0llT^^)5AAGh9_C?l0m<-4!LU6fA*wgvCVM z@c%_Z6zI`F?q3#)Yd8Hpi*WB-0CmW3XXVpgWthjm2WJAix{JG$-mTwsv)nPnpB_Iw z?G9f(7Fr5fSP$7IQekWSKSH$s#aGtzn6~YZGx`sIrM#pj{&ks}sQqj7_aBG$Qy!ck zbYyNkh;N2LdrxGA3y<@4GwUut(LZNo`^BAiLuFNK#e(6uD=GfN!wjoAA2^)c*+ZKj|~O0*3EK;pm2`U{ zB;CJX-RZ-cp*w3&-I3hsiuAquB*SrvUA@444ut%(YX7H&?kVDBU3+IsLl~0Xfsi~u z(mlgb=G*?FPfWNAY~=;DK8=>dXL2IOvyun-cEaDA;}3z^{sVQm{pfLrdJ^$5x>t=P zJa21!aqF@uZ2=$t&W;MXGGtVjfb88sHhbz{UL_P%BCO4<=;@oEkDDC%84?UH&?Y<4 zyIXv66j{?o6zOu3hLAUSw8MMfy(%ck*(5QIZKlyAwW%EG)!cM^lf&}mVn^KNu4$vFS@KUw~L%0wKbUp0z6BDdQ8 z_Un5PZyf>sR5J6kL{U+SYJZ;^F-&PZOp1+9JJJ;E=UXut;LsdwWiI{mRl3XCSF+F;BClg@>d?E_5Ko(~7ScE?8V@=1u(@m1%FF|wOc z(LLkzf9$ONJB8-;?`L+X{QnvJ|F^#f`J(i%;v#Qa_)e=I-!;iaD<4a9T&(5Jw-Wz5 zS5$ERh9P96-u)I}9d`@JYGoG!+4BHuNel(-5P znWnx8-|+E~1qk43EA`bZ$gNYH%FA3w2#1`=19A-=JJgUYwIx9XE8s7dZ!e!b!uyac z`m)DvoWfvovjg$ugWl9>yJ=%K@opdLAu>+nO{PHI{tXv}hVorgkh#>Jf9glGl5EA; z1Fk`FS{oL~n-i(wSAj|?moI+KHiFyRWqIRGxqLV}_*?E>yU*-c{&JcMm($Q&y<-5> zyeNs&8WPq{dXVkthzS{O{;1K+3?3Dk4z{qKaa<-Z{q{bIkD~Qb*{4YgZX>Y`f(*e{fj%D50gOA8wNW zaYa`;r+gv{-I8HVRC2XA!Y3ZCLqEcYkU4L|dYl8{!enrqnN>0rdHhsaeZ!^!HdaH7 zJy-)dJU}%Zkhbe#ygP}V@=nTivgXz5I}CSQlYrB2Pa;Ud)zH<%CPY>_;{L_*0ka}eg&3iLSKE=z= z67fIhsO+thTEAs054kYpUVX$+p;j~@zlpXjF0_8$!|^HaaNHG*rHm+?#i+P^0HKLo+P$@YERpHkPSi~zfAbselDP4J zyH-X1!nESi@Lb=jm(XgnNUj^ApXEc_@TT~;^U(-=z%&6g`^3h)uDEWi(61B&8xXOYO_n5F7yBK#jw!ydi{+;~uTv&9RJ$0de{pB77#h-MxrZn5) za+=pBXP9+O=enz|5opqdORQUWHqcA%<$i5fEEzA;PpEy%_E@W8Y`D8d5S@!$?0KkJ zB}23X`5et&MHm!r$nM&}>D3^&19vLDkPvgjzZpSe*|(DfB_7;khuUK{38LgS8 zvYLLX9urzqmh4n{_HmRt5rTX81dk32GqV2)A@gnY^6BUG;mo=Nj1MKdVUmnQDRs64 zD|0^l>v^>igd1ANwua^KGHu#T&mx1)EguZms|UyKt4Urj_-07K%PlFqRm`UChD{5> zpuzMKR!@~rn*#C>sQr@ytWc3x$p`ZVj@$fpWa4vlRDevh z_}7X#*O@${D&T&==#mp|UKX9-j27uH6?N*rPAvVa#MJMvM{oCL0LFeNU|YU0y=8FW zY4mpxy$LYxh;?QC?q?eHCsd#v^(&SHA+rB}k^mF;<2kjXQ=e$OHY7lO9e~|02=u{z5X5U=795^8STPQ| z8uqb;T@x4b`U&9teK`;>!tNZ^s@%LZ8RY9=gOziHMTQ%SLs_by73BK`KD6Pw6g%39 zS)wKU4yrul2(Hl%@)TaoMQG`4<7#6F(#U-vF{xw`(?U;Nt6CblwiAYoHPqHG8jkQS z7dCOSqq^_XH*vyA8PPdl&bOtxd@la2pJJ0@HXsIIDWry`;BdjrYNt3tdd?etrsyE5 zBZN`$zXip1LcIY!6yL_D!8`)5Ol5$(Q&T~;GNZ5u`%MIx)|xa z$MYaka@|&TZAA6Dp7hqH5TS>nwW9F)it8x314WfM7g;mi!SGnMk|Kg|1H}@2nqhQS z(J$PL%4?HrLeE}$STF^}-MFJGY4K|GUS(%U_e-rOM=j{d9wQa*Bnc*#e0e0fD=I01 z582JWgIp7VH7v%m-w2mnM$BJj62w(r)UDA>w9*jo*X4WkPd8pDz=T~+T-&^gwYNpM zK5dN4J3wD~PBat7MW%FPUZh3`bJTGG>q^DJ?v>{x=r&R0J2}h~A9`Li?WXC*MS|kl zMDGI|SZ#grxrR4s?BfV+94Ld|h5Awq_-fa|kczV%iv0+iMJAY7QwD@yl*a++ZF5F} zy8;a4`!rRI8GGSkJoGkEogCETk+MHpP5({U`|mZt|9K2<&>y-IF34-RlM6UZc^KE( zT{;cl26ZlI_oWaPjvFDpT3TJ!rX}y)#HPxg!jSk5dTpM3G#t<(@;Ff|mOo8;)LC}W zQJZJs<}Q5ig|F2;Nlv^n`4e9=1m@lqmm#02x=6HL{kFN_ekTyZSg`&?g2D2JlsMO; z2sFW0ayb^M_~ze+Y8&2C%*nMM=TQ|Kz(~{hGT~g_p5m^&z4Nao_g`Ji%Gm9O!hhs) z8Zw}EPA07Ul}gsa1W}r*(<%eUsrXDS_LiN6LhadB{Bifa*>*y_uRO~;ATu_7Dh1-$ zlfbY&if>>g+_HU2PcuejxF{eZA}&+2bf7$BR-^li;;VPfo|TU}!(PXX@u^BO1)x&L zcG5CC-SRE9yYGF`pM@3axoMmRzQAuI#8>dyQ!z5~z-~oMgSR(jNRLgaTfv3B? zUMf$ApOd<1b!O`uU?Em~xo;5;+)`F{>tB>0rZ14b{CX;nf;Rj}^IZy(hk*}Kw(j|#i0du&0Z45g6bWj0g% zx_6~+Q>QpGi=?l?I2yQCD$)PrF z+0Ua^7P&0aw+ImsPUMk>DrM%@4s%(FRCjk-d8=i*!vQX-;iVqHR@Q& znl{JJ!|=*p!@+MP^kV#}HIzE@6aOHtEHA8PKOIUX(v`6D?v!pI91q49b*CAAMyc zdsmZW5$m}-h^N;=J92#<7e#J)z7nvLlJ_aD;)Jlg1-s;WNfV`gt+z^d{_*dUSY`tAHTr=i?pt4bNx154I z7tB`E(Dr_#y7q3v1Hg|m*y+(C3^!?gl2m}jmACROeOT7Q>>Ba_KD8Zt7^%BD|tpaFGgnJ$X7*b*x5-*9?K@DEK z5`S?Z06B08t3K`nuYW-bH-^#@Ko}z~a9U(l8)tBF@O-5Y!aqD0Ku?iSa*y*rqSn^h z>rFzH1OE}`^S$+-%GXY;>nqTc9V#uw)>1blC^h~aBJ*cb;D6{YG-=}1WOk(CPI(UO zE5~AJC$E2I>sJ~g7tk?!^(9!3_l1_0vB1Pol7N62Q;5h@kHSsZUWn_nkJS5>0?e-H zRhYo-zejTg4G?)?z6jxcbn~`B2U@cILx+uROcSNOIRCs$wk$Wb=Ze)Jo*BY!TFG_H zOh>+atHG6HjRxADni%5hyOpZ|s{cbxI#<9~*OePKmIHo#};IV<~kxR3#L)%rya z#tqi0djn@Lgew`htGWtHjZVu3@vb&jRY%;DZQFfM=r`d=p98p?JQ=PNPha2Ge=@|% z`r(LwtdsmZs9hqTWiN@T&FJ2+CwOn{tX!b_jaq%w*csD(dGpx^s*1%72x`%$Lg$}L zj9=Lqpy~A!$kUemwbIH)|9`KNs#|Ikw8i6emmK6 zFU<5{sNW^6^`bQ{uA&z+jpykWs^idj&5#YESEk_=!|~>TJ@au~Y2f?^HEH6GXMZoO zetrBYn*Kzn{?oebx0e&I%#(O>Eh41!sf{kaZ7ud{5pzj8(tf;CgSgJsrb36TTI1W~ zTNEdLHn82mW)UA3(mv%#o)>DQ@Q{r%$4247j5g~-Y8b962JE$nz``^K;RXQ=d5!v| zwY=0u#Z7WkdEEwsYbr>p1=D1=!nEAhLs8j<(0Ej0{A12A`2tfCfBhN1)g8)*^ysVY zM&PNRQ{vx>b+eLoTou-W;x?aBVX7N6p9av^|EPHXheiEk9fSLUp%cm+sBB;zKLtd_ zv_#ibeDg|!rhVl2{BSIcTr-XpwiuG7)H{iXj0Q>UYqj_Vn3z5c}tf zd_As+lFYp$fD>Jy(8!KDc09iP8q8eq=y$Rx52F3;+&R*oZH-%`s#9lNShkVR8-Gc8tG0!H`$c`O z{m;F?bGbcY45tlB!p6gw;_0q7fOH>k#^`y-TfaQD<)z`l2vrQ8IUlYv4r))vy79v; zL!Oq%M=-t~yJ$Vz?e=C|?p~YrIj+QpAmoc*_w+EOR&OYXy_s zTpg@JkpsTmcpi{{vzm!pP7yo)1*LrOjlj<11Y75i2TjqmEul{E+Tb;jr{;P6=Jr#L zgPrJ8b@qB4Zy4Tro6#rD7P+S0JY3J;Jp^axT}nGJR(?cPC)}3GxSnR(vD>-)vb5|B zdsFZ2piLpWW5;%iM_t&xv!%~%BGC~F$y;A*GFJjq&8o}vyU&)Q%p|_Dy%pQd>31J5 z@-qx2v%!zUy~bU96)VssV1v~rnvI?~V8a@v#Dl1e-*!>=Z!xIzC1+@7;b~qin3#}< z%a`ld=3aEfgR9vPV_3ngl7g7%Ibui8A@oM0G(&jm9%$B3?kD{qD7SjOpa{DMJn1Sj zf1^^MMK-y6dpxS_v>LC56k4ZEMK~W1s$En(I}*p!otdfFcx&d2LTZ6i+KIKlmb67$ zEhcGv2U!`|DvJk~^BQvTLhj8qQ&k1I*>hC~b&7xFi(Ayma4om*$b*%$Kcp9Lm7vD% z2)8ZB^9r{d84bI&qNZ_y`!xIaoBMwclek)Ey5z;w9BWs9P&;dYK~ga&W4Q8++m-qa zx!57v(4c|L8>C80d}ZLyk!}HCY;~%xxXu_91E0a*6n}fCiGC_h>NIol%0+|1HbS+i z+Eve03pLTYCX*V~+1b&=w8x7!t!981(F&L_8e2J6Bx_|siDw#=%2!$w819JKUMoHl zOVw6`tK7CsCK`M(>STGDYtdnZp*wM?N!@=+0UArsp0uhMKc{Xe#5l0FdsQhPb5_B< zL*>43mWA7$#gG3LWBhOoI86OWe-m(`*eD1*tppn*RO@-rLJ6p z`UmfGsLHYS^nATQvk}gl7XCoDj+Cc~-sd*8ut{U+)47_isdAk*v98!zW7sK13=2bj zl4v7mdNBB1({JSb|3RyE z%hOTrSVh#ATef8G1i;%?6$@sOd$6z^z}_U>Q?CfIW8E(FH1^~;!fnlJgSp&7ahP1G zoKbj#g9X`m1=FP;E9nieWH+IEIlc{{V7V8dNe03Nfs=orfQ=}zvp-MQCGR#~qP6k= zMgZ3Y|5Q#&b>ZWn9r2b^HWR(!xXpF+0>g*XQ@u+$5FH&LxpahRPjVFvMlr%N3^p-% zINoPqC7MNn)`~jw8t*g{LDIP&2gAyrk!zw~jOOQ7E3qsFL`(7m&)9K42r1F#{4y4^N|)huxA$XACYW93%@g4mV2_0_$KxmK0Mn-x>aaYT47tPq~ghr;v{ zc8U1%Ou@YZ(7e;AZqqzYqChq(SU|Dj5O;zR{UWU=~CYq=FDg5MIzjbY8nc&toUuMxhBat$kOE@ zRc_EwpXnOEEghFJfvN7z_}23GU9S;Sa}Tsld*~mU`gLrz=UW)g3IvslvQsJwSEW~c z=ocydYKj#=$04e$Vg=Gf;#b>cDi-ML=_c_B2z_0tTMLu#Hcu6~6TAswSKsfW{7|~m zOVLq%NKbz@P6%~#Fj;(1CwxD_{rqjXq8Szz^I{NM%V8VAhGA@+)tH`iOvXE-o44=D z+JC`3OvQ2h9gd3M@)U=M3AYQu!bOEH5 zArY0V1W~;C^6}R7M7MR(~($3OR;RC2+@)Yx{yLWa?OxBz2yLDP~B@FZ{P)%yIb1lSdX8l#- zSe)&rv4gwv_v~VvL?vn7qqTX8AOTZB-#Ru^+aX;er598r$0t^VJ0jU5JgJ>sRAw|w z@tvoDsEbh+u5#t2_Z{@Wx}Ju1==zjri*m(n%g&X_%2=LjOEz}~o}sBvSyFffAXZi~ z6FFz*xXXL8eRw~a--z%uLph>3|B&jg{qp^PAoW25PdN2lWp+G1W&{$|J|B{6)|C0S zspO{QJS7&wU%j_!ZS9sCt}*g0xU}N7Ix$AeQU98@GRsp1<&JEI4VV5MWsP)J_{dJbv@*dLI4Lb<85>99y zpc*&XmUer;XC+kmO;mMb!14r-I#KNDCS3``3!=SLCx(O{sw`p8@YU~;zGmI;B zqF&+OmG+J&v?TcrP4ZEz>zl{JPAQ~1r#3ChWtHJ~k897;12ij~B_#SEUdd)^)mBy~ zz%eR#z3vaI6HZGL5m_1X#KCub4f$sy{aWMcK~?yOS{Nzkk?Dg7Ptr64T|KdMK-dxH^8A!C z+`&iamUy9ZdHJH-3M{B{VhHd^q>od?=thOm=9kSu8dV%s>kTse``ymnteA<={jU>{d(~V z#Cu-WZpv#arsY9!+!$lw%G7MKBkICqVau(0n*= zuvrJT$oP9y)^4_wWd}GUa{_aV8L)}x63&Akvb-)oB+LO2-}szI{!% zRf>9_9Z*U?l)0@nolVm%>M&O+uc1J&K9{C~(BESz3#;xRo3X`ESpZ!=IQ(1p7|&6cFfvdxkU%u2NRqe=3zsX; zjt=`v1_(RqXH^}r7aRBdDcarAGUO|Pr$z`48&=T4AXf832KFJ@M;lG>dZRjexOesE zQ{g4CutK9L$|E7~Nk~`3I~&^qe`InUt#aKSTFBms!KrShYpJp!SOJN9p4qbjg^Hek z?Y{GH3ZJ-5=%`sN<}l2SCJZ_LwWh7wm-O{l}1f$Vh~iDu&kf75Cj zw1;`RWSrhn4&K$Y{J6&W%$I0ppD=VOm8@gaI_pcE(2Fp{X!+>qCmt0mou+*{Q*yyh zgF_K~3-oq?lYyQtsf#&>?QI?*RA$|7YtM+<6A`SbPkB1j4ztt0X%o9O!u#IgwNQpB z>dY2j%>N4PqH&sFX96$0Ga;SD5byl@^VxonKJLN0Ozo@TM8`f!zFyZy?I`9k5$R8s z-kif_XO1a^)18c8@+20R8O+R;4 z3LyK z&9D**IyiDWb1?;8p-N!yk+Sb0G?cINRjloI`rw8}e5hBF=U-KT4oA$kUa6G7urQ)c-r2e^)$56F?4?q=6!VW* z2Kx~PC8QL!Y_S)9^76d6$H?TZv%@1jV8f?aoM-LkZ2BS1q)ph&nZ~!4>Ul8OQ;dR$ z-ePlZH`g>&LltN}3;`fsvCH6*)xeXIJs<1lpU>=L4*QLQQDm-}GKTY`Y@eEI2N zFIc;X^YOVqo+Vjx5FNH$dp=Z6y~y$X`v-XiQdbo@!%8P$9{H5^p^2hBi3Ok z8!tpqZ&F9vfwIdCTQdRMjnYkYU-X8(*xex!;V+jCW6zLtY0J7-@%M}7?N-`kM$Z&s z1V}2oDEZxf^Fr!={U?4>yZ$aEkh|BfQZWbA9RRce zGO7u!rSBlA0^eVAiz=q%UtGVx3kLp`N6dbTl*<69_!a;>*JZwp=BuEjDpQVTzk?3x zH7$PP0t!<_!~t|V7xx`x6bW0u)&TqUj1RBHAFZn2*6ok~{tHzhtBw(e-$9gHQ*U&( zhp2_OJj6RM9{e_jNk#R2s7TQoVFj<$fBSsN}~G zAMfC<#sHAxZv+gW*w7BUwBeQDt!MSh&`H70&H$pGfB3csLHkxFi+V1*PTWleXe_-a zbjdU$^3w)=KM)$fWY4acr?oqazJYtqE=1gbJqKIGryU%HgqD_4&uO{)cogrz2y%gt z_j&NdYO4D8XLyo;XF^X*XHe+#l?B1&rpjV%alI$N$F_dhMtJOqf;L_Pmm3`Vw`!sV zN$@s(aB)B5ntnSigLm@U<-rpq>Dyril~Vl({&Pux37$&P{zGE@mN_cdRn*9+eC+w7 zD)2$+%*A&Q%fS|$N!`1J0O{mK{~d(c(WegC|Mv3r;$-~Ng}%;x#Y*8w+EdZ~##F>* zO$~EFq8JU)-EM+}&3R{y=jnOA!OZzxKT)H9SKIvY`0oj2fBDfr-)65d$eX4Z+1Qp( z*>XZkU>+vySn7-Agu6|Mwj}Ddw$U!OVWftPP~2ndpRn;%%vH+VwD_d2XqCD00#lF$ zt;;Ww=HadcjRQQwzlQdWh}&X4xs&1~#OaO2+b*|(3JI+Z-%mx`LNwMx4R}UYoJw#Q z?%aI+r5^Qq;2c~^o<%XN?wFD%e3Gx#Q%oG!v7dBeChIL9o3U=JrtiSav`~6&QtA@5 zTbq1Sq0QcI0B-~iQlTq~Tw=h5UG~Gk6hyY{Xu_7ep6uFHEW}m{`LZ-3OcqsDOtrp=ml!;l1}q zUQm}{#?Wrjr<~$BjsO?ju7Mv};_7|Z4qBzwtZqz%GX{eeLSLR@XbJy2H|A@3ayQ%ZWvSde&}}pf!HY18&7`T z?YX%WA(XPBfH2QE=J?2Xl3v}G(m+e2uJhpmk3A4UB3c?K`Jq>7*(v;}N42XDCsS6v zwmzo^Z>32Q-R-+yd^xv@xlXXg+P2Pka6xJKO=;XN5XuFsHObu;D( zu+LgCiE{08WuoTUrePY{ErS6Qa}+ko=K;&i6kv_(AMqucnwZj<_6h}%Od)&~5pvZD zi_qip*52T5uBgeR{E1ou=2dPPPJF?OS!T3zfry4Xb6c?_H8zo!)R*Ig?XQ4bK>V%< z`;lq5UBA^)7g7|nl_J0Wt>XmasYtx1y`fCL+c9^e`3)a#^DeUK@$3zw$-jy29g+oL zM6#(Ed;Y53U);L)d$gf(ecnj)p&FrOG5U?poKq`7+SZUoA9l@#8MDdS6=<~ZaDl`4 zskWY4oYk#X za_tzuM>OHXLa1|tfbml)f^&{YK&bExSby>Bo&<|kuSASr*f)~iF*XC1;_XFy+P*H% zjWq$4`T;;-W%^q!R`66CiJ$KDkhLsV16rzr>mbjN;KH+i3zz2Rrgd%sPAdm+>sY_s zp{xZr^FBfVzfnkJze}4ga<`udZLl+EbI5Y6ACeJM`!TrAq3c>p=)T-?xM0&4OL$g! z*{E`sRq@P5tw0_}wb}fMbCmEH*I2o}A&zpOMkPtS z*RKB&7;@8KW+pJ)GTwDYo(BUns>y6o1^s*gqr|!4qz7HFK;ZK>0r~)2 zS(w_Rt(yGV(e)wTb@R%Sci(n-Tc3oB?WckckN9L&NAoWg9X{z<=JG~KdrV`pJXA52K5eR5og3C)vS{8nK!|Dp zjM+83m##mI=w|K(yg#NPE%jZ&budYKcVyV3gTCvh37B(4&=hF7NM7aGHL1cvW`OsS zmZq}5%eZO@WC7S+^}ket{wu)lx7YsP4(>Kn2-xAnm&Dm~%F`y5$+`yTEN%WMUmw0p zaa0mKf^ah{MmJ+PO?>-XZ36A_kT&fe#YQe>vvbsxtE|d}6nt|LZ?;L)b5f_{JLvQv z@>B~J=4nk3h-fV@y0aOLD}8hAr8^Hya;_lhp+;1wR_1RZnBLQRNvBR2R9bq?OV8vj zoNBvY#^myidlnFZyI=5FPi^pQ9waTucfJt?{4h|44;rxc%*Iae;CL_|tKp>068q=< z4yf_{g7X?HjWFh#-@itdcGA!#gHHpLDS!zgn%u%Gh&P%b8y*Ru?K;$2Mh?WV*=c!b07;?1Myv*@p{AB+=+Izw!mm{9vFY=wJ z3bkv$yHk5enTPeqyHSh%?=Ki4O#44aHOWVY(#B}!`>Lr)EiMwu_}q9P=czYuc@cc+@|otbFJ`H1 z(;`25*`^|QwH&WxhBeDb`5myHGlVYq-rT$?;%CtQh9OXe=*ea#`bKQ8O@aMa5qoiw zm-(*tI?>ul98Brgg;F_vXkHn%G-s(e<8#bIWfwVYpW|hs=x`ehp1?U_brSH;fS75T z(%NmMYNk@8srZ3&bYmCh{gOJCfjo-cWDU*{a~>#am#n`6jM+TTo}1w|iS4*i7lvwZ za0FmJ>OI=z^r_#In&Gv2#Y`PK!4DnSANtcUBh@xgjKh53%nDy!l9z@ZQQro4jWs*S z3k1G-;Fsr>ENyoK{u*)@jUK0s3hS40s+%Y(!gnt(Y_zGJA;OLL@6m?;fPQGy-@eYx z>3+CfoMw-y4(jzh@<7f#@V;v~m04niTvyX8bK$>!|3&m#%Z=Iatv=b=!gOn3Fz?h) zX>aWtWqpR0S?(n#S{Z)!03Dy11dHm#Xt8ocuPSzbcXwtqHWWilGV zX#8(U(e^ut`14+CqyKGV5?#YN%Us6~;iE;Umz0>h9Ay3OJ1h_L7po(>lbKhZtXXH9 zq@<-1B$ZEy;hcIr74q5%$o=?Dt;@68AXEsG>MY6jYk32W@#TYrb_!2#iXlFIFP*dG z$I-P!w6&1Qo;RlsQxzhHO9**Sh|^*t_Nmacc5O+R-l}f;42H%n3@#rubcUlPD$PR7 zE$@eQNa;@$?XOuKYcb|m4O>Yl#NS2wRUHY^hJoDke?wHVe0ADr*R#p!-xwU5bnr0} z{VAmTAGH04bpfKx-u%?%{`;)8Q75xv;J)X4#DsdJ3l3tsR+L3en&NgY^#wJ+7F-vU2imuq#Qhy-=H~qB zUz{>EN4MM0-9zHP95okv&X9PfyCZv2Ukkf`3U02%*glQ;rkn8eV_nBmG)mVa@@iKS z5HtbX(T%VX7QQzc@-9C~nC>{e*lOV4N^!se(%TcbEOBOgWNtVHUctcL`IGf$` z;mdQc%&;=G=If+bVjzYUW!JTwjc7pN6VHr(@1li5EW>eJK+7ca8G1=A`z6fv8rr3m z$1H);2wPfZb(u^!?UAH?p!kVgGjWTSO4q3jCDWj=gx`QXK&#~F< z9fqf=&kCgrNl$Yu;le6!&q?0`>h6o$t!^+h-ccRN$-2_s{8&Gkm&m#3X*NQZ5T@cHW%}^yzd%qX_3Uuf#S!;tlusr zFXVj)!U6SmY;#+W{rTC@o1CV1@4&tSTl2MCiSTVZ4}Q*A<33YQ`tr!+?erzRU$n26 zb;(}lNo!D@yftl5S-ub+-R{A1z`Lq5?hv+M%$WdONE^6Ri39g2dP>6LU7v>)93weFXppml%n9|5;0CTNTy_O~lZ6+$c zMT8sWp?WO})G0{Zx_Lf2R_*HOMaoEUM_Gt)KDm2${;X1Cw4hGj)iH?*K9AB@8X!DE z?H&}w$qpjQtR-=AVnyoSg7X2Gz$ z_u)FgUY-a&M>V4`A-gQ%KVu6^+n)uux|uoo@^R+R*Z!Y?L;N4R)jv`BfBn+m#&(L~ zO8V`buELU|;gPp3`(bwWAMsCT_kAC33lWa2Wx5}^skBj}w?rqfS*s$!t{$2J6lCnC zxCj;(!(bOGvwb0(hg%Y46e^*0(w|SY7t0T`JQ(x$RSvrOHxztz4)(*77qn}} z`)P6N_!fnC2`&4X!um?Z;-AC{IXq&(p)_~akaKH3D@wKJwqAGtlXYIHQ$&(}Z6K^w zR5I7lvC(z8sm>qib97g-eMys-HhAt9lD!loI&+g-$xUpR4X^AiK2h;nn=mZ;ZX$EP?JD ziSj;f!B+H~mJw66Kp$TJg#rKX8{QCe{e^+c|9P4IY!Bdt2JBIKMsf()_5uD^62u2r zznmdL%lM0a1ZMap(6)&z0t^7Jqslw9jem58`s24~2$14jJ1_ooQFetI%(FhOs?s4H z?Uz*3pDzAm*-6*e^0*D#x4$E>sy5VQk?}WVQ86k!G?F=p|6F~yQTv?k|7(D>D%=!M zNSl>>;^Hl-daM@vQV^s%TmB4O(-yh|XvyT6K-h36JOw~|-2P`%@}zw42*F5ajq9yjZu3AB7!(MVK9tZ&f zQTsbRh5hLhHZhs{mL%3tn}+SR+?1!<+3EGnQk=n|no+z8Q$XHl`3E^h^In~5VJRQv zIS&Q4^}1z^VpXiZG=Gt{SyVp5rE56-Mfw$b%iE7192%3|=(|q}F4`$lf*m<-yGp^? zytIbZ8!Ad_N&+W6D;NWh@h4yL+l9B(+oRY9JfVfw@}D-nDF&e&Ock}^crpSSQUQzK z{AoY%nu1(}BMhjHDZYbz`9i`tnn<)V*%PGd%AQ5=&^W((V|<51Zmy(2nfrczkaK^VLwW~sQ{)9 z>@m{~#W_`ZR8dhe`s(n2t@S8()=f{i_61ru(76^<&vzF*oB%GDnG0? zwL@O(Z{cYH=F#xK?qJt1c>}D&-g=;^Hm>F%BGsn8TqI5awFwBm$h3Vw))=p|45e>^?pV=k!4OosNb5n?aV}!t{TN3r)F=;b)F4d|ZcgXk`yR_Z|&G z*--|1l1lassTe#HV9_t+v3D7ffSs&WGbLj09q|)u)5%z4hEVaiu|3 zg^ap(T;9MXA{CkH+jh|$En&mZ*5xeIspM77OKnyLiZ=*b5LQ*)-PvL z@hJYHv-j`2rltK6+O?GvzsSpg)iX#1x^Mk{E<#dudb(5~Nr2T2beujn`|)z4+Oz|+ z+@Rn~480V43{&nbP^`_VYs4`K04BkYfyH7+)%=sD2iyrBZJ?Y1)-z@2+B*h$=amn~ zXSbs*!9~JuYr#{b@#Y#~7`h)8u;Dg1mg1O^W%Bp6!?i4$~SVk9+`g@&bDW% zCcqF;k7AlGcXrm&I`{gbKP_{cmkMou<^tGBvbe8XLGe2XaC3fr_R-VZ?-eWC;q>E* z3Z>%0Rh68RTutfnkn4>kN)ER^_}bOkN}^*X*Bu^bb$_wkQy5h8fKr8w8mZpA1lRCn zl1;HOl<_K&tbd8fnUH-|N#F(w2q(U5_Vo%=f;!-hNa?$&>-5XWbmSAkXQM~hgDk#O z>czxFe=(SC!4Z(yLE}B+0II8t$8m(L3}$jVH^jp@Eb%mtanjvcaP+73C%UH%WVyq! zWH5<|+zV3J`X`s_U$6G@v-%_sBJO6?jIf6Q<$})F;3YBodBX{du@c=`B^!iYS(&m% zWEry6t?-f|e(ef_1CY4h{SIpL5j_sx<2aS)>p#Z@3@sNPeFufn?nPOr9FjExf~PmY zq!H+33f$DV$S^Lx&IJQ55dn%9_Eo+BHt_sA2=K|Ky|h4|wbvUToN)9MLQh>|7^&PZiXx`Fmtlo&EiZq;`HIQNZ z=&K0k>|0s8K7g4$_-{9g{!dIhqV;+{Qv4`JeHs8kClB+Vz!?e461?Ow*liZhCs7GJ z_oR6B4(C1IamPpmPw$Mqj?+dc?Gx>cTV%InBscNgv??#ePZhOB7-}nGPm~YHN*~#D znPN_GCwlQjSb;iEGsTjekg`okOBJH{8JEtr4|1%+Q*tg|7-on_*aW0t+Ff zUN1{$jS(!B0#C@_(bP(SaLq!{GyB@)eT`uj~R zhBKoV=qI>*Op}LOYJb7J@n%tdyr*WZMbMLRZ(If0my~B*P);#DQ)@%XbJw``^*c$E zP9{OI^23)w?&q<7S;lhL>U+x^IH}QIT>lOd0@eQp>*RKdGlYvY)quQwEjHWQ*)rRr zGB!f+^^3PT?+%R}5E8_BB7kj~kbi~qNhI%Ei6@*_YvjUSc3P=pj$0oK?)vUOV5q#_ z^~LQ871cPUHS4)n&dTPL;d>-w4|gLB%|ay zQ}Og+ch=*y37au8uSaz_r-SX4#7@5H$t2jb73?ks+i?3YM-&Ko^Yu5g7cS;P*oC*e zg!DxR^`b#gE-W&<=;jl6zN-;4Q-qPVzpNV{`4D93p;{<9A{ewWIHWTsH^I%>Rg&l{ zD*fz7W7UL(moN9c!r{syP1UG%XY|9*u14B&N46FFa^k>Hqek!S*9b z2Gh~tgR?v8e^bBy1rGjK+BW~iEgLNC%pefHJYUbFrSBjDm3;ohczw zr97Ut?bv>!xufyG4D6tJ?>EvEcJVGVhY-cKIhrA8Sj=jAj&0RkE~Qw{`MaCM_tTbW z2{u%)jjYt#UcObLA^gBiwcwi?-c}-Fil4?G?LEacG(=hft#s?Y7~)6z3ik1w^Mh%{Mn!J!cOlVR8?rt#;~2>DS2 zQ%zBwZYuDc-UYccGI_x{`_-j>n(^eUk-imId3`E|8@aBOx!@*dmX+VvDZZ>gM|}vl zeENmKWRzi#fWorp1a;HKA@Q3h`J%iAeq6ks|7a^n)xMzzUsm$YX_LMk;s;FrCok`aacw3wZ!Pe zXQj4EP%0vGzHV=Q_mllAJS}nsg=v~K=|`3h)GMC^)Xcj+l02T_w3U&OwTiD>&9nQB zRs3r^FwNGlP04xg*i}-R3>n1EK=GBrDg|GBrlKW!;}O)RZ_aV1co}^v#KYO;;PXlZ z9R*N(F;NgY=G0c!UJ=%P%3+s-j-uZFL7lS5nYhhAC&)R>NnE;FEk8+TGgag(&hPl- zHr5(VOes;#o2iB2yu%H4$frnoxg8QtokSV}PnR!H24}7h@D>I4%1n2G7QuT7kCS)g zMt4Y7-YK-Ol3FFApT;A-XhKwFZYP?( zvlYCk^LEX5D+{qK$Sc5KvF3cNkcHz5TH7zAh{gWmzt|P!toZWss1j0`-=jet&$SCV z%qQtqsfg`IR5D{X-!h0ZJqZ^wy`O(-tyERyyQ=thh@|>`2i>DcvcsGsQH*KL>R`9b zs&Kj2np|Qt=6heXlc}aZmzU?AaF<18*u=luZud?S&4rl?IT=ddu*hfIfH>A}YsR}2 z$2m|9pQdP8D+#eZB1~^glpw4S!ads1Aj*r!55Q%yF$Vyy8wT;)hja%6X5LO?My@tf zoBN8N07Yx8>h(tRI15-Ik^Ewq=r$YTPVZa&VvCiqy@li~zFv_V<`2Y9F3pnT-JBX4 z9#@GfLn!pE>pi?`CT1wvy07biE7M+ue+R)Wrvd_!Dt%dZ>4F7!-ap=AiTF_5=5`6F z&^BIqid&i;boBG}lz#4HsPg(&m^G9uy-0a8GxTG@r$GL>*j;whr-!ZL*<&kKowIxm z<1P*J@9c#sv*p7SZRJ*zq=^v;R2;1wXP`*{nf9Z)MpuXdEJ)P2jBXCN?}}f5wcr07 ziVd?`XwN7Z0nb`(fOU+mszHd60myccN5Bv7mwC#>?Ynv-dB1+$&gv{%tO;E>U*bheqe3d| zv)YiSL&7ol644_%BKkfRitQ~9!N)&pK@`3&&zRh;Vij3jDbPHKOk6%c{9o+7XH-;e zwk=$UieyBP3`!ORB#C4J$pV6slS&Rn&Z!Uulqet|IcJJOaz-+U1W6@1X9_4{0Y!PY zcE4|{eY)@IzUSQgoiV;2j@m`gA?*FEwdR~_&Z9y{HTn#D_ab8*>v$FmC%A)3ss0VI zi!`+#sl*=Pg7Qq{+WEMYXb{b6-k6j zRz?@gqOCLyV9C8V!|a%1J^~ZAI9|wpqqKJlwV}3bS#1K>jPf&FZryZJ+B+Ja2$WXN zSQ2s2VK}{^U~agboy33`n3J#jidb@LdI)Ii`w;Q{Pr&IOp-Ah>>GF~Otwks#%+pkG*lCRi3`pE(6w}Q=SK-2-kDcUafNG z$?5j8-0k$MgJ%(GfyOW#6c~9;XUOFoyVco)Q~fK8zIl_It0plU_mi$cyF{lDE1bK#~mG}1AA9?!N`0YL==1TuG=d$R>G4KjaoC7SPoSL$A4H?g-N0|(D z_s=g^KcYo|O!q{!>GDun?agG{EY;1(&b0Y6+o&UUgDAI7Xm!B^~-Csdm0kN)!$es{KA4gJISp;{-umz;PM_VJvumZvJ;vzE( zxXRO?s}OZp%xf$n7)8h*ud6ohR;78n3buIXQ$r3n?L`^;FuQa%b|IjUi3u7}O9W!^ zY8!Rn13D1-oBv121^=}&L(co*+?BHwVj(b^*)A+Cs!LiEv@cQRnjR9g9$th7GogaK zmo(zGK5H6~Dm!`NFoiuenSpwUt_FX1v`?LAgk$buj4?fiECw9?6zXoZ>d9HV)hu(z#zz;z6(0r4;mGXhJ$dYEE$Tg> z5{zKUt*5Mucx{iC&)zVu*Ziwp?E{DuK^kz^deUG8*bwU@aH>6hw z4;Nv!T9sst#aq$dkd$vAUgzH07uhI&=`TF8%drXi{&>aF3jL3~O_Q}!b8^Nv2AL^J zPd;Raa)h@(3zq3H$PCtSis!eRTV>sz$&e*`LJBCOS3?^NE#?g+KgR%?lq&%sPm>K_ zGR;=fYk`EP6pCmr`w2ikbZ}B(&Q|X$%bfS9x$&K zR}tm0#r%#E{ASqhWgZNc)oyC87~0BdXs!ir>6`I!7I=l?J;gjksQtDX){ylrK~6Vs z^Z3tU-9oSAI9E*0vOynDuYS*x3(ZB zEV%Tlb&koPY}{Jnr&C;Kl3bv_EarZKl~EvN`*n_nqprD9HgUT4J; zL7A#+s3CAE;r$#touYUkNF-#nB9azmY&4>;Hgwo4T2zX=7jBW6oUFWU)-Y4cp&*OvQ$iE4P)-18#mhWPqu7v^<&=LBF z6h%&%l&|^kg+HN<`M8*uFEJ{RN;~>2U`MTV1iUb8m-6+^O)h3B%w9qNm8HYOlH3r` zDnnbf7R94!%XoDeMt!b|sHqIoDDCakD>p)3Tj|BcSkh0d_2nDRGRJG&W+KQ7qLg&K zzk}E@1eSFAssf92r;NUVOs%i)P8ungxh699U8|m9m)cpgdf8gGzMj{cHO;K8ta6wQq_v6eq*#O~dW+w8@vq@S?{cqZ-y?tXm&%U%Gg}097LjEi~NFa@XrX zG!&RCcaS&8$^Z_eWnPg|hJ=69evC6KWLYu0KDe)&5COuj61GdOFWMwj^()sDH;2sVea^3!#^A`$G$UpFk+B zLQcxmUv4^yhW$k!_#HIIDB*`^wV8@a-#`xCwP<2&WI8TS1O6Pt>r;o(8x;Xk{CD&> z8K#BJo@XZ$2;DfJvq{`#c*X6*xj639hu>-Ao1)MSq5cFv`^v@_+k*{}6l-bGe48WBt_&&y~_ zw-oRoAn!q(%HM&}|Dk%D<34J2;y{5D{Y;(CuHE~V*c5Hl;Z-rar-sjwpY)mIP?2iLFacI-O! zxr$nqfNA8lYJXuj*PaOY)<2cjXnTNAa`Q9AA|C z^PVrhobeZT*F|MGT0yO=sD%P|Ts;o_i835%i>Ny~9k4D|fh42ytQp9mnZsyN=j^rW zx}wbj&dHW*m!f@GS-3$L#mJu7ZX#PB2M;bO|Z^GU(Zv|4d5V(ppmq{~4PzHPLQ zDP3a~#q+xdn9I9#@Q_0;`L^LU!gej#b0}%qwSdp_szaa14p$wwEU~}(aKUCbkE%&$ zQGUmA9?2YCE>e617EE+S6zvQoxx;8hw%`LEVndt6cO-ANqp7r50@WUV=vVB%cTC=e z!s#safG8uyXEKU5sDlHG?U_Q=nrpP8<)+_$BI3I*3XvEujwBhk1cI5b`zToFT!OtsH_}9MDwr7T54nOI_+yV zp}g(y7IC$IO_i-oT$C_nkGx$i;`~@O6<%4fKkae`&{j>#IElVk-0IuVV7Y9JpT5% z5dL322hY)dvI-37>-9YV%KyOr7Kd^)%^@W7pMdh;MNb^dmFf7a20kAm?u%~{$T5=L z?30Y%Ku7UKht$)6tyUuWWU6isjK!_&cNpg0x5Y=7|bD^))tF@e8)w8Z* z@{3w)?u*WxT=}j!T}!<5O9~2%`B#GW9Vm?qEyNQd#zr37h{&mIcaKsbow@rp_t{$} z`;zUHe$JP>gj?7B1S*FGfggg^*zs}s2Jgh|#9KKnE3vO3-#{)tpKuW`2H{Xtd>gfP zs7p-l#?gw7&>NB>a?2?fGp39sxbWOg;Dmf&<@HgzZ5~&aXnSB+H36MJ{na09(NXIp z(n){JUzivlNCB|&>gtdw(!VdD!_By#zSr6J@^fn-t!pPv3CBQ?fb!0Dk9s*Ixap<+ zJOW?kMMv0EEb)mta*Y;w{FhN_ZNhzSa7m&A{2EtOPQ;>JJ)@%co7Yq7)r3vSql7f` zJV8L1-{hn6Gf0x1Qz*+Mp9|9^wTn*&aS~;6ZDt94_$~5n{AhMdGIA#k5H&Hq`#cg> zM3lmmi}$dj?}#JUW+_Wj{H|?=TQ3Pmuz>U4hBIQP<+DahzVy2maU>jK++#yK<7BmG zSgy~xD*P4FSsff8z*D^$@1!pPg?4_J$&OGG$^Qs)TO9Hi8&;?pZD4_U2)p`Hw1G_u zE>)D{18G@jTU%))iynD9qenO@@C|)WhFD1Hz??rvIadI{<8NPL3PqA5N*KFaQyP>z8 z%^ZXiSQ@eQRc0_khhULG>GbknhD~U`)lpqZn&UIy3j;O|LBb|>VR`vQ0*`RQ)XpSY z2-}0@kmh-L`oaS-)qavBy=m>*`g3Z5i9#>(;*O~DnnqSMVD64#fnsmdT?YK}X6g(7 zRHFfB&tI;8W*Gm0OL%(F-pqpcYsOe3eRo{Gi2M#25acorQ0CWu6RrF)*8iWpo-F>t zJ68U(9QF$(PL2@LW-vwR3idlLtwCA+-mWz0HM$!KqX5{w_|)_fWVYi8(AVZ;=Y4@T zb!B|jH=$Rla9kY5yCe_Ju&VN^?9aQDt?2HQU#39jL~-xbcQH+G5Cj6P-LkyU*qLVL z&e-E@b+JFqH9L&gI9K-3@Xif*jC#PrCw}sr#fhyLA_mLH^%otF0k4xr=Tu-VSAT)w zj6mgk#1qI4np379>RSC;x{>M1-N?2amcCf%LgBQ&j7Y(!H;?Y%ObUvVf&4Fl*g-@f zvI=eSg@NniD1Fs^Ge+eCERy@g+`ek*^p%fZ<*`r}RXusWwDqzh^>+DgNyg_r_sPkcdHLt^PeqA4V_N;LHNagNP)cIz2`q1e ztE(J8(!HVc-;%s$lwjzb?122cIsok00Xmpp(^pDSZ}B}U{P`Z@`b$!;E(VD*lMcKD zmZNfGyv3!Ko?zehb2$pup~Q)xm7q1d4U$lnh(xnAm04b$V=oe7=}=xI5cp ziQR}9etA~rCxf>9fsr^$H80bC!Y z-ITN!S-bY(#cb7a-87!*apNh>6XR6aZ>2h&M8XCyB{9&W;#S*%h*lbuwqiuv;}q)7 z_2*L?t5))l8>{9M#EA&KD}BxiJVF4~NCcW8(?g>lveifJnyNb^o?)Vqg0%#?hApmm z;g4D3KhP6i`2I0~;KltpJmnwxY6C;&&?e7ZIFLqWV4aJ-b?Gdm~_|Tk>3ubzjtOvPT8kQ_sY*CkJ^2lL_-4YtEgrJ|&+HEXk5c5kn%B z^Ci$0Bao;r$FYVkRh0cg%H#GK$eY8O+_DcN&l)e&JcW8PJ2DnopyVzOZ%7>pU|nR# zTt^KoU|Bi43n-?mO4{bL z4~YjRLk@eyuC1cw>X8D+>_^oiy2GjPXFZo*u(^3XH6ibUwB{f=*81ijAf1TusN6jz z&FDQ;&W`L(j|aYUxXDOnxX4ZjR8U-587Vs&HbGvok;qv0f|KyG;<`UBL#$VGn>Y0O zPRm^4THpvw`njgh9afeEW zV3l%irErZNH0r~jV|gTE{95;-P@j82xwPc4(AC=en5!>BKzj+%pJp}SfSs(#qRu=9 zp6pxFU8XOC>sT(@*t7mMrsA}E#I7D!I)qM4JQZo*puHt1Nyxtk;gYYC6AOq2+Qx1 zm}DS>brujJe0>hMt#W3SIJZJJeEY+vZC>*4(xLTv*f}>Gi zp$lAKruWb{iZr%DKRGW%Fv0R25C^HCNk5MBx?shXYDPcoAT7p?sgd{y!k@|BavyECU$O_lX)4Z-^Xa z&7Vs?iR+xs!FR;F4$jJ5$PU>kN<)VC`-_RYK3m|@purbio47E**ZKL%s-0Kb$H7aG zV7QlC!CWt)zUmjK%ljR&+Nvr|EQRNd(qRV&C7Ct$xTyG#brK1A9`^D`mh6pJGIrtj ztmqOVa6qQfGqF>@O3i-!m&gMbdW(8hZkkqjGCk(Iz1NUb+hmpJ&d|8qVSZWpqI6$M>Gv0Kf9n3XiV#uJ@>^`7AdN#jv ziZ_WdZvc9VV|&y?F>yd>q!0Q0U33&2+`PtVf0DQhEU`>C&)?ZSKEDM12I7}Tv(f|G znd#8x6`NMebYeF^0MOPpdq7@|?qk?}@8OLJsQLytg;L&lSsOdVZ>X1Cj@uc%w zi9egYic%MqN>{mg`k`9=^m01U*pI?I0MunDlK4CbP_+$`)`{$oA`RmMi_tP-ONtJs z4XszOmiyb${@d0WxT;FL+P3)}J>zr?A!uzgAW39EnAAYx0SR_*+8U0tL2#fCr}qMv zf=5;dsM{c@s>-oblvI>`OY~4;^)H`N{(jD3GYJ#g+b)Iyh=%%C_YL23k$Sd!^9$?4 z?lIGUf12q|be7v>=_TaZemIMYY*tFuV|_}8xa*UZ9_=zyQQo<^pRkUdZg zritc+%d^o--XFL-vDnUig!c46>%R|GBg2v-l4r9+FikrCbD*F!o^u@jIKLd4e<8ND zi+(%weiiDwr4c}uEaNeafHF^7Z(*|nWdejMK4?A!2=Y_XL7b8=|AI))(qJr$~eW_SI4iC45HrfGa%wI&0Gp%XlP&C}}Aca|9Zm^Xl)jRO^Wi>nyQ|h6K z)To`{rTXf|kb66}D|wuu43m|rt%UliKZ<#e2_bDGgrX7^nP4SYTVV7>}V%KkP9G+{P7DQpO?IbsJMoV zmyPM<4;suK{`9&%+ty(qpha~qdoc=5-`P0PkhL(AGuJ5k=mRx8n8py!7HU>mVcXJB z+C{~b-g9p*rVk{WrGvh%;?zGk{_3susM|qcc zTa3Zf@6(#U1o954E1PmW?Js5O^*9G=D1S1X6rS_Rs)iXITH6z-xm&>WdpYEfHeU!3 zWvA4%P@!dA0nmYajCuv;npaXlp3~Tv+$;Ltl-u0(tcH?C&Qe9v@XNN~m4g~6Hy4c6 z>Pzn1jKIkfIimT*4fb>b1imt+^9#6G=p>vpWniN{8lqvt(|R7?Ih}PlevSyQte{l3nP7`8b&4#a)`_#%o^e0{ViI0(7_+`7bR1AQt|& z-v3La^5FoX-*Nj`SAO*_BK?$#4iK(>{oYgV%I^po${BB4ow_!k?_ZO|jDA0NIq^Fr zZhdbT?YOUa&N5PBeUBFtS9x;tR*xJ9B7)b+O==;HY#c}$6QaQJ*@+?mp@)PH=yPir zE_eV2?}bJlX*y5b&Xk`|np1rKDE!nLB!}a7=-i5-Xhh28im5I58yXXWt=m8m?Gc|u ziO$uUb?3QRkbnv|zX1qX*EU_H6$8lE8SmyVbfnRlKiF5MEMRVbwE|?zOD*UeWE|N% zWo_~!$GPb|n6_JZ3ZytEQZ<*m%PG<@p_gl4QS0FQ3ihr4N7ujN0QJ zBUgh5n<`)0(92)E+|BJ&>L0w6+8s%=#EgvLCL;Tn<(JrMkB< z{Nxu4+V|xcH*RM?y_@7~u7RQo&61~VYwPYD%IorD(^D{@k zMtuFW>d+;;CeAyk?GYtUBg)7TFZ%vPco$_48iX!5!0Nf)IEa?zI$y4UI~q>PMJrq3 zg3Gue!TFbCRO)JmFCD?$pE-mU7&9{KoGi-h72QL`pgUYg$;Ni4JpEj(G5WzOZTFrH zo?BwSD{H9TovKU8o-jkSByK2&T2#NT>t=Y1OLs-0Qu*O9tbw09*4NQWuoYiEV4e%^ z{h;sWTL+N{$F3aribpuu-nknhy?|Kp`eWJnY8y*d8hUlA9%i?G4=B~P1@by9yO`P+ z(Qr5BJkyi=Ij{fy8|(v*Vzda{APVSB7 z{cT0DN0)6eH%26QP`zzTD!b>S;Sn)z+$B70TiCZai0(-2Jd9SPkKL)2lh+ zH)N}`QLiCyAq^dT`x|Hzm{ZLKuv&kcm$BBrfqpp&J$G7RDC9`>{L*Xip;gMLZppJq0!RP2w>$x7S)V(Hk(UkV3vUGucoUkB_vXQ+})SToAND&ZH5~Ab$MZA6+Q3M74I)abN|7|Y+aX0<9 zOEUkhwYW!c!6$PF45m|&)d{i6>d3k#wio$Yv3U%6LA%e?kH1-A~u zTI6*6`6I0UA)@=EXZxr83+E@lTL2*9G=OPrJ~A|gPP80+1N}T&HXP4u zqdhi^4@V`VVAB71HT&n;Nh9zItGjMvi9U=IGn6u&#WQ-BrW{8W>ijB|Vue(r-FG(6Ey0KkRqi{+ ze7hh5qIc?#rhr8+Y!&`M*`3WT&c1i=v;|UwDNoEIt|TfNaIp8VRXRVOn)V2po19qT z6qFQA1@7zJ>J3J7NPkk+=0Y3ZMcK7Wp7E)|<5tSZ_!FMby}GAY z$8}ns+1-5MAR03oSx_XPZuIUZrK}uB4|UE{U~kud<|Rga5n!Pq9MM^SyZkZi{}ZuP-z?{l7!IJ%!j;^dZ;pm1!LvPxG51L!@HPlbI#KFxzbHbLKc_?WKPJW>Rph^e zguOtl^vksIlkV$3O=rZx3iT}$#a9YEI>6AIR@oeU4q5l;50{u0+2IJ*;Yv>Y0lZ0T zw>^A9v+)EuoEci4mFzwFu}0(eL_UwuHodrrJ<7o4`c`IaQCwPD_m&nI?g)kyEjQXA zOdnrp9Y;EDxm$hY+n~dB3meP1!sLa&&Z(~@_O-Uu+ep?A3j+i?HhKs&h5-Y_IlJDb z@f$&wa!?*Jxy@uFrV+*(@cy{S1;C3cxMhMLB>^TKkV-MFiTPw2iT5O$FP+=XflrYQ z>Pzv@%jJK@t@qbw`hUn+)6bDzolX};h(HWg=E%7kriTU!t4y$$ZRgj&_ZQ#DnpWhu zt$vI_W^kD%R8Il$s*7fYtn5TO7Q&FBHsk!n4 zgFAs?5F)Y;!b=B%XhECxLjxbm{h&H}VNcY`o}TB9 zUg!AdB<}<4G~*n1&3;s$kMsmPFswU`weHO+<=W_f*&CohvNO+J>oVd^1ntiT)y0*@ zY46D|obziBKUeNaoqZ1dtl?R9^#(S|cn3Y@gQDYP`#8fr=6aUGfM_NIXEmhCTDjy6 zn_TjT9Qv7qda))`t?h#`q;MXw$My^S>0j5XszQCQ-H6qhuJ2{P(jq@C?Y0ylP2d=& zJLbq@g(igQsP(HyDode zcEnO70m*)-x1Ak@V^3?qWyhW_sC)HwB$!^J;OC7W(Ae^Uf@Ca;;8hM)|@dr6b~bh-6BGRcBZ`G zP;IW=&9)&Aj`DkE10o_C8GXVCsTg;?m>BvRQ7^lT;R?pf&=7bmg#S`0% zhFq2PMU%LsXLRg^d=p+S_D`lvN-ZtoNGg4S#aDo}+8zyKDAp67^l>+vA~yN+i^BF&ir${wSrA9Ax-wbju#eRS*HSHm z`WTnceH%9AUms^d-C<^@MV=)lB$X=3uC_|34-{-_e9_fx|6~(u zMc;cm@@DK^JjWL|JoX$-3vgjhWO|ADkhVFaMqzo?=*2WCZKl$R`fXaqQ}bkalynYP z7Q+tpT%2gi-h5X7*@3a-9{*}n)QE3RMqqn6n9iTA5sVTQDamnMoqTa5tS3yC9Cj#! zD!|z<(NZ+hn%dHVXXm}3v6enZrKatJ68M9P z;k@+5B)dJ45cbSQlSesN=}^k4rUpO4bo=*85zA*(-h_$7RG zvmi*f-YTXBveoJ@9_esMW=Ee{yTOi~(fzl}DREhv;QSJ=5YU=CvAW#+24`xfzX)xl zy0Wf_J=!PjSZ$vidjBWyxL?$CTGOK=*R{$Frn*10QBG}7gVJ&HOCS%CZfOO^Jy!97 zpOc>Mo0#p(4SICX?GF#*G)gRXOlVfNf2dbAd4sOCP|(c`Z^Xw60ZnI?9Ws#Ute15V z1nK6*x@X@-CeXr;OGPzPgdSInLM4VRSW(C1!WYS(jgWX=_UHPdQ{8lH*w!wm+F1;W z!!K~Byx4e$D#S&>yJi*Yf$$8jCd0|j;Aa$depf)gH6LVBBgv;~(Qzw}@=!1D^dcQ(9t#*TU%o;`*74HfC zs&7qyN#91egek%k=!NRSN(rXD2@(%~%|rNO&TB{bx2uY%A<&c{`iI*92FXZX5{`=44;LyLqYM8I!>r zv2!A5E*i`?0cQ4af6|b;H9!lt8!CTEAxD%K`0@04g-gowuC^hYyNSuIGta!aS~;=L zM(m!zpl*!pqrDvLCL=F;GP2~K>XL^XFk6se11t%kmpTYDo?yhZH%-HX9{mITgYCki-ZYj-l`mS9Cjvhpv0qN7xuEj zBva33V!CjxDzUR@iL%(O_Qh2!95>!z+q1A$xIe)x7AkQC@>+;mD{vuchtv`fCqf;W z6C-5g@7%*u200E-u1Lgo$}aX9tZSVX``sNeMcT$=9$SXJuj)jDJMS)1u8Jdfj3UkE z(ynb{_<{cGW&7PzyNJ^Dz40#k)a4l&#om+v7Oevn2?8`~IYD4R#Z;&NQbs--jK$M;` z%|H2asJ0$L;4#qdCom`oA)Qzk@ojo-2;Wle-!BL84#>_q9&VTBp-uH^uNr-At`YMNjlpEq)vEU+(eGC|j~zFJ zq3eV1=2pF#ZSYjpowmHp$XCJ_7_M&IO&lw?DViA9w<;I2dA?gjfq%Xt7T`tTUNWmX zMta=#$c7+E6Y*;MWp!^f&Zg``v4Wz~lv}dY@3h}@*Dg{u@L{-Aq({Nn6x5KnS9pUa z{QPP+XVIL6qiFRb@a0bnojgL1Mxy+>>_?(M5viSV5Gm%}7FZZk5eKhzaqy{5rf`w3 z-ddzHKw&F2%15Z^Yt-E>V7h^q=M8eWw|2?6Y@|Ud)1$0SVXX2>C5^8vZ3kZ~ljXFW z!p9BC&ZO2pBZL(sr5ZtmsA!}0WVOAiw*Kf zHkBdMyI$4+Ts0e0!7R_7HyAZ?Mm$uC@L^bxq%vSR4uLUOR`xGds8SLid-04c!JphI znL0G)KpSTYg%b+rZiNEzlHQG~TkEMLWzie+HD}6f*i*L<+f&|85_#IRYU?n=NW;n5 z>1tNGrTin4*|j?om6Pit+$ouQ_+XK-drUhOPF*9Z+2_{hPd;$q#|4z_0M*J!#`T>x z7mNMplL?GKs776cAYLf{_vmF)q~^w0-p}qs-C>AIvJVI39k-ZE#{0j%D=bD-J^|jl ziGo6#R_=$c5fXaSa+Xcp!8v|if@{*4wcIA30C>x+V&r^|heQCCtLCi3JjEcjR$~PD zEifemf;x{mlGD!Zw(ojH9G<4}T@Hg8Wrc5`mXly62+2CLr@l*=&e^bKX5*C(Sv`*K z;ioYD#Dn!y0wKJluX6s_!^}jLOR<}<<})I$xYpHU#rJPE?r@51E4`~VOlg$mdM(f! zF1z}SqwO}!q;po<+cS&&T*D3OoVOI$a!OzIRDZe{*EN>EzRNpGsxPWI_aLfa2u)Zs z?^jXP38|ndLfqY3@Euo+UzXb9>9)Zu5$aSH#>Mc`~I2UstOcwKLlD zKrB@T-6NWY6NV$!IBdK?d7U2P%88?|7BT(N!;v27yGRKb2`iD*H_WRdTt~pR4 z5Pc@<@GygSX`Emi-`F$G4M0jGVQOi#L7|SX?4iNr?l={#?l>~^b=HvPIF6fJzYPq&bGd#z@1B z58D9Vh74=%$#0~z!7yZT=^Lm9@*YZc-D$QiTF_W7too2VCHEVs3jn8qzlB$C1nV;@ z!Gbu`vrMiRaA3!$dW<$&sc4V|TX<#|OwJmF+PH(L6tS(kTUKYumH*@rt1SXJ^XYLH zbr~Il^y;?4cd)aG#6Je0TM&?^79 zt~^<`iJ#&PFM^oJR)2vt0P+2S`gUU&*Ng-hNDcT{%?R-|EFdR;mrzoT};;sZ&l>)G%v*8q z9*fjdK6#`C)5WW6HZyB~r!V=VZT|lH-^5`&_%4rYD4z#RzdVc}aCLN%w=SHkLMM}- zxvZDpHiS~!x%P-0SS??USPU~riJJIg(348mUadtfC1^&S=U6 zTf(vd>rn47WA>QfnJ?X3U`Fc7##|H`3HPe4tE5tziG-c9HID_ioxZg@nfApdwh9$$ zU!{$ijW0LB;j~}CPBSmZl5$b0G-n58iJ$g9Ej-{Mo$9Loj6IZAT40VS^%fqlw{<_z zPE=VeqRe^UW#%iM%~Pgav+&v4ueN0NL854=v)Ds=VRBtcmm6U(0nhV`$*}=Xmg!8x zmDn@OZ~QSA{QR0%%4I4F*Idj#??t$X!yRW7bLg0SKw1=thW>`LP@agdwVvu!x*=aJLAFCj3 z>B%8iSi@;rf9pWy>PpOIk9(<$KLNDycFf_dkB#q@QmT@0=#zB7AieO)ME1OFDDe~tO4Z5$lo>7FBK zsIrQCy9Cd0C0I=f^SNSJ{fOx^UA7w~Vlr>p&@QhC@VpCm%{P>{C)w_b_62RImBK_D zkR{Cvlkvl77Bl8M7Bh;b5j;mbfQ;u>ycimC#xJPKG zhW=>9g_5H&Eu-aj+k?j@1zUw2>Nu zbI<)68*k8I%URbHMsvy^^pWyH`B6na1lL&^uvenp&D*sG{-tQ)9U>1-`Q4$*Rl$9A zU470x+o#*W^+m4m>TAh}w^gFQvOVa4Uci6 z`FAR`2VmX30B=T=muNWAc?=gK{dCculW(zxIqH%4-h+CsED*#aBvONNq~y`HN*32g zYWHyqE8;GV`g~1r_5?H3!#)f1Ac;KbzvDSS^PTB-3NhDt9UsxGnf64pg{i3-1ICxsmo?;`yv6CA+H*9%{|^AlRhOlthpS>&3#S_w*6rQtmdkXR3pe~J)v$Y+T{|I41=%*i_jmm=+@gx3 zt8$nfi|B<&M9!X4Z&l=1q)2?>+G44~E5u^&jqh@jf2^gN_H_5XN=o?yeZPTQN^yss zB4=YU?T=U8BO+#lPvSW~tbD(E@oDYa`lhdQ(d?#!(T1c2OS`cXWqdSfC!`3bZ124g z&Uk#LxV6R2rK`2gUwSBv_4>p)s?T*NTle-gu1YAhAzH3yc9c&LSbDa>0bL*exdB3gc+O$*`CWh^n|vQ%L^=jeWVSJ*iZHSmk{RoXMQK zaMiL8(?`k*H>Qsp@>^FoLm>JQRYpm!T96eDTFZ@vCaQ}rX)MN?c~L3zeW^WY^;p%~zoH7*%y!)4P7t=EoN(hx4^}dD7%1ZVu1V;fxGTcu08n1di2v?00+ajn>e&1LLKUa+e4!o{-*NJnQ)Igcf~_ zl3wkA>%HoPOO<;9M?$Z*Q7Q^?D4+-Upr1Z-EtwG;VmL8{4u4Hs<1DPKs4jaY1ICl3 z#E?@K6cppRckxaBk$%(Re0hI^XLE2H2)wQ+!-za~7 zki>tgaUxAF=|&)H5}Lu zvA)7I^?XWb5g$$k)A{GJs^7Pm{==nK*e%p|^s86$^awx#!o4az)jA|kh;)HLyBhP^ z68qK9!`63PL%A)LAXGm#Cwl)=kPucD{{Ed!f9DQ24Ij}#hV2d!Nc}BLZHe5(0e24A z4DXH=BcJq|A5sGgd&-pI8%WcwO&z}Qh7ylN+6}kF2XS5;BX@T|ho?wGBrLq~$o~*AqaQFIlT|NJ> z_*HVu@@UVOLh=-Uv`CW5qiIlL2d{_y-atHA+OXjG1!9@Rp`kiu6hv=$^N#nhapXyG zRp`u>dC0t_bt$9^{;DKqWXB9+rPNGmJF>7SS08qa)q?SWDP@~#Q<`1$x+-7Q*0_$j z?Kc#^=f{Rc(Q%H2IWNgz7!#1}DV#DUL-Vi-85qc8(1hpwhEEc;QiM(?@h!fAyvQYo z3VP>_iTjQZRK(KMZT&G%)dvE-J78&O&`LU{_w4gp6C+B)z(o5#WHLk+W23czObg06 ztDAlceD|Z~{*3sKM%UwaEXNklhkDWBk7MTJ1DV(J>KKkJAR7S5U<&C&;cSb217Uk{ zLFAIowV;66_V(0GOr?`(k!R5JR6qyo`l5u@_*D~4%y#7#Wc#yZVPXMW%k+)pj{Mo1 zhK z)d>HFQ1tsRTkPV67y|KFi5N81x{G!K5JJj{#3MeGV2gMN@za*B~hH~^&`a_SW|H9T#tC=Vela_*+%?L)6ty1Y?%b=F|lGD zW4V(waY;`jo5d5fu2`=b?DJr^$;ky0k^-^P$rf>l(v}3h$Ms%~Va74R*UQNek>%-F|GC-cGcftyH_)A>rm~TsiJ_VPtArJ|$V^a=o&$-!JBlo(mcT@I2bik- z|FvDi|G)PARR8|#Y9E2 z)d1R4_O~ePcYG7l}2Yd z4%okUDM|bXtE}0~9C6PI&dV&eZpk#7Q;lh4K9ew zl5Fy--B$0l5l|GmDi+$NWt7HgC?)s`&Zul(jA>QN8Z+HEEclXZD?Cj3HbMGHa4wC~p=RKdipSL{k^Oyx9tM>%&!}|BE!Cv4s zU6X_P#0-*;ic3>;%bLM!FDeEi(>*9#?mEA|o=dLu%(Wjd*32y)DRVH(Gx}a0#d~z? zTiPyxQF?umlirs9IgP`LU(2wFf2~yLpmA;8m1H4=vL^G~Q6c}kwkQE;2c;fy;c6pn zY}kWaB&m({qkJ}bG1a=Cs>-8mT4?Ks7})rgP+}9%saT@a0>TilhtO}(%VoH}yMxZ% z!P070GX{H$;Z?|+s;ZWF3^RcA-Tf{+71c}+%sv=f-ATvvA9^!ypX%l?p)%I191cq) zG-*T7KsN$z7)1fK$~`~;A$4u*cZrM?k$qGdWe28KUj7%zkr4A@6acja=aZqmt@+7b z9X~qO-?h%(PMqk_Lyti-os_agaSu-$QdRP2<$OY=hAH1jG1@#jhb>9XOGD#I6XO_6 zKa%h9YpsrR179pXGaxM9f02KwvFjLNTX@-n{-p2oeG+Z_1-P?EDQZ_5UonZcQnzew zU(d-cU~rb8X?$uVn`76gr&yTH8FW!eI>{(=ZYAecMeV5ow6<&&>IwMWb99vZ{_ff@ zkd7;GON}goA}o;sBu+=}WJ(V8f@3}m%vTthIDdzc5_RnOgiY8)iL}}T> zmp;E_}fmqQV@s(y55~zJLbSCYrt*dY?a)I^rd_JI&6}x@*d>hB* zdDpu|tC-x^trM6oM{cEjBh{~dV+wKTBFTKXg}a$n_<-!^eez6yD;5Pg87j_5GG(m{ zazp)D_Ol(hPqc!WDZ9+YkqE#a5CV5NEmXRhmumwaIK@<2!LjTOW`fbD(qfb zDKHG%MYonRag;Ax`oiTZVju+)p;dde^UwstRNaR=M;&6BuJhHei{0|I$jn4)dP=j; zwj%Euol&D5#gVO$1`fFr7Zpo9|FE|!nk&tc(th;{ifWqJf8&#lX(kOH?kqPjY;zDy z{82WIXX>GYwc}JrLu>YjDZOlbC%B=_x8Wtl+jlZjr`X&M z0JWEg|C7TYK!MI^)C?FyqFOlZgT|Io{|W0>$2s|1qtoHpzfLUuIlc5xiQs>%RuJ@; z5njqa$wMH6JAcknjNq&N^LX7qUh(oiW4`P~=Um#}A>kf1mL|gS=kJbOT4U9P z%>2=upp_%GYWzJ(B!q0RZt`g5Qj!mL_3V@XPvp|~mk1A-8_N4{p z+IB(RHGZ1bxq|c4-h6hRe3`nM?WO&{A1!==PcbUVHdX3<;1Q6ONe|pX>1-3thzJsi z+VQ^f<|{=S10(GBB1nXF$hW8O)e*YlC_27fEa9WTa{neG0Wt2-eqUQy+UG{>)W`8f zK$Di@PL&DnC6?xtyCjVuRb72kV?!9bB1^PQ1V~F7w{A6h{)gp$J2FY<)iC26Dt{U} z(PJm{BdJEA$2b=xzv$*6e)o-Ggqdfd)EwzIUq3ApwpU3C39kZt*tB9>zK9ncOvAJ_ zXKz7@s;UCI$GE`Oo$oroXvmwsQpY1h-^0fEB~M_mG7d|byGL!A?jlfo+*NkU?C*9Z zEh@276ZUtE;IbfhYBZbv#`TRyt@|C&Am)Q8Kr)G-1xRp7FqHei5UfA6T@cVnV!gtk zg)cOIfe4U5SwT8t|LS02$>{38{`VfnL7KlnvNd(lJ93Btg{9yDOYt|g76^Uw2OL2~ z4e5iH3Z?VmDjX%>mUzk}`)+oT1{FW_kLDgyL&C(mf(E*lP%1R1TowV4hOFG@57=St z6(Tge%0|8@Y8|NPdCl94%Qq!Zn>v+@Nza;DGiEc9uJ?#nk!+Oa0Z0+uGOF9kMofqm z&O$J^WdaJgzmfVv=pu=mi{{z8>Jln>>TCF;Yb;kN8b@kHd$;0d;9Bv~BZ3lpQzg?YXVw67qy9 zfts0TlBvmC#mt?aH{B!DqDS}WZpDWQKiu0S=mE4w7Y-^;q<*e>{Q~)h06OIIXzpe* zjEB)L&;sBty0pkHw2%NDHboK!9dnqCp#&!Qae)}2l?`nTL~?&U8MrZ`M=}8YyH4x* zCiGBC=Fih@#bSbyRInohgB>{lSNp)oFVIHf{`Dm@T8sIwclY`)XUF_`Z4*nlo?$_B!BIA58T0zv>IU+pIHhjSVnk3Vk-x73U)*gY`FAnRrw-_t z0${R(J3uW`u^`GzsTFpHB&fC%Jkalb*yBhF>Wiw=VLYoL6Bo8^&EC=1ZX{}s3?C++ z5xC9KXX>!m!1VfBR@s!MW^9(*bbbdW41-$G$-yaPJ02_D9Tq=dk*pTL+#TmNCzA>} zr8RJ|Dkqj-hy3HZ{&AIsUiXd8FtH`k_Qn*t!_OUxZQG>`2N7Bs5{y1lSn5xf|spBjQa)eHg0gI6u&|3K> z11P^i0CrbtLx7M5i|k^;=D+`a=TY}h>lk(NtsN#IO55@3Q~8Ww>H!gr_B8XA@6_ z3Jm&WcEME}cc9KKJaXX`0=GHz%+9j6mv zGWE7|gJq&@RGPu~tg7_{_qUU}^;-h*&qXgd4TLZM-S4NRHykJ%o)}uuJ9Yuv6dkZz z{N3&=r-m;L-)txZGqSyo=tsQyJP*+#3e}8x(1Om` zEvp}vX4~9DeW*D!RD!U@z^q%2l0i#NPX$tZJdy?wE;HGVH6m-=Q~>daMqe+dsDqKL zQvMdzIWaj*_Y+;SWDWTAAh5V$PIKf5_zNRwfW+0(G_Hq zT4yV@B#nE^n3#|Avmd1}K4rVrtcE)i#1#Qzjti7~NqlW2B^7hxHujuL&@q1M3@Q{^EKE8On!(kx$7&PIXVk z>f@*BBNBBc4T2axla3Cv6E8Pon#z^{bvp5TOA(>__O2-;#-02^Dg(1@cFxm+L(T~k zg=a%tKmOWo`&%0E7ieB7wgPZbs8rzwhY#NJjyC+a5^~5XL`1}YriDr+x}C240^y$Q zleY#4fa|I^*~W0<6W+Xi=7bXyw{E}ZBy)AMy(w%z4S%9P95hP)>Y?5 zq9_oF`xEmKupx9p1}%8udG4hnw%YGEFn91esn1C$biygsVs)JZC$PtgjJIua+>JY@ zCDM~!=tJ`=AuFc76Ss11i)dS4dRWo+eiI+TZw|))g1;qg;)I3I2pKf5I$#bI7<_Xs zpBjuAuu2~4P=7Pll}ZSjeK*D8fJ$ByAn1-c)F9Up%2sx$VY?75*fk()oA9_!@losUjAhdbfz2uaSZWY?}GA2$6YwTe$| z&V{lzDej*bn4IhUJY&gL)X|IpmdcCz+Sd@exltk^V<1J&D=6{M5TZ6JE{Hp>uhw93 zWw_TaZ06W$t>&wkN!qG=xR;SbD)oTP!1-<|PpVcFx8~mjNXa zPNinC7d2`6YWx_zsuEQt;sd}GfQn%8-%@HUSqq3H>v+`@IzxqFXW?7=(!q*(lHq{~ zr<-DNddyP|`(eIzZ*<07x_^N_x#qOCe)Lqjf7QB|e|=(o1Mr>U0EED%fw9pVVzYHGvbOr&800yWa$ySMDOtH}D6eEawOH;i+4Xj)F~OEUpr`EnPf>iD zGR?2!yC;nzySqGDH8qi@tYJ3_(6}B(DW+u$-g<`{XH2ztpUX?U_l3S2czphV>ee-* zkutba$N#d`a58aTbGDyls}14#QqMOl_r*-??59eAuebN)2|V^&{O>h8&?L zQZ0<`=7r^eK?F*N&`Wp@vV$OXxhQ(9!p(m}^#vnfSX1PL1y;I3w8L zYsm7-_Q>~Is%iNUOW5h+-5QdsL_!985%!y)nQ`BHTNBe!K_!=}mHD!TLIYgIGbVAW zo#r5!+k}jKco6C?-+LoDEp?4aFAgcl@_J2ue7h%1`~ru+s~Y0J(@L5|cru_?($wT8 znq=TMumc&^Ih*NwCEG*nY^yf9sg-`do7nh=6s>pcb#xE9%Oz7*AYd;suv_?ss>U6{ zRZXS4(3adf18nprc6MN=CoBTT2{H;_|Fq2{VcE^6W3y=)YM#&BS#NAu{R3V4%nN)=uI8Uy)S|v6At@An7j3#GgG+7=+@Pac|WtG zaD)Y&%GEZE2gt`4B5wE|Y&uzEPb=ZlvSaN(Gs$N6lTdMAWj=_x81HWMmK(l?HQa*_ zCE0j6_1^5wX-ufIcl^f5c`~S0%1magwzNQMQxm&$*kg9Dt13CTqz|$dJOwkM|8DXO zWadYm(r< z($gjUO^w|1_nCc`_sWmyzATeK6q~QpN_OG)i^2d?1elw3T~AXyMml`s1gDq$HNeC6y1O9s~T|nY;j? zM%g*_r0X5sMEIlXsP{l_s_$Ezzxh>9hQaMJan`T@`%ZjpA&V8G?g_dQ+^JomXMKqu zRHeeEi9kP$!)wY=eZXe$Fd?+o9-Fv>Tizc3l{Di-kT^KceL%IYbxe8Cj${Cw6eqN2 z0a_(Kxf{$Mhyquy8xy$V)j=t~38XaMZ>=fiV6uH!L^Um5O4z|G%7`Z~xK;H}_`5$b zdD_OIr!#=Cn(wh*W3(P1R_c`_O~uCI7>^ItM*t$(!|3?oThNK$B$F^djxliuNw_nP zS-a@^E*}Z91beKtb>JGHul`8KCZ~Z!0wHJ=SRvB`i;q zBHpc6$0%|(=t6hoF4i$mT{l*h2)Xi_)24rAN?OoI@k*pu+dG4ok0M@l2+BexGjUo_ z#Z0=eLq@MgnM%UXdW#5`_M4zYX)a}V1YW_nE@IfuyJxFKiE}d}IsspO1Y20P$Yo5e zAlHE#)vyeg@eYPGei$}MU@tZ!*R)H@q{{-9?!xSCmYQ+4_rK3ed{qZym}OYsadi$G5XIZw+@T?Bdob(*uvJUyF(*?aRDS#3{0>p#Q5^WHdt(v&4JPnCKaqOGvLsy=I54S33Qf`!;N5+8qyI_cxKdvxlC$l}-U%5Q5Q8SR;DxScgN1#JO3Qq<%Vk z4Md8-n%Fer#>}=rjCV=A)MjWC1`*s%1w#b%&-ymf)4LlU9vjlXLY!oT3F(6l`7Z*x zFNgHjZkJdp$tj!GvM_U5(Fkt6{x?e*Sk_pbDu+x`sAeUl@&jaJL0|sMyXi^_+2+J@ z@;dk@rGYBQb;P6ixf8qRqxz%v-ZLMtw8DBmDsXNvDwGKyoMkL?EihfIRjGKA2niz8 zUsgn(H)1NQ+UVw*~rtbH(QeIA=-ze^*Q$%dpKeb+2fz! zD6VrH0!pgE5bByr8|@aJ(W)6Reu|cMz@?AITI zXjxO!L$Ki?u`2Idmki?2y<|LM@fD#{<&#mhBaD_PANhVsb zQ)_-MWpk$}sbF3)jRr5kh1Cv_Avsej7C;R{8(DcMpwa%t9Cb|W+yl4OkfynW28m1L zbU9LGWKDwQopL#K;Rr(ND|ipvOfH?<2eep@LcGl-M+SsdG>(bORa+Yv--?VTWR?L_ zsN;HZ(W^Y21yAXB*n!@(i>R}%rY87rdz5?@Ctp&`^{_ZV#vX}LSwckm-gw44f)VO? zs`*P%M0x=Pt~^Zrdm?yu^}c8y^<*7>JIkMScYe`0o;kiJ5JC7hQ#5!Eh3KK(Ix3`L zRNK&ZQOA?W4-ywMsf4Eo<;_HdM)DKDHN1O=wce6GI}v1T4M0D#ai;8Q4RI;R59{b* zfA3h#1}MK^2awqRgGBc?Vwf_;xj5(PTq!J$mLpjxacGwy*0Jtdjq;~C3I_!!~BB2O^uV`f-3}WgnR<)#Zir&?AmXo@?=f zdeARWm78$0%X^*Jg~<*Nd&OXL(hrfh$*PU{nj1bjDJ7*(+d=uFqm~AxUhpa15U)*u zmNnHu`8-cyq{oAs3+FEFv5v^7B zRU|@?_L;~r@Pju#iangO$TbQOmcKtx9KiM3AyQEtQc~KnF15>{anF_RU+^z=zT!7H zm>A6JpUz8ecA2S%deNK&n@wC#plP&`*~L6TYeC7K7IjS@vsgsh{b6RixEAvl6Uvnc z*bJZIZT^yDcI5$DqbckR=*uB3Q|7QVii5=MnRuQ#<5lC)CfWQQU-S1vhjSg6mZ2qL z=jYIAnIVbpXS*s8od;z={3=m2zjZM8v(HOt+ze_?B#w<0=s(b~mszdOeTcmHr$4V*q?B0FmyKxQK*a7*QvNs*=`w|~gvN73k@klZ1%o_t#k&yw#bGmv9 z;v{YmVw3|5?9M~HCv}?x=^?XP;rDMeqoVQvF`gvk*V!3K;VwRWxi%bgCJ`3q_Da`B z1Q*VF&EP>Lnen|nIs6pChRRz{&Etkwz^~Quve*cTDn%9pPs*x{aP1{ctfySt65GaOSAY+1nzksThxVwkyT+FAekRqG-M5>!2RSB zOZ={50^}nIkS7&KYz6F&f($m}?racUi+u+S(k2+E)`jj>9fT$BCvtHBYud$ zr@XMh9#*|>T3vH%1pB9*%{YY zXnjT6%8mp+Z8v8XpWZnCD#HSN}o&fih- z;cEvM+UHc%daOtE!sJ4!fM}>r&Gl?cdoz>R=TW;a(MA zFAY1uhI-qBMKjhga^rIL*?SEkecR5_`N$eTJ3#QjD){!gSKZK~q=89~mJhjb5SAt1zw?+OM_=Xep~@L1sqoN*xW+c@2qs zA^EdLg*a+F4V8z$gK-2){27q^4<3tJ*Ff8$MIIa2SCSGrsotN(TJ<-aotI9j>3pu! zv-8@_ha#SuJ@E*9NW%o?EMyLH)1iAg2%p{#dk}4HW}S-Ln@A=iB{K+g=0EVuV17_9 zn1qM`z|Gp#xzxQLWejD{Q+_C!TUp`mX!mGT{l2}HtNE#pG1s}_cSh$|c5)%)%HWz%52i>^Q!xsu@ z!*w&uUSX>|@$BTr#WbMz7QB^i>E$Khw|7(c*CSsy%L3 zb^4w?LYiI&SA5IENwus%9#k3f!3>Lr_{IafT|2vb>0ozl;k5Jqj*A%tGCcKk0(vBx z&JnTZ-AZz@RovXN5_0ezS<~_a0`U?%(J6mzS?5C`Hf>NoKT?JySQbd@zJO(<(S9sSm9sj@Bgnk=8$uYi<^f1Yv%Dk_(lJhZ0G;v z&s@!^udhE;t#rKknO?Dy(1A?E#!$(MsR_%_&Z&>FDI<+fB%0(iAohPzJG@Ydfv!jVH~+Mpav0 z0AyZ!A~=JzwmQR*uDK~3pg&0ROi#?3+E$J0v$f}FihSTfVM3BmLo2?j^(LvJ4}Rga zccc^x8<42(Ff}*W0L4Wv`K_lhhg3_@3V0P$h952W&%rl>GORw8I3Ja71@25IX1A29 zSas^BJ|I4t2{oyXLbqpHf#MJ_wp{S~ydnlxJ9qg>cKjtG8Im`#zA<_}vt~y!OtMW@ z1o@^!sbRS%QC*N0gusMe9P9G|N j7en9ZX)Fx{q~D+aJ4n)hMwI^lS3s)t|8e_}{hIn;fPelwh)=;XAsj7opf2E30T4QOb#Bj}0YA97hR4%HGDphKz zxkafVw6s-XC=o-rhL8xQjiK`N-goa?|NmQWt@r=$v%Y7ov(8z2ozK~4@3mQ9STn$3 ztFsr*0w53ofDQ)0A^{cv7bho#lYT!6Bi)hJ{B&-bqMIO1_(tip_qI zgZnKvFTdz%amlmN=Vj%!b@dI6O|M=zcXo9XdV1fz?;9cxkBp9e8K0P?%+3G4u(-5L zUElb+x%KTkb9?6}7YG3ViFGjl3HC2sf(Kk|?CfB6$WJa1TiAiXg6tgfI-El0wh(u; zu!3$J7xZ-2lbQ~0MLj!)h)2*Mk0?xkR%!hw+8<>9Kd|`!i0p5$|KXYdc)_59;DH4J zQ(${PRke`q|J2dJ=YAqKr_BmE2i@M!4X`bb#_|#uhiJ6X`IBeZFclB7K83GbZAy*i zY7=H8;X|X(kKCHXWtdWjJ<}q;nn^1`ZOu6TNlSAPUQTm%pL#*(?2t9}Ok~AFbZFVO zb7=byLw}O|oDf5#=57Wj2^=DAT-aVB`N5#J#@ql2U;W4y3W1w4y>|`B` zlqBdlg76Rsqe|*Y^<{=i_ro>aG4m09=9aHS{24D(IL6|RnJgjdM_e(LgPz#7i@@2@ zl;K$NPL!kEy2bgSH5s_|3#waaf*^ZS)7`~^;5AZV)wNy>f2&OlWt;9#lb{72TU}AM zH8wZ7bG)0`Zku)?u*m6sbC-?l?K_!kd)7NVoF4rV;J(#HbU6;D6)9k)mILKdElLIc z)C1m3Rq*FMZTG-gynvkWHG-7u2h3*?7I2%tI-_5tXT_OJfBn2aA_5JoT4X+iJEG_M z5hQa)Y+eAjpI4on!j{ho0&3z@Vwta4XaD8Ka4nuBuBoKkr?eAeH3QV0B{dm(%aSP@ z-ifb)$IJ@TyRtGihnRi4*7r=z0@l4S+{^Z~5b`e;!<(djlF{|_3nG*o;=FnjE%*$6 zYZ5_$>rQ?B?fF3nrfR~1?_d*w10!>L>Zbk0)yp;b!!wA)zKg9uRv z|0lT^;gEfMQ0443Iiz!SO7&)7y%=#zh^i2b*ET+e?9)=zx%yrz)Zxe&fd`H|H1sIJHv7ZfuEL#wAIN?gNkgwt zwLX+N%}nM-%3f4Cvtw1h#c3{`1QOC2UjKMA+Fd9^)FmPr|B}#K>imw0neBfRMfjDg zMy?JPw~)EXbCTGmgiNVsF6>i%<(xE~z94?JR4ToN1rJ{_A zu@;F}EItC4t1bP~?=f8E-ys^1wd4q$=PY2gECh4M4c3~kJZz_FLcXi?EbyV6m5G|A zkgL=>udOY3 z88$GFMp)G=td3i0K31B2bbTop_v&j=(p6n+v8i)1;*nd`HjL?b_^EZ7wFbQqJBqYK zbn%3A!GDtGuJ8jh>szev7UZgp_+|zQ^9ivUDjaG*21{tcICCy_B zDX$;d^h8#bDb`n&SJ=>pqn|FR3sE*Gw2><12%a>pxdH*JTAScThE3Ofl`h=Qn_bsQ zmjT&zSh8R5Eu)HO^-K%2w^FxX8O*9fqDQxQb8QuA(ai!p+L`Kw)*%Qsg~EKDmO$yW z*eeS=zvfRVwKOum-4V`3eD8-pW#7e{nN|W}q7#dqKR|mc>IizCBRIsYW-plFNo^O~ zil_;g-95ia@_J_>PF6OZqGTh8d!SjFm&J9+d@PD8<$q*%t15V3IHJ-^vt&6kCZiP< zKaGciJ~M-+d#AL=Xmsn5_0waQ%qu$LLjAm=f1?+F zm&@^9UKV-N^liUZcF9^xF`*6Kc@i6Q(qTGIV;;l7Z2mn9C(>|eLBK7%gfW`!UdHs+ zf}0I+J_?XOs&NB*DlOKYcSpVBH&l3X2ktklrFgQt;PZI^BJ%F0WUjfIVsxHD>uy)O z=#im?W@6>2i>5|l>VO}4ey5IK*G6RUnmSb!KjG}fxJo+DIVXI>^YI8FrHGubpOE6u zm-r#<$#~nuD=*+i(AR?4oDOE4%i!$m9I-Ck#>L{KN2KhvG+X5t=u#^BV!Fp;F$waQ zsG;yu%j8S19xsp_iD^MbsOV0e$7Hpo-UoI@k4`QSue6J1BSD zsg;j%&(}nH#W}XcM_WVl+nd%~Xq-PoVD{>2+wmr1eg6wn0hb0W88wf38+qf6N(S5_ zAPl8{UzWXL==v!fOuRp8OvDTXD|lRQYY9OAN#`!70^dvptYtG80$1{eBk~h_GxMm( z-1V5&eC;ItH%7U!`^U3y;>}sWSS8ub+vXregUix6N_B%l9LFO?t6j=ZgVKx(w zyKA*0%mS!|{JJd}No_}54G}R0v^e70SbvTaL%?%u<*Kn-jRSiq6hv#z==$hlU-z|g znkIZy!O(NhBt1&<_Y$VfoWxUkfADMDvrD4Zt>lm>Gh~}w>^6?A#|WnIySV^u`t|u z-6no56|3};e+Cn7$^z;RGO3E0J!?{3c~GX_qrP@E>W9DOx*f7M)Ve~UXDPL(v^P9$ zNuu`bi8h5dbI7zi&&VgzZwaXm*R@;+xr*J}($5R|p};07V<4xF+Ezoe^z%bA7|zfU6#eC@4`v!Ue43>M&ul5C9%B-h4z*r>pE z^J)46Vob-b$RcW_Ny@27k)H9z?aO1*!RG5tiPfW)!*E3&x{*WbML`yj{uzQlCuKEj zdqHKWg#|#JMHwo|MUBbm`z4NFj0XBg^NhnBb@A-8m9($nPR|N#Y^)F#mWy0hcrS7Yak8-RN%0Gb zh(e)Ic3v3;X)*aL;!v?;Cqxt!6x3AIjAza;id|y4B=&!P2p>Rnr$OGtVkATyAYwWq z5;`J6BZwV{la%O>2lVS9A|^RSN=AN~f|3dtfIAB!CL$prK1D)Gdg>G~IuQ6Cbc&Aj z+$B*ZGJ2Q=Ifpw~EI9i8Y0m3~9~pFdSGmM3J)Tfdp1;7z1mWi4<>MECN=Qmc%gEli zsjQ-UOHEx@Pv5}M=&rGqwT-Qvy@R8tm$#3v-$VG*XCco+!(Jd_VqeF_C%j2SrlHc& z8JStxIYq@KxYDxnipu(i#-`?$Ppxfz{R4wT!=Jx=ot&DUnVp;ezJOm_-`L#R-r3#T zKgLA_BKZ>*@ct*Tf8e46a1oz6MRJP#7#9(-FJL5er${e}lATk6kz2UabBG0>249bU zU-*%NQ(R}2!P29b@;n!Gl6&nK+8@aNcfg+fzmWX}>~CCOLDVEffOsTyAVtuQf9|5w zSbUC2QB56wGb0X}Cfx+0% z*cIQNdKI(;XVK9}f6)9q0mSlP6%DlO=BNK?W5wdjhtEz`HY<7(K;qijx+`5R2cDV) zP)f?7dtH6h@~G6&qoB2B0*G?j9`)0rDM;wi{E^ju+|?iZd#owjbMFWsMaA9epb*~e zxwrpnyO;QWw11p_ z0GubBaLE1*`QY%70Gf_CVrcGF-ETk~8U5@&Y{2V{brL}4zrjPDh`$`K&VF_(W=QlP zWiNEVV)Py0^sk78|7^1#NAF&u1H@Zs{%5>_>M1@x=HHOMhlfGHUxI^N9mY4-2%t!* zT|;JN_QmspdsaJ&8??uCwI6tD0dz$j(-nK*`xkZPc)kGT7aEg)V7iLdu99P4?)-~P z#WJt)y*&b`h51l00)D?biPU7xnTB7nGa>Q)Qo z1qmP}^TUJ$0_a7)$Xd(cBa@$k@%d%#_X~~u29qPU<=Y6Ls<`MM?7xpi9i8V53HlEF zI%b^!T9cUAGz=PW*s*2nO$IPg(bv~)G7vzm4}*4*ij$UphDT58_JoBAAo_c>%8FV2 zD+EyfD}NRLmO~+JN&=|I5C~_M0Ga{>4tX(Yz7KphQ(bqc>8Ah=R7vsq-*{z*WsZiH z*i{GBH?8LXBn?`>XMV&JwQH0N#F3hRq|r_QH4Z<8sSaEU{QNUh#lQPd$efa%0J`)G z{l6hPMBb)tj{VTT($ux`=jeX~`f^hH??9#0?_%qfF@L-z6{ zGO2rTAQ*>n-=6`&otp~tN9To45DNVbMl}Rsu7h?lIjX(x%^LOaA0z?N1e}zSEG-I{ zL=9iX$ATs;cmJkd^&ezAeaxE1KFr(_Fc_>`ocYD^h5D4glr#S)dv@{=a+z*Nb67~e zkD~pzo}H9kquLYv_5#vhI3fM2Bj0|U+)5+`RrRij^`Nw_9XpI9}5)L z)M;S|HZr7P6|H@@A3)+sJ^qbEwx9Oz?>k8N1Js<$cKJ`F|CT#sXcQiLP_P}o^N%_A zXXt;Y!z!F_KjP|=0!m@}_Z$KwZ~S}aRB*ePMC@Eqzbae{B#8C z^mE5qhF3r1V~KI<%GA;gC>&XcHDJ}pdNXS%C8A-Q5*_$Q+{igymkfU1_Z+C=Rovj@ zxv`GV@~<@g^@{LrQ4)5x9+k^gHwP1%kZ=~28RUxqb4Ts&R{{f9&-YESG^=f8n=~Mu zsG$#YYv~*;U0`_bvaro>PZ(@IUd3fZ#$1_k`>+q0cdcf3;)r!_mR!C|jVueD8DUC) zS%>4lzq%DSu|?}5z2fU-U=WS{hA#+yR}`7!WL%x31wk|2eRn6x8v85_A&6YrwvpVY zQU>24`qN<6J4Afw>6(kJ(>S@@FO()M=8=li>*GJa&gNzvms2G?smIoS?M=+nh;H)p zHLHM=Z!*^reCfX&+#!ak$~2KPO`)q&7qVsyqw7_w-fUxNmzW74p|n;PXk9( zT`jQbjsFtWp=kVhHQmJ}Qxoo4Lh~9wmD3+GTmD`n`TNYu<(bmNw8U%HSvfx&FBnHG z^*1of(kBfPQ&yb19|@^iEf4Fu;N0xy?n5bHT~ZNE)k4Mi)81b$Z0el65O^ooFHFXG zVq)h~TAr$vuRy{rrq{I9vX*bR<>JEHEUYg0WNzg+tdAs6+?#BZ$CzSTnX2ZR0%@txke9MlsmYo_Jlw;lRCUP8m!CBlTEZ`(;#%NY$8 z^q$LfdL@fgi;Z#D`h{5*oa>vq^)gTS)h0(#!_;2ZhFYZ8vfz14iOJ%pvSw8KA~GGX z|AqazE@C-vaI=nDKI&CtVE=&OIZtR~mQaT$qMQ6>oLZ{_0;3Tx>ne#Tf-V;2O^qTV z(|t0O9X#D=W9gi|iK~Vc6%12$ zz6DbTgHBorOc~_AgHW99KO!f9iiYLPE&!e3U!AXY99@dz_<%aD!1HuRvq5Ipl8CiI z9!`j%EwwGD$;>bGv-77Sx2mQz2%v(6yqwR`y9^rjNV_=K2e#&8t^E@H&oWsQgk3mg zV?0v?pwK{?ag7%q?=v`K{0#~hN+Xllv@yIv@C{MvilKM=%=Jn4^H?@pz$_$jbK%pA z8cTw)iAbXx!#<*FTnuz}{HnjV2z2+e+jek0%a>ADxSEI9UB|qjSet zECYw?Odol9K>e)i7DKN+GRHCAqWHyy=L~_-4=xi|9u^7TFx4!+8ZyOsL6C$k8MnA0 zq|eLb8fp7ZM$p+;?@qcbIr-C0l75!qt}YtkxD{W1htv`sSpQftk$w7(VjN3cH{8XD zG%a=o{F=c_izf;z9J<$SprVM!vTs{ zQuvW)Z}X<01Lu^GA9eL?x8%cZD5N^V)%nHuVCHt`k(Khj_UY(O=nB=HGz1?AFe35r z-p}`~k&2Ybz0?V>vKg5}H1G21dhiS8(iT-GXjb3Vy%>oas%kpm4clz!DP)hvdYbo| zaj9Wc-!M`K8yxbJHibt?d=jlEzwR`*wyMvOd&9HzSyJE^XHTh87Yn#c3h$Z6PNkyn zq#7(tm+xNSDd6=Ql)ocsff^mc?v^ec&3!bd6g9F`K-+-ZG6tBT)iMGt=nXe*_?zYjg}7wjl`peicpZ(6JN_EnADd-b$|u< zMIIzDsx9l9o7}*WaK$Z`%ga5a(#(&uOFFMgA1XtO${i>^F(gdgis&i=mxGdsl=-!E zO}@+A<+U3LOnh3fVgexfYY-#kVKtUU{+RjE7BB$8-6M zu!+g}V6RUN-HcI`G+U={i|tDb;n&k_P0-K!!m1sD376F?vzz5|3*@f^*J~Q9o@OSW zgD1cez3R?zxfhTreahH1hL-kaz9|rzSob&NR@Zgugz%(=NB9Lf_$9Q31!RUJsxVw| zo;UGsAf8v&b#OGg+TAZK(NuD7Vz7?RxYHj~FM3e;G{q5LqoE>P_%_^E*hGc>$$EX1+$Al$;E;MuwsGfh}xDd;rNXpus%d^>|AHDE?_IvNb11Ksal<-7^I- zvIsvAZ4E2ya9qiEFV-*X&r|t4@Yvlz9#W|4?t>1tdY6&WUoxXjcdM`5PCi&y$1|#; z$5BA1igV4S!Q}2oM`oc3k~_u@g$u@tU+1E{)@ff`p_QgMn2mIWo>DfD^p#1JuUsiH z@w9wj8jQTOe2KAHK?TDv^%?Et=T!r-tMK-H!y=4Up4js8e;HX=daKiO$ER4xAIp8! zPV?C??7ptn>f1t#xq_+6+`gpyF?buMvD+A_FcdC+j^Ucd3}mLYy%?U<(`Dl^6VSMB zpYKHad|R~66Nyok1d5sOn-DPU8tDl;URcTN{8E84OzPJ7aXC{MEHtD2ySmBO1<;vC z&r|O19Rekd45C=By+1M)ptT&B`|SfQrHZnT(}BcR zg(doK^7-EI^|#;R+MWW%rn_Hge%Sj4=e(_lTK>%^;IfcQ*NC$g%$34&wJC0VQT1`Z zekHx)g*4-b`faRsv%wdU5x(YnF{ivP4{dGrk84Ii&WXp!unM-eJC>OwRP-B5Ms%Lx z3BoxPUiBw3$nY^T>#3Nl0fqCI^8QtP7AzkSxU286q@q;R9J28I@Z|G59 zkhR{;fO_yS@F#lKv@!Sm$O=Q-cX17}_kKO^ZCkN%)-jp_4^7#n91aEh^Pi!ZvYK42 zATf8!`!Q}6pTA-`_Y0oz7QQn~TbZyXH{EPL?L_e4rnK-`Kez@H^)}v^{<9C(Z>(pJXk@@ZOs4Y3U`vD5~=xMZVQC>|_lpsSW!dHc9LT62_TOHO{i&GGKxWt!+ zI0ejA5Rdpc(nAjAc8l|U#&nE--$taPb?lB!LBG(M>pL z4u|>}y94wB5TZ|`HSj&SJSkoe@0yCcw{yiGCYQ97Bh}ZRV`vu%}Ps78g|?@%wT1<(x$r*qe447xg*%PB@#yQecbH?)#Ei=QDzD_BdFn{jcp7I)T?`DT(34`?+7jlgJk z{Y33IjG{!9e2oB_nXG_A^6vV0B^onT}CyQPgYr;_lMU=}Yz6TzOU z`(9bt%TtxXsfkKD(pcvzmhbH`{^IJJueU`uwdtDNZI$~$Ymo|zi31OKhP`1B>x?B3 z-{7X;JJ}yQN=V6ra%^n<{>75eW%2wU?5PU5051zPVA z)l~50(?z|lVWpwyJT+IVdckZs`ypF|G~W`dn^V$S_<_)n(E41Z)}8{X+t=#lJf{UG z_6|;k+@5xR=%{LuXrspaK0i1;Su`Yi7}=XxYSKCqcM*2VlFI8D%}c~1@*ABVmW~^S zo+_`RR+c=H*^0`G2CwkE6Dguc;33e{pY)tx-S7Xz_?>UMyo-ORMaxr12?A4Zk#fi| zwTGix8OL2NwPt(!%?~rUyw`~)iA9cg-TKs};NT`+4x+`Gv{Yq#)2CsqIOaYXf9Gkw zs9}|rBTJUtZXtmg3GzpKRgmK@krXu97pq72dUN+`CS@er1g1M`A;cY3Pr9RE>MSb+=DLTsy88++f%Zm+!=VI(<3=vQ%oy;92i15tNMl zX6hzdfs|&ON-pX-#T6U6jNar#4DagbraBeFoj}LhXlK_g>0Xc1GT67xhAs?o)u+`w zAG_ZplM;C+(3ZqBtp4ORvz>%{+UPgsngqA4dvVGP>28m0UNphkkn(jjQeO)^l3cOp z1s~4USlPY}a}_A4cs7Ub;_pK^HV%KiJoeT#fu44>#x7Eu#h|PBVYb>SpR%X+_g?B5 z-mdbyIR>M^`wzRa)1&7)=BbttNwyg{Ts$5ImL3}w*C|beqpm77izDkM-x?$snT=k^ z#|dc(2*f5aqBJXCh$X*AjV#z_%(F|Ob!Il21yO5EDR%}t^C+z;dJGskUvGP!#&QZQ z_vlTmMQ4+CwPmYqc%%`#kv&>K<+lS1K&aa4z=()`C0{AO>Hi^x*2lil(=~iqk(Aa`^18&Jbim|=GFYeXT5Vdd zGT8fg3;s_d0u%!MaeI;ZA8#)>fXm-MPfRrQyBhm=Rd?c%Z`D`C7`?D;Otz~-MeRHR z35w4Jt6NzGp68;Ka!g)(GapEy4a{H6*6=Si86-$mzjV`N=%C`#tYA)-(?eZ!?Bxj#UVepZM5Ou;OXc|Gxri9?`?!+O4EBe%d0+RfosZTsvq~ z_B@7R)7?Oo@#HD?Uu_+o-kX&q*CDz)#a|H&Y%1%+b?S-jai+pO3jwT7b7r{SgX|{f z=Ap{u3aHNG`YU=6ZQ2j<5P3vXK#S=X)>ET8X1tNBy?OiNzG1UrrtFptSE+)u=5_un z>B(HRRVJ~HYLTKl*EX}|*`3$#Rp)4N&J5-=nh33w!Ol+5G-Lc6&X23a*toHV$`5N- zu_#`~abI6|ox7Lm&4!w?Zzwp835V27g(h^C;7WAPPRqd|&1I#}H_Sh}N?E7Cze}(R z!VJ%ubgj48j(ACv+gCT=&(eZ<&Zho!%U$1*i*ExlBIGvXA}0-4F?s*k@h| zbyc?Cz0<^yRL1=wE>tEj3-x5W@$m&}D(l$KLtYMA=Ug@gbmodxm9j?K(Yx++Nqdpi zknBmO_sC}dhi|3Cg6I}C*Y0mLx-)_FNbJcmHZ>8=DGGVp50W06kmbY)_Ico&_#dsB zDGNib++FV?`L&tyMEff1z<4^-yWg@%x=7}*x{QZTqiq8yt&W+zammO04zlhV_t40Z z55S%Y^xv5=U)p1ZgC_R%1pVD}zROE)=@t4(RDwhCBp)LrmdIID|#vhk>Kyf!I zS9lUR0MjRR`B>81v^#yI=+cez7#uimT|SCyH^lFqsm)|&BNN^Y2bkH$dH_3>TNc1^ z=i=`rO5N;cQ|dVyrH|=pZv`#Qai)>-J1s+R%-SCaX~6sKfoYF;_x(^H@0J_}D&_q7_jYsIL>q?!sHv^5yXy8#C{u~Fxj?By?MsqR--iZQU1 zweI_w7GTxo9Ox)3@QzVk;1jb|;%JxIOLn-rI)4*)8ZFE|4dM7xjgl|z&B;^X$@Z@k zA?nCXaUuxVyA8n1qwUOp6z@H{wa>!4ny@th(76ul0ZNwSX+9R1@@XZ6s4|!u`1CA5 zE&pSnvOESl$Mca<;TZ}C){UG=I=-V_q1|zcPec4P=OBQ*UQA5Yh1DbY^6lqe;?)2E literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-81-### Delinq.jpg b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/images/image-81-### Delinq.jpg new file mode 100644 index 0000000000000000000000000000000000000000..92bbb3390722a2c3454bde8b01d4be09dbe23c75 GIT binary patch literal 4973 zcmb`KcTiJpo5oKFB=jbs34|sh0hC@O5EKYtL5k9g6pZ(^aP{{B#IJRviQBrDPb8-vuaC32SiwPd$6Ot5%OG%1LNJz`89+j3= zmXnZB)H$YnLQPXs6Mj?=sjF_Fs-da=XA=+y2M0G7wb)O#Fz#QN?3Qnp)aA zx_bJjO-#+u<`$N=b{8+%J2*OddHY=Pz3TT@U{LVwJ0YQW@7<4yeGvB$pZYlM$_ z8JYP7g+(umOJ2RMd{o{qSOqpN!wz#ysLi_P^ z^Vjb!`u5HrE)W3z8|z^G8|=Su2^?^-u(E<#p?|nQEO!nB7GPzQQG*B?TSG73JS3|g z#V&Lz~R!UtZ4gXz5f>CN;-a(s*|>Es#PuW!RDm_Wn@=%8a^ z_bIj3r%4G zExs3@16ec}B0j*%PpNnQ&Yb3$?RHVrWZy}@#RNjI!G|Z`i*@wi@6MJJ(tYW7D2@3q zBMjHi22FQTeiku-7%R#tjs0x&epMb{iQzbg2_Pv~+ek%3iSqmhP@`I+~?Qqj**Y(m!mLN?-n)3N(I znSHG^-jtUz{!^ADiC`Y-FAJlQT2~3kVXQjcti#ZCc+5BNhg|o*d71r%{XCgVh~lNB zSzal>iZA_qt0<}QbIa2$H|prU6xXxr{yjNbS((k5SDv8yshc-k zqPE3nD))2O9iI?_>qhQYn(|tGv)at_y&Os-v|oB_qZi<7Sn+|l=pcftV#Dh>yNdfk zuY{t$%;5(s=!fPSZ}S8O<8IxIE*i1lbYWBvI_#-Yu{n24^E>;dif-fjvoFZ?mKo~P zW0$)t1~JwrCCk1WxnQ3-giy#y^-s&AgA?Dy#17yx#0Wl_ly*aj)2+Lz-2{0h>C4;P zqsYY%OaSWm3W41^9A+#WxJ`~USb=lw*^Rw4YZ~5A@wyf;6ZA2e&Pi_8X_tc4QX?8a z;@+OS6CvHv8>mMr7_iGRdVb4kjF@=B!LkAQq+_yNDftBFn~kW^rqkA=T9R&ai>*yL zb5pf&wU1)ICE37+we_@1F;|MpwEWH6Rx_ZRZ=I9zSl(Lr4^h0;&NmHrC9k9KCr-)d zTAyLpdlrbhYHEx$&?an==+5)u@L>nlbqT&Fj@sU1nX5Q4gNv`t$)p8km5(QFSjE7# zpKb*@2v4)nF3v(E1=pgOfH&Vae_I9&jhBl&7tB+g?aP|=u$_Vj+M_2r-Bjuk@QG?Y z`^WX9vsn4D%Y}P_G$ccu`rb`n-B2g@8hzh(ALq2Rh7}JHG4Dmj zH0lHfs@>dH`w-IEf}lSnJBrEH4U4(wPkdO2M1lbLXdFW_WY>`iB){>`=hJJ*euT3{ zNzr)5qBWn6g9g^JjDCF9bjfYAx4>xX@XOcykWV?$Qz;RRn0G%!N83;nJGFPRoVN$d zs404$BGWDR=_gtH_w*|@d<6}XMUm;c43B6pfuf1Wr`JfN752a-Zlr^OUfqKV>9uOK7mGpRHDdm)V?8WT&I z;Y#=k8}eh(!USGV5bMDTqG!@lCPc-7XFTM3p=!o zlF|#KnJ8Cwr*l4D1K>vGFhc zZpMAT<`IRWcmg>6)+m$|vM`)=NCs4mIZs~r`TUV$0^hal#m%xWR&>8PNzK>?=!cWs z*VCRJ{i(Z;C?hF;0PJc8t9G@xFkTpT_|MPM)$hZ-Xvz&D_hYIHOJO75qqTx`l?onz z34j=hZ)R@SQiO{Ynkiqa-dKO*@{l&Q-756z-uV`Zej9!SPs73;EW$KXKdLU;tflfl zuFZXksFzfVefW}CxK!g5j=MDtOTdC*FYrM03K?e?(+zaQa>g@YCtZ;?&n^ODd1T$Uv*b5YToK-740JBq)T?7 z=3k#}m5(1%!tDpKs4TVe>^YNPiWvD`rC_lGA@jc5xr2>sIW8zucO&v8+a2t`xdiqwfmr`L z?{mex2TzUr|32lAz97Oc?I^O+1m3Z~OxcL9aT3WWzB+MSi-p}!;~9A;XYs3jqMwu) zjW6k!UU5~8x@Lltb1X>CCSnK65|tA}wn`SXd#{%AoZoccHi0e~DGvn`0DA8G_4KVvvgWcL(?Q& z%dX}^`B5%Oy0TF$rv=Rl=+-EhyLD^r;O6p1fQ$SInLA!fN^{F@M`^?lUAkoINWtDZ zYx)r@<8j=7!33&8@@^YGZ=o&5Y>ub|mXZLYx(Oz*DKVy25`LXplraIFP@Nc6jCAj; zfcG>J&e1iI5yuVAW zqt9rQYu`V!i9SW1NS(HRS8Ed@&}ji)vDze94>36D=lj<)5%EunhS##ODEl*Bv+mq} zZCn`fk+AtP+;bMg>xM{LOTLAx({Hz%yCuX;p@Fv{%aIe>mB!0)Gr5B;uj9{_M|vU< zLGbTgZ0BVYxLK`}jZy32`N7xWW^EZ-dB)wJtQat9T5)oz{)%`SZli;NaKD&lJ7T

` zap-4ojk8tPn845xCcycrfgUk;W+-Fx+dl5_4A|i0o-Fm+Y}CtV@0ftWB%YWxYajC!yAnocR6-bO?ERJW zzS-Ss$eE98fus;C_||9t{C6}$>_(gR&QwNs_ms5NJ{RoUmeITY`0R<*$7V(iRLK6F zi;#n4VzTDX^}*8T)&${vS4VfQ3HGX=symMGuK1N7 zsa4S@T9hBj&uuec3J6)70=C1y*sGdVS9Sv9*2sR&OA!A>_t@KlbOJeaL={N!xZRi% z$57sDEkg`2v{!S-TX=*8T8^})p(AgcI{wLO2}-fv{{!tHa)5k(0 zm+r>tN(~2;xsxGpUR|B$Yi=C89IDy|sHuLZ2YGYqYG+UjH25cU(%bi4#tN& zOs$ZD;}m}EkS2MpvXsh>WW&-5}qY7!pSE9 zi09iIa}%F#;i&i8Yw+#U$I|=kBt;~!u-R#eG53y!&bw)s%3aQg$_V>(+phj%0^-2t z49lK)<{tW0NP75%8c9sxPhBS9-)~J)p*Z4lHmFxc4J7BeV8WIh?$#$ zY}hqQqXS|hj~>pd+_z{Wq-bHxHk?OPf1P~(MZU~+0?PH6IOpn{h?SoX=`by?NY&s= z({b@CO#Vt4Y<8ae(lLQ>2R{prH+hONHF+$T+|b&NjAI1xd3AWuh!59F<*FaJG?~kR ztXZ=aC4pk6bl)>{rLGJy5sC}z}ZF9`)aHhn^R0+Mw;BHqzKNgri#S8oixzr&5^vT7G(i{2y2!Q z=nzd(9N`bTC<;4ZniRir5Tuowf4#LIwI!Wx%+JwIB$miQ-X2_HsYvi_S1~jp^_Ka2hz;ARGFfJb1E7{I^{ z0&|1t?SRP9Jt3f90{#++0n7+tf-a!gd= zG82!fBUH>AruaCyfLUC-qLtTdkRqYv|T}X=QEW?BaUI?XElA*Uvv7Feo@AD*8#x)7WQmDXD4c8JStxImp7-Ma8I+ z(z42`>YCcR`Uc#Gws!nSLPuv8@e^ricw}^Je3m>nzp(gaX?f-Q#^%-!>h{j=-cK$N z0R9u}==>AxU%0rBxEL52!Hm$KTp)(PqXu&`LPQmqj$Jl|I(qYnDL!U~X(ty{w6cgR znNfJ1d9H9GU3isW(gva#?=)earciQklnSPj?hjJ$$^t3mzk&IcU z1K;TYbe~*}$|XxGXAN1~QQTB@IXtcHm=JJx2KngQ( zyUi>wM1%MUw8YDM3F1BR;+ihPMoI>hq2Xe(ZDcq97wJ;0@Ecc+l=(Jv7A|r582(Fw zmwZ+^##Qfauh~TTw9s^8>w#Qd)qo#O<6Kycgum0N`zlkaJ@+ z(~6y!$J6t2J>TA%>l5;6D2I`hJ^G-T{bi5seozxt-}#ILv{vteS}Jbe^vH6 zHnDPBPn7M{TIiri8M$dp-kM=PdC})8rNSCkDvZo^D8vN$jF=_h5Z0mRuopY6fQr-R~;nu(S4;y_DOvANujB^zIP5^uM<6N($go ztvscxku_&$pA2f#0Xve)vxEA0i1ZT&Sz1!mn>SnrzM^)UQN34Y#%>#3&ai;weoY;3 z5>IUO!@N~@jf+QHOqwJqzEdAnDXC`BiH=XdMSfGZ-}*Q8m(N z^+Gg;^wsAQW!=v3%*VpO$C=5!j@f^ttw`#Td3AFKzGCSwhKNYHb?+UWSk!%#%Un;A5`~Hx zzqc`exc;0OdzrLWFziP}#B_8eJNRRkHr)fSv*w>YZ&{wizVl+@Ty2BbR37Lex~+AD zYBY;p>v5C|tvq)pZce|Y6=KZ^9d6>(9w{K*Imhtg?eE&; zrJoPF!V8UT9Ikj`2b(OwSgp-$jk?7dAN(P*Z1w0x(5ih?-ZLh&XO)AY<-?~J;qwlB z#52#AT`dXQvAmt;AhYUS>h(;c&Pg(+y1QK0q@qwT&uRp5cMnpV%9U}Q-#dj+qodkT zusVrK=V65EloIb{VpHZT-~r zGG+SxU>9Qkfk%H$5}VzqLJH*Ui8}b!xCgSn4bW?P;6#CaXS3%&7MZ+&bBpy-t{rW! z+t2z)BXl!r71r%z3yD?V8jVq7k(vYROB})YkLX~Zh+w6M{?}6o5~)9m%rz<( zUSoc%b1ox=xqGX-)Y`|dc;wXM@|L%I#TB+yIC$qR@k1EeqU$NOdW%W!bKP|C? zRSm6nZQHzF9oWq!w;)p;7?38MR&91Wm=L>_JCSJdSuW63yo5KI_M|kTACsRR@<}$e zE2%3no;J99Gs;#Db-qE>X<4dY^6&$x<;$>fSJK*IfIBOK|mzOMi7A}gMfh0aHb&+fEOd12 z0@rWc;^E`tgRtEbxyvhj^EMyv_bn#la)R!8QO8@IKGr{PY0+`oO`(!#_hnc$VngdGLVZO8_no9v&_}-kCG__~6m5 z;C=v~^b8pjj~D^Dq5&bZ6$S6xkc6`=5Atg%m3p>VZ$EqEL3EDl@)c^DtJm1ruXFJ6 z-x0WbPf+}!grt-}^Z1uq_8$%VzwH_YF5uyS zgNH{7Kmp{jEUcwE0SgR`VSy#eRktfxAVfhD3*=pa?r9B$y7eMN>zlDB|-57=g;kT>jT0LQFd@NN;SdJSjGZ-hKf>f=$!c=Y!wYVz0&`L zQ2u{3c$5nm0W9z?j##Z;(Oi4YNyjNC`;wLxk6Ntemlsjhg6>H{TEHHhJv}$qpr9yv zWk!tdRzj$h2<81&KAfA5%}zuB*<)Z+$KHL~d9G%g{~PY1da!gz*sl1iyS^f2M?sY; zqPO1KBB@aJ^^ry~kIFPRED|oLD9n3}<^*O%w9;?ir%zUv5VV9p46*i4!o99L$aiy% zQi?VI3*{pEy>FU>P~#+BWw$v?-%S6_=%$_XRt48pl6XUt0>0$wRrQaSAk1fie8r^) z438g28KiM_7g-;dU$y6;9!h%34WPRTZ9-Yo83feDdp^w3*uGZNGEaMlxvII0e#qP> zKk({BsGKy)*T_GYUOh4Nif54QrZJERXa95iZTbLj0YL$48OsJm%N{5^c+!&^Av+3wMm=nqu(yw__7XQgyBHzjN8ffr*a zMuHh|T2oa8Li%UF0gk|-KxvG z5n=CN-%alcjkQyKn}B*eCQ%ck7;$i@)SoIwt2-r4U56J7r#Pe9br-jE3CPrn2sg=k z7@{ZH=`;3$kh5!A-8lZDzS34)ka@~P7k#BKi`Rrqco35e{i?p$y)8@}Yk>b|3MK+m7+?|&?Re!q41gTcX#D8vGcBI20pWGv7vFBK0O z6lt)b3K)};8FUA)AG$iKaEjmWM0nyiYJcjSwnKrF^rzMEKeifvXY1hCO}jH(FERI> zyic8Fut0ot(0$N6aefS%CpMS|$%C*t!gSbB5duw=zDEK|?@jcf!W=uEB%D1S_s{Jg z^Z_S}_E3D!>^Z5Z*Rz~gL-UF{sba^n$g{)D2U_PB(-Mc9|=S*vgZrJ>K! zrmKeW?1}KF6p9kSFE^LInV=(DTLc<|zygx|6b?xk$%&U3WEvLe36KD6ri#+Q#S1xg zMPlCSl}-m8QS4cww>`0dM2;id`gJV8eBg}*^ew?jp`L>>#7KzHVyd2kC!5`;Df-*> zREV6-uFgf6SfAH@m@xP5?oGH8nV|b?6K=~KzA>4bo4E!hrT#|qQFR=rGBfTs6g8uz zeD-u*pPRF&7S`-wP##a`-QwSHm@BCF*=UQ~X6kE?jCls}I^yEG5;#Qzl@f}*#W~zM z#R7tA3FgkgU2aSE*CYwe;_B(&O}am&)c;uG{L}ZnnYulFc(nAh!CZ z(_QCd{@00c&i*^+*c07czQCUYWaqI{h}8h@8q+)g31h{|f)h zk)(bJ<{Sifj>gC2E{VRv8L9?ufARO|mA6!ixNFNz+Bw7{!Q#Ha2(^Qgu${%ZyU!QL z=E0>fL*ivKR-~CdaMNP+GPPO=_*L6Y z z0R4MMgY$;84+Chgt1;R#@=)hamUU=G!2Cq%-}6$huic#gKr*d1oT-p14Gp)m0Td;a z`m<@D_>yv4n9G*fYuEI;1^_f&C+_iqIHvERQ@E9mWD}?19e>HYT#79n#ZU<*nrkPW9S$nL1t-b4{74sd3NWj(RD4 z_8ILbM4Dw&rlHn=V0fQ-kio93)E>q*!7ycHxz*2DRb$_}O`w0^E5cAMZ`;Ud#4XQ$ z=kuG=?c1Ww5(S?vdGEG|F+A#x?Nk^lomO!Xn{F6@^)edJ_xeBIvs!XJU3G$IeiS56 zS<`XtwwW2?3gSjQ7+`#pV<_aDcp36B%L(RXOWR5qOI_kuDKoT1Z`Na+caxfcUUox1 zw~6yS?S^`Ofww}Doblzean&;W{LkQ=WBw{i{;HvZN+mv^HjV$K;Q9XkcNab#gL`ti zZjCoKMpXmY0$&?Gy*T%@K3({^S~vxcvTBjJxKwq4u8VIkfFVjq2wUECB2X|u%|1e*Eia% zQY05+$5c`xhHt|dD8>-Kkk@nniXgYAjR@}c<_+`b0XUc4!6nF{syAu2|)mb;{7v!ZvVFrC?8+} zjl#Yb8J();jruF&i(8qZYI#{$AlEug3-V1oO2avBphNhPN;zYB3b?MGi$_;SWS6?; zjB3ivdGw9PQJs}sU_+}a6s(~01j1b3e_zs99zSQ9s8kCNj50}+DjB_$e7Y!H=qj(B zaX`tjs<}L3zUz#vbtL43kt(&x7xg8LnBk^6m8wCN2XlHPZ0blo=HUg`S&PIZC{!DEtHU?#^%ggKMY!Xo9Ik|8PBN zf1PjmQ0FTJqQW!6GCq4^v?lLRtk0apy1gBGpjB#+IHoDHm~ZBnbeQ&W9!VnGn|&It zHPz^nRdA|*q&)!djhLq;zSi{^zm~xW>ZC3di=?TRpzJwkmBygK$s8*85OjAm>i!82 zlqA~a7xrS%--j~eFUI_13o(!NA4E5e@zj2BGk?mye@%v~nPUnqjTqmG>*$tFc-++9 z-b>F=qnPawcz&%{{e|z-Oidr)jm)}6pV=E-*}+j^s7dhaJ^@m;*{Q8m;+u!62G{bq zBMB8>o_`Aa667@U{z}e_MnZsLfD?y5j?Peyw&fe>5op76$JIz9ob;sIO1r~TFAIKZ zI+(MW6n%*1dS!@*3+HrFF_6(Ze=?P|q6F6l8=~d^U&*&$+yDE?ysD`tEU*<;n>lRC zmE{^F7o=M3ab8Z6qc7|3g+76_Clpqz_ZdPYC;ndA4ymxah&FzD*c zlzJdVeHc&tJo>t3#6884i&r>tRgU{we*%H@Ue0u#Uy36y>fMK49BxcAZFR-JMEfdB zhu}4h00zYw*6H-3*u?;M)s1Uw)B9%Iov7D^;}Z|E?<3mI7i-h2bA<}cxC=?A!fmD~ ze3=)iCGnLivyF9!IRa%2-8sAViOC;t+51)9vyUH@N^4Blw=;7?gUtDIDc_M6)>z#DHg>U6s zcF~vu(TQwf-=2>|YE0&E2Wv#gHd!0Bj$obyD-R(<_+a5dLQ&&obi?w~R*GwoEz?JH-5hPObgnC{^c zvg3$5*5^%42-ET7l$`AsG+$fx%dC!@A!~tCTD^UHmAd?AJ)p5QCjmeC1RNRD{YG5^ z9XB`kbfH-u0kM8?rVD4in zF!Lm6`)DEXZUU22Z%9pV1qX)H7vnh19NFR|01C;m!*0}^5(l1%|i33PwO@=guA!(++m=N>zZu+h2OLp6C{B%iw!x+%%sXxp7mFx6}~*ga8uwcwN7CJE+F?!&8xW zmeEY2>P^L&rrB|$? z3-eHVeR>h2bK?`kQ==O@+&AWkIPeA>@hOF$UoysH(-f-VKpCh<$O$zWEM2TcbXIbF z={{jIPn?oZZ*z9eFjZy@$lXl7FpbmMcn&vn@e_{r9-zu4d;Vxf;&YX@I&s7cA5p1P zV%VGEZ|nJ z06jba?WlsF-3|qlq}z|VZDk}nhPLuf3jIc;_H*Eia~b+tacj=DZh3q90^NT1EE73L zhQ~DSGnM!niVe{#l#8r1QmCsuODunu9YFSDhx^x@@E?8lyF2H<5{>+RFV}w^3!}VN z_yh3lJ`LZ$QN)RoVo2>^cy|$g0}Hs;VFAIqZPA)hQ?8s0+y$}Fsd4@7+;cJvCKp15 zgi5b|oQvMJI%>bwcko)L!*YafpXc1gmi=^#r)AV04)YA@)5vg988Y@!8EQp}=Y@0t zXUj~qG%J||i{xt0hi~vzK9i}2!XvHlTBEn(XDQ6dU`3PFNWrf2v91Qkq6)6-xrm`j z_GkBsN>Y<1C(Qy4`uwy<@VqNo&YT9rwE92y|e`HykSgA!IAz2 zc#0#ZS9}9{qdIwFeyTK77b?22a%W%zbU$n;z<+-qKKmEn{~=J#FYZn0*P%TQ`5%IT z?tI2SaIycpn)v^x<8~pJqSMOej;c5DdndbLEsgK=;WPg>Il1 zXw4j&^fW&)=k>?2<2F)8B|8L33=K;5@@oria&P)vyQscG$eh~0?M*R6AReDaX-u*UZ>l9O}q0e|82Rq&Xn$Cvb=_j1%Pf7d(K_qsUT7(_Q z=8QQGP@7(}NgJd2q+ln=U4V{$bx3ZuTOAwraqZHKbI1v7Ef0kS!ivq!iDN6o@C{hR z+|O9ulOsENx_&Va8M?E9*;uf*9(z_@6J})o^i>3k3-7#l$<4+ygj@0ET6l=*E z$v7mF&Sl;@BMQ$JB@3k2`tXYMoxxT-2m>a6{sBAXU!n8;eiWKK9@N_eU6}Fm==?#J za4S{5afF@8js$Kb5DE?G5{~msak8~SlA*TN{b2TLh0r5vW4wag+cWo%CBu@JI~7JY z_sz_ly}G9S^RFu8k8&2gPNJgTJ>iJBZbMeWysF(%t+v0u)!xgn}&+MQmpU8K+%^wtC`(kFYw}L zxM3VX7@XqrhpfkWmX zu|N|8*#MV;w8hnGv+HsE=brn{8W!jGybK(l4!iv()Aqs>&8;<_;4Vu8`y27Fb0v*R zuDoI=S%n}KgCim(tt5erd2)Po{nbXPYw|G-@B>y_bAOe{4MEkpH0=RaBD|937>;4( z+SwE{l#5~gC=9x@f%xjqm;S$A-35LC-nkR80k+pZUyZiB)IK70$T- z6y{x?_1Q-3jO_1ELa&Dv5Bp$n2Z)Ii0x^o7z^zzyL1Im{=TMJ<`cbWZE zipU-1YuMC{=K`u#``_z-8O));2}(d|OKmNF8a2<5dD$GxS5Q^Z8W?4N)#sp-^VwfW z!GE%7e&!kfl_IBcS-CA&SHvskL!tl7!Zgh9)C+@Ld-S(-PwGw(uOUi5_{Ig_yS5zsv zDAb?HaI)M_gkN|64Gq7a16IGY5%`~gF&sHCqam09Ogho)g)Lo6>IQk>eaO{+TVimw zs}jb*tvX?>dJgn&YitCE5wvw}MM2X*mzCwe{Y0v5G8izc$yu9GJN0XfI#pM(c zU1VK3N2jXMnn@*h(Xk`90E#<=xLj04E^Symb(E_L?`<2;N^?2CDV#I)l0S12<#V{~ z*A~#LO7Qv}VR8r`>hGw748KTVxlb?f59xw>GXF58rJEq<2RBbq{OgqRupdpecw(9# zL~4E%Ir{(R^Q25u#pRO9uaV)M!-@2|)f~=kWv|TjX6(jjE+CZGL{@Lh6-Do?9&4Ri zabD1_gUyG+d~;3rjcjER4OrmJ)JY{Wip}4^Si1H`_IZD zJ_pvi4$7TyCO$j$CzA>|Px135r=uKa(}dAa@X;@WPDC%?tz;vmGr)Phx zxu7IshPGef)|l>?489k%itLj5%k0NS!V9$n#iuJ-#Ae!B?p#>G5GlYLEa5 z@fihzNydh(%qs#HD;d~Y(66B!Oo{?cLM6c#chEPUfuXmHN7{Cxl@fsYk{)|+mBeDo zyB)`aIij3JujvTkz41UC0KJz3pXKML!U9x#)WPOvIf5(HXEBT^)Yf~f7~7FQL9?Gv z@)%ZM3UgL}N(!ToaX^|n^!578`Gq21pT2m@sTLCSAf!$N*_C3TS6V1@)lnpJzptD( zyjr3pVE=faCRVy{+H%j~BP`o;6S0i3MoRnSCnmD1D!K#T&S}}OAM&dY=KdeOaZ4+vW0svrwgW?1Ps<$Je=%)q^|AUKt@_29ZDFLtJr?(&%T^ z$Wc)+5ng-kv%|dJx<;h)qC$@3f-#NinC>$ATCB;N2_ugYQsk}M>Ki%co91z&8)wrz z#Ju4u7qUgp3ex~HC#tSySq^D+PDTL@>H<6~5c6<@*ONLjv_>)XI5&VUE!+8i8iB_2 zT+0+WXIDAdV`1KwG1Zj72V=YrowSVV3Mor8qVIG0VmZDx$Q)uiaGNVv-*zaZo0JF^ zXKAh6Q9HC-d-JT!M=)%U52s~ZcWf0xx%cjKw9wmBjd78_hT4eR*9Ih5ZP@TI?BflLSm@npn z{sr=y{S)%Sk)VvqT2U~~Ff*+&0@D(2$Sh-e4k%-_@>4F~BPt3XA;_@jpC3Aq@O-95 zUG@Ri#)<{Vwd^8-hew+&an}O~2$AK7>3r4vd}F#XRjr>kqLl~62J_TZuFG!ZDEMN| zmmVF$n4cjU4#5>!A@@g_41a=JKL~05Hkiwgv709oy5cXv=xCWn_w(rQxjMslnEH#a z6pe}Vxfh0&+-+93_@Zjg@;Ks}^J*^*+D*Hce=RY-&-_KdMN^)OmjjqV^A6-y@?*9S zp^Ia0%CAcdh31Gv)*;mw#|rju>+LQWNKmL7#Q#p7-_O0kzrFs`8~h)qVE!{PXB_%L zzL$326F`J`WfB7Fmyi3AWyqbE7`bgX?Zs^*Q@^5wUG%?3r~B8b)PJ1+M~SO9 z+Uj^UTBmMEk=>E$;`zk5cY*PAG;(9B@^^aWDpx<6=gd+&zIY8sDmlcTr%-zlG9z+X zUz1jP;MHaovWD^FJ^ge#c#QJE-UL`$~tmWYs;e@)K+M-4st5RUQA^X``M`u{nta;0s;q{pPcg$x&FM=TLo3s>PRNf5SyG`vdlQ1!5Vt0 z9cR-EkDo0d%4q0>-uNB$m>M}gy&bD%*(rh4;)Bn{sHT|*6ne&)%w5|>Z$iH~yMO7% zx2Nh})~DKQF7#!ji9-mEXB!%aBpGouTudfl;U62K&f@Lbep&yyNnE7bYrNp?3C9?v z%w8+uPHrcN<$~o0;zr*|`#IX06&F@a9E7d8rUSpuRp5juZS4gwtB1y#sRXlL9Aj|F zGEl%b{9lOTaB`~lC&M3p{){VkB*hStcNiV1GC(uY+u4CXv~$|9=6T$)u4Lk*wY*(e z8D)_olp0k5Qnt>`WhkjWUsiIb)HO%E`Q3;+l-kMughR`l`BdVg=LL zJ~){D^w;Q*;U8t+{`@+Bh^_0f(>eB}Frstk>|#ljtPF);xduan%H9(~o-`Mct)@sA z@qA6(Iv?RJ$3x!s&xixcVXu=juT0T5WAjYb4KuZ_c2r&((`NJ#yd=dK^u$_yDgFZe z<=P5{w+@F9cM51rbl=a+%*f20Kke>pc)P~YSw7vt*6><_D#Q5NPK?Xr+@4lir>u+% z$hSFK?069It&*VWu6speN}*jrf+j|eMq%{48*-%UpVS)v0z&*-(v%DKD%o$Zd{Ugn@!tlfL0CMj7mw3^Iuyq%W#Z%vC@qG+eCM^0E6bx zwk@WTiNtG>DLEd(q0m@+?%R)?!j?p`B;lNiv3J3_uI9xWS^!SJY|sw>&(O~w@Y(N+ zTK=E`|31@+BO)WCK1Gwip(@VDGYcZ?+sQYxDDeEg(-!NtYH0SA>SVZZajO7!eG?P=vP|C65{MG!D_pz2mV$a=3DNM<&;g zgM*Ye*EI$O4-#_E?#&smFFbW=b9dwb(Lh)iWxs}o_VQred7IH{84(S1n4Xugc$3uO z&7KuBntR~1DO0a+df4w#A9J4plT7GKp_Hha-DohkY3%VQlSWC7R+|6Elk+<)|K|k# zdBpd(eqE+dt8;th_1NNEAlL9A>*CS5jV$%^j&%HdQFnFz zZ{^Q=i;iV%9ffnu8VXw-4A~pPN?$*E`?lm3;oVJzp$`h%;&s6-0#fwV<11M}f2T-5u zk~*hEVD+!>#S|$1Es*-aaBr)OaZ}~uxL`?&8SMhzjJ2AyljA}DEXh!il&-7E8!8bX zrB1cuCv+Fn=Yj?7czRri4=5>fJ!u`oR-Wrl$d)~uvL#Fevy{ryOd`;)u0A`HZ1Ft& z=%r}#!01;#dZ{jby|zRwAh38yB#CaB$vdIsZP9A7c`ibx>;YxK0x4>qwbA~8_+1xd z#y7@c0B22N*?g$M)thGg^R5>r8HmTbdTpG|nQ?VdIm{UgtqXcSheOIu(nd3ZM^C?`0=~<6Z{8geWR}~Bc?Q(P(#cCe$$_D>4o`}g%Je|aIW=zF3z>BHWqWWhwa(ss^;TCjGTs2zAT5~Pr0;>1m?T~Phv3@LoFkQ&f&2Sc!-8s3OcVyA> zX0Yvz@C3?dyJkpke`z;p3AwmjYH3NFef|PzwpIEU)r|XJzAQ_q(wrlSd%WX}?6K1m z_czZP2Upm5eBy6YRr8+-E&u4)-;dt>gGT=EWevXb!Vd-LLXC0EIG+rhL}3~YA)}Tt zD8dOaOD?l&G@vu*ww4LuD_JY9X*!TdkJ#PR5_F5POsz9~EAwK0&O);*gTHB|vvNLh zouL2=+{9EmMNPCxpU%4qNgoS0?d&KNO+htj6P?+A;@%{NvnvJNAT?pl_JFhJ5d0XI z`n_?lU;q9;bCdq-k|zH>bz5LroC3FgX-HaG4gG?&6&XIrUK#MZ+97c2NYu*JR$6*6 zYFi$?aNlp(-f)S~sg-J|Vw=Q)VMt-6cBAOo%`+P*Y@?P!FE^yt zlF)k;uFw*F)w`eX4q<_nZz9?21bKSI#8?%1enAy)v3KAaeBRUmRiW4-=|fwl6ujMV)J;(QPk+@vJs#&6Tld z5lb42iKOFOZ;(M9W^nQGllxUaPHx#9;v<{f>Uy zNhT7*r%o*n5n^jXBu<0*ed|R_V8*5gxAjC&hR8|>rOoFR-o>zmNPD!MejiOhu`@`D z)jjh^qox0GJxV={l(#~8#BFcss1LJu8EKBDbT3D{WiX$QdiU7A+52woI?5&|z4st5 z;($LzVN^)3xtBt{4Pj9US}7S=U~@WYeF(O@4?R8MAJ`{6X_o4rvR)D7Y^$LVe|@d2 zyevZM_5(Y@XRi~yh0(ic_4FZ1>2#S!p-+Wy|4+)Gk00;;E2{hVwG*;~NSAm}VX`zf zyH#o(e3rcFSU?Lt!hx%nA@_@wu zPdJzi&7`i@_hhCp6Vc{Tcj6x==jQKoU#$9`(4;l+d<0BrD*IYOu_Y;}hV*U{7D?L$ z&5-4jbUfMLER@9w)vghA3$c2QSjAme$oK{ekdTFcqMI-1mmksnU>iiCZ4!EgSwZxh zEikRV;_8}(hauOMa&w6>M^22wn#!A=wQKsaehPK&Rc}=9DgEDBU1u3wiDEyB?5Y~0H;=CGi zEX?9Qd=zuS=|>%u^bT$-;7YY%%OOb{yG}XZvl^XjS#7yNgau66NLsda)!xn^N@YtT z6V&XcS?LHx-`VlEdUd9&M|(czI%K}$(>lDoln6f{uKQ^aB!4PkWP7dfrm%SGSWPVy z!S>wUl5J4Ge>yhUx)j`bA%`c4q(-3*aiBnC#Tl&$|Y2aPm2LOR(W-^nnvr|2l?0 zMO@y3m=83Gnf^np{vcQ2Z~GY}X7cDAA|-}4atz`ig+#Qi?ToSUA^yi|+$UVz6tJ#F z!@PRE5E!DIhVLA%R(W_ocn6q(|xnIh*%)9#uU3qes@dSI#p!Vh-5 zRa`y|rE>_=9x`9wxXE6!UkZl%%3qKjb&vqjSSU z_xO4VzV0!tC#8A)-cQZxlu~TJgBgm<)Z{)MIt$HF4{H>K_`8fw#-el*Fp^dC)S}*m zoNH!1I(u<~TMaO0%~I&bJTf?QeVewU6>iGa1kU2M)!}t1CgL5uVwo8HIhfD@8$= z3f&CML&Y_O?P2_ZWyNvNaF5Q*F;OPyMDLz4aw;Tf$MnvLcvyDy4gHNQ$f8u?uGrd1 z&q$AsW?v1X%#cnW7%a0Jhb$CLp`=bM+0au|P0-cY%%Ei{W{eS-dK~urj0#h(Ufdw4DWOn76kixPD zUMZihzJm;wC!=(y8&@fsF@zX?15r8$E9m?{%;^ltUSj_-a&>_!Z6Z2ort4z!VR1DS zgZDb^h$a&w^s`Y)O>;jHfykW&)piJm(mTPzqo^Pr^1EiPC0YZRU zrRmHG3l?B9fvtiyb-N=fWw(QVNJ0Ny1A{_3-9=MmCPAqyg9|e~BbrIO)VyLX^U@-& zt6-7SIpyh~cZRlBsqLE5jK%x4i`!09f30p>uMAcsCi7wl zijCthxRTV3LcN6-`pjkJM1_}A!sjC%*|e44kSej#{;Mqdt@4yOgnVaSL(XK|1NuwJ8({8%SJMy&|iBu3Yh5N`c^4&QL7i6*&z>6P04{TCqSw zFBYi#x`jB!2R%icIk0GtX?1|YJj_b(E9`4$0lc}VN-^2t?VLRCQoAAbtA^E}<0TSB zZAxOkwHA}K7ylt=DfP^bg<$(bPny?yaa35h?}Sm0-gIfgtQmJ8dh_F$1E7I2F0F6i z6VZEW+y^J%T{P#0xpF5wA5PombIsF_In|3bw-9*|V~_FFJbe>`vN$C_SvPQ&=pIeX zcWce-z!WN#cw)-MK~wiZ(($*J;96GlWb*@)NmJ( zSB+HZ`-TNHoh{P3TW%kzl1{tc`qUow*wrXJXeUe2xZ*kVP16N*0Rc3JZaA{&S&ccy zZ})sxzV3-aGSleI($Yj0$nqobmyrT5V*y2jba~7i9ZYaoI?3l5p9%w<=W)`=jNYzX zFLYs8%e7s9-4&EKf^>_R3h9iKrDFx3=f@FL!JzjoBM0LcO$^HFI8lXsL*Acv)%2Tp z2vn%HHn@E3(|pu(8k-}7DbzkELN)@6HCL1xU_ShSIIRK;@Oy^X!yNrieeI@J`78S{ z3sgZUDOKd4sZ-dg#a)bQu>?dZVL*SN6zyH;>xTu-Hu@g->OznEXufV!d8Z98yS=*3 z81;!G;*C^Br$Rjc&=B`*f1JWa^WxxUX!N;T`=jG-d?}IBspUFU32X6t`HiQsr)oik z$jqU$E@D}@VwJC@6pLV9kgebqw9E;BF=_$PX# zY~?W+*^ufr1QpiVfzQXK_HCH8g%abW6%E*&*xKxHBRFIZB;{GAV3$}zPEGaKINu)H zW^H43e>zG<$c_7?ex#uZ)7(!uF?195O`4eP;KgU8y`CkCvH-EFt)jRusLPE7-pYZ; zOPlqV_so_jgFwj?_c_|HyKGXK*$!u11l;{IUc2Q9b7iX4EJB&T(W059U=yg zS>wpbL{>oO=<<+rU{ft0iyj1|!@d}HK(<2IkI{OmNNpIV8oX?N9u|N?@#zXSzW0cn zspx+33aG(wQozP1J%Tu<>K`VUjwA-X1noWq*bz9OHlsqozBDm#^aN=ah~`64%HFJFoLQWJzLW}UrY1S%`Tq$xVZh5!baY_p5$%zJb2dnRsO@<=RRB& zu={G0+3rs*bK#;GeSr2bVnY9GfbZ$RNmk^kUe?J{P-e}drR7ZZy_`gD&L}wUN(fKA zVMuQ>*a^PRu|Sad9gL;bMZ!g8hu2~8qjO>PBwu)PWYK4P8%D0H=!1q@%u6NZ6>#m+ zn_BJ*-tvbznKbXv^&IFrEuPvXSemAx%xA`7cXBeXpQ<7Q-@t_2WC+>p3`vN`q-*mm zEb>Tn9M>PK&gnVza$W-96 zG1QS@Vsf%F7Mn4@xT#tm!krXB4N#wJAT7chEl$pS zAS-iQ*3>FAR4PJIX*^rTnz@!O>a1b@XGtlG;!D(b>CHwTuPkOql}-l;tw#q(H9hmZ z;5zmhgY<*5GxewV}nAj#->{?upgKm1;4quw3ZIqrleg| zvQXDq0e$1_OK$P-mnTfom53>aHdW|;*BnMNQ2IO0iV(9tBD}^v7b(XC(@5C#*a);@c9@*pNz>jGqx&l?h9f;2$AWjZQ$W#&dN5?L-3k0?qN?TD=lhcv?s z96I&)Yq;Q0^x`w`lj|h!T*TSXn>3RM^r8j%5x<>^;=0lNA>%<%+lEx2-jVnF#OAC#EDlPbQLrQmL$ukNSCK9`lv&! z`JjTwfwL!QHI$3E|InTGL9Z8Sxy_vMCWgw$F?P|1hSuL^MLkc@>Sk#ueaSGz61bd5 z%TDHJUIE6CxP5}R;9$L6_Q~eqLkPz;>_Oaf)^B4T48ry&pnFWRHOxve^G@fAP43mG zC-gOcgPw0bv~eIgZj3pKE{4225xI0^BM-%#vB)`Te5I%%80PxOaell@o=_@i!{iPp zX1#!9S{+b1*rO4#k^upMqLcw>w)U1-ptK@;WO$x}{r z7k;VLds+4U)fV-D+6heSaXr{-f8%_~{P=Tl;E529UY3M%jFqA@^QRaYI{kOBIWF$g z*HARskck8(tJe>F;y*5N;}lR#naB553kB-?=MZ7S=y288y7a3bzu}Dc zsx^70>FJE0X7-C6d)#`_IgVinU+t77HIMr=mzdk7fP$G(g*LI-JkkzbdyMm}4Y9kf z`wqj@n}c#59!5(=VpJC@UWCucD=pR2p|uLY(wMPV0^^;Whr~&|gfMo zPi@4hTXt~ZDn83moH3|1Er?*V8#@P!{W3hwbDVLa#M?rk19`a=g9~2hMJfVkHR$j} z7xa@$uY#UJI;JZgb5DPu72`zs*;~^MHfL*PEGS=dXG9!KN2W!_nU3hR>>^Z&;U*hLq*RcsYKs#~B5IA6kl%=vJw`wY6WBgAk~dlD z?)sdPdcbCX)!7myP5j1idK}p}!`Zjzd*-f-ApMD02JDL+$+FfP@gpQ}x#3CkBim{o zr%4GET_1^$P0Fer6yNZa6QS>&E{F=!fozPQ!Gtqz;j`rVR6T&LA7X+0qzFg{itvVW z40J2h1W|vxwD-<;6L*?9A8PY>N;=>Xwu_+n28}VIGx^6QLjim5dQPjnHL(QUIm(Z%r6s%Yp>2WH3GH{OYikfAG$ zHpaa_RZ%HJB0dapS|NOQB3R8Q0^KfEKIMcoGC+EAidv$Hnd`3I1~j>X>C7#~q&80K z58Yk{ICuytW)uyENwHeHG_brhvj+n)U-KRe@M(i}1vI{u!tc3{dtJf{8QWggbM>IL zxajn(T8CftuBMF$4JE~_QFmWAbBU|`)yio08NbETS)8&1ItvGa_6v_Hn&PMInkJAFm9pWs^eTJ3lr@oZn(JG;bpb=t z3U!)LBTP5G?dl0G*~7y1e1gbONhB2d;bcbkp(1lDqoSI%CwJS-Z<0!=|{K{VV7E&U^6WL^Hj$$kCVt=ZFubwL8n zNt>D|S1d5$D7Bk3auk2yW<0{c+#7UK9dta6tcwNDB#3~`ufsMYK(SJHtPs#V*;540 zZaX_jXW05A55&2i#u^mXc1cG){}Cnazyj;*+!z@74ip^jgGlhwzDJ<&fE_C|nhn*e zbz#;|#E;<65|@Wa*cVR>o+~NlgY$Jq*HzG8MRHeDL&>Oybo;m(ll^WC9#7Z>r9pfM zs@_yvH)suJf|6jjiiu89VtGjjrnO72H7RYWCs`aiarJ6Aq%@ZkNia__&E39M zS5iHwn5AX=myn^G;VwjMd_ZeiScD+KgjQE9ji{=}D2V&kEAv zT^8aC#(94Ous+}XDA`5Ul#ZzEi;IP-wo?J(QV+!2` zg?RAT3FqZw@+LBHAjmXp!7z5wzU88sin4rc)YYV3&Z#MI#Ko$#lDZ z1-w?cNk9M@6-8)su$u@*Sql?eGT2hOZ@kf{2K)+GI3(d$n!oU#u0kRXYeK}QSVPCh zD(B^WdB5RBRn+>p4v-SoP+T9?l{)hd#kud-qCJ!jGP$Yc0qV5A@({s))8D`gKG~NE zu)n$h7B!(6c;#LJ`lNuP9qW;3u={ZN?fa4h4y;XwpP-`QiZjv7X|CJCRZZULpP<6Z zAh8QN#HhJ*mU!8cWAnwPF>mT!QAmt1|#^L%brLwWcA$q3uQAD-6c)Cy4 zNCQz#1h_RZ@%roJb6Jq|j9;JgkM~=9%B%Cci3}GONgkz&HTGeID{U&0#`+o*)yoi5xexW)H4|2>!dw+Vy)^`7j1 zopYG(p8@1vltGWv5t|RG?;?i5VrD+0OIY4bdZkRzw3f`}6*tD}{d}<;$ zn<*Qb#J2Qv<4tPaQV6R%7CAS@qjYx}4+;1lwa~-9BooaCcW*l0UtF~-0hZAI4k$#f zQr-8|GEcfvXHCuLSnxB&3)B6k7ScY#jk8+>OWPB*TYU=aotY&+6OFIbTmNk-`EN_h zzuH$nqY$J}oHI^J(a_^(5*KKpXgMqTr2G)&mR1mQdtL)5L%=|UmZ&t6boe-w2gThm zb#5ed&8=^cd3oK=iAZ{B5IPhv8O+It&)VDN{>^1j5h2Ck*Dy)ba?h&ty^ONiS<$(x!+7Hk6tzYq_k)!5BRE6JU0 zGiNZ8Jf+QIBuW-F8FA@8^OYkL@|c%$PC(6$&MGKXjgn~ydlJFQu7UeNnZ^!u%aS)F zOkpCZ=EE#QOU-C)zVl7^P!Ra`Ep*}<=1k?LlSb31v6zgpdY(c}NTUMd~ZgV2gnY-zMHi#M@uESPw&MtYx8yvX!dTGKu2 zc#&xz?ruvxYkR8QrgKkmT(#>9FbT~J)vM%qd>oWVTRqw9!Uqpxw6c0`)sQ11oNa*# zJFKnQ?%6iEsNs{ku49~dw-R!k8=JDXJHOEFGNu0~qpie#!_k-Yr}^Q41po$!|Nm_n!sJC2XD7m&o$b4O_ofhrz}wO3PbWBPw;ZSJ}!L!&m!ki4;>^2b&OO*+Mhsc1L<$R_&b^8o+9jH&mT=d=gw9T-x&&v@dXE4h0MOIsz%Nnh zd%%mn9P%MaPp$g%_QctLI6VB@v;W^Zqg`n!<%zJGm049bxZ7^Ocb$)(QeeQw@E*4x z<=!Xi?jCYfRBlzu&}5$KZB4BECqfD&%J?kt$6|*;mSQKXkb{$-ARvYDs~vllKaT!& zY5q4?gS;_oiQ>`1>$+qi&hOF~wP`#9sH)OlQPPLG5$;(!vkI2FI<%VRCwgWvyEZuu zeskJ-;GzedZCz-4F=6IOmJ@<+Dzl!M2R#c6e)(RU1oG<#^~Y2D=d6aAM`Ej>Vp?Xm zqm8wy@+OjJxCbl3woPU6p0k24Sl(M^ReL%K?2OG%{fA=wt5x{FXb7-bG!6&iYzw!Z zy|mU83e_>AG^J?9ygt$e;EKOeboyi5ll5cVJMl5`RbkZpG= z(P?-0Wo%z(tn6ws=v2_3tk))KBM5CJ7VL(`Nrz}^!ijLoPdzfH*KT3OGuB7Mq>}9> z;zz>M@9}6r^Dv{anuB7#ls<9f+fDB}dFkpG=_PT}L`vqvv)pX%od|b@pwq9usq)!c zgSlt}zC{MrANz@0-Vq#rg4^Yt(_y^okgA)qs4@KtD-sY*xz-F0VB?`VJa}Evlwipb zk@{n=FJ_e@d*jlt(_q<8@IhmR_vCkQKItB=P&55QNwuN)ZIlKj1IMLO932n)<(z(K zQnFN>q>e?ZCq}n$kZGd1h-|wT*E0>)P7r3SB_ct6Rl>22%tl0sv4#~bCI9m(TbxG3K}ZL>R&7#j2}>H1Gvl7r^ixa2l0cq z=y{{{bXk^g{i88DSc7+P?DF~k{u74Rv5LDJe*B%U(Lq}p@c`lR8#?NMXi2l)Fo%5=R@ zKAc*;qtKZe%meqpc$rViJ1}JFa&>xL=iIm~>E^mX0?WP#Lw;(Fv#&XzsB8c!fniCO%H+*K_y7+zpQ`f|tX`RH0h zP30rolk7_0$R5$-pk18Tk&b$utqP)(a-xji$^wlm1GHo4BD|RJazO4be>0TsPto!t zQ1_yWjwsEL$q_}E_UxQqtECy1u-oxlt`eNCHs^|qPCexZH}W$IFK|SAw#h!b6Q0mS z3$Bvm#mthXR;L9f9O+)FuFs=R@SBY7594_;hiZdZ?AzhxmW>2s9470oA=wyrTi`Vy zPV&VQCrwDKf=&DcSyog>dbo35#e93y#ZM>NhNpAh2*~Sxrpt}KvqI`AxhLKnov3#> zQgVUql6GRV!{wV2Mt?5rZ}O~|gxz93{u0~HX$2(~omB=TWrrV<***|GEuwxk$_Dkn z%P{YnQpfr16}JUEoy=ZVhZo2Fw9=%Aef!N1D6>=JQ*u2LUrgkBxHOpmIM=>1gH_8u z#?eC>ov$e+>$nJA(kH#x2)a>;;K09pgBrf3l+T(KW*>EWBf{A!JH87whSisiK@$Xw z!)Z|W@#3PEXlj*F+?BxwU;{LTilaW-=yrw<6r0FG-_9JZi8KCFee4^xFZ;U9$1ywa zn{Sl}BbK1p%ahH|tqvT{FD{gY%LlSp}ul*NtK$dBQ9v;?d)4_V9jSF@5(~V=P z<_8Kjz-z0Mq$2+uJ>4wU0aQX|c(@ScBX&4(Jh0s}vs-ok0S(i%&(qPb`D`9ce%4(m zN>9A86cv2BCp)XR&~IwtJwOsbIFEJ#3b-6XF*ROJ!`X&GE%J7}T*pT?p0zkd&bls= zj--0n8hP$Nk;w?R^Q1Q_%o>Xusg4Rs7ELi;v^pp*hVGTPlvA{4M|a?5GVcsU`h|yU z-q$Q$3M)NKp;*n^jUIq;fo*&AjL2EM5hXf(lpmyUwCOTbk1)g+O)YRkd~HjJUNk9V z=`@WMyA(;)j@h&Au}ucB9wVz*fxo<$*Ryl7;p``vL0s2izd+LYd*s!XP2 zsj`om^Uvr{2_%HIo=FQo%Kp}s5i2vr`Ar0sF24laX+TmrSy8jkgSJm!RZ*EN`CV&< zAlCMYe)G*#C+JAa2jeG*mgqfD?74)TB8Zc@^F7mn_c}XszC@NYkfndioWCu*C;Mc) zUNdN0XC-o$;e_QII`LRs2TS77G~XehcV3=YPlm(--yn>C5&JF-H)N;FztzqE*&td4D~l#~S!9ofPJqud|f^ zb%nS;fY$nMSwz!D6$on)F4wV;L5MD8UFs~2C$RRN5g*up4-~cAh2SVQY{AEfYNOhH zX?rzaB@Bp&%I+?GHs`CJMVp})nEqU{h=!YXRT6|73mxL)*~~VK3(k&s&agNWE)YExXgiWU8Q*h zp5x7~auSvys8?k{ghpmjQ@vG2F@zLoZ)rxj=FfxS;L?XWtEgE=-2GhIqdY~up5P?l9VrP;CWz#A~vP%VSu zQ*P<;N_io|cet|yW&77=Cz?+=$y5{R*(r*r|@BbSg@-zQ4zB@2XA z6$G>2*={K$Y(C`8mF_`Xy-sdBBn9(@)GIk>F=HX>EZ+!(m@s~3e`0ym!4)Zu93W0H z_cyq-KUFy-+lBy5Qm>(--bS5HYq`jreUj?`Lt%eRP=&t@-(@qe7}*kCm6 zLz_I_hvnbX4aoL`yxvH-dfJ+WT&Q)Tchj~7Ng+9xHLL6^S$}mYkp|M;p-pXA`10hf zc~aLY&3kqGuW*)u3VE|7Qugm@7xScItK{*T>X9=H@RUT0x2L8!PDab(IF@uPZ!IjO zxDvchjHmNz2P->-^f?{g%>X0>;r&)}b0mBOCmgRN3e%BrlziKt5IsAAQgM-g?xNvH zoWh`>#7dA{mc6I|uATLjMX#@HUvye{{EN_x{)=(;;u7N;SoE7GfvrWpsEd<`3?r=X z7}v}x{p6E5?kx-m`)i57WxS6#O>ebCNjeH4~$QZT6xZ@){fR?y+rFDWGt zuY~)~#Zo^U?1b5b|$u_znhkZf z%fKa-9B?~!zFB+;CMv-f@ejPWZgVh_gOMC>ZJwx)aXF_?9f^{1I?j)}GUDlDW}N+g z8B_>zO}2yP+3+@5jdJXj8Vv~ddTL4z=1T0SSl8Z1WfW3?HNKZU@*S-Bv%u=LR0hag z!pvEXsT;XllmnK|b&3S^dD8sSJFW(YjT7epvM**u+Mej*Bs+^Qc zERjqmw&%~M*=-raJm^VeO&oe!$M9M*xA);aNFsA>nuq=fv0A9N`xIA=uC7I zx0>7)>cf)9(AtjYk3UhLRAGOfi~MTAedjvaMkz3Qs@i%ZCZked82e83HbnVe0Me|C zVk;?)d^p=2H`5c_N*g3gXVlYtN*$S*{k!%B{lrigtS4bFGKQz%AlFJB`@jlyH4r8C zUA#$sui;^TNzzIw=JtU|`+Syr(Ko9NpDV{JRXkGcu&kL?f)UgP$RY-ZAT|0-RIi_xM^ef^I+X>WOw{d z{8_O?^g58)Tm1Li4rsT`+N-jrI&`is#_?|iki7!*zNQ`VU#1X1ycS-Tb5bdZ+Mz7t zDczlkQ()&6ZPl{D;I8t&FgG>FE^CD2Z=#7J&_8;RCLNc z3Q(6pN3LXbU#w z&XDF)obd;3xK;Xw>M}IqrI}ZRqDbVtENkJ(9++c;f7 z-O@Gl3co18I=v{D73uFElao~&R&+itTtpZRYd=KXJLJu_ZN++CQWE)JCz zY8?I0L1qRnY8y$`&h&AjRQ`apGf1&B0mUjS4=lbdEhl)$!9vCx)A7862?1;E`#dxz z^|{MYk|sr~qqsO*)a#gj^>Jaz%uQR;?b^zSkFpmJ`gTa{(;2?q0^mG+Lt>9$ec^WU zkGWu~bCdNKq9tvRX6w7C54&uYwrc8i+6N^EYqMsl)UQ?i5IbrHdjzi6{dgCvgSL@V z=PlQRBx>j*0$6^L>~antjio)_ySMh0{etk_72JN8klWM6uajJhFA1TrF!NCY_OQfb z9{xsCk5@ zlscV=%;V0Hxc^CLl#acgLG0FXRNUSswWTa$JkJ-%NLF>$;XNT0<<1+wXhYWTq#8S; zO`mAD5v_sn_GVJ(@IZ6qOxC_q#}$m`+OMl5u&264sZC=*7iB5%*g&a<8)0Es@;Bry z|E(nW0ITHh5W)QQGcB@ANVn>d;hnb*WRpWbUM)cGd(}Am8B@lLr(1qpcXi4V7c6|y zlrcRW8?uZXB^qj<`Jzpsc?^`ky)E56ulQtJ}~=M&m|FWrv(j!N~4 z(vnbbpGis{Zl510AF`*2tPTo)c9BgxkOP`%IwzSC`lZEMl`!=5fk<%cPK zym`cleN~#5JB@1PVt(}Zgx|S~Xo{;!XnukeEg#e*I{pM{lDW-VP;)03{iv+;Vix?phLBEY9=kMfPW+fPA+t=Nr|fyI_pY4oa>XF7HKC;SRhfgU z)zHeoyUaPI?fF?PS3Rd&wnKle&%5j^C**agXF@p-Lc94NnfUS4t5~gjuxOQ6g=0?1 zlRKsZ0iYi!--j2Rld-kj(Lyip(C**M)#K!K zJ0N>!^UyG;kvMr-+UDs52~}V0F#8(V{p|(=^$Y~l#ECc#~l0iSsQLnx#9sM zf^s>p(xSE4FDa*#H$H|1ZNoFvPqW}M&(}J<`dV)eW`^HYj%>q;zLZ#wSdj*Pi~S*9 zIFwak3i5xb?vH|$V^rFTgU-MUH^g)S^d=OjwNa#FH&!noLb{~e7tYBNe}cs2b~Row zsSIIaJ8Q7I6tQv2*RU=f!;hUE1!fhmP(i6mij&JKy`aBs6 zxMQ|>V{Ev#dxH1keni(LFs`b~fLs6yIna2e4I9quFDheCp6j+oaf%04t-g;137egj zXmwsQ1TzM?QsH{?p2OoS9xl(Q1-^J)lC|0zDI<(Y9+14d&$d(xi=AJlPasjx>ez`l z^70o7TZ1-|UK&iSKh^B4BlvNU4T#sn8;f z>`7xhnAhj1CHmx>FN#uyT2hHCSY}0W_fvDJyTHSh4OZr`(`>P4Uh z;({(A#7w_^lti#yF4@k?udg zJZba|MHeI8kwwprMSLHtB+mTBiZ=?%P|~Ha>OkgV(uw{dh@XN}hWti-vdY1lmup&| z#@lnna#gwA{pn*mYJdLIbZq8!rKtKB$;IhuSX;4a-#WYu6#2QR)D1pvJ6R95Wzr6v z&cK#g^bJO=z~A$u5tU%vWw)${AD$OGYR}VKKP#-w83K7BtZT{N#hAE2b`xQbrmhTR zgUmS}zBWZJqrU^9IOjA1dX>Wi^^Xn<4s=0J_IC%;f#Q z(nDQkHmYZ7}2#j4F{Q8c8=>}uC*k@{{sB`^4W9%IS<0rnRa41vyCE|g~(WZd4J4zjEJ$r=Xt{DXBEehKxW4fQqcNav1P|2MwNz} zAj~yi3yr;455#yfYd`jE+4v#xpubB5Z87T zy)n>iQdebg;g2T8>sy!LGD;8sr>ALgiVgF&NkoVVFoc@age(?qmY*Md%&oOSR%^b+~{*C zrCy)E34$YY%F_v5|EJa@WbC@P4COp}Y4IQ`>yTk~zsu!Pk}y z-m0yvsjhI-v}}u_lo22=Kq+l7&TADz@7Uh4@*$C-yJvt&{{&epBO5$1wm-sendGD> zdtM8h^|H5YQ90x1_SpJRTl&G4VBKQE^b_??CEiD>^p%s6IIUssJ#Qr&EX8pM@)Ov}D-8q$LOE1)GgrrCOh}+-M+M@qp+Fc?R ztlYUGA_AbUIyC#fgwpiaYV|Bj+TJKP-`Dz(lcXEUzJYGXGDL)gt~t_u$=De0U1 zZ`3Oiz@O=!chb8%ztE8FrlLbvplT5$3V6ria>*X+kmLO1_BfofiKT`r2c*NN%NSxY z#%;9%Yp~%1=CD@>J0M5&xP5D({NSj!@d3IteCxbSfg^wA5@PJ{3RC-3m2_bzdA~{h zaAt-F2KC^2HCg)Td><{VL2`;6YDbAC_Y>su9arXDps1GNeXfUF7u?FN2>EsU>^|$r z^XyS5y77nY3$ToUZC~w)%UX@$Sb7j1i&e+v^ z6CQ3~dMmgln>y#u38Y@{)%F83sr@!ljGopW_*dXOGHmfJ)RkxJ!~47vD^!!goV^+q z6UY#?n1UhEL|p4bEX%K26MOF|g<}{$8#Ur5ut%jZS}(g|k1tyamhg2pc1Sp{nv|iJ zp=LsTBek+lzg9LwC>CS{8@{L6@nVQYv|_O$?46OmFM+$6pyzMaeL||xT!$^r5gXRY z_x5_40*ejPgyA69L27kS>AlIv8}M7N4TImpC*>=&>D(6R?AVX_E|NaScFrb)+J>8! z?N{#XHH3k5Tk@Nly}E-Q>1B0DvyEw+?j0=X)YE?drA+gOAmp|6=)?T18iwL6FNNHw zx98`U8uUr)PJPInVOP!U}eFENudpY%YaS>Qeiut@Yf~I}uod)B1+|Oqv=s8O^ zS5?Eg!u_9FB1K*5L!3Pq71#w%d%tk|PtZroCZT?{z3Hupiz>n4nI5+l{V^#w3f*PR zJcin?;I)1EXsgmpLLYZIp8Th|Y=29yy5eI6^!!i-i0v{%PUR=n@(o&Zw_tpeSXE>m z$y%a4oRE0V3%T!Oc5h4AGftj-PaJ^Rw!QTs97IV)+ezlpXP?v=WgIQlNQFCANoulG zTkv;XD?jA(E6%Mv%xP#JSw2y=XsjO8KTestML;?%yl0!Rr>_ZUB_;Uc>b=HBzY=4C zBK7I!TxIr-rC0+sLmiZEgfOMWRRD|;8h{ol73xpxkX^sG&C=tC*JtUNgTWDmDXK`K8I%mzx0i0-p!TX-5zD3R8pi>Nctj(jTnR1yc zP9>q^lb+e~vNMJP8%m_Z{RW_m&DxsCS>zyr_Qa$xTrq9(TO*;aWh8FMDJv;+D+e(S`q+Uvs z>Vgv}_zIuXZS$dzP1KWR?!bzCT``Xg3OylK6*aMFBx1F4crj;1*P@|$mZTnKqimY< zp!N{tGIJ4~&#~HRs?Or%C4WE8GW1~GbNTpQ*hohb2@Ovl4LgX>WRYlzoeogdtZ=n|mrM876g$K|$tOmKM#J=ZFu6 zu>u}9KbF9IxwE(P`Vz8%6I_b|=krW6juNU}v8T#qF1qD9ESlx|g7yLP-c$!VmO1@J zpgV+jprI_vdPhS87eJr@W2BoA{fK^#mSJ+f= zoXbuu(^S82kO0LjkQ@Lpc<>SO@ntI*f;J`7ojbEs`Q;P;;xQZ0W$W4Wamd%E^cj45 zD;MukrxhmHr^0{`vN}K2<8=C1sA&`;j!=tO)42Q5GLc$yKc>xNOOC~^>=8}mVbba3XAg0u1HJ}~HaZ{ZT?AH|4>h7s zZ#yF@G<+16^ZM3(+_`cbHd1|QwPf=p>4^{GpNwqh-(ocf4KHe92oTo!lb;kP^@ zqAVDnA;Tsfb0#k<-@QN&$8LV#smO$w_rWD$SePYms<&^4=xXL>1lG54f6Y*KtCp{C zt`leUBc_+}EDoHpepsyj=tz)~JE=nEUN7*|)uB%#iHJ?Vx9 zhwli00-E*GuS!d&;QT=tG|%Eppi(OS{Jpjz7GuOc*=7@eAMd?-l9*$3bn~@Q*z~Zl zQ0}E0n(-)YGIuW--A= z6Zu6t`U2(@6`F#3G;Q@5`-YSVItdtMD-3Fq*Evt4psB1BblR;nWh%nKNKO(}$RT+q zz%$vMJ=L966>A&+mJ}Z$h*KH0t7903?6{>j2Eh=f0jME0`9ev>K$L!-pAe|p zL9v&1J@B=#tx3W73cmiY2x#w|yb;|3&I+6GxE1gf54kTA5;Uva6NIP~6XtoOWdpFn zvYoqED0aRJiUe3LOh1}oouJf;HU6B0#PHl-JN8}ndKs%I#ihMsqyN5;?3%avoLwTH zQ$IP%INsFzDnAF9P^KMS!d42y%cztE)ru>H&)lR_FSDj zv%K}{{>liZ>10!Dvva&@1NH?-Icg{e+MushXx)UNxqs6e~rngfz=) zjMhVg5QTJ?(Wx7!G{#n`FF1eSW5^3%LXK%f?Go*b5ht=Nj=Gesh^-6kX0~rCZ$+JX z=(A$EG8uk)k?(3qY~>}D&Al~o4K8JRZCX*5yjb{1oUvCeQesD(tU^;WVp)X6fD0)0 za}VJ%_*SH_@ zWI{O!8%ojAPJVc6Wh6)ohd^@?Q;FS0ryyJWT6@#A>)jaqch&e;qf|m$4Fqt?}uUdI)*!H?=kuS!ZOfmjvYMxEFd%zMyK{r*X^exN2{UdM_VXQFsnh+szgx0K3}fNdhrABb3X}gT_*|0`>>zD0 zhL9r(5Uof9-tOSS{YwZyhw(^+@&4UU&@mZ6`2w;Y{tq#_Grq^o%P?6ttQX7kL{Hh~ zs?xZ>(U9YsX;81!khFtbPBnceV0k*03-D3B_a-X(W*ThOyu5oLv$z<)hloarWhh0gVa^Fx;%C^KwKWRS9S;1nFAE#1C;z29PiZFAj_?En4UF9Gf zAw$w~TR0!JOu=QK)7awBPmjzx)QjIvZ+1ExiG+C*1{%vOKTEOhl_(>2?`Mk0*8cPo zqi1*50i=NpROP|)M{KF3F1Cx6YNe4yUUj14j!}X5!u>JKDU*vdiS)&S3vK?B&|OeE7CH-^ z%n@Z8RDq`+ah)t*+={JaZyrrUNeaIB2q%BJa`V?*R)5~EDO#ZNa8_(c9@xKHNhM<1 zv|R<#$O}9Gcd22Z*e3nIQJ3m6Yfw$Uxp{hS0hV+EAeVT;T&~W9^~C$fJxh3!^-1Q8 zAYb^l^&2$lWle%CrhWe~rULrE27gJcp_odn!Kh(6QRxJrYv#zPVr`pibs&T703(VE z&-TMhb#TI=-YCo4`bEI4e-ZFk^d2AoEZ~FH+z57m@{F0Yz2b{&v#+gX6vpuAzcY;f zD5FvIf}o{i{27j_QF?}pvd?b#>fi(l2nG+%@*=T`FUt|&@}x3Z<5;{`O()!%2R|96&-kEjcF#hlV_zn zDXlk%$;8SmzU(V7_wB2ii|@NuJ)OX;*Ky)4=CJC(j09RwvKT))x(uL)5}5k;cSQfc z{`o&?#>RjB#(rre{>@Jk>#%r6P7Ix@W1`VRt23`l7pGKR6NBg2@j%NfE@Xk{LR)rux8|9V-Y5S>fGP_ zAJXT6iMRJa#I`UF^ih=F1pxFaAE+++sip+AZZ9Q%f_x?=kMy%HnvPi(j=3d}&B((} zL~P`0>W}Wr^SAsUHdYC4i9`+c18;Ef&Rx5%7tc%OLCk&+L5t`e<-_}Ix5fqA`$uHn zMV2&@$omg4Hs~IA_C4=+B{I^no`4@jR$$2u#bI8H^n$z!PUs%U>`HWr$weuRYyHr* z4|_{bZlx+RUI_7e`h7F&{-KWbTyOxxs3>iJsqU;4cMWv4zkt!jxsJwMCEV85FV$-! z(h7hd_`nxJPgCT`LY6B-0>^0Iu6qoBh>L2(4Y&l0_Jct0)~qE5>>JMegGr8G6L zaB-#GY$i_wQV?4J0iK|T*y-E&jAaqp6)k;MaJ~Gwf)rWa2^z^OMrtEU&*C|G`6B|# zXNkJQ=_C8J$=z&89PHVlpNh)OY$LvlT7x*oB#-2!VnA4;Uz}p`ZsnjPf10DHdRUc5 z^zraSd2Hh+NXy3A_bFv{@@t&7BV8}Wk43%9M3b&pZ`=Fa*5Pn?SGn|~GR!`GKi0PQ zQ=kXtQ4bAzm+JJfx8JK8w=W__nV=I2rlpy{D*Z-STnEteypLVR^QCzBZd_vN46V^<3j^&}1RxRZ7o)yn{UyMj? z8Czw%3d~Qo19cf{xJf#yvpE^P(BcW*3LhBJ<4B}vcBL8HLor%4T3oRl@JvSmF0M*h zb{ld9pN*;7ymYm7T`gAR{iZ2}ahz74jf>)ywRq*DP_lKdJH8Jicc_p0!oP=alUdHx z>$})V(zP-*5gC`WeIZH{r$4b$0gJb7?6tMAVUfd=e`s#OTtol9`7uai5Pgel(}1*M z)Xsn}nc^myR_<$!9VY^yOp`~S9JzCp7&KflxFL1 z?U>kG|0bXuQSZLpn0i*p_@R`0kFIq^Iu(rvivTyxs;AR|#H|Yr6hEl!;2;Uo9w~LT z6>8f{--hkvXHNue!*{y1zuFoVhZC=fdCE{@pGmhIUw+Xa-K{K|v*Ta#q~djiGfYaF zt0dgCL~_M+mO&{P<;VAZUNY1{JFV=QXM7Xj6BC-BXcI8|1~i;xk^G*4ic$$jEQW4* zta4}T?YV_6xL?H!IL(?N0n-NAmeX)$v#PTAkIoYT@EMd|5S)!fjrqYyyGwQ}TH4@& zP@7G@HGIQ=f6tZS`^`(C!Vn^ba3gcZ)OyD-jl+X5)?2Xx54Vm5>1X<$H?Ab?@XO0ajueV_e=Z=ThO%*_!%hy-gnTC=pXr^*IV z8>$4iH`-WcRtIgqqBEDC<^5ht@<54>oHXqo?~3;5VW_*^(MJ)%0kCSZ;z^y6o?~Q) zbH9RZwh~S_$C;Z)e?xorGg^Q5hf+?mk2^$-=K+^2^UqVnn2GDEB6=FyUoVb-S37$D4giWY z`KM6y;qy-AMoeKYY2#}&{pCgqY)^0GWig|{on$fo>ZW0UoAl?z3(^xMUdAYTJJlKr zDVw0-1D=D2o8PRo8cC}g%gdD5>|MG|oi+uZhlYK7>x$$qP_nMPhhJ>>=tVXQVHZkS zr`OttRFO!0Yg-qz)efSYJQr& z!02hGP-3*BAGTvyaEp7CBX3^O&S<+WH0ja6_Sn*+si&g4>`MK0jz-fPZYXj|^>m)O zn_lF)>W72A*{?^VD&Npmvp<$WQ)Dnrua`)B)5kLr{do@B{RvbR^^}(A8iAF^S3v^M zTF`pEtwn)>m8Rh|Z-KKbc82{kY2S~5Z)26x4gA@y5A3Ucn)Il@WFr zWIW%4WT1O{7FNAnE6(KvdJfef?2Tkl(c+kn4~~01()z4H64$Lx^6T}o(_GFiLoA|~ z)xP__n$t(RAG0pgWUHD+Tt1(6zqQLrkYO)jbn!N(VQucf4Q?@w`s_X4yMUqe@MA`6 zwmk6+x0Dp@M)8FlBEcII1Ul+eaek2q5b**>+Z;`z|Mi<5STznF)j)0XdPP5%R^1WR z`FbT`PSMM^6B83v<=itbpP8FG6?MbYLHuVW34VVWV*RwYXzQEYutGjun)WNFa1ZLW zzg5~yvh>}9w~#{`v5ulS5AchIn%MOXxyl49Z2dV-?J=>#z)PEZzbeSZeucpxwD&@| zci~r;MD6v{-TPOGUE-##s6&=84kzyr0P6RHIjX;22M{_(0c@9FZ`I33_t|aFzF_p~ z02G0)4l#1YPNF+TRElB<29Aa==RAm86M5srYc(p^XbB%leKwOmVQ-&4u(W%{x-y z5L?>aO)4rLD2y~f@11swcr>cZNzTYrG&8Q^f)WPqykMOU_eI?O=Gj)=w|r8n^to^0N@ z$CDR9i0aMSsrnXhH^sq=(TEt*-u zR?qtV5^=KCD~^ua%?ha-O^hY~u?sy?v^e#RrbwE9rS?5}NuaB{m1s6hYPG8+oQq1R zAHG7;>LTozdWt|9zwDdJGU_rVY-+mvxYTDXf&YWh;v0pQ;CKKBZM3nYGPo>Ml!j}q z=`PAqet3I+!N6ie7D9J-YJ~M2pXR}Z+1@!7v-7M7U9Xs{B~8Sid}(w*@yb}rcV&Tu=rgNBq=?iW^0p$wl25cf45a?~Gk zJ_K#VdbvRFUhHU(*XxKz@yTiqAFK7%$$zn<9jQDA^&zI@}x4voe5Vp9O8R4&uf zLS2}r_7@RIe{dL-RWz>zh1=l2(8|Fhe@!S zNMaIpN|Eey>E-#2T=C!#KrHJy++3+gJO!q*TS)*H!H>UtCvtrHJRMuD*{rm6=8@os zPf<6QXNh^Qq7I)GG0{|b>NIpq(cim{dKk)lk^H6Y!DYC<3 zzh!C5=P77Z@$>c&L;7fSJhh3(ag|Xu3g`f-sS+fQK~k@_I7%}M`JNY-c3>M~am-Nq zg4aj24x{6al6HPoZV|$zx+oK}^qgLq&R|H1_9y?{aJ5Ed4rC zSW0Pew~Z@Q#)=o;!V1F(#Wj#X*nXiG_q==xQw+Jtux6RzY?!C{>cmLUEcJ?#7d-3UF37|L7(boY$AOHYe4&Y)QxCu~@ zk&%;;Qjn9AQ&Lh;QNw7cp-^h3D_7}YY|I?&Y|N~zoV?-!oZO;3tgM3aLZXsVGBPq8 z0*a~%(kkLNWTby@0-~g(q=r&6($FwUbFp$s|DWsP9l$^V_yPf75H~==00J|BE_wit z%X5-~{x0BO1tI}MNXf`4D5;>A4UO~w2?z`(fq+R#A&|@N(93lI!a#b3OG=aMs_7$g zZax@jyeEb4}Lc%h#a`Fm_N;hw5>*(s~8{D~j z&)nj^rImxDle3Gfn|r`x#FIcID(KJXnAkXUd_qd<^R)Dg%okZj#U-U>ru{AZ-@&5(U$Xze{_C0psKKDi%L6k2>cGi)f{2M8;jpI8z+rrgzxC0~ zkj-+$4Z+&6Fc@+S4c5+dT*>-z0pKYKcJGJPpo`aTAGM_xIMr)VnD486B#&n-0rRo! zmd(5x-3it}wRaug6lI@a@%{8A`E0EMS^Q2es4po`h>&UOCiq1*9Tk0qYjOn-xt6Q= zuslZyzgvnDY5+_v_H$YOn2|^sHEu$!1w_CepUzae<%8M8+JDk}q|c#vqV2LzmLpVx z680~E>mfqSq0xco%pFEnb21lz-tOp!5`@BjofK!3CL>Qp0i-BF5cr+CLQ#p5<3K{& z;O4mAa@G%3>;*9A?H}r^P?z_8K_(xXkG6yEztrx!Hz_vN1--&_ZKA7tVxyUCV6glD?+ntPv2y2AqOy8wo$LL0xewsmFKOjGeG;tRRFq zYb?ulN4Cb}%4>pjd%9>pFD|$SK1ts-;#J1mcXd;Hg4E;TdlvwaQ%$&G%IZ$hLfB>~ zHtj}^TzIJz`&3}XMi{)l?Ws0_=!;_&5u2=f0<#BE87qEGAYd%4rrVt=h7f@&wa=mk zi}~zLKT;iyUi%hzpTIV$?jSeQ{KRyyrxw~q+OKwe$Ce3C$7+jITfEzt1WEJZl7jmk zS(k+i19f;Bt39Kc>eV;xuB)_x#d2^Qf~fsi9j+n)j~T8*KRyc}O)8O>x#cjF!Sv{4 ztp#eyx+yuRb^&D7y!)ATsQy%qw`nQ_T4b$HY2Elx_oEQx{4r?xxv4e)y9|6B8GvIx zie;B7`|=z^d&_HZZzf^w+g*Izbc9+oVYKh*(`+0wyP?Kk;-#L(QtcKUi#7{ZNa>mWyjxnl;#EiLUb}NT6YOIH;80r_MyVO7G5Y>&DrV5O!FKuY$6rw zUtR3WXB^Bt`m9+G$F@13+J)s{b_zzk+TnW_XmR4r$3$ITA<%4M_gyt0aGY`-Gat3hEX?SLoIrN$n~w_it1lzZBnZsoqP3H58*w2YGy z-*O74n%&7rmJQBEhG2}9+2jycsz26R&M?i-?+{(VgCQuwX*i{a^T@bGM?}QEDM{VN z0EUgMl410gfWtYy_E#9^ag@i{n;vvU%Uu@z?xY;&rtJ`uD%t7W(3i**%dOHj23@xL zm0%avy?4U^^6T`Q`45EQtGsC6L0!(V2l9ztE+&lbg(Lbc&O|z%eVISV1+-y*kYO;h z@ndiHgdi5<^73w#M9B&7BHYjGj&UuV9GfDmOL~g%pLVo9WF3V1Q>?W2{_L&(;+7GU z4I}>8B=*CvzjBLlAF6d%@5-FLjcbkHYVB+cmy^n~)HOYEZV0(ao)#)nCT9S7dAzH; z=8>B)Irkweck2CIH_K75VxZdl@f7e)T##107vhZ5YRK%tIvBlrE=$DL%D#S^aOgvd zY}T6eTq7J-I;_rpRz&h=q+Jclzn#Bs^W!xkwbw8}b+{>GO_MiqU{ww|YMUz@jhccx zUjQa=C&csS4!VeuNj>2s%+snPi4Nr7QZl?>21J^3(pC3($^6 z8*7=2Yb+xQ0}8BH0ir8YkJaXkm0InQx$CRjr5SEa^PdtnxdUY5VupKK&$RapmV!=VTLrfy4xq(o zJ~uwTcwv;PQIKp`Ok=2(1m$}-N37V9zD%MdJtZY?7Q`O;$tIa;GqRX7zUB0x!+tM( z3i+cC)$UTE4|~;im7S5j>#da_7$;bi0Eo04+pUlilH;SBA=x6&R*gHF@($jQX>jNw zzjw>kmCw7MdPXe|>dW^f+=%fuQOlW0M5a)R|4hswbjxTc(MEu5-xFARW<^5?AMg*!iLP-!U>Mv0$QZuxa_2on1%nbsumi(7__NnJ1Xoz{RI zE1kDFh3mYI_H9Ni4|d3Bi9ach(>J%bXVzY|;hX4wB<4&Tzr+s*_Qf1zyV@@MGvx+v zlP3m)B5nFI5gas|(`|q8&)w(i|LluY;5nH9PZmt?hbKs4SbF|~O3?Q)-tOOKHfW$EjkunZQB39ZyOMN6n+rt^lj?fT2*o*nYcpd(CLj=&@;t+axzlv(`9v z>z0PYew}1dBzsp05z&=kl$Dh2`tgAM_CtmZ5#=5`md0Q1gN7pa3tvYLkPmHIOf;e} zFp&QP^HVqKWZ%^!#ZkQWouUXT!QZJ(#i}v1?c}3*CdB549+Z}3FnyAm*(;#4_?1KL z=2xxr+2r+oi8B@RMU0ozgbbVWk?9*lyQVKOG@i3=u*YBC?Zf3`9$x@ZBHx+y7xx-O z*v2(yXhbm&xNf|BMz6U^sfJiRXBAnE+2yZ)o*s!(Qi@`TXu)uR2v5T zI~Fw|(5aPi#ED_Mo;U?MLt&+k16{{gYVOEc$(ghqq4xav^%4smy$#z=|JCo@?S zM^aVxa_Ov;Dl5Izi{U#8^*t$_g!-Yj=BJ zF{FXB#MXOC;ai^a9Sskn10g>Z5B0KL#moC;PAc41# + + + Month + NVIDIA + Alibaba Cloud + Hugging Face + Ai2 + Meta + Google + IBM + ByteDance + Baidu + Tencent + Microsoft + Zai + + + + + Jan 2025 + ~10 + ~10 + ~10 + ~10 + ~10 + ~10 + ~10 + ~10 + ~10 + ~10 + ~10 + ~10 + + + Feb 2026 + ~450 + ~350 + ~200 + ~180 + ~150 + ~130 + ~120 + ~110 + ~100 + ~90 + ~80 + ~70 + + + \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-1 NVIDIA Definitions.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-1 NVIDIA Definitions.html new file mode 100644 index 000000000..086d709da --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-1 NVIDIA Definitions.html @@ -0,0 +1 @@ +
2007 PlanNVIDIA Corporation Amended and Restated 2007 Equity Incentive Plan
ACAudit Committee of the Board
ASC 718Financial Accounting Standards Board Accounting Standards Codification Topic 718: Compensation - Stock Compensation
Base Compensation PlanPerformance goal necessary to earn the target award under the Variable Cash Plan and for the target numbers of SY PSUs and MY PSUs to become eligible to vest
BoardThe Company's board of directors
BylawsThe Company's Amended and Restated Bylaws
CAP"Compensation actually paid," as defined under Item 402(v) of Regulation S-K
CCCompensation Committee of the Board
CD&ACompensation Discussion and Analysis
CEOChief Executive Officer
CFOChief Financial Officer
CharterThe Company's Restated Certificate of Incorporation
Control NumberIdentification number for each stockholder included in Notice or proxy card
CSCorporate Sustainability
CSSCCorporate Sustainability Steering Committee
DEIDiversity, Equity, and Inclusion
ERMEnterprise Risk Management
ESPPNVIDIA Corporation Amended and Restated 2012 Employee Stock Purchase Plan
EVPExecutive Vice President
Exchange ActSecurities Exchange Act of 1934, as amended
Fiscal 20__The Company's fiscal year ended on the last Sunday in January of the stated year
Form 10-KThe Company's Annual Report on Form 10-K for Fiscal 2026 filed with the SEC on February 25, 2026
GAAPGenerally accepted accounting principles in the United States
GHGGreenhouse gas
Internal Revenue CodeU.S. Internal Revenue Code of 1986, as amended
Lead DirectorLead independent director
MeetingAnnual Meeting of Stockholders
MY PSUsMulti-year PSUs based on 3-year TSR relative to the S&P 500 with a three-year performance metric, vesting after three years
NasdaqThe Nasdaq Stock Market LLC
NCGCNominating and Corporate Governance Committee of the Board
NEOsNamed Executive Officers consisting of our CEO, our CFO, and our other three most highly compensated executive officers as of the end of Fiscal 2026
Non-GAAP Operating IncomeGAAP operating income, as the Company reports in its SEC filings, excluding stock-based compensation expense, acquisition-related and other costs, and other. Please see Reconciliation of Non-GAAP Financial Measures in our CD&A for a reconciliation between the non-GAAP financial measures and GAAP results
NoticeNotice of Internet Availability of Proxy Materials
NVIDIA, Company, we, us, ourNVIDIA Corporation, a Delaware corporation
NYSENew York Stock Exchange
PSUPerformance stock unit
PwCPricewaterhouseCoopers LLP
RSURestricted stock unit
S&P 500Standard & Poor's 500 Composite Index
SECU.S. Securities and Exchange Commission
Section 162(m)Section 162(m) of the Internal Revenue Code
Securities ActSecurities Act of 1933, as amended
Stretch Compensation PlanPerformance goal necessary to earn the maximum award under the Variable Cash Plan and for the maximum numbers of SY PSUs and MY PSUs to become eligible to vest
SY PSUsPSUs based on annual Non-GAAP Operating Income performance with a single-year performance metric, vesting over four years
ThresholdMinimum performance goal necessary to earn an award under the Variable Cash Plan and for SY PSUs and MY PSUs to become eligible to vest
TSRTotal shareholder return
Variable Cash PlanThe Company's variable cash compensation plan
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-10 Performance Payouts.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-10 Performance Payouts.html new file mode 100644 index 000000000..104bad91e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-10 Performance Payouts.html @@ -0,0 +1 @@ +
PERFORMANCE ACHIEVEMENT AND PAYOUTS
Variable Cash PlanSY PSUsMY PSUs
Performance Achievement for Period Ended Fiscal 2026 (3)$215.9 billion revenue$137.3 billion Non-GAAP Operating Income (1)100th percentile 3-year TSR relative to S&P 500 (1,055%)
Payout as % of Target Opportunity200%CEO 150% Other NEOs 200%CEO 150% Other NEOs 200%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-100 Income Before Tax.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-100 Income Before Tax.html new file mode 100644 index 000000000..7d09ab988 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-100 Income Before Tax.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
U.S.$ 123,181$ 77,456$ 29,495
Foreign18,2696,5704,323
Income before income tax$ 141,450$ 84,026$ 33,818
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-101 Tax Rate Breakdown.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-101 Tax Rate Breakdown.html new file mode 100644 index 000000000..17dc4c296 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-101 Tax Rate Breakdown.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026
(In millions, except percentages)
US Federal Statutory Tax Rate$ 29,70421.0 %
State and Local Income Taxes, Net of Federal Income Tax Effect (1)2580.2 %
Foreign tax effects
Israel
Reduced statutory tax rate on qualifying income(3,064)(2.2)%
Other1,6061.2 %
Other foreign jurisdictions7410.5 %
Effect of cross-border tax laws
Foreign-derived deduction eligible income(4,208)(3.0)%
Other(142)(0.1)%
Tax credits(1,933)(1.4)%
Nontaxable or nondeductible items
Stock-based compensation(1,475)(1.0)%
Other29— %
Other (2)(133)(0.1)%
Income tax expense$ 21,38315.1 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-102 Tax Rate Analysis.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-102 Tax Rate Analysis.html new file mode 100644 index 000000000..2b9c6c79d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-102 Tax Rate Analysis.html @@ -0,0 +1 @@ +
Year Ended
Jan 26, 2025Jan 28, 2024
(In millions, except percentages)
Tax expense computed at federal statutory rate$ 17,64521.0 %$ 7,10221.0 %
Expense (benefit) resulting from:
State income taxes, net of federal tax effect5540.7 %1200.4 %
Foreign-derived deduction eligible income(2,976)(3.5)%(1,408)(4.2)%
Stock-based compensation(2,097)(2.5)%(741)(2.2)%
U.S. federal research and development tax credit(990)(1.2)%(431)(1.3)%
Foreign tax rate differential(984)(1.2)%(467)(1.4)%
Other(6)— %(117)(0.3)%
Income tax expense$ 11,14613.3 %$ 4,05812.0 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-103 Income Tax Paid.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-103 Income Tax Paid.html new file mode 100644 index 000000000..d65faaa5e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-103 Income Tax Paid.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026
(In millions)
Federal$ 16,755
State
California1,049
Other1,041
Foreign
Israel1,287
Other156
Total income taxes paid, net of refunds$ 20,288
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-104 Deferred Tax Assets.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-104 Deferred Tax Assets.html new file mode 100644 index 000000000..63dddfd7e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-104 Deferred Tax Assets.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
(In millions)
Deferred tax assets:
Capitalized research and development expenditure$ 5,436$ 6,256
Net controlled foreign corporation tested income deferred tax assets5,3892,820
Accruals and reserves, not currently deductible for tax purposes3,6442,058
Research and other tax credit carryforwards718759
Operating lease liabilities554299
Net operating loss and capital loss carryforwards443456
Other deferred tax assets679566
Gross deferred tax assets16,86313,214
Less valuation allowance(768)(1,610)
Total deferred tax assets16,09511,604
Deferred tax liabilities:
Equity investments(2,227)(264)
Unremitted earnings of foreign subsidiaries(1,813)(891)
Operating lease assets(533)(286)
Acquired intangibles(38)(70)
Gross deferred tax liabilities(4,611)(1,511)
Net deferred tax asset (1)$ 11,484$ 10,093
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-105 Tax Positions.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-105 Tax Positions.html new file mode 100644 index 000000000..848b426a0 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-105 Tax Positions.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Balance at beginning of period$ 2,861$ 1,670$ 1,238
Increases in tax positions for current year1,9591,268616
Increases in tax positions for prior years574887
Lapse in statute of limitations(224)(27)(19)
Decreases in tax positions for prior years(157)(88)(148)
Settlements(76)(10)(104)
Balance at end of period$ 4,420$ 2,861$ 1,670
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-106 Segment Revenue.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-106 Segment Revenue.html new file mode 100644 index 000000000..2e6eedd6b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-106 Segment Revenue.html @@ -0,0 +1 @@ +
Compute & NetworkingGraphicsTotal
(In millions)
Year Ended Jan 25, 2026
Revenue$ 193,479$ 22,459$ 215,938
Other segment items (1)63,33813,30376,641
Operating income$ 130,141$ 9,156$ 139,297
Year Ended Jan 26, 2025
Revenue$ 116,193$ 14,304$ 130,497
Other segment items (1)33,3189,21942,537
Operating income$ 82,875$ 5,085$ 87,960
Year Ended Jan 28, 2024
Revenue$ 47,405$ 13,517$ 60,922
Other segment items (1)15,3897,67123,060
Operating income$ 32,016$ 5,846$ 37,862
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-107 Income Breakdown.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-107 Income Breakdown.html new file mode 100644 index 000000000..2eec26623 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-107 Income Breakdown.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Segment operating income$ 139,297$ 87,960$ 37,862
Stock-based compensation expense(6,386)(4,737)(3,549)
Unallocated operating expenses(1,997)(1,171)(728)
Acquisition-related and other costs(527)(599)(613)
Interest income2,3001,786866
Interest expense(259)(247)(257)
Other income, net9,0221,034237
Consolidated income before income tax$ 141,450$ 84,026$ 33,818
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-108 Revenue by Region.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-108 Revenue by Region.html new file mode 100644 index 000000000..7fcebafe1 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-108 Revenue by Region.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Geographic Location (1):Revenue based upon Customer Headquarters(In millions)
United States$149,617$ 77,482$ 31,533
Taiwan (2)42,34523,60014,912
China (including Hong Kong)19,67725,04812,330
Other4,2994,3672,147
Total revenue$215,938$ 130,497$ 60,922
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-109 NVIDIA Revenue.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-109 NVIDIA Revenue.html new file mode 100644 index 000000000..a43a841d4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-109 NVIDIA Revenue.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Revenue by End Market:(In millions)
Data Center$ 193,737$ 115,186$ 47,525
Compute162,361102,19638,950
Networking31,37612,9908,575
Gaming16,04211,35010,447
Professional Visualization3,1911,8781,553
Automotive2,3491,6941,091
OEM and Other619389306
Total revenue$ 215,938$ 130,497$ 60,922
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-11 Proxy Voting.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-11 Proxy Voting.html new file mode 100644 index 000000000..8ab1b2e34 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-11 Proxy Voting.html @@ -0,0 +1 @@ +
VoteChange Your VoteRevoke Your Proxy
Virtually attend and vote at the 2026 Meeting
Via mail, by signing and mailing your proxy card to us before the 2026 Meeting
By telephone or online, by following the instructions provided in the Notice or your proxy materials
Submit another properly completed proxy card with a later date
Send a written notice that you are revoking your proxy to NVIDIA Corporation, 2788 San Tomas Expressway, Santa Clara, California 95051, Attention: Timothy S. Teter, Secretary, or via email to shareholdermeeting@nvidia.com
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-110 Long-lived assets.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-110 Long-lived assets.html new file mode 100644 index 000000000..d41fe63db --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-110 Long-lived assets.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
Long-lived assets:(In millions)
United States$ 5,125$ 3,626
Taiwan3,2191,481
Israel1,471840
Other568336
Total long-lived assets$ 10,383$ 6,283
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-111 Lease Obligations.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-111 Lease Obligations.html new file mode 100644 index 000000000..832076797 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-111 Lease Obligations.html @@ -0,0 +1 @@ +
Operating Lease Obligations
(In millions)
Fiscal Year:
2027$ 493
2028485
2029457
2030381
2031314
2032 and thereafter1,494
Total3,624
Less imputed interest680
Present value of net future minimum lease payments2,944
Less short-term operating lease liabilities372
Long-term operating lease liabilities$ 2,572
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-112 Lease Cash Flows.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-112 Lease Cash Flows.html new file mode 100644 index 000000000..303622922 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-112 Lease Cash Flows.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Supplemental cash flows information
Operating cash flow used for operating leases$ 428$ 313$ 286
Operating lease assets obtained in exchange for lease obligations$ 1,439$ 877$ 531
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-113 Allowance Changes.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-113 Allowance Changes.html new file mode 100644 index 000000000..0b3ee6294 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-113 Allowance Changes.html @@ -0,0 +1 @@ +
DescriptionBalance at Beginning of PeriodAdditionsDeductionsBalance at End of Period
(In millions)
Fiscal year 2026
Allowance for doubtful accounts$ 4$ — (1)$ — (1)$ 4
Sales return allowance$ 82$ 188 (2)$ (100) (4)$ 170
Deferred tax valuation allowance$ 1,610$ 31 (3)$ (873) (3)$ 768
Fiscal year 2025
Allowance for doubtful accounts$ 4$ — (1)$ — (1)$ 4
Sales return allowance$ 109$ 151 (2)$ (178) (4)$ 82
Deferred tax valuation allowance$ 1,552$ 58 (3)$ — (3)$ 1,610
Fiscal year 2024
Allowance for doubtful accounts$ 4$ — (1)$ — (1)$ 4
Sales return allowance$ 26$ 213 (2)$ (130) (4)$ 109
Deferred tax valuation allowance$ 1,484$ 162 (3)$ (94) (3)$ 1,552
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-114 Exhibit Index.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-114 Exhibit Index.html new file mode 100644 index 000000000..2424c862b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-114 Exhibit Index.html @@ -0,0 +1 @@ +
Exhibit No.Exhibit DescriptionIncorporated by ReferenceFiling Date
Schedule/FormExhibit
3.1Restated Certificate of Incorporation10-K3.13/18/2022
3.2Amendment to Restated Certificate of Incorporation of NVIDIA Corporation8-K3.16/6/2022
3.3Amendment to Restated Certificate of Incorporation of NVIDIA Corporation8-K3.16/7/2024
3.4Bylaws of NVIDIA Corporation, Amended and Restated as of March 12, 20248-K3.13/14/2024
4.1Reference is made to Exhibits 3.1, 3.2, 3.3 and 3.4
4.2Specimen Stock CertificateS-1/A4.24/24/1998
4.3Indenture, dated as of September 16, 2016, by and between the Company and Computershare Trust Company, N.A., as successor to Wells Fargo Bank, National Association, as Trustee8-K4.19/16/2016
4.4Officers' Certificate, dated as of September 16, 20168-K4.29/16/2016
4.5Form of 2026 Note8-KAnnex B-1 to Exhibit 4.29/16/2016
4.6Description of Securities10-K4.62/26/2025
4.7Officers' Certificate, dated as of March 31, 20208-K4.23/31/2020
4.8Form of 2030 Note8-KAnnex A-1 to Exhibit 4.23/31/2020
4.9Form of 2040 Note8-KAnnex B-1 to Exhibit 4.23/31/2020
4.10Form of 2050 Note8-KAnnex C-1 to Exhibit 4.23/31/2020
4.11Form of 2060 Note8-KAnnex D-1 to Exhibit 4.23/31/2020
4.12Officers' Certificate, dated as of June 16, 20218-K4.26/16/2021
4.13Form of 2028 Note8-KAnnex C-1 to Exhibit 4.26/16/2021
4.14Form of 2031 Note8-KAnnex D-1 to Exhibit 4.26/16/2021
10.1Form of Indemnity Agreement between NVIDIA Corporation and each of its directors and officers8-K10.13/7/2006
10.2+Amended and Restated 2007 Equity Incentive Plan10-Q10.18/28/2024
10.3+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Deferred Restricted Stock Unit Grant Notice and Deferred Restricted Stock Unit Agreement (2016)10-K10.263/12/2015
10.4+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Restricted Stock Unit Grant Notice and Restricted Stock Unit Agreement (2016)10-K10.273/12/2015
10.5+Amended and Restated 2007 Equity Incentive Plan - Global Performance-Based Restricted Stock Unit Grant Notice and Performance-Based Restricted Stock Unit Agreement (2019)8-K10.13/11/2019
10.6+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2021)10-Q10.25/26/2021
10.7+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2022)10-K10.163/18/2022
10.8+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2023)10-K10.142/24/2023
10.9+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2024) (version 1)10-Q10.25/29/2024
10.10+Amended and Restated 2007 Equity Incentive Plan - Global Performance-Based Restricted Stock Unit Grant Notice and Performance-Based Restricted Stock Unit Agreement (2024)10-Q10.35/29/2024
10.11+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2024) (version 2)10-Q10.111/20/2024
10.12+Amended and Restated 2007 Equity Incentive Plan - Global Restricted Stock Unit Grant Notice and Global Restricted Stock Unit Agreement (2025)10-K10.132/26/2025
10.13+Amended and Restated 2007 Equity Incentive Plan - Global Performance-Based Restricted Stock Unit Grant Notice and Performance-Based Restricted Stock Unit Agreement (2025)10-K10.142/26/2025
10.14+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Deferred Restricted Stock Unit Grant Notice and Deferred Restricted Stock Unit Agreement (2025)10-Q10.18/27/2025
10.15+Amended and Restated 2007 Equity Incentive Plan - Non-Employee Director Restricted Stock Unit Grant Notice and Restricted Stock Unit Agreement (2025)10-Q10.28/27/2025
10.16+Amended and Restated 2012 Employee Stock Purchase Plan10-K10.152/26/2025
10.17+Variable Compensation Plan - Fiscal Year 20258-K10.13/14/2024
10.18+Variable Compensation Plan - Fiscal Year 20268-K10.13/7/2025
19.1*NVIDIA Corporation Insider Trading Policy
21.1*Subsidiaries of Registrant
23.1*Consent of PricewaterhouseCoopers LLP
24.1*Power of Attorney (included in signature page)
31.1*Certification of Chief Executive Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
31.2*Certification of Chief Financial Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
32.1#*Certification of Chief Executive Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
32.2#*Certification of Chief Financial Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
97.1+Compensation Recovery Policy, as amended and restated November 30, 202310-K97.12/21/2024
101.INS*XBRL Instance Document
101.SCH*XBRL Taxonomy Extension Schema Document
101.CAL*XBRL Taxonomy Extension Calculation Linkbase Document
101.DEF*XBRL Taxonomy Extension Definition Linkbase Document
101.LAB*XBRL Taxonomy Extension Labels Linkbase Document
101.PRE*XBRL Taxonomy Extension Presentation Linkbase Document
104Cover Page Interactive Data File - the cover page interactive data file does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-115 Signatures.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-115 Signatures.html new file mode 100644 index 000000000..7f644d48f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-115 Signatures.html @@ -0,0 +1 @@ +
SignatureTitleDate
/s/ JEN-HSUN HUANGPresident, Chief Executive Officer and Director (Principal Executive Officer)February 25, 2026
Jen-Hsun Huang
/s/ COLETTE M. KRESSExecutive Vice President and Chief Financial Officer (Principal Financial Officer)February 25, 2026
Colette M. Kress
/s/ DONALD ROBERTSONVice President and Chief Accounting Officer (Principal Accounting Officer)February 25, 2026
Donald Robertson
/s/ TENCH COXEDirectorFebruary 25, 2026
Tench Coxe
/s/ JOHN O. DABIRIDirectorFebruary 25, 2026
John O. Dabiri
/s/ DAWN HUDSONDirectorFebruary 25, 2026
Dawn Hudson
/s/ HARVEY C. JONESDirectorFebruary 25, 2026
Harvey C. Jones
/s/ MELISSA B. LORADirectorFebruary 25, 2026
Melissa B. Lora
/s/ STEPHEN C. NEALDirectorFebruary 25, 2026
Stephen C. Neal
/s/ A. BROOKE SEAWELLDirectorFebruary 25, 2026
A. Brooke Seawell
/s/ AARTI SHAHDirectorFebruary 25, 2026
Aarti Shah
/s/ MARK A. STEVENSDirectorFebruary 25, 2026
Mark A. Stevens
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-12 Proposal Voting.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-12 Proposal Voting.html new file mode 100644 index 000000000..41f136734 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-12 Proposal Voting.html @@ -0,0 +1 @@ +
Proposal NumberProposal DescriptionVote Required for ApprovalEffect of AbstentionsBroker Non-Votes
1Election of ten directorsMore FOR than AGAINST votesNoneNone
2Advisory approval of our executive compensationMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
3Ratification of the selection of PwC as our independent registered public accounting firm for Fiscal 2027Majority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstN/A*
4Stockholder Proposal: Simple Majority Vote StandardMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
5Stockholder Proposal: Report on Faith-Based Community Resource GroupsMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
6Stockholder Proposal: Report on Workforce Civil LibertiesMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
7Stockholder Proposal: Greenhouse Gas Emissions DisclosureMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-13 NVIDIA Board.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-13 NVIDIA Board.html new file mode 100644 index 000000000..7f62c4c95 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-13 NVIDIA Board.html @@ -0,0 +1 @@ +
NameAgeDirector SinceOccupationIndependentFinancial Expert (1)Committee MembershipNumber of Other Public Company Boards
Tench Coxe681993Former Managing Director, Sutter Hill VenturesAC, CC
John O. Dabiri462020Centennial Professor of Aeronautics and Mechanical Engineering, California Institute of TechnologyCC, NCGC
Jen-Hsun Huang631993President & CEO, NVIDIA Corporation
Dawn Hudson682013Former Chief Marketing Officer, National Football League & Former CEO Pepsi-Cola North AmericaCC Chairperson
Harvey C. Jones731993Managing Partner, Square Wave VenturesAC, NCGC
Melissa B. Lora632023Former President, Taco Bell InternationalAC, NCGC1
Stephen C. Neal (Lead Director)772019Chairman Emeritus & Senior Counsel, Cooley LLPNCGC Chairperson
A. Brooke Seawell781997Venture Partner, New Enterprise AssociatesAC Chairperson1
Aarti Shah612020Former Senior Vice President & Chief Information and Digital Officer, Eli Lilly and CompanyAC, CC1
Mark A. Stevens662008 (2)Managing Partner, S-Cubed CapitalCC, NCGC
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-14 Board Skills Matrix.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-14 Board Skills Matrix.html new file mode 100644 index 000000000..a366964fb --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-14 Board Skills Matrix.html @@ -0,0 +1 @@ +
Senior Leadership & Operations ExperienceDirectors with senior leadership and operations experience provide informed oversight of our business, and unique experiences and perspectives. They are uniquely positioned to contribute practical insight into business strategy and operations, driving growth, building and strengthening corporate culture, and supporting the achievement of strategic priorities and objectives.
Industry & TechnicalDirectors with industry experience and technical backgrounds facilitate within the Board a deeper understanding of innovations and a technical assessment of our products and services.
Financial/Financial CommunityExperience in financial matters and the financial community assists our Board with review of our operations and finances, including overseeing our financial statements, capital structure and internal controls. Those with a venture capital background also offer valuable stockholder perspectives.
Governance & Public Company BoardDirectors with experience in corporate governance, such as service on boards and board committees, or as executives of other large, public companies, are familiar with the dynamics and operation of a board of directors and the impact that governance policies have on a company. This experience supports our goals of strong Board and management accountability, transparency, and protection of stockholder interests. Public company board experience also helps our directors identify challenges and risks we face as a public company, including oversight of strategic, operational, and compliance-related matters, and stockholder relations.
Emerging Technologies & Business ModelsExperience in emerging technologies and business models is integral to our growth strategies given our unique business model and provides important insights as our business expands into new areas.
Marketing, Communications & Brand ManagementDirectors with experience in marketing, communications, and brand management offer guidance on our products directly marketed to consumers, important perspectives on expanding our market share, and expertise in communicating with our customers and other stakeholders.
Regulatory, Legal & Risk ManagementOur business requires compliance with a variety of regulatory requirements in different jurisdictions. We face new regulatory matters and regulations as our business grows. We are also subject to multiple lawsuits. Directors with experience in governmental, public policy, legal, and risk management areas, including cybersecurity, help provide valuable insights and oversight for our Company.
Human Capital Management ExperienceOur people are critical to our success. Directors with experience in organizational management, talent development, and developing values and culture in a large global workforce provide key insights. Human capital management experience also assists our Board in overseeing executive and employee compensation, development, and engagement.
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-15 Director Profile.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-15 Director Profile.html new file mode 100644 index 000000000..31d49cef1 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-15 Director Profile.html @@ -0,0 +1 @@ +
Age: 68Director Since: 1993Committees: AC, CC
Other Current Public Company Boards: NoneIndependent Director
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-16 Director Profile.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-16 Director Profile.html new file mode 100644 index 000000000..674d8e921 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-16 Director Profile.html @@ -0,0 +1 @@ +
Age: 46Director Since: 2020Committees: CC, NCGC
Other Current Public Company Boards: NoneIndependent Director
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-17 Director Profile.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-17 Director Profile.html new file mode 100644 index 000000000..4b0fd0d21 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-17 Director Profile.html @@ -0,0 +1 @@ +
Age: 63Director Since: 2023Committees: AC, NCGC
Other Current Public Company Boards: Conagra Brands, Inc. (since 2019)Independent DirectorFinancial Expert
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-18 Leadership & Operations.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-18 Leadership & Operations.html new file mode 100644 index 000000000..dd59b3b3b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-18 Leadership & Operations.html @@ -0,0 +1 @@ +
Senior Leadership & Operations ExperienceFinancial/Financial CommunityGovernance & Public Company Board
Emerging Technologies & Business ModelsMarketing, Communications & Brand ManagementHuman Capital Management Experience
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-19 Director Profile.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-19 Director Profile.html new file mode 100644 index 000000000..2510f83d9 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-19 Director Profile.html @@ -0,0 +1 @@ +
Age: 77Director Since: 2019Committees: NCGC
Other Current Public Company Boards: NoneIndependent DirectorLead Director
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-2 Revenue Growth.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-2 Revenue Growth.html new file mode 100644 index 000000000..7275ba1a7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-2 Revenue Growth.html @@ -0,0 +1 @@ +
RevenueGross MarginOperating IncomeDiluted Earnings Per Share
$215.9 billion71.1%$130.4 billion$4.90
up 65% year on yeardown 3.9 points year on yearup 60% year on yearup 67% year on year
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-20 Leadership & Governance.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-20 Leadership & Governance.html new file mode 100644 index 000000000..332f03d89 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-20 Leadership & Governance.html @@ -0,0 +1 @@ +
Senior Leadership & Operations ExperienceGovernance & Public Company BoardMarketing, Communications & Brand Management
Regulatory, Legal & Risk ManagementHuman Capital Management Experience
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-21 Director Profile.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-21 Director Profile.html new file mode 100644 index 000000000..02b8b2357 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-21 Director Profile.html @@ -0,0 +1 @@ +
Age: 78Director Since: 1997Committees: AC
Other Current Public Company Boards: Tenable Holdings, Inc. (since 2017)Independent DirectorFinancial Expert
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-22 Director Profile.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-22 Director Profile.html new file mode 100644 index 000000000..c987bfbfd --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-22 Director Profile.html @@ -0,0 +1 @@ +
Age: 61Director Since: 2020Committees: AC, CC
Other Current Public Company Boards: Sandoz Group AG (since 2023)Independent Director
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-23 Board Risk Oversight.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-23 Board Risk Oversight.html new file mode 100644 index 000000000..db1ec2fd6 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-23 Board Risk Oversight.html @@ -0,0 +1 @@ +
Board of Directors
Oversees management of major risks, including:
√ Business model, including AI√ Strategic execution√ Product quality and safety
√ Operational, including supply chain and sourcing√ Regulatory, public policy, legal, intellectual property, and compliance√ Financial and macroeconomic
√ Information security, including cybersecurity√ Brand and reputation√ Business continuity
√ Corporate development, acquisitions, and investments√ Management development√ Enterprise resource planning
√ ERM
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-24 Board Oversight.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-24 Board Oversight.html new file mode 100644 index 000000000..663bc4ea7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-24 Board Oversight.html @@ -0,0 +1 @@ +
ACCCNCGC
√ Financial statement and earnings materials integrity and reporting√ Financial risk exposures, including investments, cash management, foreign exchange management and insurance coverage√ Disclosure controls and procedures√ Information security and cybersecurity policies and practices and the internal controls regarding information security risks√ Internal audit performance, including auditor functions, performance, and independence√ Accounting and audit principles and policies, and regulatory and accounting initiatives√ Legal and regulatory compliance, particularly as related to the above matters√ ERM√ Compensation policies, plans, practices and programs for directors, executives, and employees√ Human capital management, including recruiting, retention, development, and other workforce matters√ Governance structure, processes and policies, including regulatory changes and other developments√ Stockholder concerns and communications√ Compliance program and effectiveness of our anonymous tip process√ CS, including environmental, social, and corporate governance matters√ Trade compliance and non-financial regulatory matters√ Board and committee composition and board evaluation√ Related party transactions√ Policies and practices related to government relations, public policy, and related expenditures
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-25 Director Compensation.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-25 Director Compensation.html new file mode 100644 index 000000000..ab648d1f5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-25 Director Compensation.html @@ -0,0 +1 @@ +
NameFees Earned or Paid in Cash ($)Stock Awards ($) (1)Total ($)
Robert K. Burgess (2)63,750278,809342,559
Tench Coxe85,000278,809363,809
John O. Dabiri85,000278,809363,809
Persis S. Drell85,000278,809363,809
Dawn Hudson85,000278,809363,809
Harvey C. Jones85,000278,809363,809
Melissa B. Lora85,000278,809363,809
Stephen C. Neal85,000278,809363,809
Ellen Ochoa (3)42,500278,809321,309
A. Brooke Seawell85,000278,809363,809
Aarti Shah85,000278,809363,809
Mark A. Stevens85,000278,809363,809
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-26 RSU Holdings Table.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-26 RSU Holdings Table.html new file mode 100644 index 000000000..3640f0bea --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-26 RSU Holdings Table.html @@ -0,0 +1 @@ +
NameRSUsNameRSUs
Robert K. Burgess (1)Melissa B. Lora (3)1,880
Tench Coxe900Stephen C. Neal900
John O. Dabiri900Ellen Ochoa (4)
Persis S. Drell (2)A. Brooke Seawell900
Dawn Hudson900Aarti Shah900
Harvey C. Jones900Mark A. Stevens900
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-27 Beneficial Ownership.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-27 Beneficial Ownership.html new file mode 100644 index 000000000..063dd776f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-27 Beneficial Ownership.html @@ -0,0 +1 @@ +
Name of Beneficial OwnerShares OwnedShares Issuable Within 60 DaysTotal Shares Beneficially OwnedPercent
NEOs:
Jen-Hsun Huang870,604,104 (1)870,604,1043.58%
Colette M. Kress4,443,523 (2)4,443,523*
Ajay K. Puri3,266,897 (3)3,266,897*
Debora Shoquist1,580,729 (4)1,580,729*
Timothy S. Teter2,687,197 (5)2,687,197*
Directors, not including Mr. Huang:
Tench Coxe30,555,240 (6)30,555,240*
John O. Dabiri13,88890014,788*
Dawn Hudson367,987900368,887*
Harvey C. Jones7,002,787 (7)9007,003,687*
Melissa B. Lora899 (8)9001,799*
Stephen C. Neal183,967 (9)900184,867*
A. Brooke Seawell2,505,707 (10)9002,506,607*
Aarti Shah25,620 (11)25,620*
Mark A. Stevens34,068,672 (12)90034,069,572*
Directors and executive officers as a group (14 persons)957,307,217 (13)6,300957,313,5173.94%
5% Stockholders:
BlackRock, Inc.1,805,935,550 (14)1,805,935,5507.43%
Vanguard Capital Management1,777,408,252 (15)1,777,408,2527.31%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-28 Compensation Breakdown.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-28 Compensation Breakdown.html new file mode 100644 index 000000000..cc48a057f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-28 Compensation Breakdown.html @@ -0,0 +1,26 @@ + + + + + + + + + + + + + + + + + + + + + + + + + +
CategoryPercentage (%)
SY PSUs44
MY PSUs44
Variable Cash8
Base Salary4
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-29 Compensation Mix.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-29 Compensation Mix.html new file mode 100644 index 000000000..c9cb3b57f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-29 Compensation Mix.html @@ -0,0 +1,30 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
CategoryPercentage (%)
SY PSUs23
MY PSUs23
RSUs46
Variable Cash6
Base Salary2
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-3 NVIDIA Revenue.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-3 NVIDIA Revenue.html new file mode 100644 index 000000000..1e73e8d2d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-3 NVIDIA Revenue.html @@ -0,0 +1 @@ +
Compute & NetworkingGraphicsTotal
Revenue$193.5 billion (1)$22.5 billion (1)$215.9 billion
up 67% year on yearup 57% year on yearup 65% year on year
Operating Income (2)$130.1 billion$9.2 billion$139.3 billion
up 57% year on yearup 80% year on yearup 58% year on year
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-30 Fiscal 2026 Goals.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-30 Fiscal 2026 Goals.html new file mode 100644 index 000000000..a53b8fd54 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-30 Fiscal 2026 Goals.html @@ -0,0 +1 @@ +
Fiscal 2026 RevenueFiscal 2026 Non-GAAP Operating Income (1)3-Year TSR Relative to S&P 500 (Fiscal 2024 to 2026) (2)
Stretch Compensation Plan Goal$160.0 billion (3)$96.0 billion (3)75th percentile
Performance Achievement for Period Ended Fiscal 2026$215.9 billion$137.3 billion100th Percentile (1,055%)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-31 Tech Companies.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-31 Tech Companies.html new file mode 100644 index 000000000..18f7eb1ab --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-31 Tech Companies.html @@ -0,0 +1 @@ +
Alphabet Inc. (GOOG)Meta Platforms, Inc. (META)
Amazon.com, Inc. (AMZN)Microsoft Corporation (MSFT)
Apple Inc. (AAPL)Netflix, Inc. (NFLX)
Broadcom Inc. (AVGO)Oracle Corporation (ORCL)
Cisco Systems, Inc. (CSCO)Qualcomm Incorporated (QCOM)
International Business Machines Corporation (IBM)Salesforce, Inc. (CRM)
Intel Corporation (INTC)Visa Inc. (V)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-32 Percentile Values.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-32 Percentile Values.html new file mode 100644 index 000000000..0e082677a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-32 Percentile Values.html @@ -0,0 +1,26 @@ + + + + + + + + + + + + + + + + + + + + + + + + + +
CategoryValue ($)
75th Percentile229.70
NVIDIA113.27
Median54.03
25th Percentile40.93
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-33 NVIDIA Value Data.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-33 NVIDIA Value Data.html new file mode 100644 index 000000000..f9f0fbc7f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-33 NVIDIA Value Data.html @@ -0,0 +1,26 @@ + + + + + + + + + + + + + + + + + + + + + + + + + +
CategoryValue ($)
NVIDIA3,471.92
75th Percentile1,913.24
Median536.35
25th Percentile242.70
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-34 NEO Compensation.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-34 NEO Compensation.html new file mode 100644 index 000000000..27afa6fd4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-34 NEO Compensation.html @@ -0,0 +1 @@ +
Fixed CompensationAt-Risk Compensation
Base SalaryVariable CashSY PSUsMY PSUsRSUs (1)
FormCashCashEquityEquityEquity
Who ReceivesNEOsNEOsNEOsNEOsNEOs (except our CEO, whose equity awards are solely PSUs)
Performance MeasureN/ARevenue (determines cash payout)Non-GAAP Operating Income (determines number of shares eligible to vest)TSR relative to the S&P 500 (determines number of shares eligible to vest)N/A
Performance PeriodN/A1 year1 year3 yearsN/A
Vesting PeriodN/AN/A4 years from grant3 years from grant4 years from grant
Vesting TermsN/AN/AIf at least Threshold goal achieved, 25% on approximately the 1-year anniversary of the grant date; 6.25% quarterly thereafterIf at least Threshold goal achieved, 100% on approximately the 3-year anniversary of the grant period6.25% vests quarterly from the grant date (2)
Timeframe EmphasizedAnnualAnnualLong-termLong-termLong-term
PurposeCompensate for expected day-to-day performanceReward for annual corporate financial performanceAlign with stockholder interests by linking NEO pay to annual operational performance and ongoing stock price performance during the vesting periodAlign with long-term stockholder interests by linking NEO pay to multi-year relative shareholder return and ongoing stock price performance during the vesting periodAlign with stockholder interests by linking NEO pay to ongoing stock price performance
Maximum Amount That Can Be EarnedN/A200% of target opportunity under our Variable Cash Plan150% of Mr. Huang's SY PSU target opportunity and 200% of our other NEOs' respective SY PSU target opportunityUltimate value delivered depends on stock price on date earned shares vest150% of Mr. Huang's MY PSU target opportunity and 200% of our other NEOs' respective MY PSU target opportunityUltimate value delivered depends on stock price on date earned shares vest100% of grantUltimate value delivered depends on stock price on date shares vest
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-35 Compensation Plans.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-35 Compensation Plans.html new file mode 100644 index 000000000..869f0aeea --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-35 Compensation Plans.html @@ -0,0 +1 @@ +
Variable Cash PlanSY PSUsMY PSUs
MetricRevenueNon-GAAP Operating IncomeTSR relative to the S&P 500
Timeframe1 year1 year3 years
CC's Rationale for MetricDrives value, contributes to Company's long-term successDrives value, contributes to Company's long-term successAligns directly with long-term shareholder value creation
Focuses on growth in new and existing marketsReflects our annual revenue generation and effective operating expense managementProvides comparison of our stock price performance, including dividends, against a capital market index in which we compete
Distinct, separate metric from Non-GAAP Operating IncomeDistinct, separate metrics from revenueRelative performance goal accounts for macroeconomic factors impacting the market
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-36 Compensation Plan.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-36 Compensation Plan.html new file mode 100644 index 000000000..f15fb6797 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-36 Compensation Plan.html @@ -0,0 +1 @@ +
Variable Cash PlanSY PSUsMY PSUs
Fiscal 2026 RevenuePayout as a % of Target Opportunity (1)Fiscal 2026 Non-GAAP Operating Income (2)Shares Eligible to Vest as a % of Target Opportunity (1)Fiscal 2026 to 2028 3-Year Relative TSR (3)Shares Eligible to Vest as a % of Target Opportunity (1)
Threshold$100.0 billion50%$46.0 billion50%25th percentile25%
Base Compensation Plan$130.0 billion100%$71.0 billion100%50th percentile100%
Stretch Compensation Plan$160.0 billion (4)200%$96.0 billion (4)CEO 150% Other NEOs 200%75th percentileCEO 150% Other NEOs 200%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-37 Performance Payouts.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-37 Performance Payouts.html new file mode 100644 index 000000000..dcd93df80 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-37 Performance Payouts.html @@ -0,0 +1 @@ +
PERFORMANCE ACHIEVEMENT AND PAYOUTS
Variable Cash PlanSY PSUsMY PSUs (1)
Performance Achievement for Period Ended Fiscal 2026 (2)$215.9 billion revenue$137.3 billion Non-GAAP Operating Income (3)100th percentile 3-year TSR relative to S&P 500 (1,055%)
Payout as % of Target Opportunity200%CEO 150% Other NEOs 200% (4)CEO 150% Other NEOs 200% (5)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-38 CEO Pay Summary.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-38 CEO Pay Summary.html new file mode 100644 index 000000000..686185d9a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-38 CEO Pay Summary.html @@ -0,0 +1 @@ +
President & CEOTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary1,500,000Flat with Fiscal 2025
Variable Cash3,000,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($6,000,000)
Cash4,500,000Flat with Fiscal 2025
SY PSUs15,500,000Up $1.8 million, or 13%, from Fiscal 2025 target, resulting in 119,607 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 150% of target opportunity (179,411 shares) becoming eligible to vest
MY PSUs15,500,000Up $1.8 million, or 13%, from Fiscal 2025 target, resulting in 119,607 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 150% of target opportunity (757,360 shares) becoming eligible to vest
Equity31,000,000Up $3.5 million, or 13%, from Fiscal 2025 target
TOTAL35,500,000Up $3.5 million, or 11%, from Fiscal 2025 target
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-39 EVP CFO Pay.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-39 EVP CFO Pay.html new file mode 100644 index 000000000..0af75399d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-39 EVP CFO Pay.html @@ -0,0 +1 @@ +
EVP & CFOTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary900,000Flat with Fiscal 2025
Variable Cash300,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($600,000)
Cash1,200,000Flat with Fiscal 2025
SY PSUs3,699,807Up $0.2 million, or 7%, from Fiscal 2025 target, resulting in 28,550 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (57,100 shares) becoming eligible to vest
MY PSUs3,699,807Up $0.2 million, or 7%, from Fiscal 2025 target, resulting in 28,550 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (49,560 shares) becoming eligible to vest
RSUs7,400,386Up $0.5 million, or 7%, from Fiscal 2025, resulting in 57,106 shares granted in Fiscal 2026
Equity14,800,000Up $1.0 million, or 7%, from Fiscal 2025 target
TOTAL16,000,000Up $1.0 million, or 7%, from Fiscal 2025 target
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-4 Returns by Period.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-4 Returns by Period.html new file mode 100644 index 000000000..5a5c3725e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-4 Returns by Period.html @@ -0,0 +1,22 @@ + + + + + + + + + + + + + + + + + + + + + +
PeriodValue (%)
1-Year32
3-Year822
5-Year1,349
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-40 EVP Compensation.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-40 EVP Compensation.html new file mode 100644 index 000000000..7141a3f84 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-40 EVP Compensation.html @@ -0,0 +1 @@ +
EVP, Worldwide Field OperationsTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary950,000Flat with Fiscal 2025
Variable Cash650,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($1,300,000)
Cash1,600,000Flat with Fiscal 2025
SY PSUs3,600,000Up $0.3 million, or 7%, from Fiscal 2025 target, resulting in 27,779 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (55,558 shares) becoming eligible to vest
MY PSUs3,600,000Up $0.3 million, or 7%, from Fiscal 2025 target, resulting in 27,779 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (47,720 shares) becoming eligible to vest
RSUs7,200,000Up $0.5 million, or 7%, from Fiscal 2025, resulting in 55,559 shares granted in Fiscal 2026
Equity14,400,000Up $1.0 million, or 7%, from Fiscal 2025 target
TOTAL16,000,000Up $1.0 million, or 7%, from Fiscal 2025 target
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-41 EVP Compensation.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-41 EVP Compensation.html new file mode 100644 index 000000000..37127ee19 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-41 EVP Compensation.html @@ -0,0 +1 @@ +
EVP, OperationsTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary850,000Flat with Fiscal 2025
Variable Cash250,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($500,000)
Cash1,100,000Flat with Fiscal 2025
SY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (57,488 shares) becoming eligible to vest
MY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (40,840 shares) becoming eligible to vest
RSUs7,450,000Up $1.3 million, or 20%, from Fiscal 2025, resulting in 57,489 shares granted in Fiscal 2025
Equity14,900,000Up $2.5 million, or 20%, from Fiscal 2025 target
TOTAL16,000,000Up $2.5 million, or 19%, from Fiscal 2025 target
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-42 Executive Pay.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-42 Executive Pay.html new file mode 100644 index 000000000..ce1413191 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-42 Executive Pay.html @@ -0,0 +1 @@ +
EVP, General Counsel & SecretaryTarget Pay ($)Fiscal 2026 Compensation ActionsFiscal 2026 Performance-Based Payouts
Base Salary850,000Flat with Fiscal 2025
Variable Cash250,000Flat with Fiscal 2025Fiscal 2026 revenue exceeded Stretch Compensation Plan goal, resulting in 200% payout under Variable Cash Plan ($500,000)
Cash1,100,000Flat with Fiscal 2025
SY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2026 Non-GAAP Operating Income exceeded Stretch Compensation Plan goal, resulting in 200% of target opportunity (57,488 shares) becoming eligible to vest
MY PSUs3,725,000Up $0.6 million, or 20%, from Fiscal 2025 target, resulting in 28,744 shares target opportunity granted in Fiscal 2026Fiscal 2024 to Fiscal 2026 3-Year Relative TSR for MY PSUs granted in Fiscal 2024 achieved at Stretch Compensation Plan level, resulting in 200% of target opportunity (40,840 shares) becoming eligible to vest
RSUs7,450,000Up $1.3 million, or 20%, from Fiscal 2025, resulting in 57,489 shares granted in Fiscal 2025
Equity14,900,000Up $2.5 million, or 20%, from Fiscal 2025 target
TOTAL16,000,000Up $2.5 million, or 19%, from Fiscal 2025 target
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-43 Non-GAAP Income.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-43 Non-GAAP Income.html new file mode 100644 index 000000000..c4779819f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-43 Non-GAAP Income.html @@ -0,0 +1 @@ +
Fiscal 2026Fiscal 2025
GAAP operating income$130,387$81,453
Stock-based compensation expense6,3864,737
Acquisition-related and other costs471602
Other56(3)
Non-GAAP Operating Income$137,300$86,789
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-44 Executive Compensation.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-44 Executive Compensation.html new file mode 100644 index 000000000..48409b695 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-44 Executive Compensation.html @@ -0,0 +1 @@ +
Name and Principal PositionFiscal YearSalary ($)Stock Awards ($) (1)Non-Equity Incentive Plan Compensation ($) (2)All Other Compensation ($)Total ($) (3)
Jen-Hsun HuangPresident and CEO20261,497,62724,800,5116,000,0004,045,691 (4)36,343,830
20251,486,19938,811,3066,000,0003,568,74649,866,251
2024996,51426,676,4154,000,0002,494,97334,167,902
Colette M. KressEVP and CFO2026898,57712,825,872600,00016,402 (5)14,340,850
2025893,73919,849,891600,00018,90221,362,532
2024896,86311,756,027600,00013,90213,266,792
Ajay K. PuriEVP, Worldwide Field Operations2026948,49712,478,9891,300,00049,840 (5)14,777,327
2025943,39119,277,0461,300,00070,46021,590,897
2024946,68911,320,3531,300,00048,40813,615,450
Debora ShoquistEVP, Operations2026848,65612,912,487500,00033,567 (5)14,294,710
2025844,08717,838,832500,00034,98419,217,903
2024847,0379,687,599500,00024,22911,058,865
Timothy S. TeterEVP, General Counsel and Secretary2026848,65612,912,487500,00015,240 (5)14,276,382
2025844,08717,838,832500,00018,90219,201,821
2024847,0379,687,599500,00013,90211,048,538
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-45 PSU Awards Table.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-45 PSU Awards Table.html new file mode 100644 index 000000000..ed9d5cf0b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-45 PSU Awards Table.html @@ -0,0 +1 @@ +
Fiscal YearJen-Hsun HuangColette M. KressAjay K. PuriDebora ShoquistTimothy S. Teter
SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)SY PSU ($)MY PSU ($)
202619,166,42418,034,3436,099,9937,350,4835,935,2617,151,9816,141,4437,400,4306,141,4437,400,430
202525,046,74833,170,2118,377,78114,564,6908,136,36614,144,9937,529,33213,089,6697,529,33213,089,669
202423,130,93722,666,27015,613,1881,937,35015,034,8111,865,42312,866,3001,596,47612,866,3001,596,476
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-46 Executive Compensation.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-46 Executive Compensation.html new file mode 100644 index 000000000..7b8650146 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-46 Executive Compensation.html @@ -0,0 +1 @@ +
NameType of AwardGrant DateEstimated Possible Payouts Under Non-Equity Incentive Plan Awards (1)Estimated Future Payouts Under Equity Incentive Plan Awards (2)All Other Stock Awards: Number of Shares of Stock or Units (#) (3)Grant Date Fair Value of Stock Awards ($) (4)
Threshold ($)Target ($)Maximum ($)Threshold (#)Target (#)Maximum (#)
Jen-Hsun HuangSY PSU3/10/2559,804119,607179,41112,777,616
MY PSU3/10/2529,902119,607179,41112,022,896
Variable Cash Plan3/3/251,500,0003,000,0006,000,000
Colette M. KressSY PSU3/10/2514,27528,55057,1003,049,997
MY PSU3/10/257,13828,55057,1003,675,242
RSU3/10/2557,1066,100,634
Variable Cash Plan3/3/25150,000300,000600,000
Ajay K. PuriSY PSU3/10/2513,89027,77955,5582,967,631
MY PSU3/10/256,94527,77955,5583,575,991
RSU3/10/2555,5595,935,368
Variable Cash Plan3/3/25325,000650,0001,300,000
Debora ShoquistSY PSU3/10/2514,37228,74457,4883,070,722
MY PSU3/10/257,18628,74457,4883,700,215
RSU3/10/2557,4896,141,550
Variable Cash Plan3/3/25125,000250,000500,000
Timothy S. TeterSY PSU3/10/2514,37228,74457,4883,070,722
MY PSU3/10/257,18628,74457,4883,700,215
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-47 Stock Awards.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-47 Stock Awards.html new file mode 100644 index 000000000..5b52d80cd --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-47 Stock Awards.html @@ -0,0 +1 @@ +
NameStock Awards
Number of Units of Stock That Have Not Vested (#)Market Value of Units of Stock That Have Not Vested ($) (1)Equity Incentive Plan Awards: Number of Unearned Shares That Have Not Vested (#)Equity Incentive Plan Awards: Market Value of Unearned Shares That Have Not Vested ($) (1)
Jen-Hsun Huang315,570 (2)59,223,022
757,360 (3)142,133,751
161,085 (4)30,230,822
179,411 (5)33,670,062
286,350 (6)53,739,305
179,411 (7)33,670,062
Colette M. Kress10,970 (8)2,058,740
61,970 (9)11,629,910
213,010 (2)39,975,587
49,560 (3)9,300,925
53,890 (10)10,113,536
53,880 (4)10,111,660
46,399 (11)8,707,700
57,100 (5)10,715,957
95,780 (6)17,975,033
57,100 (7)10,715,957
Ajay K. Puri10,560 (8)1,981,795
59,670 (9)11,198,269
205,120 (2)38,494,870
47,720 (3)8,955,612
52,330 (10)9,820,771
52,340 (4)9,822,648
45,142 (11)8,471,799
55,558 (5)10,426,570
93,020 (6)17,457,063
55,558 (7)10,426,570
Debora Shoquist9,040 (8)1,696,537
51,070 (9)9,584,307
175,540 (2)32,943,592
40,840 (3)7,664,443
48,420 (10)9,086,981
48,420 (4)9,086,981
46,710 (11)8,766,066
57,488 (5)10,788,773
86,080 (6)16,154,634
57,488 (7)10,788,773
Timothy S. Teter9,040 (8)1,696,537
51,070 (9)9,584,307
175,540 (2)32,943,592
40,840 (3)7,664, 443
48,420 (10)9,086,981
48,420 (4)9,086,981
46,710 (11)8,766,066
57,48B (5)10,788,773
86,080 (6)16,154,634
57,488 (7)10,788,773
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-48 Stock Vesting Data.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-48 Stock Vesting Data.html new file mode 100644 index 000000000..8310247ed --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-48 Stock Vesting Data.html @@ -0,0 +1 @@ +
NameNumber of Shares Acquired on Vesting (#) (1)Value Realized on Vesting ($) (2)
Jen-Hsun Huang1,113,555 (3)141,311,217
Colette M. Kress442,087 (4)63,881,560
Ajay K. Puri425,837 (5)61,550,468
Debora Shoquist368,879 (6)53,412,792
Timothy S. Teter368,879 (7)53,412,792
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-49 Unvested RSUs_PSUs.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-49 Unvested RSUs_PSUs.html new file mode 100644 index 000000000..b4c86b4d3 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-49 Unvested RSUs_PSUs.html @@ -0,0 +1 @@ +
NameUnvested RSUs and PSUs at January 25, 2026 (#) (1)Total Estimated Benefit ($) (1)
Jen-Hsun Huang1,411,679264,929,798
Colette M. Kress569,889106,951,069
Ajay K. Puri551,090103,423,060
Debora Shoquist500,14893,862,775
Timothy S. Teter500,14893,862,775
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-5 Dividend & Buyback.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-5 Dividend & Buyback.html new file mode 100644 index 000000000..b39ef4584 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-5 Dividend & Buyback.html @@ -0,0 +1,26 @@ + + + + + + + + + + + + + + + + + + + + + + + + + +
MaturityDividends ($)Stock Repurchases ($)
1-Year040.4
3-Year2.284.1
5-Year3.094.2
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-50 SY PSU Grants.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-50 SY PSU Grants.html new file mode 100644 index 000000000..3399d8d5f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-50 SY PSU Grants.html @@ -0,0 +1 @@ +
NameEstimated SY PSUs Granted in Fiscal 2026 at Base Compensation Plan Performance (#)Value of Estimated SY PSUs Granted in Fiscal 2026 at Base Compensation Plan Performance ($)Actual SY PSUs Granted in Fiscal 2026 Eligible to Vest (#)Value of Actual SY PSUs Granted in Fiscal 2026 Eligible to Vest ($)
Jen-Hsun Huang119,60722,446,646179,41133,670,062
Colette M. Kress28,5505,357,97957,10010,715,957
Ajay K. Puri27,7795,213,28555,55810,426,570
Debora Shoquist28,7445,394,38657,48810,788,773
Timothy S. Teter28,7445,394,38657,48810,788,773
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-51 MY PSU Grants 2024.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-51 MY PSU Grants 2024.html new file mode 100644 index 000000000..e8b357639 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-51 MY PSU Grants 2024.html @@ -0,0 +1 @@ +
NameEstimated MY PSUs Granted in Fiscal 2024 at Base Compensation Plan Performance (#)Value of Estimated MY PSUs Granted in Fiscal 2024 at Base Compensation Plan Performance ($)Actual MY PSUs Granted in Fiscal 2024 Eligible to Vest (#)Value of Actual MY PSUs Granted in Fiscal 2024 Eligible to Vest ($)
Jen-Hsun Huang504,91094,756,460757,360142,133,751
Colette M. Kress24,7804,650,46349,5609,300,925
Ajay K. Puri23,8604,477,80647,7208,955,612
Debora Shoquist20,4203,832,22140,8407,664,443
Timothy S. Teter20,4203,832,22140,8407,664,443
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-52 CEO Pay & Performance.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-52 CEO Pay & Performance.html new file mode 100644 index 000000000..dade7591f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-52 CEO Pay & Performance.html @@ -0,0 +1 @@ +
Fiscal YearSummary Compensation Table Total for CEO ($) (1) (2)Compensation Actually Paid to CEO ($) (1) (3)Average Summary Compensation Table Total for Non-CEO NEOs ($) (4) (5)Average Compensation Actually Paid to Non-CEO NEOs ($) (4) (6)Value of Initial Fixed $100 Investment Based on (7):Net Income (in millions) ($)Non-GAAP Operating Income (in millions) ($) (9)
Total Shareholder Return ($)Peer Group Total Shareholder Return ($) (8)
202636,343,830162,180,93614,422,31757,721,5521,445.67198.10120,067137,300
202549,866,251344,188,02720,343,288125,583,9131,100.69168.4672,88086,789
202434,167,902234,132,30512,247,41185,558,057470.88134.7829,76037,134
202321,356,924(4,118,947)9,941,838(1,364,661)157.0594.134,3689,040
202223,737,661105,543,7688,910,80238,453,071175.98111.839,75212,690
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-53 CEO Compensation.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-53 CEO Compensation.html new file mode 100644 index 000000000..f8a9c013b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-53 CEO Compensation.html @@ -0,0 +1 @@ +
Fiscal YearSummary Compensation Table Total for CEO ($)(Deduct):Equity Award AdjustmentsTotal Equity Award Adjustments ($) (b) (c)Compensation Actually Paid to CEO ($)
Add:Add/(Deduct):Add/(Deduct):
Value of Equity Awards Reported in Summary Compensation Table ($) (a)Year End Fair Value of Awards Granted During the Year which were Unvested at Year End ($) (b)Year Over Year Change in Fair Value of Outstanding and Unvested Awards ($) (b)Change in Fair Value of Awards Granted in Prior Years which Vested During the Year ($) (b)
202636,343,830(24,800,511)60,909,207105,767,276(16,038,866)150,637,617162,180,936
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-54 Non-CEO NEO Pay.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-54 Non-CEO NEO Pay.html new file mode 100644 index 000000000..a7436feee --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-54 Non-CEO NEO Pay.html @@ -0,0 +1 @@ +
Fiscal YearAverage Summary Compensation Table Total for Non-CEO NEOs ($)(Deduct):Equity Award AdjustmentsTotal Equity Award Adjustments ($) (b) (c)Average Compensation Actually Paid to Non-CEO NEOs ($)
Add:Add/(Deduct):Add:Add/(Deduct):
Value of Equity Awards Reported in Summary Compensation Table ($) (a)Year End Fair Value of Awards Granted During the Year which were Unvested at Year End ($) (b)Year Over Year Change in Fair Value of Outstanding and Unvested Awards ($) (b)Vesting Date Fair Value of Awards Granted and Vested During the Year ($) (b)Change in Fair Value of Awards Granted in Prior Years which Vested During the Year ($) (b)
202614,422,317(12,782,459)27,759,18025,940,0531,776,820605,64056,081,69457,721,552
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-55 Financial Metrics.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-55 Financial Metrics.html new file mode 100644 index 000000000..f80c519c7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-55 Financial Metrics.html @@ -0,0 +1 @@ +
Revenue
Non-GAAP Operating Income
3-Year TSR relative to the S&P 500
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-56 NVIDIA CEO Pay.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-56 NVIDIA CEO Pay.html new file mode 100644 index 000000000..468737678 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-56 NVIDIA CEO Pay.html @@ -0,0 +1,48 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
Fiscal YearCEO CAP ($ in millions)Other NEOs Average CAP ($ in millions)NVIDIA TSR ($)Nasdaq100 Index TSR ($)
Fiscal 2022105.538.5111.83
Fiscal 2023(4.1)157.0594.13
Fiscal 2024234.185.6470.88134.78
Fiscal 2025344.2125.61,100.69168.46
Fiscal 2026162.257.71,445.67198.10
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-57 CEO Pay & Income.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-57 CEO Pay & Income.html new file mode 100644 index 000000000..f15ca8fc6 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-57 CEO Pay & Income.html @@ -0,0 +1,48 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
Fiscal YearCEO CAP ($ million)Other NEOs Average CAP ($ million)Net Income ($ million)Non-GAAP Operating Income ($ million)
Fiscal 2022105.538.59,75212,690
Fiscal 2023(4.1)(4.1)4,3689,040
Fiscal 2024234.185.629,76037,134
Fiscal 2025344.2125.672,88086,789
Fiscal 2026162.257.7120,067137,300
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-58 Fee Breakdown.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-58 Fee Breakdown.html new file mode 100644 index 000000000..f1ff442bd --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-58 Fee Breakdown.html @@ -0,0 +1 @@ +
Fiscal 2026Fiscal 2025
Audit Fees (1)$ 10,166,400$ 8,067,106
Audit Related Fees (2)1,716,820724,806
Tax Fees (3)1,076,103856,439
All Other Fees (4)401,999354,000
Total Fees$ 13,361,322$ 10,002,351
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-59 Equity Plan Table.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-59 Equity Plan Table.html new file mode 100644 index 000000000..9b5132ae5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-59 Equity Plan Table.html @@ -0,0 +1 @@ +
Plan CategoryNumber of securities to be issued upon exercise of outstanding options, warrants, and rights (a)Weighted-average exercise price of outstanding options, warrants, and rights (b)Number of securities remaining available for future issuance under equity compensation plans (excluding securities reflected in column (a)) (c)
Equity compensation plans approved by security holders (1)192,391,7653,551,257,925 (2)
Equity compensation plans not approved by security holders
Total192,391,7653,551,257,925
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-6 Proxy Voting Items.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-6 Proxy Voting Items.html new file mode 100644 index 000000000..830179b80 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-6 Proxy Voting Items.html @@ -0,0 +1 @@ +
MatterPageBoard RecommendsVote Required for ApprovalEffect of AbstentionsEffect of Broker Non-Votes
Management Proposals:
1Election of Ten Directors14FOR each director nomineeMore FOR than AGAINST votesNoneNone
2Advisory Approval of Our Executive Compensation37FORMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
3Ratification of the Selection of PwC as Our Independent Registered Public Accounting Firm for Fiscal 202758FORMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstN/A*
Stockholder Proposals:
4Simple Majority Vote Standard61AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
5Report on Faith-Based Community Resource Groups63AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
6Report on Workforce Civil Liberties65AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
7Greenhouse Gas Emissions Disclosure67AGAINSTMajority of shares present, in person or represented by proxy, and entitled to vote on this matterAgainstNone
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-60 Nvidia Executives.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-60 Nvidia Executives.html new file mode 100644 index 000000000..99a557db5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-60 Nvidia Executives.html @@ -0,0 +1 @@ +
NameAgePosition
Jen-Hsun Huang63President and Chief Executive Officer
Colette M. Kress58Executive Vice President and Chief Financial Officer
Ajay K. Puri71Executive Vice President, Worldwide Field Operations
Debora Shoquist71Executive Vice President, Operations
Timothy S. Teter59Executive Vice President and General Counsel
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-61 Stock Buyback.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-61 Stock Buyback.html new file mode 100644 index 000000000..42af2e3e5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-61 Stock Buyback.html @@ -0,0 +1 @@ +
PeriodTotal Number of Shares Purchased (In millions)Average Price Paid per Share (1)Total Number of Shares Purchased as Part of Publicly Announced Program (In millions)Approximate Dollar Value of Shares that May Yet Be Purchased Under the Program (In billions)
October 27, 2025 - November 23, 20255.6$ 198.895.6$ 61.1
November 24, 2025 - December 21, 20256.6$ 179.416.6$ 59.9
December 22, 2025 - January 25, 20267.3$ 186.527.3$ 58.5
Total19.519.5
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-62 NVIDIA Stock Growth.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-62 NVIDIA Stock Growth.html new file mode 100644 index 000000000..232e42596 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-62 NVIDIA Stock Growth.html @@ -0,0 +1,48 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
DateNVIDIA CorporationS&P 500Nasdaq 100
1/31/202110080100
1/30/2022180100120
1/29/20231609090
1/28/2024470120150
1/26/20251100170180
1/25/20261450200210
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-63 Stock Performance.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-63 Stock Performance.html new file mode 100644 index 000000000..a95ffbad5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-63 Stock Performance.html @@ -0,0 +1 @@ +
1/31/20211/30/20221/29/20231/28/20241/26/20251/25/2026
NVIDIA Corporation$ 100.00$ 175.98$ 157.05$ 470.88$ 1,100.68$ 1,448.75
S&P 500$ 100.00$ 121.00$ 112.98$ 137.98$ 174.50$ 200.33
Nasdaq 100$ 100.00$ 112.60$ 95.60$ 138.07$ 173.95$ 206.01
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-64 Financial Results.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-64 Financial Results.html new file mode 100644 index 000000000..a6c4df1cb --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-64 Financial Results.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Change
($ in millions, except per share data)
Revenue$ 215,938$ 130,497Up 65%
Gross margin71.1 %75.0 %-3.9 pts
Operating expenses$ 23,076$ 16,405Up 41%
Operating income$ 130,387$ 81,453Up 60%
Net income$ 120,067$ 72,880Up 65%
Net income per diluted share$ 4.90$ 2.94Up 67%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-65 Income Statement.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-65 Income Statement.html new file mode 100644 index 000000000..df076e78f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-65 Income Statement.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025
Revenue100.0 %100.0 %
Cost of revenue28.925.0
Gross profit71.175.0
Operating expenses
Research and development8.69.9
Sales, general and administrative2.12.7
Total operating expenses10.712.6
Operating income60.462.4
Interest income1.11.4
Interest expense(0.1)(0.2)
Other income, net4.20.8
Total other income, net5.22.0
Income before income tax65.564.4
Income tax expense9.98.6
Net income55.6 %55.8 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-66 Revenue Growth.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-66 Revenue Growth.html new file mode 100644 index 000000000..6ab2c2c4f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-66 Revenue Growth.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025$ Change% Change
($ in millions)
Compute & Networking$ 193,479$ 116,193$ 77,28667 %
Graphics22,45914,3048,15557 %
Total$ 215,938$ 130,497$ 85,44165 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-67 Revenue Growth.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-67 Revenue Growth.html new file mode 100644 index 000000000..7c7deb20b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-67 Revenue Growth.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025$ Change% Change
($ in millions)
Compute & Networking$ 130,141$ 82,875$ 47,26657 %
Graphics9,1565,0854,07180 %
Total$ 139,297$ 87,960$ 51,33758 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-68 Operating Expenses.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-68 Operating Expenses.html new file mode 100644 index 000000000..59f87cd37 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-68 Operating Expenses.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025$ Change% Change
($ in millions)
Research and development$ 18,497$ 12,914$ 5,58343 %
Sales, general and administrative4,5793,4911,08831 %
Total operating expenses$ 23,076$ 16,405$ 6,67141 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-69 Other Income.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-69 Other Income.html new file mode 100644 index 000000000..889910828 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-69 Other Income.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025$ Change
($ in millions)
Interest income$ 2,300$ 1,786$ 514
Interest expense(259)(247)(12)
Other income, net9,0221,0347,988
Total other income, net$ 11,063$ 2,573$ 8,490
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-7 Board of Directors.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-7 Board of Directors.html new file mode 100644 index 000000000..31cce88b8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-7 Board of Directors.html @@ -0,0 +1 @@ +
NameAgeDirector SinceIndependentFinancial Expert (1)Committee MembershipNumber of Other Public Company Boards
Tench Coxe681993AC, CC
John O. Dabiri462020CC, NCGC
Jen-Hsun Huang631993
Dawn Hudson682013CC Chairperson
Harvey C. Jones731993AC, NCGC
Melissa B. Lora632023AC, NCGC1
Stephen C. Neal(Lead Director)772019NCGC Chairperson
A. Brooke Seawell781997AC Chairperson1
Aarti Shah612020AC, CC1
Mark A. Stevens662008 (2)CC, NCGC
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-70 Cash & Securities.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-70 Cash & Securities.html new file mode 100644 index 000000000..9b55e29fc --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-70 Cash & Securities.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
(In millions)
Cash and cash equivalents$ 10,605$ 8,589
Marketable securities51,95134,621
Cash, cash equivalents, and marketable securities$ 62,556$ 43,210
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-71 Cash Flow Table.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-71 Cash Flow Table.html new file mode 100644 index 000000000..1cb228c9b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-71 Cash Flow Table.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
(In millions)
Net cash provided by operating activities$ 102,718$ 64,089
Net cash used in investing activities$ (52,228)$ (20,421)
Net cash used in financing activities$ (48,474)$ (42,359)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-72 Debt Maturity.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-72 Debt Maturity.html new file mode 100644 index 000000000..8a19d7f34 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-72 Debt Maturity.html @@ -0,0 +1 @@ +
Jan 25, 2026
(In millions)
Due in one year$ 1,000
Due in one to five years2,750
Due in five to ten years1,250
Due in greater than ten years3,500
Unamortized debt discount and issuance costs(32)
Net carrying amount$ 8,468
Less short-term portion999
Total long-term portion$ 7,469
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-73 Stock Sale Plans.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-73 Stock Sale Plans.html new file mode 100644 index 000000000..add2e5878 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-73 Stock Sale Plans.html @@ -0,0 +1 @@ +
NameTitle of Director or OfficerActionDateTotal Shares of Common Stock to be SoldExpiration Date
John O. DabiriDirectorAdoption12/10/20253,984*12/7/2026
Colette M. KressExecutive Vice President and Chief Financial OfficerAdoption12/18/2025500,0003/23/2027
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-74 Income Statement.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-74 Income Statement.html new file mode 100644 index 000000000..7622071f0 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-74 Income Statement.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Revenue$ 215,938$ 130,497$ 60,922
Cost of revenue62,47532,63916,621
Gross profit153,46397,85844,301
Operating expenses
Research and development18,49712,9148,675
Sales, general and administrative4,5793,4912,654
Total operating expenses23,07616,40511,329
Operating income130,38781,45332,972
Interest income2,3001,786866
Interest expense(259)(247)(257)
Other income, net9,0221,034237
Total other income, net11,0632,573846
Income before income tax141,45084,02633,818
Income tax expense21,38311,1464,058
Net income$ 120,067$ 72,880$ 29,760
Net income per share:
Basic$ 4.93$ 2.97$ 1.21
Diluted$ 4.90$ 2.94$ 1.19
Weighted average shares used in per share computation:
Basic24,35924,55524,690
Diluted24,51424,80424,940
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-75 Comprehensive Income.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-75 Comprehensive Income.html new file mode 100644 index 000000000..cf0893a97 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-75 Comprehensive Income.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Net income$ 120,067$ 72,880$ 29,760
Other comprehensive income, net of tax
Available-for-sale securities:
Net change in unrealized gain107180
Cash flow hedges:
Net change in unrealized gain (loss)43(10)
Other comprehensive income, net of tax150170
Total comprehensive income$ 120,217$ 72,881$ 29,830
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-76 Balance Sheet.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-76 Balance Sheet.html new file mode 100644 index 000000000..d2f8722f0 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-76 Balance Sheet.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
Assets
Current assets:
Cash and cash equivalents$ 10,605$ 8,589
Marketable securities51,95134,621
Accounts receivable, net38,46623,065
Inventories21,40310,080
Prepaid expenses and other current assets3,1803,771
Total current assets125,60580,126
Property and equipment, net10,3836,283
Operating lease assets2,8671,793
Goodwill20,8325,188
Intangible assets, net3,306807
Deferred income tax assets13,25810,979
Non-marketable equity securities22,2513,387
Other assets8,3013,038
Total assets$ 206,803$ 111,601
Liabilities and Shareholders' Equity
Current liabilities:
Accounts payable$ 9,812$ 6,310
Accrued and other current liabilities21,35211,737
Short-term debt999
Total current liabilities32,16318,047
Long-term debt7,4698,463
Long-term operating lease liabilities2,5721,519
Other long-term liabilities7,3064,245
Total liabilities49,51032,274
Commitments and contingencies - see Note 12
Shareholders' equity:
Preferred stock, $0.001 par value; 2 shares authorized; none issued
Common stock, $0.001 par value; 80,000 shares authorized; 24,304 shares issued and outstanding as of January 25, 2026; 24,477 shares issued and outstanding as of January 26, 20252424
Additional paid-in capital10,11811,237
Accumulated other comprehensive income17828
Retained earnings146,97368,038
Total shareholders' equity157,29379,327
Total liabilities and shareholders' equity$ 206,803$ 111,601
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-77 Equity Changes.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-77 Equity Changes.html new file mode 100644 index 000000000..d973992f6 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-77 Equity Changes.html @@ -0,0 +1 @@ +
Common Stock OutstandingAdditional Paid-in CapitalAccumulated Other Comprehensive Income (Loss)Retained EarningsTotal Shareholders' Equity
SharesAmount
(In millions, except per share data)
Balances as of Jan 29, 202324,661$ 25$ 11,948$ (43)$ 10,171$ 22,101
Net income29,76029,760
Other comprehensive income7070
Issuance of common stock265403403
Tax withholding related to common stock(72)(2,783)(2,783)
Shares repurchased(211)(27)(9,719)(9,746)
Cash dividends declared and paid ($0.016 per common share)(395)(395)
Stock-based compensation3,5683,568
Balances as of Jan 28, 202424,6432513,1092729,81742,978
Net income72,88072,880
Other comprehensive income11
Issuance of common stock203490490
Tax withholding related to common stock(59)(6,930)(6,930)
Shares repurchased(310)(1)(189)(33,825)(34,015)
Cash dividends declared and paid ($0.034 per common share)(834)(834)
Stock-based compensation4,7574,757
Balances as of Jan 26, 202524,4772411,2372868,03879,327
Net income120,067120,067
Other comprehensive income150150
Issuance of common stock160644644
Tax withholding related to common stock(51)(7,948)(7,948)
Shares repurchased(282)(230)(40,158)(40,388)
Cash dividends declared and paid ($0.04 per common share)(974)(974)
Fair value of partially vested equity awards assumed in connection with acquisitions2828
Stock-based compensation6,3876,387
Balances as of Jan 25, 202624,304$ 24$ 10,118$ 178$ 146,973$ 157,293
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-78 Cash Flow Statement.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-78 Cash Flow Statement.html new file mode 100644 index 000000000..6a8fa4bd9 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-78 Cash Flow Statement.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
Cash flows from operating activities:
Net income$120,067$72,880$29,760
Adjustments to reconcile net income to net cash provided by operating activities:
Stock-based compensation expense6,3864,7373,549
Depreciation and amortization2,8431,8641,508
Gains on non-marketable equity securities and publicly-held equity securities, net(8,918)(1,030)(238)
Deferred income taxes(1,424)(4,477)(2,489)
Other(287)(502)(278)
Changes in operating assets and liabilities, net of acquisitions:
Accounts receivable(15,399)(13,063)(6,172)
Inventories(11,324)(4,781)(98)
Prepaid expenses and other assets577(395)(1,522)
Accounts payable3,0963,3571,531
Accrued and other current liabilities5,2574,2782,025
Other long-term liabilities1,8441,221514
Net cash provided by operating activities102,71864,08928,090
Cash flows from investing activities:
Proceeds from sales of marketable securities15,15749550
Proceeds from maturities of marketable securities11,22611,1959,732
Proceeds from sales of non-marketable equity securities841711
Purchases of marketable securities(40,616)(26,575)(18,211)
Purchases of non-marketable equity securities(17,502)(1,486)(862)
Groq, Inc.(13,000)
Purchases related to property and equipment and intangible assets(6,042)(3,236)(1,069)
Acquisitions, net of cash acquired(1,535)(1,007)(83)
Other22(124)
Net cash used in investing activities(52,228)(20,421)(10,566)
Cash flows from financing activities:
Proceeds related to employee stock plans644490403
Payments related to repurchases of common stock(40,086)(33,706)(9,533)
Payments related to employee stock plan taxes(7,948)(6,930)(2,783)
Dividends paid(974)(834)(395)
Principal payments on property and equipment and intangible assets(101)(129)(74)
Repayment of debt(1,250)(1,250)
Other(9)(1)
Net cash used in financing activities(48,474)(42,359)(13,633)
Change in cash and cash equivalents2,0161,3093,891
Cash and cash equivalents at beginning of period8,5897,2803,389
Cash and cash equivalents at end of period$10,605$8,589$7,280
Supplemental disclosures of cash flow information:
Cash paid for income taxes, net$20,288$15,118$6,549
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-79 Cost Breakdown.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-79 Cost Breakdown.html new file mode 100644 index 000000000..44100c859 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-79 Cost Breakdown.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Cost of revenue$ 261$ 178$ 141
Research and development4,6763,4232,532
Sales, general and administrative1,4491,136876
Total$ 6,386$ 4,737$ 3,549
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-8 Board Skills Matrix.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-8 Board Skills Matrix.html new file mode 100644 index 000000000..4f1f5b6b8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-8 Board Skills Matrix.html @@ -0,0 +1 @@ +
Senior Leadership & Operations ExperienceIndustry & TechnicalFinancial / Financial CommunityGovernance & Public Company BoardEmerging Technologies & Business ModelsMarketing, Communications & Brand ManagementRegulatory, Legal & Risk ManagementHuman Capital Management Experience
Coxe
Dabiri
Huang
Hudson
Jones
Lora
Neal
Seawell
Shah
Stevens
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-80 Equity Awards.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-80 Equity Awards.html new file mode 100644 index 000000000..156060fba --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-80 Equity Awards.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions, except per share data)
RSUs, PSUs and Market-based PSUs
Awards granted7089140
Estimated total grant-date fair value$ 9,389$ 7,834$ 5,316
Weighted average grant-date fair value per share$ 133.97$ 87.99$ 37.41
ESPP
Shares purchased133030
Weighted average price per share$ 49.13$ 17.74$ 15.81
Weighted average grant-date fair value per share$ 20.75$ 8.61$ 6.99
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-81 ESPP Option Data.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-81 ESPP Option Data.html new file mode 100644 index 000000000..faad93d37 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-81 ESPP Option Data.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(Using the Black-Scholes model)
ESPP
Weighted average expected life (in years)0.1-2.00.1-2.00.1-2.0
Risk-free interest rate3.5%-4.3%3.6%-5.4%3.9%-5.5%
Volatility26%-96%31%-75%31%-67%
Dividend yield0.03%0.03%0.06%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-82 RSU_PSU Activity.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-82 RSU_PSU Activity.html new file mode 100644 index 000000000..e6da0eb97 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-82 RSU_PSU Activity.html @@ -0,0 +1 @@ +
RSUs, PSUs and Market-based PSUs Outstanding
Number of SharesWeighted Average Grant-Date Fair Value Per Share
(In millions, except per share data)
Balance as of Jan 26, 2025274$ 44.75
Granted70$ 133.97
Vested(146)$ 39.14
Canceled and forfeited(9)$ 59.29
Balance as of Jan 25, 2026189$ 81.51
Vested and expected to vest after Jan 25, 2026188$ 81.15
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-83 EPS Data.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-83 EPS Data.html new file mode 100644 index 000000000..c29780a69 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-83 EPS Data.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions, except per share data)
Numerator:
Net income$ 120,067$ 72,880$ 29,760
Denominator:
Basic weighted average shares24,35924,55524,690
Dilutive impact of outstanding equity awards155249250
Diluted weighted average shares24,51424,80424,940
Net income per share:
Basic (1)$ 4.93$ 2.97$ 1.21
Diluted (2)$ 4.90$ 2.94$ 1.19
Anti-dilutive equity awards excluded from diluted net income per share4151150
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-84 Intangible Assets.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-84 Intangible Assets.html new file mode 100644 index 000000000..5453d1f41 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-84 Intangible Assets.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
Gross Carrying AmountAccumulated AmortizationNet Carrying AmountGross Carrying AmountAccumulated AmortizationNet Carrying Amount
(In millions)
Acquisition-related intangible assets$ 5,656$ (2,580)$ 3,076$ 2,900$ (2,264)$ 636
Patents and licensed technology528(298)230449(278)171
Total intangible assets$ 6,184$ (2,878)$ 3,306$ 3,349$ (2,542)$ 807
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-85 Amortization Expense.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-85 Amortization Expense.html new file mode 100644 index 000000000..f5f13809d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-85 Amortization Expense.html @@ -0,0 +1 @@ +
Future Amortization Expense(In millions)
Fiscal Year:
2027$ 923
2028729
2029592
2030511
2031468
2032 and thereafter83
Total$ 3,306
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-86 Debt Securities.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-86 Debt Securities.html new file mode 100644 index 000000000..09e883ed5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-86 Debt Securities.html @@ -0,0 +1 @@ +
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable SecuritiesOther Assets
(In millions)
Debt securities issued by the U.S. TreasuryLevel 2$ 21,635$ 77$ (3)$ 21,709$ —$ 21,709$ —
Corporate debt securitiesLevel 215,41092(3)15,49934515,154
Debt securities issued by U.S. government agenciesLevel 22,15742,1612,161
Certificates of depositLevel 2110110110
Foreign government bondsLevel 24014141
Money market fundsLevel 17,8307,8307,830
Publicly-held equity securities (1) (2)Level 117,72612,8864,840
Total$ 47,182$ 174$ (6)$ 65,076$ 8,285$ 51,951$ 4,840
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-87 Fair Value Table.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-87 Fair Value Table.html new file mode 100644 index 000000000..799b0a6d3 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-87 Fair Value Table.html @@ -0,0 +1 @@ +
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable Securities
(In millions)
Corporate debt securitiesLevel 2$ 18,504$ 51$ (29)$ 18,526$ 2,071$ 16,455
Debt securities issued by the U.S. TreasuryLevel 216,74942(22)16,7691,80114,968
Debt securities issued by U.S. government agenciesLevel 22,7757(5)2,7772,777
Foreign government bondsLevel 217717713740
Certificates of depositLevel 2979797
Money market fundsLevel 13,7603,7603,760
Publicly-held equity securitiesLevel 1381381
Total$ 42,062$ 100$ (56)$ 42,487$ 7,866$ 34,621
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-88 Debt Securities.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-88 Debt Securities.html new file mode 100644 index 000000000..595aa1278 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-88 Debt Securities.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
Less than 12 MonthsLess than 12 Months
Estimated Fair ValueGross Unrealized LossEstimated Fair ValueGross Unrealized Loss
(In millions)
Debt securities issued by the U.S. Treasury$ 10,666$ (3)$ 6,315$ (22)
Corporate debt securities1,332(3)5,291(29)
Debt securities issued by U.S. government agencies1,134816(5)
Total$ 13,132$ (6)$ 12,422$ (56)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-89 Debt Maturity.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-89 Debt Maturity.html new file mode 100644 index 000000000..969981dd4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-89 Debt Maturity.html @@ -0,0 +1 @@ +
Jan 25, 2026
(In millions)
Less than one year$ 20,427
Due in 1 - 5 years19,093
Total$ 39,520
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-9 Performance Goals.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-9 Performance Goals.html new file mode 100644 index 000000000..3cb2557a7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-9 Performance Goals.html @@ -0,0 +1 @@ +
PERFORMANCE GOALS
Variable Cash PlanSY PSUsMY PSUs
Fiscal 2026 RevenuePayout as a % of Target OpportunityFiscal 2026 Non-GAAP Operating Income (1)Shares Eligible to Vest as a % of Target OpportunityFiscal 2024 to 2026 3-Year Relative TSRShares Eligible to Vest as a % of Target Opportunity
Threshold$100.0 billion50%$46.0 billion50%25th percentile25%
Base Compensation Plan$130.0 billion100%$71.0 billion100%50th percentile100%
Stretch Compensation Plan$160.0 billion (2)200%$96.0 billion (2)CEO 150%Other NEOs 200%75th percentileCEO 150%Other NEOs 200%
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-90 Equity Securities.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-90 Equity Securities.html new file mode 100644 index 000000000..5237a004a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-90 Equity Securities.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025
(In millions)
Balance at beginning of period$ 3,387$ 1,321
Adjustments related to non-marketable equity securities:
Net additions17,4441,309
Unrealized gains2,369816
Reclassification (1)(848)
Impairments and unrealized losses(101)(59)
Balance at end of period$ 22,251$ 3,387
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-91 Inventory Data.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-91 Inventory Data.html new file mode 100644 index 000000000..841eb2a36 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-91 Inventory Data.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
Inventories:(In millions)
Raw materials$ 3,807$ 3,408
Work in process8,8223,399
Finished goods8,7743,273
Total inventories (1)$ 21,403$ 10,080
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-92 Property & Equipment.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-92 Property & Equipment.html new file mode 100644 index 000000000..7513453af --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-92 Property & Equipment.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025EstimatedUseful Life
Property and Equipment:(In millions)(In years)
Land$ 777$ 511(A)
Buildings, leasehold improvements, and furniture2,8912,076(B)
Equipment, compute hardware, and software12,6197,5682-7
Construction in process683529(C)
Total property and equipment, gross16,97010,684
Accumulated depreciation and amortization(6,587)(4,401)
Total property and equipment, net$ 10,383$ 6,283
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-93 Current Liabilities.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-93 Current Liabilities.html new file mode 100644 index 000000000..31b850e6c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-93 Current Liabilities.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
Accrued and Other Current Liabilities:(In millions)
Customer program accruals$ 5,318$ 4,880
Accrued purchase consideration3,9219
Product warranty2,8071,290
Excess inventory purchase obligations (1)2,7392,095
Taxes payable2,669881
Deferred revenue (2)1,379837
Accrued payroll and related expenses1,146848
Other1,373897
Total accrued and other current liabilities$ 21,352$ 11,737
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-94 Long-Term Liabilities.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-94 Long-Term Liabilities.html new file mode 100644 index 000000000..4688e62d8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-94 Long-Term Liabilities.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
Other Long-Term Liabilities:(In millions)
Income tax payable (1)$ 3,958$ 2,188
Deferred income tax1,774886
Deferred revenue (2)1,193976
Other381195
Total other long-term liabilities$ 7,306$ 4,245
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-95 Deferred Revenue.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-95 Deferred Revenue.html new file mode 100644 index 000000000..3f096bb92 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-95 Deferred Revenue.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
(In millions)
Balance at beginning of period$ 1,813$ 1,337
Deferred revenue additions (1)11,1375,083
Revenue recognized (2)(10,378)(4,607)
Balance at end of period$ 2,572$ 1,813
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-96 Hedge Values.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-96 Hedge Values.html new file mode 100644 index 000000000..431d5b689 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-96 Hedge Values.html @@ -0,0 +1 @@ +
Jan 25, 2026Jan 26, 2025
(In millions)
Designated as accounting hedges$ 1,765$ 1,424
Not designated as accounting hedges$ 2,332$ 1,297
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-97 Debt Notes Table.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-97 Debt Notes Table.html new file mode 100644 index 000000000..054601c5d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-97 Debt Notes Table.html @@ -0,0 +1 @@ +
Expected Remaining Term (years)Effective Interest RateJan 25, 2026Jan 26, 2025
(In millions)
3.20% Notes Due 20260.63.31%1,0001,000
1.55% Notes Due 20282.41.64%1,2501,250
2.85% Notes Due 20304.22.93%1,5001,500
2.00% Notes Due 20315.42.09%1,2501,250
3.50% Notes Due 204014.23.54%1,0001,000
3.50% Notes Due 205024.23.54%2,0002,000
3.70% Notes Due 206034.23.73%500500
Unamortized debt discount and issuance costs(32)(37)
Net carrying amount8,4688,463
Less short-term portion(999)
Total long-term portion$ 7,469$ 8,463
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-98 Restructuring Reserve.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-98 Restructuring Reserve.html new file mode 100644 index 000000000..df6cda560 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-98 Restructuring Reserve.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Balance at beginning of period$ 1,290$ 306$ 82
Additions2,4741,203278
Utilization(957)(219)(54)
Balance at end of period$ 2,807$ 1,290$ 306
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-99 Income Tax Expense.html b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-99 Income Tax Expense.html new file mode 100644 index 000000000..54f7b77f4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/tables/table-99 Income Tax Expense.html @@ -0,0 +1 @@ +
Year Ended
Jan 25, 2026Jan 26, 2025Jan 28, 2024
(In millions)
Current income taxes:
Federal$ 19,039$ 14,032$ 5,710
State1,218892335
Foreign2,550699502
Total current22,80715,6236,547
Deferred income taxes:
Federal(1,364)(4,515)(2,499)
State(885)(242)(206)
Foreign825280216
Total deferred(1,424)(4,477)(2,489)
Income tax expense$ 21,383$ 11,146$ 4,058
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/toc_hierarchies.json b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/toc_hierarchies.json new file mode 100644 index 000000000..b6dc0e2d3 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-fy26-annual-report/toc_hierarchies.json @@ -0,0 +1,64 @@ +[ + { + "toc_range": [ + 370, + 416 + ], + "scan_range": [ + 370, + 531 + ], + "toc_with_level": "| id | heading | level |\n|----|--------------------------------------------------------------------------------------------------------------|-------|\n| 1 | Page | 1 |\n| 2 | DEFINITIONS 3 | 1 |\n| 3 | BUSINESS OVERVIEW 4 | 1 |\n| 4 | PROXY SUMMARY 6 | 1 |\n| 5 | PROXY STATEMENT 10 | 1 |\n| 6 | Information About the 2026 Meeting 10 | 2 |\n| 7 | Proposal 1—Election of Directors 14 | 2 |\n| 8 | Director Qualifications and Nomination of Directors 15 | 2 |\n| 9 | Our Director Nominees 17 | 2 |\n| 10 | Information About the Board of Directors and Corporate Governance 23 | 1 |\n| 11 | Independence of the Members of the Board of Directors 23 | 2 |\n| 12 | Board Leadership Structure 23 | 2 |\n| 13 | Committees of the Board of Directors 25 | 2 |\n| 14 | Compensation Committee Interlocks and Insider Participation 26 | 2 |\n| 15 | Role of the Board in Risk Oversight 27 | 2 |\n| 16 | Corporate Governance 28 | 2 |\n| 17 | Stockholder Communications with the Board of Directors 30 | 2 |\n| 18 | Majority Vote Standard 30 | 2 |\n| 19 | Stockholder Special Meeting Right 31 | 2 |\n| 20 | Board Meeting Information 31 | 2 |\n| 21 | Corporate Sustainability 31 | 2 |\n| 22 | Public Policy Engagement and Accountability 31 | 2 |\n| 23 | Director Compensation 32 | 2 |\n| 24 | Review of Transactions with Related Persons 34 | 2 |\n| 25 | Security Ownership of Certain Beneficial Owners and Management 35 | 2 |\n| 26 | Proposal 2—Advisory Approval of Executive Compensation 37 | 2 |\n| 27 | Executive Compensation 38 | 1 |\n| 28 | Compensation Discussion and Analysis 38 | 2 |\n| 29 | Compensation Committee Report 49 | 2 |\n| 30 | Risk Analysis of Our Compensation Plans 49 | 2 |\n| 31 | Summary Compensation Table for Fiscal 2026, 2025, and 2024 50 | 2 |\n| 32 | Grants of Plan-Based Awards for Fiscal 2026 51 | 2 |\n| 33 | Outstanding Equity Awards as of January 25, 2026 52 | 2 |\n| 34 | Option Exercises and Stock Vested in Fiscal 2026 53 | 2 |\n| 35 | Employment, Severance, and Change-in-Control Arrangements 53 | 2 |\n| 36 | Potential Payments Upon Termination or Change-in-Control 54 | 2 |\n| 37 | Pay Ratio 54 | 2 |\n| 38 | Pay Versus Performance 55 | 2 |\n| 39 | Proposal 3—Ratification of the Selection of Independent Registered Public Accounting Firm for Fiscal 2027 58 | 2 |\n| 40 | Fees Billed by the Independent Registered Public Accounting Firm 59 | 2 |\n| 41 | Report of the Audit Committee of the Board of Directors 60 | 2 |\n| 42 | Stockholder Proposals 61 | 2 |\n| 43 | Equity Compensation Plan Information 69 | 2 |\n| 44 | Additional Information 69 | 2 |\n| 45 | Delinquent Section 16(a) Reports 69 | 2 |\n| 46 | Other Matters 69 | 2 |", + "toc_tree": { + "Page": {}, + "DEFINITIONS 3": {}, + "BUSINESS OVERVIEW 4": {}, + "PROXY SUMMARY 6": {}, + "PROXY STATEMENT 10": { + "Information About the 2026 Meeting 10": {}, + "Proposal 1—Election of Directors 14": {}, + "Director Qualifications and Nomination of Directors 15": {}, + "Our Director Nominees 17": {} + }, + "Information About the Board of Directors and Corporate Governance 23": { + "Independence of the Members of the Board of Directors 23": {}, + "Board Leadership Structure 23": {}, + "Committees of the Board of Directors 25": {}, + "Compensation Committee Interlocks and Insider Participation 26": {}, + "Role of the Board in Risk Oversight 27": {}, + "Corporate Governance 28": {}, + "Stockholder Communications with the Board of Directors 30": {}, + "Majority Vote Standard 30": {}, + "Stockholder Special Meeting Right 31": {}, + "Board Meeting Information 31": {}, + "Corporate Sustainability 31": {}, + "Public Policy Engagement and Accountability 31": {}, + "Director Compensation 32": {}, + "Review of Transactions with Related Persons 34": {}, + "Security Ownership of Certain Beneficial Owners and Management 35": {}, + "Proposal 2—Advisory Approval of Executive Compensation 37": {} + }, + "Executive Compensation 38": { + "Compensation Discussion and Analysis 38": {}, + "Compensation Committee Report 49": {}, + "Risk Analysis of Our Compensation Plans 49": {}, + "Summary Compensation Table for Fiscal 2026, 2025, and 2024 50": {}, + "Grants of Plan-Based Awards for Fiscal 2026 51": {}, + "Outstanding Equity Awards as of January 25, 2026 52": {}, + "Option Exercises and Stock Vested in Fiscal 2026 53": {}, + "Employment, Severance, and Change-in-Control Arrangements 53": {}, + "Potential Payments Upon Termination or Change-in-Control 54": {}, + "Pay Ratio 54": {}, + "Pay Versus Performance 55": {}, + "Proposal 3—Ratification of the Selection of Independent Registered Public Accounting Firm for Fiscal 2027 58": {}, + "Fees Billed by the Independent Registered Public Accounting Firm 59": {}, + "Report of the Audit Committee of the Board of Directors 60": {}, + "Stockholder Proposals 61": {}, + "Equity Compensation Plan Information 69": {}, + "Additional Information 69": {}, + "Delinquent Section 16(a) Reports 69": {}, + "Other Matters 69": {} + } + } + } +] \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/chunks.json b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/chunks.json new file mode 100644 index 000000000..3c23587c3 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/chunks.json @@ -0,0 +1,1151 @@ +{ + "chunks": [ + { + "chunk_id": "76e4a0ad-c01f-55c9-b259-e9c1d1b2ad04", + "type": "text", + "content": "", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "07632b9b-0809-5a76-b2a7-97dd194211e7", + "type": "table", + "content": "
($ in millions, except earnings per share)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Revenue$81,615$68,127$44,06220 %85 %
Gross margin74.9 %75.0 %60.5 %(0.1) pts14.4 pts
Operating expenses$7,621$6,794$5,03012 %52 %
Operating income$53,536$44,299$21,63821 %147 %
Net income$58,321$42,960$18,77536 %211 %
Diluted earnings per share$2.39$1.76$0.7636 %214 %
", + "path": "tables/table-0 Q1 FY27 Results.html", + "metadata": { + "length": 788, + "summary": "table-1\nThe table shows Q1 FY27 financial results with revenue of $81,615M, net income of $58,321M, and diluted EPS of $2.39, reflecting strong Q/Q and Y/Y growth.", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "file_path": "tables/table-0 Q1 FY27 Results.html", + "keywords": [ + "Revenue", + "Net Income", + "EPS" + ], + "tokens": [] + } + }, + { + "chunk_id": "308faa82-fa7a-5390-b7ac-a34e2a0abfb6", + "type": "table", + "content": "
($ in millions, except earnings per share)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Revenue$81,615$68,127$44,06220 %85 %
Gross margin75.0 %75.1 %60.8 %(0.1) pts14.2 pts
Operating expenses$7,449$6,666$4,99312 %49 %
Operating income$53,783$44,474$21,80121 %147 %
Net income$45,548$38,969$19,09417 %139 %
Diluted earnings per share$1.87$1.59$0.7818 %140 %
", + "path": "tables/table-1 Q1 FY27 Results.html", + "metadata": { + "length": 788, + "summary": "table-2\nThe table shows Q1 FY27 financial results with revenue of $81,615M, gross margin 75.0%, operating income $53,783M, net income $45,548M, and diluted EPS $1.87, with strong Q/Q and Y/Y growth.", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "file_path": "tables/table-1 Q1 FY27 Results.html", + "keywords": [ + "Revenue", + "Gross Margin", + "Earnings" + ], + "tokens": [] + } + }, + { + "chunk_id": "1d52513f-325a-58af-b998-bd0745d212aa", + "type": "table", + "content": "
($ in millions)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Compute & Networking$74,550$61,651$39,58921 %88 %
Graphics7,0656,4764,4739 %58 %
Total$81,615$68,127$44,06220 %85 %
", + "path": "tables/table-2 Revenue Growth.html", + "metadata": { + "length": 426, + "summary": "table-3\nCompute & Networking revenue grew 21% Q/Q and 88% Y/Y to $74,550M. Graphics rose 9% Q/Q and 58% Y/Y to $7,065M. Total revenue increased 20% Q/Q and 85% Y/Y to $81,615M.", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "file_path": "tables/table-2 Revenue Growth.html", + "keywords": [ + "Revenue", + "Growth", + "Segments" + ], + "tokens": [] + } + }, + { + "chunk_id": "69b041b5-9e8f-5237-808b-455e174816d7", + "type": "table", + "content": "
($ in millions)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Data Center$75,246$62,314$39,11221 %92 %
Hyperscale37,86933,81417,59912 %115 %
AI Clouds, Industrial, & Enterprise37,37728,50021,51331 %74 %
Edge Computing6,3695,8134,95010 %29 %
Total$81,615$68,127$44,06220 %85 %
", + "path": "tables/table-3 Data Center Revenue.html", + "metadata": { + "length": 648, + "summary": "table-4\nData Center revenue reached $75,246M in Q1 FY27, up 21% Q/Q and 92% Y/Y, driven by Hyperscale and AI Clouds segments. Total revenue was $81,615M.", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "file_path": "tables/table-3 Data Center Revenue.html", + "keywords": [ + "Data Center", + "Revenue", + "Growth" + ], + "tokens": [] + } + }, + { + "chunk_id": "11f56754-6357-53f3-b87a-ebbaa7e9f5f4", + "type": "text", + "content": "GAAP\n\n[tables/table-0 Q1 FY27 Results.html]\n\nNon-GAAP\n\n[tables/table-1 Q1 FY27 Results.html]\n\nRevenue by Reportable Segments\n\n[tables/table-2 Revenue Growth.html]\n\nRevenue by Market Platform\n\n[tables/table-3 Data Center Revenue.html]\n\nWe specialize in markets where our computing platforms can provide tremendous acceleration for applications. These platforms incorporate processors, interconnects, software, algorithms, systems, and services to deliver unique value.\nFollowing the rapid evolution in our businesses, we are transitioning to a new reporting framework that better reflects our current and future growth drivers.\nWe will have two market platforms – Data Center and Edge Computing.\nWithin Data Center, we will report two sub-markets, Hyperscale and ACIE which incorporates AI Clouds, Industrial, and Enterprise. Hyperscale will include revenue from the public clouds and the world's largest consumer internet companies, while ACIE addresses our growth opportunity in diverse AI purpose-built data centers and AI factories across industries and countries.\nEdge Computing highlights devices for agentic and physical AI including PCs, game consoles, workstations, AI-RAN base stations, robotics and automotive.", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary", + "metadata": { + "length": 1361, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "tokens": [ + "GAAP", + "Non", + "Revenue", + "Reportable", + "Segments", + "Market", + "Platform", + "We", + "specialize", + "markets", + "computing", + "platforms", + "provide", + "tremendous", + "acceleration", + "applications", + "These", + "incorporate", + "processors", + "interconnects", + "software", + "algorithms", + "systems", + "services", + "deliver", + "unique", + "Following", + "rapid", + "evolution", + "businesses", + "transitioning", + "reporting", + "framework", + "reflects", + "current", + "future", + "growth", + "drivers", + "market", + "Data", + "Center", + "Edge", + "Computing", + "Within", + "report", + "Hyperscale", + "ACIE", + "incorporates", + "AI", + "Clouds", + "Industrial", + "Enterprise", + "include", + "revenue", + "public", + "clouds", + "world", + "largest", + "consumer", + "internet", + "companies", + "addresses", + "opportunity", + "diverse", + "purpose", + "built", + "data", + "centers", + "factories", + "industries", + "countries", + "highlights", + "devices", + "agentic", + "physical", + "including", + "PCs", + "game", + "consoles", + "workstations", + "RAN", + "base", + "stations", + "robotics", + "automotive" + ], + "keywords": [], + "connect_to": [ + { + "target": "07632b9b-0809-5a76-b2a7-97dd194211e7", + "relation": "embeds", + "ref": "[tables/table-0 Q1 FY27 Results.html]", + "position": { + "start": 6, + "end": 43 + } + }, + { + "target": "308faa82-fa7a-5390-b7ac-a34e2a0abfb6", + "relation": "embeds", + "ref": "[tables/table-1 Q1 FY27 Results.html]", + "position": { + "start": 55, + "end": 92 + } + }, + { + "target": "1d52513f-325a-58af-b998-bd0745d212aa", + "relation": "embeds", + "ref": "[tables/table-2 Revenue Growth.html]", + "position": { + "start": 126, + "end": 162 + } + }, + { + "target": "69b041b5-9e8f-5237-808b-455e174816d7", + "relation": "embeds", + "ref": "[tables/table-3 Data Center Revenue.html]", + "position": { + "start": 192, + "end": 233 + } + } + ] + } + }, + { + "chunk_id": "b0cd400a-f2ee-560d-8e11-fdaafe5ddda5", + "type": "text", + "content": "Revenue for the first quarter was a record \\$81.6 billion, up 85% from a year ago and up 20% sequentially.\nData Center revenue for the first quarter was a record \\$75.2 billion, up 92% from a year ago and up 21% sequentially, driven by the ramp of our Blackwell 300 products and demand for our InfiniBand, Spectrum-X™ Ethernet, and NVLink™ solutions. Hyperscale revenue increased sequentially and remained at approximately 50% of Data Center revenue, while the remaining 50% came from a continued diversification of customers, including AI Clouds, industrial, enterprise, and sovereign customers. No shipments of Data Center Hopper products to China occurred during the quarter, compared with \\$4.6 billion in the first quarter of fiscal year 2026.\nUnder the previous sub-markets, Data Center compute revenue was a record \\$60.4 billion, up 77% from a year ago and up 18% sequentially. Data Center networking revenue was a record \\$14.8 billion, up 199% from a year ago and up 35% sequentially.\nEdge Computing revenue for the first quarter was \\$6.4 billion, up 29% from a year ago and up 10% sequentially. The increases were driven by robust Blackwell workstation demand, partially offset by slower consumer PC demand that was tempered by elevated memory and systems prices.", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary/Revenue", + "metadata": { + "length": 1275, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "tokens": [ + "Revenue", + "quarter", + "record", + "81.6", + "billion", + "85%", + "year", + "ago", + "20%", + "sequentially", + "Data", + "Center", + "revenue", + "75.2", + "92%", + "21%", + "driven", + "ramp", + "Blackwell", + "300", + "products", + "demand", + "InfiniBand", + "Spectrum", + "Ethernet", + "NVLink", + "solutions", + "Hyperscale", + "increased", + "remained", + "approximately", + "50%", + "remaining", + "continued", + "diversification", + "customers", + "including", + "AI", + "Clouds", + "industrial", + "enterprise", + "sovereign", + "No", + "shipments", + "Hopper", + "China", + "occurred", + "compared", + "4.6", + "fiscal", + "2026", + "Under", + "previous", + "markets", + "compute", + "60.4", + "77%", + "18%", + "networking", + "14.8", + "199%", + "35%", + "Edge", + "Computing", + "6.4", + "29%", + "10%", + "The", + "increases", + "robust", + "workstation", + "partially", + "offset", + "slower", + "consumer", + "PC", + "tempered", + "elevated", + "memory", + "systems", + "prices" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b8b5e430-e545-52ef-a3c6-d90afd13bb93", + "type": "text", + "content": "GAAP and non-GAAP gross margins for the first quarter increased from a year ago on lower inventory provisions, primarily due to the prior year's \\$4.5 billion charge associated with H2O excess inventory and purchase obligations. GAAP and non-GAAP gross margins were approximately flat sequentially as our Blackwell architecture remains the majority of our revenue.", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary/Gross Margin", + "metadata": { + "length": 364, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "tokens": [ + "GAAP", + "gross", + "margins", + "quarter", + "increased", + "year", + "ago", + "lower", + "inventory", + "provisions", + "primarily", + "due", + "prior", + "4.5", + "billion", + "charge", + "H2O", + "excess", + "purchase", + "obligations", + "approximately", + "flat", + "sequentially", + "Blackwell", + "architecture", + "remains", + "majority", + "revenue" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8b09f7d4-92c7-504a-88e2-a50142c44373", + "type": "text", + "content": "GAAP operating expenses for the first quarter were up 52% from a year ago and up 12% sequentially, and non-GAAP operating expenses were up 49% from a year ago and up 12% sequentially. These increases were driven by higher compensation and benefits expense due to employee growth and compensation increases, compute and infrastructure costs, and engineering development materials for new product developments.", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary/Expenses", + "metadata": { + "length": 408, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "tokens": [ + "GAAP", + "operating", + "expenses", + "quarter", + "52%", + "year", + "ago", + "12%", + "sequentially", + "49%", + "These", + "increases", + "driven", + "higher", + "compensation", + "benefits", + "expense", + "due", + "employee", + "growth", + "compute", + "infrastructure", + "costs", + "engineering", + "development", + "materials", + "product", + "developments" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d68604c8-ade4-54ab-a564-cb4ca52fd4d1", + "type": "text", + "content": "GAAP other income and expense (OI&E) includes interest income, interest expense, and equity securities gains or losses. Non-GAAP OI&E excludes equity securities gains or losses.\nInterest income for the first quarter was \\$540 million, up slightly from a year ago and down slightly sequentially. Net gains from equity securities for the first quarter was \\$15.9 billion, driven by unrealized gains in publicly-held and non-marketable equity securities.\nGAAP effective tax rate for the first quarter was 16.6%, an increase from a year ago, primarily due to a lower impact from stock-based compensation tax benefits. Non-GAAP effective tax rate for the first quarter was 16.0%.", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary/Other Income & Expense and Income Tax", + "metadata": { + "length": 674, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "tokens": [ + "GAAP", + "income", + "expense", + "OI", + "includes", + "interest", + "equity", + "securities", + "gains", + "losses", + "Non", + "excludes", + "Interest", + "quarter", + "540", + "million", + "slightly", + "year", + "ago", + "sequentially", + "Net", + "15.9", + "billion", + "driven", + "unrealized", + "publicly", + "held", + "marketable", + "effective", + "tax", + "rate", + "16.6%", + "increase", + "primarily", + "due", + "lower", + "impact", + "stock", + "based", + "compensation", + "benefits", + "16.0%" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6f03e055-b7b8-5515-8611-20d16d739a0a", + "type": "text", + "content": "Cash, cash equivalents, and marketable debt securities were \\$50.3 billion, down from \\$52.4 billion a year ago and up from \\$49.7 billion a quarter ago. These changes primarily reflect higher revenue, offset by outlays for stock repurchases and strategic investments.\nAccounts receivable was \\$40.7 billion with 45 days sales outstanding (DSO), down from 51 days sequentially, driven by timing of cash collections and customer payments received prior to next quarter's invoice due dates. We expect our DSO to return to more normal levels next quarter.\nInventory was \\$25.8 billion, up from \\$21.4 billion sequentially, and total supply-related commitments were \\$119.0 billion. We have strategically secured inventory and capacity to meet demand beyond the next several quarters.\nMulti-year cloud service commitments were \\$30.0 billion, up from \\$27.0 billion sequentially, to support the growing needs of our research and development efforts.\nCash flow from operating activities was \\$50.3 billion, up from \\$27.4 billion a year ago and up from \\$36.2 billion a quarter ago. The year-on-year increase reflects growth in revenue, while the sequential increase was driven by higher revenue and lower cash taxes. We expect a substantial increase in cash taxes in the second quarter related to estimated federal and state cash tax payments.\nWe returned a record level of approximately \\$20.0 billion to shareholders in the first quarter through share repurchases and cash dividends.\nOn May 18, 2026, our Board of Directors approved an increase to our quarterly dividend from \\$0.01 per share to \\$0.25 per share and an additional \\$80.0 billion to our share repurchase authorization, without expiration.", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary/Balance Sheet and Cash Flow", + "metadata": { + "length": 1702, + "summary": "Cash, cash equivalents, and marketable debt securities were $50.3 billion, down from $52.4 billion a year ago and up from $49.7 billion a quarter ago. Accounts receivable was $40.7 billion with DSO down to 45 days from 51 days sequentially. Inventory rose to $25.8 billion from $21.4 billion sequentially. Multi-year cloud commitments increased to $30.0 billion. Operating cash flow was $50.3 billion, up from $27.4 billion a year ago. The company returned approximately $20.0 billion to shareholders via repurchases and dividends. The Board approved a dividend increase to $0.25 per share and an additional $80.0 billion for share repurchases.", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "tokens": [ + "Cash", + "cash", + "equivalents", + "marketable", + "debt", + "securities", + "50.3", + "billion", + "52.4", + "year", + "ago", + "49.7", + "quarter", + "These", + "primarily", + "reflect", + "higher", + "revenue", + "offset", + "outlays", + "stock", + "repurchases", + "strategic", + "investments", + "Accounts", + "receivable", + "40.7", + "45", + "days", + "sales", + "outstanding", + "DSO", + "51", + "sequentially", + "driven", + "timing", + "collections", + "customer", + "payments", + "received", + "prior", + "invoice", + "due", + "dates", + "We", + "expect", + "return", + "normal", + "levels", + "Inventory", + "25.8", + "21.4", + "total", + "supply", + "related", + "commitments", + "119.0", + "strategically", + "secured", + "inventory", + "capacity", + "meet", + "demand", + "quarters", + "Multi", + "cloud", + "service", + "30.0", + "27.0", + "support", + "growing", + "research", + "development", + "efforts", + "flow", + "operating", + "activities", + "27.4", + "36.2", + "The", + "increase", + "reflects", + "growth", + "sequential", + "lower", + "taxes", + "substantial", + "estimated", + "federal", + "state", + "tax", + "returned", + "record", + "level", + "approximately", + "20.0", + "shareholders", + "share", + "dividends", + "On", + "May", + "18", + "2026", + "Board", + "Directors", + "approved", + "quarterly", + "dividend", + "0.01", + "0.25", + "additional", + "80.0", + "repurchase", + "authorization", + "expiration" + ], + "keywords": [ + "Cash", + "Revenue", + "Shareholder Returns" + ], + "connect_to": [] + } + }, + { + "chunk_id": "5ed0d14d-8600-5fe5-94f3-b8547da170db", + "type": "text", + "content": "Outlook for the second quarter of fiscal 2027 is as follows:\n• Revenue is expected to be \\$91.0 billion, plus or minus 2%. We are not assuming any Data Center compute revenue from China in our outlook.\n- GAAP and non-GAAP gross margins are expected to be 74.9% and 75.0%, respectively, plus or minus 50 basis points.\n\\- GAAP and non-GAAP operating expenses are expected to be approximately \\$8.5 billion and \\$8.3 billion, respectively.\nFor the full year fiscal 2027, we expect GAAP and non-GAAP tax rates to be between 16.0% and 18.0%, excluding any discrete items and material changes to our tax environment.\nFor further information, contact:\nToshiya Hari\nInvestor Relations\nNVIDIA Corporation\ntoshiyah@nvidia.com\nMylene Mangalindan\nCorporate Communications\nNVIDIA Corporation\npress@nvidia.com", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Outlook", + "metadata": { + "length": 795, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "tokens": [ + "Outlook", + "quarter", + "fiscal", + "2027", + "Revenue", + "expected", + "91.0", + "billion", + "minus", + "2%", + "We", + "assuming", + "Data", + "Center", + "compute", + "revenue", + "China", + "outlook", + "GAAP", + "gross", + "margins", + "74.9%", + "75.0%", + "50", + "basis", + "points", + "operating", + "expenses", + "approximately", + "8.5", + "8.3", + "For", + "full", + "year", + "expect", + "tax", + "rates", + "16.0%", + "18.0%", + "excluding", + "discrete", + "items", + "material", + "environment", + "information", + "contact", + "Toshiya", + "Hari", + "Investor", + "Relations", + "NVIDIA", + "Corporation", + "toshiyah", + "nvidia", + "Mylene", + "Mangalindan", + "Corporate", + "Communications", + "press" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "35b72c32-0b98-5e6e-9dba-8a4a5be3c032", + "type": "text", + "content": "To supplement NVIDIA's condensed consolidated financial statements presented in accordance with GAAP, the company uses non-GAAP measures of certain components of financial performance. These non-GAAP measures include non-GAAP gross profit, non-GAAP gross margin, non-GAAP operating expenses, non-GAAP operating income, non-GAAP other income (expense), net, non-GAAP net income, non-GAAP net income, or earnings, per diluted share, and free cash flow. For NVIDIA's investors to be better able to compare its current results with those of previous periods, the company has shown a reconciliation of GAAP to non-GAAP financial measures. These reconciliations adjust the related GAAP financial measures to exclude acquisition-related and other costs, other, gains/losses from equity securities, net, certain other income and expense, and the associated tax impact of these items where applicable. Beginning in the first quarter of fiscal 2027, NVIDIA's non-GAAP financial measures no longer exclude stock-based compensation expense. The historical non-GAAP financial information presented has been updated to include stock-based compensation expense. Free cash flow is calculated as GAAP net cash provided by operating activities less both purchases related to property and equipment and intangible assets and principal payments on property and equipment and intangible assets. NVIDIA believes the presentation of its non-GAAP financial measures enhances the users' overall understanding of the company's historical financial performance. The presentation of the company's non-GAAP financial measures is not meant to be considered in isolation or as a substitute for the company's financial results prepared in accordance with GAAP, and the company's non-GAAP measures may be different from non-GAAP measures used by other companies.\nCertain statements in this CFO Commentary including, but not limited to, statements as to: expectations with respect to growth, performance and benefits of our products, services, and technologies, including Blackwell, and related trends and drivers; expectations with respect to supply and demand for our products, services, and technologies, including Blackwell, and related matters including inventory, production and distribution; expectations with respect to our third party arrangements, including with its collaborators and partners; expectations with respect to technology developments, and related trends and drivers; our future cash dividends or other returns to stockholders, our financial and business outlook for the second quarter of fiscal 2027 and beyond; projected market growth and trends; expectations with respect to AI and related industries; and other statements that are not historical facts are forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the “safe harbor” created by those sections based on management’s beliefs and assumptions and on information currently available to management and are subject to risks and uncertainties that could cause results to be materially different than expectations. Important factors that could cause actual results to differ materially include: global economic and political conditions; our reliance on third parties to manufacture, assemble, package and test our products; the impact of technological development and competition; development of new products and technologies or enhancements to our existing products and technologies; market acceptance of our products or our partners’ products; design, manufacturing or software defects; changes in consumer preferences or demands; changes in industry standards and interfaces; unexpected loss of performance of our products or technologies when integrated into systems; our ability to realize the potential benefits of business investments or acquisitions; and changes in applicable laws and regulations, as well as other factors detailed from time to time in the most recent reports NVIDIA files with the Securities and Exchange Commission, or SEC, including, but not limited to, its Annual Report on Form 10-K and Quarterly Reports on Form 10-Q. Copies of reports filed with the SEC are posted on the company's website and are available from NVIDIA without charge. These forward-looking statements are not guarantees of future performance and speak only as of the date hereof, and, except as required by law, NVIDIA disclaims any obligation to update these forward-looking statements to reflect future events or circumstances.", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Non-GAAP Measures", + "metadata": { + "length": 4614, + "summary": "NVIDIA uses non-GAAP financial measures like non-GAAP gross profit, margin, operating income, net income, and free cash flow to supplement GAAP results. These measures exclude acquisition-related costs, equity securities gains/losses, and other items. Starting fiscal 2027, stock-based compensation is no longer excluded. Free cash flow is calculated from GAAP operating cash flow minus property, equipment, and intangible asset purchases. The document also includes forward-looking statements about products like Blackwell, market trends, and risks such as global conditions, third-party reliance, and competition.", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "tokens": [ + "To", + "supplement", + "NVIDIA", + "condensed", + "consolidated", + "financial", + "statements", + "presented", + "accordance", + "GAAP", + "company", + "measures", + "components", + "performance", + "These", + "include", + "gross", + "profit", + "margin", + "operating", + "expenses", + "income", + "expense", + "net", + "earnings", + "diluted", + "share", + "free", + "cash", + "flow", + "For", + "investors", + "compare", + "current", + "results", + "previous", + "periods", + "shown", + "reconciliation", + "reconciliations", + "adjust", + "related", + "exclude", + "acquisition", + "costs", + "gains", + "losses", + "equity", + "securities", + "tax", + "impact", + "items", + "applicable", + "Beginning", + "quarter", + "fiscal", + "2027", + "longer", + "stock", + "based", + "compensation", + "The", + "historical", + "information", + "updated", + "Free", + "calculated", + "provided", + "activities", + "purchases", + "property", + "equipment", + "intangible", + "assets", + "principal", + "payments", + "believes", + "presentation", + "enhances", + "users", + "understanding", + "meant", + "considered", + "isolation", + "substitute", + "prepared", + "companies", + "Certain", + "CFO", + "Commentary", + "including", + "limited", + "expectations", + "respect", + "growth", + "benefits", + "products", + "services", + "technologies", + "Blackwell", + "trends", + "drivers", + "supply", + "demand", + "matters", + "inventory", + "production", + "distribution", + "party", + "arrangements", + "collaborators", + "partners", + "technology", + "developments", + "future", + "dividends", + "returns", + "stockholders", + "business", + "outlook", + "projected", + "market", + "AI", + "industries", + "facts", + "forward", + "meaning", + "Section", + "27A", + "Securities", + "Act", + "1933", + "amended", + "21E", + "Exchange", + "1934", + "subject", + "safe", + "harbor", + "created", + "sections", + "management", + "beliefs", + "assumptions", + "risks", + "uncertainties", + "materially", + "Important", + "factors", + "actual", + "differ", + "global", + "economic", + "political", + "conditions", + "reliance", + "parties", + "manufacture", + "assemble", + "package", + "test", + "technological", + "development", + "competition", + "enhancements", + "existing", + "acceptance", + "design", + "manufacturing", + "software", + "defects", + "consumer", + "preferences", + "demands", + "industry", + "standards", + "interfaces", + "unexpected", + "loss", + "integrated", + "systems", + "ability", + "realize", + "potential", + "investments", + "acquisitions", + "laws", + "regulations", + "detailed", + "time", + "recent", + "reports", + "files", + "Commission", + "SEC", + "Annual", + "Report", + "Form", + "10", + "Quarterly", + "Reports", + "Copies", + "filed", + "posted", + "website", + "charge", + "guarantees", + "speak", + "date", + "hereof", + "required", + "law", + "disclaims", + "obligation", + "update", + "reflect", + "events", + "circumstances" + ], + "keywords": [ + "Non-GAAP", + "Financial Measures", + "Forward-looking" + ], + "connect_to": [] + } + }, + { + "chunk_id": "3f5f3829-3a53-5a3e-8765-76c4701c0307", + "type": "text", + "content": "© 2026 NVIDIA Corporation. All rights reserved. NVIDIA, the NVIDIA logo, Spectrum-X, and NVLink are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and/or other countries. Other company and product names may be trademarks of the respective companies with which they are associated. Features, pricing, availability, and specifications are subject to change without notice.", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Non-GAAP Measures/", + "metadata": { + "length": 397, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "tokens": [ + "2026", + "NVIDIA", + "Corporation", + "All", + "rights", + "reserved", + "logo", + "Spectrum", + "NVLink", + "trademarks", + "registered", + "countries", + "Other", + "company", + "product", + "names", + "respective", + "companies", + "Features", + "pricing", + "availability", + "specifications", + "subject", + "change", + "notice" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "11b6d289-eac3-547a-80db-f8572c6de8b3", + "type": "table", + "content": "
Three Months Ended
April 26, 2026January 25, 2026April 27, 2025
GAAP cost of revenue$ 20,458$ 17,034$ 17,394
GAAP gross profit$ 61,157$ 51,093$ 26,668
GAAP gross margin74.9 %75.0 %60.5 %
Acquisition-related and other costs (A)4748123
Other28(1)3
Non-GAAP cost of revenue$ 20,383$ 16,987$ 17,268
Non-GAAP gross profit$ 61,232$ 51,140$ 26,794
Non-GAAP gross margin*75.0 %75.1 %60.8 %
GAAP operating expenses$ 7,621$ 6,794$ 5,030
Acquisition-related and other costs (A)(172)(90)(37)
Other(38)
Non-GAAP operating expenses$ 7,449$ 6,666$ 4,993
GAAP operating income$ 53,536$ 44,299$ 21,638
Total impact of non-GAAP adjustments to operating income247175163
Non-GAAP operating income*$ 53,783$ 44,474$ 21,801
GAAP total other income, net$ 16,367$ 6,098$ 272
(Gains) losses from equity securities, net(15,936)(5,491)175
Other (B)26131
Non-GAAP total other income, net$ 457$ 620$ 448
GAAP net income$ 58,321$ 42,960$ 18,775
Total pre-tax impact of non-GAAP adjustments(15,663)(5,303)339
Income tax impact of non-GAAP adjustments2,8901,312(20)
Non-GAAP net income*$ 45,548$ 38,969$ 19,094
", + "path": "tables/table-4 Non-GAAP Financials.html", + "metadata": { + "length": 2193, + "summary": "table-5\nThe table compares GAAP and Non-GAAP financial metrics for three fiscal quarters, showing cost of revenue, gross profit, operating income, and net income with adjustments.", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "file_path": "tables/table-4 Non-GAAP Financials.html", + "keywords": [ + "GAAP", + "Non-GAAP", + "Gross Margin" + ], + "tokens": [] + } + }, + { + "chunk_id": "c82b7b61-7c8f-56b7-bab5-d7454fc2bf3e", + "type": "table", + "content": "
Diluted net income per share
GAAP$ 2.39$ 1.76$ 0.76
Non-GAAP*$ 1.87$ 1.59$ 0.78
Weighted average shares used in diluted net income per share computation24,39124,43224,611
GAAP net cash provided by operating activities$ 50,344$ 36,190$ 27,414
Purchases related to property and equipment and intangible assets(1,757)(1,284)(1,227)
Principal payments on property and equipment and intangible assets(33)(4)(52)
Free cash flow$ 48,554$ 34,902$ 26,135
", + "path": "tables/table-5 Financial Data.html", + "metadata": { + "length": 861, + "summary": "table-6\nTable shows diluted net income per share (GAAP & Non-GAAP), weighted average shares, and free cash flow for three periods.", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "file_path": "tables/table-5 Financial Data.html", + "keywords": [ + "GAAP", + "Non-GAAP", + "Free Cash Flow" + ], + "tokens": [] + } + }, + { + "chunk_id": "b9ae4fd5-a302-5899-8910-8803075fbdb9", + "type": "table", + "content": "
Three Months Ended
April 26,2026January 25,2026April 27,2025
Cost of revenue$ 47$ 48$ 123
Research and development$ 167$ 83$ 28
Sales, general and administrative$ 5$ 7$ 9
", + "path": "tables/table-6 Cost Data Table.html", + "metadata": { + "length": 403, + "summary": "table-7\nThe table shows cost of revenue, R&D, and SG&A expenses for three months ending April 26, 2026, January 25, 2026, and April 27, 2025.", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "file_path": "tables/table-6 Cost Data Table.html", + "keywords": [ + "cost", + "revenue", + "expenses" + ], + "tokens": [] + } + }, + { + "chunk_id": "66080d3d-b6bd-515b-994a-2105c6e2759c", + "type": "table", + "content": "
Q2 FY2027 Outlook
($ in millions)
GAAP gross margin74.9 %
Impact of acquisition-related costs and other costs0.1 %
Non-GAAP gross margin75.0 %
GAAP operating expenses$ 8,500
Acquisition-related costs and other costs(200)
Non-GAAP operating expenses$ 8,300
", + "path": "tables/table-7 Q2 FY27 Outlook.html", + "metadata": { + "length": 482, + "summary": "table-8\nThe table shows Q2 FY2027 outlook with GAAP gross margin at 74.9%, Non-GAAP at 75.0%. GAAP operating expenses are $8,500M, Non-GAAP $8,300M after acquisition-related cost adjustments.", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "file_path": "tables/table-7 Q2 FY27 Outlook.html", + "keywords": [ + "gross margin", + "operating expenses", + "Non-GAAP" + ], + "tokens": [] + } + }, + { + "chunk_id": "178629b5-a569-5b2f-bfb5-86cc63cfac55", + "type": "text", + "content": "RECONCILIATION OF GAAP TO NON-GAAP FINANCIAL MEASURES\n(\\$ In millions, except per share data)\n(Unaudited)\n\n[tables/table-4 Non-GAAP Financials.html]\n\n\n[tables/table-5 Financial Data.html]\n\n\\*Includes H20 charges/(releases), net, which was \\$4.5 billion for the first quarter of fiscal 2026, insignificant for the fourth quarter of fiscal 2026, and none for the first quarter of fiscal 2027.\n(A) Acquisition-related and other costs are comprised of amortization of intangible assets, transaction costs, and certain compensation charges and are included in the following line items:\n\n[tables/table-6 Cost Data Table.html]\n\n(B) Comprised of interest expense related to acquisition consideration discount to be paid in the future, dividend income on equity securities, share of net (earnings)/losses related to equity method investments, and amortization of debt discount.\nNVIDIA CORPORATION\nRECONCILIATION OF GAAP TO NON-GAAP OUTLOOK\n\n[tables/table-7 Q2 FY27 Outlook.html]", + "path": "Q1FY27-CFO-Commentary.pdf/NVIDIA CORPORATION", + "metadata": { + "length": 982, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CFO Commentary on First Quarter Fiscal 2027 Results, NVIDIA CORPORATION", + "tokens": [ + "RECONCILIATION", + "OF", + "GAAP", + "TO", + "NON", + "FINANCIAL", + "MEASURES", + "In", + "millions", + "share", + "data", + "Unaudited", + "Includes", + "H20", + "charges", + "releases", + "net", + "4.5", + "billion", + "quarter", + "fiscal", + "2026", + "insignificant", + "fourth", + "2027", + "Acquisition", + "related", + "costs", + "comprised", + "amortization", + "intangible", + "assets", + "transaction", + "compensation", + "included", + "line", + "items", + "Comprised", + "interest", + "expense", + "acquisition", + "consideration", + "discount", + "paid", + "future", + "dividend", + "income", + "equity", + "securities", + "earnings", + "losses", + "method", + "investments", + "debt", + "NVIDIA", + "CORPORATION", + "OUTLOOK" + ], + "keywords": [], + "connect_to": [ + { + "target": "11b6d289-eac3-547a-80db-f8572c6de8b3", + "relation": "embeds", + "ref": "[tables/table-4 Non-GAAP Financials.html]", + "position": { + "start": 107, + "end": 148 + } + }, + { + "target": "c82b7b61-7c8f-56b7-bab5-d7454fc2bf3e", + "relation": "embeds", + "ref": "[tables/table-5 Financial Data.html]", + "position": { + "start": 151, + "end": 187 + } + }, + { + "target": "b9ae4fd5-a302-5899-8910-8803075fbdb9", + "relation": "embeds", + "ref": "[tables/table-6 Cost Data Table.html]", + "position": { + "start": 582, + "end": 619 + } + }, + { + "target": "66080d3d-b6bd-515b-994a-2105c6e2759c", + "relation": "embeds", + "ref": "[tables/table-7 Q2 FY27 Outlook.html]", + "position": { + "start": 932, + "end": 969 + } + } + ] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/doc_nav.json b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/doc_nav.json new file mode 100644 index 000000000..f89d82302 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/doc_nav.json @@ -0,0 +1,132 @@ +{ + "version": "1.0", + "file_name": "NVIDIA Q1 FY27 CFO Commentary.pdf", + "stats": { + "total_chunks": 19, + "text_chunks": 11, + "image_chunks": 0, + "table_chunks": 8, + "max_depth": 3 + }, + "sections": [ + { + "title": "CFO Commentary on First Quarter Fiscal 2027 Results", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results", + "level": 1, + "summary": "", + "chunk_count": 10, + "children": [ + { + "title": "Q1 Fiscal 2027 Summary", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary", + "level": 2, + "summary": "GAAP [tables/table-0 Q1 FY27 Results.html] Non-GAAP [tables/table-1 Q1 FY27 Results.html] Revenue by Reportable Segments [tables/table-2 Revenue Growth.html] Revenue by Market Platform [tables/table-3 Data Center Revenue.html] We specialize in markets where our computing platforms can provide tremendous acceleration for applications. These platforms incorporate processors, interconnects, software, algorithms, systems, and services to deliver unique value. Following the rapid evolution in our businesses, we are transitioning to a new reporting framework that better reflects our current and future growth drivers. We will have two market platforms – Data Center and Edge Computing. Within Data Center, we will report two sub-markets, Hyperscale and ACIE which incorporates AI Clouds, Industrial, and Enterprise. Hyperscale will include revenue from the public clouds and the world's largest consumer internet companies, while ACIE addresses our growth opportunity in diverse AI purpose-built data centers and AI factories across industries and countries. Edge Computing highlights devices for agentic and physical AI including PCs, game consoles, workstations, AI-RAN base stations, robotics and automotive.", + "chunk_count": 6, + "children": [ + { + "title": "Revenue", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary/Revenue", + "level": 3, + "summary": "Revenue for the first quarter was a record \\$81.6 billion, up 85% from a year ago and up 20% sequentially. Data Center revenue for the first quarter was a record \\$75.2 billion, up 92% from a year ago and up 21% sequentially, driven by the ramp of our Blackwell 300 products and demand for our InfiniBand, Spectrum-X™ Ethernet, and NVLink™ solutions. Hyperscale revenue increased sequentially and remained at approximately 50% of Data Center revenue, while the remaining 50% came from a continued diversification of customers, including AI Clouds, industrial, enterprise, and sovereign customers. No shipments of Data Center Hopper products to China occurred during the quarter, compared with \\$4.6 billion in the first quarter of fiscal year 2026. Under the previous sub-markets, Data Center compute revenue was a record \\$60.4 billion, up 77% from a year ago and up 18% sequentially. Data Center networking revenue was a record \\$14.8 billion, up 199% from a year ago and up 35% sequentially. Edge Computing revenue for the first quarter was \\$6.4 billion, up 29% from a year ago and up 10% sequentially. The increases were driven by robust Blackwell workstation demand, partially offset by slower consumer PC demand that was tempered by elevated memory and systems prices.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Gross Margin", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary/Gross Margin", + "level": 3, + "summary": "GAAP and non-GAAP gross margins for the first quarter increased from a year ago on lower inventory provisions, primarily due to the prior year's \\$4.5 billion charge associated with H2O excess inventory and purchase obligations. GAAP and non-GAAP gross margins were approximately flat sequentially as our Blackwell architecture remains the majority of our revenue.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Expenses", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary/Expenses", + "level": 3, + "summary": "GAAP operating expenses for the first quarter were up 52% from a year ago and up 12% sequentially, and non-GAAP operating expenses were up 49% from a year ago and up 12% sequentially. These increases were driven by higher compensation and benefits expense due to employee growth and compensation increases, compute and infrastructure costs, and engineering development materials for new product developments.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Other Income & Expense and Income Tax", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary/Other Income & Expense and Income Tax", + "level": 3, + "summary": "GAAP other income and expense (OI&E) includes interest income, interest expense, and equity securities gains or losses. Non-GAAP OI&E excludes equity securities gains or losses. Interest income for the first quarter was \\$540 million, up slightly from a year ago and down slightly sequentially. Net gains from equity securities for the first quarter was \\$15.9 billion, driven by unrealized gains in publicly-held and non-marketable equity securities. GAAP effective tax rate for the first quarter was 16.6%, an increase from a year ago, primarily due to a lower impact from stock-based compensation tax benefits. Non-GAAP effective tax rate for the first quarter was 16.0%.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Balance Sheet and Cash Flow", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Q1 Fiscal 2027 Summary/Balance Sheet and Cash Flow", + "level": 3, + "summary": "Cash, cash equivalents, and marketable debt securities were $50.3 billion, down from $52.4 billion a year ago and up from $49.7 billion a quarter ago. Accounts receivable was $40.7 billion with DSO down to 45 days from 51 days sequentially. Inventory rose to $25.8 billion from $21.4 billion sequentially. Multi-year cloud commitments increased to $30.0 billion. Operating cash flow was $50.3 billion, up from $27.4 billion a year ago. The company returned approximately $20.0 billion to shareholders via repurchases and dividends. The Board approved a dividend increase to $0.25 per share and an additional $80.0 billion for share repurchases.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Outlook", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Outlook", + "level": 2, + "summary": "Outlook for the second quarter of fiscal 2027 is as follows: • Revenue is expected to be \\$91.0 billion, plus or minus 2%. We are not assuming any Data Center compute revenue from China in our outlook. - GAAP and non-GAAP gross margins are expected to be 74.9% and 75.0%, respectively, plus or minus 50 basis points. \\- GAAP and non-GAAP operating expenses are expected to be approximately \\$8.5 billion and \\$8.3 billion, respectively. For the full year fiscal 2027, we expect GAAP and non-GAAP tax rates to be between 16.0% and 18.0%, excluding any discrete items and material changes to our tax environment. For further information, contact: Toshiya Hari Investor Relations NVIDIA Corporation toshiyah@nvidia.com Mylene Mangalindan Corporate Communications NVIDIA Corporation press@nvidia.com", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-GAAP Measures", + "path": "Q1FY27-CFO-Commentary.pdf/CFO Commentary on First Quarter Fiscal 2027 Results/Non-GAAP Measures", + "level": 2, + "summary": "NVIDIA uses non-GAAP financial measures like non-GAAP gross profit, margin, operating income, net income, and free cash flow to supplement GAAP results. These measures exclude acquisition-related costs, equity securities gains/losses, and other items. Starting fiscal 2027, stock-based compensation is no longer excluded. Free cash flow is calculated from GAAP operating cash flow minus property, equipment, and intangible asset purchases. The document also includes forward-looking statements about products like Blackwell, market trends, and risks such as global conditions, third-party reliance, and competition.", + "chunk_count": 2, + "children": [] + } + ] + }, + { + "title": "NVIDIA CORPORATION", + "path": "Q1FY27-CFO-Commentary.pdf/NVIDIA CORPORATION", + "level": 1, + "summary": "RECONCILIATION OF GAAP TO NON-GAAP FINANCIAL MEASURES (\\$ In millions, except per share data) (Unaudited) [tables/table-4 Non-GAAP Financials.html] [tables/table-5 Financial Data.html] \\*Includes H20 charges/(releases), net, which was \\$4.5 billion for the first quarter of fiscal 2026, insignificant for the fourth quarter of fiscal 2026, and none for the first quarter of fiscal 2027. (A) Acquisition-related and other costs are comprised of amortization of intangible assets, transaction costs, and certain compensation charges and are included in the following line items: [tables/table-6 Cost Data Table.html] (B) Comprised of interest expense related to acquisition consideration discount to be paid in the future, dividend income on equity securities, share of net (earnings)/losses related to equity method investments, and amortization of debt discount. NVIDIA CORPORATION RECONCILIATION OF GAAP TO NON-GAAP OUTLOOK [tables/table-7 Q2 FY27 Outlook.html]", + "chunk_count": 1, + "children": [] + } + ], + "resources": { + "images": [], + "tables": [ + { + "path": "tables/table-0 Q1 FY27 Results.html", + "summary": "table-1 The table shows Q1 FY27 financial results with revenue of $81,615M, net income of $58,321M, and diluted EPS of $2.39, reflecting strong Q/Q and Y/Y growth." + }, + { + "path": "tables/table-1 Q1 FY27 Results.html", + "summary": "table-2 The table shows Q1 FY27 financial results with revenue of $81,615M, gross margin 75.0%, operating income $53,783M, net income $45,548M, and diluted EPS $1.87, with strong Q/Q and Y/Y growth." + }, + { + "path": "tables/table-2 Revenue Growth.html", + "summary": "table-3 Compute & Networking revenue grew 21% Q/Q and 88% Y/Y to $74,550M. Graphics rose 9% Q/Q and 58% Y/Y to $7,065M. Total revenue increased 20% Q/Q and 85% Y/Y to $81,615M." + }, + { + "path": "tables/table-3 Data Center Revenue.html", + "summary": "table-4 Data Center revenue reached $75,246M in Q1 FY27, up 21% Q/Q and 92% Y/Y, driven by Hyperscale and AI Clouds segments. Total revenue was $81,615M." + }, + { + "path": "tables/table-4 Non-GAAP Financials.html", + "summary": "table-5 The table compares GAAP and Non-GAAP financial metrics for three fiscal quarters, showing cost of revenue, gross profit, operating income, and net income with adjustments." + }, + { + "path": "tables/table-5 Financial Data.html", + "summary": "table-6 Table shows diluted net income per share (GAAP & Non-GAAP), weighted average shares, and free cash flow for three periods." + }, + { + "path": "tables/table-6 Cost Data Table.html", + "summary": "table-7 The table shows cost of revenue, R&D, and SG&A expenses for three months ending April 26, 2026, January 25, 2026, and April 27, 2025." + }, + { + "path": "tables/table-7 Q2 FY27 Outlook.html", + "summary": "table-8 The table shows Q2 FY2027 outlook with GAAP gross margin at 74.9%, Non-GAAP at 75.0%. GAAP operating expenses are $8,500M, Non-GAAP $8,300M after acquisition-related cost adjustments." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/full.md b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/full.md new file mode 100644 index 000000000..b86e59705 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/full.md @@ -0,0 +1,135 @@ +## CFO Commentary on First Quarter Fiscal 2027 Results + +## Q1 Fiscal 2027 Summary + +GAAP + +
($ in millions, except earnings per share)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Revenue$81,615$68,127$44,06220 %85 %
Gross margin74.9 %75.0 %60.5 %(0.1) pts14.4 pts
Operating expenses$7,621$6,794$5,03012 %52 %
Operating income$53,536$44,299$21,63821 %147 %
Net income$58,321$42,960$18,77536 %211 %
Diluted earnings per share$2.39$1.76$0.7636 %214 %
+ +Non-GAAP + +
($ in millions, except earnings per share)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Revenue$81,615$68,127$44,06220 %85 %
Gross margin75.0 %75.1 %60.8 %(0.1) pts14.2 pts
Operating expenses$7,449$6,666$4,99312 %49 %
Operating income$53,783$44,474$21,80121 %147 %
Net income$45,548$38,969$19,09417 %139 %
Diluted earnings per share$1.87$1.59$0.7818 %140 %
+ +Revenue by Reportable Segments + +
($ in millions)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Compute & Networking$74,550$61,651$39,58921 %88 %
Graphics7,0656,4764,4739 %58 %
Total$81,615$68,127$44,06220 %85 %
+ +Revenue by Market Platform + +
($ in millions)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Data Center$75,246$62,314$39,11221 %92 %
Hyperscale37,86933,81417,59912 %115 %
AI Clouds, Industrial, & Enterprise37,37728,50021,51331 %74 %
Edge Computing6,3695,8134,95010 %29 %
Total$81,615$68,127$44,06220 %85 %
+ +We specialize in markets where our computing platforms can provide tremendous acceleration for applications. These platforms incorporate processors, interconnects, software, algorithms, systems, and services to deliver unique value. + +Following the rapid evolution in our businesses, we are transitioning to a new reporting framework that better reflects our current and future growth drivers. + +We will have two market platforms – Data Center and Edge Computing. + +Within Data Center, we will report two sub-markets, Hyperscale and ACIE which incorporates AI Clouds, Industrial, and Enterprise. Hyperscale will include revenue from the public clouds and the world's largest consumer internet companies, while ACIE addresses our growth opportunity in diverse AI purpose-built data centers and AI factories across industries and countries. + +Edge Computing highlights devices for agentic and physical AI including PCs, game consoles, workstations, AI-RAN base stations, robotics and automotive. + +## Revenue + +Revenue for the first quarter was a record \$81.6 billion, up 85% from a year ago and up 20% sequentially. + +Data Center revenue for the first quarter was a record \$75.2 billion, up 92% from a year ago and up 21% sequentially, driven by the ramp of our Blackwell 300 products and demand for our InfiniBand, Spectrum-X™ Ethernet, and NVLink™ solutions. Hyperscale revenue increased sequentially and remained at approximately 50% of Data Center revenue, while the remaining 50% came from a continued diversification of customers, including AI Clouds, industrial, enterprise, and sovereign customers. No shipments of Data Center Hopper products to China occurred during the quarter, compared with \$4.6 billion in the first quarter of fiscal year 2026. + +Under the previous sub-markets, Data Center compute revenue was a record \$60.4 billion, up 77% from a year ago and up 18% sequentially. Data Center networking revenue was a record \$14.8 billion, up 199% from a year ago and up 35% sequentially. + +Edge Computing revenue for the first quarter was \$6.4 billion, up 29% from a year ago and up 10% sequentially. The increases were driven by robust Blackwell workstation demand, partially offset by slower consumer PC demand that was tempered by elevated memory and systems prices. + +## Gross Margin + +GAAP and non-GAAP gross margins for the first quarter increased from a year ago on lower inventory provisions, primarily due to the prior year's \$4.5 billion charge associated with H2O excess inventory and purchase obligations. GAAP and non-GAAP gross margins were approximately flat sequentially as our Blackwell architecture remains the majority of our revenue. + +## Expenses + +GAAP operating expenses for the first quarter were up 52% from a year ago and up 12% sequentially, and non-GAAP operating expenses were up 49% from a year ago and up 12% sequentially. These increases were driven by higher compensation and benefits expense due to employee growth and compensation increases, compute and infrastructure costs, and engineering development materials for new product developments. + +## Other Income & Expense and Income Tax + +GAAP other income and expense (OI&E) includes interest income, interest expense, and equity securities gains or losses. Non-GAAP OI&E excludes equity securities gains or losses. + +Interest income for the first quarter was \$540 million, up slightly from a year ago and down slightly sequentially. Net gains from equity securities for the first quarter was \$15.9 billion, driven by unrealized gains in publicly-held and non-marketable equity securities. + +GAAP effective tax rate for the first quarter was 16.6%, an increase from a year ago, primarily due to a lower impact from stock-based compensation tax benefits. Non-GAAP effective tax rate for the first quarter was 16.0%. + +## Balance Sheet and Cash Flow + +Cash, cash equivalents, and marketable debt securities were \$50.3 billion, down from \$52.4 billion a year ago and up from \$49.7 billion a quarter ago. These changes primarily reflect higher revenue, offset by outlays for stock repurchases and strategic investments. + +Accounts receivable was \$40.7 billion with 45 days sales outstanding (DSO), down from 51 days sequentially, driven by timing of cash collections and customer payments received prior to next quarter's invoice due dates. We expect our DSO to return to more normal levels next quarter. + +Inventory was \$25.8 billion, up from \$21.4 billion sequentially, and total supply-related commitments were \$119.0 billion. We have strategically secured inventory and capacity to meet demand beyond the next several quarters. + +Multi-year cloud service commitments were \$30.0 billion, up from \$27.0 billion sequentially, to support the growing needs of our research and development efforts. + +Cash flow from operating activities was \$50.3 billion, up from \$27.4 billion a year ago and up from \$36.2 billion a quarter ago. The year-on-year increase reflects growth in revenue, while the sequential increase was driven by higher revenue and lower cash taxes. We expect a substantial increase in cash taxes in the second quarter related to estimated federal and state cash tax payments. + +We returned a record level of approximately \$20.0 billion to shareholders in the first quarter through share repurchases and cash dividends. + +On May 18, 2026, our Board of Directors approved an increase to our quarterly dividend from \$0.01 per share to \$0.25 per share and an additional \$80.0 billion to our share repurchase authorization, without expiration. + +## Outlook + +Outlook for the second quarter of fiscal 2027 is as follows: + +• Revenue is expected to be \$91.0 billion, plus or minus 2%. We are not assuming any Data Center compute revenue from China in our outlook. +- GAAP and non-GAAP gross margins are expected to be 74.9% and 75.0%, respectively, plus or minus 50 basis points. + +\- GAAP and non-GAAP operating expenses are expected to be approximately \$8.5 billion and \$8.3 billion, respectively. + +For the full year fiscal 2027, we expect GAAP and non-GAAP tax rates to be between 16.0% and 18.0%, excluding any discrete items and material changes to our tax environment. + +For further information, contact: + +Toshiya Hari + +Investor Relations + +NVIDIA Corporation + +toshiyah@nvidia.com + +Mylene Mangalindan + +Corporate Communications + +NVIDIA Corporation + +press@nvidia.com + +## Non-GAAP Measures + +To supplement NVIDIA's condensed consolidated financial statements presented in accordance with GAAP, the company uses non-GAAP measures of certain components of financial performance. These non-GAAP measures include non-GAAP gross profit, non-GAAP gross margin, non-GAAP operating expenses, non-GAAP operating income, non-GAAP other income (expense), net, non-GAAP net income, non-GAAP net income, or earnings, per diluted share, and free cash flow. For NVIDIA's investors to be better able to compare its current results with those of previous periods, the company has shown a reconciliation of GAAP to non-GAAP financial measures. These reconciliations adjust the related GAAP financial measures to exclude acquisition-related and other costs, other, gains/losses from equity securities, net, certain other income and expense, and the associated tax impact of these items where applicable. Beginning in the first quarter of fiscal 2027, NVIDIA's non-GAAP financial measures no longer exclude stock-based compensation expense. The historical non-GAAP financial information presented has been updated to include stock-based compensation expense. Free cash flow is calculated as GAAP net cash provided by operating activities less both purchases related to property and equipment and intangible assets and principal payments on property and equipment and intangible assets. NVIDIA believes the presentation of its non-GAAP financial measures enhances the users' overall understanding of the company's historical financial performance. The presentation of the company's non-GAAP financial measures is not meant to be considered in isolation or as a substitute for the company's financial results prepared in accordance with GAAP, and the company's non-GAAP measures may be different from non-GAAP measures used by other companies. + +Certain statements in this CFO Commentary including, but not limited to, statements as to: expectations with respect to growth, performance and benefits of our products, services, and technologies, including Blackwell, and related trends and drivers; expectations with respect to supply and demand for our products, services, and technologies, including Blackwell, and related matters including inventory, production and distribution; expectations with respect to our third party arrangements, including with its collaborators and partners; expectations with respect to technology developments, and related trends and drivers; our future cash dividends or other returns to stockholders, our financial and business outlook for the second quarter of fiscal 2027 and beyond; projected market growth and trends; expectations with respect to AI and related industries; and other statements that are not historical facts are forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the “safe harbor” created by those sections based on management’s beliefs and assumptions and on information currently available to management and are subject to risks and uncertainties that could cause results to be materially different than expectations. Important factors that could cause actual results to differ materially include: global economic and political conditions; our reliance on third parties to manufacture, assemble, package and test our products; the impact of technological development and competition; development of new products and technologies or enhancements to our existing products and technologies; market acceptance of our products or our partners’ products; design, manufacturing or software defects; changes in consumer preferences or demands; changes in industry standards and interfaces; unexpected loss of performance of our products or technologies when integrated into systems; our ability to realize the potential benefits of business investments or acquisitions; and changes in applicable laws and regulations, as well as other factors detailed from time to time in the most recent reports NVIDIA files with the Securities and Exchange Commission, or SEC, including, but not limited to, its Annual Report on Form 10-K and Quarterly Reports on Form 10-Q. Copies of reports filed with the SEC are posted on the company's website and are available from NVIDIA without charge. These forward-looking statements are not guarantees of future performance and speak only as of the date hereof, and, except as required by law, NVIDIA disclaims any obligation to update these forward-looking statements to reflect future events or circumstances. + +## ### + +© 2026 NVIDIA Corporation. All rights reserved. NVIDIA, the NVIDIA logo, Spectrum-X, and NVLink are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and/or other countries. Other company and product names may be trademarks of the respective companies with which they are associated. Features, pricing, availability, and specifications are subject to change without notice. + +## NVIDIA CORPORATION + +## RECONCILIATION OF GAAP TO NON-GAAP FINANCIAL MEASURES + +(\$ In millions, except per share data) + +(Unaudited) + +
Three Months Ended
April 26, 2026January 25, 2026April 27, 2025
GAAP cost of revenue$ 20,458$ 17,034$ 17,394
GAAP gross profit$ 61,157$ 51,093$ 26,668
GAAP gross margin74.9 %75.0 %60.5 %
Acquisition-related and other costs (A)4748123
Other28(1)3
Non-GAAP cost of revenue$ 20,383$ 16,987$ 17,268
Non-GAAP gross profit$ 61,232$ 51,140$ 26,794
Non-GAAP gross margin*75.0 %75.1 %60.8 %
GAAP operating expenses$ 7,621$ 6,794$ 5,030
Acquisition-related and other costs (A)(172)(90)(37)
Other(38)
Non-GAAP operating expenses$ 7,449$ 6,666$ 4,993
GAAP operating income$ 53,536$ 44,299$ 21,638
Total impact of non-GAAP adjustments to operating income247175163
Non-GAAP operating income*$ 53,783$ 44,474$ 21,801
GAAP total other income, net$ 16,367$ 6,098$ 272
(Gains) losses from equity securities, net(15,936)(5,491)175
Other (B)26131
Non-GAAP total other income, net$ 457$ 620$ 448
GAAP net income$ 58,321$ 42,960$ 18,775
Total pre-tax impact of non-GAAP adjustments(15,663)(5,303)339
Income tax impact of non-GAAP adjustments2,8901,312(20)
Non-GAAP net income*$ 45,548$ 38,969$ 19,094
+ +
Diluted net income per share
GAAP$ 2.39$ 1.76$ 0.76
Non-GAAP*$ 1.87$ 1.59$ 0.78
Weighted average shares used in diluted net income per share computation24,39124,43224,611
GAAP net cash provided by operating activities$ 50,344$ 36,190$ 27,414
Purchases related to property and equipment and intangible assets(1,757)(1,284)(1,227)
Principal payments on property and equipment and intangible assets(33)(4)(52)
Free cash flow$ 48,554$ 34,902$ 26,135
+ +\*Includes H20 charges/(releases), net, which was \$4.5 billion for the first quarter of fiscal 2026, insignificant for the fourth quarter of fiscal 2026, and none for the first quarter of fiscal 2027. + +(A) Acquisition-related and other costs are comprised of amortization of intangible assets, transaction costs, and certain compensation charges and are included in the following line items: + +
Three Months Ended
April 26,2026January 25,2026April 27,2025
Cost of revenue$ 47$ 48$ 123
Research and development$ 167$ 83$ 28
Sales, general and administrative$ 5$ 7$ 9
+ +(B) Comprised of interest expense related to acquisition consideration discount to be paid in the future, dividend income on equity securities, share of net (earnings)/losses related to equity method investments, and amortization of debt discount. + +NVIDIA CORPORATION +RECONCILIATION OF GAAP TO NON-GAAP OUTLOOK + +
Q2 FY2027 Outlook
($ in millions)
GAAP gross margin74.9 %
Impact of acquisition-related costs and other costs0.1 %
Non-GAAP gross margin75.0 %
GAAP operating expenses$ 8,500
Acquisition-related costs and other costs(200)
Non-GAAP operating expenses$ 8,300
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/manifest.json b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/manifest.json new file mode 100644 index 000000000..62d3c5fa5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/manifest.json @@ -0,0 +1,64 @@ +{ + "version": "2.0", + "job_id": "job_009c9fcfdf0b", + "data_id": null, + "source_file_name": "Q1FY27-CFO-Commentary.pdf", + "processing_date": "2026-06-16T03:20:39.687613Z", + "processing": { + "page_count": 8, + "billing_status": "charged", + "cost": { + "micro_dollars": 12000, + "credits": 0.012 + }, + "timing": { + "started_at": "2026-06-16T03:20:03.174662+00:00", + "completed_at": "2026-06-16T03:20:39.682343+00:00", + "duration_ms": 36507 + }, + "stages": { + "timing_ms": { + "document.profile": 3158, + "pdf.extract.standard": 24810, + "toc.detect_range_llm": 1033, + "md.detect_toc": 1047, + "heading.merge_pre_pass_llm": 1009, + "heading.hierarchy_llm_call": 2189, + "heading.hierarchy_llm": 2199, + "md.predict_headings": 3331, + "md.deferred_summaries": 3796, + "md.build_dataframe": 1, + "pdf.parse_md": 8257, + "document.cleanup_unreferenced_images": 0, + "document.compress_images": 0, + "worker.parse.document": 36232 + }, + "token_usage": { + "prompt_tokens": 3849, + "completion_tokens": 206, + "total_tokens": 4055 + } + } + }, + "statistics": { + "total_chunks": 19, + "text_chunks": 11, + "image_chunks": 0, + "table_chunks": 8, + "total_pages": null + }, + "HIERARCHY": { + "CFO Commentary on First Quarter Fiscal 2027 Results": { + "Q1 Fiscal 2027 Summary": { + "Revenue": {}, + "Gross Margin": {}, + "Expenses": {}, + "Other Income & Expense and Income Tax": {}, + "Balance Sheet and Cash Flow": {} + }, + "Outlook": {}, + "Non-GAAP Measures": {} + }, + "NVIDIA CORPORATION": {} + } +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-0 Q1 FY27 Results.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-0 Q1 FY27 Results.html new file mode 100644 index 000000000..090d3a3c2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-0 Q1 FY27 Results.html @@ -0,0 +1 @@ +
($ in millions, except earnings per share)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Revenue$81,615$68,127$44,06220 %85 %
Gross margin74.9 %75.0 %60.5 %(0.1) pts14.4 pts
Operating expenses$7,621$6,794$5,03012 %52 %
Operating income$53,536$44,299$21,63821 %147 %
Net income$58,321$42,960$18,77536 %211 %
Diluted earnings per share$2.39$1.76$0.7636 %214 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-1 Q1 FY27 Results.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-1 Q1 FY27 Results.html new file mode 100644 index 000000000..1864277ee --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-1 Q1 FY27 Results.html @@ -0,0 +1 @@ +
($ in millions, except earnings per share)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Revenue$81,615$68,127$44,06220 %85 %
Gross margin75.0 %75.1 %60.8 %(0.1) pts14.2 pts
Operating expenses$7,449$6,666$4,99312 %49 %
Operating income$53,783$44,474$21,80121 %147 %
Net income$45,548$38,969$19,09417 %139 %
Diluted earnings per share$1.87$1.59$0.7818 %140 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-2 Revenue Growth.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-2 Revenue Growth.html new file mode 100644 index 000000000..13c51410c --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-2 Revenue Growth.html @@ -0,0 +1 @@ +
($ in millions)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Compute & Networking$74,550$61,651$39,58921 %88 %
Graphics7,0656,4764,4739 %58 %
Total$81,615$68,127$44,06220 %85 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-3 Data Center Revenue.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-3 Data Center Revenue.html new file mode 100644 index 000000000..ed16d4049 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-3 Data Center Revenue.html @@ -0,0 +1 @@ +
($ in millions)Q1 FY27Q4 FY26Q1 FY26Q/QY/Y
Data Center$75,246$62,314$39,11221 %92 %
Hyperscale37,86933,81417,59912 %115 %
AI Clouds, Industrial, & Enterprise37,37728,50021,51331 %74 %
Edge Computing6,3695,8134,95010 %29 %
Total$81,615$68,127$44,06220 %85 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-4 Non-GAAP Financials.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-4 Non-GAAP Financials.html new file mode 100644 index 000000000..609472d99 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-4 Non-GAAP Financials.html @@ -0,0 +1 @@ +
Three Months Ended
April 26, 2026January 25, 2026April 27, 2025
GAAP cost of revenue$ 20,458$ 17,034$ 17,394
GAAP gross profit$ 61,157$ 51,093$ 26,668
GAAP gross margin74.9 %75.0 %60.5 %
Acquisition-related and other costs (A)4748123
Other28(1)3
Non-GAAP cost of revenue$ 20,383$ 16,987$ 17,268
Non-GAAP gross profit$ 61,232$ 51,140$ 26,794
Non-GAAP gross margin*75.0 %75.1 %60.8 %
GAAP operating expenses$ 7,621$ 6,794$ 5,030
Acquisition-related and other costs (A)(172)(90)(37)
Other(38)
Non-GAAP operating expenses$ 7,449$ 6,666$ 4,993
GAAP operating income$ 53,536$ 44,299$ 21,638
Total impact of non-GAAP adjustments to operating income247175163
Non-GAAP operating income*$ 53,783$ 44,474$ 21,801
GAAP total other income, net$ 16,367$ 6,098$ 272
(Gains) losses from equity securities, net(15,936)(5,491)175
Other (B)26131
Non-GAAP total other income, net$ 457$ 620$ 448
GAAP net income$ 58,321$ 42,960$ 18,775
Total pre-tax impact of non-GAAP adjustments(15,663)(5,303)339
Income tax impact of non-GAAP adjustments2,8901,312(20)
Non-GAAP net income*$ 45,548$ 38,969$ 19,094
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-5 Financial Data.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-5 Financial Data.html new file mode 100644 index 000000000..445456766 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-5 Financial Data.html @@ -0,0 +1 @@ +
Diluted net income per share
GAAP$ 2.39$ 1.76$ 0.76
Non-GAAP*$ 1.87$ 1.59$ 0.78
Weighted average shares used in diluted net income per share computation24,39124,43224,611
GAAP net cash provided by operating activities$ 50,344$ 36,190$ 27,414
Purchases related to property and equipment and intangible assets(1,757)(1,284)(1,227)
Principal payments on property and equipment and intangible assets(33)(4)(52)
Free cash flow$ 48,554$ 34,902$ 26,135
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-6 Cost Data Table.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-6 Cost Data Table.html new file mode 100644 index 000000000..95d4fd3da --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-6 Cost Data Table.html @@ -0,0 +1 @@ +
Three Months Ended
April 26,2026January 25,2026April 27,2025
Cost of revenue$ 47$ 48$ 123
Research and development$ 167$ 83$ 28
Sales, general and administrative$ 5$ 7$ 9
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-7 Q2 FY27 Outlook.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-7 Q2 FY27 Outlook.html new file mode 100644 index 000000000..355a6bc9d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-cfo-commentary/tables/table-7 Q2 FY27 Outlook.html @@ -0,0 +1 @@ +
Q2 FY2027 Outlook
($ in millions)
GAAP gross margin74.9 %
Impact of acquisition-related costs and other costs0.1 %
Non-GAAP gross margin75.0 %
GAAP operating expenses$ 8,500
Acquisition-related costs and other costs(200)
Non-GAAP operating expenses$ 8,300
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/chunks.json b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/chunks.json new file mode 100644 index 000000000..c7bbe3c4b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/chunks.json @@ -0,0 +1,3912 @@ +{ + "chunks": [ + { + "chunk_id": "854fafe3-55ad-52a5-8b48-3d7944fe437b", + "type": "text", + "content": "20-May-2026", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf", + "metadata": { + "length": 11, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "20", + "May", + "2026" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "60795bbb-72b3-5fd9-ad61-940e12a83b80", + "type": "text", + "content": "Q1 2027 Earnings Call", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)", + "metadata": { + "length": 21, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Q1", + "2027", + "Earnings", + "Call" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e3650088-d72a-5264-97b9-484bb0b262ef", + "type": "text", + "content": "Toshiya Hari\nVice President-Investor Relations & Strategic Finance, NVIDIA Corp.\nColette M. Kress\nChief Financial Officer & Executive Vice President, NVIDIA Corp.\nJen Hsun Huang\nCo-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)/CORPORATE PARTICIPANTS", + "metadata": { + "length": 249, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Toshiya", + "Hari", + "Vice", + "President", + "Investor", + "Relations", + "Strategic", + "Finance", + "NVIDIA", + "Corp", + "Colette", + "Kress", + "Chief", + "Financial", + "Officer", + "Executive", + "Jen", + "Hsun", + "Huang", + "Co", + "Founder", + "Director" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9b93eb44-3614-53b1-9fbb-701d38a763f1", + "type": "text", + "content": "Joseph Moore\nAnalyst, Morgan Stanley & Co. LLC\nBen Reitzes\nAnalyst, Melius Research LLC\nC.J. Muse\nAnalyst, Cantor Fitzgerald & Co.\nTimothy Arcuri\nAnalyst, UBS Securities LLC\nVivek Arya\nAnalyst, BofA Securities, Inc.\nStacy A. Rasgon\nAnalyst, Bernstein Research\nJames Edward Schneider\nAnalyst, Goldman Sachs & Co. LLC\nJoshua Buchalter\nAnalyst, TD Cowen", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)/OTHER PARTICIPANTS", + "metadata": { + "length": 350, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Joseph", + "Moore", + "Analyst", + "Morgan", + "Stanley", + "Co", + "LLC", + "Ben", + "Reitzes", + "Melius", + "Research", + "Muse", + "Cantor", + "Fitzgerald", + "Timothy", + "Arcuri", + "UBS", + "Securities", + "Vivek", + "Arya", + "BofA", + "Inc", + "Stacy", + "Rasgon", + "Bernstein", + "James", + "Edward", + "Schneider", + "Goldman", + "Sachs", + "Joshua", + "Buchalter", + "TD", + "Cowen" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "752ececb-5252-52eb-92a7-28b348cecdea", + "type": "text", + "content": "Operator: Good afternoon. My name is Sarah, and I will be your conference operator today. At this time, I would like to welcome everyone to NVIDIA's First Quarter Earnings Call. All lines have been placed on mute to prevent any background noise. After the speakers' remarks, there will be a question-and-answer session. [Operator Instructions] Thank you.\nToshiya Hari, you may begin your conference.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION", + "metadata": { + "length": 399, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Operator", + "Good", + "afternoon", + "My", + "Sarah", + "conference", + "operator", + "today", + "At", + "time", + "NVIDIA", + "First", + "Quarter", + "Earnings", + "Call", + "All", + "lines", + "mute", + "prevent", + "background", + "noise", + "After", + "speakers", + "remarks", + "question", + "answer", + "session", + "Instructions", + "Thank", + "Toshiya", + "Hari", + "begin" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "646678eb-f53d-53b7-85dc-7df3d7c84f82", + "type": "text", + "content": "Vice President-Investor Relations & Strategic Finance, NVIDIA Corp.\nThank you, and good afternoon, everyone. Welcome to NVIDIA's conference call for the first quarter of fiscal 2027. With me today from NVIDIA are Jensen Huang, President and Chief Executive Officer; and Colette Kress, Executive Vice President and Chief Financial Officer.\nOur call is being webcast live on NVIDIA's Investor Relations website. The webcast will be available for replay until the conference call to discuss our financial results for the second quarter of fiscal 2027. The content of today's call is NVIDIA's property. It can't be reproduced or transcribed, without our prior written consent.\nDuring this call, we may make forward-looking statements based on current expectations. These are subject to a number of significant risks and uncertainties, and our actual results may differ materially. For a discussion of factors that could affect our future financial results and business, please refer to the disclosure in today's earnings release, our most recent Forms 10-K and 10-Q, and the reports that we may file on Form 8-K with the Securities and Exchange Commission. All our statements are made as of today, May 20, 2026, based on information currently available to us. Except as required by law, we assume no obligation to update any such statements.\nDuring this call, we will discuss non-GAAP financial measures. You can find a reconciliation of these non-GAAP financial measures to GAAP financial measures in our CFO commentary, which is posted on our website.\nWith that, let me turn the call over to Colette.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Toshiya Hari", + "metadata": { + "length": 1598, + "summary": "NVIDIA held its Q1 fiscal 2027 earnings call on May 20, 2026, with CEO Jensen Huang and CFO Colette Kress. The call included forward-looking statements and discussion of non-GAAP financial measures. The webcast replay is available until the Q2 fiscal 2027 call.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Vice", + "President", + "Investor", + "Relations", + "Strategic", + "Finance", + "NVIDIA", + "Corp", + "Thank", + "good", + "afternoon", + "Welcome", + "conference", + "call", + "quarter", + "fiscal", + "2027", + "With", + "today", + "Jensen", + "Huang", + "Chief", + "Executive", + "Officer", + "Colette", + "Kress", + "Financial", + "Our", + "webcast", + "live", + "website", + "The", + "replay", + "discuss", + "financial", + "results", + "content", + "property", + "It", + "reproduced", + "transcribed", + "prior", + "written", + "consent", + "During", + "make", + "forward", + "statements", + "based", + "current", + "expectations", + "These", + "subject", + "number", + "significant", + "risks", + "uncertainties", + "actual", + "differ", + "materially", + "For", + "discussion", + "factors", + "affect", + "future", + "business", + "refer", + "disclosure", + "earnings", + "release", + "recent", + "Forms", + "10", + "reports", + "file", + "Form", + "Securities", + "Exchange", + "Commission", + "All", + "made", + "May", + "20", + "2026", + "information", + "Except", + "required", + "law", + "assume", + "obligation", + "update", + "GAAP", + "measures", + "You", + "find", + "reconciliation", + "CFO", + "commentary", + "posted", + "turn" + ], + "keywords": [ + "NVIDIA", + "earnings call", + "fiscal 2027" + ], + "connect_to": [] + } + }, + { + "chunk_id": "84762bed-4101-52ae-8f73-9e9b9d2afc05", + "type": "text", + "content": "Chief Financial Officer & Executive Vice President, NVIDIA Corp.\nThanks you, Toshiya. We delivered an exceptional quarter with revenue, operating income, and free cash flow exceeding our prior records. Total revenue of \\$82 billion was up 85% year-over-year and 20% sequentially. This marked our 3rd consecutive quarter of year-over-year acceleration and the 14th straight quarter of sequential growth, a significant feat given the sheer size and complexity of our manufacturing operations.\nThe \\$13.5 billion sequential revenue increase was also a record. We capitalized on the inflection in inference demand by ramping Blackwell systems across our diverse end customer base from hyperscalers to model makers to AI cloud providers and sovereign customers.\nIn Q1, we also allocated capital effectively across R&D, investments in our ecosystem, and share repurchases. We returned a record \\$20 billion to our shareholders, while executing strategic investments, both upstream supply chain and downstream go-to-market ecosystem. This is critical to the market's development and our long-term position.\nData Center revenue of \\$75 billion was up 92% year-over-year and 21% sequentially, driven by sustained strength in our Blackwell architecture. And demand for GB300 NVL72 was particularly strong with frontier model builders and hyperscalers, each having cumulatively deployed hundreds and thousands of Blackwell GPUs, marking the fastest product ramp in our company's history.\nGrace Blackwell is the fastest training system as well as the lowest token generation cost at inference. Spectrum-X, our end-to-end Ethernet platform purpose built for AI, is now larger than all Ethernet network peers combined. InfiniBand has also had a very strong quarter, growing more than 4x year-over-year, driven by deployments of our next-generation XDR technology.\nFor your models, Data Center computing revenue of \\$60 billion was up 77% year-over-year, while Data Center networking revenue of \\$15 billion nearly tripled year-over-year. Before we deep dive into Data Center, we'd like to brief you on our transition to a new reporting framework that better reflects our current and future growth drivers.\nWe have two market platforms: Data Center and Edge Computing. Within Data Center, we will report two sub-markets; Hyperscale and ACIE, which incorporates AI clouds, industrial and enterprise. Hyperscale will include revenue from the public cloud and the world's largest consumer Internet companies, while ACIE addresses our growth opportunities in diverse AI purpose-built data centers and AI factories across industries and countries. Edge Computing highlights devices for agentic and physical AI, including PCs, gaming consoles, workstations, AI-RAN base stations, robotics, and automotive. For your reference, we have posted on our website a revenue breakdown based on our new platforms for the past nine quarters.\nMoving back to our Data Center results. Hyperscale revenue of \\$38 billion was approximately 50% of Data Center revenue and increased 12% quarter-over-quarter. ACIE revenue was \\$37 billion and grew 31% quarter-over-quarter, including AI cloud revenue that more than tripled year-over-year.\nOur customers have enabled rapid stand-up of AI compute capacity. The number of partner data centers exceeding 10 megawatts has nearly doubled in just one year, now surpassing 80 sites. Sovereign revenue increased more than 80% year-over-year. NVIDIA AI infrastructure is now deployed across nearly 40 countries, representing \\$50 trillion in GDP.\nAs evident to our Q1 results, our customer base is diverse and growing. Supported by our vast ecosystem and installed base, breadth of CUDA accelerated application, and the lowest token cost provider, we are well positioned to address a market opportunity that far exceeds that of any other AI computing platform.\nDemand for AI infrastructure continues to expand at an unprecedented pace. The build-out of AI factories is accelerating. The value of NVIDIA AI infrastructure is rising. The price of renting an H100 has risen 20% year-to-date, while A100 cloud pricing is up nearly 15%. Benefiting from the versatility of our platform and continuous performance enhancements enhanced by our software stack, customers are generating profitable revenue beyond the depreciable life of their GPUs. The vast and trusted marketplace for NVIDIA Compute is a critical foundation on which billions in AI infrastructure spending is being financed by the ecosystem.\nThere are two primary drivers behind the accelerating build-out of AI infrastructure. First, from search and advertising to recommender systems and content understanding, the largest hyperscale workloads continue to transition from CPU to GPU-based accelerating computing.\nSecond, the adoption of products and services native to AI is inflecting. Since the advent of ChatGPT, we have witnessed mainstream AI transition from one-shot inference to reasoning and to now agentic. AI is no longer a nice to have. AI is now a necessity for enhancing productivity across all industries and roles. This is propelling revenue acceleration across all layers of the AI cake, including energy, chips, infrastructure, models, and applications. Growth in the model layer, particularly at Anthropic and OpenAI, has been incredible with momentum continuing to accelerate, including breakout growth in OpenAI's Codex since the launch of GPT-5.5.\nWith analysts now forecasting hyperscale CapEx to exceed \\$1 trillion in 2027 and agentic AI beginning to proliferate all industries, AI infrastructure spending is on track to reach \\$3 trillion to \\$4 trillion annually by the end of this decade.\nOur Blackwell architecture is everywhere, adopted and deployed by every major hyperscaler, every cloud provider, and every major model maker. Last month, we celebrated OpenAI's launch of GPT-5.5, co-designed for, trained with, and served on Blackwell, currently positioned at the top of Artificial Analysis leaderboards.\nMicrosoft's Fairwater, the world's most powerful AI data center, is now live, ahead of schedule, powered by hundreds of thousands of Blackwell GPUs. Starting this year, AWS will add more than 1 million Blackwell and Rubin GPUs, and are collaborating on Spectrum networking. At Google, Blackwell will be offered to customers in the cloud, including confidential computing capability, a new foundation for secure, high-performance AI.\nOur share of frontier AI compute is increasing. We have deepened our collaboration with Anthropic and are delighted to be a strategic partner to expand their compute capacity. We will support the company's growth trajectory through AWS, Azure, CoreWeave, SpaceXAI, and more. Now, with the addition of Anthropic to OpenAI,\nGemini, SpaceX xAI, Meta, MSL, Microsoft AI, TML, Reflection, Perplexity, Cursor, and other major frontier labs already building on NVIDIA, our share of frontier AI models will grow significantly.\nToday's data centers are revenue-generating AI factories. Constrained by power and capital, AI factory operators must choose the right architecture. With our extreme co-design approach, we deliver the industry's lowest token cost, the highest token throughput, and the highest ROI.\nMLPerf inference results are in and once again, we swept every benchmark as Blackwell Ultra delivered the highest throughput across the broad set of models and deployment scenarios. Full stack innovations drove the 2.7x increase in throughput and a 60% reduction in the cost per token on GB300 compared to just six months ago.\nNVIDIA Compute is not just the highest performance AI infrastructure, it is the most economic and financeable. Customers do not buy GPUs. They build AI factories. And the right economic metric is not the purchase price of the GPU, it is the lifetime cost of an AI factory producing intelligence, token per watt, tokens per dollar, uptime, utilization, time to production, software durability, and asset life. NVIDIA excels at all of them.\nAgentic AI and reinforcement learning represents new growth opportunities for CPUs. Building on the success of our Grace CPU, Vera is arriving just in time to meet this inflection. Built on custom Arm cores and co-designed end-to-end with Rubin GPUs and NVLink, Vera will deliver up to 1.5x faster performance per core, 2x performance per watt, and 4x density per rack, compared to x86-based alternatives.\nVera CPU opens a brand new \\$200 billion TAM for NVIDIA, a market we have never addressed before. And every major hyperscale and system maker is partnering with us to get it deployed. We have visibility to nearly \\$20 billion in total CPU revenue this year, setting us up to become the world-leading CPU supplier.\nOur annual product cadence, a pace that is unmatched, remains a key pillar supporting our market position. We are on track to commence production shipments of Vera Rubin in the second half of this year, starting in Q3. By integrating seven purpose-built chips across five accelerated racks, Vera Rubin will deliver up to 35x higher inference throughput and up to 10x greater AI factory revenue compared with Blackwell.\nAs an early adopter, Google's A5X bare-metal instances, which can support up to 960,000 Rubin GPUs across multiple sites, can enable customers to run their largest AI workloads on NVIDIA's optimized infrastructure.\nWhile the US government has approved licenses for H200 to be shipped to China-based customers, we have yet to generate any revenue, and we are uncertain whether any imports will be allowed into the country. As a result, consistent with last quarter, we are not including any China Data Center compute revenue in our outlook.\nLet me move to Edge Computing. Our Edge Computing market platform generated \\$6.4 billion, up 10% quarter-over-quarter and 29% year-over-year. Robust Blackwell workstation demand was a strong contributor to the growth, while consumer demand fell modestly due to higher memory and system prices.\nOur physical AI continues to gain momentum, exceeding \\$9 billion in revenue over the last 12 months. Our partnership with Uber will power the robotaxi fleet across nearly 30 cities and 4 continents by 2028. And in robotics, leading companies across a range of industrial, surgical, and humanoid applications are building on NVIDIA's technology to develop and deploy at scale.\nWe remain front-footed in securing sufficient supply to support our customers' growth. In Q1, we increased total supply, inclusive of inventory, purchase commitments, and prepaids to \\$145 billion. While we are not immune to supply challenges, we remain confident in our ability to support the growth opportunity ahead, with our intense focus, scale and longstanding partnerships with critical suppliers continuing to serve us well.\nLet me move to the rest of the P&L. GAAP gross margin was 74.9% and non-GAAP gross margin was 75%, largely flat sequentially by Blackwell systems continued to account for most of our shipments. GAAP and non-GAAP operating expenses were up 12% sequentially, primarily due to higher compensation and an increase in compute and infrastructure costs.\nOur non-GAAP effective tax rate of 16% came just below our prior outlook due to favorable geographic mix. And on our balance sheet, days sales outstanding was 45 days. Due to favorable timing of collections, we expect to return to the mid-50s in Q2. We generated record free cash flow \\$49 billion, up from \\$35 billion in Q4.\nI'd now like to update you on our capital allocation plan. First, to reiterate, our intention is to prioritize R&D and strategic investment, both will enable us to cultivate our ecosystem, drive market growth, and strengthen our market position. As a key enabler of AI, we will make investments necessary to deliver the industry's lowest cost per token and the highest token throughput, which will help our customers and partners scale and expand the AI frontier.\nReturn program is another key component of our capital allocation strategy. Given confidence in our long-term free cash flow outlook and our commitment to sharing our success with shareholders, we are increasing our quarterly dividend from \\$0.01 to \\$0.20 (sic) [\\$0.25] per share. We plan to review our dividend on a regular basis, as we continue to scale our business. We are also announcing an \\$80 billion share repurchase authorization, which is in addition to the \\$39 billion remaining on our current plan. As we indicated at GTC, we plan to return roughly 50% of free cash flow to shareholders this year.\nLet me turn to the outlook for the second quarter. Total revenue is expected to be \\$91 billion, plus or minus 2%. We expect sequential growth to be driven primarily by Data Center. We are continuing to work vigorously on our supply chain ecosystem to address the incredible demand we see ahead of us, giving us full confidence in the \\$1 trillion in Blackwell and Rubin revenue we foresee from 2025 through calendar 2027.\nGAAP and non-GAAP gross margins are expected to be 74.9% and 75%, respectively, plus or minus 50 basis points. For the full year, we are still expecting to be in the mid-70s.\nGAAP and non-GAAP operating expenses are expected to be approximately \\$8.5 billion and \\$8.3 billion, respectively. For the full year, we now expect OpEx to grow somewhere in the upper-40s on a year-over-year basis, driven by higher R&D and acceleration in the usage of AI tools to enhance productivity.\nFor the full year 2027, we expect GAAP and non-GAAP tax rates to be between 16% and 18%, excluding any discrete items from material changes to our tax environment. This is lower than our prior expectation of 17% to 19% due to changes in geographic mix.\nThat puts me at the end of this part, and I'm going to now turn this over to the Q&A with Toshiya.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Colette M. Kress", + "metadata": { + "length": 13789, + "summary": "NVIDIA reported record Q1 revenue of $82 billion, up 85% year-over-year, driven by Data Center revenue of $75 billion (up 92% YoY) led by Blackwell architecture. The company highlighted strong demand for GB300 NVL72, rapid AI infrastructure build-out, and a transition to a new reporting framework with Hyperscale and ACIE sub-markets. Key drivers include CPU-to-GPU transition and AI native product adoption. Edge Computing generated $6.4 billion. Q2 revenue outlook is $91 billion. The company announced an $80 billion share repurchase authorization and increased dividend to $0.25 per share.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Chief", + "Financial", + "Officer", + "Executive", + "Vice", + "President", + "NVIDIA", + "Corp", + "Thanks", + "Toshiya", + "We", + "delivered", + "exceptional", + "quarter", + "revenue", + "operating", + "income", + "free", + "cash", + "flow", + "exceeding", + "prior", + "records", + "Total", + "82", + "billion", + "85%", + "year", + "20%", + "sequentially", + "This", + "marked", + "3rd", + "consecutive", + "acceleration", + "14th", + "straight", + "sequential", + "growth", + "significant", + "feat", + "sheer", + "size", + "complexity", + "manufacturing", + "operations", + "The", + "13.5", + "increase", + "record", + "capitalized", + "inflection", + "inference", + "demand", + "ramping", + "Blackwell", + "systems", + "diverse", + "end", + "customer", + "base", + "hyperscalers", + "model", + "makers", + "AI", + "cloud", + "providers", + "sovereign", + "customers", + "In", + "Q1", + "allocated", + "capital", + "effectively", + "investments", + "ecosystem", + "share", + "repurchases", + "returned", + "20", + "shareholders", + "executing", + "strategic", + "upstream", + "supply", + "chain", + "downstream", + "market", + "critical", + "development", + "long", + "term", + "position", + "Data", + "Center", + "75", + "92%", + "21%", + "driven", + "sustained", + "strength", + "architecture", + "And", + "GB300", + "NVL72", + "strong", + "frontier", + "builders", + "cumulatively", + "deployed", + "hundreds", + "thousands", + "GPUs", + "marking", + "fastest", + "product", + "ramp", + "company", + "history", + "Grace", + "training", + "system", + "lowest", + "token", + "generation", + "cost", + "Spectrum", + "Ethernet", + "platform", + "purpose", + "built", + "larger", + "network", + "peers", + "combined", + "InfiniBand", + "growing", + "4x", + "deployments", + "XDR", + "technology", + "For", + "models", + "computing", + "60", + "77%", + "networking", + "15", + "tripled", + "Before", + "deep", + "dive", + "transition", + "reporting", + "framework", + "reflects", + "current", + "future", + "drivers", + "platforms", + "Edge", + "Computing", + "Within", + "report", + "markets", + "Hyperscale", + "ACIE", + "incorporates", + "clouds", + "industrial", + "enterprise", + "include", + "public", + "world", + "largest", + "consumer", + "Internet", + "companies", + "addresses", + "opportunities", + "data", + "centers", + "factories", + "industries", + "countries", + "highlights", + "devices", + "agentic", + "physical", + "including", + "PCs", + "gaming", + "consoles", + "workstations", + "RAN", + "stations", + "robotics", + "automotive", + "reference", + "posted", + "website", + "breakdown", + "based", + "past", + "quarters", + "Moving", + "back", + "results", + "38", + "approximately", + "50%", + "increased", + "12%", + "37", + "grew", + "31%", + "Our", + "enabled", + "rapid", + "stand", + "compute", + "capacity", + "number", + "partner", + "10", + "megawatts", + "doubled", + "surpassing", + "80", + "sites", + "Sovereign", + "80%", + "infrastructure", + "40", + "representing", + "50", + "trillion", + "GDP", + "As", + "evident", + "Supported", + "vast", + "installed", + "breadth", + "CUDA", + "accelerated", + "application", + "provider", + "positioned", + "address", + "opportunity", + "exceeds", + "Demand", + "continues", + "expand", + "unprecedented", + "pace", + "build", + "accelerating", + "rising", + "price", + "renting", + "H100", + "risen", + "date", + "A100", + "pricing", + "15%", + "Benefiting", + "versatility", + "continuous", + "performance", + "enhancements", + "enhanced", + "software", + "stack", + "generating", + "profitable", + "depreciable", + "life", + "trusted", + "marketplace", + "Compute", + "foundation", + "billions", + "spending", + "financed", + "There", + "primary", + "First", + "search", + "advertising", + "recommender", + "content", + "understanding", + "hyperscale", + "workloads", + "continue", + "CPU", + "GPU", + "Second", + "adoption", + "products", + "services", + "native", + "inflecting", + "Since", + "advent", + "ChatGPT", + "witnessed", + "mainstream", + "shot", + "reasoning", + "longer", + "nice", + "necessity", + "enhancing", + "productivity", + "roles", + "propelling", + "layers", + "cake", + "energy", + "chips", + "applications", + "Growth", + "layer", + "Anthropic", + "OpenAI", + "incredible", + "momentum", + "continuing", + "accelerate", + "breakout", + "Codex", + "launch", + "GPT", + "5.5", + "With", + "analysts", + "forecasting", + "CapEx", + "exceed", + "2027", + "beginning", + "proliferate", + "track", + "reach", + "annually", + "decade", + "adopted", + "major", + "hyperscaler", + "maker", + "Last", + "month", + "celebrated", + "designed", + "trained", + "served", + "top", + "Artificial", + "Analysis", + "leaderboards", + "Microsoft", + "Fairwater", + "powerful", + "center", + "live", + "ahead", + "schedule", + "powered", + "Starting", + "AWS", + "add", + "million", + "Rubin", + "collaborating", + "At", + "Google", + "offered", + "confidential", + "capability", + "secure", + "high", + "increasing", + "deepened", + "collaboration", + "delighted", + "support", + "trajectory", + "Azure", + "CoreWeave", + "SpaceXAI", + "Now", + "addition", + "Gemini", + "SpaceX", + "xAI", + "Meta", + "MSL", + "TML", + "Reflection", + "Perplexity", + "Cursor", + "labs", + "building", + "grow", + "significantly", + "Today", + "Constrained", + "power", + "factory", + "operators", + "choose", + "extreme", + "design", + "approach", + "deliver", + "industry", + "highest", + "throughput", + "ROI", + "MLPerf", + "swept", + "benchmark", + "Ultra", + "broad", + "set", + "deployment", + "scenarios", + "Full", + "innovations", + "drove", + "2.7", + "60%", + "reduction", + "compared", + "months", + "ago", + "economic", + "financeable", + "Customers", + "buy", + "They", + "metric", + "purchase", + "lifetime", + "producing", + "intelligence", + "watt", + "tokens", + "dollar", + "uptime", + "utilization", + "time", + "production", + "durability", + "asset", + "excels", + "Agentic", + "reinforcement", + "learning", + "represents", + "CPUs", + "Building", + "success", + "Vera", + "arriving", + "meet", + "Built", + "custom", + "Arm", + "cores", + "NVLink", + "1.5", + "faster", + "core", + "2x", + "density", + "rack", + "x86", + "alternatives", + "opens", + "brand", + "200", + "TAM", + "addressed", + "partnering", + "visibility", + "total", + "setting", + "leading", + "supplier", + "annual", + "cadence", + "unmatched", + "remains", + "key", + "pillar", + "supporting", + "commence", + "shipments", + "half", + "starting", + "Q3", + "By", + "integrating", + "racks", + "35x", + "higher", + "10x", + "greater", + "early", + "adopter", + "A5X", + "bare", + "metal", + "instances", + "960", + "000", + "multiple", + "enable", + "run", + "optimized", + "While", + "US", + "government", + "approved", + "licenses", + "H200", + "shipped", + "China", + "generate", + "uncertain", + "imports", + "allowed", + "country", + "result", + "consistent", + "outlook", + "Let", + "move", + "generated", + "6.4", + "10%", + "29%", + "Robust", + "workstation", + "contributor", + "fell", + "modestly", + "due", + "memory", + "prices", + "gain", + "12", + "partnership", + "Uber", + "robotaxi", + "fleet", + "30", + "cities", + "continents", + "2028", + "range", + "surgical", + "humanoid", + "develop", + "deploy", + "scale", + "remain", + "front", + "footed", + "securing", + "sufficient", + "inclusive", + "inventory", + "commitments", + "prepaids", + "145", + "immune", + "challenges", + "confident", + "ability", + "intense", + "focus", + "longstanding", + "partnerships", + "suppliers", + "serve", + "rest", + "GAAP", + "gross", + "margin", + "74.9%", + "75%", + "largely", + "flat", + "continued", + "account", + "expenses", + "primarily", + "compensation", + "costs", + "effective", + "tax", + "rate", + "16%", + "favorable", + "geographic", + "mix", + "balance", + "sheet", + "days", + "sales", + "outstanding", + "45", + "Due", + "timing", + "collections", + "expect", + "return", + "mid", + "50s", + "Q2", + "49", + "35", + "Q4", + "update", + "allocation", + "plan", + "reiterate", + "intention", + "prioritize", + "investment", + "cultivate", + "drive", + "strengthen", + "enabler", + "make", + "partners", + "Return", + "program", + "component", + "strategy", + "Given", + "confidence", + "commitment", + "sharing", + "quarterly", + "dividend", + "0.01", + "0.20", + "sic", + "0.25", + "review", + "regular", + "basis", + "business", + "announcing", + "repurchase", + "authorization", + "39", + "remaining", + "GTC", + "roughly", + "turn", + "expected", + "91", + "minus", + "2%", + "work", + "vigorously", + "giving", + "full", + "foresee", + "2025", + "calendar", + "margins", + "points", + "expecting", + "70s", + "8.5", + "8.3", + "OpEx", + "upper", + "40s", + "usage", + "tools", + "enhance", + "rates", + "18%", + "excluding", + "discrete", + "items", + "material", + "environment", + "lower", + "expectation", + "17%", + "19%", + "That", + "puts", + "part" + ], + "keywords": [ + "Blackwell", + "Data Center", + "AI Infrastructure" + ], + "connect_to": [] + } + }, + { + "chunk_id": "39e78aea-6ba1-5292-8a2a-32be08317f89", + "type": "text", + "content": "Vice President-Investor Relations & Strategic Finance, NVIDIA Corp.\nThanks, Colette. We will now transition to Q&A. Operator, please poll for questions.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Toshiya Hari_2", + "metadata": { + "length": 152, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Vice", + "President", + "Investor", + "Relations", + "Strategic", + "Finance", + "NVIDIA", + "Corp", + "Thanks", + "Colette", + "We", + "transition", + "Operator", + "poll", + "questions" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e6aa7700-93ad-5ed1-8d05-94bcff67d917", + "type": "text", + "content": "Operator: Thank you. [Operator Instructions] Your first question comes from Joseph Moore with Morgan Stanley. Your lie is open.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION", + "metadata": { + "length": 127, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Operator", + "Thank", + "Instructions", + "Your", + "question", + "Joseph", + "Moore", + "Morgan", + "Stanley", + "lie", + "open" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "acc3c007-53eb-5731-a952-6f07aeb4deca", + "type": "image", + "content": "\nThanks, Colette. We will now transition to Q&A. Operator, please poll for questions. Great. Thank you for letting me ask a question. I guess, I'd like to ask, what drove the change in segmentation? What's the philosophy behind giving us the numbers that way? And then, can you talk about any competitive differences between the two segments and this kind of surprising CPU number that you talked about, how do you see that across the two segments as well? Thank you.0\n[images/image-1-Thanks, Co.jpg]\n", + "path": "images/image-1-Thanks, Co.jpg", + "metadata": { + "length": 501, + "summary": "image-1\nThanks, Colette. We will now transition to Q&A. Operator, please poll for questions. Great. Thank you for letting me ask a question. I guess, I'd like to ask, what drove the change in segmentation? What's the philosophy behind giving us the numbers that way? And then, can you talk about any competitive differences between the two segments and this kind of surprising CPU number that you talked about, how do you see that across the two segments as well? Thank you.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-1-Thanks, Co.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d6fc3d14-4406-5a60-8bef-8576de14510a", + "type": "text", + "content": "Analyst, Morgan Stanley & Co. LLC\n\nThanks, Colette. We will now transition to Q&A. Operator, please poll for questions. Great. Thank you for letting me ask a question. I guess, I'd like to ask, what drove the change in segmentation? What's the philosophy behind giving us the numbers that way? And then, can you talk about any competitive differences between the two segments and this kind of surprising CPU number that you talked about, how do you see that across the two segments as well? Thank you.0\n[images/image-1-Thanks, Co.jpg]\n\nGreat. Thank you for letting me ask a question. I guess, I'd like to ask, what drove the change in segmentation? What's the philosophy behind giving us the numbers that way? And then, can you talk about any competitive differences between the two segments and this kind of surprising CPU number that you talked about, how do you see that across the two segments as well? Thank you.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Joseph Moore", + "metadata": { + "length": 917, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Morgan", + "Stanley", + "Co", + "LLC", + "Thanks", + "Colette", + "We", + "transition", + "Operator", + "poll", + "questions", + "Great", + "Thank", + "letting", + "question", + "guess", + "drove", + "change", + "segmentation", + "What", + "philosophy", + "giving", + "numbers", + "And", + "talk", + "competitive", + "differences", + "segments", + "kind", + "surprising", + "CPU", + "number", + "talked", + "you.0" + ], + "keywords": [], + "connect_to": [ + { + "target": "acc3c007-53eb-5731-a952-6f07aeb4deca", + "relation": "embeds", + "ref": "[images/image-1-Thanks, Co.jpg]", + "position": { + "start": 503, + "end": 534 + } + } + ] + } + }, + { + "chunk_id": "07ba7e69-b593-55fd-94b1-d093b1c2967a", + "type": "image", + "content": "\nAnalyst, Morgan Stanley & Co. LLC Yeah. Thanks, Joe. First of all, Colette meant to say we're increasing our quarterly dividend from \\$0.01 to \\$0.25. I think that extra \\$0.05 would mean a lot to the large shareholders.0\n[images/image-2-Analyst, M.jpg]\n", + "path": "images/image-2-Analyst, M.jpg", + "metadata": { + "length": 255, + "summary": "image-2\nAnalyst, Morgan Stanley & Co. LLC Yeah. Thanks, Joe. First of all, Colette meant to say we're increasing our quarterly dividend from \\$0.01 to \\$0.25. I think that extra \\$0.05 would mean a lot to the large shareholders.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-2-Analyst, M.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "95c2d821-274f-5811-9bc9-beb57c4fb470", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nAnalyst, Morgan Stanley & Co. LLC Yeah. Thanks, Joe. First of all, Colette meant to say we're increasing our quarterly dividend from \\$0.01 to \\$0.25. I think that extra \\$0.05 would mean a lot to the large shareholders.0\n[images/image-2-Analyst, M.jpg]\n\nYeah. Thanks, Joe. First of all, Colette meant to say we're increasing our quarterly dividend from \\$0.01 to \\$0.25. I think that extra \\$0.05 would mean a lot to the large shareholders.\nSo, anyhow, let's see, Joe, on the segmentation and the description of the business, we wanted you to understand our business better. AI is very diverse and computing is diverse. They're diverse in several ways. The first thing, of course, is AI includes languages, and depending on the different industries, it could be 3D graphics for manufacturing and industrial robotics. It could be proteins for life sciences. It could be small chemicals for life sciences or material sciences. It could be physics for the physical sciences, whether it's in the energy sector or, of course, the science labs, higher education, so on and so forth. So, AI is diverse.\nThe second thing is the applications are diverse. It could be in enterprise. It could be in the energy sector, manufacturing sector, and such. Where it runs is diverse. It could be in the hyperscale cloud. It could be AI natives. There's a whole network of AI natives that are cropping up around the world. Enterprises on-prem, industrial in the factories, in the plants, all the way to supercomputing centers and the edge. Edge, including, of course, what most people see, self-driving cars, robotics, but a large growing network of computers inside manufacturing plants, whether it's a chip plant or packaging or computer plants, all kinds of different types of manufacturing plants. And then, of course, in the future, every single base station, every single radio network would become an AI-powered radio network. And so where it runs.\nAnd then lastly, how it's governed. It could be operated by public cloud, but it could also have industrial regulatory reasons that prevents it from being run in a regulatory cloud. It could be because of confidential computing. It could be because of national security reasons. Different data centers have to be built differently.\nNVIDIA is quite unique in the sense that we are the only company that builds all of the technology components. We build it in an extreme co-design way, in a complete end-to-end way, in a full stack way. But then, we, of course, open the platform so that it could be integrated into all the different environments. But some environments just require – an enterprise, for example, require a company who has all of the technologies working together so that they don't have to build it. They would like to buy it and operate it. And so there's many different segments of the data center market where NVIDIA's total solution, fully integrated solution with full stack, but still open, that way of doing – of producing or delivering products is really, really important.\nAnd so if you look at our different segments, the way we broke it out into three large segments. You take all of the words that I just said and you try to find the simplest factoring of it, it would be the hyperscale clouds, that would be one large segment. And within that segment, there's three different ways that we operate. First way is that we help the hyperscale clouds accelerate their data processing and machine learning workloads. We accelerate and support their AI processing inside. We also, of course, bring a lot of business – NVIDIA ecosystem business to their public clouds. And so that's one segment.\nThe second segment is AI natives, enterprise on-prems, industrial on-prems and sovereign AI. That segment is growing incredibly fast because everybody needs AI, and we're going to see AI being adopted by every industry, every country, every company. And so everybody wants to build it in a different way. And the fact that we provide the entire solution, it makes it much easier, it makes it possible at all for people to be able to build these things.\nAnd then, of course, the robotic edge. Today, yesterday's computing was largely about personal computing. In the future, it's going to be about personal AI. And that personal AI, one example of it is the self-driving car. It's a car – it's a robotic system that's essentially your personal AI. And of course, there'll be all kinds of different types of robotic systems, including even the base station radio network, as I mentioned, is going to be essentially a robotic system.\nAnd so that's the reason why we broke it all apart this way. It's the simplest way of understanding our business. Each one of them have different stacks in a lot of ways. They have different operating systems. They operate in a different way. We go to market very differently in each one of them. The easiest go-to-market, of course, is the hyperscaler, because there are only five or six of them. But the rest of them – the rest of the industry represents a couple of 250,000 companies around the world. That go-to-market is very complex, very diverse. Your understanding of AI has to be extremely diverse.\nAnd as you know, NVIDIA has the largest suite of acceleration libraries in the world from computational lithography to fluid dynamics to particle physics to molecular dynamics to the list goes on. And all of those libraries are essential for us to engage the vertical industries that represents the second and the third category. Okay?\nSo, anyways, it's really about the fact that our business has now evolved and grown to such a large scale, it's helpful to segment it so that you have a better understanding of how our business works.\nOperator: Your next question comes from Ben Reitzes with Melius Research. Your line is open.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang", + "metadata": { + "length": 5894, + "summary": "NVIDIA's CEO explains the company's new business segmentation into three categories: hyperscale clouds, AI natives/enterprise/industrial/sovereign AI, and robotic edge. He describes how AI is diverse across languages, industries, and applications, and how NVIDIA provides a full-stack, integrated platform that is still open. The segmentation helps investors understand NVIDIA's evolving business, which now serves a wide range of customers from a few hyperscalers to 250,000 companies worldwide.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Analyst", + "Morgan", + "Stanley", + "LLC", + "Yeah", + "Thanks", + "Joe", + "First", + "Colette", + "meant", + "increasing", + "quarterly", + "dividend", + "0.01", + "0.25", + "extra", + "0.05", + "lot", + "large", + "shareholders.0", + "shareholders", + "So", + "segmentation", + "description", + "business", + "wanted", + "understand", + "AI", + "diverse", + "computing", + "They", + "ways", + "The", + "thing", + "includes", + "languages", + "depending", + "industries", + "3D", + "graphics", + "manufacturing", + "industrial", + "robotics", + "It", + "proteins", + "life", + "sciences", + "small", + "chemicals", + "material", + "physics", + "physical", + "energy", + "sector", + "science", + "labs", + "higher", + "education", + "applications", + "enterprise", + "Where", + "runs", + "hyperscale", + "cloud", + "natives", + "There", + "network", + "cropping", + "world", + "Enterprises", + "prem", + "factories", + "plants", + "supercomputing", + "centers", + "edge", + "Edge", + "including", + "people", + "driving", + "cars", + "growing", + "computers", + "inside", + "chip", + "plant", + "packaging", + "computer", + "kinds", + "types", + "And", + "future", + "single", + "base", + "station", + "radio", + "powered", + "lastly", + "governed", + "operated", + "public", + "regulatory", + "reasons", + "prevents", + "run", + "confidential", + "national", + "security", + "Different", + "data", + "built", + "differently", + "unique", + "sense", + "company", + "builds", + "technology", + "components", + "We", + "build", + "extreme", + "design", + "complete", + "end", + "full", + "stack", + "But", + "open", + "platform", + "integrated", + "environments", + "require", + "technologies", + "working", + "don", + "buy", + "operate", + "segments", + "center", + "market", + "total", + "solution", + "fully", + "producing", + "delivering", + "products", + "important", + "broke", + "You", + "words", + "find", + "simplest", + "factoring", + "clouds", + "segment", + "accelerate", + "processing", + "machine", + "learning", + "workloads", + "support", + "bring", + "ecosystem", + "prems", + "sovereign", + "That", + "incredibly", + "fast", + "adopted", + "industry", + "country", + "fact", + "provide", + "entire", + "makes", + "easier", + "things", + "robotic", + "Today", + "yesterday", + "largely", + "personal", + "In", + "car", + "system", + "essentially", + "ll", + "systems", + "mentioned", + "reason", + "understanding", + "Each", + "stacks", + "operating", + "easiest", + "hyperscaler", + "rest", + "represents", + "couple", + "250", + "000", + "companies", + "complex", + "Your", + "extremely", + "largest", + "suite", + "acceleration", + "libraries", + "computational", + "lithography", + "fluid", + "dynamics", + "particle", + "molecular", + "list", + "essential", + "engage", + "vertical", + "category", + "Okay", + "evolved", + "grown", + "scale", + "helpful", + "works", + "Operator", + "question", + "Ben", + "Reitzes", + "Melius", + "Research", + "line" + ], + "keywords": [ + "AI", + "Data Center", + "Segmentation" + ], + "connect_to": [ + { + "target": "07ba7e69-b593-55fd-94b1-d093b1c2967a", + "relation": "embeds", + "ref": "[images/image-2-Analyst, M.jpg]", + "position": { + "start": 295, + "end": 326 + } + } + ] + } + }, + { + "chunk_id": "c1513822-b24c-5b5a-a472-18ad275654cf", + "type": "image", + "content": "\nAnd as you know, NVIDIA has the largest suite of acceleration libraries in the world from computational lithography to fluid dynamics to particle physics to molecular dynamics to the list goes on. And all of those libraries are essential for us to engage the vertical industries that represents the second and the third category. Okay? Hey, guys. Thanks so much. I wanted to ask, Jensen, I want to ask you about your philosophy on growth. Your Data Center business ex-China grew about 120% in the quarter. And then, you're guiding about 100%. CapEx at the hyperscalers is forecast by many, including myself, to like grow 90% to 100% this year.0\n[images/image-3-And as you.jpg]\n", + "path": "images/image-3-And as you.jpg", + "metadata": { + "length": 678, + "summary": "image-3\nAnd as you know, NVIDIA has the largest suite of acceleration libraries in the world from computational lithography to fluid dynamics to particle physics to molecular dynamics to the list goes on. And all of those libraries are essential for us to engage the vertical industries that represents the second and the third category. Okay? Hey, guys. Thanks so much. I wanted to ask, Jensen, I want to ask you about your philosophy on growth. Your Data Center business ex-China grew about 120% in the quarter. And then, you're guiding about 100%. CapEx at the hyperscalers is forecast by many, including myself, to like grow 90% to 100% this year.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-3-And as you.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "94fcb5fe-1a27-5ae4-99aa-9ced05c8c04b", + "type": "text", + "content": "Analyst, Melius Research LLC\n\nAnd as you know, NVIDIA has the largest suite of acceleration libraries in the world from computational lithography to fluid dynamics to particle physics to molecular dynamics to the list goes on. And all of those libraries are essential for us to engage the vertical industries that represents the second and the third category. Okay? Hey, guys. Thanks so much. I wanted to ask, Jensen, I want to ask you about your philosophy on growth. Your Data Center business ex-China grew about 120% in the quarter. And then, you're guiding about 100%. CapEx at the hyperscalers is forecast by many, including myself, to like grow 90% to 100% this year.0\n[images/image-3-And as you.jpg]\n\nHey, guys. Thanks so much. I wanted to ask, Jensen, I want to ask you about your philosophy on growth. Your Data Center business ex-China grew about 120% in the quarter. And then, you're guiding about 100%. CapEx at the hyperscalers is forecast by many, including myself, to like grow 90% to 100% this year.\nAnd you talked about Data Center still on track to be \\$3 trillion to \\$4 trillion by the end of the decade. I was just wondering, the goal for the company to grow faster than hyperscaler CapEx, are you comfortable in kind of endorsing that view? And do you still see hyperscaler CapEx kind of still growing after this year at a very rapid clip? Thanks a lot.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Ben Reitzes", + "metadata": { + "length": 1375, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Melius", + "Research", + "LLC", + "And", + "NVIDIA", + "largest", + "suite", + "acceleration", + "libraries", + "world", + "computational", + "lithography", + "fluid", + "dynamics", + "particle", + "physics", + "molecular", + "list", + "essential", + "engage", + "vertical", + "industries", + "represents", + "category", + "Okay", + "Hey", + "guys", + "Thanks", + "wanted", + "Jensen", + "philosophy", + "growth", + "Your", + "Data", + "Center", + "business", + "China", + "grew", + "120%", + "quarter", + "guiding", + "100%", + "CapEx", + "hyperscalers", + "forecast", + "including", + "grow", + "90%", + "year.0", + "year", + "talked", + "track", + "trillion", + "end", + "decade", + "wondering", + "goal", + "company", + "faster", + "hyperscaler", + "comfortable", + "kind", + "endorsing", + "view", + "growing", + "rapid", + "clip", + "lot" + ], + "keywords": [], + "connect_to": [ + { + "target": "c1513822-b24c-5b5a-a472-18ad275654cf", + "relation": "embeds", + "ref": "[images/image-3-And as you.jpg]", + "position": { + "start": 675, + "end": 706 + } + } + ] + } + }, + { + "chunk_id": "5072678f-dc1d-5ec3-9a5f-d4f70b1f72ae", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\nA\nYeah. Thanks, Ben. So, first of all, we should be growing faster than hyperscale CapEx. And the reason for that is illustrated by the segmentation that I just described. Our Data Center business has two large parts. It has more parts than that, but we combined it into two large parts for simplicity's sake. It's much more complex than the two large parts, but I combined it into two, so that it's at least easier to understand. Okay?\nAnd so if you look at the first part is hyperscalers. That's the hyperscale CapEx that you were just talking about. And they're at \\$1 trillion this year. I have every expectation it's going to grow from here for fundamentally good reasons. This is the way computing is going to work in the future. And if they don't have the compute, they won't have the revenues. It is very clear compute is revenues, compute is profit. And so the world is changing. SaaS didn't used to use as much compute, but AI requires a tremendous amount of compute.\nBut you could do, of course, incredibly more, which is the reason why we heard about the frontier AI companies, both Anthropic and OpenAI, growing at an incredible pace. The fact that they can grow within one month, what some of the SaaS companies would have taken a decade to grow, tells you something. And so the first category is hyperscale and the CapEx is at \\$1 trillion, and it's growing towards the \\$3 trillion to \\$4 trillion.\nThe second category is all of the AI native clouds. They're regional. They're all over the place. There are startups all over the world supporting those companies. They're enterprise. 250,000 enterprise companies around the world, many of them will have to build or want to build AI factories for themselves to operate. Many industrial companies, there's no choice but to put the computer where the context is, where the action is. You can't put that in the cloud. It has to respond reliably, quickly every single time. Can't imagine a chip plant – a chip fab being connected to a cloud service provider, doesn't make any sense.\nAnd so, the second category, and then sovereign AI clouds, and so there's a whole category of data centers that semi-custom chips just don't apply, because these data centers want to buy systems. They want to operate systems. They don't want to design – they don't want to build it themselves. And so, the second category is extremely diverse. Instead of five or six, seven companies representing the revenues associated with our first category, the second category is hundreds, thousands of companies. And in the future, it will be hundreds of thousands of companies with a large number of companies with smaller installations. And that category is going to continue to grow at incredible pace.\nThis second category, when I talk about physical AI and I talk about how the rest of the \\$100 trillion industry that has not been impacted by IT in the last 30 years, it's about to be impacted by AI, that is the segment that I'm talking about. The second cluster is growing incredibly fast. Our share of that, of course, is very, very large. We're fairly unique in our ability to be able to serve this industry. Our platform is built like it's vertically integrated, so that everything works. But when – then we disassemble it, so that people can build and buy it in the configuration they want and assemble it the way they like.\nAnd so, this second category is fairly poorly understood because they're just so many small companies, or so many companies, and each one of the installations are relatively small compared to, of course, one of the hyperscalers. And so, if you look at the segmentation and the size of each, you could see that, in fact, we're growing share in the hyperscalers, because we now have much bigger support from Anthropic, a new partner of ours, and we're helping them expand their capacity greatly in the coming years. And then, the second, very few companies have exposure into the second category, because of the platform solution that we have.\nOperator: Your next question comes from C.J. Muse with Cantor Fitzgerald. Your line is open.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_2", + "metadata": { + "length": 4177, + "summary": "NVIDIA's CEO explains that the company's Data Center business should grow faster than hyperscale CapEx, which is at $1 trillion and expected to grow to $3-4 trillion. The business is divided into two segments: hyperscalers (large cloud providers) and a diverse second category including AI-native clouds, enterprise companies, industrial firms, and sovereign AI clouds. The second category, involving hundreds of thousands of companies, is growing rapidly and represents the future of AI impacting industries not previously transformed by IT. NVIDIA's vertically integrated platform uniquely serves this diverse market, with strong share in both segments.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Yeah", + "Thanks", + "Ben", + "So", + "growing", + "faster", + "hyperscale", + "CapEx", + "And", + "reason", + "illustrated", + "segmentation", + "Our", + "Data", + "Center", + "business", + "large", + "parts", + "It", + "combined", + "simplicity", + "sake", + "complex", + "easier", + "understand", + "Okay", + "part", + "hyperscalers", + "That", + "talking", + "trillion", + "year", + "expectation", + "grow", + "fundamentally", + "good", + "reasons", + "This", + "computing", + "work", + "future", + "don", + "compute", + "won", + "revenues", + "clear", + "profit", + "world", + "changing", + "SaaS", + "didn", + "AI", + "requires", + "tremendous", + "amount", + "But", + "incredibly", + "heard", + "frontier", + "companies", + "Anthropic", + "OpenAI", + "incredible", + "pace", + "The", + "fact", + "month", + "decade", + "tells", + "category", + "native", + "clouds", + "They", + "regional", + "place", + "There", + "startups", + "supporting", + "enterprise", + "250", + "000", + "build", + "factories", + "operate", + "Many", + "industrial", + "choice", + "put", + "computer", + "context", + "action", + "You", + "cloud", + "respond", + "reliably", + "quickly", + "single", + "time", + "Can", + "imagine", + "chip", + "plant", + "fab", + "connected", + "service", + "provider", + "doesn", + "make", + "sense", + "sovereign", + "data", + "centers", + "semi", + "custom", + "chips", + "apply", + "buy", + "systems", + "design", + "extremely", + "diverse", + "Instead", + "representing", + "hundreds", + "thousands", + "number", + "smaller", + "installations", + "continue", + "talk", + "physical", + "rest", + "100", + "industry", + "impacted", + "IT", + "30", + "years", + "segment", + "cluster", + "fast", + "share", + "We", + "fairly", + "unique", + "ability", + "serve", + "platform", + "built", + "vertically", + "integrated", + "works", + "disassemble", + "people", + "configuration", + "assemble", + "poorly", + "understood", + "small", + "compared", + "size", + "bigger", + "support", + "partner", + "helping", + "expand", + "capacity", + "greatly", + "coming", + "exposure", + "solution", + "Operator", + "Your", + "question", + "Muse", + "Cantor", + "Fitzgerald", + "line", + "open" + ], + "keywords": [ + "Hyperscalers", + "AI Clouds", + "Data Center" + ], + "connect_to": [] + } + }, + { + "chunk_id": "07165cdd-0a09-5701-8661-e012f90558a6", + "type": "image", + "content": "\nThis second category, when I talk about physical AI and I talk about how the rest of the \\$100 trillion industry that has not been impacted by IT in the last 30 years, it's about to be impacted by AI, that is the segment that I'm talking about. The second cluster is growing incredibly fast. Our share of that, of course, is very, very large. We're fairly unique in our ability to be able to serve this industry. Our platform is built like it's vertically integrated, so that everything works. But when – then we disassemble it, so that people can build and buy it in the configuration they want and assemble it the way they like. Yeah. Good afternoon. Thank you for taking the question. You have Vera Rubin coming soon. And you, obviously, have great insight into coming updates to frontier models, new techniques to optimize around diverse AI workloads. With investors keenly focused on your market share in inference, how do you see Vera Rubin and your extreme co-engineering impacting your share of the inference market as we look into late 2026, 2027?0\n[images/image-4-This secon.jpg]\n", + "path": "images/image-4-This secon.jpg", + "metadata": { + "length": 1091, + "summary": "image-4\nThis second category, when I talk about physical AI and I talk about how the rest of the \\$100 trillion industry that has not been impacted by IT in the last 30 years, it's about to be impacted by AI, that is the segment that I'm talking about. The second cluster is growing incredibly fast. Our share of that, of course, is very, very large. We're fairly unique in our ability to be able to serve this industry. Our platform is built like it's vertically integrated, so that everything works. But when – then we disassemble it, so that people can build and buy it in the configuration they want and assemble it the way they like. Yeah. Good afternoon. Thank you for taking the question. You have Vera Rubin coming soon. And you, obviously, have great insight into coming updates to frontier models, new techniques to optimize around diverse AI workloads. With investors keenly focused on your market share in inference, how do you see Vera Rubin and your extreme co-engineering impacting your share of the inference market as we look into late 2026, 2027?0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-4-This secon.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "a94ba201-3e21-5985-a478-46903f9cf880", + "type": "text", + "content": "Analyst, Cantor Fitzgerald & Co.\n\nThis second category, when I talk about physical AI and I talk about how the rest of the \\$100 trillion industry that has not been impacted by IT in the last 30 years, it's about to be impacted by AI, that is the segment that I'm talking about. The second cluster is growing incredibly fast. Our share of that, of course, is very, very large. We're fairly unique in our ability to be able to serve this industry. Our platform is built like it's vertically integrated, so that everything works. But when – then we disassemble it, so that people can build and buy it in the configuration they want and assemble it the way they like. Yeah. Good afternoon. Thank you for taking the question. You have Vera Rubin coming soon. And you, obviously, have great insight into coming updates to frontier models, new techniques to optimize around diverse AI workloads. With investors keenly focused on your market share in inference, how do you see Vera Rubin and your extreme co-engineering impacting your share of the inference market as we look into late 2026, 2027?0\n[images/image-4-This secon.jpg]\n\nYeah. Good afternoon. Thank you for taking the question. You have Vera Rubin coming soon. And you, obviously, have great insight into coming updates to frontier models, new techniques to optimize around diverse AI workloads. With investors keenly focused on your market share in inference, how do you see Vera Rubin and your extreme co-engineering impacting your share of the inference market as we look into late 2026, 2027?", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/C.J. Muse", + "metadata": { + "length": 1550, + "summary": "The speaker discusses how physical AI is poised to impact the $100 trillion industry that has not been affected by IT in the last 30 years. This segment is growing rapidly, and the company has a large share due to its vertically integrated platform that can be disassembled for flexible configuration. In response to a question about Vera Rubin and extreme co-engineering, the speaker addresses how these advancements will affect the company's share in the inference market looking into late 2026 and 2027.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Cantor", + "Fitzgerald", + "Co", + "This", + "category", + "talk", + "physical", + "AI", + "rest", + "100", + "trillion", + "industry", + "impacted", + "IT", + "30", + "years", + "segment", + "talking", + "The", + "cluster", + "growing", + "incredibly", + "fast", + "Our", + "share", + "large", + "We", + "fairly", + "unique", + "ability", + "serve", + "platform", + "built", + "vertically", + "integrated", + "works", + "But", + "disassemble", + "people", + "build", + "buy", + "configuration", + "assemble", + "Yeah", + "Good", + "afternoon", + "Thank", + "taking", + "question", + "You", + "Vera", + "Rubin", + "coming", + "And", + "great", + "insight", + "updates", + "frontier", + "models", + "techniques", + "optimize", + "diverse", + "workloads", + "With", + "investors", + "keenly", + "focused", + "market", + "inference", + "extreme", + "engineering", + "impacting", + "late", + "2026", + "2027" + ], + "keywords": [ + "Physical AI", + "Inference Market", + "Vera Rubin" + ], + "connect_to": [ + { + "target": "07165cdd-0a09-5701-8661-e012f90558a6", + "relation": "embeds", + "ref": "[images/image-4-This secon.jpg]", + "position": { + "start": 1092, + "end": 1123 + } + } + ] + } + }, + { + "chunk_id": "44dfcf5f-6ca4-52b2-9cb6-6e6de465182e", + "type": "image", + "content": "\nAnalyst, Cantor Fitzgerald & Co. Well, we are growing share in inference and we're growing share in inference very, very quickly. And the reason for that is, this year, the number of frontier model companies grew. And so there's Cursor and Perplexity, and there's some new model companies TML and Reflection and the list goes on. And so the number of frontier model companies has grown.0\n[images/image-5-Analyst, C.jpg]\n", + "path": "images/image-5-Analyst, C.jpg", + "metadata": { + "length": 421, + "summary": "image-5\nAnalyst, Cantor Fitzgerald & Co. Well, we are growing share in inference and we're growing share in inference very, very quickly. And the reason for that is, this year, the number of frontier model companies grew. And so there's Cursor and Perplexity, and there's some new model companies TML and Reflection and the list goes on. And so the number of frontier model companies has grown.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-5-Analyst, C.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "70a3c18a-3a15-566b-a108-ddc11b5c6db1", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nAnalyst, Cantor Fitzgerald & Co. Well, we are growing share in inference and we're growing share in inference very, very quickly. And the reason for that is, this year, the number of frontier model companies grew. And so there's Cursor and Perplexity, and there's some new model companies TML and Reflection and the list goes on. And so the number of frontier model companies has grown.0\n[images/image-5-Analyst, C.jpg]\n\nWell, we are growing share in inference and we're growing share in inference very, very quickly. And the reason for that is, this year, the number of frontier model companies grew. And so there's Cursor and Perplexity, and there's some new model companies TML and Reflection and the list goes on. And so the number of frontier model companies has grown.\nAnd we added Anthropic to our partnership this year. They're expanding incredibly fast. We've partnered with them to secure computing capacity across Azure, AWS, CoreWeave. I forget who else we've already announced, but there's a whole list of others that we are bringing online for them. And so the amount of capacity that we're going to bring online for Anthropic this year and next year is going to be quite significant, very significant. And our coverage of Anthropic has been largely zero until just recently.\nAnd so we're gaining share tremendously fast in inference. Vera Rubin is going to be even more successful than Grace Blackwell at this point. Every single – I can't think of one, every single frontier model company will jump on Vera Rubin from the get go and that wasn't true before on Blackwell. And so Vera Rubin is off to a tremendous start, and it'll surely be more successful than even Grace Blackwell. So, I think the end of your answer, C.J., is that we're gaining share in inference.\nLet me go back again to the question that Ben was asking. Remember, so far, everything that I've just explained in the inference question is really focused on hyperscale. Remember, there's a whole second category of AI data centers that we serve almost uniquely. This segment is very fragmented, requires a really well-integrated platform solution and a very large go-to-market, and that segment, all of the inference, 100% of that, the vast majority of that is NVIDIA.\nAnd then, of course, physical AI. NVIDIA is practically the only company serving physical AI today, and we've been working on physical AI for a long time. And so, that is also growing. So, our share of inference is growing very quickly.\nOperator: Your next question comes from Timothy Arcuri with UBS. Your line is open.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_3", + "metadata": { + "length": 2645, + "summary": "NVIDIA's Co-Founder and CEO discusses the company's rapid growth in AI inference market share. Key points include: the number of frontier model companies expanding (Cursor, Perplexity, TML, Reflection), new partnership with Anthropic providing computing capacity across Azure, AWS, and CoreWeave, and Vera Rubin expected to be more successful than Grace Blackwell. The CEO emphasizes NVIDIA's dominance in inference for hyperscale customers, unique position in fragmented AI data center segment, and leadership in physical AI.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Analyst", + "Cantor", + "Fitzgerald", + "Well", + "growing", + "share", + "inference", + "quickly", + "And", + "reason", + "year", + "number", + "frontier", + "model", + "companies", + "grew", + "Cursor", + "Perplexity", + "TML", + "Reflection", + "list", + "grown.0", + "grown", + "added", + "Anthropic", + "partnership", + "They", + "expanding", + "incredibly", + "fast", + "We", + "ve", + "partnered", + "secure", + "computing", + "capacity", + "Azure", + "AWS", + "CoreWeave", + "forget", + "announced", + "bringing", + "online", + "amount", + "bring", + "significant", + "coverage", + "largely", + "recently", + "gaining", + "tremendously", + "Vera", + "Rubin", + "successful", + "Grace", + "Blackwell", + "point", + "Every", + "single", + "company", + "jump", + "wasn", + "true", + "tremendous", + "start", + "ll", + "surely", + "So", + "end", + "answer", + "Let", + "back", + "question", + "Ben", + "Remember", + "explained", + "focused", + "hyperscale", + "category", + "AI", + "data", + "centers", + "serve", + "uniquely", + "This", + "segment", + "fragmented", + "requires", + "integrated", + "platform", + "solution", + "large", + "market", + "100%", + "vast", + "majority", + "physical", + "practically", + "serving", + "today", + "working", + "long", + "time", + "Operator", + "Your", + "Timothy", + "Arcuri", + "UBS", + "line", + "open" + ], + "keywords": [ + "NVIDIA", + "inference", + "AI" + ], + "connect_to": [ + { + "target": "44dfcf5f-6ca4-52b2-9cb6-6e6de465182e", + "relation": "embeds", + "ref": "[images/image-5-Analyst, C.jpg]", + "position": { + "start": 461, + "end": 492 + } + } + ] + } + }, + { + "chunk_id": "db69f90e-cfec-5e14-9773-f4167c0d746c", + "type": "image", + "content": "\nLet me go back again to the question that Ben was asking. Remember, so far, everything that I've just explained in the inference question is really focused on hyperscale. Remember, there's a whole second category of AI data centers that we serve almost uniquely. This segment is very fragmented, requires a really well-integrated platform solution and a very large go-to-market, and that segment, all of the inference, 100% of that, the vast majority of that is NVIDIA. Thanks a lot. Jensen, I wanted to ask about the traction you're getting with some of these custom merchant things. You're doing stuff like CPX and LPX. And I just wanted to ask and see sort of you've talked before about [ph] SaaS LPX (00:37:50) being, I think, 20% of the market. So, I would imagine you're getting pretty good traction with LPX. So, can you just talk about that and maybe also how that fits into your broader platform strategy? Thanks.0\n[images/image-6-Let me go .jpg]\n", + "path": "images/image-6-Let me go .jpg", + "metadata": { + "length": 957, + "summary": "image-6\nLet me go back again to the question that Ben was asking. Remember, so far, everything that I've just explained in the inference question is really focused on hyperscale. Remember, there's a whole second category of AI data centers that we serve almost uniquely. This segment is very fragmented, requires a really well-integrated platform solution and a very large go-to-market, and that segment, all of the inference, 100% of that, the vast majority of that is NVIDIA. Thanks a lot. Jensen, I wanted to ask about the traction you're getting with some of these custom merchant things. You're doing stuff like CPX and LPX. And I just wanted to ask and see sort of you've talked before about [ph] SaaS LPX (00:37:50) being, I think, 20% of the market. So, I would imagine you're getting pretty good traction with LPX. So, can you just talk about that and maybe also how that fits into your broader platform strategy? Thanks.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-6-Let me go .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "22051198-a5c0-5751-ab63-f729b293ff76", + "type": "text", + "content": "Analyst, UBS Securities LLC\n\nLet me go back again to the question that Ben was asking. Remember, so far, everything that I've just explained in the inference question is really focused on hyperscale. Remember, there's a whole second category of AI data centers that we serve almost uniquely. This segment is very fragmented, requires a really well-integrated platform solution and a very large go-to-market, and that segment, all of the inference, 100% of that, the vast majority of that is NVIDIA. Thanks a lot. Jensen, I wanted to ask about the traction you're getting with some of these custom merchant things. You're doing stuff like CPX and LPX. And I just wanted to ask and see sort of you've talked before about [ph] SaaS LPX (00:37:50) being, I think, 20% of the market. So, I would imagine you're getting pretty good traction with LPX. So, can you just talk about that and maybe also how that fits into your broader platform strategy? Thanks.0\n[images/image-6-Let me go .jpg]\n\nThanks a lot. Jensen, I wanted to ask about the traction you're getting with some of these custom merchant things. You're doing stuff like CPX and LPX. And I just wanted to ask and see sort of you've talked before about [ph] SaaS LPX (00:37:50) being, I think, 20% of the market. So, I would imagine you're getting pretty good traction with LPX. So, can you just talk about that and maybe also how that fits into your broader platform strategy? Thanks.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Timothy Arcuri", + "metadata": { + "length": 1438, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "UBS", + "Securities", + "LLC", + "Let", + "back", + "question", + "Ben", + "Remember", + "ve", + "explained", + "inference", + "focused", + "hyperscale", + "category", + "AI", + "data", + "centers", + "serve", + "uniquely", + "This", + "segment", + "fragmented", + "requires", + "integrated", + "platform", + "solution", + "large", + "market", + "100%", + "vast", + "majority", + "NVIDIA", + "Thanks", + "lot", + "Jensen", + "wanted", + "traction", + "custom", + "merchant", + "things", + "You", + "stuff", + "CPX", + "LPX", + "And", + "sort", + "talked", + "ph", + "SaaS", + "00", + "37", + "50", + "20%", + "So", + "imagine", + "pretty", + "good", + "talk", + "fits", + "broader", + "strategy", + "Thanks.0" + ], + "keywords": [], + "connect_to": [ + { + "target": "db69f90e-cfec-5e14-9773-f4167c0d746c", + "relation": "embeds", + "ref": "[images/image-6-Let me go .jpg]", + "position": { + "start": 953, + "end": 984 + } + } + ] + } + }, + { + "chunk_id": "8253a6a6-cc73-56e8-8796-28c9bf6003c0", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\nA\nThe LPX is designed for low latency and high token rate, but its throughput is low. Its model size capacity is low. And its context processing, its ability to absorb a lot of context, for example, for software coding, for agentic workloads, its ability to absorb a great deal of context is lower. And so the challenge is simply – and I've explained before that the use case for LPX is not broad. It's intended for somebody who has a fairly large portfolio of different types of token services. And for the high token rate, maybe these services are quite premium and the number of customers is not significant, but the token rate is very high. And so, that remains exactly consistent with what I've said before. And I still expect that. And so I expect that LPX and other SRAM-based, decode-focused, high token rate generated focused accelerators will be a niche product for some time to come.\nAs you know, Grace Blackwell and Vera Rubin, we support the entire life cycle of AI from the data processing, preparing for training, okay, data processing to pre-training, to post-training, reinforcement learning, all the way to inference, Grace Blackwell is the best platform in the world to do all of that. And if we insert certain circumstances, so long as the customer – the provider already has a high token rate service that they can offer, then we can tack on an LPX and they could deliver that service even better. And so that's how I see the market.\nAnd I think whether it's 20% or 10%, just depends on where we are in the development of AI. I think today, it's a lot less than 20%. Someday, these premium tokens could be 20%. And we're ready to work with the service providers to enable this capability. I'm excited about it.\nOperator: Your next question comes from Vivek Arya with Bank of America Securities. Your line is open.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_4", + "metadata": { + "length": 1906, + "summary": "The speaker discusses the LPX accelerator, designed for low latency and high token rate but with low throughput and limited context processing. It is a niche product for premium token services with few customers. In contrast, Grace Blackwell and Vera Rubin support the full AI lifecycle. The speaker estimates LPX's market share is currently less than 20% but could grow to 20% in the future as AI develops.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "The", + "LPX", + "designed", + "low", + "latency", + "high", + "token", + "rate", + "throughput", + "Its", + "model", + "size", + "capacity", + "And", + "context", + "processing", + "ability", + "absorb", + "lot", + "software", + "coding", + "agentic", + "workloads", + "great", + "deal", + "lower", + "challenge", + "simply", + "ve", + "explained", + "case", + "broad", + "It", + "intended", + "fairly", + "large", + "portfolio", + "types", + "services", + "premium", + "number", + "customers", + "significant", + "remains", + "consistent", + "expect", + "SRAM", + "based", + "decode", + "focused", + "generated", + "accelerators", + "niche", + "product", + "time", + "As", + "Grace", + "Blackwell", + "Vera", + "Rubin", + "support", + "entire", + "life", + "cycle", + "AI", + "data", + "preparing", + "training", + "pre", + "post", + "reinforcement", + "learning", + "inference", + "platform", + "world", + "insert", + "circumstances", + "long", + "customer", + "provider", + "service", + "offer", + "tack", + "deliver", + "market", + "20%", + "10%", + "depends", + "development", + "today", + "Someday", + "tokens", + "ready", + "work", + "providers", + "enable", + "capability", + "excited", + "Operator", + "Your", + "question", + "Vivek", + "Arya", + "Bank", + "America", + "Securities", + "line", + "open" + ], + "keywords": [ + "LPX", + "NVIDIA", + "AI Inference" + ], + "connect_to": [] + } + }, + { + "chunk_id": "cfaec787-8b5f-5058-9adc-be41676a0dd7", + "type": "text", + "content": "Analyst, BofA Securities, Inc.\nQ\nThanks for taking my question. Jensen, there's a lot of excitement around CPU for agentic applications and just a lot of noise around the number of CPUs actually exceeding the number of GPUs. And I was just hoping that you could kind of give your perspective that, first of all, is this an incremental workload? Is this kind of cannibalizing what the GPU would have done otherwise?\nAnd then, secondly, the \\$20 billion number that you gave, is that for standalone Vera CPUs, or is that kind of already included in that Vera as part of Vera Rubin? So just if you could educate us on the role of CPU versus GPU. Is it cannibalistic? Is it incremental? And then, the \\$20 billion number, how to kind of put that in context with what you sell, right, which is usually the CPU as part of the GPU. Thank you.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Vivek Arya", + "metadata": { + "length": 835, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "BofA", + "Securities", + "Inc", + "Thanks", + "taking", + "question", + "Jensen", + "lot", + "excitement", + "CPU", + "agentic", + "applications", + "noise", + "number", + "CPUs", + "exceeding", + "GPUs", + "And", + "hoping", + "kind", + "give", + "perspective", + "incremental", + "workload", + "Is", + "cannibalizing", + "GPU", + "20", + "billion", + "gave", + "standalone", + "Vera", + "included", + "part", + "Rubin", + "So", + "educate", + "role", + "versus", + "cannibalistic", + "put", + "context", + "sell", + "Thank" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5cd16c58-949e-5475-9635-3daf4786eb69", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\nA\nThe \\$20 billion is for standalone CPU. And remember, we have Vera is used in three ways, as a standalone – four ways, as a – let me just start with the one that you already know. The first way is Vera Rubin. And we'll sell millions of Rubins, and every two of them is connected to a Vera. And of course, we price those too. And they're properly priced. And so that's number one use case.\nThe second use case is Vera standalone CPU. The third is Vera with CX9 and the software stack for storage. And then Vera in a – with CX9 with a software stack for security and compute isolation and confidential computing. Okay? So each one of those use cases is built on Vera. And my sense is that we'll be supply constrained throughout the entire life of Vera Rubin. There are four different use cases of it. And – but anyhow, the answer to your question is of the \\$20 billion is a standalone.\nWith respect to CPUs, an agent is essentially what people call a harness. And the agent has a harness that does the – and the harness could be OpenClaw, it could be Hermes. Claude Code is essentially a harness around Claude, around the Opus model. OpenAI's Codex is a harness around the GPT-5.5 model. And so these are harnesses, and these harnesses provide for things like IO, orchestration, memory management, tool use, connected to tools, for example, browsers and things like that, C compilers, Python compilers. And so, the harness runs on CPU and the tool use runs on CPUs. For example, if the AI were to do a search or do a browser, use a browser, that would run on the CPU.\nThe world has 1 billion users, human users. My sense is that the world is going to have billions of agents. Not today. I mean, we're going to grow into it. But it will have billions of agents and those billions of agents will all use tools. And those tools are going to be like PCs. Just like us humans using PCs today, in the future, you'll have an agent using PC.\nAnd so, if you kind of think along the lines of in the future, you pick your favorite number of agents at the moment, at the moment, call it, a few hundred thousand, but in the future, call it, eventually a few billion, I could imagine them all using – effectively having PCs that they can all use.\nEvery one of those agents are going to spin off sub-agents. And every time they spin these off, you're going to need to do inference. That's where the thinking happens. All of the thinking happens on GPUs. All of the orchestration essentially runs on CPUs and the sub-agents, when they're spun off, when they're thinking, they use GPUs.\nWhenever the agents use simulators, those can run on CPUs or GPUs, which is the reason why we're working so closely with Cadence and Synopsys to accelerate all of the world's tools. We're accelerating all of the world's tools and data processing engines and database engines because agents use these tools and they have lower patience – tolerance than humans, and they want things to happen quickly.\nAnd so we're accelerating all of the world's tools so that it runs on CUDA. And you could see us doing that, when I work with Cadence and Synopsys and Siemens and companies – and Adobe, that's because we're trying to get all of the world's tools to run on GPUs, because they already have GPUs and it's a lot faster.\nSo, we're going to need a lot more CPUs, and Vera was designed to be an agentic CPU. The CPUs of the past were designed to have many cores, so that it could be easily rentable. People rented cores. Well, agents don't rent cores. They just want the work to be done fast. The economics of the past was dollars per core. That's the economics of cloud computing of the past. The economics of AI of the future is tokens per dollar or dollars per token. And so, what we need to do in the future is to generate tokens, process tokens as fast as possible. And that's what Vera does incredibly well.\nSo we're expecting to be very successful with Vera. But ultimately, what we're doing is we're building infrastructure for AI and it needs incredibly great storage. That's the reason why we built STX. It needs incredibly good networking. That's why we have Spectrum-X. It needs incredibly great GPUs, of course, and inferencing ability. That's the reason why NVLink 72. It needs incredibly great security and confidential computing, which is the reason why Vera Rubin is the world's first platform with end-to-end confidential computing. And it needs great CPUs. We've got it all covered.\nOperator: Your next question comes from Stacy Rasgon with Bernstein Research. Your line is open.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_5", + "metadata": { + "length": 4634, + "summary": "The speaker discusses NVIDIA's Vera CPU designed for AI agents, highlighting four use cases: with Vera Rubin GPUs, standalone, with CX9 for storage, and for security. They contrast past CPU economics (dollars per core) with future AI economics (dollars per token), emphasizing Vera's speed for token processing. Billions of future agents will use CPUs for orchestration and tools, while GPUs handle inference. NVIDIA is accelerating tools on CUDA and building complete AI infrastructure including GPUs, networking, storage, and security.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "The", + "20", + "billion", + "standalone", + "CPU", + "And", + "remember", + "Vera", + "ways", + "start", + "Rubin", + "ll", + "sell", + "millions", + "Rubins", + "connected", + "price", + "properly", + "priced", + "number", + "case", + "CX9", + "software", + "stack", + "storage", + "security", + "compute", + "isolation", + "confidential", + "computing", + "Okay", + "So", + "cases", + "built", + "sense", + "supply", + "constrained", + "entire", + "life", + "There", + "answer", + "question", + "With", + "respect", + "CPUs", + "agent", + "essentially", + "people", + "call", + "harness", + "OpenClaw", + "Hermes", + "Claude", + "Code", + "Opus", + "model", + "OpenAI", + "Codex", + "GPT", + "5.5", + "harnesses", + "provide", + "things", + "IO", + "orchestration", + "memory", + "management", + "tool", + "tools", + "browsers", + "compilers", + "Python", + "runs", + "For", + "AI", + "search", + "browser", + "run", + "world", + "users", + "human", + "My", + "billions", + "agents", + "Not", + "today", + "grow", + "But", + "PCs", + "Just", + "humans", + "future", + "PC", + "kind", + "lines", + "pick", + "favorite", + "moment", + "hundred", + "thousand", + "eventually", + "imagine", + "effectively", + "Every", + "spin", + "time", + "inference", + "That", + "thinking", + "All", + "GPUs", + "spun", + "Whenever", + "simulators", + "reason", + "working", + "closely", + "Cadence", + "Synopsys", + "accelerate", + "We", + "accelerating", + "data", + "processing", + "engines", + "database", + "lower", + "patience", + "tolerance", + "happen", + "quickly", + "CUDA", + "work", + "Siemens", + "companies", + "Adobe", + "lot", + "faster", + "designed", + "agentic", + "past", + "cores", + "easily", + "rentable", + "People", + "rented", + "Well", + "don", + "rent", + "They", + "fast", + "economics", + "dollars", + "core", + "cloud", + "tokens", + "dollar", + "token", + "generate", + "process", + "incredibly", + "expecting", + "successful", + "ultimately", + "building", + "infrastructure", + "great", + "STX", + "It", + "good", + "networking", + "Spectrum", + "inferencing", + "ability", + "NVLink", + "72", + "platform", + "end", + "ve", + "covered", + "Operator", + "Your", + "Stacy", + "Rasgon", + "Bernstein", + "Research", + "line", + "open" + ], + "keywords": [ + "Vera CPU", + "AI agents", + "GPU inference" + ], + "connect_to": [] + } + }, + { + "chunk_id": "b55d1ab8-f2ac-5507-b424-115803b5f041", + "type": "image", + "content": "\nSo, we're going to need a lot more CPUs, and Vera was designed to be an agentic CPU. The CPUs of the past were designed to have many cores, so that it could be easily rentable. People rented cores. Well, agents don't rent cores. They just want the work to be done fast. The economics of the past was dollars per core. That's the economics of cloud computing of the past. The economics of AI of the future is tokens per dollar or dollars per token. And so, what we need to do in the future is to generate tokens, process tokens as fast as possible. And that's what Vera does incredibly well. Hi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud?0\n[images/image-7-So, we're .jpg]\n", + "path": "images/image-7-So, we're .jpg", + "metadata": { + "length": 854, + "summary": "image-7\nSo, we're going to need a lot more CPUs, and Vera was designed to be an agentic CPU. The CPUs of the past were designed to have many cores, so that it could be easily rentable. People rented cores. Well, agents don't rent cores. They just want the work to be done fast. The economics of the past was dollars per core. That's the economics of cloud computing of the past. The economics of AI of the future is tokens per dollar or dollars per token. And so, what we need to do in the future is to generate tokens, process tokens as fast as possible. And that's what Vera does incredibly well. Hi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud?0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-7-So, we're .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "741fe289-b9e9-5b33-8e25-4ffa221ee55f", + "type": "text", + "content": "Analyst, Bernstein Research\n\nSo, we're going to need a lot more CPUs, and Vera was designed to be an agentic CPU. The CPUs of the past were designed to have many cores, so that it could be easily rentable. People rented cores. Well, agents don't rent cores. They just want the work to be done fast. The economics of the past was dollars per core. That's the economics of cloud computing of the past. The economics of AI of the future is tokens per dollar or dollars per token. And so, what we need to do in the future is to generate tokens, process tokens as fast as possible. And that's what Vera does incredibly well. Hi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud?0\n[images/image-7-So, we're .jpg]\n\nHi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud?\nPart of me assumes the latter, but I'm not so sure. And then, by just looking...", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Stacy A. Rasgon", + "metadata": { + "length": 1192, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Bernstein", + "Research", + "So", + "lot", + "CPUs", + "Vera", + "designed", + "agentic", + "CPU", + "The", + "past", + "cores", + "easily", + "rentable", + "People", + "rented", + "Well", + "agents", + "don", + "rent", + "They", + "work", + "fast", + "economics", + "dollars", + "core", + "That", + "cloud", + "computing", + "AI", + "future", + "tokens", + "dollar", + "token", + "And", + "generate", + "process", + "incredibly", + "Hi", + "guys", + "Thanks", + "taking", + "question", + "wanted", + "back", + "segmentation", + "curious", + "put", + "neo", + "clouds", + "segments", + "Are", + "hyperscale", + "Part", + "assumes" + ], + "keywords": [], + "connect_to": [ + { + "target": "b55d1ab8-f2ac-5507-b424-115803b5f041", + "relation": "embeds", + "ref": "[images/image-7-So, we're .jpg]", + "position": { + "start": 850, + "end": 881 + } + } + ] + } + }, + { + "chunk_id": "d02181e2-f4c6-5dab-a70b-f016daa165ba", + "type": "image", + "content": "\nHi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud? Correct.0\n[images/image-8-Hi, guys. .jpg]\n", + "path": "images/image-8-Hi, guys. .jpg", + "metadata": { + "length": 272, + "summary": "image-8\nHi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud? Correct.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-8-Hi, guys. .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "74ff6db0-19d5-5517-b056-be0f86f008e4", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nHi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud? Correct.0\n[images/image-8-Hi, guys. .jpg]\n\nCorrect.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_6", + "metadata": { + "length": 353, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Hi", + "guys", + "Thanks", + "taking", + "question", + "wanted", + "back", + "segmentation", + "So", + "curious", + "put", + "neo", + "clouds", + "segments", + "Are", + "hyperscale", + "AI", + "cloud", + "Correct.0", + "Correct" + ], + "keywords": [], + "connect_to": [ + { + "target": "d02181e2-f4c6-5dab-a70b-f016daa165ba", + "relation": "embeds", + "ref": "[images/image-8-Hi, guys. .jpg]", + "position": { + "start": 312, + "end": 343 + } + } + ] + } + }, + { + "chunk_id": "419bb53b-766a-578f-ab19-a89dbb2d071c", + "type": "image", + "content": "\nCo-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. ...at the magnitude of them, I mean, they're both about the same magnitude now. It almost sounded to me like you were suggesting that you thought the latter, the AI cloud would grow faster, maybe going forward than hyperscale. Is that what you were trying to say? Or do you see like the same kind of growth coming from both segments?0\n[images/image-9-Co-Founder.jpg]\n", + "path": "images/image-9-Co-Founder.jpg", + "metadata": { + "length": 440, + "summary": "image-9\nCo-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. ...at the magnitude of them, I mean, they're both about the same magnitude now. It almost sounded to me like you were suggesting that you thought the latter, the AI cloud would grow faster, maybe going forward than hyperscale. Is that what you were trying to say? Or do you see like the same kind of growth coming from both segments?0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-9-Co-Founder.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "3e5e8808-8362-5967-a07b-f3ccc71100af", + "type": "text", + "content": "Analyst, Bernstein Research\n\nCo-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. ...at the magnitude of them, I mean, they're both about the same magnitude now. It almost sounded to me like you were suggesting that you thought the latter, the AI cloud would grow faster, maybe going forward than hyperscale. Is that what you were trying to say? Or do you see like the same kind of growth coming from both segments?0\n[images/image-9-Co-Founder.jpg]\n\n...at the magnitude of them, I mean, they're both about the same magnitude now. It almost sounded to me like you were suggesting that you thought the latter, the AI cloud would grow faster, maybe going forward than hyperscale. Is that what you were trying to say? Or do you see like the same kind of growth coming from both segments?", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Stacy A. Rasgon_2", + "metadata": { + "length": 802, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Bernstein", + "Research", + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "magnitude", + "It", + "sounded", + "suggesting", + "thought", + "AI", + "cloud", + "grow", + "faster", + "forward", + "hyperscale", + "Is", + "Or", + "kind", + "growth", + "coming", + "segments" + ], + "keywords": [], + "connect_to": [ + { + "target": "419bb53b-766a-578f-ab19-a89dbb2d071c", + "relation": "embeds", + "ref": "[images/image-9-Co-Founder.jpg]", + "position": { + "start": 436, + "end": 467 + } + } + ] + } + }, + { + "chunk_id": "efd5eb5a-1470-51e9-83eb-e1ec17d7ad24", + "type": "image", + "content": "\nAnalyst, Bernstein Research First of all, you're correct that AI native clouds. AI native clouds don't build chips – don't design their own chips and they can't really assemble unrelated parts together into an AI factory. And their patience, their tolerance for time to first token is extremely low. And their need for an architecture that has a great deal of offtake, so that it runs every model has customers from everywhere, is incredibly high. And so that's the reason why NVIDIA's architecture is so perfect for them.0\n[images/image-10-Analyst, B.jpg]\n", + "path": "images/image-10-Analyst, B.jpg", + "metadata": { + "length": 558, + "summary": "image-10\nAnalyst, Bernstein Research First of all, you're correct that AI native clouds. AI native clouds don't build chips – don't design their own chips and they can't really assemble unrelated parts together into an AI factory. And their patience, their tolerance for time to first token is extremely low. And their need for an architecture that has a great deal of offtake, so that it runs every model has customers from everywhere, is incredibly high. And so that's the reason why NVIDIA's architecture is so perfect for them.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-10-Analyst, B.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "b82f3592-70ed-5749-8c6d-9aa1c3c234d4", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nAnalyst, Bernstein Research First of all, you're correct that AI native clouds. AI native clouds don't build chips – don't design their own chips and they can't really assemble unrelated parts together into an AI factory. And their patience, their tolerance for time to first token is extremely low. And their need for an architecture that has a great deal of offtake, so that it runs every model has customers from everywhere, is incredibly high. And so that's the reason why NVIDIA's architecture is so perfect for them.0\n[images/image-10-Analyst, B.jpg]\n\nFirst of all, you're correct that AI native clouds. AI native clouds don't build chips – don't design their own chips and they can't really assemble unrelated parts together into an AI factory. And their patience, their tolerance for time to first token is extremely low. And their need for an architecture that has a great deal of offtake, so that it runs every model has customers from everywhere, is incredibly high. And so that's the reason why NVIDIA's architecture is so perfect for them.\nWe offer every component and whatever we don't offer, our ecosystems of partners offer it. And it's all fully integrated. It all works together. The number of customers that could rent it from an AI native is incredibly high, basically, every single AI builder, every AI native startup around the world, SaaS companies, enterprise companies, industrial companies.\nAnd so, our computing – our architecture is the most rentable of any computing platform in the world. So it's the most performant. It's the easiest to put together. It's the most rentable. Has the best TCO. And it's the easiest to finance. And so all of those properties are quite unique to the needs of AI natives. It's in the second category. They're very similar to even OEMs and so on and so forth, large enterprises and so forth, surprisingly. Okay? So we put that in the second category.\nIf you look at that segment, it started growing after the AI ecosystem developed in the hyperscale. Hyperscale developed AI first for a lot of reasons. They have great computer science. They have excellent data center capability. And they also focus largely on consumer applications, which, if not perfect, is not the end of the world. It enhances the service, so long as it enhances the service.\nAnd so for many of the other applications, industrial applications, enterprise applications, until the AI is very capable and does really productive work and does it safely, and it could do it in a way that can actually generate impact and income, it doesn't really get used. And so you expect the second category to develop slower than hyperscale, and you could see that in the numbers.\nHowever, long term, if you look at industrial and enterprise, clearly, that's where future economics is going to be, because it represents some \\$50 trillion, \\$80 trillion of the world's economy. And it's going to be larger than that because of AI. And so I expect the second category to be larger over time, both in the near term, over the next several years. I think it's a foregone conclusion, both are going to grow incredibly fast. I expect the second category to still grow faster, but both are going to grow incredibly fast. And then, I'm hoping that within the next five years, physical AI and robotics segment is going to grow incredibly fast.\nOperator: Your next question comes from Jim Schneider with Goldman Sachs. Your line is open.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_7", + "metadata": { + "length": 3515, + "summary": "The speaker discusses how AI native clouds rely on NVIDIA's architecture because they don't design their own chips and need high performance, easy integration, and broad rentability. NVIDIA's platform is described as the most performant, rentable, and cost-effective for these clouds. The second category of customers (enterprises and industrial) is expected to grow slower than hyperscale but will become larger over time, representing significant economic value. The speaker predicts both segments will grow rapidly, with physical AI and robotics accelerating within five years.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Analyst", + "Bernstein", + "Research", + "First", + "correct", + "AI", + "native", + "clouds", + "don", + "build", + "chips", + "design", + "assemble", + "unrelated", + "parts", + "factory", + "And", + "patience", + "tolerance", + "time", + "token", + "extremely", + "low", + "architecture", + "great", + "deal", + "offtake", + "runs", + "model", + "customers", + "incredibly", + "high", + "reason", + "perfect", + "them.0", + "We", + "offer", + "component", + "ecosystems", + "partners", + "fully", + "integrated", + "It", + "works", + "The", + "number", + "rent", + "basically", + "single", + "builder", + "startup", + "world", + "SaaS", + "companies", + "enterprise", + "industrial", + "computing", + "rentable", + "platform", + "So", + "performant", + "easiest", + "put", + "Has", + "TCO", + "finance", + "properties", + "unique", + "natives", + "category", + "They", + "similar", + "OEMs", + "large", + "enterprises", + "surprisingly", + "Okay", + "If", + "segment", + "started", + "growing", + "ecosystem", + "developed", + "hyperscale", + "Hyperscale", + "lot", + "reasons", + "computer", + "science", + "excellent", + "data", + "center", + "capability", + "focus", + "largely", + "consumer", + "applications", + "end", + "enhances", + "service", + "long", + "capable", + "productive", + "work", + "safely", + "generate", + "impact", + "income", + "doesn", + "expect", + "develop", + "slower", + "numbers", + "However", + "term", + "future", + "economics", + "represents", + "50", + "trillion", + "80", + "economy", + "larger", + "years", + "foregone", + "conclusion", + "grow", + "fast", + "faster", + "hoping", + "physical", + "robotics", + "Operator", + "Your", + "question", + "Jim", + "Schneider", + "Goldman", + "Sachs", + "line", + "open" + ], + "keywords": [ + "AI native clouds", + "NVIDIA architecture", + "enterprise AI" + ], + "connect_to": [ + { + "target": "efd5eb5a-1470-51e9-83eb-e1ec17d7ad24", + "relation": "embeds", + "ref": "[images/image-10-Analyst, B.jpg]", + "position": { + "start": 597, + "end": 629 + } + } + ] + } + }, + { + "chunk_id": "5bdff50c-ec55-520d-ac0a-40148bc4b48c", + "type": "image", + "content": "\nAnd so for many of the other applications, industrial applications, enterprise applications, until the AI is very capable and does really productive work and does it safely, and it could do it in a way that can actually generate impact and income, it doesn't really get used. And so you expect the second category to develop slower than hyperscale, and you could see that in the numbers. Good afternoon. Thanks for taking my question. Back at GTC, I believe you discussed \\$1 trillion visibility into both your Rubin and Blackwell platform revenue. But I believe that excluded things like LPX, Rubin CPX and the Vera CPU racks. Can you maybe give us a sense about whether the Vera CPUs are going to be the biggest source of upside above and beyond that \\$1 trillion? Are you contemplating other sort of combinations of products, including CPUs that would allow you to gain an even greater share of that total TAM? Thank you.0\n[images/image-11-And so for.jpg]\n", + "path": "images/image-11-And so for.jpg", + "metadata": { + "length": 960, + "summary": "image-11\nAnd so for many of the other applications, industrial applications, enterprise applications, until the AI is very capable and does really productive work and does it safely, and it could do it in a way that can actually generate impact and income, it doesn't really get used. And so you expect the second category to develop slower than hyperscale, and you could see that in the numbers. Good afternoon. Thanks for taking my question. Back at GTC, I believe you discussed \\$1 trillion visibility into both your Rubin and Blackwell platform revenue. But I believe that excluded things like LPX, Rubin CPX and the Vera CPU racks. Can you maybe give us a sense about whether the Vera CPUs are going to be the biggest source of upside above and beyond that \\$1 trillion? Are you contemplating other sort of combinations of products, including CPUs that would allow you to gain an even greater share of that total TAM? Thank you.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-11-And so for.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "99466b4b-88ae-58bd-907e-9060eba7723f", + "type": "text", + "content": "Analyst, Goldman Sachs & Co. LLC\n\nAnd so for many of the other applications, industrial applications, enterprise applications, until the AI is very capable and does really productive work and does it safely, and it could do it in a way that can actually generate impact and income, it doesn't really get used. And so you expect the second category to develop slower than hyperscale, and you could see that in the numbers. Good afternoon. Thanks for taking my question. Back at GTC, I believe you discussed \\$1 trillion visibility into both your Rubin and Blackwell platform revenue. But I believe that excluded things like LPX, Rubin CPX and the Vera CPU racks. Can you maybe give us a sense about whether the Vera CPUs are going to be the biggest source of upside above and beyond that \\$1 trillion? Are you contemplating other sort of combinations of products, including CPUs that would allow you to gain an even greater share of that total TAM? Thank you.0\n[images/image-11-And so for.jpg]\n\nGood afternoon. Thanks for taking my question. Back at GTC, I believe you discussed \\$1 trillion visibility into both your Rubin and Blackwell platform revenue. But I believe that excluded things like LPX, Rubin CPX and the Vera CPU racks. Can you maybe give us a sense about whether the Vera CPUs are going to be the biggest source of upside above and beyond that \\$1 trillion? Are you contemplating other sort of combinations of products, including CPUs that would allow you to gain an even greater share of that total TAM? Thank you.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/James Edward Schneider", + "metadata": { + "length": 1530, + "summary": "An analyst from Goldman Sachs discusses NVIDIA's $1 trillion revenue visibility for Rubin and Blackwell platforms, excluding LPX, Rubin CPX, and Vera CPU racks. The question asks whether Vera CPUs will be the biggest upside beyond that $1 trillion and if other product combinations, including CPUs, could increase NVIDIA's share of the total addressable market.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Goldman", + "Sachs", + "Co", + "LLC", + "And", + "applications", + "industrial", + "enterprise", + "AI", + "capable", + "productive", + "work", + "safely", + "generate", + "impact", + "income", + "doesn", + "expect", + "category", + "develop", + "slower", + "hyperscale", + "numbers", + "Good", + "afternoon", + "Thanks", + "taking", + "question", + "Back", + "GTC", + "discussed", + "trillion", + "visibility", + "Rubin", + "Blackwell", + "platform", + "revenue", + "But", + "excluded", + "things", + "LPX", + "CPX", + "Vera", + "CPU", + "racks", + "Can", + "give", + "sense", + "CPUs", + "biggest", + "source", + "upside", + "Are", + "contemplating", + "sort", + "combinations", + "products", + "including", + "gain", + "greater", + "share", + "total", + "TAM", + "Thank", + "you.0" + ], + "keywords": [ + "AI", + "Revenue", + "CPUs" + ], + "connect_to": [ + { + "target": "5bdff50c-ec55-520d-ac0a-40148bc4b48c", + "relation": "embeds", + "ref": "[images/image-11-And so for.jpg]", + "position": { + "start": 960, + "end": 992 + } + } + ] + } + }, + { + "chunk_id": "0e78217a-8e13-53d8-af97-651578dca760", + "type": "image", + "content": "\nAnalyst, Goldman Sachs & Co. LLC In terms of incremental above the \\$1 trillion, I would say, one, the continued growing of share of the frontier AI models. I'm expecting to grow more share. And so I'm expecting that to grow. Number two, we didn't include any Vera CPU, standalone CPU in that number. And so I expect that to be the second largest. The TAM is, of course, quite large in agentic systems, and all of our customers are quite excited about Vera, and we're going to sell a whole bunch of Veras.0\n[images/image-12-Analyst, G.jpg]\n", + "path": "images/image-12-Analyst, G.jpg", + "metadata": { + "length": 541, + "summary": "image-12\nAnalyst, Goldman Sachs & Co. LLC In terms of incremental above the \\$1 trillion, I would say, one, the continued growing of share of the frontier AI models. I'm expecting to grow more share. And so I'm expecting that to grow. Number two, we didn't include any Vera CPU, standalone CPU in that number. And so I expect that to be the second largest. The TAM is, of course, quite large in agentic systems, and all of our customers are quite excited about Vera, and we're going to sell a whole bunch of Veras.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-12-Analyst, G.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e30221a8-416d-5ad7-b6a8-65bbcfea7229", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nAnalyst, Goldman Sachs & Co. LLC In terms of incremental above the \\$1 trillion, I would say, one, the continued growing of share of the frontier AI models. I'm expecting to grow more share. And so I'm expecting that to grow. Number two, we didn't include any Vera CPU, standalone CPU in that number. And so I expect that to be the second largest. The TAM is, of course, quite large in agentic systems, and all of our customers are quite excited about Vera, and we're going to sell a whole bunch of Veras.0\n[images/image-12-Analyst, G.jpg]\n\nIn terms of incremental above the \\$1 trillion, I would say, one, the continued growing of share of the frontier AI models. I'm expecting to grow more share. And so I'm expecting that to grow. Number two, we didn't include any Vera CPU, standalone CPU in that number. And so I expect that to be the second largest. The TAM is, of course, quite large in agentic systems, and all of our customers are quite excited about Vera, and we're going to sell a whole bunch of Veras.\nAnd then third would be LPX, because as I explained earlier, LPX is designed as a – because of its SRAM architecture, it has the benefit of very low latency and very high interactivity, but it's also its throughput and its context processing ability is also quite limited. And that's just kind of the nature of SRAM type-based systems. But the combination, we'll be able to address the entire spectrum of AI from pre-training to post-training to inference agentic systems through the combination of Vera Rubin and LPX.\nOperator: Your next question comes from Joshua Buchalter with TD Cowen. Your line is open.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_8", + "metadata": { + "length": 1696, + "summary": "The speaker discusses three incremental growth drivers above $1 trillion: gaining more share in frontier AI models, the Vera standalone CPU (not previously included), and LPX with its SRAM architecture for low latency. Vera is expected to be the second largest growth area, with strong customer excitement. LPX, while limited in throughput, complements Vera Rubin to address the full AI spectrum from pre-training to inference.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Analyst", + "Goldman", + "Sachs", + "LLC", + "In", + "terms", + "incremental", + "trillion", + "continued", + "growing", + "share", + "frontier", + "AI", + "models", + "expecting", + "grow", + "And", + "Number", + "didn", + "include", + "Vera", + "CPU", + "standalone", + "number", + "expect", + "largest", + "The", + "TAM", + "large", + "agentic", + "systems", + "customers", + "excited", + "sell", + "bunch", + "Veras.0", + "Veras", + "LPX", + "explained", + "earlier", + "designed", + "SRAM", + "architecture", + "benefit", + "low", + "latency", + "high", + "interactivity", + "throughput", + "context", + "processing", + "ability", + "limited", + "kind", + "nature", + "type", + "based", + "But", + "combination", + "ll", + "address", + "entire", + "spectrum", + "pre", + "training", + "post", + "inference", + "Rubin", + "Operator", + "Your", + "question", + "Joshua", + "Buchalter", + "TD", + "Cowen", + "line", + "open" + ], + "keywords": [ + "AI models", + "Vera CPU", + "LPX" + ], + "connect_to": [ + { + "target": "0e78217a-8e13-53d8-af97-651578dca760", + "relation": "embeds", + "ref": "[images/image-12-Analyst, G.jpg]", + "position": { + "start": 580, + "end": 612 + } + } + ] + } + }, + { + "chunk_id": "ebd8d672-ee16-54a2-a1e0-40c6bb0f7b1b", + "type": "image", + "content": "\nIn terms of incremental above the \\$1 trillion, I would say, one, the continued growing of share of the frontier AI models. I'm expecting to grow more share. And so I'm expecting that to grow. Number two, we didn't include any Vera CPU, standalone CPU in that number. And so I expect that to be the second largest. The TAM is, of course, quite large in agentic systems, and all of our customers are quite excited about Vera, and we're going to sell a whole bunch of Veras. Hey, guys. Thanks for taking my question, and congrats on the great results. Colette, I believe in your prepared remarks, you mentioned GB300 is sort of the fastest ramp in the company's history. How should we think about Vera Rubin against this benchmark? It's, obviously, a new architecture at the silicon level, but similar rack. Does that mean we should expect a similar slope to the Vera Rubin ramp as the GB300? Or should it be a bit more gradual given the new silicon? Thank you.0\n[images/image-13-In terms o.jpg]\n", + "path": "images/image-13-In terms o.jpg", + "metadata": { + "length": 995, + "summary": "image-13\nIn terms of incremental above the \\$1 trillion, I would say, one, the continued growing of share of the frontier AI models. I'm expecting to grow more share. And so I'm expecting that to grow. Number two, we didn't include any Vera CPU, standalone CPU in that number. And so I expect that to be the second largest. The TAM is, of course, quite large in agentic systems, and all of our customers are quite excited about Vera, and we're going to sell a whole bunch of Veras. Hey, guys. Thanks for taking my question, and congrats on the great results. Colette, I believe in your prepared remarks, you mentioned GB300 is sort of the fastest ramp in the company's history. How should we think about Vera Rubin against this benchmark? It's, obviously, a new architecture at the silicon level, but similar rack. Does that mean we should expect a similar slope to the Vera Rubin ramp as the GB300? Or should it be a bit more gradual given the new silicon? Thank you.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-13-In terms o.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "a9daaf8b-e409-5101-b054-2ed6b6042507", + "type": "text", + "content": "Analyst, TD Cowen\n\nIn terms of incremental above the \\$1 trillion, I would say, one, the continued growing of share of the frontier AI models. I'm expecting to grow more share. And so I'm expecting that to grow. Number two, we didn't include any Vera CPU, standalone CPU in that number. And so I expect that to be the second largest. The TAM is, of course, quite large in agentic systems, and all of our customers are quite excited about Vera, and we're going to sell a whole bunch of Veras. Hey, guys. Thanks for taking my question, and congrats on the great results. Colette, I believe in your prepared remarks, you mentioned GB300 is sort of the fastest ramp in the company's history. How should we think about Vera Rubin against this benchmark? It's, obviously, a new architecture at the silicon level, but similar rack. Does that mean we should expect a similar slope to the Vera Rubin ramp as the GB300? Or should it be a bit more gradual given the new silicon? Thank you.0\n[images/image-13-In terms o.jpg]\n\nHey, guys. Thanks for taking my question, and congrats on the great results. Colette, I believe in your prepared remarks, you mentioned GB300 is sort of the fastest ramp in the company's history. How should we think about Vera Rubin against this benchmark? It's, obviously, a new architecture at the silicon level, but similar rack. Does that mean we should expect a similar slope to the Vera Rubin ramp as the GB300? Or should it be a bit more gradual given the new silicon? Thank you.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Joshua Buchalter", + "metadata": { + "length": 1500, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "TD", + "Cowen", + "In", + "terms", + "incremental", + "trillion", + "continued", + "growing", + "share", + "frontier", + "AI", + "models", + "expecting", + "grow", + "And", + "Number", + "didn", + "include", + "Vera", + "CPU", + "standalone", + "number", + "expect", + "largest", + "The", + "TAM", + "large", + "agentic", + "systems", + "customers", + "excited", + "sell", + "bunch", + "Veras", + "Hey", + "guys", + "Thanks", + "taking", + "question", + "congrats", + "great", + "results", + "Colette", + "prepared", + "remarks", + "mentioned", + "GB300", + "sort", + "fastest", + "ramp", + "company", + "history", + "How", + "Rubin", + "benchmark", + "It", + "architecture", + "silicon", + "level", + "similar", + "rack", + "Does", + "slope", + "Or", + "bit", + "gradual", + "Thank", + "you.0" + ], + "keywords": [], + "connect_to": [ + { + "target": "ebd8d672-ee16-54a2-a1e0-40c6bb0f7b1b", + "relation": "embeds", + "ref": "[images/image-13-In terms o.jpg]", + "position": { + "start": 980, + "end": 1012 + } + } + ] + } + }, + { + "chunk_id": "3615c5e5-720c-5fda-9793-4b7bc7dd2013", + "type": "image", + "content": "\nAnalyst, TD Cowen Yeah. Well, we have indicated for a while that we will be launching Vera Rubin in the second half. We will start in Q3. That will be our initial pieces together. And then once we get to Q4, we're probably going to start to see our ramping continue. It's hard to say at this point what will be a faster ramp. But again, we have demand already planned. We've got POs. We've got almost all of our major customers ready to go. And these are very complex systems that we need to put together. So I think it's just about the timing that it's going to take for us to get that into market.0\n[images/image-14-Analyst, T.jpg]\n", + "path": "images/image-14-Analyst, T.jpg", + "metadata": { + "length": 635, + "summary": "image-14\nAnalyst, TD Cowen Yeah. Well, we have indicated for a while that we will be launching Vera Rubin in the second half. We will start in Q3. That will be our initial pieces together. And then once we get to Q4, we're probably going to start to see our ramping continue. It's hard to say at this point what will be a faster ramp. But again, we have demand already planned. We've got POs. We've got almost all of our major customers ready to go. And these are very complex systems that we need to put together. So I think it's just about the timing that it's going to take for us to get that into market.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-14-Analyst, T.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "aaf4f90e-18da-53a4-9c55-6edfe2cb312c", + "type": "text", + "content": "Chief Financial Officer & Executive Vice President, NVIDIA Corp.\n\nAnalyst, TD Cowen Yeah. Well, we have indicated for a while that we will be launching Vera Rubin in the second half. We will start in Q3. That will be our initial pieces together. And then once we get to Q4, we're probably going to start to see our ramping continue. It's hard to say at this point what will be a faster ramp. But again, we have demand already planned. We've got POs. We've got almost all of our major customers ready to go. And these are very complex systems that we need to put together. So I think it's just about the timing that it's going to take for us to get that into market.0\n[images/image-14-Analyst, T.jpg]\n\nYeah. Well, we have indicated for a while that we will be launching Vera Rubin in the second half. We will start in Q3. That will be our initial pieces together. And then once we get to Q4, we're probably going to start to see our ramping continue. It's hard to say at this point what will be a faster ramp. But again, we have demand already planned. We've got POs. We've got almost all of our major customers ready to go. And these are very complex systems that we need to put together. So I think it's just about the timing that it's going to take for us to get that into market.\nNothing else other than getting from production of all of the different systems that we have ready for order. So, a little early to say, but yes, we're going to start in Q3 and continue to ramp into Q4 and Q1 of next year certainly is going to be very big as well.\nOperator: There are no further questions at this time. Toshiya Hari, I turn the call back over to you.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Colette M. Kress", + "metadata": { + "length": 1650, + "summary": "The CFO and EVP of NVIDIA discusses the upcoming launch of Vera Rubin in the second half of the year, starting in Q3 with initial pieces and continuing to ramp into Q4 and Q1 of next year. Demand is already planned with purchase orders from major customers, though the ramp speed is uncertain due to system complexity. Production timing is the key factor for market entry.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Chief", + "Financial", + "Officer", + "Executive", + "Vice", + "President", + "NVIDIA", + "Corp", + "Analyst", + "TD", + "Cowen", + "Yeah", + "Well", + "launching", + "Vera", + "Rubin", + "half", + "We", + "start", + "Q3", + "That", + "initial", + "pieces", + "And", + "Q4", + "ramping", + "continue", + "It", + "hard", + "point", + "faster", + "ramp", + "But", + "demand", + "planned", + "ve", + "POs", + "major", + "customers", + "ready", + "complex", + "systems", + "put", + "So", + "timing", + "market.0", + "market", + "Nothing", + "production", + "order", + "early", + "Q1", + "year", + "big", + "Operator", + "There", + "questions", + "time", + "Toshiya", + "Hari", + "turn", + "call", + "back" + ], + "keywords": [ + "Vera Rubin", + "Q3 launch", + "ramp" + ], + "connect_to": [ + { + "target": "3615c5e5-720c-5fda-9793-4b7bc7dd2013", + "relation": "embeds", + "ref": "[images/image-14-Analyst, T.jpg]", + "position": { + "start": 667, + "end": 699 + } + } + ] + } + }, + { + "chunk_id": "40417f08-9cf7-582d-9264-a6f2b3d92847", + "type": "text", + "content": "Vice President-Investor Relations & Strategic Finance, NVIDIA Corp.\nThank you. Before I hand it over to Jensen, please note Jensen will be giving a keynote at GTC Taipei at COMPUTEX on June 1. We will also be participating at the TD Cowen TMT Conference on May 28, and the Bank of America Global Technology Conference on June 4. Our earnings call to discuss the results of our second quarter of fiscal 2027 is scheduled for August 26.\nWith that, here's Jensen to close us out.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Toshiya Hari", + "metadata": { + "length": 476, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Vice", + "President", + "Investor", + "Relations", + "Strategic", + "Finance", + "NVIDIA", + "Corp", + "Thank", + "Before", + "hand", + "Jensen", + "note", + "giving", + "keynote", + "GTC", + "Taipei", + "COMPUTEX", + "June", + "We", + "participating", + "TD", + "Cowen", + "TMT", + "Conference", + "May", + "28", + "Bank", + "America", + "Global", + "Technology", + "Our", + "earnings", + "call", + "discuss", + "results", + "quarter", + "fiscal", + "2027", + "scheduled", + "August", + "26", + "With", + "close" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b7e3b32b-f43e-5146-b405-ca9ac59d884a", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\nThis was an extraordinary quarter. Demand has gone parabolic. The reason is simple. Agentic AI has arrived. AI can now do productive and valuable work. Tokens are now profitable. So model makers are in a race to produce more. In the AI era, compute capacity is revenue and profits. NVIDIA is the platform of this era. Of all the platforms in the world, NVIDIA Compute supports the richest diversity of demand. Let me highlight my top five things.\nFirst, NVIDIA is the only platform that runs every frontier AI model. With the addition of Anthropic to our existing partners, OpenAI, xAI, Meta MSL, Gemini, and many others, our share of frontier AI is growing.\nSecond, we are in every hyperscale cloud, supporting their core data processing and machine learning workloads, internal AI services, as well as supporting their demand for NVIDIA users in their public cloud services.\nThird, our full stack complete AI factory solution and vast global ecosystem let us uniquely address new AI data center segments, mew AI native clouds and sovereign AI clouds and on-premises enterprise and industrial infrastructure. This is that second category I was talking about earlier.\nFourth, NVIDIA CUDA extends all the way to the edge. Robotics, autonomous vehicles, embedded medical instruments. AI-RAN, telco base stations. The next wave is physical AI with billions of autonomous and robotic systems operating in the physical world. This is the third segment we were talking about earlier.\nAnd rounding out the top five things, we have a major new growth driver, Vera, the world's first CPU purpose built for agentic AI. Vera opens a brand new \\$200 billion TAM for NVIDIA, a market we have never addressed before. And every major hyperscaler and system maker is partnering with us to deploy it.\nThe world is rebuilding computing for agentic AI and robotic physical AI. NVIDIA sits at the center of these transitions. We built NVIDIA Compute platform over three decades, one architecture, vast ecosystem, extreme co-design across chips, systems, networking, and software. We built it ahead of this moment, so that when agentic AI arrived, NVIDIA would be ready. It has arrived.\nLook forward to catching up next time.\nOperator: This concludes today's conference call. You may now disconnect.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_9", + "metadata": { + "length": 2350, + "summary": "NVIDIA's CEO highlights an extraordinary quarter with parabolic demand driven by the arrival of Agentic AI, which makes tokens profitable. He lists five key points: NVIDIA runs every frontier AI model, is in every hyperscale cloud, offers full-stack AI factory solutions, extends CUDA to edge and physical AI, and introduces Vera, a CPU for agentic AI opening a $200 billion TAM. The world is rebuilding computing for agentic and physical AI, and NVIDIA is ready with its three-decade platform.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "This", + "extraordinary", + "quarter", + "Demand", + "parabolic", + "The", + "reason", + "simple", + "Agentic", + "AI", + "arrived", + "productive", + "valuable", + "work", + "Tokens", + "profitable", + "So", + "model", + "makers", + "race", + "produce", + "In", + "era", + "compute", + "capacity", + "revenue", + "profits", + "platform", + "Of", + "platforms", + "world", + "Compute", + "supports", + "richest", + "diversity", + "demand", + "Let", + "highlight", + "top", + "things", + "First", + "runs", + "frontier", + "With", + "addition", + "Anthropic", + "existing", + "partners", + "OpenAI", + "xAI", + "Meta", + "MSL", + "Gemini", + "share", + "growing", + "Second", + "hyperscale", + "cloud", + "supporting", + "core", + "data", + "processing", + "machine", + "learning", + "workloads", + "internal", + "services", + "users", + "public", + "Third", + "full", + "stack", + "complete", + "factory", + "solution", + "vast", + "global", + "ecosystem", + "uniquely", + "address", + "center", + "segments", + "mew", + "native", + "clouds", + "sovereign", + "premises", + "enterprise", + "industrial", + "infrastructure", + "category", + "talking", + "earlier", + "Fourth", + "CUDA", + "extends", + "edge", + "Robotics", + "autonomous", + "vehicles", + "embedded", + "medical", + "instruments", + "RAN", + "telco", + "base", + "stations", + "wave", + "physical", + "billions", + "robotic", + "systems", + "operating", + "segment", + "And", + "rounding", + "major", + "growth", + "driver", + "Vera", + "CPU", + "purpose", + "built", + "agentic", + "opens", + "brand", + "200", + "billion", + "TAM", + "market", + "addressed", + "hyperscaler", + "system", + "maker", + "partnering", + "deploy", + "rebuilding", + "computing", + "sits", + "transitions", + "We", + "decades", + "architecture", + "extreme", + "design", + "chips", + "networking", + "software", + "ahead", + "moment", + "ready", + "It", + "Look", + "forward", + "catching", + "time", + "Operator", + "concludes", + "today", + "conference", + "call", + "You", + "disconnect" + ], + "keywords": [ + "Agentic AI", + "NVIDIA Compute", + "Vera CPU" + ], + "connect_to": [] + } + }, + { + "chunk_id": "a0a13abb-a54b-5bf1-aa4a-accd099a41be", + "type": "text", + "content": "The information herein is based on sources we believe to be reliable but is not guaranteed by us and does not purport to be a complete or error-free statement or summary of the available data. As such, we do not warrant, endorse or guarantee the completeness, accuracy, integrity, or timeliness of the information. You must evaluate, and bear all risks associated with, the use of any information provided hereunder, including any reliance on the accuracy, completeness, safety or usefulness of such information. This information is not intended to be used as the primary basis of investment decisions. It should not be construed as advice designed to meet the particular investment needs of any investor. This report is published solely for information purposes, and is not to be construed as financial or other advice or as an offer to sell or the solicitation of an offer to buy any security in any state where such an offer or solicitation would be illegal. Any information expressed herein on this date is subject to change without notice. Any opinions or assertions contained in this information do not represent the opinions or beliefs of FactSet CallStreet, LLC. FactSet CallStreet, LLC, or one or more of its employees, including the writer of this report, may have a position in any of the securities discussed here in.\nTHE INFORMATION PROVIDED TO YOU HEREUNDER IS PROVIDED \"AS IS,\" AND TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, FactSet CallStreet, LLC AND ITS LICENSORS, BUSINESS ASSOCIATES AND SUPPLIERS DISCLAIM ALL WARRANTIES WITH RESPECT TO THE SAME, EXPRESS, IMPLIED AND STATUTORY, INCLUDING WITHOUT LIMITATION ANY IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, ACCURACY, COMPLETENESS, AND NON-INFRINGEMENT. TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, NEITHER FACTSET CALLSTREET, LLC NOR ITS OFFICERS, MEMBERS, DIRECTORS, PARTNERS, AFFILIATES, BUSINESS ASSOCIATES, LICENSORS OR SUPPLIERS WILL BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL OR PUNITIVE DAMAGES, INCLUDING WITHOUT LIMITATION DAMAGES FOR LOST PROFITS OR REVENUES, GOODWILL, WORK STOPPAGE, SECURITY BREACHES, VIRUSES, COMPUTER FAILURE OR MALFUNCTION, USE, DATA OR OTHER INTANGIBLE LOSSES OR COMMERCIAL DAMAGES, EVEN IF ANY OF SUCH PARTIES IS ADVISED OF THE POSSIBILITY OF SUCH LOSSES, ARISING UNDER OR IN CONNECTION WITH THE INFORMATION PROVIDED HEREIN OR ANY OTHER SUBJECT MATTER HEREOF.\nThe contents and appearance of this report are Copyrighted FactSet CallStreet, LLC 2026 CallStreet and FactSet CallStreet, LLC are trademarks and service marks of FactSet CallStreet, LLC. All other trademarks mentioned are trademarks of their respective companies. All rights reserved.", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/Disclaimer", + "metadata": { + "length": 2711, + "summary": "This document is a legal disclaimer stating that the information provided is based on reliable sources but is not guaranteed for completeness, accuracy, or timeliness. It warns against using the information as a primary basis for investment decisions and disclaims all warranties and liability for damages. The content is copyrighted by FactSet CallStreet, LLC.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "The", + "information", + "based", + "sources", + "reliable", + "guaranteed", + "purport", + "complete", + "error", + "free", + "statement", + "summary", + "data", + "As", + "warrant", + "endorse", + "guarantee", + "completeness", + "accuracy", + "integrity", + "timeliness", + "You", + "evaluate", + "bear", + "risks", + "provided", + "hereunder", + "including", + "reliance", + "safety", + "usefulness", + "This", + "intended", + "primary", + "basis", + "investment", + "decisions", + "It", + "construed", + "advice", + "designed", + "meet", + "investor", + "report", + "published", + "solely", + "purposes", + "financial", + "offer", + "sell", + "solicitation", + "buy", + "security", + "state", + "illegal", + "Any", + "expressed", + "date", + "subject", + "change", + "notice", + "opinions", + "assertions", + "contained", + "represent", + "beliefs", + "FactSet", + "CallStreet", + "LLC", + "employees", + "writer", + "position", + "securities", + "discussed", + "THE", + "INFORMATION", + "PROVIDED", + "TO", + "YOU", + "HEREUNDER", + "IS", + "AS", + "AND", + "MAXIMUM", + "EXTENT", + "PERMITTED", + "BY", + "APPLICABLE", + "LAW", + "ITS", + "LICENSORS", + "BUSINESS", + "ASSOCIATES", + "SUPPLIERS", + "DISCLAIM", + "ALL", + "WARRANTIES", + "WITH", + "RESPECT", + "SAME", + "EXPRESS", + "IMPLIED", + "STATUTORY", + "INCLUDING", + "WITHOUT", + "LIMITATION", + "ANY", + "OF", + "MERCHANTABILITY", + "FITNESS", + "FOR", + "PARTICULAR", + "PURPOSE", + "ACCURACY", + "COMPLETENESS", + "NON", + "INFRINGEMENT", + "NEITHER", + "FACTSET", + "CALLSTREET", + "NOR", + "OFFICERS", + "MEMBERS", + "DIRECTORS", + "PARTNERS", + "AFFILIATES", + "OR", + "WILL", + "BE", + "LIABLE", + "INDIRECT", + "INCIDENTAL", + "SPECIAL", + "CONSEQUENTIAL", + "PUNITIVE", + "DAMAGES", + "LOST", + "PROFITS", + "REVENUES", + "GOODWILL", + "WORK", + "STOPPAGE", + "SECURITY", + "BREACHES", + "VIRUSES", + "COMPUTER", + "FAILURE", + "MALFUNCTION", + "USE", + "DATA", + "OTHER", + "INTANGIBLE", + "LOSSES", + "COMMERCIAL", + "EVEN", + "IF", + "SUCH", + "PARTIES", + "ADVISED", + "POSSIBILITY", + "ARISING", + "UNDER", + "IN", + "CONNECTION", + "HEREIN", + "SUBJECT", + "MATTER", + "HEREOF", + "contents", + "appearance", + "Copyrighted", + "2026", + "trademarks", + "service", + "marks", + "All", + "mentioned", + "respective", + "companies", + "rights", + "reserved" + ], + "keywords": [ + "Disclaimer", + "Investment", + "Liability" + ], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/doc_nav.json b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/doc_nav.json new file mode 100644 index 000000000..fe9a36f87 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/doc_nav.json @@ -0,0 +1,317 @@ +{ + "version": "1.0", + "file_name": "NVDA Q1 FY27 Earnings Call Transcript.pdf", + "stats": { + "total_chunks": 44, + "text_chunks": 30, + "image_chunks": 14, + "table_chunks": 0, + "max_depth": 2 + }, + "sections": [ + { + "title": "Root", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf", + "level": 1, + "summary": "20-May-2026", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corp. (NVDA)", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)", + "level": 1, + "summary": "Q1 2027 Earnings Call", + "chunk_count": 3, + "children": [ + { + "title": "CORPORATE PARTICIPANTS", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)/CORPORATE PARTICIPANTS", + "level": 2, + "summary": "Toshiya Hari Vice President-Investor Relations & Strategic Finance, NVIDIA Corp. Colette M. Kress Chief Financial Officer & Executive Vice President, NVIDIA Corp. Jen Hsun Huang Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.", + "chunk_count": 1, + "children": [] + }, + { + "title": "OTHER PARTICIPANTS", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)/OTHER PARTICIPANTS", + "level": 2, + "summary": "Joseph Moore Analyst, Morgan Stanley & Co. LLC Ben Reitzes Analyst, Melius Research LLC C.J. Muse Analyst, Cantor Fitzgerald & Co. Timothy Arcuri Analyst, UBS Securities LLC Vivek Arya Analyst, BofA Securities, Inc. Stacy A. Rasgon Analyst, Bernstein Research James Edward Schneider Analyst, Goldman Sachs & Co. LLC Joshua Buchalter Analyst, TD Cowen", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "MANAGEMENT DISCUSSION SECTION", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION", + "level": 1, + "summary": "Operator: Good afternoon. My name is Sarah, and I will be your conference operator today. At this time, I would like to welcome everyone to NVIDIA's First Quarter Earnings Call. All lines have been placed on mute to prevent any background noise. After the speakers' remarks, there will be a question-and-answer session. [Operator Instructions] Thank you. Toshiya Hari, you may begin your conference.", + "chunk_count": 4, + "children": [ + { + "title": "Toshiya Hari", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Toshiya Hari", + "level": 2, + "summary": "NVIDIA held its Q1 fiscal 2027 earnings call on May 20, 2026, with CEO Jensen Huang and CFO Colette Kress. The call included forward-looking statements and discussion of non-GAAP financial measures. The webcast replay is available until the Q2 fiscal 2027 call.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Colette M. Kress", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Colette M. Kress", + "level": 2, + "summary": "NVIDIA reported record Q1 revenue of $82 billion, up 85% year-over-year, driven by Data Center revenue of $75 billion (up 92% YoY) led by Blackwell architecture. The company highlighted strong demand for GB300 NVL72, rapid AI infrastructure build-out, and a transition to a new reporting framework with Hyperscale and ACIE sub-markets. Key drivers include CPU-to-GPU transition and AI native product adoption. Edge Computing generated $6.4 billion. Q2 revenue outlook is $91 billion. The company announced an $80 billion share repurchase authorization and increased dividend to $0.25 per share.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Toshiya Hari_2", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Toshiya Hari_2", + "level": 2, + "summary": "Vice President-Investor Relations & Strategic Finance, NVIDIA Corp. Thanks, Colette. We will now transition to Q&A. Operator, please poll for questions.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "QUESTION AND ANSWER SECTION", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION", + "level": 1, + "summary": "Operator: Thank you. [Operator Instructions] Your first question comes from Joseph Moore with Morgan Stanley. Your lie is open.", + "chunk_count": 21, + "children": [ + { + "title": "Joseph Moore", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Joseph Moore", + "level": 2, + "summary": "Analyst, Morgan Stanley & Co. LLC Thanks, Colette. We will now transition to Q&A. Operator, please poll for questions. Great. Thank you for letting me ask a question. I guess, I'd like to ask, what drove the change in segmentation? What's the philosophy behind giving us the numbers that way? And then, can you talk about any competitive differences between the two segments and this kind of surprising CPU number that you talked about, how do you see that across the two segments as well? Thank you.0 [images/image-1-Thanks, Co.jpg] Great. Thank you for letting me ask a question. I guess, I'd like to ask, what drove the change in segmentation? What's the philosophy behind giving us the numbers that way? And then, can you talk about any competitive differences between the two segments and this kind of surprising CPU number that you talked about, how do you see that across the two segments as well? Thank you.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang", + "level": 2, + "summary": "NVIDIA's CEO explains the company's new business segmentation into three categories: hyperscale clouds, AI natives/enterprise/industrial/sovereign AI, and robotic edge. He describes how AI is diverse across languages, industries, and applications, and how NVIDIA provides a full-stack, integrated platform that is still open. The segmentation helps investors understand NVIDIA's evolving business, which now serves a wide range of customers from a few hyperscalers to 250,000 companies worldwide.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Ben Reitzes", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Ben Reitzes", + "level": 2, + "summary": "Analyst, Melius Research LLC And as you know, NVIDIA has the largest suite of acceleration libraries in the world from computational lithography to fluid dynamics to particle physics to molecular dynamics to the list goes on. And all of those libraries are essential for us to engage the vertical industries that represents the second and the third category. Okay? Hey, guys. Thanks so much. I wanted to ask, Jensen, I want to ask you about your philosophy on growth. Your Data Center business ex-China grew about 120% in the quarter. And then, you're guiding about 100%. CapEx at the hyperscalers is forecast by many, including myself, to like grow 90% to 100% this year.0 [images/image-3-And as you.jpg] Hey, guys. Thanks so much. I wanted to ask, Jensen, I want to ask you about your philosophy on growth. Your Data Center business ex-China grew about 120% in the quarter. And then, you're guiding about 100%. CapEx at the hyperscalers is forecast by many, including myself, to like grow 90% to 100% this year. And you talked about Data Center still on track to be \\$3 trillion to \\$4 trillion by the end of the decade. I was just wondering, the goal for the company to grow faster than hyperscaler CapEx, are you comfortable in kind of endorsing that view? And do you still see hyperscaler CapEx kind of still growing after this year at a very rapid clip? Thanks a lot.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_2", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_2", + "level": 2, + "summary": "NVIDIA's CEO explains that the company's Data Center business should grow faster than hyperscale CapEx, which is at $1 trillion and expected to grow to $3-4 trillion. The business is divided into two segments: hyperscalers (large cloud providers) and a diverse second category including AI-native clouds, enterprise companies, industrial firms, and sovereign AI clouds. The second category, involving hundreds of thousands of companies, is growing rapidly and represents the future of AI impacting industries not previously transformed by IT. NVIDIA's vertically integrated platform uniquely serves this diverse market, with strong share in both segments.", + "chunk_count": 1, + "children": [] + }, + { + "title": "C.J. Muse", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/C.J. Muse", + "level": 2, + "summary": "The speaker discusses how physical AI is poised to impact the $100 trillion industry that has not been affected by IT in the last 30 years. This segment is growing rapidly, and the company has a large share due to its vertically integrated platform that can be disassembled for flexible configuration. In response to a question about Vera Rubin and extreme co-engineering, the speaker addresses how these advancements will affect the company's share in the inference market looking into late 2026 and 2027.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_3", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_3", + "level": 2, + "summary": "NVIDIA's Co-Founder and CEO discusses the company's rapid growth in AI inference market share. Key points include: the number of frontier model companies expanding (Cursor, Perplexity, TML, Reflection), new partnership with Anthropic providing computing capacity across Azure, AWS, and CoreWeave, and Vera Rubin expected to be more successful than Grace Blackwell. The CEO emphasizes NVIDIA's dominance in inference for hyperscale customers, unique position in fragmented AI data center segment, and leadership in physical AI.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Timothy Arcuri", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Timothy Arcuri", + "level": 2, + "summary": "Analyst, UBS Securities LLC Let me go back again to the question that Ben was asking. Remember, so far, everything that I've just explained in the inference question is really focused on hyperscale. Remember, there's a whole second category of AI data centers that we serve almost uniquely. This segment is very fragmented, requires a really well-integrated platform solution and a very large go-to-market, and that segment, all of the inference, 100% of that, the vast majority of that is NVIDIA. Thanks a lot. Jensen, I wanted to ask about the traction you're getting with some of these custom merchant things. You're doing stuff like CPX and LPX. And I just wanted to ask and see sort of you've talked before about [ph] SaaS LPX (00:37:50) being, I think, 20% of the market. So, I would imagine you're getting pretty good traction with LPX. So, can you just talk about that and maybe also how that fits into your broader platform strategy? Thanks.0 [images/image-6-Let me go .jpg] Thanks a lot. Jensen, I wanted to ask about the traction you're getting with some of these...of you've talked before about [ph] SaaS LPX (00:37:50) being, I think, 20% of the market. So, I would imagine you're getting pretty good traction with LPX. So, can you just talk about that and maybe also how that fits into your broader platform strategy? Thanks.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_4", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_4", + "level": 2, + "summary": "The speaker discusses the LPX accelerator, designed for low latency and high token rate but with low throughput and limited context processing. It is a niche product for premium token services with few customers. In contrast, Grace Blackwell and Vera Rubin support the full AI lifecycle. The speaker estimates LPX's market share is currently less than 20% but could grow to 20% in the future as AI develops.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Vivek Arya", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Vivek Arya", + "level": 2, + "summary": "Analyst, BofA Securities, Inc. Q Thanks for taking my question. Jensen, there's a lot of excitement around CPU for agentic applications and just a lot of noise around the number of CPUs actually exceeding the number of GPUs. And I was just hoping that you could kind of give your perspective that, first of all, is this an incremental workload? Is this kind of cannibalizing what the GPU would have done otherwise? And then, secondly, the \\$20 billion number that you gave, is that for standalone Vera CPUs, or is that kind of already included in that Vera as part of Vera Rubin? So just if you could educate us on the role of CPU versus GPU. Is it cannibalistic? Is it incremental? And then, the \\$20 billion number, how to kind of put that in context with what you sell, right, which is usually the CPU as part of the GPU. Thank you.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_5", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_5", + "level": 2, + "summary": "The speaker discusses NVIDIA's Vera CPU designed for AI agents, highlighting four use cases: with Vera Rubin GPUs, standalone, with CX9 for storage, and for security. They contrast past CPU economics (dollars per core) with future AI economics (dollars per token), emphasizing Vera's speed for token processing. Billions of future agents will use CPUs for orchestration and tools, while GPUs handle inference. NVIDIA is accelerating tools on CUDA and building complete AI infrastructure including GPUs, networking, storage, and security.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stacy A. Rasgon", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Stacy A. Rasgon", + "level": 2, + "summary": "Analyst, Bernstein Research So, we're going to need a lot more CPUs, and Vera was designed to be an agentic CPU. The CPUs of the past were designed to have many cores, so that it could be easily rentable. People rented cores. Well, agents don't rent cores. They just want the work to be done fast. The economics of the past was dollars per core. That's the economics of cloud computing of the past. The economics of AI of the future is tokens per dollar or dollars per token. And so, what we need to do in the future is to generate tokens, process tokens as fast as possible. And that's what Vera does incredibly well. Hi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud?0 [images/image-7-So, we're .jpg] Hi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud? Part of me assumes the latter, but I'm not so sure. And then, by just looking...", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_6", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_6", + "level": 2, + "summary": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. Hi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud? Correct.0 [images/image-8-Hi, guys. .jpg] Correct.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stacy A. Rasgon_2", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Stacy A. Rasgon_2", + "level": 2, + "summary": "Analyst, Bernstein Research Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. ...at the magnitude of them, I mean, they're both about the same magnitude now. It almost sounded to me like you were suggesting that you thought the latter, the AI cloud would grow faster, maybe going forward than hyperscale. Is that what you were trying to say? Or do you see like the same kind of growth coming from both segments?0 [images/image-9-Co-Founder.jpg] ...at the magnitude of them, I mean, they're both about the same magnitude now. It almost sounded to me like you were suggesting that you thought the latter, the AI cloud would grow faster, maybe going forward than hyperscale. Is that what you were trying to say? Or do you see like the same kind of growth coming from both segments?", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_7", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_7", + "level": 2, + "summary": "The speaker discusses how AI native clouds rely on NVIDIA's architecture because they don't design their own chips and need high performance, easy integration, and broad rentability. NVIDIA's platform is described as the most performant, rentable, and cost-effective for these clouds. The second category of customers (enterprises and industrial) is expected to grow slower than hyperscale but will become larger over time, representing significant economic value. The speaker predicts both segments will grow rapidly, with physical AI and robotics accelerating within five years.", + "chunk_count": 1, + "children": [] + }, + { + "title": "James Edward Schneider", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/James Edward Schneider", + "level": 2, + "summary": "An analyst from Goldman Sachs discusses NVIDIA's $1 trillion revenue visibility for Rubin and Blackwell platforms, excluding LPX, Rubin CPX, and Vera CPU racks. The question asks whether Vera CPUs will be the biggest upside beyond that $1 trillion and if other product combinations, including CPUs, could increase NVIDIA's share of the total addressable market.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_8", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_8", + "level": 2, + "summary": "The speaker discusses three incremental growth drivers above $1 trillion: gaining more share in frontier AI models, the Vera standalone CPU (not previously included), and LPX with its SRAM architecture for low latency. Vera is expected to be the second largest growth area, with strong customer excitement. LPX, while limited in throughput, complements Vera Rubin to address the full AI spectrum from pre-training to inference.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Joshua Buchalter", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Joshua Buchalter", + "level": 2, + "summary": "Analyst, TD Cowen In terms of incremental above the \\$1 trillion, I would say, one, the continued growing of share of the frontier AI models. I'm expecting to grow more share. And so I'm expecting that to grow. Number two, we didn't include any Vera CPU, standalone CPU in that number. And so I expect that to be the second largest. The TAM is, of course, quite large in agentic systems, and all of our customers are quite excited about Vera, and we're going to sell a whole bunch of Veras. Hey, guys. Thanks for taking my question, and congrats on the great results. Colette, I believe in your prepared remarks, you mentioned GB300 is sort of the fastest ramp in the company's history. How should we think about Vera Rubin against this benchmark? It's, obviously, a new architecture at the silicon level, but similar rack. Does that mean we should expect a similar slope to the Vera Rubin ramp as the GB300? Or should it be a bit more gradual given the new silicon? Thank you.0 [images/image-13-In terms o.jpg] Hey, guys. Thanks for taking my question...Vera Rubin against this benchmark? It's, obviously, a new architecture at the silicon level, but similar rack. Does that mean we should expect a similar slope to the Vera Rubin ramp as the GB300? Or should it be a bit more gradual given the new silicon? Thank you.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Colette M. Kress", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Colette M. Kress", + "level": 2, + "summary": "The CFO and EVP of NVIDIA discusses the upcoming launch of Vera Rubin in the second half of the year, starting in Q3 with initial pieces and continuing to ramp into Q4 and Q1 of next year. Demand is already planned with purchase orders from major customers, though the ramp speed is uncertain due to system complexity. Production timing is the key factor for market entry.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Toshiya Hari", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Toshiya Hari", + "level": 2, + "summary": "Vice President-Investor Relations & Strategic Finance, NVIDIA Corp. Thank you. Before I hand it over to Jensen, please note Jensen will be giving a keynote at GTC Taipei at COMPUTEX on June 1. We will also be participating at the TD Cowen TMT Conference on May 28, and the Bank of America Global Technology Conference on June 4. Our earnings call to discuss the results of our second quarter of fiscal 2027 is scheduled for August 26. With that, here's Jensen to close us out.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_9", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_9", + "level": 2, + "summary": "NVIDIA's CEO highlights an extraordinary quarter with parabolic demand driven by the arrival of Agentic AI, which makes tokens profitable. He lists five key points: NVIDIA runs every frontier AI model, is in every hyperscale cloud, offers full-stack AI factory solutions, extends CUDA to edge and physical AI, and introduces Vera, a CPU for agentic AI opening a $200 billion TAM. The world is rebuilding computing for agentic and physical AI, and NVIDIA is ready with its three-decade platform.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Disclaimer", + "path": "NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf/Disclaimer", + "level": 1, + "summary": "This document is a legal disclaimer stating that the information provided is based on reliable sources but is not guaranteed for completeness, accuracy, or timeliness. It warns against using the information as a primary basis for investment decisions and disclaims all warranties and liability for damages. The content is copyrighted by FactSet CallStreet, LLC.", + "chunk_count": 1, + "children": [] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-Thanks, Co.jpg", + "summary": "image-1 Thanks, Colette. We will now transition to Q&A. Operator, please poll for questions. Great. Thank you for letting me ask a question. I guess, I'd like to ask, what drove the change in segmentation? What's the philosophy behind giving us the numbers that way? And then, can you talk about any competitive differences between the two segments and this kind of surprising CPU number that you talked about, how do you see that across the two segments as well? Thank you.0" + }, + { + "path": "images/image-2-Analyst, M.jpg", + "summary": "image-2 Analyst, Morgan Stanley & Co. LLC Yeah. Thanks, Joe. First of all, Colette meant to say we're increasing our quarterly dividend from \\$0.01 to \\$0.25. I think that extra \\$0.05 would mean a lot to the large shareholders.0" + }, + { + "path": "images/image-3-And as you.jpg", + "summary": "image-3 And as you know, NVIDIA has the largest suite of acceleration libraries in the world from computational lithography to fluid dynamics to particle physics to molecular dynamics to the list goes on. And all of those libraries are essential for us to engage the vertical industries that represents the second and the third category. Okay? Hey, guys. Thanks so much. I wanted to ask, Jensen, I want to ask you about your philosophy on growth. Your Data Center business ex-China grew about 120% in the quarter. And then, you're guiding about 100%. CapEx at the hyperscalers is forecast by many, including myself, to like grow 90% to 100% this year.0" + }, + { + "path": "images/image-4-This secon.jpg", + "summary": "image-4 This second category, when I talk about physical AI and I talk about how the rest of the \\$100 trillion industry that has not been impacted by IT in the last 30 years, it's about to be impacted by AI, that is the segment that I'm talking about. The second cluster is growing incredibly fast. Our share of that, of course, is very, very large. We're fairly unique in our ability to be able to serve this industry. Our platform is built like it's vertically integrated, so that everything works. But when – then we disassemble it, so that people can build and buy it in the configuration they want and assemble it the way they like. Yeah. Good afternoon. Thank you for taking the question. You have Vera Rubin coming soon. And you, obviously, have great insight into coming updates to frontier models, new techniques to optimize around diverse AI workloads. With investors keenly focused on your market share in inference, how do you see Vera Rubin and your extreme co-engineering impacting your share of the inference market as we look into late 2026, 2027?0" + }, + { + "path": "images/image-5-Analyst, C.jpg", + "summary": "image-5 Analyst, Cantor Fitzgerald & Co. Well, we are growing share in inference and we're growing share in inference very, very quickly. And the reason for that is, this year, the number of frontier model companies grew. And so there's Cursor and Perplexity, and there's some new model companies TML and Reflection and the list goes on. And so the number of frontier model companies has grown.0" + }, + { + "path": "images/image-6-Let me go .jpg", + "summary": "image-6 Let me go back again to the question that Ben was asking. Remember, so far, everything that I've just explained in the inference question is really focused on hyperscale. Remember, there's a whole second category of AI data centers that we serve almost uniquely. This segment is very fragmented, requires a really well-integrated platform solution and a very large go-to-market, and that segment, all of the inference, 100% of that, the vast majority of that is NVIDIA. Thanks a lot. Jensen, I wanted to ask about the traction you're getting with some of these custom merchant things. You're doing stuff like CPX and LPX. And I just wanted to ask and see sort of you've talked before about [ph] SaaS LPX (00:37:50) being, I think, 20% of the market. So, I would imagine you're getting pretty good traction with LPX. So, can you just talk about that and maybe also how that fits into your broader platform strategy? Thanks.0" + }, + { + "path": "images/image-7-So, we're .jpg", + "summary": "image-7 So, we're going to need a lot more CPUs, and Vera was designed to be an agentic CPU. The CPUs of the past were designed to have many cores, so that it could be easily rentable. People rented cores. Well, agents don't rent cores. They just want the work to be done fast. The economics of the past was dollars per core. That's the economics of cloud computing of the past. The economics of AI of the future is tokens per dollar or dollars per token. And so, what we need to do in the future is to generate tokens, process tokens as fast as possible. And that's what Vera does incredibly well. Hi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud?0" + }, + { + "path": "images/image-8-Hi, guys. .jpg", + "summary": "image-8 Hi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud? Correct.0" + }, + { + "path": "images/image-9-Co-Founder.jpg", + "summary": "image-9 Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. ...at the magnitude of them, I mean, they're both about the same magnitude now. It almost sounded to me like you were suggesting that you thought the latter, the AI cloud would grow faster, maybe going forward than hyperscale. Is that what you were trying to say? Or do you see like the same kind of growth coming from both segments?0" + }, + { + "path": "images/image-10-Analyst, B.jpg", + "summary": "image-10 Analyst, Bernstein Research First of all, you're correct that AI native clouds. AI native clouds don't build chips – don't design their own chips and they can't really assemble unrelated parts together into an AI factory. And their patience, their tolerance for time to first token is extremely low. And their need for an architecture that has a great deal of offtake, so that it runs every model has customers from everywhere, is incredibly high. And so that's the reason why NVIDIA's architecture is so perfect for them.0" + }, + { + "path": "images/image-11-And so for.jpg", + "summary": "image-11 And so for many of the other applications, industrial applications, enterprise applications, until the AI is very capable and does really productive work and does it safely, and it could do it in a way that can actually generate impact and income, it doesn't really get used. And so you expect the second category to develop slower than hyperscale, and you could see that in the numbers. Good afternoon. Thanks for taking my question. Back at GTC, I believe you discussed \\$1 trillion visibility into both your Rubin and Blackwell platform revenue. But I believe that excluded things like LPX, Rubin CPX and the Vera CPU racks. Can you maybe give us a sense about whether the Vera CPUs are going to be the biggest source of upside above and beyond that \\$1 trillion? Are you contemplating other sort of combinations of products, including CPUs that would allow you to gain an even greater share of that total TAM? Thank you.0" + }, + { + "path": "images/image-12-Analyst, G.jpg", + "summary": "image-12 Analyst, Goldman Sachs & Co. LLC In terms of incremental above the \\$1 trillion, I would say, one, the continued growing of share of the frontier AI models. I'm expecting to grow more share. And so I'm expecting that to grow. Number two, we didn't include any Vera CPU, standalone CPU in that number. And so I expect that to be the second largest. The TAM is, of course, quite large in agentic systems, and all of our customers are quite excited about Vera, and we're going to sell a whole bunch of Veras.0" + }, + { + "path": "images/image-13-In terms o.jpg", + "summary": "image-13 In terms of incremental above the \\$1 trillion, I would say, one, the continued growing of share of the frontier AI models. I'm expecting to grow more share. And so I'm expecting that to grow. Number two, we didn't include any Vera CPU, standalone CPU in that number. And so I expect that to be the second largest. The TAM is, of course, quite large in agentic systems, and all of our customers are quite excited about Vera, and we're going to sell a whole bunch of Veras. Hey, guys. Thanks for taking my question, and congrats on the great results. Colette, I believe in your prepared remarks, you mentioned GB300 is sort of the fastest ramp in the company's history. How should we think about Vera Rubin against this benchmark? It's, obviously, a new architecture at the silicon level, but similar rack. Does that mean we should expect a similar slope to the Vera Rubin ramp as the GB300? Or should it be a bit more gradual given the new silicon? Thank you.0" + }, + { + "path": "images/image-14-Analyst, T.jpg", + "summary": "image-14 Analyst, TD Cowen Yeah. Well, we have indicated for a while that we will be launching Vera Rubin in the second half. We will start in Q3. That will be our initial pieces together. And then once we get to Q4, we're probably going to start to see our ramping continue. It's hard to say at this point what will be a faster ramp. But again, we have demand already planned. We've got POs. We've got almost all of our major customers ready to go. And these are very complex systems that we need to put together. So I think it's just about the timing that it's going to take for us to get that into market.0" + } + ], + "tables": [] + } +} diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/full.md b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/full.md new file mode 100644 index 000000000..8ca65d039 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/full.md @@ -0,0 +1,443 @@ +20-May-2026 + +## NVIDIA Corp. (NVDA) + +Q1 2027 Earnings Call + +## CORPORATE PARTICIPANTS + +Toshiya Hari + +Vice President-Investor Relations & Strategic Finance, NVIDIA Corp. + +Colette M. Kress + +Chief Financial Officer & Executive Vice President, NVIDIA Corp. + +Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +## OTHER PARTICIPANTS + +Joseph Moore + +Analyst, Morgan Stanley & Co. LLC + +Ben Reitzes + +Analyst, Melius Research LLC + +C.J. Muse + +Analyst, Cantor Fitzgerald & Co. + +Timothy Arcuri + +Analyst, UBS Securities LLC + +Vivek Arya + +Analyst, BofA Securities, Inc. + +Stacy A. Rasgon + +Analyst, Bernstein Research + +James Edward Schneider + +Analyst, Goldman Sachs & Co. LLC + +Joshua Buchalter + +Analyst, TD Cowen + +## MANAGEMENT DISCUSSION SECTION + +Operator: Good afternoon. My name is Sarah, and I will be your conference operator today. At this time, I would like to welcome everyone to NVIDIA's First Quarter Earnings Call. All lines have been placed on mute to prevent any background noise. After the speakers' remarks, there will be a question-and-answer session. [Operator Instructions] Thank you. + +Toshiya Hari, you may begin your conference. + +## Toshiya Hari + +Vice President-Investor Relations & Strategic Finance, NVIDIA Corp. + +Thank you, and good afternoon, everyone. Welcome to NVIDIA's conference call for the first quarter of fiscal 2027. With me today from NVIDIA are Jensen Huang, President and Chief Executive Officer; and Colette Kress, Executive Vice President and Chief Financial Officer. + +Our call is being webcast live on NVIDIA's Investor Relations website. The webcast will be available for replay until the conference call to discuss our financial results for the second quarter of fiscal 2027. The content of today's call is NVIDIA's property. It can't be reproduced or transcribed, without our prior written consent. + +During this call, we may make forward-looking statements based on current expectations. These are subject to a number of significant risks and uncertainties, and our actual results may differ materially. For a discussion of factors that could affect our future financial results and business, please refer to the disclosure in today's earnings release, our most recent Forms 10-K and 10-Q, and the reports that we may file on Form 8-K with the Securities and Exchange Commission. All our statements are made as of today, May 20, 2026, based on information currently available to us. Except as required by law, we assume no obligation to update any such statements. + +During this call, we will discuss non-GAAP financial measures. You can find a reconciliation of these non-GAAP financial measures to GAAP financial measures in our CFO commentary, which is posted on our website. + +With that, let me turn the call over to Colette. + +## Colette M. Kress + +Chief Financial Officer & Executive Vice President, NVIDIA Corp. + +Thanks you, Toshiya. We delivered an exceptional quarter with revenue, operating income, and free cash flow exceeding our prior records. Total revenue of \$82 billion was up 85% year-over-year and 20% sequentially. This marked our 3rd consecutive quarter of year-over-year acceleration and the 14th straight quarter of sequential growth, a significant feat given the sheer size and complexity of our manufacturing operations. + +The \$13.5 billion sequential revenue increase was also a record. We capitalized on the inflection in inference demand by ramping Blackwell systems across our diverse end customer base from hyperscalers to model makers to AI cloud providers and sovereign customers. + +In Q1, we also allocated capital effectively across R&D, investments in our ecosystem, and share repurchases. We returned a record \$20 billion to our shareholders, while executing strategic investments, both upstream supply chain and downstream go-to-market ecosystem. This is critical to the market's development and our long-term position. + +Data Center revenue of \$75 billion was up 92% year-over-year and 21% sequentially, driven by sustained strength in our Blackwell architecture. And demand for GB300 NVL72 was particularly strong with frontier model builders and hyperscalers, each having cumulatively deployed hundreds and thousands of Blackwell GPUs, marking the fastest product ramp in our company's history. + +Grace Blackwell is the fastest training system as well as the lowest token generation cost at inference. Spectrum-X, our end-to-end Ethernet platform purpose built for AI, is now larger than all Ethernet network peers combined. InfiniBand has also had a very strong quarter, growing more than 4x year-over-year, driven by deployments of our next-generation XDR technology. + +For your models, Data Center computing revenue of \$60 billion was up 77% year-over-year, while Data Center networking revenue of \$15 billion nearly tripled year-over-year. Before we deep dive into Data Center, we'd like to brief you on our transition to a new reporting framework that better reflects our current and future growth drivers. + +We have two market platforms: Data Center and Edge Computing. Within Data Center, we will report two sub-markets; Hyperscale and ACIE, which incorporates AI clouds, industrial and enterprise. Hyperscale will include revenue from the public cloud and the world's largest consumer Internet companies, while ACIE addresses our growth opportunities in diverse AI purpose-built data centers and AI factories across industries and countries. Edge Computing highlights devices for agentic and physical AI, including PCs, gaming consoles, workstations, AI-RAN base stations, robotics, and automotive. For your reference, we have posted on our website a revenue breakdown based on our new platforms for the past nine quarters. + +Moving back to our Data Center results. Hyperscale revenue of \$38 billion was approximately 50% of Data Center revenue and increased 12% quarter-over-quarter. ACIE revenue was \$37 billion and grew 31% quarter-over-quarter, including AI cloud revenue that more than tripled year-over-year. + +Our customers have enabled rapid stand-up of AI compute capacity. The number of partner data centers exceeding 10 megawatts has nearly doubled in just one year, now surpassing 80 sites. Sovereign revenue increased more than 80% year-over-year. NVIDIA AI infrastructure is now deployed across nearly 40 countries, representing \$50 trillion in GDP. + +As evident to our Q1 results, our customer base is diverse and growing. Supported by our vast ecosystem and installed base, breadth of CUDA accelerated application, and the lowest token cost provider, we are well positioned to address a market opportunity that far exceeds that of any other AI computing platform. + +Demand for AI infrastructure continues to expand at an unprecedented pace. The build-out of AI factories is accelerating. The value of NVIDIA AI infrastructure is rising. The price of renting an H100 has risen 20% year-to-date, while A100 cloud pricing is up nearly 15%. Benefiting from the versatility of our platform and continuous performance enhancements enhanced by our software stack, customers are generating profitable revenue beyond the depreciable life of their GPUs. The vast and trusted marketplace for NVIDIA Compute is a critical foundation on which billions in AI infrastructure spending is being financed by the ecosystem. + +There are two primary drivers behind the accelerating build-out of AI infrastructure. First, from search and advertising to recommender systems and content understanding, the largest hyperscale workloads continue to transition from CPU to GPU-based accelerating computing. + +Second, the adoption of products and services native to AI is inflecting. Since the advent of ChatGPT, we have witnessed mainstream AI transition from one-shot inference to reasoning and to now agentic. AI is no longer a nice to have. AI is now a necessity for enhancing productivity across all industries and roles. This is propelling revenue acceleration across all layers of the AI cake, including energy, chips, infrastructure, models, and applications. Growth in the model layer, particularly at Anthropic and OpenAI, has been incredible with momentum continuing to accelerate, including breakout growth in OpenAI's Codex since the launch of GPT-5.5. + +With analysts now forecasting hyperscale CapEx to exceed \$1 trillion in 2027 and agentic AI beginning to proliferate all industries, AI infrastructure spending is on track to reach \$3 trillion to \$4 trillion annually by the end of this decade. + +Our Blackwell architecture is everywhere, adopted and deployed by every major hyperscaler, every cloud provider, and every major model maker. Last month, we celebrated OpenAI's launch of GPT-5.5, co-designed for, trained with, and served on Blackwell, currently positioned at the top of Artificial Analysis leaderboards. + +Microsoft's Fairwater, the world's most powerful AI data center, is now live, ahead of schedule, powered by hundreds of thousands of Blackwell GPUs. Starting this year, AWS will add more than 1 million Blackwell and Rubin GPUs, and are collaborating on Spectrum networking. At Google, Blackwell will be offered to customers in the cloud, including confidential computing capability, a new foundation for secure, high-performance AI. + +Our share of frontier AI compute is increasing. We have deepened our collaboration with Anthropic and are delighted to be a strategic partner to expand their compute capacity. We will support the company's growth trajectory through AWS, Azure, CoreWeave, SpaceXAI, and more. Now, with the addition of Anthropic to OpenAI, + +Gemini, SpaceX xAI, Meta, MSL, Microsoft AI, TML, Reflection, Perplexity, Cursor, and other major frontier labs already building on NVIDIA, our share of frontier AI models will grow significantly. + +Today's data centers are revenue-generating AI factories. Constrained by power and capital, AI factory operators must choose the right architecture. With our extreme co-design approach, we deliver the industry's lowest token cost, the highest token throughput, and the highest ROI. + +MLPerf inference results are in and once again, we swept every benchmark as Blackwell Ultra delivered the highest throughput across the broad set of models and deployment scenarios. Full stack innovations drove the 2.7x increase in throughput and a 60% reduction in the cost per token on GB300 compared to just six months ago. + +NVIDIA Compute is not just the highest performance AI infrastructure, it is the most economic and financeable. Customers do not buy GPUs. They build AI factories. And the right economic metric is not the purchase price of the GPU, it is the lifetime cost of an AI factory producing intelligence, token per watt, tokens per dollar, uptime, utilization, time to production, software durability, and asset life. NVIDIA excels at all of them. + +Agentic AI and reinforcement learning represents new growth opportunities for CPUs. Building on the success of our Grace CPU, Vera is arriving just in time to meet this inflection. Built on custom Arm cores and co-designed end-to-end with Rubin GPUs and NVLink, Vera will deliver up to 1.5x faster performance per core, 2x performance per watt, and 4x density per rack, compared to x86-based alternatives. + +Vera CPU opens a brand new \$200 billion TAM for NVIDIA, a market we have never addressed before. And every major hyperscale and system maker is partnering with us to get it deployed. We have visibility to nearly \$20 billion in total CPU revenue this year, setting us up to become the world-leading CPU supplier. + +Our annual product cadence, a pace that is unmatched, remains a key pillar supporting our market position. We are on track to commence production shipments of Vera Rubin in the second half of this year, starting in Q3. By integrating seven purpose-built chips across five accelerated racks, Vera Rubin will deliver up to 35x higher inference throughput and up to 10x greater AI factory revenue compared with Blackwell. + +As an early adopter, Google's A5X bare-metal instances, which can support up to 960,000 Rubin GPUs across multiple sites, can enable customers to run their largest AI workloads on NVIDIA's optimized infrastructure. + +While the US government has approved licenses for H200 to be shipped to China-based customers, we have yet to generate any revenue, and we are uncertain whether any imports will be allowed into the country. As a result, consistent with last quarter, we are not including any China Data Center compute revenue in our outlook. + +Let me move to Edge Computing. Our Edge Computing market platform generated \$6.4 billion, up 10% quarter-over-quarter and 29% year-over-year. Robust Blackwell workstation demand was a strong contributor to the growth, while consumer demand fell modestly due to higher memory and system prices. + +Our physical AI continues to gain momentum, exceeding \$9 billion in revenue over the last 12 months. Our partnership with Uber will power the robotaxi fleet across nearly 30 cities and 4 continents by 2028. And in robotics, leading companies across a range of industrial, surgical, and humanoid applications are building on NVIDIA's technology to develop and deploy at scale. + +We remain front-footed in securing sufficient supply to support our customers' growth. In Q1, we increased total supply, inclusive of inventory, purchase commitments, and prepaids to \$145 billion. While we are not immune to supply challenges, we remain confident in our ability to support the growth opportunity ahead, with our intense focus, scale and longstanding partnerships with critical suppliers continuing to serve us well. + +Let me move to the rest of the P&L. GAAP gross margin was 74.9% and non-GAAP gross margin was 75%, largely flat sequentially by Blackwell systems continued to account for most of our shipments. GAAP and non-GAAP operating expenses were up 12% sequentially, primarily due to higher compensation and an increase in compute and infrastructure costs. + +Our non-GAAP effective tax rate of 16% came just below our prior outlook due to favorable geographic mix. And on our balance sheet, days sales outstanding was 45 days. Due to favorable timing of collections, we expect to return to the mid-50s in Q2. We generated record free cash flow \$49 billion, up from \$35 billion in Q4. + +I'd now like to update you on our capital allocation plan. First, to reiterate, our intention is to prioritize R&D and strategic investment, both will enable us to cultivate our ecosystem, drive market growth, and strengthen our market position. As a key enabler of AI, we will make investments necessary to deliver the industry's lowest cost per token and the highest token throughput, which will help our customers and partners scale and expand the AI frontier. + +Return program is another key component of our capital allocation strategy. Given confidence in our long-term free cash flow outlook and our commitment to sharing our success with shareholders, we are increasing our quarterly dividend from \$0.01 to \$0.20 (sic) [\$0.25] per share. We plan to review our dividend on a regular basis, as we continue to scale our business. We are also announcing an \$80 billion share repurchase authorization, which is in addition to the \$39 billion remaining on our current plan. As we indicated at GTC, we plan to return roughly 50% of free cash flow to shareholders this year. + +Let me turn to the outlook for the second quarter. Total revenue is expected to be \$91 billion, plus or minus 2%. We expect sequential growth to be driven primarily by Data Center. We are continuing to work vigorously on our supply chain ecosystem to address the incredible demand we see ahead of us, giving us full confidence in the \$1 trillion in Blackwell and Rubin revenue we foresee from 2025 through calendar 2027. + +GAAP and non-GAAP gross margins are expected to be 74.9% and 75%, respectively, plus or minus 50 basis points. For the full year, we are still expecting to be in the mid-70s. + +GAAP and non-GAAP operating expenses are expected to be approximately \$8.5 billion and \$8.3 billion, respectively. For the full year, we now expect OpEx to grow somewhere in the upper-40s on a year-over-year basis, driven by higher R&D and acceleration in the usage of AI tools to enhance productivity. + +For the full year 2027, we expect GAAP and non-GAAP tax rates to be between 16% and 18%, excluding any discrete items from material changes to our tax environment. This is lower than our prior expectation of 17% to 19% due to changes in geographic mix. + +That puts me at the end of this part, and I'm going to now turn this over to the Q&A with Toshiya. + +## Toshiya Hari + +Vice President-Investor Relations & Strategic Finance, NVIDIA Corp. + +Thanks, Colette. We will now transition to Q&A. Operator, please poll for questions. + +# QUESTION AND ANSWER SECTION + +Operator: Thank you. [Operator Instructions] Your first question comes from Joseph Moore with Morgan Stanley. Your lie is open. + +## Joseph Moore + +Analyst, Morgan Stanley & Co. LLC + +![](images/0d85508fb3b05f1ea4581d205e234657e37d89318b6ed3527ab87c967514dd9c.jpg) + +Great. Thank you for letting me ask a question. I guess, I'd like to ask, what drove the change in segmentation? What's the philosophy behind giving us the numbers that way? And then, can you talk about any competitive differences between the two segments and this kind of surprising CPU number that you talked about, how do you see that across the two segments as well? Thank you. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/0b25a26c8581c4587355a36b9ed441d79ed5af4e7a240acbb580e02cbca82bef.jpg) + +Yeah. Thanks, Joe. First of all, Colette meant to say we're increasing our quarterly dividend from \$0.01 to \$0.25. I think that extra \$0.05 would mean a lot to the large shareholders. + +So, anyhow, let's see, Joe, on the segmentation and the description of the business, we wanted you to understand our business better. AI is very diverse and computing is diverse. They're diverse in several ways. The first thing, of course, is AI includes languages, and depending on the different industries, it could be 3D graphics for manufacturing and industrial robotics. It could be proteins for life sciences. It could be small chemicals for life sciences or material sciences. It could be physics for the physical sciences, whether it's in the energy sector or, of course, the science labs, higher education, so on and so forth. So, AI is diverse. + +The second thing is the applications are diverse. It could be in enterprise. It could be in the energy sector, manufacturing sector, and such. Where it runs is diverse. It could be in the hyperscale cloud. It could be AI natives. There's a whole network of AI natives that are cropping up around the world. Enterprises on-prem, industrial in the factories, in the plants, all the way to supercomputing centers and the edge. Edge, including, of course, what most people see, self-driving cars, robotics, but a large growing network of computers inside manufacturing plants, whether it's a chip plant or packaging or computer plants, all kinds of different types of manufacturing plants. And then, of course, in the future, every single base station, every single radio network would become an AI-powered radio network. And so where it runs. + +And then lastly, how it's governed. It could be operated by public cloud, but it could also have industrial regulatory reasons that prevents it from being run in a regulatory cloud. It could be because of confidential computing. It could be because of national security reasons. Different data centers have to be built differently. + +NVIDIA is quite unique in the sense that we are the only company that builds all of the technology components. We build it in an extreme co-design way, in a complete end-to-end way, in a full stack way. But then, we, of course, open the platform so that it could be integrated into all the different environments. But some environments just require – an enterprise, for example, require a company who has all of the technologies working together so that they don't have to build it. They would like to buy it and operate it. And so there's many different segments of the data center market where NVIDIA's total solution, fully integrated solution with full stack, but still open, that way of doing – of producing or delivering products is really, really important. + +And so if you look at our different segments, the way we broke it out into three large segments. You take all of the words that I just said and you try to find the simplest factoring of it, it would be the hyperscale clouds, that would be one large segment. And within that segment, there's three different ways that we operate. First way is that we help the hyperscale clouds accelerate their data processing and machine learning workloads. We accelerate and support their AI processing inside. We also, of course, bring a lot of business – NVIDIA ecosystem business to their public clouds. And so that's one segment. + +The second segment is AI natives, enterprise on-prems, industrial on-prems and sovereign AI. That segment is growing incredibly fast because everybody needs AI, and we're going to see AI being adopted by every industry, every country, every company. And so everybody wants to build it in a different way. And the fact that we provide the entire solution, it makes it much easier, it makes it possible at all for people to be able to build these things. + +And then, of course, the robotic edge. Today, yesterday's computing was largely about personal computing. In the future, it's going to be about personal AI. And that personal AI, one example of it is the self-driving car. It's a car – it's a robotic system that's essentially your personal AI. And of course, there'll be all kinds of different types of robotic systems, including even the base station radio network, as I mentioned, is going to be essentially a robotic system. + +And so that's the reason why we broke it all apart this way. It's the simplest way of understanding our business. Each one of them have different stacks in a lot of ways. They have different operating systems. They operate in a different way. We go to market very differently in each one of them. The easiest go-to-market, of course, is the hyperscaler, because there are only five or six of them. But the rest of them – the rest of the industry represents a couple of 250,000 companies around the world. That go-to-market is very complex, very diverse. Your understanding of AI has to be extremely diverse. + +And as you know, NVIDIA has the largest suite of acceleration libraries in the world from computational lithography to fluid dynamics to particle physics to molecular dynamics to the list goes on. And all of those libraries are essential for us to engage the vertical industries that represents the second and the third category. Okay? + +So, anyways, it's really about the fact that our business has now evolved and grown to such a large scale, it's helpful to segment it so that you have a better understanding of how our business works. + +Operator: Your next question comes from Ben Reitzes with Melius Research. Your line is open. + +## Ben Reitzes + +Analyst, Melius Research LLC + +![](images/edf18e4027af2d964bd58c4b17c2f3a62442fc6c96506968474bc0a393e71bfb.jpg) + +Hey, guys. Thanks so much. I wanted to ask, Jensen, I want to ask you about your philosophy on growth. Your Data Center business ex-China grew about 120% in the quarter. And then, you're guiding about 100%. CapEx at the hyperscalers is forecast by many, including myself, to like grow 90% to 100% this year. + +And you talked about Data Center still on track to be \$3 trillion to \$4 trillion by the end of the decade. I was just wondering, the goal for the company to grow faster than hyperscaler CapEx, are you comfortable in kind of endorsing that view? And do you still see hyperscaler CapEx kind of still growing after this year at a very rapid clip? Thanks a lot. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +## A + +Yeah. Thanks, Ben. So, first of all, we should be growing faster than hyperscale CapEx. And the reason for that is illustrated by the segmentation that I just described. Our Data Center business has two large parts. It has more parts than that, but we combined it into two large parts for simplicity's sake. It's much more complex than the two large parts, but I combined it into two, so that it's at least easier to understand. Okay? + +And so if you look at the first part is hyperscalers. That's the hyperscale CapEx that you were just talking about. And they're at \$1 trillion this year. I have every expectation it's going to grow from here for fundamentally good reasons. This is the way computing is going to work in the future. And if they don't have the compute, they won't have the revenues. It is very clear compute is revenues, compute is profit. And so the world is changing. SaaS didn't used to use as much compute, but AI requires a tremendous amount of compute. + +But you could do, of course, incredibly more, which is the reason why we heard about the frontier AI companies, both Anthropic and OpenAI, growing at an incredible pace. The fact that they can grow within one month, what some of the SaaS companies would have taken a decade to grow, tells you something. And so the first category is hyperscale and the CapEx is at \$1 trillion, and it's growing towards the \$3 trillion to \$4 trillion. + +The second category is all of the AI native clouds. They're regional. They're all over the place. There are startups all over the world supporting those companies. They're enterprise. 250,000 enterprise companies around the world, many of them will have to build or want to build AI factories for themselves to operate. Many industrial companies, there's no choice but to put the computer where the context is, where the action is. You can't put that in the cloud. It has to respond reliably, quickly every single time. Can't imagine a chip plant – a chip fab being connected to a cloud service provider, doesn't make any sense. + +And so, the second category, and then sovereign AI clouds, and so there's a whole category of data centers that semi-custom chips just don't apply, because these data centers want to buy systems. They want to operate systems. They don't want to design – they don't want to build it themselves. And so, the second category is extremely diverse. Instead of five or six, seven companies representing the revenues associated with our first category, the second category is hundreds, thousands of companies. And in the future, it will be hundreds of thousands of companies with a large number of companies with smaller installations. And that category is going to continue to grow at incredible pace. + +This second category, when I talk about physical AI and I talk about how the rest of the \$100 trillion industry that has not been impacted by IT in the last 30 years, it's about to be impacted by AI, that is the segment that I'm talking about. The second cluster is growing incredibly fast. Our share of that, of course, is very, very large. We're fairly unique in our ability to be able to serve this industry. Our platform is built like it's vertically integrated, so that everything works. But when – then we disassemble it, so that people can build and buy it in the configuration they want and assemble it the way they like. + +And so, this second category is fairly poorly understood because they're just so many small companies, or so many companies, and each one of the installations are relatively small compared to, of course, one of the hyperscalers. And so, if you look at the segmentation and the size of each, you could see that, in fact, we're growing share in the hyperscalers, because we now have much bigger support from Anthropic, a new partner of ours, and we're helping them expand their capacity greatly in the coming years. And then, the second, very few companies have exposure into the second category, because of the platform solution that we have. + +Operator: Your next question comes from C.J. Muse with Cantor Fitzgerald. Your line is open. + +## C.J. Muse + +Analyst, Cantor Fitzgerald & Co. + +![](images/4cbfc1695947a2407ad7cae3e28a210d827408661d065876f2ce396071713a23.jpg) + +Yeah. Good afternoon. Thank you for taking the question. You have Vera Rubin coming soon. And you, obviously, have great insight into coming updates to frontier models, new techniques to optimize around diverse AI workloads. With investors keenly focused on your market share in inference, how do you see Vera Rubin and your extreme co-engineering impacting your share of the inference market as we look into late 2026, 2027? + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/14cbd7f395d85ac7973319fc91432e1731f538d0fff7e39e943445a65e242cb3.jpg) + +Well, we are growing share in inference and we're growing share in inference very, very quickly. And the reason for that is, this year, the number of frontier model companies grew. And so there's Cursor and Perplexity, and there's some new model companies TML and Reflection and the list goes on. And so the number of frontier model companies has grown. + +And we added Anthropic to our partnership this year. They're expanding incredibly fast. We've partnered with them to secure computing capacity across Azure, AWS, CoreWeave. I forget who else we've already announced, but there's a whole list of others that we are bringing online for them. And so the amount of capacity that we're going to bring online for Anthropic this year and next year is going to be quite significant, very significant. And our coverage of Anthropic has been largely zero until just recently. + +And so we're gaining share tremendously fast in inference. Vera Rubin is going to be even more successful than Grace Blackwell at this point. Every single – I can't think of one, every single frontier model company will jump on Vera Rubin from the get go and that wasn't true before on Blackwell. And so Vera Rubin is off to a tremendous start, and it'll surely be more successful than even Grace Blackwell. So, I think the end of your answer, C.J., is that we're gaining share in inference. + +Let me go back again to the question that Ben was asking. Remember, so far, everything that I've just explained in the inference question is really focused on hyperscale. Remember, there's a whole second category of AI data centers that we serve almost uniquely. This segment is very fragmented, requires a really well-integrated platform solution and a very large go-to-market, and that segment, all of the inference, 100% of that, the vast majority of that is NVIDIA. + +And then, of course, physical AI. NVIDIA is practically the only company serving physical AI today, and we've been working on physical AI for a long time. And so, that is also growing. So, our share of inference is growing very quickly. + +Operator: Your next question comes from Timothy Arcuri with UBS. Your line is open. + +## Timothy Arcuri + +Analyst, UBS Securities LLC + +![](images/2004f21ede9c6ef68a82be7903b2db1f4dc772be2de351fa717d20c68a5c6bdf.jpg) + +Thanks a lot. Jensen, I wanted to ask about the traction you're getting with some of these custom merchant things. You're doing stuff like CPX and LPX. And I just wanted to ask and see sort of you've talked before about [ph] SaaS LPX (00:37:50) being, I think, 20% of the market. So, I would imagine you're getting pretty good traction with LPX. So, can you just talk about that and maybe also how that fits into your broader platform strategy? Thanks. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +## A + +The LPX is designed for low latency and high token rate, but its throughput is low. Its model size capacity is low. And its context processing, its ability to absorb a lot of context, for example, for software coding, for agentic workloads, its ability to absorb a great deal of context is lower. And so the challenge is simply – and I've explained before that the use case for LPX is not broad. It's intended for somebody who has a fairly large portfolio of different types of token services. And for the high token rate, maybe these services are quite premium and the number of customers is not significant, but the token rate is very high. And so, that remains exactly consistent with what I've said before. And I still expect that. And so I expect that LPX and other SRAM-based, decode-focused, high token rate generated focused accelerators will be a niche product for some time to come. + +As you know, Grace Blackwell and Vera Rubin, we support the entire life cycle of AI from the data processing, preparing for training, okay, data processing to pre-training, to post-training, reinforcement learning, all the way to inference, Grace Blackwell is the best platform in the world to do all of that. And if we insert certain circumstances, so long as the customer – the provider already has a high token rate service that they can offer, then we can tack on an LPX and they could deliver that service even better. And so that's how I see the market. + +And I think whether it's 20% or 10%, just depends on where we are in the development of AI. I think today, it's a lot less than 20%. Someday, these premium tokens could be 20%. And we're ready to work with the service providers to enable this capability. I'm excited about it. + +Operator: Your next question comes from Vivek Arya with Bank of America Securities. Your line is open. + +## Vivek Arya + +Analyst, BofA Securities, Inc. + +## Q + +Thanks for taking my question. Jensen, there's a lot of excitement around CPU for agentic applications and just a lot of noise around the number of CPUs actually exceeding the number of GPUs. And I was just hoping that you could kind of give your perspective that, first of all, is this an incremental workload? Is this kind of cannibalizing what the GPU would have done otherwise? + +And then, secondly, the \$20 billion number that you gave, is that for standalone Vera CPUs, or is that kind of already included in that Vera as part of Vera Rubin? So just if you could educate us on the role of CPU versus GPU. Is it cannibalistic? Is it incremental? And then, the \$20 billion number, how to kind of put that in context with what you sell, right, which is usually the CPU as part of the GPU. Thank you. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +## A + +The \$20 billion is for standalone CPU. And remember, we have Vera is used in three ways, as a standalone – four ways, as a – let me just start with the one that you already know. The first way is Vera Rubin. And we'll sell millions of Rubins, and every two of them is connected to a Vera. And of course, we price those too. And they're properly priced. And so that's number one use case. + +The second use case is Vera standalone CPU. The third is Vera with CX9 and the software stack for storage. And then Vera in a – with CX9 with a software stack for security and compute isolation and confidential computing. Okay? So each one of those use cases is built on Vera. And my sense is that we'll be supply constrained throughout the entire life of Vera Rubin. There are four different use cases of it. And – but anyhow, the answer to your question is of the \$20 billion is a standalone. + +With respect to CPUs, an agent is essentially what people call a harness. And the agent has a harness that does the – and the harness could be OpenClaw, it could be Hermes. Claude Code is essentially a harness around Claude, around the Opus model. OpenAI's Codex is a harness around the GPT-5.5 model. And so these are harnesses, and these harnesses provide for things like IO, orchestration, memory management, tool use, connected to tools, for example, browsers and things like that, C compilers, Python compilers. And so, the harness runs on CPU and the tool use runs on CPUs. For example, if the AI were to do a search or do a browser, use a browser, that would run on the CPU. + +The world has 1 billion users, human users. My sense is that the world is going to have billions of agents. Not today. I mean, we're going to grow into it. But it will have billions of agents and those billions of agents will all use tools. And those tools are going to be like PCs. Just like us humans using PCs today, in the future, you'll have an agent using PC. + +And so, if you kind of think along the lines of in the future, you pick your favorite number of agents at the moment, at the moment, call it, a few hundred thousand, but in the future, call it, eventually a few billion, I could imagine them all using – effectively having PCs that they can all use. + +Every one of those agents are going to spin off sub-agents. And every time they spin these off, you're going to need to do inference. That's where the thinking happens. All of the thinking happens on GPUs. All of the orchestration essentially runs on CPUs and the sub-agents, when they're spun off, when they're thinking, they use GPUs. + +Whenever the agents use simulators, those can run on CPUs or GPUs, which is the reason why we're working so closely with Cadence and Synopsys to accelerate all of the world's tools. We're accelerating all of the world's tools and data processing engines and database engines because agents use these tools and they have lower patience – tolerance than humans, and they want things to happen quickly. + +And so we're accelerating all of the world's tools so that it runs on CUDA. And you could see us doing that, when I work with Cadence and Synopsys and Siemens and companies – and Adobe, that's because we're trying to get all of the world's tools to run on GPUs, because they already have GPUs and it's a lot faster. + +So, we're going to need a lot more CPUs, and Vera was designed to be an agentic CPU. The CPUs of the past were designed to have many cores, so that it could be easily rentable. People rented cores. Well, agents don't rent cores. They just want the work to be done fast. The economics of the past was dollars per core. That's the economics of cloud computing of the past. The economics of AI of the future is tokens per dollar or dollars per token. And so, what we need to do in the future is to generate tokens, process tokens as fast as possible. And that's what Vera does incredibly well. + +So we're expecting to be very successful with Vera. But ultimately, what we're doing is we're building infrastructure for AI and it needs incredibly great storage. That's the reason why we built STX. It needs incredibly good networking. That's why we have Spectrum-X. It needs incredibly great GPUs, of course, and inferencing ability. That's the reason why NVLink 72. It needs incredibly great security and confidential computing, which is the reason why Vera Rubin is the world's first platform with end-to-end confidential computing. And it needs great CPUs. We've got it all covered. + +Operator: Your next question comes from Stacy Rasgon with Bernstein Research. Your line is open. + +## Stacy A. Rasgon + +Analyst, Bernstein Research + +![](images/49b2e96aca58eccb8d7e39a9f320774c2299e17c8d8795a4ee85ddbfaf9d0e03.jpg) + +Hi, guys. Thanks for taking my question. I wanted to go back to the segmentation. So, first of all, I'm just curious, where do you put the neo clouds across those two segments? Are they in hyperscale or are they in the AI cloud? + +Part of me assumes the latter, but I'm not so sure. And then, by just looking... + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/3d382a9028b2d4d428cc449e07362009df09d8b4333287f962a471e75753bb2b.jpg) + +Correct. + +## Stacy A. Rasgon + +Analyst, Bernstein Research + +![](images/da53935fa8a8d12ad9eef774abeedcb64054b495dae0fc02e6094c54e4c985cb.jpg) + +...at the magnitude of them, I mean, they're both about the same magnitude now. It almost sounded to me like you were suggesting that you thought the latter, the AI cloud would grow faster, maybe going forward than hyperscale. Is that what you were trying to say? Or do you see like the same kind of growth coming from both segments? + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/4779b804abdc539c7aecb355dc3a0c0ebbca9da2bf81bcfd29740bb023a57911.jpg) + +First of all, you're correct that AI native clouds. AI native clouds don't build chips – don't design their own chips and they can't really assemble unrelated parts together into an AI factory. And their patience, their tolerance for time to first token is extremely low. And their need for an architecture that has a great deal of offtake, so that it runs every model has customers from everywhere, is incredibly high. And so that's the reason why NVIDIA's architecture is so perfect for them. + +We offer every component and whatever we don't offer, our ecosystems of partners offer it. And it's all fully integrated. It all works together. The number of customers that could rent it from an AI native is incredibly high, basically, every single AI builder, every AI native startup around the world, SaaS companies, enterprise companies, industrial companies. + +And so, our computing – our architecture is the most rentable of any computing platform in the world. So it's the most performant. It's the easiest to put together. It's the most rentable. Has the best TCO. And it's the easiest to finance. And so all of those properties are quite unique to the needs of AI natives. It's in the second category. They're very similar to even OEMs and so on and so forth, large enterprises and so forth, surprisingly. Okay? So we put that in the second category. + +If you look at that segment, it started growing after the AI ecosystem developed in the hyperscale. Hyperscale developed AI first for a lot of reasons. They have great computer science. They have excellent data center capability. And they also focus largely on consumer applications, which, if not perfect, is not the end of the world. It enhances the service, so long as it enhances the service. + +And so for many of the other applications, industrial applications, enterprise applications, until the AI is very capable and does really productive work and does it safely, and it could do it in a way that can actually generate impact and income, it doesn't really get used. And so you expect the second category to develop slower than hyperscale, and you could see that in the numbers. + +However, long term, if you look at industrial and enterprise, clearly, that's where future economics is going to be, because it represents some \$50 trillion, \$80 trillion of the world's economy. And it's going to be larger than that because of AI. And so I expect the second category to be larger over time, both in the near term, over the next several years. I think it's a foregone conclusion, both are going to grow incredibly fast. I expect the second category to still grow faster, but both are going to grow incredibly fast. And then, I'm hoping that within the next five years, physical AI and robotics segment is going to grow incredibly fast. + +Operator: Your next question comes from Jim Schneider with Goldman Sachs. Your line is open. + +## James Edward Schneider + +Analyst, Goldman Sachs & Co. LLC + +![](images/7f937c7affb6b92ada7a37b27f6447d86b1f226c36ad7411a922189e5ebcb3ef.jpg) + +Good afternoon. Thanks for taking my question. Back at GTC, I believe you discussed \$1 trillion visibility into both your Rubin and Blackwell platform revenue. But I believe that excluded things like LPX, Rubin CPX and the Vera CPU racks. Can you maybe give us a sense about whether the Vera CPUs are going to be the biggest source of upside above and beyond that \$1 trillion? Are you contemplating other sort of combinations of products, including CPUs that would allow you to gain an even greater share of that total TAM? Thank you. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/2c7666403c8b03a5bd61f4797fb9ab85fca44839f9017bf914196207882680f7.jpg) + +In terms of incremental above the \$1 trillion, I would say, one, the continued growing of share of the frontier AI models. I'm expecting to grow more share. And so I'm expecting that to grow. Number two, we didn't include any Vera CPU, standalone CPU in that number. And so I expect that to be the second largest. The TAM is, of course, quite large in agentic systems, and all of our customers are quite excited about Vera, and we're going to sell a whole bunch of Veras. + +And then third would be LPX, because as I explained earlier, LPX is designed as a – because of its SRAM architecture, it has the benefit of very low latency and very high interactivity, but it's also its throughput and its context processing ability is also quite limited. And that's just kind of the nature of SRAM type-based systems. But the combination, we'll be able to address the entire spectrum of AI from pre-training to post-training to inference agentic systems through the combination of Vera Rubin and LPX. + +Operator: Your next question comes from Joshua Buchalter with TD Cowen. Your line is open. + +## Joshua Buchalter + +Analyst, TD Cowen + +![](images/e2b7ee37f611d6e1b72b30b49809951c7ef006d50fcab518fc2c35d690aa20f5.jpg) + +Hey, guys. Thanks for taking my question, and congrats on the great results. Colette, I believe in your prepared remarks, you mentioned GB300 is sort of the fastest ramp in the company's history. How should we think about Vera Rubin against this benchmark? It's, obviously, a new architecture at the silicon level, but similar rack. Does that mean we should expect a similar slope to the Vera Rubin ramp as the GB300? Or should it be a bit more gradual given the new silicon? Thank you. + +## Colette M. Kress + +Chief Financial Officer & Executive Vice President, NVIDIA Corp. + +![](images/38bd8e5075489a757ece927fd10e943cba2f73b82a2e1ff394c04c80d4c58029.jpg) + +Yeah. Well, we have indicated for a while that we will be launching Vera Rubin in the second half. We will start in Q3. That will be our initial pieces together. And then once we get to Q4, we're probably going to start to see our ramping continue. It's hard to say at this point what will be a faster ramp. But again, we have demand already planned. We've got POs. We've got almost all of our major customers ready to go. And these are very complex systems that we need to put together. So I think it's just about the timing that it's going to take for us to get that into market. + +Nothing else other than getting from production of all of the different systems that we have ready for order. So, a little early to say, but yes, we're going to start in Q3 and continue to ramp into Q4 and Q1 of next year certainly is going to be very big as well. + +Operator: There are no further questions at this time. Toshiya Hari, I turn the call back over to you. + +## Toshiya Hari + +Vice President-Investor Relations & Strategic Finance, NVIDIA Corp. + +Thank you. Before I hand it over to Jensen, please note Jensen will be giving a keynote at GTC Taipei at COMPUTEX on June 1. We will also be participating at the TD Cowen TMT Conference on May 28, and the Bank of America Global Technology Conference on June 4. Our earnings call to discuss the results of our second quarter of fiscal 2027 is scheduled for August 26. + +With that, here's Jensen to close us out. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +This was an extraordinary quarter. Demand has gone parabolic. The reason is simple. Agentic AI has arrived. AI can now do productive and valuable work. Tokens are now profitable. So model makers are in a race to produce more. In the AI era, compute capacity is revenue and profits. NVIDIA is the platform of this era. Of all the platforms in the world, NVIDIA Compute supports the richest diversity of demand. Let me highlight my top five things. + +First, NVIDIA is the only platform that runs every frontier AI model. With the addition of Anthropic to our existing partners, OpenAI, xAI, Meta MSL, Gemini, and many others, our share of frontier AI is growing. + +Second, we are in every hyperscale cloud, supporting their core data processing and machine learning workloads, internal AI services, as well as supporting their demand for NVIDIA users in their public cloud services. + +Third, our full stack complete AI factory solution and vast global ecosystem let us uniquely address new AI data center segments, mew AI native clouds and sovereign AI clouds and on-premises enterprise and industrial infrastructure. This is that second category I was talking about earlier. + +Fourth, NVIDIA CUDA extends all the way to the edge. Robotics, autonomous vehicles, embedded medical instruments. AI-RAN, telco base stations. The next wave is physical AI with billions of autonomous and robotic systems operating in the physical world. This is the third segment we were talking about earlier. + +And rounding out the top five things, we have a major new growth driver, Vera, the world's first CPU purpose built for agentic AI. Vera opens a brand new \$200 billion TAM for NVIDIA, a market we have never addressed before. And every major hyperscaler and system maker is partnering with us to deploy it. + +The world is rebuilding computing for agentic AI and robotic physical AI. NVIDIA sits at the center of these transitions. We built NVIDIA Compute platform over three decades, one architecture, vast ecosystem, extreme co-design across chips, systems, networking, and software. We built it ahead of this moment, so that when agentic AI arrived, NVIDIA would be ready. It has arrived. + +Look forward to catching up next time. + +Operator: This concludes today's conference call. You may now disconnect. + +## Disclaimer + +The information herein is based on sources we believe to be reliable but is not guaranteed by us and does not purport to be a complete or error-free statement or summary of the available data. As such, we do not warrant, endorse or guarantee the completeness, accuracy, integrity, or timeliness of the information. You must evaluate, and bear all risks associated with, the use of any information provided hereunder, including any reliance on the accuracy, completeness, safety or usefulness of such information. This information is not intended to be used as the primary basis of investment decisions. It should not be construed as advice designed to meet the particular investment needs of any investor. This report is published solely for information purposes, and is not to be construed as financial or other advice or as an offer to sell or the solicitation of an offer to buy any security in any state where such an offer or solicitation would be illegal. Any information expressed herein on this date is subject to change without notice. Any opinions or assertions contained in this information do not represent the opinions or beliefs of FactSet CallStreet, LLC. FactSet CallStreet, LLC, or one or more of its employees, including the writer of this report, may have a position in any of the securities discussed here in. + +THE INFORMATION PROVIDED TO YOU HEREUNDER IS PROVIDED "AS IS," AND TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, FactSet CallStreet, LLC AND ITS LICENSORS, BUSINESS ASSOCIATES AND SUPPLIERS DISCLAIM ALL WARRANTIES WITH RESPECT TO THE SAME, EXPRESS, IMPLIED AND STATUTORY, INCLUDING WITHOUT LIMITATION ANY IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, ACCURACY, COMPLETENESS, AND NON-INFRINGEMENT. TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, NEITHER FACTSET CALLSTREET, LLC NOR ITS OFFICERS, MEMBERS, DIRECTORS, PARTNERS, AFFILIATES, BUSINESS ASSOCIATES, LICENSORS OR SUPPLIERS WILL BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL OR PUNITIVE DAMAGES, INCLUDING WITHOUT LIMITATION DAMAGES FOR LOST PROFITS OR REVENUES, GOODWILL, WORK STOPPAGE, SECURITY BREACHES, VIRUSES, COMPUTER FAILURE OR MALFUNCTION, USE, DATA OR OTHER INTANGIBLE LOSSES OR COMMERCIAL DAMAGES, EVEN IF ANY OF SUCH PARTIES IS ADVISED OF THE POSSIBILITY OF SUCH LOSSES, ARISING UNDER OR IN CONNECTION WITH THE INFORMATION PROVIDED HEREIN OR ANY OTHER SUBJECT MATTER HEREOF. + +The contents and appearance of this report are Copyrighted FactSet CallStreet, LLC 2026 CallStreet and FactSet CallStreet, LLC are trademarks and service marks of FactSet CallStreet, LLC. All other trademarks mentioned are trademarks of their respective companies. All rights reserved. \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-1-Thanks, Co.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-1-Thanks, Co.jpg new file mode 100644 index 0000000000000000000000000000000000000000..345570fefc2750d212dbb9ca2d31a00afa50db10 GIT binary patch literal 1369 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<5A1R0qH8UG()kO#Vxl@SaWpn#EynT3^&or9B$8>nEb z00R>vGcywlGb<|#3s7|}P@aKBkX1<0(2-3zFp*uUP{gQl;zAB(r;P_igD!qhF-|IK z;^Yz&myncFRa4i{)G{$OGqmaka3YSZQ|TeofBv2)j3M&~ka)>xhT)6Qdr?PR-2hpUWi(FzVCJ$9Vg1iRy z8F3zKBFkrRk0JbZi-Cuk5g2*Qf(-Ty-+yVW*_$8kt9DtL*Jh^R`mc6Z81p{;yLQO- zyRlBf7xtCE_?B&0{&DWAIgZ&n^WHf9XXq^sbgo@{e~#F+w5K!sxorQkmTcR{x6Os^ zRi#SZL%YC#CX7NASGjAx=`6mQdj5;(>f}edRzI6KeS+keybR|bUm8C7kX80N@pC=D z+RuMpR~`1@?zvfQDfw1+N)rsiU)ir^DE|^J`S88K9;2mmA5ZJ-2r08!_{Hx2soho# z+GoG+es;w)a>=`+g+TawmI$i@>z>NTdem|Fi`l!&$j2-JNC<4HI>duN!kT}c_RPR zL2kaQV`|CtZG|yzZZ}2b7rxV~V$fK%U8;BPnwPvs54kJ%tVr&&+pBd#u|dm`eW3{B z*&k;QTwZ&5=XCoC_K!pUh_&3gyFa1J-l+c2f{&%I?qq4^^-gUpxc^>m|1$05LuT{* zrd{4}*zc$OyFT{Uvlw1z&-v)qbG@+pbm^IkJyv~d_CL~Iyr{wEKZEo2c&}oahLbtR zRoz4F7VVCX^*i;|vFVwa`j6r^`Nn6POZ|DG?(du>_S@$8N@lJ5fiX}1iCtAx&+*py z>Mnm>gmLR5S>E+(*+=*Eas(_?wRT{>xAV4c(r@q2N^Jqb6SvGg#lK?t^7hhn3$Yx# zHSAy4THCYxJTZOpTIxSTnVtXh3w2ur7IG)Y~NNX#Svtz#+@T^RwRJ^|4jhLvpj78 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-10-Analyst, B.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-10-Analyst, B.jpg new file mode 100644 index 0000000000000000000000000000000000000000..55705c10e61dd7491186981e1de811d62e7fe9a4 GIT binary patch literal 1243 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyF8z*5x0y+VahfyE@XR$Sk(wbzk$Z ze<6&Q0%QC-*T(%SF}Ejfv|Ik4VeRA};nojc`y2hrUMAzRJN#AsuLJ)|7{iZTo1b)z zXJYS44Za<3s^qVve_VIocaQk55^?3#>(9Qq%l|quPH*$)FI#s6es;KN|85@tmsNGV zd;c@UryTz`OZ}7mr3R~0VAn6(y{2&U&O<9y<~usbl=tM@oL-T0<)g&C`@xsL{43r5 zaJK)Z7aN5PKG%KtXKU{h&Rbn{efs8@?U6Ebs`w6nT5Iw#Y^ju%{v=U1hMN}x*h>qU z?%eif{(Dno<)7?xI?HR-Bi!qFj6@6HUby<7;fHsreMJ5Ht`oa=R-azJV~R$(W$t38 z23rq-QyiA;6I^wZf24hXJIf~PxRLYkZNK>DFaNZ*i$PoXsQLWVOH!s5mFEtAdH+wo zH}^W{quuAXyxQkAX9@TAe=L>u7wSY9)mw7H9({}M6fknx!O+ut92k1@&+XRV9Q~i+ zJjed`jgj}HH~&`a6?2n|zi0mV@|UC80W9%p+y67n6Rf|_wDFJka(A-{pIn7jA1eQ| zx~K0y!*cu9yvwm-E-8gm6CC9hHCW_x-upW{#K7=SjNbhp0#E7#e_St*ODqD8+DFMK4}u@Abx<0 zftjIzRaYc{(N!dXRl8U1(7R-lZzn#Zda0NCXH AkN^Mx literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-11-And so for.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-11-And so for.jpg new file mode 100644 index 0000000000000000000000000000000000000000..1b4856b4404389f8ae70541c158139d36e9ec33e GIT binary patch literal 1368 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrE7s{LZQWk^tYG4TlWGmBE0*qTx|5dq zxc~2W=AYMvS5?e=_wlCg%(V|*7=EAqs=7`z=(f|9_wF;My%F7cU-9<*^J^K(SFX5m z; yZwt!E&p|O&XbgIM)lUxi0eu!^{IvPudO;*y)xr-PS`XdOGAqX|JWLA)w)hz zEWJ@0qT$f@&@IyV`kkkIgnAGVUo=8C_oke-(me1SiHZK+5{x0IH~p%n+1H!NMi zwDOfz?!M(YNps!mTMs<--x{;1!RA=RJCk0Yb!BInp9tQ+@Rwt~>*e=C|Bh)-k>InG zYv1wb^~{fRH$>*HYk1n|_Qz8G%+-Fj7_qH8x2v}#9H_1T8^XAK)$`v<-{n6a{Lk=3 zIEJsY|Lty@{|x5?Bpusm#*FL%2@HA^RFxOxEt1p zhqDyCSz*uT|5=3b%5~mHYYlGQ|GSp`-EvF2v-dy0{&~GPFy^*vdEN)V#>a(CybQ(v zB(7b`5IXemqY;mr|404hSHitx_*On#y_;R1Y0HOFp2g4NSNz*5yS~%Z+LiM;r%2h9 z^slcSr2eVwzw*{pbkdEpA{-5Q%o@kn+wNU?tGxSh#?Ac8$*U&G86K`q-niywP4tS> zlD^ws&g0~q|I9Zv`c`cG8P`3$%EFU37O;O2Vfekf^J}f|l3PBLwh6LZ$hKCA%}m-= zHeYmurIuatEB~kmUv@uUoTd6~>z&h|Hf`sb?_c$9jcK&4r*Xc4ZqJO{ZSt)QvB2SxjgMxCkX$%=*r+U%WhV?(lMUj3rtG|1DzJN23}xnbQNKA54x1|Rib~s j_&IhVeI9$MgCA=5WS7?bIPKmNRLg1k&j}dV|8D{S9x_6b literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-12-Analyst, G.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-12-Analyst, G.jpg new file mode 100644 index 0000000000000000000000000000000000000000..bce1bedb86a233de2e0f48dd0a7d96afa10e0751 GIT binary patch literal 1169 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<KW=16jCP7AKLB{__803NOWMu>c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyF9$yR`d1PJQ)Y+9y**<-&*-~U+I3K z)dr_d>OJC5k=?J@^{U3V@|&-mYjFFFpUE$({>?wPX^&*)^c{N=FK}G>xBOQCYrOc) zpH{PCR{TtltG_2Be_8bMi}L;XmlCJU+11)M{|x?5 zBlhv{{q*kX=HySO%=LdC{?D-X&kE+vi@G!z7Ik$!D%rMeTKVsdE0cc&u!}#secnDe zy3}5UyZ?valm84}FTY6pZ+vm>EkJ7Z<93}X;5O80y-qbE_;_5l+wR`qdH`K5ASDJt7qxfH!`uEFMnNPcZ zar@kAk%x00*DQGQPp!e$NHeG+DB@h;6{Gpd{%iO-1m%Mr)QpW{=M~&)xVt!cSeDC@4md?_iojj X{RZkXef+O9n7)bxI^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<5A1R0qH8UG()kO#Vxl@SaWpn#EynT3^&or9B$8>nEb z00R>vGcywlGb<|#3s7|}P@aKBkX1<0(2-3zFp*uUP{gQl;zAB(r;P_igD!qhF-|IK z;^Yz&myncFRa4i{)G{$OGqmaka3YSZQ|TeofBv2)j3M&~ka)>xhT)6Qdr?PR-2hpUWi(FzVCJ$9Vg1iRy z8F3zKBFkrRk0JbZi-Cuk5g2*Qf(-Ty-+yVW$a}YN)3)l)+c_DQ-@l14nug5}*u2eR z)+-Uap2zl63#I0Fyf@|KayiQ)%cS_^udmtl!%=fDwF`7i`C>GuZf*wi7T4{3rVL*Q}SBqE0J9B^VMI zZ%u39T3egQwm9d4-%R$!<-xDC&WbQT3pa?#l$zOJ7@3%}n(trRUT)vLvCjT&Pu;lt zXH*Fu6g^jQ=?44Ooy?PXa_Z*vKg)_|Jr{NF)tz~Q&5d6l{AbW&xO=tu@5JiF<;rf3Dskg5zAM`yFZ%g5q z?lGG8h4E}fdTh_3Y2tpO*E{C+v(%Trj`c0LwY+uR+L)7RQjg?z$k*2M{by(m+J4>C zTkxsw;fEDB74G)uRf;ffeH6LX!<2R29e>$984M-%nznxRzy3{`Ro;D~chW!GGmKgv z`}fpx{M$NHy{j}*WW`ir+pFwfSB32={kBxmz~TMmNk0z#tJPqb7u_2ltiNT;WiwIt ze8(@MFRDr>WquQkRy=p`^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<KW=16jCP7AKLB{__803NOWMu>c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyF8z*7+a()x2avl&i8P^O@S^a)*Ci zV2^aI{MWhSeEOOrdMcBX<#n&#_&ERHKJ5z{42!zJn30n~$dBPw_W3Dui;par7Zh3i zpMgRCZ(?kEr@r0s#oo>B)4yl``1;p4Vju68yS7Jf^9fXDf3=za^5+HCc(KU8^FxJK z96Go2KSLuQ|Cc|nS6^ITcYmp)rDg5z_`eSi{%4Rswz6W?`L|ZqUW@ga^X2{8|I~Ax z`cV4rLA-I66nAx609~=p~-X3LqTuhHyX^{A8Vh_U9oi4{AsMu z|33cDaP4SGo!pk+YTZl|+b2}^ezpIb6V7Zh>3*E2-nB~Aq#J>gUq>+qTyJ2oy}|sq z`JV{m3!!71T|QczvYM)V|7Q@sSGzu3 zi-4QsD-Gtx4@ckMG%azn=qr3We|`S-{$sm;eb#lmy5((9hNJT7nuXt9@jvbQqHFh6 zzvPFWIYVrbTa{e<*Ru@K>---7vrGNYQ1hR`N@D@vn%awBmM-dI&;a7+Tl;_fXRv?Q THSPRA+u)D?N&+3^|K9`v0&n^H literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-2-Analyst, M.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-2-Analyst, M.jpg new file mode 100644 index 0000000000000000000000000000000000000000..c31eb9e297ea8b9ea647cc10e188a220853718eb GIT binary patch literal 1186 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<KW=16jCP7AKLB{__803NOWMu>c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyF8z);S+DKYp(1tVHkPGc|J08~@f@ zF}!}U-sWPy%H^#q_B=YpU0K9t%lMz+^@HmRFEF}_1b{G;F>BD70)~@ndw$8hu2bbX ztf{lR!~Tx!e#O}E&V0L*i#HG69CCr$!Lt6gt!+*s;uEvYp{JY$}lHv4$nuA|n)ce(Q z9sf}JetN0yY&jL_%YPqU|MD+{@n)c_2t%N&yUoe`e3f$r6IUkx2w)d~^!mJgP;{!j z4)^)r$0z?v+Lg<6`AwDSF2VM~XXaNv|JfK<_pJK&@ly9mnGxHruuQ(cG=tN|iH-eR z|KCqv7+(k}otAcctaV_e+=A<~n$;$*DZihu|BUh9;-6Rd8y|`PD)gUWU)+i~*DCpQ zd%6nF748dFE`OB4e0kBtv`$s$DWyEiuSk_9{P6q!)+(~GFYeCU{|pWD+`r1V<_9pw zeD7Xw(`%SJ<?JSc6JbUW*#2;S&a@N%a zJw0+$7@yyh-1}Q@qtwPD#dUA@Enoin&+7}>XTDkoFMl=JVJD-!M(wVy_g~&RS)aAy tZh9qo{Ta*m7xhsLyCdI!zx~=(gmED>npv&qf4$$TQ(qA1%25CRCIB+t@YDbR literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-3-And as you.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-3-And as you.jpg new file mode 100644 index 0000000000000000000000000000000000000000..5aed9503ef22b28aa8cd437c847aaa850ec79687 GIT binary patch literal 1382 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrE1u7?u87N=qmp5CraFLS%1zZi#hJ1J zRU3~MS}pHdmlwNulNv`=-aoVd4E%>SFSt}{>;8RZzD~n`2I)D>Kjv*=e9>8&ubk`o zDEY-YUhlX{^|q}U*UN8~^M8vG{`1N@HazIk(iNOLuyPa5DTN}cfQ*@`?ckzymyzFZ#$oFmFFfu zea8YGd!E;t#W~Z0_o??J8vn`ur?gA!^IQMJ-)H5tKIS!1o~XnZ{>nOdZB6jXXbrcL zneP53DgTlhOlw?My)v_!RXgXTjMjpbWufPTuYXus@}EK6Q732n>8Bp6{GLdLefwrQ z-M)LhTIb^5wi|d1ik9wo$`+HZe4nPxqN_Z4j?$O^3|tKJe!a_$4$3T>^vNK^@f>?% zT;tM+UTeRu+m@-4#jlR9Etuu{DbuQdc41D;x{&6>ieH(p?Y^z+bvdBFcFU=Mt1q@+ z31D5Nv8aJ-x!u&nBF|};#FDRWd+;x*OM^+}Ldh0OvA}@C&H?`Delh>w63zHR+nfKLuX)NIXe-i*PK}4hg literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-4-This secon.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-4-This secon.jpg new file mode 100644 index 0000000000000000000000000000000000000000..8df99a8280826a18095374ffe20ccf4524bda718 GIT binary patch literal 1377 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrtFB+a@bX)Kp><;Tf_KL+1hCF)R$Jk} z_T{waLUBhY-t>F_bk?I%*|hIrv(qAz)wy{p5B~Z!YjgHPTepYIC!ZeRPq+E!)L_-} zQEswwu6P&!Z;!)2yDqdnlAdy(Wu?-u#Z21$lMhbM+S&I^+^$9NpXo#$skZJSZcc^C zKO+5kfAsM(RKAPfHg4A(xy>1+{byBCW{qZ+pyzwL5-P_w`&s4@8Q}SZ|Gce7Y z-&-OAj5r=fR|ZkuyGzWsoo9I=DNz^qXxX9$n{&S|<*El|Y!|q=A>&TDtmC@>3|{-B z`eyT-eE7upY zHSPZyF8CHmDKbZIKL6*d=u07mY3jYZjHcT6RyMuga7ARp-zP#n{iXJQvlzZgX>L5R z>BuC-Q&w%kuXwE*xHJ|uSe};@{5zB3{d4}mELZtio4=i2eEYQgxw((67hJx5tNFXg z?m8R(C-qiaAKXrwRxZv`zouaE3;VzIOBqV89;?{wWf)VtPgTv*VzT_9d;9x;zK)%1 z=5Z?8Q z=0AG*uy^rZXEVlIuMeJI_@6=Q?BpX>TUKwmXK#NuVEGr3X}i>xuCPpPJdiwDWe@ZF z^HB^ff0svGImuk}d3#>-e}*-zRl2>4B5u0o%HQoRoF9DWU+IC{F>ybt19C64)+A0n zeoyZBAMH(-w%(2nf8ub`Leg-<+cz1^L0uXPm?lm!ILzR_a8VaS)=?>$HCfAVB+Yjf sVPw5KckdT9?HSWtzrV6;-Rzy(^N=kBD*1mC0H)^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyFArm)GVdN}Vj#=#6;G(o|H>wK2xM zJMZkV<+7`a6k`PJ_tgKov4C&YiYhM2H&#K(I^KW;VykLuEpRqsiKf|@NMgK(h{x)0v z#PCoI@BR;pC;lz|xPb3nm&O97F0E|~%XCV9zVrW}seI?#-wfuCkEYLWeVr-l7bN(! zf5q!x{~5Giec8DtXUpm8`fbr|A=3M+_E&tW{&k%UulwzVsk+dB#l1ZO@GS`1d(1;QMv%X}-?G9e$HEa;oGWEnk;> z@cQnz%Qh#UNw2EqkH7x&y1xT!(LJL*w|T^}DvO`lUnsDT`4@WS`s~-+eaqL}+4-U` zIq-#e7sCr1>GVzas*lO+ys_W$`d6Tm&{(sZIbUv9i2n#k`899JztXG*uBEFZ|L*sm zcZX+V%<|udCx&M%;5WONU%&ZlMp|;qx#RyCNzbiy;Z-?F8No3`TtD-L7N32 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-6-Let me go .jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-6-Let me go .jpg new file mode 100644 index 0000000000000000000000000000000000000000..12c09918e7576589f96db3c4aea0165def041c26 GIT binary patch literal 1375 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrEArMY+PtmudF{ppCrubl!)N`vbiZ?5 zV$Uz*hV0I(HlCGrPZiRd_!>k09DkK19nbfwJWIy$>B<+E=I;Dsb%DKL*0#CIxzV?a zYHbdNU7b3`XoBMl$8|#foLLz=19{pn#?QaPwOHoYwC=(!CfnX$SzMBTc;NzmuSFZT z_}1x$)~sacd+O%(yzRUE!)OzRaPCKIJ?_TO4pBeZ@Ss2JT|M9Yxn-X{6q`K#U4o8( z5PfF$;`*)@k0;Y+1xWn$H(|W`BiY4cf%oqgUhU5k_J^$HSFW2iF+`~~u$7sC@5&a2 z(CEK*_k*@&q&v8O*j{S4Xl%#*!v^{k*?{i=H z!nHWgcpH;<_5{m>!2NvgpH1zX)pc`A^ZYpu9{Bb3UkKyQRnLFReV6}y@IS*B;TXQo z{N2GAa%zhqQyRBEkIagK0y*hjP*6J9?>y4lFuay7MTDxuksz`p_JjtmQ zw>9qepP$8W^P}pymANZ)uKlYC`?!Gd`n$_!h2ymHKVkw6sGj-oDVj z@a4AZCRUomu};E{4}+&Eq>wK*kkk8M83-Y-DzLhGvjuf z{PSB4`WIbA7($GrZr`b2*Q9BG-=Ti4NC1oHvIxafH=~^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrE1u7@&XCKTqmp5CdwbSH6n!*%x(EDfhL=g2$h1HJC4Htl29c(;U{b zLN4Z&(Z$KX`U9h;FoX*}+Us*CuC~j}F2ufG{>c5m`5)I^iew4c)4oVE`A7JSBkOBd zC~Z7_UW4WL#{UejFR=SQT+3$6b^q>yq&r7?FYv#Ye;FHVDwy0G$R_H*P{mcsAnG1` z>FrnVVwqFSjh{T&wd&7j?R90S?B<@U9IR}u-uUaA{of?*i)->O{X3;S(WlKW8I9q#msHo?TSYdzksySRsEQfaUu3;H+Ddy|+xC*tCeRz?$m|uhx;wTP0x! zo=!>aUoE%wP^S6mC0TZN{ls3`~uzU9R&Ma__nL lC^X2W^Kjvx*X#!;EZn(gO8;EZQ2&kU_9p^e8S4Mv1OP8SI5z+Q literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-8-Hi, guys. .jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-8-Hi, guys. .jpg new file mode 100644 index 0000000000000000000000000000000000000000..c4118e604b5c61dee115e9e170b7b5d0da1c583f GIT binary patch literal 1207 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyFAP9zEj@;V(3DS@J;KLULtx*T>rD zbL&MH&FKi&VgKI$pCMX)E<@>r_?veb?n}5SO0iy@-UKe*8a_S&VItW$w$lQ zr(LR~q;IhQT&49~802fE~E z*!On*y4#D7?s(?0zwCay|7DSVN}GSDbsI42i@z8D@cLKU&hs4ieo9S_HvFj`XMazo z|FQ|=oj_L+hCo-x>Y{CB!7)PyBg-UGUND^INmtO3hiqobc=We+K?TyWT3X zY`V1efzHCF)Sds7PyQ=?{_Wa!tGN&R6!)c{7V@(_xmNy%Yiaz?C-c952w<;>Gvd%$ zBDdJ|+4aDE(^v00R@oMNqIOsRWzq6~g0bQMD%Jn~5e@s#;I{X2Qe<8-SI=!n9)1(X z7t^kDmS&5*{GJkU_~Xj#*&k}(&rVfcEI&)^&i;4v|Lv={da1#%F8!GKyx1jXS(Q`X z=P!8q>pz3m%hj#db4#WFMrDSbdT?xclDy6r`6u&_HJI2)r*}NxlXzNQ*KV%=mp`jl zt*VM$exu}s?#Itsf9}7R^M4(#!LUl|>ipYVW#n%m2w^*O}>UnhUnoUk;cXBBzv(KrM)!ukHFSXW+xmaYS SL5xL;<%!4k{ZS16Zvp_*eDKu( literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-9-Co-Founder.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-earnings-call/images/image-9-Co-Founder.jpg new file mode 100644 index 0000000000000000000000000000000000000000..f892d2fd01297ad961e45f9bc4e51894d76e5730 GIT binary patch literal 1373 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<5A1R0qH8UG()kO#Vxl@SaWpn#EynT3^&or9B$8>nEb z00R>vGcywlGb<|#3s7|}P@aKBkX1<0(2-3zFp*uUP{gQl;zAB(r;P_igD!qhF-|IK z;^Yz&myncFRa4i{)G{$OGqmaka3YSZQ|TeofBv2)j3M&~ka)>xhT)6Qdr?PR-2hpUWi(FzVCJ$9Vg1iRy z8F3zKBFkrRk0JbZi-Cuk5g2*Qf(-Ty-+yVWx_*7b%WvxotrNo+ygPm&fOXbFyRFM^ z&+)7ja}QWv_*c90O6lsmQMofyI_BsYPCkGA%GKHNTv_H^a*i?#-zI$i$;I$W_(+sb z@0K_X`?uQs&$OK+AMWt|aB7nNDg&h*FKwhNR?m2sSp1?$TP>ngR_uwGQ1r>0Q$F19 zcyNJz>BR^s{*LK^O81^;#WWt2VgJwY`r2Ox(E~r`2b#~C?X>)R*pmfvYw!GL@ISPB zb!Sa?W~!|qcb&88v#^QMdz~a^>!eO$vkyMYP-Z82@zSNM`f34rs**AP8JOnH?_Af^ z08DEQ)`1SJj^&xWZ#O$yvAO-VJ-*l9m7#L(*Cn_7mR!tJe4KFN@s3x#<-bBdTq(Jc zSHW|bk3CH`d;Z6zJN`4c?qr^qocyc*wf~2hCr3*eZM3x;zn-u^5y19u73U>=hu2T+ z|21VdT;!0tJFouviuE27SvK0_UYxS>Pxcvsy=@DYw!c@Hl2W}A=&b?CKBK)>$l$mOg%K@!yc13b|vu9uyf_g-fog@XMHUGV%9~Uq4g+oc{84 z&5ZS5eGPtQ>wI|pL#Du<*+F`rYIBHH)7{%DizF^+?eRVF&Xalme}+)U584HjUSGyc~-FBS6|EE0\n[images/image-1-Financial Growth Comparison.jpg]\n", + "path": "images/image-1-Financial Growth Comparison.jpg", + "metadata": { + "length": 186, + "summary": "image-1\nQ1 FY26: $44.1B\nQ1 FY27: $81.6B", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "images/image-1-Financial Growth Comparison.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "553b8966-4ebb-54a7-87a4-dd6e473795bf", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PeriodValue
Q1 FY26$44.1B
Q1 FY27$81.6B
", + "path": "tables/table-0 Revenue Forecast.html", + "metadata": { + "length": 304, + "summary": "table-1\nThe table shows projected revenue values for Q1 FY26 at $44.1B and Q1 FY27 at $81.6B, indicating significant growth.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-0 Revenue Forecast.html", + "keywords": [ + "Revenue", + "Forecast", + "Quarterly" + ], + "tokens": [] + } + }, + { + "chunk_id": "d7d0b5bb-78a5-50de-addd-bb549031d607", + "type": "image", + "content": "\n

0\n[images/image-2-Q1 FY26 vs Q1 FY27 Revenue.jpg]\n", + "path": "images/image-2-Q1 FY26 vs Q1 FY27 Revenue.jpg", + "metadata": { + "length": 55, + "summary": "image-2\n- Chart Type: Bar Chart comparing financial data across two periods.\n- Data Points:\n - Q1 FY26: $17.6B (represented by a grey bar)\n - Q1 FY27: $37.9B (represented by a green bar)\n- Trend: Significant increase in value from Q1 FY26 to Q1 FY27, more than doubling the previous period's figure.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "images/image-2-Q1 FY26 vs Q1 FY27 Revenue.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "308bfa4c-5008-5951-a18b-5e9ba3b08d19", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PeriodValue
Q1 FY26$17.6B
Q1 FY27$37.9B
", + "path": "tables/table-1 Revenue Forecast.html", + "metadata": { + "length": 304, + "summary": "table-2\nThe table shows projected revenue values for Q1 FY26 at $17.6B and Q1 FY27 at $37.9B.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-1 Revenue Forecast.html", + "keywords": [ + "Revenue", + "Forecast", + "Quarterly" + ], + "tokens": [] + } + }, + { + "chunk_id": "6a7e25e7-5eb4-5ef9-b0e7-b7f4fb930863", + "type": "image", + "content": "\n
0\n[images/image-3-Financial Growth Comparison.jpg]\n", + "path": "images/image-3-Financial Growth Comparison.jpg", + "metadata": { + "length": 55, + "summary": "image-3\nQ1 FY26: $21.5B\nQ1 FY27: $37.4B", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "images/image-3-Financial Growth Comparison.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e5595fd3-c434-5e91-87d0-af02e4fa91d9", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PeriodValue
Q1 FY26$21.5B
Q1 FY27$37.4B
", + "path": "tables/table-2 Revenue Forecast.html", + "metadata": { + "length": 304, + "summary": "table-3\nThe table shows projected revenue values for Q1 FY26 at $21.5B and Q1 FY27 at $37.4B, indicating significant growth.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-2 Revenue Forecast.html", + "keywords": [ + "Revenue", + "Forecast", + "Quarterly" + ], + "tokens": [] + } + }, + { + "chunk_id": "89765158-7b6e-5bea-bfd5-b5c2a7b211da", + "type": "text", + "content": "\\$ in billions\nStrong growth across a diverse end-customer base – hyperscalers to model builders to AI Clouds, enterprise on-premise and sovereign customers\nBlackwell adopted and deployed by every major hyperscaler, every cloud provider, and every model builder\nAgentic AI and reinforcement learning represent new CPU growth opportunities\nVera Rubin on track for the second half of this year, starting in Q3\nNVIDIA AI infrastructure delivers the lowest token cost and highest token throughput as shown in latest MLPerf and InferenceX benchmarks\n\nStrong growth across a diverse end-customer base – hyperscalers to model builders to AI Clouds, enterprise on-premise and sovereign customers
0\n[images/image-1-Financial Growth Comparison.jpg]\n\n
\nbar chart\n\n[tables/table-0 Revenue Forecast.html]\n\n
\nData Center\nRevenue \\$ in billions\nHyperscale\n\n
0\n[images/image-2-Q1 FY26 vs Q1 FY27 Revenue.jpg]\n\n
\nbar chart\n\n[tables/table-1 Revenue Forecast.html]\n\n
\nAI Clouds, Industrial, and Enterprise (ACIE)\n\n
0\n[images/image-3-Financial Growth Comparison.jpg]\n\n
\nbar chart\n\n[tables/table-2 Revenue Forecast.html]\n\n
\nContinued strong demand for our Blackwell architecture", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Revenue \\$ in billions", + "metadata": { + "length": 1250, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "tokens": [ + "billions", + "Strong", + "growth", + "diverse", + "end", + "customer", + "base", + "hyperscalers", + "model", + "builders", + "AI", + "Clouds", + "enterprise", + "premise", + "sovereign", + "customers", + "Blackwell", + "adopted", + "deployed", + "major", + "hyperscaler", + "cloud", + "provider", + "builder", + "Agentic", + "reinforcement", + "learning", + "represent", + "CPU", + "opportunities", + "Vera", + "Rubin", + "track", + "half", + "year", + "starting", + "Q3", + "NVIDIA", + "infrastructure", + "delivers", + "lowest", + "token", + "cost", + "highest", + "throughput", + "shown", + "latest", + "MLPerf", + "InferenceX", + "benchmarks", + "details", + "summary", + "bar", + "chart", + "Data", + "Center", + "Revenue", + "Hyperscale", + "Industrial", + "Enterprise", + "ACIE", + "Continued", + "strong", + "demand", + "architecture" + ], + "keywords": [], + "connect_to": [ + { + "target": "fbdc85d3-12a0-59ff-8076-6a7ea5545daa", + "relation": "embeds", + "ref": "[images/image-1-Financial Growth Comparison.jpg]", + "position": { + "start": 699, + "end": 747 + } + }, + { + "target": "553b8966-4ebb-54a7-87a4-dd6e473795bf", + "relation": "embeds", + "ref": "[tables/table-0 Revenue Forecast.html]", + "position": { + "start": 789, + "end": 827 + } + }, + { + "target": "d7d0b5bb-78a5-50de-addd-bb549031d607", + "relation": "embeds", + "ref": "[images/image-2-Q1 FY26 vs Q1 FY27 Revenue.jpg]", + "position": { + "start": 909, + "end": 956 + } + }, + { + "target": "308bfa4c-5008-5951-a18b-5e9ba3b08d19", + "relation": "embeds", + "ref": "[tables/table-1 Revenue Forecast.html]", + "position": { + "start": 998, + "end": 1036 + } + }, + { + "target": "6a7e25e7-5eb4-5ef9-b0e7-b7f4fb930863", + "relation": "embeds", + "ref": "[images/image-3-Financial Growth Comparison.jpg]", + "position": { + "start": 1117, + "end": 1165 + } + }, + { + "target": "e5595fd3-c434-5e91-87d0-af02e4fa91d9", + "relation": "embeds", + "ref": "[tables/table-2 Revenue Forecast.html]", + "position": { + "start": 1207, + "end": 1245 + } + } + ] + } + }, + { + "chunk_id": "3dc86686-3dde-592d-9ea5-bb8e20ab9a90", + "type": "image", + "content": "\n
0\n[images/image-4-Q1 FY26 vs Q1 FY27.jpg]\n", + "path": "images/image-4-Q1 FY26 vs Q1 FY27.jpg", + "metadata": { + "length": 55, + "summary": "image-4\n- Chart Type: Bar Chart\n- Data Points:\n - Q1 FY26: 60.5%\n - Q1 FY27: 74.9%\n- Trend: Significant increase from the first quarter of fiscal year 26 to the first quarter of fiscal year 27.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "images/image-4-Q1 FY26 vs Q1 FY27.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "cfaaa7df-6ca2-5601-aee2-c956585b36f6", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PeriodValue (%)
Q1 FY2660.5
Q1 FY2774.9
", + "path": "tables/table-3 Quarterly Values.html", + "metadata": { + "length": 304, + "summary": "table-4\nThe table shows two quarterly periods with their corresponding percentage values: Q1 FY26 at 60.5% and Q1 FY27 at 74.9%.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-3 Quarterly Values.html", + "keywords": [ + "Period", + "Value", + "Percentage" + ], + "tokens": [] + } + }, + { + "chunk_id": "1ba62209-1422-54b3-a0dd-3ff8bf818ac4", + "type": "image", + "content": "\n
0\n[images/image-5-Q1 FY26 vs Q1 FY27.jpg]\n", + "path": "images/image-5-Q1 FY26 vs Q1 FY27.jpg", + "metadata": { + "length": 55, + "summary": "image-5\n- Chart Type: Bar Chart\n- Data Points:\n - Q1 FY26: 60.8%\n - Q1 FY27: 75.0%\n- Trend: Significant increase from the first period to the second.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "images/image-5-Q1 FY26 vs Q1 FY27.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "18a8955f-f1b6-563c-8d4e-f69d3cb6e306", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PeriodValue (%)
Q1 FY2660.8
Q1 FY2775.0
", + "path": "tables/table-4 Quarterly Values.html", + "metadata": { + "length": 304, + "summary": "table-5\nThe table shows two quarterly values: 60.8% for Q1 FY26 and 75.0% for Q1 FY27, indicating an increase.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-4 Quarterly Values.html", + "keywords": [ + "Q1 FY26", + "Q1 FY27", + "Value" + ], + "tokens": [] + } + }, + { + "chunk_id": "eee301e5-988f-5e25-9d67-dd621388b161", + "type": "text", + "content": "GAAP\n\n
0\n[images/image-4-Q1 FY26 vs Q1 FY27.jpg]\n\n
\nbar chart\n\n[tables/table-3 Quarterly Values.html]\n\n
\nNon-GAAP\n\n
0\n[images/image-5-Q1 FY26 vs Q1 FY27.jpg]\n\n
\nbar chart\n\n[tables/table-4 Quarterly Values.html]\n\n
\nGross margin year-on-year increase driven by lower inventory provisions, primarily due to a \\$4.5B charge associated with H2O excess inventory in the prior year", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Gross Margin", + "metadata": { + "length": 468, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "tokens": [ + "GAAP", + "details", + "summary", + "bar", + "chart", + "Non", + "Gross", + "margin", + "year", + "increase", + "driven", + "lower", + "inventory", + "provisions", + "primarily", + "due", + "4.5", + "charge", + "H2O", + "excess", + "prior" + ], + "keywords": [], + "connect_to": [ + { + "target": "3dc86686-3dde-592d-9ea5-bb8e20ab9a90", + "relation": "embeds", + "ref": "[images/image-4-Q1 FY26 vs Q1 FY27.jpg]", + "position": { + "start": 28, + "end": 67 + } + }, + { + "target": "cfaaa7df-6ca2-5601-aee2-c956585b36f6", + "relation": "embeds", + "ref": "[tables/table-3 Quarterly Values.html]", + "position": { + "start": 109, + "end": 147 + } + }, + { + "target": "1ba62209-1422-54b3-a0dd-3ff8bf818ac4", + "relation": "embeds", + "ref": "[images/image-5-Q1 FY26 vs Q1 FY27.jpg]", + "position": { + "start": 192, + "end": 231 + } + }, + { + "target": "18a8955f-f1b6-563c-8d4e-f69d3cb6e306", + "relation": "embeds", + "ref": "[tables/table-4 Quarterly Values.html]", + "position": { + "start": 273, + "end": 311 + } + } + ] + } + }, + { + "chunk_id": "d0ddf35d-5f35-545a-9b0e-d8b8baebaf6f", + "type": "image", + "content": "\n
0\n[images/image-6-GAAP Revenue Growth.jpg]\n", + "path": "images/image-6-GAAP Revenue Growth.jpg", + "metadata": { + "length": 55, + "summary": "image-6\nChart Title: GAAP\nData Points:\n- Q1 FY26: $5.0B\n- Q1 FY27: $7.6B", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "images/image-6-GAAP Revenue Growth.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "21cb7c93-5982-58db-94df-c33e726be8ca", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PeriodValue ($B)
Q1 FY265.0
Q1 FY277.6
", + "path": "tables/table-5 Revenue Forecast.html", + "metadata": { + "length": 303, + "summary": "table-6\nThe table shows projected revenue values of $5.0B for Q1 FY26 and $7.6B for Q1 FY27, indicating growth.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-5 Revenue Forecast.html", + "keywords": [ + "Revenue", + "Forecast", + "Quarterly" + ], + "tokens": [] + } + }, + { + "chunk_id": "95da4069-d756-52e5-a23e-92c3f645c276", + "type": "image", + "content": "\n
0\n[images/image-7-Non-GAAP Revenue Growth.jpg]\n", + "path": "images/image-7-Non-GAAP Revenue Growth.jpg", + "metadata": { + "length": 55, + "summary": "image-7\nChart Title: Non-GAAP\n\nData Points:\n- Q1 FY26: $5.0B\n- Q1 FY27: $7.4B", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "images/image-7-Non-GAAP Revenue Growth.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "7e207392-6e74-5f9a-96d8-2d1cd59d8c2e", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PeriodValue ($B)
Q1 FY265.0
Q1 FY277.4
", + "path": "tables/table-6 Revenue Forecast.html", + "metadata": { + "length": 303, + "summary": "table-7\nThe table shows projected revenue values for Q1 FY26 at $5.0B and Q1 FY27 at $7.4B, indicating growth over the two periods.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-6 Revenue Forecast.html", + "keywords": [ + "Revenue", + "Forecast", + "Quarterly" + ], + "tokens": [] + } + }, + { + "chunk_id": "cee412ca-51ad-5c67-afbe-1265a6a2197d", + "type": "image", + "content": "\n
Top-line growth and operating leverage drove accelerating EPS growth0\n[images/image-8-GAAP vs Non-GAAP Comparison.jpg]\n", + "path": "images/image-8-GAAP vs Non-GAAP Comparison.jpg", + "metadata": { + "length": 114, + "summary": "image-8\n- GAAP: Q1 FY26 ($0.76), Q1 FY27 ($2.39)\n- Non-GAAP: Q1 FY26 ($0.78), Q1 FY27 ($1.87)", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "images/image-8-GAAP vs Non-GAAP Comparison.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "b12ffcab-99f1-52e6-b9a3-60441e206605", + "type": "text", + "content": "\\$ in billions\n\n
0\n[images/image-6-GAAP Revenue Growth.jpg]\n\n
\nbar chart\nGAAP\n\n[tables/table-5 Revenue Forecast.html]\n\n
\n\n
0\n[images/image-7-Non-GAAP Revenue Growth.jpg]\n\n
\nbar chart\nNon-GAAP\n\n[tables/table-6 Revenue Forecast.html]\n\n
\nInvesting to address the multitude of growth opportunities ahead\nEarnings per Share\n\n
Top-line growth and operating leverage drove accelerating EPS growth0\n[images/image-8-GAAP vs Non-GAAP Comparison.jpg]\n\nTop-line growth and operating leverage drove accelerating EPS growth", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Operating Expenses", + "metadata": { + "length": 592, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "tokens": [ + "billions", + "details", + "summary", + "bar", + "chart", + "GAAP", + "Non", + "Investing", + "address", + "multitude", + "growth", + "opportunities", + "ahead", + "Earnings", + "Share", + "Top", + "line", + "operating", + "leverage", + "drove", + "accelerating", + "EPS", + "growth0" + ], + "keywords": [], + "connect_to": [ + { + "target": "d0ddf35d-5f35-545a-9b0e-d8b8baebaf6f", + "relation": "embeds", + "ref": "[images/image-6-GAAP Revenue Growth.jpg]", + "position": { + "start": 38, + "end": 78 + } + }, + { + "target": "21cb7c93-5982-58db-94df-c33e726be8ca", + "relation": "embeds", + "ref": "[tables/table-5 Revenue Forecast.html]", + "position": { + "start": 125, + "end": 163 + } + }, + { + "target": "95da4069-d756-52e5-a23e-92c3f645c276", + "relation": "embeds", + "ref": "[images/image-7-Non-GAAP Revenue Growth.jpg]", + "position": { + "start": 199, + "end": 243 + } + }, + { + "target": "7e207392-6e74-5f9a-96d8-2d1cd59d8c2e", + "relation": "embeds", + "ref": "[tables/table-6 Revenue Forecast.html]", + "position": { + "start": 294, + "end": 332 + } + }, + { + "target": "cee412ca-51ad-5c67-afbe-1265a6a2197d", + "relation": "embeds", + "ref": "[images/image-8-GAAP vs Non-GAAP Comparison.jpg]", + "position": { + "start": 511, + "end": 559 + } + } + ] + } + }, + { + "chunk_id": "f5f4a4d6-0f47-53c6-8553-aa7e00cca420", + "type": "image", + "content": "\nTop-line growth and operating leverage drove accelerating EPS growth
0\n[images/image-9-Financial Growth Comparison.jpg]\n", + "path": "images/image-9-Financial Growth Comparison.jpg", + "metadata": { + "length": 113, + "summary": "image-9\nQ1 FY26: $26.1B\nQ1 FY27: $48.6B", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "images/image-9-Financial Growth Comparison.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ec2e3b6d-2257-547d-a660-dce84a02bebe", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PeriodValue
Q1 FY26$26.1B
Q1 FY27$48.6B
", + "path": "tables/table-7 Revenue Forecast.html", + "metadata": { + "length": 304, + "summary": "table-8\nThe table shows projected revenue values for Q1 FY26 at $26.1B and Q1 FY27 at $48.6B, indicating significant growth.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-7 Revenue Forecast.html", + "keywords": [ + "Revenue", + "Forecast", + "Quarterly" + ], + "tokens": [] + } + }, + { + "chunk_id": "c941e7ab-e113-5e57-bbb3-ae791c57edca", + "type": "image", + "content": "\n
0\n[images/image-10-Dividends vs Share Repurchases.jpg]\n", + "path": "images/image-10-Dividends vs Share Repurchases.jpg", + "metadata": { + "length": 56, + "summary": "image-10\nChart Title: Dividends and Share Repurchases Comparison\n\nData Points:\n- Q1 FY26: ~$14.0B (represented by a grey bar)\n- Q1 FY27: ~$20.0B (represented by a bright green bar)\n\nLegend:\n- Dark Green Square: Dividends\n- Bright Green Square: Share Repurchases\n\nObservations:\n- The chart compares financial figures for two quarters: Q1 FY26 and Q1 FY27.\n- There is an increase in value from approximately $14.0 billion to $20.0 billion between the two periods.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "images/image-10-Dividends vs Share Repurchases.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "406869a7-b4d8-5901-91d5-8d5f840ae4d7", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PeriodDividendsShare Repurchases
Q1 FY26~$14.0B-
Q1 FY27-~$20.0B
", + "path": "tables/table-8 Dividends & Buybacks.html", + "metadata": { + "length": 377, + "summary": "table-9\nThe table shows Q1 FY26 dividends of ~$14.0B and Q1 FY27 share repurchases of ~$20.0B.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-8 Dividends & Buybacks.html", + "keywords": [ + "Dividends", + "Share Repurchases", + "FY26-FY27" + ], + "tokens": [] + } + }, + { + "chunk_id": "28a269ec-3fcf-5432-8194-07a91613d637", + "type": "text", + "content": "\\$ in billions\nFree Cash Flow\n\nTop-line growth and operating leverage drove accelerating EPS growth
0\n[images/image-9-Financial Growth Comparison.jpg]\n\n
\nbar chart\n\n[tables/table-7 Revenue Forecast.html]\n\n
\nCapital Return\n\n
0\n[images/image-10-Dividends vs Share Repurchases.jpg]\n\n
\nbar chart\n\n[tables/table-8 Dividends & Buybacks.html]\n\n
\nReturned a record \\~\\$20B to our shareholders, while executing strategic investments. Announced new \\$80B share repurchase authorization, on top of \\~\\$39B remaining as of Q1 FY27. Increased quarterly dividend from \\$0.01 to \\$0.25 per share.", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Free Cash Flow and Capital Return", + "metadata": { + "length": 654, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "tokens": [ + "billions", + "Free", + "Cash", + "Flow", + "Top", + "line", + "growth", + "operating", + "leverage", + "drove", + "accelerating", + "EPS", + "details", + "summary", + "bar", + "chart", + "Capital", + "Return", + "Returned", + "record", + "20B", + "shareholders", + "executing", + "strategic", + "investments", + "Announced", + "80B", + "share", + "repurchase", + "authorization", + "top", + "39B", + "remaining", + "Q1", + "FY27", + "Increased", + "quarterly", + "dividend", + "0.01", + "0.25" + ], + "keywords": [], + "connect_to": [ + { + "target": "f5f4a4d6-0f47-53c6-8553-aa7e00cca420", + "relation": "embeds", + "ref": "[images/image-9-Financial Growth Comparison.jpg]", + "position": { + "start": 111, + "end": 159 + } + }, + { + "target": "ec2e3b6d-2257-547d-a660-dce84a02bebe", + "relation": "embeds", + "ref": "[tables/table-7 Revenue Forecast.html]", + "position": { + "start": 201, + "end": 239 + } + }, + { + "target": "c941e7ab-e113-5e57-bbb3-ae791c57edca", + "relation": "embeds", + "ref": "[images/image-10-Dividends vs Share Repurchases.jpg]", + "position": { + "start": 290, + "end": 342 + } + }, + { + "target": "406869a7-b4d8-5901-91d5-8d5f840ae4d7", + "relation": "embeds", + "ref": "[tables/table-8 Dividends & Buybacks.html]", + "position": { + "start": 384, + "end": 426 + } + } + ] + } + }, + { + "chunk_id": "f020fa0f-d13d-53cb-abc9-b7f2ac9ae38e", + "type": "table", + "content": "
Q1 FY27YoY %Q1 FY26
Revenue$81.685%$44.1
Gross Margin %74.9%+14.4 pts60.5%
Operating Expenses$7.652%$5.0
Operating Income$53.5147%$21.6
Net Income$58.3211%$18.8
Diluted EPS$2.39214%$0.76
", + "path": "tables/table-9 Q1 FY27 Results.html", + "metadata": { + "length": 510, + "summary": "table-10\nRevenue grew 85% YoY to $81.6M, gross margin improved to 74.9%, and net income surged 211% to $58.3M in Q1 FY27.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-9 Q1 FY27 Results.html", + "keywords": [ + "Revenue", + "Gross Margin", + "Net Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "07b055c7-c8cc-5133-84ef-fc65b788a59b", + "type": "text", + "content": "\\$ in billions, except EPS\n\n[tables/table-9 Q1 FY27 Results.html]", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/GAAP P&L Financial Measures \\$ in billions, except EPS", + "metadata": { + "length": 72, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "tokens": [ + "billions", + "EPS" + ], + "keywords": [], + "connect_to": [ + { + "target": "f020fa0f-d13d-53cb-abc9-b7f2ac9ae38e", + "relation": "embeds", + "ref": "[tables/table-9 Q1 FY27 Results.html]", + "position": { + "start": 28, + "end": 65 + } + } + ] + } + }, + { + "chunk_id": "eb119c2a-6a1e-5597-b9bc-f93d25d10677", + "type": "table", + "content": "
Q1 FY27YoY %Q1 FY26
Revenue$81.685%$44.1
Gross Margin %75.0%+14.2 pts60.8%
Operating Expenses$7.449%$5.0
Operating Income$53.8147%$21.8
Net Income$45.5139%$19.1
Diluted EPS$1.87140%$0.78
", + "path": "tables/table-10 Q1 FY27 Results.html", + "metadata": { + "length": 510, + "summary": "table-11\nRevenue grew 85% YoY to $81.6M, gross margin improved to 75.0%, and net income rose 139% to $45.5M, with diluted EPS at $1.87.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-10 Q1 FY27 Results.html", + "keywords": [ + "Revenue", + "Gross Margin", + "Net Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "07b055c7-c8cc-5133-84ef-fc65b788a59b", + "type": "text", + "content": "\\$ in billions, except EPS\n\n[tables/table-10 Q1 FY27 Results.html]", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Non-GAAP P&L Financial Measures \\$ in billions, except EPS", + "metadata": { + "length": 72, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "tokens": [ + "billions", + "EPS" + ], + "keywords": [], + "connect_to": [ + { + "target": "eb119c2a-6a1e-5597-b9bc-f93d25d10677", + "relation": "embeds", + "ref": "[tables/table-10 Q1 FY27 Results.html]", + "position": { + "start": 28, + "end": 66 + } + } + ] + } + }, + { + "chunk_id": "d73747ff-4fd8-5177-bacb-8bbfbacf3632", + "type": "table", + "content": "
Q2 FY27
Revenue$91.0 billion*+ / - 2%
Gross Margin %75.0%+ / - 50 bps
Operating Expenses$8.3 billionApproximately
", + "path": "tables/table-11 Q2 FY27 Guidance.html", + "metadata": { + "length": 269, + "summary": "table-12\nQ2 FY27 revenue guidance is $91.0 billion, plus or minus 2%. Gross margin is 75.0%, plus or minus 50 bps. Operating expenses are approximately $8.3 billion.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-11 Q2 FY27 Guidance.html", + "keywords": [ + "Revenue", + "Gross Margin", + "Operating Expenses" + ], + "tokens": [] + } + }, + { + "chunk_id": "f97eedeb-62b4-5a17-a2f4-9307e17a2822", + "type": "text", + "content": "[tables/table-11 Q2 FY27 Guidance.html]\n\nFor the full year fiscal 2027, expect non-GAAP tax rates to be between 16.0% and 18.0%, excluding any discrete items and material changes to our tax environment", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Q2 FY27 Outlook – Non-GAAP", + "metadata": { + "length": 192, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "tokens": [ + "For", + "full", + "year", + "fiscal", + "2027", + "expect", + "GAAP", + "tax", + "rates", + "16.0%", + "18.0%", + "excluding", + "discrete", + "items", + "material", + "environment" + ], + "keywords": [], + "connect_to": [ + { + "target": "d73747ff-4fd8-5177-bacb-8bbfbacf3632", + "relation": "embeds", + "ref": "[tables/table-11 Q2 FY27 Guidance.html]", + "position": { + "start": 0, + "end": 39 + } + } + ] + } + }, + { + "chunk_id": "85f5d248-d5c3-5de0-a2b2-72dcd545ec6a", + "type": "table", + "content": "
Q1 FY27Non-GAAPAcquisition-Related and Other Costs (A)Other (B)Tax Impact of AdjustmentsGAAP
Gross margin ($ in million)$61,232(47)(28)$61,157
Gross margin %75.0%(0.1)74.9%
Operating expense ($ in million)$7,449172$7,621
Operating income ($ in million)$53,783(219)(28)$53,536
Net income ($ in million)$45,548(219)15,882(2,890)$58,321
Diluted shares (millions)24,39124,391
Diluted EPS$1.87(0.01)0.65(0.12)$2.39
", + "path": "tables/table-12 Q1 FY27 GAAP.html", + "metadata": { + "length": 930, + "summary": "table-13\nThe table shows Q1 FY27 financials: Non-GAAP gross margin 75.0% ($61,232M), GAAP 74.9% ($61,157M); Net income GAAP $58,321M; Diluted EPS GAAP $2.39.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-12 Q1 FY27 GAAP.html", + "keywords": [ + "Gross margin", + "Net income", + "Diluted EPS" + ], + "tokens": [] + } + }, + { + "chunk_id": "d952a3ac-47ad-5454-b7df-e77b4fc3afd3", + "type": "text", + "content": "Reconciliation of Non-GAAP to GAAP Financial Measures\n\n[tables/table-12 Q1 FY27 GAAP.html]\n\nBeginning in Q1 FY27, NVIDIA's non-GAAP financial measures no longer exclude stock-based compensation expense. The historical non-GAAP financial information presented has been updated to include stock-based compensation expense.\nA. Consists of amortization of intangible assets, transaction costs, and certain compensation charges.\nB. Other consists of net gain from equity securities, interest expense related to acquisition consideration discount to be paid in the future and amortization of debt discount, legal settlements, dividend income on equity securities, and share of net earnings related to equity method investments.", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Reconciliation of Non-GAAP to GAAP Financial Measures", + "metadata": { + "length": 788, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "tokens": [ + "Reconciliation", + "Non", + "GAAP", + "Financial", + "Measures", + "Beginning", + "Q1", + "FY27", + "NVIDIA", + "financial", + "measures", + "longer", + "exclude", + "stock", + "based", + "compensation", + "expense", + "The", + "historical", + "information", + "presented", + "updated", + "include", + "Consists", + "amortization", + "intangible", + "assets", + "transaction", + "costs", + "charges", + "Other", + "consists", + "net", + "gain", + "equity", + "securities", + "interest", + "related", + "acquisition", + "consideration", + "discount", + "paid", + "future", + "debt", + "legal", + "settlements", + "dividend", + "income", + "share", + "earnings", + "method", + "investments" + ], + "keywords": [], + "connect_to": [ + { + "target": "85f5d248-d5c3-5de0-a2b2-72dcd545ec6a", + "relation": "embeds", + "ref": "[tables/table-12 Q1 FY27 GAAP.html]", + "position": { + "start": 55, + "end": 90 + } + } + ] + } + }, + { + "chunk_id": "5ffd9825-2e9a-5d4d-8d9f-30261f2ddf13", + "type": "table", + "content": "
Q1 FY26Non-GAAPAcquisition-Related and Other Costs (A)Other (B)Tax Impact of AdjustmentsGAAP
Gross margin ($ in million)$26,794(123)(3)$26,668
Gross margin %60.8%(0.3)60.5%
Operating expense ($ in million)$4,99337$5,030
Operating income ($ in million)$21,801(160)(3)$21,638
Net income ($ in million)$19,094(160)(179)20$18,775
Diluted shares (millions)24,61124,611
Diluted EPS$0.78(0.01)(0.01)$0.76
", + "path": "tables/table-13 Q1 FY26 Financials.html", + "metadata": { + "length": 919, + "summary": "table-14\nThe table shows Q1 FY26 financials with Non-GAAP gross margin of 60.8% and GAAP of 60.5%. Net income is $19,094M Non-GAAP vs $18,775M GAAP, with diluted EPS of $0.78 and $0.76 respectively.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-13 Q1 FY26 Financials.html", + "keywords": [ + "Non-GAAP", + "GAAP", + "Gross Margin" + ], + "tokens": [] + } + }, + { + "chunk_id": "797a48f3-cd54-59df-b9fb-f1aa47cba117", + "type": "text", + "content": "[tables/table-13 Q1 FY26 Financials.html]\n\nBeginning in Q1 FY27, NVIDIA's non-GAAP financial measures no longer exclude stock-based compensation expense. The historical non-GAAP financial information presented has been updated to include stock-based compensation expense.\nA. Consists of amortization of intangible assets, transaction costs, and certain compensation charges.\nB. Other consists of legal settlements, net loss from equity securities, and interest expense related to amortization of debt discount.", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Reconciliation of Non-GAAP to GAAP Financial Measures_2", + "metadata": { + "length": 571, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "tokens": [ + "Beginning", + "Q1", + "FY27", + "NVIDIA", + "GAAP", + "financial", + "measures", + "longer", + "exclude", + "stock", + "based", + "compensation", + "expense", + "The", + "historical", + "information", + "presented", + "updated", + "include", + "Consists", + "amortization", + "intangible", + "assets", + "transaction", + "costs", + "charges", + "Other", + "consists", + "legal", + "settlements", + "net", + "loss", + "equity", + "securities", + "interest", + "related", + "debt", + "discount" + ], + "keywords": [], + "connect_to": [ + { + "target": "5ffd9825-2e9a-5d4d-8d9f-30261f2ddf13", + "relation": "embeds", + "ref": "[tables/table-13 Q1 FY26 Financials.html]", + "position": { + "start": 0, + "end": 41 + } + } + ] + } + }, + { + "chunk_id": "e757adf1-b61f-5447-adf3-b3f031f2368d", + "type": "table", + "content": "
($ in Millions)Q2 FY27 Outlook
Non-GAAP gross margin75.0%
Impact of acquisition-related costs and other costs(0.1%)
GAAP gross margin74.9%
Non-GAAP operating expenses$8,300
Acquisition-related costs and other costs200
GAAP operating expenses$8,500
", + "path": "tables/table-14 Q2 FY27 Outlook.html", + "metadata": { + "length": 472, + "summary": "table-15\nThe table shows Q2 FY27 outlook: Non-GAAP gross margin 75.0%, GAAP 74.9%; Non-GAAP operating expenses $8,300M, GAAP $8,500M.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-14 Q2 FY27 Outlook.html", + "keywords": [ + "Gross margin", + "Operating expenses", + "GAAP" + ], + "tokens": [] + } + }, + { + "chunk_id": "504dec07-cb4d-542f-82b8-c72b1f65bf7b", + "type": "text", + "content": "[tables/table-14 Q2 FY27 Outlook.html]", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Reconciliation of Non-GAAP to GAAP Financial Measures (contd.)", + "metadata": { + "length": 54, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "tokens": [], + "keywords": [], + "connect_to": [ + { + "target": "e757adf1-b61f-5447-adf3-b3f031f2368d", + "relation": "embeds", + "ref": "[tables/table-14 Q2 FY27 Outlook.html]", + "position": { + "start": 0, + "end": 38 + } + } + ] + } + }, + { + "chunk_id": "ec646ce0-f94f-5fc4-baf0-5b203fd702e7", + "type": "table", + "content": "
($ in Millions)Free Cash FlowPurchases Related to Property and Equipment and Intangible AssetsPrincipal Payments on Property and Equipment and Intangible AssetsNet Cash Provided by Operating Activities
Q1 FY2027$48,554(1,757)(33)$50,344
Q1 FY 2026$26,135(1,227)(52)$27,414
", + "path": "tables/table-15 Free Cash Flow.html", + "metadata": { + "length": 447, + "summary": "table-16\nThe table shows Q1 FY2027 Free Cash Flow of $48,554M from $50,344M operating activities, minus $1,757M property purchases, up from $26,135M in Q1 FY2026.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "tables/table-15 Free Cash Flow.html", + "keywords": [ + "Free Cash Flow", + "Operating Activities", + "Property Equipment" + ], + "tokens": [] + } + }, + { + "chunk_id": "7fe4a603-f0e8-57b3-95fc-8d2905559763", + "type": "image", + "content": "\nB. Other consists of legal settlements, net loss from equity securities, and interest expense related to amortization of debt discount.
0\n[images/image-11-Abstract Green Wave Design.jpg]\n", + "path": "images/image-11-Abstract Green Wave Design.jpg", + "metadata": { + "length": 181, + "summary": "image-11\n- A 3D rendered abstract image featuring layered, glowing lime-green curved bands.\n- The bands create a tunnel or wave effect receding into the distance on the right side.\n- The background is black with scattered white dots resembling stars in space.\n- No text or specific data is visible in the image.", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "file_path": "images/image-11-Abstract Green Wave Design.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d57169f5-e0e7-5623-bc2d-0bc9c03535a9", + "type": "text", + "content": "[tables/table-15 Free Cash Flow.html]\n\n\nB. Other consists of legal settlements, net loss from equity securities, and interest expense related to amortization of debt discount.
0\n[images/image-11-Abstract Green Wave Design.jpg]\n\n
\nnatural_image\nAbstract green curved bands against a black starry background (no text or symbols)\n
", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Reconciliation of Non-GAAP to GAAP Financial Measures (contd.)_2", + "metadata": { + "length": 423, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Investor Presentation Q1 FY27", + "tokens": [ + "Other", + "consists", + "legal", + "settlements", + "net", + "loss", + "equity", + "securities", + "interest", + "expense", + "related", + "amortization", + "debt", + "discount", + "details", + "summary", + "natural", + "image", + "Abstract", + "green", + "curved", + "bands", + "black", + "starry", + "background", + "text", + "symbols" + ], + "keywords": [], + "connect_to": [ + { + "target": "ec646ce0-f94f-5fc4-baf0-5b203fd702e7", + "relation": "embeds", + "ref": "[tables/table-15 Free Cash Flow.html]", + "position": { + "start": 0, + "end": 37 + } + }, + { + "target": "7fe4a603-f0e8-57b3-95fc-8d2905559763", + "relation": "embeds", + "ref": "[images/image-11-Abstract Green Wave Design.jpg]", + "position": { + "start": 187, + "end": 235 + } + } + ] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/doc_nav.json b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/doc_nav.json new file mode 100644 index 000000000..e89089aa8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/doc_nav.json @@ -0,0 +1,224 @@ +{ + "version": "1.0", + "file_name": "NVIDIA Q1 FY27 Quarterly Presentation.pdf", + "stats": { + "total_chunks": 39, + "text_chunks": 12, + "image_chunks": 11, + "table_chunks": 16, + "max_depth": 2 + }, + "sections": [ + { + "title": "Investor Presentation Q1 FY27", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27", + "level": 1, + "summary": "This presentation from May 2026 contains forward-looking statements regarding NVIDIA's growth, products like Blackwell and Vera Rubin, financial outlook, and AI industry trends. These statements are subject to risks including global economic conditions, reliance on third-party manufacturing, technological competition, and market acceptance. The document also notes that many products and features are offered on a when-and-if-available basis, with no commitment to delivery. Additionally, NVIDIA uses non-GAAP financial measures such as non-GAAP gross margin and free cash flow, which are not substitutes for GAAP measures, and reconciliations are provided.", + "chunk_count": 12, + "children": [ + { + "title": "Revenue \\$ in billions", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Revenue \\$ in billions", + "level": 2, + "summary": "\\$ in billions Strong growth across a diverse end-customer base – hyperscalers to model builders to AI Clouds, enterprise on-premise and sovereign customers Blackwell adopted and deployed by every major hyperscaler, every cloud provider, and every model builder Agentic AI and reinforcement learning represent new CPU growth opportunities Vera Rubin on track for the second half of this year, starting in Q3 NVIDIA AI infrastructure delivers the lowest token cost and highest token throughput as shown in latest MLPerf and InferenceX benchmarks Strong growth across a diverse end-customer base – hyperscalers to model builders to AI Clouds, enterprise on-premise and sovereign customers
0 [images/image-1-Financial Growth Comparison.jpg]
bar chart [tables/table-0 Revenue Forecast.html]
Data Center Revenue \\$ in billions Hyperscale
0 [images/image-2-Q1 FY26 vs Q1 FY27 Revenue.jpg]
bar chart [tables/table-1 Revenue Forecast.html]
AI Clouds, Industrial, and Enterprise (ACIE)
0 [images/image-3-Financial Growth Comparison.jpg]
bar chart [tables/table-2 Revenue Forecast.html]
Continued strong demand for our Blackwell architecture", + "chunk_count": 1, + "children": [] + }, + { + "title": "Gross Margin", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Gross Margin", + "level": 2, + "summary": "GAAP
0 [images/image-4-Q1 FY26 vs Q1 FY27.jpg]
bar chart [tables/table-3 Quarterly Values.html]
Non-GAAP
0 [images/image-5-Q1 FY26 vs Q1 FY27.jpg]
bar chart [tables/table-4 Quarterly Values.html]
Gross margin year-on-year increase driven by lower inventory provisions, primarily due to a \\$4.5B charge associated with H2O excess inventory in the prior year", + "chunk_count": 1, + "children": [] + }, + { + "title": "Operating Expenses", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Operating Expenses", + "level": 2, + "summary": "\\$ in billions
0 [images/image-6-GAAP Revenue Growth.jpg]
bar chart GAAP [tables/table-5 Revenue Forecast.html]
0 [images/image-7-Non-GAAP Revenue Growth.jpg]
bar chart Non-GAAP [tables/table-6 Revenue Forecast.html]
Investing to address the multitude of growth opportunities ahead Earnings per Share
Top-line growth and operating leverage drove accelerating EPS growth0 [images/image-8-GAAP vs Non-GAAP Comparison.jpg] Top-line growth and operating leverage drove accelerating EPS growth", + "chunk_count": 1, + "children": [] + }, + { + "title": "Free Cash Flow and Capital Return", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Free Cash Flow and Capital Return", + "level": 2, + "summary": "\\$ in billions Free Cash Flow Top-line growth and operating leverage drove accelerating EPS growth
0 [images/image-9-Financial Growth Comparison.jpg]
bar chart [tables/table-7 Revenue Forecast.html]
Capital Return
0 [images/image-10-Dividends vs Share Repurchases.jpg]
bar chart [tables/table-8 Dividends & Buybacks.html]
Returned a record \\~\\$20B to our shareholders, while executing strategic investments. Announced new \\$80B share repurchase authorization, on top of \\~\\$39B remaining as of Q1 FY27. Increased quarterly dividend from \\$0.01 to \\$0.25 per share.", + "chunk_count": 1, + "children": [] + }, + { + "title": "GAAP P&L Financial Measures \\$ in billions, except EPS", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/GAAP P&L Financial Measures \\$ in billions, except EPS", + "level": 2, + "summary": "\\$ in billions, except EPS [tables/table-9 Q1 FY27 Results.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-GAAP P&L Financial Measures \\$ in billions, except EPS", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Non-GAAP P&L Financial Measures \\$ in billions, except EPS", + "level": 2, + "summary": "\\$ in billions, except EPS [tables/table-10 Q1 FY27 Results.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Q2 FY27 Outlook – Non-GAAP", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Q2 FY27 Outlook – Non-GAAP", + "level": 2, + "summary": "[tables/table-11 Q2 FY27 Guidance.html] For the full year fiscal 2027, expect non-GAAP tax rates to be between 16.0% and 18.0%, excluding any discrete items and material changes to our tax environment", + "chunk_count": 1, + "children": [] + }, + { + "title": "Reconciliation of Non-GAAP to GAAP Financial Measures", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Reconciliation of Non-GAAP to GAAP Financial Measures", + "level": 2, + "summary": "Reconciliation of Non-GAAP to GAAP Financial Measures [tables/table-12 Q1 FY27 GAAP.html] Beginning in Q1 FY27, NVIDIA's non-GAAP financial measures no longer exclude stock-based compensation expense. The historical non-GAAP financial information presented has been updated to include stock-based compensation expense. A. Consists of amortization of intangible assets, transaction costs, and certain compensation charges. B. Other consists of net gain from equity securities, interest expense related to acquisition consideration discount to be paid in the future and amortization of debt discount, legal settlements, dividend income on equity securities, and share of net earnings related to equity method investments.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Reconciliation of Non-GAAP to GAAP Financial Measures_2", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Reconciliation of Non-GAAP to GAAP Financial Measures_2", + "level": 2, + "summary": "[tables/table-13 Q1 FY26 Financials.html] Beginning in Q1 FY27, NVIDIA's non-GAAP financial measures no longer exclude stock-based compensation expense. The historical non-GAAP financial information presented has been updated to include stock-based compensation expense. A. Consists of amortization of intangible assets, transaction costs, and certain compensation charges. B. Other consists of legal settlements, net loss from equity securities, and interest expense related to amortization of debt discount.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Reconciliation of Non-GAAP to GAAP Financial Measures (contd.)", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Reconciliation of Non-GAAP to GAAP Financial Measures (contd.)", + "level": 2, + "summary": "[tables/table-14 Q2 FY27 Outlook.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Reconciliation of Non-GAAP to GAAP Financial Measures (contd.)_2", + "path": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf/Investor Presentation Q1 FY27/Reconciliation of Non-GAAP to GAAP Financial Measures (contd.)_2", + "level": 2, + "summary": "[tables/table-15 Free Cash Flow.html] B. Other consists of legal settlements, net loss from equity securities, and interest expense related to amortization of debt discount.
0 [images/image-11-Abstract Green Wave Design.jpg]
natural_image Abstract green curved bands against a black starry background (no text or symbols)
", + "chunk_count": 1, + "children": [] + } + ] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-Financial Growth Comparison.jpg", + "summary": "image-1 Q1 FY26: $44.1B Q1 FY27: $81.6B" + }, + { + "path": "images/image-2-Q1 FY26 vs Q1 FY27 Revenue.jpg", + "summary": "image-2 - Chart Type: Bar Chart comparing financial data across two periods. - Data Points: - Q1 FY26: $17.6B (represented by a grey bar) - Q1 FY27: $37.9B (represented by a green bar) - Trend: Significant increase in value from Q1 FY26 to Q1 FY27, more than doubling the previous period's figure." + }, + { + "path": "images/image-3-Financial Growth Comparison.jpg", + "summary": "image-3 Q1 FY26: $21.5B Q1 FY27: $37.4B" + }, + { + "path": "images/image-4-Q1 FY26 vs Q1 FY27.jpg", + "summary": "image-4 - Chart Type: Bar Chart - Data Points: - Q1 FY26: 60.5% - Q1 FY27: 74.9% - Trend: Significant increase from the first quarter of fiscal year 26 to the first quarter of fiscal year 27." + }, + { + "path": "images/image-5-Q1 FY26 vs Q1 FY27.jpg", + "summary": "image-5 - Chart Type: Bar Chart - Data Points: - Q1 FY26: 60.8% - Q1 FY27: 75.0% - Trend: Significant increase from the first period to the second." + }, + { + "path": "images/image-6-GAAP Revenue Growth.jpg", + "summary": "image-6 Chart Title: GAAP Data Points: - Q1 FY26: $5.0B - Q1 FY27: $7.6B" + }, + { + "path": "images/image-7-Non-GAAP Revenue Growth.jpg", + "summary": "image-7 Chart Title: Non-GAAP Data Points: - Q1 FY26: $5.0B - Q1 FY27: $7.4B" + }, + { + "path": "images/image-8-GAAP vs Non-GAAP Comparison.jpg", + "summary": "image-8 - GAAP: Q1 FY26 ($0.76), Q1 FY27 ($2.39) - Non-GAAP: Q1 FY26 ($0.78), Q1 FY27 ($1.87)" + }, + { + "path": "images/image-9-Financial Growth Comparison.jpg", + "summary": "image-9 Q1 FY26: $26.1B Q1 FY27: $48.6B" + }, + { + "path": "images/image-10-Dividends vs Share Repurchases.jpg", + "summary": "image-10 Chart Title: Dividends and Share Repurchases Comparison Data Points: - Q1 FY26: ~$14.0B (represented by a grey bar) - Q1 FY27: ~$20.0B (represented by a bright green bar) Legend: - Dark Green Square: Dividends - Bright Green Square: Share Repurchases Observations: - The chart compares financial figures for two quarters: Q1 FY26 and Q1 FY27. - There is an increase in value from approximately $14.0 billion to $20.0 billion between the two periods." + }, + { + "path": "images/image-11-Abstract Green Wave Design.jpg", + "summary": "image-11 - A 3D rendered abstract image featuring layered, glowing lime-green curved bands. - The bands create a tunnel or wave effect receding into the distance on the right side. - The background is black with scattered white dots resembling stars in space. - No text or specific data is visible in the image." + } + ], + "tables": [ + { + "path": "tables/table-0 Revenue Forecast.html", + "summary": "table-1 The table shows projected revenue values for Q1 FY26 at $44.1B and Q1 FY27 at $81.6B, indicating significant growth." + }, + { + "path": "tables/table-1 Revenue Forecast.html", + "summary": "table-2 The table shows projected revenue values for Q1 FY26 at $17.6B and Q1 FY27 at $37.9B." + }, + { + "path": "tables/table-2 Revenue Forecast.html", + "summary": "table-3 The table shows projected revenue values for Q1 FY26 at $21.5B and Q1 FY27 at $37.4B, indicating significant growth." + }, + { + "path": "tables/table-3 Quarterly Values.html", + "summary": "table-4 The table shows two quarterly periods with their corresponding percentage values: Q1 FY26 at 60.5% and Q1 FY27 at 74.9%." + }, + { + "path": "tables/table-4 Quarterly Values.html", + "summary": "table-5 The table shows two quarterly values: 60.8% for Q1 FY26 and 75.0% for Q1 FY27, indicating an increase." + }, + { + "path": "tables/table-5 Revenue Forecast.html", + "summary": "table-6 The table shows projected revenue values of $5.0B for Q1 FY26 and $7.6B for Q1 FY27, indicating growth." + }, + { + "path": "tables/table-6 Revenue Forecast.html", + "summary": "table-7 The table shows projected revenue values for Q1 FY26 at $5.0B and Q1 FY27 at $7.4B, indicating growth over the two periods." + }, + { + "path": "tables/table-7 Revenue Forecast.html", + "summary": "table-8 The table shows projected revenue values for Q1 FY26 at $26.1B and Q1 FY27 at $48.6B, indicating significant growth." + }, + { + "path": "tables/table-8 Dividends & Buybacks.html", + "summary": "table-9 The table shows Q1 FY26 dividends of ~$14.0B and Q1 FY27 share repurchases of ~$20.0B." + }, + { + "path": "tables/table-9 Q1 FY27 Results.html", + "summary": "table-10 Revenue grew 85% YoY to $81.6M, gross margin improved to 74.9%, and net income surged 211% to $58.3M in Q1 FY27." + }, + { + "path": "tables/table-10 Q1 FY27 Results.html", + "summary": "table-11 Revenue grew 85% YoY to $81.6M, gross margin improved to 75.0%, and net income rose 139% to $45.5M, with diluted EPS at $1.87." + }, + { + "path": "tables/table-11 Q2 FY27 Guidance.html", + "summary": "table-12 Q2 FY27 revenue guidance is $91.0 billion, plus or minus 2%. Gross margin is 75.0%, plus or minus 50 bps. Operating expenses are approximately $8.3 billion." + }, + { + "path": "tables/table-12 Q1 FY27 GAAP.html", + "summary": "table-13 The table shows Q1 FY27 financials: Non-GAAP gross margin 75.0% ($61,232M), GAAP 74.9% ($61,157M); Net income GAAP $58,321M; Diluted EPS GAAP $2.39." + }, + { + "path": "tables/table-13 Q1 FY26 Financials.html", + "summary": "table-14 The table shows Q1 FY26 financials with Non-GAAP gross margin of 60.8% and GAAP of 60.5%. Net income is $19,094M Non-GAAP vs $18,775M GAAP, with diluted EPS of $0.78 and $0.76 respectively." + }, + { + "path": "tables/table-14 Q2 FY27 Outlook.html", + "summary": "table-15 The table shows Q2 FY27 outlook: Non-GAAP gross margin 75.0%, GAAP 74.9%; Non-GAAP operating expenses $8,300M, GAAP $8,500M." + }, + { + "path": "tables/table-15 Free Cash Flow.html", + "summary": "table-16 The table shows Q1 FY2027 Free Cash Flow of $48,554M from $50,344M operating activities, minus $1,757M property purchases, up from $26,135M in Q1 FY2026." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/full.md b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/full.md new file mode 100644 index 000000000..c8e82c5bf --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/full.md @@ -0,0 +1,210 @@ +## Investor Presentation Q1 FY27 + +May 2026 + +Certain matters in this presentation including, but not limited to, statements as to: expectations with respect to growth, performance and benefits of NVIDIA's products, services, and technologies, including Blackwell and Vera Rubin, and related trends and drivers; expectations with respect to supply and demand for NVIDIA's products, services, and technologies, including Blackwell and Vera Rubin, and related matters including inventory, production and distribution; NVIDIA's financial position and financial and business outlook; NVIDIA's future cash dividends and other returns to shareholders; expectations with respect to technology developments and related trends and drivers; projected market growth and trends; expectations with respect to AI and related industries; and other statements that are not historical facts are forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the “safe harbor” created by those sections based on management’s beliefs and assumptions and on information currently available to management and are subject to risks and uncertainties that could cause results to be materially different than expectations. + +Important factors that could cause actual results to differ materially include: global economic and political conditions; NVIDIA's reliance on third parties to manufacture, assemble, package and test NVIDIA's products; the impact of technological development and competition; development of new products and technologies or enhancements to NVIDIA's existing product and technologies; market acceptance of NVIDIA's products or NVIDIA's partners' products; design, manufacturing or software defects; changes in consumer preferences or demands; changes in industry standards and interfaces; unexpected loss of performance of NVIDIA's products or technologies when integrated into systems; NVIDIA's ability to realize the potential benefits of business investments or acquisitions; and changes in applicable laws and regulations, as well as other factors detailed from time to time in the most recent reports NVIDIA files with the Securities and Exchange Commission, or SEC, including, but not limited to, its annual report on Form 10-K and quarterly reports on Form 10-Q. Copies of reports filed with the SEC are posted on the company's website and are available from NVIDIA without charge. These forward-looking statements are not guarantees of future performance and speak only as of the date hereof, and, except as required by law, NVIDIA disclaims any obligation to update these forward-looking statements to reflect future events or circumstances. + +Many of the products and features described herein remain in various stages and will be offered on a when-and-if-available basis. The statements within are not intended to be, and should not be interpreted as a commitment, promise, or legal obligation, and the development, release, and timing of any features or functionalities described for our products is subject to change and remains at the sole discretion of NVIDIA. NVIDIA will have no liability for failure to deliver or delay in the delivery of any of the products, features or functions set forth herein. + +NVIDIA uses certain non-GAAP measures in this presentation including non-GAAP gross margin, non-GAAP operating expense, non-GAAP operating income, non-GAAP net income, non-GAAP diluted earnings per share, and free cash flow. NVIDIA believes the presentation of its non-GAAP financial measures enhances investors' overall understanding of the company's historical financial performance. The presentation of the company's non-GAAP financial measures is not meant to be considered in isolation or as a substitute for the company's financial results prepared in accordance with GAAP, and the company's non-GAAP measures may be different from non-GAAP measures used by other companies. Further information relevant to the interpretation of non-GAAP financial measures, and reconciliations of these non-GAAP financial measures to the most comparable GAAP measures, may be found in the slide titled “Reconciliation of Non-GAAP to GAAP Financial Measures”. + +## Revenue + +## \$ in billions + +Strong growth across a diverse end-customer base – hyperscalers to model builders to AI Clouds, enterprise on-premise and sovereign customers + +Blackwell adopted and deployed by every major hyperscaler, every cloud provider, and every model builder + +Agentic AI and reinforcement learning represent new CPU growth opportunities + +Vera Rubin on track for the second half of this year, starting in Q3 + +NVIDIA AI infrastructure delivers the lowest token cost and highest token throughput as shown in latest MLPerf and InferenceX benchmarks + +![](images/dacf7cad35006a22acf99d5d134c0e1536f218dc989c12fb80887c435288130e.jpg) + +
+bar chart + +| Period | Value | +| --------- | ------- | +| Q1 FY26 | $44.1B | +| Q1 FY27 | $81.6B | +
+ +## Data Center + +## Revenue \$ in billions + +Hyperscale +![](images/0a7f3f0ba3e758995f39fca895349df9774792d61c5acd52659758a1eeb66f7a.jpg) + +
+bar chart + +| Period | Value | +| ---------- | ------- | +| Q1 FY26 | $17.6B | +| Q1 FY27 | $37.9B | +
+ +AI Clouds, Industrial, and Enterprise (ACIE) +![](images/e640ed2465ce87d093155f99ebb71b7f625cc5652653f44969e7275e4488aa47.jpg) + +
+bar chart + +| Period | Value | +| --------- | ------- | +| Q1 FY26 | $21.5B | +| Q1 FY27 | $37.4B | +
+ +Continued strong demand for our Blackwell architecture + +## Gross Margin + +GAAP +![](images/a3a674ac36c2bb3eaa605b9f0a7e4b42f92f9d4eaac14398f76e65e045b7bbb7.jpg) + +
+bar chart + +| Period | Value (%) | +| :--- | :--- | +| Q1 FY26 | 60.5 | +| Q1 FY27 | 74.9 | +
+ +Non-GAAP +![](images/4531beabe284ec3676f5c81f81016d4df947fc86de961e2a3d839a7146067791.jpg) + +
+bar chart + +| Period | Value (%) | +| :--- | :--- | +| Q1 FY26 | 60.8 | +| Q1 FY27 | 75.0 | +
+ +Gross margin year-on-year increase driven by lower inventory provisions, primarily due to a \$4.5B charge associated with H2O excess inventory in the prior year + +## Operating Expenses + +\$ in billions + +![](images/167a673bb7a14d45312d36087b16d2d192e8358c0e80642c3c3a065b426f33c9.jpg) + +
+bar chart + +GAAP +| Period | Value ($B) | +| :--- | :--- | +| Q1 FY26 | 5.0 | +| Q1 FY27 | 7.6 | +
+ +![](images/080e5ee2d2de0dacb9d365aa0bfa136bcde0527134588a3ba187bb5af05bfe38.jpg) + +
+bar chart + +Non-GAAP +| Period | Value ($B) | +| :--- | :--- | +| Q1 FY26 | 5.0 | +| Q1 FY27 | 7.4 | +
+ +Investing to address the multitude of growth opportunities ahead + +Earnings per Share +![](images/31cb23d10061dd6eb9f783eb694a61e463723f84db596a00bc6dfd4cae85fbb3.jpg) +Top-line growth and operating leverage drove accelerating EPS growth + +## Free Cash Flow and Capital Return + +\$ in billions + +Free Cash Flow +![](images/e0a4b245f69891ffeed620979d8f0c1068132aafd201cdf6285875c7a5cced91.jpg) + +
+bar chart + +| Period | Value | +| ---------- | ------- | +| Q1 FY26 | $26.1B | +| Q1 FY27 | $48.6B | +
+ +Capital Return +![](images/a200c681baf944209185508397cf72ffd7709c09fdc6c26ba472805795a32770.jpg) + +
+bar chart + +| Period | Dividends | Share Repurchases | +| ---------- | --------- | ----------------- | +| Q1 FY26 | ~$14.0B | - | +| Q1 FY27 | - | ~$20.0B | +
+ +Returned a record \~\$20B to our shareholders, while executing strategic investments. Announced new \$80B share repurchase authorization, on top of \~\$39B remaining as of Q1 FY27. Increased quarterly dividend from \$0.01 to \$0.25 per share. + +## GAAP P&L Financial Measures + +## \$ in billions, except EPS + +
Q1 FY27YoY %Q1 FY26
Revenue$81.685%$44.1
Gross Margin %74.9%+14.4 pts60.5%
Operating Expenses$7.652%$5.0
Operating Income$53.5147%$21.6
Net Income$58.3211%$18.8
Diluted EPS$2.39214%$0.76
+ +## Non-GAAP P&L Financial Measures + +## \$ in billions, except EPS + +
Q1 FY27YoY %Q1 FY26
Revenue$81.685%$44.1
Gross Margin %75.0%+14.2 pts60.8%
Operating Expenses$7.449%$5.0
Operating Income$53.8147%$21.8
Net Income$45.5139%$19.1
Diluted EPS$1.87140%$0.78
+ +## Q2 FY27 Outlook – Non-GAAP + +
Q2 FY27
Revenue$91.0 billion*+ / - 2%
Gross Margin %75.0%+ / - 50 bps
Operating Expenses$8.3 billionApproximately
+ +For the full year fiscal 2027, expect non-GAAP tax rates to be between 16.0% and 18.0%, excluding any discrete items and material changes to our tax environment + +## Reconciliation of Non-GAAP to GAAP Financial Measures + +## Reconciliation of Non-GAAP to GAAP Financial Measures + +
Q1 FY27Non-GAAPAcquisition-Related and Other Costs (A)Other (B)Tax Impact of AdjustmentsGAAP
Gross margin ($ in million)$61,232(47)(28)$61,157
Gross margin %75.0%(0.1)74.9%
Operating expense ($ in million)$7,449172$7,621
Operating income ($ in million)$53,783(219)(28)$53,536
Net income ($ in million)$45,548(219)15,882(2,890)$58,321
Diluted shares (millions)24,39124,391
Diluted EPS$1.87(0.01)0.65(0.12)$2.39
+ +Beginning in Q1 FY27, NVIDIA's non-GAAP financial measures no longer exclude stock-based compensation expense. The historical non-GAAP financial information presented has been updated to include stock-based compensation expense. +A. Consists of amortization of intangible assets, transaction costs, and certain compensation charges. +B. Other consists of net gain from equity securities, interest expense related to acquisition consideration discount to be paid in the future and amortization of debt discount, legal settlements, dividend income on equity securities, and share of net earnings related to equity method investments. + +## Reconciliation of Non-GAAP to GAAP Financial Measures + +
Q1 FY26Non-GAAPAcquisition-Related and Other Costs (A)Other (B)Tax Impact of AdjustmentsGAAP
Gross margin ($ in million)$26,794(123)(3)$26,668
Gross margin %60.8%(0.3)60.5%
Operating expense ($ in million)$4,99337$5,030
Operating income ($ in million)$21,801(160)(3)$21,638
Net income ($ in million)$19,094(160)(179)20$18,775
Diluted shares (millions)24,61124,611
Diluted EPS$0.78(0.01)(0.01)$0.76
+ +Beginning in Q1 FY27, NVIDIA's non-GAAP financial measures no longer exclude stock-based compensation expense. The historical non-GAAP financial information presented has been updated to include stock-based compensation expense. +A. Consists of amortization of intangible assets, transaction costs, and certain compensation charges. +B. Other consists of legal settlements, net loss from equity securities, and interest expense related to amortization of debt discount. + +## Reconciliation of Non-GAAP to GAAP Financial Measures (contd.) + +
($ in Millions)Q2 FY27 Outlook
Non-GAAP gross margin75.0%
Impact of acquisition-related costs and other costs(0.1%)
GAAP gross margin74.9%
Non-GAAP operating expenses$8,300
Acquisition-related costs and other costs200
GAAP operating expenses$8,500
+ +## Reconciliation of Non-GAAP to GAAP Financial Measures (contd.) + +
($ in Millions)Free Cash FlowPurchases Related to Property and Equipment and Intangible AssetsPrincipal Payments on Property and Equipment and Intangible AssetsNet Cash Provided by Operating Activities
Q1 FY2027$48,554(1,757)(33)$50,344
Q1 FY 2026$26,135(1,227)(52)$27,414
+ +![](images/51fcb213adfa05e5cb2dd45e006c8756c828d795a5224654aad00aa396a4ff0f.jpg) + +
+natural_image + +Abstract green curved bands against a black starry background (no text or symbols) +
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-1-Financial Growth Comparison.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-1-Financial Growth Comparison.jpg new file mode 100644 index 0000000000000000000000000000000000000000..3ef10b8f90d83dfbc93f698a5a70cc3bdc3cdae4 GIT binary patch literal 41505 zcmeHw2Ut^C*Y1gw08*r>NC^rGf{0S22n29wG8j}KQUn`aq)R6h0i}ZiiU>%fHMFMH8qr)iUtaW($dn<(X%no@7_(%!NSVK#(eH!#7PAbSxFW^64kX;m%RMb!!TDslf359!rT@VV2U6d45RFst9 z*=&m+r`z*-TzKNV9?!r!O{00 z#Kguud=#IO`s{gHdPZi}tNen(qT<&jrL}eS4UJ82-Zpo3b@%l4efZcvh8dsuH2HaI zdIq<&yt2Bsj^Eh)QZEQV@gLO!-~Xdx-_(m4tk*6|N(xHomwG{Vxqt@+GbPo*lhiCq z=b=~aSr3W%(y%E!G8k(%X; z4}n<3#$RviQDJR4-6gc|POt@=9)iABX zV}bdmV=W>k+KCM}^3FeRt>|L54~?Z#nqJRtAo6)imtz^9GA%X?KXSfr<3ue^Y+-EL z=pAYJFwDKL`h9a~^zrbl<~m9q=|is>)anfyb4)EwQ5l0q1|O5dhxegkPsm>Eq3UBQJF_U2`eaKn`g_%#H1#UsV+5Ol!IIqOVo6Q9VpYkH*>QNI1RAg}opN+_V% z{n-kTfN~Qo>=L#pMvTFLgQGlpK+~_LY19rWIq}MswIS~^r#ZF_J?S61TZRU%Oii<% z%s7C7)@C#C)#c%{Yu%%|oz(bT&FJ6z+0k5~R(0hK4PJ_0Uw`=;%QUwPwhXT@!`gE` z8WDY8Q<36`3?Od>4J0#JVG(tUfr&@*x(C$FO=ag9ON zV^s%c;^pr*uk}XHg+1rZ+YgNfxVMF$;t!L6!zSvRnj|2lKZ9_I1VpD}m^L>^z(+kF z>1M&_6#S6x4rn?-9wnPJpERfHVn5y55L%;dVuyPbvoJ^2M=xNBP)03 zwy4a(>At89ms6b*7erGj{B^Bg;2&og%p|c$4wOmh^IC^x&Au=4c@|NBU}6eL2^J(D zzt9G`1GOXU1pP5i-Tar88ZYiI7+w~57_uU_N8Y!=A^y7lkxZ-p2M!w74x7jyvUzb) zZgdk3J`nhx{ zfb$;z?G`DD4sFUxAiuMciy&!xS2iH8tb5^Jc&GdaI$9xMrHxjg^X0^Lis!YyBR>>% zUXa7>i=4PVTP4|rztG!-aj9X$<`hn7TWwx9y>Mo#K2v@zvfexoefs`ram0AZi7?74 zX~yD{HVPXzcpjez8Wco5Y;~CgyvhOWAK+idz=zpUb(Y z$UO^yjsQuW1#*_7TZj9vOFh@060!RN%rr`$fa%{4T`vh)*II1wuwxTST<72Q$sh`CsiK+7Q7D* zHjJZvH%&s1Ur}E@ZOP6Zd0oaPRN>C(c%jOcbn{gc)@I=F!FxUm12gbomw0o3Z!XI! z#<^RulQ@0TGE*WK_eX_0d2e|9lD1?TJDV_fyB6h63T{4Gf#_CQa=TrMn3I?Ck0>Vr z;|BdqP3m$;&qN;Fk>JRFD}KXAImfx}41H*3{S%Y>4v+wdoYUpIg|BMh%PWNyV=$l1 z`Pd4}OuXMrb6aDZ?vq;0ipQ31vR4>o_&cCb-07M;JgiaNvZzxc)`2&1u%v}jX1~H~ z-r#ZT4x%_dt*OCS&aL3%UTaNty^lwXqxWd=d)yeuqUFAmHQ=vV>#o2dcyqPh0+|Kp z-Edjm5{o*w+}GM!_C9J3)|n*W4zv!wLQewTETM3^6Mb!h=QF0?o<<{<#hkA_iC|5s z$S3ZBf!}wClK{TAB%t9u37}yLL``v!fT#%)fXNn}pOac@SumLtwPmfJLf`!QnmVXj zim{|kYSe@q9G?X*emT-2T7r7$;hHF*;$cqURRFJ(Y)(|WPXdBdzrEFVMIce%tA$WE zP>+)p9)OQQuyN_WIOtWspgJ@EchrKkA7C7Iaz9A2J(mpvpa%nT#y{~`-eAR?s-Dco}?WefGs(13`2+miPWnH(TXBua$t?z8T zsynD^@|3GjuWGO7?I^eF^~I0vLnEINZFu-fVueL;bq*o0>X`D6O*clB&|^nWahV$M+a-#G4;q z@Y0ZrU$gzyMH@d=>~c!#{E+^jrma2m5ZB8RwKFp`6FQ}VnHg5=aLaCVF8bD~faP0e zmYfvHEiCrNb-!g$c%17IT?uI@>#{X3cx@bAb5M2j#_il#RB)oq>*5m{7{k#!R~}pm zBLQ9j)9mHX-wc?4HIDU7qb&{eFF%WFo4hS^%u6G?)T}IfD(%d3tKwd2Be%fbG0aH# zqkdKEKGat|{XY%&N8s>q8E4u&u-Z=8-wIApU11WyVqN;eST8llN-By3IAEqHVE0*# z^M6YN{I*%)ge>}co_|&JpAFJqVNf!N!*!@*m7j=L=P|BN*yoeJs`)nqWNnRe!*pGA zndvWUd$Q|ii=h$eA(LyC*@h_iBfI$68WlFA{Z$n6AApvnoqTN$Zy9HB%H_s%4@4)wP ztfB)*z|yfn63{n}nxeg2Xb=CkXYfWX+Y9NL4Hq$>twz%Bh_N=2>g!}*_VN;7LbbHV z%{Xb#u$@=&I@kfe1K-2p-d9TkHul~^&El$7D0BN?CwleE*nYa_6#lEv6 z2Kx&z(18eXKd9={=ayAjL1z$yDkGhO`bw* z6;b7J(k8)n8zNkkEAUm9nCyOps_0&o&hq>bG|pqZJ7J-}z=84?To_y2a9i3w7tF03 zEQo$4yCeAX7j)4v8GN%YW&Ns0&;jqNhG*OpbQO+(hS@H1C#>@qBslm5671Fqm;Wv57-i^?PToB2q`3$+rHJf06#q&>Bi&#CpA zN)JtCs_wDc@t}+Ed=ry>u4% z@($qnkfk9P1JXL3DVYZ;IfI!KPMr={=NtApy%+>z^F)lpc!th+IYoU4Ir==jbAPuG zbEsojeT1XF!e0D`zVpTw!|)%vok~)!9NxpQG4~t^e9jy7A1*^t;LXQI2!~m%d6;p# z%DW8J1FM@E8*bjXbA=1C7{6S_On~$iY#p^-@lx))6rkh~l^L!K|H#2hbW3Iusyr2z zVI=9-TokB!;>GJ>Rym4re5OQ&*s?*7NWmu7nr8NczBQDN@oN=~ zy(6RReu%rPf>8wBti}tqADd21q3(tc-6KGA0C!|vi#%&hTwTrHW5owv^*?Ss`w3MQ z13?)>udbqFBuzHxjltNM`lL-}^yTfOg#}UKA**#4hk`y<{nKyx5@N5-Xh=>Npb3GZ zs96kRo2G;6?0Q6CDOFJ2wpd+*!od*P%4{)1%h<5GLnT1vX)dlUCP$;B+C}P*Ln2L~ za+-I`<)0HsfHVf9^j>iR($qcb-UCLiatynUVA5hT2F*@8R|cLo^u)1wU#Z?qeAYo- zJP!p^F^-}7LQPt;o3ndR7VwRu*NNrVh&IN$id63HwdlJ+Ch43+kD5rsyAZ1`_(&4O{MGjMjy{di*?6SVCN|r}vu%C12)}R4QI^)5x4-$$Qq-}t zO;etAgp6IELvLa5LD)ku_L2U&y*Z{wM#N5rj;)>GY2qF4twf=pdY>@3jK4V66Bp>j zeMA?F6*8|HhS|3%N>7-U9J4!3w=uKh$)4Z!wy*8fFnwR)HcY#xV0K@8bB(6f5igm8 zY8UL=xwChbA%alM&P;%ng9g3wOzrrt$)3lYdG8;rT~*|S?*fDUyk3)lmo`mHY>R%f zm6?_}q0ZV)ORXpGBaNzQ1$B;|zB?VMI9YL~x}OBpA@<-?tmj!Zk`TSm3S-hPa;O|_ zB(eva4+7Q??qh;{2&|TlFdu@#d`9zgt)bdTXm{RmRkwlaWcng7#Q~|Y{ZvxFaGEdZ z$D${bKR&&&{E~)J5MHlA5YE8KkpL*cw(F}+jSp11h!ckdr`X(#=r%Ujmd3R! z6PIT6Vli&?(Wjlw1o*qzAgfY!45DKd5qFQ@zdn2v67Dtr5?5@2U`qlDw8WNC|TZK8Ee0`p$uO_!*^m7VYc;(7GbfVbravQFj z3!pJt*QffKYA&soL0{t7SK>nKIeck7#TPi$Hmi(~zCGKFYw}XYUFb=qMNaPZN43DE zfGg+3M}mR%_OMtUs_4nh#S7iHSnJ>3Na^p{ZFB3~AmN$&`n-uO(i*4Pu!#yIm~5ld z>KxgtK6Z)}3uS$z{2XxYWP6T5&y3I~I0m?1)1Wufo>`NWcxWDq04g4W?oy zZP|bs*4og7_y4|6<@<=8dXEM1BeWE`?R|y>JPlqi#UhwMP-uCM>YIRX?nis<++J7t zXAGK9i<&PczBLZ}d3Po=$@~F+$?w6h2uTGcKQ=g=(3=Es-ko7zbRJCvaX2dOESMbh z2DNh1c>Rx0!1}L96SwvTlZ`thV8C`ey5hjkdQ*|d@ec?L{>MA#Db&Dp;w7gZoTx%VV0Tb*Z!7YVr6$?AZ$(NIYtCnd=K+(w7VRy-&ccqWq zrjGa;=lc9yzXrk`V>|4h;q0hMRuW)SwRU#J>gW9kxV8S*-268_|JjjY46v`^4Ndti z0Pff>x^SDf5D9oEKpa5L#Z2}5ydQz(_&>nc`cL0rJD~6lwi(lY1Ak6CmH>=QAn1~S zhlp)(lkB7L`|O|fsKgyp^+ljet9+`t=n*K&F1xn}QT=7G$>f zYocSwZ0~QLRZnJne`j?HGTZwbOT3WT-rqOxnauY7rkVj{w)gi{+#s{PzqM!$neF}E zwV%jrkIeRduP5MV-XOAX@Avtl$!w3z_I{uL@8>=tvTyG&+L{)aHCozyfY zvpq7~BeT8#&lcrQs@{;<9+~Zt+1@WaylW?)>q2IGe_^&KtoCD{GXwVxLyqW^2dRpc zBNCjps>gH8&nt362}eMU2f{X(tmk6~CT_E#YHiRYpuHQ8hk}U>%tqU6kdANur}oWn z*xTR#qc22<)~8-ZWYzrC_7l5*4A+Ng(YWXOj2E8N*xjJb@i zPui>h!}X%>)G*pX#e9GCHg5hR_|V!OT+%jG7YP7HK|vDXu#|kK%wiYzb-FimRby88 zcB0?(7o0?Wfyr-Ajmt|XWH0RIdqn~Qvd7_@yVHq~8W0@7XH)M+HlkMcAqW)yB;aoD zsbKjq_$L!PeeO7dhza

5j{AEF%-~Pz2tfQXJG>DMgKP!3Zk$1?T51Nr2Fn->f<++75t`#NKO@=#-t_m>to?9-`q~^>l(i97hIDd0Wn@RH*TAJ zb3gc34b0y7UX9$Kuz*V<*ONh^q5%^U zkd%SKGQz&hIjura@n93oz6QYDPmFLsJl?J!eH!{=nZ6)Hm~r3)IFpV7BDu zRkP|fh`LC`A`6uY?4APDDpy1R+cWPcwQDR@Xy3@mH;8FP9&QzP+xm!rD7-l$s9ph9 zE*lZFaN>tY^zY5n)W;rd5~<6ZK~>pK1(>Vt*joOxeAjf7l2_fv<1K+SH+IKV*L>oP z)>zag7-3JvjcU0!Hl2-~QBN;gzBE{x*M^a<4VO-4lK)ACr9y2K?e8SIf2Q<>?5FH@ z_JZB~S50r!98=nQW>!iIA5(7NQ|%ZDyr8pe-|Z~{8S6&ok5pB+&ut~;HF&L|kCrDq z4z;Zr=pN4%mKt*jkP{w%vQOvz!l;Fu3Y*R0oHvhrOy_QYhFUhdPFYYiXJpHUCpkX1 zLdtqgs0u%JqjxwsSB)u!2N^TOH4={(b=lRn!aZDgiTj(3y%0*P$&BZM=M?x1T^+g_ zmMs(63oPbs5Yda11%Z~1ZDI%j*Hq!MiK<1@;1NniJ>G|AmXzeH2V=GK_OI3Yy;JkN zLI>dJgokf>3Sg}pFQUvWWSgcbwlRvwc87JWnO;xO~H(@6#A(y5S?xSB=ZP9HHrv||r( zpX)9JFw4V-9nQ-b@d`d7)M$FN9I#W1WwJNs09)!}D&(2AI|~(XDtJ?>-kSlhGCFk= zUVJkbppl|r)i|{Y0xX;YymJhI4;m< z^*MArAu(o9Bi$3)WqRtO$?#?MAZsJxcXbLSiflw67{K-m_j-C@vo*U2VwK0R-f8`T zVKp%v{`+gB-&~& z@YWLJr8lg7rpZO5duc?G-(*)%;d_ZuD1LQmEb1%9m5Jp!Nm zkFm3R@n;K}52DP~`vq{4%b`l#%oYvFn5oQUt5r$47 z<3ph_@@z?V-WTz@y~0?RFH}FNx*q*Mf0&WmLG0K z@pI6O`nbx6D{TfZj|B7ZmE!~=3T1M`S^CWi{Z6?4psY6Ifavv+-kz`W9xsy_N_3n> z7HT=_^!7u0L_e2X@G<|;C37u{(fv92MV4I8 zZeP>S+hE@4%`L_yX<&)*o1%3UJi>w}3x?nFfC@>LI!lNxM0=rYMeDou;tPX=@0Jn5 z`=Zrk%&vFeSL9iKacO365FL9gZaq@)D^W+8FLJirJnY<+ z;vhaz@pe%D^#!=Wipd^)?dKX2K;bFll6L1nKx@G%?2pY%F%Mc!w0qKFmBtZlA?@Hw zw^k)J6ERE3M_5iI*|k2T?rshT_&*b2NkJ#-9Tw9wYVB1GOk>Pw}yqL)Z3YwuVK*D4(?hf=qi~$Sq}}CA8UT=<(^%bqk}bzWmoWI-u#q z_Xmtl4WH;*mDUj*m>0t@ys=MugZ^YJZ#=T4_9i>a_4WX~Jh)T2u7f2RHFcz1*+;&s zI%(;af(Qxl{z)r$c1R8!1V_d(BkPmk3}q=l?nCEkGJAh}?8o#5VhMbh4YB|$Kq7Xd z`pT2G4!fFq8R7(*))Z*{RW#cf3}>k)cj>eoLdO+|{+R0AUVL_WTH*8*nlp=V`lTMP zk9S7`Y-?GON5J?KHXT`jEA}sB3Njez8^j%1gnol~5-GIG|N(jBhKdM^+ z(=F>*^|*YY@eFrsf!`0d_8+!DSe=G3R2~>KawXy=hMj^&q zGB6qihGP`W_=w9esr6J!M-t$f*|gPaWKx5mX~G~lcF#6I5uFc=G|N)qSPHx!imnO& zWE_f13SB@mzkTKNP-NfBP@bxLZ<84%5)%Fh{Uy5JtKl>y$Bc3Fgkc7_%_NUURvrmO zahV6g;%2D6kkTsr`q%^19&XR;qaNAT=psSO`A}a_ecsQNVJ4lc(^ClBUJu`G6>V7> z?q5~6{>bpP`Rlt`H?BK`H3@JrhpV_IMqs$F*DXIP2s=0Pc3{zsC5B;@n`zvz7|up; zbb$n#u#8ahL0hhJ5V|Q8d4|Tf3}FE%`c+O*#t&^({avr*MqHJ-9PfXy+0a40(;arT z#e_dk!8_`9py%%+^D4rM)NX&eb3saPm9J9zb3*Xw)UJhp1$biN?@LQhA}`RrO4#G2 zV%S|6hR}a>q(_poW%~T>?4fg=uObvLCC~xS!Sw*N*t-?f$m8fE0kiMnD}2T$JxBm- z9lqLA^(b>?$Xxe*qFU$b>6U$wHn$6(q#ToH>=f5|6mrsxKTn2#RRVYb1{Aq3&dr0G z|J1X@a!@`>`RsDlKGd7fIiP$tVPumd%iwl)#?(6X_>bMkU!)5cytNTmt#>z7m(J?a z5~**O=#+c)rs2s@h&XF^w0k^Im^$!$xcI6KZ?l#om(9vC2>O}A2|Tnx;bd9oe1Y_+ ztZ9P4y6p+2N!IYhX6USmm>Pd%b9DV&G#@qcqTe$o-ZjYzaOYtp6h39$2ghCq7i8>; z+%5z`I5(<2Cyq!Z;-DacHOm{#o;tSr5TlV6(W=De+c-F0+!=vCN4N@#JQ|mf0Dq1u zeAYX#=`Q}%bi&^>-5Pp#oh1%A#%0X1FUoMX^o(sW3lF#I?Ifcs^U~A4MhL=DIBYwA zjs$%0=gj-;l_v6VwQX)3+g@lFTkLX5>Zqfoy+}DO;aQ%H-o+W32_zESrZj9)9(fBc zBxBdjPA23oZq0IRTTpbBYcj;uM)-H(br%nn^htU?(i2}lbLHe}KTs`@i3T!r7mwR6 zBu*}23`6cQ8tW&Nom>~Vul?sNB8^u^MJ}r2qp(wWs6nZr&ADz=a`Vt7rcz`#nm2rW z1#6&lTMely)a*bb&D#Ne{ub+k5eO=L__^z*+%#{boYd_&8frCVsXb~K;bDnI5Eg1R zd4kaRI@R#LdIQV1d)~a@Z&Ds`plN?c?W)kTtd-Qj#QC9`*T24a*dUbY z5tMKyYeTu{<7FO_L-`94tq)!?oi9$K84cB@Y7gc)T!K@J5M9XSjUBpTo3BEdN`LoG zxZ1UIh;bulX9am&B(AYMKC^Rv?#7ayj9iLFjL)ZsME?P9>)2!_C#(SBaYP7e1f)`m zXR)PK)Wz8b)A^P~o6qM?iJP|;g^%?f74LKK_wR6KSN?AxOJ-qIqxO++Ukjcdv)x<= z7oRuK^x6I4#aXjF$hE|D53dP7zAddf&6n;B_2bD|5?^H2APXZ0P!kXFe8dto@ty%T zzRDwKra$Jz-r~*Y^F0->>rW!=7NT$w4_JMhxOC}&&3&V$0q^nF7{jomI#$Ij5+};I zuHLzBD_?TWI@s2yQqINP!{d~h4F6Cje;+grfB*9PQ45I*>+^X5j~&&lop47Lo*m`t zY5#T_0Mn9iKX36)rYIpZn&`8@+GtyqP=q&b;@(;P3cL&N&?RUT5#M*0;X3*Cu`;jsu6U zsA{MJWMlw9M*0U3M}SKJH5C%3ZpwqN=8@aplHMy<7UX4Gb+Tt?pXe*xEU}xVpK+Jv;*+Jq`*E zc@i2E`z$U#;dx@x%gk3gTi@8++Q#kd{?sorfc#J0BK`lTj{U4(Or(AtqM#tB0R7Z2vP15qgPe(i z^2iw~W@TNFnG=iPS$}HQi_sZ{^)y0~*D-A74|{0Yg{3BrVt?w|fAs9{>sY|A>e+vF z>@WQq2I$GjNH33^34j7Sgiz7+LkGwIJ{chAwxVGRbo6!4PUwKxE*4kG5`iPGM8GNc zMCaRwHaZrL1g1{t-h0HD3!*_bn+Uk+?5z~PK|)3b2zM<$eM-ZrNfCjSb|T<|5<2lH zvEPMW33(YQUq%EX6A@$3uF3~`BIkBqds<%QMd={5cv1HqrNq1XLBgUB5%Ca7>$ z?jF*Mk!&X?%u2qFT=hy2;cWdTK34pYZQf0hqvth_CI=UxDc}lsyd}CK-`bJm!a4Gy znsaMar&nJ5B|LV#B`fFlmq%E}=uTf5$yfVk2V9M`6o=fkoKG|bev1&R#=Uq6cfN>?Po7w2Z@1W!auVN2wJ(fUZE=oq zYJDyDnF`!Q2c`Xj3lOb30^c@tD!Jvs_e9X7GrEnh1SJLw#wIsPc@n>F9b?WaQasr=--Lbuc~ zMBq^kC){4U8mk!N+I@I|X{T^V{;{CH0uhjZ2v#_w{kgBMmUYyD9m1F$wAsj_kTUo_ zKH?(PgGRvjg=8lH?_?YChGII4x0`vO;Kh>Edzw;(#urTzj_OtBfZh~THeITNGPvgE z%ivt4_jO;{@`a2Cy6w_upIK>8e8D}*9O{7UYcaeK1?!yay zu-t7yVX172Nud-oovUuSlV}dVqiIWhc&QIs@-JSiujr(3pf0E>j?{7kYaL>l*!;Ah zrOhpVwtoI~t!Q$8!P7RkH?KyWw-_w%07PGg{?$XQuL?Tu!=wB8|yP(VALBpa9yBicA^Ixe{h_-r6@cNIF^b(ggzBpg-yLtw2}jGr#*>@IG{> z8;P!+o!sm4^>CNZZsq6fbK$w5aqpH=Z<}6e%?eO(toi~v0Uzi>1a_CkI;vV~a>a7w zp5N}cziRqv;nhe)2FxEbIq{YMWYFI2LC3DNblb`EEMZg@VNo>WcU)x@)YxV>muqQ# z+tJ7%UjfYN>QK~;vZzkbbEhgZoymZz$wK7i&w0an4}YrhvW4Paq<| zm)FgMm+48zRNxxxlVZngvI06;vn6Om`298o4y&qq4Qgj&oLfCDYL;m=pTS}b+ujH5 z0`9)%a9n2mIpavK+nO?^5oqb~8KFnlBjfRu&O`u{gqQ2N4=q}f2k~cH zaeYCRCO~SE2mDSrEl{aVDDms7cse$XQO;GkMZUt7HjJ9v45zn7O{E>hrX2EUrLYj8 zxaA&zs!O^b?4A#Q-k!6-)W$rfezA>b|MmSxzVe>DpFtoMFNt8%xZi6@YEp#sd(+v! z+Apknb81=SLUi}HE;g_2*?E}zB$&ST5#Guxqy0#nz_WW%VH~RU*L^TH>|aB_G~i8>-{+Jgl(q-L3Z?vl@oCe{Jha(YxX@cj{tO zyNkw!H4Q|~SIX^WC0=Q^rHV}=&^Q)+zE!ChljnP+U%bxfW987hN^BcBg|Cvu=v%uQ zUQBI-JQ#tFq_OROwIFBR>67j8xF@z?8h;kvdZ`)k5O+x>>p6*6n|vXrUd!lk-%3m* ze#D4JW1PU&iL%GU60W=^jZ2emf?q$`=pu|CCo!>aK6|DARB_14aLaO-V$@j4W>E35^V>y3-&8Hi0#qX|E9m*xvr)P$3iWk|5zZ`} z88`^ zB`;q1`6s(~`YFlO$?skR#r42)M4l4(aBk!CN$&pWL5GP8!nM_9O|eY%Ry^wDec7LN zClMgHClRQRg3s0$exwZ4n1xMjg;X3HwE!<%{B|vpW%^W?#S;)p=9v zZp{YP@WS1jOlk97pQ1~yx5Agb-f-P!E4?}eXn)<{6D4ncgS?DQ&A>Eos1{=4Bi|f* z^}ayD`mX4L9L7=sPGG{_+QF(pXzJs%zmk*3Cf$>Y8l-B7#i>!U8e!F$(F z{$VxCctQB<#oEKnANAe`t$Z2oAl%H2YinV!yOi&J4*TN8Kx$^jr3uacC#l0*!xynJ z(8tfn(q1!qG3k0ABJ&k}8$?g}s_yihr?V2jM?oWBZHTL=nNqZx>j^=m3R&BsqQx)x zOSw3fdVYbLcl_LuO-UIqNUDCpAv&{&S(k&tDa9J{Lb8dU3sV}V$Y+e}ZbSG#o!S5q~|Z={e}Di&;N!+7$rXAvSQs1ZRt0ypZdl zqohLz6INMXz4z_a#YUXLp~xxUm*rq`_(ex@=5)-Jbfb6nMakRea+C7dZ&eicbGr^zz1N@)8qc^S z=qAjsa^ruWQ~sS7ehfLc?POp_C)M~g4dY%Z3#u9Lu1ZD;3 zUN2&dZE}ycw&RnBK|ajGi3mW>5`h}T3)kBJy0bv_7FhD#-0f(SvJ=jM-W|>LQulE@ z1hsGEga(owpEIvWiy#7PdJ?sZVbEQm;6g*f;I8A-4^J1%w(guk{ zUfD8bH^y^eBWh0-y}==GM9k>vS(TPX-=IA$bDCurm2LxP^RfYDDUI8xhD2FD{-Q_@K8$E;qOG1kj2K}6C7Mq@N0-iMio zs}Mpl$=Y+tYrTRaTOQbP9~~RczuqDf zmTL5xS2E^w9Y4B)$CMeTi53a`u+P#w{B)kXD=F2aBx5>3g9Gp$^Wy^Xodw$g>uHob zX_yon*e&<{8eWTitVwIji&{*1jrY9~hf2lgY!>Rlbx=lEb!jKG3YI^c$KQdix?KHE zH z+^r&4_bsV28rFh&zAz*5c-~p*b{oYrIGu@VyRln+E)b@yZ{^3%*)jzS1V&kINX^SD zPe7iG^7Q55b>Ef*qxlq}M$BDBcVE5JC*LfF&erhbx^&2KZ^d;wB#q>r-!}s#gl)E#)d|fj@#b`C zd!jygORqyG)N$*wYg^~WuI_jffh$>WahematG$lpXW>@FPYk=A)v$81Gnf>K;fO*o z`o1pCz@q#=AcK7cG4avPCoVe}h-asAjQL2CmzNofPAGi~1=fi`aE)SYJqOC%h_$xQ zadNF-*?98ddYk-u`}k(3E1eVOhGDGz&|(Eo`E*?A%UE13l7cP;I%)T^Z0sW>5*n>->NGfN(tQ2e(e|ApI-eiqXU(&ZSqIrFQyGu51a_Uy zL{@t&1FX6#fAwJd5nTl9XTuS#+~;1T8w`DwZ@8T~JYvz-W#8Im)1Rw)BB`=g=Mr~& zamJXgaT@-l2|2;j7%`Tq_Lq;QvYYh!_&tPEcl)Up$7YEgDv|=-9IB1;2zMKm!73&r z4Te?Bd)Y%Tc2=Lms5nV)8korTHMB_H@+4U63HQ6wrm|LqD-Ip{l*8gbn=T;4pLTi< z{zk07h99r=IbZ&~^Y|*~5Z4FI;8U?Ji8jx7{kvhxyrCOfjLzD)tjX|`qNUiQizhZX z=#N2gIO+M-Ja8xxpeoK?w(BL!ohyg0SZI~keZyH=GWBKT-Nk?#Y{$9-(G<}msh1r~ z;ruCYpjj&DG+ww)BLS3FpGv`Etm1~~lbE64;I-xgVw@uU$a?)maO$7Z=;eJo)CExi z(8Xj_nEq=L>VQ4Tz=2!1?`L^U!R*-g#4XldR5ca`olW;|=Tk>Yrh~1pAZU?9uydD0 zB;m2R1>AaK=20qRWb>%hz{*LNp{X;&^)`MV&roa&?DC%(k!7_bi1nPRp`07#4cd0g zcXYFiAM0=`c9-o!pG&Kf zh5gS|_gLiU~%Mq&{W1rG_DU;UK?rt@ zUhi!$Otu+TQ=v$$ehhe`A{1rev5kt7{TucQ&?_;cz1Kq4)s8$o7udziNQiT9@jZ=_ zH^MvwGr`?9^DnaS=Q{vA2}Nc#7+}4AZU9XLn56Cb+*^lRvR`-44t>4Kef-?n6&|Iz zI;Os|{MARJvPGDJT=(onMq+oD{zx~#e49D{7%J`Kt^0WTl| zutJ`Ldyz0$jO=wmH!)I-pew$B@~Vf`j(Fh5VvX0jGY1!W$4y~KvM zt8MW`9texw&LINi-rYDgd8cXXPkYoO5#=LConeB-IFE@{y3W_O?hoO?CW-ZXp1CS) zW&D9MP;K8#wgEg@6}NA~BwZipGo#^GH3oAdmr{J6k)hd^YG}IglBh8J>ZY%Idun2X z3*O0~7k*=+HKag0+iX&=N8s6NtWH(QNla|JoN2xgHqZb9_k5{{ZW$I+-50fL#N%Z#+jil|J?wl6xWP}#S2S0oo(KS9 zhrls1(9KR;td=nJeSZcgG7(W&_|+pbvbTtE9JhWE8~jLKf$I4pUuRKye5b_&0b?_L zCx9QA)4I+t)&rfm2%+CEC5S*A)_uFDq2I<*R$4|=H^^p_+{TtKXyJxtk1g%qXInsg zP$U94os&gr5_<~gJbh-6LS2k1UPrKyM~cUA_B!vdu-CiCYO`CXRea4Qcs~*b6%*@) zQiD^jfx=NazTG6p^hPY9YK9PZQcj`~3K}sUwc}oc7c6I5mKK&g$&3`KJZ%wKpsK)N zffvM{ik^ssAZ!=54|RQ9T6;ag)5e-|$4ps@tOK5yh}X+4V!l24Ndvl?v!*csbEg!EiLunoz@#tstATQ6 zLyWO#2D1d*%=63nM8NxfT5lqzwK<70J9x938kb}%?y`T=>VA*L+8F07uzQH{Osmu? z?s5$;&e!aX9shZgyVPA%?+S}gtTXi(lpfARWdZl_9x*g7juaXW_Z^l;J5(ds&0C}e zPoB~-rfu)_J&JR`j>(Sib zXLZBLf#T)DHzC)&jcQp-6bza46KXG5VQ0rZdSe-Vk!TFCH=>#XW2Fc65 zT0BPt92)~&be}2e$GFGHm|ii_Zy001>mp7ev*2hx{tNMEHLcw5X@o}#OMI%bE`=J- z%3I2VUeCo>s0CdePJI=_Q8tj~VyCvPC&z?yT5wikCjlOaPH&YY^s%?jOCn%z^RUfQ zPi;i_s)x+{SZaENy|Ye_FBn&|B@Vq>Ik;i87x>$N=Y?olXUO+H{_*CxM)Ue=uCSJfs4Ehz17>zaC3VwIa{y5OT1|~GIO}-71Kjo3KCjJMRYjfT|~ej^jIj(^ONdh za6vyRpDsxpN?M7MRrlcdYclXOZ6E^dHZ{DTz4h(ietUkbwrDN?+2z&hlCqYYysmuq zod?|I0e5>a1`fu+!5H}G#z0>=hLB^OF|?@9{#a_U)iS_g!98x_G^3g}x=0jw8G3O1 zdKm!ib&E=rY{Si%$krS4mvkc=*PnB_KRzoEg)C?osinVL)-D9MvdL1_I2>nX7C^l~ zolb>9;%-c+K6RW3@xC$HY4TxN@c}oT%LSAEwX+X2gq@33onT&QFlf=2J)p$_HgHFC z4*&7tQT?w~kO>h!HN&mQP&ulsNyp$(_5r*s`IwQB5n*i*y zjdq0qvAKmxa6h+8*L`As{qD6(Q!Bzz>a$R|c~v$V+AxUm$>Z*d_1Vu#{~9Hfa_58P zMmz6euhaOq+hGp(PZB_I#nA;I(sc-IjJ(J~;+Wn`*Ey}}MvN5v;eAuYQ6~3&ir!n~Q;no_)aU2HmW7i+M`=q*=g9nW7x9%xL ziM4ub<$krVIo9ubAwlA3b7P6doMhcDnZLd_OxI^i*Mj?&PnC61%nbZN%Xk*%D$5Ke1H zu(voSJ7}3Ef-CdZa|xc36REy=LwK=z8}c5)QEzcaszxeg7L4!G#R>s6C}pOwC6!PB z^;{z<;L1_oO*_vofqgXWTXer~L_D*D&C+rwP!x1RdBwgA=W_v`FfQ?A!@j|io$iQV z=>^rXGfo?K>_J=Zq~K97DWDExM$Bm{@#qhZ9dcRRdn)!^BHqh``#4}9Kz_V_R{3PUKM+EdQr~qCf5sGK*OAt&m$&I@WTCawm-OdOL4l?_|-d&&}q>y`m0r42h z{Hhe?s-?CEzG=+UMjrr`ifD=s+Kq4;tU5$W&p?F1r2FtDXtCQa4~?$&WvrRUQhY0`~m|~_L{T(uj7x8tDZ@x zRd4HB)!yuY{}kiHx;GY0j?o#MN;?TISs7v#koS9-=~ox^P%cp46&W;ovR{!EZmJx3 zg$O)(oYmjy#G}Q@l*Kc-U&nT*fc zPs__91!J>*U|o(!YL z+YLVE(L^(xlj6Et5kj_FrJsja#);<}FZM0+vj&N)Y1LTBy?XfegH^dqJ6VE=K;USe zw+v4Ct1!)6Rt zLP7{4rabW7h21Yi;AOm#m-$!kKtBYt2HpnS-57qO?)aVIxNj-wZTEz&yB6VPUGAa; zOgi}TemYsKGRAxH(_K)bj~)YD&xa;c%^Dj#akM0$)aeqMfmbw4-E*1eX6l{+T-Zgh z3adzi+ei9-a#6~3lb>{EuTyZYl)+0bEAeI5XH0va#8#~~`5J|#*%;wfEPP2&f9=`| zVzEJIpT<5D8~DXr?58j=^*s)!?tD9s^T#AVpF3aru8vj`mG7KZREvLhA#E2!;!ZCb z0;7XOz(&ha6IINhsKBG+Qd}0pcGtR{w#jmRStE?}(?3+Q!Ga%t3HH9UjJfIXv1;4_RC6e7FKO+>frk7$EQHu44V1JA?bKIWx$uE!*#V!g+<34d zv!Ga5dB%rp zpQeXuUaeo#RpK^u3@NVW7I}JQ(3OsJ%AqU1ROhh8!o}!V{o08O*x-Ez&shTfT_S+a zO?&Bg-U?e9I{SDcxoNjQS}xATYv;;41!c-|w#O>0WS{41WU*?&clH$H3SdS`a+e_w z!7odcUfcUhVI@M)Om@2i)BOUo9o7xjD4V){F{)5N&peMygPu{D?=2y#V3E5m#I7QL z6Z2Hjbhv=uBylmkX)--;SM-pV*kZNZ{HeVAcT38`rjI`EIJ-|hQ;~@^YSjY8cx!qU zakib{A?&sw3@6{D)7&p8khl+$d7t?TOl1T|yZ&9gYSpH0!pl#EE`{rf=v7_n>h(q>* ze?5Qlk7&ayGmVLgu{B9s z)6E?)AxLa2cUH1u)3!e#``CxMCFj9dt=lt1V1f+Oy-yBZlTX9zLB9XxiYsD`yE7sL zU(MCK@@p&3i0>~(;*`84NmN)jK2n4Tu$dfX^>Jn)0??z|?`9L6N$>Cxs1dqBPXyZ5 z5ZIfD2J(rtT`I(s2Nd^qUNm#@dD_m7M0K!yWy%;{K#_|Gq+1izNUAe5lW-V9`mN8T zXA89xfo6RT6^L;YDR-gWXGwPY>)&p#=zliXtozjLd6K3E2O;K2&yJoV&7LPMi!VOW~XYlcn>6eAmI}gSAMfGY^>A7M}GA)5h$uH%CCyN z(76tUcc$DU(HQ-xeEnp+J`s4XwqHU7hQvO1V5MiDKC(VHSXnxha7C-#B-K%cRUh@j zOojCt*&pDbN(ioNIV z#m%#kOb{Cpm|-FU4z*iXHr_{Ww~ zJypKfa3?eO*C5dLr0ckznGIbPpTaM{l&JXs+j(2~kG2I4{?Yb)BxRU21!r`cx{!MA zz2o$egqOHrHF`?%ED%L&bw2dVQ+i!A0zRygZ*8#XE-Jv#mTlw=sgTaR=`!_ZZECrqXxhCR;l z&4PYBOS3$h(#ZN@eMx-#*UZRgC=e-qO|MhrO4Z<|k;8eN$FD2O9z>_F=&bRrE3wWM zl(s^_#X5ko!KIw9L!VGKU1@jk3Q!90uIMfr#{sO@-^Sm0OJn{3Sg;=?Efc3}l1oiV z7|!1;h-7(VrQz=ZkfPxaIObD6gJODMe>&?!Y4Zi1+z8=b z8dJeCyyDgg0iYHYVf&MFuhx7%r+PVmCa+^YA2q@Bd{?y7b9HIi(NQDm~CR zzHd*gC?XqoP}bv6GGp&!k)~LV5aiOQo*`Q3yM5gVb7Ul z0nKM__fB_|Dd+`1upuwQv}ENjARtSYgRc`;dI^`0ABR=~=U5r*Zasug-omIg?VIm5 z=)<8Oj;$|qwMAHs!i7+uH`x{xnQ%E12rgWyX5=+YdSDado>^76Zo|79c<sU{AZ~62WV`7*_Ss4<0(}33~_=q+=46|n$sdp6bTeTU# z&I)PT+*Ihau6K2^(Le(uBD$7^g^PG&yc3Z0NGDu037+ncr)^IVfgTAMpAtY31+**Vj(oUagb1vm ziXJJ1`fcVp1}UvI{Z5rAj7&7T9Ed+=^%#;lLL4CKGy z8`lIcYSt&5!3ytI?|Jv06AXdG3-hreQnhW!`=K_8+(wafZL_Dg?vrBwVlW}t(B8#; zB0x(@I3NXim*UVaWb+7eDQef+kf@K&oNyOh(xNRFWjz@}&>45jy;+CL6rS61zHJ-4 z|H4-Ck^|rvMJe+w@Q*$?;F<-dWC(Q zQflbd*6_;cy>R0ld>;7fc7e8n*Z4*CJb4ALC`8DW#UXDnd4^mj1 zF{c)03b^pQhT;xFU_XS^#E)gaQM$*xD zr#5}A5QIb5uT_?ZE26a|WlYmWLKl2_aEuZVyV7`xx2;yfhq;)~mfPwmM|=Tst#&%; zIcoK*+{Gy+eOVvLs@_alB!Lq{s zEiQ~5t8(0Q)lUNw%{HSay#CQyKDJw^{bZQIw=PYV{@AbmE@UfEX6V#d6$=3m-95V4 znEQ4Rk9fN{3hu>Ax;k3M8`|EWj~1fgv5Vt*Vh*PFBs0zjyj&An9rz*O`CYBq0CZ(& z7i;+Cq737=lMiliwIaHnc{^Y-1P~2n?@05Fl=wfXhh`-YU;uoyNc^Iqe$cYZoTI$7 z!<%*=n`MfaB_g0U&UiMv-=+1+PVSY@(&4UI+}a5q-sV7?c!upc>(>ut@)2C{^_$rG zzOHP_6jQU1W{#og1cAxb`xYdDT*pDs{K#<{N8#T5rPt@}az%!tM|hmW9IQ6e8Hhjt z_mVF^E}E};(c{c$%V3ykI5O%cq{#xCw;GyCSBpW0Ic9m;LW>G!Z1b~e-43^2aQ+hD z6pFSk^H`ZGzjrc5-SPswUPN9$2j?|88?y02;iRjWrU^{gUfEt>gZEfqr4h5Jw?7B^8ke6LLR$o?dER&srBfSc9J<*Y#u~$Ac`U% zPnHB2qk!ZeUFafVg8fP}-q?L{nokoq>2B^(k1DAdI;bn;J4K8oK=64=- z|&n6-Wx+ZQ+@@FRzb2Knr7!uP;P{~c(#XIhkiINxp zj=^5?S-IbUx%>mOg8I4mGbD=5ZkABXL$b6|AcYwbQ-4P^g2#*#r@F27L06`g+rWBNyBB|U2!Av^2^h(&De1_e^|$5;7`Z1^{@MZc7#L$?0nH!!1rV9ru( zfWS#g{r#!nnrS#Gx%a4t$`7wIopJCtFh{>+zQ8GSzkwzFqlOQDzyo5W3Y9;5m4Mra z-@qXKlJWAM{S9pC0jxWKb-z*r{Fn;$^MsGzDcSSqP~ZU8{aPjPL%!6%CK>bxEN}qp zeyuzGV=C3ZAXoIiVSxi!_e=ip-#q`bM5`Ya=Xe0PB8LD(k-^RrLVY{U*8c->+f$SFF5FhbqBES?;XYm zY~3%#2mh|~2e9s!SdD+^`~j@{6~WR2w(fwf`@59XFOet*ut6ny?$QBU_e(JCA3c8n z>keSu-(LdvfUWy&s!#psm;7rA%N($Ef3H9MH}5-O>wd|B`-SrdYwNySTemS5?D^tF zLhx(^+9Be8qTjF%56fq-YaoQmUDED{wzcb|;-gU`cu~S=A7YkNweLtN?iy7bTk?MO zYf{1LZo(x;Oy-`K7O8G@8a`-jLW%q4Eyt}bQf+J~bcdm~A#!>7Ji!;S#)Eha*yMpY~sqV5|@NpOZj8aLEoKwikA? z^Mu&>J;c>NPXM(oLgIf7ev2DBreE3b{xn@`UnBlJTfV!wFb9uZp_9B4+DZ7!qjkT# z`Aa`Rf15J#Z!|>rKiU!SsrO|O;xGMRML7PY*;~@kxc&OO|1tk|t-qUp|J7|)FYdGs z!g!I?o4m2*eaZivv5=AP?p6rbnMOP?Ncw4hf7ivo+|A#Ty;fUxAdf#w@Bf}We#~+$ z`EE$+lVQ&i2KV+PmU>r*W>$98TiiS&?KU1J-H15Lkc}c$>x`XMBUL`uANrrmsb6_( z5b&m51R#|c;sfCY2uDe4r>>4A5rJ?BCJ>iE@R#(pCrEeq{QrN|B^81xD0MIs)XHuP zvl6DVn)hFYxs!HTLE(ieO7T0T`&+e>YTtc^X&zjhJP-nDRQ$Vy;PTxZl|nRt(Wbi; zojXVz3aSDeT>F#m`adg|Ek#-^ujk{Unz++?aT>x%a-q>m9h{o0>CT4Z-ZG-4DkgLm z85Dju*|O~GX6<(uBES1gx~Xwbk?%`6(;@;Bu~54{H?}?eAIkl^FCIr(5!Q{BjF&yf zNa_uV-#!0lRrlx5Fw&klLztDuX#P^teljb$L_m?W$kqpm>-q63Dn?gr%lpE+DYbmk zjx`tFlAP#F#F#yS^b-E^MSAL)?tU8MlI%g74&?Oj9Z$Dm!kG9<*DXgO?HH}f_2Ptx z)Xlmc!|U}eDgAxi0ot?5r11@Xf|z4DKl79E|K2}x-q1Ht`{d(o9ol2Tc*~;1^@Hqc>dZ zkN2txpXHoZRpZkOqQ)Qp@V{H)+Lt|e1&`mAr__zH^VrxMxe0c%cB17E7rvUg$x3GU zk1|&OhWo*0IMZYSS6W=yuX zfPCaO7CSRC^2myOzS9_&*Q$XNAKyPQHXURY{)!w|$iF%>grrUfr*D8mYWQ%uXK{`# zhDY-_6388j`$fX{17g>YY)l3AR(ZAZvy@{@Lkm85d}(_U9AIqmDQv%QbIGSZPt53v z_^FBV$Dtotax)pz1EXOhdzUZ9W(P@YrE4k_)C%}9-v8IF)7=541c45f_n&Oa>NuI;Xv}eNfKMky;Q6i?2 z;gf{2w0+hMvI!RwLCv?gfZ%~|`aI~*BgbKH)kJ@}s^*A%o$y#V5R!Cs~rqgCK za1#rZ;50EW85oa-;9?D~Z-e~|1=Kz%Z$U)2+Y2No&IebW$+dGYB^<9Q)zngyki9n* zYn#R;Jefl>5-7oPBscEuVvFMO%E40)*;PmAEr~#Gn9l$afIPHMO9wkAPYy8bcN$=+Vkb;V8`&`;@%riWPqG$vAKWbqm8Pq-`P%WN?q0;D zJKR`tq09Z7dzc`d#on;veSXEc2L^8A zo{zbNH@f+skxkU_elGCVMDy_R_Z_m&xk#nJ>Uz>Q@K|-LAr=}y0IMf4hEL47n!u!} zW@gOO15|yj1yZ($PgB`P?x#7#I6gV1&vn}1%HooSF8)dW{-p`4$6ZBlaDlD9B7v%! zi{+L9Un-s$oabdW8sP;Hv_R2oybeYu($GY#g`f9*pWdVe-}s%2QIgvwL6_#kX0zaK z`3!L-NO#6$m-~4!w+?Gx%W>(Z2y8=F$0Pg+5R}VMfjkTP%j#^1WusVTfDg@8HoOYF zcr&~!*+(I6N}S;nqeaX)-dx6)eqz64IiFwDxVKP`<vZuF_A&_yl2L(0;qXset5t%BL4+%$dOfIT&^>9u!+{iE z=AYk_qpriAe19>Iwjn)TG4|MXj3Gq3DC?^M&3mEs1fD_^cF$I+^a))vCZ&~i4kNyaKv$AF zahY!oaXj@&h{k+&<(cPA6=Avc_tpbfTGP@G`N^-h@{1^*$L5v#wL0AGr4|ol$YnLC z6M-ZmZ&qPJel^VVgSR{t<809khBmhhPJPUw;NQc1>9*4C%O~x>r2_a+W_aJiv_?wH zxPV#Us=cgLW~`FOnSoh9#W}G}s;XSH&Z8UkFsX{(bT_Z-xzgqLD|ylt#{;0Wi;8UC z;)AYy4_i$K97Y?}KSbXr?>Z+n;=!Xl%loi3%q36aDdE@$tWFb8ZY~W)e3ambn!wzi zb3Ff$2rF67oGEJ+$7`E&=gy}B?<>m#*dmOAf|xDu$XYz>R@}@1HK&n(Ggfm1HS!!k z8W$jV@-Xu!Z2)oFxHE`(xxp$35BIv;H_#@aD+vEqAhyXcW zI`7JTyzJytokFM>Z}L%5h4UcDn@+%kwB4;Q_o6QwRy;mhjxMgo)yOF1=F+nmDo_T+ z2Hw7U|K)v)AV-FI+PZd!dtKvhAiKzY3RE`SZncrXRoGi4od++~8Fb~$YED+?@V4f> zcG~E^ED)ROWUAn8REk350ud4+hiRQGIAG< z*tmguj4JXL9Nc(ueu4;uIUqMR#SB&i1v#_JF57~iHBM7ZtZSA8~R7CiMkT}fT$-mDhU!;C#b_Z4f3wF^9cl;ijy zhO?__uU@0|5~|lvHV|8i=Au9}D9I9mY#Asnblfs}I&Lazg;c%VN8@#m5^bJt;PE5= zZ9;FPHz3VE7T40o5kZzQ(;vrnAhytT)V5VQUP*fWn-{dn$5EpOCq2Q#qjkLly++3I z!jYjtJ{m2^I1Lg?l}!h87OYM9=8R)HI@z%bC+ZgrU!w$u9oM-vWIQv(qFw<%UC=2`Zi)*ua*dK z=($Qft^q#uuBPz#k!s>axXtBd@6{sk$W70xayhYGne0D?b3PsaVDWFS@5@;Ia-2pt$ zI3l!E#PqYbCS$@IV*quoh*HG;;k9=Hdtre}{JNM~z3N^?0h=NN^@TMPHM+N{d|ZB3 zikVauh{(ZrBY5p-^tF72SJyQnmfO0gUREk_)Te@QdM?uYh^EhW`dj10#Z}763=4

4e_U&7$tk7POobb2XhGjYu-S^b^ou|_$^w;)n$f^Df0N_w{!j^MQs08~$Jo9M= zSL{oY`}xBz{!@@wG*l&M79OiI7JLm-ab>y|rTEHQ|(2FJbTW(qn0xw6Ue}IkNVh2QuG2 zy3cvut+7sLd|GF^c1<2i1k9*OTWy*IFOw62qomoi*a6v3Dnn8lU$UJHXGG7~Oaxq* zOPq=hcq?iaqcKjqK#;L><1TGdt>|;iM>-JNnRiNh^PWVfgyjfHk^Qt4Bj8oKG_%a= zD;mwnb?1ryhLDewBWb!Gb|W|C1QEPRd$ps*GI31#^KuHcPgJvWh77pUl@FJfsc)x1 z>B_=3`NUcwiJXx)lz3{6VUZ!;=7Y;fJBaU z;;>YO=0_dh7r%SnIs(5#hJS)QhE-W^$w24dT3ERiZ+q(1*X90+5J)=mGLAgYHtBq5 zv23xHYHy23(J<0gt|`wwg3SD^{iot!4cE#YAuAXAL@onuWy%O@DYEM_qXz6U{YB-p zZ{=dV^lL|KMwxID=>~S5`_Bjd-)ekIc0X!qa7o{H=fi>&-W%$++DG@Os|Z%?cboLe zWoe)z5Xfv#!Oo{ydR zvK4xa&+Xsa`CrUFxkp@gkDU31XS=6atbM$#%LRAJwKGxfFeJ zP5kMyYZo~d-K%@m|EhZ4vQ2~k>Q&xa^`lbvS`%xNmt`48N5`w3ZgKwJzJ1}_iMKZ4 QSZ+m?wfIo8?f>5d0B7uk>;M1& literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-11-Abstract Green Wave Design.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-11-Abstract Green Wave Design.jpg new file mode 100644 index 0000000000000000000000000000000000000000..1816166ce1afafd4c320ccc7536d1b3ec1ce32e5 GIT binary patch literal 183056 zcmb6AbzD?i_W+C^LL`-vR#3W;E&)+d7#UzFhek?bNRg6I5tUS6Ko~+|fFY$jMUbwc zTY5-o$=|_y?{lB$`+MI%-o=NrYwfjT_E~$?IUhNn0Vvf}R8;^xJOIE0|A6yx;2v;^ zfRK=o;1c+_bm?!LvFEqYEJx2Rvl5)X7!1DhmO^emr8_ZNX= zSuO_r3;yq@V5>A0I2(LCLV%FqG65k09zXyNN<~0Oee?Dun!8%JM9pcHVc@ta*TnjW z*b5s|XSeQgJpFY(29SfuASM-{2%J~~0GY(wWKN}v5$GZhk^hYJ|Cf^WVPqvN^b#hx z>5-g+w19!DYZPLkJo3xmn1tvUj;~vU|MrW%+uJKZhsN}kxQ5bBiB3^XA1lvFZ*_U= zAe;GR8dat1xiaXwGEP?J@dV!R{nZcuX=Re!2LbuN_L=VW;;6t^K62zTHbqY-yGb*_ zgvIh}U+5oP!dx3?HtJuA@Bba~e{yOqdft@unu6M~sEO`7YLP$wr}dwLO^WGAQ|kWn zrJgGqUN*6V-k)8{9^o6d{ngGX`VZc}soF%*>y52TFozP0g&E119}(|qxJMu?Y@Dq&jvwo}q9g3{Ah;6K|Rw++C7c?)aS< zVgAt?Y0KF`&S9RAKDEi|O8f>9K>7vv^4~=9ymgkeeu_lb5t9bEy(w^q)v&Yp} zls(=zhP>+K7o7wTMt8yYvc^O{-Zy|bRkMLZ-rQx2`Lk#E`pV8acn;*(07HUHN^}i; zfYgjt0Br}OSu`p}Pw}q*~4BPHroQ5>9O-Eq>!Re*cr!$amn(^Fj8bv=uzE{r`S30 zt}p!@7!!&*jfV)JA@N0sZuR;PA<2VLm68)`Es-aQCE1E%AHCl`$$#=dJK}wM7(HiK zY{I8x!t!T5uxixijx{e--MnokX97>Yr-Kc^%V1TO0=D0GFdqoSjN*SuAjbp#{8JN> z)8RyD>}JgA8X6??sd&v2$S;lINiz@)Zm6>3Q9_AGh(pP#fl%^WN;(`LfLlzD&8cq} z5}JeM5#X`8jN%K?*whea%DbfyjXKPsAYaF+zz90U4#|34lXAR%%j0hMWovmEuc16c zZ~LK)#YfAu7QZJ|;!g)7O&*)gQhIZy+iLmkE!du|Ih#lxNl5(;j%tl_7xQ{Lw)0G& zHga+Ij$-LRN{RT9>7BN7fby-xZ$4V~WlTof&Qp2pv!U48OsV7%YdZt0J&kV&?K7k6 zOW8&iq%M$t>DxM3en=j?J`U9`gWbKOZ2T;1ZZG{)4AZraG56)W-ak5Ei@8u)Ipqa-AnQ*+!0N)g4pU3*Jci3B@Bht zkpJ+QRYL9E|#l3%l-=C1YBFZw}vWe5YJ$J zj`wV8{NHAb#cc%_KH_}@sxqbM zIZ32jb0C(N@Pwa&Jv;0T(6!%XchPOIgy=K@8wwgP7Z~oR`zj@DoG+L2+VwyKC>X&a81ZKZNEfGrxe{2|<;zPFUA2rcx?LLn>7OUa;A zRW}=_np;=F1J0D@HF>b+zNhoc(bXsh0KnH1*)CEd5WuNuhUz@6)SeWaZf^dwyo(3^ z$bqX)vXYo~MFJwzG3|S8PFg-6 z8njKoA@V<{fQd--6kAb43a{i=3_MAk{Zf3vDL=-^oDViqgozVbV^YrDPdVGuN~p{+{QyB^D?P8mi}{r56)gUhxd$|8#gLd;=uPp z+;nBguljy)Ir4S~RD9)IvmS-fAzRQe5{ek>*w!j*0082P)m-}~0K8TL^sfqh#jhZ& z3v`+R8hNQNIg_+&xKfgni|)?JMo?z?rZnz%1ZQRkpjj)s*sS<}c6 zQf2z75wegBAsg9=)~h!5e?af?F_XoxH%((WRv2e1BR`*mD3;4zFV@@0)A$ZEz0$&{ z!3jvla_N6o04JGh*z=U=jOYx&tAAA=065Kn^Cpv7enR@T@+rta=>UD^$e|IA0MtWX zcE9;IX25%VsV%83^$&zIW)4c+gz`dZW+K!ZQM=rJD=`7A_*{&zSVbOn?Y+oVl2?%iO?WIYH|nv;ktFt^R|?vuRk^*Pztgg~BTjTt&ro6KYAzPAEx(4A%gnCPv{D<^$k< zRpk9(1YXz)fB+J%P4uS&K#qWJV5z1dp~ypWMjhT`M%AwD59=)MD(uU0;NkllIOPS1 zwtxIn`adb2$JI8s9ifU=*8ot8PwJkMq8DH8F5zi$y3y|+q>$SR;}eb6hN;fQJvY<* zt1tdQ1YbzEYd{uQ1R3IFbxo#=t*sN8osKc7BXxo|)=_=oC_Y%2ar(L;{+F0oYb5|6 zcrgvDPRTTS0?R_yehbN=i`lp%B_2Ti&$RGjCi`_21=45V9m3z^iGK6C|JPXH^HOqk zA>)WtY~-JsTHf{q>?}9qg9V8}F~P|}`|R%0@9;qk=R292Vk11BHiL9dVij?V%hx`L zPuDz<9$572he4Rdi!0#*96MOj003~8_JMe#r@w|`C2><^T>XCjAend0s5-67Fte`f+YFH=EM`wEkVV z6#a4+ADwn>GBRG{_I=3_nt`I5<&i{6q6GhgjmOJqJC2R)coGpqbZd77>i-t8uq(#X zmr1@3yRVbO`w&_0(fZVZ9h#s>1M_>WYxqR-+&)`WW6y{IhuF@Sb8<^d5 z$ExsJ0bjR#Sm~v{=V$k<`S5}u}QJ=F!X-0W8d-QAyi^L%fH%SAmdmg$C~e;6o-(_D|q;7 z0%70{d!t=Z6L{z5rWl9qO0-9qjJ|()VIE(&DLUd+SX*6nTp7x#`^-+h1QcbPwQ^u3 z8K;X2zpMp5QrXD3t9v`>wzX{rj9zbu?G0$SPL?XFxkGNUMoIqAz*>k|;#VrI*Uko%3Dbhom5L7{HybjDD2@@_u!@txL|gBQSz%1= z!ypM>;SM=_|7px==)v7@a9@!h)4jVThcDX6P7Ac*n@r9fqAlO1X|dzE!OsePfgT)>mP2aa_I3gp)Jo@gC~!Z%j+_IUvN#STnGUXiQ15 zD=mK;_NZWN{5LB4@ab`$!G4LNcZSQZq#aJ5#%~W^{!2vdr#5ctv5Ea|&}i({C;k5G zOwk!9%8X~$k4BcoWd|zqILBa3{`b^$jgk!8^f}IfzS=O|_~5Rz zAK1u7^KQoc5FeOsm~No#ZIn9%REX^_1-uTZqkr{wrkY`-@kK51w=MXCM*;)vby@HEM8A54s;Xm=T@aDCYnfiDxvbzYLRtql#$PwNZ65$0;s0}_K?-P@fmV4V z*P`M{JSC868Y`75=yRYGY+f=7TPRD3=@lc@c#n%Bzxm(d!1!F_RBU&%h-J2D3j~SD zcsBkIJ(7+VGL>|!)iti@dg^UWJU=IweBEC>@bt)uEWk1_GBEY$*Rkm;o}*U)-J6B@ z7@BY0Gdk>?&8KKvoo0*9o%gvM76guHA6whqER}xBJQ5jbxmah^a^bR_!k3GY+HROU z*mOz?B{#b${U6{K{(-2=4FGG~RB~6GNZMt#wk&{R`#}s}q-!+FsoDw-Fo6q(LHXn{ z7WI+u0a4|=N4dPEJ3Qk~g7X1}FMtx~eXBXS)XE4PYXo9b&uTsf6(Q@9aXJsxO*Ee6bbPirHC@?e;hPU zXql>ZG0%U2I?9Q}vWJubp2KOOk1_7LG>L_LMG9sd`Y&CAYR0LH3RR3eVnUY`x=Za>Iftol5 zn&zKF*HIYuoC7!Rypadz%DXNxJ|@As=r}MNu8ZnYnTk!PiMs9j(QAStLW_c3IW64# zKuX;|gS_{1h4@1B< z3V!BJ_Yk2WAKEr~731D0yqQ;(3{C2*Tg!=AP`Lk{UZrz3+?=ygE8UPoQi&@7X#c{^ zOeqOm{Wva8m@XpOP9#tB>Qz}us22k_0fBwGIYkIhN_$Wj|5whGS4CZ4XkI%94XA++ zhS5^h2bEe3TR~kp6@nMckXbUkneH0{GyFxC;H%f|*`l{`#(6&)U+}+bHzW+pz%H<$ zx2T!bG3_#t5*>%JJWA7fks6BJYE61us@Ll}3}B}apn$0$ zab=r|TzPL$yr-F(dTK_e-}(XP#^7;NvT>!1%U)Y9K%I2so)_XP8$g)mq=`mAMo-7q zjd<_MNL;=F_Y`k);vojf;7NOS6N}0YJhymI?Tf{;s?ZhRRT3mkn?(KTz?Xbqqa~ z?(#~)A71x+b0@NaoO9S?M#jd$oDv`}AhZnNEczE0wCy$}WFBh!6NqN2#lujGQENoR z%G#d6ZG}v?Uw5g#tofi9BV+Ntb-;qmX%BT+#4ewmm2Y7Ca3P9b?ss&@eM*QI8%XZU zACecjZ!JCmgV(Zk8p5D_ac1n)aa!AsPTaON>1UPGo|&XLIJuV7uL#({YMXU|Aj4Tn z*E=`0*lK$f%JE+o0>e25=>v)u>LNtcB@wwb%BJ zlq}Bg1l3teEnygM26S9?z8L)!QB5(&)|dh)&OnB&E)ZHA^-4qrKJ{phyf+08Y5!93CztMz=-IO~O!6sL3(b+>k^ zl&eEI0?ZKTz_NdhQTI<1-nIGyzXcoJ&qeieO~6T_VIr)G?Z|=|4C!1(|Q0OYuY07=700F;QR z4@h8>WTRu?YUU7up1y+?;8DHSDn*!kxxaO?H_7ppG3!QNVZbiYuJY-1oC9>WiYbYW z-*&>bl>IimZ)7B2S=psNp33AcAEHe;g%>i(Qe?0QSQ+QiizOR^ z#Bb~mMSQN55pzt6NYRW~*vJ=34z^X{O3&%)5KPTSG7nnqmpet}sR&re52 za}1$m8toc}(ln=EU`ewt6?x^S)h>zBCY&kfSpa(zk-3KJNPREbS z;7A5>lb8?F^j2oXqawd`=OfLmbD+Pa>#I4{8!Pyv>N!w72$GI5@Ll!PLrxztT5FDL z-*{74Rnm2L{ebWI9GETA`-G`QhP^lkTw<$yhaLThu0$M4=8S7po&zgI{*CS$XZUkx zD$tb!lcjgu90aSiYyjTv77sSOLJz$#UG@I*up3W!I;frEeoHi5yf6R>L>Mn4X-4T! z;m?C`=7N7+K${Amt5)>fwS`>*1}@`Ae^ml%2!UGw_0&Jbi&~Np7u(FBH;6q!6{r|Y zRsMyg`ofaa$7_Fj#GC^kLGJ#4fy5q^#{OS`Gb6Q(EoNL*s%V)I9~-X?)q<&J3~U&b zr13>YKTV1C&1g~4AOH5?!~?pZff>c1?ruh#=0*P9* z-RAc5aTAjc91(S2NR*mf`KVxbpu;oR)X(fFWAn?Wkkbv?Sk78bH3hx(la~I{7283R zy+RcQtKntq;G*&q+DY4j{#cEa0$46OpIl>9Lw|2^N}(4%N}LgMmY&0of`3| zr+hw2t`Xm?ZMnA^Df;Zu)u*C_uSM0q;Q*RIkq?DH!p%S(C2sS;VP`f*#Cpy|N6cH^ z!PAlX<$*pvo*iQ&tuL{)!V;`*MvdEI=^%k*I}`-%Ln;9TZ~Y z|!PhE@YCBP#=~F(%q+S$plhpH56FoTnFW|=AW90rR5*(9F zW_~qbv{Xx)!^~k!L!r2QEukkzddR3}@UklpL0X3~AS=tJdpK?=E zhK6&wUX|%}kH#wp^H;6fYm$%RGWkD1a#4qIHN(q@N)6Rs!GlDf9E?yK&-daxa@Ay} z-aXK=jA1XIp2GxNH4W=2;XR2)!8aJ6S zj8v7XLGdLAwFgA53zi*#UPK1Myv7yKFGW%8*ee3XH@T<9z+m-y0wvgAVL6))C0W>( z?9wdaj&_|vNCp_*P@%_2A~X%jqho*!hHt{JRR$b9Ey&x#zeOWFXct&AOkg5GeG$>F zabB7WyOtv@sZho!BvtqdD9L?e@F%z<=e?&WrBdu)e3{*Q`reTJ+^B82TaQMYI3zk* z+fk3v(KzD4z1N5F`Kzi0tka$V*JN;q|cMnWTOX*bMUX| ztte|xXURm3)4BJy^$P7F{+-{K=8|>B8{aHz&O)!H7i+`na?b&4KHj*rv@Lg`vT}ES z?=r&!;E$h7u`1#6l4zU1%M zq2mzC&aY(^NquOTDRJ>ih;fR>jw~Yd_RrTeKYg;+iTi{7bJPWB_S<&f|DF_N*TqEG zn7aw=>m;dH83{K__!C9BZ8<;NkNl~w0{`Az|0z$~xIX!3Rr1MN*Q)Tyc=y0e5*KI! ziLzmxid6A*W~D6u(4V4lke!wzXn-?zv_1#0onrO5(@WN;uGV|-)3DN5z%FMeHb(p$ z@Ww=)H9-{9jX><%|3VejwtU)nA`1);}m1&FJtSStpfV8p54m zmAOb*!)NfN z-_tSoP^acKb&FH)PVAyR!jmoQXnPGd`I{Z`EVeQtq!(L7_m09$E6dYf-g8c_Rtxo` z{RuOA;5B;pC0m62e63%)o3JUyvfj<<{$cpuS>!o@WLRj{A3xJNiIirUX3bg@!j!xJ zqS$$|zkPVkd;QpOIP!OblxWQ~Rhr}U;ri0%~!t{u+bX)=u+Uo<1w zi})iR{=<*|P6-4{Y@tQE*|N@?9L2ArY-@u9INk3!k=n--C4aG#G)VeDHja85tebSF zhn3v>#{FoCLg5Th=?4{k;J?`fjq&z=dAhopNi6aTkJ3tAvTIOsE_ zh!RW~XbEe<>-XX|BPY6eE&BwPOc-i-oLM3HTZ_o z-_`#nBw9{0#;&b?U-WKo98oVYtY6!?S#u7U;cb9Q3^{@2KbSOOMIt-lGfFC+h8>K_ z6}6B|Qd1N}!Lt4%!>p5Jbb2LtSL$G3PXcMr0lNpBFxX_<=eClZ?%ee4DK)92bbO$J zob3l_-XOi8i1#iulol_3YOL1dx2z+idO2(au9^Nj-(JPEQ2iW$Ik-JdNTNm@F=I=P z@=0~ANJV@WH9WS(PI2M`VWkoeyamd4B(n7q#$7v{^5@4SZ!*jI&aond`eU6wsSUel z3+aM>swXN%4xvR3V5oxAqVDb`C4e~K@6~*{Z2l4GKpw9eXPpO-Kd9&Fn8 zmz(A!+4^37)oK`*m**&CjjUxH^b4^lB{g`E?mI?ZkzJ*hUUo!f2+RMu(wOb+pQi4b zZoPC1y#8E67HcZrUOK|;JQ@iUM?LnD6|Iav<%=L&w#8z2g}q@>+ntE@z^6;kmURdD z+oYAX&{qcoox_-rnKIVePCqTY7`T>WwpDzIh}rN}@Yz&pUgNXTv89qc5^QUrdKIwG zh!+5P97+1xdHPY&QO#Z<~Zjtkx!l%Jj3UJZVIH6 z_qd<4UBjX?iWusul}XR5@IhdHpLXrgvEwATJZcB#X?ox49AMO;=L{<)d$Pa(e8?|D zSRm;geoxRIDms&yQ9I<6FH+02c^ch><>lBcRde1U<&Qgs7}-flB-y|aM{Ol`-6ac$ z>C+8MUIo3+Q*6GKB?rAV#N_YUly8Zi1ED^oFJqqbjJJN5Ha~@O%O}HFwdo^=jL8zA zi#c6t#eVrJ2{};19Z3tKYc?_pl~#sM6_@a~0b)x!hP#sacxLMF8yeY(Ik!T|>6`E* zGf0G@7cpOjV-+#=57#y| zLyJAKR)#w2>pS}*Q12!@Lmh4>p_1=rHXI3Zp*@=G2HH~~$EfGKH^|FcOZ0~kFOUze zolte}$vclPe%}5N>bg}lzU8YzJqtl(Iq#8__UooES{a1V?N}Q*Yr><7}F7~uk9ms z^4M=UQ4aO|Q0|?raL20DHPhhd{gS-P^4ITL_1ETuFKq^m40Zl=GUlP~p}%&L(L^lH z;|`%kQ_Q$#FGf9okCV&F;CtAmv#vB->^ZRQcG4q)d-RTbu*hhqUfW4Sh^H!eAw6rr zzTkWJc zJ*3Vo>)vv@C?ss#H|E5!d|qR|QoOA@mUY9Ehguj?(MhFF2i0_QUjkQ##F$l6wfrRkhm0 z#63kT69z{a4eO$El{oKGZkW*^$AYlnLNtC~R7g9{V6oOPcp%Tj(Vc6DS0V4e>`G+n z976zj8G(Su|Kcw1Y(NWR~E46Ne8{gNXhTY-!bm?Rjd<@mIdoQ4iXf<$Tj0FGpxP$5snr zN*B#UDyr-N7*TMMFJb4Ee;|2Ec}*a7yh*)S@oT7}94Gx4PjwElfOc?1z7*GhgL8RI zw`-_=a}@@f)N5Fm_c@bWC^yGzjUrDrto}RqFTQbKI75%KQUkyS06&0iyw}czgdZAi zF)$FR>F^DXA!Sq#nwE)#=C^C+82#QU$0QO9*=OS@`~vk%JG%~oC08S-LM@J+UbGD6 zIAsoDZrE1zzuA;}V=cax=ly*+epC3{u82+rMt;%hr@6*e>&}fd$J$BN^eC-&xbNmn z{Yfvcb!!ole+B(O_iUbuJR&JEGy6#Y%#scm`%*?)v^?f4VuF+zj1`n_clD)xI&Pyj ztz46!vtb1ty6y&h;(zxgd*pz6u47dkk5- zQgv84f2Q*7v`a;rs*D>SZToudw~rxJ*u1o34N-q~JMof`M;R${n0+_Th(0D={%$AH zD5vhnj_{4Xki_duV~I95)YQZ+j!K58(qxOG6QiA>r#!c^hV!DLdqy}2Geq@0?6fq|`J zrdI700OB`&C&EJh~tpBNsS zWvaMuC_6~M>RM8lyQVcuRxvw0HD(}SmG5UO94VDwT3uDQSsMB7S*ztU-OA{rvYm;g zScxO9@sg6d>m?_9kdV;=lPBi@f$U%&B3AHiVrhAHkm+CoT)i%@=9<$Tc=xu|L$xKn z>g*|d$z)8L>_9Ty@u~0K#PM|+UWI0>Phm%qcdud?IYI#f0z6547%(Qq9Q%kO1FISp z^d!k-&c^jK=lujah+~>VP9RxURJ+a7FwdL9_k?*|Kuh<#r)wKp+%v9m>?ErB#wWAP z`?5EReK_)-ceIC;@yb(=(m2}P-%hF;SUdQ@{{UKB$}e5Fx2mGpFrB&Y?*#GLxLNuR zsUn0BOI&F2Rj;9GYTf6)@mT?RFUV72_zJq~bhpMzJ4A|CXBH)K*70pltkrMUM5 z4YdjcC*g`o>Rv`|OH+|;-Fxuq2l+@LhM8lV--d~5EBWneRMvyi&dAK;`m6mI&b}d3 z+!ETv$wun+bybUBW=pYBoCn%Yniavwt1o(>MdhtfHGclZPI?wC3>K>B9c?k|Q2%O) zW!&^bQNAI|2HX@=Zcvx!rY+>YBzasZW}xrR)FWpq_VDvNE_Kt&s?ah8_i+air$$-P6( z*Z`r1+!(rhKH*<ggD~UiwLl~yWt6g>#nh^<8RugOOJC!ah!$Y1i}Y02->6JW z1FTs-4HvJ{m3oq%M?vLxS3f1>iv8BTR7b|uG22Emv;2#rq;1thyKhcJAWVoS3c5@_T1R7K#pjqj<9o!5hZ`8~{&9i-hZ3j2S2nr? zBdefzK9D!t+eM=FuCWb^TOpCiNMr_1H+mtot!%4|=kE@5TAhY%nn3gzxH0K-^w;{0 z)xhw=5#s;Q11_Zn^|wD*ZG`3U@O_FmMbk$&G=r`9uqTrr0YXTKv7V^1pp1;URD#1F zhT}MDr~jhNgRX?HqCY<}{*cKPr}Q6caHy&{N_EdfEr4)gj*$&dnl0wv#tSU67CEa| zIz=5koQKK1RziW45dZ9l&i{4q6PiWt&dPMe-0mQtcinn}YJNJJBz}_u*Wo{SzdfJ* zy|Rc2%>;26@m=fO*o{_C@1+DvxgcGa&}>))pjgcU=fI zN)9Y?f(_W2Lx0)@nkG*7UhyhLSh$#85udO)8W?4EdGVrv7SobjZ<6WsDf}lcPyZd~ zYX=j_{pcI^QU;JBPi6G{ZyB~U=#86&*;8tfp^;e&gP+HZzh0;#B=ZSW)JpHLm$4TE z69P9&s-hn;_{pNmywy*~yZA&kPi>D2j|yjOX|8b#s0h~IfbQL4Y8s}lt47)u^rj6g zj-obq1uZ(PoS?zmX>q=ARP>uTn^e=`V2m^$<*8i+gDHV`QX<;GXAjqnvrmnvrn&Te z-LBs?dVV^l%L}T0+$EMcUw;>yl(nKgUz7;H%G~R2K>d-AP?DFI@N?X1QwW@s&Y7>Z>7_@S!cQCE34u*S- z3ULkqJnz)IPa{a|ul>w42u1Lom3q0>r}U%0d2^VA@z{;KD!rH%0uXfYbmz?65(Xo;_Mr(Sc-Acxt{zQ-%`?b zsxe0g5{9M^3JcExTiY>i3c1Hc#SqIC82l)~73cE7e#yXt%!OkAdVZ8UPkn#1jQ%gV zvU=;IiCBsC?{7o>fAKGx-hWXb9immCu3LrSIOcV5OOV6SbG6EOx9Vhp5mv)NtK6+DIN8E2y9;V1l9|w&;3gx z+ZY>Z)RoKRXh_u=iYqSj(sSBH{YJXG4HgsWze-l#Z;VQon}4VE`}S9pK6ioVmNMSa zh(qFM@WM!d=tuW_-CH%z(AQ*ac&}Y>+IFue(p>{LE`GhhF+BM+ycYq`J1Tgk48m_` z%n9u9Ptu}q1=RD{3Z?J`&K?T3LmA|vFJg`8zir)p;bb>^$R68UcSq%p%0RW6>k&Dl z44i>k?(^6^qKB-Xu5co)EE*aeO|?48T=mLYBM7cGMn$_Gy>ajagPh)QDG)gEIE+(T zjlo&6FcTR>WEG#@&R=ggiwpR^oW=t6TsV5$`R!1Qouh9@INWNqe2(wVR5C>e`@~9;PYa>~^#O3<~X4Ef8rNBHI9VlrW}ovYLYKDXZ4Qo5iB zf|}<|&{PHj{z$~FSVOnE=PU^ByhGx>*2|LR`s})6cMES+>sI;h~Yd$hE615xyv|lL#=ffQW5HgHx=D&#= zL{36Q09(EvRL`G=z3BwZ%rKjmb2rgdjj0 z=Ph!hJjs$wUZDbsj`u}0>eQyR(ikBw=jFIQiy5RM}{RLz1PojZz$AW*~=s{&e{Qm zkgAurlO77$o>KMI{v>Pv_&h~IE%}-DfyM4azPw4bxJ8sdoh!HqYoA0WS%2~;0#&*N z*~0w^mFBxDcJXUjt)ru`6vUNwqy+t*v`~8NicQ}>KX+&O>attqn1__3JL%rI8%A>4 zgR2WZh%)`~`Bm~_7bJ>5c0N+%BoO$vEkY4M{qa7&vxs`|j9JwO`MNaAYz`=y`EQn7 zhf*Wz*vLt``E7`ko_faM!Y)hQJAL(D_q+!`TOU{D*BgjWDdg_HP_!N!RNo^D5tO}t zTB>g&vfj% z6*JwqwTfemRT$D1_G9!tT(wHe@=Vm{DBswm79K19G|a+4yXl@U^y^E|UbQg|$;TSY z=_(H~G=(wQNguKJy@1tD448QKn+vF(D26 zs*2|B{t|{Gv}y6hB<$&w=!M$we?quk=&FWogOSzqDD=1C&%P5bC*7q5qf*Ses@Q=Q!|yf1_9i^v z7g~I>wv2Kuut#sN7S?*@PTIxmR_*0=o&&FHAFlcJGf%m^4Rskqa&x)X>XV=TRJa$R zb9-*;kktmgZ`_F0Kh~$2()B;ofnB)4M!!>J>ksBLotCmBY&J{ZK2UFMuJjdQf?y83 zU*6ewLDW2#i?1tnG&u+EJx%buHF^#dZ=pl%YgOi5>uG*Uu>wW~8H~^R6UxGP6$1?w zmG2q#@%UFe4@GFOo&(UhsI!qO|G2_|70>z7h<%yq5SP=aDg`_U1e<+k*R<$gt@e1b zeAa}c;$-NZzrf*_p7NfsJrF1C>SXP*|zr7egn^mulc z&&-Zb9$7O;Jv&9Z44R$;H?i+FvB77VbveyPf%VtcJ(9<*Yxa(ZtE+O)#BAh#nP_AP zOw6_QAG6NM__VEp~cb`~%2Q}c7$mEe@)1;s4LW4#)@90;BRm9M6CPOa;)52v#&e$``=$B#nx z6f9hPKW`cNvr>o}bl~iUa&6dG=W55nZ#Ji``=)eC zZO?(FbhvZoXgd9I44V5KIN-iC3AwIiTC2$8H*i6Amufb-_(?e!Nl+<*>7IG^KgoD38&zp_)EKT z#pKn#!_l(?`*UCjuK#Ts^vs^w)tZt;b^GE>9!@2H+VMIEf)ji0f$xxC7wA3t<$lxy zdT)_0x%VB1kIT@PR-8eMnhubt!E%Sp(T8K~#*2lkL7$$mvkcv z#Nn;_!qq>H2aLO*nOSG-l|!+3TZBnt89DSzRMxzwNzU@;z?Evu!vjo)+u5G*kj19! zO6M6mQD5uip^wXHyfX^^j%K3uAj`7ltYeoEp%HTXSGmO5fdaVue!sQWyU#M}9dQ2d zHw`jJRg#6fTy#;VICX~1eO1YN&gLq6lSQ%L&QIiST4%S{?&0yD#`@As(k50dlt z^;;-cY@GCmBz|t&=^abXNei3E>@B59b|Fhlo0D9Wk_NxFbP<3k6k2zn#hY2jNd@k|49h zAXH>4*j9n&lR&Gztm*Ug^is08Xe7%5tln}hcc(5i8}!uH_qP1_yQ@u(6D*NQ*M>zU z+fn$Avl+K=k23JtLJnmRJ5ihKH^EiAz{K@C#4J+HgTvj~{FR(Z6D5t*mSE z{j)m(DbKW{&jHf5Byo)P$=qUW;Nw-vLfW)zbC|^UI?Dr~69SAG)TNTG48K)rF-q&& zre6=^8gi0DB7&w!Ms5+ba~X2J&YQ2I#A8z;d<8@TfOU$KUa?7gLgtpC@Mi{yG7YNv z*8GW3rdmw0O8jC~HpWTlp+-AAc&h5!JD8TCv+5$<7h!Z~w3x(T486vK zQTIzYI3vn4=ub$p9QVuM8aRwAw=L`Pi7&NWklB%%A4be)-7EnE*m=K-g9d#PK!v0U zAfjU6P(XR$g^E+4_{loj9&YUK*=0@&2xErRzG$WvXH7j&Jk-&%tP?O~o8g#ha~j7k zk3KU9>*iSTVe&7~n!WDr5u>MNuTc~zKZx0c9L9Ti>}_zyBaLEnWTrHV5b;T|>7xbv zRRdL+Ybqom~2i*+gzjTdC(_d0Oqu06WR z%)5Mz0Ug`S;Lpz-0Df&p0%3m&en(DHf{(@MG|lG?G6>j*WjsY5kYP44Q<*{|^`?(mlqp93_7_uiI-NF-|OyIaTX=B1VS zdm*a_H#X8I4Z4;(#JNpepSZSl2>I+vZuHjni!aYT?q8o7Rzv#pO6^HF?RAXjwxHqL z`5*4oy}Q{}9+_`S6YHKk5tnJ=*3qrN%&;J%P}KIYqLBh!ri4fNLqG)nJoxoQS~Nlu zBh4wyF%xqkU52ZEyZ$59Byc-Mwa9rVbaMtJnaRhJwQidrRbc0N924cCt8fy|kL!Wh z?Av5!;|T&BTDw8Og^hPy0JbVZmFQa25v4;QZ5D&|NpU}vdz)cJ$lPt`Nc(ME$Ir`#H3f5x!SO3%!; z)04b?%Coc6_43nb-OoI=4TaO@Ujt;5ruSX>O^1@a(x*=2T7?SyB1hR%9tIu?eJ`jr znX=Pm9}@X4jl#MMlBq*Zgq4cYcxtnCjkeFYKVw{Eq~h*$I;PoMcgwgf#g=S%`%DR7 zbA)sabY_pfHksacag>nVGp*MuUr1Qu9(FBCg1(pt-=x48wbd;7FFti=O?k&*nV5SZ zx}&gGa9f$|Rzd81_|tT=tYS;caHFUo9m*f7QSJA~L`5}tz7Y)~>YC}lF<4j+LfE70 zC>tp-q0}+o`**o}9mW#sUa>olL+h$N%Ze*=k9}l8BN%&arf>4Eur;r&Wht$Y6~2n# zGxlo6J)17Cmp3EsE${nVoT!T@PWq5Nm;}G+iDu)$OBKeekgATOpC*awf?RfaUJSy*MFWi=;J8}Y6ewn_Q?eLB=Ue!nhPhd}M zv$V@m#x!1-Y?Np4;;(j;EL}WZvB%ju+GOZ-jQqGu&>vbd_U+L?-mgABDl7Y6$nCL2 zQ|lUKUiO~Y?=DFxldK$t#}adfsdsJ!19{sY_2YXA`!^V zM7uiIhA{v06F)^x#UuS?aw?DKsOp${^;U80YEXV{_#^85A+}j|_dq&Nzap251VI1z zLbJ>;(4H6)Rl5#&Z1td z4~!#E;547afati0=P7`bRhq$@HIqzvx%)?X;8}_5yvKcd9+Mpu@-mUqH~@6zI5^Ta z}f={yRb5qSmk<(WK`*p2=9j>zI7h%IRAeZ*#w_@v`A_7gF-?JFcxh4WZ zLgE|mBV@49^!W$b16D6+gbIra`O{A7pD>WLEq_Yo)jaNsHy;#!JR;}@nlwvE5GF(^i*uKC^dz(UvHKD zEp`>(P?D`)r+B-a1mS#@I%~T4!#Sz{4b^4v%50elFMRF&-eMd38<1^ zpa@CwsV$2={V;u(`0Yhq9M`$G30VgZiU${PcDJmPwQH-qL_6isJRgpRACq-LbKRGP zaK_5*z}-@ZOGI-!YKcfj4KysuqBm69lxmLA^Q%*UPHh$#xW@5eafM*!?W~J`o^R&NEUGtr#S-YKq+=Aa+U$m*y`I&gbyMk3o z>ba`YcmG^?_F(f9f7T`uAo<3*=(W(2$oc zfgJ_1CXC+U?-^^mu>|1*;{j69&&J`-pKr~IT*%rqw7@wHuO|_hBi4Dd@Js9c>1(DV zeN=npz*m9Ln+(U^YO1rd)u>SqVkvRc4uTaO7o0}P)gmeurVG1^-iINr;ntk}*i#+9 zK%yfvWA;RY$8H5K(u9KwQH9U>Y0!XvtlI-xn})E{eBn#y7nfU80LBUEbazbil?M~i zD<=1X7w?fL9EWhstZH{u)tKr|_~lk-B8A2kEc%CPMm1ngN*KHCSu7hjBOKeq=q%4Z z;K!(`yy|g~?^(4!&v`McLUkp@bMuDr0c-QEmPUfb3e{JaF@2GL~}ms6*D!^Nsfs{d$=C;W)B^tWw$I-Oj7U5DNVJg zg{1HIT}5?sB<@CgoGYz~;jjIXC=n|7OeCYef|}Q#E)GE;ws;U>rsjh}y@BKJy`MnG zHK~PVHw%lhYEngy*4=sSO~4poQ=VtVS<^ijl`YC%n!XncJn!H$MvFe@xHXn38#|A! zn_AJHm8xp{6b^k=s}~n6gc-7LJWbM-6nXdBWjT1oDm z18&y9uP{1>T|@M~;5%KYA?VN7TtZ#4|fX%vA@Hu}73=mOEh2 zEDUZbk(FdQieZx$Ns1ze$_Q^1}{0cJ=yk$-COMbpcc(EZZ>@Q;7OL zWMa9ietIY?^FS{JP}jt^qzG4qbLPr3!5!S(xXwG+_styYyhkw3_`RHfiqzR-2%$az zi5ZgR!?q#cU!EM3FYV}5P%u%cgcj$Vu`}IJU-3xN4`N>SDcLI&rNZ1+*FC|OM;clp zwFAA&(*suv+8Ggd@-&L&ZpCupTm!b>E7`w~fy=Y>6247#$Xswwz-@_d!QyeAGrJVhXLN&~8;4_M zt$z)5Kh8y77^iA6ggkd;8Taf(SLIaH!Ejr3?d_I$WWyqYy>Q%(brZH);)Z`_Zd9J; zrti1!BZkePCi>)(Qc=d#{5&vspnlJ+Dc=~NGfKJ7T$uEgX6?a8q1n4Ac^UUqIA)eR zd5G0yZW;`Mx#w>Yc}(|tmt=1h`T-O>>hy;}q(y8V>F?G&8|-cNi<#D@3a4L@{7#$O z9$w*(`JD27PQomPeFe=7OHvjlTS*&j{epv#zjz?0KxwhlxyW*c8m^BYrF>4 zy4@cA?n3e72L3_i@88dCdky{j$pd2k1@cKQMdr4YBv8l5(q`G9li2rp1xrNVvNGiP z&)Qlgm}Ok~GsUvmgJsvJtaA?7c3hFQW1;gs#eS?>iu}?T(O0zB%hf=yCrPeJFEq!-e@>t5vw2k!CWn9@QX-XEGKMXtM9 zU%$QkeqNO;8aAi<|LqiXMn}18&Er+4{wMWk&NCZz!3(bSz8~FZc~vd6oWwNW0W1cR zu#mNUPO>-`B%fcAd-_4YzE=DJ4Kv%m%AdohKAWR_mKlK5{)<}voBH51>2dcnsTrc> z?T`uT)wqN^Nvcq7(-5NHy%_r`SGO1GaZW5ZmwKEsZ`hJ6_FxMgPhS4LZ`^#O98BVl zN^lnYZ>|M|_8SvQRVwlP?@{Kb(L+qjcAKxcbKXC5ut?;VM#R*sAR$<6d{)e*!>MR8Am9#+2Q_iS)*k#h3cghBn%cKW5XG^K6_3wZ}LbB@>^rKmC26 zf4ptH%=Eqfg%GJ3ijED(nPRlB$sySo)&}Hju}C3NmEy70(p8#rpo-Ha9Ke~N3>y!W z_Szfk#K9nBm!0zvxIy4}fGlCJE zcVj~eWI~PfB7cD_%VK<^j%dSW>!!?Xm6(nItUeT+H|QVhm*^JOR)TEnb&SvB6I4Sv zFJ1$#`{f~(lat8byLr-j{^Uf$Cn@1>6J|jAZE3+W{MIbhnlDFDT6H^;HaON(u3%c1 zo%iI`kk*?1oJkZ%#AwHJW}7t_LfFWIuXn81#LY^+=>4IjAm&c3u43 zD6Y%oBc*B88MtE)zfu@nWM$oX6hGR`M`1W)1uKfe#@sX}YV7Ay9Y%hjLq!HEPy#Kh zbQ!AqMYdiG#-WD`ZeVb;$xkMz!P9=+o+zo~T43acSRDkYg(&^;#@}JXeMz3LzQ<2^ zWn|RRH8GMvqU$`^49mTyU0LCSGtvnvtDwxiaC3Ggj63e?8zHU6jDs#VOUf1y1mYiU z+$$^AO!&s(%0fL62GQV(D3uRkZQv5QtHYd-rZ(|9<+`2KgXjp!Xu(7xMtcfdk_pk{ z=~iL_GV3Qwhgq-4#xI7~DGb6rSJUGX>~d+8 znyLJGtAa(|I=oIqjstPm4N>5x9mtL#HG6QyWOv7q1M_Fy+78tTG}AuB=PUHlfHXTKl6|CdQt zN@>kFjN%fq)y?edu@Dg3vRoHdkZP7!XlhDwo^;ZP2XBd%Ic2OXB~n+!_{C|;S*zJJ zFjk=>-LNUV(qVN^y#PHrLKSjyFhHz8FPf7$QeGZPP2YL`2rYE0fIkRtnrj!JfiHaE zS-LNX>mtOx)XKsEd5d(f#_LO@%J_toV><~S(LMa_qoVW%xoULW&*Le3BSH;!uy7rl z9M5R+q%3a*3tBUVy>e(QYW^f0^X~;t`uT~XA@+K;I6fO4AK9cM4U`C3jMCn+yzQAd z!imXR7P}w^k>k;lDP}xj*OSWC{W|H0>k5z5(qbq_p-0jeZ&(8sWQJ*um9hTE)ao*t zaCu~np)@2vg+4N`-9qNC@})|6aqL`EZF-O?L#cT#+t;-Re!$j3-CfLm=v7dz5b6AO;VmOW>>yAPKj-08 z+b%U#r_So$yQZCGp#~eR2s2pNa12Y@hL~H9yb+YCo20S@`o*TdfNLJ!>I1JCt)dit zCGSh3m;-ue=XdeHc#D`Q6@{%VUEKtR1=)M~fYqr>QG z?(}+;Zf;OR%AR7$_N{?!BehML(@4%E$B}Y}Qy&+rF1%y7@mzUq{MJa*oLIpW09PeH(unn7OM^s$&t`&YpO0hS?E1^6 zUs&1-T#d*90JohUMst(=^RR$a7KIcMI*0vcc5NL!Jq5 z@trrnTeOz3HQR!VSAje7762(7e3EM{kKO$R`cUu-B+a=K38>y9fR|E0)I{D>pI8U{ zphTvF07eIZ&EZ&rc1)WA9M3u(kc{yV{1pQ0fNjll?u7jUf&Y0$0}l5b<4kyv&28tp zqj0?=9?n%u)L=S0H-J!H@M2(&DL@T7@l(Vx;JS7yA7iL&$8*G^>0=%6B(OP+<>?sh zmgBUFgUpi*N_ksa&*X3uBQ;&wL>*^&yox`XX^EioR_9fcj~(l1Nxwi$)q!iD$-zv`W~`*`EYBX)VE51a`dm?-wY-G0a-3%huG`A=4r1i_u130P1h;ttnP_MHV+@2 zD?b2d?AjTvpMrP3t+^pGh~xdt10bn=<26~MJ(MGvU!ajCd7cLPlOThJAyYQ9S^eWn zWz>oElE~rb1f0UL-9#Ut2&mT!derFEwXwp-{i$L@_&oiIImeIHXY|NAYa`u(LmoC` z;uMDMxC4@IWfWaE>aoIIOiziDneJaVAA903cW+xh+;}-I$^t;=+0QMtJ74Ln0KRed z7qA)B9F(uCwhS~glCdpr_AtzAO-ebUKy)|e93u_9o+9nBy>xJ1Jb$TotK)Rh(Zv$P zd0M00Lo*i_THW8eTLk&ACPRHIRoF=g?>B!C2*51A{f^Rrax(w&@6{im3z3{&_FoGq zX&R(W1`(eTtReXCblYirU9?HoL=l^a@2la%&rg5X|6T6L->J*M(%cJ$0X{BNTKoMP z;`4doa0Qd{Ot;(-{mkd+vk+sQ;KAFQu1|?JNxYqpk`tz~QLuGhz_j`#3(*?Gq{P=I zY|LPY62whSwnRw2UNN*$$xF4P@2Al_-=)V2Hm+&HV5Rx3-<3B>_ zoabdV<%*&rQd6@w$MrS;;QdeQ^UsA{wM@-glfNOQ%hl`@C%^p9N588uy=tQ<$j$ng z=(@RJ?*JLPwr}r#r!Rk(6TCH$GJalM2N~(wBKIl!guRnub(37x$Bm!wKv&-pK<_Vn z{8!C?mVb1``YxY*F#fzvV4EQHS^4Wop0n{s4GC5jfaCqU09qB$A6+3>K}89G?v_fE zeJ+$Vtt7Ua(;LWy^ok=DsuX;Su>GKoRcGiC*SYo59$m}iJln)Q!~A8;_v7FbAO3%? z4(P-a7G>y6ei>mFI;SS*E~7-2 z6hq$@<(MEz1XJA0jmM>BQkkubvLXdEt@#XVDV;U_%$e{$^daU>GS;K$9%Rs_h2;IEY;!P0upT*AdZQnM6z7Ewdtf-$uB9x2lv8)0*ty7iVb~MW@xrU7xDRen!RGfd06xaU#;Foz!f=;z*wl9s_CV#uZ~fsS`c4G zfn9b}{vj`UNbf-C)6dc#@AHb$){y4fJeN~@tM(Q*x7uQTT&LqI@hH1W>m7`tR#7S5 z4Sem(EPdJU0(iO31P2utS4&anMl7w&rTzlVD(93>v*+rfYtzrM>OOfs79YN0oEEsI z3p_k4k@gyWC1C+32i((GyN25rvM_)f^xCV!bd8Q)O1*P;jW7?f7guS-nqr;ZOWAiv zNeR%vW0jf3|sb`qrA+dhXOmA!TP>SoW(-w&vXaG1HfwU#}Vm6l*!N$ z;|H5Hgj(Co3fm@9PJXB1;?l@?8{Cxug2++h6$BD?jNR=yTyhd5Ta{P zvKt#PgnU)hCHGl%r4QZdH?si)ki27Wo6S-0WZ$J)uIlv4 z)vS2q=R!&XGqLIcMww1q+~b~aF{UNga?|{aXDD0x!afE@g_+uh?ToO}wFkdI>LsbP zSr2}iZ&K<@NuIZdgmx&wuSAbR%}mGjxYU=e1VA9T{F$~2QMYKA$IXja#2HE4h2-U% zXeF)o<_j-h*GXSgeC7cHm88A~#zV5#6R(cFx;&yllMpo??l0`gP%6+LV-sYjbgkM| zUvH8Li7)MfXQlgSyO`_iXWul}ny+H!d|m|b(AZ;^v`dQPlFNGGGqD}SpFPRwuZ zqD*s)U|r$M*)%DNI3AHcxT#cm8L=8i=sFaiFxExS;d9JXJ>P38KOa{-lVp_HPS3sB0({ zl8H3El~t@wyltu|{;W5v`Z?2W*S$DaFm5n>fTv|SkRsK&!Kc+F{Dxz-@ZM2pAMqzZ zaMO+4Y5OrKUx*1zf8b}F_K3f7DVk#b!Py>`R-?)@oBDVUyFLYvwKTtwQ)S|z1}MB`r~YA>1AB^Brx;DyVABIeNz z)~0VZi))we&_~lSTFW%S8`eSGyNICEE*M=r!Jz{7he_VIHB4Nyr5C;m#sk2c{8b*- z`&T4r0lQDEKdz21fpKj+WNMsU+A?+G<8bckc%t5@{7Ndtj@IhI8GrQIsS8P_=-+AB z>-iz2d$pOFQrbiiRL>jE7#XP0LcD=0iQ+FuXICpAM&W$iwfE-)v3{?<~^q^{)eOwWW{Mq$Wmp#xjR}OVrY-(g zdZ2xLKpifVLX$pLCcpPu<>7(Pd?2dNkxft;N*%>KjSZ)Hq@)wP)E9BLc{duhnGI3OdXfxkU1EUmy)D7Ow* zqV@r9S4Q^GvTPijt-YNDI?l1i*yFw1pv{2q1t3BAg}JCs&0 zW`E48A8CUX?OHeNZYo^sI~V-2FSq=!MUm?DvRrr#UwN9mq{t7!LhFgj*_M9YRbQi2 z$sHfJx)J<}F+g88lnqel|ILlK|2BQpW0)L3*r6ioNe1e&v-q&tzU5yaGjoM(*%~GJ z;=IR!l0vs(UPUag*5WU%zzHP%hx74YN4``_d~3HEt#N>O8eF3dLoto}0Z2uib2kEa zMmOz=i6{Swt$@WJU51axnnMIg#vPq$tn9J>WkB+WTP#}TKlT3CnmOT1_K>UAk4}i% z&7c1Yu#goXX5`(>8>PK%<6X^aZ=Aosvqsj~cE{LhCu4=TK9yiB)Rq|wLrp}eZX4eftgxuG zPH_5m7N~UMUyw8Rm4ulDedG0$TxlWAEQm%On?_ruD|7in+lpA@^NFdtY=LGoL;N|< z)973shrFeETpUJkHO;PslFBMht7CvMbT;VhWwiypZ_FSf(CwdtRy8tiIIbItZ5qX+ zfIJ}cCIs>*%4tVsp|{sW+SFxEj~qN+ui9g z-CAkszEyiFN?et%PUuE;x|O?aUr*^CkbyHg^o@c$=*=Fz1+uM@Y~a|uVWOpC^{SE~ zr8255E8X27Y2w{ez^C{l;PWz(@`6PpFeIl zWv8^S4S(Zx-xwHF5_|bP8fPgw9G{jwmt1ZyXuFHGT@&Q3>&|2jFxP*(igy@5xZwnA z@jT3x5!>oj*M!QAl?{o-i5aHwmBOG)Z332nr60@q$g`dpMWw`t8LF;LBEeFkX;$NO z-p#{QXCHmPl>lpA?!*avw zQOT6L^H7t}=&egbx(?0yBy^sU|J5&wLXwJyE8ensQS%+?A-?w${eN1@m0{brp|=J+ zaG2~x13e%Cq^@HxFwIl6v>e#Z^-KwW#etA`MPm%3yjdpC?&v8lFG3amvXI*u!$Dp2 z`_xj;;afwuFK(yR8H{Yf`h(6tiBooP_a+@y_>iAZ>9U1?}yHH9viO0O=#D12{T)16^?yq0}+kC3+HLdk~oU~ zp?WbSlg>NSMw*F6sg_neu5OR6yU&49PodxBDZ0UL^qK|A7VJ2bc`6~`JlF6|c8TVQ zmJDyjHmzujS?&$;^`wAHS{ZN8usZNThLy{)ioZZ;<4u=?;{3z|AN6nP@kWhq)PtJv zLAio-!M*Sg3ji15dFMITJ?pe^XxAR6N+c3>sp(gnEQlQH`1a$ad-I-UeWSu^|hi?QN8mv3H zPSNQyB8Qhcl?HP57FZRLwZHBl%ys_OXE^a0cz*J^O`G**;s$UgF^rDz2; zdLetG*X*S%ZFP>4%^$cd**{LaZ4_mCFf-FxT~%@|Pu4@$OWtH{s?Z1KdQd42x0aTz zcyO@lpj+PAK{}Y}K^mlCxOAvcv0-PA8skTh5^=0%9uS4TP9$*YeI7NKY3iV9k*U-l z_XTnk(lcU6*U6g>r9-P452UXl0#yt*?G+74GvPcFZ=xT6Ag{~BBLY}en>qH>1?fz( zxzG!vU^S`-Q_j!DU4sk1Kn+=gv~7HGnpj<9S`xd8c$!0UE@uPIYS@jr_^1TVO!PKY zVsLgHtfc>J%xTi?-;3`F@oNs1htXjcl)q~`w@c9tHB>Cy9#$)-F1|i>|Kaq}u6pNf zMlI(#HG zeSJ9bM*S=WKhe@RbW{ziPJh%Z-{#%9!>`ku0XtJC-ek==;d>rj6xlVKSkOq&-!Rp? zQK3fbEStY(Zs6251*cTI2{?#j`!mf?`xNh`uTOcoOw*it%MPU{*l!2y>eSZZI7W_& zf*!=8YltAb_$JiFVcqye{tj|b(sV9OoTn6%o`l!l(Wq@;>fhOD ziU~y%SVt1NGklsWV@xnEZRxaSZ&g#-okwkp+wD+kS2qB__t`wGQKwun?F*#am2KF> zm|EnIZ6-ob7P7BedW`vgPW3!~RWZC;CGNpQ&d+LRsfN>QvBG-k@v$YF_HQXAcREuG zE1enH(xzY@={|Ucq3$@YqT%A)QDF+7K$z{JhPL#^$jDUKu=C_}g`q)UHXe*K^K^dg zRL0*?ID5w+Hh*BwO5zALn`t@pxG*C;g&JMQUG;_lw{6`C6RdjxpYd!OzMo==KKMKw zs6yT}>vhEcT;lxHz*(-S-p2}{!158^rp3ro#hGfrR`g+WF9V8 z4ebReBpzT7_RafW?CVZ{E-4iEoHKi(=Pdl>6Nh=o5l=HH6XXm6dnB!71N`_*>khvW6ytwQZBqWlVb3Fe=}#)HcP zRaN&-v`Z%`r_wLfadp*+UUY0h&BTxar=U;?%huMFn?9}cHaw{w_eEM)OiSE~I*&g4 z{0?tL>_u}!4i(H(sBXR}U@D%tBKhu#e%FAc+smH9qAn>=^gKiTFdlJ0BVQ}P$ zU$QnvpHGr)sUlQX79kCBokKY7!|(nF%s+_*6sx~uEGBcnnWjbsa=W_QWzNuWHS#JS z^3NNeftbtY>7TBI zOb`7Z$Y>hzIb!$zL@(&Px+2(P_WH?*KfKqCvSzhSdVRd>p91O3FEiu%!1@o@r`!TU zcki;sH#qQptUkf-8fh<-budZiH`7Y6NO`4!#R0WIH&><=xBLP|6!cqecl6vZNAg= zCEx1!ZLYrAs2D*d`7lZDgYCPmZeUH&)8G9>`^E!*(Jax4JAF%(YhTClFHQ@x0mEjy z&7SuOHCMW^)=u!6K2_axS_76^W89yyywH4Yu`p8L(j0Wg5{vE0t|(j*+zkOdIT zD#Q;Z*kYzV88)dj>FVMnQRij|(puXP?#BM+4E1a)SJuOmf4r?`{?qn6PRfkw?%%;L z34{SQJvnsQ%|6tJ5R2a-XXeS2<#;m8cRbv#8<2yVDa9H{7Gc-amb%$ps*8%--N~NzW%qRnM1FZ?v>sPpxem#&uo7dmDrR%Q#MEKh1ivRqO-XPCHj9SW4*EWvWe>YFF+Y#MsN2GFu<|@JOyla#il6_=u&s zb)2l4S+J84&dcq1{-dqTpOIks0dEE?T5%GT(ooWJ$&k~Zdx9GQldy+5G0G8efu0@R zT$yVImAIbBXzl&m#dFQRlD9mouflAQ>Olq}w?>DD{Cf7J-IXDUxJ?vG>g-#$m)NENf&9`dj!pxQ&$HUZ=cB!_8uJQFSM+a>unTW zyyiwm+z#mLPk+?Ct#1A*A7|$q&@?ix2)%TE>VRYX<)KIc{-L|9u3*=SmzFe?T#(!B;t=F>CgeIO#YV&~c(0trzGe@{ljPtTFcZhaU-&XBd)$qJlJMzw_NX$~#OrM6 zrR>witqp6Jf6UVK)#q`Y6yDak@#B}yiY7^?KA~5fE!gi~ds_cMHbmL=EhLWm`dsPx z*F=$bfWLK?yBL~Nw=eflPNI2^X4o==ljZ{QqO0!sVmW8cm;ktDQTGD4=@w!K`RFij z|NQh|p1G9nuIQ^l)3JWt{1pci-GpzMN*A4uFO@8&-u#}drj3UznG{^Frl}2fQ|j+f zHpa`l4{MQjFidxx%5aa#&pkS6es)R9`*@~5>!9|{*JI>{`m&zGf@Y7chig^3V9{Zo zTkj$;DmJ-?6{8E8ZRzLNT^F!Dl=0c4*Vh5FDBvY!`uGh3g80Xb#T?8bY8#f_qVul) z27wQw&0fN=UrVcd#6Gw#B%caPC->w}Q*>y%;v#f%RfV|GT!H#lEbRKGGCJ?=kVDe( zOVRwH23WW}HXX5L)oe0o4#gM`a_<}P42%iXUn-8Y&hcJQiEw=PLy@P{SwXn> zUP~^+xb92j*YOMi9=P(P^MihAeRQT-zqYZ)pcFpuhg>#dM;zb3OE*5eBf)Wde?Bg; z`Ov{FO14J1{XvcK-o4wB1nY%Qy8_^nz|%dGXf>@WDh=OsgB&k(bmb?pc2el3qMZrY z8e*C~fe}gZOBJPgzuicLrvIhU;l3C9=|{D4Zg~M%v9>U4qhmSMqK>*A_XcwuvvJ2A ziyM1oM?Ku4q?s!oQ(OX?cj=`B zIarFw2RG||n>aL2&cP`PdmrWuv@?|5wlFeT^|k|pzpngf#W#LWW%F@`&yn#Bn0a7b z3ePO~vJvjJq9edHY zle$a0#PjNQow#?h{SLWH=z>TaiGx_P$&!Af0Q81a%VZB$1H57LH5N@ARK5aLTVrDd z_}w=vg6;3(SH;PtJ8u{sXu_;K$Gj7m}m9k+xtaE&@bigr(y=Fod z7!oz}S9BepPp3BRWrJVZ6>^sO_?lZCy#OG8k2f0AT6GJK%F77tUm!`tqVReNPkV=5 zb%xj8XIR0A+R*@Alz|NjKT4T66k9xwx7|WP1lB*e!ANMwkWJ#B&=pc`;xc zk2Lghcjp=0@J1hzeAqseAu{Rs3$$GIEb<6{#iTCeuc(&gfl!N@W3!#0)2MARjyz-f z05xty2A}^aj(!oIEqa9Sm}am+!LU?z4&qQq-4@%;2mx~4PTAePkUFy$LeH~(tAE$-miGkP9i!*JS2GDCMGSuA zGMc!)%d#Apn&Xf9j$S6 zu;UKTO^(<{Ho;V`Aq|?wUW~@TL2+kQ`qHJ8@O1)a$Na0>x(k=b!s3etaB-AG-TlVW zU!Xo^1_s>3k14JDV1zN5PYO_{d~iFqlsgJCd6gRw-w)q3QG|&^-x8-CBOLewm34b_ z{jIvYWApTalH!uQX>|+jbqcF` zPTt&$fyUuZH35kLOCc%GICSriS&96@m94SszIR}a-rDN{#c@%V)UW_w{NTVj1ONV^ zzS)w1YdyBz_#GXHQ?5#~`FE^Qy{v{Sno!v*BY)L2@Z5ovZcu+(h5c+k8}Rg~l<7rBcE4yBqw@HVFCfr; zIb$R03T_0-r<0dvkXHUD5N`ihIU#@pF_%qD^f-&~fjwp#)vW&c@Qi_a_bS|-oI=wM z4!zWvGMZZ9NX~i+iu|vpAP_8ll(BN=4A?0#J9*b-81xzhQZl55;C?*9-jFJoNokLM z`mfEMuo@74xHXdlw1zmz7JgMHqWb=uWdH&t4*z3seflY&rxkn+WPi`Hd_4X0XZ#Ho zh2H~~e-|cT7FH|G*@lnTqK$>)!%ZIFPJ2sGG$x%&E!G8c*1t;^=;3BO?fc1MM{Poi zA>eiGmvo<3+dO{Abl4EE0Civ4tjJZ12RGKq@~uFb4oYs%&1??y4?pJCcA^8r$rJx{ za>gHY*U`!F=ZJxSHv|nHKs^w>w8{giKROKy+;#bB6Fon~;; zzKC6mT9`9Zj{T(%M%{G4*sX7IGa_iAzr4`WIGpE@@kdRD~T(cEp)RprQ0 z=sI?_-v2Pkfz}%XwuYS@)>bbR1f)Q`)mijB5z}8>wc6PKDSN#Zyt==N){g=`>=Zn`RcUXa$gfr)4cnHJp~@uS^tNz~-153^>2(XY!L#yHj1_ zIUmjCq_CQ-h%QWVOzb8EM^W-<-ZT_TqOq$_;7j4m0J9c^u>a;k$>@$d-oqg($xXLF z+ZFD`XVr}}()Q~pb*T8%?z+1oX^?PyPTDHZF*;HGFP~2rK8;+LxJ&{Kop>;TD|_ZB zX!W^p@28_qg;nLMyL$|qUcZ3f9Z~g)YQ!&4sCrp?oWhtL*y8Ap~3FD7QqdO=25>mG*R(-mx^wTXrwtpIe3dYiDn8MptrL_j)`0UHVcNlda#k z)PQjqSmjyd2#2gi2nR%E!e#VzW%%`rG+*s=m_e)F$Bhn+4eqC%S3A5%O*J*U+vixC z9N=@U24mvS+LGO~Xa2XlN^XJ}t+K`+=j>M;kWTtm9YmU5HelXpe<%PSK?U*v~c$+4zHZ@K|KC=@XCk zU!XZyMls;YQp;(tTRI3#J9sop-R}*$nEpK#brCE*dp~QUuEu@O8JT6Dj8W9k=v^%Y z_Iy&?2{z?tdUZKGN<`P*>_@L!R{v^^37MfAz&{_2t`T{I)y%L{bwu*=Vzy;>G8$1R z7FV=S*`Tx)z`77-4iYRY9OFAk1Y7&_VHMSk287{OmA#p(%7~Tkt9~_Blc;#%A%}kS^_VFz$`5b|} z=9PGEnnB98D{0=|gg8VV=5Bu`DT`)(hj~cy(r#4zrgU7Z zvi3Mv@p<_5fE}yrz1G5K?l=<2va%V(!ts-=Hf2L7sj5s4Ei3Td47e`igwjn5cqks5 zFX%&B++IB~RW;9Podu1DZ|Emyqo~`7V5`cIA|W+{(hRmeq!@6LV_VFTh(g)hNxpGO zb(%#phg|qU;yJ@#pjvxL1Cd6f;2;H@1Lmz#5@)MzDG%3yUcfG3UA6gkgW(Uf2Jg8a zqxkmok`ZgV0bvmy*Z$GK8EnXtgzuLU;<(TSWe*6D z%PKwPQ&mZlWQp2KmL2JhGL8YpL0^w*1p{W*`@z;Gs)l@~dL^Z=j%6aswx39hTlJR? zC1x}VU~6uP^#hFE!0F%QjpWROqa3PFvr7(28C!bPmgae;2olSC!H2a&t0pr;cpZ5o*AAvE~-lL#!JlnvGEAym% zmpGq0ZqTx!p5}gpsWe}fWkFLJM^~e~^55v`bsz_BCPDDtPu--+q#n;2)Q5AP~x zd}rFDQPT?A1$v5&WZw*m7eoBqQJIF8f%XSdFqj^uqU z&k?UnCj+LZ?6!)f)8miI(||nvjGw2y8~7dxHQVHL=vq6PT=>E*4t08*xOoTmUcFr! zfLWaC9FHM~UI0w(;P$VEDNcT#a#i9!>G8Ld0!<`_cK|Z-OTXVW?15Z2X>*b=PPCF) z=-Qwu-d+whj z!|4+kEcv*zi+)8@9 zo}utu;0uXWWzA&odE3@9nTYIMGF;(Zp7;pTkVx%VoE+IMElV!)(X(>untD z^K*tRsdNW7lt|T2gQAJ3(LCF0I8{3Grt9=xvgGXx^^(S2RkbDXorHl6sYg{V%Z%(- zX5?Ha|4sxd{gNT4=o&j4*qT5GADK7m_fQn|)7bY@{Q%UP!JUbMvpbYlY3C9T(7r|q z(>)qAC$0WMKD=jsE#JgaoGI1v0PD%Nuz$RJ&gFVSR-u?f>vW!`tGTYLfL5OBRDpEw z@j_`p+CU_JBHm+`YqAu;?PcrY+JaOE;3uM|? z4I_=0HUEh!|9^`U@o4*RAsOco(bjwMeve}=M?usl0W;%wIb4FJ)rmMJ9h6|=jo&$T zz;-hI%RfAh^Qy0ZhtcO%TacF0tEZkyvIp0E|eBo=dp^Mq1EOCFs49>8?(@Bj1)VvRQ5{v%sO2kuqOGa9GfRsnv%sxA~+ zn*|*GzbiHTAG+QKpBM%d=*m zo&SgX&uXrZz`n*K<%+;bN7d*Tg(%>s@CX%QZ)IZETm)&7&g4DW-c4lNFhQX?CZD!T zsej_DvK=N1^9@qjrhuMa3q7K>FrZ5eJQWO(xfL^@d%`J9n@g^tpL5LmlNUuxCU>#z z%vrT`qjJ>ro;cBLsnyJ>224?r=r0<$jqmHOt1k)E^`?l7MSSl|O&4jEg=|V%&WxXk z{(d%_TlL};ckbioFMj~`I{;Z0B!%KjzH>)^N~qo5B4y!oIXc`~{ADm{D4@DFpI4PR z#i>$aLEsi_`II3DYuI`9fDU9odhgaH_xk$SxvpL-H!j zui>1tF+0*cBH%<9PzGEmy=2|uq>&tIG9AwK@8va zw4lT!-b?(b&S0jfkB~+%U9TQ;U7S!o>fdJa%-1@#%n{Rt)&1`9zM%3Q>ql8^SQ8v7 z_|5ge9Pohi+&3ZLV8&xc$YPGMRe;_Hg&R~&l=zn`zqB^bD-L+TQM>xBs&(gcv(RAUExr_z zB~(%Cdl%8xYCQeGToPdm3O7kjv?>Zf5)n&%MRQG`Uf&WVtT>8>OfHd~){Hjj->`&a zayi2f6hVYh&^E#lmSUdt=_Q|E(EIz`Y@{BUC$UnX~sL5?yqHNFzySKJ(G%ke_nlBVVn8hhx zki0He*07b)fw`WVT~o_WAxNrbhaR(OSUiw_X?JwA_}$&*9j`#(M~4!vdNd*WN7qcD zRz$q_jC{nFB03}L*zlz7nk~`65AXyZrh9Pc5T9A?rAWGKEI2ZtQ+e%HR#x8pJ6Wqw zi)1&0ucJZvKF7E*Eh4fwlPo_c4zwc+)P`$nwSE#@#jyI))=BH!Aql~T#>sGvWtn88 zQl|rov**r4Jp#JyryuDqMfjSLmO9MxU{6!Wj5X>2kiS#9NN}Y7Kb;loi#BQgz+07H zin~2}WsdXS`D4;d44j{ z699Qedu)P`oly(g1+VEfQBnR%g+S;WVQM+>%XZe?=#5Q@y1#VEbFlG`UnO{(`jWza z69tXO(MTTL+klHnITk@dI@jnuLW5s%&hH$9cel9RMYC(Y8Gc zcNYMjLUjR3wamz-19Z^*v=`BOfKr1AOmwh4oOWdpOAA-Q2fg0GvvKRrXO8ugY)KGi z3L~Wj-*OATIMFy$KN5E6aXctat#U)X+KW!8Dqmx;Q6zF@FdR1yL)D@Sy|3OhqpZD?+}Q>Nv{V2!Pt z)N^>r8RD7AZ(@0`^_x*=`C{$4vd^iI8d%WY5HFf*lSF6@BS$kN+#)-@E>%anT2i)^os zu5;5GLZcre;Io|)9K(;Ld-7fKr8*{{5@*26gDSA43P?I()y`4PNk1wyN%DyLCAx+p%{L^w5 zP>S9h!1V#hZ)19s-rHbMc+H`y%csNTnDUxQ(mR$+r_@iD(|j;nkzU?OpK&dEifV4O zUju7s=&9OL!3@-vb9+s^k>4!Zo7sS^DM!gHx4pM<3C~VEj;+OMIGgBLytlJlOZ=OG zvK_E?3l?SvVU;8|$8j{yvryjI5U-jyP6;XRg(1chJK)Qn)BP~n4!?lr;ipWbaeEHu z32Y=r()&r=VfT*A%vxU)T3BUBX>Avk+5vx{+TXG$q|vI8IC*Lal9K+{bR7J8ecn1o zksBQ6nX|)E zeo{H3ZMyquYF_?o=b23oPE+&JgqL*fl7hZT`hjSWXbom?;5>s|U#|zUHR#aopv>;j zG6kb%R9#cBTf#Us`CaY1r+1GjmK_*++S!)BJV>kba0PGZbqw54d~-{k6jjDKz^g&F z1)~ZP3Su4?$7g(KdgbPadHS24c~Z+hYYf6|$2WM;0b@`D+PmNIl*;>LH)Bt=+jiNB z4E$Wh_B1CB@l$GNyRVyE-(@T{mU58zQBIk#332230nVQ@%C3U`(%Xz^f{$C*LMPvi zCXPh~9i-^>?lo9u_&hT&j9OTW-t(^V`6V_uewOkt*Hi>yUAT2wkGdxrmA@5P^iEq0 z=&cOWUyg~BM?AMIK!{8)#pgau%(o1M*|Dg=O*3;~TTU5&_7XP1`E3!{yK=j|20RC8 z_Z0`{*BuTDqc-G{4?MR_dSDEz{RU_fb=?SoaYjZ{J=b2T6dt7c2q77dONobh(aGy| zjh?UJV3$@3X%o`?0rb9tv`4W#^uYOXO=d*Y>na}%b$jHgWcG3xIqlQ#OC$a<=Pg_| zcf0lJL0cs)`L2r*!S+ z#duXjJ`rqpqIXs*4fiU*#=Y>MtW)8c^^YAlq#B=VZpVAJNsg;>#?04D<*-f2b{$s~ z9QgvUp2+RPUX_Lw(N3wm7+3T*9pwT-gXjI}1So-hhCS0RkhBSVpRb$lL!?(@y={OL zWqxVZPqm82x|(?q53g|fS`!RZACrXN8t3Nr5pPsX*8N@lyF%qU+T98hv;5E9|IU5~ zK{`Tq|BH1CqwNB!vIC5UEeq%M^^JYP!gC%S@v3V`li>jlfy)H4apga2aRV=R&o*U} zE_vlnuVjWjJng|Ezj2I}jjc95d4UgR=2*9FKzUx>=O5PLq6*iy^17Yb^w)p)dqiWH z>&}1o4a|@DdBHfdP;Q0VC3mu|4IDn^^z2jR!i#S;Xw#A=$@g6H(i&=**!E;1PGTF4MTRc2E)6yqv#yeg+6uGu@+YbMK07c{Q1 zBPs@3%KL^^N&pwk)P5+TU?9=qg3gfFE{&uNOWQN8(|Pk|DckV}$bng1u+^eo^@bfj zs`t0seg5qji+nJD%$B#j8Of#Olc!}BUu^`m#w(<+2(q^QTtcKoe#7^tlZy>WTBF^P z#;Buf1dHnNz`}(Qh6df>8^1DwWr70|p`bu!)bZWO1l?@Xm!eagZq{IqC!^dVzA#qm8!SjE$Ocf`5Tle=EIRtJUo)8F;%xAROt_I&6q z*UUt!zMd#3t_-do}{a*G8jiDdw z35LJMAA~H2wb#q)g9(pw@`e?IttyShY^yxe>-tClC#KLEWw2ifF9+y7gud`h$FYkO zsJ5nEedz($EmCTs=~TnP8MCUEOK5_fnt&B1h+1ig$Jur$rTB7Komt3voq=D2q+rc> zDu8+O^g*@KMyhTMx5X<{SUT^GL01xdFFtrE+gYvf_r@Q8oG8%w{>8oSBFoj~^&o}s z2lk?c;eGw?U}xZc^sKYDQs|dn3!6;tUWBhFFVY`oxh^2g@9bTrzK=&rFv;uf5!A;U3Q=H_Df_EU|c4W z;is@PJRrM^18W`b6)mxy0V$+ifpjxs9y>|Cf$WFo1a%9aDNPDbryZxkVm zA{zDgEPG^nfCy^%wS%p9)l=CAg$2?uGIMr~*nz1XaWg_wRBoq+jc=<=+?4t}uD!ow zK0FO<(=fwhI!#LPGaTOFkLL2wlvZ(W? zDO&WN(JtQ*(=*ETof|@I=D7a&;V@7`>_wy0r1mw%{Cbh}YJq!k-rfT$Dn4&&6W~wr zFAt06Y46t$6k;p7<>ZTgwSq4N^)R|vZJkX`Q`%L?-p-&MC8YOG!{3{Co2x#@Bg_cA zl8I3e1+S@}jU=4}{68XpWD|6ZDG-O6@$=}b>jrtG=Bg^lo&(23CpSPrg~wc@V7z%1 z7U^KV>3?|WkUkN^N{rBzFYT31HH9Jt+Tm$Zb^2wK{rsv3Xw8-R3H&2Z_1{RHnJVkkCbNbD!Gi1v0vqt~Wo)Lm5I#;wZlG__kvs=IvO3ChpSCz`&w>avXD8fnbBw*0%4W>t*4%L@KzO(M|%xO0K6?LLpO4E z?fBZar?PwSe2$==O;y{zUw4Dzt|YtOkC9R{Z3@Oq;VU?cnvntfJGyQ;h_qE5cHENJ zrq=sl=rMyFA(L307W#&XC}%E2(|=0T%kbo*3$O} z$y$w8t@!5fV)OIz=DlCSM? ztf-&?b6l!HK^+eyeSj)Zca?$ic!u%_YK^|-gS69dexy{5G&Qs-OU`WT33!+ZF>Y|R z5-D8l`_1mAr*ID|ld@Pk*|BNSgVZ<>$Z3;pYqbGZYOkc__)vF0Wz2=9HI~#y2u?Jr-aRr5l$-=SOm<2z@UR zBGX9&l)YO6l9(|+xy6CV&c^9*$1J%*PJJ%s3wAF-k|cL%Yp6K9EyemaKA)GTEqK6A z)$NeazObzFW41*-DiDU@S|K0DMbK&(J@$s8mGzdcGv$`W<$#k)rXn$Mm~$|u0sQP% z-RA!+&Tp#Y_vS3P+3XP2#GhfSZM%^s56kQU;l!gAWhQWzbpnAB8ubUHTYH{X2=AzS zf%AQdc~OaT6#~=pkhzk<+$3YHqhLueVddQ#kGG5W!&Z49qu^Z1Y_Q!H+l7>}v}@y{ z0BE944{c{q)Qd^*&Bo@PzU6&u54Fg>J`V+vokQQnLn3^w(XPvTRUf3ojuN@Fp^mQF zx7pAZu0N=>8OzZtV%MF`EK!uNoW{b3QMhP-Syo6>r~b}h17 z*2i7c^N1MxGH;`0W4c?W_l2#<9pbXkE zKq|vliREES;*aH$CmS`4{gw?0A5`hi_`h66Sz61kyWjW53l3t%%0dUq9Tsg4Lv+Dt zDo6(y`x-UIPYM=Md&6iw}(UCZzPG2+P3R z?w&Mx{V$i!{-gEnyosQaSF^@jK=Q-D+)rd1rl)s?rCXBMGr322vj}UD>`?z!GH0wf zMdh?0_IxX#1N#P`CYlb|SvMrAdAB00_>KOKw{`w%esUR89m&#z5(XKxAc0L?=gWq` z--dxb+x|e~;uJ%1hHMsjxctec^k-%weeS}u@i6cuW|-kih4Y(!lY`JJs;@vc zl&Y+nj3{Vim>;@jLt6%JwX#HYTucu~J?rf2A~g_G)2Lr|IdV2RmYKi$K6NKH1}<7o zX8NRYoPo^=>~=l+VvGNCyul})Z&0s4kLtRjrg^Bf8Gyts(1{t zq(ByfMx;lb#|LhLg;duMU~g=g21Pf=J1+;YtIKxt7pUX5cJ!M<^Uy;^SjM=*#r35p zm>re#>S{HTWcQ((Uccv4oocVT6T)0=XbN(xTOz9{_Vo;N*|U8K-=+fr%jXM-^9u$h zLxxiyj~vwRN=7(l^zSdlgRBB$UsGMJcXLjc0$wFpB>}5!OEk9Pxov!lXCzxi#!Qr* zi2a+vl~(k^`9UczmVWDff@GGh%joGRK!z931EEC(cK)l1$D9?EF9YNidD2NHfUMq= zl6oN{)`z#`iW@L2`2U@hALGR)RIL#KTU*X8ZxDL0zRGeIdT zh5f0+ksNyXtYk#=i0trCk%Og0REhksip&7fqH?-szXRke)bOAxtNHkM+R9^J;A7e3 z(u!X#qpi4CWxLE25}?L5-g@0HX;)fm7IrZNGaHX=GX<-N>u5)<{z#pZ7B&FjqeXEI(}!4Q0o z<%4tBKP^Qll=-3XVBKPDfM(fFh}Lbz3Z0W#w{`9DAg}j+1B``pe!w?{zA%8%=_XQi}Cd@o50}7GLP2R@DnD0nyEC;Tp+! z<&G?WIEO92Dl}Y63b?#6{}|O;boEa=~h4ddt(SM8L8a zf}CsGR2{PUv)eL3DiIXyDMDRoL1_2KZ#st>-=a>&8;?Kygi*>CiZJFd@*jk28Sh{H zIwyb1m=F+fc8@vUXr*fzA>J|0uCh~x;Xo#e8j$H(Ym($JRI!gU%Vb49Si&3T^aCox zK9*wLh&55P+wyDTKWiI)o*Js-u5tsZ!y~zwo*5%$n=Le-_8chDP-javxhw*0%I|rN zj5_vU`UTftQ^{ZWWo~_bSgHIYwPvA7>-|rZNPW#^(?9O?s|=%{xpN!K*M~r_g%c32 z4-mrl_O8AW{MLA3{=#B;Ev3TS1_}H1CL@y^GPI+kPBk?U-6y=xxjY>8$I!~`V9W*f zYCL?1tl97SzJ7#it4(0yt$H!1S2{F;E{hJuTYp}XAy|?&Xty_XuapE^0K=mc$YJJS zWUZ(u^zBN(p=*ojlupHZ7S~TQ-z_51mR#)A;@7BVsCR5&Ce&w@k5U`=Qk6;75E}qsAII$wWA3mG0E)<0zS@?t zcrmp~Hd(AsZd>KZEVrM%!W(B9vvcYYPhq%Lq|VpKyUqM=yi$)!92Ci*yWcVZ&d85` zb*FkhKejryR20Ml!M|}j^Ri?UKCVCT8+*QtfKfQXiCNJXZbB#lP&go8Lfd|@GbtW z2UxY;{yh7qY{-(={H|n-;nMDz6d%tEpL0K^;u*xR%?yXo9Q)u%9-8EC2O^*k&5jU31azh|+kgHF?b*@)f){vwT`Uml)& z84B&IkfsOg;M9!+5#s!qb^R1%teUln;t~B>!KnNFzg#^>%nK=&stWxHj3Fo8NBcEd z1IBv_T?>|IN+FYq2U+GU={L@VV1<0(Vfd1moHLmzirt)(k8wr?_pk)1ZhdeT{=rYt z8%zyP6Ejr3j_xfQP&0l7=-jBCA5Zm;Yex7S$`($w85IoKB323Z}T4fk*#ejp4HcpbK5^fEaut=l6g{`X$ z#ZuKqNJxU?ca?1IW_Kkm_ZrjOziK}0PQkwC8GUE3pk4fvM=OIX#sC{vA$>=$P+q-W zEZl9w^HJE6%XM!zv`fy1vGJ~i2D=u~J+r-=%)Hc+Wk^xZr0#xW=`XqMvu}z%FDuB7 z(X|@9U@iMe!<$Z}Yq+$3xips(jcCY>Luis=Zy+guWs{1W+HzK4e?sis1cKLDJS#IO zlv|#|-NPyw-=@%(c(7G7@=pf)!V1)Rooe>3Y5;BGg6&Wn4NY(ApyyI19S*Yz>SUC@Lvt zW{g4E&uy7`TeE1S6c#m7+SekRuFGo84WvRwNr2JBstMbNzH?a7kblvJHJZk9aWSi2 zFI8PI+u89D8q<@K9(Yb7WzB+moBXiLVkzfqF)6r3*W}_H&Hc1YXl>qJDvH)>rQZj3 z>ZA+O=R@p-mMm??Q>?HF9zLul2*47Qnb*6~KjR*#>>NF;K;qml+|O8g5UA8~yp@7{ z7%Z0Jiq&urISur(uJ$1x32j{cUVa2W)OfYeUWMw6Sb^xH_?afqj590^=T!8 z62{gcaeM;*nr?kEVw151a#A)-PtM&>M5{633M>@O8AJz6cCW}4xU+slsY&{)~0AOUCS+r#DXO$Y-+m{;A{D_}a6Y>V*T|9cxvQ$lC4 zTaYNB!0EZJe$w6Rd54C!jQ7RFr?T(gg)R66>lld63|6m#*L!!yxb(Ubfj-N5Y|s^; zwV?sbxFocXbM64Q|9ARBLcOIC-WGW-J_R%V;(wO@&nz!A~csb~2 zQ8@-ujn$BQ>T7ph2{45J4;$==T%GU*a;6hg>^f2yCCCSvknC8gG#MjFoJPj7JK71F z7OZ^!z0nTkOoWS3X*$y)MqV*El=CFDY>oM0?BO+`C&1obaB-cB1^jfN>R;7~ zF&}ovB$PdQ{#6Ov_dA`e%@5k$3+C?()pHZO?6@O*t~wjKG}QBWSY#P~i6U!e+EzI{ zPmTRB6yI{{IR|)RO{y-PY*|sLI+BS5y({CXe;0L>>k`+I2f811v$@2nZk}enOUwB* zs6ilN8dKANcvs=`baSNB6C4b;=&dAZmN{G72CO2^Ebn+?&f1s9y1xG zo3B6s0e>lSt#~UGy*q>4V$|8z#!7>ik1O`6m0CYN-=+5X!Os)_e7U}$nrU!&oJZ<3 zM84-@^Uo}&V6966ygrxcmvOlV8Gxa~9?*DcXuMkPr>e@Zig?(ga@ltF$6Pqi@YeXMr%~QBDZXH5= z`XvL2MY&w>yHIEdcW|g{e1+Lmn=N_#xhiDnGD`NO;`*nQ%xxNRR@TXyC^;+FsBu$` zI!nvw_knF5QaV_gOt6{)prpv11r(bj1?r_0)1;M$Ctg?b`ie#z>-C2SZ%Z5v_<)W@ zHfwJyBuPiMRoKW^olz`PyIrpd6l*o{#dxp-yQszhD}q1J_3XNxSk~5ML7nWhMdOSV zJRwHJ{^X5?n0cd`9g4bfu=t64RHhSoQ z77M(wYEAyB>OK#1F3||1-(+ks#jf{X6$S7fX4~|Y1WW7jMTkc30;G0xuw z9K6J~hO{HXr`pIe*wN6{l=Dfp2WKp8)ofG7i7rq}C7b$AZyso$XJ5X*t<{I`*JX~V zbIfl%c)BfdjjisO8Sar8MI0qLYtrjSJvQ?V&X~{aU4-kB<@C6}X@r4bP#(7sM#0MR zAn$(8sRYSBX|L3X?pIbt|8jlO`G#`)>ZHhBflx@DCm*f_?-=97?&B2|#ZHe)I`}wj z?%44jWCYlj3)-QuLG!i;>@C=|{kk(~c^%31vy<-}sX%j=HDYJfU@WxurdBQ3?;4c~ zY4>x{z;o2?LN{cz>F~Aa$OdRE!^yrBE~%SUy!@{I%~|S?dpL5`}fD#>=DZR@5rDP;x*1`ofwPYGWzn9e=t`_%8HjlZ#Z z_8g$B2+Mp67I>OxZS8V6R<;?mx{5#=c2}5}Aw_!zswT;xQD(x@*!bv6q%fO~muVZt zyxXmr_|k@<=6_p|D=t~e0fbQw?24LusUn1R!!GrRg{ZfR)(R@&6}K}(Yc~@!NdIz~ zyZu@nST{X0zO;Kh9Rfa924u`LX5jU%@1TD9ie3So4Z{nGVv4T40bxz6t}Ao5rkchx z*j{zA|A;8Jfx6Jd(6ehkb**}}U5vt2oR>vV5mU?CXu=?wWv{K$vOmKaMuy%dIeGfx z^)Qo{4b|R7c>V%+BHSjmFuR?dUdA@EM)~t8D3{fC&`gxoJsUjwDbc39?IzWE;1qfo zr67DD?h>4dF!cfoNGrFZDUWG}OlI({8eG}a*>+iJzo0bbN+WY5!KwZc4BcX-c1yJ` z*XHwG^3@OLv@YC{`pGx^v1g83d#d={a96dPQ$K+}e-6c#>W4Xf1{hxp5u?HjpW=B8 zD(eedBpS=A9QQ@=>sk}No~cV4x?9bmO)=Ut>pS;A$?xfogYp1(3xEHonItnAby+iH z-9=B)xr2rOtSK+LH!u)1J z`K!}WD&w4him0;DYu_@6CEh93sjkWdeNO8{7LWlRTp^ft1GsZH$~TQ=1)FMS!6;Q)@AP^+ zglypv8JzTZ22GQ$3h_ieT24;J6W?@@(A#Z2PqIKxm25ZIhRn2qX@`C}D+zB|7V($! zDK)m;43CWSN4|^8r%K}7Yb3kQ&{LW&Z4xaIy{3gf2d0u)G0dzg!2F>kt&h2(sk*|R z_}!NR+!du10%2i$}%7HaAa~xyEh(C`R_iu&&4i?^S~U zik*$X+(6`bQ#K`_8QouTfFsCQf2z509kZg0nhT2>nI7FU*JF(ZRZs13!sHbf#$9VH zPy?S!V{F;93|9aN(JLRjT>2VJ93QC7m=}yFjp{4TY2hzad!N38HK~VHjAW7q`qswZ z1x!uF#75BJYZHf~u@YjNXftpwUbVO7726HjEm3Vs|oZ zN`@V)@CN>j;&~12P;X?`DF&W6v9JsPQI&BlF>AVhKgAXV!{Tv_TefR@xrcdmO96vyC z1L#ocw)%!$_W-b3z2_}j(ZCnl6X;7q&-#Xaf>g;YUw#BD91nE<9?gN4i7OZGrvkOM zkZotjllIvv%VL*2-DeuH07WdbF9B*L{GrYLBETwBv?M1b{{l5yDzj~!4G^cYeRn?$ za(UgxCVe=WEvEDkaq17*?`>SqO~A0a=sOQsa3S$0?azJVUN|4WbFAmJXP@qCp7Ref zI0lIOSMpNu7I;gp+t|1d-@o5(uuDQse^fweh+m2R2OEcC6flVT(lmE}-sIdP)QASH-dZ{VFb{ zzL;wXt8mcZbhDZLU;kP3e`h?Vr*B^Qp9M#|zp4Ee0dpx?_2G1Hl1L5}Y$K}CEG;69iiqwZL;Xr2=BTOF z`e}efj2(t*ARpa%2{Smn1MCLy-x(R}4Ij=V`Tjr0-qh_46 z&0K3mKaECs44{m_*C#WV_C#auPVl-}e=3%^r)rz*x-7u^O;3WEl{e@=aCYK{L)2TR zEoyO!Pyel7@w|4<4Gje>hdO*v=Ael3Y)TJwX{5Y2%+Kuxx(bG#ZC|I6*JU8w`4f#XQR$gk>U$)~an%U`rBRQrAkJwbd|snB_*&G% zsXECzo4C!;rd^pgezHv69Xm|=kb}12gmmbRa7Lq`bAOdBx-cGB5fU#P2GdteVl6?v ztueP}HFQhlIG#+PH>)O9Ra!mSekr)Tv`Y@MUhDoDCAi0!+{*$qKUUS%DLCWuOoU?u zcFweo)t1JRXDEPJxHWsKL#Jg9dRbCHZbM%rleaEd`5JE$rYo*Yt*rn}sDHl}p1ga3 zYVp70bK`8f~e+lS$|3q%)%`R{ye-kshQS$o6WrXFV`iLWNcL5=B}6vz%hOq z=68-9_-E0WAu=MdlNhcaWtB}_HMKJzv}n_E_Zsjcj5Pwu%Ut#|!qcH_< zIZ-+6yQ4jRer7tOso6~dT3#MOwk1nRSrl!t>CO>Va!X?o87eZBV>3d4s`ji0$az3E zYgk1VT+Yo-1MAMH>CnqBhP1P_(^xv~8mZ8}xmyA-X=&{b20LokuE!e#SqqvDpSCFJ)Tdpfj&RGt(2&iJyw6HOJnfhWq!yHdW;o6-xZB zLc#Y4EucDs3qPmFyLJ!XY#I^f_N%mQiooQGOYF2Q_vtajS2v{4ow5+-xVBXYQdC4`x4*$d5NW?z z!w?#^i_JIqL@@spr}|Z!vP5>N8_Bmd*V#!>@k!TR7{G-L1kTK2q~w?AOH|J`Xyu|~ zn^9d`eWyc}5L_kSCXJwt5tY(Pc@`X}OiVe2Hl_jyKP0JhmS|*@8YtTloc3wSh2Xd@ zGZuPEqDz%?{>kgdPjGdo!}&dv&W)n;FHV21Szijzy8Ki0dVhc=E`w4PHS09PvoBy3 z842PzyQz+8!@+C_%aW$L-l@7}QEe?pmK|XpIQT=^h)}~l^@&As{@~1-jCix+Ecg@r z7r#(&vDXl|adf@bB*G!!;8K8z-hC&`tNeetBJ}382O`705@*YFqp}ZD*#kT_{R+}v z>y8Xbcz$Bcg9sh9=xVi?(i=mv!)Zl(4u_kEvDx%K_2|i32E`tl7cOFgGYEvVn`Vic z$F@!7e>LylXJ;jS76WRBN9vS08iYCkN%OX0q_|kY z**4$j)qgAv-!2p+sJa1E_`1uIk^_P6f#Inz*_3cuadZUAH!t*;FC=OI+?rtJrw zCWgU9j)h5R?=j|I=CT`%SGAv366#4tT0SQ~-YxZb$soZKgQtN3i$8*N8!`jj7`Y?iTw7fCLKk6@bl2ujj>`BWx%aWSLgxS8u z=PTc0Oc71y9}D#%mOdG-&3%1gWlqKClGlCQ-E_?oU#)qLM*Yj>>N@%hrzR@41fE?F zs9==K8uM892iuj6p?FhvaC=(17rURJ&^Tio@k6jVKxI1L;kwFeoEVO)!#a{}VQ9xF zt7z8?^st{(0Ftez(wN>FGu^Vhyj_AB&3T>tjql5B$@csDM^$gu+NKp${*>G;NhjDY zHugRTh;4_TmNBySG`w8>z^9si8PmExxCWELqe~I%nzF8mDtdxyieP0n<@H8S5p4lF z`*<7Ka$cvKfdkA>W#Sw1b3}R0Q){(eot|%$BxQm{63P?qL*DfxV1RA`P?-SdxVUVl zG`(H*K6p_8cD2zmJ_&!{K>!AJba#&)43h8oPUGC;eF51Kx^h8S{Aw8&+TgBT-)9qU zKA_1COmQCpon2*VS6s=lqk_IKUp$HF`)w<7l3U4mi%UuD-bWKIuCpJne!c?q(T|*t zHUXrNe+}OK`Qqe}zZE_j+^LxgZ=KJFRJ^26L`H?RhYLH(MhkF!}pvV&$nemh7>C{nK#bfgpL<1*^S39<5 z(&Y5-T8@i*_8Bn0_!_KnqeXT6!P@|@rHt*j#*dRz9ZiFtYZyB_KDh(D0I2z19C-oU z0oLH+`)}xT>K}h_zxci6_r(1aZPf6#&P{M>?C^NY(=&dxktTJ9`Ib_0sn_I_Kc3_I z;eHH=dpUJa37hJ6@dy9#u{lc|=-D~w!0}G4=XV2bfgxOB4F=^Y$oW#x+MpF{<^2tH z$wWo$cAf4^8*Jk#mi;&JvRg)L+Vz80y4~`X2HWD|a(_HI`hx4qQPcI-e?}{NeXJ98 zJPLZ}(}9%P_1npiT=eoMj$c=AVf#q}T;?Vts)b|FA>joP+YE=}=V>M1zc|#*tFk0J zMxqBQvb{?oMuEt>DqkkqgyMjf zrVRZR3(0CS`{~v5p4=?3@;S!!sp*&<4|OBbD**z(&!jK^0p03K{zh{oVCtGFIKu~ z?|7nJeo)|a^YwJ6vLmn6xT39x?w+Ccxi3Be+7+xa4JC`Nl%Iva@_n}_DHt09{ zg8y9Z_db<-ZB?*Da0R$RDC0qmRxYp&4>g5+^hnv7k7~c6>KLru*g)-?l1rasTJ)AI zKr(WaZ~r>pD$Gei`71B*6gKnD&TlM=*LGtXKocj8OwG{njZS${W~)-nBuNniO<#|QSQ49{fgGP-0tnGavHOVBl z&9=t324@Akrd(hH%xyKmz8Ww-;W_V{oXK$m$~Ip+4b2V%L%AGb$O|6_?bk{V-7SZ1nsE!## zuSzWc*T=3Csu$)*C1L^tmW>X=Ba^bp0dE#Z^7++HtT(%y5vg}I|1vo?znO(Yx(FhD zDC-Hwe-;Z&lZb2cYM)0nYpu+;Y{#84GxCM{tbZp0g`{2s#N0B{ihob~QBI~nmAx8j z%keHVmh+N~%+6H>q-#^+1U?&HtZcs<0^K0RzN^xtn^Xd$1x!GwR6pSE@tj`zGB|K# z)l6s4pX#^)u`XW2rpeesf^Ql}2(si8s#m|s7nK!`!xDU3Qf>uVtQNSq2I&+~Y(kyIN0X zFzw{Lt4r>GA7Lw}X1VWQQc=j=ZNuvoYy-nUvr#r8{|zl`uqe{-*|IIs7E)7|!UzeR zcWIl=W)S_y15dMOsZ*`7^DYs%-O9p|HK5!0_e+9De@B06y9T6hK2anNl3^p344jWo z0jQ8g?SGc(9tfq{iHkmjsSVNDOQu^uCV=&&o;fARmbl!JoQL1r~<8c-tpZb`No>^;Bz3haj2dlZ)!{tzGNPHaVJ0V?_ZO1aCp>pAeeA1xG7fq zUoN>Y=g-4*ZJdr6Krix=c>2_#$aM!>{SvUM3HRjnFMcQHrKyJ~5orit*3<9$(Ks1q zNmwAeoqX&laRJTixu<4+xD?Uo8(D9^US`U1U(?hZOe}A4_IdmUkhq^rSe{{+g5AE> zT(BxHCCJ2*jkDJAeQOSh=j;mB7}v`i>->v1;Ees&4eh~5D`}uKcH13SA-1))PMvmH zww!Ex+M9nJ<6dc)*)SOZ;U^$%l%FREAbj}vv!1ToXX{(2YD;*Un#_vsR7sN z=Ui7l{==p9*Pvqd<3FxMTWDMXDumzDC(&jlaNo;UzpT-BfJ6_ohg?6n?jK1?=HmN) z^~Grb6cRjs@0SvQG;ss<7f_h}x^uLC@LB!fM=q{=&JUsfw2w0O#K&C6?q5OQz2lzM zK@8${SK`{wa<6*rZTm*U^scdyhHlb^$IU2rL}kR#xW|2c;L zWp>p6pBJa3D#h-|@PsYgxtOoUBdtJW2|M5p=s5t=Vr7}dXK(-2Ankuy{r6&Hkl^*> z{}0MLmUH~zFaB@gLj2ltRUD}mmVA5V8a90hqW#0(D?9;t76rPsy##0h}m0<+b$7l&8iVZs4Zqul-Od&79)1-7=5mOf6wn3@B9AqUVmhq zxvnHmNv_X%9G?NbGe?9d)U(8R@_@T!rA{Dry!K`J5OWyhEZd_iH`KoZvlevQb*Q!6 zD_x`nJL$}&H+(fYvD(uyPg@K6Cy1|gHl#ewfH{BAXqkxp)eM_8NSB~>n$Y#tI1v$BQk)2&ywXR;ta{HTcB$qwT#_v^#WwTx z$(>@r(>b0?@-nMoB7wxd)-)v!n?2yVx{&E^el)Oe^s;TUC@zz);$*={e$~B#WCDeI zO?r?ERbM_h`|8J)8^)g8cNv@^;PWklDElZeD8OJ=I{LGY6>=o&!A|Q8dK&33x3LYhSsnEgSW-x^@xdT0=K<$8&X3~}W#Vfz&6&f#B z0AYIowtfydUt zZqMzRdDM6kUXSFzm@bJKHNU#XrB3qns~TIbFklD>g<&=}PQs4q3>x1K*^|h+Hd$icxPR!iDDRHc8uNSJ{BXlbsCjmg;OpC5(}iO5RY?ZothYKBpw`bzO7=qeFrV z>zI)wlcX_yTfAe6>bY)9F)7HGqS5I0>F8vt4?(V=sghPR_~KCMOL#*1K0!Xx@(t3sw%l-Ut=y8B4FyXc4;73U4%ho0qL}6SnjZpt**0Ioi2@; z9hoDvD4dKE<|2m-X4%+<`(7^@N;un#c;Cs4uCwsyq{?-5b!JiZyG?s)-En3BJh&~n zXe3av*gj98EYx%pDD6(J<)q}m?J3e=DqQG93#l)vjTDb%L({tC2%?ZoQNp?|5$x=y zLJ^99sF+mmerV zREQ8|X%gKlWY4_W_r9ga&#|zB5TAN@&47Oy2Ak;W9>{TAh^4NCT|*-2N@`!1$4=U5!9lVn)YKQ&D`i8TmlNgOil=u!Ri;lKND zYr-ubh0sjAAt5;>ze7hktA|tMr;kjnd-uvdUCt>scDKk%&7@bxcVf(KJ@wg`V!kTJ zPMs+giMIG^WWa^h zQ=pCp$gc9_>T{U%6uVzFoJWoT_y0JFW9`&Qb4rf@b0Nbk*$q;XOH<5YsP8DUlPgJsE_M4yvDV zbbS(^T}I+qh~qAr|BjCIC%y1~(b-1Q-sJooPK)EqRd`i}b^D>hf0 zwy73psA|~hwuu466tjb>1Kva0+KJ-S+Ha7ED!b%<>cC=({GJ6K1{m8zz7N<-q&I{~ zHF$qJOxxC)IlVbkrK*q5Kzd#)cRK2!jNYHn3l}i6k&EVYjbB9`)c}&liW8md;FbmD zu$*QKKqxa%PTfoK+_Zq8#-9j`#4Uh`}hy6-Q0NlD1T~o6+i>@_G?RCsnj}GECG>$kL zMwQIg7l|c-fJa7UC5{87kVD(zl<2Keh7EDp8$+V+pL*dau|!23c=k!!Qj))c`CFP1 z`rdj9r=a-CMDiH+F7Z}~GD6^+8ipiEEtgHnuNCjX{FDn6x&PYRq7FFe0E}bf*z25m z)q5pF5_P<>-v{vcwdq8r(gDP>@P-h(<>F@B1T#N<+a)Az-; z_IB@W+hGQm;N`w3mm~Nr!fA>iy+NUiqo9kjrwg?!6?WH>VK6^$(_Pn;>R&^79)fi= z2koLfb*;1lA8V}&enVRp15w!oMEjU!*1C1VsG3-eim9}y3gr<&AW_Vm0p57!z5T_P zzaE30TVDM2pOOZM<45oZ06Rpzw+7MgojLy|{ylK-%o#Ks(P;ZO3(Mnp7M35~PtS_6 zv4Woa-IdI6dv*5~Jvd|aYIZOI%+psEEcV>*D%D)G{Y-!-(O#aR{*d4 z|Ju%hAMFWa#k$=)tp4X?*3g_89v)$5VY7FKpMU&kkim&T;h+C{Scti@&pNNv5^2&g zW0XnkM!|Lvm6G%1tnqaK@SL?C2h4yO+t7jBV`r;uD5_Zws<>E!5UO2s?yXsJ%##XG zl7HUXeBH}|C?;-la<`%A2R6XI45 z_Wms>;eQvx&J2G&YsCbP3CXy&HeDL8!EiOFTjlC_`>n(g{t+Nl>CZd8as^0c@O_u9 zJIi-+wW>E;16j`J^)e9$3%QKuF8++2z(=Dm-;onW9tJ3VG{DOi*r9z?{|c*zX|5KO zJ9fBu*v-F1wSyj6(DlfIBcM>~Y=m;hCx(172v~dYeA~2q1D!_o!=5u|l=p+htBLv~#>z zXlDQ6?!)zq?}0+))qnVG&@SM5trK}jZNp?L_;OG-)KA6#C^~0MR*uy*78wxUl!R4N zPR%O|qI%|Y_f^Tk4ZlmJk4}tIk9o12nwX#%M?42Ul+sE_6 zRrRX!YpQj`7IpQ;z~HgReJO-AW6cyq2gwAHIJsrS`&10*_@&mw$2~lMhEKZh#^0h- z0@c#*Ql)-_^b2zx_B>r3yOe zK~Q}6ou`L^`RRb7kbdjP?S;|-$tB^cV5uGU95h_JHH_N-8a_*}Kpx?~45APSmtFFY zg(>TQ74`)%6|U-tt$2>t=WUpIYHy4_xHo+}^u^QHRAZrV_2`I1)A(;E*Ra=@Oe$*J za+jqfg>5EvVXJ{6-?)TLV2F2RLbnZu0P0iL$Z&P;Bqz65EH2>Yj=A`I33k?`s@VXD zbmh&T^>a=l-by^mt8l8n+0s}*vs#9Rc8Mv!{zGDV~L>6cVs!s*9K>Md88mmxI3{ zeu2+VU!F&Ys`Fi48mn4;Vr(V9BgE0Zzm#~$R-T?;taEiP!(Om3YF~kE&*JCtvkVWz z35N>VeHIEu%V=Qi3+IgPf^)v%<|VVP>`G6XHl zuRr|qxpl0NX8<&doUBz}aA@`$bZ;t# zV!SO4)lnR8@60o=#PGgB8*V51*V1|t_ieo9Fg*DD%4-_l1iV~9BYzr8473N)d4G>6 z7uCX7%7%d!8q3Z7TbrdzM0~O0otLTZ@%RTS{mb#+=!ZvA;udzuM# zoV|O;JbMm5O=!`bOG0vOMU%oF_l8&7a7uCCHXAOuC7|uDZ}}Z;tT`R^SzOwYcKbXx z(eBCaKHYIx&qb*P?*;rMt31{o8#1hj^>#-P-qbVVlrWiYB7eD+S(@tOPQ?Q_`_dO? zPq6X`4XjNZ@+&>YIH%$4G*cZQrj5vCgND{it_bq@(sS^DnMx@xh5GS zf#VmR?2uTqZ+ZJ9D_$XF4`*1e4)I7{SV=^H%{N8>!-mB?+948&q^-}_0I@ZKMTrxp zUo>W@08Jv?}G-D$T@Nv-qVZTDPYP-h(cg7~~tF^L?tCTlJA z>sR{=oA)4)X#6jWi_a!z?v08lUkl5biMcpjeyI9R$mZGTZjtGe7{%KY$<2|1>&40N zD`Gvgwxo#FJ@V@2;uR;~H3N3JrmW@m4MV5IB%pp`0zT{3`VY|*n!hWM`wsd#IOtrE z{%j${H6;5nG@Cua>OPkl;oXVI(dhaKHW|P{gEFG6@c61~&e6i?>{e-ozkLC%I?bWp zBxoX)I(8D13V3L1m9sf&IO|2cq{h{tZHG zYw}htq)up55d5jq)F{Gmexhk##nx493qF8WSP}DTJ$^R5iZ##KoU^E_fCi zpusz=%fdKqnut_aq0l)p!w%TY2MLba!KD>c>hL&Rg}l1mVoQN0WY)|&G8D}%=Ed$qujdiPf$G|(M7h5x1B$fu=t?BbU zlUEguc)6CcTn*nZ1qf?F6Vz0}<{qBYZq>Lw59~;b;V`_ejDExqr9>PY%J&bJ;W;Ku zm%YmH(~0C_V|tpj?(F6A(jFuE0~v3VkCYnZVVd(PAJGo36Kj2R)G*PHm;(gbl@hy2 zMYo6Xyrl27c7ofOBs?T~8)Bgs4WmIf9rUaG)_s(;&Axcsd7F83Q@qXP_D6eWgTx_t za+>!@Z=R`QCp3@h`+bJwG47b{-P_1t(8xzmKp^GkT1|*@q0&;iB~iaW z{DZO&P(Ixo7sd3NJI?TgOFRX}&)5(t-5d)>0^eVx+R(i=z{`7Ie&Kf zvz_rKt`5m9BHsV!-v4z0p}KUJHyrp@M%S9dTJjP?s6O(znq#2W{MZY|QFapFH(h(e`rfAwE8Xc#BeUb)I&j!QGCU3QGW^ zm`=p^B^eNm1aIkX)_i8$<$7%W>!0W!py$rP+~{H=#vY(Jq!}+K z^cN3EtU_={6whj{Q!QUQx{kDBq|P_XnaNzhv#I(K+Z9z_`Rt2dN}*w7hgPjLnyNcj z4EhDqKI?KmD_G_G)eI5tnX}NIXU>1~4t>GG9F1;PmQ>(iIG>+PI)+CaO>G}{)-f5y^W)cJ zLYdD#A`KNy0G4&if#n7fs)N(eoYrRVbr4A7sW*x4Saxs`d?xZUn>;xvQ<5!Ugk3f) zdp9oGkFM6Wkg85IT%cvEdq6aTE~ct4`lktNX9}M9M48oh5rIk}%!c=vvMaHuTV5OG zid={4ctoL72Z34-@cYI_9{O?#e#JVhMo=oAGya*%3&Hsk=1U2R49%Oy+{Vwebt66| zC-wmiyXj!p++n-oCy@z+x5=nLs)o>U?AJ{TAl7PyDlH9owl)=rcJU9(OK*nntg$_x zPGgA7vvOhAfwtPXyAk3s-~52aD++Usua>%Z;zS`c?HT<{IyPwOlFWWxUgsB51B;^u z$iJAGyfmGFg}auUtFpT~wKxw`vBgC{?;HF1B~WR9zx(JFalrAmZcV+m!w+!nnI+gr zHcQe=k?w6jeN#OmT{mt+TyJ#ueQ0U~_;-s26%@frlq%l29{-s~_fW}ZW1d(eUrq6- z9DkdQR5}C@h;<8pT8^xSD^~r3ymdk+0?;v3h!WU7m$dWULHMEJqz?#S#SCJ74`Lcy zI-lJD-Ta7^ywuVG_0csq-Ha>M^s&icQO|9?osj$!3Wzff66 z?Y?S9>8kx@>zAjieuLImbn2&CEXiiV4t4|bB5?&d3VJmdA)W1$)Z!VUDe>=I4O7oti$=J+`-TEt_}@h*vbkRnpKU<` z2Ou2afKm4h*cOOt-IN0?Jci)}ux+W8^PUNd;u=~xwn8Gom%3Whq@rMLXhK7ttcSJ> zan^%s6sUz9+*PvmRx+JPop9?(q^wssnW>+m4p;~9s4MxtpR<+qe6|Q$xBs_>`hg^S<+=Du)72F&>?9|Z(?W=k>`RBR^7m5#bK zjCR))s4_803?xX!IWh9HHGbZuGXyiVg4RxZ$&s`K)Mrsfxjc- zdwI_yZgp%Y)-XT-((9x4afYQS!FA(J9d!^R-t%oKinZsiVL>)4)Lq>?l{=+&+%e*y{BlQSH;^w+=15Kk%Rw$;KH$*a zhvJ~dDdX+kIQK5{VXE5C>KNpB>`dF5@(Bg8NVl4m!;Gc+4j@w1rKy1QIIm($nt%n0 zgDm>H3a__n_XUiphRF6%1B4}_2N6wAEpALU<`Ik0;o$*z=c&AcCN(z5h>Cj)Sr_+j zF+Zq&3KUvySAN>#Z*kLRHEuI>9vL;aFca{D+;tj+Z|X(D7e_g^!%n+|`BluEDn-k& z!=rkMw!@)xu7T!$+IFDv+y*ts8>V#SP*JTXR~HY<&wduQq9d=$D3mt*^ax1&<+@DI ze(kVFo4P)kY;movvNErkEa}2;Vw25}*w8)d$8@c`?Z)uqExtVawp{E%S)vEsJMOF> z7~c0iRb%2zSzjr#%yw(L(v~*#k$<>t(XrmazqGe+@Sq}Vl8@x57(cRAS&cU-afND9 z$S^`6O2L03NUuj;Sy&lp({Atpb0)Fw$l3+$VZehKoS!puyoGXUU+?xbP269QbR0&k zBR7{92ipjBChIyo(GUy^waa1fk0)>sOaUtDBmp#I!%5xVq69mFnh?)V4ue@Fo{3}Sc?^^Zsfi)D#e zWCpkNbc|(u_`*o+KZZjP!o9z}i&+4B0vA#~CavtJzJPgydfX6J}^WHM=l$5;hEvJp)`CQXIgH&PfG%~EmrX^-$ z2K_t5NlD9(Q{9(nSf_pnRS6=WOXqMjL7mv2XzG!sR@lKRMh43iz?Yv0aNG;8B77 zrNR{|(xO2>fJ^!Cc^!^ka+vAy74JX=N}-$rRJEzO#e+26$-xtK{|~h|di)_~?{d-! zN6|(x&2*#q#HtY}`iY-*U1&j1{57X4Ah@JKBJdU?4B2?ke z^~zTaW)J*{0>2IwB1jDmrH@G7$4=C@JNg5IxZFeG!f!{J9c{V}fijpFAP;YDSJao^ z3ozC;^cKVwq}j^FOog*p<(>}6PNKNVm+v;U=^GpEs-=bibZOPsQw6~8W|@kGG#t64 z;7dygm1Yw%X)!AcOG_;*$s@RWBo-$22>vS6TYtMbP~N9El4hkUeW|*j_H(TlZ*s)* z&x6tW#8&&R9|KmOZ4RDn$LDAdXc7W1T6|!|PJOIGn~2j z4Zvx|`gO@ARnfV`d}%0@|;Pe3JJth_AqD^zWo{Zvz#X zmd9;qp5q@)|9$BHdCk}s*Q%{^y;$|nAkfTy06HIa{yq0gpjHXMIslFKeZV~Bhe=F| z3q|rU`|b330)C}+-@r*p$i zmj`W_yBma91H6y5eA$$SLP}e*WmR8KW>VWHGM(>8{QQ=0-ZRQWXWSAOT(^X9v<^jJ znQEMH2c;%>2%4vhKf02X3C~FO5|Sms#gK>d-!LD=p7w*zzCT9=U8rdiHy$Iy45H;7 z#c^Od-_@KE%xqiGMLz+y$gvoy9XkbYQ=V=4dnS^pDF#~F1&%$xL19^2v5{hAM`4M& z=z@R{;gq=7qKSyhulfrF<5N%IwE)RJ0U?NT+BzQ^|CCq#`* zMtRRl9--2$fbP3a`Q@f)ND0EP!O%%o>R0;$Q6QkSJnQ6APX^0*h+VHMLxhl4B&TRp zam)b^i&We;j}Xz)`rmw5HbKLRtqz0A$XFUnEBC>_!^u@j!g@yWnLw=X&7KH3--{oB z%vSzTw3R$21)}~(aQrX*THFs--pC7{EtuZwDoRIbC8j3*RyryOI!ov;GYLdN64UMO zhCqTT8h^)dLQ;cOy1g`Q=Mu#ML z>D#xc!HvkYN~2PoG+&3-JPEBizlHE*S)Bb3Ivr0QE9V}^^U^El2maXZd+hQr5}M1U zO5@ZX@rYbNbiM2zhQdPFK;Ld#zpu1;YQ=Rv>-if%-jLWz6!oj+-utED&mX@M-*GFE z>p~}`k01=6SCQP8)vHbuTMJlgGmxDxBofdRTvA%@)QC< zX(5xC&YqUi!`YQJRTupt5!-4#Qx%ba76c|`cM(cz$}I?D+Z~cdN*k?C`i8s9$&q`W zaVvWU`!~#wZ6;|{E^U5nH5yWm*zW8?k^tBRUGyudH*@tmvduHD8v|g(6}>-{w+&Q| zjA(_rzG#POg2&^jy#BY5ggXflmDJ}*=?sO*k08CL=lpjhtG@C)>igjGxTS*5T?ucL zc$`Qf-6vaJ0RSpQK&avQ#AeWF(2bRd)UJ8urd7F)a?g$Sz|J7TP5@YIx|;#ph3$>r z`XIhuSEEoXTjlH;pRg>v+7xvX|i(r_be0vaDz;6q;5=*|Ez==}<+L4#gFi*)%oADCcPOsL}>NPot*12C~aOcl%D z1@*oMGgkF4neH$*Pb5a{&t3YD1g2XD-jr^~x?Y&!j$~dQf(QB*?~?-_dCkre<5!r6 z?i<^AM)`bx1|;S@8t=ZP;h#1hQs3p-Wvp6vZ1}}tu%mE0ekXiKOZ?@G_N_2Ue3{ck zBOF##2dr&|4sz<6DIeTpX<%vHfp=-*b)x~;)0IJ)Cv(91b7SY7L>E^`mG)(?GA?(o zM8!I|ug1)D?~qSUIZ~#{^+~r}8f@cL9@G_}8yk91$+8NyoC3KAvC6vgb^sJ*686Ib zK-4HH#H1wAt_zcfRVwdH2cqco?*C0Rkc86wP}u*r&Mqy_u_iufo7nkGPkwZ*B_wXb4waJGR?xdd5W+efnIX| zyP7qV>P+mob}g-Bw!(sGgw7GN=kzSD(44nVRyRT)7L!c1*g}miC4H4e#@;%t`xNn+Yq-!Wi>xE2Op-IjIMl(4zbc zol~lh^|A+07Cw0CqZ=Kk;F`!8W6H`sEPH(>nMB$;wZ!Y`k8jN%?gjz|ZOWE_YNT@> zLz5724_WB|)+bDb#_MwjV4i&m@^tqJ=Sy6kdR;R8vt5u>B~-1`aZ9+!h|lau;y1{E z+rfxbv?9GGBB15lmDZ`maBR^3ZkS5XHr;1mG)z&FyqRrkUJ`1q^(?r}8sKUCLtEBQ zNuvcRc2Ex6m)oX~Jtkb=4XjwG;#yd(eaVoZnv0HxvnO)O!D!#uL2e(-4}G!CGTZ|r*rr*;BVe6@#Bny|pxQO01>jx^=$U?lT7iR8 zlL~;tN~qFYc@{je^?ybZO|t1g;GbF0F3Y#SY&osh+E$mJn-0D$W`9-Mu}Y zXR# z?OP6e9MHqua4-s?KESB1SoQ zDmw~xnT}U{zE}D|QStfkLArS5EV-;uNN@M!_jmZJOl`oyMgSVfbN53_`)1T)Uu28L zGi+db(5zhuHK6CVQ^A;_MSM-xRE#s*P7x0=4cD;__ziF-D$Oxnzi!T0W*gFzH=j3^ zd#C{vcmW+!R@^)>^Q~27UUX9rbQ4EKjv0-giaVyxI&Fz}VYJ?wnYS2`O7c)~+RI4W z7I~m2@p8t;S$kMThm>Xok;EigN57cTOKnPW-WEfC`9~k7p^99$E0^fZYS;XkO@(0c zBw{5fJ`V z2XvWlTeSl;ng<||I>7E60oYrjkqO12rx$&tN>Y#uBrD1n{pJ49$5{_H_Y;`z+m_jBeCllN=UE;EY==ljVMb$~)1(Aa=DTSUU2TJEX(IhhP!C&xUh z?pou_s{?o+aXpLFLcrSOi%?$$K0Cy*PtQRQpnQ&yKd!kbj@^icq%QKLXGAP4WE+JO zDI<+t4SG!6)_9mFhgtSEQq)7qOtINU{rvp1m;8F>D#AD>4$DpS@~0Gw;Akdyu8IZLmh`VSImCw%^Zr+>fXnuN(6 zCZ#;{Ml1f0%J%|{({eJIp?mkdnIu zs4-J);-3t+qTn`kd1hzJN(om$3-!7d0q+WLbj!>aBLu zwoIIeZ&n?`UATzfC>afjPpJKml4d}$dqT@`K&RZxmj^E|XTEr0gE+?(;^_%3E9I-C z>`P`#%_e$rvD3O)Y!G{0c?rqL?w{6Yd4F6Y2OG}Q;-ZP*B}lR=quQHDo?@wvtd@I} z_EPMq`P4+yQ0rKE3sN%T-CP&%bC-9O>hQGB(h{j6uLj|FbJFb$7aFK~#`FWly|6tG z7n{4DE&J37>Ge_5GAeODeJPa~!(A~TIGNf!r2p;+-h3#zp~GgYLm(=dxzUSaaR4JU zb-7w2WcZ#%9zcbf7Me5-Di>Mug!|+~rzQ=Q$wOFOQoQCfuBabd7$TjB^xefSXjayV zG2q&Sn$+B+dCJ->3tQMyD7z;8t_TFNM^D>8Y`f%ud48Lm<2OHtdp%)JMBrETlrojF z<8H?@bFs}={~fG)7L)*?o_Bs?%GhX+r{9#ye^GC6kapd;@&1{gwWg0r1I}b2WeBx| zl7rb*s>yIuJMRV!aXJ+8FpTK7ZQx=NVI-s+KACN+k`K|Y$dbTwD-41MNRY!QB#*z0TN!Pb*Y0jAWc#&uYeSJPaH<2TVCx`V z?@GQMfks>J-jJ^bY-XpN2l;C~pLArl>D|epStUekGLO-c+tpuYWf>6j#@>b_xISmX zxdB=$;^_r*5Kki0D|-g9c3zDp({@46%Y}*z>T4k>2TQ))*QYri&B9EEDmo^6aMe}X zPl7HCrt9}r0d!6&QM|d3k#5-XP@AAd7N9;Jr*yh+%-5A}n17!kAnQ3LrhkJ6jq@|#3eSX7a4*0LUkho}FE zk6K(!UCIKm;@u!@9|rQpaE1j3nF1}!qs``4QL0Q>BwQo~Q|2CKkN=J|O&-->KqRB! zgiL5{X^cxWaxhTm3%Um*yt+>Im?tWG7DFrixSdFOgz&EV1!-_vPH`jA8Srw6f!f$S z!k1=+mu=q|HT=7dxARl`g_~8xt0iZ?-s+8*`f7fDu1PSn!c^Ty(dg19Z(R{>5UaLc z5xrqji!@ZW@wC%c5+n$JPsUrdrFp+@ONa_sH&b$+r7HM*eMz+8vTLQhl>0IyglHBd)#~VG#1T#wvwH^|DT~R0F6PU3e;$ZB> z`+-Q29Tx{VFykg#rUU(W0ygJ6SPV{A8R;FCDa5}c5?Wi zS(6re6i?1ui&U^3rEN_i$di~xZ*`Pos*Z->7hJHEbq}hU`Yy`04A4y3>81(#q&X#y z{(-ci2Art-Ssk*zsX456P92(=aj>_y~e&MCbI%1!1XUj;3# z3Vp3S?6iw(Ny2vl4V6Yjk*kT5JiWfH7m%7wO2iXp1<=g+lMp_yT|3-|ta-{dIJTj0 zJT?p^KQk>+Ck*hfY7FTCc*rg%RS44p7&9ICtxir(-YELXBp~4Zj9tlj$Dlkw;6;UD zwL!U@s78U&HVr9c_wTwZxK7=2gESPNras`)e7az99CYNnH8(rAzffo6;-t}s4_wK4 zmHF)EC_-d$o+j6J%cw|}$AD&6+PEkpvIy0x7y#y~oaj<}9Ggz^R5-Cas%#tcNgqwq z_Al8|P5HXnM!viFW>ar+Cp_^0Let zn_iM=N4_}=7CM8J%%5lpj2q@)GcxE!n|JCJVmwfWrvjxCD^!7I#ncw~mT$QU1lY&m9bt{DqW%=?~_LCeMvkN|8iZkoy*9V898cM-69P`C`z;@l7Otwb^WnrKCTo?uFH z40sSHgv;01dla`TQ_qk-++46sE6N_KPZtq)*WCtxNcJuod8K7opOa2)n+*DKnA$=a zMB3K*qB+;pOU>JN{nDSuPLn;hjm&)Eb$;V%qQyB=CD|sGZ8tR+$bieS)Y|Iu--jrK z9SW`(s^iLeeR=a_*tj{$%T*@U2+w=Ev%FaJ=;fkOZ4Lf#Jfieuxx$HZ$SFP$%fU>K zBSfWr>(&BPQe4WMdFQ}dW${l`E#eYyI>*4Di9LUi20zDrHXkAO6a;z%I-7g(;-&Y@ zLf`M6f6s8`*5|YGiH%Uij}+f)KWz?K-iV2*K9wjThyJ)2RU9uSb`|U%s``>hyDAhm zpYE4~h_E^2k)FR2UPen6?`^2fIIM|?0|a=Ih~udEIgzrej`i^^rA)Q^wCL9d@l;PC znd(TNBA3U?PZEu`6dWFJFZOLNZrl^OC%KcX{a|3l^4a;OKWRUd!kHDz1a9)79hZ0SY%d*M5LgXF?MgqMJnf_Txq?Y~JeN_?_C-T6 zzh zBPX$8=50r$EQz(X#28x@WezvZC|`8+G)h1^O0TeMjY*FOk+$smN4jF4<?Ei7ipz-+qpCMiW zk@M|uUq8I@wy`aK}Y0@cZRxa8aj&FJ&cs9F6NX986h zd|cQn>|E<5UhG?bpElLomJ-Zz<`eJ~XJ|r~wDn$nl+8JMGnI*LN`}_v5hwIwp>$Tm z9Y>?Qc5Pvu1Fj6!?#WRPb9F=h&{M&|gwkPIQN+>Jq+yTv zJ&nXC=67^)QsWbG>0h0ba^Hl-g06x9#>rY*%O4^M@6^Y45+|$+>q$6W{yRp?q;ciz zU_;dR0S=+OR=l?Pik8iBlug3FizD`)`F|VLU?y^Rr?!-Xw(0 zv}X^wA*NAE{=AsPzRBJLJt)Z`V4;A^3TV`^*0l9`I-7E9Rp~-eKaHedPfAnLFsd_* z#H3f%q=4yvn+{8P{FkEAFY3P$dV)twX4^*RXPiEgF#4o(VGl#;vok(-&a5aZij4lG!MnYrV z#zel9_cos%KrVec%i9wFn)&(p!cedFV5kq6!Bd3Q)gy|FMOOOXtdgl;AM?3wD=P6G z9@>**K>3t+Gvhot`G8|QOsQLBJ-77;hG%DODaeK@qtn;hDm!=GNIOa_U)K)awo35x zgqxH0^LrLY{@CaHnZ`~mmT{Ikmd$DJM-9rHTTxICU{hpm!cQ1(t*cGM;xI2_OE#xQ)vj%lx%^u9d+V!VwBiiGzkStF z0#Dj%1aeDU>vbBBS$>)sd9un$rOWk=r`4F1Wp2goUE?^Cow;^B%Jm?kdZ_&IOpTA^ z>YHQVR3VFKc6mF?%sy<+=--HZV6ik&sbId4X;{CVk~K75k8NK($f=CIa+)_4qr^4r zex22f+-PFL?=|E$CgC9bI-928Ik*ccMWl=)eWr@Dp`{b^!>OwN1%>{VHFA{1KV@tq z5n1Gu?u-ZnY0(Z%a$`g-784w!!t*GvG^+)j5>fHpfQVpwE&V|1U*J3P<5kKvXotMy z^rZ;!Jl$t`ORiSDnO@)Ei!Y00#X!4h8aKjl0kMT{8>5N;H18N}+?{xKIwORiE0>{5 z0w>b-nxgRi!dic0J87ztd6Pd|k-MY~yKo2M+g6wR9|IpioHP{nBL&ZHCA3g%GPtWw9JLNo~#re z9Mr1U?E77NI0xBXK4LS(FX1zpT`E7cOi*tFtql#?T?qjrIBzXA9gI@acCIO{DU_v= z>xNo<*Uk4Nu|}j8+4PgBHUGVPI4YKOqoHI1CMdmli1CMi86z~Nk znm4i9Y&;h_;^k=C*An5m#GWx^k*%Txz}R%Ozkj#cUtC*S?&D0Zg^Jx2Y}Fzwh=4Mk z9fhRinPh8=zK$^EFoF)^j8e$?>Nbb6l7)7HN25SxCVkeQJWv|pWwn=Xv1lZW@>Vfl#%OUd z(1-2lK<#+3O3`aCY(8~4+?om&`x9SsCVKOAz_mtSeB?kj6cJ^~0D<~MXb6>#100i~ z>;PsnDo}j8UfH@86kW9BJ-tkjUH={fd^`{btqL~fc)z6H3>SzeRU^yteudvD%aJr5 zdb4I2R|Ca~$Y2Q1|@B%!-;z_E>**Na$=Od~jbOt@3C2I$Mhw?{S*#J~R&s z?J}y7MHgC6(%QEP^ukQs@xI=gBCr3Uao(()v^Omw+4FW!UN_z&tzh5}b2>QNusw`1 zbd+ke@&N{jqM^B+h%ZZRu`g@$AH~OMej1bO)N@Fvcs%vhIh5wA#X!|R`yM#IfdVc$ z|G5HzLeGDqe^T|906#>--NtF5ALoli?{^l7;u*K^AdLAb#cEQ~nyPA#VqV*_436i8 zdhSu*wVf&DuVKa-a69T(wR%emn9taKUJJX5!#?w7aDIbB#QmyjXj{Hdr^iq5e+j&~ z3M9GHg;U3d!-bL#lRV)kZ;@1by10YRm%=us4c%xNf+3z&fNfiW<5VPUe8=_ej_0ld zcth^xL>2NKk_h!+Dr1^BHX7GQ?ofSAFcr{GIx z&IrbfojZ4q>j9`8^zy-(ho3;!)}S2Kr)T2gLUjZ#MMa@Hj&GsPX7YTOg$TsOlQS;` ze=T{L^58M!+O*5`<+Hy*JUc800ib-Hc9}p`^|Mcoqa+|yZPfwCb%+)!?i1HlZ|1;{ z^EOPj157ftd6tDsQze7tjJ5>Vh1qIV9f2heS~zjlO3)T2s3EnRcwaB%bnv-y`Clj3 z)o8nKfX-@F@eui#T3p(TTI1m-UN%{)R1VF-3 zr_G6Uix)oyZl8t1GyaT$0J|WfhYM^BBDq{8D_F4q)bc;W{~s6UvtHhK`CsD<(j~vn z_I=DUxe)=|o@x}Dv#t<#bm8Y`EQC~S2VZKxsrWDW%XHpT!1A1s7E(*~>zRmH|IvJB zyh*F!;-I_%^Y&23)aIfGx7lOCXM7EUEwY3^Ut*q2?)==g!~`UC)zYPe%{l7Jt>+Jb zSLA%5%^@CLn=F-R;8f_nu6N?ydlgnA)U{%+lnKGN%{$O{gU;Dr8yav_@6J4t6%MWQ z-R0L2_a`><{2I3!@5|sYm3gej*1kT9lZ#I?t3HO<72dr1WwKAUokjCK_a6!etu&!t z<*a!9J(gk{8xKFD(cVlxtAZMlNO}EtD_51oQYTbA-O>V5EC`%{&iMi5wS%{Jf0vwp z8LAnl_^saVZAg7;j!HX^r~5#f(?o0|^ZG7p`SiDRLD86(t8rP2ywxd%o@~?X?sl58InXV3qjN`aS5%O(|}gy1-k*>9JdaiD!GlcBIRq!LJ6Lj@Sf{ z`csq1W1_q8udHgOFJ{&>41X>GIf*MsGlPAH?5&MW#jAUDVEb&&&5(Fd?t)Sal1&Ccr(zbR;UVs|cggavTQ} zHus3*(;klc2&XRGB2{{Qv_|w zFgP)b$4yZ6TI;eWTf=~M^4kpT>QGe#tbc53#H8hDSn{O9X@Pzm>xXq9i!|ZV61WN+ z-7M1aob62(rJUiZKZJUY4p8QyOnLkOB+8-Q_{u-ifXUSM9rl-_2LAH5NfTQPA;6NnzpQ4>aPOY4XSNeAT;VY^5~0-G)LYgGcu6?8e-m5n!@xlCu#|C^RNlZR@AEl(sq2*9LbT^z)2+ zL>jCXwgIt^f;+Oo3*W}MusfV$WSKP*6zbNRuBKBKGn!Kj)YoTsuR_aPjARE93DxUU z+p^`ZU|w360||~(QuR+s8HLNuqXTWbverzf@x;r+ZZsA~t#mINz~_^P|0K9p zoZ1b^=&i<2cBe>Q6RFjo{g$=5tnnK}$Ly%#vALUz45IL!hU;obR!=11{{r2#sqE|P~QlRrA@&{ZpG7!bk-i9a z%fnnT^`pdeIrp1svWaK(P=*JalYnpK-K@=wOO&*g5T#Df|H0OKhqKkkeZc7+s@qoD zqH34gv)bFPq+-vsMKuUT)uy*9wQJYRtwcnO+B0(7jo1>3phk?Sy+hLX^nRY_z1~0G zb6pbWBDP4nHqXI@q4cRzFyG^y$P+AF`SeB0r2~{&>HlpG3&P2Uj@W=ZI zwlg1yE4(^(Yc^AZxILYdv^6hC`mUeqEu9X}QQ~s>Dl(*O-)-AIY?Sw%oUI!$O7g-8 zrwt2V#}Sj<)7NP(CfHF-nNlUSM*=rG(VuE|SWFgeUcblnv_m90IenDZVtgxI&)_~b zNi4fBi5+2CV-n5^mv@|Z1X?pdDEY~M<1voJ!P=Yy9rp-Dd$ZA^tugD_#AELQq%Z?a zu~FKZqTE~Yo$+kwXpib!>APaudd!Ml5mmSmU0EHnK(el0iud{YChqUHZ?PS}<^ufK z01C$s{9i9;`6P^@QEnuA-VUMimHD}4y3%jZA%|bcrQe`Q|IOvixnrAZ913ou)cUr* za40sQIDlN_E=A`>1VlTVy2;Mh!y-O()BHR1hfH^+Y#dQ&a3>UONcF-YCwe7neJ2j=*{-YKQ0d3qK-;fUboDoDZXA11t zd*rd`8FROR&?Fed#}_`GXUy?#?wNr|jhCAo+n~Y*f$A+OJiL1NPm%tt%MN@m6=+KEj+szatoc!rML`3mo zVax41@dAIA0i_)P74rWb@v|uFGr`FHrJpo~kL^r-kek2e+PXzTg~b@t|^_}HiTaS9^jWA18lQg|%w>#qP-HV$sOC)xEtr@lV| z-I@P<+D%vBwyuDHz++7=LBPQ1JOx;!Ab5i3V^-psOEK>?Masl;9E?w$;r#b(=AZBM zf0LDrhdbxw^TPk;pIXNxr>D5ag3wqWxY8Go8;2M_0ULSJ%Rk$y}<9+Bjy%z$U3^-aF;eXmEsizvS{eF*7Xk-_H55 zv%&u`#jk$9X)yNr1nA@r23(vB?@Diuye%?Oo-saN-CnNGdKi>Yuiu)&w(5QlC0Uf^ zBV*dwqn0i7Yi9O-n#ffY0s&v|PsN!|sMDk=w%yqM$~eMX?Q;_4iI?0WJ>9Vcv*XYI z4rC=zPT2v!=f+K{?AJ_pM8as*pzuZKY6j%RxgpM;X<~&?18Jx|z_3{Kdpv|J|CV=AEDyP?5v@dat|xRG)-b?+oS|^1IEW&L;&Lw1ngsKVD%X@WXo0(ry{D(?ffdZ>yndrN? z4*kkxv}#SyWy=6+j_aZ1)74bqa=xio$qq|ArPPVgC76;Dj0Asv97sHIKa_$aIP)B6 zC0p1Uy{&x+CHwmd-J|pPzse54*Z_^=k?Icr_{op56Nd+f$R|U-EKdh#c>2T}Or7U6 z-qd95hXlh0`YfNmT1H3COx+rhTd98Xv?9fVnPM}0)y^iTYe2S-VD0eu6o>tOr=VmgP!CiApBi~SE(aRC^zWpZU&5&>y?EP_*G$G7kM44TxS z93OyUURR|$_vImK@t9Dremj&>yil!_Ty*Ma@|1t_f}^$8U!Q8Fx$IDyM&gqB z99{1t?<{`P`g}EYA`W)?XR={9--W0(gXq>@{>^6ro8sA{EV!A3weUKHybsq53^)$2 z$#F-g7)T(pn$y)I{taR*D!^HkWTu_!5 zwceL5kDUZV{1iSwQBN1eJ4C~kzEk6rJ?y+?UV2fm33^1Y;&xl27w;E$AAp>M`j0BEdX@rJE-T#BPoiz>J(O+qoyk#Viqquy4#yp!sE zkD_&JKhc1+*K}>qqimBWwx>`{y+yS#>c~rK%(g#oWBud?{%kA9c>4&puo$7y!XFlb z6>F!scz9&UZ^d#OSr~;++{ny~C_u)oOI7&`vvd0O{RU|>lh7M!C&#yb7{{(MrA2qI zCr|8bG@nDxs*5z?v+9fb(}GR8UI_F~JF|)Qfgu_Xvz2>mCY%T~GEDaU^6k}(gC0lDVW1+>P; zynWcrIOxWcNaMjOJFcH})g3rv;ZVvSXOfkK{#UP|0k5t2XPgL-L)`KZFI<#& z`R(iy4w}2FXJD0?nV+=xZ3f0;Mf z$vDjw4=G_9Gwu4+Q^%T)bAKB?TR<}*N8LSgF;znfmHmamtUVlcA;#UX@t7Eq8{ef0 z(g@;xiNlFM@HTt_e_ZOWNnsQ1Q}yJBP4$F~c6 zpgFq9F>cn`mv!UPKsIUJ9sk=$3Yjw>akYK zK29Sy3sI!w@c`=!LX_4ga$1 zG!8^-)+7!%rdSo?G407zB_*(7ExB?>O`*l3(_MLIX4t*ge-#oVJk%X#7Se-hr1NU}9BJVX0e>&|(`(#MZ;xI&N`OgjAg6IT+7 zp7V=j8_9T1 z?9U-CDMeSOL~GDnRQ6;<;{kRb^7V1js~_c3rvs)gl_Hw+zWjN2sKUmp1p$FqKWaDH z`okR;?9*d2Urx6u1z$=R(mRywEq7cj&<@tNdoz!GQRxbkl{w~;0o(t~1Xg`t;n)*+ zS8e)yqkQKNLejMjY5r3IW~xa)n(xpn=m|vr>nA@YQlfXuw=N<9W;Cdo6LdD?e~v|v z&^_k!NB{f(|Nqm4{>GTQ7j!>ga0T7|6PN;cdfPOdutUx1&B^%6NLomyJ38L`o(YD= zH{NPG6iq*+c$eq;cxQ**ulMIcAS6SEwA`1cFXlv{#EQP<^Wy67e^DM9TUAg$@hoEH zp>p!;VyJ;e@li5BmrHdFZxoENBu;YGzTD1CwO*P3O1u>FSvTnX_vepJsec>*bUxER z5mrA&>UvTI41+}s(BCKZ$9rE*iK4*UsVno5ZsE9^oGD=sVeMOuHLfb3tUHEl(43On zA(0uoE+&QBBAl7hp5x*)eLpwD=0njZ4vwD<96+`}clE44T|Dz0H-fxv*L~rM3Mpq0 zMoHy?FWRHfuXjsdVoi{fg(Z{vIOh2!XfArW4lY)-l0UxxlEJb6z|L;8E^s)r&m*o= zN-!s(8eE-|ZP?nKk*zaQJc`@-1-G#=!3qnD7>q!?&dFPzQ$d`aplRNI4+1vOjRE_C ztYXK{KX7_+>nwfp$_O@T{7 z7@@1Sg8F>};0qQ6huWoA!>^4da^E)iT>zwrr8B{&$#PN7YpW&lEpBU)>^zJgyCFAJ z_{>r+jA8+`7%5w?>3*vcf-W$*q~YYvb{@7V*FF4NRim+E0jD*0>)H}RYy+^4HQKXc ztt@$4o4x9gHP3iB$QJ_KBHXab)kVsQX_oR6s>j(+yE=tP{@mkx$b3r!$jpkO#eiQ- zkpvC0pYIniL9e~Bn4)J$Y?Ko=!etU73G*A&9Ls5Oz>X++dsDg=h({>ZyO%!R3-m5D zpMgGSYQCPU=X`NHveVL`=~Juo2a zmm2nzff&JY{8if3^1&+Fs5j-?p+Lp1`&S*3uh=iE;*#8=&5r2<)2!OqVyiteM;UE? z--p)#^=3C}$k!N>i?;4itk;zE^RFdmDK+G%UpOd3R@x@<_iP_a>>0K-N$ENBx&-hX z)s?>fG4w~(Z$KECy)2uxnu9DDMmsuzRllJgU%aNf#a&QlN6Y&I3^kWNohKL1PaDsA z-wKh^nCSwNO%T4Z6BElLg^R-C&5apsVh=CT9^NeQyLVY-d^!i&>tIPui2D83x5zVRIuU0cV%`tOf#73OQ>@J@9pM{SA_9 zs(}jHQ?)(@YUHo-5`Ha*IZ|X@KuI_VbI|TGPJ7kwfwpkn&(3Rnri7#l8ja zG~U;oXIF6)8x3!$X6H={HgS)RP`7D(lj$BGFA3pn$}TDCAooa=U`()9fX#fI15n$? zk9qCA!O!}~#a!>M4%-x77G_*|Q82c#f6hu-Ip{oij*sEDt`ztiRKBp=cdzW-p$Wx) ziO#)1Qx1;bn3zURENf6#x}xKww-3wod`gxZBoe<3A(o9$hA*@;x;9nq%1cHXlNl&I z@jWFD1vfWK&&fmCB2&Uvfn+x#8rxcT#EdSuf;p0MzrG%& zcMR2htIc*2Ue^oeECY0L5xzGJ2Ji6nzRGu(gpP9_XQ|uvJ3$7Wh$`zdok8yCV7^O- zfh%pTZ_sC*8@4~18ctZ&;+(NO{*sGBJ2jGwwd+i);nst;@yPmhN&O0N(Xa2i)V92S*YZ?vv{=$Ll) zcCMM!-d7q`NJOmla#v5No2G9&P7{1w!hyb1U=5a{*Ng#rGOh;?x^ttvh=irKC>=3N zVTJiWh+&0Mjn9|6LkW5smx{kY2ByM*fHs1|NlwCe7-33U3V?hmYc3u_bW zg#|5ZJN?=vGpNGt8jt!osEG0xTSNbp^1*EBf1Q$t)}CWcg-PTRFB*Kb(LjBRo)e>p zu;fW`djKD;@3aFWdbce1<4mV=WAdjUxotyM70ROuKVXq%e-j%?4=Vdcf-YZ17f)Dq zbkk;Kk9CTX-#6)HoM1>|YkY+@8fr~A=CEJe2{}JMzb#P1YpwzD2q_vu8TQRC+_aDH zP>BDknu_U^ebLKuU%$BWusc&KQPsPb+|jF}yen+fyrz9RI39gq&T(_f&Sf|PB3#2n z6-_y_@ve#K+ob~)90%T$-|+99@=kj>Rh1rx9rgSNT&Aagh6aFsM362$0KGVnh<_yd z;F#Uu^xx+bWg2fFZ$DYmbjVza4p8#_S4&35N>pvj(z7&D!6GuGUk)!OXRdJpavoz#?K-GtfYfitJGx>AMFe_iaZW9ugQIKy&^*O7Bvc9<-||<36bKDMP26YiSD^q-ZU2H zjFqH`+TDCorZAG<{IsjPOAv3$I5DHTy4riM>2xbvc#Of)`O#@8t2vCKz&UZdci(ya z7!!0F;ey}#0W5hET>Qf)yZzyd zak4Fx9^XLPXwGZkiGAZ5G*tG9p))CZ5Kr=X;!%2OZKRC0__J$LjEUZqdTZMWaymkV z9JteTCb1!H_HQUJ&O%<8?{b-gzpk9y;R&Dpr?wJF!{_oQ+8WbYOJ`3g-oAZSw?Efh z4;1)K78I8s2J{}!f0zCIaOvbJao`+y0Ag);|KGWAHoztAF`NA_O#iq}+QT@jU2V{b z|9R!Vm;d*H>vgL{t1e#f zbv5#g#h);iD+$suR7r~t!kz(UsCIb9P-K!s4Z@U6oO(R<(I9dQu+r$Z7Rb{d86Q_F zCOczm-u*>tpnk~H=IhItV5u*n<`l?$cdBc^8r?6}7WtJpakeoj=;8M5aHnzC{z3w% z;*YJ8GJBY?QlX}T!okZ_>y!9X5+%3}vKDwMrbutkHV#!9lgp1NB&D3+HThbpz=(}X z8$f9dRQ*%3tZ`*3arc!>|KnuY6@&9f2K4e0<7!SOFTKKT_kw9eWoWFjZSJiE6f`fX zPoKBc?CVf!z`f&?(=vB~1eAY((%m4al~c&R;84c;S&3Z*Abg?qmc5g0-^K5Qvq zsjW9I{YgR#-{ufJ)}qg3SZ_KCG_;lnG$g-)146NMWp#wQi<{iN35Bt zAXP;Eg=dRRij&$y=u|+=XBOk^l`E<(-jSjeDuS6_fhf z*rj!STRO*gvzPjIL~&w$cmd`Fb2!%QxTsJIJ3{NNh+#j*c^_R13Rb&PvW)!@+`pR5 zcdH8RQD~p)Lmri#XQ&FKb*Lf&FpxCSn;bJ?@_N24b8ha!Mr-4iYg>WAd9yxJZ5^{g zL}Hw}w2$wHNU3VkEv$!~vAde(A^@E@HaWCFOV_xLZSUn z+@B7w%c*p~dnZqp6f^OSxn05ZT?+CEV|KgBcFXRnKd6gY|fkKp3=0n*`irxz_KyC zg&Nmv@##S!(HJZ1pf=q+@JJZD0JpCN?6fbdjxk^*mQ|!jSoB#W%q<*5@UyM{%*2PF zcl=ep%^o+qkT#{b>|h1Cro;K$pV~J^eUb_pjO}?c)|SvjyjQhwj1*HrO4^Q1j%KQ^ znZK*(D}DN~&^|dziKpwVgcWr6KDFLb16uNv5g7*-xo!kN%#pq6}j<*47@430Qg~QiTO)r*96hFEu!plz)dzv+l~`qlkS%!eT;xD zfYL8VomT?G!lftie+8`FI8rQmo9kC*l2UTH89<>8$a^|v13MDu!d4micUIRn*Ubtm zn0A*TS2s=d%f^P>5BYl54F>>;c)<1%#Q1Qu@$w}_Cm&&aWupfMZpE5dH4Hb-J1%t%-NJk1GtL%`r0D8J98i8SN)#mrD4n%`q_ej z@?&b|v%BVJ(16;f7eFiVeEWMd8>cl%C+gKqKjnWCkza3? zhw=7L)8oy=X^D6iWV+5c)5#_IyrtOABwFqWWZt(I7VYN7gMNkkc<&o^1j_7MQq~A_ zA}re7`Zi0Hf;eJV%u+qv)*50Oje#mOo~PKuk?i)`QfF<0;Z_mC?$=N)40SV3LhSeH z+qcl|kTXrH5BoEAXWHV_op)WxSOw*z6HUO06g?61DMyjpsl-kFQ^Fkq=Z9gIxOd^XL~CminYW)i6_C zkoqnPLU`~|k9pBEt;M0AM=Z5cG%&g+Am0xpL4UO&KXu{&*-=Z6!XRMJyoXAVp{+jzwrf z#$l^>CpzR(QZs74`eI70zqcU60V34kXTs52DfrqvaXgtBn`V{>sTKK)o5AMm<4d)FO`{B>L+Vaj(Q`Wp{fBo2x-ACC89o%JJ7m<+P3`t}{rGOuX8$>nIKDTbfVk{;l zn%9k&?6XUHLfE(y4!-tK>1AGL&#FEYF3(_a1#y{QC@t?IHQE-|`efb;oSmSZr)&3c6&6rs7r2XDA zmt1r0TQR+^=6tAh)3svtj(4diYu)n)Q%YI_Fq0OJ?sAhF>E`+_<=M*`i$3&3(Eod%u>W^dshaP|E{=+gwuTf#|$;;Ek6& z-$AZV`V$ql3~qzYy0|#k0s+l<^RW{V^Gxm4l8^e4CKwGz1LDe#ynyyp?+d0?SxzJ; z)&`7SiH)HoAjc;rB&`(}U}bWE6uw%&BC{udLb_(!E?<@NIx_at(j&>Rnn$M|fp|{x z2W`d`PgH!XZ$d!BiNyfIe4|8ryvChikNPY^iXhG?e3 zhK@E{$=d_ZA5dq5P1DSt%$*GvTi9rwAI9i7Rhkj2BaqVnYHWp~k99K@-0yox`j?te zbN&Hfqsx5U|5^96?rk1Dk*nXlWrJ9rk!Bb5(}Yel=(~IZY|Ha0Vwv3;6B?U zKp-P1B@~g&9$`yp{bl)AWYzN~=$*UEh@XAKFReLHV6Utf`nnC^uh_^I_K&P?5ij!V&3NTbyG_*^4G_RJkT`;{UpA!=;^cW?jt*J{4I zrBxF`P4L|^BY(;1hl4=%%V6(arl>fvh_b%)Hro~TulH6|E-mv2vN9C|U}}Z5d-SY& zfE%Tnm91^wZdg;g4(JikP(%%Vp5*1*`A$T{k&Upp*S!WIn041K-2l3s?=_9?e@^{3 z1&GN4C_XWc90t!*S?>SB>HQ^C?hqMcX9Cy`2TA=}`4wS1Yw^U0bXbMB6VNw1tC+2I zbd6aYB%j2ShVHK-vh-^2v8T#cyOoL}F24Or$y6V039)ClLKMqEs#rP-ePr9+kM8KUswBTWcQp_;wg{*QHn?X#sg1E)=bRLL23jHDLvDsu~bDv)vFM`J(IK z&9>DPZ%Bzh!zuQ@B_Pw9QB_0efWL)#a2mx+F`I?|#*0`IDov4{v;l7+~<@_`j;!{YfZ(+Z_e zt|MQCD*kxSv+RJbiE!E8xogHEwDj)4`r~enVAD})5?ncYhbFi7gp6Nu;dL`N@< z=C_sbX1+MDPi|!^U~d=dBHlj2FF?7{Gl>{vnP*msm_Jlyz!Z$<#4vkKNtxn!ZU0@W zNNr9*)t~Xr)2`lnBCnuFwcpdu`6mf+PgkU!T$E%=DtIPxx#SAoWvy}V3cS1+4SIj+ z_TJigP3Q;8T@Q+ESdI7y#}g;;Hyo+d9&V3-&NhwfKZslQJK(pNk*c)5CHa7)n3^wU zdaOUC5vsv2oEKftuhBsZT1VSSBGnW{Ni*u+nKP0nQAq&68|W$^?G)n>h$%wW#GtpO zjdrewGX0aUFs985O$Dak^D|nS-ozjtriB|35>lTi2;_85DSy^8e*EYJa1B54=&na? zVcZQ0yih1fEjFch7XZTV_u=48_ch(OH~j6c{>(P8PAs!jQBhqS?{${~$Om-8kIj*- zHEt;%IYpbAcg>p?7?$g^%l8zI`vMz1euESa)og4f`h1*bRo7X9awf)F6(~>cn-=1z zZb%^YY0YOh=Wb1T?!JqizDJA)XN#*JE+mp>?is86MJKkIBy3c$m`2(53ItPQX8|s0 z*ZMA=yxdf`**B&5YsHG{5*6E>r&k)A^NzMjV2W2lac!x7Ifj;GRSSivHU z_ea;i6HlwXmrg9T`)CuZD?Y{0*Q2K-#~5S6++ND)Tdcjnn4fPm!yALE`&6pySQCxtA74ylE2k!o~{f~nf?87))skigK-yRw)HEQ;Y~< z-~FEYzwgx3V&T(=CsqePa}=Ju@FP2P75b*>M#M~-$&}3e=do&b)oV$@;^sb5fAbP~ z?lTm4{<=>njXzyzUtE5y^Kfgl2QWuVG>Uc~JS~02i~;zE7{N}xWLLOuGlc%$)Tgh} zQ-wgx;9Tk!cVBQw%^J280=^F-?r~+U@(dK}9xl7A$_1zAeaNNJE~(_9Gd5r+GHk>Y z9IL4kmn=x0)ArrkW4_XII&-+vAj{-Rq6^%lkqRfWP~y8pM`eN#FJ7}Dn_R63-x;}y zC5K!#>26)fRyY)Mh971uyo`;d+SENOnp@_SDhE@F+DQxSG%E;pB}5C5+>YfGomI z)cv9y_W@5S&Bgf*$}qU+x^x5CDaW~*ztb=}+X??G|1IsH*QyF$b|VKU;b`6CQa2hF z7~eG(x{pIQVklSz(eKMY+FlQ1jrURU^G`*M!cA5Dr5Q`x8naeYeYD2oloB|-Q=Ag_ zu`ncE52#ZY3 zxTEa_zi-(5BhE0xAF*mkuKEq4wk$PoTS}UWXa)COvjjs|loNxCJ8ezre;vlD${H+G z-fv~;DM;{rO%4Gsgasoi(J@VNu=M`Dtlr%Kb-gg0B|zm-qW z9P`|*upOet$ghT9h)AWiO0NyF_z73U$I-w;* zo-Ymj%n>QB8C-kb)LZx)^f1oWgg)Qf*F-TD{8soZp3Z{tjWJ*AI#+_)^J|x}#PX$ipZHHk|>3Tc`E+1wu} zi-v?st^oVx=ydZXBn^;mvMTUPKijz=b#B<>AN7Va@q-D!>b!yXo*d@K-C?5QH-;P3 z;lw5Br68e+HxvExA2g-(6Yu=kXiwz}Pb(_zqQ~(&;|8~#_%*wO*p_)kT9Z z^*bHG_?dvY*0mZU{quHC87;s$>x$b5+~Z7bzLQ~r3*_(8%?#y_A7MI1?aT4jZ6UsE z@q(j?Lg(jHi(ohXotv(dqI$uxxDckTL$BzuSo=+D<6%k1#T2K*Tjez0^UHq;kSj$qN|a z8f96gym3O?ba`JoY{w;Pj=jMSup>SEUKdvu71DtijlZn17!9^ zcJ+l|(P#Z%mBB?Yk6@?CMK0D`W6L8$A)y8{0<4CTscPe83vo%vvyF2J@860pdtdJn z0>dsB;0#zky)9ts=n-P+)#4?d5272q;)bK2mU7c#k6fI;S(KT7BZKI!4~{<=nI6U- zC$?Cnpw;c1hy(w)-~P4sXVe8uR{6Ywb3zR#^PAoQhQVL|XCd%-63Y4<;HmsKC7-^Y z0MJ%IfGj}f{q?WYT8X5Y?1@&X6t+*r*Gc07XLuEM?m}JT<}QV|KD(m}zdm)5FrT^= z_5RHAaEcC*E!DL(R@ov&r&sFyC6m|s1?!XXNiPCKS#EAjr$W6Q=&pE^_K~rNT4)8~ z)CouH=i!<_49a^U+^17NE4M*bSwhgj&p=?i?wptBC?spNx|k8GF;44^u*+`plQ527 z#kUhYKNr7s6?vwQqTtqwt}#(q3ZZRm_&8gG92p9v#p^31v9D$e5!HiTOP}0^8z+L~ zZiAlq-v7L3$&$Wfsa#|A*SKj}5o09Uk{S{RyX$ViyW%hBLfU(Kdr<>CeWUD&er5iT_shb;Ts; zx~;lYh03@)Yry@^MXQp1eT(j|NUC6y#yRa8toHbr6+9U}KER00!u#z@rwMj&B4iC! z^y90@LM+6O4Y~Su^Dyk1oP=F<>z=bsG9Yo+d?|g%+g*)TrIt*s*GD$3CiSNJh*}Xu z6H~YR#1ixFagC4NzRh((Z&C0?I=j)e3)io`-fcMprPQdg0PYXvq)2Csk7Sx*^dMTf zS^%6Zq(ne*q8goW1BxtTP^*X=BS$F}ma$bZs&fdlaM5!}&!y{ihI(F!E!I|M+?y8M zJr)Ra0Nh?@49`W}#;&A@(EMY9s0fZjPK|NKtC;-;1}+Ct_O5{Z(l7gfa1nNsYBzpJ7?=rH_Qy(Xr0ql339IR=#WlwW~h=CMVEQx?1POz8t z7}bnjDI~~t>$*CGW`cy=SpTVMxeKMpIK4EZCP*82$$!#g7v_?wXI{H_I_csQ;(gI~bCYJ9o&mOVoxoGEX#7Vb z7xuhjLQUi3hJ1a8DQOk15^x{iZEIo?WqTN6iF`3w7033I#b#X7js{jt>3FIXG8o9Z zXCJ3_oMto~A3oO9q!arSS>c?7z6FVP5fgLu(jJ#-pU1=Zm`)WW-=wF5u4b8Sd!&-W zE_Xt)R6J&(E-h-Uj7FSp59HHdSTp5w7zxZ<^a|7wQ&Cif;K@T%m_u=ZO(idt6dgn#Xuj8fc6XMa_#l>!Vu4PgAT&nxsVx5gFYxZG0x?A5Jy zT&5-Z+)^Xn46%8I&A{Jm>si#qt4Yb4eai z7kITmGga<}83S(Z)}F3iqCozLS|8^OI?bgI@nwFGW}N9Mxb-&aDGGna&hpf~udc*P zr*!{N5C1uFWZq8MHe4_l9qQ>9uyAeE@wPHNyc&V$R$kYCc%E4flKTVd)s`ZlP z@d_w-Z3Ai^Aud7nFx@k7q)j^n>3poa0vQ7MKF)8~+@Rek`qdC}b^~EdS9;n>%}Vy{>+*|2zzsZ0c16D{>vGu(p8VkV{3BK@$pu_M zj8uCZf;?^u+f-``Qo{;(&MrEw^EP7JGi`n-jW&dtnOC(fSM~*;%NmuNu)fLMG;Ju` z7>e#e?x}YfY#i51=DYLy1giM>tk&cSZf4Du&+Sf$)zmDaHG>xLfBgn&1_tI>Tg~#U zYJ<5Cs$yIi*To2?p2Iu!YT= z^9**wX}BZA`CaoMv)VG`FQJe(vvU`Uj5fil`4cKoW$zKCB5hgwh5TfLBKBvn)Fy8s zdX+tLwiiwyEI`tmODc9GfX31N!E8vFr#vFXtSWS0eC}R@etstNq)#o6_Cu4oaoM3L z=2_omg=|h%eb&U1EvD#qva&I#A9#y#Km_SVer|6g6 zT<(Sp-n!CcZ5;Bs9K}H&4(vaO%J@hbVRJe&{s0)X-KbajVwUD#zG}~XEr@3LR}(_w zE?(pvI<6=gw*89hm0`^xGF|Qae$57$mc0rL93qBX#>To()nD`p`nSS9?p>(?H; z2YV~!Q|Ns7&Tr6^lqk8&^f7YxmH*qFZjH=gQ?mwroc3=}sM1s*SAADzl&5bLqB+vd z?b(nw_t9`wQR0e-q)D^u8e^{*Y9caFYfvo;XDjjWu-;B_5^U7Bd`k6?ukt!f=nwzs zaWlJvT*5CxT1Xq@T3%LmV0WIRmQd(FpU+eAZSyD23)20w+a<$a@kq zYx^u8dQ}s$$z({E6~Y&C_Eu)UGGrx*V#UaCS-(tgMxwj$HFbt%$>5YLm8+MhRL7Cm zl`9wq`dm!$EvL~9h~GoXVT1LIH>-UqM-Lsxdg^ZkN26mQ%%HIZmZGp>o%vR>N=)sS zqK_d;Xpn?n<=<4T4}`=8TRPhy1*yZ+;YUlp;+tlN@$@nvyTC>o+ualk{Ls?4Z zWTY1I2CT!=2+QMJ8WkhJ8+5I|X{5}J$9t`a+;1vUgZw^|)?P0~?NG^i7co}TJmuyL zA29-~lj|*Eg6_|bR4UNS-=Ge5cxlm84Eosa(pFaF#Xov?Sgkk8 zM!2QQV^y-+x?YNG6b9Rshc7pAxHrAMK#?1su+b~~={eax^`q;nh*)?7K}x&NJP9e~ zXY9QsG6k0>GLkfh3hCYPD$IM3BR#(}QI|iS-!pmL$w?RJANc6JeiLI-I1RR_4-vpR zwo;6OKDzuvGgUKfN>p;xh!!A9*xeb*%Rlg=;??o5erpPkC zPxQJ~v3-2pqJG2}U&F>)HG+G%dvDD!=vwF28~gS98D*E&j%EUNk4QO(T^s802EVY7 zaP`AZ_SMp&SZvxRf~Tsv@8Is#1_AuUfhSdn; zk%zyglS6uzeRfqnCjboe$**d~somg<8}a$@ju64;n9gymHEcABkuGb`Fc4cgO$w|* zUi2{4`tq7P`Zwr%LOJB`y1oJ>U;p2r5ZP}*7tx%~(3L`SICFxzaKS@vf|$~aINGiV zDct%EI{B_ECVp&`*P!gmr1(BUiUz=CV`!c^ji06fzoI98U+w8ZI1QrVi;;?U>b-S#-c2?M;dZXF6Uvjp;-mOV@Zd8FtO zvWy?js&DJ~W9}T(`8+#(at{~GKI_k)HBb#SBj@#U_Ehx(*s3QxRRi<{BqT*i&vC>R zKv%e6aFoW-NjqRwe_?J&UJmwp+7zcckhUkcP%0Aa`O!k-GAsY{2rg%g{rWC4FV&z_ z2A}ARvhuV&#Aq!Za3ik~SKc4srv!7jU1s&`f??l}cn_kcYD_y*P?W{8s_OV1Ihy%Q zX1xF0$DLXv3C#bUa*)bzBZ4NVZ{Dw7cVBWs6;2ululh&tI}Gj+mYZ0-4tDgDH4YV6 zLu6gdey*LZJD%Jtdp-SEy|d79T+*zX+P*k5x@aR|WPKJtuGaz zoPLGnuly0#LfXDfQ)RLf$B0Z+B2(Dae@$Gr@;vIYh-i&AD>PHQpBXfi&^XA9bO$yJ+4?dF7*b#sz$X z@}TW9!9e5GV7zHZ%4R~{&wpOkSJ%}dCH&b$hz*G?5g}<$bJ87mmL*A9p0)eUx@{^} z50BseGrB2n74U0l>nJ6N^=UD;oa}_pf$d)mzD}I2j75(Fa{bn0a1v6H)Yo+u73+)3 zLry;E%lZlI`%Rr(zb&SgLuIA4Z)aDaf3#Y)5}!xMqM|^^l_ei1_+vje&S0Z5W5W+q zGA^^;U4@;sN&@{yN?S^W%_bXD^7kE&I@}u@LtZnI(jLZRKJDS&gD#_V3PM7S_kssu z+ehm)m+iS604nKe_T;RFn(lA^m3r5sO+FMHI%K51zyII-Dd%of)h$TVptslA+=v*B zdsPiUVA7vS=qmWZsnhQtfo=jb6vBC*G|h6|R_Oo3)O&_Cl}2sb%#3C16#;3ZH0gr$ zI)Z=%2|^%9Xd_h!5GkP-2SubSB~c(CLI@#*NGG(6(uB|xdXr8y*1Ti+^6#)53z0ku7vm8v)~gz;vYQpvIMz7>yLdcSt? z{dMb4uG?H7dAMO!w-q=^-4YclcKm1C37@Kr(l}x8WBcd;Ln?Lb-mICR2kN?YaYL>s?;?1idd|LiOcadygw3JIZ?PHG1%C3 z0~O?*9?Vi-AF^1;nRj{hQ2i8_*j+tf#Q)3PKd&tWY>$tO(JBwS_3Cl%dPo=4{&J)3 zYG|0YwCR8Xlx?|D>Qu2h3`9vg`q>p2@fh_Xl+Q;_Jt55E<@5)F)=V1Cepo|r|KL+gPsfx}yK7PDt zBf8lAnM?6MeCyT68A9^~9mYSS^~^*Is0Lb}N>;zQ3e98Zr~503`qm9erMcdKRn>|c zaQl?W#uX+j)O*OVfH~H!pOe;hpcw<$Gu+0c^`3Le=&@TUQ94FRM=;R`$)M2@17l-~ z4xQPJQP0|ew#DQ|Esobb9)!6kV<_D(6bTOZ-E9)&SN;5wDC}PF)LON+g#<65Y9ClH z=R{c!`p6e|5<@yI5CnG}th+?rw>WAfE>>&1cbrFT$-L8aFT%!fI+fqRuZw{prUIfl zF_||H)p%V~$M0}2bhvzyyw&_o_G+BDxb($3GjVSN%kWz!=c)ttgY*ZH#~>6Y|Jh)D z9iDU5#yAA=#qq{^gCHByJaGqfn{?aGf?ma2^1UFr{yAMyZ)EgmP0V`Hkd)()F&)vQYZZR! zveaBcR%aBlV!oydx8C+2)SPtoO;ZEFsP{`l;I~6l;>P!baV&Xpwcz&g`W7wI2s|_c zw0GT)`n}h=Yq&Ojr(4eJqXb0T<}F0WrER?40iHqVi6kqK|0*aj|G`i&UGp#zb77B= z6z|Ok40vmOtBU?S@ZUmyacbtxNAnx8sV~!P*28>$J5(R4#e84%&3K!IS@)Dt4_Pn^ zI!c;pM@1^YFZ%VY@TAhbBy^ghnC)%-G7?Te@RZEI*)g|?J7uZ|F46Se7qG&RVvNL!Job~Vb|$5 zEiJh12`+S4tC;uO-u7RZ?63pu`iHJ~tA%P#=FTUZC+(763g6n{WZwDF%LcX+3=w7Z zZGAYN?@m?UfJCu~?>N^VzexVVmHKN|>UCM_PZ?`+KyKnjVkqX)SwnhN{qmRxs)A(Q zbnB)y>QMNv8J*Mp`%dZd$Ph>){ceSj3)zEWaBOD8b2ea|+NK)>Sa`@<$2uE4--O*Y zk}G(pn8ks&qZ~t=N`}9zZRwzIb0jV>ru_!)4SY#$h6hfg4l6#jXgyux(@T1k%INI9 zQN&b+8+Tf)OiQ41XFtx!35uSC0OUUODC6LxujzBah#*H25+YZSpL@l*$8Hi_klk3g zIz+UV4ob79>j3-uAao5{YmH5{$Y0-no*mz5BPot-=g|yA%b&K3ubCHA>gn;i?4~dB zN7Y}}V$oiw#QD#^=lSyz*ISshqD`~GN_R_400jSTc907K3b8XY5!$wl-8`O@P(+C7ffsvgw{|ABn<@W4@>dB` zLID{JnAwD5_W4cw8Et#8m-TffpCD0LvDA8%{nv4f0dU4eQPiy~>cFK9o#{5jAtXLD zP}1k8HQlp{_LIZY76NuEuIxl*H7y|u|$=v)Wj8&CG!&Y7mCZzuYpmIwm8FbioQY1Y8*!;tB@NbEDjH$LS*F8y=w z+WQ~B+IqeEZ#DeSadpItOKNNx0Ed}ZrT!2%`<;L~-)Sgmsp0&taxrz{n?)X0>x13wu)rk<)KgcE9DZT=@S z6Qf}%2%}jtaQpKBfZ@VBwalsBOa;%z$bK_dX7AZz&!EgCBPmvP{+RS8=>sP@_UE3g zEBf6=d$ZrMNZ(4ftpy}QTUEM(pgn(bp=^G|3J>U0)zwWK$g1^HQ#?h+HRT-~AEC=S z0?0RE3g4ZTuRcU_Cz73Mr*s_8Lg~$6*Yg%oi{8&GoMJM%gEB28Iw{7hO(otc!**qi zVLDn;xuTtVW@(}b5{0gRYyE6sZ<>9%Dnm8%Rzr6XQRZK+Q!3tg!z9?iJQnPSekYTq zNGl$2TvPVw?_KUVoPHz`_HsxYSCPV`*%kehjT0{)N3_|fLcHI#n-%QDS51=u>YnY! z;GOk8Ea**?8($REVH!jsUU@-u{XGa{)JciU@L;EJ?Gg24OTXsx$j|kmesojU`-l!i zk4R|c(qXTt1rAzrAgPBGc0G*>5VHp7pKQ3U39neAqN2rN8T7o4+wtkT(DV?V&52&W zp!@A6b&fgaH4dIs^gdEiZ`kxp1&mo`6O0v*C!DlZ8F!j?RFAJuJ&s4X4yGYB3BLBW z7AS8o52NPdiSd&dX~)UIF{^_7Ss~`h8C&fe7Ih8-8CZiW!+^4apqlrY7JE2_^FBi{ zOhbyEWbsnfwTI*^)Bu`@?%fc(O!LWo+lqd}X``BUu_`z@5b568^sZMBgca+68^%zG z8pgT=^(l{Cw{4?aG=o7ok#6wNm)I3kKN3M@?sIX}MrKl@>eT(NF%Ca>sq}CN5r!0A zqlzonm^Pi2?##K@T$-5Sy=)q(We40-o%u^qP(=px9n z7@CO#Wivr$=kMEiEucaik-be7^u~Gu&?vGnX$!osD}C-jTP5Wm0o&ptdQE?+`UPw}zds-n!w#kup zftX&Dhn94UMs`|xKgqM5b)06-sThDAqgINgLNSY<3Y!U5(s=3uZ_by)I3#}t6?#(d zSd3dgzHDmJ(P`J?hv=B^owYlNG`?%c9GW=yeW-qVrN+A4G5!`k-FnFIxH09_s1|H< zwl*^=nOPrlAS3cYvk>47G;6+P_kelOxw-&kisi^oCKfNyh+baKiDG(F=SVr1q?_*0 zOC1+TrK0sX)z`fmsmCr&n%7LmtTJPiC5;=zPIpcJSlSL+#8&m=L*(3O2Or&$;$>Zd zO37#EE;*=w?ITGD?`KHX8ZkFR!>*J~dShkXEon*Ih` z_x<#YklLjz>AAi`LG)}y|JY2CX1QCkSb%t_j7H~e4S(2 zc#mCg**l49<9+5G9Zj=`% zXShQU?pWR=Tw^zezGYQHYfi#ePeZ$uSsbELA<9&W(`eH`CS|EPR1{N(-)sqGPQQul zk#TSb`cz7k-r5DTg&RIG;L9}1#J=^DgcM`Tlu9@=*qF&Jbi|;PlGRFp5#NqgEY=9xewB4b1|b(Da`P!ceYEewQyGI9=G! z!pLdv4yi5W7SZtpJL3&2lKv@{2Boyu`#D>o60ykAaSy#*?V9LuCywiN)@Nx?!W)3z%gO)g1?` zhOa@r(b29UE4`b*ax{@7kA@;jI%m>nBG1TRypauDFh$CsO;z zS1RN*O30M&8M?-U@v1M5KWsG`eA4U8-57T47dFPXZiloi?6?uZvc*$_Q3(o>3bE@m zQQz5{W{ttcj?UT7rWX(aV^J#z4*&hbTSGZ4%6Enh02J{pf#lr+z*>?i?0kV2-?B?P&eUb8=H7rVGymnfUOuZadbG49 zwqGL)h%Yb`4~Ar494Ep~&^M{NM={4#-Lfp5T=o~oq|K{->k)->;V?bQIdDZQ+kEpV{$vqxsBg;x=$3W{j{mWLKISsB z|K<9o+o)MfiNVR2)zu#5-snF7oGPHAPRNJlHonM=c*2Tom0*1v2sgt~Cuq77{3{a8L>9G9GE9T@Jb9~V zcvC~EmSi#S8FU@bGYFP4+0V5g10rD96#bN1StcNjNq#LD!TZ@dAE9|^002C1xh$5+m$O6x=L=N7p$;dSmc zmr3|G9~);h32~vwBeO79$97ygQ>UVE(?beFPTPIV8N4i7|CH@tRIJbP6caspf^BsW zrk9g;!;G7fEv+lin1}mHXg8Zf{cFD6J?hN*us`_T-{h0I`bdeg^3MOX4cCm1T9ez;mdw zUoHneVtD=7({Dkxh!L8TtL0#0jg6hUaGsW!IGhvKp$u<82i;;}_5Z7wAFf^fJ@#n|Zj#USx zv8iX@W%D(vKW+F@Dqz!c1oYf@?Lnq_aHnDA@jDM?++jzo9maH zu`yx`*`ryyK1}q?<;G8B50`}l&;a{sJG3rKi7^h7ISDLYANtV>TO``(gt%Ux&vIka z?-}R~WtlVpR-nv^?D-uPt7^^%l=H}$KjmHl%$!rM;gMX%7vxO%5aruG9vo!rip84{ zk08A4mk&evdx9FetNlqZyz^Lx`*NJ$hueWL%@+7{ci)#ibH^mWIN(#YrlB->LE|a* z@){D|La!xbpbP589gW*vmraR1)|#e3?smcPZ1SV4?s_&{*8nM!x#abamz$;edRo@z znJ+k(CXZDeJe6c2Ei0DGO&4leYbpzEH?s48H|R87yIDT{&u(bcWhu+Sp)$u2BhuIS zZe#f?-}-RYX8n1qyJMb6vafVp33cJ;!leBm36g^~5}#Ub&S)c?x5O0Yf>r3( z+hvuL5zxbmG>wX!wU1cw(X$ra4pCosNSz-c*WsZ)5R~6xW_gyYZ3mwkeC64g zpMD76=|8zpdSU@w&)<*Kp~PfPMSDHMrgr>Q{!9$-iRUqwK~ragY9G9*nv$hEJAZjr zx1pO}#KxJ1q}Pc^igoV4bms0tDIPh?&qqL3msa7{!+489TmFDM034r_eO$X)%QCl5us*;2CyZ8q7YCoimw)R;J@RE*<<;{dkk59Vp-+k;!mV zi$owE%DTtas+xux{r-%WcZ%%(asVl%D`(&PO9?zkwaQgv&|Mv5Lv@z|G$In8s*}_Q zy;(Z)^70beQ;wtUX{qyodbVIZO?gSh!e61s_J)AcToqQflkDT{Cn90|t^&1S8zOaS z6MxPP)6(~YU+vt=|9L# za|UcDjJ=$q?y%5ZBY{p^L<&ZRQ(}|h!beqTrJVbjKGV$WL@hFvzGziE8js|k7UwgG z#HC%<)?$;?=ktNcb0Vft!)cNz86Y*)>epsc@OJ)UIlo*g{rz_tU@g^wE=D`k&#mg& zrjFSl5KP4Km?@hS#V@EcX0=k2C{_uyf~@0X3dqmS*Ew2rAD zn|t$y3$`6(z8Vj|ky3q^+ph9?KpP*By_`uByRP@>!{<8!LJ4nz%`9fmPwaXeUa+cM z>Gs|BB*b1+VcS(YFf=^j&|0TPSoYfr^z|w>zvru zEp<>)K~(MT-Une7Afj6rQc9%@f&!2D{+NSik0w!_Qc*U>I_)9JAKZ5OO0Z>ib7Vf% z&j?ZWABkV$MD8@af4ct9|2~K=z4D6w9pMxwExv6CGDbaA{E!UUg!(&}h7MZTgGXk{ zLNip9rLCKpiQMZ7m>$>3X{la?)!Cop5W(toAF_X4M|)$FiRMO%c*e=+t;!GcRqG;H z&q+1v*l_z#vq`hTT>mfofp#?(%E|m&f*4y%AhNl5s3h-ZUCmL-X_Bu294!S+w)9Yr zl^;N}jU+4@$&`{Mq_yluXPwc{B$?KL`2EJ-;oYdqb2F(6i;dyoiXlbb^pFrlkN#T? z3OdzUvIlhEB0Xp1@kH{cAYRD{J*7Kx$tC2dSO~*=`9%$HV;)d}P9#=u=A(L|ak!^@ zI1RMl25;%(t`N%n}adD}SH z8sb*v(fi%6qp5OS%GSX%Af00ICL*7+dzDNhqGU^)p(X|ELQ%&z-N~BSD`8<-pZxmr zEC|Q%=BHuWu&~Jq+wqV$xKE^1ZkFa)**p?^I|_d)pL^RdD>!T(Gnj=QZTx+i&Db%V z-HxcSoIAF8!9;pU@Pm|O(X0B3oW`DTNP23rZN&9SjI`-*mgMN%$;o?x7&mX3Zq;P?wVkNg#Hmox86NxS=eZ zu$-h5$R1e=A@wJ8u)X`M`n&<>k!HeJj=Q}3>22EzCyn%?Qd?X%-iY*-% zMroEaz$G%7Ujr|C5vt$zy17~<*wzX!3^zYo=sP&d<(_G8yWuradEgN9E!&*v7xi|z zCWA4z)hY0v(;zYkJ5C?UlUb=d$;DyoaUh4Os4GEUIk6VA_TU|zsrAa)f4Pj*;txd9 zG)G-`cj0-ny|WR9QK_B^{sA(z{;vWm$wYZ%JQXP&-_(=!FBfQnA$JGoE@UrlO50r~ zqQ8Ew`0oPVt{@{cKVm#|mQpmS9S7Sb>-EE9Yinbkjgz5xcJ`D`tYQ$NuP?)Ta1{y( zCx@8)XpiRi1}|EKeVlaL_I@#TtBPpF)_|gsArjRd7T)r?`V$OC=Wu;B!_t~C=X~1u zH3B53P#1Y|XLj~z*jTK|c}dRe-61H|9M>yp;>z1veG)sqs?&SYyPFMS1Ku&~V#@XO z45^TP*`7whAcOwAjxR6jd~#C;a0Vd6v6+8+$Ci#vxU(#xV#D!|zR2F-b{Gg-80&xD z?5MM!?YqBdM5LXxdzzzl>RV7Mwao#&TwUL{MYZVwt(hcrBqr_`rMs*8&4)7+VY|MW~vY6yVLl z@Z%>3tO!ua&`#Pxrhh}$xPNc@;{4Glq|T^`UOUE_)7r}ruwLF-+f}pHdgu^Vhx>?W z$+i1l(W4kB99BdD&F1HB zE(gvOFLPYfWn|V~7|4XZH#rb;aHubwJsOQ#8)lnEh8IMJ_Mddu(jdkIea!whHA^J1 z(_35_FC8R2>egu!m&6@!vH1{B`E!)Id;Gzqdes9p>?l7AbCSJve^W*1thu z3}}iGnI_B-|9ui9cfQTADe}SgEGJ=jcg2bd*NeGV+81w_)XOV6W_pzJkxqBx0BThj z_Ap|BrFGzs4A0F{)ub;)=*0TBviO7XVms<+0duI0!;n#2#{6NxeZX-`?%W-956Y;U2+1tL8HeMPxV|ok@71yAxD!M@nb6dQVExG>BcKyn{L(gYcj`vo&on!h(Emp)z zZEstT>q^*#dO2x6paYtnE;rRkKu7rCJ(H7;baV^AXxSf6x`*x@v_Y*?D#z6$2M-MF zjpnC+#MPK>AUZBHmhfHdzv+=C+s}a$X4E^Aob28gN6A6+6mkB5$las&>sb-ts7mQI zzQ`xC0ZluHPo#;QimWwWc4GL)tbmg}Ca69VY)gpNCbfjW^By{!+wMr3+wL){j5Mlp zp4ayoX!3|uao|}4@{VHG98)7Iw%vBOF@wt&Xi~ z`s5tUrgn7h!ghzWKA91iCUMKdv0{U_mD8BG(;|k4COexZ5?~d<@T}i-kzC3|cbeyqyJmp>h&+r!G+$V6Gi*$QNx!Re#El3-A zN0|Ub&)~F!2Q}UV%j8ZA4o1`Ak6K(Z?BhOWbA{8S+r21u5E9n9t!+q!4lp%fF&UE; zDJwLOh$l+Qk`EWTCA#YKHPB8!Ov(R7f9V%8Ge3CJ<=T$Y@!Fho z*c8Khs55RC+x*KVwWt;)DbmBtmM>+DX27LrxZyF2y}6?+#uxC-UJa$AszhW~u^H-y zuMSSwdeHaq$ULdZZAx_L{;=JuQKb&f*ZMJij*9n4_v~QddAB=``rDg+66I!g7_iq^ z8E^31m)$3f<>fOY}%TCbp(88SQ!wg!e`1 zCrF45yVdL9>V>)%x9aWdxTS3DW$Z`$2rJZBminHZZsCMMJ&0C=sN9P?@6BW70RA}j zL{}Dp+$!--J=Zhx$8r7Y|1A!)SO5Hv@Vjuaw(&(8~Qk?Iarx3&%WaQpB%VdpK8dvjwzq7h&d3T~kHI<92KD8!T#l!ZwBy>+pUZigr z2S$p$GZLThu*i>9TGADKpFYRkVRl|z5@hD%9aol$wJ~$stQU0>zNPQ>=)(u#9kkD% zY6bb3C)y&us}Q}IR-ghJ*7zj7uEN&}3+Dlk$!FTCSJjKpt27&a$J`4u!B-yY$wz$( zOTZi7gMcu~9EDL{DU+-qdfT{yPrqUaLSEGM>X|p?TrymGF0U^7Ox>{A3|Os<>&&fb zAnbWQ>onUknZLKxhlS31-h3|EvHM+v%|Q}T5+PzRlA|j}de#LqsOjl5)pI#rvuXPv z(@IJm85U?HP%&NLHRWhY5O&LaqiM)5>IwESlg;oBlzM)B>LJ$);7b$xIKdP>7*NHf_4J@)}m1JLZVvT0J8f`Ui325G2?NAi)4fkxoS-o=ly)@{lmY$Jmc4sI8_NUD?A531-)S4dcEm% z>aIdn)w#dQ6FQ$w_*JLaexKlpB0O}Mq#o&-B}`Qr#L7?FcTE)JKx|2>&)tz>zWi{L z(2Nzc7qmBiqr0;F(Ir4Te=lxb#G-gF&lzkFs3QPLW}cvt@eI}2W|LKsI9;zgtCMNoEC0p32 z%|_kxp5xe1Lg~L;#gw?+OOW_V;sp>){`_b^ZS`m7o=BWAr-?Kh5-B}9z6LA7coIdY zWy1d24pOzqu4A|Pg*)VvdeNI(`-wmr$JQGJGn5F&lj%iEnC+wN;;M=Aaqo&tprpMH zAOU=V2kNju@p&oHP`k5#6<~{1%nOA^u zE6Y?my5Dx+!A8=jF2XG6A6>Vxn(Ox*CU-_W;x=FfzLKPogcG-~-xrV~g}km9dWb9! zh%$tD?pg@5BolS}qkDJ&U5Mw*m5K=dw|C4SvoQJwVs4`?;b>5GbG@faPAw{WcLBEy z?Sq(17w=`~qI$2CZk--HHU?phwOGCSDZ19Jv=af{TaL5ZHw4te4@su20fdQA`MHjaJ}VJe*v2e#;{KzVi+ntjL#&mIkOw|f8xUP`JwYcSm*CSsy{ zOe?BS;Z|$sV0OImx_+wEl<5z#h72#oOCA`U8sF}nSFc03P_N!Retfm>+@qp-58_?T z@)0{C8_+&FeeYqinYXF~Jgtc;>CeIHh(nQ;a6Y!bE7sy|eoJanSoV$~0AmmqmWY%S zs5o*c;MBj|bnOTa4!L~hr=Xc*x=*QkN#*LoV79O12{WpTgAv{YU#CSfu;#ckXo_k> z;jsZThbJ3B{4Lb~5*hnxjK@?zG{TyKsJoLM}c zjPQRaHEEfO6cTox@UcfZjNLf@>TU2(^uTx4f(&lwUe!IQIQ@rrC7X(bfG5=8(019~XySB__T%m;L{s3&4%{m+Z4;7ti(P@!hG6C*S=c*3{~<2{j9!Z&w%b zzoz&h9;+`1!{0`853C0@r?`ERhPDN*`mf7bm`%Cy*B4Y!oD(9Vs9m$YdEaPu85JQB zysb^{X@BFH<~Y6{E2li6^0e#XKX`$R8kOKV-dX!P*?GBYxrvK&&)97M}l;>0Q7M;( z_c2wR7U;n)JaHF}7M`GWEkR1M$2aQT^>X5(G$O}l^o9W0yW%W}??8KW+jNtiwzkWEEO+f8)`7B*%Jcdwfax*NeSz;1c$y$Qko zw>j;hkxB3`Q^UA5*VB4vM-Le28mB_WTd5v-5Y{o#t@}++E#;_dh~x{aJjTLOvl}|* zn^8lnS>Pp8PfOyk?Au&#{Z`}lF*9bjZ5EY)&fMen40p}_=~R;kBN_niQVX+PCt#BJ z_eg%Tb-V(h$SMzKfnp5j0uzaWng?ESezaJM;dHY&NQnSQ6*Pc6q{nUQi20O<7rf8( zfS%RS-A7HfEvKoa7gJzl91plB=>jlyf1#g2Vj3h?7|*MgJc}-8_e^mA+SRT;SL62`Vp!6d`CR9msfl zm!N~dIl#_>zNb_wS1Ko#Ifw(=@}qAawO6Y`n!>7P(sJ_}3zj)KLV>R2(yoVv zAAC8iTY5FkDn7~KE`2rNne3#xB|9VY_{S1p7WK(bNZ45@^fBLY$vV1tIBmKxXq1G_M3FZGkZANc)_YMS4?^HO%$0D(rP``VQbTwlvEO! z#1He#_T8S$RYlM7nhE&(1nE|^FeomW!!n*LV=-?N3^wAcxIwj>c{lnQ6Vm^LDkcyN zckD^REi1jF76%4E$vI8Oe*)nipzZdQn06}cs0ZIVhFDrt>$v6a&S66Cjh?Zz{DN^9 z!?Ml~X7RL>72!yqnewdRx#35EB1Na{aDXKm10CA#h==t0Q`Lfk!pq>HSb8orx20KQ zJFME19GZa`)`_(X&Z`cwLXeZKgwq9_?8AEZ*LA=R##eiXF0{}dBMxGS7>Y$OaQU+aBk!F8}fLhWvCEr z&yi%_I2E3qm84fN6;h*RgA_EAug9*ZEpq^k5&uw2ih z9tkzAnPh2|tQQ+of}1r4&^Em7h0}}MXleT_A`IAofb1R9K*ag^oWCRC1MM5yDIxhUb${8MwUQjsTW?Thqpx6=76A?vkf5}h z+s$at?>V5hL?Zk4o!=>UV6hCIAjt;*V(*9HE^M9GpO0f#_GB%Ge{yr2>A|8A|hNtz%nR#i~^C@Fqm!%jYLkfUBmtWcKG=Zv2`(|0*e3DTdFX zI1}@1kE9!PVpYa?;kc}!E$obYW;1ROh&z>%GqtkE?XZK2l#rT(5G2szSu8ESBkSVc zZR5m$xug#t0X;jjFVkBI&E|OdM=uez#g#h;(BRIRL9w9_Wlwu!)`vOjWe0@TC)gZX zCnSty;$J#gl*uliF6$xUyKBg;nC6Hsi_&SJlTdq{y$;6OGE{cRBu#Q?eh!l&CTIp} z(oz&3FDE1F6*}v;Jvt+?b~_U*YIzf*8E)o7ZuMv&uO3n`m$+UnXl79wmbvbbpH2T3 z_Ng&nx~B{5P~abmu99_Gy3_q}oFz4`Jf^r)uAUfjhho7_%%1m8{~`Z^qAsvnYJU0g z6N7^v{*oEgTuh&?*Ag7REJ?al><3nB6HUqbJ~M|e1FsuLKIrJ^cXoeU)-%U! zoXW>-rd*|Y{>x>O4=SN_=dYXUY6_WIABy|GCNy(eH33kXpQNe0}4oyYu2ORdYh0?c5UTybYMm%eg9rDhJGSg?u^Nt|GK0*vyGR}X{Kg4ef5pX>af{}m z8iV>1C}#fe@hkNHpTM8@!RP(}Ue0r!M)z}xD|pov4sEyp^GUh?h7%<5mI5WU)W#+J zQQ1i1Czgmju$Sk*adUn%?Z>smPm=w3)i+-oKH;Cfu~v558NG$K4*Zgu^2Rj5z@lbS z$Kznt84pwe?x{)h$ptk|k40*M4?4SVY+MslA8{XH`%IZz6aAxZJIHlc`QB5!14!x1 zG>RjV2a0LLHJZ0ITJ=Z9I}v+j>@`Duai>QSxe%xY0-u2D&ur`lQR0r!kDeod%r-^- ziRc$xr~7_Yfp$m0PFDe1EL zw`+D5Q~%`>mp>@qvmVISPIu;o0NmeE13vQd*e+v~u^(^%A6|o+Ogf%|+u2(Z)Ik;q zaXvFi`P2Nt>b7Ny-c1G9KBRhZDV_(uy${j9GL1b|SUyrxmUq-_=K{G@FpU`CxrR6xp<&sMC%mt}y$ z7O1?zGKNNDO(w9X-$9}m62X8AWl%tIzGgthXMM#Qtqe;|&)J?PC}-E-r2f8$!oIpb zA^!cL=%4A=5YZ1jVYf^~#g-kUo$e|yZNzPioca>dzoNp$1}VRH7Zn@n8xAO6a$vw1 zY#PaGFMdpixNB52({zdmo8sZHjBulr5J(iJsF4Q|x9n3jZmQDFs1vM|S<9`Ph`M=f ztL{!!P-c~ASYKYwR^Xjrrs438t!_%dv_b;xK&V$%?zz)Imr}MM#3=7S=UKg(b4R|jcW>HtImP?) z3%fgPcaLuO-QEU1=_<@n!hjQ+ly=mVX*xN6GhnxsctEnO1a^%dd%6u?JkA z2Xc|4#Y5ge9(Sffr>j7tX_pq?i2lA-v-b<@!a1U$kz9v5gZqUgK7dNb z$6qruVJmVrB2r=mkOZ!7{N_>GY?#MmE1XV}T~Y-!gLx(OYGoVqb;%5TWZePe(SGFIKs9J-9?4U%doPqu;IPGduo|>BOZqJRrQF; zPXPpbcLJQ9iNsCpa$e>eX0|y0zFm-f0M^M8I)QJ{IgJewT50v@;gNxWwclRLWXs_3 z6UI)+rC9awLwWsfzk#`LP`26C>Lt zRQ~0<3_vuPTir`K7}IluJ#88p7s2i_YV#p$W7{8$wL7av`80mCnuLgmx3Gb8^+7G`6NGR?a|XOlvBc92k!dsYBQB)q%3dUtsXC9 zbiUlJsL$Mx2j-OM`V?Gya*!VYH`QC*F`YT&amNkdK-V1erw4QR9fyl2cASbQy~!Pe zRiPP6U5nF@Sckb!l=bj+pGXIi-0DV3P2cRHstu)W{o{zR=H zgk6clBwBp1>B|4G{E=}3aFATeaGSrr_eY=UwNL;D+ndfMxAZ(yW)8N2poE%Za71^w1eq~6J61knCz@=vz)TLh@9221U0`2p zkx=|S-5pAN_C3?K)8s(dxnQ}xhiQmIv5wh(>18$9LYg+SaL*er--!7tW%@i=h>xPr z!Y3X=w+^K)`+go$n2IcnL4Ak{I%{#$Uc-0;MwpL2XrhR)CV9SZ&82RK1-7TGH@1~I z*mb_~7{^CSK6FDIKUwU-f-J_|4a=${D zJif#q4@Ar@5iI=EnDg7t5f7r*vjTER7U>}}_xdW{DjcXSMyYE9DMU=GEDW%|B3va< zPAQG)souL9V?mdhHI)$!-i<3#Bk&o&)v;(r-ld5}Qzzg!-N|bGGT(hfv zGhp)9HiaN-WYz7Fsf@b$!rYRnjrqhBd!k-$IZbwLmjg%q(7%puvwz}XyzAVgM6C5Z z2^fTTn9~?K$&2B{Bay*R4BHnfCfgs#bWhSzaTJ)ee;Y{v1G{w33cNYXFGLbOzvSOp z=0A-C7uE#e>!9X1XM>ZN2KA_8IZDe?X9-RcWZ{>dv(`|Ku@T&Kwsjv1?A42S1c|tT zXuIOrEbOiYA<+(7FFjbVj<6q`Z_H3?DCB(V)q0uL$-uO~XH4|Edjy5qJgulE`a;bn zW7O1D8E+!|!H3#ECk1!lnZ%xomtQCO7tV`;t z_ctxXApG7&A$^8hmOZuHHGun>f4QZ%?A&^>N}uHyG!?&d5GnaqM{SYG?j^K)2s;Pe zWQ0ZD8(Iq(*{jJrfK6Yj24?aB>!)Wv%5SdSQbj9FZ}4o486S!C3pqZWyj7SDerWVEJo~r@p2rM3fE{PcOVIOU7Xc zFBg{1oHef@Cx&`#h=&ma*5#R|8!9Jf3f4vc6O~R(or%PS2hY1HY9vuwI^E^*=e%G* z{C9C12FSlqkK_g7f;JyB#8c*9r6R~1`;gVFc-G+gsJz*5s#zmfJ~!8o8mCjR%z6M-hJ2jfEblf4Kr_MNjj7m+_;;Ki&ak==3bKbZ~nR zc}M*O@LDRz&Yn16+A|-qAc^l2j3#FS!h!3gS1(DN7Y1x*Rk#bfzFIPhpVRVmtw6&i>X2-JA0Xf zu%sI6AZExI+HSgFVM}?*Rw9iZhM9_V2(RcoEUgOTBvC?iKVhM3HHj2_zEqn!ePs9f zH;MkF>LvRo0Ak;)(Nst%8nrQguWs5h&P%e9HWo zSZaH75m@qx>+nkZAIVNh>xQO}<-3d&rM^Adxdt>-FK{JB^Y*1ScC1=nxccY4YZE%3 z%l{KGYy5KYmGGYc=@v+&;Jx&=<=17d-vGq-Q`y{yxqA6x2ITjP_t+vL`V#jS73N(J zZbM)ED9_;|b#+*{)PAeFP12^el@%5AD`a(Ky_!x6=VJ^<+H_f6Xe$+lU}j&Jf!1(C zj*!Ulr`fQ*f$@&TIje&QdX#@=a?NCf=1hlI`We=9r1`HR4f5yFq?i4@iDx@+6P1;7Jb&R9qDZ~i`A-tnj=dGz{tX&V3} z81{dff6heP@llh6Lkt#`oxb8XG2!eo<$TM;<#yn0d)Ug@pl8hGolItfNdDT|Dx(f+ zNxyLf3F|o0x;<=-gm>(i*3P+(hF>clXZn)IGD0)NBvt0eBm5vmk&ipBKCg=AF23keDl6vGVm_e$42JVR~r!(>?5cR z^5#cl_q>1)n6J?jwavJ;yR-3FAGqqu6Ed9`IK7Kl*kvk3LcNz_4XRf_m#`N#^gxCHnGs(7b~8B_=n@N zSu{el;@P9|@@I-{4JjE!qF5X?BTkV`zvr20XuPiT$Eb9)uoX%UpPRL_Ej?y7SXls> zoW#-wbGEL$4|Bup>JpZ{|N2nzZ}>*0F1WO%#OBa3sT%gZPGu@Xb?<-MGaI{PA$9%1 zN>|u=Mvv&$#Ur7)4q!8}-d-AcC%KNVH6ZmgtS=p@YdTl8J{Y1f^X9VOjz#64?VVM# zKQJ5D0`PZAMHx8*SMB@2kK2B3kL$H~`QkgXO>0K|a@Ntxc)!oW(zAQqE`ULrmkC#C zozzeZjF?L}pmfTC;aoc@ym6crZfyMxHVxD9@O>Fcq09=Kc-@Q@2}q|;>z|R<@9YVU z5Sa6T;d!qIlhNJU4=#Rbc=s*F=EpEFs1bbLWZgsy)wd~s(UHr{=k|!c`zp53l$Oza zzgD^aA3{V&-&{mH?J=8NxqSJ{(xP-&t#1@luxz~GQUKoMIA|ABy}CZZWS*NZ)=$K zAulJ)HOc!AWzD)g3Ku>df`;3_)Tf4qjCsFO^#g!Qj=*$uxnLbaPc%83e@s>He~x&F zk6l;1r$?KrS?o61v=}qqAbb>y3-=$9kCgDR*PVsIcJFmih|Y9*KBYdhp0~U(`E4m9 zXYKZT(MPOo=gx!9xv&ad0D~SC`?Oy6v#IM4ES^T3qj)T9?$cmJ{Z&=ys9J+w!IQ47 zEoutIeQvG;y~3V4b9&2QNOI2p`7Rb{MFmln^Sj(EnO zMu|K;fH%aQnr(0RM)MJ3M|U!w7%(!sSL;a{Few)!ZB<6Plt*Qsa9M__)%U3+%zgfQ z#Uh_?7J#vZ3t*LaBmi%Q^*F9-kw8QNCcFlMVkDHXjSLbnR$gCo?LjYk#aa_h%C{bZ zHSF#u_Mq|~SFB>cEbLx1Df5dXbC555O9q{jV}5epU|Z^`==%$w&VPtyJAS=ibL(e3 zXKutkvCHvoxxODH_w36`OG~`?TC1Y#7L^?DBEinMxu0Gzkt>H)6XbY{w-00p@(i%l z>f1#y6+29!@%@NXpx*fplp!8Kus;$HcXASb7V%h2L*1%j#dktCv>^J&X&__?BS_ z`@a_>Koz}u8QEZ@bK~vTdo`>*0C7t9K7Z+Si|}PAcb9Ler(cR|gAr~MP{@NRB!^`$ zh4^?KJJdi+Q)!GJ5od=A3(9d5mKP7_jQoaiOr#ztd);*01hV^uMPie7N^aEl`m0Ik z&dNU;BM}hr_t+UtyipvW^PgB+wbJrq9qRgF6iU*vv-5`A5z76=q9rY!?uf1(&`~du z{&ITlGP$<69Oz#N>U_SiwKjFtoz?^|FX=f5Owq({1)CO71-GVbc4Z{;*IiOwL?M}L ziS7s9vbZFHn*@bcrf?j-mmha>!znUVbMoQ#CVVlua;kjB?>~?=(E(1{9lhZ{4!z~r z2TbwgGS8eUO{Uuz!1Ce-Bx* zFb+kez4YK4oa~uO>$VJLts5+5tT>jp9Dyyb=HMZh{rOHIjz|^63xqDP^{Rd}ATE|k z#@okHZ2kkW744U*7T9cSscxTb@B0rK%Q`Br^$3b_l71xHsV%hCJA$2S&$8?)EV-C5 zjjd+g4=uQVz&5}3s1I%yjP$U$UgLKKlTBXHt83zmvg$Y{PNF=%Z$kficU@NYPST0G z*WWdKNk-%Xz)Y;WNA?e3)6>R7ayhe0C3z{LJN)*2 zRH5=M3YE&P-GIBB=U+WyV~spuqgnsQWu*;SXQ?#?KWI_&l$?JDkQsAmaE0~t6(i(c zc8P00V|p{-WnPe%>q(1Ku=7e8;~p$?FxROg9d#`50XF=C~B$a33+C5O9KpF6Thlu^il)RHvu znBq8DRMfn(pl^2R;qUkD(tQhkbS6$1nG#2o9FQJM{dg3`|XcYmgSM^+0?1oi7_H-&{06 z6k6Z`Gkbf}`>^iIar~8l&?R4|H>l?LsQ?md{Ig`&?`b=B)&GHR_3rAz?eV5#W3bR> zE68Oog{pj>8+56G#CBY{yX!U2!~GlG0|kO2g5H{YZfMIO3-nX5Y4LxQ6~vGSop&7RQw4rL>XCngDg z1^oSm-cf81q3?CFMbF!O8|kqJaxPB;t&L#g`PxoZw_>N|3yovY8E%?h9QR^jqnpO6 z|9t!aVt4>7m=~@_(=|TGS^PU3^78H_kc~IT-6CM1QU*X%=~#oVr2{5K(9O{cQXfE| z_6rX$oQuAZZqEGlwG;>lNC*D<-p8*CTrF;z)7A@Dpp>>s1uJ!tM30w0(_*?lnrn`W zyzaQt8j2C;D0w@r=)TUm>D*i6>a((G$0o~*5fiBwputziV#LXv8@6LfPX+9CM`QS7> zJdTiGr5Li$;>vW+BvijRZ@!k3R#WL5RHt1FOEBhv*~(@qQg(&`Eh#>%A$S_ z6j%(z%X}Z(^TQSOtjZB5gQ|B8Cr6;=(m~bXzz|3}twjzE6A%Ue62%7=ozJ z$Bc!I1EpqR?KOL%3NA;*X{NIt{GGr+UE`#evcjTU9c+NA+@XM6&|g0nr~mN@2>e_E zA71PLs5)U9UIM*L2;B=N&y)4iJtX29Mc*#FOFP^X2y0P0$M>g3UOBM}qFZow!$Q4@ zuCk%-LqIDfJ^9VTgjQ-gnB?#SuX&3ILo$W)#Z7&R*{?4yDEvb7cJ%jxo0y7$W6|+c z5$}bdjM^J^P0G@ZYPLq9H9oR>F8a;u*}R@Ns>1zvP@&#Mbi+y%!3Y)hN>_zxrz*9c zN3AI;X=-MU{yzeI1S$BFd%*e8l-O8)^6kDz=`B@fph|WT0UPAZPZ4GEU2KF=3UX%x z*j$yKm^hQ|iA8hi`AR=qpj@O30eMmu?EtuyTLJ%z7z5;K+3AD|5{$H@`S|-DdcIdVM3{o{;vW|_fg$n=`OMP|p(!Uw+EpiE0 z%@IIET9Ww`!-1B`rv&=8#oJw1ce+}Ozt-O$&$>ywG?`LWepdoasc}dl+;Yvw4*5S- zWK|sfiFrh;p=DN;kBAa}`<>0TSUF;1J{M)Hw@>pw3#{7KZEKQ^IxNsMAD&b49;-^l zD-PHQD!6QK9|1Y>$r$IoNATSF9b5vxSgOqY6#V7N%fyH-*0lZbWhUdg>NsY|(J1t#E76O8V<_I}M3Q0Gb+gNB+rsKU&;}HG1L~tZ#YCQw z_`A3-6KXJIE;9ZACsdx$G)GOY(P|DAaTij3mH%sQOOV-+)J5hGIZ2f|I`R;du-gHK zUOf^~L1Qy16+ub6HXV_-iX$os4j!WN!g_n#t`ZXb0#ot^tiB`Ns{<)7iqm+nM#&Nl zlegbso3q}!jqi1um~^Obi#bxPqp&I|jp?seXjgdVTB?cy_88CMWYr2(^XrHgwcU znAeVNU3$I)d!eBNfOoPG>Z1!6?B?C6Nwdv|`C2R~B9e-U4*C@&?z1)9!#YT|#-@oB z)I7LUUQ4sEynim-PdneCXb%-|_BR`BirBL)ieVdD>mD<47$O?EcsM{%L7@ON@U8O? zrRlHfa#9U_zh$`XENhICG*b}U`D*Q94uEF=jOJ6v+GSR+#wQy~>Y_G_uL;2mO#Orm z{>3IA@X?ZtM-a*QilI3@+bcm~fc-uRh)B;qu-Yas3$2X01Uu6qEPc+q-W9X(=}G{P zltZ8D|GRtfAe)^xT#eOAY~H)}uhnD&_e_r#(}=o%9r-&66#5F;*T6#6uApCN6Uq4I zovDgaM?wL8;+aTRTHc$3x2^>qigea&Ba%tGn_s1J@iUL)0F!N?7|cxDv4EBUExC;q zbB!Dak5fE#enqt%YF)$M%jGka2Jt*eigb#!?Y~4xSS!}Jv(J4MR;yt!5`?`58lbCQYX7(o zy;xw-`_%m~bpvcJ+$b~t*h?3Gg{(X=K8v?k6D@(>P_qO$ zK2_*Ibxt~E=@6Pt{Jfmu$Jl|Xz8!CZbl`2okA zS+B8US2$fzoXS6&^dp6oJ+XF&o-ZMN+p5>Y3SrD8-Mw;qizYrdfu!t>bMJdKpXkD* zN(y7|7||;W*423uC(@jOaGLAPGw&K{$BOt}HuDupH_1K~)oee**-#nbV^KImqv(}K zBYZ8C6ZL3m(O54?gT2Je5bl&tOqq!H9_(c4#dcc7$d7> zsqh56esO;kOSpL={$-o+eDh+ZmLMjtA*D!x%IY?p8DrFqVY1gQcyujV-_}s>sr$b) zkjSS;Xb>*C;TrZ&Y5l+BW|K9`UrSQzLMp#5$Q?8AN6TL-`2&qT_wwg?W0;iY^lhNa zA$0+$jX~$4pMcsOXPVY>G~%}tJ@cDeTbXP`T&&F7eS1z+{3J&_$6AOV~uT_ z;y*^zFx?W^ehT^mH2H7jN`kSpnx}S@kzhslR5!fZ=VzvQ?#^TMT-8@VGr%TRyTJ_> z`pK`D8;HWB4gJhS&f58DBnwR*aJ>EixUJl)i)CwTO}ASfQ@sK{G{>2~3h=vAB4*ZN z_IY(?L?W+C$81z^H95uT#{%7w?;;qJ+?jNcVOF4F8{7OgMNxD_y;0VBr<`zY-ug@2 z4&|Si==;x^uYj0Riia-?+)UBQc_xJkU@=-9E9`d|Z%ftd%c|;1x#5Vk`}@xAtgWwl zXTm{N^1_$y1SrN(c$*J&`E;9yMhmUtM<>N{+H7NTS9l( zeJlD+FzbWwIiSNFMMl-Wo>*EtKQj%C(djSzzG$%~vORH9!!?$i6l>QdH=brR>E_mf zS(Pu46<8Kw;p3EfTRUuE#+N99#&lpX+i-Hxb-A@$AD%pC1Myeb_OfYfT|=YE9K&C= zu`&MlO$~)s!Di?TGe$y6@jDFp-S@!`$p$nqE#=S}D+KmS`fp0!6`MCR0-Qc21%OV*sCRfK`qRSPc8+M z{WS#c|3LlIgM*JlAQL993?fLSYMwI7d15?3HbA$Db2-iXP+@J0(F%y|k&syF_Hd{y zJFZ_x;pN6`< zUrsOxu5&dl+B=1xK+{*0;cqgQiJCJ+Q%|{v=7ol{m_QQUE_HTY3Q*L6eNe1&jfOko zb68ci2hUpZtOvKJdHv@bn&!p4xBnEO_)e-YnOx~%CqVL%A4tfByV-K+l=(brT0uCI zgpri2tk6m+(xK9e=Vj`FSIXL{$$bJZBCXXgUY-BLBOAXPBg`E0I;Ztl+6U?3kqOI* zzLMfUU3YZzy^FwZV@U<=a`3uczr(w|r} zsM;ka19lyJT;{$qVV9bcJBd?OwcJ$XGH%%+vLhWPMD!c=Bgg#1cTldoJ3c^cM>dbb z=h?14B4KG<5Kof)OtLo!4`Qh%A7W@D(;g%j(s(Y|!>nVBvf#k^DYxAOTTVlX!rg+x ztmaWDHxCXko704O>bh-3#exr+0L$FA!j6fIy|W6zDPg*ep!{as}W1~$8(!*(I^ONaBb(yn*<<=G;o0*6)%d1Yr8(F1SoDyAiikGoKD z$1hPO{l-8tg-4Kw_nwOU#eme&W$YZ$Rfa&xD-<%5Y~sDHTfM7zqSL~pJ8X6w9oneN zFeTw1AhSNd6WHUzb$dpsQmxo?QNRCj&%uyMcZ<*BEf>QsWJuSxiCNh|~F3 zYliIZ5EI&`fC0p;G7mMce#-i$-MJP`#k89m&rQ!Ah)zom#u>a%xBj zF&A7ih8^3fY^bPo+H`Ys+fzWQS!%nX%>M&nWC!|_Y46~~dIVm?(5Xp5x{k~DeW`tC zi||bYNsrx(WubyR*;6ol8{^fV8jNp=2+H|IIK&KTSGiyY)T z!#=#j*-hemS2ZQs41^GFF7mpj3vE2Rr`{m4MluI{W)^@@suxt~SQ%h)dSSvezE zx0)}~X8XmdR?```)QQQFdqNxg%euNDi2+7mSqHIWEI|XnXHD?j4kj)RZOps|jI~~& z`?1us*pfh}E>r;|mfbmX4Aw`Q?3P>G=@~TeEB>q)34y@e+G1J#<(l8jgCdJnmXD=WmYuf{b6Mx{nOJqmz3Wyw))82PrzGc&oV z?Wc9ViNY(153xg_U1h!oW!;urZa_0Ja*`KUMLblLIW}SLvibzC1Yl^!EcR1ddYv*#915YRA%TgaHRS_VwLGlNIT0Jy?rP(y=LwXUoaji+s+N)H!aVJdDD$`5` z&-r+vBJIdmYr!u?b?5T|c;xpCNtF%lN{ywQ9__TPsWojYPm-G9`9PzD&?78XDIYi0 zxl^9>iKgp$mQ9{tSfn9mtJ2=p_R|qj#M`biERxPW6_m<~eZ?GCg9e8}2Q{P<|$>wyNK0b*md~Z3tADry3}LoZCsX!nEM!i+dd6sbR(+ke zhscwfslQL-HtVm7c$pIa1F_Ua9uf{K>a_vq)Nk6jzUE|FVQJ{6`#;cQ3|pf;Oq2p) zzl|V+E`$R!u81m;3jL>NOCm6+>GMq5;&Hjp{E0FOs!rUx=G&JPtBF~|bWb=)Aa}wS z27v4qfTitV5YvD7^8-BuGRaJEIKi;oa>I3H5X zVnJr<*6tAQ?0Pra6@yiGj(*28ef7iDKfFF09l*u!Ec5i~hd<8sf1)2Sp*da`Ud=^T z*f7KQ&Xt7SJ%7`W{Ye7{$N8uKyIsTTaPaX7oxAZ3^z<(1g7pXf=yR`uj`h?V^v&~} zP=#DfEY z+DS#iy%WQv!gW1N(-{?Xb;iZZkgsdq^199UZi|TYnYcX_|UHOuzW7 zhh-*Wvpk5+9zbD>DX6JQr*t}5SQzG>u4Pp57s;CKYM`P!0m27|XiWU{{U@s+<}du` z?p}5N{bAq4W~q+=@SXMl1(ZsOdUZeD-gRVir)V}CdskxL5}H_+)i?V;&uFIj4v6SfLKPm$z3&h; za6~Nw}9m{jWavoQ zgJ7F5cJ9eJO*}DhPa^uA9fz}+ahHBZ$Rdc1z7*^IJZ>J2GuWyS~P76hi? z5caUZ`n=8)o7N%uLp&rS`>C>sqjmgGAfV7i*}z~d+q!K>iQRM|xRF!xfcLU#Uuk(L zJ-5YgfY);hwdmVgY`UM7YnY7b27pq0$v#1fMLw8@9)huGJ~X0(45YEPS1gj<_WvXl zPA-&uKfz*8Sg^?wkL^1g!8TGw>cHkVB`0nu4jw&B{`Tiw1--tkD^>(_T%(6^uiYoI zgU9(X%W8I0w(^Q4_lT~HDIp1QR)4Qka_ck1;z~okMVuA7`>3nk9-o zOY5oLRUOH&_gDDI28+lrRtyIWC(^o!N0JNRkt1`r+g zh!jkUPKWsV$lT(J9q-I6o$f1$WG`?jXLjZGJ1XTDArV|MQ(~wR`FI5dI~1a}8DH$2 zX@pV@vT5HCs9i-M*TRcjWHH^RcgMf zrFEwEZ`ds3o_#h9g$}}|E%gINlb)ukOJNP3se=%9#jr?0Q}2~o(IP2$b)UDTu4!PD zCQBx3NM8qLd&<+>+0g{B9>Z|OpWGH!9?hI^A>58n`6j}f{vE=lAnwm9#zP#K8w0bW zYie*~$ZJOl5-xbbNGnbJi!{YIsM--fpW^PZstyIWn6FUun+#h2LWaR_?Wv45x}v{AIuPPMf&ZHnSa>qfY_AEbR%7*B`jNQ|ZO{0Dj&Yfr8eO#NO<9bXiuViHY>NCjn7b{C|Qg%LXksZ<9jooBJHv53QWQ*3yPi3$?NctlbbGjpbP(ia97X$ z@1~rkY5Cy+Q~Wd_^Z|+~@MZ%CQ#8EjC-S11$+d(S3Fdf`yKUari!tGh1#y6T@4r z{P0D_qJvb`g;~6BK3=C%XV2->`03_apjeW-2e;SD>Thc)KF_AkxKxJgBY^f z`SM>jwzqt|Eum^F!@Ib94O|+{+l&3H6!H6~$5lP(a5c)&Nz}J@C&CRlNp|Cx3$uEV zJKKSbrpUvd!#kW)w`lB}5tk{!qP{`N<(WL-_ZDbfD=O*)7Kq@eALx;Q7RltZO{a2v zB)C(XtFZFslhiC;f33VTZ%qqiD6$~u*&sY1g2aC)>ig&={>CSxw&VhoFu7c!j$QpP zz&_a%pB5y(K40xxX*x>uP}qzAJ%mD5M%4bY(0v$f-NE+&fOOX}y@kCI4|BJW+n##H z_NfJ>qbXc=;e@Txh4;#+GEKLGJq_6k2n|R- z4_5N!z-Bd~vW1peOT+BMRN5%0!JG3_UJd~K8c2#D71#!=FMkFq99 zoaO0_jDWO&7)Qk&ybp25z6v$5)3#lV2rKVAMnrsn-*$0Wky8)<*s!hE$gJCv?fnAH z`{2=fH%-MqfWHp4#j7yQTL`~+O)G3A2fraIg0xu{=6#WxnYASMsIc#yei{B^FsG{y z50C&F8moYHC@UP^stO^%+g@z(dKrFv^DaIzRd%vHaSZ`mi8L^42CMD5f&Wj8KDYnN6b}w7!eYfUf z!i;g9ZZXCW+RHp**xUd;?}Qa2n=T$b#wYggGlRQVeLtL|0-w$d-|P{jE6pHlTJP^tpQR~_aH zuYocZ1S%K5?Hsp%o}ja<`9f30H+mjT*XQVY>nviuGv)S;=}&8^qwRO8h+EEPoo%NT zB$OU-Shw+G7_a1>*#*ZhkJr%fthb$N__)50bqH@u`-dsmHN2Gi0CP8a?$18z`|H}{Zs%t=vEL`jcPP&9dd=zL?Lj&91k7mTgqzt@@$ytJSqBJg(rmt~ z{!JXI^ZRC{p|N&;$YuE$EWDP?bT>Tu;s zGE83Fy*0kH7MIh%cBvR>)!LHAt)RodzZ3M=SdNa0<%Q2jFq+*y?%7}$ZKuC|T~SFe z_3umdf88KlUQN?4jxGC&F`?b*F*3Xa``@d?5KCbY`_kSmr)X*$f!I?l<){0-I}^cd zY$9|tBWgK^Sm7cvco@{?cHYUC#_UDbvvm0T;l- zxLdam?>jTp7)_iHUA53+B#3LkB%-xTj9xgW#B3vkG{xJPOcq9DhvxkPxu@fnn)GKm z+S`ZhHKR;RgNh!-!32ms?>BDH6xV|6Y;?BUc4GNU?$(-Y0E zVW$`#Vu3-rGyR%CxY!!sdh5+s?c1kNNyMy&Embj6FV+08eD)5KR#oap%l9C*yTt*X zzBq%4Q?O2Sl&qaS^6vS&(4TM6@j0W0`Jb{Lf?E|6bC{zwGHP{xP{rcYl?%wizKNw5 zip*{_2ut#3)Jdvb#oteKvhs14GV3L}n#u^lQoxvau{R70us~+5vAXWFHqz=EVWq~8 zUo%o+$^Ni;aW0LUJw1FUL63r_F3+fT1$KF?%T&N~H06@a5h9Zm1kHltU%YPWUtzYo zp_}sgx>=*wD=NuL-xjacUl65k5L(fVa~@2_!y@ppvg9l785wA;ZjWovJn(`7fF?8N zq2kf8xpjm%dB3~jdZ`u&^HPj;R5&X0=}204>khQ#EOi@u2yss32w7i=GRlwRIt(*tB!1_1%?8+FPU6?UZXk>V!a27x$pzT9OJ~Q(KiOwmPuLWr7o`9&Yn13!=Mc1Z9}bIT9g zZ$&sBioq+B05P@K_m*hnt!ky5HGmJb6aQuYJ#=I}v56%e=xXLnSQhxET>pXJ+V0r0 zV;iVS#zP!Z#1G%ra7o;sy+V&&RE)LRkOHf19*z#NL2?CE?bhH$lAH*9u+gAlFWcxwd8z=F;dpvx%#m^L0qke2g{cT^v z?@-EZgjXU#IZR9I?ppumye=qcPoBz{| z(_GIc7?K32%H=dyqLFjGPkrJ?y!qZk$aj;rs4%E(JHX`Ypjo-~xJ5ok@q7&vm8%Iq z!jeVPN%u$g_?5B4vBd(3#E(o5G~$+G557b_GoF1Y#{QB?Bn$iNoJw8@yZ^z6^@BmT z#Ptfkj}8{OzHvnx zp5`MJzgp3d-JS~jk~#3hS9D+Up>17t!Pu*pMxG)!vmwfaXG#zJj|~Szj8bj%_nfF5 zl+K_vAoEin;^8*pS2G-{y;tfsSscV z-u=Y!wDcqTO>JrmKOgME{A+?tG(j3x`a-<>;7?tOtn8@*;$Xvw!7Sa^vvb@8tEODC zdzX*=VjVe3XkIkob?@%GgSGK8_U+O+LupMkt66c!k#<8ukE|u)BpXeD8MwfF0HWhc z)s4?D9(2fMjqc^PXWcf-t;x&Geg(sSUoJe6D=VD;d{j+7Dqh?<(Y>m6gEyIeB#nw8WNaEU zzVCWG?l>@0;PB`_P*pEL)|Nkptq7)eSlOA)I(D?bEQ6%frU)L2nlin_>v@7j4~x48 ztK57RoB<$V&XlKr-uC=Lhkv{ayON_ViGm+kME#Y#{nm{vRLp&O;Q_&$R29llcK)k( z+gpKM{=#wPWElta>q3SHasto?x6eRi*-XD+PS zvWX`dDAa~tM#vjsg&G1O5&Rtx(Z(n%GJaC?V5e~u@k@9u%A}EQRveyvj-6)h{%{OL|}P=$SYsV-(B5S3stV~7dwI&7jWzi zRf}vEB8u=VysjFy0%wZ9xV5pVHFY4px)@UN3>ZeFLiH(e>m-|mhO_onFBcCBdG#L= zA&JjYN<|#w9%ger@J>N~$u-6emHK7QdIYhzBDy8zHvlC4+G?Zgd*2;Iw&iR>&h|_1 zSN2xvGaf1qr9L(kKN4MmJ-$Efu%0j3sLqSLwXO^#^le-GNk~9Ay3y9iK+e~1g0WKB zn-rtB*u>?gP%TjtegjII0v59?LdBG{q5s`AB}MF;o<0(~lUwIJ1ajj?x0xmCyDTG#XMwZ)li^i1k zpH2!H`MP9+$bz8jzo^iN5kmyHs&M-qc#hI0isKf%X2Sh9+((3#ha+3aVz}13G1>ryT zxj8lHT)xCR;dH6i>V~i;$k0>6S7P6bDZk-om$V1=9(nicbz4Ip<>pFlr7kEj{Yc_? zNN_PrOkCrT`nsUWFZqmobpM@2Jt$vi{PJIg*H&d7gggOWDRurL5XiV%NcG$K4}K4y zSADqvYQJv*y7+|19}NdxWdMDQYQOu}@vXHFw?5py{ovN?Kkf$m^V+$j$Wm2ClH-Df$Mnnn2XGD)3+83sucmu6!)}}MPj4_CH1&Y+jdwP+zntt` zf>G$Ne>%t@q>zp8r(}EnxFc>Q3nerY$&a6*G*(Z|CrmRX1#eb`E6K!Gk$1IVDN2qE z`7ey}>-7pe{F%V?{FnQN>70f`+HE#FGE{;64sQi&T&w9G)hH7nH%?&~G1h-wF8pxr z2Vm#`@iT>gn&jcz5yFVqUMZKg(hL@+;HorLOaZI#{n~kz5-KMPyGf7J2|Frpoh0_I zJtR?2t2h^E8`l>mV?UN9!SivkW`~-!9#weuFD*w$2|$;&T&g)&72Ye3Mm)41Ac~+n zllKl1ws)3%w%hw10fw(|oaUk6ZKa1ocNsmNW0R_V&VU-CPN3k?LxkC;SB2%jD^mu>Rd{H{Flj7_ zPLpKP%Bme$-947geussVSO;3(8XVLu@<2?&M1b+jsK#h`Zae!?svHjOzN|K4(Tym` z8=oAE1($NA=}lRsc%Wc;X{v_-U4c$}S{PN`#FB0oT_09j?}ZlRbdwU)?d63WW{kk& z2i^UzwXt6tumsPO)tYH^O^ zsQlG;|AD4);`RXV?y|5fuW#n5?QwBe7iFeJc_3j0SaW9r*#L`_lB}zw9bu*Gt+A9& zD`xBB*sM>t@GP+wT(f9OOg-iw;nFiRCZXWnsLrhINnf1l04^EHIPI?rur~aT6c)+I zugj_HG`NF;f;8BJ$CRuCnD>P_+O8=$+rP++CzvS5I;}ji)3u|7Orc5B;q!wL*@@xN zR|N0f1OsOgn+3??9H@>>`eV*WQE5eOWwyP|GmmwV8iQ;GD*4~3?-u~-M^->eK$5w` z?hMf}-&4`lzcTG#^1cJ+;yRnn*t_L|zmpIogmX(^SU z3c(y$;@U+UI+D|_U#1s8lwZvp6z{J%oT$IaT;A#{0n4nB2!5dtBc*BnbfWx1919=xM7d$I;Ljc!q?LM2# z7Wl??EVV?qfOdMl{@kYQLfxE?FVzm8y`gs|QWw*+qZ$>IYk$O+#;v52c}3=J+Dw3D z+Sj~%uE%sY%VoN6MODybPqP`MsMlNYg`pFgfRa4~Z*c7Z8v!2mFCMgbxSgP2^@#PN zqhj;ANF_L1CYk)Rq7plQ#J}V}2|++IB~=AS>L3;eseCFEV>oKaX4jS2JG4#|q27Sy z`?h;i7*wU&)L>7DT4{V=q4s*w{k3rHFDU9o0ouUE?d$VCv9EVyf47T;N#zvtYsCFZ zjpAoac=hAn^|QRtgxW+jAl?$Ly#K7caA^NAjemN$3VkFXXp2xa&25Z0aa34}_@t59nbyxaZ*NgDZAbozOp zXtt&YD3Z!mrcf9}Uh)b$Hg|*4ps;9V?r*>g6(``PFf8AVV8X&KzOF6I)Ksc@t+tJ^ z!S_$mBMJxx1`x3D0X%z?o*78dhH`_V*3SJNQYcIj;4w>VaB4P4( z!J3>H-Ad^gf~nu0(wmUR(ie89G!L`O8^g&-GzGm>Kcjm6?gx{T(9Hf_x)MHyGe3QG zmfj7$)oj{U1FuaF=Q{_@cYjr-NpH1 zZTi)*DuO!Q785;HS=HNGd1;#}1Hn^(NS>GT9v?5;Z(dAU&%m2nR#iROM(x#@s|qHW z&qv8O0c0kJ(O9K3ncBEmOOft<|gjv{5n%Sv(BqLirRe|?MQsu_^vhxxb!dijhGq-ajQ z@Nc$z&~}+?-lS|&)Bn@{61xF7EykL8yy}m0%Z~eMj2T0l75>}1s?_QnEofgJ(R?Vm zG2-ze4{2Asr3kXJgz9U(WA6`j{0W2r`SvjNAD=!iSrb-kN0|E&d4HtM2K_bue(3yr z`sG%6e{~VlK@zX8T(nO)po`lVG!I3F3^8rVI#!m#aw8t+Ya4uQzFRGn-_}A46h5xu zK~BgPNQ5L_=bo;H>xQh&Db>gt>sROurezsxXLi#@$BE>2H!PmIXU9f=2{?QD13ycZ z`sh~f*j@Lp^m4Q4qt@$n5JwE1cXUqlIAg~#Q8#5~ok#2=^~1DqdxskH384}^qYS;xC1^E35;i0zLzD%gk?_xA zWxq5gHEMm%T{$@NIY_Njjk7wa^bd@Hu*|4(PMbYnG6?QEQxo#QAz{doBc`2cXN0$TcwoStt&AobTASMBb8(|p1SQWu%L_RdV0)oq*@7;JO@SG*&6 z4Ht{rd(`fC+s5G1f=HL}2Bl3gj9j2dzAZ9@Fd5?!)sw@+eI9m~uW`ysyaj@z9T6Im z&Atz(ORe2S%&lCkuIYw*J>J{nboF+7w`k$;JFHEOvhTW^XtV2Bx+Lfsznp1GqUjyC zCHn^ZcI6u4B@tn?F41I~Vf~~eX{lz{U@^Xz=S(ex8W_>Zla!C=3UqbvF@WV0F8l}L znRN46rkOs6xltSp<2_c#-=5iWtgtBrwvtLLlcEXA#Ur1af;3jW9B9A0S;hjxTQ_K3 zwlXWqyCbQ-z9g?SyWBDLK9)+$ZCM9+fr;N}m#|k;ZZEBE=Ag$hJ2%C#)25fewYy+{ z+rfM=WIzP=xwQl;E^0bpWaX-vT~H1^EGjOliKIozH)NkA4!XY*NKN(9Ti+n@Ew9CB zc=7%RdTbW7cSBsnm}%5KS298_6s4xIEw8A2O~s))%JJA3K%<&%NVN{hZ66qEMWJ+D zHqR~)=rrBl>Dt28bi`czBdkEZZLTe}W}=Oxa1sZ-)iuT2TqCL%$N^EiEG?hWkYc~a zm(zV?${=wusd)87exp1U!K($ik8L(LdOdiKVoxlW@5pQzoXU8c z%J=nD8cFYlkxQWl?*r-8Tgv`v*XwQl5DV7k!l01q1 z&bnj~GY?tHwlz#kDRenjmtx)48K$Y-^`(L@0mZ?5hwtt;^#>s@$=p9dx9@%Mi$3=x zgbcLe{|{Yf8P@ayzy0a)crY-KkT&QP1ZhxWzvP^4rVoAeBTbRrky9pXpj&xm-jL{t3(>&r@!4 zj-)%AYXT2E=t^G2;A;#W(%M{2y$JTSqvXLueml#nm;Q9&JSSelkd%C_8#a^4{IIOP zU9BG#D(nK+L>!B5eD^bSu$0zYiv}=#E+ut0$9P34w+mm;eq;vb$8#VY^Bk8oAh?vNNs7SG{dEs0h3h$QmYIAIPG@-`V!6y-n?+mJZs=FL=5ZZ3OtiIDFCRd^bI94WTHXdIHqK@ zeJAL_cmq|?xY_hmv_&;)G%$v7XJ_2zY3f(a?qk*Y_us{uOwlhP%b!Q~c&}svJW-9* z(Nw%xv8A78G(w{#@6$-Fztt7|iiu!wj*Cu)@ClPf0_TU$J$cbaT=nq6;>D^JL*fK= zOzBF^mkQOLx#e;u|L{(i0+%I3SsJ}%wSU7uy`Hgb0GsemHK2w<4(gB zP|P|uWVyML&lWm=gJh)6?=B*X9EZ^!1b&-i`3V5!?e8Ty={P?CM3{g?mRDDFbqdZY^ z?JvH2aa?<`?g4S74W;83+!*90YjM9-Es^qux|5N<2!A#N1V0RkFwtyv6I)n5mG=LF zKty*Etc$O^?M&|}PG&l={4=_dGNaDqpEA7QYKwaosNM)5E?)IldH`uZ*LsDskPtz( z?LQ%oYsUvg`P3pK9cmB2^H^Htk!y-Ot)rW3ne{8orE_EGT8BgRv70w~cPe~5F!;o! zFT*7B4)2{VIF&Nnb|a{AqR`mOKY`T`kQnguGxTIkS+r46ZM@vw8>PmD^Cy5zW^%(SRfR2Q$KUNc#ZRL!NE>Pz zeZtPwZ}eiogCppw$A|;-y$-I-@eKYGw02QyM~;;mdLs0gi(u2+Wn*nB_p`vCuHkNy zkU71%mZ)(YRx_tot$HrduBxwBBH-Ni7(UwXk$eUViMY{O%PP;^ER(=?a~A5bSCn6>cXxk-trY89+M^h-)O;!n2&Zdx zwJVIp(N4*Qi{U28`+~5HP}qw9levNL8WRsnWQRL9mlcvXiw$l98_9nirl#fl;MwzM zpFz*>Xn<%gzW`kWz2mwA`mFx8-7%Zt$B+JHOQ70-tfhWNv=iLFGe2xlF0uKn+W9p8 zX(G`iW1LNyq?5L!xKRdk!Bvqu7~e`d9kt%(ZOHvuH8{Xg6m*|A!Z+>V;N*SswP5%j zf(dn|Ld@4gwI^r_wfOGa#gL4=Y*zr zOc*u_-ymJUctJa;a2fvaD7OCXK=gKx1y*pJC>wdXmpgtUdBz`OL0M94TtetiUT>4GRSQA(Fw0l)Bgbv-^rgLK+ z^O%!+j*YJOlMtFmOqBitEDYgWszN>STS@NG<<*L)3^AOgg)jo52d$+3)$orkJ}!VV z8Gxp~#oc~urnn!R=j7jMMA>vfis}yAZ+j^qnRJyHb`LbGcg*e{*o6>YMvw{zfscrO zSpwA}f%F^H5jL6roS5%WtCZ8DF;nwu#KbeDzsuw|h%B>z%oh7-R*<-|OZ0$KQ_>hu z)qJ%|^S6~{?0#{V>%Y7F)@8;7CgtVcTC2%Ny>Y4a>Qu$f`gYO% zqX8z?e%a?vRCb@0fP`skrwv1_3*^Kjo#~2&5f6}PN!`LCn@FBuXvZ5FcsG%YvwumXey>oz$pPNNPE(tbg#*&`rIiW z4b@^E_n7fYXE+Cz?Pu!cf%WObNtLczS z5$77+E$}?T~>xy^G^8<;uoDOrWj*y z!987{E=w3r0Ol(TKMM|{pj(Q#n+j^{1rDg8m*y(xt5Laps@$dQkN94L6g5|81FD1Ya&y-dN#`SSJMKg(IUZmtl zGrk6fM}0QRnQ;x8DT}R}Q<=VGjd5PEgh_~+0 zSzt2L3!d2`*>%Ks#SPM(AF~j^77;m|k&pV{qdM9s3(S?j2Dog8RV-5{?S;SvsI5Z@ zA8xH#+qtpx=8Wja_b3WL+e^037S?@?Yn5a6F~$B zSlp5zcHf>vgBNI(owj*J4Yqu%{IB7%i}y@}ltrr=lsAOE)O*$?xwbn*2}5O6e_K68 zW2F~X>_twHYTHhMlQi3*b9v8WcDVqua9ko~U;;KbGorsQ%4iQGVeG=S)3m=9ESIGP zV70$9Gr?pIvUoJLGjSQtiM{0-uYnPQ=1s{B5S?oEI|_(Nnf2Qbobq)L=&{TdpTDm@ ze~1Y-1aXd7IRDQOGLo=SSHaVTS7$Bex1-RO)PBGB`2~Oc%}`E#ZUF$9!d;Q}%}}To zx%0F?iypI55!*pQ(=R(?t#M|6jkN-Yp#WLQ+(%Xnf6v?>r_g4Kzj#=}@$$!AM!VzT|E!OyA()nK{Dh7HUz6Xnqf zPS7k@s7`}W^OVoG9#oY$FBiz*&s+w&!7g|3?aDTg5ime7am$6=az{661t$C3Cc60_ zXV#^%EgEw7iLQ8O84YOXt~n2+8r7Wua43$O;}4n~vd2f&j2j#awo>fXX6y#z1M&;F=IA@mY?MVWPAh--fr((V(b;qsBET&+78U_-31iBoh>P+|M{uR?pp4>uc2iH)F z4>!=hJA;MkMoL~V9{+5~`;tARUB=fqmENn+08r9 zWdR9PhO(S;3V=tKF)7bQY>+jnDcR$cF8B789*)2Yt|=VJ#KUFD+X~4SaZA&+KT^Qd z=BsMhVsL;0QXv!{Z&>WxCWme}5&WL4Cp~@z|J7V;PN;jpUMq7KC^aco zRi)Vu6=S4xL1FQJ(qNZS)^*PzQ{*vc$WMz>o!g;#u%XIwgQ9&i5B<_HgI`J(LQ51% z>^)D^H1hF*)xirdq8`4`O*=Qo{z%_oyv1a-&s)8p+|B#7Ut;7}t&(zTLbzlH&A_W3 z;>I+gdGZ&P=u$f-B9)DhseO}8Yr%{rDwfRVv~ zF2+=Pn8AiWpa1C^D;l4wRh zW84g<%gn9YeU)@$vavbK(@wND!Fbwg^87_C}tVrVYV9L@n^ccSaO5SC16|*4X5cEgo zv(RzmwzrDoYW1F5E)5A0i1DDhZ@3eOI-FaKO3}90v^?&+EjJDs>y=p|xlJFei`hBZ z+{Hc@NrY+&Kv*CsW94h+mklyCRNwbP9@4$;*W-F*{4(?QEF){?~K-9+yU*5*&-mr|z8Wx5v4T3M45wG=(j5WLZ8TA(zkrhY=aeY~1tQn*;V1egqC zb*%U1A`*%1YNmWA3GK2DOwPRXIeM0vl=%`dG*gw38pT{g_d&b3_qCsyLXjC>X$y0` zt%=qUagp#WYHOnBcGV~06XCKR{qJnN9z_%2I@|REesYiD1rAdeq&@LZBi-P_;pv5= ztd~?uJ=r4sR{#f~yJE!leI3Xoyv*I;rj3t2emj?3?=;xd#I-qp(Uq!iRNT9!)a04h zm?|wKB$C@hHizI_;4-?&3OWI&%Y$f_+9OH_5l*R1)3NCYD3$SZbuu$@+314zP)yJh zO)?w)U%yivI$QsoZ>n~@zeFM}tGLSyhTxt^RFD6@z>n2fflsBHEK&@`s6&P)u^a@ywHe+tB4z#O zOqF4@3F(lFvo5CRUeQW>tnCu6I@*oe)6A0X(C?8l#JKF^ykms*FS@hRKGfPzo@}0a z3`z0W4~jq8jM$@tp4Cb6j4?^tAqxH^aSH9Tv3^S`x-;OjkJr`dCtP^%EKNoz*?6Ik zI7-1Y?Ei$@u%-s}4@pXItS#$G@O3m#FE^+ff0PbAG|10Ip7}Z#gBF{M9uUuFl5{5&4?)4yasp_fqbp(^_T$y&L>7uUaqQg+( z{1lwTza*E&9Nv9s=ksc^pfrE{Njow_od*!w4q0WOWoSNbz#pn+BlvSvjapUXlT}-s z%^y9&jwg?SlY{atlslkgqL&-YtT-1cYT2BUif|0V9#B0Z_tG<5AI`EI4vUalEFAlr zjWt@lH9a+s>Wo`NPNjp+`=~|v%lK*sUha-ngJTy;#+p@>dWX_}81#D*=-RoQps$k7(yOFo2fH4t>-O@Oa%4ZhtfaDq2)apb+BFA0rck}4-&)m}ANdECCRgY!DRy177f1-a z9P-5vRMsNS0RP^jCpmgbZc+gPzd*pi{Bp5`Xtt=0P;fdPd$w)IbOY;z=}M8(7tI+ zc7<7xMRM;JRyxTnr_7%->1<6rIqNRHzqZ1r;$x!NJ|tF}H3KoJx1LoG{vP)_+hyo5PBOolk zix#}@ymG2GtEJRQov!4`JaYXF`e?U{-D~JCK8ud8CF5rCFwTF{7d;{=a$*Fxgn=+} z+E!(B5YJbNHz0KT1hXJ9q0uNSezis6+yw>74r;sl7OlK|j$zwX8pMRUgN}|bijs(E ze05NULO7siFwn8+n?lS6>np)51g+-sTNu=t=&V)Gpn}+&dkOQ2LjDO(pfT2IOX>`~=|{IT%aWup|khFR060t9df(7LZsTPJ1jEgs z_jEm-m1ltoh*QY==6dPfB{wXDOxTvi z9txQEq%WUD+r;@T&HAiuR=dVUQulmVXP;t4;m{mov&6Z_kxo9C3R!w5o;Wn6#mx%( z-lkYaGD|+u>T!4r)?ie|=Fzy2mGTQ_wkG8lY)xP6HE9>3|NZYF%HTgx0O(=t>z5Bv zhVLHk-wgeJ{0fAHpNAi`;QAOfg{ONHD!>5c@6m=SJ@WB`cc3+TZk{P77?l9YYP4V) z8U2*2ZcSYnjmcscU|*3J{d-AoXNN0`lcr#C{?Jx_{v&ql;Ne#q(B*%b+uVcXlf4hj zI?`rN`Mi6KuGak_9V(}oWuLK9mtkah-XNqX_aj=iH(b<4jZdXswqw>wi*EXtRgFCb zQI>Ov4!L*3+2-@}4iI1kpn#w%4=^d&nbq}aS{WP>0a~($u)z3Fp=U6pHE=oD#xs~4 zpGus-Hac-qP(6d?)+f1~P51XKCc5GFh~)Cp58MnaWR?yhcl8kq8AusR#*FjJKy=0Z zY?ylj?qRC$(EL!11`@Kr-n`rk2$#mbeyyw}esXJme3Y-$SpV8dAC`i#VwTaI6=a7` zmG;fJa3Tqn>(!a*v%$=1(Jm56E-zc0BI;IRJdc3Fw_p<7ue0l4ng(C(9j~L>!30w2 z33_oex$LNm#rJrA&GMaTSWLx4SVHD^S7n9m;I?S z$U%qRkQ1c8&4JI8U_8K~0Bw!?&mm9>DrFLt8$_5B%c;l(4hE8vb&iV%rKym#NUMy+ z`9*ZI&T>P(lc&9jAiF@cFv+qQnBC4TJi5%xpxG5bi*OCWB!dBUnwmR2u;rmq|9c|hEhYRi7dKASqekSZI z-t#tV8S9^3xIcM~5 zE>eSLJ#_gFRujoSBsRE~sF3Nc75pLdXWE^?%EIHm;@q|MWs1P?Q^bPO#`&$K(c#g_ z?mC8i$AQ$?cy{V`3Nhg6B4&TWPGf4SA1h;7+_f_8?eQlc+qLe#8|A2E zuggK}`A;3;QHMLx8`%xQ?uV)B+uOfD6{Zl;$5ONYc2!mGo9O^FHwUpOLj~}_r8)BcZbS%Nuv4$Y{Z@&C!1@uh zAnHc8%?}Dpqh00KogkV>^BW@J#>$i7bdi6KYLH(q1T&KV{*Vv17OIQii@o%d0Y<-H z|2Qhkfzv;xwt>L{?u!W3C3(-z*5dR!f8M%URMags7M4(mW2#MR&OcEfpq^YUCB3pg zxS`eW&=}6*gcaZ%rVv}?b)XW1shG!*Xk&qB#wWFpm#Sm4WJ}#W0{r9y>(sg@q|bw< zqOkHFrMEZ6fcJ@GV9P!Tl9Eo?U`y^0Ga_}>!$~l|4ZElhP{wtC@>5{!qQ{g@WMSKg zY@Sgv&W5!KNG{fRMi;9qHlJ@~@1&HB@}u;}x%x#C`3N%N#FiPXE)Q2nZs@D_mP|v} z4}pnHLe=+6YH{Y*GJ;(Q=2=HXb*I=F*;XWx3>(4!3kO;$LIuXW$;{VB1Ol}{X{cPpmAXw{Ni0Mexw^= zlC(ZJlVQP-uwNsG0tObC{964FM$@?BU-_%gWF_D4)iaeDF%~%{_?03W3Kr)2*hm3p zb<_>Kho6TmnWfRMQL9AL8J8*TtP?Tq=HEz6bkzao`r=b}}1T%VFODgt-jr1`D z4b|V5b@M!$kyqX6;4imYX*kWD1!?CvJOpz8Kg6sa^^;fF7~`K5MZclJVeSNugwib|@hQyf;!G@%AU z_h7j%Y9aYMXR0R=HOe}L&7+E8Ne-)07;mCj0Uby=;i&HhCB?1TXP4H^3h z&TMthT3MiM021dxAVIuh@Qd}%NogODjQ4CfLHo;s@Tc;x#xj5vTxemQXiuCd)XD70 zFh4~-e+jhV-)_doPgMU8JR{%|^Mk9P=!hR8@9vBVcE)J%=vHZJM0Q-gMwN=?tJ9pT zDuNXeV)15DxZ+&5x0$8rfe(zs+Ji)^s_yEVX#K;lU8bu?-@Pt))9a}ML+PbijMK)k z)1#dFIlp$gE22G-`y57x7j4%M0J=%=wn>4#X~i+c=X`K_-pwnKaT+Op=y#qDbA-akjXkYIs3=}Wj2{ZCnuXuXV^(_$oJk3a_qR3-2i-w z{EC&|zcFejWmvJgxukT@L^m)&aft2J8_8%F^2K8Swco6}^lo3S=O{kpKJPiLk->5v z)#W!Ra#C_rfDXaV_Z#$jbE1KrTzaJcJ?Gncx((GOXd^B_BQU#9$*;_Bu_7hJ+wa0Q zjfN#6xtL%hw29#S1!KeZBi5py{UN`cNe78PutM_f z`4v0q25)zZfBqfwE`=n=8OBAjm9&AiT{oDH3oYuZ>?e74W$E!-uUH8AZF~)pMs|k6 z)`$AJfZ_I$p{&Ol!++--O_Z7gr`KWfzq$p5l@|31ypQKgP{nJU%H2W{U!T<&gHX&i z`r^B1HyX(ro@fj4FC$@y!3xMB8RBrUoW#IKJ$tH$EFD?0`Uc!B-yC^=mx`m>4or* zC1?stT&$7|NNm2wdZj+&AAB3UYQM5Ph7N04qYqi2#mxPXryB%}N4$Gw`6*eXl z$IGfa5Qp1}PBwn9C_P;<1(DRe7E+qqOgdBkFRw-Je^>8ZH@@7XKQo3CP=Q|Kb9H6yGVndk)GZ z{X0Rg0vw+B&Wi8X+N^2$z7m+yP%2U!GkpMh_jd*$4R}=2m|y*~2O8R@p#Jg5Ts!_f zAL;y9@#3a?GV+*?YMrV&o5k?)2+X(xY$5;1SO1_EDc}?ve!B{_?Wd+DGA-Q|Rkvq5 zcZt_E8>3~6eNd*^gk12{pocC)eqqN}7Q4v+VN<94TpNZCYM=j+`qAb1uh*dH&)=}m zY3K6ZUGedieU=mmI~exxoQclE*VZfEC??uDnXJ6R>yW2(t?PLq*QZ*QP@ht5N(mxS zKdmaOUDHdlpc4O;FGTfpC6!?C)FrGbv!FpLctJ)+Y88EFy}a4_2Z|YV?ExtH27?fA zrMMxZ$&Z{PBn=JKO&Ws3rJfyjzpb}(p9;rfF$0E$uu1YpGmigiVRBa6-J(35etdK* zCthjI#caAo?|^Yd!d^Iw=5XJZGzmnC#B}>v`~|N)lCi4)q|x9_YyfAy!UR4>l~oENr9vCiMv3o_Go=A0f9_2 z-;G{#fWVz2cGUIb%0-55vPaVy$)P|$vz7vIr)A*s)WhYrfr3Kv z4PeD%ATVJzh!d7Im>tJ`I~@j{CpySKM(Uc+@%7aVKC1dPvXn$0DB~BLdTa6D?5*L^ zv~*d2%kYh27#n%hhJ9fB?@)?xDI9EOgE9bUlPp5)cQ2U*dw~Ke(ojV zqrpN>dGx(TP0v^I!u#_p6oNt>Gs(k4Mq$f=nm2CO`DLPT$Gt;j5t6{_3^aZ2e52!R zp6%8@fWHA)Un(1_^7u{hpNb*eXpq^FVF34h0HP5>384hpN|BI(B1s+P)Yx8WecUllXqhmU z3ttmHb}((SZ3Okly z(vB&i`azu@i7}mfojhL(P}ml6V?IE`WBQoe{T_?j057-E6HNhO4*aJ!ighOy!o4Rr z$+)5l`2}ZC`<(d=a1DLT>8xXMC&1_l`j{nt_ZF^k^j6gJfsfQMW?g7~$%JhDu~7kj zfVex8DiyYd#S^+2yfoXSTdlPE{7+D$Je#=)4x!S+<*aWLx1F+HMpQJ%=WH0)fk*c~ zeBhou@#KxF%po`9p_pTbHtW*jtQIqMqn+&CXIZJhHDO^Q9}3l{?3T`VB(3ePL6>NB8Y*0dvzBBZO%)x)Ps5Rgrj-6-6OPrnQK|LgT?iE;y+`yzFPfg=ZUYw zn-k@FoRb#ynuB?|hVTL<9axLzU6#6#H`fiB{ALpQCU&#OSDZvpHasqKfyR!;Y^}WV zrBDV=v;bDGXJHv`BNUHHWM(l;eh`hS1}>d{7U8e{pW6TLKi~d&!T{;<8yL6aqn zZ8Wqjn#`g>EfV7KX>^T|8>=mD)fFqWa9`k8`p4o`<;zf)pW7_qAx`j(ZS=+-8~c>w zi4IGBw`*jvD?k z4n0ygn2j3M`7+@*5%*=-Si=^ zBhOv^ZWZGQX;bjeB)wXy6*7NyB401ggb}sdFpH9?3^OYVaD&yg*uk<-%726Qi`%JQ zGEf~E@^VUw{~_DxG;i~9V9K%2&SWwvidgaNe5oo+Hs%XnxV>;b0hqy#`(Ne_auQNBK zS7kT3bzpBh`6;#%-s>yYvuQkp0GQkRZ7n$=&cBL|(tm?`BG;Q@-$Kt8cTUt*J^`Y3 zrmq#v{DqP7LL9d)0Bs>Jzn5RY;S+OnnevHbnSBA*Lq#x~kKiP(f&;3Rxy>+ZWZX!rY0AOk*XTK z?$yP&WOhxk=#5@Lq%8q2wm9 zeP_|IQ1ZG*^0=@<9o<@t4P1dDM76grK|_V!pF*zXo@*qKJP+}rs2^kGFtCFjfun4(UW}h_W@dqY zoa1`p6J5~^Bclh_tpRtVaV{<~MOUorub)Nn3tW)UxHhR*qA))mqJUC5|byE*aao)!$W5#&hJq+reQTa%@g zRxTGdCU-0IaP`N->r93Y9YvB-TkSkV1#--uV3=Ia)N)JF=tS#W>;g`;TK@`D(?j77 zvav0|qkdP~?m$i9u;PiSHhzn0QTs4jwBm46XdZVBS28DBq4NSF-{-e)eqFlEL7ymq z6hnj8S@XRZ1!n1X;@fAeQd`d%?9=oGbEoKZ{OR2SvRLgLC5!KlyHn(^e06dh$?zPm z(W~H8=qe=g=hWS@uEu2O4+mDS1!*Z_8ijF>LYB=IQ)? z5L>yDa;4+By>hS9;pqBzoe_y`loOhY?<#+Xkx0lZS4!)V(Nl^b-0CW$Qcfy6I zxVg`+^FzYi7Lym^ieNj0`-QL6?`}_!0B-jpQk0tI_q1%xMhMU?g6_R$U1AN=Np{j4 zgt&L|9|u`{3;SEvOQ7Dw%{@3OqYfG_=!rLic*${e9dBlPCr?juP!yxna4(x zSE&T<#Ao$a8Bg#00PM7%e_RGyV$j{RqY-;e?AR2kOO^1VgMDn2)u{YlBhf<=j4OhR zs#uab-)N1DtB=CkpqsPM^=rA%u@+m=F*7Saa#=9VaFGe1DCXt4;Nj)j#i2+k>m%4v z+Rv<);!&48c^`9Kk?2N$rfcH9bB%^9>p|3m=0wSU?$>jbjhipC25e=U&*H!hS%8-GCe9y zl`To`+-mpmYW)6>pm!J&=tU#-cHi=`IWUYIy4hYUp@F}V7JM^Zha=#IY1EmU}eFA zOgnYj?=?9Gs@q!9Gw|8(bi>S^wX$2M>j(<`z zWy~Ho$rjynyh)#?B$d6@%tj0v<%S2`f(mPd^hoTP|2C~-vfk}gR2 zt!`9V%;ji@@$UisL5@m8$Qs1};tE+1*c&G<30F|JJ@%Dq!Wv@v3F~!*s|l{ec2{?o zwVea`!9i+nYNr&#Zi>7$m7uTCqsV4qvg86-vl@7qb$Mj8g7Iv1x&}=e^>Eq~`X3Y$ zw`p>30_tN*jCj4WAO}AFuC+FzS=^{uZT3~REDj6=6b-xoAg;mhXj->3sxndETBKuvzvOCa?4ly}bk)nsoesgREA|8ov7qLQZ3ovS5o& z3`EV-{t%9i-ZmcJ?XKZq5mwx6M_;CciTY)SjljMt>Jn$WQIK)wRvc$@po&|Nkw>ad z8s)K;`HALuG{fjIi$lG`3eKTo)fPg)bzdP=n5$jShuzPAQ4vuK?9?OUDa|^ zoIX^r9WK_cW>&spq)Xit`>X{-KaSF_Q;Cup&J7QWIs+SBUMpM2JkY#EdqarMn5!-0 zZ!}jv_mfD~`FA>x++{xg1s>?6bZi+Txy|ap@wud?aff}BX{&t4Y^-;@*Sfu0>Zdrd zg@D2(+<${N@Y7Bu&l7g}&DqLx9Lqkv-HLOo`m?E=faI_7vpwvx&*&4a-$a+M zSnFEFgic0@gj*h1A3WiG@Gc_$fwi@L@73ph|A8dzfzuYyr~c_w|BpwNn;}R!I!jdK zW0vIA>#4M?xf}AVyWmu+$3I~MErId#(pchquEaSY5;yIia;A)b#KE=K;Q{hH)S02|_1dWuR; zRi!cXM$f~bJegYIsv!TQiV9TTDGP&ul3Oq_wf`OZtS>{>UsKGMDn+4ODdTgT3^h!P*y^?=nScfT3dbZ zTqy|g$7Zn@dL{K90xc13aQ;O>@2rA5egR&ghfhmX65hTYiE!Gf3<_CTZDMYcD@tkD z>S;bxF*zt(=;FTB!HWhAQhxj%Z^wfV^Z-}vhHqu1(++7qUMo4A{!m&rE$e<8Gu*BrcX`9m zdfZB{A(|EH5C~rqEiy-h1S#J+^S~6}7d3KnG+v=k?P{^GQ7cDli6?I}OdoRyz8YLv zq1AtH$ZB8Pi1Isb5IWa=u{C*IpDNb2F*W2HHkrWwbQ1O|p-r>P7s0e`myENBqWB4I zoFw=uskObIOKKVO-KMwtJ89zeGTG|>o@gkuca8B(MRS6TcS_bS?cLB0Cg=FdBH9Fh z;>@F!gN2YU^9>F>uOEY)gLaVHdYQu;R>!w%`$IkrDe;#%&3x9qK7LS+T4gD_pp{TD z+6i87Hr{N*4BE$Ba8!yqI!ZBI)7p~OI; z^;7N^zlYzCY!aG(gP?bCEsS2IX%kq^StFdq+=!R$FR6L!^LFB5(yi#R_9&eq)Clp;jqnRkZiZygg%a%UDJ^Zia6D&@>3MEAsU;xn*6MrI6 z?JNRU!0)TJ(|WzKt#1D{ijrJns98jCdPuR0xn;cy$YjPo&6SLxTKp^EDn||nn&Ew-@)- z8VQ)DpV+vCpHEs5Y*a0^59FLM!waE;(`UFPxIIZHSpSH%HH0WJ&X zx#9tS_1*mJviRN@Pxc}Tdrv_vp+~|4y@+vyq);=*?DtN1vX~piFk?8+z_f~NtdAvD zVeop@hHHNsY~m-}jK4+lF#(+E;2P|Ki_*4U2CEIx{_ovsm1RqG)taWkkUJFWoP9U` z^Q97wU-UH2f8uIJ06O>QU}N$A%=uE@_Nc}b{!P6_1k)Q>7EUMr+y_}!ig}|0zT3ES z`Xw1jnI1XRfexKZ{R?aRh$80^nvqAWily*U?wqW#%Sv zqkZNUK3l@oEaiJn#(5bK`B4j%T*%bWcYXtmlW`leu|&sYFR$;wsQBfFWuQGF5CmdM znS@W6T}5~SAR#u;`uk)9*E7TN#4kdC!1T{g2l?>czxIInY6!rZDZ?3g3+;XB4<+RV z5DN0eg7@mVSo~eWokpeDl#}3I938(wRYO3k?}=+ddL8dkJ~`z}k*L8e0d~kQ5+A1m zhV>YEbUgGYn=Rp{iwKboe{tQviB;u_e6IWYn&w?P)DGZz1onz|qPQ!b$9qOoM>)dd z#k4YPI8_)+9T^w;ziIl)nlQ$M6igDG1RPAc}%Qo|^ ziQwya>+%Pous!ZaeX3*k#{u!}jmI>vzZ5CMgc+!Pv%td+8M_19nre5nUq?`v zKmuC@B{qAf(^U;1tO)bd+nAO7=phWfuzkTCtT9dTOn7Y^^3d9Wk^9aaI?y9PlWTtM zQB$?hM{_N@cgF=wxfzacQz4&HrQ1EQeE=#RnTg09k-q&-SVGRPr5=-hC0P1ZW!u>ND7pkuvL^Et~pTY4Bl0{oPgHtVqH@S!`;K^0NaB zj>};nmQG`7JA{6y4)b5WyS_z)SOAx==_*C1!QV7p370HlN?0CR5ZjKj(x(a-qC2Ox zt>*!g-;_<&C$iBYrATH07Z(N9>h=rLhG7?M9+ArZd7i)95e`5R`Y3<0%!zCzJd3mN;86+I@^2cyfAiV)!D@_xlI34}wpp!3$*=H3K$BRXTp&}!9%oKP1 zVtG~`v>+Li+!L6RKbVr{K05bhff7(+b}XW0R4__}`90DfY2i<+ee0nGXb~q0@bxWb ziNj@ZawsX26b5?$V5hESL%12Zp9o#woKHySq61x?2303?aD|R?yBY>+ntHxryjgew zM)<#1v4M*c)Ts^(XtPnf^jXJSm|qs(J%hXbwocer(xM0Jvu+UKV?B7)!=PG@?qtQz z1x1pJY^`nPE_J0ZsZF2qD*?|^_T=(s@QS@Pw4+#p0QA-6E1yB1!582Q<9O(b*86{=9!L?Mi7dWuSUEm&2q&*p^<&mZKj2ZIEn|unL9cO{A?K2kZ9i z2e}5R9S0)BPS0C6?>HG2i2xbthTZPI?nrOHny3#vL*8P>1@q$_9g-7MH{Wh)^M)hz zdpzjV;Ksb6;{MjXm&G9x8GNCD3&7g^x@xs}lq!YpGg#w30scDO%=%P!q=?pZImzcd zpZcrFMVjnzE%?)>XyT6ReZcv2cUTC36FM520gS;4yRrRL*psXqrhx$UWCtvmsgC|r!YJ1Xr2*EwHVm{=D)@!AcC$+RmB(GRiReA9Mv=*udW&6mN=dOGE zc)4Z|55EQxdk$GhBYYAX6BMbrGeRMO9u_ym|Mx$bIbdyh_8i5`dJJT0`0TtAdj2<1 zDpi6aaVl(Si`jRkTKPNNZM*yjWuTQ^p9~2fs7sR*e66~P&Qa#ZdfiJYgOW6@!cyhO zdx~*70~4z-mbCw}m^iGbZVMMUVqEf0O(Q)Zru|t{S!?6#oQwLVgKz9O+uO#>ATr@R zB6n6Ly^1#K3*Ew5J1XXA zi~dD(s^9i;fP==qMmr*rRMr-8gIqqB;49KqyqW&IW3D%%xVOeU4?ABL`RyCM+hdOK zvHOCiRHRRFLl+T zc5tB?V_RqR8wBf1Wft`pNiTG-kGTA#G(N97CQVJWi>#EU^e!BI)U)=va z>(3Ud#~*B|95LX{v>FRC`WAcx5~eBlMZ?@muPJA3HUAv8b;5d}dxFWQ+L7YSWHdBkSQTYF&>b#@b zZ2YjVt@f$nX;HhS_TGE9N=gWo*fENr#HhV#6>ZHnBvw!(G-iX?<3Y{Xdu!E7f|b0P-g=@r$kMUyBT6d8@^?pJgTs=M34R- zIS;!B#VzbA^c2&u0rKmRl}<&LZhmv#T6uSICAO%4DX7CjS#b0y8@dvH8^J9iI_1#p zRsyd7^EmI^yF+DGsMSc2{^5GUHxNw?5y;`CaL@#DrB(Tuq!LC>ZYp~jzzc%mmcMkJ z2}dF#?>wOr=|5WD5k@`pO{KiTh5#2`#-&jvM?->sreu(k{6vGF!lrTWevsr?roY5E zHZ7O?AcO+ZwF{n* z+2_P0vAm8wg-OE;YF{DHXmlQ|?i=}EiVkb;EAWCo)Hr@MB%UNAhuYc9cxNnREu*1Xdd0Cv?^%2SR4De4KZ$2A2c~ z5j(x^8mMJ&>v!&jzN7*oX2l@1vNyKiG~RUYd&t6#=(6fQqt>Y_Y;m}Q4Ywh?U{Y=*~-pf4l+pkRvh6F5m#5U{0Hd8Sg+Vdzob_^rk= zJd>7PvnBx@EqK_TrPks}Quo8V$_Wob{cPZaYh^nt+Z0@x1ZodZycHoc}mSI0}o(ZlZ7vG0q)FKRZCe(dW z@SA6cO*9T=|O#u1g!5^+E5u14f6Jgz^z4|qeavcBF&vnOatC{e`bjJ#o z%7umcmcnSudiON42{3r8o!@(RKT!`HBvsGoExd$ZbjWq)R*FR+CZm77ld_83u+XV= zlI~6aYUPMF^!`T3Xb|8t)70N&hSb3M80PB}sLP{Hr%U((cQ07SM2;b+fFnbFNRhw1 zD#6eM3YhSkR=2k_sK>^`vR-+QkNaiW9$H!ppjq9vG&paJ9e6Sbby`|Hn}I`<#1x+` zB#m&`c;wlgd!6oecMB}{62D#Qg0re+7o_iRSi-D6Jav`Hv@%uaObZ1>gvAyA#zSZPp zC{<4{D=SbJfO!2xi5D-(kGkEFME3 z&C$APxP58e?;Q$+srx|-9-KZCKLT$4Qr==XM(z5GiX0VO(XdU_E*@W0kDiUL0nf|M zt-WlPmXQL#}tPbgmAzHftj^g+l`4}8;2 zXOZuU$q>sTYTKE^HCVxXM36_)kWW;;5J>NDiMuP|*il4%hS%EQkU)F*NVRLmGv*~Y z9|>G~MHFl3JpYvg;Ma7K(PbVd=~4>5A_GF%W4n{ zOPE*WVwsC52pd{^MYEc%=m2C;ML20*DqQ@*QPZkNtqU|2E^-;Dtt`FW@;5#%zdPz! z;O9pENH|C~&af(HFH)W%PM zHW2f30D31lgeTJjX3lyAVB1Z!!eM3@g_Trm?2>!O;^H1S zBB9Uu6dLi4;c5oatytFt7MJLFqDrXyL?R=)W;%FM>-{SUKPg`{>cmUXG$LKAffgLL z$Fp@TTHhmgjQHwVL?x{=pQEd0>RJV{*E?AO2R6U+@>EsEufB4OD_Z{wS~b7eE^ciChVd{_x7yd>PS zAuh$IjvCLpKjO0*dS%nlL+X482hxkbT9 zP^H{KZ1p;Nx@f3M4g)(}jffs!%lC>u7w*3-+%1E%aUT z`++!>2F#K9qS^ynkk3%8tzs2M#dDBf>6wEQcGE0LP9sAI%JDAK-&Q<$%)0v)Qtut) zuq7IZzoa^ZOF#7cAX@nKsVLlvf|;a_$o~E^u3Lnj^C5VeIjqm#%KNFPS@1Q<(E*;tn;nva-BN}}+YdCR6eZhB zD;*0|&d(M}z*d~&vfL5L_Z{m${&f2E=(Uy}AOI;SVmcm;XP93$UHj{#c3Q_8+7Yym z$OZY~Vgti%5gt#c=fXW1O2*R5Qvb;u?go#_&O7_mL_G0}phuc`8%TjAH=`WzP-wD; zp$~@*)yt}_iSH$pw_me9d=32VMPO^A=b9=PBxF}H6~e1VLoKe$I~OKd>Fa)c;rqG7 zR=%pqMWFFK9p4}ihkh$L_G23wo+vXJsMlRHEID*qk+HI>5BZeDr#@_maILwz?Tugo z8n-DZ9aA-RJa34qb-j@Ww?u#ZEI+7n&Y>2}F3)7LCv1vHe4+JN4 zJ(FvZMI)7wWf8GxRN$Q;hs}Asu zh<^xX^G?a}E^B&A{h0#%hWACK7Ny0#Ph5hQg4)stac!Ae1qKXs+Fu9#t6c3d)bto5 zMCGh?L;BqPc%ST&zn*H1dJ;o==FErJGs%l1v4=|zMeUaoJ|)lg0vLW1qd1#Q^(qQN zvx~QkDo3{$a|(=%U7oq<9T$;jNI)L&J0t;<*?!E&292J~<=Zn2`O-?K*nr8kT!Uf@ zmK2O0)f z8H(@+G`a80Qe}6lW~m!bi=KBoU-R$Zr1jH607ULx6I1ZKFdN{Ed zQaiEI$v?Y*20GL@J|sn1RI&@tK5Ag<;o)89yAFF4@YtLJjHXFD=1S&>`ck=JUc6!c zT0b}vx;5sw5-nXdA-=>g`by!pYp6-Yc3J+myGtu+e&{94M$I<2_0;E|DOE$G#RUV? z3~d2`BJ5(wnSXaTjNNkbc!yapG2!!E2~t@vm2OMI^mi-lor3JooRY!^K8rT# zTzSpxPof-ViT+m?sNFgBkWzAG12!{9-qs)KIvZaCsaf>Ft}M@+c?b1=PZozarHVE- zNdtEk>Zu4q>#o;ZCJ*Y1!#SEH$K)l1SmtAUE6a9)7ADVLTC{_rT&@z(&B>$LyK4Z> z3wUAfO!>@SM~sv2Q+y>tnqRh9$_?Q1AW)_T$i5}eXZ6IDi_jr0IyaBM6*O?io zM-rnKr4^5(^G)z%tWknorOtsu?Sn&7*aeHH$AT_N+&ZL6v&?jkq}pVqd<3s`Bf0`% z<7^9EU|Ftn^7E}!kaR+2(>M9SN@1Um#@EF1l-}#0;;w(ee&wToY)DCe>&ZR7d)Gy8 z>wl#DNO$vtuz>~T>+8=cj7*J=`3=1*9Y2__+2uUtt#J~ns*z$@nLRjCU6>DFE%SNi zD%ZSm8Im27565Qt$JJH$Ydqu|;(X^D_IIk7%^bO0{UP#B@$+sU92m}F_Ox*w*SK{1 z*#D^CbUnHF$97-W+<}wUspOM%hv9w4TRKe`sSfBiko^M&9Vg zQxn_mmLzg8yM^85U}p%M<>cD{++-}O0t^Ye%30ab-I?RxahbF?!2eQEjxdYywu)K&5gxT(%1tD~Y5xMpe9B*z)MMQG zgLVlR%3ZAxcNA(~9LTwmlPO-MQ6mGuM%|fHp_xHW(cpv=gmLEJ8f#T6$3p^404SI2 zQSPCjVJo&ned5hmp?i2}K?&x&IsR$B629`Ddn=cU9s#C#eL*KYoEfb{l5iN^ti9u# zpE`Yra?gS={XBTmlJ)mPLuP+=j^dR5gK)?lhCcndx9Z}7n{c63dIv-A&6S@oK7G6m zq$v~lTf|Ca(kh#c-syke#0 zjR1A&3%CWuLoIl2&Mtg<3$4z#x#s^=eA0yV<}6i^5K&Un{Ds>b$1Q3?H4mx*H8`F$1z7LmIT)WXcE{-N@cLPjR_NTXnUUd6XhVIt!ar!!?5BG|(P^O;&KR@plbp19FugH~eHxk(YYIHX{;!Q$94%BUq>9&);XI0<% zy%96ldn6)(MSS8poKn_8*W@3hd+$HO+fl1bSpTrEjtPiUEbOYFR`LfOrorz}wqee( zP1ose=8y5N<*^0#rHAoG-q1Cu%LQvbz$bRi?Diwb6;y#0cb|efq7YSo^*7&9Y>C>{ zHGjqW>060t8Ax9G5`*NqIj5f@0_NCm-Z*#aTfuX@)cBM)B8kPfxFw4^9(H!SkLDrmKfA za%vFkt~fVd*w`N5<~F)(S67GD4H->o{qzQCa*H#tSlWOIy#`LJm=5w2Jmt!3>J5sA zRgv!U>QU#@L}8yYs1)vR|3#%R)vQK1JCep=3qy1Z`8qgqjv8}E(n@4x)HNjIw0Dzj zj$1$z6Wb1u=edS6V|tE2Cn&?s)m89>jFRt0Bd=?XSIxNo+S=iy`c0cFt&qs;q9#lf z6neZ`orubXoAf$vVvL_yR+(dUJ)9OOO)U`Fx_aIR&r~dkRMMLo*M(#}^}g1Rm)jwo z!6vSdZPX_Die|RMhAyS>bH0%7c0MC!X6HO4`pe8%(U+~AFMk9rD>@|}EYb~+`ZmkN z5uCQ{oGkpXmF+&heBtLT-@U=@c*@_jUbBeeRyYcA=Ge z>Jn+*uTBfx_yAD@s&*+?I!N9OzYqfbKt>)1l4$gua;e9rorKFA+01mm#O2z9NVO) zWVSzm7sb*;`H(P-Sjr zA_a$CId1|m%`+Bid?Xuo<-}Z|Xm<`u8nlIdvMEU>L(5E$l5KpZ*WYn23@(%@t#qxJ zrIlMOLbzuT3iT9Uus6pe|ItYIQ+DMD)t{ru(7nmu%6fdB$fdVqvqHaKl$=#v2??%84#$5`FDbAiFTne}VWiNvCBp%#$#Sg`-3=Yd{B z3gBK$Wj?UlUEjt-U^v5y^*JfetZJ`*%tjJMl~}(yAxMKDxKymM8-)Zw@JKe3U(%;=j3P1Cvj^q z7=CXFigMeLV)pDcoeZw5zd4+3*Z=;8E-ty34>pZ>X)O)@{KVe=V7wUdiTS3HTe7A% z>4=9#>+6_+w1A@2icd&s252|ax2ys`zFTX{kn8iMp%Q=)#{`MlG=0%T8_#y?D;&s+ z?bc7i{_v;+Q41tbtw^HtcnLmxElOCu@MX_x;EHLt6m6jclur#HR{bQ1+-46>4|jeTYqfQ zyKH6I0j5fXHH}m%w6{6Y()76rlhk8i7)G^1!F$g@w}7gbLf9KDe0uvDW8ZYnnP^9{(I-OS5KBEKT@3b;MA@Gh+S(3(+4~Fv$ zx-kC2^CdTeTnYdWZBor-tLciRwDt}mhBn;{6j=Gb=YwF5W|!c{|Ctv5Q#t=Tu1g!; z{SwCVqI5@-Z5n$6T56Y~)W~7D0~*VTDlg{aU(Oe()t3Rl#4o0*SMje-PICAiO`}r6 z6{k;pfz7DMmkyD$glB8|l4H{2Z{t#Un>T7Qq*~H6@SYxr1mgE-CdACW%TsEi{RzTat{`fda{c*pmcUAvy=rmV8T)skxR@REP zv^0>#2@*?0^^F(Hf~XHaj`%m4l-{$YeBm}*hIoOgltdy4V55#WmtQ~>v*AlkHh-(D z(jWduPrepEPc`q$3|G)<5ZBAdOwj%<5s_Qk11U$uJ5EN>n%JDvRqP7b>~h z7li71o`~W!!ye=wEU*&f%wP4mq~i?TIy)<*93KSS{>HJIq2<7b#=Yw4*tMl(Mhe&Q z+~p%=wc%!Ygr4OizdJ?0>MlTB! zxRz9KAogRAW&IY?s6^S&8U1~(NO2+&w$%MLp^NmRkEAB4xa*Jt$w=9^Pw|%%f$<9( zJkLn6VN+pBm8we;#R%hS@jYJKTJX1PaV~kQ@q=Q?0vNcNPbsGpB0hS+9z~v;pd#SdW@;Ls-Yuu5Su_!C&3kdaR$Y8$W%T;@lvQ z?^&HQ=`$Q&ED}o^IJ7HLVUN|WuOh_U?M2H5ZI8MHZhz?2M``Ii0${K`H;%LlW}ZK} zvQe-5uOE!!Fv@!+FWLQAuZLzYF}rX8odrHTO`0!g7}RsHpB>F=UG=oj+%Z(U(ru_m zL#p0aw5Mu0tw@83B7tCZDSaHARGMC-bfk9J=eOJTG2}JqF-rO^f|jZe=n;Ah{0Ggo z!7U2iT01+N-JnDs2+&pkNq$?lHrnz?cGJ%&4AGiP@C!=&EligntmnuAM5GHG{h*8O znMLj#t~ofUp8rde3SpEtX6H<*&;4Y+#98H~b7*|tHOfMRtM}TePo8i;5xVz9k8in# z&U{<B(8Ef2hFlalG@IktI+NmPYGF=XV8&-uuiS;qo_Du?nY+4Tkf=9>2K+j#aq7DyZI)tEPggdo?k@pMP)kx zijdA^8VJs`3_~yGhk*OK7IB@OF5UnaQ*|_1A<*z0(SUyfzZnbx+G#wcF>143dqV0T zI0J`$%l_J)Z5tUQ!w~w&TNpf z0IB(*h}qV~sH|ei2f#3#K@^_wPHFgo-HCu#_*r`66@+wm=TrvyIh%nj_PAf&ipZKHA`AagniDN7Wc!m7FKKn z4w>Y5skGrC$2FUx?Bzu6vIrX(T30Uvax#5zbi&#Awz|ts0gko(c2ZH!Cp4a#7uThB zmoX?vXwXnCG|i)&Phf`4MA*?m0N|bMR=KxKmH_v#L#Yy#o(P=>K#QN8xy!9%mFW8Y zKI(;hLMb)h4uk0MqCQh2Q1!>GH_h*T;O3LHP)&hL`)oXUA+sS(SqSH&>XR13+^;m_ z5qxRO2k+pj(ZbbzVhbK%_9?W2*(N<0InPJuE;iKs-jNvXUC5l-4ibqL)3CsUg^K0E z&Ld!($904H5^O(h8WJFwwQ&JLcXcVwKuopNZ7`a3l~W@*$pq;wEv=og?_^#zWG~H6 z(%y!Oy+~$bHhdJl0Y-eLc!Wj8qOvJI%52=erbZp`JdV2ji(= zLQKcwo{x?Qi)&43D zf1k(bA77u{+BAjlo8#R+S;cMDVaWcw3S#*|#z-w(mi`NLVYB{jvS ziobGmjhtE44Pl*g=mgVCR+$f=?Q_Yog!~v_6Ukzw7oqTjs5a zMhG60c_NiE&r$KDP7g!h0X28Y;rjh})G_ZsY@o9HD_#Pr{fDHURAjb_wVb=vI@^LG z8n6}NMR_P=6>%xW{7fimI!7TaDK^UKN92EU=?_ z@sYA@gtx$eX8XF<&*7901St!#hml(4g)i+^u=KTvpTn8=qzwhzW!_Gzdd_VYp@3O;tH^Nxo|K!e zmY^)X;r?Vh)arX*dSG)rG8f^Kw}B!c(&;H4{r6B&fL>8X%51PxRS7!YyYA`IOMxi9 zH=}ec482x*O*~I{1hk5U*&N^*JZANBvEA#o)my?1oNTrhpe#Avq1n-XF4p79mA3vOhr2d;5_i!_!Y7*J8D3(Kq3@v|mayQCP@x+>g^~$g0^5;lmb>Hsl zzZAEUE(=tEt{KK=Af_l;S$bia;6gZy)%A31f!QnE*MR1o>#_~dMUv%sJZEP$MoU{q z(|m|w!s9mOx61F3zU;*0`+?#Xtj5k|M}*^$^I1xv!aUD7uZBAgTVDE0Aj9S+yVnY) za$}!+kqcS`C!wfKffMm^k?#G8obl#&NFKeD)N#4Q{^%gv<6nS)5zJ|ZM#L^E9kI$9 z%w`ceAHIb<+aHv;i?>X2)UEM}1gVvDk7bBOaIMOAydKHfJxcw)l5v&Jo&5Igw`AM< zPvR(L5y-9;#ik$HO(_olw1vKiiLLz7(|#fmPeWVb!yd=u!^4%g6Oz_N4S2I^#ixM9 zncLsc9g~0fawE5A3~*rYwIXo~R`ig}Wtr?!TB!_$!*jR8cI2TCq;&+MOCL>K%^VLZVT>Z|*UuiFJthJQv| zaZ5=nBY;-a*ot(XFV7=4@K=B1Pr{I?R0-Q4AKUbdmrab4)_8B(*QWPMn!25pyL^^y zcuUXcbclQ)NCPWg5_gjgg)=e``APQ06QLQ{aA##BX}GCxAtr2Uf^{$;=mMi)(*+0bMw^dA z3bN1l9bDx4vxjoo%hOa(^_KG&U6@@Z$IAsJPh1vFafvdS_@tr%}hyMo_kp#mc84z1n{G1 z;o&yFN{!2Z2Wze7@ODX04xjr^X%|s1ScA&em!n(F$GU@poYZX1%KS|FV6)crJ68!} znx|#K<*>}4AhOT8{3<5!aBvM%0WDn>J(72DCw!IeGvd?hbpIHs{a!2GQEGvy>voc6 z&p>XBH9wt7dNohZO-$(RGtp03g`=2nk!CPC`Uc8MM={nd^`Sr>ljWVK_j%Zn zckU`Epd)U|UD-zY{poKR6gi-ADBY5?PK?%F^OO28FLxn1q+r2mb+FLdw{g6UCDfCAFDDH&UwjHfqxHWFCh-m2)+oV-*nsMlzKF9aJp(I5H>< ztVEMv8fp72v}c-G7neU?8Iq#49+pn`N|QGRRV!+$M$IG7Ju;7?0=>-cOx0|T6OVa` zy~S*jTI+-B2=_T!32=eOEemYH&Z<;ss(Vij7?v*SOfIA1~`;FD6R{EAk-z?m;EldZ#``$;)l z4b8}gDs0ivUU;&q>XkNW|HTe(dw5w3_f*NE)gaDd&@jBED7C&mWl^bS5DaX?lHGT& zAWB30is!^Ng2yCY{&yPCEg`_ekgOKC{2}>7a4F{LCz?&}B3?v%?s4dJ5|6*ZHGfQvlNGwHON4YkXalQ&WcgG{ldq;NwAJf%CX@9`Iw3 zIi-Ap4!7YgAr0tgmqnsAt&WgU8vm6(k?rtBrnwqRTWsK=>qN-5bEX`bBi?>9vSb@+ zaw+C#=;UUC4lwN)$es;$ zlq~Ocx5%FuYI>usaF{Pe7-@^2raGeMTYasir%&k;C%^D zLH@r3*omc_3E)itrn~RePhOj@`!*5|^0afVX?yc0%Ep`4@QJFrZ+O$UQDgffKR5M! z{f|Qu!Gs`Nna=Sae)YR|le|D^nAViXF&SUsY;>qTcsF0Ee)^(&4isAy9g%{^&rbK+lp(+47$+m$`nLIN|`9 z(SH1g04slk?3sp*i=UF?gC>ubXYCr7j}(Jktn5$N?q6jACNU&X^BYmnTG4+caV0Q$ z>b*YZ81^gYw1X_`CX0c)7z5P<>T|c2FuMFZBX67!EZ0{{_a`U zy{&!FA`P<#(tl=WAxbbT=eC|uXnOGoOW;oJ0b3lehxbU%S~>U9hjOTOvvK9;)!&U7 zivLoKim`$;KPh~q)de2Ra@jlY8Ha)BMZbW)i@()a%)SJKL} zo@t|8)ex2Um;4#_#=_dp$s&CVbbt!%L!exC-=}Ks7`NS=J?84>nYk&4Yug{aNysPYz6+``BNX6Ehmec*C zL(h&x1U=*%?O&Gvp($5k>{7`qPYqR57n@>W`Vy$@ruHCoBYx7~&!ga%6bm4Le07%mge+mcWo!%fFrd_AUk?-g zat=N#9!!>V{ZsVvQs0bOPQ=7J<%+&r%N}quh19_K+%x8q5j^+47Ti&T)MqCV_UsQL9|ta%6P12-JmENUSn0Wc~f%$9rY3 zalg7gbgQLrFu16@UAdu!Fkm3-zKl02B;4+_4re#QR9?smB;V`aRc1P4mX_ zrPz;i*GyNNrEq_Z%a@P2ZrJ7)&@1?!kKleV%tvs@9?ieo_VQ;Pyi{x^_`d7BX~LcR zMW$uCg6&v`RDqdC1COfCLvt*CtNfPBHyF&J^U6{htwzpTPOBB!ek)p5&l+r{e8rK| z&NP0qD44*W;NXgxl=wUA<{Gl8S%pQh;zPNd@Px2wnNZE?5|soMrvje5rt$8sV@6^U zvzN7A^_S7rTp+BLviZbbQZG-h?T5skBlJIX9EiHHR!DvopqWHlrU$9Ac+CMd1~xd; znHV1u*&$`-8P8ba?WUsXatj-~sm@<*6`1BTd-mP^Z%a^2zDP#S9<#)OaVqxwSdy%aNo^078Bq)u zzGs_Ny69|b16TJcA)AQ1_CYh!ZjZblMBB@gPy)mnNJnWeDd;s#_n?m|qr`!M~i|HY8eYK`~@tW=tg}YEB)f4&; zfOPsuL8*J~-e@n%IHjEzTfuvss_*bI_hT(WB92-w7j&j@OE|t;;CrtO3uo zB*gb?BV;@$CgI_olfDc?r);+U%ZDFoXR>{uh1-4_O=Gdf#m$#OoQ41-vp)JK=|Z%^ z0XWf56kLV`Zs^yq8!dSFi>SYM=}nm3$1rMOw5t3GYE0>g%Bc7Vc;imS}jQQsWfq{BZ`cBJq9rDkBSMx+Oi{h+RxXqI1 zNLY*hL86$W-FAf4_x0|aBo6&Bi&~4#q#HM9wd@|&jIrw{nj#(nCg&G+Ox<{vr>-U4 zk&nzLvPP95a+)@n zwtSo)TLf97_X>Z@@un<}nDBmayi2&)FLkHe*TR^X<1!j?#A6hE4G-;EFy@0DL8DE6 zOWfV)_lf!;YHT1e!8Q{#7$7s&RB`*TUij0MMLEJ?tR}acx{caL)kDnrS08@E8fhZ)Tf1okPs0R*$(l)19fFEW31Glrkb)?RWU>U&gaO2mrQ#)`YrBw*C0{ zay9B2+;b+Rw5!{3zz^Hy59mstJZEnN)MCbgC4{H-Sts7XZ{HJ5( z9B=lGri2vsZ-Vl?-+Jp+k>ip2GlkXpkqPV!i8I^wUh`G_I4uD|y$lJOIUH0DBM;qo zA;#Rt23Uh&b(3=@Ld~a=z3#&U;A&?^)@s~kr_N8p2ey9b#+xABEbg@g}-XP zH2zY8T1WA34uF-`o{FlhlPzjZyC2;Pb4u+S-UAQ8|BLtJBBRD5Jzs z?f1tEU)h%?XjZ&<7H6?CyCS|L|Ic3!2s?*~OLuL(2Vluv%s$gIldvM}nYy#456%Rl zo+MB;?Acs)ADc{4SFj!1JLqx?&VWu2v5KF))(iJ3lVnJ6)e$M$s zC3WUxrhM0Gsw(zXAt9=Pt*mB_s}(BktZ6~J{aE}p0MWVs-v!@v)Z8G?C8UH}c?xAS z^zu(ij+UFt1-#^L(>#&wVXD9T{DS5^W5Es#*34tgyH|J^O?}58t1^5n|I0-ci0IM> zsh3|$ez3N*93+Cpe)N0YHX+K&TG1!exrhb^#%vsl4^Lok4 zTYa{3$aUOj|GqftJJ$?e9VQm69AJKH2f2045F-~8^kND5nki)C?2%Y_)%->SWA+D# zwQF$}ykEc|(T6R96K3BPlhc7olpmDaU1`49apw4Rtax@t6q!g_6~LaUHep+JNS@{s zTQu ziG9%XVZU8Q!bMT(@%tNhx`E+=e4_Mv7DOrS(dqZ)tBJnT;O@tSB~Jl|s5&2$DE3-eIonW+f&F2PjAVh%1?F-fTHYB6t7F7O|Kpo}a)L z8SS3-9$sXZuZBqK`OYtkc@uYnCV(P6`7_ndf&QWit!|q}Pk&cqeDXkd=SgWN8M2O# zgaHx!de9-zdumx-py}DG)!SQ7uU&iibO?1z4+r+_75a`BjNCjEP;z7teLry85Xh&B z=3@nNXc9~g+LO{w-z<*BIHG~RCw^DnhVb~jOI`8x^55#;8?$-qf3K3z<)1*dsdd1$ zsj0d}d})_Q#w_+)+u1zKf86mbMIvmbNcj9j$%Y}BvJ6Ea-=Wg}GMwox^%8XLi= z8(tMQ^X`LU zIn2z4kMM+hbD!~-!^PkSco_5!$`v2Wi@opu8r!HYn= z1!Gah*f(5mnjt94Ja94!2&D18q~pDA9t2m^LJRN?f7IlCRKs_V0_CkqIqIUsZTLw2 z^=R2P)LKuY9jZSblUCR-$BJ=lld0iPD^?AV^<$H4tU*)#^xJKvnfr3;LqvF{7Rjb>o7hKp^uIMRT<8Zw$Qh!wx|xjld+pXT;0& zut)!+z@tbbS&HjGKm9e)*F8gDXkvP}ngJmgc1Qk#-W1OE5qOqSp1gfFcXTs9YNaO{ zj#p=Ick=swH1!aSGMo!bNo5n0ho^B=J0LI`jyiKHY+Owku$u- z0{m{Hya4CMMU;Y0X&UjG5pYc*VkU2TOUssAuV-ar`tO$Lot2cXsMn8&Q4WL~yv%X) zq`wF;5B1tg)4%$b79aN7WoKmFQJ(zZg`00`FI6%tXq4$54XxDQdztEyrTyV>H|Z~< z=$s@@FI%SZQu5|S0#f+E3lGXK;=Q6#oWC%eGmYa!dVZ51F_25Gs))WYpUIK$7=oQX z6o4sb(*y}jEgjjP7t&i(=?02ozJ1>z7=Lo(hO(USf9n*OuGC2(45w5&`%a@gb0w(1 z>UsCFR?jj4S^y}bpCR7Zu*j?MC(#mm-5gNTq&%sfdGj{4WKi={d#5|^_YzFY4C1HK zt6=$ZuGyt@U>~n9>6^ZB*)PQ|L^&)W1)iJCs#UA*ega0FOOf4mL_84fCJ8Q2PeMZf z&SKpT56Pr8DVy%;WjaKTeaeuYKXjz~C|)3n>Ji`q%CkWr{S6&pY9)(}-+l}b)&5&7 z00#q~^fkdD$gZy_S+S1=BRObg`_w6Ff9yo@(4kIg$^BBfE-upgs##8DLE(^axcLY5 zr)ce6)smWW%ldW#cQro_Tp7PPC@yW758+G?p9pk)AkxL`PoLn~IFVfDF8HIf9sYB5 z_{#Z;Zwj7h%^%Qzq>(x^#&)oyF(Ej*@8m;_SU0xaHhu2G`j4YskKn6ais%VIW9;u;*e0i;J}6Zp8Y#HNwa*S zeIG<;e|^VFf>Lh7_61|o3Qd|e0_8Zn17H)V+uZ&(s-CUm!XPOoCmvh9A;K!`qTn1+ zlI~VC+^>M_nn{7#Eq#Q;O|*YkPZH)o9p~2^c)LVMwT3&x@2!kQJ^Ww>K01N!)SM=FpFn;gp8a()^YVnes;ZNha?WudjB z$LMv~N`7h%YR5a9J=jM#UE+PTFXL2n>yAcKM4k>PU|OtoFtQoPToq4c;D^=YO={`1 z?&E-B<2BVA9_TYQnr}q9ncLi=uKN_ZJsV$%&FACe&)B%U^}hYqoFuUKFBigx&-+K5 zNQa2OJc8y_zbi61l>Hi~>0S6MSAeOpsIHR9Ju^9-_;6OL?!|z?nw)&a1C6Pq6YW*K zg|!%gyOT+tu`+7{Y2(P@5>Sw~j~$7l1q@bUn-QUt@(@#OlAD){yaD-QlAs&1he5iX zhR3ImhOBzG+N|Ztj8q7UTZjiX*Y4~isukEN^c5E@4X086d^!p5^8_|(nFOdO;8c=2tg#F*8?F5p-Pu7ASHy} zJA%|uq<8Snxxf3a^|IC^ftCEhWM;n4?EUP|QEc!5|5S5 z@7m&@sr!O;I18sM63q+MWX5)_rHeO)5f$}mJR|rbN;Z>$93NdatD{FvF@D??$w~}* zEh_zVVAy$DwQz1x%4`NM@!l`B#3H8{r zoHDA6DG6m!%0cUg#~w)=o1J^ycU=ldfi^GP0yhPlWOzZs%jysnY~B(AX8Oll^}KG1 zPHS&I$wgq-_t5Ha%m!_UiuOD zWdZLe#p(Bskw#7pe9oWj5|hoe{AR;4c4@ol6$(glE-&+pl7+41ew)56y2~%!-(JVyG`-D=TOPmz6LStF z-Rrlrb6!5+0imgNIRXHg_M3FsV%+Lx83nGK_vWPASNe<=cb|#Rt0dJUHudWa6#+Kt zW}WxWKf-~n&=h1B(0)lxQZuE99F zbAo~VI3tVbBib6^-s-gIpN*oRjd~9zOwwNjgG?m{!lJiV-I_qw)PCiF_?y~8uOORa zlCfu9oi{M(UofRJdL-E&?1H|dLYWA7>NK6idgMPJ&`Bp~i6(sUFwh1LU~;I+Khy>E zU`H}h3XOLfUv8SdyZ!H{hxecX0)IWDvs33Khblgd-enT2;9%4Ajm@Q9WI_?7Lm5$W z+*10fZfwrIa-3Zrb`v=ux3M5XrO_e~>8WBN>Ta#hBm~YJw;$&W&1@21dR+^tE&RH) z+x)iy8lKr_mYHc9Bs%(Gfh#{gxBV2;c4d`9oPeKDz1ZAUs&X01R(sxLD%5e{R9(Lt z;to!WP}MSRf7OHa8O8RNBMEo|D;|3>E(KAG;-OGxsg5lhl{>Gw_trVWV)DAX6I$#j z_@o1c!8d(7EpF4417qbB_4KA&GDwc@F9zVXSI;zRBap%TFe=rk2o6pD6Mn12%}+ zhM-%w=Q|F2cIH>K9_w|<+&;)G`%%yNc`9+tP31VnZ@}QmiPE7LAyqI7M(tMJWYs^n*)|4q3u0+$|S{g+!-R%Cd4V|!0#J-iTF4!S5R>kTzsNmm>Sqm7~e z@frH-n@W}vqg)eJbCT}?`ykxOKgIbC2c)6|=O$NN`bcmLt-vNQGL+*uB%zXJq1=q4 z-cHdt&Iw!+aSMw4wY6Drwl!528NgL}Fi+Fuw0V{cmh!&kM+yAE z77i`=TAQVCnsTI+e9*3yJG9xRGgO$EE4wtZlie3~$NB*%jLYbwDp$kXuCZdToR}5H z1W?mmzvK3{_^5BgtHUvGeM+yIs>*A5-hNye-1F5sy3Q$RFIehIi61xi+TV^^j@v6o z>Lj&4=^5|t@wt=%rU`kyN_mExG!LYzc0tPVx0IUdej8_;523NnyABvXn~W0?12xiQe>K&1=pN2|{KfN{tU}-1529Y%3%@Gk z&-o_~?vERXZ-cA;kc9}_cBkVD+lH*2IDG!aofX9fx6m#ED1Bi|>#ks zRtNX$S5oD^ey`foc-vLpiZZc4>F9sl}`an-pJ(I zd(N~XS|oB$e95$!ea1?wyGghbBktzw6Q0%cGg`jnn^~^8aP2bSz5}iam1-mll z65rZL2#!b_l_)`axyC$nV_XyUd_Og?8GW1jHkh%q@C`_90gt*tf(p_HwBC!OLfl>= z*sMbt-F>%9mohPwIOZh}^%q5{+w5^G4rsYF>tFVp@a=M?APYEdF*Wm*69(pxoVNK4 z_wamtfZIEi3XGM9>xPV=O~EQp$^@X`tPpTM`)Zi^2b5q{h0SW?+JFT z#v6OE`a$1gL}8=P4A=iPzaDKaS@IVod$V2y?2AzlykcAVb`o^{b=_jZU!1YBaw>a7 zx4}5l?6tD;*QLUC^!Cq<^_1ZcFHm!80t1_mRh^&6)iz^{Q~R;LVqT2(90>*CfvPGd zBR->Tq?cLx^{IU3rK{DUk?2W>4mj?|yT1=zhW>XZlfUl%yuZ>;+#m&b9G(tG6opr2 zE$Ut%=V;hp=Rlf;O~QL3-!i7pUD8gcFs=_DGfzb_ED%pqcXlwc$R95sIf1Vg!I`LIz7q zj6%gdzDBxB&ZkbZKRPsqaQB4>X1-+W`3{LOJ+K}{{@1Coe$pI7p#)S2Ox(Qg77qvs7C~ymfvhxsU@mLy+Q?GSE63$mBjxzA;8;%hNZzffM$S!Tz_^ zof!d7qbr+ztkezXRJ_sx`an<4v%w@P|C(u^^}ar;6KbR{yu;yd16=It;@pvFjmjA1 z3cuI08lSBqJm_)Q=Q#F{Yr#QB)>~fcAiWy9*hwM-(K)R=LDu>QOopOH2S{cW08)tl zc}K?a?`=mo=mwjhWAyHGIFmQu7Y@M%$UrWruUmw_N2ie)B}Lf!{KZT^b$OneSR|#h z<^Kf&p!xWT1NfM*JCjBC_g|(w7tIDH*-OQw4Dk`U=|E|&Sp-X&Dx5}Q6%JHkm7OTC z%Sjp?FlklA;TX$9IXYbj92IB-pXSaYVP%Dh*Fqp4dTr36e^DAR;^)=N%nFw6NM_ z*AEllxxZUtn4-$%926sS^yHjx6hHQ1wX1hG;4$=c?}XxvzZ^F)9AFhoomve+)R~z{ zH^|&JGb{)zTTf`JME@`=lKEKYQ)@b{cNXbene=&$q@-FHV9kKqJi7J@);C$Ci&{+9 zL~4a)fLQR3$&{Ux?OV=ZpY~Y2#nZ_vGwz@SQ+e)bxN^Mx|3S$ z-ra7*hJ9HO_`}^)KYh%o`o(z3?BUN@H&Z#Rr|EhhbouBfca#=5*v8O2fA*m5$)WO_ z`YaW0(u;z;nHj$ahIhlCaPE&|AqpG1qKg(2kVwlggX$*z_&z@$Qru}JcPRVeN_mn@ zmrJ6P#&TY;6Jk;{r(Rk;?EP)PWA(ocm>Bvm4mR`4B-u?eN}x+fy_PE#B$#a}@X!S( zD(ioE-e;FCTkwS~S`xEJjudTy1_qCL??%4fJ<|Xn?h)h;rbFJ zC^XbOBeamIiLqP)&C&*|ntekT?O|SBhd!HTdbIbsQ>^7MnqV<)vEH*7y;0FCHD1%M0uG~4dHxl zx6o>i{5(&$bF+1cS?{oz6&)@S3BJadr`BE&b7PiC9#K=5_Q!ma2Kf)Hplz$1Lf3qI z@y1VDO#Cw6(0zE>&IZxfdEKg`RRqH7dq$Prmf>fk^>TM4Xzt1jT94+}y-s}6W+(&% zaZp=4jMFzee5Ag34F<~qJQX2_w;;PB@Cy`TX*9)T=)>X&zFK|z*VIHr$(`8&dRuZ` z)!xa8L@7im#?l6RQ4?jcI*hC8+_}3avEv42)nD9RcXzC}Zw--i(pgoLJ(m)!x{#YX z6K>nCNOW79^1OI(4Ptvvljo@`I2mOzW?e1C7GCR}CRe6C;eLB0GgsX8asnmh28q=v z#toIm|Fn01V8*&I>(Oackf9sCp8s z=c8~YBrsM&%jHR`b@k(Mw?DgNGy9=p(8UCJyzPd@*EER_`zd?D4A*zUR zJmbbeJI;Zx$-tX1x>e(nA{w}odKdx&IJ_CjQsq;a;Oq*WY<}q?Z{u_kk+UpIg&LG? zj=!aCM80)$xZv649B?}X1Yh$r zg(-%E+cG{R_j}J5Tyv^|g6dU5Ia$fVb9er`Vsf;G$gWx9`1pisYv||RgwZ6+n|hJ> z9-wnVyz`Xyiol&2XtxxZOXDy=+f<3DM>S_~!3Ee;9%ReMe;=#=c3k0kA~jJY)JP+Y6C5ApJoaJ8&vCrQfs$_7uO6VaTyQK+00B?UJ(s!dSa$`}Yi~Bt4@NunzaZ&OLGKDE~b$NL{Oy5PkdvC_~xD#MN2059V zfv6w$gz4~I8gpWXPr#VjdMmb`JR{eVe-N)F5IhttF(=@B$Cw#% zPr8U7;!$}Oj@%VcvG*eUp5UZNM>JJZVOtz47^Bk&l7f1k^YBGw^=;3Jc?uddxgm`r zxhs>y(2?J4+TU-DV+sn*l@5RH@Q*~*ya}HGsNWf!LiI%R_uqsEqPFpyjXGetBT=Cl zC=e%SYYO1eqs4`mI5I9inc$&~Yc7)`OJc99SKvT~>GI5sKx$o`AIe*xdqA{qe@?#w zW)0S_1m#~rEvx-zTY;#V^)6cBgXC5DDX2E0vMQat{ZSPIKwTztCA+0=+%Ec}Mn_`> znc?y7S-@AlLdwMiwj02bn(wI3kmx`VDrOrbuq69`d)+AZdk!ks$9mBmccKUdVL}HQ zNQ{eppM%~)E;ESYQnJV9Df>mDb<0C|D-|{ZE^y!+q-m})Ai8m(% zW@?6YHvMIRyQy<5ki|yjwH2`uW%g{otl^AP=^{TKH=aWU9>A87tlA zHko+;%^$K`Qy-@sM&UpQEw-m;@=*Rrnd`|=X%_o@C#3Vt23I+&LSIUN}vn=Kyj73MG5vE_*PY{s7^O)w;w5F|f!4@6Kt%mktoZnApc zcFaxYXr)aGO?i4khMr3QBE4@0qww}E9=+G=D<&|C4(0d)F)XEl^iww%ckl;v3Kw^% zM(g-X61D<5Of!HdN7QV)OpR1Qm;B2VkW)eojBSQF->$$8KRfWE)wpiqM%jy2RqT&m z29?JMhIPx~;(`f8V*P6o8HUiT{Mo@1C9zp2_VB%RPm&YYlfg1YKBADrgS&8?wUBsI zMfls&jSJ}+`6vGQF0qrtxyp+fk}h2XEb>2~9+iwv(T;`!B7cn#_@3lV0k$^4G?sir zmh49UAnPp<@>A;zS-1xc6Y#Q|Nai|h#KuOGu7N_*2@<#o%T-{p*hiKw*j+E^*pFme zKlQsb0|pkZi1qYZjw+DbH!2 z1GT|BlclM`{&I!>@@_$LB(Zn{s*%EMwh~qBibfy2*em2~;aHAu6?S6`LO{CXqg@Xj z*1jKws0=5`{q~l1NC`~wcat}Up#;NmFxIX2xy%eGxtN$%&lmc6X zNivZ%@vg-3u@c3g6#9V8mU

NBe);)_?3+8?QW5+8 znw0&Hpw1`8MGK4BxGqcYH+I*Rg8_|U-um&XT(*1R<-?|urbp6t(eVzx1<&5l@xn69 zZ!@UX#yME1I2p&q|ZO$SuHp5bM(jI}@DlX%G z0|Q;`L%Snpkz@9z%5t_yyN{2!(}YkFdSH1$S{;{yX6np3)vGCG3W%?mj(H5T;TCj0M)QX_BR#kpE;MOm~ zyAD>@z1^y1&G^IMNvPvtuXvSEdV8gouA3UvxOPEU+(;PU*IdtBbgWO(zF{<`)UF3r6bXJRA z3%))rEf_~?H?da>60Lc=_t& zEeO6UeJTUX+lC5```Gk{A1REhO;#rgSUqtNul-4;5 zb;<_cXfZHNlaEsmzc12Smc4d|aEY{^bk|4!Hfc(-4v(XdU+CL{05l60YAWG*>1k7? zzOwlB)CU+Pk*mU;airCv$(r{v!W&Wx0M~uc14e2CW}kz@OhR0itb@`=2FUK_AHSpM zgun%f5_w%O_*9Ab!I8r?GpZ({|AvH$qf6yioV(%7e3#4>d$?Hm^w3i7~6Q6 zv149+NXv7zJFX$|_uTxqWQl>6`Pq&a>MAq43CC6(FrN2S_$hcPi^7pZ%!#Mh!at{4 z896zeLTi)QDD=jgTi9lEgH22PeY4`)W8I^dR|8_DZb#xSianvValP#QcoD!BjuXgk zIC*TkStm9HfC*W|cDgN>7Ha^@SN2x%$^yTDXm;k~!*ol%{QT!BcrBi)4m08|$Tp%5 zKylYDyNpKfKOL+O4PZMGPUBJkSq3nCSZ_1rK!77F%(>X>+l`M3wyh3hsAA#8JVPT+lKoce*^{DP_kTUO z2v#yF^a$E!U&RQQ&X*43#LXHxlcx)Pl#Z+)Bo(QaW(p)aA>x#bCkH6Go`^y{Tn4Lb zQH+i@qHFmu=_x5(oyQjoAadoV@6($)pF502pDEs^Pv^U{y!_5d6rd*XJOqjW5BYC5 zqL9*!h_Z)PJgQNE4ZG~&{pc|5(08;u&qJeoYTTas&X$HzcoaGBMexomvF7hh0*){} z-_R_nxDGxsK&=I*V*4jMUZQ|LT^hD8B+KAb)2`6n#C5W0a6FCFYB}~KeZ;xLV71x+ zS45tQOES_A7yLuUGq@hu1EyAE7|_Y{aPOHXNvLkh^>rM)8iv*mdyoz^y+_xFlDkrd zOA{CG0~XT@%QRSg@6ml&t%0Jxl2mxKMay%u%nBDoJZ(oT8%iLx>+d?J(KVwts2oV$ z(}<}f;EN%v3}x!}E+>=z&chc`(xTFhqW#iGm4p)Q^>fb4-Rbo^lo;{5pWV5~=>6IO zWg)?DpYpsomMM9w9gu5{GK|?q6RgF)p{ck*f(+w)mxq}Y5vZ8;R&1hZ8nV!O#*!vG zE@{zW{gXwv>t$|_ul_ieG{mpBS{|(1rx?pUr#a8sW5Ov=I&rooEsL; zqPhy3lF2b{SCj;w!!S0kXu!{`A(DqxG8LW{}2JV!Mo`xdU*en8xQ}J95?O$f=U}{ zj@B5Zi>D=LroETpT9i2yg`Egbs@RSNHj7JA5Q|KG~l1HT;)_|cqjJn!1Cv(&q}(`Jx484M^iK)Ex; z7yIvQ!vmsQTDX6J#%4Fhtqt2x z(wXzn!U~ML8CR6AkwhfJu!lYYa;#PUsF^7;+^J9eCf@EfNG&y^*on(GJ z_L3kmQIjI^hb-?{8Z_MBa)CKgkgp`beAUXHKN+r@trH$<@_QMot~};yvU25Ehdu(8 zG8~<EOr-sjT}$ z7Fq+K+*e;%oOhwm?*BBrm8&A}ixW<+i0f0W!=0jmg4wfVm3yYofhUw#Wz%@?ygG zXQseKrMF%RNR&cVLc~vJ5h}-m1D*kvu|5Vefc9nNtz_AkDU_dp`tz|fz5Us@bxnVb z7;1G|*XIy1!h^k-9tjcx^ci&ldIIkCtV+_TX$TB8{o+*hM^BHO9m!(!xbR86w_Czl z;2G^zU)8Hd9%Xl-G8Z#tBIi~MSZ(2_eWGP1id4Ty7q{P2Jm$GZAsowqJvUHMTko~P zWx9}qnJ^&zF7&05yU?XkH9PY}gB9UL*&nhy-uF^=ninJy@)P$y90&Z|gzOC)oyz!F zfL6s{Gra_AVC|@tf+nejWxpL))0!fJWhS0jf$aK2=11R_v(y>LU%TNyS?2BYY^j8@ zb#3=sRWrY+#B;CcqJm+gNC^QVXN+N57+ccjbkWo-(CR&VVLRx~qwgO2q$vJ%ToY8Q z0rC3$P`-Hs?s^brH9cx|!4cSSwc&5eW43g@v!Od6Lw>~{=N#PEzW(*dnu=VG$Ux*HF*RBmxC<1Gr3hSv?&HE8063Rx~t+cP^j>< z>h$yDfI=_?<=Mg?nXiFKNq+u#Lrfa#W$FbLrAz!-vGvRmF+>F%Ri)JF1bg(gT8o7Jn2x|{F_yr8+bkK*MO+u3mvpSAwX++Zibt1giuU6+#=>UP zR@X=ZjHAb~pb$d}fOWH&t`rO8!Ox4gfQx?bD;wbP;H9FB6u7qw5codeuk{%O(SyX48VEWMIV>1@Yj;ew5p@NN32n^ssfEvp=yeRsHDz5`JY$1nmhI%wo%+uoS|%vlbK1e0{o)- z0uO=@AW}(va``*$&Cb!^p$PZsa`6c}nIo=rTyg zb_o2DQV*HGOi7KWK)OPiW2FGGkwN~0*VAhBM?x;k-~d?b{vK!T$AG^whie4pVr0{HSM0DVTpLjwiT!bGO9 zzGt67ifp__CH)CNOBgrWCyo`$DtGr_08xDklIMS!{^S>^xR;4L4qx`1X5!6*-4ajS z<0CoJkC}qFkk|+D227R}@wstZy$8StvJs|}LuHmo{iYffr?>#YBh_F~K7Ld`Sj7PK z8pHIPXYHs-b@k`bp|BIjtAx&~!c4O}No>oPC*thjd0@&RNX_^!MK-&kl^$urmu&5q z%4~yQA9?8j=f+L?n@?$O&?3IvUITXOyZ`ryLxM{_b+f?qzJy~LiJIrgzPJmwQhI%v zF;@kA$6yibD8@)mP#0}jsO+R@x3MucUBo4V%UvQpZorP4w_nc&fiFIMtMIMNGR4>` zm<_w>=T|nW3hq)HH4j|$?rpXlsM-6N--Y^2P*}SHY@5=n3(J;W5~lX}EGNFYMEtk* zn3w-1NgCs$j>bK1#^WyI*K-aRI?6ZC-=+|*S2-m%&l~im{*Zl|RqLXXQ*qK^IJx%j za`C#REL-jaHEFk=tImcK3uXh8QKJ$#x2L5(*FZe(~=WFNSN~@707}p5&F5$3<3Y;nxziK5&%Qg`D}3&=czJfa=1D z#ERyE)n6mNyG53$J|)!0Hoeyi_10F>7_+KU9f+HK+m!={94?f&1L30hXYuN%k0NK-9v}YK9!?e<|6xankXo1b zG%$Joxnpmg9Wab>%JU?pNrI?r64wD(7$SrgJH}%_d3eA1gL@UZu1%yJTCQoXuPU*? z98caLP&<Gn2IgXv(DKMP{l`AkTx0nRf$b@Y58gZNf z`lh(^LtCZmzg1S!Ja4K_1YbK?7%CQiU- zXH&bhzfH@zLkWk*@ye=2J`G17(&8cP?MkMyMHm)HMAVqtM6S5gAt_F1#LY`wx#V@2 z2AcVXKMJbL!52U8{EvQ+uW4(G;#kh(N-E>5!}XAgVaLh`&bE|Ry61m6VFcxmf(d0qQB}2PmD$-!I??mC3i5MY$l!|% z?@nwINPoDp#P=X`B1belLBy}X+8jp^VE1>Z#~27p+VV}Ty1zi%EagQt35|t6tnlvA zLqGF#Q!t9YF#30N^;#m2Q=qCEje~#A53BCYQqmoOzM;)1s_w=^5b)vn#hA?lsd-p` zF=}n5Q;8jtyjp736l;=f118h>Z-pnL=lgk6^DCvsJ@6!bvNolcRn7Or8b=LGxd>J% zBr58c^XsYy^qAil`n_2k+FL7pjn@=?_3Y>32|oC-H$Zau$6tl#66e|Jfy~nZd6gr- zSVyjehfhe6^Civ} zB`pJ2{36>QReLr zOu}<*>-3YEj>&nG{Jdhs)PG~Od@)NGbj(_~pQ`s7mfZW{3eOGm4{R2Tc^f4}iQknU z`23K&KJm#26Grt}mK|Pkhy4v3FY&(+gj?!Aeo>9S50bnRSjiSg<5O*>w%O$L4Z1W( z&}dhaB25Pbw$F8TtlF{rNW^Dky5rW|6EPL zP`By$ld4Da*5Q%042tv8ve_m*u`%PGi)xhKFF&u$3ZN^gWM5-{!Ny=AshMUxb>Ces zLFJQvVr;8NY5@TW=GIFG_?+4R3jGG}N)pV-{-)y>1{ph~@6$(^_b=!>{_%Jxn{J@( zp&F&H4ObM&YUc)3RnR7;yqMQ^iu%V{F@Ud;(YtO~T|}T}RjA6w&;iKn-HVeNZIkRJ z7^OP*pB6(vl+9SE8b2GL9i>dd^n=q#fs%T^W4@Zm;ZC8IGB)ncA+OlS^n?B7o#R59 z?i$>psRU|WURhK_(3w_@I9Am|N4EZfZ=1{Kf^+et*BAqVkolsZ*|VLfw^2PSA0{oV z@q$ZGHHGoR37X9Pj(MDnG9LY(~kx7@)Qjkvm@YE4) z5`lKvRoCMk^o&lMLG)62Ggi^dF$@}arljeR_F4ZY z0nji0N=u{mIc5$Slf?gidtzSS$A<;HJ7ZfdvK0+>iT5r4qHe@;P$~ z)#|b1px12h8erTOpt!%kyA5=o9sj>vMY8N=AX5A3O#qYuB)Z?e;eX5*;KVRlV?znj z9AbMZ%E~=0!MtzZTBmN2Fb@nCLG%wFIIG6nUh(06)C!^(ICaDihZJ?Z8+hqegi_8iqC(`%(2xvAoEi$Axcgk?v=q7Mzfczt#ED+$H6Y4%=``PZquqUMsdL9?C2UO<4rIGh+d4Mrs zh6H3;fd(^y?a@14HFYu?#Jd+nfOIp7*Q@{we|ngSVy-Yr?2NYPo~kYQ#oNa_Pcypg z*Av8Uu;v7ePi+$K-*)*&bt=i-{=NM$Y3pH&qP~Al5v;!tV*|wwK-;Gy zplDNdll*c)f)iN6IjtQ>3I>R6zee}MRIbbGP2+BN>lp|9hO#V8@}=&$rg=;7H0)3t70f)_;6*tL~b-ipjK zuv)Uaec!Q_mBTb}r@QNWGdZ=8TV~5mMVpW`Nl94VBc-Z&_*f#>76I*+T!c5CRtVtR z5Ssun5;4_v;$`1TP6|g;E;^Ec6!zivx3ZQB>DDPLvS;n%ake|3eS0X`wtwuB1UH!8 z($4kK9GVIVz2!EI63*E1M!4F!B6j(O~belhu=8bLO)qIz;@-Fh7M5-*FT>J;WM*=i}VrbOU76>F@LE0>IkO^rb2zP2w0nJ#mDC zs{QOec{Tw5S;~d-B6aRBa8CDEE1&oT$&EirpTu$jiHtBz*7+%q_xm^4u@&Lo*zm}g z0BhtlE>><`QMY=>h`;lUWOYNs!$f*g!%lpxeII-YE_yjk_v4~hEmM%36Cj)X&WOnWN zI}l>!f9#Gnq-8MQ&=|uh`7UQ#^VZ`efc0bSl`EZ3*~8_S`sdw94_9xNA*L(%2iy&7m4a%}y_Iae%7maSM0NBRqtVz^m^+ zZaa=mc3qN)~uYhDgvvqRIfwZQC6Gg$nrQs4_1j}6t^jvdq4NrE#r!8>JMPJk;<_UrKB zgM-m5{G8R-5-d=D9w zP+Z@VGMZvM{8%yk>EGrpcK`KP{&T`D(!ZDgE9U2yzfz2tX3Ntvs9(2{c-R^RJ8I_X zVFQNwaK>V|`X(gO#ZostOW<6!xF4#c0D3^J`LWV`8uzP7Eww>jZTz z>7D*$K)FrzjWe)a|@Wj#fMaqT3W4 zbjn%w>F6|65NgqCRDoZB_|dwHu7!FQ_d=lBJ?&IN6VP zO!+dGyMlcN6m`>fTu1S!_Kz%CLvq6MF9FNGrggTVNe5)vSoEVq=PoaTqN^$BF0t&^gh;MRx=EGgUzcAl+T zhWqAdv1OdCH{|UAPHfUo)pb}|`<&`x=u5H>Jm znt5o>0n?Mr0N~YGB2++Vn3o=Z^c#(#em#m{b*`bSo4mn0n$KNMJVl^D}!_c zUHp=o_QG)K3aBtPtp9=e?d1;9F!k{kh47|3Q9>)6wmOZ*2pOIxPxB|}{l-fL4&41L zL2UvXdll9hQ4!>efaeG@!gQqUfrP?L6et4tZ%!U!0qmILlvq-j?rnu~5K$p3rYWu! zWTrt3>#6Y)Mv{6V>y9X??hSU7n@(}6%i#Z1*)R4Opv4NKsQ}d2b3k(C|0l|lJB3uz zCYT%&KEt3$kYa#~c2W#Th|ST##o7{Km_1zBQ?gL&zcR{hYPJD?W2(y~5U)rYTi9m!tL@X9L-PIcJnh^Z!@Dtt8k>V-3JsxsJU?9yDZ6HE_%C z{;G&Jdhx%9_mqe)z#zlJ|No{Vlvl?~ys_0 z4Bw4@T9x4^*7_LZrEFGzxVY7($r=dm)FQZ_DEzXRGvi}5<%@40R^cw?2X%YIcp>7) zZ?+QgU>S_1|7y7)k>0Q{jM!1kxw-A|F^Ot-qWEF+_NzT8Ule6Fx!B?Ez=;3l>5;N= z3Rh89dT|{rCh|@Mn@*sFSgjDR8{<6Rdlee<#gP&{QWg zs=BtOc{Sks`6bp5olfAe{3c^HULPQU;J&~Qb0u*!ISy>IT>j7KN4UhN^H7~; z|5?kRF5mFcQ|X26yHPDuXAUJ6t^LE^hvNz#_H!JjcJId8M*Cf6>;5707P%^0xV8*` zo@hQ;N>>fs@{KQHN`CMeln6aGpr7g}`PzHG;5v6sBDJEt%ADN%=l>0Yu8Y6@GL8ND zEl<7F-A|k**EGOfq$1JA!UscY#9)HLXbB0$-an%Dka80Mn@wLZx z{g@NFA@lkG>%wyneCbY<^;>eoiecevuk&g)@iDl|DLP$Vzy5Fun!-U`WooSCsC9*HZt^f^c=Ut3#<>FZF!C!WnyZ7^lVV zvxMr2bMjii#usr}$3x-yf4chaXf_-0?RcqHQLR<8TEvVMqoq_!DX~hd*uqi$Yh;fu?PJY}cP=tL2I_zv| zO=B%QIXmdSlV8L#uw~Vd`T*+*>}OlkksPWq!kR=j2*ZQvZ-HNtS@CD(j&W!NQIAZ*5UMQorDW8?H<*4DHQ6 zC#YYJzR3ONZmEH18m!t)Q40Ht83|Bva2-E3k{Mpfv%uZ|o?<+3ATy5#G#1S!)3dc> z{VCwXXNoH{Q?%oIB{&UH6Vu4nopLtIjvH>Ae?YVDEk=q5Qc3*0mnCzi^E2vE2Q^XO zFGap)T(XGKYL`J&hv&yBsrQVC z(;$k4mdhkyPEz_AUlM)P&@@dar5;Kc zWBAL#LP#OvX9$J(bjJqNrP)h=Cm(+lx;v%|V|Vc7z~m8c-zfS{bh z`kkM@>=xYb8W?@3jm4@8r`(Dsg4Y5*^``45tz_DXZrjOv@bw9f&hB~cB35Dua0L>} zvPf(LU%F;adi+ zd!XRY3PgGzxp;&)As&?Vk>LGwlDN73XC-U?BE|dQq@Vjhi+Eqh&^uMNY72Vczb*$B z(%ldzh47IbwS6~>>EZ75c?YaEFi-A+K!5c%ApXC~C9-4>tDX<6_f9F)21{m3>~cE5 zvx8VWK137C>L)q1N`xy)FE|}M`1Q!acm&XG$sURU<9%|I5&%PX{{x0ZB>E1;iN7Nt zXKD6+Uy-?Rif_Xk-{5P09mdXVhqN(id_m%dwaQnN?eoR;pS^Vcpg}Mu5^QTw9#{ny z(w&4StzTd|d0SFvINlV$PMs7}gJgcpZo@Ah;X^Rv{2lpRt_4oMRUhwO&gPEmF{PRM zOBoNw4fdHDh6~}gp)RQj20eZE;VMa*ZLCYg+jXlsod_t@h__lm{TKA{wGvy(XR6NE zkMA|10c9P;P_uI?5TG|_4eSPiOjiHVg8m4O#D2TTaU^ds;P;l{H ztfXeUC~7N~Ali9vdOkvJ~ZE*iy(l~0Uv;;dxh z13-~OPNyK}E|z-3Vj7DxN{6C13^K>hCCh-G_z+fs71O;_1udMN07Pc3gm0&Ds6gli&2~Of{U&i`pN1mjVmL#uMyA$YBjWn}Yn2 zDcZRlYl92%>}P$Y+=tW;R#+^v5_|$lqG6M#XVCNXu-JaM=Wcg%l_j6ZqU@8>2Tf9E z%qxmINFcP=kbF%|0C=a3cK1m6#7TVX0ym@po2qA;s;nX?FI5B&M`N~~aXWalL&_ng zQwCUC)qF~`ckN@r*U2rYmOXVD)tt%FWa18zAmR&#Dx}=C6||hNLvha14x^=}f76Wc zA8Z`*1K;=_WH!m)f4i%o_2#QdzfSH@W+6OX-aSY}XkWtE_4L9)r&Z$4^n9`Xmh3VI zH9M+{FqUT|$d$9aIHtjQg?kq*y*|LFzZVGYV6B2l;2W7$A;XYsluuvZ{JR3A2`A^d z%Sic{ED_CO4yu$DfZ5*e*|Up8QEgk$w}T^s=1Z|C$q`4kthLbyZ{q>~i0-y+cayPY zw0ZjC#shnABF6wQV#TVvT~ZhpMY#Y)vpbGsgYj-Xf$JnD(0rhuqXD7lxU*gIYqeL_ zo^B3f>GRSE*mGU|)#sQhxzJ1*FBMGSx4;-0)IR$I`qlblVE7Mcgo~}J|1F^<%L#+1Koow#{zoB^*Y^$)9-+YOrcA`ul7R_v;`SOOv9x1GJJ z#Ls#7fJ*5ExC>PE$iAn;FRESD9MT_brPYMqnfMcqV1=`WmLR&2!kks~B##OmtSJL?h<0O0AX z?lO_qA3lXXTDFF}glK2I6xw-l`4F2iKFQCjQoRLieQJWwy>)Ta(V@+(1gu_dv$Gfz zK_5^pIX{qSYxKyVEnMrHmMhP@_S0M+6J{7ny5v1qR_7fLLUD5KvPm<$AIvzTFPsAl z`41)?R7;z`x0tlPCNNz9qB3Pt!@uicZ{dW?ts)bH156gz5T*iWIrO$B8>izy8Va=h z?EiE`8Pp;^4FpxZ*L`kUq^oSzNb~a-OoRNIo8S!fNv2=@aSE|BIZ(hqVqR>i&kTaL zKh+xFYpf~$1DZaVhI^E*)oWX}+kjmP#^&R;g8fOgPTpe0mz{}r`|LxKXLVs;97M`~ zyKc06JZh>qN#dk42=1K7Mjki=K~ZqT&wRav*Z#)MyOo@`^;ee9vi8r0E|-_{oo}80 zavZX7HjKT*mtS<=j@15DY@vSS)}>kZ_N=qQjZO$-uJg-6@uw7Af6eGqu9IOZ#E4|L zlHUnc>i2ie9uxInnZBZd=PYCKGO>qHN0Pa@ zQiTkYoW%LEy5E6Cp5j9;s7bpkK_apky9uS5=<(Hu6RS?eR>k7;F6txvtUq16$n zwV<~mXjoIQHxtG{9r(bsGEl?T0LlPqya{U)cr%)HE0Y<37n(8dM-nc)8l?A$0W7sb zx&dX~8a9({u_BBn4{uSJvlO8>^Pax@y6PhQwCdIxeW1M?tO|Vd6S`hKaHl(Fj>x;4 zdidzwXEQGCY=O<`^FV@M4CCkF~i zUqtQ|v?gvGoFxk!ZVWJZ-p+N-ZD~y$8}ew#V`=@0EdG(cI7cn7(99N7;5w3`t&kmR zTQ8?yIPxPjEH$CFr=ELUGclOu9uKdn{5TKP7DIHbzE*W;SJ$RQOb3!cZ5mkl-qc-D zf2_vJ3X-|^o}%hbZxPWfj(qJgWo!(TVaCg1nM;!~>1mE`yh`SpZgXF^5Q&F_l#(gX z*AcslWNb`LRR)HJr&ByuLK9eM2)#9~j5Yt}1=VPXa&RoxX^65%G^6w$+DK@WNa*CH zm;=DXx5F4W3h2$93X>QCqRnMg75I?t*!FxYwzk+uTx~MrP<-Ip!9pXF*o>R&iL?7{ zHIy=wtmm{!w`BGqT}!|6Amq+jcBp(#ERd!L|Mjw0x#fA5p~+UmR?Q4fPPd(S+VTmv z)!QDgyM9l4X@JUh^*N!6Sg-Byb%0Y==6!t+BpaxaEX6Q8lWnqn!A%mA zU9M7G1^`U3YbGk-A@)Rr5SV7_Xvu~TV=kPgF}=V#)zGJ-R6kWO2_V`4%rq{|g113W zrP~548{kE)PIi*g?S@!#3B$r1-vDu>y^L>chiGX5DoJ2y*uXkO6KBXI+|x7S60+|>NSUt2=|K@+HjtUIUW>&Y z3B~IZ@(0+~U0_F(@RwYJ#}14h`X!s80~Lk;W(s=uIE_W3h*<5cst*nkA#~{}5U0wk zm9{ooKN^$T`oEfCCcGl?R;JdkZ^^R#Lp&8uL{q!U2)Bk`sbgmXnOavq|NnllQlcwj z))Bz$7${PdM<_!~+A=cRN|=W-=w+9$Dox{g>xt_hJ#@319e*x7YG&IBybf<0tSS?b z?=*IJTl6!%k5;iZhdnO&t8XvMA5fLK#vVb0@NEYOI~qCR3K-rr*~oA+7dkD*xTtjb zo>P1C>3s4JD6P|Yvo|MQEKjY{TW z=m7cyn7gzz@g-uGcndz#*5m=R?>^(pGwnv7mHYwS!ntK|PFL5aoEx2y65hu3%QNor zI?uTNI6-{MYWq^Vfz(XIY>Vct#6*Pzg%~51q&+>}bItYJldk$V|NiA~xkYt6*028& z4zdEIUC^k=$en-K=yR5nX7dDpqT5m+ecGm26G)S(1ZHw`%Iu^5yOmBl%lz*q^HXCZ zzt@!(#ur4HtRJCus^zc9ZpY)Z0aw|>gI@!5@V17AE#zXWo2A2Td8U$(?J?M*N>k=p zv)y=R@Ses5iR(d*-kjA%rJIq?M2+*bXa!8#O}Hz$aGb8^P*LKd>_zL%ojc9PCSyF_ zh2!pisiJ7-f?&SglT`1cKcIu-E)P>h`_)~d;;G6=8|2ZNR3qzh>3n_boXH>1(zCE1 z&rWJJ&yEb?C&OKP4jZ2&&!nf@B4sV3?Eiom3ck1E5w$P zvmA(y8@v!lRPeZeotk4}QOU4SovRl*7K~J-iQ^X6!VwcNIeI6`C*lqu)3SCDw!h{2Xc3cA>48Ch znQdkpTkFjL1Vqa#P{)5MS zZ|?ivxSs5f3o%Ui1CmQI8l&5Jm;T$$VtzyM#K15#!CM1(s=a_UaI1F%dIbm)v`ne1 z(BUehZNXEVySooIfBF~jVtAVfcRl*9EIC>XnP*By`oydr!UflDcU1P7l|G@gI-YH7o{9RVDr~)u%eM)Dt#-nXSYldA5XW&s0hf?^f$VYz8QM$@({~MCgIJ-8A=egr z(#J~d8@IxdBy%CNNB#$1_{oqWPO`;%?6kNyUTmQs5D>L?dZD8m#z!tj55TWtuZ}Z<&lUjD zzZsRT^6ih7bs*GR1I97&94~{8lAXFc1U*J$?!QI{=fuR~b?69&d!$5>A<8BCNunX~ zAMXb6lPlpMNDUy!l4S)X(>!2*D_)I_ z9%52fRr;Y&90?PAOjjcCoj{;aPb-+G?n|@3(EivKkF{@pmITSU&}2faeSugNI|JIC z41AgLSfsP`h9t#2&6r%qA)Pig69y%u{U*$Xa%X<628d)^4!VO1ak9vSHEG{Eu8tc} zb~DWo_Fnb7O`oT$(mUr+KH`oAGv<2m_iP5a1;z*39Mlvh+0c!y%owa}yTcQACEK-q zYclrCrIToEEyNnWv?W7zHJ7SFP*qjI=sF>FiX}9eRR0Eom9?8qChL8-2Zh6+RMva4 z$5AW-Q9HWLy=qv*TD2ew|T7h3|$MX(RARGYDhfgr4%}A&bMN7Lxqc@|wp0|lc1*YXH|0f0?$}k52 z2F(SFNkxuOnL>0meleX?BJ?^J02&voc?3vk-PDFfAu{y;RoiMu?YIwigG#`?U5k&00H6Ze7%~3l$xr9UW;pc zHd^ov)k(^Xa~Ajp|Gf*XW;{FBexEb?7jW5zXnL>t#Hk9h;D-JKTz(z;sYnh|I(V|{ z5q_bPXbZ45_BH@HD9o~=XHNaCi>JU)&8e{42d zrpGU`YPBSL=t_^y7Rpz@Li*s7o>`#w8%{0UA5AZp-Oq!hGfa*b@<>^%CP94Kp@FxP$XJv+bqmLo?slct!?-dS=PHZ0pA_yf8U@`5X0?BbV1<v7#2f&2MgMm4*TX%Ke)BdfyIf>web?hgsLu7vDR12r=t`?W z4s5zm^i<(K-3GL-M62-oACTH<*5q;rW_j^^DCFd&jczvEFwWpXzZm_yjf`KW0wl%h z-!!WF6AVt6i=Irgk=S!O;{JmH1?IVg(Sv4fP9w&pF4ARkPp`0ypRh>x3OGAiscHkz zI>F%p;QhNyA~88`UTX(e#;f)6BGX5h)9hxuc!al+iQ~dFsd=$3Ncof%S!|t@ixvw% zm$I`9B^9<#8eh6#m8I1fJ&~cnGL~E-Ws@NTW?&rp&NhZpuNjxi){@)mC|KI_TB_%! z{o;f-+ez=V*w<<7(NVZDS8KaFx)WqKL0GRj!$CYIPdF{LipO)%m_2(`yj`>Fw);qq z?w)tMrtyrjE9+;14<1 zKrq{_rJ{eW+lGym6H2FX!e}lMzVV5EJonhyEl@ zQ#NZE1Tol85jN!BD7FQ9a=lTtkFO?lEYJ3YwP0G)3q}sWBK4BNp~TrW5%}N$L-1yB>4N zWb+4usRIs+$g?f$E2 ziR;&W#P8}&*My4OUo-kxc-nYqeOB!_prbb`cu76IdorN-&2K3K%H8MBt#ZnH6%nF0%-7tz86A4TvACu)TUs3IVpun@)v0^wm?>?x65TdpqTuAA zz0y;-Jz+T-<2a2nG^j3WwSif7HxGKh5O>q=-_8l#67E`5PcwV5D?n6?qU<1hc)!{b z>KTvQ>nnckxhAnj`!Dly=vC3LOl<7n#xIVrf8yNY*CZ?RvG48X5&)Q>JrvCaAIHm>vUi)XYWRXAD?BOI$nYQ{Y34=A#r zH4polbOlVeTc(|Hq&pwSmJ8T45G0<$@{KW3XQ>qod`k3f1Ytwbo6&ZnHevv=gu=m^&1BvWBy5=y{$&o=)7AlQkI*wfe)?1D!@z)Vh(f#2a@I&aReI9o zAPz1}sHfsfNKKLR9+9^sgqfw~4+S)3xvC`1R+`s=q1^s^_xH%i!+65(-W2M~a@ax6H&C8w#k@y%-u1&DB5wuBo@? z(_uWnBAAX)FMn$+u0dBBl`q3K`B#=(o-tyYW_RaPru1#XYxUEINP|~p-NDd3)3opK z8eBX+A!uETpD)<$&RK4fNK>v%gl7u;;WTqzufVg16>z}O1sPyR0!Sg<1tH}Q&*{n@ zaUWf%f5<4b&l_r#8MSEUE*fc}VcSL}+NlDm?(#%x~zLksf8SJkrMk*Ryo z-pZIm057t}YKWxU>X{?P4N!Uj5<;uqA6-`{pIjd762+_Pxp)__tPXZb0qYT>yYE+* zK9&{Ko~qD0WVw_Tda@ndkP|{9sks%$a#js)eg?zCT~>4}lnS-n>0H`L@6b^uAql1q zFL0byFR{Q1M2NrT+p!6lRKoPOMLw4t;0&9(AlUICLh|fnWW0HqS!w1dKK=AkQ)(6I zMbkJt5s4Z~T&YYR354iAIsr?{F;nnwr~~O--@CtegYE-Y#edg&_i;f~pzT{AIi*WnkUZ>vu|_&g5pPVOPOFW3R;WbMZ&vMJ zA%?6_-BbR|!6;q>%LsNTP?j0@1t8maZF9XjK5=qas{2aR*trl?f4R$1xXeKB}+Nz$TO!l)YQRmXwh9+J!T&;cMH2?A@}hGH^p&`YN*x zMn_c|CT+%F-Ff3EHMr_!S5D`Osq=il`HuI>LVDZpUqE9TVxqn{VPR5`Ite=mlZ z!F!W3@-=Ai{bM%z?m&>^|2;iq4Csd0#5~7dul3)fUtz%q;0POrb{R(lrvk@LwO*

=B_YfVn=wNjj+u>_b4<3>dX%p7-n^7Mi&GdoV^bhw1M5XE&r zH$o7zwAk1am?ACw!_qm|A*hvX^TwN=Ba*M`T6{Hd3c_| zpf8EB+)cRLOv~dciFo9AK62D{geZQY@O|D<$!HpkP_*{t8nod5Njg^a-Qgl7-KjTM zgr5Y4bz1fwI`_3WPvF(ndIhWsg%ypAYgMDW@{%ytaiygJ2P_9-QEWh5JdhtRrkkL* zB~wmGZG}?goyL-1X$VnY{+le{y9mnEBD{LNWaT<7YrL!yKGj$u#8LEtU281xdBBdb zm78DrJmn?sgN|6yRITT-*uQP|&8k$h#fvKz^jQ{2!Dl~qRH@qcWcp`rwlo)>h;CM# zcm4sTC%dqrr)Iq`&-H;Z^mZvrk?*CLZ~WVnX4O_<2P9k}un3&{uhBTv$za%te`d-6 zqjjbJF{JC4|HPBgVczCVmqx?|tk|;e)~_0ZLisk($>NT7Iboyn7)@rHp%{IbP4{t1 zYayk~1Bu%)R?759p5IM#eOLQh&y{+^SWPDv^PC%egBZt9$}I3vvA z=gVb)ojorZo)t#3e0y!j9v2u)*XEB1D@$t3j6 z%*48!G6O)ZbYfZL*GA|eW9$36!k^xH2mDu6jZ#Lw z@!2nvxZZBBM*1io!#9@n^)wmZ&%I+oHcwcPm_Dpt8R%45z|z;LVHhv zV5xZc)gN5Fy&4B>?6Gj|cgH?c0OOkpJe<<@(arOYH`#oRFsakrUH-7OoJU(gr~}im zk7GYR%n&kdVqmNS+pcAIY#C0my;Q5!m?g!1rJaT*DRClI| z*>;#3ds#^m>2R4oM0QugQa9jcC71X*3&UU8-%L6YGe7Ot>*QTwlW(eKlWqhBN~v9) zs%W7PzgsJyU9fmfrAO;0Z-R=o!!u3YvKN?b9G5Rr2*lAK+C*zwLNLt4WMXm`Tt_ff zpJf+>Q(&{?CXbtfKro;NW0suJ;vuufUDo^|(Rxsa&6&B6$juvmI0Wlmve+0E?C3xG zB^|#IAe(CC!JVl1TIvCp7wB4C)8E{uf3Mznx&6SDwZfpxITnG{xPB`^NZs0MKe@Qa z{P80P#+=S3AY#1gvjWsnmzemTHrncME4^U&V=(9mZ+SS`UGlpivOAICFt9H#h}y=B zjLf$iP$J$0%3U6dMm&NMx5;do?Cd{F-|gI+igzV1V8H{$CLeB0mf8R*khejyzj}xV>*6BlhCyJp1%0^KArf(C(_ zaI-`GxM$`AIhx@q`&(0c5EtJj)!>H=u%ozV=Zc9g6i2(Ums zTDd<;Uc-VL{W^G6tyqT;AN=o)4>^;|Bb2Nl-wwQBfprFGE6z}`3?%Uz zcYUC2PTn{WQI>Bxl2Pz|GaxPf8x$J6ARq9p>47Z}Yu%K2-Qr)atFkgHO{QvDt#>(| zt`g~I$?&!Y2LfabY|0Yh?$=wQ?hleb|DWkjUM|d%uI>Yxb(ZL@jMw;h)mRWvXneSl zSkwU?ENBe(jbK&x1+8BJ1M~b}&bwW_HW{Gx=nkpB2N&J>&7uCGuDlEIR22kG541H^P-1RK!7O^<>n8XEI@f!-AFWEYD`-Sm$NDL+CZWw$xC@?7OXLT zz*%sX(1Nh-Yka=?(gc{FRCjNj)Bw5`qYqcDbNUzbzrzQeQY(^!Ki{z^r1E^0U)8u0 zdmNw5MmPqY|9(Vo=;fvqiZ>7mmuFQq(n{$CeE}=-E|GIT1_80*WQp%VGLM0m3j!pz zAei0aU7Dd3dKp23&ikNc45zu=k^agO4zlP9v|z~9;`6MHU}KIzv{_T2gLXk+(qk^+ z(uK~x3kN)|;>|@>NSxUV5}|yjR(5W!uc04fDXn3*^P6E&TTl#WnxG?DE^u6UsLb{G zJH-6~5wnIRfDd&(!|{MV>Lk literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-2-Q1 FY26 vs Q1 FY27 Revenue.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-2-Q1 FY26 vs Q1 FY27 Revenue.jpg new file mode 100644 index 0000000000000000000000000000000000000000..ee835f542ca757617c80283763240877942c7c62 GIT binary patch literal 39357 zcmeHw2V7Ixw*H9$LJ<)J=>!EuMQPFn0xDfZY$zb8NEZ$bUKnO)qQHm4+5du=B zgWCJJLjC8z4tj;YkhmIZ*B5t@-VPl zOHES^fIt8M0{#KyKHwBUOG5*rp{9kwV03h}^gCG@cQP>SipQKUD&uqodo&u#{lE@p2$UH@t^(j-JEF6226LNP0J0MW#4k{=$ zH5ChvIhpCAK+_J-DJOV>z{gtPo*$YTI<@ds=T4E4=AUm&OK5 zA{U4IS50;tf*0cYuS0z-1t)=+!DrKOTr;0A8HmdWA_K?+WMIq@GWJwIiwsoCx}Zr93q?1H z&&fbA_NB|K%2i`FJlmH5_oEX11|QOj1)s{D>%RqUT|Y;o!>X?{{0(ALDqG?r8RWDO zUmNWp-pA^?tk{xC)e1YV>TbXJ8ZEtMS}jfU4L;)$ft@WvB)P^05t-8o0(yozHwP|i z&;`1LOie&HlGk~-fJw^?^vLQ%LHxJJg6#>VhWkS$2TfHmvizo;e7RHU8wzUgFBRL| zw3QusOz01E?en9O*x>QB=@41g z^^?a|UX!1_&|%);NQb_5JwPwCRz^v`v|wi{be|?#ytSjoafn z1#m|u`?~si{%&V~FMb!Fir#~75~uYKGnRZ48G*eB?zopZ&a=~br97jk7X0~14v1C; zjACeKEcDtqP~um8B(J}|B?c8x*sZ4OssqSt3PxUzRJRBVx%y%}Vn)Y($RCZC674B6 zvGvlA;xncVip#G>ZzS!hA zw7Z=t0M5vM6&IPz-?7wc;CV<*0BTuZ2i^Dd9*st=7O>WPIO(B^poGF5c?oeUul?`I zfQ_}u((WQMfN3I0oy{-i9C?Lecgd(#^Wi_U_N*4CCTC8>7@dP<5RU{C4-cnb7^Nw( zR|uMT@ijj3q_e{lZ};pQ>X}lb@};Rx=`*Wt(O-S%?<7fR3{K{^I-<`LMn>HoTrGx6 z1em7_PadgDZLOmL$bdJhGirez5FT1C(Hbj5X0Q0}4gAtOd8yaD0Oil!S+v`o4`@kp zp>|*O5y{aFgzvn8Q%iEobF21KYG`_rX*6dB`*!z!7`6m(5sDU#o^(ZhnIFK|55yX+ zV+#$8jR>nNZe1q&4)*Q|S_a(y6Lrem5&d(5?%guPe#{%x3KHwg=zad&4ycJd_K^=#W*JU&1f<({-&P{U?0V{w6Yv2geGY zXwF6fi*P^r$mC?AO51atsEfx31kcwz8a*xaW#@?}w>I=o;a9j_N`=ncpy$)R5@UpP z4{Ip}&}um6AQE$r^?o>0pL*iV9wrg)prZS?#fKRuLSUoEjZfhXE2F+V7&{Rl)mZL% z;M=$OqD#F8j9>H-Wv4^>&%u~fSp57WOY#UCDQLv=xN-ZPUD;Ol zwBaLj34-b|D$rCT=VhMJ_tt?;8wyjj@3eW1&jCCUi`|HaBw?55+3hKAyop0=7vsn%D{ z3h#ClXYR^LYh;%8z~of4eXBLP<ZYr*7q;YTPKj(dP6GkE*84l@rdr1SDtK&G0hrZ8;xbRQIhCF+jGsg$6BsUp+FQ zhbBHF17D}MmPpugGJpVEP^b9*M2wxPn>z6bYCRY^jwS;ZTlQBZzgnH3 zWT!`b5hVk+kt=SCZd>lrN6f}+l1bF@G~DraP+Joso@|%S#=&=kTA}G32%E--);Ztz zw&R(IPkPatUOHspO;i)Q6;vg=lpcRGq-NDK*tTYDSNDFTUg35nR^%rW(q~YQE`M3x za+@5Gt8zq)q=o*lMWUW68(xS*Pn8weh7B^Tw2usgy}CpO(Dj(5<5zQ&M&WCmxhCtx zS7hLn)GL!WZ6qE~=D|Erg5_>kE4HI&KU}ANqQ@aJ@B*_5&O7$ZHqCCiO(4pjcUbeX zY(|&0Xv7B!kDe8a4?F~H!v-z$I7J2oq|?ZNNd$7rJljUQ3bAn98ne76L-Mnc#+*tZ z%B~BP+y(`X^>#I4J398mb)q#0pyer!+%N@=S?nw9*89{%2Kpx#h;7j(4`RYpS@o-4 z-`J*U4wyG#i%#i&zD47B`KuG`m(<-pgF>NmyV~$CHHBjwhD#vb1hEJ>(P@E}`R+b} za^z#n#9D$p2CBF}@DMscwqlz+Gfa@JN+n(NJEgO^3xyk-3R%tsWopWHwPMFM_3nr2 zEfE+|CsvbzITWeI*zUU*SnPdS^^`#c{D*IR(5+g)c`<4{o_nckD=$i05CP%n)}j1s;ko6&%|x_@DsW zqEm8;zf(C&@`y@Q{E6;zRo2T1+trAFsd~5A#}o{+Js9nu#Yhwk^DjD;O2IJOg35jv zyZi$oQZUT+fU+OPF7bbF&v&*l1;cC;M*FAn(RZftclU{cVg3)}>9*L$6b$pnqcaMI z`S*&vqhOeA@-6+dxaH^PQ!oq#!~D7rhvJ9%1?Hh(7z&2@%`i=hALhGIUJ8c!As@+~ z;jf`!m~HSX{V;C%d0Qxcn13n%%Rg996bwVbFuwvA{?%g1Qqyhkh-QrD=K6jqc==GB z$0pKT5n!A5f)aSsrwzV5%dnB@z8rIUTC_*v*8N&sL>X*K2XK)B&rl-3(L**|Z;KMH z8JHYhEFQXb{=8~^b6X4@T+bLTDbd%AYvEQudizpa4SiTAv+~aPLK7;2kRE<8OXStt zCfl7^C(SJi%o)6uYa8fVq&|)z9?4HWSQ1aOmg#kD<-B|>Z+_x~!{!8($E_6>s|Uto zKO!;Mki>9HTwTGWmu$t=4->!f+vkKjJQB!H2f#4!ugLEym`)L+b`MDVaQXfj%V|xg zv?B+*PRXBKV3*^!e+tEb7d`t)nzBvDFcS`pitd?CP3_mQOS+;_l;9q!T}TF84$aML zq(`3u*^n&tlWoh5jGC<$4dqYrGvx#j=MO<^Qc^SKM2?Dl$tm(3De!I3MUp(HO~M*rV>4_;@_OxnG=Ppx7b_a6QI0M+R^$NBY&rv}qh#U3roC^)Uro z-1919+Pn?hI={uUbu?c-hwo98*&(4+jB*2pr+(#e`jObZ|Lz)!FYFh-Fvtuhd$}ZM zh=rl~UAwA+A-&J0aO2#`^<72C zk#y_6%N=G5^XYx}!X;>@RCvpI^h;b!N3nhirlZ#m@|srI@D(iXS8-)Yd;rhtzgy<+ zpv>p+&ZYJe&Z36V?p#N*r;lDlbDqe_GesXS!?C&dyoB*QgGceagSqxLcDBdLsS~_+ z7w@!)N$y-lJ9QTyz!lZR2zTaumXZVc(kwaig53SE=L*A!hW?fF!WG1;BSAy7E+MmO znBMjZm8dapj# zRLvGj%mtoZR!St+;*tp3Z+i*H z%iWTzbPU9c8ZF=J6=pWeu9}yvLye{0%{Uu0r#qz@*WLQ~F)S$zm5ONaK8)4FeLEFY z<)zfNRF8FJ(Jj5StV;cuXpgG-=(c3{zz4zRIJ7<~L-MwSUKm3JNZIWsj3ODWDn`CK z5>H1)&S*H79S965XokHN4r;$stUt_np4XWB#GI0kpyj5=XXRwylqVS(TM48Tw#T|9Ie6!t%vSAAhoxwC3T8b ziC)HLPEpxbWMH=v?=>p~t5fQ?vdrbKF(UB-MUoEaZE2#lI11u)XiLa z8mcnGD80Rt%@o7wJHzQdVwb`mSO`(&;7jS0$OHQ%Qu;nl)SAuJUYIlBLVHj4PrN9t$|h1kPgN@m3(Zt z+~j&29wpO5q+{NPTE5g2ixkFkyR2TsPc*$=PbzsWYQ-5${Z8T`UCW7=A2STRB^}eD zSFi?`+8v7;$9B`G*fMBlD;|GPt2L#>X~Rsg8$BE0?BbCUX-ZXH@R@_HeyLmv4q6d2 zBp`A@=g44#>HT*fJX3kI@?~qfO$;V7glt~Dyu$-5e7w)lAvhTbEZGI~?de=X@eIwh z*a=McK8kVFKl1TRs6^t5+@|nLoUX5w?p~Mh_72^=urZc2U9k%v1Yv?%Q?he|g&2C` zJ-m>=9;<`i78z%`5=d05lqa_-(2F1+I)XkFS??Q-- z!0{E|u5`4At<{Y?BfS-!>W5*n=5n=I4&toy<@}i06_fV;`F@&-i=lJD+VEx6XZ=Sl z3Rry>T?O>fc5#ca=BoiSn)5bo#Zvg9s?xo<&9}j(rit~o)=7HFfhQ*gV>Ci{k)A>- zkTQ{?C;AS*yVOG?W>XqZ1MKqe(AhyTt_r3{PXzxd;DbreaB ztdM(K>$>@7C3ulG&-H|OXsuj7vdAdt+7+N@Hh{TSPfhi_qJ!3Tf z1W7f`Zk=ed37Kv9{rtc0!VUj2VWF_Lr{AHqpNGu+-hF$MJ(zKWJdoN7GN4-disMU6 zwn)w2^e+FeB1r$b9#^J>|85TSkNRx1R0vs@YYS3e!2|+vims||m@R_y_a}G!%Q|G? zYQLsC{(#RR)gIs3AHHGeo$H^qthCfVUr2NEGn)>#?f;TTt@n$Fx$w+wUf*f!gIy&pbi_Z@(Xj zfNJK?j)$dyw?Bpy_vfcAp@6sF4@1D2!9PE%2?f0U5y7f|de##Pc>CQ*;Mo?aFr3h^<6 zAxa%&V3;Qs%(cCxsLbt1Ipb%;Lm7{MT@$mO%-*o6rOWI@A!Be=ho1@DpBGIIZ$1wqsS|ig zn>~Yjg&9bD(6q=_kfMv*r35MV+-EiGxKnPLk304lX23?WQZzDXNB`RX7UYnN;)sjR}BWl5B!!5Le zca2^5Yrl$r@X}!@h9QrxeoCAG+zrOTYvbW!Rx8z6u}81H{Wgi}F7ULg?8W$tdey$S zTI^}*JeuZR*`<3@vD#X+N|!Uer56_HA8ebE6ZWKG&EY6?>vx4&3nzkQDb!Nqt&R>oM0Xzo`S5&6M(X=ST(T+8$i5 zHWO64bKSC%8OzDjZRU=x?R#nGAuzm=`dH$VpE|$kQPY#iXqRIE+ne~QDAZ#o*`N&~Vbi51>b&AKY>tV-NbQE<&cOQEr8Bo+~zA}bYK-ph2sTrbGx$&-#fYzX2KbuMAo3N2~n6stL zA7RuAi@5wHdw336x$cxsN>5fy%I#K&48w|s#TM>za|@gy?IIjg&>cn1F)Twh+VP5& zhyin+cd>UQJpD^7JoNC{1Z;JO(dpU^d8}AZvTfYL&}D^9ab=#Ghyy~G-e(rMThK2sXthfHr5<#>U?8lh^^1?v51sxmc)Zbk9Q5ashk1` zExYnfQ61c>uN3OBTh0zsv%(XnZ?t)H3iOAw^7XgE7;SKAwK9>73QF9?%$aqI(ltY? zjk}RrlewM@ms&+gjcHb^(ifORvupJ4F7qYy>szD{ZSe!k8~vx!HbW$OH28g` z-(B^~@Hv)~sIxXfc=qZRj@i|PvxQCFu@Jj~z8k5QKp$!#|rJlD^|@+o7LaQO4*G$ z`+Rl0cC*pAt!T_3L36kDkl88KYnqP?@2orc#Ga@?s*Y;9 z)%-LNo~WeGzfKP^lkO2>MGEmj6gc;Lic6rNjvwK<7PT)b1TE;u+h;jvpGBd%y zLVnfPn7Vsu`e;+xF=q42J7Gp_g3j*^R=+XovR4oJvh(p2J1OzEvr6TJxZb7NO%j+u zJV4WKu(kA&D#sDM#Vx`OwM`h=K9mt&nac}~i!2&Vj&{_M*T!V0SZRGoQFgnP1=nAc_emVtPF$o+-*1Aa8V+-q#u6R94i;~W*gf`; z8A0vIOCIIcOkNm8m>qghbr0#{GnA>At;}}Yi``{NFW)c-$vC)nTViM2X^nL`nk*+Y zZX+L8fPQjo8dJb^^AyWf^N#Z$3bG66IjZeB)yk*tx}x8dvJhr#yd=6Z=5yWq9NTUS z96Q0_r1MFgp(u|=bKM;q?4g0zB2?C__9kjk=baXFdx7=V#iGZQd7}Sd`uq1mIloor z9V#yfHwo4d*m-^%AymuRn2()g8jst?n)Lf`}Mv;=b@Quad1IKsR2d zL3i9HkFYUP|5d?Ks_JC7{8Q`GX$=9a*Pm^P^B-@$fx4P)KDIGUdMT=i=vX6r2yBYb0M;|)Udu|j? zMbMD;EuC|ijsg~R1MCw`Sc*?R7Q@&&pi-Oe<)TNUFWST$>2gxtc&>k53o{=nyD5H^ z6*3>ogBe%F>2BU71HD$p?8ejhM$%~$Dlk%jY4h!~5l28j!DA;G2v18!V&yPPP-6$< zVBB$%RtA_OzzAG;wv`cZV-F%hX9flFcaq;%zVA83+h3y^+*YqQWp5-JthCcwi5}Nq z*fof@P_Qo)dd?+ge$;+2)9dga@XF^4*nG&30=YnFkG4Hk5ihJtSqD;eNps|J2Ud*}6(}ZE<=&W_@yeb)p?=GJAKA4+ZQ1UUPx0_BV=2Wgg0K;*(GVEXp zvH}Us5D0KWa}QJ|S=}8~&vymx^MwrT_Oh5ZEZmn#P{GAxt1veMC#$$JO%5l*o$vOu zcOKFUhh12A844&iLlX*TTe_V%Er5xAu-o=zpfKUV@~kqi1!&a3uex9^WoNcM4Hh_a zJ~U5y?a5s_=!aoIxQ)mgcXf6sdm`!CGA=9gc=4*k(tgRO*{1!&^lODB9XgS=x+~X8 z^0NBdd(T!lo_$00pk*ZIi}yZMEgY8PzrmLLaH2F{_vwJnjuVK2!5)%{7BY|oC+Joi z?~JV;gnUx_%wF(;vj%8nOn!D=)Vz*YP%|Pisc#c<#5_634t;>&gxRHUh8=JyRhJ*_i zX#`VkED>_FH@G-3binZAr!=9HRrkjiPB`&g9;#W3YxQ6ov4g4a&*{d$4}b7^zby=Z z0CNPuy<3I%v}_^+$iQs(PwC9M<`LPk>!6Vd{xsero2H>CeS%PGgt{#$4^` zA|rWWEOJ2&bQ};r*>&uvOY?(LV=J~e%%-FNnj)Z22K;QgL339Toc{B1Jpt*F|*c z2-g(OdL7|@f%#A!+e<9IYhF9FK9;|lHes;pstJD01yls7fQg7w=F&+>YidA-g)fCACJdV`-|Y{q)d%HK7Su-$UJYZ*d|V zS6MZ9{8o6wr}CS+GSB5S@T8XeM1)eb{Q=!m?aplIUb=*cfQY1 z$JL1~x7b#&Lsp|r{bsjt}f*gm_4e>4oaO4Zk$ggSyJ@o{5SS?yaEWTHCP|lc=MXaX=S818(tk*w`yw=yj<>|-8Cb(=Hr{TeJ9(r; z2>kUT-kET7FawsRM|cjMSO2D-bRL}H9ven)(t-9OfeD0ud}!u)?-QV~CHRyX)I?jc zKk`60yfsR!e(P#_Y1fPC%6zvp-0bLqA@AJ7s8CVK#<;yRT_G~aoYGCN+-i_DGro=~ zi`Ke;v`-u~_%!Lz_pZT5>4jP^V&`Si5U-;>)_kDR`N*(*{YLxJdFi_9lf^-st#PlR z{o$Huq4?j=L;t}%G4cp%LdQj%QeI(XR6|4MEgYi%r6SXkVp`{Xu>}rUBjH(8CC}It z5>ITvr!m%I#baO1sM%Cl#xcE-!~|j(i|Ls2a2a7K8WDAGof{!2qaqd!&SA(hTP@ML z-PXM8n8LN>*~3V)_i>hOiPU3dQ?P3#B`BKWjHg-rJ2atjOL0_ab%O5Lrf^yKEaN-F z!B3($G{!cL2wx8#aF$r#z`p+b!eDN?k4yJi$zv*XxGV0jtfnDv)!hm|t-{>P!Lz$+ zmDkrb_it&2^pg-(#}1Uzi~3NPd|L^PaDJ7g@Y=O#eZEE4dePtzfYu?X{N_Cm=9F7xpraT`bbBcjOk&76yQf@cd zzY^0X3B4!Yn`c5hOV3wu*j+!Kco^)(IRCi`S`ln1sEe0Nf aH1{Q%^VUoD{xQV5?T8@xtJTRw?)eY#j1(6D literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-3-Financial Growth Comparison.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-3-Financial Growth Comparison.jpg new file mode 100644 index 0000000000000000000000000000000000000000..ffd0212da1da718fa0936e08e4f35a75c38064cd GIT binary patch literal 39423 zcmeHQ2|QH&-alh7$Pyt5Lx>2;mUXDdRtarW)`~)s>|-4zWQh=p7?dQ-RJKU=y=2K! zlzm@ktbuKg$wZW^6?3ZNQ(*X+AGY* zC$6+-?>437zp=35|}q8ygpY z?|wqs!$;|lGoEBVeUX=6P+0V`xTLDOrnauWp|Pp`eMe_k_lJ)?gG0k3qhsR}lT-MG z#iiwyRl?f(mwrJ2=wG@8{{2hGzUdb`*e@DdS|}~-OTQpA9$>C-Q)vF*H^~##Wd2Sy<1F3BcHFm}oQVV1I6W?j`KwPrr9;27X)x9rP>DyIiAQhCBMdCmahpe?k1 z0&CWdwxIx9?RvcXCY4qaN_NG`)-dVmZU~ORAG=uSH8qU#AAa#*s}6%aa2J`|)m zK1-6emQ5-25ZNmCZj`IK6{5{)1kv^o8#KHnjN?p}dFgHHEgIRp_3+!>)*^srZ-h&- z;o-qf4;5@?Eg;G@2gkk3zEdmGC!HVs)~G@~7m@hhaAtOIN=`=D% zS_|Rg*U4w+tlQ?8qYu~K!^CGQv=HD!?gax*E-Ym-dmo2C$$#l=r8`rB-Y1)|QVu07 zRwnQ3RM;bgYkXfI^Q=*H)okxkx;5M_PXTssc{#d2h5bh8F=v_w_(m?75aOvU)GK6+acAz) zXxTO43~jZRyCfZtEuVi?Yj+QbAtK?smFLC`~xCqMzc^SWOQxw7@%I(6zbhR?XAw} zz2KcUuBZsDS|$tEzgHMfu^w>4`Z{Z{`|Mzlh2L5pY$S_U!Z#!^8X+f@ z9`K6ea}%B1XCWK1Vd$oPuks4wdtQbi*nOCaiuJcuuL=fq**=QW?L2Y^EdgL^R+UuB zXIe5W?WNoMvY)RVf3w=ldXg`>_!w!rB5Oh>tSeq)AMRq&tx~0|N{_)Tp|}?}q=NQd zJF`A~pb|<`=KB-Ks2?DQU&AU9%p&-rzygQE^~b}f6oaoQ0zU7&h&4;J)@)8)rRxPm z#A+!l?Rvaz!H&ez79|A3T5@}&x=;X?Uvxn z)iC25R$j&`;|=Q(+Xz8b1{z(ez7p-)Jrv+lV^sWm&j}5GBYNrb$7ZPRp}HKK6rGqx zH}5@R<4lieoIk3+-2F`L&|nU-swca|J;xSjuyZR z*`!OdJU;VcL7W0uKY*+uqxFTGB^!nF8b4$~N>00} zR}pLcV&4-zJ4ZV-L<}yy^+{wn&eQ!&g(37_bu08Gnh z_CtgAwtF2!C3inaS3K-zacmpBMpYno5gS&jWjBQ8($D;8=VTW*$9}wP@A+$hm_JM` zIaVWS`16UQICTneGqotse#^LwNJRKW+S!RiRs>xm`}ormmBB_XngWX(T4z4@$po?F zWG~JZ#dyT-xZ?ZnC6>zob|&h^;~1Z}UXZIJnD*C-GCeZ2ZNEh2CZ*m9<-Ek`jqWvZ z@oOvFC!zU%Nn0b05lt8cT+7(oWqfwvBx@E?LEL#8m-WsUB?j-jvuo7wpyAbm#m}yN zl6VSmi)gnCeWgD>NbcV3uu)Xza*qneCC027;562zwl6d|m89~1c0DOQnF7enZWA%Z zC%wAIJX~%vPXPu`qC4y)x~2=$Vcd^ClugGH~BCHSaIFFVh`RKgF}x9bn&++rM(s=bLo#e<~(0>O~P?c5mO2jAiA8Cv}w7AfEXvj*I)&~F3l5I zJqi$LOaY9s#8?V2GqHK?fBLSTt(xqp?WIeUh+Myo9>G!on@!6rWo&B{07llDJ^#Dp zL=@oAEd;~@8e>yABsZRO;%S9*&ai(C3gl82Qe zc=J0|oL~7F5`&&7FjB5M+sx?UK!8kT<>+CikLeVka-c z=kKhMkG99xMJlC*O3p@9q=kwDKS6^YsqSBpv$I@YVZ8b=J!pDIj6S-~J(Xt02c)qW2XQPG0q6Hf~TeumbLTP23~AW43QC_rlH>J$#nhHhw1M|Y1P)_Jgg*!UI- z5W;pSp=<)A#@gZ^!c4GA1W?P_7rEg;0bU>Nwv@s)OP|q23NUo2bqreq(xc3$;*a1Xn+5;3?CnM5 zYJ(4Gd$l_g_}R`6{#*5+Cq`(4ZX z_p+Kz(qC^ggG;}QE;}^7CU@A97eub&*h1AmL>u`NdhS?cHVYTXH^OHP9J7QwOUhYu zH$b{z_WdF3WV7Vg{+h$>eMnk>BN`Kq6WYd>jJ18AfA}MGkw4=#snF&R(QJMI^rAwW zKP^n6LYqG|M502QKW$_M725nMp%N9^{BV2i4}?inX!9q8H2($NsL1jQ6AO2M)(V+zR4dSzE8KgUDlq#cRlJ=%0|X(7kIuxa{4uS)eY-719qtf z7M|Vm&L}ZH=G=Y0S@DFJ`s)5*%nBQt6z8v8&PRYEaomsMdi?br#u{%-?VY*9`tU=< z@Y;JK9d7Xye&wME_tChOJLf)`H!z-T(S*6LkWZPK;m2>OX{SuCE2g`P<~@hLp`%|B z7UZ^S6y!VUrF46l1U*#VmOuTTB^I~qB-4Jz3Q^N1BFsjxCq-50pu%WU#Oe_9h2>*i zG^msGOM>I@gZ5E`KpYsbRjK&=vU_((rtQKV?$PaZjFT^|!v$*id?nva_siFm#m>*CdK4 zz;Yd&1RxVzAr{XE<^$sVL~#y9w;W521>*VUCOAypB=*&?-78tH^;GxPD0>Y57+;OQ z{4n6H5I#XNUWji@JjV0*K{sjyIxs9rwC!#-5fbP@eOaAJ~j@Q?@ zi>&1Pw%~7Hb6Bt z{taW}$rU?C(HTY$$=WOJ;&H#2*CN&KsNyvR!UpDVmkHtu69wX(Xn7xFczXF0cYx^eN*`CiCMu$3uOj^R%9t%0J6?dQ^Uq&zi>lfwPL2D^>&FT^SZ8X>VX! zX{VTzeuO5D{aLix@VvhUP94SPbt;mNfUMyy3Q&Vuolom;ap4Mk^gNs8UgVo<`~vwv z3Nz@L?`Ly0kURJS?{U+{PX0(8Q(K(WXv~AHh6e_2unBh44MuL6DwFQAYF`RNE4uHz ze{WFOVy6BX_8h+UK2Ep7K&fi|L0i)4v1{bxh3-D1y*NJR^@IYP^Sfz7!jU0g>w^pST#5=)izvP%mMNxicw?|X zZ0XGotfLRYn=YTpLf`zrDgI`_Md52^GBTY_umXchm8#2^*Y84#@n8&R4h0}Nfl(xy zTw}|VaDMd2(|+(%*q*wDa?UYZZh>e8a z6pLFC9HvmLwmkzCrjww4&g}7;!z^dqE%!Rl3(B;RJHEEiwCALU%Y&};nQ4%zNEoR;J}`79$2;KIJU5YEwu$j zxW3FW**fb2RHt8d1%!~W>==UA9_SA+*&dK9EgXN`<}AO_IKX`gVbFYiT9XTXshz$U z%dwyCsS^dDQ4pa3+lgEFYo7G8%$$vFkQ%9MY_TjhCN~cx*;_(0^9@N1-HaZ%>biMP zRcsKpDL-86gYu5`R@gI?&9K-_H+`J|)>uTaMvu>v@q#l|SpqP@Yrd>3sO}mJ%#0ve zn|G$gdDZEd8sRE@UN76Qj{?{$w~U+maS@b;B}ray`|OU}twc3p_pUN$6=*e9ywfL2 zP=Hxz&c~_b14a4yu@@D&`;NZN5LCW!n54@I8R__ggQ}?z@NY4FXcn%%2F-;!C?l-j z@!7!zlXo-?j8O*s<_ANUC;+$5l`biw%;2P}xuK!T?az6NDZ<+BLC5wfM!T;LgyKxZ zX~@rWro6~0tLp_b-5!RO(Wl?RQuu)tzSI4|Wg@7&ssy|D`R1lGh3RjD5P3BeK(_AK zS^XKw*ff82D0#bM4H#mCUN{VXv$s19P7HNNC?xG4Ht0|&S7{$*cCKI1Lpmov6oSC7 z=SqYE5`FQReYQPGW$EXYVmz)4={@S%Y0pX!3?PJNf*{|%gq$(taN58SS-fE|;xDN` z`qUb7y~`*f~bNgMD2!(`#5 zF7ov9302&}0@m);*C;?)2t3`hFDvUZZ(ph?Im1)xa|g!jtf|=KjC$aSG=C*Eg<%%& z%GGTxsg|T@{SJ}aBaagOmEpd}Bp-nR;k|e;ei2TL2KAz0xA7$-n9UR7S$tRK)mAe? z#E4Ecv*0f`qn5SlU$!6N&=*P8SD*X=sX@O7j=_CDr)7()3uRSliHn+iVS%>w#aP!# zB_AUW?0M(GHjfmuz~bu3+F6)&q9X|+4pk*{AxS#u|H>5j?`5n`n!nmObtjbw|E`gs z;qn6j!_Nu|yvnQLYuvf$X?*z-EmjTDtTDUto}LVDW%fbjnRbQeJulYT&W z_)l!alk6n(fIz*#17by->{4hLM%v`fdG6MJ;7*Uyy-QHLkoU@B>Gz@=F2wd zxXKT(yZ^+tb7U3~G}}g99w(RY1kEXj7eY6`O-=rsA0Ur@D?Xpn8}fNKVF4wY69^Y^PGeNzTxGXwo&DjiW_ z-48J4{OU0I*9X;8VcmDZxNo5!71n*Hf=Bh%fw3@DSVx6*f4ATu)m!&Z298i+-S@({ zZ_lH`y6;f7E*dNe#mMPEXwzh)0EWR9N>9MUYUv zb-xv@_782Rdh7l{yFE1slNyBi4>785A>EgCDy*Zzy8orJMD^BvKa~3#=KWw;m;USI zU(S2PQ&=3k4>=+4gYr8RHZG-!D>R zj~A|WH|CjL79T$7v32jjG*~di8@Wdta7W+@)NCk#mg^VtA-<6BGOC4qW0C^+Uqj=$ z>Www|BZ*RmJHYs3a|Di+jl3(8V3I5crc9Nf2l(JbZKn<+7?}6Ae8z;FJgnqvDxP zVV@zLAdhXrqe=9ZWMVzYQrkejELlqdmUf^w7U2|Nn^`*N>j$brXZbCOd$HR=o_>)6 z%)dnw4dA#JrC`x2t#}Gh4rY4c6`G87^Mb9pb&=m*Lid|9o$ngY41T1G`rq8!`+&Qa zwHx}IdWJ#_aTZGm_Ecc{hs1n5q>ns z<>U#Y&`zNbe^)tu3QYlL!0iRUPRtWDjurkz12ZU~0Bh@@ehUnOnv^6a&0mG>n~k7) zwdthdb@BhB7g7?#`xUu3N|hGvyEl3;f-YP0jt-lUGlHr5J8ibVm8(eI{!h+nht(w5 z;hNBVE$XAAl`q<)*yiU@Z}_aAjz>I*Msa;?$tJSmmV%o2S?7_~sun{E`{^}`J&Zw# z7H$L!GPYP`3((l%bEM$D8r(@bDtp&aq^L5EHILj(WCfgo^?pu3vA9_z5+7n@iadDt4n;gyo ze5SR%(f4s|WKT=72V&UFCNzWK$rh5I>7!($UKjC@q_RWcl+D^JZe>0%BN$BEE56n- zxaU%U)nim-)l_pzzfKBqbVYdwOzZ*s8oPz5CLPzb6JcjHc{|~-kM2;IF6K-*1&G|@ zBDec<^=QL!x%#&*y)?yygFpv^09o4;gl{9r~kSR8J$YXP-!Hro8K>*tjt?pI`51^4=9aKl(WE(q&n^K=)tUb?i_dUn8G#Ah{R$W+P2cek_5?TSLVhq_LO zYGw!v4hKwXa!&kPzS8W;mF*^vQytG=++yHzZRo(TMRt8Ua60YvFnKFH7{jtO?OE#P zz8;aaf1dkV-Z^y1Q!U}bGyc93Jf&JVRg*hsTLycZ<&~PG`I+ce-#W_@>%^hewpcwv zEbb-YAcI1Zz|0gcU7LW}N9MMe2HFS)^?T@bVxK>gWM>(ifEet)CGRcLt)}HcU+f|q zZc-^|krT*Q=ZP>GFWZAUVu(YAC+fWyc^Uo5v|f5Ti237{@jF1hz+Fh~uSOHlyJA}D z2gf^g^;#Bqwl@pAew2LtX|Cn7u)5^A@``#UTUa?d|7pdA$wuy!I;Y z9lapi74f|v=1`oF!F*)t2ZGWd>~>QAJnQA6(<=9AYjRr_;j7j>jwyoOk*J5&@nzfa zEPc2AIGeA2QXzGpXgEKk4Oq2Vl&7MQgDhPERvKADSsxS_g-*W=xd;eHHNZ>MNzJzC zDt_D%axhq?V1tMtbpA>pOHRKX1Y~6yM_#9V;`EyI)yJnoq^*zhGezqe`LFxA4d;JX z1p3$Zt$$(}GViC5L=?h7tJk%o7BD9nY%UHwySNBn6k$t8iF%jU0gf8%1c`N-1TMXf zs1Nso5KG%EwsRggl+fmok5;8SX&ha%1Q^lH!jQ$S3e!@j+YiQ=9(_T>X2d2up`$x& z@qFVKyH@lS+k_Db&i9-reLqMA{^tCs(zI=_V6ePdd zuazQjxy#GQh*GzU0~i*9!@Qa*{WEEY3gf+@WfmP4_jRT*SMNd=a=2#ehu?(+LGgmR?h82$CFjq z82t7H)XqVN&)qezl4Qzw6TUhenVQ13i|~9HO*mS5h5}qf6JpPd zZPDgu;J3sJFVGZADVcY=qUNmL7iD<1BGZKm2{t(Y zp$W`ZRJ6=Mg`-p%#vM^FWnnX19^%y>Db3}`)N6+?oX8#&v(?B(-`XH+8qv2@_$sP` zwO+g052pq`;y4DXi{21jav;(zk_J^jWO>^y$)deg3e%I!lPmq`L!wbE#4Gy1jO;N| zQUhp%3%;g)MFfO`ad~7>AN}qE8P0yY72d!)O!~*8t%n@#g>La@&?R2ouMgMKWa%tp zMa@^WDyB0_9thF5lb>vKH9Lp*udz~xV3m{`TP)VF#|V~qxf&H@nCr%t=EGwf2Br7U zMen@NoT$V{f83KttcX67c)2ylRVG9!>SNvWYByfxk$^XQc^xa=E z66@EZa#Sf<=~c&WqyBRRd+*P@bst~&Y;nka+AjG0AelRi4I?W z6Uj~}KhZJA_dL#>fpOyrD<@kD&9kRn8PML#MR)Ylts^{iV9%Mg4OoMp8a_14sy(kd z*?J(t^bTg%^9e=@pst$(cq6MkduuHjOAQCZ-jAhcU(8u`dA2_e@ZI4;=)TeY$_H}7 zVX%LDf8#>MvnPAKu04`ZC@1Gg1nABm+V61dq)#g2lCIb3*1YdF4cJ6qXFiLqm(5V(WG539FVL z>$JAw)dCOE2JM#c9!7IPeNb&sb_&2WRRIprmMSzp0l=;I;4(s_iDE;h>X+Uo;Slfc z5B-AhH0(G%UjrBTNciv}48htvG3`)(EwHn+>cjL+i+iinX4-(aVVxy&?=<(iN2HHN zkg2+X@l|fwXc09@_9?%~RcKefg55UoE>*Vqw|Hbi+-jv%_pGHIS0-NXQEyZbo0-QN zqDta}uD02{2+{r|ROeIpX+NMr(5OM)C2Hci#zTnqnU>}c9AZ;L_c;q|ivS2rM*=Vw z)uV*ek(hGk;oItac`hsGve#&h%T)aPJ_EiBBAyr*I)3`gf!^?&WgK>=6KE~O@#IJH zB;^X!tkM-5Sk%kq@r|sOcK-wOf$OXpxv15{zA-Qc9-k1ghv*n*xGwqOEs&-THcf_O7A?n6Ue&@NMrtEejB^7!4Uz+;~kg zqSP*Dv@o=R6vuPXJp1XRk*GO+#L?5glp*le>p6w#8n1n&n)5<-$?TEH_~_-*if=xP zU%lNDv3$h*pn-H>6DsA*C@Mrs&O>C~i?zBqh4t`XNk{BQd#}Po2laxWOmBO-ikdd9 zYZEV+t;T=>{In+L7oe%yT|)E*Ybu#mmM>R+iCtsLT3zqUd0NYC5|Qcc$p>W@t*vel zhR+0v@xu%c1F63{u4nPOWIf9p8d^#b(t zonp|411U^gORyQzMH%8QKk?FM+A1+o%*^#XLX5_kqX}!|{QQ1)doOuI?%3=pZrO(3 zB&n;(g7qDN!n^%d*ZgQv>z>JW=kh)I!{ih`h=xoVoO(^*A2!{xGpunoyJnwGd+~OO zW1Tw?O~sLfW5n~|RP7dEp_+gl<%Fl!Wr}z+T&WxiV-hWPJU0@(=QQkzV9_prc~A{| zj9@@bRiAh%V!~H_P1pJ=BaxMD(7ZpA%rdPKRP2OB8K3Wz|E$$qmF*&KE!9}3$6TyT z%ZtC<*QRDaWXbx=`F4e@H)mWhdo1^0%rk^R7Rus(`SCD zyUC^s{xlrAcAD;oBujmpJoWe1xjK9_@KywwIz_LtE^nuMbL?LE4&4I70R*HB$UwUz z%Q%(7a}C2vS)3fa=?IIWMI3Z*RwWE~s8`mpUhb~Ku%I@oyHD7bbX^}qUBz*|3>`u> zgh$#tw5RkM9Ua-*y5_> z)}GM&t7C>vDx#VFwl}8|{F(+YblSK^4-l*NeY#`X5qXzQ(|+vf+>&9)dPLH|=M&W? zf!SU;N$u>go_iRfkdgJMPt4&0_BT^l!zwHnjMNYm;5-9*N!C1g777+=SEc~dO>j~| zXKajFzUy`EdQ%!f7z)7H zAC8!???&LPlqdk#S+9tkO#AIL!zVhVv5!ynh!o#XKUa4D!ZC5EZl(M4K0^PY!n>E0 zWl~SWUiL|rl2RQ-Fhe5lZw=!eFY;(jM(blGie4tS52=RPJqkj7kayOpJy03)H&w3u5^6@_aPZ%GU literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-4-Q1 FY26 vs Q1 FY27.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-4-Q1 FY26 vs Q1 FY27.jpg new file mode 100644 index 0000000000000000000000000000000000000000..d65907d552ea2155c29bcf275d0ea70a79a6717d GIT binary patch literal 37490 zcmeHw2|Sc*-~TPkD9fZQ$ufnaBxTPwN=QN~l(kX`NwPBxDtjggSw|#A_U!u}LiUi1 zT_(&}2eZ6)&-uZea8z&~ z-w}RpZXxMIM~)qrkdS~0o=}hxlRqjhA@;o!s_omiGcqvlVPe`Nwx4^y*zbN&DgYLG z>S5Yr)Kq-HHWn&s7Ai^&00YNKOZB%0@b?GRHfkDLItcyt9SqJgWq#Yj($3z&(aG7x^})kOKE8hb z0Z&7pg@s2vkBmx4e3|qr`SqKW?3{PGdHDr}MOD=`wRQFH8yed?I=i}iKK1qu4UdeD zjZaK|ox;z}FDx!C6INEg(?tbP|0h}C^M4Yyfi4!1u5C0l)HIOqbWv?{0vk084ekEp zbh}RKK(5*D<`eUxXFV12BCB#czxV|l+x7cxJJ>8ZmrD#ar5fFeeZI1?MA)Q5EP*#F zju#Pv_|NR)PPnciCGm%cn<)UoTl=fg!;ekY;??rntgju5WW#PMKIDA{Gzu%Le#|x9 z`6W<+uga|{=*G)KG0+RHR5O-KrI?IegzEwkSK4N*`RpErdERJ!H6~!*=XvaQ!aI8s zM8C^Axv*oQxT!#-?$%R!&Td8V`JKmP;u)ue5Wd`o7@p zH|8&^?VyO7mQ=9J!)r8_ zBXN-BOY<{UjUG(Sk_heLvC@*0q==;9;roLQ&)vJ=eho`(X#B-G>k&Aok+w9EnJ=HnGbBd=I0?an>RA`0V>`~OHL;qd!sZ)l&93wrnSN2Wu-L6> zEB-K8g7ZkdC#{=Qv6BtXtrCuV{53m~ka{Ud@x^`LXnX#d2mJ*w|Alzyset*DsG*)f zgxt_5U8_p=(}8h_;+d{Q9vePLzap3o_deN-K#2)yZO=VE2_FuDp2o%#anw{=Ud&o;bTZ7X-G z2({#OUTH>%x;c$3>AcF&cCT+DMm&@?9RKawSI3jn4j+b^A;HsT9mh3tu0{zXN z&M0KuXUX2B8e(XJp^s-aJza{f9L%XHYI#Bd%sX?12M^6$oK2Y#uAHulalO$=0k-E3 zXxBxl`4gpwxq26c^|Hk3dbDoX-;FswkxswBmt6u)oiq_7SUX^NY%T?!Ex1Bl6Mpla zB{T-}RH?x9+>|%C%yT!~0Kj(~x^5SNF zGl<$5{kK=`-DU*c3%>;ppfb68l8NegTHgmHCpfCnj16;j;weTMU&2}ss4pszG8L?> z)H`kvE)EG@%gyIdl*7}r^jF2Rduh+rUz~D(WS32Z;MuBApiSDq({UO!&+K_$Gk4#% zXJI)W%Ab%nw%V}=5l zTpK?lBXduk0@MSr$Ig}z$z&jqmf1&KB)?{Zn(R?_FG zpn-XnH6#kfO7|uEA8qfOlF-dDe;mU&;X-wCVh`uCWXF^vj5hbN)`(6y{#K%7S9Qcw zRSPlj9C^B!&hwD2IED^IUPEW;ntPLbbDQfzkKC}_RcPIoULA9jYOU~_)kHb3f%AnX zqhL|U7fq++6`&+L6fG{e@1}m9oZf_2iCYrEyn=kNP1Jf;ps_qtA+eTs=PpYFp4t!WwN>X3 zr8sd%M*993^Dpnn2c4s_G?%5x;we2A*s7R$0%~vcpOq*PLA@ww>D;4Yg^5!w&;5C^7AELrblc?cy@<^iYwZ#w8L?Y_T zqgLJtX%%}k&XW*L_Cl}l!pVnG1SNzf1yE`wqF@w&O|9qH;QSoimDzduM3L}I3h)q8 zGq}J=0p8;&0RGbYlPDJYPylqu+DE(k!sa$(LE=8egA^d%oUDeT00D{Qov_`=kKz=7 zzn%ir>im5W58@PB?go5aH$d3B5v5-%pBVL6;Kt9`+3SN`g&yi(KFLV|LPjXS@av63 za<^lm0C4{0_z9FK-XvP_W4h1-lTr;BpIX%%mFA+i%;CVSIivsjwn?aoVd18*(7HjGi7ox(dUlVif zH!4!USLuX!5`4+3*81dfkOzKjzSSM!haA1tg~ZO;kQlqIJ$%&&whL;mpM(+uIyY+! zzjw@=`hAlSTq5~81y};Pzj=oAhINXj0EMk%D~Yn@)Czn7LK*%tRhwkdfl}e2#Gt+m zk_RZ~UPLvgiS7i&S#t^n=ouYc<$T*mVn*RcCW99~#)GDaj0*um31tE!n`ID_ht$C5#4XDgSER{4R#NXultCR%DwP75h9W0#$HVo{*8Y4b-M8u zhS?m1_Cq*koqD{5Vg4#n3+wdbEe!MLSxb?-)2|2-TE(%D)x1*= z6eDS~R?b5L+VmG=xz2x-yt2@&)jK_dxg#q*d2$Atc84HdAv+sv!^?A!W@jx*cHAsa zX8u#mF6~U3j}TE6Hy2V&`upRe0dJW;!KM(QJ=kJ+l1oG&ktK~FsAH&k{j<=M-K=#nR89O1(M_ z>34tI{vp3u+kfPB%R;}wLeGXg-4UFT{>dWc-raP5U&ZL>Q0yh)R|H3KJXduum_I#P zzHk^=$OzuMI1oD%i!p{a`_z5um47cXAE`cc$*U6(N4d^nHQ=lSy&0bCBJhhHqh1f* z&e{lAEeLqhD~8R&cb6cq;g)Mc6;Vn3Mugy10hgKdlYDzB`cjTRXxUz;_NLi(6*kS1 zmL4BM@UE)EYCoBdcpsFQ68G}P6OV^>SS8+@!)Ugtvg2JX$*E0Tt?&5!M(9%##j6#PP-3R%!Np!LRZY=Inj8=BRbE%#)t<86!3Ml2 zhW(YK^2Ah>)cBSD@aqrmU3WkG6q<%0nP73IWOY=q_%h-Z1!&V$;$<^7t(GAU`g`yi zA)BUPP3W&iN%o3#Jt7F{5o6)YrRuzW-PUtQuCmwBV_<23J$jF%MkgUDO^s=+=5}xZ z+kvZ)nVw9^YNVB=>B7!p3Q*ommYgmsuS3xqVmYK;&AqR8n#V3{xHx1>KN;cV)R!6Q z|Ej-WZ8kJm+wIOiTb(u8SN-Cc`i|GvmB<|DrCEBc7|%mO5R#`02$4gJ=<$oL4`(I> zUz4nqN4V!T@%jeyk*M%V|F}jL+kmRBJCe|Cgmx>PgQE6LPeA?xp<%Omu}AW&^cPuz z*)Z160nIuo8{=BV=ilCz!#(DT+)29UcH<+phxl$=OgYs+>%4J;*?iVo{bC-wL zR5Jf*3FA~T5tZa_{)#Z(Ju=J@yvUHs1Dvs5XyURdYh;F)HSAkhkoOd#u36+ch1Ko6 zD>{42B0ZhELEv$he@fR>--M^U$XkN!d<~hpQ}jS^1qFC@#TXT<-+52YZcu){)rtXO4T9~MsaT2L##^}ZqRl`7*mR-poSRogYmWq#*+^||S!{x{ zVHH-a!NH8~tL-ORdbm8SF;`wUSradgN%u{)pug+DByy^G+x>|hY!$pEboI6-&^MM& zMRkc%FJ6CcuyYc>`}7L4I(#D6njlvnHS0BN;?6gOcw;+Zcurd7pn}>SMa%;w?y3Hk zzDKVPsP2{rj*VFG((t>_u?kTHd4()-*fQrjyvE#5ivi9zMlLJvUK!;;W&n%`{7f z^6gB^K2`J^wmtJF9!Gx;R`u}I)!s>T#~;nOx@6#xmL9r0z1vx=ooaFq4o-!31u6)Y zuMMY6StTRWV&rdQS#C~Vxpathk}Vbkqa7fO)MQzFyeSg89C2~PFL9O-=UwQnar~CR z)BEbP#&-kpyY+F)k3=Py4)@jdii%Hs;e+>aDdgX+7zvIgxx_Q0hRYv|94$dXobB!y z(1rQ6hC9cp%+5gI7el`ql!@pSl442(MG8b%Ib@z^sK86O$?%qvTg5iC#B-fQ*iFUf z)*=oHe3+T1jnhsC-9o;`VZ4Q?GYE=lkZKCRJJ7PDFIpiw`h_*BehMy&PnjiN!5f~( z^ZH51F^Scyv~T~s)UA#>wph83Drnx5_TJ`)GXA_Sy;3eRn?6YuIj0ZqzmWK=_wmpZ z30nuZCIQI#cU=iZt-A@rWR_+rxz=guJDcY!ISsu*J||j6NPVM|f7tNyvl?0;ZO=A9 zoL}o?fAqbr>F)ATgFJ8x2ylOjkGv%brKZD9#6Z2lMMC*1bNo#Q`F%gHFBzKn zOLoI%ybhNMBH!m)gS$k4Mv!;~7Zum?n%Nchv~nySxd6#nZ-n|GclJ-~YKE!)lK!v> zuT-@NEpXcv%YJe>xWk2#>s-Jh(tmJH#1PptOaUgLCVxU|;m_%cRt(x`oc|L)QwI=Y z!EG74@<|?|WY5JhwkZm*fpz|IYu`_4&2#76M3(`C4KGJlNwrxenv$rnjlx~z4P5gN zTTOmitH`R@CdNcD3$If{^izo&nC2heT;?auIHX#*xrqT&-1_TukPQ9xO!JQ$27XrH zPd(ZABkrN!_#2sw^%{st#yYO~$F~OF0&jm5sQshgZ-KYX+f;H3y!|PAQ2y)poZJF$ ze?}PYzkE;1E%5e-k;r!u@vjav-U4s`z$g5B&3*rdJuSDu+wTaz{a>tre{QNJm7V5Z zivpWz)#A%iJg>TRqaD?ZYZ%SoNtSnTNaAu2eD(vJsB5=y9&$3Lk^;;R8iBDi$G%sI z_}PwUP=KmdTq|+Ka_|{gZUctPA%aye6bdZOx>Zgmt_)b*t!`!|U zK8}M@fVkBO(uyxwOeWnU0GO0=hLhc}=;d)3`S8}^pKDr*pp4FmRdJka`p=@ka#Xi_ zV;gqX4Gc~rb4`#c=6cVBPgYre^?%tF`3;VHS0ee}>{pE_BFcpPh3)F2bc~R_m*JyC z)0L2KFFq~72nHs5!6;l2!#UL7iWad)_>@mH=@ApbVneC*uM^jVAw)Fj(LA zG=fE)N?Zfr*_p@&Lis%OCUbozD4%}S6>R+knJB2h{yrnV*}i6QwSuIkt21-QrLLjQv&N)K2~vUC5qQShEJ4*g0+Fc8>$ABn~eAo?Uh-u zz92>!L_e%lqn-NKhFX1ATLUDGFP`g|IYK)g_3dfxkq=3kwd$r3AtBS9`iC5>CBr(q z^I6{6RgIru@!63n!{^DD5T+Yj#TR}{24GfCPpw5lb5V4cFArGl>P`F-RDbtr_%p^E zGkYrPl(5~NC(eGS(~GDw!Wk=Jd}88Ei$J8`nU1OYpfzTztO636Xv#@{hpBJ%?1&k&w%V$2sR-6|VS_=IBXQPw?7aCCZrI5Xm%!7s1MY3_&wgMHPAb#!YApq9?e+CRwaT|>(ECL;`ofw-UXSPV1 z5`G*7Fu=d$$lq(5=hN44GRp1Ej(sCC?;$aaJ#tLXmi4hW&AE)VX251H+6%7h|+KqK2^;p7l?|+_c04c<4vnGb~{Vz8JlVg->J9!Tvc_ZVL`wi zZgd_ZtYX}5pW1Avo8KWpc$n`bnk{gI!*=y4kls&bMb^MpsglWbCqe1*5fZ0k0&4a< z;KNW7&_MyH_kj`x8-j=WESm0^NWkG; z5^?+S2V-BJzSm>-*08*2`zV?EU}?ng^5cWJASn-8+`Grg{r!e-3Z7Eu;#`IJLqZx9 zv^;Vv5uCa<0ewd?G8Xi!E7qEXx(|bFxy`-9r>Vb=r2)29PWL0=7KZ+Mh>2!qW%C{d zt1o5jgcoG~+p=gc;>GNLE$^CVZ(3dd`Bt@lUoU40IFRjtH} zeDg^o$ui#+F|wT3AuqakL2V&5N;JS>koibc;#7pxsq2j19R0;S;ww>|gs7DA-SVe6 zlKV{x&jQOR)NXZKOer93eN{bq>P)$XkRmo@E^DDxc}4+Sm)fVu=p4|Y-!V((#U{Il#|K$U z!-_sN2`!Lu%aCqQam5`3RVHP<0=!l_o5bTioEqxO2Is}#wA>_HqX-ny6#f{jah`JK ziuL?hzbevue+OPcyD&qQkde|$5Ut9CGfNy|W~0fU3+9}E4;esxaL*L3xlP>xsl}Gc z4UM6EN)&F{NBd>xbJ!}Ldr){WD0O!yS+&p_USD>kL$rNvutqi{%+Pk%=aPUgAKsqf z$6hHq_#lOV(lKP>tFtN&pSWQbzbF_F7FCZ{A*c^g0Gj^VbfR*%h?jEd5ti(?)iJzt zu$z09`8@dv%dGljUfiqlBMfY$&vG|`tC)m5_%`hcO~OT>jc^B$!w6iXhm+U-6P_ty4f46_`SNe*TJu2Momi@=BWg(b9K zi=4T8ikov=;~A3^qZr#}z}YczWR~4BV%EiOnCbMH<|ljw&Rq&&mpmcudAZO)+3^gF z8m=ytm17MJw=Qft%X&2k`3{Qb##v>dMp`10;Jp&2)V<2df{iIR=szf&p;v;(T3AI1 zKJpMIurJrdadmWPb$??m!SOvhqST;A?eo5cD}$A4p0tS(HUDgTRo-dbtM;Z!B(tQA zpLFJt{2LN_Ud4c;r-HOJ)X-%zwD&t#O z?-O)oj|g1DX-+j4oF7R()*D~EmZ!WYa@ktY(RM|@WpM_cgI)#0DM4@kTV(Sjb@gk| zsUkdx&~L2tA@}<6y0E;gaF>shO|m$4<|SKT_NK5WIqQzL^}ajiK0KXkEFq?v@8$U(kl6_nBx*n0i3qTM{2;PYiCeJ7_*DN;eKi&a zScswW2h_%=#04)vP1h#zMS}H+q4qXxhozJ3nKGw`^<0p?kD;G$*v0ww6;2}! zOYpqKE~$fx*uY4J{Dj~}19{cK?sv9za1$dcmk6A`ecjhIjvC(cB9X*5X_r zBB+;x%3jDp9qUJoMg57^L#9-OQxlbBI%KY~(Y6Tvp?s(vIP7@Z}ax?bqojGP~1bFzDI5vwU{?zJt}dj&uiit$wYXpawfTzS_OPqcP_p z#YC=ddF{8&ubsgy3-*_}Lp2{~=bUvsLuhDyt<%~W04_DRbArtt=c2XC&)C#W?+78i zk5{3Z77vzc45uerV{E7`&`^m;NEF`RX4Hg%aRR*@EjUT*eDhE7xc;NSoZs@ly99AV zUT%DNXlfav*y@qRaZ+`)$HShNL>C;Wr}rNtlxr*7zA~+fyOA}=Qc)q;>vPPi&Z=FW z2_LA33-fHZu{P^^7ql~PUTW_#*GG4F75w0jfLjRh&?>bWZS&!NZLz*Nej4$dp{sUv z!|+f?vPm@t?qfdG8sLz@V(DZn6>4#3@J7KAvG*?5tFr*JL3w>KMPZ_p+S)lJbNt%* z72j2PD!aw{g){~DaG5Wjmr1#bER-9lnpK1%vD-f%+A&9cyrPwpo2t)KUfQ&a%Mj0; zm~yPLxWFaTjqbVGGEdMxl(qAdE^pGym>{}?r|c_sdF)%-E+q}ac|XPm57rD z6|*4*ym4OPl4zdGVeIHinTgXOm>Z<0-Sa5oA_aJg9)_*1!QboeHy|JCfe#(PtG-Bc zkN!4z=ULO}>|jp6BOUVa`}l3eq7Ny++@%sYm{0*D*Ma8P9YBummu+btLus!lG2hNO~3Q;HrxO7 zkZx0u8MkV=K`l-{Skh)2mTv9bP5URbabM1T3v-9Q;6Yge<9_>|sM2S+5O}Pl9%E)% z2|lEAvUZhzy##31O_?-2%yNI;hjuCl7BC~f6gob>| z5-h-?hk1#Y&^);en!+l$S?{-_%y;;%9IU@{w6nEv1vPd-v_6(gS@%KJ7lyp-M=_l0 z{k%hQ9+!A*7IJ#!;lw6Fryf!1LGU#x25I28INSW2`Y&*|8GmV68Q_I+M4betghwAd-JAe0CZX@sVWh5j?6 zzUo)#eOjdYqEBM-6!EjILq?KrPDIf(4zhojFPSEp0&u=+z>lap$3m)$fy*rDcge67 zI!}46Qnw|ywxJSy*m$a}kFiu|ws}rg9dcjS=-vL<=ZrT(QD>*gJP7cHeh68+NIqa{ zvxd-s5fLti$KyP(<{7!bRp)%e=o%B@&c!RA!V6*p9uS3?Sv`OqwMn-=e;W|9Q~y&! Z=Kh(+w{`4K$d39qY3u)2*A7PM{SOR>2i^bx literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-5-Q1 FY26 vs Q1 FY27.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-5-Q1 FY26 vs Q1 FY27.jpg new file mode 100644 index 0000000000000000000000000000000000000000..1e9d1778e62adf6058e8d3685e30c819981d1143 GIT binary patch literal 36832 zcmeHw2Ut_ty7tCUq(%@F5r|TR5fr8OfCvbT0v3V|6r>AC?*vgnsUiY`N+>EILPUB; zdhfj_fFL!Y1_;UBo^#KC2hYrzx%24!b03M%!aQVWvC7`x`@P@#3b~g&3hdET(@+B_ zC;)&0{0EQ+02N?2H8l-2)ovOZ8d}=jbPTMF4D|F22Uzw(S-Ci1+*}--oQL?13moDR z=H=uRl=?~d#7S{+ahQONg0!goaWQexFP%`(($X@}Gq5uVr|t^{DGthsKfu9%L_qMT zqv8^hQqnTADyLP|)X!*WUbuKk521hgit#NIQ!{f5O9w|MXBSsD_j~sr`1wD46!0Yc zX+&hyv*?(F#89=e3KH5<*GyWnL7!T)+llrldN2lA1+X zm*%G3ULMiAyID_##b7aEqua+Pj^W3Dnc6>Q_QxjX`=@61|0ec7^Xdl} zAQa%{L6`wJuucjPjNA1?v|*f6~b?gND;3NFlgfX&K$TB^HcD?qQzQ zi;gQbhO0{G<9IW_c~5?<$I!md_lh$yw;&XqL8;z{0)VQ7!WO?$XjC;KN-E~Xf>~T| zyhrNm!{_W?zCFXMP&ccXU2m1BaozH>BRq=?n5IVDP1hF>qQP?lGBZIs)K09rN>n@T z53T{i7MD~Gspgi>&VDIoY)uMtn!2P-Pj~@+V-9Kal;H3@LQr(@G0rOw?rpVIy1MeM2>*g3bnKevD>vgT989C ziqyn-)2OxJul{^gESukWRz8y?VOXb!WmDTY-t|k4 zgaxceT1LNlfppiKU58!RG7sp-Qd7^DK3h%qKVOWzKfEWr2{orEhlef*M2CFx;a*>D z;ZB#n&mm6bAc5( zvL(45H2t>67oyziW`ca(WX3{X*1Z#XD68m?CdzksA`>jaEbO(U!rZKO?^5XFJp@qO zD^M%PL3tAb1c9CQ$JPMtNZ20;qaw6idT4|dyp#4hN9yKA&x5X+W43x7)@IomoeS31 z^VCNI=>1QJ2YHrpa!Y^m`AKY_QKP*^-I0i0Q984iG!7gxgM2=l;G^JwnZPCS$vwb3-?w-V0kTF zL=~@Ai0LtHTd^uURwMkr{#3@Afj^!N?S+R|_XihV9dv!N*>KaB_vQiK={OExUQy1< z-JC<oCvJS*zE&HJ?iVaHQVr8z^ij9jTsviN@BQ-_T0s!fSwcg2ewOyFtJsyW^BsEU z51UC-)z6u<7^$6{7veP_>>pE*x)8}FTV|=}5gH+XK4HScM(@zQ4UeAaWf``vUeBNL z>Cv}yx=VOVm<7w<1@VM;S6FY`I~83AR@ zG@nP^(-S4psl%31u$lBXZm8)R3B^n-mvN=kE-%CL1;t_BpJPsa^u8BI6cY7Oc#g^S zpOrs=$@k`UjZtJi(?0in;ml+r#k7Yz14?s=(WMeG%!}vT=uQ-fwiCaL@4M)zJ4?f6 zuz9U(Ma$KCg;8%~o*mw$Zg*!<*wc7|^0D;tTULsB6uZ3KTi2FNNaZkB2I|6x*eYY& z5Ak7-$77Wy^SLW8+3vfdg@O8TApEnH4tLqw5XnI16Dgi~VTa{)J`sf8FSKnB>4I7p)DJlY4k&D# z3}BN2Vx5>bL znhN+b*nRCH5`S^~m~!8bG>GIQpLw{(3aEORZJ$r#*Q44h_9I>PY9Lk&RN`fY2DZ;= z+mQaX5ZuE8kJ=3SzMjt4W3s(j8fj=x22QNve}9=1H$51MA2y@b{dZRRpSRo2Dt~#t zzTLfS2jAn!D5gScCN<|Q$}dPYZjDLWWx`xh7p`J5-Pr@)KHXcmU~4cId`aMzow%;5 zyI;I2s}98v(5R#QOUF&yOJ^2gTMUR&gb%@kq-HHdDm@HJuPTof&k`i@j5}x7nzAY@ zqUG6lA#!e|Rmk=G$N}4}C&fHPBiK@6IeadcRA&&WOInAb#+!cfG+%NTOepD{r4vl> z`4O6UkphSuk`ZLU@G%*fGAiCEqNx~J6tgf~S%;GRgho;)mOvO$Y5jyQ0N10$vrI4{ z0QlBTsEH&*1{!%u1IXEk&upJSgd_u>n8|>3>8j@PhtRcJPcm@CLTgO@d7CQRYRwH@ zRaW2!XcVt83+o56&>KX5GN8h?&^4;a4FdafO)`M4MJ}H-*fJDBkpUKNe+jQvT~)TF zPv1q97}(+sNjG7u17zS9`V3l%fCd3Kbkmv)d_cT@iY!r?To=MtdnaN^b?3|<#=p3wGbSy$C7qnA}z75EXFh3j@~afNQjo`y-# zPURopWatR4hUdD^_mCur=R^bui>Y}5KiJ#V6xnd0~cnp@w*6x9D?++I$zy z<{Qwj9cuFjJx~66FP$A~^UW=`-;yTnP@BK)3D6F;*`YSSDLb~?G{3go?3~){oZ9>i zDb5bH`4%+fd-SnGZFZ>5e-aowr#8QGYP3Ubep5a9mK5dB_O(N8cBsvNuy_2@U1P@~ z?H^QY=hSBB)aG}c*tYp>>^P+Tx7M^nZN3LB`4)ZdP@5fUvqNoug}(eJ=eiwgvqNou zCuDfXA?;V`)y}ESubRye)2()6UORb z=9@YzY{YHNwnM$FcT`SRkGhAvLMkM`BSEg@eTZD#n5e0@>22}#^46glKdC>$tjKzK z%AM0&0GGYqDK2xuP?)nCN^uWblt0$#(Mk!UmS;4<23yWt=w`mE(cY@T10M=&KEp_= z1;Pn>^QD{0(~qa2R+wPf@&PTM+51h*rcRs?A0^$n(CRc#sUsbuG(pD`)=i902*YQd zgic$5DKDl*7znAFgn`%B2@aJ7PT3_HtZG^w4{1p9(RJsUL`9X^mR8f=v?NZi#aOpi zu4amBuUrV2Rb)Je^VDl|W|1mP^zaS-nNY=2fk#~1eS+h(@rvN9RvJdb)2~S{afX=4 zFr@Jt6_@mBe5!bR$b<@KSaaLsT|uTL&gfU3$6i(8JuufJU%u-fzI^tLk({;a#oOT! zn~SjG8cO$&EQ;zrQk#Kr0N!A}K`V9yF0Zxsa80I-G*LMf&LMt`V?n7+-Zrs6ExGH2 zK=4IDqlUbbkl?{R+U{K)WWY>Lpp@PYJ08m>W1cowUR%+@-wfx~AK`McIA*AcdypQj z5E=ERnnU2Jog#JX#WlcN=7^8@k*-O?b)17zek%*RllATO%wciMS!$^pw~T$14veMk znO-955~R?BGoKmya~^v39`BY=<30HP4lurpN&53?S#ts!7k;{Z*`v^;M_lZK)@781 zWA$KYGMz(5a?(pK(thI6ZgFc(ENzi8I?s<@$a%O7_7o{L_k;>?gFYydKw?MsT(#hE zH@*D)lf9yTtCR68l{>r=3Hz!5En!boa3*fj?o!hGoA8EC)9N;ei01+JGlHO-Kln5sD9mr%A@AdLQ?qZt=8b@xlawGZ6r5-x_NIdj}GU) zOa}Z=%#Ka_wF+^$0cmG@{1<5X9t_?z?g`10=a!+or9<-!QAj-YM5AQ4YyH9YvrjEV zB1+>F_0GXLVQHs5p*T|u@7`GPm&YCrts=7o54qoXvoLen8%_v|7kLo%(VHgM+%c?j zjY@MpQKqb=(cX=3(HvrhIm%@DcqW0Rz7!+mWM;Jk=*I$cXmCI4RHurM~e_1lJ?r} zy*Jv(C{%mSS9i1V<;aYpt+wTInm`SS4nKK6O&kid%_`2)$uLNkmS$&Qhh&W<^U|yX zuD8XstVPdKqFIw&gc8|I6yri7=1JmtQQ*UxnG=>iUD4PX7qP3c;uG>kJp3VmdaV~& zR!aTu)_#PdMC3H~@_Y$k1*izxmkt4am7DmZZC|^#!pT19B z!41rNgs;U*yAFlbs9o?C>htC_*NbIqH6uh|g2kZ>YA&WqH)ox@S$m|1)r~xu zD)4bkOBls`C#(o7p|q<15zF)m{m6Ur5AP11LdQ`krHf&51G78uDq4}&=jUy6hIY^I z)3_X|G|1^|SEKtm0F6tn&WUBjvyy@Su*}cy_m2!un%61!WsGo6_V!pLOpyTw&jab` z07SL1Zb)ymwJ>ks+TD!BdxK*N<(sWE6nMRQMmW~H*w*GQ&!*CyxI6SmGqyC+WwVJF z((ACcd#7bPQzo(ptIEyf5DB4WfDq5GtAgJ!ynDxDy<@TdJ6f!Z4Kbm_Km>e+FtV`= z@f49tJPF2*TTYTn4}ytkG=Vwu?-0rVuY=O{=Ko~x`h%bSm4HM{u_Y%dq; zBC&&6%Ca56G&aFtDmH#M)ot4u$JG4?A?MHJ8-Tu;`UXh)qq+(ipm8-M)l}35(RPbM zsQ%x!5^&D?27vTir?{Sz-vCR0M0@)R(ht6P>>3&98ADFd8t87@MnOpSH$b9aqS7zl zLndXv0f_#@C1ga7dR{~vBr+xWl^t^2e3^j{qzvYiO*(7Hcq*x9a?d@VfxL}>i^^q~r8 z=gAjNRs5_Jt+#PV;#v=Up%G5JbWk2f1{79WHkY7a+T?3nKg!2WvFv!{J)**!I*Z(E z;hxpk)=#VfNMwIea{#bvOSoLJhYZAWOm3}%sgpy%vIz82<*=m+&|pz|0!(9BkBv2`VI4E7aO27)eH*C_J(o*}~{RB36)r8lh_QIH)NZc<>wpUB{#&N1h=G6EIw+ zCz!EZUgDqY&51q$)>nB>+Gl$oKIsRZ*m&Or3xhpHmN()2iMW66kbRNq&WLu_=>Ot< zB+D2J3YFN6e?CP4iDvi6z+}mP-{#mq#@-W={4Bm?E`xE|uV<;YVc@sjHbTazJ;%AStYgT2wpS`NKPjD@Qe$urxw9{#0J z%@qROt4x z`$;!v<9k?#FQRukUhrv?9KMHHUOaf1l< z%kIGHJ=<2B6=`yV420v6pSbI}LP5~l)*Jkni(lj(0RQkFSw|>cg6=H!&QgEJyE-)% z4-b(432XadJ>}iURWCw+80Y)|sb^DSa#`=`2cpD*lyXU{PRkb;3e=S|K>@Hd@-Kz% z;xBb#hh@Y~v;Ta74$c4Z&#A{}Fou~8t8UrhiWL)&cpe^~wR<6H7gO;kDa*e+w|4&j z#`SGdHxvU?T<@I^)oy<`QBV**U&B5(t*xZc!Zw`)&y0Fxmdj#b8DpIqh7#WYIMhf9 zve*Q;4zW#<1PT1u5r)1xcbBu}d)==$lH>+mqvD^@^WJqvAfe5>+kUmZ^i7O;M%r+7(Ds$MVb}atmA$y^h;5cW*wf$qd z^D$N{>!nSqTsBc5*>?Y-`QoN{2+6;Fh?XBfX=oG%~7e zgRW*6fk5H~)2JM0TB_$kuh#6{Zq_7oW^fa#fIG{G_?c2Fd8_!#3S+$$ZVbi_&t+Ub zgkFEO3kg`YeE0Z<-!|A4)088ri3n`N>EZzP+qYjG&_J`N)&IADX7$I!~+-fT3<=N#eqk65Y5tCMn?CvR?%=2I+N;)Al zjto>l_q!%)oYg+lUb$-1n0h!bSXB8V<{3vde!t&}A>Fk9*>39x#l6Z$KJt{-yK|=S zW&^kWaYB3D(3iwLqbToD)y~~^HtvEzKcCQAX3!0o%5@i_SWPf|vLu`g+yxmzr4CKF zt^q2=lLqe`@p{N||6U8zgXS?7yUtb4>4t}fquE+4P)ymyW*>VLk93S@=4Is!O`duJ zlaX9mcOFYpaMz*XRb@k}wxKgPhtQ=|CYspexdx15-4E|EU}tR ze~jx^c7Bl7-mSN6gRgFJ?yF5T#3hmrfex9BaGWTVblgXNFD*D~vKoS$i#%7w-OJsmfQ+wgX0*~gymEa|-$2F1LQKVu-aq8D^wElV zw&nfh3(PZ$JOr098Hr#`>?q-6A(HT98Nwxwc^q>qN{M~0=zK)L`KP*nh2uY;CV z8utyzm~&FHMah2L-Hej_Y!QmV8xi3ywJ-0up;4PXT?Vo^ls{1yQ_7fLj6EZM$#;-F z=!06UjjluRxMSJp86_CV7Rq(8&<&<(Bw@b*p-Zj_tg)pN##daNpYO=77o%sN*c)Eu zahRZio85)PnDin?&z6#b{RYSE!)lzw@01WCdU?r!E97Gw)IHOYP=^bi9Y0)jHi01? z?*20ql(?G#w)()e#*ZlDfi-05H;+tKtz4WnQr~#5xCD!9GIxbkEfIvOatdCV1m`PW z?=&W06_CuG9swJWYle)Tav!?A4}C7=FX}c{Hhp%`C9fILi+Z*t2x;*abq(k^)A3G` z+x+Z@25nU-6YUZ2cy`=~y8vOjF5mP?Ut-Yt({ zb8%8w8Sils70+z^q`?)UbyDU?;%N&mUDnKu^0z@xyaXX}o$?;orgQ7_^YSVwZtn(s zLMGY}TV3Y?nC2>+b_OvlM>vc%mp}b!?wv02d9<2-#Mx<$d1Ws5rE5|5Nz_&xFCe;j z{_-hzqu{wuVUuhd+U;@A9^;WVXpix^5wQiF35KI11yhYk`R%@3U{TLmzr z{jN4JR-#fz*5=a@%U3P3DD3G2HP}0Fj%jaU&$Hd>MHgHWrZ(=ca$LW+_40gIqp6w! z`;$EB7BGq^YSl%@o9EizPJNBg&CgcxJ)X80m|wx%Tu3N};1qnTanWGp+KGd=jP&2W z*jsIP_wI@~#pEzEArp(F8;Lg97(1VBayZRa!0g(XFKyj5*tzUl%l>o*o>wOs8i%n7M4yte%WU2@N658gs(2Qv=T$X&<|L^`Pm#gc-F27O9wFt!dP9 zy5bEeof<`$3R~+$PnDjNDs=k=M2y8THdLZaa(NY9@>K>fdbS8G9Y4)35j zsK*$GtR5>jIJ&O~v`?_SRNpg5y_xJTL^|+s^+<}RGAMoIkKb0iht+p_gK~?ie zTG)j-Nqh?rdaYR*9zu~m`7^G^m&8ETRiym(coLTN=a1HV4xCcQCFzoujC<+uhWiLH z3sofk!frGxZ}W(-r9H25dGZDwkZXa5R;}*4(|^U)k95#ghtKX+9$jJ_TdS*f6>^U! z0&8`Qqo=Yza^?}>vZP&mFVcAV09pp^LU6F4sR;Fp2@RECuNhdlQ2#!7BLeP1+K@+? z#L=N(^4A0kvv3by%?V3V+$sor?jWC6M!Df*@70p+`cm7?O)=g>VWQJd>(RLAnYpfK zFT!rEGv&V5C&Aq{*!#@$Ck>d*Z)SI5 zNUb?f^m(4T;i!8Xlj*U;|MQyq;4%*Cwxr6S0drK zr^+b%k>Z*ro>%ARL51%G%~BJf?SV^F$G@Hr#)?qZvARoiCg;_ zOZekKZ|vMo&+8Viu1K3_CClgbyRt~q!hGnj=Ag2uqh1g7_f3&ld@IyhyoqX;3@xTF zT?C-s8ZFswYgGrKOjliB7`R$p3|Z{nQf$ZnBz(hkl5%rWT0Uf%qN1n36RZ&W#Cj}r z1B+a~(DR_jX%zi%1m{n%y4D7kr7Y>$N`BxulI7-&s71W#h|QlCggp4-lDc3pgI^%2 z+ERm#t6&3zYmOF{auY#g-u*Gnws)l;N~*SE9!PB^HZF<2m9=KyEZH)L-2d`Tv@;=* z4DgA3M4mw|YoKY<7pmaQWB|jDJ06HZ&7w1rreMYFGn;B;K+KYDvrtWTG9`|Qr5H6T z>3@SM6Fa2FAL00n4A`@qSy>A=DwMIquHXZfk^XSD3uUsZB-D}n1y8Qm^BMIh>O*W! zs8u%sH(eXBOvPhYKl62p@Lg3HuQEGmBv@l3hjM7L&z>8>m1G%i2uQ*REGVMs+iAmt z%N|?D=bPjyj3F)b)%;iQ^V9E=LZ)zu%_xE1@f0g(92Xu%;u}AzjB${%{im|YBkHvf%jlFvH;ryYn#287YkR&R7``r?=IWSJ~ zp$k9zwN#VXa=M0@ALLzB6Bw~#g66>=0C!-F%Yj2?H5=lRx^m(O-6mc^thgR!hC%^C z3qQX~U;3zn?>PHFe-*>C*})80INe||Vd9O&m0LO9w7Azf5qyc!eaz0N>knlV$4S&h zrqiE122yO&P29|)8|rvpL>ku}@!|Qh7k_m_pChng1s&lJX>A=bhdPZU2cF?zhtQwY z;2cXFmyF&W&|qLv=28WmQLOR@-7qlrVczoc@@c)8_S>HCUQOC~so4b#5X=%?;Z^;h zrs?Ozx+(C`%vQ8{{khzG=}TON<1VmsA_U~vtIx0Agkk6QF}k-Y>GC~yc`R$gK$o#L zx@C1vBuEj`4x76oP(d&*G0^RoPVD!3wo4tObWm395qP$Fzl0~hYnZ=Fq%Zyv^HR9) zqkDu*FQG}zH7{qhhR2xdGj-%rf3&`DHXy98+MK;i8T_ zoDAID4c3*t?mq`11N<^%V5T0n6;rgS+y#G6R8E>M(SErqcOO+hlUV=xW169bC1=ylHt`FWV)!Ac* zTfj;Y51pJ&*@{?Q$gNI6Ww{v`NI9%sg?kLVzSHP(QZY&VAK6bl;Ikm`^7^Hc0R#A^ zY)|+MauL>voS8P<+TEjt8>tz=p|b-ELGW(L-AaH8n4u=D*Y7qv=eQ&Iy&P5>HlL?G zm;Zd_Y<{z-Mm3TH2-ke?bpestF(&O1-M+_5>IAI2cnq5xXrstwQDEjv6jb#Mys@Os zUNOga_g?Q4T7fiIgGVVyptHS3Y=yVZB>I@*gPCs!YIpiP!+im0sQ zpH;8hvV|y!-i(7X#aP%JaZ3uDh-m)gV$@jE8QvXjFa(<9XH>~Rf&*x8zxxvzsJ3b@ za>EW79lH7q1o4JXcg~1qgx5> zhhAKCvgWPWS6cYtefDKas}8$$#Nwqbt9noPTW|wi{k6M1L*cV7v{Dg;Hd{&3%}ns4 t-JmcILSkg_ax;cEHup_E`La=p3!ZbZdLJ40KkG!h^Sy5#T@EMr{TCrjmq7pk literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-6-GAAP Revenue Growth.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-6-GAAP Revenue Growth.jpg new file mode 100644 index 0000000000000000000000000000000000000000..4ec1800e69779ec946ec6b9faaef3b7498119245 GIT binary patch literal 43665 zcmeHw2V7Ixy6+A}Isv4Y1O)^IL=XgN0TB@xL_h(lDkumd(j?SSq)JC*Y*gt&6e&sz zy-KgrLl*=Rqyz}b%gnj=zH#1pZ|2N#?mh2J;%_1R*sQ(Q+Iy}4SN`Ak5l4vA0E^BA z?F#@11OOn?KY%y}XaKZ_4$&NTd1%+Ad( zEG~Us#%_Gw+}hs3?e6_FFAzZaw`q~S{%vBvninf+UKCVRlvFf7%?m`~Notg=RMbbL z53#8m(A;o2d{icomi=6O`rA4>q0@#Kj+?H%^qj)7Xc6pBQ~P;lzi(nee`sd^HL?Fa zuTg-J5=44SynAm-^g_rsD$NMx3ju}|zhIJO)F{NvA1@^Ns zOU2E-I{dd>;#~TXU(=YaetzM^i~3c0aj7Pl7d0a9inaP}w1vlQL7$A2!mV=B@wt@o z8Ih#^D~JHS?`f=Pv}~IP6ECfBd87>`&oNq0fG>UfhWQjGUD=Zes6Ig8*h@eQS4_O} zeR;8yNtBoNge)#(3XirYr0&lNv|u>uE0kSjUB$)ojwFTBs)h#eX|i7ik=t*s0acua zNN3~yb0PCdK}VTiH%^1Bbsgl*5VeS~Geg^#HC>QXn63liIPlYMzeg`XYbx4ER-DZp zft|)Fu!dm!qsBzY=dZDYj_(dK)`g$93>m*lJu#(!u7J?5lzGspFJjN#_x6cB zebw%IKHbgx0Dq!66cdh1CWNe-69M2A5ty;3EWkCnpv8#*XS7EBaD6lBWIfg2W_akNN%tjb_)Q+G(k%AH)^McvBBV*#$FjoOB=si+q|sSdSopqN6K zT?7)mR%&_omeMJiD5-<5+>%?Ju2b)JA`r#CjwfSJ&rDr?f?-UPd?Q53vdpWmKF1eyGyt7LbL(g-@9Y(XVU)n!kJDF{2V{L2` zh+p}_pkk+mDXK&hfw1`e#A7eUw26Q%qb;Jm%+Rl;8jk$3k513Ul`UGmH`sPMAr=2T z=h<}N?#bEJCRbH4Y;{)__8fxaTB{rNXmz%?mrERrzo2cbn;Lr+0-cRXj^0ucKF*?b za^KabaZ2XkKC^1tu$&bUz)tCuV`)O%rT9O0RGf(xN(!?+q5gqxPg-bsaGD*o=~>%m z8;s@(Ux;Ye78q<0IzE3fNQM^4M+8EWXPx(WfPqR#H1kvgNYop?E<&l$WBFQySzQD~z$lwT}2O`j)-CvT*^SL*lZ^$&~mW5Y( zT-80TBKD{Q74fDLC$7uGTNm9|PY&rm&-;=3yv6MJ9%wuzX&na1DeomfaGnjSu5GRf z7a9a~G+ED$tR^&UrGZ~AWkdpE(J* zI?^-8Y5969zuct&zu932eZ890KQd)L(Hy=`RdAvpD({3_@M@EHIbPN}3bELsNd!=A z2yFYzmQ?8E>AoUfzE~!Qlj_a;aK={usHYrfk}#FcqmAj3__Ka2qSR}d8|*&w*uBd% zNMFASpQ+Fq{>sN<2|hp0UC#_K;}nIygqH{S%1%mP%bvb+I>sXczOQAUynx7|q#Tdj zc8-1P7lxh+-Q2Z}U1hx#dDXmVBr-v1R$+1cDlA}BuO=A%(yfIz{1e*+Y-0=Zaj^SC zzl)U1*#Ytd891~lMYg(MQdZX-@ugDik?+_1+ZPl*FR?TF>3KE3#OTjJlJm@#Im?Fb zxHz5Y&%%}D#M zO4CbnzD^_Jwo_M6>luvxj;S4iO1#~db=omnf9AC+f6XIt8`n;6>E_m0STGA0fl7>W zOj&7q)kYJ?zGoawfw<=o_f%vOedg(XX6+?oWpFO?7^z>nS=8>=cSN8zt0;EXy3y}u zcXQ^&9uLRoVt0=)oGp8ty=%U3w#9-eb7Y}viomHAjaafDL|{Zv%>k<680#R9bWA=G zsH-_1Dm*S6^Qq53kakKYMX6wpaCM_}^~IjwR#Q?t_nPFKeU=7ZB8STVX0Sv5Up1>`r4<9hT>MO3obfm zSr70;07XJ2IP;?M4(=xk$lLL=utqA@1Q>5UX)>RtQ};x-iNMY>lyIz6_5VDd){(JeF=GRWU@@fNWemX~^2p+Sf2NLd~e|ULg`B`uWY2kWXtRWX2b3xwR+!eN7JUG93nYvo{3TjBoYR9lDWo@3Y`R?p( z)lVwb1^)Ws{znt*3eeJS?A~f>DOc|2zXg@MPIxJ&n$6q)tP(J#GU76gF)@V)Iho_M z_k7F3pQzP6o)rw^{5TFR-Tb^RN=dmIn)a0$J zCD-{vp9y1`A$)h$u-%i8olMl_looO&xhY?5NP3U3~+C8_(T}Rfr-Gm10t{}PXq*zEJZ{B%@qtoOJN^h zm_MC8a&u-WB|3N`X9l5WPg#k3iH3rfP)s)mHwy(`ORYbXdK+=;aa)kanX4}XHx|Eg zOqVV}xvK1l0G(%4mie92*`~KAN0_67zrsizh<7hhtR-emIC@uSKfMLnf1XF>g8ZZx&x-y-oy}R4JD$?o~?|I6Ql&5ZR+}n>xV4Wf-uXjojS}pcl&^ zx8J)4OC?|dTxNrGc!{6G@+kkDMjSSnutEXT$wC*NYtg}qgskPk5<`me50znxm&6avp5a;%N|8rZn`A- zgb&2ttgrUmj(v(ki}kL)3(qW;C>UXzgU*RjHsxkiGoXenIy^Z2V;ivNNfAPRG($C0a!0l8L^jXFUu$8E>l9^Nc~J%uTpTWR z`Q_SPdp98EzRUQ;#FMkjyF`Xv6nm(^K#kK`j$uN*VRF*K;{8&;PM6k#Ln+y?`S;W) z8f(l#*U)~WJgeoHh6{JeZ90h?wMzXjAoypSL;$B%qQ6J^ne+`~D%#2!gCbmTOx^qH ze840=R50iEEHUPIZdr+zJj69{bH2nO!kh%}1Qxum95dt8ud?!4H3r-NDCHgdLFVPrZ&-Lt2KejoRAfYlPg$e#_5o{~`3y;JuSyjB+0FKo ztL;0sUp53_QTZv*rt|dOev*G86r8`!o)_gCCKsD;Tt1*tWagf5Te}*$pybgg!$r)ye2XMxXzAWtogV z|H4Q{MxTFWhZ7ln{`JjDWZmcYnIeB_R{U@Lo{T4qLI;uj6VN=qK^+(-8$I5Y!Wh5;co<$~vH7cbg{YBFc+o-?jcoTs) zXRm*1=}3%MgD?$D8{wZK7D=wjsbgjMBlK;vot*4WOCH?k%B0^r^UqXI%|M$Ajp}No zKlD1uR*HwGpLdoqzTQ|IJhINa>MO||xSsff5Y3_LcLZL2OOij6lhmCb3O5|< zki4G1JpAloc`IjJ4JtO;IstX0rSqs%Q}#}$xqz7Y z8i{2$57XBhdzXsnE3b0QUP2wN;cJ_H3F&l_^O1Po+Rt+x1wW%0$2>jZq6XozyWy2P z?w8Ws=aUyM+7TqP8aPGyte520)M}jCJh?oDqZTDGCoEfP!Mn$?h~ZaLt;Xw*R26WQ zn%H}V*=Es7U$)Ej*?m^qoYJOk*V4ey>fTa4ISuR2x*s^|4)wRt_{Skm?W(jSsuc5#>%4KX6Z0%%nf zkU|ALOQT86NiO*sHcBJrZ?4^=5D3?4UK1v^fKw;YU?b&bgz@-o8)XumL0H z>9-BlA=CaGvk)IaSykiHddWq7}Bu0=$q7Stht>^mHR=d#Ou% z2kC&_X%IMk4V$%Aio~_IXva(oMoxU!Eq)00SMZ>YJ;$i?;wJu%i=NJ z&5HB2pD3GKqY#g5ol*w_(wOb0!T85G)JmW8oOm8qt5%=QG3EH6oUWg)&c%CcR<_q5 zdbQH6T`W<>#sM$^h-#2+3%1`?IX z$<`MA9Jwfq;)DlZsE%1{Fs%-~a{a=wn}uK06@*5HU*mntL)SGiMJQo$frD|r+qHE< zH_O01iMu^Cp}l`!xA>k`H?p4c8+wXWI;LgDDJS&xNnfQIB_ppi%fvV?ukaMU`L&$> z+-32Or8IC3o{tpA7mC8FRA9QQkA!~Jhp(mIQ78}Jy3V7)=(F(QebqJ0Ec)I{4-y?G zxZ|~>o7ViJ2)gSi<5~;57h;RmUNm98S99SvJUHH%ayxJqI?6u_c@sU%Ep+J35+$Oc zMl?|q8;uITS9lZU@uEY*P3%aAov>Y|TbvK?R`rG9>GjyP=O%s*e#`{78h8Zjtyn{i z;}-jQNa~CL)qs8}e>VInsl3k>QKJh*`R5gG95K|o$$s?Ul71E(8NRiTWQWJBUqjVC zS=MVZ$&gaZ9090pd(|KjSPp_fO;Q$^I%NO4=<1KxTKF!&hFpeK#EsC%#%l2&dWw{V zX7=4&(k<+4pX+*En+u{*a%&ZeL+Fc=f<9VteA69z+jX-AAwyj&#o++~zSJAN@R(Z#YWfs-|-7ZSy4>%53rQ-w-x` zjT(ROXC>!f`9Q67za@aox}kQ;mtu6LSPV2d=AQm3M2l|;1nu8iapqv_GDLs@-&+a} zUm!(T?)B$>AGd7vN(0ILD|yFyYY|Kc5TmTKM`ntpfnO~BOa3SR0=tNh@ctV~>5&>l z@;kVM|H=k$pW!ePS;MRkwbBJW08E00tKkTOW}!HtigbAPPy=`VELw!g+1c)0wymGJ zOqrY&ii^r#MBu2^f6xJL_Ru`~4tM^`1F6*(=`rqdmLT`OlAf~~DgI~O3QD4XIO>qt zISPnPkQ(omKgpm$lH9%$&R749&e2p9^PwApXiff^ki%U9ni=>GxBUzIdQHCFr*O${ zH}d5}N&goE-an08ZYj#77f9-FOe$%|%p$jz)D4lp(cmsf0`SuBaO1zMm679D+tZ5S zS38&;l=U4D@9#1DK^BX#usPkI`HN9zODklc{jRLqZ}%q|X#dSvg}>USWT5>M^tNvm zqW>mJFBxe6Es6AGp#66ZUm*kSzcqcI47C5Iu`^_#B?IkmxPAUn&6ljv{+V&2WS}Jj z?Vl6TN;Vz;nL(;#pd|w>8EF4dqK5w&S~wYK$v{g6+JB04N;VyneYD>z2MgJBOg0^V zXAUr#4E5iU`GsscCYz4GbM_cA8R}0=c18wTGSHHN_K(+6`X5WNMh03k(2{}nk5$9` zRrKwDOIA@b(2{|c476mR{YSDBlYO*gAMGEFB_fCS{73X1IlSlhrQ`ZXlIoLvwB$J2 z@18WAtkIG+TCzq<)@XkZ1DVV^CbN$JXf~*S>x31?61U5OHqM@0^&g~IbXUEhH_q?d z5Z?txUtLOL4&FC!Ky8Yr;<>d6>OQ!4fF-887z!I#qN{Lh`msE77?qMogYr8xQv19_nol3-ur)C5wf>)iH zN%>5Q5SSaHID>;i^wb_lH-S-@RI+XHU+Lx?{iArJpsqeX(<*t|NbVuo&auRRQ%XK2 zFAUTmPf7WITwqJ)!K9+~sXrC33wfWqDOpDZ0`&JLq#Nru;A=b5b?uLo%Jek<1q~si4fF zrLaX&wfFC{q5rx^*>|70k^o<`;-z5tj-fc)cS}=i;1ufXs$czFtW7@x?JV_Q=Oy%)e^%8V zshUp85SQw>#(Z&GMV`R9K|+(i9gU=Fo(x`y|2&n&KjvmGb6f9j*G!YU{7>v5ujNm8 zOMZ#|&h9Z$%EM)94sP){7ge2N+?aJS{|rn63N2|I?FMFs4F<1v^O{8G z{j)dct%xSXyWM4L(LCe2M$@@(q}`HD`#N9jk16E8NVz&5!tx%i`kOu@wl*t&#|J!QiZ(h6%* z_uQk$x>1#Gu=!nf#uIqC_{Dt1xQ+&!_q8yy4PHR3sG!!0tppodDi^(0%ACk}pT~Fi zT-+r!deE({tyN2{SbtV?xO{#5%23Tw%S*?PnLZYIe^qcXu5L_4jIzT%gXG{WL@?Yr zwNi7UZ?hl*qHu4`kPD7|PvH{(Y}?WZ&DEgCw)x2+<@AS=5A@tkvRHAzoBwVam@=ap zMd0YJ*WZIqkyMWxL}11-uFqE-O{()5@fGK^^*pD~_J%;xrp%Gad38|C`1KbtN;3h< zgvifTUTB6`DqQ}e_TzW;FfZrR;H~(Su$7Vh$c5SH`FNXXU8T7EG}{DkdccX}vz0u~ z^A(P-p4WDGcvyc%X^_#09huC%+=xw45b%=INAJ{k>b7W0u2^KJzN&rD=6Q6jjXy3^ zNMTJIxJ*N->674l9Ct?|qf08>p!Bnn*4}v$U2hvyzg#U|9p{t}r`R~+a>8=Nf=#e8 z`a!4=&h!B+B8~ZETPQKzZdD8qQUApS9uCJNYcfmw6TJ0|76QdOq zl74D|Cn@Xc2c(kj$Do8`0m|nhH&J{zjao&iFulbvU*pl2xTGuQPcxk#FN)HqVf6=9 zc|RvZ83nPiq0T4QBCk)5UApq)+XfqZs5>D{7PKI1U=@t%K!3E7OCaoAsd&Pfd1=0a{M zND|s8s$#k(?s+iy+!1DZUHtm)7u9J|s?;pN@WSSzTa)|(LRS_1jM^!B0 zg4Zbayh)MYz$3f|TIit(GqCJ#M1}R@RR&()?Z3M3^&uilxMBWrU5j2z;r)R>YJ&Qw z|L$t_$+l8Sv#uru$ESUQh{@})!YhT{hg#pl$If{wTtz)hKuocNR-h1%(B9XZ&eV5a z)!kq!?l>XJSG#)|dXH{zIApkx(92qZKOfktWL2+{Gt4+R&r&D(v4fFSx%fkU#_`*t zBlN%~GO$MULvNLrbGBF1T@U33iSzhMj82(~Tg$8|jNM(VN(IYh^0r*<$%4A9j^c^; zOo5xik+JB<`l5~9;G>l9tE$J)@{d#li|-T;PF2q8exIm~*)ph|@ZSob29t`7WE&^93xcHrHB zk9M76M$KiyeF&%wlBQ5F;V*=HjxI```7Bwbk1nV@xOgtAzlH712`$QFnuIH$uOf

U|R(GoS_CTKftQUk}Hbb{Gg zgU1||Lep+Ny4kT)yBkiU8WHJj>U$b@{(N#jqF#MQ>;r2q(|q|p$^$lizC6pzE>;mNRN3rDrU)J0!lLktjn}6}jmB~9p zbDJLmR)Cq!i!Y(VRAmv76;@ejFlPhD$L8@if%Y`RT7KKW+UpkuqfELhqMzRxeL3;y zMRiFx1Jv-f)@B<0_ehfakC1wPx|U-NMV%4WkV1ilMGHq;&9K|2i@v!!5!)0=xQf2I zHpvJNdv&zSGi)&*^9;BzS(O9jrYTpbWaRX<($@Uo*#=tY8_24kQ3$3zVwEs484 z0+;CHdxC$+gV1G0p0yG+o$G$`~)DcdfprQs5vdYAm zH^(no2_~K!_+)-@5aqs11U!V5Dg0t5U}0GU9FNY0T@*M1Xm{1RQRo_9kRLlAtz3v)|s5jrg?W|Fmdq=3b8DG?dcR zSGji{dgqDs$;=a5f*a~Y08R*C-WrATHFNF_H%QMEnl)N1`wK*xHa7ZxJjblQB_81S z)mnDNta`Hg^~7>{AJ>P9xYosfm9RJb;4DB7J;l20&oLx4DEqWg-XQ3M7&nKCI?dbs z(V{%B!sT+sBz&jbV8c^kw`D(-O3>BxQ1hp=LA0aVw%2pRWlU27=(k^jr@Ior!}K*CR|V5z7|G;=i*@1O1&B<_HxvAfH_-^At{eGZz3m=Mp1Hn?o z22Z}?bJvT>^LVVlnRNx^Wh|muTx?SdGaM+3&b||(qtewGv+TWU^@)I;+c$~n0a;aJ zVD;PUeqOVZ)^yo-Xgm2}8iG8c1PanY*1p-PI%_;A>ArG)%Ej$O6YpMvi&Zby&aez) z610g5^VSGxJ|g)QNge9ez2BL*NfO)2nmvUr({v{#eVYup76S zBXmNafA}d*r=_`u8tYe+4i51+emndz+^lIt5khNB@A5^{)mt>&1Hm-KFaaSrDn(r{ zacTfxpb3REnjoMpM1WuT=$T~w__wjrh5pVxfqR!}ibrwc4bSAmg>f-MUq7Soj(Xj- zeFVAk(6=+N_icQv(X-vKChyG@QS7D6*X>z)15Jba7^EgNfT9us#yewGaw4AWTU0t} zU3%>0$@!e)%8|8502H>;Gt4yOGa+Sw71^46WV>zVTl^&S(A^|xzlzIP7GOJE0Y?(_ z5*t#tMqvwEXpc(ISxNu>2F^*o52&eEV=TSqyVB9G?4YMCTk@9eDd|blBJ*?fF#fSK z7%{wQy~p7JgCgc~yNQ6+8&WxWD~}upF0)Hn&bg)?^OokV!jXFEWp9@5tJ7c$%!ojE zsWrPpbe^NU^_?nPTHB351%X5CpuNk7EJF%O=+vleJw=mdJbYZ zfP~*PRm2&z!$Y8A>BKy+4!_jyEZUm+dJ{xbCcavbM};?rKB0glB1i zWj(;gq&J1Pf6GaJ+tUc`b1iGI^q52^uv=?);-YH1wDsBHTbcN)*rLEJ_3S&(hU4w- z(KkUq-Ehi0yWFZhw<8NQU@BXF&{pLgpi@pD9csuc9?q2T%yVxXSB=hBmXsQ@cbu4} zfODWo2FUUm=m8^aSHzhJG_Rbt#yy=OmCyRDpY6zt8ofJ_^bGdy{8Pa>*Ie+5nm~7* z(!QRFCKejFwnhccKi7G$rP67%)+dHpCHlU1OG7M|{j-Np+?M3Oa@>_D>ydr?ev2_J zcE#f3;Z8>vfu<@Xd+!_Dgfl1Fwa(ElIreT6ff&Nsx22|#jiGh>+Ju>LF1*exEE;%_ zO6v>tnKqS!@!+ygXOFcwwDF(zEm9Xu&@$)&z@lTeaI z*h$668f&CaRCHzZsf8VP=6x8gU(`FW|Kiz#{U6HN@sDuF{@%4nE1bYg0fTO}R{gV% z{E-!gGMPn&1=f+>#g<~efao=j7J zsN33@rL{lhvsJ*0YytBx$8nn&M7Htel{^%J;#k|Df$X3m-#dd6RcF!BQ1sdp)52om za>ffXfRg{K1XBVBdT@zw3KwfLYfUv`wp?KYi*T}b9_tA)g1Y)7_{0QksEBuoQ4SIT zJJjxRebVM3)fSN}VrYhKD6J5IZMOGBpf}n8=hIk$3h@xiT`EW1s+9f_-z^E9S7>dZ zP@+%w@>EUtbMbK}+-ho!mOuu`pXN^ttex9Vqy4EY$&MwIpl3sRX^Tk3=`xM0PL$uk zK0!I7ulb(lAWT4vCAKGjuoI5+-=3IITQ|M(lgQH05V;tRSg^V^}aD6pBwMCjzq`4O3R*4kkNX!>(S*@@%rjH9hjuXGJLkyR!gq zsakVdq`#{V#Vmsign_ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-7-Non-GAAP Revenue Growth.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-7-Non-GAAP Revenue Growth.jpg new file mode 100644 index 0000000000000000000000000000000000000000..697cb9a316101a7efec89aa3f7f2d05ff01142df GIT binary patch literal 43938 zcmeIb2S5{R);B&FAc#n$N)u2}sv^Az1mx0HiXsXKDk>rZA|OZ$MXErgT9BfMs0b11 zHFN|8q((qW5_*w@8bV0Exx4TC{#WmJ_uX~9_uKzUbWkRf$vo$oIp_S&`90^*KG8;j zT}FC_dH@|A0MLPd0NNm+128c%LKzvDpit=c?M%!poI6=|>|o(#=YVnU;S&gA28Ko_Pg$I{Jag8{{?g?u4p$woxqEot^76jz6LKds zEIcAIDn21GDf#|`hmW3SKFfNZ{o>`Tw*`eo#U-U><#qK9jZMw(T3WlhdwTmYpZW)| zIQ-c7#N^cU3~_O3d1ZBtv`+qYTyy~Bufqa={&isAjEfB%*EV{32tD-IanWt_0yhX7 zJ;PpEMs{s8=mj^9eR4reoJZrHla2^UuFs$LoREz1 zNuS%K0sB=Ev#qY1$${#9G{B^?Ya#0`x;S}Yh9h!clF{3=$`Fg3>GYU1NO%MJShZ+p zu<^_3ve+*w>vBz%_w?89qe|gM>yH8tPM2R^EkNvvO>sKd`}tu%i*Z>BkPx&?-}%z< z{(uH0SuN>xafut=PcE;|CQwiRW5_vl4GmbAdZfHjKm(@2Gw}#YQk%o(C;0MFQ)pVJ zE+=62&1TUiZ`k9RlJE5V%;#Ynw)8{8`{t20tRcg%yjh**l*+w@ON8urbh#s{d z%mo>=^G$HU>|1XfF0`Dqzar0`D0O3G}qu{V{P4>1~?510kEvoKA6!V3$jnOWN&?2;~& z;k26`xb=kf$z_)Gva-nfFmpLNk?82rjgT{ibgsv`7FJS7a~^|+d0(Ej7(6lma6`_U z%&e$!hz97PefJo3IT^pGImp>+5hx;pe1dW$?aZ>C8qll4=NjG`9H{$P*W9VM&a{Tr zqUahDm-I_|3sYgud^biI8jM+9LoF^;E5TDBBxZo%BmOjy8PlUKlQ(U3i7)xdv!TJW zCr;_A=X2?OmOJcUa{xtw6S}=in#fn;PdyMkw?=;}GT;jN5{th9Dw}eW`yqZ4(kRGY zAn95-uDhd&``ilRLRCE4Kq`$Sso%{> z*YNGF_p|UKG(hFH4 zy6AF?bg>A#Ye8bmnj_jas3t^sNL~4-K81JTtB0PBCkn5hPh31Xbz6!Q+Z#l5pcBU z2frB>6M(J^?~oAYIcr*f03+BCpf6H%(a)6z#FxFM0Tobcgb~n9(kAqEG{Uc8R#y9; ze2Qo09yoAbS;_kCi(0r{uIFy_9tu3;vm!FoPt2uHHnrMO%0)=vlDN6sm)om3rU_Zp zmM-IKx70h?L|Cbp`7gPLmj=_ZQ|Mqx#;|M4 zaV8%SpYaK1WkOxqOu6ia8@edmY=0(9>QmglQEK)QN}gDLhqxBetH*imzA)Y3TCNt4 zf+d;;G^r!f=b7PnL3qb%C~!WYJ$btuzGASsR;l=4LI1-ZlvxHNmhdAEx zi$=EUJLd9JLzb-;5Lf!+{OodSOxjoZYgXCC;pAN?6(R-ZO5|v8E3Kt=m0nnuV}~CV zhon9&m!bE51Y;tl?Y!bI>&X?m=a-$rTE-GKlTIEj^R^w>j#zF%Vd8<5Od3$f(ga-XA_UflL9&%S-JXA%M#e46Lk=#^-|E@^X~cttgv zOLt5e{rdeSku_)*gsY!c)as=n$1IrvJCP7EgJuj|nX)bLh)Gm@^C`QtG6`=0Ot59wa%`l4$@^k+V+vxB%~HqRGR2 z>XNhFD~tpI*_6;vvD`RL=|6WZ0bf0Sr%nTO2N+CpCDbx^p>wzJ2y0k zkQ~ye)D7#YRb6v9f`-`c7pg}J)Jqx*g%zEONM(p^MUH7t9I=FRq5L zBJv$*K(!bRz)eBgs&`LYT9q{9G9Eeeo<~Nc>|Rh^Ge7jAzX}ieo>bvxrt9AHW+m?Y zH;xDAc7Led$V!cXj_|-EavxF^y>;rye#OI#gwiN49(iN-hKF<3^ac2jaVw}@y`I)X zt`=oIE4%}~758OCV?NUWEj7AN%s{p-rzzdmW=dz9y#YtGoUkJ=rqF=PDCxX}TEW9* z1bBpAzLOKzhu{Nu-zVNX{Am)I1=x{5cVNFsgU|{AzRZ{O) zT*l(OcTV*IOcoIyDO2P?h{IuhIZ^=VkbANolL5#7j|^ zPQ2>im)c7ME`4D^Q_BhP7#gr>shGZYo(6PoKU2*}?8_cKczI!Y#n3((e#q&zVZZR) zi6bE&H~p`rN2?!)N2{B8<$WG(c0Kq`tFy__tY{lCeeEh_P?PV0x)%X{ig5KIH@Qb` zU;YY0Nl9~X5zb6FZfV$~aQvwC&J$hM)-PiDR4t)18v3Q9Jp<4PPZO4F#A}aqyRC=x zo3QULYgh!0hDoRccmjz_;T*A2RBFuU0`$s?o%mt33y$oy6`AmJn^`n(&U`%bh|$9Q@X zmBuB=YfkB4M=mX)g-B0M-r%;l5%Q~xh@35x++{l%NbD!{1EdZ=6HE>-;Q_V$(|AGS6|h{TW9&OQ@bSgNt6@( z%v<3Rwxa?0j{`jK=TJM-gv*1Tc6sC+JR-79-(DWoEi5y!6VTq!N3`g?p#g7BA;?Nr zvz3Af>={3%3L?j=yzCc9wO#le_n7&N#+@vOd$_U%e9bIi6>Jpvs@oZ6#wTjAW;&P6N_6>?rgmgiPx0 z4kUb%ilG5?92BWG@eJ6r%8Sqj8W6-?EgWM}Bt^gROAu7gB>BP|(|h}Vr!W!(i=0RD zSEJ-yio5KW81)%Xztbk|Y+Kp(hO(V_(wv}_;QL(Xg0sJ)r`QAgQc;Hew@OS*q2nfm zjz;5-?cVmLogS%&2IPGfRm%@G{UTkV2Or{TM)w7(*>O`miRzkCJ=l+5<`a&ExpVSo zSW54X>DRca4Uf49+U0fPB!vbH(16i>h|t?voOH{mI+@7bRE@IULKufSM`iK`ZU>JR zakHN@2(utFYs-MkWprin&NlCKXmK)(k`lt5cr_Ar55cE?5O)N~J@T zP`H+|ydGsY?$|tu?JWtp)SVkYKp)9aSKxYtNhJ0$u2c;We;T#5{fRP=AnJ;CNxO*{ ziU>%)Xc7B#QfkUKM&}r5@2K}%Q#k?FsG*PDj2X{OPIvMsk=xGm;mj_rW*$ZGX{c2Z-?IyuSLfz09$X6Nu^C$%uTEMF1vt%&=tKcI z6D@Io`3zQD?4$WE?OGm1@M$sGe#Tp1x_ed&#_n{jZg`vA$6q+}JD#ZUYL5C~qA15e zTy!5NL*7hS|H!yMIpp;}d?at$pJ05J- zyZ`8GIK&jwrO8Ir&2DNFGXJP~0u3k7@R%A56Z7!es$d1CzP=L^2u}y)fiHBv zx2@##Ch)=-ea#$qd`CaVv`6vtasTIS8#(bA@?26OEA_kluR>~^P$Z|(HDY#juE*-h zRTX6Ft@^_A%jAS05@xUM684G^OcP?~9ea4UMIZkqnIY}{Cxx`jUO|xqdqYgu4-ALQjeO(d3JzdO~)ivy{rT-Q9L{h|^)^q1 zzhMjNENL%p-iyjnagNV=BPCaJUX)#1A+-0#voPN2?vh-$boRnoXEB^o+2t#D_f&d` zK9$k^3_0(+?J6YcX_Ppywf#dfptE==pcel+TR1iYQ$7ub9VlExNPY4Eyn$yJV_M8_ zkJh@;%=4L7Thio)-pDjo!4%qEjtZ?7D5)P}?lvVV)E_NMvn(p*I5gcNc5_TjWio@Y zatQ}(aAve)ZXE4kuqsqjJ1l>q)v~R&9e~TWj3u<>F>W+kg+0NZQ?4(N-uB)kZtTvh zU1RyGE$uKnqxy_(Rd!8javweILWcQ1Ds0dIWwQcV{s{GlyRWbXtw)8br!`;_&9oRi z;jqEF_qZ1w(XyLy)CFoSN z-etANsj!qO)H87ueDeYg$VMP3+ceN|Oa5HTp`3`<@Z|*}82T3+dAD92MLrq@?-L?> zkwAw7<~rs#qU*m`S7;Wtd4z%(g_bO8x<0MG_Y}DjPEDo436Q~cA>~dx12cVcJqbEUhc5#|fLAce8Zl(gh2$2c0)HJg=4LF0C9>LQ9;ODy_ zsWz$<4fr5J9Yo9}b%(TT7Fa2i9YnyTRwcP^r@p>j#c5{S|K=yTvFWm~3+udvC^99S z2Iz1t^p0wNb5GmJq=R81!EiCDYVA$ZFpb5iHTvYD=b{wGc|T@aW3=WmYQzEnMDDJmL_mVRvXUGav6qsOj*0h2|c zO%_YE<@-q*&=CHl&CUaqhTnSNAE8YwQz(`+;DP>TF_@#2>iSBpOZ>o97V)JRT^p9T#OQA(l#cK1LZ=wcPQ+73u_d94nBuwW~^%N+P^`$=q&wg$BpM9g^THzymmnaXj zWMTwebvZ2u_XPbUQ8LE+Uz?Xe(%0t4uNpxy@Db=;(+0)T@!$_pCVy-A_x-JYOE#mf z!_vpvB>b)}fz+z%Pjf%q2l>%H%?~=CLglx%r4stJ^%-9RQ|dY&Cs5^}Osf5A{s>+2 zwc-C(y}d=Qy#w9E1CGXc{gh5!uGQxAe}psgZ+OlvX!CE(X5NA}e}r%IBhI-6ZT=9? z<_Ey9Eok#^3zN2>&A&DOV+-2+SQ>l$YEJ82OSL$C0~jEWrLq_F;g@ykw=(r>adhUeVRH^CW5LoO&FO` zmL0g;FT!oO*C0vjJe^ssx~F=tlgVAag*yQvGcUbvyCX91PRAP?3I+5IBYMBUB?ME& zA@61@@=2=rT%)>SGDgp`q%`JoOw4ByNj>e_>#)Ie$*^5J>)j{U1&>JG@0KSFoX0A@H2wh4eVG>9+= zrDo%Zuy93_9oKT7J8{Bq_e}yU|2leoCctaLUNq=Vk`VyJ&jxP zz8Hndwf0|U4Gt9VDPHX!wrO^C-qEK1noA_cSf1k}pw?qC51xMc{T?B0Or__GUqm;y zDcrc^|M}2_$mcOwd!XtpEU;Ros`%`)l7d2MIg!%R`hdOKjqbLRBrlvR2c== z^VSa@81ia9+yldK&imrvQ7fC~*8M9vpCGal43+Z&Zm~5CB1^NjSp-B=iLIJpY4MdX!}dvG|8eQt8aKC(T?|y0KmI1=RL{J}#?7Zr4^ML*x}hvv(@~-bCc_CG zf(jtllWanzhtmKyo3#5`n&HHY?~U$_=PG)HYi1Y#@FXvU;7YEg2`PwE$trccP_I=< zf@SV*YFj&921`nMKe%#Z*U=6>M+^(}bJQd~4H(ga^;J%u>hj#lVH#h5vM{-CMt(_G zaukVh`yzFszFAPR+a(@{X;6zV7r^G7OWZsa&qR(|LB~J6`Q*ClIz^t)XwO|@}gb;D1sSVtuE?L zITbJU%aWMWRkL13V*~e~pb>tjJD5t@*QATy!hP5G7sPuvo4SWQ_Pnb$2_#cShlm=XLY*Lx?i67%Q#6y#p#Tf#OONbAa67G?<_j#bIP>%# za8eBVMH%fwM9g2pXeu_oY;fmldK<&LVt1I1Om_}84+H7y+(Kjm5@Z2i1Ev*_%oJFN zX=F#lz+4!;{%!YSIe)_4=AJ&+=0U_RL(NoA(OYOjbd%DVqD&W41FEKqD64D}gLnGL zAlnUJXZP%Y_fL0T7E~S3RU&>kT*G@a6PnEnHv$NfKPVvg_X3Yw!Mfk9EdFmDeHs<6 zuK$u`c{bdl%VE*Q9v_$FsbLX*!$}fc-v_T9+3n%VL&`BK>Q$GZ_vx;F=bGR&cf>?E zu1w!n&Pl$(j?ZIc=QW?)`(b_(PR=hP$6}rOxh#Fdg!7r@J}L3emizTZ1-Hhk?}KWR)+iHNpP1T=+y&XjwBcyLN!Ra+sGOsS98K3G{7Fuyn7z) zNCTdyP$xR-7c3>P=^WPr*!!@(9xp2^B1n?;Qv1?g)t4$FbphYRbY2Q?9(@^6ED&3Y z1{jmfRlNeF>6%Lv!||{fj#)nbtAYYu7dsWQ&&7&}_{wH@Z%opV_9HB{aEffSM7`mbaqJK899i{Z+?DE5$jMl4t$V31gV3B4T z&}c>jn1mt`Q~V%34J3Ktat5wuz__=rWlJvqcK3^>`&MN1yYsa^=z5`b?mWaS5xq>0 z5=U8*>}kNh-j2-xWolp|-9YCzr9}SQ;QxSs>6TJ|0A=MTj+;vLQh7kDPy{XswTl7u z`TpCwUrlEQYv1p|yC1M0_826${L3-hpEa)fm6exj{moPYw+jD25~Sbc4~WtJjB(c$ zL>PSWAPDj?c*N9pTeGjGQ5;hA11!<+dDM_ewI6_Se_{-4nM08Uz1eG%R5a+-Y8fp? zeLb66dvbq(J^DRMETsAaOzF=XudK4eA}Mz*;VYyO@;1xFZ|4%2zxo4=(eH7q*4ux8 zz5F@(DqJ%(fCVvTRR*qb0xLd!WUqYs)l`~7oqm8R`n?q;&~}d>U`2mYKFilwAZh$J zQyF-~>IYb)-;20z!McBe)BhC?*n)N6%WC_U3j3!p{eOc6wqV`w`NV&B|No1X0AJyO zEm-%x{`8-{Qs8SKum$VB)0h5#vX0?9}vv>%aZTEUBdY1EYaSAbw8vw#1^dk zP6ux*TL*G-wqV^BtovWNaa-BCe~~z{1?#>S#(n#_Em-#*4&PR`ZYx{&{k1B$VBP;} zApdWn`=6Bzz6I;HVBPn%j@%+)e(UZ0Grwwb@|_D}8K%GUjpar-S2<`xO__b6-s z@cu1Ww*~9|L2~0(w(k3(+}~#JwqV^BtotWOnOm@K3)XGHy1#A2@aL3c-hy@i6ny)= zY~B{E+k$m}>;3$_|8HgM{+YPb-#hk4!#W8|7i`55_jWa0AbZr5k~f^{JDa{XAw-po zNdt>BrI9B$*TKd;Q5mcO|FLg>%`#;xs0G{6jDU-H(TBl;Di+-&_a;ql!k5(j&u+}h zLlfVZ#qW&{lwGTO0qMM&XH4Zrgdry1pQdDjWl&zwfY5Yo`ubuo8tlSBdIijG9kC^k zpXB~nWR}Y4AEg273wel<2ME%uzttvzs^qPA|3|m}2TDWyn~ovgRiVvyt}F6eHzwHJ z3OK2dw_`Tus(zn;_7S0dG)>)lshCqyW$RHOl#Rwd(qDUG$NDZh^36l`^{)4loa zdtYD6CNi38*9hIZ%)gnuZzZ-R&Hw-0n=4;;wE~O$hGt<=|KsUQR+`|D%-$(;)z_R3 z`=z+_YX0J{-+$^a-y1RLmlmqa#ns$^Gv*W@Rw%ZdI#RuYX5Ta zq~EJov9 zy)s`8P1?tIt6ukqYH5T5UV`GJ{kRd9{uZBoA!dcqE?xx}KZNu=0jOtDDkvS215w6; z@Eq;r?0#E;tgy&5S(5Z#`ut-Tn`IHyo@>ob3)=@tPEAj7O^UJa&5`p=QHNix*9Xl% z76LodFE+K}&Cw#XxUdL@mc%3^%6c?DP$o8zMcS)SR=ZXi^BdGFY;)lyLD zJU;yzsCMsUHV7^6=>coNA2K1n@_Lj+eX`Hc*Vp~#)Ztu(*W>#6Vo|^+Vt8;75;EiV zX=a*x-F`_Vo?ALrO@JM`uWVp#L)j!cooTrJD1XC%a4W~!vlixun&m*thu`xN+B#Ov^EXa~sJ~BR!BrI8B)DQaN48{tJyXKW$lV^^m6vy5QWcAmq zv^%I{$0mLclI>b0Kc;#bN$;4auh?TpV;d>*7K+8{$Y;0!0bw zwP-HV+fJE_Vf_3OjOI>FQk6ch0u2R8kl}J`bboz=o0cu+boU`q$fmXHu?B@Vh+rlQ_mn0a7O5;cX|VQD~^;31-3{G+tEtjM{cho^Pi_%4Nn)E`!i!O85LspjfsO?f@I!}m?_ z@nBs<;nWu{_>vpv(v-pnnmCQrH66vZmV766g z^HHFBG019QMb#o;p6P-bxM|qc=N}~!=GnP6`kM~Jp5#?fdGSr@OCyHhg9nD~a8j$mXP8a}5*t3@$aB60VvdM?G( zp4KbB$s^csr9vo#sw~J(G%c<~kP>H9!bX*cbUN(_?XY|Wx=BxjZYRo5`}M%Im7ne9 z-uHL3ydj^{YFLkZj&S?2`|;zZ8Qvlq@;2lL4o_vIKZHZbZ0At*8k!hTwm~w~y-sY% z!VccMCnC$=sIwp{5nBFNA+42XvBo1%Co!X*LkIVlpJK%g9L36iWY?M$pep8K5upJG ziO#Qv9m(FYZPoLQ?;|>^AC4h!$l44I=V)f<8NQman?89Qo_g3mkTQn6Fi^3;=Xbex zg(N*!D54pjUC*3T?|i85%AS%#tRD{p_~Ag2-T(#?-nh^=si3&(6z`?N6!)OHXWIfo z6en3@_}{7}d8%B#5Da^QI!bcG(*Vvh-rdH<)ZVI!;_@fci65(OGHG^lfLKvGs$7eE7hVbFII2 zJ&7v%P3IEkZk*8FIcaQRA#o=~iI8D{H@v~6uO*d1NROfllTZzPR*I;|qJtQAU|}LX z4l3O|C9lO(Ekbg}p|?+Gn)Ma-78l7e)!&U6nNY2%%h!mx={cV+q;4ZtQAh}VHZQG2 zcd0xs#a+F$hMpXF)4!w0JN1WTaKE?wj&Dg5A_40aY^H*Ie*$W&_8>z_ZH$e#Tb& z8jd7yR%ha7kMHZO*Ui!%ukNmidt*QWtbj$eHIrcU;_*s18en5Wjq)Lsf3pK1l!F;0 z)|C~@*h(?*_T~glq!Y?Jeka%Y5=If}xa^cf8~p7(ht+&AtyQx_0bnc8ce$e@Wbp4c zQ+g;?tp3e*5QW{p06rN?G=Or>yQo!vPfusv2F~(4a&r{M`1EO@8WQlXMlg66V@^fq zmx#SDk46aSqi||SH#;q-yI55CEvt@(J<69#9(}kf-3@VB<89sGSQ5(eL)p#~^{EDS z!PH;7S|?};3VC^Itc+p2TJF);z|b+5Ixpb=nG;Ce}hAEVy#%uer)SigfZA%5q4GmktJ zi3khR>BYO8N)KrMY=I0T>wYP~Z}Rbrr#_N;F%g&O#wWGyOxloC7HQ3wVWeR?em9Hr z{Kc?iw_MI3TS9@`p}&`Q{Ui7LBIbsntIZMoq{gTAR5E6zUrkJkB#-rQ4Ny!8&Xafi?7fjYcIXKGwSJb47=+c_;x>C_nb=OS9S$D>>`t53>;A66|!Y}B3-BV4yq9pG;?!%e{IG7 zW-|jZmJQ#$twRF}EQLm8*2M+E7{i-8a57I`{b~>mh<{82R>on}HZ4JGFv$k3ptg&@ zg>huQ(Y<<+0eN9*$yB+{Bs??{&OS#cMh$QVQ#j=bt6nq!=?XUV##AC&X6rA}fHNZ} zwf3MIgtn91>n0kmYVHrWa$I}p_fo`-TQ+)5EYHNw2nRmdnh^wn0oTR_pb?wvqet!R z5{>CITN@C*!9>{g>SIkH-XTq1dItp?pT(RIEln7Yg6AN#6Xws5 zEkb4N zLs;c*!udnFxSeL6@Lf-@?6?}($_@`zL$0L=)>uZ7N*XW$GQ@+!Mcz*YZ>`&(jN@bF z(*Jl`z1!GLzD*q|o#!QA!F@&ZSz2K$AT>l47|Ya#f3bGX)kDApI z)VO_H^NFv2qxhd5nN@^L6xq3H!7m0B@%5BNuC1(YC^cWpw$#xIe+@mdfo-S+J>sGN3{7%N3 z&W3a}Tax8~Pm%=y8HsnFxAiCL)z>Evi|tP1>lLXiLYov;8X5EOmep#mx5GrLZY-P9 zox=oIiy-QRu(|H~)Yc8V)u&ub0iEOPc6)78PWVi|$m9MiHkYa(E;JxWClqq|-fu|F z+4a9Ti5O7tkeAGS-ifSO=qok)D_<{fp%KnT1F-uEhwft2=THTRE7xg&!G<0UkiW*Y zQK6?c`7)0^3Xtlg3Q#Oefa%?2zensdhF;=vcWu_s?<}q5=U1ErpQjQ;9k~-R_4pHF z)CfU*l=IThwR2&)qC$bd5Din8;Yo7Yl_k;&^3F zt5a&(t3nY9o0ct0)y+2&)~$;d2Hc}Z@r$|02thdz;LkQpp51_ca%a{ZTOW2E>QEi1 zu1*LDtpy(uzG~=Ie1Im_8#gOy1u#Kl2kAIxq%Wf8u15$V z>Zhh9HC4-atgQj&?u2>DZ3#c97bXQP! z_km4LewnDJ0ozW}fM`sr_flhVwOk*r=b*wFk3<7{J*LNpMcKVBNs`I8@~onOE&?@Y zGZeQ6Zn3CN1Mw7_I;R0HG>XJpcdz literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-8-GAAP vs Non-GAAP Comparison.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-8-GAAP vs Non-GAAP Comparison.jpg new file mode 100644 index 0000000000000000000000000000000000000000..2c9563458d7181b9aa15a75b7ffac9356509d671 GIT binary patch literal 49234 zcmeFa30#v`);JtQrB)PEL{v5l7L-k5*q2}{OREW#h7fiHN+KX2`ws4qu(_ZlVQWDX z2@*th2p|wZAwV&PeTRg7-}f)hOvQIvXWE(0y#4+!=r1`r&%Ni~=iGDeIrlv0+-IwG zs~2$Sg5EcJ04^>7fQ$1Fu+;%L57@)Kd-rbcJ)Dm{d-m+*;r)b%^T)S;|2|$I-)Elz z`G7$FBf^6GhmQ*Yfk!_-dR#7xoUfea}&FE zbM5(*clWN{`##;ZpYxM@_ny5#o=-meJp0YzBOsM8O+39HJmeP;6hEtMWQ?%QYWmLo zdMx~-AV_8kSQdffl<1>@uwrxcaWBfZKid~Qmq>Dqtqa4RK;HqB{5h|~1D_CM&r z?maxa_HuD^E`oua_PO?Q*aqO*&Ha}m@7h1B=#jM-$Zxdgv+vx?1>pDYH+_Cs9Q4)6 zFKz9%+5x zdq66EMf~JZKTPs&S18oa&ep}VaSpaID-4MP7fG5Hf zYaXdF5O5I(=z zwCsrT9!k}!)Q0&m*^#W>z%|4Dqle#5p{Ld@UVAaeyqVW1Q9&&t)>=V9TmVE7KD<*jtN>US~kmt9- z{nB->LxG0oMUvpN+~6_e?!ve}=j0pNR6n&W1nBUx4}p!!$PW(>DmZ|CZD=}h>UDOA z@PJ~XAen81Yb&fLb~ft35Ec}lmeYBAK1aG^Ydpce*z{v6=hs)BB}!6{`*- z_^p8aKr6U00>odYTe8W##9TR|0(H7lXNWS{7$-%QuAPS9?8 z^%5hBX$PggH#fmLFUz7NHHhROMN`s{ReI>jIXjiU&r6U9ALm$lx_KD=mw$E(s@EFRB1Uv zz{;YHW|>lUomekW4?4^Qf~k$SS;w>Q>xL>*ghm71wfK$gWV*GrS1y^{yX=kLUEo6T z3&cB@?qx5JF&(_>{W{~0l^_^UxDpiJurlA01EoQk0xmhw3(t{2pb+=$AHO{B*WCi} zguri?&|Gw%x*NlhRcg?6(aq!1h1Lw?&|+Pq%~?tYa$N+MuH;njaIeKmvAc$K7^yec zhm|3qpJDb~X+Fl*fR83Lev1;6E{veNBSC_a=nw)K-?5^vqdyyTWZZ_}ORRq#orFtq zSgThsjm!jBIV=}@ThMGXk*I|ts$vTa4yA1YPKVj|HJ)~H{!!6~`@-PL-q?~GFB!%b zg%Rael!>5GVj6)cPlHZ@2d{aQoKGATW~Zxw#~aVqz~bGV=>3x|mKH>D@zbQlCSZ=f zPH>EY*vPYx%;rJ}t}bm0Q0u8_bu_?f^AmE2H+%(dFsB`R;iuW!hBHnO!%oGwIJ)v@ z)hddaMQ)5c`ScV5Ig45a0)v4J7T$gf(7V*6WVaydP2+`OrE_E35=8g9&cjtB?#+?> zH;*f0=C3WX1giF{3&tE56KB8`zhB+e74IwCq$N=ARi$z{u}EcEB&}zHKuDI4`aG<* z;K8kCybj#+G}G^byO(7OjaWR}9)EvIAXoG;-E+N0>y3B+94~^7 z*aEEgS+1{+;&H2@pG>RA9a=^{ojdq7@X5j+Ki-g(H^vC;%dT>(_$;e6#Nk)Qo*hJ|}5zZegmt!RL}?@0w~%=tHM0oYLbrpXg?#Yv;6M$hrR zusDnO0(yvK_wg#Ah!WDu2&TPQyWqZdgOF-8y#>&lp3}9Tg^FIhYuW=mN{5+@mV}(f zI;LO7tU0Iq&%QnAR{l+ryvKJ;OC=goHF65pqIL(NWU#>GSF$8Zgf5mI!6d&Jl=gmw zquALy#_$G5WW7$mo}kx|Wt# zIVXn=l!lIK90}o*qf0N>aOZgG7H5MIAWp|hgSBhLgMJtjKD7l{2pO9)?cM@xAYkpj z51R%X6CBidBt(!2X#5w?mgFngEu%DtC6su?e7)*rcEb#)50^ar(?E>y@oZzO%GtnI zDGU51{7UT|p{N59_aI^du~#E7s}agWwzDz?`|@?`imL==@=;R|`w&dK3{f(qt?fJK zRhu{3`ltbwQ^7_w+2};@&AR96Xc$9rD$ZeZ|2=-HuD6R2gP&0!9^S1<0U6Jc@oQiH z@iY1#QJfSG3mrOkH$WB%hldr1ecrQD{k>>}z}vS62kwvY##_1bgj;gAv^XV*YAe&2 z_w2^f1E!bkGWxr~kbRhyx1lA5H|zuiG3wfCKM`6K-n31mr5?LcQjhJ2xQt2><6}2+ zm_pD{EviQLdT^DUg60oOu3(-JR-|s^;zAwra=Oz~GGU4MH2Ea0k~%7^EELZ%u}R@C9F@G59=N zJi73<-lz~ceW!18e}wR<>h3B>VJojDY(+QOsqhBHiCqJo zuhN`~S7cIYbT7K0ZqOFMBDCIFbsrczXjo;btP(JtmubZ@ST-Xqmt_z-HA4+VOvgm) z@%)qWf+L$>I1~-+k9TOEF}owU$LVa{!yZbP@)_jxJpn`I0DC`D@Y3)dd;X+Zx4uWp z8E339)z0HJ`?UAq*!h$y<@6E}fhvya)h^{X995<^{epUZva|C0byyjS;!tOXC>q8k z;c&!p8u@lRk-Wl$(i@@ll`qk%uPZi1Mx%KV)$)g0X}t!-Ojt>BckR(1scQ&Tg@_&tDY?J}U@(Jkyp@ zAN*q))5KkQf^j>A^8NZABPAGouC$e4pE}NS7Aupy2fTSvgrk9~_0gb#5D5^?2nLU3 zAv4V}QEYjn)scStjrQ~{fHItw`XFu#z?Bio%vvk$K9AqLh-H1Ao?iQ0k>wxaa+B6w zHysR|RvrNAdbLLtsh`46hw|Pic87mEpkxDFIPzn4i&A$DN|2&ASEd~8Rdu+A1p{$D z4z*;MmVp|H^|fhhcKAnda~3Wk-W087au$;xUT#0tS_qwoB4D*C_M0`(KJ=*)+sX_v z`-|WEi{fdQ#=8oAVO9Er3#`t?baTXmmu@tz1~7 z($V-_rY_+$@idVj9&kE5JiHxJ6ODhJ)gET#uyVvgzrt>g7k1PcG&s~^tYbJEwX_*w zQJZXJmm6~u#*+?~mE_@zq0dG5y1{Sd()0PJO>ylwJU)CSs}Jw2Jso=4$&(^suS8U* z`yNcr&LyPB#H+?mV`f5bx-chUlPSx@ghZ}nKhuBFt zm8s=7^e~LeXD1k@7mFfJ^xx41dY)qX33)R;aKp<^moY^29J!ORaXBrvpGVSFdhCF6 zf>UIO?-TJ&hseTk0$R*+P8`i{n%6wc$kCtm8n^i*7jovlfp!?-{-Q-p!;1^)Rf7w? znL+aNO(kAU3?K;1?-4#c4c&FiMcb}v_CkeO-_obvjY6Ip#=4#I$Db9S!VZjdCFbMS ziG{g_`uAl_ED!SW$JF#qtI8;K)9hRssQM!NfOJY)I1#tbMtvSsP*x5zis?;??hA>j z87zQGnu+ldkP)={w6c)X-IC&`6Z@UXEO=Sl*J}BabXbsCuO78e-Z`{y&RKK4MW)cl zm+Xo}Y6ov9dNpnVN*)@~+6z$=K5j;c;zy$d=ai@Ij<`gkr1ojcXv6hg{Yc*BCl8(C z3#;*kH*y=YqjQ}_km@Y#fK!V2<#0Tv{pot&t3tNaYBU@=no#vfuB zH=?Ct>EgR5u6n_m6ugp)%+Xf~NpexF-ZYB?l6`pw|B-t6*RBbL@+>qvR8mej5He*` z;`0?A(yg+LCtzy`J!S`f)=hNZtihYN0LugDANScsvhvz}>U2%4YFl$mGzg z%y5&OxFf?a(SDmw`kO_m{u3+X!JZrYdS#CXB69Z+>Gb_EFCm%1`JHf?2j5S$gI9Jb0kYOurPOZqAS-Dq3f>X#qr|U z4wXa1+3ROpmgQiNU`V-+VnT6`)B(eYno%u>3cbS&LP0LZI z@vNnBE4oHXq+CR7Rh=RNRk*S^K*V(W;BeEE?DdDL@x=LVzHhHCSx+5@!z~DBY{K!8 zO0P+HC4&w+9H9|3jA7R+Hs18z)eRe(xNA0!{DK#9=2%yIAhrl8Kyt=Yy3bAw6Gl?! zu#%2s`w2|UaI~tk)qx+DW!!FPD%W3b3`|+KX_&L{*6Aojs>=}Uo(>bpgv+a75KK|W z4Ghk;8!}{V>d+hSe2H2CqWzWOcip8UJCtAc%FrAXo9q${#N$1E{Nn;n4?*3uKuuy! zFA#m3cds3Bdywb-pjS=S-C3@h-72iCB49UeJH6z7xXEL_LCg&gZCLP{_mYvbxe$t{ z{;;>M5@vJXY;!zP1R+Kx`2EN!lT%DT({5gP!pIx#|@~FTO4#>x*XYcY)QZ?iHA#P z%40zQ$^>pMyqm@4f1SnInk1zdKG|vKi&tZGM}-J?s5g(n#tKH|wg4#v$9x}rw3inG zjObbL4G*6#EwM>t2`dc-#$FzOp!U+vdoQ`{Xu=x>96=t5;`qkMPHg%F2x^|sSY$3r zch4=xztSQdi&(3la?~!)^7yL%3-dEk$6PuY3h`zpXlO6!VR!M|DeNLa2A>ojPB}fY zCTVHfd8Fmsj9L38GQ#K*J(Hz5Z zu6AkLfll`ZN8)IG*uB9$uhURFNbaxMEOQ1=FQ%`KALzYX(ULG zk1vU}aA|Vjs!vvH0E(sLIc^y>oXHtLs=pLz)^^cqQhF$Xw{gJ)|E6e6x2x6_h2OiE z&7)S)uaC#VX{_q+C%@=4G1G%2XsOjJRhe)k zS*XRBQPzaPktON?Z&DD$meSJgL61mW|f1jTqf3d8Yi zKf569QJRjQN?bWDN~lsUupC;6rGj419QPD>&_ir?k41Vo<>$hI4J4wSGB(A|v0YMS z7_-M2s?&JmlbA%{N>R%$kI6$Fb=esgtArairTT|c3tsSQGIi-*jd6)WZa^&|T_YBL zc=+xhwJql41O7$X1~b~GK_MH`d@*7ALS*WhGE4IWA0dH~F`$<43VY5ek>ogx-~5i< znc|1mKd74S31yi^6pSB~@1h!5AOc~v8J&q~oEZxi(5c?E#_*iMIwkM+`vw!kD`(S+aeR+5#YeQU8rHACW$2{r!7&hCJ4^wEjAYf7KnwC=J z6h$PY_~4QoeeeD&GWGA)SvyP32GxUNFU@uyloG;Z4!2dd$3JAQrx7J>PWqXY@jL}3 z2o9z_Q4+mb-sgfYB5yQCf&(9kPx#~wdAcyDE@FD(;=a+C#Qenan&q_e^(}zzpt_sV zaR9e&`E*!Jc;b&Q4XJ1iZd5*=EId4Td#zqW`vvHao zu=c8eO(ZkQ6ICnx8>?FdH0yC>1E+`5o~S)dv3gMjFPxn->=*?+pn-R-IAUip6yZJ z;aN|?ju4!PA+Iqb7*mM8CA?Y9gvzi~e>w2Xg0Z|_TQqYdFbPj7BxCR^n@3kcy`Bx< zd1{|g@A1bcN4o>5TYxt_>LKIi#UU)^XB}i#Eg9=b#C1F2r*-rn+1!3O=xNV=e|S=S zMzjHG4}N>x^9+%i$87g+azznF3H_Sdqcu>k34UN7k3i;~&M#BXELW5OpN>>``uFTt zbLn>Na4TJJ&a^6yMJTpF8O8<-{w9!olYQ*c@d-2#$4s|d5Z?6ab1pYq{xo&j2W#gu z>&9QkZY^|mD7?ZBP(ilVjmx*TUH!GIf#z=CcICtB z%#JvEU{j=t zhFUP0NI1pW*s0(~hQobJ_qr#Cnz@g;=D=oVnlilLbsNq;|ZybMEgp9QGM2u6K^AbBD!FGKuEDcN+i4xQ=(X- z4=Ez1b*$F){kuZ`(fS4(`!$H(hWnN?E9BIET#kAQI&KJI_^c&!;n&3=^&N1a5Ojj$ zasz=hy9vG+lMBjgpSR?O?(&v2t4!3Ucuj1sdQcnpO&1VmJCu-}-dlhJMnX)G=)QDU z&KhT@&ed5k@{%`%TJybMiP*FGO%?v@44kH<7OLUN3Urm6T30k$|IQmr5 zNUjrEf={ugfiTuCo|qckkDhHCs+BoitJEk2<$Zeu!=(aaF1D$jRtVy3a-_mBE6#%rxz<7f8**AZI^Icz?1SkJJ+pW1 zYYob^-ZB8FJ@riPY$IB|P{om;0EAH$`=z6CsmWSe;KAre*K0$yn1?+unF>t&7sp)a zFDAtbp8HG&eZhY_-x#68uS0Vh~jcTrBKgqL@c z;MK-aE@(Z<+YB@$)8oHVz_>NaYe!Q76mS4QV8+hLxn7bkhOEil_BuB_E zSz-OVK+~}_gf55m+8e%@@Da-dH+Oz;vuf#?84Gaz(B}-b5OG%eO0s&FfE78(wfQM7 zd>E6+mZ$^)o~;`a391j$hEtS@2|a>9vA9a(7O{{Nx=elcc%@RI;aRGyBwL*9I}P)$ zw+L-t{^(r%4_@GL>>U!xDgu9F)nxtv=1^38!II(}xmQ_SIvcD(H!`MGGqg0IUWPYp zs~&0ST%Eh4Y1XF)j~lV?tP3+|A_qa8E4=?~J*2Nx;&$hkB_x?@ zJ?(u~1dypP$1AbyEHG7%4<8=Vq0uH5W#c|ufuIVzdBWzMM3AHx!X(4NBhS@{(rg)` zGm(WrOj{0zla-Ts>7h~wS#FeBIZxUcVkF|B&FdGDrOg(%X;7%Iju9Aa>Bh)gaN&m> zjavVq#=~d}kf0Z_5Vi$yaEpi~#bmgRQS~>5nqGVIhn3H40rc(M{US^QkIbx(PFhh; z5pirJ1FUsuc}1sR)~Mh}IOAy9ie@guz!hd)Afn`LjiD z@jiwI-;}$^d0u&Np^XT7V5sk9FA4Et=HGGb_T2(F6_r-2+^wglAEa0M*RIVV99dxp! zc&KW7sB%o!bJ@6KE8jZHc1IR-rz*iZMGsCI*@2IbU+S;#9w&q{bd{Jw41>*mB)3{i zK8d&f)!(Ek6vfEW<=g6?=dYzSOJ`7!g4$UBCXn$ISW6>ILx?-m@N~4wR8uBMthFz6 z8rKD@DU}_)a#Qf`5TSY#ELYvJtO>(Z^o>NsPL*~rTqQL{zwy+Hx}GyKdBUkn>j~IJ z?g3h<&m*W`s(3ofuFEJez(1SHgfc}$MELotueY@4e~EgCGQX%rWBE?%zf1-Na?@oeX|doBcq^^g{0MwqF^NkZykubRu&W%xEGC{43U znOlHQE4KiF&N;!>eM=)7sjJ?nCOPc-O{JvJUwzmPL@##>&}SMt<>C{Q-U->z&IvWh z5b2z=N1y*K#n6-ZcHbCYuOgm*(!tAnz74ZV$GWZd|2jh;nOE25H=8!pb9_C!aZp`^M&u&`^PWHdP^{jU;&Acf!uMgRu6x-({&Nma`h6gL45POz2klAVK)i{wQ@m-h z7X#r1_aO9(KgC4j5~=)r-4$y-RKrKrZ={nqp$3gvR!X6?tgpVkKm2!7?O*x1o8v3{ z>DEhmpS#&8uSYO#ZfF0x>9zG&oH_Fd-g{*WaNW7BX1vCckF0hkeR=~So4&u)!aVe8 z<_5q1@wjw@=6d$#flRbm_D|}~Nonf|I}QCCtW&-+u?6_tdGoP!oX{c(tF~~M z0aL76^$CwETk?{-MGh6#bV7G0B~ik1debkAmYVA{ZiLvL9Re#ED&Ob6VEV7f-fyq7 z%f4a*4yI}=;8qQv%M0=bS&ncdXT>|SDQgSxwQ~Y!#*-9m zl?%otal41o85$K_yI-g=SQP2j;?DXE=pH@Fm3miqGXm5 z-#+ZWyBmGtBNRS)XgClS1zlq>m_OxxmbQL%CM`K2AfQOJfiv8!X!EZRDA|3a z`X7YecdO=&jwIy4`2O3KVuI?v%;qw^>K0Y|{vm_aqL$&-R{e0)gg1hk<^uY9m(aD3 zw8~{t_|-Clu60n1{KOIA9D628v+}3DRkC&C4KQ;gu}^*|+H&zLy=0jaA0hC2(EOi3 zcMXqhuaN?eAx{TOGdaWj;@R;PTxPfLcS!6xV8XSJ5EVygM~f%qG>| zR`hdD#D~f3lbos(pSXUk(LLs@{R4&y^Qs}_hfffcZ%Y;D_|AuorUgPiX+#ozZxtu0 zdA9949`x+m$0+;`JoIyg{iz=?+vM(Q>iyL+KIV6|&z;$JW&T~?Sf2*yd3=l{#&}cJ z?`>m=mX_6iTj}H)zT@~M5-I)12|o!nM-rP~-jwzD8pGto$LjwvcpNnNBQYUhuIt1-6CUKfWxXWsUFtY|@X_ z`29)8*%u!p@H_Ba3LKjudYjo<=G&`Ic_tpaG35P)-(#+0A6wOSeF0zY8-PRR=NJB` zcFFCHWD%2b&_er1_qk!F*M3Kzet*$jwxIcY`(hri?drQ)Q`L~T@J^xRHXL0pWRy~J zSCy54{l|)}5N^^I(C+|udOy~{2SRR+*p~jd?B2+Q0JPu5!htxgyZ1=7A7jSH=>8gh zXNlkD%FYtM&5Pd%?JV&-OZ;yXm7OL214ivE@gFebH-S4={Eij>n*@bZXvd1*=G2ZA zzs-x^2JKk!JC2&)CMf{Gj-zIqQ#+2DZC?CFXvb0W{xQIgqvjvW%#NdGn?E~_nr&YE zMrg-Tv*W1wjiR#SsQG|VJC2$UnDLvy9Y@WMqvkgW3a8MY+)={`6WOgd^|ncie=wQn zyhoolZ*C|*I316ASeI5dtvHu;G=jM<5`VbyQ4wikR!4`v5g{1wVL|IJxcI<9&BX1o zyi>ec)LPt3Y!6vNK!+jRakwRCO0n^QYgGBSlFEs9TfNy)Z`H-Vvs2t8)w%Y9_ltma z;ruYK#v4UW(b%-)la)VAXFh1iYcbpHbRpBBC{&yh+GL+?!ElYwT5JLe)oV`}G(lkH zHTqZk#UwMFpUih{eDlcuMKCtkTxw znVe;(o0*XBQj?PmTW#@t`)}j+8nP$aMhE>rBC!>UOw432M$hm(YZd{xY zpRYLqAi>wK7QB!6XdikP@$=VsgimI8yC~%RhjPTzfb&;BM(KkpB9LHP^AwNRg zv;hmOHH==E`Fa=lW66KJKM44+*tZ>!z~@|ktPZBFfcL18JGqyT^*g%sVNou{W!rtE zNior}12<9%jU;S*7QNIA|cWgc!Zk^M1k+->{1^FZPh&BUD%PL#ET9>13xI z!U&20m6g{U&7u7;Nmr&yyZ0oMVg(P)8W<8ROzdX-c&9{S8fN@tZG8J2*$ST*@EGb1?$+_1(dy>{FN4a*({HEV}{My;#h_f(%6e>yo^GH*7;sw zwLG0Mo)O&lg91|Oz+K1j(kd$CAf~+sLKgRDi0;FY;(?1UD=(;*>*h=7jkgl)U%!Yy zCQzaumF;`Bye&M!t25sRODqT<-aM5Q-YqUw!Bq3Z z*_zph)n;_nxo0z`PW{CQRK6oNB|wGbgx+ku>)C;nw@6QPEqa=~cfO&%@>aW{T0h^R zvx!o1He~_>ZaUV*4P?7|+z4fG3m|P5N@-#_mRdb>yP0dzry?1-1;A$MYRx#wWXBHX zGDVbKra%by4ZEVU&d0}xWH#Kc*DD$Dgtgl}B|%4otg7q?PTleOBhfCk?4V+o`kq;G zP)400tF57A5E(Z@5o4OjDAjsid^n_rIzFhow#X*4rQ%B$Y8p~}W+xz~2*M=NX96jY zM0hhb=$!uz8(qm}0QcKB`YM)iXZg6qJ^o zkg7dUokpM4$jiLsTm{!vaWR;;q+m}Ii&BEJlyv4-g5+w#EQtlAFeXk&Jxne2Th`Iy zdDE)iEYph?mid-*PMmWKIN3Q6KAw&Pp%1&Uuji$XXUmd4&oU5_rm_AsH5H6y2jD6v z1gGQBE~6x(%1AhFgp;^sz{t)*oA|4=_PQp}?A;0Z0*@z#RMT-niVRT5qPQcePp(C? zb>*Irj#tgxcfV+%*8=vDZYojc25aun2vzajuc5d-0(<;8!x@?2Nu<=5B=G^3F5AM znOyzzl{`ZySzGWpMTNpD7;Y~Z!QtCghRKq?xON9~HAZ1Wx=a3DvvskL<0# z{Z~5`29A6aUBxop|G*lF@0AK%wqm0sT>9-@*y8qnxMAEd@-dC>ck+m@czxmN%U7Bk zQfo=b(S*8vmR1$EAd{86ltF`e`dnIyx>TJ;&0xQ#O^aIcXG)Xy>Dd>(%!(t(;$2v& zfGh1pPKqNxt6jM}D4Uvl&})M@m-q78hZ^qh;2jg>6jI<2k`Y}raR2Y-8RzHDqt`#^(d(}5iIg6p|B_FjFqBl*{I$wY zzvTJfFj6o1d-nvezFer3S-9feB7CoT<=fz^wHp>9-ZdwGS!;KN{fIaxzI$;`#dbjV zhlvvsG_cQK(k5_^xemPiC4s+U(k*`sa%78!aRRJ!$hozG)k(I)S2qx6(B;QQ=NqQu zJZt0wKbJ6k__4jtU#Z|maMxU@puUF8etB;b8_%%X4GWRqcJ>{*98U|>$fr%y8I{O2fHer($mk@f-RMO z@l1YX1?Ln6inj)0kbF|u*~D>t?BY;s!h>%e3L=X^m5BOcy9Et|*!MN}?Sep{Ci#1n zqK2AZx3qmhUE{i5?ts^A?QQ^rIW7jP@L7fs3~o4-*I2({mF!-JT~(21y5A%(_&#rQ zL0&P%$`rafM|&e0?~qBvkXPM4F2Z>WMEj_IiiA$@jNsM6y@|Q#19!^J$d0)U%j!-9 zB5oKT)sO2b2R8NXO|OC&QuX>4FRD$VuUDXs>4L@fMb%}lD-{PWOFnIvCaQ3b<#b6P zaE@Nd3JUWMQC5vX@_EYSg9yGxm!?|FZj=}^U1cIXk?P`{4GV%lKPJ)|9A^fB@@X>F&_`g@AgSh;fkrgBoCt>IRm^yab(wq%5cjs zYrTk8iR>(E<$2w)ENmeN_13GiPz!RwH+d+Bd!#>qkJ1$Htfjmgd0JHy!Y_cCbVis} z8~LD0T@WrHL&F>R5xW9JrHBKK+B|9Vc$Sa_w-a|4oJ1TMH^oeQjwOZTXE|xti|X34 zR{pmdyj^*|bR5_9cC43hF31RdOej6Rt^@Zd#81=zwzFYxnvrK+FBb9XULE3!E=hF0 z7?-sYWh#x5Z*k%nTTDxrdXiHHjZW8V0BVRV0PiR5m$!sC6s3Gg9 z)N&x_2yg928}j?Gl|2>x^0qU|u4W}pX+&1h2%8Br(ttt@Z&(e<^-i6#KH5J%CL1qN zZA!BhcSWcQ!>GDO23;UdCaizcA|r>BQF-_+o`Oz7n|2PURUt8ScVBV|rq~K3fb<9g zgE+5^C>!Lx^s!m`4lj8Kx2Nfin6ZZql9TmfW+L--u+i&a8%`pIgQ?OS$c8%kIP>r| z3AK;W(AiG;y8Zlvyr4xYFd>IAw=YiCapbl!C$N{MwsicaWp4V-&R$zpV9VdT_=$ zhHqaCFZbo22xZ9OAaAqg2lZKPR<%Sqq)!X1ILX?<2&BkdoRIkjHyg$*^xfr!A~(z? z#2EE7c=GxPi8&v;&2I8w64J1oS0v~dF#2$~^@QvjRHL7om%Xulz5*T^1sQ6-K-FiG z+Iw6fV*7o36Ng-gKHWa)M0?KZ6=7y?%>y4ai`^h7Cos&1mZRiu(cSYsRx`|SC6R^3 zw61)U=6<`=iK%|<<)9NUN9(myXDMfD=?%{j{FkmCSnH}1klCJ0i+JtX@&El@QB45}rYk4HLj`BUh z8D<&@F$_zJxR=*6BITG!Vl#5wG+Y=Ob*U)xwHuvRUdX1Ef{D{=#M9@2kwK zXWWo>;|PKF;zYaAl&7;bm57|;Jh5<1k=|;2{q(v7jU5nY!hR%mV5VBRoShfF zw#(f!+-&pSAgN+q{!z7M-Q>_Q->d~VwQ;)>F0CQFU-sY=OpO2!f<=3qxXyp++`ZvXOt&CMY5mUeWwnt{}T zG_a*hd~2Oama3+Kd~tKuVT7fwks=c;G@->W0@45r=bjZVHsmbvYS|ClQmPUh(p76G z;6Na(E_TYgX|UTqbhK-JeAcEf&(Rstn=2Zeku57?v;W#lq&;3&#h!>yXTwGKS$JH` z!3Mr6IW&HT+Y&{{RwiOWs%zBYeEp_e9eri2L`Y&9QCu5>-%I}G#NNj2yCuox<+bf( z#eT5N9SEW$r!>-JI2mdU=eDc-4+!`{qSuuV5|OkrccbML*DD9#1`O5v+BXKtJdZAUP{1I|l zQyZm>q*!y`p)2GoFH%3gCnvWR@*djbjqmpTs%rh#h+!4D%+dLI(0l09VOg732C?gK zjm1aBJnta@<;L$T9{=ipg(bh4;bWYU+Ep}uhgCIk=K33Tazgrc6XY8}d5KQd!+clG${81~2gHsx*tg?v9|QVI_IgEeiB zD@$hJvlwLChLb0c;Wqrsas87<(e8i0Z@xg069Z_Pur`yn0A?q8&i%?Q{1uf9;5b3X z9`dyCW-_x{EIsrswl~zjvW4a6xlruh2yuPASS=o@2^^>Evj744?B-hq;JWE?xC6tLhh6n z^{V^>LtLOZHZ$^qmJNC#jV_|S4gX39|C|r_KNwBvf}9}0j78jP;KuhigMZ}?{)*Za z#Mw6>DaK-!XUj}#tlVFVxNVQ}ukaNr3Az6{Xg>LprTljGZ{haw0~MO*Rh-Egtc#A= zbB8)V*6gR-9ub)fAGk^ugMdky(5ZdCnjS+hK2uyV`xOFQA52s72lMw@=dX14=lUj- z-<5(+Ed(bwlN7u*x8a|!Z-4(ugT%H6lia`U0y{9j-SNMJ_D%0z=CUr_c6E7|ZV&R` zmF4{W6Fdvpy**3&pMEfXx!r%+-JdaB7KIm#v6<vfFcZXD7R}lik_L{%-}w z@9bpXdl>)Uip<$jvj1->+4Cb=*Q@$Z4fRd8Ou_kc>W{FWV%F5T=J>MvsddxqHqX-+ zAu{wbzN)oP%Ps}kk1uDQUDp;Im)?lM&Yh{838~0z6+622zKH$%T0GT|#`i_u*SbCv z^E(I&f8Y!CLx-M{yZ7@O`|iOFXB)3LuO~GAp(Fes6P^DyVO(&i0NW<4f8(}I z){M2i5WN@Vz6BU$uI(E5;ACa7#|o{uH)e_kqVrIP<DgdzUR?DW7>DyM`9mo;V`t&JT6P_TR*#cbjob?EOZ!i4k zOo~5ZQv7#~pp%^Q?Z+a2HoG1eAaPS19bv~&G}+_(UACV)mtSq%k0pf#1Q4FL^xL1w zr3MLaM1^rI8;*ux-Cn+(7&@8}TC+j&R(fnm2(6?C_$o_HE|nxC)??mjpHso=YpR^r z-f{E*$Ap>+{EG?2(HsVe!7;M$Gz!OR;;5K+MiYnN;7CjBPT_xt8-x!+JKiq+btuyL zPVp9t_+pPm`AnOV)0TxqK&Xte!L;?Hx*M0tXL9K^8=N!3i*%}79V8~++tl~oFNccE zi2vN0`H{uAYlU{kqmr-sZ|dY)>r0&{%lGJ>Ywiq4j_JHJ#G-So~(T)o+iEY=TIZ)w>0a!RI) zxKsSU$eC}VyfaeEpDFQBGkpld2X1(mX5+iPU4y|ZY{9^<|1@It=MnIa^u!UAksFtY zpQ6ey5|$58l1{pRinYTkH9sK|?8L>v7n^{_ypP1NEasQ>Xjj_O-x0O%yiF~rkjoYZ8Co}wwC5BTLT|}aYw3dRuZQeF2 zm31aODR?wApnF0|VZ>_*(OlV_MjS7^tYJOY(cv<&}%uol-Xr zxHMaNWgDemYXj6=d%L+%((YvYCY8bSNC>@+f_%4(8R{l2WlS>Q2Oo;!G@o~Dqnftw zjDYicI$zG(x!Dg#mhx_}OE`nQlIfh`98g&3i$pV9kWi+Htq&@6lU)E|=qhdjXu?1K z`a&-P$=+CVKBEEd@w1u`c@HT7!2dzC^1+Rtl)}w-GYJo z?;$AQ&ww{%st=hngy89O%bR9Vty_S}|KdWuqago9(_!W>md2;K)j4F}bdw<_)XPon zHs{S|lYchx`ZM?EzzwCm7jxNKOSik<*jqTCN*pjj7jFUPmI7gZ!Pm2v<}&QWC6Gfi zD|+_6Ilc=SRE~-Fj`25><+PI3&f8CoS@?78zX6W@w|((CiBo=Dk@y_xS>Pjuk za2X2uVJT{oVaX?_75#1N+G=yR?bs*+_`%UkO zJ5qQaCDoj|68ygX==a27S2c4McW#a|awK+!+cP!_Z8DS%?!(hLHqf+3_!b~s?ffN< z3F)%%SHaO(Y_|46mrfVAU*c(L4lQ$W7(xY3?Jr4v4-o--|C8wQp1%n3=H`^+OX@7;5U z02?q6P(-q6@#d_yI}Bo`!^q*Dm6Y%rfA<#WsKh90`uy*2i{ zzY$Pu3-E^1h@o=GYo5W7LQbud{~%xC;~n_m%D(W8U5(spsWB;;0@6a2$^OmoxX1E}7{iFHhPC$2r{#PwBniv8u0&O?&8p3v7IgSi~9 zo3oPkjvY@vApTL1VDI0i*}7)OwZd<#W2k=}u#er?`t=Dh%9VF*wJ1Xv@nUWZJcBT! zobe4^5G%0^r_H@U$4|SRYCXhhrA@PK?WEYfRd!?Bsk`Ss+RXNaePNdd>ng_cZP5Ni zPqI7W(tLDc<&w2?fHb?_odI(F|4RGvs5H{_T{7K?oi>R>8`K1sY3*p-Lr{r(_r#^U z<1$_ZG;SS?sHnK$u8~gCQE<&-6vSmlC!n~+xNk8=L`Bf12`=Dna9?nZ8pX?;d(WA_ zes_NN{Qj$|^HrUyQ}w;?^FD8V&!e{;epvv8@Ef9|2Xp$K1hn`;fBOhwCL^yVG}m2e zY=>cUF7P^uLc%M!^q0)x+&pT3WQ@4bcJ|!U_aB?+6l_P2{BT#I%)|F?4UVs9TPU!l z>WApfZeHhJDMUw)o>knJ~HTM&ofobwm=;GhkAm zIEzz)X4Yy|0!f_BIlam@BR8OvfJ@41rjU@*>mH4%b2hJf#+I5x%zEAI9z0XV#twad z3h>)=Jq4u77_;erg4nxuJ~f(V+0pA7fp;XkOu``Q6!0?vMK+*oSzRiwbl%dk8g5`A zN{*fLM7?A6**&}{&ESlBh@!hIEG5)IA}MxDRM)iKP2$-o%r>6_9&w^n2dfv3T+=Ct zi4+avvMag5I;7r-msdT=er!=-*=HBte(@Bb#I=QJNQ>i>njfD6yk#d6FT!umjTCa? zE9iS+Z?6P%Ds|uSlpy2Ig{I1(bJbkeLKgXj%KUAfRV_&x+Y0WMqUt8c%)n^6NacM zVaHC{7O#k7!Le`z&Z>a-0s z9=_FF8-Wa84ufIJ>(jdI5Du2gkVX@HU}<%8cFvqk`sFVDLPVR6Pb2}DJ*VXVkZ*pL zUE^;w*5f8ITM^L;P2Yxw*@cNE>#ichzL&`I!JFRNZrwiL#!;UwC!!&rbW2V`z>rP^ zVt)q{c*Hq*T6;WRQ%)lm&;o;qSqneh-%^d9+4$lL5|tYvuf_ZrjWU5JDMl49w>oY` z^W2%kDPUQDI{`KoN0jwVC=Xg;pcZg-o$1HB^S>n^bKgxhdwX%C*l$zRtLaqe zWjCioP1y3XD1N+a%TIr{M*g=#;}5DoGUsc>ttq#$BXZvjE>iK}l=j#a2Bs#foTe+; zRM${=Mh^xTQVuF9qCZreH41_6;!a^>t+g9Y})E_lBxK$GWNkFO4vbxB!71@ z#=A6G5i(3H*`^>#l1Jlr>)snac&A3FXjqhgT@a`7bT*jAU8w{4)Rws0^}6vJ2tER{ zx9Kz`*B76TKiuLjRhyI&m$rT-6G?|y|I>YV# z&bmw-(lDCl{NEXsaqo)sQ%AhBpvR9vgfvC;=17t0$py7l2yGlmd@F{6mW6>_g; zMrH88yHUwJ+r6+;z|8Ao=$sj;FLG`_u)@Tsq&qlNdcE84or`)#-fJ-bGCy{L3*jyw z!GUTZr^9bQVUi1XrSgWo{<`zUkC@IGrjr@SI(O$d^9IgNFtL>nZDOASEW|@%FR$(Xxl*8 z{777;cyPZM_)s#j@(q~n1SBfcW|&>yq)+uSgqdN$0f{hcMu2yPES5L6!M z5|{IB#2_a2R#EQ-+^e2@HTNP+L}l_18}ZA0Oir0;*bIe2A#aD*JMt+Gwzk6(b0ZWw zd_Hn4-N2(+V-@ATuJOCFVk$xc^$f~#uG;EQdH2!t(hq?}cu*1uF`U-kbP70YXV>S- z+djb3tQh`THIp2KC@;*uG6sckcT?O&2bS8>ANnW27jIpZFO&U|aAtQzQ?D*pSf*S& zT~^ixK{#4y5^^~L$RS_d9#<1r@|Eph{;M$X-#dP(d4T$4%ti~F9Bcfyxs`Xz-Wdv$ zlqSuB)_N0%9i_Qs>JA3uPt(-FQPQr_-@iEqF{*U5Z1m@Ee3nUkq&K76%6cStrTK7i zBZ5HIF%xUNYS=egK=fn-gY~Z%rqeP|^#-u_zD&6K*DIpnEGe5-5fF7Or(phLaI$dd zr&kdcFk&m_Y9zAymsq+28LqCfz_fGO;ms_63dUAe1W*mhWDW7H&awX}l9z-AXO)2s$~#RET|81&n;ub`EypUfE;Lp)jnz;9DpfDWx6rf(G~`jY8Bu3sV##N_ z+~>MJ|M0&4Gg5N6aSts3;wC$s+1E?0E;A93KOcCiIq&7j`Y;O{GN_YxOX>DA8!!4+ zY|CTH;t^hdxPl#(7#VBe|9aP(rbjPvXW~l}+IP^fw7CTPrFdD5!`rkwuuYwo%Io&S z_m za^i@fD~k7Uaq?vOG1cS?C5rBI=t-3oZ13^};(_%biDu4sBKQ>n)m0Yza#hstTm@?~ zlkvT3_|!)HkIP9QZ-(GVlK~-9=nFz&qMj(esG||#WLkYD+&)q+D-M!4ua>{;Y?akI*&%(COeHlC@A?EJrp3%srk99;QlZTQRO72Q_q6uqpijt>@e ziMW*M!=dSw$^`++-5vt%vU3#5>!i=;$lc_`Bp@B%dFNbt#`Eju6MH9U5al4LagjaT zi4(mpgEFZKxkC?m@oJD)xAJ>L$`MzT=%wYA#_h-xZY+5UT_lsuT()EF08MtI?+m7@ z7RILyTw9TBhmLrW2?d)O|Pp%!`?Y^xrL9e;d8k+pF5N5ShkA)6YM@&qJ9( zOj%o|f_>H&Dy>wLQcNMI`hPcf3iz@S=kIX4>cIGU@G)Jn6G#z%(we=|3ARVzuhS}s zpV5SWHv8D;N~GpDB+ce?+Lme#j(Ib*_5@$KQmH?w-PIn5+~N|26_hC}8x`zTKhuxh zOw~VF&ujmRDX;`(F$`vQ^dNZTXkaOsk+5gy#5oL>iIJ~;JzE)ipB#3FJjQPk9jujR z+DATO`ZY@8jP6b^aNQ5Dk#OGkes7VgM#`J~;rV2&%lA?X6*Xs;F~T4+()YK+op$@4 z#|8iWKgK0(D_g9hx{6%PR1YO)1Y{Yq-gm*!oEHJXIOgO=#fpJ9c!zD54PLo z`dHfCff8~z7;LMnfzTin6+xyub(7L^9jf=4Lnccqf)MtB#1^g;3*&gp?}M; zsieMJ`fY?G@ygKc_Y>OoIs8^XixJHQ&B{hW6hn_(V@4MGQ@dQGpr1>=G|1OBK5UPy z7sMX5GekDu|CJw=(EciV^V*4LEi_Bv88#H>N#)}vRgSGePJH*fbn0*c2CY>BeN$E4 zr=tc-6-+#Bq?Pmrj>rBG0xE(J_06xOhYA;Z)89sO`qIM6=#vq9XZY z-q+S#2b&(YN{}nH6bB;`qq^)t*@;dn;!~-fC?dMptlPosP$)FT~*;%KW>8zea>x%wK95cx_%Z5di&_?};B%9xf<2 z2Cv50;~#g%TGoPqe9i*vcv9LwRm(n8zD=8Lg7*{=(i-U@?Y{QZnIO)M%afFXYB+Ms zB-|Pbl&RD0u3Oujab=eaFSq8jZ?(BfQQ&yKg~IkhElm9s8g(#Yy23cczSJN|3?EyCsdz zP|G}Hj7r9sm+cv}SjcD&KhQ0pV+sEZlLxcIq1}?zGz-G7hvziYYwlI^vOl2KhmV`I zYl|+siugNCpEaLg8F5A3|1l`Y2K@Pr>N9*O)os~1pgZ&v*x{V1b;*;#@KGdBrNF?;bNjWG8pmZs%0+n z?O@C~7?=?$SUIkRydQnS5+`K!AIRrAxlD`^g(OnRchM}#CC9?~aEvF-7{VO-w1ryj zj3Njb6%4}q%ilgK?41JK2?sK9zLO(_2<)RAsf^=csFz2*1UazF8KoCfoe>p zsh$0i8F@j#n-Ma~j$T789~yP{PX~#?uP_*`ko*pt)MoVn6V^Y5qKD%AE&8GIZITlxo`Y)wlzHwXMj%5}VXp=o>=0r^{m9&PQeU z%&uh3wI46hNWx?&nA5k|aHy(d&Z!$`1*dbNR?ZwGOgXl4EaQ_-VhzfD)6{kGCVMF> zgwfz)CWZaHNamCQiOmue?%oQqU}^dRJ-#@r`-6-NYW*Qn<8Ra481?4@kBk^^MwOXj zDx&NpyIIKw;sb20~<3BG0Nf1YLI>@futvc~Ed z+o}l9Y-ik__ZZU+3N>M#0xVW(DMK;PnV>>#9%VmgE>C@=V^%4tSS-Lx$c3h!VXKAx zwF_Jk-U%O4keC=p7Cn&Jo@C$7Tku$`R}0k0M{mmhA=O*F+lBtfiz0nSmoKNe_eAB* zBQoWp;MH{?eiJWh;VKuUb+Sy4h~wTVd*kceVK_=@Qh;=3GFLj^tUaHMT`=-&M1Yco zi|aH!1}3>fn9R6idneGF-Lh3_9DLE4W}0<*^@36y7BjJsk7ZP;Yn6pGI7PE~{L3s} zhaGFua@&HQNc-Xbq3ybAZv3F4%x1&$6wp@Ptc9f$8Vd&+Q(q12?g!GhIrE?=>>Uqc z-@Pya(EflZG@sP)Ka23*{EqPbU)jqEbm?E(Xcn-n+u=Q@02g*6bnJnDMOBdK$YKbe zQT7+6??KYoR}7o4+o{YQn%VS&W%}U9Uw`?Ob@WS3IXSB=tmLVZ>r)MvHbDbtcZk~A ze3A%idrPf91^h(iwe|e3r8Qq2Wg5QF7e!D+hx*nkxZ925FxtUI^W7b{zm!Mxf7WB& zpSU(!=)k8~dp1?~tmg+PRWwRodQoT;1|wJ+!-pQaHRpLZBO+F_3(Cw|sVizam3XI^ z@#2^@7xlIUi)7$On(cMrx0Or9+N)ep-sE6I6PxTpzs9x_V&WINX!Ui#dDCG1~oq-P5L(9N#RV(EP8Rn^k z%x3#xVpP(>po!pPO>C*ok)1$_gH*?4(?%#Ss|MW{Z&ituu^&8}lYs8d?mR`P>`keG zIm3GnPe&}MKh8noc1Q5(?2WlK7Y!g%79d-3fhnPz>VRZntvL&8=ty2{eZsCm?_6-^ z*bC`BGJ})hQvg!;h*3RH(>nT}*4_YeQJ(&@{Rht%D43erbEPc)qxUNl#;~&~t-^vx znGwR}&3z6YV>4bBZlGWGubj4;E_3{0VDv~NA=-cKX^eN^*9Q-KCLcVZ+sqE#E17-m z!3beY78;7iyu*N~NBYf*Ja@-Ou_aAbHyW;nr&$=M-Ax0tzaG+1dwxIfO%l1_rYiCq zhVjf6$|lSmAK0z+WVp#~ zR!~;`WTVvPpJ~6JT|7z5o0$_#FKN{3(7L~bd1r|XXmKn!gYm>-+ zvW%-d1!&{%tg794B{|I2a9vcXYMTo1Z{2DbXJ{IWWMCb+H6si>9@8y)S0c~Y{<-8t zb61v3El~)s+&30@OK$W=T+(!$V|HJJIM z#k5M4@87I``p{C}D6s>&--@aF{qFEjvxfer zW(Az>LTgdUS=@piljj^0)@3HMfCx9F{(Gk8ec=^p?6c~9HuGG?hr`scCv|bNVRxA* zMQ77>$WT_PMJy#}<w60+vV(41B zVGJ(Zk~)yOV19)Vs!R zOWC@!Nr$sa2_*r$&Pa%dZ3b>CJ1mk*&08QBgfNOeWm9s*mQGPtT^gB#bt94EDcbvr zAWTb*U=^mFzAjKdPb*(0#e^#B`kKr}sP#ECGQH_xrQ$Vf#s*=eAnxp=Ars#p271zt zJ|o13S;x+a&wI^N?9p=upX;9WvhTKPlsG)I@uw>(YsEk0P0yNv`H$XU&wX2KC&GE zxh$RWZw0KYdJJh30L*y8^6#|?nSCtf{bkvD2ITbKs~5bsE?JIyzj2QS%|s&7J29c& z?P&~6Mm2}Yh4N;mw2sL?uQJ+SqP)2X1QZGtKlPN_wHnNtNIT|~{Bj+_%cHziJq9Ga z@DfxH${yYc*&q7QCz4gmA;qg<@_n5FMTkOdm|-QB*NJ#lVYD zyGXECP@#>E!Yfm4+v|<#H3gge)EZ4A1hBzrQ}JV{pYeqMDPZnW+sRcckMrct@4{j|@IwOln1dN~JC z((VYHEx KpC=Jc$Nx8i89Imn literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-9-Financial Growth Comparison.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/images/image-9-Financial Growth Comparison.jpg new file mode 100644 index 0000000000000000000000000000000000000000..05f58cdac4ab882ad731afa2eb7e75d12e59529f GIT binary patch literal 36801 zcmeI52|QG7|Njqzu}@?VqbMTEGL|fp%3dUu5Gs`|B->a=Bs(D#g+!E?TlRgYlzq)^ z$SyOoOpKZT>3)98UH|*}J@@Z<>h*vA_dVlvaprZ#IoJ1^bDj5feb49me#wL631FA5 zrj8~+MFjv<;156^22KFa2*AO5(o+5U0{-JgMGc{$rGwHlFfxH3DB1;3Q$ZlqG!R-^ z8XEA^zTkd3m?y~<2>7i6Op}oJAi?ai<^g+Z~p;N zv4i4r2zdoXrK2ZKYG`U{>*yMtJAc9WqKT>1wd*&mZEWpaT;1F~P@Z0acZ2Q)hlGa3 z#yyBnNPPGxDg9YS=JTxV7cXBI6c!bil$Mp(H#9aix4e1V+TGLJ*FP}$VQ3ONHT`jB zc5Z$FzxsJ?ePfgGW$UY5Q~=~}W`V!|X4rRjv4M6`)6hU@pkM8xqV@m_gpG!FuQVOI zx*_z6^X`2zcj!5e$2`riV-S`-i{rd{yNi)aL~e3F{;O%fTK2~b3;3rj`!B=(+pZyi z8A1hqc@Q=L349@iiKS6f3Z)JH=r!=l&5j;$(BKi^t?}43^CqJ{X@d)U2l9 zQ<1+okhHpEpxx_gR$%*LaKvPU)u(Jq%y~h!*L$GYh77O~O+0bojb6l|0s-BjEs+32 z=F9tH%c$I^_*2sJ*PX(W6e39cqqZc;*M~93(Va6z;k-9*eNj35jP=We7l2ylFt)xAD8Zw8X4hSkgoy(y8`mRS8B??637L!#1*$pw*>U>$TZ?{2y5-&lhDbR+i#i3L6pi_E zh9(IfP^pL2eqXK`?%_tS`1$;iM{6UukftRsiSqb_aJvBol3=CwuIm{Z?6iVewu*4m zb*FGk_H_*XVhq8oflgAUznkV=3wweRFJl=EKzEdRKzYI-f+Q(;@R!$p7Q@BSGiWPZxg}s4gi(~EppatijwuW0V~j6N z0L3nWq=+LhqEW?7Kf6EsaulSRH?WuA2b;_K0UrkI!qFOX-SmAS0_96 zj1U|*-x(%IQ%OO)7!X55eL@U&XE*jhZkBEN^v1^h@(}E~_YPapRY2;(hEh=%f&Cj3%P==vIwdAeTOD5nw`Cqa^2A1 zT>DFP23u+-Y6svj)S#4q&<31KXqb=hWO6E<&IEdouvE&$hf>Exr1Wp;u4V|T&vziE z;%m?oZQiGFZCN3XOM;T}%f(H5RUStKXkGTU&INp_o)eGanl>A9txwboQIg9FhPTh(qJ8 zckT+2Y|Sm}B(_H_>r%wK5lYumb?yys`q>`J|8!v_EhP}(cTaH^-!~}-LmM^}& z;6=EA#pL^ghZC0OxuD|5?<;joi}wz}YZpUPY%7hj-yVuE8BIEVF(o5a+<#v;db+>J za*K%El9p(3E&H_WX0NlhGk!DbnvFbj>_lX#z!yyR&9#o;a9r4GH;d`V%0oRf_UUFQ z-*+t|9hclOVz|?Z_Xeigf~tdMN9OQvOowzr_Ya=b_k<9tD$YP(q6+F$c=4LSS$1Pc zl#l8wp5wAjs4#1%#c-g=t~O8J!G>oJXJAKfF}oLD-3}q?W;w1GHv*)WZX5KMF0*4d2F!m+)wkusi=JSJUU7U=%RyCx zmnQ>ey(-?>?=VBYvFjIwN}v+_mwnO%r}$^9?PD0P)+}z)C7UE1C zsyyqj7D;e>s*drYN;N30G?Os-^yHHx4A-2O89iOKNTp(n^FEIL#e|?{uFKQjUy;P& zB3k-+4>E1*+}Q2}2fYoufu)-Yh_*t^%s#gdWtYgnR2ea+ehxBPsXlQSN|>tKR6BrU zOt#NCFR6O}4HU9pm)*TsiD!aTR8Qn11F^a3=o(Q{xZN@tu(p}KOGwzbl$TPxkk_{9 z!P2tEOI(Z~9B&=xUP~}2z5W2xV|50aMHruy55~fI?D%pAB6H=weB#wJQ?WUPI-~U7 zf1^V;ya7(5BFu_m{!DnOdbzQ#zaPQ}6@Dzw8wx*VoJh3#deS`fz)3RD$yJJ)MrD`? zyK5Q6Usn{`3OpwG2>`+(8m3`KUBZ^9@!KbJ!tSmP1m)pE? z#`VTq*`ntpxh&K_Kk0h%VQW$cLqi$G{E<;il|e0>dtaxkCT={maD3UE<<9v_bg15S z^V|k0&t^@DcVcOYf%N*{`)CB ziS;DfuJ%V+rx@FmvyI(Uj{$KV?xKnq*^+^EJ#)szllPY6^vh$eC9&iLUy0zKQVl*~ zQV1Aoa8S)9Sg3X{S2J4>l$KXmpQ(`STG@Qk6mo8RSSuP0m6|IUmUknR9W+~wINJ&X zv`5@8+vRdlt+4x?Cb&%{iyV zDgAL{d!Bpht6-*e2x4OG3{-vUFt;UPpwY-~bMnkD3>;#97w@|ol2qU3B{q#6!A0dy zNJ{8*%N;1}%Z;$KDVLAY6dAg!(EjS;^u($}dv+RY!{kaM(+ZV6%$fQ4OzQ+qd6AX> zR6ewT)xPs%KZ(v>=E?ET+7$(*F6G*6`;QrP%MX9KS}79-tdH{&O?$54dI)B9d?r0| zS+qCe9|)JHqyYvN(~pxTiB#A<28@JZ{eH}|6g~SJR~M-RX2rFD`^sAb2o1oyJqs^8 z$yJd34xLRnakjMLVW0?x#`7jKy2DBRJeClRi`Kt?X?~2h)c(Ool||GqFh{*&q&RLt z4Dy-~i03|wCq3gRk=~yhU#(H~QvSI03(>O8+C2F%F(92%zP16ix0!Oz{HY5GJ(sk( zowc7y9b2zmev^a7Mcbp-puI>OFBzyWHwYpFtFqudb|4u&XKPVpA&8Vu;VUVKFp_>^ z%xzVRo)qjS19Px!^n@%K2!em}-&I>-Skm|lVo0>(wjhKI&~KEGfn&ZW$N*I;QH%`S z^)@2|X=+c&fPV*xbEVgCr^e{ZE^m)P9Gsw4RI>$HfD@QOW@!&`!OilY*lgDo_&JvJ=UFE0_d9 z5)8kqjl_8Nvu(euRiTZQ{0ZbYRz!bO7mFPY`0zS123q19P4%4#9O&$%#t$Y!1gKdM*N$;4zR^JjG;+N?VcYL(>dAnd z!wKxy?o(a3z$ljP|5qr=PTvn1^l&vKP5acX&rUlm68$+-E;XAab5XkEy9_lrryqem z=-3Z0-`vqQTJcCY`psfCdKkIGRkS>fT#+FIOvJ7tSjfy=de~M^wpe-?@K-3y)!by@ zO`8iDuy7y)PcyyxJP(7H=@;HS^kQ4hnjFRy6WLA%#>Uf?Bz(Y*V;ptS?b}k!mkx{718G{Wf<-gZxu-Ez)8ci z^7F>)oKdhkKkD!%uhyYOud0Tnu=kuUOz!4cL5kD){O+3o-cDj7Tn>9{cD>5(xX+M*=vIHlR*6wHgo#aXtp#7-S}xZ+eT8lY?UIs{4!MHM z8QHuTCQ!b&1)sc-0}rl-E^LhtJ}r-JpE04i?ObtpN>Op~A{oGS$p}{O`bC&jK?XKB zsIVNyU8YtjJBHQswmS;SLcQW|fFQ zmeW>7{_|sdN>0O08Rq+>sbFgi>wURn7A?+xT7A>4{r*$Sq4&3S)(`lrf;unL{iLjI zH9Iaol`EJciQ;gjTHy5yku6#%A`xEi6uf1^dQKl|ZAOD|d+_o2(GG@j$*`8V)W zoiwhXTHZUWspE@qHIxk8KD@Krg4<#q??TdExkCmT_{U8ht*7*yQXBo>z7?*i(&pO*3y^JvLvv zj!ALMvd<~YME!^jUH7KUBg36PhDnO=i{`WaQ=L5r{2`5#|uD?pYRu;vz zPy2)Vegm2psyyz94#Q*cla{E%wU%rMRHeq`_-?cL&DU=V4D2TTZnW!LiO9r_L|gH632kF>$?ZKT6x?)8(M>A#d`r)QV&uKBZJCw}X$6 zNxY5?k91h(>U)09^5C^Gx;{uAG!t;O5?v zKzN>UOEnaib+7QEs!&4xem9Oa;zuVGYidv2qXjdE!W ztx@lz6S>O3Ce0&qr=9-O(aU`EnFP)>yR*qv+$|SUKeT&Dt?TH*PS@2U?_-$L{Z&7o z8zAn-rPgf@aR(LA_92Ep9z=8>x%}8lB#Mqo@r+J~>1TMbsc5i|I##O?ul8JSn+>A) za3n+NmL}aSjfaY{um)g{N`I%{<71Hh!B%14qQO0PJARKLfckXprd5Jc0Zf~g;3eUy*5h|Rbgy99*vDp!{X~ADjP<=u0xQORU&8+&^ z80NE`-K}w!`~EAfOi-;Fq=#E1gqy3{NLrPBl(L#bbl5%%Q^(nZxVl1y73fs z4Jb`@d|_YzRi*fh%U{Lx2b}h=jO0xW43y6mWugZjBewtyEyfE+28iLN#yg(i_nXED zbf^C$=T@FeOX}+Wzv)XgC0*UxQJ+aCtNtbqZUPrG{($rTwIBO`+e0Pl?7#2C{&LyB zGt2vDsPz5+NIb^R5<|WIrtA(k7jQ{Y0nu^zZ{kR48%WcSIPqWJ?fiXT#hHHFo6}gt ze*pA;SK&RY;tBG=;y=W#R|m8xw4M85vTF)$|E!B0zU`kB+Wv`_8Bu8aCtqhlq3xe@ zkrRcs6x#lPx#rgZ1pieibK9T1sFy-p3T-KK+kbJH@&5}!J%zRu+WwZ5KVq9gTMBLe zrC0|PPx5z9eJGyfKSux$3T-L0{Sk!y?J+2{rO=i_+u!zx|Lr!#lT7g>{{Yz%^>2?s zQAz#GAa#`Iw#t9j@Q@TwGR2cj@g!3`$=}fQw`BdjZ3=DwS=#1Xu~aopJDZ#5W=KBi zk&z9Wm$_=BP$SnR{YgAloNvj*W7(v=mr+N`DMcoedVJOwzYI?+}(@HGUOr@`L{9Rwd7g6^<@ zYV-+?@ySa&l-+-|W|^~KK{JEN-V#2|HnGX5&>f+DFsgTuv{K{&$2}s*lJ1@RN*yvl z2*a651hOS`HX00_tJ=|ySA08~vJS3oPYp!QT{j?7r;>q?XQbUoI&^1iDuEVUB*kXI zqyYw_zIsI&S($3Jd#q+s7H7S~{Nig9gUcMjgc{Zp9b3HMHjqUIuo87P&$f3neAt;> zy6QbZJm3dr2YDKg#4(~t`=SY_h+xW+6XNIz4scP921=3x{p)d6BMiL2L>(P zBq{?k&?_Y_ z^*4*cA`#rKuHW!OYeOCkfNx1F}h0GQQ*(}b&<7B9IHB2~D@S331vk$I1qXEv{vU~XWq>v|fidsx1&mCN&S zgX`W*>MoFnA0xMNSkrSlsWcofp_DC)a3dCbOM6lz#V>h!ur5HZO^^!27DCD^>wM=TDdoU zas7$-!kd~=TO-DC%{es0@)`M~h=@dwy-W{|({Z?o z!~$u|t>W!Zuef9`>>eD`6fV2ds(mNW;Y#l{b_ui&!Ffu+vCq+!48X>8AQ7*k(h*k{ znAl?0(pW9Rh6jA$*uyNe4%S-bg_?@&N3S!ijmXFn`sJJ}yE6=*sYD%rnf+E|4LAd_ z$%`g&l7R^H-1+_*9GwB~*{^SF{_|}b>+P_bmo{F-!6MEJL)S;c-!SbPsqadFl)jnC z{_#nD{%DdQ{V0WmbCDbcWsGSa2a?VFSV9lri-Y$>gS zdxx8Kh7#D81s~nOh@`>L1c#{}JS$Tw4WaGq+DK#b*vALZt8}-Vi6^ohD~-cZ$tRK< zEcW-xdA83MK5X8z{f;P#=RxG+nvH|8*1?Y$XvX%Dpn+R_V}unxExi6(b{%T(jjd(K z)*0yjNlWep5_9kMny!p6JY1ajylH)^ zj#Qjei~l_8ofQTq#rmM4LCk(`bRa)XTrJq^Yv(NzepS=>6j!&s#}#oY+wM0sFo zhWz4Qwb|Pg;}t_WwP~UZga#fGkAg{2BPUn3NMJCe;}|zEqoeg%d`Wmlzk{2Zo=plz zgqFCplHouYfX5Kbs=>HVPh)3EglIAl`YIRM?A4YSF^Sh3mD`GK6^R*!wDHs#MKz)F z78{XNZU(FbzwL{q{Y%%vx_3+LYn~`pW?a<;HpU~oSW-g@RgqY!yA`>go}I5c&;MeM zXFG5;H4TF#I8S`=almdnC^H|LQgavyBv?UX)>WbX707@T@nTykA}lxNKw^%>p}ieU zdwC?5Oa0Qk2K4gGhJ4rvzMffwBd>-w3M5|Y)A3&A1-2c)NsBK6aJJ-AvF0knXr;~H z%y~8sFN?I=#t89EID4TbHGy$eRo37J{^aWAHW|pCgLCRoMMI*>wWQTp z9g(SUCL*IvV`^z9FT|06J z!iJrd#?SLG&N9m-j~&9Dwol*b)2A158LLdM+FGgeqa~zH=70%aaZi+ui_F_>G;8%` zIO9Z^H?B{q+SiWqDO__CZ@s0yWJ6W>4{})llvGEHmNY)XS6*0-6Ov6&Iw)(IFT7@c z5x3{HmT#K1V1{w*Kn3d1PV$C5UargMwMy(g;gNl0KpdF%vBD=e2@3TcAhq>6%2gpQ zhFmFf*|WT4(jtXw*t+YYlQU{3UsvPpNru~VH)DIV?^*Z9DKcoykMu%b;(qL&M zW*;p)JcMW@7?j-sxu|}viB0Id@Qu#OXgjkpV4U~>oS0*H z(MbceQ}8bW;B5y4pHVtTZ7d_UvSj}r z%+}yu=>!Z0RmrA>62PA_Wei zcecHN&*n}oq=R_Dl$!|OtM(OV>xd=iA;M257}sGzrd3%{D!kb~NBgG*FNaj!jeHBZ z-1(Fhvt*FzX&FJ3PzY4fkq>uv<9wgLPgZA_kWYi+Oo6L|>B{kbNrWCRvxod^hFv?# zaI0>A7Au{K&h_wIsV$jA`EH(3-xlX3uYuxu-Bf5b``efApZe(KiR}`GCG8iE-kug{ z8M$L_gtBa?VRtHog{WOKG&{lOO)z^V!|Xr)>Ap)!8Y8YFv|28%c}yw=&l65B%aY_J zg!C4Cj(qfQM#nD5rdKY5;swfD?hig_RP5)!-f=+1VnpmttO6IwvU@P++u^c4#- zaAQ8C>9RKV?9#Jm(wFrO(rPi{d5QHY)OhZQJZf7mR$O3kdE9i6+AFUeI)fs>o4Xz@ zxL~Cd)FP4Qc+4p}<>>qE6r}Wg*enJC9 zHmO`H`>{#}frr?)wF^=PhlVVagZ75h8oq52U{D(|?hriCIo{wWT-`0Wu;+R4{j4WU zju*%PWI>q^c#CnxTN0B+gOQ7c=(ffB#csEnkfZvUIBy+Xg#2bswn@2M>`|c*f3v47 zsaSz*NZZ%=CG*t+=(n1<%^kD`h;m~iJT3-*+z41r~vnK1c zLf2whVZ(dD44qA0(aZd3Y;jM`Vl8^k!NMTZe1BB_bq_Hser~qbIxP_iYUhKyq4zu7xje&E%XW z_Z0i9GTfZrV#FnFjI$8W;Z$oHs+8+V!d-3>la5H#3+5a04vhf}_XSB0sAf|n387d6 z-uL4}iA$0$@rA+Dp=eezU=(*L7YunP3a3TSY1e?>QxxbuU3VEn;E@gHMh&-z2-XeU z?tIDGL81vFXE@*wuW65J6#?LQo8dsh<|HgqE{FXzLCaNRyHO4W-ju7gF-$^%0Y@}k ztD*SkPbZvq9mu*$6Y}o0x4WW=QP1F(VBh?v;C{kEtZh_gj?I9(LbpJ7`k^r13Nn!6 zkhSQ48nE_jzhw^Rv*Q-yG1aTg5P8CUr5 zqAGQ8D@}qdww#XFF2o8MoLP(@F+cl^VaT;)C4?K|ic9EnGuu91s`p;atLh8BWkVgm z9^anqs~SHo$?~4nDpr9i$sI5h}1Kmg>H=$IWgOlK%^`K5SI?%BDzB65FL6hpr#%M9z|wwaGVqMAM2!p8 zV1UyKPrlGQS)MR5tfK4?Z}M8Zg}K%$+H}Tkn^0uHX6JR6agn8Oq#(VNDxe^Nx7o8U zjd=@Xw(77J8Hh9KB?F5E=(#=&=_>pEm0|bPc851xC>HccxojZlyd(J8!-w&T@JNqT_Uc39^InlHK`J7`-D``G z4fh0Es=+9FBchO?;_O)Ic#xW0j}Hs>QPF9ArWa}67w$^FmXOD`qC@>#1dB#Si0T9d zPdCZFmOuor=%aUUmT5e1h>DPbyfhdr$=30J=ku89`1To(t0vJeL~~`TDoDGd18UYp zEWw9eDrlltrT4{@)>+f4RQ)??uVX!I_FV}GF=&N}rftem^@718i59|y$mgc*PnRdH zT*OoEiVFFma#HzJw4`eA>>HBP!?rC;zhJJEtiYQJzA(PP6*M39xI;AED$Zi*?c8Gx z);3382TFLz?7VnsAOt~QIpZ&D^OGV^iTy8*OC>bT l&&MDVtFN>26MOG(d-A`sT@f`}rr7Qb!r^D%gs8|L{trg}8{q%| literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/manifest.json b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/manifest.json new file mode 100644 index 000000000..153fdffe2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/manifest.json @@ -0,0 +1,65 @@ +{ + "version": "2.0", + "job_id": "job_840184a62587", + "data_id": null, + "source_file_name": "NVDA-F1Q27-Quarterly-Presentation-FINAL.pdf", + "processing_date": "2026-06-16T03:25:59.990364Z", + "processing": { + "page_count": 17, + "billing_status": "charged", + "cost": { + "micro_dollars": 25500, + "credits": 0.0255 + }, + "timing": { + "started_at": "2026-06-16T03:24:55.507977+00:00", + "completed_at": "2026-06-16T03:25:59.984711+00:00", + "duration_ms": 64476 + }, + "stages": { + "timing_ms": { + "document.profile": 5113, + "pdf.extract.standard": 44922, + "toc.detect_range_llm": 991, + "md.detect_toc": 1020, + "heading.merge_pre_pass_llm": 1193, + "heading.hierarchy_llm_call": 2280, + "heading.hierarchy_llm": 2288, + "md.predict_headings": 3599, + "md.deferred_summaries": 8986, + "md.build_dataframe": 1, + "pdf.parse_md": 14089, + "document.cleanup_unreferenced_images": 0, + "document.compress_images": 291, + "worker.parse.document": 64423 + }, + "token_usage": { + "prompt_tokens": 4543, + "completion_tokens": 317, + "total_tokens": 4860 + } + } + }, + "statistics": { + "total_chunks": 39, + "text_chunks": 12, + "image_chunks": 11, + "table_chunks": 16, + "total_pages": null + }, + "HIERARCHY": { + "Investor Presentation Q1 FY27": { + "Revenue \\$ in billions": {}, + "Gross Margin": {}, + "Operating Expenses": {}, + "Free Cash Flow and Capital Return": {}, + "GAAP P&L Financial Measures \\$ in billions, except EPS": {}, + "Non-GAAP P&L Financial Measures \\$ in billions, except EPS": {}, + "Q2 FY27 Outlook – Non-GAAP": {}, + "Reconciliation of Non-GAAP to GAAP Financial Measures": {}, + "Reconciliation of Non-GAAP to GAAP Financial Measures_2": {}, + "Reconciliation of Non-GAAP to GAAP Financial Measures (contd.)": {}, + "Reconciliation of Non-GAAP to GAAP Financial Measures (contd.)_2": {} + } + } +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-0 Revenue Forecast.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-0 Revenue Forecast.html new file mode 100644 index 000000000..ca7d889a6 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-0 Revenue Forecast.html @@ -0,0 +1,18 @@ + + + + + + + + + + + + + + + + + +
PeriodValue
Q1 FY26$44.1B
Q1 FY27$81.6B
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-1 Revenue Forecast.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-1 Revenue Forecast.html new file mode 100644 index 000000000..092179859 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-1 Revenue Forecast.html @@ -0,0 +1,18 @@ + + + + + + + + + + + + + + + + + +
PeriodValue
Q1 FY26$17.6B
Q1 FY27$37.9B
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-10 Q1 FY27 Results.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-10 Q1 FY27 Results.html new file mode 100644 index 000000000..e20a4bc89 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-10 Q1 FY27 Results.html @@ -0,0 +1 @@ +
Q1 FY27YoY %Q1 FY26
Revenue$81.685%$44.1
Gross Margin %75.0%+14.2 pts60.8%
Operating Expenses$7.449%$5.0
Operating Income$53.8147%$21.8
Net Income$45.5139%$19.1
Diluted EPS$1.87140%$0.78
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-11 Q2 FY27 Guidance.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-11 Q2 FY27 Guidance.html new file mode 100644 index 000000000..18066e517 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-11 Q2 FY27 Guidance.html @@ -0,0 +1 @@ +
Q2 FY27
Revenue$91.0 billion*+ / - 2%
Gross Margin %75.0%+ / - 50 bps
Operating Expenses$8.3 billionApproximately
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-12 Q1 FY27 GAAP.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-12 Q1 FY27 GAAP.html new file mode 100644 index 000000000..fc93a7e81 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-12 Q1 FY27 GAAP.html @@ -0,0 +1 @@ +
Q1 FY27Non-GAAPAcquisition-Related and Other Costs (A)Other (B)Tax Impact of AdjustmentsGAAP
Gross margin ($ in million)$61,232(47)(28)$61,157
Gross margin %75.0%(0.1)74.9%
Operating expense ($ in million)$7,449172$7,621
Operating income ($ in million)$53,783(219)(28)$53,536
Net income ($ in million)$45,548(219)15,882(2,890)$58,321
Diluted shares (millions)24,39124,391
Diluted EPS$1.87(0.01)0.65(0.12)$2.39
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-13 Q1 FY26 Financials.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-13 Q1 FY26 Financials.html new file mode 100644 index 000000000..4fbf310b8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-13 Q1 FY26 Financials.html @@ -0,0 +1 @@ +
Q1 FY26Non-GAAPAcquisition-Related and Other Costs (A)Other (B)Tax Impact of AdjustmentsGAAP
Gross margin ($ in million)$26,794(123)(3)$26,668
Gross margin %60.8%(0.3)60.5%
Operating expense ($ in million)$4,99337$5,030
Operating income ($ in million)$21,801(160)(3)$21,638
Net income ($ in million)$19,094(160)(179)20$18,775
Diluted shares (millions)24,61124,611
Diluted EPS$0.78(0.01)(0.01)$0.76
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-14 Q2 FY27 Outlook.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-14 Q2 FY27 Outlook.html new file mode 100644 index 000000000..6033c084f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-14 Q2 FY27 Outlook.html @@ -0,0 +1 @@ +
($ in Millions)Q2 FY27 Outlook
Non-GAAP gross margin75.0%
Impact of acquisition-related costs and other costs(0.1%)
GAAP gross margin74.9%
Non-GAAP operating expenses$8,300
Acquisition-related costs and other costs200
GAAP operating expenses$8,500
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-15 Free Cash Flow.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-15 Free Cash Flow.html new file mode 100644 index 000000000..a7dd8feb3 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-15 Free Cash Flow.html @@ -0,0 +1 @@ +
($ in Millions)Free Cash FlowPurchases Related to Property and Equipment and Intangible AssetsPrincipal Payments on Property and Equipment and Intangible AssetsNet Cash Provided by Operating Activities
Q1 FY2027$48,554(1,757)(33)$50,344
Q1 FY 2026$26,135(1,227)(52)$27,414
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-2 Revenue Forecast.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-2 Revenue Forecast.html new file mode 100644 index 000000000..f32b3e53a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-2 Revenue Forecast.html @@ -0,0 +1,18 @@ + + + + + + + + + + + + + + + + + +
PeriodValue
Q1 FY26$21.5B
Q1 FY27$37.4B
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-3 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-3 Quarterly Values.html new file mode 100644 index 000000000..4ed07b86f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-3 Quarterly Values.html @@ -0,0 +1,18 @@ + + + + + + + + + + + + + + + + + +
PeriodValue (%)
Q1 FY2660.5
Q1 FY2774.9
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-4 Quarterly Values.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-4 Quarterly Values.html new file mode 100644 index 000000000..64a35f7c8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-4 Quarterly Values.html @@ -0,0 +1,18 @@ + + + + + + + + + + + + + + + + + +
PeriodValue (%)
Q1 FY2660.8
Q1 FY2775.0
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-5 Revenue Forecast.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-5 Revenue Forecast.html new file mode 100644 index 000000000..13666ba3e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-5 Revenue Forecast.html @@ -0,0 +1,18 @@ + + + + + + + + + + + + + + + + + +
PeriodValue ($B)
Q1 FY265.0
Q1 FY277.6
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-6 Revenue Forecast.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-6 Revenue Forecast.html new file mode 100644 index 000000000..82e96a95f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-6 Revenue Forecast.html @@ -0,0 +1,18 @@ + + + + + + + + + + + + + + + + + +
PeriodValue ($B)
Q1 FY265.0
Q1 FY277.4
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-7 Revenue Forecast.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-7 Revenue Forecast.html new file mode 100644 index 000000000..11a284724 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-7 Revenue Forecast.html @@ -0,0 +1,18 @@ + + + + + + + + + + + + + + + + + +
PeriodValue
Q1 FY26$26.1B
Q1 FY27$48.6B
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-8 Dividends & Buybacks.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-8 Dividends & Buybacks.html new file mode 100644 index 000000000..6d83f9aa2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-8 Dividends & Buybacks.html @@ -0,0 +1,21 @@ + + + + + + + + + + + + + + + + + + + + +
PeriodDividendsShare Repurchases
Q1 FY26~$14.0B-
Q1 FY27-~$20.0B
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-9 Q1 FY27 Results.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-9 Q1 FY27 Results.html new file mode 100644 index 000000000..dc043ef29 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-presentation/tables/table-9 Q1 FY27 Results.html @@ -0,0 +1 @@ +
Q1 FY27YoY %Q1 FY26
Revenue$81.685%$44.1
Gross Margin %74.9%+14.4 pts60.5%
Operating Expenses$7.652%$5.0
Operating Income$53.5147%$21.6
Net Income$58.3211%$18.8
Diluted EPS$2.39214%$0.76
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/chunks.json b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/chunks.json new file mode 100644 index 000000000..e689980c4 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/chunks.json @@ -0,0 +1,8228 @@ +{ + "chunks": [ + { + "chunk_id": "fa2f820e-6a52-5366-9655-85a5ae3fac5a", + "type": "image", + "content": "\n☒ QUARTERLY REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 NVIDIA.0\n[images/image-1-☒ QUARTERL.jpg]\n", + "path": "images/image-1-☒ QUARTERL.jpg", + "metadata": { + "length": 132, + "summary": "image-1\n☒ QUARTERLY REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 NVIDIA.0", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "images/image-1-☒ QUARTERL.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "955dbc28-dd00-578c-a1b9-ff24c41246b4", + "type": "image", + "content": "\nThe Nasdaq Global Select Market Accelerated filer0\n[images/image-2-The Nasdaq.jpg]\n", + "path": "images/image-2-The Nasdaq.jpg", + "metadata": { + "length": 84, + "summary": "image-2\nThe Nasdaq Global Select Market Accelerated filer0", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "images/image-2-The Nasdaq.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "6499ea7b-b724-5e19-88e1-ec0cc484f474", + "type": "image", + "content": "\nIndicate by check mark whether the registrant has submitted electronically every Interactive Data File required to be submitted pursuant to Rule 405 of Regulation S-T ( $§232.405$ of this chapter) during the preceding 12 months (or for such shorter period that the registrant was required to submit such files). Yes ☑ No ☐ Non-accelerated filer0\n[images/image-3-Indicate b.jpg]\n", + "path": "images/image-3-Indicate b.jpg", + "metadata": { + "length": 379, + "summary": "image-3\nIndicate by check mark whether the registrant has submitted electronically every Interactive Data File required to be submitted pursuant to Rule 405 of Regulation S-T ( $§232.405$ of this chapter) during the preceding 12 months (or for such shorter period that the registrant was required to submit such files). Yes ☑ No ☐ Non-accelerated filer0", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "images/image-3-Indicate b.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "8de6e618-1f4b-59ce-8aaa-c65d90cd18ee", + "type": "image", + "content": "\nLarge accelerated filer Smaller reporting company0\n[images/image-4-Large acce.jpg]\n", + "path": "images/image-4-Large acce.jpg", + "metadata": { + "length": 84, + "summary": "image-4\nLarge accelerated filer Smaller reporting company0", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "images/image-4-Large acce.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "1c97bf51-5a1d-58af-9742-38600515c014", + "type": "image", + "content": "\nAccelerated filer Emerging growth company0\n[images/image-5-Accelerate.jpg]\n", + "path": "images/image-5-Accelerate.jpg", + "metadata": { + "length": 76, + "summary": "image-5\nAccelerated filer Emerging growth company0", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "images/image-5-Accelerate.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "a8a2d756-46ac-5fb6-9035-1e6e3382106f", + "type": "image", + "content": "\nNon-accelerated filer If an emerging growth company, indicate by check mark if the registrant has elected not to use the extended transition period for complying with any new or revised financial accounting standards provided pursuant to Section 13(a) of the Exchange Act. □0\n[images/image-6-Non-accele.jpg]\n", + "path": "images/image-6-Non-accele.jpg", + "metadata": { + "length": 309, + "summary": "image-6\nNon-accelerated filer If an emerging growth company, indicate by check mark if the registrant has elected not to use the extended transition period for complying with any new or revised financial accounting standards provided pursuant to Section 13(a) of the Exchange Act. □0", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "images/image-6-Non-accele.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "cb977c3b-f874-5d6f-a90d-ff3a1d30e9b8", + "type": "table", + "content": "
Item 1.Financial Statements (Unaudited)
a) Condensed Consolidated Statements of Income for the three months ended April 26, 2026 and April 27, 20253
b) Condensed Consolidated Statements of Comprehensive Income for the three months ended April 26, 2026 and April 27, 20254
c) Condensed Consolidated Balance Sheets as of April 26, 2026 and January 25, 20265
d) Condensed Consolidated Statements of Shareholders’ Equity for the three months ended April 26, 2026 and April 27, 20256
e) Condensed Consolidated Statements of Cash Flows for the three months ended April 26, 2026 and April 27, 20257
f) Notes to Condensed Consolidated Financial Statements8
Item 2.Management’s Discussion and Analysis of Financial Condition and Results of Operations23
Item 3.Quantitative and Qualitative Disclosures About Market Risk30
Item 4.Controls and Procedures31
Part II : Other Information
Item 1.Legal Proceedings31
Item 1A.Risk Factors31
Item 2.Unregistered Sales of Equity Securities and Use of Proceeds38
Item 5.Other Information39
Item 6.Exhibits40
Signature41
", + "path": "tables/table-0 Financial Report.html", + "metadata": { + "length": 1611, + "summary": "table-1\nTable of contents for a quarterly financial report (Form 10-Q) covering financial statements, management discussion, market risk, controls, legal proceedings, and exhibits.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-0 Financial Report.html", + "keywords": [ + "Financial Statements", + "Management Analysis", + "SEC Filing" + ], + "tokens": [] + } + }, + { + "chunk_id": "f5d9f468-07f9-5707-af87-1410cd1c8c4b", + "type": "text", + "content": "UNITED STATES\nSECURITIES AND EXCHANGE COMMISSION\nWashington, D.C. 20549\nFORM 10-Q\n☒ QUARTERLY REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934\nFor the quarterly period ended April 26, 2026\nOR\n☐ TRANSITION REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934\nCommission File Number: 0-23985\n\n☒ QUARTERLY REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 NVIDIA.0\n[images/image-1-☒ QUARTERL.jpg]\n\nNVIDIA.\nNVIDIA CORPORATION\n(Exact name of registrant as specified in its charter)\nDelaware\n(State or other jurisdiction of incorporation or organization)\n94-3177549\n(I.R.S. Employer Identification No.)\n2788 San Tomas Expressway, Santa Clara, California (Address of principal executive offices)\n95051\n(Zip Code)\n(408) 486-2000\n(Registrant's telephone number, including area code)\nN/A\n(Former name, former address and former fiscal year, if changed since last report)\nSecurities registered pursuant to Section 12(b) of the Act:\nTitle of each class\nCommon Stock, \\$0.001 par value per share\nTrading Symbol(s)\nNVDA\nName of each exchange on which registered\nThe Nasdaq Global Select Market\nIndicate by check mark whether the registrant (1) has filed all reports required to be filed by Section 13 or 15(d) of the Securities Exchange Act of 1934 during the preceding 12 months (or for such shorter period that the registrant was required to file such reports), and (2) has been subject to such filing requirements for the past 90 days. Yes ☑ No □\nIndicate by check mark whether the registrant has submitted electronically every Interactive Data File required to be submitted pursuant to Rule 405 of Regulation S-T ( $§232.405$ of this chapter) during the preceding 12 months (or for such shorter period that the registrant was required to submit such files). Yes ☑ No ☐\nIndicate by check mark whether the registrant is a large accelerated filer, an accelerated filer, a non-accelerated filer, a smaller reporting company, or an emerging growth company. See the definitions of “large accelerated filer,” “accelerated filer,” “smaller reporting company,” and “emerging growth company” in Rule 12b-2 of the Exchange Act.\nLarge accelerated filer\n\nThe Nasdaq Global Select Market Accelerated filer0\n[images/image-2-The Nasdaq.jpg]\n\nAccelerated filer\n\nIndicate by check mark whether the registrant has submitted electronically every Interactive Data File required to be submitted pursuant to Rule 405 of Regulation S-T ( $§232.405$ of this chapter) during the preceding 12 months (or for such shorter period that the registrant was required to submit such files). Yes ☑ No ☐ Non-accelerated filer0\n[images/image-3-Indicate b.jpg]\n\nNon-accelerated filer\n\nLarge accelerated filer Smaller reporting company0\n[images/image-4-Large acce.jpg]\n\nSmaller reporting company\n\nAccelerated filer Emerging growth company0\n[images/image-5-Accelerate.jpg]\n\nEmerging growth company\n\nNon-accelerated filer If an emerging growth company, indicate by check mark if the registrant has elected not to use the extended transition period for complying with any new or revised financial accounting standards provided pursuant to Section 13(a) of the Exchange Act. □0\n[images/image-6-Non-accele.jpg]\n\nIf an emerging growth company, indicate by check mark if the registrant has elected not to use the extended transition period for complying with any new or revised financial accounting standards provided pursuant to Section 13(a) of the Exchange Act. □\nIndicate by check mark whether the registrant is a shell company (as defined in Rule 12b-2 of the Exchange Act). Yes □ No ☒\nThe number of shares of common stock, \\$0.001 par value, outstanding as of May 15, 2026, was 24.2 billion.\nNVIDIA Corporation\nForm 10-Q\nFor the Quarter Ended April 26, 2026\n\n[tables/table-0 Financial Report.html]", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf", + "metadata": { + "length": 3844, + "summary": "This is a Form 10-Q quarterly report filed by NVIDIA Corporation with the U.S. Securities and Exchange Commission for the quarterly period ended April 26, 2026. The document includes company identification details such as its Delaware incorporation, IRS Employer Identification Number, principal executive offices in Santa Clara, California, and its common stock trading under the symbol NVDA on the Nasdaq Global Select Market. It indicates NVIDIA is a large accelerated filer and not a shell company, with 24.2 billion shares outstanding as of May 15, 2026. The filing covers required financial statements and disclosures under the Securities Exchange Act of 1934.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "UNITED", + "STATES", + "SECURITIES", + "AND", + "EXCHANGE", + "COMMISSION", + "Washington", + "20549", + "FORM", + "10", + "QUARTERLY", + "REPORT", + "PURSUANT", + "TO", + "SECTION", + "13", + "OR", + "15", + "OF", + "THE", + "ACT", + "1934", + "For", + "quarterly", + "period", + "ended", + "April", + "26", + "2026", + "TRANSITION", + "Commission", + "File", + "Number", + "23985", + "NVIDIA.0", + "NVIDIA", + "CORPORATION", + "Exact", + "registrant", + "charter", + "Delaware", + "State", + "jurisdiction", + "incorporation", + "organization", + "94", + "3177549", + "Employer", + "Identification", + "No", + "2788", + "San", + "Tomas", + "Expressway", + "Santa", + "Clara", + "California", + "Address", + "principal", + "executive", + "offices", + "95051", + "Zip", + "Code", + "408", + "486", + "2000", + "Registrant", + "telephone", + "number", + "including", + "area", + "code", + "Former", + "address", + "fiscal", + "year", + "changed", + "report", + "Securities", + "registered", + "pursuant", + "Section", + "12", + "Act", + "Title", + "class", + "Common", + "Stock", + "0.001", + "par", + "share", + "Trading", + "Symbol", + "NVDA", + "Name", + "exchange", + "The", + "Nasdaq", + "Global", + "Select", + "Market", + "Indicate", + "check", + "mark", + "filed", + "reports", + "required", + "Exchange", + "preceding", + "months", + "shorter", + "file", + "subject", + "filing", + "requirements", + "past", + "90", + "days", + "Yes", + "submitted", + "electronically", + "Interactive", + "Data", + "Rule", + "405", + "Regulation", + "232.405", + "chapter", + "submit", + "files", + "large", + "accelerated", + "filer", + "smaller", + "reporting", + "company", + "emerging", + "growth", + "See", + "definitions", + "12b", + "Large", + "Accelerated", + "filer0", + "Non", + "Smaller", + "company0", + "Emerging", + "If", + "elected", + "extended", + "transition", + "complying", + "revised", + "financial", + "accounting", + "standards", + "provided", + "shell", + "defined", + "shares", + "common", + "stock", + "outstanding", + "May", + "24.2", + "billion", + "Corporation", + "Form", + "Quarter", + "Ended" + ], + "keywords": [ + "NVIDIA", + "SEC Filing", + "Quarterly Report" + ], + "connect_to": [ + { + "target": "fa2f820e-6a52-5366-9655-85a5ae3fac5a", + "relation": "embeds", + "ref": "[images/image-1-☒ QUARTERL.jpg]", + "position": { + "start": 444, + "end": 475 + } + }, + { + "target": "955dbc28-dd00-578c-a1b9-ff24c41246b4", + "relation": "embeds", + "ref": "[images/image-2-The Nasdaq.jpg]", + "position": { + "start": 2265, + "end": 2296 + } + }, + { + "target": "6499ea7b-b724-5e19-88e1-ec0cc484f474", + "relation": "embeds", + "ref": "[images/image-3-Indicate b.jpg]", + "position": { + "start": 2663, + "end": 2694 + } + }, + { + "target": "8de6e618-1f4b-59ce-8aaa-c65d90cd18ee", + "relation": "embeds", + "ref": "[images/image-4-Large acce.jpg]", + "position": { + "start": 2770, + "end": 2801 + } + }, + { + "target": "1c97bf51-5a1d-58af-9742-38600515c014", + "relation": "embeds", + "ref": "[images/image-5-Accelerate.jpg]", + "position": { + "start": 2873, + "end": 2904 + } + }, + { + "target": "a8a2d756-46ac-5fb6-9035-1e6e3382106f", + "relation": "embeds", + "ref": "[images/image-6-Non-accele.jpg]", + "position": { + "start": 3207, + "end": 3238 + } + }, + { + "target": "cb977c3b-f874-5d6f-a90d-ff3a1d30e9b8", + "relation": "embeds", + "ref": "[tables/table-0 Financial Report.html]", + "position": { + "start": 3791, + "end": 3829 + } + } + ] + } + }, + { + "chunk_id": "9fa09e5b-feab-511c-93b3-2c6f9f297e30", + "type": "text", + "content": "Investors and others should note that we announce material financial information to our investors using our investor relations website, press releases, SEC filings and public conference calls and webcasts. We also use the following social media channels as a means of disclosing information about the company, our products, our planned financial and other announcements and attendance at upcoming investor and industry conferences, and other matters, and for complying with our disclosure obligations under Regulation FD:\nNVIDIA Corporate Blog (blogs.nvidia.com/)\nNVIDIA Technical Blog (developer.nvidia.com/blog/)\nNVIDIA LinkedIn (linkedin.com/company/nvidia)\nNVIDIA Facebook (facebook.com/nvidia)\nNVIDIA Instagram (instagram.com/nvidia)\nNVIDIA X (x.com/nvidia)\nNVIDIA Investor Relations (investor.nvidia.com)\nNVIDIA YouTube (YouTube.com/nvidia)\nThe information we post through these social media channels may be deemed material. Accordingly, investors should monitor these channels, in addition to following our press releases, SEC filings and public conference calls and webcasts. This list may be updated from time to time. The information we post through these channels is not a part of this Quarterly Report on Form 10-Q.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Where You Can Find More Information", + "metadata": { + "length": 1227, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Investors", + "note", + "announce", + "material", + "financial", + "information", + "investors", + "investor", + "relations", + "website", + "press", + "releases", + "SEC", + "filings", + "public", + "conference", + "calls", + "webcasts", + "We", + "social", + "media", + "channels", + "means", + "disclosing", + "company", + "products", + "planned", + "announcements", + "attendance", + "upcoming", + "industry", + "conferences", + "matters", + "complying", + "disclosure", + "obligations", + "Regulation", + "FD", + "NVIDIA", + "Corporate", + "Blog", + "blogs", + "nvidia", + "Technical", + "developer", + "blog", + "LinkedIn", + "linkedin", + "Facebook", + "facebook", + "Instagram", + "instagram", + "Investor", + "Relations", + "YouTube", + "The", + "post", + "deemed", + "Accordingly", + "monitor", + "addition", + "This", + "list", + "updated", + "time", + "part", + "Quarterly", + "Report", + "Form", + "10" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9545a6d2-95db-5275-a8be-5392be22e9be", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c19796ac-7ecc-5937-9d80-bcfac6375451", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
Revenue$ 81,615$ 44,062
Cost of revenue20,45817,394
Gross profit61,15726,668
Operating expenses
Research and development6,3213,989
Sales, general and administrative1,3001,041
Total operating expenses7,6215,030
Operating income53,53621,638
Interest income540515
Interest expense(102)(63)
Other income (expense), net15,929(180)
Total other income, net16,367272
Income before income tax69,90321,910
Income tax expense11,5823,135
Net income$ 58,321$ 18,775
Net income per share:
Basic$ 2.40$ 0.77
Diluted$ 2.39$ 0.76
Weighted average shares used in per share computation:
Basic24,28624,441
Diluted24,39124,611
", + "path": "tables/table-1 Financial Results.html", + "metadata": { + "length": 1473, + "summary": "table-2\nThe table shows financial data for three months ending Apr 26, 2026 and Apr 27, 2025. Revenue increased from $44,062 to $81,615, net income rose from $18,775 to $58,321, and operating income grew from $21,638 to $53,536.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-1 Financial Results.html", + "keywords": [ + "Revenue", + "Net Income", + "Operating Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "4b3fc8ff-2efb-5df6-9180-6d26431f3e7c", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
Net income$ 58,321$ 18,775
Other comprehensive income (loss), net of tax
Available-for-sale securities:
Net change in unrealized gain (loss)(78)139
Cash flow hedges:
Net change in unrealized gain3719
Other comprehensive income (loss), net of tax(41)158
Total comprehensive income$ 58,280$ 18,933
", + "path": "tables/table-2 Comprehensive Income.html", + "metadata": { + "length": 708, + "summary": "table-3\nThe table shows net income of $58,321 for Apr 26, 2026 vs $18,775 for Apr 27, 2025. Other comprehensive loss was $41 vs gain of $158, resulting in total comprehensive income of $58,280 vs $18,933.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-2 Comprehensive Income.html", + "keywords": [ + "Net income", + "OCI", + "Comprehensive income" + ], + "tokens": [] + } + }, + { + "chunk_id": "699d3faa-22e1-57a5-a7d2-7f3883789334", + "type": "table", + "content": "
Apr 26, 2026Jan 25, 2026
Assets
Current assets:
Cash and cash equivalents$ 13,237$ 10,605
Marketable debt securities37,09839,065
Marketable equity securities30,23712,886
Accounts receivable, net40,71038,466
Inventories25,79721,403
Prepaid expenses and other current assets3,9163,180
Total current assets150,995125,605
Property and equipment, net12,40310,383
Operating lease assets4,2582,867
Goodwill20,89420,832
Intangible assets, net3,1203,306
Deferred income tax assets11,70713,258
Non-marketable securities43,36422,251
Other assets12,7338,301
Total assets$ 259,474$ 206,803
Liabilities and Shareholders’ Equity
Current liabilities:
Accounts payable$ 13,097$ 9,812
Accrued and other current liabilities29,78721,352
Short-term debt1,000999
Total current liabilities43,88432,163
Long-term debt7,4707,469
Long-term operating lease liabilities3,8782,572
Other long-term liabilities8,7687,306
Total liabilities64,00049,510
Commitments and contingencies
Shareholders’ equity:
Preferred stock
Common stock2424
Additional paid-in capital10,27510,118
Accumulated other comprehensive income137178
Retained earnings185,038146,973
Total shareholders’ equity195,474157,293
Total liabilities and shareholders’ equity$ 259,474$ 206,803
", + "path": "tables/table-3 Balance Sheet.html", + "metadata": { + "length": 2499, + "summary": "table-4\nThe balance sheet shows total assets of $259,474 as of Apr 26, 2026, up from $206,803. Liabilities are $64,000 and shareholders' equity is $195,474.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-3 Balance Sheet.html", + "keywords": [ + "Assets", + "Liabilities", + "Equity" + ], + "tokens": [] + } + }, + { + "chunk_id": "dcfbc1d7-e729-550b-844c-f366db9c0c01", + "type": "table", + "content": "
Common Stock OutstandingAdditional Paid-in CapitalAccumulated Other Comprehensive IncomeRetained EarningsTotal Shareholders’ Equity
SharesAmount
(In millions, except per share data)
Balances as of Jan 25, 202624,304$ 24$ 10,118$ 178$ 146,973$ 157,293
Net income58,32158,321
Other comprehensive loss(41)(41)
Issuance of common stock37515515
Tax withholding related to common stock(12)(2,129)(2,129)
Shares repurchased(108)(157)(20,013)(20,170)
Cash dividends declared and paid ($0.01 per common share)(243)(243)
Stock-based compensation1,9281,928
Balances as of Apr 26, 202624,221$ 24$ 10,275$ 137$ 185,038$ 195,474
Balances as of Jan 26, 202524,477$ 24$ 11,237$ 28$ 68,038$ 79,327
Net income18,77518,775
Other comprehensive income158158
Issuance of common stock50370370
Tax withholding related to common stock(13)(1,532)(1,532)
Shares repurchased(126)(92)(14,411)(14,503)
Cash dividends declared and paid ($0.01 per common share)(244)(244)
Fair value of partially vested equity awards assumed in connection with acquisitions2222
Stock-based compensation1,4701,470
Balances as of Apr 27, 202524,388$ 24$ 11,475$ 186$ 72,158$ 83,843
", + "path": "tables/table-4 Shareholders' Equity.html", + "metadata": { + "length": 2736, + "summary": "table-5\nThe table shows changes in shareholders' equity from Jan 2025 to Apr 2026, including net income, stock issuance, repurchases, dividends, and other adjustments.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-4 Shareholders' Equity.html", + "keywords": [ + "equity", + "stock", + "dividends" + ], + "tokens": [] + } + }, + { + "chunk_id": "125ac24b-3661-5289-9fd5-f78ecf1c4c37", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
Cash flows from operating activities:
Net income$ 58,321$ 18,775
Adjustments to reconcile net income to net cash provided by operating activities:
Stock-based compensation expense1,9281,474
Deferred income taxes1,584(2,177)
Depreciation and amortization997611
(Gains) losses from equity securities, net(15,936)175
Other(94)(98)
Changes in operating assets and liabilities, net of acquisitions:
Accounts receivable(2,243)933
Inventories(4,420)(1,258)
Prepaid expenses and other assets(983)560
Accounts payable2,210941
Accrued and other current liabilities7,7637,128
Other long-term liabilities1,217350
Net cash provided by operating activities50,34427,414
Cash flows from investing activities:
Proceeds from maturities of marketable debt securities1,9463,122
Proceeds from sales of non-marketable securities26
Proceeds from sales of marketable debt securities25467
Purchases of non-marketable securities(18,582)(649)
Purchases of marketable debt and equity securities(8,000)(6,546)
Purchases related to property and equipment and intangible assets(1,757)(1,227)
Acquisitions, net of cash acquired(87)(383)
Net cash used in investing activities(26,429)(5,216)
Cash flows from financing activities:
Proceeds related to employee stock plans515370
Payments related to repurchases of common stock(19,312)(14,095)
Payments related to employee stock plan taxes(2,129)(1,532)
Dividends paid(243)(244)
Principal payments on property and equipment and intangible assets(33)(52)
Other(81)
Net cash used in financing activities(21,283)(15,553)
Change in cash and cash equivalents2,6326,645
Cash and cash equivalents at beginning of period10,6058,589
Cash and cash equivalents at end of period$ 13,237$ 15,234
", + "path": "tables/table-5 Cash Flow Statement.html", + "metadata": { + "length": 3090, + "summary": "table-6\nThe table shows cash flow data for three months ended Apr 26, 2026 and Apr 27, 2025. Net cash from operations was $50,344 and $27,414; investing used $26,429 and $5,216; financing used $21,283 and $15,553. Cash ended at $13,237 and $15,234.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-5 Cash Flow Statement.html", + "keywords": [ + "cash flows", + "operating", + "investing" + ], + "tokens": [] + } + }, + { + "chunk_id": "29bdbd71-9e9f-5c85-a650-19125e27c9b0", + "type": "text", + "content": "NVIDIA Corporation and Subsidiaries\nCondensed Consolidated Statements of Income\n(In millions, except per share data)\n(Unaudited)\n\n[tables/table-1 Financial Results.html]\n\nSee accompanying Notes to Condensed Consolidated Financial Statements.\nNVIDIA Corporation and Subsidiaries\nCondensed Consolidated Statements of Comprehensive Income\n(In millions)\n(Unaudited)\n\n[tables/table-2 Comprehensive Income.html]\n\nSee accompanying Notes to Condensed Consolidated Financial Statements.\nNVIDIA Corporation and Subsidiaries\nCondensed Consolidated Balance Sheets\n(In millions)\n(Unaudited)\n\n[tables/table-3 Balance Sheet.html]\n\nSee accompanying Notes to Condensed Consolidated Financial Statements.\nNVIDIA Corporation and Subsidiaries\nCondensed Consolidated Statements of Shareholders' Equity\n(Unaudited)\n\n[tables/table-4 Shareholders' Equity.html]\n\nSee accompanying Notes to Condensed Consolidated Financial Statements.\nNVIDIA Corporation and Subsidiaries\nCondensed Consolidated Statements of Cash Flows\n(In millions)\n(Unaudited)\n\n[tables/table-5 Cash Flow Statement.html]\n\nSee accompanying Notes to Condensed Consolidated Financial Statements.\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Unaudited)", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Item 1. Financial Statements (Unaudited)", + "metadata": { + "length": 1304, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "NVIDIA", + "Corporation", + "Subsidiaries", + "Condensed", + "Consolidated", + "Statements", + "Income", + "In", + "millions", + "share", + "data", + "Unaudited", + "See", + "accompanying", + "Notes", + "Financial", + "Comprehensive", + "Balance", + "Sheets", + "Shareholders", + "Equity", + "Cash", + "Flows" + ], + "keywords": [], + "connect_to": [ + { + "target": "c19796ac-7ecc-5937-9d80-bcfac6375451", + "relation": "embeds", + "ref": "[tables/table-1 Financial Results.html]", + "position": { + "start": 130, + "end": 169 + } + }, + { + "target": "4b3fc8ff-2efb-5df6-9180-6d26431f3e7c", + "relation": "embeds", + "ref": "[tables/table-2 Comprehensive Income.html]", + "position": { + "start": 363, + "end": 405 + } + }, + { + "target": "699d3faa-22e1-57a5-a7d2-7f3883789334", + "relation": "embeds", + "ref": "[tables/table-3 Balance Sheet.html]", + "position": { + "start": 579, + "end": 614 + } + }, + { + "target": "dcfbc1d7-e729-550b-844c-f366db9c0c01", + "relation": "embeds", + "ref": "[tables/table-4 Shareholders' Equity.html]", + "position": { + "start": 794, + "end": 836 + } + }, + { + "target": "125ac24b-3661-5289-9fd5-f78ecf1c4c37", + "relation": "embeds", + "ref": "[tables/table-5 Cash Flow Statement.html]", + "position": { + "start": 1020, + "end": 1061 + } + } + ] + } + }, + { + "chunk_id": "5f36dd67-797e-5f22-9c78-fc7289732083", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "315081e4-2af5-5914-aef1-6a292645d85d", + "type": "text", + "content": "The accompanying unaudited condensed consolidated financial statements were prepared in accordance with accounting principles generally accepted in the United States of America, or U.S. GAAP, for interim financial information and with the instructions to Form 10-Q and Article 10 of Securities and Exchange Commission, or SEC, Regulation S-X. The January 25, 2026 consolidated balance sheet was derived from our audited consolidated financial statements included in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026, as filed with the SEC, but does not include all disclosures required by U.S. GAAP. In the opinion of management, all adjustments, consisting only of normal recurring adjustments considered necessary for a fair presentation of results of operations and financial position, have been included. The results for the interim periods presented are not necessarily indicative of the results expected for any future period. The following information should be read in conjunction with the audited consolidated financial statements and notes thereto included in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026.\nCertain prior fiscal year balances have been reclassified to conform to the current period presentation.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Basis of Presentation", + "metadata": { + "length": 1273, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "The", + "accompanying", + "unaudited", + "condensed", + "consolidated", + "financial", + "statements", + "prepared", + "accordance", + "accounting", + "principles", + "generally", + "accepted", + "United", + "States", + "America", + "GAAP", + "interim", + "information", + "instructions", + "Form", + "10", + "Article", + "Securities", + "Exchange", + "Commission", + "SEC", + "Regulation", + "January", + "25", + "2026", + "balance", + "sheet", + "derived", + "audited", + "included", + "Annual", + "Report", + "fiscal", + "year", + "ended", + "filed", + "include", + "disclosures", + "required", + "In", + "opinion", + "management", + "adjustments", + "consisting", + "normal", + "recurring", + "considered", + "fair", + "presentation", + "results", + "operations", + "position", + "periods", + "presented", + "necessarily", + "indicative", + "expected", + "future", + "period", + "read", + "conjunction", + "notes", + "thereto", + "Certain", + "prior", + "balances", + "reclassified", + "conform", + "current" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "881cc4f8-0d85-506d-b4e1-80e47a74a3e8", + "type": "text", + "content": "There have been no material changes to our significant accounting policies disclosed in Note 1 - Organization and Summary of Significant Accounting Policies, of the Notes to the Consolidated Financial Statements included in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Significant Accounting Policies", + "metadata": { + "length": 298, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "There", + "material", + "significant", + "accounting", + "policies", + "disclosed", + "Note", + "Organization", + "Summary", + "Significant", + "Accounting", + "Policies", + "Notes", + "Consolidated", + "Financial", + "Statements", + "included", + "Annual", + "Report", + "Form", + "10", + "fiscal", + "year", + "ended", + "January", + "25", + "2026" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3732b8a0-095a-56d4-83e9-7bb74ca7a881", + "type": "text", + "content": "Fiscal year 2027 is a 53-week year and fiscal year 2026 was a 52-week year, both ending on the last Sunday in January. The first quarters of fiscal years 2027 and 2026 were both 13-week quarters. The fourth quarter of fiscal year 2027 will be a 14-week quarter.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Significant Accounting Policies/Fiscal Year", + "metadata": { + "length": 261, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Fiscal", + "year", + "2027", + "53", + "week", + "fiscal", + "2026", + "52", + "ending", + "Sunday", + "January", + "The", + "quarters", + "years", + "13", + "fourth", + "quarter", + "14" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d6130ee9-5661-5f4d-ac29-42dfdd59ff41", + "type": "text", + "content": "Our condensed consolidated financial statements include the accounts of NVIDIA Corporation and our wholly owned subsidiaries. All intercompany balances and transactions have been eliminated in consolidation.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Significant Accounting Policies/Principles of Consolidation", + "metadata": { + "length": 207, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Our", + "condensed", + "consolidated", + "financial", + "statements", + "include", + "accounts", + "NVIDIA", + "Corporation", + "wholly", + "owned", + "subsidiaries", + "All", + "intercompany", + "balances", + "transactions", + "eliminated", + "consolidation" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f9db58cc-839e-5f0f-bc6d-c7b9b31384eb", + "type": "text", + "content": "The preparation of financial statements in conformity with U.S. GAAP requires management to make estimates and assumptions that affect the reported amounts of assets and liabilities and disclosures of contingent assets and liabilities at the date of the financial statements and the reported amounts of revenue and expenses during the reporting period. Actual results could differ materially from our estimates.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Significant Accounting Policies/Use of Estimates", + "metadata": { + "length": 411, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "The", + "preparation", + "financial", + "statements", + "conformity", + "GAAP", + "requires", + "management", + "make", + "estimates", + "assumptions", + "affect", + "reported", + "amounts", + "assets", + "liabilities", + "disclosures", + "contingent", + "date", + "revenue", + "expenses", + "reporting", + "period", + "Actual", + "results", + "differ", + "materially" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8cc6c4c7-5a91-51f6-a1ea-b6aa73e66eaa", + "type": "text", + "content": "Recent Accounting Pronouncements Not Yet Adopted\nIn November 2024, the Financial Accounting Standards Board, or FASB, issued a new accounting standard requiring disclosures of certain additional expense information on an annual and interim basis, including, among other items, the amounts of purchases of inventory, employee compensation, depreciation and intangible asset amortization included within each income statement expense caption, as applicable. We will adopt this standard in the fiscal year 2028 annual report. We do not expect the adoption of this standard to have a material impact on our Consolidated Financial Statements other than additional disclosures.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Recently Issued Accounting Pronouncements", + "metadata": { + "length": 671, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Recent", + "Accounting", + "Pronouncements", + "Not", + "Yet", + "Adopted", + "In", + "November", + "2024", + "Financial", + "Standards", + "Board", + "FASB", + "issued", + "accounting", + "standard", + "requiring", + "disclosures", + "additional", + "expense", + "information", + "annual", + "interim", + "basis", + "including", + "items", + "amounts", + "purchases", + "inventory", + "employee", + "compensation", + "depreciation", + "intangible", + "asset", + "amortization", + "included", + "income", + "statement", + "caption", + "applicable", + "We", + "adopt", + "fiscal", + "year", + "2028", + "report", + "expect", + "adoption", + "material", + "impact", + "Consolidated", + "Statements" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d270d060-0939-5af0-a29b-749b3ba5e1d5", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Cost of revenue$ 68$ 64
Research and development1,4591,063
Sales, general and administrative401347
Total$ 1,928$ 1,474
", + "path": "tables/table-6 Cost Breakdown.html", + "metadata": { + "length": 437, + "summary": "table-7\nThe table shows costs for three months ending Apr 26, 2026 and Apr 27, 2025. Cost of revenue: $68M vs $64M; R&D: $1,459M vs $1,063M; SG&A: $401M vs $347M; Total: $1,928M vs $1,474M.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-6 Cost Breakdown.html", + "keywords": [ + "Cost", + "Revenue", + "Expenses" + ], + "tokens": [] + } + }, + { + "chunk_id": "b591e76f-1c39-59c5-b169-36fc84398fb0", + "type": "table", + "content": "
Number of SharesWeighted Average Grant-Date Fair Value Per Share
(In millions, except per share data)
Balance as of Jan 25, 2026189$ 81.51
Granted44$ 181.73
Vested(31)$ 48.04
Canceled and forfeited(2)$ 95.82
Balance as of Apr 26, 2026200$ 108.92
", + "path": "tables/table-7 Stock Grant Activity.html", + "metadata": { + "length": 512, + "summary": "table-8\nThe table shows restricted stock activity from Jan to Apr 2026: shares increased from 189 to 200 million, with 44 million granted at $181.73, 31 million vested at $48.04, and 2 million forfeited.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-7 Stock Grant Activity.html", + "keywords": [ + "shares", + "fair value", + "grants" + ], + "tokens": [] + } + }, + { + "chunk_id": "225644a7-158e-5fd1-a950-e02f84f14ad6", + "type": "text", + "content": "We recognize stock-based compensation expense from grants of restricted stock units, or RSUs, performance stock units, or PSUs, and market-based PSUs, and issuances under our employee stock purchase plan, or ESPP.\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)\nCondensed Consolidated Statements of Income include stock-based compensation expense as follows:\n\n[tables/table-6 Cost Breakdown.html]\n\nEquity Award Activity\nThe following is a summary of our equity award transactions under our equity incentive plans:\nRSUs, PSUs and Market-based PSUs Outstanding\n\n[tables/table-7 Stock Grant Activity.html]\n\nAs of April 26, 2026, aggregate unearned stock-based compensation expense was \\$20.8 billion, which is expected to be recognized over a weighted average period of 2.6 years for RSUs, PSUs, and market-based PSUs, and one year for ESPP.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 2 - Stock-Based Compensation", + "metadata": { + "length": 927, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "We", + "recognize", + "stock", + "based", + "compensation", + "expense", + "grants", + "restricted", + "units", + "RSUs", + "performance", + "PSUs", + "market", + "issuances", + "employee", + "purchase", + "plan", + "ESPP", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Continued", + "Unaudited", + "Income", + "include", + "Equity", + "Award", + "Activity", + "The", + "summary", + "equity", + "award", + "transactions", + "incentive", + "plans", + "Market", + "Outstanding", + "As", + "April", + "26", + "2026", + "aggregate", + "unearned", + "20.8", + "billion", + "expected", + "recognized", + "weighted", + "average", + "period", + "2.6", + "years", + "year" + ], + "keywords": [], + "connect_to": [ + { + "target": "d270d060-0939-5af0-a29b-749b3ba5e1d5", + "relation": "embeds", + "ref": "[tables/table-6 Cost Breakdown.html]", + "position": { + "start": 425, + "end": 461 + } + }, + { + "target": "b591e76f-1c39-59c5-b169-36fc84398fb0", + "relation": "embeds", + "ref": "[tables/table-7 Stock Grant Activity.html]", + "position": { + "start": 625, + "end": 667 + } + } + ] + } + }, + { + "chunk_id": "85bdcf7d-b72c-50d1-9a0d-f32fddcf6b10", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions, except per share data)
Numerator:
Net income$ 58,321$ 18,775
Denominator:
Basic weighted average shares24,28624,441
Dilutive impact of outstanding equity awards105170
Diluted weighted average shares24,39124,611
Net income per share:
Basic (1)$ 2.40$ 0.77
Diluted (2)$ 2.39$ 0.76
Anti-dilutive equity awards excluded from diluted net income per share4762
", + "path": "tables/table-8 EPS Data Table.html", + "metadata": { + "length": 864, + "summary": "table-9\nThe table shows net income of $58,321M (Apr 2026) vs $18,775M (Apr 2025), basic EPS $2.40 vs $0.77, diluted EPS $2.39 vs $0.76, with weighted average shares and anti-dilutive awards.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-8 EPS Data Table.html", + "keywords": [ + "Net income", + "Earnings per share", + "Weighted shares" + ], + "tokens": [] + } + }, + { + "chunk_id": "024ce11b-4b82-5841-b259-e6f93907c57b", + "type": "text", + "content": "The following is the basic and diluted net income per share computations for the periods presented:\n\n[tables/table-8 EPS Data Table.html]\n\n(1) Net income divided by basic weighted average shares.\n(2) Net income divided by diluted weighted average shares.\nDiluted net income per share was computed using the weighted average number of common and potentially dilutive shares outstanding during the period, using the treasury stock method.\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 3 - Net Income Per Share", + "metadata": { + "length": 553, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "The", + "basic", + "diluted", + "net", + "income", + "share", + "computations", + "periods", + "presented", + "Net", + "divided", + "weighted", + "average", + "shares", + "Diluted", + "computed", + "number", + "common", + "potentially", + "dilutive", + "outstanding", + "period", + "treasury", + "stock", + "method", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Continued", + "Unaudited" + ], + "keywords": [], + "connect_to": [ + { + "target": "85bdcf7d-b72c-50d1-9a0d-f32fddcf6b10", + "relation": "embeds", + "ref": "[tables/table-8 EPS Data Table.html]", + "position": { + "start": 101, + "end": 137 + } + } + ] + } + }, + { + "chunk_id": "af21bb32-3527-5c94-b9fb-d7395cabec5c", + "type": "table", + "content": "
Apr 26, 2026Jan 25, 2026
Gross Carrying AmountAccumulated AmortizationNet Carrying AmountGross Carrying AmountAccumulated AmortizationNet Carrying Amount
(In millions)
Acquisition-related intangible assets$ 5,658$ (2,759)$ 2,899$ 5,656$ (2,580)$ 3,076
Patents and licensed technology525(304)221528(298)230
Total intangible assets$ 6,183$ (3,063)$ 3,120$ 6,184$ (2,878)$ 3,306
", + "path": "tables/table-9 Intangible Assets.html", + "metadata": { + "length": 766, + "summary": "table-10\nThe table shows intangible assets as of Apr 26, 2026 and Jan 25, 2026, with gross carrying amounts, accumulated amortization, and net carrying amounts for acquisition-related assets, patents, and total.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-9 Intangible Assets.html", + "keywords": [ + "intangible assets", + "amortization", + "net carrying amount" + ], + "tokens": [] + } + }, + { + "chunk_id": "a56cbb8a-3168-5a2b-b341-df1b726c41f7", + "type": "table", + "content": "
Future Amortization Expense(In millions)
Fiscal Year:
2027 (excluding the first quarter of fiscal year 2027)$ 689
2028754
2029610
2030516
2031468
2032 and thereafter83
Total$ 3,120
", + "path": "tables/table-10 Amortization Expense.html", + "metadata": { + "length": 430, + "summary": "table-11\nThe table shows future amortization expenses from 2027 to 2032 and beyond, totaling $3,120 million, with the highest in 2028 at $754 million.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-10 Amortization Expense.html", + "keywords": [ + "amortization", + "fiscal years", + "future expense" + ], + "tokens": [] + } + }, + { + "chunk_id": "9e83010b-c1fe-593d-b46f-73e77e1b9db9", + "type": "text", + "content": "The components of our amortizable intangible assets are as follows:\n\n[tables/table-9 Intangible Assets.html]\n\nAmortization expense associated with intangible assets was \\$232 million and \\$159 million for the first quarter of fiscal years 2027 and 2026, respectively.\nThe following table outlines the estimated future amortization expense related to the net carrying amount of intangible assets as of April 26, 2026:\n\n[tables/table-10 Amortization Expense.html]\n\nIn the first quarter of fiscal year 2027, goodwill increased by \\$62 million from acquisitions and was allocated to our Compute \\& Networking reporting unit.\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 4 - Amortizable Intangible Assets and Goodwill", + "metadata": { + "length": 745, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "The", + "components", + "amortizable", + "intangible", + "assets", + "Amortization", + "expense", + "232", + "million", + "159", + "quarter", + "fiscal", + "years", + "2027", + "2026", + "table", + "outlines", + "estimated", + "future", + "amortization", + "related", + "net", + "carrying", + "amount", + "April", + "26", + "In", + "year", + "goodwill", + "increased", + "62", + "acquisitions", + "allocated", + "Compute", + "Networking", + "reporting", + "unit", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Continued", + "Unaudited" + ], + "keywords": [], + "connect_to": [ + { + "target": "af21bb32-3527-5c94-b9fb-d7395cabec5c", + "relation": "embeds", + "ref": "[tables/table-9 Intangible Assets.html]", + "position": { + "start": 69, + "end": 108 + } + }, + { + "target": "a56cbb8a-3168-5a2b-b341-df1b726c41f7", + "relation": "embeds", + "ref": "[tables/table-10 Amortization Expense.html]", + "position": { + "start": 418, + "end": 461 + } + } + ] + } + }, + { + "chunk_id": "d7478cd0-e8ad-5893-acb5-950f047f0e5f", + "type": "table", + "content": "
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable Debt SecuritiesMarketable Equity SecuritiesOther Assets
(In millions)
Debt securities issued by the U.S. TreasuryLevel 2$ 21,883$ 43$ (8)$ 21,918$ 470$ 21,448$ —$ —
Corporate debt securitiesLevel 215,09248(8)15,1321,53313,599
Debt securities issued by U.S. government agenciesLevel 22,0092(1)2,0102,010
Certificates of depositLevel 2132132132
Foreign government bondsLevel 24014141
Money market fundsLevel 110,21210,21210,212
Publicly-held equity securities (1) (2)Level 129,88721,0238,864
Publicly-held equity securities (1) (3)Level 29,2149,214
Total$ 49,368$ 94$ (17)$ 88,546$ 12,347$ 37,098$ 30,237$ 8,864
", + "path": "tables/table-11 Debt Securities.html", + "metadata": { + "length": 1886, + "summary": "table-12\nThe table details debt and equity securities by pricing category, showing cost, unrealized gains/losses, estimated fair value, and how they are reported as cash equivalents, marketable debt, or equity securities.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-11 Debt Securities.html", + "keywords": [ + "Debt Securities", + "Fair Value", + "Pricing Levels" + ], + "tokens": [] + } + }, + { + "chunk_id": "917e224a-0fba-5653-87f1-0566d90cdd09", + "type": "table", + "content": "
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable Debt SecuritiesMarketable Equity SecuritiesOther Assets
(In millions)
Debt securities issued by the U.S. TreasuryLevel 2$ 21,635$ 77$ (3)$ 21,709$ —$ 21,709$ —$ —
Corporate debt securitiesLevel 215,41092(3)15,49934515,154
Debt securities issued by U.S. government agenciesLevel 22,15742,1612,161
Certificates of depositLevel 2110110110
Foreign government bondsLevel 24014141
Money market fundsLevel 17,8307,8307,830
Publicly-held equity securities (1) (2)Level 117,72612,8864,840
Total$ 47,182$ 174$ (6)$ 65,076$ 8,285$ 39,065$ 12,886$ 4,840
", + "path": "tables/table-12 Fair Value Table.html", + "metadata": { + "length": 1718, + "summary": "table-13\nThe table details fair value measurements of debt and equity securities by pricing level, showing cost, unrealized gains/losses, and estimated fair value, with breakdowns by type and reporting category.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-12 Fair Value Table.html", + "keywords": [ + "fair value", + "debt securities", + "equity securities" + ], + "tokens": [] + } + }, + { + "chunk_id": "cd6a3c2a-3542-5c81-b540-8dde21e63605", + "type": "table", + "content": "
Apr 26, 2026Jan 25, 2026
Less than 12 monthsLess than 12 months
Estimated Fair ValueGross Unrealized LossEstimated Fair ValueGross Unrealized Loss
(In millions)
Debt securities issued by the U.S. Treasury$ 12,238$ (8)$ 10,666$ (3)
Corporate debt securities2,658(8)1,332(3)
Debt securities issued by U.S. government agencies1,291(1)1,134
Total$ 16,187$ (17)$ 13,132$ (6)
", + "path": "tables/table-13 Debt Securities.html", + "metadata": { + "length": 792, + "summary": "table-14\nThe table shows estimated fair values and gross unrealized losses for debt securities (U.S. Treasury, corporate, and government agencies) as of Apr 26, 2026 and Jan 25, 2026, with totals of $16,187M and $13,132M respectively.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-13 Debt Securities.html", + "keywords": [ + "debt securities", + "unrealized loss", + "fair value" + ], + "tokens": [] + } + }, + { + "chunk_id": "a4c67a78-cf3e-50ea-8ebb-550783331e2b", + "type": "table", + "content": "
Apr 26, 2026(In millions)
Less than one year$ 24,307
Due in 1 - 5 years14,926
Total$ 39,233
", + "path": "tables/table-14 Debt Maturity.html", + "metadata": { + "length": 211, + "summary": "table-15\nThe table shows debt maturities: $24,307 million due within one year, $14,926 million due in 1-5 years, totaling $39,233 million as of Apr 26, 2026.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-14 Debt Maturity.html", + "keywords": [ + "Debt", + "Maturity", + "Schedule" + ], + "tokens": [] + } + }, + { + "chunk_id": "193f4f29-8e56-5e38-aca5-74276f8101cd", + "type": "text", + "content": "Cash equivalents and marketable securities including debt and equity securities are measured at fair value using quoted prices in active markets for identical assets (Level 1) or for similar assets or use of other observable inputs (Level 2).\nThe following is a summary of cash equivalents and marketable securities:\nApr 26, 2026\n\n[tables/table-11 Debt Securities.html]\n\n(1) The balance as of April 26, 2026 included \\$27.4 billion of investments, which are subject to short-term lock-up restrictions on the ability to sell.\n(2) The long-term portion of publicly-held equity securities, which are subject to lock-up restrictions through December 2027 of \\$8.9 billion as of April 26, 2026, was included in Other assets.\n(3) The publicly-held equity securities classified in Level 2 include investments in warrants and preferred stock convertible to common stock in public companies.\nPublicly-held equity securities are subject to market price volatility. Net unrealized gains on investments in publicly-held equity securities held at period end were \\$13.4 billion for the first quarter of fiscal year 2027. Net unrealized losses on investments in publicly-held equity securities held at period end were \\$222 million for the first quarter of fiscal year 2026. Unrealized gains and losses are recognized in Other income (expense), net, in the Condensed Consolidated Statements of Income.\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)\nJan 25, 2026\n\n[tables/table-12 Fair Value Table.html]\n\n(1) The balance as of January 25, 2026 included \\$10.5 billion of investments that are subject to short-term lock-up restrictions on the ability to sell.\n(2) The long-term portion of publicly-held equity securities, which are subject to lock-up restrictions through December 2027 of \\$4.8 billion as of January 25, 2026, was included in Other assets.\nThe following table provides the breakdown of unrealized losses, aggregated by investment category and length of time that individual debt securities have been in a continuous loss position:\n\n[tables/table-13 Debt Securities.html]\n\nGross unrealized losses are related to fixed income securities, driven primarily by changes in interest rates.\nThe estimated fair values of debt securities included in cash equivalents and marketable debt securities are shown below by contractual maturity.\n\n[tables/table-14 Debt Maturity.html]\n\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 5 - Cash Equivalents and Marketable Securities", + "metadata": { + "length": 2696, + "summary": "The passage discusses the fair value measurement of cash equivalents and marketable securities, including debt and equity securities, using Level 1 and Level 2 inputs. It provides summaries for April 26, 2026 and January 25, 2026, noting investments subject to lock-up restrictions (e.g., $27.4 billion and $10.5 billion, respectively) and long-term publicly-held equity securities with lock-ups through December 2027. Net unrealized gains of $13.4 billion for Q1 fiscal 2027 and net unrealized losses of $222 million for Q1 fiscal 2026 are reported, recognized in Other income (expense). Tables detail unrealized losses by investment category and maturity of debt securities, with gross unrealized losses primarily due to interest rate changes.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Cash", + "equivalents", + "marketable", + "securities", + "including", + "debt", + "equity", + "measured", + "fair", + "quoted", + "prices", + "active", + "markets", + "identical", + "assets", + "Level", + "similar", + "observable", + "inputs", + "The", + "summary", + "cash", + "Apr", + "26", + "2026", + "balance", + "April", + "included", + "27.4", + "billion", + "investments", + "subject", + "short", + "term", + "lock", + "restrictions", + "ability", + "sell", + "long", + "portion", + "publicly", + "held", + "December", + "2027", + "8.9", + "Other", + "classified", + "include", + "warrants", + "preferred", + "stock", + "convertible", + "common", + "public", + "companies", + "Publicly", + "market", + "price", + "volatility", + "Net", + "unrealized", + "gains", + "period", + "end", + "13.4", + "quarter", + "fiscal", + "year", + "losses", + "222", + "million", + "Unrealized", + "recognized", + "income", + "expense", + "net", + "Condensed", + "Consolidated", + "Statements", + "Income", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Financial", + "Continued", + "Unaudited", + "Jan", + "25", + "January", + "10.5", + "4.8", + "table", + "breakdown", + "aggregated", + "investment", + "category", + "length", + "time", + "individual", + "continuous", + "loss", + "position", + "Gross", + "related", + "fixed", + "driven", + "primarily", + "interest", + "rates", + "estimated", + "values", + "shown", + "contractual", + "maturity" + ], + "keywords": [ + "fair value", + "marketable securities", + "unrealized gains" + ], + "connect_to": [ + { + "target": "d7478cd0-e8ad-5893-acb5-950f047f0e5f", + "relation": "embeds", + "ref": "[tables/table-11 Debt Securities.html]", + "position": { + "start": 331, + "end": 369 + } + }, + { + "target": "917e224a-0fba-5653-87f1-0566d90cdd09", + "relation": "embeds", + "ref": "[tables/table-12 Fair Value Table.html]", + "position": { + "start": 1515, + "end": 1554 + } + }, + { + "target": "cd6a3c2a-3542-5c81-b540-8dde21e63605", + "relation": "embeds", + "ref": "[tables/table-13 Debt Securities.html]", + "position": { + "start": 2099, + "end": 2137 + } + }, + { + "target": "a4c67a78-cf3e-50ea-8ebb-550783331e2b", + "relation": "embeds", + "ref": "[tables/table-14 Debt Maturity.html]", + "position": { + "start": 2397, + "end": 2433 + } + } + ] + } + }, + { + "chunk_id": "7a709bdf-6751-54a5-99c2-3b15ec32dd2c", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 6 - Non-marketable Securities", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "95399265-72b7-55ea-b7eb-2a7cebe439fc", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Balance at beginning of period$ 22,251$ 3,387
Adjustments related to non-marketable equity securities:
Net additions17,899649
Unrealized gains (1)2,60363
Reclassification (2)(389)(843)
Impairments and unrealized losses(28)(16)
Balance at end of period$ 42,336$ 3,240
", + "path": "tables/table-15 Equity Securities.html", + "metadata": { + "length": 690, + "summary": "table-16\nThe table shows changes in non-marketable equity securities over three months ending Apr 26, 2026, with balance increasing from $22,251M to $42,336M, including net additions of $17,899M and unrealized gains of $2,603M.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-15 Equity Securities.html", + "keywords": [ + "non-marketable equity", + "unrealized gains", + "balance" + ], + "tokens": [] + } + }, + { + "chunk_id": "04f5ee14-491a-584e-827c-ee8247cceeea", + "type": "text", + "content": "Our non-marketable equity securities are primarily in privately-held companies carried at cost less impairment, and adjusted for observable price changes. We value investments using observable comparable transactions and other inputs including volatility, expected time to liquidity, the risk-free rate, and security-specific rights and obligations.\nAdjustments to the carrying value of privately-held securities:\n\n[tables/table-15 Equity Securities.html]\n\n(1) Unrealized gains are recognized in Other income (expense), net, in the Condensed Consolidated Statements of Income.\n(2) Includes primarily reclassifications to marketable securities following public market trading.\nNon-marketable equity securities had cumulative gross unrealized gains of \\$5.3 billion and \\$396 million, and cumulative gross unrealized losses and impairments of \\$199 million and \\$110 million as of April 26, 2026 and April 27, 2025, respectively.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 6 - Non-marketable Securities/Non-marketable Equity Securities", + "metadata": { + "length": 928, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Our", + "marketable", + "equity", + "securities", + "primarily", + "privately", + "held", + "companies", + "carried", + "cost", + "impairment", + "adjusted", + "observable", + "price", + "We", + "investments", + "comparable", + "transactions", + "inputs", + "including", + "volatility", + "expected", + "time", + "liquidity", + "risk", + "free", + "rate", + "security", + "specific", + "rights", + "obligations", + "Adjustments", + "carrying", + "Unrealized", + "gains", + "recognized", + "Other", + "income", + "expense", + "net", + "Condensed", + "Consolidated", + "Statements", + "Income", + "Includes", + "reclassifications", + "public", + "market", + "trading", + "Non", + "cumulative", + "gross", + "unrealized", + "5.3", + "billion", + "396", + "million", + "losses", + "impairments", + "199", + "110", + "April", + "26", + "2026", + "27", + "2025" + ], + "keywords": [], + "connect_to": [ + { + "target": "95399265-72b7-55ea-b7eb-2a7cebe439fc", + "relation": "embeds", + "ref": "[tables/table-15 Equity Securities.html]", + "position": { + "start": 415, + "end": 455 + } + } + ] + } + }, + { + "chunk_id": "80d24a2a-d202-5997-8e21-8066d1b849d8", + "type": "text", + "content": "We have \\$1.0 billion of investments in infrastructure funds accounted for using the equity method as of April 26, 2026. Our maximum loss exposure under these investments, including invested and future committed amounts, was \\$2.3 billion as of April 26, 2026.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 6 - Non-marketable Securities/Equity Method Investments", + "metadata": { + "length": 260, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "We", + "1.0", + "billion", + "investments", + "infrastructure", + "funds", + "accounted", + "equity", + "method", + "April", + "26", + "2026", + "Our", + "maximum", + "loss", + "exposure", + "including", + "invested", + "future", + "committed", + "amounts", + "2.3" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4737724c-0f81-5485-9ae0-b95ec0386184", + "type": "text", + "content": "Total Investment commitments were \\$27 billion as of April 26, 2026, subject to certain contingencies, which we expect will be made through the remainder of fiscal year 2027.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 6 - Non-marketable Securities/Investment Commitments", + "metadata": { + "length": 174, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Total", + "Investment", + "commitments", + "27", + "billion", + "April", + "26", + "2026", + "subject", + "contingencies", + "expect", + "made", + "remainder", + "fiscal", + "year", + "2027" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d2ea1bd6-6b61-57ec-9a99-5ea16dbd4776", + "type": "table", + "content": "
Apr 26, 2026Jan 25, 2026
Inventories:(In millions)
Raw materials$ 6,647$ 3,807
Work in process9,9498,822
Finished goods9,2018,774
Total inventories (1)$ 25,797$ 21,403
", + "path": "tables/table-16 Inventory Data.html", + "metadata": { + "length": 396, + "summary": "table-17\nThe table shows inventories in millions for Apr 26, 2026 and Jan 25, 2026: raw materials $6,647 vs $3,807, work in process $9,949 vs $8,822, finished goods $9,201 vs $8,774, total $25,797 vs $21,403.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-16 Inventory Data.html", + "keywords": [ + "Inventories", + "Raw materials", + "Finished goods" + ], + "tokens": [] + } + }, + { + "chunk_id": "3665bae0-7d48-59b4-bd43-92740071cd46", + "type": "text", + "content": "We refer to customers who purchase products directly from NVIDIA as direct customers, such as add-in board manufacturers, or AIBs, distributors, original design manufacturers, or ODMs, original equipment manufacturers, or OEMs, cloud service providers, or CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. Three direct customers accounted for 30%, 18%, and 16% of our accounts receivable balance as of April 26, 2026. Three direct customers accounted for 25%, 18%, and 13% of our accounts receivable balance as of January 25, 2026.\nCertain balance sheet components were as follows:\n\n[tables/table-16 Inventory Data.html]\n\n(1) We recorded inventory provisions of \\$0.8 billion and \\$2.3 billion for the first quarter of fiscal years 2027 and 2026, respectively, in Cost of revenue.\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 7 - Balance Sheet Components", + "metadata": { + "length": 1019, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "We", + "refer", + "customers", + "purchase", + "products", + "directly", + "NVIDIA", + "direct", + "add", + "board", + "manufacturers", + "AIBs", + "distributors", + "original", + "design", + "ODMs", + "equipment", + "OEMs", + "cloud", + "service", + "providers", + "CSPs", + "AI", + "model", + "makers", + "system", + "integrators", + "Certain", + "internal", + "resources", + "party", + "complete", + "build", + "Three", + "accounted", + "30%", + "18%", + "16%", + "accounts", + "receivable", + "balance", + "April", + "26", + "2026", + "25%", + "13%", + "January", + "25", + "sheet", + "components", + "recorded", + "inventory", + "provisions", + "0.8", + "billion", + "2.3", + "quarter", + "fiscal", + "years", + "2027", + "Cost", + "revenue", + "Corporation", + "Subsidiaries", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Continued", + "Unaudited" + ], + "keywords": [], + "connect_to": [ + { + "target": "d2ea1bd6-6b61-57ec-9a99-5ea16dbd4776", + "relation": "embeds", + "ref": "[tables/table-16 Inventory Data.html]", + "position": { + "start": 698, + "end": 735 + } + } + ] + } + }, + { + "chunk_id": "fec267b6-eeeb-5d99-a40d-3bdfeff0783f", + "type": "table", + "content": "
Apr 26, 2026Jan 25, 2026
Accrued and Other Current Liabilities:(In millions)
Taxes payable$ 10,638$ 2,669
Customer program accruals4,1825,318
Accrued purchase consideration (1)3,9573,921
Excess inventory purchase obligations (2)3,1212,739
Product warranty2,9482,807
Deferred revenue (3)1,7141,379
Accrued payroll and related expenses1,0331,146
Other2,1941,373
Total accrued and other current liabilities$ 29,787$ 21,352
", + "path": "tables/table-17 Accrued Liabilities.html", + "metadata": { + "length": 823, + "summary": "table-18\nThe table shows accrued and other current liabilities as of Apr 26, 2026 ($29,787M) and Jan 25, 2026 ($21,352M), including taxes payable, customer program accruals, and warranty costs.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-17 Accrued Liabilities.html", + "keywords": [ + "accrued liabilities", + "taxes payable", + "customer program" + ], + "tokens": [] + } + }, + { + "chunk_id": "681c5cf2-d629-50b6-8a27-9dd5e8d1def6", + "type": "table", + "content": "
Apr 26, 2026Jan 25, 2026
Other Long-Term Liabilities:(In millions)
Income tax payable (1)$ 4,830$ 3,958
Deferred income tax1,7981,774
Deferred revenue (2)1,4031,193
Other737381
Total other long-term liabilities$ 8,768$ 7,306
", + "path": "tables/table-18 Long-Term Liabilities.html", + "metadata": { + "length": 488, + "summary": "table-19\nThe table shows other long-term liabilities totaling $8,768 million as of Apr 26, 2026, down from $7,306 million on Jan 25, 2026, including income tax payable, deferred income tax, deferred revenue, and other items.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-18 Long-Term Liabilities.html", + "keywords": [ + "liabilities", + "income tax", + "deferred revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "95d1081f-e62e-52c2-b885-74d7e53ae909", + "type": "text", + "content": "Property, equipment and intangible assets acquired but not paid for the first quarter of fiscal years 2027 and 2026 were \\$1.1 billion and \\$408 million, respectively.\n\n[tables/table-17 Accrued Liabilities.html]\n\n(1) Related to the Groq, Inc. non-exclusive license agreement.\n(2) We recorded \\$0.3 billion and \\$3.0 billion for the first quarter of fiscal years 2027 and 2026, respectively, in Cost of revenue.\n(3) Includes customer advances and unearned revenue related to hardware and software support, cloud services, and license and development arrangements. The balance as of April 26, 2026 and January 25, 2026 included \\$297 million and \\$160 million of customer advances, respectively.\n\n[tables/table-18 Long-Term Liabilities.html]\n\n(1) Primarily comprised of unrecognized tax benefits and related interest and penalties.\n(2) Includes unearned revenue related to hardware and software support and cloud services.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 7 - Balance Sheet Components/Property and Equipment:", + "metadata": { + "length": 930, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Property", + "equipment", + "intangible", + "assets", + "acquired", + "paid", + "quarter", + "fiscal", + "years", + "2027", + "2026", + "1.1", + "billion", + "408", + "million", + "Related", + "Groq", + "Inc", + "exclusive", + "license", + "agreement", + "We", + "recorded", + "0.3", + "3.0", + "Cost", + "revenue", + "Includes", + "customer", + "advances", + "unearned", + "related", + "hardware", + "software", + "support", + "cloud", + "services", + "development", + "arrangements", + "The", + "balance", + "April", + "26", + "January", + "25", + "included", + "297", + "160", + "Primarily", + "comprised", + "unrecognized", + "tax", + "benefits", + "interest", + "penalties" + ], + "keywords": [], + "connect_to": [ + { + "target": "fec267b6-eeeb-5d99-a40d-3bdfeff0783f", + "relation": "embeds", + "ref": "[tables/table-17 Accrued Liabilities.html]", + "position": { + "start": 169, + "end": 211 + } + }, + { + "target": "681c5cf2-d629-50b6-8a27-9dd5e8d1def6", + "relation": "embeds", + "ref": "[tables/table-18 Long-Term Liabilities.html]", + "position": { + "start": 695, + "end": 739 + } + } + ] + } + }, + { + "chunk_id": "2e275abd-dc22-53bd-93c4-2035303b3839", + "type": "table", + "content": "
Apr 26, 2026Apr 27, 2025
(In millions)
Balance at beginning of period$ 2,572$ 1,813
Deferred revenue additions (1)2,5306,493
Revenue recognized (2)(1,985)(6,228)
Balance at end of period$ 3,117$ 2,078
", + "path": "tables/table-19 Deferred Revenue.html", + "metadata": { + "length": 429, + "summary": "table-20\nThe table shows deferred revenue balances for Apr 26, 2026 ($3,117M) and Apr 27, 2025 ($2,078M), with additions and revenue recognized during each period.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-19 Deferred Revenue.html", + "keywords": [ + "Deferred Revenue", + "Revenue Recognition", + "Balance Sheet" + ], + "tokens": [] + } + }, + { + "chunk_id": "65a6c73f-1917-5985-b56c-78a5c3cd1b46", + "type": "text", + "content": "The following table shows the changes in short- and long-term deferred revenue during the first quarter of fiscal years 2027 and 2026:\nThree Months Ended\n\n[tables/table-19 Deferred Revenue.html]\n\n(1) Includes \\$1.7 billion and \\$6.2 billion of customer advances for the first quarter of fiscal years 2027 and 2026, respectively.\n(2) Includes \\$1.6 billion and \\$6.0 billion related to customer advances for the first quarter of fiscal years 2027 and 2026, respectively.\nWe recognized revenue of \\$451 million and \\$265 million in the first quarter of fiscal years 2027 and 2026, respectively, that was included in the prior year-end deferred revenue balance.\nAs of April 26, 2026, revenue related to remaining performance obligations from contracts greater than one year in length was \\$2.6 billion, which includes \\$2.3 billion from deferred revenue and \\$304 million, which has not yet been billed or\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)\nrecognized as revenue. Approximately 40% of revenue from contracts greater than one year in length will be recognized over the next twelve months.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 7 - Balance Sheet Components/Deferred Revenue", + "metadata": { + "length": 1171, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "The", + "table", + "shows", + "short", + "long", + "term", + "deferred", + "revenue", + "quarter", + "fiscal", + "years", + "2027", + "2026", + "Three", + "Months", + "Ended", + "Includes", + "1.7", + "billion", + "6.2", + "customer", + "advances", + "1.6", + "6.0", + "related", + "We", + "recognized", + "451", + "million", + "265", + "included", + "prior", + "year", + "end", + "balance", + "As", + "April", + "26", + "remaining", + "performance", + "obligations", + "contracts", + "greater", + "length", + "2.6", + "includes", + "2.3", + "304", + "billed", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Continued", + "Unaudited", + "Approximately", + "40%", + "twelve", + "months" + ], + "keywords": [], + "connect_to": [ + { + "target": "2e275abd-dc22-53bd-93c4-2035303b3839", + "relation": "embeds", + "ref": "[tables/table-19 Deferred Revenue.html]", + "position": { + "start": 155, + "end": 194 + } + } + ] + } + }, + { + "chunk_id": "4af4cc8b-ffab-5e8b-8074-1485315acb77", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 8 - Derivative Financial Instruments", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a0880006-66e3-5c79-8c8d-09c73661584e", + "type": "table", + "content": "
Apr 26, 2026Jan 25, 2026
(In millions)
Designated as accounting hedges$ 2,114$ 1,765
Not designated as accounting hedges$ 1,850$ 2,332
", + "path": "tables/table-20 Hedge Designation.html", + "metadata": { + "length": 293, + "summary": "table-21\nThe table shows designated accounting hedges at $2,114M (Apr 26, 2026) and $1,765M (Jan 25, 2026), and non-designated hedges at $1,850M and $2,332M respectively.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-20 Hedge Designation.html", + "keywords": [ + "hedges", + "accounting", + "derivatives" + ], + "tokens": [] + } + }, + { + "chunk_id": "01356978-fcda-51ae-9019-3fa6a4125e84", + "type": "text", + "content": "We primarily utilize foreign currency forward contracts to mitigate the impact of foreign currency exchange rate movements on our operating expenses. These foreign currency forward contracts for operating expenses are designated as accounting hedges. Gains or losses on the contracts are recorded in Accumulated other comprehensive income or loss and reclassified to Operating expenses when the related operating expenses are recognized in earnings. During the first quarter of fiscal years 2027 and 2026, the impact of foreign currency forward contracts designated as accounting hedges on other comprehensive income or loss was not significant and all such instruments were determined to be highly effective.\nWe also entered into foreign currency forward contracts to mitigate the impact of foreign currency movements on monetary assets and liabilities. For our foreign currency contracts for assets and liabilities, the change in fair value of these non-designated contracts was recorded in Other income or expense and offsets the change in fair value of the hedged foreign currency denominated monetary assets and liabilities, which was also recorded in Other income (expense), net.\nThe table below presents the notional value of our foreign currency contracts outstanding:\n\n[tables/table-20 Hedge Designation.html]\n\nThe fair values of our foreign currency contracts were not significant as of April 26, 2026 and January 25, 2026.\nAs of April 26, 2026, all foreign currency contracts mature within 18 months. The expected realized gains and losses deferred into Accumulated other comprehensive income or loss related to foreign currency forward contracts within the next twelve months were not significant.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 8 - Derivative Financial Instruments/Foreign Currency Derivatives", + "metadata": { + "length": 1717, + "summary": "The company uses foreign currency forward contracts to hedge against exchange rate fluctuations affecting operating expenses and monetary assets/liabilities. Contracts for operating expenses are designated as accounting hedges, with gains/losses recorded in Accumulated other comprehensive income and reclassified to Operating expenses when recognized. These were highly effective and not significant in Q1 FY2027 and FY2026. Non-designated contracts for assets/liabilities have fair value changes recorded in Other income/expense, offsetting hedged items. Notional values are presented in a table; fair values were not significant as of April 26, 2026 and January 25, 2026. All contracts mature within 18 months, with deferred gains/losses expected to be realized in the next twelve months not significant.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "We", + "primarily", + "utilize", + "foreign", + "currency", + "forward", + "contracts", + "mitigate", + "impact", + "exchange", + "rate", + "movements", + "operating", + "expenses", + "These", + "designated", + "accounting", + "hedges", + "Gains", + "losses", + "recorded", + "Accumulated", + "comprehensive", + "income", + "loss", + "reclassified", + "Operating", + "related", + "recognized", + "earnings", + "During", + "quarter", + "fiscal", + "years", + "2027", + "2026", + "significant", + "instruments", + "determined", + "highly", + "effective", + "entered", + "monetary", + "assets", + "liabilities", + "For", + "change", + "fair", + "Other", + "expense", + "offsets", + "hedged", + "denominated", + "net", + "The", + "table", + "presents", + "notional", + "outstanding", + "values", + "April", + "26", + "January", + "25", + "As", + "mature", + "18", + "months", + "expected", + "realized", + "gains", + "deferred", + "twelve" + ], + "keywords": [ + "foreign currency", + "forward contracts", + "hedging" + ], + "connect_to": [ + { + "target": "a0880006-66e3-5c79-8c8d-09c73661584e", + "relation": "embeds", + "ref": "[tables/table-20 Hedge Designation.html]", + "position": { + "start": 1278, + "end": 1318 + } + } + ] + } + }, + { + "chunk_id": "a52430b7-d936-59cd-9d5d-cd4b2ee1c115", + "type": "table", + "content": "
Expected Remaining Term (years)Effective Interest RateApr 26, 2026Jan 25, 2026
(In millions)
3.20% Notes Due 20260.43.31%$ 1,000$ 1,000
1.55% Notes Due 20282.11.64%1,2501,250
2.85% Notes Due 20303.92.93%1,5001,500
2.00% Notes Due 20315.12.09%1,2501,250
3.50% Notes Due 204013.93.54%1,0001,000
3.50% Notes Due 205023.93.54%2,0002,000
3.70% Notes Due 206034.03.73%500500
Unamortized debt discount and issuance costs(30)(32)
Net carrying amount$ 8,470$ 8,468
Less short-term portion(1,000)(999)
Total long-term portion$ 7,470$ 7,469
", + "path": "tables/table-21 Debt Notes Table.html", + "metadata": { + "length": 1246, + "summary": "table-22\nThe table lists various notes due from 2026 to 2060 with interest rates from 1.55% to 3.70%, remaining terms, and carrying amounts. Net carrying amount is $8,470 million as of Apr 26, 2026, with long-term portion of $7,470 million.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-21 Debt Notes Table.html", + "keywords": [ + "Notes", + "Interest Rates", + "Maturities" + ], + "tokens": [] + } + }, + { + "chunk_id": "83b5aa29-c87e-588b-ac81-1d66305a6b45", + "type": "text", + "content": "In fiscal year 2026, we entered into agreements to guarantee partners' facility lease obligations in the event of their default in exchange for warrants. The maximum gross exposure under all agreements is \\$3.5 billion, which is reduced as the partners make payments to the lessors over terms ranging from 5 to 7 years. The partners have placed \\$712 million in escrow to mitigate our potential exposure. The guarantees, classified as credit derivatives with changes in fair value recognized in Other income (expense), net, were not material.\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)\nNote 9 - Debt\n\n[tables/table-21 Debt Notes Table.html]\n\nAs of April 26, 2026 and January 25, 2026, the estimated fair value of debt was \\$7.4 billion and \\$7.5 billion, respectively. The estimated fair values are based on Level 2 inputs.\nOur notes are unsecured senior obligations. Existing and future liabilities of our subsidiaries will be effectively senior to the notes. Our notes pay interest semi-annually. We may redeem each of our notes prior to maturity, subject to a make-whole premium. The maturity dates of the notes are stated by calendar year.\nAs of April 26, 2026, we complied with the required covenants, which are non-financial in nature, under the outstanding notes.\nAs of April 26, 2026, our commercial paper program had a capacity of \\$25.0 billion, with no amounts outstanding.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 8 - Derivative Financial Instruments/Facility Lease Guarantee", + "metadata": { + "length": 1511, + "summary": "In fiscal year 2026, NVIDIA entered into agreements guaranteeing partners' facility lease obligations with a maximum gross exposure of $3.5 billion, reduced by partner payments over 5-7 years, and $712 million in escrow. The guarantees are classified as credit derivatives with immaterial fair value changes. As of April 26, 2026, the estimated fair value of debt was $7.4 billion (Level 2 inputs). Notes are unsecured senior obligations with semi-annual interest and make-whole redemption. The company complied with non-financial covenants and had a $25.0 billion commercial paper program with no outstanding amounts.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "In", + "fiscal", + "year", + "2026", + "entered", + "agreements", + "guarantee", + "partners", + "facility", + "lease", + "obligations", + "event", + "default", + "exchange", + "warrants", + "The", + "maximum", + "gross", + "exposure", + "3.5", + "billion", + "reduced", + "make", + "payments", + "lessors", + "terms", + "ranging", + "years", + "712", + "million", + "escrow", + "mitigate", + "potential", + "guarantees", + "classified", + "credit", + "derivatives", + "fair", + "recognized", + "Other", + "income", + "expense", + "net", + "material", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Continued", + "Unaudited", + "Note", + "Debt", + "As", + "April", + "26", + "January", + "25", + "estimated", + "debt", + "7.4", + "7.5", + "values", + "based", + "Level", + "inputs", + "Our", + "notes", + "unsecured", + "senior", + "Existing", + "future", + "liabilities", + "subsidiaries", + "effectively", + "pay", + "interest", + "semi", + "annually", + "We", + "redeem", + "prior", + "maturity", + "subject", + "premium", + "dates", + "stated", + "calendar", + "complied", + "required", + "covenants", + "financial", + "nature", + "outstanding", + "commercial", + "paper", + "program", + "capacity", + "25.0", + "amounts" + ], + "keywords": [ + "Debt", + "Guarantees", + "Fair Value" + ], + "connect_to": [ + { + "target": "a52430b7-d936-59cd-9d5d-cd4b2ee1c115", + "relation": "embeds", + "ref": "[tables/table-21 Debt Notes Table.html]", + "position": { + "start": 671, + "end": 710 + } + } + ] + } + }, + { + "chunk_id": "63321a11-501f-50d5-82bc-cbef72d80710", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f48f8892-6b7c-50ac-87a2-ca0cf12d6c84", + "type": "text", + "content": "Manufacturing, supply, and capacity commitments reflect data center-scale production and longer future ordering horizons across current and future product architectures. We enter into agreements with our supply vendors that allow them to procure inventory based upon our defined criteria, and in certain instances, these agreements are cancellable, able to be rescheduled, or adjustable for our business needs prior to placing firm orders. Changes to these agreements may result in additional costs. As of April 26, 2026, these commitments were \\$119 billion for which \\$95 billion will be paid in the remainder of fiscal year 2027 and the remaining balance will be paid in fiscal years 2028 through 2031.\nMulti-year cloud service agreement commitments as of April 26, 2026, were \\$30 billion for which \\$6 billion, \\$7 billion, \\$7 billion, \\$5 billion, \\$3 billion, and \\$2 billion will be paid in the remainder of fiscal year 2027, each fiscal year from 2028 through 2031, and fiscal year 2032 and thereafter, respectively. Cloud service capacity may be reduced or terminated. Cloud service agreements will be primarily used to support our research and development efforts.\nOther vendor commitments were \\$6 billion as of April 26, 2026, of which the majority will be paid through fiscal year 2027.\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies/Commitments", + "metadata": { + "length": 1414, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Manufacturing", + "supply", + "capacity", + "commitments", + "reflect", + "data", + "center", + "scale", + "production", + "longer", + "future", + "ordering", + "horizons", + "current", + "product", + "architectures", + "We", + "enter", + "agreements", + "vendors", + "procure", + "inventory", + "based", + "defined", + "criteria", + "instances", + "cancellable", + "rescheduled", + "adjustable", + "business", + "prior", + "placing", + "firm", + "orders", + "Changes", + "result", + "additional", + "costs", + "As", + "April", + "26", + "2026", + "119", + "billion", + "95", + "paid", + "remainder", + "fiscal", + "year", + "2027", + "remaining", + "balance", + "years", + "2028", + "2031", + "Multi", + "cloud", + "service", + "agreement", + "30", + "2032", + "Cloud", + "reduced", + "terminated", + "primarily", + "support", + "research", + "development", + "efforts", + "Other", + "vendor", + "majority", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Continued", + "Unaudited" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d6303f84-f5ac-5bd8-8d2a-06bb9472d77a", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Balance at beginning of period$ 2,807$ 1,290
Additions330870
Utilization(189)(80)
Balance at end of period$ 2,948$ 2,080
", + "path": "tables/table-22 Balance Changes.html", + "metadata": { + "length": 439, + "summary": "table-23\nThe table shows balance changes over three months ending Apr 26, 2026 and Apr 27, 2025, with beginning balances of $2,807M and $1,290M, additions of $330M and $870M, utilization of $189M and $80M, and ending balances of $2,948M and $2,080M.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-22 Balance Changes.html", + "keywords": [ + "balance", + "additions", + "utilization" + ], + "tokens": [] + } + }, + { + "chunk_id": "44c31497-4131-591c-a746-91db85010656", + "type": "text", + "content": "The estimated amount of product warranty liabilities was \\$2.9 billion and \\$2.8 billion as of April 26, 2026 and January 25, 2026, respectively. The estimated product returns and product warranty activity consisted of the following:\n\n[tables/table-22 Balance Changes.html]\n\nFor the first quarter of fiscal years 2027 and 2026, the additions in product warranty liabilities primarily related to our Compute & Networking segment.\nWe have provided indemnities for matters such as tax, product, and employee liabilities. We have included intellectual property indemnification provisions in our technology-related agreements with third parties. Maximum potential future payments cannot be estimated because many of these agreements do not have a maximum stated liability. We have not recorded any liability in our Condensed Consolidated Financial Statements for such indemnifications.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies/Accrual for Product Warranty Liabilities", + "metadata": { + "length": 883, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "The", + "estimated", + "amount", + "product", + "warranty", + "liabilities", + "2.9", + "billion", + "2.8", + "April", + "26", + "2026", + "January", + "25", + "returns", + "activity", + "consisted", + "For", + "quarter", + "fiscal", + "years", + "2027", + "additions", + "primarily", + "related", + "Compute", + "Networking", + "segment", + "We", + "provided", + "indemnities", + "matters", + "tax", + "employee", + "included", + "intellectual", + "property", + "indemnification", + "provisions", + "technology", + "agreements", + "parties", + "Maximum", + "potential", + "future", + "payments", + "maximum", + "stated", + "liability", + "recorded", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "indemnifications" + ], + "keywords": [], + "connect_to": [ + { + "target": "d6303f84-f5ac-5bd8-8d2a-06bb9472d77a", + "relation": "embeds", + "ref": "[tables/table-22 Balance Changes.html]", + "position": { + "start": 235, + "end": 273 + } + } + ] + } + }, + { + "chunk_id": "aab742aa-9b18-5847-8816-34503e3ece0f", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies/Litigation", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "706584ce-12f5-5b08-bcf7-81f7149d670e", + "type": "text", + "content": "The plaintiffs in the putative securities class action lawsuit, captioned 4:18-cv-07669-HSG, initially filed on December 21, 2018 in the United States District Court for the Northern District of California, and titled In Re NVIDIA Corporation Securities Litigation, filed an amended complaint on May 13, 2020. The amended complaint asserted that NVIDIA and certain NVIDIA executives violated Section 10(b) of the Securities Exchange Act of 1934, as amended, or the Exchange Act, and SEC Rule 10b-5, by making materially false or misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand between May 10, 2017 and November 14, 2018. Plaintiffs also alleged that the NVIDIA executives who they named as defendants violated Section 20(a) of the Exchange Act. Plaintiffs sought class certification, an award of unspecified compensatory damages, an award of reasonable costs and expenses, including attorneys' fees and expert fees, and further relief as the Court may deem just and proper. On March 2, 2021, the district court granted NVIDIA's motion to dismiss the complaint without leave to amend, entered judgment in favor of NVIDIA and closed the case. On March 30, 2021, plaintiffs filed an appeal from judgment in the United States Court of Appeals for the Ninth Circuit, case number 21-15604. On August 25, 2023, a majority of a three-judge Ninth Circuit panel affirmed in part and reversed in part the district court's dismissal of the case, with a third judge dissenting on the basis that the district court did not err in dismissing the case. NVIDIA filed a petition for a writ of certiorari on March 4, 2024. On June 17, 2024, the Supreme Court of the United States granted NVIDIA's petition for a writ of certiorari. After briefing and argument, the Supreme Court dismissed NVIDIA's writ of certiorari as improvidently granted on December 11, 2024, and issued judgment on January 13, 2025. On February 20, 2025, the Ninth Circuit's judgment, entered August 25, 2023 and corrected August 28, 2023, took effect, and the case was remanded to the district court for further proceedings. On March 25, 2026, the district court granted plaintiffs' motion for class certification and certified a class of investors consisting of all persons or entities who purchased or otherwise acquired NVIDIA common stock between August 10, 2017, and November 15, 2018, inclusive, excluding certain persons and entities, such as NVIDIA's officers and directors, and members of their immediate families, among others. On April 8, 2026, NVIDIA filed a petition with the Ninth Circuit for permission to appeal the district court's order pursuant to Federal Rule of Civil Procedure 23(f).\nThe putative derivative lawsuit pending in the United States District Court for the Northern District of California, captioned 4:19-cv-00341-HSG, initially filed January 18, 2019 and titled In re NVIDIA Corporation Consolidated Derivative Litigation, was stayed pending resolution of the plaintiffs' appeal in the In Re NVIDIA Corporation Securities Litigation action. On February 22, 2022, the court administratively closed the case, but stated that it would reopen the case once the appeal in the In Re NVIDIA Corporation Securities Litigation action is resolved. The case has not yet been reopened by the court. The lawsuit asserts claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty, unjust enrichment, waste of corporate assets, and violations of Sections 14(a), 10(b), and 20(a) of the Exchange Act based on the dissemination of allegedly false and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs are seeking unspecified damages and other relief, including reforms and improvements to NVIDIA's corporate governance and internal procedures.\nThe putative derivative actions initially filed September 24, 2019 and pending in the United States District Court for the District of Delaware, Lipchitz v. Huang, et al. (Case No. 1:19-cv-01795-MN) and Nelson v. Huang, et. al. (Case No. 1:19-\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)\ncv-01798-MN), were stayed pending resolution of the plaintiffs' appeal in the In Re NVIDIA Corporation Securities Litigation action. On March 7, 2025, after the Supreme Court issued its judgment dismissing the Company's petition for writ of certiorari as improvidently granted in the In Re NVIDIA Securities Litigation action, the district court adopted the parties' stipulation to extend the stay until the final and complete resolution of the In Re NVIDIA Corporation Securities Litigation action. The lawsuits assert claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty, unjust enrichment, insider trading, misappropriation of information, corporate waste and violations of Sections 14(a), 10(b), and 20(a) of the Exchange Act based on the dissemination of allegedly false and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs seek unspecified damages and other relief, including disgorgement of profits from the sale of NVIDIA stock and unspecified corporate governance measures.\nAnother putative derivative action was filed on October 30, 2023 in the Court of Chancery of the State of Delaware, captioned Horanic v. Huang, et al. (Case No. 2023-1096-KSJM). This lawsuit asserts claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty and insider trading based on the dissemination of allegedly false and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs seek unspecified damages and other relief, including disgorgement of profits from the sale of NVIDIA stock and reform of unspecified corporate governance measures. On August 11, 2025, the court granted the parties' stipulation to voluntarily dismiss with prejudice plaintiff City of Westland Police and Fire Retirement System. This derivative matter is stayed pending the final resolution of In Re NVIDIA Corporation Securities Litigation action.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies/Litigation/Securities Class Action and Derivative Lawsuits", + "metadata": { + "length": 6339, + "summary": "The text details multiple legal actions against NVIDIA Corporation. A securities class action (4:18-cv-07669-HSG) was filed in 2018, alleging false statements about channel inventory and cryptocurrency mining's impact on GPU demand. After dismissal and appeals, the Supreme Court dismissed NVIDIA's certiorari petition in 2024, leading to remand and class certification in 2026. Related derivative lawsuits in California and Delaware, asserting breach of fiduciary duty and other claims, remain stayed pending final resolution of the securities action.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "The", + "plaintiffs", + "putative", + "securities", + "class", + "action", + "lawsuit", + "captioned", + "18", + "cv", + "07669", + "HSG", + "initially", + "filed", + "December", + "21", + "2018", + "United", + "States", + "District", + "Court", + "Northern", + "California", + "titled", + "In", + "Re", + "NVIDIA", + "Corporation", + "Securities", + "Litigation", + "amended", + "complaint", + "May", + "13", + "2020", + "asserted", + "executives", + "violated", + "Section", + "10", + "Exchange", + "Act", + "1934", + "SEC", + "Rule", + "10b", + "making", + "materially", + "false", + "misleading", + "statements", + "related", + "channel", + "inventory", + "impact", + "cryptocurrency", + "mining", + "GPU", + "demand", + "2017", + "November", + "14", + "Plaintiffs", + "alleged", + "named", + "defendants", + "20", + "sought", + "certification", + "award", + "unspecified", + "compensatory", + "damages", + "reasonable", + "costs", + "expenses", + "including", + "attorneys", + "fees", + "expert", + "relief", + "deem", + "proper", + "On", + "March", + "2021", + "district", + "court", + "granted", + "motion", + "dismiss", + "leave", + "amend", + "entered", + "judgment", + "favor", + "closed", + "case", + "30", + "appeal", + "Appeals", + "Ninth", + "Circuit", + "number", + "15604", + "August", + "25", + "2023", + "majority", + "judge", + "panel", + "affirmed", + "part", + "reversed", + "dismissal", + "dissenting", + "basis", + "err", + "dismissing", + "petition", + "writ", + "certiorari", + "2024", + "June", + "17", + "Supreme", + "After", + "briefing", + "argument", + "dismissed", + "improvidently", + "11", + "issued", + "January", + "2025", + "February", + "corrected", + "28", + "effect", + "remanded", + "proceedings", + "2026", + "certified", + "investors", + "consisting", + "persons", + "entities", + "purchased", + "acquired", + "common", + "stock", + "15", + "inclusive", + "excluding", + "officers", + "directors", + "members", + "families", + "April", + "permission", + "order", + "pursuant", + "Federal", + "Civil", + "Procedure", + "23", + "derivative", + "pending", + "19", + "00341", + "2019", + "Consolidated", + "Derivative", + "stayed", + "resolution", + "22", + "2022", + "administratively", + "stated", + "reopen", + "resolved", + "reopened", + "asserts", + "claims", + "purportedly", + "behalf", + "Company", + "breach", + "fiduciary", + "duty", + "unjust", + "enrichment", + "waste", + "corporate", + "assets", + "violations", + "Sections", + "based", + "dissemination", + "allegedly", + "seeking", + "reforms", + "improvements", + "governance", + "internal", + "procedures", + "actions", + "September", + "24", + "Delaware", + "Lipchitz", + "Huang", + "al", + "Case", + "No", + "01795", + "MN", + "Nelson", + "Subsidiaries", + "Notes", + "Condensed", + "Financial", + "Statements", + "Continued", + "Unaudited", + "01798", + "adopted", + "parties", + "stipulation", + "extend", + "stay", + "final", + "complete", + "lawsuits", + "assert", + "insider", + "trading", + "misappropriation", + "information", + "seek", + "disgorgement", + "profits", + "sale", + "measures", + "Another", + "October", + "Chancery", + "State", + "Horanic", + "1096", + "KSJM", + "This", + "reform", + "voluntarily", + "prejudice", + "plaintiff", + "City", + "Westland", + "Police", + "Fire", + "Retirement", + "System", + "matter" + ], + "keywords": [ + "Securities Litigation", + "Derivative Actions", + "Class Certification" + ], + "connect_to": [] + } + }, + { + "chunk_id": "8c15817a-009a-5534-b52e-81215dd05751", + "type": "text", + "content": "As of April 26, 2026, there are no accrued contingent liabilities associated with the legal proceedings described above based on our belief that liabilities, while reasonably possible, are not probable. Further, any possible loss or range of loss in these matters cannot be reasonably estimated at this time. We are engaged in legal actions not described above arising in the ordinary course of business, as well as regulatory and government inquiries and investigations, and, while there can be no assurance of favorable outcomes, we believe that the ultimate outcome of these matters will not have a material adverse effect on our operating results, liquidity or financial position. These matters are subject to inherent uncertainties and if the ultimate outcome is unfavorable, there exists the possibility of a material adverse impact on our operating results, liquidity or financial position in the period the outcome becomes estimable and probable.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies/Litigation/Accounting for Loss Contingencies", + "metadata": { + "length": 954, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "As", + "April", + "26", + "2026", + "accrued", + "contingent", + "liabilities", + "legal", + "proceedings", + "based", + "belief", + "probable", + "Further", + "loss", + "range", + "matters", + "estimated", + "time", + "We", + "engaged", + "actions", + "arising", + "ordinary", + "business", + "regulatory", + "government", + "inquiries", + "investigations", + "assurance", + "favorable", + "outcomes", + "ultimate", + "outcome", + "material", + "adverse", + "effect", + "operating", + "results", + "liquidity", + "financial", + "position", + "These", + "subject", + "inherent", + "uncertainties", + "unfavorable", + "exists", + "possibility", + "impact", + "period", + "estimable" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5dc664dc-0f25-5e78-a0b9-4eb8d86bcb79", + "type": "text", + "content": "Income tax expense was \\$11.6 billion and \\$3.1 billion for the first quarter of fiscal years 2027 and 2026, respectively. Income tax as a percentage of income before income tax was 16.6% and 14.3% for the first quarter of fiscal years 2027 and 2026, respectively.\nThe effective tax rate increased primarily due to a lower percentage of tax benefits from stock-based compensation relative to the increase in income before income tax.\nOur effective tax rates for the first quarter of fiscal years 2027 and 2026 were lower than the U.S. federal statutory rate of 21% primarily due to tax benefits from foreign-derived deduction eligible income, income earned in jurisdictions that were subject to taxes at rates lower than the U.S. federal statutory tax rate, stock-based compensation, and the U.S. federal research tax credit.\nWhile we believe that we have adequately provided for all uncertain tax positions, or tax positions where we believe it is not more-likely-than-not that the position will be sustained upon review, amounts asserted by tax authorities could be greater or less than our accrued position. Accordingly, our provisions on federal, state and foreign tax related matters to be recorded in the future may change as revised estimates are made or the underlying matters are settled or otherwise resolved with the respective tax authorities.\nWe are currently under examination by the Internal Revenue Service for our fiscal years 2023 and 2024.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 11 - Income Taxes", + "metadata": { + "length": 1458, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Income", + "tax", + "expense", + "11.6", + "billion", + "3.1", + "quarter", + "fiscal", + "years", + "2027", + "2026", + "percentage", + "income", + "16.6%", + "14.3%", + "The", + "effective", + "rate", + "increased", + "primarily", + "due", + "lower", + "benefits", + "stock", + "based", + "compensation", + "relative", + "increase", + "Our", + "rates", + "federal", + "statutory", + "21%", + "foreign", + "derived", + "deduction", + "eligible", + "earned", + "jurisdictions", + "subject", + "taxes", + "research", + "credit", + "While", + "adequately", + "provided", + "uncertain", + "positions", + "position", + "sustained", + "review", + "amounts", + "asserted", + "authorities", + "greater", + "accrued", + "Accordingly", + "provisions", + "state", + "related", + "matters", + "recorded", + "future", + "change", + "revised", + "estimates", + "made", + "underlying", + "settled", + "resolved", + "respective", + "We", + "examination", + "Internal", + "Revenue", + "Service", + "2023", + "2024" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4bb0997c-62f4-56a8-829a-997fc3252e66", + "type": "text", + "content": "Capital Return Program\nWe repurchased 108 million and 126 million shares of our common stock for \\$20.2 billion and \\$14.5 billion during the first quarter of fiscal years 2027 and 2026, respectively. As of April 26, 2026, we were authorized, subject to certain specifications, to repurchase up to \\$38.5 billion of our common stock.\nOn May 18, 2026, our Board of Directors approved an additional \\$80.0 billion in share repurchase authorization, without expiration.\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)\nWe paid cash dividends to our shareholders of \\$243 million and \\$244 million during the first quarter of fiscal years 2027 and 2026, respectively. On May 18, 2026, we increased our quarterly cash dividend from \\$0.01 per share to \\$0.25 per share to all shareholders of record on June 4, 2026. Our quarterly cash dividend will be paid on June 26, 2026.\nThe payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 12 - Shareholders' Equity", + "metadata": { + "length": 1113, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Capital", + "Return", + "Program", + "We", + "repurchased", + "108", + "million", + "126", + "shares", + "common", + "stock", + "20.2", + "billion", + "14.5", + "quarter", + "fiscal", + "years", + "2027", + "2026", + "As", + "April", + "26", + "authorized", + "subject", + "specifications", + "repurchase", + "38.5", + "On", + "May", + "18", + "Board", + "Directors", + "approved", + "additional", + "80.0", + "share", + "authorization", + "expiration", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Continued", + "Unaudited", + "paid", + "cash", + "dividends", + "shareholders", + "243", + "244", + "increased", + "quarterly", + "dividend", + "0.01", + "0.25", + "record", + "June", + "Our", + "The", + "payment", + "future", + "continuing", + "determination", + "declaration", + "interests" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0d1c0272-9cc9-5720-8ea0-8b60982df4b9", + "type": "table", + "content": "
Compute & NetworkingGraphicsTotal
(In millions)
Three Months Ended Apr 26, 2026
Revenue$ 74,550$ 7,065$ 81,615
Other segment items (1)21,2154,12425,339
Operating income$ 53,335$ 2,941$ 56,276
Three Months Ended Apr 27, 2025
Revenue$ 39,589$ 4,473$ 44,062
Other segment items (1)17,5352,83320,368
Operating income$ 22,054$ 1,640$ 23,694
", + "path": "tables/table-23 Segment Revenue.html", + "metadata": { + "length": 774, + "summary": "table-24\nThe table shows Compute & Networking and Graphics segment revenue, other items, and operating income for three-month periods ending April 2026 and April 2025, with totals.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-23 Segment Revenue.html", + "keywords": [ + "Revenue", + "Operating Income", + "Segment" + ], + "tokens": [] + } + }, + { + "chunk_id": "380f5b46-c989-504f-84e8-ea97b6651a4e", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Segment operating income$ 56,276$ 23,694
Stock-based compensation expense(1,928)(1,474)
Unallocated operating expenses(565)(419)
Acquisition-related and other costs(247)(163)
Interest income540515
Interest expense(102)(63)
Other income (expense), net15,929(180)
Consolidated income before income tax$ 69,903$ 21,910
", + "path": "tables/table-24 Income Statement.html", + "metadata": { + "length": 774, + "summary": "table-25\nThe table shows a consolidated income statement for three months ended Apr 26, 2026 and Apr 27, 2025, with segment operating income of $56,276M and $23,694M, leading to consolidated income before tax of $69,903M and $21,910M.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-24 Income Statement.html", + "keywords": [ + "Segment operating income", + "Consolidated income", + "Tax" + ], + "tokens": [] + } + }, + { + "chunk_id": "42ce1654-35c3-59ef-a13a-aa569eff10fc", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Geographic Revenue based upon Customer Headquarters Location:
United States$ 63,769$ 25,685
Taiwan12,0067,648
China (including Hong Kong)4,5509,659
Other1,2901,070
Total revenue$ 81,615$ 44,062
", + "path": "tables/table-25 Revenue by Region.html", + "metadata": { + "length": 576, + "summary": "table-26\nThe table shows geographic revenue for three months ended Apr 26, 2026 and Apr 27, 2025. US revenue was $63,769M (2026) vs $25,685M (2025); Taiwan $12,006M vs $7,648M; China $4,550M vs $9,659M; Other $1,290M vs $1,070M. Total revenue $81,615M vs $44,062M.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-25 Revenue by Region.html", + "keywords": [ + "Revenue", + "Geographic", + "Quarterly" + ], + "tokens": [] + } + }, + { + "chunk_id": "e89ea2a1-ad27-5fb5-9994-86c817ccd875", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Revenue by Market Platform
Data Center$ 75,246$ 39,112
Hyperscale37,86917,599
AI Clouds, Industrial, & Enterprise37,37721,513
Edge Computing6,3694,950
Total revenue$ 81,615$ 44,062
", + "path": "tables/table-26 Revenue by Market.html", + "metadata": { + "length": 573, + "summary": "table-27\nThe table shows revenue for three months ended Apr 26, 2026 and Apr 27, 2025. Data Center revenue grew from $39,112M to $75,246M, with Hyperscale and AI Clouds segments contributing significantly. Total revenue increased from $44,062M to $81,615M.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-26 Revenue by Market.html", + "keywords": [ + "Data Center", + "Hyperscale", + "Revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "6513b862-0e62-50f4-aabf-64c82356f43a", + "type": "text", + "content": "Our Chief Executive Officer is our chief operating decision maker, or CODM, and reviews financial information presented on an operating segment basis for purposes of making decisions and assessing financial performance. Our CODM assesses operating performance of each segment based on regularly provided segment revenue and segment operating income. Operating results by segment include costs or expenses directly attributable to each segment, and costs or expenses that are leveraged across our unified architecture and therefore allocated between our two segments. Our CODM reviews expenses on a consolidated basis, and expenses attributable to each segment are not regularly provided to our CODM.\nThe Compute & Networking segment includes our Data Center accelerated computing and networking platforms and AI solutions and software, and automotive platforms and autonomous and electric vehicle solutions including software.\nThe Graphics segment includes GeForce GPUs for gaming and PCs, and Quadro/NVIDIA RTX GPUs for enterprise workstation graphics.\nCertain expenses are not allocated to either Compute & Networking or Graphics for purposes of making operating decisions or assessing financial performance. The expenses include stock-based compensation expense, corporate infrastructure and support costs, acquisition-related and other costs, and other non-recurring charges and benefits that our CODM deems to be enterprise in nature.\nOur CODM does not review any information regarding total assets on a reportable segment basis. There are no intersegment transactions. The accounting policies for segment reporting are the same as for our consolidated financial statements. The table below presents details of our reportable segments.\n\n[tables/table-23 Segment Revenue.html]\n\n(1) Other segment items primarily include product costs and inventory provisions, compensation and benefits excluding stock-based compensation expense, computing infrastructure expenses, and engineering development costs.\nDepreciation and amortization expense attributable to our Compute & Networking segment was \\$526 million and \\$296 million for the first quarter of fiscal years 2027 and 2026, respectively. Depreciation and amortization expense attributable to our Graphics segment was \\$194 million and \\$109 million for the first quarter of fiscal years 2027 and 2026, respectively. Acquisition-related intangible amortization expense is not allocated to either Compute & Networking or Graphics for purposes of making operating decisions or assessing financial performance.\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)\nReconciliation of segment operating income to consolidated income before income tax for the first quarter of fiscal years 2027 and 2026 was as follows:\n\n[tables/table-24 Income Statement.html]\n\nRevenue by geographic area is based upon the location of the customers' headquarters. The end customer and shipping location may be different from our customers' headquarters location.\n\n[tables/table-25 Revenue by Region.html]\n\nRevenue from sales to customers headquartered outside of the United States accounted for 22% of total revenue for the first quarter of fiscal year 2027 and 42% of total revenue for the first quarter of fiscal year 2026.\nWe refer to customers who purchase products directly from NVIDIA as direct customers, such as AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. We refer to indirect customers as those who purchase products through our direct customers; indirect customers include CSPs, AI Clouds, AI model makers, enterprises, and public sector entities. Our revenue is concentrated among a limited number of direct and indirect customers and this trend may continue.\nFor the first quarter of fiscal year 2027, three direct customers represented 21%, 17%, and 16% of total revenue, all of which was primarily attributable to the Compute & Networking segment.\nFor the first quarter of fiscal year 2026, sales to two direct customers represented 16% and 14% of total revenue, which were attributable to the Compute & Networking segment.\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)\nIn the first quarter of fiscal year 2027, we changed our presentation of revenue by market platform, and the comparable period has been recast as follows:\n\n[tables/table-26 Revenue by Market.html]", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 13 - Segment Information", + "metadata": { + "length": 4616, + "summary": "NVIDIA's CODM reviews financial information on an operating segment basis, with two reportable segments: Compute & Networking (Data Center, AI, automotive) and Graphics (GeForce, Quadro). Segment operating income is assessed, but certain expenses like stock-based compensation are not allocated. Depreciation for Compute & Networking was $526M (FY2027 Q1) vs $296M (FY2026 Q1); Graphics was $194M vs $109M. Revenue concentration exists: three direct customers accounted for 21%, 17%, and 16% of total revenue in FY2027 Q1, primarily in Compute & Networking. Non-US revenue was 22% in FY2027 Q1 vs 42% in FY2026 Q1. The presentation of revenue by market platform changed in FY2027 Q1.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Our", + "Chief", + "Executive", + "Officer", + "chief", + "operating", + "decision", + "maker", + "CODM", + "reviews", + "financial", + "information", + "presented", + "segment", + "basis", + "purposes", + "making", + "decisions", + "assessing", + "performance", + "assesses", + "based", + "regularly", + "provided", + "revenue", + "income", + "Operating", + "results", + "include", + "costs", + "expenses", + "directly", + "attributable", + "leveraged", + "unified", + "architecture", + "allocated", + "segments", + "consolidated", + "The", + "Compute", + "Networking", + "includes", + "Data", + "Center", + "accelerated", + "computing", + "networking", + "platforms", + "AI", + "solutions", + "software", + "automotive", + "autonomous", + "electric", + "vehicle", + "including", + "Graphics", + "GeForce", + "GPUs", + "gaming", + "PCs", + "Quadro", + "NVIDIA", + "RTX", + "enterprise", + "workstation", + "graphics", + "Certain", + "stock", + "compensation", + "expense", + "corporate", + "infrastructure", + "support", + "acquisition", + "related", + "recurring", + "charges", + "benefits", + "deems", + "nature", + "review", + "total", + "assets", + "reportable", + "There", + "intersegment", + "transactions", + "accounting", + "policies", + "reporting", + "statements", + "table", + "presents", + "details", + "Other", + "items", + "primarily", + "product", + "inventory", + "provisions", + "excluding", + "engineering", + "development", + "Depreciation", + "amortization", + "526", + "million", + "296", + "quarter", + "fiscal", + "years", + "2027", + "2026", + "194", + "109", + "Acquisition", + "intangible", + "Corporation", + "Subsidiaries", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Continued", + "Unaudited", + "Reconciliation", + "tax", + "Revenue", + "geographic", + "area", + "location", + "customers", + "headquarters", + "end", + "customer", + "shipping", + "sales", + "headquartered", + "United", + "States", + "accounted", + "22%", + "year", + "42%", + "We", + "refer", + "purchase", + "products", + "direct", + "AIBs", + "distributors", + "ODMs", + "OEMs", + "CSPs", + "model", + "makers", + "system", + "integrators", + "internal", + "resources", + "party", + "complete", + "build", + "indirect", + "Clouds", + "enterprises", + "public", + "sector", + "entities", + "concentrated", + "limited", + "number", + "trend", + "continue", + "For", + "represented", + "21%", + "17%", + "16%", + "14%", + "In", + "changed", + "presentation", + "market", + "platform", + "comparable", + "period", + "recast" + ], + "keywords": [ + "Segment", + "Revenue", + "Operating Income" + ], + "connect_to": [ + { + "target": "0d1c0272-9cc9-5720-8ea0-8b60982df4b9", + "relation": "embeds", + "ref": "[tables/table-23 Segment Revenue.html]", + "position": { + "start": 1742, + "end": 1780 + } + }, + { + "target": "380f5b46-c989-504f-84e8-ea97b6651a4e", + "relation": "embeds", + "ref": "[tables/table-24 Income Statement.html]", + "position": { + "start": 2829, + "end": 2868 + } + }, + { + "target": "42ce1654-35c3-59ef-a13a-aa569eff10fc", + "relation": "embeds", + "ref": "[tables/table-25 Revenue by Region.html]", + "position": { + "start": 3056, + "end": 3096 + } + }, + { + "target": "e89ea2a1-ad27-5fb5-9994-86c817ccd875", + "relation": "embeds", + "ref": "[tables/table-26 Revenue by Market.html]", + "position": { + "start": 4552, + "end": 4592 + } + } + ] + } + }, + { + "chunk_id": "55f90935-b9ec-5196-b6ce-f2d906604312", + "type": "table", + "content": "
Operating Lease Obligations (In millions)
Fiscal Year:
2027 (excluding the first quarter of fiscal 2027)$ 460
2028626
2029602
2030530
2031462
2032 and thereafter2,924
Total5,604
Less imputed interest1,260
Present value of net future minimum lease payments4,344
Less short-term operating lease liabilities466
Long-term operating lease liabilities$ 3,878
", + "path": "tables/table-27 Lease Obligations.html", + "metadata": { + "length": 709, + "summary": "table-28\nThe table shows operating lease obligations totaling $5,604 million, with a present value of $4,344 million after imputed interest, resulting in long-term liabilities of $3,878 million.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-27 Lease Obligations.html", + "keywords": [ + "Operating Lease", + "Future Payments", + "Present Value" + ], + "tokens": [] + } + }, + { + "chunk_id": "baddd1e0-f79a-5bb3-93c8-e8e958deea50", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Supplemental cash flows information
Operating cash flow used for operating leases$ 185$ 96
Operating lease assets obtained in exchange for lease obligations$ 1,516$ 98
", + "path": "tables/table-28 Lease Cash Flows.html", + "metadata": { + "length": 445, + "summary": "table-29\nThe table shows supplemental cash flow information for three months ended Apr 26, 2026 and Apr 27, 2025. Operating cash flow used for leases was $185M and $96M; lease assets obtained were $1,516M and $98M.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-28 Lease Cash Flows.html", + "keywords": [ + "operating leases", + "cash flows", + "lease obligations" + ], + "tokens": [] + } + }, + { + "chunk_id": "f496b9df-fbd9-594a-ac54-2e242b10fd19", + "type": "text", + "content": "Our lease obligations primarily consist of operating leases for our data centers and offices, with lease periods expiring between fiscal years 2027 and 2075.\nFuture minimum lease obligations under our non-cancelable lease agreements as of April 26, 2026 were as follows:\n\n[tables/table-27 Lease Obligations.html]\n\nBetween the second quarter of fiscal year 2027 and fiscal year 2033, we expect to commence leases with future obligations of \\$32.4 billion, primarily for data center leases to support our research and development efforts, with lease terms of 3 to 20 years.\nOperating lease costs were \\$171 million and \\$101 million for the first quarter of fiscal years 2027 and 2026, respectively. Short-term, variable, and finance lease costs for the first quarter of fiscal years 2027 and 2026 were not significant.\nOther information related to leases was as follows:\n\n[tables/table-28 Lease Cash Flows.html]\n\nNVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)\nAs of April 26, 2026, our operating leases have a weighted average remaining lease term of 10.4 years and a weighted average discount rate of 4.61%. As of January 25, 2026, our operating leases had a weighted average remaining lease term of 8.8 years and a weighted average discount rate of 4.38%.\nItem 2. Management's Discussion and Analysis of Financial Condition and Results of Operations", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 14 - Leases", + "metadata": { + "length": 1424, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Our", + "lease", + "obligations", + "primarily", + "consist", + "operating", + "leases", + "data", + "centers", + "offices", + "periods", + "expiring", + "fiscal", + "years", + "2027", + "2075", + "Future", + "minimum", + "cancelable", + "agreements", + "April", + "26", + "2026", + "Between", + "quarter", + "year", + "2033", + "expect", + "commence", + "future", + "32.4", + "billion", + "center", + "support", + "research", + "development", + "efforts", + "terms", + "20", + "Operating", + "costs", + "171", + "million", + "101", + "Short", + "term", + "variable", + "finance", + "significant", + "Other", + "information", + "related", + "NVIDIA", + "Corporation", + "Subsidiaries", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Continued", + "Unaudited", + "As", + "weighted", + "average", + "remaining", + "10.4", + "discount", + "rate", + "4.61%", + "January", + "25", + "8.8", + "4.38%", + "Item", + "Management", + "Discussion", + "Analysis", + "Condition", + "Results", + "Operations" + ], + "keywords": [], + "connect_to": [ + { + "target": "55f90935-b9ec-5196-b6ce-f2d906604312", + "relation": "embeds", + "ref": "[tables/table-27 Lease Obligations.html]", + "position": { + "start": 272, + "end": 312 + } + }, + { + "target": "baddd1e0-f79a-5bb3-93c8-e8e958deea50", + "relation": "embeds", + "ref": "[tables/table-28 Lease Cash Flows.html]", + "position": { + "start": 871, + "end": 910 + } + } + ] + } + }, + { + "chunk_id": "ce2aa934-f8d5-599d-a632-a902f20a807a", + "type": "text", + "content": "This Quarterly Report on Form 10-Q contains forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the “safe harbor” created by those sections based on management’s beliefs and assumptions and on information currently available to management. In some cases, you can identify forward-looking statements by terms such as “may,” “will,” “should,” “could,” “goal,” “would,” “expect,” “plan,” “anticipate,” “believe,” “estimate,” “project,” “predict,” “potential” and similar expressions intended to identify forward-looking statements. These statements involve known and unknown risks, uncertainties and other factors, which may cause our actual results, performance, time frames or achievements to be materially different from any future results, performance, time frames or achievements expressed or implied by the forward-looking statements. We discuss many of these risks, uncertainties and other factors in this Quarterly Report on Form 10-Q and our Annual Report on Form 10-K for the fiscal year ended January 25, 2026 in greater detail under the heading “Risk Factors” of such reports. Given these risks, uncertainties, and other factors, you should not place undue reliance on these forward-looking statements. Also, these forward-looking statements represent our estimates and assumptions only as of the date of this filing. You should read this Quarterly Report on Form 10-Q completely and understand that our actual future results may be materially different from what we expect. We hereby qualify our forward-looking statements by these cautionary statements. Except as required by law, we assume no obligation to update these forward-looking statements publicly, or to update the reasons actual results could differ materially from those anticipated in these forward-looking statements, even if new information becomes available in the future.\nAll references to “NVIDIA,” “we,” “us,” “our” or the “Company” mean NVIDIA Corporation and its subsidiaries.\nIn addition, statements that “we believe” and similar statements reflect our beliefs and opinions on the relevant subject. These statements are based upon information available to us as of the filing date of this Quarterly Report on Form 10-Q, and while we believe such information forms a reasonable basis for such statements, such information may be limited or incomplete, and our statements should not be read to indicate that we have conducted an exhaustive inquiry into, or review of, all potentially available relevant information. These statements are inherently uncertain, and investors are cautioned not to unduly rely upon these statements.\n© 2026 NVIDIA Corporation. All rights reserved.\nThe following discussion and analysis of our financial condition and results of operations should be read in conjunction with the risk factors set forth in Item 1A. \"Risk Factors\" of our Annual Report on Form 10-K for the fiscal year ended January 25, 2026 and Part II, Item 1A. \"Risk Factors\" of this Quarterly Report on Form 10-Q and our Condensed Consolidated Financial Statements and related Notes thereto, as well as other cautionary statements and risks described elsewhere in this Quarterly Report on Form 10-Q and our other filings with the SEC, before deciding to purchase, hold, or sell shares of our common stock.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements", + "metadata": { + "length": 3423, + "summary": "This passage contains forward-looking statements under the Securities Act and Exchange Act, based on management's beliefs and assumptions. It identifies terms like 'may,' 'will,' 'expect,' and 'anticipate' as indicators of such statements. Actual results may differ materially due to known and unknown risks, uncertainties, and factors discussed in the company's Annual Report on Form 10-K and this Quarterly Report. The company assumes no obligation to update these statements. References to 'NVIDIA' or 'we' mean NVIDIA Corporation and its subsidiaries. The text also cautions against undue reliance on beliefs and opinions expressed, which are based on information available as of the filing date.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "This", + "Quarterly", + "Report", + "Form", + "10", + "forward", + "statements", + "meaning", + "Section", + "27A", + "Securities", + "Act", + "1933", + "amended", + "21E", + "Exchange", + "1934", + "subject", + "safe", + "harbor", + "created", + "sections", + "based", + "management", + "beliefs", + "assumptions", + "information", + "In", + "cases", + "identify", + "terms", + "goal", + "expect", + "plan", + "anticipate", + "estimate", + "project", + "predict", + "potential", + "similar", + "expressions", + "intended", + "These", + "involve", + "unknown", + "risks", + "uncertainties", + "factors", + "actual", + "results", + "performance", + "time", + "frames", + "achievements", + "materially", + "future", + "expressed", + "implied", + "We", + "discuss", + "Annual", + "fiscal", + "year", + "ended", + "January", + "25", + "2026", + "greater", + "detail", + "heading", + "Risk", + "Factors", + "reports", + "Given", + "place", + "undue", + "reliance", + "Also", + "represent", + "estimates", + "date", + "filing", + "You", + "read", + "completely", + "understand", + "qualify", + "cautionary", + "Except", + "required", + "law", + "assume", + "obligation", + "update", + "publicly", + "reasons", + "differ", + "anticipated", + "All", + "references", + "NVIDIA", + "Company", + "Corporation", + "subsidiaries", + "addition", + "reflect", + "opinions", + "relevant", + "forms", + "reasonable", + "basis", + "limited", + "incomplete", + "conducted", + "exhaustive", + "inquiry", + "review", + "potentially", + "inherently", + "uncertain", + "investors", + "cautioned", + "unduly", + "rely", + "rights", + "reserved", + "The", + "discussion", + "analysis", + "financial", + "condition", + "operations", + "conjunction", + "risk", + "set", + "Item", + "1A", + "Part", + "II", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "related", + "Notes", + "thereto", + "filings", + "SEC", + "deciding", + "purchase", + "hold", + "sell", + "shares", + "common", + "stock" + ], + "keywords": [ + "forward-looking statements", + "risk factors", + "cautionary" + ], + "connect_to": [] + } + }, + { + "chunk_id": "737c3375-fd4b-5d9f-bb69-6bbfd4a18dfb", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Overview", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b35f7cf1-de14-5f02-aee9-4da7d30d83c0", + "type": "text", + "content": "NVIDIA pioneered accelerated computing to help solve the most challenging computational problems. Since our original focus on PC graphics, we have expanded to several other large and important computationally intensive fields. Fueled by the sustained demand for exceptional 3D graphics and the scale of the gaming market, NVIDIA has leveraged its GPU architecture to create platforms for scientific computing, AI, data science, autonomous vehicles, robotics, and digital twin applications. NVIDIA is now a data center-scale AI infrastructure company reshaping all industries.\nOur two operating segments are “Compute & Networking” and “Graphics.” Refer to Note 13 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q for additional information.\nHeadquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Overview/Our Company and Our Businesses", + "metadata": { + "length": 946, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "NVIDIA", + "pioneered", + "accelerated", + "computing", + "solve", + "challenging", + "computational", + "problems", + "Since", + "original", + "focus", + "PC", + "graphics", + "expanded", + "large", + "important", + "computationally", + "intensive", + "fields", + "Fueled", + "sustained", + "demand", + "exceptional", + "3D", + "scale", + "gaming", + "market", + "leveraged", + "GPU", + "architecture", + "create", + "platforms", + "scientific", + "AI", + "data", + "science", + "autonomous", + "vehicles", + "robotics", + "digital", + "twin", + "applications", + "center", + "infrastructure", + "company", + "reshaping", + "industries", + "Our", + "operating", + "segments", + "Compute", + "Networking", + "Graphics", + "Refer", + "Note", + "13", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Part", + "Item", + "Quarterly", + "Report", + "Form", + "10", + "additional", + "information", + "Headquartered", + "Santa", + "Clara", + "California", + "incorporated", + "April", + "1993", + "reincorporated", + "Delaware", + "1998" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "acfd6c99-73da-5b84-9792-14f4c7e29973", + "type": "table", + "content": "
Three Months EndedQuarter-over-Quarter ChangeYear-over-Year Change
Apr 26, 2026Jan 25, 2026Apr 27, 2025
($ in millions, except per share data)
Revenue$ 81,615$ 68,127$ 44,06220 %85 %
Gross margin74.9 %75.0 %60.5 %(0.1) pts14.4 pts
Operating expenses$ 7,621$ 6,794$ 5,03012 %52 %
Operating income$ 53,536$ 44,299$ 21,63821 %147 %
Net income$ 58,321$ 42,960$ 18,77536 %211 %
Net income per diluted share$ 2.39$ 1.76$ 0.7636 %214 %
", + "path": "tables/table-29 Quarterly Financials.html", + "metadata": { + "length": 1140, + "summary": "table-30\nThe table shows Q1 2026 revenue of $81.6B, up 85% YoY, net income of $58.3B (up 211% YoY), and gross margin of 74.9%.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-29 Quarterly Financials.html", + "keywords": [ + "Revenue", + "Net Income", + "Gross Margin" + ], + "tokens": [] + } + }, + { + "chunk_id": "2e674fb5-1d8c-510c-a615-afc4728e6150", + "type": "table", + "content": "
Three Months EndedQuarter-over-Quarter ChangeYear-over-Year Change
Apr 26, 2026Jan 25, 2026Apr 27, 2025
($ in millions)
Revenue by Market Platform (1)
Data Center$ 75,246$ 62,314$ 39,11221 %92 %
Hyperscale37,86933,81417,59912 %115 %
AI Clouds, Industrial, & Enterprise37,37728,50021,51331 %74 %
Edge Computing6,3695,8134,95010 %29 %
Total revenue$ 81,615$ 68,127$ 44,06220 %85 %
", + "path": "tables/table-30 Revenue by Market.html", + "metadata": { + "length": 893, + "summary": "table-31\nData Center revenue rose 92% year-over-year to $75,246M, with Hyperscale up 115% and AI Clouds up 74%. Total revenue reached $81,615M, a 20% quarterly and 85% yearly increase.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-30 Revenue by Market.html", + "keywords": [ + "Data Center", + "Revenue Growth", + "Quarterly Results" + ], + "tokens": [] + } + }, + { + "chunk_id": "9032a1e3-7f4d-5462-9cf1-b549c507a289", + "type": "text", + "content": "Revenue growth in the first quarter was driven by data center products for accelerated computing and AI solutions. Blackwell continued to account for the majority of our system shipments.\nThe availability of data centers, energy, and capital to support the buildout of NVIDIA AI infrastructure by our customers and partners is crucial, and any shortage of these or other necessary resources could impact our future revenue and financial performance. Expanding energy capacity to meet demand is a complex, multi-year process that involves significant regulatory, technical, and construction challenges. In addition, access to capital can be particularly constrained for less-capitalized companies, which may face difficulties securing financing for large-scale infrastructure projects. These limitations could delay customer and partner deployments or reduce the scale of accelerated computing and AI adoption.\nWe expect our Rubin platform to start shipping in the second half of fiscal year 2027. The complexity of bringing up our product architecture and sophisticated system configurations has caused and may in the future cause delays in production and create challenges in managing supply and demand. This could further result in revenue volatility, quality issues, increased inventory provisions, decreases in product yields and higher material costs, and/or increased warranty costs. Customers may postpone purchasing new architectures or may adopt new technologies more gradually than anticipated, affecting our revenue timing and supply chain expenses.\nBeginning in February 2026, the U.S. government, or USG, granted licenses that allow us to ship small amounts of H200 products to specific China-based customers. To date, we have not generated any revenue under the H200 licensing program, and do not yet know whether any imports will be allowed into China. The license requires that the H200s go through an inspection process in the United States prior to any shipment to the customer. As a result, any H200 shipped under the new licensing program will be subject to a 25% tariff upon importation into the United States.\nThe recent rise in high-quality, open-source foundation models is making advanced AI capabilities broadly accessible. Open-source AI is dependent on developer adoption, and if deployed on our competitors' platforms, it could reduce demand for our products and services.\nWe have made, and expect to continue making, investments in our ecosystem to enhance our growth opportunities, cultivate our ecosystem, and strengthen our competitive position. In the first quarter of fiscal year 2027, we made the following investments:\n- \\$18.6 billion in private companies and infrastructure funds. Some of these investments include AI model makers that may indirectly purchase or use our products in the cloud.\n• We made investments in publicly-held equity securities where the value may fluctuate significantly and could adversely affect our financial results.\nOur global supply chain for our networking products, including our Israel operations of approximately 5,900 employees supporting research and development, operations, and sales and marketing, has not been significantly impacted by the conflict in the Middle East. If the conflict escalates or extends, it could affect future product development, supply chain, and revenue, and create business uncertainty.\nMacroeconomic factors, including tariffs, inflation, interest changes, capital market volatility, global supply chain constraints, and global economic and geopolitical developments and conflicts, have direct and indirect impacts on our results of operations, particularly demand for our products. While difficult to isolate and quantify, these macroeconomic factors impact our supply chain and manufacturing costs, employee wages, costs for capital equipment, the value of our investments, revenue, and competitive position. Our product and solution pricing generally does not fluctuate with short-term changes in our costs. Within our supply chain, we continuously manage product availability and costs with our vendors.\nRefer to Part II, Item 1A, \"Risk Factors\" for a discussion of these factors and other risks.\nFirst Quarter of Fiscal Year 2027 Summary\n\n[tables/table-29 Quarterly Financials.html]\n\nWe specialize in markets where our computing platforms can provide tremendous acceleration for applications. These platforms incorporate processors, interconnects, software, algorithms, systems, and services to deliver unique value.\nFollowing the rapid evolution in our businesses, we are transitioning to a new reporting framework that better reflects our current and future growth drivers.\nWe will have two market platforms – Data Center and Edge Computing.\nWithin Data Center, we will report two sub-markets, Hyperscale and ACIE which incorporates AI Clouds, Industrial, and Enterprise. Hyperscale will include revenue from the public clouds and the world's largest consumer internet companies,\nwhile ACIE addresses our growth opportunity in diverse AI purpose-built data centers and AI factories across industries and countries.\nEdge Computing highlights devices for agentic and physical AI including PCs, game consoles, workstations, AI-RAN base stations, robotics and automotive.\n\n[tables/table-30 Revenue by Market.html]\n\n(1) In the first quarter of fiscal year 2027, we changed our presentation of revenue by market platform and the comparable periods have been recast.\nRevenue was \\$81.6 billion, up 85% from a year ago and up 20% sequentially.\nData Center revenue was \\$75.2 billion, up 92% from a year ago and up 21% sequentially, driven by the ramp of our Blackwell 300 products and demand for our InfiniBand, Spectrum-X Ethernet, and NVLink solutions. Hyperscaler revenue increased sequentially and remained at approximately 50% of Data Center revenue, while the remaining 50% came from a continued diversification of customers, including AI Clouds, industrial, enterprise, and sovereign customers. No shipments of Data Center Hopper products to China occurred during the quarter, compared with \\$4.6 billion in the first quarter of fiscal year 2026.\nEdge Computing revenue for the first quarter was \\$6.4 billion, up 29% from a year ago and up 10% sequentially. The increases were driven by robust Blackwell workstation demand, partially offset by slower consumer PC demand that was tempered by elevated memory and systems prices.\nGross margin increased from a year ago on lower inventory provisions, primarily due to the prior year's \\$4.5 billion charge associated with H20 excess inventory and purchase obligations. Gross margin was approximately flat sequentially as our Blackwell architecture remains the majority of our revenue.\nOperating expenses were up 52% from a year ago and up 12% sequentially. The increases were primarily driven by higher compensation and benefits expense due to employee growth and compensation increases, compute and infrastructure costs, and engineering development materials for new product developments.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Overview/Recent Developments, Future Objectives and Challenges", + "metadata": { + "length": 7141, + "summary": "NVIDIA's first quarter of fiscal year 2027 saw revenue of $81.6 billion, up 85% year-over-year, driven by Data Center products for accelerated computing and AI, particularly Blackwell architecture. Data Center revenue was $75.2 billion, with hyperscalers accounting for about 50%. Edge Computing revenue was $6.4 billion. The company faces risks including supply chain constraints, geopolitical issues, tariffs, and competition from open-source AI. Investments totaled $18.6 billion in private companies and infrastructure. No H200 revenue from China has been generated yet, and a new reporting framework with Data Center and Edge Computing platforms is being adopted.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Revenue", + "growth", + "quarter", + "driven", + "data", + "center", + "products", + "accelerated", + "computing", + "AI", + "solutions", + "Blackwell", + "continued", + "account", + "majority", + "system", + "shipments", + "The", + "availability", + "centers", + "energy", + "capital", + "support", + "buildout", + "NVIDIA", + "infrastructure", + "customers", + "partners", + "crucial", + "shortage", + "resources", + "impact", + "future", + "revenue", + "financial", + "performance", + "Expanding", + "capacity", + "meet", + "demand", + "complex", + "multi", + "year", + "process", + "involves", + "significant", + "regulatory", + "technical", + "construction", + "challenges", + "In", + "addition", + "access", + "constrained", + "capitalized", + "companies", + "face", + "difficulties", + "securing", + "financing", + "large", + "scale", + "projects", + "These", + "limitations", + "delay", + "customer", + "partner", + "deployments", + "reduce", + "adoption", + "We", + "expect", + "Rubin", + "platform", + "start", + "shipping", + "half", + "fiscal", + "2027", + "complexity", + "bringing", + "product", + "architecture", + "sophisticated", + "configurations", + "caused", + "delays", + "production", + "create", + "managing", + "supply", + "This", + "result", + "volatility", + "quality", + "issues", + "increased", + "inventory", + "provisions", + "decreases", + "yields", + "higher", + "material", + "costs", + "warranty", + "Customers", + "postpone", + "purchasing", + "architectures", + "adopt", + "technologies", + "gradually", + "anticipated", + "affecting", + "timing", + "chain", + "expenses", + "Beginning", + "February", + "2026", + "government", + "USG", + "granted", + "licenses", + "ship", + "small", + "amounts", + "H200", + "specific", + "China", + "based", + "To", + "date", + "generated", + "licensing", + "program", + "imports", + "allowed", + "license", + "requires", + "H200s", + "inspection", + "United", + "States", + "prior", + "shipment", + "As", + "shipped", + "subject", + "25%", + "tariff", + "importation", + "recent", + "rise", + "high", + "open", + "source", + "foundation", + "models", + "making", + "advanced", + "capabilities", + "broadly", + "accessible", + "Open", + "dependent", + "developer", + "deployed", + "competitors", + "platforms", + "services", + "made", + "continue", + "investments", + "ecosystem", + "enhance", + "opportunities", + "cultivate", + "strengthen", + "competitive", + "position", + "18.6", + "billion", + "private", + "funds", + "Some", + "include", + "model", + "makers", + "indirectly", + "purchase", + "cloud", + "publicly", + "held", + "equity", + "securities", + "fluctuate", + "significantly", + "adversely", + "affect", + "results", + "Our", + "global", + "networking", + "including", + "Israel", + "operations", + "approximately", + "900", + "employees", + "supporting", + "research", + "development", + "sales", + "marketing", + "impacted", + "conflict", + "Middle", + "East", + "If", + "escalates", + "extends", + "business", + "uncertainty", + "Macroeconomic", + "factors", + "tariffs", + "inflation", + "interest", + "market", + "constraints", + "economic", + "geopolitical", + "developments", + "conflicts", + "direct", + "indirect", + "impacts", + "While", + "difficult", + "isolate", + "quantify", + "macroeconomic", + "manufacturing", + "employee", + "wages", + "equipment", + "solution", + "pricing", + "generally", + "short", + "term", + "Within", + "continuously", + "manage", + "vendors", + "Refer", + "Part", + "II", + "Item", + "1A", + "Risk", + "Factors", + "discussion", + "risks", + "First", + "Quarter", + "Fiscal", + "Year", + "Summary", + "specialize", + "markets", + "provide", + "tremendous", + "acceleration", + "applications", + "incorporate", + "processors", + "interconnects", + "software", + "algorithms", + "systems", + "deliver", + "unique", + "Following", + "rapid", + "evolution", + "businesses", + "transitioning", + "reporting", + "framework", + "reflects", + "current", + "drivers", + "Data", + "Center", + "Edge", + "Computing", + "report", + "Hyperscale", + "ACIE", + "incorporates", + "Clouds", + "Industrial", + "Enterprise", + "public", + "clouds", + "world", + "largest", + "consumer", + "internet", + "addresses", + "opportunity", + "diverse", + "purpose", + "built", + "factories", + "industries", + "countries", + "highlights", + "devices", + "agentic", + "physical", + "PCs", + "game", + "consoles", + "workstations", + "RAN", + "base", + "stations", + "robotics", + "automotive", + "changed", + "presentation", + "comparable", + "periods", + "recast", + "81.6", + "85%", + "ago", + "20%", + "sequentially", + "75.2", + "92%", + "21%", + "ramp", + "300", + "InfiniBand", + "Spectrum", + "Ethernet", + "NVLink", + "Hyperscaler", + "remained", + "50%", + "remaining", + "diversification", + "industrial", + "enterprise", + "sovereign", + "No", + "Hopper", + "occurred", + "compared", + "4.6", + "6.4", + "29%", + "10%", + "increases", + "robust", + "workstation", + "partially", + "offset", + "slower", + "PC", + "tempered", + "elevated", + "memory", + "prices", + "Gross", + "margin", + "lower", + "primarily", + "due", + "4.5", + "charge", + "H20", + "excess", + "obligations", + "flat", + "remains", + "Operating", + "52%", + "12%", + "compensation", + "benefits", + "expense", + "compute", + "engineering", + "materials" + ], + "keywords": [ + "Revenue", + "Data Center", + "AI" + ], + "connect_to": [ + { + "target": "acfd6c99-73da-5b84-9792-14f4c7e29973", + "relation": "embeds", + "ref": "[tables/table-29 Quarterly Financials.html]", + "position": { + "start": 4248, + "end": 4291 + } + }, + { + "target": "2e674fb5-1d8c-510c-a615-afc4728e6150", + "relation": "embeds", + "ref": "[tables/table-30 Revenue by Market.html]", + "position": { + "start": 5280, + "end": 5320 + } + } + ] + } + }, + { + "chunk_id": "0216f33a-487e-5220-b951-c0aff2320860", + "type": "text", + "content": "Refer to Note 13 of the Notes to the Condensed Consolidated Financial Statements for disclosure regarding segment information.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Overview/Financial Information by Business Segment and Geographic Data", + "metadata": { + "length": 126, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Refer", + "Note", + "13", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "disclosure", + "segment", + "information" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "26d1a843-83db-5dda-8da8-1e296b447522", + "type": "text", + "content": "Refer to Part II, Item 7, \"Critical Accounting Policies and Estimates\" of our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. There have been no material changes to our Critical Accounting Policies and Estimates.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Critical Accounting Policies and Estimates", + "metadata": { + "length": 235, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Refer", + "Part", + "II", + "Item", + "Critical", + "Accounting", + "Policies", + "Estimates", + "Annual", + "Report", + "Form", + "10", + "fiscal", + "year", + "ended", + "January", + "25", + "2026", + "There", + "material" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ed640f8b-fc52-58c1-bedc-ccb62b095597", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
Revenue100.0 %100.0 %
Cost of revenue25.139.5
Gross profit74.960.5
Operating expenses
Research and development7.79.1
Sales, general and administrative1.62.4
Total operating expenses9.311.5
Operating income65.649.0
Interest income0.71.2
Interest expense(0.1)(0.1)
Other income (expense), net19.5(0.4)
Total other income, net20.10.7
Income before income tax85.749.7
Income tax expense14.27.1
Net income71.5 %42.6 %
", + "path": "tables/table-31 Revenue Analysis.html", + "metadata": { + "length": 1073, + "summary": "table-32\nThe table shows revenue, cost, gross profit, operating expenses, and net income for three months ending Apr 26, 2026 and Apr 27, 2025, with percentages of revenue.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-31 Revenue Analysis.html", + "keywords": [ + "Revenue", + "Gross Profit", + "Net Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "46166253-fe70-51c1-b38c-469465205798", + "type": "text", + "content": "The following table sets forth, for the periods indicated, certain items in our Condensed Consolidated Statements of Income expressed as a percentage of revenue.\n\n[tables/table-31 Revenue Analysis.html]", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations", + "metadata": { + "length": 204, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "The", + "table", + "sets", + "periods", + "items", + "Condensed", + "Consolidated", + "Statements", + "Income", + "expressed", + "percentage", + "revenue" + ], + "keywords": [], + "connect_to": [ + { + "target": "ed640f8b-fc52-58c1-bedc-ccb62b095597", + "relation": "embeds", + "ref": "[tables/table-31 Revenue Analysis.html]", + "position": { + "start": 163, + "end": 202 + } + } + ] + } + }, + { + "chunk_id": "b9121a0b-a8d4-50e2-90eb-b2dd73c683a1", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Reportable Segments", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6fe48a35-b1c7-5fa9-a40c-7dddee05708e", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change% Change
($ in millions)
Compute & Networking$ 74,550$ 39,589$ 34,96188 %
Graphics7,0654,4732,59258 %
Total$ 81,615$ 44,062$ 37,55385 %
", + "path": "tables/table-32 Revenue Growth.html", + "metadata": { + "length": 488, + "summary": "table-33\nCompute & Networking revenue rose 88% to $74,550M, Graphics up 58% to $7,065M, total revenue increased 85% to $81,615M for the three months ended Apr 26, 2026 vs Apr 27, 2025.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-32 Revenue Growth.html", + "keywords": [ + "Revenue", + "Compute & Networking", + "Graphics" + ], + "tokens": [] + } + }, + { + "chunk_id": "bb864106-6ddb-54e7-9440-097c2252746a", + "type": "text", + "content": "[tables/table-32 Revenue Growth.html]", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Reportable Segments/Revenue by Reportable Segments", + "metadata": { + "length": 41, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [ + { + "target": "6fe48a35-b1c7-5fa9-a40c-7dddee05708e", + "relation": "embeds", + "ref": "[tables/table-32 Revenue Growth.html]", + "position": { + "start": 0, + "end": 37 + } + } + ] + } + }, + { + "chunk_id": "6cf4cb43-f1ce-5f7d-b82c-89a784855c4e", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change% Change
($ in millions)
Compute & Networking$ 53,335$ 22,054$ 31,281142 %
Graphics2,9411,6401,30179 %
Total$ 56,276$ 23,694$ 32,582138 %
", + "path": "tables/table-33 Revenue Growth.html", + "metadata": { + "length": 490, + "summary": "table-34\nCompute & Networking revenue rose 142% to $53,335M, Graphics up 79% to $2,941M, total revenue $56,276M, a 138% increase from $23,694M.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-33 Revenue Growth.html", + "keywords": [ + "Compute & Networking", + "Graphics", + "Revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "81d3c3f5-2459-5371-b7d2-cc3b03592e60", + "type": "text", + "content": "[tables/table-33 Revenue Growth.html]\n\nCompute & Networking revenue – The year-over-year increase in the first quarter of fiscal year 2027 was due to growth in Data Center products, driven by the ramp of our Blackwell systems and demand for our InfiniBand, Spectrum-X Ethernet, and NVLink solutions.\nGraphics revenue – The year-over-year increase in the first quarter of fiscal year 2027 was driven by sales of our Blackwell architecture.\nReportable segment operating income – The year-over-year increase in Compute & Networking segment operating income in the first quarter of fiscal year 2027 was driven by the growth in revenue and the non-recurrence of a \\$4.5 billion charge associated with H2O excess inventory and purchase obligations in the first quarter of fiscal year 2026. The year-over-year increase in Graphics segment operating income in the first quarter of fiscal year 2027 was driven by the growth in revenue.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Reportable Segments/Operating Income by Reportable Segments", + "metadata": { + "length": 930, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Compute", + "Networking", + "revenue", + "The", + "year", + "increase", + "quarter", + "fiscal", + "2027", + "due", + "growth", + "Data", + "Center", + "products", + "driven", + "ramp", + "Blackwell", + "systems", + "demand", + "InfiniBand", + "Spectrum", + "Ethernet", + "NVLink", + "solutions", + "Graphics", + "sales", + "architecture", + "Reportable", + "segment", + "operating", + "income", + "recurrence", + "4.5", + "billion", + "charge", + "H2O", + "excess", + "inventory", + "purchase", + "obligations", + "2026" + ], + "keywords": [], + "connect_to": [ + { + "target": "6cf4cb43-f1ce-5f7d-b82c-89a784855c4e", + "relation": "embeds", + "ref": "[tables/table-33 Revenue Growth.html]", + "position": { + "start": 0, + "end": 37 + } + } + ] + } + }, + { + "chunk_id": "508dc9fa-45df-5be6-b461-ba76a4953c79", + "type": "text", + "content": "We refer to customers who purchase products directly from NVIDIA as direct customers, such as AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. We refer to indirect customers as those who purchase products through our direct customers; indirect customers include CSPs, AI Clouds, AI model makers, enterprises, and public sector entities. Our revenue is concentrated among a limited number of direct and indirect customers and this trend may continue.\nDirect Customers – For the first quarter of fiscal year 2027, three direct customers represented 21%, 17%, and 16% of total revenue, all of which was primarily attributable to the Compute & Networking segment.\nFor the first quarter of fiscal year 2026, sales to two direct customers represented 16% and 14% of total revenue, which were attributable to the Compute & Networking segment.\nIndirect Customers – Indirect customer revenue is an estimation based upon multiple factors including customer purchase order information, product specifications, internal sales data, and other sources. Indirect customers primarily purchase our products through system integrators and distributors.\nWe generate a significant amount of our revenue from a limited number of indirect customers, some individually representing 10% or more of our revenue. Certain companies purchase cloud and related services through various direct and indirect customers. We estimate that one AI research and deployment company contributed to a meaningful amount of our revenue by purchasing cloud services from our customers in the first quarter of fiscal year 2027.\nRevenue by geographic region is designated based on the location of the headquarters of direct customers. The end customer and shipping location may be different from our customers' headquarters location. Revenue from sales to customers headquartered outside of the United States accounted for 22% of total revenue for the first quarter of fiscal year 2027 and 42% of total revenue for the first quarter of fiscal year 2026.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Reportable Segments/Concentration of Revenue", + "metadata": { + "length": 2156, + "summary": "NVIDIA's revenue is concentrated among a limited number of direct and indirect customers. Direct customers include AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. In Q1 FY2027, three direct customers accounted for 21%, 17%, and 16% of total revenue, primarily from the Compute & Networking segment. Indirect customers purchase through system integrators and distributors, with some individually representing 10% or more of revenue. One AI research and deployment company contributed meaningfully to Q1 FY2027 revenue via cloud services. Revenue from customers headquartered outside the US was 22% in Q1 FY2027 and 42% in Q1 FY2026.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "We", + "refer", + "customers", + "purchase", + "products", + "directly", + "NVIDIA", + "direct", + "AIBs", + "distributors", + "ODMs", + "OEMs", + "CSPs", + "AI", + "model", + "makers", + "system", + "integrators", + "Certain", + "internal", + "resources", + "party", + "complete", + "build", + "indirect", + "include", + "Clouds", + "enterprises", + "public", + "sector", + "entities", + "Our", + "revenue", + "concentrated", + "limited", + "number", + "trend", + "continue", + "Direct", + "Customers", + "For", + "quarter", + "fiscal", + "year", + "2027", + "represented", + "21%", + "17%", + "16%", + "total", + "primarily", + "attributable", + "Compute", + "Networking", + "segment", + "2026", + "sales", + "14%", + "Indirect", + "customer", + "estimation", + "based", + "multiple", + "factors", + "including", + "order", + "information", + "product", + "specifications", + "data", + "sources", + "generate", + "significant", + "amount", + "individually", + "representing", + "10%", + "companies", + "cloud", + "related", + "services", + "estimate", + "research", + "deployment", + "company", + "contributed", + "meaningful", + "purchasing", + "Revenue", + "geographic", + "region", + "designated", + "location", + "headquarters", + "The", + "end", + "shipping", + "headquartered", + "United", + "States", + "accounted", + "22%", + "42%" + ], + "keywords": [ + "Direct Customers", + "Indirect Customers", + "Revenue Concentration" + ], + "connect_to": [] + } + }, + { + "chunk_id": "065a4ad2-f012-5b06-b2e3-baea743d4993", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change% Change
($ in millions)
Research and development$ 6,321$ 3,989$ 2,33258 %
Sales, general and administrative1,3001,04125925 %
Total operating expenses$ 7,621$ 5,030$ 2,59152 %
", + "path": "tables/table-34 Operating Expenses.html", + "metadata": { + "length": 524, + "summary": "table-35\nTotal operating expenses increased 52% to $7,621 million, driven by R&D growth of 58% and SG&A growth of 25%.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-34 Operating Expenses.html", + "keywords": [ + "R&D", + "SG&A", + "Expenses" + ], + "tokens": [] + } + }, + { + "chunk_id": "1683defa-145f-5703-8eef-5e90262c9a57", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change
($ in millions)
Interest income$ 540$ 515$ 25
Interest expense(102)(63)(39)
Other income (expense), net15,929(180)16,109
Total other income, net$ 16,367$ 272$ 16,095
", + "path": "tables/table-35 Other Income.html", + "metadata": { + "length": 523, + "summary": "table-36\nThe table shows three-month financial data ending Apr 26, 2026 and Apr 27, 2025. Interest income increased by $25M, interest expense rose by $39M, and other income surged by $16,109M, leading to total other income of $16,367M.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-35 Other Income.html", + "keywords": [ + "Interest", + "Other Income", + "Change" + ], + "tokens": [] + } + }, + { + "chunk_id": "5f3a5c90-8317-5673-a350-df0cdb742c5b", + "type": "text", + "content": "Gross profit consists of total net revenue less cost of revenue. Cost of revenue consists primarily of the cost of semiconductors, including wafer fabrication, assembly, testing and packaging, board and device costs, manufacturing support costs, including labor and overhead associated with such purchases, final test yield fallout, inventory and warranty provisions, memory and component costs, tariffs, and shipping costs. Cost of revenue also includes acquisition-related intangible amortization expense, IP-related costs, and stock-based compensation related to personnel associated with manufacturing operations.\nGross margin increased to 74.9% for the first quarter of fiscal year 2027 compared to 60.5% for the first quarter of fiscal year 2026, primarily due to the prior year's \\$4.5 billion charge associated with H2O excess inventory and purchase obligations.\nProvisions for inventory and excess inventory purchase obligations totaled \\$1.1 billion and \\$5.3 billion for the first quarter of fiscal years 2027 and 2026, respectively. The first quarter of fiscal year 2026 includes \\$4.5 billion associated with H2O excess inventory and purchase obligations. Sales of previously reserved inventory and settlements of excess inventory purchase obligations resulted in a provision release of \\$103 million and \\$436 million for the first quarter of fiscal years 2027 and 2026, respectively. The net effect on our gross margin was an unfavorable impact of 1.2% and 11.0% in the first quarter of fiscal years 2027 and 2026, respectively.\nOperating Expenses\n\n[tables/table-34 Operating Expenses.html]\n\nThe increase in research and development expenses for the first quarter of fiscal year 2027 was primarily driven by a 112% increase in compute and infrastructure, a 31% increase in compensation and benefits, including stock-based compensation, reflecting employee growth and compensation increases, and a 204% increase in engineering development materials for new product introductions.\nThe increase in sales, general and administrative expenses for the first quarter of fiscal year 2027 was primarily driven by compensation and benefits, including stock-based compensation, reflecting employee growth and compensation increases.\nTotal Other Income, Net\n\n[tables/table-35 Other Income.html]\n\nTotal other income, net primarily consists of realized or unrealized gains and losses from investments in non-marketable securities and publicly-held equity securities. The change in Other income (expense), net compared to the first quarter of fiscal year 2026, was primarily driven by unrealized gains on investments in publicly-held equity securities of \\$13.4 billion and non-marketable equity securities of \\$2.6 billion.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Gross Profit and Gross Margin", + "metadata": { + "length": 2730, + "summary": "The text discusses gross profit, cost of revenue, and gross margin trends for the first quarter of fiscal years 2027 and 2026. Gross margin increased to 74.9% in Q1 FY2027 from 60.5% in Q1 FY2026, mainly due to a prior year $4.5 billion charge for H2O excess inventory. Provisions for inventory and excess purchase obligations were $1.1 billion in FY2027 and $5.3 billion in FY2026. Sales of previously reserved inventory led to provision releases of $103 million and $436 million, respectively. The net unfavorable impact on gross margin was 1.2% in FY2027 and 11.0% in FY2026. Operating expenses increased due to higher compute and infrastructure costs, compensation, and engineering materials. Other income included unrealized gains of $13.4 billion from publicly-held equity securities and $2.6 billion from non-marketable securities.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Gross", + "profit", + "consists", + "total", + "net", + "revenue", + "cost", + "Cost", + "primarily", + "semiconductors", + "including", + "wafer", + "fabrication", + "assembly", + "testing", + "packaging", + "board", + "device", + "costs", + "manufacturing", + "support", + "labor", + "overhead", + "purchases", + "final", + "test", + "yield", + "fallout", + "inventory", + "warranty", + "provisions", + "memory", + "component", + "tariffs", + "shipping", + "includes", + "acquisition", + "related", + "intangible", + "amortization", + "expense", + "IP", + "stock", + "based", + "compensation", + "personnel", + "operations", + "margin", + "increased", + "74.9%", + "quarter", + "fiscal", + "year", + "2027", + "compared", + "60.5%", + "2026", + "due", + "prior", + "4.5", + "billion", + "charge", + "H2O", + "excess", + "purchase", + "obligations", + "Provisions", + "totaled", + "1.1", + "5.3", + "years", + "The", + "Sales", + "previously", + "reserved", + "settlements", + "resulted", + "provision", + "release", + "103", + "million", + "436", + "effect", + "gross", + "unfavorable", + "impact", + "1.2%", + "11.0%", + "Operating", + "Expenses", + "increase", + "research", + "development", + "expenses", + "driven", + "112%", + "compute", + "infrastructure", + "31%", + "benefits", + "reflecting", + "employee", + "growth", + "increases", + "204%", + "engineering", + "materials", + "product", + "introductions", + "sales", + "general", + "administrative", + "Total", + "Other", + "Income", + "Net", + "income", + "realized", + "unrealized", + "gains", + "losses", + "investments", + "marketable", + "securities", + "publicly", + "held", + "equity", + "change", + "13.4", + "2.6" + ], + "keywords": [ + "Gross margin", + "Inventory provisions", + "Operating expenses" + ], + "connect_to": [ + { + "target": "065a4ad2-f012-5b06-b2e3-baea743d4993", + "relation": "embeds", + "ref": "[tables/table-34 Operating Expenses.html]", + "position": { + "start": 1564, + "end": 1605 + } + }, + { + "target": "1683defa-145f-5703-8eef-5e90262c9a57", + "relation": "embeds", + "ref": "[tables/table-35 Other Income.html]", + "position": { + "start": 2262, + "end": 2297 + } + } + ] + } + }, + { + "chunk_id": "0c971af4-7dbf-553b-8fbd-e4ff2c28208f", + "type": "table", + "content": "
Apr 26, 2026Jan 25, 2026
(In millions)
Cash and cash equivalents$ 13,237$ 10,605
Marketable debt securities37,09839,065
Cash, cash equivalents, and marketable debt securities$ 50,335$ 49,670
", + "path": "tables/table-36 Cash and Securities.html", + "metadata": { + "length": 384, + "summary": "table-37\nThe table shows cash and cash equivalents, marketable debt securities, and their total for Apr 26, 2026 ($50,335M) and Jan 25, 2026 ($49,670M).", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-36 Cash and Securities.html", + "keywords": [ + "cash", + "securities", + "financial data" + ], + "tokens": [] + } + }, + { + "chunk_id": "27a41ebc-884b-5486-bec0-98f35e523721", + "type": "table", + "content": "
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Net cash provided by operating activities$ 50,344$ 27,414
Net cash used in investing activities$ (26,429)$ (5,216)
Net cash used in financing activities$ (21,283)$ (15,553)
", + "path": "tables/table-37 Cash Flow Data.html", + "metadata": { + "length": 456, + "summary": "table-38\nThe table shows net cash provided by operating activities of $50,344 million and $27,414 million for the three months ended Apr 26, 2026 and Apr 27, 2025, respectively. Net cash used in investing and financing activities also reported.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-37 Cash Flow Data.html", + "keywords": [ + "cash flow", + "operating", + "investing" + ], + "tokens": [] + } + }, + { + "chunk_id": "011b326c-c08c-5356-b2f9-b803e3b6e617", + "type": "text", + "content": "Income tax expense was \\$11.6 billion and \\$3.1 billion for the first quarter of fiscal years 2027 and 2026, respectively. Income tax as a percentage of income before income tax was 16.6% and 14.3% for the first quarter of fiscal years 2027 and 2026, respectively.\nThe effective tax rate increased primarily due to a lower percentage of tax benefits from stock-based compensation relative to the increase in income before income tax.\nOur effective tax rates for the first quarter of fiscal years 2027 and 2026 were lower than the U.S. federal statutory rate of 21% primarily due to tax benefits from foreign-derived deduction eligible income, income earned in jurisdictions that were subject to taxes at rates lower than the U.S. federal statutory tax rate, stock-based compensation, and the U.S. federal research tax credit.\nRefer to Note 11 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q for additional information.\nLiquidity and Capital Resources\n\n[tables/table-36 Cash and Securities.html]\n\n\n[tables/table-37 Cash Flow Data.html]\n\nOur fixed income security investments include highly rated, diversified investment types and credit exposures with shorter maturities.\nCash provided by operating activities increased in the first quarter of fiscal year 2027 compared to the first quarter of fiscal year 2026 due to higher revenue.\nCash used in investing activities increased in the first quarter of fiscal year 2027 compared to the first quarter of fiscal year 2026, primarily driven by higher purchases of equity investment securities.\nCash used in financing activities increased in the first quarter of fiscal year 2027 compared to the first quarter of fiscal year 2026, mainly due to higher share repurchases.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Income Taxes", + "metadata": { + "length": 1792, + "summary": "Income tax expense was $11.6 billion and $3.1 billion for Q1 of fiscal years 2027 and 2026, with effective tax rates of 16.6% and 14.3%, respectively. The rate increase was due to lower tax benefits from stock-based compensation relative to higher income. Both rates were below the 21% U.S. statutory rate due to foreign-derived deductions, lower foreign taxes, stock-based compensation, and research credits. Cash from operations increased due to higher revenue, while investing and financing cash uses rose from higher equity purchases and share repurchases.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Income", + "tax", + "expense", + "11.6", + "billion", + "3.1", + "quarter", + "fiscal", + "years", + "2027", + "2026", + "percentage", + "income", + "16.6%", + "14.3%", + "The", + "effective", + "rate", + "increased", + "primarily", + "due", + "lower", + "benefits", + "stock", + "based", + "compensation", + "relative", + "increase", + "Our", + "rates", + "federal", + "statutory", + "21%", + "foreign", + "derived", + "deduction", + "eligible", + "earned", + "jurisdictions", + "subject", + "taxes", + "research", + "credit", + "Refer", + "Note", + "11", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Part", + "Item", + "Quarterly", + "Report", + "Form", + "10", + "additional", + "information", + "Liquidity", + "Capital", + "Resources", + "fixed", + "security", + "investments", + "include", + "highly", + "rated", + "diversified", + "investment", + "types", + "exposures", + "shorter", + "maturities", + "Cash", + "provided", + "operating", + "activities", + "year", + "compared", + "higher", + "revenue", + "investing", + "driven", + "purchases", + "equity", + "securities", + "financing", + "share", + "repurchases" + ], + "keywords": [ + "income tax", + "effective tax rate", + "liquidity" + ], + "connect_to": [ + { + "target": "0c971af4-7dbf-553b-8fbd-e4ff2c28208f", + "relation": "embeds", + "ref": "[tables/table-36 Cash and Securities.html]", + "position": { + "start": 1020, + "end": 1062 + } + }, + { + "target": "27a41ebc-884b-5486-bec0-98f35e523721", + "relation": "embeds", + "ref": "[tables/table-37 Cash Flow Data.html]", + "position": { + "start": 1065, + "end": 1102 + } + } + ] + } + }, + { + "chunk_id": "900f175d-1c97-53fd-83ff-1cce7ef2c91a", + "type": "text", + "content": "Our primary sources of liquidity include cash, cash equivalents, marketable debt and equity securities, and cash generated by our operations. As of April 26, 2026, we had \\$50.3 billion in cash, cash equivalents, and marketable debt securities as well as \\$30.2 billion of marketable equity securities. We believe that we have sufficient liquidity to meet our operating requirements for at least the next twelve months and for the foreseeable future, including our future obligations. We continuously evaluate our liquidity and capital resources, including our access to external capital, to ensure we can finance future capital requirements and commitments.\nOur marketable securities consist of publicly-held equity securities, debt securities issued by the U.S. government and its agencies, highly-rated corporations and financial institutions, and foreign government entities, as well as certificates of deposit issued by highly-rated financial institutions. These marketable securities are primarily denominated in U.S. dollars. Refer to Note 5 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q for additional information.\nExcept for approximately \\$1.7 billion of cash, cash equivalents, and marketable debt securities held outside of the U.S., for which we have not accrued any related foreign or state taxes if we repatriate these amounts to the U.S., substantially all of our cash, cash equivalents, and marketable debt securities held outside the U.S. at the end of the first quarter of fiscal year 2027 are available for use in the U.S. without incurring additional U.S. federal income taxes. We made no federal income tax payments in the first quarter of fiscal year 2027, whereas our second quarter of fiscal year 2027 is scheduled to include two payments.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Liquidity", + "metadata": { + "length": 1834, + "summary": "The passage discusses the company's primary liquidity sources, including cash, cash equivalents, and marketable debt and equity securities. As of April 26, 2026, it held $50.3 billion in cash, cash equivalents, and marketable debt securities, plus $30.2 billion in marketable equity securities. The company believes it has sufficient liquidity for at least the next twelve months. Marketable securities include U.S. government and corporate debt, equity securities, and certificates of deposit, mostly in U.S. dollars. About $1.7 billion held outside the U.S. has not accrued taxes for repatriation, but substantially all foreign-held cash is available for U.S. use without additional federal income tax. No federal income tax payments were made in the first quarter of fiscal year 2027, with two payments scheduled for the second quarter.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Our", + "primary", + "sources", + "liquidity", + "include", + "cash", + "equivalents", + "marketable", + "debt", + "equity", + "securities", + "generated", + "operations", + "As", + "April", + "26", + "2026", + "50.3", + "billion", + "30.2", + "We", + "sufficient", + "meet", + "operating", + "requirements", + "twelve", + "months", + "foreseeable", + "future", + "including", + "obligations", + "continuously", + "evaluate", + "capital", + "resources", + "access", + "external", + "ensure", + "finance", + "commitments", + "consist", + "publicly", + "held", + "issued", + "government", + "agencies", + "highly", + "rated", + "corporations", + "financial", + "institutions", + "foreign", + "entities", + "certificates", + "deposit", + "These", + "primarily", + "denominated", + "dollars", + "Refer", + "Note", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Part", + "Item", + "Quarterly", + "Report", + "Form", + "10", + "additional", + "information", + "Except", + "approximately", + "1.7", + "accrued", + "related", + "state", + "taxes", + "repatriate", + "amounts", + "substantially", + "end", + "quarter", + "fiscal", + "year", + "2027", + "incurring", + "federal", + "income", + "made", + "tax", + "payments", + "scheduled" + ], + "keywords": [ + "liquidity", + "marketable securities", + "cash" + ], + "connect_to": [] + } + }, + { + "chunk_id": "4eb06f36-7b8b-545e-b20a-d4a74a7b6538", + "type": "text", + "content": "In the first quarter of fiscal year 2027, we repurchased 108 million shares of our common stock for \\$20.2 billion. As of April 26, 2026, we were authorized, subject to certain specifications, to repurchase up to \\$38.5 billion of our common stock.\nOn May 18, 2026, our Board of Directors approved an additional \\$80.0 billion in share repurchase authorization, without expiration.\nWe may execute repurchases from time to time, subject to market conditions, operating requirements and other investment opportunities, in the open market, in privately-negotiated transactions, pursuant to a Rule 10b5-1 trading plan or in structured share repurchase agreements in compliance with Rule 10b-18 of the Exchange Act. Our share repurchase program may be suspended at any time at our discretion.\nWe paid cash dividends to our shareholders of \\$243 million during the first quarter of fiscal year 2027. On May 18, 2026, we increased our quarterly cash dividend from \\$0.01 per share to \\$0.25 per share to all shareholders of record on June 4, 2026. Our quarterly cash dividend will be paid on June 26, 2026.\nThe payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders.\nThe U.S. Inflation Reduction Act of 2022 requires a 1% excise tax on certain share repurchases in excess of shares issued for employee compensation made after December 31, 2022. The excise tax is included in our share repurchase cost and was not significant for the first quarter of fiscal year 2027.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Liquidity/Capital Return to Shareholders", + "metadata": { + "length": 1580, + "summary": "In Q1 FY2027, the company repurchased 108 million shares for $20.2 billion, with $38.5 billion remaining authorized as of April 26, 2026. On May 18, 2026, the Board approved an additional $80 billion in repurchase authorization without expiration. Repurchases may be executed via open market, private negotiations, Rule 10b5-1 plans, or structured agreements, and can be suspended anytime. Cash dividends of $243 million were paid in Q1 FY2027. On May 18, 2026, the quarterly dividend was increased from $0.01 to $0.25 per share, payable on June 26, 2026 to shareholders of record on June 4, 2026. Future dividends depend on Board discretion. A 1% excise tax on certain repurchases under the Inflation Reduction Act applies but was not significant for the quarter.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "In", + "quarter", + "fiscal", + "year", + "2027", + "repurchased", + "108", + "million", + "shares", + "common", + "stock", + "20.2", + "billion", + "As", + "April", + "26", + "2026", + "authorized", + "subject", + "specifications", + "repurchase", + "38.5", + "On", + "May", + "18", + "Board", + "Directors", + "approved", + "additional", + "80.0", + "share", + "authorization", + "expiration", + "We", + "execute", + "repurchases", + "time", + "market", + "conditions", + "operating", + "requirements", + "investment", + "opportunities", + "open", + "privately", + "negotiated", + "transactions", + "pursuant", + "Rule", + "10b5", + "trading", + "plan", + "structured", + "agreements", + "compliance", + "10b", + "Exchange", + "Act", + "Our", + "program", + "suspended", + "discretion", + "paid", + "cash", + "dividends", + "shareholders", + "243", + "increased", + "quarterly", + "dividend", + "0.01", + "0.25", + "record", + "June", + "The", + "payment", + "future", + "continuing", + "determination", + "declaration", + "interests", + "Inflation", + "Reduction", + "2022", + "requires", + "1%", + "excise", + "tax", + "excess", + "issued", + "employee", + "compensation", + "made", + "December", + "31", + "included", + "cost", + "significant" + ], + "keywords": [ + "share repurchase", + "dividends", + "stock buyback" + ], + "connect_to": [] + } + }, + { + "chunk_id": "8a28477a-a3a6-5f15-809a-ed15fa77eaab", + "type": "table", + "content": "
Apr 26, 2026
(In millions)
Due in one year$ 1,000
Due in one to five years2,750
Due in five to ten years1,250
Due in greater than ten years3,500
Unamortized debt discount and issuance costs(30)
Net carrying amount$ 8,470
Less short-term portion(1,000)
Total long-term portion$ 7,470
", + "path": "tables/table-38 Debt Maturity.html", + "metadata": { + "length": 570, + "summary": "table-39\nThe table shows debt due in various periods totaling $8,470 million net carrying amount, with $1,000 million short-term and $7,470 million long-term portion.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-38 Debt Maturity.html", + "keywords": [ + "debt", + "maturity", + "carrying" + ], + "tokens": [] + } + }, + { + "chunk_id": "65528dc9-974e-5acd-9df1-e705f33f5323", + "type": "text", + "content": "Our aggregate debt maturities as of April 26, 2026, by year payable, were as follows:\n\n[tables/table-38 Debt Maturity.html]\n\nWe have a commercial paper program to support general corporate purposes, pursuant to which we may issue unsecured paper notes, from time to time or all at once, up to \\$25.0 billion. As of April 26, 2026, no commercial paper was outstanding.\nRefer to Note 9 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q for further discussion.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Liquidity/Outstanding Indebtedness and Commercial Paper Program", + "metadata": { + "length": 522, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Our", + "aggregate", + "debt", + "maturities", + "April", + "26", + "2026", + "year", + "payable", + "We", + "commercial", + "paper", + "program", + "support", + "general", + "corporate", + "purposes", + "pursuant", + "issue", + "unsecured", + "notes", + "time", + "25.0", + "billion", + "As", + "outstanding", + "Refer", + "Note", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Part", + "Item", + "Quarterly", + "Report", + "Form", + "10", + "discussion" + ], + "keywords": [], + "connect_to": [ + { + "target": "8a28477a-a3a6-5f15-809a-ed15fa77eaab", + "relation": "embeds", + "ref": "[tables/table-38 Debt Maturity.html]", + "position": { + "start": 87, + "end": 123 + } + } + ] + } + }, + { + "chunk_id": "12ba28dd-63b8-5079-8e90-8b296298face", + "type": "text", + "content": "For descriptions of our facility lease guarantees, long-term debt, purchase commitments, and operating lease obligations, refer to Note 8, Note 9, Note 10, and Note 14 of the Notes to the Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q, respectively.\nWe expect to continue investing in our ecosystem. Refer to Note 6 and Item 1A. Risk Factors for additional information regarding our investments.\nUnrecognized tax benefits were \\$4.5 billion, which includes related interest and penalties of \\$439 million, and were recorded in non-current income tax payable as of April 26, 2026. We are unable to estimate the timing of any potential tax liability, interest payments, or penalties in individual years due to uncertainties in the underlying income tax positions and the timing of the effective settlement of such tax positions. We are currently under examination by the Internal Revenue Service for our fiscal years 2023 and 2024. Refer to Note 11 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q.\nOther than the contractual obligations described in Notes 6 and 10, there were no material changes outside the ordinary course of business in our contractual obligations from those disclosed in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. Refer to Item 7, “Management’s Discussion and Analysis of Financial Condition and Results of Operations - Liquidity and Capital Resources” in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026 for a description of our contractual obligations. For a description of our facility lease guarantees, long-term debt, purchase obligations, and operating lease obligations, refer to Notes 8, 9, 10, and 14 of the Notes to the Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q, respectively.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Liquidity/Material Cash Requirements and Other Obligations", + "metadata": { + "length": 1941, + "summary": "The passage discusses various financial obligations and commitments, including facility lease guarantees, long-term debt, purchase commitments, and operating lease obligations referenced in specific notes of the financial statements. It mentions unrecognized tax benefits of $4.5 billion including interest and penalties, and ongoing IRS examinations for fiscal years 2023 and 2024. It notes no material changes in contractual obligations outside ordinary course of business from the prior annual report.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "For", + "descriptions", + "facility", + "lease", + "guarantees", + "long", + "term", + "debt", + "purchase", + "commitments", + "operating", + "obligations", + "refer", + "Note", + "10", + "14", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "Part", + "Item", + "Quarterly", + "Report", + "Form", + "We", + "expect", + "continue", + "investing", + "ecosystem", + "Refer", + "1A", + "Risk", + "Factors", + "additional", + "information", + "investments", + "Unrecognized", + "tax", + "benefits", + "4.5", + "billion", + "includes", + "related", + "interest", + "penalties", + "439", + "million", + "recorded", + "current", + "income", + "payable", + "April", + "26", + "2026", + "unable", + "estimate", + "timing", + "potential", + "liability", + "payments", + "individual", + "years", + "due", + "uncertainties", + "underlying", + "positions", + "effective", + "settlement", + "examination", + "Internal", + "Revenue", + "Service", + "fiscal", + "2023", + "2024", + "11", + "Other", + "contractual", + "material", + "ordinary", + "business", + "disclosed", + "Annual", + "year", + "ended", + "January", + "25", + "Management", + "Discussion", + "Analysis", + "Condition", + "Results", + "Operations", + "Liquidity", + "Capital", + "Resources", + "description" + ], + "keywords": [ + "contractual obligations", + "tax benefits", + "lease guarantees" + ], + "connect_to": [] + } + }, + { + "chunk_id": "8aab6ac4-a395-5919-b0e6-6b542026561d", + "type": "text", + "content": "There has been no adoption of any new and recently issued accounting pronouncements.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Adoption of New and Recently Issued Accounting Pronouncements", + "metadata": { + "length": 84, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "There", + "adoption", + "recently", + "issued", + "accounting", + "pronouncements" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f97eda7e-e9f5-54ab-be8c-93d8aaa1a3a7", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 3. Quantitative and Qualitative Disclosures About Market Risk", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1bd4d830-678d-5e73-86ee-22b2980e2b86", + "type": "text", + "content": "Financial market risks related to investment and interest rate risk are described in Part II, Item 7A, “Quantitative and Qualitative Disclosures About Market Risk” in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. Our marketable equity securities consist of publicly-held equity securities, while our non-marketable equity securities are investments in privately-held companies. Publicly-held equity securities are subject to market price volatility. A hypothetical 10% decrease in our publicly-held equity securities would decrease the fair value of the publicly-held equity securities balance by \\$3.9 billion and \\$1.8 billion as of April 26, 2026 and January 25, 2026, respectively.\nNon-marketable equity securities are measured based on cost minus impairment, if any, and are adjusted for observable price changes in orderly transactions for an identical or similar investment in the same issuer. Valuations of our non-\nmarketable equity securities are inherently complex due to the lack of readily available market data and observable transactions, and impact of macroeconomic factors.\nFor a description of our equity investments, refer to Notes 5 and 6 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q, respectively.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 3. Quantitative and Qualitative Disclosures About Market Risk/Investment and Interest Rate Risk", + "metadata": { + "length": 1318, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Financial", + "market", + "risks", + "related", + "investment", + "interest", + "rate", + "risk", + "Part", + "II", + "Item", + "7A", + "Quantitative", + "Qualitative", + "Disclosures", + "About", + "Market", + "Risk", + "Annual", + "Report", + "Form", + "10", + "fiscal", + "year", + "ended", + "January", + "25", + "2026", + "Our", + "marketable", + "equity", + "securities", + "consist", + "publicly", + "held", + "investments", + "privately", + "companies", + "Publicly", + "subject", + "price", + "volatility", + "hypothetical", + "10%", + "decrease", + "fair", + "balance", + "3.9", + "billion", + "1.8", + "April", + "26", + "Non", + "measured", + "based", + "cost", + "minus", + "impairment", + "adjusted", + "observable", + "orderly", + "transactions", + "identical", + "similar", + "issuer", + "Valuations", + "inherently", + "complex", + "due", + "lack", + "readily", + "data", + "impact", + "macroeconomic", + "factors", + "For", + "description", + "refer", + "Notes", + "Condensed", + "Consolidated", + "Statements", + "Quarterly" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0f9d764d-18bc-5d08-9365-766527108a87", + "type": "text", + "content": "The impact of foreign currency transactions related to foreign exchange rate risk is described in Part II, Item 7A, “Quantitative and Qualitative Disclosures About Market Risk” in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. As of April 26, 2026, there have been no material changes to the foreign exchange rate risks described as of January 25, 2026.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 3. Quantitative and Qualitative Disclosures About Market Risk/Foreign Exchange Rate Risk", + "metadata": { + "length": 381, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "The", + "impact", + "foreign", + "currency", + "transactions", + "related", + "exchange", + "rate", + "risk", + "Part", + "II", + "Item", + "7A", + "Quantitative", + "Qualitative", + "Disclosures", + "About", + "Market", + "Risk", + "Annual", + "Report", + "Form", + "10", + "fiscal", + "year", + "ended", + "January", + "25", + "2026", + "As", + "April", + "26", + "material", + "risks" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "936043e7-b966-5bb9-926a-c7125e6be145", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 4. Controls and Procedures", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2bb8e2d9-6055-5686-b3a5-fbc44fb35b9e", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 4. Controls and Procedures/Controls and Procedures", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fddfc899-5916-51f6-a41b-4010214c0252", + "type": "text", + "content": "Based on their evaluation as of April 26, 2026, our management, including our Chief Executive Officer and Chief Financial Officer, has concluded that our disclosure controls and procedures (as defined in Rule 13a-15(e) under the Exchange Act) were effective to provide reasonable assurance that the information we are required to disclose in reports that we file or submit under the Exchange Act is recorded, processed, summarized and reported within the time periods specified in the SEC rules and forms, and that such information is accumulated and communicated to our management, including our Chief Executive Officer and our Chief Financial Officer, as appropriate, to allow timely decisions regarding required disclosures.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 4. Controls and Procedures/Controls and Procedures/Disclosure Controls and Procedures", + "metadata": { + "length": 727, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Based", + "evaluation", + "April", + "26", + "2026", + "management", + "including", + "Chief", + "Executive", + "Officer", + "Financial", + "concluded", + "disclosure", + "controls", + "procedures", + "defined", + "Rule", + "13a", + "15", + "Exchange", + "Act", + "effective", + "provide", + "reasonable", + "assurance", + "information", + "required", + "disclose", + "reports", + "file", + "submit", + "recorded", + "processed", + "summarized", + "reported", + "time", + "periods", + "SEC", + "rules", + "forms", + "accumulated", + "communicated", + "timely", + "decisions", + "disclosures" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2f7e4866-fe14-5d15-ac85-c75122f1c36b", + "type": "text", + "content": "There have been no changes in our internal control over financial reporting during the first quarter of fiscal year 2027 that have materially affected, or are reasonably likely to materially affect, our internal control over financial reporting. We are continuing a phased upgrade of our enterprise resource planning, or ERP, system to update our existing core financial systems. The ERP system is designed to accurately maintain our financial records used to report operating results. We will continue to evaluate each quarter whether there are changes that materially affect our internal control over financial reporting.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 4. Controls and Procedures/Controls and Procedures/Changes in Internal Control Over Financial Reporting", + "metadata": { + "length": 623, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "There", + "internal", + "control", + "financial", + "reporting", + "quarter", + "fiscal", + "year", + "2027", + "materially", + "affected", + "affect", + "We", + "continuing", + "phased", + "upgrade", + "enterprise", + "resource", + "planning", + "ERP", + "system", + "update", + "existing", + "core", + "systems", + "The", + "designed", + "accurately", + "maintain", + "records", + "report", + "operating", + "results", + "continue", + "evaluate" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "23cf2a4e-acd2-553b-825b-46081c78a5ac", + "type": "text", + "content": "Our management, including our Chief Executive Officer and Chief Financial Officer, does not expect that our disclosure controls and procedures, or our internal controls, will prevent all error and all fraud. A control system, no matter how well conceived and operated, can provide only reasonable, not absolute, assurance that the objectives of the control system are met. Further, the design of a control system must reflect the fact that there are resource constraints, and the benefits of controls must be considered relative to their costs. Because of the inherent limitations in all control systems, no evaluation of controls can provide absolute assurance that all control issues and instances of fraud, if any, within NVIDIA have been detected.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 4. Controls and Procedures/Controls and Procedures/Inherent Limitations on Effectiveness of Controls", + "metadata": { + "length": 751, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Our", + "management", + "including", + "Chief", + "Executive", + "Officer", + "Financial", + "expect", + "disclosure", + "controls", + "procedures", + "internal", + "prevent", + "error", + "fraud", + "control", + "system", + "matter", + "conceived", + "operated", + "provide", + "reasonable", + "absolute", + "assurance", + "objectives", + "met", + "Further", + "design", + "reflect", + "fact", + "resource", + "constraints", + "benefits", + "considered", + "relative", + "costs", + "Because", + "inherent", + "limitations", + "systems", + "evaluation", + "issues", + "instances", + "NVIDIA", + "detected" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "00ef5721-60f4-57af-9737-abe243fd5deb", + "type": "text", + "content": "", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c0b75c00-d41a-57da-a7f8-e7b2e3ab1912", + "type": "text", + "content": "Refer to Part I, Item 1, Note 10 of the Notes to the Condensed Consolidated Financial Statements for a discussion of significant developments in our legal proceedings since January 25, 2026.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information/Item 1. Legal Proceedings", + "metadata": { + "length": 190, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Refer", + "Part", + "Item", + "Note", + "10", + "Notes", + "Condensed", + "Consolidated", + "Financial", + "Statements", + "discussion", + "significant", + "developments", + "legal", + "proceedings", + "January", + "25", + "2026" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cf183df1-c729-5828-b029-f8ee9fa6a942", + "type": "text", + "content": "Other than the risk factors listed below, there have been no material changes from the risk factors previously described under Item 1A of our Annual Report on Form 10-K for the fiscal year ended January 25, 2026.\nPurchasing or owning NVIDIA common stock involves investment risks including, but not limited to, the risks described in Item 1A of our Annual Report on Form 10-K for the fiscal year ended January 25, 2026, and below. Any one of those risks could harm our business, financial condition and results of operations or reputation, which could cause our stock price to decline. Additional risks, trends and uncertainties not presently known to us or that we currently believe are immaterial may also harm our business, financial condition, results of operations or reputation.\nCompetition could adversely impact our market share and financial results.\nOur target markets remain competitive, and competition may intensify with expanding and changing product and service offerings, industry standards, customer and market needs, new entrants and consolidations. Other companies compete\nwith us on a wide range of parameters including price, total cost of ownership, and performance, which has resulted and may in the future result in lower-than-expected selling prices or demand for our products. Some of our competitors operate their own fabrication facilities, and have longer operating histories, larger customer bases, more comprehensive IP portfolios and patent protections, more design wins, and greater financial, sales, marketing and distribution resources than we do. These competitors may be able to acquire market share and/or prevent us from doing so, more effectively identify and capitalize upon opportunities in new markets and end-user trends, more quickly transition their products, and impinge on our ability to procure sufficient foundry capacity and scarce input materials during a supply-constrained environment, which could harm our business. Some of our customers are developing their own ASICs and other products, including designs optimized for certain workloads that may not require all of the features and functionality our data center systems provide. Others may offer cloud-based services that compete with our AI cloud service offerings, and we may not be able to establish market share sufficient to achieve the scale necessary to meet our business objectives. If we are unable to successfully compete in this environment, demand for our products, services, and technologies could decrease, which may negatively impact our business.\nLong manufacturing lead times and uncertain supply and capacity availability, combined with a failure to estimate customer demand accurately, has led and could in the future lead to mismatches between supply and demand.\nWe have long manufacturing lead times and build finished products and maintain inventory in advance of anticipated demand. In periods of shortages impacting the semiconductor industry and/or limited supply or capacity in our supply chain, the lead times for certain supply may be extended. We have previously experienced and may continue to experience extended lead times of more than 12 months. To secure future supply and capacity, we have paid premiums, provided deposits, and entered into long-term supply agreements and capacity commitments, which have increased our product costs and this may continue. We may still be unable to secure sufficient commitments for capacity to address our business needs.\nIf we inaccurately estimate demand, or our customers change orders, as we have experienced in the past, we may not be able to reduce our supply commitments in time, at the same rate, or at all. Significant mismatches between supply and demand have varied across our market platforms, resulted in both product shortages and excess inventory, significantly harmed our financial results and could reoccur. If we underestimate demand, and our foundry partners and contract manufacturers are unable to increase production or provide sufficient supply, we may not be able to meet increased customer demand in a timely manner, or at all. Our reputation and customer relationships could be damaged and we could lose revenue and market share. Additionally, since some of our products are part of a complex data center buildout, supply constraints or availability issues with respect to any one component have had and may have a broader revenue impact. For example, our ability to sell certain products has been and could be impeded if components necessary for the finished products are not available from third parties.\nIf we overestimate demand, or if customers cancel or defer orders or choose to purchase from our competitors, we may not be able to utilize on-hand inventory or reduce purchase commitments accordingly. We have had to reduce average selling prices, including due to our channel pricing programs, increase prices for certain of our products as a result of our suppliers' increase in prices, write down our inventory, incur cancellation penalties, and record impairments, and may have to do so in the future. The impact of these risks would be amplified by our non-cancellable and non-returnable purchase orders placed in advance of our historical lead times and could be exacerbated if we need to make changes to the design of future products. These risks have increased and may continue to increase as our purchase obligations and prepaids have grown and are expected to continue to grow and become a greater portion of our total supply. All of these factors may negatively impact our gross margins and financial results.\nFactors that have caused and/or could in the future cause us to underestimate or overestimate demand, and impact the timing and volume of our revenue, include:\n• changes in product development cycles and time to market;\n• competing technologies and competitor product releases, announcements or other actions;\n• changes in business and economic conditions;\n- sudden or sustained government lockdowns or public health issues;\n• rapidly changing technology or customer requirements;\n• the availability of sufficient data center capacity or energy for customers to procure;\n- new product introductions and transitions resulting in less demand for existing products;\n- new or unexpected end-use cases;\n• increase in demand for competitive products;\n• changes in end-user demand;\n• purchasing decisions made, and inventory levels held by, distributors, ODMs, OEMs, system integrators, other channel partners and other third parties;\n- the ability of developers, end customers and other third parties to build, enhance, and maintain accelerated computing applications that leverage our platforms;\n• the demand for accelerated computing, AI-related cloud services, or large language models;\n• changes that impact the ecosystem for the architectures underlying our products and technologies;\n• government actions or changes in governmental policies, such as export controls, increased restrictions on gaming usage, or tariffs;\n• our customers' and partners' ability to secure capital and energy and to build complex data center infrastructure timely; and\n• the availability of third-party content on our platforms, such as GeForce NOW.\nThe availability of data centers, energy, and capital to support the buildout of NVIDIA AI infrastructure by our customers and partners is crucial, and any shortage of these and other necessary resources could impact our future revenue and financial performance. Expanding energy capacity to meet demand is a complex, multi-year process involving significant regulatory, technical, and construction challenges. In addition, access to capital can be particularly constrained for less-capitalized companies, which may face difficulties securing financing for large-scale infrastructure projects. These limitations could delay customer and partner deployments or reduce the scale of accelerated computing and AI adoption.\nChallenges in estimating demand could become more pronounced or volatile in the future on both a global and regional basis. Extended lead times may occur if we experience other supply constraints caused by natural disasters, pandemics or other events. Geopolitical tensions in regions where we rely on suppliers, contract manufacturers, and assembly partners that are critical to our supply continuity, could have a material adverse impact on us. Supply availability affecting memory, and other components, as well as rising prices, may drive the prices for data center buildouts higher. Publicly announced intentions by governments or other companies to purchase our products can further complicate our demand estimates, as such announcements are often non-binding and may not result in committed volumes.\nWe continue to increase our supply and capacity purchases with existing and new suppliers to support our demand projections and increasing complexity of our data center products. We have also entered and may continue to enter into prepaid manufacturing and capacity agreements to supply both current and future products. The increased purchase volumes and integration of new suppliers and contract manufacturers into our supply chain creates more complexity in managing multiple suppliers with variations in production planning, execution and logistics. Our expanding product portfolio and varying component compatibility and quality may lead to increased inventory levels. We have incurred and may in the future incur inventory provisions or impairments if our inventory or supply or capacity commitments exceed demand for our products or demand declines. We are increasing our U.S.-based manufacturing and investing in specialized equipment and processes to support domestic production. We may experience delays or difficulties in scaling production as planned. Our ability to increase manufacturing capabilities will depend on the domestic manufacturing ecosystem's capacity to ramp production supply to the required volume timely. Delays or shortfalls could impact our ability to meet demand.\nIntroducing or offering multiple architectures concurrently is complex and we often ship multiple architecture products simultaneously as our channel partners prepare to ship and support new products. We are generally in various stages of introducing and/or offering the architectures of our Data Center and Edge Computing products. The computing industry is experiencing a broader and faster launch cadence of accelerated computing platforms to meet a growing and diverse set of AI opportunities. We have introduced a new product and architecture cadence of our Data Center solutions where we seek to complete new computing solutions each year and provide a greater variety of Data Center offerings, including our Rubin platform which is expected to start shipping in the second half of fiscal year 2027. The increased frequency of these architecture introductions and the larger number of products and product configurations may magnify the challenges associated with managing our supply and demand which may further create volatility in our revenue. Qualification time for new products, customers anticipating new architecture introductions, and channel partners reducing channel inventory of prior architectures ahead of new product introductions can reduce, or create volatility in, our revenue. Customers may delay adopting new architectures if their data center infrastructure is not ready, which could affect the timing of our revenue. We have experienced and may in the future experience reduced demand for current generation architectures when customers anticipate new architectures, and we may be unable to sell multiple product architectures at the same time. Our financial results have been and may in the future be negatively impacted if we are unable to execute our architectural introductions as planned for any reason. The increased frequency and complexity of newly introduced products could result in unanticipated quality or production issues that could increase the magnitude of inventory provisions, warranty, or other costs or result in product delays. For example, our gross margins in the second quarter of fiscal year 2025 were negatively impacted by inventory provisions for low-yielding Blackwell material.\nWe incur significant engineering development resources for new products, and changes to our product roadmap may impact our ability to develop other products or adequately manage our supply chain cost. Customers may delay purchasing existing products as we increase the frequency of new products or may not be able to adopt our new products as fast as forecasted, both impacting the timing of our revenue and supply chain cost. While we have managed concurrent architecture introductions and/or offerings and have sold multiple product architectures at the same time, these efforts are difficult, may impair our ability to predict demand and impact our supply mix, and may cause us to incur additional costs.\nDemand estimates for our products, applications, and services can be incorrect, which may create volatility in our revenue or supply levels. We may not be able to generate significant revenue from them. Because our products may be used in multiple use cases and applications, it is difficult to estimate with any reasonable degree of precision the impact of accelerated computing and AI models on our reported revenue or forecasted demand.\nThe use of our GPUs for new, mercurial, or trendy applications, has impacted, and can impact in the future, demand for our products, including by leading to inconsistent spikes and drops in demand. For example, several years ago, our gaming GPUs began to be used for mining digital currencies, such as Ethereum. It is difficult for us to estimate with any reasonable degree of precision the past or current impact of cryptocurrency mining, or forecast the future impact of cryptocurrency mining, on demand for our products. Volatility in the cryptocurrency market, including new compute technologies, price changes in cryptocurrencies, government cryptocurrency policies and regulations, new cryptocurrency standards and changes in the method of verifying blockchain transactions, has impacted and can in the future impact cryptocurrency mining and demand for our products and can further impact our ability to estimate demand for our products. Changes to cryptocurrency standards and processes including, but not limited to, the Ethereum 2.0 merge in 2022, have reduced and may in the future decrease the usage of GPUs for Ethereum mining. This has created and may in the future create increased aftermarket sales of our GPUs, which could negatively impact retail prices for our GPUs and reduce demand for our new GPUs. In general, our new products or previously sold products may be resold online or on the unauthorized “gray market,” which also makes demand forecasting difficult. Gray market products and reseller marketplaces compete with our new products and distribution channels. Our inability to accurately predict our demand that arises from new use cases may create volatility in our revenue.\nWe are subject to complex laws, rules, regulations, and political and other actions, including restrictions on the export of our products, which may adversely impact our business.\nWe are subject to laws and regulations domestically and worldwide, affecting our operations in areas including, but not limited to, IP ownership and infringement; taxes; import and export requirements and tariffs; anti-corruption, including the Foreign Corrupt Practices Act; business acquisitions; foreign exchange controls and cash repatriation restrictions; foreign ownership and investment; data privacy requirements; competition and antitrust; investing; advertising; employment; product regulations; cybersecurity; environmental, health, and safety requirements; the responsible use of AI; sustainability; cryptocurrency; and consumer laws. Compliance with such requirements can be onerous and expensive, could impact our competitive position, and may negatively impact our business operations and ability to manufacture and ship our products. There can be no assurance that our employees, contractors, suppliers, customers or agents will not violate applicable laws or the policies, controls, and procedures that we have designed to help ensure compliance with such laws, and violations could result in fines, criminal sanctions against us, our officers, or our employees, prohibitions on the conduct of our business, and damage to our reputation. Changes to the laws, rules and regulations to which we are subject, or changes to their interpretation and enforcement, could lead to materially greater compliance and other costs, and/or further restrictions on our ability to manufacture and supply our products and operate our business. For example, we may face increased compliance costs as a result of changes or increases in antitrust legislation, regulation, administrative rule making, increased focus from regulators on cybersecurity vulnerabilities and risks. Our position in markets relating to AI has led to increased interest in our business from regulators worldwide, including the European Union, the United States, the United Kingdom, South Korea, Japan, and China. For example, the French Competition Authority collected information from us regarding our business and competition in the graphics card and CSP market as part of an ongoing inquiry into competition in those markets. We have also received, and continue to receive, broad requests for information from competition regulators in the European Union, the United States, the United Kingdom, China, and South Korea regarding our sales of GPUs and other NVIDIA products, our efforts to allocate supply, foundation models and our investments, partnerships and other agreements with companies developing foundation models, the markets in which we compete and our competition, our strategies, roadmaps, and efforts to develop, market, and sell hardware, software, and system solutions, and our agreements with customers, suppliers, and partners. We expect to receive additional requests for information in the future. Such requests have been and are likely to be expensive and burdensome and could negatively impact our business and our relationships with customers, suppliers, and partners.\nGovernments and regulators are also considering, and in certain cases, have imposed restrictions on the hardware, software, and systems used to develop frontier foundation models and generative AI. For example, the EU AI Act became effective on August 1, 2024 and will be fully applicable after a two-year transitional period. The EU AI Act may impact our ability to train, deploy, or release AI models in the EU. Several states are considering enacting or have already enacted regulations concerning AI technologies, with new state laws that took effect on January 1, 2026, which may impact our ability to train, deploy, or release AI models, and increase our compliance costs. Restrictions under these and any other regulations, if implemented, could increase the costs and burdens to us and our customers, delay or halt deployment of\nnew systems using our products, and reduce the number of new entrants and customers, negatively impacting our business and financial results. Revisions to laws or regulations or their interpretation and enforcement could also result in increased taxation, trade sanctions, the imposition of or increase to import duties or tariffs, restrictions and controls on imports or exports, or other retaliatory actions, which could have an adverse effect on our business plans or impact the timing of our shipments. Additionally, changes in the public perception of governments in the regions where we operate or plan to operate could negatively impact our business and results of operations.\nGovernment actions, including trade protection and national and economic security policies of U.S. and foreign government bodies, such as tariffs, import or export regulations, including deemed export restrictions and restrictions on the activities of U.S. persons, trade and economic sanctions, decrees, quotas or other trade barriers and restrictions could affect our ability to ship products, provide services to our customers and employees, do business without an export license with entities on the U.S. Department of Commerce's U.S. Entity List or other USG restricted parties lists (which is expected to change from time to time), and generally fulfill our contractual obligations and have a material adverse effect on our business. If we were ever found to have violated export control laws or sanctions of the U.S. or similar applicable non-U.S. laws, even if the violation occurred without our knowledge, we may be subject to various penalties available under the laws, any of which could have a material and adverse impact on our business, operating results and financial condition.\nFor example, in response to the war in Ukraine, the United States and other jurisdictions imposed economic sanctions and export control measures which blocked the passage of our products, services and support into Russia, Belarus, and certain regions of Ukraine. In fiscal year 2023, we stopped direct sales to Russia and closed business operations in Russia. Concurrently, the war in Ukraine has impacted sales in EMEA and may continue to do so in the future.\nThe increasing focus on the risks and strategic importance of AI technologies has resulted in regulatory restrictions that target products and services capable of enabling or facilitating AI and may in the future result in additional restrictions impacting some or all of our product and service offerings.\nConcerns regarding third-party use of AI for purposes contrary to local governmental interests, including concerns relating to the misuse of AI applications, models, and solutions, has resulted in and could in the future result in unilateral or multilateral restrictions on products that can be used for training, modifying, tuning, and deploying LLMs and other AI applications. Such restrictions have limited and could in the future limit the ability of downstream customers and users worldwide to acquire, deploy and use systems that include our products, software, and services, and negatively impact our business and financial results.\nSuch restrictions could include additional unilateral or multilateral export controls on certain products or technology, including but not limited to AI technologies. As geopolitical tensions have increased, semiconductors associated with AI, including GPUs and related products, are increasingly the focus of export control restrictions proposed by stakeholders in the U.S. and its allies. The United States has imposed unilateral worldwide controls restricting GPUs and associated products, and it is likely that additional unilateral or multilateral controls will be adopted. Such controls have been and may again be very broad in scope and application, prohibit us from exporting our products to any or all customers in one or more markets, and could negatively impact our manufacturing, testing and warehousing locations and options, or could impose other conditions that limit our ability to serve demand abroad and could negatively and materially impact our business, revenue and financial results. Export controls and other restrictions targeting GPUs and semiconductors associated with AI, which have been imposed and are likely to be more restrictive, would further limit our ability to export our technology, products, or services, creating a competitive disadvantage for us and negatively impacting our business and financial results. Export controls targeting GPUs and semiconductors associated with AI have subjected and may in the future subject downstream users of our products to restrictions on the use, resale, repair, or transfer of our products, negatively impacting our business and financial results. Controls could negatively impact our cost and/or ability to provide services such as NVIDIA AI cloud services and could impact the cost and/or ability for our CSPs and customers to provide services to their end customers, even outside China.\nExport controls have and could in the future disrupt our supply chain and distribution channels, negatively impacting our ability to serve demand, including in markets outside China and for our non-data center products. The possibility of additional export controls has negatively impacted and may in the future negatively impact demand for our products, benefiting competitors that offer alternatives less likely to be restricted by further controls. Repeated changes in the export control rules are likely to impose compliance burdens on our business and our customers, negatively and materially impacting our business.\nIncreasing use of economic sanctions and export controls has impacted and may in the future impact demand for our products or services, negatively impacting our business and financial results. Reduced demand due to export controls has and could in the future lead to excess inventory or cause us to incur related supply charges. Additional unilateral or multilateral controls are also likely to include deemed export control limitations that negatively impact the ability of our research and development teams to execute our roadmap or other objectives in a timely manner. Additional export restrictions may not only impact our ability to serve overseas markets, but also provoke responses from foreign governments, including China, that negatively impact our supply chain or our ability to provide our products and services to customers in all markets worldwide, which could also substantially reduce our revenue. Regulators in China have\ninquired about our sales and efforts to supply the China market and our fulfillment of the commitments we entered into at the close of our Mellanox acquisition. On September 15, 2025, China's antitrust regulators published their preliminary finding that our compliance with applicable U.S. export controls, which required us to offer degraded products to the Chinese market, discriminated unfairly against customers in the China market and therefore violated the terms of China's approval of our Mellanox acquisition. If regulators conclude that we have failed to fulfill the terms of our Mellanox acquisition or we have violated any applicable law in China, we could be subject to financial penalties, restrictions on our ability to conduct our business, restrictions or other orders regarding our networking business, products, and services, or otherwise impact our operations in China, any of which could have a material and adverse impact on our business, operating results and financial condition.\nWe continue to be subject to a series of shifting and expanding export control restrictions, impacting our ability to serve customers outside the United States.\nIn August 2022, the USG announced export restrictions and export licensing requirements targeting China's semiconductor and supercomputing industries. These restrictions impacted exports of certain chips, as well as software, hardware, equipment and technology used to develop, produce and manufacture certain chips to China (including Hong Kong and Macau) and Russia, and specifically impact our A100 and H100 integrated circuits, DGX or any other systems or boards which incorporate A100 or H100 integrated circuits.\nIn July 2023, the USG also informed us of an additional licensing requirement for a subset of A100 and H100 products destined to certain customers and other regions, including some countries in the Middle East.\nIn October 2023, the USG announced new and updated licensing requirements for exports to China and Country Groups D:1, D:4, and D:5 (including but not limited to, Saudi Arabia, the United Arab Emirates, and Vietnam, but excluding Israel) of our products exceeding certain performance thresholds, including, but not limited to, the A100, A800, H100, H800, L4, L40, L40S RTX 4090, GB200 NVL72, and B200. The licensing requirements also apply to the export of products exceeding certain performance thresholds to a party headquartered in, or with an ultimate parent headquartered in, Country Group D5, including China.\nIn April 2025, the USG informed us that it requires a license for export to China (including Hong Kong and Macau) and D:5 countries, or to companies headquartered or with an ultimate parent therein, of our H20 integrated circuits and any other circuits achieving the H20's memory bandwidth, interconnect bandwidth, or combination thereof. As a result of these requirements, we incurred a \\$4.5 billion charge in the first quarter of fiscal year 2026 associated with H20 for excess inventory and purchase obligations, as the demand for H20 products diminished.\nBeginning in August 2025, the USG granted licenses that would allow us to ship certain H2O products to certain China-based customers. USG officials expressed an expectation that the USG will receive $15\\%$ or more of the revenue generated from licensed sales of our products, but the USG did not publish a regulation codifying such requirement.\nBeginning in February 2026, the USG granted licenses that would allow us to ship small amounts of H200 products to specific China-based customers. To date, we have not generated any revenue under the H200 licensing program, and do not yet know whether any imports will be allowed into China. The license requires that the H200s go through an inspection process in the United States prior to any shipment to the customer. As a result, any H200 shipped under the new licensing program will be subject to a 25% tariff upon importation into the United States.\nIn the event that we are able to sell licensed products into the China market, we may not be able to pass along all or any of the tariff to our customers, and may be subject to litigation, increased costs, and a harmed competitive position.\nThe export controls applicable to China are complex and address a variety of parameters, including the total processing performance of a chip, the “performance density” of a chip, the interconnect bandwidth of a chip, and the memory bandwidth of a chip. Under the current rules and geopolitical landscape, we are unable to create and deliver a competitive product for China’s data center market that receives approval from both the USG and the Chinese government. As of the end of the first quarter of fiscal year 2027, while we were able to ship uncontrolled products to China, such as gaming and workstation GPUs, we were effectively foreclosed from competing in China's data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide. Unless we are able to return with a data center system that meets the approval of both the USG and the Chinese government, our lost opportunity and the benefit to our competitors will have a material and adverse impact on our business, operating results, and financial condition.\nIn addition to controls targeting D:1, D:4 and D:5 countries, the USG has also imposed worldwide export controls impacting our products, and may impose additional controls in the future.\nIn January 2025, the USG published the AI Diffusion IFR in the Federal Register. The IFR would have imposed a worldwide licensing requirement on our data center products, such as our H200, GB200 and GB300. The AI Diffusion IFR would have divided the world into three tiers, relegating most countries to “Tier 2” status, and would have created a complex and burdensome scheme for licensing approvals.\nIn May 2025, the USG announced that it would rescind the AI Diffusion IFR and implement a replacement rule. The scope, timing, and requirements of the forthcoming rule remain uncertain. The replacement rule may impose new restrictions on our products or operations and/or add license requirements that could have a material impact on our business, operating results, and financial condition. For example, in October 2025, the Senate passed the GAIN AI Act in the National Defense Authorization Act. The GAIN AI Act would restrict the Trump Administration's ability to adapt the Biden Administration's export control rules and could also allow private U.S. persons to review and overturn licensing and foreign policy decisions made by the Trump Administration. Congress is also considering legislation such as the Remote Access Security Act, or RASA, which could prohibit the provision of cloud services to any company with an ultimate parent headquartered in China. If enacted, RASA could impose new restrictions on cloud service providers and OEMs, and could have a material impact on our business, operating results, and financial condition.\nOur competitive position has been harmed by export controls, and our competitive position and future results will be further harmed, over the long term, if the restrictions remain in place or are expanded in geographic, customer, or product scope, if customers purchase product from competitors, if customers develop their own internal solution, if we are unable to provide contractual warranty or other extended service obligations, if the USG does not grant licenses in a timely manner or denies licenses to significant customers or if we incur significant transition costs. The licensing process may not be resolved before significant business opportunities evaporate. Even if the USG grants any requested licenses, the licenses have already and may in the future be temporary, impose burdensome conditions regarding the installation, maintenance, and use of such products, or include financial or economic requirements that we or our customers or end users cannot or choose not to fulfill. The licensing requirements have already and may in the future benefit certain of our competitors, as the licensing process will make our pre-sale and post-sale technical support efforts more cumbersome and less certain and encourage customers in China, the Middle East, and other regions to pursue alternatives to our products, including semiconductor suppliers based in China, Europe, and Israel.\nGiven the increasing strategic importance of AI and rising geopolitical tensions, the USG has changed and may again change the export control rules at any time and further subject a wider range of our products to export restrictions and licensing requirements, negatively impacting our business and financial results. In the event of such change, we may be unable to sell our inventory of such products and may be unable to develop replacement products not subject to the licensing requirements.\nFor example, the USG already imposed license conditions that limit the ability of foreign firms to create and offer as a service large-scale GPU clusters, such as imposing license conditions on the use of products to be exported to certain countries, and may impose additional conditions such as requiring chip tracking and throttling mechanisms that could disable or impair GPUs if certain events, including unauthorized system configuration, use, or location, are detected. Such government mandates in chip designs could introduce system vulnerabilities and expose us to significant risk and potential liability, negatively impact demand for our products, and could have a material impact on our business, operating results, and financial condition. Even if not enacted into binding legislation, draft bills have impacted and may in the future negatively impact our business. For example, following U.S. legislative proposals calling for mandatory features in our chips, China's government publicly questioned whether our H20 products have built-in vulnerabilities, discouraging customers from purchasing our products. We provided a public response explaining that our GPUs, including H20, do not include such built-in vulnerabilities, and will respond to any follow-up questions we receive.\nOpen-source foundation models are rapidly growing in popularity with developers worldwide. The demand for open-source foundation models and applications promotes use of our products worldwide. Any regulatory control or other restriction that limits our ability to provide products and services that support third-party applications and models, including applications built on foundation models originating in China such as DeepSeek, Qwen, or KIMMI, could have a material impact on our business, operating results, and financial condition.\nThe USG already imposed export controls restricting certain gaming GPUs, and if the USG expands such controls to restrict additional gaming products, it may disrupt a significant portion of our supply and distribution chain and negatively impact sales of such products to markets outside China, including the U.S. and Europe. For example, the French Competition Authority (FCA) is questioning whether gaming GPUs and data center GPUs are separate product categories, an inquiry that may impact the export controls applicable to gaming products sold in France and Europe. In addition, as the performance of the gaming GPUs increases over time, export controls may have a greater impact on our ability to compete in markets subject to those controls. Export controls may disrupt our supply and distribution chain for a substantial portion of our products, which are warehoused in and distributed from Hong Kong.\nExport controls restricting our ability to sell data center GPUs may also negatively impact demand for our networking products used in servers containing our GPUs. The USG may also impose export controls on our networking products, such as high-speed network interconnects, to limit the ability of downstream parties to create large clusters for frontier model training.\nExport controls have and are likely in the future to have a disproportionate impact on NVIDIA and may disadvantage us against certain of our competitors that sell chips that are outside the scope of such control. Export controls have already and may in the future encourage customers outside China and other impacted regions to “design-out” certain U.S. semiconductors from their products to reduce the compliance burden and risk, and to ensure that they are able to serve\nmarkets worldwide. Export controls have already encouraged and may in the future encourage overseas governments to request that our customers purchase from our competitors rather than NVIDIA or other U.S. firms, harming our business, market position, and financial results.\nAs a result, export controls have in the past and may in the future negatively impact demand for our products and services not only in China, but also in other markets, such as Europe, Latin America, and Southeast Asia. Export controls increase the risk of investing in U.S. advanced semiconductor products, because by the time a new product is ready for market, it may be subject to new unilateral export controls restricting its sale, resulting in excess inventory and purchase obligations as we recently experienced with the H20. At the same time, such controls may increase investment in foreign competitors, which would be less likely to be restricted by U.S. controls.\nThe increasingly complex export controls impose complex and burdensome compliance obligations on our partners, suppliers, and customers. We have provided and will continue to provide assistance to authorities regarding attempted diversion, but as we do not have physical control of our products after sale, we must also rely on the compliance programs of our customers and partners. While we seek to strictly comply with all applicable export control regulators, reports of diversion of controlled products, even when unsubstantiated and untrue, or any compliance failure at a customer or partner, may negatively impact our business, relationships with partners and customers, and our reputation. Incorrect allegations that our compliance efforts satisfy the letter but not the “spirit” of the applicable regulations, as well as incorrect allegations that legitimate and appropriate business is using supposed “loopholes” in the export controls may negatively impact our business, relationships with partners and customers, and our reputation.\nIn addition to export controls, the USG may impose restrictions on the import and sale of products that incorporate technologies developed or manufactured in whole or in part in China. For example, the USG adopted “Connected Vehicle” restrictions on the import and sale of certain automotive products in the United States, which if adopted and interpreted broadly, could impact our ability to develop and supply solutions for our automotive customers. The USG is also considering restrictions that would limit our ability to support third-party applications and models built on open-source foundation models originating in China. Such restrictions, if implemented, would favor our foreign competitors and negatively impact our business.\nAdditionally, restrictions imposed by the Chinese government on the duration of gaming activities and access to games may adversely affect our Edge Computing revenue, and even if we are able to participate in the China data center compute market, increased oversight of digital platform companies may adversely affect our Data Center revenue. The Chinese government has encouraged customers to purchase from our China-based competitors and discouraged customers from purchasing, importing, or using our data center products, including any China-specific product designed to comply with U.S. export controls. As another example, an agency of the Chinese government announced an Action Plan that endorses new standards regarding the compute performance per watt and per memory bandwidth of accelerators used in new and renovated data centers in China. Although we are already effectively foreclosed from the China market by U.S. export controls, if those controls changed to allow us to return to the market, the Chinese government could modify or implement the Action Plan in a way that effectively prevents us from being able to design products to meet the new standard, which may restrict the ability of customers to use some of our data center products and may have a material and adverse impact on our business, operating results and financial condition. Further restrictions on our products or the products of our suppliers could negatively impact our business and financial results.\nFinally, our business depends on our ability to receive consistent and reliable supply from our overseas partners, especially in Taiwan and South Korea. Any new restrictions that negatively impact our ability to receive supply of components, parts, or services from Taiwan and South Korea, would negatively impact our business and financial results.\nItem 2. Unregistered Sales of Equity Securities and Use of Proceeds", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information/Item 1A. Risk Factors", + "metadata": { + "length": 43060, + "summary": "The passage details risk factors for NVIDIA common stock, focusing on competition, supply-demand mismatches, and complex export controls. Competition could harm market share and financial results. Long manufacturing lead times and inaccurate demand estimates may cause supply-demand mismatches, leading to inventory issues or lost revenue. Export controls, particularly targeting AI and GPU products to China and other countries, have significantly impacted NVIDIA's business, causing charges, limiting market access, and benefiting competitors. The text also mentions risks from new product introductions, regulatory changes, and geopolitical tensions.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Other", + "risk", + "factors", + "listed", + "material", + "previously", + "Item", + "1A", + "Annual", + "Report", + "Form", + "10", + "fiscal", + "year", + "ended", + "January", + "25", + "2026", + "Purchasing", + "owning", + "NVIDIA", + "common", + "stock", + "involves", + "investment", + "risks", + "including", + "limited", + "Any", + "harm", + "business", + "financial", + "condition", + "results", + "operations", + "reputation", + "price", + "decline", + "Additional", + "trends", + "uncertainties", + "presently", + "immaterial", + "Competition", + "adversely", + "impact", + "market", + "share", + "Our", + "target", + "markets", + "remain", + "competitive", + "competition", + "intensify", + "expanding", + "changing", + "product", + "service", + "offerings", + "industry", + "standards", + "customer", + "entrants", + "consolidations", + "companies", + "compete", + "wide", + "range", + "parameters", + "total", + "cost", + "ownership", + "performance", + "resulted", + "future", + "result", + "lower", + "expected", + "selling", + "prices", + "demand", + "products", + "Some", + "competitors", + "operate", + "fabrication", + "facilities", + "longer", + "operating", + "histories", + "larger", + "bases", + "comprehensive", + "IP", + "portfolios", + "patent", + "protections", + "design", + "wins", + "greater", + "sales", + "marketing", + "distribution", + "resources", + "These", + "acquire", + "prevent", + "effectively", + "identify", + "capitalize", + "opportunities", + "end", + "user", + "quickly", + "transition", + "impinge", + "ability", + "procure", + "sufficient", + "foundry", + "capacity", + "scarce", + "input", + "materials", + "supply", + "constrained", + "environment", + "customers", + "developing", + "ASICs", + "designs", + "optimized", + "workloads", + "require", + "features", + "functionality", + "data", + "center", + "systems", + "provide", + "Others", + "offer", + "cloud", + "based", + "services", + "AI", + "establish", + "achieve", + "scale", + "meet", + "objectives", + "If", + "unable", + "successfully", + "technologies", + "decrease", + "negatively", + "Long", + "manufacturing", + "lead", + "times", + "uncertain", + "availability", + "combined", + "failure", + "estimate", + "accurately", + "led", + "mismatches", + "We", + "long", + "build", + "finished", + "maintain", + "inventory", + "advance", + "anticipated", + "In", + "periods", + "shortages", + "impacting", + "semiconductor", + "chain", + "extended", + "experienced", + "continue", + "experience", + "12", + "months", + "To", + "secure", + "paid", + "premiums", + "provided", + "deposits", + "entered", + "term", + "agreements", + "commitments", + "increased", + "costs", + "address", + "inaccurately", + "change", + "orders", + "past", + "reduce", + "time", + "rate", + "Significant", + "varied", + "platforms", + "excess", + "significantly", + "harmed", + "reoccur", + "underestimate", + "partners", + "contract", + "manufacturers", + "increase", + "production", + "timely", + "manner", + "relationships", + "damaged", + "lose", + "revenue", + "Additionally", + "part", + "complex", + "buildout", + "constraints", + "issues", + "respect", + "component", + "broader", + "For", + "sell", + "impeded", + "components", + "parties", + "overestimate", + "cancel", + "defer", + "choose", + "purchase", + "utilize", + "hand", + "average", + "due", + "channel", + "pricing", + "programs", + "suppliers", + "write", + "incur", + "cancellation", + "penalties", + "record", + "impairments", + "The", + "amplified", + "cancellable", + "returnable", + "historical", + "exacerbated", + "make", + "obligations", + "prepaids", + "grown", + "grow", + "portion", + "All", + "gross", + "margins", + "Factors", + "caused", + "timing", + "volume", + "include", + "development", + "cycles", + "competing", + "competitor", + "releases", + "announcements", + "actions", + "economic", + "conditions", + "sudden", + "sustained", + "government", + "lockdowns", + "public", + "health", + "rapidly", + "technology", + "requirements", + "energy", + "introductions", + "transitions", + "resulting", + "existing", + "unexpected", + "cases", + "purchasing", + "decisions", + "made", + "levels", + "held", + "distributors", + "ODMs", + "OEMs", + "system", + "integrators", + "developers", + "enhance", + "accelerated", + "computing", + "applications", + "leverage", + "related", + "large", + "language", + "models", + "ecosystem", + "architectures", + "underlying", + "governmental", + "policies", + "export", + "controls", + "restrictions", + "gaming", + "usage", + "tariffs", + "capital", + "infrastructure", + "party", + "content", + "GeForce", + "NOW", + "centers", + "support", + "crucial", + "shortage", + "Expanding", + "multi", + "process", + "involving", + "significant", + "regulatory", + "technical", + "construction", + "challenges", + "addition", + "access", + "capitalized", + "face", + "difficulties", + "securing", + "financing", + "projects", + "limitations", + "delay", + "partner", + "deployments", + "adoption", + "Challenges", + "estimating", + "pronounced", + "volatile", + "global", + "regional", + "basis", + "Extended", + "occur", + "natural", + "disasters", + "pandemics", + "events", + "Geopolitical", + "tensions", + "regions", + "rely", + "assembly", + "critical", + "continuity", + "adverse", + "Supply", + "affecting", + "memory", + "rising", + "drive", + "buildouts", + "higher", + "Publicly", + "announced", + "intentions", + "governments", + "complicate", + "estimates", + "binding", + "committed", + "volumes", + "purchases", + "projections", + "increasing", + "complexity", + "enter", + "prepaid", + "current", + "integration", + "creates", + "managing", + "multiple", + "variations", + "planning", + "execution", + "logistics", + "portfolio", + "varying", + "compatibility", + "quality", + "incurred", + "provisions", + "exceed", + "declines", + "investing", + "specialized", + "equipment", + "processes", + "domestic", + "delays", + "scaling", + "planned", + "capabilities", + "depend", + "ramp", + "required", + "Delays", + "shortfalls", + "Introducing", + "offering", + "concurrently", + "ship", + "architecture", + "simultaneously", + "prepare", + "generally", + "stages", + "introducing", + "Data", + "Center", + "Edge", + "Computing", + "experiencing", + "faster", + "launch", + "cadence", + "growing", + "diverse", + "set", + "introduced", + "solutions", + "seek", + "complete", + "variety", + "Rubin", + "platform", + "start", + "shipping", + "half", + "2027", + "frequency", + "number", + "configurations", + "magnify", + "create", + "volatility", + "Qualification", + "anticipating", + "reducing", + "prior", + "ahead", + "Customers", + "adopting", + "ready", + "affect", + "reduced", + "generation", + "anticipate", + "impacted", + "execute", + "architectural", + "reason", + "newly", + "unanticipated", + "magnitude", + "warranty", + "quarter", + "2025", + "low", + "yielding", + "Blackwell", + "engineering", + "roadmap", + "develop", + "adequately", + "manage", + "adopt", + "fast", + "forecasted", + "While", + "managed", + "concurrent", + "sold", + "efforts", + "difficult", + "impair", + "predict", + "mix", + "additional", + "Demand", + "incorrect", + "generate", + "Because", + "reasonable", + "degree", + "precision", + "reported", + "GPUs", + "mercurial", + "trendy", + "leading", + "inconsistent", + "spikes", + "drops", + "years", + "ago", + "began", + "mining", + "digital", + "currencies", + "Ethereum", + "It", + "cryptocurrency", + "forecast", + "Volatility", + "compute", + "cryptocurrencies", + "regulations", + "method", + "verifying", + "blockchain", + "transactions", + "Changes", + "2.0", + "merge", + "2022", + "This", + "created", + "aftermarket", + "retail", + "general", + "resold", + "online", + "unauthorized", + "gray", + "makes", + "forecasting", + "Gray", + "reseller", + "marketplaces", + "channels", + "inability", + "arises", + "subject", + "laws", + "rules", + "political", + "domestically", + "worldwide", + "areas", + "infringement", + "taxes", + "import", + "anti", + "corruption", + "Foreign", + "Corrupt", + "Practices", + "Act", + "acquisitions", + "foreign", + "exchange", + "cash", + "repatriation", + "privacy", + "antitrust", + "advertising", + "employment", + "cybersecurity", + "environmental", + "safety", + "responsible", + "sustainability", + "consumer", + "Compliance", + "onerous", + "expensive", + "position", + "manufacture", + "There", + "assurance", + "employees", + "contractors", + "agents", + "violate", + "applicable", + "procedures", + "designed", + "ensure", + "compliance", + "violations", + "fines", + "criminal", + "sanctions", + "officers", + "prohibitions", + "conduct", + "damage", + "interpretation", + "enforcement", + "materially", + "increases", + "legislation", + "regulation", + "administrative", + "rule", + "making", + "focus", + "regulators", + "vulnerabilities", + "relating", + "interest", + "European", + "Union", + "United", + "States", + "Kingdom", + "South", + "Korea", + "Japan", + "China", + "French", + "Authority", + "collected", + "information", + "graphics", + "card", + "CSP", + "ongoing", + "inquiry", + "received", + "receive", + "broad", + "requests", + "allocate", + "foundation", + "investments", + "partnerships", + "strategies", + "roadmaps", + "hardware", + "software", + "expect", + "Such", + "burdensome", + "Governments", + "imposed", + "frontier", + "generative", + "EU", + "effective", + "August", + "2024", + "fully", + "transitional", + "period", + "train", + "deploy", + "release", + "Several", + "states", + "enacting", + "enacted", + "state", + "effect", + "Restrictions", + "implemented", + "burdens", + "halt", + "deployment", + "Revisions", + "taxation", + "trade", + "imposition", + "duties", + "imports", + "exports", + "retaliatory", + "plans", + "shipments", + "perception", + "plan", + "Government", + "protection", + "national", + "security", + "bodies", + "deemed", + "activities", + "persons", + "decrees", + "quotas", + "barriers", + "license", + "entities", + "Department", + "Commerce", + "Entity", + "List", + "USG", + "restricted", + "lists", + "fulfill", + "contractual", + "found", + "violated", + "control", + "similar", + "violation", + "occurred", + "knowledge", + "response", + "war", + "Ukraine", + "jurisdictions", + "measures", + "blocked", + "passage", + "Russia", + "Belarus", + "2023", + "stopped", + "direct", + "closed", + "Concurrently", + "EMEA", + "strategic", + "importance", + "capable", + "enabling", + "facilitating", + "Concerns", + "purposes", + "contrary", + "local", + "interests", + "concerns", + "misuse", + "unilateral", + "multilateral", + "training", + "modifying", + "tuning", + "deploying", + "LLMs", + "limit", + "downstream", + "users", + "As", + "geopolitical", + "semiconductors", + "increasingly", + "proposed", + "stakeholders", + "allies", + "restricting", + "adopted", + "scope", + "application", + "prohibit", + "exporting", + "testing", + "warehousing", + "locations", + "options", + "impose", + "serve", + "abroad", + "Export", + "targeting", + "restrictive", + "creating", + "disadvantage", + "subjected", + "resale", + "repair", + "transfer", + "Controls", + "CSPs", + "disrupt", + "possibility", + "benefiting", + "alternatives", + "Repeated", + "Increasing", + "Reduced", + "charges", + "research", + "teams", + "overseas", + "provoke", + "responses", + "substantially", + "Regulators", + "inquired", + "fulfillment", + "close", + "Mellanox", + "acquisition", + "On", + "September", + "15", + "published", + "preliminary", + "finding", + "degraded", + "Chinese", + "discriminated", + "unfairly", + "terms", + "approval", + "conclude", + "failed", + "law", + "networking", + "series", + "shifting", + "licensing", + "supercomputing", + "industries", + "chips", + "produce", + "Hong", + "Kong", + "Macau", + "specifically", + "A100", + "H100", + "integrated", + "circuits", + "DGX", + "boards", + "incorporate", + "July", + "informed", + "requirement", + "subset", + "destined", + "countries", + "Middle", + "East", + "October", + "updated", + "Country", + "Groups", + "Saudi", + "Arabia", + "Arab", + "Emirates", + "Vietnam", + "excluding", + "Israel", + "exceeding", + "thresholds", + "A800", + "H800", + "L4", + "L40", + "L40S", + "RTX", + "4090", + "GB200", + "NVL72", + "B200", + "apply", + "headquartered", + "ultimate", + "parent", + "Group", + "D5", + "April", + "requires", + "H20", + "achieving", + "bandwidth", + "interconnect", + "combination", + "thereof", + "4.5", + "billion", + "charge", + "diminished", + "Beginning", + "granted", + "licenses", + "H2O", + "officials", + "expressed", + "expectation", + "generated", + "licensed", + "publish", + "codifying", + "February", + "small", + "amounts", + "H200", + "specific", + "date", + "program", + "allowed", + "H200s", + "inspection", + "shipment", + "shipped", + "25%", + "tariff", + "importation", + "event", + "pass", + "litigation", + "processing", + "chip", + "density", + "Under", + "landscape", + "deliver", + "receives", + "uncontrolled", + "workstation", + "foreclosed", + "foreclosure", + "helped", + "developer", + "ecosystems", + "challenge", + "Unless", + "return", + "meets", + "lost", + "opportunity", + "benefit", + "Diffusion", + "IFR", + "Federal", + "Register", + "GB300", + "divided", + "world", + "tiers", + "relegating", + "Tier", + "status", + "scheme", + "approvals", + "May", + "rescind", + "implement", + "replacement", + "forthcoming", + "add", + "Senate", + "passed", + "GAIN", + "National", + "Defense", + "Authorization", + "restrict", + "Trump", + "Administration", + "adapt", + "Biden", + "private", + "review", + "overturn", + "policy", + "Congress", + "Remote", + "Access", + "Security", + "RASA", + "provision", + "company", + "providers", + "place", + "expanded", + "geographic", + "internal", + "solution", + "grant", + "denies", + "resolved", + "evaporate", + "Even", + "grants", + "requested", + "temporary", + "installation", + "maintenance", + "pre", + "sale", + "post", + "cumbersome", + "encourage", + "pursue", + "Europe", + "Given", + "changed", + "wider", + "firms", + "GPU", + "clusters", + "imposing", + "exported", + "requiring", + "tracking", + "throttling", + "mechanisms", + "disable", + "configuration", + "location", + "detected", + "mandates", + "introduce", + "expose", + "potential", + "liability", + "draft", + "bills", + "legislative", + "proposals", + "calling", + "mandatory", + "publicly", + "questioned", + "built", + "discouraging", + "explaining", + "respond", + "follow", + "questions", + "Open", + "source", + "popularity", + "open", + "promotes", + "restriction", + "limits", + "originating", + "DeepSeek", + "Qwen", + "KIMMI", + "expands", + "FCA", + "questioning", + "separate", + "categories", + "France", + "substantial", + "warehoused", + "distributed", + "servers", + "high", + "speed", + "network", + "interconnects", + "model", + "disproportionate", + "burden", + "encouraged", + "request", + "harming", + "Latin", + "America", + "Southeast", + "Asia", + "advanced", + "recently", + "At", + "assistance", + "authorities", + "attempted", + "diversion", + "physical", + "strictly", + "comply", + "reports", + "controlled", + "unsubstantiated", + "untrue", + "Incorrect", + "allegations", + "satisfy", + "letter", + "spirit", + "legitimate", + "supposed", + "loopholes", + "developed", + "manufactured", + "Connected", + "Vehicle", + "automotive", + "interpreted", + "broadly", + "favor", + "duration", + "games", + "participate", + "oversight", + "discouraged", + "importing", + "agency", + "Action", + "Plan", + "endorses", + "watt", + "accelerators", + "renovated", + "Although", + "modify", + "prevents", + "standard", + "Further", + "Finally", + "depends", + "consistent", + "reliable", + "Taiwan", + "parts", + "Unregistered", + "Sales", + "Equity", + "Securities", + "Use", + "Proceeds" + ], + "keywords": [ + "export controls", + "competition", + "supply chain" + ], + "connect_to": [] + } + }, + { + "chunk_id": "9cc51452-bdb5-54ac-885e-d2a6ca0f0552", + "type": "table", + "content": "
PeriodTotal Number of Shares Purchased (In millions)Average Price Paid per Share (1)Total Number of Shares Purchased as Part of Publicly Announced Program (In millions)Approximate Dollar Value of Shares that May Yet Be Purchased Under the Program (In billions)
January 26, 2026 - February 22, 20268.2$ 186.358.2$ 57.0
February 23, 2026 - March 22, 20269.0$ 183.389.0$ 55.4
March 23, 2026 - April 26, 202691.1$ 184.9891.1$ 38.5
Total108.3108.3
", + "path": "tables/table-39 Stock Buyback.html", + "metadata": { + "length": 723, + "summary": "table-40\nThe table shows a company repurchased 108.3 million shares from Jan to Apr 2026, with $38.5 billion remaining in the buyback program.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-39 Stock Buyback.html", + "keywords": [ + "share repurchase", + "buyback program", + "stock" + ], + "tokens": [] + } + }, + { + "chunk_id": "03f3b2b0-e0b7-5825-9b53-806644880b43", + "type": "text", + "content": "We repurchased 108 million shares of our common stock for \\$20.2 billion during the first quarter of fiscal year 2027. As of April 26, 2026, we were authorized, subject to certain specifications, to repurchase up to \\$38.5 billion of our common stock.\nOn May 18, 2026, our Board of Directors approved an additional \\$80.0 billion in share repurchase authorization, without expiration.\nWe paid cash dividends to our shareholders of \\$243 million during the first quarter of fiscal year 2027. On May 18, 2026, we increased our quarterly cash dividend from \\$0.01 per share to \\$0.25 per share to all shareholders of record on June 4, 2026. Our quarterly cash dividend will be paid on June 26, 2026.\nThe payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders.\nThe following table presents details of our share repurchase transactions during the first quarter of fiscal year 2027:\n\n[tables/table-39 Stock Buyback.html]\n\n(1) Average price paid per share includes broker commissions but excludes our liability under the 1% excise tax on the net amount of our share repurchases required by the Inflation Reduction Act of 2022.\nWe may execute repurchases from time to time, subject to market conditions, operating requirements and other investment opportunities, in the open market, in privately-negotiated transactions, pursuant to a Rule 10b5-1 trading plan or in structured share repurchase agreements in compliance with Rule 10b-18 of the Exchange Act. Our share repurchase program may be suspended at any time at our discretion.\nEmployee Equity Incentive Program Share Withholding\nWe withhold shares of our common stock associated with net share settlements to cover tax withholding obligations of awards under our employee equity incentive program. During the first quarter of fiscal year 2027, we withheld approximately 12 million shares, for a total value of \\$2.1 billion through net share settlements.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information/Issuer Purchases of Equity Securities", + "metadata": { + "length": 2088, + "summary": "In the first quarter of fiscal 2027, the company repurchased 108 million shares for $20.2 billion. As of April 26, 2026, authorization remained for up to $38.5 billion in repurchases. On May 18, 2026, the Board approved an additional $80.0 billion in repurchase authorization. Cash dividends of $243 million were paid in the first quarter, and the quarterly dividend was increased from $0.01 to $0.25 per share, payable on June 26, 2026. Additionally, 12 million shares were withheld for tax obligations under the employee equity incentive program, valued at $2.1 billion.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "We", + "repurchased", + "108", + "million", + "shares", + "common", + "stock", + "20.2", + "billion", + "quarter", + "fiscal", + "year", + "2027", + "As", + "April", + "26", + "2026", + "authorized", + "subject", + "specifications", + "repurchase", + "38.5", + "On", + "May", + "18", + "Board", + "Directors", + "approved", + "additional", + "80.0", + "share", + "authorization", + "expiration", + "paid", + "cash", + "dividends", + "shareholders", + "243", + "increased", + "quarterly", + "dividend", + "0.01", + "0.25", + "record", + "June", + "Our", + "The", + "payment", + "future", + "continuing", + "determination", + "declaration", + "interests", + "table", + "presents", + "details", + "transactions", + "Average", + "price", + "includes", + "broker", + "commissions", + "excludes", + "liability", + "1%", + "excise", + "tax", + "net", + "amount", + "repurchases", + "required", + "Inflation", + "Reduction", + "Act", + "2022", + "execute", + "time", + "market", + "conditions", + "operating", + "requirements", + "investment", + "opportunities", + "open", + "privately", + "negotiated", + "pursuant", + "Rule", + "10b5", + "trading", + "plan", + "structured", + "agreements", + "compliance", + "10b", + "Exchange", + "program", + "suspended", + "discretion", + "Employee", + "Equity", + "Incentive", + "Program", + "Share", + "Withholding", + "withhold", + "settlements", + "cover", + "withholding", + "obligations", + "awards", + "employee", + "equity", + "incentive", + "During", + "withheld", + "approximately", + "12", + "total", + "2.1" + ], + "keywords": [ + "share repurchase", + "dividends", + "fiscal 2027" + ], + "connect_to": [ + { + "target": "9cc51452-bdb5-54ac-885e-d2a6ca0f0552", + "relation": "embeds", + "ref": "[tables/table-39 Stock Buyback.html]", + "position": { + "start": 998, + "end": 1034 + } + } + ] + } + }, + { + "chunk_id": "727dfcb8-3748-54f2-9b52-0cd315c4277e", + "type": "text", + "content": "On February 17, 2026, we acquired a company and issued to a key employee a total of 37,890 shares of our common stock, valued at approximately \\$7 million based on our closing stock price on the issuance date.\nOn April 10, 2026, we acquired a company and issued to key employees a total of 72,972 shares of our common stock, valued at approximately \\$14 million based on our closing stock price on the issuance date.\nThe above securities were issued in transactions not involving a public offering pursuant to an exemption from registration set forth in Section 4(a)(2) of the Securities Act (and Regulation D or Regulation S promulgated thereunder).", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information/Recent Sales of Unregistered Securities and Use of Proceeds", + "metadata": { + "length": 650, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "On", + "February", + "17", + "2026", + "acquired", + "company", + "issued", + "key", + "employee", + "total", + "37", + "890", + "shares", + "common", + "stock", + "valued", + "approximately", + "million", + "based", + "closing", + "price", + "issuance", + "date", + "April", + "10", + "employees", + "72", + "972", + "14", + "The", + "securities", + "transactions", + "involving", + "public", + "offering", + "pursuant", + "exemption", + "registration", + "set", + "Section", + "Securities", + "Act", + "Regulation", + "promulgated", + "thereunder" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "993e44f9-cbc8-50ac-832f-a0a23a30862d", + "type": "table", + "content": "
NameTitle of Director or OfficerActionDateTotal Shares of Common Stock to be SoldExpiration Date
Tench CoxeDirectorAdoption3/19/20268,000,000*10/30/2027
Colette M. KressExecutive Vice President and Chief Financial OfficerTermination4/10/2026500,000**N/A
", + "path": "tables/table-40 Stock Sale Plans.html", + "metadata": { + "length": 455, + "summary": "table-41\nTench Coxe, a Director, adopted a plan to sell up to 8,000,000 shares from 3/19/2026 to 10/30/2027. Colette M. Kress, CFO, terminated a plan to sell 500,000 shares on 4/10/2026.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-40 Stock Sale Plans.html", + "keywords": [ + "Director", + "Officer", + "Stock Sale" + ], + "tokens": [] + } + }, + { + "chunk_id": "e589e208-8253-56b6-8606-8458665a4674", + "type": "table", + "content": "
Exhibit No.Exhibit DescriptionIncorporated by ReferenceFiling Date
Schedule/FormExhibit
10.1+Variable Compensation Plan - Fiscal Year 20278-K10.13/6/2026
31.1*Certification of Chief Executive Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
31.2*Certification of Chief Financial Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
32.1#*Certification of Chief Executive Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
32.2#*Certification of Chief Financial Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
101.INS*Inline XBRL Instance Document - the instance document does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document.
101.SCH*Inline XBRL Taxonomy Extension Schema Document
101.CAL*Inline XBRL Taxonomy Extension Calculation Linkbase Document
101.DEF*Inline XBRL Taxonomy Extension Definition Linkbase Document
101.LAB*Inline XBRL Taxonomy Extension Labels Linkbase Document
101.PRE*Inline XBRL Taxonomy Extension Presentation Linkbase Document
104Cover Page Interactive Data File - the cover page interactive data file does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document.
", + "path": "tables/table-41 Exhibit List.html", + "metadata": { + "length": 2074, + "summary": "table-42\nThe table lists exhibits filed with the SEC, including a variable compensation plan, CEO/CFO certifications, and inline XBRL documents, with filing dates and incorporation references.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "file_path": "tables/table-41 Exhibit List.html", + "keywords": [ + "Exhibits", + "XBRL", + "Certifications" + ], + "tokens": [] + } + }, + { + "chunk_id": "5ed2d7ba-7ced-5989-845f-4292d580edef", + "type": "text", + "content": "The following members of our Board of Directors and/or officers adopted, modified, or terminated a trading arrangement that is intended to satisfy the affirmative defense conditions of Rule 10b5-1(c), or a Rule 10b5-1 Trading Arrangement:\n\n[tables/table-40 Stock Sale Plans.html]\n\n\\* The Rule 10b5-1 Trading Arrangement is solely for gifts to charitable donor-advised funds.\n\\*\\* The Rule 10b5-1 Trading Arrangement was adopted on December 18, 2025, for sales through March 23, 2027. No shares were sold under the Rule 10b5-1 Trading Arrangement prior to termination.\nItem 6. Exhibits\n\n[tables/table-41 Exhibit List.html]\n\n+ Management contract or compensatory plan or arrangement.\n\\* Filed herewith.\n\\# In accordance with Item 601(b)(32)(ii) of Regulation S-K and SEC Release Nos. 33-8238 and 34-47986, Final Rule: Management's Reports on Internal Control Over Financial Reporting and Certification of Disclosure in Exchange Act Periodic Reports, the certifications furnished in Exhibits 32.1 and 32.2 hereto are deemed to accompany this Quarterly Report on Form 10-Q and will not be deemed “filed” for purpose of Section 18 of the Exchange Act. Such certifications will not be deemed to be incorporated by reference into any filing under the Securities Act or the Exchange Act, except to the extent that the registrant specifically incorporates it by reference.\nCopies of the above exhibits not contained herein are available to any shareholder upon written request to:\nInvestor Relations: NVIDIA Corporation, 2788 San Tomas Expressway, Santa Clara, CA 95051.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information/Item 5. Other Information", + "metadata": { + "length": 1680, + "summary": "The passage reports that members of the Board of Directors and officers adopted, modified, or terminated Rule 10b5-1 trading arrangements. One arrangement is solely for gifts to charitable donor-advised funds. Another was adopted on December 18, 2025, for sales through March 23, 2027, with no shares sold prior to termination. The document also lists exhibits for the quarterly report, including management contracts and certifications, and provides contact information for shareholder requests.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "The", + "members", + "Board", + "Directors", + "officers", + "adopted", + "modified", + "terminated", + "trading", + "arrangement", + "intended", + "satisfy", + "affirmative", + "defense", + "conditions", + "Rule", + "10b5", + "Trading", + "Arrangement", + "solely", + "gifts", + "charitable", + "donor", + "advised", + "funds", + "December", + "18", + "2025", + "sales", + "March", + "23", + "2027", + "No", + "shares", + "sold", + "prior", + "termination", + "Item", + "Exhibits", + "Management", + "contract", + "compensatory", + "plan", + "Filed", + "herewith", + "In", + "accordance", + "601", + "32", + "ii", + "Regulation", + "SEC", + "Release", + "Nos", + "33", + "8238", + "34", + "47986", + "Final", + "Reports", + "Internal", + "Control", + "Over", + "Financial", + "Reporting", + "Certification", + "Disclosure", + "Exchange", + "Act", + "Periodic", + "certifications", + "furnished", + "32.1", + "32.2", + "hereto", + "deemed", + "accompany", + "Quarterly", + "Report", + "Form", + "10", + "filed", + "purpose", + "Section", + "Such", + "incorporated", + "reference", + "filing", + "Securities", + "extent", + "registrant", + "specifically", + "incorporates", + "Copies", + "exhibits", + "contained", + "shareholder", + "written", + "request", + "Investor", + "Relations", + "NVIDIA", + "Corporation", + "2788", + "San", + "Tomas", + "Expressway", + "Santa", + "Clara", + "CA", + "95051" + ], + "keywords": [ + "Rule 10b5-1", + "Trading Arrangement", + "Board of Directors" + ], + "connect_to": [ + { + "target": "993e44f9-cbc8-50ac-832f-a0a23a30862d", + "relation": "embeds", + "ref": "[tables/table-40 Stock Sale Plans.html]", + "position": { + "start": 240, + "end": 279 + } + }, + { + "target": "e589e208-8253-56b6-8606-8458665a4674", + "relation": "embeds", + "ref": "[tables/table-41 Exhibit List.html]", + "position": { + "start": 586, + "end": 621 + } + } + ] + } + }, + { + "chunk_id": "75a20916-6ff3-5d47-8161-f62bc23bc42a", + "type": "text", + "content": "Pursuant to the requirements of the Securities Exchange Act of 1934, the registrant has duly caused this report to be signed on its behalf by the undersigned thereunto duly authorized.\nDate: May 20, 2026\nNVIDIA Corporation\nBy: /s/ Colette M. Kress\nColette M. Kress\nExecutive Vice President and Chief Financial Officer (Duly Authorized Officer and Principal Financial Officer)", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Signature", + "metadata": { + "length": 375, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Pursuant", + "requirements", + "Securities", + "Exchange", + "Act", + "1934", + "registrant", + "duly", + "caused", + "report", + "signed", + "behalf", + "undersigned", + "thereunto", + "authorized", + "Date", + "May", + "20", + "2026", + "NVIDIA", + "Corporation", + "By", + "Colette", + "Kress", + "Executive", + "Vice", + "President", + "Chief", + "Financial", + "Officer", + "Duly", + "Authorized", + "Principal" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "27190189-b621-5b6c-9a1f-faff4d0e9651", + "type": "text", + "content": "I, Jen-Hsun Huang, certify that:\n1. I have reviewed this Quarterly Report on Form 10-Q of NVIDIA Corporation;\n2. Based on my knowledge, this report does not contain any untrue statement of a material fact or omit to state a material fact necessary to make the statements made, in light of the circumstances under which such statements were made, not misleading with respect to the period covered by this report;\n3. Based on my knowledge, the financial statements, and other financial information included in this report, fairly present in all material respects the financial condition, results of operations and cash flows of the registrant as of, and for, the periods presented in this report;\n4. The registrant's other certifying officer(s) and I are responsible for establishing and maintaining disclosure controls and procedures (as defined in Exchange Act Rules 13a-15(e) and 15d-15(e)) and internal control over financial reporting (as defined in Exchange Act Rules 13a-15(f) and 15d-15(f)) for the registrant and have:\n(a) designed such disclosure controls and procedures, or caused such disclosure controls and procedures to be designed under our supervision, to ensure that material information relating to the registrant, including its consolidated subsidiaries, is made known to us by others within those entities, particularly during the period in which this report is being prepared;\n(b) designed such internal control over financial reporting, or caused such internal control over financial reporting to be designed under our supervision, to provide reasonable assurance regarding the reliability of financial reporting and the preparation of financial statements for external purposes in accordance with generally accepted accounting principles;\n(c) evaluated the effectiveness of the registrant's disclosure controls and procedures and presented in this report our conclusions about the effectiveness of the disclosure controls and procedures, as of the end of the period covered by this report based on such evaluation; and\n(d) disclosed in this report any change in the registrant's internal control over financial reporting that occurred during the registrant's most recent fiscal quarter (the registrant's fourth fiscal quarter in the case of an annual report) that has materially affected, or is reasonably likely to materially affect, the registrant's internal control over financial reporting; and\n5. The registrant's other certifying officer(s) and I have disclosed, based on our most recent evaluation of internal control over financial reporting, to the registrant's auditors and the audit committee of the registrant's board of directors (or persons performing the equivalent functions):\n(a) all significant deficiencies and material weaknesses in the design or operation of internal control over financial reporting which are reasonably likely to adversely affect the registrant's ability to record, process, summarize and report financial information; and\n(b) any fraud, whether or not material, that involves management or other employees who have a significant role in the registrant's internal control over financial reporting.\nDate: May 20, 2026\n/s/JEN-HSUN HUANG\nJen-Hsun Huang\nPresident and Chief Executive Officer", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/CERTIFICATION", + "metadata": { + "length": 3249, + "summary": "Jen-Hsun Huang, President and CEO of NVIDIA Corporation, certifies the Quarterly Report on Form 10-Q. He confirms the report contains no material misstatements, the financial statements fairly present the company's condition and results, and he is responsible for establishing and maintaining disclosure controls and internal control over financial reporting. He also discloses that he has evaluated these controls and reported any significant deficiencies or fraud to auditors and the audit committee. The certification is dated May 20, 2026.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Jen", + "Hsun", + "Huang", + "certify", + "reviewed", + "Quarterly", + "Report", + "Form", + "10", + "NVIDIA", + "Corporation", + "Based", + "knowledge", + "report", + "untrue", + "statement", + "material", + "fact", + "omit", + "state", + "make", + "statements", + "made", + "light", + "circumstances", + "misleading", + "respect", + "period", + "covered", + "financial", + "information", + "included", + "fairly", + "present", + "respects", + "condition", + "results", + "operations", + "cash", + "flows", + "registrant", + "periods", + "presented", + "The", + "certifying", + "officer", + "responsible", + "establishing", + "maintaining", + "disclosure", + "controls", + "procedures", + "defined", + "Exchange", + "Act", + "Rules", + "13a", + "15", + "15d", + "internal", + "control", + "reporting", + "designed", + "caused", + "supervision", + "ensure", + "relating", + "including", + "consolidated", + "subsidiaries", + "entities", + "prepared", + "provide", + "reasonable", + "assurance", + "reliability", + "preparation", + "external", + "purposes", + "accordance", + "generally", + "accepted", + "accounting", + "principles", + "evaluated", + "effectiveness", + "conclusions", + "end", + "based", + "evaluation", + "disclosed", + "change", + "occurred", + "recent", + "fiscal", + "quarter", + "fourth", + "case", + "annual", + "materially", + "affected", + "affect", + "auditors", + "audit", + "committee", + "board", + "directors", + "persons", + "performing", + "equivalent", + "functions", + "significant", + "deficiencies", + "weaknesses", + "design", + "operation", + "adversely", + "ability", + "record", + "process", + "summarize", + "fraud", + "involves", + "management", + "employees", + "role", + "Date", + "May", + "20", + "2026", + "JEN", + "HSUN", + "HUANG", + "President", + "Chief", + "Executive", + "Officer" + ], + "keywords": [ + "CEO", + "Certification", + "10-Q" + ], + "connect_to": [] + } + }, + { + "chunk_id": "86814477-98c5-57d6-9a48-f8eff5be6f58", + "type": "text", + "content": "I, Colette M. Kress, certify that:\n1. I have reviewed this Quarterly Report on Form 10-Q of NVIDIA Corporation;\n2. Based on my knowledge, this report does not contain any untrue statement of a material fact or omit to state a material fact necessary to make the statements made, in light of the circumstances under which such statements were made, not misleading with respect to the period covered by this report;\n3. Based on my knowledge, the financial statements, and other financial information included in this report, fairly present in all material respects the financial condition, results of operations and cash flows of the registrant as of, and for, the periods presented in this report;\n4. The registrant's other certifying officer(s) and I are responsible for establishing and maintaining disclosure controls and procedures (as defined in Exchange Act Rules 13a-15(e) and 15d-15(e)) and internal control over financial reporting (as defined in Exchange Act Rules 13a-15(f) and 15d-15(f)) for the registrant and have:\n(a) designed such disclosure controls and procedures, or caused such disclosure controls and procedures to be designed under our supervision, to ensure that material information relating to the registrant, including its consolidated subsidiaries, is made known to us by others within those entities, particularly during the period in which this report is being prepared;\n(b) designed such internal control over financial reporting, or caused such internal control over financial reporting to be designed under our supervision, to provide reasonable assurance regarding the reliability of financial reporting and the preparation of financial statements for external purposes in accordance with generally accepted accounting principles;\n(c) evaluated the effectiveness of the registrant's disclosure controls and procedures and presented in this report our conclusions about the effectiveness of the disclosure controls and procedures, as of the end of the period covered by this report based on such evaluation; and\n(d) disclosed in this report any change in the registrant's internal control over financial reporting that occurred during the registrant's most recent fiscal quarter (the registrant's fourth fiscal quarter in the case of an annual report) that has materially affected, or is reasonably likely to materially affect, the registrant's internal control over financial reporting; and\n5. The registrant's other certifying officer(s) and I have disclosed, based on our most recent evaluation of internal control over financial reporting, to the registrant's auditors and the audit committee of the registrant's board of directors (or persons performing the equivalent functions):\n(a) all significant deficiencies and material weaknesses in the design or operation of internal control over financial reporting which are reasonably likely to adversely affect the registrant's ability to record, process, summarize and report financial information; and\n(b) any fraud, whether or not material, that involves management or other employees who have a significant role in the registrant's internal control over financial reporting.\nDate: May 20, 2026\n/s/ COLETTE M. KRESS\nColette M. Kress\nExecutive Vice President and Chief Financial Officer", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/CERTIFICATION_2", + "metadata": { + "length": 3271, + "summary": "Colette M. Kress, Executive Vice President and Chief Financial Officer of NVIDIA Corporation, certifies the accuracy and completeness of the Quarterly Report on Form 10-Q. The certification covers the review of the report, the fairness of financial statements, the design and evaluation of disclosure controls and procedures, and the disclosure of any changes in internal control over financial reporting. It also includes reporting of significant deficiencies, material weaknesses, and fraud to auditors and the audit committee. The certification is dated May 20, 2026.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Colette", + "Kress", + "certify", + "reviewed", + "Quarterly", + "Report", + "Form", + "10", + "NVIDIA", + "Corporation", + "Based", + "knowledge", + "report", + "untrue", + "statement", + "material", + "fact", + "omit", + "state", + "make", + "statements", + "made", + "light", + "circumstances", + "misleading", + "respect", + "period", + "covered", + "financial", + "information", + "included", + "fairly", + "present", + "respects", + "condition", + "results", + "operations", + "cash", + "flows", + "registrant", + "periods", + "presented", + "The", + "certifying", + "officer", + "responsible", + "establishing", + "maintaining", + "disclosure", + "controls", + "procedures", + "defined", + "Exchange", + "Act", + "Rules", + "13a", + "15", + "15d", + "internal", + "control", + "reporting", + "designed", + "caused", + "supervision", + "ensure", + "relating", + "including", + "consolidated", + "subsidiaries", + "entities", + "prepared", + "provide", + "reasonable", + "assurance", + "reliability", + "preparation", + "external", + "purposes", + "accordance", + "generally", + "accepted", + "accounting", + "principles", + "evaluated", + "effectiveness", + "conclusions", + "end", + "based", + "evaluation", + "disclosed", + "change", + "occurred", + "recent", + "fiscal", + "quarter", + "fourth", + "case", + "annual", + "materially", + "affected", + "affect", + "auditors", + "audit", + "committee", + "board", + "directors", + "persons", + "performing", + "equivalent", + "functions", + "significant", + "deficiencies", + "weaknesses", + "design", + "operation", + "adversely", + "ability", + "record", + "process", + "summarize", + "fraud", + "involves", + "management", + "employees", + "role", + "Date", + "May", + "20", + "2026", + "COLETTE", + "KRESS", + "Executive", + "Vice", + "President", + "Chief", + "Financial", + "Officer" + ], + "keywords": [ + "certification", + "financial reporting", + "internal controls" + ], + "connect_to": [] + } + }, + { + "chunk_id": "8ef0e44e-fa18-5af6-8765-f95bae292d29", + "type": "text", + "content": "Pursuant to the requirement set forth in Rule 13a-14(b) of the Securities Exchange Act of 1934, as amended (the “Exchange Act”), and Section 1350 of Chapter 63 of Title 18 of the United States Code (18 U.S.C. § 1350), Jen-Hsun Huang, the President and Chief Executive Officer of NVIDIA Corporation (the “Company”), hereby certifies that, to the best of his knowledge:\n1. The Company's Quarterly Report on Form 10-Q for the period ended April 26, 2026, to which this Certification is attached as Exhibit 32.1 (the \"Periodic Report\"), fully complies with the requirements of Section 13(a) or Section 15(d) of the Exchange Act; and\n2. The information contained in the Periodic Report fairly presents, in all material respects, the financial condition of the Company at the end of the period covered by the Periodic Report and results of operations of the Company for the period covered by the Periodic Report.\nDate: May 20, 2026\n/s/JEN-HSUN HUANG\nJen-Hsun Huang\nPresident and Chief Executive Officer\nA signed original of this written statement required by Section 906 of 18 U.S.C. § 1350 has been provided to NVIDIA Corporation and will be retained by NVIDIA Corporation and furnished to the Securities and Exchange Commission or its staff upon request.\nThis certification accompanies the Form 10-Q to which it relates, is not deemed filed with the Securities and Exchange Commission and is not to be incorporated by reference into any filing of the Company under the Securities Act of 1933, as amended, or the Exchange Act (whether made before or after the date of the Form 10-Q), irrespective of any general incorporation language contained in such filing.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/CERTIFICATION_3", + "metadata": { + "length": 1655, + "summary": "Jen-Hsun Huang, President and CEO of NVIDIA Corporation, certifies under Rule 13a-14(b) and 18 U.S.C. § 1350 that the Quarterly Report on Form 10-Q for the period ended April 26, 2026, complies with Exchange Act requirements and fairly presents the company's financial condition and results of operations. The certification is dated May 20, 2026, and is not deemed filed with the SEC.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Pursuant", + "requirement", + "set", + "Rule", + "13a", + "14", + "Securities", + "Exchange", + "Act", + "1934", + "amended", + "Section", + "1350", + "Chapter", + "63", + "Title", + "18", + "United", + "States", + "Code", + "Jen", + "Hsun", + "Huang", + "President", + "Chief", + "Executive", + "Officer", + "NVIDIA", + "Corporation", + "Company", + "certifies", + "knowledge", + "The", + "Quarterly", + "Report", + "Form", + "10", + "period", + "ended", + "April", + "26", + "2026", + "Certification", + "attached", + "Exhibit", + "32.1", + "Periodic", + "fully", + "complies", + "requirements", + "13", + "15", + "information", + "contained", + "fairly", + "presents", + "material", + "respects", + "financial", + "condition", + "end", + "covered", + "results", + "operations", + "Date", + "May", + "20", + "JEN", + "HSUN", + "HUANG", + "signed", + "original", + "written", + "statement", + "required", + "906", + "provided", + "retained", + "furnished", + "Commission", + "staff", + "request", + "This", + "certification", + "accompanies", + "relates", + "deemed", + "filed", + "incorporated", + "reference", + "filing", + "1933", + "made", + "date", + "irrespective", + "general", + "incorporation", + "language" + ], + "keywords": [ + "CEO Certification", + "SEC Filing", + "Financial Report" + ], + "connect_to": [] + } + }, + { + "chunk_id": "92a83f27-56f7-537a-847c-4a2b4758e911", + "type": "text", + "content": "Pursuant to the requirement set forth in Rule 13a-14(b) of the Securities Exchange Act of 1934, as amended (the “Exchange Act”), and Section 1350 of Chapter 63 of Title 18 of the United States Code (18 U.S.C. § 1350), Colette M. Kress, the Executive Vice President and Chief Financial Officer of NVIDIA Corporation (the “Company”), hereby certifies that, to the best of her knowledge:\n1. The Company's Quarterly Report on Form 10-Q for the period ended April 26, 2026, to which this Certification is attached as Exhibit 32.2 (the \"Periodic Report\"), fully complies with the requirements of Section 13(a) or Section 15(d) of the Exchange Act; and\n2. The information contained in the Periodic Report fairly presents, in all material respects, the financial condition of the Company at the end of the period covered by the Periodic Report and results of operations of the Company for the period covered by the Periodic Report.\nDate: May 20, 2026\n/s/ COLETTE M. KRESS\nColette M. Kress\nExecutive Vice President and Chief Financial Officer\nA signed original of this written statement required by Section 906 of 18 U.S.C. § 1350 has been provided to NVIDIA Corporation and will be retained by NVIDIA Corporation and furnished to the Securities and Exchange Commission or its staff upon request.\nThis certification accompanies the Form 10-Q to which it relates, is not deemed filed with the Securities and Exchange Commission and is not to be incorporated by reference into any filing of the Company under the Securities Act of 1933, as amended, or the Exchange Act (whether made before or after the date of the Form 10-Q), irrespective of any general incorporation language contained in such filing.", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/CERTIFICATION_4", + "metadata": { + "length": 1692, + "summary": "Colette M. Kress, Executive Vice President and Chief Financial Officer of NVIDIA Corporation, certifies under Rule 13a-14(b) of the Securities Exchange Act and 18 U.S.C. § 1350 that the Quarterly Report on Form 10-Q for the period ended April 26, 2026, fully complies with Exchange Act requirements and fairly presents the company's financial condition and results of operations. The certification is dated May 20, 2026, and accompanies the Form 10-Q as Exhibit 32.2.", + "page_nums": [], + "document_top_summary": "This document includes: Where You Can Find More Information, Part I. Financial Information, Forward-Looking Statements, Item 3. Quantitative and Qualitative Disclosures About Market Risk, Item 4. Controls and Procedures, Part II. Other Information, Signature, CERTIFICATION, CERTIFICATION_2, CERTIFICATION_3, CERTIFICATION_4", + "tokens": [ + "Pursuant", + "requirement", + "set", + "Rule", + "13a", + "14", + "Securities", + "Exchange", + "Act", + "1934", + "amended", + "Section", + "1350", + "Chapter", + "63", + "Title", + "18", + "United", + "States", + "Code", + "Colette", + "Kress", + "Executive", + "Vice", + "President", + "Chief", + "Financial", + "Officer", + "NVIDIA", + "Corporation", + "Company", + "certifies", + "knowledge", + "The", + "Quarterly", + "Report", + "Form", + "10", + "period", + "ended", + "April", + "26", + "2026", + "Certification", + "attached", + "Exhibit", + "32.2", + "Periodic", + "fully", + "complies", + "requirements", + "13", + "15", + "information", + "contained", + "fairly", + "presents", + "material", + "respects", + "financial", + "condition", + "end", + "covered", + "results", + "operations", + "Date", + "May", + "20", + "COLETTE", + "KRESS", + "signed", + "original", + "written", + "statement", + "required", + "906", + "provided", + "retained", + "furnished", + "Commission", + "staff", + "request", + "This", + "certification", + "accompanies", + "relates", + "deemed", + "filed", + "incorporated", + "reference", + "filing", + "1933", + "made", + "date", + "irrespective", + "general", + "incorporation", + "language" + ], + "keywords": [ + "CFO Certification", + "Form 10-Q", + "SEC" + ], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/doc_nav.json b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/doc_nav.json new file mode 100644 index 000000000..db7e8993d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/doc_nav.json @@ -0,0 +1,804 @@ +{ + "version": "1.0", + "file_name": "NVIDIA Q1 FY27 Results.pdf", + "stats": { + "total_chunks": 120, + "text_chunks": 72, + "image_chunks": 6, + "table_chunks": 42, + "max_depth": 4 + }, + "sections": [ + { + "title": "Root", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf", + "level": 1, + "summary": "This is a Form 10-Q quarterly report filed by NVIDIA Corporation with the U.S. Securities and Exchange Commission for the quarterly period ended April 26, 2026. The document includes company identification details such as its Delaware incorporation, IRS Employer Identification Number, principal executive offices in Santa Clara, California, and its common stock trading under the symbol NVDA on the Nasdaq Global Select Market. It indicates NVIDIA is a large accelerated filer and not a shell company, with 24.2 billion shares outstanding as of May 15, 2026. The filing covers required financial statements and disclosures under the Securities Exchange Act of 1934.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Where You Can Find More Information", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Where You Can Find More Information", + "level": 1, + "summary": "Investors and others should note that we announce material financial information to our investors using our investor relations website, press releases, SEC filings and public conference calls and webcasts. We also use the following social media channels as a means of disclosing information about the company, our products, our planned financial and other announcements and attendance at upcoming investor and industry conferences, and other matters, and for complying with our disclosure obligations under Regulation FD: NVIDIA Corporate Blog (blogs.nvidia.com/) NVIDIA Technical Blog (developer.nvidia.com/blog/) NVIDIA LinkedIn (linkedin.com/company/nvidia) NVIDIA Facebook (facebook.com/nvidia) NVIDIA Instagram (instagram.com/nvidia) NVIDIA X (x.com/nvidia) NVIDIA Investor Relations (investor.nvidia.com) NVIDIA YouTube (YouTube.com/nvidia) The information we post through these social media channels may be deemed material. Accordingly, investors should monitor these channels, in addition to following our press releases, SEC filings and public conference calls and webcasts. This list may be updated from time to time. The information we post through these channels is not a part of this Quarterly Report on Form 10-Q.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Part I. Financial Information", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information", + "level": 1, + "summary": "", + "chunk_count": 33, + "children": [ + { + "title": "Item 1. Financial Statements (Unaudited)", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Item 1. Financial Statements (Unaudited)", + "level": 2, + "summary": "NVIDIA Corporation and Subsidiaries Condensed Consolidated Statements of Income (In millions, except per share data) (Unaudited) [tables/table-1 Financial Results.html] See accompanying Notes to Condensed Consolidated Financial Statements. NVIDIA Corporation and Subsidiaries Condensed Consolidated Statements of Comprehensive Income (In millions) (Unaudited) [tables/table-2 Comprehensive Income.html] See accompanying Notes to Condensed Consolidated Financial Statements. NVIDIA Corporation and Subsidiaries Condensed Consolidated Balance Sheets (In millions) (Unaudited) [tables/table-3 Balance Sheet.html] See accompanying Notes to Condensed Consolidated Financial Statements. NVIDIA Corporation and Subsidiaries Condensed Consolidated Statements of Shareholders' Equity (Unaudited) [tables/table-4 Shareholders' Equity.html] See accompanying Notes to Condensed Consolidated Financial Statements. NVIDIA Corporation and Subsidiaries Condensed Consolidated Statements of Cash Flows (In millions) (Unaudited) [tables/table-5 Cash Flow Statement.html] See accompanying Notes to Condensed Consolidated Financial Statements. NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Unaudited)", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 1 - Summary of Significant Accounting Policies", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies", + "level": 2, + "summary": "", + "chunk_count": 7, + "children": [ + { + "title": "Basis of Presentation", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Basis of Presentation", + "level": 3, + "summary": "The accompanying unaudited condensed consolidated financial statements were prepared in accordance with accounting principles generally accepted in the United States of America, or U.S. GAAP, for interim financial information and with the instructions to Form 10-Q and Article 10 of Securities and Exchange Commission, or SEC, Regulation S-X. The January 25, 2026 consolidated balance sheet was derived from our audited consolidated financial statements included in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026, as filed with the SEC, but does not include all disclosures required by U.S. GAAP. In the opinion of management, all adjustments, consisting only of normal recurring adjustments considered necessary for a fair presentation of results of operations and financial position, have been included. The results for the interim periods presented are not necessarily indicative of the results expected for any future period. The following information should be read in conjunction with the audited consolidated financial statements and notes thereto included in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. Certain prior fiscal year balances have been reclassified to conform to the current period presentation.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Significant Accounting Policies", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Significant Accounting Policies", + "level": 3, + "summary": "There have been no material changes to our significant accounting policies disclosed in Note 1 - Organization and Summary of Significant Accounting Policies, of the Notes to the Consolidated Financial Statements included in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026.", + "chunk_count": 4, + "children": [ + { + "title": "Fiscal Year", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Significant Accounting Policies/Fiscal Year", + "level": 4, + "summary": "Fiscal year 2027 is a 53-week year and fiscal year 2026 was a 52-week year, both ending on the last Sunday in January. The first quarters of fiscal years 2027 and 2026 were both 13-week quarters. The fourth quarter of fiscal year 2027 will be a 14-week quarter.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Principles of Consolidation", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Significant Accounting Policies/Principles of Consolidation", + "level": 4, + "summary": "Our condensed consolidated financial statements include the accounts of NVIDIA Corporation and our wholly owned subsidiaries. All intercompany balances and transactions have been eliminated in consolidation.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Use of Estimates", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Significant Accounting Policies/Use of Estimates", + "level": 4, + "summary": "The preparation of financial statements in conformity with U.S. GAAP requires management to make estimates and assumptions that affect the reported amounts of assets and liabilities and disclosures of contingent assets and liabilities at the date of the financial statements and the reported amounts of revenue and expenses during the reporting period. Actual results could differ materially from our estimates.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Recently Issued Accounting Pronouncements", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 1 - Summary of Significant Accounting Policies/Recently Issued Accounting Pronouncements", + "level": 3, + "summary": "Recent Accounting Pronouncements Not Yet Adopted In November 2024, the Financial Accounting Standards Board, or FASB, issued a new accounting standard requiring disclosures of certain additional expense information on an annual and interim basis, including, among other items, the amounts of purchases of inventory, employee compensation, depreciation and intangible asset amortization included within each income statement expense caption, as applicable. We will adopt this standard in the fiscal year 2028 annual report. We do not expect the adoption of this standard to have a material impact on our Consolidated Financial Statements other than additional disclosures.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 2 - Stock-Based Compensation", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 2 - Stock-Based Compensation", + "level": 2, + "summary": "We recognize stock-based compensation expense from grants of restricted stock units, or RSUs, performance stock units, or PSUs, and market-based PSUs, and issuances under our employee stock purchase plan, or ESPP. NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) Condensed Consolidated Statements of Income include stock-based compensation expense as follows: [tables/table-6 Cost Breakdown.html] Equity Award Activity The following is a summary of our equity award transactions under our equity incentive plans: RSUs, PSUs and Market-based PSUs Outstanding [tables/table-7 Stock Grant Activity.html] As of April 26, 2026, aggregate unearned stock-based compensation expense was \\$20.8 billion, which is expected to be recognized over a weighted average period of 2.6 years for RSUs, PSUs, and market-based PSUs, and one year for ESPP.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 3 - Net Income Per Share", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 3 - Net Income Per Share", + "level": 2, + "summary": "The following is the basic and diluted net income per share computations for the periods presented: [tables/table-8 EPS Data Table.html] (1) Net income divided by basic weighted average shares. (2) Net income divided by diluted weighted average shares. Diluted net income per share was computed using the weighted average number of common and potentially dilutive shares outstanding during the period, using the treasury stock method. NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 4 - Amortizable Intangible Assets and Goodwill", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 4 - Amortizable Intangible Assets and Goodwill", + "level": 2, + "summary": "The components of our amortizable intangible assets are as follows: [tables/table-9 Intangible Assets.html] Amortization expense associated with intangible assets was \\$232 million and \\$159 million for the first quarter of fiscal years 2027 and 2026, respectively. The following table outlines the estimated future amortization expense related to the net carrying amount of intangible assets as of April 26, 2026: [tables/table-10 Amortization Expense.html] In the first quarter of fiscal year 2027, goodwill increased by \\$62 million from acquisitions and was allocated to our Compute \\& Networking reporting unit. NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 5 - Cash Equivalents and Marketable Securities", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 5 - Cash Equivalents and Marketable Securities", + "level": 2, + "summary": "The passage discusses the fair value measurement of cash equivalents and marketable securities, including debt and equity securities, using Level 1 and Level 2 inputs. It provides summaries for April 26, 2026 and January 25, 2026, noting investments subject to lock-up restrictions (e.g., $27.4 billion and $10.5 billion, respectively) and long-term publicly-held equity securities with lock-ups through December 2027. Net unrealized gains of $13.4 billion for Q1 fiscal 2027 and net unrealized losses of $222 million for Q1 fiscal 2026 are reported, recognized in Other income (expense). Tables detail unrealized losses by investment category and maturity of debt securities, with gross unrealized losses primarily due to interest rate changes.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 6 - Non-marketable Securities", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 6 - Non-marketable Securities", + "level": 2, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "Non-marketable Equity Securities", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 6 - Non-marketable Securities/Non-marketable Equity Securities", + "level": 3, + "summary": "Our non-marketable equity securities are primarily in privately-held companies carried at cost less impairment, and adjusted for observable price changes. We value investments using observable comparable transactions and other inputs including volatility, expected time to liquidity, the risk-free rate, and security-specific rights and obligations. Adjustments to the carrying value of privately-held securities: [tables/table-15 Equity Securities.html] (1) Unrealized gains are recognized in Other income (expense), net, in the Condensed Consolidated Statements of Income. (2) Includes primarily reclassifications to marketable securities following public market trading. Non-marketable equity securities had cumulative gross unrealized gains of \\$5.3 billion and \\$396 million, and cumulative gross unrealized losses and impairments of \\$199 million and \\$110 million as of April 26, 2026 and April 27, 2025, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Equity Method Investments", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 6 - Non-marketable Securities/Equity Method Investments", + "level": 3, + "summary": "We have \\$1.0 billion of investments in infrastructure funds accounted for using the equity method as of April 26, 2026. Our maximum loss exposure under these investments, including invested and future committed amounts, was \\$2.3 billion as of April 26, 2026.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Investment Commitments", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 6 - Non-marketable Securities/Investment Commitments", + "level": 3, + "summary": "Total Investment commitments were \\$27 billion as of April 26, 2026, subject to certain contingencies, which we expect will be made through the remainder of fiscal year 2027.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 7 - Balance Sheet Components", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 7 - Balance Sheet Components", + "level": 2, + "summary": "We refer to customers who purchase products directly from NVIDIA as direct customers, such as add-in board manufacturers, or AIBs, distributors, original design manufacturers, or ODMs, original equipment manufacturers, or OEMs, cloud service providers, or CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. Three direct customers accounted for 30%, 18%, and 16% of our accounts receivable balance as of April 26, 2026. Three direct customers accounted for 25%, 18%, and 13% of our accounts receivable balance as of January 25, 2026. Certain balance sheet components were as follows: [tables/table-16 Inventory Data.html] (1) We recorded inventory provisions of \\$0.8 billion and \\$2.3 billion for the first quarter of fiscal years 2027 and 2026, respectively, in Cost of revenue. NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)", + "chunk_count": 3, + "children": [ + { + "title": "Property and Equipment:", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 7 - Balance Sheet Components/Property and Equipment:", + "level": 3, + "summary": "Property, equipment and intangible assets acquired but not paid for the first quarter of fiscal years 2027 and 2026 were \\$1.1 billion and \\$408 million, respectively. [tables/table-17 Accrued Liabilities.html] (1) Related to the Groq, Inc. non-exclusive license agreement. (2) We recorded \\$0.3 billion and \\$3.0 billion for the first quarter of fiscal years 2027 and 2026, respectively, in Cost of revenue. (3) Includes customer advances and unearned revenue related to hardware and software support, cloud services, and license and development arrangements. The balance as of April 26, 2026 and January 25, 2026 included \\$297 million and \\$160 million of customer advances, respectively. [tables/table-18 Long-Term Liabilities.html] (1) Primarily comprised of unrecognized tax benefits and related interest and penalties. (2) Includes unearned revenue related to hardware and software support and cloud services.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Deferred Revenue", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 7 - Balance Sheet Components/Deferred Revenue", + "level": 3, + "summary": "The following table shows the changes in short- and long-term deferred revenue during the first quarter of fiscal years 2027 and 2026: Three Months Ended [tables/table-19 Deferred Revenue.html] (1) Includes \\$1.7 billion and \\$6.2 billion of customer advances for the first quarter of fiscal years 2027 and 2026, respectively. (2) Includes \\$1.6 billion and \\$6.0 billion related to customer advances for the first quarter of fiscal years 2027 and 2026, respectively. We recognized revenue of \\$451 million and \\$265 million in the first quarter of fiscal years 2027 and 2026, respectively, that was included in the prior year-end deferred revenue balance. As of April 26, 2026, revenue related to remaining performance obligations from contracts greater than one year in length was \\$2.6 billion, which includes \\$2.3 billion from deferred revenue and \\$304 million, which has not yet been billed or NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) recognized as revenue. Approximately 40% of revenue from contracts greater than one year in length will be recognized over the next twelve months.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 8 - Derivative Financial Instruments", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 8 - Derivative Financial Instruments", + "level": 2, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Foreign Currency Derivatives", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 8 - Derivative Financial Instruments/Foreign Currency Derivatives", + "level": 3, + "summary": "The company uses foreign currency forward contracts to hedge against exchange rate fluctuations affecting operating expenses and monetary assets/liabilities. Contracts for operating expenses are designated as accounting hedges, with gains/losses recorded in Accumulated other comprehensive income and reclassified to Operating expenses when recognized. These were highly effective and not significant in Q1 FY2027 and FY2026. Non-designated contracts for assets/liabilities have fair value changes recorded in Other income/expense, offsetting hedged items. Notional values are presented in a table; fair values were not significant as of April 26, 2026 and January 25, 2026. All contracts mature within 18 months, with deferred gains/losses expected to be realized in the next twelve months not significant.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Facility Lease Guarantee", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 8 - Derivative Financial Instruments/Facility Lease Guarantee", + "level": 3, + "summary": "In fiscal year 2026, NVIDIA entered into agreements guaranteeing partners' facility lease obligations with a maximum gross exposure of $3.5 billion, reduced by partner payments over 5-7 years, and $712 million in escrow. The guarantees are classified as credit derivatives with immaterial fair value changes. As of April 26, 2026, the estimated fair value of debt was $7.4 billion (Level 2 inputs). Notes are unsecured senior obligations with semi-annual interest and make-whole redemption. The company complied with non-financial covenants and had a $25.0 billion commercial paper program with no outstanding amounts.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Note 10 - Commitments and Contingencies", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies", + "level": 2, + "summary": "", + "chunk_count": 6, + "children": [ + { + "title": "Commitments", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies/Commitments", + "level": 3, + "summary": "Manufacturing, supply, and capacity commitments reflect data center-scale production and longer future ordering horizons across current and future product architectures. We enter into agreements with our supply vendors that allow them to procure inventory based upon our defined criteria, and in certain instances, these agreements are cancellable, able to be rescheduled, or adjustable for our business needs prior to placing firm orders. Changes to these agreements may result in additional costs. As of April 26, 2026, these commitments were \\$119 billion for which \\$95 billion will be paid in the remainder of fiscal year 2027 and the remaining balance will be paid in fiscal years 2028 through 2031. Multi-year cloud service agreement commitments as of April 26, 2026, were \\$30 billion for which \\$6 billion, \\$7 billion, \\$7 billion, \\$5 billion, \\$3 billion, and \\$2 billion will be paid in the remainder of fiscal year 2027, each fiscal year from 2028 through 2031, and fiscal year 2032 and thereafter, respectively. Cloud service capacity may be reduced or terminated. Cloud service agreements will be primarily used to support our research and development efforts. Other vendor commitments were \\$6 billion as of April 26, 2026, of which the majority will be paid through fiscal year 2027. NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited)", + "chunk_count": 1, + "children": [] + }, + { + "title": "Accrual for Product Warranty Liabilities", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies/Accrual for Product Warranty Liabilities", + "level": 3, + "summary": "The estimated amount of product warranty liabilities was \\$2.9 billion and \\$2.8 billion as of April 26, 2026 and January 25, 2026, respectively. The estimated product returns and product warranty activity consisted of the following: [tables/table-22 Balance Changes.html] For the first quarter of fiscal years 2027 and 2026, the additions in product warranty liabilities primarily related to our Compute & Networking segment. We have provided indemnities for matters such as tax, product, and employee liabilities. We have included intellectual property indemnification provisions in our technology-related agreements with third parties. Maximum potential future payments cannot be estimated because many of these agreements do not have a maximum stated liability. We have not recorded any liability in our Condensed Consolidated Financial Statements for such indemnifications.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Litigation", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies/Litigation", + "level": 3, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Securities Class Action and Derivative Lawsuits", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies/Litigation/Securities Class Action and Derivative Lawsuits", + "level": 4, + "summary": "The text details multiple legal actions against NVIDIA Corporation. A securities class action (4:18-cv-07669-HSG) was filed in 2018, alleging false statements about channel inventory and cryptocurrency mining's impact on GPU demand. After dismissal and appeals, the Supreme Court dismissed NVIDIA's certiorari petition in 2024, leading to remand and class certification in 2026. Related derivative lawsuits in California and Delaware, asserting breach of fiduciary duty and other claims, remain stayed pending final resolution of the securities action.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Accounting for Loss Contingencies", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 10 - Commitments and Contingencies/Litigation/Accounting for Loss Contingencies", + "level": 4, + "summary": "As of April 26, 2026, there are no accrued contingent liabilities associated with the legal proceedings described above based on our belief that liabilities, while reasonably possible, are not probable. Further, any possible loss or range of loss in these matters cannot be reasonably estimated at this time. We are engaged in legal actions not described above arising in the ordinary course of business, as well as regulatory and government inquiries and investigations, and, while there can be no assurance of favorable outcomes, we believe that the ultimate outcome of these matters will not have a material adverse effect on our operating results, liquidity or financial position. These matters are subject to inherent uncertainties and if the ultimate outcome is unfavorable, there exists the possibility of a material adverse impact on our operating results, liquidity or financial position in the period the outcome becomes estimable and probable.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Note 11 - Income Taxes", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 11 - Income Taxes", + "level": 2, + "summary": "Income tax expense was \\$11.6 billion and \\$3.1 billion for the first quarter of fiscal years 2027 and 2026, respectively. Income tax as a percentage of income before income tax was 16.6% and 14.3% for the first quarter of fiscal years 2027 and 2026, respectively. The effective tax rate increased primarily due to a lower percentage of tax benefits from stock-based compensation relative to the increase in income before income tax. Our effective tax rates for the first quarter of fiscal years 2027 and 2026 were lower than the U.S. federal statutory rate of 21% primarily due to tax benefits from foreign-derived deduction eligible income, income earned in jurisdictions that were subject to taxes at rates lower than the U.S. federal statutory tax rate, stock-based compensation, and the U.S. federal research tax credit. While we believe that we have adequately provided for all uncertain tax positions, or tax positions where we believe it is not more-likely-than-not that the position will be sustained upon review, amounts asserted by tax authorities could be greater or less than our accrued position. Accordingly, our provisions on federal, state and foreign tax related matters to be recorded in the future may change as revised estimates are made or the underlying matters are settled or otherwise resolved with the respective tax authorities. We are currently under examination by the Internal Revenue Service for our fiscal years 2023 and 2024.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 12 - Shareholders' Equity", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 12 - Shareholders' Equity", + "level": 2, + "summary": "Capital Return Program We repurchased 108 million and 126 million shares of our common stock for \\$20.2 billion and \\$14.5 billion during the first quarter of fiscal years 2027 and 2026, respectively. As of April 26, 2026, we were authorized, subject to certain specifications, to repurchase up to \\$38.5 billion of our common stock. On May 18, 2026, our Board of Directors approved an additional \\$80.0 billion in share repurchase authorization, without expiration. NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) We paid cash dividends to our shareholders of \\$243 million and \\$244 million during the first quarter of fiscal years 2027 and 2026, respectively. On May 18, 2026, we increased our quarterly cash dividend from \\$0.01 per share to \\$0.25 per share to all shareholders of record on June 4, 2026. Our quarterly cash dividend will be paid on June 26, 2026. The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 13 - Segment Information", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 13 - Segment Information", + "level": 2, + "summary": "NVIDIA's CODM reviews financial information on an operating segment basis, with two reportable segments: Compute & Networking (Data Center, AI, automotive) and Graphics (GeForce, Quadro). Segment operating income is assessed, but certain expenses like stock-based compensation are not allocated. Depreciation for Compute & Networking was $526M (FY2027 Q1) vs $296M (FY2026 Q1); Graphics was $194M vs $109M. Revenue concentration exists: three direct customers accounted for 21%, 17%, and 16% of total revenue in FY2027 Q1, primarily in Compute & Networking. Non-US revenue was 22% in FY2027 Q1 vs 42% in FY2026 Q1. The presentation of revenue by market platform changed in FY2027 Q1.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Note 14 - Leases", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part I. Financial Information/Note 14 - Leases", + "level": 2, + "summary": "Our lease obligations primarily consist of operating leases for our data centers and offices, with lease periods expiring between fiscal years 2027 and 2075. Future minimum lease obligations under our non-cancelable lease agreements as of April 26, 2026 were as follows: [tables/table-27 Lease Obligations.html] Between the second quarter of fiscal year 2027 and fiscal year 2033, we expect to commence leases with future obligations of \\$32.4 billion, primarily for data center leases to support our research and development efforts, with lease terms of 3 to 20 years. Operating lease costs were \\$171 million and \\$101 million for the first quarter of fiscal years 2027 and 2026, respectively. Short-term, variable, and finance lease costs for the first quarter of fiscal years 2027 and 2026 were not significant. Other information related to leases was as follows: [tables/table-28 Lease Cash Flows.html] NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) As of April 26, 2026, our operating leases have a weighted average remaining lease term of 10.4 years and a weighted average discount rate of 4.61%. As of January 25, 2026, our operating leases had a weighted average remaining lease term of 8.8 years and a weighted average discount rate of 4.38%. Item 2. Management's Discussion and Analysis of Financial Condition and Results of Operations", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Forward-Looking Statements", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements", + "level": 1, + "summary": "This passage contains forward-looking statements under the Securities Act and Exchange Act, based on management's beliefs and assumptions. It identifies terms like 'may,' 'will,' 'expect,' and 'anticipate' as indicators of such statements. Actual results may differ materially due to known and unknown risks, uncertainties, and factors discussed in the company's Annual Report on Form 10-K and this Quarterly Report. The company assumes no obligation to update these statements. References to 'NVIDIA' or 'we' mean NVIDIA Corporation and its subsidiaries. The text also cautions against undue reliance on beliefs and opinions expressed, which are based on information available as of the filing date.", + "chunk_count": 18, + "children": [ + { + "title": "Overview", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Overview", + "level": 2, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "Our Company and Our Businesses", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Overview/Our Company and Our Businesses", + "level": 3, + "summary": "NVIDIA pioneered accelerated computing to help solve the most challenging computational problems. Since our original focus on PC graphics, we have expanded to several other large and important computationally intensive fields. Fueled by the sustained demand for exceptional 3D graphics and the scale of the gaming market, NVIDIA has leveraged its GPU architecture to create platforms for scientific computing, AI, data science, autonomous vehicles, robotics, and digital twin applications. NVIDIA is now a data center-scale AI infrastructure company reshaping all industries. Our two operating segments are “Compute & Networking” and “Graphics.” Refer to Note 13 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q for additional information. Headquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recent Developments, Future Objectives and Challenges", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Overview/Recent Developments, Future Objectives and Challenges", + "level": 3, + "summary": "NVIDIA's first quarter of fiscal year 2027 saw revenue of $81.6 billion, up 85% year-over-year, driven by Data Center products for accelerated computing and AI, particularly Blackwell architecture. Data Center revenue was $75.2 billion, with hyperscalers accounting for about 50%. Edge Computing revenue was $6.4 billion. The company faces risks including supply chain constraints, geopolitical issues, tariffs, and competition from open-source AI. Investments totaled $18.6 billion in private companies and infrastructure. No H200 revenue from China has been generated yet, and a new reporting framework with Data Center and Edge Computing platforms is being adopted.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Financial Information by Business Segment and Geographic Data", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Overview/Financial Information by Business Segment and Geographic Data", + "level": 3, + "summary": "Refer to Note 13 of the Notes to the Condensed Consolidated Financial Statements for disclosure regarding segment information.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Critical Accounting Policies and Estimates", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Critical Accounting Policies and Estimates", + "level": 2, + "summary": "Refer to Part II, Item 7, \"Critical Accounting Policies and Estimates\" of our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. There have been no material changes to our Critical Accounting Policies and Estimates.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Results of Operations", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations", + "level": 2, + "summary": "The following table sets forth, for the periods indicated, certain items in our Condensed Consolidated Statements of Income expressed as a percentage of revenue. [tables/table-31 Revenue Analysis.html]", + "chunk_count": 11, + "children": [ + { + "title": "Reportable Segments", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Reportable Segments", + "level": 3, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "Revenue by Reportable Segments", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Reportable Segments/Revenue by Reportable Segments", + "level": 4, + "summary": "[tables/table-32 Revenue Growth.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Operating Income by Reportable Segments", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Reportable Segments/Operating Income by Reportable Segments", + "level": 4, + "summary": "[tables/table-33 Revenue Growth.html] Compute & Networking revenue – The year-over-year increase in the first quarter of fiscal year 2027 was due to growth in Data Center products, driven by the ramp of our Blackwell systems and demand for our InfiniBand, Spectrum-X Ethernet, and NVLink solutions. Graphics revenue – The year-over-year increase in the first quarter of fiscal year 2027 was driven by sales of our Blackwell architecture. Reportable segment operating income – The year-over-year increase in Compute & Networking segment operating income in the first quarter of fiscal year 2027 was driven by the growth in revenue and the non-recurrence of a \\$4.5 billion charge associated with H2O excess inventory and purchase obligations in the first quarter of fiscal year 2026. The year-over-year increase in Graphics segment operating income in the first quarter of fiscal year 2027 was driven by the growth in revenue.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Concentration of Revenue", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Reportable Segments/Concentration of Revenue", + "level": 4, + "summary": "NVIDIA's revenue is concentrated among a limited number of direct and indirect customers. Direct customers include AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. In Q1 FY2027, three direct customers accounted for 21%, 17%, and 16% of total revenue, primarily from the Compute & Networking segment. Indirect customers purchase through system integrators and distributors, with some individually representing 10% or more of revenue. One AI research and deployment company contributed meaningfully to Q1 FY2027 revenue via cloud services. Revenue from customers headquartered outside the US was 22% in Q1 FY2027 and 42% in Q1 FY2026.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Gross Profit and Gross Margin", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Gross Profit and Gross Margin", + "level": 3, + "summary": "The text discusses gross profit, cost of revenue, and gross margin trends for the first quarter of fiscal years 2027 and 2026. Gross margin increased to 74.9% in Q1 FY2027 from 60.5% in Q1 FY2026, mainly due to a prior year $4.5 billion charge for H2O excess inventory. Provisions for inventory and excess purchase obligations were $1.1 billion in FY2027 and $5.3 billion in FY2026. Sales of previously reserved inventory led to provision releases of $103 million and $436 million, respectively. The net unfavorable impact on gross margin was 1.2% in FY2027 and 11.0% in FY2026. Operating expenses increased due to higher compute and infrastructure costs, compensation, and engineering materials. Other income included unrealized gains of $13.4 billion from publicly-held equity securities and $2.6 billion from non-marketable securities.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Income Taxes", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Income Taxes", + "level": 3, + "summary": "Income tax expense was $11.6 billion and $3.1 billion for Q1 of fiscal years 2027 and 2026, with effective tax rates of 16.6% and 14.3%, respectively. The rate increase was due to lower tax benefits from stock-based compensation relative to higher income. Both rates were below the 21% U.S. statutory rate due to foreign-derived deductions, lower foreign taxes, stock-based compensation, and research credits. Cash from operations increased due to higher revenue, while investing and financing cash uses rose from higher equity purchases and share repurchases.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Liquidity", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Liquidity", + "level": 3, + "summary": "The passage discusses the company's primary liquidity sources, including cash, cash equivalents, and marketable debt and equity securities. As of April 26, 2026, it held $50.3 billion in cash, cash equivalents, and marketable debt securities, plus $30.2 billion in marketable equity securities. The company believes it has sufficient liquidity for at least the next twelve months. Marketable securities include U.S. government and corporate debt, equity securities, and certificates of deposit, mostly in U.S. dollars. About $1.7 billion held outside the U.S. has not accrued taxes for repatriation, but substantially all foreign-held cash is available for U.S. use without additional federal income tax. No federal income tax payments were made in the first quarter of fiscal year 2027, with two payments scheduled for the second quarter.", + "chunk_count": 4, + "children": [ + { + "title": "Capital Return to Shareholders", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Liquidity/Capital Return to Shareholders", + "level": 4, + "summary": "In Q1 FY2027, the company repurchased 108 million shares for $20.2 billion, with $38.5 billion remaining authorized as of April 26, 2026. On May 18, 2026, the Board approved an additional $80 billion in repurchase authorization without expiration. Repurchases may be executed via open market, private negotiations, Rule 10b5-1 plans, or structured agreements, and can be suspended anytime. Cash dividends of $243 million were paid in Q1 FY2027. On May 18, 2026, the quarterly dividend was increased from $0.01 to $0.25 per share, payable on June 26, 2026 to shareholders of record on June 4, 2026. Future dividends depend on Board discretion. A 1% excise tax on certain repurchases under the Inflation Reduction Act applies but was not significant for the quarter.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Outstanding Indebtedness and Commercial Paper Program", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Liquidity/Outstanding Indebtedness and Commercial Paper Program", + "level": 4, + "summary": "Our aggregate debt maturities as of April 26, 2026, by year payable, were as follows: [tables/table-38 Debt Maturity.html] We have a commercial paper program to support general corporate purposes, pursuant to which we may issue unsecured paper notes, from time to time or all at once, up to \\$25.0 billion. As of April 26, 2026, no commercial paper was outstanding. Refer to Note 9 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q for further discussion.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Material Cash Requirements and Other Obligations", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Results of Operations/Liquidity/Material Cash Requirements and Other Obligations", + "level": 4, + "summary": "The passage discusses various financial obligations and commitments, including facility lease guarantees, long-term debt, purchase commitments, and operating lease obligations referenced in specific notes of the financial statements. It mentions unrecognized tax benefits of $4.5 billion including interest and penalties, and ongoing IRS examinations for fiscal years 2023 and 2024. It notes no material changes in contractual obligations outside ordinary course of business from the prior annual report.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Adoption of New and Recently Issued Accounting Pronouncements", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Forward-Looking Statements/Adoption of New and Recently Issued Accounting Pronouncements", + "level": 2, + "summary": "There has been no adoption of any new and recently issued accounting pronouncements.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Item 3. Quantitative and Qualitative Disclosures About Market Risk", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 3. Quantitative and Qualitative Disclosures About Market Risk", + "level": 1, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "Investment and Interest Rate Risk", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 3. Quantitative and Qualitative Disclosures About Market Risk/Investment and Interest Rate Risk", + "level": 2, + "summary": "Financial market risks related to investment and interest rate risk are described in Part II, Item 7A, “Quantitative and Qualitative Disclosures About Market Risk” in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. Our marketable equity securities consist of publicly-held equity securities, while our non-marketable equity securities are investments in privately-held companies. Publicly-held equity securities are subject to market price volatility. A hypothetical 10% decrease in our publicly-held equity securities would decrease the fair value of the publicly-held equity securities balance by \\$3.9 billion and \\$1.8 billion as of April 26, 2026 and January 25, 2026, respectively. Non-marketable equity securities are measured based on cost minus impairment, if any, and are adjusted for observable price changes in orderly transactions for an identical or similar investment in the same issuer. Valuations of our non- marketable equity securities are inherently complex due to the lack of readily available market data and observable transactions, and impact of macroeconomic factors. For a description of our equity investments, refer to Notes 5 and 6 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q, respectively.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Foreign Exchange Rate Risk", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 3. Quantitative and Qualitative Disclosures About Market Risk/Foreign Exchange Rate Risk", + "level": 2, + "summary": "The impact of foreign currency transactions related to foreign exchange rate risk is described in Part II, Item 7A, “Quantitative and Qualitative Disclosures About Market Risk” in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. As of April 26, 2026, there have been no material changes to the foreign exchange rate risks described as of January 25, 2026.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Item 4. Controls and Procedures", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 4. Controls and Procedures", + "level": 1, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "Controls and Procedures", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 4. Controls and Procedures/Controls and Procedures", + "level": 2, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "Disclosure Controls and Procedures", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 4. Controls and Procedures/Controls and Procedures/Disclosure Controls and Procedures", + "level": 3, + "summary": "Based on their evaluation as of April 26, 2026, our management, including our Chief Executive Officer and Chief Financial Officer, has concluded that our disclosure controls and procedures (as defined in Rule 13a-15(e) under the Exchange Act) were effective to provide reasonable assurance that the information we are required to disclose in reports that we file or submit under the Exchange Act is recorded, processed, summarized and reported within the time periods specified in the SEC rules and forms, and that such information is accumulated and communicated to our management, including our Chief Executive Officer and our Chief Financial Officer, as appropriate, to allow timely decisions regarding required disclosures.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Changes in Internal Control Over Financial Reporting", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 4. Controls and Procedures/Controls and Procedures/Changes in Internal Control Over Financial Reporting", + "level": 3, + "summary": "There have been no changes in our internal control over financial reporting during the first quarter of fiscal year 2027 that have materially affected, or are reasonably likely to materially affect, our internal control over financial reporting. We are continuing a phased upgrade of our enterprise resource planning, or ERP, system to update our existing core financial systems. The ERP system is designed to accurately maintain our financial records used to report operating results. We will continue to evaluate each quarter whether there are changes that materially affect our internal control over financial reporting.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Inherent Limitations on Effectiveness of Controls", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Item 4. Controls and Procedures/Controls and Procedures/Inherent Limitations on Effectiveness of Controls", + "level": 3, + "summary": "Our management, including our Chief Executive Officer and Chief Financial Officer, does not expect that our disclosure controls and procedures, or our internal controls, will prevent all error and all fraud. A control system, no matter how well conceived and operated, can provide only reasonable, not absolute, assurance that the objectives of the control system are met. Further, the design of a control system must reflect the fact that there are resource constraints, and the benefits of controls must be considered relative to their costs. Because of the inherent limitations in all control systems, no evaluation of controls can provide absolute assurance that all control issues and instances of fraud, if any, within NVIDIA have been detected.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Part II. Other Information", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information", + "level": 1, + "summary": "", + "chunk_count": 6, + "children": [ + { + "title": "Item 1. Legal Proceedings", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information/Item 1. Legal Proceedings", + "level": 2, + "summary": "Refer to Part I, Item 1, Note 10 of the Notes to the Condensed Consolidated Financial Statements for a discussion of significant developments in our legal proceedings since January 25, 2026.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Item 1A. Risk Factors", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information/Item 1A. Risk Factors", + "level": 2, + "summary": "The passage details risk factors for NVIDIA common stock, focusing on competition, supply-demand mismatches, and complex export controls. Competition could harm market share and financial results. Long manufacturing lead times and inaccurate demand estimates may cause supply-demand mismatches, leading to inventory issues or lost revenue. Export controls, particularly targeting AI and GPU products to China and other countries, have significantly impacted NVIDIA's business, causing charges, limiting market access, and benefiting competitors. The text also mentions risks from new product introductions, regulatory changes, and geopolitical tensions.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Issuer Purchases of Equity Securities", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information/Issuer Purchases of Equity Securities", + "level": 2, + "summary": "In the first quarter of fiscal 2027, the company repurchased 108 million shares for $20.2 billion. As of April 26, 2026, authorization remained for up to $38.5 billion in repurchases. On May 18, 2026, the Board approved an additional $80.0 billion in repurchase authorization. Cash dividends of $243 million were paid in the first quarter, and the quarterly dividend was increased from $0.01 to $0.25 per share, payable on June 26, 2026. Additionally, 12 million shares were withheld for tax obligations under the employee equity incentive program, valued at $2.1 billion.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Recent Sales of Unregistered Securities and Use of Proceeds", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information/Recent Sales of Unregistered Securities and Use of Proceeds", + "level": 2, + "summary": "On February 17, 2026, we acquired a company and issued to a key employee a total of 37,890 shares of our common stock, valued at approximately \\$7 million based on our closing stock price on the issuance date. On April 10, 2026, we acquired a company and issued to key employees a total of 72,972 shares of our common stock, valued at approximately \\$14 million based on our closing stock price on the issuance date. The above securities were issued in transactions not involving a public offering pursuant to an exemption from registration set forth in Section 4(a)(2) of the Securities Act (and Regulation D or Regulation S promulgated thereunder).", + "chunk_count": 1, + "children": [] + }, + { + "title": "Item 5. Other Information", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Part II. Other Information/Item 5. Other Information", + "level": 2, + "summary": "The passage reports that members of the Board of Directors and officers adopted, modified, or terminated Rule 10b5-1 trading arrangements. One arrangement is solely for gifts to charitable donor-advised funds. Another was adopted on December 18, 2025, for sales through March 23, 2027, with no shares sold prior to termination. The document also lists exhibits for the quarterly report, including management contracts and certifications, and provides contact information for shareholder requests.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Signature", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/Signature", + "level": 1, + "summary": "Pursuant to the requirements of the Securities Exchange Act of 1934, the registrant has duly caused this report to be signed on its behalf by the undersigned thereunto duly authorized. Date: May 20, 2026 NVIDIA Corporation By: /s/ Colette M. Kress Colette M. Kress Executive Vice President and Chief Financial Officer (Duly Authorized Officer and Principal Financial Officer)", + "chunk_count": 1, + "children": [] + }, + { + "title": "CERTIFICATION", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/CERTIFICATION", + "level": 1, + "summary": "Jen-Hsun Huang, President and CEO of NVIDIA Corporation, certifies the Quarterly Report on Form 10-Q. He confirms the report contains no material misstatements, the financial statements fairly present the company's condition and results, and he is responsible for establishing and maintaining disclosure controls and internal control over financial reporting. He also discloses that he has evaluated these controls and reported any significant deficiencies or fraud to auditors and the audit committee. The certification is dated May 20, 2026.", + "chunk_count": 1, + "children": [] + }, + { + "title": "CERTIFICATION_2", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/CERTIFICATION_2", + "level": 1, + "summary": "Colette M. Kress, Executive Vice President and Chief Financial Officer of NVIDIA Corporation, certifies the accuracy and completeness of the Quarterly Report on Form 10-Q. The certification covers the review of the report, the fairness of financial statements, the design and evaluation of disclosure controls and procedures, and the disclosure of any changes in internal control over financial reporting. It also includes reporting of significant deficiencies, material weaknesses, and fraud to auditors and the audit committee. The certification is dated May 20, 2026.", + "chunk_count": 1, + "children": [] + }, + { + "title": "CERTIFICATION_3", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/CERTIFICATION_3", + "level": 1, + "summary": "Jen-Hsun Huang, President and CEO of NVIDIA Corporation, certifies under Rule 13a-14(b) and 18 U.S.C. § 1350 that the Quarterly Report on Form 10-Q for the period ended April 26, 2026, complies with Exchange Act requirements and fairly presents the company's financial condition and results of operations. The certification is dated May 20, 2026, and is not deemed filed with the SEC.", + "chunk_count": 1, + "children": [] + }, + { + "title": "CERTIFICATION_4", + "path": "927dc2d6-a76c-4006-9f34-8769b2c665fb.pdf/CERTIFICATION_4", + "level": 1, + "summary": "Colette M. Kress, Executive Vice President and Chief Financial Officer of NVIDIA Corporation, certifies under Rule 13a-14(b) of the Securities Exchange Act and 18 U.S.C. § 1350 that the Quarterly Report on Form 10-Q for the period ended April 26, 2026, fully complies with Exchange Act requirements and fairly presents the company's financial condition and results of operations. The certification is dated May 20, 2026, and accompanies the Form 10-Q as Exhibit 32.2.", + "chunk_count": 1, + "children": [] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-☒ QUARTERL.jpg", + "summary": "image-1 ☒ QUARTERLY REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 NVIDIA.0" + }, + { + "path": "images/image-2-The Nasdaq.jpg", + "summary": "image-2 The Nasdaq Global Select Market Accelerated filer0" + }, + { + "path": "images/image-3-Indicate b.jpg", + "summary": "image-3 Indicate by check mark whether the registrant has submitted electronically every Interactive Data File required to be submitted pursuant to Rule 405 of Regulation S-T ( $§232.405$ of this chapter) during the preceding 12 months (or for such shorter period that the registrant was required to submit such files). Yes ☑ No ☐ Non-accelerated filer0" + }, + { + "path": "images/image-4-Large acce.jpg", + "summary": "image-4 Large accelerated filer Smaller reporting company0" + }, + { + "path": "images/image-5-Accelerate.jpg", + "summary": "image-5 Accelerated filer Emerging growth company0" + }, + { + "path": "images/image-6-Non-accele.jpg", + "summary": "image-6 Non-accelerated filer If an emerging growth company, indicate by check mark if the registrant has elected not to use the extended transition period for complying with any new or revised financial accounting standards provided pursuant to Section 13(a) of the Exchange Act. □0" + } + ], + "tables": [ + { + "path": "tables/table-0 Financial Report.html", + "summary": "table-1 Table of contents for a quarterly financial report (Form 10-Q) covering financial statements, management discussion, market risk, controls, legal proceedings, and exhibits." + }, + { + "path": "tables/table-1 Financial Results.html", + "summary": "table-2 The table shows financial data for three months ending Apr 26, 2026 and Apr 27, 2025. Revenue increased from $44,062 to $81,615, net income rose from $18,775 to $58,321, and operating income grew from $21,638 to $53,536." + }, + { + "path": "tables/table-2 Comprehensive Income.html", + "summary": "table-3 The table shows net income of $58,321 for Apr 26, 2026 vs $18,775 for Apr 27, 2025. Other comprehensive loss was $41 vs gain of $158, resulting in total comprehensive income of $58,280 vs $18,933." + }, + { + "path": "tables/table-3 Balance Sheet.html", + "summary": "table-4 The balance sheet shows total assets of $259,474 as of Apr 26, 2026, up from $206,803. Liabilities are $64,000 and shareholders' equity is $195,474." + }, + { + "path": "tables/table-4 Shareholders' Equity.html", + "summary": "table-5 The table shows changes in shareholders' equity from Jan 2025 to Apr 2026, including net income, stock issuance, repurchases, dividends, and other adjustments." + }, + { + "path": "tables/table-5 Cash Flow Statement.html", + "summary": "table-6 The table shows cash flow data for three months ended Apr 26, 2026 and Apr 27, 2025. Net cash from operations was $50,344 and $27,414; investing used $26,429 and $5,216; financing used $21,283 and $15,553. Cash ended at $13,237 and $15,234." + }, + { + "path": "tables/table-6 Cost Breakdown.html", + "summary": "table-7 The table shows costs for three months ending Apr 26, 2026 and Apr 27, 2025. Cost of revenue: $68M vs $64M; R&D: $1,459M vs $1,063M; SG&A: $401M vs $347M; Total: $1,928M vs $1,474M." + }, + { + "path": "tables/table-7 Stock Grant Activity.html", + "summary": "table-8 The table shows restricted stock activity from Jan to Apr 2026: shares increased from 189 to 200 million, with 44 million granted at $181.73, 31 million vested at $48.04, and 2 million forfeited." + }, + { + "path": "tables/table-8 EPS Data Table.html", + "summary": "table-9 The table shows net income of $58,321M (Apr 2026) vs $18,775M (Apr 2025), basic EPS $2.40 vs $0.77, diluted EPS $2.39 vs $0.76, with weighted average shares and anti-dilutive awards." + }, + { + "path": "tables/table-9 Intangible Assets.html", + "summary": "table-10 The table shows intangible assets as of Apr 26, 2026 and Jan 25, 2026, with gross carrying amounts, accumulated amortization, and net carrying amounts for acquisition-related assets, patents, and total." + }, + { + "path": "tables/table-10 Amortization Expense.html", + "summary": "table-11 The table shows future amortization expenses from 2027 to 2032 and beyond, totaling $3,120 million, with the highest in 2028 at $754 million." + }, + { + "path": "tables/table-11 Debt Securities.html", + "summary": "table-12 The table details debt and equity securities by pricing category, showing cost, unrealized gains/losses, estimated fair value, and how they are reported as cash equivalents, marketable debt, or equity securities." + }, + { + "path": "tables/table-12 Fair Value Table.html", + "summary": "table-13 The table details fair value measurements of debt and equity securities by pricing level, showing cost, unrealized gains/losses, and estimated fair value, with breakdowns by type and reporting category." + }, + { + "path": "tables/table-13 Debt Securities.html", + "summary": "table-14 The table shows estimated fair values and gross unrealized losses for debt securities (U.S. Treasury, corporate, and government agencies) as of Apr 26, 2026 and Jan 25, 2026, with totals of $16,187M and $13,132M respectively." + }, + { + "path": "tables/table-14 Debt Maturity.html", + "summary": "table-15 The table shows debt maturities: $24,307 million due within one year, $14,926 million due in 1-5 years, totaling $39,233 million as of Apr 26, 2026." + }, + { + "path": "tables/table-15 Equity Securities.html", + "summary": "table-16 The table shows changes in non-marketable equity securities over three months ending Apr 26, 2026, with balance increasing from $22,251M to $42,336M, including net additions of $17,899M and unrealized gains of $2,603M." + }, + { + "path": "tables/table-16 Inventory Data.html", + "summary": "table-17 The table shows inventories in millions for Apr 26, 2026 and Jan 25, 2026: raw materials $6,647 vs $3,807, work in process $9,949 vs $8,822, finished goods $9,201 vs $8,774, total $25,797 vs $21,403." + }, + { + "path": "tables/table-17 Accrued Liabilities.html", + "summary": "table-18 The table shows accrued and other current liabilities as of Apr 26, 2026 ($29,787M) and Jan 25, 2026 ($21,352M), including taxes payable, customer program accruals, and warranty costs." + }, + { + "path": "tables/table-18 Long-Term Liabilities.html", + "summary": "table-19 The table shows other long-term liabilities totaling $8,768 million as of Apr 26, 2026, down from $7,306 million on Jan 25, 2026, including income tax payable, deferred income tax, deferred revenue, and other items." + }, + { + "path": "tables/table-19 Deferred Revenue.html", + "summary": "table-20 The table shows deferred revenue balances for Apr 26, 2026 ($3,117M) and Apr 27, 2025 ($2,078M), with additions and revenue recognized during each period." + }, + { + "path": "tables/table-20 Hedge Designation.html", + "summary": "table-21 The table shows designated accounting hedges at $2,114M (Apr 26, 2026) and $1,765M (Jan 25, 2026), and non-designated hedges at $1,850M and $2,332M respectively." + }, + { + "path": "tables/table-21 Debt Notes Table.html", + "summary": "table-22 The table lists various notes due from 2026 to 2060 with interest rates from 1.55% to 3.70%, remaining terms, and carrying amounts. Net carrying amount is $8,470 million as of Apr 26, 2026, with long-term portion of $7,470 million." + }, + { + "path": "tables/table-22 Balance Changes.html", + "summary": "table-23 The table shows balance changes over three months ending Apr 26, 2026 and Apr 27, 2025, with beginning balances of $2,807M and $1,290M, additions of $330M and $870M, utilization of $189M and $80M, and ending balances of $2,948M and $2,080M." + }, + { + "path": "tables/table-23 Segment Revenue.html", + "summary": "table-24 The table shows Compute & Networking and Graphics segment revenue, other items, and operating income for three-month periods ending April 2026 and April 2025, with totals." + }, + { + "path": "tables/table-24 Income Statement.html", + "summary": "table-25 The table shows a consolidated income statement for three months ended Apr 26, 2026 and Apr 27, 2025, with segment operating income of $56,276M and $23,694M, leading to consolidated income before tax of $69,903M and $21,910M." + }, + { + "path": "tables/table-25 Revenue by Region.html", + "summary": "table-26 The table shows geographic revenue for three months ended Apr 26, 2026 and Apr 27, 2025. US revenue was $63,769M (2026) vs $25,685M (2025); Taiwan $12,006M vs $7,648M; China $4,550M vs $9,659M; Other $1,290M vs $1,070M. Total revenue $81,615M vs $44,062M." + }, + { + "path": "tables/table-26 Revenue by Market.html", + "summary": "table-27 The table shows revenue for three months ended Apr 26, 2026 and Apr 27, 2025. Data Center revenue grew from $39,112M to $75,246M, with Hyperscale and AI Clouds segments contributing significantly. Total revenue increased from $44,062M to $81,615M." + }, + { + "path": "tables/table-27 Lease Obligations.html", + "summary": "table-28 The table shows operating lease obligations totaling $5,604 million, with a present value of $4,344 million after imputed interest, resulting in long-term liabilities of $3,878 million." + }, + { + "path": "tables/table-28 Lease Cash Flows.html", + "summary": "table-29 The table shows supplemental cash flow information for three months ended Apr 26, 2026 and Apr 27, 2025. Operating cash flow used for leases was $185M and $96M; lease assets obtained were $1,516M and $98M." + }, + { + "path": "tables/table-29 Quarterly Financials.html", + "summary": "table-30 The table shows Q1 2026 revenue of $81.6B, up 85% YoY, net income of $58.3B (up 211% YoY), and gross margin of 74.9%." + }, + { + "path": "tables/table-30 Revenue by Market.html", + "summary": "table-31 Data Center revenue rose 92% year-over-year to $75,246M, with Hyperscale up 115% and AI Clouds up 74%. Total revenue reached $81,615M, a 20% quarterly and 85% yearly increase." + }, + { + "path": "tables/table-31 Revenue Analysis.html", + "summary": "table-32 The table shows revenue, cost, gross profit, operating expenses, and net income for three months ending Apr 26, 2026 and Apr 27, 2025, with percentages of revenue." + }, + { + "path": "tables/table-32 Revenue Growth.html", + "summary": "table-33 Compute & Networking revenue rose 88% to $74,550M, Graphics up 58% to $7,065M, total revenue increased 85% to $81,615M for the three months ended Apr 26, 2026 vs Apr 27, 2025." + }, + { + "path": "tables/table-33 Revenue Growth.html", + "summary": "table-34 Compute & Networking revenue rose 142% to $53,335M, Graphics up 79% to $2,941M, total revenue $56,276M, a 138% increase from $23,694M." + }, + { + "path": "tables/table-34 Operating Expenses.html", + "summary": "table-35 Total operating expenses increased 52% to $7,621 million, driven by R&D growth of 58% and SG&A growth of 25%." + }, + { + "path": "tables/table-35 Other Income.html", + "summary": "table-36 The table shows three-month financial data ending Apr 26, 2026 and Apr 27, 2025. Interest income increased by $25M, interest expense rose by $39M, and other income surged by $16,109M, leading to total other income of $16,367M." + }, + { + "path": "tables/table-36 Cash and Securities.html", + "summary": "table-37 The table shows cash and cash equivalents, marketable debt securities, and their total for Apr 26, 2026 ($50,335M) and Jan 25, 2026 ($49,670M)." + }, + { + "path": "tables/table-37 Cash Flow Data.html", + "summary": "table-38 The table shows net cash provided by operating activities of $50,344 million and $27,414 million for the three months ended Apr 26, 2026 and Apr 27, 2025, respectively. Net cash used in investing and financing activities also reported." + }, + { + "path": "tables/table-38 Debt Maturity.html", + "summary": "table-39 The table shows debt due in various periods totaling $8,470 million net carrying amount, with $1,000 million short-term and $7,470 million long-term portion." + }, + { + "path": "tables/table-39 Stock Buyback.html", + "summary": "table-40 The table shows a company repurchased 108.3 million shares from Jan to Apr 2026, with $38.5 billion remaining in the buyback program." + }, + { + "path": "tables/table-40 Stock Sale Plans.html", + "summary": "table-41 Tench Coxe, a Director, adopted a plan to sell up to 8,000,000 shares from 3/19/2026 to 10/30/2027. Colette M. Kress, CFO, terminated a plan to sell 500,000 shares on 4/10/2026." + }, + { + "path": "tables/table-41 Exhibit List.html", + "summary": "table-42 The table lists exhibits filed with the SEC, including a variable compensation plan, CEO/CFO certifications, and inline XBRL documents, with filing dates and incorporation references." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/full.md b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/full.md new file mode 100644 index 000000000..83b7fcea5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/full.md @@ -0,0 +1,1059 @@ +# UNITED STATES + +# SECURITIES AND EXCHANGE COMMISSION + +# Washington, D.C. 20549 + +## FORM 10-Q + +☒ QUARTERLY REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 + +For the quarterly period ended April 26, 2026 + +OR + +☐ TRANSITION REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 + +Commission File Number: 0-23985 + +![](images/9dd2f02ed25c1dbce8e283c38f39045324dad9ebd66175e4c5421295e4729bed.jpg) + +NVIDIA. + +## NVIDIA CORPORATION + +(Exact name of registrant as specified in its charter) + +Delaware + +(State or other jurisdiction of incorporation or organization) + +94-3177549 + +(I.R.S. Employer Identification No.) + +2788 San Tomas Expressway, Santa Clara, California (Address of principal executive offices) + +95051 +(Zip Code) + +(408) 486-2000 + +(Registrant's telephone number, including area code) + +N/A + +(Former name, former address and former fiscal year, if changed since last report) + +Securities registered pursuant to Section 12(b) of the Act: + +Title of each class + +Common Stock, \$0.001 par value per share + +Trading Symbol(s) + +NVDA + +Name of each exchange on which registered + +The Nasdaq Global Select Market + +Indicate by check mark whether the registrant (1) has filed all reports required to be filed by Section 13 or 15(d) of the Securities Exchange Act of 1934 during the preceding 12 months (or for such shorter period that the registrant was required to file such reports), and (2) has been subject to such filing requirements for the past 90 days. Yes ☑ No □ + +Indicate by check mark whether the registrant has submitted electronically every Interactive Data File required to be submitted pursuant to Rule 405 of Regulation S-T ( $§232.405$ of this chapter) during the preceding 12 months (or for such shorter period that the registrant was required to submit such files). Yes ☑ No ☐ + +Indicate by check mark whether the registrant is a large accelerated filer, an accelerated filer, a non-accelerated filer, a smaller reporting company, or an emerging growth company. See the definitions of “large accelerated filer,” “accelerated filer,” “smaller reporting company,” and “emerging growth company” in Rule 12b-2 of the Exchange Act. + +Large accelerated filer + +![](images/b84723b19101fb2d602a34c7c1dafb860fe6beca41db4d69acf5a6b06ac18dff.jpg) + +Accelerated filer + +![](images/6f9712903031ea0abaf169e132c58093c9891d5819e6fe52381b4a54f5e9aefe.jpg) + +Non-accelerated filer + +![](images/5fcbb2f88ac366cd3e2d744639f987270b013aa18ec5d8e54c682711dbd73c13.jpg) + +Smaller reporting company + +![](images/fa7bbf2bb9168f14c68dd0f1c114175ee74e50e9976ef0c346c8c20913ed761e.jpg) + +Emerging growth company + +![](images/c0596cced0662fcf6495e9b2343657380e4906809d245137302767be4748029a.jpg) + +If an emerging growth company, indicate by check mark if the registrant has elected not to use the extended transition period for complying with any new or revised financial accounting standards provided pursuant to Section 13(a) of the Exchange Act. □ + +Indicate by check mark whether the registrant is a shell company (as defined in Rule 12b-2 of the Exchange Act). Yes □ No ☒ + +The number of shares of common stock, \$0.001 par value, outstanding as of May 15, 2026, was 24.2 billion. + +# NVIDIA Corporation + +# Form 10-Q + +# For the Quarter Ended April 26, 2026 + +## Table of Contents + +Page + +Part I : Financial Information + +
Item 1.Financial Statements (Unaudited)
a) Condensed Consolidated Statements of Income for the three months ended April 26, 2026 and April 27, 20253
b) Condensed Consolidated Statements of Comprehensive Income for the three months ended April 26, 2026 and April 27, 20254
c) Condensed Consolidated Balance Sheets as of April 26, 2026 and January 25, 20265
d) Condensed Consolidated Statements of Shareholders’ Equity for the three months ended April 26, 2026 and April 27, 20256
e) Condensed Consolidated Statements of Cash Flows for the three months ended April 26, 2026 and April 27, 20257
f) Notes to Condensed Consolidated Financial Statements8
Item 2.Management’s Discussion and Analysis of Financial Condition and Results of Operations23
Item 3.Quantitative and Qualitative Disclosures About Market Risk30
Item 4.Controls and Procedures31
Part II : Other Information
Item 1.Legal Proceedings31
Item 1A.Risk Factors31
Item 2.Unregistered Sales of Equity Securities and Use of Proceeds38
Item 5.Other Information39
Item 6.Exhibits40
Signature41
+ +## Where You Can Find More Information + +Investors and others should note that we announce material financial information to our investors using our investor relations website, press releases, SEC filings and public conference calls and webcasts. We also use the following social media channels as a means of disclosing information about the company, our products, our planned financial and other announcements and attendance at upcoming investor and industry conferences, and other matters, and for complying with our disclosure obligations under Regulation FD: + +NVIDIA Corporate Blog (blogs.nvidia.com/) +NVIDIA Technical Blog (developer.nvidia.com/blog/) +NVIDIA LinkedIn (linkedin.com/company/nvidia) +NVIDIA Facebook (facebook.com/nvidia) +NVIDIA Instagram (instagram.com/nvidia) +NVIDIA X (x.com/nvidia) +NVIDIA Investor Relations (investor.nvidia.com) +NVIDIA YouTube (YouTube.com/nvidia) + +The information we post through these social media channels may be deemed material. Accordingly, investors should monitor these channels, in addition to following our press releases, SEC filings and public conference calls and webcasts. This list may be updated from time to time. The information we post through these channels is not a part of this Quarterly Report on Form 10-Q. + +## Part I. Financial Information + +## Item 1. Financial Statements (Unaudited) + +NVIDIA Corporation and Subsidiaries +Condensed Consolidated Statements of Income +(In millions, except per share data) +(Unaudited) + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
Revenue$ 81,615$ 44,062
Cost of revenue20,45817,394
Gross profit61,15726,668
Operating expenses
Research and development6,3213,989
Sales, general and administrative1,3001,041
Total operating expenses7,6215,030
Operating income53,53621,638
Interest income540515
Interest expense(102)(63)
Other income (expense), net15,929(180)
Total other income, net16,367272
Income before income tax69,90321,910
Income tax expense11,5823,135
Net income$ 58,321$ 18,775
Net income per share:
Basic$ 2.40$ 0.77
Diluted$ 2.39$ 0.76
Weighted average shares used in per share computation:
Basic24,28624,441
Diluted24,39124,611
+ +See accompanying Notes to Condensed Consolidated Financial Statements. + +## NVIDIA Corporation and Subsidiaries + +## Condensed Consolidated Statements of Comprehensive Income + +(In millions) + +(Unaudited) + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
Net income$ 58,321$ 18,775
Other comprehensive income (loss), net of tax
Available-for-sale securities:
Net change in unrealized gain (loss)(78)139
Cash flow hedges:
Net change in unrealized gain3719
Other comprehensive income (loss), net of tax(41)158
Total comprehensive income$ 58,280$ 18,933
+ +See accompanying Notes to Condensed Consolidated Financial Statements. + +NVIDIA Corporation and Subsidiaries +Condensed Consolidated Balance Sheets +(In millions) +(Unaudited) + +
Apr 26, 2026Jan 25, 2026
Assets
Current assets:
Cash and cash equivalents$ 13,237$ 10,605
Marketable debt securities37,09839,065
Marketable equity securities30,23712,886
Accounts receivable, net40,71038,466
Inventories25,79721,403
Prepaid expenses and other current assets3,9163,180
Total current assets150,995125,605
Property and equipment, net12,40310,383
Operating lease assets4,2582,867
Goodwill20,89420,832
Intangible assets, net3,1203,306
Deferred income tax assets11,70713,258
Non-marketable securities43,36422,251
Other assets12,7338,301
Total assets$ 259,474$ 206,803
Liabilities and Shareholders’ Equity
Current liabilities:
Accounts payable$ 13,097$ 9,812
Accrued and other current liabilities29,78721,352
Short-term debt1,000999
Total current liabilities43,88432,163
Long-term debt7,4707,469
Long-term operating lease liabilities3,8782,572
Other long-term liabilities8,7687,306
Total liabilities64,00049,510
Commitments and contingencies
Shareholders’ equity:
Preferred stock
Common stock2424
Additional paid-in capital10,27510,118
Accumulated other comprehensive income137178
Retained earnings185,038146,973
Total shareholders’ equity195,474157,293
Total liabilities and shareholders’ equity$ 259,474$ 206,803
+ +See accompanying Notes to Condensed Consolidated Financial Statements. + +NVIDIA Corporation and Subsidiaries +Condensed Consolidated Statements of Shareholders' Equity +(Unaudited) + +
Common Stock OutstandingAdditional Paid-in CapitalAccumulated Other Comprehensive IncomeRetained EarningsTotal Shareholders’ Equity
SharesAmount
(In millions, except per share data)
Balances as of Jan 25, 202624,304$ 24$ 10,118$ 178$ 146,973$ 157,293
Net income58,32158,321
Other comprehensive loss(41)(41)
Issuance of common stock37515515
Tax withholding related to common stock(12)(2,129)(2,129)
Shares repurchased(108)(157)(20,013)(20,170)
Cash dividends declared and paid ($0.01 per common share)(243)(243)
Stock-based compensation1,9281,928
Balances as of Apr 26, 202624,221$ 24$ 10,275$ 137$ 185,038$ 195,474
Balances as of Jan 26, 202524,477$ 24$ 11,237$ 28$ 68,038$ 79,327
Net income18,77518,775
Other comprehensive income158158
Issuance of common stock50370370
Tax withholding related to common stock(13)(1,532)(1,532)
Shares repurchased(126)(92)(14,411)(14,503)
Cash dividends declared and paid ($0.01 per common share)(244)(244)
Fair value of partially vested equity awards assumed in connection with acquisitions2222
Stock-based compensation1,4701,470
Balances as of Apr 27, 202524,388$ 24$ 11,475$ 186$ 72,158$ 83,843
+ +See accompanying Notes to Condensed Consolidated Financial Statements. + +NVIDIA Corporation and Subsidiaries +Condensed Consolidated Statements of Cash Flows +(In millions) +(Unaudited) + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
Cash flows from operating activities:
Net income$ 58,321$ 18,775
Adjustments to reconcile net income to net cash provided by operating activities:
Stock-based compensation expense1,9281,474
Deferred income taxes1,584(2,177)
Depreciation and amortization997611
(Gains) losses from equity securities, net(15,936)175
Other(94)(98)
Changes in operating assets and liabilities, net of acquisitions:
Accounts receivable(2,243)933
Inventories(4,420)(1,258)
Prepaid expenses and other assets(983)560
Accounts payable2,210941
Accrued and other current liabilities7,7637,128
Other long-term liabilities1,217350
Net cash provided by operating activities50,34427,414
Cash flows from investing activities:
Proceeds from maturities of marketable debt securities1,9463,122
Proceeds from sales of non-marketable securities26
Proceeds from sales of marketable debt securities25467
Purchases of non-marketable securities(18,582)(649)
Purchases of marketable debt and equity securities(8,000)(6,546)
Purchases related to property and equipment and intangible assets(1,757)(1,227)
Acquisitions, net of cash acquired(87)(383)
Net cash used in investing activities(26,429)(5,216)
Cash flows from financing activities:
Proceeds related to employee stock plans515370
Payments related to repurchases of common stock(19,312)(14,095)
Payments related to employee stock plan taxes(2,129)(1,532)
Dividends paid(243)(244)
Principal payments on property and equipment and intangible assets(33)(52)
Other(81)
Net cash used in financing activities(21,283)(15,553)
Change in cash and cash equivalents2,6326,645
Cash and cash equivalents at beginning of period10,6058,589
Cash and cash equivalents at end of period$ 13,237$ 15,234
+ +See accompanying Notes to Condensed Consolidated Financial Statements. + +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Unaudited) + +## Note 1 - Summary of Significant Accounting Policies + +## Basis of Presentation + +The accompanying unaudited condensed consolidated financial statements were prepared in accordance with accounting principles generally accepted in the United States of America, or U.S. GAAP, for interim financial information and with the instructions to Form 10-Q and Article 10 of Securities and Exchange Commission, or SEC, Regulation S-X. The January 25, 2026 consolidated balance sheet was derived from our audited consolidated financial statements included in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026, as filed with the SEC, but does not include all disclosures required by U.S. GAAP. In the opinion of management, all adjustments, consisting only of normal recurring adjustments considered necessary for a fair presentation of results of operations and financial position, have been included. The results for the interim periods presented are not necessarily indicative of the results expected for any future period. The following information should be read in conjunction with the audited consolidated financial statements and notes thereto included in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. + +Certain prior fiscal year balances have been reclassified to conform to the current period presentation. + +## Significant Accounting Policies + +There have been no material changes to our significant accounting policies disclosed in Note 1 - Organization and Summary of Significant Accounting Policies, of the Notes to the Consolidated Financial Statements included in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. + +## Fiscal Year + +Fiscal year 2027 is a 53-week year and fiscal year 2026 was a 52-week year, both ending on the last Sunday in January. The first quarters of fiscal years 2027 and 2026 were both 13-week quarters. The fourth quarter of fiscal year 2027 will be a 14-week quarter. + +## Principles of Consolidation + +Our condensed consolidated financial statements include the accounts of NVIDIA Corporation and our wholly owned subsidiaries. All intercompany balances and transactions have been eliminated in consolidation. + +## Use of Estimates + +The preparation of financial statements in conformity with U.S. GAAP requires management to make estimates and assumptions that affect the reported amounts of assets and liabilities and disclosures of contingent assets and liabilities at the date of the financial statements and the reported amounts of revenue and expenses during the reporting period. Actual results could differ materially from our estimates. + +## Recently Issued Accounting Pronouncements + +## Recent Accounting Pronouncements Not Yet Adopted + +In November 2024, the Financial Accounting Standards Board, or FASB, issued a new accounting standard requiring disclosures of certain additional expense information on an annual and interim basis, including, among other items, the amounts of purchases of inventory, employee compensation, depreciation and intangible asset amortization included within each income statement expense caption, as applicable. We will adopt this standard in the fiscal year 2028 annual report. We do not expect the adoption of this standard to have a material impact on our Consolidated Financial Statements other than additional disclosures. + +## Note 2 - Stock-Based Compensation + +We recognize stock-based compensation expense from grants of restricted stock units, or RSUs, performance stock units, or PSUs, and market-based PSUs, and issuances under our employee stock purchase plan, or ESPP. + +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +Condensed Consolidated Statements of Income include stock-based compensation expense as follows: + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Cost of revenue$ 68$ 64
Research and development1,4591,063
Sales, general and administrative401347
Total$ 1,928$ 1,474
+ +## Equity Award Activity + +The following is a summary of our equity award transactions under our equity incentive plans: + +RSUs, PSUs and Market-based PSUs Outstanding + +
Number of SharesWeighted Average Grant-Date Fair Value Per Share
(In millions, except per share data)
Balance as of Jan 25, 2026189$ 81.51
Granted44$ 181.73
Vested(31)$ 48.04
Canceled and forfeited(2)$ 95.82
Balance as of Apr 26, 2026200$ 108.92
+ +As of April 26, 2026, aggregate unearned stock-based compensation expense was \$20.8 billion, which is expected to be recognized over a weighted average period of 2.6 years for RSUs, PSUs, and market-based PSUs, and one year for ESPP. + +## Note 3 - Net Income Per Share + +The following is the basic and diluted net income per share computations for the periods presented: + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions, except per share data)
Numerator:
Net income$ 58,321$ 18,775
Denominator:
Basic weighted average shares24,28624,441
Dilutive impact of outstanding equity awards105170
Diluted weighted average shares24,39124,611
Net income per share:
Basic (1)$ 2.40$ 0.77
Diluted (2)$ 2.39$ 0.76
Anti-dilutive equity awards excluded from diluted net income per share4762
+ +(1) Net income divided by basic weighted average shares. +(2) Net income divided by diluted weighted average shares. + +Diluted net income per share was computed using the weighted average number of common and potentially dilutive shares outstanding during the period, using the treasury stock method. + +## NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +## Note 4 - Amortizable Intangible Assets and Goodwill + +The components of our amortizable intangible assets are as follows: + +
Apr 26, 2026Jan 25, 2026
Gross Carrying AmountAccumulated AmortizationNet Carrying AmountGross Carrying AmountAccumulated AmortizationNet Carrying Amount
(In millions)
Acquisition-related intangible assets$ 5,658$ (2,759)$ 2,899$ 5,656$ (2,580)$ 3,076
Patents and licensed technology525(304)221528(298)230
Total intangible assets$ 6,183$ (3,063)$ 3,120$ 6,184$ (2,878)$ 3,306
+ +Amortization expense associated with intangible assets was \$232 million and \$159 million for the first quarter of fiscal years 2027 and 2026, respectively. + +The following table outlines the estimated future amortization expense related to the net carrying amount of intangible assets as of April 26, 2026: + +
Future Amortization Expense(In millions)
Fiscal Year:
2027 (excluding the first quarter of fiscal year 2027)$ 689
2028754
2029610
2030516
2031468
2032 and thereafter83
Total$ 3,120
+ +In the first quarter of fiscal year 2027, goodwill increased by \$62 million from acquisitions and was allocated to our Compute \& Networking reporting unit. + +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +## Note 5 - Cash Equivalents and Marketable Securities + +Cash equivalents and marketable securities including debt and equity securities are measured at fair value using quoted prices in active markets for identical assets (Level 1) or for similar assets or use of other observable inputs (Level 2). + +The following is a summary of cash equivalents and marketable securities: + +Apr 26, 2026 + +
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable Debt SecuritiesMarketable Equity SecuritiesOther Assets
(In millions)
Debt securities issued by the U.S. TreasuryLevel 2$ 21,883$ 43$ (8)$ 21,918$ 470$ 21,448$ —$ —
Corporate debt securitiesLevel 215,09248(8)15,1321,53313,599
Debt securities issued by U.S. government agenciesLevel 22,0092(1)2,0102,010
Certificates of depositLevel 2132132132
Foreign government bondsLevel 24014141
Money market fundsLevel 110,21210,21210,212
Publicly-held equity securities (1) (2)Level 129,88721,0238,864
Publicly-held equity securities (1) (3)Level 29,2149,214
Total$ 49,368$ 94$ (17)$ 88,546$ 12,347$ 37,098$ 30,237$ 8,864
+ +(1) The balance as of April 26, 2026 included \$27.4 billion of investments, which are subject to short-term lock-up restrictions on the ability to sell. +(2) The long-term portion of publicly-held equity securities, which are subject to lock-up restrictions through December 2027 of \$8.9 billion as of April 26, 2026, was included in Other assets. +(3) The publicly-held equity securities classified in Level 2 include investments in warrants and preferred stock convertible to common stock in public companies. + +Publicly-held equity securities are subject to market price volatility. Net unrealized gains on investments in publicly-held equity securities held at period end were \$13.4 billion for the first quarter of fiscal year 2027. Net unrealized losses on investments in publicly-held equity securities held at period end were \$222 million for the first quarter of fiscal year 2026. Unrealized gains and losses are recognized in Other income (expense), net, in the Condensed Consolidated Statements of Income. + +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +Jan 25, 2026 + +
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable Debt SecuritiesMarketable Equity SecuritiesOther Assets
(In millions)
Debt securities issued by the U.S. TreasuryLevel 2$ 21,635$ 77$ (3)$ 21,709$ —$ 21,709$ —$ —
Corporate debt securitiesLevel 215,41092(3)15,49934515,154
Debt securities issued by U.S. government agenciesLevel 22,15742,1612,161
Certificates of depositLevel 2110110110
Foreign government bondsLevel 24014141
Money market fundsLevel 17,8307,8307,830
Publicly-held equity securities (1) (2)Level 117,72612,8864,840
Total$ 47,182$ 174$ (6)$ 65,076$ 8,285$ 39,065$ 12,886$ 4,840
+ +(1) The balance as of January 25, 2026 included \$10.5 billion of investments that are subject to short-term lock-up restrictions on the ability to sell. +(2) The long-term portion of publicly-held equity securities, which are subject to lock-up restrictions through December 2027 of \$4.8 billion as of January 25, 2026, was included in Other assets. + +The following table provides the breakdown of unrealized losses, aggregated by investment category and length of time that individual debt securities have been in a continuous loss position: + +
Apr 26, 2026Jan 25, 2026
Less than 12 monthsLess than 12 months
Estimated Fair ValueGross Unrealized LossEstimated Fair ValueGross Unrealized Loss
(In millions)
Debt securities issued by the U.S. Treasury$ 12,238$ (8)$ 10,666$ (3)
Corporate debt securities2,658(8)1,332(3)
Debt securities issued by U.S. government agencies1,291(1)1,134
Total$ 16,187$ (17)$ 13,132$ (6)
+ +Gross unrealized losses are related to fixed income securities, driven primarily by changes in interest rates. + +The estimated fair values of debt securities included in cash equivalents and marketable debt securities are shown below by contractual maturity. + +
Apr 26, 2026(In millions)
Less than one year$ 24,307
Due in 1 - 5 years14,926
Total$ 39,233
+ +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +## Note 6 - Non-marketable Securities + +## Non-marketable Equity Securities + +Our non-marketable equity securities are primarily in privately-held companies carried at cost less impairment, and adjusted for observable price changes. We value investments using observable comparable transactions and other inputs including volatility, expected time to liquidity, the risk-free rate, and security-specific rights and obligations. + +Adjustments to the carrying value of privately-held securities: + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Balance at beginning of period$ 22,251$ 3,387
Adjustments related to non-marketable equity securities:
Net additions17,899649
Unrealized gains (1)2,60363
Reclassification (2)(389)(843)
Impairments and unrealized losses(28)(16)
Balance at end of period$ 42,336$ 3,240
+ +(1) Unrealized gains are recognized in Other income (expense), net, in the Condensed Consolidated Statements of Income. +(2) Includes primarily reclassifications to marketable securities following public market trading. + +Non-marketable equity securities had cumulative gross unrealized gains of \$5.3 billion and \$396 million, and cumulative gross unrealized losses and impairments of \$199 million and \$110 million as of April 26, 2026 and April 27, 2025, respectively. + +## Equity Method Investments + +We have \$1.0 billion of investments in infrastructure funds accounted for using the equity method as of April 26, 2026. Our maximum loss exposure under these investments, including invested and future committed amounts, was \$2.3 billion as of April 26, 2026. + +## Investment Commitments + +Total Investment commitments were \$27 billion as of April 26, 2026, subject to certain contingencies, which we expect will be made through the remainder of fiscal year 2027. + +## Note 7 - Balance Sheet Components + +We refer to customers who purchase products directly from NVIDIA as direct customers, such as add-in board manufacturers, or AIBs, distributors, original design manufacturers, or ODMs, original equipment manufacturers, or OEMs, cloud service providers, or CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. Three direct customers accounted for 30%, 18%, and 16% of our accounts receivable balance as of April 26, 2026. Three direct customers accounted for 25%, 18%, and 13% of our accounts receivable balance as of January 25, 2026. + +Certain balance sheet components were as follows: + +
Apr 26, 2026Jan 25, 2026
Inventories:(In millions)
Raw materials$ 6,647$ 3,807
Work in process9,9498,822
Finished goods9,2018,774
Total inventories (1)$ 25,797$ 21,403
+ +(1) We recorded inventory provisions of \$0.8 billion and \$2.3 billion for the first quarter of fiscal years 2027 and 2026, respectively, in Cost of revenue. + +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +## Property and Equipment: + +Property, equipment and intangible assets acquired but not paid for the first quarter of fiscal years 2027 and 2026 were \$1.1 billion and \$408 million, respectively. + +
Apr 26, 2026Jan 25, 2026
Accrued and Other Current Liabilities:(In millions)
Taxes payable$ 10,638$ 2,669
Customer program accruals4,1825,318
Accrued purchase consideration (1)3,9573,921
Excess inventory purchase obligations (2)3,1212,739
Product warranty2,9482,807
Deferred revenue (3)1,7141,379
Accrued payroll and related expenses1,0331,146
Other2,1941,373
Total accrued and other current liabilities$ 29,787$ 21,352
+ +(1) Related to the Groq, Inc. non-exclusive license agreement. +(2) We recorded \$0.3 billion and \$3.0 billion for the first quarter of fiscal years 2027 and 2026, respectively, in Cost of revenue. +(3) Includes customer advances and unearned revenue related to hardware and software support, cloud services, and license and development arrangements. The balance as of April 26, 2026 and January 25, 2026 included \$297 million and \$160 million of customer advances, respectively. + +
Apr 26, 2026Jan 25, 2026
Other Long-Term Liabilities:(In millions)
Income tax payable (1)$ 4,830$ 3,958
Deferred income tax1,7981,774
Deferred revenue (2)1,4031,193
Other737381
Total other long-term liabilities$ 8,768$ 7,306
+ +(1) Primarily comprised of unrecognized tax benefits and related interest and penalties. +(2) Includes unearned revenue related to hardware and software support and cloud services. + +## Deferred Revenue + +The following table shows the changes in short- and long-term deferred revenue during the first quarter of fiscal years 2027 and 2026: + +Three Months Ended + +
Apr 26, 2026Apr 27, 2025
(In millions)
Balance at beginning of period$ 2,572$ 1,813
Deferred revenue additions (1)2,5306,493
Revenue recognized (2)(1,985)(6,228)
Balance at end of period$ 3,117$ 2,078
+ +(1) Includes \$1.7 billion and \$6.2 billion of customer advances for the first quarter of fiscal years 2027 and 2026, respectively. +(2) Includes \$1.6 billion and \$6.0 billion related to customer advances for the first quarter of fiscal years 2027 and 2026, respectively. + +We recognized revenue of \$451 million and \$265 million in the first quarter of fiscal years 2027 and 2026, respectively, that was included in the prior year-end deferred revenue balance. + +As of April 26, 2026, revenue related to remaining performance obligations from contracts greater than one year in length was \$2.6 billion, which includes \$2.3 billion from deferred revenue and \$304 million, which has not yet been billed or + +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +recognized as revenue. Approximately 40% of revenue from contracts greater than one year in length will be recognized over the next twelve months. + +## Note 8 - Derivative Financial Instruments + +## Foreign Currency Derivatives + +We primarily utilize foreign currency forward contracts to mitigate the impact of foreign currency exchange rate movements on our operating expenses. These foreign currency forward contracts for operating expenses are designated as accounting hedges. Gains or losses on the contracts are recorded in Accumulated other comprehensive income or loss and reclassified to Operating expenses when the related operating expenses are recognized in earnings. During the first quarter of fiscal years 2027 and 2026, the impact of foreign currency forward contracts designated as accounting hedges on other comprehensive income or loss was not significant and all such instruments were determined to be highly effective. + +We also entered into foreign currency forward contracts to mitigate the impact of foreign currency movements on monetary assets and liabilities. For our foreign currency contracts for assets and liabilities, the change in fair value of these non-designated contracts was recorded in Other income or expense and offsets the change in fair value of the hedged foreign currency denominated monetary assets and liabilities, which was also recorded in Other income (expense), net. + +The table below presents the notional value of our foreign currency contracts outstanding: + +
Apr 26, 2026Jan 25, 2026
(In millions)
Designated as accounting hedges$ 2,114$ 1,765
Not designated as accounting hedges$ 1,850$ 2,332
+ +The fair values of our foreign currency contracts were not significant as of April 26, 2026 and January 25, 2026. + +As of April 26, 2026, all foreign currency contracts mature within 18 months. The expected realized gains and losses deferred into Accumulated other comprehensive income or loss related to foreign currency forward contracts within the next twelve months were not significant. + +## Facility Lease Guarantee + +In fiscal year 2026, we entered into agreements to guarantee partners' facility lease obligations in the event of their default in exchange for warrants. The maximum gross exposure under all agreements is \$3.5 billion, which is reduced as the partners make payments to the lessors over terms ranging from 5 to 7 years. The partners have placed \$712 million in escrow to mitigate our potential exposure. The guarantees, classified as credit derivatives with changes in fair value recognized in Other income (expense), net, were not material. + +## NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +Note 9 - Debt + +
Expected Remaining Term (years)Effective Interest RateApr 26, 2026Jan 25, 2026
(In millions)
3.20% Notes Due 20260.43.31%$ 1,000$ 1,000
1.55% Notes Due 20282.11.64%1,2501,250
2.85% Notes Due 20303.92.93%1,5001,500
2.00% Notes Due 20315.12.09%1,2501,250
3.50% Notes Due 204013.93.54%1,0001,000
3.50% Notes Due 205023.93.54%2,0002,000
3.70% Notes Due 206034.03.73%500500
Unamortized debt discount and issuance costs(30)(32)
Net carrying amount$ 8,470$ 8,468
Less short-term portion(1,000)(999)
Total long-term portion$ 7,470$ 7,469
+ +As of April 26, 2026 and January 25, 2026, the estimated fair value of debt was \$7.4 billion and \$7.5 billion, respectively. The estimated fair values are based on Level 2 inputs. + +Our notes are unsecured senior obligations. Existing and future liabilities of our subsidiaries will be effectively senior to the notes. Our notes pay interest semi-annually. We may redeem each of our notes prior to maturity, subject to a make-whole premium. The maturity dates of the notes are stated by calendar year. + +As of April 26, 2026, we complied with the required covenants, which are non-financial in nature, under the outstanding notes. + +As of April 26, 2026, our commercial paper program had a capacity of \$25.0 billion, with no amounts outstanding. + +## Note 10 - Commitments and Contingencies + +## Commitments + +Manufacturing, supply, and capacity commitments reflect data center-scale production and longer future ordering horizons across current and future product architectures. We enter into agreements with our supply vendors that allow them to procure inventory based upon our defined criteria, and in certain instances, these agreements are cancellable, able to be rescheduled, or adjustable for our business needs prior to placing firm orders. Changes to these agreements may result in additional costs. As of April 26, 2026, these commitments were \$119 billion for which \$95 billion will be paid in the remainder of fiscal year 2027 and the remaining balance will be paid in fiscal years 2028 through 2031. + +Multi-year cloud service agreement commitments as of April 26, 2026, were \$30 billion for which \$6 billion, \$7 billion, \$7 billion, \$5 billion, \$3 billion, and \$2 billion will be paid in the remainder of fiscal year 2027, each fiscal year from 2028 through 2031, and fiscal year 2032 and thereafter, respectively. Cloud service capacity may be reduced or terminated. Cloud service agreements will be primarily used to support our research and development efforts. + +Other vendor commitments were \$6 billion as of April 26, 2026, of which the majority will be paid through fiscal year 2027. + +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +## Accrual for Product Warranty Liabilities + +The estimated amount of product warranty liabilities was \$2.9 billion and \$2.8 billion as of April 26, 2026 and January 25, 2026, respectively. The estimated product returns and product warranty activity consisted of the following: + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Balance at beginning of period$ 2,807$ 1,290
Additions330870
Utilization(189)(80)
Balance at end of period$ 2,948$ 2,080
+ +For the first quarter of fiscal years 2027 and 2026, the additions in product warranty liabilities primarily related to our Compute & Networking segment. + +We have provided indemnities for matters such as tax, product, and employee liabilities. We have included intellectual property indemnification provisions in our technology-related agreements with third parties. Maximum potential future payments cannot be estimated because many of these agreements do not have a maximum stated liability. We have not recorded any liability in our Condensed Consolidated Financial Statements for such indemnifications. + +## Litigation + +## Securities Class Action and Derivative Lawsuits + +The plaintiffs in the putative securities class action lawsuit, captioned 4:18-cv-07669-HSG, initially filed on December 21, 2018 in the United States District Court for the Northern District of California, and titled In Re NVIDIA Corporation Securities Litigation, filed an amended complaint on May 13, 2020. The amended complaint asserted that NVIDIA and certain NVIDIA executives violated Section 10(b) of the Securities Exchange Act of 1934, as amended, or the Exchange Act, and SEC Rule 10b-5, by making materially false or misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand between May 10, 2017 and November 14, 2018. Plaintiffs also alleged that the NVIDIA executives who they named as defendants violated Section 20(a) of the Exchange Act. Plaintiffs sought class certification, an award of unspecified compensatory damages, an award of reasonable costs and expenses, including attorneys' fees and expert fees, and further relief as the Court may deem just and proper. On March 2, 2021, the district court granted NVIDIA's motion to dismiss the complaint without leave to amend, entered judgment in favor of NVIDIA and closed the case. On March 30, 2021, plaintiffs filed an appeal from judgment in the United States Court of Appeals for the Ninth Circuit, case number 21-15604. On August 25, 2023, a majority of a three-judge Ninth Circuit panel affirmed in part and reversed in part the district court's dismissal of the case, with a third judge dissenting on the basis that the district court did not err in dismissing the case. NVIDIA filed a petition for a writ of certiorari on March 4, 2024. On June 17, 2024, the Supreme Court of the United States granted NVIDIA's petition for a writ of certiorari. After briefing and argument, the Supreme Court dismissed NVIDIA's writ of certiorari as improvidently granted on December 11, 2024, and issued judgment on January 13, 2025. On February 20, 2025, the Ninth Circuit's judgment, entered August 25, 2023 and corrected August 28, 2023, took effect, and the case was remanded to the district court for further proceedings. On March 25, 2026, the district court granted plaintiffs' motion for class certification and certified a class of investors consisting of all persons or entities who purchased or otherwise acquired NVIDIA common stock between August 10, 2017, and November 15, 2018, inclusive, excluding certain persons and entities, such as NVIDIA's officers and directors, and members of their immediate families, among others. On April 8, 2026, NVIDIA filed a petition with the Ninth Circuit for permission to appeal the district court's order pursuant to Federal Rule of Civil Procedure 23(f). + +The putative derivative lawsuit pending in the United States District Court for the Northern District of California, captioned 4:19-cv-00341-HSG, initially filed January 18, 2019 and titled In re NVIDIA Corporation Consolidated Derivative Litigation, was stayed pending resolution of the plaintiffs' appeal in the In Re NVIDIA Corporation Securities Litigation action. On February 22, 2022, the court administratively closed the case, but stated that it would reopen the case once the appeal in the In Re NVIDIA Corporation Securities Litigation action is resolved. The case has not yet been reopened by the court. The lawsuit asserts claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty, unjust enrichment, waste of corporate assets, and violations of Sections 14(a), 10(b), and 20(a) of the Exchange Act based on the dissemination of allegedly false and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs are seeking unspecified damages and other relief, including reforms and improvements to NVIDIA's corporate governance and internal procedures. + +The putative derivative actions initially filed September 24, 2019 and pending in the United States District Court for the District of Delaware, Lipchitz v. Huang, et al. (Case No. 1:19-cv-01795-MN) and Nelson v. Huang, et. al. (Case No. 1:19- + +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +cv-01798-MN), were stayed pending resolution of the plaintiffs' appeal in the In Re NVIDIA Corporation Securities Litigation action. On March 7, 2025, after the Supreme Court issued its judgment dismissing the Company's petition for writ of certiorari as improvidently granted in the In Re NVIDIA Securities Litigation action, the district court adopted the parties' stipulation to extend the stay until the final and complete resolution of the In Re NVIDIA Corporation Securities Litigation action. The lawsuits assert claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty, unjust enrichment, insider trading, misappropriation of information, corporate waste and violations of Sections 14(a), 10(b), and 20(a) of the Exchange Act based on the dissemination of allegedly false and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs seek unspecified damages and other relief, including disgorgement of profits from the sale of NVIDIA stock and unspecified corporate governance measures. + +Another putative derivative action was filed on October 30, 2023 in the Court of Chancery of the State of Delaware, captioned Horanic v. Huang, et al. (Case No. 2023-1096-KSJM). This lawsuit asserts claims, purportedly on behalf of us, against certain officers and directors of the Company for breach of fiduciary duty and insider trading based on the dissemination of allegedly false and misleading statements related to channel inventory and the impact of cryptocurrency mining on GPU demand. The plaintiffs seek unspecified damages and other relief, including disgorgement of profits from the sale of NVIDIA stock and reform of unspecified corporate governance measures. On August 11, 2025, the court granted the parties' stipulation to voluntarily dismiss with prejudice plaintiff City of Westland Police and Fire Retirement System. This derivative matter is stayed pending the final resolution of In Re NVIDIA Corporation Securities Litigation action. + +## Accounting for Loss Contingencies + +As of April 26, 2026, there are no accrued contingent liabilities associated with the legal proceedings described above based on our belief that liabilities, while reasonably possible, are not probable. Further, any possible loss or range of loss in these matters cannot be reasonably estimated at this time. We are engaged in legal actions not described above arising in the ordinary course of business, as well as regulatory and government inquiries and investigations, and, while there can be no assurance of favorable outcomes, we believe that the ultimate outcome of these matters will not have a material adverse effect on our operating results, liquidity or financial position. These matters are subject to inherent uncertainties and if the ultimate outcome is unfavorable, there exists the possibility of a material adverse impact on our operating results, liquidity or financial position in the period the outcome becomes estimable and probable. + +## Note 11 - Income Taxes + +Income tax expense was \$11.6 billion and \$3.1 billion for the first quarter of fiscal years 2027 and 2026, respectively. Income tax as a percentage of income before income tax was 16.6% and 14.3% for the first quarter of fiscal years 2027 and 2026, respectively. + +The effective tax rate increased primarily due to a lower percentage of tax benefits from stock-based compensation relative to the increase in income before income tax. + +Our effective tax rates for the first quarter of fiscal years 2027 and 2026 were lower than the U.S. federal statutory rate of 21% primarily due to tax benefits from foreign-derived deduction eligible income, income earned in jurisdictions that were subject to taxes at rates lower than the U.S. federal statutory tax rate, stock-based compensation, and the U.S. federal research tax credit. + +While we believe that we have adequately provided for all uncertain tax positions, or tax positions where we believe it is not more-likely-than-not that the position will be sustained upon review, amounts asserted by tax authorities could be greater or less than our accrued position. Accordingly, our provisions on federal, state and foreign tax related matters to be recorded in the future may change as revised estimates are made or the underlying matters are settled or otherwise resolved with the respective tax authorities. + +We are currently under examination by the Internal Revenue Service for our fiscal years 2023 and 2024. + +## Note 12 - Shareholders' Equity + +## Capital Return Program + +We repurchased 108 million and 126 million shares of our common stock for \$20.2 billion and \$14.5 billion during the first quarter of fiscal years 2027 and 2026, respectively. As of April 26, 2026, we were authorized, subject to certain specifications, to repurchase up to \$38.5 billion of our common stock. + +On May 18, 2026, our Board of Directors approved an additional \$80.0 billion in share repurchase authorization, without expiration. + +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +We paid cash dividends to our shareholders of \$243 million and \$244 million during the first quarter of fiscal years 2027 and 2026, respectively. On May 18, 2026, we increased our quarterly cash dividend from \$0.01 per share to \$0.25 per share to all shareholders of record on June 4, 2026. Our quarterly cash dividend will be paid on June 26, 2026. + +The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders. + +## Note 13 - Segment Information + +Our Chief Executive Officer is our chief operating decision maker, or CODM, and reviews financial information presented on an operating segment basis for purposes of making decisions and assessing financial performance. Our CODM assesses operating performance of each segment based on regularly provided segment revenue and segment operating income. Operating results by segment include costs or expenses directly attributable to each segment, and costs or expenses that are leveraged across our unified architecture and therefore allocated between our two segments. Our CODM reviews expenses on a consolidated basis, and expenses attributable to each segment are not regularly provided to our CODM. + +The Compute & Networking segment includes our Data Center accelerated computing and networking platforms and AI solutions and software, and automotive platforms and autonomous and electric vehicle solutions including software. + +The Graphics segment includes GeForce GPUs for gaming and PCs, and Quadro/NVIDIA RTX GPUs for enterprise workstation graphics. + +Certain expenses are not allocated to either Compute & Networking or Graphics for purposes of making operating decisions or assessing financial performance. The expenses include stock-based compensation expense, corporate infrastructure and support costs, acquisition-related and other costs, and other non-recurring charges and benefits that our CODM deems to be enterprise in nature. + +Our CODM does not review any information regarding total assets on a reportable segment basis. There are no intersegment transactions. The accounting policies for segment reporting are the same as for our consolidated financial statements. The table below presents details of our reportable segments. + +
Compute & NetworkingGraphicsTotal
(In millions)
Three Months Ended Apr 26, 2026
Revenue$ 74,550$ 7,065$ 81,615
Other segment items (1)21,2154,12425,339
Operating income$ 53,335$ 2,941$ 56,276
Three Months Ended Apr 27, 2025
Revenue$ 39,589$ 4,473$ 44,062
Other segment items (1)17,5352,83320,368
Operating income$ 22,054$ 1,640$ 23,694
+ +(1) Other segment items primarily include product costs and inventory provisions, compensation and benefits excluding stock-based compensation expense, computing infrastructure expenses, and engineering development costs. + +Depreciation and amortization expense attributable to our Compute & Networking segment was \$526 million and \$296 million for the first quarter of fiscal years 2027 and 2026, respectively. Depreciation and amortization expense attributable to our Graphics segment was \$194 million and \$109 million for the first quarter of fiscal years 2027 and 2026, respectively. Acquisition-related intangible amortization expense is not allocated to either Compute & Networking or Graphics for purposes of making operating decisions or assessing financial performance. + +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +Reconciliation of segment operating income to consolidated income before income tax for the first quarter of fiscal years 2027 and 2026 was as follows: + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Segment operating income$ 56,276$ 23,694
Stock-based compensation expense(1,928)(1,474)
Unallocated operating expenses(565)(419)
Acquisition-related and other costs(247)(163)
Interest income540515
Interest expense(102)(63)
Other income (expense), net15,929(180)
Consolidated income before income tax$ 69,903$ 21,910
+ +Revenue by geographic area is based upon the location of the customers' headquarters. The end customer and shipping location may be different from our customers' headquarters location. + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Geographic Revenue based upon Customer Headquarters Location:
United States$ 63,769$ 25,685
Taiwan12,0067,648
China (including Hong Kong)4,5509,659
Other1,2901,070
Total revenue$ 81,615$ 44,062
+ +Revenue from sales to customers headquartered outside of the United States accounted for 22% of total revenue for the first quarter of fiscal year 2027 and 42% of total revenue for the first quarter of fiscal year 2026. + +We refer to customers who purchase products directly from NVIDIA as direct customers, such as AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. We refer to indirect customers as those who purchase products through our direct customers; indirect customers include CSPs, AI Clouds, AI model makers, enterprises, and public sector entities. Our revenue is concentrated among a limited number of direct and indirect customers and this trend may continue. + +For the first quarter of fiscal year 2027, three direct customers represented 21%, 17%, and 16% of total revenue, all of which was primarily attributable to the Compute & Networking segment. + +For the first quarter of fiscal year 2026, sales to two direct customers represented 16% and 14% of total revenue, which were attributable to the Compute & Networking segment. + +# NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +In the first quarter of fiscal year 2027, we changed our presentation of revenue by market platform, and the comparable period has been recast as follows: + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Revenue by Market Platform
Data Center$ 75,246$ 39,112
Hyperscale37,86917,599
AI Clouds, Industrial, & Enterprise37,37721,513
Edge Computing6,3694,950
Total revenue$ 81,615$ 44,062
+ +## Note 14 - Leases + +Our lease obligations primarily consist of operating leases for our data centers and offices, with lease periods expiring between fiscal years 2027 and 2075. + +Future minimum lease obligations under our non-cancelable lease agreements as of April 26, 2026 were as follows: + +
Operating Lease Obligations (In millions)
Fiscal Year:
2027 (excluding the first quarter of fiscal 2027)$ 460
2028626
2029602
2030530
2031462
2032 and thereafter2,924
Total5,604
Less imputed interest1,260
Present value of net future minimum lease payments4,344
Less short-term operating lease liabilities466
Long-term operating lease liabilities$ 3,878
+ +Between the second quarter of fiscal year 2027 and fiscal year 2033, we expect to commence leases with future obligations of \$32.4 billion, primarily for data center leases to support our research and development efforts, with lease terms of 3 to 20 years. + +Operating lease costs were \$171 million and \$101 million for the first quarter of fiscal years 2027 and 2026, respectively. Short-term, variable, and finance lease costs for the first quarter of fiscal years 2027 and 2026 were not significant. + +Other information related to leases was as follows: + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Supplemental cash flows information
Operating cash flow used for operating leases$ 185$ 96
Operating lease assets obtained in exchange for lease obligations$ 1,516$ 98
+ +## NVIDIA Corporation and Subsidiaries Notes to Condensed Consolidated Financial Statements (Continued) (Unaudited) + +As of April 26, 2026, our operating leases have a weighted average remaining lease term of 10.4 years and a weighted average discount rate of 4.61%. As of January 25, 2026, our operating leases had a weighted average remaining lease term of 8.8 years and a weighted average discount rate of 4.38%. + +## Item 2. Management's Discussion and Analysis of Financial Condition and Results of Operations + +## Forward-Looking Statements + +This Quarterly Report on Form 10-Q contains forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the “safe harbor” created by those sections based on management’s beliefs and assumptions and on information currently available to management. In some cases, you can identify forward-looking statements by terms such as “may,” “will,” “should,” “could,” “goal,” “would,” “expect,” “plan,” “anticipate,” “believe,” “estimate,” “project,” “predict,” “potential” and similar expressions intended to identify forward-looking statements. These statements involve known and unknown risks, uncertainties and other factors, which may cause our actual results, performance, time frames or achievements to be materially different from any future results, performance, time frames or achievements expressed or implied by the forward-looking statements. We discuss many of these risks, uncertainties and other factors in this Quarterly Report on Form 10-Q and our Annual Report on Form 10-K for the fiscal year ended January 25, 2026 in greater detail under the heading “Risk Factors” of such reports. Given these risks, uncertainties, and other factors, you should not place undue reliance on these forward-looking statements. Also, these forward-looking statements represent our estimates and assumptions only as of the date of this filing. You should read this Quarterly Report on Form 10-Q completely and understand that our actual future results may be materially different from what we expect. We hereby qualify our forward-looking statements by these cautionary statements. Except as required by law, we assume no obligation to update these forward-looking statements publicly, or to update the reasons actual results could differ materially from those anticipated in these forward-looking statements, even if new information becomes available in the future. + +All references to “NVIDIA,” “we,” “us,” “our” or the “Company” mean NVIDIA Corporation and its subsidiaries. + +In addition, statements that “we believe” and similar statements reflect our beliefs and opinions on the relevant subject. These statements are based upon information available to us as of the filing date of this Quarterly Report on Form 10-Q, and while we believe such information forms a reasonable basis for such statements, such information may be limited or incomplete, and our statements should not be read to indicate that we have conducted an exhaustive inquiry into, or review of, all potentially available relevant information. These statements are inherently uncertain, and investors are cautioned not to unduly rely upon these statements. + +© 2026 NVIDIA Corporation. All rights reserved. + +The following discussion and analysis of our financial condition and results of operations should be read in conjunction with the risk factors set forth in Item 1A. "Risk Factors" of our Annual Report on Form 10-K for the fiscal year ended January 25, 2026 and Part II, Item 1A. "Risk Factors" of this Quarterly Report on Form 10-Q and our Condensed Consolidated Financial Statements and related Notes thereto, as well as other cautionary statements and risks described elsewhere in this Quarterly Report on Form 10-Q and our other filings with the SEC, before deciding to purchase, hold, or sell shares of our common stock. + +## Overview + +## Our Company and Our Businesses + +NVIDIA pioneered accelerated computing to help solve the most challenging computational problems. Since our original focus on PC graphics, we have expanded to several other large and important computationally intensive fields. Fueled by the sustained demand for exceptional 3D graphics and the scale of the gaming market, NVIDIA has leveraged its GPU architecture to create platforms for scientific computing, AI, data science, autonomous vehicles, robotics, and digital twin applications. NVIDIA is now a data center-scale AI infrastructure company reshaping all industries. + +Our two operating segments are “Compute & Networking” and “Graphics.” Refer to Note 13 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q for additional information. + +Headquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998. + +## Recent Developments, Future Objectives and Challenges + +Revenue growth in the first quarter was driven by data center products for accelerated computing and AI solutions. Blackwell continued to account for the majority of our system shipments. + +The availability of data centers, energy, and capital to support the buildout of NVIDIA AI infrastructure by our customers and partners is crucial, and any shortage of these or other necessary resources could impact our future revenue and financial performance. Expanding energy capacity to meet demand is a complex, multi-year process that involves significant regulatory, technical, and construction challenges. In addition, access to capital can be particularly constrained for less-capitalized companies, which may face difficulties securing financing for large-scale infrastructure projects. These limitations could delay customer and partner deployments or reduce the scale of accelerated computing and AI adoption. + +We expect our Rubin platform to start shipping in the second half of fiscal year 2027. The complexity of bringing up our product architecture and sophisticated system configurations has caused and may in the future cause delays in production and create challenges in managing supply and demand. This could further result in revenue volatility, quality issues, increased inventory provisions, decreases in product yields and higher material costs, and/or increased warranty costs. Customers may postpone purchasing new architectures or may adopt new technologies more gradually than anticipated, affecting our revenue timing and supply chain expenses. + +Beginning in February 2026, the U.S. government, or USG, granted licenses that allow us to ship small amounts of H200 products to specific China-based customers. To date, we have not generated any revenue under the H200 licensing program, and do not yet know whether any imports will be allowed into China. The license requires that the H200s go through an inspection process in the United States prior to any shipment to the customer. As a result, any H200 shipped under the new licensing program will be subject to a 25% tariff upon importation into the United States. + +The recent rise in high-quality, open-source foundation models is making advanced AI capabilities broadly accessible. Open-source AI is dependent on developer adoption, and if deployed on our competitors' platforms, it could reduce demand for our products and services. + +We have made, and expect to continue making, investments in our ecosystem to enhance our growth opportunities, cultivate our ecosystem, and strengthen our competitive position. In the first quarter of fiscal year 2027, we made the following investments: + +- \$18.6 billion in private companies and infrastructure funds. Some of these investments include AI model makers that may indirectly purchase or use our products in the cloud. +• We made investments in publicly-held equity securities where the value may fluctuate significantly and could adversely affect our financial results. + +Our global supply chain for our networking products, including our Israel operations of approximately 5,900 employees supporting research and development, operations, and sales and marketing, has not been significantly impacted by the conflict in the Middle East. If the conflict escalates or extends, it could affect future product development, supply chain, and revenue, and create business uncertainty. + +Macroeconomic factors, including tariffs, inflation, interest changes, capital market volatility, global supply chain constraints, and global economic and geopolitical developments and conflicts, have direct and indirect impacts on our results of operations, particularly demand for our products. While difficult to isolate and quantify, these macroeconomic factors impact our supply chain and manufacturing costs, employee wages, costs for capital equipment, the value of our investments, revenue, and competitive position. Our product and solution pricing generally does not fluctuate with short-term changes in our costs. Within our supply chain, we continuously manage product availability and costs with our vendors. + +Refer to Part II, Item 1A, "Risk Factors" for a discussion of these factors and other risks. + +First Quarter of Fiscal Year 2027 Summary + +
Three Months EndedQuarter-over-Quarter ChangeYear-over-Year Change
Apr 26, 2026Jan 25, 2026Apr 27, 2025
($ in millions, except per share data)
Revenue$ 81,615$ 68,127$ 44,06220 %85 %
Gross margin74.9 %75.0 %60.5 %(0.1) pts14.4 pts
Operating expenses$ 7,621$ 6,794$ 5,03012 %52 %
Operating income$ 53,536$ 44,299$ 21,63821 %147 %
Net income$ 58,321$ 42,960$ 18,77536 %211 %
Net income per diluted share$ 2.39$ 1.76$ 0.7636 %214 %
+ +We specialize in markets where our computing platforms can provide tremendous acceleration for applications. These platforms incorporate processors, interconnects, software, algorithms, systems, and services to deliver unique value. + +Following the rapid evolution in our businesses, we are transitioning to a new reporting framework that better reflects our current and future growth drivers. + +We will have two market platforms – Data Center and Edge Computing. + +Within Data Center, we will report two sub-markets, Hyperscale and ACIE which incorporates AI Clouds, Industrial, and Enterprise. Hyperscale will include revenue from the public clouds and the world's largest consumer internet companies, + +while ACIE addresses our growth opportunity in diverse AI purpose-built data centers and AI factories across industries and countries. + +Edge Computing highlights devices for agentic and physical AI including PCs, game consoles, workstations, AI-RAN base stations, robotics and automotive. + +
Three Months EndedQuarter-over-Quarter ChangeYear-over-Year Change
Apr 26, 2026Jan 25, 2026Apr 27, 2025
($ in millions)
Revenue by Market Platform (1)
Data Center$ 75,246$ 62,314$ 39,11221 %92 %
Hyperscale37,86933,81417,59912 %115 %
AI Clouds, Industrial, & Enterprise37,37728,50021,51331 %74 %
Edge Computing6,3695,8134,95010 %29 %
Total revenue$ 81,615$ 68,127$ 44,06220 %85 %
+ +(1) In the first quarter of fiscal year 2027, we changed our presentation of revenue by market platform and the comparable periods have been recast. + +Revenue was \$81.6 billion, up 85% from a year ago and up 20% sequentially. + +Data Center revenue was \$75.2 billion, up 92% from a year ago and up 21% sequentially, driven by the ramp of our Blackwell 300 products and demand for our InfiniBand, Spectrum-X Ethernet, and NVLink solutions. Hyperscaler revenue increased sequentially and remained at approximately 50% of Data Center revenue, while the remaining 50% came from a continued diversification of customers, including AI Clouds, industrial, enterprise, and sovereign customers. No shipments of Data Center Hopper products to China occurred during the quarter, compared with \$4.6 billion in the first quarter of fiscal year 2026. + +Edge Computing revenue for the first quarter was \$6.4 billion, up 29% from a year ago and up 10% sequentially. The increases were driven by robust Blackwell workstation demand, partially offset by slower consumer PC demand that was tempered by elevated memory and systems prices. + +Gross margin increased from a year ago on lower inventory provisions, primarily due to the prior year's \$4.5 billion charge associated with H20 excess inventory and purchase obligations. Gross margin was approximately flat sequentially as our Blackwell architecture remains the majority of our revenue. + +Operating expenses were up 52% from a year ago and up 12% sequentially. The increases were primarily driven by higher compensation and benefits expense due to employee growth and compensation increases, compute and infrastructure costs, and engineering development materials for new product developments. + +## Financial Information by Business Segment and Geographic Data + +Refer to Note 13 of the Notes to the Condensed Consolidated Financial Statements for disclosure regarding segment information. + +## Critical Accounting Policies and Estimates + +Refer to Part II, Item 7, "Critical Accounting Policies and Estimates" of our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. There have been no material changes to our Critical Accounting Policies and Estimates. + +## Results of Operations + +The following table sets forth, for the periods indicated, certain items in our Condensed Consolidated Statements of Income expressed as a percentage of revenue. + +
Three Months Ended
Apr 26, 2026Apr 27, 2025
Revenue100.0 %100.0 %
Cost of revenue25.139.5
Gross profit74.960.5
Operating expenses
Research and development7.79.1
Sales, general and administrative1.62.4
Total operating expenses9.311.5
Operating income65.649.0
Interest income0.71.2
Interest expense(0.1)(0.1)
Other income (expense), net19.5(0.4)
Total other income, net20.10.7
Income before income tax85.749.7
Income tax expense14.27.1
Net income71.5 %42.6 %
+ +## Reportable Segments + +## Revenue by Reportable Segments + +
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change% Change
($ in millions)
Compute & Networking$ 74,550$ 39,589$ 34,96188 %
Graphics7,0654,4732,59258 %
Total$ 81,615$ 44,062$ 37,55385 %
+ +## Operating Income by Reportable Segments + +
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change% Change
($ in millions)
Compute & Networking$ 53,335$ 22,054$ 31,281142 %
Graphics2,9411,6401,30179 %
Total$ 56,276$ 23,694$ 32,582138 %
+ +Compute & Networking revenue – The year-over-year increase in the first quarter of fiscal year 2027 was due to growth in Data Center products, driven by the ramp of our Blackwell systems and demand for our InfiniBand, Spectrum-X Ethernet, and NVLink solutions. + +Graphics revenue – The year-over-year increase in the first quarter of fiscal year 2027 was driven by sales of our Blackwell architecture. + +Reportable segment operating income – The year-over-year increase in Compute & Networking segment operating income in the first quarter of fiscal year 2027 was driven by the growth in revenue and the non-recurrence of a \$4.5 billion charge associated with H2O excess inventory and purchase obligations in the first quarter of fiscal year 2026. The year-over-year increase in Graphics segment operating income in the first quarter of fiscal year 2027 was driven by the growth in revenue. + +## Concentration of Revenue + +We refer to customers who purchase products directly from NVIDIA as direct customers, such as AIBs, distributors, ODMs, OEMs, CSPs, AI model makers, and system integrators. Certain direct customers may use either internal resources or third-party system integrators to complete their build. We refer to indirect customers as those who purchase products through our direct customers; indirect customers include CSPs, AI Clouds, AI model makers, enterprises, and public sector entities. Our revenue is concentrated among a limited number of direct and indirect customers and this trend may continue. + +Direct Customers – For the first quarter of fiscal year 2027, three direct customers represented 21%, 17%, and 16% of total revenue, all of which was primarily attributable to the Compute & Networking segment. + +For the first quarter of fiscal year 2026, sales to two direct customers represented 16% and 14% of total revenue, which were attributable to the Compute & Networking segment. + +Indirect Customers – Indirect customer revenue is an estimation based upon multiple factors including customer purchase order information, product specifications, internal sales data, and other sources. Indirect customers primarily purchase our products through system integrators and distributors. + +We generate a significant amount of our revenue from a limited number of indirect customers, some individually representing 10% or more of our revenue. Certain companies purchase cloud and related services through various direct and indirect customers. We estimate that one AI research and deployment company contributed to a meaningful amount of our revenue by purchasing cloud services from our customers in the first quarter of fiscal year 2027. + +Revenue by geographic region is designated based on the location of the headquarters of direct customers. The end customer and shipping location may be different from our customers' headquarters location. Revenue from sales to customers headquartered outside of the United States accounted for 22% of total revenue for the first quarter of fiscal year 2027 and 42% of total revenue for the first quarter of fiscal year 2026. + +## Gross Profit and Gross Margin + +Gross profit consists of total net revenue less cost of revenue. Cost of revenue consists primarily of the cost of semiconductors, including wafer fabrication, assembly, testing and packaging, board and device costs, manufacturing support costs, including labor and overhead associated with such purchases, final test yield fallout, inventory and warranty provisions, memory and component costs, tariffs, and shipping costs. Cost of revenue also includes acquisition-related intangible amortization expense, IP-related costs, and stock-based compensation related to personnel associated with manufacturing operations. + +Gross margin increased to 74.9% for the first quarter of fiscal year 2027 compared to 60.5% for the first quarter of fiscal year 2026, primarily due to the prior year's \$4.5 billion charge associated with H2O excess inventory and purchase obligations. + +Provisions for inventory and excess inventory purchase obligations totaled \$1.1 billion and \$5.3 billion for the first quarter of fiscal years 2027 and 2026, respectively. The first quarter of fiscal year 2026 includes \$4.5 billion associated with H2O excess inventory and purchase obligations. Sales of previously reserved inventory and settlements of excess inventory purchase obligations resulted in a provision release of \$103 million and \$436 million for the first quarter of fiscal years 2027 and 2026, respectively. The net effect on our gross margin was an unfavorable impact of 1.2% and 11.0% in the first quarter of fiscal years 2027 and 2026, respectively. + +Operating Expenses + +
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change% Change
($ in millions)
Research and development$ 6,321$ 3,989$ 2,33258 %
Sales, general and administrative1,3001,04125925 %
Total operating expenses$ 7,621$ 5,030$ 2,59152 %
+ +The increase in research and development expenses for the first quarter of fiscal year 2027 was primarily driven by a 112% increase in compute and infrastructure, a 31% increase in compensation and benefits, including stock-based compensation, reflecting employee growth and compensation increases, and a 204% increase in engineering development materials for new product introductions. + +The increase in sales, general and administrative expenses for the first quarter of fiscal year 2027 was primarily driven by compensation and benefits, including stock-based compensation, reflecting employee growth and compensation increases. + +Total Other Income, Net + +
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change
($ in millions)
Interest income$ 540$ 515$ 25
Interest expense(102)(63)(39)
Other income (expense), net15,929(180)16,109
Total other income, net$ 16,367$ 272$ 16,095
+ +Total other income, net primarily consists of realized or unrealized gains and losses from investments in non-marketable securities and publicly-held equity securities. The change in Other income (expense), net compared to the first quarter of fiscal year 2026, was primarily driven by unrealized gains on investments in publicly-held equity securities of \$13.4 billion and non-marketable equity securities of \$2.6 billion. + +## Income Taxes + +Income tax expense was \$11.6 billion and \$3.1 billion for the first quarter of fiscal years 2027 and 2026, respectively. Income tax as a percentage of income before income tax was 16.6% and 14.3% for the first quarter of fiscal years 2027 and 2026, respectively. + +The effective tax rate increased primarily due to a lower percentage of tax benefits from stock-based compensation relative to the increase in income before income tax. + +Our effective tax rates for the first quarter of fiscal years 2027 and 2026 were lower than the U.S. federal statutory rate of 21% primarily due to tax benefits from foreign-derived deduction eligible income, income earned in jurisdictions that were subject to taxes at rates lower than the U.S. federal statutory tax rate, stock-based compensation, and the U.S. federal research tax credit. + +Refer to Note 11 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q for additional information. + +Liquidity and Capital Resources + +
Apr 26, 2026Jan 25, 2026
(In millions)
Cash and cash equivalents$ 13,237$ 10,605
Marketable debt securities37,09839,065
Cash, cash equivalents, and marketable debt securities$ 50,335$ 49,670
+ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Net cash provided by operating activities$ 50,344$ 27,414
Net cash used in investing activities$ (26,429)$ (5,216)
Net cash used in financing activities$ (21,283)$ (15,553)
+ +Our fixed income security investments include highly rated, diversified investment types and credit exposures with shorter maturities. + +Cash provided by operating activities increased in the first quarter of fiscal year 2027 compared to the first quarter of fiscal year 2026 due to higher revenue. + +Cash used in investing activities increased in the first quarter of fiscal year 2027 compared to the first quarter of fiscal year 2026, primarily driven by higher purchases of equity investment securities. + +Cash used in financing activities increased in the first quarter of fiscal year 2027 compared to the first quarter of fiscal year 2026, mainly due to higher share repurchases. + +## Liquidity + +Our primary sources of liquidity include cash, cash equivalents, marketable debt and equity securities, and cash generated by our operations. As of April 26, 2026, we had \$50.3 billion in cash, cash equivalents, and marketable debt securities as well as \$30.2 billion of marketable equity securities. We believe that we have sufficient liquidity to meet our operating requirements for at least the next twelve months and for the foreseeable future, including our future obligations. We continuously evaluate our liquidity and capital resources, including our access to external capital, to ensure we can finance future capital requirements and commitments. + +Our marketable securities consist of publicly-held equity securities, debt securities issued by the U.S. government and its agencies, highly-rated corporations and financial institutions, and foreign government entities, as well as certificates of deposit issued by highly-rated financial institutions. These marketable securities are primarily denominated in U.S. dollars. Refer to Note 5 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q for additional information. + +Except for approximately \$1.7 billion of cash, cash equivalents, and marketable debt securities held outside of the U.S., for which we have not accrued any related foreign or state taxes if we repatriate these amounts to the U.S., substantially all of our cash, cash equivalents, and marketable debt securities held outside the U.S. at the end of the first quarter of fiscal year 2027 are available for use in the U.S. without incurring additional U.S. federal income taxes. We made no federal income tax payments in the first quarter of fiscal year 2027, whereas our second quarter of fiscal year 2027 is scheduled to include two payments. + +## Capital Return to Shareholders + +In the first quarter of fiscal year 2027, we repurchased 108 million shares of our common stock for \$20.2 billion. As of April 26, 2026, we were authorized, subject to certain specifications, to repurchase up to \$38.5 billion of our common stock. + +On May 18, 2026, our Board of Directors approved an additional \$80.0 billion in share repurchase authorization, without expiration. + +We may execute repurchases from time to time, subject to market conditions, operating requirements and other investment opportunities, in the open market, in privately-negotiated transactions, pursuant to a Rule 10b5-1 trading plan or in structured share repurchase agreements in compliance with Rule 10b-18 of the Exchange Act. Our share repurchase program may be suspended at any time at our discretion. + +We paid cash dividends to our shareholders of \$243 million during the first quarter of fiscal year 2027. On May 18, 2026, we increased our quarterly cash dividend from \$0.01 per share to \$0.25 per share to all shareholders of record on June 4, 2026. Our quarterly cash dividend will be paid on June 26, 2026. + +The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders. + +The U.S. Inflation Reduction Act of 2022 requires a 1% excise tax on certain share repurchases in excess of shares issued for employee compensation made after December 31, 2022. The excise tax is included in our share repurchase cost and was not significant for the first quarter of fiscal year 2027. + +## Outstanding Indebtedness and Commercial Paper Program + +Our aggregate debt maturities as of April 26, 2026, by year payable, were as follows: + +
Apr 26, 2026
(In millions)
Due in one year$ 1,000
Due in one to five years2,750
Due in five to ten years1,250
Due in greater than ten years3,500
Unamortized debt discount and issuance costs(30)
Net carrying amount$ 8,470
Less short-term portion(1,000)
Total long-term portion$ 7,470
+ +We have a commercial paper program to support general corporate purposes, pursuant to which we may issue unsecured paper notes, from time to time or all at once, up to \$25.0 billion. As of April 26, 2026, no commercial paper was outstanding. + +Refer to Note 9 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q for further discussion. + +## Material Cash Requirements and Other Obligations + +For descriptions of our facility lease guarantees, long-term debt, purchase commitments, and operating lease obligations, refer to Note 8, Note 9, Note 10, and Note 14 of the Notes to the Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q, respectively. + +We expect to continue investing in our ecosystem. Refer to Note 6 and Item 1A. Risk Factors for additional information regarding our investments. + +Unrecognized tax benefits were \$4.5 billion, which includes related interest and penalties of \$439 million, and were recorded in non-current income tax payable as of April 26, 2026. We are unable to estimate the timing of any potential tax liability, interest payments, or penalties in individual years due to uncertainties in the underlying income tax positions and the timing of the effective settlement of such tax positions. We are currently under examination by the Internal Revenue Service for our fiscal years 2023 and 2024. Refer to Note 11 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q. + +Other than the contractual obligations described in Notes 6 and 10, there were no material changes outside the ordinary course of business in our contractual obligations from those disclosed in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. Refer to Item 7, “Management’s Discussion and Analysis of Financial Condition and Results of Operations - Liquidity and Capital Resources” in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026 for a description of our contractual obligations. For a description of our facility lease guarantees, long-term debt, purchase obligations, and operating lease obligations, refer to Notes 8, 9, 10, and 14 of the Notes to the Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q, respectively. + +## Adoption of New and Recently Issued Accounting Pronouncements + +There has been no adoption of any new and recently issued accounting pronouncements. + +## Item 3. Quantitative and Qualitative Disclosures About Market Risk + +## Investment and Interest Rate Risk + +Financial market risks related to investment and interest rate risk are described in Part II, Item 7A, “Quantitative and Qualitative Disclosures About Market Risk” in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. Our marketable equity securities consist of publicly-held equity securities, while our non-marketable equity securities are investments in privately-held companies. Publicly-held equity securities are subject to market price volatility. A hypothetical 10% decrease in our publicly-held equity securities would decrease the fair value of the publicly-held equity securities balance by \$3.9 billion and \$1.8 billion as of April 26, 2026 and January 25, 2026, respectively. + +Non-marketable equity securities are measured based on cost minus impairment, if any, and are adjusted for observable price changes in orderly transactions for an identical or similar investment in the same issuer. Valuations of our non- + +marketable equity securities are inherently complex due to the lack of readily available market data and observable transactions, and impact of macroeconomic factors. + +For a description of our equity investments, refer to Notes 5 and 6 of the Notes to Condensed Consolidated Financial Statements in Part I, Item 1 of this Quarterly Report on Form 10-Q, respectively. + +## Foreign Exchange Rate Risk + +The impact of foreign currency transactions related to foreign exchange rate risk is described in Part II, Item 7A, “Quantitative and Qualitative Disclosures About Market Risk” in our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. As of April 26, 2026, there have been no material changes to the foreign exchange rate risks described as of January 25, 2026. + +## Item 4. Controls and Procedures + +## Controls and Procedures + +## Disclosure Controls and Procedures + +Based on their evaluation as of April 26, 2026, our management, including our Chief Executive Officer and Chief Financial Officer, has concluded that our disclosure controls and procedures (as defined in Rule 13a-15(e) under the Exchange Act) were effective to provide reasonable assurance that the information we are required to disclose in reports that we file or submit under the Exchange Act is recorded, processed, summarized and reported within the time periods specified in the SEC rules and forms, and that such information is accumulated and communicated to our management, including our Chief Executive Officer and our Chief Financial Officer, as appropriate, to allow timely decisions regarding required disclosures. + +## Changes in Internal Control Over Financial Reporting + +There have been no changes in our internal control over financial reporting during the first quarter of fiscal year 2027 that have materially affected, or are reasonably likely to materially affect, our internal control over financial reporting. We are continuing a phased upgrade of our enterprise resource planning, or ERP, system to update our existing core financial systems. The ERP system is designed to accurately maintain our financial records used to report operating results. We will continue to evaluate each quarter whether there are changes that materially affect our internal control over financial reporting. + +## Inherent Limitations on Effectiveness of Controls + +Our management, including our Chief Executive Officer and Chief Financial Officer, does not expect that our disclosure controls and procedures, or our internal controls, will prevent all error and all fraud. A control system, no matter how well conceived and operated, can provide only reasonable, not absolute, assurance that the objectives of the control system are met. Further, the design of a control system must reflect the fact that there are resource constraints, and the benefits of controls must be considered relative to their costs. Because of the inherent limitations in all control systems, no evaluation of controls can provide absolute assurance that all control issues and instances of fraud, if any, within NVIDIA have been detected. + +## Part II. Other Information + +## Item 1. Legal Proceedings + +Refer to Part I, Item 1, Note 10 of the Notes to the Condensed Consolidated Financial Statements for a discussion of significant developments in our legal proceedings since January 25, 2026. + +## Item 1A. Risk Factors + +Other than the risk factors listed below, there have been no material changes from the risk factors previously described under Item 1A of our Annual Report on Form 10-K for the fiscal year ended January 25, 2026. + +Purchasing or owning NVIDIA common stock involves investment risks including, but not limited to, the risks described in Item 1A of our Annual Report on Form 10-K for the fiscal year ended January 25, 2026, and below. Any one of those risks could harm our business, financial condition and results of operations or reputation, which could cause our stock price to decline. Additional risks, trends and uncertainties not presently known to us or that we currently believe are immaterial may also harm our business, financial condition, results of operations or reputation. + +## Competition could adversely impact our market share and financial results. + +Our target markets remain competitive, and competition may intensify with expanding and changing product and service offerings, industry standards, customer and market needs, new entrants and consolidations. Other companies compete + +with us on a wide range of parameters including price, total cost of ownership, and performance, which has resulted and may in the future result in lower-than-expected selling prices or demand for our products. Some of our competitors operate their own fabrication facilities, and have longer operating histories, larger customer bases, more comprehensive IP portfolios and patent protections, more design wins, and greater financial, sales, marketing and distribution resources than we do. These competitors may be able to acquire market share and/or prevent us from doing so, more effectively identify and capitalize upon opportunities in new markets and end-user trends, more quickly transition their products, and impinge on our ability to procure sufficient foundry capacity and scarce input materials during a supply-constrained environment, which could harm our business. Some of our customers are developing their own ASICs and other products, including designs optimized for certain workloads that may not require all of the features and functionality our data center systems provide. Others may offer cloud-based services that compete with our AI cloud service offerings, and we may not be able to establish market share sufficient to achieve the scale necessary to meet our business objectives. If we are unable to successfully compete in this environment, demand for our products, services, and technologies could decrease, which may negatively impact our business. + +Long manufacturing lead times and uncertain supply and capacity availability, combined with a failure to estimate customer demand accurately, has led and could in the future lead to mismatches between supply and demand. + +We have long manufacturing lead times and build finished products and maintain inventory in advance of anticipated demand. In periods of shortages impacting the semiconductor industry and/or limited supply or capacity in our supply chain, the lead times for certain supply may be extended. We have previously experienced and may continue to experience extended lead times of more than 12 months. To secure future supply and capacity, we have paid premiums, provided deposits, and entered into long-term supply agreements and capacity commitments, which have increased our product costs and this may continue. We may still be unable to secure sufficient commitments for capacity to address our business needs. + +If we inaccurately estimate demand, or our customers change orders, as we have experienced in the past, we may not be able to reduce our supply commitments in time, at the same rate, or at all. Significant mismatches between supply and demand have varied across our market platforms, resulted in both product shortages and excess inventory, significantly harmed our financial results and could reoccur. If we underestimate demand, and our foundry partners and contract manufacturers are unable to increase production or provide sufficient supply, we may not be able to meet increased customer demand in a timely manner, or at all. Our reputation and customer relationships could be damaged and we could lose revenue and market share. Additionally, since some of our products are part of a complex data center buildout, supply constraints or availability issues with respect to any one component have had and may have a broader revenue impact. For example, our ability to sell certain products has been and could be impeded if components necessary for the finished products are not available from third parties. + +If we overestimate demand, or if customers cancel or defer orders or choose to purchase from our competitors, we may not be able to utilize on-hand inventory or reduce purchase commitments accordingly. We have had to reduce average selling prices, including due to our channel pricing programs, increase prices for certain of our products as a result of our suppliers' increase in prices, write down our inventory, incur cancellation penalties, and record impairments, and may have to do so in the future. The impact of these risks would be amplified by our non-cancellable and non-returnable purchase orders placed in advance of our historical lead times and could be exacerbated if we need to make changes to the design of future products. These risks have increased and may continue to increase as our purchase obligations and prepaids have grown and are expected to continue to grow and become a greater portion of our total supply. All of these factors may negatively impact our gross margins and financial results. + +Factors that have caused and/or could in the future cause us to underestimate or overestimate demand, and impact the timing and volume of our revenue, include: + +• changes in product development cycles and time to market; +• competing technologies and competitor product releases, announcements or other actions; +• changes in business and economic conditions; +- sudden or sustained government lockdowns or public health issues; +• rapidly changing technology or customer requirements; +• the availability of sufficient data center capacity or energy for customers to procure; +- new product introductions and transitions resulting in less demand for existing products; +- new or unexpected end-use cases; +• increase in demand for competitive products; +• changes in end-user demand; + +• purchasing decisions made, and inventory levels held by, distributors, ODMs, OEMs, system integrators, other channel partners and other third parties; +- the ability of developers, end customers and other third parties to build, enhance, and maintain accelerated computing applications that leverage our platforms; +• the demand for accelerated computing, AI-related cloud services, or large language models; +• changes that impact the ecosystem for the architectures underlying our products and technologies; +• government actions or changes in governmental policies, such as export controls, increased restrictions on gaming usage, or tariffs; +• our customers' and partners' ability to secure capital and energy and to build complex data center infrastructure timely; and +• the availability of third-party content on our platforms, such as GeForce NOW. + +The availability of data centers, energy, and capital to support the buildout of NVIDIA AI infrastructure by our customers and partners is crucial, and any shortage of these and other necessary resources could impact our future revenue and financial performance. Expanding energy capacity to meet demand is a complex, multi-year process involving significant regulatory, technical, and construction challenges. In addition, access to capital can be particularly constrained for less-capitalized companies, which may face difficulties securing financing for large-scale infrastructure projects. These limitations could delay customer and partner deployments or reduce the scale of accelerated computing and AI adoption. + +Challenges in estimating demand could become more pronounced or volatile in the future on both a global and regional basis. Extended lead times may occur if we experience other supply constraints caused by natural disasters, pandemics or other events. Geopolitical tensions in regions where we rely on suppliers, contract manufacturers, and assembly partners that are critical to our supply continuity, could have a material adverse impact on us. Supply availability affecting memory, and other components, as well as rising prices, may drive the prices for data center buildouts higher. Publicly announced intentions by governments or other companies to purchase our products can further complicate our demand estimates, as such announcements are often non-binding and may not result in committed volumes. + +We continue to increase our supply and capacity purchases with existing and new suppliers to support our demand projections and increasing complexity of our data center products. We have also entered and may continue to enter into prepaid manufacturing and capacity agreements to supply both current and future products. The increased purchase volumes and integration of new suppliers and contract manufacturers into our supply chain creates more complexity in managing multiple suppliers with variations in production planning, execution and logistics. Our expanding product portfolio and varying component compatibility and quality may lead to increased inventory levels. We have incurred and may in the future incur inventory provisions or impairments if our inventory or supply or capacity commitments exceed demand for our products or demand declines. We are increasing our U.S.-based manufacturing and investing in specialized equipment and processes to support domestic production. We may experience delays or difficulties in scaling production as planned. Our ability to increase manufacturing capabilities will depend on the domestic manufacturing ecosystem's capacity to ramp production supply to the required volume timely. Delays or shortfalls could impact our ability to meet demand. + +Introducing or offering multiple architectures concurrently is complex and we often ship multiple architecture products simultaneously as our channel partners prepare to ship and support new products. We are generally in various stages of introducing and/or offering the architectures of our Data Center and Edge Computing products. The computing industry is experiencing a broader and faster launch cadence of accelerated computing platforms to meet a growing and diverse set of AI opportunities. We have introduced a new product and architecture cadence of our Data Center solutions where we seek to complete new computing solutions each year and provide a greater variety of Data Center offerings, including our Rubin platform which is expected to start shipping in the second half of fiscal year 2027. The increased frequency of these architecture introductions and the larger number of products and product configurations may magnify the challenges associated with managing our supply and demand which may further create volatility in our revenue. Qualification time for new products, customers anticipating new architecture introductions, and channel partners reducing channel inventory of prior architectures ahead of new product introductions can reduce, or create volatility in, our revenue. Customers may delay adopting new architectures if their data center infrastructure is not ready, which could affect the timing of our revenue. We have experienced and may in the future experience reduced demand for current generation architectures when customers anticipate new architectures, and we may be unable to sell multiple product architectures at the same time. Our financial results have been and may in the future be negatively impacted if we are unable to execute our architectural introductions as planned for any reason. The increased frequency and complexity of newly introduced products could result in unanticipated quality or production issues that could increase the magnitude of inventory provisions, warranty, or other costs or result in product delays. For example, our gross margins in the second quarter of fiscal year 2025 were negatively impacted by inventory provisions for low-yielding Blackwell material. + +We incur significant engineering development resources for new products, and changes to our product roadmap may impact our ability to develop other products or adequately manage our supply chain cost. Customers may delay purchasing existing products as we increase the frequency of new products or may not be able to adopt our new products as fast as forecasted, both impacting the timing of our revenue and supply chain cost. While we have managed concurrent architecture introductions and/or offerings and have sold multiple product architectures at the same time, these efforts are difficult, may impair our ability to predict demand and impact our supply mix, and may cause us to incur additional costs. + +Demand estimates for our products, applications, and services can be incorrect, which may create volatility in our revenue or supply levels. We may not be able to generate significant revenue from them. Because our products may be used in multiple use cases and applications, it is difficult to estimate with any reasonable degree of precision the impact of accelerated computing and AI models on our reported revenue or forecasted demand. + +The use of our GPUs for new, mercurial, or trendy applications, has impacted, and can impact in the future, demand for our products, including by leading to inconsistent spikes and drops in demand. For example, several years ago, our gaming GPUs began to be used for mining digital currencies, such as Ethereum. It is difficult for us to estimate with any reasonable degree of precision the past or current impact of cryptocurrency mining, or forecast the future impact of cryptocurrency mining, on demand for our products. Volatility in the cryptocurrency market, including new compute technologies, price changes in cryptocurrencies, government cryptocurrency policies and regulations, new cryptocurrency standards and changes in the method of verifying blockchain transactions, has impacted and can in the future impact cryptocurrency mining and demand for our products and can further impact our ability to estimate demand for our products. Changes to cryptocurrency standards and processes including, but not limited to, the Ethereum 2.0 merge in 2022, have reduced and may in the future decrease the usage of GPUs for Ethereum mining. This has created and may in the future create increased aftermarket sales of our GPUs, which could negatively impact retail prices for our GPUs and reduce demand for our new GPUs. In general, our new products or previously sold products may be resold online or on the unauthorized “gray market,” which also makes demand forecasting difficult. Gray market products and reseller marketplaces compete with our new products and distribution channels. Our inability to accurately predict our demand that arises from new use cases may create volatility in our revenue. + +We are subject to complex laws, rules, regulations, and political and other actions, including restrictions on the export of our products, which may adversely impact our business. + +We are subject to laws and regulations domestically and worldwide, affecting our operations in areas including, but not limited to, IP ownership and infringement; taxes; import and export requirements and tariffs; anti-corruption, including the Foreign Corrupt Practices Act; business acquisitions; foreign exchange controls and cash repatriation restrictions; foreign ownership and investment; data privacy requirements; competition and antitrust; investing; advertising; employment; product regulations; cybersecurity; environmental, health, and safety requirements; the responsible use of AI; sustainability; cryptocurrency; and consumer laws. Compliance with such requirements can be onerous and expensive, could impact our competitive position, and may negatively impact our business operations and ability to manufacture and ship our products. There can be no assurance that our employees, contractors, suppliers, customers or agents will not violate applicable laws or the policies, controls, and procedures that we have designed to help ensure compliance with such laws, and violations could result in fines, criminal sanctions against us, our officers, or our employees, prohibitions on the conduct of our business, and damage to our reputation. Changes to the laws, rules and regulations to which we are subject, or changes to their interpretation and enforcement, could lead to materially greater compliance and other costs, and/or further restrictions on our ability to manufacture and supply our products and operate our business. For example, we may face increased compliance costs as a result of changes or increases in antitrust legislation, regulation, administrative rule making, increased focus from regulators on cybersecurity vulnerabilities and risks. Our position in markets relating to AI has led to increased interest in our business from regulators worldwide, including the European Union, the United States, the United Kingdom, South Korea, Japan, and China. For example, the French Competition Authority collected information from us regarding our business and competition in the graphics card and CSP market as part of an ongoing inquiry into competition in those markets. We have also received, and continue to receive, broad requests for information from competition regulators in the European Union, the United States, the United Kingdom, China, and South Korea regarding our sales of GPUs and other NVIDIA products, our efforts to allocate supply, foundation models and our investments, partnerships and other agreements with companies developing foundation models, the markets in which we compete and our competition, our strategies, roadmaps, and efforts to develop, market, and sell hardware, software, and system solutions, and our agreements with customers, suppliers, and partners. We expect to receive additional requests for information in the future. Such requests have been and are likely to be expensive and burdensome and could negatively impact our business and our relationships with customers, suppliers, and partners. + +Governments and regulators are also considering, and in certain cases, have imposed restrictions on the hardware, software, and systems used to develop frontier foundation models and generative AI. For example, the EU AI Act became effective on August 1, 2024 and will be fully applicable after a two-year transitional period. The EU AI Act may impact our ability to train, deploy, or release AI models in the EU. Several states are considering enacting or have already enacted regulations concerning AI technologies, with new state laws that took effect on January 1, 2026, which may impact our ability to train, deploy, or release AI models, and increase our compliance costs. Restrictions under these and any other regulations, if implemented, could increase the costs and burdens to us and our customers, delay or halt deployment of + +new systems using our products, and reduce the number of new entrants and customers, negatively impacting our business and financial results. Revisions to laws or regulations or their interpretation and enforcement could also result in increased taxation, trade sanctions, the imposition of or increase to import duties or tariffs, restrictions and controls on imports or exports, or other retaliatory actions, which could have an adverse effect on our business plans or impact the timing of our shipments. Additionally, changes in the public perception of governments in the regions where we operate or plan to operate could negatively impact our business and results of operations. + +Government actions, including trade protection and national and economic security policies of U.S. and foreign government bodies, such as tariffs, import or export regulations, including deemed export restrictions and restrictions on the activities of U.S. persons, trade and economic sanctions, decrees, quotas or other trade barriers and restrictions could affect our ability to ship products, provide services to our customers and employees, do business without an export license with entities on the U.S. Department of Commerce's U.S. Entity List or other USG restricted parties lists (which is expected to change from time to time), and generally fulfill our contractual obligations and have a material adverse effect on our business. If we were ever found to have violated export control laws or sanctions of the U.S. or similar applicable non-U.S. laws, even if the violation occurred without our knowledge, we may be subject to various penalties available under the laws, any of which could have a material and adverse impact on our business, operating results and financial condition. + +For example, in response to the war in Ukraine, the United States and other jurisdictions imposed economic sanctions and export control measures which blocked the passage of our products, services and support into Russia, Belarus, and certain regions of Ukraine. In fiscal year 2023, we stopped direct sales to Russia and closed business operations in Russia. Concurrently, the war in Ukraine has impacted sales in EMEA and may continue to do so in the future. + +The increasing focus on the risks and strategic importance of AI technologies has resulted in regulatory restrictions that target products and services capable of enabling or facilitating AI and may in the future result in additional restrictions impacting some or all of our product and service offerings. + +Concerns regarding third-party use of AI for purposes contrary to local governmental interests, including concerns relating to the misuse of AI applications, models, and solutions, has resulted in and could in the future result in unilateral or multilateral restrictions on products that can be used for training, modifying, tuning, and deploying LLMs and other AI applications. Such restrictions have limited and could in the future limit the ability of downstream customers and users worldwide to acquire, deploy and use systems that include our products, software, and services, and negatively impact our business and financial results. + +Such restrictions could include additional unilateral or multilateral export controls on certain products or technology, including but not limited to AI technologies. As geopolitical tensions have increased, semiconductors associated with AI, including GPUs and related products, are increasingly the focus of export control restrictions proposed by stakeholders in the U.S. and its allies. The United States has imposed unilateral worldwide controls restricting GPUs and associated products, and it is likely that additional unilateral or multilateral controls will be adopted. Such controls have been and may again be very broad in scope and application, prohibit us from exporting our products to any or all customers in one or more markets, and could negatively impact our manufacturing, testing and warehousing locations and options, or could impose other conditions that limit our ability to serve demand abroad and could negatively and materially impact our business, revenue and financial results. Export controls and other restrictions targeting GPUs and semiconductors associated with AI, which have been imposed and are likely to be more restrictive, would further limit our ability to export our technology, products, or services, creating a competitive disadvantage for us and negatively impacting our business and financial results. Export controls targeting GPUs and semiconductors associated with AI have subjected and may in the future subject downstream users of our products to restrictions on the use, resale, repair, or transfer of our products, negatively impacting our business and financial results. Controls could negatively impact our cost and/or ability to provide services such as NVIDIA AI cloud services and could impact the cost and/or ability for our CSPs and customers to provide services to their end customers, even outside China. + +Export controls have and could in the future disrupt our supply chain and distribution channels, negatively impacting our ability to serve demand, including in markets outside China and for our non-data center products. The possibility of additional export controls has negatively impacted and may in the future negatively impact demand for our products, benefiting competitors that offer alternatives less likely to be restricted by further controls. Repeated changes in the export control rules are likely to impose compliance burdens on our business and our customers, negatively and materially impacting our business. + +Increasing use of economic sanctions and export controls has impacted and may in the future impact demand for our products or services, negatively impacting our business and financial results. Reduced demand due to export controls has and could in the future lead to excess inventory or cause us to incur related supply charges. Additional unilateral or multilateral controls are also likely to include deemed export control limitations that negatively impact the ability of our research and development teams to execute our roadmap or other objectives in a timely manner. Additional export restrictions may not only impact our ability to serve overseas markets, but also provoke responses from foreign governments, including China, that negatively impact our supply chain or our ability to provide our products and services to customers in all markets worldwide, which could also substantially reduce our revenue. Regulators in China have + +inquired about our sales and efforts to supply the China market and our fulfillment of the commitments we entered into at the close of our Mellanox acquisition. On September 15, 2025, China's antitrust regulators published their preliminary finding that our compliance with applicable U.S. export controls, which required us to offer degraded products to the Chinese market, discriminated unfairly against customers in the China market and therefore violated the terms of China's approval of our Mellanox acquisition. If regulators conclude that we have failed to fulfill the terms of our Mellanox acquisition or we have violated any applicable law in China, we could be subject to financial penalties, restrictions on our ability to conduct our business, restrictions or other orders regarding our networking business, products, and services, or otherwise impact our operations in China, any of which could have a material and adverse impact on our business, operating results and financial condition. + +We continue to be subject to a series of shifting and expanding export control restrictions, impacting our ability to serve customers outside the United States. + +In August 2022, the USG announced export restrictions and export licensing requirements targeting China's semiconductor and supercomputing industries. These restrictions impacted exports of certain chips, as well as software, hardware, equipment and technology used to develop, produce and manufacture certain chips to China (including Hong Kong and Macau) and Russia, and specifically impact our A100 and H100 integrated circuits, DGX or any other systems or boards which incorporate A100 or H100 integrated circuits. + +In July 2023, the USG also informed us of an additional licensing requirement for a subset of A100 and H100 products destined to certain customers and other regions, including some countries in the Middle East. + +In October 2023, the USG announced new and updated licensing requirements for exports to China and Country Groups D:1, D:4, and D:5 (including but not limited to, Saudi Arabia, the United Arab Emirates, and Vietnam, but excluding Israel) of our products exceeding certain performance thresholds, including, but not limited to, the A100, A800, H100, H800, L4, L40, L40S RTX 4090, GB200 NVL72, and B200. The licensing requirements also apply to the export of products exceeding certain performance thresholds to a party headquartered in, or with an ultimate parent headquartered in, Country Group D5, including China. + +In April 2025, the USG informed us that it requires a license for export to China (including Hong Kong and Macau) and D:5 countries, or to companies headquartered or with an ultimate parent therein, of our H20 integrated circuits and any other circuits achieving the H20's memory bandwidth, interconnect bandwidth, or combination thereof. As a result of these requirements, we incurred a \$4.5 billion charge in the first quarter of fiscal year 2026 associated with H20 for excess inventory and purchase obligations, as the demand for H20 products diminished. + +Beginning in August 2025, the USG granted licenses that would allow us to ship certain H2O products to certain China-based customers. USG officials expressed an expectation that the USG will receive $15\%$ or more of the revenue generated from licensed sales of our products, but the USG did not publish a regulation codifying such requirement. + +Beginning in February 2026, the USG granted licenses that would allow us to ship small amounts of H200 products to specific China-based customers. To date, we have not generated any revenue under the H200 licensing program, and do not yet know whether any imports will be allowed into China. The license requires that the H200s go through an inspection process in the United States prior to any shipment to the customer. As a result, any H200 shipped under the new licensing program will be subject to a 25% tariff upon importation into the United States. + +In the event that we are able to sell licensed products into the China market, we may not be able to pass along all or any of the tariff to our customers, and may be subject to litigation, increased costs, and a harmed competitive position. + +The export controls applicable to China are complex and address a variety of parameters, including the total processing performance of a chip, the “performance density” of a chip, the interconnect bandwidth of a chip, and the memory bandwidth of a chip. Under the current rules and geopolitical landscape, we are unable to create and deliver a competitive product for China’s data center market that receives approval from both the USG and the Chinese government. As of the end of the first quarter of fiscal year 2027, while we were able to ship uncontrolled products to China, such as gaming and workstation GPUs, we were effectively foreclosed from competing in China's data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide. Unless we are able to return with a data center system that meets the approval of both the USG and the Chinese government, our lost opportunity and the benefit to our competitors will have a material and adverse impact on our business, operating results, and financial condition. + +In addition to controls targeting D:1, D:4 and D:5 countries, the USG has also imposed worldwide export controls impacting our products, and may impose additional controls in the future. + +In January 2025, the USG published the AI Diffusion IFR in the Federal Register. The IFR would have imposed a worldwide licensing requirement on our data center products, such as our H200, GB200 and GB300. The AI Diffusion IFR would have divided the world into three tiers, relegating most countries to “Tier 2” status, and would have created a complex and burdensome scheme for licensing approvals. + +In May 2025, the USG announced that it would rescind the AI Diffusion IFR and implement a replacement rule. The scope, timing, and requirements of the forthcoming rule remain uncertain. The replacement rule may impose new restrictions on our products or operations and/or add license requirements that could have a material impact on our business, operating results, and financial condition. For example, in October 2025, the Senate passed the GAIN AI Act in the National Defense Authorization Act. The GAIN AI Act would restrict the Trump Administration's ability to adapt the Biden Administration's export control rules and could also allow private U.S. persons to review and overturn licensing and foreign policy decisions made by the Trump Administration. Congress is also considering legislation such as the Remote Access Security Act, or RASA, which could prohibit the provision of cloud services to any company with an ultimate parent headquartered in China. If enacted, RASA could impose new restrictions on cloud service providers and OEMs, and could have a material impact on our business, operating results, and financial condition. + +Our competitive position has been harmed by export controls, and our competitive position and future results will be further harmed, over the long term, if the restrictions remain in place or are expanded in geographic, customer, or product scope, if customers purchase product from competitors, if customers develop their own internal solution, if we are unable to provide contractual warranty or other extended service obligations, if the USG does not grant licenses in a timely manner or denies licenses to significant customers or if we incur significant transition costs. The licensing process may not be resolved before significant business opportunities evaporate. Even if the USG grants any requested licenses, the licenses have already and may in the future be temporary, impose burdensome conditions regarding the installation, maintenance, and use of such products, or include financial or economic requirements that we or our customers or end users cannot or choose not to fulfill. The licensing requirements have already and may in the future benefit certain of our competitors, as the licensing process will make our pre-sale and post-sale technical support efforts more cumbersome and less certain and encourage customers in China, the Middle East, and other regions to pursue alternatives to our products, including semiconductor suppliers based in China, Europe, and Israel. + +Given the increasing strategic importance of AI and rising geopolitical tensions, the USG has changed and may again change the export control rules at any time and further subject a wider range of our products to export restrictions and licensing requirements, negatively impacting our business and financial results. In the event of such change, we may be unable to sell our inventory of such products and may be unable to develop replacement products not subject to the licensing requirements. + +For example, the USG already imposed license conditions that limit the ability of foreign firms to create and offer as a service large-scale GPU clusters, such as imposing license conditions on the use of products to be exported to certain countries, and may impose additional conditions such as requiring chip tracking and throttling mechanisms that could disable or impair GPUs if certain events, including unauthorized system configuration, use, or location, are detected. Such government mandates in chip designs could introduce system vulnerabilities and expose us to significant risk and potential liability, negatively impact demand for our products, and could have a material impact on our business, operating results, and financial condition. Even if not enacted into binding legislation, draft bills have impacted and may in the future negatively impact our business. For example, following U.S. legislative proposals calling for mandatory features in our chips, China's government publicly questioned whether our H20 products have built-in vulnerabilities, discouraging customers from purchasing our products. We provided a public response explaining that our GPUs, including H20, do not include such built-in vulnerabilities, and will respond to any follow-up questions we receive. + +Open-source foundation models are rapidly growing in popularity with developers worldwide. The demand for open-source foundation models and applications promotes use of our products worldwide. Any regulatory control or other restriction that limits our ability to provide products and services that support third-party applications and models, including applications built on foundation models originating in China such as DeepSeek, Qwen, or KIMMI, could have a material impact on our business, operating results, and financial condition. + +The USG already imposed export controls restricting certain gaming GPUs, and if the USG expands such controls to restrict additional gaming products, it may disrupt a significant portion of our supply and distribution chain and negatively impact sales of such products to markets outside China, including the U.S. and Europe. For example, the French Competition Authority (FCA) is questioning whether gaming GPUs and data center GPUs are separate product categories, an inquiry that may impact the export controls applicable to gaming products sold in France and Europe. In addition, as the performance of the gaming GPUs increases over time, export controls may have a greater impact on our ability to compete in markets subject to those controls. Export controls may disrupt our supply and distribution chain for a substantial portion of our products, which are warehoused in and distributed from Hong Kong. + +Export controls restricting our ability to sell data center GPUs may also negatively impact demand for our networking products used in servers containing our GPUs. The USG may also impose export controls on our networking products, such as high-speed network interconnects, to limit the ability of downstream parties to create large clusters for frontier model training. + +Export controls have and are likely in the future to have a disproportionate impact on NVIDIA and may disadvantage us against certain of our competitors that sell chips that are outside the scope of such control. Export controls have already and may in the future encourage customers outside China and other impacted regions to “design-out” certain U.S. semiconductors from their products to reduce the compliance burden and risk, and to ensure that they are able to serve + +markets worldwide. Export controls have already encouraged and may in the future encourage overseas governments to request that our customers purchase from our competitors rather than NVIDIA or other U.S. firms, harming our business, market position, and financial results. + +As a result, export controls have in the past and may in the future negatively impact demand for our products and services not only in China, but also in other markets, such as Europe, Latin America, and Southeast Asia. Export controls increase the risk of investing in U.S. advanced semiconductor products, because by the time a new product is ready for market, it may be subject to new unilateral export controls restricting its sale, resulting in excess inventory and purchase obligations as we recently experienced with the H20. At the same time, such controls may increase investment in foreign competitors, which would be less likely to be restricted by U.S. controls. + +The increasingly complex export controls impose complex and burdensome compliance obligations on our partners, suppliers, and customers. We have provided and will continue to provide assistance to authorities regarding attempted diversion, but as we do not have physical control of our products after sale, we must also rely on the compliance programs of our customers and partners. While we seek to strictly comply with all applicable export control regulators, reports of diversion of controlled products, even when unsubstantiated and untrue, or any compliance failure at a customer or partner, may negatively impact our business, relationships with partners and customers, and our reputation. Incorrect allegations that our compliance efforts satisfy the letter but not the “spirit” of the applicable regulations, as well as incorrect allegations that legitimate and appropriate business is using supposed “loopholes” in the export controls may negatively impact our business, relationships with partners and customers, and our reputation. + +In addition to export controls, the USG may impose restrictions on the import and sale of products that incorporate technologies developed or manufactured in whole or in part in China. For example, the USG adopted “Connected Vehicle” restrictions on the import and sale of certain automotive products in the United States, which if adopted and interpreted broadly, could impact our ability to develop and supply solutions for our automotive customers. The USG is also considering restrictions that would limit our ability to support third-party applications and models built on open-source foundation models originating in China. Such restrictions, if implemented, would favor our foreign competitors and negatively impact our business. + +Additionally, restrictions imposed by the Chinese government on the duration of gaming activities and access to games may adversely affect our Edge Computing revenue, and even if we are able to participate in the China data center compute market, increased oversight of digital platform companies may adversely affect our Data Center revenue. The Chinese government has encouraged customers to purchase from our China-based competitors and discouraged customers from purchasing, importing, or using our data center products, including any China-specific product designed to comply with U.S. export controls. As another example, an agency of the Chinese government announced an Action Plan that endorses new standards regarding the compute performance per watt and per memory bandwidth of accelerators used in new and renovated data centers in China. Although we are already effectively foreclosed from the China market by U.S. export controls, if those controls changed to allow us to return to the market, the Chinese government could modify or implement the Action Plan in a way that effectively prevents us from being able to design products to meet the new standard, which may restrict the ability of customers to use some of our data center products and may have a material and adverse impact on our business, operating results and financial condition. Further restrictions on our products or the products of our suppliers could negatively impact our business and financial results. + +Finally, our business depends on our ability to receive consistent and reliable supply from our overseas partners, especially in Taiwan and South Korea. Any new restrictions that negatively impact our ability to receive supply of components, parts, or services from Taiwan and South Korea, would negatively impact our business and financial results. + +## Item 2. Unregistered Sales of Equity Securities and Use of Proceeds + +## Issuer Purchases of Equity Securities + +We repurchased 108 million shares of our common stock for \$20.2 billion during the first quarter of fiscal year 2027. As of April 26, 2026, we were authorized, subject to certain specifications, to repurchase up to \$38.5 billion of our common stock. + +On May 18, 2026, our Board of Directors approved an additional \$80.0 billion in share repurchase authorization, without expiration. + +We paid cash dividends to our shareholders of \$243 million during the first quarter of fiscal year 2027. On May 18, 2026, we increased our quarterly cash dividend from \$0.01 per share to \$0.25 per share to all shareholders of record on June 4, 2026. Our quarterly cash dividend will be paid on June 26, 2026. + +The payment of future cash dividends is subject to our Board of Directors' continuing determination that the declaration of dividends is in the best interests of our shareholders. + +The following table presents details of our share repurchase transactions during the first quarter of fiscal year 2027: + +
PeriodTotal Number of Shares Purchased (In millions)Average Price Paid per Share (1)Total Number of Shares Purchased as Part of Publicly Announced Program (In millions)Approximate Dollar Value of Shares that May Yet Be Purchased Under the Program (In billions)
January 26, 2026 - February 22, 20268.2$ 186.358.2$ 57.0
February 23, 2026 - March 22, 20269.0$ 183.389.0$ 55.4
March 23, 2026 - April 26, 202691.1$ 184.9891.1$ 38.5
Total108.3108.3
+ +(1) Average price paid per share includes broker commissions but excludes our liability under the 1% excise tax on the net amount of our share repurchases required by the Inflation Reduction Act of 2022. + +We may execute repurchases from time to time, subject to market conditions, operating requirements and other investment opportunities, in the open market, in privately-negotiated transactions, pursuant to a Rule 10b5-1 trading plan or in structured share repurchase agreements in compliance with Rule 10b-18 of the Exchange Act. Our share repurchase program may be suspended at any time at our discretion. + +## Employee Equity Incentive Program Share Withholding + +We withhold shares of our common stock associated with net share settlements to cover tax withholding obligations of awards under our employee equity incentive program. During the first quarter of fiscal year 2027, we withheld approximately 12 million shares, for a total value of \$2.1 billion through net share settlements. + +## Recent Sales of Unregistered Securities and Use of Proceeds + +On February 17, 2026, we acquired a company and issued to a key employee a total of 37,890 shares of our common stock, valued at approximately \$7 million based on our closing stock price on the issuance date. + +On April 10, 2026, we acquired a company and issued to key employees a total of 72,972 shares of our common stock, valued at approximately \$14 million based on our closing stock price on the issuance date. + +The above securities were issued in transactions not involving a public offering pursuant to an exemption from registration set forth in Section 4(a)(2) of the Securities Act (and Regulation D or Regulation S promulgated thereunder). + +## Item 5. Other Information + +The following members of our Board of Directors and/or officers adopted, modified, or terminated a trading arrangement that is intended to satisfy the affirmative defense conditions of Rule 10b5-1(c), or a Rule 10b5-1 Trading Arrangement: + +
NameTitle of Director or OfficerActionDateTotal Shares of Common Stock to be SoldExpiration Date
Tench CoxeDirectorAdoption3/19/20268,000,000*10/30/2027
Colette M. KressExecutive Vice President and Chief Financial OfficerTermination4/10/2026500,000**N/A
+ +\* The Rule 10b5-1 Trading Arrangement is solely for gifts to charitable donor-advised funds. +\*\* The Rule 10b5-1 Trading Arrangement was adopted on December 18, 2025, for sales through March 23, 2027. No shares were sold under the Rule 10b5-1 Trading Arrangement prior to termination. + +Item 6. Exhibits + +
Exhibit No.Exhibit DescriptionIncorporated by ReferenceFiling Date
Schedule/FormExhibit
10.1+Variable Compensation Plan - Fiscal Year 20278-K10.13/6/2026
31.1*Certification of Chief Executive Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
31.2*Certification of Chief Financial Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
32.1#*Certification of Chief Executive Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
32.2#*Certification of Chief Financial Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
101.INS*Inline XBRL Instance Document - the instance document does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document.
101.SCH*Inline XBRL Taxonomy Extension Schema Document
101.CAL*Inline XBRL Taxonomy Extension Calculation Linkbase Document
101.DEF*Inline XBRL Taxonomy Extension Definition Linkbase Document
101.LAB*Inline XBRL Taxonomy Extension Labels Linkbase Document
101.PRE*Inline XBRL Taxonomy Extension Presentation Linkbase Document
104Cover Page Interactive Data File - the cover page interactive data file does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document.
+ ++ Management contract or compensatory plan or arrangement. +\* Filed herewith. + +\# In accordance with Item 601(b)(32)(ii) of Regulation S-K and SEC Release Nos. 33-8238 and 34-47986, Final Rule: Management's Reports on Internal Control Over Financial Reporting and Certification of Disclosure in Exchange Act Periodic Reports, the certifications furnished in Exhibits 32.1 and 32.2 hereto are deemed to accompany this Quarterly Report on Form 10-Q and will not be deemed “filed” for purpose of Section 18 of the Exchange Act. Such certifications will not be deemed to be incorporated by reference into any filing under the Securities Act or the Exchange Act, except to the extent that the registrant specifically incorporates it by reference. + +Copies of the above exhibits not contained herein are available to any shareholder upon written request to: + +Investor Relations: NVIDIA Corporation, 2788 San Tomas Expressway, Santa Clara, CA 95051. + +## Signature + +Pursuant to the requirements of the Securities Exchange Act of 1934, the registrant has duly caused this report to be signed on its behalf by the undersigned thereunto duly authorized. + +Date: May 20, 2026 + +NVIDIA Corporation + +By: /s/ Colette M. Kress + +Colette M. Kress + +Executive Vice President and Chief Financial Officer (Duly Authorized Officer and Principal Financial Officer) + +## CERTIFICATION + +I, Jen-Hsun Huang, certify that: + +1. I have reviewed this Quarterly Report on Form 10-Q of NVIDIA Corporation; +2. Based on my knowledge, this report does not contain any untrue statement of a material fact or omit to state a material fact necessary to make the statements made, in light of the circumstances under which such statements were made, not misleading with respect to the period covered by this report; +3. Based on my knowledge, the financial statements, and other financial information included in this report, fairly present in all material respects the financial condition, results of operations and cash flows of the registrant as of, and for, the periods presented in this report; +4. The registrant's other certifying officer(s) and I are responsible for establishing and maintaining disclosure controls and procedures (as defined in Exchange Act Rules 13a-15(e) and 15d-15(e)) and internal control over financial reporting (as defined in Exchange Act Rules 13a-15(f) and 15d-15(f)) for the registrant and have: + +(a) designed such disclosure controls and procedures, or caused such disclosure controls and procedures to be designed under our supervision, to ensure that material information relating to the registrant, including its consolidated subsidiaries, is made known to us by others within those entities, particularly during the period in which this report is being prepared; +(b) designed such internal control over financial reporting, or caused such internal control over financial reporting to be designed under our supervision, to provide reasonable assurance regarding the reliability of financial reporting and the preparation of financial statements for external purposes in accordance with generally accepted accounting principles; +(c) evaluated the effectiveness of the registrant's disclosure controls and procedures and presented in this report our conclusions about the effectiveness of the disclosure controls and procedures, as of the end of the period covered by this report based on such evaluation; and +(d) disclosed in this report any change in the registrant's internal control over financial reporting that occurred during the registrant's most recent fiscal quarter (the registrant's fourth fiscal quarter in the case of an annual report) that has materially affected, or is reasonably likely to materially affect, the registrant's internal control over financial reporting; and +5. The registrant's other certifying officer(s) and I have disclosed, based on our most recent evaluation of internal control over financial reporting, to the registrant's auditors and the audit committee of the registrant's board of directors (or persons performing the equivalent functions): +(a) all significant deficiencies and material weaknesses in the design or operation of internal control over financial reporting which are reasonably likely to adversely affect the registrant's ability to record, process, summarize and report financial information; and +(b) any fraud, whether or not material, that involves management or other employees who have a significant role in the registrant's internal control over financial reporting. + +Date: May 20, 2026 + +/s/JEN-HSUN HUANG + +Jen-Hsun Huang + +President and Chief Executive Officer + +# CERTIFICATION + +I, Colette M. Kress, certify that: + +1. I have reviewed this Quarterly Report on Form 10-Q of NVIDIA Corporation; +2. Based on my knowledge, this report does not contain any untrue statement of a material fact or omit to state a material fact necessary to make the statements made, in light of the circumstances under which such statements were made, not misleading with respect to the period covered by this report; +3. Based on my knowledge, the financial statements, and other financial information included in this report, fairly present in all material respects the financial condition, results of operations and cash flows of the registrant as of, and for, the periods presented in this report; +4. The registrant's other certifying officer(s) and I are responsible for establishing and maintaining disclosure controls and procedures (as defined in Exchange Act Rules 13a-15(e) and 15d-15(e)) and internal control over financial reporting (as defined in Exchange Act Rules 13a-15(f) and 15d-15(f)) for the registrant and have: + +(a) designed such disclosure controls and procedures, or caused such disclosure controls and procedures to be designed under our supervision, to ensure that material information relating to the registrant, including its consolidated subsidiaries, is made known to us by others within those entities, particularly during the period in which this report is being prepared; +(b) designed such internal control over financial reporting, or caused such internal control over financial reporting to be designed under our supervision, to provide reasonable assurance regarding the reliability of financial reporting and the preparation of financial statements for external purposes in accordance with generally accepted accounting principles; +(c) evaluated the effectiveness of the registrant's disclosure controls and procedures and presented in this report our conclusions about the effectiveness of the disclosure controls and procedures, as of the end of the period covered by this report based on such evaluation; and +(d) disclosed in this report any change in the registrant's internal control over financial reporting that occurred during the registrant's most recent fiscal quarter (the registrant's fourth fiscal quarter in the case of an annual report) that has materially affected, or is reasonably likely to materially affect, the registrant's internal control over financial reporting; and +5. The registrant's other certifying officer(s) and I have disclosed, based on our most recent evaluation of internal control over financial reporting, to the registrant's auditors and the audit committee of the registrant's board of directors (or persons performing the equivalent functions): +(a) all significant deficiencies and material weaknesses in the design or operation of internal control over financial reporting which are reasonably likely to adversely affect the registrant's ability to record, process, summarize and report financial information; and +(b) any fraud, whether or not material, that involves management or other employees who have a significant role in the registrant's internal control over financial reporting. + +Date: May 20, 2026 + +/s/ COLETTE M. KRESS + +Colette M. Kress + +Executive Vice President and Chief Financial Officer + +## CERTIFICATION + +Pursuant to the requirement set forth in Rule 13a-14(b) of the Securities Exchange Act of 1934, as amended (the “Exchange Act”), and Section 1350 of Chapter 63 of Title 18 of the United States Code (18 U.S.C. § 1350), Jen-Hsun Huang, the President and Chief Executive Officer of NVIDIA Corporation (the “Company”), hereby certifies that, to the best of his knowledge: + +1. The Company's Quarterly Report on Form 10-Q for the period ended April 26, 2026, to which this Certification is attached as Exhibit 32.1 (the "Periodic Report"), fully complies with the requirements of Section 13(a) or Section 15(d) of the Exchange Act; and +2. The information contained in the Periodic Report fairly presents, in all material respects, the financial condition of the Company at the end of the period covered by the Periodic Report and results of operations of the Company for the period covered by the Periodic Report. + +Date: May 20, 2026 + +/s/JEN-HSUN HUANG + +Jen-Hsun Huang + +President and Chief Executive Officer + +A signed original of this written statement required by Section 906 of 18 U.S.C. § 1350 has been provided to NVIDIA Corporation and will be retained by NVIDIA Corporation and furnished to the Securities and Exchange Commission or its staff upon request. + +This certification accompanies the Form 10-Q to which it relates, is not deemed filed with the Securities and Exchange Commission and is not to be incorporated by reference into any filing of the Company under the Securities Act of 1933, as amended, or the Exchange Act (whether made before or after the date of the Form 10-Q), irrespective of any general incorporation language contained in such filing. + +## CERTIFICATION + +Pursuant to the requirement set forth in Rule 13a-14(b) of the Securities Exchange Act of 1934, as amended (the “Exchange Act”), and Section 1350 of Chapter 63 of Title 18 of the United States Code (18 U.S.C. § 1350), Colette M. Kress, the Executive Vice President and Chief Financial Officer of NVIDIA Corporation (the “Company”), hereby certifies that, to the best of her knowledge: + +1. The Company's Quarterly Report on Form 10-Q for the period ended April 26, 2026, to which this Certification is attached as Exhibit 32.2 (the "Periodic Report"), fully complies with the requirements of Section 13(a) or Section 15(d) of the Exchange Act; and +2. The information contained in the Periodic Report fairly presents, in all material respects, the financial condition of the Company at the end of the period covered by the Periodic Report and results of operations of the Company for the period covered by the Periodic Report. + +Date: May 20, 2026 + +/s/ COLETTE M. KRESS + +Colette M. Kress + +Executive Vice President and Chief Financial Officer + +A signed original of this written statement required by Section 906 of 18 U.S.C. § 1350 has been provided to NVIDIA Corporation and will be retained by NVIDIA Corporation and furnished to the Securities and Exchange Commission or its staff upon request. + +This certification accompanies the Form 10-Q to which it relates, is not deemed filed with the Securities and Exchange Commission and is not to be incorporated by reference into any filing of the Company under the Securities Act of 1933, as amended, or the Exchange Act (whether made before or after the date of the Form 10-Q), irrespective of any general incorporation language contained in such filing. \ No newline at end of file diff --git "a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-1-\342\230\222 QUARTERL.jpg" "b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-1-\342\230\222 QUARTERL.jpg" new file mode 100644 index 0000000000000000000000000000000000000000..c056a9a6638cbc2f8aefeeb9b52b4b3afc314e1c GIT binary patch literal 2333 zcmbW1XH?Tm8pi)2giw_#9Rz}uAYMX{ZmwcNTLA?_FCqw1M5Ki#BA`-ofh*03fOLV- zLz50tz0^Pe1*I!Za_I=pbQX3C=|-X#K_Fb&Bn^Y!phIdb(&jPKvYCnKuG8;OkV1&xSWKLkhHRl z+%F19BvMpLRZ|6_A+Lx;oS1-^n3z~uSozr4_z+@3Vu=4)v}S;V5ikIJ!60#fjspbd z0MXh3(c^bQKtBZhQy@ApJ%j@%rW>$7iZeISI zg7S(=QdMG3T+`+dk%@jX@FpKH~@*LM9?FAdNLu|Q3$Cm-RCNF9UP+qqJE zVG~^|O+U9PLw?%0;xl7c|6#ZRCB-jIFqBR-6zcU>hDE{Q&7b~cdww{J4#sd6En*N! zizdReUd7o`+~Y8_fr>b8pK~>TguMH}VmU4nz)LJsDZt^T+$1H1`@)PVE1P8XIo87j zQ(^xZcYWjJ-U=*QgDs2jI`^(}!0N)z?aI^&!Z&v}gvWe8Rtvx$O zf5UE}oRaV@_zdiALUF*Fs*-OoNCS1xBQtoYv#Ve4Vb-i8*10HCZW1h#sMl-ML@plm zAV*_S7cmegH$+hB+hotcx8l+V?m@(jj_g3@@iHp@)hOruMXryVMA2RC&6g=hW^rMY zo@N5okE>*oy6ZwfL?MOBWJFMxFF{$pQuBgoc)iQZl>6OQ*s|$DqC#}R3-K**j-ug6 ziMM*m`Di!e4Y87|!&`#ZclzOB@f$MvMs2gnaH|Ry%)`9+?*i~rs#HXbWHAxiJ(A#C zjG`A3Ax+QRA8zYU!f-}MS1?Ow)<$YVPkuOL2hWWYJi1(RM9plZfR4U#Ri(G$4Y{>6 z>X1`f?9C3LQ}so`G~gwLuwBcp%da&MW15Uh)^%4lvM>Cer)q4)KPqyqpzyZr;F4dO zsHJCC>u3)EPw%L)<#OSDzH~!Nk^R|yUeeuNWit-lZ5c~YVH44uVPkrCpI^XU0g){o?&F7x$L#AufE^7o#CX-q4w z1yuDS0(Vh1yD8E2NupF(P@&1wGqp$+JeN)FaPa^#W`Ju$M(Hct8%xQ!T7Te*;%L2J zgL&YGH=p(GSGpR`)PFqZq$bm{Jl4Y_Fwii=rgvWNpk5&Y9J8a;^qF{ZLD@A=ER6|nHs^u%De7iK-CI-dWYE?#F(J7D3R$(t0D`P|B@ z-F2~ANtxA3wyZXHSIkYidGmRmT!zxyQC?ufzv>NR5uUDI13_#w;OsbR3VHpr#~wrX zNyC~>8djd(`deKEDOmoz!C81HCW{ZRi|9V;sgo;f;=ecIs!vTVm3@bcXrA>87$W;bYw8=`a+qCJ z+52@_Kl!veId^pulDUQESyx-@errw2^2m0cklwfxFWl*?T5Nl-kp{>+<;1oHyqmpq z^;X*IbtcS(9d8onT{k@5wBuK}mR-$iT`g|9VWb#$TB2-@uq>x>>Y3JLkKhQ}dvtqL z&B4+LX6K_+LSCK`Rq}JVq^m!pfJ8B&ozJ*th}dL^jF2DY*YE*YVHyxsx%|tOE|}-I zYXUjQ(a)RtL##}2AjN64G)3h>bE9*Mv#&)J$?+BoleKx-4%+{>*WSr~9&(<~$lE;h z<=u+O@x2O5*AVrrzL(7MNb%u|?*mSU(ttSu8t_d(dR_S-%z*~r25og8(Ez{5VZ&){ zYc#iMu9?Y$tG~Z7HEP}`CnONAnTPg&!f6}UOC|0&xDAr^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nkLGv?oqKewqS@;`%8-PZpMamMv8*!C>S zH9lO&yLrp5>wgTFWl5ziX30%ADVx}tJZFKOcb}2OC;9D`Ul;J_@ho{@VDIc57yfWQ zPcQRvDV;KjqjNC`-=RF3XfJTWk~zcb}joo@2hFkPVTFG ztJ-IBLVCtZhLn%j_I&oc^tDlO$8{Uwm06p*Q@6BP=H>iZU9qjOy-xSNH?}=hsJ`8yQgQLm-i^;JB{oR=ampHKU&>h2VE_Ln0FSFU A{{R30 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-3-Indicate b.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-3-Indicate b.jpg new file mode 100644 index 0000000000000000000000000000000000000000..f3cc796805eaa23b0d8a580ec4ffd575f01bf9f9 GIT binary patch literal 799 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}muHk}8bvJ)fz6zy5ih|0CHo;=9adXP0vM zvh3Qr$Wx|!XQWTFLf@k`^@nzUNdL=q%l#jd`i}s1FIj&#`MIK>-}h|U%`1QOR_nak z>8wjP%{bA;!m(h+bsH`X=Iej5)9WjjMsAy%@_qB|EV1uuEE5H127W4tX`EAi+5Vx* z_5Tbn+^+8bV08XJLw}$vLxG*bhy7hOuIuk*bw2#LRw!v<@~me}ca+xEg)6VHeX*#E H;r~qlcOW7o literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-4-Large acce.jpg b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/images/image-4-Large acce.jpg new file mode 100644 index 0000000000000000000000000000000000000000..9d1d27d73fa9b8665bbc3b640023e9e7a31fba4b GIT binary patch literal 789 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}m`GMw$DP>TBz*p6ydwk>7sn(gs%FeH^ww zMJE2S`H^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}m`PlIl#epQ(Sp{&`*WBbQ9`#{9(-ofq%X zI(1l~St~_q`ajY8&Gs+CF4=z&v_I5f6>#b^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}mw0_Ik(%dp=YDe*N=0^_Yt$dy5~1w(eWK z{C3=-Sw8Dl7D^OIpLLwTUvItem 1.Financial Statements (Unaudited)a) Condensed Consolidated Statements of Income for the three months ended April 26, 2026 and April 27, 20253b) Condensed Consolidated Statements of Comprehensive Income for the three months ended April 26, 2026 and April 27, 20254c) Condensed Consolidated Balance Sheets as of April 26, 2026 and January 25, 20265d) Condensed Consolidated Statements of Shareholders’ Equity for the three months ended April 26, 2026 and April 27, 20256e) Condensed Consolidated Statements of Cash Flows for the three months ended April 26, 2026 and April 27, 20257f) Notes to Condensed Consolidated Financial Statements8Item 2.Management’s Discussion and Analysis of Financial Condition and Results of Operations23Item 3.Quantitative and Qualitative Disclosures About Market Risk30Item 4.Controls and Procedures31Part II : Other InformationItem 1.Legal Proceedings31Item 1A.Risk Factors31Item 2.Unregistered Sales of Equity Securities and Use of Proceeds38Item 5.Other Information39Item 6.Exhibits40Signature41 \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-1 Financial Results.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-1 Financial Results.html new file mode 100644 index 000000000..eca58d190 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-1 Financial Results.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
Revenue$ 81,615$ 44,062
Cost of revenue20,45817,394
Gross profit61,15726,668
Operating expenses
Research and development6,3213,989
Sales, general and administrative1,3001,041
Total operating expenses7,6215,030
Operating income53,53621,638
Interest income540515
Interest expense(102)(63)
Other income (expense), net15,929(180)
Total other income, net16,367272
Income before income tax69,90321,910
Income tax expense11,5823,135
Net income$ 58,321$ 18,775
Net income per share:
Basic$ 2.40$ 0.77
Diluted$ 2.39$ 0.76
Weighted average shares used in per share computation:
Basic24,28624,441
Diluted24,39124,611
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-10 Amortization Expense.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-10 Amortization Expense.html new file mode 100644 index 000000000..b8a0c0116 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-10 Amortization Expense.html @@ -0,0 +1 @@ +
Future Amortization Expense(In millions)
Fiscal Year:
2027 (excluding the first quarter of fiscal year 2027)$ 689
2028754
2029610
2030516
2031468
2032 and thereafter83
Total$ 3,120
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-11 Debt Securities.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-11 Debt Securities.html new file mode 100644 index 000000000..9fa07c99f --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-11 Debt Securities.html @@ -0,0 +1 @@ +
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable Debt SecuritiesMarketable Equity SecuritiesOther Assets
(In millions)
Debt securities issued by the U.S. TreasuryLevel 2$ 21,883$ 43$ (8)$ 21,918$ 470$ 21,448$ —$ —
Corporate debt securitiesLevel 215,09248(8)15,1321,53313,599
Debt securities issued by U.S. government agenciesLevel 22,0092(1)2,0102,010
Certificates of depositLevel 2132132132
Foreign government bondsLevel 24014141
Money market fundsLevel 110,21210,21210,212
Publicly-held equity securities (1) (2)Level 129,88721,0238,864
Publicly-held equity securities (1) (3)Level 29,2149,214
Total$ 49,368$ 94$ (17)$ 88,546$ 12,347$ 37,098$ 30,237$ 8,864
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-12 Fair Value Table.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-12 Fair Value Table.html new file mode 100644 index 000000000..606b720da --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-12 Fair Value Table.html @@ -0,0 +1 @@ +
Pricing CategoryCost or Amortized CostUnrealized GainUnrealized LossEstimated Fair ValueReported as
Cash EquivalentsMarketable Debt SecuritiesMarketable Equity SecuritiesOther Assets
(In millions)
Debt securities issued by the U.S. TreasuryLevel 2$ 21,635$ 77$ (3)$ 21,709$ —$ 21,709$ —$ —
Corporate debt securitiesLevel 215,41092(3)15,49934515,154
Debt securities issued by U.S. government agenciesLevel 22,15742,1612,161
Certificates of depositLevel 2110110110
Foreign government bondsLevel 24014141
Money market fundsLevel 17,8307,8307,830
Publicly-held equity securities (1) (2)Level 117,72612,8864,840
Total$ 47,182$ 174$ (6)$ 65,076$ 8,285$ 39,065$ 12,886$ 4,840
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-13 Debt Securities.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-13 Debt Securities.html new file mode 100644 index 000000000..d06c33edd --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-13 Debt Securities.html @@ -0,0 +1 @@ +
Apr 26, 2026Jan 25, 2026
Less than 12 monthsLess than 12 months
Estimated Fair ValueGross Unrealized LossEstimated Fair ValueGross Unrealized Loss
(In millions)
Debt securities issued by the U.S. Treasury$ 12,238$ (8)$ 10,666$ (3)
Corporate debt securities2,658(8)1,332(3)
Debt securities issued by U.S. government agencies1,291(1)1,134
Total$ 16,187$ (17)$ 13,132$ (6)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-14 Debt Maturity.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-14 Debt Maturity.html new file mode 100644 index 000000000..19ba171cf --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-14 Debt Maturity.html @@ -0,0 +1 @@ +
Apr 26, 2026(In millions)
Less than one year$ 24,307
Due in 1 - 5 years14,926
Total$ 39,233
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-15 Equity Securities.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-15 Equity Securities.html new file mode 100644 index 000000000..83cb7fd1a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-15 Equity Securities.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Balance at beginning of period$ 22,251$ 3,387
Adjustments related to non-marketable equity securities:
Net additions17,899649
Unrealized gains (1)2,60363
Reclassification (2)(389)(843)
Impairments and unrealized losses(28)(16)
Balance at end of period$ 42,336$ 3,240
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-16 Inventory Data.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-16 Inventory Data.html new file mode 100644 index 000000000..ac34e7eef --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-16 Inventory Data.html @@ -0,0 +1 @@ +
Apr 26, 2026Jan 25, 2026
Inventories:(In millions)
Raw materials$ 6,647$ 3,807
Work in process9,9498,822
Finished goods9,2018,774
Total inventories (1)$ 25,797$ 21,403
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-17 Accrued Liabilities.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-17 Accrued Liabilities.html new file mode 100644 index 000000000..7ba50616b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-17 Accrued Liabilities.html @@ -0,0 +1 @@ +
Apr 26, 2026Jan 25, 2026
Accrued and Other Current Liabilities:(In millions)
Taxes payable$ 10,638$ 2,669
Customer program accruals4,1825,318
Accrued purchase consideration (1)3,9573,921
Excess inventory purchase obligations (2)3,1212,739
Product warranty2,9482,807
Deferred revenue (3)1,7141,379
Accrued payroll and related expenses1,0331,146
Other2,1941,373
Total accrued and other current liabilities$ 29,787$ 21,352
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-18 Long-Term Liabilities.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-18 Long-Term Liabilities.html new file mode 100644 index 000000000..5ce7bbbb3 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-18 Long-Term Liabilities.html @@ -0,0 +1 @@ +
Apr 26, 2026Jan 25, 2026
Other Long-Term Liabilities:(In millions)
Income tax payable (1)$ 4,830$ 3,958
Deferred income tax1,7981,774
Deferred revenue (2)1,4031,193
Other737381
Total other long-term liabilities$ 8,768$ 7,306
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-19 Deferred Revenue.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-19 Deferred Revenue.html new file mode 100644 index 000000000..3a635bbe9 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-19 Deferred Revenue.html @@ -0,0 +1 @@ +
Apr 26, 2026Apr 27, 2025
(In millions)
Balance at beginning of period$ 2,572$ 1,813
Deferred revenue additions (1)2,5306,493
Revenue recognized (2)(1,985)(6,228)
Balance at end of period$ 3,117$ 2,078
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-2 Comprehensive Income.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-2 Comprehensive Income.html new file mode 100644 index 000000000..699531450 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-2 Comprehensive Income.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
Net income$ 58,321$ 18,775
Other comprehensive income (loss), net of tax
Available-for-sale securities:
Net change in unrealized gain (loss)(78)139
Cash flow hedges:
Net change in unrealized gain3719
Other comprehensive income (loss), net of tax(41)158
Total comprehensive income$ 58,280$ 18,933
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-20 Hedge Designation.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-20 Hedge Designation.html new file mode 100644 index 000000000..6515c2f41 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-20 Hedge Designation.html @@ -0,0 +1 @@ +
Apr 26, 2026Jan 25, 2026
(In millions)
Designated as accounting hedges$ 2,114$ 1,765
Not designated as accounting hedges$ 1,850$ 2,332
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-21 Debt Notes Table.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-21 Debt Notes Table.html new file mode 100644 index 000000000..50500cd47 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-21 Debt Notes Table.html @@ -0,0 +1 @@ +
Expected Remaining Term (years)Effective Interest RateApr 26, 2026Jan 25, 2026
(In millions)
3.20% Notes Due 20260.43.31%$ 1,000$ 1,000
1.55% Notes Due 20282.11.64%1,2501,250
2.85% Notes Due 20303.92.93%1,5001,500
2.00% Notes Due 20315.12.09%1,2501,250
3.50% Notes Due 204013.93.54%1,0001,000
3.50% Notes Due 205023.93.54%2,0002,000
3.70% Notes Due 206034.03.73%500500
Unamortized debt discount and issuance costs(30)(32)
Net carrying amount$ 8,470$ 8,468
Less short-term portion(1,000)(999)
Total long-term portion$ 7,470$ 7,469
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-22 Balance Changes.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-22 Balance Changes.html new file mode 100644 index 000000000..2b63c301a --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-22 Balance Changes.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Balance at beginning of period$ 2,807$ 1,290
Additions330870
Utilization(189)(80)
Balance at end of period$ 2,948$ 2,080
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-23 Segment Revenue.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-23 Segment Revenue.html new file mode 100644 index 000000000..2af1d8935 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-23 Segment Revenue.html @@ -0,0 +1 @@ +
Compute & NetworkingGraphicsTotal
(In millions)
Three Months Ended Apr 26, 2026
Revenue$ 74,550$ 7,065$ 81,615
Other segment items (1)21,2154,12425,339
Operating income$ 53,335$ 2,941$ 56,276
Three Months Ended Apr 27, 2025
Revenue$ 39,589$ 4,473$ 44,062
Other segment items (1)17,5352,83320,368
Operating income$ 22,054$ 1,640$ 23,694
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-24 Income Statement.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-24 Income Statement.html new file mode 100644 index 000000000..3cf30e202 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-24 Income Statement.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Segment operating income$ 56,276$ 23,694
Stock-based compensation expense(1,928)(1,474)
Unallocated operating expenses(565)(419)
Acquisition-related and other costs(247)(163)
Interest income540515
Interest expense(102)(63)
Other income (expense), net15,929(180)
Consolidated income before income tax$ 69,903$ 21,910
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-25 Revenue by Region.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-25 Revenue by Region.html new file mode 100644 index 000000000..961f16257 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-25 Revenue by Region.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Geographic Revenue based upon Customer Headquarters Location:
United States$ 63,769$ 25,685
Taiwan12,0067,648
China (including Hong Kong)4,5509,659
Other1,2901,070
Total revenue$ 81,615$ 44,062
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-26 Revenue by Market.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-26 Revenue by Market.html new file mode 100644 index 000000000..83f269b97 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-26 Revenue by Market.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Revenue by Market Platform
Data Center$ 75,246$ 39,112
Hyperscale37,86917,599
AI Clouds, Industrial, & Enterprise37,37721,513
Edge Computing6,3694,950
Total revenue$ 81,615$ 44,062
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-27 Lease Obligations.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-27 Lease Obligations.html new file mode 100644 index 000000000..64cb75840 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-27 Lease Obligations.html @@ -0,0 +1 @@ +
Operating Lease Obligations (In millions)
Fiscal Year:
2027 (excluding the first quarter of fiscal 2027)$ 460
2028626
2029602
2030530
2031462
2032 and thereafter2,924
Total5,604
Less imputed interest1,260
Present value of net future minimum lease payments4,344
Less short-term operating lease liabilities466
Long-term operating lease liabilities$ 3,878
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-28 Lease Cash Flows.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-28 Lease Cash Flows.html new file mode 100644 index 000000000..cc51240dd --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-28 Lease Cash Flows.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Supplemental cash flows information
Operating cash flow used for operating leases$ 185$ 96
Operating lease assets obtained in exchange for lease obligations$ 1,516$ 98
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-29 Quarterly Financials.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-29 Quarterly Financials.html new file mode 100644 index 000000000..1b24e7a25 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-29 Quarterly Financials.html @@ -0,0 +1 @@ +
Three Months EndedQuarter-over-Quarter ChangeYear-over-Year Change
Apr 26, 2026Jan 25, 2026Apr 27, 2025
($ in millions, except per share data)
Revenue$ 81,615$ 68,127$ 44,06220 %85 %
Gross margin74.9 %75.0 %60.5 %(0.1) pts14.4 pts
Operating expenses$ 7,621$ 6,794$ 5,03012 %52 %
Operating income$ 53,536$ 44,299$ 21,63821 %147 %
Net income$ 58,321$ 42,960$ 18,77536 %211 %
Net income per diluted share$ 2.39$ 1.76$ 0.7636 %214 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-3 Balance Sheet.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-3 Balance Sheet.html new file mode 100644 index 000000000..b3c80b2f1 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-3 Balance Sheet.html @@ -0,0 +1 @@ +
Apr 26, 2026Jan 25, 2026
Assets
Current assets:
Cash and cash equivalents$ 13,237$ 10,605
Marketable debt securities37,09839,065
Marketable equity securities30,23712,886
Accounts receivable, net40,71038,466
Inventories25,79721,403
Prepaid expenses and other current assets3,9163,180
Total current assets150,995125,605
Property and equipment, net12,40310,383
Operating lease assets4,2582,867
Goodwill20,89420,832
Intangible assets, net3,1203,306
Deferred income tax assets11,70713,258
Non-marketable securities43,36422,251
Other assets12,7338,301
Total assets$ 259,474$ 206,803
Liabilities and Shareholders’ Equity
Current liabilities:
Accounts payable$ 13,097$ 9,812
Accrued and other current liabilities29,78721,352
Short-term debt1,000999
Total current liabilities43,88432,163
Long-term debt7,4707,469
Long-term operating lease liabilities3,8782,572
Other long-term liabilities8,7687,306
Total liabilities64,00049,510
Commitments and contingencies
Shareholders’ equity:
Preferred stock
Common stock2424
Additional paid-in capital10,27510,118
Accumulated other comprehensive income137178
Retained earnings185,038146,973
Total shareholders’ equity195,474157,293
Total liabilities and shareholders’ equity$ 259,474$ 206,803
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-30 Revenue by Market.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-30 Revenue by Market.html new file mode 100644 index 000000000..6651a39ee --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-30 Revenue by Market.html @@ -0,0 +1 @@ +
Three Months EndedQuarter-over-Quarter ChangeYear-over-Year Change
Apr 26, 2026Jan 25, 2026Apr 27, 2025
($ in millions)
Revenue by Market Platform (1)
Data Center$ 75,246$ 62,314$ 39,11221 %92 %
Hyperscale37,86933,81417,59912 %115 %
AI Clouds, Industrial, & Enterprise37,37728,50021,51331 %74 %
Edge Computing6,3695,8134,95010 %29 %
Total revenue$ 81,615$ 68,127$ 44,06220 %85 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-31 Revenue Analysis.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-31 Revenue Analysis.html new file mode 100644 index 000000000..de4cfcac5 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-31 Revenue Analysis.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
Revenue100.0 %100.0 %
Cost of revenue25.139.5
Gross profit74.960.5
Operating expenses
Research and development7.79.1
Sales, general and administrative1.62.4
Total operating expenses9.311.5
Operating income65.649.0
Interest income0.71.2
Interest expense(0.1)(0.1)
Other income (expense), net19.5(0.4)
Total other income, net20.10.7
Income before income tax85.749.7
Income tax expense14.27.1
Net income71.5 %42.6 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-32 Revenue Growth.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-32 Revenue Growth.html new file mode 100644 index 000000000..18cf82056 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-32 Revenue Growth.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change% Change
($ in millions)
Compute & Networking$ 74,550$ 39,589$ 34,96188 %
Graphics7,0654,4732,59258 %
Total$ 81,615$ 44,062$ 37,55385 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-33 Revenue Growth.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-33 Revenue Growth.html new file mode 100644 index 000000000..af4f74227 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-33 Revenue Growth.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change% Change
($ in millions)
Compute & Networking$ 53,335$ 22,054$ 31,281142 %
Graphics2,9411,6401,30179 %
Total$ 56,276$ 23,694$ 32,582138 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-34 Operating Expenses.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-34 Operating Expenses.html new file mode 100644 index 000000000..613464e56 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-34 Operating Expenses.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change% Change
($ in millions)
Research and development$ 6,321$ 3,989$ 2,33258 %
Sales, general and administrative1,3001,04125925 %
Total operating expenses$ 7,621$ 5,030$ 2,59152 %
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-35 Other Income.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-35 Other Income.html new file mode 100644 index 000000000..92161dca6 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-35 Other Income.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025$ Change
($ in millions)
Interest income$ 540$ 515$ 25
Interest expense(102)(63)(39)
Other income (expense), net15,929(180)16,109
Total other income, net$ 16,367$ 272$ 16,095
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-36 Cash and Securities.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-36 Cash and Securities.html new file mode 100644 index 000000000..eacd323d0 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-36 Cash and Securities.html @@ -0,0 +1 @@ +
Apr 26, 2026Jan 25, 2026
(In millions)
Cash and cash equivalents$ 13,237$ 10,605
Marketable debt securities37,09839,065
Cash, cash equivalents, and marketable debt securities$ 50,335$ 49,670
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-37 Cash Flow Data.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-37 Cash Flow Data.html new file mode 100644 index 000000000..245a557ef --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-37 Cash Flow Data.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Net cash provided by operating activities$ 50,344$ 27,414
Net cash used in investing activities$ (26,429)$ (5,216)
Net cash used in financing activities$ (21,283)$ (15,553)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-38 Debt Maturity.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-38 Debt Maturity.html new file mode 100644 index 000000000..43019de36 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-38 Debt Maturity.html @@ -0,0 +1 @@ +
Apr 26, 2026
(In millions)
Due in one year$ 1,000
Due in one to five years2,750
Due in five to ten years1,250
Due in greater than ten years3,500
Unamortized debt discount and issuance costs(30)
Net carrying amount$ 8,470
Less short-term portion(1,000)
Total long-term portion$ 7,470
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-39 Stock Buyback.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-39 Stock Buyback.html new file mode 100644 index 000000000..212c9b721 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-39 Stock Buyback.html @@ -0,0 +1 @@ +
PeriodTotal Number of Shares Purchased (In millions)Average Price Paid per Share (1)Total Number of Shares Purchased as Part of Publicly Announced Program (In millions)Approximate Dollar Value of Shares that May Yet Be Purchased Under the Program (In billions)
January 26, 2026 - February 22, 20268.2$ 186.358.2$ 57.0
February 23, 2026 - March 22, 20269.0$ 183.389.0$ 55.4
March 23, 2026 - April 26, 202691.1$ 184.9891.1$ 38.5
Total108.3108.3
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-4 Shareholders' Equity.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-4 Shareholders' Equity.html new file mode 100644 index 000000000..6202206f7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-4 Shareholders' Equity.html @@ -0,0 +1 @@ +
Common Stock OutstandingAdditional Paid-in CapitalAccumulated Other Comprehensive IncomeRetained EarningsTotal Shareholders’ Equity
SharesAmount
(In millions, except per share data)
Balances as of Jan 25, 202624,304$ 24$ 10,118$ 178$ 146,973$ 157,293
Net income58,32158,321
Other comprehensive loss(41)(41)
Issuance of common stock37515515
Tax withholding related to common stock(12)(2,129)(2,129)
Shares repurchased(108)(157)(20,013)(20,170)
Cash dividends declared and paid ($0.01 per common share)(243)(243)
Stock-based compensation1,9281,928
Balances as of Apr 26, 202624,221$ 24$ 10,275$ 137$ 185,038$ 195,474
Balances as of Jan 26, 202524,477$ 24$ 11,237$ 28$ 68,038$ 79,327
Net income18,77518,775
Other comprehensive income158158
Issuance of common stock50370370
Tax withholding related to common stock(13)(1,532)(1,532)
Shares repurchased(126)(92)(14,411)(14,503)
Cash dividends declared and paid ($0.01 per common share)(244)(244)
Fair value of partially vested equity awards assumed in connection with acquisitions2222
Stock-based compensation1,4701,470
Balances as of Apr 27, 202524,388$ 24$ 11,475$ 186$ 72,158$ 83,843
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-40 Stock Sale Plans.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-40 Stock Sale Plans.html new file mode 100644 index 000000000..563c659b7 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-40 Stock Sale Plans.html @@ -0,0 +1 @@ +
NameTitle of Director or OfficerActionDateTotal Shares of Common Stock to be SoldExpiration Date
Tench CoxeDirectorAdoption3/19/20268,000,000*10/30/2027
Colette M. KressExecutive Vice President and Chief Financial OfficerTermination4/10/2026500,000**N/A
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-41 Exhibit List.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-41 Exhibit List.html new file mode 100644 index 000000000..531219f60 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-41 Exhibit List.html @@ -0,0 +1 @@ +
Exhibit No.Exhibit DescriptionIncorporated by ReferenceFiling Date
Schedule/FormExhibit
10.1+Variable Compensation Plan - Fiscal Year 20278-K10.13/6/2026
31.1*Certification of Chief Executive Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
31.2*Certification of Chief Financial Officer as required by Rule 13a-14(a) of the Securities Exchange Act of 1934
32.1#*Certification of Chief Executive Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
32.2#*Certification of Chief Financial Officer as required by Rule 13a-14(b) of the Securities Exchange Act of 1934
101.INS*Inline XBRL Instance Document - the instance document does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document.
101.SCH*Inline XBRL Taxonomy Extension Schema Document
101.CAL*Inline XBRL Taxonomy Extension Calculation Linkbase Document
101.DEF*Inline XBRL Taxonomy Extension Definition Linkbase Document
101.LAB*Inline XBRL Taxonomy Extension Labels Linkbase Document
101.PRE*Inline XBRL Taxonomy Extension Presentation Linkbase Document
104Cover Page Interactive Data File - the cover page interactive data file does not appear in the Interactive Data File because its XBRL tags are embedded within the Inline XBRL document.
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-5 Cash Flow Statement.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-5 Cash Flow Statement.html new file mode 100644 index 000000000..cb6efee80 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-5 Cash Flow Statement.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
Cash flows from operating activities:
Net income$ 58,321$ 18,775
Adjustments to reconcile net income to net cash provided by operating activities:
Stock-based compensation expense1,9281,474
Deferred income taxes1,584(2,177)
Depreciation and amortization997611
(Gains) losses from equity securities, net(15,936)175
Other(94)(98)
Changes in operating assets and liabilities, net of acquisitions:
Accounts receivable(2,243)933
Inventories(4,420)(1,258)
Prepaid expenses and other assets(983)560
Accounts payable2,210941
Accrued and other current liabilities7,7637,128
Other long-term liabilities1,217350
Net cash provided by operating activities50,34427,414
Cash flows from investing activities:
Proceeds from maturities of marketable debt securities1,9463,122
Proceeds from sales of non-marketable securities26
Proceeds from sales of marketable debt securities25467
Purchases of non-marketable securities(18,582)(649)
Purchases of marketable debt and equity securities(8,000)(6,546)
Purchases related to property and equipment and intangible assets(1,757)(1,227)
Acquisitions, net of cash acquired(87)(383)
Net cash used in investing activities(26,429)(5,216)
Cash flows from financing activities:
Proceeds related to employee stock plans515370
Payments related to repurchases of common stock(19,312)(14,095)
Payments related to employee stock plan taxes(2,129)(1,532)
Dividends paid(243)(244)
Principal payments on property and equipment and intangible assets(33)(52)
Other(81)
Net cash used in financing activities(21,283)(15,553)
Change in cash and cash equivalents2,6326,645
Cash and cash equivalents at beginning of period10,6058,589
Cash and cash equivalents at end of period$ 13,237$ 15,234
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-6 Cost Breakdown.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-6 Cost Breakdown.html new file mode 100644 index 000000000..3f0beb044 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-6 Cost Breakdown.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions)
Cost of revenue$ 68$ 64
Research and development1,4591,063
Sales, general and administrative401347
Total$ 1,928$ 1,474
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-7 Stock Grant Activity.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-7 Stock Grant Activity.html new file mode 100644 index 000000000..b2fe90c48 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-7 Stock Grant Activity.html @@ -0,0 +1 @@ +
Number of SharesWeighted Average Grant-Date Fair Value Per Share
(In millions, except per share data)
Balance as of Jan 25, 2026189$ 81.51
Granted44$ 181.73
Vested(31)$ 48.04
Canceled and forfeited(2)$ 95.82
Balance as of Apr 26, 2026200$ 108.92
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-8 EPS Data Table.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-8 EPS Data Table.html new file mode 100644 index 000000000..1f0c6501e --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-8 EPS Data Table.html @@ -0,0 +1 @@ +
Three Months Ended
Apr 26, 2026Apr 27, 2025
(In millions, except per share data)
Numerator:
Net income$ 58,321$ 18,775
Denominator:
Basic weighted average shares24,28624,441
Dilutive impact of outstanding equity awards105170
Diluted weighted average shares24,39124,611
Net income per share:
Basic (1)$ 2.40$ 0.77
Diluted (2)$ 2.39$ 0.76
Anti-dilutive equity awards excluded from diluted net income per share4762
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-9 Intangible Assets.html b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-9 Intangible Assets.html new file mode 100644 index 000000000..85f7e9601 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/tables/table-9 Intangible Assets.html @@ -0,0 +1 @@ +
Apr 26, 2026Jan 25, 2026
Gross Carrying AmountAccumulated AmortizationNet Carrying AmountGross Carrying AmountAccumulated AmortizationNet Carrying Amount
(In millions)
Acquisition-related intangible assets$ 5,658$ (2,759)$ 2,899$ 5,656$ (2,580)$ 3,076
Patents and licensed technology525(304)221528(298)230
Total intangible assets$ 6,183$ (3,063)$ 3,120$ 6,184$ (2,878)$ 3,306
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/toc_hierarchies.json b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/toc_hierarchies.json new file mode 100644 index 000000000..d120a17bc --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q1-fy27-results/toc_hierarchies.json @@ -0,0 +1,17 @@ +[ + { + "toc_range": [ + 50, + 52 + ], + "scan_range": [ + 50, + 222 + ], + "toc_with_level": "| id | heading | level |\n|----|--------------------------------|-------|\n| 1 | Page | 1 |\n| 2 | Part I : Financial Information | 1 |", + "toc_tree": { + "Page": {}, + "Part I : Financial Information": {} + } + } +] \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/chunks.json b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/chunks.json new file mode 100644 index 000000000..697c7231d --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/chunks.json @@ -0,0 +1,5285 @@ +{ + "chunks": [ + { + "chunk_id": "6c75d29d-90aa-5237-b67f-5e437aaca413", + "type": "text", + "content": "25-Feb-2026", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf", + "metadata": { + "length": 11, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "25", + "Feb", + "2026" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "783ee690-9a04-50d4-9476-5cff7e71b686", + "type": "text", + "content": "Q4 2026 Earnings Call", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)", + "metadata": { + "length": 21, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Q4", + "2026", + "Earnings", + "Call" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5449c78b-8c75-5762-90cf-9e7106a71989", + "type": "text", + "content": "Toshiya Hari\nVice President, Investor Relations & Strategic Finance, NVIDIA Corp.\nColette M. Kress\nChief Financial Officer & Executive Vice President, NVIDIA Corp.\nJen Hsun Huang\nCo-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)/CORPORATE PARTICIPANTS", + "metadata": { + "length": 250, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Toshiya", + "Hari", + "Vice", + "President", + "Investor", + "Relations", + "Strategic", + "Finance", + "NVIDIA", + "Corp", + "Colette", + "Kress", + "Chief", + "Financial", + "Officer", + "Executive", + "Jen", + "Hsun", + "Huang", + "Co", + "Founder", + "Director" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cfecf30e-182d-5c3b-a831-7eefca0a4b80", + "type": "text", + "content": "Vivek Arya\nAnalyst, BofA Securities, Inc.\nJoe Moore\nAnalyst, Morgan Stanley & Co. LLC\nHarlan Sur\nAnalyst, JPMorgan Securities LLC", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)/OTHER PARTICIPANTS", + "metadata": { + "length": 129, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Vivek", + "Arya", + "Analyst", + "BofA", + "Securities", + "Inc", + "Joe", + "Moore", + "Morgan", + "Stanley", + "Co", + "LLC", + "Harlan", + "Sur", + "JPMorgan" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "72fceaad-4b93-5709-a9e6-348ebc45c365", + "type": "text", + "content": "Analyst, Cantor Fitzgerald & Co.\nStacy A. Rasgon\nAnalyst, Bernstein Research\nAtif Malik\nAnalyst, Citigroup Global Markets, Inc.\nBen Reitzes\nAnalyst, Melius Research LLC\nAntoine Chkaiban\nAnalyst, New Street Research LLP\nMark Lipacis\nAnalyst, Evercore Group LLC\nAaron Rakers\nAnalyst, Wells Fargo Securities LLC\nTimothy Arcuri\nAnalyst, UBS Securities LLC\nJames Edward Schneider\nAnalyst, Goldman Sachs & Co. LLC", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)/C.J. Muse", + "metadata": { + "length": 407, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Cantor", + "Fitzgerald", + "Co", + "Stacy", + "Rasgon", + "Bernstein", + "Research", + "Atif", + "Malik", + "Citigroup", + "Global", + "Markets", + "Inc", + "Ben", + "Reitzes", + "Melius", + "LLC", + "Antoine", + "Chkaiban", + "New", + "Street", + "LLP", + "Mark", + "Lipacis", + "Evercore", + "Group", + "Aaron", + "Rakers", + "Wells", + "Fargo", + "Securities", + "Timothy", + "Arcuri", + "UBS", + "James", + "Edward", + "Schneider", + "Goldman", + "Sachs" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2d3beced-8304-5688-8cdf-fa393cc7d2bc", + "type": "text", + "content": "Operator: Good afternoon. My name is Sarah, and I will be your conference operator today. At this time, I would like to welcome everyone to NVIDIA's Fourth Quarter Earnings Call. All lines have been placed on mute to prevent any background noise. After the speakers' remarks, there will be a question-and-answer session. [Operator Instructions] Thank you.\nToshiya Hari, you may begin your conference.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION", + "metadata": { + "length": 400, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Operator", + "Good", + "afternoon", + "My", + "Sarah", + "conference", + "operator", + "today", + "At", + "time", + "NVIDIA", + "Fourth", + "Quarter", + "Earnings", + "Call", + "All", + "lines", + "mute", + "prevent", + "background", + "noise", + "After", + "speakers", + "remarks", + "question", + "answer", + "session", + "Instructions", + "Thank", + "Toshiya", + "Hari", + "begin" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "899a1688-b331-5679-bfc5-8fbd3bebbdb6", + "type": "text", + "content": "Vice President, Investor Relations & Strategic Finance, NVIDIA Corp.\nThank you. Good afternoon, everyone, and welcome to NVIDIA's conference call for the fourth quarter of fiscal 2026. With me today from NVIDIA are Jensen Huang, President and Chief Executive Officer; and Colette Kress, Executive Vice President and Chief Financial Officer.\nOur call is being webcast live on NVIDIA's Investor Relations website. The webcast will be available for replay until the conference call to discuss our financial results for the first quarter of fiscal 2027. The content of today's call is NVIDIA's property. It can't be reproduced or transcribed without our prior written consent.\nDuring this call, we may make forward looking statements based on current expectations. These are subject to a number of significant risks and uncertainties, and our actual results may differ materially. For a discussion of factors that could affect our future financial results and business, please refer to the disclosure in today's earnings release, our most recent forms 10-K and 10-Q, and the reports that we may file on Form 8-K with the Securities and Exchange Commission. All our statements are made as of today, February 25th, 2026, based on information currently available to us. Except as required by law, we assume no obligation to update any such statements.\nDuring this call, we will discuss non-GAAP financial measures. You can find a reconciliation of these non-GAAP financial measures to GAAP financial measures in our CFO commentary, which is posted on our website.\nWith that, let me turn the call over to Colette.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Toshiya Hari", + "metadata": { + "length": 1605, + "summary": "NVIDIA held its Q4 fiscal 2026 earnings call on February 25, 2026, with CEO Jensen Huang and CFO Colette Kress. The call included forward-looking statements, non-GAAP financial measures, and a replay available until the Q1 FY27 call. Content is proprietary and cannot be reproduced without permission.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Vice", + "President", + "Investor", + "Relations", + "Strategic", + "Finance", + "NVIDIA", + "Corp", + "Thank", + "Good", + "afternoon", + "conference", + "call", + "fourth", + "quarter", + "fiscal", + "2026", + "With", + "today", + "Jensen", + "Huang", + "Chief", + "Executive", + "Officer", + "Colette", + "Kress", + "Financial", + "Our", + "webcast", + "live", + "website", + "The", + "replay", + "discuss", + "financial", + "results", + "2027", + "content", + "property", + "It", + "reproduced", + "transcribed", + "prior", + "written", + "consent", + "During", + "make", + "forward", + "statements", + "based", + "current", + "expectations", + "These", + "subject", + "number", + "significant", + "risks", + "uncertainties", + "actual", + "differ", + "materially", + "For", + "discussion", + "factors", + "affect", + "future", + "business", + "refer", + "disclosure", + "earnings", + "release", + "recent", + "forms", + "10", + "reports", + "file", + "Form", + "Securities", + "Exchange", + "Commission", + "All", + "made", + "February", + "25th", + "information", + "Except", + "required", + "law", + "assume", + "obligation", + "update", + "GAAP", + "measures", + "You", + "find", + "reconciliation", + "CFO", + "commentary", + "posted", + "turn" + ], + "keywords": [ + "NVIDIA", + "earnings call", + "fiscal 2026" + ], + "connect_to": [] + } + }, + { + "chunk_id": "7eea403f-aabc-51b3-9f05-bc55c3c7d364", + "type": "text", + "content": "Chief Financial Officer & Executive Vice President, NVIDIA Corp.\nThanks, Toshiya. We delivered another outstanding quarter with record revenue, operating income and free cash flow. Total revenue of \\$68 billion was up 73% year-over-year, accelerating from Q3. Growth on a sequential basis was also a record, as we added \\$11 billion in data center revenue across a diverse and expanding set of customers, including cloud providers, hyperscalers, AI model makers, enterprises and sovereign nations. Demand for our Blackwell architecture extreme co-designed at data center scale continues to strengthen as inference deployments grow, in addition to training. The transition to accelerated computing and the infusion of AI across existing hyperscale workloads continue to fuel our growth. Agentic and physical AI applications built on increasingly smarter and multimodal models are beginning to drive our financial performance.\nOn a full year basis, data center generated revenue of \\$194 billion, up 68% year-over-year. We have now scaled our data center business by nearly 13x since the emergence of ChatGPT in fiscal 2023. We look ahead, we expect sequential revenue growth throughout calendar 2026, exceeding what was included in the \\$500 billion\nBlackwell and Rubin revenue opportunity we shared last year. We believe we have inventory and supply commitments in place to address future demand, including shipments, extending into calendar 2027.\nEvery data center is power constrained. Customers make critical architectural decisions based on performance per watt, given these constraints and the need to maximize AI factory revenue. SemiAnalysis declared NVIDIA Inference King as recent results from InferenceX reinforced our inference leadership, with GB300 NVL72 achieving up to 50x performance per watt and 35x lower cost per token, compared with Hopper. And continuous optimization of CUDA software helped deliver up to five times better performance on GB200 NVL72 just within four months. NVIDIA produces the lowest cost per token and data centers running on NVIDIA generate the highest revenues.\nOur pace of innovation, particularly at our scale, is unmatched, fueled by an annual R&D budget approaching \\$20 billion and our ability to extreme co-designed across compute and networking across chips, systems, algorithms and softwares. We intend to deliver X factor leaps in performance per watt every generation and extend our leadership position over the long term.\nQ4 data center revenue of \\$62 billion increased 75% year-over-year and 22% sequentially, driven primarily by sustained strength in Blackwell and the Blackwell Ultra ramp. With NVIDIA infrastructure in high demand, even Hopper and much of the six year old Ampere-based products are sold out in the cloud.\nNearly a year has passed since the release of our Grace Blackwell NVL72 systems. Today, nearly nine gigawatts of infrastructure on Blackwell are deployed and consumed by the major cloud service providers, hyperscalers, AI model makers and enterprises.\nNetworking, a cornerstone of our data center scale infrastructure offering, was a standout this quarter, generating \\$11 billion in revenue, up more than 3.5x year-over-year. Demand for our scale up and scale out technologies reached record levels, both growing double digit sequentially, driven by strong adoption of NVLink, Spectrum-X Ethernet and InfiniBand. On a year-over-year basis, growth was driven primarily by NVLink 72 scale up switches as Grace Blackwell Systems accounted for roughly two-thirds of data center revenue in the quarter.\nNVLink scale up fabric has revolutionized computing and demonstrates the power of extreme co-design across all of the chips of the supercomputer and the full stack. In Q4, we announced that we will enable AWS with NVLink to integrate with their custom silicon.\nMomentum is strong with our Spectrum-X Ethernet scale up and scale across networking, as customers work to unify distributed data centers into integrated giga scale AI factories. For the full year, our networking business exceeded \\$31 billion in revenue, up more than 10-x compared to fiscal 2021, the year we acquired Mellanox.\nOur demand profile is broad, diverse and expanding beyond just chatbots. First, there's a fundamental platform shift from classical machine learning to generative AI. Strong evidence of ROI as hyperscalers upgrade massive traditional workloads to generative AI, including search, ad generation and content recommender systems, is encouraging our largest customers to accelerate their capital spending.\nFor example, at Meta, advancements in their GEM model drove a 3.5 increase in ad clicks on Facebook and more than 1% gain in conversations on Instagram, translating into meaningful revenue growth. With the same NVIDIA infrastructure, Meta Superintelligence Labs can train and deploy their frontier agentic AI systems. Frontier agentic systems have reached an inflection point. Claude Code, Claude Cowork and OpenAI Codex have\nachieved useful intelligence. Adoption is skyrocketing, and tokens are profitable, driving extreme urgency to scale up compute. Compute directly translate to intelligence and revenue growth.\nAnalysts expectations for 2026 CapEx across the top five cloud providers and hyperscalers who collectively account for little over 50% of our data center revenue, are up nearly \\$120 billion since the start of the year and approaching \\$700 billion. We continue to expect the transition of classic data center workloads to GPU accelerated computing, and the use of AI to enhance today's hyperscale workloads and contribute toward roughly half of our long-term opportunity.\nEvery country will build and operate some parts of its AI infrastructure, just like with electricity and Internet today. In fiscal year 2026, our sovereign AI business, more than tripled year-over-year and over \\$30 billion, driven primarily by customers based in Canada, France, the Netherlands, Singapore and the UK. Over the long run, we expect our sovereign opportunity to grow at least in line with the AI infrastructure market, as countries spend on AI proportional to their GDP.\nWhile small amounts of H200 products for China-based customers were approved by the US government, we have yet to generate any revenue and we do not know whether any imports will be allowed into China. Our competitors in China, bolstered by recent IPOs, are making progress and have the potential to disrupt the structure of the global AI industry over the long-term. To sustain its leadership position in AI compute, America must engage every developer and be the platform for choice for every commercial business, including those in China.\nWe will continue to engage with the US and China governments and advocate for America's ability to compete around the world. We unveiled the Rubin platform last month at CES, comprised of six new chips; the Vera CPU, Rubin GPU, NVLink, 6 Switch, ConnectX-9, SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet Switch. The platform will train MoE models with one-fourth the number of GPUs and reduce inference token costs by up to 10x compared to Blackwell.\nWe shipped our first Vera Rubin samples to customers earlier this week and we remain on track to commence production shipments in the second half of the year. Based on its modular cable-free tray design, Rubin will deliver improved resiliency and serviceability relative to Blackwell. We expect every cloud model builder to deploy Vera Rubin.\nMoving to gaming. Gaming revenue of \\$3.7 billion increased 47% year-on-year, driven by strong Blackwell demand and improved supply. GeForce RTX is the leading platform for PC gamers, creators, and developers.\nIn Q4, we added several new technologies and advancements, including DLSS 4.5, which uses AI to bring game visuals to a new level. G-SYNC Pulsar, bringing incredible clear graphics even in motion, and 35% faster LLM inference across leading AI PC frameworks.\nLooking ahead, while end demand for our products remains strong and channel inventory levels are healthy, we expect supply constraints to be the headwind to gaming in Q1 and beyond.\nFor Professional Visualization, it crossed the \\$1 billion mark for the first time, with revenue of \\$1.3 billion, up 159% year-over-year and 74% sequentially. During the quarter, we launched the RTX PRO 5000 Blackwell Workstation with 72 gigabytes of fast memory for AI developers running LLMs and agentic workflows.\nAutomotive revenue of \\$604 million was up 6% year-over-year and was driven by robust demand for self-driving solutions. At CES, we introduced Alpamayo, the world's first open portfolio of reasoning, vision, language, action models, simulation blueprints and data sets enabling vehicles that can think.\nThe first passenger car featuring Alpamayo built on NVIDIA Drive will be on the road soon in the new Mercedes-Benz CLA.\nPhysical AI is here, having already contributed north of \\$6 billion in NVIDIA revenue in fiscal year 2026. Robotaxi rides are growing exponentially with commercial fleets from Waymo, Tesla, Uber, WeRide and Zoox, and many others are expected to scale from thousands of vehicles in 2025 to millions over the next decade, creating a market poised to generate hundreds of billions of dollars of revenue.\nThis expansion will demand orders of magnitude more compute with every major OEM and service provider developing on NVIDIA's platform. We continue to advance robotics development with the new NVIDIA Cosmos and Isaac GROOT open models, frameworks and NVIDIA's powered robots and autonomous machines for leading companies, including Boston Dynamics, Caterpillar, Franka Robotics, LG Electronics and NEURA Robotics. To accelerate industrial physical AI adoption, we also announced new expanding partnerships with Dassault Systèmes, Siemens and Synopsys to bring NVIDIA AI infrastructure, Omniverse digital twins, world models and CUDA-X libraries to millions of researchers, designers and engineers building the world's industries.\nLet's move to the rest of the P&L. GAAP gross margin was 75%, and non-GAAP gross margin was 75.2%, increasing sequentially as Blackwell continued to ramp. GAAP operating expenses were up 16% sequentially and up 21% on a non-GAAP basis, related to new product introductions and compute and infrastructure costs. Non-GAAP effective tax rate for the fourth quarter was 15.4%, below our outlook for the quarter, primarily due to the impact of a one-time tax benefit.\nInventory grew 8% quarter-over-quarter, while purchase commitments also increased significantly, as we have strategically secured inventory and capacity to meet demand beyond the next several quarters. This is further out in time than usual and reflects the longer demand visibility we have. While we expect tightness in the supply for our advanced architectures to persist, we remain confident in our ability to capitalize on the growth opportunity ahead with our scale, expansive supply chain, and the longstanding partnerships continuing to serve us well.\nWe generated free cash flow of \\$35 billion in Q4 and \\$97 billion in fiscal year 2026. For the year, we returned \\$41 billion, or 43% of free cash flow to our shareholders in the form of share repurchases and dividends. We continue to invest in our technology and our ecosystem to cultivate market development, drive long-term growth, and ultimately yield total shareholder returns superior to the market or our peer group. Importantly, we will continue to run a strategic and disciplined process as it relates to our investments, and we remain committed to returning capital to our shareholders.\nLet me turn to the outlook for the first quarter. Starting this quarter, we will be including stock-based compensation expense in our non-GAAP results. Stock-based compensation is a foundational component of our compensation program to attract and retain world class talent. Let me first start with revenue. Total revenue is expected to be \\$78 billion, plus or minus 2%. We expect most of our growth to be driven by Data Center. Consistent with last quarter, we are not assuming any Data Center compute revenue from China in our outlook. GAAP and non-GAAP gross margins are expected to be 74.9% and 75%, respectively, plus or minus 50 basis points. For the full year, we continue to see gross margins in the mid-70s. We will keep you updated on our progress as we prepare for the Vera Rubin transition.\nGAAP and non-GAAP operating expenses are expected to be approximately \\$7.7 billion and \\$7.5 billion, respectively, including stock-based compensation expense of \\$1.9 billion.\nFor the full year, we expect non-GAAP operating expenses to grow in the low 40s on a year-over-year basis as we continue to invest in our expanding opportunity set. For the full year fiscal year 2027, we expect GAAP and non-GAAP tax rates to be in between 7% and 19%, excluding any discrete items and material changes to our tax environment.\nWith that, let me turn the call over to Jensen. I think he has a few words for us.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Colette M. Kress", + "metadata": { + "length": 13039, + "summary": "NVIDIA reported record Q4 revenue of $68 billion, up 73% year-over-year, driven by strong Data Center growth (up 75% to $62 billion) and Blackwell architecture demand. Full-year Data Center revenue reached $194 billion. Networking revenue hit $11 billion, up 3.5x. Gaming revenue rose 47% to $3.7 billion. Professional Visualization crossed $1 billion. Automotive revenue was $604 million. Q1 FY27 revenue outlook is $78 billion. The company highlighted AI infrastructure demand, sovereign AI, and the Rubin platform launch.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Chief", + "Financial", + "Officer", + "Executive", + "Vice", + "President", + "NVIDIA", + "Corp", + "Thanks", + "Toshiya", + "We", + "delivered", + "outstanding", + "quarter", + "record", + "revenue", + "operating", + "income", + "free", + "cash", + "flow", + "Total", + "68", + "billion", + "73%", + "year", + "accelerating", + "Q3", + "Growth", + "sequential", + "basis", + "added", + "11", + "data", + "center", + "diverse", + "expanding", + "set", + "customers", + "including", + "cloud", + "providers", + "hyperscalers", + "AI", + "model", + "makers", + "enterprises", + "sovereign", + "nations", + "Demand", + "Blackwell", + "architecture", + "extreme", + "designed", + "scale", + "continues", + "strengthen", + "inference", + "deployments", + "grow", + "addition", + "training", + "The", + "transition", + "accelerated", + "computing", + "infusion", + "existing", + "hyperscale", + "workloads", + "continue", + "fuel", + "growth", + "Agentic", + "physical", + "applications", + "built", + "increasingly", + "smarter", + "multimodal", + "models", + "beginning", + "drive", + "financial", + "performance", + "On", + "full", + "generated", + "194", + "68%", + "scaled", + "business", + "13x", + "emergence", + "ChatGPT", + "fiscal", + "2023", + "ahead", + "expect", + "calendar", + "2026", + "exceeding", + "included", + "500", + "Rubin", + "opportunity", + "shared", + "inventory", + "supply", + "commitments", + "place", + "address", + "future", + "demand", + "shipments", + "extending", + "2027", + "Every", + "power", + "constrained", + "Customers", + "make", + "critical", + "architectural", + "decisions", + "based", + "watt", + "constraints", + "maximize", + "factory", + "SemiAnalysis", + "declared", + "Inference", + "King", + "recent", + "results", + "InferenceX", + "reinforced", + "leadership", + "GB300", + "NVL72", + "achieving", + "50x", + "35x", + "lower", + "cost", + "token", + "compared", + "Hopper", + "And", + "continuous", + "optimization", + "CUDA", + "software", + "helped", + "deliver", + "times", + "GB200", + "months", + "produces", + "lowest", + "centers", + "running", + "generate", + "highest", + "revenues", + "Our", + "pace", + "innovation", + "unmatched", + "fueled", + "annual", + "budget", + "approaching", + "20", + "ability", + "compute", + "networking", + "chips", + "systems", + "algorithms", + "softwares", + "intend", + "factor", + "leaps", + "generation", + "extend", + "position", + "long", + "term", + "Q4", + "62", + "increased", + "75%", + "22%", + "sequentially", + "driven", + "primarily", + "sustained", + "strength", + "Ultra", + "ramp", + "With", + "infrastructure", + "high", + "Ampere", + "products", + "sold", + "Nearly", + "passed", + "release", + "Grace", + "Today", + "gigawatts", + "deployed", + "consumed", + "major", + "service", + "Networking", + "cornerstone", + "offering", + "standout", + "generating", + "3.5", + "technologies", + "reached", + "levels", + "growing", + "double", + "digit", + "strong", + "adoption", + "NVLink", + "Spectrum", + "Ethernet", + "InfiniBand", + "72", + "switches", + "Systems", + "accounted", + "roughly", + "thirds", + "fabric", + "revolutionized", + "demonstrates", + "design", + "supercomputer", + "stack", + "In", + "announced", + "enable", + "AWS", + "integrate", + "custom", + "silicon", + "Momentum", + "work", + "unify", + "distributed", + "integrated", + "giga", + "factories", + "For", + "exceeded", + "31", + "10", + "2021", + "acquired", + "Mellanox", + "profile", + "broad", + "chatbots", + "First", + "fundamental", + "platform", + "shift", + "classical", + "machine", + "learning", + "generative", + "Strong", + "evidence", + "ROI", + "upgrade", + "massive", + "traditional", + "search", + "ad", + "content", + "recommender", + "encouraging", + "largest", + "accelerate", + "capital", + "spending", + "Meta", + "advancements", + "GEM", + "drove", + "increase", + "clicks", + "Facebook", + "1%", + "gain", + "conversations", + "Instagram", + "translating", + "meaningful", + "Superintelligence", + "Labs", + "train", + "deploy", + "frontier", + "agentic", + "Frontier", + "inflection", + "point", + "Claude", + "Code", + "Cowork", + "OpenAI", + "Codex", + "achieved", + "intelligence", + "Adoption", + "skyrocketing", + "tokens", + "profitable", + "driving", + "urgency", + "Compute", + "directly", + "translate", + "Analysts", + "expectations", + "CapEx", + "top", + "collectively", + "account", + "50%", + "120", + "start", + "700", + "classic", + "GPU", + "enhance", + "today", + "contribute", + "half", + "country", + "build", + "operate", + "parts", + "electricity", + "Internet", + "tripled", + "30", + "Canada", + "France", + "Netherlands", + "Singapore", + "UK", + "Over", + "run", + "line", + "market", + "countries", + "spend", + "proportional", + "GDP", + "While", + "small", + "amounts", + "H200", + "China", + "approved", + "US", + "government", + "imports", + "allowed", + "competitors", + "bolstered", + "IPOs", + "making", + "progress", + "potential", + "disrupt", + "structure", + "global", + "industry", + "To", + "sustain", + "America", + "engage", + "developer", + "choice", + "commercial", + "governments", + "advocate", + "compete", + "world", + "unveiled", + "month", + "CES", + "comprised", + "Vera", + "CPU", + "Switch", + "ConnectX", + "SuperNIC", + "BlueField", + "DPU", + "MoE", + "fourth", + "number", + "GPUs", + "reduce", + "costs", + "10x", + "shipped", + "samples", + "earlier", + "week", + "remain", + "track", + "commence", + "production", + "Based", + "modular", + "cable", + "tray", + "improved", + "resiliency", + "serviceability", + "relative", + "builder", + "Moving", + "gaming", + "Gaming", + "3.7", + "47%", + "GeForce", + "RTX", + "leading", + "PC", + "gamers", + "creators", + "developers", + "DLSS", + "4.5", + "bring", + "game", + "visuals", + "level", + "SYNC", + "Pulsar", + "bringing", + "incredible", + "clear", + "graphics", + "motion", + "35%", + "faster", + "LLM", + "frameworks", + "Looking", + "end", + "remains", + "channel", + "healthy", + "headwind", + "Q1", + "Professional", + "Visualization", + "crossed", + "mark", + "time", + "1.3", + "159%", + "74%", + "During", + "launched", + "PRO", + "5000", + "Workstation", + "gigabytes", + "fast", + "memory", + "LLMs", + "workflows", + "Automotive", + "604", + "million", + "6%", + "robust", + "solutions", + "At", + "introduced", + "Alpamayo", + "open", + "portfolio", + "reasoning", + "vision", + "language", + "action", + "simulation", + "blueprints", + "sets", + "enabling", + "vehicles", + "passenger", + "car", + "featuring", + "Drive", + "road", + "Mercedes", + "Benz", + "CLA", + "Physical", + "contributed", + "north", + "Robotaxi", + "rides", + "exponentially", + "fleets", + "Waymo", + "Tesla", + "Uber", + "WeRide", + "Zoox", + "expected", + "thousands", + "2025", + "millions", + "decade", + "creating", + "poised", + "hundreds", + "billions", + "dollars", + "This", + "expansion", + "orders", + "magnitude", + "OEM", + "provider", + "developing", + "advance", + "robotics", + "development", + "Cosmos", + "Isaac", + "GROOT", + "powered", + "robots", + "autonomous", + "machines", + "companies", + "Boston", + "Dynamics", + "Caterpillar", + "Franka", + "Robotics", + "LG", + "Electronics", + "NEURA", + "industrial", + "partnerships", + "Dassault", + "Syst", + "mes", + "Siemens", + "Synopsys", + "Omniverse", + "digital", + "twins", + "libraries", + "researchers", + "designers", + "engineers", + "building", + "industries", + "Let", + "move", + "rest", + "GAAP", + "gross", + "margin", + "75.2%", + "increasing", + "continued", + "expenses", + "16%", + "21%", + "related", + "product", + "introductions", + "Non", + "effective", + "tax", + "rate", + "15.4%", + "outlook", + "due", + "impact", + "benefit", + "Inventory", + "grew", + "8%", + "purchase", + "significantly", + "strategically", + "secured", + "capacity", + "meet", + "quarters", + "usual", + "reflects", + "longer", + "visibility", + "tightness", + "advanced", + "architectures", + "persist", + "confident", + "capitalize", + "expansive", + "chain", + "longstanding", + "continuing", + "serve", + "35", + "97", + "returned", + "41", + "43%", + "shareholders", + "form", + "share", + "repurchases", + "dividends", + "invest", + "technology", + "ecosystem", + "cultivate", + "ultimately", + "yield", + "total", + "shareholder", + "returns", + "superior", + "peer", + "group", + "Importantly", + "strategic", + "disciplined", + "process", + "relates", + "investments", + "committed", + "returning", + "turn", + "Starting", + "stock", + "compensation", + "expense", + "Stock", + "foundational", + "component", + "program", + "attract", + "retain", + "class", + "talent", + "78", + "minus", + "2%", + "Data", + "Center", + "Consistent", + "assuming", + "margins", + "74.9%", + "50", + "points", + "mid", + "70s", + "updated", + "prepare", + "approximately", + "7.7", + "7.5", + "1.9", + "low", + "40s", + "rates", + "7%", + "19%", + "excluding", + "discrete", + "items", + "material", + "environment", + "call", + "Jensen", + "words" + ], + "keywords": [ + "NVIDIA", + "Data Center", + "AI" + ], + "connect_to": [] + } + }, + { + "chunk_id": "c1b56c6c-d6e6-5d89-affc-e607a47f2fb4", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\nThis quarter we significantly deepened and expanded our partnerships with leading frontier model makers. We recently celebrated OpenAI's launch of GPT-5.3-Codex, trained with and inferencing on Grace Blackwell NVLink 72 systems.\nGPT-5.3-Codex can take on long running tasks that involve research, tool use, and complex execution. 5.3-Codex is deployed broadly inside NVIDIA, our engineers love it.\nWe continue to work with OpenAI toward a partnership agreement and believe we are close. We are thrilled with our ongoing partnership with OpenAI, a once in a generation company we've had the pleasure of partnering with since their first days.\nMeta Superintelligence Labs is scaling up at lightning speed. Last week, we announced that Meta is deploying millions of Blackwells and Rubin GPUs, NVIDIA CPUs, and Spectrum-X Ethernet for training and inference. This quarter, we announced a partnership with Anthropic and a \\$10 billion investment in their company. Anthropic will train an inference on Grace Blackwell and Vera Rubin systems.\nAnthropic's Claude Cowork agent platform is revolutionary and has opened a floodgates for enterprise AI adoption. Between Claude Cowork and OpenClaw, compute demand is skyrocketing and ChatGPT moment of agentic AI has arrived.\nWith partnerships spanning Anthropic, Meta, OpenAI, and xAI, NVIDIA deployed across every cloud, and with our ability to build full-stack AI infrastructure from the ground up or support them in the cloud, we're uniquely positioned to partner with frontier model builders at every stage; training, inference, and AI factory scale out.\nFinally, we recently entered into a non-exclusive licensing agreement with Groq for its low-latency inference technology, and welcomed a team of brilliant engineers to NVIDIA. As we did with Mellanox, we will extend NVIDIA's architecture with Groq's innovations to enable new levels of AI infrastructure performance and value. We look forward to sharing more at GTC next month.\nOkay, back to you.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Jen Hsun Huang", + "metadata": { + "length": 2065, + "summary": "NVIDIA's CEO highlights deepened partnerships with OpenAI, Meta, Anthropic, and xAI. OpenAI launched GPT-5.3-Codex on NVIDIA systems. Meta is deploying millions of Blackwell and Rubin GPUs. NVIDIA invested $10 billion in Anthropic, whose Claude Cowork platform drives enterprise AI adoption. A non-exclusive licensing agreement with Groq was also announced.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "This", + "quarter", + "significantly", + "deepened", + "expanded", + "partnerships", + "leading", + "frontier", + "model", + "makers", + "We", + "recently", + "celebrated", + "OpenAI", + "launch", + "GPT", + "5.3", + "Codex", + "trained", + "inferencing", + "Grace", + "Blackwell", + "NVLink", + "72", + "systems", + "long", + "running", + "tasks", + "involve", + "research", + "tool", + "complex", + "execution", + "deployed", + "broadly", + "inside", + "engineers", + "love", + "continue", + "work", + "partnership", + "agreement", + "close", + "thrilled", + "ongoing", + "generation", + "company", + "ve", + "pleasure", + "partnering", + "days", + "Meta", + "Superintelligence", + "Labs", + "scaling", + "lightning", + "speed", + "Last", + "week", + "announced", + "deploying", + "millions", + "Blackwells", + "Rubin", + "GPUs", + "CPUs", + "Spectrum", + "Ethernet", + "training", + "inference", + "Anthropic", + "10", + "billion", + "investment", + "train", + "Vera", + "Claude", + "Cowork", + "agent", + "platform", + "revolutionary", + "opened", + "floodgates", + "enterprise", + "AI", + "adoption", + "Between", + "OpenClaw", + "compute", + "demand", + "skyrocketing", + "ChatGPT", + "moment", + "agentic", + "arrived", + "With", + "spanning", + "xAI", + "cloud", + "ability", + "build", + "full", + "stack", + "infrastructure", + "ground", + "support", + "uniquely", + "positioned", + "partner", + "builders", + "stage", + "factory", + "scale", + "Finally", + "entered", + "exclusive", + "licensing", + "Groq", + "low", + "latency", + "technology", + "welcomed", + "team", + "brilliant", + "As", + "Mellanox", + "extend", + "architecture", + "innovations", + "enable", + "levels", + "performance", + "forward", + "sharing", + "GTC", + "month", + "Okay", + "back" + ], + "keywords": [ + "NVIDIA", + "AI partnerships", + "frontier models" + ], + "connect_to": [] + } + }, + { + "chunk_id": "d01243b6-a695-5833-a349-de5d11a35eb0", + "type": "text", + "content": "Vice President, Investor Relations & Strategic Finance, NVIDIA Corp.\nWe will now transition to Q&A. Operator, please poll for questions.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Toshiya Hari_2", + "metadata": { + "length": 136, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Vice", + "President", + "Investor", + "Relations", + "Strategic", + "Finance", + "NVIDIA", + "Corp", + "We", + "transition", + "Operator", + "poll", + "questions" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1b336282-5396-507e-9f20-873215435528", + "type": "text", + "content": "Operator: [Operator Instructions] Your first question comes from Vivek Arya with Bank of America Securities. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION", + "metadata": { + "length": 127, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Operator", + "Instructions", + "Your", + "question", + "Vivek", + "Arya", + "Bank", + "America", + "Securities", + "line", + "open" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "acc3c007-53eb-5731-a952-6f07aeb4deca", + "type": "image", + "content": "\nWe will now transition to Q&A. Operator, please poll for questions. Thanks for taking my question. I think you mentioned that you now have growth visibility into calendar 2027 also, and I think your purchase commitments kind of reflect that confidence. But Jensen, I'm curious, when you look at your top cloud customers, cloud CapEx close to \\$700 billion this year, many investors are concerned that it would be harder for this level to grow into next year. And for several of them, their cash flow generation capability is also getting compressed. So I know you're very confident about your roadmap, right, and your purchase commitments and whatnot, but how confident are you about your customers' ability to continue to grow their CapEx? And if their CapEx doesn't grow, can NVIDIA still find a way to grow in that envelope? Thank you.0\n[images/image-1-We will no.jpg]\n", + "path": "images/image-1-We will no.jpg", + "metadata": { + "length": 873, + "summary": "image-1\nWe will now transition to Q&A. Operator, please poll for questions. Thanks for taking my question. I think you mentioned that you now have growth visibility into calendar 2027 also, and I think your purchase commitments kind of reflect that confidence. But Jensen, I'm curious, when you look at your top cloud customers, cloud CapEx close to \\$700 billion this year, many investors are concerned that it would be harder for this level to grow into next year. And for several of them, their cash flow generation capability is also getting compressed. So I know you're very confident about your roadmap, right, and your purchase commitments and whatnot, but how confident are you about your customers' ability to continue to grow their CapEx? And if their CapEx doesn't grow, can NVIDIA still find a way to grow in that envelope? Thank you.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-1-We will no.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "2a456310-ba0e-5f46-ba14-9d171d2b577e", + "type": "text", + "content": "Analyst, BofA Securities, Inc.\n\nWe will now transition to Q&A. Operator, please poll for questions. Thanks for taking my question. I think you mentioned that you now have growth visibility into calendar 2027 also, and I think your purchase commitments kind of reflect that confidence. But Jensen, I'm curious, when you look at your top cloud customers, cloud CapEx close to \\$700 billion this year, many investors are concerned that it would be harder for this level to grow into next year. And for several of them, their cash flow generation capability is also getting compressed. So I know you're very confident about your roadmap, right, and your purchase commitments and whatnot, but how confident are you about your customers' ability to continue to grow their CapEx? And if their CapEx doesn't grow, can NVIDIA still find a way to grow in that envelope? Thank you.0\n[images/image-1-We will no.jpg]\n\nThanks for taking my question. I think you mentioned that you now have growth visibility into calendar 2027 also, and I think your purchase commitments kind of reflect that confidence. But Jensen, I'm curious, when you look at your top cloud customers, cloud CapEx close to \\$700 billion this year, many investors are concerned that it would be harder for this level to grow into next year. And for several of them, their cash flow generation capability is also getting compressed. So I know you're very confident about your roadmap, right, and your purchase commitments and whatnot, but how confident are you about your customers' ability to continue to grow their CapEx? And if their CapEx doesn't grow, can NVIDIA still find a way to grow in that envelope? Thank you.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Vivek Arya", + "metadata": { + "length": 1675, + "summary": "An analyst questions NVIDIA's confidence in customer CapEx growth, noting cloud CapEx near $700 billion and compressed cash flow. Jensen is asked about growth visibility into 2027 and whether NVIDIA can grow if customer CapEx does not increase.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "BofA", + "Securities", + "Inc", + "We", + "transition", + "Operator", + "poll", + "questions", + "Thanks", + "taking", + "question", + "mentioned", + "growth", + "visibility", + "calendar", + "2027", + "purchase", + "commitments", + "kind", + "reflect", + "confidence", + "But", + "Jensen", + "curious", + "top", + "cloud", + "customers", + "CapEx", + "close", + "700", + "billion", + "year", + "investors", + "concerned", + "harder", + "level", + "grow", + "And", + "cash", + "flow", + "generation", + "capability", + "compressed", + "So", + "confident", + "roadmap", + "whatnot", + "ability", + "continue", + "doesn", + "NVIDIA", + "find", + "envelope", + "Thank", + "you.0" + ], + "keywords": [ + "NVIDIA", + "Cloud CapEx", + "Growth" + ], + "connect_to": [ + { + "target": "acc3c007-53eb-5731-a952-6f07aeb4deca", + "relation": "embeds", + "ref": "[images/image-1-We will no.jpg]", + "position": { + "start": 872, + "end": 903 + } + } + ] + } + }, + { + "chunk_id": "6482acbb-d24c-5066-91cd-ab73b854838a", + "type": "image", + "content": "\nAnalyst, BofA Securities, Inc. I am confident in their cash flow growing. And the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents across the world in enterprises everywhere. You're seeing incredible compute demand because of it. In this new world of AI, compute is revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues. So in this new world of AI, compute equals revenues. And I am certain that at this point, with the productive use of Codex and Claude Code and the excitement around Claude Cowork and, just the incredible enthusiasm about OpenClaw and the enterprise versions of them. All of the enterprise ISVs who are now working on agentic systems on top of their tools platforms, I am certain at this point that we are at the inflection point. We've reached the inflection point, and we're generating profitable tokens that are productive for customers and profitable for the cloud service providers. And so the simple logic of it, the simple way to think about it is computing has changed. What used to be software running on computers, modest amount of computers, call it \\$300 billion or \\$400 billion worth of CapEx each year has now gone into AI and AI in order to have, in order to generate tokens, you need compute capacity. And that translates directly to growth and that translates directly to revenues.0\n[images/image-2-Analyst, B.jpg]\n", + "path": "images/image-2-Analyst, B.jpg", + "metadata": { + "length": 1468, + "summary": "image-2\nAnalyst, BofA Securities, Inc. I am confident in their cash flow growing. And the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents across the world in enterprises everywhere. You're seeing incredible compute demand because of it. In this new world of AI, compute is revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues. So in this new world of AI, compute equals revenues. And I am certain that at this point, with the productive use of Codex and Claude Code and the excitement around Claude Cowork and, just the incredible enthusiasm about OpenClaw and the enterprise versions of them. All of the enterprise ISVs who are now working on agentic systems on top of their tools platforms, I am certain at this point that we are at the inflection point. We've reached the inflection point, and we're generating profitable tokens that are productive for customers and profitable for the cloud service providers. And so the simple logic of it, the simple way to think about it is computing has changed. What used to be software running on computers, modest amount of computers, call it \\$300 billion or \\$400 billion worth of CapEx each year has now gone into AI and AI in order to have, in order to generate tokens, you need compute capacity. And that translates directly to growth and that translates directly to revenues.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-2-Analyst, B.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "cb60366a-1b21-526a-a823-40aa12e6b454", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nAnalyst, BofA Securities, Inc. I am confident in their cash flow growing. And the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents across the world in enterprises everywhere. You're seeing incredible compute demand because of it. In this new world of AI, compute is revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues. So in this new world of AI, compute equals revenues. And I am certain that at this point, with the productive use of Codex and Claude Code and the excitement around Claude Cowork and, just the incredible enthusiasm about OpenClaw and the enterprise versions of them. All of the enterprise ISVs who are now working on agentic systems on top of their tools platforms, I am certain at this point that we are at the inflection point. We've reached the inflection point, and we're generating profitable tokens that are productive for customers and profitable for the cloud service providers. And so the simple logic of it, the simple way to think about it is computing has changed. What used to be software running on computers, modest amount of computers, call it \\$300 billion or \\$400 billion worth of CapEx each year has now gone into AI and AI in order to have, in order to generate tokens, you need compute capacity. And that translates directly to growth and that translates directly to revenues.0\n[images/image-2-Analyst, B.jpg]\n\nI am confident in their cash flow growing. And the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents across the world in enterprises everywhere. You're seeing incredible compute demand because of it. In this new world of AI, compute is revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues. So in this new world of AI, compute equals revenues. And I am certain that at this point, with the productive use of Codex and Claude Code and the excitement around Claude Cowork and, just the incredible enthusiasm about OpenClaw and the enterprise versions of them. All of the enterprise ISVs who are now working on agentic systems on top of their tools platforms, I am certain at this point that we are at the inflection point. We've reached the inflection point, and we're generating profitable tokens that are productive for customers and profitable for the cloud service providers. And so the simple logic of it, the simple way to think about it is computing has changed. What used to be software running on computers, modest amount of computers, call it \\$300 billion or \\$400 billion worth of CapEx each year has now gone into AI and AI in order to have, in order to generate tokens, you need compute capacity. And that translates directly to growth and that translates directly to revenues.\nOperator: Your next question comes from Joe Moore with Morgan Stanley. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang", + "metadata": { + "length": 3033, + "summary": "The speaker, an analyst from BofA Securities, expresses strong confidence in cash flow growth driven by the inflection of agentic AI. They argue that in the new AI era, compute capacity directly equals revenues, as tokens require compute to generate. Citing productive use of tools like Codex and Claude Code, and enterprise ISVs building agentic systems, they assert we have reached an inflection point where tokens are profitable for both customers and cloud providers. The speaker notes that annual CapEx has shifted from $300-400 billion on traditional software to AI, where compute translates directly to growth and revenues.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Analyst", + "BofA", + "Securities", + "Inc", + "confident", + "cash", + "flow", + "growing", + "And", + "reason", + "simple", + "We", + "inflection", + "agentic", + "AI", + "usefulness", + "agents", + "world", + "enterprises", + "You", + "incredible", + "compute", + "demand", + "In", + "revenues", + "Without", + "generate", + "tokens", + "grow", + "So", + "equals", + "point", + "productive", + "Codex", + "Claude", + "Code", + "excitement", + "Cowork", + "enthusiasm", + "OpenClaw", + "enterprise", + "versions", + "All", + "ISVs", + "working", + "systems", + "top", + "tools", + "platforms", + "ve", + "reached", + "generating", + "profitable", + "customers", + "cloud", + "service", + "providers", + "logic", + "computing", + "changed", + "What", + "software", + "running", + "computers", + "modest", + "amount", + "call", + "300", + "billion", + "400", + "worth", + "CapEx", + "year", + "order", + "capacity", + "translates", + "directly", + "growth", + "revenues.0", + "Operator", + "Your", + "question", + "Joe", + "Moore", + "Morgan", + "Stanley", + "line", + "open" + ], + "keywords": [ + "Agentic AI", + "Compute Demand", + "Revenue Growth" + ], + "connect_to": [ + { + "target": "6482acbb-d24c-5066-91cd-ab73b854838a", + "relation": "embeds", + "ref": "[images/image-2-Analyst, B.jpg]", + "position": { + "start": 1508, + "end": 1539 + } + } + ] + } + }, + { + "chunk_id": "90bb2dc3-92a6-57c6-971c-c3f5c8ddb00b", + "type": "image", + "content": "\nI am confident in their cash flow growing. And the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents across the world in enterprises everywhere. You're seeing incredible compute demand because of it. In this new world of AI, compute is revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues. So in this new world of AI, compute equals revenues. And I am certain that at this point, with the productive use of Codex and Claude Code and the excitement around Claude Cowork and, just the incredible enthusiasm about OpenClaw and the enterprise versions of them. All of the enterprise ISVs who are now working on agentic systems on top of their tools platforms, I am certain at this point that we are at the inflection point. We've reached the inflection point, and we're generating profitable tokens that are productive for customers and profitable for the cloud service providers. And so the simple logic of it, the simple way to think about it is computing has changed. What used to be software running on computers, modest amount of computers, call it \\$300 billion or \\$400 billion worth of CapEx each year has now gone into AI and AI in order to have, in order to generate tokens, you need compute capacity. And that translates directly to growth and that translates directly to revenues. Great. Thank you. And congratulations on the numbers. You talked about some of the strategic investments that you've made into Anthropic and potentially OpenAI [ph] Claude as well (26:23) but also partners Intel, Nokia. Synopsys. You're clearly at the center of everything. Can you talk about the role of those investments and kind of how do you view the balance sheet as a tool to kind of grow the NVIDIA's position, the ecosystem and, and participate in that growth.0\n[images/image-3-I am confi.jpg]\n", + "path": "images/image-3-I am confi.jpg", + "metadata": { + "length": 1906, + "summary": "image-3\nI am confident in their cash flow growing. And the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents across the world in enterprises everywhere. You're seeing incredible compute demand because of it. In this new world of AI, compute is revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues. So in this new world of AI, compute equals revenues. And I am certain that at this point, with the productive use of Codex and Claude Code and the excitement around Claude Cowork and, just the incredible enthusiasm about OpenClaw and the enterprise versions of them. All of the enterprise ISVs who are now working on agentic systems on top of their tools platforms, I am certain at this point that we are at the inflection point. We've reached the inflection point, and we're generating profitable tokens that are productive for customers and profitable for the cloud service providers. And so the simple logic of it, the simple way to think about it is computing has changed. What used to be software running on computers, modest amount of computers, call it \\$300 billion or \\$400 billion worth of CapEx each year has now gone into AI and AI in order to have, in order to generate tokens, you need compute capacity. And that translates directly to growth and that translates directly to revenues. Great. Thank you. And congratulations on the numbers. You talked about some of the strategic investments that you've made into Anthropic and potentially OpenAI [ph] Claude as well (26:23) but also partners Intel, Nokia. Synopsys. You're clearly at the center of everything. Can you talk about the role of those investments and kind of how do you view the balance sheet as a tool to kind of grow the NVIDIA's position, the ecosystem and, and participate in that growth.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-3-I am confi.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "26cc8720-f666-5b69-b61a-7dbb9a11dc71", + "type": "text", + "content": "Analyst, Morgan Stanley & Co. LLC\n\nI am confident in their cash flow growing. And the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents across the world in enterprises everywhere. You're seeing incredible compute demand because of it. In this new world of AI, compute is revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues. So in this new world of AI, compute equals revenues. And I am certain that at this point, with the productive use of Codex and Claude Code and the excitement around Claude Cowork and, just the incredible enthusiasm about OpenClaw and the enterprise versions of them. All of the enterprise ISVs who are now working on agentic systems on top of their tools platforms, I am certain at this point that we are at the inflection point. We've reached the inflection point, and we're generating profitable tokens that are productive for customers and profitable for the cloud service providers. And so the simple logic of it, the simple way to think about it is computing has changed. What used to be software running on computers, modest amount of computers, call it \\$300 billion or \\$400 billion worth of CapEx each year has now gone into AI and AI in order to have, in order to generate tokens, you need compute capacity. And that translates directly to growth and that translates directly to revenues. Great. Thank you. And congratulations on the numbers. You talked about some of the strategic investments that you've made into Anthropic and potentially OpenAI [ph] Claude as well (26:23) but also partners Intel, Nokia. Synopsys. You're clearly at the center of everything. Can you talk about the role of those investments and kind of how do you view the balance sheet as a tool to kind of grow the NVIDIA's position, the ecosystem and, and participate in that growth.0\n[images/image-3-I am confi.jpg]\n\nGreat. Thank you. And congratulations on the numbers. You talked about some of the strategic investments that you've made into Anthropic and potentially OpenAI [ph] Claude as well (26:23) but also partners Intel, Nokia. Synopsys. You're clearly at the center of everything. Can you talk about the role of those investments and kind of how do you view the balance sheet as a tool to kind of grow the NVIDIA's position, the ecosystem and, and participate in that growth.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Joe Moore", + "metadata": { + "length": 2409, + "summary": "The speaker expresses confidence in cash flow growth driven by the inflection of agentic AI and enterprise adoption. They argue that in the AI era, compute equals revenues, as generating tokens requires compute capacity. The discussion highlights strategic investments in companies like Anthropic and OpenAI, and views the balance sheet as a tool to strengthen NVIDIA's ecosystem position and participate in growth.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Morgan", + "Stanley", + "Co", + "LLC", + "confident", + "cash", + "flow", + "growing", + "And", + "reason", + "simple", + "We", + "inflection", + "agentic", + "AI", + "usefulness", + "agents", + "world", + "enterprises", + "You", + "incredible", + "compute", + "demand", + "In", + "revenues", + "Without", + "generate", + "tokens", + "grow", + "So", + "equals", + "point", + "productive", + "Codex", + "Claude", + "Code", + "excitement", + "Cowork", + "enthusiasm", + "OpenClaw", + "enterprise", + "versions", + "All", + "ISVs", + "working", + "systems", + "top", + "tools", + "platforms", + "ve", + "reached", + "generating", + "profitable", + "customers", + "cloud", + "service", + "providers", + "logic", + "computing", + "changed", + "What", + "software", + "running", + "computers", + "modest", + "amount", + "call", + "300", + "billion", + "400", + "worth", + "CapEx", + "year", + "order", + "capacity", + "translates", + "directly", + "growth", + "Great", + "Thank", + "congratulations", + "numbers", + "talked", + "strategic", + "investments", + "made", + "Anthropic", + "potentially", + "OpenAI", + "ph", + "26", + "23", + "partners", + "Intel", + "Nokia", + "Synopsys", + "center", + "Can", + "talk", + "role", + "kind", + "view", + "balance", + "sheet", + "tool", + "NVIDIA", + "position", + "ecosystem", + "participate", + "growth.0" + ], + "keywords": [ + "AI", + "compute", + "revenues" + ], + "connect_to": [ + { + "target": "90bb2dc3-92a6-57c6-971c-c3f5c8ddb00b", + "relation": "embeds", + "ref": "[images/image-3-I am confi.jpg]", + "position": { + "start": 1908, + "end": 1939 + } + } + ] + } + }, + { + "chunk_id": "9be796f9-431c-56f7-9108-78c8b58bb6c4", + "type": "image", + "content": "\nAnalyst, Morgan Stanley & Co. LLC As you know, fundamentally, at the core of everything NVIDIA is our ecosystem. That's what everybody loves about our business. The richness of our ecosystem, just about every startup in the world is working on NVIDIA's platform, we are in every cloud, we're in every on-prem data center, we're all over the world's edge and robotic systems, thousands of AI natives are built on top of NVIDIA. We want to take the great opportunity that we have as we're in the beginning of this new computing era, this new computing platform shift to put everybody on NVIDIA. Everything is already built on CUDA and so we're starting from a really terrific starting point. But as we build out the entire AI ecosystem, whether it's in AI for language or physical AI or AI physics or biology or robotics or manufacturing, we want all of these ecosystems to be built on top of NVIDIA. And this is such a wonderful opportunity for us to invest into the ecosystem across the entire stack.0\n[images/image-4-Analyst, M.jpg]\n", + "path": "images/image-4-Analyst, M.jpg", + "metadata": { + "length": 1035, + "summary": "image-4\nAnalyst, Morgan Stanley & Co. LLC As you know, fundamentally, at the core of everything NVIDIA is our ecosystem. That's what everybody loves about our business. The richness of our ecosystem, just about every startup in the world is working on NVIDIA's platform, we are in every cloud, we're in every on-prem data center, we're all over the world's edge and robotic systems, thousands of AI natives are built on top of NVIDIA. We want to take the great opportunity that we have as we're in the beginning of this new computing era, this new computing platform shift to put everybody on NVIDIA. Everything is already built on CUDA and so we're starting from a really terrific starting point. But as we build out the entire AI ecosystem, whether it's in AI for language or physical AI or AI physics or biology or robotics or manufacturing, we want all of these ecosystems to be built on top of NVIDIA. And this is such a wonderful opportunity for us to invest into the ecosystem across the entire stack.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-4-Analyst, M.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "df88a375-f6a8-5b52-b4dd-d517dfe22709", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nAnalyst, Morgan Stanley & Co. LLC As you know, fundamentally, at the core of everything NVIDIA is our ecosystem. That's what everybody loves about our business. The richness of our ecosystem, just about every startup in the world is working on NVIDIA's platform, we are in every cloud, we're in every on-prem data center, we're all over the world's edge and robotic systems, thousands of AI natives are built on top of NVIDIA. We want to take the great opportunity that we have as we're in the beginning of this new computing era, this new computing platform shift to put everybody on NVIDIA. Everything is already built on CUDA and so we're starting from a really terrific starting point. But as we build out the entire AI ecosystem, whether it's in AI for language or physical AI or AI physics or biology or robotics or manufacturing, we want all of these ecosystems to be built on top of NVIDIA. And this is such a wonderful opportunity for us to invest into the ecosystem across the entire stack.0\n[images/image-4-Analyst, M.jpg]\n\nAs you know, fundamentally, at the core of everything NVIDIA is our ecosystem. That's what everybody loves about our business. The richness of our ecosystem, just about every startup in the world is working on NVIDIA's platform, we are in every cloud, we're in every on-prem data center, we're all over the world's edge and robotic systems, thousands of AI natives are built on top of NVIDIA. We want to take the great opportunity that we have as we're in the beginning of this new computing era, this new computing platform shift to put everybody on NVIDIA. Everything is already built on CUDA and so we're starting from a really terrific starting point. But as we build out the entire AI ecosystem, whether it's in AI for language or physical AI or AI physics or biology or robotics or manufacturing, we want all of these ecosystems to be built on top of NVIDIA. And this is such a wonderful opportunity for us to invest into the ecosystem across the entire stack.\nOur ecosystem is also richer today than it used to be. We used to be largely a computing platform on GPUs, but now we're a computing AI infrastructure company, and we have computing platforms on, well, every aspect of that.\nAnd everything from computing to AI models to networking to our DPU, all of that has computing stacks on top of it. And as I mentioned before, whether it's an enterprise or in manufacturing, industrial or science or robotics, each one of these ecosystems have different stacks, and we want to make sure that we continue to invest into our ecosystem. So, our investments are focused very squarely, strategically on expanding and deepening our ecosystem reach.\nOperator: Your next question comes from Harlan Sur with JPMorgan. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_2", + "metadata": { + "length": 2842, + "summary": "The speaker, a co-founder and CEO of NVIDIA, emphasizes the company's ecosystem as its core strength. They highlight that startups, clouds, data centers, and AI systems are built on NVIDIA's platform, starting from CUDA. The goal is to expand this ecosystem across AI domains like language, physics, biology, robotics, and manufacturing. NVIDIA has evolved from a GPU computing platform to an AI infrastructure company, with stacks for computing, AI models, networking, and DPUs. Investments focus on deepening ecosystem reach across enterprise, industrial, science, and robotics sectors.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Analyst", + "Morgan", + "Stanley", + "LLC", + "As", + "fundamentally", + "core", + "ecosystem", + "That", + "loves", + "business", + "The", + "richness", + "startup", + "world", + "working", + "platform", + "cloud", + "prem", + "data", + "center", + "edge", + "robotic", + "systems", + "thousands", + "AI", + "natives", + "built", + "top", + "We", + "great", + "opportunity", + "beginning", + "computing", + "era", + "shift", + "put", + "Everything", + "CUDA", + "starting", + "terrific", + "point", + "But", + "build", + "entire", + "language", + "physical", + "physics", + "biology", + "robotics", + "manufacturing", + "ecosystems", + "And", + "wonderful", + "invest", + "stack.0", + "stack", + "Our", + "richer", + "today", + "largely", + "GPUs", + "infrastructure", + "company", + "platforms", + "aspect", + "models", + "networking", + "DPU", + "stacks", + "mentioned", + "enterprise", + "industrial", + "science", + "make", + "continue", + "So", + "investments", + "focused", + "squarely", + "strategically", + "expanding", + "deepening", + "reach", + "Operator", + "Your", + "question", + "Harlan", + "Sur", + "JPMorgan", + "line", + "open" + ], + "keywords": [ + "NVIDIA", + "ecosystem", + "AI" + ], + "connect_to": [ + { + "target": "9be796f9-431c-56f7-9108-78c8b58bb6c4", + "relation": "embeds", + "ref": "[images/image-4-Analyst, M.jpg]", + "position": { + "start": 1075, + "end": 1106 + } + } + ] + } + }, + { + "chunk_id": "c9666f6f-1178-5ce6-857f-552d80559bf8", + "type": "image", + "content": "\nOur ecosystem is also richer today than it used to be. We used to be largely a computing platform on GPUs, but now we're a computing AI infrastructure company, and we have computing platforms on, well, every aspect of that. Good afternoon. Thanks for taking my question. Networking continues to rise as a percentage of your overall data center profile, right. Through fiscal 2026, your networking revenues accelerated on a year-over-year basis every single quarter, right. With 3.6x growth, as you guys mentioned year-over-year growth in Q4, obviously, on the strength of your scale up and scale out networking product portfolio, I would seem to remember that first half of last year, your annualized run rate on your Spectrum-X Ethernet Switching platform was around \\$10 billion annualized. It looks like that may have stepped up to around \\$11 billion, \\$12 billion in the second half of last year.0\n[images/image-5-Our ecosys.jpg]\n", + "path": "images/image-5-Our ecosys.jpg", + "metadata": { + "length": 936, + "summary": "image-5\nOur ecosystem is also richer today than it used to be. We used to be largely a computing platform on GPUs, but now we're a computing AI infrastructure company, and we have computing platforms on, well, every aspect of that. Good afternoon. Thanks for taking my question. Networking continues to rise as a percentage of your overall data center profile, right. Through fiscal 2026, your networking revenues accelerated on a year-over-year basis every single quarter, right. With 3.6x growth, as you guys mentioned year-over-year growth in Q4, obviously, on the strength of your scale up and scale out networking product portfolio, I would seem to remember that first half of last year, your annualized run rate on your Spectrum-X Ethernet Switching platform was around \\$10 billion annualized. It looks like that may have stepped up to around \\$11 billion, \\$12 billion in the second half of last year.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-5-Our ecosys.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "97d5d0cc-5a8d-5730-b76f-f72cfc9794b2", + "type": "text", + "content": "Analyst, JPMorgan Securities LLC\n\nOur ecosystem is also richer today than it used to be. We used to be largely a computing platform on GPUs, but now we're a computing AI infrastructure company, and we have computing platforms on, well, every aspect of that. Good afternoon. Thanks for taking my question. Networking continues to rise as a percentage of your overall data center profile, right. Through fiscal 2026, your networking revenues accelerated on a year-over-year basis every single quarter, right. With 3.6x growth, as you guys mentioned year-over-year growth in Q4, obviously, on the strength of your scale up and scale out networking product portfolio, I would seem to remember that first half of last year, your annualized run rate on your Spectrum-X Ethernet Switching platform was around \\$10 billion annualized. It looks like that may have stepped up to around \\$11 billion, \\$12 billion in the second half of last year.0\n[images/image-5-Our ecosys.jpg]\n\nGood afternoon. Thanks for taking my question. Networking continues to rise as a percentage of your overall data center profile, right. Through fiscal 2026, your networking revenues accelerated on a year-over-year basis every single quarter, right. With 3.6x growth, as you guys mentioned year-over-year growth in Q4, obviously, on the strength of your scale up and scale out networking product portfolio, I would seem to remember that first half of last year, your annualized run rate on your Spectrum-X Ethernet Switching platform was around \\$10 billion annualized. It looks like that may have stepped up to around \\$11 billion, \\$12 billion in the second half of last year.\nJensen, looking at your order book, especially with Spectrum-XGS, upcoming 102T Spectrum-6 Switching platforms launching soon. Where is the Spectrum run rate trending now and as you foresee exiting sort of this calendar year?", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Harlan Sur", + "metadata": { + "length": 1873, + "summary": "The passage discusses the growth of networking revenues in NVIDIA's data center business, highlighting a 3.6x year-over-year increase in Q4 driven by the Spectrum-X Ethernet Switching platform. The annualized run rate for Spectrum-X is estimated to have risen from around $10 billion in the first half of last year to $11-12 billion in the second half. The speaker asks Jensen about the current trend of the Spectrum run rate, especially with upcoming products like Spectrum-XGS and the 102T Spectrum-6 Switching platform.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "JPMorgan", + "Securities", + "LLC", + "Our", + "ecosystem", + "richer", + "today", + "We", + "largely", + "computing", + "platform", + "GPUs", + "AI", + "infrastructure", + "company", + "platforms", + "aspect", + "Good", + "afternoon", + "Thanks", + "taking", + "question", + "Networking", + "continues", + "rise", + "percentage", + "data", + "center", + "profile", + "Through", + "fiscal", + "2026", + "networking", + "revenues", + "accelerated", + "year", + "basis", + "single", + "quarter", + "With", + "3.6", + "growth", + "guys", + "mentioned", + "Q4", + "strength", + "scale", + "product", + "portfolio", + "remember", + "half", + "annualized", + "run", + "rate", + "Spectrum", + "Ethernet", + "Switching", + "10", + "billion", + "It", + "stepped", + "11", + "12", + "year.0", + "Jensen", + "order", + "book", + "XGS", + "upcoming", + "102T", + "launching", + "Where", + "trending", + "foresee", + "exiting", + "sort", + "calendar" + ], + "keywords": [ + "Networking", + "Spectrum-X", + "Revenue Growth" + ], + "connect_to": [ + { + "target": "c9666f6f-1178-5ce6-857f-552d80559bf8", + "relation": "embeds", + "ref": "[images/image-5-Our ecosys.jpg]", + "position": { + "start": 937, + "end": 968 + } + } + ] + } + }, + { + "chunk_id": "98135a8b-2eb0-5179-b28c-7ad8b87a38a9", + "type": "image", + "content": "\nGood afternoon. Thanks for taking my question. Networking continues to rise as a percentage of your overall data center profile, right. Through fiscal 2026, your networking revenues accelerated on a year-over-year basis every single quarter, right. With 3.6x growth, as you guys mentioned year-over-year growth in Q4, obviously, on the strength of your scale up and scale out networking product portfolio, I would seem to remember that first half of last year, your annualized run rate on your Spectrum-X Ethernet Switching platform was around \\$10 billion annualized. It looks like that may have stepped up to around \\$11 billion, \\$12 billion in the second half of last year. Yeah. As you know, we see ourselves as an AI infrastructure company and the AI computing infrastructure includes CPUs, GPUs, and we invented NVLink to scale up the one computing node into a giant computing rack.0\n[images/image-6-Good after.jpg]\n", + "path": "images/image-6-Good after.jpg", + "metadata": { + "length": 924, + "summary": "image-6\nGood afternoon. Thanks for taking my question. Networking continues to rise as a percentage of your overall data center profile, right. Through fiscal 2026, your networking revenues accelerated on a year-over-year basis every single quarter, right. With 3.6x growth, as you guys mentioned year-over-year growth in Q4, obviously, on the strength of your scale up and scale out networking product portfolio, I would seem to remember that first half of last year, your annualized run rate on your Spectrum-X Ethernet Switching platform was around \\$10 billion annualized. It looks like that may have stepped up to around \\$11 billion, \\$12 billion in the second half of last year. Yeah. As you know, we see ourselves as an AI infrastructure company and the AI computing infrastructure includes CPUs, GPUs, and we invented NVLink to scale up the one computing node into a giant computing rack.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-6-Good after.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "09c7f944-c571-5615-9cf4-fd12f45c86ca", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nGood afternoon. Thanks for taking my question. Networking continues to rise as a percentage of your overall data center profile, right. Through fiscal 2026, your networking revenues accelerated on a year-over-year basis every single quarter, right. With 3.6x growth, as you guys mentioned year-over-year growth in Q4, obviously, on the strength of your scale up and scale out networking product portfolio, I would seem to remember that first half of last year, your annualized run rate on your Spectrum-X Ethernet Switching platform was around \\$10 billion annualized. It looks like that may have stepped up to around \\$11 billion, \\$12 billion in the second half of last year. Yeah. As you know, we see ourselves as an AI infrastructure company and the AI computing infrastructure includes CPUs, GPUs, and we invented NVLink to scale up the one computing node into a giant computing rack.0\n[images/image-6-Good after.jpg]\n\nYeah. As you know, we see ourselves as an AI infrastructure company and the AI computing infrastructure includes CPUs, GPUs, and we invented NVLink to scale up the one computing node into a giant computing rack.\nWe invented the idea of a rack scale computer. We don't ship nodes of computers, we ship racks of computers. And those – that NVLink Switch scale up system is then scaled out using Spectrum-X and InfiniBand, we support both. And then further, we also scale across data centers using Spectrum-X scale across.\nAnd so the way we think about networking is really an extension, it's – we offer everything openly, so that people could decide to mix and match in different scale, and however they would like to integrate it into their bespoke data center. But in the final analysis, it's all one big part of our platform.\nAnd the invention of NVLink really turbocharged our networking business. Every rack comes with nine nodes of switches, and each one of them has two chips in it and in the future they'll have more. And so the amount of switching that we do per rack is really quite incredible.\nWe're also now the largest networking company in the world. And if you look at Ethernet, we came into the Ethernet market about a couple of years ago into Ethernet switching. And I think that we're probably the largest Ethernet networking company in the world today, and surely will be soon. And so Spectrum-X Ethernet has been a home run for us. But we're open to however people want to do networking. Some people just really love the low latency and the scale up capability of InfiniBand. And we will continue to support that, of course. And some people love to integrate their networking across their data center based on Ethernet. And we created an Ethernet capability that extends Ethernet with artificial intelligence, way of processing in the data center and we're incredibly good at that and our Spectrum-X performance really shows it.\nThe difference of when you built a \\$10 billion or \\$20 billion AI factory, the difference of 10%, and it could be easily 20% on the effectiveness and the utilization of your network for your data center that translates to real money. And so NVIDIA's networking business is really, really growing fast. And it's, I think it's just because we built the AI infrastructure so effectively and the AI infrastructure business is growing incredibly fast.\nOperator: Your next question comes from C.J. Muse with Cantor Fitzgerald. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_3", + "metadata": { + "length": 3484, + "summary": "NVIDIA positions itself as an AI infrastructure company, with networking as a key growth area. The company invented NVLink for rack-scale computing and uses Spectrum-X and InfiniBand for scaling out. Networking revenue is accelerating, with Spectrum-X Ethernet being a major success. NVIDIA is now a leading networking company, and its AI infrastructure business is growing rapidly due to the significant value of network performance improvements in large AI factories.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Good", + "afternoon", + "Thanks", + "taking", + "question", + "Networking", + "continues", + "rise", + "percentage", + "data", + "center", + "profile", + "Through", + "fiscal", + "2026", + "networking", + "revenues", + "accelerated", + "year", + "basis", + "single", + "quarter", + "With", + "3.6", + "growth", + "guys", + "mentioned", + "Q4", + "strength", + "scale", + "product", + "portfolio", + "remember", + "half", + "annualized", + "run", + "rate", + "Spectrum", + "Ethernet", + "Switching", + "platform", + "10", + "billion", + "It", + "stepped", + "11", + "12", + "Yeah", + "As", + "AI", + "infrastructure", + "company", + "computing", + "includes", + "CPUs", + "GPUs", + "invented", + "NVLink", + "node", + "giant", + "rack.0", + "rack", + "We", + "idea", + "computer", + "don", + "ship", + "nodes", + "computers", + "racks", + "And", + "Switch", + "system", + "scaled", + "InfiniBand", + "support", + "centers", + "extension", + "offer", + "openly", + "people", + "decide", + "mix", + "match", + "integrate", + "bespoke", + "But", + "final", + "analysis", + "big", + "part", + "invention", + "turbocharged", + "business", + "Every", + "switches", + "chips", + "future", + "ll", + "amount", + "switching", + "incredible", + "largest", + "world", + "market", + "couple", + "years", + "ago", + "today", + "surely", + "home", + "open", + "Some", + "love", + "low", + "latency", + "capability", + "continue", + "based", + "created", + "extends", + "artificial", + "intelligence", + "processing", + "incredibly", + "good", + "performance", + "shows", + "The", + "difference", + "built", + "20", + "factory", + "10%", + "easily", + "20%", + "effectiveness", + "utilization", + "network", + "translates", + "real", + "money", + "growing", + "fast", + "effectively", + "Operator", + "Your", + "Muse", + "Cantor", + "Fitzgerald", + "line" + ], + "keywords": [ + "AI infrastructure", + "Networking", + "Spectrum-X" + ], + "connect_to": [ + { + "target": "98135a8b-2eb0-5179-b28c-7ad8b87a38a9", + "relation": "embeds", + "ref": "[images/image-6-Good after.jpg]", + "position": { + "start": 964, + "end": 995 + } + } + ] + } + }, + { + "chunk_id": "fb3a920e-0c4d-5f9c-8c05-31684e1549bf", + "type": "image", + "content": "\nWe're also now the largest networking company in the world. And if you look at Ethernet, we came into the Ethernet market about a couple of years ago into Ethernet switching. And I think that we're probably the largest Ethernet networking company in the world today, and surely will be soon. And so Spectrum-X Ethernet has been a home run for us. But we're open to however people want to do networking. Some people just really love the low latency and the scale up capability of InfiniBand. And we will continue to support that, of course. And some people love to integrate their networking across their data center based on Ethernet. And we created an Ethernet capability that extends Ethernet with artificial intelligence, way of processing in the data center and we're incredibly good at that and our Spectrum-X performance really shows it. Yeah. Good afternoon. Thank you for taking the question. I guess with CPX for large context Windows and Groq likely adding a decode specific solution. Curious how we should think about your future roadmap. Should we be thinking about customized silicon either by workload or customer, as an increasing focus by NVIDIA, particularly helped by your move to a dielet architecture? Thanks so much.0\n[images/image-7-We're also.jpg]\n", + "path": "images/image-7-We're also.jpg", + "metadata": { + "length": 1272, + "summary": "image-7\nWe're also now the largest networking company in the world. And if you look at Ethernet, we came into the Ethernet market about a couple of years ago into Ethernet switching. And I think that we're probably the largest Ethernet networking company in the world today, and surely will be soon. And so Spectrum-X Ethernet has been a home run for us. But we're open to however people want to do networking. Some people just really love the low latency and the scale up capability of InfiniBand. And we will continue to support that, of course. And some people love to integrate their networking across their data center based on Ethernet. And we created an Ethernet capability that extends Ethernet with artificial intelligence, way of processing in the data center and we're incredibly good at that and our Spectrum-X performance really shows it. Yeah. Good afternoon. Thank you for taking the question. I guess with CPX for large context Windows and Groq likely adding a decode specific solution. Curious how we should think about your future roadmap. Should we be thinking about customized silicon either by workload or customer, as an increasing focus by NVIDIA, particularly helped by your move to a dielet architecture? Thanks so much.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-7-We're also.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ceade533-367b-5026-86a8-229b591ed1bc", + "type": "text", + "content": "Analyst, Cantor Fitzgerald & Co.\n\nWe're also now the largest networking company in the world. And if you look at Ethernet, we came into the Ethernet market about a couple of years ago into Ethernet switching. And I think that we're probably the largest Ethernet networking company in the world today, and surely will be soon. And so Spectrum-X Ethernet has been a home run for us. But we're open to however people want to do networking. Some people just really love the low latency and the scale up capability of InfiniBand. And we will continue to support that, of course. And some people love to integrate their networking across their data center based on Ethernet. And we created an Ethernet capability that extends Ethernet with artificial intelligence, way of processing in the data center and we're incredibly good at that and our Spectrum-X performance really shows it. Yeah. Good afternoon. Thank you for taking the question. I guess with CPX for large context Windows and Groq likely adding a decode specific solution. Curious how we should think about your future roadmap. Should we be thinking about customized silicon either by workload or customer, as an increasing focus by NVIDIA, particularly helped by your move to a dielet architecture? Thanks so much.0\n[images/image-7-We're also.jpg]\n\nYeah. Good afternoon. Thank you for taking the question. I guess with CPX for large context Windows and Groq likely adding a decode specific solution. Curious how we should think about your future roadmap. Should we be thinking about customized silicon either by workload or customer, as an increasing focus by NVIDIA, particularly helped by your move to a dielet architecture? Thanks so much.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/C.J. Muse", + "metadata": { + "length": 1699, + "summary": "The passage discusses NVIDIA's dominance in networking, highlighting their success with Spectrum-X Ethernet and continued support for InfiniBand. It also addresses a question about future roadmap, specifically regarding customized silicon for workloads or customers, leveraging a dielet architecture.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Cantor", + "Fitzgerald", + "Co", + "We", + "largest", + "networking", + "company", + "world", + "And", + "Ethernet", + "market", + "couple", + "years", + "ago", + "switching", + "today", + "surely", + "Spectrum", + "home", + "run", + "But", + "open", + "people", + "Some", + "love", + "low", + "latency", + "scale", + "capability", + "InfiniBand", + "continue", + "support", + "integrate", + "data", + "center", + "based", + "created", + "extends", + "artificial", + "intelligence", + "processing", + "incredibly", + "good", + "performance", + "shows", + "Yeah", + "Good", + "afternoon", + "Thank", + "taking", + "question", + "guess", + "CPX", + "large", + "context", + "Windows", + "Groq", + "adding", + "decode", + "specific", + "solution", + "Curious", + "future", + "roadmap", + "Should", + "thinking", + "customized", + "silicon", + "workload", + "customer", + "increasing", + "focus", + "NVIDIA", + "helped", + "move", + "dielet", + "architecture", + "Thanks", + "much.0" + ], + "keywords": [ + "NVIDIA", + "Ethernet", + "InfiniBand" + ], + "connect_to": [ + { + "target": "fb3a920e-0c4d-5f9c-8c05-31684e1549bf", + "relation": "embeds", + "ref": "[images/image-7-We're also.jpg]", + "position": { + "start": 1273, + "end": 1304 + } + } + ] + } + }, + { + "chunk_id": "23463bd0-e7b3-54b7-89fd-16eed0fcca01", + "type": "image", + "content": "\nAnalyst, Cantor Fitzgerald & Co. We don't use – we want to – everybody should want to extend, push out dielet as long as they can. And the reason for that is because every time you cross a dielet you have a dielet, you have to cross an interface. Every time you cross an interface, you add latency, you add power unnecessarily. We're not allergic to dielet. We use dielets already. But we try to use dielets only when we absolutely have no choice but to do so. And so we – if you look at the Grace Blackwell architecture and the Rubin architecture, we used two giant reticle limited dies [ph] and we abut them (33:53) and that reduces the amount of architecture crossing.0\n[images/image-8-Analyst, C.jpg]\n", + "path": "images/image-8-Analyst, C.jpg", + "metadata": { + "length": 706, + "summary": "image-8\nAnalyst, Cantor Fitzgerald & Co. We don't use – we want to – everybody should want to extend, push out dielet as long as they can. And the reason for that is because every time you cross a dielet you have a dielet, you have to cross an interface. Every time you cross an interface, you add latency, you add power unnecessarily. We're not allergic to dielet. We use dielets already. But we try to use dielets only when we absolutely have no choice but to do so. And so we – if you look at the Grace Blackwell architecture and the Rubin architecture, we used two giant reticle limited dies [ph] and we abut them (33:53) and that reduces the amount of architecture crossing.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-8-Analyst, C.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "64da31c0-bb30-50f2-90fe-9075d36b1300", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nAnalyst, Cantor Fitzgerald & Co. We don't use – we want to – everybody should want to extend, push out dielet as long as they can. And the reason for that is because every time you cross a dielet you have a dielet, you have to cross an interface. Every time you cross an interface, you add latency, you add power unnecessarily. We're not allergic to dielet. We use dielets already. But we try to use dielets only when we absolutely have no choice but to do so. And so we – if you look at the Grace Blackwell architecture and the Rubin architecture, we used two giant reticle limited dies [ph] and we abut them (33:53) and that reduces the amount of architecture crossing.0\n[images/image-8-Analyst, C.jpg]\n\nWe don't use – we want to – everybody should want to extend, push out dielet as long as they can. And the reason for that is because every time you cross a dielet you have a dielet, you have to cross an interface. Every time you cross an interface, you add latency, you add power unnecessarily. We're not allergic to dielet. We use dielets already. But we try to use dielets only when we absolutely have no choice but to do so. And so we – if you look at the Grace Blackwell architecture and the Rubin architecture, we used two giant reticle limited dies [ph] and we abut them (33:53) and that reduces the amount of architecture crossing.\nThe [ph] dielet tax (33:59) shows up in the architecture effectiveness of the competitors. If you look at NVIDIA, people call it our software advantage. But where software starts and architecture starts and ends is kind of hard to tell. It's, our software is effective because our architecture is so good. And so the CUDA architecture is unquestionably more effective, more efficient, delivers more performance per flop, per watt than any computing architecture out there and it's because of the way we architect.\nWith respect to, how we think about Groq and the low latency decoder, I've got some great ideas that I'd like to share with you at GTC, but the simple idea is that our infrastructure is incredibly versatile because of CUDA. And we're going to continue to do that. All of our GPUs are architecturally compatible, which means that when I'm working on optimizing models today for Blackwell, all of that work and all of that dedication to optimizing software stacks and new models, also benefit Hopper and also benefit Ampere. It's the reason why A100 continues to feel fresh and continues to stay performant years after we've deployed it into the world. Architecture compatibility allows us to do that. It allows us to invest enormously in software engineering and optimization, knowing that our entire installed base in the cloud, on-prem, everywhere from generations of architectures of GPUs will all benefit. And so we'll continue to do that.\nAnd allows us to extend the useful life, allows us to have innovation, flexibility, and velocity, which translates the performance and very importantly, performance per dollar and performance per watt for our customers. And so what we'll do with Groq is, you'll come to see GTC, but what we'll do is we'll extend our architecture with Groq as an accelerator in very much the ways that we extended NVIDIA's architecture with Mellanox.\nOperator: The next question comes from Stacy Rasgon with Bernstein Research. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_4", + "metadata": { + "length": 3404, + "summary": "NVIDIA's CEO discusses the company's approach to dielet usage, preferring to minimize dielet crossings to reduce latency and power. He contrasts this with competitors, highlighting NVIDIA's architectural and software advantages through CUDA. The architecture compatibility across GPU generations (A100, Hopper, Blackwell) ensures software optimizations benefit the entire installed base, improving performance per dollar and per watt. He also hints at extending architecture with Groq as an accelerator, similar to the Mellanox integration.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Analyst", + "Cantor", + "Fitzgerald", + "We", + "don", + "extend", + "push", + "dielet", + "long", + "And", + "reason", + "time", + "cross", + "interface", + "Every", + "add", + "latency", + "power", + "unnecessarily", + "allergic", + "dielets", + "But", + "absolutely", + "choice", + "Grace", + "Blackwell", + "architecture", + "Rubin", + "giant", + "reticle", + "limited", + "dies", + "ph", + "abut", + "33", + "53", + "reduces", + "amount", + "crossing.0", + "crossing", + "The", + "tax", + "59", + "shows", + "effectiveness", + "competitors", + "If", + "people", + "call", + "software", + "advantage", + "starts", + "ends", + "kind", + "hard", + "It", + "effective", + "good", + "CUDA", + "unquestionably", + "efficient", + "delivers", + "performance", + "flop", + "watt", + "computing", + "architect", + "With", + "respect", + "Groq", + "low", + "decoder", + "ve", + "great", + "ideas", + "share", + "GTC", + "simple", + "idea", + "infrastructure", + "incredibly", + "versatile", + "continue", + "All", + "GPUs", + "architecturally", + "compatible", + "means", + "working", + "optimizing", + "models", + "today", + "work", + "dedication", + "stacks", + "benefit", + "Hopper", + "Ampere", + "A100", + "continues", + "feel", + "fresh", + "stay", + "performant", + "years", + "deployed", + "world", + "Architecture", + "compatibility", + "invest", + "enormously", + "engineering", + "optimization", + "knowing", + "entire", + "installed", + "base", + "cloud", + "prem", + "generations", + "architectures", + "ll", + "life", + "innovation", + "flexibility", + "velocity", + "translates", + "importantly", + "dollar", + "customers", + "accelerator", + "ways", + "extended", + "Mellanox", + "Operator", + "question", + "Stacy", + "Rasgon", + "Bernstein", + "Research", + "Your", + "line", + "open" + ], + "keywords": [ + "dielet", + "architecture", + "CUDA" + ], + "connect_to": [ + { + "target": "23463bd0-e7b3-54b7-89fd-16eed0fcca01", + "relation": "embeds", + "ref": "[images/image-8-Analyst, C.jpg]", + "position": { + "start": 746, + "end": 777 + } + } + ] + } + }, + { + "chunk_id": "729262d1-c4b6-54d9-b115-752c4d0c5a3d", + "type": "image", + "content": "\nWith respect to, how we think about Groq and the low latency decoder, I've got some great ideas that I'd like to share with you at GTC, but the simple idea is that our infrastructure is incredibly versatile because of CUDA. And we're going to continue to do that. All of our GPUs are architecturally compatible, which means that when I'm working on optimizing models today for Blackwell, all of that work and all of that dedication to optimizing software stacks and new models, also benefit Hopper and also benefit Ampere. It's the reason why A100 continues to feel fresh and continues to stay performant years after we've deployed it into the world. Architecture compatibility allows us to do that. It allows us to invest enormously in software engineering and optimization, knowing that our entire installed base in the cloud, on-prem, everywhere from generations of architectures of GPUs will all benefit. And so we'll continue to do that. Hi, guys. Thanks for taking my questions. Colette, I wanted to dig a little bit into the call for sequential growth through the year. So, I mean, you grew this quarter more than \\$10 billion sequentially in data center, and the guide seems to imply the bulk of the increase \\$10 billion sequential in data centers. So, how do you see that as we go through the year, especially as Rubin ramps into the back half? Blackwell has been a pretty massive acceleration with a sequential growth. Should we expect something similar as we get to Rubin?0\n[images/image-9-With respe.jpg]\n", + "path": "images/image-9-With respe.jpg", + "metadata": { + "length": 1519, + "summary": "image-9\nWith respect to, how we think about Groq and the low latency decoder, I've got some great ideas that I'd like to share with you at GTC, but the simple idea is that our infrastructure is incredibly versatile because of CUDA. And we're going to continue to do that. All of our GPUs are architecturally compatible, which means that when I'm working on optimizing models today for Blackwell, all of that work and all of that dedication to optimizing software stacks and new models, also benefit Hopper and also benefit Ampere. It's the reason why A100 continues to feel fresh and continues to stay performant years after we've deployed it into the world. Architecture compatibility allows us to do that. It allows us to invest enormously in software engineering and optimization, knowing that our entire installed base in the cloud, on-prem, everywhere from generations of architectures of GPUs will all benefit. And so we'll continue to do that. Hi, guys. Thanks for taking my questions. Colette, I wanted to dig a little bit into the call for sequential growth through the year. So, I mean, you grew this quarter more than \\$10 billion sequentially in data center, and the guide seems to imply the bulk of the increase \\$10 billion sequential in data centers. So, how do you see that as we go through the year, especially as Rubin ramps into the back half? Blackwell has been a pretty massive acceleration with a sequential growth. Should we expect something similar as we get to Rubin?0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-9-With respe.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "19c19dcb-381c-5faa-a3fa-28ed619074a3", + "type": "text", + "content": "Analyst, Bernstein Research\n\nWith respect to, how we think about Groq and the low latency decoder, I've got some great ideas that I'd like to share with you at GTC, but the simple idea is that our infrastructure is incredibly versatile because of CUDA. And we're going to continue to do that. All of our GPUs are architecturally compatible, which means that when I'm working on optimizing models today for Blackwell, all of that work and all of that dedication to optimizing software stacks and new models, also benefit Hopper and also benefit Ampere. It's the reason why A100 continues to feel fresh and continues to stay performant years after we've deployed it into the world. Architecture compatibility allows us to do that. It allows us to invest enormously in software engineering and optimization, knowing that our entire installed base in the cloud, on-prem, everywhere from generations of architectures of GPUs will all benefit. And so we'll continue to do that. Hi, guys. Thanks for taking my questions. Colette, I wanted to dig a little bit into the call for sequential growth through the year. So, I mean, you grew this quarter more than \\$10 billion sequentially in data center, and the guide seems to imply the bulk of the increase \\$10 billion sequential in data centers. So, how do you see that as we go through the year, especially as Rubin ramps into the back half? Blackwell has been a pretty massive acceleration with a sequential growth. Should we expect something similar as we get to Rubin?0\n[images/image-9-With respe.jpg]\n\nHi, guys. Thanks for taking my questions. Colette, I wanted to dig a little bit into the call for sequential growth through the year. So, I mean, you grew this quarter more than \\$10 billion sequentially in data center, and the guide seems to imply the bulk of the increase \\$10 billion sequential in data centers. So, how do you see that as we go through the year, especially as Rubin ramps into the back half? Blackwell has been a pretty massive acceleration with a sequential growth. Should we expect something similar as we get to Rubin?\nAnd then I was also just hoping you could comment on, your expectations for gaming? I understand the memory issues and everything else. Do you think gaming can still grow year-over-year in fiscal 2027, or will that be under more pressure given memory? So, those two questions, please. Thank you.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Stacy A. Rasgon", + "metadata": { + "length": 2385, + "summary": "The passage discusses NVIDIA's GPU architecture compatibility, emphasizing how CUDA enables infrastructure versatility. The speaker notes that architectural compatibility across GPU generations (Blackwell, Hopper, Ampere) allows software optimizations to benefit the entire installed base, keeping older GPUs like A100 performant. The second part addresses a question about sequential growth in data center revenue, noting a $10 billion quarterly increase and asking about similar acceleration for the upcoming Rubin architecture. A follow-up question asks about gaming revenue expectations for fiscal 2027 given memory constraints.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Bernstein", + "Research", + "With", + "respect", + "Groq", + "low", + "latency", + "decoder", + "ve", + "great", + "ideas", + "share", + "GTC", + "simple", + "idea", + "infrastructure", + "incredibly", + "versatile", + "CUDA", + "And", + "continue", + "All", + "GPUs", + "architecturally", + "compatible", + "means", + "working", + "optimizing", + "models", + "today", + "Blackwell", + "work", + "dedication", + "software", + "stacks", + "benefit", + "Hopper", + "Ampere", + "It", + "reason", + "A100", + "continues", + "feel", + "fresh", + "stay", + "performant", + "years", + "deployed", + "world", + "Architecture", + "compatibility", + "invest", + "enormously", + "engineering", + "optimization", + "knowing", + "entire", + "installed", + "base", + "cloud", + "prem", + "generations", + "architectures", + "ll", + "Hi", + "guys", + "Thanks", + "taking", + "questions", + "Colette", + "wanted", + "dig", + "bit", + "call", + "sequential", + "growth", + "year", + "So", + "grew", + "quarter", + "10", + "billion", + "sequentially", + "data", + "center", + "guide", + "imply", + "bulk", + "increase", + "centers", + "Rubin", + "ramps", + "back", + "half", + "pretty", + "massive", + "acceleration", + "Should", + "expect", + "similar", + "hoping", + "comment", + "expectations", + "gaming", + "understand", + "memory", + "issues", + "Do", + "grow", + "fiscal", + "2027", + "pressure", + "Thank" + ], + "keywords": [ + "CUDA", + "GPU", + "Data Center" + ], + "connect_to": [ + { + "target": "729262d1-c4b6-54d9-b115-752c4d0c5a3d", + "relation": "embeds", + "ref": "[images/image-9-With respe.jpg]", + "position": { + "start": 1515, + "end": 1546 + } + } + ] + } + }, + { + "chunk_id": "78f76571-185e-5670-9d9b-86a6c03d03e2", + "type": "image", + "content": "\nHi, guys. Thanks for taking my questions. Colette, I wanted to dig a little bit into the call for sequential growth through the year. So, I mean, you grew this quarter more than \\$10 billion sequentially in data center, and the guide seems to imply the bulk of the increase \\$10 billion sequential in data centers. So, how do you see that as we go through the year, especially as Rubin ramps into the back half? Blackwell has been a pretty massive acceleration with a sequential growth. Should we expect something similar as we get to Rubin? Thanks, Stacy. Let me start with the revenue going forward. Again, we're trying to look at revenue quarter-by-quarter. As you think about the full year, we are absolutely going to be still selling and providing Blackwell probably at the same time that we're also seeing Vera Rubin come to market.0\n[images/image-10-Hi, guys. .jpg]\n", + "path": "images/image-10-Hi, guys. .jpg", + "metadata": { + "length": 874, + "summary": "image-10\nHi, guys. Thanks for taking my questions. Colette, I wanted to dig a little bit into the call for sequential growth through the year. So, I mean, you grew this quarter more than \\$10 billion sequentially in data center, and the guide seems to imply the bulk of the increase \\$10 billion sequential in data centers. So, how do you see that as we go through the year, especially as Rubin ramps into the back half? Blackwell has been a pretty massive acceleration with a sequential growth. Should we expect something similar as we get to Rubin? Thanks, Stacy. Let me start with the revenue going forward. Again, we're trying to look at revenue quarter-by-quarter. As you think about the full year, we are absolutely going to be still selling and providing Blackwell probably at the same time that we're also seeing Vera Rubin come to market.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-10-Hi, guys. .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ff820dba-a764-5c8a-959e-cc155b3299d5", + "type": "text", + "content": "Chief Financial Officer & Executive Vice President, NVIDIA Corp.\n\nHi, guys. Thanks for taking my questions. Colette, I wanted to dig a little bit into the call for sequential growth through the year. So, I mean, you grew this quarter more than \\$10 billion sequentially in data center, and the guide seems to imply the bulk of the increase \\$10 billion sequential in data centers. So, how do you see that as we go through the year, especially as Rubin ramps into the back half? Blackwell has been a pretty massive acceleration with a sequential growth. Should we expect something similar as we get to Rubin? Thanks, Stacy. Let me start with the revenue going forward. Again, we're trying to look at revenue quarter-by-quarter. As you think about the full year, we are absolutely going to be still selling and providing Blackwell probably at the same time that we're also seeing Vera Rubin come to market.0\n[images/image-10-Hi, guys. .jpg]\n\nThanks, Stacy. Let me start with the revenue going forward. Again, we're trying to look at revenue quarter-by-quarter. As you think about the full year, we are absolutely going to be still selling and providing Blackwell probably at the same time that we're also seeing Vera Rubin come to market.\nThis is a very great architecture that helps them just today quickly standing up and have already planned on many different orders across the different customers to provide that.\nIt's too early yet to determine how much in terms of that Vera Rubin, that beginning ramp will start in the second half and we'll get through it. But no confusion in terms of the strong demand and the interest. We do expect pretty much every single customer to be purchasing Vera Rubin. The question is, are, how soon are we in market and how soon are they able to stand that up in terms of in their data centers? That was your first part.\nThe second part was focusing on our gaming. As much as we would love to have additional more supply, we do believe for a couple quarters, it is going to be very tight. If things improve by the end of the year, there is an opportunity to think about what that is from a year-over-year growth. But it's still too early for us to know at this time. And we'll get back to you as soon as we can.\nOperator: Your next question comes from Atif Malik with Citi. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Colette M. Kress", + "metadata": { + "length": 2327, + "summary": "The CFO discusses sequential growth expectations for NVIDIA, noting a $10 billion sequential increase in data center revenue. Blackwell sales will continue alongside the ramp of Vera Rubin in the second half of the year, with strong customer demand expected. Gaming supply remains tight for a few quarters, with potential year-over-year growth later in the year.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Chief", + "Financial", + "Officer", + "Executive", + "Vice", + "President", + "NVIDIA", + "Corp", + "Hi", + "guys", + "Thanks", + "taking", + "questions", + "Colette", + "wanted", + "dig", + "bit", + "call", + "sequential", + "growth", + "year", + "So", + "grew", + "quarter", + "10", + "billion", + "sequentially", + "data", + "center", + "guide", + "imply", + "bulk", + "increase", + "centers", + "Rubin", + "ramps", + "back", + "half", + "Blackwell", + "pretty", + "massive", + "acceleration", + "Should", + "expect", + "similar", + "Stacy", + "Let", + "start", + "revenue", + "forward", + "Again", + "As", + "full", + "absolutely", + "selling", + "providing", + "time", + "Vera", + "market.0", + "market", + "This", + "great", + "architecture", + "helps", + "today", + "quickly", + "standing", + "planned", + "orders", + "customers", + "provide", + "It", + "early", + "determine", + "terms", + "beginning", + "ramp", + "ll", + "But", + "confusion", + "strong", + "demand", + "interest", + "We", + "single", + "customer", + "purchasing", + "The", + "question", + "stand", + "That", + "part", + "focusing", + "gaming", + "love", + "additional", + "supply", + "couple", + "quarters", + "tight", + "If", + "things", + "improve", + "end", + "opportunity", + "And", + "Operator", + "Your", + "Atif", + "Malik", + "Citi", + "line", + "open" + ], + "keywords": [ + "NVIDIA", + "Blackwell", + "Vera Rubin" + ], + "connect_to": [ + { + "target": "78f76571-185e-5670-9d9b-86a6c03d03e2", + "relation": "embeds", + "ref": "[images/image-10-Hi, guys. .jpg]", + "position": { + "start": 906, + "end": 938 + } + } + ] + } + }, + { + "chunk_id": "44889a26-b9a1-54b6-8940-974999c65aaa", + "type": "image", + "content": "\nIt's too early yet to determine how much in terms of that Vera Rubin, that beginning ramp will start in the second half and we'll get through it. But no confusion in terms of the strong demand and the interest. We do expect pretty much every single customer to be purchasing Vera Rubin. The question is, are, how soon are we in market and how soon are they able to stand that up in terms of in their data centers? That was your first part. Thank you for taking my question. Jensen, I'm curious if you can touch on the importance of CUDA as now more of the investment dollars in AI are coming from inference workloads?0\n[images/image-11-It's too e.jpg]\n", + "path": "images/image-11-It's too e.jpg", + "metadata": { + "length": 653, + "summary": "image-11\nIt's too early yet to determine how much in terms of that Vera Rubin, that beginning ramp will start in the second half and we'll get through it. But no confusion in terms of the strong demand and the interest. We do expect pretty much every single customer to be purchasing Vera Rubin. The question is, are, how soon are we in market and how soon are they able to stand that up in terms of in their data centers? That was your first part. Thank you for taking my question. Jensen, I'm curious if you can touch on the importance of CUDA as now more of the investment dollars in AI are coming from inference workloads?0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-11-It's too e.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "8c9e3630-9720-5183-817d-c7b2aa7bebea", + "type": "text", + "content": "Analyst, Citigroup Global Markets, Inc.\n\nIt's too early yet to determine how much in terms of that Vera Rubin, that beginning ramp will start in the second half and we'll get through it. But no confusion in terms of the strong demand and the interest. We do expect pretty much every single customer to be purchasing Vera Rubin. The question is, are, how soon are we in market and how soon are they able to stand that up in terms of in their data centers? That was your first part. Thank you for taking my question. Jensen, I'm curious if you can touch on the importance of CUDA as now more of the investment dollars in AI are coming from inference workloads?0\n[images/image-11-It's too e.jpg]\n\nThank you for taking my question. Jensen, I'm curious if you can touch on the importance of CUDA as now more of the investment dollars in AI are coming from inference workloads?", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Atif Malik", + "metadata": { + "length": 871, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Citigroup", + "Global", + "Markets", + "Inc", + "It", + "early", + "determine", + "terms", + "Vera", + "Rubin", + "beginning", + "ramp", + "start", + "half", + "ll", + "But", + "confusion", + "strong", + "demand", + "interest", + "We", + "expect", + "pretty", + "single", + "customer", + "purchasing", + "The", + "question", + "market", + "stand", + "data", + "centers", + "That", + "part", + "Thank", + "taking", + "Jensen", + "curious", + "touch", + "importance", + "CUDA", + "investment", + "dollars", + "AI", + "coming", + "inference", + "workloads" + ], + "keywords": [], + "connect_to": [ + { + "target": "44889a26-b9a1-54b6-8940-974999c65aaa", + "relation": "embeds", + "ref": "[images/image-11-It's too e.jpg]", + "position": { + "start": 660, + "end": 692 + } + } + ] + } + }, + { + "chunk_id": "182ae2fb-e17b-5a69-8a3a-d8d52fdf74c3", + "type": "image", + "content": "\nAnalyst, Citigroup Global Markets, Inc. Without CUDA, we wouldn't know what to do with inference. The entire stack from TensorRT LLM that we introduced a few years ago, which is still the most performant inference stack in the world. Optimizing it for NVLink, requires us to discover and invent new parallelization algorithms that sits on top of CUDA, to distribute the workload and the inferencing to take advantage of the aggregate bandwidth across NVLink 72. NVLink 72 has enabled us to deliver generationally 50 times more performance per watt. It's just an incredible leap and it's sensible. NVLink 72 is a great invention. It was hard to do, the creation of the switching technology, disaggregating the switches, building the system racks, all of that. We did it all in plain sight and everybody knew how hard it was for us to do. And, but the results are incredible. So performance per watt is 50 times, performance per dollar, 35 times and so the leap in inference is incredible.0\n[images/image-12-Analyst, C.jpg]\n", + "path": "images/image-12-Analyst, C.jpg", + "metadata": { + "length": 1023, + "summary": "image-12\nAnalyst, Citigroup Global Markets, Inc. Without CUDA, we wouldn't know what to do with inference. The entire stack from TensorRT LLM that we introduced a few years ago, which is still the most performant inference stack in the world. Optimizing it for NVLink, requires us to discover and invent new parallelization algorithms that sits on top of CUDA, to distribute the workload and the inferencing to take advantage of the aggregate bandwidth across NVLink 72. NVLink 72 has enabled us to deliver generationally 50 times more performance per watt. It's just an incredible leap and it's sensible. NVLink 72 is a great invention. It was hard to do, the creation of the switching technology, disaggregating the switches, building the system racks, all of that. We did it all in plain sight and everybody knew how hard it was for us to do. And, but the results are incredible. So performance per watt is 50 times, performance per dollar, 35 times and so the leap in inference is incredible.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-12-Analyst, C.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f6b17ea3-85c6-5133-acae-fdd39315cbf7", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nAnalyst, Citigroup Global Markets, Inc. Without CUDA, we wouldn't know what to do with inference. The entire stack from TensorRT LLM that we introduced a few years ago, which is still the most performant inference stack in the world. Optimizing it for NVLink, requires us to discover and invent new parallelization algorithms that sits on top of CUDA, to distribute the workload and the inferencing to take advantage of the aggregate bandwidth across NVLink 72. NVLink 72 has enabled us to deliver generationally 50 times more performance per watt. It's just an incredible leap and it's sensible. NVLink 72 is a great invention. It was hard to do, the creation of the switching technology, disaggregating the switches, building the system racks, all of that. We did it all in plain sight and everybody knew how hard it was for us to do. And, but the results are incredible. So performance per watt is 50 times, performance per dollar, 35 times and so the leap in inference is incredible.0\n[images/image-12-Analyst, C.jpg]\n\nWithout CUDA, we wouldn't know what to do with inference. The entire stack from TensorRT LLM that we introduced a few years ago, which is still the most performant inference stack in the world. Optimizing it for NVLink, requires us to discover and invent new parallelization algorithms that sits on top of CUDA, to distribute the workload and the inferencing to take advantage of the aggregate bandwidth across NVLink 72. NVLink 72 has enabled us to deliver generationally 50 times more performance per watt. It's just an incredible leap and it's sensible. NVLink 72 is a great invention. It was hard to do, the creation of the switching technology, disaggregating the switches, building the system racks, all of that. We did it all in plain sight and everybody knew how hard it was for us to do. And, but the results are incredible. So performance per watt is 50 times, performance per dollar, 35 times and so the leap in inference is incredible.\nIt's very important – it's really important to realize that inference equals revenues now for our customers because agents are generating so many tokens and the results are so effective. When the agents are coding, it's off generating thousands, tens of thousands, hundreds of thousands because they're running for minutes to hours. And so these systems, these agentic systems are spawning off different agents working as a team.\nThe number of tokens that are being generated has really, really gone exponential. And so, we need to inference at a much higher speed. And when you're inferencing at a much higher speed, and each one of those tokens are dollarized, it directly translates into revenues. And so inference equals, inference performance equals revenues for our customers.\nFor the Data Centers, inference tokens per watt translates directly to the revenues of the CSPs. And the reason for that is because everybody is power limited. And so, I mean, no matter how many data centers you have, each data center, 100 megawatts or 1 gigawatt has power limits. So the architecture that has the best performance per watt translates because each token, the performance tokens per watt, each token is dollarized, tokens per watt translates to dollars per watt, which translates in a gigawatt directly to revenues.\nAnd so you could see that every CSP understands this now, every hyperscaler understands this, that CapEx translates to compute, compute with the right architecture translates to maximizing revenues and compute equals revenues. Without investing capacity today, without investing in compute, there cannot be revenue growth and that I think everybody understands. Compute equals revenues. Choosing the right architecture is incredibly important, is more than strategic now, it directly affects their earnings. And choosing the right architecture, the one with the best performance per watt is literally everything.\nOperator: Your next question comes from Ben Reitzes with Melius Research. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_5", + "metadata": { + "length": 4064, + "summary": "The speaker, a co-founder and CEO of NVIDIA, discusses the critical role of CUDA and NVLink 72 in AI inference. Without CUDA, inference would be impossible. NVLink 72 enables a 50x improvement in performance per watt and 35x in performance per dollar through new parallelization algorithms. Inference performance directly translates to revenue for customers, as token generation has become exponential. For data centers, inference tokens per watt equals dollars per watt, making architecture choice crucial for maximizing revenue under power constraints. The speaker emphasizes that compute equals revenue, and investing in the right architecture is strategic for earnings.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Analyst", + "Citigroup", + "Global", + "Markets", + "Inc", + "Without", + "CUDA", + "wouldn", + "inference", + "The", + "entire", + "stack", + "TensorRT", + "LLM", + "introduced", + "years", + "ago", + "performant", + "world", + "Optimizing", + "NVLink", + "requires", + "discover", + "invent", + "parallelization", + "algorithms", + "sits", + "top", + "distribute", + "workload", + "inferencing", + "advantage", + "aggregate", + "bandwidth", + "72", + "enabled", + "deliver", + "generationally", + "50", + "times", + "performance", + "watt", + "It", + "incredible", + "leap", + "great", + "invention", + "hard", + "creation", + "switching", + "technology", + "disaggregating", + "switches", + "building", + "system", + "racks", + "We", + "plain", + "sight", + "knew", + "And", + "results", + "So", + "dollar", + "35", + "incredible.0", + "important", + "realize", + "equals", + "revenues", + "customers", + "agents", + "generating", + "tokens", + "effective", + "When", + "coding", + "thousands", + "tens", + "hundreds", + "running", + "minutes", + "hours", + "systems", + "agentic", + "spawning", + "working", + "team", + "number", + "generated", + "exponential", + "higher", + "speed", + "dollarized", + "directly", + "translates", + "For", + "Data", + "Centers", + "CSPs", + "reason", + "power", + "limited", + "matter", + "data", + "centers", + "center", + "100", + "megawatts", + "gigawatt", + "limits", + "architecture", + "token", + "dollars", + "CSP", + "understands", + "hyperscaler", + "CapEx", + "compute", + "maximizing", + "investing", + "capacity", + "today", + "revenue", + "growth", + "Compute", + "Choosing", + "incredibly", + "strategic", + "affects", + "earnings", + "choosing", + "literally", + "Operator", + "Your", + "question", + "Ben", + "Reitzes", + "Melius", + "Research", + "line", + "open" + ], + "keywords": [ + "Inference", + "NVLink 72", + "Revenue" + ], + "connect_to": [ + { + "target": "182ae2fb-e17b-5a69-8a3a-d8d52fdf74c3", + "relation": "embeds", + "ref": "[images/image-12-Analyst, C.jpg]", + "position": { + "start": 1062, + "end": 1094 + } + } + ] + } + }, + { + "chunk_id": "17c4a52f-d6ea-5d53-a4f4-515f19c5f483", + "type": "image", + "content": "\nFor the Data Centers, inference tokens per watt translates directly to the revenues of the CSPs. And the reason for that is because everybody is power limited. And so, I mean, no matter how many data centers you have, each data center, 100 megawatts or 1 gigawatt has power limits. So the architecture that has the best performance per watt translates because each token, the performance tokens per watt, each token is dollarized, tokens per watt translates to dollars per watt, which translates in a gigawatt directly to revenues. Yeah. Hey, thanks. First, let me say kudos on including the stock-comp in non-GAAP. I think that's a great move, but that isn't my question. My question is around gross margins and the sustainability of the mid-70s long-term. Should we read into the visibility on supply being available into calendar 2027 that it's sustainable until then?0\n[images/image-13-For the Da.jpg]\n", + "path": "images/image-13-For the Da.jpg", + "metadata": { + "length": 907, + "summary": "image-13\nFor the Data Centers, inference tokens per watt translates directly to the revenues of the CSPs. And the reason for that is because everybody is power limited. And so, I mean, no matter how many data centers you have, each data center, 100 megawatts or 1 gigawatt has power limits. So the architecture that has the best performance per watt translates because each token, the performance tokens per watt, each token is dollarized, tokens per watt translates to dollars per watt, which translates in a gigawatt directly to revenues. Yeah. Hey, thanks. First, let me say kudos on including the stock-comp in non-GAAP. I think that's a great move, but that isn't my question. My question is around gross margins and the sustainability of the mid-70s long-term. Should we read into the visibility on supply being available into calendar 2027 that it's sustainable until then?0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-13-For the Da.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "c7b2fcc6-3e0d-5914-a3e4-141e6fe619d1", + "type": "text", + "content": "Analyst, Melius Research LLC\n\nFor the Data Centers, inference tokens per watt translates directly to the revenues of the CSPs. And the reason for that is because everybody is power limited. And so, I mean, no matter how many data centers you have, each data center, 100 megawatts or 1 gigawatt has power limits. So the architecture that has the best performance per watt translates because each token, the performance tokens per watt, each token is dollarized, tokens per watt translates to dollars per watt, which translates in a gigawatt directly to revenues. Yeah. Hey, thanks. First, let me say kudos on including the stock-comp in non-GAAP. I think that's a great move, but that isn't my question. My question is around gross margins and the sustainability of the mid-70s long-term. Should we read into the visibility on supply being available into calendar 2027 that it's sustainable until then?0\n[images/image-13-For the Da.jpg]\n\nYeah. Hey, thanks. First, let me say kudos on including the stock-comp in non-GAAP. I think that's a great move, but that isn't my question. My question is around gross margins and the sustainability of the mid-70s long-term. Should we read into the visibility on supply being available into calendar 2027 that it's sustainable until then?\nAnd then Jensen, what about after that? Are there innovations in memory consumption you can unveil that makes us feel better about the ability to keep margins at that level for a long time? Thanks.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Ben Reitzes", + "metadata": { + "length": 1474, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Melius", + "Research", + "LLC", + "For", + "Data", + "Centers", + "inference", + "tokens", + "watt", + "translates", + "directly", + "revenues", + "CSPs", + "And", + "reason", + "power", + "limited", + "matter", + "data", + "centers", + "center", + "100", + "megawatts", + "gigawatt", + "limits", + "So", + "architecture", + "performance", + "token", + "dollarized", + "dollars", + "Yeah", + "Hey", + "First", + "kudos", + "including", + "stock", + "comp", + "GAAP", + "great", + "move", + "isn", + "question", + "My", + "gross", + "margins", + "sustainability", + "mid", + "70s", + "long", + "term", + "Should", + "read", + "visibility", + "supply", + "calendar", + "2027", + "sustainable", + "Jensen", + "Are", + "innovations", + "memory", + "consumption", + "unveil", + "makes", + "feel", + "ability", + "level", + "time", + "Thanks" + ], + "keywords": [], + "connect_to": [ + { + "target": "17c4a52f-d6ea-5d53-a4f4-515f19c5f483", + "relation": "embeds", + "ref": "[images/image-13-For the Da.jpg]", + "position": { + "start": 903, + "end": 935 + } + } + ] + } + }, + { + "chunk_id": "182ae2fb-e17b-5a69-8a3a-d8d52fdf74c3", + "type": "image", + "content": "\nAnalyst, Citigroup Global Markets, Inc. Without CUDA, we wouldn't know what to do with inference. The entire stack from TensorRT LLM that we introduced a few years ago, which is still the most performant inference stack in the world. Optimizing it for NVLink, requires us to discover and invent new parallelization algorithms that sits on top of CUDA, to distribute the workload and the inferencing to take advantage of the aggregate bandwidth across NVLink 72. NVLink 72 has enabled us to deliver generationally 50 times more performance per watt. It's just an incredible leap and it's sensible. NVLink 72 is a great invention. It was hard to do, the creation of the switching technology, disaggregating the switches, building the system racks, all of that. We did it all in plain sight and everybody knew how hard it was for us to do. And, but the results are incredible. So performance per watt is 50 times, performance per dollar, 35 times and so the leap in inference is incredible.0\n[images/image-12-Analyst, C.jpg]\n", + "path": "images/image-12-Analyst, C.jpg_2", + "metadata": { + "length": 1023, + "summary": "image-12\nAnalyst, Citigroup Global Markets, Inc. Without CUDA, we wouldn't know what to do with inference. The entire stack from TensorRT LLM that we introduced a few years ago, which is still the most performant inference stack in the world. Optimizing it for NVLink, requires us to discover and invent new parallelization algorithms that sits on top of CUDA, to distribute the workload and the inferencing to take advantage of the aggregate bandwidth across NVLink 72. NVLink 72 has enabled us to deliver generationally 50 times more performance per watt. It's just an incredible leap and it's sensible. NVLink 72 is a great invention. It was hard to do, the creation of the switching technology, disaggregating the switches, building the system racks, all of that. We did it all in plain sight and everybody knew how hard it was for us to do. And, but the results are incredible. So performance per watt is 50 times, performance per dollar, 35 times and so the leap in inference is incredible.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-12-Analyst, C.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "4ffcd475-f4af-5c05-ac93-51decafa24a2", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nAnalyst, Citigroup Global Markets, Inc. Without CUDA, we wouldn't know what to do with inference. The entire stack from TensorRT LLM that we introduced a few years ago, which is still the most performant inference stack in the world. Optimizing it for NVLink, requires us to discover and invent new parallelization algorithms that sits on top of CUDA, to distribute the workload and the inferencing to take advantage of the aggregate bandwidth across NVLink 72. NVLink 72 has enabled us to deliver generationally 50 times more performance per watt. It's just an incredible leap and it's sensible. NVLink 72 is a great invention. It was hard to do, the creation of the switching technology, disaggregating the switches, building the system racks, all of that. We did it all in plain sight and everybody knew how hard it was for us to do. And, but the results are incredible. So performance per watt is 50 times, performance per dollar, 35 times and so the leap in inference is incredible.0\n[images/image-12-Analyst, C.jpg]\n\nThe single most important lever of our gross margins is actually delivering generational leaps to our customers. That is the single most important thing. If we could deliver generationally performance per watt that exceeds dramatically what Moore's Law can do. If we can deliver performance per dollar dramatically more than the cost of our systems, than the price of our systems, then we can continue to sustain our gross margins. That's the simple, most important concept.\nEvery – the reason why we're moving so fast is because number one, the demand for tokens in the world as a result of the inflection points that we've gone through has now – has gone completely exponential. I think we're all seeing that, to the point where even our six-year old GPUs in the cloud are completely consumed, and the pricing is going up.\nAnd so we know that the amount of computation necessary – the amount of compute necessary for the modern way of doing software is growing exponentially. And so our strategy is to deliver an entire AI infrastructure every single year. This year we introduced six new chips, Rubin next-generation will do many new chips as well. And every single generation, we are committed to deliver many X factors of performance per watt and performance per dollar. And that pace and our ability to do extreme co-design, allows us to deliver that value and that benefit to the customers. And that is the single most vital thing as it relates to our value delivered.\nOperator: Your next question comes from Antoine Chkaiban with New Street Research. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_6", + "metadata": { + "length": 2673, + "summary": "The passage discusses NVIDIA's advancements in AI inference, highlighting the critical role of CUDA and the new NVLink 72 technology. NVLink 72 enables a 50x improvement in performance per watt and 35x in performance per dollar, representing a generational leap. The speaker, a co-founder and CEO, emphasizes that delivering such dramatic performance gains is the key driver of gross margins. The strategy involves releasing a new AI infrastructure annually, with six new chips this year and more planned for the next-generation Rubin, to meet exponentially growing compute demand for modern software.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Analyst", + "Citigroup", + "Global", + "Markets", + "Inc", + "Without", + "CUDA", + "wouldn", + "inference", + "The", + "entire", + "stack", + "TensorRT", + "LLM", + "introduced", + "years", + "ago", + "performant", + "world", + "Optimizing", + "NVLink", + "requires", + "discover", + "invent", + "parallelization", + "algorithms", + "sits", + "top", + "distribute", + "workload", + "inferencing", + "advantage", + "aggregate", + "bandwidth", + "72", + "enabled", + "deliver", + "generationally", + "50", + "times", + "performance", + "watt", + "It", + "incredible", + "leap", + "great", + "invention", + "hard", + "creation", + "switching", + "technology", + "disaggregating", + "switches", + "building", + "system", + "racks", + "We", + "plain", + "sight", + "knew", + "And", + "results", + "So", + "dollar", + "35", + "incredible.0", + "single", + "important", + "lever", + "gross", + "margins", + "delivering", + "generational", + "leaps", + "customers", + "That", + "thing", + "If", + "exceeds", + "dramatically", + "Moore", + "Law", + "cost", + "systems", + "price", + "continue", + "sustain", + "simple", + "concept", + "Every", + "reason", + "moving", + "fast", + "number", + "demand", + "tokens", + "result", + "inflection", + "points", + "ve", + "completely", + "exponential", + "point", + "year", + "GPUs", + "cloud", + "consumed", + "pricing", + "amount", + "computation", + "compute", + "modern", + "software", + "growing", + "exponentially", + "strategy", + "AI", + "infrastructure", + "This", + "chips", + "Rubin", + "generation", + "committed", + "factors", + "pace", + "ability", + "extreme", + "design", + "benefit", + "vital", + "relates", + "delivered", + "Operator", + "Your", + "question", + "Antoine", + "Chkaiban", + "New", + "Street", + "Research", + "line", + "open" + ], + "keywords": [ + "NVLink 72", + "CUDA", + "Gross Margins" + ], + "connect_to": [ + { + "target": "182ae2fb-e17b-5a69-8a3a-d8d52fdf74c3", + "relation": "embeds", + "ref": "[images/image-12-Analyst, C.jpg]", + "position": { + "start": 1062, + "end": 1094 + } + } + ] + } + }, + { + "chunk_id": "373364c6-5cee-517a-b03d-ed6c891343ec", + "type": "image", + "content": "\nEvery – the reason why we're moving so fast is because number one, the demand for tokens in the world as a result of the inflection points that we've gone through has now – has gone completely exponential. I think we're all seeing that, to the point where even our six-year old GPUs in the cloud are completely consumed, and the pricing is going up. Hi. Thanks a lot for taking my question. I'd like to ask about space data centers, which some of your customers are considering. How feasible do you think that is and what kind of horizon? And what do the economics look like today? And how do you think that could evolve over time? Thank you.0\n[images/image-14-Every – th.jpg]\n", + "path": "images/image-14-Every – th.jpg", + "metadata": { + "length": 678, + "summary": "image-14\nEvery – the reason why we're moving so fast is because number one, the demand for tokens in the world as a result of the inflection points that we've gone through has now – has gone completely exponential. I think we're all seeing that, to the point where even our six-year old GPUs in the cloud are completely consumed, and the pricing is going up. Hi. Thanks a lot for taking my question. I'd like to ask about space data centers, which some of your customers are considering. How feasible do you think that is and what kind of horizon? And what do the economics look like today? And how do you think that could evolve over time? Thank you.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-14-Every – th.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "99d8f407-e2a4-56c8-892b-eec02de61893", + "type": "text", + "content": "Analyst, New Street Research LLP\n\nEvery – the reason why we're moving so fast is because number one, the demand for tokens in the world as a result of the inflection points that we've gone through has now – has gone completely exponential. I think we're all seeing that, to the point where even our six-year old GPUs in the cloud are completely consumed, and the pricing is going up. Hi. Thanks a lot for taking my question. I'd like to ask about space data centers, which some of your customers are considering. How feasible do you think that is and what kind of horizon? And what do the economics look like today? And how do you think that could evolve over time? Thank you.0\n[images/image-14-Every – th.jpg]\n\nHi. Thanks a lot for taking my question. I'd like to ask about space data centers, which some of your customers are considering. How feasible do you think that is and what kind of horizon? And what do the economics look like today? And how do you think that could evolve over time? Thank you.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Antoine Chkaiban", + "metadata": { + "length": 1004, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "New", + "Street", + "Research", + "LLP", + "Every", + "reason", + "moving", + "fast", + "number", + "demand", + "tokens", + "world", + "result", + "inflection", + "points", + "ve", + "completely", + "exponential", + "point", + "year", + "GPUs", + "cloud", + "consumed", + "pricing", + "Hi", + "Thanks", + "lot", + "taking", + "question", + "space", + "data", + "centers", + "customers", + "How", + "feasible", + "kind", + "horizon", + "And", + "economics", + "today", + "evolve", + "time", + "Thank", + "you.0" + ], + "keywords": [], + "connect_to": [ + { + "target": "373364c6-5cee-517a-b03d-ed6c891343ec", + "relation": "embeds", + "ref": "[images/image-14-Every – th.jpg]", + "position": { + "start": 678, + "end": 710 + } + } + ] + } + }, + { + "chunk_id": "185b6f2a-1e5a-5e88-95b2-c3161ff8d2ec", + "type": "image", + "content": "\nAnalyst, New Street Research LLP Well, the economics are poor today, but it's going to improve over time. As you know, the way that space works is radically different than how it works down here. There's an abundance of energy. But solar panels are large, but there's plenty of space in space.0\n[images/image-15-Analyst, N.jpg]\n", + "path": "images/image-15-Analyst, N.jpg", + "metadata": { + "length": 329, + "summary": "image-15\nAnalyst, New Street Research LLP Well, the economics are poor today, but it's going to improve over time. As you know, the way that space works is radically different than how it works down here. There's an abundance of energy. But solar panels are large, but there's plenty of space in space.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-15-Analyst, N.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "63304068-a83b-59aa-8ffa-e1dad4468401", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nAnalyst, New Street Research LLP Well, the economics are poor today, but it's going to improve over time. As you know, the way that space works is radically different than how it works down here. There's an abundance of energy. But solar panels are large, but there's plenty of space in space.0\n[images/image-15-Analyst, N.jpg]\n\nWell, the economics are poor today, but it's going to improve over time. As you know, the way that space works is radically different than how it works down here. There's an abundance of energy. But solar panels are large, but there's plenty of space in space.\nThe heat dissipation -- it's cold in space. However, there's no airflow. And so the only way to dissipate heat is through conduction. And the radiators that you need to create are fairly large. Liquid cooling is obviously out of the question because it's kind of -- it's heavy and freezes. And so the methods that we use here on Earth are a little different than the way we would do it in space. But there are many different computing problems that really wants to be done in space. And so, NVIDIA is already the world's first GPU in space, Hopper's in space.\nAnd one of the best use cases of GPUs in space is imaging, to be able to image at extremely high resolutions using, of course, optics and artificial intelligence and to be able to do that computation of reprojection of different angles and be able to upres and do noise reduction and just be able to see, be able to image at very large, very high resolutions, extremely large scales, and very, very fast. It's hard to do that by sending petabytes and\npetabytes of imaging data back here on Earth and doing that work. It's easier just to do it out in space. And then, ignore all of the data collected and processed until you see something interesting. And so artificial intelligence in space will have very good, very interesting applications.\nOperator: Your next question comes from Mark Lipacis with Evercore ISI. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_7", + "metadata": { + "length": 2056, + "summary": "The passage discusses the economics and technical challenges of computing in space, highlighting NVIDIA's role with the first GPU in space (Hopper). It contrasts space conditions (abundance of energy, cold, no airflow) with Earth, noting heat dissipation via radiators and the impracticality of liquid cooling. Key applications include high-resolution imaging using AI for reprojection, upscaling, and noise reduction, processing data in space rather than sending petabytes to Earth.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Analyst", + "New", + "Street", + "Research", + "LLP", + "Well", + "economics", + "poor", + "today", + "improve", + "time", + "As", + "space", + "works", + "radically", + "There", + "abundance", + "energy", + "But", + "solar", + "panels", + "large", + "plenty", + "space.0", + "The", + "heat", + "dissipation", + "cold", + "However", + "airflow", + "And", + "dissipate", + "conduction", + "radiators", + "create", + "fairly", + "Liquid", + "cooling", + "question", + "kind", + "heavy", + "freezes", + "methods", + "Earth", + "computing", + "problems", + "world", + "GPU", + "Hopper", + "cases", + "GPUs", + "imaging", + "image", + "extremely", + "high", + "resolutions", + "optics", + "artificial", + "intelligence", + "computation", + "reprojection", + "angles", + "upres", + "noise", + "reduction", + "scales", + "fast", + "It", + "hard", + "sending", + "petabytes", + "data", + "back", + "work", + "easier", + "ignore", + "collected", + "processed", + "interesting", + "good", + "applications", + "Operator", + "Your", + "Mark", + "Lipacis", + "Evercore", + "ISI", + "line", + "open" + ], + "keywords": [ + "NVIDIA", + "Space", + "GPU" + ], + "connect_to": [ + { + "target": "185b6f2a-1e5a-5e88-95b2-c3161ff8d2ec", + "relation": "embeds", + "ref": "[images/image-15-Analyst, N.jpg]", + "position": { + "start": 368, + "end": 400 + } + } + ] + } + }, + { + "chunk_id": "22cb2d0d-cea3-5951-9a89-b5e53e5f0bc1", + "type": "image", + "content": "\nAnd one of the best use cases of GPUs in space is imaging, to be able to image at extremely high resolutions using, of course, optics and artificial intelligence and to be able to do that computation of reprojection of different angles and be able to upres and do noise reduction and just be able to see, be able to image at very large, very high resolutions, extremely large scales, and very, very fast. It's hard to do that by sending petabytes and Hi. Thanks for taking my question. I want to pick up with the comment you made on the script about revenue diversification. I believe, Colette, you said that hyperscalers were over 50% of revenues, but growth was led by the rest of your Data Center customers. And as a clarification, I just want to make sure I understood that, does that imply your non-hyperscale customers grew faster. And if so, what are the – can you help us understand what are the non-hyperscalers doing different. Are they doing different things than the hyperscalers or the same things on a different scale and does this – do you expect this trend to continue? Would you expect your customer base to evolve to a point where non-hyperscalers are becoming a bigger part of your, the larger part of your business? Thank you.0\n[images/image-16-And one of.jpg]\n", + "path": "images/image-16-And one of.jpg", + "metadata": { + "length": 1282, + "summary": "image-16\nAnd one of the best use cases of GPUs in space is imaging, to be able to image at extremely high resolutions using, of course, optics and artificial intelligence and to be able to do that computation of reprojection of different angles and be able to upres and do noise reduction and just be able to see, be able to image at very large, very high resolutions, extremely large scales, and very, very fast. It's hard to do that by sending petabytes and Hi. Thanks for taking my question. I want to pick up with the comment you made on the script about revenue diversification. I believe, Colette, you said that hyperscalers were over 50% of revenues, but growth was led by the rest of your Data Center customers. And as a clarification, I just want to make sure I understood that, does that imply your non-hyperscale customers grew faster. And if so, what are the – can you help us understand what are the non-hyperscalers doing different. Are they doing different things than the hyperscalers or the same things on a different scale and does this – do you expect this trend to continue? Would you expect your customer base to evolve to a point where non-hyperscalers are becoming a bigger part of your, the larger part of your business? Thank you.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-16-And one of.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f8ceb64e-02cc-5470-b4eb-832730f8f12c", + "type": "text", + "content": "Analyst, Evercore Group LLC\n\nAnd one of the best use cases of GPUs in space is imaging, to be able to image at extremely high resolutions using, of course, optics and artificial intelligence and to be able to do that computation of reprojection of different angles and be able to upres and do noise reduction and just be able to see, be able to image at very large, very high resolutions, extremely large scales, and very, very fast. It's hard to do that by sending petabytes and Hi. Thanks for taking my question. I want to pick up with the comment you made on the script about revenue diversification. I believe, Colette, you said that hyperscalers were over 50% of revenues, but growth was led by the rest of your Data Center customers. And as a clarification, I just want to make sure I understood that, does that imply your non-hyperscale customers grew faster. And if so, what are the – can you help us understand what are the non-hyperscalers doing different. Are they doing different things than the hyperscalers or the same things on a different scale and does this – do you expect this trend to continue? Would you expect your customer base to evolve to a point where non-hyperscalers are becoming a bigger part of your, the larger part of your business? Thank you.0\n[images/image-16-And one of.jpg]\n\nHi. Thanks for taking my question. I want to pick up with the comment you made on the script about revenue diversification. I believe, Colette, you said that hyperscalers were over 50% of revenues, but growth was led by the rest of your Data Center customers. And as a clarification, I just want to make sure I understood that, does that imply your non-hyperscale customers grew faster. And if so, what are the – can you help us understand what are the non-hyperscalers doing different. Are they doing different things than the hyperscalers or the same things on a different scale and does this – do you expect this trend to continue? Would you expect your customer base to evolve to a point where non-hyperscalers are becoming a bigger part of your, the larger part of your business? Thank you.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Mark Lipacis", + "metadata": { + "length": 2106, + "summary": "An analyst asks about revenue diversification, noting that hyperscalers account for over 50% of revenues but growth is led by other data center customers. The question seeks clarification on whether non-hyperscale customers grew faster, what they are doing differently, and if this trend will continue, potentially making non-hyperscalers a larger part of the business.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Evercore", + "Group", + "LLC", + "And", + "cases", + "GPUs", + "space", + "imaging", + "image", + "extremely", + "high", + "resolutions", + "optics", + "artificial", + "intelligence", + "computation", + "reprojection", + "angles", + "upres", + "noise", + "reduction", + "large", + "scales", + "fast", + "It", + "hard", + "sending", + "petabytes", + "Hi", + "Thanks", + "taking", + "question", + "pick", + "comment", + "made", + "script", + "revenue", + "diversification", + "Colette", + "hyperscalers", + "50%", + "revenues", + "growth", + "led", + "rest", + "Data", + "Center", + "customers", + "clarification", + "make", + "understood", + "imply", + "hyperscale", + "grew", + "faster", + "understand", + "Are", + "things", + "scale", + "expect", + "trend", + "continue", + "Would", + "customer", + "base", + "evolve", + "point", + "bigger", + "part", + "larger", + "business", + "Thank", + "you.0" + ], + "keywords": [ + "hyperscalers", + "non-hyperscale", + "data center" + ], + "connect_to": [ + { + "target": "22cb2d0d-cea3-5951-9a89-b5e53e5f0bc1", + "relation": "embeds", + "ref": "[images/image-16-And one of.jpg]", + "position": { + "start": 1277, + "end": 1309 + } + } + ] + } + }, + { + "chunk_id": "672e58f5-030b-5fd4-b408-334d4ef4610d", + "type": "image", + "content": "\nAnalyst, Evercore Group LLC Yes. Let's see if we can help on this question. So when you think about our top five, as we articulated as being our CSPs, our hyperscalers, and they have right now, [ph] as I said (49:15) about 50% of our total revenue. There's a big organization, therefore, of diversity of all different other types of companies that we are working with. That it goes through our AI model makers, that goes through our enterprises, that goes to supercomputing, it goes to our sovereigns. There's a lot of other different facts on there. But you are correct, it's a very fast growing area as well.0\n[images/image-17-Analyst, E.jpg]\n", + "path": "images/image-17-Analyst, E.jpg", + "metadata": { + "length": 646, + "summary": "image-17\nAnalyst, Evercore Group LLC Yes. Let's see if we can help on this question. So when you think about our top five, as we articulated as being our CSPs, our hyperscalers, and they have right now, [ph] as I said (49:15) about 50% of our total revenue. There's a big organization, therefore, of diversity of all different other types of companies that we are working with. That it goes through our AI model makers, that goes through our enterprises, that goes to supercomputing, it goes to our sovereigns. There's a lot of other different facts on there. But you are correct, it's a very fast growing area as well.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-17-Analyst, E.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "1ec996ae-6b9c-53a1-a2cc-888e6420c9bd", + "type": "text", + "content": "Chief Financial Officer & Executive Vice President, NVIDIA Corp.\n\nAnalyst, Evercore Group LLC Yes. Let's see if we can help on this question. So when you think about our top five, as we articulated as being our CSPs, our hyperscalers, and they have right now, [ph] as I said (49:15) about 50% of our total revenue. There's a big organization, therefore, of diversity of all different other types of companies that we are working with. That it goes through our AI model makers, that goes through our enterprises, that goes to supercomputing, it goes to our sovereigns. There's a lot of other different facts on there. But you are correct, it's a very fast growing area as well.0\n[images/image-17-Analyst, E.jpg]\n\nYes. Let's see if we can help on this question. So when you think about our top five, as we articulated as being our CSPs, our hyperscalers, and they have right now, [ph] as I said (49:15) about 50% of our total revenue. There's a big organization, therefore, of diversity of all different other types of companies that we are working with. That it goes through our AI model makers, that goes through our enterprises, that goes to supercomputing, it goes to our sovereigns. There's a lot of other different facts on there. But you are correct, it's a very fast growing area as well.\nWe have a strong position in terms of all of our different cloud providers on our platform, and now we also have a extreme diversity of different customers that we are seeing all the way across the world. And this will really benefit, seeing that diversity and being able to serve all of those parts. Let me see if Jensen wants to add a bit more", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Colette M. Kress_2", + "metadata": { + "length": 1640, + "summary": "The speaker discusses Nvidia's revenue composition, noting that its top five customers (CSPs and hyperscalers) account for about 50% of total revenue. The remaining revenue comes from a diverse range of customers including AI model makers, enterprises, supercomputing, and sovereign entities. This diversity is seen as a strength, with the company serving various cloud providers and customers globally.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Chief", + "Financial", + "Officer", + "Executive", + "Vice", + "President", + "NVIDIA", + "Corp", + "Analyst", + "Evercore", + "Group", + "LLC", + "Yes", + "Let", + "question", + "So", + "top", + "articulated", + "CSPs", + "hyperscalers", + "ph", + "49", + "15", + "50%", + "total", + "revenue", + "There", + "big", + "organization", + "diversity", + "types", + "companies", + "working", + "That", + "AI", + "model", + "makers", + "enterprises", + "supercomputing", + "sovereigns", + "lot", + "facts", + "But", + "correct", + "fast", + "growing", + "area", + "well.0", + "We", + "strong", + "position", + "terms", + "cloud", + "providers", + "platform", + "extreme", + "customers", + "world", + "And", + "benefit", + "serve", + "parts", + "Jensen", + "add", + "bit" + ], + "keywords": [ + "CSPs", + "hyperscalers", + "revenue diversity" + ], + "connect_to": [ + { + "target": "672e58f5-030b-5fd4-b408-334d4ef4610d", + "relation": "embeds", + "ref": "[images/image-17-Analyst, E.jpg]", + "position": { + "start": 678, + "end": 710 + } + } + ] + } + }, + { + "chunk_id": "185b6f2a-1e5a-5e88-95b2-c3161ff8d2ec", + "type": "image", + "content": "\nAnalyst, New Street Research LLP Well, the economics are poor today, but it's going to improve over time. As you know, the way that space works is radically different than how it works down here. There's an abundance of energy. But solar panels are large, but there's plenty of space in space.0\n[images/image-15-Analyst, N.jpg]\n", + "path": "images/image-15-Analyst, N.jpg_2", + "metadata": { + "length": 329, + "summary": "image-15\nAnalyst, New Street Research LLP Well, the economics are poor today, but it's going to improve over time. As you know, the way that space works is radically different than how it works down here. There's an abundance of energy. But solar panels are large, but there's plenty of space in space.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-15-Analyst, N.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "fa65feb6-bcbd-5daf-8cdf-0dc4525699bd", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nAnalyst, New Street Research LLP Well, the economics are poor today, but it's going to improve over time. As you know, the way that space works is radically different than how it works down here. There's an abundance of energy. But solar panels are large, but there's plenty of space in space.0\n[images/image-15-Analyst, N.jpg]\n\nYeah, this is one of the advantages that we have with our ecosystem. I'll build on top of CUDA. We have -- we're the only accelerated computing platform that is in every cloud, that's available through every single computer maker, available at the edge and, we're now cultivating telecommunications. Obviously, the future radios will all be AI driven radios and the future wireless network would also be a computing platform. That is a foregone conclusion. But somebody has to go and invent the technologies to make that possible. And we created a platform called Aerial to go do that.\nWe are in just about every single robot, every single self-driving car. Our ability – CUDA's ability to have the benefit of the performance of specialized processors, one the one hand, with the Tensor cores inside our GPUs, on the other hand, the flexibility of CUDA allows us to solve language problems, computer vision problems, robotics problems, to biology problems, physics problems, and just about all kinds of AI and all kinds of computation algorithms. And so the diversity of our customer base is one of the greatest strengths that we have.\nThe second thing, of course, is without our own ecosystem, even if our processor was programmable, if we didn't cultivate our ecosystem and talking about some of the things that we're doing today, investing in our future ecosystem and continuing to enhance our ecosystem, without our ecosystem, it's hard for us to grow beyond what design wins we capture for somebody else's ecosystem. And so we could grow and expand our ecosystem very naturally because of our – the platform that we created.\nAnd then lastly, one of the things that's really important is the partnerships that we have with OpenAI and Anthropic, with xAI, with Meta now makes – and of course, just about every single open source in the world. There's 1.5 million AI models on Hugging Face, all of it runs on NVIDIA CUDA. And so an open source in totality probably represents the largest – the second largest model in the world, OpenAI is the largest, second largest, probably all the collection of all the open sources.\nAnd so NVIDIA's ability to run all of that makes our platform super fungible, super easy to use, and really safe to invest into. And so that creates the diversity of customers and the diversity of the platforms and available in every single country and – because we support the whole world's ecosystem.\nOperator: Your next question comes from Aaron Rakers with Wells Fargo. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_8", + "metadata": { + "length": 2917, + "summary": "The speaker, an NVIDIA executive, discusses the company's competitive advantages: its CUDA platform's ubiquity across clouds and devices, its cultivation of a diverse ecosystem spanning robotics, self-driving cars, and AI across multiple domains, and key partnerships with OpenAI, Anthropic, xAI, and Meta. They highlight that 1.5 million AI models on Hugging Face run on NVIDIA CUDA, making the platform safe and easy to invest in.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Analyst", + "New", + "Street", + "Research", + "LLP", + "Well", + "economics", + "poor", + "today", + "improve", + "time", + "As", + "space", + "works", + "radically", + "There", + "abundance", + "energy", + "But", + "solar", + "panels", + "large", + "plenty", + "space.0", + "Yeah", + "advantages", + "ecosystem", + "ll", + "build", + "top", + "CUDA", + "We", + "accelerated", + "computing", + "platform", + "cloud", + "single", + "computer", + "maker", + "edge", + "cultivating", + "telecommunications", + "Obviously", + "future", + "radios", + "AI", + "driven", + "wireless", + "network", + "That", + "foregone", + "conclusion", + "invent", + "technologies", + "make", + "And", + "created", + "called", + "Aerial", + "robot", + "driving", + "car", + "Our", + "ability", + "benefit", + "performance", + "specialized", + "processors", + "hand", + "Tensor", + "cores", + "inside", + "GPUs", + "flexibility", + "solve", + "language", + "problems", + "vision", + "robotics", + "biology", + "physics", + "kinds", + "computation", + "algorithms", + "diversity", + "customer", + "base", + "greatest", + "strengths", + "The", + "thing", + "processor", + "programmable", + "didn", + "cultivate", + "talking", + "things", + "investing", + "continuing", + "enhance", + "hard", + "grow", + "design", + "wins", + "capture", + "expand", + "naturally", + "lastly", + "important", + "partnerships", + "OpenAI", + "Anthropic", + "xAI", + "Meta", + "makes", + "open", + "source", + "world", + "1.5", + "million", + "models", + "Hugging", + "Face", + "runs", + "totality", + "represents", + "largest", + "model", + "collection", + "sources", + "run", + "super", + "fungible", + "easy", + "safe", + "invest", + "creates", + "customers", + "platforms", + "country", + "support", + "Operator", + "Your", + "question", + "Aaron", + "Rakers", + "Wells", + "Fargo", + "line" + ], + "keywords": [ + "NVIDIA", + "CUDA", + "AI ecosystem" + ], + "connect_to": [ + { + "target": "185b6f2a-1e5a-5e88-95b2-c3161ff8d2ec", + "relation": "embeds", + "ref": "[images/image-15-Analyst, N.jpg]", + "position": { + "start": 368, + "end": 400 + } + } + ] + } + }, + { + "chunk_id": "7d8c7af8-e14a-5929-953e-ff2ca52deb68", + "type": "image", + "content": "\nAnd then lastly, one of the things that's really important is the partnerships that we have with OpenAI and Anthropic, with xAI, with Meta now makes – and of course, just about every single open source in the world. There's 1.5 million AI models on Hugging Face, all of it runs on NVIDIA CUDA. And so an open source in totality probably represents the largest – the second largest model in the world, OpenAI is the largest, second largest, probably all the collection of all the open sources. Yeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis.0\n[images/image-18-And then l.jpg]\n", + "path": "images/image-18-And then l.jpg", + "metadata": { + "length": 780, + "summary": "image-18\nAnd then lastly, one of the things that's really important is the partnerships that we have with OpenAI and Anthropic, with xAI, with Meta now makes – and of course, just about every single open source in the world. There's 1.5 million AI models on Hugging Face, all of it runs on NVIDIA CUDA. And so an open source in totality probably represents the largest – the second largest model in the world, OpenAI is the largest, second largest, probably all the collection of all the open sources. Yeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-18-And then l.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "02cfdf07-d0d6-5068-8a59-8bc02f71e7d6", + "type": "text", + "content": "Analyst, Wells Fargo Securities LLC\n\nAnd then lastly, one of the things that's really important is the partnerships that we have with OpenAI and Anthropic, with xAI, with Meta now makes – and of course, just about every single open source in the world. There's 1.5 million AI models on Hugging Face, all of it runs on NVIDIA CUDA. And so an open source in totality probably represents the largest – the second largest model in the world, OpenAI is the largest, second largest, probably all the collection of all the open sources. Yeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis.0\n[images/image-18-And then l.jpg]\n\nYeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis.\nSo, I guess, Jensen, I'm curious of what's the importance that Vera plays in this architecture evolution as we move forward? Is this being driven more by the proliferation or the heterogeneity of inference workloads? I'm just curious of how you see that evolving for NVIDIA, particularly on a standalone CPU basis? Thank you.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Aaron Rakers", + "metadata": { + "length": 1394, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Wells", + "Fargo", + "Securities", + "LLC", + "And", + "lastly", + "things", + "important", + "partnerships", + "OpenAI", + "Anthropic", + "xAI", + "Meta", + "makes", + "single", + "open", + "source", + "world", + "There", + "1.5", + "million", + "AI", + "models", + "Hugging", + "Face", + "runs", + "NVIDIA", + "CUDA", + "totality", + "represents", + "largest", + "model", + "collection", + "sources", + "Yeah", + "Thanks", + "taking", + "question", + "guess", + "sticking", + "idea", + "platform", + "extreme", + "design", + "news", + "quarter", + "ability", + "push", + "bring", + "Vera", + "CPUs", + "market", + "standalone", + "solution", + "basis.0", + "basis", + "So", + "Jensen", + "curious", + "importance", + "plays", + "architecture", + "evolution", + "move", + "forward", + "Is", + "driven", + "proliferation", + "heterogeneity", + "inference", + "workloads", + "evolving", + "CPU", + "Thank" + ], + "keywords": [], + "connect_to": [ + { + "target": "7d8c7af8-e14a-5929-953e-ff2ca52deb68", + "relation": "embeds", + "ref": "[images/image-18-And then l.jpg]", + "position": { + "start": 783, + "end": 815 + } + } + ] + } + }, + { + "chunk_id": "63763a3e-40d9-55b3-9fef-a271526681d9", + "type": "image", + "content": "\nYeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis. Yeah, thanks. And I'll tell you some more about it at GTC. But at the highest level, we made fundamentally different architecture decisions about our CPUs compared to the rest of the world's CPUs.0\n[images/image-19-Yeah. Than.jpg]\n", + "path": "images/image-19-Yeah. Than.jpg", + "metadata": { + "length": 484, + "summary": "image-19\nYeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis. Yeah, thanks. And I'll tell you some more about it at GTC. But at the highest level, we made fundamentally different architecture decisions about our CPUs compared to the rest of the world's CPUs.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-19-Yeah. Than.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "0bd3e813-6e30-50a5-8b67-009d8f182054", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nYeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis. Yeah, thanks. And I'll tell you some more about it at GTC. But at the highest level, we made fundamentally different architecture decisions about our CPUs compared to the rest of the world's CPUs.0\n[images/image-19-Yeah. Than.jpg]\n\nYeah, thanks. And I'll tell you some more about it at GTC. But at the highest level, we made fundamentally different architecture decisions about our CPUs compared to the rest of the world's CPUs.\nIt's the only data center CPU that supports LPDDR5. It is designed to be focused on very high data processing capabilities. And the reason for that is because most of the computing problems that we're interested in are data-driven, artificial intelligence being one.\nAnd the single threaded performance and this ratio with bandwidth is just off the charts. And we made those architectural decisions because in the entire phase – the different phases of AI from data processing before you even do training, you have to do data processing. So, you have data processing, pre-training, and in post-training now the AI's are learning how to use tools and the usage of tools, many of those tools run in CPU-only environments or they run in CPU or GPU accelerated environments. And Vera was designed to be an excellent CPU for post-training and so some of the use cases in the entire pipeline of artificial intelligence includes using a lot of CPUs.\nWe love CPUs as well as GPUs. And when you accelerate the algorithms to the limit, as we have, Amdahl's law would suggest that you need really, really fast single threaded CPUs. And that's the reason why we built, Grace to be a extraordinarily great at single threaded performance and Vera is off the charts better than that.\nOperator: Your next question comes from Tim Arcuri with UBS. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_9", + "metadata": { + "length": 2102, + "summary": "NVIDIA's co-founder discusses the Vera CPU's unique architecture, emphasizing its support for LPDDR5 and focus on high data processing capabilities for AI workloads. The CPU is designed for data processing, pre-training, and post-training phases, with exceptional single-threaded performance and bandwidth ratio, making it suitable for CPU-only or GPU-accelerated environments.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Yeah", + "Thanks", + "taking", + "question", + "guess", + "sticking", + "idea", + "platform", + "extreme", + "design", + "news", + "quarter", + "ability", + "push", + "bring", + "Vera", + "CPUs", + "market", + "standalone", + "solution", + "basis", + "And", + "ll", + "GTC", + "But", + "highest", + "level", + "made", + "fundamentally", + "architecture", + "decisions", + "compared", + "rest", + "world", + "CPUs.0", + "It", + "data", + "center", + "CPU", + "supports", + "LPDDR5", + "designed", + "focused", + "high", + "processing", + "capabilities", + "reason", + "computing", + "problems", + "interested", + "driven", + "artificial", + "intelligence", + "single", + "threaded", + "performance", + "ratio", + "bandwidth", + "charts", + "architectural", + "entire", + "phase", + "phases", + "AI", + "training", + "So", + "pre", + "post", + "learning", + "tools", + "usage", + "run", + "environments", + "GPU", + "accelerated", + "excellent", + "cases", + "pipeline", + "includes", + "lot", + "We", + "love", + "GPUs", + "accelerate", + "algorithms", + "limit", + "Amdahl", + "law", + "suggest", + "fast", + "built", + "Grace", + "extraordinarily", + "great", + "Operator", + "Your", + "Tim", + "Arcuri", + "UBS", + "line", + "open" + ], + "keywords": [ + "Vera CPU", + "AI data processing", + "single-threaded performance" + ], + "connect_to": [ + { + "target": "63763a3e-40d9-55b3-9fef-a271526681d9", + "relation": "embeds", + "ref": "[images/image-19-Yeah. Than.jpg]", + "position": { + "start": 523, + "end": 555 + } + } + ] + } + }, + { + "chunk_id": "22cb2d0d-cea3-5951-9a89-b5e53e5f0bc1", + "type": "image", + "content": "\nAnd one of the best use cases of GPUs in space is imaging, to be able to image at extremely high resolutions using, of course, optics and artificial intelligence and to be able to do that computation of reprojection of different angles and be able to upres and do noise reduction and just be able to see, be able to image at very large, very high resolutions, extremely large scales, and very, very fast. It's hard to do that by sending petabytes and Hi. Thanks for taking my question. I want to pick up with the comment you made on the script about revenue diversification. I believe, Colette, you said that hyperscalers were over 50% of revenues, but growth was led by the rest of your Data Center customers. And as a clarification, I just want to make sure I understood that, does that imply your non-hyperscale customers grew faster. And if so, what are the – can you help us understand what are the non-hyperscalers doing different. Are they doing different things than the hyperscalers or the same things on a different scale and does this – do you expect this trend to continue? Would you expect your customer base to evolve to a point where non-hyperscalers are becoming a bigger part of your, the larger part of your business? Thank you.0\n[images/image-16-And one of.jpg]\n", + "path": "images/image-16-And one of.jpg_2", + "metadata": { + "length": 1282, + "summary": "image-16\nAnd one of the best use cases of GPUs in space is imaging, to be able to image at extremely high resolutions using, of course, optics and artificial intelligence and to be able to do that computation of reprojection of different angles and be able to upres and do noise reduction and just be able to see, be able to image at very large, very high resolutions, extremely large scales, and very, very fast. It's hard to do that by sending petabytes and Hi. Thanks for taking my question. I want to pick up with the comment you made on the script about revenue diversification. I believe, Colette, you said that hyperscalers were over 50% of revenues, but growth was led by the rest of your Data Center customers. And as a clarification, I just want to make sure I understood that, does that imply your non-hyperscale customers grew faster. And if so, what are the – can you help us understand what are the non-hyperscalers doing different. Are they doing different things than the hyperscalers or the same things on a different scale and does this – do you expect this trend to continue? Would you expect your customer base to evolve to a point where non-hyperscalers are becoming a bigger part of your, the larger part of your business? Thank you.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-16-And one of.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "41f339e3-6b08-56f7-8680-f35cbb2eb176", + "type": "text", + "content": "Analyst, UBS Securities LLC\n\nAnd one of the best use cases of GPUs in space is imaging, to be able to image at extremely high resolutions using, of course, optics and artificial intelligence and to be able to do that computation of reprojection of different angles and be able to upres and do noise reduction and just be able to see, be able to image at very large, very high resolutions, extremely large scales, and very, very fast. It's hard to do that by sending petabytes and Hi. Thanks for taking my question. I want to pick up with the comment you made on the script about revenue diversification. I believe, Colette, you said that hyperscalers were over 50% of revenues, but growth was led by the rest of your Data Center customers. And as a clarification, I just want to make sure I understood that, does that imply your non-hyperscale customers grew faster. And if so, what are the – can you help us understand what are the non-hyperscalers doing different. Are they doing different things than the hyperscalers or the same things on a different scale and does this – do you expect this trend to continue? Would you expect your customer base to evolve to a point where non-hyperscalers are becoming a bigger part of your, the larger part of your business? Thank you.0\n[images/image-16-And one of.jpg]\n\nThanks a lot. Colette, I was wondering if you can talk about the deployment of capital. I know that you really jacked up the purchase commits. But it sounds like maybe you're over the hump on this, and you're going to probably generate about \\$100 billion in cash this year.\nSo, and, pretty much no matter how good the results have been, the stock hasn't really gone up much. So I would think that you probably feel like this is a pretty good price to be buying back a bunch of it here. So I was wondering if you can talk about that like, question being, why not put a big stake in the ground and just, have a huge share repo here? Thanks.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Timothy Arcuri", + "metadata": { + "length": 1950, + "summary": "The passage discusses GPU use in space for high-resolution imaging via optics and AI, and a Q&A about revenue diversification at UBS Securities LLC. Colette notes hyperscalers are over 50% of revenues but growth is led by other data center customers, implying non-hyperscale customers grew faster. The question explores whether these customers do different things or the same on a different scale, and if this trend will continue. Another question addresses capital deployment, noting increased purchase commitments and potential $100 billion cash generation, suggesting a large share repurchase might be warranted given stagnant stock performance.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "UBS", + "Securities", + "LLC", + "And", + "cases", + "GPUs", + "space", + "imaging", + "image", + "extremely", + "high", + "resolutions", + "optics", + "artificial", + "intelligence", + "computation", + "reprojection", + "angles", + "upres", + "noise", + "reduction", + "large", + "scales", + "fast", + "It", + "hard", + "sending", + "petabytes", + "Hi", + "Thanks", + "taking", + "question", + "pick", + "comment", + "made", + "script", + "revenue", + "diversification", + "Colette", + "hyperscalers", + "50%", + "revenues", + "growth", + "led", + "rest", + "Data", + "Center", + "customers", + "clarification", + "make", + "understood", + "imply", + "hyperscale", + "grew", + "faster", + "understand", + "Are", + "things", + "scale", + "expect", + "trend", + "continue", + "Would", + "customer", + "base", + "evolve", + "point", + "bigger", + "part", + "larger", + "business", + "Thank", + "you.0", + "lot", + "wondering", + "talk", + "deployment", + "capital", + "jacked", + "purchase", + "commits", + "But", + "sounds", + "hump", + "generate", + "100", + "billion", + "cash", + "year", + "So", + "pretty", + "matter", + "good", + "results", + "stock", + "hasn", + "feel", + "price", + "buying", + "back", + "bunch", + "put", + "big", + "stake", + "ground", + "huge", + "share", + "repo" + ], + "keywords": [ + "GPU", + "space imaging", + "revenue" + ], + "connect_to": [ + { + "target": "22cb2d0d-cea3-5951-9a89-b5e53e5f0bc1", + "relation": "embeds", + "ref": "[images/image-16-And one of.jpg]", + "position": { + "start": 1277, + "end": 1309 + } + } + ] + } + }, + { + "chunk_id": "672e58f5-030b-5fd4-b408-334d4ef4610d", + "type": "image", + "content": "\nAnalyst, Evercore Group LLC Yes. Let's see if we can help on this question. So when you think about our top five, as we articulated as being our CSPs, our hyperscalers, and they have right now, [ph] as I said (49:15) about 50% of our total revenue. There's a big organization, therefore, of diversity of all different other types of companies that we are working with. That it goes through our AI model makers, that goes through our enterprises, that goes to supercomputing, it goes to our sovereigns. There's a lot of other different facts on there. But you are correct, it's a very fast growing area as well.0\n[images/image-17-Analyst, E.jpg]\n", + "path": "images/image-17-Analyst, E.jpg_2", + "metadata": { + "length": 646, + "summary": "image-17\nAnalyst, Evercore Group LLC Yes. Let's see if we can help on this question. So when you think about our top five, as we articulated as being our CSPs, our hyperscalers, and they have right now, [ph] as I said (49:15) about 50% of our total revenue. There's a big organization, therefore, of diversity of all different other types of companies that we are working with. That it goes through our AI model makers, that goes through our enterprises, that goes to supercomputing, it goes to our sovereigns. There's a lot of other different facts on there. But you are correct, it's a very fast growing area as well.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-17-Analyst, E.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "463ccb46-071c-5232-9414-484e0bb2efd5", + "type": "text", + "content": "Chief Financial Officer & Executive Vice President, NVIDIA Corp.\n\nAnalyst, Evercore Group LLC Yes. Let's see if we can help on this question. So when you think about our top five, as we articulated as being our CSPs, our hyperscalers, and they have right now, [ph] as I said (49:15) about 50% of our total revenue. There's a big organization, therefore, of diversity of all different other types of companies that we are working with. That it goes through our AI model makers, that goes through our enterprises, that goes to supercomputing, it goes to our sovereigns. There's a lot of other different facts on there. But you are correct, it's a very fast growing area as well.0\n[images/image-17-Analyst, E.jpg]\n\nSo, thanks for the question. We look at our capital return, very, very carefully. And we do believe that one of the most important things that we can do is really supporting the extreme ecosystem that's in front of us. That stems from everywhere, from our suppliers and the work that we need to do to assure that we can have the supply that's needed and help them from a capacity, all the way that we are in terms of the early developers of the AI solutions, that will be on our platform.\nSo we will continue to make this a very important part of our process and strategic investments. But of course, we are still repurchasing our stock. We are still with our dividend as well. And we will continue to find the right unique opportunities within the year for doing those different purchases.\nOperator: Your final question comes from Jim Schneider with Goldman Sachs. Your line is open.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Colette M. Kress_3", + "metadata": { + "length": 1596, + "summary": "The Chief Financial Officer and Executive Vice President of NVIDIA Corp. discusses the company's revenue distribution, noting that top customers (CSPs and hyperscalers) account for about 50% of total revenue, with diverse other clients including AI model makers, enterprises, supercomputing, and sovereign entities. He emphasizes careful capital return management, prioritizing support for the AI ecosystem through supplier capacity and early developer investments, while continuing stock repurchases and dividends. The passage ends with an operator introducing the final question from Jim Schneider of Goldman Sachs.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Chief", + "Financial", + "Officer", + "Executive", + "Vice", + "President", + "NVIDIA", + "Corp", + "Analyst", + "Evercore", + "Group", + "LLC", + "Yes", + "Let", + "question", + "So", + "top", + "articulated", + "CSPs", + "hyperscalers", + "ph", + "49", + "15", + "50%", + "total", + "revenue", + "There", + "big", + "organization", + "diversity", + "types", + "companies", + "working", + "That", + "AI", + "model", + "makers", + "enterprises", + "supercomputing", + "sovereigns", + "lot", + "facts", + "But", + "correct", + "fast", + "growing", + "area", + "well.0", + "We", + "capital", + "return", + "carefully", + "And", + "important", + "things", + "supporting", + "extreme", + "ecosystem", + "front", + "stems", + "suppliers", + "work", + "assure", + "supply", + "needed", + "capacity", + "terms", + "early", + "developers", + "solutions", + "platform", + "continue", + "make", + "part", + "process", + "strategic", + "investments", + "repurchasing", + "stock", + "dividend", + "find", + "unique", + "opportunities", + "year", + "purchases", + "Operator", + "Your", + "final", + "Jim", + "Schneider", + "Goldman", + "Sachs", + "line", + "open" + ], + "keywords": [ + "capital return", + "AI ecosystem", + "strategic investments" + ], + "connect_to": [ + { + "target": "672e58f5-030b-5fd4-b408-334d4ef4610d", + "relation": "embeds", + "ref": "[images/image-17-Analyst, E.jpg]", + "position": { + "start": 678, + "end": 710 + } + } + ] + } + }, + { + "chunk_id": "ebd8d672-ee16-54a2-a1e0-40c6bb0f7b1b", + "type": "image", + "content": "\nSo, thanks for the question. We look at our capital return, very, very carefully. And we do believe that one of the most important things that we can do is really supporting the extreme ecosystem that's in front of us. That stems from everywhere, from our suppliers and the work that we need to do to assure that we can have the supply that's needed and help them from a capacity, all the way that we are in terms of the early developers of the AI solutions, that will be on our platform. Thank you for taking my question. Jensen, you've previously outlined the potential to get to \\$3 trillion to \\$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter.0\n[images/image-20-So, thanks.jpg]\n", + "path": "images/image-20-So, thanks.jpg", + "metadata": { + "length": 794, + "summary": "image-20\nSo, thanks for the question. We look at our capital return, very, very carefully. And we do believe that one of the most important things that we can do is really supporting the extreme ecosystem that's in front of us. That stems from everywhere, from our suppliers and the work that we need to do to assure that we can have the supply that's needed and help them from a capacity, all the way that we are in terms of the early developers of the AI solutions, that will be on our platform. Thank you for taking my question. Jensen, you've previously outlined the potential to get to \\$3 trillion to \\$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter.0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-20-So, thanks.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "981f0a25-944f-5cde-b391-efcfa287f2ac", + "type": "text", + "content": "Analyst, Goldman Sachs & Co. LLC\n\nSo, thanks for the question. We look at our capital return, very, very carefully. And we do believe that one of the most important things that we can do is really supporting the extreme ecosystem that's in front of us. That stems from everywhere, from our suppliers and the work that we need to do to assure that we can have the supply that's needed and help them from a capacity, all the way that we are in terms of the early developers of the AI solutions, that will be on our platform. Thank you for taking my question. Jensen, you've previously outlined the potential to get to \\$3 trillion to \\$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter.0\n[images/image-20-So, thanks.jpg]\n\nThank you for taking my question. Jensen, you've previously outlined the potential to get to \\$3 trillion to \\$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter.\nThe question is, what are some of the key application areas that you believe are most likely to drive that inflection? Is that physical AI, agentic or something else? And do you still feel good about that \\$3 trillion to \\$4 trillion envelope? Thank you.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/James Edward Schneider", + "metadata": { + "length": 1352, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Analyst", + "Goldman", + "Sachs", + "Co", + "LLC", + "So", + "question", + "We", + "capital", + "return", + "carefully", + "And", + "important", + "things", + "supporting", + "extreme", + "ecosystem", + "front", + "That", + "stems", + "suppliers", + "work", + "assure", + "supply", + "needed", + "capacity", + "terms", + "early", + "developers", + "AI", + "solutions", + "platform", + "Thank", + "taking", + "Jensen", + "ve", + "previously", + "outlined", + "potential", + "trillion", + "data", + "center", + "CapEx", + "2030", + "implies", + "acceleration", + "growth", + "rates", + "sort", + "guided", + "quarter.0", + "quarter", + "The", + "key", + "application", + "areas", + "drive", + "inflection", + "Is", + "physical", + "agentic", + "feel", + "good", + "envelope" + ], + "keywords": [], + "connect_to": [ + { + "target": "ebd8d672-ee16-54a2-a1e0-40c6bb0f7b1b", + "relation": "embeds", + "ref": "[images/image-20-So, thanks.jpg]", + "position": { + "start": 794, + "end": 826 + } + } + ] + } + }, + { + "chunk_id": "570655e7-fd3d-5e83-9e1f-244ca5e99803", + "type": "image", + "content": "\nThank you for taking my question. Jensen, you've previously outlined the potential to get to \\$3 trillion to \\$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter. Yeah. Let's back that up and just reason through it from a few different ways. So the first way is on first principles. The way that software is done in the future using AI is token driven. And I think everybody talks about tokenomics and, talks about data centers generating tokens and inference is about generating tokens. And, we generate0\n[images/image-21-Thank you .jpg]\n", + "path": "images/image-21-Thank you .jpg", + "metadata": { + "length": 647, + "summary": "image-21\nThank you for taking my question. Jensen, you've previously outlined the potential to get to \\$3 trillion to \\$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter. Yeah. Let's back that up and just reason through it from a few different ways. So the first way is on first principles. The way that software is done in the future using AI is token driven. And I think everybody talks about tokenomics and, talks about data centers generating tokens and inference is about generating tokens. And, we generate0", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "file_path": "images/image-21-Thank you .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f2773083-0cc3-57af-8055-e7e77c5cb62d", + "type": "text", + "content": "Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.\n\nThank you for taking my question. Jensen, you've previously outlined the potential to get to \\$3 trillion to \\$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter. Yeah. Let's back that up and just reason through it from a few different ways. So the first way is on first principles. The way that software is done in the future using AI is token driven. And I think everybody talks about tokenomics and, talks about data centers generating tokens and inference is about generating tokens. And, we generate0\n[images/image-21-Thank you .jpg]\n\nYeah. Let's back that up and just reason through it from a few different ways. So the first way is on first principles. The way that software is done in the future using AI is token driven. And I think everybody talks about tokenomics and, talks about data centers generating tokens and inference is about generating tokens. And, we generate\ntokens. We're just talking about tokens. How NVIDIA's NVLink 72 enabled us to generate tokens at 50 times better performance per unit energy than the previous generation. And so token generation is at the center of almost everything that relates to software in the future and relates to computing.\nIf you look at the way we use computing in the past, however, the amount of computation demand for software in the past is a tiny fraction of what is necessary in the future. And AI is here. AI is not going to go back. AI is only going to – only get better from here. And so if you think about it and you said, okay, well, the world was investing about \\$300 billion to \\$400 billion a year in classical computing. And now AI is here and the amount of computation necessary is a thousand times higher than the way we used to do computing, the computing demand is just a lot higher.\nAnd so if we continue to believe there's value in it, and we'll talk about that in a second, then the world will invest to produce that token. And so the amount of token generation capability that the world needs is a lot more than \\$700 billion.\nAnd I'm fairly confident that we're going to continue to generate tokens. We're going to continue to invest in compute capacity from this point out. And fundamentally, because every single company depends on software, every software will depend on AI, and so every company will produce tokens. And that's the reason why I call them AI factories.\nAnd whether you're a company in the cloud data centers, you have AI factories to generate tokens for your revenues. If you're a enterprise software company, you're going to generate tokens for the agentic systems that are on top of your tools. If you are a robotics factory and self-driving car's first indication of that, you have huge supercomputers, which are basically AI factories to generate tokens that goes into your cars, that becomes its AI. And then you also have to put computers inside the cars to continuously generate tokens. And so, we're fairly sure now that this is the future of computing.\nNow, why is it so certain that this is the future of computing? And the reason for that is because the way we used to do software was pre-recorded. Everything was captured a priori. We pre-compile the software, we pre-write the content, we pre-record the videos. But now everything is generative in real-time. And when it's generated in real-time, it can take into context of the person, the situation, the query, and the intentions could all be taken into consideration to generate the outcome of this new software we call AI, agentic AI.\nAnd so the amount of computation necessary is far, far greater than prerecorded, just as a computer has a lot more computation capability than a DVD recorder, a DVD player that was pre-recorded, artificial intelligence needs a lot more computing capability than the way we used to do software in the past.\nNow, the question about computation, about sustainability at the first level is just at the computer science level, this is the way computing is going to be done. Now, from an industrial level, because all of our companies, in the final analysis are powered by software and the cloud companies are powered by software. And if the new software requires tokens to be generated and the tokens are monetized, then it stands to reason that their data center build-out directly drives their revenues. And so compute drives revenues. And I think they all understand that. I think people are increasingly starting to understand that as well.\nAnd then lastly, the benefits that AI produces for the world ultimately has to generate revenues. And we're seeing right in front, right being developed as we see – as we stand here, agentic AI has turned an inflection point, and it literally happened in the last couple of two, three months. Of course, inside the industry, we've been seeing it for a while, probably six months or so, but the world is now awaken to the agentic AI inflection. The agents are super smart. They're solving real problems. Coding is obviously supported by agentic systems now. And all of our coders here at NVIDIA are using agentic systems, either Claude Code or OpenAI Codex enormously to – and oftentimes both and Cursor, oftentimes all three, depends on the use case. But they have agents and co-design partners, engineering partners to help them solve problems.\nAnd you could see their revenues skyrocketing. These companies, in the case of Anthropic, I think their revenues 10xed in a year. And they are severely capacity constrained because demand is just incredible. And the token demand is incredible. The token generation rate is growing exponentially. And the same thing with, of course, OpenAI, their demand is incredible. And so the more compute that they can stand online, bring online, the faster their revenues will grow. And that goes back to the comment that I was saying that inference is revenues, that compute equals revenues now in this new world.\nAnd in a lot of ways, that's the reason why we say it's a new industrial revolution. There are new factories, new infrastructure being built and this new way of doing computing is not going to go back. And so to the extent that we believe that producing tokens is going to be the future of computing, which I believe, and I think largely the industry believes, then we're going to be building out this capacity from this point forward and continue to expand from here.\nNow the thing that is – the wave that we're seeing now is the agentic AI inflection and the next inflection beyond that is physical AI, where we take AI and these agentic systems into the physical applications such as manufacturing, such as robotics. And so that's a giant opportunity ahead. Okay.\nOperator: This concludes the question-and-answer session. I'll turn the call to Toshiya Hari.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_10", + "metadata": { + "length": 6933, + "summary": "The speaker argues that AI-driven token generation is the future of computing, requiring far more computation than classical methods. They note that data center CapEx could reach $3-4 trillion by 2030, as token generation drives revenue for cloud and enterprise companies. The rise of agentic AI and physical AI are key inflection points, with demand for compute growing exponentially. Inference is directly linked to revenue, making AI factories a new industrial revolution.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Co", + "Founder", + "President", + "Chief", + "Executive", + "Officer", + "Director", + "NVIDIA", + "Corp", + "Thank", + "taking", + "question", + "Jensen", + "ve", + "previously", + "outlined", + "potential", + "trillion", + "data", + "center", + "CapEx", + "2030", + "implies", + "acceleration", + "growth", + "rates", + "sort", + "guided", + "quarter", + "Yeah", + "Let", + "back", + "reason", + "ways", + "So", + "principles", + "The", + "software", + "future", + "AI", + "token", + "driven", + "And", + "talks", + "tokenomics", + "centers", + "generating", + "tokens", + "inference", + "generate0", + "generate", + "We", + "talking", + "How", + "NVLink", + "72", + "enabled", + "50", + "times", + "performance", + "unit", + "energy", + "previous", + "generation", + "relates", + "computing", + "If", + "past", + "amount", + "computation", + "demand", + "tiny", + "fraction", + "world", + "investing", + "300", + "billion", + "400", + "year", + "classical", + "thousand", + "higher", + "lot", + "continue", + "ll", + "talk", + "invest", + "produce", + "capability", + "700", + "fairly", + "confident", + "compute", + "capacity", + "point", + "fundamentally", + "single", + "company", + "depends", + "depend", + "call", + "factories", + "cloud", + "revenues", + "enterprise", + "agentic", + "systems", + "top", + "tools", + "robotics", + "factory", + "driving", + "car", + "indication", + "huge", + "supercomputers", + "basically", + "cars", + "put", + "computers", + "inside", + "continuously", + "Now", + "pre", + "recorded", + "Everything", + "captured", + "priori", + "compile", + "write", + "content", + "record", + "videos", + "But", + "generative", + "real", + "time", + "generated", + "context", + "person", + "situation", + "query", + "intentions", + "consideration", + "outcome", + "greater", + "prerecorded", + "computer", + "DVD", + "recorder", + "player", + "artificial", + "intelligence", + "sustainability", + "level", + "science", + "industrial", + "companies", + "final", + "analysis", + "powered", + "requires", + "monetized", + "stands", + "build", + "directly", + "drives", + "understand", + "people", + "increasingly", + "starting", + "lastly", + "benefits", + "produces", + "ultimately", + "front", + "developed", + "stand", + "turned", + "inflection", + "literally", + "happened", + "couple", + "months", + "Of", + "industry", + "awaken", + "agents", + "super", + "smart", + "They", + "solving", + "problems", + "Coding", + "supported", + "coders", + "Claude", + "Code", + "OpenAI", + "Codex", + "enormously", + "oftentimes", + "Cursor", + "case", + "design", + "partners", + "engineering", + "solve", + "skyrocketing", + "These", + "Anthropic", + "10xed", + "severely", + "constrained", + "incredible", + "rate", + "growing", + "exponentially", + "thing", + "online", + "bring", + "faster", + "grow", + "comment", + "equals", + "revolution", + "There", + "infrastructure", + "built", + "extent", + "producing", + "largely", + "believes", + "building", + "forward", + "expand", + "wave", + "physical", + "applications", + "manufacturing", + "giant", + "opportunity", + "ahead", + "Okay", + "Operator", + "This", + "concludes", + "answer", + "session", + "turn", + "Toshiya", + "Hari" + ], + "keywords": [ + "AI tokens", + "data centers", + "agentic AI" + ], + "connect_to": [ + { + "target": "570655e7-fd3d-5e83-9e1f-244ca5e99803", + "relation": "embeds", + "ref": "[images/image-21-Thank you .jpg]", + "position": { + "start": 686, + "end": 718 + } + } + ] + } + }, + { + "chunk_id": "246fa2dc-12b6-53f2-a5d4-c797bce952a6", + "type": "text", + "content": "Vice President, Investor Relations & Strategic Finance, NVIDIA Corp.\nIn closing, please note, Jensen will be participating in a fireside chat at the Morgan Stanley TMT Conference in San Francisco on March 4. He'll also be giving a keynote at GTC in San Jose on March 16. Our earnings call to discuss the results of our first quarter of fiscal 2027 is scheduled for May 20. Thank you for joining us today. Operator, please go ahead and close the call.\nOperator: Thank you. This concludes today's conference call. You may now disconnect.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Toshiya Hari", + "metadata": { + "length": 535, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "Vice", + "President", + "Investor", + "Relations", + "Strategic", + "Finance", + "NVIDIA", + "Corp", + "In", + "closing", + "note", + "Jensen", + "participating", + "fireside", + "chat", + "Morgan", + "Stanley", + "TMT", + "Conference", + "San", + "Francisco", + "March", + "He", + "ll", + "giving", + "keynote", + "GTC", + "Jose", + "16", + "Our", + "earnings", + "call", + "discuss", + "results", + "quarter", + "fiscal", + "2027", + "scheduled", + "May", + "20", + "Thank", + "joining", + "today", + "Operator", + "ahead", + "close", + "This", + "concludes", + "conference", + "You", + "disconnect" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c6f6c60a-4fce-51a0-afe5-e2d478f5d2c9", + "type": "text", + "content": "The information herein is based on sources we believe to be reliable but is not guaranteed by us and does not purport to be a complete or error-free statement or summary of the available data. As such, we do not warrant, endorse or guarantee the completeness, accuracy, integrity, or timeliness of the information. You must evaluate, and bear all risks associated with, the use of any information provided hereunder, including any reliance on the accuracy, completeness, safety or usefulness of such information. This information is not intended to be used as the primary basis of investment decisions. It should not be construed as advice designed to meet the particular investment needs of any investor. This report is published solely for information purposes, and is not to be construed as financial or other advice or as an offer to sell or the solicitation of an offer to buy any security in any state where such an offer or solicitation would be illegal. Any information expressed herein on this date is subject to change without notice. Any opinions or assertions contained in this information do not represent the opinions or beliefs of FactSet CallStreet, LLC. FactSet CallStreet, LLC, or one or more of its employees, including the writer of this report, may have a position in any of the securities discussed herein.\nTHE INFORMATION PROVIDED TO YOU HEREUNDER IS PROVIDED \"AS IS,\" AND TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, FactSet CallStreet, LLC AND ITS LICENSORS, BUSINESS ASSOCIATES AND SUPPLIERS DISCLAIM ALL WARRANTIES WITH RESPECT TO THE SAME, EXPRESS, IMPLIED AND STATUTORY, INCLUDING WITHOUT LIMITATION ANY IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, ACCURACY, COMPLETENESS, AND NON-INFRINGEMENT. TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, NEITHER FACTSET CALLSTREET, LLC NOR ITS OFFICERS, MEMBERS, DIRECTORS, PARTNERS, AFFILIATES, BUSINESS ASSOCIATES, LICENSORS OR SUPPLIERS WILL BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL OR PUNITIVE DAMAGES, INCLUDING WITHOUT LIMITATION DAMAGES FOR LOST PROFITS OR REVENUES, GOODWILL, WORK STOPPAGE, SECURITY BREACHES, VIRUSES, COMPUTER FAILURE OR MALFUNCTION, USE, DATA OR OTHER INTANGIBLE LOSSES OR COMMERCIAL DAMAGES, EVEN IF ANY OF SUCH PARTIES IS ADVISED OF THE POSSIBILITY OF SUCH LOSSES, ARISING UNDER OR IN CONNECTION WITH THE INFORMATION PROVIDED HEREIN OR ANY OTHER SUBJECT MATTER HEREOF.\nThe contents and appearance of this report are Copyrighted FactSet CallStreet, LLC 2026 CallStreet and FactSet CallStreet, LLC are trademarks and service marks of FactSet CallStreet, LLC. All other trademarks mentioned are trademarks of their respective companies. All rights reserved.", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/Disclaimer", + "metadata": { + "length": 2710, + "summary": "This document is a legal disclaimer stating that the information provided is based on reliable sources but is not guaranteed for completeness or accuracy. It disclaims all warranties and liability for any damages arising from its use, clarifies it is not investment advice, and notes copyright ownership by FactSet CallStreet, LLC.", + "page_nums": [], + "document_top_summary": "This document includes: NVIDIA Corp. (NVDA), MANAGEMENT DISCUSSION SECTION, QUESTION AND ANSWER SECTION, Disclaimer", + "tokens": [ + "The", + "information", + "based", + "sources", + "reliable", + "guaranteed", + "purport", + "complete", + "error", + "free", + "statement", + "summary", + "data", + "As", + "warrant", + "endorse", + "guarantee", + "completeness", + "accuracy", + "integrity", + "timeliness", + "You", + "evaluate", + "bear", + "risks", + "provided", + "hereunder", + "including", + "reliance", + "safety", + "usefulness", + "This", + "intended", + "primary", + "basis", + "investment", + "decisions", + "It", + "construed", + "advice", + "designed", + "meet", + "investor", + "report", + "published", + "solely", + "purposes", + "financial", + "offer", + "sell", + "solicitation", + "buy", + "security", + "state", + "illegal", + "Any", + "expressed", + "date", + "subject", + "change", + "notice", + "opinions", + "assertions", + "contained", + "represent", + "beliefs", + "FactSet", + "CallStreet", + "LLC", + "employees", + "writer", + "position", + "securities", + "discussed", + "THE", + "INFORMATION", + "PROVIDED", + "TO", + "YOU", + "HEREUNDER", + "IS", + "AS", + "AND", + "MAXIMUM", + "EXTENT", + "PERMITTED", + "BY", + "APPLICABLE", + "LAW", + "ITS", + "LICENSORS", + "BUSINESS", + "ASSOCIATES", + "SUPPLIERS", + "DISCLAIM", + "ALL", + "WARRANTIES", + "WITH", + "RESPECT", + "SAME", + "EXPRESS", + "IMPLIED", + "STATUTORY", + "INCLUDING", + "WITHOUT", + "LIMITATION", + "ANY", + "OF", + "MERCHANTABILITY", + "FITNESS", + "FOR", + "PARTICULAR", + "PURPOSE", + "ACCURACY", + "COMPLETENESS", + "NON", + "INFRINGEMENT", + "NEITHER", + "FACTSET", + "CALLSTREET", + "NOR", + "OFFICERS", + "MEMBERS", + "DIRECTORS", + "PARTNERS", + "AFFILIATES", + "OR", + "WILL", + "BE", + "LIABLE", + "INDIRECT", + "INCIDENTAL", + "SPECIAL", + "CONSEQUENTIAL", + "PUNITIVE", + "DAMAGES", + "LOST", + "PROFITS", + "REVENUES", + "GOODWILL", + "WORK", + "STOPPAGE", + "SECURITY", + "BREACHES", + "VIRUSES", + "COMPUTER", + "FAILURE", + "MALFUNCTION", + "USE", + "DATA", + "OTHER", + "INTANGIBLE", + "LOSSES", + "COMMERCIAL", + "EVEN", + "IF", + "SUCH", + "PARTIES", + "ADVISED", + "POSSIBILITY", + "ARISING", + "UNDER", + "IN", + "CONNECTION", + "HEREIN", + "SUBJECT", + "MATTER", + "HEREOF", + "contents", + "appearance", + "Copyrighted", + "2026", + "trademarks", + "service", + "marks", + "All", + "mentioned", + "respective", + "companies", + "rights", + "reserved" + ], + "keywords": [ + "Disclaimer", + "Investment", + "Liability" + ], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/doc_nav.json b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/doc_nav.json new file mode 100644 index 000000000..a8d215fd8 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/doc_nav.json @@ -0,0 +1,425 @@ +{ + "version": "1.0", + "file_name": "NVDA Q4 FY26 Earnings Call Transcript.pdf", + "stats": { + "total_chunks": 63, + "text_chunks": 38, + "image_chunks": 25, + "table_chunks": 0, + "max_depth": 2 + }, + "sections": [ + { + "title": "Root", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf", + "level": 1, + "summary": "25-Feb-2026", + "chunk_count": 1, + "children": [] + }, + { + "title": "NVIDIA Corp. (NVDA)", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)", + "level": 1, + "summary": "Q4 2026 Earnings Call", + "chunk_count": 4, + "children": [ + { + "title": "CORPORATE PARTICIPANTS", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)/CORPORATE PARTICIPANTS", + "level": 2, + "summary": "Toshiya Hari Vice President, Investor Relations & Strategic Finance, NVIDIA Corp. Colette M. Kress Chief Financial Officer & Executive Vice President, NVIDIA Corp. Jen Hsun Huang Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp.", + "chunk_count": 1, + "children": [] + }, + { + "title": "OTHER PARTICIPANTS", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)/OTHER PARTICIPANTS", + "level": 2, + "summary": "Vivek Arya Analyst, BofA Securities, Inc. Joe Moore Analyst, Morgan Stanley & Co. LLC Harlan Sur Analyst, JPMorgan Securities LLC", + "chunk_count": 1, + "children": [] + }, + { + "title": "C.J. Muse", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/NVIDIA Corp. (NVDA)/C.J. Muse", + "level": 2, + "summary": "Analyst, Cantor Fitzgerald & Co. Stacy A. Rasgon Analyst, Bernstein Research Atif Malik Analyst, Citigroup Global Markets, Inc. Ben Reitzes Analyst, Melius Research LLC Antoine Chkaiban Analyst, New Street Research LLP Mark Lipacis Analyst, Evercore Group LLC Aaron Rakers Analyst, Wells Fargo Securities LLC Timothy Arcuri Analyst, UBS Securities LLC James Edward Schneider Analyst, Goldman Sachs & Co. LLC", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "MANAGEMENT DISCUSSION SECTION", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION", + "level": 1, + "summary": "Operator: Good afternoon. My name is Sarah, and I will be your conference operator today. At this time, I would like to welcome everyone to NVIDIA's Fourth Quarter Earnings Call. All lines have been placed on mute to prevent any background noise. After the speakers' remarks, there will be a question-and-answer session. [Operator Instructions] Thank you. Toshiya Hari, you may begin your conference.", + "chunk_count": 5, + "children": [ + { + "title": "Toshiya Hari", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Toshiya Hari", + "level": 2, + "summary": "NVIDIA held its Q4 fiscal 2026 earnings call on February 25, 2026, with CEO Jensen Huang and CFO Colette Kress. The call included forward-looking statements, non-GAAP financial measures, and a replay available until the Q1 FY27 call. Content is proprietary and cannot be reproduced without permission.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Colette M. Kress", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Colette M. Kress", + "level": 2, + "summary": "NVIDIA reported record Q4 revenue of $68 billion, up 73% year-over-year, driven by strong Data Center growth (up 75% to $62 billion) and Blackwell architecture demand. Full-year Data Center revenue reached $194 billion. Networking revenue hit $11 billion, up 3.5x. Gaming revenue rose 47% to $3.7 billion. Professional Visualization crossed $1 billion. Automotive revenue was $604 million. Q1 FY27 revenue outlook is $78 billion. The company highlighted AI infrastructure demand, sovereign AI, and the Rubin platform launch.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Jen Hsun Huang", + "level": 2, + "summary": "NVIDIA's CEO highlights deepened partnerships with OpenAI, Meta, Anthropic, and xAI. OpenAI launched GPT-5.3-Codex on NVIDIA systems. Meta is deploying millions of Blackwell and Rubin GPUs. NVIDIA invested $10 billion in Anthropic, whose Claude Cowork platform drives enterprise AI adoption. A non-exclusive licensing agreement with Groq was also announced.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Toshiya Hari_2", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/MANAGEMENT DISCUSSION SECTION/Toshiya Hari_2", + "level": 2, + "summary": "Vice President, Investor Relations & Strategic Finance, NVIDIA Corp. We will now transition to Q&A. Operator, please poll for questions.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "QUESTION AND ANSWER SECTION", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION", + "level": 1, + "summary": "Operator: [Operator Instructions] Your first question comes from Vivek Arya with Bank of America Securities. Your line is open.", + "chunk_count": 27, + "children": [ + { + "title": "Vivek Arya", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Vivek Arya", + "level": 2, + "summary": "An analyst questions NVIDIA's confidence in customer CapEx growth, noting cloud CapEx near $700 billion and compressed cash flow. Jensen is asked about growth visibility into 2027 and whether NVIDIA can grow if customer CapEx does not increase.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang", + "level": 2, + "summary": "The speaker, an analyst from BofA Securities, expresses strong confidence in cash flow growth driven by the inflection of agentic AI. They argue that in the new AI era, compute capacity directly equals revenues, as tokens require compute to generate. Citing productive use of tools like Codex and Claude Code, and enterprise ISVs building agentic systems, they assert we have reached an inflection point where tokens are profitable for both customers and cloud providers. The speaker notes that annual CapEx has shifted from $300-400 billion on traditional software to AI, where compute translates directly to growth and revenues.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Joe Moore", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Joe Moore", + "level": 2, + "summary": "The speaker expresses confidence in cash flow growth driven by the inflection of agentic AI and enterprise adoption. They argue that in the AI era, compute equals revenues, as generating tokens requires compute capacity. The discussion highlights strategic investments in companies like Anthropic and OpenAI, and views the balance sheet as a tool to strengthen NVIDIA's ecosystem position and participate in growth.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_2", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_2", + "level": 2, + "summary": "The speaker, a co-founder and CEO of NVIDIA, emphasizes the company's ecosystem as its core strength. They highlight that startups, clouds, data centers, and AI systems are built on NVIDIA's platform, starting from CUDA. The goal is to expand this ecosystem across AI domains like language, physics, biology, robotics, and manufacturing. NVIDIA has evolved from a GPU computing platform to an AI infrastructure company, with stacks for computing, AI models, networking, and DPUs. Investments focus on deepening ecosystem reach across enterprise, industrial, science, and robotics sectors.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Harlan Sur", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Harlan Sur", + "level": 2, + "summary": "The passage discusses the growth of networking revenues in NVIDIA's data center business, highlighting a 3.6x year-over-year increase in Q4 driven by the Spectrum-X Ethernet Switching platform. The annualized run rate for Spectrum-X is estimated to have risen from around $10 billion in the first half of last year to $11-12 billion in the second half. The speaker asks Jensen about the current trend of the Spectrum run rate, especially with upcoming products like Spectrum-XGS and the 102T Spectrum-6 Switching platform.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_3", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_3", + "level": 2, + "summary": "NVIDIA positions itself as an AI infrastructure company, with networking as a key growth area. The company invented NVLink for rack-scale computing and uses Spectrum-X and InfiniBand for scaling out. Networking revenue is accelerating, with Spectrum-X Ethernet being a major success. NVIDIA is now a leading networking company, and its AI infrastructure business is growing rapidly due to the significant value of network performance improvements in large AI factories.", + "chunk_count": 1, + "children": [] + }, + { + "title": "C.J. Muse", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/C.J. Muse", + "level": 2, + "summary": "The passage discusses NVIDIA's dominance in networking, highlighting their success with Spectrum-X Ethernet and continued support for InfiniBand. It also addresses a question about future roadmap, specifically regarding customized silicon for workloads or customers, leveraging a dielet architecture.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_4", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_4", + "level": 2, + "summary": "NVIDIA's CEO discusses the company's approach to dielet usage, preferring to minimize dielet crossings to reduce latency and power. He contrasts this with competitors, highlighting NVIDIA's architectural and software advantages through CUDA. The architecture compatibility across GPU generations (A100, Hopper, Blackwell) ensures software optimizations benefit the entire installed base, improving performance per dollar and per watt. He also hints at extending architecture with Groq as an accelerator, similar to the Mellanox integration.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Stacy A. Rasgon", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Stacy A. Rasgon", + "level": 2, + "summary": "The passage discusses NVIDIA's GPU architecture compatibility, emphasizing how CUDA enables infrastructure versatility. The speaker notes that architectural compatibility across GPU generations (Blackwell, Hopper, Ampere) allows software optimizations to benefit the entire installed base, keeping older GPUs like A100 performant. The second part addresses a question about sequential growth in data center revenue, noting a $10 billion quarterly increase and asking about similar acceleration for the upcoming Rubin architecture. A follow-up question asks about gaming revenue expectations for fiscal 2027 given memory constraints.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Colette M. Kress", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Colette M. Kress", + "level": 2, + "summary": "The CFO discusses sequential growth expectations for NVIDIA, noting a $10 billion sequential increase in data center revenue. Blackwell sales will continue alongside the ramp of Vera Rubin in the second half of the year, with strong customer demand expected. Gaming supply remains tight for a few quarters, with potential year-over-year growth later in the year.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Atif Malik", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Atif Malik", + "level": 2, + "summary": "Analyst, Citigroup Global Markets, Inc. It's too early yet to determine how much in terms of that Vera Rubin, that beginning ramp will start in the second half and we'll get through it. But no confusion in terms of the strong demand and the interest. We do expect pretty much every single customer to be purchasing Vera Rubin. The question is, are, how soon are we in market and how soon are they able to stand that up in terms of in their data centers? That was your first part. Thank you for taking my question. Jensen, I'm curious if you can touch on the importance of CUDA as now more of the investment dollars in AI are coming from inference workloads?0 [images/image-11-It's too e.jpg] Thank you for taking my question. Jensen, I'm curious if you can touch on the importance of CUDA as now more of the investment dollars in AI are coming from inference workloads?", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_5", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_5", + "level": 2, + "summary": "The speaker, a co-founder and CEO of NVIDIA, discusses the critical role of CUDA and NVLink 72 in AI inference. Without CUDA, inference would be impossible. NVLink 72 enables a 50x improvement in performance per watt and 35x in performance per dollar through new parallelization algorithms. Inference performance directly translates to revenue for customers, as token generation has become exponential. For data centers, inference tokens per watt equals dollars per watt, making architecture choice crucial for maximizing revenue under power constraints. The speaker emphasizes that compute equals revenue, and investing in the right architecture is strategic for earnings.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Ben Reitzes", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Ben Reitzes", + "level": 2, + "summary": "Analyst, Melius Research LLC For the Data Centers, inference tokens per watt translates directly to the revenues of the CSPs. And the reason for that is because everybody is power limited. And so, I mean, no matter how many data centers you have, each data center, 100 megawatts or 1 gigawatt has power limits. So the architecture that has the best performance per watt translates because each token, the performance tokens per watt, each token is dollarized, tokens per watt translates to dollars per watt, which translates in a gigawatt directly to revenues. Yeah. Hey, thanks. First, let me say kudos on including the stock-comp in non-GAAP. I think that's a great move, but that isn't my question. My question is around gross margins and the sustainability of the mid-70s long-term. Should we read into the visibility on supply being available into calendar 2027 that it's sustainable until then?0 [images/image-13-For the Da.jpg] Yeah. Hey, thanks. First, let me say kudos on including the stock-comp in non-GAAP. I think that's a great move, but that isn't my question. My question is around gross margins...the visibility on supply being available into calendar 2027 that it's sustainable until then? And then Jensen, what about after that? Are there innovations in memory consumption you can unveil that makes us feel better about the ability to keep margins at that level for a long time? Thanks.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_6", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_6", + "level": 2, + "summary": "The passage discusses NVIDIA's advancements in AI inference, highlighting the critical role of CUDA and the new NVLink 72 technology. NVLink 72 enables a 50x improvement in performance per watt and 35x in performance per dollar, representing a generational leap. The speaker, a co-founder and CEO, emphasizes that delivering such dramatic performance gains is the key driver of gross margins. The strategy involves releasing a new AI infrastructure annually, with six new chips this year and more planned for the next-generation Rubin, to meet exponentially growing compute demand for modern software.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Antoine Chkaiban", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Antoine Chkaiban", + "level": 2, + "summary": "Analyst, New Street Research LLP Every – the reason why we're moving so fast is because number one, the demand for tokens in the world as a result of the inflection points that we've gone through has now – has gone completely exponential. I think we're all seeing that, to the point where even our six-year old GPUs in the cloud are completely consumed, and the pricing is going up. Hi. Thanks a lot for taking my question. I'd like to ask about space data centers, which some of your customers are considering. How feasible do you think that is and what kind of horizon? And what do the economics look like today? And how do you think that could evolve over time? Thank you.0 [images/image-14-Every – th.jpg] Hi. Thanks a lot for taking my question. I'd like to ask about space data centers, which some of your customers are considering. How feasible do you think that is and what kind of horizon? And what do the economics look like today? And how do you think that could evolve over time? Thank you.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_7", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_7", + "level": 2, + "summary": "The passage discusses the economics and technical challenges of computing in space, highlighting NVIDIA's role with the first GPU in space (Hopper). It contrasts space conditions (abundance of energy, cold, no airflow) with Earth, noting heat dissipation via radiators and the impracticality of liquid cooling. Key applications include high-resolution imaging using AI for reprojection, upscaling, and noise reduction, processing data in space rather than sending petabytes to Earth.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Mark Lipacis", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Mark Lipacis", + "level": 2, + "summary": "An analyst asks about revenue diversification, noting that hyperscalers account for over 50% of revenues but growth is led by other data center customers. The question seeks clarification on whether non-hyperscale customers grew faster, what they are doing differently, and if this trend will continue, potentially making non-hyperscalers a larger part of the business.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Colette M. Kress_2", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Colette M. Kress_2", + "level": 2, + "summary": "The speaker discusses Nvidia's revenue composition, noting that its top five customers (CSPs and hyperscalers) account for about 50% of total revenue. The remaining revenue comes from a diverse range of customers including AI model makers, enterprises, supercomputing, and sovereign entities. This diversity is seen as a strength, with the company serving various cloud providers and customers globally.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_8", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_8", + "level": 2, + "summary": "The speaker, an NVIDIA executive, discusses the company's competitive advantages: its CUDA platform's ubiquity across clouds and devices, its cultivation of a diverse ecosystem spanning robotics, self-driving cars, and AI across multiple domains, and key partnerships with OpenAI, Anthropic, xAI, and Meta. They highlight that 1.5 million AI models on Hugging Face run on NVIDIA CUDA, making the platform safe and easy to invest in.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Aaron Rakers", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Aaron Rakers", + "level": 2, + "summary": "Analyst, Wells Fargo Securities LLC And then lastly, one of the things that's really important is the partnerships that we have with OpenAI and Anthropic, with xAI, with Meta now makes – and of course, just about every single open source in the world. There's 1.5 million AI models on Hugging Face, all of it runs on NVIDIA CUDA. And so an open source in totality probably represents the largest – the second largest model in the world, OpenAI is the largest, second largest, probably all the collection of all the open sources. Yeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis.0 [images/image-18-And then l.jpg] Yeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis. So, I guess, Jensen, I'm curious of what's the importance that Vera plays in this architecture evolution as we move forward? Is this being driven more by the proliferation or the heterogeneity of inference workloads? I'm just curious of how you see that evolving for NVIDIA, particularly on a standalone CPU basis? Thank you.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_9", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_9", + "level": 2, + "summary": "NVIDIA's co-founder discusses the Vera CPU's unique architecture, emphasizing its support for LPDDR5 and focus on high data processing capabilities for AI workloads. The CPU is designed for data processing, pre-training, and post-training phases, with exceptional single-threaded performance and bandwidth ratio, making it suitable for CPU-only or GPU-accelerated environments.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Timothy Arcuri", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Timothy Arcuri", + "level": 2, + "summary": "The passage discusses GPU use in space for high-resolution imaging via optics and AI, and a Q&A about revenue diversification at UBS Securities LLC. Colette notes hyperscalers are over 50% of revenues but growth is led by other data center customers, implying non-hyperscale customers grew faster. The question explores whether these customers do different things or the same on a different scale, and if this trend will continue. Another question addresses capital deployment, noting increased purchase commitments and potential $100 billion cash generation, suggesting a large share repurchase might be warranted given stagnant stock performance.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Colette M. Kress_3", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Colette M. Kress_3", + "level": 2, + "summary": "The Chief Financial Officer and Executive Vice President of NVIDIA Corp. discusses the company's revenue distribution, noting that top customers (CSPs and hyperscalers) account for about 50% of total revenue, with diverse other clients including AI model makers, enterprises, supercomputing, and sovereign entities. He emphasizes careful capital return management, prioritizing support for the AI ecosystem through supplier capacity and early developer investments, while continuing stock repurchases and dividends. The passage ends with an operator introducing the final question from Jim Schneider of Goldman Sachs.", + "chunk_count": 1, + "children": [] + }, + { + "title": "James Edward Schneider", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/James Edward Schneider", + "level": 2, + "summary": "Analyst, Goldman Sachs & Co. LLC So, thanks for the question. We look at our capital return, very, very carefully. And we do believe that one of the most important things that we can do is really supporting the extreme ecosystem that's in front of us. That stems from everywhere, from our suppliers and the work that we need to do to assure that we can have the supply that's needed and help them from a capacity, all the way that we are in terms of the early developers of the AI solutions, that will be on our platform. Thank you for taking my question. Jensen, you've previously outlined the potential to get to \\$3 trillion to \\$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter.0 [images/image-20-So, thanks.jpg] Thank you for taking my question. Jensen, you've previously outlined the potential to get to \\$3 trillion to \\$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter. The question is, what are some of the key application areas that you believe are most likely to drive that inflection? Is that physical AI, agentic or something else? And do you still feel good about that \\$3 trillion to \\$4 trillion envelope? Thank you.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Jen Hsun Huang_10", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Jen Hsun Huang_10", + "level": 2, + "summary": "The speaker argues that AI-driven token generation is the future of computing, requiring far more computation than classical methods. They note that data center CapEx could reach $3-4 trillion by 2030, as token generation drives revenue for cloud and enterprise companies. The rise of agentic AI and physical AI are key inflection points, with demand for compute growing exponentially. Inference is directly linked to revenue, making AI factories a new industrial revolution.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Toshiya Hari", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/QUESTION AND ANSWER SECTION/Toshiya Hari", + "level": 2, + "summary": "Vice President, Investor Relations & Strategic Finance, NVIDIA Corp. In closing, please note, Jensen will be participating in a fireside chat at the Morgan Stanley TMT Conference in San Francisco on March 4. He'll also be giving a keynote at GTC in San Jose on March 16. Our earnings call to discuss the results of our first quarter of fiscal 2027 is scheduled for May 20. Thank you for joining us today. Operator, please go ahead and close the call. Operator: Thank you. This concludes today's conference call. You may now disconnect.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Disclaimer", + "path": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf/Disclaimer", + "level": 1, + "summary": "This document is a legal disclaimer stating that the information provided is based on reliable sources but is not guaranteed for completeness or accuracy. It disclaims all warranties and liability for any damages arising from its use, clarifies it is not investment advice, and notes copyright ownership by FactSet CallStreet, LLC.", + "chunk_count": 1, + "children": [] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-We will no.jpg", + "summary": "image-1 We will now transition to Q&A. Operator, please poll for questions. Thanks for taking my question. I think you mentioned that you now have growth visibility into calendar 2027 also, and I think your purchase commitments kind of reflect that confidence. But Jensen, I'm curious, when you look at your top cloud customers, cloud CapEx close to \\$700 billion this year, many investors are concerned that it would be harder for this level to grow into next year. And for several of them, their cash flow generation capability is also getting compressed. So I know you're very confident about your roadmap, right, and your purchase commitments and whatnot, but how confident are you about your customers' ability to continue to grow their CapEx? And if their CapEx doesn't grow, can NVIDIA still find a way to grow in that envelope? Thank you.0" + }, + { + "path": "images/image-2-Analyst, B.jpg", + "summary": "image-2 Analyst, BofA Securities, Inc. I am confident in their cash flow growing. And the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents across the world in enterprises everywhere. You're seeing incredible compute demand because of it. In this new world of AI, compute is revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues. So in this new world of AI, compute equals revenues. And I am certain that at this point, with the productive use of Codex and Claude Code and the excitement around Claude Cowork and, just the incredible enthusiasm about OpenClaw and the enterprise versions of them. All of the enterprise ISVs who are now working on agentic systems on top of their tools platforms, I am certain at this point that we are at the inflection point. We've reached the inflection point, and we're generating profitable tokens that are productive for customers and profitable for the cloud service providers. And so the simple logic of it, the simple way to think about it is computing has changed. What used to be software running on computers, modest amount of computers, call it \\$300 billion or \\$400 billion worth of CapEx each year has now gone into AI and AI in order to have, in order to generate tokens, you need compute capacity. And that translates directly to growth and that translates directly to revenues.0" + }, + { + "path": "images/image-3-I am confi.jpg", + "summary": "image-3 I am confident in their cash flow growing. And the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents across the world in enterprises everywhere. You're seeing incredible compute demand because of it. In this new world of AI, compute is revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues. So in this new world of AI, compute equals revenues. And I am certain that at this point, with the productive use of Codex and Claude Code and the excitement around Claude Cowork and, just the incredible enthusiasm about OpenClaw and the enterprise versions of them. All of the enterprise ISVs who are now working on agentic systems on top of their tools platforms, I am certain at this point that we are at the inflection point. We've reached the inflection point, and we're generating profitable tokens that are productive for customers and profitable for the cloud service providers. And so the simple logic of it, the simple way to think about it is computing has changed. What used to be software running on computers, modest amount of computers, call it \\$300 billion or \\$400 billion worth of CapEx each year has now gone into AI and AI in order to have, in order to generate tokens, you need compute capacity. And that translates directly to growth and that translates directly to revenues. Great. Thank you. And congratulations on the numbers. You talked about some of the strategic investments that you've made into Anthropic and potentially OpenAI [ph] Claude as well (26:23) but also partners Intel, Nokia. Synopsys. You're clearly at the center of everything. Can you talk about the role of those investments and kind of how do you view the balance sheet as a tool to kind of grow the NVIDIA's position, the ecosystem and, and participate in that growth.0" + }, + { + "path": "images/image-4-Analyst, M.jpg", + "summary": "image-4 Analyst, Morgan Stanley & Co. LLC As you know, fundamentally, at the core of everything NVIDIA is our ecosystem. That's what everybody loves about our business. The richness of our ecosystem, just about every startup in the world is working on NVIDIA's platform, we are in every cloud, we're in every on-prem data center, we're all over the world's edge and robotic systems, thousands of AI natives are built on top of NVIDIA. We want to take the great opportunity that we have as we're in the beginning of this new computing era, this new computing platform shift to put everybody on NVIDIA. Everything is already built on CUDA and so we're starting from a really terrific starting point. But as we build out the entire AI ecosystem, whether it's in AI for language or physical AI or AI physics or biology or robotics or manufacturing, we want all of these ecosystems to be built on top of NVIDIA. And this is such a wonderful opportunity for us to invest into the ecosystem across the entire stack.0" + }, + { + "path": "images/image-5-Our ecosys.jpg", + "summary": "image-5 Our ecosystem is also richer today than it used to be. We used to be largely a computing platform on GPUs, but now we're a computing AI infrastructure company, and we have computing platforms on, well, every aspect of that. Good afternoon. Thanks for taking my question. Networking continues to rise as a percentage of your overall data center profile, right. Through fiscal 2026, your networking revenues accelerated on a year-over-year basis every single quarter, right. With 3.6x growth, as you guys mentioned year-over-year growth in Q4, obviously, on the strength of your scale up and scale out networking product portfolio, I would seem to remember that first half of last year, your annualized run rate on your Spectrum-X Ethernet Switching platform was around \\$10 billion annualized. It looks like that may have stepped up to around \\$11 billion, \\$12 billion in the second half of last year.0" + }, + { + "path": "images/image-6-Good after.jpg", + "summary": "image-6 Good afternoon. Thanks for taking my question. Networking continues to rise as a percentage of your overall data center profile, right. Through fiscal 2026, your networking revenues accelerated on a year-over-year basis every single quarter, right. With 3.6x growth, as you guys mentioned year-over-year growth in Q4, obviously, on the strength of your scale up and scale out networking product portfolio, I would seem to remember that first half of last year, your annualized run rate on your Spectrum-X Ethernet Switching platform was around \\$10 billion annualized. It looks like that may have stepped up to around \\$11 billion, \\$12 billion in the second half of last year. Yeah. As you know, we see ourselves as an AI infrastructure company and the AI computing infrastructure includes CPUs, GPUs, and we invented NVLink to scale up the one computing node into a giant computing rack.0" + }, + { + "path": "images/image-7-We're also.jpg", + "summary": "image-7 We're also now the largest networking company in the world. And if you look at Ethernet, we came into the Ethernet market about a couple of years ago into Ethernet switching. And I think that we're probably the largest Ethernet networking company in the world today, and surely will be soon. And so Spectrum-X Ethernet has been a home run for us. But we're open to however people want to do networking. Some people just really love the low latency and the scale up capability of InfiniBand. And we will continue to support that, of course. And some people love to integrate their networking across their data center based on Ethernet. And we created an Ethernet capability that extends Ethernet with artificial intelligence, way of processing in the data center and we're incredibly good at that and our Spectrum-X performance really shows it. Yeah. Good afternoon. Thank you for taking the question. I guess with CPX for large context Windows and Groq likely adding a decode specific solution. Curious how we should think about your future roadmap. Should we be thinking about customized silicon either by workload or customer, as an increasing focus by NVIDIA, particularly helped by your move to a dielet architecture? Thanks so much.0" + }, + { + "path": "images/image-8-Analyst, C.jpg", + "summary": "image-8 Analyst, Cantor Fitzgerald & Co. We don't use – we want to – everybody should want to extend, push out dielet as long as they can. And the reason for that is because every time you cross a dielet you have a dielet, you have to cross an interface. Every time you cross an interface, you add latency, you add power unnecessarily. We're not allergic to dielet. We use dielets already. But we try to use dielets only when we absolutely have no choice but to do so. And so we – if you look at the Grace Blackwell architecture and the Rubin architecture, we used two giant reticle limited dies [ph] and we abut them (33:53) and that reduces the amount of architecture crossing.0" + }, + { + "path": "images/image-9-With respe.jpg", + "summary": "image-9 With respect to, how we think about Groq and the low latency decoder, I've got some great ideas that I'd like to share with you at GTC, but the simple idea is that our infrastructure is incredibly versatile because of CUDA. And we're going to continue to do that. All of our GPUs are architecturally compatible, which means that when I'm working on optimizing models today for Blackwell, all of that work and all of that dedication to optimizing software stacks and new models, also benefit Hopper and also benefit Ampere. It's the reason why A100 continues to feel fresh and continues to stay performant years after we've deployed it into the world. Architecture compatibility allows us to do that. It allows us to invest enormously in software engineering and optimization, knowing that our entire installed base in the cloud, on-prem, everywhere from generations of architectures of GPUs will all benefit. And so we'll continue to do that. Hi, guys. Thanks for taking my questions. Colette, I wanted to dig a little bit into the call for sequential growth through the year. So, I mean, you grew this quarter more than \\$10 billion sequentially in data center, and the guide seems to imply the bulk of the increase \\$10 billion sequential in data centers. So, how do you see that as we go through the year, especially as Rubin ramps into the back half? Blackwell has been a pretty massive acceleration with a sequential growth. Should we expect something similar as we get to Rubin?0" + }, + { + "path": "images/image-10-Hi, guys. .jpg", + "summary": "image-10 Hi, guys. Thanks for taking my questions. Colette, I wanted to dig a little bit into the call for sequential growth through the year. So, I mean, you grew this quarter more than \\$10 billion sequentially in data center, and the guide seems to imply the bulk of the increase \\$10 billion sequential in data centers. So, how do you see that as we go through the year, especially as Rubin ramps into the back half? Blackwell has been a pretty massive acceleration with a sequential growth. Should we expect something similar as we get to Rubin? Thanks, Stacy. Let me start with the revenue going forward. Again, we're trying to look at revenue quarter-by-quarter. As you think about the full year, we are absolutely going to be still selling and providing Blackwell probably at the same time that we're also seeing Vera Rubin come to market.0" + }, + { + "path": "images/image-11-It's too e.jpg", + "summary": "image-11 It's too early yet to determine how much in terms of that Vera Rubin, that beginning ramp will start in the second half and we'll get through it. But no confusion in terms of the strong demand and the interest. We do expect pretty much every single customer to be purchasing Vera Rubin. The question is, are, how soon are we in market and how soon are they able to stand that up in terms of in their data centers? That was your first part. Thank you for taking my question. Jensen, I'm curious if you can touch on the importance of CUDA as now more of the investment dollars in AI are coming from inference workloads?0" + }, + { + "path": "images/image-12-Analyst, C.jpg", + "summary": "image-12 Analyst, Citigroup Global Markets, Inc. Without CUDA, we wouldn't know what to do with inference. The entire stack from TensorRT LLM that we introduced a few years ago, which is still the most performant inference stack in the world. Optimizing it for NVLink, requires us to discover and invent new parallelization algorithms that sits on top of CUDA, to distribute the workload and the inferencing to take advantage of the aggregate bandwidth across NVLink 72. NVLink 72 has enabled us to deliver generationally 50 times more performance per watt. It's just an incredible leap and it's sensible. NVLink 72 is a great invention. It was hard to do, the creation of the switching technology, disaggregating the switches, building the system racks, all of that. We did it all in plain sight and everybody knew how hard it was for us to do. And, but the results are incredible. So performance per watt is 50 times, performance per dollar, 35 times and so the leap in inference is incredible.0" + }, + { + "path": "images/image-13-For the Da.jpg", + "summary": "image-13 For the Data Centers, inference tokens per watt translates directly to the revenues of the CSPs. And the reason for that is because everybody is power limited. And so, I mean, no matter how many data centers you have, each data center, 100 megawatts or 1 gigawatt has power limits. So the architecture that has the best performance per watt translates because each token, the performance tokens per watt, each token is dollarized, tokens per watt translates to dollars per watt, which translates in a gigawatt directly to revenues. Yeah. Hey, thanks. First, let me say kudos on including the stock-comp in non-GAAP. I think that's a great move, but that isn't my question. My question is around gross margins and the sustainability of the mid-70s long-term. Should we read into the visibility on supply being available into calendar 2027 that it's sustainable until then?0" + }, + { + "path": "images/image-12-Analyst, C.jpg_2", + "summary": "image-12 Analyst, Citigroup Global Markets, Inc. Without CUDA, we wouldn't know what to do with inference. The entire stack from TensorRT LLM that we introduced a few years ago, which is still the most performant inference stack in the world. Optimizing it for NVLink, requires us to discover and invent new parallelization algorithms that sits on top of CUDA, to distribute the workload and the inferencing to take advantage of the aggregate bandwidth across NVLink 72. NVLink 72 has enabled us to deliver generationally 50 times more performance per watt. It's just an incredible leap and it's sensible. NVLink 72 is a great invention. It was hard to do, the creation of the switching technology, disaggregating the switches, building the system racks, all of that. We did it all in plain sight and everybody knew how hard it was for us to do. And, but the results are incredible. So performance per watt is 50 times, performance per dollar, 35 times and so the leap in inference is incredible.0" + }, + { + "path": "images/image-14-Every – th.jpg", + "summary": "image-14 Every – the reason why we're moving so fast is because number one, the demand for tokens in the world as a result of the inflection points that we've gone through has now – has gone completely exponential. I think we're all seeing that, to the point where even our six-year old GPUs in the cloud are completely consumed, and the pricing is going up. Hi. Thanks a lot for taking my question. I'd like to ask about space data centers, which some of your customers are considering. How feasible do you think that is and what kind of horizon? And what do the economics look like today? And how do you think that could evolve over time? Thank you.0" + }, + { + "path": "images/image-15-Analyst, N.jpg", + "summary": "image-15 Analyst, New Street Research LLP Well, the economics are poor today, but it's going to improve over time. As you know, the way that space works is radically different than how it works down here. There's an abundance of energy. But solar panels are large, but there's plenty of space in space.0" + }, + { + "path": "images/image-16-And one of.jpg", + "summary": "image-16 And one of the best use cases of GPUs in space is imaging, to be able to image at extremely high resolutions using, of course, optics and artificial intelligence and to be able to do that computation of reprojection of different angles and be able to upres and do noise reduction and just be able to see, be able to image at very large, very high resolutions, extremely large scales, and very, very fast. It's hard to do that by sending petabytes and Hi. Thanks for taking my question. I want to pick up with the comment you made on the script about revenue diversification. I believe, Colette, you said that hyperscalers were over 50% of revenues, but growth was led by the rest of your Data Center customers. And as a clarification, I just want to make sure I understood that, does that imply your non-hyperscale customers grew faster. And if so, what are the – can you help us understand what are the non-hyperscalers doing different. Are they doing different things than the hyperscalers or the same things on a different scale and does this – do you expect this trend to continue? Would you expect your customer base to evolve to a point where non-hyperscalers are becoming a bigger part of your, the larger part of your business? Thank you.0" + }, + { + "path": "images/image-17-Analyst, E.jpg", + "summary": "image-17 Analyst, Evercore Group LLC Yes. Let's see if we can help on this question. So when you think about our top five, as we articulated as being our CSPs, our hyperscalers, and they have right now, [ph] as I said (49:15) about 50% of our total revenue. There's a big organization, therefore, of diversity of all different other types of companies that we are working with. That it goes through our AI model makers, that goes through our enterprises, that goes to supercomputing, it goes to our sovereigns. There's a lot of other different facts on there. But you are correct, it's a very fast growing area as well.0" + }, + { + "path": "images/image-15-Analyst, N.jpg_2", + "summary": "image-15 Analyst, New Street Research LLP Well, the economics are poor today, but it's going to improve over time. As you know, the way that space works is radically different than how it works down here. There's an abundance of energy. But solar panels are large, but there's plenty of space in space.0" + }, + { + "path": "images/image-18-And then l.jpg", + "summary": "image-18 And then lastly, one of the things that's really important is the partnerships that we have with OpenAI and Anthropic, with xAI, with Meta now makes – and of course, just about every single open source in the world. There's 1.5 million AI models on Hugging Face, all of it runs on NVIDIA CUDA. And so an open source in totality probably represents the largest – the second largest model in the world, OpenAI is the largest, second largest, probably all the collection of all the open sources. Yeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis.0" + }, + { + "path": "images/image-19-Yeah. Than.jpg", + "summary": "image-19 Yeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis. Yeah, thanks. And I'll tell you some more about it at GTC. But at the highest level, we made fundamentally different architecture decisions about our CPUs compared to the rest of the world's CPUs.0" + }, + { + "path": "images/image-16-And one of.jpg_2", + "summary": "image-16 And one of the best use cases of GPUs in space is imaging, to be able to image at extremely high resolutions using, of course, optics and artificial intelligence and to be able to do that computation of reprojection of different angles and be able to upres and do noise reduction and just be able to see, be able to image at very large, very high resolutions, extremely large scales, and very, very fast. It's hard to do that by sending petabytes and Hi. Thanks for taking my question. I want to pick up with the comment you made on the script about revenue diversification. I believe, Colette, you said that hyperscalers were over 50% of revenues, but growth was led by the rest of your Data Center customers. And as a clarification, I just want to make sure I understood that, does that imply your non-hyperscale customers grew faster. And if so, what are the – can you help us understand what are the non-hyperscalers doing different. Are they doing different things than the hyperscalers or the same things on a different scale and does this – do you expect this trend to continue? Would you expect your customer base to evolve to a point where non-hyperscalers are becoming a bigger part of your, the larger part of your business? Thank you.0" + }, + { + "path": "images/image-17-Analyst, E.jpg_2", + "summary": "image-17 Analyst, Evercore Group LLC Yes. Let's see if we can help on this question. So when you think about our top five, as we articulated as being our CSPs, our hyperscalers, and they have right now, [ph] as I said (49:15) about 50% of our total revenue. There's a big organization, therefore, of diversity of all different other types of companies that we are working with. That it goes through our AI model makers, that goes through our enterprises, that goes to supercomputing, it goes to our sovereigns. There's a lot of other different facts on there. But you are correct, it's a very fast growing area as well.0" + }, + { + "path": "images/image-20-So, thanks.jpg", + "summary": "image-20 So, thanks for the question. We look at our capital return, very, very carefully. And we do believe that one of the most important things that we can do is really supporting the extreme ecosystem that's in front of us. That stems from everywhere, from our suppliers and the work that we need to do to assure that we can have the supply that's needed and help them from a capacity, all the way that we are in terms of the early developers of the AI solutions, that will be on our platform. Thank you for taking my question. Jensen, you've previously outlined the potential to get to \\$3 trillion to \\$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter.0" + }, + { + "path": "images/image-21-Thank you .jpg", + "summary": "image-21 Thank you for taking my question. Jensen, you've previously outlined the potential to get to \\$3 trillion to \\$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter. Yeah. Let's back that up and just reason through it from a few different ways. So the first way is on first principles. The way that software is done in the future using AI is token driven. And I think everybody talks about tokenomics and, talks about data centers generating tokens and inference is about generating tokens. And, we generate0" + } + ], + "tables": [] + } +} diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/full.md b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/full.md new file mode 100644 index 000000000..3092ab058 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/full.md @@ -0,0 +1,527 @@ +25-Feb-2026 + +# NVIDIA Corp. (NVDA) + +Q4 2026 Earnings Call + +## CORPORATE PARTICIPANTS + +Toshiya Hari + +Vice President, Investor Relations & Strategic Finance, NVIDIA Corp. + +Colette M. Kress + +Chief Financial Officer & Executive Vice President, NVIDIA Corp. + +Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +## OTHER PARTICIPANTS + +Vivek Arya + +Analyst, BofA Securities, Inc. + +Joe Moore + +Analyst, Morgan Stanley & Co. LLC + +Harlan Sur + +Analyst, JPMorgan Securities LLC + +C.J. Muse + +Analyst, Cantor Fitzgerald & Co. + +Stacy A. Rasgon + +Analyst, Bernstein Research + +Atif Malik + +Analyst, Citigroup Global Markets, Inc. + +Ben Reitzes + +Analyst, Melius Research LLC + +Antoine Chkaiban + +Analyst, New Street Research LLP + +Mark Lipacis + +Analyst, Evercore Group LLC + +Aaron Rakers + +Analyst, Wells Fargo Securities LLC + +Timothy Arcuri + +Analyst, UBS Securities LLC + +James Edward Schneider + +Analyst, Goldman Sachs & Co. LLC + +## MANAGEMENT DISCUSSION SECTION + +Operator: Good afternoon. My name is Sarah, and I will be your conference operator today. At this time, I would like to welcome everyone to NVIDIA's Fourth Quarter Earnings Call. All lines have been placed on mute to prevent any background noise. After the speakers' remarks, there will be a question-and-answer session. [Operator Instructions] Thank you. + +Toshiya Hari, you may begin your conference. + +## Toshiya Hari + +Vice President, Investor Relations & Strategic Finance, NVIDIA Corp. + +Thank you. Good afternoon, everyone, and welcome to NVIDIA's conference call for the fourth quarter of fiscal 2026. With me today from NVIDIA are Jensen Huang, President and Chief Executive Officer; and Colette Kress, Executive Vice President and Chief Financial Officer. + +Our call is being webcast live on NVIDIA's Investor Relations website. The webcast will be available for replay until the conference call to discuss our financial results for the first quarter of fiscal 2027. The content of today's call is NVIDIA's property. It can't be reproduced or transcribed without our prior written consent. + +During this call, we may make forward looking statements based on current expectations. These are subject to a number of significant risks and uncertainties, and our actual results may differ materially. For a discussion of factors that could affect our future financial results and business, please refer to the disclosure in today's earnings release, our most recent forms 10-K and 10-Q, and the reports that we may file on Form 8-K with the Securities and Exchange Commission. All our statements are made as of today, February 25th, 2026, based on information currently available to us. Except as required by law, we assume no obligation to update any such statements. + +During this call, we will discuss non-GAAP financial measures. You can find a reconciliation of these non-GAAP financial measures to GAAP financial measures in our CFO commentary, which is posted on our website. + +With that, let me turn the call over to Colette. + +## Colette M. Kress + +Chief Financial Officer & Executive Vice President, NVIDIA Corp. + +Thanks, Toshiya. We delivered another outstanding quarter with record revenue, operating income and free cash flow. Total revenue of \$68 billion was up 73% year-over-year, accelerating from Q3. Growth on a sequential basis was also a record, as we added \$11 billion in data center revenue across a diverse and expanding set of customers, including cloud providers, hyperscalers, AI model makers, enterprises and sovereign nations. Demand for our Blackwell architecture extreme co-designed at data center scale continues to strengthen as inference deployments grow, in addition to training. The transition to accelerated computing and the infusion of AI across existing hyperscale workloads continue to fuel our growth. Agentic and physical AI applications built on increasingly smarter and multimodal models are beginning to drive our financial performance. + +On a full year basis, data center generated revenue of \$194 billion, up 68% year-over-year. We have now scaled our data center business by nearly 13x since the emergence of ChatGPT in fiscal 2023. We look ahead, we expect sequential revenue growth throughout calendar 2026, exceeding what was included in the \$500 billion + +Blackwell and Rubin revenue opportunity we shared last year. We believe we have inventory and supply commitments in place to address future demand, including shipments, extending into calendar 2027. + +Every data center is power constrained. Customers make critical architectural decisions based on performance per watt, given these constraints and the need to maximize AI factory revenue. SemiAnalysis declared NVIDIA Inference King as recent results from InferenceX reinforced our inference leadership, with GB300 NVL72 achieving up to 50x performance per watt and 35x lower cost per token, compared with Hopper. And continuous optimization of CUDA software helped deliver up to five times better performance on GB200 NVL72 just within four months. NVIDIA produces the lowest cost per token and data centers running on NVIDIA generate the highest revenues. + +Our pace of innovation, particularly at our scale, is unmatched, fueled by an annual R&D budget approaching \$20 billion and our ability to extreme co-designed across compute and networking across chips, systems, algorithms and softwares. We intend to deliver X factor leaps in performance per watt every generation and extend our leadership position over the long term. + +Q4 data center revenue of \$62 billion increased 75% year-over-year and 22% sequentially, driven primarily by sustained strength in Blackwell and the Blackwell Ultra ramp. With NVIDIA infrastructure in high demand, even Hopper and much of the six year old Ampere-based products are sold out in the cloud. + +Nearly a year has passed since the release of our Grace Blackwell NVL72 systems. Today, nearly nine gigawatts of infrastructure on Blackwell are deployed and consumed by the major cloud service providers, hyperscalers, AI model makers and enterprises. + +Networking, a cornerstone of our data center scale infrastructure offering, was a standout this quarter, generating \$11 billion in revenue, up more than 3.5x year-over-year. Demand for our scale up and scale out technologies reached record levels, both growing double digit sequentially, driven by strong adoption of NVLink, Spectrum-X Ethernet and InfiniBand. On a year-over-year basis, growth was driven primarily by NVLink 72 scale up switches as Grace Blackwell Systems accounted for roughly two-thirds of data center revenue in the quarter. + +NVLink scale up fabric has revolutionized computing and demonstrates the power of extreme co-design across all of the chips of the supercomputer and the full stack. In Q4, we announced that we will enable AWS with NVLink to integrate with their custom silicon. + +Momentum is strong with our Spectrum-X Ethernet scale up and scale across networking, as customers work to unify distributed data centers into integrated giga scale AI factories. For the full year, our networking business exceeded \$31 billion in revenue, up more than 10-x compared to fiscal 2021, the year we acquired Mellanox. + +Our demand profile is broad, diverse and expanding beyond just chatbots. First, there's a fundamental platform shift from classical machine learning to generative AI. Strong evidence of ROI as hyperscalers upgrade massive traditional workloads to generative AI, including search, ad generation and content recommender systems, is encouraging our largest customers to accelerate their capital spending. + +For example, at Meta, advancements in their GEM model drove a 3.5 increase in ad clicks on Facebook and more than 1% gain in conversations on Instagram, translating into meaningful revenue growth. With the same NVIDIA infrastructure, Meta Superintelligence Labs can train and deploy their frontier agentic AI systems. Frontier agentic systems have reached an inflection point. Claude Code, Claude Cowork and OpenAI Codex have + +achieved useful intelligence. Adoption is skyrocketing, and tokens are profitable, driving extreme urgency to scale up compute. Compute directly translate to intelligence and revenue growth. + +Analysts expectations for 2026 CapEx across the top five cloud providers and hyperscalers who collectively account for little over 50% of our data center revenue, are up nearly \$120 billion since the start of the year and approaching \$700 billion. We continue to expect the transition of classic data center workloads to GPU accelerated computing, and the use of AI to enhance today's hyperscale workloads and contribute toward roughly half of our long-term opportunity. + +Every country will build and operate some parts of its AI infrastructure, just like with electricity and Internet today. In fiscal year 2026, our sovereign AI business, more than tripled year-over-year and over \$30 billion, driven primarily by customers based in Canada, France, the Netherlands, Singapore and the UK. Over the long run, we expect our sovereign opportunity to grow at least in line with the AI infrastructure market, as countries spend on AI proportional to their GDP. + +While small amounts of H200 products for China-based customers were approved by the US government, we have yet to generate any revenue and we do not know whether any imports will be allowed into China. Our competitors in China, bolstered by recent IPOs, are making progress and have the potential to disrupt the structure of the global AI industry over the long-term. To sustain its leadership position in AI compute, America must engage every developer and be the platform for choice for every commercial business, including those in China. + +We will continue to engage with the US and China governments and advocate for America's ability to compete around the world. We unveiled the Rubin platform last month at CES, comprised of six new chips; the Vera CPU, Rubin GPU, NVLink, 6 Switch, ConnectX-9, SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet Switch. The platform will train MoE models with one-fourth the number of GPUs and reduce inference token costs by up to 10x compared to Blackwell. + +We shipped our first Vera Rubin samples to customers earlier this week and we remain on track to commence production shipments in the second half of the year. Based on its modular cable-free tray design, Rubin will deliver improved resiliency and serviceability relative to Blackwell. We expect every cloud model builder to deploy Vera Rubin. + +Moving to gaming. Gaming revenue of \$3.7 billion increased 47% year-on-year, driven by strong Blackwell demand and improved supply. GeForce RTX is the leading platform for PC gamers, creators, and developers. + +In Q4, we added several new technologies and advancements, including DLSS 4.5, which uses AI to bring game visuals to a new level. G-SYNC Pulsar, bringing incredible clear graphics even in motion, and 35% faster LLM inference across leading AI PC frameworks. + +Looking ahead, while end demand for our products remains strong and channel inventory levels are healthy, we expect supply constraints to be the headwind to gaming in Q1 and beyond. + +For Professional Visualization, it crossed the \$1 billion mark for the first time, with revenue of \$1.3 billion, up 159% year-over-year and 74% sequentially. During the quarter, we launched the RTX PRO 5000 Blackwell Workstation with 72 gigabytes of fast memory for AI developers running LLMs and agentic workflows. + +Automotive revenue of \$604 million was up 6% year-over-year and was driven by robust demand for self-driving solutions. At CES, we introduced Alpamayo, the world's first open portfolio of reasoning, vision, language, action models, simulation blueprints and data sets enabling vehicles that can think. + +The first passenger car featuring Alpamayo built on NVIDIA Drive will be on the road soon in the new Mercedes-Benz CLA. + +Physical AI is here, having already contributed north of \$6 billion in NVIDIA revenue in fiscal year 2026. Robotaxi rides are growing exponentially with commercial fleets from Waymo, Tesla, Uber, WeRide and Zoox, and many others are expected to scale from thousands of vehicles in 2025 to millions over the next decade, creating a market poised to generate hundreds of billions of dollars of revenue. + +This expansion will demand orders of magnitude more compute with every major OEM and service provider developing on NVIDIA's platform. We continue to advance robotics development with the new NVIDIA Cosmos and Isaac GROOT open models, frameworks and NVIDIA's powered robots and autonomous machines for leading companies, including Boston Dynamics, Caterpillar, Franka Robotics, LG Electronics and NEURA Robotics. To accelerate industrial physical AI adoption, we also announced new expanding partnerships with Dassault Systèmes, Siemens and Synopsys to bring NVIDIA AI infrastructure, Omniverse digital twins, world models and CUDA-X libraries to millions of researchers, designers and engineers building the world's industries. + +Let's move to the rest of the P&L. GAAP gross margin was 75%, and non-GAAP gross margin was 75.2%, increasing sequentially as Blackwell continued to ramp. GAAP operating expenses were up 16% sequentially and up 21% on a non-GAAP basis, related to new product introductions and compute and infrastructure costs. Non-GAAP effective tax rate for the fourth quarter was 15.4%, below our outlook for the quarter, primarily due to the impact of a one-time tax benefit. + +Inventory grew 8% quarter-over-quarter, while purchase commitments also increased significantly, as we have strategically secured inventory and capacity to meet demand beyond the next several quarters. This is further out in time than usual and reflects the longer demand visibility we have. While we expect tightness in the supply for our advanced architectures to persist, we remain confident in our ability to capitalize on the growth opportunity ahead with our scale, expansive supply chain, and the longstanding partnerships continuing to serve us well. + +We generated free cash flow of \$35 billion in Q4 and \$97 billion in fiscal year 2026. For the year, we returned \$41 billion, or 43% of free cash flow to our shareholders in the form of share repurchases and dividends. We continue to invest in our technology and our ecosystem to cultivate market development, drive long-term growth, and ultimately yield total shareholder returns superior to the market or our peer group. Importantly, we will continue to run a strategic and disciplined process as it relates to our investments, and we remain committed to returning capital to our shareholders. + +Let me turn to the outlook for the first quarter. Starting this quarter, we will be including stock-based compensation expense in our non-GAAP results. Stock-based compensation is a foundational component of our compensation program to attract and retain world class talent. Let me first start with revenue. Total revenue is expected to be \$78 billion, plus or minus 2%. We expect most of our growth to be driven by Data Center. Consistent with last quarter, we are not assuming any Data Center compute revenue from China in our outlook. GAAP and non-GAAP gross margins are expected to be 74.9% and 75%, respectively, plus or minus 50 basis points. For the full year, we continue to see gross margins in the mid-70s. We will keep you updated on our progress as we prepare for the Vera Rubin transition. + +GAAP and non-GAAP operating expenses are expected to be approximately \$7.7 billion and \$7.5 billion, respectively, including stock-based compensation expense of \$1.9 billion. + +For the full year, we expect non-GAAP operating expenses to grow in the low 40s on a year-over-year basis as we continue to invest in our expanding opportunity set. For the full year fiscal year 2027, we expect GAAP and non-GAAP tax rates to be in between 7% and 19%, excluding any discrete items and material changes to our tax environment. + +With that, let me turn the call over to Jensen. I think he has a few words for us. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +This quarter we significantly deepened and expanded our partnerships with leading frontier model makers. We recently celebrated OpenAI's launch of GPT-5.3-Codex, trained with and inferencing on Grace Blackwell NVLink 72 systems. + +GPT-5.3-Codex can take on long running tasks that involve research, tool use, and complex execution. 5.3-Codex is deployed broadly inside NVIDIA, our engineers love it. + +We continue to work with OpenAI toward a partnership agreement and believe we are close. We are thrilled with our ongoing partnership with OpenAI, a once in a generation company we've had the pleasure of partnering with since their first days. + +Meta Superintelligence Labs is scaling up at lightning speed. Last week, we announced that Meta is deploying millions of Blackwells and Rubin GPUs, NVIDIA CPUs, and Spectrum-X Ethernet for training and inference. This quarter, we announced a partnership with Anthropic and a \$10 billion investment in their company. Anthropic will train an inference on Grace Blackwell and Vera Rubin systems. + +Anthropic's Claude Cowork agent platform is revolutionary and has opened a floodgates for enterprise AI adoption. Between Claude Cowork and OpenClaw, compute demand is skyrocketing and ChatGPT moment of agentic AI has arrived. + +With partnerships spanning Anthropic, Meta, OpenAI, and xAI, NVIDIA deployed across every cloud, and with our ability to build full-stack AI infrastructure from the ground up or support them in the cloud, we're uniquely positioned to partner with frontier model builders at every stage; training, inference, and AI factory scale out. + +Finally, we recently entered into a non-exclusive licensing agreement with Groq for its low-latency inference technology, and welcomed a team of brilliant engineers to NVIDIA. As we did with Mellanox, we will extend NVIDIA's architecture with Groq's innovations to enable new levels of AI infrastructure performance and value. We look forward to sharing more at GTC next month. + +Okay, back to you. + +## Toshiya Hari + +Vice President, Investor Relations & Strategic Finance, NVIDIA Corp. + +We will now transition to Q&A. Operator, please poll for questions. + +# QUESTION AND ANSWER SECTION + +Operator: [Operator Instructions] Your first question comes from Vivek Arya with Bank of America Securities. Your line is open. + +## Vivek Arya + +Analyst, BofA Securities, Inc. + +![](images/5783e08f73a3f442d8fb8ab3147ce1d0528ecd89d97d98d99b35b3698ae8ba90.jpg) + +Thanks for taking my question. I think you mentioned that you now have growth visibility into calendar 2027 also, and I think your purchase commitments kind of reflect that confidence. But Jensen, I'm curious, when you look at your top cloud customers, cloud CapEx close to \$700 billion this year, many investors are concerned that it would be harder for this level to grow into next year. And for several of them, their cash flow generation capability is also getting compressed. So I know you're very confident about your roadmap, right, and your purchase commitments and whatnot, but how confident are you about your customers' ability to continue to grow their CapEx? And if their CapEx doesn't grow, can NVIDIA still find a way to grow in that envelope? Thank you. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/7f2f6cc7c8d282228a0bdc656d68b4ccda9edf90ff6d4cae928b26733fac4aab.jpg) + +I am confident in their cash flow growing. And the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents across the world in enterprises everywhere. You're seeing incredible compute demand because of it. In this new world of AI, compute is revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues. So in this new world of AI, compute equals revenues. And I am certain that at this point, with the productive use of Codex and Claude Code and the excitement around Claude Cowork and, just the incredible enthusiasm about OpenClaw and the enterprise versions of them. All of the enterprise ISVs who are now working on agentic systems on top of their tools platforms, I am certain at this point that we are at the inflection point. We've reached the inflection point, and we're generating profitable tokens that are productive for customers and profitable for the cloud service providers. And so the simple logic of it, the simple way to think about it is computing has changed. What used to be software running on computers, modest amount of computers, call it \$300 billion or \$400 billion worth of CapEx each year has now gone into AI and AI in order to have, in order to generate tokens, you need compute capacity. And that translates directly to growth and that translates directly to revenues. + +Operator: Your next question comes from Joe Moore with Morgan Stanley. Your line is open. + +## Joe Moore + +Analyst, Morgan Stanley & Co. LLC + +![](images/2f05be69cc66b57bac9dc9ed11ebaaaddc6c1aee952a965303ab4a7af5ec8dec.jpg) + +Great. Thank you. And congratulations on the numbers. You talked about some of the strategic investments that you've made into Anthropic and potentially OpenAI [ph] Claude as well (26:23) but also partners Intel, Nokia. Synopsys. You're clearly at the center of everything. Can you talk about the role of those investments and kind of how do you view the balance sheet as a tool to kind of grow the NVIDIA's position, the ecosystem and, and participate in that growth. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/5de23582bf34f7a871f852886d3cfd3edfa7dfd8053bd8fd213936cc066600d3.jpg) + +As you know, fundamentally, at the core of everything NVIDIA is our ecosystem. That's what everybody loves about our business. The richness of our ecosystem, just about every startup in the world is working on NVIDIA's platform, we are in every cloud, we're in every on-prem data center, we're all over the world's edge and robotic systems, thousands of AI natives are built on top of NVIDIA. We want to take the great opportunity that we have as we're in the beginning of this new computing era, this new computing platform shift to put everybody on NVIDIA. Everything is already built on CUDA and so we're starting from a really terrific starting point. But as we build out the entire AI ecosystem, whether it's in AI for language or physical AI or AI physics or biology or robotics or manufacturing, we want all of these ecosystems to be built on top of NVIDIA. And this is such a wonderful opportunity for us to invest into the ecosystem across the entire stack. + +Our ecosystem is also richer today than it used to be. We used to be largely a computing platform on GPUs, but now we're a computing AI infrastructure company, and we have computing platforms on, well, every aspect of that. + +And everything from computing to AI models to networking to our DPU, all of that has computing stacks on top of it. And as I mentioned before, whether it's an enterprise or in manufacturing, industrial or science or robotics, each one of these ecosystems have different stacks, and we want to make sure that we continue to invest into our ecosystem. So, our investments are focused very squarely, strategically on expanding and deepening our ecosystem reach. + +Operator: Your next question comes from Harlan Sur with JPMorgan. Your line is open. + +## Harlan Sur + +Analyst, JPMorgan Securities LLC + +![](images/4399df0c24da704190b3568bf21bccfb1d8c451b52d7a2d96038e9c8a39bc8d8.jpg) + +Good afternoon. Thanks for taking my question. Networking continues to rise as a percentage of your overall data center profile, right. Through fiscal 2026, your networking revenues accelerated on a year-over-year basis every single quarter, right. With 3.6x growth, as you guys mentioned year-over-year growth in Q4, obviously, on the strength of your scale up and scale out networking product portfolio, I would seem to remember that first half of last year, your annualized run rate on your Spectrum-X Ethernet Switching platform was around \$10 billion annualized. It looks like that may have stepped up to around \$11 billion, \$12 billion in the second half of last year. + +Jensen, looking at your order book, especially with Spectrum-XGS, upcoming 102T Spectrum-6 Switching platforms launching soon. Where is the Spectrum run rate trending now and as you foresee exiting sort of this calendar year? + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/587cfeba26d126d25a593e96e7a20e4b703313a48fd9a954af961f49ffe96feb.jpg) + +Yeah. As you know, we see ourselves as an AI infrastructure company and the AI computing infrastructure includes CPUs, GPUs, and we invented NVLink to scale up the one computing node into a giant computing rack. + +We invented the idea of a rack scale computer. We don't ship nodes of computers, we ship racks of computers. And those – that NVLink Switch scale up system is then scaled out using Spectrum-X and InfiniBand, we support both. And then further, we also scale across data centers using Spectrum-X scale across. + +And so the way we think about networking is really an extension, it's – we offer everything openly, so that people could decide to mix and match in different scale, and however they would like to integrate it into their bespoke data center. But in the final analysis, it's all one big part of our platform. + +And the invention of NVLink really turbocharged our networking business. Every rack comes with nine nodes of switches, and each one of them has two chips in it and in the future they'll have more. And so the amount of switching that we do per rack is really quite incredible. + +We're also now the largest networking company in the world. And if you look at Ethernet, we came into the Ethernet market about a couple of years ago into Ethernet switching. And I think that we're probably the largest Ethernet networking company in the world today, and surely will be soon. And so Spectrum-X Ethernet has been a home run for us. But we're open to however people want to do networking. Some people just really love the low latency and the scale up capability of InfiniBand. And we will continue to support that, of course. And some people love to integrate their networking across their data center based on Ethernet. And we created an Ethernet capability that extends Ethernet with artificial intelligence, way of processing in the data center and we're incredibly good at that and our Spectrum-X performance really shows it. + +The difference of when you built a \$10 billion or \$20 billion AI factory, the difference of 10%, and it could be easily 20% on the effectiveness and the utilization of your network for your data center that translates to real money. And so NVIDIA's networking business is really, really growing fast. And it's, I think it's just because we built the AI infrastructure so effectively and the AI infrastructure business is growing incredibly fast. + +Operator: Your next question comes from C.J. Muse with Cantor Fitzgerald. Your line is open. + +## C.J. Muse + +Analyst, Cantor Fitzgerald & Co. + +![](images/cd0a5e77eb9ea6d98b3b45883334f0e6ba8b2ee9fd799dc7d536544e166f73f2.jpg) + +Yeah. Good afternoon. Thank you for taking the question. I guess with CPX for large context Windows and Groq likely adding a decode specific solution. Curious how we should think about your future roadmap. Should we be thinking about customized silicon either by workload or customer, as an increasing focus by NVIDIA, particularly helped by your move to a dielet architecture? Thanks so much. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/3eba27b96ac02d2fa913446a61360fd53153ea4da14506acea108a958b6c081a.jpg) + +We don't use – we want to – everybody should want to extend, push out dielet as long as they can. And the reason for that is because every time you cross a dielet you have a dielet, you have to cross an interface. Every time you cross an interface, you add latency, you add power unnecessarily. We're not allergic to dielet. We use dielets already. But we try to use dielets only when we absolutely have no choice but to do so. And so we – if you look at the Grace Blackwell architecture and the Rubin architecture, we used two giant reticle limited dies [ph] and we abut them (33:53) and that reduces the amount of architecture crossing. + +The [ph] dielet tax (33:59) shows up in the architecture effectiveness of the competitors. If you look at NVIDIA, people call it our software advantage. But where software starts and architecture starts and ends is kind of hard to tell. It's, our software is effective because our architecture is so good. And so the CUDA architecture is unquestionably more effective, more efficient, delivers more performance per flop, per watt than any computing architecture out there and it's because of the way we architect. + +With respect to, how we think about Groq and the low latency decoder, I've got some great ideas that I'd like to share with you at GTC, but the simple idea is that our infrastructure is incredibly versatile because of CUDA. And we're going to continue to do that. All of our GPUs are architecturally compatible, which means that when I'm working on optimizing models today for Blackwell, all of that work and all of that dedication to optimizing software stacks and new models, also benefit Hopper and also benefit Ampere. It's the reason why A100 continues to feel fresh and continues to stay performant years after we've deployed it into the world. Architecture compatibility allows us to do that. It allows us to invest enormously in software engineering and optimization, knowing that our entire installed base in the cloud, on-prem, everywhere from generations of architectures of GPUs will all benefit. And so we'll continue to do that. + +And allows us to extend the useful life, allows us to have innovation, flexibility, and velocity, which translates the performance and very importantly, performance per dollar and performance per watt for our customers. And so what we'll do with Groq is, you'll come to see GTC, but what we'll do is we'll extend our architecture with Groq as an accelerator in very much the ways that we extended NVIDIA's architecture with Mellanox. + +Operator: The next question comes from Stacy Rasgon with Bernstein Research. Your line is open. + +## Stacy A. Rasgon + +Analyst, Bernstein Research + +![](images/6177a334fd259bfbec04c382cb7c0e5c7a9ee7a7529b0198aff38bd466562b36.jpg) + +Hi, guys. Thanks for taking my questions. Colette, I wanted to dig a little bit into the call for sequential growth through the year. So, I mean, you grew this quarter more than \$10 billion sequentially in data center, and the guide seems to imply the bulk of the increase \$10 billion sequential in data centers. So, how do you see that as we go through the year, especially as Rubin ramps into the back half? Blackwell has been a pretty massive acceleration with a sequential growth. Should we expect something similar as we get to Rubin? + +And then I was also just hoping you could comment on, your expectations for gaming? I understand the memory issues and everything else. Do you think gaming can still grow year-over-year in fiscal 2027, or will that be under more pressure given memory? So, those two questions, please. Thank you. + +## Colette M. Kress + +Chief Financial Officer & Executive Vice President, NVIDIA Corp. + +![](images/d3c04f1d6a7332e087e486e0a76ba1fca9a3caa125072f6b1702a7a2ccc1217d.jpg) + +Thanks, Stacy. Let me start with the revenue going forward. Again, we're trying to look at revenue quarter-by-quarter. As you think about the full year, we are absolutely going to be still selling and providing Blackwell probably at the same time that we're also seeing Vera Rubin come to market. + +This is a very great architecture that helps them just today quickly standing up and have already planned on many different orders across the different customers to provide that. + +It's too early yet to determine how much in terms of that Vera Rubin, that beginning ramp will start in the second half and we'll get through it. But no confusion in terms of the strong demand and the interest. We do expect pretty much every single customer to be purchasing Vera Rubin. The question is, are, how soon are we in market and how soon are they able to stand that up in terms of in their data centers? That was your first part. + +The second part was focusing on our gaming. As much as we would love to have additional more supply, we do believe for a couple quarters, it is going to be very tight. If things improve by the end of the year, there is an opportunity to think about what that is from a year-over-year growth. But it's still too early for us to know at this time. And we'll get back to you as soon as we can. + +Operator: Your next question comes from Atif Malik with Citi. Your line is open. + +## Atif Malik + +Analyst, Citigroup Global Markets, Inc. + +![](images/f7c9e2a9e906a805881d264a25c6274c768340a7d2a5806e20c3803eae53496b.jpg) + +Thank you for taking my question. Jensen, I'm curious if you can touch on the importance of CUDA as now more of the investment dollars in AI are coming from inference workloads? + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/0865d7136bf6e3c0290dea97cae2566810065f825815f2f5a1c5b6ce1da1954e.jpg) + +Without CUDA, we wouldn't know what to do with inference. The entire stack from TensorRT LLM that we introduced a few years ago, which is still the most performant inference stack in the world. Optimizing it for NVLink, requires us to discover and invent new parallelization algorithms that sits on top of CUDA, to distribute the workload and the inferencing to take advantage of the aggregate bandwidth across NVLink 72. NVLink 72 has enabled us to deliver generationally 50 times more performance per watt. It's just an incredible leap and it's sensible. NVLink 72 is a great invention. It was hard to do, the creation of the switching technology, disaggregating the switches, building the system racks, all of that. We did it all in plain sight and everybody knew how hard it was for us to do. And, but the results are incredible. So performance per watt is 50 times, performance per dollar, 35 times and so the leap in inference is incredible. + +It's very important – it's really important to realize that inference equals revenues now for our customers because agents are generating so many tokens and the results are so effective. When the agents are coding, it's off generating thousands, tens of thousands, hundreds of thousands because they're running for minutes to hours. And so these systems, these agentic systems are spawning off different agents working as a team. + +The number of tokens that are being generated has really, really gone exponential. And so, we need to inference at a much higher speed. And when you're inferencing at a much higher speed, and each one of those tokens are dollarized, it directly translates into revenues. And so inference equals, inference performance equals revenues for our customers. + +For the Data Centers, inference tokens per watt translates directly to the revenues of the CSPs. And the reason for that is because everybody is power limited. And so, I mean, no matter how many data centers you have, each data center, 100 megawatts or 1 gigawatt has power limits. So the architecture that has the best performance per watt translates because each token, the performance tokens per watt, each token is dollarized, tokens per watt translates to dollars per watt, which translates in a gigawatt directly to revenues. + +And so you could see that every CSP understands this now, every hyperscaler understands this, that CapEx translates to compute, compute with the right architecture translates to maximizing revenues and compute equals revenues. Without investing capacity today, without investing in compute, there cannot be revenue growth and that I think everybody understands. Compute equals revenues. Choosing the right architecture is incredibly important, is more than strategic now, it directly affects their earnings. And choosing the right architecture, the one with the best performance per watt is literally everything. + +Operator: Your next question comes from Ben Reitzes with Melius Research. Your line is open. + +## Ben Reitzes + +Analyst, Melius Research LLC + +![](images/0fa007cd4eb21d44c43adce1cb0d9a25b12cf58c058b3a2fd00a2723ceaf5a49.jpg) + +Yeah. Hey, thanks. First, let me say kudos on including the stock-comp in non-GAAP. I think that's a great move, but that isn't my question. My question is around gross margins and the sustainability of the mid-70s long-term. Should we read into the visibility on supply being available into calendar 2027 that it's sustainable until then? + +And then Jensen, what about after that? Are there innovations in memory consumption you can unveil that makes us feel better about the ability to keep margins at that level for a long time? Thanks. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/4ee7a393ea814059bdb5f40f512a929e4a3292d7688d1e76d3c379f78417a821.jpg) + +The single most important lever of our gross margins is actually delivering generational leaps to our customers. That is the single most important thing. If we could deliver generationally performance per watt that exceeds dramatically what Moore's Law can do. If we can deliver performance per dollar dramatically more than the cost of our systems, than the price of our systems, then we can continue to sustain our gross margins. That's the simple, most important concept. + +Every – the reason why we're moving so fast is because number one, the demand for tokens in the world as a result of the inflection points that we've gone through has now – has gone completely exponential. I think we're all seeing that, to the point where even our six-year old GPUs in the cloud are completely consumed, and the pricing is going up. + +And so we know that the amount of computation necessary – the amount of compute necessary for the modern way of doing software is growing exponentially. And so our strategy is to deliver an entire AI infrastructure every single year. This year we introduced six new chips, Rubin next-generation will do many new chips as well. And every single generation, we are committed to deliver many X factors of performance per watt and performance per dollar. And that pace and our ability to do extreme co-design, allows us to deliver that value and that benefit to the customers. And that is the single most vital thing as it relates to our value delivered. + +Operator: Your next question comes from Antoine Chkaiban with New Street Research. Your line is open. + +## Antoine Chkaiban + +Analyst, New Street Research LLP + +![](images/b77c6bd84e53e741c4fb712a28f1f5fa1f8744ff6737b13a18ad3756854bb2ed.jpg) + +Hi. Thanks a lot for taking my question. I'd like to ask about space data centers, which some of your customers are considering. How feasible do you think that is and what kind of horizon? And what do the economics look like today? And how do you think that could evolve over time? Thank you. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/928c1edbf954bb409b08b1029898ff3c17177d5dd5fbc218470bf4623194e328.jpg) + +Well, the economics are poor today, but it's going to improve over time. As you know, the way that space works is radically different than how it works down here. There's an abundance of energy. But solar panels are large, but there's plenty of space in space. + +The heat dissipation -- it's cold in space. However, there's no airflow. And so the only way to dissipate heat is through conduction. And the radiators that you need to create are fairly large. Liquid cooling is obviously out of the question because it's kind of -- it's heavy and freezes. And so the methods that we use here on Earth are a little different than the way we would do it in space. But there are many different computing problems that really wants to be done in space. And so, NVIDIA is already the world's first GPU in space, Hopper's in space. + +And one of the best use cases of GPUs in space is imaging, to be able to image at extremely high resolutions using, of course, optics and artificial intelligence and to be able to do that computation of reprojection of different angles and be able to upres and do noise reduction and just be able to see, be able to image at very large, very high resolutions, extremely large scales, and very, very fast. It's hard to do that by sending petabytes and + +petabytes of imaging data back here on Earth and doing that work. It's easier just to do it out in space. And then, ignore all of the data collected and processed until you see something interesting. And so artificial intelligence in space will have very good, very interesting applications. + +Operator: Your next question comes from Mark Lipacis with Evercore ISI. Your line is open. + +## Mark Lipacis + +Analyst, Evercore Group LLC + +![](images/2db053192d9aa84442ecdee057ab03f2566080aa4453be3635347fa0fe75b088.jpg) + +Hi. Thanks for taking my question. I want to pick up with the comment you made on the script about revenue diversification. I believe, Colette, you said that hyperscalers were over 50% of revenues, but growth was led by the rest of your Data Center customers. And as a clarification, I just want to make sure I understood that, does that imply your non-hyperscale customers grew faster. And if so, what are the – can you help us understand what are the non-hyperscalers doing different. Are they doing different things than the hyperscalers or the same things on a different scale and does this – do you expect this trend to continue? Would you expect your customer base to evolve to a point where non-hyperscalers are becoming a bigger part of your, the larger part of your business? Thank you. + +## Colette M. Kress + +Chief Financial Officer & Executive Vice President, NVIDIA Corp. + +![](images/c7bddcae25599a5841d1eb0fada6b34b8d44b6f34fdc68d55f0cd5fd3938edba.jpg) + +Yes. Let's see if we can help on this question. So when you think about our top five, as we articulated as being our CSPs, our hyperscalers, and they have right now, [ph] as I said (49:15) about 50% of our total revenue. There's a big organization, therefore, of diversity of all different other types of companies that we are working with. That it goes through our AI model makers, that goes through our enterprises, that goes to supercomputing, it goes to our sovereigns. There's a lot of other different facts on there. But you are correct, it's a very fast growing area as well. + +We have a strong position in terms of all of our different cloud providers on our platform, and now we also have a extreme diversity of different customers that we are seeing all the way across the world. And this will really benefit, seeing that diversity and being able to serve all of those parts. Let me see if Jensen wants to add a bit more + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/e7621c2778ec89c26d8fcc05400825ee6df08de97ac5cc0a559f45ed19ab6b42.jpg) + +Yeah, this is one of the advantages that we have with our ecosystem. I'll build on top of CUDA. We have -- we're the only accelerated computing platform that is in every cloud, that's available through every single computer maker, available at the edge and, we're now cultivating telecommunications. Obviously, the future radios will all be AI driven radios and the future wireless network would also be a computing platform. That is a foregone conclusion. But somebody has to go and invent the technologies to make that possible. And we created a platform called Aerial to go do that. + +We are in just about every single robot, every single self-driving car. Our ability – CUDA's ability to have the benefit of the performance of specialized processors, one the one hand, with the Tensor cores inside our GPUs, on the other hand, the flexibility of CUDA allows us to solve language problems, computer vision problems, robotics problems, to biology problems, physics problems, and just about all kinds of AI and all kinds of computation algorithms. And so the diversity of our customer base is one of the greatest strengths that we have. + +The second thing, of course, is without our own ecosystem, even if our processor was programmable, if we didn't cultivate our ecosystem and talking about some of the things that we're doing today, investing in our future ecosystem and continuing to enhance our ecosystem, without our ecosystem, it's hard for us to grow beyond what design wins we capture for somebody else's ecosystem. And so we could grow and expand our ecosystem very naturally because of our – the platform that we created. + +And then lastly, one of the things that's really important is the partnerships that we have with OpenAI and Anthropic, with xAI, with Meta now makes – and of course, just about every single open source in the world. There's 1.5 million AI models on Hugging Face, all of it runs on NVIDIA CUDA. And so an open source in totality probably represents the largest – the second largest model in the world, OpenAI is the largest, second largest, probably all the collection of all the open sources. + +And so NVIDIA's ability to run all of that makes our platform super fungible, super easy to use, and really safe to invest into. And so that creates the diversity of customers and the diversity of the platforms and available in every single country and – because we support the whole world's ecosystem. + +Operator: Your next question comes from Aaron Rakers with Wells Fargo. Your line is open. + +## Aaron Rakers + +Analyst, Wells Fargo Securities LLC + +![](images/d24152dd54faca566c2ef0a130eeb5de5bc41b3f6965525c09d59be2aa165d86.jpg) + +Yeah. Thanks for taking the question. I guess, sticking with the idea as a platform and extreme co-design, some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis. + +So, I guess, Jensen, I'm curious of what's the importance that Vera plays in this architecture evolution as we move forward? Is this being driven more by the proliferation or the heterogeneity of inference workloads? I'm just curious of how you see that evolving for NVIDIA, particularly on a standalone CPU basis? Thank you. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/45623a6211da3738a0566694c430c25d853db1c34887904939718d894d285695.jpg) + +Yeah, thanks. And I'll tell you some more about it at GTC. But at the highest level, we made fundamentally different architecture decisions about our CPUs compared to the rest of the world's CPUs. + +It's the only data center CPU that supports LPDDR5. It is designed to be focused on very high data processing capabilities. And the reason for that is because most of the computing problems that we're interested in are data-driven, artificial intelligence being one. + +And the single threaded performance and this ratio with bandwidth is just off the charts. And we made those architectural decisions because in the entire phase – the different phases of AI from data processing before you even do training, you have to do data processing. So, you have data processing, pre-training, and in post-training now the AI's are learning how to use tools and the usage of tools, many of those tools run in CPU-only environments or they run in CPU or GPU accelerated environments. And Vera was designed to be an excellent CPU for post-training and so some of the use cases in the entire pipeline of artificial intelligence includes using a lot of CPUs. + +We love CPUs as well as GPUs. And when you accelerate the algorithms to the limit, as we have, Amdahl's law would suggest that you need really, really fast single threaded CPUs. And that's the reason why we built, Grace to be a extraordinarily great at single threaded performance and Vera is off the charts better than that. + +Operator: Your next question comes from Tim Arcuri with UBS. Your line is open. + +## Timothy Arcuri + +Analyst, UBS Securities LLC + +![](images/0e2950011ad05f4bedbe3b156ecded511542731c5a0af92932604dc95dbe590d.jpg) + +Thanks a lot. Colette, I was wondering if you can talk about the deployment of capital. I know that you really jacked up the purchase commits. But it sounds like maybe you're over the hump on this, and you're going to probably generate about \$100 billion in cash this year. + +So, and, pretty much no matter how good the results have been, the stock hasn't really gone up much. So I would think that you probably feel like this is a pretty good price to be buying back a bunch of it here. So I was wondering if you can talk about that like, question being, why not put a big stake in the ground and just, have a huge share repo here? Thanks. + +## Colette M. Kress + +Chief Financial Officer & Executive Vice President, NVIDIA Corp. + +![](images/642f640c7700871d13d763dd05c7fc8af1ab6b594ddd6a475b7883536c1c8b57.jpg) + +So, thanks for the question. We look at our capital return, very, very carefully. And we do believe that one of the most important things that we can do is really supporting the extreme ecosystem that's in front of us. That stems from everywhere, from our suppliers and the work that we need to do to assure that we can have the supply that's needed and help them from a capacity, all the way that we are in terms of the early developers of the AI solutions, that will be on our platform. + +So we will continue to make this a very important part of our process and strategic investments. But of course, we are still repurchasing our stock. We are still with our dividend as well. And we will continue to find the right unique opportunities within the year for doing those different purchases. + +Operator: Your final question comes from Jim Schneider with Goldman Sachs. Your line is open. + +## James Edward Schneider + +Analyst, Goldman Sachs & Co. LLC + +![](images/abcaeccd9b0528001b1eafb35ebf57c2d8412def6b60e06c246e39edac1553e9.jpg) + +Thank you for taking my question. Jensen, you've previously outlined the potential to get to \$3 trillion to \$4 trillion of data center CapEx by 2030, which implies a potential acceleration in growth rates, which you sort of guided to this, at least this next quarter. + +The question is, what are some of the key application areas that you believe are most likely to drive that inflection? Is that physical AI, agentic or something else? And do you still feel good about that \$3 trillion to \$4 trillion envelope? Thank you. + +## Jen Hsun Huang + +Co-Founder, President, Chief Executive Officer & Director, NVIDIA Corp. + +![](images/94b7e94a356eb1a7e021424be7fbbdb22428bd277769b712737bfbddf32f980c.jpg) + +Yeah. Let's back that up and just reason through it from a few different ways. So the first way is on first principles. The way that software is done in the future using AI is token driven. And I think everybody talks about tokenomics and, talks about data centers generating tokens and inference is about generating tokens. And, we generate + +tokens. We're just talking about tokens. How NVIDIA's NVLink 72 enabled us to generate tokens at 50 times better performance per unit energy than the previous generation. And so token generation is at the center of almost everything that relates to software in the future and relates to computing. + +If you look at the way we use computing in the past, however, the amount of computation demand for software in the past is a tiny fraction of what is necessary in the future. And AI is here. AI is not going to go back. AI is only going to – only get better from here. And so if you think about it and you said, okay, well, the world was investing about \$300 billion to \$400 billion a year in classical computing. And now AI is here and the amount of computation necessary is a thousand times higher than the way we used to do computing, the computing demand is just a lot higher. + +And so if we continue to believe there's value in it, and we'll talk about that in a second, then the world will invest to produce that token. And so the amount of token generation capability that the world needs is a lot more than \$700 billion. + +And I'm fairly confident that we're going to continue to generate tokens. We're going to continue to invest in compute capacity from this point out. And fundamentally, because every single company depends on software, every software will depend on AI, and so every company will produce tokens. And that's the reason why I call them AI factories. + +And whether you're a company in the cloud data centers, you have AI factories to generate tokens for your revenues. If you're a enterprise software company, you're going to generate tokens for the agentic systems that are on top of your tools. If you are a robotics factory and self-driving car's first indication of that, you have huge supercomputers, which are basically AI factories to generate tokens that goes into your cars, that becomes its AI. And then you also have to put computers inside the cars to continuously generate tokens. And so, we're fairly sure now that this is the future of computing. + +Now, why is it so certain that this is the future of computing? And the reason for that is because the way we used to do software was pre-recorded. Everything was captured a priori. We pre-compile the software, we pre-write the content, we pre-record the videos. But now everything is generative in real-time. And when it's generated in real-time, it can take into context of the person, the situation, the query, and the intentions could all be taken into consideration to generate the outcome of this new software we call AI, agentic AI. + +And so the amount of computation necessary is far, far greater than prerecorded, just as a computer has a lot more computation capability than a DVD recorder, a DVD player that was pre-recorded, artificial intelligence needs a lot more computing capability than the way we used to do software in the past. + +Now, the question about computation, about sustainability at the first level is just at the computer science level, this is the way computing is going to be done. Now, from an industrial level, because all of our companies, in the final analysis are powered by software and the cloud companies are powered by software. And if the new software requires tokens to be generated and the tokens are monetized, then it stands to reason that their data center build-out directly drives their revenues. And so compute drives revenues. And I think they all understand that. I think people are increasingly starting to understand that as well. + +And then lastly, the benefits that AI produces for the world ultimately has to generate revenues. And we're seeing right in front, right being developed as we see – as we stand here, agentic AI has turned an inflection point, and it literally happened in the last couple of two, three months. Of course, inside the industry, we've been seeing it for a while, probably six months or so, but the world is now awaken to the agentic AI inflection. The agents are super smart. They're solving real problems. Coding is obviously supported by agentic systems now. And all of our coders here at NVIDIA are using agentic systems, either Claude Code or OpenAI Codex enormously to – and oftentimes both and Cursor, oftentimes all three, depends on the use case. But they have agents and co-design partners, engineering partners to help them solve problems. + +And you could see their revenues skyrocketing. These companies, in the case of Anthropic, I think their revenues 10xed in a year. And they are severely capacity constrained because demand is just incredible. And the token demand is incredible. The token generation rate is growing exponentially. And the same thing with, of course, OpenAI, their demand is incredible. And so the more compute that they can stand online, bring online, the faster their revenues will grow. And that goes back to the comment that I was saying that inference is revenues, that compute equals revenues now in this new world. + +And in a lot of ways, that's the reason why we say it's a new industrial revolution. There are new factories, new infrastructure being built and this new way of doing computing is not going to go back. And so to the extent that we believe that producing tokens is going to be the future of computing, which I believe, and I think largely the industry believes, then we're going to be building out this capacity from this point forward and continue to expand from here. + +Now the thing that is – the wave that we're seeing now is the agentic AI inflection and the next inflection beyond that is physical AI, where we take AI and these agentic systems into the physical applications such as manufacturing, such as robotics. And so that's a giant opportunity ahead. Okay. + +Operator: This concludes the question-and-answer session. I'll turn the call to Toshiya Hari. + +## Toshiya Hari + +Vice President, Investor Relations & Strategic Finance, NVIDIA Corp. + +In closing, please note, Jensen will be participating in a fireside chat at the Morgan Stanley TMT Conference in San Francisco on March 4. He'll also be giving a keynote at GTC in San Jose on March 16. Our earnings call to discuss the results of our first quarter of fiscal 2027 is scheduled for May 20. Thank you for joining us today. Operator, please go ahead and close the call. + +Operator: Thank you. This concludes today's conference call. You may now disconnect. + +## Disclaimer + +The information herein is based on sources we believe to be reliable but is not guaranteed by us and does not purport to be a complete or error-free statement or summary of the available data. As such, we do not warrant, endorse or guarantee the completeness, accuracy, integrity, or timeliness of the information. You must evaluate, and bear all risks associated with, the use of any information provided hereunder, including any reliance on the accuracy, completeness, safety or usefulness of such information. This information is not intended to be used as the primary basis of investment decisions. It should not be construed as advice designed to meet the particular investment needs of any investor. This report is published solely for information purposes, and is not to be construed as financial or other advice or as an offer to sell or the solicitation of an offer to buy any security in any state where such an offer or solicitation would be illegal. Any information expressed herein on this date is subject to change without notice. Any opinions or assertions contained in this information do not represent the opinions or beliefs of FactSet CallStreet, LLC. FactSet CallStreet, LLC, or one or more of its employees, including the writer of this report, may have a position in any of the securities discussed herein. + +THE INFORMATION PROVIDED TO YOU HEREUNDER IS PROVIDED "AS IS," AND TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, FactSet CallStreet, LLC AND ITS LICENSORS, BUSINESS ASSOCIATES AND SUPPLIERS DISCLAIM ALL WARRANTIES WITH RESPECT TO THE SAME, EXPRESS, IMPLIED AND STATUTORY, INCLUDING WITHOUT LIMITATION ANY IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, ACCURACY, COMPLETENESS, AND NON-INFRINGEMENT. TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, NEITHER FACTSET CALLSTREET, LLC NOR ITS OFFICERS, MEMBERS, DIRECTORS, PARTNERS, AFFILIATES, BUSINESS ASSOCIATES, LICENSORS OR SUPPLIERS WILL BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL OR PUNITIVE DAMAGES, INCLUDING WITHOUT LIMITATION DAMAGES FOR LOST PROFITS OR REVENUES, GOODWILL, WORK STOPPAGE, SECURITY BREACHES, VIRUSES, COMPUTER FAILURE OR MALFUNCTION, USE, DATA OR OTHER INTANGIBLE LOSSES OR COMMERCIAL DAMAGES, EVEN IF ANY OF SUCH PARTIES IS ADVISED OF THE POSSIBILITY OF SUCH LOSSES, ARISING UNDER OR IN CONNECTION WITH THE INFORMATION PROVIDED HEREIN OR ANY OTHER SUBJECT MATTER HEREOF. + +The contents and appearance of this report are Copyrighted FactSet CallStreet, LLC 2026 CallStreet and FactSet CallStreet, LLC are trademarks and service marks of FactSet CallStreet, LLC. All other trademarks mentioned are trademarks of their respective companies. All rights reserved. \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-1-We will no.jpg b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-1-We will no.jpg new file mode 100644 index 0000000000000000000000000000000000000000..345570fefc2750d212dbb9ca2d31a00afa50db10 GIT binary patch literal 1369 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<5A1R0qH8UG()kO#Vxl@SaWpn#EynT3^&or9B$8>nEb z00R>vGcywlGb<|#3s7|}P@aKBkX1<0(2-3zFp*uUP{gQl;zAB(r;P_igD!qhF-|IK z;^Yz&myncFRa4i{)G{$OGqmaka3YSZQ|TeofBv2)j3M&~ka)>xhT)6Qdr?PR-2hpUWi(FzVCJ$9Vg1iRy z8F3zKBFkrRk0JbZi-Cuk5g2*Qf(-Ty-+yVW*_$8kt9DtL*Jh^R`mc6Z81p{;yLQO- zyRlBf7xtCE_?B&0{&DWAIgZ&n^WHf9XXq^sbgo@{e~#F+w5K!sxorQkmTcR{x6Os^ zRi#SZL%YC#CX7NASGjAx=`6mQdj5;(>f}edRzI6KeS+keybR|bUm8C7kX80N@pC=D z+RuMpR~`1@?zvfQDfw1+N)rsiU)ir^DE|^J`S88K9;2mmA5ZJ-2r08!_{Hx2soho# z+GoG+es;w)a>=`+g+TawmI$i@>z>NTdem|Fi`l!&$j2-JNC<4HI>duN!kT}c_RPR zL2kaQV`|CtZG|yzZZ}2b7rxV~V$fK%U8;BPnwPvs54kJ%tVr&&+pBd#u|dm`eW3{B z*&k;QTwZ&5=XCoC_K!pUh_&3gyFa1J-l+c2f{&%I?qq4^^-gUpxc^>m|1$05LuT{* zrd{4}*zc$OyFT{Uvlw1z&-v)qbG@+pbm^IkJyv~d_CL~Iyr{wEKZEo2c&}oahLbtR zRoz4F7VVCX^*i;|vFVwa`j6r^`Nn6POZ|DG?(du>_S@$8N@lJ5fiX}1iCtAx&+*py z>Mnm>gmLR5S>E+(*+=*Eas(_?wRT{>xAV4c(r@q2N^Jqb6SvGg#lK?t^7hhn3$Yx# zHSAy4THCYxJTZOpTIxSTnVtXh3w2ur7IG)Y~NNX#Svtz#+@T^RwRJ^|4jhLvpj78 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-10-Hi, guys. .jpg b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-10-Hi, guys. .jpg new file mode 100644 index 0000000000000000000000000000000000000000..9ced9f56d7e306ff2baf79bf659aed1207bbb388 GIT binary patch literal 1203 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<KW=16jCP7AKLB{__803NOWMu>c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyF8z)&(EE)icqPb%~LYfX#8O8{e3J zUSJP*Es#=MUYNP&$aZ0`!2W5k-uM}R{eHCk3qzo*2p9`7sLWxEtKz(FK4a#|3``a;~!!1&Fjm2 zO()CARCa!qt@o)5VY~rxiBV<{M_jmtOVF9xR))fP`g=Za<#-$NXTv)G-^c$mtnGTS z$2RkuuNbQ`zrttfk1u~E?cjVm?b|0$uW6IQbQ_GLltZsGFnAX*ys+8-Z!1Ijr(>I) zK3bfznwt0~`|P4hlimcw-`_qx_}x%YZy5HUVe8iWKVKLAXQ=zn@ayW_vg<}?(s%Jc zvkyx>mce{Q@S4Zf3r{1M_*X0slW)zhySk(!;^_lx`v-X0&9~GaUbxl^}se)X@ukU{={%A10Z_fFC=WNB{j$=oE&D+2H^SV=; zOStVfW_(bTD3?qA>;Io2nrX`lhBf^{k$*S4O=s2@e=q*w^{=$8FFANPlodX>G8EW} z@BOW|ZsW!y#r-;etLygEnclxXGfV93@~p{DOPZyh@^25Q*9l+v@_T1>BI^kS@tBD( zs{id=z#V$zZvF8m_rG4;KUXAxJ!^kSt!4%&kt}Mk%q{=tl>6_+$~X3pPuU;;IZI;! H|Nol+IpXhE literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-11-It's too e.jpg b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-11-It's too e.jpg new file mode 100644 index 0000000000000000000000000000000000000000..c11422ea2b500e5e895531ea2a1475dae57db46d GIT binary patch literal 1370 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<5A1R0qH8UG()kO#Vxl@SaWpn#EynT3^&or9B$8>nEb z00R>vGcywlGb<|#3s7|}P@aKBkX1<0(2-3zFp*uUP{gQl;zAB(r;P_igD!qhF-|IK z;^Yz&myncFRa4i{)G{$OGqmaka3YSZQ|TeofBv2)j3M&~ka)>xhT)6Qdr?PR-2hpUWi(FzVCJ$9Vg1iRy z8F3zKBFkrRk0JbZi-Cuk5g2*Qf(-Ty-+yVWSeKWyd3)vaf{hDKnlS1*nm%2-|JUn| z#ox-AcJu}x*$Lv+b4JBy@=KADZ%Gn^4x`*T~KV$a%A7Jj?bU-RmaZIF8|cJ`j}9Sb)~vek25k85>L9x$RA#~fZuD;#xK5A zx}iRKEah58!8YbQ)^;&yw|)#;lDGcfsh*k@d>mi%{xeMXKexMbiNHaX<%=|ve{g-4 z5L{o%aq*xMeK^-K@xb)&8dBlURP7Jh)-Bsrf->L%*OAls_TAQ~KS zdF@y2WS$rN-y=$cUuoSHVSE;Rv~WpQs!>=*yw>6U3x7#{v9m_e9s%||Yvq9s2yYkOj^AAp*wr%Hj^_GMOwN?Kav>5JOE&khI-T(8!e}-S2>)2h6 zf4e)c{`rda?z0R{gj}kurrLL`H*CChI-vKCz=c!X`~{T@_yZSBjErU0JhkD3@{Xko zn7T9;@D(46_;;rH`Z@VuoU7!x&)?SDYqRmsD}SjNrf=I0uV8cg=loM@_Q7dSbKkAw zcz8qMkJWz$D~4Cv?H}W2)?{5saQplW%W-kLjmH*Gy4TNn6Q|4R8Et-aejOsx%{cAnPy zxGiwE{Iv!%8_)7p(OKH2+vkezQjoqrRek08{PUNl`+cnXnX~=l!>gYB62I@=c#~}w zFZ!xTG2o(F^{V>5Y_nU(J_kXnAKUi*4gUMnso-JzpQ)ifA!5*)xSp;*BK?}Nz{oj zRslnaY2mC(w;unk-4j&5ui>IAgUhV2dF|TCdHn0AX^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<KW=16jCP7AKLB{__803NOWMu>c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyFAHPfXN%DH~97S|)gP@5kQfGfR1A z%~O`X{i>>N{<#LLC-WqubBY$v+Y!dLWWD93=aW6ID;qP-j!tR#CI5NWmn;U2MO|Q= zz3W)dyMk4m>KDT2-(_slr{bY{{1_x%61)oU@l%>cV+;jSIg-ky2# z?~X0)-^)-Kr?cm`+v+DPPdV(af8aj(FZ;vTxX#RPx=Qzy433tYU-|r>;o5q;KJB?~ zA6qg~=Q%|*-ru>F|4ncCzluBc_s{>^%TUFvQL1%V%Z=mfdXt$cmrC2XzrTHY>iG`# z{xhNbkrW2+z5M-0-IBNL|_mw$Jeql$0; zx(sITgWGEE#BKP`uq^Y>`7i6Xt$6XRKJn7p2g*F+#s3-hRGL;baGkxr=HK1L=OrXd zr7yp-|8-NoH#5*duKeNc`x~b?{0s5`Rmat)yi|Lm-%L{@|OuXk$iV2*Ggsf jiLu9K*M94@;!Hc`70xFiP-J1kSk^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrt9G^ivC3T><5MA{)wiaWl_63^LuCTP zE8legN4$E>ZF}78I=Jh#9tx>KuYMc4ELGJjo{1f%P z3?-hAYE^H`XKKGYasOeI+4r*VnfvagOs#&K^Ju51jkrbp-8i?iFHVI2$(DL{I^g0u z(K$A{e~c6#{N?x;!uT>I+g0;684jfq2z}5BamiF%QrV@VX9)YYV4dydnr8P_DdM>=<=p>+vRg z-zEM1TS`u><1?Sw&b*m#|KlIm_ zDb0yqo1G>tJZ04u{3i1YL!g^m;k;Un1$@QFTci&(Jq(;e!oz*GsB74gjr8 zEs=Y6@2dZuptoOb?WeyEw~@|C-ZbyQe}>cHTz1|68D9U@U|zb;`sl1+mHW2@R%-n5 z?mNEmPyP9*H<|0Ym5i*~!Xy;>I2Q2PAFlcpt*~iN?Z)Tdi{!fB)UMtb#= z*ArRugeJzEUM_H=^3#k@cJ2EgN37Ocy;Nx7y4+(`GL8R4823Jk=JkvISf1B&(XBD$ z$bEx%epdfN!~H*|p50_po{@gQI(jkli(2>Bk!#mxZSI|)J>&VNf_kq>mmRLWzh6G- zVpCDd?59A-&SF@zx^zueg}lvYGtW2nzYgt{I(N-l+hXma-y7H2?OV1(BK3-mWnHmE zxH5~usU6?!AI8>ATz~u3^3x$1&$)Ljl$7{p!noE|B!Cf!nHbm|San4LSj^2f#2!l8 i(V60Wvo-M2m+n}jXAf?_|GDk&<2n3}t|E;8Zvp_Edp&~y literal 0 HcmV?d00001 diff --git "a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-14-Every \342\200\223 th.jpg" "b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-14-Every \342\200\223 th.jpg" new file mode 100644 index 0000000000000000000000000000000000000000..c9908f46e57d5730467c1d8809b76d357ea2f231 GIT binary patch literal 1354 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrtE9HA6q}lnT0ir=;PUVNKLS{HF7LW$ zta3>u!)mh(*YWvBXQjAvTsr&jQGK8Ggjv;p^L{Yz%8Y)u=u6<+7urSF5B8msdi6r_ z-;-pEKfVlc{t_?W+{jobQZ<$R*R`ed)+W~<-Fs+reZhzO6W2uAi06fW4=-9>zhy!H z%Lu7yoz+V=Og#4D^O22T*zaGsz`i&zW_|k~w@W9sm7kxMr#$7mUHhNxRSTH*ygy?8 zwtm&=<{69BOyvERe>A-xd-_SFzxkify|X__t~a~9@VDFYT)`(ctN3b-&oY$D$z8ha zclqMY`b(;eb0(hnBD&sHgdsSkSFLmLL}z6lqj@dCD;DsZT{HQ-^p!u?&%5Qr1pP_$;|8G6}fcm8j zC02{3UCI|JHL$aPKRJf2>tN|>qpHeP_Euk}1-uB|$&zq=hPTYDvPCEAuQ{sA{jiZ@ z_{ycFU3ux*Ez8H2o+1W+vZ5F?R@|xx3hjLAI&o*H4x_zePyoC6V>4#o#_E}yf60Gb z6~`Tvw{G?Z2mAPcTaAyICq>`L4c(*|v(483KSKonxh*yFSxJ|d?yya(;a~Kh;p@o- ze0@9PMPKr&|M5}qnP6xjH+R;qRU9V{zI2c=Z{Oo!^8QQooY_f5Yb-VD5ABXfZ+m8_ zVzi{Adga;TUCUb@&pmUEXQSbhWA{t`+G@|MRSw@&^n}J-7e> literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-15-Analyst, N.jpg b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-15-Analyst, N.jpg new file mode 100644 index 0000000000000000000000000000000000000000..15506b4a877e31fae97e12befacb111724af1918 GIT binary patch literal 1200 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<KW=16jCP7AKLB{__803NOWMu>c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyF96CLI@gH}R^t#c7$~*F7J5pU*7k zoi$Hc`u3};y7}iCtbV*xoVx9}NnYU`ZRVixIWH%lT>N_C%tqg}n;dHSf1a%}We^E; z1!L2@$-Q&Wi8`wM;`;nsG&lXoYUQ5Qr&1ECHJ-~a5WW0k+2=RAoZXi5Ti&pn`=8

BxYv-S8u+9ScOk+jf zI@gxYAWG zi*LP}d|X7xq^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrEA;ghH*T%&yq%^0%zwf61)Q(u{94<( z^T_rID`mGg)hovMEnj~xt4PExW1$Vlgs;DTU0z$Gzw*w4VkxKP%V!nTD^7Nht;*gN z)7w}rVE;i>n<{WcKhV@^}K(1 z8LDDM+ZR?en%dv-6SJOR!Mymd?cdc!4O|cY1g?}m>v4PK->W;d`@`PVFD(~7rMP`s z@I`^`PqfdhO*Ws^Vz^V~v<6RYQRM=@{fDiLqN7#cyx{r7v#h?c{>XJ#4~9?%9>xPq z(;B$eSEsW6-RxK}d&(EvgI`6Hx)?rd@6P-(*K_e5mx%@`g?CEX_Hz5|lj@txdxAks z-axkF!FB7$-UU+Ep78l4C;#ey9aN!Ll(1{NdTVNd-5&WT8qDRPCokz2{%5d}|IgrP zdf>$jsk`&)pRZW&KFv@^U%I4Jt6u7}#OWKaIL;Y-@ZywvRsZV><13Tkz-u1YTr9&F znqTo+HE?MxYOp*nDfo9L!~5s_e_5{bvo?P_z4-QN`E&ChTTe*cwv~M<>+O#Z>XpN6 zB=drLjM`Ej<3)?yX(vZw+8l}e<~B~*DrNn{$uYYGw&Z~S~PFH zF!21M{OelIJhiJ!V)tj>vcKuT|7q2?U11AXT4pRL<0(qXV}F*xILY?h#+B1<@2b4t zU|s*UVZFzT{4%c>UK78aes zGvh{){jmm<6^ptUG$*;rtls8v_4s0*i>?fYI&T+SdRK(}YGGOilw4cA{MIYM%5p>P kzT-Q7NuBucZ$9s|g)2C|3$+G%%N3sd&%mXzfdBtZ06|zmQvd(} literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-17-Analyst, E.jpg b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-17-Analyst, E.jpg new file mode 100644 index 0000000000000000000000000000000000000000..8e9be56cffcd8aea2e7d40d81cc0a12ba23ce00f GIT binary patch literal 1227 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<KW=16jCP7AKLB{__803NOWMu>c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyF8z)} z{&@l4tgeS!Z1dihRxIAVbK>I(KU1vJS-#lMEjM8ZbQJ;P@OfvB$9&Xd*Hqc<{=4}< z!y4n*`VM`&ql?cxczCBW^Q-;eoa^=vUi*HZ9Bc5%^w$31{|wjU*D{2Dbe(_GwBMBN z%1q~;GUeuf_Pu3Ru@ClNPS|n&%;~oO3>5|SQD;8p&c9`{(5hz-@9uw0750}ut&-#2 zd)Pna`1<$4KjmLK$d-x(I+_k_qKoEFQ@wNlz5KuJ zpVz+Uy;VEewOyHC|Go6bm%oyBSf8G@XV0vMzfL^e`0{O(_M_TB5v?GwbsaQB=qSyR|>n*lU^Hq;<&)>_^8pHQ^;a}sh{|pZ`ukL@otN#AR$m!umaWQ*7 z3#1fJQ%GvC^^CA;3!bX1(D!9)_rvb{TVHbY6sxeGw)eZgT)s6ofH9&@YtL`92}!~2 zbKd;1n*ZTXcF_)f(YV^>;f9*09L(x${eRSnFs2{;mK1$5Z)VX(y{qTH_y1>z{&_up zWBo3Bt`!%?ErzT$qr|7FqD4xC_#V;@aBc3yl^61Y$J84usx{|x?5qeFUD zy$Ed!ZR7YV!g#16?E9Nr>n9!OEHL}eaA0?R;E(H-af#(Dv+|a!I869(^3(h)@=xRX zQ%kmQ3%Il1Q$(gp?(ydZ+@jBRE)C|(ivJk`<$o{!8T^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<5A1R0qH8UG()kO#Vxl@SaWpn#EynT3^&or9B$8>nEb z00R>vGcywlGb<|#3s7|}P@aKBkX1<0(2-3zFp*uUP{gQl;zAB(r;P_igD!qhF-|IK z;^Yz&myncFRa4i{)G{$OGqmaka3YSZQ|TeofBv2)j3M&~ka)>xhT)6Qdr?PR-2hpUWi(FzVCJ$9Vg1iRy z8F3zKBFkrRk0JbZi-Cuk5g2*Qf(-Ty-+yVWidwduZ+b@B^l3Iq<;f0WPo`NU8+pCz zu-Yuc7%BW?v3JV@pGy<}GuT}C&kz$XTU=hd_MS%WC2RRfCNJ!oe_cDezB#>=x3p-Z z@E;9x`Ew1X!bfidKZYYu>m2GCkx`zx%>J_gT-Me2B{Zd-XNje}+sW z`?(4CUy9#y>H4I3d;YYD^OC>5nJ~T-2{(NBpFwcz;k9!=^Ba7WnSbw_-Pg+(7~?7* z={m088p&Y4t7XQ8TC4d-u2+}pPfWMD|M{F*{)x0b*ZteyT20;Jd!jN#=2!YGhPQvL zm#WQA*tDyEix0D*@>KRk*UpLrF#61rc&C`L)Zycf!nIxCL`iJEK?B<1K)myF>bi}<(rCgso=Yd&*f(<;C(NY@vHUe^IH$N zm6TdTg4-mX|Gbvl-kHDUYehO+!h!}Bd!za5zpP60O?r}}U#nn$c)I=4sLda?EiQ4K z%DSuSPrxXiA7w>q|Ti_oQ0x96UBu$~;(pPMhcbW7Ps9+m7Z;S9mSAD+H@z?I&6 zdG V{d9YW!l}L`AJ4B931I(!696noI0yg$ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-19-Yeah. Than.jpg b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-19-Yeah. Than.jpg new file mode 100644 index 0000000000000000000000000000000000000000..520f3ea735b8a48a92b9ee9e0e0c6b1e9f3fbb94 GIT binary patch literal 1183 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyF96CLI@gH}R^t#c7$~*F7J5pU*7k zoi$Hc`u3};y7}iCtbV*xoVx9}NnYU`ZRVixIWH%lT>N_C%tqg}n;dHSf1a%}We^E; z1!L2@$-Q&Wi8`wM;`;nsG&lXoYUQ5Qr&1ECHJ-~a5WW0k+2=RAoZXi5Ti&pn`=8

BxYv-S8u+9ScOk+jf zI@gxYAWG ri*LP}d|X7xq^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyFArmsjT|OH6teD7Enx=ew+rhk z&84Q4tvu0M$^85He}=VP46oLfW=r4n3A@g%+&FiE;PJ=d+2<58OKb~m&hO=J`0;M} z`WM+38+jPn6+Uq+U|rOu!2rZeme0H^RA!zNGS#VT+F>hYeznGJ(+T6xGv|MJ{_{b& zmA#`&?QP!n|KND?pTTSWgWJB_KHuH&^mfu;cJ-hAry8_=oI0O+Mm)=6Vww4s zlE3y-b1Txmznx{1b==7L_qJbr^Ot{GS>u@g?W|M@vtIL=_!a*dUZ?G>=e+lKc8G!D zA-~=A51c3db${$2dsigTfmI|l*Xc#X4gaG0r?V#6o=^OFfnD%X_4%!@g|y}@Q9dmn zc>U|Y&}*4NkvB7^ef!Ul)nK&X?(c)&`_;d$6FYY5ap>MpNeU+pY)rSQ_09Uva7Z3# z-Fff)-*5giv@(?0=Wn+E^qu|hAJOOG9Xh3vtb4j#g7SK&JV{=$*8hj|SNWeWbN@5c z&i}9~YFcyU(gU0Ql2g)dPf>aBFNE>_%{yzK&N1iwo)U25%eR%UXM8AqKQmi*wp_CH z<-eDKt}4BJfiY%(&w8_ith46D{OtcCYajD3^h&tk+RUnT3s;@Bx@FmK9X#>7`{N(~ zOc?7AZmUk1_px2CX6r9L|JVJec6F^@9>pzd#hrLnfbG{;`*m4Y8AQFqH~+4iZo`n; z_0;~JOxJE#hRSW-)!%rdBqJVj-uTbp42=8u59>YGNr!&;`ti#1?C7KqyYqrL-k+48 cTUwPfoqM^%_5hE~tKKFae0i+u0{j1)02G4+h5!Hn literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-20-So, thanks.jpg b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-20-So, thanks.jpg new file mode 100644 index 0000000000000000000000000000000000000000..3baf85e0ddb1fe6870cea3ad10731af0b018de7f GIT binary patch literal 1378 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<5A1R0qH8UG()kO#Vxl@SaWpn#EynT3^&or9B$8>nEb z00R>vGcywlGb<|#3s7|}P@aKBkX1<0(2-3zFp*uUP{gQl;zAB(r;P_igD!qhF-|IK z;^Yz&myncFRa4i{)G{$OGqmaka3YSZQ|TeofBv2)j3M&~ka)>xhT)6Qdr?PR-2hpUWi(FzVCJ$9Vg1iRy z8F3zKBFkrRk0JbZi-Cuk5g2*Qf(-Ty-+yVW$a}YN)3)l)+c_DQ-@l14nug5}*u2eR z)+-Uap2zl63#I0Fyf@|KayiQ)%cS_^udmtl!%=fDwF`7i`C>GuZf*wi7T4{3rVL*Q}SBqE0J9B^VMI zZ%u39T3egQwm9d4-%R$!<-xDC&WbQT3pa?#l$zOJ7@3%}n(trRUT)vLvCjT&Pu;lt zXH*Fu6g^jQ=?44Ooy?PXa_Z*vKg)_|Jr{NF)tz~Q&5d6l{AbW&xO=tu@5JiF<;rf3Dskg5zAM`yFZ%g5q z?lGG8h4E}fdTh_3Y2tpO*E{C+v(%Trj`c0LwY+uR+L)7RQjg?z$k*2M{by(m+J4>C zTkxsw;fEDB74G)uRf;ffeH6LX!<2R29e>$984M-%nznxRzy3{`Ro;D~chW!GGmKgv z`}fpx{M$NHy{j}*WW`ir+pFwfSB32={kBxmz~TMmNk0z#tJPqb7u_2ltiNT;WiwIt ze8(@MFRDr>WquQkRy=p`^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<KW=16jCP7AKLB{__803NOWMu>c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyFAPs%Pj1*nC#;@?$JxNL#v2`%fB+t3{uS1$i%eJ@Mz^%-(M&>~GrE zJJ0%3P-LXRu&4`+87H0CaOBg@LrT+%WT0TGR zQq5G0%G>#WSc>Ww{K=Z$tZOKH^fBKOx8+~$FMnFVwa;+#r}s&_1COlUx^MA+hP8iI zS0A3fKlkDzonn>v-T8-?zwC+^y!m&t>kP)}-QT-^l>Rl2dESy=)9bl*#-GXS<=^wU zznsPJPGeCQgT|r-dv``#Yp#=Ck`XPpe8U1h<)h~Fv##DUwe(;+`1Soi`PR~3w!3a- zhJ6rXW<6be=TrUM%CB3uo?Bj#+>t-?({XjVC)fJ_DBW8B=cE7c9~#Vib$U2-mdJ6O ztz3I0f9b1t8|Rn^J>}mX_$z7Me}+T4tNt^b7x~YyziCzci|e~`yj9O62F}hP&Tu>-hJ6&Pmyl{MqSd{CEHV4Dt4JOQtY{8$XsluXjn?)aKOf z*)Oc>_Mh67_0aTouG@cx@L8doc4S<6! zJmkx6{|zrEL_g+P{keWmrTyBN1xzcl!|L-crMK|xyLH8aA zeK_H#T>NzZh5rm+KV)Bgeg12%&^z6r!u)KL3DfTMba)(mF5m9;X3OoF@qP_=j%xXC Y=ab|svN-?vT7!O|s|W)UX8eB>0Mrou(f|Me literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-3-I am confi.jpg b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-3-I am confi.jpg new file mode 100644 index 0000000000000000000000000000000000000000..5acc713315d6ea047c5faf90ee244dac634bf665 GIT binary patch literal 1362 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrtE|>7=bN69HhH>@QhD+}5yrg8#ZAU4 zn{+a)Hp_4wpMNweMx9$_()?d(^B;FTQQx`m+IxrGODFCZnZDTXcxzj3jkNZP?A23y z{xcMcm%nz9@_fu|I-6BbplTEUi)_F1A)EI9XXwvvDf{={?N7gHt9sKT?cb}9^ZvWU zq|F|1aec{jeuY!%Pp26lu)lDDeeuPZ?d^ZuE}z&|etx=*yXHUT7qx$1uUx>i=kpQs zx3#NIHveAX*~@-;I@$ z$>lddLoZM7q5AK2_5t-v z8A_}cO}msYQfgpl|9)}|Ti3zT)kam73&NM|YIh5JahPq-A(zR1SG?wbo#6R%LI(2{ zjTLX7d(W+Uq-A;YQt-|1(6?ZC(7}b+Fic(}S;#AJmHfXISUU zP*QrTV!cP2{VfBv0z(7&#o^(O3{4y-Y*#S1fB4?PHqY*KiT}-$r#3uK{LjGkHffgT zV^cI&x~D|1-3Ofq|3e4|`#uiUMB*3lUcIuey*7wr&l5B~7< z*~cm88INl`+SMz3ggLaB`@l(~b^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyF8z)`cJ4HD!a1*tSo5a{CQW8M0sC zvzKCcwPGE&%hh#jzjjP|_9(6Q{It``lY73re&6*zgJDsZ1{kX|t1qp5sUGuDTd2|K`?2 zBOaAk63JHz>KFdWUUx~q?*7t)1%5H>2J&ZL@AGp1r`QX3y4{Oe| z-FrJNTg}Gcv+yem`^%pe@Ez*XSisb!IqB4fL!WjYQks_hMuWNMW9#$T^|GtxPZK>| zzo-7!jX$p!et5h5rbz^&L?!!-`~CixMGxFLtNQfGW&_nHNjGG!bzC=H8o=_2!Gifz z{D%d6|1zdk#HmZV2d%N~t=uByxg?wG`1{+Z+rBeA_?NWyKSSQ@`p+NL|NfGH7JMWC_!?b%Yu&+>F)3I*3)?tOdM{^_QiS7LzHff{b=>ui*%kw|{(avk&I`}d5 z`OPou75(JOZvSUsl$rm+Uh3XD_M_tSv#v>*T6(Y@{QCaa#~%T#=h-9Q@0#_tU2xse zU-R}a|Ge(hW*htTZ@N+}%6yN{!@rOIvHCZ2ZIv72 zl+`><2968(xDQ?5pL+G0nuX`F!(ZS2lkY9R`po>=`{h|hhd9~PPs!&k_!qI>b;IuH z+Z=h0@_Pb2zStgr?jV-6IrE#((OKv2Y_wE4veSO4NC4}+DS@sGB0%h@J9o|bebE^y b_3v#GW)($8A6pp2G1c8cfYDWi@&8Q#K^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrEA-wy+_RGaqwRCxge5r_4pay(eDc@7O1H*%&BWjfrwV_SF5o+9y4T3< z*7gn0?{56_>XpYNjf}gI{!Yu4c68dvuUPanj+1{u68|f06SKvO?mRZSee$~O>|at0 zUv_B=JAY)#YCbtbr(7YOdGYW54{LWZXg7cKTax!xIH&!)+4DTx=$-$j-tpIHyXWce z;&gw)dZTT<>rFWp7kSPKkofiFvx97YTS-XMUH#<_=@Zfn>Kkk8mhJL#4Ok`W#vs8E zbb&E+{hEX;y0f@j?kr#UZuynaSsKh|)~$Vh$<(aBFj6sRao@kTz1%)~W1aomp74mv z8+=i2j(OgzdmuY{f~CT-<6q^kE&UMlbiuCe>a7V6YAftd1hBnZJ^xMjo%-|a{~0d$ z7Fem-Z+r9kKZC7nmCeS^#ERRCR|fyMKI6#wuS}&rLm~N^1Nj7ZzKMAi~1L8D=PHub9R0Cbo}%3ALpJ(=B{;o>Ns(~9Q!BJ zrlWz5fv&8e_NeVf5^DqsV3(+p_-*QvYQCGjJ`;U-N33 z{4VcK5q)VD$>ZB?HJI~{tY39_h3?TE^Q}%x{9x>Wb{@SH9cNEeI2zF2%R#n0G=9@0vb?!Z+37*~?ysRn0h}7Z#Ip z?2h0+6UN1^A_0tnr&)q|yoEe$B^?)aG4$%>Jht?X5c$<{V6{k~gJ}Mi*RgK8(+%0b iUlk literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-6-Good after.jpg b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-6-Good after.jpg new file mode 100644 index 0000000000000000000000000000000000000000..5beba9380bd5feb5cb62908ea85e66e26cb6f3db GIT binary patch literal 1209 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyF9=ugEI9mEx&-+JNu2L*V3a?Y57h zmt$YO&Xqr-$jBgj05GmW^~qH_Wukq|E4g89}BCPbnm14 zJ-O)Gs{YqM!{)cHue-XWW5&}5*8G>Z|ELo!j^6w`Z(^Zzn5cw?I2^{@!sy{np2La zTz_xHiG&EB`apo&Rxx?LWh}Z_|PNjdfcO##aWsDmGoxxL7E!clw4`_FDJ;GqmW%|NQI! z`&<376XHaPyVD5-?88McWCRWLW+YA~&l zO@DgRy9^e7@jt`8L*L|1J9kXF^raNz{NOuRyBM_AEzQv1^73tec>MI&SL(KAbuoOl z)q3;t)*9n&4q>+*-}sPyneoJlmoau*7kU0&YklZq%Y^p~lMmWfS3TS1tM|;~%L(Nf SY!k&Z6}}#4{<47o|4jh+?gk$K literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-7-We're also.jpg b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-7-We're also.jpg new file mode 100644 index 0000000000000000000000000000000000000000..0fe27edb2db98327e8e556b5495f5fa588b2b2f4 GIT binary patch literal 1370 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhrEA-wy+_RGaqwRCxge5r_4pay(eDc@7O1H*%&BWjfrwV_SF5o+9y4T3< z*7gn0?{56_>XpYNjf}gI{!Yu4c68dvuUPanj+1{u68|f06SKvO?mRZSee$~O>|at0 zUv_B=JAY)#YCbtbr(7YOdGYW54{LWZXg7cKTax!xIH&!)+4DTx=$-$j-tpIHyXWce z;&gw)dZTT<>rFWp7kSPKkofiFvx97YTS-XMUH#<_=@Zfn>Kkk8mhJL#4Ok`W#vs8E zbb&E+{hEX;y0f@j?kr#UZuynaSsKh|)~$Vh$<(aBFj6sRao@kTz1%)~W1aomp74mv z8+=i2j(OgzdmuY{f~CT-<6q^kE&UMlbiuCe>a7V6YAftd1hBnZJ^xMjo%-|a{~0d$ z7Fem-Z+r9kKZC7nmCeS^#ERRCR|fyMKI6#wuS}&rLm~N^1Nj7ZzKMAi~1L8D=PHub9R0Cbo}%3ALpJ(=B{;o>Ns(~9Q!BJ zrlWz5fv&8e_NeVf5^DqsV3(+p_-*QvYQCGjJ`;U-N33 z{4VcK5q)VD$>ZB?HJI~{tY39_h3?TE^Q}%x{9x>Wb{@SH9cNEeI2zF2%R#n0G=9@0vb?!Z+37*~?ysRn0h}7Z#Ip z?2h0+6UN1^A_0tnr&)q|yoEe$B^?)aG4$%>Jht?X5c$<{V6{k~gJ}Mi*RgK8(+%0b k^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}nhriyFAv9=+mr;V(3DS@J;4;^fNJT_0Ju!Fe-gd;W7Fq1 zzrrUi=eN%N&(P54|3$tx_(SS4r5(EyPYX)TPdxc={#1srkG=D!SMuKKQ;l8wRsL`O z)98INdw=Kki#?vH^!Ik1ti4arkFN7?XPF$}*;BtK{_y&jv!4A?t^QW)xpu~%$?MpE zAD;L(gL&toE)9l7UF&O0wuKqTZ1=b@r)u8e&kOj}kDAZVx+Z06>A`mJ>-%3He}u;_ zaTT35HJ`DBM_BhW|DH>klf0Sme|MR*3-#;#fcRd>> zI8NlSf6Mi`ddF>_t4|_k&q%$~etX5gsn!1keZ&7%9{O0%dA8`G+M6N<-&SVN`cV6RcB<-P`B}m{>+jnCt$!Y?%OHBD;@IcZHD+0r z6W-@9c=_u;gVxK}t=DzFYX6PO3_DjWVJ+~+mi?#t#|6CnY&RcSzLzLEe%gO|@Lzka zSF4sy&(JBkq5JWa{@H&YpZwdJ*TA)K&CS2NJ?9A+{VrDr2FpMB-uOigHjkb5{^ncH z)ACT_P5r|E46h&j6MgaZz^hxLPpX#uXGq&R%_DNVrt67=&*fXoq;ij(UG9*VG*yke Rot3Z1;{4-l4fg+U0sy+x{eu7i literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-9-With respe.jpg b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/images/image-9-With respe.jpg new file mode 100644 index 0000000000000000000000000000000000000000..0893f230ed1d07700f994c6aea1c7c7b17d6c5dc GIT binary patch literal 1362 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<5A1R0qH8UG()kO#Vxl@SaWpn#EynT3^&or9B$8>nEb z00R>vGcywlGb<|#3s7|}P@aKBkX1<0(2-3zFp*uUP{gQl;zAB(r;P_igD!qhF-|IK z;^Yz&myncFRa4i{)G{$OGqmaka3YSZQ|TeofBv2)j3M&~ka)>xhT)6Qdr?PR-2hpUWi(FzVCJ$9Vg1iRy z8F3zKBFkrRk0JbZi-Cuk5g2*Qf(-Ty-+yVW(ARsoacgzw?JSKyo?qm?FxrO9581HI zV%j@Vt>!-SOhT`46L~Fod%|>h-w$O6@NHyWDeks^-TL>5%-@<^p@{!?~imZzugKVXI^dwm-1{QTE~`3|g(s3_J{) z84RnwZwts0JIi_~FZjau;8$8lMHrujmwkR|HQ9Yfr&7X+$NLMiFKjt|amC9EcN!QI zKl3|Tn4YV+bcb#0PUguoZv0#P>&iavIWzRuo_rV?!@r*WCl|xrtHpmOR`>sW@Souq z=Q?(mH<(qNwRaiaXq34H$w{xe+g4fqgOceA9rGWqA_Kj}q0+paD6v|x9g4gZs@ zR^I0@$8+`mGjQu-)vnk)srYobmCMD*Ufs`dr=<_flH*PwI)_trrp|zi|J$ z=A@1PsuR2GE{OlV!f-V^R_{MUP|<}0Z+AR7aYz4I2IHoa#cQ%sx4oN{_Ru{)e}U+W zS+%nwB97+FX_1*9eCJ>3!HBzeUPyII>K^-Gw#nk)@r~=Q`(AtV)p47cwuYqPh9~vC z3>g<)MHqrr+AP~tdu6JOnq5U0r6Qw_6-~Ra@~dED@I_aKpsPA-*S%Tt`C+)ml9jfO d-%QW_&=omb@k6?41^We==RdhL7V!VS2>=HvL&X39 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/manifest.json b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/manifest.json new file mode 100644 index 000000000..5c1e8b210 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-nvda-q4-fy26-earnings-call/manifest.json @@ -0,0 +1,92 @@ +{ + "version": "2.0", + "job_id": "job_9886373d369e", + "data_id": null, + "source_file_name": "NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf", + "processing_date": "2026-06-16T03:21:34.059667Z", + "processing": { + "page_count": 19, + "billing_status": "charged", + "cost": { + "micro_dollars": 28500, + "credits": 0.0285 + }, + "timing": { + "started_at": "2026-06-16T03:20:41.173722+00:00", + "completed_at": "2026-06-16T03:21:34.044762+00:00", + "duration_ms": 52871 + }, + "stages": { + "timing_ms": { + "document.profile": 5839, + "pdf.extract.standard": 34365, + "toc.detect_range_llm": 1021, + "md.detect_toc": 1049, + "heading.hierarchy_llm_call": 3562, + "heading.hierarchy_llm": 3573, + "md.predict_headings": 3859, + "md.deferred_summaries": 7460, + "md.build_dataframe": 4, + "pdf.parse_md": 12589, + "document.cleanup_unreferenced_images": 0, + "document.compress_images": 0, + "worker.parse.document": 52799 + }, + "token_usage": { + "prompt_tokens": 6334, + "completion_tokens": 508, + "total_tokens": 6842 + } + } + }, + "statistics": { + "total_chunks": 63, + "text_chunks": 38, + "image_chunks": 25, + "table_chunks": 0, + "total_pages": null + }, + "HIERARCHY": { + "Root": {}, + "NVIDIA Corp. (NVDA)": { + "CORPORATE PARTICIPANTS": {}, + "OTHER PARTICIPANTS": {}, + "C.J. Muse": {} + }, + "MANAGEMENT DISCUSSION SECTION": { + "Toshiya Hari": {}, + "Colette M. Kress": {}, + "Jen Hsun Huang": {}, + "Toshiya Hari_2": {} + }, + "QUESTION AND ANSWER SECTION": { + "Vivek Arya": {}, + "Jen Hsun Huang": {}, + "Joe Moore": {}, + "Jen Hsun Huang_2": {}, + "Harlan Sur": {}, + "Jen Hsun Huang_3": {}, + "C.J. Muse": {}, + "Jen Hsun Huang_4": {}, + "Stacy A. Rasgon": {}, + "Colette M. Kress": {}, + "Atif Malik": {}, + "Jen Hsun Huang_5": {}, + "Ben Reitzes": {}, + "Jen Hsun Huang_6": {}, + "Antoine Chkaiban": {}, + "Jen Hsun Huang_7": {}, + "Mark Lipacis": {}, + "Colette M. Kress_2": {}, + "Jen Hsun Huang_8": {}, + "Aaron Rakers": {}, + "Jen Hsun Huang_9": {}, + "Timothy Arcuri": {}, + "Colette M. Kress_3": {}, + "James Edward Schneider": {}, + "Jen Hsun Huang_10": {}, + "Toshiya Hari": {} + }, + "Disclaimer": {} + } +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/chunks.json b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/chunks.json new file mode 100644 index 000000000..f49669e63 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/chunks.json @@ -0,0 +1,1301 @@ +{ + "chunks": [ + { + "chunk_id": "27d6fee5-9ea2-5b39-bb59-dec0c5d8310a", + "type": "text", + "content": "Transformer network\nJan Platoš, Radek Svoboda\nMay 5, 2026\nDepartment of Computer Science\nFaculty of Electrical Engineering and Computer Science\nVŠB - Technical University of Ostrava", + "path": "dl_10.pdf/Deep Learning Transformer network", + "metadata": { + "length": 181, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "tokens": [ + "Transformer", + "network", + "Jan", + "Plato", + "Radek", + "Svoboda", + "May", + "2026", + "Department", + "Computer", + "Science", + "Faculty", + "Electrical", + "Engineering", + "Technical", + "University", + "Ostrava" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b9402e80-2196-55ca-ae71-353fa1f39582", + "type": "text", + "content": "- Recurrent neural network works in synchronized serial way.\n- This prevent efficient computation in parallel.\n- The attention-mechanism looks at an input sequence and decides at each step which other parts of the sequence are important.\n- The attention mechanism helps the decoder to focus on the important part of the input sequence.", + "path": "dl_10.pdf/Deep Learning Transformer network/Transformer network", + "metadata": { + "length": 335, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "tokens": [ + "Recurrent", + "neural", + "network", + "works", + "synchronized", + "serial", + "This", + "prevent", + "efficient", + "computation", + "parallel", + "The", + "attention", + "mechanism", + "input", + "sequence", + "decides", + "step", + "parts", + "important", + "helps", + "decoder", + "focus", + "part" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1eaeda53-06bd-57e7-b424-15c56a5602a6", + "type": "image", + "content": "\n- Inspired by how humans read – we don't process all words equally.

0\n[images/image-1-Attention Mechanism Process.jpg]\n", + "path": "images/image-1-Attention Mechanism Process.jpg", + "metadata": { + "length": 112, + "summary": "image-1\n1. Input: sequence of words (e.g., \"The animal didn't cross ...\")\n2. Compute attention scores for every pair of words (represented by a grid matrix)\n3. Form weighted sum of word representations\n4. Output: context-aware representations", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "images/image-1-Attention Mechanism Process.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f1724798-2e92-5e82-a75c-a615cee8c88c", + "type": "image", + "content": "\n$$
0\n[images/image-2-Attention Matrix Heatmap.jpg]\n", + "path": "images/image-2-Attention Matrix Heatmap.jpg", + "metadata": { + "length": 47, + "summary": "image-2\nChart Title: Words attended to (X-axis) vs. Words attending (Y-axis)\nData Source: Attention weights between word pairs\n\nKey Data Points:\n- Highest attention values (0.4):\n - \"animal\" -> \"animal\" (Self-attention)\n - \"didn't\" -> \"didn't\" (Self-attention)\n - \"cross\" -> \"cross\" (Self-attention)\n- Notable cross-word attention (0.3):\n - \"The\" -> \"animal\"\n - \"the\" -> \"the\"\n - \"street\" -> \"street\" (Self-attention)\n- Lowest attention values (0.1):\n - Most off-diagonal interactions show minimal attention (0.1)\n\nScale Range: 0.10 to 0.40\nWords Involved: The, animal, didn't, cross, the, street", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "images/image-2-Attention Matrix Heatmap.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "6907e0d7-0e26-5f71-962b-2fff66d237f0", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
Words attended toTheanimaldidn'tcrossthestreet
The0.20.30.10.10.20.1
animal0.10.40.20.10.10.1
didn't0.10.20.40.10.10.1
words crossing0.10.10.10.40.20.1
the0.10.20.10.20.30.1
street0.10.10.20.20.10.3
", + "path": "tables/table-0 Attention Matrix.html", + "metadata": { + "length": 1223, + "summary": "table-1\nThe table shows attention weights between words in a sentence. Each row indicates how much a word attends to others, with diagonal values often highest.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "tables/table-0 Attention Matrix.html", + "keywords": [ + "attention", + "weights", + "words" + ], + "tokens": [] + } + }, + { + "chunk_id": "3e6d769c-acf4-584b-9847-90f9e015564d", + "type": "image", + "content": "\n
0\n[images/image-3-Word Alignment Matrix.jpg]\n", + "path": "images/image-3-Word Alignment Matrix.jpg", + "metadata": { + "length": 55, + "summary": "image-3\nThe image displays a diagonal alignment matrix visualizing the correspondence between English and French phrases.\n\n**English Text (Top Axis):**\n\"The agreement on the European Economic Area was signed in August 1992 \"\n\n**French Text (Left Axis):**\n\"L'accord sur la zone économique européenne a été signé en août 1992 \"\n\n**Alignment Pattern:**\n- The matrix shows a strong diagonal pattern, indicating a direct word-for-word translation structure.\n- Key alignments visible include: \"agreement\" <-> \"accord\", \"European Economic Area\" <-> \"zone économique européenne\", \"signed\" <-> \"signé\", and \"August 1992\" <-> \"août 1992\".", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "images/image-3-Word Alignment Matrix.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "6d3df7f5-0233-521d-8ada-14352cc1fe99", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
The agreement on the European Economic Area was signed in August 1992
L'
accord(not labeled)
sur(not labeled)
la(not labeled)
zone(not labeled)
économique(not labeled)
européenne(not labeled)
a(not labeled)
été(not labeled)
signé(not labeled)
en(not labeled)
août(not labeled)
1992(not labeled)
.(not labeled)
<end>(not labeled)
", + "path": "tables/table-1 EEA Agreement.html", + "metadata": { + "length": 1249, + "summary": "table-2\nThe agreement on the European Economic Area was signed in August 1992.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "tables/table-1 EEA Agreement.html", + "keywords": [ + "European Economic Area", + "agreement", + "1992" + ], + "tokens": [] + } + }, + { + "chunk_id": "423b548a-1bc4-5fea-8bc7-000e30d5e142", + "type": "image", + "content": "\n- For each input that the LSTM (Encoder) reads, the attention-mechanism takes into account several other inputs at the same time and decides which ones are important by attributing different weights to those inputs.
0\n[images/image-4-Transformer Architecture Diagram.jpg]\n", + "path": "images/image-4-Transformer Architecture Diagram.jpg", + "metadata": { + "length": 260, + "summary": "image-4\n- **Input Side (Encoder):**\n - Inputs flow into Input Embedding.\n - Combined with Positional Encoding.\n - Passes through N identical layers (labeled Nx).\n - Each layer contains:\n - Multi-Head Attention block followed by Add & Norm.\n - Feed Forward block followed by Add & Norm.\n\n- **Output Side (Decoder):**\n - Outputs (shifted right) flow into Output Embedding.\n - Combined with Positional Encoding.\n - Passes through N identical layers (labeled Nx).\n - Each layer contains:\n - Masked Multi-Head Attention block followed by Add & Norm.\n - Multi-Head Attention block followed by Add & Norm.\n - Feed Forward block followed by Add & Norm.\n\n- **Final Output:**\n - Linear layer processes the final decoder output.\n - Softmax layer produces Output Probabilities.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "images/image-4-Transformer Architecture Diagram.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "9d610904-929b-540c-b017-aef3358c325d", + "type": "text", + "content": "- Allows the model to focus on relevant parts of the input when processing each word.\n- Inspired by how humans read – we don't process all words equally.\n- For each word:\n• Computes how much attention it should pay to every other word in the sentence.\n- Creates a weighted sum of all word representations.\n\\- Output: a context-aware representation of each word in the sentence.\n\n- Inspired by how humans read – we don't process all words equally.
0\n[images/image-1-Attention Mechanism Process.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"Attention\"] --> B[\"Input: sequence of words\\nThe animal didn't cross ...\"]\nB --> C[\"Compute attention scores for every pair of words\"]\nC --> D[\"Form weighted sum of word representations\"]\nD --> E[\"Output: context-aware representations\"]\n```\n
\n• Each input word is projected into three vectors:\n- Query (Q) – What we are looking for.\n• Key (K) – What each word offers.\n- Value (V) – The content/information of each word.\n\\- For each word, compute attention weights:\n$$\n\\text { Attention } (Q, K, V) = \\text { softmax } \\left(\\frac {Q K ^ {T}}{\\sqrt {d _ {k}}}\\right) V\n$$\n\\- The result is a new representation that captures the context of the word.\nSelf-Attention Heatmap\n\n$$
0\n[images/image-2-Attention Matrix Heatmap.jpg]\n\n
\nheatmap\n\n[tables/table-0 Attention Matrix.html]\n\n
\n\n
0\n[images/image-3-Word Alignment Matrix.jpg]\n\n
\nheatmap\n\n[tables/table-1 EEA Agreement.html]\n\n
\n- The attention allows you to look at the totality of a sentence to make connections between any particular word and its relevant context.\n- For each input that the LSTM (Encoder) reads, the attention-mechanism takes into account several other inputs at the same time and decides which ones are important by attributing different weights to those inputs.\n- The Decoder will then take as input the encoded sentence and the weights provided by the attention-mechanism.\n- Transformer is a novel architecture that utilize Attention to process Sequence2Sequence tasks.\n- It consist on Encoder and Decoder but they are not based on the recurrent connections.\n- Encoder and Decoder are composed of modules that can be stacked on top of each other multiple times (Nx).\n\n- For each input that the LSTM (Encoder) reads, the attention-mechanism takes into account several other inputs at the same time and decides which ones are important by attributing different weights to those inputs.
0\n[images/image-4-Transformer Architecture Diagram.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"Inputs\"] --> B[\"Input Embedding\"]\nB --> C[\"Add & Norm\"]\nC --> D[\"Multi-Head Attention\"]\nD --> E[\"Add & Norm\"]\nE --> F[\"Feed Forward\"]\nF --> G[\"Add & Norm\"]\nG --> H[\"Feed Forward\"]\nH --> I[\"Output Probabilities\"]\nI --> J[\"Softmax\"]\nJ --> K[\"Linear\"]\nK --> L[\"Add & Norm\"]\nL --> M[\"Multi-Head Attention\"]\nM --> N[\"Masked Multi-Head Attention\"]\nN --> O[\"Add & Norm\"]\nO --> P[\"Multi-Head Attention\"]\nP --> Q[\"Add & Norm\"]\nQ --> R[\"Feed Forward\"]\nR --> S[\"Add & Norm\"]\nS --> T[\"Multi-Head Attention\"]\nT --> U[\"Masked Multi-Head Attention\"]\nU --> V[\"Output Embedding\"]\nV --> W[\"Positional Encoding\"]\nW --> X[\"Positational Encoding\"]\nX --> Y[\"Nx\"]\nY --> Z[\"Positational Encoding\"]\nZ --> AA[\"Inputs\"]\n```\n
\nFigure 1: The Transformer - model architecture.", + "path": "dl_10.pdf/Deep Learning Transformer network/Transformer network/Transformer network - Attention- What it is?", + "metadata": { + "length": 3397, + "summary": "The passage explains the attention mechanism, which allows a model to focus on relevant parts of input when processing each word, inspired by human reading. It describes how for each word, attention computes scores for every other word, then creates a weighted sum of representations to produce context-aware outputs. Key components include Query, Key, and Value vectors, with the formula Attention(Q,K,V)=softmax(QK^T/√d_k)V. The Transformer architecture is introduced as a novel sequence-to-sequence model using attention without recurrent connections, consisting of stacked encoder and decoder modules with multi-head attention, feed-forward layers, and normalization.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "tokens": [ + "Allows", + "model", + "focus", + "relevant", + "parts", + "input", + "processing", + "word", + "Inspired", + "humans", + "read", + "don", + "process", + "words", + "equally", + "For", + "Computes", + "attention", + "pay", + "sentence", + "Creates", + "weighted", + "sum", + "representations", + "Output", + "context", + "aware", + "representation", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "TD", + "Attention", + "Input", + "sequence", + "nThe", + "animal", + "didn", + "cross", + "Compute", + "scores", + "pair", + "Form", + "Each", + "projected", + "vectors", + "Query", + "What", + "Key", + "offers", + "Value", + "The", + "content", + "information", + "compute", + "weights", + "text", + "softmax", + "left", + "frac", + "sqrt", + "result", + "captures", + "Self", + "Heatmap", + "heatmap", + "totality", + "make", + "connections", + "LSTM", + "Encoder", + "reads", + "mechanism", + "takes", + "account", + "inputs", + "time", + "decides", + "important", + "attributing", + "Decoder", + "encoded", + "provided", + "Transformer", + "architecture", + "utilize", + "Sequence2Sequence", + "tasks", + "It", + "consist", + "based", + "recurrent", + "composed", + "modules", + "stacked", + "top", + "multiple", + "times", + "Nx", + "Inputs", + "Embedding", + "Add", + "Norm", + "Multi", + "Head", + "Feed", + "Forward", + "Probabilities", + "Softmax", + "Linear", + "Masked", + "Positional", + "Encoding", + "Positational", + "AA", + "Figure" + ], + "keywords": [ + "Attention", + "Transformer", + "Self-Attention" + ], + "connect_to": [ + { + "target": "1eaeda53-06bd-57e7-b424-15c56a5602a6", + "relation": "embeds", + "ref": "[images/image-1-Attention Mechanism Process.jpg]", + "position": { + "start": 458, + "end": 506 + } + }, + { + "target": "f1724798-2e92-5e82-a75c-a615cee8c88c", + "relation": "embeds", + "ref": "[images/image-2-Attention Matrix Heatmap.jpg]", + "position": { + "start": 1265, + "end": 1310 + } + }, + { + "target": "6907e0d7-0e26-5f71-962b-2fff66d237f0", + "relation": "embeds", + "ref": "[tables/table-0 Attention Matrix.html]", + "position": { + "start": 1350, + "end": 1388 + } + }, + { + "target": "3e6d769c-acf4-584b-9847-90f9e015564d", + "relation": "embeds", + "ref": "[images/image-3-Word Alignment Matrix.jpg]", + "position": { + "start": 1424, + "end": 1466 + } + }, + { + "target": "6d3df7f5-0233-521d-8ada-14352cc1fe99", + "relation": "embeds", + "ref": "[tables/table-1 EEA Agreement.html]", + "position": { + "start": 1506, + "end": 1541 + } + }, + { + "target": "423b548a-1bc4-5fea-8bc7-000e30d5e142", + "relation": "embeds", + "ref": "[images/image-4-Transformer Architecture Diagram.jpg]", + "position": { + "start": 2543, + "end": 2596 + } + } + ] + } + }, + { + "chunk_id": "110b737c-863a-5280-b5d2-8ef52a99d69f", + "type": "image", + "content": "\n$$
0\n[images/image-5-Heatmap Visualization.jpg]\n", + "path": "images/image-5-Heatmap Visualization.jpg", + "metadata": { + "length": 47, + "summary": "image-5\nChart Title: None visible\nAxis Labels:\n- X-axis: time step (range 0 to 196)\n- Y-axis: input dimension (range 0 to 11)\nColor Scale: Range from -0.8 (dark purple/black) to 0.8 (light beige/white), with 0.0 represented by red/orange.\n\nKey Data Patterns:\n- Input dimensions 0, 1, and 2 exhibit a high-frequency alternating pattern of dark and light vertical stripes across all time steps.\n- Input dimensions 3 through 5 show lower frequency oscillations with broader bands of color variation.\n- Input dimensions 6 through 11 display large, solid horizontal bands of uniform color, indicating stable values over the entire time range.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "images/image-5-Heatmap Visualization.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f6798ffe-dc58-54f3-879b-58a56696955a", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
input dimension01234567891011
0-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
1-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
2-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
3-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
4-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
5-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
6-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
7-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
9-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
10-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
11-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
", + "path": "tables/table-2 Constant Matrix.html", + "metadata": { + "length": 3684, + "summary": "table-3\nThe table shows a 12x12 matrix where every entry is -0.8, indicating a uniform constant value across all rows and columns.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "tables/table-2 Constant Matrix.html", + "keywords": [ + "constant", + "matrix", + "negative" + ], + "tokens": [] + } + }, + { + "chunk_id": "1010af9e-309e-5f34-a3fa-da7abc85e111", + "type": "image", + "content": "\n- Positional vectors are trainable.
0\n[images/image-6-Heatmap Visualization.jpg]\n", + "path": "images/image-6-Heatmap Visualization.jpg", + "metadata": { + "length": 80, + "summary": "image-6\nChart Title: Embedding dimension heatmap\nX-axis: Embedding dimension (0 to 15)\nY-axis: Position (0 to 48)\nData Pattern: The visualization displays a gradient of values across embedding dimensions. Dimensions 0-2 show high variability with alternating red and blue bands, indicating fluctuating positive and negative values. Dimension 3 shows a transition zone. Dimensions 4 through 15 are predominantly solid red, suggesting consistently high or positive values across all positions for these dimensions.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "images/image-6-Heatmap Visualization.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "faa0f7be-7ef5-5301-a204-4815d171fd9e", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
Position0123456789101112131415
48LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
45LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
39LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
36LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
33LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
30LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
", + "path": "tables/table-3 Low Values Table.html", + "metadata": { + "length": 2481, + "summary": "table-4\nThe table shows six rows (positions 48,45,39,36,33,30) each with 16 columns all containing the value 'Low'.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "tables/table-3 Low Values Table.html", + "keywords": [ + "Low", + "Table", + "Data" + ], + "tokens": [] + } + }, + { + "chunk_id": "1367db09-0636-5d8c-9612-6180cb429703", + "type": "image", + "content": "\n
0\n[images/image-7-Heatmap Visualization.jpg]\n", + "path": "images/image-7-Heatmap Visualization.jpg", + "metadata": { + "length": 55, + "summary": "image-7\nChart Title: Embedding dimension heatmap\nX-axis Label: Embedding dimension (Range: 0 to 15)\nY-axis Label: Position (Range: 0 to 48)\nData Representation: A grid of colored blocks representing values across embedding dimensions and positions.\nColor Scale: The visualization uses a diverging color scheme ranging from dark blue (likely low values) through light blue/white (neutral) to orange/red (high values).\nNotable Patterns:\n- Strong positive values (red/orange) are concentrated in the column for Embedding dimension 7, specifically at Positions 27 through 39.\n- Strong negative values (dark blue) appear in columns 2, 8, and 14.\n- Most other areas show neutral or weak values (light blue/white).", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "images/image-7-Heatmap Visualization.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "916ae609-9e8b-5734-a470-d9b3466dc51c", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
Position0123456789101112131415
0BlueOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
1BlueOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
2Dark BlueOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
3OrangeOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
4OrangeOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
5Medium BlueOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
6Medium OrangeOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
", + "path": "tables/table-4 Color Pattern Table.html", + "metadata": { + "length": 3449, + "summary": "table-5\nThe table shows a repeating pattern of colors across rows, with Blue, Orange, Light Blue, Red, and Dark Blue in columns 0-15 for rows 0-6.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "tables/table-4 Color Pattern Table.html", + "keywords": [ + "colors", + "pattern", + "table" + ], + "tokens": [] + } + }, + { + "chunk_id": "5cb2697e-5ddb-570a-93f7-578372834aa0", + "type": "image", + "content": "\n$$
0\n[images/image-8-$$ _detail.jpg]\n", + "path": "images/image-8-$$ _detail.jpg", + "metadata": { + "length": 47, + "summary": "image-8\n$$
0", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "images/image-8-$$ _detail.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "93577511-1d14-53bd-b03e-58ee85033f71", + "type": "image", + "content": "\nE --> G[\"K\"]
0\n[images/image-9-E --_ G[_K.jpg]\n", + "path": "images/image-9-E --_ G[_K.jpg", + "metadata": { + "length": 57, + "summary": "image-9\nE --> G[\"K\"]
0", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "images/image-9-E --_ G[_K.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "cf59c8f4-65f0-5cca-878f-1e98ebb2a9d0", + "type": "text", + "content": "- Transformers do not have recurrence or convolution.\n• Therefore, they have no built-in sense of word order.\n- Positional encoding injects information about the position of each token in the sequence.\n• These encodings are added to the input embeddings.\n- Uses sinusoidal functions to ensure unique and continuous position representation.\n$$\n\\mathsf {P E} _ {(p o s, 2 i)} = \\sin \\left(\\frac {p o s}{1 0 0 0 0 ^ {2 i / d _ {\\mathrm{model}}}}\\right)\n$$\n$$\n\\mathsf {P E} _ {(p o s, 2 i + 1)} = \\cos \\left(\\frac {p o s}{1 0 0 0 0 ^ {2 i / d _ {\\mathrm{model}}}}\\right)\n$$\n- pos is the position in the sequence (e.g., 0, 1, 2, ...).\n- $i$ is the dimension index.\n- $d_{model}$ is the total dimension of the embedding vector.\nSinusoid absolute positional encoding\n\n$$
0\n[images/image-5-Heatmap Visualization.jpg]\n\n
\nheatmap\n\n[tables/table-2 Constant Matrix.html]\n\nThe data is a heatmap with a color scale ranging from ~-1 to ~1 for each row and column of the time step axis.\n
\n\\- Sinusoidal functions are used to provide:\n- Unique yet smoothly varying encoding for each position.\n- Ability to interpolate or generalize to longer sequences.\n- Simple mathematical patterns that can capture relative distances.\n• Alternative: learned positional encodings\n- Positional vectors are trainable.\n- More flexible, but limited to sequence lengths seen during training.\n\\- Without PE:\n\\- Model loses word order completely (e.g., “dog bites man” = “man bites dog”).\nSinusoidal Positional Encoding\n\n- Positional vectors are trainable.
0\n[images/image-6-Heatmap Visualization.jpg]\n\n
\nheatmap\n\n[tables/table-3 Low Values Table.html]\n\n
\nLearned Positional Encoding (simulated)\n\n
0\n[images/image-7-Heatmap Visualization.jpg]\n\n
\nheatmap\n\n[tables/table-4 Color Pattern Table.html]\n\n
\n- Instead of computing attention once, we do it in parallel multiple times (heads).\n• Each head learns different projections of Q, K, V with its own parameters:\n$$\n\\text { head } _ {i} = \\text { Attention } (Q W _ {i} ^ {Q}, K W _ {i} ^ {K}, V W _ {i} ^ {V})\n$$\n\\- All heads are concatenated and projected again:\n$$\n\\operatorname{MultiHead} (Q, K, V) = \\operatorname{Concat} \\left(\\operatorname{head} _ {1}, \\dots , \\operatorname{head} _ {h}\\right) W ^ {O}\n$$\n\\- This allows the model to jointly attend to information from different representation subspaces.\nScaled Dot-Product Attention\n\n$$
0\n[images/image-8-$$ _detail.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"MatMul\"] --> B[\"SoftMax\"]\nB --> C[\"Mask (opt.)\"]\nC --> D[\"Scale\"]\nD --> E[\"MatMul\"]\nE --> F[\"Q\"]\nE --> G[\"K\"]\nE --> H[\"V\"]\n```\n
\nMulti-Head Attention\n\nE --> G[\"K\"]
0\n[images/image-9-E --_ G[_K.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"Linear\"] --> B[\"Concat\"]\nB --> C[\"Scaled Dot-Product Attention\"]\nC --> D[\"Linear\"]\nC --> E[\"Linear\"]\nC --> F[\"Linear\"]\nD --> G[\"V\"]\nE --> H[\"K\"]\nF --> I[\"Q\"]\nG --> C\nH --> C\nI --> C\nC --> J[\"h\"]\n```\n
\nFigure 2: (left) Scaled Dot-Product Attention. (right) Multi-Head Attention consists of several attention layers running in parallel.\n- Multi-head attention = multiple attention mechanisms in parallel.\n• Each head uses its own learned projection matrices:\n$$\nQ W _ {i} ^ {Q}, \\quad K W _ {i} ^ {K}, \\quad V W _ {i} ^ {V}\n$$\n- Different heads can focus on different types of relationships (e.g. syntax, semantics).\n- The outputs of all heads are concatenated and linearly projected:\n$$\n\\operatorname{Concat} \\left(\\text {head} _ {1}, \\dots , \\text {head} _ {h}\\right) W ^ {0}\n$$\n\\- Encoder and decoder use multi-head attention differently:\n- Encoder: attends to all positions in the input.\n- Decoder: attends to previous positions (masked).\n- Encoder-decoder: decoder attends to encoder output.", + "path": "dl_10.pdf/Deep Learning Transformer network/Transformer network/Transformer network - Positional encoding- What is?", + "metadata": { + "length": 3834, + "summary": "The content explains positional encoding in Transformers, which injects word order information using sinusoidal functions. It covers the formulas for PE, the benefits of sinusoidal encoding (unique, continuous, generalizable), and contrasts it with learned positional encodings. It then introduces multi-head attention, where multiple attention heads run in parallel with different learned projections, allowing the model to attend to different representation subspaces. The encoder and decoder use multi-head attention differently: encoder attends to all input positions, decoder attends to previous positions (masked), and encoder-decoder attention allows the decoder to attend to encoder output.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "tokens": [ + "Transformers", + "recurrence", + "convolution", + "Therefore", + "built", + "sense", + "word", + "order", + "Positional", + "encoding", + "injects", + "information", + "position", + "token", + "sequence", + "These", + "encodings", + "added", + "input", + "embeddings", + "Uses", + "sinusoidal", + "functions", + "ensure", + "unique", + "continuous", + "representation", + "mathsf", + "sin", + "left", + "frac", + "mathrm", + "model", + "cos", + "pos", + "dimension", + "index", + "total", + "embedding", + "vector", + "Sinusoid", + "absolute", + "positional", + "details", + "summary", + "heatmap", + "The", + "data", + "color", + "scale", + "ranging", + "row", + "column", + "time", + "step", + "axis", + "Sinusoidal", + "provide", + "Unique", + "smoothly", + "varying", + "Ability", + "interpolate", + "generalize", + "longer", + "sequences", + "Simple", + "mathematical", + "patterns", + "capture", + "relative", + "distances", + "Alternative", + "learned", + "vectors", + "trainable", + "More", + "flexible", + "limited", + "lengths", + "training", + "Without", + "PE", + "Model", + "loses", + "completely", + "dog", + "bites", + "man", + "Encoding", + "Learned", + "simulated", + "Instead", + "computing", + "attention", + "parallel", + "multiple", + "times", + "heads", + "Each", + "head", + "learns", + "projections", + "parameters", + "text", + "Attention", + "All", + "concatenated", + "projected", + "operatorname", + "MultiHead", + "Concat", + "dots", + "This", + "jointly", + "attend", + "subspaces", + "Scaled", + "Dot", + "Product", + "flowchart", + "mermaid", + "graph", + "TD", + "MatMul", + "SoftMax", + "Mask", + "opt", + "Scale", + "Multi", + "Head", + "Linear", + "Figure", + "consists", + "layers", + "running", + "mechanisms", + "projection", + "matrices", + "quad", + "Different", + "focus", + "types", + "relationships", + "syntax", + "semantics", + "outputs", + "linearly", + "Encoder", + "decoder", + "multi", + "differently", + "attends", + "positions", + "Decoder", + "previous", + "masked", + "encoder", + "output" + ], + "keywords": [ + "Positional Encoding", + "Multi-Head Attention", + "Transformer" + ], + "connect_to": [ + { + "target": "110b737c-863a-5280-b5d2-8ef52a99d69f", + "relation": "embeds", + "ref": "[images/image-5-Heatmap Visualization.jpg]", + "position": { + "start": 775, + "end": 817 + } + }, + { + "target": "f6798ffe-dc58-54f3-879b-58a56696955a", + "relation": "embeds", + "ref": "[tables/table-2 Constant Matrix.html]", + "position": { + "start": 857, + "end": 894 + } + }, + { + "target": "1010af9e-309e-5f34-a3fa-da7abc85e111", + "relation": "embeds", + "ref": "[images/image-6-Heatmap Visualization.jpg]", + "position": { + "start": 1574, + "end": 1616 + } + }, + { + "target": "faa0f7be-7ef5-5301-a204-4815d171fd9e", + "relation": "embeds", + "ref": "[tables/table-3 Low Values Table.html]", + "position": { + "start": 1656, + "end": 1694 + } + }, + { + "target": "1367db09-0636-5d8c-9612-6180cb429703", + "relation": "embeds", + "ref": "[images/image-7-Heatmap Visualization.jpg]", + "position": { + "start": 1770, + "end": 1812 + } + }, + { + "target": "916ae609-9e8b-5734-a470-d9b3466dc51c", + "relation": "embeds", + "ref": "[tables/table-4 Color Pattern Table.html]", + "position": { + "start": 1852, + "end": 1893 + } + }, + { + "target": "5cb2697e-5ddb-570a-93f7-578372834aa0", + "relation": "embeds", + "ref": "[images/image-8-$$ _detail.jpg]", + "position": { + "start": 2509, + "end": 2540 + } + }, + { + "target": "93577511-1d14-53bd-b03e-58ee85033f71", + "relation": "embeds", + "ref": "[images/image-9-E --_ G[_K.jpg]", + "position": { + "start": 2788, + "end": 2819 + } + } + ] + } + }, + { + "chunk_id": "ea4f6fa9-15a7-55d0-99d2-066cbc5f164e", + "type": "text", + "content": "- Uses teacher forcing – the true previous output is fed into the decoder.\n- The target sequence is right-shifted by one position.\n- A token is added at the beginning, and at the end.\n- A causal mask ensures decoder only sees previous positions.\n- Loss is computed between decoder output and original target sequence.\n- The target sequence is right-shifted to prevent the model from copying input.\n- A token is added at the beginning, and at the end.\n- A causal mask ensures that each decoder step sees only previous tokens.\n- The loss is computed against the original (unshifted) target sequence.", + "path": "dl_10.pdf/Deep Learning Transformer network/Training a Transformer (Seq2Seq model):", + "metadata": { + "length": 621, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "tokens": [ + "Uses", + "teacher", + "forcing", + "true", + "previous", + "output", + "fed", + "decoder", + "The", + "target", + "sequence", + "shifted", + "position", + "sos", + "token", + "added", + "beginning", + "eos", + "end", + "causal", + "mask", + "ensures", + "sees", + "positions", + "Loss", + "computed", + "original", + "prevent", + "model", + "copying", + "input", + "step", + "tokens", + "loss", + "unshifted" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8a8721e3-5d3a-5a6e-bf94-a66ff306e428", + "type": "image", + "content": "\n3. Predict the first output token.
0\n[images/image-10-Attention Mechanism Visualization.jpg]\n", + "path": "images/image-10-Attention Mechanism Visualization.jpg", + "metadata": { + "length": 80, + "summary": "image-10\nTop Layer (Source Tokens): A, man, is, walking, .\nBottom Layer (Target Tokens): , Un, homme, marche, .\nRelationships: Dense connections (attention weights) link each source token to every target token.\nContext: Illustrates a sequence-to-sequence model translating English (\"A man is walking\") to French (\"Un homme marche\").", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "images/image-10-Attention Mechanism Visualization.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ca042cfe-7e47-58d7-bdc6-8e5a938016f4", + "type": "image", + "content": "\n- As opposed to directional models, which read the text input sequentially (left-to-right or right-to-left), the Transformer encoder reads the entire sequence of words at once.
0\n[images/image-11-Transformer Encoder Architecture.jpg]\n", + "path": "images/image-11-Transformer Encoder Architecture.jpg", + "metadata": { + "length": 222, + "summary": "image-11\n- Input Sequence: W1, W2, W3, [MASK], W5 (where W4 is masked)\n- Embedding Layer: Converts input tokens into embeddings\n- Core Component: Transformer encoder block\n- Output Representation: O1, O2, O3, O4, O5\n- Classification Layer: Fully-connected layer + GELU activation + Normalization\n- Final Output: Predicted probabilities w'1 through w'5 via embedding to vocabulary and softmax", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "images/image-11-Transformer Encoder Architecture.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "40f3c966-8ee4-5a7e-acb8-a9dd94ca1ff5", + "type": "table", + "content": "
Input[MASK][MASK]
[CLS]mydogiscute[SEP]helikesplay##ing[SEP]
Token Embeddings $E_{[CLS]}$ $E_{my}$ $E_{[MASK]}$ $E_{is}$ $E_{cute}$ $E_{[SEP]}$ $E_{he}$ $E_{[MASK]}$ $E_{play}$ $E_{##ing}$ $E_{[SEP]}$
Sentence Embedding+++++++++++
$E_A$ $E_A$ $E_A$ $E_A$ $E_A$ $E_A$ $E_B$ $E_B$ $E_B$ $E_B$ $E_B$
Transformer Positional Embedding+++++++++++
$E_0$ $E_1$ $E_2$ $E_3$ $E_4$ $E_5$ $E_6$ $E_7$ $E_8$ $E_9$ $E_{10}$
", + "path": "tables/table-5 BERT Embeddings.html", + "metadata": { + "length": 1224, + "summary": "table-6\nThe table shows BERT input embeddings: token, sentence (A/B), and positional embeddings for two sentences separated by [SEP], with [MASK] tokens.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "file_path": "tables/table-5 BERT Embeddings.html", + "keywords": [ + "BERT", + "Embeddings", + "Tokens" + ], + "tokens": [] + } + }, + { + "chunk_id": "81edf90e-6892-5363-a7b8-3f571e070f9c", + "type": "text", + "content": "1. Input the full encoder sequence.\n2. Start decoder with token.\n3. Predict the first output token.\n4. Append it to the decoder input and repeat.\n5. Stop when the model predicts .\nAt each step, the decoder sees its own past outputs and the encoder context.\nTransformer Inference (Autoregressive Decoding)\n\n3. Predict the first output token.
0\n[images/image-10-Attention Mechanism Visualization.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"A\"] --> Un[\"Un\"]\nman[\"man\"] --> Un\nis[\"is\"] --> Un\nwalking[\"walking\"] --> Un\nUn --> homme[\"homme\"]\nhomme --> marche[\"marche\"]\nmarche --> u[\"u\"]\n```\n
\n• BERT - Bidirectional Encoder Representations from Transformers.\n- Uses a Transformer-based network with pre-trained deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers.\n- BERT makes use of Transformer, an attention mechanism that learns contextual relations between words (or sub-words) in a text.\n- Transformer includes two separate mechanisms — an encoder that reads the text input and a decoder that produces a prediction for the task. Since BERT's goal is to generate a language model, only the encoder mechanism is necessary.\n- As opposed to directional models, which read the text input sequentially (left-to-right or right-to-left), the Transformer encoder reads the entire sequence of words at once.\n- Therefore it is considered bidirectional, though it would be more accurate to say that it's non-directional. This characteristic allows the model to learn the context of a word based on all of its surroundings (left and right of the word).\n- The input is a sequence of tokens, which are first embedded into vectors and then processed in the neural network.\n- The output is a sequence of vectors of size H, in which each vector corresponds to an input token with the same index.\n- When training language models, there is a challenge of defining a prediction goal.\n\n- As opposed to directional models, which read the text input sequentially (left-to-right or right-to-left), the Transformer encoder reads the entire sequence of words at once.
0\n[images/image-11-Transformer Encoder Architecture.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"Embedding\"] --> B[\"W1\"]\nA --> C[\"W2\"]\nA --> D[\"W3\"]\nA --> E[\"[MASK\"]]\nA --> F[\"W5\"]\nB --> G[\"W1\"]\nC --> H[\"W2\"]\nD --> I[\"W3\"]\nE --> J[\"W4\"]\nF --> K[\"W5\"]\nL[\"Embedding to vocab + softmax\"] --> M[\"Classification Layer: Fully-connected layer + GELU + Norm\"]\nM --> N[\"O1\"]\nM --> O[\"O2\"]\nM --> P[\"O3\"]\nM --> Q[\"O4\"]\nM --> R[\"O5\"]\nN --> S[\"Transformer encoder\"]\nO --> S\nP --> S\nQ --> S\nR --> S\n```\n
\n- Before feeding word sequences into BERT, 15% of the words in each sequence are replaced with a [MASK] token.\n- The model then attempts to predict the original value of the masked words, based on the context provided by the other, non-masked, words in the sequence.\n- In technical terms, the prediction of the output words requires:\n1. Adding a classification layer on top of the encoder output.\n2. Multiplying the output vectors by the embedding matrix, transforming them into the vocabulary dimension.\n3. Calculating the probability of each word in the vocabulary with softmax.\n\\- The BERT loss function takes into consideration only the prediction of the masked values and ignores the prediction of the non-masked words.\n- In the BERT training process, the model receives pairs of sentences as input and learns to predict if the second sentence in the pair is the subsequent sentence in the original document.\n- During training, 50% of the inputs are a pair in which the second sentence is the subsequent sentence in the original document, while in the other 50% a random sentence from the corpus is chosen as the second sentence.\n- The assumption is that the random sentence will be disconnected from the first sentence.\n\\- To help the model distinguish between the two sentences in training, the input is processed in the following way before entering the model:\n1. A [CLS] token is inserted at the beginning of the first sentence and a [SEP] token is inserted at the end of each sentence.\n2. A sentence embedding indicating Sentence A or Sentence B is added to each token. Sentence embeddings are similar in concept to token embeddings with a vocabulary of 2.\n3. A positional embedding is added to each token to indicate its position in the sequence. The concept and implementation of positional embedding are presented in the Transformer paper.\n\n[tables/table-5 BERT Embeddings.html]\n\n\\- To predict if the second sentence is indeed connected to the first, the following steps are performed:\n1. The entire input sequence goes through the Transformer model.\n2. The output of the [CLS] token is transformed into a $2 \\times 1$ shaped vector, using a simple classification layer (learned matrices of weights and biases).\n3. Calculating the probability of IsNextSequence with softmax.\n\\- When training the BERT model, Masked LM and Next Sentence Prediction are trained together, with the goal of minimizing the combined loss function of the two strategies.", + "path": "dl_10.pdf/Deep Learning Transformer network/Inference – autoregressive decoding:", + "metadata": { + "length": 5109, + "summary": "The passage describes two main topics: autoregressive decoding in Transformer inference, where the decoder generates tokens step-by-step using past outputs and encoder context until an token is produced; and BERT (Bidirectional Encoder Representations from Transformers), a bidirectional model using only the Transformer encoder to learn contextual word representations from unlabeled text. BERT is trained with masked language modeling (predicting masked tokens) and next sentence prediction, using special tokens like [CLS] and [SEP], along with sentence and positional embeddings.", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "tokens": [ + "Input", + "full", + "encoder", + "sequence", + "Start", + "decoder", + "sos", + "token", + "Predict", + "output", + "Append", + "input", + "repeat", + "Stop", + "model", + "predicts", + "eos", + "At", + "step", + "sees", + "past", + "outputs", + "context", + "Transformer", + "Inference", + "Autoregressive", + "Decoding", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "TD", + "Un", + "man", + "walking", + "homme", + "marche", + "BERT", + "Bidirectional", + "Encoder", + "Representations", + "Transformers", + "Uses", + "based", + "network", + "pre", + "trained", + "deep", + "bidirectional", + "representations", + "unlabeled", + "text", + "jointly", + "conditioning", + "left", + "layers", + "makes", + "attention", + "mechanism", + "learns", + "contextual", + "relations", + "words", + "includes", + "separate", + "mechanisms", + "reads", + "produces", + "prediction", + "task", + "Since", + "goal", + "generate", + "language", + "As", + "opposed", + "directional", + "models", + "read", + "sequentially", + "entire", + "Therefore", + "considered", + "accurate", + "This", + "characteristic", + "learn", + "word", + "surroundings", + "The", + "tokens", + "embedded", + "vectors", + "processed", + "neural", + "size", + "vector", + "corresponds", + "index", + "When", + "training", + "challenge", + "defining", + "Embedding", + "W1", + "W2", + "W3", + "MASK", + "W5", + "W4", + "vocab", + "softmax", + "Classification", + "Layer", + "Fully", + "connected", + "layer", + "GELU", + "Norm", + "O1", + "O2", + "O3", + "O4", + "O5", + "Before", + "feeding", + "sequences", + "15%", + "replaced", + "attempts", + "predict", + "original", + "masked", + "provided", + "In", + "technical", + "terms", + "requires", + "Adding", + "classification", + "top", + "Multiplying", + "embedding", + "matrix", + "transforming", + "vocabulary", + "dimension", + "Calculating", + "probability", + "loss", + "function", + "takes", + "consideration", + "values", + "ignores", + "process", + "receives", + "pairs", + "sentences", + "sentence", + "pair", + "subsequent", + "document", + "During", + "50%", + "inputs", + "random", + "corpus", + "chosen", + "assumption", + "disconnected", + "To", + "distinguish", + "entering", + "CLS", + "inserted", + "beginning", + "SEP", + "end", + "indicating", + "Sentence", + "added", + "embeddings", + "similar", + "concept", + "positional", + "position", + "implementation", + "presented", + "paper", + "html", + "steps", + "performed", + "transformed", + "times", + "shaped", + "simple", + "learned", + "matrices", + "weights", + "biases", + "IsNextSequence", + "Masked", + "LM", + "Next", + "Prediction", + "minimizing", + "combined", + "strategies" + ], + "keywords": [ + "BERT", + "Transformer", + "Autoregressive Decoding" + ], + "connect_to": [ + { + "target": "8a8721e3-5d3a-5a6e-bf94-a66ff306e428", + "relation": "embeds", + "ref": "[images/image-10-Attention Mechanism Visualization.jpg]", + "position": { + "start": 363, + "end": 418 + } + }, + { + "target": "ca042cfe-7e47-58d7-bdc6-8e5a938016f4", + "relation": "embeds", + "ref": "[images/image-11-Transformer Encoder Architecture.jpg]", + "position": { + "start": 2177, + "end": 2231 + } + }, + { + "target": "40f3c966-8ee4-5a7e-acb8-a9dd94ca1ff5", + "relation": "embeds", + "ref": "[tables/table-5 BERT Embeddings.html]", + "position": { + "start": 4552, + "end": 4589 + } + } + ] + } + }, + { + "chunk_id": "bab774c6-87e7-54fe-8c90-4d6daf3c590a", + "type": "text", + "content": "\\- BERT can be used for a wide variety of language tasks, while only adding a small layer to the core model:\n1. Classification tasks such as sentiment analysis are done similarly to Next Sentence classification, by adding a classification layer on top of the Transformer output for the [CLS] token.\n2. In Question Answering tasks (e.g. SQuAD v1.1), the software receives a question regarding a text sequence and is required to mark the answer in the sequence. Using BERT, a Q&A model can be trained by learning two extra vectors that mark the beginning and the end of the answer.\n3. In Named Entity Recognition (NER), the software receives a text sequence and is required to mark the various types of entities (Person, Organization, Date, etc) that appear in the text. Using BERT, a NER model can be trained by feeding the output vector of each token into a classification layer that predicts the NER label.\n1. What is a Transformer? by Maxime,\nhttps://medium.com/inside-machine-learning/\nwhat-is-a-transformer-d07dd1fbec04\n2. A Beginner's Guide to Attention Mechanisms and Memory Networks https://wiki.pathmind.com/attention-mechanism-memory-network\n3. What is Teacher Forcing for Recurrent Neural Networks? by Jason Brownlee https://machinelearningmastery.com/ teacher-forcing-for-recurrent-neural-networks/\n4. Positional Encoding: Everything You Need to Know by Darjan Salaj\nhttps://www.inovex.de/de/blog/\npositional-encoding-everything-you-need-to-know", + "path": "dl_10.pdf/Deep Learning Transformer network/Transformer network - BERT - How to use BERT (Fine-tuning)", + "metadata": { + "length": 1456, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "tokens": [ + "BERT", + "wide", + "variety", + "language", + "tasks", + "adding", + "small", + "layer", + "core", + "model", + "Classification", + "sentiment", + "analysis", + "similarly", + "Next", + "Sentence", + "classification", + "top", + "Transformer", + "output", + "CLS", + "token", + "In", + "Question", + "Answering", + "SQuAD", + "v1.1", + "software", + "receives", + "question", + "text", + "sequence", + "required", + "mark", + "answer", + "Using", + "trained", + "learning", + "extra", + "vectors", + "beginning", + "end", + "Named", + "Entity", + "Recognition", + "NER", + "types", + "entities", + "Person", + "Organization", + "Date", + "feeding", + "vector", + "predicts", + "label", + "What", + "Maxime", + "https", + "medium", + "inside", + "machine", + "transformer", + "d07dd1fbec04", + "Beginner", + "Guide", + "Attention", + "Mechanisms", + "Memory", + "Networks", + "wiki", + "pathmind", + "attention", + "mechanism", + "memory", + "network", + "Teacher", + "Forcing", + "Recurrent", + "Neural", + "Jason", + "Brownlee", + "machinelearningmastery", + "teacher", + "forcing", + "recurrent", + "neural", + "networks", + "Positional", + "Encoding", + "Everything", + "You", + "Need", + "Know", + "Darjan", + "Salaj", + "www", + "inovex", + "de", + "blog", + "positional", + "encoding" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5bc569ab-8bf7-519b-be49-f56f4787c1b8", + "type": "text", + "content": "5. Luong, Minh-Thang, Hieu Pham, and Christopher D. Manning. \"Effective approaches to attention-based neural machine translation.\" arXiv preprint arXiv:1508.04025 (2015).\n6. Devlin, Jacob, et al. \"Bert: Pre-training of deep bidirectional transformers for language understanding.\" arXiv preprint arXiv:1810.04805 (2018).\n7. BERT Explained: State of the art language model for NLP by Rani Horev\nhttps://towardsdatascience.com/\nbert-explained-state-of-the-art-language-model-for-nlp-f8b21a\n8. Understanding BERT Transformer: Attention isn't all you need by Damien Sileo https://medium.com/synapse-dev/understanding-bert-transformer-attention-isnt-all-you-need-58", + "path": "dl_10.pdf/Deep Learning Transformer network/Transformer network - References ii", + "metadata": { + "length": 659, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Deep Learning Transformer network", + "tokens": [ + "Luong", + "Minh", + "Thang", + "Hieu", + "Pham", + "Christopher", + "Manning", + "Effective", + "approaches", + "attention", + "based", + "neural", + "machine", + "translation", + "arXiv", + "preprint", + "1508.04025", + "2015", + "Devlin", + "Jacob", + "al", + "Bert", + "Pre", + "training", + "deep", + "bidirectional", + "transformers", + "language", + "understanding", + "1810.04805", + "2018", + "BERT", + "Explained", + "State", + "art", + "model", + "NLP", + "Rani", + "Horev", + "https", + "towardsdatascience", + "bert", + "explained", + "state", + "nlp", + "f8b21a", + "Understanding", + "Transformer", + "Attention", + "isn", + "Damien", + "Sileo", + "medium", + "synapse", + "dev", + "transformer", + "isnt", + "58" + ], + "keywords": [], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/doc_nav.json b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/doc_nav.json new file mode 100644 index 000000000..cc552ad43 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/doc_nav.json @@ -0,0 +1,153 @@ +{ + "version": "1.0", + "file_name": "Deep Learning - Transformer Network.pdf", + "stats": { + "total_chunks": 25, + "text_chunks": 8, + "image_chunks": 11, + "table_chunks": 6, + "max_depth": 3 + }, + "sections": [ + { + "title": "Deep Learning Transformer network", + "path": "dl_10.pdf/Deep Learning Transformer network", + "level": 1, + "summary": "Transformer network Jan Platoš, Radek Svoboda May 5, 2026 Department of Computer Science Faculty of Electrical Engineering and Computer Science VŠB - Technical University of Ostrava", + "chunk_count": 8, + "children": [ + { + "title": "Transformer network", + "path": "dl_10.pdf/Deep Learning Transformer network/Transformer network", + "level": 2, + "summary": "- Recurrent neural network works in synchronized serial way. - This prevent efficient computation in parallel. - The attention-mechanism looks at an input sequence and decides at each step which other parts of the sequence are important. - The attention mechanism helps the decoder to focus on the important part of the input sequence.", + "chunk_count": 3, + "children": [ + { + "title": "Transformer network - Attention- What it is?", + "path": "dl_10.pdf/Deep Learning Transformer network/Transformer network/Transformer network - Attention- What it is?", + "level": 3, + "summary": "The passage explains the attention mechanism, which allows a model to focus on relevant parts of input when processing each word, inspired by human reading. It describes how for each word, attention computes scores for every other word, then creates a weighted sum of representations to produce context-aware outputs. Key components include Query, Key, and Value vectors, with the formula Attention(Q,K,V)=softmax(QK^T/√d_k)V. The Transformer architecture is introduced as a novel sequence-to-sequence model using attention without recurrent connections, consisting of stacked encoder and decoder modules with multi-head attention, feed-forward layers, and normalization.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Transformer network - Positional encoding- What is?", + "path": "dl_10.pdf/Deep Learning Transformer network/Transformer network/Transformer network - Positional encoding- What is?", + "level": 3, + "summary": "The content explains positional encoding in Transformers, which injects word order information using sinusoidal functions. It covers the formulas for PE, the benefits of sinusoidal encoding (unique, continuous, generalizable), and contrasts it with learned positional encodings. It then introduces multi-head attention, where multiple attention heads run in parallel with different learned projections, allowing the model to attend to different representation subspaces. The encoder and decoder use multi-head attention differently: encoder attends to all input positions, decoder attends to previous positions (masked), and encoder-decoder attention allows the decoder to attend to encoder output.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Training a Transformer (Seq2Seq model):", + "path": "dl_10.pdf/Deep Learning Transformer network/Training a Transformer (Seq2Seq model):", + "level": 2, + "summary": "- Uses teacher forcing – the true previous output is fed into the decoder. - The target sequence is right-shifted by one position. - A token is added at the beginning, and at the end. - A causal mask ensures decoder only sees previous positions. - Loss is computed between decoder output and original target sequence. - The target sequence is right-shifted to prevent the model from copying input. - A token is added at the beginning, and at the end. - A causal mask ensures that each decoder step sees only previous tokens. - The loss is computed against the original (unshifted) target sequence.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Inference – autoregressive decoding:", + "path": "dl_10.pdf/Deep Learning Transformer network/Inference – autoregressive decoding:", + "level": 2, + "summary": "The passage describes two main topics: autoregressive decoding in Transformer inference, where the decoder generates tokens step-by-step using past outputs and encoder context until an token is produced; and BERT (Bidirectional Encoder Representations from Transformers), a bidirectional model using only the Transformer encoder to learn contextual word representations from unlabeled text. BERT is trained with masked language modeling (predicting masked tokens) and next sentence prediction, using special tokens like [CLS] and [SEP], along with sentence and positional embeddings.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Transformer network - BERT - How to use BERT (Fine-tuning)", + "path": "dl_10.pdf/Deep Learning Transformer network/Transformer network - BERT - How to use BERT (Fine-tuning)", + "level": 2, + "summary": "\\- BERT can be used for a wide variety of language tasks, while only adding a small layer to the core model: 1. Classification tasks such as sentiment analysis are done similarly to Next Sentence classification, by adding a classification layer on top of the Transformer output for the [CLS] token. 2. In Question Answering tasks (e.g. SQuAD v1.1), the software receives a question regarding a text sequence and is required to mark the answer in the sequence. Using BERT, a Q&A model can be trained by learning two extra vectors that mark the beginning and the end of the answer. 3. In Named Entity Recognition (NER), the software receives a text sequence and is required to mark the various types of entities (Person, Organization, Date, etc) that appear in the text. Using BERT, a NER model can be trained by feeding the output vector of each token into a classification layer that predicts the NER label. 1. What is a Transformer? by Maxime, https://medium.com/inside-machine-learning/ what-is-a-transformer-d07dd1fbec04 2. A Beginner's Guide to Attention Mechanisms and Memory Networks https://wiki.pathmind.com/attention-mechanism-memory-network 3. What is Teacher Forcing for Recurrent Neural Networks? by Jason Brownlee https://machinelearningmastery.com/ teacher-forcing-for-recurrent-neural-networks/ 4. Positional Encoding: Everything You Need to Know by Darjan Salaj https://www.inovex.de/de/blog/ positional-encoding-everything-you-need-to-know", + "chunk_count": 1, + "children": [] + }, + { + "title": "Transformer network - References ii", + "path": "dl_10.pdf/Deep Learning Transformer network/Transformer network - References ii", + "level": 2, + "summary": "5. Luong, Minh-Thang, Hieu Pham, and Christopher D. Manning. \"Effective approaches to attention-based neural machine translation.\" arXiv preprint arXiv:1508.04025 (2015). 6. Devlin, Jacob, et al. \"Bert: Pre-training of deep bidirectional transformers for language understanding.\" arXiv preprint arXiv:1810.04805 (2018). 7. BERT Explained: State of the art language model for NLP by Rani Horev https://towardsdatascience.com/ bert-explained-state-of-the-art-language-model-for-nlp-f8b21a 8. Understanding BERT Transformer: Attention isn't all you need by Damien Sileo https://medium.com/synapse-dev/understanding-bert-transformer-attention-isnt-all-you-need-58", + "chunk_count": 1, + "children": [] + } + ] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-Attention Mechanism Process.jpg", + "summary": "image-1 1. Input: sequence of words (e.g., \"The animal didn't cross ...\") 2. Compute attention scores for every pair of words (represented by a grid matrix) 3. Form weighted sum of word representations 4. Output: context-aware representations" + }, + { + "path": "images/image-2-Attention Matrix Heatmap.jpg", + "summary": "image-2 Chart Title: Words attended to (X-axis) vs. Words attending (Y-axis) Data Source: Attention weights between word pairs Key Data Points: - Highest attention values (0.4): - \"animal\" -> \"animal\" (Self-attention) - \"didn't\" -> \"didn't\" (Self-attention) - \"cross\" -> \"cross\" (Self-attention) - Notable cross-word attention (0.3): - \"The\" -> \"animal\" - \"the\" -> \"the\" - \"street\" -> \"street\" (Self-attention) - Lowest attention values (0.1): - Most off-diagonal interactions show minimal attention (0.1) Scale Range: 0.10 to 0.40 Words Involved: The, animal, didn't, cross, the, street" + }, + { + "path": "images/image-3-Word Alignment Matrix.jpg", + "summary": "image-3 The image displays a diagonal alignment matrix visualizing the correspondence between English and French phrases. **English Text (Top Axis):** \"The agreement on the European Economic Area was signed in August 1992 \" **French Text (Left Axis):** \"L'accord sur la zone économique européenne a été signé en août 1992 \" **Alignment Pattern:** - The matrix shows a strong diagonal pattern, indicating a direct word-for-word translation structure. - Key alignments visible include: \"agreement\" <-> \"accord\", \"European Economic Area\" <-> \"zone économique européenne\", \"signed\" <-> \"signé\", and \"August 1992\" <-> \"août 1992\"." + }, + { + "path": "images/image-4-Transformer Architecture Diagram.jpg", + "summary": "image-4 - **Input Side (Encoder):** - Inputs flow into Input Embedding. - Combined with Positional Encoding. - Passes through N identical layers (labeled Nx). - Each layer contains: - Multi-Head Attention block followed by Add & Norm. - Feed Forward block followed by Add & Norm. - **Output Side (Decoder):** - Outputs (shifted right) flow into Output Embedding. - Combined with Positional Encoding. - Passes through N identical layers (labeled Nx). - Each layer contains: - Masked Multi-Head Attention block followed by Add & Norm. - Multi-Head Attention block followed by Add & Norm. - Feed Forward block followed by Add & Norm. - **Final Output:** - Linear layer processes the final decoder output. - Softmax layer produces Output Probabilities." + }, + { + "path": "images/image-5-Heatmap Visualization.jpg", + "summary": "image-5 Chart Title: None visible Axis Labels: - X-axis: time step (range 0 to 196) - Y-axis: input dimension (range 0 to 11) Color Scale: Range from -0.8 (dark purple/black) to 0.8 (light beige/white), with 0.0 represented by red/orange. Key Data Patterns: - Input dimensions 0, 1, and 2 exhibit a high-frequency alternating pattern of dark and light vertical stripes across all time steps. - Input dimensions 3 through 5 show lower frequency oscillations with broader bands of color variation. - Input dimensions 6 through 11 display large, solid horizontal bands of uniform color, indicating stable values over the entire time range." + }, + { + "path": "images/image-6-Heatmap Visualization.jpg", + "summary": "image-6 Chart Title: Embedding dimension heatmap X-axis: Embedding dimension (0 to 15) Y-axis: Position (0 to 48) Data Pattern: The visualization displays a gradient of values across embedding dimensions. Dimensions 0-2 show high variability with alternating red and blue bands, indicating fluctuating positive and negative values. Dimension 3 shows a transition zone. Dimensions 4 through 15 are predominantly solid red, suggesting consistently high or positive values across all positions for these dimensions." + }, + { + "path": "images/image-7-Heatmap Visualization.jpg", + "summary": "image-7 Chart Title: Embedding dimension heatmap X-axis Label: Embedding dimension (Range: 0 to 15) Y-axis Label: Position (Range: 0 to 48) Data Representation: A grid of colored blocks representing values across embedding dimensions and positions. Color Scale: The visualization uses a diverging color scheme ranging from dark blue (likely low values) through light blue/white (neutral) to orange/red (high values). Notable Patterns: - Strong positive values (red/orange) are concentrated in the column for Embedding dimension 7, specifically at Positions 27 through 39. - Strong negative values (dark blue) appear in columns 2, 8, and 14. - Most other areas show neutral or weak values (light blue/white)." + }, + { + "path": "images/image-8-$$ _detail.jpg", + "summary": "image-8 $$
0" + }, + { + "path": "images/image-9-E --_ G[_K.jpg", + "summary": "image-9 E --> G[\"K\"]
0" + }, + { + "path": "images/image-10-Attention Mechanism Visualization.jpg", + "summary": "image-10 Top Layer (Source Tokens): A, man, is, walking, . Bottom Layer (Target Tokens): , Un, homme, marche, . Relationships: Dense connections (attention weights) link each source token to every target token. Context: Illustrates a sequence-to-sequence model translating English (\"A man is walking\") to French (\"Un homme marche\")." + }, + { + "path": "images/image-11-Transformer Encoder Architecture.jpg", + "summary": "image-11 - Input Sequence: W1, W2, W3, [MASK], W5 (where W4 is masked) - Embedding Layer: Converts input tokens into embeddings - Core Component: Transformer encoder block - Output Representation: O1, O2, O3, O4, O5 - Classification Layer: Fully-connected layer + GELU activation + Normalization - Final Output: Predicted probabilities w'1 through w'5 via embedding to vocabulary and softmax" + } + ], + "tables": [ + { + "path": "tables/table-0 Attention Matrix.html", + "summary": "table-1 The table shows attention weights between words in a sentence. Each row indicates how much a word attends to others, with diagonal values often highest." + }, + { + "path": "tables/table-1 EEA Agreement.html", + "summary": "table-2 The agreement on the European Economic Area was signed in August 1992." + }, + { + "path": "tables/table-2 Constant Matrix.html", + "summary": "table-3 The table shows a 12x12 matrix where every entry is -0.8, indicating a uniform constant value across all rows and columns." + }, + { + "path": "tables/table-3 Low Values Table.html", + "summary": "table-4 The table shows six rows (positions 48,45,39,36,33,30) each with 16 columns all containing the value 'Low'." + }, + { + "path": "tables/table-4 Color Pattern Table.html", + "summary": "table-5 The table shows a repeating pattern of colors across rows, with Blue, Orange, Light Blue, Red, and Dark Blue in columns 0-15 for rows 0-6." + }, + { + "path": "tables/table-5 BERT Embeddings.html", + "summary": "table-6 The table shows BERT input embeddings: token, sentence (A/B), and positional embeddings for two sentences separated by [SEP], with [MASK] tokens." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/full.md b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/full.md new file mode 100644 index 000000000..bf44145d7 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/full.md @@ -0,0 +1,458 @@ +# Deep Learning + +# Transformer network + +Jan Platoš, Radek Svoboda + +May 5, 2026 + +Department of Computer Science + +Faculty of Electrical Engineering and Computer Science + +VŠB - Technical University of Ostrava + +## Transformer network + +- Recurrent neural network works in synchronized serial way. +- This prevent efficient computation in parallel. +- The attention-mechanism looks at an input sequence and decides at each step which other parts of the sequence are important. +- The attention mechanism helps the decoder to focus on the important part of the input sequence. + +## Transformer network - Attention- What it is? + +- Allows the model to focus on relevant parts of the input when processing each word. +- Inspired by how humans read – we don't process all words equally. +- For each word: + +• Computes how much attention it should pay to every other word in the sentence. +- Creates a weighted sum of all word representations. + +\- Output: a context-aware representation of each word in the sentence. + +![](images/aadc6745f5528a2c9d30319197caea907c6320b6139360732c84599b72cfc199.jpg) + +
+flowchart + +```mermaid +graph TD + A["Attention"] --> B["Input: sequence of words\nThe animal didn't cross ..."] + B --> C["Compute attention scores for every pair of words"] + C --> D["Form weighted sum of word representations"] + D --> E["Output: context-aware representations"] +``` +
+ +• Each input word is projected into three vectors: + +- Query (Q) – What we are looking for. +• Key (K) – What each word offers. +- Value (V) – The content/information of each word. + +\- For each word, compute attention weights: + +$$ +\text { Attention } (Q, K, V) = \text { softmax } \left(\frac {Q K ^ {T}}{\sqrt {d _ {k}}}\right) V +$$ + +\- The result is a new representation that captures the context of the word. + +Self-Attention Heatmap +![](images/05f53829b739652e4ad3cca2ee9e35661295321a084084045027daf042ba6c84.jpg) + +
+heatmap + +| Words attended to | The | animal | didn't | cross | the | street | +|---|---|---|---|---|---|---| +| The | 0.2 | 0.3 | 0.1 | 0.1 | 0.2 | 0.1 | +| animal | 0.1 | 0.4 | 0.2 | 0.1 | 0.1 | 0.1 | +| didn't | 0.1 | 0.2 | 0.4 | 0.1 | 0.1 | 0.1 | +| words crossing | 0.1 | 0.1 | 0.1 | 0.4 | 0.2 | 0.1 | +| the | 0.1 | 0.2 | 0.1 | 0.2 | 0.3 | 0.1 | +| street | 0.1 | 0.1 | 0.2 | 0.2 | 0.1 | 0.3 | +
+ +![](images/29bf1d6b322a526e071027bf0b3a59ef8bdc5a67e9dd0c4d78830c7e6eeb553d.jpg) + +
+heatmap + +| | The agreement on the European Economic Area was signed in August 1992 | +|---|---| +| L' | | +| accord | (not labeled) | +| sur | (not labeled) | +| la | (not labeled) | +| zone | (not labeled) | +| économique | (not labeled) | +| européenne | (not labeled) | +| a | (not labeled) | +| été | (not labeled) | +| signé | (not labeled) | +| en | (not labeled) | +| août | (not labeled) | +| 1992 | (not labeled) | +| . | (not labeled) | +| | (not labeled) | +
+ +- The attention allows you to look at the totality of a sentence to make connections between any particular word and its relevant context. +- For each input that the LSTM (Encoder) reads, the attention-mechanism takes into account several other inputs at the same time and decides which ones are important by attributing different weights to those inputs. +- The Decoder will then take as input the encoded sentence and the weights provided by the attention-mechanism. + +- Transformer is a novel architecture that utilize Attention to process Sequence2Sequence tasks. +- It consist on Encoder and Decoder but they are not based on the recurrent connections. +- Encoder and Decoder are composed of modules that can be stacked on top of each other multiple times (Nx). + +![](images/6bbe323bb2ef1f2aee0f563922e4c1bc3fe4e4dab8143f342eba8640b6117b40.jpg) + +
+flowchart + +```mermaid +graph TD + A["Inputs"] --> B["Input Embedding"] + B --> C["Add & Norm"] + C --> D["Multi-Head Attention"] + D --> E["Add & Norm"] + E --> F["Feed Forward"] + F --> G["Add & Norm"] + G --> H["Feed Forward"] + H --> I["Output Probabilities"] + I --> J["Softmax"] + J --> K["Linear"] + K --> L["Add & Norm"] + L --> M["Multi-Head Attention"] + M --> N["Masked Multi-Head Attention"] + N --> O["Add & Norm"] + O --> P["Multi-Head Attention"] + P --> Q["Add & Norm"] + Q --> R["Feed Forward"] + R --> S["Add & Norm"] + S --> T["Multi-Head Attention"] + T --> U["Masked Multi-Head Attention"] + U --> V["Output Embedding"] + V --> W["Positional Encoding"] + W --> X["Positational Encoding"] + X --> Y["Nx"] + Y --> Z["Positational Encoding"] + Z --> AA["Inputs"] +``` +
+ +Figure 1: The Transformer - model architecture. + +## Transformer network - Positional encoding- What is? + +- Transformers do not have recurrence or convolution. +• Therefore, they have no built-in sense of word order. +- Positional encoding injects information about the position of each token in the sequence. +• These encodings are added to the input embeddings. +- Uses sinusoidal functions to ensure unique and continuous position representation. + +$$ +\mathsf {P E} _ {(p o s, 2 i)} = \sin \left(\frac {p o s}{1 0 0 0 0 ^ {2 i / d _ {\mathrm{model}}}}\right) +$$ + +$$ +\mathsf {P E} _ {(p o s, 2 i + 1)} = \cos \left(\frac {p o s}{1 0 0 0 0 ^ {2 i / d _ {\mathrm{model}}}}\right) +$$ + +- pos is the position in the sequence (e.g., 0, 1, 2, ...). +- $i$ is the dimension index. +- $d_{model}$ is the total dimension of the embedding vector. + +Sinusoid absolute positional encoding +![](images/ed82c0264cec9185992b5b44d52a77d7ff7b87ae92f39cda602dab0ab109651a.jpg) + +
+heatmap + +| input dimension | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | +| --------------- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | +| 0 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +| 1 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +| 2 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +| 3 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +| 4 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +| 5 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +| 6 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +| 7 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +| 8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +| 9 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +| 10 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +| 11 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | -0.8 | +The data is a heatmap with a color scale ranging from ~-1 to ~1 for each row and column of the time step axis. +
+ +\- Sinusoidal functions are used to provide: + +- Unique yet smoothly varying encoding for each position. +- Ability to interpolate or generalize to longer sequences. +- Simple mathematical patterns that can capture relative distances. + +• Alternative: learned positional encodings + +- Positional vectors are trainable. +- More flexible, but limited to sequence lengths seen during training. + +\- Without PE: + +\- Model loses word order completely (e.g., “dog bites man” = “man bites dog”). + +Sinusoidal Positional Encoding +![](images/19c35a0d92f60fa563b8877e47749c0d837e6f2d39f29b9394783080b0695d34.jpg) + +
+heatmap + +| Position | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | +| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | +| 48 | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | +| 45 | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | +| 39 | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | +| 36 | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | +| 33 | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | +| 30 | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | Low | +
+ +Learned Positional Encoding (simulated) +![](images/a70d75abdfe0004c118abcc80faace9771e7ce95602a2c74deb170af66018106.jpg) + +
+heatmap + +| Position | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | +| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | +| 0 | Blue | Orange | Light Blue | Red | Light Blue | Light Blue | Light Blue | Red | Light Blue | Dark Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | +| 1 | Blue | Orange | Light Blue | Red | Light Blue | Light Blue | Light Blue | Red | Light Blue | Dark Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | +| 2 | Dark Blue | Orange | Light Blue | Red | Light Blue | Light Blue | Light Blue | Red | Light Blue | Dark Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | +| 3 | Orange | Orange | Light Blue | Red | Light Blue | Light Blue | Light Blue | Red | Light Blue | Dark Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | +| 4 | Orange | Orange | Light Blue | Red | Light Blue | Light Blue | Light Blue | Red | Light Blue | Dark Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | +| 5 | Medium Blue | Orange | Light Blue | Red | Light Blue | Light Blue | Light Blue | Red | Light Blue | Dark Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | +| 6 | Medium Orange | Orange | Light Blue | Red | Light Blue | Light Blue | Light Blue | Red | Light Blue | Dark Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | Light Blue | +
+ +- Instead of computing attention once, we do it in parallel multiple times (heads). +• Each head learns different projections of Q, K, V with its own parameters: + +$$ +\text { head } _ {i} = \text { Attention } (Q W _ {i} ^ {Q}, K W _ {i} ^ {K}, V W _ {i} ^ {V}) +$$ + +\- All heads are concatenated and projected again: + +$$ +\operatorname{MultiHead} (Q, K, V) = \operatorname{Concat} \left(\operatorname{head} _ {1}, \dots , \operatorname{head} _ {h}\right) W ^ {O} +$$ + +\- This allows the model to jointly attend to information from different representation subspaces. + +Scaled Dot-Product Attention +![](images/e10a7f36a671dd9b101cd518e9a3ca79b90e1be5a3e8b23febf47b4f8144e078.jpg) + +
+flowchart + +```mermaid +graph TD + A["MatMul"] --> B["SoftMax"] + B --> C["Mask (opt.)"] + C --> D["Scale"] + D --> E["MatMul"] + E --> F["Q"] + E --> G["K"] + E --> H["V"] +``` +
+ +Multi-Head Attention +![](images/e0038bd5adb07e91faea8279d7634575f945d1929525720d3d2b5fc366650e98.jpg) + +
+flowchart + +```mermaid +graph TD + A["Linear"] --> B["Concat"] + B --> C["Scaled Dot-Product Attention"] + C --> D["Linear"] + C --> E["Linear"] + C --> F["Linear"] + D --> G["V"] + E --> H["K"] + F --> I["Q"] + G --> C + H --> C + I --> C + C --> J["h"] +``` +
+ +Figure 2: (left) Scaled Dot-Product Attention. (right) Multi-Head Attention consists of several attention layers running in parallel. + +- Multi-head attention = multiple attention mechanisms in parallel. +• Each head uses its own learned projection matrices: + +$$ +Q W _ {i} ^ {Q}, \quad K W _ {i} ^ {K}, \quad V W _ {i} ^ {V} +$$ + +- Different heads can focus on different types of relationships (e.g. syntax, semantics). +- The outputs of all heads are concatenated and linearly projected: + +$$ +\operatorname{Concat} \left(\text {head} _ {1}, \dots , \text {head} _ {h}\right) W ^ {0} +$$ + +\- Encoder and decoder use multi-head attention differently: + +- Encoder: attends to all positions in the input. +- Decoder: attends to previous positions (masked). +- Encoder-decoder: decoder attends to encoder output. + +## Training a Transformer (Seq2Seq model): + +- Uses teacher forcing – the true previous output is fed into the decoder. +- The target sequence is right-shifted by one position. +- A token is added at the beginning, and at the end. +- A causal mask ensures decoder only sees previous positions. +- Loss is computed between decoder output and original target sequence. + +- The target sequence is right-shifted to prevent the model from copying input. +- A token is added at the beginning, and at the end. +- A causal mask ensures that each decoder step sees only previous tokens. +- The loss is computed against the original (unshifted) target sequence. + +## Inference – autoregressive decoding: + +1. Input the full encoder sequence. +2. Start decoder with token. +3. Predict the first output token. +4. Append it to the decoder input and repeat. +5. Stop when the model predicts . + +At each step, the decoder sees its own past outputs and the encoder context. + +Transformer Inference (Autoregressive Decoding) +![](images/6779eee546f4cb9f7c3e11a6a716f9201dc11f5fbfc609650259170013c3b78e.jpg) + +
+flowchart + +```mermaid +graph TD + A["A"] --> Un["Un"] + man["man"] --> Un + is["is"] --> Un + walking["walking"] --> Un + Un --> homme["homme"] + homme --> marche["marche"] + marche --> u["u"] +``` +
+ +• BERT - Bidirectional Encoder Representations from Transformers. +- Uses a Transformer-based network with pre-trained deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers. +- BERT makes use of Transformer, an attention mechanism that learns contextual relations between words (or sub-words) in a text. +- Transformer includes two separate mechanisms — an encoder that reads the text input and a decoder that produces a prediction for the task. Since BERT's goal is to generate a language model, only the encoder mechanism is necessary. + +- As opposed to directional models, which read the text input sequentially (left-to-right or right-to-left), the Transformer encoder reads the entire sequence of words at once. +- Therefore it is considered bidirectional, though it would be more accurate to say that it's non-directional. This characteristic allows the model to learn the context of a word based on all of its surroundings (left and right of the word). +- The input is a sequence of tokens, which are first embedded into vectors and then processed in the neural network. +- The output is a sequence of vectors of size H, in which each vector corresponds to an input token with the same index. +- When training language models, there is a challenge of defining a prediction goal. + +![](images/99b79e7ff6eb2a1ccc1fb87ceb9ad2cd1b3c8b461156899d828b0d1f50a56fef.jpg) + +
+flowchart + +```mermaid +graph TD + A["Embedding"] --> B["W1"] + A --> C["W2"] + A --> D["W3"] + A --> E["[MASK"]] + A --> F["W5"] + B --> G["W1"] + C --> H["W2"] + D --> I["W3"] + E --> J["W4"] + F --> K["W5"] + L["Embedding to vocab + softmax"] --> M["Classification Layer: Fully-connected layer + GELU + Norm"] + M --> N["O1"] + M --> O["O2"] + M --> P["O3"] + M --> Q["O4"] + M --> R["O5"] + N --> S["Transformer encoder"] + O --> S + P --> S + Q --> S + R --> S +``` +
+ +- Before feeding word sequences into BERT, 15% of the words in each sequence are replaced with a [MASK] token. +- The model then attempts to predict the original value of the masked words, based on the context provided by the other, non-masked, words in the sequence. +- In technical terms, the prediction of the output words requires: + +1. Adding a classification layer on top of the encoder output. +2. Multiplying the output vectors by the embedding matrix, transforming them into the vocabulary dimension. +3. Calculating the probability of each word in the vocabulary with softmax. + +\- The BERT loss function takes into consideration only the prediction of the masked values and ignores the prediction of the non-masked words. + +- In the BERT training process, the model receives pairs of sentences as input and learns to predict if the second sentence in the pair is the subsequent sentence in the original document. +- During training, 50% of the inputs are a pair in which the second sentence is the subsequent sentence in the original document, while in the other 50% a random sentence from the corpus is chosen as the second sentence. +- The assumption is that the random sentence will be disconnected from the first sentence. + +\- To help the model distinguish between the two sentences in training, the input is processed in the following way before entering the model: + +1. A [CLS] token is inserted at the beginning of the first sentence and a [SEP] token is inserted at the end of each sentence. +2. A sentence embedding indicating Sentence A or Sentence B is added to each token. Sentence embeddings are similar in concept to token embeddings with a vocabulary of 2. +3. A positional embedding is added to each token to indicate its position in the sequence. The concept and implementation of positional embedding are presented in the Transformer paper. + +
Input[MASK][MASK]
[CLS]mydogiscute[SEP]helikesplay##ing[SEP]
Token Embeddings $E_{[CLS]}$ $E_{my}$ $E_{[MASK]}$ $E_{is}$ $E_{cute}$ $E_{[SEP]}$ $E_{he}$ $E_{[MASK]}$ $E_{play}$ $E_{##ing}$ $E_{[SEP]}$
Sentence Embedding+++++++++++
$E_A$ $E_A$ $E_A$ $E_A$ $E_A$ $E_A$ $E_B$ $E_B$ $E_B$ $E_B$ $E_B$
Transformer Positional Embedding+++++++++++
$E_0$ $E_1$ $E_2$ $E_3$ $E_4$ $E_5$ $E_6$ $E_7$ $E_8$ $E_9$ $E_{10}$
+ +\- To predict if the second sentence is indeed connected to the first, the following steps are performed: + +1. The entire input sequence goes through the Transformer model. +2. The output of the [CLS] token is transformed into a $2 \times 1$ shaped vector, using a simple classification layer (learned matrices of weights and biases). +3. Calculating the probability of IsNextSequence with softmax. + +\- When training the BERT model, Masked LM and Next Sentence Prediction are trained together, with the goal of minimizing the combined loss function of the two strategies. + +## Transformer network - BERT - How to use BERT (Fine-tuning) + +\- BERT can be used for a wide variety of language tasks, while only adding a small layer to the core model: + +1. Classification tasks such as sentiment analysis are done similarly to Next Sentence classification, by adding a classification layer on top of the Transformer output for the [CLS] token. +2. In Question Answering tasks (e.g. SQuAD v1.1), the software receives a question regarding a text sequence and is required to mark the answer in the sequence. Using BERT, a Q&A model can be trained by learning two extra vectors that mark the beginning and the end of the answer. +3. In Named Entity Recognition (NER), the software receives a text sequence and is required to mark the various types of entities (Person, Organization, Date, etc) that appear in the text. Using BERT, a NER model can be trained by feeding the output vector of each token into a classification layer that predicts the NER label. + +1. What is a Transformer? by Maxime, +https://medium.com/inside-machine-learning/ +what-is-a-transformer-d07dd1fbec04 + +2. A Beginner's Guide to Attention Mechanisms and Memory Networks https://wiki.pathmind.com/attention-mechanism-memory-network + +3. What is Teacher Forcing for Recurrent Neural Networks? by Jason Brownlee https://machinelearningmastery.com/ teacher-forcing-for-recurrent-neural-networks/ + +4. Positional Encoding: Everything You Need to Know by Darjan Salaj +https://www.inovex.de/de/blog/ +positional-encoding-everything-you-need-to-know + +## Transformer network - References ii + +5. Luong, Minh-Thang, Hieu Pham, and Christopher D. Manning. "Effective approaches to attention-based neural machine translation." arXiv preprint arXiv:1508.04025 (2015). +6. Devlin, Jacob, et al. "Bert: Pre-training of deep bidirectional transformers for language understanding." arXiv preprint arXiv:1810.04805 (2018). +7. BERT Explained: State of the art language model for NLP by Rani Horev +https://towardsdatascience.com/ +bert-explained-state-of-the-art-language-model-for-nlp-f8b21a +8. Understanding BERT Transformer: Attention isn't all you need by Damien Sileo https://medium.com/synapse-dev/understanding-bert-transformer-attention-isnt-all-you-need-58 + +## Questions? \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-1-Attention Mechanism Process.jpg b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-1-Attention Mechanism Process.jpg new file mode 100644 index 0000000000000000000000000000000000000000..080fd2bfc44a36e3003398df72bbf90d0db5aebf GIT binary patch literal 21937 zcmcG$1z43!*FSv2rW>SdBOwivl5RpkKthHv%%$na?x0fTAGXQ~tth_9MgaiOc;6LDU9FPVu z(9qD)P%+Ta(J?VGuyBZRaj>y*D2T4%6Vp&aX=x~_sp%NGnCa*_7^taPZn1LQ;2^oTdiiVDXiG>X|C?x=pksuIc6bLFR3JTcT8>|OV2vLcy^GKpy zQ+e|lk-u}Vi(ecUYm0U;w z zKDBL*~={ zU;#b`>H%r+EXj-$t98a=#Y8T0ZO2-XI48i=G!kfAj?K8JcDMv2Xq5M!Ujm_hgo(rk zQQtaQkk9Pg9Zxol+BC-=v0egp6Yw5%_4AoaVEzsKw?g#OK;TabYCo$rezc8t2?Wyq z7@+r$lI9o-^4r6M)(gNH_6;8)D)-xGZHpu$RfO zCdh`e9PMW3yk=%XkLjMb(QQjE5ULCskOk#EM8_&mf%M;OYTaW@Z1E9TN&Oh!5XfqQ z#A=b6gdE87>(OHYQa{G{5C!Tf0Iiq6``9By%|_k6@Y_(#DnVi?A8uy>#Kbv z&cKv4GQPCW*+OmRCae4CJIzkfONML}0|#NRNd-N`Se+_fcxSZ zKb;6S%5luHZf^lz-l9yV$NH0&V73u!|Ox0!?LTxvKXxxD5VpBNtyp7)2Fvzb;4O}IN~MNd!9*!cio512}b0&1s>hoOA6(W zWcj|h$9I_S^-D;{MzW(QgY~*G%E9P;m!o?OI%w0ybln~2cy324?)lb@iYYqh_jhDU z`*gmr#*BY==*Y})=||+!o;+S%V~=&Nvd2A_BmGc*30U}@d7M0Rv^(dCJ2QJ1rdhZH ze^ttU349t|JB)e|^->0210NNq+U`0=JpV#7uKo3jd}GG^<8h(euX@fXD-FH3msmDx!R89QEC2T@uN!*`3 zFPA*gPU-`)t8Se?TZbK-)}1Qj>Sd3?ew2={N`O_b4CnRNGMr2RClmOKSXEI1Q-)Xt zNl{(~Ec|>-E`hI|o|k~3%k(Mr!PA`SF$1RUx}rPL!H-yz&Bi}VGASgg=vIi)Ov6PBj>FfuCWx;PwOT4JnLgeuRgUSuC3e1;Q`cA*Gi zxv>KH3aM2#q^)+0Xb%ukL6U+(~}W!v;Bq)BT{=8G^4+HSlP_j(=R&{?QltDJHIp z)$Ve8g}}2$j~f|ITm>aT?C)F9-*aGnb?9ddg;N$DR!^TIPhA4jZLs4-sP*SdAc^zn z;{0l}Re_}U+i>9)#&(QJTYcF$YRc$%*i!&yC8rT*?S5#^wC~z&2{F8r#t@%^c1JzV zXHsGT;jFf$mC!P{$=tgY!{2h(LRmfRWOYmZA|F&j`Sj5Q^!%w0g~LZnZI2gvFIzrX zb1n$+eA{7lZoNlaN4k%J%<=&2SIMy8eRIGSZr+=4pr@zzFmz;=qXE$a|4iaV+ zw&7a0k6L>P)c!ayyniuwrBMkYeI4D=E(d|qRD#YY<4bPySf=(`3|R{Y2RS6lSN+gH z75(b>V`jmb0B2T$j4eOx5`b29Up(VRL^a;KCE`2F@80OLcOwA5$)9;&z+SO9rjNpK zwm097iBCa&^SMtF&LLJd-!#QMKaxivOX|&!fh`2;?E~7^`j5zul*b4uP=Z7 zg)lT-0=wJ>@Os!jkKWuTF(bISWpXD;*91X3*EFYTII9=$GH}prGd8I|HJ*D|+$_c! zUvy{o&HQse7G$fxj^JN^Ndp!10G5w|0PBzOV%c#L}M5+GD_b0#}uc`$!e&MVee?((R$?o&Oj50YL9QrJcHDTF3DmRJ z9S17^UfU}{K0RLcPbI9}rmh-XP(JGL9g<=hzyxlm3wPPO1b}pKVY0%1uP*#|H49>* z2(ht@K)QbkG--7aBP~yf7nkfP`EkzZN4VnzZu9BhLw|4+#(dB50J+?$x{F)BZ=>YQ zVhcrZCt&yj%O=hh#;iWPlOGxAN*Hoc`H#41e^72F_6N6?-x>RJu~WmWp)0eEKnVPDhpQ5@sB=!3OxFcKK>jQ6!;IOb6nn*4fj%Tn#*e{4UM_0 zugIBvw>mJHYWd{m-{MxUT@iZsaqEr`g!Oi+@%+WS1SS8EeN%E(g1AQX5}?CYK2k=! zuvRNN|Gnfx!QsbZD2wt!tqI>MD|PLY)u}*C27$#~wXeBM6q4U3^f;(T!VyA@3$`8w zXWKJ?-YXf5swW-p$vwOTMxGYmInS9ju75aT9!EQ)f}V+{hBR^0gZd!ZK!+$7gPNLh z*ISBR5r$(;%GEM`uVhDzCp(3N;Eo=tam87Zok^?0FsX-TSWt4wc9njrE9K)apZ&f* zr-0bMX~MfjqyrfU?*EkUfwx#M>W?p-P^!?aT~b?DTjbjw+dd@h*w5kHBJM_Cb0VOK zc@Eu1^ifrqR_N!TT10$#%<@5>A@j@3@@KfxM8ErjPBizp!ofF`rv})e(*$xTLM1T%CE!a$+bwn z;Mku~!p@$r`5a%x(NL@Efo-}4rPoo6YmDt2sfHr!`?)|qoEHt4%fKf2zEFlWX`>9S zKwglP4b9 z@m7UyPaE$HLces`#AfxSvQ;^n^B`9~B@1aRgxz}JnkmMa{sle2$vW@$dSlrRKM{J6 z()-|Yhc`2yU~c|j+L!Z4A^kcp)#L%Z)It3XDQsM>WtE6>NtFBSHgpRtjPuihOJX*h z)9TybMUqvM+nwgVD>z((#dsxMl*M$T`#C(g=!%KrbsHMK=Qn0cN+WDW*aDFx8Ii|i zK?gQ4ltidqlkkot1~w181?)DzlXiY>l=X99Wz!aNan$4ZX4Cdd879d&8R`g2RB z@kW_KQb{{Gcsl9)Zzjhjby7QvIp?FR_HP#CZ`EL0ZQ+eqma@dVVPhKc-V}eaiQa1g z8GOoD&LHk^w}>H6&XT{q}M|jiN=d9KQ>}NDBjsh%qLY%Te2u_U*!< zH&2c|`kQIN9pr`L_eZ*Vl216^5gbnGGNrZ}eK5mzj0F{rAwoJvAunF6I~G$+o9LRZ zA8_99?RseQsx`)|+$+(jGT3uCan$m)?RKZukPRoe=Q;%Slq>MDF~`mN(WWe-%(yOR zWi1wMr={1;?9J|(6)U>v1ev3>LnD~K9j`r=V@^>1ObY3%Y2qcuOKW9v>=LPIRno_$ zYSOaVUOBBYvQonkpmIkaXK>{h33sg2j%K6RrO}_u{c`%O@&TjN?iNJT{$ij^H_22P{7wr$-LMQ!)2%cjzZObH-Osps+5LNYwTZm{YVe z2l~&e^GphqevWUR$V^Tpv_TfzZ#RQ5_n0Ccr6BsQ2BlfPxOjC51SR2OBnM)sk(=C| zlflXiM9!q~p+S|jOxMw=2{V6eYq#rt=+7iI{Q%FZ{4o0H0@#9WeTSVJK*zK%fqh8z zB~bV264-GnMnOfaUE-DY5Q?6sXlRSPq6(_cBhxuima~|&|yBnwNRz8krelUYO z)>;#hFwCU@ZUHKfXbXX8yabwiFH*e0E5Cl-*%|&NKqf)DR9I3OE1Xg4r~Bv<(Apw5 zQ%)9VW$gc+UZO8=voSULgp@RZn0)nR3tRTu*QjQir9?-IqEB~Xq=H|D$uQ~6B#e@G zQ?!$aZ=C9us0O4ZxH~R3VxYdm%ZLOZ43QEGc&REVIp$Uojgs)^$vBm;p}s!58)MW~jVfErX}ECq^baBfu&o8?xl7oFXx$GQ zpm*%z!`+3WO;Ezu64^436PoNxJ2T**tJi2{`>4!c^eF}pqw#h0%vOc0iwsv#N1W1# zvnzWU%n=>2&pTgH1TeL<)pCPH{P3kEu4QMlXsBeT07yQld~GitP=&FI%~j)kklm>K z#y3o{q1(I8JCPLAuTYaPSs7~F%i^#G5fVD}Rr+NX)4$9Ddq)ztirAkqM3MA5F(AL@ zwrM45d%L4+!Bbz+jaJFvV*smtKDvPVnZhN2@NtD7ok9^@QD=%OR|Y+Of%s`#b2xE{nIse!F)|4moSmkFS2+&;3Zi=uA=ds`K#A&Z&t0 z0r(2U8C+_!;108bJZ2d~Xjr=7BG~RU?cw}dIx)vYz^bUO2VP3lmxIE()w+`(B`oJSm&D>Vn9LO%BY4TW+#$^Zz#D>=f{g7pa{Yt;Vn^! zb8u3Uu=lTg9GA3IaS#c51z$k2My!I6PY4Hem*40T7>+JSyb`CS0ej#UoPh&K@k?F; z3^JF%g}Xj{#jzhEvTQVDJe1O}E)^*Eo+*a(>vu$>(h@4$G1`gl@Wj+c%UgH(r34R>#Jxagt zajic2fhcp0?YOb-K$}Y>6?WXVBziwuPGiC7jX17S_8j(WRT5wFd!t=$H|{+V66#JC z5*rrB^Yx^CZ&UXO*?9+hs*XL8oo24LNmH%4me(dmpKXq zP7(J6&kJf5bXBy{Gy{XYwAu`FfzB2t9g}$74#MzyNSw)M2?TxL)Y!%Mf(>r7qSp&C z(+FAoE)MdnsP<=7#?na3L#)+i!QpFLfn&jwu!kk5}FO#y6d( z8-CnDpHLVd7DrtI(#qRO zCeHQ*tFbpn-My`G8)Xey9=~Ci54(4oz_+eqS#>(uz{M4kRsu8j-+5h0m4ZheU+-Hg zD`d0ReNRg(>tU8;uJu#>Rc6&7t&+Q{O_?9*vpY182wc2IPbAmea@)_@bJ|VUI3wQk zck#klQy<(r+%0Y=JY;}~3n$6RSSvLf3wa1k_7*&V^?f^Jv5L|5V> z{ky8s2J&||h&uu+TvaD}T=vFlf%s*&+LZcv3lsO)} ziDG;}ev}bPzjOT(NN>Fa2oVd3;zy*RE+_G;uhEmmbeUAP=)SHx?I;^633p^cFd}m+xSU~ zlAN?I`|O1^IPpRE{8y|9#`N_%-J)b|(>2Z-G#smaj8J9W*YC+aq^(X<%qokgO|}~j zY`%BBWux9SER#B7sDwSdu>U|`nUu7sw7RrEts+LhDb^y@@80uQ0p=c+Hbg=+-Z|-f z4=w>qjN9#qpoO=$ZX2JSv~+#;aEz7a=+$l5Qy1}4r(q`JyeU_HCUyS0j_Sw>#@!YA zIY-uqkHc-~z*q?NBlCJO#RdA}eGu?o;7J9~QEr5fyinTjxHu&q6MNWJ>43v1Nrn@Fzgy;x|>2`vzStE$TK(C_55xn!)gKi0`=E;N0rNC4eaho?k1%Ae=Cx2M>X6zj1>jp2ylFP&1Bu zAT=9~2rvGvb6hw=K+)dPUEtJk`khKs?x!!99N9{|%>fib|?7DRcAorhJAnlGBjvID68XHI0VOSc+2YSt*PLM-Fk^OFZ;KeP9 z{WcEB_wH_+P|)XqK((_U1F-GbW%fmu(~p z&UPXv$SDMR352M_W?8nuvkjep`W#z%C*|L|1ZewJ2WXjU1$)Nj=-YZFzHVGsqj=hc zgX7~6>M)f4xws;7jjB^vm5W|Wswj(pcjfb{>@YFD9LSFqtI% zqvx+l`q4A=@1BW2r}O6+z2Iakzp7giO^44sfX_?d?>RTNj+87Oo$1Ftqt)qNR@b#e zfcFJQ5UqX7Es8ppdl%b>_X)BOx_D*#QVzU9eRk^{WP#lqD5zKy{WlT-P}CQwzw>0c`CrB zdkU~?Uv2v$Bm@k83}gjVWr%-1it-Yu#3gV3`Qm2q26{@K=g?OqrhEy| z9}J&Ez@w9-VC+XIk~H!Pdu2&W0%2A?K!Mfb*N4AgQnr{p{+^M!$wNhs!iIw2?^Cmi zNE1HtPzAEaDjn4&z1N8~QH2rXYw}%yL5hGnf^dTY;kyl==7%?kr^0(~ z8ezwBC1I!NUoHW!9@HwnU-W4Hp$FpGAiYV7?8F}=@IewO|1g7O7YKSi7>5zy&0z($ z75fi664V(@>jr8@cxm6F9YjiakvMp6oi;rJ<|}c__XPSaT;6JlFr|Gy=$x2LHTf1j z?qDvK;Eg0i+W_w0|K|H@&~*AmlK)i;;hj#J%Ku0K8t6}UjzM;Sav9L|{z=a3Z#7(z zOTvDm3HzrP=z{*H)}`RV|4cEK57g}6T37tnaDQw4-@`d+$!4VMuX;GQq-3y`B@79E zu(6R+AU4h6mJa8MGj%4(St>{aREOa6Q{#v_PHX($K=!YHrMYQu9I# z4w~0rCDX8THxt1h9|uVfFKdZ-XC_Q7$*6fn&>0jFz+Mo45ii$-5xe>KH{GTja_YvL zgj*P3^D&f0>=E3Ut-Mu0)v?)dkj{`-Jm|C`CC^Y)S6b%RgW|D+OY|OP`Rk2Y~y;RU?zIB?P9prJ*Shhhx&1W z?FRH;5M;VaLlG2W6{<9c z4-H|YEc@W-G8Aq_S~QV4(bLLX?7Ln7Jk|_NjGEr`l>YDDM zjdWbhw3dl3rAc-!POHLxH_24l`&k;P)!g~o-Py)Q`bBCYBo5&d3C4&OTvFtX zTJTcxMeJ}aKCY*pk3ZsDZeiGAfY>vEF!a!U;d61<+`NHD?5eD8_p+8~UitubYvXuR z?rr%-5;a91=BVlKtKY2+T~~+!Y;2oUL@^ocS!lk}_{9i)lMok9U*)kZJ{Q~}eM_uW zko^U`xqh(8PB#@@9F1NR34ZoQ2M;l9*w0n+a$x9m)E3Kcm%p=&S3*>`*vFmZp#T*H zr%9D!r%QRXUn^Tj8g-Np(l<=zz&&xTxrXI?+wn_@?&&w8?7o+)*U);`sK0yrFk161 zdL*whO7MWAj-Fz&0l!FEOq#i8{;0h^l?iT@;68rwlJU4dCf&U7$1je3^Fv?|FugKJ zy*LsBk12v#tuRqmxc=PwMt5_#wH|6&M#|?D{-QetxPuMJuXPLkTb_)ph{mQ@dB+(M z(PHFXzk8#%n9-UXYtf9nZN(w#L0Pn5!Zc1vkiL2OneFF`K*N@IdkpZB%%qnbC9e3Y5f?@}C*)Kld_ zlm^8#p@E36++1APQd}ye2_r4Zq#PIDX&SM=S@Lu8$~%Xlf63E7d3)2su~5EcE|-?z z7Gwev24c*;wwuP)uY|^xaLZJfWnkd)nY5e73A&wrO4`PDmaAoxs^SITlt5<6KQw^jM(Mw*h z--T(BBl&BD;^9KXk(YJ)Ir8UAss?Q7D&NvR;QgAtwd$jEUt~|K-|2)Q^hl+&EK<3m z_+F;|3g%J;xxJWphM`-+a5GQCaQUro>%_hGkxR(LcNoY!jU8UDNUo}J4*90LJ#U)c zj`offwxgk(-mrjqS}$f|(#i?h)@LsB>&SUJ!p%*S^YdqX6=KkIk+*rGx*tZI45(X+ z7*t8@t!X)M)}s;aWfDPp@|9sP-8Z8$6d9-SH=Ki>?lPr~@;u<012kC3X=iNLd%{#z zt+v0I&*MJj1aYN1{dwtr#HEz+5fO&KJ66Iy0vy5bnakl8pJ<~zvn@|8Wb10nN{Ohi zCp^C2I?fy$zj-}$GM_P)-&fl`;vFjXu<&Y_#~xM1_kDfu8{gf6l&r_SuRp}h)#sIS z#gj*MCD1(n0&!T4=&QcK(>Z|Rj3}IXjLC>+^AtZO4<>=$;OTK%mH1kReVkPzWw)tl z*{du2jb3N7{42?W{8ch?6^Dp!g~{T9?*imgid5^?Z=_#YFll3Xw~1-T%9&7}bNOkM zpiNlGl%fX0Vig}(+0M*I0yN@Cc@|v!A<`{F!?YCEmYj3_8TuzS^}C;3P86EmCc~TX zlFWQ*N^4V0TzLY=4yP;yF|p*>3dcMIkygB|x}GNDm7Iy%LsrTO#+sr!A5%oCJohw3 ze(I5k_>`X?{&kB0-d3IsA5H~t1#@A>ISq^w=VXoID$3V< z;$94TM@w`3ZL&kwVLFeLlxIZu3-U|zK;nS*a#TMb;q$E6mZU1BAn(i~GJSp5xFiH; zwh~NXb{~Gt`YlA`XTR5|XBk;8+AdXr~Jc6jLb)WJZX!yPW_>=+s<>Oj3kPP>E8%FSb?sd_I2hbV2E z7>vvDc5`HNJLF+33wy(fb{5>PowfQ?`pyRslOlD8j_r$K75MKSkN49m;ScmtOkm8b zB+>hPzA2*iLdMKEo1!XQSRiC>LupxdJ3TFPH2O@YMs$9HmoH*K%f!(7^hRE4yc%Na za9FM4j6Xuoe(@qfDtP_VLX6Ry=LsHAD`2h3;ROq!R6-ISuC%SIxdY?2UN`pV;*+PT zHkcuWaReeyto`3MYdcuG$d2oXysSA59BZev%q?3BZk8P_^y_ThAy9H7;YypMk!tB| z9wzleO=v-2BwdV9;Cxzk%{u9ZDJ!@TPN}i6#jquj3NzxBDk_!@aOO=qb4uEqnNQFd z6QYnqS%A3UfUuXDJM9OncFm_}x zk^9s`rTy^d)kGNPJ#vf}OwG}1#`=~G`ycL@G|03h#pYu24;`Mh1$6}Q2a&ql#%(Oe z4I9kb$~d!AEZjG>ddM_hnRou4e#R@m#&LQxv@V2BaEd>Qs#hPKfkB9W4NT#SAS8H5 zqG2usAudrt0uz_eTmxVW`h&3bJ}}*#ekhQPk5Q-Hx|T`sl;M zDaM&G>h^KZD4@eGKgwEG$KB$t%GVTX(veEx4b7D8uu;3j?;=BKmJb?>2Ukr%xSS`1j4$T$uwqDz%se2QoACR^zyw3ekY zY)=PkiVCR&r_t*RDxRoV$eF0b0J~LrGPH8d8qP2j9Hs}b zzpdH6N5(sqE0UgUY4U7Iu~Ucw`vq) zzt%0N1<_kt(_BR|Jcl;hR&h2&&g|fT0C_39{#l-=N+<*V7-t#HPE>quelA8gLvY~O zM<3y1uj>!-63s1&9~73?l5Ow_@!r457~>fFR9UFi=26CwmzaQPr}dfjSqMy}OzTc% zIhMR^i)yql4lmN`7m{K{3F^U&7ep;%C!D1r_qpu3NY@k#cyJPoRXz)D(9K8Fh1~;q zJn!^G$$56Zs!rHFJQz|8dhO@8sKMq#3;>W%c~<|VF@YIN5YZ;+F6Ua}!Ri&*H!uyJ zOqG9~Oe(0}e;hqSw~H?U+QaFC6Z=abMXT%&LVxE*<#V@^_83@E>OGJoZljNT836$t z$%F*Pw6D)YPp~W#4_ZJdNGk9JI_E`1I=n})`w~FvKWD8Q6)_*neRb%ZK)H#^&}JwZ z@l-@}&HlVr2R{5Addg!N@V#!9x~Fah8-&2v;4jsNykH+&lh^QKZztrI^$|3TBB5I@ zr#X>GgTPog8W`j-EV?>Q!#iROX?rOfCPPCqh#slQmnylPf~#>m)WT>V6~t;tU=4XA z@HUZgL?!nw$s->Yw?{&=y>30`HF{Yb;U0y&liE(E7Fhn+^a?tMWmZA7eT1ur{E zXNNmF5;QK5VkXs0XN#-*U3_7Oaz63}fEQdN#-6w14HyY{kr1`2Y zrP{R+QgRZ==QnBHvmSm7La6kd(-+PliYl8YQ1N~taXZQyyf9M@z_(dJ^zLKD4s9h< zN#^Odq_#F`_6LGu(!ZYKrOg1eJBYQ$|0Nx$0*NWObBV|5Lw2VM#)s6iN{ z0*fs`;Fy6i!R(7y=g;%t>w)m=TI>vJuU^g*tR$$I8NbzL!!CT}AQmsbkmxazn}`Q* zW4Z*^ch@fgxpuvQe`7^gWu*8y{1egx<UzJ{*n)oDXg;PRxmQTf&U5{P2v0CvcpJnc4d22} ziTRQ6N1W|Xakvr41;<0szdKtzPBFpBoj$mt{6`6>mQh2Df-uwy6UYO6 z_}t>p67t4lE0?!hS$c*et+uRztdC3+%)++b(D{+nRh|xp{WenAMhG~n;LlN6myJLK z@EN#be0SkHv00gV)<-?f(2@P7-Q_@=i!$l>P%7wJ^*l+zMKQQKdK9`5pze-)|0wk% zCv3nGzNhvGbP!Q(vBDk^IP1g?vj4Oiw0x&`^UUcL1$p;rqDTLqEdN=8ybF>To_~Gq z`>EwvlKDm?d?(xWkhTsCD~#DA(5~p_r2Xn)`MUxbeyM|y^ts9H_*ACS?zhYCk!=k( zI8wZVk5PORKd-PWWWjA#;rm43I;vYo@c&WpH~9EpYLQ$5)BcdP?~l%47H!H!X&U~? z>(O7cro7Pw+Ix)AF+t+I1ldaVPYdAqVF6(qlMD#IGT5)W+)H3%68hH{a)*auV01x2gHQ;+)a5m)|u#=_4=-Yw7LqCQ~ zfFlZVLY!#+3gOlY!n=Wyfk4>zZFo1BowXGqHwpr~2|S=w^x&q=aK;baYk{3DsCmcY zpUguqcG>h;kYUB(1y9Tb1b}6jN0a_)p7vXF)k6!|axna;Eel+k%GV^Zjy+Za!l6+g z!?v)7{-UX6{~CYmzFNOY5}Xfu*hWa)mG zaGZpT#rF1`;ZW#oIAdZObOM^MAm@u$M+NZh+i(Pz4_B{7-7M`V#K9H<{jchS|6aWT zR@eRA8$L?|_CETj_ixqUHu8V3TG&1iI=16xYgbbJo{%9NJ0O~*Jc|K6Tp2esA-&2L zi$)3;uUA0vxdVLBKfVpJBnG<3!SmZ0xBT~)sdT0yR_vF-gLCx{2Kx!YH3TZ?9~r=q zgAMGnH@GG3foG1Bzu`x4DjGx|UaW7O<0pbyV_573B5b}3v{G)-B(T^t3bC5!vEbzlF_52aBb~E$@zw=mh#(o8sSBorrz&WNA3Whtiv^R@y?t(A zMLj99TTh7RX`?AERQ6Q02+?`*tbi@liw``i&Pa6Z z1dP;ZE6-kusnN(b$#M7}PB7jtlikzC@W$*$s~O!7xjjA8^`d92nfgfZkk)CXB4Ur@ z&SKSt@O-dSp2=(%$ulMyt18WRWr#N36J(Oz&fH{ak``>>E0Wl*1*BR|mq6s^q( z3xxN`2;S99i@>;rrYlT@Ta5}xzInk^p)c>S@KE1+>md6%n$#VDj35Orl+tinUl@W% zZ$xo3N;$99Yx)J$m*Jlvt4|z{RgJ|Jze>^tS8Xx4Ozed{(16+oTDvb4$56crfJQ$i zpFo%L)Q(Yh&peK5sJ%WsA9|-U^xL!&8&byx+mQ|=AzuzUkuxTq_I^<2TVmF+8*0Cq zE@w$V;7Q?(E%zSbZk!3o_ooN@*tRn8BHIFiYp5UODT}k>bqNDjpgzw&ujOaVQF)RU z5|HOV>-VI|v3c`s%Z=Hg#S<^#U|IW?%c#;isW&YD7w)M~&4Cuz2_o9d;AYG?- zHv~JezVmLrL&FX>_wD|a?_!cs?&n22q@2bXSOXn1yMv4Ip}paQPb#!-*kL`m*I3++ zD!}ggHEE8gcii!Uu@alKG^Xf%oj{_GEEZ8@O;kqag}-^rvUlg(YT@G@ z`|BIXDOG-oS?VpY-FJE%NZyFVL(aT)1}1D$Cz5mItXB4DH3IlGw>&;D&pINk2G9FL=DLsG zJRiYoh@SpthC0*qIO`78?-~~#HZugiguNmPaLqbi=Q2gsHx6UW63m%!MeV84-k%GY5to#>#B zT8Z|y1wSQmK11KTkA4NfkwTK-ZIsX>FeKYER+7Uf0XpdXur2UdO)jvuoi9f%7L z1a4G@m^Tem(&pqyKOvbl;5XFqso+>s;Ps5`qJ>RmR~fxN5FO~{U?RMcdF0rb`SQU_ z9i%CVjf!{M-n2fVFit^~k!p0AR^n=Hmx~k^o;b(Az2+Tg0DShE5hYcp{inElPTj@( z&!54D%ta6P65h{~6x({|Sv}(#w3)5-8&~d6_t`kRee~W?z~w>IZfMP}J8fJHB&{N0 ztlYnvhlTU^^80v#>Cc+JlEwk6t4@Wud=)%)pMRYWf z`*?j_eum~@QYJc_vQ|!{yD~h?L%1Yu&-gkbB0Mz^D9g;m(8R)QuisK_w6?deYcCHJ zU*;|ZCcJlsDy+i@tgDP(avXS8uiU5@ykGJ}Hs(cIRya}DOQzHlPgjPXe)=7)Jo|@p zsCx^a^~Jdq$E{bLRyhmla~@OZ023PwG{s5{BAm*N!+h2^*G>(SY2qhbmPT^R8Z0{O zChI$*g*S2UCtE=7l0|kq99CD4hm|OEzhN-=A{CM?0uXy7R0;$NKnYgwrx>l>8XDFb z3u)5Y(`&UaZsWy&lVZ;K+{bYkOcD+qHW<2JTbXBI(7=B|zqAW0l0)wjX|AU5I?u?E z968-a8n-oQE(>Wg5nB4FsAw@C$zT3jO-eC43F-BE#8ibiezRWFSWNWtg!sPpvh7#a zwEM_oL}|}Cf^T+Uj6HS#CZo5#6(B&>_1N~nK$Fv6v2)3iPv_mYFLCZ_*Y>uN+9o3s zPyv_l@kV=sK=PE(>%D|hVQuQhC1quRq&=|&k;EPB_VvD+sn%A?&kI_yB~Lr~UCFTU z-qXGnBqH^$yf%;ObMpEdK7C2PGo0vw{&D1Cze;I&kX%Ldqt|)mCtnz}$66U6ai)%4 zq>zdP3_BAZWgf`9S((w1zw0awv$7-79$^q=e2M0v=Zx>dt47- zRi!kZXRz4lODU@=so|Bk72|sspB#~5JWdXMuOvb^q0QiEn_1U<*X#3L*~HP{SsY+i zY@ow=)4fZ|OWGSZx`|YEtl8bUpff}IdeB0!X%queNY_j86QNXer|fJDO~70X%VfM^ zH?;uAmfm}q4bvz&Uu0f)s6;eR{eDf2|LNWR_s&%MiFC4GqWtocW`d*Xb|%!ut*S24 zx?kB9Y}w{>mcAlrx22{M^S6kVVH<5fs}NpK;E|wtugO84IU$4&Mm}cuxe2Upd2E*s z9O!)C!s*NEGb>`Px~Xp&C?&h+gD1&QfLt95DEvx(;VHpCfk-3v6HlqM;>|3eyJ)X2 zR$=>sd29gN?{1M%?uN8?=|=j;La{6ix5c{y#VB50^>|8H1(3LF!3ivoM}c2KUoT2} zD_4gVu|B~Tae%U*3g%qVz67sYd7@wP_drfDzQ>Q}JSRN_-%L!l5Qaw- zx~V7xKT&GxKKlqJFTIHRFO)regT*lI7)Tj3Y7tF)UR&SmqB_29jJw)S7mICl+3Btl zj9OL2BvVK7S3Q^QV<#+ip_2?1NB8Q7%`$?Rp#X>(s$Jy>MuMmxC5Za5!dgP^7*BfL z4qR5M?$g#EFz(IAzBTe_cl%}N97#vac!J9^^cBgVpCxn8|J7{B}~uf&}(-lSCL#kCY94qIA5A)%JpuzQ|bW`!TE2#)2N*tv4x zIR+0bi)DyyDFOVL5v+g|4*B(VF|F=%##piow34lJtRIPd_puSr^;{9C%*QQm$f z9J;vwBh}u_4n1vuJO1{(w`Xv*ilS0-zdmISfX9cz&;JS1Jad0O>nAtOmNXM*&pace zR^tt(67^gYYZCUD$13_jME87bjhA9lX8X{7kNEvQc*q|kCCS>peUYdXOFrfek7(bR;;gqu}9Z&PbWSZU?tt`V%BzUX$~%zY$=sk89u zQFUPjq+F=GoCHJ!F>m~C-2XN|D^&e3ye`4ChgmoB9B~gyVCFHJp}brSOgjE*sW{PV z6)pF9?J;z^GbYCPhuQj#1e2mtkQlFB3vY7>75_35w*;mQ{x>V(o81mJ1(Q%HqbI?1 z^c>yKUz3Sj3A;N2vx%QN_ACspmS9_^^!BE`!)4Qqp(flT2}z-<*f;5??yvhVYUx(S zs$LUxC94`d_)f}4u`%;bZnqW#HIBus-Xu6N-#R6?`&gcz`xF<_fA8DnoX{AKFE@Ak-9wi|)>f24!x?|aW2dkRjs$gi)0wTBG9J^=TSpOcQ)G)ruyw)!Na%tfjL$&urM0WcM?L-NiO*eg zEjWGErfW;pH6YxXnZ2gD}tShC1P>~0ybPx>OYRkR^%ncZ{%KS%)Mkik>(-pE)2i`)i zJWKMa_iepr66XW{mlDK)iifJ?iUo4w)w@CFh*^YL$vQXqiXk@$K~})u=PCTa#Fv(K zf{iRH6>*^&14*Th)Wd=i$ss!4Gu7|N7;1a0?Ue^Ooegedud}`=>(5`AaD^(hj#V0= zSjAAO-4N^yIFGt3-LF%dfZ>VP>DI2TL*YzShwZ zw&k0A^RQd+-MS~2VnsG|Y8LMZcWSoX_-JWLXZlh#s{T}?r=wUZGTo)hpJJcBrdx+p zq${jR9thkNGLd^-@uY~ekirPO1BkEL+VxBMw>FTj*lz5P7PQ7M2;F%!wADy>G4JCo zd6lariDF71m-(@#ShuXK(1^ma3M+ecUV1Ox5AtrV;smim&C0?nv5!{nj4Rcsx+C|3 zbEeXN_ObpeQ#l+ga*Sn~u;X51kvy1s=>THz(9G;ipmAJ2ngOrko4Ch8PV~|#TEy(g z{uYVMlc7^Lvf-QZVoHVb#1s#J$+sD$F0s`UlC%2DeWR(BimbR<=Ev+w*i9R=KzA9`i&t+@mfRW_Fcc)p3wc zX`acW(&}H7-O(r>UwU}M?s0`Bu|n9lyNh4c>yhJMIu2ta1&8z1XTPOP%gaxDoYC=> ziyOtC2|HX0&}S}JeSk7X8=uuWpVK++z<)T*>m47LI1yn>g{Hh&<{9cnTrMbLizt{0 zn$(&a)2=QIpHLJDo4WZDYm+&vo$(nz(yA`ta~Ea*W6fLDq9~o=b0zM#@@IF5Uiev` zwfZaG^(tlA0a97onV*QoEoJQ7dmDZu{*b-We1NH+*jjDm{LA;2TMCt}-SOl2YS@TMpyLX?$G)(u?c5ng5>m&nZSB6{KFIWUm#KEZ2GY879vdU6h9 z*y}qbjZaOhV{U0pFbPpR{Xfl|cU03?8^i4tOcZ5_ z0m7E06e`vclR!bH5|kmzEF%m787WHygg}%42|@%J0+u18iSN(0r|0ee?;r2q&$;K^ zd(S=3dG7Q5e4j3ziHXbir2E~wWZ}ZqJgsCT5{1tpZ=FADe$ObdxGP&`svZs1Y( z)AM!{k0G%vl}QDaE6}qmt&aFh+0Q{!5pu8{@rtM&?Vf;Rqk~+1zpA>jrmE_B^mL6q z^bB^u>f360??rfjK{&35T{K_IKQJ6n+Lxo4^Q2#{puJS${r4?GlAI!Q*{IQ8b*CX+ zhX1%(-!!3lxbgMD%2==lF17ZCWqzhM-6sjO2$7fw?_Q!#v*>5f4h94yFI{-Ou;^{5 zlST5~do1U^omyr}t8mfgWYi36PZ;q{{uePe)Ry^S_RB0E+XR~#uqwtw9 zfySubxfp2Dq&f$y?s=dN_B=9e=dYvRYHJgvt#Ngpv6e@Z6PB6ka;~3(4^b8hmbyf>YQ$Z>w67#51m13CfNWEb(xP_oEGVPl#`%t_x-+S;mLHf* zR#*R|l5!t&Z+(ThhblP1JdVp!i>L$du0MZUJqjHh#!S%2Dx`gn7&^4qsX<}y;9R?5 zUj2|(wDLJXUMrH{Bavj}pr?nB$Y2amCP=sI2wo}{a)Z+98Gz+GAE)we{M$wm&zgCC zUBQf!^7*F5jj_NOIr=Tf7q0xo6`vNKhJa2`ie%-@y!h?Va z^CsG6^(-*S;OmsUM+1L*;12zjYqGsJVlcaXN5B%-?5-3X8~Wy?B6?@5X2(Cg2w-CA ztr%48!3_sBPR?MHE#;scy%j}X=#^RjftZgeb+;~IJ ziqi57BeN4;l5LLRmR(g+W~G%IKG4`oDL%-%i+piD?g#H*61w5SV2jCym->=&>^AOm zA8{tSq@;qpuH>cRj3gY^9y?XuJ-Q+EO5eYr+VT6$orZ-p<9^GoRf)$yJ?iMMp0Q|x#x3Y< zA9AY#YpWb8hP4?>(Tt>MGs|T_;_P$*e*v*fiZ_?uGyQi(dl>B1KMXhXVHnAng-bTM z)-Th{iS7-ap99t0>dd8wWjrZQWzfOc-Lw1kB{Lqdtsd5LW3uM&WaS&7t5ZaDM+bGI zK$XasCuFr&#$1gz4%j!^=-xMtCX{}7QB!xtwJ{WYU6A3cwBq~g`~?bR$E_Edhy8xJ zC>Y1+j0Mtu~}7QeiG5-0BpgM?tY|4x@Ahz z0$x@_1+nk3$OLV9c&Oi)99iM`{w?4W3sf_*&`26cdxv{+XUFlXL zdgJlK#M@Z`aiD{&BddBLh!GdimBC3scm7IoJiGejoO@U07_nWa$fy#!8NaIV2CB5K z33Ozv0rF=%jt&rnqT7XW{mo))!Ia&>ajPi)mNoPP5D*4IL`Pu=0R^RdXz5fslG?^?dmJCqSkwuOts(VF3UZ_y;&2 z17raLJUo27O9c4%_=JQ6M8uRN#FsA<(^6a|rDULkFf!25(=)O0aWJvmWTmI)6y>@p zASf&>4B@yVDJCSrcS~64;v!gtgoMPGiD^hkXoQ&QnT7txU*}%|asn(|oD3W+766+Z z3x^!*yb*wa_jw8H*B9`|7Zx@S?j<~Y0z#t8;0YyU05%p54mK{%rAxTD;Mu<5?*Ux$ zOBBq4GI&=tP4QWrDTQ8yClRpTEBr?FuxE?ymYGWcA<;Ez8d|y=>>Qk2+`=NaMaAxj z%ifohS5Q<^*3yRQ=<4Ykm|Ivrd1_^CMpu)ZEhA*52FKKQK5nJTf{xGdnlGu(-6mvc0prw|{VW zbbN9#E-V1&_prdfe-G?m#zhW}3mX>~2N(ZhTv*ti;DJMqdx=>Pk3vQh-_-dki_i-K z%6s8Sh2IESZ#~?iGIQx6y2d6vePjD#Xun4G-wiC_e>Jkd2KJ9}jRM3tSYYsQ$N?yD zc*2?NPsEw(sn_#)m^-tLBasp)){SYH&8L1AL~LE@}O|7qfE zjS9MAn+N0dO!pjM=dQRh9Q^RIN_>o*S5rdhbGbB|!P}EueZ`xFQ);C>lh!*@tl{Mv zD`!9Bg?M_Fg$L)kxOfti5nRNa$q3*-A4Szl`36Vk4TrH!KcBoBxm75&8zBv2oD1W` z(LwCq*yeAvO}i?W@yz1~y@jkZTj%PU@^-A}!D3_n@%*&O*@x`87;~uYmBqB4N7jG4 zZBESiQPepgQO$D>L{hq_6!_?QO7bCg)z`b;nCy1c*BuK!h-`Vpv$d_#MX_9|J*u;> zzt|Tqy@e-648v@-(KcP7FviQ4sbcqbr}TfN(Jq5)K|UFsT9Az31pcd|=(%eC*fVfo z1~t7BZ?HoOGbY2|)Mp;DA!c+N6Ug)4t1>1`cieXlC*FCmkAWXDc7`Kx@6)bw-u-VbR+ragSI#hebnY&kV(*H)BDQ`JIHbEO9SmWVpi| z+Tz3ID7B+0HRXW+cxa1yY7CEbc{)uq2i%U3mwZng`N-CBR{i>F0bf$evbRWlg204> z2qBxbF{wPW7%lxwtm@i>)Rc-9O5Iqv!*k_xApQ~6InaiHmPy@!ckvUQ19MGMN*)}e z4He67|K=R}h&9GRhAx$V8lmRj12-4*_@H3J+n8!0VJv~7ZZReZzGH^`N7xkDs_;vK zU3-`E`0L{Z>-tNw_DHPSA5@Fv^ zT|D}k@z>$(0>p230fOt5i*?@1gY7NY${X`t5O^zI|A~!G!pR-m!PS=Pr?-rrJ*1QuC?^RnZfCbPvIq^P?V|MeP4{l&in3+hGw!!wt@XrTM&#fBrs+ za7_+nQ(!en;^ji$n3L^693J@&QzDC@h;uJLl_(Y6R$Hwkh-a>JPMBUio~7KRGVE|o zA((zM4==vwksRSfPXXuVU!R9xv%Z?c*P}o=Jq06xFPgAInp_nYJ?IycemrY;w9W;rZELevOEUS`2EhB`Xl2R0sFQd zW$#s4jicTrjgNgS{ps07lX<3tgw%KRyK}E*WLlaAT*1eCTV-?2sC}Xf)(b8wa?5kT z=;}EjL*Ye z1cR%Bu*#SyQ~;c(N9u5!l5JPa{Z{ze4eN6tMdwSk3hvOgQf!*|dzC>|1#%oj^c3m^ zxmjzSU~nm@s^zX5tCG}yhP#Bd&9?~)1|S+T=K`G9c<#zm$LKYPa4B9_y6dOe z_$XpOW|a`B%N@p2C-dyQg=TV{l(QT+0epH_qm)rwj)uB1mS*_+lZu`n#G`>@9g81ZQo?^iHJs}e>A#fDl>72tmPqMxV`)wI`cZEFE8n8!R@WKiiua> z#v_!@)Wkhdv4ermDBIFBoR+!aR|*#Fw^+)VO$r(NjYE5e`zr>SD+|6<^1vO5%feG1 ze9+=k$9;!xIHHS~Z47J5G}db}_i=T$FRuJXBSHRYE?nDjzIs4-0_vCGO!8g)rIeeW zK9jvfQT*zru3MFt`8H$Ot;SXQ+-oh|(smqTlFJj3mi9^(+YAV6G8>+SHiib=S8NYj zH{dTxACb;0na>dLyKAM#8KQjBUUs&LV+G+}XA5Fdg%n)=$OS$hVwTQfGFSX-14 z23wfTGBIX^$r%+^`R5uZ#`8fhYx$~o&&OZ z@fAosw{22%up@L&e7vAKV%vc@x@hcJt6G_jW|DGicK9GK^Eu~k_H#R_tMJycY*N)H z7jH~PzMR_Ovq?Chtv(03m-3!gEGd}RTmT{N(iA&Kw~i{<<6H@T8jeCs){}u)PMoG{ z)){+_2V3*;;V2Ws<;v&;nzHiJ>ahFm9nJD<7L3*2gPUjUjI`%KQZQI>zqXzO%~h*E zsFnLINPW}wC&ZTezUC<`aHl*oY*0*#lj+sIvd}4r=7GUdU8}_%4ka2)2G%4roztwu zZ;TrR^$%^SFqfLLG*E~=@nB^lCOSZ5n5SEhdWUBWkB;%UpGqkEWA5c?D;sq#6a)km z5pgE}cW5W??o4QcXXX*nzFm^_22%u2R1ltiqYDq2$a{yw@^9;w`0;Pd>C$b4j{E~iC$vv z1;dI(f|Vxc4#MJW#h5tQ#RhJzzCgqxcvutvGZOxdmS?57^o&iO(M{(}d(D@`U(e^l z(;pNi{(5B*7$hkRITLhsh{&tSXeobcQAYGm5>be&n>lEKx!#@k#M(jFSX|$5QzAOE z5v@e!9QP(lS=r~4=)XQh1GS&v$onwr|$T+a_DN{2IRuaxFIa9X}QUM5GWBR~Bgp%tYDp8A`; zV|NpxmC@%kyLr59A-5(FVNiozD9KTSEK=r6NnL+#4!?bpo&#^*Jh_BJ3NG332G1}` zYV%&N2;uXw3GU_vwYY^c1?WdA%^0rF$~LxIfncB;8lZauV4!IsM?L2NVel*Xk|`J< zA?SAA5Qbd+3cR~64gQlJjQbf&JUzIW;1`QY42*vMXl=8Jj&Uas<|-*MxMmx`^DN-A z3jaD^RVM%X5OVy!R3hG=PbK5P0V#@ak@}x+AmyJgY@GkEcYyrY3;%JQOFlE|m!}MA zOd3a@n==ImYutaVP%dJqHc-u+-f^?%D74=8jWnt7NxC(soRfGnucRC)Y&n(rw-%fw|S8LFstVxiDT_W-;%8h6{}Hc*<6seKtrg;H1*?xC^e-oT^z9fn`ZrX@q>=#`!)z*{_1@nrTZEa zs`eaPNP#tYMn-SW8+Y4VZ=Sa9N4b#B@c8MO#|uSpT8gMYk2YaDjvn5k-_c!_rqY5{ z%T0)6gZRtqio&7}%OEqnhW_myEe6I_6a@Ydhvvf$-K}qD;@T8r?RBfUDdi@cOrv{P z_WbJ>r(gJdNgr+%JvzSm!5DF-R8 zs75#Q%r`@|VVBD6?%Q-)lP{gnVLLjYPEPC8mE&YpKXRO;VJ{;+$z~DT62GS$A=6b&6qC<>5Jiq=k|$ z>I6?fl!2x<`T)kjz4>2oI@*&g39f^TA_K)rmAEe$4-ef+_2fLT_ ztkSwmmahqQH+h^oj=FLBG)m2jYnw5vAte?9reQLS$yt)0S_LWJa>KjsG{X&YbO^={ z3p<`nP0Alz;XFd5zV)DX6BX>NTSu;N`3(ufSJ#b5H<`ydvFE8C-Cea-y9Og!8ix$# z{D(GvU>b zSDFaG0X!(NKHnsH&i=B!-PE}taD8epCBU#ep-l*f=^Ti5IS0rNY|21X4AD*4M9yoe zJ2KSUhq)~&X!vGgQ`N~mE+gm`JvBmh=JN0W+5x1_+SNm)EZ4tFCb&TNsO^@zAbb@*6{?9EA5MzF)zSJ%O$wDqblK_WYB-WcGedkzrP| z{@CVScadvJhu~D?EuSr61*I3785Nh#p*M08X1Hg>TVM19axKY;XGvB^cUl(DF&bR$ zMoW?a#>q)#6`|9FvH@b2k}`)`ng8y|1b-J`*zxR{mo`MEEs=038*U$=zJIL}`67TevFt9mJ#9-YZ4H?NG)tq-z$3Xnw^MdD>s1<8#kq zcYUlQt}exg4v4jVmSLnqS8$kp4p_!Rjt)M}o<{6IrAltp3`f0l;g1nm2|h+@>TqHw zj2ZApuizokNxQuKBX%3F_Ozkv%-_)5QwFFTFK>?Z2`ZtNIytfGEaIX2qK?RoR#76` zxG+JwkYW>L&spmieAv%a^_TMxQ6--G-E>}6Rb zZCR}gbhLq}9HwXL&pL^cO0a8jNeZc}ewho1if-@gSu!?erL5;AtlG1ixL*%mRv6is zIR{|w)o>43*9zyUYCn86SA}%*zUzqwm=a3Nk-}Vi4$#kfy1KMF#F8P})X}cbBdG9C zi}IkKx%$@Gj>EANHsu@P+!w3lls*fSy#qUNqS3@8K&&hwh*+Qk{INwArnFacBS9d8 zFU6(>&p5cJd1_JmT4s9A>sii~hT(0G+Xfe51JeR3dP6gyc z;^p_+&RxU}f~_HYA~*g<8*0y0c%=%AJ}#I}g}TEVyK!TvRS3zGKe9Vk2A-S{+1<`v zAl%TCm~>x6>~flzvY3*<#pZ6L_9feRhnnRLh&;nMRG(TtE{*Ok}b0u9X8voPfM7M z&~s3C5U}-^9x6Ezd{d&{WOb*+q8{0Jd=6+z?PKPuF~yg+X~$fz*88gS91pbL!=hm# z-;^c@E#kCz{Phs;GmNoba%}oZ5L#pV*e!LJAIxL=G@qf>pG4=SsX4Q^Ccg=eLTt|g z=W=M2-rt%#*Suyk2N zSfs-MV%@KG29^*4_ud~=+TatQPClvpii29vVL>|>vj?PJHc%csX+EU zf*a|I0DP%5R>%b#x!79JL9@H5Q@;EY2r&iJ=awBxESKcy1p@e>GHOj#=71v-?c0xB ztK&p->*cIZ&ogWIs_F~gw z_(NIH5I}a(xC@g`PH_(SJ*tc19%dyQy3$v6s_^0nR!R7^ZCRQin@U;(UNU`E9HJC6*TD|C)ak~WRkw#U+kTOKJ%Djg;gtglOy3AImJETR}E>vMYK z04go@b3j)V?EzoZSxHa8wV39Nx8F5a8HEfixaMNr&!Y7;f z`84Rgn`3?w>ir0-l154G(zi$tobScyhJW6Y&8mt@+X&@pJ@`H5dl2%GZMP-f% z>9x%pKd2EiFLysg$vDp9kP@dbeABvE{BYaP(7}=NLwOQqk?Uqo-+K}i9&%z&FarSu zt&$-k?4dI6!wr{Q)f0($iA(Js4O73~=|1s6ZlN^P85(_N?LueEPx_;De6$w44%X*? zO!Z)jv1}0ej&3*1uV;l?p0>g>a`h5{ccC^p(PUp7k#6c>Wrix3$N^dtK8}Y`4p;l1 z=N@a#)CR~8>w)nacP<`?4b@}$9d!-d_wLzn^Jqr<3^44G3nrKnh+PCDU!nzMO8w#? zz+#Km>HSo!k^uXOiGmEzBhhl>Z!sMq^*JrVu6#=qdOmc%fYz>~+36Nr%HZv9XCa*Z zruY=WFHaQ9(KZvzSg#0heY&yP-Y~U`7aZJ4nlQiD`AyrMJNC3*s^ zx8cezHrLiXzI+(#l_~Y-)l|&M5lSUrA>sX1XS9pKUIW#2*s_Cif7pkS%7;!t+H<>q zX*946UL7+AB_&;O*+83fz$(sMX&ENtf~_)A7les&NQtFce3$w*{46^&+K__6#Ue1d zW5Ta)R+>69m6%rMA%P@8S|UMu*(>Ka<~oO)8otppSL_#Wz$^G!cUN^e6N$`@8lBxQ z%VD~YV|GzWEXq|W?xrhOY-sbD-p+dxw;WMgB-CftlQD&BlW%%n4$H>#1hb`~vmW$q{Ph*Y_TWWFk#M#_-%r)hCFj>Owv z`D#A@kOic;Vz29gWlNU^zkdfV9(zIfga)$v6Xi-yw(OXi2$>JNnYe1!G{&fzm59kS zs1E!7k&^o9yecEKYn)3+76S_u~~gNcw<;G;d& zK{&MZOMVRt!&3r$x`CyDp&n zOl`ntDpUtk?Aj~mLvx5bv~i#vkb}t)C0=3x(tb@X0^t5f4agl(g`$n7_D9Zv&w$~- zYvQui9dCU3XfMuazlx}*Jt6r|5@9v&L3>5hep^@Up zuAvig;~JN;f2sM&{l7AytQQ;D&B=J7Ju&aXxxaOc=`ykx_IowISXak8JRBo2z?Wpd@<_6G7~E zk-@NVEnGMg2`r02cxj=io_FVL1yWE9U*WSv4u5Xg5dlj%)}(2@z+96hB$G($UgPkka>odkoIA4c#E{P0|*s*?t-uFO>vgJSgy+EXvLmoyW9N8oRoS_iMS^wrVV7rN zqelKm!KIYMLlNA~gPq4OU|~+Z1+5TuZ$-C4i@GIOp{s4AyLgG0$qek^;!k!tppD-A z1t<~qbywM@@3^F!EucU^S4g?7R+StMCVn6yb}52b97THXmRhnu_wCPku8c=h!gc74}u)(g%;mse@5+%b3k0 z*vrZn)4TT(S8LS-H8yI>PcuuGSH{ccVN(inG3$bbKA{V0u-at*o7ZQ#nsRvgPfgt6 zw-iS5IKZOvrX9?+KmFWql=QYgu&Xsf4c99 z5u)dVhyWkv10~aiQ4gHci02*Wz+lRHJ}I?^5b0+A_d2vqg@%z)S~(EVIpvP8C4KoI zQzDW-HVywFiS8KhamNN?tZ45k*cJI(6FPI2mml9)L6Lto*%o<WC$akKpFZ7s?7l~d$qL??q7c%ln&oibUlHE$pSL17xfze@bH!Y$aX|rghYgFfOw?#yP_$IFp4lYYoD@H zu;7!%F28$q`7z_8#Urg%^|BH@B@cF`xXnAOU<|3)1X^%SS&c{+AO16E2onbNVpsUk zQ;a*PssB3?=T9={&sRCgYEQzF7Bv|LnS(5=L?RW5=y$A@cf+>6z>#UjD?(hfuznPv zy^oh=7yWfXpm)quEhj-;6T~IyfAzQte{mC+M0(gB<|VyYlM_b(>2}Y*#OWJcMNguR zH4oxmLuY45I?zWIXzqDREJ{&c0mJMOf&rkDxBqK3_$Ni6BbC8v^s~m?G zKS2`M1(8{{{c22Hm;^KY6C+|J05ZaXV9_vE!AMw~K|dQn+L=VA`awH<%23hihQ?&bH?$SjmXU%}Tv(U_aYg`X zjk3J=fd2aZckDsc6~Xchq=jh#n?A?4KuQdAwWp>D5Le-`_B$g`&!ow$`!WC@P#5<5 zRs=Pu&=?2TEp5l$UuB-B?V!qn&%w+RvfJZ)mYo!f+5MQSZ(qae`0*T&3dn1*+ZBu5 z#0}LE?15ht`_!1hZBF@6MPP}6ph+)ufzWR*b4otTIW|D$u9x2#r|&ibh(~o;;^x*l z3}H^_>5!+A)aVCaO(N;T-S{p{GfxY)zb`o(c*XzdJ%hlzl>%?758VxOylHf{s zPy@#VFUU!U24=f%PZ;Uruv$7pOXjTvp~P|{lUHCt{80S4YWa*6IzAD8Oc%E%Sih@P}Y)wC|8 zM4V9CBOc_VhnrPzp>NNokGY)#wQ*;hSP=#-c2yC}KN*9HpscW_jpS!#FZgs%7db(S z)cY=hW?8vV8+!SQ%@4r*xCnaolDLO-BY7GsI9SZ9nrLK5vyEw6$kC?IH&k}n%JWsg z-}^=2nny$TsUPy6^6!!kB85))>xf;D+IEoEZrrkKSx#7YbWJK(`?WR4tAz+0z7jvv z$DeTA5>C;4Nv~XV55Al{yPt&&cPBSWeeHY23fq7?blL6-#I1B94#hFl|Eo!6{oN!} zz6`>RxZwF!ZmR&pzDN(4`2MBd z%ZjA|nKf;aHthY6J6WfAyyYuE?gR!EKQP(%Sf$wyuQ&^WSj4OEA37);nqQX#RGsV( z%zn0%^{fh)zXQqKnnlM0><#trd$|N__9=JBu_tre&^BLbmZfR&4z(U}lql+Ayt(*K zJhEn9@C~_6RyUMu75m*Q=B_RkP#!y2gKs8%OT?G6b|gJ1e-AOT=pRtGU$2|61Rs=R z!`+MdK@4m4pWugkO5at-*NY)O3uvrcus1e)@;cifD(vH zjQX7IamdlfQkr1rp0cJcDysDbxxwFFhxP`Z$84q8J+y>xGz(cLESEth%G6Jn>%exA z(kXAaN85+U7tjf(d?yJ`F!sKHRV0f~SBzNyOp zZEF0Vk}>~E&|s+<>LO%Cl86>J<^8~d@gmE9BjP7{KoS3>{0g^szO&62vX$mLF&pKp zM<)oNP)_^R^2>nBVHA&Aj!c$yeo>vji}!EE28$LpWXt*3ha5%{kThY>T0a>Wb(Rfg z@5sy_?j-L5tnR;U>7pL$&Hv_(yCPA61)Y$9w5Rmf%{Hqy9QQ|z%7_QDJua2n`iX1? zRiLb6Y7KZ++z*otm7(1<>Xl&o@Af#}Wjq%F=c0-law&N16R7f!TB5*4t}MuAGo|?u zY?rSj(0yuP3rjdKM7H+d8l9-Le-irm&O`ZT$UPqEFf6v8SN4_CqxTVp6k(xsCMICp zM+bcO_3*hP)#2*^ed4! z6e*pnXD{0j+~w##q{PaS(+nM;#f^@6$Q>W(F?~$u;W)EWs3sZO_lss3CHSOHS?E(V z=Vz9qAaRQr%v;^vxbH9gY3^V%^-4wMBBiS5ar3V&#h|q#nlP)MVAZp^FrOn>sY0bL zX3ZR$3+Ti>i0Ygc|Kf4;1N1}oj**SG%YqE$c4*VqRB*1zn2+-zVk{%^IBqwi$sE_d zIxGAfsAC?t!bvFE=8i)}*($A%e0si{MIg4%U~9n7y@eyP(@wg@Y7d#Qv31ieuh!LQ zH*%8q(RNunSW~%=Yrp&s+3wFTN2$rjBB!~=wI6D4QF_|bN_~2;0pGi-b#JE)WSS1! z=is+bYLT0UseuC!NGEa~*JQu@2c+Tw@&XoXDEQd(Pkg+!q{i(evjoDv z(5Oc$t5%_OFpeHCXd0jf$_?0(D^8r_WL^CQM$2L-K z!_oKBOACqW^`nRr>azv1_7aJ$+=BwdJ~rPYEp^VF2vsHE~~lJj6W}XfbGd)`c}ysCZO&EYXo(l>yXKJ8;~f#C>f;yzvwD>yrJ!IgCYV}riI_U#Ra^xZI0j@ z(rD;@*T-J|3hSc<>;NQevS<8we=@`C_;GetrLl^3YHh7!AqOD6PX{_Bp9HssDn9tm zZW24Sz~FPDlUOM9Y2rceIZz}8JImr{RZMTX({m}-xqrvq3v)Ma#}I!MFkj&0{b4a#sdt(=XbS9H>2#l}0C&JgF#WhF3zMZUgr z#ya5RD04vr8~h1_|Nfc z&%|%m46R~ByuPh{k@2Z5DL)h6$uef4Wcp%pX2_T*l zYxVP8^*bs#1!uPcQ-Gi60t9{}K?w)=Br}>!YaVJXOB0iKJeO4?cJe@0`NHe-vV%0T z0h~n2;$Qv3r91ZVw8VoSh`p@3Kx87qThnWVjRzY7^(QJx-SQIs*WolRLg*pLP`}0{ zME;Ga;tXDj#eOHavW4gaHZr43r7Fg*J=A@FAvGY8DB&Myh)AMixb}>Tb3dd2&Z-P* z^;|^7T}y$;&d84Ha!{LM`I<1HX1JS(r-T{>i#+6JH$;iDH*GJ*{I)Z8-EL1O{`_D% zqIr_>L z*FON5?Q6bf(=WFIdl3H6TN>xUlW|3tP+okrkeT`sthqqWY;Xf~Ti^#CJ@hJbg6F<| zkT8kNpO>am3zu>k+WCR`K91~758<!E8oy zxQ$*)dJivXgbX6fKy}^NUfA!{1*#0(FOL{$)P2hROg~t5e^10C8(q{ntxT&C_)TZ_ zxL@{r;XGmNP`69l-ae?YDVo1GQyd5`|Dh}HrtEsT4P=r`%g?gEyRc<4-X_jSqM{$t z7&`}Y!f=K?p+SIb(FuRmO;=I{8t8|_DG6E8YrXRLIwl`958i?uPrpndm|r1j*k2Vr zjoSMlvh|?INwKQqQrhu!cMh|%eYRr4w=;{j2*AaxJuh3-5=pDd8w?V}tjit%2`e1Y z{EtT<;faLIUjnXHXt|yPz7?RZpABE$hIetrV2+{F3S;%~`~bb0cChIC^Zf&|_V*ac zik?)!YF{P&DPR6#OtEOGhHNu0D9xk-0#8Hbtpu6&U)kWI`GgLATi`zg%>_iU*E1dP zH|B6p+*r4=vI%pU5@KYkzFZ*y@xC{ALZm1D$;X#j=r(VR8_ExHZ=6+nYW)i7ESMNm zb%+zQa2aP%h<;sA3IVjf;jgMN*7uAAd3} z`^%^UTpSNnTu(&{)5@xL>6^Gg3OC#56ajqsmxPJ+zvmk?EXdk$OkaqRR2L|%q1AG8mRJh-2zNV&}+qT%NO)TVPrqf&u9tctXAz)%@J^Z^B z43!0Md6y8pi;~?yrUATvejMe)Tijevm^fEd0WRPjdAPnPIa?&n+AaF2CKgDPUrG|{ z(tO=rGxocFn-iG$$$baT2!eX~ByA16=VTLl^PeyEIMy&tcTMZe)C4^FEx{j&n59Hd zat%>5=YmiC4J!4uFW-I?FFKy>Mus2nr>y6aH}#W1PHlJj7eQQ^IcW7+GzX#E&&Z8e zCvRW61nRvwFOEQnjIjLx3N89f_}f9On795)B1ACuhn+FT|{4MYB{f7B3ZS|FY#>*!PfSnu)WlI;+YUG%T+Ny=(|c z1B}wE-_UeaX6|!P>8=fd6P#Pg4XvV>-Opf8u*+~(U0+%ByvX{#_UxLP83*?#V6mP0 zYD>ILS>tOJ<5xsUe*suIFGo>Ap~xjQ(s|YLD)Iuvja_z0N{>US z_dpEFhIPWpKmIMUTD?TW@bU;jc-nXAdHeY}Klc1@RvF{%a*J|r?H7*H0XQ>~r z?>?Oc%^Ioh;hwB`jEM{H{pAT-v-3@_t#3vDHfmfnk&=gDzI*z8r)>>>FgD$S2Ohzr zphW6N0_Y_t#%>UZdEsre-+@NS{gd(!=tYa^D@AF%mQAlUR^^^rKY9iLSvJ^D8 zC*lqs0~MI7fjjzomU`rC);1{X!Nz%wqR@a$Q z>;CK~_qX8q&sRVtceD2GrW}34z!lOtYxxw4PY)vQ!N(p!#mi(a7N|6FsNOs+XdNp2 zA%#*hwP)^nPgSEe4^6wZRz`jfWL4RLoOBh~zrTk;rFQuDqdS=P-_#|6HY*3UA>OOr zid9zF1FV-Unf+PIg`4AVyl%BH4y?7wyHN+dy_8!cehc*IClVFOEVk>adIaJxMtPZh z^%M~%l!{N4kaChJ4oA=K;}w%%6DB}VDM)kmW8sQAsX2v;!rivu-0?}VgS0K1thgt8 z(9g05J|Cd{>NOy40Q)|pqzuQ#r`Fe-p2sj{eLsttFVP)$jU5m4)|o9t=Iq^_C%BNyKE7%{KoAdx^vdbtSX@)h-`*d-Yx(g(mNAl z4VWSre%~Dt(y>_3=Goh}A>VV-jq}R~p%iisRL`w5vc-SxZop_0A+`ns!PLvf2YGBNS8{Mnr``>Wr56)qq5y9_IN zLwhIgfnAj#xTnI`Jqt8A3Oo*lL3Y2d{}N-FW-P%nG5X2R4U5>ov}mDI8andcp=)+9 z?bP-DKg<+dH(lPPph(SQy=YgZNkZ>OdCIRe^wYV+BI>W%Z*K&FoE~kHsAwq~_RINl z!4378$gtYo1i8v`Fy`I3uHQgo4VzJDOZf>57HfE$&_$)G=gHoI<2I`QUDA6LAhSXO z00CI|Oox;u_gi@1@Dd2gOiHcpfwNOW+|za8B#Z?jz%_waZ8duZ^iDnbBEldwoO8*1 z`1#&EA;YRpwhhFE^kgHU)Boz4OU4PRO?PCgAD+>{mSW)b`4naOm>K(JxUUHMU;E@Y53-eA}3b-Nn}>Jn_F?uyV8 zj(JhtRQ)ksQXk{^^=V=v9v_l^$Lo;D+%G%46#E=_gL))4ZnBami`;0h>(_jHbPgbB z70SUcxz7N_-7jnzcq-Q)`xTkTAA*Q-hwO+o>bTb=KOze3mrsbB>8u5qd8!Xvq4gBn(- zdVaIJE}IcNp;QSpW9d_)Q;&m}A2#{e3)U-F>`6cd{aJSTc|ko*{!7MT=;yDoWR7Lqbq*)>gxQ7^;GVg#0Ek0Ayi z^9l=C`Y7=>XlG%4X>um_x0?6=4>kVxdZ1X^(mVu?1^co6erAw5 z2A%g7B!JrkY+IS`>hGYVyhMy=qF$|bB+YKA<9EYi(H3nAjuE1*~1FkDUS|9nnHfBSXUUpEv4U{>|qd}SDq39{s4g&bR@M^+o z<2AmSvj*IXbsP8Lj}6p|AC z)me}cXl6kS62J!<5x&kpRR3rwVAghz!^`~5;hW7)Ha{q_)Tv9+MB^NF?NifgzpaO< zpvpmI^N~%7dp#V7^d2YHUYnUx3Kz>Ot>FR3LSduP?>*;0!IV(XLDtWi1svpNQ|nn} zH;!z_V`pY--czWKdVVGwdjRZ!1$gQnSnBe!N`eairhd4Qu=L`wuL`tJ{#e z&{l6Ez`~RQ!7G&QOI9`nAdXj}o4zWy4smg4!#;O6H$XR>M^ zh}4Tap>tWeJ-6;5Se6RTaCo?p>3e}a7Jt@fKP%KfK9oCKAua46z5S&KxuLYne}r(g z9-P~dvx*zMg>Vc9Z6c^}roAY5CYXbYWSjbl15cNC`Og7^f7lCi4^XecHo-vfg6u_e zQ@9;Zabq6Trp(y@z1filMHd!sHp~sEI|j6Sw#v(>d;tGX5inIh|LuK&Z?ptnL;Ne!}IW6$I&** zzqbnGTI$kR;)2c5i7)EYF?YtmGYss>g#>;Lw6JRN7vBA4VLg!$LLR)(y08=eHj&m1 z5ZIWH53BBUCaI^s2|OM+xtH{c1{>T6qp6~&e!EO9zV#Rq1*P8hM1X=zt@@GHWc{FQ zBvIgnY4z!_GCZr|qhy%_2(uuLBbZu3v%}`;V?_Mj4;+xH@atnDcpXQWwSso&*hA7~ zjdoDQpPf~k!1P;O#Ry}Wt}#EEVq7vg?fK;QeqspRAMK?xEvY+ilgak#*6WRoGU!B< z3)G+>Miklr`seh<@OKgklWPNXM#@0}-np4XKbvC0Q2Dm;*KIJE_h)x0!|SDgL=E$x zoA1VeVA(L9V1sppWzIvgr5AN`XQUu>ZsRrRdXUfXEt#HK!BaozTQDfdr)qp^d-JvT zY%CB)CNI;PocUqC@NN=l_1``;Ni#%PEnXN3L8$UK*Tu?&cHcQL@B`$;5`Kx9e^?s- zPIdne{!O=3#H^sv?j5Mj=GiBYgYUmb>}PQfkQ<257c>bp7NU^O(ljxLNU4ajLWWa6 zC33$Kk>>PD)oQ4981?h20I!xr3@pXg-2Umpq8PVPQplgjxYSqH0_7 zwXrK^PCwnJXgI(f2?EMg5>?kdOeYn7UR{mf=O4Kd7njur8Ar}TX2AtSFcZK@ZKs$m z83c+6|0>h|)Aw3z2#~puGu`0Y)3_v?aX&+#7*sj0BimCVCRZH>tsJO(<-mPVXi6;P z>8gV!$4hV0Nsc3gbHC|FVqIqx*GjAFi91!Uriwjr!(lj&uN+{mn$whF|9YGa#U!P5 zwLL*TMYA$O(VlO8m*v4yl>*c4XCmtQTmA#=ru133de;?cAKTi>^)`p6H9cMUh_Lub zW9|(0>Ml0e?RvSMUJom+**E zt3bs*w+6rYtcS zvP||SB$_NoLeZfJVH#8uhD;bsrm{v6GSze1epZj~?_xJO?mk+$?c8J#R9_akTk{yZi*wEo_Y9 zX*01$o*Lmi*kbQw<2mC;SC8G$!hH5y^yO}n*+K5Y#^C~^%4K?<4k5W5zI05vJo{(6 zH+Txkf%P9DMb5$mrdQGj9bWb=xemX!cC}k;#}^#ptM~2lZ8HY2s+`f|_cKVqWk)6n zzP%x%A;EN?_B#goXkcF3dSMSI=8_Ch)LjIK0A!grqkx&ZyQl=pJex9VKaZ7y7#`7D zoK!}>8I)KR{A{ll(QauNZBjS=@W8BMr4~b`bjJ(nvBDr#OJqr7C3b+!1_}UI3h1)y z+t`o|5N-n4mtdpHtb`5FG0Gm?$X8rro;%@qw1Qk!?d8(>SJZ=CvNp~33C)szKyRX# zLlI8>a6QwL&kSm5L> zIWOUHFP;?Ea!CNG%u+XB|9cO-X5VrSf^V!umfq9!>)CnrH@VZz(QLeKN zUTETp4N29rm;;=_4)kwGizuH2-g9RwE?2Ohwiy8#+dhb;=c z>875tu}%vKv^!}11Ds%O3Vt79vNa|BAFj9S}5SQCiYTBD4oKNUFLc_oB0mtTq=X1DAg6IR(hYrUPX%? z`AN|FS{u>lv!{eIb&T3~N3lQ<*oihPKb7?(^Z+oIf23*T3kSt4^6me!m)mm9h<9n8 z9((d~Q1Tso3JQ!~Fb~R6zpwWi;V#X4C5&M<#{s{t1Ur!pGL>qPCvE^os^)A(9Hu{H zKh#Bc$I^I3b%55v4mDE3OTN417O`Or481x&t{KP$?Qf9SUPq%MT7y44v-iGbj6g4v zHKt9He)1(Ty`bBfUt!CFRU4o#Uzej1j~?V{_QyX|xwaId54i2LVeB6S>1{Q*eXM77 zx+_u^hq)j4kmDW=I)~?o@6>wvDlQ(x(d|WMvzqeo*?doPg5c}R8Q_DoyD^VLq=kismSDc8aLD?Q$bJ^9o2YPZW(odru z4IRw#Q8u$|&u*T1m*Y zCWxuc^%hCnN>dGe0spy**PUsSdVyE{W3#FrzJ)$UReb>1fxhii*DWu#GrnSkPTepo zq-#;FJ-JH|5pO~l&gV$v$xmbs+v7tWEL1(DMW`1G2KzqAI+l|t+gg870YEAxK03d) z0Xo70@&){GDr}XP70KBE(Kupn*`~#`a4`9HrM4K}U0Aqn7(cYza0&I~aZO}&mNsqQ zc9Q^poT}`+Fq_=<89&I^S2oai3bq8lOFl zmO>0GNQDUYkYEd5nV3%48q|M#4ac7Fa@z*b@>P^;StUZ1`t@B{!N)5k9o;J4quy(v zpC5e>hPOjWBfjs$LTwZ{(FXy4*d1OKx8IE^m^8(BnB*78Vff|=BW&q4r~K^D2z(&W z-SF*DY3H^8^krW-=@d3M`P^|>M1d!1POV4RpXTzUzRuUyk$&sYtuh@`{i48{;9jJ5 zxubuI{xxD1>hhLI1mChIHVO;-bKW-#G>7iE9Sc|N`;_=8Vpkq20^5=3H%H?4+j+(! zSh8#%J+Ve^Di?Y)+N~nmhjI@0ZKt=S8!0n9Vid?f=O4Y*!CPf5=4a(+jK$ zI5U-maG1@k>PMaD59C=3C4>r#sn!=j(9v!sFsJf$h0sPBtsK|7x^9x?QS|Dw?6Mak zv`M%krmL&sXve-e7fg6cOlylHWLX9`&niDfoCykwO^2vx2gj!GRL&=K{&wO2^YPWm?s~uF)GWw5sEY#SlqJ>*Csg2Y1GH98kF_VHb)tiybHTDW#5VQeL$GU zd=|I{_BxyE=Rc9r4>8$ZvtUk+kgo0>f*55|M)y_seCEgQHgr%gOuH^dvHS$dXa$;l zYJ)ZFI~nQ7o~ zSuO_r^1ln;jj8pvz!WlW58&d5)f}Tt<$bnbDxmiM08>z) z>>o=e#`QOsDyTQ#d#iYU62D-p2_2Ks6DL?UR!8$h@yGnUFWHlrdbKkhbW4d>T zhMe4$6_l2-*5uvVSH8)twz`|z5FxE)kaR<{-54jJWDaztn*Rho7Yn=S>)goMmOI<% zr2BlK7=KsfZ6jSdKuF$`YoLJkK`^$GZAZr(s56L0V{if3@KAw4;M2B$JXGo!D0ujp zMN=l?WBtBvKOj%!jPcraBz0Z|gv+=%8gTyI^x1N>UX_>rpD;5QIQnmH9Df04BAx=y zLg^-~yn=$+2x7_P$9^8Y0tED3(uiv$Dq{&i!!5fqti7h1Lk{298$(x z%3dZaoW7XCp~QzAr)EollHH!i0qv3k939H$AiQLIpDc62^I+%etL}ubqVAOe{_cTp ze(*%5R?h9pc+g^CQeTjh1z;i#n3sDdF_Vyl(AD?L#OoXph$ts^1;!Gh03=_E5r@`! zKjD}Ma4jk4A9_f4+dmo@NnjfwZLSTT>D=$`_F$1AccFj=iB`!dtcz=MRX8q?aD&b) zoip&8+V}Y(?wdP>d3#^Q&HEVyPgj#eD|`uQ^3h!#n40hJJm9{wYA2y4YSpKEWR9QY zji7(%*p(Te)QSK#@ONF+?C?VZVVGa$8bJ9ODa+o0S^z-#|A}q+qs86caq{jS%-md* zp*+dFTUM=|@O4uFq=bEA?jFXfJ?u&YOz;WN z=+l!9!qvDImJI4nW`z4(*`CdpBsyc8I_nDiFfWA8%$0oHJwGq^xK~wSZ~xwYLE)~R zl!Ph?;z=iemDarm`eOZoo}`~x+~#p@K{J>tc*S@W?GC>CjXSVUU&%U^Ijr^aX%$HN zf|S5zt3Kj4Be|?jZ7BPcAR6Tllq8$q^ZP%U{l3*P&stoF)c&dtE+MGBxc)xEeb&Y0 zimEY+@MBkneeC~`~~FUmiJZs3iA9BK7HT4wsdNNttiY=_Fj6OU0+xv))o4d9F?yhFyf9rXeB+x z1o!<85fQ9-tYm5z4qNwytcarf$Ug6+;wNNmlorL7k8gXmlF6QHckFswF0nH#{7TcF+VaM;}RUBN#g*yX>C zTv)1aMC8!0i?^hBVidZ33iF(KOCLPz4^ZZLlDSTcX#N?B2b~jcECdi7TIzb-> zgMX&^1iLg(J#A1${M_Ddj4l-zLj!>djpkX=-6ke|K1ob3BMlB46R@)7` zLRnOs`8Qo`$tAIY4!*eEuc%hygY#Z>$K;vp0^uNAX^BZS)tqzP&Z0SvB!iw9by0OC?u zJZ{;X$8n#L}KmnYOEyn=^@IDmSHlN!9)&m>mW0tQAkWrj% z#VlYlx6jgIM&WQfM}GV(sZx$@? zT*2Ex%MB1jzRK<#E5_-`uY8ZcDd+#I0cXqaa2+9_{jfvCLNrOs@G1h*0Oa7Q&>hSY z1L*$$Fa7*ZXb|q(E5v&pwlgf!s@nNFMARzQbL|n_KPI7*PMZD`<*=8%#s>L5kNHWv zYu0;%bC%@sz+C_Twx<8H%ztcEQucjMS5B+$tuWTNyYv1>M4fi$Q7yb)=8aSh>sGDwr@kx?Xq~Hr>XMke_goMYA6P`G6{5bfw z5BNQBobm(}`;}`%XSMDTb2v~7`NzCD$thpdNTc0{Qxlj;}c+F3fdP5~?53b@>b zo0Sq!yLl8o?-ioIDq+}t&*6bLhkAIRff9xiAsEv#*8Ke;>?7SQ#Q|Fx><~qV2hPf| zl+iTffnOi~i$_vHT0R}LkoL-&WQ(S;|0p^9R%Z9@8>pg$D2sTuaBGdO)fWpR zmPo+GSxW(+KDzYyXQR(7U-e9m3VG8=UHx{p)c@V#v({QV@s*xyw;PJA#CXO^3577Jx`UA{$5NYWuc%ab% z56ouaI?y3_;JRWQ9(Xc`2U>0s+bTFfMj>bLz&ERY7}6Mq&VdyBvg3i?z6?AtU5N)! zx4)x$SF;&%z$gO8(6QkzMz6x3!nQ@MA!q>z=FUG_PKK@6>?Ktb?`s)1l{kd&-xe!( zY5`og^&m5Xlq$2QWnpDw;kNRhUCV3yn1i0V5%)$ZI8y^!3gD$>Jg#I7n0O_Fol+}Wu&)>SGuXe5~DPwTK2@7MCU0--p1Cs zl^#`%jDj}$Dk;vAwA#c&^R`9!TK8oyACS1k>4#<@Lu|7SK#+vsZ63Bxc2RRM7Y|%m zxvdwA2cT#?fD1ra;sDt0ZP+h&UYN4p#3F83(a(EH)>#af_Z&-Uvr>m>Z~K|JiMj7`|4*&2)uWG*jtJn(@6 z_bK|fTYJq)AJvA2yQjoRIW=tIpI;ftmP&s6qInvQ6)e(x$JLOnZ@42P47W3bk>KQ}{%SX{iE3jbJz2MFBb?x3hv%n-xKv2F|R z;T38nb~#`=1rHbktGsDp0AE-?Y`y>;jjQE{?b@tCf4K$LtiV@}UYmWlIH%E29yoO% zSz|c+7R11SJs!AAiwF9K5WNHn)VByt`bEcAb7WLn*b_v@618g9=XVbDWaicwHM~9? zltr8fYV=Q`3L*Bye4r?}f*$E{W91#hsAO6tmdADC0qUVA8ZJ+KA5vZ10#{AADMGMa zP?vX_?nZrsvPM@z5rgmN$SqkZ9ckhGXiBtekFGu=f#WO8I|)AY$P4#l>~m9R-3!gC z63jEl2`o>loWfKxv`RW-p_wR!g{qlcD_dz9nqYXwsG&%)$>iyd_`@QCXO2Erg!O5~ z()>B&&=k4>7pWRpNsrU9cGNUdc`aqAQ0@e^mK?KQfH*27ir9XQw_#ge4{EFz0GoF} z^*gg19F(f+xt_ndI?6;Yrvwfi}7Xr*PUFSNj#*!Ws(G z4f~Yaf(J-Y=*UqM)PRU1(4bYmIQpuDwIUR z!@aM0%w2Hizpo+dW@ImEQ6~RxXbYDY>uVj_Ed)5~%XZyY_C%sO!)fdfO}FY^Oc2h&)z3Mbav* zm9v7#>tQcVuJ1Xc%lVb?weIKWi!~CQVY|@@q`I!vvT35W+>_`V_s?>#1bWT>Qsi^Nh{`6)1Tb{$Rf(7r+H)nb4dZp^|q%huW z(Nkr}8XUx1UkYX|hznEA)vc$4we5vNR=FAuF7RcNqJ+D<8iMOIZ={9sRifnIV=O?- znU>)JQ%BI2;I81KXf@kB4>jBQ3bK(E!6ay5#0Y}MNa0DyHi6Nd@EMAIs%^*tG5};A znzHZgd74VG>$YTqX)3G2xg-7veSgFswSsMu8Mx!<=!+PYjQ1Fd7LgYlmg#%pH$*67cup7?I4sb`|K@?D#)DWSGYGM$|SRCwBKn}t}Tz2 zX}Z!xuqMYEcf{ky(Jd7F+=vmZqZ#%zw4gv~HUtxA3&m9l;(|*F21a3RkqzK?UlnN0 zP;;PUF;3jqk~=#hk&WK_cp#h3AN(e7tkH#5thxECqYz(uA@QK;SKrHM`07DwFvp(z9g(k?nY(#wwtgB5{OJVkHUKV-dbn zH2kP#GTAYa_UlXOX6Xmal>0-U_M7CZUnwIk-sNYapUk+>@3;mze|`SQ!usa5oBN-g z^a^VgX1ykybnDdi`2`-?H(zc!KdV5M3UO}rcoc=(CfIOud#rD$=H_y-<@iNYCA%dm9H5QIgBP;jw~=UsV`*8Z$2w~ zYleiLwU*q9aC{{+IgpLfmi2m>Wp7595G*wCQ!_$pPDLPCN@wFm*C&sYz`o&uWMpv(cinh)+UbP{BS3`K#ZIB38T zHsOKhUZTQKZ@M&ev6|RW{-kcH4~oXIQWlq%;P(DH*!7y@caj4s~&-Z(?}DTt3;*B*z-6&&q@x!Eo*|HaFOSdaBgkaE0mW zz6W5t>@<-70Q8y|X4wwdc3>D{P`-zPMoka6g?sS;XdO*ULogbEYpEU$9yn?F6!r}^ zga?qUY&h=de*ovm**EXuvkjqdBi3(}mWRG@CKFP|TFH`(yta*Nt8QXnK7Kg@#}0)x z;h2k%6X8aqp$r*LTm!0O8Sy=+D;si*+ke_24tmFVTEpeXiF-%&#zRX?@|`BRk(DE+ z!|RAmwbM7d;_O|{9M6)u9nh$=WN=*x9au|2y%gwK4)59eZl^UA%YJukK$SmT14<&L zU8V0nO#_Aww)2g!Ogci3=N(22`8m!u*9c)y-I5bwa^czQsGG#8^F#Kl(cfJj7+L@i ztn^`HzI$JSp!YR}^#mr6Bf%mKB-Whj{r9Z$0}`9A&Z?}$LgA=H|1v@9m8)X!S7OGD z5M*yNhEtqDNm+aCF3G`=j6$vTTQzH%#R!q7DeGEf?4{s0Z55NbQ$IX15UnP7U?i11 zSFNI^Y~bz_+>$_n*5fTMQ+@N*D)HR-7T5!Ckg?vRtAQ?_8z_CFe_3ZQF;v09Y?KWo z9q8ZvIoiAO(7Ano6={RJPu=-X`;p1M6anf_v11J8Jm{WFKEKQ6aEkC?#MXe78M#v! zT-s1HCLLQ-mEiGJ!Xv}Z;Kf1G$ahPUESly5jKpR+oeQA~fNY%<&`4f5ZE`aExKA); zTh^`9b}Ec&Op414^k+Wo(KNM8Z%|qVO(RF1&Xo^@(?$)ie^KY?=s4x{`k|jx4qq6q z{SVhP5p+#AaCL0IEjpEa7_>`a9c6*41(_lCp%a#bWpqwl5QU(t6HbX zcQpGA`#F+E?gNHl9+#x=WvD+>M=I-Sep`O5G?_|TP3LdF>aAd#XLGPLXi-veb@)p_ zr%YP7NPY1g4v19vH%nxT9bwzJUdT2RF) zdwVht-guC6l$=bGzguuLUsDA4eJ_oj{a{PYyT|cRblIf~UD1d{zc4aZa5Cyi^SuJ@k;I#zls$_Rfqf*NV>@7Q-oOw*=B~fd$~C5yB7fTzSNU8U z6nj5?r}Wzo>+^*`L20!gs#{OHtj)^2~+np z_P)r8c7E4of zd%cyitwl#*MXARZuo0E^(rMa+Xrtw73TMJlRJx5n$K!ck;EYfm$wEoPullO* zcYWX|*?Fj2_daEmCzdGwRH(W51D?tPJ$a<=8_UARJgDK!;5z@^tGi$Pl*%-<$CF!2 zNC33H76&58Aj0vpE8v2wI|6{0BD)iwv*gw4mxKuJUU@=c`kY_@;qIjP-W@TZQ67{Y(HwNot0cx{$Tb`)U}tvR*9G!dYh?xI2ksiUPzeB(u|jvm_oHOv`_*ReiTOW3U48Y) zPH*+$y0}W?T_WX!X+|J{?WDWwyLL2 zp9`#BZX;&yCB6fjiKMt)OXr5Tli%{er2aw76Ne=gml<24T`B!_=T|D)ee!aQm%YoF z0h!#SUW`ZXikPcYVX!45|7yguNyWoaxo@9Sy>onQ0;j0$@?BPp_$IhXy}W9gtmzN$ zmwntN?~4jRrIL?=YgUZmuy)OPpPO-4QTC3VK1IH2HETlX!(8)Y;ezPhkFe}HHo>`?lO2nMuZUKlD_5Ik#Scq#ws zw*4tbf;Mr{^Oktvr}aQert!c#_#S9;MzR@XE@Hhu_KWc5;ek~&WGNGify^oWZJxgk zsu)BA|7*ooxSjx95^SmoC*t!E4&T`#RqjnI!2>^!Jc@)3Xda?LEs!^`g{!}3%}A{a z`)lBzFtL;53mfU;=adiY`D@&tF(~^TKS3E*S_XLFuW|oc@38eg9x%IH`fsAx>Yy6} zSy`ZsnQLJ1mYI#TZ1Lnua2!*xvZErs^y#`sTXNTx53_ZNDs-J0h=L4Gy0pvOn`4d% zAoe=lOEfFQa4hXkkt&i^+oA*iqLsZ-`2Po)I3SctIT|yEUrNn*ZKxErY2(Xt;#!>o z`fK6Sx&gENw2S&;+D0Akm|og`J*I77$6=e{A9ZaJ{?>ExZ`3caY?O4VEFOBy3~NTL zq{H`@lzxyf3i1B|7yOUGHO(v=o%Gqw0>@^mcZ-!^05%64JZnTRqZYJ3nno z}`57xs|Yr|MFh>AA_a; zSHV$StHzoO39-M|N2Zq5!JurYi^$Ss_)MXi_>-9VNqqf~2J;QaoY~HNi_G=OUYqYn z%<9u$3M}qnvgIOkmd3@e-Dz>>)U059nC>@+<4!bT#sgp`M(}T3{GS(&+>D>}Kh=m_ z8dEL$eDRIpSyy>Cqa+b61)IwSUah|Lpvh9vwDB%~rZlcg#r%a*1N1`sr+Le*g5WS< zs|Fq-S(UO56+9bmEy_#Fe*^pU@^k;lb}{e3=zLdC#c*i46)i%hk{dBed1`%G@k+O5 zmW{6Aopci$IxQokLLwD?UK(9CDyu|Bqh$x1@@R90rqAb@s1K8XIgA7{sqK3I&KSwY z2rD{VL9x#7s!jkR(iXfo^|MO9X7c7o@GDW>=?l7EO9Ke%Iy%JN zrs75lmrOpPdzv(P5~|nyZG+W4R`!&1blPTZ$xY6e-5#=K($opAq>A)Ac<3jt?0eSi zZOE>AJ+HNhhoHv%@W(E`EBrN9K5c|=Jz3K;zATp$ecBynm~JDvNCeCU1l(aqS%$1% zVmx5E-@SIH*_QXYEt%&ruRCM5+O3-lP%miZ!z40~tOyeI?QXyvt<+fU7Fas}WRjP0 z@wm(Lj1qFW3BlfLG#pWe)*pRQS>MP)A5+1a+yDZEre)++QqND_;<{D^9JDMHtS{*vW z#uIVaq%M$8<`j2oK^H~^@-Z5HQ^ek=2Xj0QbHV5OVd23(>B<@018ZzGWF3qI_;x!< z4!=2xY6$kHI;Oh_yy@821!D^{;(c}Y5G0i*vcSv;#S%Xi2Q7nHE zcT5FG{Rh~@!|a&$fgebc{ef#R5cW4%(|=j7>>KO|);r)=Y_ZuzAI5@NzY+sDNRY`t z2%?1tNLnLTJV4(!78b;I3%7!+zqa!fCM1IUF-`QJ$jR&Ziypkc5-JPk0^;)AK{Mnf zBkadCN@MA``G&pKVauNY`WenaNHu8QJ@dKG))fi)rzs#?M6sY~87_KK0S|;Yaxn}( zBYU$LU?TSCtRp0|8mM!tJ0tujfM)-%EFb>o zV*7q@gy_p(>G2bxb|}^xR_59M&*8f&`0?d4Wl{O1$_}|E0f9X5%fs-$IEkjyovGQ? zP$R04WmT_8vVdGt)uoPu9ZLLSC{g{?kYn{J*>OFKP7OuuFfi`H1jr zzC2|P`Vr+h9xeJ$tiC#iu)jE5Pk6TQ%&3iLVen#E)Y@F};Ek=gM{_o|oH7zCKM0Bg zH@%Y6kKOVgm7l+!heYb}-lSws(dE4LQ~6(&lrrJu zrx%g2)V9s<8@WqhphAV!xyqJ%gR#773T3cxf16B z)9vGVkCF{=O=2z{OJ>P`-3Z7UzjdEj>hau_vpH;_?iL=*L>!tU}ozSa)jrd}GQ?iW- zGJml}Yq(*a7%ZSN-2f}2lCpA3L*q(~gfQnbo2zd~M&xkS*0MSmzweNkr%0tQ2rI4? z2J`3jVDiJX2+Y-Rl-cBzFRW{DJ*RmR#+%puSLIId`ugO@exuy%Sb7fa9?sV%Sa({l zqo5!ACzP1fJWkgZ7E5*Hzr1U{y5J|3#3wM)%;>Yw;Fi|J$<(PFYoIul2z74UFns<; zshvUpif#<_{kPTghwAT3Ki#l=*f+x8%h((}d(~egk(AJ+>-|msN#(UK*^aK$te9S6 z1BG)EK-6|ApmlpkM!Qq-z1N!JjO|+b2WuvA262{)B%))yJ5U+1y?oJeMg5oN=dVWV zuE{Yvf6w36E>{DgzA=ft$1 z-W#X*JpX_o`Ym9R7RC#t6_1#`hf#H;62mIm>;w#4QSV<9!n4wEqfP`z`(%zz)Z~q9 zzDPY)b;7Sl_BlcJuS0)=!YM^qNVRNb^{T3Cuf1e1_Ur14-Yx3)ldR@TDMJ+mdClY| z`yy65%pYZR2T9Gf-O|ev4WejeBkd{qq@fY`0`al?z0*Jvu#uJLBiM|vE6SO&E{|rf zDjHkl;|9isE+F7fx7Rpt50tJ{xKs`@XAtO}dnVig;(So|X8x><)Q#GKd_j|86M1`X zAorQkRDW3(kQUZ~Vtcy1acYS_dpTCxHK&o3V;!vIPCaeIclH*_H0B-J+WVSlqpie8 zp$__qXUzm{b4FLHI~dy=T>Ln{SrPm~uRm#boNy4VV&${iuDh-|%zs z_(FBkqBIx>J@-oU01s4vN#qvjwZkPhp+d~nLlzclEd$0NusG?cKih5H4C5v2@fCil=fys-O2EU1&7c;h$?DBW? zIGN3@m=`luzuD*6y6|1dE4-_+_D;K4agII6;j#C5B*v*}Vcf<|_p-Qq>0x-emoq0e zpj9S|@I`q;`Fj*R+}?KMCj3#-`CJ7DE{iU;=;!gvH?Oo*wF&OpPP_)jm?aTz|J zG~Ti?(!J2gq9vp{k~xWHQWz9;G0J`M)@_J?cyh)KQ=|8e*j#@FnXWWtaj80`M0!}8 zZ%l=M?#fe&nOpA1{KPK^?$iluL9t#s!)wf@nCi$Pqj1`2w^O$%qgCr>t9E+O%4J5K zuj(vHOK(8=@);GJ%k~}Bi!6CkpvCq9Eo5aB1`z5|W4Z)H6HT5@n})YMA5eV|!7os$f6=+Ri9 zreKs$F;i`3uq+1ULLmL0H!$6(goJLGc3Mv-(>Cc8)t3b;XB3rQeVo?N&MD(41^3L8 zW&LI@ufawKd7#!)9zBVfe7H8@FaTmQrYJ-k^*Mc3DmVJf@%DQQW5 z;%##0jYI(EO5yeJMIo(e>*JA? zV=WnIP3A^L$UWTF$adxs;~5Wa$_&s7f(y0JkKo<4P-8to2MYXltNH*YHW z@@24{s+HyeIe2cwMJS8x+*I^faVv z(@^ileq`mkd0e*>qT(Z#LT^N%%tK(u2wL;nuCjL zPOh21yI$=TM0mJQK+QPkj8$rOGo?qll$EWgQy-ik)E?vS41U2)^5DKClCK&{d+!58O$P zIe9N?V+Y#dlFaRPNE@FX{_?)gX6+_9q37BP`i=~b7D}czi97#M{kasQ?DRv+G4=Zh z{i%Y+8cv4xg3g&j|Bw(2kBe=NL2KNywsa!xJ{b*Dn&p=}-H@4-@XV<(W1n=b%HX~| z1j5)vxQWy=2Yz#qpc8$+%@Cy&CauBW_Ovj6&3Qswc`Y?GSa{4WCrW&_j;*QMb|Eh9 znguMoBYV6pc|==a*gIBK2BQtq>P?yzem$l-9J z`1=E*wGxPjKHVK$^EeTLuaXaZdE#ah!y9IpC2z6MZNW+jQ^sw_Ovuw=g$C}OjkJc5 zT_mENe5nlH$WVtML+msub0H1xa<~zK1 z+1OBJ!Dt-u$vpPz?manHvpJue>kBKhry~*tcFT8NTX{o2z2SMSW0x;z?`_~g`^w>b z-8bZ*CRxmDmQMtPs8+tZ=$T-DX2uNlz0KsCB2mjKcBD0H62x3Cxh3wdJ&HB2nrW{l zig^a=%8#BiAl{(#BuD{E#G02_1shPlg)?4iQlUDl<%MoJmn;)3_T|~KiMR4Q8=gu& z>p$}JXK02lF@E|4eY#M6XLI!VV{wVa!Wnh+y&0y+(`Aunw~CDTy0xd<`S*yfD8x{rSJ@@I)Ev9Y# zZ;HcpP-PpTf(xh>gX3lpB{J*~t=>(KABg;^&e%PCs0?7_caVsw{Z7KT%|0*f7myd$O%B=p3GP&fQ2Q zGQCWB>=dC(`$m86GAqz0Dj^ywaN_ftakHIkNAwQN&7Rpgm0alIQ0;};jniGCH7n>H zakI5s)&br=qEpQw&K`G{^ch0Jrt?dqaZ}xfk>KC74Vmit5PU>K!c+%bMX2YOJ z?r31W>q~ga3{rxA-aLhm+$;z|mS{I?D=&57{8q+5|58v%y;E$mWG@j_M>cw@^&mLj zEql`1wc7SIIeY0dPU+Zl_nY?!#W8#-D7zRZ2a6dkw^?h^3v{})aI11}vU^W?utfu+ zOP*7H&iOW2!C|inS>%m#)_DSCmr&6&k6frb;UT7MmB_Ct!7)AdwwAm}BY zYD7t_1%|lN`m0Yt)@#8cXBw(D zD|{Sf`Q*hjputgPZBuvGf101C3DH$uS6flCeD?ZR4Vl3O#b=fhTzlVSpOlJ5pSC_& zTZg-NatY}xTLkjmeGtKX5OIq*0_yoWofV$t94 zp`Y(~y5vREqp$8YyB$yj52z<;@ZF5(O4O-*PKEOSQraWlcQW1i3@SWaNS$mwP8L_zT;%!EY++_C1Bv&ncq+~Q z4*e9-DLjstKv=Hs={ml68$tPmkbH9JnlH1}^=q$GU}c{Y&*nGpQK=hRo7}wiP{xW- z=_7)_H9C{6Y&p>|sHOFMtXeQJ8CbmkjV}NNYlhZ<)tm{#2;B;5Ei*}D-hg9k@@(-E z`&A2dCe$1k$-^%5N$K=HDuZH$b{D(0oJS*&wvSG?sBSG^wjw|%X1OG@3&!>@^tO|; zS7{Vf?$9inoo00&kk)ZHQL1N%9< zhE+UJ^x>9d=fb`hk#MV+`A$J>=S*gZ^C63NKSC-fS|{jUHTQjk4?W#+7$qL=lDp_@ zZ;UQkljqJJkGgz}WBC#_L9aOr@3BIEJTRZ*K7~cYwi#tPE=RZ36UVpfx6BpH4V>4o zn)2rDEi5yNRGwVk-J@gj9vB~^;HFPYCR`aSvW+lIf{c_>p zEh38MG`3b&qM(}(Bt!3}7Xb4kE(JDCH~KT`ue#~IIu*@5B-ZZe@FDLs@l1X6o8^w2 zvKCyr594Yu^f2JcQ&>q#{FhtAacIuj^Aos+9E8bIQdfs4T3PK{e3`cF%GTy^X6thU z@8g^f@z5{S(YFk1PV>kW_u09StVwt8ZS8N_KZ^X)#`@&dEE{M)z||Rj-z<0>yOX$Q zTe_|j(+#+rg@1k3h@&0E(71LD3zzD7s^?>yI1;0mGm_So8&Gah1>1B(>)UG(^`c;f zsJ>n4f@$yOgHx;0TGRPVEeAezDY>M@MwRX_s7re)#kunG;dN&-hEQ+K99!hq=Dsg$&yI~bUUQ0ml}fYXOW6Rw zjSPEAK9F>7vM_NTYG}m!wA~_HDrqFae=;7jBNAD5nS3Ge1lWady0V7l$EX|29y zVtUrWI#lK%SUflwk~}tS0klbeTKWGS8^p;b8><=Gtkv<61_16n|=VFQ{sm#z}zrr{w=-$ z6Ul*`$~f#Ej!5P2M^^dYKeC|H%xUs|gwe}|x+wZ*H8rLxLBNBQvKEX)Ehx^SrUzA- zzPu}|yjz4VR(}z?&s6x98O;B_#7-fvClziMHsA=g34!_MT2G@PA*VN!D=wo z9=v-YzktEJpA*m3 z=ms6wsR6Z{ohUTCy0NAWNK!D+df)DQRQ%@% z`EUQHnZsQ{yUbvfX6*t;u8h~%1kbD+t&x#j$%$`!Cyc0jQrSQS>rg}5A6CE*j@=?Q zCzC(P8LCF3aEg>UE?P(P&vUUyXJh}-wP_!UJJBZ!qxwcS5-Qd017pdUs=q5;%L=UZ z*FiJ}X8KFNUE=Qwra!~s5urcr88=*I#tLiK`3Tu%hz1Xk+pK|M^rKMvKfaCymt`2P zAT(NEdM)@Q&A7I6ItE6ohi$bN_(vPb3w;CcRg z(~X~p`+e}E)*U}D=pQT(PL7^X)N9p}V4=yQCOjD0AXypN8<7l-&TC_|sdnOxNjr>9 zC(Dgi?YqG3n_hXsA zcfF9+Y=-Y2A~&#y3DG>WniwqF4TtUgj+%W46>M>%sbO#3P#PWz+1i_#I^5$2q}t${ z)W~RTFnA`W*RV!)%fze+e#Rv{Qqp>(lmj>7{G zUly8d0Hj>uLUS~qPWkQbs;h6#ZE1bgx!u*^2xA|$cm>U z&7O~9>uL-qjZ}yCT)Q0a$9P8Uu{7o*{@Aoxi%C*Y51#lZZHx)`y$Fo?-&M^)&MIKv zyfIuV_cV{&b*sPBb1@+Di>Y5Pv5$hoKM%g^P>Z)H9(eY!4iDTr+`uwg|AZWiZ@$dv zfYe4enzG#|oh}~ya~mq+9!w5|Vpg}^q(OTRZ4i9_+=d@)jB z;o8^qpE1=7wkLUva*u2Fq|LR2&b~6bT2}Y$I?dOYQrg#hFH1)pj5z$6Gt#><3!|D? z8rk){8vU@svu?H|O?gaPyiL!1NsdK^rWquuCd3B0dHCqr?0=BlAiU^B-1T~pOz%NI z^4n>RpVnRWGS}T{mb}pIow24*YOeJj$HKXyjwfF@iN}(F?Iga?pWZF~=a?bbEcEh| z!)?+Cr>|KtYF-bv^BYLrL%o&W)~?y@Ry>q?^bILZblSno0%EvN|| zw)oz|f(ccDCC>((p6%~m+Ap_%w6OT;H|gL2jNmT#a|B{<@NmO3fkKcmYBsRU(9%Mw zeo#Q4=fn2ECG`q#gIcfI37@? z@{>~ivsn=@RI)qDKcgpmS@@*wU6apk`xKsA&gi9g!D^w~EV?ZGV>VMFW;`QHDyaR3 zvgAc~=@I$MOc5}*$!Z9d@i#a!dxIr~Fta@6hCAo+wSBq0&%A^M#kCcKkM7o%##>Y* zACIZtw%|NBxHBtDiPFuU-tZq5>&zIRY}jDrA>4c1R`X<0f-o*ZuSEazh@nqe6Tqk+ z-_AmCju9w2l6S_sF>Nh9lwhhes1Vl;w(c<0#KceixR+ze|1+J+e0v;YUEX6$@)Q(3 zLx>R?K}^Cc*db7Xi~7;uU`ARGrKXOYED-VX;;YlqX8NgMIeEq2?&a?2!7XCw^v2Tzz3}yiMap)UTzH8dlvbG(5N8#9s?}`PxdF2> z;nuj|J@(XA&GZr{mmf)Eo41};y8BqOwtMc)0heTf;YofCF8BM9uX-v-d)p~MIiB{t z{f`1QL5uq(^A?=ARj5cG7mw*gpB5nKe;Q~dd@8RqE$9o8K@ESq`?m^j)td02bHBIIt2 zY-YAPdzqib3!I z#>tcdN);Oju*1n1PE8vRd;wtJc2mjw|7Zat!%O%CrP(4Q+*IARE1DlanVapO_@qya z`e3Ms*jur`_MY^vgMv&#Ij4%!rzQbzfke`DZ@|3}$aVna=d>V0Frl7oSodp3#76wVEo)B3 z?~&^bjQ@=pS)dZ+4S*^r)&Efhoh`x`EvF_ z#j>5l087cih{~K%G98XaAPuA|-Fpr)$|qyK>b~Nsl}BZ53IEB6^p~vDi@!3`ZU@`6UIn!LIO zZb`bADE?^II>G#R?v-HeIM^>a!f9*{IcE7pSSb-ft$q3R&5eo}^fVW4#hZG~@3 z3F1fxFh{=fPj`gbLn1Z&mKt!)LH>R(gG6ZS#K}VG=xNQM*ctUi&%9xqv4rk*m<#;o z%NASMN6|UW0~Qc4&;}KK=|K^CNi>}0;#^yMV&a>K3)mF+)ycYVgjp!T;6#1oovDyK zGTl(FWAycAv6rtz5!>-SY#m2@F)}J4x^g2+dBYFH9Wv5=02cxlus3?bA_Xy=rl4wi zGYb2|?#-uZ`DIGb?EPs0U{i*XvuQCrEpV#p&0I^Hz^Lwywv}thwuslZFedNJg0bVo zTiv#3`zGB<=F)i11f%d*AzO9=4<$T|2Ek1a*!CSYFguc9H(f!g%M*z>OR(;}yas)R zf&_C93q0E4zq+?7S(Y$E_SKE=!g8KeSofh=^P3O#$T`2%Ta#v_CxMJ@#f6+9~|vG+WaAnmzX{v zD2XB8E=OY?>yGzvA!~(rAGlonut!u0wI@T@G_^m%$mA|go08FVJ0l4Jr!%rI2*nEmno}5t z))lgl*tPt!Qr(5oRe>}?yLQhr>F5xffcBqgHf$H3i(WJfo;8sI(cNJ$)yPiS*_7Rw zN?!3%k^tEjd}%wOO6)WsmB9(bs))0MNb62GK>kgtA9V=)i3`|%ezBJTCAZk*s)o_X zF}&-E+Xgj>MnjOxnx1Y+9qMK#R9e^3<2}NlRaa;({fga9zVh#AKUSG3)3wyt}9h~ABwpug`s8@!??J zVn+u4cXQMCrw8d^GL5QXajyvm`dfhX9b}ChN;ggPg5h%KEkdG&PbLv_*aC7JGYW4Z zT+cHwHO9s=nVmlu>-z9Cywgrf4`!?x^VbT0UC>bzxSv<}8!VND$CD?y zQMu$Hw~g5{crWBzqB!nqmB}2Vk){h{Af-r^*4+q}mZAtQEwPD{Y2IufFg8A7=&DOr zgf>UYxX9Zyk|)6B`7Jdj2{}USfYI#{g%}2w>wIKqyuyGMCx+@OQ%T28l zV76o=r3uORv2XlM!pfb>(#VSQuPY>;dfg?PEpM1%L}g;h!F~7LOVC6sVt!T)m)ygf zlDY;ARc6LIne^ZI=O+y>OnG~q{JM1LL%!U&n7j@-*t!XHgvIQz>D1O0+ z8ioHxpHw9+g|nf_V^Rhu0LrBbak7kOnWd3_h91t4<+ipk2(`h(?v@0a)UeJx$v zYM&3Sy0g*Cpm^OdTAQ=UiPQ+5=m?y)G?OSO4n8Duhw6&o)8(?v5!Zh!?(% z$k0~4c_u};#Y6MM)M;on&2SOd-YvCEpBy6t^W*DVX|^>XMGJK~|6J5{wp z1b>uruJgZxCJXn3F`ZiE+FC*MrN>j<3{e*z@Q|dC%+>Shh5;y;zc15>2+z=ilw}f; z;mU;3x#R0+Zyc&onD;E1_8FTsY3?5b+eGX#feFIC)*97+82kMDcxX(r-o|A%9XV13 zrQN5n#ZuT_Z~EF9SPy6gZM?bJrFSE~1s#V|H4A`EfM@l3scI-nK`T@y9S_t8IaYto ztB{pK)@`$!BKqtCH;L#qKD}|;X-w-9h^h{Y{@XZ82(Z^}N*}03<-;#mbETyp-eB88 zKn}lY?LLKkpA-B=C~p?7j~KHXjXR+BNu$_bW5aIqmXM~YUWEKLC7V?M&Pxdo_?)`* z@muV_eu_B^$nNZ#CqakfAq0COx3oa1m9}`9F(f#njP#=6a;Ql2<^3~+vCy2qf%K5> z#`D{-zpYdD2TTS*edlS|{=!7`Ur^EBb5=RR{~x&c3;zl)OsKT9us@LTubM)S204Pe zmb8Zr*=X3dE@=B3-HZ$U`Oo*Siaywgg!{}z+i74%@yYD3I?aNp$a%bYM0{!Nu%veEWC%iyg8#e6(ec_*?rJ{ev?~W&G^_u+sgz z?7!_s^_2hZ^KAwG{Zl~WP+54D3t@)Uj7BfeEk%|GHrNtgyTucj|89SWucl^Hb5clM z9(PKddgEjxXg=NZ;Sr-N7TFXl&?MY!u^-V%Eb)H-?J?}^WJ~~<*|H2dT>9i&EQans z>gzVMaX^%b-JO~A4~tCVbd5+EjcQ z?D&J+!xHM?4((A#P2nhDy`qqfaxi8%kroZr*@sWEerAaY`aWsHH%9p07%cLr(U2F; z9}F(~RG^2~sJ)IRMeaUcjSvL;#evqO9OymOkA3$fg`1|fPir2o{Xgw}c|4SB`}jzX zq=+mtJU3 z-8vEH?VR^~-uL@^zwh_+Ie++!dFHvF=en=$zOL)OuInzX%(IHew}2E6-&ej@F0Y1X zslU+*r1FL8g1Dq9SkcjEM?_fH00f%ht$>=27n{Yxp3(?og!XM_tKk7!7T{+YRJ)Iw z*c58R1iK|$^*i9${s3pGwcQU;5iWj5)8T_C*S}HsH|qZ8y1xZ>SqlEDuCoT(sqx4K zra4Xp3U20yN(^$Q%I*rEa6)Pf;&hu1i^^zpm-96?Vv|a(Z70mP*c_y4L$3)u^$fH& zIFiBsaKXo6-L-*x#j=afBLVYnW;2SEs464owe^i}m_GvgP|KcpCtL`Z1y%Z|{{oi>3 z1YPCkSI7)V{##&4ZPhda+gbu&AvE_=&fDnBEdcn}Cf3xR#wg*g;$ed|bkI;t9MkJ5|5cvbl`lwetk zp=KFsQgRqGLxj?OB+h12LC{C|!&9&i0~T6P*D7Iz;Xf2hiiz_RX9Z<>f9j84aT2)#)8`94V2Qd=+m**y(#N?3;a5&m>-oE7>EYB z_W=s~g}>!}WvwiXvzzfecFr*N0`XQy^&3`Pn#Z|jd*`FPT0xa~ZC~_xAI%GGyUy9w zD7OsDMLt!QtxCbhV_#Ui^Ei}>c~%81G)ZgQXx${kj^{*Q*>6@SP@>!{=Q4J7o%?~= z34JX412?IA3m4Eiot-tddAgeRERMnor!0mqA~TFf3j*^V*k;tunpUAh=rFa>lCLJlw~982HeWeWnQ-UB)X7KO@UvpC9pWbLlnC7t+^VBju8?gwKRkmNd{*`UWHj-;B?J6(z3UvwLnqz}CWU3oNg#5w zOLVWMcfv)tcx3A*)R) z9>k=&ZrUArV+T*Vvt%uC@D|fs5l4%bmW#rq==4-Xv^-C42vh$3_Rr?~Uvc#(%bB9X z_E;x)sC}IAJk8G(G>JF%HtOtrQ~ip4Rx8<#zQ>9u&r#DqA4%+~iyjx-9vfK9dR_o# z(t;(zPi6a;$>1y2yXLQ^eT@=8>sK}vfYo@xMBgFdrdR}?g-U2yscZRq($1$DV4>1@_0&4 z<;Gl%1JC-M(h8*{^hkRi#BU z7=};yPCOP{X_^q&#=Hw{v?W4IAdpMCgH=(;zT>ird`Jy1i4! z^umzRuKl15z6B*tZ!7z;gaEJ5Jd@TYIKlI*dv)9`r^x3$GBGu`7juuYyWw+M#xmnL zjZM|XG2<^Y{QM@OY(H#UEa~gw`I6Ader&J9QH1b2KbfOe)iW`w>e~f#uDCTa{Qda@ z5BHup3wg=CF&Zj@FD?cOSKPXl$&ocy^msx4etd@Y!WHB3eR?+*vr_5~Ohe8e<*Xe& z)6!wmV!l7BQ0kIhR_q8oKV2x_b5`4AB3M9b$+UI-BXzcA!`toL{?3G>oa1^DpW;rq zN6&9R31iKYxqCGs^aD6a%ai|Q0aQC2#0B9L)oJ?$cd)UT1ilS_BV+39 z4D9LC#f@gQL7)oDlb;?9s;u^Yg;c(jW3^|Y`0KUWN+uLq=L{F6z?v|Fb33L#XcA06 z#BXXda;_-GzVWdZ5ew-%!Xj|vq5>)Nje#39aHg&t7vNLdkN*(JOZ8Y6o9Ep3STiq$ z`u^xW=>g3|)I0WgRC5oV3HGhlOF8f>1dgI`eTDRy?$%5PJmv0&d0+PFEd3lPOSj|3JERO2fosg$}AwG?e$#V7@*&_;lj zR$e{`>hqXCmJWj@Y#ruXB>Bh4jK?Cl0JPE3V))?YvArnq!=-Cj61Hj}A^`v(fXQU$c>^8V}~mrnv*TH%~f17>6T9ix?7Y@XU9Sa>eE&pu2Nr zb;1M8jOXo!q?0RX1WDX~ljq-Jr+lg1_8V=w54BKbSqi7pTDE}G89ffug$@%hM#{{; zEHXcK?}>L^i+fvus~!Ox7D=|$ixAMK)+MDGY#gh5dg`epe~|i$J`x1sgw=xfe_hG^ zr&XebYpM>Z;cgDgjQfOL<7Nk1=8b*dMiCiWHZj3nZrqpAlu9jEE9PEQ8C}Bh+_dK7)yv{gCoDVpKqgB1&_2D@bGoDh6eWc-XaPI zvPk#X-nBd9nas#v9>hH&mr;?h9;RW%Z5N&SG&tKfON5%)5oKUG$RYH5{ zp$LYap<`E@oJUhx2+e)X8KNhGOFnXLTBmkJ0wjC&D#`bDbQrU5?lhIJuP^H`J5^~M z{#I;=;r{jRXVV#x>XfJHCOuZtFRXHN@pjn(m&5k@D`ux07J1GlK9Kniam3j4AXBT| zwhM1>tb3&)(`yj|KdNq3L7K%&*l$*w9o}YT*!~1sQXAZ$z2R~MsSvYnG*2OTm?u{;5$k8;@5HI~G2KFC(O;U( z>3PJUAp8EYq}XjSn?9vi;8`+Q%&GSpAix-UMMMofqSniA!riI0x_I2Ipjx&9Mzo}R z+QBBe9L@!LSt34Zhu|O5FbL3q&36!-F##gW(P~q2r3l#6=nar4_+`-$*SdRuB2Dd78BRL>rdg*1kZ)J?3wk< zq)tMg#xE+j(QmQ8Z=x+fD}H&xJ6@yr(>`I^XG^E5y{@6h7R~1SVY1q@p!7sJ4>~uF zUV+1*SE+2yd<8(*cRWht8>kDNsGb7l@6-JFUjU}1Y|`*-De0v=2GC*VPX1g5Xu$ss z1#PlPW+{}1=|VmZy-l5=?bV~Eu;m$e09?$i|C<(ng=O07DRXcj7kBv$XpLVXdLy3d zP6vi(OU`>6^9Pw| z{^~Ub|1K1o-=!&WMfx?DIp^^IcQD)JpIZZtjvmDnKu3heJvTe`EYE%P*QSXlRF$?f zs)ujb+lR7Le{NpF#%rg)5iGZ&u709V*YZJ|JelzuLvPhzV(4>6FrWwN57kZPq*;*I zGA30Og&tgex}W*|-btVjI~DE!Idho&t18yo>}4IlMpf;#jn$`bU{0bs*niWs>RNUK z$H+O?W4rE;RALiS_wGY{f`^ZdTY*ivCa^R2lg;C!+UK%9j^QWHvIpc@#`M6+66P~8Bh7+nMoKujiG z6-CNOGcp!~7?(rFww$F!u+I~qG?Ex22qxZ3B(N`R1Y6KcLtq4JS2em%={ysN7!)+T z0-2^u8Nek>&}(<-u3*F7VNCFLwNic~eD`hX{!TDHf4g_{GUs;)!- zct6AewdOWWs+9BwRHFgdJuamf7|vS3hVG)~-o%eJq8dQ1=)`T69dXn%q&r(+Se1sf z)38LRS%uk%s5sMh6HO&Hnau6gaH4~`E$oqn=N=(tAO!#A^ z&H`^Y-|hhQSR>z)v^TnirC7wbW4mJ=1&UZ(`wwx~WBio2FhlXBPo(MG21dF57bxk8 z)@{kVbFj@9$(Oy(RbS}{Ir4cUkts1dfYsGoAa^5VU=51VYe@8J1$xaNNbo)kTYXbJ zohSS*?281*BQT>%q%F!pr{2L9KkF`IAckF{*~61i?InV6dlLRWH=HMoT(KvFa?PaQ zX)fcRj%I_ee!kf6*kg2sh@OAQLJl5q(w&#OsbUQ-58qI$Gg4P3bt~YMn>ZJ&ceK)-tKc@+kfeII=tyQ2MQXK*{kUMI3+0mYOA)0;mi{_n9pX0>GKeuc1VzJpG4gMAjjqqr5aEy|aEm&V8O zu7$_Q>(hB6lhx&7u2j|LD*HxsWU%%MgqC1<9%mRGd(>k*vR`9Ypi4*)!SJ+V0N;ew zw;i7sZqcb0@pyS$tg}E7W8mZt$6*7ByRGSQIAtm8SzaIpubS}Woi{?KHr-R;=z*MJ zdPUZJ2kBe;tI^h`uO< z!Q*rA>~7U{$B)}Ti#c3fC4gwAFDs-Kr2J2)!@nfa=O`Eg^YwUr-Do2J?fGC+)8cZ4 zGorJT30Wr;L}IHikJ%iyl^hC;55w_lY!m$%+i)Hxz)!ZnRo=Ke2CO)3Aa);^vFvuVTxD*A2i+ z7+#pI(WH8mt?F5?NG{>x#u}FSd0K;^Pk4em?L?z#i2p%KIDidRswbh|k}#W0`yE*gW2?#^p_t$NOAae;3~h-%fwRFpT5F-cO#Bxaz!P5Ud&B{=|xhgSsPG%zxpdUN%lLIdK14wLYnQr zLN4)u0-B$87GP{D%JCm`prfw!TS)mMRv)G&6 z-QHo>D4lp8F~}*#@M!L=1B#*BTY9Zvkr5ym`ITmktFkR1W;L9NWb zGf;MfvP?6PpqAKf-DZb#p&w6|f1lnOeQX0OF#<9t{yih@P$sC{s_yH<`h%0iZ#tC`T7j7qLqKPSq(Oe5G_b;|(XptuyRZewUui&)0js(w z7y0&%_A?MZNo=k5v1TP6uq-Wh;jX1>kUC5DlGN!@$ZhULEK-#!wBBbyfK3?%p?(J^ zsi#mI`OjbAX4_XcbG`6E#gQ^It~6&P+k{bTn3PO~9UkA_Ubj7~XIthG)P3cBW(?{# z1!93CnM9nB&5VdF%GL25{GbSNYdgO?^73ai6UFg#t7O7`r~9h3xOZ%Tnide4Xk~+_ol_n8JJ=ryRW@7QM-aJ;O?HPP~4(F&YzErthsVbM!_w* zwNUz<9MavaCB8UNBQdE4vFR$ymO9bPAlH*qR)uS(foBv$%Ag*wf8NcJIVxUgd8it4 zHFBQ4GT}AZ_kq0hUHygVknEBq{s+<#TwTJ*LoVmX-CKWE<|uRb1i4mSQZqqOmvUbk zmtPcyDwU`*V@S`{NpW+he9z{7ly7^9bx$bBQkU%3V|&q$d8z}sWbGwg#Ha8;w3%vn zbRziD)eF+GljAW@Ztl1Q6854GwNg=_4gGPxv@ni#LrwYVat8UaUm1MJ-i5Co8*lZH z!C_MiX_}_3wwtLqDNFWG!~iP=XUic;o34hky|K+62M&g!Hb-M)%Y*q71Bg-U2%+9% zr%|#h>?Dtcz|!(kPOCP>#Cl?S(Se#R&+g=8)l@2DzhDPTSP`ic59zyF`d=31<%M;N zwT=r}m1UD$hLU@W#l1JK?bHC2qx4&|NjcXJ`n!ifXwH6brn0tILucAjCx=kb$>9m; zs>(>UZ;pjVd8brVK+i~VM=!eJ&C7Z*wg46WfkJJUP z&kldME2$d8NJ6#|b$w(v!&Ht4}>zx&{hvU9nq|g+}IBie+!;{qz{^^pR1X@GEUd zq|JwcfKS^S;5J03`m+reva~x`aPF2L+AjKX_eW&<+7=xkpqW3}gc}7#huILN>YuGs zH*CMvdHuS>iS>-8ZDsE1r>EXFYQLBd+!HC}rzP@f@;H0#`$KPqy%q6Gs;1vd3{`Uw zS2`L3l4B93&JM6|tsl&0YEaYiF>KLwbR{G}2KZH6EE{yv;72X=m)810qE}a0JA<5Mk{aL5LldQe z`N<4W|J^e)?pFgYBuHg0QHNUk(p{^73+V^xF9jYXY}wCTGMU&{{M5WL5(u4wQ!FQf zHQNcm6+8f3L1FL1(yQ-EfO0GC&|=hQ7C<>ry+WHtWK=Id@9c`aEOh^%=R?6kTQ+Yt zY_A!M$NAK5%7jsFtJ7kzz31NR7a4O|AOtvcp8*%cFbvp1&=tpX$)j8i_^c~xC;&BA z9sdI!l^x-0Zd$UgwcmnueA&FFMMLNNptMP`MJ3o&C2X<(470G;UWz7kcie}r&`S>R zVO`2WiFA0mVuRlvY#5G}u!AS-{eB`Jdy%GX2qJrOCv>s~HW7T*X{B74jlV*}leyCE zrk{FR+d*N_c`+GODgyclIS8wux&Uq{8*At6veJ8u zkETUb4Bv9Q#q} z_~YR@9EZPS$?hwUq#2&vt`_0`(t-<8fXPUBYFIHoTJoGfgS__Dk(CvWFX;J!@_$n- zwvzSU*0LZkK#u%sMiql`6$xsvr38cCE+9JaD`hmT%i4>BhtD~oa6IT*v{;aEfZ!PO z_Cpz&gznUXrJw60qE1$PE)DcZ!CATRTa{#W&t<`81pK|0uvV1R&rKB?8LVvawI1lEf|@wL zL#DVclNTY}=dLUo_)ugb<4regDn8ufzg|fxHvYC56jO>}vh&+~T?*T!iZaL0& z`~>@AT*7Z!D0x($`P@h8G)APt7A2L1E9RNHDuT{HH46-P3w}g8*>i|4c+Lcvcikzk zF!?46LK|CtkET)lq0WI|;}Kb%Wf?xhyzAeb{7y>B=YjRPX%K<;Ah^-COeBdK`jJru zBg?yO@aFVgT~0UQO1c7B8tg8N0ng4bpYsx`1mR2`1TQA@EvriMViqiRAZlL1ox*2UV z0!*W~hZ<6hqQ3^cuyPpiKa0En*);>+&*=H>O4gt&23Y;$Lh>@HT8Y*X6=9*boP>*t z59I~inZjhAOuyUnYX&;r6!v>qGUB`sqx^a%wTn~%_InohXB0qZfPGc~ z3w1OK4^$E>5yZC0vv%BUnqkZq5x3Ji`aqJ3&Q!T`11s1C)nOyalW&N?tbaR0lj& zhtDIGtO*^Tu!qxUHgw5#`sYIE69SH-;7GiU;$e?d4H0g(ih*~#k!fXO6cG_}>TgZ$ zenw;iUa~!1>tZQbJ%MgeGaA84X>@2cCzTdGuFJiBvzoGpvCZ#o3)PY!-zX*REYV@v zEYHcaG*9=*4Rj}Pik;zE*T*OQ%MVydIl*!g18QGFwD3oUNa-r7U(_KW5%y0gy!JQ2 z%j6f8C!7Res)|i_d%>Gwp`LJ2YjLh% z?#3DY?Z3n$v@HWI5UvsOi!aV>QTFjxdTa=~EpMdqHWQk$&@5IfbNBLd^ZhrSB^g}? z0^M4Jr7qjPcQ|!Y3#NKzq~Lrs_IX`WPuU>8E^s(R)@Q@;IFoT8j~GsF+j$0+SuV)P z9~Ie6vIj~WyyG52`A{z_6;R1?qifp^OL8?8(KZ}X#xa>E;cqT-~16?iIJ! zzdurP7#3xkESW}V9*cREclPBQA2}xfEs!_;lQwSu5TRSDQxq?zCEQ%_taui)4TNDY zd2Xn<5a+FY;br3lRdZL5Pm)19fs(mfjW#H13rzya2xTclWS#E9-Y$)~v0r*l`=9nB z=1)dm0;P;5Xk0NFNzmd`zb_w5pJIYwQj+&HOWH|Z%MNMUij)Zr5%QK`91&yuKxH2q z_Bz@kce9}I;9&i(d;4}zdC*T^Wu`a|U&LHcWWSRG+uR_za1|3S7b9_E{6mMP`ufMW ziN-DeP<8s9w;B3+_>OlAw&X}}H;k3RA2HSr8Yv5)AEu7h%L~ACNjQwP#kxYR-nv8O zB_Eou(C-?w%KSzr^-*jK$3n9azN?nk=imX_`yGRdkC>o0S1|92lTPOc&7Hy)-1FCz(-zL!RxPx}M!-NF^d{ev(;^jSW#Us7 zE4m{{Q{I`wLjCx*^V=XUf6PweBL{UyjIP!{d9!Wk;WrX^yiNLYlfK)yf%yX(&8B)w z(krchfO!~QS(1{B>s}w6aKgd>RrFRxYDSJ>YVP(pmVK*Qul@GJEH?$C1TnVb1QpLa zH#*%MU`FH0P&vHWgR38%w`DLb8Yc8=6v@VC_B9R`|I&BzWkK`c%dn=ci0`x+lYps` j?)wvxLZ|EC`?OT_gF6c^Ic*mu6IJ-1tkPVlukHU25xWhgGzTdh;(-gn-CDB5l|42Zjf#e>6VsG>F(To3(u`L z_kEsozVp1#`~3CdZ@bo9d#=6LTx-rTu5pcP?3?d5^PmS(Vv=GY7#I);2KWcMnF2in z-9tn~LPWTSgoJee{yk(=Y&2996jTDNhZxu-gk+>7gv7+;)GV~*luT5_#B@CLOpn<( zIXTH_p9u1@3$So-vfq9P%>DcKQBhFw(a`YODTpc9|J#2zUqG1mV0vLq;b161u$VA# zm@qf3ATr=O5nz7)f&Tb|frW!dKt#HC9~lMcQ1Ji+3j+rS3lE2Y01prJ_5$7q!DAv| zQLu?1K2$V7qO`|m_l-)uNAUJk67kuBd@i2MEAy;=lKFuCYDog8swPRGl&Kw9QY#>`sPi zbK#K1+zS#q_j#^-AKif5AxSE9&0RDyn|-3?7Qh7#4BvoOzpfr?ffr2IZ$O6&j^~B$ z0XQ?q41(dT#cDeqkSnHZF!U7*gex72>z#4~62b!~jK5bDAIUkv_`ltukmO+pZo*1N zrNX26qj*$t?3lpOVy4e1@Oj>wYd*A zx{pfJUgWbIkbu!uo8yt8hZ)ij{W=J!)w$G;=Fac697EqF(S6SuOEf3Z`{S;5_*einq56>p1ESbia0zxO3orBuQf z-?!bapJR@6)bnfHx0x%!!NMX+%@^^q0n}mb&*(#8@8xqR;|^{W?#%m5e4S}q?i~;a zUlDkK&Cm7%Q4T~!uvq_rQ8NA8+uEA$E78%y%#cmdx@ZY&rM4YoA?+ysh5G7dIDAwd zZZB+C&Nyr1qS-UIH{!4cA=So?MDfjaR4dMud-od16)~;UpM_ewJIyGVG%Uw9}jGZi6CT`}_3j|u{ zy+juwSoZ&s(yLZka|0q@f}Z6CwZ6&SEBlZZ!dO24)4Kh#a(9e}<(PPI)AD`R%bu%^ z^((WG6gXZjy0iw4-oKo1cfsmZID;{6K+6^p&>HYTLH-R0SugNAWU*u9xRx%ho@xy? zgznS7edTU69iS%eL0=ZB3aUdzV!X+El_Oe%Q@-}U452G3drkNAR@~l>5D>#`;thxb z-@9ojPql#u`tg^a%@y8so|o_AfzoTTo3N(*6-`hA1C_aC)oRs|K!UJMp(PpEO& zd@@nm!cj2_HGc0nRcJAexwtaS!gc1fZK$0nTbj`JadH2J_=|Z%v}WEXR|p{5fFt_y zJ_8e5v_31M-oC!}0vj3aM#Bm@JB=l3#|>HAvxFTKX7WCJ@^CPl*y8<_Qcjw0V)-_e zdK}dwrZ9e5jae~B=W-6|F}^WpCRG&!=SXCyZ7M?4qVfSB3m$ejBFm!KgteaVCP=GE1?{JTk zt+3mJ^RVk=(>%*uwy-O}_yrI9wi-bP0a3}B3ND6S!cpAd?0~B+HtjYVoep&kkuU=i z@3h2rN1<-jP+{R~J;is}E|D=5k0t`|+wMB=Zp#cinQ;~|uhh^FbWvvpN_GjH4tON8 z%@A@biYiiQlG5COBB`5pSq}?6?o3o&a2TfYyeWz(wtAMkulmt@%7hiQ5AeaLSh7E& zg;oVM1Ro2$&G3Dz+0nKA}!aoqZM}7=Xs3eNz?mfGh2I-gxm_C>y{n#$6 z=P6p66;D5Jl>UVr3ohnZl;WsfiwnZDosMs|AYa%D3Er&Iy58RwRFwA6JyU za09vu1029OEmSoYvLO9!3gb%SBDo3}$7?T;ns~4+3)4|7cd5Iwyd%QrcOaRUk~kMN^$;3`LTuyPk__)Vmip)CX+sd`hC9v_%UW3##w}f3%OO zeSF8*g;JFTW9Tb^1{kyV2){kNPM$~cl|RiD_|j2leiKg!Dv9WbO?-P+Jxk2^zrvsZUsJ} zZr@^GMFN?^ll6ZJD|)$)t@dqB-XW1-XH2;6CDM)|=*8-%qdr}KiOMQCLmB=4hn!)I zxGr#T_B+6Z+sl>?e)?SPo|>@A&M&bs6kX7fOVuFzekY8OE-fcA-^2FmwZX*=NGkWY z>(4A>0Aulm(52O&&ljX%&B}(Q)M2K>DpD4^3}F`DxB2$&XL&$yxchb%_A|FL;)e%p z#?0-aB=9gxqS}GG@58q|TIU7ZKOxsq!_L{o(S2AZ5ZdI#@vwU-)G0$2c7m&Yw5@HS z#`wERzm)ZIz|=4(+WwE5&v3mUDb?hr1SAOuoAjRK9&a*it2qYyH!Vwjy8&%}#kdBk&AdDVL!UQb zK6RF=VI|(IjUI0N=tj}s6gV+))QbE zr(Lag3J7lHmvq!v&&+L!!b78c#Fs+y|A6|8Qf1D6F`#|G1jz;|iemqM0ZaA|hqd4K?Mpv;Q_Dbc9f5P4KT`Z! zptsv!GbdK5aJ}gOGidXgUfM*(o7x%Z>Oj(bx9ja=QYxOs_K6Om%cW!N_T~v)PY}O$ z`3V9yMt?wv#GLx5pmmafa}3#NhLF+K8(n!*thsGxF#Fqq4sNy;6&P<_PbK zX%*|r;uyM^$I{1cQj8nU9Rh?ir3Fo^K~Momd&E{g>p5UsBgpoPA}qr74XECy)k6L} zn10F719gKebzGU9#{%pQWh5FxwwXtXcwT$!l^G%=t?u&Je#S2Ku1CmogS zDrz@3b&+y)O?J5e`5Li(Ef4jMxZZ18&?{yam@}ZNNvpQ~^Pu<%H2Vew7Kfhc@1=i0 z`W!VqsKayj*d;{y29yGLmRGE2ioL*P?U)r2{pH|wQ!Wr^4gw)7-WA=7$QHFL*T@z4 z1{9SCzF58ib=SMz?XIr7H>ERTwEtRaoH~L=eqLWb(^PXdQS4nVZ=d9mP()fy#J2X_ z`3__^3W^P}m<2xC5%}_my<)Nnj?t5SJ`cwC(9s6qrmaHH$ri*<3U{wg!Ka-7AQX(j zuzi^QFFzoC;_-j;bR1jRrrP9ls%YI#kK`;7SY*i;;)B8E^yr>F2tXWFqfwU)mmRqZB)q=Yx7T>Meu{Sky2rF_O=X^q zi#tSZSdhpqG?zOS(VpqAAS>P$q!&23?{taAAhGPutV=MHr2jG>q>&TxdE<+E&&b)t zQRJCoQ=Y}}dVfA{1msnwD~I;Iy~YmDBKgmK3h`|xS#y$c?cIk8&=Ild8S;DC$T0j> zrAvjPi_+H&8{Pcc>s)Ve&02G;URJAF`EkeKZ$f)K?KK~3uiKxg3rDo^S%1}e26rzG z#a#X!FFqw^z{@(f<9-HYeoM*anu>69>&!|Qg2%lorPnUeU`n$BTSVpqH)n*s8nVXt ztxYG%cOv!Y28aDFOT|Lv3P^w>QGijDbwp^K6Ea(z8TJeqSXCMW2U{PdY`qN=R~=?9 z9%{c&qnd!@0Md!93nG}LgA-xhTDA*zh2tOwQNzcaj|`kz_Bsty^HnsooN2egd98(y#eVMK~+1(4fY10;&e@$p()*I*#K1)RTSf&V#Rr zuiB6$bv(lvj!i9B=G|PCxI=p!Pc#M6a&!VG?E^qD16G8fm4YMB%R-lN_3FL$7!(^Z z_M&vEQqW@}(>TnJtRQH`BzUG+aPNo@UfH|{J!$bht&*?WvhJI~rpc~}P$n4bg(KiTRH8PAZR3uO}EE=ic!!x~LBYW0H&IiQo?-A@E_>ko!J;=CO z(-E8QkH0jv6rt2!61jUEE8PGw5YvwwjsK|T1BD-LoI0cHoPaemGo}DK6CJuww_TDKla9I}4!tVoQhsh~$6ZUE9Z; ztEil5Fb`M(yvXZs^raG!?L!h9Z_=51?LfygNt!}bVqVk_LiaOU*`@M0-j9hgcX-1r z26E+dTJOB5do61%;P~7Pg9s66ABUJQAa-N`t>H&~Y4)tdGHTB;Y-xF24q|$a4h7}p z&JQfad-has+824(+c}0q0=rKJRQItYv7^M|UP>&4+8vZEB$B){jL>V0Q5eXu_<&HRMs7Z}P$rtXT`W?g?Bn=GYkY3v?Qx*Bo_bO>J=VsB>s*&uaGx z5X)ja1e)D|WMo6u1OM*W+)sbUlccFg+5>D zKI~ao=WZXOl!uQ_D_R_6%IRmD`+*dG1e0&>X$b)(_pcDWktCNy3DUV1<(^Y4sjp6+ zojjN|FYJ6MPMn&+?s)1W;3Vk92q)H9LD-+~PK0Uf>4>Vf16-sr=2ffxun)&D|lG$psZ&TwS zsJa=7Kr@5!)8jf5XrYXOc{+P*1ob~slm$p zquT%4O0k<@=?3Hm7j-rc&X}8o*jiNWv*F^&TH2z%C9;mTdd$1@?bCo9-@a_s2v( z)Xy!h^`~S(Zd|vz6Vg=xLas&Xs6>i~oMLul{*#Uz& z+3_kK0Kc^wU*5v+%;0aEAS%0&-TSd(`e?liVY9W>fkij$ed)_6Dv(YUQup+ivN{(Sjy4NM~^d z@bokE0s2wuJRyZhC@$s8Ut%YQ8Wt5F5+e48&y>a?K;qR#TC<|E^m<>UmS*s9ThKm8dx-)?c5uMBQL z(a!Dw%1i_NZkuat%avAD^5(+zwAx-5XQ0L#iIEa8?-rvJxHTddM+ouNpr}5yWwdDx z&+`W4vkL?a1{-rnC_r@Z#T5t$PFNy%Q~zxb7N7s|`?^6b;B#W&dW+a04699_k8Ljy zfT(4j{{|G`1cfcH8M;y=8B4ciPO1(b>9GssgbzoQLy(#LVW^$lC#nFQ9s;k?8{dF@ z??X8uHs)kU*g!=1@maI%q&?x-I*`Jd!k1M!2W7FXidm?z+0y$tZ*FRpz11KE1 z8nH1l@iSA)Tx|wvh6!N`v{UJ7Rh>)Y0NJcri#Q=#T|DIm=^x3RGop`Qi*PQF)D=g+> zGjPn@@OYJ_c#p@PPXBSyd`3R85uBeJ@BFw-Z%*b1zZ?SUk03f2^Wsx;%~4^dFR;P{ zf|42krFB!`ia19Ck3jgEe(%|zMRA;I_5sAtm|Xz2qA6H+RF#|8)Ov;pluLZRmG^nK=e%hsdInYWd)FZY{{Q z6J(h&{r;8N$L|C4ori+u*16n+l?^S;f_{RJxCVkwCM`Q(9u}mbP=>w(e4b(BJ;X%I zbW3r?6!nJM5^G7OgrUKgVXab<)0zhk4$W>Ppu*dvQ)dm}`wAcz_X-)#C$3~Jq#&?t zyXvh*VoVuB=Wp%{9K{6lNBMVO5_?(FklmYES4*HWca5w#`GEa7uIAZ~-slH926E~v z3gSA3IOB>+nz0Mp6i3#45?LQiYpau`6N7PF@+G!oe7~cEPO#Mqu|ewei|5RRHszP) zBRf{kqF9frtIBy~8AvjlLt!NrMQkCBp%*fRH83g8Y&61?+C^^4|6p7@0NzMd;bo;lq@h_a@>%1`B^$cb4xYNtaD zg`8HbeoIUVth$szH*|eY9C53#xAzUzmB}Z;REV@Re<>M5e`m;GUX09uvbx|?Sjz~phf7C+2C{=NGL zLqGGAp;r`D8L^F*ufRIDW-KePZ_h?Z3I`3?l%rOODh==UttC;vZn3wejc4+#dNKc) z{1pCy1-|nW;Cfz`)s=E@v-7t;Pd&gRKqA5D6%=n^4EO+KYm^_TlkIarpHDzG%OU-B z$V*TfNM|DJwM)eLXv?DTiGUnnUY8!Hh16x559G}imUfn~bVN%MQBzXJnJg871Qcc} z5OSqL21hw7K1*xPSL1QV+kc>lH)(0Xd7=LCfq3q=a@WGS!Cv)PMv}^hXs)WMLo2&W zo#Zd?zZ>o!<*qy{V*}Q3DuOHc9%MPg{06iGu*qqY)fRtW$EMn64o560?Gc$dX(9U& z&*{=~EowtO;-VDJ{;ylbRd#tK>OpVd*!!oE@tCx2=tUEb9nW$Z-rbzAC;)idn z!Hp=g>_qG@Q#QI%JzmUNvn7LgFWM-wo2XaJM8S^FtL6L~V&N~_^K+}ptgR$VD;K@J zWd682n9z`At65ddOT$+f&#YjEQN{DzX1j>FvCq|PqqMEbt%c6m*b3DT&!W0Btne#26pTOdO!!^r9I z;hB4gW<2m6>z{5@K0e4m8aNt|I%L2Dv7n@g4o_=pA|uAd8el@y9d*Q8?GyQd)?BB$ z7~a|7lM+aC)GCqLl+t#(wWqoO5$WJ3NM~IZU~PZe8p45ZkYzo~4@Pe(k7aI>V<+U8J z&mOyuEV;~toJ=d*GI@vcy52_s_P<8(v~;ehlEBO`7V#PC)QuPoriOAc0|*l*zQ}xd zZfZ&}U5v$=MQM?Oo&rDAFT+QoliC!CG&WZh^wONqJJXP}OvV~zc1E!6yPDx=E98(M zIIz8g1j&XVWsgSH__1_dCzIObepGx{wW~8jsxWyw67PVX;HG!TZRX@La%usj^9eNl zeDvbi-ta%v1vbEXE_=-oSZK_=UGNf7G?^gU>~z(f!O=gw)ySL6I|Q+uy})XBdo^KE zzRbPKmDFgXpp_CHIaLl(zPmwq!-AP*KsiD&VOKF^@W(qGtTukO=L6j#nTSZa1GIbX zqu2#_fIEKgSA7EtPC20I0l;%Q)cN~eD^o;LF33M8JK59UdbuM(z++msH`DrV zs@v9UBc?B0Zayhblqf2q@g0+R?)$X8s9~9x`~e6KpLVjwg25eArx$>JrBCZpUV{&m zI05AjkYOLR*5Ue_-6<=)dg6g6za&01qA2Rtk%p60v|gNVjx{Mt*}=UWO=A#ALd>^1 z#lWG(sep(Uo|69z;+)gV6wo(a!W5Me-bGAz?_AsLQEBS|XeAnj*+9U+KBo)eO-USl z6@m?MCoIQj>qb3x*Ki*D@T{t}44Au&LBWixY>etq@+<-FlQD^v;E%8BmYkjY`LR9` z4yE}#2ACk#s>Yh8_nIhgyOs{qH3{N)6(7S^e$y-)AV>Ggt&G!Z^=7Fbq48_glt;CL zjYl_vL4D0Vj8DMdF1EuqGIova1Kfib!2I|baFRRrR#XZ5g^QX&)&83<$l?-$2qLloEB~J77=5pik?hmb}SBCp96rjj9$14)IwN z0Fh<@h~$#Mop55~W?H%WwdK8!rcdPyX@zz@q8kv!ChCcemrgX?Z2; z$p%lOOE{-gQ&m}AFJCS-2)IZ*cOrKtS30=OK3A@0w-%MXZD<*;e}0E^xd@}c$$l@% z{qNjc4jivbA19#1LWmbZXA}T&`St~_Cixg_$&22BRfyP2?24I$EkoeX$cU? zb|SYx8vtVIo2XtM9}gi((6wha)Q@3Axs5L_P$f#FneK~^7jH2-M{P3|pOo$Gu_>w%#ZglFGtw;Rqoj%V3nDwC#byYI$f1-WATRri zrL*Ya)+mbyLSeWcu(6&OyE^UrX|#LW@Ld&#nuY>5G&}Sa8~9DAw}$T+qvRl#xA2D$ zY%9a-ID!p01%zTWSyy*^h`Vd<`l&2!+*`zyCvm(_{rlxzy0-X`*DOEmWyL!hI_Bxh zW@?)&>egGuMe;Qw7ZC(2Akh+I$Q;wuT7c+*_~~De5qcNjXSB+tnkC7XM?>i3CjHgb z9yfgko^Ep2IU)Gh&P-bIq{P7YFJGa}@la3NWwSrFmM$TJ8UFw0$1NYWc_Y;BqX(Wm zn0@xJJY*_KlGux}IxQiedM9A(g`4^notHY)pBr+zodXaJ7{MxkeY@_LjpzH?JWG(*Z3HdW->6@yaLH|2llqb$;Ru`;?&MGfCNcywC$eLTm)S^1>Jzem{lzM9TZ~*LsZgndTWVKuUNxap+4g&r~5#_ z8L*!T42tj&3T1Xv#Th@=(hU`?zf9h|0ii4bxtvd2tJhb6-uRU{lRICX;_%^yii@8 zzL3B4U_D>F0liQEwNt>r7(db4WX<9F>&TtA=uYf0{)g}6G9SC`&`P|PK_NCuSk3z_ zWsJ!tLgzA?6bHbg`je5nHtBElZw;6@baDd7=L1MpQ>QpA(7F&jXS>X31>VODq&<^a z^dL^eWfNm403ing%X@k(&1j^f+TnBFo;*&rgSi;ueU- zi4WIw&Z&Tq&^{*c3_Z-PM4oJ?vCZD?z@~llt%)#6*W86#pItJ=Tg?%!O+j{Rj+Uyt zZRWPB$=9zt2yuua$UIj>e6Q)?JX>lC9m6WhnLWN($olpB)vL^b?=u~L2|VokCec+9V9 zgFE}A&^Rp1JIXHuSE!{Bq|H@mj0ItDaUYsWJ1 z44`wV=YYgQQ6!LDXe*?LU4U)!*FJZT#pohtag`KI~lzS*t8EskIH7 zI}liJ(gYuAs|+l>s1lq6z&M=nU03@5r*^yB_it{s*R9uNa`@J*)_Sbezu>U71l%ga zSkOLZIMtcVlD!y%Zn<@_+-i*BV0rO5+yrDC9{RLMRhAVh!VNiY14rgNFZV2(R$v6& z26YNQgF2EWo>J)0P>gbEee$=0R#`*Tb~Ea8m&kpGJh6FkZ$UH{uCBAMN7{PppY=AX z2-RZf-VM0ISqZ(EDnWhE4};L})<13H!+fS-mTT>hAXu^d%*{M`hagT8KRh#&DA9Mr zUdv-N1MdPi)!dOM2L9wgoVo_?Wh{A>kbgq!p7naIwG_XXl$X48;B(xnn54E~o!X!; zvoIZfJ7RK-+O5Ix$@rs6Yf{#eLyEM~?3XUY+a&#b3P=w#`V4Yrnu#w8qV;t*9?UTJ zCRkPO%YDw7AQr7&tJ!dj>(JUgM%lFHn;NK%>4~t@*=8R2hP=vLo&-_mGaF?+>KNWO z3jaYkgo`Do{;ppmfS{ST-@~b>5W0O2`DC0;cPg_Lt(KNRt(k!AnbIJjUeK7Os;v%0 zA@Vd&K&-D1pUD#Dp^TJl7a^dqPw|+@tGe)N%(k%JQ4O(CS1M4A_3bpKiWJ$Qg9EpW z^I$>DCw1gb5!?5^#0c@QTZt2V0Yxzb5UlS)DGPG91x6Q#JNuiWcL%TA6B;@~1ndt~c zn_10EioTf8{VY<_uD39B=?5*b(bCZ0F1ur&`A1uFbd&MBHm1|Z+V_Rw+o_`zJ`WIp zKAJXIsd{b_d5QH)OXr*dYGHN}cw)bM%MXOf!{2{GZ9w)67;(1d_WZ!2)UR=sYutWa zPL46GzCKm8?1m@IOdmHxXzxyC7!|#314wMnPIy!z}O=9$N6RJ4am%#(8M(A zWVg#MU5b897)Da4FnqLs z+H$;#IUaWb9u4++BxM>SaZ(;4%)paP1)5xs1h2L?Nns5wR1z7hj~pYiXSs+=OBvQ~ zCEu_ALrYgMYSDr*gp7E^*O5kf39OkIaL# zMf)9Ru8H*Z({{~Srz#`1E*c(#cwTClj}IVxe%#d&t%_64UV_~SrgfU!v7uzuh@j>H zTbb1gy&#DSAGBn9TTYq&p%T*?+t>f&mAS@-^9BnZ%W)A(8>Oj#^iwm9{a``fi~A3WXee|(=7 z`yaFHZvpqJfEI_XHS`d|u(p}9uh5G-w@dazYw&C^46;xb1Bi>`fhwiO)enXZ!{B8c zfO^$YK1Z%lb(%Xz)B+;yM^IIZ`YxLRJuot0GNMFoWk!1Td3qZ1i;Po(D9YpEMWvWr z73G}<7&}!+Hh@e>yhvJSkgLIaQR7k7H_8&UH$Q5sEOXx+PrBr#w0=XFL(UG4o^;+P z(u}G0O6kP+s$cEx=t)j-DLM;R#0YfOl&!A5Z_emW9kpVB(bFMtsR~AMGlXI$dcFZj z?44biUi{!U_enWgEqE0Nq~x%TyTS7fNxxIad-J09&xk?{~NBE#G?!3}7DJ zI&n3O;Pw|)fF=Xb(##fsu|ivHq5FmOugtouzVsUcl4TvmknNh)@7qqh=ZKU4c<3hi zKql(|y%In;(pPzH?gul;87{!nSHi!W9aGld8O|*}H4|nJq06TRuH$E=(m%F&7Rr(c zF_J{qFT}xN_jSL^PUuvLI+9JsHyuvjfaa*K=%6FNC9_(dG{)PDPDko&*SSF2TyWs} z01W7arGJTTJ!wd1Y32Xr>EfjZ|IM=*Rz7$+3a&{SGu279BcLZrEvcr>gaHz;GOoW& zgugxH0?5Plv4kRj!OSg6GvZt$>d*yW5{C6US$2z8Oy`s#rx9nuuh?!t!WlsDS9}9% zQv}qa7=hq5?3B~v8_-$av0%MC|GPa8##!(J%wESeYz!3IbOXW>zakgH2LebX$lUl1 zXffwxA6>b5tr&P@uaR2Fj*#Uqf#QcNRkA&wGefAaxGqW_A0lxI$ z2YO!>t@(U3-$UuiT+jR*uI?gd6?HhEauYEZkY!r~%ID%UU|e^bN*CpqAG}OR!Y@MR zAq(WA;MMJ>-B>{G0F;pa*$^reS~!01!oR`j3yjNO1rTuq6RZ*74gS^yJ}qd>;QMY) z5fT&&O&RJv1w99};CJ^ohIhy`gA5!0#L9=ZXA)fU=a;sZKOkQ#u9yop@Puee0xhEl z%Ob@F2aqPS+}!D^9C8;aS7IZy02kXU_~PFn>9>m%?*Un++t{d~n~w=Tc$Uiv%4w=T zZz>qVCf}xBSEk+h+L~z}G*yNgfU&}_ZqQ;_==F3;a&LEPXU@2xq1u()Xdb<#<|K!+ z_(4{t*Q3xe9v8f~50VUOQQ|{C+8}6z*JFMRX9#-3+4FWTXh>5te}GTqVH$^V#=(TX zjl7b}(th4&{V?%1>ta#!S8=$6I#D&zqf4$)NjhoiAPrwKgV0I@yM_TVekwjP|%n1e?o zBS)JaMXYyE=49q}>73|_{Hr*JuI1;RbgJNbK}7Fe*Q4`wyR=)=ygCy>III0Kyt#R_c;HBjoDqYzKRC4T4#*kt50Vkn5Dm0pop3%KB& zIyO_2*^bcEhSRHC7FY@Egtz*Bbm_x}8T@5V{6a~6oaPe)hX_>rfC*i{t6To|}t)+$h;D378O6F_;CJrrT1Dpg< zicMGkM!;A6@ws{b^tlC1n2=7yRVc6yQjtgg(gKK4n(QzA$bb-8=HTnyZ>SO=1L|8$ zQUJt<0rj&)|7+WV;R@@>P1KIf1$0EH`wrE6f3LPTq~bJyg0D5E$_f)(jsLeZ|DBF??1;0RNI|@#YJKxLx}EKLRf_ zjY2W|zYY5E!8GHHZD8rPS=V;YG6q*+L~F_zlGY&4b-#u$S5fQZpGwzhjgk@Xe{=aY z#n3mUJ2Rh}KDvh(lM_fbx2H4h*nv(tFIKO#E3A&rP2si_^}n8K2dap1_{}y~#vFY@ z`)(4l_$ZD5OX9n+F8%}Y66{Wv?V`7DJ>O)pB^Cduw$A>xlv?E4!V?$O^QuJRt+i)P zrxk2OoG50lLqkcmgyXjnA%eysqBbG&is2orexu-MIudu-$M`;i7)Y~`BtVUj4D2uN z{a>;0mbL8xGk$rYafUt!d?i;BNGDD}xoWNM({VTgQVsiPuORqKWKw^rtV9 zZxJO&&C)$Rfo5(0voZKhR-3D}JyXdud+g}v_LNk93YWganT{T4#~~KA1!IR|OsuR_ z;iKu7(LF!VF4>wd8$65%BP1G!@D>qEEX^HF9(lT9w>`{ZrYvu3M^YnLZ7L+^12D{c z1jSLHn3{+dPz5!pI77}6yN=2F|0w2C`B}^*A)(6D3VK3)k_puNRNvk}A*Rnc7{06@oO{}{~ZC0uP zRy$L69DbV7WdzG`9sW|*V2VZHELjTp$0*}w%Q4f}4DLbQ zPDeirbG<16kp|kp719mpDNe5DUz&iEkd^oShy$$W4?hCv5Z*KRNzTK}Mwgf4qU6(3 z=yZ_3i3ycZ^n>!3NXd1TwCQFj#L<&Sb}#`IczGKlWC2gMvqK)mNl;5n?bK0#uF)(l z)W(?&46#dUYW)aERU!3JyXTI{o+-U5`6RtU#0}bS-Cs}Rkr*~~K@MFSGzOyAK zm29=828CPGXqVkHM3nIR&7lCnnrQBoA-+$uGmDbQnJkSh7Dn=nGfff&7D=sxw&9Oy z@z=7+f(A1a-o~s7Mjj1ubMI8@OS225Od2G6Q z7zQW|5pF?&O0Pr#_eqp;h5>!esC2i*0Q}5B0`M~j47WdX(DU0l+pTE|Nx$t__U+vR2NMXl!nO6YtXhT%`i1>dI3TIuBK4c4+92{S2UjK;YAVrl@~z z0dekz<$oEj{gO6}US#H_OTO3$(O9-a^!!8UU=k9)RYi9ARsV8+@I*V}e1Y|P&oXaq zv5Yb;^OCvaYA+so#_1Ke1a$+lH?Ig;{CY$Fi!;n>QP>$6^h>891NR}j{^Aoh;MLbn zr_E_=Or819!~kR9cLS=m$o_RazgEW8Nfgzi$c>FV??^b%V$V37I0JVnt2?|He3 zz6B7|CGSJ#UPe@C10{)WO}izc?{%?8N%Wy2rCxx94E(!Cu0`3`i&r7h?Y$E7k7)~4 zlxu0M9m4a0(alQF_QQCv;}D2uoDgOEfq`#m3*H*Nyf3jfD zh$`83AtTh!MgQ~~D9^pKA1>*F?sm$mr~;`hFTS`x0ht40LPGG}#*;_ipDax`0N76z zj>M zn9iJPe#}c1nlMm3nrcvLX+ufdP+z=eVM^Gb4`n2xqROjXaAUlcC?8lR?buZESun=$ zCyf@k8Gf>vSEA@hym~Mg(Fliji+zbPQ5% zI=XAil%k|fWo||p{8}=mx(pUj9y=fjGo}SyKZVrNANx;FSUKv^4RJMCN>!Ftq{xQ` zg=o&Idh3$tSU0apo#v0DEK*rpS!WKleNke8kLpxNeyrdh$+zcmsn&oUXzb=GZv9Y( z`(WBDZBa9$I$0s{nIsji6)P`+05YT6KUgG2l(q{it&~7pgX)nT&Q0e)u>>#()51JYgu|U( zgcGZr;~C9;aeUrR>IZ=C_bCH2t%kMS#>K2OLE)ovXE9ApNs4$oFg;c3@h3E?Uyl9f zHczW4lQh=&Bq;2MNfc!`-F~gO3347shw>~SX;tBDP;{oQR`Zkgd~5#d!$PE=;%9t^ z`|uCG`0py3C$v})23&uPwMpDp}|S^m*$5;Stla}->gLPfTrdxgjYV~}Oi zxn1O+?wDy62n{y?%)n4_z2)JRt|L)!W-R*rvcLCf4#AEjtPj(m$`%)lhiO=<8?EnG z>4%Ya67UbiZ*|4T>U^)&;&GL=KeXU$PH1q?O7PMWJ4!8;aRM;p<)3~fZ~U}m{X|Ju ziXfKlAgAl)?Lv(N6N1V_l|GLMeLrlW#;~=AftC2^?;{4}Y92sRbK+H#z4>ds1ex%dc4AAsoWl-l>~ z8$vfDzx1eB+Uesi>qix%`L7cQ?(ue_W;~D}_rD-C4*xu!(07sua5e)L!Q|!y5>Xn} z@Yb|scE>9|T)_PTF=Din9nb@rn_kG;)2hF;isJXFHbOTc=z5|19ssmDTm|l}C%-1E z%~I8y98*D!d6UJQpX#CZ>gE=?Yx6W4DK*Giw9=+KI4f6%6e*vbPPlvwIPH)cD!?Gvpq1ULxO$u2_i6zC z4W)$4+oYgfAMag!uokF&@QN!(+O0~xb~|(Cb(P{=fZv><%qAg&aTTU-i8jc=3khf3 z0yo9F^1yr09;p}ojLg#8`Rl~@but^EaHHf3Q!ul8TC7j7E#lq(*UEK9HMMQ)ASfUp zMFf-*lp=_92t){o5=e*&A{`6Ti$Oqo2LUA%=_&%!dr)dZQ=|k$si7r6kSYS9Nk=$u zAMZI{@4NSnH{KZU#~yp{u}kKf*=w!2zVDkeUJzi8cZvEVX zuxtJZTvTMAT^d)4_0z>ShW4B5UK}Hh<$ZP#LK_n_^j%!z+S5GR905V2FFYmBMwQsZ z$@^xRGBs;-#Xos{P*1#RED+vvBR{8)T5aO5Q9L$Ix_YW{T1ZHQunrNt&0gg}aLUY% z1LE9kNB%~4DmJNV0apm%ilqJP)vd;|7sJ9WW72bjUE||3NLU>)S|(KG0M&W=u-lN+P0JVLIE^eEO(OI7Boo`e1;qYC6h0G9mdT6 zr46s#e7#GL+i$LtP&Wm_OuwLp3t2w{u(Urzx5?vKB_vdUCC8C$``K9g?KE*~mnN`P z^j`|zi`X}l_4tMq9?j>Eq+jyHy_QnE#?0sDrhd;cwt4g>+7i`uWeC>prnv~wl-;6n z9ryz94;tp~`frzksYmYDDxz?aOW*N~b)05r#WIo&cfI37uo}s%5l(Byv*k$+^TKiW zw;ccEww9wQA}s943T^uMQT3LjWHtA&!202kTxP}PPY}(U`%+NbrjI|=>VLhz%;Dbp z$tP=vT1A6h5{Fk@4%IvQ0Y<|g@ooMmqfx1Na0#u8w3j4txu=LvQ4sol9c47_mvyGs zu$RPWX>}_H1qY-yPh+V==A>!Uv!+oDS(-vBZw@`>R z_h}0u2L^P#-wK}{;NXAAZcEj9f;tKcgN6anJ_{&w_AjS(_gbG7%l|0`mTv=0+U`j| zj?2H3my*S|E^gms@B-s^ZMst>;D zPd7ai{(C-y-ms+eLwEiwL*2Sk3r-F~;oE5PEP@_ZTe6`?&nOUCLRm1{8}CXQ!WH?@ zxwLJ?QQ*aIR^+Q|(p%s2Wf&7&`fg|qYG*wzkLE%A|@Q`U(EV_+RK_#}OQPUsv{vLsz!XC_q?!c4OP zWKE`EFis_YZlOL?Pix8bV^Y)e@+gJXQB9i+1B~oodT7W_DkuR1xA9=km5Yzl=upr* z()U-bC4KgI6dWXP*T@U|ws8wz`j4Z{R_`YWgEh6V_=QUit_$3cn8KdjN;e2$!h`6W zqeGh7xS9B2TdzgNDX-*o*wIqV-R~!koxc|7%sOf8={HRG-0WRUnRl{6N6Ippz#yXT z+z|gFgqNfE#uW0y%&5wBw(KEyd<=*g1lVeZI_W@OflUT2+!( zw0eB=vKFGS^C_tbt-VoCTX;r`Gs67VSrxP3%`FhWzDmk#I5!jX&pQLORkN9VY;55S z;gO!?99q?k7D`lj_|I$88~1pTK_%F*HFJC_1^FeFWXs@bB_ww)qT@km?gq6@MSIN_ z>zR6V@G#wpCF3oGVEHl?G4V-?ki{=Texf~Yzdgr=?3`pliq6Ha3%*=w4-rPtU5GL@3L@0;Y|j5 zWvh;5c{AT@73$GHw0RM2>}by-RjHemA>^p=7;46**>=qoXI|X9iI|-><(b68We+&9r7c?RI?6>0er$8p!%+ZSDyh$3ygTUtC%lu~Eznk0{fMJ5|)FDXfI z&|^KhC_$v0edFg}sRDM%>9s4--~VBE8$cEM-rHS%z5OxgmuGPEIphvlH1q%D$)7E5 z`cF?{o%UMnhuD!Uczi8BL&eEEipnCC#+=4S^YJ`=p3TCU;in2!#H`eI-PGl+4Rb+^ z7Xv4u3!hNSBeBvZFr=|$A+i#Fz7CQn)u;M$=+^q{7a;Eg$BExsY}Ye&-5ruES+#f? zsM>WiaOB$3JF6v%k#y3AKGgZEz{bZP4i{6gjg4=%0;*1aU|xNfBd_VK`B?ecaV>4o z`VC#3pVk8r(Z9F52>7#Zd7Q`0^>;!no1T{yfoQC=X+l;Q8IND$DILPxCV0$!=DiS0+254?HBbnCzD*FlY^u zufOh-%2DQK8P1UCFV^b;m-<{smVXW%35+RI6lkJ~wcfN4J)LcC_SD<1jLoa&M6h9B z*#mueOuim1dmNO0&nTI5Al2(il0ovzxuDgO74>-$`h$IipyoG#2s<+gyuCD>fAbFb z4t8vAFGisKCE&JSSNZ^!`rY@#P`#ju$~)4LswlN*MPyR@7O(eQWUi8L7?#hJ4ip@$ zHJ!GuEl_Q&JD|AR;81j>-&8T$sHXo#OFeoFEOJ5yDO9v{^v_>V%h4|H(Ou2ox-X!J z;ICr;`1_3A%*M2-*jY9G9x}V}h7Ga95Wf0fpEQ6$Ex(Ao1mF-S% zZIu;!pR28oQ#X4iZR50<61sb&FdfHxN$GHjz19I%O z3~}c6oyP^FPdOR^BBjz^EVIS!@`6&qW1^Q~T8`J<7#=owWSUNBCXfzN7i=(a=9mk+ z796Z};dbo_WOuc3oK1#g%7}awA6O zIBbJtSydzD>s8~2g5_12>3nT-(A-=#9@KyV}OYS)ZF)GHvmR^s5hs!Z~RY$jp+DO0R!K zrt6EuYJ{vGy^Eg?a*SL5V6sH?_8EL)A)R+f+Za5SVuRR&|2ifspc&@ z`>@_($j(S8AjPGRW7tsCy|#+ZPkkWkp*q!%DiCmHOzeT8+uvPipvXR?HXfmnd1lH7 z1d#FyqF*{jKVc=Xel!|LuECs;!~vfDGEw3duehsekTTsZyX5vS7LZQ5%C zn7gHUl{R7?BQ%qqGn<(wDln~eYqs%r-do>MO3N?x9gHYi7%t@|9}u}9eqb#=HFlq1 z26^q}g;uRxsg&4C41{O>O$fNN43KgB-ebgQ? zSomnVZh2i`F#NLa+Pi1SuhiD76tQ7Gei^Y%4$r>r!{xam0Z!sN%gMpX0aeQm78z&| z#SME^Y82l!)h-l7jaZCmHnG$mO-dH=7WSlFL>7DN?%yDls#i-{^k*nWmgdBs%i^R7 zx6FGk=zAb*WV>b*e=0q_vqyN*?{#JSL)uS5(<}ZMCT6dW=|-hy@~G2Zz#vJ?3Ps9# znw{J!aUV&aHjM?rgC@NOmJ_LF)nh5L-OrgEX4`{1u8OhZJmu*9_lK|ECL~Xz%vY#d z^*DTGM6-k8*A#k_Em7!2E!s!hzC$U`rzF*lfp22GAa;@sVTs3i-_H7?e1b*-5TXA{ zNU!jSVXsz+lE6#nSJqeB$)e6kthO*SSZGjpjjB0mLe7U{XZxyI6w7Q#exAcOc1|0~ z(Kl^%v|T*pO>*QmHghibDPSV7y7D1;OO$E6q&8Jdf@#|M+_4K{G@SP@e^lnz^)>rv z@;h_76{2oX?(M=hmrH{!^c*7F!^Q8%0z!3~NG>+@)|ZspVys$Re0mmTf)xoa#wQCg zpA?RM{9Qn!ECK^+65y;6;3HNg*WzQUs4-~w7aT&FP0kwpjCAawj0LF4kJ9cWw1=eG z@`6^p2*#zH(0hrmM0AiAy4f9=Tn`6XbWMJIQ}t?*Mc1^x!916Nr=2g*=!Aw|F|?Z% zIxk^HV2bk7#4J$DW>0_qj0vyZ&^v1K_dxUixYyEKFbuw+#Q^(epUJA+|AW7jv}0$C zOSW{{nTb^Bp)m7cC;XwX3S`%9qCd1`Ue8{s*BtKSgLY8hu9?#dYxrCtUm_sKkJvs| zUnfQzMFi?yeDKS?ds|q|VxVkgIYFtjWiBVU_=b{|;K^sV^cMC+>-pbDE-CEu`W1Sl zueYlvhONKMHtDR0?-}n98cwsUFBYQ>AFPtRpR3hhDln?IuXdKoYHR zdbZN9R?@w2NlG$nK!6ngoKCKSjmJ(fG=Slac6P9Gp#m@U+*h4}sxDC(i-@VyOJ*DO zV#$Nsk>H}vb=GbCdA>z&?Hlij?xWQJ+Aa~fQh+C&T-M|N#0)0j<}**6``SXrcBfBC zf|u4xjfbUu(4IC&nYgqVK5ajd08oJ{_kEbzejB^frQ{?dMzrnLM~-e0htnj}7jdBD Y5Rl5cukrQY`0(Gk@&Ds-&>y7#0x5fFeETN%yW@lkb{|-_bQ*@ zWl?c)aW;N7$ z1fLp$YXR86c@jc?djY?{Ab9u$gha$=NJz=R1Io_=co2MiJOX?|LIMKtXn$}&KtN4+ z?vkhi(RrQw#O%&AVu4X_&v0BX`9yoWcZ>7N1D7BYQabtz42)ddJiL7T;u4ZqrLIXU z-cV9jQB_mdy`!gZaM#et;^8ApD{C8DS2uSLPcLtur_Y`Tzj*m7BswNGF8)nIVp2wC zR(4KqUVg!c(z5c3%Bt#`rskH`w)T$B&wc#^gG0mLe~e7e%+Ad(EG{jtptg5*_x2Ca zhexODf&lnGmj%B6T-e{%MGdYCkAMK5fcSJ>5Ij$C;8PP2UJ@lbr=UZ8-}yYdSl}6& z>rrn@K9O)-xxGdEz@?Xzj#GS^3w64*-&XeDEiCANwX%OK?9X+L0Oa@(PxHNbRty=xO(p(^WH(O4vgZv+|qZs(-~0CGA&AykcOSFv zuJA%W64x;i*GY!}4xcuaKW*O12w?jE=|SMsNO}MBkTPdAX{LLdt>Ev>DJk}mwS+^DqnQ91+rM4)S{ITT{p9QX`a&K14XJw^JVX`mMAr8N^s38UP*;JdkSQT! z)1D3FJyW`-nQOwVdeO#l30J6D*jgHd)XTy0yIz~fuWOTgs z%D8Haahc8=e&=r8Ev2~Z>}Gx=t%dYMd%;JV8nZUS28O*ueX4IGJ0@!*XUlD^RdkQF zj!fsyb4zo~^$@Em(P%?{8N?AQ(!QNe!%2-FqgeN5$V=$>pXRd0O?a_=!2!?gZ$#C4 zB{sbvV5y$M>UE}isJU0=)DQSoCG8bWj5Tzeh0_Uq*a}-aI1X?`eCT@Z$Sa zh3RS!IP7(W;@lw+MXiAis!#Az-&Z%apLDzT=J}%yfT2lpZ$A!_RRinK=- zp)J+zLnZVx!YLN#PC~YB7&~mISV|Sy?Sd%bN0eR%XAnajSe_Mow8=7Xd6Gx zjNg`X)IPU-wcFtg^^>(JpB`eklIsLRZ9_x!(aJib)WgUN)K&ad3pI=YZ$`+U1Cc_W zG%fB(GhvD@Fqb9lFJ@aCYHmn)wt=AB2;oKG{g;DMq}|clsEtp^Ds8IN+f0W|gn1!x zx>Il+`7VxbO3mY)f$u|#I$JH)*2kC1tEHcL7&w<~n<)vBaHIqd6?j2!Y52ajaooG> zDE!jGasCpF_z4BzIq)|X^tL|vb;?~w) zjRyR3gm)AE*q|m;8>2_1&rrn|zg8g9p8{|b|9z#m2j3~b%>iYo@80OBzEi?=GM-h# z&z1JTx&A)e=S*3vi8?C7VBawA839XnOA|iYc{&44d-uE_t&g1d6GI4<{h}4-76Rlv z+Ri&#(I{2Y>U7^NsBv>qynjwz&8c>w=%=}u`Oc8I=ari(PYg{gxArSva)a^*9u)Qz zMxba@V-MyR2fAwmoZco_Y$Hfy*(_Xh&1{GII-Rz%3g^YUKXN!qVeH-&3pBdYRS#XH z=U$Mq)KjDMzHm@H6hJBvWNOfA6gIy%lq^?O_8~DtvgnMMMuhUZz8jLhv(RqYis1|m zL{RC;BL~chHGbLeR)EjZ!WfpFlPY+{ywygqu;OauEK#-9o8||fKTAm@E2RzTidD=8 ztzDBCK)=4p7#`&k)q-?>e`p-b%(&C_a&7WPjK*uFSX<61idnAGVfk3d%GP^pj(eX1 zb#^E*x+@iF5v=KWs+VUZ2z;FN|7Lr?+Trip#TIu_K{hp6lG>w*Kjx@ES}~DG^beGWyiuO?! zQPF#`oBwHoG3RbK;0wFPK*JiRds8j5UubA*;aw&zaY&MzAr1WmRa454JdEN}(T-qw;G zh9f>(_GP+y96XFU9tO@IY7?b;e+!v=acj61UG7fyW2`DXmb20t-YDMn z)sTaku=H{65Q?VU19So#aiUyN846DOPbYhs7w>ExmQun8N(kdlqUKV#`?;Tm18hjxj&;? zZ@gwkDnbsQ*|L0rf^}aJxN69-YDPb=bUVIXasozM~p)>x2HxK$Rp3J~s z8hD!N<*XQxNYp*UGJOd8Dv%B9bh=x(FR@Yv(J)A}v*(|&b57KA$*8iI{5o@@zoB(j zVS)Y{>+PUx%cftE?26eYs!PRhz4fJtY74A|*Vn>xaln};DMBqNS4T$Wd}adVd;on~ z!b)cMy?g=uX2u_b_md{<+$*cgM?SPYi(_E8Ah-JDClJ~>@v-~3N&U^nBI|YO8CHu# zk6Dh8UDBNfIxQQEBH8!mEAjNVh?Dl?Ew!;JJ^12KTC=;RT~^+iT<@sWOElO!C|T;4 zW{Z5}oXlv-`Xsspkix!&iO)K@THa`1_>tm5LOanzoNn%wWY4fi;wDG69@)yllqfr1 z<$%$PAh)HHt|%M!l)hIoFv9hisDt0f*RrHmjet>Ua7+zpqG0dQx)^?8u#uBOHKK?bIrA@}S#4D0!V&9<>;M z*x(sNmF+pw`bFPIX|ScUo_AKTYek@2DCv@Lk92#W{e5(b9X3B1ewi*W|4jxy_qOM(?_^3J)6&J;CoB?~5O! zeMU{|?7C%c95btaR*e@6Z7*prcYAUT+mmK2pmA*>kkfKvD@77`2Pamy7@iKNc(AH% zi0|FSnf&aA+qZ1MM<~=q>CriLSN72Jcjd2^`PQ)ty)SFJ_O$Gtl@m*t+h8}k&-bPH zOUw(e-gz?dmrgn|DLq7zFT8U*;g6%xexHP=uD*PE=1KX)@S73M`;PCHEfj}y2UyjD z#*`n8(aLH11T4cCj@mM69MU;QW4_?e>&w}_?7vt$U?!)!B8mFE-l3*)bGHZwFoa3U z$!g_k%@Z7FN_D&F_oFT4r0GlR#;-Y7(A-~LhOuT+-r%dQWOnhlnGf2#c!!QKER<`Y z(0-0}@bxTHqL6q;j>;o4Uco#22y%P%DK%$V^Y7H%(aMIDjFn&@bIx62XNpe#Qap93 zL;DsCBM?`D%1%NT+i=xFEhJS7;}1O-)7xr9Al(w}FVYMX6Uqm&HQ=6}Y}tU@+?1<{ z-gxWei}KaN!E)b^zHJlawuX^`0heVA4Dw1$<)>pKZ#-6(k6U1Ld_5at%BD%?E8l!fnVnwu@D-brC`PfQIH#J5$ku$s^Js3FLzZas;6_T=M_&VT z7;|OFV~I}THAgq43acQ^K9aFKfe=kH<00WaqKX0dhqj4Adr?a^9g~IHkz?C!;fyy8 z_i#YG<2{clczIIgR%cm!KN|&;?E|aHhD48zLsuB<^Sfz$;p4l{?GkdMr54mA8ztZE4ByvZE7tupPgvv*V~PD3s6Xkwi0ys zc~|SB+ULQe9l`UreCCEf^blE=$mLf`dh)KtP&Dv%d-s}OJ_$B8zGJ6#<*m+Y|FT(F zaj8Pqn2FNaXzO9#Gu__f3&kVJrOQQ+&wa^HY@CW2-9b=Y?{iD`WhfR#2qhO~2eMoi zHF`L8&R5S~+A^()kHI6lHpG2=jLw<3DoV#N2(LX~E)V~WBdzUKgEyi!6?m%abE2iH zXF6Hq`j-30OpFBI=)9bhyJL|0WvJ#!q5m8@h}ZAO>oUvhl1ZE5d^bjHrK;)Kc6*o? z0>oEShc+;3_E#iR9dpa`*q+SX&Z#VxoxFre@lH5&?U$GxXD##^(E$Swk&h7_CD9zios= zl4-MP4jPYTHFVAZpXXWG_Z;b%)i+3c4?HFl;bEN)^QJn+yU|`rd^BR0KM~h?7r(ah zhrVL(b9fM`k~1bTu`O?)(m{sSCe9koP_buhyBr9^AEo62?g7%Iza%UF1TbZk&JM26 z4dwKGI>D$GH-F#FsvEpm`7xyRD_5$q)f6xO6_xdWwi$4yZ zeu$RAJ{@X6qIKD3;6by;*h^(>x8Y5-wv#Glbf_249NHziyg(yin59OrVGDqxOIRK+ z>JE_^GCGd8*Q8GPncy$sl1?N&UVV z_$H?p^m8%1X?RQIgghRehyzH`J*2mZc=6X{kgv?CL@c&9+{fkhcAd09tMb~Jj+NTp zGv!lVy5SF1KP01xh3q?Ggy48Mpr`Zn>Ny%HmB#aL)%+Cht@ui+V)alJe7yLpH~_6x z&j9byRl@-d@Dtlg`c^jJ(*@O9#Ro5?`RAzS%^7=PY?f>x^#u16y}~=54lv_@=el4X zj02{la6m$F3=VjL1C9n!Cp3_KxrUZ+FDeT3Cr$fj2f|9i10H6)pFw`o!M8a%9l)ko-2`On`hR z_$%a;`4NL44!DA)1+RX-o^4!dUJLnMHSy~(LBzZ{%al$ajP1SxN?>|EuZ3|T_3W+^ z$I+e1t#S+*4p8JuUw}8&OX>`ZSAjEBLuXG}nOuGOu0t*y-c-Q~CC-wRg6Xs%A)lT@onA6CD#OR9_I#C3cwu9|+z3juo zU_u#d1XT%i+RlsVk9#9#9>NNwl4{P75$g{8CZXB995GZY;2&s(VwxU*k+^T_(RArj z7Xbm)ygvLdx77hafETwAX@x>Aarr4dAeoz$#l|*lmCF(tdtx~CjqAKm6=*D+4?s|T z_=`x`{(74lvl_vPP0mCFJi6HM^6}m9#Lo6*t-lHZgR7Hku&B0dJ$f?LenB z*=YO-J}?l!e60BXlY@&2C>Q=voyx_1@N}TA*m?a!3EYk>^B=uiMwu@UlQ`S)epXLq ztUt#uYHo&M!2!{xnK+Mvn=-<#QpBOL1I;$g-JwS362k1AbyQ?rE;~Oq7-BIF?}}!0xvqWHwoa) z6t@gut>YUQ#PTOMVV!nLV(B{4Ax#6p%VI0w_x|^$pN#P~=O9DunpegmC zc)S%pN;QHMcLbGsf$OC4E7pOIdS~6|-{b^dRVC|{yzg%|4Xz%fHSmU-}M9d&NvQ`9z_Q@&lVYA!5gZ+ z$cxuLx`6{0KVse28wjQ%v`cF2%Fa`#&@a|`>G^(ZgdY-*W1(LIJI zmh9qy!=2aI!@>b=;x~Ra{Y66ctQwq>sLYd`OL43V_4deIc`10O)EjJl2G+9P;-9c) zghBw9yy%Xu4eDeCh@p899|BlJ#J|m-+(-)gEiL1rM6-!z`r-hueia1`7+D<-sI}#> z6za_9mpmQ-66MPt9`)(K$YNB)gW>}yZTM^`{?sjyJ!0;g>mxM+#O&6fPaC#1%HRi7 zI})gI|5sa{ST&Cb>@?OgK&Iwzlr)~$nPThfeVDVziN6x2+pp8?ZMIjk9x8hY%y1*h zeuyQK`#SH4D7HG_fF=qA{98I!)*8Ph4`~kuR3;dxDo~8qpq%cfB$oO_*^-g_gD#Y_ zFjhGw4SvG70XK43&OL2>>)d;~L800NZg-LinFs-8KATZx>qZ1o|LpNV0?ph7aR?Vp ztTrBWJdeK-NMU>fMyCKi9Rt6YC%oGny9m-*L^dMy9B(T?%jBLfdXpmwcto>CdYKoJ z5uX(X;St9HLjJsXq>B2o<-0jGye5I5*HEGXbbcSCM&;zZy+JAxJU?c!>wu2LKHI0s zL#`3%@r%QxL$0FPktaq+lT(|R=o7Y+wo`p6_?y1`QAoLlq(KMmC`+P7j5&)0o(pl% z>Jiu5;?MYw!q?o=m<)VZ8%j$;?5CBlq=aJ!GiR8XMwbXYD)xs-pX_=PJTLZ{Y&?*x zxrs-wR9jJFR~5+?DarzJcKPs`S49$Gw>s^6pQqNk6w*OO9oo_nwn_Cx7F>gSDQB;L z3i5Vw;4Uq~2;`61GR9%wijg2MhO@l*K|#z{=z27C-bMUZ^rm%hois51})N5-7YVwPipSX5u~Efx?SgEnom zs;2JK3nRH@Ru)@06u<&r0Vq5u@CkgfcNal$dPLq6)KQ;q#01nx9ZeON_Y(4yPlP|jspfs;{!N6fwNi?d&p0lhg?U{ z<1kQTkzeL?p$ZujtbKkVk*oCTOD&F%ue-2kHA2y9EQc!m0ehN1_%m$76aRgEh6+01Mse>`A(PkLvK2i33Np-()sX`0dgcLpj3VLAa^L9n zsQ0d%quU!kSZP zV{L7|%~TF15Tz>7-naW)UickZUidNZ)%JYkhdB=f z>D^44@tLrh0Kw~0`OaD;rSS`5y@3%$#@Np4UA9y}r`2Z3;zS4(SYE>i2Y=#4<23rM z^}Th>t8@96u5@V=rs#w$L#Igw;Z^Ae-!rjiQw@$^fp{+jnB5b_268#&7WuS&oEB2= zS4*;?mdslu7x+x23nTf>`YvRGLQ?(@g|PaTe_$*yh+Hve8UCJ544->rjRQ8=7Oe3< z{-xIUGku0I1aXKi5oucFzS?c8x=Wt@&>^s@+t#Fmy*&PUcF%-%@3i*@A)-|j+fUe`{d0q^oo3-ii#rLv3#OB=P_i3ZYC7Qf7 z51g7gkNjJGFjab-U+MGhA3bjo+X9F1UmSvg7K#e1Ql$GE^ji2ke?V*m#K`I-nSs`586&=84{%#| zgZ{X0iTLEgP`w=o7*iuGjA?+x;*ZmZIN((a4%q8D8s;)}YzE;NIcU=+4Ml*DvUN9& zHW5{*7Do4~x%y#tJ?dz@o$aLgUz;6&-TIuqBvSSr9x{IwM#o!_Xw-c#hf-4ZPJWmSaAb805 z%3dg{Qjq%T3SH`AYv^JAQ*2Mc_T-t(nerd$qGTK2ZfX-OOq}OQcmP_>M7zXAC5kHE z@m0;83}V&?R{c)7M7NzToL^1?Y=Xs}lL|KvSUlB9U=Ny5ps@|eKh9(+Fc-z!n5f$8 z7E@8FjIj{#%YFOJ7C@K&mx4pMgVp00j-k*9qwwE0vW#sK6$KwFEZNq%$Meg#QsHle z0E_aSP|UXN&-ayz2nFTdC0@t`Q=j8vnFY(cP&rsp!_06Lybdy&@a%IX`6Wza?C|Ri1ldcYeaXx5OL!5Veo9H~+wx@kN zz0BQAMw=LnEF|)vT`TZHrJ>V2z0f5t;ze`u!#*(UGXV3pFjlvQok1}D%Y?BW&NQ4| zzRJo{9WH=2?U=$hmpfII#QDf#@=%qJj6Wec4J)6kW<~%@lFjNbDC!S1_3Nj>ujXzy zTl1>|0iJFkJf~eW6HnS$Z1Jd|c+6b}ZR|U|A1-ItYa5)Mu{=_fF*%YoO=;k13+%iO zELc90t=Ja(^l%%Yz?^3JDNWhV&bX|s2P18+1W&WcR8#M4AoX(0r5#Y29_MD_Db?U@ z{dh9BNEF@Q?4uMareyz!*}1DZ@0&QJS&f`gGSmJfN=9wr({Ur4hxh8bUjHqUOcOgI zHHx{beadyYt?Q7lk`wzEG=3`34A&rX6(i?Sd zHDA^WSz>Q1XYT{;4Rt!DzD3q(myvx#*=k{7>UItn zh(e95?*gOJqZjQSAMzY0?joqSdpN z!&Gx4(Vc;tunnuQ+h+H+*PTtEwU6t|u?4pKRq#&0EjyLFTesf$Acg8uY}__$JJZw( zyQl5>H)K5OOT%quLb-QRP%u`zlia-S44cTd94`lVjjV(#>x7Wmg>_S_=YpF>!j4@k z%=(X?KD3jHab(^x<-u_HsO&rL!@pY`bI(i|NOi7iSiQBWxp5_#>_$LM>Zt2C8VG8# z1$xL@u649}B9Ks={)?IYrFG>ePXfZlxFN9O+8zqs2$@?gejnBy%E>`p=7Auw&X>)H zR!zpS-bo*G{2U;sHQkO0GG8?q~LSc(2LD+Cd4bb9Gf!)FU^ZcbEio>btHabuBaI)Ck8?$g7aCNj);x zfw3m4w(!UE0ASbC(jJSk2}(R()I{(yg*;hsn?hH(^Bo@4lf0ydj88rZjrGa;M!IXm zn*!umziER>s&~MKlGOP36aGQ(=i!wFOLU&kJ&QnP_mv~hzAAFi%x(#~ zf4ml&$GOKE6d)%ZvZDhp`bJcepPYT?ebv3F&~1bjQIS&K%E-aX_|HtwgX#KbtNb1^ zSDI)OBXs`&R|bhx7iYvS#@^gv6sM1WUl}|3}!F?ydFurw;3pVCBpsBC)n!E zU4I=}h6N*jr#y&gS9^K!7BTEb#3y+6r>w^GxgO}!d3zk-H}n-uDgDM#!3&O5cBoHT zszc7ypRYSf8zjS<^pNnL+l$yEq#}A3I+Oc)`9z=t1!ID4#o0~n(o}BVcLix(L)L+o}DL=SsoK(A@}@A8ko@B6P{RDn-UqGRFm_777&QPpdoJ{5u; z5i+@6hL3Z|55&|M^ObHdzP+$ex`6h7BuXv5^I9+Nk?ON|m+#COJW>K4jVyWWCq13u z)~LEThq}9iKKI#hflRQ=Ugk}_$mQu*tEP`=AfGy?88&3?y6^vi|VcA!sfBSMRvy$j@I~oNU`t-?B(>dscbhY9hKj(j8k2d`C^}xa!K9)9W77jm zd0+yxiGpJG%#Wu|Gq(m1h3yQ&Hv?gG@^bKwZ}j~~22sO`GwC#w=;A#GAxbfCgZ4h*h4I;4~;)I@$QQHAP&{cf4y+DZQ#&-KrtK6$qXRuifjZSCQR zAJ84LD0^*YG)#YgN9$uirgEan;zCDJ`uQctq8}M8=y;CmJG&N}hCW)sB7%E-dotC* z`w@<@x%^cZz_Q3)^b0T@9m@a7UD5Vq{$lI@Xw1_YijQ>9M&L#6W@pLqLQZ@3_{ZQY z!^?URdFI?f1M0*Xygt+60Il3625tf$R>ry$0<>ML=BOjPykc{(r6Qsq#&Ftz zcPZnk52omyR<930mVK$Zr6S9ZP65%e0YFKi^fCd+zO zsuEm2cUAp6!8+D0jebxTExfjIzba_^vPQeihU_*k{=E51^FPwpE2mx*{)ZFo2Do`! z!C0G+NmHhc=awnX8XF5#XNt%E49U8Jqj$*N@~jag7#v_L3yz==AfY>LAkDE{NeR$c zIKIu`I+B?Yz*Zl&`ZleNxV=34Bj+Hf4lOTQ=8@W6Sr?@B`_gQdu=KUUwN~{Hvs8^H;IT~c8#WLweAfW zy?vE%zS~#ATO}MmCp{SkxO}8wqEJ%;Zghp%tMS-^F7Ry*OYD`enG?eV^w~st?#$`n3b}m+jeF`{~Yd%kr8aK5lE1^XKIIREA1b zg6C^`%Do=+S(VmQSBMH|UQIAgi(7KjCQe1*eXhY!4~vU#FiOWi+I zhLOlRi7H<7=brhjtW_t7wcHRW$kt&c zprker&1Tg-(k8HsPD6btR?Yb*fa%ZvU!q_T`(L{Iqp3-E3Pi%kUi#Y)y(oezarrp9G~F>8G&`I0#NBNd>B% z6-ym|3D!Uv2BC{D!8n(P&sv+=4&7#PY^!U)Qm5)-y&{fRYK>>O$|~eJtaD9~AlU2F zQCkpVeb9sda9yW=Jg|+g`qb@;$Nt8_-tN1-@UvWV`W?8^|0>@7jXJ=4jxBkVE1lg^ z<*}3*bExMrn7YSwcY0g1te1i8%5;_7i)>eUv66t5DD%wsHP0`wUmf?`e}`7i%r~(< zpt7`Iu$R>1M+Agw;RVQepjhMIP>`7*9!c?VFzj|qFSbbAXQi1`k}W)@Dx*%#trz&R zshM5>VqYUnQmaxB?M>>x&^}Rg!lM$r*|y+YsK9CdL%X09$8TSs z4sT6&?iNV1<{hP_%f@uFrQlm~Jzi%i{K}?LFHM(N!=O~lnA&v{0f(IO2nNsHM!iHF zH>TpRKUApln9?Tx0{=xLYW`xzfKl4y+X8eU-Rx7?`ChdUQ5f|YyZ~dHZCr%|^w5Lq z|L~xHX>$gZ+=x$9nHOSXQsw2M(Ar|62dH)vpefi&e400fBI7V8`cIQ2h7p|9Pl6V_8;Nb+7+= zxXSfrsoX=+(EGr}d(M67AiSVwB`K*Mz>W&F_Fqk0cr4~uz1yDc_KMtj_5lDJjg-+{ zxyBt%oycOaVhvorVIi6@e1D~CYQX$a(8Poexf<)8MeBCkH(y^{JBoUwy%29p<3qgZ za(~SCm8u-@Ge?pA$>~3d+`a8`G=G)2nwr_-iuA=L86TRfs_P z6e?>K{mxenjSihXNKt! zNeWe9O+!}kSQ;$kPHt!Z{)x<~Zw7@d{~kVBqVnAyx;R92yw8ijz+lhw44?C6^>UQX zn&U>&zT>O-9wnQ57Qy0Q!cq%}ZFj<6pY(#QtVC_tdF_rpdIbJ+QeiJJ7M}TAq__)ppUsHSl!{6d&gr@)Ve^;ja2U)cPJ^#9}flVT- zr1VG82XTup->)vF+4b{>XJ&r4&7<2Ne}o^E#*5#bu2HQiqft_oq#UAAtu`_@#rlJ` zI?IqdZpL@j*YUCG^r6xilz#IS64Xw4fFan+8|^G-0Z*sA5#B(5_Z5jF#%bgP3G0K`W+IVwEy>)e`rHYA(GDk-vbzVFLf!R$) z*bwlTWB~4v;B@PH0-~(ZSR;J0^P`e_qZ3v!iM27W5)rA6dZ6c$hMX0^%@)G z3Px1DuV7+6xdj3j$2}=OMQkcdnL0{vY@Kr|$_%XEkd8Jr?4GFrqpCnB+=}Xu%r1Se z33@_WAKVSNz_*HO1#kfDwd3(jwaCi};;sF}X{Oi7nv!R-%|Z*msqdu>Q=`D-xRuHs z>&pln($hRR?S2r|xl=up@w4_G+6z){zhWNr#@~YN?>%&5#C65LHg<&k@_{ zHH~0z&g=SLX@OG|^-JmIU$Qy$;+D~%ZTtfI{@CVo1(Bt&&F_Isu6f@H#IO}$if_UJ zwxSg-Jsxgn#Y2s^2Nol_D_?xM%A2Q>Lum)@4-On?SnS%Pf$VzNy+jwlXI_B47II!K7dnF^;Q4L_WG2{Nl(sv0cydC)w)M@&5rf4B z2go~qqowr+R7;{>_p8Hwnm(vuaG2 zmKvX$a0gn@CC(NW`uaCXlB){dx<%;eKoS;|B1%j=Q8j8#X4~Txx>A%U>j70L+A zyT2-$Nj6Krs}bXWzWRf^{w3YYThaqazmZtRD=Vuq%WDPlN}TcSoU_>;jEgV!C>~fn z8^5sr@WO*L9jpn3^hsO;MtbG(slo&>zRtEOLcnE$B5BmnR9E%>rG5*hShYCWa?PwC z3dA8fnrEfIw>&dB*HXyr9OMH&9wk3B+B~cuKk}^0-!*mEga}BP?Z-=<#(~XKOZp3R zRYDH5uS~+#Heqc1|0|zx>KFdgz>AL#w}#kNqrtKR$9V4HhQRZMkLk}qZ1ilSA9Dbr zDG3@^6>$jRCdO+FKFp2P+j{aAMC1a9XyB^~)=BmF4bFz+5hBR_`aTHUboZ z-$`T&3>6LB-b0*`Op;9UrvbWi-rUa(FDrwvHmlwViI3bxoXN}1I8958z^*Rv(N};w zEGMOG1&FZ8rj8ds@`L*XaBLfoL~zv=w9-KE8oxM+>cXwMR?0ceR|&0qhz$1q>I+p) z-;Bz%iB!I4?a}y$Xz3X76wyFZ_xa?MRFfWm6(b!SDJLF7kf*QZH`JEFA8DS<7uE3M zr>A|@a3YIzUDmdjC@MJVz~+?lLXtJz>WPMhYOl>$OOTa&ypn#lI#fNvi@!pJwq^qx zC%+YKo^11;8s=65=4b-+$L#g@G6Lk2(`Pq>Ei0qBDr(18?)YWq)9#$K2XAZ$KT0%} z?3i!!C+W6O!InJ9sACk_Q^PJHC}ptJc9Wa!hhQFgrhM~Y7PDHG5gzAywlB#V48#BY;@=ShAW@+2K3DnLBdVQR@=MVI zG?!UFOw4ib9J~e|2l(Be9iW&^y#q#>B5d5Zz(7-~2y(M0}aa!Z` zA&_|M(b->xA1?9YM-4U1vG1ju67IjsGaoD4-@4M~L#@{^O%Im07JKvX86sJ|kx2P} zPymh#cJsMgU*EknEV>J}h^;t2{cJED9GRP^{7<`l-g>|LZYaP|Y%vB6yi12E7;I7Y zVwH%>6cnxY3E4qSjVr3w+9kZL9F%|1X4icDX-yQy@)RA?BqS;_)=>f-zn%bM0Lbdi za+G|srxjP<#|&ksRrdA!XvVo(0~_xr;zc*wU1=`o8*)^dRB@sL^>kIj?TnYFrDL~0 zPfArsFjXNa`?XMJZ}Lq~l8EEX8HjF-ypxOY`GTSk7c5)9qAgkf@Xcjq^~e{No0@}( zzFA*|#BWqoJajvg_53(<;{E19`!t1}s}mjg#?T@=n30??}jPw@!C^EeJu=9*#ukCX0riWf)gbzZLI-q z0`b3l*Rt5M-w1)&(;b;bT)e9=vOgjk_=#@~MrZyT%Q^$E`oI<<8f}92dlC|gRJyv_ zkee|Z{mUQxM}Q#EZ36DJIV6qmJ`Lzcg4IsuP+*?l*4PF{EL3) zaRB?8UT+Vfse-|=JUrIFd32;E{W)~|R@t{q_>yw%rg@*p#?77P{BPvO(82r*%VsaO z5~#OkH3;Sc#&AGm7~oO%_cefj6X0F;(Y?+e3IO1LO+XT;-1^n&$|{%#(sTyjkEGg2 zpVtC?jtDw1q1LmYFzg-I(v!F^vQL>Ak2CD4)T-mM({Gl>D<6N|;^RTFgGk_+t0IUb z(HejC(O_@FoVG;z(dy|fl|7IdIM4`ju@c+lVp`{NPj6?s! zNlF})WSsVpAb|W2Q|`#pzNIOxW?)6p?GK9Z-w69Gi7uPhZP`47JY*FpLa+tk4+;KX z_aS zvCel9!%-u0f+q;yVu7D(y_-4sdyt5d25TjNSNG=YT|UA28w2+QKf@q3D6`jF`RW*% yetdFfQLua*<_6u0ux{8TDFGGQ9aQKBa#7D#M3Y;kXFRU{rEp0JR7F~ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-4-Transformer Architecture Diagram.jpg b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-4-Transformer Architecture Diagram.jpg new file mode 100644 index 0000000000000000000000000000000000000000..ad21a4f2d4d1911631322b1180bf7b0357f825e5 GIT binary patch literal 25178 zcmb@t2Ut|yvMsvEl0|YB$ypGP+$c&EksKr_pd`s6nMM$hoCE}v2FY1+2FXd1M4Q}z zWM~kffu{Q{_dfTY-+#Yz?%U@+_!i&S^hz~n%~3T*)x^$Xmw~GqD(Wf#4h{g|fWH8C z4p0WJ5E2p*5?mo7A|fWfLPADEPDV;f#(0gIl7^LuosE@=h2=Vz2=Db9LYyote7E_9 zZi-1rNU-zDC`gOTi`lcJUm={JOToIeDLf5@OuEC zir^ZD*d0P@ou@=M+-SsKCuCpYyj$5pt2>V5x@F}aOiV&Y&%nsU&BM#bFCZZ)bz53S zR{7q26;(BL4LyAWLnC7oQ)`=N&u#7O9XwuozVh<+@eO$s8W#RGA~G@QU2;n5`?T~= zIk|cH1%*Y$Rn;}Mb@dI6O`TocJ-vPX1B2ftCa0!nX1~K%5UXqJKQ=bEwhxbf9{)Ny zMV+1hwhITq`=?pp-+vnRH@m1nyKwRG@$iX$+l7Pc1s-@*_yincgxBuq5IuFHz9Ihl z3eDYw?8**e&Re=jS}XT)5;`u474E~|rv1aR|8Ir`|Bo#D--i9CU2uR54+oq)JSqSJ zoSyT23L^P8j$6Nkt(LSB8BAq18Z&t0{Jc+_aiTxG1Z>>ZSmI0rGm7UN64b_g-X20y zg>$2BNY3<$MuNF;xt~>hc-tI|0qW-Dz-RJnH}u| zm3=S%VdZt7g|AV@qOsbrSFzcad$ye?2TPT5s@5EtO=j^G%D4xsMMZg^9m}NHcVh$) ziv|<9^d@h!lOxL~=b_V5%MX{e1AW%;TAEsd2PG2`N|E2Vo~+8)J}HJ=Q!3hdGhH@P zUub5|Du8fQ>j@G4_`-cogfSN$eXEnrdqZKT$;^p?fnnL22B^`w577Ky9JV`5+26{n zlAUrJ6HO9nA{h9E`tTB!it83{L@#JL7Co-Zkavr4a}%JokzphwW^s}sOzloLORxQy z*F+>xpP5#WOiMh{pj3{6tBjQc~)h386&=(U8WR(&6)wtp}Z*7f!E7?-KK2htF*{91BMUR=&rkoJbJziW_vT_e)9#tag9d5Eh`-~oat7SFPb3Fd+XY-YKU1O`U zj*-~9CgF7p&WP_AVXmUk#F~_#;1#x$`>_N*)Lif1v_v0v^v$rom3$@a{H4dvwlR3| zajsD;KaY8zf>=n%D2vgpyiM8UxLK~m3AeBl8CHWC4#n=zNdvoQr(aej?^2REoSfU# zx!?EVg_k}FD!ulFilp~yPAWkVk#O?pqBP?+&LU5qweXHfpAVk7h}YdJ+*MZZ9mA6N ziQ_VrIkTX=nt>OodYXJ`tBF?`Afr@Rz_`38?9OTmTk^~p&V8Ck|MSDm2u3H*ZtEvm z;vhc6C8G0{u1YOCT&Gb#bwbYOv>r#@I9h#Gn* z0sZqFeN=MA4qT;mMY<-+ggZ=2$>+kDQf{4dh*3VVd2~|Pmrn*bpNa8qN8Aqgbkghm z`F-fs;p}#lo2%BloEJ|SY?C(?N@O_+KBLLwjJyT_eV-0v3kHZ8}T81_nVT) zlQ2dn`N<#+8-bDGAZBL@dJl@TKFK7{TlvT6aRqLUdHOAQ{N+~tc?{?x6V zRziCZLgg*4(%KM&1?)@3-o9lfCa6D>k-2{v;5gCiD(>z#ljP7u6ek(=Q3`R;&7^Xv zxMqH*ZXjq8hXp+!O>O!(kPAB2^$EI+2R+nx|NNIXp!i>Jpi_@fy(DAe#Q?Xl0IjMZ zX=cKGpSGn4_T!s&Fch$aHgNK$>^4oQcPfHA5<5$>itNINxEpTB1U~8rl8>d-(2{Hz zX^P~_{U5>9gVak+EYM_s{D|nizwnblR(qwZZa+eH{b$NMtDG@3-irH2j`#(y6RAQfFO0yTP7v@uOT%B~&TNilg>HAZ?r;F#Sb4SlxI zuNX8{fzoFKX9p(Nj;722ry>VKOBB++F>-(f@P1$cMBxT0qn_Oi(*7jsMIxc2cj)p$ zOMge|tH7rXt6qVsv=)Bkdb?GhOh>+QKCh8;)v5Bv3*_v@VH_o5v8Ih!ZKp&DcAWQx zmkx(c2wzUmC7o=Irc#JCAGi$=ZRR(-A;UYmZFjw(DE8k2tocFc@%Szr&bVmsRAS~@ zGu?}i`6`0P%xZ`lN9w=77#0Y$9#Hy(toNj0t->XKjqM{CYW zmBymKjYd-dY1g@BSk2VLamM|+Ix@3Dt`zIZL?7Py$w4Fh7+JngxljKmVa2TCZHtFf zjst(Lv5A?(`9%5QGsoypSio0B{`&chx~Ft)b>>RgJ)P==W5#mIm5<<8ojgsKU#>Zr z4oU7!_LVsEI$YAa0$Ie+l^dwwrJG3S%&cR>DHYWjV$cbQDZ53_sgIY?OjzKAHtLh5 z_f8Mo)bW+IquCxRtM>dg!ZrGPX}Ag9R8J7R4H-T0~CU7$#N{ zB=6r8SPK6MvtD7?TYi%1SjKi8^@flr`|Qql4(jDUF|<7b(*cG9G;w3uHsU zSlCg41)i#+xgb~DH?=3ANJlI%Yhbwxdka05+Jd2ku)uC1W&jlq;)-(`76=~2492-= zHyGFcZA@vPoyv!#m^YW89}>1-h8q0GMbH1^q8fSH3J2K_re&GcCmsQL7R>fMLA2^B zV;!G;U43YEiWu$QzlHv7B-40f{0;I%;ganH4OJF484p_?hei9o`@$6<_&L&eBgQ83 zzV}@GpgZNNb_q_CO6AiEd)IB#XM<*@U&Rwew<)vP!YYhl-tGiQ5DXQpRco1rBn7!A zr>Rb@-zoON<(!w(kD2~$pw26W_Jb=P)0&F~#t_`mC3^PooQ1mLq`TEotvfi{ z8G~5BC}yrQCbvy&a)vSh3y@ssPnPeOwa}Ms3ryVfD^n!K7nhc{YH+1gID7=PMX_Td zMkCt~zAmBN8y6Yt;C|uaaya?sbhX|ucq7%y$h@MTKeBZZzp>0ZpP~x4=t3#)c zVWxwaQ6%J&8b%9UN^3{0`>F=*w~{GtInddvZS0v+bqyZop@0t+O5r@V{; zV^B_y*E>MxKFz;#;Ak3Sfp}IdaOtOx1+voD6g&?{Rk9`G-PAcQw62-*$?(D>)EJ63*Q9lRKH^-vsd?&X5=f>g9@<1KQ zz6g;Veorp3T;gcJ>8@N>Kp9M6Fqp^VBfo=3lF!wC<%(y7xs#D93 zv@=*(GP9AwTug%NcKosV#gIv&qzNta*#51t~ozf58wS-=>3AU7zL)Pk)v4GVDxDXG`*Z*y% zuR9G{BZ7i6^bNYn)2JP8dGz3===q`Ly17Lq7Pu~l1lhRXGCfrB zg1)-(dNQNb_W~Z6{31u@nzsDS47+%S!Ik$CZ$jdIZ2~>xdn`g*P9a0tgf_2bY{+7+vLvQ%az(06sF zau3_0JoD!ky`%1^OESk*D=GG(>xeO@4}tRpQ`jtws(n)~3uIVef_#C~hk$vw zHW+$zE+Y<^F(FNq7#s+grQI!};njGtUGg{g`1E&^%i4)h4p0jCV(Z>!%cSVO*6`QC z40<;JD47KawA71)$@%r%ylmC~TYs`~{`b-0*^VODxW59^K*@GW-P5+sbfvtukfJPd z4Z&D2X)FQ3lVJ~Xc!12o&{g~n5Y{zdaA@xgl4JV6KLJtR#$)dYRme}y--b`jRjTJs zI6X+np_q#RVEK;py-PG`aSX{c=Q#R~^ZA>Uukk|9>TM4DSY5SDkxOxb>~@N*pC^Pp z=lF7h33RhqQu??KCqE=9(L+{jKp?tn6xoi%2k8p5vp*H{rh*o_d<9I>i~G>RmSBt+ z>$ArKanMk9WUW6K2$fk2FbWtN97_z{z4l{n?-I}#Re$psv`_?eQ{&JG1H@qjN}-)o zUT_3pqHZZH;q=ATq-~(N^av2MpVSxsa@$k2!BI7RP5Tci4zM>dX5PLz0`Qu~W zUePJ(HE8E7sAOS*dFV3D68b(nvQC~I3qVfUQ3PP7O9cYEt^a*6myQGfba+4y4WJ#S z`#B}|i%lGfT`sfp7BWnp)#Iy*4l>*%O6b@!U`j~&djHm;b>}@Rxspaplyz*c)z<)` zA=OYPNmk45(?!wA-L=mqWz!>_-<-LOm7RrFE1&PplI*EaKG2-eY%y&f$e&Wj&M&;XZTUX#*mV?fyD=dxF|4= zX-c-AT7zj*k@?-LpNqHv8k;eE%O}+!)m4KW%;xOUL4R%$U0%=*3>@-mryz`L0(gz5 zb0Q)YdSQ-j*SiYI)QFI!Wy*-W^^~HgFWSgvH0DzXxPpXQtlt427kozKJd{}3su3<% zyaRVl9qO9`YCFsq^{xvM6#*=UUT7gmc>6Kk1eiAtfRpEt`nuZ)BXpR&b`r|%H>>g; zRQua>SG3KnCnU=dS`W!Zeq2FZ$Omd%=C3?>k{W$gib6J_A@1dADYwP9_0PM8;N*uR zPcgx8=;av%t;LS~j9~-U`(x0#+Ig`+M2QhRllEC{WsphKY#Dr2K%IH=_E7AHgby2Y zt(wn#xPMewCE6o996fA>I{|~JCw=u^ctm90j4Zzp1v_katYZ!jVp?bmQS?UD6Sd6z zF5Dq6)LHI{$*@Y7&1t`712p+)f0@h6_GZ6+`>XGw)_3vjagX>Nj))iE)1|%cWph2D zp;eAfA#ayoWjO6lz3uc2VV`K6C2QJs1-d}iy+4AtaZzALG;#fSnBDk@lm-i|JVh&j zF{TcTje(#M)F3APIbx3FPZX2LhqjQzEI04Xc(~@?;E5% zSWa0Zd`G#^tIxi&?d1v)=R5D(LxqAM`M2>YV`i=(LVX+YegTpMPd=md6fvg6#$Wh0 z0S?n2<-L-I$0Ck3jn(k)xr_{Wf^{}@CK3w>Y}dw};VfZ+ znHtzRak=G6Jvzw}ME99Qr~ojE(z(GN23MnKw3N6xs{Es!uuQ99yTTXV;(oU(shA`^E)J0^uz)K8h-`z+J%mP6{U;OT4Uy>4s`sj4`7)ghFb%B`HL1x5pK)Ye}EQ&|~;_d1Y_b|EOU*5xy+wUg<%(|wj zvMq)9Mi_H$`)^W4e6xy!{4@V((Oux8K|yW-c8^?sUDy265xXEY8wdRLmyPVRpF0)- z|8l#(nZ-5$y=iiyV5t(Z_ac+gJ=->Qi0pf}-)zN7v{lRgP}fWwU(;b(TLyx(Rp_H} z%*ias<(ia~*_e->n!sikL4^KfkAR}DxM1Wg!k0-~NdgEp14thxX=$3k1*2q=UL-mo(el$zGH63C zy~S5Gth}El!eL(IiN`{WdvZORYRBBkO&Cn!YazQ9kn=}^Zq&U0P6F@7PLBoFQXHZC zJs?ri%|@kpDim}%X_c*swIuWrKij5V&}mnZybXSP#+|-&Q3>6gEQZoT*3y@t?JnRY zi9xapx3z1()y(;kN~Z{Vb5iw2-UMiE9rhPqH!}?5W#wg2DkKi|C^8-z*8An!)t$Gd zln9uSEecVRz9G3YFT-ME^`WUf=9+yz8(z5o>^16=UU%RNtE?w_*d{;p*(y6TD!aDJ z%{_%)Y!2Qn!g`JCs6(1Wl`DN5MAfG&kaMD8EbuxBdiXNH^ww$GyAr-;T24#qZmaLh z1jN*wxzM?&&*DL}b;qN)c{Q+}l-tDeig@OSy;gC7b)}ql?+9cei3TAYdeF|#NZ93X z)%43iJE|!gx=cxGYBu+{Cuc|~Dg0^kqcngt`e(y=BpWT%h<=)iqM~HyTsRm0`{}`t z;WCX}j(1XccvB4Ug55n{&!4AHF||kS->ZfJUl7~e>bD)n<2P>wsX*Y6k2-8P$a{&x zdBX{h!|`77oIeK=KG?px73-$fMvNrAdsO;B=O!_IlGlpsZI8^^9e7;vJ%y%S4Hz=1 zJHC%Sh{`(Rni#|MjH5unLKmv?z%XzDalj8o%Qv{JQKU*lC!GPky(<&52Ii12Y~+(U zDQZGI{4DtjOHLVyVwj~?5WfC4aJrDGuAM81!j`@9Unf+TrSC;YF+$*nM`zoRS{X$r2@_#j z^v;Y7{uIdCL4hpI4)P1}TLe?2g$_u6!&E~zXUZVd7Um|3X~H$7k@aLhZ$zIZl)m`d z9^Ze8Tm6^Q%A3?~q^;U>8rlu;sf`fhvyUZMfU@?;Y)KExSEi{URHP4n=CgPC(oY_s zvt0N_e+k3BkT{K07it!1q{J`@d2|sQ$>95}6Q3u%eoStg6T$iItJF3EuntxE#;L;4 zD+{RgwF=1w#ty9ACvT)ofj4m1wW*^Se*OBi^V#I|R)wSweOIdcy?Wdg>*R#aM8cUT zivHH}FY~u$_tsOgsTHCRdB5l0kb3rH-zJoP|C;wARCbd*ud&Y2nuOu8ToMb*2iWA~ zZT_bo9I-1_3CO1cU0j46f_opcFf>W#;kNUgN!|4kWL>-cJVA1w?Q)r?4AeST1wYOX zEL*oZXvSEb|B!aRZv3hC_Sly|>mOImmHnQqMf-YWPUXL36;3uN9S<7ux~6H%NCth& zLP6n6;ywOyDvg^i;oxrGLumq6oW2J^z^DcP;*FZ9Ujrygj0%JaI;#Uc)-^IQgMqc5 zHS5EI%TN%6?sGeaq0&AEB+SqJRbMi4{F7i7hzURkUW8+MPQmOD(LA_3fzh;qoQSq# z;%FEDrJhVgbqK|%Kd;B^TAoV^z7biLfU@qG+u^PZ$ z_#I6(SrSv==O_sl=wDxTxwz5?;(AKc_L{#S7T`>cY2W2!$3&X$loYd*NIRfk2r6D^ zgkdn#es2*e(sXy$`(M8nuFnh*JD)Q5`Ej9=Mx*JvBectUJ!9xOu9SkiT#)aiR?+vNIfVr>%X^~km3xy6)_)0hB}!N=0&@~1It>95|O zIR@Q|g*>eFZA8_U>@+9Ph&}!^z^?|s`w&oC`w-w3SF{(|nl-QGkw@&z!03NAxl)f4 zVFplU0VMVj1%qyHn&Ef)XPmMGm$!fen&+LOIOe}uX+L(Hd1hu% z@Y-MV)4dZ@%1A`r1b!^?jzaL@I3aiBet-XfBrFF;x8oaIW^KRf$CL4(;pziO)lAVN zcSYy4dBtMKSYsaP1TMRpVlIIZc_&e_`eog{nf2DG1sweeK6G-=WJzP*gEwh57K{lH!#mO285$ z0IP(>x8~wK{8nz;dcT35Nc(byD%sme`hkDmK(EJ4125~aokXiCAE`2Hr{w?(cfJoX z!5g6h`CpXLQ^I!M^-8b@VW+((YVeq`+HefF#V~#3GQF{G82q2LKfHJ;s?^wUn3h)bZW|#A~uhoGI)9NR5OR0Ri0p!|=bL^K< zU{ub-Z6Z_G`wKc45&of6r}pS+xvVIdI{IWXCQjc?BSEz2yS`+2w)PtyGc-{WS7F9iIU2pa> zi#+`=&1KU6o6TDNX0!Sj5WOVZK-L5@0#$A^R6D3rSAzlhmFOi?vNoasoh0{jRa*Qp1v5~z z&=NT)Lxb%v9sXe3i?1ZDLyktTZVsbd^)L*f8ph?l(DPHm<{htghJachyJ)r~*`ho$ zv}e#vi+pwPeHW{&a~~%}R9ZexvMLKP|IxF9zS18$S@zPp#`JXZxlaL&g< zz_Q<3{e-v6!z5h;`uqiXL;*(zyLW&uM#9gKt&;h!TmAjoR}+n0OFSb*Rs^iYyL z^Z>HLh02g>wsv~iL~yA3C|>3UsCxWbnz0lT0ka?cET{9Si#ApS7fZ zKcQEAd-s0r47)>i0wgz`gZkxZ=Yhm!Z!7AOmt-e^*P895kmg*S5BUvp)F zUU$O9xuPbDIpa;ed)2UWX%~sx0Asl_NVeXz$oPTvLgj|0v6}A%iQ|lGiBpX4&l6Q8 zaK^UT{HWK1w(H=4QM}zdWF3X1iE!zl5xKnZdUww1GssUH|g6QcQ6z zDKTIe0C6)kW_jtH@#W|RrZx%zWwT?-kMhXbtd4QfX^ZZn%}Wq&$Q0j|Hu$xX2ZkMc zZ*~k+9)bm+97TJ-u)r0#?Jm4Lio*BIi%JB_sYHch@BEBX?mTI-u@w)cbXC4q#*gR5 zxG8-pm{STxY086){>~)`mAK#tFhN#?kQ5=q^d zJ4=hde2DMksKD$K%-!Z9jG} z+|S83-H{bKciQ#aD}O51-`4%=v{dweNpk-HG4lt|?WD_bB^K@vfr`k5Ci~Cdx6Utf zIc@00%`$xID1w#l;J@y@Z;Y8g1gm(&br6)4C5ETjln*cR&ws`3je=ZxECZx2A@j=;yu-fN0G+!+)tG;f*7+p;{mQiTP| zi9yy7xq;yZNrmJg(WOu8OC}?9uu%I>Fm&@qkns3FtCBc*G^m<&krTy@DrFvlZ;05GfX=(G0^ z+iC}Q+MXxQz7m9Xew%=75+NXHIt6xQP0i>z4k`}Bz&JhiatD6T0yD_>L36l7!o6rE zLmv)+_OS%OMl*LOCFL3wtB(cS|?tT zizRT0sU;fUkaw0X;b$alVinkbJSXKdg-o!I!+ZF)2Lv5#zh<|5Ys2ab`U3J7?t3`#E+pPcI6hIF zYKV3E0}0k$!EA}Uiv5EAcfJsL@Sk0u<(9otT7_EKk5yeyj-;|oWs-_zJNWeuf{9qq zjdXMd@O4|Xxg7SWU7|k^=mm=_P!O}n0B(lx-6IG1ha(@lnG-`Q?yurWm1b%iK&)#Q zfi@_%`5N8Na+q+W`)Q_0`TcFulUlb3t1nx;*$S?5R0Cyr&m5dB*5>ZndK%bEk$nVsN zDtY$aab|H{7IZCB>&n;b!rglzF|oK62sIjOcn-3| zRhB4Nw`^tmdm=|a<{TF?#to~`o!B%5hu_|KDRuFvs=>+SG1c6)S-%-CuGRf&_SAx> zNlI71W+U(Vp7dX}6>s+?qJ#zP_EsaOX{;ck8@j*`?UpNk23q!UqQ~?R6N3; z2iLo&U0XYYkh`#}ui}yXbXWQZggL7YxxZhOT ziw+PNNhsW+gC|FSv$UH=Ob_hkCWu<|81jf;>DW4=9+i<~`6U-U)Pju7E>-u;TJ%0J zPM4wS6OgQV${uRf*k>jV>n&AYX*E^QzF>m9HElAx?_5>-94_=?Kqn;DEf%KnQBqE= z_Ar31GLOjqz1i29!vrDbw);A0p&VylTcV;4dKZ^t$1;umeFdmyq>Bi(qbPjrT>@EJ z|L6h-GT@GcpZY0oOnq2wNwc3%4rp6wgVG^k@b#}(Mjz;nhI)P$59D%C{s1iwuUI#5 zDwjaj9M`7mj^}2oIZf={IKOrB{64VC%ZpFPu&q8KFyaq6;rk?;A8GQ!nM0;28gq2F zK_MVc!eg}In(7MWRn6uhR7A1jL`WUOL&^dTuW;wuGS0qgZ(H4IuBo!MiM9oriS8mU z8}-i>Lz8T^ep*v6NlCYI&8^jA>Fa}U7I`w9r3S7>pVMx@tA#}OBrRiv6=5FalQ-=w z%o3GWri_Q`~*zs$8?+FJ5+z~T?u54 zSt6GJe3`k}^E^%e+Por&TwYzt3|7^#0);7=X4cz1fmC&vo^QX}$ravE{GhB)RZljF zdK&z}Eom6Crzn0z#tvsAI*J$>KC+e>af4Yn#0AN7W;rfW)=kw02rOPaYEH4|ZD=sx z-}~`-J0PeWhM`>wY2P&aUBntISJ%(aJ)nN3p=GflAMc$Bx#iRs6#hPsJxxY$(sOH1 zcTOF#qqZpLcYvo-aoX7eyFHWX;CMjneRs%wqLec9c$FNkW@;(APb)wl=RSLmD4Fwy zb|fw=sbFZTk5=B_CyvD3VgBO1v|=>YE(yE|UHfbygfq zyz?XQ;QEKkQPLx@G&1AU@|25^z8!h%-sle+VGe?N=aE}8@*C41MZ^8^*5u|;uLfV- z{~<_fUCn*FP&k&D0llrW0>wGgSVG2rm!l)D6SO*WCl(N!45A2-8z0p^Q~s{Fu8=A4 zGdE+;0{k)qqCNpN*`=Bls(+$vkImSWB4Zh?x_wT zz}DFiAOKgOM*PSrFfNQ7)tR8fs2PQtq-cc(?Zx|?(k}cVoEiSGDf%OJ2!?2%3CDvC zmY}zu%NzM95zy%Taj>ZN{Ff?!`JcdEaHxk-FFUll0 z2zdi&!B!jfxTn;&PF02JLRT-@eUb)USDeym5;! zSy(oRd6OWz@G91IR(hxNa4ynf`spshov9+Yc0pUCsWQA{@oG(?4WGz$cQ$dqxTG?Z z@J{$w!^3Rk?cgPc(8Uq=zJgQP8S9nrB0c#x3_b#HB$A|h``$Imi+;V+zriXHQ9nf# zzNei*dRG+j^h*5`)P*jrwVCa#Ps0KZKIsXlnTVD_$3m9dFE+srADZya^D#z6T;wD6 z19r@fI^o)n58&Yq&2XC-`J-~;;JaPdlNhL?omuud@wQFj9~Y+6d{`IIV{gKq)$?V? zfE`Rk0=NC!EC4)?>H}@~B!wCyXXp7iBO5N#=<*nfid;e(X7gkqh|#zFRu*onHTk*J z^}_85WgbaQWa9MmG?iRcfxWoWK_znMoMasf{HO!j)hwa*B^&>@GJcJ(2K`z!PCTRY`tQi6$O1-u@iFL559D9k_BJr;Dd57g;0N3GZmk%hlVU`7Zi8)m zyCpFw15glI)jS{m-L_ZbufyFfzNI@S|N2vdy~$PBqCoeqe(V{5k*<)^_-VHnRigvd z8(*Ew{mEn-F-;V~2n=59@91mbyHc#R;ysmf)67ZVKwHGP`)!Ajvs(X<*JxwuoRg{T z%v!tpsj0b-*q742Cvji-8z|DO6DeH{6Ib zO9^%oKSF64UU_n1e}sszqFj5;Gi9Dd$lj@{6S9==pzhCWXB{g|W>535N|h|Wq?_H| zP%fZjgJI3K@hg3L=m4^IRTaOuQzr6jMJT=;Ag=!6aO=U8r<6YpwbmaR zgls~+ZgQwHN9^qM=PXjbtnnAC8}w5XeQ?vPugdjv;lM_6FYbnPFC|{(k0z2eoVlj9%N#ycdz&dQ5@Yaf=Ky=e6Ze6 ze<-{!@ojc{Cke@pjt&G->GLNL>2k}f8X;f~-ru*XeL*4$wh_*dAKsF^;B5SFb++b* zR=>sb|7`k9?)Yn?$baz8-`w+`$Dfqb01^p1(sF5Est4K6fRW;#siGbVL@}=Ei^~h> zCCfiGvG|uXH8>sKk1?w4%wQpXy&V%lJ8Q(c;&R4@b^)vMDIh3u8ChEdxFBv-rGANI zN+}_H@6=tmx1mP8&hF~vf$8rbozBYW_BEpi1D|`8kX|lXUihNxm4gGVNeQkg7oNm@ z6MSK!_z0TNn*|3OtR7&rT9P!E2vFh>*TSz#yZ)|`%h?9F?JQ0YiMgx8Uk7ICj$2D$ zF8Z+ka-bX5!Y(Skak#o;uy#n7Ud81$-y=8xrpqSu@$cuwwTWP3=Z0GKp|)=yt8yh& znVvbsPN;Kun&^xU&WlnVGj25Wn6`|G@Ma#ard+M(m%aM^>|R1|*2PcuHGBAtmKeQa zW1(+%XX1&ibiZ-LKT~4lx+G9{^mFtrovx?)dUek#nu7ev0?PMje?$}aq;pQg@Kp;tpq?+tOb& zds_<|43C%cipr~OsSI0|s|!8ssjfTDZ>`8Ed~wCTT*~VCUve;(NzFxTd}csb#$0%X z_A1B}!}BAgDtu%T_cgr<8iOH%_PlOG)D0{zA6Sn~``UTAaAYDnzn3#@GpBma4N&2~ zux)tq`0!UYw^R2)Kx~H8FmvDgNdB%026mlC+CJ7TD(l={p$qyVi2j?*Y^|!SJMXKI z!?T*+8a3h9=%OxKz+~e9`m-hd$LHKX| z1%FoA8vo1UTo%PFxX44m<^c*#6BM*y8{X2CD;rjnZ$467-3D{FqIO~+2%}JAF{1R$ z-8yNfdV%Iz6h4GxxH(L?V8%Dk1`8JUv*Xq+hRj@|TKE+fF5}+5ly-vF)tr&ONmJ6y zyN+Ax$NtXAc{a~RuBsj_$OQ@y^*!vUo1md~V6tWdJC;J_fSXC!FJsi?UCOVMLV{{5 z`BvU&QI`i~c-=_FP?uR^-e^IOKtdatll+Rb`r~10Xc?M&p%hcN17yK*A#_>73 z-AC&=dWLz|?$f=;vR*Ng(h_raD@dNk&ngXkm4NrQqVuKQx_R(!5{5PrIyQ%hLo+~6 zK;5=V`<&Td2b9w&F}VcJp~I#5a=B|O@|cnWC8&+uJ3ig= z3EqC*eQmxedyJARkL_O?dO6v%~O>hy*dh4p8yIaMV|%; zzW<-Kw(dV0#{Zk9(!YLHi2DXcdc2-#IPt5@ODkB_UUAsgSC`w${Z%=IOu1eclL`!B z4z>K2b?6e&wdI8?$@$x?_n=wdyC?@+6C22tyWnx5Xg$%l=v#Sni*pP2DlN9Q4S5N? zDXH0Hagq;SqeS%OsYz675(z;*1yiJPPz)b{Y&spPV#pvDOn-BUe|3$iP>R19!@W$^6ru^pw7ykB$sK0uIq~#+dL4I0+*^;?; zR}V2kix3G892 z{Uh7R$N%?v%(8apunV5F+Q2#SwK7!wJr`<}$f;)VeRbP65+c4tx{Mxs@q$@XFNd6L z)te_==sjB}wP-?SqXRuR(!C-?(d5qQqERpQ7Z+!JUNyD02n*?{NSm*?>wNj5*pNjk z>(#x-MEb1mkSi|7VXW|RlwaX2c~_B2FQfLO`(_||WG`2oGdTTjQZKIZe}sH6`%^US zaKiIl0T!^yhjtEvNf$Z}4Wwd$w^^j^NN{%==vL(aX!nM8j)7@AnbhxgZUa%2i3J|m z^O^v*@?ZN+75;y`maKveeHWQou&m%PIBHgx{=?BwZPjc&u0Z41To7}I&hl3WfwP;o_Wsrd)ueo4&(!nyc zzxYsH&q2jLC9#XsGk!0{5CNhcrE3lwUOl74d~0QSJ2#J+AAy@S&Ku&c*uiTTo4pWA zSHB3mh;YzNzmltGHold}fCU0#iI2+{g9IE3zirW1W(`SDWceoMH;RVo3c<)<;)Y*0 z6@XcZUqN@2BU;AhwwEYs<*FU?=N&TXn$rr5_vrnnMf!wrp)@u>`?P=HERCC~p(j3$6*w((Ffw$w1sSI!at0G{9gH%{pdU3sRwPl5SU@ozx||Jem{|Z`E+Jvws+cfy zF#E<$LW3dj%|$2VSbPhLqJgcNjiPzl$)OwHDA6W+s#hY!z|~W^q@2P zV?BxXnv7)9ifu-Bn?&28@7xgKd!gyhFN1leo=dH{9cyIEl%|E5w^1(4;hJ{IulqUG zT~#s4B~|`f6Z;eW$o;gU9s06vH)XrSEXbZ*5RKm$!CSS|RcC4#uM)>0n!A7jqtXmt zYoT2!S>Ri8WSXxWF5ix=aJRmQ{}j!ozMIqTVea(5H!B&oOunc=hMgQO&MMwb%O!5! zN>5z;LE2#GXrGhLb5X%es0Z$8^}jn+{Uox~g=2jVkc7uYkk3wKQJN8#xVs9I*gyHWS(J!_g;8 zw8NRBz9>1zuC$5^ubeJ+J0^m zpv@ya&55VD{dz>@HW~7!V}2HeG9s7{WsSH-1MSOrb#A1bdjo9vm__Gc6jt{z{pHq( zvh^BEO3$d<9|@+Ws=2OiY$P|V7!{!fvyppyWhbwnap!gGed>H`Dn%YfA`WiMeTS3x zpYQPSQxNk%K0PXya|k8qvXf@5;VbitaZXg2_eR#e53j8QV~+phRsyyZ%ip(11x zZBcw8eOk)!ORiPw*z7U66=G=UXhD*8mTY7*fj|gnt#V)u=Rk?ImGJYb@uDfc*gWrn) z5&6T(vw5K6$Hd+fbrJOUw68x|-xlh--rY!?CwnD-QfSnZI>YZf-Son8Q)=-g!leH! zWH34^ROSayhvKIo*+IeN)LvW#4L+&I?{hQxaEa?5$PYJIWnIk?lICLZ*c3GCGP-5q zZTq0HA$~xju77`{DzixOIqfDRQZ1@Iq#?@wzZ9iCLmoA zLT^EZ(4|TVDAg#Pgepi8A+&_v5u{27K|q2iy+q;Ad+$w(^d5>7kt#?zch0)!uJ7D) z@A}rdf6bcxM`q35GuiV#&-4Cf8YWL(+j7|n;zfBh8qt}rFa+9AU<~y^pJzW@`3L!inC6` zu+A^t@4}N~by#&jOLI#m0K(h;D2lu98{pElw|;jH%a@qNWwXfc-06#tS2|OiyEf-e z6g`P5m@?T)cN`2Ib*`?B{E*Vak`bT7%vbiMv2_SeEUEC2k=#`ZW%DiGo{(p3uX?~Q zzFj`7?M&f~`MDK9{kWbEe3TCZ<&1%x%yRQ?;4P&UlR^pTYZSz$JXP0u;qicoHC+%P zClgVSCdTLbkum5X4OjV1T$mn^@Ez0bmmyJ`mt`R{?-g~d+E{VaJ9z6dT&kuN`g9#8 zIXW|n%HU)FE^9XswQNM}G;Pq1Swi=)Pw1^gr#nS1qxNXkd1EiRX_ZGFS&5}c>pQsg ziD{BaGnEdLspJ8kt^JHkLUViz4W&HqMi;Y-rMw?kgtpo3d@ zb}GbH5a17ctqO1md(poh0W_D5t zARZXSVD7owQTSxy2;}$E!#lGbog3r}^ZU`%xokB&#cEPL z{YPsiSA_ijYRco>6Zo&<&>GIVv|B>&YafgZF@7&=B{3w524GpN*VjsxCN7LtGi3HY z?dFe(Bkjxt;v=YrM2eVV%~ByrEl)`S8)H-IL}Ux0Yi#OtkkhrBPa2Z-P31V6Ip$qT zHR$akYhf!fYcP!^88L_b!2YX5m1TLQ6){pW(Y2;7n8*@qv_Fd1cva;k&e083Y==)_ zahIyz72^;yc-VE)buS$>t-umxcdi0?bc;k))^Qo4WeogG_z>OHEK7TK~{R=1+prCZ@hf5uDN1#(Rxzbf*zh>70mdfD<8E6?UiID2l% z+`^M*j8i=kYIQq8`48E+lmas;eshTa#B*-rh6Xuy83JOx|T3x zoV$8s1uL13%d-@QLeC3FFw^2wkxHxu>P)_c|fiNRT~)n)m2E53Ga1!m04nj7KlG;hCG-Cq zKson>UwPlTJb(%#n!_KZGx)lDJ;unsQZ(G1jSg_b!EBuJdG~?zPl!Y_q5X-ZPX&@O zakjh#3&)oMjlz2-#>h@BLSsA0cs3{SG`DsiXZ*9WVU^>-r&m+qyD;mk-mG2Uhyyf- zV)k#DQ(ZKlj8o*zwGwCSI_n9p7vE22B2KLN6-*}nUD$SA34oP(wlp7c#9dwV{s}Mp z>-oeW`|A0lj=3k6f>Tc*zF@v((*n7h-;fs(fg6c|i`=Oqm#JxyfBTWI3olnLi8|Rm zPlWQRW*^3hGV=FhUj-9%H?k|k4trsH38tA9I_>*yb6rsxM~sUVRvx#(!=>WAZxjq% zzifVpdJa&2rePDXkrC26f%~x5bep5ENaN$Wxd)xr`bhth`_TnF5++v+KVQN&VT_k& zNk9==5*LaXk!o+e|B%uK@0~VP(9a=p2H-6G(`n9DIW7ogypM;84yVvu&fj0xsQ+0op^YN(${V7yQYocOEr`Eu20?v$d_UOB0G! z9huT$)%HloNJ|-MY}Q>kX#I?qU^lj(GSxD_uQ5R_2{9oI4zg!D5?VGq7S!af?}c|R z5`suCAaw6MMgExT(WfZMle&!A7`Xs3niJQcUG*i4N;ynw(CHRN8#I&b;R-dpNsYtQM%e3c7H6V+PF<{z={s&`cM(0 zQQRWvu0oq2e!W5(_f2=^U#<2hr$yn2AQ@y`uF)ISxu}0l4WDU1|G?quN?iDji_|9Z z31Ps)#NM3K`g~4XBt?TmjmpUeez3nUaL+5i#7=zejOSd00Li8*%V8g~GZHuq6y?B* zPPV^N_fuMZ@liVgSCqpY4SM`Yi+-&3W;nlW#x{%pqdpVa9AV;*r~=EJxB|QZnV(dC zTkCJ0LQIC_d%VenB}|%35k36Sr%~2SllR#|=Z3`-#0G{ShgJy*?Ibf~oOpCMn?1vi zp%g1+!`wLga)u~RRd-9nFbm8y1nb(XQ29v(WNqdJ2tG&+!D-7xF3JQI9&Ys@3n+g) z8Bwe+8sP{;a^VSRVgg=*aY-#zev`wg<~L721j#%xX;b|Mss0VFzMETpu3+x^d_OZtc9dQ&%fdsm(pQ!R*6-WwmO~_yw5~#~pE+``f~_6H(^(( z>}M9X^-`p*;-zNNXm8Em#M?TJh6{AQ53Kk9W z8DnYp$n~qNsa2mhuvS5zkCqtTP#?l*Nb|^VlON9;*1beClf;QIaw$xlcWZm>@Rk>a zQ^Yl5R}we$>TN4(;aw^F44x^I{lq4pRlzMxt0Nm8((jRK=C*ch(a!XkdrToC&A^bQp`RDTQbB_seTfo7(HM~9u{2KhFQb*XGop@omPm2DdAZ27U-JE#tNA?XvH7|lQTL(shR(ajw3Sx&Is(+xvDP%a#1k&7W+o{g!pmap0C?ay4{03f!q&7(nM>mLz)ZMXk?oIc^tMPuni3h)1^^iR9h+d6e zEu_D`xw%u}>&y2ckuT3yOA7s*H1O^DtX!b)*sWYsg*Rxgv2IF60S4c2Di0W_27z8^x9Nxk#G|tRLhg4^LTVSl zN~71hq^}&Oe)G4XiWL>J*sG|!MI!$n$X9ORP<8)#$jT*3cjTyE9x;q~v2K&N-Q#7Z zXU7^$COrr{S6kt@Bu{~zN&o@nMusQ^pC(+{d-5c%1NQ62PT&vw=8Gp>;|g5tO%!E- zbXEkEND`alI@Su_o^svVlNYXehFXEX%WMizl$)3ak9~2d&iLg_rUWd4I(w|Oip0=b zWFi%Kg8?FCG=rA!k8VSQS})zZ4I5-K@q9hz&)(fU!o6;*a?U^1gID;5*NB>MN3uSP zT8%z``7+C1AOveRq-@VY8?HqH7;sAyfv4%gTjuNw`~poB6(lT}augSs)hZYyPvzrC zZ$v5=CsAFue_UFpL=)_wD-Jfe`U>1u>8Y*qKHFpreqXZL29-}%LY2($d3b}NKU`oi zNY?7s;un=y_?h2;iA>{CZ^yhj2IqbVlFNTM=I6dlCH}$#%U(VqD2Pw$sbgH>-W)Qt zQC~VVhRp=h!ZtMkxudllXuH^~&@G2z)ek@I!Xh}1ATxnW)3DQYSVa>j$62Bv?1w`Z z(2pWWZ$vW#uU%)vLiTZ4yyf#=3SD>V-YQY=8Yy^k#(!Kpx>oW>QvY`luwT659^*J+ z(7L)^bYb*mZTvXvTKU#Z6PjJ?Fi|iE^s`_yHpc4tE8eiS`Wj?;4oR50Oet>1eiLoLvC!D&)TvYJVTY2iyfB zFrxc3m)_ODGOGhrD?tDu0DSemj|%T#7GUQd5q9%UgT%7bL~HJ^x{!`Zc+8MgR))=! zLvQdC`ov$mOKAL;R-(T?>eA*BSqHZOqaYk`_D0z2i+a9Z$UEXp?et9H{E``7)d|F1 zOv$taJ>LdV{}X7JME)rr?0($OOwjA>gx75Lz{1i-ObslGjh2&zD1$a zRS!-pP~5#{Qx9C;I&ocP`oiNgOtxxkYb#g++r&|3!cr^(Ilx6RDL4h_>w_=oUJxm8 z&@EJu`wJ%XSxn{)M{2}#NOy36yC|%pk7P~?_AcNUKM7lJ6d`ZtaY_pCV&wCl7Nl4Ven< zm2nb{Cm&J-IzDCc=T0sAw(Qb!Yt^48%H&kXw8Z2*sW$-+{}2YP#wHV_=Ar7w%T;V4 z45M?VrVB+qHe6Qfxieyt_~VvelwVU)S4f-3nY&DEqF#*Z2~ExseGF|`jmJ>ilV9*t ze{ytn^h6V*D?4g66s!|cTGD1LNXe{H)K{sZG5_gnQuAsw}X`cMa zffHhhU1?2An)9%`5gFH(?k6H3B0r#B^~HoLaVa!nU2^zma?-HhVS01EU4q?$uR!^r zM0oX0SQI>bsqp{wk#H;}Z<6F$i%%S7D1GnN;fAq6Z zM$&yo?J3`pEXs z20B)u;a$@8e8mOWMZxT6N6m#A2_qZZ{b?X#^xqlr-pWt1_kmnYj&T zkoQTG15BF5K&h`8LZSIj)>ErnQ*`d#^3a74u_T8ayz>MPf-Jwy41Fq^Z;S#vKCf2w z=1rMz_K|w?n9Bkg3c+9UrF-Q2qkjCNk&iaq0;dRRLG= z^UQ~7m>GVTaU7sFxs#OWiqM0XN3}u0y9th#{8kZ^{BJD!SER^ugZA)id6EZ7rP1h2snf~h_Fky#!*da(4*ULv}9X`wr#A3 Wvba391rIaHM;@2{C-V6B>3;y#pXbv6 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-5-Heatmap Visualization.jpg b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-5-Heatmap Visualization.jpg new file mode 100644 index 0000000000000000000000000000000000000000..1d878edfeb3e215a364946ebe56a0234d2f6e3e4 GIT binary patch literal 56322 zcmeFZ2{@GB`!GIIDJ2zBqNyZXk}Q=lNtQ^mXDn06G9<~8{b@rHlY}V7zAwoV!W5y! zzJ}ss-}i0C%sk#RgVOi6T<`DyzuxQp|F7#?*D&Wf_qn%opZnbRea>U~$MQdjovKO~ zl@JUJ2m}N8k67+NC?K|M-psUl(-tNsrmb7IFmGpP*}iSt_Ip5`*17s z9?re{xcH9p3kV8HNJ^bLeMVa0f})bL%0*Re9bG;Bs|IM(n`Y(~mRKu$2S=xS&MvNx z9(#HFJbCK-BIspsNa(Au@R-=R_=LoyWIk|c7^9u?~%RZM^RDSvTt)a20xux|- zTYGm;Z(skw;Lz~M^vvws{K6uUv_zAO0l~N;7Wla$*uTiN3&^!$<3`4fOf2(Ixxe6YSyt zBguXX_D8OE#CApoFnNr-5GcfAcb!(lj)x)lnFV52KDYd{w(V#d34rEBE{;hSu(rP@ z$KY~LEh@?@n$}diXrk6V;2t~k@Akj-@!LTDZ4dv4)7j%7#dJp8FhyOeNni5> z6_21}e1)|S#`p>#z=z?**ZzdDdr5LFP@!O}&rXY$PrQcWc0($+V{z#= zLkorTMKeDsbz*7t~~0$W?t%)PjFBEF*^aES3>aE5-8f7VG~>kVe#Rs=w)g zXS}~TS=ac#3V>Zv&0jpMK7)VInEW@u|9|z+x^4pD^^LXZUjOXh0Q9ARg>vU=m$tM+ zUGvKbF2=Un?Bkj`%ZT+4QR^H3hT!_|_0bR12q3!XI;Az%Y3$F#|LYKFsOt}>tBt=o z{r>_)4$#<)y~v9mkdt|D4adM{eb_ z!1BE0bq<$yGarRFSfjHX@~Q_Z8(W&K6+4egHpj8+%BpGSHrTLKcf7rHe?zjl=@6nX z9H-iHYyK6Np%V0b+#!(pB;HZR_s8eDn74z|=0UbeRIk3!svSG?pFXpy%<_Dg_h~x= z)4>@52Fn;pDoQEE4tZFQ-^I-Sz z;n_zOo_a>TiMK5ooeJ2#{-|=xPK@xzx!tm$wXVe-Xd(OjJeW&;e+g zjOW-j6%oHbIWKeW3jo2_e>4?0_+A2N;?@zpvY}p)9ih4Ib49xT+~~O-F}{L-Jzft& z`>r2D;ax^N&x>6~7;!Em&Sy$zDiW2K5r=G);e1ipOPBTO0Z(gq5H*@2Yvn7D5&KW+ znn`edD#uZ_=%Y&BJ7)W%;LXCXe;Mr!#YTqgIhBCEXj^v{8>k>sr=4YXBgHe0Q8Bp7 zKYT{As4?gBd;Y9;@uD5Kv~m?*O>2BlwtT0yj5yFUtiV<$ zrG!dXj=i|wxy<0j?(OFg9j=@oQIW#2Mep*OM8Z2KIKuag_}4XCdoZcx1e_ipi}AnV zF~lh&B^Ew(26v{&+d$%%skT}cPmX@ZWMc5fZ#r)%zvSf}7Aw`iZ#zeJn()8eby?_H z33vTK3HesIyNG>6=XLIZTm<5p!-_gVBpk=HNrXNlE*S|6cZt3;ovj+JrXLgvHLX;X zYCNwoGSsHqIxBSw9KS7=F!EgWdr0Img8v@MVfSTOJ7dA@_or@*eJ*j`=bYfezZJp5~{F`y+nHGyGS4^$edY5oO=y@Sw{FHr@Wo* z;RjO7h(O?LzlS09sRha;OgC)tP}5{_8L`E}2Q|cIOF~Y^!aHksKy^pEW>CG#Ct1h8 zp}s96SZ8H+|Mw`*Jcn*;!umug?4OtV#mk7&eciQ78#`cSisdrmv3MDNZtHsrj#yvy zhm9MO>_A!;)x3Ge-S4B zMc8RYgPUWD2jQJv>orrdUNgVeG?PSU@Ko74&HPUp#M(10j3fIYA5tWXqFXKH&>u>; zznJCzME$U0S&uYLX2Bf%rW7Z~GBpA9tymeDk2STp3U{LZOSK>u`d6s3D!Ntvw^Uc{ z`H?Sr5h5xStyiCqv)k`^)BiKWhkfagvg>AeJ>+p@$4e*}Mh-YDa!f2Erq=hNx_3-1 zt{ITgAKUU3{1Nme0G7y~9$-Fa%eq!g|ITldF15^bTJ2({ zuO6Vzdp?9A4%$ME>0VpHkLyjw*Y|)WJjn{x3DK6YV%gey+Zt2cv93dza;WnFSVp?E zW#s3xre-!pxA*INz~mOu6qHJLBbfYSjYuzmr~YaP|&zKa?}1qzL$Bouw^{vPv3A71q) z!$2PpO;b!jIv~TJ5$h(a5WOf0x=??R9qL|2tZ?uLn|Hkl2m!bF_<)+hD!L0YzsF>t zetSqMUx0317gYK`N4Ji;=0`NAO*wo3 z2`1Kvu5HG(>1T~8E`xWmvH~}HoUT>m)Dve2esITAnSy0nn|`FCo6le4QQf=hP(z$F zHQxVimt{kAn#$*XllgxG_E{j)7D%uT8|!0-?a-GMHW}8e{E4PX2hc&t12A3rD|62m z1+5ChiCp*v{KSJE9sNA8vi@`+r=D6O=eE%l1D3*V;5a`{3ki2X-MiPs!Vxdzur6)- zw6dh$1K;(Jb1-ANn=R!qnw7fYAKI=(qx)!$ZU(Z$0r&+bUrm!%&=E=N__D$fe-u#f zNxFI&s4J-AImV%>r$eZ@?W>dH%1UKC2$0tOLDT$SqXD*GNZS_}%PljlfR;4mfzEHP zE+n=(XlNg8YJfSch@YdM3ze?pf}e7Djb8m7xP!JK$e?`}@Wcyq$feCIOKq1d6a3)$ ziW#h6Y0}RDN?eBqXmL$~bC69v*yj(3D7pbl3+JQ>1HubmOAJjy2_vXOho&6A)_(~P zGjQ-O^=&jlN_jG<*x`1w&$VV)?qeg2924UD>f|rIl><4W*Zb8Kf4!d5RCGihUkhc; zMhCepDBJZ#N(yavy}@D@>hyU_imD8$-DB5GT`}Ue=WdOdQp=2W?;ha3H4;m$$S5hi zgD*Qh{jJ$kUXfj2vtkX?!688`lf2rwP3)TIxTR7vts6QpUBeBAhHVC_(b@8%V4E3eY}>^Z|0pmh37a)vu^+%K!m6DpSnY-;XDWO%5gZ&N>7 zQ!!%w>d-ccA3NIm9t5CN5g#;b;rJBQwZjip5d7NT0YUojq~L%0Js~n@1UpJ+do70E z@GnULX#W+}GpHwOTha*bT_3l<)KomUR;{J|-VLz)PC_oKU8Ipvp+ArxR%D7tNUc$M zq`3LP({-5&14t`2)7APi=jHD-0iZwRW!y+;K{X`n{yIxFr3wvGRDAK3V29W*LF*UTh8Whl)rC7L51TlOmYB_ecwoJMwT_M$(Ro$p=8+EYcUo9ghrPMnZAI7I9E% z+iUoN*fOH47iM4I;231ryoqdc*65>PNRffdh;8P!OB*Xexd}(q{Hvw^MpN|*>fIFn zE0HiOi-h-l2B88DB;m-Z*!7Lf(=mRBmJw#xQ9;wtfGdh3@{};dhNQx%ZjcsjyYU-6 z4dEwh=wn!;d>Iio1Y(=Kqrk0Pf)QWW?h2gl=n^ijsb-=&$B%wBkYd+Zsj;5#0!OdODU+ zcu!i_9P^57nP=kT%KqSlx~;1J!ueNoHcOF{bu5j4##5A1djUw1dWyOMq^5*QiXa$C z<0g*Y60Y$x$56Isbq0&Zx5KJL$#tBn7pk@*Usx!SX$KcBx`xAdKiW1SJT)wjF)>66)RC729Bf_dM&G+onNYf=Ywn zJ?@F{by}rfgQ$;h@HMJhg9EWK(O?E_Re<&iMgwpnQYW_3nq5L)zV#RAx=sx7Z_qo# zYLW8P;8j!Vf=a7H;RjFYG-*aGg<0CnPo<)|jaW7WxG=5h2wLBzOgZ)y8rq9roCDK# zMM$m zN1A_8Z(zPSWUYthuBa5ys zphBRDKv62AHVorwYy83*C;4cBh3ttnuMyEoIpzWlajmG1pK=Tf4RNpPaZYPJU@hgC z!S6ogR2GfnyL3gC1-#60GgxFDt6I@y7rf^fVQvRqZm=AWtbNb7Q&a2E$cM~Vyvsd+ zgf<&=B*3qg0l~hmC~~K?R>MN$gy>k6C~Y;?bijMK*RTML+;|#xB2=3HKutLhe+RBA z1w{{n9%nt zVEq(>A5f_a028HBmJy>@WbOe^AG!%>3Qx7_9acQhnRo}TDT73NYRX`?WYl0eG&uIHtdkBjlX-_xuXD=?k+_Q?2 z0JDXy#|Sk48>8hqj4inRU!eX~%0P!MzXuf@Qd6djLIjj+ve$P3I4o-L>k7LCLTK6I z%iUDGH;kzoW%7?h&GfBsq)a*b4yt{zqM5r7LvXg*`H)37A#_$clc)}+9}motrVPwq zt$snQn`=Zl74HKOju5&fz#N3aFf{^X+RU{JUCe*~MP zsU=rL6wvihV5JbknxIlMehMj*3hHMqu`nAy;6ZX3Q36uP=I;%@K#razZzKs4r;!4d zJXj4)65w3q)GNZ;s0@<_Ei^$^5rPT1wv3n^AS`x)sSASHnAV}Js;AqGW*I2H4>*85 zJM=f2Sa>(!;Wu#rNs}d(Cd)9P`#40J#gJ+2gT=Mgei>041GETHtp=1;>Dr#>M$Nz$ z7eS}>pH2)>4Fh{{!en4=`jTMn>r99y)zL7bx~3SjmtOtBCS>at zzRL_Fr)3qL4q#IiZ;ZfvISrEg_wVC~i=5GYs*CdB-zS+%XKaf!Gg`h*kMr4^%5KBT z-7?N?1Ot56Mue#KGr&_vbQZeXbI%IAmlCgWPkmUB6g1R;6JEw~1&2;B+ zJ2tuCp`&cG5ahyrbf;?dn@1mHo-L%yB2}AiH1#*-8ouh!KSA1(l~BiahB9l|bcMmF z_A`>SD_3KY6V%c<8ueKHk7YBRz=fVs6OXOwOz#HTVt@x;IVp zQ#bShivzb9fQfN_#?NgFg2|Zav=J*e)@(?}{Kl14QC z&ML_zVGvY3Mt+9ownH>302@UaV5_8aie+j^f~3jywan-tR5#O~;eU;}{~H&`s2}rw z2Fc|ruJbaAe{iv`@c)rrZy7;3BzG8W0>Fg3tZ7AEwjR-?%>?QKp;tvNSAo>p`HUIe z36KeYM$NI%k%huLGk4PVbr7b;1353P!tGMDg-SVlK!Qvku1q;ZO9i69%IQ*rB_6HN z{KSK8&X?KOvmGP}pw3Q5fFUY4g4NH2qR#!HDkOXl-uVkuT56XO9-{PGf)fFVFbLEU z3Q8|4>Ub>On0m|kEand#!j#Ww*BmQE9)>D{kSkC=YKG9%6YI6dpmgEbwZufAkvr#M zj+@Oa-up5aOxczIu>Q^gp8zP(Ohe5W;B{N$UjhPiULKzgssOpQq$xUuPztfQ)rI5h@i+>xE_T@Dj3D1-cqb`97(nxFMI&y{`pq2!zRK+B8an zQMEx(mjVTAB$uT|3hgx0YDuKWgQK$*Fgi@U%ed@TYgg!n>&2u*ZHHOR7V%xuOFY zm)EMo%49|zfJ=!|NvxCLxdeSr^)#hmjhXTi0ng$?>Bun=HQ2*&7TJ4|>XQuB(z-1R zS#2}`yH{dzP+5{l(t>T37Ib3-H9OYzD!CAuOB*VvsZYqtWu^6LQgI+EmK!Je9`_o* zPWck6XbgV=dkNl&QjTmp)ipZ-i2;t%&c;M=*pr18VP_;56AC6(hIK%o5SioSpu;0* zRo7@M2FJlSaATtcQV(s2(e#8ad#(jpZWMOLS7|P6%Ov&Dp1V&^=z{^!+n*2MBrqWB zCsZp=YAEECaeK~;emxch^e(R4X0eQjH1~@dF#dwI(#XDqx?jKG0_6kQPx2;e7d=IN zQ0qr6TIL&XkB^_}Xz))TK4+wgK4BThpen6>S71kMF+%RarCGKXoHS}u(_DDBmakte zj@;xCbBN+#;1wn!kmT9Ox>(t-a{M1a>CsGVh_RRX)tD%<8G$l~4Ce2!sifo}gSlUf z?J)}%B6Oq^29qfM{2k*E?YG=Tr)?*jIrPa&t*{&GmwIYEb{Qe-(}#6Cvv4C@#21|) zDJ-^(5b^1=gqwirMV8G-P+yY83_=C1LKo6QDT~O!STg!M?1uYNKb8UslOI+U30ggA zPE)45r*)ZB!k$chmJ5=oU}|Z$s8%!UxF0KthzyK}f4l{=^MJ(VS&=wH$X6UeM^m$} zFC$QlGZN6PVX%)+p?H*X#@-Fn&x3oEwq794xO$w?GS>5dgI?i3{~!Vvl4;X78I%|bp|R8g?F=v!k(cp3B<~8QvvJ18p3sQ!LY5HLuq4|5$?2%gbvoa4lKvV z{xZl~7@$gO7xgZ(rf3n~%>z);VT|9Gd0ehpMralrVjn+y`|uo7fVI0MS76F(KXSma zQafiXh*47`aiCb2;r?p1V8u}5{Ij?0^D?5*49*?s7_BdIvg-!Hk8#ja>nqz#ULFc2 zaNa=1)|-?aHPv@7RQk2z^}Bhkn~!$z9-TrlkBrUXILhzs|4Dqf?>f_F$M8khQXIsR*E!U~f6*k20 z9?_rDKn9+expw!50Ukj~1nwLCfz$tpR)+64SMPo7VEW?ng8&YsVZ^@ZuQ#7W*>*VJo$H!U_ttVJE) zrX}avogUcP-#!=YlypP<{V@iVx3!J~M_5|!VB76Idu?t$`+S1`$cDB)Zl`jDz2Wg3 z!(UR)^KGJ@&C(a9RXQGRmU;JV8aHZO?NZ~1<@0$W(sXrLOtp_K4{Pu+a%Asw7FS7x zGJE}{{oevH)*Lp8r+Jf-g}obPPI(?75Z)Dsd>fxk;4^#~c+c*+{e!uTAk`p4@gmnW zy$2kg1|r|xE4w5ZQvbX9W&m}bCRBaAD1(imX zIKD&iaT51g4h^efSyeYnJ#T2>dhy)*#mfdQuKg-lEX&a=?X0Q-vVttz<0Nvt1w@JOsTv9NW~W5}go^tq8YLRlp`hs7v@Ms7o~ZOEieE ztN_QIn2(diYM>Qb0#&dHr&L|QM={#_m1|@5iTS9ULv&Qz6!sUNXzVCH(IB#2!uw?d z7xM+1OA-qST-y}XPXNF=^j>UYbRrja2pdzsFM-P?fh(r2v96ItW=^6^$5(+xyjZ63 zY;&}~M8&8E_@EDab3JSFj+KmINC*gNO^UTne}Y(mbt6}t6X{UzG% z5}IT7dOvRnSI~QULrbecmsg6es4v%n(sax}gIgJ4SpwsVJjr2MIhu;%Qhal*ej zRCGQ(lD8k27PwbrbTj7Y>AD5nHa?6nnGIl2 zG?)b|6Fw)--!LM4&Ql+NJou*&fHC~=2CDo_SZs@%R(ngxx>?diV|%{0;kAkC@d;?K z2SXlfgq!WaF5C|yuWF5!&}(od;F~E`OFzUfH2?*PQWvnL0}yGFpSobV)G#JG779ul z4~FNwi>i!(LS(^aP*SaOoF8o04N#X-XiMmI$P6_=VO5dviFf&u{~C4!rs~0U=l3_c zcR(mp*!JU`kb=>cNL$h|wGXHkF9MLW8li zZd}^&PxDNaO00@CZWbg0kvJN5&13l4dN7SCvV`X)Ac=Z4YAGumvimD_fiisI1>`}H zHtG0+(e|hbdER<<0JYdLo46H89w*h0yFVJKEhqtA?emVVPE>?|6Ff~BT6i!Ii5oSl zn{T$AMAp37SFACp_D0sfFnn{kRb=l7_9I?f`9yKpwMO*|Tjvus?b^$d6CzXvHt?!T znld)LLCmpVJu{}A5MIYDyXkl{JVZ(AW3}!x=fbAn@uSw0v>sz7aNls|DBHdp;n?)? zr_^Rj@&zoloRYL)a(&+G#~{t?6>5l}@(IDG6%WYuz&dWo6H5)oJB%9l$2)rJKtD0x z-W?M~lvmhl5)EJq}hl>a=d(OTLB6TR$|ixUvSaQf6?VKcs2ts@mY3>nB#Z9GL;as3V zgIzTCNdvE}hwx)$QPLbp=m@41K~(k_VX&qPrm{|@sG#1*Qt&tmWz&i#Ls7sisIJg?jFJ9P|JV)r(=x|}9=B{W* zyn$=!+Hl?(;FEvhqDew09Gmp&$OKKMsym?$vcFw4vY zDEk%=vV$h1=K2x1w2V@?PGbyzvO&L5S)G<{tO$5v3757BfO=O@ zY4J#bxk)FiOpT$j*L?X8(kp<8U;U4&R2(|MOaQ|NGr{g*In8txRz6m}D<% zdxzX{&1J+Txc1Q?#+1-$BdO)x=T(C6znu;{@orHjgp!d*s zEgfquy;};Q=`z}!=`uQEVm*<94}rvG-_en9c8oyd&0Kg zNo2N6WWJe2=Swi=>d2jhMz525yDziX(*`9g@(CoU@mywazmvijz=Opyq0u)3*t^x) zyDtSQ(lBaY$ikMzOTCfS-Cdr@Y;#)s2-;C9LgyfR%yTW@8>Y$1s;Zg|;d}w?R|FCc zVX=bPCup=P`UblLxK=A@xig1Zfc@*9ms(oB*VspFq#uBKGVN+}6q8d-K+~0jG5`5> z3x>pW_5(RrPSCn(X~pxc@-dIjeU6Kdi$h=q)%h5A&Vw9aR9(XN9F|2ePQshdXyrOX zb2a@sLo)3;10(%9!xWCV@06M$bnj z*C0+g_}6EG}9tX&Q}{h~@^&$Lx78 za_a@@q;b*#wA>x}K^YI>XuTcLFZ*^qw_Lpf4~okBLQ6xvu4exHz9Z`PzRD6|X9te& z&e?|9jeZ@*R`qRRQJeOtJi33+V?v%M>)bxBLv@~QW(&4Et+h(}+t{<+)7&;l&1K4G ziq+2>UX*-&1hX4mRV(D>e=AUJwvMMrKM~w#x7Q}VHT{;I{lYCTQbp@f*Yz^7g+AS4 zq+$5djN$%{Q#nQcGS!K($L#`%)93cI%PKkc?ks$ zzBx)Bo&~3^0Gy+Z>w2>3Rf@)Jb^vO}oLJp2I+C&}vLep}VG@*_m`B@VW!25A_nc*; zjP>E}=Und`gkjdHq>hhE=Z0>h=rAMg(XQVvwVbX}p3M%i-8^h{S{$G^r3f9J3JH(; z2L!Di%ZT~Z1W&xbeo%DU&9?gOx)uvzHTdn=(YorD-W1L$>WjIPU>t^OHH6I72}Aj%b>U|d?7IP+HK0nha{6= zPmN-ocsQp~I;!Jm>XDkTiqk2R&`FE^&I`#$C+Aa6&RY$~?=7w`RB;<=6>@GP=~x(k zK+e{}q?C7#DAkN#!X@m2#XbD&7`cPRJTjT&f!Cn8!vj*DNX#M-#3Y;Rm0RnjJ55k9 zcQ8+$B-DMh2}EchBu6dH`IMu4R<+VED>zkwq&e7>EZNj9C(=|Y&qM{F ztqI7zBafmU!9;L))WSQL^g23aDW0`onR~R!SiP~==Bg8I^S7X2eJWi)j@y8Up_bSx z*St-0UKRN*z~Ba`e2Gho_m1D(5z@6d2Cjgfpqy{1l($|1l=vldbOh_6=Q9(#^+qZr zE^R-(-2xhC`nZ&X$F<+!RC&YJ@=k4S!5rTMmwcf zmP@(F{tGg?wgVP+pnhcWIWlCW4>^@rq&92@c+?=aZc{|+#Zrpe*K=or^Uwzbll)7!3a?Fwoyq!Si~Sjq>Ih{5<*Q-T@)_Dc1Mdr(*hKh2V&- zTtZiud@CIODdc%udd2M%PmKGymRHaGY{?5yMOtNV`t_{9RioR-si}WsNG`Wr*U%J` zd>!n;Ja?U2^={l10b+CkD^IKJ8CQSuLvzfT4i(F>j9Ah1pdb43Ze&4o2Vj#WepYWKCDj!P|#8qlL#}T_ih7ceoSC0}1Np z!&!Jyn3QI@#Ffoy8!xHlQFE~_q2y~uc>JddCr{mzBW#f$AfYN;t^w|SOYXD&pZyU6 z$GsnRs=AI-Qm)${k|hYm1YdJZ*9of?P|m_y*bi$Rd#0#=x%uQ=y0-aAm6u|Y2a~*1 zg)iK!WKd<-+5tj_r2Fjxz6V+xStnp!HY;u1gs1NQW~Xrj)w>fmTy|rZ+KayoXTzkV*XR%X zmJ#|u>Sq0<{lzy&GzuEIB*=U4m}2cpnlhSLdkp6*fUg00NTyEQh)`U_CUcM@H*S~Y zU(k$#g_}a7eCW=i*+s@I<@lY0E9wm8pHI2FxYR_pim1uIHiUZXA0}g)nki39F=~m; z`qOYu)HOq>D;t*d6P-)hJ;xCzR6Q>p{h$o>HsArwYn@t$o(KL)nPSBRtS;lus;v#& zq&e5bQRAi#w{Q=IZImY?uX~^;3^E2QwzQ8J-*c$F7N#PnGvt@TAGh8z<*ci4=01s* z=LsVw_lT;Y$TKt_B1C-Vw7UHKB=_rmdvlgs@Dsd)JpvTIH0+K~j$1R2UhPJC__292X2_1# z9}9j_@oWF&F;&)9cFEw_%ly_?FO|i4i{*RDTDj6+)dRN<^q2j-phi_#xdK{eM#n*J&IptrlooA+Q_ z=$TvF@0DZT)|Y%wF|raj`YJP%8S(+QC9I~NIE&7?!mv{EPfhyO@#BY@y`xY$Gs2)F z3cPccl+gF$V4SQqyK`AtWkNb|s~%e5l-M9DGCaP3VvRR_KhjJHq*k>`lo{Zsf`)t& z;+Y1ODM!%q!_ZcW6?g1UNI#u|n=l(ls@2xrQIK#-?FX(`9d3Cmdpwphdg}C|h-OmB zck)vZf;}6NvWPGBbg207{`QOHw0ORsEhk?5+CoKfW%4U%<ON(je!nM_1qdI0I%?)bZ z5!h^hDf`-ZouJb2=^BX-IPyT@uwsVK`OliPixaPdl9zOyhK;!+S*N$kU{9eItIv0` zOz%R@Y$*Vq&blt)^ECa$eUKJ=o$uTRE_N_XK$GFmh@#+92NS5!mRP5pG_Fz29Y*Dp z%tnHECF9ATJ}apy%v015t0VBWEpMhLT$+XC$Lebcs!t2KhPpm^9;u`}0MTx^R}n|g zB_uHj1i=#2xU+BUUwB-$OWBhVSj5?*vBU@-OKx+fQ>3wFx$t}`tg@#O>T&8D11*4>!0IadZ&u4EswD97N7; zrD3x&wt}ikW>2Zo4og1*0VE4M*;*@&kN47ckkZ1T8ixPX>&PEoS zQ{IwUt?KxhB*U`skE4mzB+YT)fliXk=dHTFzymky(aL2@Z%416>Q|ROkbvs{hWUn< z!CYBNhx#CU@uAd8r6y&=GmwzRn7;z-afFhR;WZE5qX0v-*t_+X5%0iWGT5T=?IreJ z8EhO!N%s4pi!q~Cr3v;V!qvsd84$b$pB>{gg|DLe&$$U^zJq%tll5~?8~ohyHdn^* zl5|BAd-U;-#o+pA%swU^iK(=+x_g57K$xX-=>7<{=W3-Mdv#2`*;$>Kl=LdzG(Qc_ zklwUZ`TBj81$I`tDd*uP#|Xh*t!(7X6h7jpV`J*3g+LcK?Ml@cNwr(4Be} zs%?TI6l9rohedNr<}8^*_Yb>NlQ>mLN^T;vmpz-L`$tu6UNr5e%iS1yzkb872HkH`hrRQAYnvqhXo)?_I5 zcOs2)3-OX3Q1x9%>2zR-uHB+6B& zOF5j6=cr-jBHF@__H-@QCoB8ApIS!9)A9K39y1@V9t*3ZMNI|My%?5WSb+wWh~siz z9gz6_c#oTPkBX4A`clS2xfudST(pwLc_cN%1*B0>-b7?rD zngm;;&U*^TO&-TyxLy&@K(R6kwWFeGFyZF0R?#38+wshl$2{-!l#yg^rTSGah6x!FFgnNI9Y8{%nS}bZ}PBkpA0JXl-gEY zx9*Xtnw*cO>Th6G{i*oVS@Q+K3;?F8BaP4EQv?np5K+eIOX$Xvwhr8;AYvJH899o z=6qlyST~wuoZ;cZjS=!c$qIVsbpUV3J$w%te4%r1OQo?C^bJS9NhF%u0xK_y?`_TX ztU=C*S!3ph@Jl={YR>jo_toH0{U&8(+2PPUxq~fSBW*o;x#za+Nc0fYd8BtK)F+jr zS|F!*-y93Oroq0ob6IZD@=Vfh4SJmQY1U+N+7`Y5LOl%71$ydMOCTDl8>ITf_VwA? z$BD2o$ZRvQsCc(>XVFpx5MY8`sWk@_uSOCKBlYf1J_WZheT6gx*8bpte4Q@O4F
Wonbb%M`kb`8~xR!_%`+qhiBW|vvt||qSLZpL+6?uNk7uMbECD0 z+`PjFOm4aKiCyA6(^(UV#ZF5{Lv7=+_2nB)G|jYnsc8sHM%BOHy8qAr@Ixdc$lwh8 zmYOX;cnDQtTJlHmF^A_oM&>XLNtK%f6=Lk9mZj?2jICEd^Dn5@$y zdtxW`dEwy8a`Sgl!#V*%UP)x#m*cvz<9gb;)@%Vczcw`or_g?+->_GSkY)~c^?geJ z*~u^SR9W@8j%mD#@P!+dzIR4JX|*MJXC=69@%hlOimO)SOP_eg3pSeJfhpowzCYcs zquv>B_4GrN_|@kqOlg#B#1hQG*mCmJ1LKl$P_;^BS+pvOV7bnlvdKUi=Fn5k<;Rq! zd^N~^w@FZ%IaSm(mtS7jZ1RfJxkORzxfC`>4o{QBVV$ER!Ya3&&c%XGWU*mTR0^)M zm|3aZVw9-EqD=sFtR@r8t~>*`BN|y8Vrx>wQjR90W2|yq=2gO@t-@0FB;2V2Oew{K z6%sWNxJ^gB0dQHU^JeCd=RBKEW@n6*(k)XRrFg3+bqag2Ilbyp&trTAcG9kx+!+Pc z=9V`C8x+U2nEl;!9zLuWYVy-eSIIeCf6^&gFXnQyQ1ku^VOG~$qS3D$6C5rCpVdj# zL z^%@nO)d3j`Ysh)=_ZAytLDBtanlcQZzZqy>{WM5 z$aBZ|lp}&HB8Q?jN4{*lEGw80B)CiRL}U9at9>ztth|=E!csVgtOfU9$e@lek(vu!CDo8N%T2ndq~f(w@uc=(eDEiGuo_FL>`+_m?-Z zJ^%~%)BczTc%Ly-hj;}{0`2`m#>Yeu-W`?)&fyifV(xs@y0hds2(SX6(E-$`2~&7` zWM#eC-CWkNP6{t-+6dIYFH(g&M?aSALJ~7L;o>+rcruscl+0LtC*ccl9*=6)-ca1> zz=7+D%MB_EwRW?UEG}Slcut-EQgK>#+K@3Sjg|A_#&DKj*~aqQqjwus{a|K%#x@>! zj$1Y_GCa8XB}=mNt4)qGk>{fhjd}LFdg3ixq{Iwmt6D|&ehR;_?Y-=@ky2n!`uSeX zoV?ct4M$lIrykQ{sh-4j3+OS|U~vxX^}ZNbr0 zu3CIRU}fPKw;OY+ZLNr~JTuM=4n(gMHss8c^Q6iG>X~~}YSX+Ow|e+2MyexSi#g{q zB9|I|N!dJ-64g6H@vM?AbIMk4qD#GEP5BUu1vrYKUTKjzuHt(<# z7|af@d978lqg9Y>c>Q22^3XvCIQ>@R- zj~gA3RI9J0EZR<|z*R5E$~bKa+zei$vOO))ie%s}pEd}eZ|x%P?CRAQ$)#Qku1R$R zmrptQ2TSapm zG);cBS$gb@*rZyl=xqKv)k;KYGogSNJ1PwMJGrWlH>RDlz~ldZNncdSw6V-*Dv=i%w=TJNn2XghFEJ>X-jlb@RKJ89 zss$09|4sECV+Rn2vqgYCZv%#6t?cDdD|ENdNv;4m)7CE{Jsxg$%ZR5fAa;8o3QG)8 z)G|W$wK%d;wnEcM&Pq!!vbH+N<$4ml&dtxV5S}I_W`K|Y7?ppip@^%RB zkS)BrPgSP%YysV28=vuu`l;qlC1r%Pbnam(JR-BGY4qIZj2F73d^EY@Vx6vmTWb=Z zSM;d$R&a>UjMRA74%LHN8{tOYQtX5ocZ(o}Kk$1KRp{|AMc3wLuU9o9*U5yR9MxTD z!wvZl4}F4n8GJiXHFtva(qcj5EZy1X&0kQ|X^(QRjLJY;T*C~=YPe36W?JGjn{r&9 zVlun0NM6Yg{B+uBc=T#3C))-w;Q&;Tb(23oM?r6oxaFsgm%}V4S~>WB)lw+nVva}n zGvl9wupA!orx*^_z) zaq&wnN4n`obpe*$$k4(TN1+s}K}T-3&%%kOVPId$)X!>63*W1T28Lz7cRnNIyJEvD zxpAqwxl!6Oy>qNaq~m4TuF!8GGjjL9=A+-72lLeL&j`HXJb9bYt*`9ua5xq|(hk9l z{6zf=b{EreLU!3%)8xgZ@RH*Esg}naS+Q54e@;SPX2#C<6gSz$UB!{_VWyHI#gwDY zKf-JGD<%1RuqIm;YNd=f+AvkQo+or0=_Mu_91OCFV;mgcn>O6rt7TM_7_Y_~q39^N zct8i6`|~H-5b@*TA#izvBTTaBu)-5?&xT`&&(Yoo@amTxTx0HA7qQcnnMp2Kv102| z-5ph_{fcLps)VRHmQMZhgAETVP6_PD={9EHGZk{?DJ$|<`e_l7fS;c@H71x3_zrp^ zMT0HOgt_l7ByV!P9h_)zwydy$@y(!^@3=!Tr>diRyG=BSTuSIR@wnuE{kVyCUdBaf z)d}kfnXOienMoqb4$4){&GOi;Ch@@Z3R6_t z>j&?hPWR@zvX8i_{Oruz)!p2YKdmKl?dSJ{ygSMQD&F-M=I{MH1JcXnH>h^8#D&XE zIuG+1UF`8hVkk_OlsI*de_wuR=3=#izLfX9EmWVyh!L>Ew{qIronaEc*h;R*v*~{x zQsqx-(TeJCnoBaDN=_7PS@aS(P@bX z6wulW9hSE$sY0c59R27BYM+&p@g>5U;X@pcj%OxscE|i;C0lj%8ozQso01V$Y_u_^ zpaAEJ4wDWi<@Fjihwg2fWJ1mY@A!@%xWXvl3Xy6)zyXad`59~M&LQ#OQd>dXGTlLZ zZNlVlePtBfr38-8ug3$rNAaFV(i~sHTYtyXzZ{m3H`7&wd%wu2XYRC5x8i{Y`78x$x3e3+bv@0 zv=uEa-Vkmju=VWSi#=&OrCLU{AEciApv}U0#LL}PPqOcwPzH~WtiRcrnF#je*>IaH z1O+3hWEY3PypE^JwQbdQ;Q^^hX)nf03ijS^cu#E7@@zLsRcJ{J{-!xFAR9<-VqHuv zy)7AWWphRBy++s#FsWWqkjezK2o?fOhEW%HeEaE9&+fJnkIhupmXxn`tu?2>HpEbo z>;_(2cu#DWYMpC`=Z}s;4la{mXVq9R!naFf2eM%8wmve!U+Z1akFO@DVmL0#0#Ecui zc&1_1j=j_+;Y_#r>Keb0#unn^ow~N6`UD~HH{8Ja&G`(Z3Ux4;Fz|gg(z=ndi5=_V zq{(5wb4aW+D|Vvg$^o1IMc11LLb>+;<2s#=Qz~szwmBs!k`^M{R1PXJCt_rosZ?S@ zLUwnhB4U!HggHX8%uG&c?2}!|&P>?`W1WdH!%F|* zulI79Ui3n+11mxCaG%9*|BCcRY5-1bdGrt9Ta_;8c!c0-BE(|6n11_e#okD5O4{e% zXfJ5~-c&OHn3!(pw16c7NzoNQ%ezuNysP42=Ufups)G9guvuA7dGyT=>dESfxlk%81ZTyNC-6Fn@aE?&QU~*5r@IQ%Sg@HQ5Prhr$W2zb9(8H!7s0+@al8|DW~p zFO3GJxFa00^q2RqC4|JQLm)+Uo4(E?No<-y%K0G&5~wPphW-^uKz^{0ybdmDv$G^e zw%~q_8iz0X_&rb=2>{q|aRDrGCUy)%f+8KLrf;zq-V*jAt>LmewC>6&Fu=kMKU!R1 z0vjBjEc7!B?;G$d;1}A`NCUwr;?I_W&I3ixUl83T#ER1EwKAQo=JGV@$-!Bn1a|*GT2UXxYr$ejH41hZ9Ir?F!FZw|R?*IR}EI^)kcW{)Ctsrnz#`7iJQFg%< z5D_;Jsm92) zlC_jkQ+1^5cSP(0si`>!R%!;!IqNqmGV|6EbHqG36W?TQ>Y4L_6`dLtJzcmJ1cGGp zkkU-gS5f@S)Hw^bvmL`p>|EMBKp|_1e@e$8EZnI`Bm{B+k}Z?in;&vPbOiXj8Nm5a z^u;&9h`DqoA0b>sa@H(paS3*JzNz_T!xgSg<*n$p#5N6P{B$wrlDLyEKyv*OZAZb3-b0gxc`P;^_x zRRMOL{iLD|F{dpHBH4pBD)jjrSp$g5Q<;;hnr!-fJJVi5gy%_F14U7~&wFavP?c`} z2NvTr^Uuh3qcH>YpJGR?>FB5YX^W?y!V!5P=J>sSS3LNL(-Ve>3WB4D2jlD_$?>eD zfNwZOA6$BCnz5TGKuIRcKh-S2hFvTMke?b{#!C*8iOE_yYX043{!3+6&WuZ%`qmF~ zpzz_6dywnmH}2Q!FLbuB2hDLLkePPaYKO4ANG8K+wS73*S= zN)GI*Nlgxb-$v<9J2$soMW`O~GeHm?{HuIsn=Sc;c7nuHqa8O-R*-4$XF+MlRja<} z(tO{}sB2c;gI{B0UZbwh7-72m75RkwK5kLtd5)uAJy!@D`>s_37$042^O4Gxg#d5c zyg=UH6>bU3!tr+?rkDel9*gv3evE}{a^TX1Vce^!+Qu-D;$7|KgYG zq}Do>`^51;>O=m~zV&DP9C1-MNZ$JIEh~HJVOCdfswF)P!WZLI z7MgWj-Q!;2gM7=r53U|~Bq$~;U8`UpFYwve;cey^c4^nOj87_lj(70Uz<&O@Cfs3g zf12O^`0aPs4Xjwne?>Zb?4RCyht3s$7y1SoSsc9EKY8i;9Okq9!PcOHRBQFX@4Bw* zFbbKc^mdrm4=AJtbdqM@ugG?R_x4V7Ne#o^oQUI{`15{r?Bq%@tuYX}0UEdNLk?nP zzy8rS-{nC3wfZIYMbr6E$sg5w$T+{sw2Ys*=Wtq>{;ze48S59>Z&wZtCWQo0q`Ft>!Xk_X;eL?A zy^szAnawo4lcC1 z_9Znu6R3Vf+T z>3sWO^SgT=M_G{p<&*D1F*NN9yPu9QGn{U9k5N9M11Yo$H!2?w^Dlb@g$uk*o4!vr zk4!s6k6*v~{()j;48avI2_M^^Y#kAp?v|Bs{t#gQKZGWgo<c88aUl--P4cdOrMFouO|X*yNs{E*YJGB(lT$~yZt#ep<^ z-kc~j4^rqKlEa(*xjwXUHvuL>4B|AXsvULdx%;e`>(iI1J!h&0;LBQrO5ztf#WUbj z6Jx%ZjlY*}ttvACX+(9fMT@C`ZW;{onR9%qd2Kq-%r`vU9&PqSFnZ)kUP)A#;O~8$ z^pD=|oosR^Ea)b3d0&^@U@|;=(*!SW5Tn)IJuWXMGw#(p6v~8yJRIndav3mq>E~tY zrpA=&>Pb!gH}10C+f34}q**L7IP{|dQYY30*6Dz%%=Yij9{2B?HvL+QJNIGKB_puS zI(YAe(X;kplO&H|tJV@nO#uq&TUjzsu#0Ja@FnwZOXgjC|Iuv zJDsReezDi-&CXBCYz{TrSRDy>N^4YiHk3Q>8uk5$oU)K^iR0qXE2y2<#bA;>Y&iaG z2=K|JzFjq?2kW}rey62+_eD#}S=>n1zIF(_3IZ?6yTa9Vto*9c-s^8{JOmp~yyYWL z4|iJt|A;=aUuQ!VDz|>*2Jil$B8tLY)ySF{OT2lDn_gicdTCuGN6_LkX5k^ zvZ|gN8F4hCck3j`h96uWUh%e4&b;qqvl%F?y+>I6{@GeP{uBs!z+{~e<;4Qjpy9Nm zd%{rX!UVoyw!K5YTzD!s5GMTBj(n^VI_74(V=+NUQ2V=PN#SoaW(o;Xrg0`;(@XlT z-|sA>LXM3bSbe5bC$+Nt=&nGZwS<>-A4(2i3QJmlv@tx1DA)`T%@qcK)rq*fWQQC< zW{2-Y8TLwB3n@@9yH;`NnI$9RoSq~}IYUhAYliYJ4F+TL!dpG3xIx+Wn6bC@#16%i zO!M+KTg3$OZ&9s3Yd;DkaErYwqkO30?PndFy>sNkDhalp6ncn| zCd+U1J#TsGa|dWtT9)`}v7V1xkWb^qJ;T!13EMF_f{8%dY>ggkgs`Y;K>(`d;}(hT z%e(OU^Ot~u5FtI)4eSLS)T6l~hh7o)SnH>1Os-V0CO4~Qz4HQ17o*#r46kB*8X$0q zdrDXD=Xsp-T_ho_f5<&USbF%z+DLV4Q%@Z=_(sxRcRHrQ6Y!E^aS z@7EqHD|pj>rey=LQnxE#s(Z>rG4Iydd(H(Kzn7HG%xrgAcmZ$nUgh?0P5N90Y~N~> zICu)R1l3i$`d9R&1i)o7UDs5W&DGz1?2D>~Iy7p89(hfFCJa!k;SM2M4;pb(@SX7U zTd%sTD_{Bg{{YVDClLMgWL`W`=iyO}VdWJM-jMg~4wGi7e2B}i1FEF+ONs6E*{h~u zlNG92WPWi16@aqpbxs|o>i2U_Ns>#B4OOWX7IdifK%+a#N=c4f{QDy}QK^xoUv z+_^DtS@y%MCojE<`lYUYbw=WK1nd&Esgwxov4q(F+N3Zi%8)CVTY9`@(8S`)pltqgABaEX=vn?1054q0xRGqoS2Kl6sz7scsw2KZ%^J2p18 zr@dvV=)&9i`%HEF0^etgo5o4TV=uI0_L0(BaSuPTlkxLKN07nPVPYATw&q;+cAu`X za}7?pv+|yg2&0(PgI93_Z&9z8?H%F2>|MSDOu0*so^@PCK;M_1+>4@J;06~~0(193 zFlzB6_mX&F?uXnvSg92oY1DJ^+lRU7DRTf*??+f?_ks{8zR$!)NZ-N@F0aw-tsULt zBxDA*lFih;{59{FWz6!Mg-*Y8UWlD*-O+3|R;h=Q1yS!`9E6m+s~_!~FxJ*|_glHn z-ss+@1N}Sv>z)L57;dNSj-6|YwJ0z88$ajIdh8Q@YCp#)`|h0|@5j7Qw=Gmm%Ov3W z3OBNSeUntpra<*J$H8WR?b7~8o2&cu-9F6V_eBXZKD}^dg0B6t0}9&8*vgNjfXzXH z&8bQn$QILvsLtZEYKB_x?c+D6qB)gWMT(7y=iApM{{5QLeee#JQGDL4-h;^jur*}8 z4-&@xAXl3~lz0Zro_i!vu87hy`(=Ai<=yg`1FQwiPaYdQlMXwtPh7`P)J&{-UL0jQ zN0HQ36k&*qWD5e7t6{Gx&=-VnbCe$K@oVk~xp`hLr!EGCH)ZaQ(-paTi?w|vp!4s$wR{99;9kx9aEhqQA!NxjndsZ^-6~8o{pT` zz45iI@gg+(TUkQx*^Gd8pFYYXj_=UvfsZ#10(~hLK}73nD=%YfpKj}eUk)DVK75Tc zTcgDaI8nH7_N06bkaGztLFKQMgnvv}m^=1{b1B7fu{h`6D?*)nfKwqd8g1^w7@H(F zFK-zzPb$5o!o0#klcMvkayGsILpTBI!OhiQJPkj1*Kp!MOn^=2g&R+GblZ*0T{{}n zLRdAIH;@)$C`hL3>K&C@?kgx+y7;c>aU@5DeZ58^(`}e~nw5L$O;0Xk=Uc&iDUmli z^&)T~vaIm1bbt4p!ePu%Vzag?R9Df81q9_S623yW;_10yo3wqmcA8kbYWN1sr<|;Z z@-ZO3@RI6oWdm&R*PSJHI{|ps`{6^)L;iUehan>3bZ~%@=espGgFU7SnEkQ!>W6k+ z(%;|Fk+iHtfy%>c{-T8aG-!V-BM?7N#`Zyd6bZkM#fbJ76qy~_U0mquU|wZ`y`SN~ zM>7sm9H@Vy^_ZCz@JDIIK*nBes$pfoa<~#KppF>M*?6`zjm~tZj63UOo_2}85RbCI zhTSyud;Nxn0CsvFAf0>?WXx0S^2ugp*A#)4XYpU$30cW802y#z0B9ygAZq=UhJ>-_ zdq>OeZfkI6eQeH(z8+$@@CXWx!hvQ`FMrXtV1k&BlC3YFhvq7lV?%KT{aPk7SA600(i8Y}*ZT{7Q!7}RX0uAN=kWwB08$cNZ8!2Vg% z+o`{-kBs~$7I{^qL*4j5Yxtto7H2QxPPFn>E`; zqDI2uEvloJb1>^>gGJz|h$ta-kW_g_k_htyH+^dzeZLSa_ftdUhp!=lXPdd3kN3FP z+z_nD823#uWSSUiYwsz_VRK7oZZckI5OGo`G6GdswobEI3_#_}akHtbho6^tv{b|! zJ6N~VSE`}UnGb2@*tDbXW#;*6G;~Mxw-Gqifnq9%_}6>u8=xKBA3R(-@-?vSBBc=k znln?EDUi|10|+ir+uqk>=g!QduHlD4+vKZg!+I77ub~4I(IekPMrw@0- zoLaA7iI(5XNWp4^D&;i z5|JXeGPFI6^v-gAeR)wrpwHutdv}D`#uwNhpxHMW%-Ev?+_r;wpG8%ne_#up2c-I3 zt&}iF8{bOu=9MnWhe%Hw+{b+{#rm7!f54N>ezUybJD?iVBW+W zDFaO?X#M$Hxt34lJt@|w*-DbM^#sF+}4$suSmvj9+(PzfA=`ED-`5$-QwXOzJ@bDHyjNse>E)OzX~55HviD@v-Auo(4Pc|*%@1c6 zJvdUdea)@RjwHMA%^8=-+Lt3UT}~GdULdt!>&_oWxGo-zUL&kCu4;swC*=8iBf=4aH@(swh0|x7wO9>X9h`2R z%CfrJOgcZaxj4kC-a6z+Ux!kCylmWe-Ff~~UB~;Yj-GBUp9#kfA~6W>}RP54IZ{w&k@eI7=SFT+$&92PtVP z<##B}Jl2u-juxqDU9b$>T5vGw>yzgFMr#8+jCBI@i~I0@iW{{NmvhT2gLMW&0P)rA zBt}D?BOW-{4-p9lx0`aa8N1g9xR?lH45PN5bms2(#Y*eer>s<#&fDt{Yisgy3T)h+6Q&Z9~U~WYEKEyQ*wKJyX8=&o4Jms^SW<`UC-rv>0YDk|8H?f zv4X<1-{qi3AI=3Pd%y`Jv#zpN*EVg^FlPomAIsUiE9a`fB2BNz{@|p)<+88;cPR4w z84b1G8vBQwLNwyy5FI2YoSGX5ir$0n<}La?*N>8aG4QdsYxGd@#5tC7vRSlhXic|$ z%#8b1JXi5}uR2Y>OzV%$75mUWX>VmorWurHtY4W~xDOU{Pr3+5=S{a#-+N9MrWNiAc0L7`z<}8Jh?lP`djxttSmswmfSkioTh4;%=yD2q#*YP~ncb}Wv z4RwrbhjP~ZD^|gE_l)Yt`>+3c>AhCi(G0~I_ok8&U!2XR(*9#Voq9W)V%8hs<6C4K zSeL2t#IGw{txh(Io(hxNW-w3?^L?d$OfU4JoFeO>S5iLEIV%Y1d`&F&>3UJAhgQ>#Sx6+wqA zcWO@UGj7x)#2T)1DJymPgnu?DWP7}O5->WN0(Ck`)X3} z`;5$+CVLFWjIKkU4|~>iY4-sIzZ7+!L8elX`C2tMd+%$M-~LCGvo(8E9esjgo6r|$ z-F)HO?y$m*!n@5UgX5=Lw%ae11~!I;zGvxLIuS8rj{^QW6zzlq-!~yP z!W-D8UR3Q6e4NzbE3BMS{`Pf!X`_9{NcWSHQO{ku<;)V|Kk*L9^DKAnVa{$gB|{Fz zwr&jr^@MK;Ol(X;lj0FUG=R#@=+Y!v9i@}2W!LZ1>=^=9@OiHVE6UF0FYdDx zlCI&)GbhyZBSXh~NEr zp@t({YiX9dogMNO80mF)=k%PR{R3Kreaj|SxaV|+zoBP;QC!}h2t%#5uLb{(gQ#!k z_m`4vc0Kc6;nGfH>^^FkKn_)Q?aH}7*Dbf}lJfNzGnl(XyXhAT&np~LZGI(2sO_WD zaxkxo?G$1!8@F6qhnucZ>?<93p8c1L?dQC-$mb(#dwjKSg$3j|`0|Pd0(JDubX^Au z#AqkwTJlzrTUC+SdWbm!-M{EQgLXan%k-YLw-Sk2R%d)_D|h=_Kjo*scT3ZgMCmTi z+4n$n(hWV|)S6B1(kn-2#B;#lYhT;5TNm7lt#+MgE$23RCg4(f#g8hg4du19JOiyH zkAKKj;fQ}pIE$o9w%%q(Rn?EJ7?0$^+xvw5ZX&KW#33s*?(Cs0aX_`+YT11_$kli&A{3#WUbFehteB9pL?tOF3`x+dL`X z%R3a6YyV#;%_n$H~>Kxr~V$}WD{u%*D!5gZ!5U+&hYTuBX zKYVU>BnJ^upy^{=6WVAS!y(k%g$4WTutjVRx(c|RXHV^lZU`$1lC zYG_eRqv5X*T{|>=!gGxRQtS5J2)kk*KDGPy5goub_m-YoE^#O@e?T!qk0GTofHY!{ zLx(ZyB&hj20{rGGK*ar!<0)2x8%vh9h2AO$gUQ}Pl``)O$a}K$-#w{!Yiw{M!xU&9 z!Nwi~6zFeke*s#A?igOj2KllTaOvV&@B!;T{E!QVQg2>DCL)uoQG#&f;VBdXRG(Nn zfKLYff{7FfNa$CA95)6DPA7bj2Nno4nW*{;n7}qcCW|xDUnjmm!Qs{@= z$5oMiV?er}08mvA9X@}7`tZ;cJT3cqo-Dlw(v&1G5xT2pBj~xcAAo@!`d9 zv2ZzI0&6X_fvrsah05?f?=k8_o(@5pKo?Ri+hA?54?h0lMFg$NSG;Bk1y+xz@)E9= z=aW^V?^fJktk|AEKP}eCpKk`;R7id;H8ny5-1W}PDOn)0y((}NFMY5NV#h!_OXi;7 zOFG1pP-egwHBPoZo}rGA1c?|;Sl&iRJc1-%42_We&fq9cxYZm28=U~2WrYl6zgWZ?LIMG+BfACxHNOUfF5HoMmRv>HJQyh+N9Ngqtur@Zh(~2Qh&YJL zE&w00L#F#lf?eWvQd1xk3V5K@6KCf`+qeH3?F7XVqr7 z5E8RWXFQk7f72!}ra?7G!gz#?BMpsq8i6+bkQ>(r^#pW7_(U~N0GlCo=8+E}o^mh5 zP6wajE`*qo@wFi2kAAd}pePi^f^%^e$k`U#f!9j}hkpfO$8l4Y|BwHO#Map>$Uc(` z_eoDKPq%e>Dgu>zLNoV zcxM^N{k-t6Ww3pz>>21e`Sl_s6cE@l-AR-@L<%BS~@1Rpkb(a1pHjFm_V67F5(y z6@N@yn(zVQ zw;@v_!707};quKf2#@?7r!Vlq6wm7elc@wIg#8JhDBA}0!G?@TuxpP=MPM=Ic}gGa zVKSF&kFP~orX7NJfst<^!i&Hn(Xn7C=@;}_CPTDUQFaT(cBj@W3SnSM$-(fY3Oq=B zwY-Jce?hJb+FXrNKsZ{CM>5A0MZXR$TlL34@#KJFEM)?*(G=0-UcnhFA>0u6Fzn7puaNN=V0W ztP*xU%b36e_VW&OdWjK>i6=o~z}iEncR zy99f^62`z5Kx_lW^!8{I@DkwjmS02f$^xbK_&3x+P_scuqB4P7glkX>hDnOT9s@96 z(q{S#0~~vIqJsq`lXD`KfOb-4@54o{^$6aEcVfagZkn+QY9WJxpBh>E!lKSSV6mz} z>p`^5My@4%Vk?9H9T@r8tkgNg7UcL2Um!efz}JS&cx=OtzB;Tw(zOV^m}ab}j_c=w zIv>WTq9J|ixCi(M5=vTB2dhIZ*i)bwftrG@z(OsRKtJ6HBJ_S;AwlzXO>flBNG?-R zC<1N(G38E=ja~#veHku~8!L?ju-YBJK*H5i!=XPxTHzkN_RK+ucL1VWPAkSzdFy*D zh|)GUF2|N3<|AABQ2+NDJp~EWIBputQIKl4t1zTdwJ?WIV%I`!KOp(wjEMadu*Ek3 zXuws>W5okh0UCsrz)FXtwJscei{Mb{p3_Ib{^&q-Enx0U$FS1-Q$VuL%5Z~j@G+nU z?rEYm^PFY#OMt6w_#}5EH!b_M;;gh|-#ZCYHi`irvj?^?R1|C$rOzuMYmmcA!1=F3 zGMRq>Lxg-mdW30tM3r?@WR;?H>PBdMwLcK}W}e|l4RAY;V<6x^Km*iHurr$&a3y{m zR~u;u?9MM>jV;AA7i`=xMVg>2jH#nGf;T$Xyy~)8B+SK&LjKo58 zDNKNC+(hZv)liEyL)HYJ%hLZLcOEH=q%0WJ0I&8qkt#`)7WEcMm8U%sKu`NL`18QO z`BT|I_;nhDz`$yRL$V-fT-SdTctZw6w}PD>a*bd!AUNV_utew=mSYdW4__L^aO@p- zpxC`s4q*zDJzo!7nn1DuNcJ;54czO0x}Jvx2KTQYa!FutWk*l~Y}NtFsNg|(3apom|07%n5Q zvBuXxBU~D=%)yHgkZIw350d>5F6~L`*T}a}meH_>$PHM&Nt!GG(Sq*i7l@xj!-k8B z|L2;)g(HMFTxgEFXyZ777yU|>0cE-$B@a--&Q!7_Mnvz(0tWtZJlF@S5Z!qT$=mDt zL#_;&38pjS^)0cY`}weD!?IAdfa`EsK73dA%Ph5zKn(Y4V2Bj3;Eh{A54=@aCPn;? zPqtB%8jICL_+~}Rf#~g^(OqA>h?MAsXcz$}=D=nV#2ZM6Wd{aClWYm_EIoh;iIn!& z0IkSgvyjKt$b$r*p{1Dxd>nA)n);FiX`vtnFggI6V0ER9XSoRJy(`P(2~#oikoYT- zJptgtX=FVnRt*~WM4m_3dP7_Rfa->m&_*!}IB6&^{3?)I1T?=HQHwAPd$NEI!9OujT?b@Z@Hvl+G(z1 z)axlV7U?)J{VIOq--ih*8Vd&l=J{0ioIV?SWR9F`VVn_Z>Mxcgi0N%^MW8g!SN?K1 zwpNo6XA2DC6W|ra_BGx@jX>^3h~15y!P+_pNAKJ{R~N6V>qfJF@@z)!mx!VlW@S5v zD1n~Z%Vp<}eY(z@9Qmt1WzcQVEh?QicxRuf;DSc1`9jROQq<-)uQztyUYGVqcPPI! zHK;s7iU??v3kq`bFcF*E+I^_Gw^5DzP}%&3^|e10H2>SF;LEjrN=mIdZk491&i1C< zaiIT?p^(}o#BC9$OJZ*&Pa!wtguQ3yY&twSvhlr_=I$y>%!f?<3qxyTCg$>KH97Mg z(%V5Cu>d>a9^23%l*Pc+FGcB7*=}DOo9Mku)oC=J2v8XnrD!rAS<>W3sUY3r8}rG~ zlhDgOv-fT`jq8l-uI#U|tbrAqszXL8QV0dss$LY|qjX}B?{~$TOu3tnyBTeCv_oUS zo?el#Z~FE zrI)Jmo8Q`L{q_CffNU2b^%S3UxQ-eESM-o!f@Z{XC{y?|vIRh#8un8~Yh=yUxJJ(rDFfK1p{Bs0|+_79-osqKx1ZZ=q_L z?uT59l2zU9lkU`Wg#~(H^r1G`rqASEE~PT(0{i{V{GBxc+$iElZ^h*O>*_Cjx8Q@U zvLbc-JUBywBh&S7M(}STb%!O}957{X5Xu$7sy z!{tNrf*W5D6Rqp3Zjs+a0!jT%!VoV(r6=Fpz9Q~I{?VBkjQl7^wAu=uft?op?;yvQ55Nrr zCn{^Yr@oOK$rKOLHcjU7$&y=)#E$gfgs>E6e9$5NHDL~>^pKHq&!nT2mIO&3>m%BY zdl){^4Qy6En2>-`4u(INu%Ov<6Y7r`k}cVBt_wv2IK(9n_gbxIz3E5%jGcd`KZra< zNw;(SBj|xORxfPL$?3hr04^k_Ge9H)$6(q(e6lmM88};Z7V4{WRFL5xVrn_{2qBCY zoZ+wj#k4C;a9dih&Spfb41WB8mgGOWYQVWBk(rkWnEHs@jQMIMA3S|W23JM2Ka}v5&mZ{9-(0EZkZs|+`h~Y z_&%_fF@qc;&?)t54~u?k`MBEtrb!_7B2^q$69X6_4ecLtZp&`rJaSR<>^;J+D8i*7 zMN+ww-(aC^>4)3{6RaF+jq9Ibo{me|eYY3M&h)&btB)F3LdVQ}WAc;Hzfm{?04Vk- zK^pD@&;c^}O~a_@Tz=H29HhUZH)VGg#SCgZLXopRvk5E!MKv^aJF!ca^1vvfy4tMcRn^L1KZB#SS-+iW zPGojE%QrmFO`rVOkoeoinp=;U$r#H~1XWQMNnTJ|6g&V_nmaI4h8@NcD#>oclk*Mz zsO`Su@R@R+Q^RJ>Os)wpyg-Qst<#P-zpBT3HP)NtqIkL-j1o2{QrYhy9WEOyQn2Iu zPkEf1;HMk)PZ`3wgU(g`w@Ucz4uxsnY^v&k4+lO@#b7N$YiJ{_xHv7OZjj}yhs}K? zv-u1VdNA>sU+u=RSu~tFk*{T=8)*_9duH^zcZ8lSkC0btVcS`o#EKdrBO05-#g?^3 z&SRzryy6h=6qy=#s}veN&~QcR@zD;^uZn=Y$qP4*$8_z*`kvXL=!IrQiNzv~0qzPF zoZiWX*T=q;=^W0o3Pr{F=A&bRHmz4YXVJGfbq_i2+(Ts(_=auLQviDafdk>;7qs4* znnvS(LCBGy=yzxqrSv^DdlIuQS+krFQ3y{*F_ZY5&OuA<*zbM26JNI4E8Q{I9j~_J z5Cm=EA@%+u1velu1#l{9cWk4xZ0$AFfaN#ePviodF{Ssj1wN#=E$7N=*97!eiH(Pp z*FMVce1xW@lcn*8pbF%o&OBl(Kw`im3}Pox>*6M#&GfO)HWg?e5i}8yJC57ir!swtqa6^P(Gjh3P#A-`eHtOx1^zG$xpx1 z=z;b?2iuvl|HGo0V#ARhqU_s&lfR69vYd6(SQQ*EvEOvyjC;(2S_t`)5WunDs!{Mt zX=5Y@@KWB<2_ENkK(5c%ckOQ3qb%)czpM&^^K_c?S!Z z+VU2i+^H0IWw6scNIWIOO%wHDs{w>~jPSWk3-z_vHX00PoyeJ`5^Qgk@SxOP;Ub3W ze$~wSZDZ9hM~52r`d-zz)+@`McUjMGTb_9H7_?OIe#mK&&44mIzK7t%a|#rT9X(8* zm&xpNFO*K97t8t4X{;%yfH_K+ z^$A;*jGv-I>MihP=zP9`xP{Ru3$0*#hPWfj&9MR2Ri~_Ju zfUjYJ-YyECoEB-*&EUeRuQx?=w0X&Hf+UyrA@iU?fnqHE2>aseU$PF|98}hL{XQ~4 z!mqjpg{~{vZXZ^=NX~qZ6#eQ3RV_0L(FmyJFC@DXAm9nIWiqF-Ja6wpcPMFima0lH z#(h}G8&u6sj>fQT!oN#Z_jzj#i(?sGA|?5*@A{5YcZBExNM(*#;Bq9_7JJ zqYqsduqZplw>3R};uAU@n-4y`vGQ*C8ZPv3gWc{T(_IdSWKVAvp5&a*S_fQOjI9_C zBDzYlG=T!mA$g>*=&?d}HbuWrDxx}$N4Sh9HZ5Ke#h#59Ko2Mf^qR6Ot~Ix0x`(Ha zngI)oa0aAx3t$*W-y^6-o#+#Ins<|%dn2;&=%{lQjo#xp&4~l2TMooQp^+muGPp4_ z7^N|s+nwl!+WltPQu$$kY(6rgknC?zo-Gr_HzM#^j260*iZmg8=(_*eVL~q)vV6P% zd-Sg6P}O}IOE#xFGqj#H_`a+xPMRZZj4T*a6pCYH47P{SJ%|oM)(R@E@?g;75}{T6 zybD9a$(L5Zrs`PyZ|NtuZ<)`lZWo;m)Zt$+MU&YnHU>w=_HJy^ zHg!?Ep*s#l%7uCsGX>~$i&9{2!FwAo&UTp2*{zm@M%a}jRkJ*q3U0T15IfyfpVfYi zva=d>*xCEr!ul_oULo7YYVHB-cKZUjks?4qnz@?!rL{VnlES%o&?i-ZS(RvPav*8J}NyJ+yE=Y%1JtFU|! z;4+n=f)cI&1ao00FoDq}FRDUKAkN4vopJ`IKjvJd4Lfje@)o9D+LZ?j$#m?fVViip!AZQT}f{5KK)oZ({Z9cbJMn5U>^40#^f zL}fPB!xa>sQtJuwUz5(&Lv4AM15Pe({r%XU79Mt0OEOEhx(^D#KWBE=-0qqC4O-Lx z%1-r2?Z*o!SCOp)fGiKGY%uE@bAO%=p8z~(p#M1Ty489w9m=%l2@K7m-p|f|kTB*{ z)$J29_qe5XoV??d`l_&}8!L9Kf0S?jYh#oz5bd3L4Dik=%qk{AqyQ|qd+dV2+ph_q z4Tzn(J;0q&DEnjG=DWwX&~`F_GmEp?C!ZIBIdp|kN8h<~*DlgeYEzR+3tQ7bbmtun z5a~2F7Gi-IVjLoRJqjj#gn8g7^y-Mcs3= ziBrlO5l?@}Z6lirW4Ao{A?N)*;9WuTjdNaR@e6nbuvb|48{pYr1?h{F_ATQF((0Zi z!!7<1KDgC4y8tEra)`o;EHFFNOlqz%a#w1f*jZga;<=K@Dgy{EM<2fL&S+E=tpz;Q zD%c5H0;`oAa~?R6bAV=H_c$G$VU*X$=kY_|3`<#XY$rH<>vCA_f1O4lpDqnF76vi+ zB4xM@duWdQ+QL*V(%4@tO#o+#ke^eW7nk(R6)lC9xB!Uh>F*o5^rSSm?1h|GcsLis zzcL!-*Uvlpn(H*qEoimwvR)Z|qRME*(Q6D4$n;!lcTH9?%g=T&yU=9_d#nIgV0UR` zyy|2zO18M}rb$ie8*h`>_(BYeZjx9^tT(8+eA?{oAEfBBEL`=gKW#K$dDQPNFJ|9+ zT2i*5Q8g8L(C%YKFvq20{D_gsQm*)GDl#}>rN)Tbg^hyn%x`Q=vX$KBPiaj8G^A3^~|od3&%e!LUXgtbY9$$JAtL&8r#M@{P23bn!P%jtGYBp? zyRj!5szh=RrvNsWaU5$Kx>@Egs{|?c^dM9<@KoKyKZBSB9RAZP(>>$gh??}g2MTrx z`K&8{!zgpQECf4+MFLW>J5p*xWq0SUJ}&((tts6skOq^d!0V5d;TDAjpehagl; zZkdtl3g)5{WVo4a$Z>TX&}I5^G@>LCAgw@}=n;aqewie|Ns9=ZrRUX^kb~3W)&(>9 z|3pZX31fl!nL7dlyg)+O0b2^X0p%w@U)~(V2B-oYU~4`GiR6cYGJ;3d#2yghVGCe2f&dtCF&_uc zCm^i_s|dB6)?fA)IgzV@sjbeJbo`S%uV?T!CCGZDMbnI*Kmx}bUI2uq^&`x51(LTP zoC!y@OcuUaHmMLBJa#-;0R%j5>Jj-39U=u=5Wo|pPN5!j@mJRT%ae~fga%QHj`K#+ zM>GJ3+JJ*mS;FB(Tjr31z_1@4@fU3;{*&LFHvCRU@b(TuWsLUn5~vXxH3>kfnV%iO z90g0000HKiuK8yWB%oDr)PU$&QP~VBF1gbLDPen*p++n_nT>$1n(D!T$8iGy%$-&k zDuItz5M(i}(T1R)m`Rf#=QC6@J4(4Eim0jG%<7^lFL_?Fh%t0 z$iqO{?JVI6ZJuFQ0<3EM7BSanezr>}fz5!=$cCL3LgJ8gcZhxqzFL#a6-YzRgA|su zXp*x5A^|BR%}U!mW8huY8@R|h1KIhII%G6LECNZ_8P2*Na&OATAmuZO_*Eu^{^rN? ztBl}RQ!He1&UUFw>D_aROREnzEPQ@H$xoTEFVfxs6-60)>0 zC~E`I1~*`HNl-7{2+XFACctcvKF6RPi-2TkD!!DV2!@I^V6N#5V;yx|&mYKaz8L+S zg@|09C(u))3W_Kga5YHG+isL@57)RQ&n%dZ*k1&+UABW^A^n#2x$l|t(DbfdN)1@wm zyO`oosvuOFlm=E^gp>sWi|P#Q#m-s0EC4luH378-NbThnq1D8916%D;0)E*mEn17> zUiQ})=tLKD6~V3Yw)IhmvE!vv&#=4=F)}VDmWzNbRVb2H>9Z^8Kzql1+s_>*jOu(s%$HwE_`fRX2wLa|LpIXFZUQpmArwTIqfva=*+J zEFV*#IfYOHsbc{P`M78~v4k&D6pIjFoRI(NDmouAqL1nv#d3N?uvfRMvNXo8CSQbp zF6@A<1mHOmFqzp<%MmsUN)5w50#Ap9#M7nBZsAOLW>Ad|RkI1?rT#;7i%tfvW5)t4 zSR{K0!(dV?Ttj_Bwna#_SV`9_g{9`(Sz?g-%LiYfC&NuJHZFopLtc~>85Gt);xPuT z#)FHng__0ZMlJSAW8e$BgNAQcL^?qKygy6NNItVBE^eCaY@h+oZ{{fN^mdkF8iMTcKEQF{@1+q^6+ch z@vSo;ocnjXzg%y)EQsDJ+CJWApAlIcw4HDR{{g7`Kr>@|+sSKi zunv%^rCkF4IPU%7eeZ)oIL}oZV-eYfl+s7Q7fSs#xN#4VPDkk=w;b43FYWO;paXw9 z;*_$tr-6?kR&5_KQ>Y)V@83)PI1*mcy1B62*1ej%*VwxEN1O@&(anpShc_|$ZSj}SKdk=4$d4rKWQ$`JUyntXqHn~|2Dz*vJYuDEC&CF z%+W6x=z;SSMA9-&-@O8AW@^Ega{#+lh5dB!+9JUzs@b{2@&pU<7wT}vegTv$C4pE$ zvTP)~*J$Qk>{uF)Kx>?q?`rTKGIJj&d@7i}vUL5V^KWyYV2$S-z_6Sx4I$kkg4Y=d zZUMakDox?{xa`4w9oYaZAY0Lw>_Gt(e7OYtrC`c2$AiQ>X&`l@H_dtk$dBRDJ)mM^ z(e`($z-m$w+yE5_qtuHSAQa%eZkxJZ(*@rCA9B4>2+tLO>uN-t3jVRg1Dr-t@Ql!d zVhj6HM^MCLoU3U`+RcElrMt)!0K?Ok=}QI}2`*Y-6!*|fD?uvRvyOTa-SSmhSQp1JDu$H^Kn+EQlKLWI*FP4gjhkQ~t&!|1H(?wtXP? z&$!}Q9vx5l3n-)k8sY(f@Lsq5_4kPYF~ydpvhV+x-DP2eZ^BvbkH$+Ih^gY<4~XJ-*smRyDIHb;P09e$v%=z% zdH_)iOg|(3=K`h^CV%%P{$~>QKc^QB%$2n2rr*7)Zq_(beZBMa<(8y^>Ccaas|N^H z(?1u6?~mKD6l|mAisv=aO8tvI*SgqGA3~#|$|`N-MDFlh#`XGXdVr{xZ&TMtgiQ+1 zmqy)nB4h0#i}udhR{e-9qSo)_UU{03{cn||SlX|!;g=CNGsYCY`Vu3EFJH`gQ1)#d z!QN|qYJHmQCBE%yC;2c2#A49F+$4zZ@zn|5FKt;=FMkmXWpd8Pq#i!x+ zNdgG`BD(9t2f>AL{mfcfb~w{i{+wgaWHVPj*s$$w4T6ny_4lF{%dXziAJd=$40*^? zZ~5=SFNVQ7+ACy@w1~3d;i1DF53=B=5|kefPPb!*_8Z!Q)>uwUYJs5gfoNhqQcXiN z?x(2ymT2#;c{GX9#KKbp=7QlsRS&S?caUJSM;ws@qsO6%@B{Xkpok$ynUJ^PJob;y z?2$%xc7294@H;`%pDp?4BGOlSSx;_dvv1k%I{T#^gO8z5{$DqbzuO2Iy1e|d!OuuV z+cl%0px?3qc4D2*;2GHW3E7i<*s%)(8TZqi^lW@n7jDsf}?lw;CSc6WceBGYd^$!Si1+$f-tmokPV~!D=3* zaAH5A_j^`OS-4~8sZj`IxDs1s93#Fo=Akjck;AV8Sr@i`YUv3;v|4|rUA`wS9Rah6 z|K8Y>qPcR@XauE$NLBQo6QuQ4tbtq#Oj^=SYaRNnLz|+_eQ_@4+5G;gbz-F1MWjNN ziJWNCyb^j40{rd9eBE9jbV$Kt9@qB=4bNRX)3B=jXB%1WD*(~#_BHS>73Dc1zvHU zWXsli%NK8XImD;Ze`xMP9Is~G(_&nh$6T@Lu6E?s{%j>Y@fw@)p8l^2;PG@&a7V?z ziY|##mVMXPD{_;z@H3-jZ`fjL`eooCQLI<^n! zSecSclFFcoc*aK2#}CrBR4Q(72dZox4$0TrpzkyaQKPYZd>)ethJT7L-1qd$08(d` zqSS8lJC>us%exO?*nBb&9<61Mq`Mp?wAaP^ecz6OMmdan^=_HBkQ6L-T);&et}|~_ zFZ^GHh2_6T9wkNchEaa~Roa*xCJ=3%=R9XiAN9Q6lQ+A-MMtNfrqf`j$81OQ{=nVy z1Le`1BAx=#E$QPUz$BQaIdm@J{W;~VGvuNyeTOyB8{0LbqoC(wXxEd!R&E}t2Nv?0 zO;=%LCgeqU!2C~FQo`I0a4#tW1U{o@i$g7ZH0|G#OSgL6JLi8>tqDho`>rvFOtDAz zhOqAwXXyDj2BC`BWBIE^&9zoW!y$f!X&2r3elccWjU7&;rqgNLN3nVt`Wp4Z^}hf6 zzHJas)_r+p=n3ve*b=E9B}eCV9Gig1$LG8-*4NJt4J*9W@Z=`<5U*3d(e2MGsgD0F z=-}}dwkS&x=5H(MTKFqRXR)t(e(a52#%8TaL#29GEI)hCL_@td09nWJclkIM3>mS1vE&MBsc0vwETquh1&~mY={Z7L>(O_`VFgd@ce0lC0e87mp zpkR!_aS-F#{i+DeaBGG3lbg^r%C>GXyUpX?8SyhwH>Q2xVQ?@IpMu~s4NhI0-8VIe z{W`7WUtQAxc!=L7IbS%7JX5Hz6He&AK|+a9oAXZPQ-2MY^-k9FaY%|%E1ZvyP3J4Q z#}&nvPR=L`rzgJSAfA=})3MwL#Yrw3*t7EAi!Y z-5xHk8Ng)$brT;2k=xQf-4ao?)E4zY0)lOn^6Dv{ z47J6@7r;38i)o(Wp#ygWE%qIW_rN6BwObMUa6Z@+?r_xE@%~YCYsP>vTJVRhn;B~b z&}GsO#qu?7?}gi>$LUU$`EvYK|MBT^akq|PrID&^P7xI6 zwm~L>!Th1LkDKxFgR8Rvx^lpR2dEucMC3_C+d7V?qT!{IuM?C&Z`fK4QhCcnn1Al2 zrU&P2qcR(TQDweU2bdmIGyIUF*!Gla@zFwY{tU$7yxjxjua{9+z?)8wWprJLp_u`8 zle1Y`Qz z(k!i}r}y`CS|Fi{o1XK`bc&ftn6!;*xiVUzmT731G#PC0y4?Z>EB~y!Qd%BVF;w>M zHMBrK{l~|!Lb>;?158f67M2=Pt#iOA`=2P3oGuAd*u0kMH@&0QgmNVci@8)%KGo=T zs|S)Q1tq2cJfqTrk^IBOEo>ej7#bQHEok8M44$B^*fKSB=*-6bbE?FZ^iA~fW)P)ydN+uby*uM6Xol*0l+v1v*&~LGm-&ElO2tpJ3yI|nZHm3e;r}t4m+{F z`2;Pz#Mw)nipmmOf(WU;%$F{7iW^9ACo?f}zFW5oA{iPz3TQXt`}%5PzH<2=C2DG9joi5M#@2Q5MCA6M zK5!l-B`&_B@Zwl|d75#2J)`25Z(=%IP_9D~_L#(nhW(s4LqGHPJyg@e!ol{(bP(;8 zmuZ$tMR}T1oLm)QWgo4)uCLp#R@$S?p-T$MbXlf9gmQLo`>-X4y}k8JIp{U>w_F_p zfEZZzUor4wqc0ys3w?{!9)_ov`B8_2@9>Q_sYPi`fjZ!d$`#SWM;@T8OYk80M?g;_ zM^u{NgV0f17+kglEQI%HOrGYHUwFX(NDB^0)KO^)`4Sw!d`c&Fp^E_r~7qjwb0tgkdLOb#lbx@Su!vNiqn98n3G zgNISiN>*{&2)A#_R7g6Qldrl$U-$4_R*MBkLyp=>nr%dE^<~Fci>f(F4ADxKKgm$J zh1w(C@TP~6KU0K4ckIwu*FUvmQ?p-2L2$m|fTX$CQGlx+{PsP!IUl7g{R8tDI;Y0r z{6yWbvRo&b+d%21q=tGQtjS!SK#mo2U~SqofTL_1vaNC7R6W~8=9sE9FRp|Q@qfo( zIcZz}!}>G*e83uW#N^(3>MLi*p|6)9H`QeKaFq6P;%^@`ahniPUx~LW;hVc!$#_D` zAMwViyv7iP&dwUs2|CSFbPAL%uk>*V76-ydcg-v^SG5Rw3+KTBg5piD)r0g<%8SO^ zh_E%1s%1525&Eh8>iFZlWcS}qU*l3XFfr-uv>%malRODR^*b>3Vs*??KKQF;xX{K( zi)Aa#aKjE4J9VmWxyE8gwDIMcJu>b!JS-R*Cn=3vs^KkD`&)$9@tq_!jo710z6Azn|zblwdKCwRQbJujo*5nZHy7A7UKjuvP;O6YcRg1&qsaoV6R`aX< zWe-3nJ7Hft32$Ma-23KMM%l)^7Hp^y<=Hy_4DAq}X-Z9`5r5;m5zoZv{?#)GKCE?2 zND@HHqT1@+kn&a){|8HqpM;To#g1CRb{6>4D45AuS7ME`7xNrO!|}H67%!*wYwa2) ze8}Auj~H4T!#j53JB;SDr5o+-RiZ)_*H!h>gvvS*J>$i$+3(}hDXw-Oj$efGWd9AheHy!RDQ}spUh}JEOP}~1(y#6qbBj-tn)4+- zi%{_d4}?e(y_u#os0Q18?iA1!;Oo7gA@93)rnuCf;f4q72C{;}s@PvLNerllpIw#( zy-;4QYzd>z=e6jI<;X}Sx8>AovD%q*(pZ&L+*J)9-6xgR&dcQ+BFBsymCe?iLN6=2 zD&7&xN_8q(ALPkEXbUe!$I15Y8lIb5?97?ND?{ECkJX-tXV_sz=PgzW+_UXTKx`%V zyU2WM?_wUeVPNx05;8lfIH6sMW|vMHU1Mb17nDp{NQ zT2-$_V?EHm^G`V>?3))w7lh)yWXVEpT5`+uT`UDLemcb`xE4UiP22j)ndh8-PxIT9 zZhrHnNv1wTgw99}uS3g$dH_Rbjga*)l;1)d6`t$mW9!D5)-s*h_o2r@m+2XQRM_^e z1=Ouh{7Ky#9Ygvrw;y8?(fd`72{GN?U)%^49=by7$rI6JMx}*S7smt(XE+~q8qcGs zOFj5LHZh7?_>r{l`ft=h@lMLq&y_UcbG}B}0r#YMNKep{pV3A` zqKiT?j+o+l#~I0F!WfcLDlV((d#!HwLQH5Hb~3MK6!UkR&tJT!mo2#9#5;>3#U^rP_C~M>bhnz8WRiFK343aJ(u2;2T8KILu`mW>&DFP4w3cH^f z?K*-5SwHvNYD%MfKHIjgG3#pX^yJ5PQp(>Bt>|}-nD8;U+O?oFltbIGAo)qMj(|Bi zp&~)WJn==g{IIPr`>X&|_f|DhnVUfw=~ml4qQd0DBB>~C zikxv{NYSnv&+2y#nxs^{{erHejCJovjGpbXR@;Ca4w-NZy>@Jm#El+O)RN@zF`W*N zbBupt0k;3HKn$MF*up%6F)G(LOt?L698+UXhissFH|?dmjq0h9R z>c&glepJ!9N-YU*>usB6mEgCLrW6n3jb>wNR`-6a0aLk4qwYCU+9qLwTs|gE4U@== z3CcX|1I<|-xlc>QTR9|SR|t*jg0rc(0MJx|^#@BYKn#{g%Bi4`pa#?I9O)A~0OO_D$St&nUio1<%VMf%+H+(ni}sB25?T z?-`HNnXGQ@DE`z2DnQJc7q7K*lDuj!+8J+*(w*W8y3cA*CiCTli-IjpTpeK`XqVeF zMcF=yF>J`sU9wA~Q}9hezTu@u4>JZCXTj56y%;YYr!2Y_#8hF;WG%AE8Df-;jrI5? zntv8NeFX>9gy(uPXBZEPR{mA}bfGh7R|KcG6`a<45qGRH)QCX~mn96$7yn3=O3TU! zb0_deoB7mjI1>;4_O7XQmgCLNynup!-IF`^3q&b3bN^doQ?Ol}btLbd<^sp(@#TPQ z;%$nSJqFT7A6~(RE>96dY>xgBY}4Il%Z=8^%39Y`4KXFRs?emwEil$u&a)IL=Bd_k z_c~J>HIi&H^#fVG5M@gXJa0o`aIm$T&vORRJVI{&<*1qON_kh+lc469Q|&Zmi7JUB zC5G^Lsm}?Mb<)qyq4fJ3IvLZDi98%u>=V)P{8C@w<>H4x?8w0wB{;_Cr|sW?+H zO}KNp;wIi8lLO(KHQa0+vp1-M!|-(%sZO1mv{~}K$I}~j{Ua|E*IK-Th`x>78WwTw z2@4y6-ejpJ8Q^5#6~!S#xE8LOWGb8ECwBEiIlhe02(d7p75-V|9*1l%;^1`qCH?RV z?6b%ZFf@uZQV8EOf7zU=p@c{LR<3jWSARE8$axa~ysv=yDmJIwV=YrK-CNsSMyjI^&{}QsG{oRjNg*<6d83w+ZKp@4str_ zf2JTn)FVt@XeP8|Ss<)`Oae=F=V~s2xhgNL|H}PAV4+-^hM!%Gr>LQ3&JO=Ak|^HG##x1SLq&vDw+(>Cc5t&M;|8FqRYPO@hn=hBWotUVVMB_HpF!l9rYqo(;?nqM$LDI~#57N5zA1a*c-Slgz zJ0EthOH&t4%gPoEnxp) zt*_yQ!|hUmg}KFqTDpjB!qBZLf7jqx7+Ut0u_ul>PghE`RPI}@!w|om6w1y=n4hwK zP;;ud-HWNK3I4Z{1lNbgvfAShGR%>#O z!!tenzhhbpb-ULJe_xv0A+19&GrAgXC3&+*lnB1S5^4a(BwWirOU`(@-I85`Mi#i0we%KI&*|&8C0{me_RK`lZWSLi&n&V6wK+M5a zyS+I`T$N#MAnWe?M*&T)_4OC9PlAT)(IN`Xt}fwIKPKZ`+EPQKni`vH{%8vETgBd- zmx5c3XE(uL2-#pI19(EIKGXP$F0g<@ra5=HlIA8R;d-|!DKg4 znITykdDQDx(2ddhXDx)zwE^8DZBs(J zXx6EfI#3b@t>9FAgGb((v)?pNfwsMWxYE$l2#=MruhM?Ve%;r*qVI6bo#`ex@5c6T zXbPA^bnfJ5mK9giS~!1vZUjg2XJ(u|Ssd?*FtHe5Xj1nan7We(f_Uk&JJ;{IcABnJtvakRlpV6}?E^c7}xJL-4&ciwo---2N z>R9GQQIw66W!<(|QWVqLNW)hH?A52xvU|-zwJPVixzC}q9;viBqpHg=LGcAc$OJ^g zSNHw;nX|B7*TyWxYP$>8HqV;l=_G2u7)b5UhAOT?*F3a?hFIC)Y z{?{PjT6b|l)?L9jY_s1Bk&(+Iu0y=4$9S~Bs@9ZQX#|uFcvMNBo18{$D{df z0R?pdeputG^C4>@Yi$-uTS64w=RJ7=&i54#R7SB6--qoPJY}%4GH_!}(a!5xVm22= zSFOhTa+4z5{q>V9Iu^w>q!v-k3sgF#91fyrRtAx1|H1!cuYFl&Xy8tftQfhI(3qb*cg<#xX6vc!0&b=oS#zFE{03=wQ4e;6 zJ3j2L#a&JLP-sL?pSSEG|I7&OH;VF};`7cNm(^-NlGIa2OqZJ?*M&OdF^V23|B)?C zi->r4E=liw+j3C)y83{A>Lz?QZRTePf4P8j05t|REf-g_`}IrL^7>wZ!NA88)NE%n zJd!nbE{Vx>{&vU$AiT|S6ndbpj2}U&SO}*U43(Qyd-Q0r6j+ibukI~8`L1aMb{A#b z)zgt7iSIy)&omnw?c=|;13I3Qg|&z*$HHVLFoSs9mkq6|5@3W2@Gx2s>Y>WALE3pv ze#ExnOM|JhVta|_in0YiLtf_s^xAbzj|<#5(o-v59+mVma)fYF{s-l|m zCyhy?Nv>n#A0k<}tY1&I35J4!8$Q@`;mLdI*_EcL)Dz$X&&19%dwtcJ}wl}!e zT5F<~H3!SiNG>X;gq&Ro%izhGBp+E^ZgnUT!@e6E#(EG*t^%4Y80=>2S_yhnrPxBE z$RVtIR2Ns;XV+M(3n0J6eF8m@b&kO3n)gZe1+we2zp6@hWY+Js{GsK4NKt*!>+!}8 z-|o$-cnxXhAvDuv3TU4GJ|DTtipuKVV*c6T#ycm_Ld_h%UWM|y=W$6S4XG+wQrAq;hOkt4HQ z%QA6v9mV@v`fr#@*%PKdMokvyJh6g@&u;pD=TLvJja#USdPrhPPQTnUzv|iZ(gfrL z9!)rEhA?)$8+(3W6e$-~^gBkuHWqz5av*Nlvu>X3)38l7t=_2U;jXy`F}mTUW?fsQ zGsouz9JJT?PkPlUw%S(ECvDY1=9Y53^h^8PLu+I4o`EeKWj{kWW=%SaW);A$O*UVw z6|5JoXqB6qk~dl>;+55`S*%KXuR=eCECPMgpCP#Hq;c0orAwjw$avFga zfObilk<0aF;fBB}cIIMv3j*KYqC0X)G(7EnH9S=z^6aD*+TzJa2x1np1Ln zwjU$-EN|>tawOXFzGhLbBGLog2w8XPeYdHVdE#~IF_U4~8_sz^xOwxi&6OL!Wp%=p z&ic`7RJMOq&~RzxBATYm&r7h}OI~!5(Q{VuVhf-uPGe5bd-^!rF?m(p4W%WrjSLYj zD|R~zeE+b6ni#m|UMHj#Jvz=nuXXoblaAKPo3pU=jpWiiSR5IWC+evJwVG=D-l(VM zy_rJAh6*ewb>Wn%VYyPjAEiw0yp5u1;ZV~$8i_724=n8I5Mf3c`6p!1mgGFkVy{fw z5$E_8Out_HqmMqbGJZq3efegq$f=_>!2nslKDjD?bp-Ak?TU0LU+8}`vZpV%3(E26dc z!7SB}-X`G1_X1OoIR}&%5inu=-U`#*irP6BdG!>nRP@j{vxu?xumJU|z+99GCz;{v zdFA|%^jz`c4q3J6<6vZ^iLrv4)6iH>N3!QuMaiyN!OZ7(EXPcj36`2Jn=gIC=3u=S zXqT|Yk&z{>+888~p7=!+1Ab^j|J~C9Gv9C*RcGIg7|8v)Cg{*VigqMKK){?xZQ_6_ zzQZJQL8vl*C7*mWS5Wa#Yj(8(_BMQCbz+A&Mwa(c-dt8l%=JBok;c(QLf+8H_JvM6 zBMfQJ5;sGYHZt>p+8VlpM13C5Q=uJXXb^#%t+f@P+2G+{jCS3qzc+d@ny=tJ2{f%c z6Rkh)hyr4KtdWR7$(2j##Iwa)#b)9+XuTW&rd*tti7m#t zz_#^_%`t4hyUVug{fkER8i$bSsmZ3!NK4gyZxgZJifQa zMM4bBbVo(5>>RY{j@l+lBqNzvldPR0oWeoyR)JfC$+jBgbFL3YCUXRJ9Hn z>`_9oG9FoQS4yGu?n<+6l5(NsDzV%zEBW)heV94DyDFu=ESl4TSLZ6$(I-NqlnD_u z&43ImX&v=^D_FoO<9M?20|T=}w0(DK>~T;-!8d3Sa@BFa5bj9vRM%?&!>dJ6?4nf9 zF;0mzt^nNe8)VuSGxMm;_x*Q|!>l`G%GVEU;8kIGnszIs$zF{p++xRLE-je=+t(lE zyA`4K^;PsI4vhI1W9L8wBZpkP=x}zf#3kgO!CH$vjaKT=F{K2FD|5lh!-h0A2_^c* zUiV4mQTJCUw^c{zRHcoq_$aXNTbm)1jxOb^tBtwKe>Omvo|4z8UW`gQV&>P@!9%-b z-xb#_KgT$0-?a#aUa2BSkjJ$@pfKGPa&jg{)-H%%|D9yn7?;2;2oW>ZK%~9DaJgpC?*=NO-8CM^h9iSQXo)TDSB;kUw45?{0Eyvc26|`P{#hf+ZYW!tIADg zj3CGm^QWSPv0`#`h;cr5Y7CoRh^{3{{ocNj5krI(!@d=xr+MP?sulV8$1gQ1Phwnh zv5-z|RPDaRtDN5pN)VUAZmQ#qywF+f)3P*Jl5@FfVA{SgtJOe28NMtrdgn0I@iI-J zFplM`wc}$#8%DIgoFmC5Z`iJG%uDZp;-ZAXkIVKx4?9Wg4+flu?h1q~i4l&NuM{=e zZ7?yO^hvysp>fG`HWf;EH6P~-ql^_LA@gQ$QVq!~GDA1F{2bJGqhDjX#I>FFJ|3Lp zj{6z+cn)qpS<6AqKLYYiv8!ma)|N|{A<6j?7sXH-E5j>8;~Z>TvR&W;E#K#1h&)cOU2KzI zKG;$wUxR>T0_w7zhM$rr5nD7{(4>s!nSX)}NUHQ#D0hM$d}?nEk!76g7{y<=_;h4r z`Lx?B%|50m(80dI%ot@V?o5ANEY)EwCD?RVZ-1QI>$Pk@nW9p zeUbd~D1j`eXKz7AwW9oAX~bb6TB*{%TSLSa*u!h1Qjf{w)!xeX0A9} z`wcB+#H1)*>pUPFUu1tOV8&||4}Y#pOg&kcZl%Ex`w<9P<0A1iM+r=52Ch3%FN=V1 zqDyC@xy*sCDb$795!ddV0WI$J!=7ReVeeKZI3pL+sTEm|4l(cY*U!yDmZIPH<9f(c z{tS!OXQ;@*B!}tFtQgwNc2kx4l>j(dM%yn*Lb39Z zTfO1$T)EFx++N*E4y8aj_2lW5=E8cEdz7#vx4nlcHOJ%M{FPWcx_Vtd#$Dr-*2fcv z&2LcoV@%)JR~<7lh(;N)p=jn4TowDU3k@jryi$&o|5+L4TwH(g2Lo#l$OT8b0X!b$ zU_;+15`Bs5)xJGbU_RL_oVP8^yxP$t@`|IjI$?4$LB}_}GkNjVWHM?kJo;15wFK=E zkZcH1Qv&d(!m%zfFL?-~@}=z2_<{HrOzN+{E?|RQrp^G0PpE`}qa?cB>!eB-pP;IT zy;chQ=tO}$LG5F{0p$^gtPbi=g+L^na)-(A5J$KH8N%;M9E%tT4R9B^U zr@)v#HDyj0Tln8Yk+x2Be;%Rv?axHOpTHtJ8ttlN-_Lwyg5LI%S4n|9F8$8;&9Ni6 z4q1?{aJ4=c=tw$f~Uaa0m4>f4f0Y}Og}tDf`H#anHn&`f@riCRCV z1g#o1pYJX<`6tpuGE@&W__+?SMKhGi@5NX1j`?P-xb})7%ST`VD1|SanvN7cYfnj* zM0pusgCUe9hC99B+@2Ox0|V`NDMs}Psm5qBSxAaU8SK*HIHtLv%uTk(<;5JFOGIDZ zIbui?qqHD}Z;-2`$%`w@#+xE4yxEDjR5PL%;?8$OX(a~ zjm5Z2vuwUb(fp^XYh3(iSzU(a*&%1-<*PAU8lfhM))}OfKml%kl54my3H;7-h*z%I z^7c3RFsM)-_x+I)t|}pb#Vg*!metC^h-Cf$UjO+t>)#Wt|CbkL#1ur*f|hz=?n|CC z@g#1+hZB6hB_ZA!M&dypQYV7@BZB_rtgZ9!$#xHEK;c@;1n${B4}nFs8ZXrYkxgRA6-_$t0YtWR<`y_GV*0I zW&B(5q;KdaQJ$hSNz7GifA-*_@5&!KdPpA2L)>jg0=UtKjoqG@35OpFw80fx1Hyht zO@?4T?%h@s4i}_!k=xu=tFO`VEM)@y7<8e7m<2jwy1hZtBhMTmw9Swg{R>(_671!HMnne8yl8(?M|!h7vqBG(wdNn9XVXP15-Y zOH-}mn+KcWs?R(8lZ>$wA#C7>S5pG_pspFyG(}D*vo^oy z5288FrYT``)mb~N-j_+Kd5*tTfrE|s3rf2d(NxaF&J?)pQRp)G|Hne~zX>enzs>$X Dcz8+> literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-6-Heatmap Visualization.jpg b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-6-Heatmap Visualization.jpg new file mode 100644 index 0000000000000000000000000000000000000000..8615f85548a6a3a551eb9271da7bcd7e251a36ff GIT binary patch literal 30651 zcmeFZcT^PH);-#Yh=Pa`mDDOK2r32;3DPPeB4Ue37FtCmw1|l0P>Lc7A_4-6KugZF zNRrqJ5|q%AGf2*%X`plU+j#CdN6&cQyTAATdG{ORa>l{hP}S^RReSEa=3HyDJJ~-F zo6numIfLNfKp;54e+YIDLKCrm-8$}dT<>ca8yB3`74StVUyM;?&$Dxz!1of|R?{pM8{4gYLz4WP^&%&3Q7Iq!E ze$RIUkBI2jZDPCk$jI)MQ&c*td`v}E^R(6(?Xx=P3@#fQ8Jn1zS>3pK%i6})&gs6h zi>sTv$D_x7PyC-g3kVH+6&?}!`b|`PLSj;KN@`mAhupmUg2InQ#TAubs;X;hzkX|O zX>Duo=`~#lKwq@E3ez@iS>;e08tax&iVu;n* z;YQ9Qh38+MKnQ-$`c~D(RCgBdoS#*a&pHYpW8yUiu-xazvGQxybMJE9rpUqN|8(Bm zps-~*NWkJKHmjolq^h38^LIHGb+yOVE~<%rZ+Eu2k3Ibn=hFUidyhEkm`uE{ry(Dz z@`z@dW~yCO!&sW*9r0R$bZrT(C%jxI4&OcdmAJMIgPd|TGxA%z@h@Lx z6LL3q3%l5!dzIdBN-g&spH1(lu+t9`o@bBBIWDOOgfT9l=VYcWwb+Oav4v~|Y6BbL z6UmfiBOVjY*$CWYHiDoF3;Vaf^T2d=zL=)dq)J#RG8DmOyc!8pWsg2V+DN}I=AD>R zovNvFJ394vaG4cUVV%WB5G$a8BsPKvYb)X-jq3IYRgAo}P@j2O;_fl2ie$Ic9?gsm z5uRevov)(psqh*$qGF*Pnh?H5BT$>z2;NFlHsYP3{}>xFtOt5?&!Y+wGCR9;lDp)tT-kO08)o|vju2v*N90O0 zt{2^2nsR<2HpD)?t5=Kj+0HkTIxnL)e+pQ-9n~LxQ;{gDn4on*qvEKMA}VhL9`eb9 zY`?O;u&}5-&!akGq%gT5e#5~}Mf*iwRApt(CNoan?Y}j5bawL5{1sQG0-rVf&2-&h$ItJxz9i}a-(E&iEyakeHJK=7HbM+F=!Qit z@w{gaFtK&1Y2%Sj1cm}P8==uL_I*Nf?!Whr!?Ji|Qx!N*K@ly6}pYB#YFa5=Q1Z@l5? zvU+a^hkwEorxfwy{Q7_NvFVQ&DP72@=368i@$;7Fk5C}7{6NpH9~OtdWZpZxMwk33 z*la0^mSMm~yv!eGxk9Y_)VV2iVQSoc2v;P>c3m!>Jy-*;b;s&N!6K$Ni7&Ea^x zcRRsp884rfg&JPrr%we0h1b=34h$$Y;1M8 za_K^K7*}6@rl5M*(}1th9QiNYG}^MR@K@n!B3Yt>?$*$BVEOfVX`N-{*kYb_K89>|~H)MIDRQz;P|$z^bC z>vc7j{8$tP@<%T2?DW3LMl=N(U(hQ<%cy+580UW}Nn8znKKI*%vkMZ=A+PPAkF-ZQ zw|%P;S3FqyK;p;C4fwsk|Fu)?gb~hEHliaBN8f;dD=!y{wVJqhez(}d+YIsU8t|L{ zr;%8Lkre&)ZnF`e1~%iFmpdxP8oQCXPgiK3MSUgNXogbfzfhnC=J#e0~bRC)d&k+B9% zPSa>g21^DO&6vc7O|{HYrNB(9M@5p@hzD+we_VEAq1gZX_FLOj?w}df#RGY^1afx^ zOgAI*Z_WHbjT)WhEZuFeZu>b_Q*5+xO88Y;M!r%~F3nfx(N?RR)RY9zTTc^DUU0kET&)5 zh*Y55C0P>Qplz0f?4^9OZ=?E9;o|%h^r-fso_f!e$EXqgX2NtI)-H!BduUMaAkx3p6@q|q6s=<=s#>c`T3YO9{76NVt{P6)zU${ zB~_b^*bZQdPGcjwwA}eTC>1Cg#{jsH2!E452W=S{hvx1rW8O?$*4Mg^6c8p~NMR#( zV2SiOaM2cgQpQ*e8a~y5Tkc~cmZVJ$C%loHn1&jdG@}67O#U?0fhUATH8ZF~1+~O! zIc~Xap##>&qUqx@@B?h`bbMrmmkR_hm5Gj31juIbZzkaak&8+_;O|F46AoAQyJCKa zOtUfE7h-uG)VG(;rDw}VAm2kBTCg+UFUK-B3anc|Z=G>bes@tKgb}kmFuSS%MoE=S z{qWfj)#if>F@bxQ^oH@3YX7@d5I$7~k=(R7W;q=YIco3~OuKpfDKaCe(=5 zdpe*P31az4MT3sq3~*lsm~YzjwSaKJYu zokuo4N8|Yt8}VoWGNo&>5xE^{#t}>@2A$Z9WW|=9E4F3IR9d6vb7?+jW#a>K>oi>eoPXqac0fidq17{>h8mREiAQ`Eq<=le9qd~U<}s#cqD5^%Q;mz%r7f=8tI?cg7K~FK zV&p@eT|+mbn~i2%uL0tap;cqy8Odygm!<3=k5uot<9tXPwB{A{8q9}->i18Fbk2-( z0n{@aRxen}uBnnC>ROZqIy2?pE+ShEW#T5UvAS)AN(+_}>Kx9MjG$5zY9aj#W4$%m zZq~sYCqDuB>_K>*%1^O;?jz)-1r61;1)5kM0n3b05Ar@$+g)16X55hD!+WW9+0IE#+6HX6S*`t|SPfBc}|(1mm2#P&kNZzG%lSH=20LF{Hg83sxQJqLyz)9lRh%yB&+Iw(`<`;GAp&Dk83+TcajxOH8Ytc~^ zOcR@BBc>ZrOUl_IMJc2s^B)E%=X@R-m_ao&$mJ#|M)U7Oov!_FmfApPt;gbiV?EjQ zkH|pG;(mZmLu|wZzX8fGhGxy^LmAY7yD_S?>#oLY zPPuvhNSpEn=!i}@j`~3vX!m&u7KG&~!%+fvIh?!X&qi3|FJ(tT8y32Vj2Rdc3Xmgn zkr=}?2NT;T3ruV@b6Tvx>$hf+)c6Dy`cc{BbJ9Yga-N3FbD&el|r zEGZ5b9mD@XH^JI1tpFd9F@BElq3=q%gYJ$=+I*FkCgR(Egkts$?+y=tug&##`>^!K z?1})ya9)Wdbh#}rqYtB#)11(v`SK-ttTeOMz~&Q&LG-Z9OS$nEB29TFTBFg%`mDIS zp@7*&qq2>JLF<8itKiYX0VJCrm` z1utI{8?l}N5o-Z>q3Fw#Y{bG-6EFomgBzfVVCw08(0EzR7-3L%>h%M-w~_)mj*F=7 z@P+#w)RiOU3XU#+CT6*yY{0W#&JxG*c6E2x-3_~7w(9_g^xB8GbK}gFCE*6k@N@Qi z-(IS!6e)_!3S=XGo}Ms^oPqN-i4G^VWBYyID~@c-U?UROL;l;~@W>S_)_il`sD_KO zWAM4)T8x2K!}XcWi;P1sWm3BhT`X2GEvu zp$=xC5pJ1EpmVko2d#ZfVG6pH)GZrdgf8QuIQXOfighdnEoT?T;nU{)nFRWw3^Vgs zDF~h+d?D;{1|$VwhW786VgEm12JSzDKnN43RU1=85*4%f(Y|v_Cs&?{jX4Q9#$C_& zE}*mH1=wsbxx6EC!KgbWfz>|7VSH8UsQ zn~T>($lfe?T};an84zE?vG(@$^|fkfMKUw|hslPe3w?FA9+%6tqD}U0yGY%2 zWARgLV#mpGzy*(sle5juIF@xxpA7qru((`DyRM`f-b-x_C)5tt+`dBbX&a1@z;WmXSjQt`8Y5Wd_~%F`Q-aV7`?Pe;xO35FBc-`{91r;*kVSo`3JU zlRs9H-4`LxrF1o+dqQiDrzsawg2yrPUxbFm-j5o6r#DJc(k`&-U@u%hOO`wWfZ+1a zQ_ifwBT()j(5O2)%J zaV9>1{rp8f`2QpyZKD2vbB|g;z5h>w5zJq~$Q`eX#to`CrXLp`X1yx+01@O5xhFdS zvw;MNC8K!>^A;?+dFqBVEige8;F!mJ#zQLLzdPGf zSUw8ygXJSs(}4)6tYul<1%zk@^QX#s28qn41xQn9$;$X~M~RmTxLl-iS^<7;*B&wM z50<+U1(L$o7(*d_bDWg^@6iW#a{q7PNkLA=cY3T8Rpvj1fZW8Iv9h0RH;pU)7(4j% zNq@88zTk%u+c#A{$2pqi$2cdk60Wr5E4{Tc+tu`{CuVw2FFE$0z8=EJzsk&_@hu-y zKt1HElVqIUTT2y-3NGhkawny?cc_u?>)Lt2JWHX)`SxI-Ait@?q z2sJtHkQJdXCSv|!mx0(ZNy^oAv$1X&Qp)*J^IA`i9k^L|?QW*o_QxUlA88MT%f;eT zjE;Ejv=)`k7f$niqV(yp)(13<9gLFGF0(-wkmvB7Gu}Vv z_>gV@Hz-q3`k0ff1Wnp7B_VQZ89Y8#JsCOErwMyyL-Q?=pNIb=hyDA!4%s_+I&qaw z5sOu#r8nkCZzt<96bE`)T8_nxOw})`;~UqIs66B&SG-j&o{J81onWE0_kFsIH?59$ zG@708z^zn=vk@o76&NsKSo_z8rP6o}YyNxF&+i>XZ##=%PsS!>J<+H=-{PfJiUxO z=j~E+!~F>LwGr|;VFcX#mS&QMQJkwYOuU28{rfQ!k<$h(r_B`{U-}q%KtsOkt^d* z7@QACsauw7(h`2Q%x(U=oKdOqPlUODiHd$fWFz_`U@^)oR96RdtqXeQbABo{KA@Mn_yWMvc1dlxED>O(o>vb>2a8$cXoaS` z(X6!{4f%kfwJuOTqb>{SYhCx=1_XD zu#WJW9~z*EFFVlb(6~uk?yrqn3QD_@jaVlYxXANMN~wP(fQa#UQ3CwQcxX_|3rW!B)ip^w)!B2>io* z1av!NoQDMwzp~YG9B3E3;%8Ugnt8?`ikfcw{XnTDRFJvriM%arXzr^SMA%!4c#+T! z5Q}=MhRFs|)(8^*Ry@lKKJT%EPN43hM$@Q>#x2&HuQJK}_^kWa-m?)VxZS&53Cm7_ zNXrPG&@{biPa)Kxg&i9~CR>3Jjlf0-YBS}tFtE>UXuc2(0UHNC?K=pE{NXbo1ORc% zFD+~d11ZSF1+WoEYTW zXYFvypPHPmsIq3fjmO+1h4jxd9^lQXfG2L({YzTi?J-c{KwE?PLt2%aWNL@SKt*+J zKGyF5mVEaz`7fLKU(aToVCjy1iX|ny=2?5@^>gl2ro}vSkl_fFR z_(F|?^1qoY7~S6?A~)?j{eToT-HUK3OGw`YchQ;fyChP2JBI(^NSn=-|0CXMKt02GN3apa21`6qEimdT0KP!xl-MuF7rx#mCIiwijZ)(I zhri<5_K$ey4Rhao9jbql9;*hFmA#a>-^7s$Y+pHeQU8)$vARx`fIRg@QX@wi(}6)m z43?e#jfE^l{DY>Lv`u)U`Mm^}PihGn$uz~{7Ch&tQD>nQk7c>VfnTi`J=A0B?Qzx$ zr0|tXNZ-iay76}})?5Qbb-!4%!A?tFsD*12gzedpgA`_Pw=&BM(!A34tNr5G8*RKG zGXYem$tl_R;^4-LBAk$IBt`7Sn>3Tj_TrFyop?|4XKOgexGH^5ZWDKGNf4XNh$ZnA zK0T0s{P2F{MXdN40ao^nCo1cYU5ZBC+3GkY%lWp&OD)sPY@%4ATQB-zr=|9NdbHEV zqvDP{ulqiZUn6cMHshHWjUD*R*B0s!dqhWFy18qaONvpEPrP*ZMLFpV?RA!oYu%q5 z+Lv|h>1`cOkzRy4NA#xrCTxp&<@F$^UHWk^5jU=`yX$Q?*|uU}sKzawHkX~^Ut`OA zCRx{}sdiNR7}IvZssPOFy)@%w7U3M=7$ph}YM%FBhxwDx9sCFr`oYUM3{+dr`tllk zHbUKCF)NVwSypDV>@J((wS$*`Y{KRslgQj2op$Hz}F-q%_SNrrgWo^xZIZG>W`G zp*hRj-@u?)*ZlpXi2dSM`8`}JK|Eqm8u?}J{>22Iy{YDwRmT&WgA$s5^I-{xdYJoY zx7rTX)QQ3kmLWxDdhUcue1wIzx)%@~XWIIIn!}zSWt}dfJIETG>Nv)IAmtyp<6v-R zY*`8I_A`1~b%>`|w8I*0_R_TWA3ks$lcZ68FQuA=cENq}&G7; zM$HAU@(n=3zk6N%9ZdeIo>*V}FSh&#V@tMZm1CIqpVr>q;c}j$j7K7a@zLTT{Sk_1zb2^oMld-c1xy1 z*_>~3ze_KpG(rFDMnS&NZghUp&6yL#4i#b1FOmY@&`IC+z78+Rb zJ}2J)F_zZx9CyaeIz)lOCv}?8tJIb9VEM+0c-?!5YO_n|@c>um_ToayEMXov;yh$l z%!QgZ;FVlsBVr_sS1j;>Q}2FDE18 z>Ph#icZ83J5G8|LuO3toG^|K!&L=7>)enm8ZL$RtLgvCPWQm zPif6JXs!c|dUpHH)zG!3g|$hda)&|>O`G&iK5mdf=X-$6p9XEN(J*i#3jsbn81D?QGXkwl(Vk;@Nqda{Eg3< zani*BIZ@Q^fiIC68WEA7ktZ@SqEKy8+RvO4Q_yf4JdTDxiq%iVF-_$^Qb;wOl6ppIG~9B8>~v{>dYyt6vJicP_+S4OX3>AFU!51Gzvgn^9lXn4)PQ(aX3 zo9_N%1wHz&74-kN6=W=*WnrfrE2}3$Gz`?$$`##X#iO(9OUZ#Y{!hGxGQK|7nu`-! z0*3N4B+(WTyu4WomCA*iKjrEQ*MA7xcIzsShh@;7RgDru1h`n3o#9tcTz2HP`LyOpWlqc=a?~hEkJhA=K?p)vDujJ2jNbZel zH?A_A-|<3oddyNE-qOJs)T7uo&Ior=cIS4*y+&mcYFe13H}Hbu#4sH&pJ&8)5S(7I*q zZ%KrQEO{?QE7hFmD?9B|3!bB?CR@OiP$A4y*a)-W-&&=t1HA8UN^A*vd6Jpd@8rP) zl2&Jw7vX)wD{X2q6z0D5z~Yb?pfK2otSc*N$w7sANAA32EHnB z3WRF}xXcxK13j2M)f>3PNf)8UhpdZr^tdOm;oWxK8PLf4< zTRg8$`7MS&%~)l{d0FYl7`5|y%C?g;!Es-0E?2gMrnSA&XoSk>`YTSrnZ~nXy!DXQ z7z0Z^BKEc3G52-op+-42q8ru4vU?__-8Zh07whT4!|-~8;o`@Q5g#|kkXG}d2rI!W zkeNmkOq{-~zOp3bK06`bI;x&q-tyKZ=8j_pLp0qXHF-X*jfw z&z#27EA7gGXy&CT9P6%9PE<9?ni&uk$#PT*Z;X$5~BueEmF#bS5MhFe^=BF!I`w~D8k`H& z>6bJ(rQfvU)hsx$u+7ymbr-D-L6|jO^UC?RG#C0m@?h*U?UpoB{L44c-vPl^aDs^b zMILr{_&iU@WP#J0~l_q)mgaTn%UY# zbPXi+;1cHLuL^Ot-LDC;$L21`Qw5hmmI&#Gaa6x;PWF2`K79B(<^A46J>}||>i*Z% zK32O~K0tabr4N^0=$@MD*lhc~ztnX_zsI~`18sjIS-d`^h<2g+(k7JuY95`9+c{ph z+jCLk$Bp`R_Vme5`5*q)bAnq%1%W{EJhC)z(Ed^&rtXo{|JHtYX2&|+Q|XJv9}p$4 zB@kPqU72zaZznNXC&M(%<@WXBHT8XG8>`gw_CL9^teaVMK@5ANe}lbBOCpd$$n( zTD+im_}YZcj`^u8*^%YNVwd8l^$aY|qp6zrQM4ro27O>yo4KPH$)qc<)?I+aXmfoK za!m&_s5WqWD>P0YhG8+N=g1sPCNS1P^)GVa&3*i^w>S(JQaf-|oEc%EpP&($RKFwd z>E6?_;%if7LN`q5*+E!46q)4+gdM&e`Go|OO{uTS_#H8w%SR;cswfek)mymT`!?S? zU-w}k-OyRiG{byR1x?jkW+P}!3}X=o(g^p^ff5cYuxRLoY&=8y0_d1H#8{DqYg-88 z)dMgqU~uK-jOUdlC~Q%s4kfx84*Kw{%S1uuV@0pXwj%Va8N-weO!pkGCMjlc% zHP4XuUH|-sMEm?t58hg3@9pPujiOfqs-X~M%BlyRGHv$LnZN9A%W9%s-b#Kg_ImNL z`LRYPx{+%XOh~e&zFcc=UwWeuUGe@8OWr*cX!Kre#8Iz1_SuIWfY?k_VnOHBJC<2; zv8!hMME^iPD#oKgy+@X9itw>W_mZ3MxLIi*0Y2lzD zB28FHfj)%2mBR!aduMhMh!pj1+}DS9dFv168N8SMl&+Cy{~NGpuH*@N~MxZ||hf`?HSd~~>VjO-m(eWGqnW@vH zJE)Ql%_$-$Hk@lMm@v1EnDaAf1F96hk^PVvN%|>5{`YmC{thslA%tlad z!na$`brp3QI*K{-n+A)T5jD@%s0xQYkU#{5wux z_s>gduWs(P$;kCm(mI!Ui|!R1-Eq48R8Y4)h>p!a!mkLO#RL|o0%0C79zRc+l?Hz% z6B-VJg`nG=5U~b6HZW-7Z479f#y~IeooTNczLZ3mvB1C&BL^rT{H!Pwt%y8Ihy33a z5TfLXXHIR}*k#c0<|UmaPQ?;&P-TEW(zL`#Y@gW{|4DR9 z7@2>bXBoG@On!{~jVI>p?#`9P^oYl0=fLwzSxm5k?!$(zAm9T%-!4F=E7+prjxF9#5sa<0A-uMcVggf}kys>llZr&EsfZl3g54nE%O83!jaDE*1C2$@Ax*0|ew&3_ppbp_IRVG=GX-g;6(*rP-!SNzF zOhbTf1wYXZ+l~q#j2HtS`dQ|S0MwV@boC1X7Aeg^Db1Pm5c#%p;lEj%rC}c3IGriK zwaEb_$FF7qBq=;yBnMNpr(S^6#q6-*ihmn>#_-33$NHD$zcl_|%l|*a@{b_xNpdUh z$0eFi<=AfNt=g45J8WcqQ*Ny-VU&AJ=A|2eG3+bSl|tjfmo^a|w<5h7?ojS!CnhCy zCta3%ph4Q-QK0sn%LK2)CvMrYU;Im5X3|WxOhJU~Q^X2NJLUcP#h&mNTGu%xen`-b zjYn~I!?!iIvr6;}^5FWe@ozbEj597H52NiV*~1QH_d321MOst7qc<*C&V=L1I~Ixg zQl=|u1&pS`{`V`VR%&9I=ftnKII$*CD`sR?MH%mqTaU;acnQbY@7|7`Ih=HRmoa{f zC@n-wFF@T*xSM?Ic}M(JPZ@uHra=_Max`mgxIw)QYCe>Kybdtue8;#OZW!GLTemc0 zN`m7hqiRWLW}sLw2oy7-zLK;xwzi^vVh&~%b1VZa^6WE;mWGVey?{;hQfVHTK{N5C zx%M8oioqZ;%FnQIm0c+ScIE7oIyVs?Fr{(g--L~;jvn%FNAJRa9X+p}|1?K0LzgSQ zU&~Xjx^R6E&)FJo?B;XuZtYPc?3)gC6_K(bdCaW5VsVSrhc}mXd~@ne4|c*Uy+2ky zPevg+BaW-mSJw9pMT~tzt4IqtiQCJEzM~(So;HQ|y{t+4V!Vy~X`I!biG|i52mzC| zHKILjPa4-JhBFiUB;>rqn5b3Vy4%jgV@v6Ttx9SQ@)bs5T`D?aw9PNm-4;hsRZGH$ z@10w4u|6Xbm0@-;wD=}6UuS|4jONsp5QnpGZjrIEZb;Hfih;_zOJhdFblp>L1{+g7m=w^BIbEJnFBRri?ZdnePSrkq07|ax)R00`{O&iRk{=$$ zVSRe`rzmgO(R=wrM(U^zf=;e7A$hloM4d-FDmw0nMl^ofID2Prb#Q@C<4V!b5{ccO z65K0C-vGC+2gO(*3~K+{Fk&ML?i1KLS52fGC4dMNU3%Mvw++HZvunn+uV7u}Ev%wqu+Dlc>KxF5B89r=&H1 z?}1GFq@u#RYwCDJ4IXfOmRcT&ycx)A*m$ovKi@LedXKVV#DORv_&b1kbc(Q24v5i7 zaRLJd+SIR&i_o>Hfh9f=4iWYP=JHEJcorV7TTsWlHMc0~ozgREkb);el1~^WUtiyQ zY}zvDBOT5kQ@sK@fqE-bJ!}RJQ8Y4}Yl;z`jcC?kxf5=H1aC zWL@gVfgEV@7DgpG8*@X$=@m7=NhBuL#}+V?AMcOL7n8p7NxtZJ_W?ED!!Ct!c;k&( zIy=qfe|YFWoHROG7?*(c5bHhQk>XDnv3w8GYvj$G-@J2t_eWS{if?p8RpT<4#Eb;y zG5QXy{1?LF{sF*Qq|M%B;spSP7J`hTD?!&ms&Pl!$X2OWS3UF}O`hRTGF>|PaMChv zp%OkpAuI#5{SdIoM4nsbe}Y=%{udG%<@p?`ZvdVmrgqK-r0{5*{WlCH)tjIHulnxb zl>Z9b{~5yeGN^CW9=~?Q^@aJ_st=ZWshS>5=cK(g1m&NeU86VcyAg6#k$Rn7otxAg zq35&TprN|QxWJ}h{O&Cy?&CRmET`EM601v8?wOqAN}w-yzjDmwJM1~(+g&?&_S@!l zW4t9?CDy{8Lw(Vda*)p4`_a+kxr^1YPMBaeF|sftZy#^kO$f71st=XhZ1N%gaQe8q z_YbGpi*6Oox($0|hk8C<5DyC;bRcCdv^>5dJ1fU9nZV1l27N(eocX_ zz>7`H^|@Xq$k;UBj$%v&&!`c=M3_$pmRa8WfyNgO8AGU$SXWjscivI3D&MnpS)XN!`Ky>;K%_De3$Uz|`HzxC$1r0`7C z9X4XjW7{o^>T3 zd|m)|-Mj@98@oJrQ)#!Ez&)c}^^L<^61&e<@1H7;Ed8@AoHjYH4NjWACw06YFS2E) znf*8yA<0Qiyt!EK{<8?apB3Sv-yH0wlC#7&jamdvo~d5Sur%=0A63pbxw0h5A0$wZ zq>ad*O0{9ytULhs^N?{aCVB`8LIFqYnl{yO=~O%BDaiY_AONsAlSel7kp_uwE)2Lx z>`Oncx(vNLK2qpd!r;(PAY=kK0=`xm=x9^_4;Q82X7TH72L5bn#hERtd@_x3O za-6mByVt)f@^V?*?2`Ra?AasO-92Rugy>z}-_jrIY~6MTZTb4$Mt)vmYS)3ct|IwR zT9@V2;jW?C($D)&sOY>531>u;eGi{GmPLH3u|2TTW20zj)mTQ~K9hdY&`iMcKiS2` zq`E!n%iFJeu{e09J2OLYl5lIoibZXGiTS+1(Faab9{QvU)j$*7*rT;$J8#t;R4+uz zoXUGQ=X^Cj>1lv><5n;9!w2%oD%^2`{0o31_{H$E){Fo)P|1`BT>Y_2Csbz&p%)V1 zEjV?#vO8oE{)frcLn_OnRh~9`j&11yF zh?^-5vWLjsm=;j=@O^%yS@CsdoZv*7CY}b;9x^ z#=t{jO6BBjNrz4GFScvg>TlWB)2PZY#OLwP!TAYJ_$#{h`)z)6=woS8Axb!IBA)(cQmx$hq*`N%B^%3 zxJ8~z8Tvldm|m;q#6{88ma@Dx!Cgd9JpR+lH=N($ttWMg9;NbNaw(@!w?OKt+?ZX~ zgPv=MrISWSFtOMFz#3~!s{WE@uBp#X6bTLTfuzk$GY2-42O0D95fC1&B8_Z(OhHh- zcJa|N~GwrdM?|I^Z?#||~ug0`JK3#3PLMnN;OW}NMjQp*f#9R|E6K9K)5_fy`vppA;ro`QP;huk)=&gW(g`i1@{FyL6yeicGee52O zrGGtv(JwNZTU+m_%PPx&zMq?h;izROC$7u;3uH5dieb$x;i+eq2(%vq6cVBOBfnsu zUg9*ltE-@P^Ik*gjS16ekUje7W_j;NkUfI^+n04Gxx?gguPY$Z%Z4q67Wo+UxwHF|*W zjmlBP)473!n`1#R3ATEx**MhhkQlxtiOci>YWt*_29^$ow+&2+-8OTs59#blNId_o zZtL33svTR(2&Ir}b98Il_cu-r5T-eJ)vQkW`O}|%wwZ~MavrWfuY~q#D+`PMibl8R zdL1E`HO}1?#SdHX|8&ls8jW6NUR*S2Sx#6z0Z&_X%k@W7DB7d?jUorkNZ75CYg)UM zgsz-^cfU%boOjDls^gP|d`@<9pcsQqsAS_y_{08;*M{UjI*SDKPHom6jVpevCz!0$UF&%x_RMrm~`?#gu&PNKEOpDopcE8ZSFuyAav6XNgf0e ziH|v$t>0Qv;}z2|vyf^G#Lo)zRRxs#vtj`_Ex)vcC!jm-3Z$VNuA7)zX+0B zc=3$IfkCqZW1|(33Y{->z?Z&$H)} zPEGsA*~(VKLxVxLI-iYwke7he3SXxn?xD>hfg9YmuyiR&O>N36Witp5E5ip)c~`dT zfAk1=*65}6Ec{{;=ggz15@chTiXwHKtR2H+mXmr2I{RI*p3tk%34_M zVfEe~RvYt0Gg#j$NB><$vz;7LvyzSYb5|_s$;SU(yJG!TqvfXef4eOf*dt5ef7m0h zpgi;+yHtQZ^3n$9eSK18SNHlk((=nro0jIk6mU(ZEa3P6JuN4W(zdS(>-%T}RfI+5 zse$DsHli^n!}_;Y$ed5GKfLj{=!kGLQv<}f3{Z@W51znA{aV3wK`kr*UduxMu~Vi# zt<#uySUyP7nW1K{Ss`s)6f2>5cX~6X6mk(erE!NUmEUKGoi@jXK<%h#$)B@2GiCVd4BuH#le0 zC=zLm;NC43B99Vm=}J}fF3SeTUCr|6a?xXXo|oma(Z>**ij)j$#I|)Tnb>R9gnSsh z(U?iOsy$X{4|qiAF5A<*ZO{1!wdU4#ia z)B+Odh3TTI1m;eIg6CHm#$k`f=DT7Bl+PnEW^;85@>>imu)3H<_1>1l&uvmp8N1G` zv`%^@uOt9%ViYx~j$GP=0-_BYk&^%meg>NIFv#!&x`EgV^xkF#a{mEghf2$bA)}xt z!5(fypl{#x+@B0u7>);V^}SjF!b;Cr_1BWd#>0njC0AlR+%^gw3~}`B9sP1W{koX5 z>ZV@V-o1Io?v1VkeCsBi_dzZwR#X{xokrUyFNJMy4cLgEKM;2)^MX}V_PZ;fYx_5s z-Z=@JXZ74!gIP)fu$eutm%Vw068uO>C10|TusDnAR@#Iew^M7N;|zc14!dC00Ihf_ zInHuYf|PLCNNOn9U*^cApg59rL(ZAKHW|ks9h9xJa~}?%a(1`r&xrAA zil;JmR}Xnrh8`W4CvOb-P|`iZG)|hPh`z_zdL+%bgw%ZSls^-+%PQ10&Ut&Zg#EEb z)$i*R{FffBBgAJF`pdLrXNqXG;$wu3#t(%&_%Wu}Xt?NMmG9s-Z|2Q=+IQ+0C9=qb zG}adGwh}VPVI;`nY9YsyGZTI2kqcUCu;6}8*lpF{0T~Td z##}25JPmN}7r_gu)oJ%yN-eK_OT9fSNrr`&Q2$qF=NZ*hy6*8%6ow8;M`|cirG#EZ zszg8%S}4*&4ZQ{t0RfGn3kV{D(j_2}5NQK~B2DQ;LXkT3qDT=mbM~CMXPrB<*17kt z`(dwq%HC_e+3)`;@BjH_`oyc9y?+OAf+()*?H2&4O*e`1J%}RR5W&cExJ8-AQB8xaC$~ zs}=6SZHx=LsPl)IYW%@e{MkA%p1zNr1`;F(pb=b3;asOPZ=}(+r09NLhU3qi^{mEG z^*5(;#=Rm$0YNh5d2NkEna;Ctr;JH zr*ip(b0*v!*xHOhy5JYT`yObS+r2=xFE8eEY}!+{flLG~xV+^mRtRgB2Y3=Dl_`}U zph^~_g)jW?5bd~M)igeFxMyMcg#+_lC@EtA7pO~EJF|&*$0C|2`3h&fr60_~&hZrhtj@heMOEKM$oj1@q9r4(q?u z0D41GHk}gU(B8V}{uGn80xMMS)408G>WA>~a8~6hPshZ}f{^(=#`-4X@LvqC{pin(cK(`345Wl~0`)q&poA?X2uW0z+ zb~$L|C+zzY-b$Fz^@!L@|C>v=0~k*K@vq@@PS|JRN!uQepeww@$gW$8r;|1RKb{|a zG^x&{S_o(V-)zJ=IQ()rK=$Lbv-%aPJ1r58WLB|SjB}o(`oV|`z^p1@~Vy@71S zsoLICtw*$_GbFIWzx!+ZY3%@+6*(}RjbaJ-}1vMVi`3&d{*!YqX|qn z9^s6uim=W_-VwcZI)E*<0FGaD#KVnm=rJ}ZIDh;DIdf0hyu~N*n+($Dk(t=LvVxl; zn?qIl%Q{;wLaZl&ztOKHnT(ZTj9Ai}8!`us4d^(ZEa7g0dwQn0*(55Xct6VyQ9Hhk zR#HVj+pQ1VW)9rovL#2Rd3hnvc5sWP&*P?NOKu4q*Bs^SzVqZFtLA&!lQ)`!xqA1c z&pG`JmiA60t51{ZkE+0nRMi{Vg93`$jjd%qqy(}p%MFj8PMZ-L5-R=C=jLA= znW9|Cn`=tav-aBO<4ULvBKYrvViy?Z&LEvX_8(GCID2M?`GKO>paAk{a>_ySD-Kq9 zD*AZl_-=)g;@8{LXR`vNSo1#lfi=rUcQZT@e5?qBqB@nIa;S1OuE=8jtDXm<&TSzd zLp9y&NQ%p}6LlVqO@y&sg>CUigs(IGd`+TXFRXbjX1LxmK|8&C^?h}=?YYysBjquX z?A6ZUnCJ6G^-RMi@ru5Vk!zfSWfp^(n_pkF^hh<`w>rckVd7we4>RZnsi`Xu@9OI_ zknk2ap3<&2oEe9Ztp^2rfOSH~cWVH(`v~_9;C%&tA%^bOC*549wNsKk8#*5vtR<(v zB}rZFVn2{(Q=+fyel2uSE+xBlo>+O3hk(MQ!UCFl`@TL0-S-faj0C-RiLM-IV52oHt@tQ7ViCE#r>tfzP zm-^%bZwuAFLGsM@dnxsDSG%WzVo4&4GvlcEMIZc5B+4iV_sL~O1+&Fb>%er8b#AfC zDRvpB#A58!!}V)`ZQDmnJ=b$jLhZytadzg(k@=gpA?+ zj@oXEyK?vhRmCj=iymL#!^#saS3=&>`<~M~eMldH>lL*DbsfgM2FnoKl8!^T0;{Xv zwLl+H2?*t^m_43L_zpxU8J{<*4e|02nbWNJ{`qXIVRr3{yRQT1PEEY>SwE^>t91H? zBw-YPBF;-~inF%ABMb+kGf;YHdjKV)k(rT_*JZQ$8*9pX4{j?5)o6!G9k}OgH##$_ zDjKh3%;8HT(?fdEB%`Bfy)|t7abLa2gm}i8q*JRiRLtOIOFU~cYw?ox8@+gW;~zBX zCbYHNCJTVyk=xGhU`;#b14& z?&pCsv-?RsGy}fB^A4Y%HNBV(D|QMnF=1+JG~uODsZHZ_eBfaA^S8hpkA)5whg*aZ zm6NsmbkTt-4UEYFi_NXn25){Nm?Uef*YbulORn=&)W5(~I6Ov> zyYRNz28v=BWNxQ8B`$p`b>Q}A&F#|mGKIX_ZFr&e3TAa(SpcL14d#QqwLA4rmkS6i z&EAZidBN~rSav!ABzCE3zMx>#cOE9U&>D&hS3UAO=Ua0NGro#{XdXQXsc(M3*U$lv zl%A@sid<0TsT;?pr*;^GED5pte0_3R#7CrA=7MRSyl`s5T_^c>0_i85hVY31kCXi8 zbnQf&OSxoqUl_{Wy-Rx^s`gH^mTz*1!qME8C?ZN#?bNqGFB-JqbeNU8!eZ%1uepXN zE!&-z7Uo2VvAg4lp-zwLw3Aw$( zjzvDHKt#*H+hf(k8KvrSwLeR&nX|Q6EC;hBiD`SnT&O3$o&310+;P}gYnKldQL3V^ zSlqYNrpbbTTVwnp$`;{<)6U~qGI+%gnNJmx;AVYq{VX3RBXZ{VY>0AzYZ_H~su%wW zKPZDb`Mfey6Zqa@A#7sulB?@zxYo;9|Aj&DcFhyHf{8%X3t1}0oc9t$naq<1Q7F39 z#dO5|7g_Z<9+}s)37VeEBQ~NpL!$!L<$DXt$#Z_(?1&ydg#wjd?fXGdUH$3{H+`ef zkxdndO;ytjRPd8!55X~$5iF+biva1kHLnO2UVRAV6ApADLs`I&g)%Sgw!qJ3y(NRn zg_ykaf>Oy=*;A&}Muh8)&#rjyFOjEw1mWj8sSbbsbvCEmh>aS)`eW@o z`tYdS(|4x^HAMN6rI1)L;rS}5=Zg6k%OmgTZ_cAHlM)&4kwL>wB%`3k4u-Ksc{#kZ zCU52{IDL}y(P`p&qY(1A<4If~%U&6$@BONy7E!%kLXD9sp|6uGX-_C=Nz_;j z2uz1DRdxzb*rL5sik+tIXAs=Izk?*ip`SL2SkQ$8BM@*?(9PCSeA_gZghoAmTjp{5 zl}xj48NA93Gk-CKXG}d41v8?8p+TB9!2L7t4#4jUW7%TIM)f&y;WPYPNPHua=zZ2J zmCG0xEdAl46_wFNE7}l)iww3r*$`R#dsbB7m13=&o4|uO;rJQ4Htry=0Fqc)ueiM4 zg8hfE&un$#&HF3N7l|Qzoy}>Kelc)uiasx_nhqTDrshG_o05}w$I065!(pYdPlKzQ z;}!!aw2M(q9+m1xB`jpj3gWGc2-M~C^*F))K!uW+ohLg1wvelB_c|p))#~Ryn!WEp z`rQ$kfG}FJ%)asA_!-e6C(-QChiwH9gnn98q6y!aQK4Fg&~mBfjrGfNy)ptMiG z;m=_0*1n1nxjc=@{EygX@`dh~ury;1e*rdVe*(@YA@oWPB7~N~sK3MmAYYzs8~9KJ5b42#eW8s^{dmoVkb1O+^2!F%OP> zp?z6oEvfL=%IJqnEf6htaSUZZnKnx9m8Y>1_DXPA(JeF|c|zN!)-bkSJTKoL(v0OY zb3r7gNk4v@3_GW%s!XD(tOs36RyBQ5sG1J5OH1Lk_hjFddEEF#=UZWMVLth`>XBoa zg51lp)iP*}5C?M!HEb<01RFNhQ2VRo%-ozqA4_4t?k*So;`I+CQ{Lv6_}CB7e1!#&* zhsEY`&e1j7(G|K1sYJ@u}1L=2|2EhbnuY-GwjdG^+v@eqqt76Z z_kF3#)-%pcM<+C%uROWY)nTK)s^=)CV5xvxnt#r*ajg|3I$ z;9JT+O}A`Fij^l1VR5cE@Dty*VxtM_f(T?v=ytNcdmTeUk-)yCMJ&lupR-tb;z8;PYsusmd zmC(XrRyEXBJRl~}<(!$G2pWSBrm$T91mFJNNc2DfY0l& z(2154UNyb4MQEAIO{SZsT)9huyJJAo<@5QKn@qS^H~-LqD-}NQKEY|uKny6IF&b53 zFSnR3mFeb@cC|{d>t+v=?SAc*YgyNl819$q3wHgXTUGL@W#xS>&hq`<%JB>hasA{)XQ)Hg)U?_U1K`QKb5OuHL1O3TE6syhHp$>B1eov0f869AOREj+inftbck)kn12}76o znGwUnWW)!$eBxeAigM8!HYOhD;S_y&g_tZn>6vIB;Tv%+*NrwM4$$yl*3~G&k$~O; z2Mdc#h?gkSO*EFseY|r4GGTqc{RR6Fg1ZQ%p`DWrb)GEr%Nc>-kC-h5vxs>Q4|V7-u=j=maffj{+6;LXbm}YQJ6Sr>Ir(#3DF183 zny#zsog$mV45k@BH9p+rM0UGcVK6@0-L>c(5;YKZ{A=KOP2k&1%p5az6YXvlE_2W- z%b-ex1x{`*b^81o1Bf9eQXNdZabA-Iuk%T(>B#*+Sa$gQX~2_MWDKSW&Gqt9?-M>v zpB*&c9-b_ebc(4qjR>QmkKOaw;|<`$x-)?p5$r$ z9R1IjjbdO4fF>gSW&!>(bFXnOzX{R%bZ+$fop$tb&8c^4gq(nZuXi1Lx&HU`=^c0( z?BnfDWxXik+(#|okzmRNhju-juU>$#UlhlX=hxpslY7F;ylXs16Rj*vAny{4aOLS%gH Zd;)q6>jCs3{wq)TSHAEcK52fR{s(rXxC8(I literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-7-Heatmap Visualization.jpg b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-7-Heatmap Visualization.jpg new file mode 100644 index 0000000000000000000000000000000000000000..c00450da1793275acd37c499b188b3d233d2199c GIT binary patch literal 22458 zcmbTdcU%)+6fYP>ET|w#06{?k=|y@cqC)5xI-x1O2uSZlMMXjrkQSN{I)om2Kt+o5 z5_*%~K}zU>9q_mBy?wiX>~0+A!`wUNo?GU8&-Z-ibm;UC=$4YaqCDur1rX>0@DFr4 z3VH^*dinB|%cNJYT)A@X+Eua}R5x#2zkcKH?K>1y_vsiJ?$h17_ke|$?Ew=v^SyiQ zq8!|h`Gtgp7}><7#RQ~y1%(98ZgSz;wQD!7-=MvDlUCs2y@vw-pMR&{L6lc7e7dkl za)AkSk@5lw<%QE|jEwpk z9TOWDk4i(QXJlq&=j4`@mSM{)Dyyo0HvDRAYHs=6+S`Zk9~c}O9+{e+nVp+oSX^4( z+}hqD?C$L!9G=;A0Yvi8EMWi7u>Z|2O2Dp*moAZ9x^iaMg^M1*e-g?|q!0No-zQ9QL_k5J=#1o?cB2ecZR+H|7F?#FzkQYH3GUp zase1T5=syRw0Fp!_Evy^v$ml$$ajx29v@enGZ=mLwpZl(y_g3e5KEN4$=|QG{X1ig()c&(igxUYSJsLmF^})OV#^xyCn~g^&9k*Ba5ukxy8{A;U@gus9`FPE zW+z)(pMq@LPC-(A1oV-P`oVk(29{RZU9z{EpYp$+-A~7_IPyJsa-;;&B}BA7>5n>l zj+T==CwE(|^8YBtpFdv9NyVLW2n-8JU*%BZpfeeX{C5%l{33my(Sgvl6}Zn`$~7ii zgstNF)3fW#EzjH4e{*d`@trQ2`+m^ZjlU?yQNe!~4=0>GQGcK6rB33PRj5!XxPQ*< z1OtI1{%dN7Y=|E|Ii- z9*pqCP@!H+UR6mt#7OTc=-k4Ab5rya!ZDykqyviF-R2Zzr~F?}!A;J4>O1;k(L&Zx zU!=vxnhjzB%tiA^ z=lphz|2^A|&&Rv3`N4~&(Knh^Xd>o(D8v6f0GlGG&Ut_Mo3Osine86C>F2}LdEci0 zj*d5&+T`S4Cqw_MNR^Y7+gMXPZyVyjfzG{*6ptJeH{knUhyJ_;FVe!S16hDk{=asD z)*JTkd9{jV|4$VU$oOuZRh0bKZT6$-GskVDoF!x00h>``fWgdU}}97iz9;?_T%h zkNV3;D6`A_p*&tmDKTh#8@%)&(rAp9nZ5GxwpYgQc3~NfCiX@u*lOUKf+Y)O05aSM z;n}w?fq53OV*HWz(IscC=-DCS8gWK`_hd8E+`A`Y1-_NWYjmF!O#A9x%kh#2U-t1z z@R!ATMma&}!|={%1-9&*uiOQ~Ts0JFNG^KR)<@yDY$!M3&t33pht;6Gha%xhz@`!o zlZR$w^4r%xj!1L7k>;pkNQ(^9`MKzSC4-jXYb*PlUqjrN3n(}FBwM8>+xAM_MI?I; zmqu-hZ=cPU-?nPO5X9}?7QC1PIzgooUdx%#t$3$L4&0F1ldbKO_Qdd>HuC-KTN)Of zGW#`3FkfqHO#;5B3a83DN1$>rffs@s!BG0xbakKtwgVPnFHM!$X&&};)S$XWbiH%wW9!xJ+@kG|X5 z&W=WkWwoe_!4&0;K3A3vPTAN*5?iyO6HlYM9n>**Ww2myf6J?O;GP<6T-v$MkFa2l zMJ| z4HPqI`oQKyu_oB3v*s`FY5eKS&Z*RMxGD3%1F7VS@XyRFwty-!*|Im%7LY%8b1BLm zS9Bm%MBdcmx#FZ)m3M2+QAwmCLs`ylEh44F+`ynWrR-2M$3?N<@#Sp!j7W+Su=!he z@Ors3a{GjW0y-p_0yuV%SE$jx7K;(?ouVF(yLl|(ZQvg_~}Dg<+wP;6^jPZ zm#bh6!W)lUzA?zr-@KskpkE~aku#iqjD3tNe7F=nLaL3sl>5>n&*0-TiFi=6?Z^i& zTh{hvWfTvUh-d~qeQ*8&&l)sxqJPOpdu(>r!aj6K=hoY|j7lbta5*2=SQzk$U(2uS z`2!8;z&m8LN=7z8<2XCzG(XSFfgNoMMen)^J2X4oWutL&W^YL!u^zDtk57_{Sm(9= zb^~Vg0yU>Ao^_>tl?o}Y)FLK@oOQ)e{2HWt zlohz)YCwNo(yRBGr^AOxup-I(T=Yib%VOpnnf@iJ7o^hgs}5y(pLv-SP2C!pLq0J~ zS0zYFs>iKb!&Y!G1Xwl&0Ow_fIZN>5=Ifty&fH1Us z5AdgqlXD2rpQa(&+jm+;sRkvTG)&VH-g181VEOpS!l}**%V33tq6CekJ~zildL{Q8 zRhjxNtMn<%7jQgR9kXO_gxJ&D9t0K1{X*If#;G^lgx2aW#m&bYN~#yS~)hK*c+u_2t%qig*N343;2CA4S}Hq^(5)@ zKCw%|P!vkp2Jv?!w~7Vjt$`Ib${Zr7L{E~z%8~wOKs$xoAvC)grDL9O*6*-&pf`W7 z-R)mB8;bJ4f*EGy7)&pA!PVCyB*JZNR<8z^s|z_g!lSHrsO%Zu*F_7PEsS_DPfTE0$S17lL($!uGa$&t zl7hjK*whU0QOjYBC?g!$?7>q;$z7cHD-Zi(BRZgJnvFJ!E=o<>xS{deFJbRg&YeIt zF5>qZ?J-#}Yf{vTJgF8_no7c#cp>w?A^=WQv;Jg*`w3|b2!@V#Fs^{$Y4?{tW+y1=UQRc@RRqk4uahqTXouvIlVlc`UwzE}Jq_vK6a3{d#~b#v zFfAE(TB^Y*XO8}RW6=G6&G+-qI~}7##{8_u{A}-~Xv(YS6h0e9N5v&L5i;(DpaR(N z_=u>AoB~5bEt|2WpEY%eGtL<#~2#OB+7! zCT$wI)Iz-L?Qe2(#aSXRb4m0yj@XZf5}QWjSc`%(E6XZBH%qWzTCiEZdXKUhXEY#G znman0)-n6r(GA=xGv2`d{ZB|3nt3z7p3a(@t~}Vq&B=B5j)6g|+>UZgm_U~21HUtoA#J+)#;NvV> z?bgW5@9m`j*vUaACUXF;?NFG9f)t}rx1{#unT!5l?0Uwc(b;5+>`{)8 zm-B=cJz_^QCDw*<*^>A1|FXNlp7p0Ez^HbCg1t5TN^jPWOml5{-8;`cWxl>V)QYDv z*WMO*&E%HV%g35zNZ=Ia2~E8;Fbz|dvs$wls<&d8xhj6vB#2NIxFn&xNsV(BxDw>j zo8FS%LGD<@9xO+tD;&BGR)!q~5ro}yY$48&oPrQe;*Lrl#ObRT8mU6}*ltv6(=*R0 zV6O0fZt_i=pCPZ*ARlyu)6*i}=$ij)$%%nCY_1~11#qB)5fRTR;%c?)I3I*uWzk#- zppop*TWx@jn^mHb_TaX5SLbl5vi=N=>39fh?Bg;Gn0AiHGwUpv@^JLpWjqaHnF6nk zD)f9(0fE7LvTP#=Oc}~Q&@xEXpP@f-c?@r`4wFRP8Ld6C8e3@8{vi3wdY5g9zv_GG zal2>+{3=p_>u2%Gn1gh7mN`uB{emW7qSFHH1Y0-FFKasFz|_dV0AGeP`Bm+JbBNKot-9;;{38p}JW!mM=>PiOE%^)gtx3u}>pIe} zdN^!NYjyG`g1G2C_b_%xLuTFSxXOqJfvw~n%%o)G5>bY8$*DHkPc2KTK21MxkNxyy zMkXGNWzS(2Z*3f(nFyVkNCSN8hV9@mllbrJmLYw*^Bt1l1s$F2o<*3)`x`wdet9lt z!-^d8EHpK4u zQaoB{tOvC+CsrUSOm3^vZz4%d87N0z-Y^p{dEJ8w>^uco5zyy*3^6g}AmSDIO^BQ9 z^$k=0Fi!hALYD@IvZkJ>J?XgovTEIq{a*gs#li<2chQbG-brtd&-uSP?#{&6_m!wJ}Vu|CwkNbCoNk;?UzFTJR5HN}7hd)JWsKl#cO=J4j z!aFiDIa#oUzJq>r~w}2EhPv3gPbj(D*!-Ky4Oa9|uESBzK#@9$NF5VTv zy@L(a;ZajJjllM^Z%UVzAGT;X{}COZNwHlOhOBOyZrOesgFn~Nv3q~;2WBz3c9X?1@@w~WX)B$Eq zP4U|6d)oH%>Fz>e!zUa?XW13%Y=&I4|qZ0QE zrxtm)<+sn!rEzG)g3+gKEm6AqG95qH7s{Fvp^LmT@-vEZj%G|9d5n*VoD;Smt2h@J zqRrYJ;WHS;s+~3+5YP`o1J5AEJbGS9NZ+iYC<{eRyho?ucolZVyy5|q-sjko8{Lt~%*WGk-qjV(V9yY0 zWE~MKmgj8#TKCmUdC91@4x#82Bo|EV(U;LU)TzBEOUpP8ZVdH67UOOXv$8h= zUVbpi(sleQlxjqmFb?C+R&wGD(gKW)yfKa2hADpw-K)tnez?n%A_FdRqh^ z$9(Zc;ytS+JlV`_xA4%;xWnKHB~C2qO=&5n^aiNCZON(N zqxh%AvFG=HGDlFpy!ZC_4LO4B$3QLrbOPo&tU{ey)wy?6;XR4j#cPThBJ0g}my(CS zktgq{I;K|tOq(|UOnT$UBO)y83OK;x5x$7Jwp`gu^d*z>{0;Agi`Ar|Q$5T`*+2zt zJ*!lZWl3dja1n|g$*(pY##t3q)X>PPjuFq{ipdy@A5=zb3NrFDDLFf zGCqU92p`Mp)|Gi^9cr_#DEA2b+x#_4;GSjPEPjjP3!YXcdcjOac+uOF@!jDk)w@D3 z4`BgIhE_JFqhE$|ia6;Vz`5S*84l0A+(vgpRLCB4SYw3vYwnhLlZk}O$}&q^w-7M> z?K?No-C8s+uaaMJ(wz0~_i2JTD89T5)r4!*OR&1nS&(rgzea5LAS9%QPC*N4CvM9_ zGmN?z&=!KObK49cJcADgK1OlLe)%3c9{Ohf`W{3u@5=GhrgK`UilNNhd+@V&#T}Af zq+vLm{d$rx__ybbrZUB9+&b3FwT3?UlXB3XmnAPBa$flHeLN&$ZB%H3Y|J{!L`A6U zjLDQ`sqB;-0%HFJZIR5nVcX7LdCRCp0cu}A}=>fedUoM!mQ%+6v`>GlI87V zlH2OEmrFK$HD9@Xv!k@FHtZf7F%YVdu=sr;J^a4*yN6$QT8^Eb+?QKwOsrR0*ek@? zFLz&0;+&@Tx+U?a%t1S_J50_+#QnO(wV+3+CkF18(dmvIL1LI3IpLhdQ3Gn=3i`o; z_@bVof!Z(YP@$k_B;CDzlnNaXaHBTdp;=o}Ebk-LS%hsA(V?(1FDl(3>V1e8P>>p2 z*$WkIXCM%q!ai>r;EQp7w)1gqyZsBr`Sp)|`o|vym}D~q9_Z*r+Y0!rRKq)%(PPrq zW70O2Xz;ewDuI{?1(+vuO(%Jd9L3;Bu519daR6^J;`+}rpRkq!7w5ylXUKU`GJ*hp zQuw%DGn~erTeS*^JL<6HfCn|s;%G81eKlVpox47@q93CoHE;@YFLVI6>BwxR8-J(? z#Lx#~+ z=6B{ihVCivZ~WU7Qj@bVCj>}7dc%@^~P#X9p)f3Q!! z8x?@L<|Xn~8K0?Vr!CqlO;C)B+d?NydlcogTjrRVqZE;D;Ki)ep95$oEjO3oi|TtH zt##b1noyG)WFrB@Sw98To_^Pc30$-gBMVrnTVWOgEa*YmwBrGmF!^mFSKJs^_TN*G zgwrXgAn;H(2_?THV`WUPljHL2oYVteP>H)%d29L%eM@qRat!U#dH1Iij%t7fa9y(U zwegAj=LuC>C)%q9WE7-fj|DddzI7K6t~=f~T&zGP!qYIhW0;>mi+@w{RG57@@LCqT zf0MTk6}pNTE*Z8KeL8Qo^%57+5-I$`B{Xp13PepjbKkhtk9t>C#7I)M==O6<%BA+X z$xAmosxCvtI{g54Le%ZP0FPq+h$0KLtjZv6`0CMj%7Elwf+U1o2G+lI3Dd#+cD}~q zPh_;Z%ZG0<(2|T%{}aKpO$ZxZi!4z{OU<0(vraYhJs>TmZ$@S<0 zVX9M*>nVt{Z9Ab`=Tk6;z2fC7^scY_Z4s!X#SDT3^r)59>oYH1WjV7bwI7?nof7`FXB0wSJJ zX^sn#rkRyQV9ADG@=qinwP{}wva>ePN)7io%Hmvrn1jd04QU&_uXNI#ho%Np zFxS7?md*yYA@(7zSk2OLwhGArt)Xm86FAp)JssUr;DycLvt9*Y7Y{g#Tcs*pkSZ=~ z4Q-9-e{R+)2#%Uo^i6u0a`#{mIXuF}_-pe@Puy+)!-0L9EI3y?GGb4^31J<6FpJp8 zg_J>$w;+4{&|R&QZP?jPw-?TKfo=CQ;D4YehXrjvx4lLb{VItoH>`$q@0_1uGo|;- zTlR(e;lRn$+T#xBkZ;G|!G-Rze}LC#T$;jDP>wb&+*Nvp{uj#C1NF;00XqF{cdM3( z>+;W(FE^LA-((F|br}E6z?Ce!baMMXtzf;MV%ye{s*z%f~4oAnx5xcmzlXPISt;nmFqP06A$SS}KuGCTf-9=MXczEn; zJeeMR0rJ}Iote`YdG|#fN)qs6)ebXAIO3x)Fi`iq-;lojnM(Frbb!6(Z&)YjSvOsj z@}O&y_w7S_BCa>QHhb7&*J2q$`GmT#ex}tV!{tD@+XT`474T~--UfHO-uy^c^;Obz5y-{uI>*1|9 zVQF$7Dk{LczS%vdJsv(>enQ6iO51)t$$1owPV^g*I=R@;hNI2I5F=z3J zpZ^J{w?~sRC=X04v>srMwoXH~jj!5lc-%~@vQMi5;4Q8&*8`K#JHT)ci9!;@C}^1v{9Cahm(?_~3M1(k1@;e(P(5B32lO|6Mj&;%rj zeiC3Y^N9)TGUftKK`3#1QU*+-m)`40>xQh54s zjKK)S3l871lf~n_t|lnTrl1-9Av^VQyt)$s?L^2+h9g}4wY+SIJqNmMC4QZSjtmVv zL7c;coTc1UZ;;Ga?64B`doeDxZTss0Zp`9H0ob3OCm-_1-7cZ*xNe00EL#yZoKbb$ z`)L2^M95De=dWPQ5x2=W(fWLEGc1!ClA>v%m@^-GUSOBm4KwRQ7Ls_KM0d;{qLy5F zUAIpTPNEy)G8d0KLj=n@I615fNh0>78Qg}HIIPhTd*p6Ig);~O+;QY&m>9kru)Aqh z?>`>su$CgyG1Mttqpg1RI%I9Me)WR45MOD*MwwTgNlML${y`#f5*H?+b0sK0wW`1N z8P9q^fwAp_cNycO1}(p@QgRrDkYCvly<^n*Zu)Y6^vhV~etTAh0ENAn(A$DpSkzXg@Mi})byXQ1If@yI*TO|Seml3Z$-l_Y+N9r-{+{nhy)R)8AK9{}o|fb~ zs#KwQ*t}lwBEy9>^Gc(1{tNe-$Z5es4O5N$;$jqp6O zY_X|uD$}gZ+Cc*a1=f44?epVk;kE0g%w2K?Aq(@VBR`0QZr-|WW{}`^U<}FBjlBzYeW7&q-?$>%>(gM8M~^gkpB<)}JKd((TJ#U6NlDs$a2y?qTJ z_-Gk4qGtCH2#+HOXG9q8A;)#DzfPHUoP)!{>prylyP9nGLpD;ONi@d<^`k`{8dP$9 z27|5lN%Ul^YUrc{-@$UqUU$Ks%7Z2M^joq)UdLR&eenb$*MN|kE7xra+B*lj3Gwj9 z;W;?cJl-C)ALz@q4(6`T#bsrb?@+2gtN3qTm~=h zlH7hx1p}CAoewKR>;R|cz<{kU6_{t#AL(Hn?l}3IHyVhsT7k$UKi%>m8>P6e9i9Vz zv|gIjLZCADMyhk1f@%(X<{*0r_}-4xrn9u1s|4S%XUbmJPJu@&V*Lc$V~y1L{$6|n zk>PA+zglOB_h*&|UcfC?gRj^#(tBU=VN&sTGLU4L07?*-iiz#Yb!~^1RJzXmctqCXOols{#%pHQ2MjPzPx1nZF)JbYe;@&In`ku zXkwQDG0m7_64Hs@j_vOE3EFhyYdf~8$H^_6f=Uku2OF)!Io6GUx`B#7mYewxK*WI) z?M|;#(CP^mJ7wdsQXE=&Kwpo26>L(Og656gq?`bf0lv}U#;ZykHWYA?n>#c@mQg~M zVQ*$>gxiR9eCzeITv;7&h6@e243GD=h-%!Xb8|xVWlSe4Cr`uIuGUwdg1+Pfu?hi) z5-aGiwlv7(eEa^gzimnj0;W~=u6nNdh!L;T9J^sic=Vf#2JGgX_2ncj{2@`!V2il8 zMO=Z{>T-N^^{t83#G7UHrfy%ssb}-@r7!dykmZ%vNm6YXJ9eB{jZV=VNdfQ@($286 zICht&5C*f)lILeiankSGAI!mMRWYXK$%%Sk>2Z1MtG=M+`C~&tN4t|F1m|LfkFNOi zLL;t_v1zgun}*xD!d2E&*5}*hI1ny4N+|eyzihB@(PRtMcM4h=hkPky*p{w`6V}Mo z_rDAQ!4dD$QOBc>6BV`s}Q~#NbsoAK8vQRVdP(IA(?c!1{cl zYR-KvPF`3X5V4*U-r{#F%ARRWCsl`@9p);giqd2|pxx;G7DqRrwS2Cs>D%%R>Lf@* zU1v<{ucc*ZQNO`HTv_8=$gEr5P__dNzu81thQZ`>wUp|1zGCI&`Xolv*nWgW@4r68 z@3zm(3i>zl<({}(< zFmCtlfP43E&)fKFlGV(5IReCy%M36>iFaMm8|VQ*u(iLET*0ovP==*1zWF2PYku$t zhh`Ye0RT^;7^_bE)!%mETQvAV-71`2e4!m%^udWq&ip)O1z*<=6Hw7J%a@a5U1+~dk_ocIcK`xL#>~5CUKC^2(l!qj&l{`t zPH(yR$A0BxF&FYWst3VYk;iVYX6hw{c&Js*m*JcIFJ#-JRrWJZNjnboD^Zs_&e=3- z>3i_d>!pO20ZnhpZT5I_z%^<7TU?J8JAgHAsd`skqC?88o(q2lGKHGX`4B8sd&*f}eno3XY>dFD^3{T0 zer840*}eZ~QhHp5WExMvL}-kMzr0EuCaC_DhO4XdPOJ zF@7?3>&~KGJYI@fBq?=SA}Ho?M!W255~r*SOZdH4fiZDi4|c?_JlL|U#0#^R9lSP+ zxR2VJHE6E5Co=Fl<@Lh?cXc`=7UyNIZ|x&g0wvOW$?wYZ>}@XjDZQug{bNOSp?igK z>sr+*$mMX5qQe6fO9!w5>w|2S-pHLIjWWiZHX~z&uK6&#XKw%v8$()+bXG(lyG^~b zdUjqvY0V8$rpy*X0oh1LsizI4b$SZ{%{)+A>>;JJMzdCWj3%17?`6BQ?dfji=^{1{ zAVgQk1()Z8iJwfcJDoDt)@erQnY1CVdU+M1ql{htY6IciAdDF|R9xOyYEIC(o5 z&}SKa?cJ1T@_mbT3w}rmKQy=1L|(gkJpN29CX{N>J;*A6(kckczXzbK9O{{6Krx`I zjK$rg2dbi5y9|;j9b_`AYEv|g)CP8hTh|%VaH|mpBfj;TEgf{eDUfc%b9q7fn6w;y zV0NMBmh#A}_KlSFQxF`VkK+kUJ`S%T+UYtUB4;q0C@26H#_G~S+m^YQ!pyV*m-Tfr za0{%KF_477W*r3n-8%)DYu{WVP8|%km?FX*Vk8;`W`Rm4gA) zE}JZ;cqj9*tqDJ*u09FMmB&hQ(J%MUq(FB(ywRu`h&|CxKxn$G!l?AvkE>v@>6cIV zUigAWW9xR85Fm44@#S%v$zSsA%x8!u{l0@H=}iJuS<>M;GjF8Xvn-L^G!iKGw3=|C zADebdIcN?%Y7XI>VkPo6{?}x~+`G>8x}i?b4lm;8Udr0MB*$MBBa&kz&f%|%N1t)Y zsJPtvR@m-uanGd-zO$yTOcyfZTIU!upldqoTQ$BVLAlnDY@3-e)xukLUtI-@U_O!? zpK~WI>Xe_p>?iSp6z3okD_0%L>Zu=fQDL*nYwZhk$YYoI|ok~ zGd$=DkKeCD#tTGn+_(1(#GU9A2P6o7TS0w(I5+n1`T6VXVk zIO(Ou@0eGg)uS}w6cgQm9t_~@rn*;a?tp)p>pG#Pr80L5?bs{o0i1TjjX0UI$O?!O z>;6eIeW1w(eR|g>+4d0_8-NyWEqPi}3>nWGF7QTfalBT?`B_iYVJ};EK-t#;dR?M+ z*7X3fL#g%Hk8`)+k8Kn=A`pm|z|eLR@BMt;&4uMtkl>CK%@G#zx3C2`$>3FRw&LP_ zM3US*A5QVo+b-IL4T=T*-u;C~*p!#JKlSvyry2TO4+PX-jrROn_TG z$3MgjB&fYvJO1c6T{TI~@z?-SJK0*0s1Zr$o#@}UB;U8BxhnKpc)+Ly5X%tOuD$Jk zec|5%`TAOTd4xs9RgxiP8Hie`m^omA_J=&s5 zUhKPYUaf9Qh)s4D(SdOW7sDxiV2t zuh^zFJ>;EAhbqn3v0VZUeV@eR-m-PprX94Iw^o^M{A$mlkB(e1jy=a3J^XDq|4i+@ zEY|7K1phGNA8V^Y$mf}_QE({Kx@bTBu;lxfC%?(6m_vsP1P1K#SOH51)uF6f-@mY~_5%=5As184lj2!eq=eb{^t0a9ASwYer zE;rq#lSJj~`Ymwf@9(qfgbX*EQ}WI{6_ruv)MF6&dv)W<{Q|6z`=KztR3V}%C&ojJ zBze9v&!^3&^0>`r8ui$zTJ(nCQiM@mQEFMjm+!YUzt|z4>orkoL%-1$s$ar2FFTux zK0g;8s&w2f7^S_OBdg}_e`_V zp~;aA$WMeGFW=KX%5yt197`bp2aKhKIfWLlX}Czg ztUvg(eSF7q;CP&O?|eOBUGTN&73fz#o6#;TeHYeuZr<_RB>Q1sk9&xQqYQrM;FGG^ z=lM6KBHwzWe4Ab0EqS@EYi20AFp{PbJ$h8=t*yv zlA%8DqwVvm?kPnEd<&YBKW7(u^PFSOnNT1Ic`@Vh;}=?OXWVlD(Ccxwgp?g|CjCw0 z8m~?^PTq4y`^xhXr~M1JA(l_LU|@U z2kSTm_!_?04QT9!B9J|R#`%%@2jm=lt24p5`18{9toa;w9Ot38DZb@1-f3ccKgQog zyK|`BS1&+X)#A8|MquGbw-H|1XS=-YnzRDo(+2|PX8p9>tLAeD({}PJWx|$l|L(=b zawQ6dTmI3%gZE$b@tsGf^wy(>BrNYr1wuUX_VX9(^r^!79lIQ54Y7a5%7g!{ zFAmlg*|$f#b!0j%wx9dl&x}G~MZNrS*^Thf3h)v6nCNSKvpOJhq=Bmx@ zYh@rs2i#@vevg4N9`z?ayOr-(RJd~&9HwgCfCOe9tT$hBQqI?lC-Bhis9!Q|olCcf z`y){HI*@Vg=Jyk~&iyF461*FPPodv_6??rX(&&krS#=PLN!$!NAVk#ojk5&Et-9j) zeF5N2c&j%krE;c@2Ht_^D35bH%I-u=USW*(>c=+D{8H`Z*05Sn&uEgV1TW7m|Mkye zWuBJDfs^Kc6L$gsY<_a&@3!b8;$qmR%hla`P`GlaebwBVTrZ_59S{eke)r%xD}&9~ z7cOZ7Un-zoyiww}47XNdAhro9GXQpTkiiOnXS9hWZXYjmlM$#J?l=(&f`6zWrXB?Y zJTCy;s#eJsYpJdrvlJr+MNhP&5mUM!xZ6F&~#|m~|85l-G3%x^N2WmV1#m@$kTH^otlH-^qI}K_`On=03)*tN+Am zQS8x4hx#$s)}x$ep zF-}F_TP%m3#IbD^N^JE3mKG}aQr%?edUuk6h<2J`tVRZzBBRdryMN|+8_eb!^+!0x zzHieYOW!T&X(wm=u$BX?)v9Fp`qEp+cWx`%-1Gyfv`w{$RlX_Kbk&fVp!tm?@r z)BN(PGCxJe_wui*oRR~-eF>UnL_Q3N@nre|5wwY{Aw$2rg!tXd zs*1EXs)UGni|pU;b4g&dLi3Fze~hUHFUdFVWTaK$kfOWoz!9Ln_Mu`JPU4}0X~_mn zJ6^Xr;V3~)9LQKRwtw9HbakkibNGhz=K@{Dzuf%$PdOUek4bL-e5sC6>t706=16^2s(Xhs&jf0D9B~@^6`s#ivsu6md{ZRikSmKXw3F z44prLfc8G@oA>)@F=k1TS}3(Jfujc!_=yKOE5>0LDmvUo3H+FD@jW2iQWuuC-qswz z-MM4Zus{?r1gK`nGFNH~0o<|Z%r=s~un)<`Z1DTpeX4a7nBHOi$G--)o4}e>X^Q!5?)hV=kGi!+?&)$?nct`6&qCK7_YC;~@5D zQBwNLmlM>n`?|BCsw+VBgpXOTw+WtS^!3=wN0g5O%~0>g55&F^UL`9rkEbH1z{s)7 zcpjo{yZt+jL)P~wVDeMToP0bu_N~$0`lrPQuXv}Hxdio@+kzwT{TnYdk+RDmx zo8~w2mQc)HhO+U{CHbZujPiZp$vhv}Ayq?^ES9NiMPwKweXHy&l+qiHbz#$cpiCon zplZ3!2*i0tgL>Ce7xq;x98KVRkBf=zYIKE@7NHG5f&%>ph@O_~&PQkv%8}Gmdb2rs z2Y`T2dNXIXJ<_)|kNVvy&E6354*5I30f~ld`QgG49Ne4gvzzdblRJ9TLK$qcM$(oM zeLZUbrsD!GSDMj+x51_%1JI_h67DaFyDSE_&g9*f6a6<%8LN{o7oKHj9t5w*F?8&b zySf63U4>4~W)HYgW{M4K-qa~*&GW|S6V8vXSp4RBpSPe*lL*%t$qV1J1*$&w zOMRgKir!u6gV@Oa$hF`16`OJ1H|k8j{ZB?-3$)l7@&%rp!Z^S_~!x*euPxPNt4H(^wtCtY4_LjG>2MRYnO z;kI3$TX7QIBBGdR*R|_2SQu*%Q7;&NXOoGM`v^9dVw+%vqnBRknHLW}l-=qz++)tL z@d{qF{S;5Fq2`0EpFIU(0Y!>Y^N3{9t$KE#%6gloyXU+ZcD~m!o-*E?qhTX*Is&{C zUO$HyF|&OBr}mDrS)Ad(!iS-o-jF@|10AsazKWG`2%>1RwY2bhgt(qZ`Js`E4L|vHs)HhCxfGcWhm6awv2}vnMw&;iPK?dbTqfU}}axySlN=75&#`1D{3PypG5&=mRfIv-B%d$9yy?tBtzzUwc~OMtgH1@$40fmK^*BKSEQmya`<_fLA8 zsB=5=WEH_;VJ~c@e#urRtJnO*v;grxPu7In_#4eqELgopI#o=rz|@=;j57(#&|&4C zq=m$~PvPQzH+BC(wWCqS$QY@)9pPtMGN#b5Uxb{!BW=6VD%}-To+q7KU04SF-JJ5C z@_cg{{oO*1o=-x`4^KfVKr|M=R;b=Hc*0T+ypYDtLO-&OzN)L1_N9 zy+&1J5X}uv+a@%=5_74MHY2Tc-ePD$3M$Ybb>ZEdx;nPMzp`SK`ZciGLUjS+ZQA-3q0Gt4SfcDD+u#C z@*M38J5!g04+KAcG-9XzgtnjR6jWlz2*^(Su{l!*70bCIGqVnaj!^aG?H$V@#B?)$ zRTNg8Am?i|yknT{z8o}rp-46hf5#7q;%gqu5Da9RS^42f-5WrEF*qO6eM0W~p`_|8 zRTubD(^AvIEd%0{uv9Q(*%9Xn;lAWU9sy|Yq;_#dm4K06=2#2aW6e(wL9naYQL9={%s$-ZLly5AilLCK9^eQIy+P<6LcBFvhkTX^{P zee!k1OCP}k&#jHBcR{xiWK(N&*N0S!uYY{uKCK5AJa~L%v*1W`K{bUqGFy(^d81*= z+sxxyfSj~Y*Fmr5EaR13BTPo&RHl7-MX$Q;n-EF*BV46Pan9kzA@T(mq#_0|^Qq;a z%Q>5dZbtLNwIX+fEc5h@GnOicLh7D(T`TFR^bySv5`kX3c{B<# zgQ7r^rZaa;A!c_=rBy*TlGIPaMt*f>S<-74vvHMQ=E3K)bzHQ|m*MCs`(*Qm2|&r` zZ!-RjO@)m!1jurfc2D-LAb*N4VKbh`|ficRx6=ghx1D`DU1?{&6`f&SQI ztP5=Pf))e40!P9hA8D%#6A?-nvuyemn8d|zrM%>jkH}My((>%0g^OmpO#keWU7znD zkT-2Fu`9jKU=TUp0DRHMv2g2w>9w|8r}ZZ%SXNcHNdvNG#Xpm=8Yw3LBMLJrm@$NY zxJDOp3`~E5w!syc#AN_#e+PIOeZuB`Cb1SsG5ddf@3@RVXDYLH zeN841?V6E4A4TX1&ffAyMvgeS+232fN-l6ANAY`yDl0Kv7bTp>)obt0mX_yL2xPO< zVm4DFFNw424)KS{XiWCjD|$QlTbUActHrbrd(Th#N+C9N(5wW?19`Ox`_?~NYEHx4{$@y?e%-y<1i7GCfXvTJYaX=(gE zo!@^|?;x+syC!PIuz$+N+D;8lo|ZFag2Bxtm2y#6TWT|Z8_B3T&K*$4n6l`}5=b2%3oh7n zyMakE5oJ5mCpN+Ao87J9QZJGs!~sA=+7#|7>1_$#gd^Y9+~8sbCXypX!7+T>pF^Fr zMXL(%*Om94TfIqGWIJ;`$;qWq5AdYiht(J7*W@XFGdk^l zaf{oVD6_mXzvhVK7yiWO8I(8TN}HNoKlg;M7g5OWhuvr{M|V46oTUlFRwFd|1xkVtjuTn+*-PM;L#lv8CihfsUnWB~(@H~9@BO!=D_Md}Uy%ZP1 zQohJ!o>D_pvb~ixd-?%KA+ys49N{5O>g4a{&u|bM0`NhpU~s-Pu1ZcUy`dC<`N=aU z2Uke?TYR_QiEVut*Q^VNn}b>pgSZ^+DyO#y}2ue4~LIQ>N@e)}L~7qrA7i0lUt1x;*-oJ?Yr(&|{1!#^+LNXr4M zwe^{onU#i+Dsw(Gw`oRTYrDoZDA}DK;lLQ*9>Owem+qolFCqS|C18g6>@;khPi$?~x1>P` zwYN>dIUly&JY9Nh)ipX~HA%0Ik#h@_e&k+^i(8x~+a0cZ`~|$pI0FI=8uovD@(pps zXt@nPlkA-Uw_z)E;t>nEEPckMnS`@$q>s( zHLmo#W2A}B`--)ja?SKDvN0;sxlT<#8*~YHc+pp)TxpzgH%=u*k2!`#Q8<(xQ_y+G zO37R@E!9Q7$oM+~w!7RmtQU6!Z)#T}g~K{YGCCQdX6CLTB+8umlN-a`flF zRX;{J{_{SzIxD?SpdKf-78|k`%5lc{?XkYhw(!>=+mC+%QxJ`F>K7sVLK%Z=-*2Y4~7D{9CQS5r;nYXT6y}t;A#uaPe&G<9T5&|LgU`m@XSayzo)uAyw zB75(w_D;ZtRgmGY=fS1r)7Ee+wl&-n*o$CZbm4%SZldrhVGQHey1M|RUA@poSYZ_`g+QX&=urb54rJt45S^Fa-S(wL#BP~ zW}{HaB8bP2r`fp$%$4~058SiL%XC^Zm_ys2Oh_kB-vqq`J4-lbO1_JMhU#$~eo?YG zzB(Q(Qq*0tfh^znb?@g3Ll;-a*o=YHY|&Phuh zya0`AUL*<-XPG_Y6E^2EzQJS3Pr=i>=&Lzz6rvVWheAG0}ILX#t*~kEX+R zZa;QqMNQsj4#kl^liN2_@7A2Y%_KDqt=dC!nQLA%RLjDVEMjAwImN0xBM}!%*!rBw z9JHnK{G3G$=RXahJsr8BGU3Lq2}>?`U2jh|92>Y}w3By}B==C`iqXfi3I&JpD&Qcy z2q(yZ{ERiZ4-xAt;`80JR(M_pAz=FTc#9_B8L#Wzcb5)D<;$FJiY*F10$UI(2 zPE}z&8})MTS09d_-NLv{);mw&+`YYiCP!6(>KwGgzxyR51!n#%O+1+2!7*r|+&?&a z;;eeiR{>FTZ(ay59F4Em$EW$-bi4IbXjFehz9xlp_eZs1jC9Z{?0Tnym+=w?Zauc9-+?BqKGh$hnPom+!T>Fb{-Y3H9HzK7jMT;AjS`f4y}<($1&A zY*&=oqPdLC^5=*_iw9e3YTA9R#mc}<)4 zp4glDy4hi?xV^{w_^K zj}Q&elxAg1IPCH^J`ie7WBXBnlZss(Q>LrdeJ0fLOw)KED=8v-4~iV-wxP(ym``n9 z?&>m_=zI^5qot3e`JLdYtMmP|AFtD1k61f;YCcCE?Nx9Tuqc*7%~2S*0Z4Z9snNnX z89jYJ2$6VQx0QgBnc$+L!{^`=)Qxkh zR1RGHdK?++j$8isyVy+UNaXV!5F&l`_@l$a^$WU$39K}rFiskP0m+M3+)KXRD|xH- zA&Vo%B>3^qeCjx$Sm$FSkMC@5u1@}Is5Yd&o=jhxifA~5)0O&oGRHA+^Oqm&7i*qR z|K&P6jIPDM+ICbU%*(-|U~~{)q`Kr)`Z-0z*U%lB%Z=HJX{y75X0Ya4lWSkAnjRTn zo|@TIWN}TFjY%@-3@@j9#hG?2PLu7I&rXV?6e*Y8PZDyu50rD(8vRb9vy+;)3_qrg z_Q;qeF=_y*?=)@M`{$lM7;hf@GBrEX4aN~)^?InLQp>RvIqMHlF^EMB0(?ZCRox14 zKLv@vyP$g@k8aFvwL>jfWEmQW`t(`n+8C=<{5DF>eNn}q=~4d=*%3ASrG0~~N$l1e zv58|3_V!{eM=pqYSJBI)feePBDOoNL=h=7oa5?q>;UlV_N8j$ItDtnB=+RkPIB^D2 zZdV5!F}~LLn;k{>%c8b@et#mS`_rtx49mGCZ9UThLE?6F#d8>d~6I^SXeZRLWdP* zH+1!7_0`HS0COW(czg+2wHhQ!NE7G)$*{r&}zW+B11ClzpMp zW`q0$_s-YFLrqHBIY!1bWPUubYI1G)_WiP@q(vsrTub|$N`I_j&* z<(`g(@X1QF-@~UMqGFYO?5*#wxJZ8IkRPFtK45HKcOtksUIGnMDJ_|KjW#Z^`B3<{ zh15=``to%x_sexlBN0yhJ4Z5jo_O&1Fzdii>iS*g+icrY42St_9V15{c6IyUenz(u z%jQJQ)+X3tXB>dSvg#>0VtolRhtbx;pK zOY!wwo-W5Unj2?qM!JAUZ}rE(-_v=0wr&@Ga4<9AHt#M1sGh+QN;!}=y_8{->KZ=* zBOT@fu`yBeK4}0J8K{>PBAtF{QdO5EJ94N^&^+oDQdRbf0~DylqURVpPHf*7iZk&B z`5@vTRLL#mQW$x}O0&#bM^Ns|JW83l=qDY;?lnL-Bdu`$1n~70!^#|&nlVuRNAK9{ zMBQh3uVjw7s9=lt%c?$NYRYS(VW-1__i%5}%+sd?fA35^@=5s>zDtEs%AHUQZ9Jk3ILyR479$D~ChuUyt4cy)i zG~>CrY@sA{)#w!d3}dVRN}Ti;Fo_Uf;agGNqSVDSzKM*oIOQde%$qx|_K|6ck~cx+ zJ?q)KoEf__+T(P`lYnH7Nq0HI|?9)(+>w$W8XkEgF!jzF%XKg9Ee8uM_j?cnnB-H)72=hWSUh;GC zmfkM1ET!LVmT}ZPMlF;zU?n$z+)0uC$^o_mF;6!#sJDB`LLhi+aH9vG&Ocv7yg&bA z6@PI_;x8c5AFhG8ef2K@aTxs1iT`2EibiyZh?{giT?qT)%;JJc(UyioG}#W~>QrqR zniR@%O&l>quhlbXS2=po|K;f>W~t{JY+v&sMt&_rt#a=l%x#Kc`upl!hN}kgX?+lr7Sh`>d0a7nz9WK4DX!R&yLF-s2uqOBxP@cpix2(1Eoyvd)&%Ack QUzExJqH+NL&GqyPW_ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-8-$$ _detail.jpg b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-8-$$ _detail.jpg new file mode 100644 index 0000000000000000000000000000000000000000..f6806d57620cfe4edbcbd2ae3496a56e0bf2a3ee GIT binary patch literal 6542 zcmbVQ2UJtrnm(b0&_b6cy-5`WLQkkCayr1uhvfJC~A2ntA5kRnBpBE6}A z^e$D9o=7o3Nak?wtb6DAX3e^9cGgZi=lpwrfBpVV94F2J^g0^a8UO?W01)s65T}4E z03`(ll!BZR3WdU8lvFg#v^3PzH0+E_aOTq-Jlv-_xVU%)r9^o7B?P#*M9-g*I4dJ3 zC&wdlL0M5&N$Q-O?1>Nv3N);B|nEft-;~MwNoe z&<4uy!7Lk|{EAXQ?Oh9t(Evv9oULaBjEeOX8#{-Pu!!gxF*$jK^NJTPUb(8Sp{b>< zbK|D5iK*Ewb31#7JC1jqoV~n#eEs|b0wW(pMaMjhMW;M|@-+2X+Vk|B+`RmP!q;z# zDk`h0-`CWBsB3L&@96C6?)fx0G(0joHa;;qzp%LUeR<`_>KgXv&hFm+FWkZ52`&gg z`X?;#{ZC-O;bH)Bk&uy*l0i>!K}h_-Nyqc@nGVY4X0#QOMdmP1txIL z2*YCQIY7lKC^s*JJ%RQIvVR6F;{S#0FJOP;ngnP_A)xU{82|+E>rgZ&jOw?^*+>(u z+~&_Np6RiEy^LX?!LZ~rb|0tf=qa~qO@B`;R4e>IP@r5g>u9a9?%$sNQ<}1PbBQSu zyI)wmcAi^k*!w6bXhO1N^X;r{QJL3TK*8ht-SCi{1_Xg#`3fOM4@Cq>;J86lE9^D= z;H5kfz*8a?O*itOwFAi0M1WHG6%n{n&_o2nq4>K*-~rr#2)r7yA_C2k1i7Fykipfg zpYG;k#}WKQ;A>Q%gTE|Z{;724L9l6w)eLGoc{2}(XqN7U0O8RWM>4op@|P3mY*nfYyfE>l=hpZNv0GeB}= zXeJ%z~jcc1;}Mv?{D zp3oUb!iP<-jLLuuV$oeDWD$A(~f1LN&5;t_M z%CBQOu7)vXMBO+$$@RuabC^jlx=TIc6#7!yQ{>$3tXcR+N&(DbJ}Zg($oGq`jXhUq zB)^6wGzUa^9Z+Sc2vwO;GwS#D5CQ)YzQR>CV-X`^!Hj;bYwPcLE_81XXTnOTwX4=I zIYfxp`bv2`S6%bcgM@ytv?J(PMIhGKtd2?Qcs8k}4~`-U1gd3u&d)(*Xe4#?x4Yky zrG)&<#uY}!0~Fu9!Gv#m8>)Lg6wS#HjAh(P9%1&g3s<4bDL($#PYBH>0!rGqNJ_C z7v@nh=_~CaUTwPhp&IOd zxjV>~+G`!;M3&a#aOp$;g?{;UBH&g@1hQTuw&$|xP7#=&g(H425&;}N5m?m3pOGd( zeH9`C7(OD@h>;vuJ|_=HOhD&ey#oQYL3)i(+TfVlTn1t|O@S^yz|?FFZiU zjGWP&!K1bt&}>2MZNC;t?#^#x7L5_)2|mMxH8c%(&Z<&r);RB_1EM)-K)eeda_^fc zuCM@i=Wu!6`}GFL>Rhy1rriqH;wrKQsb7NX2?3Qy^Z z7T&qBzjd2*V)IBl<@BuImqv&FA>l^44^#C4Q$V=<6`Qoad@2kQ8WF z?dWTrca&C)No=Kil}eoh=Y3STvR8m;b)#s%sm;>*OwnNMo@@wIK0n`5JAmIQ-K(eY ziW{dAy2Y~XAg-RmQy~JbZ#Y=Fm3eftp(aJQIgzDg@^tbUqwW?zn%(}@>l@Nii_lhm z3fOx}N5=A*noZZ*L}b+>CZna*kO4iX)tuDiaxZy3Za7G|YPc$1{kxNW<&a$fTi7|y zv+(9!C<4P#DKtt1m`;0rQCD^IVx~KoWK?dzj=vA|C5hMcn|8yC@xW01)nf#51a5!` zP{p)b?QHsh9-vQ<_mNk=e(8f`{8ap?t7pYb>${?giY05Ws>s=7^H8o4$I+7fk54EK zGi)swcDfSk8~d#a&RC*9+0}s?z7htY#XH-hUU8P z;nm=MCpq$&3OTc(3c3HUc=y&Fzd=?oE2+s_~ZVe15DunNuVAD*h3#Y3$yZfbnFrq)M#Kt1^O?Dyd{X*XV6iEb>C6wy~NFKzPNb3Ske24qPAL0SiiFLhI{wss(N)|bm_KK3GdJb zcaTxB>?5B882jnQGvi}x1cxk~1%VXQTm;Y2o(LF$TC=Ab$kvRjaasao0Vj)qsabQ^ z>W~an&fG;*)K%{gGE2#KLlVKGYV)Ce4X=5bhNcaz59_xAwxgBo=0Xq)jt}?P%Oq`w zmdq}x7?8(_I-}OL9|ZU%SnPJY+25vB zg-|#-#LLw)Vim?HW;E!|t#04;OZ2X%J!g!`-;By*tr0LmBV@@absl#s=Cpr$gl_D= zm%W=4IhY)~oJE0HqAf#>-&s6B@7C?;NAmon7;D-%=Yj7Cv;>$9)d3s-F~ZdlKSF~r zxNlld9y-KQ=H0Bl4Md3_lCS*nP0Y8#!d8{k3B}$h)+shr&Mrk*lE>RD%4^d4@NASZR%R;7W>R- zP{I41TY^7-U3zwh91J(#eq=;?sK~9e*1^(c^j#tf%K9sM&lXIwWMx0kTl%VpvdKrQ z9)5N0rCUnA-B5Y^7WJ^%VgtnHd}4LWBO#@4xi7L_^Jhv`vkS!>oiW$8i}1FSsGm^l zeoi^t>Cr5LEs9wXeloz8(aJFp7p8Dp$&R#;MFPci&(_!sFImQG7)5Ot2UqQd^m4Cg zhRAvP3olrlu}`M0w3&$weX*jnQtBMnIjQ?n3pxAQb>D1g>NLC95wc~3XxhybKtO3F!MQK4hr3Ds7+=QHwpy;s-;gJIbd)gY(KUo_v*KP= zTMLR+5-*S7t!qWKC;!p|6jYO}anq;dECOHX? z3lh1WnreRU;osAjKY~oA3{!R-ZA0)efj`_=*y)$WxwQ)|3kXhl1J;m!eQZ%p`cop% zu1|lL)%UZ4OvR+*af}Dsdip}JUeaKDe?|;z0u1V-zD1#c&wdKNSzCehUN>fT$RuS z?8ly?S|^5yzz>x%BCtK4OJMCR1PcrW*uaE#NZtKK3U4Mzj%|E^qv_g)co}2@&G*0jXko@h7dzUt0L5H?k9Bro%u|zou0Z2uX^NM8sZV%7iy|DO zFXT?cuQ%`Jk`6X>5m@{FUgR=7?rcYp8(qzOqlIQorYgtubk|XNSGGv?Wru0F?jMgl z;fhOmas&87HDUiC8J{F%I;KS)d`hp&aXT1z9d7>qkom^5knnx8rdLF( zJGmBiDZGv7^Yu;q^!*ooU@3)`K;}!a)arXfYh+e;i{F@~bv+pI8PxPK!`DEqPmqgdo1wR&^U-^*ug@$>zdz!M}+?iqPl%tQ`>q%!rx%ZD@ zEZ0ZpA_~7oc9))7w2j0}{zxA5*CGNRJG?(~y-S5{m*%6_5K7rr)jHf7R2P#JPos2} z31HXTBVSJhN-?10*rJ`p(9QeG{6hpKu=^-=!52A4_y7HU?^;CUAR4X%{FWVmbMU{q z^Ro?h^YaU3ur-eJ8Q> zA2`Ay`@jLqS7`ScRef0kw0C7LSiww*05Wk?{}EzW;s;pSSct$_nNgmt8Y6`k7u-D2 zM!($aTn`^sx_$l8Q3hKTEyI>;#Pi%!!P)l~m40=Y6NYghFIVeJ1evGRao`Fb3XiqJ zO)Xs5niPS$P1)W$_s%M9AKNz#gcdY@`cvG$tzWqp0$q?!u#V9j-SF-`cg-n9H`eA^ z(9;_)Ajl4Nc@5YMZ#^dm<>E~w8em=%Eg=KxhC!eKG(8yf&?JRMB4Cq|xGX-n-g}t_ z)y;_n8-?Dk9hOyZz)+nTFlAY^dg$SAbU1Xadg#8qp2@0imn*E38O6jsd+faF(uD^( zlF;N3p#=&eky|L;09!Xqb?^-$aFwtPHW->XkjGJ*b|7~3x49?ufU5;4UbeedgTvzBoo zgKq1*F5+Y}3UJHry+s%;KQgR7x{34`WbXIq;W0;kQf~L)K%j$-^ZkczW1)CHQzKeK zJ1dbs($C{EV_eI1ZaDLgAji!?`GI;Dz>Bd5TixTUO8+J!XEamY5XX6J#(wbeh2yGe zwVu7xugQ?2w&m*SS8OYsrwX>epp{<)g;nl|FnD{B9a3-W) zzou@YvN!JX^a?xkamv;?QM?IOEv}~R=nGwKEzi7VQ)IA}!=fJ3k=oz|KJTOsCC0>b zk<+U9b&hwTR3ZJWsBv4JK6gWgwW<^InOOhk-2Z#C01}#}rFwJJI4>AR8jTSR>s_JH zQ6GK!3IE4!e!f^|6HG<`{|Bi3!|Q&L2S+ zYL3Bf{MV3)Hw??Ya=UIIC#$VD#D!5;A}Ay3-r_-MPMz{w4Kt3**`qsx1NV<)626Ts zL%nTPI+}Q)c#253*V;lA)%o52ZdWVB9+*XJDV6!ot?U#d#t|${YhbheF+FjE2R*w7 zcHy&nH~JYE;>8vgH%52N>pD))v?Rx}iOyp*5t~%@9LRjJ4&~`vwF%2?U0>i^xd;?i zFzZtxp#@h$1VHY8ygQ7t#+wjlP^MwU6clF@WNz3H&=X!Y)=^CBW4z?#RB!iK0q$ANo_<+;42jYR-wxwm)&Y%!vzU(3Poh2iQf> zm2(dE7~$x*ROJ;9=`|=1)B*1k_W2)W6A%tW(`>|g!28Yi7OCu2#jEU&w(iV9DJpmu z64x6Y3as%(maBn5<~iG`lQ6Ts*jcjpwblR}c7;7v%fj8o4lcMWLR+V3)!{o}c z6e+{%823p2{O95+@mM+9Yf6tTgx7;gnk3tmAtKoxUlTjSF6x{dHusFlI}UWj74Ys4 zSqd^;P?U+8JAb?vjvK8u7~LHn*|q3Z?jys>QV>?Vpcq9th{oC)EvLay)#k-dQ&DIC0{?0NZVSCIA2c literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-9-E --_ G[_K.jpg b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/images/image-9-E --_ G[_K.jpg new file mode 100644 index 0000000000000000000000000000000000000000..1e53ea0503ed09c2d27757a57fac0164497f604b GIT binary patch literal 9871 zcmch72T&B-wssQ-lpr}XN=A}oiGv{dkQ5|?2uPBoWW)iK97O>Mk`a(3N=}31AX#$G zISg@t32!|2oqNu?b#MLe{Z;?pQ+v9rd)4aw?Y-9CYkjLRlbA)|hK7o|3V?$H065qm zfSCc503t#{5Fr5(2m~S~Cb~vKO-ge8Itc?M6*)BvBbb$ik(v1>hamS&Hhy+yW*$ji z{yV~AVq#!!DY?5MvVx*wBEOX25EB!VTqmI?C8Za+#e7TTe|=zD016@;PGAiWhYi4` zz`>)y!L$QlY(EKbem{W!K5%gH@CgV(M8wywV+E>i0Ju1Kc)0j@1O)i_Sm^-la{!-$ zfby2`JwhryD-fGIwMfw0Y$A5WitjWJh7UPJpF9sHzD7$&&%nsZ#m&QeTTEO+@~)J$ z(tTwWRW)^uhx!JO42_IUtZi(c+SxlezVPt$^7irdd;KOPG%P$KGU46(#H8d8DXBTR zdHJ6T3X48}{Z?63T~k|E-`dvR(b?7A(>pTyb8LKKa%y^Md1ZBNePeTL8-8?ra(Z@- zxVZd<3kSga2Q2LWKY;xUE($CzTzq^ye9$jkIJn-}hDU)7u;BkIWd8*AC$4FL1P=#0d3Y261USFs z$qBsnZ<^G7c*i@W>bq7v*Yuf{SE`*O(1zNz#9zE1wGNu!vHO(7~tBD^-tM;mjIc* zu0&idn4C8)eHRjR`l+vY#$?c~=Ecl+W?KW}dTohk8Vt~S zoma#tO6MD@!sjAru7Pz<*jxz7jh4XxIiB2TP&vZTNW4k7L~FD8_3rm-21)mZ(VacD z%k^ZIRP(A6Tw=1H0aLU7K?e$EYmg-fd;vvO1UZz0wpPkY!r9D~+9@^}(IM_Rjt_bL zJ>RzFwRqwLGLJ8fS9aW>3IH9n2wQ!h7mbzkEfeYrBnKmi!z^pMMfWOoWim>Ysdew< zMDhUtlUB}mn2Ah`V}VYnK~AVFiY%0hs|yx|!c(hjJ^Y(ARmUksMvOn5hbq&5x75HN z_`Ix{Mowd99wWWKPF%_sV?yvS^wINsJ*+uJ;vr+`29{%^K4z{X2CpB4`uy5?@x+^I zCFKZHFUW}iA_@Zlu*(x@>+DwyknP%Ti6FoL6Isf84^?64$&5>mdL!D4rybQJyUhw( zz4SNSpp2o{K-`xhC!Hs}3DZK!>bokUm9l?3C$4+h%0YfR( z5w=KHAfVR!Uo~d#SfqKeLuy&d(jb|sl3-O4p{Fkf7;9rZeIj!ro+LIHx%y(+nI3$3 zB36$9kjXGI=%k?Q`q7z|dcm1&&XF%-7f=8YFHppxO1 z=+FQr?DX}6r=VEV-?STs9g<9FCz-63e9=OE68!GQ9jcZ{&AY7|^fpc@eJF#w+BY7R zx@n>=l#px-dOw(MUFN0(tkA^O6b;8EGE*YhmqmQCk@sa)v94&}1e=ONeb}2yUjc_W zL2Z$SSo_FS!TPAExj=bjsY}LWlxgi=nOPUjk8oU368Hpr_6Gg(M^MDxx()QJd45Sl zV{ua4>N_Cg&T|o*PE|2~zwF*v>novU7>eXl5{;9p^y#G`@D__dQ|BwIQ&uh9NK-SR z-gDLa`m>1}9N~3|DpGjf&!S(yHf1zeS=jQLc%XRLSP{yzh)c}cKD0At8g^T10mrA| z;kZzh7O0(e_+F`l?aT2u$am}>R$EvzE?aMGYh5@pdufWb9xn{g+a%wxc#)GPYNbqV z0O%=G|DPJpy7XVCWyW#(xnupaahEm8K!b_TcWxFTi3sapEC?MV*x1P+CBntX&}dZY zX5@=EsS?|HXR@h%=qW@o22hMm#sEQ?D8JQmmFJmano;U3GR8YK{rGev(@L()kw@y! z_L`jwZaxmGq?&lQWnewzDQrg=e`JCIo_-gpeNTIAOmlwQxf7%yn%VjVR^5CX0}N&7 zV}M0FEV8)0@<`YTj{_9JfCXLB6k^e~uBwRvh)%*(Ms`{CAvPb>R_)g5t%5K>fPv3R zfUj8hN$r7i>E2>^7fZXZ$lYnhHMQ22U^Tho6jokDk|%pX-1RMp6NPW5P<8a~fat_{ z9xFQ{-DUcxkDS8Die*ecXIq{aolfaHwGeh9c4A0ESJohCQ zez0XV^mI#_#$r<{zL^#SB&;kiTprh;4Yp2O5ujGTvgAU?=38*;p`McM5brL~YzSRW zoF%%-+hBp9k2-~fCnqxdKfCD^htZ0ib%ICwt3p;$ne+qrWE6Ayft9G(HJ^-W6RcFM@ zFV5r8S2;h(GNLE`y)xsVb+487uCu7r)jXdoiS?ReX_Hl+v$=E$zQV84gGd4Yh!i?P z{zlVFMQzYZI%9k2%0mngvhm_(2zFuIb-T8|;sQ67INHHm=A2!yi*x^K(1`)uHCjB9 zLeJ5ZQy&ArTYmKR&>s6a!5mfmDW7vQM6x^R+5!D?X}~wEF=hH-fP;Fk#~vsD0j+V+ z=&)JWA5e1tg#8qxqx=H=Y#g>R54qa^Cu-#nEOInI{-RrcsuVid2tA*p{rgL{!1_P; z6q42_eHD-$rgw8%JgN0*zeR=9-Wv{8)=|U7uHCq~sW!oMNh;UE^FL&?=(&rSn)b-X zRIxw!q4QWueCfPd6kZT>tNvZvIPDh)G7Fy7y77H|<8F=21#o8Ip)}|9Vz~k|a*{PC>w_h8wWLpR*CkXSgMbFV{SeOK0_{OZzvupM@>@=IDL`6pT5pE6)) zxGEA}jkTPen19NsyyClbFKAn;@TaG)eFP3jxB^4m=GOq zeK{``Eh*(|*+yUe9LE55?EQCV(#Jpj*-7e`f&VW%`Tte-^}i3UxW%zC{4C_zbZD-X zom{$EufQs2qJOHCE=yzmL^wa)9-UW3<@NI)t_h%04_B|6X_9XRw})2^Maha;zb)~) zK9N#)?j!H5@_=4YC8KLc627o#M__=PwUAkwTw_1{;r_hVCvVMorZv(RV{@wbZvKjm zMP*en0NX$jc7aa7?#^6|-);)p)PN7DhOlNQz4>c>-SJZWh*k5G z{#7$VgK2UF|6n=Qd@jQ1`nmiHVmEURB1YXpBjwBkik1S+QlmEa)ykOo@5Zf^k=l2z zTjLcXcLc}#3`W^9s6&KUmvk=%775&fv=@EZqu~apq(TODvD~F5v zZ^(M*dl8mM@A3gof%8t9Y$fiuo*U4Yxyc8AdS_ap9PuWGJ2ma+{g8o$2f*SZc6)Dk zbAjPt;~>gnRnqS<$s%+hDd2Ry#Axe$`hKuc$+(r+Emra550j#jkGADqUcEOD@|$?% zpGmG+#G`0nYtyAjT@8ur=iNcdX(dZ zFKKW6{c|q+dEUq*&{n_97shkHkDaBy&Q6>q?lagl#w5BUisXzfN{GU=pW5lUqFKe` z2X~BEO7y=#=2l9@Wg*TIrT#B`otcZ+IxFa;?&`fx4H$@7ACy)sq9(U-jxX#SQQzBA z$uh1P-*w3&O1RUTq`0Xs9h4TP>0+NXAg=7!d=NY^?&MjmCa@Z^ z@Pjmow~``u%*Dr3PO($dn18vZqoJ#|*7=7D;j9PPM6hML>Z40(?nfm8Pkh1$i26UC)}hY;JR^{IPViG zzfE3L=KJi8>q`!=I(T`6i$~3eO>TRc$@fli%Y2vU#9hyRgs9G_hY-osq_MAF2P641 zgY|6Y3&YcoE-VbYzpc?SMMOW_b+1_y$b9ty>2y=v{&O{B%%=j^jj4feD)dEsld)N` zosmH4GUS5RQ|IyoyEohuU&|zgt!8dn?n|OW$F^UwV{;fgeGE{p4#j3~#)mU#b#x+n z1-j{MGCWIG7Rgf<6!J$VanIi)>Cp?bp5D&czWZ`-Y5O?$V*T-STwb> ze&nFk1P)5^r#_$=@G87DHsRLNvZeqD${XW%DdTRsGoxM1J~O~T^Srlqwh`Vid8t1a zG{jYG#aZ{!Wl0#XvKby2aJ#B2++&h0SkIFxZW9jQTBU9WF@X{k-ceTvNRF za{`^Wi~@^@l=#!LPS2k%g0#bV(dPFAZ{Qjc+zJOL(mi(i_+Xme#Iz_UdG!{YcH|FF2mN5sJSxf_OA9nq6}KB4*UWj3hzFB7pA$!gZfRp({<0Q2+Rrlv#Ri4XRtrc<~ z6Gl$t1m0|_(wt$~vl-~gwVz>q^Q2K#=L@_%Xm23AncEl+e#!B>}%);NUPOIr+Z9|K?$ZCEanZ~8M|n6PONTeg=0$L(01 zFQ3m6@Y<<5Ex3}_P3@87e#iISIh^itYKwMvrOuP%EvSnLe7$e1qJPPH zJFG1YMxhQjG?yH@D>9(fyASGd(TMAdYi(j`g&axxgh7taprx!M3hu)`-#Q+4Z-Ik^ zrs9_Z-@8F?r=cmq4A5c7Pw2rWmIgYBp#o3)#i6{#hv2G}@vPWfoR~ZswwbvNIX{7h zfDhvledq{uE(UE9tkmTtk5QB){ZG&=6E++jI6QZ6qLZuu|s z^oue3%QQB9{v~xdlS%KBF`$;P(Q;lP+g{>08fdg~m)TF>n^h9dnRloJ2^Pvtcz$2Q zmf@vm7Y%u*Og|J&3kn+_m>Zwx#sI1pfya5ehnqoR^4CFhCV=%)W~)lb#O#$w|5Hm^-Qh zo1NA#jMQ^>tTXAA+9>5Vvql z$WCWuLpefQ@$cl)%b8hn0%YKq;K4h?Y=m-(6v+WiqW=%j5{7 zSGTXEO_?$)%RkoDfp$KTO;%QnJ9A~|t3~~GHDH-aH1$?pg`xWl`0LmZjdk=eeEJjE_)hR-t%=&0UyV#hKw#i!# zS?`rg{?40@120~+uo%ltuQTqhyV=LGk$t~|zl)|Jx9Ls5Rl zVzUi#=A}iY3Bw{AU>4YfL>Ic?8Pm`SJ|5^pUy0duLXLZXT@dttygWx9y+&UaOl~Lh z`JZWapJ_%%GY9jrYfe9O8fYWj2J&Agsz^0gNat8AOSSk!%L7h{qH zXrdL>17}y$325TV5%2WO%{|P?#<&|PA~=1Sh#9ZOIp52RYiaL0JgN8*=n+2zl7b6e z=^-r_F@PK!Wk8M&KjA^pcum!n*7FtPA={%R)bO*H=+-(&I0%b05%F-(;ThM}{S58# z?g7FV=EYM!pSSi3?Ds^8dSlR)L{mv!#&xw=5;e-><|1fuy)1MhuqzRSIyY)Tw2=6E zS)Nbj4@N^b5|dG9_Se&7P7f~Duutzxf848Db+c@z`f*iKaC4krdcES)_8n6vTKH#e zcjmX%_v915l}enC&Wt+(oA1}|<$vlzmjv+s{KZB9if+Uw<*fB^?`Wh|;X=V1{yi$q z%ET;5Lv7X6m9ilj!}5MK)>r1RQ3N4uRU69#eu>5amo&IfX%n-M4~BP@^I>OF(Za5& zj|S9>BkbT-boPvbkeyt^YzN6>i~b^xY=!Cu`-k@E8=rm8SI3m$oRwOu-L`dgcF8tg zhr*nCXPJHDL9Dg~K1$1vQi``3)34_Rw%9 z1Nrb4VB zrWU4be?y{Lw;`3Kn%utxj(CVjlJlKVlZ*hxL;PmEHpDB=ETiO8g-}Sx{jdmrlW{3M z7gp7`xwJbJBKww&*Uq|5sU;)$8w;m-_U*h$UKVC^jykltuzU%y#|ya<>d4fhL!-sj zvwa%cOaen)3aGH>gv8JZFBnR$@r&@0laE003R8T;`%zhHG_!~4;Ji$(^o{6}d0mLC6Q|j)bN#fSk^Jz9c&rG8 zHD`?s4U6<@^pha7qdC)7OUkN`n%d3v<~REu?;h?YRwUecVbbRSx6%jd1wkk=$Nh=+ zBrf8QFAANiWg!SLqx1fE&yf@og&y zyR~-~TEi`)kium~(>`S*lctuCWYUgr3E9|9=Y=_zv*oz-H4VsF%KJD+d5c1*PPmq0 zGJ1jmj`gf|SX)foMU*l!2Jm<#5U-Ic8hfq*)tDG--cQ2-=E~3K1)Kf-1xQdSvuXJN zdA`S?vTw1DM2%uYeJAw&o$@Q9S?H$0 zc8Me}D7&2b+ZT9TTXDk`siXpWc0^E+ad~d(J$qL}LJ29Zot2Yoi9=0&MCib@kBdw6 zX1|+mpn(d}bO6tYhiS2#uK5q^&Ythrp5OEJjvJ{PL})d-Vk5TwV+?=`LSVx-;+hyN zxA1Wux~PB+nFyX-9G3p1%I}&Q*?pfN5pC8e^MYMEUe&Ux`x7M8Q~6s)yE6V;E!VY9 z<+dD!c9toN@}2=~@<(PFX~7ZVGOdF*kkcP*2F(}R?*RgR&S?Y++T&;}H+GwG>yrU+ z(xNw#hg5q(anyyMv~8BCO@obIDcw0nADuOkhKRgeZd*xzK0Vpvdux>1S1%phG+qN) z5{JW3R1JejBJk1!1f-1-oxj3|4ONn+Wf(FjT%DiB)OH4L@CVp4sw{+g1ZW?$)3M$h z%cA|N7R=-Py~lvHY7i-b&8MQez-M%>YZj%kd_CnJnKTIYj{4^OY=`S5TYi-}zPHpR z)JZ;Go>JpUCsmI#0^Mmka22qGhJPPa(TPDJNhl(1+G^V z&vyDg%XjUJlkS(0qG{>(RUu39QcfQA9u##hC3BY1kEmgkNIlZ_vm+ncrKiES#0p)K zdBPyVD+Popws1)MbnL=cAIf#}Hn$l!=~wy$DU&`?j;6J@0VN8Xz0hlf9&~5*kxqw>61qy#0^eGCIP!@&->)q(+v|d*dV&+)z>m7r4ZI{s^Q6HYI-1mj}I@ZKBy%)q00W0d@ ze&s*}EhEafc@rebc+?W$IBFl515!aqrE)Op^ zKe?bP*V^*=otYZ0)H7m<#pg1QCq&_&RiZS8IPQyxY*)m;E@o!Ey@2GJm&B$H-q;dd z5FYY&^rcQz^Q|2d%d^gKF>`jEbTg5jo4%axLnK~VUJ|7xr{%QfTQT0Bsxz&gwpz|h zn5gVUo_rW$BVY-UpnoIGZ=DmZ|Ni&o;g2lzwp7XpF{9Z0lDfgET!^|d@UqxJ5Jnf% zc<icwysR zpAB$t(|7&bfjl_&|DELW$YE(mc`LZ`Lb$WMfRlhb8~a>qMdP&#RNG^o%R}PJ)N=^_ z7Wnaqb&@h_H4=g|-MC{gv|x_PF0<}@65Pw1cL{)w#rM8+vPI~r0Q6a);Xg}cIL8wm z0e0=u-W>uGNBDPU+VHv@^TQ74@YN>I2S6iDR7=gV)}s&3Y%e~%tUF(w5QZCSB%2~~ zm-vXfL^=eab0F>$nw{x@9Ajf;m)F-K2>_FhKS+>&IkVaw9-uYA?chsv+=aNGt#4)< z?1Be+1G;kwSsmMY|JSM=&bo;qGM%dd13V6grA~aBNO7<{a!Wlx1+$@@EYki~u4Q#t zcD)dkC&e-|PMIkDiTsIFpER^)+-vC0^u{k9XUpTSs{eY8vkoQ0)`GK<*Z{vAo1MNh z{A8Gdl);izVQ>%{mW$T1NcsD_Z1bT#txm8rqNs~fKWE>hL=3=9T!sNA*@V`swkrRw zJcH|3rE78F@~jR^4r}?D`xZpNR%;QG=%%t?9MXR%J&PFU1o9>+EBv6Qz1t%P8qYMA z=Nu&~U%G0zFY}bInHIepslg+i8b&&Xz5x~VB!21%lo5pNMA1*T2g7Qsn=BPz;IwQ z-@CykoZwZS)hB$JI{k;YVb8alPID|L&5XM>^5B)v3=8qC;|3zJeO(W-3&>K<8*J=g zcC6?oW4RWW7_{5Q&~HD}Ic(a+)I0LEkWi)5baojltntUkJ1PbqJS{Xc0RMvqL!m literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/manifest.json b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/manifest.json new file mode 100644 index 000000000..c5b190bde --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/manifest.json @@ -0,0 +1,62 @@ +{ + "version": "2.0", + "job_id": "job_5184790cddae", + "data_id": null, + "source_file_name": "dl_10.pdf", + "processing_date": "2026-06-16T03:40:35.535316Z", + "processing": { + "page_count": 34, + "billing_status": "charged", + "cost": { + "micro_dollars": 51000, + "credits": 0.051 + }, + "timing": { + "started_at": "2026-06-16T03:39:35.098598+00:00", + "completed_at": "2026-06-16T03:40:35.529163+00:00", + "duration_ms": 60430 + }, + "stages": { + "timing_ms": { + "document.profile": 3886, + "pdf.extract.standard": 44922, + "toc.detect_range_llm": 1241, + "md.detect_toc": 1261, + "heading.merge_pre_pass_llm": 896, + "heading.hierarchy_llm_call": 2029, + "heading.hierarchy_llm": 2038, + "md.predict_headings": 3302, + "md.deferred_summaries": 6795, + "md.build_dataframe": 1, + "pdf.parse_md": 11556, + "document.cleanup_unreferenced_images": 0, + "document.compress_images": 0, + "worker.parse.document": 60372 + }, + "token_usage": { + "prompt_tokens": 5588, + "completion_tokens": 169, + "total_tokens": 5757 + } + } + }, + "statistics": { + "total_chunks": 25, + "text_chunks": 8, + "image_chunks": 11, + "table_chunks": 6, + "total_pages": null + }, + "HIERARCHY": { + "Deep Learning Transformer network": { + "Transformer network": { + "Transformer network - Attention- What it is?": {}, + "Transformer network - Positional encoding- What is?": {} + }, + "Training a Transformer (Seq2Seq model):": {}, + "Inference – autoregressive decoding:": {}, + "Transformer network - BERT - How to use BERT (Fine-tuning)": {}, + "Transformer network - References ii": {} + } + } +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-0 Attention Matrix.html b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-0 Attention Matrix.html new file mode 100644 index 000000000..185812596 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-0 Attention Matrix.html @@ -0,0 +1,69 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
Words attended toTheanimaldidn'tcrossthestreet
The0.20.30.10.10.20.1
animal0.10.40.20.10.10.1
didn't0.10.20.40.10.10.1
words crossing0.10.10.10.40.20.1
the0.10.20.10.20.30.1
street0.10.10.20.20.10.3
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-1 EEA Agreement.html b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-1 EEA Agreement.html new file mode 100644 index 000000000..d9d564577 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-1 EEA Agreement.html @@ -0,0 +1,70 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
The agreement on the European Economic Area was signed in August 1992
L'
accord(not labeled)
sur(not labeled)
la(not labeled)
zone(not labeled)
économique(not labeled)
européenne(not labeled)
a(not labeled)
été(not labeled)
signé(not labeled)
en(not labeled)
août(not labeled)
1992(not labeled)
.(not labeled)
<end>(not labeled)
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-2 Constant Matrix.html b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-2 Constant Matrix.html new file mode 100644 index 000000000..11dd4e597 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-2 Constant Matrix.html @@ -0,0 +1,201 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
input dimension01234567891011
0-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
1-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
2-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
3-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
4-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
5-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
6-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
7-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
9-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
10-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
11-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8-0.8
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-3 Low Values Table.html b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-3 Low Values Table.html new file mode 100644 index 000000000..04c53d1bc --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-3 Low Values Table.html @@ -0,0 +1,139 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
Position0123456789101112131415
48LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
45LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
39LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
36LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
33LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
30LowLowLowLowLowLowLowLowLowLowLowLowLowLowLowLow
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-4 Color Pattern Table.html b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-4 Color Pattern Table.html new file mode 100644 index 000000000..4e534c684 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-4 Color Pattern Table.html @@ -0,0 +1,158 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
Position0123456789101112131415
0BlueOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
1BlueOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
2Dark BlueOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
3OrangeOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
4OrangeOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
5Medium BlueOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
6Medium OrangeOrangeLight BlueRedLight BlueLight BlueLight BlueRedLight BlueDark BlueLight BlueLight BlueLight BlueLight BlueLight BlueLight Blue
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-5 BERT Embeddings.html b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-5 BERT Embeddings.html new file mode 100644 index 000000000..17e077a7a --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-deep-learning-transformer-network/tables/table-5 BERT Embeddings.html @@ -0,0 +1 @@ +
Input[MASK][MASK]
[CLS]mydogiscute[SEP]helikesplay##ing[SEP]
Token Embeddings $E_{[CLS]}$ $E_{my}$ $E_{[MASK]}$ $E_{is}$ $E_{cute}$ $E_{[SEP]}$ $E_{he}$ $E_{[MASK]}$ $E_{play}$ $E_{##ing}$ $E_{[SEP]}$
Sentence Embedding+++++++++++
$E_A$ $E_A$ $E_A$ $E_A$ $E_A$ $E_A$ $E_B$ $E_B$ $E_B$ $E_B$ $E_B$
Transformer Positional Embedding+++++++++++
$E_0$ $E_1$ $E_2$ $E_3$ $E_4$ $E_5$ $E_6$ $E_7$ $E_8$ $E_9$ $E_{10}$
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/chunks.json b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/chunks.json new file mode 100644 index 000000000..83c0a8416 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/chunks.json @@ -0,0 +1,2186 @@ +{ + "chunks": [ + { + "chunk_id": "84debb14-fe3d-57f3-a62e-868f197a7ed3", + "type": "text", + "content": "Carlo Ciliberto\nDepartment of Computer Science, UCL\nOctober 5, 2018", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory", + "metadata": { + "length": 67, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Carlo", + "Ciliberto", + "Department", + "Computer", + "Science", + "UCL", + "October", + "2018" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "29f78cbc-320c-5371-ac77-cf126a372b55", + "type": "text", + "content": "▶ Class times: Fridays 14:00 - 15:30 $^{1}$\n▶ Location: Ground Floor Lecture Theater, Wilkins Building $^{2}$\n▶ Office hours: (Time TBA), 3rd Floor Hub room, CS Building, 66 Gower street.\nTA: Giulia Luise\nWebsite: ccilib er.github.io/intro-stl\nemail(s): ccilib er@gmail.com, g.luise.16@ucl.ac.uk\n▶ Workload: 2 assignments (50%) and a final exam (50%). Final exam requires to choose 3 problems out of 6. At least one problem from each “sides” of this course (RKHS or SLT) \\*must\\* be chosen.", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Administrative Info", + "metadata": { + "length": 490, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Class", + "times", + "Fridays", + "14", + "00", + "15", + "30", + "Location", + "Ground", + "Floor", + "Lecture", + "Theater", + "Wilkins", + "Building", + "Office", + "hours", + "Time", + "TBA", + "3rd", + "Hub", + "room", + "CS", + "66", + "Gower", + "street", + "TA", + "Giulia", + "Luise", + "Website", + "ccilib", + "er", + "github", + "io", + "intro", + "stl", + "email", + "gmail", + "luise.16", + "ucl", + "ac", + "uk", + "Workload", + "assignments", + "50%", + "final", + "exam", + "Final", + "requires", + "choose", + "problems", + "At", + "problem", + "sides", + "RKHS", + "SLT", + "chosen" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f7e9b473-e838-5303-9b4b-10bf784b0422", + "type": "text", + "content": "Main resources for the course:\nClasses\nSlides\nBooks and other Resources:\nS. Shalev-Shwartz and S. Ben-David Understanding Machine Learning: From Theory to Algorithms (Online Book). Cambridge University Press, 2014.\n▶ O. Bousquet, S. Boucheron and G. Lugosi Introduction to Statistical Learning Theory (Tutorial).\nT. Poggio and L. Rosasco course slides and videos from MIT 9.520: Statistical Learning Theory and Applications.\nP. Liang course notes from Stanford CS229T: Statistical Learning Theory.", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Course Material", + "metadata": { + "length": 497, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Main", + "resources", + "Classes", + "Slides", + "Books", + "Resources", + "Shalev", + "Shwartz", + "Ben", + "David", + "Understanding", + "Machine", + "Learning", + "From", + "Theory", + "Algorithms", + "Online", + "Book", + "Cambridge", + "University", + "Press", + "2014", + "Bousquet", + "Boucheron", + "Lugosi", + "Introduction", + "Statistical", + "Tutorial", + "Poggio", + "Rosasco", + "slides", + "videos", + "MIT", + "9.520", + "Applications", + "Liang", + "notes", + "Stanford", + "CS229T" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "83696650-7104-5ddd-bc17-8eb72d435a59", + "type": "text", + "content": "▶ Linear Algebra: familiarity with vector spaces, matrix operations (e.g. inversion, singular value decomposition (SVD)), inner products and norms, etc.\n▶ Calculus: limits, derivatives, measures, integrals, etc.\n▶ Probability Theory: probability distributions, conditional and marginal distribution, expectation, variance, etc.", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Prerequisites", + "metadata": { + "length": 327, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Linear", + "Algebra", + "familiarity", + "vector", + "spaces", + "matrix", + "operations", + "inversion", + "singular", + "decomposition", + "SVD", + "products", + "norms", + "Calculus", + "limits", + "derivatives", + "measures", + "integrals", + "Probability", + "Theory", + "probability", + "distributions", + "conditional", + "marginal", + "distribution", + "expectation", + "variance" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "89e9a239-be09-5517-8475-22cd2ed2d51c", + "type": "text", + "content": "SLT addresses questions related to:\nWhat does it mean for an algorithm to learn.\nWhat we can/cannot expect from a learning algorithm.\n▶ How to design computationally & statistically efficient algorithms.\nWhat to do when a learning algorithm does not work...\nSLT studies theoretical quantities that we don't have access to:\nIt tries to bridge the gap between the unknown functional relations governing a process and our (finite) empirical observations of it.", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)", + "metadata": { + "length": 457, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "SLT", + "addresses", + "questions", + "related", + "What", + "algorithm", + "learn", + "expect", + "learning", + "How", + "design", + "computationally", + "statistically", + "efficient", + "algorithms", + "work", + "studies", + "theoretical", + "quantities", + "don", + "access", + "It", + "bridge", + "gap", + "unknown", + "functional", + "relations", + "governing", + "process", + "finite", + "empirical", + "observations" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "44b7e150-420d-51d0-9ef0-b6d363ec5ac5", + "type": "table", + "content": "
Living area (feet $^{2}$ )Price (1000$s)
2104400
1600330
2400369
1416232
3000540
", + "path": "tables/table-0 Housing Data.html", + "metadata": { + "length": 281, + "summary": "table-1\nThe table shows living area in square feet and corresponding house prices in thousands of dollars for several properties.", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "tables/table-0 Housing Data.html", + "keywords": [ + "Living area", + "Price", + "Housing" + ], + "tokens": [] + } + }, + { + "chunk_id": "df07e316-9d81-5a84-b29d-2556a47ed4e7", + "type": "image", + "content": "\nWhat to do when a learning algorithm does not work...
0\n[images/image-1-Housing Prices Scatter Plot.jpg]\n", + "path": "images/image-1-Housing Prices Scatter Plot.jpg", + "metadata": { + "length": 98, + "summary": "image-1\nChart Title: housing prices\nX-Axis Label: square feet (Range: 500 to 5000)\nY-Axis Label: price (in $1000) (Range: 0 to 1000)\nData Points: Blue 'x' markers representing individual data points.\nNotable Patterns: The chart displays a positive correlation between house size and price; as square footage increases, the price tends to increase. Most data points are clustered between 1000-3000 square feet and $200k-$600k.", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-1-Housing Prices Scatter Plot.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e3b44510-fade-50fc-b09c-082b23a357d9", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
square feetprice (in $1000)
1000200
1200250
1400300
1600350
1800400
2000450
2200500
2400550
2600600
2800650
3000700
3200650
3400600
3600550
3800500
4000450
4200400
4400350
4600300
4800250
5000200
", + "path": "tables/table-1 House Price Data.html", + "metadata": { + "length": 1410, + "summary": "table-2\nThe table shows house prices (in $1000) vs. square footage. Price rises from $200k at 1000 sq ft to $700k at 3000 sq ft, then declines symmetrically back to $200k at 5000 sq ft.", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "tables/table-1 House Price Data.html", + "keywords": [ + "square feet", + "price", + "trend" + ], + "tokens": [] + } + }, + { + "chunk_id": "7fb3602f-2fd2-5e4a-8add-fbb72a65566d", + "type": "text", + "content": "[tables/table-0 Housing Data.html]\n\n\nWhat to do when a learning algorithm does not work...
0\n[images/image-1-Housing Prices Scatter Plot.jpg]\n\n
\nscatter plot\n\n[tables/table-1 House Price Data.html]\n\n
\nImage credits: coursera", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)/Motivations and Examples: Regression", + "metadata": { + "length": 290, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "What", + "learning", + "algorithm", + "work", + "details", + "summary", + "scatter", + "plot", + "Image", + "credits", + "coursera" + ], + "keywords": [], + "connect_to": [ + { + "target": "44b7e150-420d-51d0-9ef0-b6d363ec5ac5", + "relation": "embeds", + "ref": "[tables/table-0 Housing Data.html]", + "position": { + "start": 0, + "end": 34 + } + }, + { + "target": "df07e316-9d81-5a84-b29d-2556a47ed4e7", + "relation": "embeds", + "ref": "[images/image-1-Housing Prices Scatter Plot.jpg]", + "position": { + "start": 102, + "end": 150 + } + }, + { + "target": "e3b44510-fade-50fc-b09c-082b23a357d9", + "relation": "embeds", + "ref": "[tables/table-1 House Price Data.html]", + "position": { + "start": 195, + "end": 233 + } + } + ] + } + }, + { + "chunk_id": "01ba1b9a-cf54-523f-b929-fd0c48191149", + "type": "image", + "content": "\n
0\n[images/image-2-__details_.jpg]\n", + "path": "images/image-2-__details_.jpg", + "metadata": { + "length": 55, + "summary": "image-2\n
0", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-2-__details_.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "a2b7f45b-2661-5cf5-b628-22e2afc8baa4", + "type": "image", + "content": "\ngraph LR
0\n[images/image-3-graph LR _.jpg]\n", + "path": "images/image-3-graph LR _.jpg", + "metadata": { + "length": 53, + "summary": "image-3\ngraph LR
0", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-3-graph LR _.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "12e16a3b-1414-5c9e-9c3a-bcb0e96a077d", + "type": "text", + "content": "Spam detection: Automatically discriminate spam vs non-spam e-mails.\n\n
0\n[images/image-2-__details_.jpg]\n\n
\nflowchart\n```mermaid\ngraph LR\nA[\"Stacked Document\"] --> B[\"{ legit, spam}\"]\n```\n
\nImage Classification\n\ngraph LR
0\n[images/image-3-graph LR _.jpg]\n\n
\ntext_image\nf :\n→ {dog, not dog}\n
", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)/Motivations and Examples: Binary Classification", + "metadata": { + "length": 391, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Spam", + "detection", + "Automatically", + "discriminate", + "spam", + "mails", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "LR", + "Stacked", + "Document", + "legit", + "Image", + "Classification", + "text", + "image", + "dog" + ], + "keywords": [], + "connect_to": [ + { + "target": "01ba1b9a-cf54-523f-b929-fd0c48191149", + "relation": "embeds", + "ref": "[images/image-2-__details_.jpg]", + "position": { + "start": 92, + "end": 123 + } + }, + { + "target": "a2b7f45b-2661-5cf5-b628-22e2afc8baa4", + "relation": "embeds", + "ref": "[images/image-3-graph LR _.jpg]", + "position": { + "start": 287, + "end": 318 + } + } + ] + } + }, + { + "chunk_id": "c5a45a02-6fb4-5f90-b557-4c71066aeef1", + "type": "image", + "content": "\n→ {dog, not dog} ### Motivations and Examples: Multi-class Classification0\n[images/image-4-Image Type_ Photo.jpg]\n", + "path": "images/image-4-Image Type_ Photo.jpg", + "metadata": { + "length": 108, + "summary": "image-4\nTitle: Image Grid of Objects and Scenes\n\n- Row 1 (baseball): Images include a player in uniform, wooden bats, a baseball field diagram, a batter swinging, and novelty baseball bats.\n- Row 2 (dog): Photos show dogs with life rings, on beaches, lying on blankets, groups of dogs, close-ups, and one dog in gear.\n- Row 3 (basketball-hoop): Various basketball hoops including backboards, nets, rims, and scenes with balls; includes text \"America Conference\".\n- Row 4 (Kayak): Kayaks in water, people kayaking near waterfalls, colorful kayaks lined up, and kayaks loaded on vehicles.\n- Row 5 (traffic light): Traffic lights at intersections, digital displays, pedestrian signals, and traffic control panels.", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-4-Image Type_ Photo.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "74dc7b10-e74b-5e26-babb-75698fbf50f2", + "type": "text", + "content": "Identify the category of the object depicted in an image.\nExample: Caltech 101\n\n→ {dog, not dog} ### Motivations and Examples: Multi-class Classification0\n[images/image-4-Image Type_ Photo.jpg]", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)/Motivations and Examples: Multi-class Classification", + "metadata": { + "length": 186, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Identify", + "category", + "object", + "depicted", + "image", + "Example", + "Caltech", + "101", + "dog", + "Motivations", + "Examples", + "Multi", + "class", + "Classification0" + ], + "keywords": [], + "connect_to": [ + { + "target": "c5a45a02-6fb4-5f90-b557-4c71066aeef1", + "relation": "embeds", + "ref": "[images/image-4-Image Type_ Photo.jpg]", + "position": { + "start": 155, + "end": 193 + } + } + ] + } + }, + { + "chunk_id": "4a7c125e-3ceb-542b-9dbe-590032b2a172", + "type": "image", + "content": "\nIdentify the category of the object depicted in an image.
0\n[images/image-5-Visual Data Grid.jpg]\n", + "path": "images/image-5-Visual Data Grid.jpg", + "metadata": { + "length": 102, + "summary": "image-5\nA large grid of small thumbnail images displaying a diverse collection of subjects. The top half primarily features nature scenes, including flowers, plants, and landscapes. The bottom half contains a wide variety of objects such as tools, musical instruments, vehicles, furniture, clothing, and household items.", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-5-Visual Data Grid.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "7fb0692d-5766-5652-8993-f710d8ef80ef", + "type": "text", + "content": "Scaling things up: detect correct object among thousands of categories. ImageNet Large Scale Visual Recognition Challenge\n\nIdentify the category of the object depicted in an image.
0\n[images/image-5-Visual Data Grid.jpg]\n\n
\ntext_image\nGrid of numerous emoji icons and text, likely representing a collection of various application or service types.\n
", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)/Motivations and Examples: Multi-class Classification_2", + "metadata": { + "length": 388, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Scaling", + "things", + "detect", + "correct", + "object", + "thousands", + "categories", + "ImageNet", + "Large", + "Scale", + "Visual", + "Recognition", + "Challenge", + "Identify", + "category", + "depicted", + "image", + "details", + "summary", + "text", + "Grid", + "numerous", + "emoji", + "icons", + "representing", + "collection", + "application", + "service", + "types" + ], + "keywords": [], + "connect_to": [ + { + "target": "4a7c125e-3ceb-542b-9dbe-590032b2a172", + "relation": "embeds", + "ref": "[images/image-5-Visual Data Grid.jpg]", + "position": { + "start": 192, + "end": 229 + } + } + ] + } + }, + { + "chunk_id": "3642f12c-21b7-567e-876a-26843beffb08", + "type": "image", + "content": "\nGrid of numerous emoji icons and text, likely representing a collection of various application or service types.
0\n[images/image-6-Grid of nu.jpg]\n", + "path": "images/image-6-Grid of nu.jpg", + "metadata": { + "length": 157, + "summary": "image-6\nGrid of numerous emoji icons and text, likely representing a collection of various application or service types.
0", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-6-Grid of nu.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "c2c86116-9667-5259-9673-c2c211d97e44", + "type": "image", + "content": "\n
0\n[images/image-7-_details_ .jpg]\n", + "path": "images/image-7-_details_ .jpg", + "metadata": { + "length": 54, + "summary": "image-7\n
0", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-7-_details_ .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "9069a803-e353-54ec-86fb-e0e460e392a0", + "type": "image", + "content": "\nCoffee mug Speech Recognition \"Ok Google\"0\n[images/image-8-Coffee mug.jpg]\n", + "path": "images/image-8-Coffee mug.jpg", + "metadata": { + "length": 76, + "summary": "image-8\nCoffee mug Speech Recognition \"Ok Google\"0", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-8-Coffee mug.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "3d6e9cd9-5f01-5937-abe7-61f3ae4302b4", + "type": "image", + "content": "\nNewspaper Speech Recognition \"Ok Google\"0\n[images/image-9-Newspaper .jpg]\n", + "path": "images/image-9-Newspaper .jpg", + "metadata": { + "length": 75, + "summary": "image-9\nNewspaper Speech Recognition \"Ok Google\"0", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-9-Newspaper .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "010ac87f-546f-5206-a4ca-154e96ce8229", + "type": "image", + "content": "\nMovie Ranking Speech Recognition \"Ok Google\"0\n[images/image-10-Movie Rank.jpg]\n", + "path": "images/image-10-Movie Rank.jpg", + "metadata": { + "length": 80, + "summary": "image-10\nMovie Ranking Speech Recognition \"Ok Google\"0", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-10-Movie Rank.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "50f0b756-3174-5261-84b6-e0269f6ee370", + "type": "image", + "content": "\nuser:127
0\n[images/image-11-user_127 _.jpg]\n", + "path": "images/image-11-user_127 _.jpg", + "metadata": { + "length": 54, + "summary": "image-11\nuser:127
0", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-11-user_127 _.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "326ffa21-4abb-50ef-88be-5ca242cb4721", + "type": "image", + "content": "\n
## Formulating The Learning Problem0\n[images/image-12-_details_ .jpg]\n", + "path": "images/image-12-_details_ .jpg", + "metadata": { + "length": 81, + "summary": "image-12\n
## Formulating The Learning Problem0", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-12-_details_ .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "500c756d-2e8a-5e5a-a3f8-dc178293d610", + "type": "image", + "content": "\nnatural_image ## Formulating The Learning Problem0\n[images/image-13-_summary_n.jpg]\n", + "path": "images/image-13-_summary_n.jpg", + "metadata": { + "length": 104, + "summary": "image-13\nnatural_image ## Formulating The Learning Problem0", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-13-_summary_n.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "fa1854a8-3cee-50e9-888c-b468deb65629", + "type": "text", + "content": "Image Captioning\n(also Localization Segmentation Classification)\n\nGrid of numerous emoji icons and text, likely representing a collection of various application or service types.
0\n[images/image-6-Grid of nu.jpg]\n\n
\nnatural_image\nCat sitting at a desk with papers and a cup, viewed through a window with curtains (no visible text or symbols)\n
\n\n
0\n[images/image-7-_details_ .jpg]\n\n
\nnatural_image\nA cat sitting at a desk with a book, viewed through a window (no visible text or symbols)\n
\nGray striped cat\nBusiness suit\nCoffee mug\nNewspaper\nMovie Ranking\nNETFLIX\nuser:127\n\nCoffee mug Speech Recognition \"Ok Google\"0\n[images/image-8-Coffee mug.jpg]\n\n\nNewspaper Speech Recognition \"Ok Google\"0\n[images/image-9-Newspaper .jpg]\n\n\nMovie Ranking Speech Recognition \"Ok Google\"0\n[images/image-10-Movie Rank.jpg]\n\nSpeech Recognition \"Ok Google\"\n\nuser:127
0\n[images/image-11-user_127 _.jpg]\n\n
\nnatural_image\nAbstract blue waveform visualization with no text or symbols\n
\nProtein Folding\n\n
## Formulating The Learning Problem0\n[images/image-12-_details_ .jpg]\n\n\nnatural_image ## Formulating The Learning Problem0\n[images/image-13-_summary_n.jpg]", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)/Motivations and Examples: Structured Prediction Image Captioning", + "metadata": { + "length": 1307, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Image", + "Captioning", + "Localization", + "Segmentation", + "Classification", + "Grid", + "numerous", + "emoji", + "icons", + "text", + "representing", + "collection", + "application", + "service", + "types", + "details", + "summary", + "natural", + "image", + "Cat", + "sitting", + "desk", + "papers", + "cup", + "viewed", + "window", + "curtains", + "visible", + "symbols", + "cat", + "book", + "Gray", + "striped", + "Business", + "suit", + "Coffee", + "mug", + "Newspaper", + "Movie", + "Ranking", + "NETFLIX", + "user", + "127", + "Speech", + "Recognition", + "Ok", + "Google", + "Abstract", + "blue", + "waveform", + "visualization", + "Protein", + "Folding", + "Formulating", + "The", + "Learning", + "Problem0" + ], + "keywords": [], + "connect_to": [ + { + "target": "3642f12c-21b7-567e-876a-26843beffb08", + "relation": "embeds", + "ref": "[images/image-6-Grid of nu.jpg]", + "position": { + "start": 190, + "end": 221 + } + }, + { + "target": "c2c86116-9667-5259-9673-c2c211d97e44", + "relation": "embeds", + "ref": "[images/image-7-_details_ .jpg]", + "position": { + "start": 411, + "end": 442 + } + }, + { + "target": "9069a803-e353-54ec-86fb-e0e460e392a0", + "relation": "embeds", + "ref": "[images/image-8-Coffee mug.jpg]", + "position": { + "start": 715, + "end": 746 + } + }, + { + "target": "3d6e9cd9-5f01-5937-abe7-61f3ae4302b4", + "relation": "embeds", + "ref": "[images/image-9-Newspaper .jpg]", + "position": { + "start": 791, + "end": 822 + } + }, + { + "target": "010ac87f-546f-5206-a4ca-154e96ce8229", + "relation": "embeds", + "ref": "[images/image-10-Movie Rank.jpg]", + "position": { + "start": 871, + "end": 903 + } + }, + { + "target": "50f0b756-3174-5261-84b6-e0269f6ee370", + "relation": "embeds", + "ref": "[images/image-11-user_127 _.jpg]", + "position": { + "start": 957, + "end": 989 + } + }, + { + "target": "326ffa21-4abb-50ef-88be-5ca242cb4721", + "relation": "embeds", + "ref": "[images/image-12-_details_ .jpg]", + "position": { + "start": 1170, + "end": 1202 + } + }, + { + "target": "500c756d-2e8a-5e5a-a3f8-dc178293d610", + "relation": "embeds", + "ref": "[images/image-13-_summary_n.jpg]", + "position": { + "start": 1275, + "end": 1307 + } + } + ] + } + }, + { + "chunk_id": "1fee5460-93df-5326-ae84-19ed917a5e29", + "type": "text", + "content": "", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ba241116-ef0c-54c9-969e-815f74640826", + "type": "text", + "content": "", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5ae7f40e-9137-565a-b028-e374630eb7ff", + "type": "text", + "content": "▶ X input and Y output spaces.\n▶ ρ unknown distribution on $X \\times Y$ .\n$\\ell : \\mathcal{Y} \\times \\mathcal{Y} \\to \\mathbb{R}$ a loss function measuring the discrepancy $\\ell(y, y')$ between any two points $y, y' \\in \\mathcal{Y}$ .\nWe would like to minimize the expected risk\n$$\n\\underset {f: \\mathcal {X} \\to \\mathcal {Y}} {\\text { minimize }} \\mathcal {E} (f) \\quad \\mathcal {E} (f) = \\int_ {\\mathcal {X} \\times \\mathcal {Y}} \\ell (f (x), y) d \\rho (x, y)\n$$\nThe expected prediction error incurred by a predictor $^{3}$ $f : X \\to Y$ .", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Main ingredients:", + "metadata": { + "length": 541, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "input", + "output", + "spaces", + "unknown", + "distribution", + "times", + "ell", + "mathcal", + "mathbb", + "loss", + "function", + "measuring", + "discrepancy", + "points", + "We", + "minimize", + "expected", + "risk", + "underset", + "text", + "quad", + "int", + "rho", + "The", + "prediction", + "error", + "incurred", + "predictor" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "72758b2f-862d-5730-afd6-66ea5f213ad5", + "type": "text", + "content": "Linear Spaces\nVectors\nMatrices\nFunctions\n\"Structured\" Spaces\n▶ Strings\n▶ Graphs\nProbabilities\n▶ Points on a manifold\n▶ ...", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Input Space Linear Spaces", + "metadata": { + "length": 122, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Linear", + "Spaces", + "Vectors", + "Matrices", + "Functions", + "Structured", + "Strings", + "Graphs", + "Probabilities", + "Points", + "manifold" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "43e87468-5a10-5f4a-a1dd-8a05a8902a1b", + "type": "text", + "content": "Linear Spaces, e.g.\n▶ $\\mathcal{Y} = R$ regression\n▶ $Y = \\{1, \\ldots, T\\}$ classification\n$\\mathcal{Y} = \\mathbb{R}^{T}$ multi-task\n\"Structured\" Spaces, e.g.\n▶ Strings\nGraphs\nProbabilities\nOrders (i.e. Ranking)\n▶ ...", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Output Space", + "metadata": { + "length": 217, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Linear", + "Spaces", + "mathcal", + "regression", + "ldots", + "classification", + "mathbb", + "multi", + "task", + "Structured", + "Strings", + "Graphs", + "Probabilities", + "Orders", + "Ranking" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9d88d4eb-79b4-5537-8153-b394ceafca5a", + "type": "text", + "content": "Informally: the distribution $\\rho$ on $X \\times Y$ encodes the probability of getting a pair $(x, y) \\in \\mathcal{X} \\times \\mathcal{Y}$ when observing (sampling from) the unknown process.\nThroughout the course we will assume $\\rho(x,y)=\\rho(y|x)\\rho_{\\mathcal{X}}(x)$\n▶ $\\rho_{\\mathcal{X}}(x)$ marginal distribution on X.\n▶ $\\rho(y|x)$ conditional distribution on Y given $x \\in X$ .", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Probability Distribution", + "metadata": { + "length": 385, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Informally", + "distribution", + "rho", + "times", + "encodes", + "probability", + "pair", + "mathcal", + "observing", + "sampling", + "unknown", + "process", + "Throughout", + "assume", + "marginal", + "conditional" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "09a454e5-02d8-5366-83a9-616342f1e42a", + "type": "text", + "content": "$\\rho(y|x)$ characterizes the relation between a given input x and the possible outcomes y that could be observed.\nIn noisy settings it represents the uncertainty in our observations.\nExample: $y = f_{*}(x) + \\epsilon$ , with $f_{*} : X \\to R$ the “true” function and $\\epsilon \\sim \\mathcal{N}(0, \\sigma)$ Gaussian distributed noise. Then:\n$$\n\\rho (y | x) = \\mathcal {N} (f _ {*} (x), \\sigma)\n$$", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Conditional Distribution", + "metadata": { + "length": 396, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "rho", + "characterizes", + "relation", + "input", + "outcomes", + "observed", + "In", + "noisy", + "settings", + "represents", + "uncertainty", + "observations", + "Example", + "epsilon", + "true", + "function", + "sim", + "mathcal", + "sigma", + "Gaussian", + "distributed", + "noise", + "Then" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a77e64ce-1612-51c9-9a4c-a813e4916637", + "type": "text", + "content": "The loss function\n$$\n\\ell : \\mathcal {Y} \\times \\mathcal {Y} \\rightarrow [ 0, + \\infty)\n$$\nrepresents the cost $\\ell(f(x), y)$ incurred when predicting $f(x)$ instead of y.\nIt is part of the problem formulation:\n$$\n\\mathcal {E} (f) = \\int \\ell (f (x), y) d \\rho (x, y)\n$$\nThe minimizer of the risk (if it exists) is “chosen” by the loss.", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Loss Functions", + "metadata": { + "length": 337, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "The", + "loss", + "function", + "ell", + "mathcal", + "times", + "rightarrow", + "infty", + "represents", + "cost", + "incurred", + "predicting", + "It", + "part", + "problem", + "formulation", + "int", + "rho", + "minimizer", + "risk", + "exists", + "chosen" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3348fe10-65e0-5889-bf3f-ac5697e82fd9", + "type": "image", + "content": "\nL (y, y ^ {\\prime}) = L (y - y ^ {\\prime})
0\n[images/image-14-Loss Functions Comparison.jpg]\n", + "path": "images/image-14-Loss Functions Comparison.jpg", + "metadata": { + "length": 88, + "summary": "image-14\n- Chart Title: Comparison of Square Loss, Absolute, and Epsilon-insensitive loss functions\n- X-axis: Error values ranging from -1.0 to 1.0\n- Y-axis: Loss values ranging from 0.0 to 1.0\n- Curves displayed:\n * Square Loss (blue curve): Parabolic shape increasing quadratically with error\n * Absolute (purple line): Linear V-shape representing absolute value function\n * ε-insensitive (yellow line): Flat region near zero with linear increase outside the epsilon margin\n- Legend identifies all three loss function types", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-14-Loss Functions Comparison.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "b77ecb8a-2fe4-50a2-9fc5-d95d3ea1d5f8", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
xSquare LossAbsoluteε - insensitive
1.01.01.00.7
0.50.20.40.2
0.00.00.00.0
0.50.20.40.2
1.01.01.00.7
", + "path": "tables/table-2 Loss Functions.html", + "metadata": { + "length": 707, + "summary": "table-3\nThe table compares three loss functions (Square, Absolute, ε-insensitive) across x values from -1.0 to 1.0, showing symmetric values with ε-insensitive loss being lower at extremes.", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "tables/table-2 Loss Functions.html", + "keywords": [ + "Square Loss", + "Absolute Loss", + "ε-insensitive" + ], + "tokens": [] + } + }, + { + "chunk_id": "1a393e24-9e76-5c8f-9041-823a8ae5f56c", + "type": "text", + "content": "$$\nL (y, y ^ {\\prime}) = L (y - y ^ {\\prime})\n$$\n▶ Square loss $L(y, y') = (y - y')^{2}$ ,\n▶ Absolute loss $L(y, y') = |y - y'|$ ,\n▶ ε-insensitive $L(y, y') = \\max(|y - y'| - \\epsilon, 0)$ ,\n\nL (y, y ^ {\\prime}) = L (y - y ^ {\\prime})
0\n[images/image-14-Loss Functions Comparison.jpg]\n\n
\nline chart\n\n[tables/table-2 Loss Functions.html]\n\n
", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Loss Functions/Loss Functions for Regression", + "metadata": { + "length": 391, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "prime", + "Square", + "loss", + "Absolute", + "insensitive", + "max", + "epsilon", + "details", + "summary", + "line", + "chart" + ], + "keywords": [], + "connect_to": [ + { + "target": "3348fe10-65e0-5889-bf3f-ac5697e82fd9", + "relation": "embeds", + "ref": "[images/image-14-Loss Functions Comparison.jpg]", + "position": { + "start": 246, + "end": 293 + } + }, + { + "target": "b77ecb8a-2fe4-50a2-9fc5-d95d3ea1d5f8", + "relation": "embeds", + "ref": "[tables/table-2 Loss Functions.html]", + "position": { + "start": 336, + "end": 372 + } + } + ] + } + }, + { + "chunk_id": "4c26451c-14b4-52f5-b064-93f0762fe772", + "type": "image", + "content": "\n$$
0\n[images/image-15-Exponential Functions Graph.jpg]\n", + "path": "images/image-15-Exponential Functions Graph.jpg", + "metadata": { + "length": 48, + "summary": "image-15\n- Chart Title: None visible\n- Axes:\n - X-axis: Range from approximately -0.5 to 2, with major tick marks at 0, 1, and 2.\n - Y-axis: Range from 0 to above 2.0, with labeled ticks at 0.5, 1.0, 1.5, and 2.0.\n- Key Data Points & Features:\n - All plotted curves intersect at the point (0, 1).\n - One curve is a horizontal line at y = 1.\n - Two curves show exponential decay (decreasing as x increases), one steeper than the other.\n - One curve shows exponential growth (increasing as x increases) and intersects the decaying curves again near x = 1.5.\n- Context: The graph illustrates the behavior of different exponential functions passing through the common point (0, 1).", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "images/image-15-Exponential Functions Graph.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "c1e363f6-8706-5645-88f0-ab3f71838ded", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
xLine 1Line 2Line 3Line 4
01.01.01.01.0
10.00.00.00.0
21.00.50.20.3
", + "path": "tables/table-3 Data Table.html", + "metadata": { + "length": 574, + "summary": "table-4\nThe table shows three rows of data with columns x, Line 1, Line 2, Line 3, and Line 4. Row 0 has all 1.0, row 1 has all 0.0, and row 2 has 1.0, 0.5, 0.2, 0.3.", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "file_path": "tables/table-3 Data Table.html", + "keywords": [ + "table", + "values", + "rows" + ], + "tokens": [] + } + }, + { + "chunk_id": "017456a5-d7bb-52d0-b8fb-55742caee3d1", + "type": "text", + "content": "$$\nL (y, y ^ {\\prime}) = L (- y y ^ {\\prime})\n$$\n▶ 0-1 loss $L(y, y') = \\mathbf{1}_{\\{-yy' > 0\\}}$\n▶ Square loss $L(y, y') = (1 - yy')^{2}$ ,\n▶ Hinge-loss $L(y, y') = \\max(1 - yy', 0)$ ,\n▶ logistic loss $L(y, y') = \\log(1 + \\exp(-yy'))$ ,\n\n$$
0\n[images/image-15-Exponential Functions Graph.jpg]\n\n
\nline chart\n\n[tables/table-3 Data Table.html]\n\n
\n— 01 loss\n— square loss\n— Hinge loss\n— Logistic loss", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Loss Functions/Loss Functions for Classification", + "metadata": { + "length": 443, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "prime", + "loss", + "mathbf", + "yy", + "Square", + "Hinge", + "max", + "logistic", + "log", + "exp", + "details", + "summary", + "line", + "chart", + "01", + "square", + "Logistic" + ], + "keywords": [], + "connect_to": [ + { + "target": "4c26451c-14b4-52f5-b064-93f0762fe772", + "relation": "embeds", + "ref": "[images/image-15-Exponential Functions Graph.jpg]", + "position": { + "start": 254, + "end": 303 + } + }, + { + "target": "c1e363f6-8706-5645-88f0-ab3f71838ded", + "relation": "embeds", + "ref": "[tables/table-3 Data Table.html]", + "position": { + "start": 346, + "end": 378 + } + } + ] + } + }, + { + "chunk_id": "e7b8bc0b-b341-5068-8975-eaf4e3107cae", + "type": "text", + "content": "The relation between X and Y encoded by the distribution $\\rho$ is unknown in reality. The only way we have to access a phenomenon is from finite observations.\nThe goal of a learning algorithm is therefore to find a good approximation $f_{n}: X \\to Y$ for the minimizer of expected risk\n$$\n\\inf _ {f: \\mathcal {X} \\to \\mathcal {Y}} \\mathcal {E} (f)\n$$\nfrom a finite set of examples $(x_{i}, y_{i})_{i=1}^{n}$ sampled independently from $\\rho$ .", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem_2", + "metadata": { + "length": 444, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "The", + "relation", + "encoded", + "distribution", + "rho", + "unknown", + "reality", + "access", + "phenomenon", + "finite", + "observations", + "goal", + "learning", + "algorithm", + "find", + "good", + "approximation", + "minimizer", + "expected", + "risk", + "inf", + "mathcal", + "set", + "examples", + "sampled", + "independently" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "126d2fc4-7b4b-5654-bf78-652d6b0fb30d", + "type": "text", + "content": "Let $S = \\bigcup_{n \\in N} (\\mathcal{X} \\times \\mathcal{Y})^n$ be the set of all finite datasets on $\\mathcal{X} \\times \\mathcal{Y}$ . Denote F the set of all measurable functions $f: \\mathcal{X} \\to \\mathcal{Y}$ . A learning algorithm is a map\n$$\n\\begin{array}{l} A: \\mathcal {S} \\to \\mathcal {F} \\\\ S \\mapsto A (S): \\mathcal {X} \\to \\mathcal {Y} \\\\ \\end{array}\n$$\nTo highlight our interest in studying the relation between the size of a training set $S = (x_{i}, y_{i})_{i=1}^{n}$ and the corresponding predictor produced by an algorithm A, we will often denote (with some abuse of notation)\n$$\nf _ {n} = A \\left(\\left(x _ {i}, y _ {i}\\right) _ {i = 1} ^ {n}\\right)\n$$\nNon-deterministic Learning Algorithms\nWe can also consider stochastic algorithms, where the estimator $f_{n}$ is not automatically determined by the training set.\nIn these cases, given a dataset $S \\in S$ , an algorithm $A(S)$ can be seen as a distribution on $\\mathcal{F}$ and its output is one sample from $A(S)$ .\nUnder this interpretation a deterministic algorithm corresponds to $A(S)$ being a Dirac's delta.", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Defining Learning Algorithms", + "metadata": { + "length": 1084, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Let", + "bigcup", + "mathcal", + "times", + "set", + "finite", + "datasets", + "Denote", + "measurable", + "functions", + "learning", + "algorithm", + "map", + "begin", + "array", + "mapsto", + "end", + "To", + "highlight", + "interest", + "studying", + "relation", + "size", + "training", + "predictor", + "produced", + "denote", + "abuse", + "notation", + "left", + "Non", + "deterministic", + "Learning", + "Algorithms", + "We", + "stochastic", + "algorithms", + "estimator", + "automatically", + "determined", + "In", + "cases", + "dataset", + "distribution", + "output", + "sample", + "Under", + "interpretation", + "corresponds", + "Dirac", + "delta" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2440ca50-2ea9-5213-9794-ecfa7ddab495", + "type": "text", + "content": "Given a training set, we would like a learning algorithm to find a “good” predictor $f_{n}$ .\nWhat does “good” mean? That it has small error (or excess risk) with respect to the best solution of the learning problem.\nExcess Risk\n$$\n\\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f)\n$$", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem_3", + "metadata": { + "length": 304, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Given", + "training", + "set", + "learning", + "algorithm", + "find", + "good", + "predictor", + "What", + "That", + "small", + "error", + "excess", + "risk", + "respect", + "solution", + "problem", + "Excess", + "Risk", + "mathcal", + "inf" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1009893c-ef52-5e1d-ac7e-a8c7294cdad0", + "type": "text", + "content": "", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7a3daf48-3a51-525f-b5a3-f7c7c93c591d", + "type": "text", + "content": "Ideally we would like the learning algorithm to be consistent\n$$\n\\lim _ {n \\to + \\infty} \\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) = 0\n$$\nNamely that (asymptotically) our algorithm “solves” the problem.\nHowever $f_{n}=A(S)$ is a random variable: the points in the training set $S=(x_{i},y_{i})_{i=1}^{n}$ are randomly sampled from $\\rho$ .\nSo what do we mean by $\\mathcal{E}(f_{n})\\to\\inf\\mathcal{E}(f)$ ?", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Consistency", + "metadata": { + "length": 433, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Ideally", + "learning", + "algorithm", + "consistent", + "lim", + "infty", + "mathcal", + "inf", + "Namely", + "asymptotically", + "solves", + "problem", + "However", + "random", + "variable", + "points", + "training", + "set", + "randomly", + "sampled", + "rho", + "So" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "61b37d98-1574-56b2-891e-52e01646ea23", + "type": "text", + "content": "Convergence in expectation:\n$$\n\\lim _ {n \\to + \\infty} \\mathbb {E} \\left[ \\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) \\right] = 0\n$$\nConvergence in probability:\n$$\n\\lim _ {n \\rightarrow + \\infty} \\mathbb {P} \\left(\\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) > \\epsilon\\right) = 0 \\quad \\forall \\epsilon > 0\n$$\nMany other notions of convergence of random variables exist!", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Convergence of Random Variables", + "metadata": { + "length": 422, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Convergence", + "expectation", + "lim", + "infty", + "mathbb", + "left", + "mathcal", + "inf", + "probability", + "rightarrow", + "epsilon", + "quad", + "forall", + "Many", + "notions", + "convergence", + "random", + "variables", + "exist" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "aec6578f-54bc-56d6-a3db-10b27a6817f3", + "type": "text", + "content": "Note that we are only interested in guaranteeing that the risk of our estimator will converge to the best possible value\n$$\n\\mathcal {E} (f _ {n}) \\to \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f)\n$$\nbut we are not directly interested in determining whether $f_{n} \\to f^{*}$ (in some norm) where $f^{*}: X \\to Y$ is a minimizer of the expected risk\n$$\n\\mathcal {E} (f ^ {*}) = \\inf _ {f: \\mathcal {X} \\to \\mathcal {Y}} \\mathcal {E} (f)\n$$\nActually, the risk could even not admit a minimizer $f^{*}$ (although typically it will).\nThis is a main difference with several settings such as compressive sensing and inverse problems.", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Consistency vs Convergence of the Estimator", + "metadata": { + "length": 626, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Note", + "interested", + "guaranteeing", + "risk", + "estimator", + "converge", + "mathcal", + "inf", + "directly", + "determining", + "norm", + "minimizer", + "expected", + "Actually", + "admit", + "typically", + "This", + "main", + "difference", + "settings", + "compressive", + "sensing", + "inverse", + "problems" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "96478707-8087-5f16-ae05-5323763af418", + "type": "text", + "content": "However, the existence of $f^{*}$ can be useful in several situations.\nLeast Squares. $\\ell(f(x),y)=(f(x)-y)^{2}$ . Then\n$$\n\\mathcal {E} (f) - \\mathcal {E} (f ^ {*}) = \\| f - f ^ {*} \\| _ {L ^ {2} (\\mathcal {X}, \\rho)}\n$$\nLipschitz Loss. $|\\ell(z,y)-\\ell(z',y)|\\leq L|z-z'|$\n$$\n\\mathcal {E} (f) - \\mathcal {E} (f ^ {*}) \\leq L \\| f - f ^ {*} \\| _ {L ^ {1} (\\mathcal {X}, \\rho)}\n$$\nConvergence $f_{n} \\rightarrow f^{*}$ (in $L^{1}$ or $L^{2}$ norm respectively) automatically guarantees consistency!", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Existence of a Minimizer for the Risk", + "metadata": { + "length": 498, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "However", + "existence", + "situations", + "Least", + "Squares", + "ell", + "Then", + "mathcal", + "rho", + "Lipschitz", + "Loss", + "leq", + "Convergence", + "rightarrow", + "norm", + "automatically", + "guarantees", + "consistency" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6e90785b-128f-5052-a6df-9feba3b1d65d", + "type": "text", + "content": "Is consistency enough? Well no. It does not provide a quantitative measure of how “good” a learning algorithm is.\nIn other words, question: how do we compare two learning algorithms?\nAnswer: via their Learning Rates, namely the “speed” at which the excess risk goes to zero as n increases.\nExample: Expectation\n$$\n\\mathbb {E} \\left[ \\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) \\right] = O (n ^ {- \\alpha}) \\quad \\text { for some } \\alpha > 0.\n$$\nWe can compare two algorithms by determining which one has a faster learning rate (i.e. larger exponent $\\alpha$ ).", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Measuring the “Quality” of a Learning Algorithm", + "metadata": { + "length": 589, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Is", + "consistency", + "Well", + "It", + "provide", + "quantitative", + "measure", + "good", + "learning", + "algorithm", + "In", + "words", + "question", + "compare", + "algorithms", + "Answer", + "Learning", + "Rates", + "speed", + "excess", + "risk", + "increases", + "Example", + "Expectation", + "mathbb", + "left", + "mathcal", + "inf", + "alpha", + "quad", + "text", + "We", + "determining", + "faster", + "rate", + "larger", + "exponent" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "be940107-f71a-5ebd-9c4b-3297062fc171", + "type": "text", + "content": "Sample Complexity: minimum number $n(\\epsilon, \\delta)$ of training points the algorithm needs to achieve an excess risk lower than $\\epsilon$ with at least probability $1 - \\delta$ :\n$$\n\\mathbb {P} \\left(\\mathcal {E} (f _ {n (\\epsilon , \\delta)}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) \\leq \\epsilon\\right) \\geq 1 - \\delta\n$$\nError Bounds: Upper bound $\\epsilon(\\delta,n)>0$ on the excess risk of $f_{n}$ which holds with probability larger than $1-\\delta$\n$$\n\\mathbb {P} \\left(\\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) \\leq \\epsilon (\\delta , n)\\right) \\geq 1 - \\delta\n$$\nTail Bounds: Lower bound $\\delta(\\epsilon,n)\\in(0,1)$ on the probability that $f_{n}$ will have excess risk larger than $\\epsilon$\n$$\n\\mathbb {P} \\left(\\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) \\leq \\epsilon\\right) \\geq 1 - \\delta (\\epsilon , n)\n$$", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Sample Complexity, Error Bounds and Tail Bounds", + "metadata": { + "length": 886, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Sample", + "Complexity", + "minimum", + "number", + "epsilon", + "delta", + "training", + "points", + "algorithm", + "achieve", + "excess", + "risk", + "lower", + "probability", + "mathbb", + "left", + "mathcal", + "inf", + "leq", + "geq", + "Error", + "Bounds", + "Upper", + "bound", + "holds", + "larger", + "Tail", + "Lower" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c2bd4750-b3ff-5917-ad13-1c86a0acb4a2", + "type": "text", + "content": "If $\\rho$ is unknown... how can we say anything about $\\mathcal{E}(f_n) - \\inf_{f\\in \\mathcal{F}}\\mathcal{E}(f)?$\nWe have “glimpses” of $\\rho$ only via the samples $(x_{i}, y_{i})_{i=1}^{n}$ . Can we use them to gather some information about $\\rho$ (or better, on $\\mathcal{E}(f)$ )?\nConsider function $f : X \\to Y$ and its empirical risk\n$$\n\\mathcal {E} _ {n} (f) = \\frac {1}{n} \\sum_ {i = 1} ^ {n} \\ell (f (x _ {i}), y _ {i})\n$$\nA simple calculation shows that\n$$\n\\mathbb {E} _ {S \\sim \\rho^ {n}} (\\mathcal {E} _ {n} (f)) = \\frac {1}{n} \\sum_ {i = 1} ^ {n} \\mathbb {E} _ {(x _ {i}, y _ {i}) \\sim \\rho} (\\ell (f (x _ {i}), y _ {i})) = \\frac {1}{n} \\sum_ {i = 1} ^ {n} \\mathcal {E} (f) = \\mathcal {E} (f)\n$$\nThe expectation of $\\mathcal{E}_{n}(f)$ is the expected risk $\\mathcal{E}(f)!$", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Empirical Risk as a Proxy", + "metadata": { + "length": 786, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "If", + "rho", + "unknown", + "mathcal", + "inf", + "We", + "glimpses", + "samples", + "Can", + "gather", + "information", + "Consider", + "function", + "empirical", + "risk", + "frac", + "sum", + "ell", + "simple", + "calculation", + "shows", + "mathbb", + "sim", + "The", + "expectation", + "expected" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1294e2d3-ebe2-5f95-a138-451c5beb6654", + "type": "text", + "content": "How close is $\\mathcal{E}_{n}(f)$ to $\\mathcal{E}(f)$ with respect to the number n of training points?\nConsider i.i.d. random variables X and $(X_{i})_{i=1}^{n}$ . Let $\\bar{X}_{n} = \\frac{1}{n} \\sum_{i=1}^{n} X_{i}$ . Then\n$$\n\\mathbb {E} [ (\\bar {X} _ {n} - \\mathbb {E} (X)) ^ {2} ] = \\operatorname{Var} (\\bar {X} _ {n}) = \\frac {\\operatorname{Var} (X)}{n}\n$$\nTherefore the expected (squared) distance between the empirical mean of the $X_{i}$ and their expectation $\\mathbb{E}(X)$ goes to zero as $O(1/n)$ (Assuming X to have finite variance).\nIf $X_{i} = \\ell (f(x_{i}),y_{i})$ , we have $\\bar{X}_n = \\mathcal{E}_n(f)$ and therefore\n$$\n\\mathbb {E} [ (\\mathcal {E} _ {n} (f) - \\mathcal {E} (f)) ^ {2} ] = \\frac {\\operatorname{Var} (\\ell (f (x) , y))}{n}\n$$", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Empirical Vs Expected", + "metadata": { + "length": 758, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "How", + "close", + "mathcal", + "respect", + "number", + "training", + "points", + "Consider", + "random", + "variables", + "Let", + "bar", + "frac", + "sum", + "Then", + "mathbb", + "operatorname", + "Var", + "Therefore", + "expected", + "squared", + "distance", + "empirical", + "expectation", + "Assuming", + "finite", + "variance", + "If", + "ell" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b5c7932a-1d9c-5a45-a20d-3d26e5af7a70", + "type": "text", + "content": "If $X_{i} = \\ell(f(x_{i}), y_{i})$ , we have $\\bar{X}_{n} = \\mathcal{E}_{n}(f)$ and therefore\n$$\n\\mathbb {E} [ (\\mathcal {E} _ {n} (f) - \\mathcal {E} (f)) ^ {2} ] = \\frac {\\mathrm{Var} (\\ell (f (x) , y))}{n}\n$$\nIn particular\n$$\n\\mathbb {E} [ | \\mathcal {E} _ {n} (f) - \\mathcal {E} (f) | ] \\leq \\sqrt {\\frac {\\mathrm{Var} (\\ell (f (x) , y))}{n}}\n$$", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Empirical Vs Expected Risk", + "metadata": { + "length": 348, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "If", + "ell", + "bar", + "mathcal", + "mathbb", + "frac", + "mathrm", + "Var", + "In", + "leq", + "sqrt" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "af70faba-faa4-5d71-842e-dcf45d775686", + "type": "text", + "content": "Assume for simplicity that there exists a minimizer $f_{*}: X \\to Y$ of the expected risk\n$$\n\\mathcal {E} (f _ {*}) = \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f)\n$$\nFor any function $f : X \\to Y$ we can decompose the excess risk as\n$$\n\\begin{array}{l} \\mathcal {E} (f) - \\mathcal {E} (f _ {*}) = \\\\ \\mathcal {E} (f) - \\mathcal {E} _ {n} (f) + \\mathcal {E} _ {n} (f) - \\mathcal {E} _ {n} (f _ {*}) + \\mathcal {E} _ {n} (f _ {*}) - \\mathcal {E} (f _ {*}), \\\\ \\end{array}\n$$\nrecalling the definition $\\mathcal{E}_{n}(f):=\\frac{1}{n}\\sum_{i=1}^{n}\\ell(f(x_{i}),y_{i})$ of the empirical risk. Note that this in particularly then also holds for $f_{n}$ , which we will use below. We can therefore leverage on the statistical relation between $E_{n}$ and E to study the expected risk in terms of the empirical risk.\nThis perspective leads to one of the most well-established strategies on SLT: Empirical Risk Minimization", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Empirical Vs Expected_2", + "metadata": { + "length": 915, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Assume", + "simplicity", + "exists", + "minimizer", + "expected", + "risk", + "mathcal", + "inf", + "For", + "function", + "decompose", + "excess", + "begin", + "array", + "end", + "recalling", + "definition", + "frac", + "sum", + "ell", + "empirical", + "Note", + "holds", + "We", + "leverage", + "statistical", + "relation", + "study", + "terms", + "This", + "perspective", + "leads", + "established", + "strategies", + "SLT", + "Empirical", + "Risk", + "Minimization" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f6c6e734-a3e3-582a-9842-1a9676e583ba", + "type": "text", + "content": "Let $f_{n}$ be the minimizer of the empirical risk\n$$\nf _ {n} = \\underset {f \\in \\mathcal {F}} {\\operatorname{argmin}} \\mathcal {E} _ {n} (f)\n$$\nThen we automatically have $\\mathcal{E}_{n}(f_{n}) - \\mathcal{E}_{n}(f_{*}) \\leq 0$ (for any choice of training set).\nThen\n$$\n\\mathbb {E} \\mathcal {E} (f _ {n}) - \\mathcal {E} (f _ {*}) \\leq \\mathbb {E} \\mathcal {E} (f _ {n}) - \\mathcal {E} _ {n} (f _ {n}) \\quad (\\text { why? })\n$$\nWe can focus on studying only the generalization error\n$$\n\\mathbb {E} \\mathcal {E} (f _ {n}) - \\mathcal {E} _ {n} (f _ {n})\n$$", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Empirical Risk Minimization", + "metadata": { + "length": 554, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Let", + "minimizer", + "empirical", + "risk", + "underset", + "mathcal", + "operatorname", + "argmin", + "Then", + "automatically", + "leq", + "choice", + "training", + "set", + "mathbb", + "quad", + "text", + "We", + "focus", + "studying", + "generalization", + "error" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "48483a48-d13f-5208-9429-d1a028799da9", + "type": "text", + "content": "How can we control the generalization error\n$$\n\\mathcal {E} _ {n} (f _ {n}) - \\mathcal {E} (f _ {n})\n$$\nwith respect to the number n of examples?\nThis question is far from trivial...\n(and it is one of the main subject of SLT)\nIndeed, $E_{n}$ and $f_{n}$ both depend on the sampled training data. Therefore, we cannot use the result\n$$\n\\mathbb {E} \\left[ \\left. \\left| \\mathcal {E} _ {n} (f _ {n}) - \\mathcal {E} (f _ {n}) \\right| \\right] \\leq O (1 / \\sqrt {n}) \\right.\n$$\nwhich indeed will not be true in general... (next class).", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Generalization Error", + "metadata": { + "length": 529, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "How", + "control", + "generalization", + "error", + "mathcal", + "respect", + "number", + "examples", + "This", + "question", + "trivial", + "main", + "subject", + "SLT", + "Indeed", + "depend", + "sampled", + "training", + "data", + "Therefore", + "result", + "mathbb", + "left", + "leq", + "sqrt", + "true", + "general", + "class" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "19d4fe47-fb59-58f3-b8e2-77e3f3d2a95a", + "type": "text", + "content": "", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d830c289-073c-5073-b904-81e5e21cb6c1", + "type": "text", + "content": "In practice we can have many different problems and scenarios:\nParametric Vs Non-parametric learning\n▶ Fixed design Vs random design\n▶ Transductive Vs inductive learning\n▶ Offline/batch Vs online/adversarial learning\nDifferent goals and assumptions but similar tools to study/solve them!", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems/A Taxonomy of Supervised Learning Problems", + "metadata": { + "length": 287, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "In", + "practice", + "problems", + "scenarios", + "Parametric", + "Vs", + "Non", + "parametric", + "learning", + "Fixed", + "design", + "random", + "Transductive", + "inductive", + "Offline", + "batch", + "online", + "adversarial", + "Different", + "goals", + "assumptions", + "similar", + "tools", + "study", + "solve" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a643e31b-cca0-52f2-af2e-45d84811a08d", + "type": "text", + "content": "How much do we know about the model?\n▶ Parametric: assume the predictor to be modeled by a finite number of unknown parameters. Goal: find the parametrization that best fits the observed data. In several scenario the goal is not in (only) having good predictions but rather use the recovered model for other purposes (e.g. identification).\n▶ Non-parametric. allow the parametrization of the model to increase in complexity as more examples are observed. Goal: find an estimator with optimal generalization performance (i.e. lowest expected risk E).", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems/A Taxonomy of Supervised Learning Problems/Parametric Vs Non-parametric", + "metadata": { + "length": 548, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "How", + "model", + "Parametric", + "assume", + "predictor", + "modeled", + "finite", + "number", + "unknown", + "parameters", + "Goal", + "find", + "parametrization", + "fits", + "observed", + "data", + "In", + "scenario", + "goal", + "good", + "predictions", + "recovered", + "purposes", + "identification", + "Non", + "parametric", + "increase", + "complexity", + "examples", + "estimator", + "optimal", + "generalization", + "performance", + "lowest", + "expected", + "risk" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6d5a8381-4e83-5d13-b078-0e8775a1c3c7", + "type": "text", + "content": "From experiment design...\n▶ Fixed Design. Given training examples $(x_{i}, y_{i})_{i=1}^{n}$ , the goal is to achieve good estimates for $\\rho(y|x_{i})$ on the prescribed training inputs. No distribution on the input data $\\rho_{X}$ is assumed/considered.\n$$\n\\frac {1}{n} \\sum_ {i = 1} ^ {n} \\int_ {\\mathcal {Y}} \\ell (f (x _ {i}), y) d \\rho (y | x _ {i})\n$$\n▶ Random Design. Agnostic about where the learned model will be tested. The goal is to make good predictions with respect to the distribution $\\rho(x,y)$ .", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems/A Taxonomy of Supervised Learning Problems/Fixed Design Vs Random Design", + "metadata": { + "length": 514, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "From", + "experiment", + "design", + "Fixed", + "Design", + "Given", + "training", + "examples", + "goal", + "achieve", + "good", + "estimates", + "rho", + "prescribed", + "inputs", + "No", + "distribution", + "input", + "data", + "assumed", + "considered", + "frac", + "sum", + "int", + "mathcal", + "ell", + "Random", + "Agnostic", + "learned", + "model", + "tested", + "The", + "make", + "predictions", + "respect" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7b4e6719-4965-51b0-8e0c-c6df7699b7aa", + "type": "text", + "content": "Do we have access to the test set in advance?\nTransductive: the goal is to achieve good prediction performance on a prescribed set of test points $(\\tilde{x}_{j})_{j=1}^{n_{test}}$ provided in advance. Transductive learning ignores the effect of $\\rho_{X}$ on the risk but focuses only on\n$$\n\\frac {1}{n _ {t e s t}} \\sum_ {j = 1} ^ {n _ {t e s t}} \\int_ {\\mathcal {Y}} \\ell (f (\\tilde {x} _ {j}), y) d \\rho (y | \\tilde {x} _ {j})\n$$\n▶ Inductive Agnostic about where the learned model will be tested. The goal is to make good predictions with respect to the distribution $\\rho(x, y)$ .", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems/A Taxonomy of Supervised Learning Problems/Inductive Vs Transductive Learning", + "metadata": { + "length": 585, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "Do", + "access", + "test", + "set", + "advance", + "Transductive", + "goal", + "achieve", + "good", + "prediction", + "performance", + "prescribed", + "points", + "tilde", + "provided", + "learning", + "ignores", + "effect", + "rho", + "risk", + "focuses", + "frac", + "sum", + "int", + "mathcal", + "ell", + "Inductive", + "Agnostic", + "learned", + "model", + "tested", + "The", + "make", + "predictions", + "respect", + "distribution" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6a93ef9f-3177-570b-b2fc-6f0018369505", + "type": "text", + "content": "How do we observe samples from $\\rho$ ?\n▶ Offline/Batch: a finite sample of input-output examples independently and identically distributed. Goal: minimize prediction errors on new examples\n▶ Online/Adversarial: We observe one input, propose a prediction and then observe the output. Goal: minimize the regret (i.e. choose the estimator that would have made less mistakes).\nNote. The distribution could be adversarial: $\\rho(y|x, f(x))$ instead of $\\rho(y|x)$ can make things “hard” for us.", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems/A Taxonomy of Supervised Learning Problems/Offline∕Batch Vs Online∕Adversarial Learning", + "metadata": { + "length": 490, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "How", + "observe", + "samples", + "rho", + "Offline", + "Batch", + "finite", + "sample", + "input", + "output", + "examples", + "independently", + "identically", + "distributed", + "Goal", + "minimize", + "prediction", + "errors", + "Online", + "Adversarial", + "We", + "propose", + "regret", + "choose", + "estimator", + "made", + "mistakes", + "Note", + "The", + "distribution", + "adversarial", + "make", + "things", + "hard" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a5944f38-db40-5e9b-bbe2-3a3af7fbc77f", + "type": "text", + "content": "This class:\n▶ Motivations and Examples\nFormulating the learning problem\n▶ Brief introduction to Learning Theory\n▶ A Taxonomy of supervised learning problems\nNext class: overfitting and the need for regularization...", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Wrapping up This class:", + "metadata": { + "length": 215, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Class 1 Introduction to Statistical Learning Theory", + "tokens": [ + "This", + "class", + "Motivations", + "Examples", + "Formulating", + "learning", + "problem", + "Brief", + "introduction", + "Learning", + "Theory", + "Taxonomy", + "supervised", + "problems", + "Next", + "overfitting", + "regularization" + ], + "keywords": [], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/doc_nav.json b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/doc_nav.json new file mode 100644 index 000000000..f28494754 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/doc_nav.json @@ -0,0 +1,447 @@ +{ + "version": "1.0", + "file_name": "Introduction to Statistical Learning Theory.pdf", + "stats": { + "total_chunks": 62, + "text_chunks": 43, + "image_chunks": 15, + "table_chunks": 4, + "max_depth": 5 + }, + "sections": [ + { + "title": "Class 1 Introduction to Statistical Learning Theory", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory", + "level": 1, + "summary": "Carlo Ciliberto Department of Computer Science, UCL October 5, 2018", + "chunk_count": 43, + "children": [ + { + "title": "Administrative Info", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Administrative Info", + "level": 2, + "summary": "▶ Class times: Fridays 14:00 - 15:30 $^{1}$ ▶ Location: Ground Floor Lecture Theater, Wilkins Building $^{2}$ ▶ Office hours: (Time TBA), 3rd Floor Hub room, CS Building, 66 Gower street. TA: Giulia Luise Website: ccilib er.github.io/intro-stl email(s): ccilib er@gmail.com, g.luise.16@ucl.ac.uk ▶ Workload: 2 assignments (50%) and a final exam (50%). Final exam requires to choose 3 problems out of 6. At least one problem from each “sides” of this course (RKHS or SLT) \\*must\\* be chosen.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Course Material", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Course Material", + "level": 2, + "summary": "Main resources for the course: Classes Slides Books and other Resources: S. Shalev-Shwartz and S. Ben-David Understanding Machine Learning: From Theory to Algorithms (Online Book). Cambridge University Press, 2014. ▶ O. Bousquet, S. Boucheron and G. Lugosi Introduction to Statistical Learning Theory (Tutorial). T. Poggio and L. Rosasco course slides and videos from MIT 9.520: Statistical Learning Theory and Applications. P. Liang course notes from Stanford CS229T: Statistical Learning Theory.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Prerequisites", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Prerequisites", + "level": 2, + "summary": "▶ Linear Algebra: familiarity with vector spaces, matrix operations (e.g. inversion, singular value decomposition (SVD)), inner products and norms, etc. ▶ Calculus: limits, derivatives, measures, integrals, etc. ▶ Probability Theory: probability distributions, conditional and marginal distribution, expectation, variance, etc.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Statistical Learning Theory (SLT)", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)", + "level": 2, + "summary": "SLT addresses questions related to: What does it mean for an algorithm to learn. What we can/cannot expect from a learning algorithm. ▶ How to design computationally & statistically efficient algorithms. What to do when a learning algorithm does not work... SLT studies theoretical quantities that we don't have access to: It tries to bridge the gap between the unknown functional relations governing a process and our (finite) empirical observations of it.", + "chunk_count": 6, + "children": [ + { + "title": "Motivations and Examples: Regression", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)/Motivations and Examples: Regression", + "level": 3, + "summary": "[tables/table-0 Housing Data.html] What to do when a learning algorithm does not work...
0 [images/image-1-Housing Prices Scatter Plot.jpg]
scatter plot [tables/table-1 House Price Data.html]
Image credits: coursera", + "chunk_count": 1, + "children": [] + }, + { + "title": "Motivations and Examples: Binary Classification", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)/Motivations and Examples: Binary Classification", + "level": 3, + "summary": "Spam detection: Automatically discriminate spam vs non-spam e-mails.
0 [images/image-2-__details_.jpg]
flowchart ```mermaid graph LR A[\"Stacked Document\"] --> B[\"{ legit, spam}\"] ```
Image Classification graph LR
0 [images/image-3-graph LR _.jpg]
text_image f : → {dog, not dog}
", + "chunk_count": 1, + "children": [] + }, + { + "title": "Motivations and Examples: Multi-class Classification", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)/Motivations and Examples: Multi-class Classification", + "level": 3, + "summary": "Identify the category of the object depicted in an image. Example: Caltech 101 → {dog, not dog} ### Motivations and Examples: Multi-class Classification0 [images/image-4-Image Type_ Photo.jpg]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Motivations and Examples: Multi-class Classification_2", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)/Motivations and Examples: Multi-class Classification_2", + "level": 3, + "summary": "Scaling things up: detect correct object among thousands of categories. ImageNet Large Scale Visual Recognition Challenge Identify the category of the object depicted in an image.
0 [images/image-5-Visual Data Grid.jpg]
text_image Grid of numerous emoji icons and text, likely representing a collection of various application or service types.
", + "chunk_count": 1, + "children": [] + }, + { + "title": "Motivations and Examples: Structured Prediction Image Captioning", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Statistical Learning Theory (SLT)/Motivations and Examples: Structured Prediction Image Captioning", + "level": 3, + "summary": "Image Captioning (also Localization Segmentation Classification) Grid of numerous emoji icons and text, likely representing a collection of various application or service types.
0 [images/image-6-Grid of nu.jpg]
natural_image Cat sitting at a desk with papers and a cup, viewed through a window with curtains (no visible text or symbols)
0 [images/image-7-_details_ .jpg]
natural_image A cat sitting at a desk with a book, viewed through a window (no visible text or symbols)
Gray striped cat Business suit Coffee mug Newspaper Movie Ranking NETFLIX user:127 Coffee mug Speech Recognition \"Ok Google\"0 [images/image-8-Coffee mug.jpg] Newspaper Speech Recognition \"Ok Google\"0 [images/image-9-Newspaper .jpg] Movie Ranking Speech Recognition \"Ok Google\"0 [images/image-10-Movie Rank.jpg] Speech Recognition \"Ok Google\" user:127
0 [images/image-11-user_127 _.jpg]
natural_image Abstract blue waveform visualization with no text or symbols
Protein Folding
## Formulating The Learning Problem0 [images/image-12-_details_ .jpg] natural_image ## Formulating The Learning Problem0 [images/image-13-_summary_n.jpg]", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Formulating The Learning Problem", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem", + "level": 2, + "summary": "", + "chunk_count": 13, + "children": [ + { + "title": "Formulating the Learning Problem", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem", + "level": 3, + "summary": "", + "chunk_count": 9, + "children": [ + { + "title": "Main ingredients:", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Main ingredients:", + "level": 4, + "summary": "▶ X input and Y output spaces. ▶ ρ unknown distribution on $X \\times Y$ . $\\ell : \\mathcal{Y} \\times \\mathcal{Y} \\to \\mathbb{R}$ a loss function measuring the discrepancy $\\ell(y, y')$ between any two points $y, y' \\in \\mathcal{Y}$ . We would like to minimize the expected risk $$ \\underset {f: \\mathcal {X} \\to \\mathcal {Y}} {\\text { minimize }} \\mathcal {E} (f) \\quad \\mathcal {E} (f) = \\int_ {\\mathcal {X} \\times \\mathcal {Y}} \\ell (f (x), y) d \\rho (x, y) $$ The expected prediction error incurred by a predictor $^{3}$ $f : X \\to Y$ .", + "chunk_count": 1, + "children": [] + }, + { + "title": "Input Space Linear Spaces", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Input Space Linear Spaces", + "level": 4, + "summary": "Linear Spaces Vectors Matrices Functions \"Structured\" Spaces ▶ Strings ▶ Graphs Probabilities ▶ Points on a manifold ▶ ...", + "chunk_count": 1, + "children": [] + }, + { + "title": "Output Space", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Output Space", + "level": 4, + "summary": "Linear Spaces, e.g. ▶ $\\mathcal{Y} = R$ regression ▶ $Y = \\{1, \\ldots, T\\}$ classification $\\mathcal{Y} = \\mathbb{R}^{T}$ multi-task \"Structured\" Spaces, e.g. ▶ Strings Graphs Probabilities Orders (i.e. Ranking) ▶ ...", + "chunk_count": 1, + "children": [] + }, + { + "title": "Probability Distribution", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Probability Distribution", + "level": 4, + "summary": "Informally: the distribution $\\rho$ on $X \\times Y$ encodes the probability of getting a pair $(x, y) \\in \\mathcal{X} \\times \\mathcal{Y}$ when observing (sampling from) the unknown process. Throughout the course we will assume $\\rho(x,y)=\\rho(y|x)\\rho_{\\mathcal{X}}(x)$ ▶ $\\rho_{\\mathcal{X}}(x)$ marginal distribution on X. ▶ $\\rho(y|x)$ conditional distribution on Y given $x \\in X$ .", + "chunk_count": 1, + "children": [] + }, + { + "title": "Conditional Distribution", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Conditional Distribution", + "level": 4, + "summary": "$\\rho(y|x)$ characterizes the relation between a given input x and the possible outcomes y that could be observed. In noisy settings it represents the uncertainty in our observations. Example: $y = f_{*}(x) + \\epsilon$ , with $f_{*} : X \\to R$ the “true” function and $\\epsilon \\sim \\mathcal{N}(0, \\sigma)$ Gaussian distributed noise. Then: $$ \\rho (y | x) = \\mathcal {N} (f _ {*} (x), \\sigma) $$", + "chunk_count": 1, + "children": [] + }, + { + "title": "Loss Functions", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Loss Functions", + "level": 4, + "summary": "The loss function $$ \\ell : \\mathcal {Y} \\times \\mathcal {Y} \\rightarrow [ 0, + \\infty) $$ represents the cost $\\ell(f(x), y)$ incurred when predicting $f(x)$ instead of y. It is part of the problem formulation: $$ \\mathcal {E} (f) = \\int \\ell (f (x), y) d \\rho (x, y) $$ The minimizer of the risk (if it exists) is “chosen” by the loss.", + "chunk_count": 3, + "children": [ + { + "title": "Loss Functions for Regression", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Loss Functions/Loss Functions for Regression", + "level": 5, + "summary": "$$ L (y, y ^ {\\prime}) = L (y - y ^ {\\prime}) $$ ▶ Square loss $L(y, y') = (y - y')^{2}$ , ▶ Absolute loss $L(y, y') = |y - y'|$ , ▶ ε-insensitive $L(y, y') = \\max(|y - y'| - \\epsilon, 0)$ , L (y, y ^ {\\prime}) = L (y - y ^ {\\prime})
0 [images/image-14-Loss Functions Comparison.jpg]
line chart [tables/table-2 Loss Functions.html]
", + "chunk_count": 1, + "children": [] + }, + { + "title": "Loss Functions for Classification", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem/Loss Functions/Loss Functions for Classification", + "level": 5, + "summary": "$$ L (y, y ^ {\\prime}) = L (- y y ^ {\\prime}) $$ ▶ 0-1 loss $L(y, y') = \\mathbf{1}_{\\{-yy' > 0\\}}$ ▶ Square loss $L(y, y') = (1 - yy')^{2}$ , ▶ Hinge-loss $L(y, y') = \\max(1 - yy', 0)$ , ▶ logistic loss $L(y, y') = \\log(1 + \\exp(-yy'))$ , $$
0 [images/image-15-Exponential Functions Graph.jpg]
line chart [tables/table-3 Data Table.html]
— 01 loss — square loss — Hinge loss — Logistic loss", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Formulating the Learning Problem_2", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem_2", + "level": 3, + "summary": "The relation between X and Y encoded by the distribution $\\rho$ is unknown in reality. The only way we have to access a phenomenon is from finite observations. The goal of a learning algorithm is therefore to find a good approximation $f_{n}: X \\to Y$ for the minimizer of expected risk $$ \\inf _ {f: \\mathcal {X} \\to \\mathcal {Y}} \\mathcal {E} (f) $$ from a finite set of examples $(x_{i}, y_{i})_{i=1}^{n}$ sampled independently from $\\rho$ .", + "chunk_count": 1, + "children": [] + }, + { + "title": "Defining Learning Algorithms", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Defining Learning Algorithms", + "level": 3, + "summary": "Let $S = \\bigcup_{n \\in N} (\\mathcal{X} \\times \\mathcal{Y})^n$ be the set of all finite datasets on $\\mathcal{X} \\times \\mathcal{Y}$ . Denote F the set of all measurable functions $f: \\mathcal{X} \\to \\mathcal{Y}$ . A learning algorithm is a map $$ \\begin{array}{l} A: \\mathcal {S} \\to \\mathcal {F} \\\\ S \\mapsto A (S): \\mathcal {X} \\to \\mathcal {Y} \\\\ \\end{array} $$ To highlight our interest in studying the relation between the size of a training set $S = (x_{i}, y_{i})_{i=1}^{n}$ and the corresponding predictor produced by an algorithm A, we will often denote (with some abuse of notation) $$ f _ {n} = A \\left(\\left(x _ {i}, y _ {i}\\right) _ {i = 1} ^ {n}\\right) $$ Non-deterministic Learning Algorithms We can also consider stochastic algorithms, where the estimator $f_{n}$ is not automatically determined by the training set. In these cases, given a dataset $S \\in S$ , an algorithm $A(S)$ can be seen as a distribution on $\\mathcal{F}$ and its output is one sample from $A(S)$ . Under this interpretation a deterministic algorithm corresponds to $A(S)$ being a Dirac's delta.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Formulating the Learning Problem_3", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Formulating The Learning Problem/Formulating the Learning Problem_3", + "level": 3, + "summary": "Given a training set, we would like a learning algorithm to find a “good” predictor $f_{n}$ . What does “good” mean? That it has small error (or excess risk) with respect to the best solution of the learning problem. Excess Risk $$ \\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) $$", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "The Elements of Learning Theory", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory", + "level": 2, + "summary": "", + "chunk_count": 13, + "children": [ + { + "title": "Consistency", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Consistency", + "level": 3, + "summary": "Ideally we would like the learning algorithm to be consistent $$ \\lim _ {n \\to + \\infty} \\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) = 0 $$ Namely that (asymptotically) our algorithm “solves” the problem. However $f_{n}=A(S)$ is a random variable: the points in the training set $S=(x_{i},y_{i})_{i=1}^{n}$ are randomly sampled from $\\rho$ . So what do we mean by $\\mathcal{E}(f_{n})\\to\\inf\\mathcal{E}(f)$ ?", + "chunk_count": 1, + "children": [] + }, + { + "title": "Convergence of Random Variables", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Convergence of Random Variables", + "level": 3, + "summary": "Convergence in expectation: $$ \\lim _ {n \\to + \\infty} \\mathbb {E} \\left[ \\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) \\right] = 0 $$ Convergence in probability: $$ \\lim _ {n \\rightarrow + \\infty} \\mathbb {P} \\left(\\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) > \\epsilon\\right) = 0 \\quad \\forall \\epsilon > 0 $$ Many other notions of convergence of random variables exist!", + "chunk_count": 1, + "children": [] + }, + { + "title": "Consistency vs Convergence of the Estimator", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Consistency vs Convergence of the Estimator", + "level": 3, + "summary": "Note that we are only interested in guaranteeing that the risk of our estimator will converge to the best possible value $$ \\mathcal {E} (f _ {n}) \\to \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) $$ but we are not directly interested in determining whether $f_{n} \\to f^{*}$ (in some norm) where $f^{*}: X \\to Y$ is a minimizer of the expected risk $$ \\mathcal {E} (f ^ {*}) = \\inf _ {f: \\mathcal {X} \\to \\mathcal {Y}} \\mathcal {E} (f) $$ Actually, the risk could even not admit a minimizer $f^{*}$ (although typically it will). This is a main difference with several settings such as compressive sensing and inverse problems.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Existence of a Minimizer for the Risk", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Existence of a Minimizer for the Risk", + "level": 3, + "summary": "However, the existence of $f^{*}$ can be useful in several situations. Least Squares. $\\ell(f(x),y)=(f(x)-y)^{2}$ . Then $$ \\mathcal {E} (f) - \\mathcal {E} (f ^ {*}) = \\| f - f ^ {*} \\| _ {L ^ {2} (\\mathcal {X}, \\rho)} $$ Lipschitz Loss. $|\\ell(z,y)-\\ell(z',y)|\\leq L|z-z'|$ $$ \\mathcal {E} (f) - \\mathcal {E} (f ^ {*}) \\leq L \\| f - f ^ {*} \\| _ {L ^ {1} (\\mathcal {X}, \\rho)} $$ Convergence $f_{n} \\rightarrow f^{*}$ (in $L^{1}$ or $L^{2}$ norm respectively) automatically guarantees consistency!", + "chunk_count": 1, + "children": [] + }, + { + "title": "Measuring the “Quality” of a Learning Algorithm", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Measuring the “Quality” of a Learning Algorithm", + "level": 3, + "summary": "Is consistency enough? Well no. It does not provide a quantitative measure of how “good” a learning algorithm is. In other words, question: how do we compare two learning algorithms? Answer: via their Learning Rates, namely the “speed” at which the excess risk goes to zero as n increases. Example: Expectation $$ \\mathbb {E} \\left[ \\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) \\right] = O (n ^ {- \\alpha}) \\quad \\text { for some } \\alpha > 0. $$ We can compare two algorithms by determining which one has a faster learning rate (i.e. larger exponent $\\alpha$ ).", + "chunk_count": 1, + "children": [] + }, + { + "title": "Sample Complexity, Error Bounds and Tail Bounds", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Sample Complexity, Error Bounds and Tail Bounds", + "level": 3, + "summary": "Sample Complexity: minimum number $n(\\epsilon, \\delta)$ of training points the algorithm needs to achieve an excess risk lower than $\\epsilon$ with at least probability $1 - \\delta$ : $$ \\mathbb {P} \\left(\\mathcal {E} (f _ {n (\\epsilon , \\delta)}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) \\leq \\epsilon\\right) \\geq 1 - \\delta $$ Error Bounds: Upper bound $\\epsilon(\\delta,n)>0$ on the excess risk of $f_{n}$ which holds with probability larger than $1-\\delta$ $$ \\mathbb {P} \\left(\\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) \\leq \\epsilon (\\delta , n)\\right) \\geq 1 - \\delta $$ Tail Bounds: Lower bound $\\delta(\\epsilon,n)\\in(0,1)$ on the probability that $f_{n}$ will have excess risk larger than $\\epsilon$ $$ \\mathbb {P} \\left(\\mathcal {E} (f _ {n}) - \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) \\leq \\epsilon\\right) \\geq 1 - \\delta (\\epsilon , n) $$", + "chunk_count": 1, + "children": [] + }, + { + "title": "Empirical Risk as a Proxy", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Empirical Risk as a Proxy", + "level": 3, + "summary": "If $\\rho$ is unknown... how can we say anything about $\\mathcal{E}(f_n) - \\inf_{f\\in \\mathcal{F}}\\mathcal{E}(f)?$ We have “glimpses” of $\\rho$ only via the samples $(x_{i}, y_{i})_{i=1}^{n}$ . Can we use them to gather some information about $\\rho$ (or better, on $\\mathcal{E}(f)$ )? Consider function $f : X \\to Y$ and its empirical risk $$ \\mathcal {E} _ {n} (f) = \\frac {1}{n} \\sum_ {i = 1} ^ {n} \\ell (f (x _ {i}), y _ {i}) $$ A simple calculation shows that $$ \\mathbb {E} _ {S \\sim \\rho^ {n}} (\\mathcal {E} _ {n} (f)) = \\frac {1}{n} \\sum_ {i = 1} ^ {n} \\mathbb {E} _ {(x _ {i}, y _ {i}) \\sim \\rho} (\\ell (f (x _ {i}), y _ {i})) = \\frac {1}{n} \\sum_ {i = 1} ^ {n} \\mathcal {E} (f) = \\mathcal {E} (f) $$ The expectation of $\\mathcal{E}_{n}(f)$ is the expected risk $\\mathcal{E}(f)!$", + "chunk_count": 1, + "children": [] + }, + { + "title": "Empirical Vs Expected", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Empirical Vs Expected", + "level": 3, + "summary": "How close is $\\mathcal{E}_{n}(f)$ to $\\mathcal{E}(f)$ with respect to the number n of training points? Consider i.i.d. random variables X and $(X_{i})_{i=1}^{n}$ . Let $\\bar{X}_{n} = \\frac{1}{n} \\sum_{i=1}^{n} X_{i}$ . Then $$ \\mathbb {E} [ (\\bar {X} _ {n} - \\mathbb {E} (X)) ^ {2} ] = \\operatorname{Var} (\\bar {X} _ {n}) = \\frac {\\operatorname{Var} (X)}{n} $$ Therefore the expected (squared) distance between the empirical mean of the $X_{i}$ and their expectation $\\mathbb{E}(X)$ goes to zero as $O(1/n)$ (Assuming X to have finite variance). If $X_{i} = \\ell (f(x_{i}),y_{i})$ , we have $\\bar{X}_n = \\mathcal{E}_n(f)$ and therefore $$ \\mathbb {E} [ (\\mathcal {E} _ {n} (f) - \\mathcal {E} (f)) ^ {2} ] = \\frac {\\operatorname{Var} (\\ell (f (x) , y))}{n} $$", + "chunk_count": 1, + "children": [] + }, + { + "title": "Empirical Vs Expected Risk", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Empirical Vs Expected Risk", + "level": 3, + "summary": "If $X_{i} = \\ell(f(x_{i}), y_{i})$ , we have $\\bar{X}_{n} = \\mathcal{E}_{n}(f)$ and therefore $$ \\mathbb {E} [ (\\mathcal {E} _ {n} (f) - \\mathcal {E} (f)) ^ {2} ] = \\frac {\\mathrm{Var} (\\ell (f (x) , y))}{n} $$ In particular $$ \\mathbb {E} [ | \\mathcal {E} _ {n} (f) - \\mathcal {E} (f) | ] \\leq \\sqrt {\\frac {\\mathrm{Var} (\\ell (f (x) , y))}{n}} $$", + "chunk_count": 1, + "children": [] + }, + { + "title": "Empirical Vs Expected_2", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Empirical Vs Expected_2", + "level": 3, + "summary": "Assume for simplicity that there exists a minimizer $f_{*}: X \\to Y$ of the expected risk $$ \\mathcal {E} (f _ {*}) = \\inf _ {f \\in \\mathcal {F}} \\mathcal {E} (f) $$ For any function $f : X \\to Y$ we can decompose the excess risk as $$ \\begin{array}{l} \\mathcal {E} (f) - \\mathcal {E} (f _ {*}) = \\\\ \\mathcal {E} (f) - \\mathcal {E} _ {n} (f) + \\mathcal {E} _ {n} (f) - \\mathcal {E} _ {n} (f _ {*}) + \\mathcal {E} _ {n} (f _ {*}) - \\mathcal {E} (f _ {*}), \\\\ \\end{array} $$ recalling the definition $\\mathcal{E}_{n}(f):=\\frac{1}{n}\\sum_{i=1}^{n}\\ell(f(x_{i}),y_{i})$ of the empirical risk. Note that this in particularly then also holds for $f_{n}$ , which we will use below. We can therefore leverage on the statistical relation between $E_{n}$ and E to study the expected risk in terms of the empirical risk. This perspective leads to one of the most well-established strategies on SLT: Empirical Risk Minimization", + "chunk_count": 1, + "children": [] + }, + { + "title": "Empirical Risk Minimization", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Empirical Risk Minimization", + "level": 3, + "summary": "Let $f_{n}$ be the minimizer of the empirical risk $$ f _ {n} = \\underset {f \\in \\mathcal {F}} {\\operatorname{argmin}} \\mathcal {E} _ {n} (f) $$ Then we automatically have $\\mathcal{E}_{n}(f_{n}) - \\mathcal{E}_{n}(f_{*}) \\leq 0$ (for any choice of training set). Then $$ \\mathbb {E} \\mathcal {E} (f _ {n}) - \\mathcal {E} (f _ {*}) \\leq \\mathbb {E} \\mathcal {E} (f _ {n}) - \\mathcal {E} _ {n} (f _ {n}) \\quad (\\text { why? }) $$ We can focus on studying only the generalization error $$ \\mathbb {E} \\mathcal {E} (f _ {n}) - \\mathcal {E} _ {n} (f _ {n}) $$", + "chunk_count": 1, + "children": [] + }, + { + "title": "Generalization Error", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/The Elements of Learning Theory/Generalization Error", + "level": 3, + "summary": "How can we control the generalization error $$ \\mathcal {E} _ {n} (f _ {n}) - \\mathcal {E} (f _ {n}) $$ with respect to the number n of examples? This question is far from trivial... (and it is one of the main subject of SLT) Indeed, $E_{n}$ and $f_{n}$ both depend on the sampled training data. Therefore, we cannot use the result $$ \\mathbb {E} \\left[ \\left. \\left| \\mathcal {E} _ {n} (f _ {n}) - \\mathcal {E} (f _ {n}) \\right| \\right] \\leq O (1 / \\sqrt {n}) \\right. $$ which indeed will not be true in general... (next class).", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "A Taxonomy of Supervised Learning Problems", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems", + "level": 2, + "summary": "", + "chunk_count": 6, + "children": [ + { + "title": "A Taxonomy of Supervised Learning Problems", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems/A Taxonomy of Supervised Learning Problems", + "level": 3, + "summary": "In practice we can have many different problems and scenarios: Parametric Vs Non-parametric learning ▶ Fixed design Vs random design ▶ Transductive Vs inductive learning ▶ Offline/batch Vs online/adversarial learning Different goals and assumptions but similar tools to study/solve them!", + "chunk_count": 5, + "children": [ + { + "title": "Parametric Vs Non-parametric", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems/A Taxonomy of Supervised Learning Problems/Parametric Vs Non-parametric", + "level": 4, + "summary": "How much do we know about the model? ▶ Parametric: assume the predictor to be modeled by a finite number of unknown parameters. Goal: find the parametrization that best fits the observed data. In several scenario the goal is not in (only) having good predictions but rather use the recovered model for other purposes (e.g. identification). ▶ Non-parametric. allow the parametrization of the model to increase in complexity as more examples are observed. Goal: find an estimator with optimal generalization performance (i.e. lowest expected risk E).", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fixed Design Vs Random Design", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems/A Taxonomy of Supervised Learning Problems/Fixed Design Vs Random Design", + "level": 4, + "summary": "From experiment design... ▶ Fixed Design. Given training examples $(x_{i}, y_{i})_{i=1}^{n}$ , the goal is to achieve good estimates for $\\rho(y|x_{i})$ on the prescribed training inputs. No distribution on the input data $\\rho_{X}$ is assumed/considered. $$ \\frac {1}{n} \\sum_ {i = 1} ^ {n} \\int_ {\\mathcal {Y}} \\ell (f (x _ {i}), y) d \\rho (y | x _ {i}) $$ ▶ Random Design. Agnostic about where the learned model will be tested. The goal is to make good predictions with respect to the distribution $\\rho(x,y)$ .", + "chunk_count": 1, + "children": [] + }, + { + "title": "Inductive Vs Transductive Learning", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems/A Taxonomy of Supervised Learning Problems/Inductive Vs Transductive Learning", + "level": 4, + "summary": "Do we have access to the test set in advance? Transductive: the goal is to achieve good prediction performance on a prescribed set of test points $(\\tilde{x}_{j})_{j=1}^{n_{test}}$ provided in advance. Transductive learning ignores the effect of $\\rho_{X}$ on the risk but focuses only on $$ \\frac {1}{n _ {t e s t}} \\sum_ {j = 1} ^ {n _ {t e s t}} \\int_ {\\mathcal {Y}} \\ell (f (\\tilde {x} _ {j}), y) d \\rho (y | \\tilde {x} _ {j}) $$ ▶ Inductive Agnostic about where the learned model will be tested. The goal is to make good predictions with respect to the distribution $\\rho(x, y)$ .", + "chunk_count": 1, + "children": [] + }, + { + "title": "Offline∕Batch Vs Online∕Adversarial Learning", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/A Taxonomy of Supervised Learning Problems/A Taxonomy of Supervised Learning Problems/Offline∕Batch Vs Online∕Adversarial Learning", + "level": 4, + "summary": "How do we observe samples from $\\rho$ ? ▶ Offline/Batch: a finite sample of input-output examples independently and identically distributed. Goal: minimize prediction errors on new examples ▶ Online/Adversarial: We observe one input, propose a prediction and then observe the output. Goal: minimize the regret (i.e. choose the estimator that would have made less mistakes). Note. The distribution could be adversarial: $\\rho(y|x, f(x))$ instead of $\\rho(y|x)$ can make things “hard” for us.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Wrapping up This class:", + "path": "lec1.pdf/Class 1 Introduction to Statistical Learning Theory/Wrapping up This class:", + "level": 2, + "summary": "This class: ▶ Motivations and Examples Formulating the learning problem ▶ Brief introduction to Learning Theory ▶ A Taxonomy of supervised learning problems Next class: overfitting and the need for regularization...", + "chunk_count": 1, + "children": [] + } + ] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-Housing Prices Scatter Plot.jpg", + "summary": "image-1 Chart Title: housing prices X-Axis Label: square feet (Range: 500 to 5000) Y-Axis Label: price (in $1000) (Range: 0 to 1000) Data Points: Blue 'x' markers representing individual data points. Notable Patterns: The chart displays a positive correlation between house size and price; as square footage increases, the price tends to increase. Most data points are clustered between 1000-3000 square feet and $200k-$600k." + }, + { + "path": "images/image-2-__details_.jpg", + "summary": "image-2
0" + }, + { + "path": "images/image-3-graph LR _.jpg", + "summary": "image-3 graph LR
0" + }, + { + "path": "images/image-4-Image Type_ Photo.jpg", + "summary": "image-4 Title: Image Grid of Objects and Scenes - Row 1 (baseball): Images include a player in uniform, wooden bats, a baseball field diagram, a batter swinging, and novelty baseball bats. - Row 2 (dog): Photos show dogs with life rings, on beaches, lying on blankets, groups of dogs, close-ups, and one dog in gear. - Row 3 (basketball-hoop): Various basketball hoops including backboards, nets, rims, and scenes with balls; includes text \"America Conference\". - Row 4 (Kayak): Kayaks in water, people kayaking near waterfalls, colorful kayaks lined up, and kayaks loaded on vehicles. - Row 5 (traffic light): Traffic lights at intersections, digital displays, pedestrian signals, and traffic control panels." + }, + { + "path": "images/image-5-Visual Data Grid.jpg", + "summary": "image-5 A large grid of small thumbnail images displaying a diverse collection of subjects. The top half primarily features nature scenes, including flowers, plants, and landscapes. The bottom half contains a wide variety of objects such as tools, musical instruments, vehicles, furniture, clothing, and household items." + }, + { + "path": "images/image-6-Grid of nu.jpg", + "summary": "image-6 Grid of numerous emoji icons and text, likely representing a collection of various application or service types.
0" + }, + { + "path": "images/image-7-_details_ .jpg", + "summary": "image-7
0" + }, + { + "path": "images/image-8-Coffee mug.jpg", + "summary": "image-8 Coffee mug Speech Recognition \"Ok Google\"0" + }, + { + "path": "images/image-9-Newspaper .jpg", + "summary": "image-9 Newspaper Speech Recognition \"Ok Google\"0" + }, + { + "path": "images/image-10-Movie Rank.jpg", + "summary": "image-10 Movie Ranking Speech Recognition \"Ok Google\"0" + }, + { + "path": "images/image-11-user_127 _.jpg", + "summary": "image-11 user:127
0" + }, + { + "path": "images/image-12-_details_ .jpg", + "summary": "image-12
## Formulating The Learning Problem0" + }, + { + "path": "images/image-13-_summary_n.jpg", + "summary": "image-13 natural_image ## Formulating The Learning Problem0" + }, + { + "path": "images/image-14-Loss Functions Comparison.jpg", + "summary": "image-14 - Chart Title: Comparison of Square Loss, Absolute, and Epsilon-insensitive loss functions - X-axis: Error values ranging from -1.0 to 1.0 - Y-axis: Loss values ranging from 0.0 to 1.0 - Curves displayed: * Square Loss (blue curve): Parabolic shape increasing quadratically with error * Absolute (purple line): Linear V-shape representing absolute value function * ε-insensitive (yellow line): Flat region near zero with linear increase outside the epsilon margin - Legend identifies all three loss function types" + }, + { + "path": "images/image-15-Exponential Functions Graph.jpg", + "summary": "image-15 - Chart Title: None visible - Axes: - X-axis: Range from approximately -0.5 to 2, with major tick marks at 0, 1, and 2. - Y-axis: Range from 0 to above 2.0, with labeled ticks at 0.5, 1.0, 1.5, and 2.0. - Key Data Points & Features: - All plotted curves intersect at the point (0, 1). - One curve is a horizontal line at y = 1. - Two curves show exponential decay (decreasing as x increases), one steeper than the other. - One curve shows exponential growth (increasing as x increases) and intersects the decaying curves again near x = 1.5. - Context: The graph illustrates the behavior of different exponential functions passing through the common point (0, 1)." + } + ], + "tables": [ + { + "path": "tables/table-0 Housing Data.html", + "summary": "table-1 The table shows living area in square feet and corresponding house prices in thousands of dollars for several properties." + }, + { + "path": "tables/table-1 House Price Data.html", + "summary": "table-2 The table shows house prices (in $1000) vs. square footage. Price rises from $200k at 1000 sq ft to $700k at 3000 sq ft, then declines symmetrically back to $200k at 5000 sq ft." + }, + { + "path": "tables/table-2 Loss Functions.html", + "summary": "table-3 The table compares three loss functions (Square, Absolute, ε-insensitive) across x values from -1.0 to 1.0, showing symmetric values with ε-insensitive loss being lower at extremes." + }, + { + "path": "tables/table-3 Data Table.html", + "summary": "table-4 The table shows three rows of data with columns x, Line 1, Line 2, Line 3, and Line 4. Row 0 has all 1.0, row 1 has all 0.0, and row 2 has 1.0, 0.5, 0.2, 0.3." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/full.md b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/full.md new file mode 100644 index 000000000..f72565ab6 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/full.md @@ -0,0 +1,658 @@ +# Class 1 Introduction to Statistical Learning Theory + +Carlo Ciliberto +Department of Computer Science, UCL + +October 5, 2018 + +## Administrative Info + +▶ Class times: Fridays 14:00 - 15:30 $^{1}$ +▶ Location: Ground Floor Lecture Theater, Wilkins Building $^{2}$ +▶ Office hours: (Time TBA), 3rd Floor Hub room, CS Building, 66 Gower street. +TA: Giulia Luise +Website: ccilib er.github.io/intro-stl +email(s): ccilib er@gmail.com, g.luise.16@ucl.ac.uk +▶ Workload: 2 assignments (50%) and a final exam (50%). Final exam requires to choose 3 problems out of 6. At least one problem from each “sides” of this course (RKHS or SLT) \*must\* be chosen. + +## Course Material + +Main resources for the course: + +Classes +Slides + +Books and other Resources: + +S. Shalev-Shwartz and S. Ben-David Understanding Machine Learning: From Theory to Algorithms (Online Book). Cambridge University Press, 2014. +▶ O. Bousquet, S. Boucheron and G. Lugosi Introduction to Statistical Learning Theory (Tutorial). +T. Poggio and L. Rosasco course slides and videos from MIT 9.520: Statistical Learning Theory and Applications. +P. Liang course notes from Stanford CS229T: Statistical Learning Theory. + +## Prerequisites + +▶ Linear Algebra: familiarity with vector spaces, matrix operations (e.g. inversion, singular value decomposition (SVD)), inner products and norms, etc. +▶ Calculus: limits, derivatives, measures, integrals, etc. +▶ Probability Theory: probability distributions, conditional and marginal distribution, expectation, variance, etc. + +## Statistical Learning Theory (SLT) + +SLT addresses questions related to: + +What does it mean for an algorithm to learn. +What we can/cannot expect from a learning algorithm. +▶ How to design computationally & statistically efficient algorithms. +What to do when a learning algorithm does not work... + +SLT studies theoretical quantities that we don't have access to: + +It tries to bridge the gap between the unknown functional relations governing a process and our (finite) empirical observations of it. + +## Motivations and Examples: Regression + +
Living area (feet $^{2}$ )Price (1000$s)
2104400
1600330
2400369
1416232
3000540
+ +![](images/130f421007cfaaea477fe6676367db25b8f408a5fb8193f4e8bd47f92e4e84f4.jpg) + +
+scatter plot + +| square feet | price (in $1000) | +| ----------- | ---------------- | +| 1000 | 200 | +| 1200 | 250 | +| 1400 | 300 | +| 1600 | 350 | +| 1800 | 400 | +| 2000 | 450 | +| 2200 | 500 | +| 2400 | 550 | +| 2600 | 600 | +| 2800 | 650 | +| 3000 | 700 | +| 3200 | 650 | +| 3400 | 600 | +| 3600 | 550 | +| 3800 | 500 | +| 4000 | 450 | +| 4200 | 400 | +| 4400 | 350 | +| 4600 | 300 | +| 4800 | 250 | +| 5000 | 200 | +
+ +Image credits: coursera + +## Motivations and Examples: Binary Classification + +Spam detection: Automatically discriminate spam vs non-spam e-mails. + +![](images/933c4cfa11e9b6b09b80ddad813e476c342506f43c76addb08ee2c3572cef92c.jpg) + +
+flowchart + +```mermaid +graph LR + A["Stacked Document"] --> B["{ legit, spam}"] +``` +
+ +Image Classification + +![](images/899609955dd03e4c21fb21432eab58790a1f67b1c381b757d9380d40b107d30b.jpg) + +
+text_image + +f : +→ {dog, not dog} +
+ +## Motivations and Examples: Multi-class Classification + +Identify the category of the object depicted in an image. + +Example: Caltech 101 + +![](images/b73272eefd03c715ce737ff948f4134759e71d88a36eced7bcb7d9a005b51f02.jpg) + +## Motivations and Examples: Multi-class Classification + +Scaling things up: detect correct object among thousands of categories. ImageNet Large Scale Visual Recognition Challenge + +![](images/c49218950f69d09bcbf89b28e7feb5ecb3ba222777df2ace4ac9b97f9487fb88.jpg) + +
+text_image + +Grid of numerous emoji icons and text, likely representing a collection of various application or service types. +
+ +## Motivations and Examples: Structured Prediction + +## Image Captioning + +(also Localization Segmentation Classification) + +![](images/25476d2009878f7a2e04bf7c55086a7c487132079fc2fd0594393228659457ae.jpg) + +
+natural_image + +Cat sitting at a desk with papers and a cup, viewed through a window with curtains (no visible text or symbols) +
+ +![](images/2c299d79f71e9b33d82df388ffa2c92c0454718c9def6fcfa7bf9166786c36b4.jpg) + +
+natural_image + +A cat sitting at a desk with a book, viewed through a window (no visible text or symbols) +
+ +Gray striped cat + +Business suit + +Coffee mug + +Newspaper + +Movie Ranking + +NETFLIX + +user:127 + +![](images/b459366b99a647b3228e24d712bdbed7960a5000d2636990f6f9ae69b333301d.jpg) + +![](images/46f73a7eab5bea2922df3ec9989cc3d858e8b9dd5280086ea72441d93e85f638.jpg) + +![](images/e8bb55e54ca29d04d82c2bf1b521cc63a2cbedd0a86db911cb99d08d4b54a1ce.jpg) + +Speech Recognition "Ok Google" + +![](images/e6f03e84ca7dcee4c061f343371cc6b235116ee114ce0172b82464b7144cd8f2.jpg) + +
+natural_image + +Abstract blue waveform visualization with no text or symbols +
+ +Protein Folding + +![](images/b01a930ba66ad1ecf8aafc1df14ffd85b78f8105dafa245723557e1d712dfba9.jpg) + +![](images/f222c147bee68cd28f4291bef0775a0f8335ca96ad13eaf890dfc15c074d0043.jpg) + +## Formulating The Learning Problem + +## Formulating the Learning Problem + +## Main ingredients: + +▶ X input and Y output spaces. +▶ ρ unknown distribution on $X \times Y$ . +$\ell : \mathcal{Y} \times \mathcal{Y} \to \mathbb{R}$ a loss function measuring the discrepancy $\ell(y, y')$ between any two points $y, y' \in \mathcal{Y}$ . + +We would like to minimize the expected risk + +$$ +\underset {f: \mathcal {X} \to \mathcal {Y}} {\text { minimize }} \mathcal {E} (f) \quad \mathcal {E} (f) = \int_ {\mathcal {X} \times \mathcal {Y}} \ell (f (x), y) d \rho (x, y) +$$ + +The expected prediction error incurred by a predictor $^{3}$ $f : X \to Y$ . + +## Input Space + +## Linear Spaces + +Vectors +Matrices +Functions + +## "Structured" Spaces + +▶ Strings +▶ Graphs +Probabilities +▶ Points on a manifold +▶ ... + +## Output Space + +## Linear Spaces, e.g. + +▶ $\mathcal{Y} = R$ regression +▶ $Y = \{1, \ldots, T\}$ classification +$\mathcal{Y} = \mathbb{R}^{T}$ multi-task + +## "Structured" Spaces, e.g. + +▶ Strings +Graphs +Probabilities +Orders (i.e. Ranking) +▶ ... + +## Probability Distribution + +Informally: the distribution $\rho$ on $X \times Y$ encodes the probability of getting a pair $(x, y) \in \mathcal{X} \times \mathcal{Y}$ when observing (sampling from) the unknown process. + +Throughout the course we will assume $\rho(x,y)=\rho(y|x)\rho_{\mathcal{X}}(x)$ + +▶ $\rho_{\mathcal{X}}(x)$ marginal distribution on X. +▶ $\rho(y|x)$ conditional distribution on Y given $x \in X$ . + +## Conditional Distribution + +$\rho(y|x)$ characterizes the relation between a given input x and the possible outcomes y that could be observed. + +In noisy settings it represents the uncertainty in our observations. + +Example: $y = f_{*}(x) + \epsilon$ , with $f_{*} : X \to R$ the “true” function and $\epsilon \sim \mathcal{N}(0, \sigma)$ Gaussian distributed noise. Then: + +$$ +\rho (y | x) = \mathcal {N} (f _ {*} (x), \sigma) +$$ + +# Loss Functions + +The loss function + +$$ +\ell : \mathcal {Y} \times \mathcal {Y} \rightarrow [ 0, + \infty) +$$ + +represents the cost $\ell(f(x), y)$ incurred when predicting $f(x)$ instead of y. + +It is part of the problem formulation: + +$$ +\mathcal {E} (f) = \int \ell (f (x), y) d \rho (x, y) +$$ + +The minimizer of the risk (if it exists) is “chosen” by the loss. + +# Loss Functions for Regression + +$$ +L (y, y ^ {\prime}) = L (y - y ^ {\prime}) +$$ + +▶ Square loss $L(y, y') = (y - y')^{2}$ , +▶ Absolute loss $L(y, y') = |y - y'|$ , +▶ ε-insensitive $L(y, y') = \max(|y - y'| - \epsilon, 0)$ , + +![](images/154a0bee1083f4c26439233b1afe778c7b25a0d7626c8cb51f3e20bdf8dff08f.jpg) + +
+line chart + +| x | Square Loss | Absolute | ε - insensitive | +| ---- | ----------- | -------- | --------------- | +| 1.0 | 1.0 | 1.0 | 0.7 | +| 0.5 | 0.2 | 0.4 | 0.2 | +| 0.0 | 0.0 | 0.0 | 0.0 | +| 0.5 | 0.2 | 0.4 | 0.2 | +| 1.0 | 1.0 | 1.0 | 0.7 | +
+ +# Loss Functions for Classification + +$$ +L (y, y ^ {\prime}) = L (- y y ^ {\prime}) +$$ + +▶ 0-1 loss $L(y, y') = \mathbf{1}_{\{-yy' > 0\}}$ +▶ Square loss $L(y, y') = (1 - yy')^{2}$ , +▶ Hinge-loss $L(y, y') = \max(1 - yy', 0)$ , +▶ logistic loss $L(y, y') = \log(1 + \exp(-yy'))$ , + +![](images/ee108933fe14119599a5efa57665e51009c4875b53d10c79e157a603e1e1575b.jpg) + +
+line chart + +| x | Line 1 | Line 2 | Line 3 | Line 4 | +| ---- | ------ | ------ | ------ | ------ | +| 0 | 1.0 | 1.0 | 1.0 | 1.0 | +| 1 | 0.0 | 0.0 | 0.0 | 0.0 | +| 2 | 1.0 | 0.5 | 0.2 | 0.3 | +
+ +— 01 loss +— square loss +— Hinge loss +— Logistic loss + +## Formulating the Learning Problem + +The relation between X and Y encoded by the distribution $\rho$ is unknown in reality. The only way we have to access a phenomenon is from finite observations. + +The goal of a learning algorithm is therefore to find a good approximation $f_{n}: X \to Y$ for the minimizer of expected risk + +$$ +\inf _ {f: \mathcal {X} \to \mathcal {Y}} \mathcal {E} (f) +$$ + +from a finite set of examples $(x_{i}, y_{i})_{i=1}^{n}$ sampled independently from $\rho$ . + +## Defining Learning Algorithms + +Let $S = \bigcup_{n \in N} (\mathcal{X} \times \mathcal{Y})^n$ be the set of all finite datasets on $\mathcal{X} \times \mathcal{Y}$ . Denote F the set of all measurable functions $f: \mathcal{X} \to \mathcal{Y}$ . A learning algorithm is a map + +$$ +\begin{array}{l} A: \mathcal {S} \to \mathcal {F} \\ S \mapsto A (S): \mathcal {X} \to \mathcal {Y} \\ \end{array} +$$ + +To highlight our interest in studying the relation between the size of a training set $S = (x_{i}, y_{i})_{i=1}^{n}$ and the corresponding predictor produced by an algorithm A, we will often denote (with some abuse of notation) + +$$ +f _ {n} = A \left(\left(x _ {i}, y _ {i}\right) _ {i = 1} ^ {n}\right) +$$ + +## Non-deterministic Learning Algorithms + +We can also consider stochastic algorithms, where the estimator $f_{n}$ is not automatically determined by the training set. + +In these cases, given a dataset $S \in S$ , an algorithm $A(S)$ can be seen as a distribution on $\mathcal{F}$ and its output is one sample from $A(S)$ . + +Under this interpretation a deterministic algorithm corresponds to $A(S)$ being a Dirac's delta. + +# Formulating the Learning Problem + +Given a training set, we would like a learning algorithm to find a “good” predictor $f_{n}$ . + +What does “good” mean? That it has small error (or excess risk) with respect to the best solution of the learning problem. + +Excess Risk + +$$ +\mathcal {E} (f _ {n}) - \inf _ {f \in \mathcal {F}} \mathcal {E} (f) +$$ + +## The Elements of Learning Theory + +## Consistency + +Ideally we would like the learning algorithm to be consistent + +$$ +\lim _ {n \to + \infty} \mathcal {E} (f _ {n}) - \inf _ {f \in \mathcal {F}} \mathcal {E} (f) = 0 +$$ + +Namely that (asymptotically) our algorithm “solves” the problem. + +However $f_{n}=A(S)$ is a random variable: the points in the training set $S=(x_{i},y_{i})_{i=1}^{n}$ are randomly sampled from $\rho$ . + +So what do we mean by $\mathcal{E}(f_{n})\to\inf\mathcal{E}(f)$ ? + +## Convergence of Random Variables + +Convergence in expectation: + +$$ +\lim _ {n \to + \infty} \mathbb {E} \left[ \mathcal {E} (f _ {n}) - \inf _ {f \in \mathcal {F}} \mathcal {E} (f) \right] = 0 +$$ + +Convergence in probability: + +$$ +\lim _ {n \rightarrow + \infty} \mathbb {P} \left(\mathcal {E} (f _ {n}) - \inf _ {f \in \mathcal {F}} \mathcal {E} (f) > \epsilon\right) = 0 \quad \forall \epsilon > 0 +$$ + +Many other notions of convergence of random variables exist! + +## Consistency vs Convergence of the Estimator + +Note that we are only interested in guaranteeing that the risk of our estimator will converge to the best possible value + +$$ +\mathcal {E} (f _ {n}) \to \inf _ {f \in \mathcal {F}} \mathcal {E} (f) +$$ + +but we are not directly interested in determining whether $f_{n} \to f^{*}$ (in some norm) where $f^{*}: X \to Y$ is a minimizer of the expected risk + +$$ +\mathcal {E} (f ^ {*}) = \inf _ {f: \mathcal {X} \to \mathcal {Y}} \mathcal {E} (f) +$$ + +Actually, the risk could even not admit a minimizer $f^{*}$ (although typically it will). + +This is a main difference with several settings such as compressive sensing and inverse problems. + +## Existence of a Minimizer for the Risk + +However, the existence of $f^{*}$ can be useful in several situations. + +Least Squares. $\ell(f(x),y)=(f(x)-y)^{2}$ . Then + +$$ +\mathcal {E} (f) - \mathcal {E} (f ^ {*}) = \| f - f ^ {*} \| _ {L ^ {2} (\mathcal {X}, \rho)} +$$ + +Lipschitz Loss. $|\ell(z,y)-\ell(z',y)|\leq L|z-z'|$ + +$$ +\mathcal {E} (f) - \mathcal {E} (f ^ {*}) \leq L \| f - f ^ {*} \| _ {L ^ {1} (\mathcal {X}, \rho)} +$$ + +Convergence $f_{n} \rightarrow f^{*}$ (in $L^{1}$ or $L^{2}$ norm respectively) automatically guarantees consistency! + +## Measuring the “Quality” of a Learning Algorithm + +Is consistency enough? Well no. It does not provide a quantitative measure of how “good” a learning algorithm is. + +In other words, question: how do we compare two learning algorithms? + +Answer: via their Learning Rates, namely the “speed” at which the excess risk goes to zero as n increases. + +Example: Expectation + +$$ +\mathbb {E} \left[ \mathcal {E} (f _ {n}) - \inf _ {f \in \mathcal {F}} \mathcal {E} (f) \right] = O (n ^ {- \alpha}) \quad \text { for some } \alpha > 0. +$$ + +We can compare two algorithms by determining which one has a faster learning rate (i.e. larger exponent $\alpha$ ). + +## Sample Complexity, Error Bounds and Tail Bounds + +Sample Complexity: minimum number $n(\epsilon, \delta)$ of training points the algorithm needs to achieve an excess risk lower than $\epsilon$ with at least probability $1 - \delta$ : + +$$ +\mathbb {P} \left(\mathcal {E} (f _ {n (\epsilon , \delta)}) - \inf _ {f \in \mathcal {F}} \mathcal {E} (f) \leq \epsilon\right) \geq 1 - \delta +$$ + +Error Bounds: Upper bound $\epsilon(\delta,n)>0$ on the excess risk of $f_{n}$ which holds with probability larger than $1-\delta$ + +$$ +\mathbb {P} \left(\mathcal {E} (f _ {n}) - \inf _ {f \in \mathcal {F}} \mathcal {E} (f) \leq \epsilon (\delta , n)\right) \geq 1 - \delta +$$ + +Tail Bounds: Lower bound $\delta(\epsilon,n)\in(0,1)$ on the probability that $f_{n}$ will have excess risk larger than $\epsilon$ + +$$ +\mathbb {P} \left(\mathcal {E} (f _ {n}) - \inf _ {f \in \mathcal {F}} \mathcal {E} (f) \leq \epsilon\right) \geq 1 - \delta (\epsilon , n) +$$ + +## Empirical Risk as a Proxy + +If $\rho$ is unknown... how can we say anything about $\mathcal{E}(f_n) - \inf_{f\in \mathcal{F}}\mathcal{E}(f)?$ + +We have “glimpses” of $\rho$ only via the samples $(x_{i}, y_{i})_{i=1}^{n}$ . Can we use them to gather some information about $\rho$ (or better, on $\mathcal{E}(f)$ )? + +Consider function $f : X \to Y$ and its empirical risk + +$$ +\mathcal {E} _ {n} (f) = \frac {1}{n} \sum_ {i = 1} ^ {n} \ell (f (x _ {i}), y _ {i}) +$$ + +A simple calculation shows that + +$$ +\mathbb {E} _ {S \sim \rho^ {n}} (\mathcal {E} _ {n} (f)) = \frac {1}{n} \sum_ {i = 1} ^ {n} \mathbb {E} _ {(x _ {i}, y _ {i}) \sim \rho} (\ell (f (x _ {i}), y _ {i})) = \frac {1}{n} \sum_ {i = 1} ^ {n} \mathcal {E} (f) = \mathcal {E} (f) +$$ + +The expectation of $\mathcal{E}_{n}(f)$ is the expected risk $\mathcal{E}(f)!$ + +# Empirical Vs Expected + +How close is $\mathcal{E}_{n}(f)$ to $\mathcal{E}(f)$ with respect to the number n of training points? + +Consider i.i.d. random variables X and $(X_{i})_{i=1}^{n}$ . Let $\bar{X}_{n} = \frac{1}{n} \sum_{i=1}^{n} X_{i}$ . Then + +$$ +\mathbb {E} [ (\bar {X} _ {n} - \mathbb {E} (X)) ^ {2} ] = \operatorname{Var} (\bar {X} _ {n}) = \frac {\operatorname{Var} (X)}{n} +$$ + +Therefore the expected (squared) distance between the empirical mean of the $X_{i}$ and their expectation $\mathbb{E}(X)$ goes to zero as $O(1/n)$ (Assuming X to have finite variance). + +If $X_{i} = \ell (f(x_{i}),y_{i})$ , we have $\bar{X}_n = \mathcal{E}_n(f)$ and therefore + +$$ +\mathbb {E} [ (\mathcal {E} _ {n} (f) - \mathcal {E} (f)) ^ {2} ] = \frac {\operatorname{Var} (\ell (f (x) , y))}{n} +$$ + +## Empirical Vs Expected Risk + +If $X_{i} = \ell(f(x_{i}), y_{i})$ , we have $\bar{X}_{n} = \mathcal{E}_{n}(f)$ and therefore + +$$ +\mathbb {E} [ (\mathcal {E} _ {n} (f) - \mathcal {E} (f)) ^ {2} ] = \frac {\mathrm{Var} (\ell (f (x) , y))}{n} +$$ + +In particular + +$$ +\mathbb {E} [ | \mathcal {E} _ {n} (f) - \mathcal {E} (f) | ] \leq \sqrt {\frac {\mathrm{Var} (\ell (f (x) , y))}{n}} +$$ + +## Empirical Vs Expected + +Assume for simplicity that there exists a minimizer $f_{*}: X \to Y$ of the expected risk + +$$ +\mathcal {E} (f _ {*}) = \inf _ {f \in \mathcal {F}} \mathcal {E} (f) +$$ + +For any function $f : X \to Y$ we can decompose the excess risk as + +$$ +\begin{array}{l} \mathcal {E} (f) - \mathcal {E} (f _ {*}) = \\ \mathcal {E} (f) - \mathcal {E} _ {n} (f) + \mathcal {E} _ {n} (f) - \mathcal {E} _ {n} (f _ {*}) + \mathcal {E} _ {n} (f _ {*}) - \mathcal {E} (f _ {*}), \\ \end{array} +$$ + +recalling the definition $\mathcal{E}_{n}(f):=\frac{1}{n}\sum_{i=1}^{n}\ell(f(x_{i}),y_{i})$ of the empirical risk. Note that this in particularly then also holds for $f_{n}$ , which we will use below. We can therefore leverage on the statistical relation between $E_{n}$ and E to study the expected risk in terms of the empirical risk. + +This perspective leads to one of the most well-established strategies on SLT: Empirical Risk Minimization + +## Empirical Risk Minimization + +Let $f_{n}$ be the minimizer of the empirical risk + +$$ +f _ {n} = \underset {f \in \mathcal {F}} {\operatorname{argmin}} \mathcal {E} _ {n} (f) +$$ + +Then we automatically have $\mathcal{E}_{n}(f_{n}) - \mathcal{E}_{n}(f_{*}) \leq 0$ (for any choice of training set). + +Then + +$$ +\mathbb {E} \mathcal {E} (f _ {n}) - \mathcal {E} (f _ {*}) \leq \mathbb {E} \mathcal {E} (f _ {n}) - \mathcal {E} _ {n} (f _ {n}) \quad (\text { why? }) +$$ + +We can focus on studying only the generalization error + +$$ +\mathbb {E} \mathcal {E} (f _ {n}) - \mathcal {E} _ {n} (f _ {n}) +$$ + +# Generalization Error + +How can we control the generalization error + +$$ +\mathcal {E} _ {n} (f _ {n}) - \mathcal {E} (f _ {n}) +$$ + +with respect to the number n of examples? + +This question is far from trivial... + +(and it is one of the main subject of SLT) + +Indeed, $E_{n}$ and $f_{n}$ both depend on the sampled training data. Therefore, we cannot use the result + +$$ +\mathbb {E} \left[ \left. \left| \mathcal {E} _ {n} (f _ {n}) - \mathcal {E} (f _ {n}) \right| \right] \leq O (1 / \sqrt {n}) \right. +$$ + +which indeed will not be true in general... (next class). + +## A Taxonomy of Supervised Learning Problems + +## A Taxonomy of Supervised Learning Problems + +In practice we can have many different problems and scenarios: + +Parametric Vs Non-parametric learning +▶ Fixed design Vs random design +▶ Transductive Vs inductive learning +▶ Offline/batch Vs online/adversarial learning + +Different goals and assumptions but similar tools to study/solve them! + +## Parametric Vs Non-parametric + +How much do we know about the model? + +▶ Parametric: assume the predictor to be modeled by a finite number of unknown parameters. Goal: find the parametrization that best fits the observed data. In several scenario the goal is not in (only) having good predictions but rather use the recovered model for other purposes (e.g. identification). +▶ Non-parametric. allow the parametrization of the model to increase in complexity as more examples are observed. Goal: find an estimator with optimal generalization performance (i.e. lowest expected risk E). + +## Fixed Design Vs Random Design + +From experiment design... + +▶ Fixed Design. Given training examples $(x_{i}, y_{i})_{i=1}^{n}$ , the goal is to achieve good estimates for $\rho(y|x_{i})$ on the prescribed training inputs. No distribution on the input data $\rho_{X}$ is assumed/considered. + +$$ +\frac {1}{n} \sum_ {i = 1} ^ {n} \int_ {\mathcal {Y}} \ell (f (x _ {i}), y) d \rho (y | x _ {i}) +$$ + +▶ Random Design. Agnostic about where the learned model will be tested. The goal is to make good predictions with respect to the distribution $\rho(x,y)$ . + +## Inductive Vs Transductive Learning + +Do we have access to the test set in advance? + +Transductive: the goal is to achieve good prediction performance on a prescribed set of test points $(\tilde{x}_{j})_{j=1}^{n_{test}}$ provided in advance. Transductive learning ignores the effect of $\rho_{X}$ on the risk but focuses only on + +$$ +\frac {1}{n _ {t e s t}} \sum_ {j = 1} ^ {n _ {t e s t}} \int_ {\mathcal {Y}} \ell (f (\tilde {x} _ {j}), y) d \rho (y | \tilde {x} _ {j}) +$$ + +▶ Inductive Agnostic about where the learned model will be tested. The goal is to make good predictions with respect to the distribution $\rho(x, y)$ . + +## Offline/Batch Vs Online/Adversarial Learning + +How do we observe samples from $\rho$ ? + +▶ Offline/Batch: a finite sample of input-output examples independently and identically distributed. Goal: minimize prediction errors on new examples +▶ Online/Adversarial: We observe one input, propose a prediction and then observe the output. Goal: minimize the regret (i.e. choose the estimator that would have made less mistakes). + +Note. The distribution could be adversarial: $\rho(y|x, f(x))$ instead of $\rho(y|x)$ can make things “hard” for us. + +## Wrapping up + +## This class: + +▶ Motivations and Examples +Formulating the learning problem +▶ Brief introduction to Learning Theory +▶ A Taxonomy of supervised learning problems + +Next class: overfitting and the need for regularization... \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-1-Housing Prices Scatter Plot.jpg b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-1-Housing Prices Scatter Plot.jpg new file mode 100644 index 0000000000000000000000000000000000000000..bfd00b70c566225b60a88a83b56e1474458e083c GIT binary patch literal 11300 zcmeHtbzGF&-tRLE&5)7;0s=})3kVD$r6MIYAfTYq-2y`hBGL#*cZVWFw=|+4F|;&@ zG!jxX@37DDZuPwH*?Zr+&;94#htC3@S!>qwi|_j7gC0fC0Ax25RTKdb2mnBs9{@c8 z$OHJexOliY_;`4D1O)hm5K3YQ5fOxr;v6X@<9VhFjOQ5`nA!L)GhgIkWnkdA%E`ko zAS^7*bXiwj`>k5iW z$|^Us?r7`i>gmHw%^o~_WNu;U1b8_?Y z3ku(qBPuGZs%vWBwzjo*bar+33=9qpBS%KZ#%Jf|7Z#V6Kd*e*-ub$_w}0^M@aT*$ z5CHxMTbSSfVC)yZ$T56jVPk`_@y_@HVYy)jm>e62MF5vVRs-*z!?}xse)yDfQ5j_| z1gt`uTU7TQKN3>23D2@`pRx8EXMc_{|Nkw{{ug7v^ED1Yz#xqBz~lfFI4C%8G8!T_ z>NdFu&&b`ZLj$Q~XkdCC4YUol790)qW7BV-f#oOAgKubHTiEWW1-t71fA)8A{U1nU zSk+D=?be$RtH(|btZEECddw7bzG7^CdfeND?Il^Y=e?K5vLof#4xvaJh%SA3vy^u} z%x~x?b$x8yM7hhO0eA!&IA=ID5^i&^T6EP)jD0EiM!V=nSOUCWC^|130S%iA`EpHn z&~{w^b{D>YpOCErv)44o1ry1A+`!W2-N}`!Q)r+MAHKPb2IkDC-Ozv}3mRzZL<7%? z#ZQbOFzv9Nl><%h;x1}tbwO*Lhuys4!Jg;Fi(Z9kYEYJ~97x|8K=5;0<35DYFmD_x zI*kl|#WT(RX`?Lf>65LeC^W!?x>V!#&CiNq;H~q!LAD$6g)i>25mo0AuT1ASODGml zm*pyW6RxWW?Zgp>GPq!eV5|Mer{Zzb;VkJX3VNcdrU1zPKR%^^kpc4;9|R)c?Bs$J zDAmF2abg!O9!9Z-;p$J$fWD{{8c&QvPjQOjTP)CyBKKn3t;$SN_on<34&G3dZ#iS&NH3va1tZf$QLj;@(KY8P8+h&KtBv~hjs46{HF(s zNL7SVBVRYODw1DJXIH$9%k_VOI4*iCeBN}uC+5wMqO*{`EI(27w~2t`SQ7fR1A4rm zjssm^XF~ZSq2I{T=9TvMF{f76is=MjZMz`-wLkUC;QqrmcH$bHfaz{QV;||sb_D8> z5~OzgSP2cJS45z&FMs>^lMF4F$7V6sTL5FdFGl>JO!au({$*{7FR;gfCe0#m8OZzh ziu=mluk=2e%GxkLDL)=MZV*7$Uitkia+PGflkpc|59Y5=AWUp{oX|jo&kIl>W14>^ zIktpEup1@TotZMWgAEmE*rg)7SsTsv^(2kgzNyjeJPPr#m+Ac&)(sb2OJjZqh|=%G z`HDl=Ug1?H<*acfeCics;r~>*gIZ)blRbdVMtCz z)gz~W-{1%Ch^Otvr_0^>&2R^B@9*4ZSE~s%#Xxz;OX{9=B{KfVnZMNjnFJs-(s0)b zpLjrRF;|;#-n>)f>hHVxa-xBs0S^5U$%fxGrtuy2>4fkf&>iuE8g-@rsVFK)CD~13 z3fg|!m#;xe;+oJBK>jCX{Q`KW|6WjkA7xsGqp}VZ8DTUOtP30{YT}52-0yt-8zdH` zJ$j0A|Lwqj=j=~IFe|l(V0|ID!&f4Bc*P%u-#QWpzzlE;h&kHJ9 z`WIcC@5Me4$My`+=(Zz`o34q-K#}V=Qidwb-WB9`q{g93Cl2Ar^d)nF!eFvPrxxbm zgl*B9eDe{9irQB88bo#cJ+10`b3Lcj*Z`Ng64)@`{9VOy(+B5nm*bDnm~o|X0Pd3W z5gJBMJ}R!!ebkRPu5l3eA^PZ>f7BQAxNx`4; zqy)2_0y=Mz-@T>PZl5$8%Vs$5$j=H51XM7d?aaIV-2-fVYT!-$(A^59P;w1bU1D65 zO7E_p8_TMRk24B7Lc)WTrRfHP#c^@ci7kJ^4HkW!OZ}VFv2G$^8#-pNui~|B!$L1; zMffK@B+A7qvfH&MSn}H_pq2No8O-{rj*x>W*)?l+cR-jJW?XSodewo_idp?)sJKlQ zI8|@QyXdvHHmy3aPZc%y>X2H5^l0%^!TN9K_q+VS?Aqop1^xLwL*8%oojixRo!Y|& z;d?f_A_=pY$0Ih6<`AGFjKHx-@>XvuOQeaMF1xu;Hl<2^wg>jN%NF zN7cYJ45*g5v!+JP$Jb)?L{g7UqE#By#vgp8P4Ll#KgoY*vflx`lJ&n7Bvr!Q+vfJ zfXomMFw+mvE{0RVyUOz6BUip~-~W#s+cQ&e9u0h%_-rNAYP`d1F}>GrrHTfItNM|T zf|?d!gH5X;{kSER8Z!82diyOr1Tn0#%3f?`4%f`)j-9037uz%>a!)e~xD~yHFrD|P(1$2jal6oPa567y%HuQYarG$He25gn0A1bqj2|sd@llQ-3;*-iZ2%TF( zZdUM;>W0tV^w2JDZ?ht6c`hipW*U=LGqQTOmz29RA(LFx=d<^uw-rsBsA!xI{jkwE zlGfd&%vP;CRyWp|9}g{8HPCh>Ex+Hv$NUnVS@qOgA*v1P#2k$RpCOe*Ado zR+(2TssvYXJs>Pgs(d<}18X{@X+duYx*ESnRc4r4_G;`)>IEcg;6t@bj1xJkg)7xz zJ&!{Rm;#-*3}s!I66}N>a~*0$AwqkUAYP23F}(fxK}FR*BpUUMk4O-Z@23^{JmgvfikmIE1&&aAhoQC8kh zl?7sU4bR=a7BITHM^9TJCHm#9*WpzG;Va>FdSvhv$JSV9OGyk9DZ=l(#CpzHE6^_kkXY3k^h>s-uBY z;~|}PfKo%@FAaGoE1U%*?Y8^MUsFo@PnA5wcT}v`ZW_o9D*NI@u*L27?LouDU)dnH zZHVSv=?KD=_l~qf>Z&J_Ojs(SmMX$kXiVQM#$%ftIcaGq+#L4Bh}8G&SfoOowXK`njf zdXSFAbq-7o)7e}9crjOqe&eGB{8*dQAbwrWen|8!B&aQpg10Eux* zSVT$g8|%6LHlGZAi)`-a(XJ_7tSP~^ppnR+9E@q~M@Krw4wa1CCVTgdO zEJd*d=`;8i3aYFOR2$*;>co*(bYKz}xw>blI z)9X#28wjMNAC?Fu&C%CD0u*p9YwLQf4DUX7f5)#0LDXAXLlDE77s z4R}1&IU4@TdH0>eIc@rhn@P`uFVF{2C_1jj9JfSuRv2ZzFq_*5!W>%VKeB8;&i`uV za)0r<=3pI6EP~75)2IkP)qd-KQ@>vI(c0CK==YG1UH5cDW9Uc#T{}(>=I!noCeyp0 zF?E{i3om$^gEkg<)-TmWiYrIfVQcov(9MwKR_kn67VriRx2ag18_OP{X#3phc?mz1 zAykxtxICHs>M3m=Cs1hCUFiYTwrjTCs6)FFs{p4 zY=xA_ds^XM((BtwJ!cl0|6o#a@p67oRGQ_=Tq@_FmtgkuWLrh-M_cIx-_0mTW)juF#kj8IsH_fA><;uPpYTZ-YsK5^xr11eK$tmqKSdq*r#OcMpwft+98XtZUwO zr<-{gmoGS7nJ>$_hT{-^)wUBAaI`$0Z0eJ)S6S!3O*GNPrf#q2+?S)KJ~7^EnPU>F zn7t6Vl9>1mhT3RWuSX=_>?+lnjplgTM}~cvyquP2?Njk^e1oW~H*SZ3jKP*sNVTrE zs!E2B$+CMV!f9nMCkUS>-R_c9)ag#4R0G#Uy*5gY`L+K(=fkT`Ji(m;zCJU3@|{o^ zU*KV*??lx8y@+dUW?xUBBT}_@1ScmyNeZsK;hwNS4^i$XP+@?cED|Qu0#SG(2I#_j(*_*1VM(RP`GTxNZ3}L&Cl12qfz_$6aS2 z%i9GoKX$4VoQ2BD`>NW};x{92M8u4eqMG162v8SZs<&kh9vSr+?tZZ~|IG;94qubu9I$ z7GS&G5!|f0Xc)1MpJg?*;;}G}t~c!T|q zwM-CIm2U2)_2MLO|OMibA23CzP?%V?)|bGRIDY~Gtp4Jmq%FPqO3 zhP$RXL2utdUws1Q62YsxX+5U+My>jufv=KR>X4(ft!@@pv8&MJ-7Xg(dLrCJD>P3^ zttn$2b}MB2h4{E4DI3#=Of7DGEO~+u=NZP)lcGzHYORD3jSbXQ1Kr0}!RMknr}m`N zk5gkaZ)bf%qF&kc8qibkV~Wa)6CGu_P8Ev45y>#)uA3lCa9o)+IjW*hwhykcuGt($ zRqs&tj>}h>ZU0~?ySOGX_JEG6uqUpp_k$~tkq+U$w85YM5aI8ksc#BdIlwe3NC$E=6KyQd!V8(d z8nl03k12%xlw1$5R;QX1=CnnU3DxpvryYOBq&ra&kd;>P)3-b3U>GJSz-ClK#zmeL z8G$oL@MB=a{bRY{w{wm!-D$V~s=`CQZ)M4?L&96C_E93=tj9WA2q(n0m2f8Q#nHCb zNA2E?G!yPIS@woP1)f}e8u7?IJ13Nb4o7B;o0|!S*1xS#_?N2-Kv-m+IZj#gWvbTq zbEkKm<7$tkLbVLohYrY7HcDUjbeGYyF0uG*<0X65&($Snh(60P#J@i7QtVF$`M43x z7=C-Mbu!fl!IGSUti!}HFoc_mwMGd}8x|nXl|biOse*ly(8s9UNhgJ%P~mx;?U!La zZ$+jy;%|j#G_WR&21qOi^v7DjACeZ$`#GIb^BAqN(q;bx^}@5WjZNd&O^MwJ?sYiO z_U&9!P~Ymq)Ciyb+0rS;W$@+nbjXi&q8ehYHtM5FzoNHs2V@@h9wCD`Sa*1IJ!et{ z^GZppM39|4j-{Wc?s`7BtpH%+tlz!E-}wKpHL(6vy}4(h(Oi%VdUNF`f=;t3V&SQB zzZRANbwydc3=G5P-vGj$3i34z5D(SYH}FUfYighZO|Ny20DDNel8|me{YNmeBN|| zAA#u?{`+|glA2S`aj$Kf>y#V@EtyA0TdrP+HwfCy5Jb z{)NJbAE^iX{Nemk2GXV7A~co0⪼Lbbxi%UTk$WG3giLfa%QFKM|PqM+@?G^+#|f z!@aJ$-V;dIKECe7&l@vR&3&kjcT>RxZfl8Y^zz)r^l-1kF-^&(S7<=4eHw+Wgc1fd z{o%ZmYVm5&3WK}tW)HrsF+!<(&2wmaBC+X|hz9tH6&O z4PY9@|9)Pcue?;aU(nrlT|+McZ=CfoXaEmx__t82EAgRvo8xJ#oB5~v6h0rBrXWNl z4IN7JT0e_HhJIHE?}E$bnf--OwUaK_C(s3+r>X|hP*qcaT+y+F9nSTw-VjxVA274*vT()Dwe z2jsB2h;;&A@S5YC{9-_rUyUil(eX(7=Rq}xEdBQ`X;O2VhqP?-1&oCochf{5AY3ATvqaVno?y;gKMC!KiY6!AO#lJgl3RzFpq zb=kfqow_cpCcbbTA1Z(d9_oS;WW(GzE(rsXl#lqPc4Rt5J`ZUiX9qgtt_Hj0sK7p< z=B!?j@Z95UA7WNHZr%C%wxDZ1@1UrBD`jG8Y%=OG`(B}HB~gO7AaQ!Cy@aTOGwp>J zri>giduOE%F(|o}E}46rfq|5MMjUNby=G+D`sgH=@7)P8aWFu#9Mb#Tc!;>+4PU`E zNXAb1fD#<$E4C}9?m3*V=1a0s!g?UcVeOUY0HnpU zmOj>bB7J3WlV^`^?MaMC=%xEf4)w&NEk^}Zw6W|2y4Xx#-LmX^4cYW$&*t`l-m@#ExmIyi zQLlLE_$${W8=g3HIG!FZ4^#=OorA4bjfn17?gY;&*olmQp7G#!=5txOF)XFD1X5y( zx{o?W2J}`O5W1U>1b9qat@UY1C>xQTj=0j+_iOHKljJ^lsg|!=X-NMxELfC$Q}!#T zacr<@AM2u&=SV&VHl8jSpk+_dih@ZiYKM1lp}JcX9QbpO4W%zbITmI;K10v%-r$$w z$g?N>fLv~@8d$KAm|Cj2Pwr~N4*D|629CJ#F*C<+5xx~=1s~fVcwW~wM4U-o;oc&x zJN8a^Q3fiyJu&K5bhFg>G|voH(e|1Z-d6E}bRwy@Mbq`98RERxXSrLf;kPnScr+=Y zcNWEJ59QTV4%%KSue2(9A!iz>j5G1Nn^`yYed(&`Y93-lDCq^}g^}w{iz;p`F<^k3 z8^F7@jF*}|lB=O)@KpYtfpH^Y>#^-;y_c;Gqw4y6uHyU=@wswd4IZ7IyYH0ZZgO03 znGjOURoJz5gi6GN`!xrvaabAiJNqBTg~)i^l4cq5>7nqxZ<8s+YSi5Ru~hWIJs_QR zsqbBADvti$$9HU|Er#-Z%S|;2q4UiiJw;WE^W3Rw$Oa+sKsAg;4uu=yZWq%-nXf;VL?-hSc+yW6LUQnLpi zPERm|?6rb8MsRBm2@pWwM3LSWd~lFNZ^xkbOua%2tXgl#pzpF?PnD zFm|IX(PV8bjWtOKGgs%_=icYJf8BfEpXdAi=lgj+&*$}g8Iz1T;1^?*5efi-004Ac z0OKQ|53sVZKv|esp-?Cr8!I~}FBc~V2dCh#Jlwp(Cq+esPYMZ%Nh(Q;NhlzMgk;Xk zDxCREO-)Ty8mXnOqN${+rt)(Lh>eYnlY?*IhG{HyqH zx-2|qe?TPyc~xR?d8`P%syBS*@4iT?x&_6uvGWVS1y7!ql9rK`Q#*U^ygKrN{v`tx z+R(_@!qUpx#@5c>{Z9|iYhKq=Ck~Q!lIYO zC8e*bYijH2i4BcS9i3gI?w;Ph{*lqK@%IyxA1L#ms0)ir%b#g0TVJ=o?R?+e+yC*? z1p>f-vySh-VgKQR9lMwy5HJM#(*gjxJImS1(vpnnANNT|fH^6KI8s@||6 zRL#Hexdpvr=a*EQKfU#n_LuCxgT?+|vVXz;?V<)VhHSWlw3f!Yc zDZIa~vwE8WoTQA2jUKuLNPh_BOP73LJ)P!Sko<@n*%Q!F`Hca@KanvYZ8OK6N^Kdp zgM|gqa$%ZQVXsC_n3l5@8-_w$zGV1%!WZnu?(8N#K^NI~ojkrwHoH%k_EGioBd<02L0~ z1I%-e)ElkTArISjdVR6NFo(Vy8+=g56LnEIUw5U5V!Sj1h@(=K_i647;P&iMqVi=CzI|KZmU5hpaieF(vzAy?K^jS5Pxk)u1BrYN@t+YLX+<2F~F z-_57|vY#=EaLUu;E#y903%Z5xUKlO&D>8Cz=*K0eY|qpwRA>(Eby zrLChH>$*GM${Xg(zH-Yx=&u#~KgJ?YcX6xss9kJ}pq|Yt!Ib*nFKW7*VKMM9kdDDq zxbTuFqjDy0b8UA^$q>^Q9I>8N=}l3+Q2S*~tHRkLoZhK{?!bGWzd8Nh8L>e4u<(s6 zA8y}2DOe2`Ir6eX+7_8+xi4>4U1k89AHU|eDi0+b;INif2bbK}3M0{75hs)q=w zLZu9rk>ASY>U{M2`Bxj)T)$XWDKwQ_94Xi< zjDo(7eE;}roEV4yE4Z=EON?XW0nfgTmewcj(w2I69!a$3TpC1l`2$d+dOPEeEMp`Vro=ddG zUwb~eWQ)_Uq292DYW|?M$$!KmWTLqpm{hbsh(t@3nVf3KYvBLt2cf!Zk$p1=CZ=iY z1+v}BY4fMQ9Ob0vjqRB9y}SI@Gl|ESMMqXlq<|2Tx@XhL-{j+{7yC)8nd7o77_uzf zoIO69NP&$XBFOv%f>RP1vG z>=x5EtZOwBd;z<{((tsN)3`9^Bf*DaNHWOrPoE#)e&nQYQ9WIYln7XMxR8*j<~tx> zunkXXsb$|P@JD;x$G_Chf?>e{cM|r!=zI8ZlP%osXqMtkBZ8!l9-dt zu`B9txvvw0e4i5xN#2dM;<73g-_@SS-&=3FOFs;;$=YcY)+Px{KOn@-#HYe`i{Vst z4ToooAy3yg7LELQBdVfah}y44xsgIJy*)qXkPILcPYyW~;J@_DWXsDf*R7=6k(rFT zwY3MPIqirKsSS$D8|3gemUC$vN?gS&jIb751Qd+g%ayH0LNzj-x3DB3CGtz7T({UV ztE9PT0Y0D2j^WUfth`T1i3XM0ttXZmwtn5S!>@06ug)JFgxfypwjcq}6Np3^r)gf$ zn{Lx0Z~r&EgwkGnj0q{&NUN6HH-Nq0`kLLD5g2iitb!gI9TPQP)kNkwdvD2eU_3m@ z%h!4|teV`l;MS-trPAnft#o2yfVH@f(o)z%@k~c10rmESBhzSdeRom>>}WGtNeYcK z;lkk>WAsiT4)V}c3z==NAs(~Ihbx#I${ec1F}MbyLvuZG&Y&!AV<>^S&N=wX z3lNnwqMsk%U8B1A>AbAl-1QsU=`kf3f7Ts&6HNJ^P2RYozK=vN1T@8wo<6j;c&Z2(t~^W2J>{A~o!>IFxOr5LAc_SJ6EKSD6Nw69_vCF6SL1q5 z=q|>OV0VkR&$h|&d5BJHs0yjZ5fIJeEy@Lsw$U%T+_P@8NiS(G)85iUNj`LelqRk*!`a*pTmdbix}A>w3|K z>Y8b;Gn2?e^ypbKWkp-jP~x6iE$mW}nUj8~&@Ntr0g##@aqC8`mgh5&(gA~5{{pYF zYo_ppqs4&L-tnLiR9yFie22_G)6NNI^r$1|Mhk+7JcYdmbDUV<;z2#l=$-63?AaOTTdO5-t1M^_<+i-|`EJOG?Yi zEACX@z5n20U46r&#>ef?n5^d=oiDoj`UeJwhB+gn?&tO~xZ z{cynm;!l=%{S)>V7e?%YBasLs>W2#kj~7n_1}UMlO;X9l2Ne>ltZVhFl*+EGB1W6E zp0)3iYACr^Mol00-hlgq_EYxX!Os3)vVX(=;~E8M1WbH+2n=ur?*w&RQ{qza#m}El z79E>@6=itB7T*!KqTQStfPjGP;`1E^DeD3V%n!aB#ftVEvJ z`R0&Ibot!J7a?KT9C-*lecO#ClOCrl7ZC2r6Mnm~RlRhF9rc;$X3Am*1c;e}L*o_@ zsKEa0tXjEqrHp5KZ(`2keh9y^xgv3gE%j01w^#_YVAm6Kc32A-kJm4YzNhf#+2?JX zVh555dAyv3*|l4=Gv)%iiPBL~NFm>OJu&h7L9U0$bgyt(46gR&4R_byw-@%#7XMQt z#lP%((Zbz$M;C4w3njRhviQnTq}Ff=0!<__nv|(1^kX{tt<_;?HZ)X@&8NI4Q457b zEvvJc0)og$#IITZR(*Vk{gM{f@s?0I^RrUngkNK&&=mrm^W_lWIHY_H8@@VRBkF+w zZ+mphihS_;M*g(wWJg#}?&O5Ve6wa${V4m2*+TVoC%uzt3T~s)@J0;VyRPxQY1&TMV>iaJD<4EI_7w1NyyR3 z+Ala8B?fP)lpE~jS?93(1(rVR(3wM}-1#7Jmv3NQ zt7WG6D-$E_HzNI3qhBBtz3o*wdrLAMHK}?w@VUb-cHLvNcK1)0HR(Im`n>^VX6mb| zx4rvLX6(&?;Hn_gIYR6{%`A>`W=yySCYJ-R$bA)=f5eJ`7%k-`OPoqs%DCD%8qN3^ zAsazu`9|%Qvu(XhYRMCzcsA+B4)isd%fd9aPc5c8bNd75O3_LIO$Pd{(=Aac-!|di z2~ifJj707d=c>|^>)HWBO(`oEAduDX1%X;J1U`1u=&-054!^9CBXDJ&!S>Z{EFF00 zgFAD0Yx0bCunoU-2FBrb~vT(#!`Cvkrd6#7s z--xyvjAU7)eV5bDxsyPO>&MaWwl1(kcc1LbsL;>0XLwtde|fcf)jHS6@P;E(@i^Ym zf8XQm!xKZ1Wpv=;&+bA^5t&i8y;GF&6}ybAikAu~fPJ1%)d}}~sW=srO`UwySl1vU z6_!fL7{u=-Cgppm?{`XLdp3BpDN%U&@}OgX%N%!pE1H0SMnkmf&3f`V^$<<9&}Ycm zM-c-fS%J4M6MDuOPaC|Lre8WyHSh-ojlIeYUJ3TeDRk|&)7n##^$QEPL0F}O?1IQE zVVcnIs$H!0AkXZghp;(^{DDc%OFeSDv-Mljkka|G;>bL_hF{da=S39GT0z$vX@kO< zov~l`mc@Ea?lHacyimT*tZx_5O)t*U%lIF$pA27*x}^~fmqAAD{BY4^<$CcTji(1| zKD^f_FktPH+KQQi!E2K42fv8T2sYtSU;StE(7(KX-KKLZX};su{qERhg&+T6#(#VA z>yjk7GWu3CUhOZtwX#X9vpsx$KAzH6z?vy+uX?(frbE3$FVRdNE3DCFT(}q>?>NTi zS7&w@zH2AS^f_(OqOJN^{7w!KGE?}k5&|0?7K!6B6!Aw-(SBGo-kd#oAbIn zTx2qv)}7Sf+(WuI=&)b%O4$=Onp*B*O!f+wJ?CpSJ>(tPn|2ujQ6@J?q2sG|^%Zyt z3NbQ_j~U7e(pVAXc+ncr0nfz zb^rzg0JggTGzM$|T2vGzDuPC#P#6qa42ze*ii>0SOUp>&l@(M~logbe)U=GW)iew= zm6UYMbPWy}lgVUNZHpu3CRRoyvdP{hFboER701d;NXVP0E2*3O*8#l%I5c1a8Um&P z;5Zlp2ZP#y>TaJRu-^jwJz#JIQbZJm#)ygU7Bou%I1GV+BM~AZNaSvH+-@EqaU#;{ z#x|ld3^q!GgEzU9Q-;<&-q=Q9_I=VKd4(lm#AM~<_bU(&XzS?ekq?`inOpq9_5{U_ zdXi?(a{A8sw96S+Zy#Sj|A4@t@QBE$=<_kLmy=UcuUx&BmYa7Yzo4+F_-1)UWmR=e zZQZ@5N4(~i)*l}~dHJgSbw_8{&)xk4gZ#e?4UddYOioSD%+Ad(2v$}ItJM7h7*Shzi~t-A)3U4Xz`NrLbb1X2`hAaI8Rf%mU=2!{CodkSC({0ByPHG~J1JQ-8odnuQooRFm8O7NP<%z({@1`Jx+c(Y|n z5Bn8iuX%ZkC(7nu{1?7{fM!-2mnM7xIOzC;(i z^VOWr_M_TnE0}28|&Ug8SM*c2d#b-v$E(!ir@tJTr zZsOFfMx}&N;)hIogD`FN_=>vwzXYBR^$<@9Fw@JYfRtmL63Lfk!+s-H&KF~wEG<@v ztC#W8B3)WaD*3jKxl^p%nVYLFGpP=V)Xbty=U#4;tJ{N{-h}^JB@^_DAbjfW#O<*Y$rm zRv{l2^Fl34*Mrc#6i#P7H(d$|mSP?0!5@b}gr`s5m#7DCzSz;Xwx|$@PqD2H3LkGb zgg_I2mS)(6TXL+s7sc7EM{eOW5iwwKEB{XX=Te@p7OFkr=QkYx0JlXrn+AC??X48L=` zp0^Y~_noXmt*iTqtJ&w;?x;QPY7<~Ay13&;JMUSIY0E7#%>6+Sgh2hQ6l7;%UhMBR|Sh83yY!Odcdt|g= zOqxovqy>Y)AhKo1zDuk$_MbKc+cIotv67_isY+Qu3H zfdBxs+W_tmU;)7R_@I2eFenrXhr{><#e@U}1O%l-ME8ivNh`?9Nz2MAqO{c%ky^^K zvil9ywGQf{(P#xVBh$lrCfbM4dOwnY;BdI0fS{z1kfffHtdidUEbeOn0Rub$TQCR- z@E|~71c>_NmZoSObR!<=1SSzTM-*kpa* z`oRSPz<*=yj(>yw2NzE0UA(t0g8$4?7p;!mC~a^jk~z-1g45clj%e0{gavDYh+(n!rLlj{j#h;hK{ zq*9#JVG$+F;>n_`cWv#J3beVv*B6-0NiMMLe6EoT6m@Zd(JU^|;@p%dC0)(P+x;@!qLeZ|velwEt1-ow-4c%bu|4&#OgD2ARzEASR}xr% z7NsR-dL~pUELL!?X;{!oewMC<|+)*QZ zK~-?ZmPsne22!-gq$vA!^r^Q|(wOUIjUu}Bftu);FTF*t9O+cD6SZdu8xxjh7jnc$ z$NM0n%cZ;cg;lJB@Xm*uVMc4O_H@@mjIx zlT44Hd7i<>Bl2b_;#Y-xXQD|QO#x*^D??$THKA*ZBkuiWr^hS!?q=#_0&9!R_G-Ya z(va}&wiOa1;vQv7jV0^g1r4Sp1cmPFf3-qWcWJ8^ir18h$*fAh0jkM6VCUH>b{rmp zM`k!VXfg*R-jmZY3zZ7DD)RLgYQ+{Shv@2|cx5BOQjK-Y!AJSqr!E8!G(LMl?zBMR zp)=uWgFn6PDv=g3XZIdo+TecrWRsHBtxBzUZ0xF#qc=yCikMtc$vFF z8gLzz|Jb730XEY&(|?(VMv6$OKW^z}877*3{9J;@^7p{L2rnNdPSn%$Q4$8Veg-Fb zbe#%oX>8Ark(RE{JWotqeQI@*0#1Tfhua55m6hfi6b1iw5PMr+31n(W?4(o8FF?-H zvXAAk&IP>qEV1^H$Vf=h(%L@4u*`(}X5tF*@f)W0ULkKjk6B=eF|MgWArr+guV~`R z;Rn6BhWm0g0PW`f?+T%-_QGJ(HerzViu8R)92`}_|7D*?1O znbi?f?$yO)|JFWMe-Tq1_foTyBZ5~Urvx?*%iHC*A1Ex;k>wa?Oq@>kWaS$2|o<(F0Y*R?VTU7qg7JdWlAw9bc;@H+j_#;N|-S(njw|yCBd4YR39hhq;btW zS3JsHnUEb|#77wwjgFf)BtGxE0b|_EO68AUbxwdEeVMI4oiBe=Y_r~NbZ{9sZRwl$ zvSFU(CKIr%&Bnj-(r!gdh*HvJQg0+5-3gwaiWMt~Pp9>_F#MmQ422F9E)b@Mt=)=G ztW}ShNkP+=u)UEP^WRJQ7F3Jyle+H^=N;FdKPVnRO$P~SvaZ^|_-Xx?B8Q^#A__8kLWjAI|g2*xIa zj^>*z@xi&RUg^MWRKaGa#vStYm=M_|K-`=@jS*xKn{mn?d!=~Fo?rU(HH!uA?b9Ke zj93=L%7-j9dp0a_fg6h^IuXe}lzSW4!tYKQ7Bk0oc;kxor#%c4N%JpNN#y9{2W*`} zT9QFQ3QBx^%%{KHmG%7A!~Gq`wKcPrr`#-~6lUAlOfGPrKj*>EZ{S?l;rNcV!I+)E zUK?ulPdBo#my{ZASTMfPzxQJ)PNn|-*)a6HM>qXgWok!(+~cE6{cotil6_raBa_%- zucWeSdu2^cdHb^9t&Fr^9iysj0_nGOwIPL)t-#x#%ZKs9Asb?Of zImFe+g0}iE0w(g``8tFz#l$I_f7Kn7GAR^r%Atj)6gCO0p=uYWWV9|^2?ykjM?sLw* z|K8_6`~LghJ9!2?u)b&Vjromy$2%XUAC>?-c^QxlfPer12=Fi9VHS7^prfLqp`xIp zp`l@5pkrba;b3E7VUrOM;u2Aj(@;~9Q&Q40K4Ye(f672f$?}5rDJK^%FE0(Vpr`=1 z$TJ>Z?%y{-z`(%3#=<7W!6D_Qqom{hx37mE06sbb8{%6e1bP4w9{~v;;h_tlf$tLq z;cpl4pBDlm5;6)Z8af6h7JNYs9)O5|goKEUgo1*M3}5XH{~SQZMENIk|cH1%*YGRn;}M zb@dI69i3g>J-vPX1CX)tiOH$y8R+uL>e~9o=GOMk@yY4g`Niec_04a+5CEkAPz(P3 zKNR~Xz3}0BAtECqA*21)3jxs$E=c&uD0Ez?1d=LfZygBfx&6?Iq@pt`+A$b-RF8;F z97i!p78h zn%%0(FmX`^l*b@YXEW4jXCQ{K{2MYb;-XWd=h>S^y5i~C53bE7L1r+fj1ro(EnBI| zH`t-0)vAN61}Rj6=3a%#Jl@K76ZLSI^cC`8N9!RE-Tp)#bHNIwdFE*YVd+OV4|}nZ z6WLR?o%}PXypk`hM}-WbisA{+*dc<%Ut@kJ$s%q@54};l~L8?4}-85TOdabG{pxys7=1(GtLTO0sev@F@qrH5U|~< zUcq|8#*S5-DTZ?3FMFBQDBSwt8s+h3=KEX+_e1AGl1xer4}YH1Yn$}A&GkM3!C_*^ zs08H$5H4^uN^^E?{s821X-VpDMl>&8Ni&`Be_XVzo@o%o8l@X9>&+d zR8?O-2D+7#+~~_sx=bzbeE_`s0T1kS`Pq@P>QQF}Q9(|xFF)SXbj@qS7-O#{iCa@> z?ifitOoU%2AnjjPz0fJ8VW&R4^RAf1GO)4WppROc9E3kfc5bPE$HL1QAAqk~p{t3x zwP2b!nEwi=eEJo^JD(pYc9ZYh@lb*do4pKdQ-=bY-A6=rR(qrUJF@QRCPG-%V(*goYU*2?G?DtZKq3_hl)oUq zPgZH&Zx6=#$>^(yD~1~LVyB&{`8Ymzb0&dH`p+v?1;zqOzOR}uwr1*&#e zEEe`>VElEz%T99#+pML1BoW9~{k3OL^eE4lea9nA$RJio<5s-@U!Rpov;- z$h(S!J$Xzbgz5O4W~iap%DwL??rpE-1MnkUUt>Nrc3r=C1GMbU1pDbNvyM4$sY+2u zbS_PctaOEmLZd%9Pox^V;YJIOAyJsWBjCRwoV>$V@mDbqfP#pmTgtAH1PD=s=7^pG zhH9k?>%oy9P*8~Hpor*2Q8K|b!KIxv|GE@!oi$+-RWq5w3?I1Bu8%LLMqVN{Eyyt1Tj}&0+)N86&Lii?7~`O z;?>5oK}t_9=9>pCjxQ-jZVI<#9TakQe4zXXOp*P~e$7ou7#lBO8RuyA1<&MnzmXHE zNYnqHBY6bt@^`Kv^(b~?ldwnVdkzaR3MuY+5oFBaOtA+H0w}EkEr9XWjtXudQ(Shv zLa_`qv?1ad!_uRTx~B)y>p5Mz9OvJK+B^V82|hP!#Lj;`9B7F? z01O;3kFASJhiwso2vZgo%jPlNHKfw(x4E0D_VHw&n>}lzcII{So7j*X`Z6s zQw~b-MmB9tV6Zwe!QlgdAkn*nVyQd((#!$}+=v`Rgr5~LfC>~F?s>odT96eo^!DFK z6@O1FljR~pK2@d9ogXjhzDTi3yEp(Ti_Z%uZoy*)+L!jRjjE|tn&Y+ZXBBsvhw#95 z*^l`E91-qn;dn1R07(?*k;U4>R2`N#mbNW{uKkbfpPB3^KLFK2h3Bi8V0%fxA6*nZ zw}8^j+_5G6j66I}%XPSs6;(zCU9_F*iwM3ZJL0)s=Ox54s!+&lKV>VBF~C1z%|WCh7xh17my$y~j#+mnH9e|DEB?wauIG!@tJK$E zau%d7M_m=A)nl`A#d+OX(2}wy^h(6u*y+IC?H0#5&MBaNp`|t&w?+=Q z{DpvsVrE4iHd3wY?W;L%D|vFlO>cPeqWu01ic-;z%C_q#D+UGj@(5D3L;20D$m2b3 zOAZ>vn4e~i%`S;54RrZY&eUI8Lt4;QJjb4~G7Sxq+onEBd{;Y?V5=Y!;Rq4j8S;#= zoG1QPw!>}F_&P?z{dLECL z20D*1(~k%ZP}CsKTW6_k&>n7WDL;j@(s`Z82FCQBQneJt?Q;JRmcqYM4*cVqqLwp< z9aeMtN-{NP+fTdE6SDR2v~W{2!8iuwDG)hz91ZxbM$1>GeWpgwAAODe9D9(F!81K#_cqv)g**!9wFAsD)O4iATx3A!i81P?$K++?Si#o5y@jc5Y+oQa*L77h^c z$mGkBsnJDu5y#UXE#{@uiO;w_50w=^X&=6;WFC0{aM!mDP$WN^qh^|OKhyz`u#pp)l!g-36{f0g+t z4?65#Y1(h+%af*1+5al!Y64sHHmY>#<%*NGPW@5lL>A5?+ z50>Hw>BvL%>h79)Jku{`+cJKfgBw|L2;!%;#A)u{h-jaY(LT04#(wOyCNtDa?E#SZ$?yOmtvvv%8uA=|P10|@^Si~*up#0{VoESR)t3${ zY9c1y!V21&A+DTJOH0nM_%0+)aE0f^bg zWvUvvoR|NpF}`ySx)UQ!)$Uc7P)7$aXHvLt_=<9etnQzGr9YMprBAp|_%#5*NE8?F zjdSe6{Xwbe6X+s*eJB4kr!`H%Dx5^P#=`8;-um%|&6AK|@d8_#pMQTByAReFLUK!%C2BW(~Ooy-)Cgz`4JK|a0jggyEy z_W@u$Re1nBXzt15Cu+J>NKM@kLGIz66-dlyLG!1s``iF1m3@oo#I*}2qb3iWhBE5E$HP%e??g@y$&ABIIoLwtop z*n=T16jxpb#Cb)R%=?P=HqYqNhiV%o9Er;As+EZr_$%< zzKAi|tIw+9Q%Lb!uf&T5lR^+tT*8yv(fXY{u*Q?*MttIl>K6&Ix^uMd?_W|B9^@E@ z^dC(OwY^JGXmY$vwz_`QTki8PIsC#t*0_cdMxQ6#0ntxA?lV9UGLMA%=NM{XQ?!%Ei>_R`1NRlfrl zxG+0c8>cw}v-M4!T|>3cMFkKF^ukb>5#4W+X0HRLVZ4K(ijGx=A;tNYc$Tgp7uz?J zeP(ZiWGIQ>P6xCXtADlpH3WA=WO&5}{Fykxcprdho3`DP;Rit8q}0D*m`cX%D5s*N zjF4|4>nkBe;yt~Xj3n7jfL?;@p_K(Cf&NeDKF;^)wwE{iz$Xc1Bb)@HReNZK1Y@siA7uQ)jGgC!?;!Pvu>qA*mKEV zHNA?SsIKUqkT1S$MPUG%1Y2wRThwo6RN#6#!a~!8_aX_*4a_l?6a0b-S<@_81y({a z+d}G>o+AB3C?Rx{Akt+bZxG_=6xUbZ;HnX(6BoHuLB!slc_vWeULjK6>w9QP>OiTJ zp_=i;&1D<<72IX`j>%83v>UsIHRmH1;iHTbU9;ST9Odhz5O>=~&@i^2o0@1UKN`O7)2O8H>0ErKjDj^!C_iq!Qr$de z^NeGewi~;=m>z!&iSiS1{PH4>kNF3!M$KHD-0qg2<0naRwqZFDdKSo!zI8jtDWkfi z2cLP1`%Iq|3%ux|_j!h5&?o+R$x%?mbKLb4*+}&@fr2P`$NVTc-v{u7oaT-I6^`Yo z1kn=2c~%d#yl!m&ZI%5?y{3GNdNWMl9cx`st!J*MG@v3M`Ev-MR2;!iW++=4h`(?2M+)EwMbtUyI$= zO@3dOYo)+o;Smnr>5_1=V{3(Yn4?9+r>?#c;g!7?gXi7t*BT@L0%;>Qytu#-`=!^E zID>8a)FYy^Y2^dZ6vI7vl5c0*vFNDx{gnc%1*uaZTFp@!eT`1hJ1;>yNQ&16~Uub^esQ zei=gxB^^f7MALJ|$>>YdA^LkF4bR!W5)R#ujuOinGCI>|%EGZa{4q^rk8-^x3!-K- zVoscDNr#nO#rGoeAAmBYu|3U>xx7=xQnx{Nm*`RCaRg^N?v3x`v{k>0Sc zwX>}djG2_T$V0IQHx+I8rr{W@b%-Ie#9p8uLL!-wLvkOitN8)Q3{}6&fY(~iL@}OK z??Zb2@sC&4p7k3b3d8Xn4P$wMV4R3v{D8<@u9MOS;2T&+jA_c`XfgQUT*u^N^XCDn zbPpL36dE<%iJz7Zb2~7mx~eZCoZ3C+7R=RdAv&bQ08RXxi8+r4053=4-n<)exE*Ij zQH=B;i<5Dyj80jjVh(9ol>I&XEhsZay}>Wx8527#J-1NM*`3v3LZ@2I5lx5}-38Wm zcS_wfBS{Q$8j10(ax{FC$J9aNvA@rQV>Q?Rkhjs@2|-yW$QMk z(L|^piGDyD)l#YSlxfl%bgeb4!cd26?q^Q?4SeNS-3?)^386@`JIt!##&B1@zD@*! zws+XW>R`}_2LRME&LMJ(WuR(TJ*TJJh!sAHK-C@^!|8}YX`U9>3puoDE~Kw&;^)4~ z6K?uoTIYI_+52l<$<$X&lJ!4&$VOSyziDoY)LX487qyt-f|1*6ELvVuL9ql4#TFD{v7<{8+{j^?SdVKZpJ_xDeppr z5dD5X*WgKtFrgjuoriqy13(wMBkU2GYc>8j!NEwkrTNFctt1+X+?M2?WKpBbTAplq z3&Va?*Z~sayS>$$h~A@;sm5y$XX=~8+z`QKcQc9O#(1;0QIbmm+G$h?3jFmW&-ANn zNl&OExo6Ysz5CkL4qDG!v<(^j^&IR1E!alC4lZ)jtjU+(MH92Hv zv4kFl;mR6a=$);(=IOVr54FC}g##bHe*%0UeLo(J_hM!B!(V@_nL<2W6br{9Fn}e` zrF(YApQULK{+1w8Te)ip+3PR3Rm7@3blMC5RHOZm=pud+TY?FqUkqpo^GCJjG0vSw zTt+zee3dyF&2Bo(6My@DaTpTXV~4lcpV!r8QM(^HL-YWE?XFjF^Do<;ZPZO@5V83d zp#n$%rYE@<){rhsR&V~i zFC7KtOe7KW-Y$D1JPl%9wTXS@k?r9bv2;mnTg69Vk@?+H6V$l;*t!Du+KwjwwzEJy z*rUnH31_|94Nq)q?j6nPLLSH6bBETAC(#c8fuf-e##K!mUr(AR%JA^xJLKgztED2e3sEq=UM*Z7T0E_GR6_eb(Lj`*10{fNLju_b@@kDh}m zhm1||KEf1pV3Kz`pUX&_SW$lV3B9|{<7OLJ|NW!iwcHBSOjw3{RCqyqQPi&e+-!Z*41Bh17B!&^OjatYDDHZC)ZQwv?inHIx0&Xu?vzPf4vvf zLI_@ALG<`D5WQ!j%shGB*_-cTexNBY$?4)lTMO)XD|;%C#l7vs{RCd)_SnHTIO+i*F8slzk7t$Rl4@#BUCo`a>0_W~iy=#JQE$AzXn} zN)5~nqM%~9!a;|{zd?MwCberTs=XpCWA}c{3LLDr_*%Bwd{D>L=&Y2Hi+m00k3jV5hp+efW=6!+xpNnHSQor z`l-)G2kddi2Pb*OJH(;l8Jm#2@v?WpgEiF%hvt3ZO0THqTY`7&*{rZw_BcMXC~z_5 zM7N%Wc~o5T1ze`HT}Rr0GJ}NO2}qS0KV)jiO0b@(Xwv>M;xHpU`L+QM$fAU>8=_6= zXfmpo?AUlRlEg1>(c~V041>h`E2E=AuX$91Q_}|^?JIl=hKXDe>${r{R)w`g#J3)s zJpi8`-?8>*ZL7#t%H}~C3b9Md3uq%gJIJ5zkv}eG<*RO($S<#`O<03A_PTJ#YcHE1 z%*Hql5&27?0b#3r#rmaRmnTNgy=Al@8EJZ{QhN1eM=9~|)(V+7kR1UsWYg%_Q2YrM z^>a~nrO6WALA9!&gNZL3L<3|X$_O?E3M@tL0d{r@_x*hn3eyn2j)Ds!gh-baEG!%= zHJ#Gup)=axt`YPq^4^5&#IS^=&F>CM5HFp4s~5ZGwuveO7lUq((-FcH@PEwZoVU|{ z;!7!ZcT#VeJRhtV>s?#4K#)$n;+}n;0*f0E$zG=v7wl_CDxA8l-`>*Ovq7cJ{7 zmfq>8h3zw!zQDfsoJeYxws3H_Sq@qs0Wpes$dF-2`|G&Fx3?p>pWHFp{JdD($lXu} z)@2H*o~lZJ_e+5X87N@k7JIYilCO7VrySaCs_l>EiOM}-`?ICfxQkX=Q$Is1vIKzevi>Y|Y6_auqM9`(dgU~`?;VZ=#`qU*-RfA=y*u|R=3l`XUY z`t-{uiBFQg@=N|6=OvzY7*?)^R7dlbp%~A3Qzc6cL@x0gUwj*_HD?$)+t<@I^m8F8 zUIDov*LW&ws=t$4doTP-65UuG;kzp{7_gEqn*HgVuv3;$_gMG}THx_+v+RmI*J1_C zf2vkj{*n#kMyP0cP_OsdF!zmjdwYAYC|i2q-E*#{zKW9_o=QIPE&M2Qg6$r_p7 zTNWy{4Jnealqe!zIBZ&w2PCvr{NSd!w!^=Ti7e&xCZ(~Q&1X4j`aL4|Wls}7Rx~=; z&YlOIdgu=HmQ`}yMOKssedFcQZ)YT=?q zim_r~=l4XFNq+GKm9cTp0KIB|l`41?^q2Zu%fK#dRNqj2bdc}#9-|>-QuYf%dJ-$c zc?J$Sd&E&!rEJ-K3>me8EQF(!GB)bH$&XrZZR~V?4LdR`T$22Jz4&qZARMXjSmKrFhm?k>$zRt+#d|hIp>K8#!>!+8d05P!9 zBMuGO15}_AcJxJqRz<^g9qTN?^1vXT2S72&BYl}FSX7Z$-PGH#JDk45M@UgiMdji) z5Xyi?zaW;ybvT!{IgL4$6(SW^dI0aK5HX7jE%7WooGd^Tic+z{+BO2#$ z-hZf7b_j<^xEm8qqd9lYCLe5N7=VyA-_ORo|C9iGC! zSc`58e%n$Tv*0pM5r%&+n^~mmGCn#4wb`|W&sJpV;26yD1K_zMgnxAj7RoP}4xQFA zz1G6#zcNFLv)jRy5)S#we9Y0)MC5Wy5rntX=x94`?bsI(DaLcTBCgCWN{TsJ4u|0P zXj^K$kA^biE^lb0gz#0ir+QA9L23hypZC(9-cO!WXPl!}8o^k>Rzur4mfUx+giD;o z8KbsrpB6pIqUg$NpQyB8SvalXx04{Ksg7k%JWYn$491W#z6=WQrHi|uolvAvN!Chy zGtOU{ZH;<|jS*WNy5||uOJ7sWqdi7_Oy5xL5~^Iw1YQ%xEqYN@_wtH*{BQ8~W zOHh&S$4qk8=rE@riyq8W>JPxnIpc|)SJh*`J;)D*>O+<#&;5pT$;)^^5TasQLu}k7Nsj z8HDt*#X0)U-0l5vnOk-WQ!#nmHr5BgnOYQtSj~Hj<^5bKXkf;*6LVkxd+ZMA0JN5_ zrUNAo?5Kv;=G2ed0dwDmXI5!k>CS#B6*5C}TB6T|^Ts%)OIRao2J$n+PljmJ7Dv1- z#$!U(jgIyR>bv0R-uiUZY2IvGpRf`w?70hCe&J?~^_XcmeYk13R2EQ_VGIbX6$|vt z9`JXBuoU6_b$Zd?ClCLMl7x^_0F`G_lh3Jz-__eIPsf0$3)`; zFxX`}c3-o<+jpp2n`m#OEA3)K8$+hf&3V9Pxw(nf>9+400AqTgloCkAkpu74d@7X| z;N+vRxlzxZ`8WO{2*~do{{T)pr3poKFN5A`FQ_JJhB3UhwbqyY(T^k^ZmToLH<9Di zQaw4e`h@J_(5Mt5yJkhvlYYWccmL_XV|~PGs^f1Vd&wI%Kl{|fQFUcSP*mn(B_K94 z#OLAi3cMlEQ4#~0wfMTQGwRZXkQSvS;qgYzynl78g2T1q=ff^=B5*0 z4?qRZvm`aL32p?2Z`>4Vx6=m1XQghSp@aOkgE80Ny6SmnBX z|KUZUZO6@`T1ChSM3Rx9blq-C)DK3M$)F;J(uZ@c<{KxxaPg)_*2^Py`1WLRa1wFH)L{aoT6Vu0@R&D zue%@?x}iuuhzA6wh$~|Ll84g~z0+{~UHqey%P#DF3z7N@H{}_fpDy;dbV2+vGH1%% ziVWPA>dE8XZ}W==iwxdLgb#rm^#)Pyk33D1MkkQgEzk2qh%Z@NeC8H+psC%^X=ai^XIz+)yH{H zAmvzWh2ESf>DK2Y@*qs|yQWb=jJ$Gvhw2^^cyH^Uo9=-2&k)1Kpn9PJg$tQF-)({j zAAqN7WB03zC;KWt9@V(k*@~|iA-$hhf{*3k=7w9oe|@9bZR)0SDjHntCrzaB1@1fl zCg=CBuV*XGG)cU;#UfpQ*~+R_Vk?9Xqw3bpiUk#OvTRTNbdHg9^cU`XG?Pt*n~i&+ zWH(r2TJD2B$NMF>8bi;ifQ);uP+IntlBkfCrKW0Mzo)sMCvfcuA7^hZ^wwo= ze8M)zThQb6HcU)t05P2rXT_4Au~9cIL}A^+e4!Bc881 z43UiTz~4dNlUxe1E@V392oq&^)mGK!Ex#o(#hTx3M&!`vTH66KaXpTaHT7k^`7U%n znQduO4q2}ZwN6UDk$%g3S^MRSaZAv0>@cNiW>z$s=&G64QD-qQV7m`bM>J#I_8;%fVaA|9!c*}2!DT|a? z3iU1SMUF3Nr1HE8R;bi-4;AeH$n;59R#YdlkoV`~#)kC13Oi)Y+T+=F@?IiKVtg4y zzY+B~6}jItdw?yWxM+5p#OFi(Tz3+qiVXKj!&Z@SEU0K_5-MGu;Qi}hmM(~^5%m2` zgKGHXP~EwOZAEW*Kit&Hs4_l?SWw}Wct>lA&)23iN{t@2Imk4c1-V&!qqmU;rVE*t zrA$7%70unL+S~*1u_@h{_3|~eWp?|RfP-G3DOzbk{Irke9&0w@{8_dek9?y`tjUf_ zLTo{xaK+^wS{Vck)qQDZnQ-(BAI&L7$x+GCinV|IcOd7>G zzi1x*e>EY9m3CR)LfbXM?>HzBF6!02*J*mtE`xW({A&u}-FQ$^-&%ybrreC-3;jY$ ze;lV_3zy}uy59Vc8z7Vk$=!mUjvm!78*lPgMA@d{r6bkCLb6Tg@wOeEsAzoT+NYW- zjW^jdcO%K{GlsL4{L%9>L$Q*y1C7+US2UIt+FHqc-MMSY37?NUo#^I8OTOXF-G29; zdH_CrqMZOg*Q*;(-JhYJgwfsZg6gA1d*2($kP+Jo844rh|87XM#NENmyW4yMajE>v zRi5OL=ZbL5&i4$CDojsF$_>ETZYXl?Xqg!W13yYiGvk6(xzT^SwD^bK&Z*;VNbLPL z`{}Bv^z-NN{0usS9sXv&VKY&{33&gFm#y+~fhd9FAmsr3*pZU+XO~yNG#q{{ud0ri zQ{>xUs$$8(MBfH__bU#5geIzbl=v1oYW; z`&FdR^{4w?yRBzfWGmdQxhqpKcFno%)d`#yxkg63yK z(qWpu_XW(?83|IVB8|aV@)@2HIe){)|6hcSON7JzVwa6VS436BcEfAtp>>}lHiO6e6jR=W z$1fEc9hdxV-rH8PzzBSr#Z`vEmJ?RHEQkBAxe0Rc`swf2-i`*>7sPmidqVlri#|*y zSIWG{r97tWbi%SpVAuiQ3sJ9x^Iw3%5)zF$_DD~Gt<|csihEq*Of8IasSlen`Gs3P zPpa^ns;Ky|K3v0DWQwMgk(Y@K)0*qxiI)9FecfToqxyk<3RJbLkmQPIUVL^J!A)b# ziLaZT5@Kd47{&K{K55=8(Re_P6K=$JZ|;$8Mzv|6$&=qsc0A0X%QPqF`DT%Km<*@5 zR~VKp`MXffg6^k>SurY?natLq=P!e3zT_-3l}RSpS!awtU9WDCI2z#Wyor<&+Ni2d znOSPEp;7UkQhByzE`DrUAN(EXcI??`>ofGLpvNak@uF{-4gvtf`aDB zRpHaE@Tm^xSG3iWw^XIMXou_ zvA1Y|!OJIV+Y6em!|JEKIfDXZRiPGF7Z#KD+y z9KJdJ&zyJOr||FypI?l4HyrAo!C#}pMi^0fWoiuPRz*c+#7H2;C@KeLeZwBrm!v$B zi|M8h2H^*=vyvg>kK-$4{CUb@6UYzphP^I)(#vv7N>w5h5b@bV89DA>QrrHKQAZqK zkdEh_4`T|=qwvm#iM{r#T096dsriucNQUD@cZJoIM}ZAq%g<_(Zk}hr*IZ){x6{aV zt*TYrgDs!S@x^6b*W}!q{r&wMG4qsnJF%!z*-x8-c*c#-3(h%ghvi*BgH=fjCKaVC z6?rcNFUY9xj&K0w;V|*0%m-jm%_mS}%*M<7Z7vk5AQtVyi2u|^WSHwCI_X?5m6Glb zL-docOyL1NJ7n5$;xvwsui)KP=WQd+{rQV44ZT4lvU$*lZLvsne+TxbcANy^CHya_ z#Ed^>XlY0TPS?l5m(Jq1J9bCeVw|vpV0gKs2qPX*k)Q;(l!Yy|G}l|4eI!_pD|vlt z6N-s|)l1g%DGzMkLo*NOVjmW^bu?nKASE}3_GppBN`bSwAPlt8)W&^sJ+f5NB%@aY z3zMWe)jGPo8r;G|I(?-5tA(ZTnf&JI-ye;-?b#y9{EmnBF8>$sI|z3H_i`Pbq5EZN zV-+9Hr_R^)oO8lv31O9^&`qr7|2{MW6e5Ift47%{7b|ADXS|{8Kf|I8BpC;`NR3Y;dv_3uKOSFRdhx-Z zm=cV5LPc$$4(ofVm-juKz_+OO00e4!8hv?QBZcRx9I*6d_Au`;`XTxIN8I9=`&*d?mPyoR=F8$V z(+nJ?b!K?nO$E2=5Vm42I;p-A&J17CK&y7UV>bMtmZLS;Ny%@TPllF_w3^Mqj}vjK ztp$Nri1g3Ft50Y;;MG$OUE|?zt58x-THjlxI)O9c8_o1%3dIUNb!tM)=7XPuJQZtI zgyXTltm+zT{4a2yY%nE=4&3VHX#}v|p`#rY77`Ebn+7zD!I)uRkSF5SU%)~fyO0I@#Yf^anV7(+b zle=3-Izr4Dj%;gMF1y|FIVvp*N%`p4V{y@jXKig#`|(cPvH$xZZ)}2f%twLCZk>Nv z`4o+?Q2S0|+&F}|X=JZspcJA95->s2*dCb-uW1V@+0U-?=fI#3dn;i9H&k5~km1rF z48Euwdi9G#erG#>bpDClgfZMT&D`D<^s3dHZ?P9;B~ik>-rmpS!fsOTSSz20u6-RT z4SK;!Hvjxe>(v`nCHbMTnp>PV4#Z8hhCIB|LC6(BPy=rn_*=tju#7e73=djkz%O_s1DmY(p9;6jh-Y7Z(s3o}ZO2pa43R1cOep z9ZrvNkAyn&T%nxxnISta?I{QCF^ccE*pZ*yQGK6`wN;A^`BZlo)X({>PFEPTv|bAC zYHnBb#lDK^XTTrSnQmPYn!BA$LWpU#Z_6pJa#H? zQu*Sb`pKUWi8a4`3HgPKV!wy>=8e5rjZvqrngxj)CmRT-_>CbHDvNC_g*JfBh)~Z3 z5C8(;qQCU=FYp`PfeWJG$|-m~l+jMAfvS>AKRB2dM`qwu#)}Y=jb+ z<4<9Agy5W+`SumjPa9UQ5m?{oJ}^Gb5zy{PLtu0i_S1p>`c^P+8D2Q6Be&tr=E;E7 zh#EfQr~1j-`1Cs=v=PdJm=I~={F9`oJ>64JuNT8kwFuQm9?qTlSIGSzKSOw!`CrS+ B)vy2n literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-15-Exponential Functions Graph.jpg b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-15-Exponential Functions Graph.jpg new file mode 100644 index 0000000000000000000000000000000000000000..d33c1aac2f0069d657177f454af2c3fd3f039430 GIT binary patch literal 10793 zcmcI~1zc2%w)aL_N|ct80Rbr~L23jJCDPI*(jXu)pdgHZgrv06-9tAB3P?(K3=P82 zoip=>bMO7$x#zz3`|kI?^Y;AK?Afz(^?$9k(Ua%};MP-lC3ye~3jnY%9{@cA$N~g- zc=&j@1o-&)goFe)i77~kiHL}4Z{H!KV4!1SWT2y`XJ+MPXJ+9B(bIE?aB@H77ZMU; zVt*_tDj>ltC?s%w5iCMNLSiCf8WIv3fqV4#1pc2F^f!Q<0E-=~85@fQxIvDEO^$_b z2beI=iHr670{nf!x`B;@i-%7@c#{ZoK=m!)1{OB<4IFG-TpS$C(cYN-01i3s?R)&v zcy~0+@L8NG1bm}12tYFB-zhbR_E-g9y7&>^q@t#wrDMC#&cXRWNcfS6=wmV2-{j;K z6qTN8J=fOJ)zddHx3ILbwz0Kyb%VHjym{;C{~;hSDEMPYbWChqe8T51iJ4j1Ik|cH z1%(xrRn;}Mb@dIcZS6lgI=i}ihDS!n#wRAHrWco%S60`4t#2Uq4-SuxPfpLyFRuB* z0N?m3!rRjA3-e z`|tqfh49Jm%=_Pq#s<#BQBO{xCzPY$qY{I44j?n+ucyN4fLMe0G|A+tgEg7#h^)cL z*Ke;L^ihzAi)RVSe2qU;n}S5v%slp6A=@p5X3&^z*{hcAf|s

3dwCT1K;B&pQS zt#eW{;=b;P`8y@Sk>qQw|rt3gEvC^)qKttQ1| zc6g-8*qAQYLegsdyk_MsqaxWt(-^MS-yD+aeKy_LQ^&^CvE(TJR|wZZrh`XWEw%sM zx$8Ym1&wuuI#JMuwj0-!;})@p-lap)XWl0}n^>G-k&zv9=a8@KL7`FI%*xamF69 z4oidAUnyy!_KtExBZ^{YGebCUQE12klxdI>TX%ynI`6S7a^wCd5$x}%*wZhP={?5- zT;^DsDpOJOPg8r<-3~MBq7{WUauh6IXuSI&ll4hJi3a-s*x+?EtMbjOi!%S{S?MDa zYX8}B;f`Tfg}f3iWBP*sy7V%o|5A4J$@HXQ_ON%TdQIvg)uZz)V!}?QM`HV_aqIy+ zj?Dhg6D{5hZ)zmIA^(meuLvw0b>pCffK*a6@JVtW82}+ea>X~?Tz&`2lHc^mzYBjN zI<`9yU8{O;388T+7_ycBzhZ9JCY;arYH^g?s;bU-=ve@b7n7Q2Y6zCNou!RPjGE0! zZFtfq`vE(6{@c0fdawUk+S9RONkbHyK~8f;)04@XuV~vhymC ztRZo-SOvn^uP9tckx!#h`j8x8w!X6AF=FGw1ZNi1zca`3R_zQ+ z^nw!B(>uYPTVEZq_LHo3swKrKRfOIpXev-)iOjCn%$qdFjARAp^Q!ST#`IJCS$hAY z+{*+WU4E+&j_fBIauiqtZ-iqvL_%mB=}I!9QWY+Yss|@{7GP36ca7yQxWwHWlb*Ne zTkFLtF;lxN2%59gFxVHQ-w5IO>#1_-L@zML>x5ayFyp~+)1dtF4pKf$k2r2TPlAyW0=^(6L8=I4q=fH%j(f(J$CaC zQEEFmXu$0Ryp`%xEzxItZ>Bs?jzT+_ix9k|T>AM!&(2X_5>(frcp*xA;lF8Et)*2K zsWq}%6&pt?Q1VqH=t0};EwV)wN~)O3$y%e+Q2k+^d(ld_(16_x=j5vFjWcH7Q7^id zOvG)IX$jpYY{!*vML5J9ncBfqW`RrwwXW9oZ$Q*KgZ&_97DDw$;|w&`8H`mQMpS>k z+XP9ow6Sz@YLZWR;EFCpMWyAqrrxTpD|1j#YYOn0zzMdfkI-k}#paa-ww^N98!aEg zE<=fwF&=DF6KK3oW9o@F8&sHlA{ARx<39e7fQLDPAn3PnnaE*Z`n(U&niKfBxf-eu zHPO<)jdGV7sl#2T+&g)*0`AEAl3c{iyTJ=q3`d){E)>n=+i^|DA?x6w83m<&^ z=_+eKuFnRv%laVaKpk4Vc$=4X0hx(H(T`{>AYI@9OQ_YHZOfw=(KiZL^|R$%XaMke zihRu2nm3x%BlYs!qe0=}c|r{crGMwkOQx-~Wya1ln9Vf&yBCl+tsO1cp9gry?A&W;J|wz=}yT;{+n)o)~;+S6XL_Kb%UC* zxIwXpmS&%%yI2C%d$iv_kSCBNG27sFOIjOTW$&K0D07{Y+DB@EPrDfdi(F6~=UMI; z7nSZm6Q;1$EvCQ|E6Iq7P{X+A5GLJ!F4dXWj5f?q9jaE|ry#-`mp!R>QQi@+956H1 z{>#<(M{|(&uy*jq3Pe!E3-(XYem`amnyYN?E!aq&dgkO1bk#Tu-lc$6vO}|cAmwuR zKRT`0yt|9q=1WQXeZj54wt|Jfy@G~b)^$5`n3@WMwlkM_v2!_xVm_4tG+DOQK+iVYJP2f6go!1s4)iwb|*yuMD3SDt|#2Niux2E)Wuo07I4C;0;be*yo4JJmOWG4xr ziSc5ak&+a*2HB^q)ltt9-DWBnrS^Z-o6HqO%IY<9?h9vaoa{deEf5-+?ZirAj?1?Txlw9oyT}3)+B;kQ5od~bE&2Oa?Gw{zfP?Y(dH~dMw?(m+v*>drm9A}*{7fGUWP#d1el@&_y;`*Sc{|F z$;+E>8exs+-ln?`FV=TYnZh}e3@$oS$8*ofzMpre(^(?t-Q-%3jk=i|vtN^I*TCZf zp*{f;-Z{k#?mq5Pi3%wZ@$znVdg$@{qAGR#iz*URGR6C0+PDOvmAjq_q zIBLG*M4$zJ7|h>tSwR+y3YxC=?dsalJG6B%;7XAM#ZRGuaXn=etYvlpE(Mp0)9D#! zCv`Kde`OhlI_R;S4e`EtC+?yEds8!e*UH+KPsr`Lj>5^;9|?4@nD#7`3C|(7sg`d& zlswXNxVGMJhca=_X6<8x#48%6cJ18HocERGD6}~KA>{w*`ayg_erL`UE!;qXl~sdp zE^Yv>u>-w0wzJKu^;61)L~x8&Hqzr_7YK*v%JLR@HjGUR$YUAO(7)&iDh>lT&8IKd z=Z1zDehn3qN?}XZoHefK&NsewN-BmuMU9N3Cu={2iE(5rJ2aMI;O8xjZ2*5Fte&&` z2wR?GE~}9p^AJhRE7(Ytk5PSNLI76p3bZOKkG3RLFDsQg@VRdIXb~52Ab!4Zz{2i1 zHXIF1>SBWKaJ_wcIx&x!t&rE)PzgK{e_)c`HKhl<8pE{9v4N2M4!a9~ba%MIGP5oc zX|$g#1`R%u(5hyTB2n$$&cnC;uB#O&Pami?>EEX{3?ckEY@7HITiw^SiI^z(f{^;0 zd=f<=lwB?4bFgL@k{CA5TV|K>{LY6)>ELkLFA9_?5m_GepoP zF_9$_n+De4Zk8etF{RljIZN;8&8O#8R|7U;)8Z~I$9<6~zc)uLl^zinSt67?ZGGdf zz!S2JSp+CiXZm!n0&&AV(1y;d{pT}934zM?4+__bqS2d<{KhS_T9|(nmD}GAq_0E9q==p0SS;(#UIh)n6qc@We z!NCQb@qf(Q|Nd`1&+zsJ*yj4Urht69cf%?fe*qeZw}Upf9nc@M_+{O#c!UN*qS3&M zUkcSRza&d?v_{|*$H!8w4L2j^2w)D&wSK}FGkm}jhKpUTr1Bxho9eq9g(mBNXY#A> z?&KNDIu)3*JgiH@6dN>MzZYLUs`#p{4TYG!!kdN&4P{$81Kt9km&q#et{AX$)8X8C zz!d{_DANiz{53*ELS%7QNSfysh(FKFkEzLshYV)==8iN7R;4_fXZ|`khI(8M ziDVozd|#In$c{1UGVBu&|036}Dg1RQjC=pMGmA7#GIB#hwxAoQ*S?^m>{%!I&lW_E zRA)IFut3qm#3F21>H{(og`-7iF!;}2gu&^y(`3y?9|Jh#Fv=L*aL2Z2JX2jinyv!Nc zl=ry5X0W@{Z0~twEK1qnNUJ=JWg{bjvBipB&Yi9N^WNKi_-L>_;ZU z$r{cU*2%(^^;$bn`ihHtooX-o!tHQlcYsq8xv#o;=read2*$bLVld^tu;wdqbxaC<_*|a$QWO2GowHb@WaAqKF$1M_HX4d z%7j9Mi(xlIfFJjC>@t7GWqIjarXdBy8w!&9@K1eO>M14EPe%wY-HPg0KwJ(?r!%9N ziq1po+ZY_I=KrkH(Ehc%pZ=E!{mEl1$!zyof__OA~xGVU{%x!b$Q;7vq)L5y}T4OqV&Q8ty z<=}aY?ZKpIT<}dAzmkWfuU$Le4dPAD%{)rwYcD#{qE1<3#QSLtj?)~3ly1*}yco(N zrXqJN?qp|bADU>e78&prqQ$?e z93@eW;=z36Mhx%8l0!J~eFVQP!`%NCy>)a7a+O&(@;82#+wt=bHSRo1JM2S#B{M%8 zhf$C`N*PjkHpVi#V5M!7_pwIJiGBX#SF$5Xf}*jwf(;iHF$A4gFT^-@AZLW{Bn9aOVQlYJ2I6!N9C+*JN;S?ZzC);;$o&I z&Vmq<_A(&}6sIg}keb|g$5k!ChQ?`){hFq3j%Yx(&t_EQQJKk;L8IR)n=C;|1%*@*Qz#ZJgIqm32>pd`?V*vLh>>ASHyU_HU%DX=nNcmW&zF}Z3{=CoI?RsKu$(?L zH&AgOw^+g9NSBHP4XE)if_KZ~koKyanW_A&0<6KF0uPx{EU8bJ{RBv?l+$f16fGmL z9r!U(v8MQ?Q>UZg*;=^lyopr`Wy}eW$plIgB2*_mEZ*e{KAko{#*5rm-h`eZhuN`F zo(2!UKzZ*TWg^>BXNu@Xb-ZipbB&c&JKxeIpx!DSfRP82!Wk;!v58*SkiYd`+ALgL znqwxk6KW&NhT%`;x@3G56D|)J+hz*ZF-`Kud(*jv;}r^)mNa#vd9)EMKF>{depN(H zq~~pT8f*-89Xz#eO5UlfGxG;QHFK1T z_4mTrjmn(Vn&(d9X3ZMGtCId?3sA?P+8lx6oMNA4m+R^}$vIa*b;8rlZ*H|70^;vQ z9bBM+s>c^2+rq+$P8G8q4yC|A?5}%4g-lCKrTLUTMer%ZdD&GXd0Oe7 zd}n*T2xKpXj>0sbPy}x;T%R=IW_>h2YLY?WQg)wVQ_|gS9TSay@1hYnI`EW&4t5-tzJME9@R+VEu~x7H^&>6#6qDk8ap-Pf4=N5O$8}#h z%w)a$05aor?BSjex*=h89}|%iKxlv=>(T^4c6K$iGYIvW{g`lc-m*VEo*7CM2m6wO z_cccqWiZzs4IOx_7LNu_dtZzPGl%(qvg6!dsE@R4=kG~_N6C_nR}KWZ9kEUqk2`GI z3(A9cKe?k~?HDf(g>8z{%IwOZoi*iazC(@+x=3AHl$&>vp+C|+$;k!$)xp^x<6w16 z{h>i%2=ZOH9T)qh4*?oj-Wk3anKN{^#5`sU=(y<7baT=hX9uVg7ZhSad(|-HCM?>V zGtF7DNhjEdD@SW({vN-e806uxbi;fkW!r5o(ko1&ZJ61U>||hZ^Vu+ez3@(fWZ%d} zgQ#x3345K{=k4Z}IjG##-h;KITyqsG- zPmbwl>aV$Ct-JB@5!Z->`M*OXfBgT~JGx+@qKTUs5k>iy<(N349N+K&14<|nGYCy( z2+>qcfe%Z-iwEmI6TC{^ogUCd%0JCqGzM^VA4=Tmh;CoD2ZaSPes~_Mjz<=iZ$m## z$agd}KAGHy5!H*dX7H56#hu~mS5DgyE;DCafnHisQAA!HC||0hcwGDBBmL?9lYXo6 zmQKNTQK+>^gjCZ7xthyP&c37Fklx1(Icn#ES`q#uRQCnCg6# z^K*IMd%5~VI6qrQ_et0H*f2j*X^QHX{~Ad0*AlsGOl(&bGo|A_UI2_MgE(4vQIfba5an&tm7$CS*nr{XhkMRw;Mw2-8qo4RE^2U(bjTC@ zNi^WWjNBgWbc@_hZJT`IrB&-Pd;1Kvd~?24=JH(v;>3ODVS{&7duG@Qb-%1g84crc ziBR8D0Jo-^Q)9^Sg%WV3Ym2lu2uT8o(0q;iDkR{sxrbaAy39beM-pBIn_UbcOFgKs z-mTXbp+?!*n6j)f>4T&8i$rOb0CbUcg!Jd33kQJMm=h6&e_L ztlbh-q^-KFG*WVS5=x}6w6o6%?!lxCUJZTlUW9EoD#N-)GW_e$sO3G95O4)=7G@ym zZHJW1K3HKY{B@kUTV`Ucdwhn;7rIl&lg(=xmZC@TQ?V$!`Qeytx-C|5kxqAbV;6aAeOl_Bh^s-#>d8 zX>h&Co6$;z70gO};5;7x{w62w*werlMD)CBS{uqwa48K?sCkuYYHDsBL2hRY;;Vi{ zC`Nt{i-dp)xhPr&*}oNZ<_x3`5($GlsGj=ns#vG@Nc`t$;C-@}&WJ!JCiZC4a;pLq zRYp1YYxs5BS%y9sj8$&zCKG;~$M$BWCp)9!GGkzsdoyKBV=*p zBX~EyC+%`wlRczrm1)CoseQ%&B>!oN^!M`)E&p9P|EcE0=$bUC)HP#{(s)&qzJBtS zS@1Wg+Ap20Y8t-U2Rjj%cE`9hb6ei1r&q)SW&0V!_IZ0(;!WgwHZk~V;6dy3)y|ye z(w%_B`)wVb%i!RJef@*FEAZ>4FYX~dD)L_VdJy#t!d4UJb_diR3j4zAq{!Kh>d;P? zilBW~n!;6)|C=r^#jloGJLl+^<#r*QA|-oCflC6^63Y|Eg}17-mA`7XsnIrlt*u1^ zB^3cENa`GWO`=WAHtQ(vCC+Mf|=xGZCxq1lc`_dewiv9MM%?A z)`cWd!OY)75C~ctfz*W+%|R?VmWF2cGEq@cNwOTCYkSrFnc-`R!1{gP{p0z{*YL3I z0ny`YcYp~ao2XC5HTmz%7FC(k-}($6aLgkK#4+BH2^PJP{n}q|i%jT{Mu+YGz|!;;ufopFM=Nk5x0K> zTio|4DgVv))?2C|fSByuTod}j4hJv5^rzp{>P8R* zlD|umjc^@pOFcIK6EnuR*7gk*Hi;6a+q9oDw14Xt5r3#nyWLHJ_2W#d&7jft^uX{< zi)O99u;J2r%{Z(+pgqKJE%?BFWjQPrpln0~1ap{I3uOB-{o(iSX;)TNN8#F~C6K$E zQB-TKX7mA4n~v&>fu65Dy>wvWzMbC};3jhj#2OzRqDXjz>)E_zwRlLrMrT|n;8w)6 z7ru#MMFP=shW|S&`I(`j=6}E?zYe&Dh!R<2Yb>itM5kn;#kNy1iO>21fOYt<5~(u*HY-k zVA4O`zwm^SWbCLt%Ei4|BbFzBi_Brc6~G{m|*`*=64VkRw)h!UXq1~j^L8tfGDsf<~E9Xs~oyr=jh ztUObo-eK|6BQ)$Zq_QY#B{+3;k&v~g~9sTqeD;VA(8%& z6tN*52MPOv&YQQjeY!CTL}{fnHG3@bO#TWpXp-&6%;eweq+QuHBywO!RPQ-fTK+;} zY(DYrLDeuc(;%haIp(KN>$ScA%jMmqNXB0@)i;J>4j*u>?Q^OyNSTqxXEecQj59}( zw_FdP4hx_QloYbC~EZOt%OUq(%F9JNInLY-<|i5k|-*` zM6Npv+Xlum(Uv{y=ORp8&9 z?pd%l=fcy9IAg|)-vyP1RnwfI9^6GdV@Iv}`nq;u9NK`z9)MJ_#k>ge@3e9$zvOLb z>f2qw4vn}&y;&VWfA3{D?As24`Z1=4Lxp;w!py8BFZx*c V_HkqSzOhf3E!C`Nz6YbH{sW?=#;^bY literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-2-__details_.jpg b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-2-__details_.jpg new file mode 100644 index 0000000000000000000000000000000000000000..701c0afc94ff06b4230ff3667247fab4d6afd797 GIT binary patch literal 9375 zcmds7c|6ryyZ@PGNQ4l!F+$2*!Y1v6Q|5V$%9IL~DSqM@GKSD-XP%Rc6?Pd@r_6-V zHe^a{)85#7|L*F%=iFPp_r0I{ofw@4bQXI_xY~pdDccgAfKRJ1}F4S zKr}QEL<9aHWCYTMm>C(F7#WzEn3#6#U}j zlM@#dlvdj>cSu25Ss5jDMDwtshP;xp;&RW&d#SODk!S>Pk)e?5GONq zo3@UQMhv3mq@m-aL0TXbkduMt_Yd^PhlZApo`I2xc?SzCXi&WiqNSmuqot=~V4$Z5 ztwX?jh@O*yOH@IJakt4CCNUpw#T&_4%;Lu?Uh$ZAFH0!7_=fLb;oZZ>FCe*3N_xMH zvWlwO;Unt0$1x}LPU;(&nOj&|S=-pUo;`Qo&E3P(@5wRO*)H#N7kwza?Rc+=C{*FP}$VQ3gXF*!B;d1iKQ zp0KjIw!T3mZEk(jiw2_mn_A%eZ;Jh-7bnn*mY$xDp6Q!jG_?Mp&~ef;h$=90>6kE` z@!2h=c!QbySaMdys~zG>rpr7ozTGUm63Tcm?YKnF-ZsZznwGDv~2$Dje2^o%E z6}xP1jBSh{ke=iPV@Cgt(Q+~Yhd?iPAP^}QU9$+!Bp?tKGD4sh%P74ve&H$_ETz^9 zfix~5&@wGd6-!!t;j}pkBT(3G1S0TcB`=)ND}&q+$a@%pI`<+F8Nwn^OwRVjL2AG_ z0)3(fk1+0MgSwL!wD}R}N0FqdBh?6$sRpK}K_d{SEtv6aP!}Wo7bMHA>G!_)81sC` zfN5mY+e;FGP}s?jIhuMiR*ITl@^iR}yxu49^uQzn9g7N6anBKpg`~6pwMrXd&(?kP z7PSU}x;tz$R2{CDyX-BzSYe++Ol0KE^xYcv*R31fMIg*<)d!w<{TpG?j@ z>-zS@*W7hiZC>0HWq6BW2!X~U5r_r-L(S;RLgyCSD0o+g12Cy9N6#wxBT(oB0F^Wi z{6pLSTRnC0i+s0Er=3gg+GM42!)}gn*ZCzXr4``cgz^w9wa$IqHgOL?rZ)FgZ>dvH zD`O{yn}+wB&X0TLYIEMbC9GFbMtIGWd^OwNp?TV{)FEvwsm152B^x@TkL9*UZ8O!? zA12PDsA6Htv&HHvzh&GzknmgCa~o$pzxcCh^!@YyQXu>hfL2(-oYZ!H<#h}-zmtk9 z!>##Rrp=Ip^r>@?SVB<1U4hkBbI0#1WHP9 zrQ&cK@#Zws&u3cDTtfj7rd8qVynOO`i7JOgn@Vmm`pn8iljS;j;7&43+KI(GBo+{~ z+Ku|l<*>S|0%QI$xH~yMmf~ptMwxQ*fig4|;FKJTBcHN*oq7j>BJWO=Z-R*sh^Ch; z{Zn1G06eG$UtKsLv=su~%%SoQU;z^Vw!=74_a)sDZ2x|8z3(dy=pW`oE!R{b&=vs2 ziVUDYzr`^J87nv6cC72snux} zflySs0+Pv(RV%>AtX|+N3q~NZas=9~_oGZAN0LZ3nLr*a2@tgqx~ZXD_jx+8dCU9b z;&YEhmy{E9{O7gx+xrF|B2d$CF9OXvA<*j*0QAA!xeTvD4^{snRC?_$i7cguS+@m4 z`E%Q1GX=nEx%~pj!x~7>XBA$ZG{xL)?)S|&QS@+-;-lHgxxjF~= zx)*AlfL^8|3uI-7SlCgvwD);mZh2awd;iSm!w4L9z>!#da#!+E-MXuWJhgJY!N4~LF z0D7qdfdrt+e{Wa?pC=y#Fc~Mu9kBeV85(~Y5TGN+_6pnIwiKTzzU>ftg{#!J6TF!jzxLPJP}qaCmh-&Jb(GCZD2Rhd_}xbHEnnGkpQ zT1{;t*Gp#w3p#Y(KIW1ifo9|Q6B4VUxJ@w7Un}~t`)8~SUII0fBIbbTwV=UAo@d)lERl@|2FoZ;{E0pa@IHE zGxNiEW2GO_{!?Y5UJsAs(R?>nfFI{JVt;$R+xla6bzRe>`=q$URbuIQ6pG7@aD8xC zwZij&%Dcm;BYVQ|c3y(lM4vDdA93hq$4Wz>(B>Bu4Y?0r8IH=joW5~OeooS_DKRY> zk3f6f8zmd-Dk{<~8fq~qhJlmPcxHN4jth!Z7l6G<)ygx)Z}CR#d20u~3+IbA3KE@< zwUw4cz4{cos)0b)UNmqtIij8t_GL{ZDAbRPYICQXsIIl?hz9R84>hOh?Z#jv`|Hix z=~G)$I*8(KH1}tZcF|Y;u^X-dag_LW&to2lORj22l{j|=srTA_O=Qz z!1(qF8EEC9e{<*oY>WDXfq{&7^clWY=gxjA@;z~6F18baezj|ZSEmc%@lI-kw-?N% zq3c@ise0M)qFDI6^({V~qZTo!N5!b+s`^t1q*;brdRfw!Qa~0PQvybDK$l8|*Jpnm z1t2yuZ{pJCBA}2^hVOgk7A)m;WbV}V~qtW=OunXJ?U8p1f#dsejK)e z8fzt#1R{X%*CjYKlDjJ)_?M_@8@I$B$y*9jhqQNwtlDxLk2bYgMd7AwdeI95XbQb$ z*|`8OD(0aQwZRqa1je(6&jiB_{gq;sMC~45F#^ZqHcLAKW#^SUmkU#JM)ks!r{vUx zOqlIgv}5hDeW6!qX0<@fkP?V5-)67{=TzR~@FyB<2`J!9u;+xAK6iK`vY}-^vm2O{UE9es_DFwjd0!MoTpxLfdN``5o%aq$Ghs^EVJEUsg z$ZGj79tpCYS`8jhmVO*+#B0aMxHm92zGig$B+~<*R{%wh)mJD6gJxmxy!NfLS4}xb z9QQIyt+L+VcDf@SxN+%aUU_ZhzB6){i*qDaZI(&J-En~gL)!gqpG~#52(4UY%rDdm zkF*|Dy=1ORG#(9XG>lrwRcJhVXnuc9MwgRZ8Qn$BlP<1F6GHGq^YD!(X8F9@gqpl& ziR{B2`%WufR;RdM;0m~Jbg@Ar1MWHUt6y4RnK)-Q&Mo!_Vs^81;W9e*RanSi9?~Ya28bt+dqcLXHfZ0OLsVWP~kQL9w!RTvLF~UH? z#7!017|C9Zh9@F>om?{(MvlDN;29odowy{hAaJlnh_!hmXs^SN=TVK3iWkH{-?A>( zX=A>bfv}^+ei^ztFCG6+V@Xit5q8J84S;HgOpkBW+d}JUI#6!SW?a z4iDrMqS`w8@JrdyovDhO`cf;o)VIL3oM{@QlFLaJ&!uS#HXMPQt;LHYPQy(}UmDQV z>j&VaH-J*@vcqKHb8`r^iU$iHYDJd*FZ>d2@(LIg35I|Rpenw^ygKwek=3UE3l$LGg0R8 zU3|zMy~Y=fPoG2`eeogtvKXVLHU7Sn&P+-XsF5yv6hwrdY_wQM}d-LlrhIV~4hBgRvOA>)7 z!N5g+bZ@3QFqUL=G9p;!w5UEy{KFT4?Q;%0D-lS<5JY>#RNzgk+=RWA8@YPr4@8}> zKIvu>Kf{8iz>z*KW*7D0gY$msQvolZ-h%Ds~BW#?Mj?H_Z_eDk|)i z)R;D51%vt1JM=ZLJVl`Jqxq9XHU|bx2oHzqj0rX^aRS13op<_>?v2;g#?JFbo#oie zw#(~b-?Zn6LYjmqB*N81!Aq|!K^c_A2bp*XrAGl#J1r_`K zdZcg7ANJRd&*71j{}R!(E973=#`v`*8A7jC9gEDRomcwdFQNz63N?h!sfXYnaH@@< ze(NMYc}DSs|9n@JwcJ~Kr)ZIjgAp9R!k4x-;VIed$4*Uq(h0U5XIF|pRyQYe&R=HQ zBy;2$mIkynV+f}c@S789kE>0`%%nH1gsW>f3r?uYM=y7G4;bCNkl&}F>F<3}NmeFJ zxX#&U%d*-#qGh=|_x^|DipJx!?UQIA1d>x+5gRl?j;GpLzr%UXi!P1u98A)P)0rEEsC z@`PVA$j_WCT(mE%*ukb*Lj6J{y$t2GU@|p$K;fE?)w5B{`%^{oR!oDHgIgh(<;I7s z3jzaKVkIXJp1rB4^>wcVHD{%)EckS^X^*^1GIw1<{Dq(rSMl{$-{)F&$_akviF155 zoxvhD!QxS`IzL%fS1)mmsH6^gB+7sNYTOpi$hJp=>*QAV4jr9*iy(_JNdqm*5Z<_z z`J?&=iUW2kXMWOcCp6s5iA%#(KKdCtxwp z8f+;BohVUX2hz$>c4mT%P8k6oA@Xa_E^xA?#1+GX#jr$?rPpU~69dKcbRzn7M*A*3 zsVJ%Bu01!ClV5~e(`^ZS=#>-gts;46VT=pub}7_Nri3gUY{RbX9}D^BdJfcPeC5cy zkg}1m3EzFSA6?w0c?3JX`_6KCM@c}q3{e32e zco0z>0!i?_(znWd;R z7h(5uK7GaLfP}e(`zg8hV)5`xeUz@PU_p6+ff29%83f|_g_rX*xAX8b$t2(4oukr^ z6JsISNWwW{oScSNo>AWLNBdcy;opddUI|ai)p^H+92tq|@^k3trfI*fyj<4#D_L5k z&tAXS@jWr4bf?}u(xlc)j&_v~-9N^){}U0_PS!y~MgRRVZJkw* zSL2=6T0u~9mNZEX@Ny#dVJA;FJzH9{3WL9Uq_<&|h@Oe6MxYRVsw;0z=t!3|sBc0c~y6oecomS-LuMeuS+Qyl zw!eAUv{@s&NU1;wN{nZ*3jeNj3pq{%DLnu`nbUg?XD2*sVY z{RVm}69iMQEOA@=y&jdzFFP#7lSOlN9(j!k*5ShY0xbvAP37*!?+$C$aCY8FTIgxe zvGZko#n1n0UYbUT>-gELmM8emow=o#mBs0bIc{8-!%}<*TH~IVrhWFL>OkEH=E2O@ z#0F3I$&;?ulbGWsVa11-@Tap=^e@PYb6E6Uyr9DO>qcYfeX&nZvs@CuHe``0GP%z; zQn`)}4=KGb+kSGU&3cNK<6`5eS3$mb8POA-Z3-CaR!LWbYWu0_>ZW4&+y^(bOGLiHbi8XX&&Ajh`*-?f zcXyB2y~bxU4~4yy)DbZ)tCu}``bpb9pQu~XAN8_SB)J(rbY#U&h;tJpDl!Y_`A{!= zXR~i`w$m$UBnh5O5`Fu$kRhpPz37|cy$K?IH}Zpn3GgBy9)q3+qn1p`!P2RFZ;vgv zPOlG+CQcmVf&!*e{BBTWGf5W9;1Z=N!#1&@{A2FcSRS?VOlLaZ()G;9xU21g9JbKf z{7-j%f8PE5r|$g{r=F4xM&eM_JN6dE+om+mPh|z&62k_($79c__HcaB)V>_}8%(x< zyh#5z!;;ZlpvBvBTH(HQ0J!F3iPQN$e?KG8*)kf`?9eBNH$=;feLk6b<@m$m7B=!& zzaEvhJqCfE0lp430tD$GUoe@pW7u;reNany)h)_DQD5sF=h&+lDT%46ej}2#FdM)i z;1@|irvh1gxaArRJ<3}YNI#N_(fA7iN!#h@Z<$hp1t1qzx>wjWIS{6lqQAN39E@-m zaL!VmfJ7>-0{nezbGYdP0Y#L9zvVtBO1!ANOcCf=;O%u-!)-}FO$&j$p2&&`^j8qm zi~}dH)dWoC+5&$B41tqEpxCnVpL(G2&H+l)NGv%mn(R)u=O?8>k|&81fqdIR>Xs@~ zgZsWQnCZr+&MjVe#S%q$#O6e@6BegVCs8)hWQ$$^FD2C^X&~x6yHt#WZ6mksq$ho(#5kKjyL+Nv9bxhTJ>T8o_ zYE*5y{E>5{rd-gIhk*#egVKd`!rS<$$U@IOYL%z1OV*|3U4AOf@1o!Sw)b8Hl}9QX zH`DknTlVcD=(l|+i$3hVZG4~DJ_)*kbcl&IHv(0{^P(WZ`soUeYyoZ{)_3)R2wmNY zDg$!h)OQF}7*&8kH+sOjn8W80s881nT)@o%@{oQ+x2;Ca7Qj-OfEU|pLb;DXChUNd z=<>i)7iNcXd>P-0@q|f22LVrkn-|J)6*PHg1~7NnhxmA0{+6--mbL$%E_BGyzX9M1(R2U+ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-3-graph LR _.jpg b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-3-graph LR _.jpg new file mode 100644 index 0000000000000000000000000000000000000000..9676bf17a2fc3f4deec1c84471eec598949d3c7e GIT binary patch literal 8368 zcmeI1cT`i`x9>O71Vj*tAieh@(mSYh0qGr7YCu4G5eTA4C(=6zp;r+?4@fUc?^=6}S^2It*PLsAf}TLn0mRDkO7Z{(1^{5( zya4nx@D#X>gM*8MeH#}S7Z2|?{vA@nI|KxG9+KQABBiFFrJ<&vqM~EsVWnf>Vx*#C z6K3an#3vviK+7sBDe_o?hhO0F?=Qi?!^68naEF|bko++{75(G?aYKIx?%f97-GXCc zFaWpiVPM|FK(_(3H@_1b;~x$9vtiuA#KOkGy^V)YaN__c25wh(%Gw5O>*DI>?&0a>{o!Nar=Z}F(CC=hxcG#` zq~t7Ec1~_yenH{4ib{A@bxmzuOKV$uM`u@e&(QG5=-Bwge~9o{=wnV z@yRLb?EJSb3;^>FSvT*02>Ul(_il9E!otGD!u_oaL$2+>AcW3pRHYYD3wQklb!>=MkL16DJqodjzWvrW`lR4p?I#$aP zmC9sMZUwWydcApet!nT|^W!S2OB~>Ok8JxoU72Y-64ITLqYLl4@o&XByvHWhkokYY zv_~K#ycY{4qpDA@o}ht7wp)3@qh_)-gyApUq!t?L4z>(yADvkuoC&?H!?b?x^$s~s zewK1nlk?*GG_J;l2CSo3WJFl=4ohtZ8=}Ua=!P-IU)$yXP_1OUxdu{Z{l#+_em5>H z-}aRMxyLicuy3W(IjWX#EirsT@FejTVDWsT6>8SpBw`3|OxnltJ*$VM<W-BA_tQU=Zk{`fo8Goh+#$}C*k z*~x8sh0w#UGpFsVj}Y23-O_op(x56qUyST__}Wa$a0gh+R63t?F&&*JqL4^2M!FWYg3sUy4*2amG-(Ru`0n5n=M-Zt4gd zl5`nNtsO8K`I;vq&;tz+w49wT8@cPaoR4QrNLKoD#gH#uZ?d?U#_jDO>1z;dWO1fx z=i@rkpE|i!`+h>Yl$wwrheDz=JmHJ1lF_8Vza4iZK(`1eah73FO9QHIulv(f@$F}? z9|-8zFN9)$%^Ux%%D8_ev6c!yHpi)!@c9<0fj4UIO3>iH{&J$-zh)-L0lMIjg@tTXKPlEe$mAqG>l2__-fjf*Z)oq=}(J z&bq?mmWBjlOl*)eDGDQ)mw^F8^j)_mE^c29>0?DplBZS?|4>{q(Rd2+XFA5t zvF?b{jB<9_1G03};V*|DoXEfUI-ht{KtJ4^s^YxrvW5LZxp|?g)%P=vDe}AiB+m|Cq9hN6${&VL` z@=|>uBdbV0*I?ylU{p8NQE3*lMx5nJ(d;DG+C=y&=w4h{sza$Gn9Mz4KY%qNgVqfy zxY39SP^PlcavbfNi5OMJ>$QIV@)k#=%nxPswwf6E2#dNXYO3zRDpwQ|`e*RUq5*mq zG~lXJH{hQY$_D%m!)(76^^DQ3Y8;f{&qFGKX>wcplKDS1zpUxc6I_#exQ83$!)=MB z>-~nvi3YDO4zs5}(#UDa54226g$9;jDDjP^a}m9!KY~f_#W3&G#dgYP-YDn2NSDui z{hN9l=FH5lOv8Gza-=$U{u#Xg8N^uy8-{6Bac_8d5{oSZEZ8M7@`hsEM^;HHQpXN>L*lP z=*8#D-be;3kQhL@6mtUR-gi2kuL@522<#J%hf2||d&5tpMj5^-wxU8sRewd!Tnj2J zvR+2pzq_t&M*}d;!ppm8AX+IK(jm7DZOVvuWO+z~e zMpuUd9IPk4*sF5E_}y^%pnJK&UDM0{dQy|@n89D7ehv{748JYtNMd9gU#^h9@eIj0 z9sk(=EJQ0KmiEzj?Ko&JM%A8o{!<7+z5!Ek-^4B=3=<9bM46NTPu&kXt;N5mtmsd4 zWX!%4p-pp!+=W9P*q&Q1rE?8qNA##yf$*xG&4;IR?Y>4o`@|YAj~|?q1eL+nGDxNk z5rB}$a-BP59(P{X8M&tKq_{LYX6s)-7J)I?0O&QQwcgt;S*rgP#l&Vd#>Is8nQ zDTG#M(*g6ZLuj1WrJUzjjI*X<%*f%kv1eQX^5bNhRi>+4QvwwmbHnes&ZW+BM%|jpf_KX=+Or)poj8;#6$=v$-5DC%*M|aM&D!G$$p@ z;ht9S_ZjpDrjV%*=8ZfIH?}UH<2`)qL2C%dV~OE+4%3sP4e>Un2&4RJpeSmdj==T; z^-N7iR5jObsCusk@A^&D%;YHb<4!e9YYZ<$?=Wj(a38)#U{#9HEK`fM8E^1;y?Oco zv>~j=!e$QjYHaf=KNktye}AAy5be&*{Q5MRu2iC4*Tf09F{ z|B_zTC5i9wM%Y@xm0ebBS!STk#PO?tCWL(HC8l3Hq18sI^EX)@%Vp-N9c_}6fRjS1 zLsS6t=cw6s3FmcyJLL2V(h;LSDDaCIywa*u5xFz+)toe`nK84};~J#?9edQv2;=9oGf~#}Mhw3eoQ|~VW z|J!TpDrd@6as?qhRY7mF{dz(0;oSX^eHog*$ZvG6FMH%=ap{H&IgM=IY{ROm!xiq> zf{0xK_X3-Ll%CxBY#WnF9R=pZIea$@R8VU?4ePLBtHymVrqxr;Sn0$sMECcP_E8s$ zIPde;q}QLSah`-ME|c9QKim%28_+qt%C3gs_%Rx+In^O;#EJ+52fV^!B4O!wrQk={ zz^hF|L_6ZAJB81?oh6KK@=wA~(SYOzZdjWTAEGmZuuCh3yiajA{OHQ&!w)p@abeTr zyr13uim(5uG_n>(rqNw>`mH=%D-GEPY4NzV(sw;~D@&!*hh`Grmb7Wqt=sy#m#-O2 z`NqIFwU*imdN>q8hAThj*FC&_-^&PsByTs~OP8Lfgm>^CTGP}+(STuj_A)cR$J+Ga zNU*7KdDD|GBW{uu8|255F@|04XM?K&>OO~PV6*6arFhD|?l!5CLxNU%$O|ZHMx!qWO6dsh7dl;~qc84+OPh3l3$aesR{{)r>U?^=%^gsXNO!#Y3?| zIPJ1b?0s&YqmpGuCKj>XMs|=Kg&^jnT)pD0c#P=G{t@jAU zjK&U!!>q6LXR|^%^C!CYdGe%UV3NqMPZ{doH&{5^JqGQwlu(#5x9Ms~|BQMhI{m)8 zvrN(!a#P>C%7w37)mKf(*9`<_rW%7RED8p~V|Avz5$@GziKaB8rMom6KQv?#;){jN zVVGUgJq5i*w#gkiE|#z}ji-(W-A*iir(fE+BO&uV`#7_Wn&#IiL$jN5)x|I;B{A=O zfd&w$YdCwg&99Xd?QbsVZx z<-T!9{DP8uLY4aoxf1>&GwcQbs?1nBNXg5Wj5i^D7C}(p5Ff)c0an zOH*MwE2%;QJgYh|=BWvVlcdZjWp-%KlNw`>Zb6t%zxgCndJDNbAH#H1&HI75A5dDYVyp29ZJ+I+v<6vN(^FL6i+b8@tb6lSs(?R>LPusLgeNjIhTbE{4K zim$q9F?2fy9HDG(g)`&YE^^U_BzdN78*4nIT6H=$h6aKj2}D@!Z8(F)ij@;&9z=}q z)QuQ6!u`(cz+%emDL^;Rj5dLGghI~_ z*({$Ibh?ZN0!T9Mt(|PUaShCNr3I*PKPcwVtYc>9kHLzPv$&0Py0Aj9Dk5j+?@KAj`p-1 zrS}C@VWv)i($g4Am*(G!a?rj$B)J96RV<@LLFVF7k=u(Iq-qx{aeDk8{OoJmP9|e9qV=LCOKA4?m zOn-Oi4ul9(Dyhu2xSp4AtuRvxXU(7!p9>yX9?7HGqwf>fshvg_rxkrnJ!{ zZ@%8}$6e#-C4GU(639ar{ZP-_Tg|lOuek!Hwth@IeYb0bzt?_3258*Qv z#{3!~|8AjnlUuDp{2Omkb*eY9f@pEcJ5~6|qqfIgu)@*c(r?i(FWXLwLykqwe5{_3 zJ9T=m=!&K1a)4JDv;zQ1rTk-60h8~AO`e)bIY!?~w_cOF-O$<^I z$d6+i9v#i&JNt&`+GqeADIzhZRPB_`zisqpV79dT(_;87vCzJhr-<_K*~C^lM`Xwsl%U=OJDMa8z6<71$y+V>L%{{DS|-7FMBjyaPs;l*E1+k|L&+mOusX@ae?tdjt0KQ= z>u(swob>p$nzF6jvYRK7ynkt^%x>zQ@Io8&qhU50{Kg+<}=>sx>umEtQ32pvf7l~ z(#xNYMBwOk_XOP`XzUEXyb`YSOF&I3PLrEodvJ$^!ZC)sxY$^M6@$njV>NW04L0#K zjyb1_$f`Jc&BZZi!yB>0TaXB=3jTBe6TJiJKP;M>O?!}uWy4B5Bk%ESvQx0cpswR{ zT{e5wQ_{hzwI%8Zb`50`C6mUhXLGS;MzNTlQT9@O=G9=K#lByM*5kPx)t_Eo2$e(b zKevqK@Ct_@`fR5SP#JM6Og`MT%{*wee#n7Mv3SQ5{6nJw2|W^dojbJjMB z_FkV^5mZ$$szOy{8*ooWFl>xHu^)+OPbj*#-fy^5Bw>>bIawbI$0)nZpi@?J&vkjf zZU~xkF!zM1)r<;=a@%VQ#rEjD#OZ?#NzUdDOaL@TWRqGm?pf>dyJ<=H+I7R!%d8!aafArMQNuZX`H)r(y7E z{#E+aprW;Yh|VJjV4gY_N2aG&?40;Hs*wGG@^@h@`*^0ox%gGh2CtoAp z>?6-~_qpE1lV?a-_)}S=e-J5c3=LAXA6V0`&CRj=SvRdk2fnr0O1z%%l9Rwyei!af zb&XR|!muxt|K^n7V9(G)w5widloOjSNUKvF!g~yQ~1|bB8c+39j?XzON@?*x?25(VBe#;(q%|X zY^Kd5XR=o5vR1O%l##t8_e5p^Z?GGseCYhkZ}5lU?1*k=g?Zge(o7$5a;-QLDWd!t z+lF)LU^qF&?4ef3EAzuQ;kOj>K!f^|PxB@_?QSgykc0TboNIGz%e8%z?eu(wE+#uJ zEp8sTRm&jbB27oOI%e;G9ViJvPgl{vV86f`^40{)qP|3!i^X**eI?lwwql^ufS-)xOd zW6I)8*}wMpoEesXdQNqsead zJG}j_$C;ltbhv4}DobGe`iUCYBVoryae$Ujaa~RSp_-c47xEAUISA4<{sm=gWTK~% zp^6Xz%N|UeI6Sh8uN=k-^kfyB0H58XvPK#w9XWm#t?*;$IEufg6=0KJzv;|f#l$Z554D?;H@Tari7RD^BG0fpM zX;%h02q)zM&m$SrcW(htPK47*)YBZh5gJ?)OKYRtwTw2y5-gZc2vk zm0x5xNb=}UO7$oD{O@$6Vf<`(T8Tt?prSy7!OhG`b5Mqg@seW@PAJC<->ko~)$iXK zI(gj%8@8E?w5yF22o;PJyd4VYX!zsii2X@w{|USPIOoXCyv~%I(W%QR_^RS&C4$i` zfRVIlWU%YQ=!;#c%2oWqmaSVNbYG-_ztplKd-mnsz2wuf?QIdFnvEcjJCLh+9DJtyewk>!phD!It=i2P1 zZm^5Q-s$C)qTgp}l-791Ipx*lsE^A`7R??Hz(x+SFCA7o_yxR3>;JblzOC|&acDh=0ma4kBQzy7~fdU!`ugzdo zu6N9qFaOm-LuYUo`QS9_w^~fk>8I)lfyHZNHX6VWN1e8|mW;qQ9d49p&Lt_^(~2QI zdD|TCb%gk-Upd+K(%|nu`ARz>2E9Vro_qOt%&)B94CC+3ZcDB@xefHxArTEs-@pxg zmSi5{&uEt8;all$I$oQI3+OT#(}M&Z+li({{6o2(yjZ$f*UFe_UzYNTafJ( zNT8zNl>i#};f@BH?5{z}Ld<>3?{8QQ|4y7kMM>#=LlZCN&WoD9+t+R|ZJEEZrT9DB zkpJU7I)txOCLU?TBsVGcJbc^A;4dO;$Dd9-^5cu~j2k&wa!1KEZGwjIGz+8RfsTN-iz=}_n| zFWK`@H&P6PEpCg80eQ%rpsop~YQnA&>$0Ki3tVa84@ccnTe9kKjhN;aWqeYcrCI(W z@q42`X+2gz1P1;0TIovSSFNfj3{w?{pHTb`S O=>K}H{#`%xKKBI1~yLcPLtlyOrXF6nAKG2<{#V1&RhKQYh|TB)Gdvu;T8;-NKjm zJ?Eb9f9}lr=H5GVpUjiV%9iZ4pS73#*2?pr=QY4fB{@Yo01^@afP{Dfo)-Z!0E`ze z&|jcoprfN>Vq#$75aHrrW8;t!zQQM>BB!CIBB!LJWqkXFmY#!wl9EM`mE#>ZA0Hph z8)0!F9=5FkDG z0B8{FL__-b1^CYk2^j?y?FBjpCKfiL1N0>T83_di85IQ$4HXs9+aJ*mKqWvUq~n%; z@k;FrI=wRy&)2wY42Jhroy6*sr;NO&Elkd>2H zP*l>;)Y8_`)zddKx3ILbwy||}bNBG{^7aY-77`lvJv<^l;YVUpa>~!toZP(pg2JNW zlIogTXkC3nV^ddmPj6rUz~B&UYI`jF(P0$4OX>W!QrfnLBpm z>XP)+Z~BFhnb?$SA?!w_D=`8o>ScF#s{CKcyyo)c$96ce>^NSl@(iLo{ZFptxZaAV zM+X_0U&Ho|)#JWSA4>#P7YmZ8>xk2OP02IR{i{K-eY7oROSBk9^I&!9HOgpz24EIIo&n-aG`jov(Qco4ca$ObyT%qd z%T!+oGG5es;|~??Y*O9(Xw5>|wybujs#MEtVYjieeFoG&^KK!)S*v&of!L7MC=<<{i_{Y;*T4P#l|XX;y#}qj;&+LzLK6y`(MLk~`!+A=Y5#Yn1mbS}{tV(R z;U{+Exvr+iFx-l`KCdO)@+}hhxp>FzoC+X49{xT~%(0-cLlI18{H~37=2vxHJ2O&0 z7Pv1D(y4Pk@cO~S0cc2bmUr^UhHFPW@3tMNHAoonKFg7EUwLliiG9Gq56LS+an8cd zCm7Bz>xbs{47hkTJ_Edk&D|-fLG-AnS0pB<&1>*`z|lJ3aD7y8fPK5R$yp^u^7_@1 z9qDS%@@dC2Alm$Z1{cbnE~mHIQX}iJ!f^LH7Zx(I(v`{KkvQ(^7_mYoo!}yQk`3oK zYU93D!5{UI^=kG~R*Q1ADj4|FDJ%w0OfL4)(QXJE4pXg-x?z1|0*h3x5;Pk3lq7GV zmE1lWpR|!&clQFqeN_YEQtx{G7|J@<^M1$3qBDBu#u>RBz~tr=x(G5{&zHU~Qvb*w zsgp`Q_NA00dt?$DBs3gu9h*VJ*lB8g)8o<9I^2os!cKoI8sax^VZ~hPR3*PYz{zvR zJLp1TB`$RgS`@=UKBy@Y?vc`3RIW17$UF1=jvms&>0eIaDg>2U`DEuXj5K#vN%af} zO@zQh`Vk@pfkS-+dr$6|Qcs>-fKRxf&4_qNR{MIs*)t&9`5AD^@C+zdd!4# zj>}h?C)>OGrg1zk>O-CH#w!Y(9K1JaCxh@kcx{@ZQG~jhj#VlCf*NU{n7~#>pdt9G z)TMuuy0xf}5qiqGnZ4I%lzT-yXhR~R8k#c2_vb@y*#$n2%-vR65L)hx%5P(<`2PF%YUpB&o?MV!I`&Laet_77%I9@#>(?Q;)p zYP74u(+V@mL!58tFgIwzYV_XEHSz7-Pq=8%v;ezX-PJ3x+Ob9Gv=r?fHDJ$x|MULb z2I1kOsu_^e*xkPF4erYzjnpG4m{w$XlVsq*ePY-j-Ns`B$2`Nfz7a9%?TI8G++hsF z17n$dr8!qV0w1Ee7spAs;kF2BM_Gr=(#KsTz!KP`F-w)U5R$xyA3ilpfjInA}HA||OJMM_Rd@dIM9 z=)Y0KYq{Mbjbsm$fxZzM>(PcdPca{TzDiMFARZD^9(V%l1pW-*y49)N^CJWxD&s#V7vvs_-rs5NTU z$D_CNTX}wOZ!UlX;rzi?R=&>pHoW@iJN&OO*OXl`oue$gA6YskO63Y&g~EfFczTkpHoCWG&ycP12%O9(2b|;dk73Zb`0UC zbi7;qMJ?#uv7d2eX#gQGnjd#$q&DCa_3QcMNwO9WVyM8O^`ef8x#gmFJ&|XgZxrSQ%EBJrdRe!!W^^?Fd{_rXt? zCK@MFFOSb&-?*&~J_GuqSLor#X?K#hv%N$}F$O)-ii-tje-XUtsogfY6268)oKb|# zqUN-p0nTlqErSvpPiO!it$|+)L!2K?H@}>Z1SaAx3hfeaCfovzlMOPZTU$@>GBwj4 zRMnH>@4jlwE?3Lkcz00s+g=@g9B49{dT&AaJtaF_M#8yo!YB3Y!FSXSBrs)Y~CDtsiFUt!cH*92~pY- zQ!AX88>W;}ic${B%@U7x469^gZ8+}RFK*^dnsPr(XFg(Xh%tNVe-|UwLu2?^Cr^LH z!1j6{j2fz&nHxNC&ELJBDjq#s=}5HlIsE&QKbl9XN_kQoQBqw4!E38)#+!T%0+G1w z6cM%HCQ*F!JhM*?=KibA{Xn0EIrvWIGr*+k8KB>L8vLR?(?(saE%(qs{~53}^$a+? zmONCVAmAMJVXYR5#^&xI-ZqvzGcHK|C6MaAINwc*SIsQx0}uP-<5|e+Qd(WyIc8XN z@~F6h4aVJ@vVE1SSc7C{T-Kj;T_pBq!>a{Y&^ojTZKLsmMk1=$t?$jlBm>#TQ;tS!|p^Zex;1f zlfn}xrJn%|?yFBeg%|w6XUtxor|gZLupB>w*_@Zu|D0}~QL?f-ThRg@(=ddouPpJy zJq|DoC~_3~vGA|Tt=*oGF{8q*BVc|#BD?3~N^|v@D$!2iSQqnvQgz@F+2X+V`fb?x z%T(Emc-Pn;+qdC-Lq4jk|rbFRM?_qyqiv!|&~yLi>)QO(q5 zl2KwNk|;nnf3#-uBymtrJqh!M4K=Ih-}4^Sm)vvtK4Kz0L2|D4+?RJC6oa$vx@4!% zjiDB<0~9_H8P>|O3<`y>Jzpg#FDp4mrX4mcPQOLpV`KelVaA;A+x&ZF6;`j&Uo2*- zz_-7pErf1eazoAAKnJ-N3&cB$80`cLSJpz#Dccj;fGa=&alMq;tMbgy(dxS!7tHtT z)zO1x*SCS&5d}#_~HwKuPeV9IN=ep_)W;xzfAlcc@em9^2 z>H~*--O~OYp)4ajy#wn#iCjHbx>~*8%Dg9^>$|#^|CdMN^M83HKIYURYbMA7xrCtC zLzNCl*7g@8dDMjq&3Y?}Og*`1URW=XAyVv-x6T`@zYX8U`M#y~7{LQBIRdjQfzy{~ z&SVW=o$1dv#X-Y;ti#J-F5YB{_gZwL==Cy~9)NeGu1_a$sB+u1Op{*fJCWU3wwL&- zS1z0+8%OU>H+Y{mqvj;2U4OuQ-G%opg~mgafge~5+%ez$@~b!6^Q#FP50F($64I8b z`U$bzH@Fd)JIU{~CM1bK!%71z2Z~rW-pTaR>8O{966l*Tud~0uzTmI?X`1bzD49yS z&69^rCWj6@EnGAN60LQN?!1Z-ou)I^L#9tML#W>Udvg!&yLhyC4@z_Yg`MDEtGpfx zJ5gss2jOD|l#oa<6zXyZ$D{WKwpK~sPpjcZ_!pJ5Zq{4Y%SzSpdDJ~RRk!1P&elWM zbtBB}U&ftj8gH&EDU)V)1@@wG#vlelN2Kg#zGIGczrFit-!VPylSr? z-*4;mA>optlH~wjOQ=QGQrD_yOs>Cx*{} z@B9VT?V+oGH$2_BF6xBF-ifQom{s;Wx^vvnzw3pXC~*{_f|PI~0NUlgOMhzA6Z~tIh@CVEFt+PD*%jJE%R;yOoGBsaGW4;V6xo zer8IyT@aJJ4|A`CC;;GA`5R923>cK0K&anI_6?;6M?^^WnDB%TZzcTs!p)c|xO;_q z5FHG++yFPC9x|xlv7iideaftZ5PMiE&b5Whs z0jVy%1}A`My5USN<4>f1$l0SHL9JZj0)GT#1H-QqJ_|<)hzMv`4sVP~sz+Ouu+`r#zd96}E_;m=h=_k>mq+#MXU!TCOb=#5n(P;YGslWmxr!R~ z+GQ=fi&Te@>TNKb@?)b2GFvap2{h?wjs+*?e!f{U$*NQVH>}DSn`b%)1O z)lR@2UFwi49e-0Sndq}qi++u^L-g((h@QRCvNA;B^~PYF%kHbvaR-|8(R=6>jSREB ziF$Q|E`Kz?UDS{^8J+#!=pN9_FflaW)zjOPC@(bh{a1tE6&L!P8`!!>_6*C!_GDp*@k0+Oa>j1=(jk!%5G{dj0#^F9{MS@ za~ACY4Tf=X$Lx^Mhfim8arOdfS^4KFZ?~U036tGx zD&wqUxAq;d0#%~%(d{4XZlwx`;ZSJh)2Jl8-3$&5Tieo_b|AG%Y0mUn`iCR(-R@1J z51cu5a#9tOcq4;(imU8+>td5n;O}ovTP-EJ`!vU8;xFlWHhE~0EvDRZU_;1ch&EF!0y?#`s(Op zC#G9ptG3AwH3eLfFqP5QFz9Y14{Q?gs77nc3C3o3rW(56EdXKdlL*8my__&l3J<%_ z@rgV7u&8uKCDaE@a#*o~Maba9geJ8_c2V;+PUeLHrwx~F>_Z~fRd)k3*B#YPjGIKX zOrMg@8Sp(jW>{-EZxx@DRE8q_3$89ri z&kEI7&NRXk`*uUE@{x>wsx5g2`6ek<0Hd$gEhpzUJ4JL9$$W3l(9De;!v!+4kC#7x zyRMeoFW5+kBvRP_#o`{@8NbX{D}2mxO?I8MU$5h_Ux505hGO)5O}!xH1C#%(t0=ZB zKZMH~NYI>hn^QR~1)x8&T=P8xeh_UhU9xP}GsL5DWST3a9B=u~WzeACE>C%0{{?_s zOg)$a>V9IX0?GFDq-k$!>R@aRRL&f+OJYwl*)XW=~FWx0M@&Af1Xw1b`0;b6`V0alw z)qXE~UC#}Xek-e<)_rdnG0>Dhp)rT511$0iuxj?GC*b1>Eb0DnLnM>6)XU>Y;QiUm zB@oZwmP*laKt1T#Y4tR8pT(8xY1TifT)tq?I-7R(WJdEdjuZP-OEVhjJN$>8DyTR$ z41PPLwW$)3qTTPelfz%4cu3*DE#hpXs*Y{&bL;828C>6p#XcuN)r(doaJ`+%A1UmR zmoUgbw14lBY)K_sR?RDLY~1XKLAt1wzVZD-0j&U}+#YkZS-j0$sFywb%ona`jBsHM z5iyyJ{_RTJz-{X@fUy!G2^9zsCA-|uQJ)ZyNE;j^LVnE!@S5j3l1nR6keDOf5OBP+ zZOha@TkunFAuOvBme)x9?ERvqSVc2BW#jL7epYR*ckeU^0rN3rmn@9s^P$mwJ6hsx zaKG0PT;qD(S=N1Rki1yLw*Di75vAd#B22f+`%FyFN-erAccGzhlvIfTX7Y0|B7Yd2 z%}6P8VIia(=AUKm`O%DarumkkCFzsYY9F3ArMAx>6JukRhNWhy;f(HuI(K$t};!Ks949r#gVIc4k6U2TPZNW`j zJU3xv5-cKI8ul+~$^qWc=&1V;n89M-IKwYMF8`&~SFzVcxAO9@aS6o|daj#E&y0>(jK;qeI3@TllCofK)RY3L_|(C5p> zZhdVLc8PoKoA$|*z9d|FY%}TsNu^~%e_Tb-(ck(nnS6Y_MI+f0+uTo9M_v!fuV9yV z&YAF5%RvkBHz{Llo}ymk`GPOi8{fAr+=3CCh>%A_Wc8sk(lQ2B%Irt8=O$@o?LXWW zHZpmx*U|Q$Pis@%^Gf(x%>I-Z;{1aD(k7xI;1|_Hns(FMVr#fS#uU1b*ta;A zI&?5)GcuI*-n7FI$FrQ>NPoiundYU^KvK`4(7z5xtSO! zIm`0x5-B%W4ufcB-4%SKzX8Ob25-)!(R0*exxH;Ck2u@f!EzIEiGFYWmfd@)7xi& z4+;7mdE8$K*kWPLRH5Su-GT~Zs-Hkr63HCNgEuE-9%NU2!EBp4e&HtQdg%$5P2Q%Z zRwql;{X)AKS9+b{iN|aD@*rzPYKzhM~PaLtq%Ni6imtQ@n5VGS#8s@}HtO z$5{4vTI_U&39#C7G1>A3f!QQIsRNW3!j&BG(b;md!HBIKzlZf?FVbg#`79*F9bCIy zzrdLBQdsW~z;!?7%2q(BG+n2k7e3X<Zer4N zo#tRi+>zwpsdOLZS&9iS4+v_aUDQt*WV}SUN$5Sv`hFS0t&I@bI&ZrZY$I8w=@Le1 zdbVw(#nwN^D6#LO=Gt{4ZetvY3nSl>0s~N1z6C9;EYMe!9DGV*@W3LuYkp}vC~jgX zc?MjeBzQ>ZUpIDiWPBM92iY$uzo=9|98ys~(G-eGSW6+~4VvalVc|nPmHC{_s%8+_ z%00JCdt~Kvf?swr>RGSV%D#ul4@KguT37nGz)ex#1a+HPr%n)MRR!7L;FwphpssRv z;Yeygz)T2+*IL4$@KMBZt{apmJyf7He;@1jUyO*3A3+bkAKGAn@RxX<2YA?Dk8%~HBPd~jOjf$hPoSHMnn z_5&TI;(A*%i3a`&gobeiB96`_W!KvCUP;of?N1Ic>nNDM+ERhxl zsIMC5-9Q8F_`RwJN7UVDpZwW_tA5V}Him4b z54i|(4W;bY*Kr$7uQCU>d)WtJa(@n=&&jcmO6N^2Dh-|F5ep(dRNiM zjw7GONRO|Hm(KMol{=cDRgGAq!^$S&4KN2==cCadqvya^2bm*t>}Q~HcyG-)ChpF! zxId;%MgM?Vz!3|1`r65699h3@bU*vE#c^C@UKZtJ<19|I90*#L85kaLvtpFoAp~(o zMY31Kyqed0m~=r(dy}}>tXwzVW9g0LkuNNom8%X<`A&hhVLE=TDxCD*eD_XkvdC66 zT%aT1%_Hlnt-xr&2=hm?*~(sjoP=LW4OFT&55kwRjk%BcB-2o4l^ishQ!Y4EF5w`< zWq<(^(;ICcEw?nJdes#pTwq{g;FuUm*C8}2Q>c^Ij{GD0u|UHR^g;@*AAmYbUBAI; zpps|`au*GCA#e6Nu^T(;;{jt0Prp`Z^FLoSsG@XWNX@j;6F0XZQ*WyhDr(UwHzYKz z)u^fDtM-cQc~f2lhlCHEpHHoG_L;DSFvVf>X zxwN7NPlMHr8D^V-czd(=oK0SlhH>lBc!vh=Lf!^d_uBk9sqMxg@AL-D1-QdL24`BNsEn!Gj}c38@n=D4J7b!%p(c6FvW4-5 z=8aje--+j@k^Zil-okn!@i{S&w9?|Xs-#SU>|NjC#TPH(%5;4F!aw+VeBhVzCHo%r zAA)pomu^msL1zrzRpzFV`5Sc^c@1CYkp2Xl_(%W##+E4I5FL5;WYa8M1kx_?P1S=bKs@q{iV`v zeB0tbt#}r88^SciIrPIDrx(T|Eh^^yW1gtiS13_k@(7NFi^Se_!wH57GW?ommw2Nh z09)p_V9TEZA_e(2dpVBWU}p5YIngpd$!P}dUxxejN|F)gAggCU+mATo+yt84catB;1oP6ILY6vGYxZJ35OMXA24WT1RY8vStnx&r_!htZ; zyhGsd*5(AaTDOl>8|=e2gx%V|ry2UlJo&87zeN1qz4gsrRq9g_w^LLLgQgU7uEwO4 zgRyT-7rVKEl&@Wu?nay4^30JIBb1Vz+{w18ZykF@K$Qz_ps?@m55jo(Yhm}AE=zwx zdw1eb6{nsIYT`6vKHfzG&Xs*rqu3 z2|h7WHyMgr@&Lf-Z}n9ne$OMO+~8%P*3ZGYc&8J)xRm^~R7Lp}(dNFLr!(uaFoqeS z#nM4ZNxnPRk;)eP9(Z}7VkNZvBR$4eC@*M;b~wH?9!*COYf7=-tn+50d2{{n#X;pW zz<-+NG~9-9cL@COcTT-1zM2<0rpihmC$;#?oOQpSc<4YsALdm!pnlqyuR*h7+$gyV z_366RQPzl~%)FT|@eDwjIN>5WurA;!IYL#Er$nT#(lnr)=4XxF}ekMnoaR| zqLkXU*VA?rIn`qAFTvwu>Cb?G%t!0Wj3WmA`9<#V8T#Yn{xjeP zarXuC@0CHgdT1^#5D@t);wJx(nGkuIf6avQe;!Qnzb!M`1>T zb-ont-&EPUt{RV{w8WKt`SKnB*ljE&Pl~fB?TcIKV$u};1e8(HzYkx{zVHtVa=Y{V z$@F9M4ei$iAf!uxSv^NjY7Cve^8siN6v)>Eme0o%Ms!skxN* z32K7D?rww1r>=|M#3x7V=}c8=TANFVTzHpYz7>zGKBqrDs zXH6Pq5Ywck)m2k2(WhFD8ul?^P>5ln_2vN|;fH0QT;%@yL#FfHw7-!jzU??=WOvC~ zsR&yYB}7K|zC81f!w7dzg;?u6?Ui`x3NY*vQA_i-`y$T~-^RWZDx+7%pMET8Ak)oP zF5p|k%1-nJ0}w_#SKvkdroJWhb*v&(=!+5E=~x`10w-MCt~QcWLW8N6jw}=t9cqu| zSFxq!sV(gI&xA!k@lES;I`pY@VYw@zleW8Wldkzf7g0O$&C3KHs7-UTrJYCfQq9B> zu~!7^;Cm!p@M@g=BOUeYXFwKl*wY;m(wRd-5iYM<#C}$co5qDa?Z6j}!~5x`okFJJ z*92}60%G*{SPxwAmk?}(p~vcmKSoHdW1O}>0m>v#2J(k0omaEZ9O6n1_ZFf&lPa_( z&PCrHq*q zz<)SrGkg^vYoh#rTrIJw1Z_^+>i;|m3rfO219wDq6QLtr?QOBoNMp*p+$(f+vALE% zHrGBDMRiqmF)IK?g1!bo>ge<%)%ua;?aG0O3UqYxQ=466QpX+(%8rqxf=PFRjElJ* zwxiK8snvu|8a4E~`%&cdTQGi>Ui}Q4yh`r%I#0>^$jzz7>Ur-5ySOa5!f zBRdccv1yHnf|h_iMBReTTKoNL4)Dn`vV?80dYBfBlUz^jP2l=+S~_&pD_M)Ypa@1c zVk=NNp)+T;ULP#gae5RZxs>^nwB;_?~R$U!L^VDYzf45Svmp|tq6E*#DdMrj_%Jsd|d zX*G`}^QQqRW-P{sUq~6o_-Rkezwe8$Jp<}?0@=H*oLwoc{vI+)CCOjJcppBNc=0_n z`-U?{kanUFRJnNF5YwJ6S4UwtpNd!Day>F6tzO(95QwjwQsh-9=W`={kA8plQ~eQ- zqe~QVA|ySZ0k?$H;C~WGd9!uLe@9U*(oP7ocD~v;;Cl7bYAOKy$|I?Tp&aCge z@;KIi#7Y7|qf^IR6FLrl!glhH{s>WBQ&ukwtsU6Zd|8DBATw0~fF@;$0BQ=lEWBA+GIHR6R>0I+>0!XrS%f(L-P%0&(0!4M$bxzW{HyifWd2I~3QanIlPK+9jd}k;i9|SD!A~Fe z{)@W&H=rnZm>suJM&VXt<%A;mt^#xp*_4&s?jY0?a`FG`q&od`Ye&Mat3IEI70TPT7_!HSS7SsNS;E$PFbjZ~awkRVh!P6lDnN zgBgk568SS6 z&h=~Eztpu`e|xsdzj2NP)G-cr^qEcW`~=}If2+-y7zxl28n->EB!wvs9?p1(kzZCU zD}k2VGAtyqRp`!0;wXmZ^}_V<`S*RAc6Blnron8Rd|Vi5_MRV8mXF_kqzHMKnFn#a zD3?UEta7C${&m)zwAEzY1MVvRzRw&B`C(NqIYJHHTTY459XGXsYxvH36f;JH$Yt2& z8Ac+=RJjwhEnr+n90d*WSLyWXtIUcW*T*1X)e$Im0_h^vKC0E~MG3WhGGo;^_?04Z z9*j0NRMT>;zR<^hJiNBqJT@{fE_M^kh4(6lw?RpOD`mWW_fu1#x1H5WL%DKqw6~8$ zzbhllJI#b`lTP-T%f=M!}@&i%* z2`7dsnGuy)vO&Ft4OX;_F0`hKmnWaR?}I5yqt)3V!dlxNcxG%mIgnwR@WLam-#gXL6qvS++6l)a^NMjr3s|XMW6U3iIF;`>;V#` z)~8Sxv*CE1h9%io-C|I*WaK3ym$Wl?#4{|m{xeje^RQoOvup=ylV~;dsYkO>bE$G1 zfb28Nv!l<9PaweEV zD$G{{a%ab_t$#^|YTXIZ2%gYkxjz7+aE`oq=Ne*#_G5GLqUc4A!+K*-*S7sheiVkj z+WBtL4ph;=^+lmSNto+SJo>1#3;6q7%>D}hbeWizn9tSR32mG`;Vi*+s7t zQZaR@oNiQ=lVH3!dzb|MG-Hwm?(r5OPZUxg9yhF6DD!5Bef#N6pN6$DYkTN=sxs!= ztqY=ajUBOwm2aD5zK3!jH(BME#T!6Pm5)#LPv??d$q%lG3}9v*BJh;I)mCh)H4;67 zyQCSVJ@#(dU*$j8|8^>;kLqJY%_sp)2K0*Z#aJAXH#HXc$rQ1F6a3?6WwZE8;~7xT z&xMd&-}+rq?CpFZ4*j#);HId0cm1 z1yP7|&sIg-5<1G7$*blVWF$Y;7bDWpQ{YvGM;7=Z0{7Y7Z$F8H6GPmK7IjDB41dd; zuXqmtO?}Dp?KdcVOH*iL$ixPvKjYuX`%Gox*L`!DfnJF?9<>7D0cr3%nsxc>6+~$b+^%l zm}}6_b6cRtz5{JRf0P-x-s>cWD=$cBRHQEv?2n!`_oa0)XFK$qJp-sWA8nX!=Qf?P zbEoYW7yoXw47^J3FkAVR--*tM1p8g(lM(d{plI<3X1+$4RHR-(`0L~t=h3TnI7-CD zW@EL1q>>Chq#@YRZ$dO<@pN^gKjKk-GhI@7+9AtVFVYbkL*0pYwprY&&zK&1CXsDr zJz&57qvSZsNdmeliH>&?L9W&!FWw@Sthy|e$?ksfgDQ1A=~CSRH9$yx%_l7IO@9M zTc15MjFux`WYb35*4P=pj{p0kcI9i=-XZa}-8qCdousY zOSCt_9<{{crDBGINi7*%9<p3-C9AcH0ypHESBlhyWw*OV^#jm2 zDfmNguR4(CHl$;8NeO%y+v`dqLF+pzPCQ`w(C{VJcIbm%J*k(t)X}Z>rD3I|nLamk zaz6w846nFs#YnbDA*QO2OsPP4RI_zXOq#lo&v}_mewRpU0vRxdE?nRzB$XO+J=GuI z;^S(V0X@T4gwAnsz8^c6cN%bN0EG6siI3T<_G?c2$%68gok_}R(g;Zy{%)dQRY)Mh z;{cK086UhioEvy}-ea)HfuR1$sZKG*+SYw*;us~NvrO6_smcWERK4;h47WM_EH-^3 zw}MpU<~xD@j&V)AEBXVTw+tN(6|9W+yvbT zlX?AdXOxfnJk2t9=7G1JWS482e>~~JHguvgcs?F%3DYpulU|KF^qvW7o@UX8J6+j`m!wF+Ro)-dtB9Z-J^5x zpShU%$_VP~1taF&J{Ltta{p|1f;wJ_D+(tsi%!4IEWpm!jH-=3_X!K@=_DNPh>OMw z%ftq;!{q# z>xxt*%cDo-sSXY41pd7m232Ua?_@884Kr;A&rlOO?J2}_fVOKt4bU0R_l~m?U6*y1yQdPs>e0Rvz)>i)H%HA$oT6wzV^P&oG2WDF5 zJTaKiywc2ac__4wiT$CfXlzEUhO`#hCz9Y+gI%OYlc z2KZ_NWfTvsB8U2mR&G(Z3mT-tR3N4C(f2v8=jql2d^9@+7W{$g`K9aXhmJjawcCA# z_MOiFUEyBo4YF%cGJD469o9Z01he32xl49X{I-IjSL5#)=bv#ongN~q1uHYCpOlQXH(lPYN%qX;GUqmq6SLa6>CfgnS{3rm z5eCv*Bv?ywcm8GPZt>nS?z-0Y#4Ew0KG;t5zOJCBIE+kjtZckspz%@HOR*}<%F!g$ z>lAwiaPtV;UG}|x1f{wYUyCqMaOzD_IdmO!x`0UK7ahz@paZa)(zi~4@m6yccjC~{ zxmVTbU2$KBUjnh%?8#?Y%PIn^&13epK4tzG5X6%@jFFdAswuY7ut6kLG|$4?W0{Q? zafP+|G-$G#k)-bjU`57zx9xB@RGd}caE>DumwZ{9__Nsoa)%Ap9 zS0&V60=v0X@#WFGFEuByhO@O%wWGK3ll&N2tk+*1#X=^xG-6X6!o+cf#vAdNXWR4l z>g%f)_fHM}o4!KUU`YDI(t55pb#)ckoMD5n8I~`T1O|V(`gp+wiq?( zrkEp9Hf_Pop7U`*ngUlIE7U=(WiNQ_0I20-Y96@h@au7R`m2X&o*yM z%*JoN2Y=(2+X1ldF@;W?_$Zd~*{1eas}jZQ_T-x?Kh1TwQ>s|>@L))&?uSb{;^YTO zb+~lZ)iy+A7}VzGHEvYgnWsgL8eVb-*DQL*N@pi6K`*RQ;-?m>tvC-P5>2V}dkdbn z%cS)^%(j=}0>!^eB>Hsu%k}kO(uRxW@eqM>gQM;t>L7t#lQbZ*aymUzk73i2gRt2i zE+$yH!x&wSZ`ynU+$w?=Z5^un{?4&_M`4nlglsbW-qKF2 zD+SwNlX96RN+B`D!9Pj$Nmv}FPC(>?+>{}=^2bcc!{`J}gibzdl;nRL@#cZAESc(; zl`|+bb6=iQ}4ILQ@JhrG;68FwQ-)HaYy`st* zqZysDAqD=R#AFp1ftKGYZ4712=K1D}5|G3zn0u*Hkas8ytGbkC%j=Pq*@|UVqiE94 zlnc6OB;u=Bew;@C!Tq7oMiSO$Hp97-U^Jr?B4IR~aPj$k7wU^1y4F~An_tK@p7zjZ zA+L7M^L^}v>mK{iqvZ`lZ}Oxnz#`|<@LznOzywCBp3D%ALZ!? z!r?T=J~KwV-2w|*1yHK1#;3ofObI>D0Cy?V!VBb9tY?5~wFVrM z`1;e7-J@)AnlIH``yNcoKdN21h3JKbT7By3A4iLtv#MHVxstVJYC)u{Dw!`VW7h|$ zk~5$a^B^CAOmE^%<9FdlN&NCO@4ifH(cs;hTAbDuqD`&*QAXi4=xpD9XnX_>{wDH;L`5JJ2ocMEawAk)dCGy}#&dqf5t6YGOz=Ct8HGpN{ipnOE?#vgSXr#2U{(Uhx#z#BjYu3_jf?k+%wY9d5i*#aF5z0!d z<zQQPuP?EP2R!fuE-U5(8#^nLTDqs(jpulk>2>3-Mg zM~b!5akNL%NqdJ1x?c)qhm1xNC$up$bGP;C5&bf!^w^6Ng2~-R(AO{2OS{fHW$E<<9Q7I-^Z}aasBr#98+Y<963%>dDYpH( zrWEZNlVUr*PjV3e01iuui!CeT4u2v_A)4Z$=A(RZx}Z!`o5D>K-2z}t{BZYSHC!jz zMNCbWOFl-2_2@#PxBu5>s`KCu|qh~F|B_*+`SFcQY_B_DE?XVB;K8qByiWR3Dr%S z{we69xf=}8zoHmcE7BXrmTzQKFT?EhE}!cfu8rrfZPBV5CgH)UHov5d@=2tFk-#P~%B_dWE3J2ns&lHObmnA+aZ@P5^y!Q+ zTOC51-^-n7L&b{(gWlial)KiB*I=wJ=f+lknp&Un5XJ|&5X!MMnRuyQhYiW&hzoy; zey4dGf=+LHVS(-;57}+!1ALzelW=kh7RDu{?)i18arz=gdHS)9f{?`SBOD*s3+eznoqVsz_w{FZ zVeW-vaj!V+sspEsEm=3a31uy(}gL2PQ!BjU@$R;5v&skGxx<;|z9a+H|(Thqs9Ha{u}V&?oV z%}-U(pFJ(fmQBT4n@Nt}f@lDtca&u~5)4He4~{0$VNIk*NA`S|b;(-pO+ULk>70|V zFLxSm%hj#*tu+;*J0(qQXC})&>L9P#r7Ys)DDcP^h~t8kD91)Ne`LxiUfy!<+!0`x zebHCR;-$XhtE)S5b^D#&4N%II(aO}OSP+dTu@uYH&<~)A(aBH_xukQ2K}!`Ui%AuZ zqP?Ei@YLigvZtX@&Pa?ooh^^Q%6yI#|6&x-B$0@KSgwF-PD`10;xZ;At$3W54?EOLW= zpa&Vsn}uZ6hQ=^869U^8?8?N5DsRR(WOI#2y`wSqbdixZ_Oljlli@8HcH&TpY%|I= zh=%dcBEtpi1v?NgCicrnOfqd6nIpB?r)YJt$ew!g@|#E<{rEjwA3MLE>x1_qcqt|) zcvmh%>o1zEHVj(%TT-TrIY~TUJ<3b=8c&+#=_y)R_N&$hQMH-&)>?56l2+o~&Q%_~ zyp5m|!Hw0I3;Y(f=~;!G7RK9{$9DcuH*p0<15)+HIlC2&;?%#L3{eGKSA@|HX?}R{ zO?XY*{m6Be^t1fdg^FLx-qCJ?y$iU4avN6e*=A=e9Eg+!lh%?Zv`XytVzH>iEstxlO^x$S#o5r15*)FF3*{Kb{hVaY+Zu`O1$4s*`T{{Gs5oN2h-BKtB(QbyB*=WzE+L4mRC|p^sH@T;pLsD?`$=jj1ZvJnJQBnf z{HxM@t2m-rZ}nTOk;6-J%b%dn^skt#lHq1q;gCeia;{D=YuJ1tC9_$_aj3-i7i4E~ zK4yLLcv=S^Gt%@KEN-=%SB7iZ_B$0aDP#u+_?waVSC;Dk01Wjjy=waYTj(S*%J><`REr_>t6KL`y>?lWhPV#P>ec&TE?M zX*c?PmAtWBEIWeSqq$L0!5e$~)t|9LXR0C5=DaUFW6F1s5>WC%$iU{f3$#rSNEY5(B=g=R zyTRd(02HwlV^^#5=KXz&2irZKFU5NTK%Bxkiz~@$5zKr%DYb%#ItzM zPzU~d5Pp>FE5y7r_lrXJ%(a3q;E)_2aa=sNtp>9!c?-3;ImLBoaETbfP)XW-D+9z@ zTo)4?q$W8qHyrTQ!y0;?l&WY;sd&b9w}?jDL|=I0wRGMd_=~4q*|dIPS;FIM^)=@* zS)`WoBe*4oGn(gabgR8Rl2lg8oU5s=;Z8F3Ij1>C(D(bTek;8)RPzZNc~M>=;u~9s zx;|0;)5-FtFc&p)-s@4h(=Ki-u2?)M8c5r&I%HQjHTI!B%DYO$jAQFvbmPk&Q;J4YU;xp9U$JanvBq${wg#?ziV)6(=wG2g>yhOA+9&1zYr-u@ZPCYt9D zyAZEGjcne#rOc#4%mbW}Tra^(u|BP;AUv!KmGt?m(5FA?68`|cf1P!sjz&bdEA9}n001j#Pw_4m$-V!67z-3;w-#^0d53{osy2&{|ysC<9C86xpQAJRU zuM$l5{vE%aF}^rihJ8S<0=&MS=FRWoReQm?opX)DJXgK=r=JVzL;jtg{{TTX^E=IX z3mGGdPbdO6g(XLN&QXM&r?Jt8tvXG(J20}-ZZ%84w(4C6-9z910IgjvmWgX|C@r;I za1To7f3_yKXx?~aXypKh3xIubS~2Q&)453H{o?gT8&B(Ael1m$^ru2-7Sc%2vBNV%k?m|T|RFRTtt?Htg7Ib9Rl|zwYB|9 z=RjRH-%pcjFe>D)EI{j9b~e!1L=rCl0C|Yy*Ei!<4ElMPBD|ck2e!m z6)HDPSwHIMrdvPSG~F^+$t=jgj@m37TX+uU2sq{R?j`-ftEDxHMO-UQQOpKh&l zXv-em(U}H68qM*ikF9kJi7jGcEumq~e);tt)pOz(jcx5u{3H5oh^(LshF$<4+!+09 zoA9=g2C1ys+}vqqduR{NkuK%Q{_y>4HkVLZohO1Mmh(|JqS|<(aq>$TUI6ZWE34DI zzY+MB=2<3|YYRl!ta;pj&P`j^uC%QKL7&LBiV0+l#~f$^l1RpJ=qoqC%YSR(x#RLY zr@CfnCmeys0V0HwMcCi)pNZ$yJV~f&6Kq*@%toK>$BrB8(!6KlZ-=#)@tM#1M3)ln zh@k79g1c`9coSam=B;_H-D}XKl7FmN<1S?x&r&$&(z}%)^C&u~~v!p|K_YlJFF(fi|M(!)d zyi(IlOU%+r#ALf3MOW~2_78IZhO>P){sx` z?=41p5a3j@+*->204>aWy`&Z8;@3!(A9jU#ZrD3dm_R>F_53RLhx`*PkM_MpN|Ay4 zwGD#b*0;Ji?9S(2wig;*lx|~a^5xm$E^+PYT(ypj-g?fjZ}+ywl0wP=@A_8Pi!E-Z z@b;T+WpzEc+`)`VyC8NonQI2IXKL#nvW#Ov3i;6ymprk^&-oRi*c4mR#NsQ9J9~n+ zains5inuHQKAd_|n=9nJwOe+AVGu3y5R4C@1annoo5W5nZB5;xxozxnPT}YX82swx zz22`i+{RlQ8Jab~Z!~T^f1dR0HnrKA(nEhGtkFxSvn1000Lw@a6gmF&UAXY{^VnQR zzSAs#1nv0oOqTxuSCY^78dNrnuk#t@-QTDKHK1kHCAjk2Q-(1jU@F8h9DCMAu@D99w0zK=WaqC?jx5JyAM|^TaZEnbZ_nyy@>;-l@KZG?MF5m3-HwkNHAjF0U zNsymi$C_3|^2@3^j|f`ao2whWO5#ZG9e!(h9At2Kz^th?n=MuK3sl}ER%KOjw{vde z9=!ATR-cEwHx8e#Sxs@dr;N9iEc4~tDtaw*Hr_GP=kXg_+v#^Na+eLt#11gnE%d0A zZh}x(T*sgI$49w_NN%ha;KnC*k%-15>Ol6c64vPxT7r2Z1{9bUL)>SY`bufiS+!{0 z(WQYUXrpXlNC4;g*UZ<}{{UyzrGw5%aVOtHl{x1kl{q&XGUlw$tUNOMNYhy~tFv;A zhm&-h%91)(F0rQRT71%6%*^OV`l%amU!d(+-w3qE)AapOyw5T>DYj5YKQ2h(rtz+B zY;JDtuaO7c9<$M=Tz`$~>7BaL7Vild-rVoCSlR)@peoj%gr zQjX~fntTLgz$62o=UjLO>hwF7b&fRyC+cbQI}bDLh{!~M=b`%6%1fOoY$X<<%#U%o z@Ri=7qClc#mMMspyuekJyB<2%E3bIMTU~Hy_Lf%hwXqP%AQ9u6>H5V+3s>_uXm%_G&u*$K4OFS z)rUO#@m_hU_@`6RZwwb+R7!E=MYHaXhPz(~>hr&b;WH1BBRgF80=&241=L!XiK2*4 z5NOnLPBET;l|&-tH*#fV3m&K8%d3rl!urxLmF6|EkAu*Vai2=?``-ZTdVR&T7j}{C zkb&o+%7O1*&EUx`BJhT|wwEwTbZz{{FrGmQK+SpL_^(#D)7A;zCL3@Wmn7BHe7YkA z6&ooXwvQF2nPn0|BCHU^0c`PsS)L@Id9=DLiXZVEv%-Vcu5ijA9Rme^{GB5UNyD8=%o!5Wp?m49AIuW$6b6!n?%!X z?6jz)i%|2LYq?4SBqZmLTGFKDWT`8+_k7IPES}?j6lu*D#JxC}+&oO-!5?>mIj*zC zP6v%OES{GHQ~n>{=z8CataWR6Uf)uF}Iv#6U!%BO$PQ3P}iDr!YJURij)N?cGcQF)ghn{}& z{dnTJX)H8nMR+xbh%n1W@e=BN0q0G3~57WH=05kWb z6Vr8fM`Mak%;sbU>fJ8T$ky=>CL-gL>-yGlY%^!@u9H<$5L-y&2?T$27x~s5-M-k1 z4!vs$KX}#~4vI{~0P1SHw#iO?X}0fx2c<_WpEu)5Ba=Ow;6iQj2C#9~)j#wq^%=ky z5*`Q6YlHYHBoe*jq-rJd6gv*5n(rkqF?0IM3FL2=hB9zwoDx8p`_Z#fmW{@sR6+ zKp$Fd=AkYAn^Qn96>>I2``zKtw1A$NM6xDXqmPeUEaYfI(NP z#ZxFkxgA2W7$@me2DY(e`i7BrXB#TK(K0D+fRT#gv~L1x+HROFB)hb5Nh|?kr$13% zhec&pjB2Y$Y1raVf@gi9{=`D`1rHykK`+DD<&es4?0nPJfFt@>dqrZfRF>EK!Jv|#W{Lfe*991(y>$LU>_6>kq09#-AGLgyxYCvmq6bf9~Z zf2CsUe-AYsL|gecmidi|VVEK#y0RsaZ*RauDDJ$GpJPI#>n&R5s$0oC{CT@4huQ2CjF{vVegx|S&YeoW+77~cxD39qM`eQ};L z50t!sr`%%|>QqwXh;y@mzW8Y!{i0jRco7vs;2XC>Zq+bQ5?+S zb^QxfWzz%~H;neiM|ZUWLmt42rv{zkvkY)Mq}*<2xU^jEKaF}d6#7^bv$^MT>E0|| zQ*naEOH@Fr#@H6lI{;~HwBHm(HOoA)yiQKYOe2gZ}^mY*%|l7ZQ_7$CbwiisfY~VXCZ&!OOz}KjaN*HiP0zg0lp- ziz5;w^1O_HP%F@=qT;(ajUU4Lg@(OvbvM|a-9aFwxekXNGEG&7NWZhbxtTQSSV&l9 zRAl9|UV}wT)mW&@l=8{EFRDL?^$kc^F@x;V#T2!;WIn)s-bsVwkc&wYB3hV9Q-CD*A z37%Q5LnLp2O7oIAuVtd6*(=IJfNuQMTas?N)q0;Z0*t z{?WC%&~7s(6{c@TXVR zd_^twwUJ0CL~YK;a-0B0KMMK+mYgHX?u6xjhm!a&!1MTTOV_g%&z*5?<$&pc$NiqU zt_nX4>k0;RZ9xA3+q?1lSJL!SoH?D@oZ_T>;+`GVz+d>2U!3!wA-eum9G)K5!OCd& z#e3%6iT+jel@*_`UC8!Jx${^2BRa7;I!2;Ty}SPag-aiYbx^^-urD`b88*$I&lU8A z6nh2SUdzn!KLcqRjjxL=q?<{(w~0$-+TD~6JDRWKjT=|E@m=)R`eo$Nw4W=-3lM)0 zE8Qt7y^^(!sXu42^Qq9aq-G;Wz9fIsyL0+hEM67y1;(K7CXJ{74na20kL6!d(NpXf zbEE4IvGas}4b_%H=SsW^qd)3fC-kn9!u}q;8HYsn(Y+RVUqR6N%&K8vdFJ9Nvu4mgFs!o9wV*2XQ|`L|H;&b2Iz%cI{OKj(vg zg<{R&KM*`*e_`qe<^KSmXa4{S`m&0b1j>xsK05yZghRy1E(eG8JbL|={{X_Ge+YP> zvHt*ydsICC0MD!c0EK;(MG6TG)NJvuf_hf3rrh5D0BLFWHmGgXA(|qYJGyc!(5@hv zr<3AXpA z5`ioeLP-lqTbJCcr-hkrd=vGz7i}l6>Jk8IN&$aG=_T> zx=9_hNi?2ps(?G=8;&_RscN>NJ6h?tHwXb}qGuWP12yIvH^mFxT%=nPD~265WlbqN zD~&6ju|;{M*N?C4QKX&W5wBCTrDVzRjop;T3aA4F0g%VnG@~wUklb3__8Ke1uD&hn z8?$a90Q4ST0;sRWuM@ifTGBAYfL<^M9`&aq5>4!T-4)^&UlBZI3<79g%6RtVnBoMF z!mKNSP_t$**xs7*bJbmbk<^feXpyL@BuXYjqJ{rMHrcx&R8vvc9E5^CCva7x-- z2`osdv@s-%9DM~_f zE59-R?1tjW~-_3a`}!wvu*zX@1s_!N$g*+pJUdjuPu+_o#cR@uv_ul zCZ!(}FN)+?TmCF-MTMyP3)s)G>9kjyKg6rRszro6{{XaUf8k`_AG=`@_sPXq)aw5L zOkT=ek4U1t`20b<5Qi3UXVY<_FCX5%3t8K!{{VQ?cpA&6;xA_-&}gmz{6M-~lV=Xm z+%c$!#7nYxvx6T~OX^j-ej)7SdLv8dbM+*_`j zZdNH-0W93$;;^kwR9`H}m%F*^6jzid#cfHy<{5AQ0DUz${8ZGtX_n{y1l5l^C!HRv zMR_xPS=875VRruj_tRBrJ~3)&f8-Txf8b3|odS|)u~A+nFUH+gFn{7vYySX!wRT$n z0FJM8oi{|fhRPWZp>O9%8IQ6ZBRV@v-43sX?Y z2Tg;Y_Kiojfn1FBN-N7x#Y^B1{7azY{i8@eDqmoq?CXE~=+W(rK_Cvr9;wVr&K^FZj0&xr3dts-lqWDKz`>^A|1Mr#RBoMp=QHC1BO(bG{}UxoZ%mHKmHTA{{ZL3zwf$Hqd7K`*ooGJ zJ|=b)S0DcX3EY44;{O2m-7ownQJ?v7fBWvM>eYJ@^{QRYzKY`J_=#t4{=OBD{pmoe zpYXM`aB-FY0KU)oR&JY)kMl@w{u6Wu zA7X#|?y58=EmFCh>p~9G(fOuzv{xnn013Q*^W@+6-CnoxD$k_eU;UVYY%*7O-laNI zgn6X22v&sOEMBJl6~QOOxSl|RO^|w>_c}eCyB&QM z!XJq-hQn#H5Bq8V0M?=YBu0t})1ROEXgK{TeQu9(y^Q-E-4(@u!hGI9oBJkzEdKx& zc>Yxv{498}pX~Sl0KbBdt%9*zN0nE+<3LE^w=*N#QyYWX8u5|CitWki2lib4nNk9@T&UFCDerm z_B)+VQnS)@`EM_-A%X-Pyv`MK*b-~UybE!r>K-h#)Zw4Pkj-%;k!cINTqZ%`L2aSE zNF6g?x2|7U>DpAcI*ctfs^fAj22y|EAQEee@ZP_qU3kvuLloXk&z3A=I#b2u6;r&I zC;-paw0d?bWG;oM6Pycq zGmrO8e4`$-Cy=QEDyReIBcIEqR9amMifp@jZ#8M!d%M3adKKNr<`chf7?9(=8*6A{FvvGB z97sn6n0tFwXf7py!W?;1ZVI0+`BdX7JvNckrC{efZfcZnSD#DNEN7BNQ4>N0N_Sxj z_9xdJ>lwMZAmtS^;Wsdp7q-JE;X0I9yUi%%55UZ_4%5(hGDk54+>mKZ}9uaQ$%p< zXBG(`?GarGq8^%_OzZ*Xl~2DFTWh)8bNJTAjbfL#inBh*SC=^bDJ{y2a%V%LYKsn! z1X05zt&$D~6n#mpNp%aT-4Z)%UopSCA!0thm59p@oh*?7=h}u0im#G0*N&B~1^u{X zg^ZDIL$nx(C(z>+ykAD_b;Y|`mwwZ#+sJ}Qty$lwM<6e!IIA)1SewYZSAt$JPt2VA zj-QQEk66>)nVv@83VfjnBL|=6YMsUF-pM7r2+_3e5#9J8XPgeb0Ik&8*d3P4>_n`< zE^-H2l{YJJ!|<%A=PCeBFfsM1A#x<#K*e;a#XE(Wy9$|~Vv^2Aa!(YoG7pk@08zlr zE}rqnovbp$(ybUR<&hRpRYzUJt!?dTrE@14T%jbwLRK(XjsOFSi)*MS4jA+E3fq<& zI6-EXIKca$o=s%k-`lLXMppx?V~We#rzD&oL4SqDzl**tF~Gc1I# zz{dmDvM03)3F%J5Qf=9sMg6uEk}K2vUP}% zg=J!To)76%^7_kB{dS($CA*5`-$anzEOELvS70Y29=}=zzM}~F?x1Bq#L)i$JQ{*s zT1UBxMzkZ!4f2D*_b;vM7S-;?FeyLe~lD@M`xmGp5O+Z@tcFhW==^8LsZhCj> zd97;*2lkby$NaWK`qo9A{hhkZCG!i&<+p9e&~@$7w6xtu%1u(#30xC0V<7s9nPKWD ztV&g&`kilzhI~tV{{H~aFXQ>MK^%gKnc~ z7~`6xzv?Da#7ejs{BvF!ZudPWtWQ1oWH${|5%;aeYr$_IMAYt)MlmxCekQ(w_-qtj zYiNgZpkxQ=#8-&wv7&1iqi#QZ{{YUtMeo;g@?Vk94|bisOt!W~0Gid%tS#?#Sj5qW zfjMM62KGMHQ(z!uR-T;e699@gFvrYz vu3C4t-J?w9TaCPFUZFA7!F+3v+Xc0Kr zk$`}D5z@Hd7wDFj(#3aZo?1#sk}v>J-mC0ucGh=mPv3&UR|}EB_pE;t-bZoeTf}#5 z`JWvDubj$fS>lvdx7_*+(v2wL+@76}OZY<=kKyK^7yf$5{HlNzr|_M2{6D`R^wLk| zSh|JmT6L>UCP~f_2Ul*uSJB0&PNRD!e2sWYql99v*q67~`g|T_?bP8>Mp@6G6-h1i zJy37Bk*sacNTWZzBi#0>w9PU@H};O9hrMy~ukgaZ-Cyyd^H`P*MeMDtCApChI}~65 z2DvKNbJgg(f0^Hh!>VtVl2Vf0yIFc)PoLaPZK3JVaxGQcyGipC`PIL(TJ-xgSoGK{ zFlt)M50gd5`|}Du)8ZLE?wx3#?@gM=nk!}a8%mXFJIZVS00Lj;Tc1MHE{0mwQM){U zDIb+ml08dA8`a-Xu^)LG{cK~@9`yy*qpsM=Yxae*RGMXv5rR)O)B{9p-Q%Mi-AZvv zqf?9Nzx+4Ir)_Zs?Xp~3NF<565!m_>*wK4p!Bp?%BeR6|9)YkST#-DP* zE@NJZ-JDmX_*eHjee+}x2q5+&6|Gv{(uAWe(Z=wqjcC-T@XK#6C+Koo-j4>Gej>Ns zwZFa8*509}M<0bP8C&LD9QHZuT>ajoCDrp>6Otc*Ye!MlUj7K9Rr2CRA0~Lse6n%Z zL)B4}vCOn=&mq7dbQMzic?dtWKS>?)4LkF9EJ*0wqfcF!%sBTL=f1t+#DR9ja_iOMTWnm#c|?X*oa7tGPF2Gi7g z*PA;^WNfnMC$)51kQ3M$n&F;e0|3Nh9l0j34X`Xpl}3H)NlUqKTVvf}aW+fjTbNS@g&>SbS>&G4* z)KC38N&f(#n)vb=knLP-J4Xh-!SM(Di@KNndddF)pqlu-S&AWUp+Kri<-pS{ufQPJ zb>OLgv+P;o07)~KkoVfXE6q`vSZ(}ktMKj2&Z`SPz{<)$8j8OOtaWBvT*#NQ%@mBR zWM&Sh2a{S>ej%RUdQLp$Wx#w7FmsWHKj)gpnM@Z_+rJV<%C`G^kS8w8rc)#HUwE&Iw?1S{?xII47YuVz0-8VZoA7&W!z$$&^FX!bwUQ)%vLCvA(;-MhrtjX*7jQ`9ywQ|YCoxsvF|kO)*B_!+Fj zCgjN(1LR!)01!PK{VSo+tfg%(-vvN%xUoK_lw_}{qjt;DZk@nnVf5`<)>>SWNswIQ zwOO$ix@3FiQ#6B|V~V|T<8-vzmUdR|<$m_xx_v4oR4+QGxrX zl7AZM=X+R&mgl5=Ji1fj$nE5oBPz!DSl5hk-m-N_q=Q$vExPbE|ws?3EsBy+p;Badp{b!;3uv%UmGec{J?!ZPMpM>4hbL#_fW zQ-#4XoHo?!Sa+=&$C7YJ$gL+y)MxI3HpuSSABVkQc!K5|CR=F!c@aTr4+I}fQ`W+5 zUiK>5`C>0rpZtgMc!rTveW~wsY?X9%6-MKl0P* z?_J-GXB| z{*xqMZHZ=+VBPYQ(T+MA>Gt&%z2<2eH067uDEvDkvXN}Y7fr%KdK~doZ+tk$MIoLS zLA|44)OY=Ad%5pr)hC);$-ZH?%FY7fwuQ`>dIDTde|-$zR^mC8ah^X+){7BRI;k>! zjX7xIa!aOpfS;7$k<-xCiw!MCr)b+({c<1FRhxBqq6)=Wf_WcWvt)nbXKBYCV*da! zT1OdIt1@^>X`K&=hx}AGpYMN4#d4@VW&`SRS{^Nydyf#`!DPW%8xJE66yu=xu6kV( z&R8=#$#CpICPyUo=Dv=M+ou~YhnZ3fo!#{;n~QwlPZ$06DR8VZ2qV-~5NR57MkD(a z7*bIDo`=?{wuj^^Hk&N4$gWDLV!{20HKUxCwk1L@sp%dA(#H99DHkMu?mw?{UV;V} z-nlOX-xt!Ou$~o;>HrIbJTN`|y$G(foDA2_;wt-Ejz_IZ9FgT22N413T(^$3^w2ap z3}7|X=gWZy@j2=E4yWl|E6TREuPm&nhzm0@zyq%};riCD{w3Dr)YzZ0CS$Tg{m$-3 ztA8w2VW_&*IZ-ugDsa0xUxl!nbkt*J-r@jRzr4hfPv=}zx&@8irE>dPM&8cgIOCe` z{vG^{s-BpGIj=8GRW1BUB=MZXE3!Y(@%-zf6Naf$w3R71*)58};vq_!l}IUHxvYFU zqe4*45dn?|?|PbZSonG*1#OwKLV~VH?s-2!nr!-P_h}2q6K?Y)WK-^(ezcJ3)5oGg z8ieyEeXi%EKxaApKb>6iGOaahT7V*v3fd^%aS( z+t^!NEv=kV2^3~roPd1=blQj>9(ZQB2+b5m8(2Z-10cwLr#SSkF|KBs>PekLJCaGR z+#VSx?IPX1cR4HKT$d!0J$vDB$AIsKKjoyK%DB*fwERIUzc5&xh`z7f*KhFESv(PZ zS$5^*IQ%Py(9Zt=w$1+lJuY0<6`ktIcV$vsuyg8_zW4q{wbk*oNYx;cd8E$B)2YW& zPhnVAS2~os{rszE63A5*(d7H1864ueRD`I z$tyk`t%g{isIFI5yNVq$=sc*T#=rvHo-3{Jm6~4sJr z@_K$d8M+Ex&wF?0Jd>4T2IHNhJqM+0ym%xy+BrW;cY#vYElTc7o1fiF7F7ttWSsP@ z?-c4Ob*jFXr*4t5JkP1y zZAZjW_^-uQLi*{VwOJIh$lo9!cJv(uOXJJ%vPYm;{_ZF~dyEWbOnu+NxxEibm&7{7 zo||eTis0vTk{%lz^!g5!wJH#&7g4)u_al~79Nwp~rjuc;TBP?{ovf0^OT}vhyMf12 zM{lK4Yh$?Kc+ahIz6G|p(!5`J=UUI@1d<^ymQ{0_?wUj_PVV)=UYdltQ@Pnnq@=EU zuB&QqG^>~e%$B9l{`xPxp5nZR!Z%SiucoEesjtg!mn$8+Tu7c=tjNraRB{0UvCnR6 z*7c2A$3@cSzSM4GwzffrmOx5^dS?~lo(t6cKd5{~wo3~=BG%<3V)Lx?NVg|zW5_}_ zryWV;ax30A4`}fR{EM2G{d&p&0HB)q#vocK!>%$04Sk=lkUTr7XQ$a@{{YZUd~cO& zSd2-y05=oWc&dq{2zG3c0N5V2xui}P=H2&On#R~=!vXVSky=_kliW=3p}}xD&rWGi zR9ey8*SWFtd|M6N@G-J%O44KHZKEV->&0x`Y8pO$GXi@aaro9!i_vx(YUWOWUeegg zz@(Cms7S{h@m2gsDm>`{P1{E1$EmIP?Hp(=a@TR*4L)NFTq|TcV{(!Cn)B3J$ zVov64k?BoF^S@!;DARi247c9+ zbH(t*5s}*35D*OR`I@zE%Is12m#KxP+Cit=t+Jd7o(CLCt zVyCa=T~zvYsf4s^sqgE|KQ1biAURe#;C zMgiP91Kf8Ny`;l?{gG-d;CE1%CbTq|Kee=|Nx6g)Hi+Dl$o&O$b{-8$l||M4+Limu z1Ym{q>T4NdVM$(3@M2K5rX+e)(j-P(#xEvw6ibsIrfL~{IeZkclql*;9AH;*ainQF z29+$ayU8r8v>7JL{dSI(gAJXTfn=WUNg!tb09MWychnAX_~x>PKBUu>TeqOpV)_}@ z{u{HoWGSedqqj0`LIMhy_%h8|LB5h-Fl9WDr#{D;>D@G&J4l|&2BK(N1hc{5x5(A4xA5K=Jo4~=yi+9VLJlK>Ih$cmDcICx4{Fa zO5n99wM{Qu)uUM$Mv}=U#v!;Z(;dBNs@$TSndV~>_L>&-ZxAkxhmFI0qZsuS z+h6#W71Yseq5l9$okXzwylo?&p|3T%mfGstCqpY-XXo6w2l1^vD%->*qMF*)<)Kat zhBNm&jGom{+fPWT&B72wY@MRM}pNX68c*a6D1>Igjk z74#0KG_&D<3(J3Z9k5+J3JF!pk+Cz9p!DOdB@0}rtxq|&{@8g5noB*Q$R8=s^Qqy~ zE@Hi8Z?d`)2Kboe8{8b%Wp0`c%CFe(w406J-MI*J^k6Djd^tNjjrMeyKTV8;0qhlx zROz;jDZ4wI>uNGhzFUnhDcx`squ|JW`eLHL*Pc672G4A04#?xj1mhrjn&{WU`h!gQ zvzb?&zLf>nhA!rXm2~M0xWH$PkJNqMhuXR6%Cy^4R!41GbSbF1QoH<)XH2~p@wxjf z&=~$!60SL|3pmI2jix{6qW=Io+lRuB3<|N?T}8%9q)tFTh@?*qYH;4$Ngce5iz1R8 zLL6t@*F%e1qk`Jlui^`VVdAYyE1~7ZB)j9A_<0=(?NuY0NLJi7NH}5BK9#%Ts8U^C z@JSevH+;x@isWX}EN7F*guD_L3=A?A;~D5UucoCZ8fkeRJmRG${LL_-By~_SPU17` z-lfyp)ipbDb27%8_a5Z_9c!MGOOzer>g|Z-qmp(24|X5J{(Wm#!dhj-n%oI*Wp1({ zVqK*TyU||9qUGG{d~@UZ{70cp9)S#xY6|_VIl*uaSr;8gsqa}jZT-HfVHMS>9$OE) z-A|#fG19e*-wo;xGrBnj9S5y zN>f_=k9P17{ob$t0ABw9<6a=kBF`>(JCF0PZuosPbK2^Y!r|k4ZOX%;7&YR}=F4zo zD}~&Dop#e}>l68YMp4p~erKd;S52vSX55h8bXkT*UI+vDhpEMATgSfSGe)FHxphI&F6DF9{v5kY;G5FmXLN&)rFr?$b(;xf{?EBX136%f0mf^( z_-g)Nfi(w|;48>U^cbvbZAQ-Rird`BAp;~SuO@V%Dzl4yPLHhAqZHzkO>1>@bEi?) ztvR^UmOwv=GAnY<@*Q5o`$B6rj`3vj@W-6vujN`bX52^QzTkwb1Gm00SyaRFtKR0c zvC3}Ti9YJ+^Zx)7OBRhM*{m%~UAlenq==C60QdK-?FQ~ED|tTGroGJ0+qW?3`PZf0 zM+EmzD2Yg8!DSh)Tf{yVThP~fdGfB8l7Qh~{0(P3B<;+Zy)4oz+?03ga9B=RWZ`ML#Thp#o&bx@TVWjpzzm;UOC2>sVRa?r?v;qb0iit*bZ za&cOIH^#b-1&Io<%&dOrw53Ep$2;i0ow3t{Q2hw=qBNax!9%(M5V6!-e9n zbDk0QkKT48xADcb%SU1?UOAyqk=+SCVpMk+>BVnoTD%cnOMWd^`%`Jy8-;l>!TQ&q zczVEUekqn%B!3|!W>+KTZ%!&5PJ8Gt{MDVd#@`~7fDUn*&ZQLPXz0$|J}S}nbyqZh zFG3#_>J2A~uViE@#lUjiOJkNlQ&xN-EY|)f)9>JgnmcKOGX>z|sQTi%Ek8t$#@5ZM zybI^)spj1V$XkfVMFe$Yj1O8{SdEsJBTF{^(71f1sxCaG9na8pHFBt(K328Z!o3*A zrPI3I9*wTw+B^Ymp=4=6k@DfN4@_dYo#gXN9Bk2ZA$C;@yVk2@jw{IA>!y-gA4y6W z#Njc&Y1_M|dsmQX2gT3gojObF+odtw`4U3&#$$NGvqN?zBxo!n`_F@Kx#zE&deB zxsMAa_4p*BpF5t8U!7b?H9vH~3bZaQL3x zTV=KRStIh(_q^>ed-;t>w#+?AZRbfp}d)TkD2GGR2DmKLcF_wt~J|dDqdq zFz8Ps@T)fQv;|h+AGKKDwro|BR+)X1Ms>Ji}j(JYBHWfEJAoj502Q(={>V2+0;_xS z$MmRYzL^z{8+S|+NFPyG3RRl2Y@TB@z_xL?5T*|$Wsriy+yk1d+LgVlhQgya2W%u~ z{6;G0k#Q`+=9_G+I4Hc4{VGPby0njc+Y|x_EW4SjrWjdEOI#g9x|oJjcHbjlHcrTs zX>W2(MkBn6P*qBfxO_1+&?3oq8%VaoHU9Zb9QxE34}EUKYD^cVN@SYqh7L{H*%Iq3 zXkPFwoXO&(nd49lP0kK`de;NvT*c!lX49BSbu6VFFnJBfeluRB;TWfrUP<0S07LR^ z!LADO!rDDv>gCLm%9kaSxA5Sf(z3+DG@4R(JnT%lr*oaK7Z!2c1PEn;_bQQ*yU^EZ zW8=F$Lr9e0S=?K?fUW`DGJPwbV6t6EV~mZ*bK|)^O-f?|7C@K{_l7#u@+rj`!frbq zHkso+Ht$h}#@Qs{bG_FCj@8^je2#$buMoSpxPg4@Y^2~t?v|QOJSC$a-Z0KK+lkP?<&^%)cH-Wq@ZVnl=)m;Ao zz!JRT^4{*|SGc@ql2$x*gOG8A(cR+a6awiF%+okU1a% zI&o0N9o?jZ6BsehOBBXE>#bHN_)q0Yb844X@<$|j3^B;5Rpr!|O&po=HPhZH`(&=c z!tXK=GwYv9u=@6`8px@r&J%81%e!OO=zXgi-@=z|uK7+s(b0pm;PO2yvKlpiL z9AM)jwRMS5IfzcPe4^quyOJaQ>xTJ<=UuOc;+7Mp$8ZUiWE--?tkTLqD&8+H zFnCd%3csd?8m+I&f7Y*WddgKCi&04Jd`6qKo3WBIagp1KahjTVEKtIbP z`PXT$DgN5II6r!@F5ogcsEd{dELV@EeJrDTF>A6sh|bj{nU5sQqxYvHJ4rlqkGg*j zYogFD_JXJEyxO)4Lcn69n|f!(CCJ>nx*Q8zYRvPLRpFH3b)%q(r6(A8SQs%R&vM4 zM$K$kZ3&JH=7}RN8ut}9QH=5Ph+}>5?r~413 ztA>k5Ui_bGmX?A}`nO|zD%e*ot(!p7dR(r%OnqGLQ|*2}w!I1(F^gn3c;$&#eAM3* zZJgHYJGAQm3RlCqxTiYt?LgC%W463$Gcya8@2W{-hM%l|QHENU>-)JsO?L5PT30e; zi-uENKZQ-@uW?0Sj!yFZ!7hPampL22zvnCQ9Pp=M&D*{uIzwgk9T+Pj_n#qGrj1m$ z3Fx9yDQ{;!>MnpW;SfVIv6e3$wS9GFg+T;eiyQeg&I}kgFy1=mdfKIDG}kGq=>Rm> z^Ccacr&7HSej(qP}*K zm-Y53%twfG|CYI)6dh#R^!Lp0nU~eDqnj>Yeb1~C^|fJ};%0|X|754{M~ll(Sxg^v zi7o+N#S<=>$eod_EWuvG>Z4r26w+4F-_DNd&a_I|*5d5Go6T9E-blwly7ZH!e%!ZN z;#n?~iy-uXk^7|BHq!K=mbT2er9StY@||cm9eqMFA%m<+2s_*9$@PoE3T#n!fPEFa zns;AFxvOoG&DtWxr!S}-%O`#1EW}IzfHd0^yGwR#8ZJ0@w8#Q|;q?_Hc}YuV`h10{ z78-R1;^!eptr84F>*!-pe-nN<4x2@PoHYAd^1vU>qx3oAXJVXCII~FlkisQ;dx-5R?N`Uc#Ob&<+p zVtn0prKp<{3|AHN=7RX4$7j{a8*JF>jV;l?>M}boPBcUA(^uNYHG~t%-bs4~oc-9O z@%w#u!<*i+3oS8{MOrl+HN;9WN z8BZPGU=hj~;59*xbfSo@n$m*hGANmu@BT1cmD{ud!65QO4|fRVp|N7$wHZBSnwnB< zDes~aQ~+18zgi=_t$etuS@NP!bF@YDOV+=`i-;wr{7<7XlyOf_dn-QFf6?Bqx-JiI z?5?`2G$Gg&f5WQAP7Lkii)jNT_D9j0stWR!Zwxqd$M90Xn(S^tLSY$f_0XdQg%^5h zOxa45rWeYM_R$RQ0$$jU(@NIs1*P%D0bJx|336ExC*HLQlTWUp^r!`FcnCi!zC~e6 zC`uy$heQ-fb!x9n{nWo6{Pzi(zJw35(pI9{VG>2^pYVI5e;wE(06PxmnQUxHJC2*E z3&_gSNUhZNHk4;l#IRC`bX|jHa^nw5OMG5Dspz%`63&}P~;eXXsRZGrhN_ji1eoLzQa1#!H zfmBN)fS$05{%e;ysF%{T7U1@E#n0x;@YImq&A#|8{sEu-ENU;uMaF*ZfM+H}CGes?%JS^M?2Iw$DQXt;uHPlZ$lvjP?PuuFZ2H=M+ z%9l%Kea2P{>i4z|e?pvbP^%@>$vIR-9UNm{s!h(zJT#oPs#TvDfNlL843nqJPC(gN zB`fFt9H>Jl3mo4Kg7A%$#lWo(!^V91%GZ-;eW@bnNv0YM*=us+V*1pqHr>b0SdEt2_|;_?+H z{mguVH`VahmMiDhDE9fW7kF<|y`jIpZV+5NXaw%SP|7Ei8?H@rQ~L8Lq6>QvVg~WM zRxT?93mYN|MZU?@Mr4xziO;0eAJ5H85Dj-rdX&j=cPAg7b!of6qzw(SqUcLsWKW-6 zv$ki{C)=}Nmo~|bk-H^dNJb6wP_&)3@ffm9_T%c~VmjG)q3iitcaQyZDTn1Qwgk$| zxE*WS%9YwJM7>91RW)Xk0GRin-Nwg(YoV7y0a+|65aUtuup+wYXwwK9oKnV59|_42 zwPHdFIBDlU$!|Z`=SlOUwU1Z)g@j$6VqgeXoLb^^p%ccBeL~gLYY=0Y;>8%Um$UTD z-kmp+tdRi)`1Z;8$8d=rez!3(Vv7##UHYa$v?z)Ci@3sH=^`LKxv#6}IM(T;Pm<{E z8k71Qbwhpc%CEHw*9`AHV%VTRDCwRRX31_EJMzWsZb1$q%I~^>!7Oi%s)cJ2#u;nt zIl8=uudl2t%iU-8)U@YqI4ixIJ4=GqI2NI&EdetYtQ2aTt)M$^?@|ooF7A8Y2_qX{ zjgG-_Plp`(?a#Pz?|`?a$w8EtLx6mDg|*^z?@n~xuf5?_p3Zvk zK`dNU_O|GS6Tq~~k}h=e#L{x>M_L_SsiU5((~Amxl7^ir_ABPc<$n7hSQJPVCnfpS zYRjnP)E)doz=B&Fx11g=u=Q3yyc0Wt=Rf&tbuN)@0i+GAEI3k=BtzLFp#>qn1<%2Q z)g@52-U2WVdG>U6$Ai|fNCX4!J1hB_vBhH>n&=mR=6x<1iTs@?%2;2dC)nF&R)()J z?ian%8B)_`b$w2xeo%aAN7^m1Rr6;|mvwekhph^h@F~xFM`Oi4QW9H5zBl_mr) zNBrhjy?6XX$Z`x6^&iW(bK*3L7PLzjzX)nZc!AggHEM5p@x8*KNIfO@>47rjkjd-EW@Dl&kxsWd$xZf?A8x5?^J-&H(_BT$ zAOB$#aRr(C6g*sy)Y}(;W!G;aoz_ZIYLMi@{uBzD{W%MA*B^3;NB;U&@oQ5U165GH z@oKJDv+-mh>=$-KNSQ5r-l}oV{fV6o1-H;@KI>@qxRoL*mlG;wa3v5v)Gn*lq06P7 z76gi`(%@WdsYJDXZwsfENdP&pln||7%<&coCDYNlt?#Zp5(d!($eV<3_B5o-t=X+i zI&(OQBc*Skp3ngFY?>MDr2`vIop>AdIvN{E{7AgOM{E|hQqeFn<_84q#7CmSnZo-| zZo9_df(qg``~73j)Rz|UPSk%0IL0VAFU+4~Bn9+A7!*LeiA6&!KtTu~`AD+BRUqsH z>jUKr`Wav`llKqdMBugxQ3W|<>J344U0}vW)IpKx;F7MvFY+gT3-d3NI zl|NCccaj9=EMofied=?atwD{%1N1Ha(;w!%*i~*lrJabdTQ-_j?otg9pBYNH zL-j6`lOfYz=i2*YzgK=KY@B3~vxx0Sz>s?x5L=fI9{ZnCP3{$BpC2iofY%K*?6dXO z_7_MT9x1CN*I(pMrKOUSkvxO@;J2m61rpgIBBF3`+c=2}U6C7)58)-^g)ZQouy%gD z9r5&M0spBpFE9%s+AYtN`vo}`oy;Y>D~X-4XMup9ss`N06yzxgk|kw zhYIUP+byMHfji#)nWC`V=2Dx6nQSFafrtY_aQr0jT$bW{|8pL>;LaE-Goc{HFXt?LT?5E@|*@3 z4Syd1|F4S-&-6fz{wLE9EphGQr5eYShS24+>)3bEItRa>_*3h>t%JY?COK+ybewfg za;Kbd-T@!p+A`oDVAu)u0F`Ali2ZFlR>EU;>$yHpOjZb(WChy+V-u0>7xBKh!-?1j zD~1C&@8`;bk3~@TL(j|HhuII5mxCr#&$i+X>wqJa23+ddo_8*rSv0AlJ82fa6ax2E ze89fQdvd~l*x}d)tW5&iZi~=R3kb~yZL{3|=VVdybi$fks@lAQEODZ(04;j&M+D)S zXe-zCvUTyj8Ng4nFoWJpVSh8cZ;8)=J;F`4GP83-m6C*YRD4t3ee8wLpGB3Afh9)S!hKpDd(fiZCy)TVuW9VXQ-8SQ{>seCa~tdJt|Th0QXXZ2+bTgj ziOCPPRW(emT|QbjZa@?O8YQx$$P3rK#8xyQ`h;-v z>l`%Qo&$S@Zz4sL>zjCA8b`0shq%1`M&>{3VR5g-74``A>ZT0!C4LQALJ$>W47an> z$bvlU9Lk?1MOyd&j>INd8j?Nb94PG0lxRoViprC2zEmCL8IpOv90I{%t_#5H>`6Zo zn`CoKUkk0cE}5~Eq;@!t0~6lsh&15RBU*xBs_Lqm5$1L>hBdOD9DPr^ox6x$Yx1CH z=IZ+^yEl~Qv4$7_5L7YX_}xYa-2Fwak|;ua05|nCU*$#0?w8KXWKJI_z3I!kdEQZY zcfr#T>9DJ1T;8bsouuN-<{A+~*enyNnWM9#;Y2FRuagLq} zVE;LUOSU~2W_5O5$$DO%KAT#?S3>zcXvU_vDsuO`!E%k|TbhAU0nrw1av0Rk(AHZP zp2mM!$((+4e1Q@2M+4ERTg2TsudLY!_vUZH=eIxugvh%|I0Y!j_azik7 z*ViZ-WAl;~qtE-!vy!FV2PIoHZS=+IfFa2plTrn@1r;`7V-dDZ5WGWA3)#5n5{~G4 zK8%giHRu>k6ij3EHjkC2YeIjqC&$*`^^)P7iBw&+`RIO{iLse|?Yws`wcc>aM76EH z(kH^R6T6J&gB0Td{1lvJdLR^M{sA$m2JZ_^?C39pC(Vs^t^suDu-yprR^}e%WHkFH zjf;0BaVVT+-Z!uf-Qwm=3#Vv`%*_(36Va7E+m`DV-}X>5F6vGqQ51XP-4_2PSR*o4 z3nyZ`{K3PQG+WllG9_xYo}C|}PcwzS8n!*^uL)6yeGTHL5J8vXSt3f>tXeCL{}6&* ztsJsW)wwrxqe&3EZSh zl2>Gh^CoZV%V(BGyOwM>zuj6$KqF-N0M>`oE9%G5Z`Hc|=|hXQ4vx@W7l2B`QO$2B ztz5j=gWBfCyiU?+8W7A15fYbB+MoUq&ZeeZL7~h1D7@P4IQ^*Uhj~HD{%K!W?n>_D zraD4N9(8Q;L~fY5wbLgj{&3hL(!@Vi>$NmAOxSxjRG=e^(^QI<+@>c@sqVBXsa1Rq z_CCeZ7BPA=v{2%|(vBSul`PH7Z5>yMEVR z0r{vsyk>J=jX*sko-z*)a0b(z8w}E%KQHYF?%*?vbK~yHUwtcZq~mcZHY>D;Qt(%J ze!qJ7)zUk+uK<{<5k>ukhye^s7RiFV<^~o@|d)?k4ep`|dS4bH8 z+mL8Xz>mdacW>3luk(3satUo~RBt)Ey|W*>g6q)5kw!lw{nkiT1_`SDKn(oWYx=vV8`8z1B4G6*U^;N8<;j>@<$^IQsqtSJyD?@vm)L(W+m30!k__2jrj3nyi9KJS2O`jjWw!%i1Sq=1``B{&9 zuuIT~x^g>$C0k{bW!=+MkkK_Mi$*sbF?D3jAeLBM1rR!WkUMAc1QDJdGD=SIEZ=Wi zUL-1kh?P27L?cE&g(^sBGQt%)!aiJb2HfmR8-|H$tSXQ9Y`yk8gRJ0`ye1HjmIC`* z6)fe?X$W9?6{{!M!7K5hy(in()%FTAr<0pnoggcVB~~eB>GjwS zk$-Raxz){PmNtK}rKZ*T#WV0mf%58mo_+V~j(G<0pJ`<^?`Mzyks_Lp~&KMP+c%VelU$<=t ziP>XIKym;89)dzIXLtx22^sFxt0{^KNIuj}6Aez$_d2|`_y)T;F_MxelTS$y825)< zx>r!Aye+Egp4v?fbo__#8itz6Jh4~CLd}7`IAjZzOW2|(`-a`_;ccIhBCgOs!9mqB zdw~(o^kxY0OzYXROLuKCy* z!WJ+_m}Np*$v}dEThSDAEO^l_TXhz(qiwTW8vd<0T#vho2|{I@upRY`42K;WZHizm zg<5|~d0B>Uvt^yX=avH`(`s++@hQ(TXO%km=c0a^&>lDJGpvh6?1>8oOQ!qu54 za);6XuBl5uTxY9OHyOqx_(5hxY+ z%n{W8zKF2B70kTN_@fo#t_2k{YUe%yEoz4?YzIk}m?IQe2J|)7yerPeH4jL3K)E(s zY>DKcJgy=DowK&epgD0H28FoV)AC=g2us!o*u;^YSWKtaou{M@si&bfxQUhtjB>Sz zrfFuQt2;f{D4q!|iBrmAl=IC(KCmS~-k@GVhdL>B|I&DT|1c`|X147wX_`gamieoa zrA$v=`#pYo!7ih+Y*~dy`}x^IcFtJ7@{@WF02eyI;jajz zdT!j>*Ci6Zgu?w~ZU6cHgQB2=~l9m2PMF+7zNeVi;7$IzI)%UML|**(;wGZt0UOx z+;5{5N7>&fAuH`MyJ1+6L5>0f3s>#@XkB$}QPo=Xka`Y0*j87ExnnnAV1v>Dnquo1!^%XgZ2{Q2agYwlf&2rR?vR*n$F}i@ zIPo7v+7728me){G4*k{;Ss_>ShEfc6VB2LBOfuABvp2iP>dhIK8Q)LwEv-Oich}@2 z7m9#}1etZctF|Lcb|DHApx!VWZsF)HPtlY)*xWU!fn>WmMq%NX?UHxT)^b~yOB$owl)2B z?44!0{B8e90@^hcV9)bYE@+p?xZb(B4v>p8?->)PUnsI9^>-uw!LKh|8u2F^fA|p} zL86-0EI^UG_Pxli)O)_0P;W4*WU0hk_Omm-ETh050o1k11EaSE8`Yd#6~AejL+B)5 zeG~>{{-^k5H+{3@glYN>Z*D5~^ry?6SiP1MN#8Y_Qi^N^gY`$--v%gV99l@>Gvhg; zIuBn#Wj6BbzI}8}#S8={Rb{+b4PU{Bw7e;XYXc=x_zvx#Y2ca4LK{hRNAikc_fPb1x)f&b0@;HT5Pp0NAu4$IsF6g zk+9;CJak*HkxM019x40g%owxb+sf(lj|9;!dGud?GqLzf28xC4>stPit}XlmbpAMc z9SH6G^cO^&9_Wf*fbp7{<@}MkXn7uCi!9}qu%pvZulL*rzwHS1gPW7uZ^)MISsXaa z+`#bg4Sj!60L>}ec;c~r^F`<}Vm*eh1<2(AeJIWiP*L3`;ETwO*Uq3p6lM>~I)atS z+z2LAcWJyCl2YM#Cr85H+F+1|x5x0Iquy@~rh_FIrb~`?RJG^mrXQB(i+r4lYt)kg zKMWgiKBt%-I-zz1sn`44_6O#O&`D?jYBu#8<7rBAkm9qnrhn1VJ~n zo6WY4)%6FKA*Z2CoMTSo*o#JL2)K%6L(L#(i@nV(j1M)x8OFwz@;=#!uT> zL#+>hR(E~TQqr_jy@Bm^ink#eD!C7Wjxcxm#9PE`q;Zm!$(vSx#V_@ZH70c}DE8qF ziTeoU%+1CAk!o)CS{35T^)q$k()Gem(>hvA$WgLxh5pqyk2{!)gGMBjDG2bR< zRo-tNbx#^B6=Z@~*jieU;v@tI7@GQDSja4`>4>I4-e-<65H5-IoBRb#Qu)N!uql@w)QdzZ%wD|`iq_JnBK*K`bHRQ7id^Tb+)ZeL^w##ec3hrhLWOFX1!YkT-f@%Bm?%Vj8V!sXRFgt6_c#FTU4k6ejf%ENg*um7qqB ztzw~)+=`1^G%t=|*@@wN>%7GAri}Cb;w`}y(NyQOwdlVbmWRccOP!jz z({#h^3h}T2P_dH4z2U;X1F*cgGqsFol_EEUpt@4&Jh=`z`_oWm$iAOnhkl5y4mpPM zqK-aNNJb_Zf;9K~6L6^5h|QLfAd;m*H`n;5?`-u0gO_YAFr|>Ja1cd8KN7$KYRqO2_9WT zxtIG=D41sFh8{0P$-a?QC(9Mt!zs`A-V~b{+i&u_f5RTaWg?>JxWR{vUX109&?@o{ zLncJ>Y+2EWSpC2^W#jc=bVHdRv#ggg-;;&;xiCHcbWt}u-^h?tGF>(<=fm0nhJab~f)w%NS z!xi?+)rWDvz2z*J%91kA!S8f73a6)=9P&W;~aCYY3YKNy?6w z2nx|)9A_eYM9iR=zPhzB4u;gjbw=ZMs49I|TE;nFZcuWlGA>R6^$kStZCNFu*VpeqrLbwtsrpv1GemE$N_Wg!w5ytb`XjKHJ~?i5 zQ(RadBdwP=oHDU(TIq>8+YC%@4J&U;8r$_KFhUu<_(*7`x%Ns>+AX?M((yyvd)$el zw+MO?gM0swae=*>@$|iv9!u*!2x%lsw|NOgUM^hr|w)XFhVl5d@iz~VXh&~62m)X3GB;G`w6a-=7uRKosqj-c`F z1yX}eDkDz2-(5*zrc6dezi7C+L7#&9_=UBkuJ`eFy%hV-a;q#+IWXene4H6-PLW+r z$h3%Gti}4feWj$RdCNyYx!ybyFWjc+bx z8rfRC8(5mPqnDt9u!kejV(X9QCnW?DAD9&ksb~a>^Z)kSM4MB;R31{8lqaHfoxaMM zESR~Ls~bXYY0RRRPaP6i&$I5FOHb6CP)RFapo-&9FWqq;-@|*FLm9X$8skPq$c7Xp zKCB0M6nEOx+RY@tf$}$-ud?9MtX8&GeEf)AVNbtk@FUmPqSZ!yJK*cHg($PH zl}G)uYn?UGk?wHfPYqnTD1NE=n@V|w74)9P{+Zf#-TInb^_@@w*D0bs`nu7g9b3!; zR);H`x~ik7OVl0RMd^a}`g=mvr$$5-|80*V?uRtKPY)Z$BLNb~BLwZV~(xE_(PtYS6WOSK!9!OZwj**Ghz+DfM0tDdW+y66a7(g7Tc!$ zq zo)O9TvGc}l)y*{yc2$;*{KcS)Zo8&!3xVu8UnhwIudV@8lABeA=Zk1V;Pa9N%`LTt`Cbb2++=#Xh7eLs5fIxQ!L9Z`FZQK`i`4+gAKcJUV5 z`aA*>)x|Co#psKoFu}4Vdzn?Es7PY<9M3FeXtW%5hH(Py z?nZQV@;eU@3`c=6h0NuLcrIb;l6CUs2}fR}BntFs_q0^Dxg!16A|LwI`MN?2+Mj=} zdz}y+-@NV@y>!p3V(m|X7JTJT-yVlk-K0(t|7ag-J{uZ%1AM?VgGVL1k2aelL|s*I$s>%E}mOjzSxR7wa`d5i_C9 z6y1Z9N29Bayep4ftQ=h6UnY$a(n2SAgD>d_cooG7FUjbVMPZsps2>w@JqPvpn8li` zGF(+My_W%dZTOxT|Gr=v4#IWg2SZg`=b*lr-X8~>tdZOMu>7&GczlH62`H^^L<+&9HeAdwG z=pke;!2@>@!FKKHvzuxT>G72tT$wQHyl+bH7NepqwuiTn&g6vRK>>Uz(MTdr1^*D* z2#c!AnWWuW$cZ2~%q%0zojlpDSFeOCxTDNuw~Y7jyO*B%fMk%??CB|lVR4r; zrgViRgLPx^dXF+_s6@AI@w@pV$jrJ})jsI!sfCSRLzP(*@`#_ET?;`OBE%rlHpo~< z39Nr{{(yumTaC9__Dv39Zhot#MPSrmps*!x&!stOpT5~C$9?vhBTw(3gD%B@Dc8sH z`-WCjrKrFmi?>!yU0btfQ$v_BDtqv4rRcrB>@zKhk`L^9bf)QZWH{|0qUecSKLsP` z5temgNhx5`u)|H|R&-zfGZU?EcoALiM>h9^ZF`FM&B;lnK39>`k((A0%qi*r84FW`lLHXf?p4Rv(MCR@yign0#$28|0e&u>u2z@83eZRlX z%Kb;!6>mH^*DhztTgR(`TaG}p{@aF&D%lI8#*Ud;%a`&+FQ}tS(Ti3S&5Wx01IaB^ zH}9VPzK}1&`rw*(XHmR+=&m_tOFfG^T30Y4$lP^__mE^9W4-Jd0d>KoXneZSi;2TC zhjV}-6=xJZ<`Zu!ZT{PKznKFFsZyMqv}Ns1m2F;k0Zu=foey_wvUdTU}r6J|i$qQMi{oX;EyVJCKVxGhHUx({6HQ zK^cFEeA383p@Qgc@KMKAMVNfXrMgG)C0EGP9vrV?lRXg;7ltVYl8UqcP$)ds@Ao*T zZdD|_lm^|TpR>`OfvuUCl$7s{Lx-w4mn&31`RVjuuvkt;wxjwy?2i>P=0a*;Vf@t@ z^dkT4AiHC1zkyAhgJYMtl!5<^G4a(WH*f3T*Zq_Qe*U@rp=2mJL{xY98dx+c^((jj zq&AJ24gJ9bcNK;@xcFP%7y1LY&92F@>0)ffjTb8$U9W87o3})~%FB2gR!9}@O_Ca9 zd@4(w>e%QXCFoW&Y=yrDPyQ`;J?JmQerpqKYRhMGw*mfJSEQJd@V$lQ&A`$8!?*qV z+m<^PMyiiirv6BOC_`oaaniT>g0>C#_SVFQ=cd8uF*Z5ZZ)yMv1p|(Ec8o<91-4q) zf;Uy(>(p(Q_(v!1$cLaw?avfw(3SxVii$Q zmNnUM54EnaL`x#?OTuSni@byHrt=0CQ-k>0%zOzkVMKmf78)8*3dtiT(`M*lOj4jb15;voS(@e8xP zGGAX;7L;{8xO0_LhR}NsFY~nAf&?e(VPY@n!*~84r(=}hNAzz3(}?NUdo}g(L+y1+ zuoA{MJURc((}K^wYL(jfCUZ3pIvFn2j!LLcQHIJzWz~bJU>YR zBqzGze$H_kFd^uAD2e)#v0JGo?fnnozj6X_+OM+nY~p*198c?-w?D*dte;^yz9H>) z_impS|3+O^x6bL^VyHgcM#As$_D|S0124Kr_)CjX!qs^#6a+0xrZwH>i%K0{RxgFu zzZ92O4+i9GHv6uBX*6cs?z`53DMd}P;;>j*df(ik zQ3$+Bz0wR5Ah#c!>zyO>Y`b>s4WBq-Y-MF%{|FXR8~t`rBhuex@`*)%^%m{+;cW)m zLp_5xfTcF41=?^^=V@J=Oor78G$k#iF)62}n_W zAopVy;7vNavW!s`wcqdlb%0utW(#abf}R=SU6bPb#Q%N3q99PLsZ&2(X60UbHXMId zsz_wk>?X>s!QCfh`t+qE4F30jCuS0_@`!yz(F35b{}kL972>_attT?s2o%e4uhSnO z@7hRwDGoIa0A*-3TtK^f?P(0S=;tm6G%uqrSymkDhA;#%QyN?(w@vF@Lj|6h*m=rW zN|X|p%9_^`4U-Q`6L;lqTztS~ledn`51;IVn^*P8swiO+d9tPa}MtvXcC6!54Qr=^xxbl29{cVyITVNiOxKr(0zPi%> z*GS=YbQ5)tZsyHC%lP#flATaNoyO`iv|I_vL3?OxO_XJRbJF31TDAUXh!C=&R;M`t zkYr1iuKS*ZC7W&V%N-wAR9H77>Z^|;hz_j3^#du#LzemYdQ5xrHApcm{voi2vLXCk z4rDg^@)?YD_%@0-Xo!Id(Ek1IN0IYhVF+=Bu{+_|{qaImqEOG^pmX73sc3;O`7&eo zHN~$z=m%a;i>-$*D+s0%W9n+X_H@&~+46kJGLh92P4r@seGDz79a|ilU2nGL4NRgVL4Eg`^5OG?FBk~se5y|;cmg>Hs-*ttmG_@Cac!2y(?(;=7iB6P_N!pTh|Nt`^wW~(ZcE@u zk?D?l9DjYS<-hin0~L0Kso3Dd)V1h4Qr?F+sg6&AZQ=`;SL#0dGLAHgx^RCoGVTk- z->DAJZ^=+6u>V!Xkn?fq@_I}^VQw?!x7$_aGtFqWMgoRn_f}QecB<*FY>~RLBX&=_ zih)LBRDHf5w0&@KVVZwC3)zK1 zy8a=g0^%zjfXM!LK<;#bv_kBLN=Sj_?NE>8>8XI&=uq31;DO;;B%GE%EI`xZS8pNj z^j-X!?A5EUfOGjHJg$k5n-de`NI6A~^R{RgIkkLD zQ|ruely>~!4!SQ$1h~L=XRm$QW@~L(?nHx=EWm7K7;;@O9R9`Q8kD&lU^ek9%Z$KV zgFslq(9O1ylBllLitc)7QdZk2KXx{i3+Dh21dI#*9!y-k^;p{2ZJ6md7OS^{v+rNb z4|vVievd6ZHO!ffv04Vn{+=*4z=o+}E0Mn7p6L0NnUZ7%*iK*^KjE+bLpXhI=6Q%S zqDYnY1^NhL4|_}5l)o3+qo|VZ2anTd7xoGv(lLcrsTkzBJq^-ct200Jbbd*a~)-Q(1?D-(Ioi9-Qw3kNI=PN&y zdf;12aaLUHZhFuwlKv9?<^40Z3P~HoJw<9m(Xt@j{CI9L*seRL=^AX;`vS{b(mWND z_u`lU`isT!e+aJo@VW4(SlPz^Nn0CJjrN}m5p*!z<42kyuEWdFAC zLZ7{4_Lr5Yb?+RvXBjWtQngr_0A<-fU&zddl7N8*tp;k9HovB zCdO!va31lW8fj+nB)Z3R*gtmzb{G2|pzOxzXN~Y2w4CQR_>|8K9rUPh>Hp+Oolqiw z0_>ixD*`FCj{V3IU}z}vIEVtsF|`bn%!FL`$rG)Iw<&6MV;OUk`sSo=-##l{K&v0;q+ON3L?<Gsu|V^y0L_z2&_C0gHQ|k{z^1Y;+E9*>ozmWFa%WMb5sH6if#ID^XAo>>uY_}9 z_stA^$YA=J=#Mku21?db-hA z)@;g#y7W`Lcv{xcNteJ{MaMiNiC@NxhRZ#)$w}c{mh7Ucc8byT^yfGA<%`j6b_pq0 zjaI5vP{CBbqoBXF)uK1itp>W~a}&}#ztCYow3Wg^023L~CRScVEoX!4&epxpZnKfX zXxH)eP4l!4T{I^tGQ;p4{`9?yu-OIrNKcfLSpU`+fh8F4`{a$zpb*Hkb01VyYm)6l z`{c}BFOM@zalaPA%`L;LG9S&}lpvO=*XLNNOg-oC2PxGLS?Vv8?Ah}#`jKQVkgd$( zBv9J;dyGli&nZepq&9Y)t5aT%r5J|B)B-^dARe0GV>{t3?YXf+eQ^5+U<%CSM5MbFE4Tz#4n!)dS=gVAa25JA;8YqghlKhnb8B$& zwGF%L_$20oHLwCq!GUZ0jtXMrG;5m_uxj8WmPII*o4|p0aHWXJNt$|9h{O7g`n~J^ z8*znH0zz5M+`T8`wUE2c0^hOTqc83EjT=u|*>adw$5$3lLkWJMy!~eQY~w@i6K4U53{~2iHU|@1y;E%mLQEsT#ROzk%Y!YXBZ82My-ynCLY7prqDeyXm z``8}+Jej#PwIRbB4#ayXGU-?T zR+-2e94`jw{3Te%27pMMP?|jOxWd(?{rV9Wmf$)cLcRf4LF{71+dz{7i;om?x33ItGz~p(cLM-EOoCv`(({FlA zu&*+#7EA}1(zR3d$q~69>zd8Ro~!%e$WHfDW$@S%Mik&xnb&5Uut;5j!Smhpj!0we z7w={#QRlYxxLC+P1gT2r3xA@`DJ$&(Y4<;Sxsa!Tnz|#?-x_(X_BSFe6YAxY^H4i) z?)NXpplF{<`0i34$egiDdCN=2+l6IzGcAu=)jl+{1m z{}5_8m#1T+`PE~LR1g%YUX~rC{`xynZ00s3|8@lepmCbb z`j))L z^j>C6v97_gZU*aRiHfV4A@-LjvaisRyrV0w$q;RgNN_94*dXdBFv(3U%JP1~f(Qw3 z(nYzcZ=xLHYT&%5*i2!l)TU50KU|8>;{WOED}&mK+O45bD8&lJwM7brLUE@+DGtSo z6n6{3-9ph4+=~XcKvLXGad!{yF2TJw_xG_t|SdYdvc{rT;@a zEUI&y6r6zg^iQTzb?5#=D}s#8JR%&L5)bs+jToW%yneY-v42E#FgDsEc)T3dM64io zu};u5rqn#^B5a>C4%eRLx3jn$GsPlZs&^$Z-4CIfoeIz@M$O8T^xmik@!V%0`B}dN zmpf-2-M-wpFpi}9)VThlrh#MUr=l}Wm7Doq%=GyJfj1LG!$n2O+?5kjvS9Q@J^_bJ zD%M=>bf5=pJ!k5pS$D3sr=w7^Cr}@sSw7Al+ogpv$A_@a&GXRUxf$z;DOTR7jr8p-VNDOK_lrB9-YIPEQehm*bw7aEFU06-`@EXh&2|N2i|4*n zFO|@lw=G3&>EnmSf&gr{`+p0USn!i0%s6SPbO0EZ6w`{Vck#rBmuz^=939!~J${!t|y; zbhPvWs{28Zh241DxV7=v^R;!akL8qIl@xOZYkW8c)Pml4*Y4u!ZPr}jFKVf^z5@fR zWYQZdF@i%fTH*QN`p;78H>=>W3E0M~CuGnOMLV3DT%pW)YX%YU;qo_y*SC-|1*0W` z^hv`^Au!U*aRZ~!j;z0LQ-{>MOux1I#fW|mX)(s)W42h|-y_N$MB`goppjzt`tN%`7hiD^@ERqW7Q;)gWx)ZW*k>>;+XSy{0c-@zDKbYPOMAO!xc)>k%xQjV{i+ETOXxz$W|vQEvtn8c9}6(>2mZX{ddQ8n2iK(2cEwjq zv!EmUFi!z|S+#5mET-vStMLoF$m*`xuRw45`6}UT_l=_!Y7_2PEmf!t8087?D*l#p zmlqsoIgB}fPjuFO7xXhnVg|rEtkC$;2`KKtmwNYxP_tOktA0Qyy+O?TIpQy_01d>P zQgOQ}BV|Rlc0BsX*^&g?e5LJZ&~XUz(7n$6ekR+zo_Em#FxJ4m~5|exJxF>UHb-9V(`BdWmY7 z4L0mAMdK8t=M7lg1bxu3BlONobC2coHvRNvCBerzW;)fe)Q)IR(=GyQl)0rxHqL=^ z!;9J(GSi&GMpB*yFY3X?*C=xVd+6WNJ{1>t*>i4quf=>{4~6Mw6PpAF`1d~o*!0=y zZ{$qsQ#C_P6ck>=*;ETT3Ntx!jj)^5Qt@ENz08F7a*YK~Ba`{37urRq9KV0XGXzP$ zMwGoCXXb2ei2(E^YSc~A^|U)UW)Zz6s(LGWP<9$uhnk5`U16Xa|M*kxo)O`@+4rR~ zMa+dkhKmDbBtOB$9EPePD!*@jnB=r&+b_at>^Ck8g*jDL)RZ(~NLC5NlH@J+Wx2N* zagTYTdHyC({hQ#qneYghx$!Fs3yOElZ=2~_w0)Dc7UeLQW_csE8=W&PeZ^xl#iWX) zv14IwM?4&rkvZ`_`hj1CMvDM-kTuTYQiC+uCA*TrrF?K7h-SS(MM;y(t4u`?gCv1t zP_fZ|XIp64LuKQHbYGytDPC{BL&xQJR3)gGZ+}?In_4|zWKyjNHatwfCId_JU;-L> zS}D_3vSPit|125 z(RHZvQWqJ2WVotcUF>5pfZ%2MsvdAxZ*~>a?w+B=$ur!$WkBCec*|i#{twLwBy#Nb zbN1e0_~Ey_J=U=hjCwu%Aq9^653NBW4d!R~2mbN2ZK_Q03c{y}OmIL_|3jPQf6|I2 z5u5pk7D@i;4d|MzM4HT3yHVmkl>RhdO!M)gRq0Nd8OTX6w)FE;rj9hnbdnEE-Pzih zTnu?BL)D>yp6x7e^2j3cRujn2?#%&XkZz2_Ka|6ELgmPwx{A1D^XwT71KXtQC`PTr zqxqW~Ti22oawI>sjbtmwD0wR-vlJu{?kVM<@##wLjh+{6(Pt`iEp@s`h}>q%-5+^A z?%Y<76+W=~kR`97>)z(JVr@uqk&T{I32nHuAS_&-abZH%r6-CX_ZaOXdX3n@o{`rO zRB~S5^!G>-n_%|l?VW8Vlv!&^oHd5HEg($8#nY-^7daYbe8#&=FNMnIn639}AF?eR zK>ws0|8rD5bl5AlhyIrGf-ncQt`6oOyd~xj`L5CY2KGi9)Qt-*R?a>a@7-n7vD_S< zgGlI!+Zp?9U3cb`8qhNViV_YM3gpPLj1F(XA3{Aud^%UCG>d@bppDXjNj7bWQXNMBT^2YQz(&>;4qGnhES*K`v+ip;_{e8+<%jyq^^> z?k7QnCRNH;y-f|%jPv53ROV&(2f$K#GmKsTeiF2mT`{PcB66xHIy4j4L9IC>MW2n# z*$H~e3rO7!?~#0C!h{qHT^7FIPK>KzrDsIQccREvw%$?xx0~vJOKqtP?g&?B;8*|9 zupc89cy#;BQRp3K6C@Tux<~D`XDseVIfIq8fmP+vrBexPm*{DUo#emJM|J(Hq_j3*`6^^=p>R`K`@L@H~;>~FVHon zr=fWTm-#8DKBkY>Nge|xL3Q|v8@Wcy0)Gf7R?UTB8zHCC{pTky`tr}s|0Y;dwqIb} zjmKUcX?!!R@#M=Nv>r0f=^-;=$vHH=_^5cX~hE*Oz{H`EAn~8DcG5*yW2k>WJa3zm8 zoL7Cc;#fV~jnpeYb*^=a#MuYy>oq=CN)A0eT)^$XaP1^-+T_~%`SGtcrFD7NB9(b& zT1gaVt?0^xgZv<1@580){d&U^b?@xwTcEppjQpkvl&3@^e1(R{FyG*_!NW#{wI098 zL}}UXYp6)9PUX_gsMeqHmQS-DIBA%Du&3`J4xHNzn}Z+U9+=maqJ6cdlZ}t4?3wXv z`$BpXzJi9dgD&*8~RT>})QZvs1Oy=Eg04jPZ6q0 zF-#M+P7?KyznH0ERTIEYD;NYbtrpa7vVgm1@`}D2!*@V^N~8ZEuUz_c&vyaQH`c5o zM_t(L+tR&LRF{P^>Va6FlFCW>;$slqm6hk5sRJx;P7J^smnWpQ9_0bu(9ev*ggDCa z`XC}dDbF>>mv#A;g%$TWT67Erq1Y0FbTNbdxArP-U@;!ah0w(?;3izklZNReJ+E~E*3Z}Tu}^gC$&LC^*6R+d1@!$rS8TBrad z$ZVo*0omIEXUw!E+q~|yR0Kr7oL-KmdE^OL+J2RohDhtoAi8YhWMRv48I3PLF7uBM z2eeJc0ABB}m?#OUY@glf&sssxIm@XnddXLK!o$b@r@C!O$mYw;Q=63&01lqI1?XHoYPR=ty=;T(K& zVx(k|?gEu@cYBf^2$G8qFlpIAo2~Mq?$q;0cK0u}xjt|&GU_O$7##|si37;bZNLht zT}7Kc+q7-H0N!hlYV$~~18m7~y>hkR*!O$% z9Rq&dph_If&<+&aL%3~gespN|;dL_9bhStT8c1_}@ei~DLG zedK@Zi|0vDC3}nilkz2UmPsd9j_@nTSH&~K06BgIh5EtDJt}&tFVY9(!$PcbG153Ip897&OKqx@} zuZC!eTvTDR$XxLuq8wc_;zx!*$@*jLN?eFsL8y}D#+fZS89h6!Pq@{v zKq8O4?@%&*e+kv8Q=~+%b-Gt%=c%3Mum`+q?`6p=PRuhpC%LU*)nYF&=GfxA2jKo%4)tU^&5%Z`a%*GT`8r~bF{=?$g`G1f z*)_2)%92+WT|Wtj{dd2FG@@wSBDqDz7HC-|oc0sTeK{w7F~FVt%6&qwbr50Li8ex; zCyH~IJzXp<)V-hq&{xn~89EzH>W{jD%^dWQllfp}71+jND4NGRuFAatI(9L3IO0vT=}9yEwGoHMOizvhi-Arcl3WWpBqIo#I-b z*$!j;73#FhXl=D8U3E;R%~BHzfq-y|YZ%X({n1c@0O|MOB%072#%IlKVNFHV&R#Qj z{^c6GM{j&HO3^in9iG8USG7Ld8R9c!9gc6)y-OTKZWmzcha*<%@3o*godrt1jCR+8 z+@BM6EErGjUn3wJ3*SwUSB0cW_R=C%W%%7`Nv~s#T-Cf>wmVguK341vyR!|S$sT?S zyGFI*OMg{1RCPW*tV5B$R;cUUAon{)15^yACClCR-7IzPV#Z*r2&hWDtR+oX*QBKV z$S|bsVC6u6TuSTE+AH`gRdM)9dFC_b(9{cf>~?g;P6{@K&0@a4RJVU#u~R$gS!!1n z?>&95PZYYOg7JuPzBbZhY(zA(UbT>uSGkYxFz|i#=i_Amcd2Cv<1bEEXXU}C{Fhew za4E;t*%@gJ^7)79Pc0TY{Z|6yO_?3kxZhz+;`yHeh)WfWuC~)+pd0*s_A&qSb(*Y< zh=ofG(oxOiesbzLBs(;`xjB7XLph<*gJmjnVxiN$)v7+mp!C)dTUx;xC|5z4;a7|yh#pq{!X$~}6-lW?W1;kbUHDDrU`0%MH? zhpv~qk3+;rnjx}YAQeLv$*M}JE^fA`WdEM$s)1;9$*(c8Yr8H*2-w%>BcF(}Xohg2 zjB-r~c(?#vA8Pg-wFq+|p)v7* zP1%$&TBb8wnvTer+s&MImfI@!HG2w>(4eeM7{OP+uYdsfrsJTktMOgqqMqGB(L
q$qu{7JQN182lt~iu|+T5jC7auyAoy@Mw+^`%-|Cs=!cPl4LpzZlc@JA7N z&gWwEM#(qY{mlnuIDkR!jXjWA_w|X9yJ_Bv_eAUm45D^CXGo!s;h|bAxgNI4@TS0! z0M}R7%k|o}suar)@vKannK~1b-R0shIu;6Vt8v&z(|6)l_Q>N~KA2O+*98`f%@)df zd;t#$K1Vy(>s5%>@|*jXEUe8?rr>}H!{WlpU7yx1IXfh)cGyuxYRXT*+#>xI&pM9; zk@ z17grjUXBvUtPOMeOKY5Mgkm3`E;f5m6f|RII2>+%`T%%R3DM)PXjh7mPXSg$mxy3o zN(|KbVOuhXhhFTx^YDfKZ0W`Phvt=-ctd0)U5E>(y&39vy8+Ab8gTE?NY`J$jYKkx zy23`j)!aro6Z33gqs0heW7U+BlBB9vlz&GWra($`mGDlx94LS~I4P#DycU%ppyO?*0-(bOxd_Pg=o(8ns|>-VX)7@Xl*T{hBf&O z{`AC!9DWPGS=gS2w56wu68@#reGJR1uHZoTJ(CcqQ9*V5v{x>b;&GA{KvzslF_N8%wA6 zE^*fm;Q9Oan8`?iN|Q>wkaYh=w|Y#^5+ClMz|`CM_tb^hAtitMi+k6)YW%&SZ8C)ukwzt`UcfT`!f}uz>A9KxR|*~$PC8o$Ck7uN z^C!g!vL-r%@7#SP+5Uja8&4+Un;lh)L|@$RpEKR*Kk~WSHt1f^6I^YnN1 zEXXF8_5=mtOd;hz6cAX>Jx;##9jzZ_sJ;kR8ZYd{TDrUA$PFC!J@;OiOkC8_>qOuW zYEMu9(SLxS3un0%4_us!8j`I%v7{uBn`!kQ?6Zg8K>i0F0$RuB1S%Z91qr%iBU;Ya zm$q8NQJ%5{hK7ojQY2DY3wcqq5b$;~C9Dn02s8927ZT9kwc_&1p~ zK?uCEhxwHBwW&f}N+tHW1RaETeXq*>mGo|0%a${s2N}2#eHcU5_N#8Cz!sHN_Rm4f zcf-#zgJ_UO?+E#_@R(SeUOTYY8@C)l!=tvhhH6C?Mk(}c$az%!`8VU(O3tprB)CTR z47O%tAH-eFOW;>sXZ}#fWYR7vzbcV4iBVG0>&OSf(i8oq&m$%`OO3HYd*8!$)&PNq zH52~xP0Tib+;eS<}gFvK~vU zJ=_Un`*a;=tPlOb?~*-Eyx>w?v9ym=tUI}*63HfzpUdQ?!Rs$xT)V1BylQ!;z+zC> ze{fE<(`5>$R?p>oK+)x&*F+@(Z1zfE)=y;QAt)yjVZbfqH2;5oq_5qDu|N3_9$PIV zAm@xhH=QUO#7{pi=*X;61sYUv~|AD$RKCL4VN8h~RVoc&ndYgWeF7$zHNT>$_bH!rf`N3bj zSz(j*XSdGPWLoqFI}6}7F=JPFjb|i#>g4=SR|>R`$_}*6Iso1bfO$l!hAmF0#e2HX z5DjO;t?0VU0NU4n2d~onM)?;ohR}{Svv1&h30jGoo$a+@XpC*Go-B&NI;ZZ1p+_ro znELYGTqJ@LUha%osSJ7yh1%|hnHddJpP-Hn!eAIwy@Du^1k)Pgu5s9r!Iv1wq?)i;nZt*=K;)n`wP+uGa}j*{>7m>M3CC`-y> z`t{B-GLmeN(&Dq|x$nEGEM-IU=5DN&^3I(@1R z-5RZ-3eC-%i?j)ni7vwG!vrA1P?OUzI+xXx41YiOPR}8Rv{Lxp}VeNul4U zf8j)^G9D_Tm?a3PUWT*O`0CisW17%Z)1cqXrr3Ep?~`i>EqYgacaK;0A$(@p-Pe~i zwZjjuedj5FXoG(rxQ7+4dTL*NZgxdy`t49~Rn9HsHC>ZrKYL4*+R=(5G?hHb^B}X{ z{HUbv>z*xt7!NM6aiXRjEiJqAjHsoh+GDC|a0z)Bt0sntzbO3(T&>Pr{K?n|-uXZT zp;q4@)?b&v zQcUY}xWKfSX6b?8kac5IIqWDyYIRQ&uD$Ju)Uge?<=0K!HCpQ-LH(7kytG z-zWSkD*BvJ_S+f3a$ZXo@aQB&4Ka^vAYeGuT>o-k$H+o05=%=KeFc$E{zGfCDA!7a z*9qkkdxzcSiRl7mkx?#-pjvv;1i+# zEwrxHKq>4u)70{BV+3%BSr-YO{g5F+&(+Ols;O(Pi4o24Dw3SXWx%43&jSOTWF4}K zboCbuy#)lukI@ABk}P~^UhXDL%mZJOA@O8MOtCR0!0-F=nGY+NrxyDGxTxMK5|fKP zJ*RH^Ict^3-(k<4+t^$ zfqWeF)S#+zkL>)zz2H0b$I$%8l|v})ABeRwh>f*~iGI#Ub?Iras<(K4@`2sDQ^Yb9 z24kwPtP7J2!HQwlz@|mLB>JISYMeCFeKwPMOraFau}e=j%>zXg+D=xyLr^=WJ*yLO z>&pY@6ZiZmvb{=iVXlp?VKZiuBQ6p&6HXX2oBWSk$F%K6&)& znG+c)#p=Cw)f|Er1~gf2!8+@a%mW`+oc4}whxhcqsi}*35{y}9vZvER!-(I)o-Z; zvR{};k(EA?!jUwDhpb^j!YI!s_IT6~DFa+bH{zdW*-q>~MAujs36VyWJ>6~17-#7x zP{-Ni<;GaKrDzucSFFN?4z|FP)4-fWc`Yw zYhF8${kC@686>SRb;Q$-VGQHL3jmg31=Og2>!2qVyl^5TOhOOaAoczpQe_&;HosFr z@{Kyfh>W@RQPg>aA(?AL7?-Hl6sFp_TNmOB%QlbOQK9zsUpxpuL5YGdR(W|0{z~VncKsPS2 zOIVB&J@cT(KK(@3KeQV>$eNQCP~_nz}Tr~fVg+X4_M$bsYlC@25`%F6-xw+fH}U}0cj zVxVJTVq#)rW4*>9!Nqy=28V)(7@vfOl8%;!lA4;HiJO(4fs>J%`n?bv=SLoXettSu z5eZ>lac(|--v1c{1sfY1=M4@yE-pFmJL-45|DW5x9snU0%I>QQR1^llD?$`hLX>|4 z0J@j=L`V5w4e)<9lvk)|=opw-*stHbbZ8_1yh1@meT9aKj*f=*(mUYgIRK3io#-9U zCk$dWQ%nX|65imXLM+D5b-kqOGe{;rGq(`z*Kf(lDJYp)Sl_d;^9u+H35$rz{3k01 zlm{tjXliNe=<4a4TUc6I+t}KIH49~c}O9vK~*ots}+Tv}d%Ber*T_x2BdA0Ay?Uj4bgxxKsp`#)SL z0M!4D^>Y2+VE+#;!WXVrXlSTtnE%6t^2+<=L?uK+f5(GC^hpiV)RmZlHyDfLb5dbl zFE%5eI+E1PZRYh`CVm9-#s8rFFS7r4U?Km%$o?;||C1(uZ z4$3IhO_k<1;i|HWb5uBCx@TOZ+P(JE(xdPAGve@1ZEi!R>|%_!qZc3U^t+ucNxmEl zj|0+Nh3L1A>Et2~ekxHroA}@e4NF>PK4pk1kazA?H zKJv&mmhd_aKaOv3x zaa^Vgxjw5XCU()-Z_4IC;!}n_dtDC%jH$MD_RWZY9cwP_kT9l8uav&$;Dp!Wsd!Bz zS+Z-^z4q0Yyngk@yp<3TPUYxtERjRwZV0&8;6mqHV=ED}!N?a*Uf|_9Yci&Ui-ceF zVi-`Hda844B9tN+C#CX*33oLV?AnftXo^# z8e6hXmtjoV#Xu)uvq>Q|rZ+1yTPu9(J>je{3&vI#wi>PLQV#K*4vs4+Rox-%L#(-9 z-kCqB1cTjE%(1^YgH2;VU$dr3LT-Kap>iwhe&4Sm%bM3>poC!c8zEjMBgKiR))U?} zMvY^`XnMvgEfvvA;Dd<7agf6q(fD@CujT9aUEFuvJ{{%PsUKjp8aKzQ@SN_qkL2fd zo?%u~9~pv>wdo%6--#!|mT^`Q?m(&>XDR)1`1eu&Y--F8;TFHwZf%xSIas;pu&>?0)2SewBY8SNrt$Ww885S#xLiopmq80)?6(5^D-! z8vEo$&Yi}V330h~4Bs!6+)f&$h_dnrr-(JSAZI2J!&r+aPU{~=hN9Mm69>c7-AM_#;dYV5RwOV_s7a5XlsnM(yUf12r+DlO^{8I@IRRj-& zsx@Oj?I!~4G#49I-ths09A@yK%xL&S9ef%~9h>-v+BaEdA6>Q)1J=869j?$RUw5tN z@iKYKoN-Ql!^C1sIf9mhc*)dB4Kh)uS)i;a^~HYcIni;JzpGEaLsUxHm+)t;EOAf7 zjZwdvMrgxhL}-q$&X3K_H8@yq7F-fcdVq2h>tIw?BU8I_c*E@27tz=KCb3-@R`9mjWM6#d`zMYG zaTAAhI(#QcQ?ib2-n-r4L~Q@qAojfu3B)=BK9-%(-)w%-1`!^)M^Z)z>52^1q^xG% z>GFOJNY1}bGImb#-JP?}0>hj`?fCjgX|iBcIz3Lx$RE-w%t-#@%+hG8Qm?3{7Mg@o z|MM-z0H2*s-tE1e)f@XTTskx>7U<-zAEPMzXBcVpcVp~ZnEUy;N)tnoV%icdZA4@1 z37a2IwH6ZvssN5}XY9{We(#_BYRNSHAtVn=IW|Ikik^IHV{`Qe@+3SDo9~`I586D7 z*N^^&eK*yDM1UW}Gvt?Pu?OFfzVEw|%h%GeJDC0lSoL+fHaUIcIN?$}VVG5a>&jF? zC679qTW~vUs57osy-0Z)td|Sb-j(4-B;5|F*w1#@vPS6)1kZYetm|fAxfBG$#20@K zh;S@xeH_E($m={{y{&*T6rGv9ir;U+#{oYmFPxdpv9!cFkKvfVvv5e1h6P4kK=TLr z1~J1LDOj?{KB4+QM^Jk}vk%=-K{dqE*)4c!6Q}y5Q#D+B71wdYvo&;ja zwq)WdSDs}YBcoAO*e%SMDp|9pB2!O}ZnG{P-hP9v4g_UHF8TRKhz0m+#c-)FsfCp` zru(wNRd^=B6QXm6lpz)7od1z#@Cd)j3u)iP8+AmSVqE%-mnx#?`>QAQ$h{`{>OJX8 za_oK-n?va(_Ksd^fXzE}dI{HYdlMhNqFT{gF@St3h+1${xk~xQvqU%6ri^J&cY&9Y z6`s$QjLGaKhef{?!l66$7k4gM$7TUPZ8Z;Q7xM6>dCt-J=*%)2Ye;Xk0BZFQpwZTF zK^U3Zs|6>U&f`r}CS z)04H9H5_}Lxwx)9S`!t!^Q0fzLz{UNn92h!WhI@4P%*Zc(1AT`Dd+6~6A*O&rhkCf zjjr4442!3ZR&5-7+*@UQBac;+mKStkc)4N%16J~*TCcl53XH7+1-rAOJj>$u*S{eY zQZc32Fe}9A`Zqe3caH6LtD_|L#Fa}yJdwuqNrr{$SiO&SCa3FC+>-O(A)GwZIzs|$I1s`tY9J^(!_4>)T8vdmf z8!7vIG4i@rBU;|MAa8jM$eR%?64*~;AC82U3LJbdZOb+Am0Vn(zgW61V})1KC*{dr z9;k1~;61n2ejv+!GH(9mI=kr;>oU1+V&z4d9_ew)0eYTqYgMyOdRRb@kPy{oyURpa z#+8U>eK+aw5~3@0*uqC~<9yc$%kB}LC z;+QKmnlC9^fZD5KQy8C{A4+fx=g}@Y%jU)x)vIWBy(au7S=5xRQU58u{X4!l>D50# z@P3-3nV#Zfo!*FUmD3+`iC}7vW7UP)lD4kZ{m7kuz;~MAe3fghDt6b_@yN0Y#Z=no ztgw>mBC&GaykJA>1tZ`9rBkwNfFB8G(D%F`_KQulNuRe}nL23#t|6*_yRcC=XIo(e zHB+TEM#^p{6Na%?b!lAhS z%{o*O*Lxe;+q$MvQQ?N+h8G&(v}aoZpKTvv2VVJtu%2NF*TwpZIM)zcJ5N)WqxMUQ zEGdG+dgyMO9BpwH=ux(8Rb_h8gLK{^1lRw*Y+cbl$l0ZONO`N(#l`&uU&&9$c_(AV z3F^cE_p1Df-YI?hA5ThhXGC;AJD1*>FTXM8EURPP`c-@wR*grdMU@!HqaDz27y_9)FQ2L)6_BmX*GG$@je`qo zYV4y97ue=ssq!*W6NajRWp~itj`SGi`7bW@qwag4e}Gry^!dK|y?N~E>LK=QW^vq$ zf{7C?ZxV%fXnb`9H#j>$uv{N!zH_SRo_rW-SosikXM8kwsW+IhEPit^-F+Ajb4Y!x_2Wgc?nmcTrv{{JR70KUeC17-n9dDD)Ck51x*>35^Gs(*0tT0m0yYOY5 zHcA+rcoA)GD!@p1Fh^0D5=H$|46cUMmM4urOr-Q169#ao8)QPw(O$h`3^*FPOGO0O z8_su3ijJ~IITo92jJ&dtpOC|U#Hjq;B~U-l+@R?Z@6$|TR6sIJ{_MX<7uTU&YjBt4 zr|75-J{?f|8B3r^ZIPF=pU(e|26yLms#Q&^tvJti9K$pA$)CHkm|a!!Oz|d6YJe)~ z?=(?Bb`R24Xu9;*cM)}Otf_6F!1kMMRRX{)w2HkgxH$}l72j z{@vcU>2LSXK>OMxfv#lNyUs+4yox@^Zf%4-@Tf}J5dw%RRTV(*|2QP=V$ z&)bt3mlo{2_yy#H-n5h#C>OLbIavE4YQ*2S2;*_JD25Tybv4!^{QGA37*e1)D=Bt2)}yylt62Nuqx!El@y|Pm z+Nj-54?kvItr9J(TsIvlZj6B78WWtm4})L8VgBokCyV&hl~*8Mp7D1SJ-2_CfI&Nd z=DED!wnyn<-+ZvN;e2yRf1d2ik@?ZpjoBu3pP3EH%{g}qUc0y^%9@jF?AOh4dV&t! zzvm4j{DHLW-7}SP`LM~A;oYanMMQWTb?)%~UGhfUXU;sP+mzAfJ za?OM?+Yf<+;NI)cH*mXA%q7(y|@}02>ehIfB)>+mb|>%ChSb&FLcoei*6Jf=lUqnIIubs zJHj{IT8$k@cz1bxZfm#vCl~*^);t2V=(#R;vmIc3>VQ5UMIyZw%R`q6s` zdyAwUs1l}(=v}HF8W&gEFSP}QHMH(F81&ufV%c(}%5&7Whj zq#6%4Moz~%$id0=Jw%($8O48LuaP9wU!UwYax3BEF5KLV zVW>QggVHPKS>uE!=FF!;)>`Z-Y7I(a`LAH4-{bUUz~Z< zyR=)A_*+e6^1VH_deJto`*G}pPU)T7vyT9F^_kViI9Vp0l&?6_^+;4Jfn>cbzXayR z7&}h+7Uwq}pr$yizY_H4<@-=*<{VV)N}}BbZRZMYYx;KHq&YTBv-D+Zg`9l`^hj=Z zqUs~(Cf$x14l zw8^7XnCMGKA3vCy6`%Z>!xz_#Nm?mJm zBfVOy=JCNgb)hxi{f6s&oFlTVwWY<&o;pwi&50WcUO+eH#e=fDarlv2R$PzAi<#qL z@WI<@G=_d1PQyRl#Kc6=u25>NiKb$9mqYF9AllS%HjZ*`6;j^r=FbEoC6Z>133)Dx zA4G)DwW0nzm9oXmu^&Gl;7!&cg@}h)AFm7tn`3B|B?QQsG{LZQdi)5d4v)igBibbI zs`VqCx7$bVjuJ^NVVV7=&53fZzk#IuEkE|R_j@4}=H$3hI$nI& zyE^t{-$t`1k&+w+D@|YKQXT5AlgzsG4W!joseZ3$<(pQ z?ONhNh<=nnRVJ%qtC=_cidNt)^|inptuCv1ewGQ5m?JBKCO0p_r^MJKpk!@=nKvAM zG=@b^{Wt5>pBd5O&ruA`bs8+v=6pk)tg3Dw+_a#)E2)%~6XP|WFY*!p9{}1$hqAt* zSb5~t>^-;Ez%zuI_0V#VZ*IgLTks=`|3{^Ot%HYecoO>>{9e;M z471ant$HK7RA`9>+-%CE@xw5W7vD{|V$QgQvRYJJ=$yKq-i>cSnlLHvo`xfHyP^-m z$_>#dMl+>m2M#7lV~XR1!bpc! zCIc!I7IgwYENplBDD>3OFF%LoMPwytH(v-vSQ5|?+eLa?)MiEpX_VbM?004^-5nrq z>T2z(1WE*_|NAsPwnYSjNsX1;Zel{hYRX(mF5gq_aF>$Sjqa3By6eYxZ-WF*KMO88 z!QY?JsJ(G1mT0NiTD<=A`ogZ6r;gd1gy&EHlO z-$5K5olH)Ysk=Qmbr0aJ)*+?p%Jr#0(lO7aJk*8@%m_y=PYq6;#0oXukK8D%p?RH^ z1xTaCq%xp&ficjP@)ZS3ioHdi3h=U$_qeUAwCu3?;^-g1jCeKOCR~h%Vwbi@TPZ8P z3T_(IdhHk8k$PSQ1EP}#st^nbuN08mVMS7{ml6*s4m$sc2`ZjER9XB zWN}DJ_lKotYwb<~{)Sy1*7NAkSXIfp9#+)Cy{eIdJI`$aya)vmF-`eqtae}^3Uq&I zU=#ii0G}*7#^;G&SNNVHCh0n9&1gq=b@=a^tV<6nz;2HE5_ zr#K(5Rv8j8FnBRdYV7RjaBFeLI>-dKB{sSDdrh3=aAnMW%Hb<^IL=mDif-dqU)k0~ zpuSgwLKWmhM`T2Kg^E@`Yr`SyfLbzw8!9 zYp2f%Bu;#2O)Xvo>fA5M)75!_p4xDmF6<)>H(~Fq9o=!{R}?dvdbZ${N#;v%9{c&X zhJB{qmr*p?1Fjy`jrj1YEY-`|X&1zz7N+^T?599N1iS%Wx;H^ALqgp8HTF(K!sqcH zKy+#L(a4mg)YjQ*tF*ryE$D54eP-vpXLhcf?rEc9W|1Dn=T_XtUn7T_^(~X{9gL_Q zDvZ@W9B$%q!}rCqA#$tZvQ+LvfCFUlAT7~SqZF~k{M(@4YN_l>q4?iKZid(wgFP!( zH`vnjC_wk=W#?nfjn0$?heKLzoOKaw6F%#w=L&ibHh#^W<9UKw`ajyJC-R6F&gkU} zn!#V(wpr8-mbZ``b&KLO=Cy}{UrhK0x$05&%TQffdL>hb6)&9nMoZo~8}SZ(bUHe<_1Xe%(kF)AAtVIPNoNA3#`6N$ z#pr@coxv%+mq+PuYb@SHi-DEwuo>x6eZs3)TFW>;VAQyEnRxDP_FvsRQa1sLVO#JLUZTL9o+|NA&&h&d0XokQ0x(W-#x{Xn$;2d?p)au^ zA7#NhksYMfx<}rUvh9jC{l3(D(xY2{J!+l1J7QGp(#`@lG!)45wI6yTn7jL0-X!4LE0$-V zSr*2gj!=_gh8oX7?jXu}8fGaSOFYGVDA->B*MFeqdUs(rRFNAryoUwpgvHg;kPOw& z8;!BQHC)t^7Q$*o*~&e6Pdx=&^59d#DmRvfAwe)Z#O~ZRg>!9elbiUvQeE+s?g|l> zPD+!=q(ftd1ci=SxIv~u*vCo)NcezaSbsWw=$YES&eTR5dt&enl_$GF&Ml3q?|vVm zFwFR*nH?NnRx-1v_FF$LFORh)ZCSpe@zlcMt+t3*g6dQ|jPx*Fe>0qDNd^2kO!(0Z z%n>~DcpNx4&?YdhNRk-P?gAUPDfsFd-*L%)R$FE3B{!_6zL{xQ@EoGrhgBMx7*QeX$k&^kIj5WQKxX@Y93|o0BO1T-G%{ z9ylMkxS#YjnIbw~;{<`fBymJXVAn+#p_WDAF(NrbVguvoiN|6L9{p1zVy?3uBUEUz zGM-sHOsl#4Mex%WKyJn;HTJT9E0=NnGwPaZ>kl^!po(b9<5QK7r{k+54H93+0!x~c zVhJ%k`>IA!!Q#^Yisj;Jy1te(7^3EZZ_C3--y9h;l2i_-4QH@4Tbgqwu<@x#W4w=d z!IWh!pk04hvMkk??A)Tr+gOufYxB|Om$-U78n@L~&PgJhsowqGd``-#`PJQkQLAp| zYhy^i@;?ADbJ;QUH9>G5uv0%)eIdtAy>*A|uQB8%r2cd7r{j%IgC0u$Hr0q3hsl~mJ_tcu%hU;Nj= z24zR6VN1O5hV2Jt>$u>llRVMe+AlNwh>}A=ZIRJ%f|9>^*AW3wcmTv-D7qK)T6( z(;B>dFMYG|{T3}_6baq{9(`caxbnCMOJm*Op>OV4)|DcTZIq@@^g4b^%_ffA!GNGU zwR4uXH)Eotp{xF>z>jCU-ZK4h;#U}sbEtU-oysK}2CH^ofKS;@B*I!eLzP+y6eBg- zuZsK^MNVp42f*!!_+sLJ0K0Jyr&K6T%6D}yC}`K$H){^L)Vs=&%iDLQ-PQ(5Hfdn` zm>3NFf{%8+-+o+6tZS$!(Al}7N1s?y(BEP`8)HJOTRNcqYps1;>@?;UUCwp3?lz#b zw$|Lh)*bb~MThE*8|*VR3a5QE1%%Io6>}!99i!D)*>X@p+mr& z+MDd=hHzxcbzeV9FX}k>$u`S422o=rOf^1xZl<}G^pq-tAEc(5qNeW;!GBSCY}W^E zIlZ}@?V1uBXw57wzJB`#X*O=0_?>Pdt_rUf@ll(vWK~>^E|0%H_G@U*!y5I{XIC{fXLB~h?n)st zaolg~xU|i7C2XhsstZX@UImAQOHN(n3y%|;KNMsf&j}cH=-67?zqX+?m!Y0*+KXpK z^}-a!-278vFsPqYpXKT_HS+0lnO^unnBoAU03&{_wKKL+Ha}lvaaSq!4}kqU%jhYr zpf#K-im5fKm5mA|x9d^cqpkZ}Bgg#G{=B0_OY=8Z)(zPy;P}&)vbQAN0@Xm9=dAH+ zi8mpS3Ad#-JN{=)FNR-Gab`qm`oYhcfr7LR2a?!-fH>8Kipwf7ThoeBK@Fr_BVDAw zUVt~GNHE*k?ww+A&e&jZ56m=Su@-!D#Kv#G>Yd1l))%_ABVhG>pn~)cr~1+yy~4Qm zs0Qmv-&wZO)bQWdp^j8qX-3#U)cex)|vuMmdAB)lKBUG4xx zfLi>nOB2~QiifR z@{|g3ruO`Qg>n8a6PR0d8c@fjlU{Gm#wwaZu1uAoUD?K){8NH_Oj=l0 z)Bo+G6UJ#J$=;aJ=Ta{;VSa*j=Ig>yz3```V_tyt1EF{N#N|T{H%IAoW44jHo_LKM z_KUDIE)X&9@a!2^bxdiYe{z$iZ=0?0+5aDaeQCmh`fJLvMyVWOjPjaWf_rv7^7&yL z$#qlTR~3Gq%6bJg9~>tQZ)eiwdf)gRNRG*b)w6}ti`lx}$qqA$%24!n#?37j)(e~t z;r&D{#8}8kDr2MHTdM7EmQH4;SPf5Hpwzb+v3NBkSMrKV^ ze%X;Jd0BYP82{N=ocHYbx>`(?=+2TN^v-KmH-oUzRCe@m zsoVFIf8ARlfP2m0cLfP$-$wr{*uz~8WT8EwaVA}E=S>#*jf1SK2bkf$VXKv z{i}Q`TbDx~!v;5|C{K?$_MJ1DN<92px*A_`p&5Z=7;}kJw!%<02`O#n8JoZx7(VZ3 zczhCdrWIjDm`M)KKY&8XtVr8$Fn78qQ}ZO}nOQVXT2hzVVf1xvIOWgzMqITd=>eWZ z_uPK2P_@abI}w`gG%7-h#H#-s$P#L$YMk_rHz;RLoN607c}#Uv-)i8z!#2E_gNkfJ zb;EbY_KL8tI;TRON6U}{O&L!INyGE*PaMfUmO)`DrF{e!G#O69n}af}gUXUVdNuo& zuO)S(L-*a6AFit|!DH~ftc|Nr-m`gh0!}g3NuZM&6_}PG<$w;rHb&N4dALUO{GtV5hC-4qBX+oom3lah0#gX zsav77>~6g8hOF)o(l1Bxh@fC|rbK65rt|HtWM%fQTR81pDm#c-HUZM?rjuzDIwY2^ zQGeZaO)xc5soM#VGMwguUwRs;P;;1qhyY-zG>)ImiybJH9T&FGXwjhHuN01E4oNw; z`-w--sP#$oZAd0#+gN!fd)*%&r%;ZE1=cOsw&f8YEOag~icnEwAUhW|=v`q9mh{d$ z`F1DQ_+(%!2^H`ba2qM@obzmjb;^3ygzpGDr{63R^yko3Rd6mb>5#nDdv@qe?X{f zDa4{42M-!FA{UbnW3nYA5t{x52JELV(xbM2jZtBhiN>U3LkzaKRKdFcrO~jRrr0h` z_NsVL9tYzPfuV6}c+SjyaCF#L)iIvr{=63I0Q=d-e*lU*oxbJ^Tb(AyF5%FkL4c)} z$PmX{x~M=RAq-HBMWScK6WXi16&OETOZ%p&U}mso&Bh&LueMiu8ktVIvBx%ZIB%e| zoKW6y)*juqTr0-U59V{t6FT<&yLI>Ob5aL2zS`Xv`M?>+6V}FA*3S655scS`I3=h< zcj5QPX+6Eyo>RHJOO6^ADjYYavk~;x=?h% z>k&V3vB65Wn1J*63AxZ15AlKAiEtS&qS?l^6_2~qJoTSs^5rn-lamgJ z4f=cSVK)zdz4P~1_0xh|HKs15Q?(~alWyRMPb`6ZYk;!P(__g5Y{7m$VIYp%kyt(X zK$@4UwCjGlep`tY4JAHPs?<&GQEu9kUb*>OGWoRSZt52(&oFWckA=5Z(z~#W*6zC@ zr{$eQ^IerPyPN;aW1QRUoxN__(*)8Oh?d>dnqrrzDk7DyX53r=?(cWAQ;Y1Nb95-L zKKG`2w;G%h++cArt?3HhL_C+Nzv1HguC)B#V}0ui7AU&#C8jKkrAF1}+&t;d6=JA= zSGlp$k22jsfIL^n=5BRVB` zL<=Ss=drX&%%4jnuL5!OW>Crg!Id0K_u1R?=C)cQXooIh_LQEMT1YRLmV(Jb9Th6Q znh>Nf=#6h($kIwFnP3Z^ksql$O=LKue?m7|nBdE$p`@TCn+=yAp?8s!H8prtpEWo8 zATFFu;>H_u7k|A_8!Im0V+uLauwmXyV~J-btm36NYdN4zB`VKxSl+?NozYO8T^tg@ zOA2^Pox6D6%e3Hp^5UbN{R8aw`~z&;gF4hqZQY!>8-{I)9jixe@;}gR;Hv%uge4)m z|6;t{_m$#fS=( z%-Q{`Gb?%>)>KirzPzDe)|ADYEKKi6r(x`g`MrY-Qq`07{n^$bN?&&DNzAOxr`5t@ z$1OehtcFSGQ$#v_quBhV91xA}i|7)jY4u3jl`L0piV`nbRH+l__K&!$aXZA8mOS}B z4qDT8zhsKf9?I5>N^dSOp^7YiNrT$K2U|uN1Qgv?K@_7@t&-Pq#k|~9a zhm)+XoV?HQ#))s4zjV%mKsg|GT^wYPROwskI@ zmdd}rcWpoiW9HS&FQ(OO{R7~GDij4PwjsQXUst0TKbQO=KR(o(im}jK`0diJQj4}GNn@1iz}Aqo=v zxe`WH|4=KTsyfte`zHpiUeUTe=9Z_@GIF$Mm{9^;&HOzjZgyVtkJCZuo&JWwo9mXJ z4dB#^s{@MO1LSr)m;Wc9O5rr=6h%j>w!8wAA*m0GZ81`>ZO#Kfgcq2@gBI$rN}l*s z7CTrWCU&^qPp!^Q4`1HDQvuMO1R~$<-O(Qt4~T5>lB-tlb(S(!e>0IG?R4S2UOw({9Id%&qplQ*N0#U zWZRk<4YcFD9sRq|_e0ae+fv(c>{QpEma!?FeOK;jW(unU%F@u$ppEnDdSj@ z8V{<_9k+h}`nXlodpNKgp+>3RveriX^>4={CHXi(58xw;`IFc@(vD~Y>}_l<^HYYx zjTe=WcPA$(-?#2V*yH=#S&8nL!4ABYWg0x;>vf~}^fIuw8ghLjxbP^cV-DSi$y&!f z9FzIe;gA$%`HUqST;U*bun8ocIme8r_)+2_UW7W~DtZ>#MK+LjOvo-IxYvt$Mq500 zY~=RHArJkWD|gt38ZM{@*nOI7uTd>}WLFkNIH&8S5E98}Em?sXml$jHQx#mj4RTWk zG+ga=mG^zfV{?1(V%eY5iE-8EU$*VEs-9ox_2NW#)n+Z@{TiZ6}mi7>n zQyZ<~B;De;(z>aS%1*zOy!*DQeY&P3ro^=K4ix9l&)&Yb0G%7si#NFi_2m|>jR(~n zdh;5{kduj_vWfjVe5sZxzXSH^P{#hvX+vDmOoXW;^MvMKw}oE`qY;j%ZM1$Q8~@adJ0(@)#~LX~B^4=1L9@wVMMb5*l=Iu~B*MqwYK;A8>-+6XGMTNevv_b0 z&P9$UrE@rx6ROYZk&9Ek4Qzaunc(KD`9K-r-mx%x+hMr#yG;ji zaNd)$X2)U&qo?4g0F}<`qX~6Oc6kwsgTYrHbGZM$F} zSt$<0RD2E%J6>e@6UXAFxog``QIV>m^JBrPYKvv1*hc&}e2yEx2bw~^IBMt_#3Tt6 zgs2Y(dWYngx(Ba{YX!W|1}AAcCsIo;pqDyn2It{_aW^`PLQr)Im<0een;m(DbCj`S z#(AC4EJr1;GkuEgtk89bBO`+Yx2{NsO$@IWt;b)%w| zj?MQ%1fMWa)`ZRaQ=(UpFAhLM99!1*;=({dB2%Uc;t$NOs`9Wc#tzp-Ak08jSZZaK z*MP?Bx;#vTOCe{@m-@svAg$EYWhq?>{&o?%_cdS);W{TW%6JHGQ6+J3RFGmWO8N7%0D5-cD?Q_Co7AP?N`MF06biIBa5{8Q#XW zj1*i5;?n!_7d0!+pRjBL#JmHJDoHI&rmwCnbo)-r_uJj@hfC^@?UK~urHYaxV?p`e zIe|BN{4E=BD({R_-by*m3XNWl<%3SCFVovC0aEJm?-`Q~8j=xknSG(jSd?Xt?FRmT zfHyMM_Do)Rj9&w_tpy9}-b@Fa{kTS^z66<(L`TYCOk2`{&wzrDr?;I|c3_A9*f$Fl zDqZyL+`GsY_+faPc1d=E(}OfG`Dgj!jnz()cv0GG#bKSlPMv=CPd|D~UxyeZrE~u^z<>VA);@qBVE`rbr(gzP%`b?Iba}25 z>V8d+v{=O5F*38=`EIz%A40!*IemaxG32Wai*u~rMb<0wV>`S{6cpAokcc?goUJ;d z!{L@{bmr!~&U7rxlcD!);3UgIdVAj9RDAG9udrJRbSi}!uG4W5qBXI;25R&|2kqab z`Rlw%h!AW}^ZU_Q#Pkg(F-%U10bK(r*c=cz;sjwJKUK5DM176V^&DVdmwYSzbGCTe z=|wEGlwVv&=$L%WF+V#Hm|U^hAg(t!4mz#+4*CXqJe+o*_zgO|o7->Ovmq9ti?VeV z8Z;0-qk22(NFJ0sIbsAP#IT)BVo(XPLxKC z2`}Hs(Kv$`b**WF(&Bz9vWbRIcA+2Kta5o)Am6ds#vry3Y~H>~ik=?2GpB2iPjl$~ z{z(p%Ky2%5yZz&}IlA@ypO3m3>MFcPV?wVM_-O)o2lM*fgCC_jl%9iT>t6Ev1Z8F*}{BFf`% zB{h>%>Eu`|9HO{T*la115q^}VoLZ9< zJ@vkrIRmPj_YYv0WTX`l-Y{NO5>4Yq->x7mj8D%Fq+XJs`_@5GDtlwo;Vz1hzh{lv zJCMrdY@^9<4py4u$ebxHy$2fa2ba2?s{!H@BguXDkt?i@5Y&oZdq|=PFePsXGNK zl|NdO#mCiC&;A&Qy=hT$&5Ws{M~la2rTL|=&tMBrtErHoNq&NEcx`eFkKO^6)b}l8Cu44^=2Rwg z)x?T^oyN0*&EJX8BGAFZAM$Hm;v3p3gX}2>f~7U&tEgZ?#(_>17{XvP09!R-Q{xs+-x?(d-YiYk}CE38;lC z%cLk%II}!A_I*Z(*H&H6?V!7fe+$`5%t`7}Nt1R0@|!QP) zA3SR2Et}QPe5vor2i|2vGKw_$tiU0;{Zx27@g-TS$lToUT1R)EYJEkPTCs$NUh(e) zaw2N;*0DFI>M{^xBkJS?%r&Oi-xv9LzvwzGb9=`(;*A2#wT7TkT6pf_ zP|4gE0=Mma86LCajdN7g*Uq<&?5r^s`9=WAAa$xVz9V|f&cA>q+4FYO)X1>a=DE}%hA2Q~ z8>D`PPIx2QuXvZuyw$V_rIO)elHK>TC|H5!9&kF;R{k5ggUDfRdebHje5M7-EwR@~EgklD;_{4K28T{^i} zNXhb>KQ2GKI_A0iT?51xejaz!^y~QUXTE70bZxn^gC~x>)!@CfZtJU_T~!zG9qo&0 zlSd_qx3?DiN&pp05C#d(I`_?S+V6#=vePeY(rwoFH#XmAmO+j_e1lsynEu5oEbp!F zGMj+6Zb%s14m*G=16H)vCh(Q>n{7r*k#lZAk~U+;{znz*;v8KyZOkY|s+Z_kvha?X zYiSfS#SD^4FiWp2d;Kb&gQ8t{R()dP&Uf=#z?3PKW&P@K575@8m9GB)#I6&m>H3%3 zBN$+|ZgPEkR9+rB?}~L@Wd@UT25dIl#PO8ee8eY~z|SA0eEteJwN>-&*Zu+M;v6|j zx_^naZBFDv_WeO_0+-mbxx#_zo^y)X&~?3D_fLE>9!EtwJ)l`|WWyyWwqJ6Ab(rRsKeQ-5XK-otUig0Z*V z&Pb@<|F=TrT$9a|0fE9+gThJ6LV4 zaRt*W$XKC?z+z88F~F`VINHr>2RC=&<<`7itEZpkd2`7mjkh>m-jxo!;Yc*SB;VUz z8G^paBV`+b#&eKyiqO=?v!{5YQ<~39ye!kW$+u=#T>Uw$okBklY7zNTX;+B|P9uud zm2>pRHPdM}n%L=5rr~C_Jr~3JNYwR4n&R$%BHGH)j7cd2CMCe@$3t5>=7DLd>2pV6 zb9rxh5|)q3nV2yhh$j`5plLoK(zVaFHiM{K>DP8P>xkHZ7($~Zax>2y^scsXs8}l< zFT?luR}!HLXH9`VzomE>+E~Sk=B0MJcKL-h4o!D4{3}15;axE%W;3h=J7+Hc06QoH zKdoucZ6ix-6!4pdKfDhR!Qkg(9)x?s>9&mU6vC~J9MYdsUNp-DhFJQ{XVXQ|QA#)%H%ab_IYTtvw{ui)8 zV>OoPCyYkrx|5ve>zcur!yYQN@Y$AIJBxee)g!q_kYu*l-<~tvf%w+(*N)k5FSJ|l zv|9+mszA@Uk9@g2RD*S+4j!&&#SOf`e z6xWZv{{VLpd03Y}on&cJc#mHH0ED9Y-$}WS?nJf_Mj
0\n[images/image-1-Transformer Encoder Architecture Diagram.jpg]\n", + "path": "images/image-1-Transformer Encoder Architecture Diagram.jpg", + "metadata": { + "length": 46, + "summary": "image-1\n- Input: input token -> Embedding (pink trapezoid)\n- Positional encoding added to embeddings\n- Residual stream processing loop (repeated N times):\n - Layer Norm -> MultiHead Attention -> residual connection (+)\n - Layer Norm -> Feedforward -> residual connection (+)\n- Output processing: Unembedding -> Softmax -> output probabilities (over tokens)\n- Key components: Embedding, Positional, MultiHead Attention, Layer Norm, Feedforward, Unembedding, Softmax", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-1-Transformer Encoder Architecture Diagram.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "7b7a45a5-5c07-5273-b864-780baed0c0c9", + "type": "text", + "content": "8\nTransformers\n\"The true art of memory is the art of attention\"\nSamuel Johnson, Idler #74, September 1759\nIn this chapter we introduce the transformer, the standard architecture for building large language models. As we discussed in the prior chapter, transformer-based large language models have completely changed the field of speech and language processing. Indeed, every subsequent chapter in this textbook will make use of them. As with the previous chapter, we'll focus for this chapter on the use of transformers to model left-to-right (sometimes called causal or autoregressive) language modeling, in which we are given a sequence of input tokens and predict output tokens one by one by conditioning on the prior context.\n\n8
0\n[images/image-1-Transformer Encoder Architecture Diagram.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"input token\"] --> B[\"Embedding\"]\nB --> C[\"Positional\"]\nC --> D[\"MultiHead Attention\"]\nD --> E[\"Layer Norm\"]\nE --> F[\"Feedforward\"]\nF --> G[\"Unembedding\"]\nG --> H[\"Softmax\"]\nH --> I[\"output probabilities (over tokens)\"]\nstyle A fill:#f9f,stroke:#333\nstyle B fill:#ccf,stroke:#333\nstyle C fill:#cfc,stroke:#333\nstyle D fill:#fcc,stroke:#333\nstyle E fill:#cff,stroke:#333\nstyle F fill:#ffc,stroke:#333\nstyle G fill:#cfc,stroke:#333\nstyle H fill:#fcc,stroke:#333\nstyle I fill:#ffc,stroke:#333\nsubgraph N times\nJ[\"residual stream\"]\nK[\"Input\"]\nend\n```\n
\nFigure 8.1 A transformer decoder for language modeling, showing the residual stream for processing an input token. A single token is embedded and passed forward in the network, with the feedforward and attention components adding information. The multihead attention layer takes inputs (not shown in detail) from the neighboring token streams. This is thus one column of an autoregressive transformer language model, taking an input token and outputting a distribution over next tokens.\nFig. 8.1 sketches the transformer architecture following a single token as it is passes up through the layers of the network. Each token is first converted to an embedding from the embedding matrix E. Recall from Chapter 6 in Section ?? that E is a linear layer that maps a token id to a vector embedding representing that token. Each token in the vocabulary has an initial embedding representation in E.\nTransformers also have a special mechanism for encoding the position/index of the token in the input string, which is simply added to the embedding. The resulting embedding represents both the word and its position. and is then passed through a set of N transformer blocks.\nIt's common to think of each of these transformer blocks as part of a stream in which the input embedding is directly passed up to the output, while simultaneously being enriched by the application of various processing modules: the multi-head attention layer, feedforward networks and the layer normalization. The value of the stream at any layer is the sum of the original embedding and all the outputs from all the previous layers and blocks.\nThe core intuition of the transformer, and the component that distinguishes it from the feedforward layers we saw in Chapter 6, is this multi-head attention layer, also called a self-attention layer. Attention can be thought of as a way to build contextual representations of a token's meaning by attending to and integrating information from surrounding tokens, helping the model learn how tokens relate to each other over large spans. It can also be thought of as a way to move information from one residual stream to another, augmenting the stream at one token position with information from another token position.\nAfter the $N$ transformer blocks we take the output embedding that is produced by the final transformer block, pass it through an linear unembedding matrix $\\mathbf{U}$ and then a softmax over the vocabulary to generate a distribution over possible next tokens. These last two components (the unembedding matrix and the softmax) are sometimes called the language modeling head. In the rest of this chatper we'll introduce attention and the rest of these modules in more detail.\nFig. ?? shows the transformer architecture applied to a context window with the words So long and thanks for, showing at each token position what is the most likely token to be generated. In this full figure, the set of N blocks maps an entire context window of input vectors $(\\mathbf{x}_{1},..., \\mathbf{x}_{n})$ to a window of output vectors $(\\mathbf{h}_{1},..., \\mathbf{h}_{n})$ of the same length. A column might contain from 12 to 96 or more stacked blocks. The arrows in the figure show how information from the hidden representations of preceding tokens is incorporated into the transformer block.\nTransformer-based language models are complex, and so the details will unfold over this chapter and the next few chapters. Chapter 7 already discussed how language models are pretrained, and how tokens are generated via sampling. In the rest of this chapter we'll introduce multi-head attention, the rest of the transformer block, and the input encoding and language modeling head components of the transformer. Chapter 9 introduces masked language modeling and the BERT family of bidirectional transformer encoder models. Chapter 10 shows how to instruction-tune language models to perform NLP tasks, and how to align the model with human preferences. Chapter 12 will introduce machine translation with the encoder-decoder architecture. And we'll see application of the transformer to speech recognition, as well as further use of the encoder-decoder architecture, in Chapter 15.", + "path": "8.pdf/CHAPTER 8 Transformers", + "metadata": { + "length": 5592, + "summary": "This chapter introduces the transformer architecture, the standard for building large language models. It focuses on left-to-right (causal) autoregressive language modeling, where input tokens are processed to predict output tokens sequentially. The core components include token embedding, positional encoding, multi-head self-attention, feedforward networks, layer normalization, and a language modeling head (unembedding matrix and softmax). The residual stream allows information to flow through N transformer blocks, with attention enabling contextual token representations by integrating information from surrounding tokens. The chapter outlines subsequent chapters covering masked language modeling (BERT), instruction tuning, machine translation, and speech recognition applications.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "Transformers", + "The", + "true", + "art", + "memory", + "attention", + "Samuel", + "Johnson", + "Idler", + "74", + "September", + "1759", + "In", + "chapter", + "introduce", + "transformer", + "standard", + "architecture", + "building", + "large", + "language", + "models", + "As", + "discussed", + "prior", + "based", + "completely", + "changed", + "field", + "speech", + "processing", + "Indeed", + "subsequent", + "textbook", + "make", + "previous", + "ll", + "focus", + "transformers", + "model", + "left", + "called", + "causal", + "autoregressive", + "modeling", + "sequence", + "input", + "tokens", + "predict", + "output", + "conditioning", + "context", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "TD", + "token", + "Embedding", + "Positional", + "MultiHead", + "Attention", + "Layer", + "Norm", + "Feedforward", + "Unembedding", + "Softmax", + "probabilities", + "style", + "fill", + "f9f", + "stroke", + "333", + "ccf", + "cfc", + "fcc", + "cff", + "ffc", + "subgraph", + "times", + "residual", + "stream", + "Input", + "end", + "Figure", + "8.1", + "decoder", + "showing", + "single", + "embedded", + "passed", + "forward", + "network", + "feedforward", + "components", + "adding", + "information", + "multihead", + "layer", + "takes", + "inputs", + "shown", + "detail", + "neighboring", + "streams", + "This", + "column", + "taking", + "outputting", + "distribution", + "Fig", + "sketches", + "passes", + "layers", + "Each", + "converted", + "embedding", + "matrix", + "Recall", + "Chapter", + "Section", + "linear", + "maps", + "id", + "vector", + "representing", + "vocabulary", + "initial", + "representation", + "special", + "mechanism", + "encoding", + "position", + "index", + "string", + "simply", + "added", + "resulting", + "represents", + "word", + "set", + "blocks", + "It", + "common", + "part", + "directly", + "simultaneously", + "enriched", + "application", + "modules", + "multi", + "head", + "networks", + "normalization", + "sum", + "original", + "outputs", + "core", + "intuition", + "component", + "distinguishes", + "thought", + "build", + "contextual", + "representations", + "meaning", + "attending", + "integrating", + "surrounding", + "helping", + "learn", + "relate", + "spans", + "move", + "augmenting", + "After", + "produced", + "final", + "block", + "pass", + "unembedding", + "mathbf", + "softmax", + "generate", + "These", + "rest", + "chatper", + "shows", + "applied", + "window", + "words", + "So", + "long", + "generated", + "full", + "figure", + "entire", + "vectors", + "length", + "12", + "96", + "stacked", + "arrows", + "show", + "hidden", + "preceding", + "incorporated", + "Transformer", + "complex", + "unfold", + "chapters", + "pretrained", + "sampling", + "introduces", + "masked", + "BERT", + "family", + "bidirectional", + "encoder", + "10", + "instruction", + "tune", + "perform", + "NLP", + "tasks", + "align", + "human", + "preferences", + "machine", + "translation", + "And", + "recognition", + "15" + ], + "keywords": [ + "Transformer", + "Attention", + "Language Model" + ], + "connect_to": [ + { + "target": "babc058d-28ee-535b-9ffe-892d38458e93", + "relation": "embeds", + "ref": "[images/image-1-Transformer Encoder Architecture Diagram.jpg]", + "position": { + "start": 744, + "end": 805 + } + } + ] + } + }, + { + "chunk_id": "57935e48-86ad-5e38-93ff-8cfc4558d833", + "type": "image", + "content": "\nAfter the $N$ transformer blocks we take the output embedding that is produced by the final transformer block, pass it through an linear unembedding matrix $\\mathbf{U}$ and then a softmax over the vocabulary to generate a distribution over possible next tokens. These last two components (the unembedding matrix and the softmax) are sometimes called the language modeling head. In the rest of this chatper we'll introduce attention and the rest of these modules in more detail.
0\n[images/image-2-Transformer Architecture Diagram.jpg]\n", + "path": "images/image-2-Transformer Architecture Diagram.jpg", + "metadata": { + "length": 522, + "summary": "image-2\n- **Input Tokens**: Sequence \"So\", \"long\", \"and\", \"thanks\", \"for\" (x1-x5)\n- **Input Encoding**: Embedding layer (E) with positional encoding (1-5)\n- **Stacked Transformer Blocks**: Multiple layers of processing blocks with internal connections\n- **Language Modeling Head**: Output layer (U) producing logits for next token prediction\n- **Next Token Prediction**: Predicts \"long\", \"and\", \"thanks\", \"for\", \"all\" sequentially", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-2-Transformer Architecture Diagram.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "add5d859-e260-5ee7-9d3b-e43e88cb3bf9", + "type": "text", + "content": "Recall from Chapter 5 that for word2vec and other static embeddings, the representation of a word's meaning is always the same vector irrespective of the context: the word chicken, for example, is always represented by the same fixed vector. So a static vector for the word it might somehow encode that this is a pronoun used for animals and inanimate entities. But in context it has a much richer meaning. Consider it in one of these two sentences:\n\nAfter the $N$ transformer blocks we take the output embedding that is produced by the final transformer block, pass it through an linear unembedding matrix $\\mathbf{U}$ and then a softmax over the vocabulary to generate a distribution over possible next tokens. These last two components (the unembedding matrix and the softmax) are sometimes called the language modeling head. In the rest of this chatper we'll introduce attention and the rest of these modules in more detail.
0\n[images/image-2-Transformer Architecture Diagram.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nsubgraph Input Tokens\nA1[\"Input encoding So\"] --> X1[\"x1\"]\nA2[\"Input encoding x2\"] --> X2[\"x2\"]\nA3[\"Input encoding x3\"] --> X3[\"x3\"]\nA4[\"Input encoding x4\"] --> X4[\"x4\"]\nA5[\"Input encoding x5\"] --> X5[\"x5\"]\nend\nsubgraph Stacked Transformer Blocks\nB1[\"Stacked Transformer Blocks x1\"] --> X1\nB2[\"Stacked Transformer Blocks x2\"] --> X2\nB3[\"Stacked Transformer Blocks x3\"] --> X3\nB4[\"Stacked Transformer Blocks x4\"] --> X4\nB5[\"Stacked Transformer Blocks x5\"] --> X5\nend\nsubgraph Next Token\nC1[\"Language Modeling Head long\"] --> U1[\"logits U\"]\nC2[\"Language Modeling Head and\"] --> U2[\"logits U\"]\nC3[\"Language Modeling Head and thanks\"] --> U3[\"logits U\"]\nC4[\"Language Modeling Head and thanks for all\"] --> U4[\"logits U\"]\nend\nU1 --> X1\nU2 --> X2\nU3 --> X3\nU4 --> X4\nU5 --> X5\nX1 --> Y1[\"E\"]\nX2 --> Y2[\"E\"]\nX3 --> Y3[\"E\"]\nX4 --> Y4[\"E\"]\nX5 --> Y5[\"E\"]\nstyle Input Tokens fill:#f9f,stroke:#333\nstyle Next Token fill:#ccf,stroke:#333\n```\n
\nFigure 8.2 The architecture of a (left-to-right) transformer, showing how each input token get encoded, passed through a set of stacked transformer blocks, and then a language model head that predicts the next token. The embeddings at each token position in the residual stream are passed up the stack, and the arrows in the figure shows how information from the hidden representations of preceding tokens are also incorporated.\n(8.1) The chicken didn't cross the road because it was too tired.\n(8.2) The chicken didn't cross the road because it was too wide.\nIn (8.1) it is the chicken (i.e., the reader knows that the chicken was tired), while in (8.2) it is the road (and the reader knows that the road was wide). $^{1}$ That is, if we are to compute the meaning of this sentence, we'll need the meaning of it to be associated with the chicken in the first sentence and associated with the road in the second one, sensitive to the context.\nFurthermore, consider reading left to right like a causal language model, processing the sentence up to the word it:", + "path": "8.pdf/CHAPTER 8 Transformers/8.1 Attention", + "metadata": { + "length": 3032, + "summary": "The passage discusses how static embeddings like word2vec represent words with fixed vectors regardless of context, using 'it' as an example. In contrast, transformer models use context to resolve ambiguity, as shown in sentences where 'it' refers to 'chicken' or 'road' depending on context. It also describes the transformer architecture: input tokens are encoded, passed through stacked transformer blocks, and then through a language modeling head (unembedding matrix and softmax) to predict the next token. A flowchart illustrates this process, showing how information from preceding tokens is incorporated.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "Recall", + "Chapter", + "word2vec", + "static", + "embeddings", + "representation", + "word", + "meaning", + "vector", + "irrespective", + "context", + "chicken", + "represented", + "fixed", + "So", + "encode", + "pronoun", + "animals", + "inanimate", + "entities", + "But", + "richer", + "Consider", + "sentences", + "After", + "transformer", + "blocks", + "output", + "embedding", + "produced", + "final", + "block", + "pass", + "linear", + "unembedding", + "matrix", + "mathbf", + "softmax", + "vocabulary", + "generate", + "distribution", + "tokens", + "These", + "components", + "called", + "language", + "modeling", + "head", + "In", + "rest", + "chatper", + "ll", + "introduce", + "attention", + "modules", + "detail", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "TD", + "subgraph", + "Input", + "Tokens", + "A1", + "encoding", + "X1", + "x1", + "A2", + "x2", + "X2", + "A3", + "x3", + "X3", + "A4", + "x4", + "X4", + "A5", + "x5", + "X5", + "end", + "Stacked", + "Transformer", + "Blocks", + "B1", + "B2", + "B3", + "B4", + "B5", + "Next", + "Token", + "C1", + "Language", + "Modeling", + "Head", + "long", + "U1", + "logits", + "C2", + "U2", + "C3", + "U3", + "C4", + "U4", + "U5", + "Y1", + "Y2", + "Y3", + "Y4", + "Y5", + "style", + "fill", + "f9f", + "stroke", + "333", + "ccf", + "Figure", + "8.2", + "The", + "architecture", + "left", + "showing", + "input", + "token", + "encoded", + "passed", + "set", + "stacked", + "model", + "predicts", + "position", + "residual", + "stream", + "stack", + "arrows", + "figure", + "shows", + "information", + "hidden", + "representations", + "preceding", + "incorporated", + "8.1", + "didn", + "cross", + "road", + "tired", + "wide", + "reader", + "That", + "compute", + "sentence", + "sensitive", + "Furthermore", + "reading", + "causal", + "processing" + ], + "keywords": [ + "transformer", + "context", + "embedding" + ], + "connect_to": [ + { + "target": "57935e48-86ad-5e38-93ff-8cfc4558d833", + "relation": "embeds", + "ref": "[images/image-2-Transformer Architecture Diagram.jpg]", + "position": { + "start": 940, + "end": 993 + } + } + ] + } + }, + { + "chunk_id": "59eddb21-9635-5244-8db6-08b2133942fa", + "type": "image", + "content": "\n(8.5) I walked along the pond, and noticed one of the trees along the bank.
0\n[images/image-3-Self-Attention Visualization.jpg]\n", + "path": "images/image-3-Self-Attention Visualization.jpg", + "metadata": { + "length": 120, + "summary": "image-3\nDiagram illustrating the self-attention mechanism between Layer k and Layer k+1.\n\nKey Components:\n- Input Sequence (Columns): \"The chicken didn't cross the road because it was too tired\"\n- Layers: Two distinct layers labeled \"Layer k\" (bottom) and \"Layer k+1\" (top).\n- Focus Token: The word \"it\" in Layer k+1 is highlighted as the target of attention.\n\nRelationships & Attention Distribution:\n- The diagram visualizes how the token \"it\" at Layer k+1 attends to specific tokens from Layer k.\n- Strongest Connection: A thick blue line connects \"it\" (Layer k+1) to \"chicken\" (Layer k), indicating the highest attention weight. This demonstrates that the model associates \"it\" with \"chicken\".\n- Secondary Connections: Thinner lines connect \"it\" (Layer k+1) to \"road\" and \"because\" in Layer k, showing lower but present attention weights.\n- Visual Representation: The thickness of the connecting lines corresponds to the strength of the self-attention distribution.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-3-Self-Attention Visualization.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "90ff64c4-8c9c-5379-90b1-5a28aec57a20", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
LayerTokenToken Description
Layer k+1TheThe
Layer k+1chickenchicken
Layer k+1didn'tdidn't
Layer k+1crosscross
Layer k+1thethe
Layer k+1roadroad
Layer k+1becausebecause
Layer k+1itit
self-attention distributionthethe
self-attention distributionchickenchicken
self-attention distributiondidn'tdidn't
self-attention distributioncrosscross
self-attention distributionthethe
self-attention distributionroadroad
self-attention distributionbecausebecause
self-attention distributionitit
self-attention distributionwaswas
self-attention distributiontootoo
self-attention distributiontiredtired
", + "path": "tables/table-0 Self-Attention.html", + "metadata": { + "length": 2020, + "summary": "table-1\nThe table shows tokens from a sentence about a chicken not crossing a road, with self-attention distribution layers listing each token.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "tables/table-0 Self-Attention.html", + "keywords": [ + "self-attention", + "tokens", + "distribution" + ], + "tokens": [] + } + }, + { + "chunk_id": "435936fb-3270-58ee-b0b6-2f96fad6e361", + "type": "image", + "content": "\nLet's now turn to how this attention distribution is represented and computed.
0\n[images/image-4-Self-Attention Layer Diagram.jpg]\n", + "path": "images/image-4-Self-Attention Layer Diagram.jpg", + "metadata": { + "length": 123, + "summary": "image-4\n- Component: Self-Attention Layer\n- Input nodes: x1, x2, x3, x4, x5\n- Output nodes: a1, a2, a3, a4, a5\n- Internal structure: Contains five parallel \"attention\" blocks.\n- Connectivity: Each input node (x) connects to every attention block, indicating full connectivity or global context processing within the layer.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-4-Self-Attention Layer Diagram.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f13fb6d3-b2e3-5788-b20c-20e9196b0929", + "type": "image", + "content": "\n$$
0\n[images/image-5-Self-Attention Mechanism Process.jpg]\n", + "path": "images/image-5-Self-Attention Mechanism Process.jpg", + "metadata": { + "length": 47, + "summary": "image-5\n1. Generate key, query, value vectors from inputs X1, X2, and X3 (dimension [1 x d]) using weight matrices W^K, W^Q, and W^V.\n2. Compare the query vector of x3 with the key vectors for x1, x2, and x3.\n3. Divide scalar scores by the square root of dimension K (√d_k).\n4. Turn scores into attention weights (α_i,j) via softmax.\n5. Weigh each value vector based on the calculated attention weights (e.g., α_3,1, α_3,2, α_3,3).\n6. Sum the weighted value vectors to produce a combined vector of dimension [1 x d_v].\n7. Reshape the result to [1 x d].\n8. Output of self-attention (a_3), obtained by multiplying with weight matrix W^O (dimension [d_v x d]).", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-5-Self-Attention Mechanism Process.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d5d65d23-91b3-5928-bd1e-4adec5dedbd0", + "type": "text", + "content": "At this point we don't yet know which thing it is going to end up referring to! So a representation of it at this point might have aspects of both chicken and road as the reader is trying to guess what happens next.\nThis fact that words have rich linguistic relationships with other words that may be far away pervades language. Consider two more examples:\n(8.4) The keys to the cabinet are on the table.\n(8.5) I walked along the pond, and noticed one of the trees along the bank.\nIn (8.4), the phrase The keys is the subject of the sentence, and in English and many languages, must agree in grammatical number with the verb are; in this case both are plural. In English we can't use a singular verb like is with a plural subject like keys (we'll discuss agreement more in Chapter 18). In (8.5), we know that bank refers to the side of a pond or river and not a financial institution because of the context, including words like pond. (We'll discuss word senses more in Chapter 9.)\ncontextual embeddings\nThe point of all these examples is that these contextual words that help us compute the meaning of words in context can be quite far away in the sentence or paragraph. Transformers can build contextual representations of word meaning, contextual embeddings, by integrating the meaning of these helpful contextual words. In a transformer, layer by layer, we build up richer and richer contextualized representations of the meanings of input tokens. At each layer, we compute the representation of a token i by combining information about i from the previous layer with information about the neighboring tokens to produce a contextualized representation for each word at each position.\nAttention is the mechanism in the transformer that weighs and combines the representations from appropriate other tokens in the context from layer k to build the representation for tokens in layer $k + 1$ .\n\n(8.5) I walked along the pond, and noticed one of the trees along the bank.
0\n[images/image-3-Self-Attention Visualization.jpg]\n\n
\nbar chart\ncolumns corresponding to input tokens\n\n[tables/table-0 Self-Attention.html]\n\ncolumns corresponding to input tokens\n
\nFigure 8.3 The self-attention weight distribution $\\alpha$ that is part of the computation of the representation for the word $it$ at layer $k + 1$ . In computing the representation for $it$ , we attend differently to the various words at layer $k$ , with darker shades indicating higher self-attention values. Note that the transformer is attending highly to the columns corresponding to the tokens chicken and road, a sensible result, since at the point where $it$ occurs, it could plausibly corefer with the chicken or the road, and hence we'd like the representation for $it$ to draw on the representation for these earlier words. Figure adapted from Uszkoreit (2017).\nFig. 8.3 shows a schematic example simplified from a transformer (Uszkoreit, 2017). The figure describes the situation when the current token is it and we need to compute a contextual representation for this token at layer $k+1$ of the transformer, drawing on the representations (from layer k) of every prior token. The figure uses color to represent the attention distribution over the contextual words: the tokens chicken and road both have a high attention weight, meaning that as we are computing the representation for it, we will draw most heavily on the representation for chicken and road. This will be useful in building the final representation for it, since it will end up coreferring with either chicken or road.\nLet's now turn to how this attention distribution is represented and computed.\n8.1.1 Attention more formally\nAs we've said, the attention computation is a way to compute a vector representation for a token at a particular layer of a transformer, by selectively attending to and integrating information from prior tokens at the previous layer. Attention takes an input representation $x_{i}$ corresponding to the input token at position i, and a context window of prior inputs $x_{1}..x_{i-1}$ , and produces an output $a_{i}$ .\nIn causal, left-to-right language models, the context is any of the prior words. That is, when processing $x_{i}$ , the model has access to $x_{i}$ as well as the representations of all the prior tokens in the context window (context windows consist of thousands of tokens) but no tokens after i. (By contrast, in Chapter 9 we’ll generalize attention so it can also look ahead to future words.)\nFig. 8.4 illustrates this flow of information in an entire causal self-attention layer, in which this same attention computation happens in parallel at each token position i. Thus a self-attention layer maps input sequences $(\\mathbf{x}_{1},..., \\mathbf{x}_{n})$ to output sequences of the same length $(\\mathbf{a}_{1},..., \\mathbf{a}_{n})$ .\n\nLet's now turn to how this attention distribution is represented and computed.
0\n[images/image-4-Self-Attention Layer Diagram.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\na1[\" a₁ \"] --> A1[\"attention\"]\na2[\" a₂ \"] --> A2[\"attention\"]\na3[\" a₃ \"] --> A3[\"attention\"]\na4[\" a₄ \"] --> A4[\"attention\"]\na5[\" a₅ \"] --> A5[\"attention\"]\nx1[\" x₁ \"] --> A1\nx2[\" x₂ \"] --> A2\nx3[\" x₃ \"] --> A3\nx4[\" x₄ \"] --> A4\nx5[\" x₅ \"] --> A5\nA1 --> x1\nA2 --> x2\nA3 --> x3\nA4 --> x4\nA5 --> x5\n```\n
\nFigure 8.4 Information flow in causal self-attention. When processing each input $x_{i}$ , the model attends to all the inputs up to, and including $x_{i}$ .\nSimplified version of attention At its heart, attention is really just a weighted sum of context vectors, with a lot of complications added to how the weights are computed and what gets summed. For pedagogical purposes let's first describe a simplified intuition of attention, in which the attention output $\\mathbf{a}_i$ at token position $i$ is simply the weighted sum of all the representations $\\mathbf{x}_j$ , for all $j \\leq i$ ; we'll use $\\alpha_{ij}$ to mean how much $\\mathbf{x}_j$ should contribute to $\\mathbf{a}_i$ :\n$$\n\\text { Simplified version: } \\quad \\mathbf {a} _ {i} = \\sum_ {j \\leq i} \\alpha_ {i j} \\mathbf {x} _ {j} \\tag {8.6}\n$$\nEach $\\alpha_{ij}$ is a scalar used for weighing the value of input $\\mathbf{x}_j$ when summing up the inputs to compute $\\mathbf{a}_i$ . How shall we compute this $\\alpha$ weighting? In attention we weight each prior embedding proportionally to how similar it is to the current token $i$ . So the output of attention is a sum of the embeddings of prior tokens weighted by their similarity with the current token embedding. We compute similarity scores via dot product, which maps two vectors into a scalar value ranging from $-\\infty$ to $\\infty$ . The larger the score, the more similar the vectors that are being compared. We'll normalize these scores with a softmax to create the vector of weights $\\alpha_{ij}, j \\leq i$ .\n$$\n\\text { Simplified Version: } \\quad \\operatorname{score} (\\mathbf {x} _ {i}, \\mathbf {x} _ {j}) = \\mathbf {x} _ {i} \\cdot \\mathbf {x} _ {j} \\tag {8.7}\n$$\n$$\n\\alpha_ {i j} = \\text { softmax } (\\text { score } (\\mathbf {x} _ {i}, \\mathbf {x} _ {j})) \\forall j \\leq i \\tag {8.8}\n$$\nThus in Fig. 8.4 we compute $a_{3}$ by computing three scores: $x_{3} \\cdot x_{1}$ , $x_{3} \\cdot x_{2}$ and $x_{3} \\cdot x_{3}$ , normalizing them by a softmax, and using the resulting probabilities as weights indicating each of their proportional relevance to the current position i. Of course, the softmax weight will likely be highest for $x_{i}$ , since $x_{i}$ is very similar to itself, resulting in a high dot product. But other context words may also be similar to i, and the softmax will also assign some weight to those words. Then we use these weights as the $\\alpha$ values in Eq. 8.6 to compute the weighted sum that is our $a_{3}$ .\nThe simplified attention in equations 8.6 - 8.8 demonstrates the attention-based approach to computing $\\mathbf{a}_i$ : compare the $\\mathbf{x}_i$ to prior vectors, normalize those scores into a probability distribution used to weight the sum of the prior vectors. But now we're ready to remove the simplifications.\nattention head\nhead\nA single attention head using query, key, and value matrices Now that we've seen a simple intuition of attention, let's introduce the actual attention head, the version of attention that's used in transformers. (The word head is often used in transformers to refer to specific structured layers). The attention head allows us to distinctly represent three different roles that each input embedding plays during the course of the attention process:\nquery\nkey\nvalue\n- As the current element being compared to the preceding inputs. We'll refer to this role as a query.\n- In its role as a preceding input that is being compared to the current element to determine a similarity weight. We'll refer to this role as a key.\n- And finally, as a value of a preceding element that gets weighted and summed up to compute the output for the current element.\nTo capture these three different roles, transformers introduce weight matrices $W^{Q}$ , $W^{K}$ , and $W^{V}$ . These weights will project each input vector $x_{i}$ into a representation of its role as a query, key, or value:\n$$\n\\mathbf {q} _ {i} = \\mathbf {x} _ {i} \\mathbf {W} ^ {\\mathrm{Q}}; \\quad \\mathbf {k} _ {i} = \\mathbf {x} _ {i} \\mathbf {W} ^ {\\mathrm{K}}; \\quad \\mathbf {v} _ {i} = \\mathbf {x} _ {i} \\mathbf {W} ^ {\\mathrm{V}} \\tag {8.9}\n$$\nGiven these projections, when we are computing the similarity of the current element $\\mathbf{x}_i$ with some prior element $\\mathbf{x}_j$ , we'll use the dot product between the current element's query vector $\\mathbf{q}_i$ and the preceding element's key vector $\\mathbf{k}_j$ . Furthermore, the result of a dot product can be an arbitrarily large (positive or negative) value, and exponentiating large values can lead to numerical issues and loss of gradients during training. To avoid this, we scale the dot product by a factor related to the size of the embeddings, via dividing by the square root of the dimensionality of the query and key vectors $(d_k)$ . We thus replace the simplified Eq. 8.7 with Eq. 8.11. The ensuing softmax calculation resulting in $\\alpha_{ij}$ remains the same, but the output calculation for $\\textbf{head}_i$ is now based on a weighted sum over the value vectors $\\mathbf{v}$ (Eq. 8.13).\nHere's a final set of equations for computing self-attention for a single self-attention output vector $\\mathbf{a}_i$ from a single input vector $\\mathbf{x}_i$ . This version of attention computes $\\mathbf{a}_i$ by summing the values of the prior elements, each weighted by the similarity of its key to the query from the current element:\n$$\n\\mathbf {q} _ {i} = \\mathbf {x} _ {i} \\mathbf {W} ^ {\\mathrm{Q}}; \\quad \\mathbf {k} _ {j} = \\mathbf {x} _ {j} \\mathbf {W} ^ {\\mathrm{K}}; \\quad \\mathbf {v} _ {j} = \\mathbf {x} _ {j} \\mathbf {W} ^ {\\mathrm{V}} \\tag {8.10}\n$$\n$$\n\\operatorname{score} (\\mathbf {x} _ {i}, \\mathbf {x} _ {j}) = \\frac {\\mathbf {q} _ {i} \\cdot \\mathbf {k} _ {j}}{\\sqrt {d _ {k}}} \\tag {8.11}\n$$\n$$\n\\alpha_ {i j} = \\text { softmax } (\\text { score } (\\mathbf {x} _ {i}, \\mathbf {x} _ {j})) \\forall j \\leq i \\tag {8.12}\n$$\n$$\n\\mathbf {h e a d} _ {i} = \\sum_ {j \\leq i} \\alpha_ {i j} \\mathbf {v} _ {j} \\tag {8.13}\n$$\n$$\n\\mathbf {a} _ {i} = \\text { head } _ {i} \\mathbf {W} ^ {\\mathrm{O}} \\tag {8.14}\n$$\nWe illustrate this in Fig. 8.5 for the case of calculating the value of the third output $a_{3}$ in a sequence.\nNote that we've also introduced one more matrix, $\\mathbf{W}_{\\mathbf{O}}$ , which is left-multiplied by the attention head. This is necessary to reshape the output of the head. The input to attention $\\mathbf{x}_{\\mathbf{i}}$ and the output from attention $\\mathbf{a}_{\\mathbf{i}}$ both have the same dimensionality $[1\\times d]$ . We often call $d$ the model dimensionality, and indeed as we'll discuss in\n\n$$
0\n[images/image-5-Self-Attention Mechanism Process.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"1 x d\"] --> B[\"×1 × dv\"]\nC[\"2 x3's query with the keys for x1, x2, and x3\"] --> D[\"×1 × dv\"]\nE[\"3 x3\"] --> F[\"×1 × dv\"]\nG[\"4 x3\"] --> H[\"×1 × dv\"]\nI[\"5 x3\"] --> J[\"×1 × dv\"]\nK[\"6 x3\"] --> L[\"×1 × dv\"]\nM[\"7 x3\"] --> N[\"×1 × dv\"]\nO[\"8 x3\"] --> P[\"Output of self-attention a₃ [1 × d\"]]\nQ[\"9 x3\"] --> R[\"W⁰ [dv × d\"] [1 × dv]]\nS[\"10 x3\"] --> T[\"Sum the weighted value vectors [1 × dv\"]]\nU[\"11 x3\"] --> V[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nW[\"12 x3\"] --> X[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nY[\"13 x3\"] --> Z[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAA[\"14 x3\"] --> AB[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAC[\"15 x3\"] --> AD[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAE[\"16 x3\"] --> AF[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAG[\"17 x3\"] --> AH[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAI[\"18 x3\"] --> AJ[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAK[\"19 x3\"] --> AL[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAM[\"20 x3\"] --> AN[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAO[\"21 x3\"] --> AP[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAQ[\"22 x3\"] --> AR[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAS[\"23 x3\"] --> AT[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAU[\"24 x3\"] --> AV[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAW[\"25 x3\"] --> AX[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nAY[\"26 x3\"] --> AZ[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nBA[\"27 x3\"] --> BB[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nBC[\"28 x3\"] --> BD[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nBE[\"29 x3\"] --> BF[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nBG[\"30 x3\"] --> BH[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nBI[\"31 x3\"] --> BJ[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nBK[\"32 x3\"] --> BL[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nBM[\"33 x3\"] --> BN[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nBO[\"34 x3\"] --> BP[\"α₃,₁ → α₃,₂ → α₃,₃\"]\nBP --> BQ[\"X1: k q v W^K W^Q W^V X1: [1 × d\"]]\nBX[\"X2: k q v W^K W^Q W^V X2: [1 × d\"]]\nBY[\"X3: k q v W^K W^Q W^V X3: [1 × d\"]]\n```\n
\nFigure 8.5 Calculating the value of $\\mathbf{a}_3$ , the third element of a sequence using causal (left-to-right) self-attention.\nSection 8.2 the output $h_{i}$ of each transformer block, as well as the intermediate vectors inside the transformer block also have the same dimensionality $[1 \\times d]$ . Having everything be the same dimensionality makes the transformer very modular.\nSo let's talk shapes. How do we get from $[1 \\times d]$ at the input to $[1 \\times d]$ at the output? Let's look at all the internal shapes. We'll have a dimension $d_k$ for the query and key vectors. The query vector and the key vector are both dimensionality $[1 \\times d_k]$ , so we can take their dot product $\\mathbf{q}_i \\cdot \\mathbf{k}_j$ to produce a scalar. We'll have a separate dimension $d_v$ for the value vectors. The transform matrix $\\mathbf{W}^{\\mathbf{Q}}$ has shape $[d \\times d_k]$ , $\\mathbf{W}^{\\mathbf{K}}$ is $[d \\times d_k]$ , and $\\mathbf{W}^{\\mathbf{V}}$ is $[d \\times d_v]$ . So the output of $\\mathbf{head}_i$ in equation Eq. 8.13 is of shape $[1 \\times d_v]$ . To get the desired output shape $[1 \\times d]$ we'll need to reshape the head output, and so $\\mathbf{W}^{\\mathbf{O}}$ is of shape $[d_v \\times d]$ . In the original transformer work (Vaswani et al., 2017), $d$ was 512, $d_k$ and $d_v$ were both 64.\nMulti-head Attention Equations 8.11-8.13 describe a single attention head. But actually, transformers use multiple attention heads. The intuition is that each head might be attending to the context for different purposes: heads might be specialized to represent different linguistic relationships between context elements and the current token, or to look for particular kinds of patterns in the context.\nSo in multi-head attention we have A separate attention heads that reside in parallel layers at the same depth in a model, each with its own set of parameters that allows the head to model different aspects of the relationships among inputs. Thus each head i in a self-attention layer has its own set of query, key, and value matrices: $W^{Qi}$ , $W^{Ki}$ , and $W^{Vi}$ . These are used to project the inputs into separate query, key, and value embeddings for each head.\nWhen using multiple heads the model dimension d is still used for the input and output, the query and key embeddings have dimensionality $d_{k}$ , and the value embeddings are of dimensionality $d_{v}$ (again, in the original transformer paper $d_{k}=$\nmulti-head attention\n$d_{v}=64, A=8,$ and $d=512$ ). Thus for each head $i$ , we have weight layers $\\mathbf{W}^{\\mathbf{Q}\\mathbf{i}}$ of shape $[d\\times d_{k}]$ , $\\mathbf{W}^{\\mathbf{K}\\mathbf{i}}$ of shape $[d\\times d_{k}]$ , and $\\mathbf{W}^{\\mathbf{V}\\mathbf{i}}$ of shape $[d\\times d_{v}]$ .\nBelow are the equations for attention augmented with multiple heads; Fig. 8.6 shows an intuition.\n$$\n\\mathbf {q} _ {i} ^ {c} = \\mathbf {x} _ {i} \\mathbf {W} ^ {\\mathbf {Q c}}; \\quad \\mathbf {k} _ {j} ^ {c} = \\mathbf {x} _ {j} \\mathbf {W} ^ {\\mathbf {K c}}; \\quad \\mathbf {v} _ {j} ^ {c} = \\mathbf {x} _ {j} \\mathbf {W} ^ {\\mathbf {V c}}; \\quad \\forall c 1 \\leq c \\leq A \\tag {8.15}\n$$\n$$\n\\operatorname{score} ^ {c} (\\mathbf {x} _ {i}, \\mathbf {x} _ {j}) = \\frac {\\mathbf {q} _ {i} ^ {c} \\cdot \\mathbf {k} _ {j} ^ {c}}{\\sqrt {d _ {k}}} \\tag {8.16}\n$$\n$$\n\\alpha_ {i j} ^ {c} = \\text { softmax } (\\text { score } ^ {c} (\\mathbf {x} _ {i}, \\mathbf {x} _ {j})) \\forall j \\leq i \\tag {8.17}\n$$\n$$\n\\mathbf {h e a d} _ {i} ^ {c} = \\sum_ {j \\leq i} \\alpha_ {i j} ^ {c} \\mathbf {v} _ {j} ^ {c} \\tag {8.18}\n$$\n$$\n\\mathbf {a} _ {i} = \\left(\\mathbf {h e a d} ^ {1} \\oplus \\mathbf {h e a d} ^ {2} \\dots \\oplus \\mathbf {h e a d} ^ {A}\\right) \\mathbf {W} ^ {O} \\tag {8.19}\n$$\n$$\n\\text { MultiHeadAttention } (\\mathbf {x} _ {i}, [ \\mathbf {x} _ {1}, \\dots , \\mathbf {x} _ {i - 1} ]) = \\mathbf {a} _ {i} \\tag {8.20}\n$$\nNote in Eq. 8.20 that MultiHeadAttention is a function of the current input $\\mathbf{x}_i$ , as well as all the other inputs. For the causal or left-to-right attention that we use in this chapter, the other inputs are only to the left, but we'll also see a version of attention in Chapter 9 where attention is a function of the tokens to the right as well. We'll return to this idea about causal inputs in Eq. 8.34 when we introduce the idea of masking the right context.\nThe output of each of the A heads is of shape $[1 \\times d_{v}]$ , and so the output of the multi-head layer with A heads consists of A vectors of shape $[1 \\times d_{v}]$ . These are concatenated to produce a single output with dimensionality $[1 \\times Ad_{v}]$ . Then we use yet another linear projection $W^{0} \\in R^{Ad_{v} \\times d}$ to reshape it, resulting in the multi-head attention vector $a_{i}$ with the correct output shape $[1 \\times d]$ at each input i.", + "path": "8.pdf/CHAPTER 8 Transformers/(8.3) The chicken didn't cross the road because it", + "metadata": { + "length": 18334, + "summary": "The passage explains how transformers use attention mechanisms to build contextual representations of word meaning. It describes how attention computes a weighted sum of context vectors, with weights determined by similarity scores between query and key vectors. The text covers both simplified and formal versions of attention, including single-head and multi-head attention, and explains the roles of query, key, and value matrices. It also discusses causal (left-to-right) attention used in language models.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "At", + "point", + "don", + "thing", + "end", + "referring", + "So", + "representation", + "aspects", + "chicken", + "road", + "reader", + "guess", + "This", + "fact", + "words", + "rich", + "linguistic", + "relationships", + "pervades", + "language", + "Consider", + "examples", + "8.4", + "The", + "keys", + "cabinet", + "table", + "8.5", + "walked", + "pond", + "noticed", + "trees", + "bank", + "In", + "phrase", + "subject", + "sentence", + "English", + "languages", + "agree", + "grammatical", + "number", + "verb", + "case", + "plural", + "singular", + "ll", + "discuss", + "agreement", + "Chapter", + "18", + "refers", + "side", + "river", + "financial", + "institution", + "context", + "including", + "We", + "word", + "senses", + "contextual", + "embeddings", + "compute", + "meaning", + "paragraph", + "Transformers", + "build", + "representations", + "integrating", + "helpful", + "transformer", + "layer", + "richer", + "contextualized", + "meanings", + "input", + "tokens", + "token", + "combining", + "information", + "previous", + "neighboring", + "produce", + "position", + "Attention", + "mechanism", + "weighs", + "combines", + "details", + "summary", + "bar", + "chart", + "columns", + "Figure", + "8.3", + "attention", + "weight", + "distribution", + "alpha", + "part", + "computation", + "computing", + "attend", + "differently", + "darker", + "shades", + "indicating", + "higher", + "values", + "Note", + "attending", + "highly", + "result", + "occurs", + "plausibly", + "corefer", + "draw", + "earlier", + "adapted", + "Uszkoreit", + "2017", + "Fig", + "shows", + "schematic", + "simplified", + "figure", + "describes", + "situation", + "current", + "drawing", + "prior", + "color", + "represent", + "high", + "heavily", + "building", + "final", + "coreferring", + "Let", + "turn", + "represented", + "computed", + "8.1", + "formally", + "As", + "ve", + "vector", + "selectively", + "takes", + "window", + "inputs", + "produces", + "output", + "causal", + "left", + "models", + "That", + "processing", + "model", + "access", + "windows", + "consist", + "thousands", + "By", + "contrast", + "generalize", + "ahead", + "future", + "illustrates", + "flow", + "entire", + "parallel", + "Thus", + "maps", + "sequences", + "mathbf", + "length", + "flowchart", + "mermaid", + "graph", + "TD", + "a1", + "A1", + "a2", + "A2", + "a3", + "A3", + "a4", + "A4", + "a5", + "A5", + "x1", + "x2", + "x3", + "x4", + "x5", + "Information", + "When", + "attends", + "Simplified", + "version", + "heart", + "weighted", + "sum", + "vectors", + "lot", + "complications", + "added", + "weights", + "summed", + "For", + "pedagogical", + "purposes", + "describe", + "intuition", + "simply", + "leq", + "ij", + "contribute", + "text", + "quad", + "tag", + "8.6", + "Each", + "scalar", + "weighing", + "summing", + "How", + "weighting", + "embedding", + "proportionally", + "similar", + "similarity", + "scores", + "dot", + "product", + "ranging", + "infty", + "larger", + "score", + "compared", + "normalize", + "softmax", + "create", + "Version", + "operatorname", + "cdot", + "8.7", + "forall", + "8.8", + "normalizing", + "resulting", + "probabilities", + "proportional", + "relevance", + "Of", + "highest", + "But", + "assign", + "Then", + "Eq", + "equations", + "demonstrates", + "based", + "approach", + "compare", + "probability", + "ready", + "remove", + "simplifications", + "head", + "single", + "query", + "key", + "matrices", + "Now", + "simple", + "introduce", + "actual", + "transformers", + "refer", + "specific", + "structured", + "layers", + "distinctly", + "roles", + "plays", + "process", + "element", + "preceding", + "role", + "determine", + "And", + "finally", + "To", + "capture", + "These", + "project", + "mathrm", + "8.9", + "Given", + "projections", + "Furthermore", + "arbitrarily", + "large", + "positive", + "negative", + "exponentiating", + "lead", + "numerical", + "issues", + "loss", + "gradients", + "training", + "avoid", + "scale", + "factor", + "related", + "size", + "dividing", + "square", + "root", + "dimensionality", + "replace", + "8.11", + "ensuing", + "calculation", + "remains", + "textbf", + "8.13", + "Here", + "set", + "computes", + "elements", + "8.10", + "frac", + "sqrt", + "8.12", + "8.14", + "illustrate", + "calculating", + "sequence", + "introduced", + "matrix", + "multiplied", + "reshape", + "times", + "call", + "dv", + "Output", + "10", + "Sum", + "11", + "12", + "13", + "AA", + "14", + "AB", + "AC", + "15", + "AD", + "AE", + "16", + "AF", + "AG", + "17", + "AH", + "AI", + "AJ", + "AK", + "19", + "AL", + "AM", + "20", + "AN", + "AO", + "21", + "AP", + "AQ", + "22", + "AR", + "AS", + "23", + "AT", + "AU", + "24", + "AV", + "AW", + "25", + "AX", + "AY", + "26", + "AZ", + "BA", + "27", + "BB", + "BC", + "28", + "BD", + "BE", + "29", + "BF", + "BG", + "30", + "BH", + "BI", + "31", + "BJ", + "BK", + "32", + "BL", + "BM", + "33", + "BN", + "BO", + "34", + "BP", + "BQ", + "X1", + "BX", + "X2", + "BY", + "X3", + "Calculating", + "Section", + "8.2", + "block", + "intermediate", + "inside", + "Having", + "makes", + "modular", + "talk", + "shapes", + "internal", + "dimension", + "separate", + "transform", + "shape", + "equation", + "desired", + "original", + "work", + "Vaswani", + "al", + "512", + "64", + "Multi", + "Equations", + "multiple", + "heads", + "specialized", + "kinds", + "patterns", + "multi", + "reside", + "depth", + "parameters", + "Qi", + "Ki", + "Vi", + "paper", + "Below", + "augmented", + "8.15", + "8.16", + "8.17", + "8.18", + "oplus", + "dots", + "8.19", + "MultiHeadAttention", + "8.20", + "function", + "chapter", + "return", + "idea", + "8.34", + "masking", + "consists", + "concatenated", + "Ad", + "linear", + "projection", + "correct" + ], + "keywords": [ + "attention", + "transformer", + "contextual embeddings" + ], + "connect_to": [ + { + "target": "59eddb21-9635-5244-8db6-08b2133942fa", + "relation": "embeds", + "ref": "[images/image-3-Self-Attention Visualization.jpg]", + "position": { + "start": 1983, + "end": 2032 + } + }, + { + "target": "90ff64c4-8c9c-5379-90b1-5a28aec57a20", + "relation": "embeds", + "ref": "[tables/table-0 Self-Attention.html]", + "position": { + "start": 2112, + "end": 2148 + } + }, + { + "target": "435936fb-3270-58ee-b0b6-2f96fad6e361", + "relation": "embeds", + "ref": "[images/image-4-Self-Attention Layer Diagram.jpg]", + "position": { + "start": 4955, + "end": 5004 + } + }, + { + "target": "f13fb6d3-b2e3-5788-b20c-20e9196b0929", + "relation": "embeds", + "ref": "[images/image-5-Self-Attention Mechanism Process.jpg]", + "position": { + "start": 11959, + "end": 12012 + } + } + ] + } + }, + { + "chunk_id": "770bece3-4834-5835-864b-56439a28e1c0", + "type": "image", + "content": "\nresidual stream
0\n[images/image-6-Multi-Head Attention Mechanism.jpg]\n", + "path": "images/image-6-Multi-Head Attention Mechanism.jpg", + "metadata": { + "length": 60, + "summary": "image-6\nThe diagram illustrates the architecture of a multi-head attention mechanism, likely within a Transformer model.\n\nKey components and flow:\n- Input: A sequence of vectors (represented as $x_{i-3}, x_{i-2}, x_{i-1}, X_i$) with dimension $[1 \\times d]$.\n- Parallel Heads: The input is processed by four parallel \"heads\" (Head 1 through Head 4). Each head applies its own set of weight matrices ($W^K, W^V, W^Q$) to attend differently to the context.\n- Output Dimensions: Each head produces an output vector of dimension $[1 \\times d_v]$.\n- Concatenation: The outputs from all heads are concatenated to form a combined vector of dimension $[1 \\times Ad_v]$, where $A=4$ in this example.\n- Final Projection: The concatenated output is projected to the final representation using a weight matrix $W^O$. This results in a final output vector of dimension $[1 \\times d]$.\n- Note on Dimensions: The diagram specifies that usually", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-6-Multi-Head Attention Mechanism.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e28cb543-3349-5d91-bd08-a8accb95486d", + "type": "image", + "content": "\nstyle C3 fill:#cfc,stroke:#333
0\n[images/image-7-Transformer Block Architecture.jpg]\n", + "path": "images/image-7-Transformer Block Architecture.jpg", + "metadata": { + "length": 75, + "summary": "image-7\n- Input sequence: x_i (with preceding steps x_{i-1}, x_{i-2})\n- Output sequence: h_i (with preceding steps h_{i-1}, h_{i-2})\n- Core components:\n - MultiHead Attention layer\n - Layer Norm layers (pre and post attention)\n - Feedforward network\n- Structural features:\n - Residual connections (indicated by green diamond \"+\" symbols)\n - Skip connections bypassing attention and feedforward blocks\n - Normalization applied before and after attention mechanisms", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-7-Transformer Block Architecture.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ef9fe40b-a3cd-5af2-a7ce-ef4883f409aa", + "type": "image", + "content": "\n$$
0\n[images/image-8-Diagram of Token Residual Streams.jpg]\n", + "path": "images/image-8-Diagram of Token Residual Streams.jpg", + "metadata": { + "length": 47, + "summary": "image-8\n- Illustrates the flow of residual streams for two tokens (Token A and Token B) across vertical lanes or channels.\n- Red arrows indicate the path of the residual stream:\n - Token A's stream starts in a lower lane and moves upward into an adjacent lane.\n - Token B's stream follows a similar trajectory, moving from one lane to another in an upward direction.\n- The diagram suggests a mechanism where token information is passed or shifted between different processing paths or layers.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-8-Diagram of Token Residual Streams.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "aaec2f26-726c-584e-b1f5-5e2afc732cde", + "type": "text", + "content": "residual stream\nThe self-attention calculation lies at the core of what's called a transformer block, which, in addition to the self-attention layer, includes three other kinds of layers: (1) a feedforward layer, (2) residual connections, and (3) normalizing layers (colloquially called “layer norm”).\nFig. 8.7 illustrates a transformer block, sketching a common way of thinking about the block that is called the residual stream (Elhage et al., 2021). In the residual stream viewpoint, we consider the processing of an individual token i through the transformer block as a single stream of d-dimensional representations for token position i. This residual stream starts with the original input vector, and the various components read their input from the residual stream and add their output back into the stream.\nThe input at the bottom of the stream is an embedding for a token, which has dimensionality d. This initial embedding gets passed up (by residual connections), and is progressively added to by the other components of the transformer: the attention layer that we have seen, and the feedforward layer that we will introduce. Before the attention and feedforward layer is a computation called the layer norm.\nThus the initial vector is passed through a layer norm and attention layer, and the result is added back into the stream, in this case to the original input vector $x_{i}$ . And then this summed vector is again passed through another layer norm and a feedforward layer, and the output of those is added back into the residual, and we'll use $\\mathbf{h}_i$ to refer to the resulting output of the transformer block for token $i$ .\n\nresidual stream
0\n[images/image-6-Multi-Head Attention Mechanism.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"Project to final representation W^O\"] --> B[\"Concatenate Outputs\"]\nB --> C1[\"Head 1 W^K1 W^V1 W^Q1\"]\nB --> C2[\"Head 2 W^K2 W^V2 W^Q2\"]\nB --> C3[\"Head 3 W^K3 W^V3 W^Q3\"]\nB --> C4[\"Head 4 W^K4 W^V4 W^Q4\"]\nC1 --> D1[\"x_{i-3} ... x_{i-2} ... x_{i-1}\"]\nC2 --> D2[\"x_{i-3} ... x_{i-2} ... x_{i-1}\"]\nC3 --> D3[\"x_{i-3} ... x_{i-2} ... x_{i-1}\"]\nC4 --> D4[\"x_{i-3} ... x_{i-2} ... x_{i-1}\"]\nstyle A fill:#f9f,stroke:#333\nstyle B fill:#ccf,stroke:#333\nstyle C1 fill:#cfc,stroke:#333\nstyle C2 fill:#cfc,stroke:#333\nstyle C3 fill:#cfc,stroke:#333\nstyle C4 fill:#cfc,stroke:#333\n```\n
\nFigure 8.6 The multi-head attention computation for input $x_{i}$ , producing output $a_{i}$ . A multi-head attention layer has A heads, each with its own query, key, and value weight matrices. In this figure, we show A = 4, a smaller value than is usually used, just to fit on the page. The outputs from each of the heads are of shape $[1 \\times d_{v}]$ and are concatenated and then projected into a different space by the $W_{0}$ matrix. Usually the dimensionality $d_{v}$ of the heads is set so that $d_{v} = d/A$ , with the result that $W_{0}$ is a square matrix of shape $[Ad_{v} \\times d] = [d \\times d]$ . usually of the same size, then projected d, thus producing an output of the same size as the input.\n\nstyle C3 fill:#cfc,stroke:#333
0\n[images/image-7-Transformer Block Architecture.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"x_{i-2}\"] --> B[\"MultiHead Attention\"]\nC[\"x_{i-1}\"] --> D[\"MultiHead Attention\"]\nE[\"x_i\"] --> F[\"MultiHead Attention\"]\nG[\"h_i\"] --> H[\"MultiHead Attention\"]\nI[\"h_i\"] --> J[\"MultiHead Attention\"]\nK[\"Feedforward\"] --> L[\"Layer Norm\"]\nM[\"...\"] --> N[\"...\"]\nO[\"Layer Norm\"] --> P[\"Layer Norm\"]\nQ[\"+\"] --> R[\"+\"]\nS[\"+\"] --> T[\"+\"]\nU[\"+\"] --> V[\"+\"]\nW[\"+\"] --> X[\"+\"]\nY[\"+\"] --> Z[\"+\"]\nstyle A fill:#f9f,stroke:#333\nstyle C fill:#f9f,stroke:#333\nstyle E fill:#f9f,stroke:#333\nstyle Q fill:#f9f,stroke:#333\nstyle M fill:#f9f,stroke:#333\nstyle O fill:#f9f,stroke:#333\nstyle R fill:#ccf,stroke:#333\nstyle S fill:#ccf,stroke:#333\nstyle T fill:#ccf,stroke:#333\nstyle U fill:#ccf,stroke:#333\nstyle V fill:#ccf,stroke:#333\nstyle W fill:#ccf,stroke:#333\nstyle X fill:#ccf,stroke:#333\n```\n
\nFigure 8.7 The architecture of a transformer block showing the residual stream, showing how most information flows up through the residual stream, and only the attention module is sensitive to information from other streams at prior token positions. In this figure and throughout the chapter, we use the prenorm version of the architecture, in which the layer norms happen before the attention and feedforward layers rather than after. The first\nWe've already seen the attention layer, so let's now introduce the feedforward and layer norm computations in the context of processing a single input $\\mathbf{x}_i$ at token\nposition i.\nFeedforward layer The feedforward layer is a fully-connected 2-layer network, i.e., one hidden layer, two weight matrices, as introduced in Chapter 6. The weights are the same for each token position i, but are different from layer to layer. It is common to make the dimensionality $d_{ff}$ of the hidden layer of the feedforward network be larger than the model dimensionality d. (For example in the original transformer model, d = 512 and $d_{ff} = 2048$ .)\n$$\n\\operatorname{FFN} \\left(\\mathbf {x} _ {i}\\right) = \\operatorname{ReLU} \\left(\\mathbf {x} _ {i} \\mathbf {W} _ {\\mathbf {1}} + b _ {1}\\right) \\mathbf {W} _ {\\mathbf {2}} + b _ {2} \\tag {8.21}\n$$\nlayer norm\nLayer Norm At two stages in the transformer block we normalize the vector (Ba et al., 2016). This process, called layer norm (short for layer normalization), is one of many forms of normalization that can be used to improve training performance in deep neural networks by keeping the values of a hidden layer in a range that facilitates gradient-based training.\nLayer norm is a variation of the z-score from statistics, applied to a single vector in a hidden layer. That is, the term layer norm is a bit confusing; layer norm is not applied to an entire transformer layer, but just to the embedding vector of a single token. Thus the input to layer norm is a single vector of dimensionality d and the output is that vector normalized, again of dimensionality d. The first step in layer normalization is to calculate the mean, $\\mu$ , and standard deviation, $\\sigma$ , over the elements of the vector to be normalized. Given an embedding vector x of dimensionality d, these values are calculated as follows.\n$$\n\\mu = \\frac {1}{d} \\sum_ {i = 1} ^ {d} x _ {i} \\tag {8.22}\n$$\n$$\n\\sigma = \\sqrt {\\frac {1}{d} \\sum_ {i = 1} ^ {d} (x _ {i} - \\mu) ^ {2}} \\tag {8.23}\n$$\nGiven these values, the vector components are normalized by subtracting the mean from each and dividing by the standard deviation. The result of this computation is a new vector with zero mean and a standard deviation of one.\n$$\n\\hat {\\mathbf {x}} = \\frac {(\\mathbf {x} - \\mu)}{\\sigma} \\tag {8.24}\n$$\nFinally, in the standard implementation of layer normalization, two learnable parameters, $\\gamma$ and $\\beta$ , representing gain and offset values, are introduced.\n$$\n\\text { LayerNorm } (\\mathbf {x}) = \\gamma \\frac {(\\mathbf {x} - \\mu)}{\\sigma} + \\beta \\tag {8.25}\n$$\nPutting it all together The function computed by a transformer block can be expressed by breaking it down with one equation for each component computation, using $\\mathbf{t}$ (of shape $[1\\times d]$ ) to stand for transformer and superscripts to demarcate each computation inside the block:\n$$\n\\mathbf {t} _ {i} ^ {\\mathbf {1}} = \\text { LayerNorm } (\\mathbf {x} _ {i}) \\tag {8.26}\n$$\n$$\n\\mathbf {t} _ {i} ^ {2} = \\text { MultiHeadAttention } (\\mathbf {t} _ {i} ^ {1}, [ \\mathbf {t} _ {1} ^ {1}, \\dots , \\mathbf {t} _ {N} ^ {1} ]) \\tag {8.27}\n$$\n$$\n\\mathbf {t} _ {i} ^ {3} = \\mathbf {t} _ {i} ^ {2} + \\mathbf {x} _ {i} \\tag {8.28}\n$$\n$$\n\\mathbf {t} _ {i} ^ {4} = \\text { LayerNorm } (\\mathbf {t} _ {i} ^ {3}) \\tag {8.29}\n$$\n$$\n\\mathbf {t} _ {i} ^ {\\mathbf {5}} = \\operatorname{FFN} (\\mathbf {t} _ {i} ^ {\\mathbf {4}}) \\tag {8.30}\n$$\n$$\n\\mathbf {h} _ {i} = \\mathbf {t} _ {i} ^ {5} + \\mathbf {t} _ {i} ^ {3} \\tag {8.31}\n$$\ntoken-mixing\nNotice that the only component that takes as input information from other tokens (other residual streams) is multi-head attention, which (as we see from Eq. 8.27) looks at all the neighboring tokens in the context. The output from attention, however, is then added into this token's embedding stream. In fact, Elhage et al. (2021) show that we can view attention heads as literally moving information from the residual stream of a neighboring token into the current stream. The high-dimensional embedding space at each position thus contains information about the current token and about neighboring tokens, albeit in different subspaces of the vector space. Fig. 8.8 shows a visualization of this movement. We therefore call the attention function the token-mixing component of the architecture, because it mixes information from neighboring token streams into the current stream.\n\n$$
0\n[images/image-8-Diagram of Token Residual Streams.jpg]\n\n
\ntext_image\nToken A\nresidual\nstream\nToken B\nresidual\nstream\n
\nFigure 8.8 An attention head can move information from token A's residual stream into token B's residual stream.\nCrucially, the input and output dimensions of transformer blocks are matched so they can be stacked. Each token vector $x_{i}$ at the input to the block has dimensionality d, and the output $h_{i}$ also has dimensionality d. Transformers for large language models stack many of these blocks, from 12 layers (used for the T5 or GPT-3-small language models) to 96 layers (used for GPT-3 large), to even more for more recent models. We’ll come back to this issue of stacking in a bit.\nEquation 8.26 and following are just the equation for a single transformer block, but the residual stream metaphor goes through all the transformer layers, from the first transformer blocks to the 12th, in a 12-layer transformer. At the earlier transformer blocks, the residual stream is representing the current token. At the highest transformer blocks, the residual stream is usually representing the following token, since at the very end it's being trained to predict the next token.\nOnce we stack many blocks, there is one more requirement: at the very end of the last (highest) transformer block, there is a single extra layer norm that is run on the last $h_{i}$ of each token stream (just below the language model head layer that we will define soon). $^{2}$", + "path": "8.pdf/CHAPTER 8 Transformers/8.2 Transformer Blocks", + "metadata": { + "length": 10353, + "summary": "The text describes the architecture of a transformer block, focusing on the residual stream concept. It explains that a transformer block includes self-attention, feedforward layers, residual connections, and layer normalization. The residual stream is a pathway where token embeddings are passed through and progressively modified by attention and feedforward layers, with layer normalization applied before each. The feedforward layer is a two-layer network with ReLU activation. Layer normalization normalizes a vector by subtracting its mean and dividing by its standard deviation, then applies learnable gain and offset parameters. The block's computations are broken down into steps: layer norm, multi-head attention, residual addition, another layer norm, feedforward, and final residual addition. Only the attention component mixes information from other tokens, making it the token-mixing part. The input and output dimensions are matched, allowing stacking of multiple blocks (e.g., 12 to 96 layers) in large language models.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "residual", + "stream", + "The", + "attention", + "calculation", + "lies", + "core", + "called", + "transformer", + "block", + "addition", + "layer", + "includes", + "kinds", + "layers", + "feedforward", + "connections", + "normalizing", + "colloquially", + "norm", + "Fig", + "8.7", + "illustrates", + "sketching", + "common", + "thinking", + "Elhage", + "al", + "2021", + "In", + "viewpoint", + "processing", + "individual", + "token", + "single", + "dimensional", + "representations", + "position", + "This", + "starts", + "original", + "input", + "vector", + "components", + "read", + "add", + "output", + "back", + "bottom", + "embedding", + "dimensionality", + "initial", + "passed", + "progressively", + "added", + "introduce", + "Before", + "computation", + "Thus", + "result", + "case", + "And", + "summed", + "ll", + "mathbf", + "refer", + "resulting", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "TD", + "Project", + "final", + "representation", + "Concatenate", + "Outputs", + "C1", + "Head", + "K1", + "V1", + "Q1", + "C2", + "K2", + "V2", + "Q2", + "C3", + "K3", + "V3", + "Q3", + "C4", + "K4", + "V4", + "Q4", + "D1", + "D2", + "D3", + "D4", + "style", + "fill", + "f9f", + "stroke", + "333", + "ccf", + "cfc", + "Figure", + "8.6", + "multi", + "head", + "producing", + "heads", + "query", + "key", + "weight", + "matrices", + "figure", + "show", + "smaller", + "fit", + "page", + "outputs", + "shape", + "times", + "concatenated", + "projected", + "space", + "matrix", + "Usually", + "set", + "square", + "Ad", + "size", + "MultiHead", + "Attention", + "Feedforward", + "Layer", + "Norm", + "architecture", + "showing", + "information", + "flows", + "module", + "sensitive", + "streams", + "prior", + "positions", + "chapter", + "prenorm", + "version", + "norms", + "happen", + "We", + "ve", + "computations", + "context", + "fully", + "connected", + "network", + "hidden", + "introduced", + "Chapter", + "weights", + "It", + "make", + "ff", + "larger", + "model", + "For", + "512", + "2048", + "operatorname", + "FFN", + "left", + "ReLU", + "tag", + "8.21", + "At", + "stages", + "normalize", + "Ba", + "2016", + "process", + "short", + "normalization", + "forms", + "improve", + "training", + "performance", + "deep", + "neural", + "networks", + "keeping", + "values", + "range", + "facilitates", + "gradient", + "based", + "variation", + "score", + "statistics", + "applied", + "That", + "term", + "bit", + "confusing", + "entire", + "normalized", + "step", + "calculate", + "mu", + "standard", + "deviation", + "sigma", + "elements", + "Given", + "calculated", + "frac", + "sum", + "8.22", + "sqrt", + "8.23", + "subtracting", + "dividing", + "hat", + "8.24", + "Finally", + "implementation", + "learnable", + "parameters", + "gamma", + "beta", + "representing", + "gain", + "offset", + "text", + "LayerNorm", + "8.25", + "Putting", + "function", + "computed", + "expressed", + "breaking", + "equation", + "component", + "stand", + "superscripts", + "demarcate", + "inside", + "8.26", + "MultiHeadAttention", + "dots", + "8.27", + "8.28", + "8.29", + "8.30", + "8.31", + "mixing", + "Notice", + "takes", + "tokens", + "Eq", + "neighboring", + "fact", + "view", + "literally", + "moving", + "current", + "high", + "albeit", + "subspaces", + "8.8", + "shows", + "visualization", + "movement", + "call", + "mixes", + "image", + "Token", + "An", + "move", + "Crucially", + "dimensions", + "blocks", + "matched", + "stacked", + "Each", + "Transformers", + "large", + "language", + "models", + "stack", + "12", + "T5", + "GPT", + "small", + "96", + "recent", + "issue", + "stacking", + "Equation", + "metaphor", + "12th", + "earlier", + "highest", + "end", + "trained", + "predict", + "Once", + "requirement", + "extra", + "run", + "define" + ], + "keywords": [ + "residual stream", + "layer norm", + "feedforward" + ], + "connect_to": [ + { + "target": "770bece3-4834-5835-864b-56439a28e1c0", + "relation": "embeds", + "ref": "[images/image-6-Multi-Head Attention Mechanism.jpg]", + "position": { + "start": 1679, + "end": 1730 + } + }, + { + "target": "e28cb543-3349-5d91-bd08-a8accb95486d", + "relation": "embeds", + "ref": "[images/image-7-Transformer Block Architecture.jpg]", + "position": { + "start": 3133, + "end": 3184 + } + }, + { + "target": "ef9fe40b-a3cd-5af2-a7ce-ef4883f409aa", + "relation": "embeds", + "ref": "[images/image-8-Diagram of Token Residual Streams.jpg]", + "position": { + "start": 8900, + "end": 8954 + } + } + ] + } + }, + { + "chunk_id": "566ee9cf-60a0-5204-90d6-23e7ab35e530", + "type": "image", + "content": "\nParallelizing attention Let's first see this for a single attention head and then turn to multiple heads, and then add in the rest of the components in the transformer block. For one head we multiply $\\mathbf{X}$ by the query, key, and value matrices $\\mathbf{W}^{\\mathbf{Q}}$ of shape $[d\\times d_k]$ , $\\mathbf{W}^{\\mathbf{K}}$ of shape $[d\\times d_k]$ , and $\\mathbf{W}^{\\mathbf{V}}$ of shape $[d\\times d_v]$ , to produce matrices $\\mathbf{Q}$ of shape $[N\\times d_k]$ , $\\mathbf{K}$ of shape $[N\\times d_k]$ , and $\\mathbf{V}$ of shape $[N\\times d_v]$ , containing all the key, query, and value vectors:
0\n[images/image-9-N x N Grid Structure.jpg]\n", + "path": "images/image-9-N x N Grid Structure.jpg", + "metadata": { + "length": 652, + "summary": "image-9\nq1·k1, q1·k2, q1·k3, q1·k4\nq2·k1, q2·k2, q2·k3, q2·k4\nq3·k1, q3·k2, q3·k3, q3·k4\nq4·k1, q4·k2, q4·k3, q4·k4", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-9-N x N Grid Structure.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f2fd271f-e2d3-5541-a56c-a44c42964946", + "type": "table", + "content": "
q1·k1-∞-∞-∞
q2·k1q2·k2-∞-∞
q3·k1q3·k2q3·k3-∞
q4·k1q4·k2q4·k3q4·k4
", + "path": "tables/table-1 Causal Attention Mask.html", + "metadata": { + "length": 257, + "summary": "table-2\nThe table shows a causal attention mask where each query attends only to current and previous keys, with future positions masked as -∞.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "tables/table-1 Causal Attention Mask.html", + "keywords": [ + "attention mask", + "causal", + "triangular" + ], + "tokens": [] + } + }, + { + "chunk_id": "e6309067-0eca-5a32-92ab-5a898171f6b2", + "type": "image", + "content": "\nN
0\n[images/image-10-Attention Mechanism Process.jpg]\n", + "path": "images/image-10-Attention Mechanism Process.jpg", + "metadata": { + "length": 47, + "summary": "image-10\nTop Row: Linear Transformations\n- Input (X): N x d matrix containing Input Tokens 1 through 4.\n- Query Generation (Q): X multiplied by weight matrix W^Q (d x d_k) results in Q (N x d_k).\n- Key Generation (K): X multiplied by weight matrix W^K (d x d_k) results in K (N x d_k).\n- Value Generation (V): X multiplied by weight matrix W^V (d x d_v) results in V (N x d_v).\n\nBottom Row: Attention Calculation\n- Matrix Multiplication: Q (N x d_k) is multiplied by the transpose of K (K^T, d_k x N) to produce the attention scores matrix QK^T (N x N).\n- Masking: The QK^T matrix is masked (indicated by \"mask\" and infinity symbols -∞) to prevent future tokens from attending to current ones.\n- Final Output (A): The masked matrix is multiplied by the Value matrix V (N x d_v) to produce the final output A (N x d", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-10-Attention Mechanism Process.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ccca02ee-7994-5974-a303-4829ba166a93", + "type": "text", + "content": "This description of multi-head attention and the rest of the transformer block has been from the perspective of computing a single output at a single time step $i$ in a single residual stream. But as we pointed out earlier, the attention computation performed for each token to compute $\\mathbf{a}_i$ is independent of the computation for each other token, and that's also true for all the computation in the transformer block computing $\\mathbf{h}_i$ from the input $\\mathbf{x}_i$ . That means we can easily parallelize the entire computation, taking advantage of efficient matrix multiplication routines.\nWe do this by packing the input embeddings for the $N$ tokens of the input sequence into a single matrix $\\mathbf{X}$ of size $[N\\times d]$ . Each row of $\\mathbf{X}$ is the embedding of one token of the input. Transformers for large language models commonly have an input length $N$ from 1K to 32K; much longer contexts of 128K or even up to millions of tokens can also be achieved with architectural changes like special long-context mechanisms that we don't discuss here. So for vanilla transformers, we can think of $\\mathbf{X}$ having between 1K and 32K rows, each of the dimensionality of the embedding $d$ (the model dimension).\nParallelizing attention Let's first see this for a single attention head and then turn to multiple heads, and then add in the rest of the components in the transformer block. For one head we multiply $\\mathbf{X}$ by the query, key, and value matrices $\\mathbf{W}^{\\mathbf{Q}}$ of shape $[d\\times d_k]$ , $\\mathbf{W}^{\\mathbf{K}}$ of shape $[d\\times d_k]$ , and $\\mathbf{W}^{\\mathbf{V}}$ of shape $[d\\times d_v]$ , to produce matrices $\\mathbf{Q}$ of shape $[N\\times d_k]$ , $\\mathbf{K}$ of shape $[N\\times d_k]$ , and $\\mathbf{V}$ of shape $[N\\times d_v]$ , containing all the key, query, and value vectors:\n$$\n\\mathbf {Q} = \\mathbf {X W} ^ {\\mathrm{Q}}; \\quad \\mathbf {K} = \\mathbf {X W} ^ {\\mathrm{K}}; \\quad \\mathbf {V} = \\mathbf {X W} ^ {\\mathrm{V}} \\tag {8.32}\n$$\nGiven these matrices we can compute all the requisite query-key comparisons simultaneously by multiplying Q and $K^{T}$ in a single matrix multiplication. The product is of shape $N \\times N$ , visualized in Fig. 8.9.\n\nParallelizing attention Let's first see this for a single attention head and then turn to multiple heads, and then add in the rest of the components in the transformer block. For one head we multiply $\\mathbf{X}$ by the query, key, and value matrices $\\mathbf{W}^{\\mathbf{Q}}$ of shape $[d\\times d_k]$ , $\\mathbf{W}^{\\mathbf{K}}$ of shape $[d\\times d_k]$ , and $\\mathbf{W}^{\\mathbf{V}}$ of shape $[d\\times d_v]$ , to produce matrices $\\mathbf{Q}$ of shape $[N\\times d_k]$ , $\\mathbf{K}$ of shape $[N\\times d_k]$ , and $\\mathbf{V}$ of shape $[N\\times d_v]$ , containing all the key, query, and value vectors:
0\n[images/image-9-N x N Grid Structure.jpg]\n\n
\ntext_image\nq1·k1 q1·k2 q1·k3 q1·k4\nq2·k1 q2·k2 q2·k3 q2·k4\nq3·k1 q3·k2 q3·k3 q3·k4\nq4·k1 q4·k2 q4·k3 q4·k4\nN\nN\n
\nFigure 8.9 The $N \\times N$ $\\mathbf{QK}^{\\top}$ matrix showing how it computes all $q_{i} \\cdot k_{j}$ comparisons in a single matrix multiple.\nOnce we have this $\\mathbf{QK}^{\\top}$ matrix, we can very efficiently scale these scores, take the softmax, and then multiply the result by $\\mathbf{V}$ resulting in a matrix of shape $N\\times d$ : a vector embedding representation for each token in the input. We've reduced the entire self-attention step for an entire sequence of $N$ tokens for one head to the following computation:\n$$\n\\mathbf {h e a d} = \\text { softmax } \\left(\\operatorname{mask} \\left(\\frac {\\mathbf {Q K} ^ {\\top}}{\\sqrt {d _ {k}}}\\right)\\right) \\mathbf {V} \\tag {8.33}\n$$\n$$\n\\mathbf {A} = \\text { head } \\mathbf {W} ^ {0} \\tag {8.34}\n$$\nMasking out the future You may have noticed that we introduced a mask function in Eq. 8.34 above. This is because the self-attention computation as we've described it has a problem: the calculation of $\\mathbf{QK}^{\\mathrm{T}}$ results in a score for each query value to every key value, including those that follow the query. This is inappropriate in the setting of language modeling: guessing the next word is pretty simple if you already know it! To fix this, the elements in the upper-triangular portion of the matrix are set to $-\\infty$ , which the softmax will turn to zero, thus eliminating any knowledge of words that follow in the sequence. This is done in practice by adding a mask matrix $M$ in which $M_{ij} = -\\infty \\forall j > i$ (i.e. for the upper-triangular portion) and $M_{ij} = 0$ otherwise. Fig. 8.10 shows the resulting masked $\\mathbf{QK}^{\\mathrm{T}}$ matrix. (we'll see in Chapter 9 how to make use of words in the future for tasks that need it).\nN\n\n[tables/table-1 Causal Attention Mask.html]\n\nN\nFigure 8.10 The $N \\times N QK^{T}$ matrix showing the $q_{i} \\cdot k_{j}$ values, with the upper-triangle portion of the comparisons matrix zeroed out (set to $-\\infty$ , which the softmax will turn to zero).\nFig. 8.11 shows a schematic of all the computations for a single attention head parallelized in matrix form.\n\nN
0\n[images/image-10-Attention Mechanism Process.jpg]\n\n
\ntext_image\nX\nInput\nToken 1\nInput\nToken 2\nInput\nToken 3\nInput\nToken 4\nN x d\nW^Q\nx\nd x d_k\n= Q\nQuery\nToken 1\nQuery\nToken 2\nQuery\nToken 3\nQuery\nToken 4\nN x d_k\nX\nInput\nToken 1\nInput\nToken 2\nInput\nToken 3\nInput\nToken 4\nW^K\nx\nd x d_k\n= K\nKey\nToken 1\nKey\nToken 2\nKey\nToken 3\nKey\nToken 4\nN x d_k\nX\nInput\nToken 1\nInput\nToken 2\nInput\nToken 3\nInput\nToken 4\nW^V\nx\nd x d_v\n= V\nValue\nToken 1\nValue\nToken 2\nValue\nToken 3\nValue\nToken 4\nN x d_v\nmask\nQ\nq1 X K^T = QK^T masked V A\nq2 \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u03bb \\u2667 a1\nq3 d_k x N q1·k1 q1·k2 q1·k3 q1·k4 q2·k1 q2·k2 q2·k3 q2·k4 q3·k1 q3·k2 q3·k3 q3·k4 q4·k1 q4·k2 q4·k3 q4·k4 N x N N x N N x d_v N x d_v\n
\nFigure 8.11 Schematic of the attention computation for a single attention head in parallel. The first row shows the computation of the Q, K, and V matrices. The second row shows the computation of $QK^{T}$ , the masking (the softmax computation and the normalizing by dimensionality are not shown) and then the weighted sum of the value vectors to get the final attention vectors.\nFig. 8.9 and Fig. 8.10 also make it clear that attention is quadratic in the length of the input, since at each layer we need to compute dot products between each pair of tokens in the input. This makes it expensive to compute attention over very long documents (like entire novels). Nonetheless modern large language models manage to use quite long contexts of thousands or tens of thousands of tokens.\nParallelizing multi-head attention In multi-head attention, as with self-attention, the input and output have the model dimension d, the key and query embeddings have dimensionality $d_{k}$ , and the value embeddings are of dimensionality $d_{v}$ (again, in the original transformer paper $d_{k} = d_{v} = 64$ , A = 8, and d = 512). Thus for each head c, we have weight layers $W^{Q}_{c}$ of shape $[d \\times d_{k}]$ , $W^{K}_{c}$ of shape $[d \\times d_{k}]$ , and $W^{V}_{c}$ of shape $[d \\times d_{v}]$ , and these get multiplied by the inputs packed into X to produce Q of shape $[N \\times d_{k}]$ , K of shape $[N \\times d_{k}]$ , and V of shape $[N \\times d_{v}]$ . The output of each of the A heads is of shape $[N \\times d_{v}]$ , and so the output of the multi-head layer with A heads consists of A matrices of shape $[N \\times d_{v}]$ . To make use of these matrices in further processing, they are concatenated to produce a single output with dimensionality $[N \\times Ad_{v}]$ . Finally, we use a final linear projection $W^{O}$ of shape $[Ad_{v} \\times d]$ , that reshapes it to the original output dimension for each token. Multiplying the concatenated $[N \\times Ad_{v}]$ matrix output by $W^{O}$ of shape $[Ad_{v} \\times d]$ yields the self-attention output A of shape $[N \\times d]$ .\n$$\n\\mathbf {Q} ^ {\\mathrm{i}} = \\mathbf {X W} ^ {\\mathrm{Qi}}; \\quad \\mathbf {K} ^ {\\mathrm{i}} = \\mathbf {X W} ^ {\\mathrm{Ki}}; \\quad \\mathbf {V} ^ {\\mathrm{i}} = \\mathbf {X W} ^ {\\mathrm{Vi}} \\tag {8.35}\n$$\n$$\n\\mathbf {h e a d} _ {i} = \\text { SelfAttention } (\\mathbf {Q} ^ {\\mathrm{i}}, \\mathbf {K} ^ {\\mathrm{i}}, \\mathbf {V} ^ {\\mathrm{i}}) = \\text { softmax } \\left(\\text { mask } \\left(\\frac {\\mathbf {Q} ^ {\\mathrm{i}} \\mathbf {K} ^ {\\mathrm{iT}}}{\\sqrt {d _ {k}}}\\right)\\right) \\mathbf {V} ^ {\\mathrm{i}} \\tag {8.36}\n$$\n$$\n\\text { MultiHeadAttention } (\\mathbf {X}) = \\left(\\mathbf {h e a d} _ {1} \\oplus \\mathbf {h e a d} _ {2} \\dots \\oplus \\mathbf {h e a d} _ {A}\\right) \\mathbf {W} ^ {\\mathbf {0}} \\tag {8.37}\n$$\nPutting it all together with the parallel input matrix X The function computed in parallel by an entire layer of N transformer blocks—each block over one of the N input tokens—can be expressed as:\n$$\n\\mathbf {O} = \\mathbf {X} + \\text { MultiHeadAttention } (\\text { LayerNorm } (\\mathbf {X})) \\tag {8.38}\n$$\n$$\n\\mathbf {H} = \\mathbf {O} + \\operatorname{FFN} (\\text { LayerNorm } (\\mathbf {O})) \\tag {8.39}\n$$\nNote that in Eq. 8.38 we are using X to mean the input to the layer, wherever it comes from. For the first layer, as we will see in the next section, that input is the initial word + positional embedding vectors that we have been describing by X. But for subsequent layers k, the input is the output from the previous layer $H^{k-1}$ . We can also break down the computation performed in a transformer layer, showing one equation for each component computation. We'll use T (of shape $[N \\times d]$ ) to stand for transformer and superscripts to demarcate each computation inside the block, and again use X to mean the input to the block from the previous layer or the initial embedding:\n$$\n\\mathbf {T} ^ {1} = \\text { LayerNorm } (\\mathbf {X}) \\tag {8.40}\n$$\n$$\n\\mathbf {T} ^ {2} = \\text { MultiHeadAttention } (\\mathbf {T} ^ {1}) \\tag {8.41}\n$$\n$$\n\\mathbf {T} ^ {3} = \\mathbf {T} ^ {2} + \\mathbf {X} \\tag {8.42}\n$$\n$$\n\\mathbf {T} ^ {4} = \\text { LayerNorm } (\\mathbf {T} ^ {3}) \\tag {8.43}\n$$\n$$\n\\mathbf {T} ^ {5} = \\operatorname{FFN} (\\mathbf {T} ^ {4}) \\tag {8.44}\n$$\n$$\n\\mathbf {H} = \\mathbf {T} ^ {5} + \\mathbf {T} ^ {3} \\tag {8.45}\n$$\nHere when we use a notation like $\\mathrm{FFN}(\\mathbf{T}^{3})$ we mean that the same FFN is applied in parallel to each of the N embedding vectors in the window. Similarly, each of the\nN tokens is normed in parallel in the LayerNorm. Crucially, the input and output dimensions of transformer blocks are matched so they can be stacked. Since each token $x_{i}$ at the input to the block is represented by an embedding of dimensionality $[1 \\times d]$ , that means the input X and output H are both of shape $[N \\times d]$ .", + "path": "8.pdf/CHAPTER 8 Transformers/8.3 Parallelizing computation using a single matrix X", + "metadata": { + "length": 10905, + "summary": "This passage explains how multi-head attention and transformer blocks are parallelized for efficient computation. Input embeddings for N tokens are packed into a matrix X of shape [N x d]. For a single attention head, X is multiplied by query, key, and value weight matrices to produce Q, K, V matrices. All query-key comparisons are computed simultaneously via QK^T matrix multiplication, yielding an N x N matrix. A mask function sets upper-triangular elements to -infinity to prevent attending to future tokens. The scaled, masked, softmaxed result is multiplied by V to produce attention outputs. For multi-head attention, each head has its own Q, K, V projections; outputs are concatenated and linearly projected. The entire transformer layer computation is expressed in matrix form, with residual connections and layer normalization applied in parallel across all tokens. Attention is quadratic in input length, making long contexts computationally expensive.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "This", + "description", + "multi", + "head", + "attention", + "rest", + "transformer", + "block", + "perspective", + "computing", + "single", + "output", + "time", + "step", + "residual", + "stream", + "But", + "pointed", + "earlier", + "computation", + "performed", + "token", + "compute", + "mathbf", + "independent", + "true", + "input", + "That", + "means", + "easily", + "parallelize", + "entire", + "taking", + "advantage", + "efficient", + "matrix", + "multiplication", + "routines", + "We", + "packing", + "embeddings", + "tokens", + "sequence", + "size", + "times", + "Each", + "row", + "embedding", + "Transformers", + "large", + "language", + "models", + "commonly", + "length", + "1K", + "32K", + "longer", + "contexts", + "128K", + "millions", + "achieved", + "architectural", + "special", + "long", + "context", + "mechanisms", + "don", + "discuss", + "So", + "vanilla", + "transformers", + "rows", + "dimensionality", + "model", + "dimension", + "Parallelizing", + "Let", + "turn", + "multiple", + "heads", + "add", + "components", + "For", + "multiply", + "query", + "key", + "matrices", + "shape", + "produce", + "vectors", + "mathrm", + "quad", + "tag", + "8.32", + "Given", + "requisite", + "comparisons", + "simultaneously", + "multiplying", + "The", + "product", + "visualized", + "Fig", + "8.9", + "details", + "summary", + "text", + "image", + "q1", + "k1", + "k2", + "k3", + "k4", + "q2", + "q3", + "q4", + "Figure", + "QK", + "top", + "showing", + "computes", + "cdot", + "Once", + "efficiently", + "scale", + "scores", + "softmax", + "result", + "resulting", + "vector", + "representation", + "ve", + "reduced", + "left", + "operatorname", + "mask", + "frac", + "sqrt", + "8.33", + "8.34", + "Masking", + "future", + "You", + "noticed", + "introduced", + "function", + "Eq", + "problem", + "calculation", + "results", + "score", + "including", + "follow", + "inappropriate", + "setting", + "modeling", + "guessing", + "word", + "pretty", + "simple", + "To", + "fix", + "elements", + "upper", + "triangular", + "portion", + "set", + "infty", + "eliminating", + "knowledge", + "words", + "practice", + "adding", + "ij", + "forall", + "8.10", + "shows", + "masked", + "ll", + "Chapter", + "make", + "tasks", + "values", + "triangle", + "zeroed", + "8.11", + "schematic", + "computations", + "parallelized", + "form", + "Input", + "Token", + "Query", + "Key", + "Value", + "u03bb", + "u2667", + "a1", + "Schematic", + "parallel", + "masking", + "normalizing", + "shown", + "weighted", + "sum", + "final", + "clear", + "quadratic", + "layer", + "dot", + "products", + "pair", + "makes", + "expensive", + "documents", + "novels", + "Nonetheless", + "modern", + "manage", + "thousands", + "tens", + "In", + "original", + "paper", + "64", + "512", + "Thus", + "weight", + "layers", + "multiplied", + "inputs", + "packed", + "consists", + "processing", + "concatenated", + "Ad", + "Finally", + "linear", + "projection", + "reshapes", + "Multiplying", + "yields", + "Qi", + "Ki", + "Vi", + "8.35", + "SelfAttention", + "iT", + "8.36", + "MultiHeadAttention", + "oplus", + "dots", + "8.37", + "Putting", + "computed", + "blocks", + "expressed", + "LayerNorm", + "8.38", + "FFN", + "8.39", + "Note", + "section", + "initial", + "positional", + "describing", + "subsequent", + "previous", + "break", + "equation", + "component", + "stand", + "superscripts", + "demarcate", + "inside", + "8.40", + "8.41", + "8.42", + "8.43", + "8.44", + "8.45", + "Here", + "notation", + "applied", + "window", + "Similarly", + "normed", + "Crucially", + "dimensions", + "matched", + "stacked", + "Since", + "represented" + ], + "keywords": [ + "Multi-Head Attention", + "Parallelization", + "Transformer" + ], + "connect_to": [ + { + "target": "566ee9cf-60a0-5204-90d6-23e7ab35e530", + "relation": "embeds", + "ref": "[images/image-9-N x N Grid Structure.jpg]", + "position": { + "start": 2850, + "end": 2891 + } + }, + { + "target": "f2fd271f-e2d3-5541-a56c-a44c42964946", + "relation": "embeds", + "ref": "[tables/table-1 Causal Attention Mask.html]", + "position": { + "start": 4784, + "end": 4827 + } + }, + { + "target": "e6309067-0eca-5a32-92ab-5a898171f6b2", + "relation": "embeds", + "ref": "[images/image-10-Attention Mechanism Process.jpg]", + "position": { + "start": 5164, + "end": 5213 + } + } + ] + } + }, + { + "chunk_id": "713ea161-334e-59c7-8683-6fc082f8e6ca", + "type": "image", + "content": "\n$$
0\n[images/image-11-$$ _detail.jpg]\n", + "path": "images/image-11-$$ _detail.jpg", + "metadata": { + "length": 48, + "summary": "image-11\n$$
0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-11-$$ _detail.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ebe2cef5-fc30-5a80-893c-a0fc27eef34b", + "type": "table", + "content": "\n \n \n \n \n \n \n
", + "path": "tables/table-3 .html", + "metadata": { + "length": 133, + "summary": "table-3", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "tables/table-3 .html", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e5077fef-2d32-56c3-a333-837657417fb8", + "type": "image", + "content": "\n
0\n[images/image-12-__details_.jpg]\n", + "path": "images/image-12-__details_.jpg", + "metadata": { + "length": 56, + "summary": "image-12\n
0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-12-__details_.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "135b24bb-f81e-555e-b9cb-8d27ceafbdab", + "type": "table", + "content": "\n \n \n \n \n \n \n
", + "path": "tables/table-4 .html", + "metadata": { + "length": 133, + "summary": "table-4", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "tables/table-4 .html", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "22eb67ef-70b8-59ac-a0b0-c498ff36b4a5", + "type": "image", + "content": "\npositional embeddings
0\n[images/image-13-Transformer Block Architecture.jpg]\n", + "path": "images/image-13-Transformer Block Architecture.jpg", + "metadata": { + "length": 67, + "summary": "image-13\n- Core Component: Transformer Block (top level)\n- Input Definition: X = Composite Embeddings (word + position)\n- Process Flow:\n - Word Embeddings and Position Embeddings are combined via addition (+) to form the composite input.\n- Sequence Elements:\n - Token 1: \"Janet\" (Word) + \"1\" (Position)\n - Token 2: \"will\" (Word) + \"2\" (Position)\n - Token 3: \"back\" (Word) + \"3\" (Position)\n - Token 4: \"the\" (Word) + \"4\" (Position)\n - Token 5: \"bill\" (Word) + \"5\" (Position)", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-13-Transformer Block Architecture.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "619d4324-d7af-50f0-ac02-0f1b204d35e5", + "type": "text", + "content": "embedding\nLet's talk about where the input $\\mathbf{X}$ comes from. Given a sequence of $N$ tokens ( $N$ is the context length in tokens), the matrix $\\mathbf{X}$ of shape $[N \\times d]$ has an embedding for each word in the context. The transformer does this by separately computing two embeddings: an input token embedding, and an input positional embedding.\nA token embedding, introduced in Chapter 6, is a vector of dimension d that will be our initial representation for the input token. (As we pass vectors up through the transformer layers in the residual stream, this embedding representation will change and grow, incorporating context and playing a different role depending on the kind of language model we are building.) The set of initial embeddings are stored in the embedding matrix E, which has a row for each of the $|V|$ tokens in the vocabulary. (Reminder that V here means the vocabulary of tokens, this V is not related to the value vector.) Thus each word is a row vector of d dimensions, and E has shape $[|V| \\times d]$ .\nGiven an input token string like Thanks for all the we first convert the tokens into vocabulary indices (these were created when we first tokenized the input using BPE or SentencePiece). So the representation of thanks for all the might be w = [5, 4000, 10532, 2224]. Next we use indexing to select the corresponding rows from E, (row 5, row 4000, row 10532, row 2224).\nAnother way to think about selecting token embeddings from the embedding matrix is to represent tokens as one-hot vectors of shape $[1 \\times |V|]$ , i.e., with one dimension for each word in the vocabulary. Recall that in a one-hot vector all the elements are 0 except one, the element whose dimension is the word's index in the vocabulary, which has value 1. So if the word “thanks” has index 5 in the vocabulary, $x_{5} = 1$ , and $x_{i} = 0 \\forall i \\neq 5$ , as shown here:\n$$\n[ \\emptyset \\emptyset \\emptyset \\emptyset 1 \\emptyset \\emptyset \\dots \\emptyset \\emptyset \\emptyset \\emptyset ]\n$$\n$$\n\\begin{array}{c c c c c c c c c c} 1 & 2 & 3 & 4 & 5 & 6 & 7 & \\dots & \\dots & | V | \\end{array}\n$$\nMultiplying by a one-hot vector that has only one non-zero element $x_{i}=1$ simply selects out the relevant row vector for word i, resulting in the embedding for word i, as depicted in Fig. 8.12.\n\n$$
0\n[images/image-11-$$ _detail.jpg]\n\n
\ntext_image\n1 5 |V|\n0000100...0000\n×\n5\nE\n= 1 d\n\n[tables/table-3 .html]\n\n
\nFigure 8.12 Selecting the embedding vector for word $V_{5}$ by multiplying the embedding matrix E with a one-hot vector with a 1 in index 5.\nWe can extend this idea to represent the entire token sequence as a matrix of one-hot vectors, one for each of the $N$ positions in the transformer's context window, as shown in Fig. 8.13.\none-hot vector\n\n
0\n[images/image-12-__details_.jpg]\n\n
\ntext_image\n\n[tables/table-4 .html]\n\n0 0 0 0 |1 |0 0 ... 0 0 0 0\n0 0 0 0 0 0 0 ... 0 0 |1 |0\n1 |0 0 0 0 0 0 ... 0 0 0 0\n...\nN\n0 0 0 0 |1 |0 0 ... 0 0 0 0\n×\nd\nE\n=\nd\nN\n
\nFigure 8.13 Selecting the embedding matrix for the input sequence of token ids W by multiplying a one-hot matrix corresponding to W by the embedding matrix E.\npositional embeddings\nabsolute position\nThese token embeddings are not position-dependent. To represent the position of each token in the sequence, we combine these token embeddings with positional embeddings specific to each position in an input sequence.\nWhere do we get these positional embeddings? The simplest method, called absolute position, is to start with randomly initialized embeddings corresponding to each possible input position up to some maximum length. For example, just as we have an embedding for the word fish, we'll have an embedding for the position 3. As with word embeddings, these positional embeddings are learned along with other parameters during training. We can store them in a matrix $E_{\\mathrm{pos}}$ of shape $[N \\times d]$ .\nTo produce an input embedding that captures positional information, we just add the word embedding for each input to its corresponding positional embedding. The individual token and position embeddings are both of size $[1 \\times d]$ , so their sum is also $[1 \\times d]$ , This new embedding serves as the input for further processing. Fig. 8.14 shows the idea.\n\npositional embeddings
0\n[images/image-13-Transformer Block Architecture.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"Transformer Block\"] --> B[\"X = Composite Embeddings (word + position)\"]\nB --> C[\"+\"]\nC --> D[\"janet\"]\nC --> E[\"will\"]\nC --> F[\"back\"]\nC --> G[\"the\"]\nC --> H[\"bill\"]\nB --> I[\"Word Embeddings\"]\nI --> J[\"Position Embeddings\"]\nJ --> K[\"1\"]\nJ --> L[\"2\"]\nJ --> M[\"3\"]\nJ --> N[\"4\"]\nJ --> O[\"5\"]\n```\n
\nFigure 8.14 A simple way to model position: add an embedding of the absolute position to the token embedding to produce a new embedding of the same dimensionality.\nThe final representation of the input, the matrix X, is an $[N \\times d]$ matrix in which each row i is the representation of the ith token in the input, computed by adding $\\mathbf{E}[id(i)]$ —the embedding of the id of the token that occurred at position i—, to P[i], the positional embedding of position i.\nA potential problem with the simple position embedding approach is that there will be plenty of training examples for the initial positions in our inputs and correspondingly fewer at the outer length limits. These latter embeddings may be poorly trained and may not generalize well during testing. An alternative is to choose a static function that maps integer inputs to real-valued vectors in a way that better handles sequences of arbitrary length. A combination of sine and cosine functions with differing frequencies was used in the original transformer work. Sinusoidal position embeddings may also help in capturing the inherent relationships among the\nrelative position\npositions, like the fact that position 4 in an input is more closely related to position 5 than it is to position 17.\nA more complex style of positional embedding methods extend this idea of capturing relationships even further to directly represent relative position instead of absolute position, often implemented in the attention mechanism at each layer rather than being added once at the initial input.", + "path": "8.pdf/CHAPTER 8 Transformers/8.4 The input: embeddings for token and position", + "metadata": { + "length": 6355, + "summary": "The passage explains how input embeddings are formed in transformers. Token embeddings are vectors from an embedding matrix E (size |V|×d) selected by token indices or one-hot vectors. Positional embeddings, often learned or sinusoidal, are added to token embeddings to encode position. The final input matrix X (N×d) combines both, with each row representing a token's composite embedding.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "embedding", + "Let", + "talk", + "input", + "mathbf", + "Given", + "sequence", + "tokens", + "context", + "length", + "matrix", + "shape", + "times", + "word", + "The", + "transformer", + "separately", + "computing", + "embeddings", + "token", + "positional", + "introduced", + "Chapter", + "vector", + "dimension", + "initial", + "representation", + "As", + "pass", + "vectors", + "layers", + "residual", + "stream", + "change", + "grow", + "incorporating", + "playing", + "role", + "depending", + "kind", + "language", + "model", + "building", + "set", + "stored", + "row", + "vocabulary", + "Reminder", + "means", + "related", + "Thus", + "dimensions", + "string", + "Thanks", + "convert", + "indices", + "created", + "tokenized", + "BPE", + "SentencePiece", + "So", + "4000", + "10532", + "2224", + "Next", + "indexing", + "select", + "rows", + "Another", + "selecting", + "represent", + "hot", + "Recall", + "elements", + "element", + "index", + "forall", + "neq", + "shown", + "emptyset", + "dots", + "begin", + "array", + "end", + "Multiplying", + "simply", + "selects", + "relevant", + "resulting", + "depicted", + "Fig", + "8.12", + "details", + "summary", + "text", + "image", + "0000100", + "0000", + "Figure", + "Selecting", + "multiplying", + "We", + "extend", + "idea", + "entire", + "positions", + "window", + "8.13", + "ids", + "absolute", + "position", + "These", + "dependent", + "To", + "combine", + "specific", + "Where", + "simplest", + "method", + "called", + "start", + "randomly", + "initialized", + "maximum", + "For", + "fish", + "ll", + "learned", + "parameters", + "training", + "store", + "mathrm", + "pos", + "produce", + "captures", + "information", + "add", + "individual", + "size", + "sum", + "This", + "serves", + "processing", + "8.14", + "shows", + "flowchart", + "mermaid", + "graph", + "TD", + "Transformer", + "Block", + "Composite", + "Embeddings", + "janet", + "back", + "bill", + "Word", + "Position", + "simple", + "dimensionality", + "final", + "ith", + "computed", + "adding", + "id", + "occurred", + "potential", + "problem", + "approach", + "plenty", + "examples", + "inputs", + "correspondingly", + "fewer", + "outer", + "limits", + "poorly", + "trained", + "generalize", + "testing", + "An", + "alternative", + "choose", + "static", + "function", + "maps", + "integer", + "real", + "valued", + "handles", + "sequences", + "arbitrary", + "combination", + "sine", + "cosine", + "functions", + "differing", + "frequencies", + "original", + "work", + "Sinusoidal", + "capturing", + "inherent", + "relationships", + "relative", + "fact", + "closely", + "17", + "complex", + "style", + "methods", + "directly", + "implemented", + "attention", + "mechanism", + "layer", + "added" + ], + "keywords": [ + "token embedding", + "positional embedding", + "embedding matrix" + ], + "connect_to": [ + { + "target": "713ea161-334e-59c7-8683-6fc082f8e6ca", + "relation": "embeds", + "ref": "[images/image-11-$$ _detail.jpg]", + "position": { + "start": 2328, + "end": 2360 + } + }, + { + "target": "ebe2cef5-fc30-5a80-893c-a0fc27eef34b", + "relation": "embeds", + "ref": "[tables/table-3 .html]", + "position": { + "start": 2438, + "end": 2460 + } + }, + { + "target": "e5077fef-2d32-56c3-a333-837657417fb8", + "relation": "embeds", + "ref": "[images/image-12-__details_.jpg]", + "position": { + "start": 2841, + "end": 2873 + } + }, + { + "target": "135b24bb-f81e-555e-b9cb-8d27ceafbdab", + "relation": "embeds", + "ref": "[tables/table-4 .html]", + "position": { + "start": 2916, + "end": 2938 + } + }, + { + "target": "22eb67ef-70b8-59ac-a0b0-c498ff36b4a5", + "relation": "embeds", + "ref": "[images/image-13-Transformer Block Architecture.jpg]", + "position": { + "start": 4397, + "end": 4449 + } + } + ] + } + }, + { + "chunk_id": "0984ef58-fdb1-5436-a0e6-df196704605d", + "type": "image", + "content": "\nThe job of the language modeling head is to take the output of the final transformer layer from the last token N and use it to predict the upcoming word at position $N+1$ . Fig. 8.15 shows how to accomplish this task, taking the output of the last token at the last layer (the d-dimensional output embedding of shape $[1 \\times d]$ ) and producing a probability distribution over words (from which we will choose one to generate).
0\n[images/image-14-Language Model Head Architecture.jpg]\n", + "path": "images/image-14-Language Model Head Architecture.jpg", + "metadata": { + "length": 476, + "summary": "image-14\n- **Input**: Sequence of words (w1, w2, ..., wN) processed by a Layer L Transformer Block.\n- **Intermediate Output**: Hidden states from the final layer ($h^L_1$, $h^L_2$, ..., $h^L_N$). The diagram focuses on the last hidden state $h^L_N$ with dimensions [1 x d].\n- **Processing Steps**:\n - **Unembedding Layer**: Transforms the hidden state using matrix multiplication where $U = E^T$. This produces Logits with dimensions [1 x |V|].\n - **Softmax**: Applies softmax over the vocabulary V to convert logits into probabilities.\n- **Output**: Word probabilities vector [1 x |V|], representing the distribution over the vocabulary V for the next token prediction.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-14-Language Model Head Architecture.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ced50707-0b91-5279-bf94-adaf780459b2", + "type": "text", + "content": "language modeling head\nhead\nThe last component of the transformer we must introduce is the language modeling head. Here we are using the word head to mean the additional neural circuitry we add on top of the basic transformer architecture when we apply pretrained transformer models to various tasks. The language modeling head is the circuitry we need to do language modeling.\nRecall that language models, from the simple n-gram models of Chapter 3 through the feedforward and RNN language models of Chapter 6 and Chapter 13, are word predictors. Given a context of words, they assign a probability to each possible next word. For example, if the preceding context is “Thanks for all the” and we want to know how likely the next word is “fish” we would compute:\n$P(\\text{fish}|\\text{Thanks for all the})$\nLanguage models give us the ability to assign such a conditional probability to every possible next word, giving us a distribution over the entire vocabulary. The n-gram language models of Chapter 3 compute the probability of a word given counts of its occurrence with the $n - 1$ prior words. The context is thus of size $n - 1$ . For transformer language models, the context is the size of the transformer's context window, which can be quite large, like 32K tokens for large models (and much larger contexts of millions of words are possible with special long-context architectures).\nThe job of the language modeling head is to take the output of the final transformer layer from the last token N and use it to predict the upcoming word at position $N+1$ . Fig. 8.15 shows how to accomplish this task, taking the output of the last token at the last layer (the d-dimensional output embedding of shape $[1 \\times d]$ ) and producing a probability distribution over words (from which we will choose one to generate).\nThe first module in Fig. 8.15 is a linear layer, whose job is to project from the output $h_N^L$ , which represents the output token embedding at position $N$ from the final block $L$ , (hence of shape $[1 \\times d]$ ) to the logit vector, or score vector, that will have a single score for each of the $|V|$ possible words in the vocabulary $V$ . The logit vector $\\mathbf{u}$ is thus of dimensionality $[1 \\times |V|]$ .\nweight tying\nThis linear layer can be learned, but more commonly we tie this matrix to (the transpose of) the embedding matrix E. Recall that in weight tying, we use the same weights for two different matrices in the model. Thus at the input stage of the transformer the embedding matrix (of shape $[|V| \\times d]$ ) is used to map from a one-hot vector over the vocabulary (of shape $[1 \\times |V|]$ ) to an embedding (of shape $[1 \\times d]$ ). And then in the language model head, $E^{T}$ , the transpose of the embedding matrix (of shape $[d \\times |V|]$ ) is used to map back from an embedding (shape $[1 \\times d]$ ) to a vector over the vocabulary (shape $[1 \\times |V|]$ ). In the learning process, E will be optimized to be good at doing both of these mappings. We therefore sometimes call the transpose $E^{T}$ the unembedding layer because it is performing this reverse mapping.\nunembedding\n\nThe job of the language modeling head is to take the output of the final transformer layer from the last token N and use it to predict the upcoming word at position $N+1$ . Fig. 8.15 shows how to accomplish this task, taking the output of the last token at the last layer (the d-dimensional output embedding of shape $[1 \\times d]$ ) and producing a probability distribution over words (from which we will choose one to generate).
0\n[images/image-14-Language Model Head Architecture.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"Layer L Transformer Block\"] --> B[\"Softmax\"]\nB --> C[\"Unembedding layer U = E^T\"]\nC --> E[\"Softmax over vocabulary V\"]\nB --> H[\"y1\"]\nB --> I[\"y2\"]\nB --> J[\"...\"]\nB --> L[\"u1\"]\nB --> M[\"u2\"]\nB --> N[\"...\"]\nB --> P[\"h^L_1\"]\nB --> Q[\"h^L_2\"]\nP --> R[\"w1\"]\nQ --> S[\"w2\"]\nR --> T[\"...\"]\nS --> U[\"w_N\"]\nT --> V[\"h^L_N\"]\nU --> W[\"h^L_N\"]\nV --> X[\"1 x d\"]\n```\n
\nFigure 8.15 The language modeling head: the circuit at the top of a transformer that maps from the output embedding for token N from the last transformer layer ( $h_{N}^{L}$ ) to a probability distribution over words in the vocabulary V.\nA softmax layer turns the logits u into the probabilities y over the vocabulary.\n$$\n\\mathbf {u} = \\mathbf {h} _ {\\mathrm{N}} ^ {\\mathrm{L}} \\mathbf {E} ^ {\\mathrm{T}} \\tag {8.46}\n$$\n$$\n\\mathbf {y} = \\operatorname{softmax} (\\mathbf {u}) \\tag {8.47}\n$$\nWe can use these probabilities to do things like help assign a probability to a given text. But the most important usage is to generate text, which we do by sampling a word from these probabilities y. We might sample the highest probability word ('greedy' decoding), or use another of the sampling methods from Section ?? or Section 8.6.\nIn either case, whatever entry $y_{k}$ we choose from the probability vector $\\mathbf{y}$ , we generate the word that has that index $k$ .\nFig. 8.16 shows the total stacked architecture for one token $i$ . Note that the input to each transformer layer $x_{i}^{\\ell}$ is the same as the output from the preceding layer $h_i^{\\ell -1}$ .\nA terminological note before we conclude: You will sometimes see a transformer used for this kind of unidirectional causal language model called a decoder-only model. This is because this model constitutes roughly half of the encoder-decoder model for transformers that we'll see how to apply to machine translation in Chapter 12. (Confusingly, the original introduction of the transformer had an encoder-decoder architecture, and it was only later that the standard paradigm for causal language model was defined by using only the decoder part of this original architecture).\ndecoder-only model", + "path": "8.pdf/CHAPTER 8 Transformers/8.5 The Language Modeling Head", + "metadata": { + "length": 5809, + "summary": "The language modeling head is the final component of a transformer, used to predict the next word. It takes the output embedding from the last token at the final transformer layer and projects it via a linear layer (often using the transpose of the embedding matrix, called unembedding) to logits over the vocabulary. A softmax layer converts these logits into probabilities, enabling text generation through sampling. This architecture is also known as a decoder-only model.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "language", + "modeling", + "head", + "The", + "component", + "transformer", + "introduce", + "Here", + "word", + "additional", + "neural", + "circuitry", + "add", + "top", + "basic", + "architecture", + "apply", + "pretrained", + "models", + "tasks", + "Recall", + "simple", + "gram", + "Chapter", + "feedforward", + "RNN", + "13", + "predictors", + "Given", + "context", + "words", + "assign", + "probability", + "For", + "preceding", + "Thanks", + "fish", + "compute", + "text", + "Language", + "give", + "ability", + "conditional", + "giving", + "distribution", + "entire", + "vocabulary", + "counts", + "occurrence", + "prior", + "size", + "window", + "large", + "32K", + "tokens", + "larger", + "contexts", + "millions", + "special", + "long", + "architectures", + "job", + "output", + "final", + "layer", + "token", + "predict", + "upcoming", + "position", + "Fig", + "8.15", + "shows", + "accomplish", + "task", + "taking", + "dimensional", + "embedding", + "shape", + "times", + "producing", + "choose", + "generate", + "module", + "linear", + "project", + "represents", + "block", + "logit", + "vector", + "score", + "single", + "mathbf", + "dimensionality", + "weight", + "tying", + "This", + "learned", + "commonly", + "tie", + "matrix", + "transpose", + "weights", + "matrices", + "model", + "Thus", + "input", + "stage", + "map", + "hot", + "And", + "back", + "In", + "learning", + "process", + "optimized", + "good", + "mappings", + "We", + "call", + "unembedding", + "performing", + "reverse", + "mapping", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "TD", + "Layer", + "Transformer", + "Block", + "Softmax", + "Unembedding", + "y1", + "y2", + "u1", + "u2", + "w1", + "w2", + "Figure", + "circuit", + "maps", + "softmax", + "turns", + "logits", + "probabilities", + "mathrm", + "tag", + "8.46", + "operatorname", + "8.47", + "things", + "But", + "important", + "usage", + "sampling", + "sample", + "highest", + "greedy", + "decoding", + "methods", + "Section", + "8.6", + "case", + "entry", + "index", + "8.16", + "total", + "stacked", + "Note", + "ell", + "terminological", + "note", + "conclude", + "You", + "kind", + "unidirectional", + "causal", + "called", + "decoder", + "constitutes", + "roughly", + "half", + "encoder", + "transformers", + "ll", + "machine", + "translation", + "12", + "Confusingly", + "original", + "introduction", + "standard", + "paradigm", + "defined", + "part" + ], + "keywords": [ + "language modeling head", + "weight tying", + "decoder-only model" + ], + "connect_to": [ + { + "target": "0984ef58-fdb1-5436-a0e6-df196704605d", + "relation": "embeds", + "ref": "[images/image-14-Language Model Head Architecture.jpg]", + "position": { + "start": 3592, + "end": 3646 + } + } + ] + } + }, + { + "chunk_id": "5de60f6a-7b06-5f67-a585-29abffb11588", + "type": "image", + "content": "\nFig. 8.16 shows the total stacked architecture for one token $i$ . Note that the input to each transformer layer $x_{i}^{\\ell}$ is the same as the output from the preceding layer $h_i^{\\ell -1}$ .
0\n[images/image-15-Transformer Architecture Diagram.jpg]\n", + "path": "images/image-15-Transformer Architecture Diagram.jpg", + "metadata": { + "length": 242, + "summary": "image-15\n- Input Encoding: Input token (Wi) passes through Embedding (E) and positional encoding to produce x1i.\n- Layer Structure (Layers 1 to L): Each layer contains a sequence of operations:\n - Layer Norm\n - Attention mechanism\n - Layer Norm\n - Feedforward network\n- Residual Connections: Indicated by arrows bypassing blocks (e.g., h^1_i = x^2_i).\n- Language Modeling Head: Located at the top, consisting of:\n - Extra Layer Norm\n - Linear projection matrix U\n - Softmax function producing logits (u1...u|V|)\n- Output Generation: Produces token probabilities (y1...y|V|) used to sample the next token (w_{i+1}).", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-15-Transformer Architecture Diagram.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "433b23c9-f753-593c-b6ab-4540d0d35173", + "type": "text", + "content": "The sampling methods we introduce below each have parameters that enable trading off two important factors in generation: quality and diversity. Methods that emphasize the most probable words tend to produce generations that are rated by people as more accurate, more coherent, and more factual, but also more boring and more repetitive. Methods that give a bit more weight to the middle-probability words tend to be more creative and more diverse, but less factual and more likely to be incoherent or otherwise low-quality.\n\nFig. 8.16 shows the total stacked architecture for one token $i$ . Note that the input to each transformer layer $x_{i}^{\\ell}$ is the same as the output from the preceding layer $h_i^{\\ell -1}$ .
0\n[images/image-15-Transformer Architecture Diagram.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"Input token W_i\"] --> B[\"Layer 1\"]\nB --> C[\"Layer 2\"]\nC --> D[\"Layer L\"]\nD --> E[\"Layer 1\"]\nE --> F[\"Input Encoding\"]\nF --> G[\"Output w_i\"]\nH[\"Input Encoding\"] --> I[\"x^1_i\"]\nI --> J[\"h^1_i = x^2_i\"]\nJ --> K[\"Layer 2\"]\nK --> L[\"h^2_i = x^3_i\"]\nL --> M[\"Layer L\"]\nM --> N[\"...\"]\nN --> O[\"Layer L\"]\nO --> P[\"u1\"]\nO --> Q[\"u2\"]\nO --> R[\"uV\"]\nP --> S[\"y1\"]\nQ --> T[\"y2\"]\nS --> V[\"Sample token to generate at position i+1\"]\nT --> V\nU --> V\nV --> W[\"w_{i+1}\"]\nW --> X[\"Output w_i\"]\nstyle A fill:#f9f,stroke:#333\nstyle H fill:#f9f,stroke:#333\nstyle I fill:#ccf,stroke:#333\nstyle F fill:#ccf,stroke:#333\n```\n
\nFigure 8.16 A transformer language model (decoder-only), stacking transformer blocks and mapping from an input token $w_{i}$ to a predicted next token $w_{i + 1}$ .", + "path": "8.pdf/CHAPTER 8 Transformers/8.6 More on Sampling", + "metadata": { + "length": 1605, + "summary": "The passage discusses trade-offs in text generation sampling methods between quality (accuracy, coherence, factuality) and diversity (creativity, novelty). It then describes a transformer language model architecture (decoder-only) shown in Figure 8.16, where stacked transformer layers map an input token to a predicted next token, with input to each layer being the output from the previous layer.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "The", + "sampling", + "methods", + "introduce", + "parameters", + "enable", + "trading", + "important", + "factors", + "generation", + "quality", + "diversity", + "Methods", + "emphasize", + "probable", + "words", + "tend", + "produce", + "generations", + "rated", + "people", + "accurate", + "coherent", + "factual", + "boring", + "repetitive", + "give", + "bit", + "weight", + "middle", + "probability", + "creative", + "diverse", + "incoherent", + "low", + "Fig", + "8.16", + "shows", + "total", + "stacked", + "architecture", + "token", + "Note", + "input", + "transformer", + "layer", + "ell", + "output", + "preceding", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "TD", + "Input", + "Layer", + "Encoding", + "Output", + "u1", + "u2", + "uV", + "y1", + "y2", + "Sample", + "generate", + "position", + "style", + "fill", + "f9f", + "stroke", + "333", + "ccf", + "Figure", + "language", + "model", + "decoder", + "stacking", + "blocks", + "mapping", + "predicted" + ], + "keywords": [ + "sampling", + "quality", + "diversity" + ], + "connect_to": [ + { + "target": "5de60f6a-7b06-5f67-a585-29abffb11588", + "relation": "embeds", + "ref": "[images/image-15-Transformer Architecture Diagram.jpg]", + "position": { + "start": 734, + "end": 788 + } + } + ] + } + }, + { + "chunk_id": "91427ec5-385d-59a6-8064-c61a29788ba0", + "type": "text", + "content": "top-k sampling\nTop-k sampling is a simple generalization of greedy decoding. Instead of choosing the single most probable word to generate, we first truncate the distribution to the top k most likely words, renormalize to produce a legitimate probability distribution, and then randomly sample from within these k words according to their renormalized probabilities. More formally:\n1. Choose in advance a number of words $k$\n2. For each word in the vocabulary $V$ , use the language model to compute the likelihood of this word given the context $p(w_{t}|\\mathbf{w}_{< t})$\n3. Sort the words by their likelihood, and throw away any word that is not one of the top k most probable words.\n4. Renormalize the scores of the k words to be a legitimate probability distribu-\ntion.\n5. Randomly sample a word from within these remaining $k$ most-probable words according to its probability.\nWhen k = 1, top-k sampling is identical to greedy decoding. Setting k to a larger number than 1 leads us to sometimes select a word which is not necessarily the most probable, but is still probable enough, and whose choice results in generating more diverse but still high-enough-quality text.", + "path": "8.pdf/CHAPTER 8 Transformers/8.6 More on Sampling/8.6.1 Top-k sampling", + "metadata": { + "length": 1176, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "top", + "sampling", + "Top", + "simple", + "generalization", + "greedy", + "decoding", + "Instead", + "choosing", + "single", + "probable", + "word", + "generate", + "truncate", + "distribution", + "words", + "renormalize", + "produce", + "legitimate", + "probability", + "randomly", + "sample", + "renormalized", + "probabilities", + "More", + "formally", + "Choose", + "advance", + "number", + "For", + "vocabulary", + "language", + "model", + "compute", + "likelihood", + "context", + "mathbf", + "Sort", + "throw", + "Renormalize", + "scores", + "distribu", + "tion", + "Randomly", + "remaining", + "When", + "identical", + "Setting", + "larger", + "leads", + "select", + "necessarily", + "choice", + "results", + "generating", + "diverse", + "high", + "quality", + "text" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "fa832dee-549f-5b42-b3ff-5c1d65e8ae03", + "type": "text", + "content": "One problem with top-k sampling is that k is fixed, but the shape of the probability distribution over words differs in different contexts. If we set k = 10, sometimes the top 10 words will be very likely and include most of the probability mass, but other times the probability distribution will be flatter and the top 10 words will only include a small part of the probability mass.\ntop-p sampling\nAn alternative, called top-p sampling or nucleus sampling (Holtzman et al., 2020), is to keep not the top k words, but the top p percent of the probability mass. The goal is the same; to truncate the distribution to remove the very unlikely words. But by measuring probability rather than the number of words, the hope is that the measure will be more robust in very different contexts, dynamically increasing and decreasing the pool of word candidates.\nGiven a distribution $P(w_{t}|\\mathbf{w}_{0\n[images/image-16-Transformer Language Model Architecture.jpg]\n", + "path": "images/image-16-Transformer Language Model Architecture.jpg", + "metadata": { + "length": 61, + "summary": "image-16\n- Input tokens: So, long, and, thanks, for...\n- Input Encoding: Embedding (E) + Positional Encoding (1, 2, 3...) -> x1, x2, x3...\n- Stacked Transformer Blocks: Multiple layers of processing with skip connections between blocks\n- Language Modeling Head: U projection layer producing logits\n- Loss Calculation: Negative log likelihood (-log y) for each predicted token (long, and, thanks, for, all)\n- Overall Loss Formula: L_CE = (1/T) * Σ(from t=1 to T) L_CE", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-16-Transformer Language Model Architecture.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d7c64d40-e8cc-568e-9ebb-8a512b5696e2", + "type": "text", + "content": "We described the training process for language models in the prior chapter. Recall that large language models are trained with cross-entropy loss, also called the negative log likelihood loss. At time t the cross-entropy loss is the negative log probability the model assigns to the next word in the training sequence, $-\\log p(w_{t+1})$ .\nFig. 8.17 illustrates the general training approach. At each step, given all the preceding words, the final transformer layer produces an output distribution over the entire vocabulary. During training, the probability assigned to the correct word by the model is used to calculate the cross-entropy loss for each item in the sequence. The loss for a training sequence is the average cross-entropy loss over the entire sequence. The weights in the network are adjusted to minimize the average CE loss over the training sequence via gradient descent.\nWith transformers, each training item can be processed in parallel since the output for each element in the sequence is computed separately.\nLarge models are generally trained by filling the full context window (for example 4096 tokens for GPT4 or 8192 for Llama 3) with text. If documents are shorter than this, multiple documents are packed into the window with a special end-of-text token between them. The batch size for gradient descent is usually quite large (the largest GPT-3 model uses a batch size of 3.2 million tokens).\n\n## 8.7 Training
0\n[images/image-16-Transformer Language Model Architecture.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nsubgraph Input Tokens\nA[\"Input Encoding\"] --> B[\"x1\"]\nA --> C[\"x2\"]\nA --> D[\"x3\"]\nA --> E[\"x4\"]\nA --> F[\"x5\"]\nG[\"Stacked Transformer Blocks\"] --> H[\"...\"]\nG --> I[\"...\"]\nG --> J[\"...\"]\nG --> K[\"...\"]\nG --> L[\"...\"]\nM[\"Language Modeling Head\"] --> N[\"...\"]\nM --> O[\"...\"]\nM --> P[\"...\"]\nM --> Q[\"...\"]\nM --> R[\"...\"]\nM --> S[\"...\"]\nT[\"Last Layer\"] --> U[\"...\"]\nT --> V[\"...\"]\nT --> W[\"...\"]\nT --> X[\"...\"]\nT --> Y[\"...\"]\nZ[\"Long Layer\"] --> AA[\"...\"]\nZ --> AB[\"...\"]\nZ --> AC[\"...\"]\nZ --> AD[\"...\"]\nAE[\"Long Layer\"] --> AF[\"...\"]\nAE --> AG[\"...\"]\nAE --> AH[\"...\"]\nAE --> AI[\"...\"]\nAJ[\"Long Layer\"] --> AK[\"...\"]\nAJ --> AL[\"...\"]\nAJ --> AM[\"...\"]\nAJ --> AN[\"...\"]\nAO[\"Long Layer\"] --> AP[\"...\"]\nAO --> AQ[\"...\"]\nAO --> AR[\"...\"]\nAO --> AS[\"...\"]\nAT[\"Long Layer\"] --> AU[\"...\"]\nAT --> AV[\"...\"]\nAT --> AW[\"...\"]\nAX[\"Long Layer\"] --> AY[\"...\"]\nAX --> AZ[\"...\"]\nAX --> BA[\"...\"]\nBB[\"Long Layer\"] --> BC[\"...\"]\nBB --> BD[\"...\"]\nBB --> BE[\"...\"]\nBF[\"Long Layer\"] --> BG[\"...\"]\nBF --> BH[\"...\"]\nBF --> BI[\"...\"]\nBJ[\"Long Layer\"] --> BK[\"...\"]\nBJ --> BL[\"...\"]\nBJ --> BM[\"...\"]\nBN[\"Long Layer\"] --> BO[\"...\"]\nBN --> BP[\"...\"]\nBN --> BQ[\"...\"]\nBR[\"Long Layer\"] --> BS[\"...\"]\nBR --> BT[\"...\"]\nBR --> BU[\"...\"]\nBV[\"Long Layer\"] --> BW[\"...\"]\nBV --> BX[\"...\"]\nBV --> BY[\"...\"]\nBZ[\"Stacked Transformer Blocks\"] --> CA[\"x1\"]\nBZ --> CB[\"x2\"]\nBZ --> CC[\"x3\"]\nBZ --> CD[\"x4\"]\nBZ --> CE[\"x5\"]\nend\nsubgraph Input Tokens\nCF[\"Xo\"] --> CG[\"E\"]\nend\nsubgraph Input Encoding\nCH[\"E\"] --> CI[\"E\"]\nend\nsubgraph Input Tokens\nCJ[\"E\"] --> CK[\"E\"]\nend\nsubgraph Stacked Transformer Blocks\nCL[\"x1\"] --> CM[\"x2\"] & CN[\"x3\"] & CO[\"x4\"] & CE[\"x5\"]\nend\nsubgraph Stacked Transformer Blocks\nCD[\"x1\"] & CN[\"x2\"] & CO[\"x3\"] & CE[\"x4\"] & CE[\"x5\"]\nend\nsubgraph Stacked Transformer Blocks\nDD[\"x1\"] & CM[\"x2\"] & CO[\"x3\"] & CE[\"x4\"] & CE[\"x5\"]\nend\nsubgraph Stacked Transformer Blocks\nDF[\"XO\"] & DG[\"XL\"] --> DH[\"XL0\"], DH[\"XL1\"], DH[\"XL2\"], DH[\"XL3\"], DH[\"XL4\"], DH[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nID[\"XL0\"] & ID[\"XL1\"] & ID[\"XL2\"] & ID[\"XL3\"] & ID[\"XL4\"] & ID[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAE[\"XL0\"] & AE[\"XL1\"] & AE[\"XL2\"] & AE[\"XL3\"] & AE[\"XL4\"] & AE[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAF[\"XL0\"] & AF[\"XL1\"] & AF[\"XL2\"] & AF[\"XL3\"] & AF[\"XL4\"] & AF[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAG[\"XL0\"] & AG[\"XL1\"] & AG[\"XL2\"] & AG[\"XL3\"] & AG[\"XL4\"] & AG[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAH[\"XL0\"] & AH[\"XL1\"] & AH[\"XL2\"] & AH[\"XL3\"] & AH[\"XL4\"] & AH[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAI[\"XL0\"] & AI[\"XL1\"] & AI[\"XL2\"] & AI[\"XL3\"] & AI[\"XL4\"] & AI[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAJ[\"XL0\"] & AJ[\"XL1\"] & AJ[\"XL2\"] & AJ[\"XL3\"] & AJ[\"XL4\"] & AJ[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAK[\"XL0\"] & AK[\"XL1\"] & AK[\"XL2\"] & AK[\"XL3\"] & AK[\"XL4\"] & AK[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAL[\"XL0\"] & AL[\"XL1\"] & AL[\"XL2\"] & AL[\"XL3\"] & AL[\"XL4\"] & AL[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAM[\"XL0\"] & AM[\"XL1\"] & AM[\"XL2\"] & AM[\"XL3\"] & AM[\"XL4\"] & AM[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAN[\"XL0\"] & AN[\"XL1\"] & AN[\"XL2\"] & AN[\"XL3\"] & AN[\"XL4\"] & AN[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAO[\"XL0\"] & AO[\"XL1\"] & AO[\"XL2\"] & AO[\"XL3\"] & AO[\"XL4\"] & AO[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAP[\"XL0\"] & AP[\"XL1\"] & AP[\"XL2\"] & AP[\"XL3\"] & AP[\"XL4\"] & AP[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nAQX[\"XL0\"] & AQX[\"XL1\"] & AQX[\"XL2\"] & AQX[\"XL3\"] & AQX[\"XL4\"] & AQX[\"XL5\"]\nend\nsubgraph Stacked Transformer Blocks\nARX[Yx1, x2, ..., x3, ..., x4, ..., x5, ..., x6, ..., x7, ..., x8, ..., x9, ..., x10, ..., x11, ..., x12, ..., x13, ..., x14, ..., x15, ..., x16, ..., x17, ..., x18, ..., x19, ..., x20, ..., x21, ..., x22, ..., x23, ..., x24, ..., x25, ..., x26, ..., x27, ..., x28, ..., x29, ..., x30, ..., x31, ..., x32, ..., x33, ..., x34, ..., x35, ..., x36, ..., x37, ..., x38, ..., x39, ..., x40, ..., x41, ..., x42, ..., x43, ..., x44, ..., x45, ..., x46, ..., x47, ..., x48, ..., x49, ..., x50, ... |\n```\n
\nFigure 8.17 Training a transformer as a language model.", + "path": "8.pdf/CHAPTER 8 Transformers/8.7 Training", + "metadata": { + "length": 5616, + "summary": "The passage describes the training process for large language models using cross-entropy loss. At each step, the transformer layer outputs a distribution over the vocabulary, and the probability of the correct word is used to compute the loss. Training sequences are processed in parallel, with the average cross-entropy loss minimized via gradient descent. Large models fill the context window with text, packing multiple documents with end-of-text tokens, and use large batch sizes (e.g., 3.2 million tokens for GPT-3).", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "We", + "training", + "process", + "language", + "models", + "prior", + "chapter", + "Recall", + "large", + "trained", + "cross", + "entropy", + "loss", + "called", + "negative", + "log", + "likelihood", + "At", + "time", + "probability", + "model", + "assigns", + "word", + "sequence", + "Fig", + "8.17", + "illustrates", + "general", + "approach", + "step", + "preceding", + "words", + "final", + "transformer", + "layer", + "produces", + "output", + "distribution", + "entire", + "vocabulary", + "During", + "assigned", + "correct", + "calculate", + "item", + "The", + "average", + "weights", + "network", + "adjusted", + "minimize", + "CE", + "gradient", + "descent", + "With", + "transformers", + "processed", + "parallel", + "element", + "computed", + "separately", + "Large", + "generally", + "filling", + "full", + "context", + "window", + "4096", + "tokens", + "GPT4", + "8192", + "Llama", + "text", + "If", + "documents", + "shorter", + "multiple", + "packed", + "special", + "end", + "token", + "batch", + "size", + "largest", + "GPT", + "3.2", + "million", + "8.7", + "Training", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "TD", + "subgraph", + "Input", + "Tokens", + "Encoding", + "x1", + "x2", + "x3", + "x4", + "x5", + "Stacked", + "Transformer", + "Blocks", + "Language", + "Modeling", + "Head", + "Last", + "Layer", + "Long", + "AA", + "AB", + "AC", + "AD", + "AE", + "AF", + "AG", + "AH", + "AI", + "AJ", + "AK", + "AL", + "AM", + "AN", + "AO", + "AP", + "AQ", + "AR", + "AS", + "AT", + "AU", + "AV", + "AW", + "AX", + "AY", + "AZ", + "BA", + "BB", + "BC", + "BD", + "BE", + "BF", + "BG", + "BH", + "BI", + "BJ", + "BK", + "BL", + "BM", + "BN", + "BO", + "BP", + "BQ", + "BR", + "BS", + "BT", + "BU", + "BV", + "BW", + "BX", + "BY", + "BZ", + "CA", + "CB", + "CC", + "CD", + "CF", + "Xo", + "CG", + "CH", + "CI", + "CJ", + "CK", + "CL", + "CM", + "CN", + "CO", + "DD", + "DF", + "XO", + "DG", + "XL", + "DH", + "XL0", + "XL1", + "XL2", + "XL3", + "XL4", + "XL5", + "ID", + "AQX", + "ARX", + "Yx1", + "x6", + "x7", + "x8", + "x9", + "x10", + "x11", + "x12", + "x13", + "x14", + "x15", + "x16", + "x17", + "x18", + "x19", + "x20", + "x21", + "x22", + "x23", + "x24", + "x25", + "x26", + "x27", + "x28", + "x29", + "x30", + "x31", + "x32", + "x33", + "x34", + "x35", + "x36", + "x37", + "x38", + "x39", + "x40", + "x41", + "x42", + "x43", + "x44", + "x45", + "x46", + "x47", + "x48", + "x49", + "x50", + "Figure" + ], + "keywords": [ + "cross-entropy loss", + "gradient descent", + "batch size" + ], + "connect_to": [ + { + "target": "8c3c4bcc-1b48-585e-861b-197bda1fa206", + "relation": "embeds", + "ref": "[images/image-16-Transformer Language Model Architecture.jpg]", + "position": { + "start": 1450, + "end": 1511 + } + } + ] + } + }, + { + "chunk_id": "632928a5-fe02-5c55-8329-bfe79fb0cf78", + "type": "text", + "content": "Large language models are large. For example the Llama 3.1 405B Instruct model from Meta has 405 billion parameters (it has L=126 layers, model dimensionality d=16,384, and A=128 attention heads) and was trained on 15.6 terabytes of text tokens using a vocabulary of 128K tokens (Llama Team, 2024). So there is a lot of research on understanding how LLMs scale, and especially how to implement them given limited resources. In the next few sections we discuss how to think about scale (the concept of scaling laws), and important techniques for getting language models to work efficiently, such as the KV cache and parameter-efficient fine tuning (PEFT).", + "path": "8.pdf/CHAPTER 8 Transformers/8.8 Dealing with Scale", + "metadata": { + "length": 654, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "Large", + "language", + "models", + "large", + "For", + "Llama", + "3.1", + "405B", + "Instruct", + "model", + "Meta", + "405", + "billion", + "parameters", + "126", + "layers", + "dimensionality", + "16", + "384", + "128", + "attention", + "heads", + "trained", + "15.6", + "terabytes", + "text", + "tokens", + "vocabulary", + "128K", + "Team", + "2024", + "So", + "lot", + "research", + "understanding", + "LLMs", + "scale", + "implement", + "limited", + "resources", + "In", + "sections", + "discuss", + "concept", + "scaling", + "laws", + "important", + "techniques", + "work", + "efficiently", + "KV", + "cache", + "parameter", + "efficient", + "fine", + "tuning", + "PEFT" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cb68a4da-dcf1-5a92-ac51-4499257fe233", + "type": "text", + "content": "The performance of large language models has shown to be mainly determined by 3 factors: model size (the number of parameters not counting embeddings), dataset size (the amount of training data), and the amount of compute used for training. That is, we can improve a model by adding parameters (adding more layers or having wider contexts or both), by training on more data, or by training for more iterations.\nThe relationships between these factors and performance are known as scaling laws. Roughly speaking, the performance of a large language model (the loss) scales as a power-law with each of these three properties of model training.\nFor example, Kaplan et al. (2020) found the following three relationships for loss L as a function of the number of non-embedding parameters N, the dataset size D, and the compute budget C, for models training with limited parameters, dataset,\nscaling laws\nor compute budget, if in each case the other two properties are held constant:\n$$\nL (N) = \\left(\\frac {N _ {c}}{N}\\right) ^ {\\alpha_ {N}} \\tag {8.49}\n$$\n$$\nL (D) = \\left(\\frac {D _ {c}}{D}\\right) ^ {\\alpha_ {D}} \\tag {8.50}\n$$\n$$\nL (C) = \\left(\\frac {C _ {c}}{C}\\right) ^ {\\alpha_ {C}} \\tag {8.51}\n$$\nThe number of (non-embedding) parameters N can be roughly computed as follows (ignoring biases, and with d as the input and output dimensionality of the model, $d_{attn}$ as the self-attention layer size, and $d_{ff}$ the size of the feedforward layer):\n$$\nN \\approx 2 d n _ {\\text { layer }} (2 d _ {\\text { attn }} + d _ {\\text { ff }})\n$$\n$$\n\\approx 1 2 n _ {\\text { layer }} d ^ {2} \\tag {8.52}\n$$\n$$\n(\\text { assuming } d _ {\\mathrm{attn}} = d _ {\\mathrm{ff}} / 4 = d)\n$$\nThus GPT-3, with $n = 96$ layers and dimensionality $d = 12288$ , has $12 \\times 96 \\times 12288^2 \\approx 175$ billion parameters.\nThe values of $N_{c}$ , $D_{c}$ , $C_{c}$ , $\\alpha_{N}$ , $\\alpha_{D}$ , and $\\alpha_{C}$ depend on the exact transformer architecture, tokenization, and vocabulary size, so rather than all the precise values, scaling laws focus on the relationship with loss. $^{3}$\nScaling laws can be useful in deciding how to train a model to a particular performance, for example by looking at early in the training curve, or performance with smaller amounts of data, to predict what the loss would be if we were to add more data or increase model size. Other aspects of scaling laws can also tell us how much data we need to add when scaling up a model.", + "path": "8.pdf/CHAPTER 8 Transformers/8.8 Dealing with Scale/8.8.1 Scaling laws", + "metadata": { + "length": 2454, + "summary": "The passage discusses scaling laws for large language models, which describe how performance (loss) scales as a power-law with model size (non-embedding parameters N), dataset size (D), and compute budget (C). It provides equations for loss as a function of each factor when others are held constant, and gives a formula for estimating the number of non-embedding parameters. The example of GPT-3 with 175 billion parameters is included. Scaling laws help predict performance and guide decisions on training data and model size.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "The", + "performance", + "large", + "language", + "models", + "shown", + "determined", + "factors", + "model", + "size", + "number", + "parameters", + "counting", + "embeddings", + "dataset", + "amount", + "training", + "data", + "compute", + "That", + "improve", + "adding", + "layers", + "wider", + "contexts", + "iterations", + "relationships", + "scaling", + "laws", + "Roughly", + "speaking", + "loss", + "scales", + "power", + "law", + "properties", + "For", + "Kaplan", + "al", + "2020", + "found", + "function", + "embedding", + "budget", + "limited", + "case", + "held", + "constant", + "left", + "frac", + "alpha", + "tag", + "8.49", + "8.50", + "8.51", + "roughly", + "computed", + "ignoring", + "biases", + "input", + "output", + "dimensionality", + "attn", + "attention", + "layer", + "ff", + "feedforward", + "approx", + "text", + "8.52", + "assuming", + "mathrm", + "Thus", + "GPT", + "96", + "12288", + "12", + "times", + "175", + "billion", + "values", + "depend", + "exact", + "transformer", + "architecture", + "tokenization", + "vocabulary", + "precise", + "focus", + "relationship", + "Scaling", + "deciding", + "train", + "early", + "curve", + "smaller", + "amounts", + "predict", + "add", + "increase", + "Other", + "aspects" + ], + "keywords": [ + "scaling laws", + "model size", + "dataset size" + ], + "connect_to": [] + } + }, + { + "chunk_id": "6c81f386-8261-5df0-b76e-d8c322827536", + "type": "image", + "content": "\n$$
0\n[images/image-17-Attention Mechanism Matrix Multiplication.jpg]\n", + "path": "images/image-17-Attention Mechanism Matrix Multiplication.jpg", + "metadata": { + "length": 48, + "summary": "image-17\nThe image illustrates the mathematical process of calculating attention scores and weighted values using matrix multiplication.\n\n1. **Query Vector (Q):** A single query vector `q4` with dimensions $1 \\times d_k$.\n2. **Key Matrix Transpose ($K^T$):** A matrix containing keys $k_1, k_2, k_3, k_4$ with dimensions $d_k \\times N$.\n3. **Dot Product ($QK^T$):** The result of multiplying Q by $K^T$, producing a row vector of dot products ($q4\\cdot k1, q4\\cdot k2, q4\\cdot k3, q4\\cdot k4$) with dimensions $1 \\times N$.\n4. **Value Matrix (V):** A matrix containing values $v_1, v_2, v_3, v_4$ with dimensions $N \\times d_v$.\n5. **Final Output (A):** The result of multiplying the dot product matrix by V, yielding a final output vector `a4` with dimensions $1", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-17-Attention Mechanism Matrix Multiplication.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "6e7590a7-d0be-5ada-8f7f-fb6e586f8d4d", + "type": "text", + "content": "We saw in Fig. 8.11 and in Eq. 8.34 (repeated below) how the attention vector can be very efficiently computed in parallel for training, via two matrix multiplications:\n$$\n\\mathbf {A} = \\text { softmax } \\left(\\frac {\\mathbf {Q} \\mathbf {K} ^ {\\intercal}}{\\sqrt {d _ {k}}}\\right) \\mathbf {V} \\tag {8.53}\n$$\nUnfortunately we can't do quite the same efficient computation in inference as in training. That's because at inference time, we iteratively generate the next tokens one at a time. For a new token that we have just generated, call it $\\mathbf{x}_i$ , we need to compute its query, key, and values by multiplying by $\\mathbf{W}^{\\mathbf{Q}}$ , $\\mathbf{W}^{\\mathbf{K}}$ , and $\\mathbf{W}^{\\mathbf{V}}$ respectively. But it would be a waste of computation time to recompute the key and value vectors for all the prior tokens $\\mathbf{x}_{0\n[images/image-17-Attention Mechanism Matrix Multiplication.jpg]\n\n
\ntext_image\nQ\nX\nk1 k2 k3 k4 = QK^T\nq4", + "path": "8.pdf/CHAPTER 8 Transformers/8.8 Dealing with Scale/8.8.2 KV Cache", + "metadata": { + "length": 1387, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "We", + "Fig", + "8.11", + "Eq", + "8.34", + "repeated", + "attention", + "vector", + "efficiently", + "computed", + "parallel", + "training", + "matrix", + "multiplications", + "mathbf", + "text", + "softmax", + "left", + "frac", + "intercal", + "sqrt", + "tag", + "8.53", + "Unfortunately", + "efficient", + "computation", + "inference", + "That", + "time", + "iteratively", + "generate", + "tokens", + "For", + "token", + "generated", + "call", + "compute", + "query", + "key", + "values", + "multiplying", + "But", + "waste", + "recompute", + "vectors", + "prior", + "steps", + "So", + "recomputing", + "store", + "memory", + "KV", + "cache", + "grab", + "8.18", + "modifies", + "show", + "takes", + "place", + "single", + "showing", + "details", + "summary", + "image", + "k1", + "k2", + "k3", + "k4", + "QK", + "q4" + ], + "keywords": [], + "connect_to": [ + { + "target": "6c81f386-8261-5df0-b76e-d8c322827536", + "relation": "embeds", + "ref": "[images/image-17-Attention Mechanism Matrix Multiplication.jpg]", + "position": { + "start": 1288, + "end": 1351 + } + } + ] + } + }, + { + "chunk_id": "82eb92b1-4186-582c-aa11-631c45cad190", + "type": "text", + "content": "dk x N", + "path": "8.pdf/CHAPTER 8 Transformers/8.8 Dealing with Scale/8.8.2 KV Cache/1 x dk", + "metadata": { + "length": 6, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "83c01058-d48b-5f22-b304-c7572551ef66", + "type": "text", + "content": "V\nx\nv1\nv2\nv3\nv4\nN x dv = A\na4", + "path": "8.pdf/CHAPTER 8 Transformers/8.8 Dealing with Scale/8.8.2 KV Cache/1 x N", + "metadata": { + "length": 29, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "v1", + "v2", + "v3", + "v4", + "dv", + "a4" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e7716ae2-e525-5084-aba7-aa2e3812b355", + "type": "text", + "content": "
\nFigure 8.18 Parts of the attention computation (extracted from Fig. 8.11) showing, in black, the vectors that can be stored in the cache rather than recomputed when computing the attention score for the 4th token.", + "path": "8.pdf/CHAPTER 8 Transformers/8.8 Dealing with Scale/8.8.2 KV Cache/1 x dv", + "metadata": { + "length": 224, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "details", + "Figure", + "8.18", + "Parts", + "attention", + "computation", + "extracted", + "Fig", + "8.11", + "showing", + "black", + "vectors", + "stored", + "cache", + "recomputed", + "computing", + "score", + "4th", + "token" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6b6445af-0d60-56d4-bb42-9f96bfb50197", + "type": "image", + "content": "\n$$
0\n[images/image-18-Pretrained Weights Decomposition.jpg]\n", + "path": "images/image-18-Pretrained Weights Decomposition.jpg", + "metadata": { + "length": 48, + "summary": "image-18\nThe image illustrates a mathematical decomposition of a pretrained weight matrix W.\n\n- **Central Component**: A green square labeled \"Pretrained Weights\" with the symbol **W**. Its dimensions are defined as **k** (height) and **d** (width).\n- **Decomposition Structure**: The diagram shows that W is approximated by the sum of two components:\n 1. **Top Component**: A yellow horizontal rectangle labeled **h**, with dimensions **1** x **d**.\n 2. **Right Component**: The product of two red matrices, **A** and **B**.\n - Matrix **A** is a vertical rectangle with dimensions **k** x **r**.\n - Matrix **B** is a horizontal rectangle with dimensions **r** x **d**.\n- **Input Vector**: At the bottom, a blue horizontal vector labeled **x** is shown with dimensions **1** x **d**.\n- **Arrows**: Arrows indicate that the input **x** interacts with the decomposed components to reconstruct or approximate the original weights **W**.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-18-Pretrained Weights Decomposition.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "c369905a-3415-5608-96f5-8dc41fe2c09c", + "type": "text", + "content": "As we mentioned above, it's very common to take a language model and give it more information about a new domain by finetuning it (continuing to train it to predict upcoming words) on some additional data.\nFine-tuning can be very difficult with very large language models, because there are enormous numbers of parameters to train; each pass of batch gradient descent has to backpropagate through many many huge layers. This makes finetuning huge language models extremely expensive in processing power, in memory, and in time. For this reason, there are alternative methods that allow a model to be finetuned without changing all the parameters. Such methods are called parameter-efficient fine tuning or sometimes PEFT, because we efficiently select a subset of parameters to update when finetuning. For example we freeze some of the parameters (don't change them), and only update some particular subset of parameters.\nHere we describe one such model, called LoRA, for Low-Rank Adaptation. The intuition of LoRA is that transformers have many dense layers which perform matrix multiplication (for example the $W^{Q}$ , $W^{K}$ , $W^{V}$ , $W^{O}$ layers in the attention computation). Instead of updating these layers during finetuning, with LoRA we freeze these layers and instead update a low-rank approximation that has fewer parameters.\nConsider a matrix W of dimensionality $[k \\times d]$ that needs to be updated during finetuning via gradient descent. Normally this matrix would get updates $\\Delta W$ of dimensionality $[k \\times d]$ , for updating the $k \\times d$ parameters after gradient descent. In LoRA, we freeze W and update instead a low-rank decomposition of W. We create two matrices A and B, where A has size $[k \\times r]$ and B has size $[r \\times d]$ , and we choose r to be quite small, $r << \\min(d, k)$ . During finetuning we update A and B instead of W. That is, we replace $W + \\Delta W$ with $W + AB$ . Fig. 8.19 shows the intuition. For replacing the forward pass h = xW, the new forward pass is instead:\n$$\n\\mathbf {h} = \\mathbf {x W} + \\mathbf {x A B} \\tag {8.54}\n$$\nLoRA has a number of advantages. It dramatically reduces hardware requirements, since gradients don't have to be calculated for most parameters. The weight updates can be simply added in to the pretrained weights, since AB is the same size as W). That means it doesn't add any time during inference. And it also means it's possible to build LoRA modules for different domains and just swap them in and out by adding them in or subtracting them from W.\nIn its original version LoRA was applied just to the matrices in the attention computation (the $W^{Q}$ , $W^{K}$ , $W^{V}$ , and $W^{O}$ layers). Many variants of LoRA exist.\n\n$$
0\n[images/image-18-Pretrained Weights Decomposition.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nh[\"1\"] -->|d| A[\"k A\"]\nh -->|d| B[\"x r B\"]\nA -->|k| W[\"Pretrained Weights w\"]\nB -->|d| X[\"x 1\"]\nstyle h fill:#f9f,stroke:#333\nstyle B fill:#bbf,stroke:#333\nstyle W fill:#cfc,stroke:#333\nstyle X fill:#fcc,stroke:#333\n```\n
\nFigure 8.19 The intuition of LoRA. We freeze W to its pretrained values, and instead fine-tune by training a pair of matrices A and B, updating those instead of W, and just sum W and the updated AB.", + "path": "8.pdf/CHAPTER 8 Transformers/8.8 Dealing with Scale/8.8.3 Parameter Efficient Fine Tuning", + "metadata": { + "length": 3267, + "summary": "LoRA (Low-Rank Adaptation) is a parameter-efficient fine-tuning method for large language models. Instead of updating all parameters during fine-tuning, LoRA freezes the pretrained weight matrices (like W^Q, W^K, W^V, W^O in attention layers) and trains low-rank decomposition matrices A and B. This reduces hardware requirements, speeds up training, and allows easy swapping of fine-tuned modules without adding inference time.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "As", + "mentioned", + "common", + "language", + "model", + "give", + "information", + "domain", + "finetuning", + "continuing", + "train", + "predict", + "upcoming", + "words", + "additional", + "data", + "Fine", + "tuning", + "difficult", + "large", + "models", + "enormous", + "numbers", + "parameters", + "pass", + "batch", + "gradient", + "descent", + "backpropagate", + "huge", + "layers", + "This", + "makes", + "extremely", + "expensive", + "processing", + "power", + "memory", + "time", + "For", + "reason", + "alternative", + "methods", + "finetuned", + "changing", + "Such", + "called", + "parameter", + "efficient", + "fine", + "PEFT", + "efficiently", + "select", + "subset", + "update", + "freeze", + "don", + "change", + "Here", + "describe", + "LoRA", + "Low", + "Rank", + "Adaptation", + "The", + "intuition", + "transformers", + "dense", + "perform", + "matrix", + "multiplication", + "attention", + "computation", + "Instead", + "updating", + "low", + "rank", + "approximation", + "fewer", + "Consider", + "dimensionality", + "times", + "updated", + "Normally", + "updates", + "Delta", + "In", + "decomposition", + "We", + "create", + "matrices", + "size", + "choose", + "small", + "min", + "During", + "That", + "replace", + "AB", + "Fig", + "8.19", + "shows", + "replacing", + "forward", + "xW", + "mathbf", + "tag", + "8.54", + "number", + "advantages", + "It", + "dramatically", + "reduces", + "hardware", + "requirements", + "gradients", + "calculated", + "weight", + "simply", + "added", + "pretrained", + "weights", + "means", + "doesn", + "add", + "inference", + "And", + "build", + "modules", + "domains", + "swap", + "adding", + "subtracting", + "original", + "version", + "applied", + "Many", + "variants", + "exist", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "TD", + "Pretrained", + "Weights", + "style", + "fill", + "f9f", + "stroke", + "333", + "bbf", + "cfc", + "fcc", + "Figure", + "values", + "tune", + "training", + "pair", + "sum" + ], + "keywords": [ + "LoRA", + "Parameter-Efficient", + "Fine-tuning" + ], + "connect_to": [ + { + "target": "6b6445af-0d60-56d4-bb42-9f96bfb50197", + "relation": "embeds", + "ref": "[images/image-18-Pretrained Weights Decomposition.jpg]", + "position": { + "start": 2745, + "end": 2799 + } + } + ] + } + }, + { + "chunk_id": "9e38b0bc-3309-53e1-923a-ab9ff5ae216a", + "type": "text", + "content": "interpretability\nHow does a transformer-based language model manage to do so well at language tasks? The subfield of interpretability, sometimes called mechanistic interpretability, focuses on ways to understand mechanistically what is going on inside the transformer. In the next two subsections we discuss two well-studied aspects of transformer interpretability.", + "path": "8.pdf/CHAPTER 8 Transformers/8.9 Interpreting the Transformer", + "metadata": { + "length": 365, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "interpretability", + "How", + "transformer", + "based", + "language", + "model", + "manage", + "tasks", + "The", + "subfield", + "called", + "mechanistic", + "focuses", + "ways", + "understand", + "mechanistically", + "inside", + "In", + "subsections", + "discuss", + "studied", + "aspects" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a7f75d28-d16e-5507-a85d-b3663a9d5742", + "type": "image", + "content": "\nThe term in-context learning was first proposed by Brown et al. (2020) in their introduction of the GPT3 system, to refer to either of these kinds of learning that language models do from their prompts. In-context learning means language models learning to do new tasks, better predict tokens, or generally reduce their loss during the forward-pass at inference-time, without any gradient-based updates to the model's parameters.
0\n[images/image-19-Prefix matching and copying.jpg]\n", + "path": "images/image-19-Prefix matching and copying.jpg", + "metadata": { + "length": 475, + "summary": "image-19\nThe diagram illustrates a sequence alignment process involving two sentences: \"She owns vintage cars.\" and \"He dreams of owning vintage cars\".\n- Prefix matching: The word \"cars\" in the first sentence is matched with the word \"cars\" in the second sentence.\n- Copying: The word \"cars\" from the first sentence is copied to the corresponding position in the second sentence.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "file_path": "images/image-19-Prefix matching and copying.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "a5f59c1f-a097-50a5-ba32-b07d2edd344d", + "type": "text", + "content": "As a way of getting a model to do what we want, we can think of prompting as being fundamentally different than pretraining. Learning via pretraining means updating the model's parameters by using gradient descent according to some loss function. But prompting with demonstrations can teach a model to do a new task. The model is learning something about the task from those demonstrations as it processes the prompt.\nEven without demonstrations, we can think of the process of prompting as a kind of learning. For example, the further a model gets in a prompt, the better it tends to get at predicting the upcoming tokens. The information in the context is helping give the model more predictive power.\nThe term in-context learning was first proposed by Brown et al. (2020) in their introduction of the GPT3 system, to refer to either of these kinds of learning that language models do from their prompts. In-context learning means language models learning to do new tasks, better predict tokens, or generally reduce their loss during the forward-pass at inference-time, without any gradient-based updates to the model's parameters.\nin-context learning\ninduction heads\nHow does in-context learning work? While we don't know for sure, there are some intriguing ideas. One hypothesis is based on the idea of induction heads (Elhage et al., 2021; Olsson et al., 2022). Induction heads are the name for a circuit, which is a kind of abstract component of a network. The induction head circuit is part of the attention computation in transformers, discovered by looking at mini language models with only 1-2 attention heads.\nThe function of the induction head is to predict repeated sequences. For example if it sees the pattern AB...A in an input sequence, it predicts that B will follow, instantiating the pattern completion rule AB...A→B. It does this by having a prefix matching component of the attention computation that, when looking at the current token A, searches back over the context to find a prior instance of A. If it finds one, the induction head has a copying mechanism that “copies” the token B that followed the earlier A, by increasing the probability the B will occur next. Fig. 8.20 shows an example.\n\nThe term in-context learning was first proposed by Brown et al. (2020) in their introduction of the GPT3 system, to refer to either of these kinds of learning that language models do from their prompts. In-context learning means language models learning to do new tasks, better predict tokens, or generally reduce their loss during the forward-pass at inference-time, without any gradient-based updates to the model's parameters.
0\n[images/image-19-Prefix matching and copying.jpg]\n\n
\nflowchart\n```mermaid\ngraph TD\nA[\"She\"] --> B[\"owns\"]\nB --> C[\"vintage\"]\nC --> D[\"cars\"]\nD --> E[\".\"]\nE --> F[\"He\"]\nF --> G[\"dreams\"]\nG --> H[\"of\"]\nH --> I[\"owning\"]\nI --> J[\"vintage\"]\nJ --> K[\"cars\"]\nK --> L[\".\"]\nL --> M[\"Copying\"]\nstyle A fill:#f9f,stroke:#333\nstyle B fill:#f9f,stroke:#333\nstyle C fill:#f9f,stroke:#333\nstyle D fill:#ccf,stroke:#333\nstyle E fill:#ccf,stroke:#333\nstyle F fill:#ccf,stroke:#333\nstyle G fill:#ccf,stroke:#333\nstyle H fill:#ccf,stroke:#333\nstyle I fill:#ccf,stroke:#333\nstyle J fill:#ccf,stroke:#333\nstyle K fill:#ccf,stroke:#333\nnote1[\"Prefix matching\"] --> C\n```\n
\nFigure 8.20 An induction head looking at vintage uses the prefix matching mechanism to find a prior instance of vintage, and the copying mechanism to predict that cars will occur again. Figure from Crosbie and Shutova (2022).\nablating\nOlsson et al. (2022) propose that a generalized fuzzy version of this pattern completion rule, implementing a rule like $A^{*}B^{*}\\ldots A\\rightarrow B$ , where $A^{*}\\approx A$ and $B^{*}\\approx B$ (by $\\approx$ we mean they are semantically similar in some way), might be responsible for in-context learning. Suggestive evidence for their hypothesis comes from Crosbie and Shutova (2022), who show that ablating induction heads causes in-context learning performance to decrease. Ablation is originally a medical term meaning the removal of something. We use it in NLP interpretability studies as a tool for testing causal effects; if we knock out a hypothesized cause, we would expect the effect to disappear. Crosbie and Shutova (2022) ablate induction heads by first finding attention heads that perform as induction heads on random input sequences, and then zeroing out the output of these heads by setting certain terms of the output matrix $W^{0}$ to zero. Indeed they find that ablated models are much worse at in-context learning: they have much worse performance at learning from demonstrations in the prompts.", + "path": "8.pdf/CHAPTER 8 Transformers/8.9 Interpreting the Transformer/8.9.1 In-Context Learning and Induction Heads", + "metadata": { + "length": 4689, + "summary": "The passage discusses in-context learning, a concept introduced by Brown et al. (2020) with GPT-3, where language models learn from prompts during inference without gradient updates. It contrasts this with pretraining, which updates parameters via gradient descent. A key hypothesis for how in-context learning works involves induction heads, circuits in transformer attention that predict repeated sequences by matching prefixes and copying subsequent tokens. Olsson et al. (2022) propose a generalized fuzzy pattern completion rule, and Crosbie and Shutova (2022) provide evidence by showing that ablating induction heads reduces in-context learning performance.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "As", + "model", + "prompting", + "fundamentally", + "pretraining", + "Learning", + "means", + "updating", + "parameters", + "gradient", + "descent", + "loss", + "function", + "But", + "demonstrations", + "teach", + "task", + "The", + "learning", + "processes", + "prompt", + "Even", + "process", + "kind", + "For", + "predicting", + "upcoming", + "tokens", + "information", + "context", + "helping", + "give", + "predictive", + "power", + "term", + "proposed", + "Brown", + "al", + "2020", + "introduction", + "GPT3", + "system", + "refer", + "kinds", + "language", + "models", + "prompts", + "In", + "tasks", + "predict", + "generally", + "reduce", + "forward", + "pass", + "inference", + "time", + "based", + "updates", + "induction", + "heads", + "How", + "work", + "While", + "don", + "intriguing", + "ideas", + "One", + "hypothesis", + "idea", + "Elhage", + "2021", + "Olsson", + "2022", + "Induction", + "circuit", + "abstract", + "component", + "network", + "head", + "part", + "attention", + "computation", + "transformers", + "discovered", + "mini", + "repeated", + "sequences", + "sees", + "pattern", + "AB", + "input", + "sequence", + "predicts", + "follow", + "instantiating", + "completion", + "rule", + "It", + "prefix", + "matching", + "current", + "token", + "searches", + "back", + "find", + "prior", + "instance", + "If", + "finds", + "copying", + "mechanism", + "copies", + "earlier", + "increasing", + "probability", + "occur", + "Fig", + "8.20", + "shows", + "details", + "summary", + "flowchart", + "mermaid", + "graph", + "TD", + "She", + "owns", + "vintage", + "cars", + "He", + "dreams", + "owning", + "Copying", + "style", + "fill", + "f9f", + "stroke", + "333", + "ccf", + "note1", + "Prefix", + "Figure", + "An", + "Crosbie", + "Shutova", + "ablating", + "propose", + "generalized", + "fuzzy", + "version", + "implementing", + "ldots", + "rightarrow", + "approx", + "semantically", + "similar", + "responsible", + "Suggestive", + "evidence", + "show", + "performance", + "decrease", + "Ablation", + "originally", + "medical", + "meaning", + "removal", + "We", + "NLP", + "interpretability", + "studies", + "tool", + "testing", + "causal", + "effects", + "knock", + "hypothesized", + "expect", + "effect", + "disappear", + "ablate", + "finding", + "perform", + "random", + "zeroing", + "output", + "setting", + "terms", + "matrix", + "Indeed", + "ablated", + "worse" + ], + "keywords": [ + "in-context learning", + "induction heads", + "prompting" + ], + "connect_to": [ + { + "target": "a7f75d28-d16e-5507-a85d-b3663a9d5742", + "relation": "embeds", + "ref": "[images/image-19-Prefix matching and copying.jpg]", + "position": { + "start": 2661, + "end": 2710 + } + } + ] + } + }, + { + "chunk_id": "30bff2da-bf25-5dd1-b542-5f288ddf6d95", + "type": "text", + "content": "logit lens\nAnother useful interpretability tool, the logit lens (Nostalgebraist, 2020), offers a way to visualize what the internal layers of the transformer might be representing.\nThe idea is that we take any vector from any layer of the transformer and, pretending that it is the prefinal embedding, simply multiply it by the unembedding layer to get logits, and compute a softmax to see the distribution over words that that vector might be representing. This can be a useful window into the internal representations of the model. Since the network wasn't trained to make the internal representations function in this way, the logit lens doesn't always work perfectly, but this can still be a useful trick to help us visualize the internal layers of a transformer.", + "path": "8.pdf/CHAPTER 8 Transformers/8.9 Interpreting the Transformer/8.9.2 Logit Lens", + "metadata": { + "length": 767, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "logit", + "lens", + "Another", + "interpretability", + "tool", + "Nostalgebraist", + "2020", + "offers", + "visualize", + "internal", + "layers", + "transformer", + "representing", + "The", + "idea", + "vector", + "layer", + "pretending", + "prefinal", + "embedding", + "simply", + "multiply", + "unembedding", + "logits", + "compute", + "softmax", + "distribution", + "words", + "This", + "window", + "representations", + "model", + "Since", + "network", + "wasn", + "trained", + "make", + "function", + "doesn", + "work", + "perfectly", + "trick" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "01cde55e-8527-54b9-9c86-dba0cc08a844", + "type": "text", + "content": "This chapter has introduced the transformer and its components for the language modeling task introduced in the previous chapter. Here's a summary of the main points that we covered:\n- Transformers are non-recurrent networks based on multi-head attention, a kind of self-attention. A multi-head attention computation takes an input vector $\\mathbf{x}_i$ and maps it to an output $\\mathbf{a}_i$ by adding in vectors from prior tokens, weighted by how relevant they are for the processing of the current word.\n- A transformer block consists of a residual stream in which the input from the prior layer is passed up to the next layer, with the output of different components added to it. These components include a multi-head attention layer followed by a feedforward layer, each preceded by layer normalizations. Transformer blocks are stacked to make deeper and more powerful networks.\n- The input to a transformer is computed by adding an embedding (computed with an embedding matrix) to a positional encoding that represents the sequential position of the token in the window.\n- Language models can be built out of stacks of transformer blocks, with a language model head at the top, which applies an unembedding matrix to the output H of the top layer to generate the logits, which are then passed through a softmax to generate word probabilities.\n- Transformer-based language models have a wide context window (200K tokens or even more for very large models with special mechanisms) allowing them to draw on enormous amounts of context to predict upcoming words.\n- There are various computational tricks for making large language models more efficient, such as the KV cache and parameter-efficient finetuning.", + "path": "8.pdf/CHAPTER 8 Transformers/8.10 Summary", + "metadata": { + "length": 1712, + "summary": "This chapter introduces the transformer and its components for language modeling. Key points include: transformers are non-recurrent networks using multi-head self-attention; a transformer block consists of a residual stream with multi-head attention and feedforward layers, each preceded by layer normalization; input is computed by adding token embeddings to positional encodings; language models stack transformer blocks with a language model head that generates logits via an unembedding matrix and softmax; these models have wide context windows (up to 200K tokens or more); efficiency tricks include KV cache and parameter-efficient finetuning.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "This", + "chapter", + "introduced", + "transformer", + "components", + "language", + "modeling", + "task", + "previous", + "Here", + "summary", + "main", + "points", + "covered", + "Transformers", + "recurrent", + "networks", + "based", + "multi", + "head", + "attention", + "kind", + "computation", + "takes", + "input", + "vector", + "mathbf", + "maps", + "output", + "adding", + "vectors", + "prior", + "tokens", + "weighted", + "relevant", + "processing", + "current", + "word", + "block", + "consists", + "residual", + "stream", + "layer", + "passed", + "added", + "These", + "include", + "feedforward", + "preceded", + "normalizations", + "Transformer", + "blocks", + "stacked", + "make", + "deeper", + "powerful", + "The", + "computed", + "embedding", + "matrix", + "positional", + "encoding", + "represents", + "sequential", + "position", + "token", + "window", + "Language", + "models", + "built", + "stacks", + "model", + "top", + "applies", + "unembedding", + "generate", + "logits", + "softmax", + "probabilities", + "wide", + "context", + "200K", + "large", + "special", + "mechanisms", + "allowing", + "draw", + "enormous", + "amounts", + "predict", + "upcoming", + "words", + "There", + "computational", + "tricks", + "making", + "efficient", + "KV", + "cache", + "parameter", + "finetuning" + ], + "keywords": [ + "Transformer", + "Multi-head Attention", + "Language Model" + ], + "connect_to": [] + } + }, + { + "chunk_id": "5c61f648-fb0a-5d26-b97e-f4c25b8401b3", + "type": "text", + "content": "The transformer (Vaswani et al., 2017) was developed drawing on two lines of prior research: self-attention and memory networks.\nEncoder-decoder attention, the idea of using a soft weighting over the encodings of input words to inform a generative decoder (see Chapter 12) was developed by Graves (2013) in the context of handwriting generation, and Bahdanau et al. (2015) for MT. This idea was extended to self-attention by dropping the need for separate encoding and decoding sequences and instead seeing attention as a way of weighting the tokens in collecting information passed from lower layers to higher layers (Ling et al., 2015; Cheng et al., 2016; Liu et al., 2016).\nOther aspects of the transformer, including the terminology of key, query, and value, came from memory networks, a mechanism for adding an external read-write memory to networks, by using an embedding of a query to match keys representing content in an associative memory (Sukhbaatar et al., 2015; Weston et al., 2015; Graves et al., 2014).\nMORE HISTORY TBD IN NEXT DRAFT.\nBa, J. L., J. R. Kiros, and G. E. Hinton. 2016. Layer normalization. NeurIPS workshop.\nBahdanau, D., K. H. Cho, and Y. Bengio. 2015. Neural machine translation by jointly learning to align and translate. ICLR 2015.\nBrown, T., B. Mann, N. Ryder, M. Subbiah, J. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert-Voss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. M. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. Litwin, S. Gray, B. Chess, J. Clark, C. Berner, S. McCandlish, A. Radford, I. Sutskever, and D. Amodei. 2020. Language models are few-shot learners. NeurIPS, volume 33.\nCheng, J., L. Dong, and M. Lapata. 2016. Long short-term memory-networks for machine reading. EMNLP.\nCrosbie, J. and E. Shutova. 2022. Induction heads as an essential mechanism for pattern matching in in-context learning. ArXiv preprint.\nElhage, N., N. Nanda, C. Olsson, T. Henighan, N. Joseph, B. Mann, A. Askell, Y. Bai, A. Chen, T. Conerly, N. Das-Sarma, D. Drain, D. Ganguli, Z. Hatfield-Dodds, D. Hernandez, A. Jones, J. Kernion, L. Lovitt, K. Ndousse, D. Amodei, T. Brown, J. Clark, J. Kaplan, S. McCandlish, and C. Olah. 2021. A mathematical framework for transformer circuits. White paper.\nGraves, A. 2013. Generating sequences with recurrent neural networks. ArXiv.\nGraves, A., G. Wayne, and I. Danihelka. 2014. Neural Turing machines. ArXiv.\nHoltzman, A., J. Buys, L. Du, M. Forbes, and Y. Choi. 2020. The curious case of neural text degeneration. ICLR.\nKaplan, J., S. McCandlish, T. Henighan, T. B. Brown, B. Chess, R. Child, S. Gray, A. Radford, J. Wu, and D. Amodei. 2020. Scaling laws for neural language models. ArXiv preprint.\nLing, W., C. Dyer, A. W. Black, I. Trancoso, R. Fernandez, S. Amir, L. Marujo, and T. Luís. 2015. Finding function in form: Compositional character models for open vocabulary word representation. EMNLP.\nLiu, Y., C. Sun, L. Lin, and X. Wang. 2016. Learning natural language inference using bidirectional LSTM model and inner-attention. ArXiv.\nLlama Team. 2024. The llama 3 herd of models.\nNostalgebraist. 2020. Interpreting gpt: the logit lens. White paper.\nOlsson, C., N. Elhage, N. Nanda, N. Joseph, N. DasSarma, T. Henighan, B. Mann, A. Askell, Y. Bai, A. Chen, et al. 2022. In-context learning and induction heads. ArXiv preprint.\nSukhbaatar, S., A. Szlam, J. Weston, and R. Fergus. 2015. End-to-end memory networks. NeurIPS.\nUszkoreit, J. 2017. Transformer: A novel neural network architecture for language understanding. Google Research blog post, Thursday August 31, 2017.\nVaswani, A., N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin. 2017. Attention is all you need. NeurIPS.\nWeston, J., S. Chopra, and A. Bordes. 2015. Memory networks. ICLR 2015.", + "path": "8.pdf/CHAPTER 8 Transformers/Historical Notes", + "metadata": { + "length": 3827, + "summary": "The passage discusses the development of the Transformer model, drawing on self-attention and memory networks. Encoder-decoder attention was developed by Graves (2013) and Bahdanau et al. (2015), later extended to self-attention. Key, query, and value terminology originated from memory networks. The text includes a reference list of related works.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 8 Transformers", + "tokens": [ + "The", + "transformer", + "Vaswani", + "al", + "2017", + "developed", + "drawing", + "lines", + "prior", + "research", + "attention", + "memory", + "networks", + "Encoder", + "decoder", + "idea", + "soft", + "weighting", + "encodings", + "input", + "words", + "inform", + "generative", + "Chapter", + "12", + "Graves", + "2013", + "context", + "handwriting", + "generation", + "Bahdanau", + "2015", + "MT", + "This", + "extended", + "dropping", + "separate", + "encoding", + "decoding", + "sequences", + "tokens", + "collecting", + "information", + "passed", + "lower", + "layers", + "higher", + "Ling", + "Cheng", + "2016", + "Liu", + "Other", + "aspects", + "including", + "terminology", + "key", + "query", + "mechanism", + "adding", + "external", + "read", + "write", + "embedding", + "match", + "keys", + "representing", + "content", + "associative", + "Sukhbaatar", + "Weston", + "2014", + "MORE", + "HISTORY", + "TBD", + "IN", + "NEXT", + "DRAFT", + "Ba", + "Kiros", + "Hinton", + "Layer", + "normalization", + "NeurIPS", + "workshop", + "Cho", + "Bengio", + "Neural", + "machine", + "translation", + "jointly", + "learning", + "align", + "translate", + "ICLR", + "Brown", + "Mann", + "Ryder", + "Subbiah", + "Kaplan", + "Dhariwal", + "Neelakantan", + "Shyam", + "Sastry", + "Askell", + "Agarwal", + "Herbert", + "Voss", + "Krueger", + "Henighan", + "Child", + "Ramesh", + "Ziegler", + "Wu", + "Winter", + "Hesse", + "Chen", + "Sigler", + "Litwin", + "Gray", + "Chess", + "Clark", + "Berner", + "McCandlish", + "Radford", + "Sutskever", + "Amodei", + "2020", + "Language", + "models", + "shot", + "learners", + "volume", + "33", + "Dong", + "Lapata", + "Long", + "short", + "term", + "reading", + "EMNLP", + "Crosbie", + "Shutova", + "2022", + "Induction", + "heads", + "essential", + "pattern", + "matching", + "ArXiv", + "preprint", + "Elhage", + "Nanda", + "Olsson", + "Joseph", + "Bai", + "Conerly", + "Das", + "Sarma", + "Drain", + "Ganguli", + "Hatfield", + "Dodds", + "Hernandez", + "Jones", + "Kernion", + "Lovitt", + "Ndousse", + "Olah", + "2021", + "mathematical", + "framework", + "circuits", + "White", + "paper", + "Generating", + "recurrent", + "neural", + "Wayne", + "Danihelka", + "Turing", + "machines", + "Holtzman", + "Buys", + "Du", + "Forbes", + "Choi", + "curious", + "case", + "text", + "degeneration", + "Scaling", + "laws", + "language", + "Dyer", + "Black", + "Trancoso", + "Fernandez", + "Amir", + "Marujo", + "Lu", + "Finding", + "function", + "form", + "Compositional", + "character", + "open", + "vocabulary", + "word", + "representation", + "Sun", + "Lin", + "Wang", + "Learning", + "natural", + "inference", + "bidirectional", + "LSTM", + "model", + "Llama", + "Team", + "2024", + "llama", + "herd", + "Nostalgebraist", + "Interpreting", + "gpt", + "logit", + "lens", + "DasSarma", + "In", + "induction", + "Szlam", + "Fergus", + "End", + "end", + "Uszkoreit", + "Transformer", + "network", + "architecture", + "understanding", + "Google", + "Research", + "blog", + "post", + "Thursday", + "August", + "31", + "Shazeer", + "Parmar", + "Gomez", + "Kaiser", + "Polosukhin", + "Attention", + "Chopra", + "Bordes", + "Memory" + ], + "keywords": [ + "Transformer", + "Self-Attention", + "Memory Networks" + ], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/doc_nav.json b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/doc_nav.json new file mode 100644 index 000000000..ab15e77c4 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/doc_nav.json @@ -0,0 +1,300 @@ +{ + "version": "1.0", + "file_name": "Speech and Language Processing, Chapter 8: Transformers.pdf", + "stats": { + "total_chunks": 46, + "text_chunks": 23, + "image_chunks": 19, + "table_chunks": 4, + "max_depth": 4 + }, + "sections": [ + { + "title": "CHAPTER 8 Transformers", + "path": "8.pdf/CHAPTER 8 Transformers", + "level": 1, + "summary": "This chapter introduces the transformer architecture, the standard for building large language models. It focuses on left-to-right (causal) autoregressive language modeling, where input tokens are processed to predict output tokens sequentially. The core components include token embedding, positional encoding, multi-head self-attention, feedforward networks, layer normalization, and a language modeling head (unembedding matrix and softmax). The residual stream allows information to flow through N transformer blocks, with attention enabling contextual token representations by integrating information from surrounding tokens. The chapter outlines subsequent chapters covering masked language modeling (BERT), instruction tuning, machine translation, and speech recognition applications.", + "chunk_count": 23, + "children": [ + { + "title": "8.1 Attention", + "path": "8.pdf/CHAPTER 8 Transformers/8.1 Attention", + "level": 2, + "summary": "The passage discusses how static embeddings like word2vec represent words with fixed vectors regardless of context, using 'it' as an example. In contrast, transformer models use context to resolve ambiguity, as shown in sentences where 'it' refers to 'chicken' or 'road' depending on context. It also describes the transformer architecture: input tokens are encoded, passed through stacked transformer blocks, and then through a language modeling head (unembedding matrix and softmax) to predict the next token. A flowchart illustrates this process, showing how information from preceding tokens is incorporated.", + "chunk_count": 1, + "children": [] + }, + { + "title": "(8.3) The chicken didn't cross the road because it", + "path": "8.pdf/CHAPTER 8 Transformers/(8.3) The chicken didn't cross the road because it", + "level": 2, + "summary": "The passage explains how transformers use attention mechanisms to build contextual representations of word meaning. It describes how attention computes a weighted sum of context vectors, with weights determined by similarity scores between query and key vectors. The text covers both simplified and formal versions of attention, including single-head and multi-head attention, and explains the roles of query, key, and value matrices. It also discusses causal (left-to-right) attention used in language models.", + "chunk_count": 1, + "children": [] + }, + { + "title": "8.2 Transformer Blocks", + "path": "8.pdf/CHAPTER 8 Transformers/8.2 Transformer Blocks", + "level": 2, + "summary": "The text describes the architecture of a transformer block, focusing on the residual stream concept. It explains that a transformer block includes self-attention, feedforward layers, residual connections, and layer normalization. The residual stream is a pathway where token embeddings are passed through and progressively modified by attention and feedforward layers, with layer normalization applied before each. The feedforward layer is a two-layer network with ReLU activation. Layer normalization normalizes a vector by subtracting its mean and dividing by its standard deviation, then applies learnable gain and offset parameters. The block's computations are broken down into steps: layer norm, multi-head attention, residual addition, another layer norm, feedforward, and final residual addition. Only the attention component mixes information from other tokens, making it the token-mixing part. The input and output dimensions are matched, allowing stacking of multiple blocks (e.g., 12 to 96 layers) in large language models.", + "chunk_count": 1, + "children": [] + }, + { + "title": "8.3 Parallelizing computation using a single matrix X", + "path": "8.pdf/CHAPTER 8 Transformers/8.3 Parallelizing computation using a single matrix X", + "level": 2, + "summary": "This passage explains how multi-head attention and transformer blocks are parallelized for efficient computation. Input embeddings for N tokens are packed into a matrix X of shape [N x d]. For a single attention head, X is multiplied by query, key, and value weight matrices to produce Q, K, V matrices. All query-key comparisons are computed simultaneously via QK^T matrix multiplication, yielding an N x N matrix. A mask function sets upper-triangular elements to -infinity to prevent attending to future tokens. The scaled, masked, softmaxed result is multiplied by V to produce attention outputs. For multi-head attention, each head has its own Q, K, V projections; outputs are concatenated and linearly projected. The entire transformer layer computation is expressed in matrix form, with residual connections and layer normalization applied in parallel across all tokens. Attention is quadratic in input length, making long contexts computationally expensive.", + "chunk_count": 1, + "children": [] + }, + { + "title": "8.4 The input: embeddings for token and position", + "path": "8.pdf/CHAPTER 8 Transformers/8.4 The input: embeddings for token and position", + "level": 2, + "summary": "The passage explains how input embeddings are formed in transformers. Token embeddings are vectors from an embedding matrix E (size |V|×d) selected by token indices or one-hot vectors. Positional embeddings, often learned or sinusoidal, are added to token embeddings to encode position. The final input matrix X (N×d) combines both, with each row representing a token's composite embedding.", + "chunk_count": 1, + "children": [] + }, + { + "title": "8.5 The Language Modeling Head", + "path": "8.pdf/CHAPTER 8 Transformers/8.5 The Language Modeling Head", + "level": 2, + "summary": "The language modeling head is the final component of a transformer, used to predict the next word. It takes the output embedding from the last token at the final transformer layer and projects it via a linear layer (often using the transpose of the embedding matrix, called unembedding) to logits over the vocabulary. A softmax layer converts these logits into probabilities, enabling text generation through sampling. This architecture is also known as a decoder-only model.", + "chunk_count": 1, + "children": [] + }, + { + "title": "8.6 More on Sampling", + "path": "8.pdf/CHAPTER 8 Transformers/8.6 More on Sampling", + "level": 2, + "summary": "The passage discusses trade-offs in text generation sampling methods between quality (accuracy, coherence, factuality) and diversity (creativity, novelty). It then describes a transformer language model architecture (decoder-only) shown in Figure 8.16, where stacked transformer layers map an input token to a predicted next token, with input to each layer being the output from the previous layer.", + "chunk_count": 3, + "children": [ + { + "title": "8.6.1 Top-k sampling", + "path": "8.pdf/CHAPTER 8 Transformers/8.6 More on Sampling/8.6.1 Top-k sampling", + "level": 3, + "summary": "top-k sampling Top-k sampling is a simple generalization of greedy decoding. Instead of choosing the single most probable word to generate, we first truncate the distribution to the top k most likely words, renormalize to produce a legitimate probability distribution, and then randomly sample from within these k words according to their renormalized probabilities. More formally: 1. Choose in advance a number of words $k$ 2. For each word in the vocabulary $V$ , use the language model to compute the likelihood of this word given the context $p(w_{t}|\\mathbf{w}_{< t})$ 3. Sort the words by their likelihood, and throw away any word that is not one of the top k most probable words. 4. Renormalize the scores of the k words to be a legitimate probability distribu- tion. 5. Randomly sample a word from within these remaining $k$ most-probable words according to its probability. When k = 1, top-k sampling is identical to greedy decoding. Setting k to a larger number than 1 leads us to sometimes select a word which is not necessarily the most probable, but is still probable enough, and whose choice results in generating more diverse but still high-enough-quality text.", + "chunk_count": 1, + "children": [] + }, + { + "title": "8.6.2 Nucleus or top-p sampling", + "path": "8.pdf/CHAPTER 8 Transformers/8.6 More on Sampling/8.6.2 Nucleus or top-p sampling", + "level": 3, + "summary": "One problem with top-k sampling is that k is fixed, but the shape of the probability distribution over words differs in different contexts. If we set k = 10, sometimes the top 10 words will be very likely and include most of the probability mass, but other times the probability distribution will be flatter and the top 10 words will only include a small part of the probability mass. top-p sampling An alternative, called top-p sampling or nucleus sampling (Holtzman et al., 2020), is to keep not the top k words, but the top p percent of the probability mass. The goal is the same; to truncate the distribution to remove the very unlikely words. But by measuring probability rather than the number of words, the hope is that the measure will be more robust in very different contexts, dynamically increasing and decreasing the pool of word candidates. Given a distribution $P(w_{t}|\\mathbf{w}_{0 [images/image-17-Attention Mechanism Matrix Multiplication.jpg]
text_image Q X k1 k2 k3 k4 = QK^T q4", + "chunk_count": 4, + "children": [ + { + "title": "1 x dk", + "path": "8.pdf/CHAPTER 8 Transformers/8.8 Dealing with Scale/8.8.2 KV Cache/1 x dk", + "level": 4, + "summary": "dk x N", + "chunk_count": 1, + "children": [] + }, + { + "title": "1 x N", + "path": "8.pdf/CHAPTER 8 Transformers/8.8 Dealing with Scale/8.8.2 KV Cache/1 x N", + "level": 4, + "summary": "V x v1 v2 v3 v4 N x dv = A a4", + "chunk_count": 1, + "children": [] + }, + { + "title": "1 x dv", + "path": "8.pdf/CHAPTER 8 Transformers/8.8 Dealing with Scale/8.8.2 KV Cache/1 x dv", + "level": 4, + "summary": "
Figure 8.18 Parts of the attention computation (extracted from Fig. 8.11) showing, in black, the vectors that can be stored in the cache rather than recomputed when computing the attention score for the 4th token.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "8.8.3 Parameter Efficient Fine Tuning", + "path": "8.pdf/CHAPTER 8 Transformers/8.8 Dealing with Scale/8.8.3 Parameter Efficient Fine Tuning", + "level": 3, + "summary": "LoRA (Low-Rank Adaptation) is a parameter-efficient fine-tuning method for large language models. Instead of updating all parameters during fine-tuning, LoRA freezes the pretrained weight matrices (like W^Q, W^K, W^V, W^O in attention layers) and trains low-rank decomposition matrices A and B. This reduces hardware requirements, speeds up training, and allows easy swapping of fine-tuned modules without adding inference time.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "8.9 Interpreting the Transformer", + "path": "8.pdf/CHAPTER 8 Transformers/8.9 Interpreting the Transformer", + "level": 2, + "summary": "interpretability How does a transformer-based language model manage to do so well at language tasks? The subfield of interpretability, sometimes called mechanistic interpretability, focuses on ways to understand mechanistically what is going on inside the transformer. In the next two subsections we discuss two well-studied aspects of transformer interpretability.", + "chunk_count": 3, + "children": [ + { + "title": "8.9.1 In-Context Learning and Induction Heads", + "path": "8.pdf/CHAPTER 8 Transformers/8.9 Interpreting the Transformer/8.9.1 In-Context Learning and Induction Heads", + "level": 3, + "summary": "The passage discusses in-context learning, a concept introduced by Brown et al. (2020) with GPT-3, where language models learn from prompts during inference without gradient updates. It contrasts this with pretraining, which updates parameters via gradient descent. A key hypothesis for how in-context learning works involves induction heads, circuits in transformer attention that predict repeated sequences by matching prefixes and copying subsequent tokens. Olsson et al. (2022) propose a generalized fuzzy pattern completion rule, and Crosbie and Shutova (2022) provide evidence by showing that ablating induction heads reduces in-context learning performance.", + "chunk_count": 1, + "children": [] + }, + { + "title": "8.9.2 Logit Lens", + "path": "8.pdf/CHAPTER 8 Transformers/8.9 Interpreting the Transformer/8.9.2 Logit Lens", + "level": 3, + "summary": "logit lens Another useful interpretability tool, the logit lens (Nostalgebraist, 2020), offers a way to visualize what the internal layers of the transformer might be representing. The idea is that we take any vector from any layer of the transformer and, pretending that it is the prefinal embedding, simply multiply it by the unembedding layer to get logits, and compute a softmax to see the distribution over words that that vector might be representing. This can be a useful window into the internal representations of the model. Since the network wasn't trained to make the internal representations function in this way, the logit lens doesn't always work perfectly, but this can still be a useful trick to help us visualize the internal layers of a transformer.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "8.10 Summary", + "path": "8.pdf/CHAPTER 8 Transformers/8.10 Summary", + "level": 2, + "summary": "This chapter introduces the transformer and its components for language modeling. Key points include: transformers are non-recurrent networks using multi-head self-attention; a transformer block consists of a residual stream with multi-head attention and feedforward layers, each preceded by layer normalization; input is computed by adding token embeddings to positional encodings; language models stack transformer blocks with a language model head that generates logits via an unembedding matrix and softmax; these models have wide context windows (up to 200K tokens or more); efficiency tricks include KV cache and parameter-efficient finetuning.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Historical Notes", + "path": "8.pdf/CHAPTER 8 Transformers/Historical Notes", + "level": 2, + "summary": "The passage discusses the development of the Transformer model, drawing on self-attention and memory networks. Encoder-decoder attention was developed by Graves (2013) and Bahdanau et al. (2015), later extended to self-attention. Key, query, and value terminology originated from memory networks. The text includes a reference list of related works.", + "chunk_count": 1, + "children": [] + } + ] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-Transformer Encoder Architecture Diagram.jpg", + "summary": "image-1 - Input: input token -> Embedding (pink trapezoid) - Positional encoding added to embeddings - Residual stream processing loop (repeated N times): - Layer Norm -> MultiHead Attention -> residual connection (+) - Layer Norm -> Feedforward -> residual connection (+) - Output processing: Unembedding -> Softmax -> output probabilities (over tokens) - Key components: Embedding, Positional, MultiHead Attention, Layer Norm, Feedforward, Unembedding, Softmax" + }, + { + "path": "images/image-2-Transformer Architecture Diagram.jpg", + "summary": "image-2 - **Input Tokens**: Sequence \"So\", \"long\", \"and\", \"thanks\", \"for\" (x1-x5) - **Input Encoding**: Embedding layer (E) with positional encoding (1-5) - **Stacked Transformer Blocks**: Multiple layers of processing blocks with internal connections - **Language Modeling Head**: Output layer (U) producing logits for next token prediction - **Next Token Prediction**: Predicts \"long\", \"and\", \"thanks\", \"for\", \"all\" sequentially" + }, + { + "path": "images/image-3-Self-Attention Visualization.jpg", + "summary": "image-3 Diagram illustrating the self-attention mechanism between Layer k and Layer k+1. Key Components: - Input Sequence (Columns): \"The chicken didn't cross the road because it was too tired\" - Layers: Two distinct layers labeled \"Layer k\" (bottom) and \"Layer k+1\" (top). - Focus Token: The word \"it\" in Layer k+1 is highlighted as the target of attention. Relationships & Attention Distribution: - The diagram visualizes how the token \"it\" at Layer k+1 attends to specific tokens from Layer k. - Strongest Connection: A thick blue line connects \"it\" (Layer k+1) to \"chicken\" (Layer k), indicating the highest attention weight. This demonstrates that the model associates \"it\" with \"chicken\". - Secondary Connections: Thinner lines connect \"it\" (Layer k+1) to \"road\" and \"because\" in Layer k, showing lower but present attention weights. - Visual Representation: The thickness of the connecting lines corresponds to the strength of the self-attention distribution." + }, + { + "path": "images/image-4-Self-Attention Layer Diagram.jpg", + "summary": "image-4 - Component: Self-Attention Layer - Input nodes: x1, x2, x3, x4, x5 - Output nodes: a1, a2, a3, a4, a5 - Internal structure: Contains five parallel \"attention\" blocks. - Connectivity: Each input node (x) connects to every attention block, indicating full connectivity or global context processing within the layer." + }, + { + "path": "images/image-5-Self-Attention Mechanism Process.jpg", + "summary": "image-5 1. Generate key, query, value vectors from inputs X1, X2, and X3 (dimension [1 x d]) using weight matrices W^K, W^Q, and W^V. 2. Compare the query vector of x3 with the key vectors for x1, x2, and x3. 3. Divide scalar scores by the square root of dimension K (√d_k). 4. Turn scores into attention weights (α_i,j) via softmax. 5. Weigh each value vector based on the calculated attention weights (e.g., α_3,1, α_3,2, α_3,3). 6. Sum the weighted value vectors to produce a combined vector of dimension [1 x d_v]. 7. Reshape the result to [1 x d]. 8. Output of self-attention (a_3), obtained by multiplying with weight matrix W^O (dimension [d_v x d])." + }, + { + "path": "images/image-6-Multi-Head Attention Mechanism.jpg", + "summary": "image-6 The diagram illustrates the architecture of a multi-head attention mechanism, likely within a Transformer model. Key components and flow: - Input: A sequence of vectors (represented as $x_{i-3}, x_{i-2}, x_{i-1}, X_i$) with dimension $[1 \\times d]$. - Parallel Heads: The input is processed by four parallel \"heads\" (Head 1 through Head 4). Each head applies its own set of weight matrices ($W^K, W^V, W^Q$) to attend differently to the context. - Output Dimensions: Each head produces an output vector of dimension $[1 \\times d_v]$. - Concatenation: The outputs from all heads are concatenated to form a combined vector of dimension $[1 \\times Ad_v]$, where $A=4$ in this example. - Final Projection: The concatenated output is projected to the final representation using a weight matrix $W^O$. This results in a final output vector of dimension $[1 \\times d]$. - Note on Dimensions: The diagram specifies that usually" + }, + { + "path": "images/image-7-Transformer Block Architecture.jpg", + "summary": "image-7 - Input sequence: x_i (with preceding steps x_{i-1}, x_{i-2}) - Output sequence: h_i (with preceding steps h_{i-1}, h_{i-2}) - Core components: - MultiHead Attention layer - Layer Norm layers (pre and post attention) - Feedforward network - Structural features: - Residual connections (indicated by green diamond \"+\" symbols) - Skip connections bypassing attention and feedforward blocks - Normalization applied before and after attention mechanisms" + }, + { + "path": "images/image-8-Diagram of Token Residual Streams.jpg", + "summary": "image-8 - Illustrates the flow of residual streams for two tokens (Token A and Token B) across vertical lanes or channels. - Red arrows indicate the path of the residual stream: - Token A's stream starts in a lower lane and moves upward into an adjacent lane. - Token B's stream follows a similar trajectory, moving from one lane to another in an upward direction. - The diagram suggests a mechanism where token information is passed or shifted between different processing paths or layers." + }, + { + "path": "images/image-9-N x N Grid Structure.jpg", + "summary": "image-9 q1·k1, q1·k2, q1·k3, q1·k4 q2·k1, q2·k2, q2·k3, q2·k4 q3·k1, q3·k2, q3·k3, q3·k4 q4·k1, q4·k2, q4·k3, q4·k4" + }, + { + "path": "images/image-10-Attention Mechanism Process.jpg", + "summary": "image-10 Top Row: Linear Transformations - Input (X): N x d matrix containing Input Tokens 1 through 4. - Query Generation (Q): X multiplied by weight matrix W^Q (d x d_k) results in Q (N x d_k). - Key Generation (K): X multiplied by weight matrix W^K (d x d_k) results in K (N x d_k). - Value Generation (V): X multiplied by weight matrix W^V (d x d_v) results in V (N x d_v). Bottom Row: Attention Calculation - Matrix Multiplication: Q (N x d_k) is multiplied by the transpose of K (K^T, d_k x N) to produce the attention scores matrix QK^T (N x N). - Masking: The QK^T matrix is masked (indicated by \"mask\" and infinity symbols -∞) to prevent future tokens from attending to current ones. - Final Output (A): The masked matrix is multiplied by the Value matrix V (N x d_v) to produce the final output A (N x d" + }, + { + "path": "images/image-11-$$ _detail.jpg", + "summary": "image-11 $$
0" + }, + { + "path": "images/image-12-__details_.jpg", + "summary": "image-12
0" + }, + { + "path": "images/image-13-Transformer Block Architecture.jpg", + "summary": "image-13 - Core Component: Transformer Block (top level) - Input Definition: X = Composite Embeddings (word + position) - Process Flow: - Word Embeddings and Position Embeddings are combined via addition (+) to form the composite input. - Sequence Elements: - Token 1: \"Janet\" (Word) + \"1\" (Position) - Token 2: \"will\" (Word) + \"2\" (Position) - Token 3: \"back\" (Word) + \"3\" (Position) - Token 4: \"the\" (Word) + \"4\" (Position) - Token 5: \"bill\" (Word) + \"5\" (Position)" + }, + { + "path": "images/image-14-Language Model Head Architecture.jpg", + "summary": "image-14 - **Input**: Sequence of words (w1, w2, ..., wN) processed by a Layer L Transformer Block. - **Intermediate Output**: Hidden states from the final layer ($h^L_1$, $h^L_2$, ..., $h^L_N$). The diagram focuses on the last hidden state $h^L_N$ with dimensions [1 x d]. - **Processing Steps**: - **Unembedding Layer**: Transforms the hidden state using matrix multiplication where $U = E^T$. This produces Logits with dimensions [1 x |V|]. - **Softmax**: Applies softmax over the vocabulary V to convert logits into probabilities. - **Output**: Word probabilities vector [1 x |V|], representing the distribution over the vocabulary V for the next token prediction." + }, + { + "path": "images/image-15-Transformer Architecture Diagram.jpg", + "summary": "image-15 - Input Encoding: Input token (Wi) passes through Embedding (E) and positional encoding to produce x1i. - Layer Structure (Layers 1 to L): Each layer contains a sequence of operations: - Layer Norm - Attention mechanism - Layer Norm - Feedforward network - Residual Connections: Indicated by arrows bypassing blocks (e.g., h^1_i = x^2_i). - Language Modeling Head: Located at the top, consisting of: - Extra Layer Norm - Linear projection matrix U - Softmax function producing logits (u1...u|V|) - Output Generation: Produces token probabilities (y1...y|V|) used to sample the next token (w_{i+1})." + }, + { + "path": "images/image-16-Transformer Language Model Architecture.jpg", + "summary": "image-16 - Input tokens: So, long, and, thanks, for... - Input Encoding: Embedding (E) + Positional Encoding (1, 2, 3...) -> x1, x2, x3... - Stacked Transformer Blocks: Multiple layers of processing with skip connections between blocks - Language Modeling Head: U projection layer producing logits - Loss Calculation: Negative log likelihood (-log y) for each predicted token (long, and, thanks, for, all) - Overall Loss Formula: L_CE = (1/T) * Σ(from t=1 to T) L_CE" + }, + { + "path": "images/image-17-Attention Mechanism Matrix Multiplication.jpg", + "summary": "image-17 The image illustrates the mathematical process of calculating attention scores and weighted values using matrix multiplication. 1. **Query Vector (Q):** A single query vector `q4` with dimensions $1 \\times d_k$. 2. **Key Matrix Transpose ($K^T$):** A matrix containing keys $k_1, k_2, k_3, k_4$ with dimensions $d_k \\times N$. 3. **Dot Product ($QK^T$):** The result of multiplying Q by $K^T$, producing a row vector of dot products ($q4\\cdot k1, q4\\cdot k2, q4\\cdot k3, q4\\cdot k4$) with dimensions $1 \\times N$. 4. **Value Matrix (V):** A matrix containing values $v_1, v_2, v_3, v_4$ with dimensions $N \\times d_v$. 5. **Final Output (A):** The result of multiplying the dot product matrix by V, yielding a final output vector `a4` with dimensions $1" + }, + { + "path": "images/image-18-Pretrained Weights Decomposition.jpg", + "summary": "image-18 The image illustrates a mathematical decomposition of a pretrained weight matrix W. - **Central Component**: A green square labeled \"Pretrained Weights\" with the symbol **W**. Its dimensions are defined as **k** (height) and **d** (width). - **Decomposition Structure**: The diagram shows that W is approximated by the sum of two components: 1. **Top Component**: A yellow horizontal rectangle labeled **h**, with dimensions **1** x **d**. 2. **Right Component**: The product of two red matrices, **A** and **B**. - Matrix **A** is a vertical rectangle with dimensions **k** x **r**. - Matrix **B** is a horizontal rectangle with dimensions **r** x **d**. - **Input Vector**: At the bottom, a blue horizontal vector labeled **x** is shown with dimensions **1** x **d**. - **Arrows**: Arrows indicate that the input **x** interacts with the decomposed components to reconstruct or approximate the original weights **W**." + }, + { + "path": "images/image-19-Prefix matching and copying.jpg", + "summary": "image-19 The diagram illustrates a sequence alignment process involving two sentences: \"She owns vintage cars.\" and \"He dreams of owning vintage cars\". - Prefix matching: The word \"cars\" in the first sentence is matched with the word \"cars\" in the second sentence. - Copying: The word \"cars\" from the first sentence is copied to the corresponding position in the second sentence." + } + ], + "tables": [ + { + "path": "tables/table-0 Self-Attention.html", + "summary": "table-1 The table shows tokens from a sentence about a chicken not crossing a road, with self-attention distribution layers listing each token." + }, + { + "path": "tables/table-1 Causal Attention Mask.html", + "summary": "table-2 The table shows a causal attention mask where each query attends only to current and previous keys, with future positions masked as -∞." + }, + { + "path": "tables/table-3 .html", + "summary": "table-3" + }, + { + "path": "tables/table-4 .html", + "summary": "table-4" + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/full.md b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/full.md new file mode 100644 index 000000000..eb2d05395 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/full.md @@ -0,0 +1,1364 @@ +## CHAPTER + +## 8 + +## Transformers + +"The true art of memory is the art of attention" + +Samuel Johnson, Idler #74, September 1759 + +In this chapter we introduce the transformer, the standard architecture for building large language models. As we discussed in the prior chapter, transformer-based large language models have completely changed the field of speech and language processing. Indeed, every subsequent chapter in this textbook will make use of them. As with the previous chapter, we'll focus for this chapter on the use of transformers to model left-to-right (sometimes called causal or autoregressive) language modeling, in which we are given a sequence of input tokens and predict output tokens one by one by conditioning on the prior context. + +![](images/3079c00a53ef3a9108b19a0c19e3065112eb09cf88a92b00d3fd800feb259912.jpg) + +
+flowchart + +```mermaid +graph TD + A["input token"] --> B["Embedding"] + B --> C["Positional"] + C --> D["MultiHead Attention"] + D --> E["Layer Norm"] + E --> F["Feedforward"] + F --> G["Unembedding"] + G --> H["Softmax"] + H --> I["output probabilities (over tokens)"] + style A fill:#f9f,stroke:#333 + style B fill:#ccf,stroke:#333 + style C fill:#cfc,stroke:#333 + style D fill:#fcc,stroke:#333 + style E fill:#cff,stroke:#333 + style F fill:#ffc,stroke:#333 + style G fill:#cfc,stroke:#333 + style H fill:#fcc,stroke:#333 + style I fill:#ffc,stroke:#333 + subgraph N times + J["residual stream"] + K["Input"] + end +``` +
+ +Figure 8.1 A transformer decoder for language modeling, showing the residual stream for processing an input token. A single token is embedded and passed forward in the network, with the feedforward and attention components adding information. The multihead attention layer takes inputs (not shown in detail) from the neighboring token streams. This is thus one column of an autoregressive transformer language model, taking an input token and outputting a distribution over next tokens. + +Fig. 8.1 sketches the transformer architecture following a single token as it is passes up through the layers of the network. Each token is first converted to an embedding from the embedding matrix E. Recall from Chapter 6 in Section ?? that E is a linear layer that maps a token id to a vector embedding representing that token. Each token in the vocabulary has an initial embedding representation in E. + +Transformers also have a special mechanism for encoding the position/index of the token in the input string, which is simply added to the embedding. The resulting embedding represents both the word and its position. and is then passed through a set of N transformer blocks. + +It's common to think of each of these transformer blocks as part of a stream in which the input embedding is directly passed up to the output, while simultaneously being enriched by the application of various processing modules: the multi-head attention layer, feedforward networks and the layer normalization. The value of the stream at any layer is the sum of the original embedding and all the outputs from all the previous layers and blocks. + +The core intuition of the transformer, and the component that distinguishes it from the feedforward layers we saw in Chapter 6, is this multi-head attention layer, also called a self-attention layer. Attention can be thought of as a way to build contextual representations of a token's meaning by attending to and integrating information from surrounding tokens, helping the model learn how tokens relate to each other over large spans. It can also be thought of as a way to move information from one residual stream to another, augmenting the stream at one token position with information from another token position. + +After the $N$ transformer blocks we take the output embedding that is produced by the final transformer block, pass it through an linear unembedding matrix $\mathbf{U}$ and then a softmax over the vocabulary to generate a distribution over possible next tokens. These last two components (the unembedding matrix and the softmax) are sometimes called the language modeling head. In the rest of this chatper we'll introduce attention and the rest of these modules in more detail. + +Fig. ?? shows the transformer architecture applied to a context window with the words So long and thanks for, showing at each token position what is the most likely token to be generated. In this full figure, the set of N blocks maps an entire context window of input vectors $(\mathbf{x}_{1},..., \mathbf{x}_{n})$ to a window of output vectors $(\mathbf{h}_{1},..., \mathbf{h}_{n})$ of the same length. A column might contain from 12 to 96 or more stacked blocks. The arrows in the figure show how information from the hidden representations of preceding tokens is incorporated into the transformer block. + +Transformer-based language models are complex, and so the details will unfold over this chapter and the next few chapters. Chapter 7 already discussed how language models are pretrained, and how tokens are generated via sampling. In the rest of this chapter we'll introduce multi-head attention, the rest of the transformer block, and the input encoding and language modeling head components of the transformer. Chapter 9 introduces masked language modeling and the BERT family of bidirectional transformer encoder models. Chapter 10 shows how to instruction-tune language models to perform NLP tasks, and how to align the model with human preferences. Chapter 12 will introduce machine translation with the encoder-decoder architecture. And we'll see application of the transformer to speech recognition, as well as further use of the encoder-decoder architecture, in Chapter 15. + +## 8.1 Attention + +Recall from Chapter 5 that for word2vec and other static embeddings, the representation of a word's meaning is always the same vector irrespective of the context: the word chicken, for example, is always represented by the same fixed vector. So a static vector for the word it might somehow encode that this is a pronoun used for animals and inanimate entities. But in context it has a much richer meaning. Consider it in one of these two sentences: + +![](images/c45c88589306f9bcac236eed8b566d8822fd384d45e73330763f503ed0122ab0.jpg) + +
+flowchart + +```mermaid +graph TD + subgraph Input Tokens + A1["Input encoding So"] --> X1["x1"] + A2["Input encoding x2"] --> X2["x2"] + A3["Input encoding x3"] --> X3["x3"] + A4["Input encoding x4"] --> X4["x4"] + A5["Input encoding x5"] --> X5["x5"] + end + + subgraph Stacked Transformer Blocks + B1["Stacked Transformer Blocks x1"] --> X1 + B2["Stacked Transformer Blocks x2"] --> X2 + B3["Stacked Transformer Blocks x3"] --> X3 + B4["Stacked Transformer Blocks x4"] --> X4 + B5["Stacked Transformer Blocks x5"] --> X5 + end + + subgraph Next Token + C1["Language Modeling Head long"] --> U1["logits U"] + C2["Language Modeling Head and"] --> U2["logits U"] + C3["Language Modeling Head and thanks"] --> U3["logits U"] + C4["Language Modeling Head and thanks for all"] --> U4["logits U"] + end + + U1 --> X1 + U2 --> X2 + U3 --> X3 + U4 --> X4 + U5 --> X5 + + X1 --> Y1["E"] + X2 --> Y2["E"] + X3 --> Y3["E"] + X4 --> Y4["E"] + X5 --> Y5["E"] + + style Input Tokens fill:#f9f,stroke:#333 + style Next Token fill:#ccf,stroke:#333 +``` +
+ +Figure 8.2 The architecture of a (left-to-right) transformer, showing how each input token get encoded, passed through a set of stacked transformer blocks, and then a language model head that predicts the next token. The embeddings at each token position in the residual stream are passed up the stack, and the arrows in the figure shows how information from the hidden representations of preceding tokens are also incorporated. + +(8.1) The chicken didn't cross the road because it was too tired. + +(8.2) The chicken didn't cross the road because it was too wide. + +In (8.1) it is the chicken (i.e., the reader knows that the chicken was tired), while in (8.2) it is the road (and the reader knows that the road was wide). $^{1}$ That is, if we are to compute the meaning of this sentence, we'll need the meaning of it to be associated with the chicken in the first sentence and associated with the road in the second one, sensitive to the context. + +Furthermore, consider reading left to right like a causal language model, processing the sentence up to the word it: + +(8.3) The chicken didn't cross the road because it + +At this point we don't yet know which thing it is going to end up referring to! So a representation of it at this point might have aspects of both chicken and road as the reader is trying to guess what happens next. + +This fact that words have rich linguistic relationships with other words that may be far away pervades language. Consider two more examples: + +(8.4) The keys to the cabinet are on the table. + +(8.5) I walked along the pond, and noticed one of the trees along the bank. + +In (8.4), the phrase The keys is the subject of the sentence, and in English and many languages, must agree in grammatical number with the verb are; in this case both are plural. In English we can't use a singular verb like is with a plural subject like keys (we'll discuss agreement more in Chapter 18). In (8.5), we know that bank refers to the side of a pond or river and not a financial institution because of the context, including words like pond. (We'll discuss word senses more in Chapter 9.) + +contextual embeddings + +The point of all these examples is that these contextual words that help us compute the meaning of words in context can be quite far away in the sentence or paragraph. Transformers can build contextual representations of word meaning, contextual embeddings, by integrating the meaning of these helpful contextual words. In a transformer, layer by layer, we build up richer and richer contextualized representations of the meanings of input tokens. At each layer, we compute the representation of a token i by combining information about i from the previous layer with information about the neighboring tokens to produce a contextualized representation for each word at each position. + +Attention is the mechanism in the transformer that weighs and combines the representations from appropriate other tokens in the context from layer k to build the representation for tokens in layer $k + 1$ . + +![](images/1fb4e57c34a18f66351c512509d3462694aa198b255b0c8a589d32c625236170.jpg) + +
+bar chart + +columns corresponding to input tokens +| Layer | Token | Token Description | +| :--- | :--- | :--- | +| Layer k+1 | The | The | +| Layer k+1 | chicken | chicken | +| Layer k+1 | didn't | didn't | +| Layer k+1 | cross | cross | +| Layer k+1 | the | the | +| Layer k+1 | road | road | +| Layer k+1 | because | because | +| Layer k+1 | it | it | +| self-attention distribution | the | the | +| self-attention distribution | chicken | chicken | +| self-attention distribution | didn't | didn't | +| self-attention distribution | cross | cross | +| self-attention distribution | the | the | +| self-attention distribution | road | road | +| self-attention distribution | because | because | +| self-attention distribution | it | it | +| self-attention distribution | was | was | +| self-attention distribution | too | too | +| self-attention distribution | tired | tired | +columns corresponding to input tokens +
+ +Figure 8.3 The self-attention weight distribution $\alpha$ that is part of the computation of the representation for the word $it$ at layer $k + 1$ . In computing the representation for $it$ , we attend differently to the various words at layer $k$ , with darker shades indicating higher self-attention values. Note that the transformer is attending highly to the columns corresponding to the tokens chicken and road, a sensible result, since at the point where $it$ occurs, it could plausibly corefer with the chicken or the road, and hence we'd like the representation for $it$ to draw on the representation for these earlier words. Figure adapted from Uszkoreit (2017). + +Fig. 8.3 shows a schematic example simplified from a transformer (Uszkoreit, 2017). The figure describes the situation when the current token is it and we need to compute a contextual representation for this token at layer $k+1$ of the transformer, drawing on the representations (from layer k) of every prior token. The figure uses color to represent the attention distribution over the contextual words: the tokens chicken and road both have a high attention weight, meaning that as we are computing the representation for it, we will draw most heavily on the representation for chicken and road. This will be useful in building the final representation for it, since it will end up coreferring with either chicken or road. + +Let's now turn to how this attention distribution is represented and computed. + +## 8.1.1 Attention more formally + +As we've said, the attention computation is a way to compute a vector representation for a token at a particular layer of a transformer, by selectively attending to and integrating information from prior tokens at the previous layer. Attention takes an input representation $x_{i}$ corresponding to the input token at position i, and a context window of prior inputs $x_{1}..x_{i-1}$ , and produces an output $a_{i}$ . + +In causal, left-to-right language models, the context is any of the prior words. That is, when processing $x_{i}$ , the model has access to $x_{i}$ as well as the representations of all the prior tokens in the context window (context windows consist of thousands of tokens) but no tokens after i. (By contrast, in Chapter 9 we’ll generalize attention so it can also look ahead to future words.) + +Fig. 8.4 illustrates this flow of information in an entire causal self-attention layer, in which this same attention computation happens in parallel at each token position i. Thus a self-attention layer maps input sequences $(\mathbf{x}_{1},..., \mathbf{x}_{n})$ to output sequences of the same length $(\mathbf{a}_{1},..., \mathbf{a}_{n})$ . + +![](images/557411430fcc9a5afeea99f2dabf61c6c01efa8ad19483eb299e4b4970a58e2f.jpg) + +
+flowchart + +```mermaid +graph TD + a1[" a₁ "] --> A1["attention"] + a2[" a₂ "] --> A2["attention"] + a3[" a₃ "] --> A3["attention"] + a4[" a₄ "] --> A4["attention"] + a5[" a₅ "] --> A5["attention"] + x1[" x₁ "] --> A1 + x2[" x₂ "] --> A2 + x3[" x₃ "] --> A3 + x4[" x₄ "] --> A4 + x5[" x₅ "] --> A5 + A1 --> x1 + A2 --> x2 + A3 --> x3 + A4 --> x4 + A5 --> x5 +``` +
+ +Figure 8.4 Information flow in causal self-attention. When processing each input $x_{i}$ , the model attends to all the inputs up to, and including $x_{i}$ . + +Simplified version of attention At its heart, attention is really just a weighted sum of context vectors, with a lot of complications added to how the weights are computed and what gets summed. For pedagogical purposes let's first describe a simplified intuition of attention, in which the attention output $\mathbf{a}_i$ at token position $i$ is simply the weighted sum of all the representations $\mathbf{x}_j$ , for all $j \leq i$ ; we'll use $\alpha_{ij}$ to mean how much $\mathbf{x}_j$ should contribute to $\mathbf{a}_i$ : + +$$ +\text { Simplified version: } \quad \mathbf {a} _ {i} = \sum_ {j \leq i} \alpha_ {i j} \mathbf {x} _ {j} \tag {8.6} +$$ + +Each $\alpha_{ij}$ is a scalar used for weighing the value of input $\mathbf{x}_j$ when summing up the inputs to compute $\mathbf{a}_i$ . How shall we compute this $\alpha$ weighting? In attention we weight each prior embedding proportionally to how similar it is to the current token $i$ . So the output of attention is a sum of the embeddings of prior tokens weighted by their similarity with the current token embedding. We compute similarity scores via dot product, which maps two vectors into a scalar value ranging from $-\infty$ to $\infty$ . The larger the score, the more similar the vectors that are being compared. We'll normalize these scores with a softmax to create the vector of weights $\alpha_{ij}, j \leq i$ . + +$$ +\text { Simplified Version: } \quad \operatorname{score} (\mathbf {x} _ {i}, \mathbf {x} _ {j}) = \mathbf {x} _ {i} \cdot \mathbf {x} _ {j} \tag {8.7} +$$ + +$$ +\alpha_ {i j} = \text { softmax } (\text { score } (\mathbf {x} _ {i}, \mathbf {x} _ {j})) \forall j \leq i \tag {8.8} +$$ + +Thus in Fig. 8.4 we compute $a_{3}$ by computing three scores: $x_{3} \cdot x_{1}$ , $x_{3} \cdot x_{2}$ and $x_{3} \cdot x_{3}$ , normalizing them by a softmax, and using the resulting probabilities as weights indicating each of their proportional relevance to the current position i. Of course, the softmax weight will likely be highest for $x_{i}$ , since $x_{i}$ is very similar to itself, resulting in a high dot product. But other context words may also be similar to i, and the softmax will also assign some weight to those words. Then we use these weights as the $\alpha$ values in Eq. 8.6 to compute the weighted sum that is our $a_{3}$ . + +The simplified attention in equations 8.6 - 8.8 demonstrates the attention-based approach to computing $\mathbf{a}_i$ : compare the $\mathbf{x}_i$ to prior vectors, normalize those scores into a probability distribution used to weight the sum of the prior vectors. But now we're ready to remove the simplifications. + +attention head +head + +A single attention head using query, key, and value matrices Now that we've seen a simple intuition of attention, let's introduce the actual attention head, the version of attention that's used in transformers. (The word head is often used in transformers to refer to specific structured layers). The attention head allows us to distinctly represent three different roles that each input embedding plays during the course of the attention process: + +query +key +value + +- As the current element being compared to the preceding inputs. We'll refer to this role as a query. +- In its role as a preceding input that is being compared to the current element to determine a similarity weight. We'll refer to this role as a key. +- And finally, as a value of a preceding element that gets weighted and summed up to compute the output for the current element. + +To capture these three different roles, transformers introduce weight matrices $W^{Q}$ , $W^{K}$ , and $W^{V}$ . These weights will project each input vector $x_{i}$ into a representation of its role as a query, key, or value: + +$$ +\mathbf {q} _ {i} = \mathbf {x} _ {i} \mathbf {W} ^ {\mathrm{Q}}; \quad \mathbf {k} _ {i} = \mathbf {x} _ {i} \mathbf {W} ^ {\mathrm{K}}; \quad \mathbf {v} _ {i} = \mathbf {x} _ {i} \mathbf {W} ^ {\mathrm{V}} \tag {8.9} +$$ + +Given these projections, when we are computing the similarity of the current element $\mathbf{x}_i$ with some prior element $\mathbf{x}_j$ , we'll use the dot product between the current element's query vector $\mathbf{q}_i$ and the preceding element's key vector $\mathbf{k}_j$ . Furthermore, the result of a dot product can be an arbitrarily large (positive or negative) value, and exponentiating large values can lead to numerical issues and loss of gradients during training. To avoid this, we scale the dot product by a factor related to the size of the embeddings, via dividing by the square root of the dimensionality of the query and key vectors $(d_k)$ . We thus replace the simplified Eq. 8.7 with Eq. 8.11. The ensuing softmax calculation resulting in $\alpha_{ij}$ remains the same, but the output calculation for $\textbf{head}_i$ is now based on a weighted sum over the value vectors $\mathbf{v}$ (Eq. 8.13). + +Here's a final set of equations for computing self-attention for a single self-attention output vector $\mathbf{a}_i$ from a single input vector $\mathbf{x}_i$ . This version of attention computes $\mathbf{a}_i$ by summing the values of the prior elements, each weighted by the similarity of its key to the query from the current element: + +$$ +\mathbf {q} _ {i} = \mathbf {x} _ {i} \mathbf {W} ^ {\mathrm{Q}}; \quad \mathbf {k} _ {j} = \mathbf {x} _ {j} \mathbf {W} ^ {\mathrm{K}}; \quad \mathbf {v} _ {j} = \mathbf {x} _ {j} \mathbf {W} ^ {\mathrm{V}} \tag {8.10} +$$ + +$$ +\operatorname{score} (\mathbf {x} _ {i}, \mathbf {x} _ {j}) = \frac {\mathbf {q} _ {i} \cdot \mathbf {k} _ {j}}{\sqrt {d _ {k}}} \tag {8.11} +$$ + +$$ +\alpha_ {i j} = \text { softmax } (\text { score } (\mathbf {x} _ {i}, \mathbf {x} _ {j})) \forall j \leq i \tag {8.12} +$$ + +$$ +\mathbf {h e a d} _ {i} = \sum_ {j \leq i} \alpha_ {i j} \mathbf {v} _ {j} \tag {8.13} +$$ + +$$ +\mathbf {a} _ {i} = \text { head } _ {i} \mathbf {W} ^ {\mathrm{O}} \tag {8.14} +$$ + +We illustrate this in Fig. 8.5 for the case of calculating the value of the third output $a_{3}$ in a sequence. + +Note that we've also introduced one more matrix, $\mathbf{W}_{\mathbf{O}}$ , which is left-multiplied by the attention head. This is necessary to reshape the output of the head. The input to attention $\mathbf{x}_{\mathbf{i}}$ and the output from attention $\mathbf{a}_{\mathbf{i}}$ both have the same dimensionality $[1\times d]$ . We often call $d$ the model dimensionality, and indeed as we'll discuss in + +![](images/c969ffdb2babae14653950354072c94ba2da859ea4681e30bdaa3dbe016fd1a3.jpg) + +
+flowchart + +```mermaid +graph TD + A["1 x d"] --> B["×1 × dv"] + C["2 x3's query with the keys for x1, x2, and x3"] --> D["×1 × dv"] + E["3 x3"] --> F["×1 × dv"] + G["4 x3"] --> H["×1 × dv"] + I["5 x3"] --> J["×1 × dv"] + K["6 x3"] --> L["×1 × dv"] + M["7 x3"] --> N["×1 × dv"] + O["8 x3"] --> P["Output of self-attention a₃ [1 × d"]] + Q["9 x3"] --> R["W⁰ [dv × d"] [1 × dv]] + S["10 x3"] --> T["Sum the weighted value vectors [1 × dv"]] + U["11 x3"] --> V["α₃,₁ → α₃,₂ → α₃,₃"] + W["12 x3"] --> X["α₃,₁ → α₃,₂ → α₃,₃"] + Y["13 x3"] --> Z["α₃,₁ → α₃,₂ → α₃,₃"] + AA["14 x3"] --> AB["α₃,₁ → α₃,₂ → α₃,₃"] + AC["15 x3"] --> AD["α₃,₁ → α₃,₂ → α₃,₃"] + AE["16 x3"] --> AF["α₃,₁ → α₃,₂ → α₃,₃"] + AG["17 x3"] --> AH["α₃,₁ → α₃,₂ → α₃,₃"] + AI["18 x3"] --> AJ["α₃,₁ → α₃,₂ → α₃,₃"] + AK["19 x3"] --> AL["α₃,₁ → α₃,₂ → α₃,₃"] + AM["20 x3"] --> AN["α₃,₁ → α₃,₂ → α₃,₃"] + AO["21 x3"] --> AP["α₃,₁ → α₃,₂ → α₃,₃"] + AQ["22 x3"] --> AR["α₃,₁ → α₃,₂ → α₃,₃"] + AS["23 x3"] --> AT["α₃,₁ → α₃,₂ → α₃,₃"] + AU["24 x3"] --> AV["α₃,₁ → α₃,₂ → α₃,₃"] + AW["25 x3"] --> AX["α₃,₁ → α₃,₂ → α₃,₃"] + AY["26 x3"] --> AZ["α₃,₁ → α₃,₂ → α₃,₃"] + BA["27 x3"] --> BB["α₃,₁ → α₃,₂ → α₃,₃"] + BC["28 x3"] --> BD["α₃,₁ → α₃,₂ → α₃,₃"] + BE["29 x3"] --> BF["α₃,₁ → α₃,₂ → α₃,₃"] + BG["30 x3"] --> BH["α₃,₁ → α₃,₂ → α₃,₃"] + BI["31 x3"] --> BJ["α₃,₁ → α₃,₂ → α₃,₃"] + BK["32 x3"] --> BL["α₃,₁ → α₃,₂ → α₃,₃"] + BM["33 x3"] --> BN["α₃,₁ → α₃,₂ → α₃,₃"] + BO["34 x3"] --> BP["α₃,₁ → α₃,₂ → α₃,₃"] + BP --> BQ["X1: k q v W^K W^Q W^V X1: [1 × d"]] + BX["X2: k q v W^K W^Q W^V X2: [1 × d"]] + BY["X3: k q v W^K W^Q W^V X3: [1 × d"]] +``` +
+ +Figure 8.5 Calculating the value of $\mathbf{a}_3$ , the third element of a sequence using causal (left-to-right) self-attention. + +Section 8.2 the output $h_{i}$ of each transformer block, as well as the intermediate vectors inside the transformer block also have the same dimensionality $[1 \times d]$ . Having everything be the same dimensionality makes the transformer very modular. + +So let's talk shapes. How do we get from $[1 \times d]$ at the input to $[1 \times d]$ at the output? Let's look at all the internal shapes. We'll have a dimension $d_k$ for the query and key vectors. The query vector and the key vector are both dimensionality $[1 \times d_k]$ , so we can take their dot product $\mathbf{q}_i \cdot \mathbf{k}_j$ to produce a scalar. We'll have a separate dimension $d_v$ for the value vectors. The transform matrix $\mathbf{W}^{\mathbf{Q}}$ has shape $[d \times d_k]$ , $\mathbf{W}^{\mathbf{K}}$ is $[d \times d_k]$ , and $\mathbf{W}^{\mathbf{V}}$ is $[d \times d_v]$ . So the output of $\mathbf{head}_i$ in equation Eq. 8.13 is of shape $[1 \times d_v]$ . To get the desired output shape $[1 \times d]$ we'll need to reshape the head output, and so $\mathbf{W}^{\mathbf{O}}$ is of shape $[d_v \times d]$ . In the original transformer work (Vaswani et al., 2017), $d$ was 512, $d_k$ and $d_v$ were both 64. + +Multi-head Attention Equations 8.11-8.13 describe a single attention head. But actually, transformers use multiple attention heads. The intuition is that each head might be attending to the context for different purposes: heads might be specialized to represent different linguistic relationships between context elements and the current token, or to look for particular kinds of patterns in the context. + +So in multi-head attention we have A separate attention heads that reside in parallel layers at the same depth in a model, each with its own set of parameters that allows the head to model different aspects of the relationships among inputs. Thus each head i in a self-attention layer has its own set of query, key, and value matrices: $W^{Qi}$ , $W^{Ki}$ , and $W^{Vi}$ . These are used to project the inputs into separate query, key, and value embeddings for each head. + +When using multiple heads the model dimension d is still used for the input and output, the query and key embeddings have dimensionality $d_{k}$ , and the value embeddings are of dimensionality $d_{v}$ (again, in the original transformer paper $d_{k}=$ + +multi-head attention + +$d_{v}=64, A=8,$ and $d=512$ ). Thus for each head $i$ , we have weight layers $\mathbf{W}^{\mathbf{Q}\mathbf{i}}$ of shape $[d\times d_{k}]$ , $\mathbf{W}^{\mathbf{K}\mathbf{i}}$ of shape $[d\times d_{k}]$ , and $\mathbf{W}^{\mathbf{V}\mathbf{i}}$ of shape $[d\times d_{v}]$ . + +Below are the equations for attention augmented with multiple heads; Fig. 8.6 shows an intuition. + +$$ +\mathbf {q} _ {i} ^ {c} = \mathbf {x} _ {i} \mathbf {W} ^ {\mathbf {Q c}}; \quad \mathbf {k} _ {j} ^ {c} = \mathbf {x} _ {j} \mathbf {W} ^ {\mathbf {K c}}; \quad \mathbf {v} _ {j} ^ {c} = \mathbf {x} _ {j} \mathbf {W} ^ {\mathbf {V c}}; \quad \forall c 1 \leq c \leq A \tag {8.15} +$$ + +$$ +\operatorname{score} ^ {c} (\mathbf {x} _ {i}, \mathbf {x} _ {j}) = \frac {\mathbf {q} _ {i} ^ {c} \cdot \mathbf {k} _ {j} ^ {c}}{\sqrt {d _ {k}}} \tag {8.16} +$$ + +$$ +\alpha_ {i j} ^ {c} = \text { softmax } (\text { score } ^ {c} (\mathbf {x} _ {i}, \mathbf {x} _ {j})) \forall j \leq i \tag {8.17} +$$ + +$$ +\mathbf {h e a d} _ {i} ^ {c} = \sum_ {j \leq i} \alpha_ {i j} ^ {c} \mathbf {v} _ {j} ^ {c} \tag {8.18} +$$ + +$$ +\mathbf {a} _ {i} = \left(\mathbf {h e a d} ^ {1} \oplus \mathbf {h e a d} ^ {2} \dots \oplus \mathbf {h e a d} ^ {A}\right) \mathbf {W} ^ {O} \tag {8.19} +$$ + +$$ +\text { MultiHeadAttention } (\mathbf {x} _ {i}, [ \mathbf {x} _ {1}, \dots , \mathbf {x} _ {i - 1} ]) = \mathbf {a} _ {i} \tag {8.20} +$$ + +Note in Eq. 8.20 that MultiHeadAttention is a function of the current input $\mathbf{x}_i$ , as well as all the other inputs. For the causal or left-to-right attention that we use in this chapter, the other inputs are only to the left, but we'll also see a version of attention in Chapter 9 where attention is a function of the tokens to the right as well. We'll return to this idea about causal inputs in Eq. 8.34 when we introduce the idea of masking the right context. + +The output of each of the A heads is of shape $[1 \times d_{v}]$ , and so the output of the multi-head layer with A heads consists of A vectors of shape $[1 \times d_{v}]$ . These are concatenated to produce a single output with dimensionality $[1 \times Ad_{v}]$ . Then we use yet another linear projection $W^{0} \in R^{Ad_{v} \times d}$ to reshape it, resulting in the multi-head attention vector $a_{i}$ with the correct output shape $[1 \times d]$ at each input i. + +## 8.2 Transformer Blocks + +residual stream + +The self-attention calculation lies at the core of what's called a transformer block, which, in addition to the self-attention layer, includes three other kinds of layers: (1) a feedforward layer, (2) residual connections, and (3) normalizing layers (colloquially called “layer norm”). + +Fig. 8.7 illustrates a transformer block, sketching a common way of thinking about the block that is called the residual stream (Elhage et al., 2021). In the residual stream viewpoint, we consider the processing of an individual token i through the transformer block as a single stream of d-dimensional representations for token position i. This residual stream starts with the original input vector, and the various components read their input from the residual stream and add their output back into the stream. + +The input at the bottom of the stream is an embedding for a token, which has dimensionality d. This initial embedding gets passed up (by residual connections), and is progressively added to by the other components of the transformer: the attention layer that we have seen, and the feedforward layer that we will introduce. Before the attention and feedforward layer is a computation called the layer norm. + +Thus the initial vector is passed through a layer norm and attention layer, and the result is added back into the stream, in this case to the original input vector $x_{i}$ . And then this summed vector is again passed through another layer norm and a feedforward layer, and the output of those is added back into the residual, and we'll use $\mathbf{h}_i$ to refer to the resulting output of the transformer block for token $i$ . + +![](images/ac23c1c97b00da820b37fe355c65936205cb97e0f8c30d3a66c95da8c6896c22.jpg) + +
+flowchart + +```mermaid +graph TD + A["Project to final representation W^O"] --> B["Concatenate Outputs"] + B --> C1["Head 1 W^K1 W^V1 W^Q1"] + B --> C2["Head 2 W^K2 W^V2 W^Q2"] + B --> C3["Head 3 W^K3 W^V3 W^Q3"] + B --> C4["Head 4 W^K4 W^V4 W^Q4"] + C1 --> D1["x_{i-3} ... x_{i-2} ... x_{i-1}"] + C2 --> D2["x_{i-3} ... x_{i-2} ... x_{i-1}"] + C3 --> D3["x_{i-3} ... x_{i-2} ... x_{i-1}"] + C4 --> D4["x_{i-3} ... x_{i-2} ... x_{i-1}"] + style A fill:#f9f,stroke:#333 + style B fill:#ccf,stroke:#333 + style C1 fill:#cfc,stroke:#333 + style C2 fill:#cfc,stroke:#333 + style C3 fill:#cfc,stroke:#333 + style C4 fill:#cfc,stroke:#333 +``` +
+ +Figure 8.6 The multi-head attention computation for input $x_{i}$ , producing output $a_{i}$ . A multi-head attention layer has A heads, each with its own query, key, and value weight matrices. In this figure, we show A = 4, a smaller value than is usually used, just to fit on the page. The outputs from each of the heads are of shape $[1 \times d_{v}]$ and are concatenated and then projected into a different space by the $W_{0}$ matrix. Usually the dimensionality $d_{v}$ of the heads is set so that $d_{v} = d/A$ , with the result that $W_{0}$ is a square matrix of shape $[Ad_{v} \times d] = [d \times d]$ . usually of the same size, then projected d, thus producing an output of the same size as the input. + +![](images/2fa93647b739c2075d59a7dc012e40c7e4ae83b5a5d536f7e6ff8ddca1cd8c38.jpg) + +
+flowchart + +```mermaid +graph TD + A["x_{i-2}"] --> B["MultiHead Attention"] + C["x_{i-1}"] --> D["MultiHead Attention"] + E["x_i"] --> F["MultiHead Attention"] + G["h_i"] --> H["MultiHead Attention"] + I["h_i"] --> J["MultiHead Attention"] + K["Feedforward"] --> L["Layer Norm"] + M["..."] --> N["..."] + O["Layer Norm"] --> P["Layer Norm"] + Q["+"] --> R["+"] + S["+"] --> T["+"] + U["+"] --> V["+"] + W["+"] --> X["+"] + Y["+"] --> Z["+"] + style A fill:#f9f,stroke:#333 + style C fill:#f9f,stroke:#333 + style E fill:#f9f,stroke:#333 + style Q fill:#f9f,stroke:#333 + style M fill:#f9f,stroke:#333 + style O fill:#f9f,stroke:#333 + style R fill:#ccf,stroke:#333 + style S fill:#ccf,stroke:#333 + style T fill:#ccf,stroke:#333 + style U fill:#ccf,stroke:#333 + style V fill:#ccf,stroke:#333 + style W fill:#ccf,stroke:#333 + style X fill:#ccf,stroke:#333 +``` +
+ +Figure 8.7 The architecture of a transformer block showing the residual stream, showing how most information flows up through the residual stream, and only the attention module is sensitive to information from other streams at prior token positions. In this figure and throughout the chapter, we use the prenorm version of the architecture, in which the layer norms happen before the attention and feedforward layers rather than after. The first + +We've already seen the attention layer, so let's now introduce the feedforward and layer norm computations in the context of processing a single input $\mathbf{x}_i$ at token + +position i. + +Feedforward layer The feedforward layer is a fully-connected 2-layer network, i.e., one hidden layer, two weight matrices, as introduced in Chapter 6. The weights are the same for each token position i, but are different from layer to layer. It is common to make the dimensionality $d_{ff}$ of the hidden layer of the feedforward network be larger than the model dimensionality d. (For example in the original transformer model, d = 512 and $d_{ff} = 2048$ .) + +$$ +\operatorname{FFN} \left(\mathbf {x} _ {i}\right) = \operatorname{ReLU} \left(\mathbf {x} _ {i} \mathbf {W} _ {\mathbf {1}} + b _ {1}\right) \mathbf {W} _ {\mathbf {2}} + b _ {2} \tag {8.21} +$$ + +layer norm + +Layer Norm At two stages in the transformer block we normalize the vector (Ba et al., 2016). This process, called layer norm (short for layer normalization), is one of many forms of normalization that can be used to improve training performance in deep neural networks by keeping the values of a hidden layer in a range that facilitates gradient-based training. + +Layer norm is a variation of the z-score from statistics, applied to a single vector in a hidden layer. That is, the term layer norm is a bit confusing; layer norm is not applied to an entire transformer layer, but just to the embedding vector of a single token. Thus the input to layer norm is a single vector of dimensionality d and the output is that vector normalized, again of dimensionality d. The first step in layer normalization is to calculate the mean, $\mu$ , and standard deviation, $\sigma$ , over the elements of the vector to be normalized. Given an embedding vector x of dimensionality d, these values are calculated as follows. + +$$ +\mu = \frac {1}{d} \sum_ {i = 1} ^ {d} x _ {i} \tag {8.22} +$$ + +$$ +\sigma = \sqrt {\frac {1}{d} \sum_ {i = 1} ^ {d} (x _ {i} - \mu) ^ {2}} \tag {8.23} +$$ + +Given these values, the vector components are normalized by subtracting the mean from each and dividing by the standard deviation. The result of this computation is a new vector with zero mean and a standard deviation of one. + +$$ +\hat {\mathbf {x}} = \frac {(\mathbf {x} - \mu)}{\sigma} \tag {8.24} +$$ + +Finally, in the standard implementation of layer normalization, two learnable parameters, $\gamma$ and $\beta$ , representing gain and offset values, are introduced. + +$$ +\text { LayerNorm } (\mathbf {x}) = \gamma \frac {(\mathbf {x} - \mu)}{\sigma} + \beta \tag {8.25} +$$ + +Putting it all together The function computed by a transformer block can be expressed by breaking it down with one equation for each component computation, using $\mathbf{t}$ (of shape $[1\times d]$ ) to stand for transformer and superscripts to demarcate each computation inside the block: + +$$ +\mathbf {t} _ {i} ^ {\mathbf {1}} = \text { LayerNorm } (\mathbf {x} _ {i}) \tag {8.26} +$$ + +$$ +\mathbf {t} _ {i} ^ {2} = \text { MultiHeadAttention } (\mathbf {t} _ {i} ^ {1}, [ \mathbf {t} _ {1} ^ {1}, \dots , \mathbf {t} _ {N} ^ {1} ]) \tag {8.27} +$$ + +$$ +\mathbf {t} _ {i} ^ {3} = \mathbf {t} _ {i} ^ {2} + \mathbf {x} _ {i} \tag {8.28} +$$ + +$$ +\mathbf {t} _ {i} ^ {4} = \text { LayerNorm } (\mathbf {t} _ {i} ^ {3}) \tag {8.29} +$$ + +$$ +\mathbf {t} _ {i} ^ {\mathbf {5}} = \operatorname{FFN} (\mathbf {t} _ {i} ^ {\mathbf {4}}) \tag {8.30} +$$ + +$$ +\mathbf {h} _ {i} = \mathbf {t} _ {i} ^ {5} + \mathbf {t} _ {i} ^ {3} \tag {8.31} +$$ + +token-mixing + +Notice that the only component that takes as input information from other tokens (other residual streams) is multi-head attention, which (as we see from Eq. 8.27) looks at all the neighboring tokens in the context. The output from attention, however, is then added into this token's embedding stream. In fact, Elhage et al. (2021) show that we can view attention heads as literally moving information from the residual stream of a neighboring token into the current stream. The high-dimensional embedding space at each position thus contains information about the current token and about neighboring tokens, albeit in different subspaces of the vector space. Fig. 8.8 shows a visualization of this movement. We therefore call the attention function the token-mixing component of the architecture, because it mixes information from neighboring token streams into the current stream. + +![](images/1415c4fb26602e8ddf6e7ba57246ab86c0013ce3b99ad32d6a6ca78bbaafab36.jpg) + +
+text_image + +Token A +residual +stream +Token B +residual +stream +
+ +Figure 8.8 An attention head can move information from token A's residual stream into token B's residual stream. + +Crucially, the input and output dimensions of transformer blocks are matched so they can be stacked. Each token vector $x_{i}$ at the input to the block has dimensionality d, and the output $h_{i}$ also has dimensionality d. Transformers for large language models stack many of these blocks, from 12 layers (used for the T5 or GPT-3-small language models) to 96 layers (used for GPT-3 large), to even more for more recent models. We’ll come back to this issue of stacking in a bit. + +Equation 8.26 and following are just the equation for a single transformer block, but the residual stream metaphor goes through all the transformer layers, from the first transformer blocks to the 12th, in a 12-layer transformer. At the earlier transformer blocks, the residual stream is representing the current token. At the highest transformer blocks, the residual stream is usually representing the following token, since at the very end it's being trained to predict the next token. + +Once we stack many blocks, there is one more requirement: at the very end of the last (highest) transformer block, there is a single extra layer norm that is run on the last $h_{i}$ of each token stream (just below the language model head layer that we will define soon). $^{2}$ + +## 8.3 Parallelizing computation using a single matrix X + +This description of multi-head attention and the rest of the transformer block has been from the perspective of computing a single output at a single time step $i$ in a single residual stream. But as we pointed out earlier, the attention computation performed for each token to compute $\mathbf{a}_i$ is independent of the computation for each other token, and that's also true for all the computation in the transformer block computing $\mathbf{h}_i$ from the input $\mathbf{x}_i$ . That means we can easily parallelize the entire computation, taking advantage of efficient matrix multiplication routines. + +We do this by packing the input embeddings for the $N$ tokens of the input sequence into a single matrix $\mathbf{X}$ of size $[N\times d]$ . Each row of $\mathbf{X}$ is the embedding of one token of the input. Transformers for large language models commonly have an input length $N$ from 1K to 32K; much longer contexts of 128K or even up to millions of tokens can also be achieved with architectural changes like special long-context mechanisms that we don't discuss here. So for vanilla transformers, we can think of $\mathbf{X}$ having between 1K and 32K rows, each of the dimensionality of the embedding $d$ (the model dimension). + +Parallelizing attention Let's first see this for a single attention head and then turn to multiple heads, and then add in the rest of the components in the transformer block. For one head we multiply $\mathbf{X}$ by the query, key, and value matrices $\mathbf{W}^{\mathbf{Q}}$ of shape $[d\times d_k]$ , $\mathbf{W}^{\mathbf{K}}$ of shape $[d\times d_k]$ , and $\mathbf{W}^{\mathbf{V}}$ of shape $[d\times d_v]$ , to produce matrices $\mathbf{Q}$ of shape $[N\times d_k]$ , $\mathbf{K}$ of shape $[N\times d_k]$ , and $\mathbf{V}$ of shape $[N\times d_v]$ , containing all the key, query, and value vectors: + +$$ +\mathbf {Q} = \mathbf {X W} ^ {\mathrm{Q}}; \quad \mathbf {K} = \mathbf {X W} ^ {\mathrm{K}}; \quad \mathbf {V} = \mathbf {X W} ^ {\mathrm{V}} \tag {8.32} +$$ + +Given these matrices we can compute all the requisite query-key comparisons simultaneously by multiplying Q and $K^{T}$ in a single matrix multiplication. The product is of shape $N \times N$ , visualized in Fig. 8.9. + +![](images/187fec547e0fd396ddcf2d47baeb28832c5381e7ce9bc71d2b07c3c9ed6f0be6.jpg) + +
+text_image + +q1·k1 q1·k2 q1·k3 q1·k4 +q2·k1 q2·k2 q2·k3 q2·k4 +q3·k1 q3·k2 q3·k3 q3·k4 +q4·k1 q4·k2 q4·k3 q4·k4 +N +N +
+ +Figure 8.9 The $N \times N$ $\mathbf{QK}^{\top}$ matrix showing how it computes all $q_{i} \cdot k_{j}$ comparisons in a single matrix multiple. + +Once we have this $\mathbf{QK}^{\top}$ matrix, we can very efficiently scale these scores, take the softmax, and then multiply the result by $\mathbf{V}$ resulting in a matrix of shape $N\times d$ : a vector embedding representation for each token in the input. We've reduced the entire self-attention step for an entire sequence of $N$ tokens for one head to the following computation: + +$$ +\mathbf {h e a d} = \text { softmax } \left(\operatorname{mask} \left(\frac {\mathbf {Q K} ^ {\top}}{\sqrt {d _ {k}}}\right)\right) \mathbf {V} \tag {8.33} +$$ + +$$ +\mathbf {A} = \text { head } \mathbf {W} ^ {0} \tag {8.34} +$$ + +Masking out the future You may have noticed that we introduced a mask function in Eq. 8.34 above. This is because the self-attention computation as we've described it has a problem: the calculation of $\mathbf{QK}^{\mathrm{T}}$ results in a score for each query value to every key value, including those that follow the query. This is inappropriate in the setting of language modeling: guessing the next word is pretty simple if you already know it! To fix this, the elements in the upper-triangular portion of the matrix are set to $-\infty$ , which the softmax will turn to zero, thus eliminating any knowledge of words that follow in the sequence. This is done in practice by adding a mask matrix $M$ in which $M_{ij} = -\infty \forall j > i$ (i.e. for the upper-triangular portion) and $M_{ij} = 0$ otherwise. Fig. 8.10 shows the resulting masked $\mathbf{QK}^{\mathrm{T}}$ matrix. (we'll see in Chapter 9 how to make use of words in the future for tasks that need it). + +N + +
q1·k1-∞-∞-∞
q2·k1q2·k2-∞-∞
q3·k1q3·k2q3·k3-∞
q4·k1q4·k2q4·k3q4·k4
+ +N +Figure 8.10 The $N \times N QK^{T}$ matrix showing the $q_{i} \cdot k_{j}$ values, with the upper-triangle portion of the comparisons matrix zeroed out (set to $-\infty$ , which the softmax will turn to zero). + +Fig. 8.11 shows a schematic of all the computations for a single attention head parallelized in matrix form. +![](images/2ad1f620305b178f211dfa04ede8865bbce7265a6cdc78840eade6b0134c2be4.jpg) + +
+text_image + +X +Input +Token 1 +Input +Token 2 +Input +Token 3 +Input +Token 4 +N x d +W^Q +x +d x d_k += Q +Query +Token 1 +Query +Token 2 +Query +Token 3 +Query +Token 4 +N x d_k +X +Input +Token 1 +Input +Token 2 +Input +Token 3 +Input +Token 4 +W^K +x +d x d_k += K +Key +Token 1 +Key +Token 2 +Key +Token 3 +Key +Token 4 +N x d_k +X +Input +Token 1 +Input +Token 2 +Input +Token 3 +Input +Token 4 +W^V +x +d x d_v += V +Value +Token 1 +Value +Token 2 +Value +Token 3 +Value +Token 4 +N x d_v +mask +Q +q1 X K^T = QK^T masked V A +q2 \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u2667 a1 +q3 d_k x N q1·k1 q1·k2 q1·k3 q1·k4 q2·k1 q2·k2 q2·k3 q2·k4 q3·k1 q3·k2 q3·k3 q3·k4 q4·k1 q4·k2 q4·k3 q4·k4 N x N N x N N x d_v N x d_v +
+ +Figure 8.11 Schematic of the attention computation for a single attention head in parallel. The first row shows the computation of the Q, K, and V matrices. The second row shows the computation of $QK^{T}$ , the masking (the softmax computation and the normalizing by dimensionality are not shown) and then the weighted sum of the value vectors to get the final attention vectors. + +Fig. 8.9 and Fig. 8.10 also make it clear that attention is quadratic in the length of the input, since at each layer we need to compute dot products between each pair of tokens in the input. This makes it expensive to compute attention over very long documents (like entire novels). Nonetheless modern large language models manage to use quite long contexts of thousands or tens of thousands of tokens. + +Parallelizing multi-head attention In multi-head attention, as with self-attention, the input and output have the model dimension d, the key and query embeddings have dimensionality $d_{k}$ , and the value embeddings are of dimensionality $d_{v}$ (again, in the original transformer paper $d_{k} = d_{v} = 64$ , A = 8, and d = 512). Thus for each head c, we have weight layers $W^{Q}_{c}$ of shape $[d \times d_{k}]$ , $W^{K}_{c}$ of shape $[d \times d_{k}]$ , and $W^{V}_{c}$ of shape $[d \times d_{v}]$ , and these get multiplied by the inputs packed into X to produce Q of shape $[N \times d_{k}]$ , K of shape $[N \times d_{k}]$ , and V of shape $[N \times d_{v}]$ . The output of each of the A heads is of shape $[N \times d_{v}]$ , and so the output of the multi-head layer with A heads consists of A matrices of shape $[N \times d_{v}]$ . To make use of these matrices in further processing, they are concatenated to produce a single output with dimensionality $[N \times Ad_{v}]$ . Finally, we use a final linear projection $W^{O}$ of shape $[Ad_{v} \times d]$ , that reshapes it to the original output dimension for each token. Multiplying the concatenated $[N \times Ad_{v}]$ matrix output by $W^{O}$ of shape $[Ad_{v} \times d]$ yields the self-attention output A of shape $[N \times d]$ . + +$$ +\mathbf {Q} ^ {\mathrm{i}} = \mathbf {X W} ^ {\mathrm{Qi}}; \quad \mathbf {K} ^ {\mathrm{i}} = \mathbf {X W} ^ {\mathrm{Ki}}; \quad \mathbf {V} ^ {\mathrm{i}} = \mathbf {X W} ^ {\mathrm{Vi}} \tag {8.35} +$$ + +$$ +\mathbf {h e a d} _ {i} = \text { SelfAttention } (\mathbf {Q} ^ {\mathrm{i}}, \mathbf {K} ^ {\mathrm{i}}, \mathbf {V} ^ {\mathrm{i}}) = \text { softmax } \left(\text { mask } \left(\frac {\mathbf {Q} ^ {\mathrm{i}} \mathbf {K} ^ {\mathrm{iT}}}{\sqrt {d _ {k}}}\right)\right) \mathbf {V} ^ {\mathrm{i}} \tag {8.36} +$$ + +$$ +\text { MultiHeadAttention } (\mathbf {X}) = \left(\mathbf {h e a d} _ {1} \oplus \mathbf {h e a d} _ {2} \dots \oplus \mathbf {h e a d} _ {A}\right) \mathbf {W} ^ {\mathbf {0}} \tag {8.37} +$$ + +Putting it all together with the parallel input matrix X The function computed in parallel by an entire layer of N transformer blocks—each block over one of the N input tokens—can be expressed as: + +$$ +\mathbf {O} = \mathbf {X} + \text { MultiHeadAttention } (\text { LayerNorm } (\mathbf {X})) \tag {8.38} +$$ + +$$ +\mathbf {H} = \mathbf {O} + \operatorname{FFN} (\text { LayerNorm } (\mathbf {O})) \tag {8.39} +$$ + +Note that in Eq. 8.38 we are using X to mean the input to the layer, wherever it comes from. For the first layer, as we will see in the next section, that input is the initial word + positional embedding vectors that we have been describing by X. But for subsequent layers k, the input is the output from the previous layer $H^{k-1}$ . We can also break down the computation performed in a transformer layer, showing one equation for each component computation. We'll use T (of shape $[N \times d]$ ) to stand for transformer and superscripts to demarcate each computation inside the block, and again use X to mean the input to the block from the previous layer or the initial embedding: + +$$ +\mathbf {T} ^ {1} = \text { LayerNorm } (\mathbf {X}) \tag {8.40} +$$ + +$$ +\mathbf {T} ^ {2} = \text { MultiHeadAttention } (\mathbf {T} ^ {1}) \tag {8.41} +$$ + +$$ +\mathbf {T} ^ {3} = \mathbf {T} ^ {2} + \mathbf {X} \tag {8.42} +$$ + +$$ +\mathbf {T} ^ {4} = \text { LayerNorm } (\mathbf {T} ^ {3}) \tag {8.43} +$$ + +$$ +\mathbf {T} ^ {5} = \operatorname{FFN} (\mathbf {T} ^ {4}) \tag {8.44} +$$ + +$$ +\mathbf {H} = \mathbf {T} ^ {5} + \mathbf {T} ^ {3} \tag {8.45} +$$ + +Here when we use a notation like $\mathrm{FFN}(\mathbf{T}^{3})$ we mean that the same FFN is applied in parallel to each of the N embedding vectors in the window. Similarly, each of the + +N tokens is normed in parallel in the LayerNorm. Crucially, the input and output dimensions of transformer blocks are matched so they can be stacked. Since each token $x_{i}$ at the input to the block is represented by an embedding of dimensionality $[1 \times d]$ , that means the input X and output H are both of shape $[N \times d]$ . + +## 8.4 The input: embeddings for token and position + +embedding + +Let's talk about where the input $\mathbf{X}$ comes from. Given a sequence of $N$ tokens ( $N$ is the context length in tokens), the matrix $\mathbf{X}$ of shape $[N \times d]$ has an embedding for each word in the context. The transformer does this by separately computing two embeddings: an input token embedding, and an input positional embedding. + +A token embedding, introduced in Chapter 6, is a vector of dimension d that will be our initial representation for the input token. (As we pass vectors up through the transformer layers in the residual stream, this embedding representation will change and grow, incorporating context and playing a different role depending on the kind of language model we are building.) The set of initial embeddings are stored in the embedding matrix E, which has a row for each of the $|V|$ tokens in the vocabulary. (Reminder that V here means the vocabulary of tokens, this V is not related to the value vector.) Thus each word is a row vector of d dimensions, and E has shape $[|V| \times d]$ . + +Given an input token string like Thanks for all the we first convert the tokens into vocabulary indices (these were created when we first tokenized the input using BPE or SentencePiece). So the representation of thanks for all the might be w = [5, 4000, 10532, 2224]. Next we use indexing to select the corresponding rows from E, (row 5, row 4000, row 10532, row 2224). + +Another way to think about selecting token embeddings from the embedding matrix is to represent tokens as one-hot vectors of shape $[1 \times |V|]$ , i.e., with one dimension for each word in the vocabulary. Recall that in a one-hot vector all the elements are 0 except one, the element whose dimension is the word's index in the vocabulary, which has value 1. So if the word “thanks” has index 5 in the vocabulary, $x_{5} = 1$ , and $x_{i} = 0 \forall i \neq 5$ , as shown here: + +$$ +[ \emptyset \emptyset \emptyset \emptyset 1 \emptyset \emptyset \dots \emptyset \emptyset \emptyset \emptyset ] +$$ + +$$ +\begin{array}{c c c c c c c c c c} 1 & 2 & 3 & 4 & 5 & 6 & 7 & \dots & \dots & | V | \end{array} +$$ + +Multiplying by a one-hot vector that has only one non-zero element $x_{i}=1$ simply selects out the relevant row vector for word i, resulting in the embedding for word i, as depicted in Fig. 8.12. + +![](images/08f9d83e4e6a1f27712a4f87cd95fef80bac5ce53ec46181a78302d9e4117d06.jpg) + +
+text_image + +1 5 |V| +0000100...0000 +× +5 +E += 1 d +|V| +
+ +Figure 8.12 Selecting the embedding vector for word $V_{5}$ by multiplying the embedding matrix E with a one-hot vector with a 1 in index 5. + +We can extend this idea to represent the entire token sequence as a matrix of one-hot vectors, one for each of the $N$ positions in the transformer's context window, as shown in Fig. 8.13. + +one-hot vector + +![](images/7a2ef0d0f306cc92a464e320223ccdcb5677279123697b7a07f0e1a88f202230.jpg) + +
+text_image + +|V| +0 0 0 0 |1 |0 0 ... 0 0 0 0 +0 0 0 0 0 0 0 ... 0 0 |1 |0 +1 |0 0 0 0 0 0 ... 0 0 0 0 +... +N +0 0 0 0 |1 |0 0 ... 0 0 0 0 +× +d +E += +d +N +
+ +Figure 8.13 Selecting the embedding matrix for the input sequence of token ids W by multiplying a one-hot matrix corresponding to W by the embedding matrix E. + +positional embeddings + +absolute position + +These token embeddings are not position-dependent. To represent the position of each token in the sequence, we combine these token embeddings with positional embeddings specific to each position in an input sequence. + +Where do we get these positional embeddings? The simplest method, called absolute position, is to start with randomly initialized embeddings corresponding to each possible input position up to some maximum length. For example, just as we have an embedding for the word fish, we'll have an embedding for the position 3. As with word embeddings, these positional embeddings are learned along with other parameters during training. We can store them in a matrix $E_{\mathrm{pos}}$ of shape $[N \times d]$ . + +To produce an input embedding that captures positional information, we just add the word embedding for each input to its corresponding positional embedding. The individual token and position embeddings are both of size $[1 \times d]$ , so their sum is also $[1 \times d]$ , This new embedding serves as the input for further processing. Fig. 8.14 shows the idea. + +![](images/cdd2a7ae6c8e243a3af268e23d657c2aaaef94c003d4c58c6d6662a55e1085e8.jpg) + +
+flowchart + +```mermaid +graph TD + A["Transformer Block"] --> B["X = Composite Embeddings (word + position)"] + B --> C["+"] + C --> D["janet"] + C --> E["will"] + C --> F["back"] + C --> G["the"] + C --> H["bill"] + B --> I["Word Embeddings"] + I --> J["Position Embeddings"] + J --> K["1"] + J --> L["2"] + J --> M["3"] + J --> N["4"] + J --> O["5"] +``` +
+ +Figure 8.14 A simple way to model position: add an embedding of the absolute position to the token embedding to produce a new embedding of the same dimensionality. + +The final representation of the input, the matrix X, is an $[N \times d]$ matrix in which each row i is the representation of the ith token in the input, computed by adding $\mathbf{E}[id(i)]$ —the embedding of the id of the token that occurred at position i—, to P[i], the positional embedding of position i. + +A potential problem with the simple position embedding approach is that there will be plenty of training examples for the initial positions in our inputs and correspondingly fewer at the outer length limits. These latter embeddings may be poorly trained and may not generalize well during testing. An alternative is to choose a static function that maps integer inputs to real-valued vectors in a way that better handles sequences of arbitrary length. A combination of sine and cosine functions with differing frequencies was used in the original transformer work. Sinusoidal position embeddings may also help in capturing the inherent relationships among the + +relative position + +positions, like the fact that position 4 in an input is more closely related to position 5 than it is to position 17. + +A more complex style of positional embedding methods extend this idea of capturing relationships even further to directly represent relative position instead of absolute position, often implemented in the attention mechanism at each layer rather than being added once at the initial input. + +## 8.5 The Language Modeling Head + +language modeling head +head + +The last component of the transformer we must introduce is the language modeling head. Here we are using the word head to mean the additional neural circuitry we add on top of the basic transformer architecture when we apply pretrained transformer models to various tasks. The language modeling head is the circuitry we need to do language modeling. + +Recall that language models, from the simple n-gram models of Chapter 3 through the feedforward and RNN language models of Chapter 6 and Chapter 13, are word predictors. Given a context of words, they assign a probability to each possible next word. For example, if the preceding context is “Thanks for all the” and we want to know how likely the next word is “fish” we would compute: + +## $P(\text{fish}|\text{Thanks for all the})$ + +Language models give us the ability to assign such a conditional probability to every possible next word, giving us a distribution over the entire vocabulary. The n-gram language models of Chapter 3 compute the probability of a word given counts of its occurrence with the $n - 1$ prior words. The context is thus of size $n - 1$ . For transformer language models, the context is the size of the transformer's context window, which can be quite large, like 32K tokens for large models (and much larger contexts of millions of words are possible with special long-context architectures). + +The job of the language modeling head is to take the output of the final transformer layer from the last token N and use it to predict the upcoming word at position $N+1$ . Fig. 8.15 shows how to accomplish this task, taking the output of the last token at the last layer (the d-dimensional output embedding of shape $[1 \times d]$ ) and producing a probability distribution over words (from which we will choose one to generate). + +The first module in Fig. 8.15 is a linear layer, whose job is to project from the output $h_N^L$ , which represents the output token embedding at position $N$ from the final block $L$ , (hence of shape $[1 \times d]$ ) to the logit vector, or score vector, that will have a single score for each of the $|V|$ possible words in the vocabulary $V$ . The logit vector $\mathbf{u}$ is thus of dimensionality $[1 \times |V|]$ . + +weight tying + +This linear layer can be learned, but more commonly we tie this matrix to (the transpose of) the embedding matrix E. Recall that in weight tying, we use the same weights for two different matrices in the model. Thus at the input stage of the transformer the embedding matrix (of shape $[|V| \times d]$ ) is used to map from a one-hot vector over the vocabulary (of shape $[1 \times |V|]$ ) to an embedding (of shape $[1 \times d]$ ). And then in the language model head, $E^{T}$ , the transpose of the embedding matrix (of shape $[d \times |V|]$ ) is used to map back from an embedding (shape $[1 \times d]$ ) to a vector over the vocabulary (shape $[1 \times |V|]$ ). In the learning process, E will be optimized to be good at doing both of these mappings. We therefore sometimes call the transpose $E^{T}$ the unembedding layer because it is performing this reverse mapping. + +unembedding + +![](images/c0f60283117c3f6f387ed00e782203c35743e82bd350fa6b6a088b9dad8fb018.jpg) + +
+flowchart + +```mermaid +graph TD + A["Layer L Transformer Block"] --> B["Softmax"] + B --> C["Unembedding layer U = E^T"] + C --> E["Softmax over vocabulary V"] + B --> H["y1"] + B --> I["y2"] + B --> J["..."] + B --> L["u1"] + B --> M["u2"] + B --> N["..."] + B --> P["h^L_1"] + B --> Q["h^L_2"] + P --> R["w1"] + Q --> S["w2"] + R --> T["..."] + S --> U["w_N"] + T --> V["h^L_N"] + U --> W["h^L_N"] + V --> X["1 x d"] +``` +
+ +Figure 8.15 The language modeling head: the circuit at the top of a transformer that maps from the output embedding for token N from the last transformer layer ( $h_{N}^{L}$ ) to a probability distribution over words in the vocabulary V. + +A softmax layer turns the logits u into the probabilities y over the vocabulary. + +$$ +\mathbf {u} = \mathbf {h} _ {\mathrm{N}} ^ {\mathrm{L}} \mathbf {E} ^ {\mathrm{T}} \tag {8.46} +$$ + +$$ +\mathbf {y} = \operatorname{softmax} (\mathbf {u}) \tag {8.47} +$$ + +We can use these probabilities to do things like help assign a probability to a given text. But the most important usage is to generate text, which we do by sampling a word from these probabilities y. We might sample the highest probability word ('greedy' decoding), or use another of the sampling methods from Section ?? or Section 8.6. + +In either case, whatever entry $y_{k}$ we choose from the probability vector $\mathbf{y}$ , we generate the word that has that index $k$ . + +Fig. 8.16 shows the total stacked architecture for one token $i$ . Note that the input to each transformer layer $x_{i}^{\ell}$ is the same as the output from the preceding layer $h_i^{\ell -1}$ . + +A terminological note before we conclude: You will sometimes see a transformer used for this kind of unidirectional causal language model called a decoder-only model. This is because this model constitutes roughly half of the encoder-decoder model for transformers that we'll see how to apply to machine translation in Chapter 12. (Confusingly, the original introduction of the transformer had an encoder-decoder architecture, and it was only later that the standard paradigm for causal language model was defined by using only the decoder part of this original architecture). + +decoder-only model + +## 8.6 More on Sampling + +The sampling methods we introduce below each have parameters that enable trading off two important factors in generation: quality and diversity. Methods that emphasize the most probable words tend to produce generations that are rated by people as more accurate, more coherent, and more factual, but also more boring and more repetitive. Methods that give a bit more weight to the middle-probability words tend to be more creative and more diverse, but less factual and more likely to be incoherent or otherwise low-quality. + +![](images/ddd6ed825da89b3ecb395938eda9eceaa5c76e0acede677834be053dec84ef47.jpg) + +
+flowchart + +```mermaid +graph TD + A["Input token W_i"] --> B["Layer 1"] + B --> C["Layer 2"] + C --> D["Layer L"] + D --> E["Layer 1"] + E --> F["Input Encoding"] + F --> G["Output w_i"] + H["Input Encoding"] --> I["x^1_i"] + I --> J["h^1_i = x^2_i"] + J --> K["Layer 2"] + K --> L["h^2_i = x^3_i"] + L --> M["Layer L"] + M --> N["..."] + N --> O["Layer L"] + O --> P["u1"] + O --> Q["u2"] + O --> R["uV"] + P --> S["y1"] + Q --> T["y2"] + S --> V["Sample token to generate at position i+1"] + T --> V + U --> V + V --> W["w_{i+1}"] + W --> X["Output w_i"] + style A fill:#f9f,stroke:#333 + style H fill:#f9f,stroke:#333 + style I fill:#ccf,stroke:#333 + style F fill:#ccf,stroke:#333 +``` +
+ +Figure 8.16 A transformer language model (decoder-only), stacking transformer blocks and mapping from an input token $w_{i}$ to a predicted next token $w_{i + 1}$ . + +## 8.6.1 Top-k sampling + +top-k sampling + +Top-k sampling is a simple generalization of greedy decoding. Instead of choosing the single most probable word to generate, we first truncate the distribution to the top k most likely words, renormalize to produce a legitimate probability distribution, and then randomly sample from within these k words according to their renormalized probabilities. More formally: + +1. Choose in advance a number of words $k$ +2. For each word in the vocabulary $V$ , use the language model to compute the likelihood of this word given the context $p(w_{t}|\mathbf{w}_{< t})$ +3. Sort the words by their likelihood, and throw away any word that is not one of the top k most probable words. +4. Renormalize the scores of the k words to be a legitimate probability distribu- + +tion. + +5. Randomly sample a word from within these remaining $k$ most-probable words according to its probability. + +When k = 1, top-k sampling is identical to greedy decoding. Setting k to a larger number than 1 leads us to sometimes select a word which is not necessarily the most probable, but is still probable enough, and whose choice results in generating more diverse but still high-enough-quality text. + +## 8.6.2 Nucleus or top-p sampling + +One problem with top-k sampling is that k is fixed, but the shape of the probability distribution over words differs in different contexts. If we set k = 10, sometimes the top 10 words will be very likely and include most of the probability mass, but other times the probability distribution will be flatter and the top 10 words will only include a small part of the probability mass. + +top-p sampling + +An alternative, called top-p sampling or nucleus sampling (Holtzman et al., 2020), is to keep not the top k words, but the top p percent of the probability mass. The goal is the same; to truncate the distribution to remove the very unlikely words. But by measuring probability rather than the number of words, the hope is that the measure will be more robust in very different contexts, dynamically increasing and decreasing the pool of word candidates. + +Given a distribution $P(w_{t}|\mathbf{w}_{ +flowchart + +```mermaid +graph TD + subgraph Input Tokens + A["Input Encoding"] --> B["x1"] + A --> C["x2"] + A --> D["x3"] + A --> E["x4"] + A --> F["x5"] + G["Stacked Transformer Blocks"] --> H["..."] + G --> I["..."] + G --> J["..."] + G --> K["..."] + G --> L["..."] + M["Language Modeling Head"] --> N["..."] + M --> O["..."] + M --> P["..."] + M --> Q["..."] + M --> R["..."] + M --> S["..."] + T["Last Layer"] --> U["..."] + T --> V["..."] + T --> W["..."] + T --> X["..."] + T --> Y["..."] + Z["Long Layer"] --> AA["..."] + Z --> AB["..."] + Z --> AC["..."] + Z --> AD["..."] + AE["Long Layer"] --> AF["..."] + AE --> AG["..."] + AE --> AH["..."] + AE --> AI["..."] + AJ["Long Layer"] --> AK["..."] + AJ --> AL["..."] + AJ --> AM["..."] + AJ --> AN["..."] + AO["Long Layer"] --> AP["..."] + AO --> AQ["..."] + AO --> AR["..."] + AO --> AS["..."] + AT["Long Layer"] --> AU["..."] + AT --> AV["..."] + AT --> AW["..."] + AX["Long Layer"] --> AY["..."] + AX --> AZ["..."] + AX --> BA["..."] + BB["Long Layer"] --> BC["..."] + BB --> BD["..."] + BB --> BE["..."] + BF["Long Layer"] --> BG["..."] + BF --> BH["..."] + BF --> BI["..."] + BJ["Long Layer"] --> BK["..."] + BJ --> BL["..."] + BJ --> BM["..."] + BN["Long Layer"] --> BO["..."] + BN --> BP["..."] + BN --> BQ["..."] + BR["Long Layer"] --> BS["..."] + BR --> BT["..."] + BR --> BU["..."] + BV["Long Layer"] --> BW["..."] + BV --> BX["..."] + BV --> BY["..."] + BZ["Stacked Transformer Blocks"] --> CA["x1"] + BZ --> CB["x2"] + BZ --> CC["x3"] + BZ --> CD["x4"] + BZ --> CE["x5"] + end + subgraph Input Tokens + CF["Xo"] --> CG["E"] + end + subgraph Input Encoding + CH["E"] --> CI["E"] + end + subgraph Input Tokens + CJ["E"] --> CK["E"] + end + subgraph Stacked Transformer Blocks + CL["x1"] --> CM["x2"] & CN["x3"] & CO["x4"] & CE["x5"] + end + subgraph Stacked Transformer Blocks + CD["x1"] & CN["x2"] & CO["x3"] & CE["x4"] & CE["x5"] + end + subgraph Stacked Transformer Blocks + DD["x1"] & CM["x2"] & CO["x3"] & CE["x4"] & CE["x5"] + end + subgraph Stacked Transformer Blocks + DF["XO"] & DG["XL"] --> DH["XL0"], DH["XL1"], DH["XL2"], DH["XL3"], DH["XL4"], DH["XL5"] + end + subgraph Stacked Transformer Blocks + ID["XL0"] & ID["XL1"] & ID["XL2"] & ID["XL3"] & ID["XL4"] & ID["XL5"] + end + subgraph Stacked Transformer Blocks + AE["XL0"] & AE["XL1"] & AE["XL2"] & AE["XL3"] & AE["XL4"] & AE["XL5"] + end + subgraph Stacked Transformer Blocks + AF["XL0"] & AF["XL1"] & AF["XL2"] & AF["XL3"] & AF["XL4"] & AF["XL5"] + end + subgraph Stacked Transformer Blocks + AG["XL0"] & AG["XL1"] & AG["XL2"] & AG["XL3"] & AG["XL4"] & AG["XL5"] + end + subgraph Stacked Transformer Blocks + AH["XL0"] & AH["XL1"] & AH["XL2"] & AH["XL3"] & AH["XL4"] & AH["XL5"] + end + subgraph Stacked Transformer Blocks + AI["XL0"] & AI["XL1"] & AI["XL2"] & AI["XL3"] & AI["XL4"] & AI["XL5"] + end + subgraph Stacked Transformer Blocks + AJ["XL0"] & AJ["XL1"] & AJ["XL2"] & AJ["XL3"] & AJ["XL4"] & AJ["XL5"] + end + subgraph Stacked Transformer Blocks + AK["XL0"] & AK["XL1"] & AK["XL2"] & AK["XL3"] & AK["XL4"] & AK["XL5"] + end + subgraph Stacked Transformer Blocks + AL["XL0"] & AL["XL1"] & AL["XL2"] & AL["XL3"] & AL["XL4"] & AL["XL5"] + end + subgraph Stacked Transformer Blocks + AM["XL0"] & AM["XL1"] & AM["XL2"] & AM["XL3"] & AM["XL4"] & AM["XL5"] + end + subgraph Stacked Transformer Blocks + AN["XL0"] & AN["XL1"] & AN["XL2"] & AN["XL3"] & AN["XL4"] & AN["XL5"] + end + subgraph Stacked Transformer Blocks + AO["XL0"] & AO["XL1"] & AO["XL2"] & AO["XL3"] & AO["XL4"] & AO["XL5"] + end + subgraph Stacked Transformer Blocks + AP["XL0"] & AP["XL1"] & AP["XL2"] & AP["XL3"] & AP["XL4"] & AP["XL5"] + end + subgraph Stacked Transformer Blocks + AQX["XL0"] & AQX["XL1"] & AQX["XL2"] & AQX["XL3"] & AQX["XL4"] & AQX["XL5"] + end + subgraph Stacked Transformer Blocks + ARX[Yx1, x2, ..., x3, ..., x4, ..., x5, ..., x6, ..., x7, ..., x8, ..., x9, ..., x10, ..., x11, ..., x12, ..., x13, ..., x14, ..., x15, ..., x16, ..., x17, ..., x18, ..., x19, ..., x20, ..., x21, ..., x22, ..., x23, ..., x24, ..., x25, ..., x26, ..., x27, ..., x28, ..., x29, ..., x30, ..., x31, ..., x32, ..., x33, ..., x34, ..., x35, ..., x36, ..., x37, ..., x38, ..., x39, ..., x40, ..., x41, ..., x42, ..., x43, ..., x44, ..., x45, ..., x46, ..., x47, ..., x48, ..., x49, ..., x50, ... | +``` +
+ +Figure 8.17 Training a transformer as a language model. + +## 8.8 Dealing with Scale + +Large language models are large. For example the Llama 3.1 405B Instruct model from Meta has 405 billion parameters (it has L=126 layers, model dimensionality d=16,384, and A=128 attention heads) and was trained on 15.6 terabytes of text tokens using a vocabulary of 128K tokens (Llama Team, 2024). So there is a lot of research on understanding how LLMs scale, and especially how to implement them given limited resources. In the next few sections we discuss how to think about scale (the concept of scaling laws), and important techniques for getting language models to work efficiently, such as the KV cache and parameter-efficient fine tuning (PEFT). + +## 8.8.1 Scaling laws + +The performance of large language models has shown to be mainly determined by 3 factors: model size (the number of parameters not counting embeddings), dataset size (the amount of training data), and the amount of compute used for training. That is, we can improve a model by adding parameters (adding more layers or having wider contexts or both), by training on more data, or by training for more iterations. + +The relationships between these factors and performance are known as scaling laws. Roughly speaking, the performance of a large language model (the loss) scales as a power-law with each of these three properties of model training. + +For example, Kaplan et al. (2020) found the following three relationships for loss L as a function of the number of non-embedding parameters N, the dataset size D, and the compute budget C, for models training with limited parameters, dataset, + +scaling laws + +or compute budget, if in each case the other two properties are held constant: + +$$ +L (N) = \left(\frac {N _ {c}}{N}\right) ^ {\alpha_ {N}} \tag {8.49} +$$ + +$$ +L (D) = \left(\frac {D _ {c}}{D}\right) ^ {\alpha_ {D}} \tag {8.50} +$$ + +$$ +L (C) = \left(\frac {C _ {c}}{C}\right) ^ {\alpha_ {C}} \tag {8.51} +$$ + +The number of (non-embedding) parameters N can be roughly computed as follows (ignoring biases, and with d as the input and output dimensionality of the model, $d_{attn}$ as the self-attention layer size, and $d_{ff}$ the size of the feedforward layer): + +$$ +N \approx 2 d n _ {\text { layer }} (2 d _ {\text { attn }} + d _ {\text { ff }}) +$$ + +$$ +\approx 1 2 n _ {\text { layer }} d ^ {2} \tag {8.52} +$$ + +$$ +(\text { assuming } d _ {\mathrm{attn}} = d _ {\mathrm{ff}} / 4 = d) +$$ + +Thus GPT-3, with $n = 96$ layers and dimensionality $d = 12288$ , has $12 \times 96 \times 12288^2 \approx 175$ billion parameters. + +The values of $N_{c}$ , $D_{c}$ , $C_{c}$ , $\alpha_{N}$ , $\alpha_{D}$ , and $\alpha_{C}$ depend on the exact transformer architecture, tokenization, and vocabulary size, so rather than all the precise values, scaling laws focus on the relationship with loss. $^{3}$ + +Scaling laws can be useful in deciding how to train a model to a particular performance, for example by looking at early in the training curve, or performance with smaller amounts of data, to predict what the loss would be if we were to add more data or increase model size. Other aspects of scaling laws can also tell us how much data we need to add when scaling up a model. + +## 8.8.2 KV Cache + +We saw in Fig. 8.11 and in Eq. 8.34 (repeated below) how the attention vector can be very efficiently computed in parallel for training, via two matrix multiplications: + +$$ +\mathbf {A} = \text { softmax } \left(\frac {\mathbf {Q} \mathbf {K} ^ {\intercal}}{\sqrt {d _ {k}}}\right) \mathbf {V} \tag {8.53} +$$ + +Unfortunately we can't do quite the same efficient computation in inference as in training. That's because at inference time, we iteratively generate the next tokens one at a time. For a new token that we have just generated, call it $\mathbf{x}_i$ , we need to compute its query, key, and values by multiplying by $\mathbf{W}^{\mathbf{Q}}$ , $\mathbf{W}^{\mathbf{K}}$ , and $\mathbf{W}^{\mathbf{V}}$ respectively. But it would be a waste of computation time to recompute the key and value vectors for all the prior tokens $\mathbf{x}_{ +text_image + +Q +X +k1 k2 k3 k4 = QK^T +q4 +1 x dk +dk x N +1 x N +V +x +v1 +v2 +v3 +v4 +N x dv = A +a4 +1 x dv +
+ +Figure 8.18 Parts of the attention computation (extracted from Fig. 8.11) showing, in black, the vectors that can be stored in the cache rather than recomputed when computing the attention score for the 4th token. + +## 8.8.3 Parameter Efficient Fine Tuning + +As we mentioned above, it's very common to take a language model and give it more information about a new domain by finetuning it (continuing to train it to predict upcoming words) on some additional data. + +Fine-tuning can be very difficult with very large language models, because there are enormous numbers of parameters to train; each pass of batch gradient descent has to backpropagate through many many huge layers. This makes finetuning huge language models extremely expensive in processing power, in memory, and in time. For this reason, there are alternative methods that allow a model to be finetuned without changing all the parameters. Such methods are called parameter-efficient fine tuning or sometimes PEFT, because we efficiently select a subset of parameters to update when finetuning. For example we freeze some of the parameters (don't change them), and only update some particular subset of parameters. + +Here we describe one such model, called LoRA, for Low-Rank Adaptation. The intuition of LoRA is that transformers have many dense layers which perform matrix multiplication (for example the $W^{Q}$ , $W^{K}$ , $W^{V}$ , $W^{O}$ layers in the attention computation). Instead of updating these layers during finetuning, with LoRA we freeze these layers and instead update a low-rank approximation that has fewer parameters. + +Consider a matrix W of dimensionality $[k \times d]$ that needs to be updated during finetuning via gradient descent. Normally this matrix would get updates $\Delta W$ of dimensionality $[k \times d]$ , for updating the $k \times d$ parameters after gradient descent. In LoRA, we freeze W and update instead a low-rank decomposition of W. We create two matrices A and B, where A has size $[k \times r]$ and B has size $[r \times d]$ , and we choose r to be quite small, $r << \min(d, k)$ . During finetuning we update A and B instead of W. That is, we replace $W + \Delta W$ with $W + AB$ . Fig. 8.19 shows the intuition. For replacing the forward pass h = xW, the new forward pass is instead: + +$$ +\mathbf {h} = \mathbf {x W} + \mathbf {x A B} \tag {8.54} +$$ + +LoRA has a number of advantages. It dramatically reduces hardware requirements, since gradients don't have to be calculated for most parameters. The weight updates can be simply added in to the pretrained weights, since AB is the same size as W). That means it doesn't add any time during inference. And it also means it's possible to build LoRA modules for different domains and just swap them in and out by adding them in or subtracting them from W. + +In its original version LoRA was applied just to the matrices in the attention computation (the $W^{Q}$ , $W^{K}$ , $W^{V}$ , and $W^{O}$ layers). Many variants of LoRA exist. + +![](images/07364a7c79fa2f97e156b1f66e6afd4b83ae8903ad63c3eb1dd5e29c7890a18f.jpg) + +
+flowchart + +```mermaid +graph TD + h["1"] -->|d| A["k A"] + h -->|d| B["x r B"] + A -->|k| W["Pretrained Weights w"] + B -->|d| X["x 1"] + style h fill:#f9f,stroke:#333 + style B fill:#bbf,stroke:#333 + style W fill:#cfc,stroke:#333 + style X fill:#fcc,stroke:#333 +``` +
+ +Figure 8.19 The intuition of LoRA. We freeze W to its pretrained values, and instead fine-tune by training a pair of matrices A and B, updating those instead of W, and just sum W and the updated AB. + +## 8.9 Interpreting the Transformer + +interpretability + +How does a transformer-based language model manage to do so well at language tasks? The subfield of interpretability, sometimes called mechanistic interpretability, focuses on ways to understand mechanistically what is going on inside the transformer. In the next two subsections we discuss two well-studied aspects of transformer interpretability. + +## 8.9.1 In-Context Learning and Induction Heads + +As a way of getting a model to do what we want, we can think of prompting as being fundamentally different than pretraining. Learning via pretraining means updating the model's parameters by using gradient descent according to some loss function. But prompting with demonstrations can teach a model to do a new task. The model is learning something about the task from those demonstrations as it processes the prompt. + +Even without demonstrations, we can think of the process of prompting as a kind of learning. For example, the further a model gets in a prompt, the better it tends to get at predicting the upcoming tokens. The information in the context is helping give the model more predictive power. + +The term in-context learning was first proposed by Brown et al. (2020) in their introduction of the GPT3 system, to refer to either of these kinds of learning that language models do from their prompts. In-context learning means language models learning to do new tasks, better predict tokens, or generally reduce their loss during the forward-pass at inference-time, without any gradient-based updates to the model's parameters. + +in-context learning + +induction heads + +How does in-context learning work? While we don't know for sure, there are some intriguing ideas. One hypothesis is based on the idea of induction heads (Elhage et al., 2021; Olsson et al., 2022). Induction heads are the name for a circuit, which is a kind of abstract component of a network. The induction head circuit is part of the attention computation in transformers, discovered by looking at mini language models with only 1-2 attention heads. + +The function of the induction head is to predict repeated sequences. For example if it sees the pattern AB...A in an input sequence, it predicts that B will follow, instantiating the pattern completion rule AB...A→B. It does this by having a prefix matching component of the attention computation that, when looking at the current token A, searches back over the context to find a prior instance of A. If it finds one, the induction head has a copying mechanism that “copies” the token B that followed the earlier A, by increasing the probability the B will occur next. Fig. 8.20 shows an example. + +![](images/ce91af053069b524e9a76621eef719d2c6092bb7f122cba2d2cb05941d7ef088.jpg) + +
+flowchart + +```mermaid +graph TD + A["She"] --> B["owns"] + B --> C["vintage"] + C --> D["cars"] + D --> E["."] + E --> F["He"] + F --> G["dreams"] + G --> H["of"] + H --> I["owning"] + I --> J["vintage"] + J --> K["cars"] + K --> L["."] + L --> M["Copying"] + style A fill:#f9f,stroke:#333 + style B fill:#f9f,stroke:#333 + style C fill:#f9f,stroke:#333 + style D fill:#ccf,stroke:#333 + style E fill:#ccf,stroke:#333 + style F fill:#ccf,stroke:#333 + style G fill:#ccf,stroke:#333 + style H fill:#ccf,stroke:#333 + style I fill:#ccf,stroke:#333 + style J fill:#ccf,stroke:#333 + style K fill:#ccf,stroke:#333 + note1["Prefix matching"] --> C +``` +
+ +Figure 8.20 An induction head looking at vintage uses the prefix matching mechanism to find a prior instance of vintage, and the copying mechanism to predict that cars will occur again. Figure from Crosbie and Shutova (2022). + +ablating + +Olsson et al. (2022) propose that a generalized fuzzy version of this pattern completion rule, implementing a rule like $A^{*}B^{*}\ldots A\rightarrow B$ , where $A^{*}\approx A$ and $B^{*}\approx B$ (by $\approx$ we mean they are semantically similar in some way), might be responsible for in-context learning. Suggestive evidence for their hypothesis comes from Crosbie and Shutova (2022), who show that ablating induction heads causes in-context learning performance to decrease. Ablation is originally a medical term meaning the removal of something. We use it in NLP interpretability studies as a tool for testing causal effects; if we knock out a hypothesized cause, we would expect the effect to disappear. Crosbie and Shutova (2022) ablate induction heads by first finding attention heads that perform as induction heads on random input sequences, and then zeroing out the output of these heads by setting certain terms of the output matrix $W^{0}$ to zero. Indeed they find that ablated models are much worse at in-context learning: they have much worse performance at learning from demonstrations in the prompts. + +## 8.9.2 Logit Lens + +logit lens + +Another useful interpretability tool, the logit lens (Nostalgebraist, 2020), offers a way to visualize what the internal layers of the transformer might be representing. + +The idea is that we take any vector from any layer of the transformer and, pretending that it is the prefinal embedding, simply multiply it by the unembedding layer to get logits, and compute a softmax to see the distribution over words that that vector might be representing. This can be a useful window into the internal representations of the model. Since the network wasn't trained to make the internal representations function in this way, the logit lens doesn't always work perfectly, but this can still be a useful trick to help us visualize the internal layers of a transformer. + +## 8.10 Summary + +This chapter has introduced the transformer and its components for the language modeling task introduced in the previous chapter. Here's a summary of the main points that we covered: + +- Transformers are non-recurrent networks based on multi-head attention, a kind of self-attention. A multi-head attention computation takes an input vector $\mathbf{x}_i$ and maps it to an output $\mathbf{a}_i$ by adding in vectors from prior tokens, weighted by how relevant they are for the processing of the current word. +- A transformer block consists of a residual stream in which the input from the prior layer is passed up to the next layer, with the output of different components added to it. These components include a multi-head attention layer followed by a feedforward layer, each preceded by layer normalizations. Transformer blocks are stacked to make deeper and more powerful networks. +- The input to a transformer is computed by adding an embedding (computed with an embedding matrix) to a positional encoding that represents the sequential position of the token in the window. +- Language models can be built out of stacks of transformer blocks, with a language model head at the top, which applies an unembedding matrix to the output H of the top layer to generate the logits, which are then passed through a softmax to generate word probabilities. +- Transformer-based language models have a wide context window (200K tokens or even more for very large models with special mechanisms) allowing them to draw on enormous amounts of context to predict upcoming words. +- There are various computational tricks for making large language models more efficient, such as the KV cache and parameter-efficient finetuning. + +## Historical Notes + +The transformer (Vaswani et al., 2017) was developed drawing on two lines of prior research: self-attention and memory networks. + +Encoder-decoder attention, the idea of using a soft weighting over the encodings of input words to inform a generative decoder (see Chapter 12) was developed by Graves (2013) in the context of handwriting generation, and Bahdanau et al. (2015) for MT. This idea was extended to self-attention by dropping the need for separate encoding and decoding sequences and instead seeing attention as a way of weighting the tokens in collecting information passed from lower layers to higher layers (Ling et al., 2015; Cheng et al., 2016; Liu et al., 2016). + +Other aspects of the transformer, including the terminology of key, query, and value, came from memory networks, a mechanism for adding an external read-write memory to networks, by using an embedding of a query to match keys representing content in an associative memory (Sukhbaatar et al., 2015; Weston et al., 2015; Graves et al., 2014). + +MORE HISTORY TBD IN NEXT DRAFT. + +Ba, J. L., J. R. Kiros, and G. E. Hinton. 2016. Layer normalization. NeurIPS workshop. +Bahdanau, D., K. H. Cho, and Y. Bengio. 2015. Neural machine translation by jointly learning to align and translate. ICLR 2015. +Brown, T., B. Mann, N. Ryder, M. Subbiah, J. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert-Voss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. M. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. Litwin, S. Gray, B. Chess, J. Clark, C. Berner, S. McCandlish, A. Radford, I. Sutskever, and D. Amodei. 2020. Language models are few-shot learners. NeurIPS, volume 33. +Cheng, J., L. Dong, and M. Lapata. 2016. Long short-term memory-networks for machine reading. EMNLP. +Crosbie, J. and E. Shutova. 2022. Induction heads as an essential mechanism for pattern matching in in-context learning. ArXiv preprint. +Elhage, N., N. Nanda, C. Olsson, T. Henighan, N. Joseph, B. Mann, A. Askell, Y. Bai, A. Chen, T. Conerly, N. Das-Sarma, D. Drain, D. Ganguli, Z. Hatfield-Dodds, D. Hernandez, A. Jones, J. Kernion, L. Lovitt, K. Ndousse, D. Amodei, T. Brown, J. Clark, J. Kaplan, S. McCandlish, and C. Olah. 2021. A mathematical framework for transformer circuits. White paper. +Graves, A. 2013. Generating sequences with recurrent neural networks. ArXiv. +Graves, A., G. Wayne, and I. Danihelka. 2014. Neural Turing machines. ArXiv. +Holtzman, A., J. Buys, L. Du, M. Forbes, and Y. Choi. 2020. The curious case of neural text degeneration. ICLR. +Kaplan, J., S. McCandlish, T. Henighan, T. B. Brown, B. Chess, R. Child, S. Gray, A. Radford, J. Wu, and D. Amodei. 2020. Scaling laws for neural language models. ArXiv preprint. +Ling, W., C. Dyer, A. W. Black, I. Trancoso, R. Fernandez, S. Amir, L. Marujo, and T. Luís. 2015. Finding function in form: Compositional character models for open vocabulary word representation. EMNLP. +Liu, Y., C. Sun, L. Lin, and X. Wang. 2016. Learning natural language inference using bidirectional LSTM model and inner-attention. ArXiv. +Llama Team. 2024. The llama 3 herd of models. +Nostalgebraist. 2020. Interpreting gpt: the logit lens. White paper. +Olsson, C., N. Elhage, N. Nanda, N. Joseph, N. DasSarma, T. Henighan, B. Mann, A. Askell, Y. Bai, A. Chen, et al. 2022. In-context learning and induction heads. ArXiv preprint. +Sukhbaatar, S., A. Szlam, J. Weston, and R. Fergus. 2015. End-to-end memory networks. NeurIPS. +Uszkoreit, J. 2017. Transformer: A novel neural network architecture for language understanding. Google Research blog post, Thursday August 31, 2017. +Vaswani, A., N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin. 2017. Attention is all you need. NeurIPS. +Weston, J., S. Chopra, and A. Bordes. 2015. Memory networks. ICLR 2015. \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-1-Transformer Encoder Architecture Diagram.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-1-Transformer Encoder Architecture Diagram.jpg new file mode 100644 index 0000000000000000000000000000000000000000..88527d57a4eab521a1a920a6ee697149d9cf384c GIT binary patch literal 33126 zcmdSA2UJvDvnG5Hp+Pc|vx0&kK_#aakt~vP5)ny~b8Lb{2?7F=b50VPC^@U-oTKE> zO>7CxocEi1XI`B<^UZhXu79n+S%C<&~?NIeRRkg3Du9pB}d1*Om00RR6Fu*_H zdKQobaImo<*jP9a2m}`w=N3L00X`lcJ~ioWA~Jd!Mh1EsIyxp6E>@pF#hoY{_TTt1M?;pHUtOv79RKjoEW%)fr)wJCMFiv&70uU-r)DZ zO%g29JC7u=Z>tzX?%I<*_KVKMVSZN8My@)HV0rS|!5{Y)1tk?V&At1qY!BFZ`S=9{ zg@h%aOG(Sf%E_y#YiMd|>*$)iF@0-hZei)@P z02DYoXUp=v_1jgH(2$p&_g>&^Vj^?k_F$r#{b}F2FXymIatP(Od0vx#!KjDgj+0Tk zK}RapUxF-H>q7BW5_NQ1MZDe5R(v^Cw3CR!oaS96s>ZX^*q9rXpF7AmevD4dPpjJ? z&o(QZvLyK$uQf5r7ECzwt&HsAqgw`^y88Y7DLuScd&-Y?&l|f((|zKiO;$>w%1H6d zjLxmo54;f{+{2zH`?0VMI6UeHY=%`79kITsi6T3xmc4rLVka^A$kXzT3#1QqjP&$# z%>E`?Lw&5|o_F-Ztcbgh3xx)&*;&KYqR*%Jq}-sa7_q<;_)t_3&x`Wvb245&r_FbkKSina!pvbh6kKkY0j!wjxog>h#$`E%6xqBgu~Yx7ZL{4 zOF7h}aeR%{7~|M-gX*Iv*wOFG zOKo(SLvKFe_#}=JRmN&y|3uTA=;(H8XUQKkf8SM-ZhN3^Ck}EqL1V1+*!uo~YNPcR z-OiDEdOK`8vCSJl$)n9mAL?z<|V1gpX;gD=&@#Y zLz=+J*Wl(e5P4}nagLrC>ubkdSD!!bTd#sciTQ#D*R(^40(~T_J!^_$!0p z8#`HJO^nWKg^(m!*hGe%JnXB8449R>1D2!(%~coNIzESAu75#N&;Z z@T68CYkFCY3Nw$AimKrFfM|0@d~tW4ePAf$GUJt>^6r3aFa48#z^!)LOHiEN4Z}wD z8;6iH6|5THSpCVnS6=Nr?eF~e8h%}7dC`6XymZ&nV2}erc(~l|=ry zD8(OhmC3>LEQ648rNN?-2=D%vUf$CQ=GzB?292~qrzfXH4XHIz`Il0o%!p&nFQPNn zi>L2hN|8!Ivz-FM5AI`P#ckH52_@J?^FNO3#({b^8ID{EWNwB& zw(yXOOX^2gOWepqZrTYkjA5*P+|+%WjwtS+9BD1a&UAj)&^3LrXQO=?KL6Yu=h3L2 zfx%VThkP#?L_6v=halJ6Nav==PWEEGo4s6=QBC2u4c|9dhixrRe>b&x=;980OTrdq zd*-rAby`wRXwghnnXDYuaqb;h{qd#send5Uu$zW;j^6<0E#U9~Z)@RJ@-lX=qc=m^ zq$L+Y=ZN*u>Zd2~+Mqm8S3qM&ZlK&dTW?`2$$FyJHdFH9(T*eIeLmQJPxZcIq!QIh zxs*5_qETPc&Py!!*U&Xk46D4am4fUk=QB}93)O*73zWl4W9wDcoL&a(a%)~+Zr-U=Nn>U?KjxcXm^D9#b@tzd5=mKM_UF&T%w^sb z75v*O+7D_}Sym##yo2r-ZSiDHTAICskK$;g+VPiJu{hTX=2avxHQdrMYxm!~xs@msBN^J6@lY@fMRdflN zbo$vTf`4k3iuXb`qVn-mRdS;TdIkDKpBR?Z)y!%5x>J&zW~AQ7v(3g&?|tkpjd36g z$Bcmq1BbS^#ePsQ5;wPgr+a(ecaQmw=8X_W5qG@WGLg4A=i zcf*YMJz)>`mnCizQr(_&F4vDz{1Uv!pb=N;oq-`f^~t@CR(fA(HPTZ*U&`VC>4^7-bhMrk)MVmf`fLQx%X%dAd3EHBJYv2cfW}BGTB+4mged0 z2rDimJ2ht1hi1X_Nya>fAP~;xSQye#Ik;10=0~)}P%0NkYbis!QqnhC|LJbemHJKI z@p{BH5VnRu5tw9c?|A39o3SzQ5wOdk`+fyM=q7XWgE;xbZgqd}oZzR)IqfCxW4Y7D z9MNId%YuCs8SG7(0&D%n)%Y{ry`Sq83r6iJUid$)MX48HOZ+;(JqZl+e!%; zgCqutO+`W*)pw5znd%U^^G!YPA3m)Q(NoA{Y4&c9KDyH?;z!+QgRw2;NuL^+mTq1X zulZ3=kISV4-%;Ruk~k13S_a zLVOR(Y_#^c8dHhoJ6^}0X+;K1q^|ahNGJ4&6Q-+FesPXpar+sw&T?n$xqLANov%A^ zR(3uyFwu>Hm0Hb@@Yj~PbyHccPms<*j*lWU*J7rcgD}_F%9E7;MB%gM)6#8B%znUq zAfl$xuQ=vw4SpZ0yYtMxJ!FGt<7qibPT-rW3+ zHqMGb;~l>gAM)Mv@*rM%2lMrDCr{dGlVye#EC)!Q?AD$Zw5r7GyM=g8R!Eepe8H7WwKqTomI^!$)$(%+bzttP65!62KOmgbpwa~T!Ir{VlYM5do z#=c*bGo7l|<$BuDG4r!gUhyigfCjJFq2O`!_W-|0iB^W7CjBj8i&3qY!rp}y$~R`I zw7vm!;arc}HgZ+RyzVW}8~(C&rYCyY*g$o)JF)%4ev~B=Mgy!GHwUXAp&gC$;?%}r zp0oi!<-1ai4-+s#*N!5UbdMX_YlpPmYOF1nv!=<5W^^Q}t^xJLxwVTJV6`%AKI29h z+rNC>qFE>=49;A%jP*Bx2j;R1S6~J*tSu3hy2-R&*=pAyC*T$(GVBDZ7NVnf)8kKY zU6fAc)b7iuReNX+Idsr%Q)rKm@w}pX^M)<}MNBlN_1WuTkkD7DRpdj}%$%Tk~51-8aiI4ypsYJ*(e`y)cpq}$ zm0tsrKhx28udjg>wm;e&#lz0YiZHogl}hfl@9A-qR(Bv++Ad7tntn_-lq6(88FZ|t zwq3Au#)qIC)T6uxNOU}u{Ly#JU4;X?Q&%|zO~-6#*0-A5Bi3c*?lqeVZo5&Dk4>Q# zx=;IU(?$*#zUk${Ytq!IQWL}cGf1EI9)xWbtiRe|u}hzBA-@JTy{fK(ws%ICQjCAJ zU7b|dU!TSZD(xY?xGPv!iM7ZTOm2`k~R+2iJfhCFA+cxE*-b)C|}5vJ{yb8(UTw z@LzFp#G_)b0hZBD*vT?CI>12DVUZNJmgE+N{(A2k7&6DCY;8GY+GIRgCPMpiq>mgY zCw8spF13)r;nkC>yfp&xS=ybg%RO&jT4f-(FGGYOS8kHmK*lZ81HWpHovm_wN{XWa zMEW-kF9!jD*qHrRg;KGSSwS*{Etc1+@m9x2y7Y4FYDTHb z5m8eLzt?6ohWYv5&iG;GA49&8z{mFI>IVcxC$+c(d4;NZhxJ~DXy2KOe!#BLu#ZnD z>AAscj;0uKdT>8!0GFV1n?i0Q@n`fWC)&@kQN~g|Kcd(JI&{@8#SrtuLAJu)s%&de ziv+9}xi7JI34t{&Eym!7&M^8NVehT_nD&uxwDnU(!-;niPG>K5`Q&P_Fa^6I`F!;yGc#7_r-c#17dr9dr2LESr&gZQ z({t~Zb#ZbHOb`Fjo-9(4Z+qizbjc1Qxdw*iC!hzD*MOP!Icor)n9q)8Z6zwXKSN`P((HtcHq0hiO(^ z13oPm!qDHFC-R^yI@b5iG*ZiZ?kDpyx=j-&MjU-%o@F;>Jp!pkPY&kJllC{WKDYlD zOb(9C0cQ#4n})A-VG9e765`Epip1nZZ*9PqO;FtEpa|A$;6}owWqIc{0GUJ0wjJb6 z>z4?qyr_M|an+|IK}$Qef7x0_*=C0lF^o5$)j~uj=Aj}7f7VyVOiS;=Xv_)@7Lio< z1YTz|n3)Vi|4k|fC{^py{2B<1T536`;Vr+!HA0pgptR7$*FdI~!Zpxp3Qj6?*$5?x z4lL!q1~3vXA2`x@QqF;s{r4YbqbZ>NpbIb`X+hkBbyrw>(J$#urgBzu(!8zx*u|sZ zBtqw`)`J}a-q$(cdbl!FZB}19yYakcQ&+5mnVIfw+p=MtcuJ>xzIz3w^U3HJNtFkBkaBo_Fm&T(pX(K+f-| z_S=jM6y1B)A7rF5sJs0dVe6OQm^Q%)k@5WC-Rxv#6BujP+uK7;Zh{8A;(O?(?g;2` zsbTH^P%-;oQBlKl5F16)PQtekkljY@{B%Sw1MBk=VXBJ=n|odfP@aty?M3y@L3sZ&Y2D zK8n{LuT-jaHuKnGaerVL+l}M=NzxRi>s7_x_p^$Um0hl{bxF%L-_6SceRqsc#bx9} z-JRg?&V5D?Nn)mdg%0L|=B}OS8XzNEphSb)NqhGius9TcYxHD)|5A8};_W?nStATd z+qspqZMXGh7jE&^%Kk}}yBFI{UxwDW$w6lX2>oV{4&4M6eJwN;Td>8|V-s2wwX+s(4=wbp7})Mb>(A1f z)>enKk=}4@!Af>S@F>X-G$NoOa0L+&0-3Hy3TCFFs@pMOF~86qJ=K{? z69IvPTde;0sN6Vb^>@z4u*w2Yl_&c0z(egA{&FV!e#bQGmd5H~Czdo?6S2WiuDdBz z9q;euuw;q`M7QmP?Qn^56wg=*oGma+tWjH5 zx8N@?=H1KRwk@F6WiXv@{0S9+tnZvFi->$i?uhfW4DprP&>N1-aPqJGd>ycBe0dTY zML7M-jcPURB+q%zw$r04FClwDGS@F(3g&D#gq!@>drX4`-?@*k*kT!4i89d~_L*(G zc&3m5GJkg;3szdkfhXHf_VgEO&4_}Xtsg8;a8-ZktzN74mMHPM(0l&^2!C=|w=J~k zk%Uzw;6+`uBQQC05(($Mt(b4{5CnRyz`PO-4<~44qM+H;O18Q_Bbm8TQP^A=%t<5Bb|?Jg_ovdEkB!r{YrDz z4r44I<{#^$jf@&}mo~oBKW)xOMyYvnK+-zeG)G>qY!Xx&eStbVUP8YYZU7$W#5^7I znRhfUsn+!H9K1@wkFPMUlnM`&Dk|Wk?~E%e{Y5;@iCOHWd5R#$n(PA>aD!0K|!%@GTarn-2FpLBLefwyW?*T8wClr?Aq)hIiR=%${>PcV|+zj#bg z!1_hY6f>jmwIUQv%ox>kQ$VQ@gYp+sx{66yJ!4wPPYF(P4rhpR|NUFd}>}rLKXj zNGL)#ebq2uiUY#imYho=x~#T-t_=_u z|H~cA{&N`q67LL&CBE(urWHnpQJ^W8aF6&VNiSd*Ti~<9dQR}5qDXwbh>97p$fQU zzhguPmTr8^esza=qK+u&$xU%KRiewFeCW60Pm!)Bv(1y%*8r3Rw)pykXQOAQYfajN zIJ)j`0etdcW&1g;`z2Iw=tjH*u7Qt+Afz-dZn?rywd@Elb8%9Uk~98iiKze7_v9Ok zG|s<{iGn6eEq4bt9ZU!Hwh`-RkY# z>0=gL;#n_5NfQxEt)!DeL)YN3>YsIc*Jt;!rrKmAhHmX$QZ9cm!a?bNm#>Y`eExKj z5b{`F=0^{5YZD+3hpxepL|1lIXjQeywZUaRjzKiF-)Rfw9eTdPmQ-g-tx4U83k#Y@ zH}u{CkVW6*xTCP^sk-SXRg2!v57|a#re5k%`vkrT= zyq+`R?%;=x9l}wc^ANYvm2u}$jA~CAN~KhXPsqlP6$lh5Ee-W#?26O;#`vMSjsLol zaOd6wS-@Swba3~Xw;b!o?P#pERR0wKgIxt?g7(-!8IrG+W=Ue8xUxttW%9Vq<*w*#iV25<>1RtL?8{F#Do64C%FIO^v=-FBt<~m&2zWGC)b+w{GG5rZ-3od z!aW^@ebL}eZwxA`1fw-|% zVJlJGBWg`Zi48QADBmVKTAfz4as2&+i|)H5kw!2)JMmMUBcYX+`IWeXG}L`swx|Eh zeM4GaGqP52j{bOC{IVu@C6@pF7JmHp&^BeDs}u+Dtf!XQ^8r(Q@9ej=HNUln{P)dF z+x80zH;*iJRV^uRn3rl}G`;!txm9*PwinYi*E2O0-jtf6IT8I(SvpgN>=lN>65}sy z%k)!9um{6IwaRN@GE(=f3AGf5#qZB;)f{T0W)&5zeWlZ<&69yu3pFZsnHMJE8s#V% zACQ?19p%H8nxW{sOS=Oc`>N?Q}Qjo zLAQ;Lnguq5Pj=oQO%?A$#3Q<#B3wDR#8XYlEhy=$QA)-FdB3XD^+bQpmldE2Yc%Gd{fEo>ZS~2Y7%3~ znxPv;i7E`g6=hPyq4yY{su{)PNz%rU9%z0hZ8;(Km~WN}QY!P?!pbY~!n_8U`qj|w zXRvdgv0kS$;f7@&HdK_x(6KlFq*NSJ%eZZJr9!asg6zAWzGCM-@OiC7FK}FqmIg0~ zBrdl#*P~t+AY{TT20Kd?#3w)-{s9CqQDDw!SbnSpK?OrPkD;=4Y>C_~FEnLu?pZS+ zY59ld{3ekl7o0NPl%#rJXZT~{N65MQe166$sX`L^dfrVcvG7;47L8k;U54ADo{5@( zTrV zr>pI|61zN$8#lbM6`^Jro*tV(K--^A; zTz3mn=Ant7##OE8Bf=81puk_@S>+OYqlF`Ic587Wp~|18RAcuC1wtU4TE4b3Ro0(J zW*S@!lp4G(L56pS-;wqxqWrwAEZ@h4i71niI>4eP8q)FHl9Y(e%Zm*ptjsK5Zr04> z;wb3|Vkp6_^(-m0KJuQ{#virU&$Fl&vz>pmf0mC;2-nCqBl$yZ=rRs63Jk+4fRVVr zUTILdtx5E_y|$*G!&l>Gk&fTy(?O3E#oI52N(xM$;xQS=LGq-aW`ymhCn717mvN#?1M`Wv356=2;qZsw)oK2da*Cf*yG{h=bD>%$PFMa%;byTR^ zGBUQDKuvFaSC#j$-{(3 zEYTl3i9L{Q`<_ng4UA#1!xwxYvp5A~&Vh~r=$8~d=vKvRAe*59`ulhNx{AyV$u_uE zuC0rQN|zuWd{u08MFwxtX21rc@37^T3oKnQYMEPl9`i3pala;<23`Zv5T$G2OEvVI zOa7$j&}fDd{hS_(ZXW`r*0W)b76&mNDvL^jI0#I-eTti##D}hBxCNu@K=$(BL63tO zBEj+JfZXQ%P7umR0%;q8fUVz!o_ZWiOXBnNyD$3PX5c{9t3@<~fXeR3)W=L6>jt+% z5AnyLyWFmz&IEF_wofgg0BC}qsDnLeT11EQJ!Vha&@wbd!meNe&|NXeQMqFIBUJ0N z{-p_4!0OAJv~dncxp&{wN$XLrrLaj1FkW}dCu+HHcVglIO)6%y@yyh+p+k{LE zjGk2k`u@_Uy~f#j2{ME%4^rMQF1SvVCR&|6Q2BX%PP>3#w?iJzG3j7C6TIQbfLNmiK61v{j6hl!iP@ z48u~ZKB30UY|a11ZSL-!o0d-W-V2X>j#x6sV1QOx=+T-;0;#e`i7XF-m}>Q{ zSujQl`ih$Z_ZGvM-oBYYk6J)7*?|B?)zV*+<@ZXzYGzFc-c>MFdibp1239x!YTl!n z40P|P&v&fyQ}6ugy(!U0?i`i(>2HO9so zo~|KP^K#lfhn5#3lFNHE3y!?P8Zoj4Z|1(RiF`8F%fE@)mf32VCA$i%bW(tY;^rIa z=tvn8CK0Auvsm7p)BhEyj?l|*Qe2{{?)6ex**u1E?^-$HsF3_3+ZFL&4L1|x_%e2E z{X{p=cx;!moLMqXXO*b>!Yo(3M}G9Ub)~&~W=s&{Jqf!_YwG~SSSdkOf4{`Of58wQk>U9jYiQ>xDT%^$$)&CWDh6((Abs$x z%r$UZ=JG+Z?K$nG3YbpFz6L_e6|+*ON}~SC9XtW~yCf_Oir|GI!N^Tw4QPT|x)@P3 zP*l?K3|k`}^N*J07qhUT#g(W_0>(>4Mns-<%cd<-28|tszTdLdGBd>jNcYU~LkC!C{CcSeMw!-Q)a!3e|! z>&S;Ml>*Rz-ZO5L8vo+4018wa^M5lGpd``nXtMs=D*`l#&Rao<(+OU;-p-UDvnpB+Akx0KyqFlg(B7emjBJGfrR%L9#M18nkr(t`s zmNH43ZKdndvP$4{p-``vtQj`0siBI&Bun@h+sbGK;=6b0Xklv0JJN3anRF9{fr}OS z=!l;ZQ@nJ1mU#HNhO3k1<+0H4wY9+;Wl z?2x1E=is@hZvK6byj2$vZeS~-nXUF9%-ZZSur_6HUGgV8QnqxY3GN!i+)5akp0f&h z8yeOUcQb~KxnQnH;*vlUT>X;S%hn1h}7Cg<4QyJ1v#f2fq*?-)Z4Mlfg_a7LS&}1>mD~$?j9rNc3q9w(N|; zLpYbIv;(-`vmeY;(9J0q7s+63EXatz8Ht|`yUG3dmFCz4TaT5R}?)o zS@HKiaIor+iSjMCna`B)*X+3R_}}MWir5(QYdzzgU_WOI42S)nwObrFdNy2|Ro7Hq zBb%GZ^sHvNa^Ei7Z%6AI_`BN8wQF?wpuxi{Z)H5DwY|hlz6L_F(UeBM&`thD6a>VD zBnGahnyJ@-I7Q3ZPgr}@Zm9^`ixYP6?XNtioGEhVjh5!rEUN`l=_5D8wO6|?? zalMK8+Gr^~74f&B2RA*O{xM$hyS96LB2>2TY?$ldiLNpat>cXKTmHUvE5xyHgC^kD zYTq>d<+g8n!(&3ox|QwU+>pWhc8sZP>A(nqt$o@|muoq89KC#U4cz1ChUqC? zQCOryWTCg5UNDTpHaChVeXzs4T`siXlkie;&-pY5T?Nw286H9NzO4hpH%S{Yu!-Dg z)XZ*fccST4h3A{9s;b$oFIDsUfFRCO5p@Q+E2~eHV_whB1;XxBHH~}C)2z^b&=})S z4vw-pi$IPM6FwQd+faWAS#QeyhKkaqi72$EKa#sGSAqP*{C4E)8~c=Dv*jr-nwh)` z{F&~eNlUO!SmbSt7M(2}p`C{r;F2l-KpQ{Rq?_{b++dM~YL&&gVWe$tgwJl}H>gYL zF2ISJ;7>1d3!|=Xyii9v{~W9AIh{J-RM!=?J`2M*dz9iYCRA3{xVB`T#$~>OHZUapR1Nbef5r`)*4>&pU2nXc_dmUnYl~y5tDzU z?p;81z$`f(`lR)Uz)ZQJYB&RtSA56wN!N=7T3=_*6 z%~f9;0wD*_1%I_}($b6EkSld_npaXm#usMRhQi(6J$zUi_BmIDH&W88>-6Hy5`K!7 zobgfJ(fG0HG|f=1l;2^pmgD8c~X>6?}Jx zEjO{sdQ*%@zgRH9N?YPyU~8O2ETLJ$z+&>(V_cIR2shpKo;}^Bij?MD{@&F)iSlL+ zYyPE+qmkyA6Z`vnzt?hw{Z-2qr(+<=fvrh>lg^)`*Ljg_hbFlFfUIX_k)VfQJ|t9! zBzgCYeZNtMiVoo}XfnH{1-GT&DK6Mbqe5s#Sh*I$%zB)K>K*RZh8MJ7Ee|rc*iZP1 zu^G!-{Pf+`v&jXi#(Xs{pf{O~mK9 zY>o#I)}PGugA$COLMSyXJnus9e?h>#oI}7+<8hrFX8HGl|ReZzxRWG z=)QmguOHc)r8jB9lPNoBA!NX|riW|bF^3%WE^|{|W6SF=8OCQ4bJU}LFqnL@j>WdQk|)gsGVn8s(S{J?j*OU40d78>}HYyygz3YmOrom zl1ze_Swu_dEmI1~N1QdsuGx7ag)wJ5k&9V8&kv1v6{{;VOh^QOIu0R(_QKK?LCRu~ ze`T+9byFkhALYIre>#bB-ST@^nP92)CtBU^qK|wEz`I*ORKBIoRhY?Es4E&+;SJ0>UMWFe%oK$}r_#P?!ej6lD z`7NM337n1b*-E1_u7)TDtxWPTOWCDu-L-h`2t?sdx>`jsU(dfHw znJN+wyOF+t>y9>>9HED*eHyG)feKv7#1G$sm(#Lqglwe0f{_?(blxe{M@jFz7$Huc zf9m}iDi!t^>bC0%_yu{(V#a{EN8o<0Nw$0fj^2wgJS8ip&6wzqROn$({rxY$Qzh~#Ek^L2azWsj8ZKV&*$P?)n%fM zral!JG45>j{!UtUF5vn}>(QzaRnIabPHo6txWO}?;rCL>uk6+U$yetaxlR?$sBvUO zE1fcCL7B-*WQy=YdU=12#hDhR@bkl2T6g};GN4lNX%1Q8G%P?39@M>r#u0(VuB#Ad7jx;8mvEtA@-ewQm zABTqyBDp@xhv@Oxix>sCBS; z_$i}+MKdE3tH;;N$~JE^sjDW`;bW?ZQX~EmCFz6LAAIe7J&0b`mF4~78w_4}Qj2`9 zORx?eM5R%usI{LdwBTHmQW?VR`6gD+w+B`y2b&{MAG!_sked0<%NPCZ1PUFM)6*Xc zzMZg>Jz{c@Wo1ZDeQmIlcE1^=?1cyMhX*#*_`R-6&9wF-=`eL$OTmcx&Ub7|hH+FI zcAVVnrDM$+dMQx4lO#B5y;@OQ!9^-`gU!_XYe1vxGoqL9XC2QbA^xJSS*(F@JdqED zoClnxnyhxrM;KE%hd&^zZ3jj#RGyOWlE=P;YpRxf$zTd6wqx)5w zd{p!WI*yG{3SPE7iqB2{is!8D2lipITlMmakzNCONY~Jx6-dtYn3x6LieC?+9g;uW zLYSFI!owr2WQO|Rd5TH?mI10WLOt~YqY_dO4UiYYu2N-|l`v3Y=ih$qU3wSkKJ<8x zeB8qzS=o%2o?N)2b~ghqmKO#Z9?bW zg~LvZ|C$SMJN!hvbrsf@#>hZ{@JYphVLJ(fw2 zPl-p|<%~Z*ZIls7pqvkqcG!Tww#&a?WzfH~uR!}h^Nzs$pX|#HEN17uFwk2$2;jY# zdS6q^KN&3_-0-=VZ_e$HFidIzO0`EFsTsRM{j*-LiSv!N3iWgAudnuRlEQAVVYZ8! zc)KsR+_VJC9r!|N;3isl+uq>Y zcmKyibRJsJRombO)+i=;l*Yw#Ro!~?yc$z2;6%SNI-qT9rs0!vNP#e0Yd_!x9zww1 zqA;XJ#ldXNRkynpb~xNcCq?ey*wwK@wjI2;YJ@p!Lq#PJiwa{jZ>y4@-gg^$mS$P- zc1BjHjp=QZICqe+I%U$@&~{R%os|MlLzQ0Op@2NkK#|sg2dSdoXfvaQ@Y}*nOX4}h z>ITl}%38urjgS71QHDxlU%<25{l|>KY4nY#{1q7CptZ=R3yFj%jeGsa=asS_;+HFx zlplLe@}K7w=3ELlBGAd4qk3{HiB6rk{FB757$4|g&G)p_pYR>Hh*MOIUg{LNMLw{h zIZ}-jT$eF@RPYp-c@#aG605z3=;%1yq$!JfV({IzU{UO%u{TTh?K?;6TaUu+Q%;1y zBZy_mnL9K@AC5WpFxv!tXhQ~=nvR^+lSD$2fuB0{OmC3CG&OnxZ_9pZX+XLL^NC5z zq^=v0TTm`|77R%4&Ie-B#a?lYSGNQa&+bqp#lu$!^b9;n6;rfE0_dKDF$TuY1u)>K z0t=|=BRls_2;bItxC&DAw!w>aXYYrv9UG zF0IUug^iXba{k18-7Ok}vQ$&X1S@NrE>bf=WgejeDL}UbN{LFFZ9`KvPVBG5nD4dT zU^{Po1mK^pos@oCzjQxMM@iat)dW-2bV*;}KDij{oey`=CX^QHT}=YoM=!lsq{p04 ze72bS)3e*`KK+ii9tj+cHtOQzzc5x~cXqAf5wHj^HKGtNYx}~@$;UIgPX>8aw=H*G zsR|gAMmkW1n@U4+-)ydaN4^Y8CweHVkOk5Mn)<$ks1*jxV>Hr@LP}eY)AiS?o_pZKoq|~EE^0yCj=!j)s`@jeE%+n|*Y@K%KX37DK@Ek@3Yu!C$HQ^UZR5hxH8N37c|(@DceycB(f(!C2Rik(G}Jx zuCjM!cWN8z{W9WZ6IGSTE7hZedf=tFNmDg0T^i6`8xKiNS^e^@WN} z>7mW@#IwYt0!*rRnALi zQyaX0<#!fatJ3qL6tC^e?6%>Zw9^ZHcx~8`mm`0+s1?g=zI)TP1gk={ksc;+2lY=W z$?*|%k8RGkOl;-G?z`uX!l~jH)oOu0Vj&&M>gcph7>}FiIE)aL{|vDZU073Iq^C7B?9Q0Y-9wu*9=j-CueKx1g8&1$zAhmfXI1O%!HkDxfqN-hcd0XVuz$)W zSozH|*A5-+gjE>onIg0#(*_=Jl{xQN`9s|)PyOrE7V#;gH{H$&4+RrFw!Z(>H0r&5`R1g$sIPd0I{}Y2ke-1683M&nU9FYtDEq}f23i6 z|4@?bA{K9m^cA#|;Z9XOql5Za2n79$@rN=bPsPwE7p&Gdgdp$b2YZ=1Gz6u3Ef@q< zQd%rd-#gnHY*Bp@rTjZ>uLF)a%Df5Ow7OC!>}^iX{A{?PVF_$vVC z7q-MJ2Rop&Nqz;7Q+(xgxu(nZw)~n0TVFXj%Bt9kVnK8)GMB7?kNU>k8ZmQfB{Co_ z5o3&~y-4MQMHQC@rzgXXGV1R{GhgaHLSi{s*VY%dMqLD(G6eaRG|hbtv4OHMwgNuU zZwD(fA5U< zUISyOVsytO;{~S&R%f$Rb){cwh#o^7bvb{{;~aLuoXF2A;aSOowx}24tgxW~Fvk!8 zmII-RpvyccFs}ZLpd5`4Lf4gIrPj1Zpwa#=6Tht>VPTJ-%s_n6tcsxt!SpBEQnBz6 zph=H}b7F9kTffUPyU>X}sQgG}LXW{>CRSv~)nKh-aplTb5fpjx3HFEN0#Itlit7~) ze+%kS6u5}WmgZXT=!7Ffl4N}SCuOXMs1YmFeBZs z9>~z@>fO8Y>fV&F8K7NZC|!-PZAyp{@^^O@{E2#iX%-t*gf4O$WC zdj{sG0>mixne_+Ot4=A4-FeFD#8aHnX?161m8UD5#*<1hU`2)8*DBYzAFC|>*hASd z!iT~na!XArd{MaC23_Xy_STPjZt!c6<37bWXUjvv$!txpyt;OoJ@s4v)F&H#n<#Yg zm}v$i@!e1~djk;x>cfSj;@7WfEH!#o^`!$w%p@r`EUVN|s z$GwGa+kMlypne4sCxYewbe1Y=#+`P)2iL;KIL#8pkL_QSd_Mu!)NJ`YY5CpWe67*T z2pqwVldQpZqc=Z4F(_#vJ|*6FU_^ZOdI@H7vB4|^HaN>j4Xnw|7D|*i=|$(lEjyotTdKdVezw99hG#HOze2WC`pXr_A+lz}+zM z+y7TN(Ep&$_3uSL9Q9z<&Q)mL07p_70F~;Vl)6c}XGvTXSC>6;-?MKZ=wf7<7%af(TO5Lr6=ibci4+ z-Q|Etm$WoU3Ji#JOAa6*-8sO}%#b57-t9T}oZqYGo_p^5Tj#EI|6t9UnTK7o_w$WU zJZ&*)Dt2fc0mLIZ4_cP!A0B!mJ)t|{$ z^hILB*B_wo_=eC4cC1`2>$lvo#m$ zU6WC^(Iq3PA*{HSrs*|8jMvBjCzpDWEr5sk|ZsZTp@)b;{>QM2k zzt=W1bGll}LjdTx^U52~OlpF6M8#S}pD$Dh&oq?s4cx8YSWw%_y2D^tXsso`#2%DI zsa`p(Xcpq%`r_Gy1H78V8hlGn-(Ukssu`B&@9I(S%nQcgklAy|$$8OPq#WurvM)_q zBwALA3$rE!h!UCg++JbG$O6NRZACf*yKOj~?zbQi54iVg;e+sgINi5mS#bzV`#Vug z(*Vbvl5F&RC__4Zqg#M%?f&>IC;(}&$W8C8+d_K&KrnmpD`x`wew`(K?nQs)P>Ic1 z*E+EowrbsNS}xgQ)i+7CGN4B;ikk!1Fa*x3jpL`ray4Uoc)DxOc86=Z#^1SGNS!Gn zE$tFqFG*jSdkcEN8K+;-Ng?{ak=oX>LWVWRBhhjzu7RN-!pZhxVf_n_3)L1?v zuB5zyx2Lcr+X_@A#=W#BM%>JGxO0_X>@JMHcQ7YEo6Bl z)u(~jVy_ubUUfF($n;SqYIP#uXe;wNYpV5|W#rEAEbc*$1Kf^q<+<%lbX!BW4|u7; zjF2R-K(6?9ylss}FknCOA+H`Fa`_e#oi#CZd6)kHapn5!Hq|l3!6;|RsMApgSjFAI z=5+d6>u&j#esOfR4r8f_%m?Nc&q9yT_5E4Alo_piaonUNq z)Q8lylFZCaVNR(s=Ei^ayc<%40~}X+QEU?sbiTUC0_{?04%!j)Os>EXb|KSNKMe~Jnn!0Xx;B>= zUfPC++9#I;eLIPz3_GJ7d^sid+H0`1X!GdO>G>_1+L-=?#mrYF?`jyEAJO5htcA#l zC3=6donIBe#CgqAp>g+)T`_f)8D=%k1+|f6zC-NU;LePMk%Gv3%R`lC#o>mMbMwTq z=GBqNZprT4Z#78~@Df-HC3>v`NU12+t!Z(fsV3vTuG5E+)bN7TPK!I40HbW^$<^VnVUVV?=et_3o>i6Y4E> z{xY*%k;N)2INkH|I8lz{wfR+}w5|2kkr#1=;k8xKV&-GV3Xx7!WUowugoQTr1ykN; zkW3sMj`BCjd6ey9Cu|)ptVakhsf$%qI(&&b(tK;U-E8a-pIqaLNGF+58!3CcKZy51 z*EZq?rNwi;8cU@6D^r->w{ta%(pf`2IPPlRT^HC~;*BB&q}r4DG8w0j#7#lcFz zGK{UFu}zl+x@c68l|&oJar+ODJlwGJw+}SrG37@xnQa3HfP(BD36NtN0n^MnQ{BQZ z)q{&3_qm_T-susL5lWE{{F*MTtBL(X_s)JC|5x3>Z>FIC17YNEG~j>AUHE$zpzUnH z15`Nv|H))1pzq0Svl1;LQSn3u>837cM!X?ytX??9buW+Rl`}`u7YCyT!IVBPARFnx zIm2-+`x28xS5!i9{{_>($UEJTOhet;qcqhUb}lmRdPbB%6Lx;#-CYN-B zyaDvr(g^x37@+H27O+5=mjhm1cU;7I?-q=@hBoGg;c_uXN?35%7R zD~);H{nx{$`=v6j0O_G4YKmFtkf~tvsW<6N9|G{_`1=+ClGy zG59sftH!8GZpgX}zzV!XBZZdR_LKJ)QoxoiGd z$UR#u+{D(l!9DIKyn)^U)bF(LYqa}a8l$Zk)9@Q|UP6ik?STM={u>xz=rVo*0u;f2 z>W}=lv{xJU4_X;XhV)-S-g6`93MfA9e9zAcrjn<{X4a62KaXA5mGq7?)oL1eGTMXgV7Y3Q} zh{Eo38#!>Iw$yOaZ!s|lA@*P?^XB@=Ltbm_gH-J(9XEN}sgl?F$UP*xIA)_?V_1wH zL3GW0diP`*xAUmP$u4%Rg#1zQ#q6%o>2gDF*xE9!vaX#Si_T&oajYlEzoh%q&MXQ(UPQ1W`KF)nF)V&3iRe?3mKp{;EK%m6KYldGe zBH^Lj0LQFSoMQKyyLwe5+ZfIZap8w{F*qa^M6l_AU4)LQI{L?2a?MUdI0~hUB3KqOFQN?DoAo4rC)? z9jLAa7v)tRZRR?4RdK$IEG^R39h%{pRvInXWl#5qTKS@pjf8I9Gt|K+ek>N>&WK;y zKyH-@aFMEiivE(pE?-w2OKx5vgd6XthSbFpp?9;FqVYGT1>pYY%KX)Z&vK)Bs?*BF z%)E#C8vJ+owE>iSrVf}D@H-(GJj-7X78tSNs;~YzTytdiUcr6FS{b@j+AALyBaQ({ zGBV|00i*8@IlTRofCB0VqpEkI>rw#ldnpAlD`iU3c>`ajdrdh}c~tpX(RQKTonn0D z%5$D44o8r?R9$=-1TZd{WoI3mhcwW0#(B2Pp7)(dnU{57eA=W8sZDWN4G4uD3Ev_q z8J!d4TQpg$&kMn!%8A#6Xz`}{Yr{&%?+KcE2JpA!d3W(n!o=L>sYi1v7-%L0uA>RX zwm5Ck5wU_uagU8%PMl@6(Ef=w>e80-DutHHoc%s0si11)o)X`FbW*@B=Qo#htwDof z+bVGh6Rx~o{77^}wBuydTW@XQUtw%_|3+Kj_PJVqKREEnEl#1LCP+w&NoA3sh=E&U z7X9T2#!YZ8+lZF<>~a>3za^sg{nhgF8WVH(wjvJoj(*1PoI}uoC}}uN;98Zv3A>+< z>=ma0CO)xi``86Z_$j_Rjo^k}P6~HC zYqzrR3RK-ZAnq8rWPv~S>*d?FjtojJw(<%J zE!16yJ$SJBU_hSiA|vHf-PX~7M|4+;ec+Ke$DoGpGgoRF zfn>=kYO2+`MYAwuGqq(*e?e)Br?a!OH*!`MKbh}U5adD7WhC8VT3SkGm!4}i3{ggG zbrM6!ReHPWCp9lZQ$tDXxhqX`175UqUOT_0f-7xaRYX2diIlMLPN(%mv>Mda28OX{ zn=T35JM4eK&1U`sl$81+>IODvH`7s1PcXm~LF?E)GxT$E9AQq^%lLMgC< zwu+hYU~a?Sp?nw(>vy$eZHQf+N;hnaznx|ar|R_TDg)|JeL$^t@|W@V#Y#w?EY_UI z+EQ*ou+gz)K2QG8+UAIEwg1jC%{>mi0J9qXLXOiek@~+Hv%h%+9DF;S$h4AL#*CFM zt^mrCM`~DMqi`ed0)>z08Ux`07oe%1q7~80p2hrMbEQ($epRsXanJ$}-MqM5+|o%% z0$ByUh!02@!dMy<6VIFQ{A$b_Qm=p~AsELZs~yyq^yTA)yOtsC{nT1?*FCpVeD~h^ z&=y*2i@ZHIH27qd!KG>$@Q#P%rlbg|(i`OBgx_wdS$d^P1d{bdZ2}qiKg_08LEF)eE>UKVsI0a&smzm|1Ub}Vi`&_D+!gsaM5j1TdFd*JdK z&;>6%ZP>b9SFq&EsrXHd*(STk`6WlmBZF&O)pB;m(U0RWPet6{o-^ky$hg~W8qLPp z#Tkpmd{H~O@jmdp2(V!LY_?)Q{$#VT&9 zuy*LNju%)I_6sx^P6c0AL?X=5L5p;B(HaKk6{yUm`OptWn@J*%GE{Q8_@Md914C(~ zsUJ5m7?VPx)*hW7)>L9Dv~9ZiJ=70FLFSW^v8>rcVxUrvbHpQ&hO%_!jcO8G*xMT#n=kM zD3c&O7ZXONGV0R&i&uq}&*~S&!i)DG_~+Q_2$mcl>j1afO%~ zx2Tp=I@cUN`Rw1j-PVhT_Ox4GEYqEl!Eam*jc~{A>+fhyO)C1tv@yt!=gSn-ZS!j< zL)!+>@)x(RLHF&k4*|o$Ju>LfPvsQb4-mo?uONTO=p~RFi*|w!e~ONKS}=29R7E53 zL_cabJ_n|X$Zz=nrZIR0owgij1}NM9z={ytXAGPBPX;tu?_Vd?a%eoh>g(Tuq0gP) zkRa0~3U%30#9evh@FJc*s$lebq0}++){_qKO)*)=r6x^6EZvsdryeT~MX}WCjy!^U zuT}k=F9^0uje-O+o^-R zZ2w(j0G(Zzp{6KzaE%t{!MwFyV)eJOd)qp<{15Tb%;O)SZ5?PBa`&DDwRNb&ET>_w zX@sw$Bt%2I@xdaO`B_JdM-5yXVAML09&!>0zZqPZYY{UQW~Tz}`alXKVwPZ>k7EMF z)6)dE<)t+)XRhi!{r|K&@=Rzsk6!?=%$X!CpkOTHtpNgdDgu1w1n^V&|Dw9=J*b7B z8`A>Hl*>R>6?3!UB=uSqo{Gcu;ubik-|Vd%Vt_H%2_ydCMP~8fqeQ$9FcJO1n8*(V zav{qE5c?HM7$rSLHj2uWH3lz!>VTc}KeUH|!c+A;|Oxd|KN2BDW3kGgdjJMkG zZQG%#nOw&5$PQwPb5%8jB+b)5S@Zw19s&89I<)68IuWt~085W1@+UY+-{z@N*rc=@ z79HBo%hLLov^1q0T!YHjjW_cO!17_P`HB~@;K6F47dQxe z4LC-->S56SOyV9uPAG+5cpaF&&^``PX6BwBa%{0bdvrKhrEm4_LvIVJOrY!(Ts3&M z{S$oDZi8MEXwsX|Od$=SnG5SI93)~<#l*z#BX5$Zc^z|agHM6G5Dqy#0tX&a^Ajnr zzf5O)`0&8+a#m!~&O3?Pk8bNvpKLlnP{3*z09{h119CE(dF~dCd4gpE687w6@Y${! zR@A6UR?z~G0kvhKDKE%>n$|2hLDAEYB{fVqHY_*pCm?0AilU0P}R*P8-)a-ZJ? zfGy~dIsg)$03Cb!{F+u@&2>+a=njAqrB`!P+RXmFm(M`^p}S<{t+9 z7ti%wE))Mgi30^FWX{I2C+;sCKrlu?FxJ#BX6w$yYx6P+_1%wu#gl^-ETs@2o*nB^M!arHD|5ui&2WLB~kXuBlGc6CqL<{;hZ4(7O@8?B+(tD(?Z~bZM zL9Cp@Iwx);6llA&xiZ`r;NITbUh6ZSIstEqwGQ+rZ!A#l7&g~@FqoNduFLMawuH0e z=6SUy3?ds(61lLO!J~cS<;cs?;E1R!=Bl6`7)y{sh?zYf%R^xj(ytZN{Z3;lO8Y{^ zYj2}AT4bM)R|&AvtEMO^3C67I;lq7zPbOC(>&^@cBxKX$Jk`sd zQ>B$3iIGapPN0lKw#Bs36^3J#3!>Y@Rtm*PwyxgF=V+s{EX?b{b~ib=s^@q&U8nuM zAlcwiMhFi>N19uw=SA>&+QSg&_6^d5s7ij`MrkaSTDUg6F(xuiXn>A-JXntzx zDI!dT^yw$m7&5eK2dq{f&`xMi8FwjkNq5&ddal?{QW%t`?P!}dCxbZK@yvZYJ!nGz zIg8!niB?|gz#Z#rFXiB{f&)! z7cbxQ9{c(M3oEX4?7D9|F<&clljbM%Kuhtwfke7R&Q~%@6`YM|H_bEa>fuQtb#KH- zlU~vzw0(&Ah^>#*MUuVAONXSJO}V9eE*rgzynX><87dYw&7^3-I^uE4>J;3*NH&x4 znFq6NshUd#E)HOzRH& z6`ElT_8x)NCIgsIUTJ(s;mSeg-~X7-ifT&-_(fA-ADgVB>c9|ZTx%AiIk6a=%N z5=y2uhI=b-Y?ACD1xBh!uls4Eh(l!9Aq$8E!m-gv2%!8Y?Z*FUy(LNiZ?T>9tit9kjbauG*7v7B zHl_5J5(V1^EkcX$CC@akhn2SIjK&v4zShivf`-|iJNW1?em^{^tF5A_ zajQA2{&dA8K00ex#J%6^8DJ*p-OLMI3rJhb$!N>DA?+{kyz>5U3_wg*2cS(mMPAk4 z&lM9kHFbTP;Xxd+#24qMIQV@{qTv0MoWO>C=xDxAq^}gNnit5C*wWmP@@r|(xlg~7ISdPgN?n&-H zK-8|L_I4w>7=~fvc?I6+!upJWA^DAo(#rvgHD)$Iz*u{@$Ls-Zo!*66mTK4`LQ3od zA_4+V7ta8P`d2m!hbpdso*PK~$fu{KW~`1KruC%X>9=JwN8*u7ZgcV(qP3EjreG$` z54i-!_t-?e-ID=ts3j@!P0z@tF#QSuLJll1gANG*)gSX;%{Yim0af=0n=-6CK%ItN zFn7Aqj4c6cXO5b_yX!MM`St1|w6Yh^x7aPpr0DLE5Co%7b8>OZ5V{Zf=~z-Nuy@K< z59IYwOv?5V?U^sPYHdq{Q{pi?ndLblqh!zG)9M4bt1Wql{DWTqQ3UlBGtD*97Y%8U zbFN5lY-nlAcQBT%PQQ!{6NOlIO$wLybL|Y9QM2K*Urp~*elVBbvhzA&dsL&3q0{}p z+xZq^tYmfEVpvn!{&t{wj9}W87m~solS3aj$Gsk@D|#}@X1LTu1hU^pUKJ|%xiMkV z|K?2YRVf=vdw1JTg-`Xb^w$xzKWE^Zcq2^il*HprKD!9yHh=sv0O)pdqGde=$Xa56 z&X-N&2fzlOv#x#m-0aOY>G_$C44Sb@f>Q8To=tNtsj{`0hMfZDIf3C-uBQD`3 z5Wb=+J(wub{PZ?T$RNYwC#AJBpcZ#Y@)`KWG(3#u>gfA~nGs zr8Hum;C^1`zv!`R9#TlmqtIa*GxPmfPve7^;u(?F!?Rb0Uq7s>v&fJa}>Unuv5TDNF<9(Pbl*mJ*y`vlxhF$Z}#LF1h zdScJ=>M7Y%oYgO4*PT1v^a|dV2bvKAS!1J}&8&pr8IQ zt#4o>0-H~=>_u%H^yc5z-dzbK{=(pw_N2B>w|`S)EvLjITrb=`>_t%9;*9?N@T>8} zAl6*=y_;%t8__V`$}ut7Wzl{jsnPq~E>;->`o+l5M^C<%AK#(IH-C+W!(66=12mCO z>qhx6EHsDn8bF3zsrgeORM&!@1i`7P@@Fx(RBgBfIcZb-C6Fq_)e~;}r2Keitu>da zZuN9;Np25-)$2za>B}%hmQ~dDrOYF?n$zA{)Cf{WjYdBlTb8t#jg*$$!wvU%v-6ovw&R=4=QF$%b zBEo@(1B}tuP~i$U_6XK(idIWzs|vGvVofeg-F9t0S!_8xJmEb}a}Dgf*Qi@c+Na(O z<~PImTdQ5IPPMzjVpnhrb3K6(^^;8R=~a&{^x359dp=jFxkOux5UpkZ1lSV| z(cC9-Lbn-6oMVh&cAc;f=EAI(*imKl&8WuH zsEBE<(u8GwU9%)26Y(!LM8>JC?TdEmyplc0DsL9W)sE1#9d!r?)gxc6QT{U(tu`Cl zGZ(;e#jS@nA_$CZ(*5ve@ZRT^Q7nlyrp0)}b%F|dN8VnMk$X%SHU<)j5{yVYo& zjK>)2_0t_6-0@hAtl_-7E>)7+pzOV+Xlu6w6swxC@=U8b3|V0ZSK>4=e1*xT2YctF z&fJ3x?urGX3^`+pkm!lT+ySrnog(qlFf%@R$1BclOUp4bIV5jyyt6uzWBj`d%fDC} z_4jc`Htr~y_FMCVj#i!}8^wG{miO%1qO)_;#u@)?uc`)GYMGDle zC~+tz46-C=2|eV&WCEHKAgqRzsekaRTLr%S?O=o*Q73JQSV#pb&Lg~|#k#J-``yAc z(I~^MT~`kJ=I}kEHB~$npRwZg4EB1=0$!0H+ed!r^muh(Z%qtsf{xCyd6a~_S}K0i zCFddNZ#xALl3$PNa5DFnbZp^sdKpmr6^l_f{?q^cQv^E3zx^U!fn>OY+`nJ%bnurJ z;_pLMYk8#$(j`558>l@c`pK17k)^SDtpj7>vV)f6z z=D^gBZAq}s4#!x-#%Lsyoh2%F7pIQiQIfK-^nsXXz7@(Uk(XZn$>{xi%=IDq(-g5w zYP8i+mloqDy+(J?l#=1?wxt6T>UwY|E_Ps^Po#E^EKQdKoX@xJl$t7|$CE!=$7~Gi zS?l@sb(_`bpUDO9Toq;|-0~j~k{C1dXi<{5VSfuh=}h2hojv#37u$kp7|33m(0ge| z)jqP`yg6}dgmIW*4mm3n=bc_V;Kj3}8V=`EMBl^s=oqImTb7jG%e*l0vF9533g`pP zOMM=pRR-Ul^G;N|9Vj~tGcmI^D1{z?+!6Uww=J4Lu9e!{-iRi_E>3bAE-n47$-Z(c zv~GDF0SOVWW2LX7t^^~x(!H4=^~pCo^P1YAVZiOSJd2Ci%1a9xI%rIi zYN|o)u!$9G$~c5~^QrNi{wZgR<9kEGlpox20MV;h4@ojOFLB)m=w~U9 zPADGUkOv1b`sc<{sDH+%mR4A5s-F;D*7u%+$W0W)sV4Vj+~-wyKpnCjxm2V3N!sB>OD)Q((&{i zoVT6hnP{bfO%;(-W&I{bE(#CZDMXYM+cI5r?St7dAY<3k*2dzr1ockqC$~RZrqXu) zq)Q$J0x$#eabS+LzhA%>hcH>LJB-v);nW<_TyUBaRk2=QSwi4A1q~W#k(A|Z)*l1r z97gnf>QyRwr`>up-&qV^MaEMrnR3n^Oqxwp0NXLKS!$ARM;f4fOWd6|M4C#5Cl|;q zcpiTZRx}_%zJt$d3%|_Gn@`AJcc;KUOx2JJ65eJ@KvacfQ$u;ORRXA~*u}1%aNNQ1 z2h`1;_`zZpbcxPzu7PYXH0L83rIK3B959b~$d)1g8{4%?lti$-Gx57Y`gkRDcb+bR W#yhDMn)&|$EA_uvBNOst>OTNY=z`k- literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-10-Attention Mechanism Process.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-10-Attention Mechanism Process.jpg new file mode 100644 index 0000000000000000000000000000000000000000..17be571781b039612b0d8a43826d4c8960c69856 GIT binary patch literal 91348 zcmeFZcUV)~x;{E7(h)&=RX{Ae%0 z^d9NGB)QYI_g+iazUMsW+~@x8KX;=OGBRhz9Al2Jz3=ypV8^htz@__&DvAIO4glbQ z{{ZYHa2FsVARr{TKtxDLNK8ybLPkwaMoLP?Ky{gdnuU>#m4%U+`3k21*A;d?4rXR< zu^W6hZ;6PAuyIMqhzm&z2#W~)>;#9Hn3#-|jGmmFUg#?GRiXd&5B4iSNrdZ;uZxGn z4&YMa;8EgWn*la(oELEZ_yPX;gM*8Qe}RCIh?s;Fya0X)z{SDC!^OwDZ~-45yxJE$ z58zW?pt^cXp5U^EDIvQfwUB>IDiOz>l13WMfo)FVr%nOHB(!w&42;*Vb8+9`5fQyD zCN3d)_nv~HlCsMEM~}6%b#(Rg&7PS*x3ILbc6M=fbNBG{3Van59P&CeEcWfYxcK)U zJ|?84XJlq&=j7&Smz|3@?Xb7FtZYXTs{!vPl$ zj}m|YM~SsLwzey^)i41f4y$KQ#1nZ!P}Xl>blQ^HT*=Ce<2bHF-d@3n&D2!S_nlmJ z?K>mv!vbj&MOKhq!V)Y{?12ULDdia0wv#L&d!uJq;KF;zxqbBU!*aG0lrI(t=t{we z8L?2~^uoTme#QcQ23zN}hFD;y1~Mnv4_hfb!jKBUj=#IOs1r(H0pig>EFj~6*w%B# z0~^MeK~&}a@0hAAC$xsVX!H3OS1UBLp8 zNF@cpIpsg&`OowDkNo^czy4!?{$mgS<1hZ>Pyc@wZwMV@fu=9byIqbRF=_D@cDENV z;Ki!!o@!{Hn_~fXvDc6vOkjkSmJkNO!Xb-lNXQR12qxURIs*%|v$kRC7qCFT!lNCF z^U#_VbRZa44cO_3q@3W5VF6?v?3Cy;={pStfcn=%ivgY_Iip=Um{k`hb@ob-jO@iA z4wALJCb%Sf4Ql%y=_q%k`TB{Dbo;&^u>rtt3>i*lt533J=Qj z84@bWU*AlpUNE|+B%;)2a}((e%nj!RVDI>*ujRVJGT8ro^TXQ zGNx!_u={&mdqtyk6Yp(m0pVERTXL5*9`E;Pw&|6qZb&W(*gfm*cqC=OnhbaHBjHZ_ zKYrYj{@Ysrz5V~%+B>mPy5NtTiz5dNGm|qjFMC>ewBd3YD|4M?kTEV{n2jOLM%GY9 z>)2yT*_N<=gMv6 zDZ~Q7u7VWD(_JRp^3!~PpB<+_efeFEv*X)k>SDszD45>eJb6(i&IL&VycMW5 z*j5eTC0`)DzF#c%gW@3d|G9Io}lB7--+*c6>Nxy$NB2eVZJ})-^m9 zI3(JD8_|T-eEwP$8l`QD`n0zt#~P$St&a1*JP2ua+%YP~2rOWzg9YyD&lT3m(>1fW zD{e1;wlBv5-lOF{%G;EB zckmRLWzEVze%ly6FG!$ob#it++nm`+&yY*^x&nP8OX*EVIhJRa6zuR{$TKQXThmYy z;QIIDXljcE6}jf8P)v%wwv}zmyD~0BOgYky5#>>CL`9`lSORHb-2c^qE3q6kLCh6b z(Z=}RZ67bznq2Hr=_nt2!%$ILHT)e_Z1eb|z(DQx_Qo}1hIxbh861yz)Hg4djr#hO zP(5SZU9&TZ?-lJ|-S90~vA})c&`wNe=)fguA5~SPN&G-a#Nw)p+(0-Ud&ks>R4DoFx-+Qb1q0zpPSo64-cq4hF z(9oiVh22Q!>(L|qz{3rRc+6N->Vj0&xv(6IKuNyds=mT#F6?U5=r~#IO@=Gp7nie3 zYojo&)4NyZB^LB9Np9Fu#>7n2O|~Qyyc8X*yP7s}+YGrBwLU9o$$>M))`quT4Amk^ zeGGhlf-|O$JMh=AaY7daPblj^H2-8W1x-Dy@q1Gp?_3XbO!gqke{uqH3I@fBJ}gD90O=QrWar`KNfv5ITg-T- ziM*#heQGQrS!#$z+e-!K2K+EF6DnE-t~0N#SjYqoGA}QJ8}P!BBmG~>_rK|bik~tk zS=77oe`*R%_l&|qpzp1r&2!0$#OQaGv}NV~unBfK;!_e}EvLw)F`+j5BvKVoDFl{ZWBBn_?xIDN^1H6AKdJ z{qB1UJH_%P6|udoOrG}3wH`>TWZ&AEsSUtll^eI^pSFY{KACOshim@#r=*6~bg9~& z6N*laUO2ILqD}o(AR#XKDVGdvO7*cobMJ&OwL3A#jWMrWz^`eY-1@&aNdKL&`p3CP z*Y|urN+%`T50$-i+iYWKvj%)=T8yX8@05P2s&zP47=IF)afQECSshvocX`P=Y(8=> znrZFK&RH4%VoEa0Enkx0ibaJQVD*T}-Ha9*yeg4LzyX73mVb^ugk}vF*Yx zHO{JwwruE}5owd1u;qkN{o9AYo93!10>_l@(Uw1+>0NieFZybHv>T2>SDyvx-iRsr z7NCqLyIgmsa_n*18dPBDQ6p;SV7UCWfaymVWF(EUt*db%w^%97Pa*}6f>Qo7DFxN! zMqQaRiD4*B%dpvIM{dc33;i5+w^OdIwpLISy4A?8i^^rnsx4kafawYBZ z4p%nww&M}+q79~v`4rqvFYVuIGhq48P=hBv-__G1yY7(Dd7X|#andD@gTD_Ve2G93 z$?v{kN16vjO3D9%@A28H3 z1dXqLp%V;-B-n6Z`8#arkYdRGRU^Z0O$K^b66N#!ciVKQ#gtfJ%nSVB9hwnCSX((_ zlWj?k{!kmKLrXiAS+q~{{ zSh1A5BTK(egjboB!>t(7xEN$VaMYU&`9L$R_a+t0rIeAh8!tZIsnW8Gxaf6h>XK*b za6!$NQhwin1xx^P=7a(ZyiWf?$F@%kGJ5T>a|P$Mg&YANg$!N22_rQd7<1uDxcj$v zFY3LXFfSf1jMl8P=;yo9DzJSqg!Su#jOTc1T+pa_C%qmKi{D5n8%Az5 z!lJbGGoT-Y>E~%>Ex&y-Sdy)zm4jq1fTM(eIE7zmQ17(XQxeaz!XB<&a`PLCmYU@wPL zD#==ovnujyjHc~YCbA=sJK5b|UY19vh3_BW(ar&Xyju(kil8eNcsU2I6VE$?vCd4( zM{Y6I3Z^Gu7_m=Rg9Q*Tz`0R-{@z3<3ZH%#b_#3yy$9~UH4klND<4dyeL2s^0;ip>aV3i0HtGz$}eH2*?9H zJ5O2AUQI{2N(aW#K*|MSt2A6_40I>`?9LI!UlTU7hy@sZdk)~A;&-MVl+<@8KfVHz z^0S{PzOr=|XdKk|cZQ|>GK9g~nO=FiYnroU;Gm&9`R_Uq!z!3H%wu2s>^Dwn2706G zgG{gR`AYKry!;fc^)u5!si$hf)PsNf6^x3N{r-y7x{(JcHgYQ}>x6LSijva&(yLEv zUjwsl9iPc1c{!R{uMJ7xO6@4Hb{5jh>er)ll!_uCd|8g_qqM4!WrST)yT$1ARY1DM zF!mMhO&L#Sbpp>FX~cF(ND!twS!JkUa%EGh3}HXzrS*6x;uI17FnkWBytz@x#2BB- zK}pSp@ywDvJ9D9FI*vj<-hs5#>Waw7t>yZDzLBu><|G$SS!F`%)~xq4&3F0^bGRHX#S?Qbvg-{F&T-LKhu?H)XWv3Fd zCqdWCn?5F8j4puY7(ww5cbd%1yMw^tGHCp**HC@=Gz{fo ziZAQISaCBO^ewTVks!!9=D^%zy9gT-0H2WpzcCWFXgrJsP=h9DV1x}16d?b(zOxIH zu+_|hR)>EXHyw^FScldQAv%x`u`hutDpGoY1!D8(1QD7s!1xFvw)#Oefj@>DX$~n4O`jfg%0=Yq-Y|=QlJ38$V0zHe5V*vKE`qFlJHwzU zK;<%o^#vWFPuUi9ff4=tbwsI``?~86o$~ ze>DmPHp7%>%o(>QKDmpTSnLY0zcqbB{D=YYb36etNFYX$9)cQ#qRDw5)>^6cl01k9 zX&%cfe~4*SG5z{PaOgaOc&f!^Ez8IV9ycXz_Y}2zdH)I$vV+%B(;By6b79C{Uq)C} z*J4B+Cq%`Z2_D9g}_}(a$Gf5!hT| zW$MXyE8f6O2d~?S5YN=~tsD#Cw|qST5#PJbNFIgWdXJ;KGWX;*B3|BwM7oa{Kubf^ zh6)SphWT#q@F~mSMv#>K4z+1|kHd(HMEPYU()BKPe%LUL`PB9pQu+h3(o1nRtKFZr zSnTZO!o1+p+Zwij_}BXy==Y}Js=t8E;O4JyRcDWwIG3bVZT+aFIN1xt98<&0$6$S* z*p9xYuWwcPY8^>2@hAambdxx{=l4i^gRIJ1WT<8gq3bsUR--Oz+US1fA+B z9gZC~j6$Nr8Hp}z?SU)uA8ilFj~n0(GjEL&-avvf{eKo<#cWfYrclDhypBKwUzE3i zk;%jYp!elj^1;KTH$x6tL6z)B<$Al#`$;l)0GCjlsX2SYCRBspt-dSWR7eT*aE;kHZ zFZ2Fr_vvnW+mmU7Tiszxp$vFSji+aiy>!A|Jrgi1I)REMLoTT7ikg&-SSk|#&Hz8= zSVB(5{<@vyPcCvkO5> zJr%(vy|KMJ94}A4XdPZ|OEtE+3*9b$EyC2Lrr2(UeD6T>&8Us9kA$~N@MATRA}3xe z5k686^19sHa*H!j!=I`r!0?tmhuMG%=IEj1o)#9E6Wl&JbTUaY9@ei5zV@a<_JxoR z%iSj~(nC&fR|#k%=hQ4zU>EQAnTf;`JJZZ;vB>K^Bd?Zy6nm?G0cF&pIe6d;(JwnQ zPIP7IW(uRdI}oC-r0yH}Txs54FoI5w(RIo*KqYdxU4zMt`7 z{FSyVhB<8yZ8D24&-c_u6Dj3?6PAtkB;Tb(l6)0gRRg425SO6AK1OM)UC+zzdOZIA zP&c|PU0eYY&N#56^xAyrAc^8mMWssC=(SibzGhDio~2J_xyx1Ih2kF5O>H&ZY_dzL zMlx+^shqbQ>&r#QcORKB`p5?wvWZCAhmugpOV0?uyh5gqKSSo4s781%<(q!_&V+T% zgTdTj3!>){CFU+2O#TFAZ<{MpsAR7mTZz23P~O*0*u(<8YFXD?TOLFNsIY|67b}Ah zG8J_Of$t|u4SwQ)l9UPdCoAZbZ##t8iju0Hv9OQ!E)VFjdq3N4&|W7pBsh1LV*(%0 zpUk#73g*;>X?TAk%74A`h!25UgT2)p9X^!)VbNND@%}UxFxg@0IiJuv$VxG@d2!f( zQqlT6sekTcgDKqk4X_fiT(HW@8?9;JZ7|l%nAjyvrNDaM;g#R9?56->)Z=Rd=g@SR z<>awz`3e6Pm(2&`b!|nso5Tw_#fuq>ha6m}$K^9*s8d`0=V@>B`tlkbU9T-_PEo5w zw7pj(8Dl2HmGmC5HnS{udE`<1Anb}6ofs9oIeyK-0mqYNLEMM?hYXYk?--@1RlbWs ziXctT^F0caGWjBChKq#gFZX&qLEgsT4NpjUcjXi~IB?X|#H*UR^4+R=(Ac1(@kSmm z=gMeh$oW<0OzR4sdZmOSh&kFHv9QvTa>(`*U$z*z zI>{1%=BAhyM29aO&@?oz!(o5$>b5ECldD$4+dkwWD(nurLLA9i!;SI|gf)L}IGBc4?mG|MmJ>^S6l! zm2pl-#&}*vIKGl*_b8J0hY`L+S z`RT)XI!^m^+lS^?ekv_q(i7%cnK@Py`659Z>2EuYZ-0rAlk#Wv)Ao-fd#R^zBf^d& z!j~uM9<8*~;eyJjXiT2bG=s?MxiW`p;EOfJR=bcpMD%{2NkBGEaYcW~ z_JGMq*1J7^cP;ELfzuB;J`WZG!bdF+GJ9-rUVaoi4TXG%bf!E=MBhu!J z>cWuaVA?1juBrqEu-8J+-r(QQVYSBd!hfbhnsj2QM2Mvh8fKodurgSml6`qtF(n z3V{i*>Dz>J?|0E(izUqs1a8^k-B)5!R)4`;NwCrVAd+Mb4kPUZgB#s?33J?Dj>5@H*a(tMvZuKmnQ9*&F zo2rNJRu5AbG45k+{8SDOjtV7F{^u8NVF3;@wu3b;OmJNGZ(yR{P;5Jf;ze2*uj5Zb znAxI5MrP|J+e{D8(-NvWro7I0oBZ6r$i2ypqwOWt&dV>Wuz8;?nufXdD9%YXj5Kkv zW<2+D=zJWl^7s2S;_Xko0;IyCdOL?6mPZY?jxrfowo=h^5QWe9iBsb&!AOcTcT8Rj zoZodStZ8~!IYb9WtMPvJ?+;qMzH5~Z*@`sB@bMKEYU&6nf4Nr@;ye;X$ay-F2XLo_ z{dP1WK?n*$O(X;2ldExoP0B7U2$WF~Grf|Ld!uVG1Dq)w!GWF{YnNafy?P81Kr$m4(!N zP}g==XSM_}uWpE-XodW&D)=9Gr ztq?Z1)!Rtb=sy)aviDTc<{=`CQd!t8I{Kww7nqeA$olZlNjm84u*?m{IJ-i>*lnkHC^EVqC zy1Qds2wzAIbt?1`Mm;`ALp`i{+s5dY7V~_N>BfAlT2t`zh@4jE0qBwr7AOZJzRyNS zOCbABr*!ykW}md-JRPKa*G22UA|AW4AwV%OQ9)Zle!Wo3zL+z?o7#pcM`_~i^4sT9 zr?>Lb={7DUx5EmuV$MkCwL>ky5}t9ee;T)Ylk1rU$(Jxb6Kc|_;u?k zPJKF{Kgx{`wVNA(tJ+kAk*d;PeK#e%J^g$O!)b;T4f3_fw>o|QI+mP7O;hH{jo@eT zqK{>D@5F9Bxf5Y=g*ChWS*5?|2M5|yE}`#M74atKN-;i(xlgq(GC;4)&faPz?lBtg zO}|*MDd3}#I&R!z;hft@-kf^Bbp_x4tQJw5Jl<;c_PA260&zw*QCMmiDxJ>-fV5mII>bVN~KEvRQJG&R!h};jEbg{`^If2#VJ?yarzc(=v<@{hBEr{c$+2T z^5IcW;8FNTYNBnGzB8^Oayruva#1POvqsbBM+Bsy-(X!*9$mG*K;BE{QP|v-emgj{ zOr|o~)QfFyFf)>#CORY3NQ`71lm-ZBTPYJLH@DhI>{+h||%WK!V5U0}8%am5G(*cCd`0 zJ}e>UHIaf=N5|JT(X?E9f3tRhV%J1|4&C6>=SHrd1C-6flqgB5I@vn$;&}J$IhME9 z!V*7COgCnuXd_KINAIIDrqy^9HzWh@MIPxEIpOpH^Yx_(uVW(E^P;NosL{o5bQgUj zENEjE)Epj^31!Ak8a(_`S9nFo?8;LyDn+*l^o{9jrEnj zH$Mj%NDz|iGKPe0{BYGPUmSunb<9O_zL~sJkjEsYsCx0`BbD;X%U0Gsr@RTE_B6x# z5n%K-^XjK9A=p_Y_N7U*)E`UMoqi0xDxt+gkH)kh_cHrf=U1Yg3(~5Kj~^(ZNl$0c z$54zwzR|fHyu`-5s{f!|OqhtB%>w$dos$Bwn!THzAk&mA(_0g^jF`K{wl9c+9nC?G zYl?jrPIGb#Lh(4t|K%8x-fX4IJ2J_A) z_}`^m%Sc!~PvU`m+nNLSjvC^V$goEeGkRJ-x@fl3%3alRP?yE zr`u^3nMt^K@oNN`ed|MF8cN^$vsPjs-sc;fvlWDQytZu`z^;wJuD>1%WWC=Qh3y>> zqR*+hdDp*oZu`Wl&8c#TMDmia)!eOmQ`9L0H3o8AR4w_Xb&6ZS;-0qBu^*AR5gRYR zRbakbVF8FbY^G6ZvJkTJ#{d<&vTBJYen%Eg4Ri@N;09&XuV>`~dA@>q2puhBBiOvq zXbX!CMovE(?G5JCJ}2~SbDh7PA%~6eAb(~PtZX2=c*79X1Qd-w>fjGqO!lzBG-@MG zZ`fQ-`m^f6NfLS<2cZI)2-zkuRgqk?zUY#2!ee|Lj9h#}_LBX?ToEUPf@q(~pP(UM zO9wPKZ=66g(1XB4-G>{$6rNXT&d-Il^Xy@&ARs90EoUeyc9|ixy!)tXsPOs#rhm{t zDC4c<1`gh(7r?bB3DLFgyj6{1p)Za2&nWefFWCl)_m9iZ>Ulc7#T}RAIY~67T_$e% zAM1!bi!EbVrjuN57_xd~wJPd;CKY*qV#1xL!HGSGiuuVc<_IvdAI~2`F=k~tW>}r7 z9cFsl$tEKs9m$8a_tdVR-&TzPk_NmilH=BHuEo>NkI9?c(Ze+>=GZls@ES?|cLO)H z1kQy&&lr$;7cxN;C^_y^ZoG-yI7*~n_f9@A9+U}aBTqL(HE?Z@u6uxXnE}pgp?04m_k(Kq?N+ZuQqwjdELgCg;+2aP1b+m3fLd{n- zV|ZfdR|D}j&s=i(H$InFs(M?R1=F`ui4XJS>;o@&%sCI`oL*iWvJ)w$^2i(Q91sScqj1!$<&U4=Wjb58i>o=HiNl z0w{mmu_(wsFBA*-zXA778TOyw70Ayq9y45v9h=f`q0W%bnTx+O`y}@u5^>+!(d)XX zc;&MmuwEnQr&Giu1PQY@6(AdD_#ESi#LLq+X`25o*&?l!*JkI*(p(A*_$1zbLwDqJ z5o(C^1u?t88o)9OQ2!TsHK*Tk}8J6-RFfa7NT8wmH0rIvn)eDS2xh-*kmD^$L~n z@qD3`ET6C97BtH4`tl%)A_sp)sd6NOgg)T&lE)(KeyZuL9zU&=vuidwfhbu#nO{QF zy)%y0Y;gs|7DR@lEr<}G;C_mUO4zgJ0W6T;miXb&qPPn+9n>>ylafmqDR>YmCZDjS z;;noBnJo4`u{o39iQ-KtSD~HBpTH?c^t<4hjY7>(-4<7$dc*VYlb(cGie8V8{sLcJ z&o4jh>qnttj$^=_X)J6^u??(2OasXewou3}cQN{vEo?uJXRnVBl>;K*=AHNFucxno zR?Dsr6K2i*wdCQ=7T=8Z*J^I5{L;srsAU}bhNvif^JdBKw zJQ!`Ew$O|=e?wMTh<`uvcVzWzB1x-e=US}@6K6Ozl;ac64YPe3lLy0ga9^QG^)jbJ zBxNYa(+sNYY=R^hILfnskiK!pI`=C_u|O%mVDj9C+4iCk5tVzS>XvC$5C^B0;*G2^ z)@lWh-bC+==^0Da`Fibpa9?6wrtr!z2~Mh;U0+P&zb|o2gT(e6b;^qg?nb-BA@NmQ45&6A_ZF{%dkdX7e?cJY1C8OA=xDSr*Fj$T z`8^6V?J(}*C*6Or6-7|Jx{>7_^&wl)TjsdB1smb6F(1QA59|X;ADoH3Myx)Eo#TX` z^Gaw%7=3(~e@CDjwOTASEKsLmQ*1>`Jb%ZmUYPKDD_>-2mDx_ANts7DCfEYBI`oKd z?<@QoN}$!q0b^JZbly(*p}~hbWxdFrI&mBBlF;kQ9c!!2qtWAa3ZRh`#safsCUZ&^ ze!fproQu<{`0b&Wtr9M#!`MPeRqSl|IpoVM5zwvQDx3#YG% zNx3gmj)$V&@J=eEY4=TRHrW@3V1O13P3|1aovlaPBk+;^aQ+TYGf$e^!Q;GF#5cZ$ zI~@>xK}zG_}O1b z%J06~-ahsC3A1xl!$`g<%qDlN#oJsGYA3X*aygYIwR?Xgt@02og6J2*0vL2L7NAl3 zy{Rvv&Rf*#{`;H%ehr;bu!kE|OfBY6(!Tn`hGT%N^C&!D_%4{jxPd?|JD|x%lbq}O zp0bfh{48^5+X_egfCyqjgTe5_hXM<*O$&mWdX6}~&WOsT{uxBmu(X|ft77IbTGtdn zsKQzXtAfP=&L-rr#RLy$OdUH$Yrw=gd+HyFkm{}AnO!w$2{3g+30oB%F!{4A3GaHB z!eH{Z?}-n}_?~L2_&Yr$ghSCG86Yjtg#~bs=RChbo3JI`O-pr5(;#G(4us2oC=4~s zbd7DFv=kIGS1dq+Ft+)89HJM-t0vjXZNKRNCzm63xc;Ps&*_s3eoSD5ujHd>;dMd~ zmHl-K?qB<(>1bKtciJhRAKv_1)dL$*nfe|zLpuQwn zYHQ=Nsq>Pu9^qz0-l{SozFFWyYAp>Z*6;$uJ6K%S{`#e@K_v%&AEuiF%)-bNr zffB-QoQatSHOS_kyF6Pc3purhrV7PxQSj#?*w>U@sI(|px}9%=riSZdNp3iALoM29Bz1WYTfKicaOu^AdEN zeYb3rJ@T?*Q+L$OzS=Ir{FqO^vT>6VPWiE3Bp*_w+-_gIc=LtT}z znz;ulMr8)fUC`P6BX`k1@XrEz9gYlGhsF*B?V$_%qf`l$o!IpGl+&FdEbwU`r?ObF z%&Oz^V*1qA0vwL$Ti#*eXCzH$Z$XRvAmv;OS_o^)fE*kUVj}e6hV<2<*PcIqu=RtN zViO8RenPOF$1qb}m=o-r82m(7A>)d)O^(r80<-XVw`p8)VHzJGp&h3cBKtYUFi+DS z$vueq_5}kD`S>!|AGVDqOu#*WaQU=K@+thd*kW#ZRHe!p@7bhT{xdoP!C6>fc5l2B> zT!M4O&D$k?2EIwqclgT~Azun0)esCmc$XIKNE+$gO#-dB$oiDmFem2c{pd9YED+)K z)*$Xf%I?kR(|yP(eWB#R&CUIhcJCl+4FjU(It$QrdYPbu_!0X&m0HrjH4Q&Rhcb&7 zhtkW6{j8S;W2NF7u>H#hBY_LB6H0jB=`JV^@}l*+`x2OT9kFe2`Bg9>potANy59yzu?&$lO)k+u0zt6@{!E_v=1#+WStAAu(Wx0>x2oLeKNL1WxtaBgD&cJ^mfKjQTxt-c9yu^2Z>)MQH!Hp7n*G1I~vyhrcO zEa=@j&JhfO+bOP3Daa=Bm8jxYICUMpfK6P0NtW2n7xHISmekf)glN9VzgFe*>P^It z??bZRZPcU^)~4zsbC7~DAs%+O9Ug3ur_}M@ z8C@rx)ZMguuJSp~o%8j$HtXkM_rHQ;74L#a(KQE>r;9H+K}}4-0*<6?Xr3_8eZIT~ zYySj8VcPzPmhrkI*TEw7|Fb*loMb_05G~`4X*vR9YjLEOBPeBqSOE0UP6$vg7P^eV zRYrsI5+(Xq9Xz$mFGcPT5ySiC=|xTa7$@zc zX411CCi@ma$7$jB5WwH>k)K!1N8yj`TW;Blde~hX+GH2Im~}5Eo8fg8UQxfI*Yom2 zt9sm(jN8%H6ARWo_1QM2wi;g^Sw0Syu3|S09sLU}!AntJ=9?t!p zX7a%_em3xW6XR`qKjDD3gl{3LrK9$rO9y6k_s?Y1rwy$iec8MI(XfoqsIO-?UYwqu zAy-4k9j|}=&yp(RJjv*p5>Ul6Fy*V@76QDn5ir3y_D4~JgjC-Nw-Cr9LC}l`nR=Be zyXqiS_vJC{R1|f;%Erfb&ylU!LJG`hw=c}492LR#wOq7%QIK8nV+%AY1x;f_b9bDf zJ|<{>>jZ}j4I788J=Ut#H5LHXlM04JzsJNW7?itEl$sq^{$p>&5gQ1}4nKlmCm-_D z3lZhX^8X93`L==kd+W16hN%xz1^OcHRkmkvtvX`R#lK@a*a2OA`eMGQ0qDwafbh(O z4st4cf0CkB@+RW4q$GD*P4Dba#4l)Y7H@K}kLbv?68$stQR)hzv6K|;8;A4jkB-`< zR;XTi6xjc>O6pwS-uio#j_S_!_p@WJwRUcfc5d(Y3^$zc3|lwiJa!e&!7m2lv}@(% zl^e19A>iVxKflAb?~VGds5_yDtC+V)P5##JA-$ik&7cTcRc&;mVZ_w&!W7Tay_h#& zZ({1geE6SOBzLlL^h8s2k@&|=sgNmE6DH)-I~UqVU#lA46r6>73fi?s%4Yi3sGfyh z4f-IyJCfvy$Ml*Iz-YPODSfLqB4Wez{k}mVq@&J-cBqp|cU1XMU@zDe{}b!%Mdf1c zXF8e%pL5rVW{3QMCoi~{C~UG`@1ZYm3?2v1G&!qEMBhtL4ZZdp(h!MQn7TX@YzxND zlKUhbWkUr%vhR^Y)3qX7%Tq?i_tOC~oKFDt#W=S%jnd(bpxzWu2X`ECs>#k$ftR1> z6WHxf(|SQi!1@S=?4CJgD3t65P5CeR3&Snvv5;?3N{wtg;QJJ36rcl3yYdstOENKpl!FE<4XlDm7X|dG=JGtKQ)2H~DoDvlU zpJAj0V5qk+4#T6r&Om-UBS-TCvd98rBg3~R;7-_I8nN>nGUYLRgLjPXhjO$o_Bp;$ ze3@3KRGoV5i52G3(EenX!H4J778UW!k1ux03s;zk8wXaR5|*UT3b~AS@T^(6Mwh+} zQcXOLpjW+bNTrVXVZjBMv3QL}l`Jr=miuZ?uRN$8I1nAmtSz^Q;k-!g9FXDHsoXAG zzkb45E+>#W`w>y;k}EP<+hv(nRaXJp8)csmz9^CR@ErD60}x0q(=VNsTS>SIF>koXTuYkZcp4@{=twMUA6yJebua> zC~r6$3rh83G|Y!_8p&__^xF=X(I$Kb$-#CW&A5er0mo=+Yc&vTL6Opu)o=ClITo#DSv@OxSoszl7_wGU_yIro}O z4oSeN52zYlaK_++p|DQ)^q;t8M zb?9s=NWxF#h$vi#JwH!a%!P^YZXoxGM#kN%GG z=$UzZ3ms0gsQuKFP{BxiNQaKu=#g-Lc@>z-TI>ngzpN*V11^;N&xQKA(WB5;2P1ccIN16LyEwkG?LMsg z2~cYt(6sQ^yVD>j1)&boyL3>(Zfiq(;ZBu-_AA)K4 z3`#?|c3u%E6|5jrjvQt?y=6V(dE6IDcc~zS(lTZ8)Yc7VxqI8NNfw=0wp8U~#!+-D z!N^n8FUH`9Ll`!khh8(yIed(tV@XCrT)i#Z=TemYvy)i$O_!kJ)~ zbUShidiOgH<#F@J|dp=&vd|5u*YP5*`N0zAs$YNju=IXxJV5`lN|7EbQYcEM(tpT!ss{KAmRvHBj|0?wULq9pECcjmSCd zqv7uzePT6WSQJx$vYK~Hvw@87NynSmP&|uDen>CSM%MFbC3rzxT1QKB#4qa?jMfc3 zp;tnCv7yPdlf0^6bL^~h3e?))B%o|6j833OVCVuAdwF^s^5_jVweW1<9}ID)*~ zPI|#YwDR{JZb&9eL|%2B&qo;tYowmGyB$=9cLwhXvlHTH z54)xDcIhoyrd9{5P~)~o5d-_2BoddrMW?#9vl6er|J>%lC)C8Dy!&D4!~L%6&?P@S z>B9w-KF9|Avm%ZsDu14+ysN!np|q}{Hda_L#xhJ#otakZqa4FR5!4hKu30&<&t1I1 z5OLYge5#6L83>TNBu_2nR8CbIu3GXP+yTUkxd%eiV$`2-6+{tnAV|=~0w79QN!I)S zU{BCbS!I>nLWgXD^trRnzv0y;BQO}iq=Lv|q9#B3`0zct={yHFDh|Xa;ew}gKfl@G zyQx*p#Sf=O>F^xB{!}h!C%0%X(8dxSf)~k#gVBfBF*5QRQ$4J>uK&xab`n3lr#zmv z@WK#XvM@Yxm3T*4YiMBCP!KUbeZLT z#p4<*UDyr@Hmx%ZL>3WqXz{9nAiWn7eN*FKC= zA|*gVHe4T@E22T>?stNH<8gG>Cw7cgN5SHH`1c-uJ$@?&p5) z_xbeyz)wbob1tq}=UT^ctYhU^ipx|!a*^ruNH@H?Ri99W?XRjnK|eS10vEqvbZXi| z3}QTF_1bAoaiuft-qhL1MxFLS=1&X*iTHK&e0E@{r1zT048QaQei?~%cPc(>bI4|PAgp$53{TtxsUx?7{U6@mKV zmvY#Rfr-h;R)o2qoA1ELbe9#Ci<}TE`&5B>Z~hT6NtoEJQ`;UXvXb+_T-cAHEP#Fr zSAf3D)Yh53%L*W30)VSOE7vJ_mA0pyX5N!tCA`V4t0Ay(NuoHh)DpXrK&X6n{*fy( zH{K~YC+YEzaMJZ#l**W!$Vrb>^hecg0M&LR6u?GBKzNlX(;I*+l9~S5oGM1DFS_iJ zVT!y5WJls^set1^&Y&zx-_WN@&2~L?rB>4{xD@Gy9S&z=qNEZTvdeY3ftFx~Y|O8K z-qB9&SJQc`O*Q2X4vHG5NADm-YB=bQp)z8w$WW1qp8Mh3b{xaiVFLz13y`^2;ioq) z^lCo6=x@~;tRVp4JDDL;z%VD%};B@inui^`5jtnuI$^>A`KJjf~QWAi$%dfX{IQ1|dG>7m(r z#x%1gne9@*J|h9qjsoRcd%k_Yi?2F$=}0%t*u>PB`F6HmyALv_aZgpDYN=|Rzp+rR zVFw+Wqbp&bWv-^K+=?!c{-kbvdi_DysAk7N$+&3S9G3kJN)dw)ul#j>A$qME$_Ih0 z;ao$G;n%vdp!S+_wzO_4xwM4HQA0nKJ4NQIXj8)Es$uHhKZ5Qw56S9r& zPuXevC|oz3ouey-T{|9EeWi7DjB=cECB>Ro*vfxA6nnRp^vNWtj3HsF*jXGD4*w?eOw3@rOT6bE z-QXFBA1H9H+7zFYVwi=WTsdwAY!__x9Si7FZiMXJ*bP4q(Eo`cJzG#6DR$A~@M0R) z{vPlQ-3>8vF91;39@-8J1VFtHoZ!Mw?oIzu@L!bmpbq-0;0Hf~gCEg=Fzo<2#mGO~ zKEa(_K)dqpab3a=K>6mwHlkp+)mX?o9UF?Y5eW+TOKhAsRzNarkvUcX>$Ma_pF;ox zpvM_j6Hnz9;A=2gzPU<4ZYh$3w%vLZXsN{eRl8@1EosC-Ft$_gw24EGM}9>@`8V&A;77> z*%=CTDlb=m(Ymb7TFJ~55*n4&gL@T|?U9tNT=-Y}E>gHf&A_XOF3hZfgCTE>FwqqT zTLWzj&rM>PH6AE@e?lH`ydcwWRV)FyOY*q$`6&$j0M1q86?+O9LrHt)xt`SzS(A#W zeYYIdSm9q0tP@mYanaXSxUL5!KOIgqN-CxwFKvsN4aaabYcj@g=EG#%*LMHqXtRDj zG19t@n9ocxnV9N${v~@f-5i0elqUf?8V;zI44LYcsFMVRvrM$+#fenzTE_7Y2m9k+ z>PJ$G)3R8<47qHJ$E7?G1f91+;bw=o#rROnU>{q=eU;o`0^Tl0X0UWCYFMR8#f<7o zg!GZjLgJZ0U7Cr>v`gyKqMf~s5rGit^y=L7uU$BV0~iFrn=XE!B8km%A(?erCy37Z zzOrve2Oq<3VUh>WYsehWK)$tcE$nhZDlhVOBJteH&O0+`C)-HIm%jFzZgQ@OWBT@J(nh*-A&7y(wvBD&v%2sC{N(x} z`w;~Q)mo;%A4=11Y)PT)9IM#foxGhCiSMEAc=-E7jU`VFwtP4|Iu58=lF$>tG~lW- z*j*gxQL6M4Z*GC7RR;FwF{FR%TryetjUWZJfZBjVQGntg>%*3hqgMAcjtmPPO9C?R zH`KW++cWg+(&>-ITG9Y`}T0|>IIj=Y5=bHcw&`w4^2L^hU0`xKUKyFA|~ ztU=f!l{472^e-o%Vs=`_gFVhwo>V;;ocM|J;f6OdnrUo4eFsJ21oFm5o|j;|6DS zi~-8+rY{|~*?ICd8kT@QXQej<9|8y3>IPZeSpBYZ&&7}Ub2RQo0h`o*h{Pg_YaM%^a6XS$QkNN@0LUHwLe=*--~0I|2EZXq&t2CpaEJ>zslv(qX@KSz%0W&pTGd8S zZeZ)se*s%<%~e&m+!lj1xhfd5(zyJSsD&JIESC)T18UO{mnG z|K{?-t6#DMZ!L;LM-4s~>fYRyrn|P5IjyU09-KKP#6_hg1^)}yYIJ+{G;V&BO{1Acv<$dMHp`p};h%{Q$O5B@%+dhXi zxm}(zrxWr+>~A_)nbhjz$D}3i8e>ybLLNtsq!cXtwkWA*DmaIqJR8v1^yoXe$uI4z zao@5zxE4tBKxwOu)Pf?cFB#!HndFh5tZZi-I@4v*7!bm=PB-7zD#*7sDw1L;-t+as z*+ekWwbT2VpZ~gAc&P>afEc8NUsFLPtYoEN!%=%bZ9QIACa&5li6Wj?)|m_yccYgn zH!jeArhyKl)sam6zZ_k9#?)6gSVt3N0|51_2BjP1J+g7TxTnuaS|qaDI-OTmZZ?%V z;zJOGBJ%^i$um7qd^8yE-?=fRVRxlanK+$Ge6mEsCFmI5+Sl=hs&tQ@Kd*Om!XQt*)3{&%AB+M*%w#l1dztg;&=n6JLM<;58g~+l{D4@fFMLJxvDD z;#Z2g@wf>_&g<*wZJW88jaL%wF6GJch(`Vy&NV!ggr#R{oBjd857^Et@M~0gmMOhm zTS;WUa@O0WE?sEvcw#gG5I!$(d(XPos?}fVYb}6(uDv|wQ677$VjyUA6HlipSlb7#zK4_>-{B(tmsaJAvGM@p;MOs98RMInK6?b$F>TNod30h`hg>dOa*; zd*dJEciuLn4RS#Eqv6jZ%=Wl*nsD5uCy+-T(T${`%Y|}2t8t#(lYNl5B4kkdxo-cp zKxxl-kk1hU_6ojvQ)(PZ&(zx+S5v-eAD(Up+qD-Kx_dG?@@m@2K~3HdnA#9WvZELw zV{e@1mSCnQ2~G=16KCfvT3(MFdr$BIVtCkQy+UW_R+m3m&(+7|f1kj1LYGD6T8!jv zr^N%R_BCC|yRF;hg9k^+DxuGn35vh=80OFeEj9`b;2F=LjZ`aH&T3!1p6+DVObzz2 zQdvqaXqJ!|)GbO+w=9(zgX1xO0;SCYA4cM$Z7W+~8kE}cJ4M_h1pyuILGtwCZ z8XRJddC~3=Lt7@CYGHaYU4ezh{)sjJ>8It9R*M==MfYyE1W=t#*KHJ_hX)!~ki=)3 zis$*6u-;AjqtO)Rf={VrgI{-fvhA(2Ta7I(-7k0Haa!<v^Ne5)*B;eDBL_lxvVn=H;HZziV5Y0wD_HJ{l(!#zhnz^{{U)Ao#~Elf_cCaK%*LTbM#oq%q5-2l|98@Z9kaR7P~0$1R~7<3LQs^cyHKWP*U z0Ie&;Kx-kK1zSb_=8N)%9@ZfU%CsWne+NKR`WqHWz%lDEfM-}jyYRYp$rJp3wWe7g zF%k|er`eI@V`z&Os#-EMeQg*ScjG69p_J?0?t|XJVjbGc(tI0-g_5xxTrl7!p_pEMmd2crwOghkE7Iq9zszm$HSyBE(YbKW=YILwS7XQRIwNdcXP)$O#gbJ6 zYNt{)AIsZhV;=B?Bun82)y@6HpzO?FkI?8eDtcy_;oAPO#;90^!5Eh|WYJf1_n3Np zpoJ$>ez0)!HG-=cN;ar)Ly-4_MlQ>JTpuDQH23Fvtg>=)%Z&B-&=6!_=M9~weI-3# zcvfcWu5?=2VYDhQV%}+PiQg)owjGrU> zw6-j*CHtr``mWujCb%-1d#x1oDkuF0WZ1U|sKmoth@({oA3B+vccr@<3D%kdLkrOi1A7u zYj^en4Y^pdE6zSQXBrS>kut=m$5jAf*i++0-6;Wga&O_EM{j`=e%Cmg`z$0%MhWUS zbGwU&DLQ&v`_+-DBpFLdW}c_J!tnO+rBrBcIw{k4R-=lWrkT?Wlq>uBK`dDazTRUg z4_lC?Z|Wc41%F6zLG>{I=Q4l6z_JH z`u$~#o1R0Z(QTJgDX*LqPuRS)5nL`5Puf2M+vNYUd{KdiGxU0%%{W(_wY_lllYK_f z6DfkF(OIGo+r2XW?C(hQgx^!YiFR~$b+~);$ur+?E~B^vV-sQa5v~h#WH~yAH~%(RP})%N?}I0 z8>(gkLfPNSgBADW77}SHD0G6{?`R4I(q0JqT!^4HBdA*b(p6(n_zqIM+y#YL3B0yz zUwNn}9Sm_d*?VKa`-a?$G^ml!({;TCebz`e%JDXsOD^e;ugsnN_jV2m7q8J8Co z$ZP+0OB9=DC1SsB7DxU3D9))!hwzJVYEG5jMpUNnOb}cOT-o1zSoo)u8>#Q*&XMkX zFGrBWsUI{iM%=tpMP9aeZBwg>vJGmOa6M%lA;Z;=KrZi9{;`Ja`ta} z-$wh3k+e=m<^opV;n@Vs=a5_ih_RmyW4!u9p2b8%?_P`bjl`;*QfmK(Y(0Gmtnv{j z+X+yF#RmJd*4Pf&`i|F_b|a9(qV* ztY3dO`>}$O;3tLv*@-xrRoxW^ET;mim8SU8wr}o@=Gk&4;jb|#JMm)gh)F%nZ4P-E za?R7CJpFTmtQ>=Eh0CoW)7Ymjh>AzoWZ_&(VKz8K3zH7ZCCgE(xQ?&fiBb&-m+BV? zy6W)R@aV!Gv_7Vcx?<>m7oXOGOvMS)TOmJ!S`VDrzpYpb44Z-X#RWou2k;JVhr`8r z$h$6WGDr3kY*03jP+Ulu4NxM0g}0w;IK!C$B~h`qTR6%WY+M^4c(2|oLf-^sEg!IM z@dd2w4{QU@bRa6jxcDlN0zh?C()`Tn!N5<9_ZLk3XIQ7)uJySasDQs&U`!1+Ec63M zGf=}wvTEi0UjMEvh%o9rDRJyJp}X41wi37~mCTIs+>~ujY^7YzT%BW~MmxCs@)g9V zSXb=x+?YW8YoCo7XPq#kpi%dEqN(bwH(Q&&TFve!Miig;+LH`kO6;e5(xJmzRanO9 zo~*mA%05T_;g;bS(=LA+%I7#Y%oi6!9JjR|m>=RNH<`Y}!;8TP-;QctIz3foA+&2& zZm9oK_CaV^{<>o1V3rmCq8)^KzuUO*Vd*yIn$p^MXgB&Z83NsF&KUUU(dXi6oIan- zTM;mURUxKK@l2PJ7K}9oOGX4~aS!X&)XZ3KuV_l9H_|#QLi5o+fKD0_6_jEATyi=7 z?{YO`Xp7yFE8*k8#0^ikch7m|hIlJp%3i}bTu%9wNFTi*(pMsRo2F!}2}FwTwh1|n z(2rl;-LHLiegQlrOD7HVc4e*p)(kV4vTZ6k{E8*a;aJVTYEk5 zSceOd4 zpU!gw_oETZZxne!R8<4mSoRcvO8Y=Y;PSC%|8E?4K~U8U9_+h4y6K4ij1}Jm-^4)` z4ifT?LqOvQw1ay|SIy6K6}Be~g(HcIK?W12nob-tL5LDht^4AuSe)FmhCk8c$M~Yv zE;Fm+R(Mkt(vEaXEZ zR>)r)tIIZ{yHiT|1v*hmu7RLy6x>CxU8jzcbAQpll|Nrx9i}USzq*NaZM5Qj@fP0q zb}CzRkEoF9e!RQ!nxRK|CLVY|~ISF(wwE7;b zBiJ5bS5Eq;s${M!?;Qt0XOaOx4xUaMLzvQ#-L=V;3%Qo z*8Vo+;$%``o#ErSwjZ`Vld}d820B-CJqy5v_s{{`>dAv{HIUl`y?~9h_M*e1%A2=(LfBh+!h)C!N-8hz{;py4 zQrx*G`<{iC+`s!D3zXX`s*#Z|Ph%6Z#@nyfDxO>)`$(kO6kX$jM%soAk+<8A@F8S8 z^sTM)*6-Ut<_>u=Azb)AJ=rAA`6})-6DEf0>MPfe0LU&(e|9SlfsE67%X{OyQEl;c z@mnAKjXAa`8!@q{2m4Tr<0dJl9!EB{5fYCD#4U2J1uwYgNMo32L4snzS&vt}BUA@NrPvKMh4)E+eoZxrWE*{vv+7vaiQ1Zl@b zDm%0v+1_7A_~2nU;gr-e6-+^8v+jP6RcA5h3uJ#ScCClU$H}L=g9$h4KH^i-VpUi; z2tNOX2;OMEAc7lC2wd9H&8C3(ox9L|qXeTpBTM3w-x{P$TYYRe|L_#G{2}Dw&H4>gRU-@ah!iD>#o0}zIc;9`I3-HxpmnF zjWGF{N|;e~(6@V065d1M-yTcpm9;^~<#QdK7i`U;Hg5*7KLuOF><@|UlhCB_bu0{* za0^m8vLo<3I8t=SCfEXPt?e4dwyjs5iS8|2QK2wyw!5?uvahu#;k}lHRDZ0kBhZNS zd)G;RUT?s1DN4}$2dzq^{#Y+@65-@^i5ioG2(qcQ2)B`#sZ}jRL2%w$RHMDSyE82X zgQBd#^P+71$CF`RS-5zSmGnA8YCrVQG;L)*d+WpI4^!j0`{KXxPVg};&PL@XmUD!m zl#tL$9VEVc4YY+FYD>CG)ZfSVk~ovBcU^?Hc)*(NGnDQL6w0fywf!peInU-O-|Hen zFEK%t1@4=HX$2n_Jk7>6@MWniS%XeqYz9tJU24O@!?c23ZQf04UROIhHC)pw)k;*x z8Su+flExl8k*R&l9xKWaan2#bp)-ux=Z$q)fcoz`@bT@ufEfhuh$3t*-Qu7F5Zb621Y#Jc);eVg-`KCv@Dv*A6tl;eh%MhcA5Y z3mlml*s!Wq0ko7r9y`xMzM#B&X}3I4At2=jTasfr1aZkHL)Ig4a{-G;aAmwEXJrev zm}1_5yb216zVgEkcw{4a4emcyN#2!AohPr_*^O)7xvi`9*vkTA+nP;YWQEpfWxNij z>caZ9ea0WQ$w`k6ihWn-1O(@SureVzpx+#7KobOjjvNg=&OwhSXp%v4YvM*yV+uBI zjM0+-%b4kr-)P`Fhgw)#%UZa|xS)nr9gqz)W&;FI1!nG&i}vfHM^VGk+`5FM1vOVU zivNH6Uq8Dw{sMG>@GS2P@T4O?a=8hxu=~Q7{0AB5gOUj!-?JNy|UafLBq@m~aSn#3@Da zjT+0W-0K-1V`K-_hVBF9VkT|gcLy%4*SF-JkYH9lwUM5d(yiPGw~!6LVOCx&{%|~| zQaW}2&h3&qPf39h7M{ogJg%;6gf*t7_}zri$7Ycqj4UI?X)SNr)sh_KPSfq255se{ zbh^xTTgs#F@0(~+(Jc}BNb)l<&Kgd~Cr*@kt&^jtbp;VS&zLGWK9YStsVLxV6Brgr znrG<>l^^Z_X&E`}wO@u;a5#sTj*$R@MuDM^Vk5!#98N5W zHxt8R5?+~mI?lVZN|#Cv+zaE?ePVdWvhuDCDYk(IPGi@l5M?W^ORFOPy}Vuiv%D=M zKEauSjZmN^LG=pzrFgZP0cC{13c3oi$L+hbjh1e>pmOPO(OxVhN!-As@DFn!I{|vs z`(6szNI4vp0YWw%FL0VLDLp9rU*XyMx+InII{7z_c+(O(*M|1<^a9_G&C z(ooVw6lO^Fyy>#ox5SLnHshn$@_n+)9rrHv`e@^uqEdsWvGe`UA_ot~zhglB^amS$ z;hC4TURRqBpjfTvq0P{gBoaOP$8(?jwjOgpKH&fWy@(PT%>H_3?=1K&5IOnE^8liB!O~w zioA|(dOB7bdGyT-lQ)YCA&>4qT~JV|Ehz)sXo;^jq0*@tZZg%|Dr0_gbxeQ$gB+yZ zEBmo^mf&}mtm##iF$Vs3*|t_Lsn|&uKJ4cj9q^j>JHGj3tNxS!WxzmQ@Yw4@W;udB4fN1V1u z8lO~{5?8UxvrXc)I(#+WFC%4_0U=DM7#ui1=S_K%-JjyBGleFkbU%5w()fwl3*q zWWq|e?X~EgS%91@d;%_HYhVP%$Zp(eC_TCfM8|tTTCRK6G2c=nO-me-wrjRyQjC1( zAVxgKNMsnRG}_MhZGpS0p$)d*E(T;01F&7LdO?sX5`$!~kRKA`z#V$Pe^NIU<_rt@ ziLnv}wsKeo2)a=Cj-X4C3xm`dm)3pX&90@7ci-~}n&O9yr2N)qEO9^`SA9VD>j!Xq z&-u=yHbkDRzJysQ)bhVa&9K>e_k$4oOZAMm+LE67!y2VlpG=Qh&aS0D2O$zKtnNG? z34v_#2)^m8v6?y(cD7GkuJ|8q$;=HN6GYpbIiNeTX#1jn<~PbzJ(n7|0GEQc*9G4F zu-a{nAEn1A)AT2Lhn8Efs$w0q*S;F0JQIRYcFr37AtwKXc)nrU=dh7vFrgMdrgu6- zbvw@XL~NtY#{H1iq{L)LK0xS;V6z=&H`1GWS|%%M#pzIVRIS&Bg{Qe?Jj6-M7**|A z>vnfBa-uPCY$a3!5ejF(6*5f>IpmgYN@g*SEM1-Gc9*P5Mw*W)4ZG zqloaV#Sd>Dm2BjbJJ{l}kPdWR9?j^taIp&5q? z3Jz?uG{(iPb4B~b7r%Yqjqu(XqvXGtB)lsBkn~eeShGS||7L1QN$@uxp*cJ0PJYJd zPR9Lc2_T+x<2~yFcj?vET~@!LF3%Lx!L$mW8^NqRuzQ+}O#ZA@b{GhFPbkhhbOvkf zh=KxQmK2E(bbQ`EaS z{lk5bgQjz+Us_7~f8oapa#Oq^5MBARU~A(4XpDv|5(A=V)cOTD!hl~XSd69tqapt| z^Do04g;5{wEj2QjA-?3h#8DJIFaR3^Nkcd0C0R^8JnN_X1wC<{z>B(>J z5$ZNXdp?Bi+y{mSQ@Zhr1lMK6Zj%;cj;iT>oEL_?__2AY9xtd-@Jchric$3eO@?ba z>=x5LclgNx$RoaHf;__E_fIVrNinvlDPTj)DP8&=g-t0;d@uca+xHwsrSCOfxKYAy z-s^W<)iVLVEfNbkCI#2jDx4D>GhCJ)brpu@eIBPj@OhY(tGrU&-YIMsHe_xWQkgj>rW{s81P zX`#yVv^9l=2#@>+w4PD3?TKA%_F;9{f@l&h0JtoJq|Flx zJGl-m4oSKq3+G}KC|vB=5lLo&guUNK@-%Jf4acjAkDm^pedD*$xMiH6?vEU4`iY@k zcuvZamSyK8X(it==b<>~7a%Ex9vg=9mInva8Sqzzn67h1u8QWv|Ndp^-#VhDIE(BB zX%zxdY(>WknsH}PEzela@{u}&{I^0Rld5%sO^F3LSYU|nwhp-7YG;M8N|l`ul+`Gmh&A~peeLDMNXBpnZgwO0xa+U0ruGgt zth5oM@E|A&-N>HwyT#4=$JfydF9z>gLY`Nl`)ygeSU1;MN3)1{m@i8#FvJMW4(Cmi z)LIjPffRNVxPxq{RXmgN+_#igkBK|nOv@aS-=u?E_7QAW8f-_*`-ki&yI^y=3{o8u zi3s~=;o1SJQTnETAbeT=Swe#>&KcCuR6j8$T%ZMMF90Wmp{gs37IXfWZ9i!H1JCfS+}=~b z0-fT0BGA7ml|3#>xPAo9Jn83ARj%I zxw7&@1ppcmpmpSoenTzqm*#3cPTTsKkx4TXa0iN|17!^?pZ>2mYN-r4+P;6;%-6xc z9J4TSdgj#&v;w{$guzFDVrW6;ZuJ9E4rI7A>+>$+-1F;yUPI9ZiLRgJ1cthQzWXCv zLpr7YIS-`T6RP%?DQj?h#o-Y~H$7T3mi`cN(W!$_>I1|s9sn}+z&ERR3PwspE+8|Y z28Jv;0N{7>LWD*Ql!}E8h1Y;F#M$}dndg-iHP8IntQP(HBCfe;B$9M&YYqj!e;tx5 z@r{r{zUmBq!mMR|?o$8=Kg&AZJrA(FXg4uvbh8JLylH2uOOwL235$kTKiVVBSu0}B zQ+d1V7GAG(AeI&=w-GwuLe6L{{N&rE{^4h&rKb=AS*1z=VZO&2EtS`;FTU%a&Gzxu zmmjHTgzHm=XOe~uS2ilnrRZV&$P}d{&k|*{cB=`um1@aN%(_>t?(0rX}HX`BmD`E?q3v~tqrWb z7PdtUYDmGU(T~^8t3etH;Py8(DCRQ<*nZmz8pV68n&W4cKfNxi`jKEa%Q@txs=Up! zqsD?6_bNu1RXEyZ66CG^z_LnYiXOFF1=3JLH0hb_1aviCTe{zDD2O~{(TYye!i{?i zTfai;Cx(y}3>XXe(f|6pWfJ6ZHta51uXu7@RE=~*;htB4>#Ey``v2(X{Z@D}1ZSprtFop>2Xv%8TW8lJIY(Rs$%t>H^TdtU#== zs=udH3uMkZ4Jt`81k42_qnl^!CkFu1XfQ;v_)=II+kNIeQvksXzF~caYFLo-6qk#1s#{7W7JmTW zP=c9`!Ftmh&d=!Qy!iLu;k zM<9ZoRN}yU# zB!CYMXCqHVs=Tc6UOyWO>b8M&08 zea6|*_+U#~I24!Ifd`h#Q4c)#)#_&q8s?zpTGA8c=Vt*6P90d_?xLX!X+|)Yk88M6C2eBHeke+i z^*df95R?MH6h>dz9D?2R^?H>e)AsX~p7n{*JoCVe3_%{|Lr55aCCdhF#=X5|K`*Ss z@GzGgdgE%4tG$IEAs|gP%qZM9^s}R_H@JwMvbsc}F!% zeoLkASX@5vp^vmpq4~ovjNWnfczEnIwmt#E!U6pG3SXvRKb|O@A>ngUV^$5Tx0%k* z6n63-*3f7=aUg2%#IBaFs1K$W2l0gAt2y9ZGiDiNHu)-K#YE;tcX{|m8d;R#c^p+3 zLD7k~9GsVScP*$42}+AAAX#`b0AdsC$T=1evNV{ZNwqsz7ZnGARNDeCOWT6i&2u)m z)7-M0Zg++6_D}$B$~vc#c!u`sDS;Y!&38qzMyJW|vWrpfQPPl5tzXt5!=9>Znur=H zn0>q?!`%B_82kR~gM040hLwNTF<^(^H~*uKxfnDtb8>APG~^8cw8@|V^ZWk`$0rCe zWRWljHVXY8#mmy0o3M@>K;=@@4anE`!j9_ny9oBPfRh*4bievJWv9bB?sTCmK?wd7 z9fy+2wJJD26btcgk>*z_9 z^m6Cmg}3u-nY4rn7Jo{@uqhj8L4U1jCX_d0xPL%p`U_>6E7yr4gUw^r4+W&>o=|}TrCGn6*)tPh^xZGZxr#Jg`8j3;v zDBWL*s+~U6%*EBJmN(11V%$F=tKIl+`XJA)1eCG&YU$j3Xw0PtP7b}8nUsaH)Sg*` zJ!yA-V#q#vN01OciR;8Mke$mdpWoCTbL=k7b5wt=zWzt1CYG~p2f|2qE;3{4=H8`= zFq?Mojc&JeGRX~%qA!E5OV5I1WNZ4HrXOC{H|t#cacOc;M*cbe-~3;p>Mt{p|G@>y zte|D+M$UcmL2~kXC{un&h&|onsLS1<0%~`M#5HzgU?R#Z>eAkny^n&vRyOD><^d`f zVeW7g-85`>d?ktEzTOJ!bHT)_0t2Z2p1`sGh9jW!{2XE=(DW{EH0P5ghHh%uFV#RP z{Gu6fal04pfLQgG1_Yht_;xhWHslx~`{^udCx1ej{U0r?*t+ik*r0x2gI?WfQeV}n?AuKc^#3#Ac)!F2SFFON^|>ISc$ z=CejDDp%6nSvf)yMH+QgMK7E(7YzC%WO#0k*4&{wRHxz}?op}eYnBvfHz@tFU@eHb zWyKlhvGF2sGQSW@x?0Wgd; z>AyNy2g}NY!v(2!3&j)Rt(?M#L*>z*r`Dk+Xx)}#XS;cSGWHKkVHLmyg{WzoKPs zbvula&(!&n*ulNXZA*g?>AT;V{ha;HF|a-D*E&!eo#{n`Cfc$a3EDNfkTc0sDe2O- zYL2*zeD*!Z^`3BKrxj&+Ebor4|qD<%l0 zVI!Xga;p7a(T@#OF+Zn8t#}BnDi7DsHkhoXY*1Qq0bQb=*7p!XRm!aiw`S#!CkX|!d zY`EdD6B3b2_Urd^b={Y1OPWVctIq?Z`?o%oQB?b9^mj9gym}F}Xh}u8KV2vPC0~d) zZFEvhxGbadrE6MfZM&eB@<(E(M0i*EgH*=oN?ON;ceT=4I_6p`>PVZ(E@yX3TL;UM zN>NEFWhYGf=kxiR$EoV!>g9b~ixu%^Wrl{rEw+|7q!y`E{5r>{E}g0(l)sj%kJl$~ z$-FA9h+Ur7XPbW-*BwVM@Ho*o!i{9O5RPLOAS)-N{XHxCJnebKW4zk!leOionVNEY zwIvLGR#U}cM}-^XjX0@frE2lVWumvHB9Wda)*noB3EabF4!*5KH&SCcWlUM#yJ{sd zX6wit-Gg88=6s?gtaiT#nv$IuUFLN^8U~ghKxH5NOYEtN!3hHWFS^a&U1PWYbdCK> zQ}-YKyW(kXJ{g zS}Lx4`OAIc^iCu= zyF}UPniDhY%7JsS_w$=_bA1IkIQD9-US3UfCIf}BfW#y_VOyLjqxJ2Wm`_Uh$$2gt z+p61Klw0Wo^Xfr#0^P%k=CrqP9M5W3yaZ;6hG?UvXLbAMs(fY2!uAZg``d)Zh@G}i zhpL(LU;N)Bj@SNft+t;bf-EKgr{u<+iz2w1rr~%Wj;^ZMD{!7`Kt<~l0doNQb1XFm zfgpf!U*OzV7aiD7?(4Aw`34XDDOL;+HUdtdQ+gNZ!HU2W#Q1;UfLNlB1A%gXs^R=? z9@vcR)HPMP=^pp0m210?dW^mkwE>-@l#33 zP8PD_Llbnz2HNN|{-V=$-frr~SLLn{GlWw6wgI z*l&0YSS%!4#0y&YpI#{t(tJ}*5zL;t7gO3V@{LLW^D3l!ql?TJR>!ZeGe5cA_$)EZb2li;n%ulNw3v11 zdz(km#GzWN#2%;skGh%I2f>~qClF{}YIcr4Z01iPfz2|l)|WMk1@ z6I;9{hD8l_f#U<~qd369lgT#6E1QjTIfc`bbXKJ~R6Kk^&NxfX4^bZZp*r#R`$(5| zYGPRS1d$g`toHyQ=Lhx|T`=A8v%p1bR|2^A36IX5$^gRuZlhNywcz5?+-TF^V{q|B zH&#GS?`MJ5#9v;0!aXke(-LzZJf`Kj@d_w!FavoSa*m}6`QLSf7*zmh!DRqsuBco1 zxj!hv4^lRMolyWf7A_4!HaTFwX$Vy=N&QYEd@1dgX>`cthEBYEdgaRFLVD(-T>bI* zU%-Z4V+!Y#l;yY^(nN6uaq(tc$X$NjCmI^LcdPI#w`j9^E-`$%m$e?%jMOzCCeURa z4vRKl#xWl^yS~`N`=DNda`YLluU-jo@s0HLgaQIk9c50TIH33>ed%-7Dpn5eem?2U ztDDMM)9csFIW=c=-W_rckvV1;Fh$-m)8Gj8WxtPiB1PGLD>)-t!Ef7&3jsM(;y=iC z3=OJw9PGZPUsB~VthV$%f)C3(ie;E6GxM5Qec7!;990nMB?cV^U*x)*O(%d{kn^|{%En1TFbg`0=pbqU|lQ<+s%`}=3CdrCKb zW54EoEl4ygl@=qd_>do2D1{ybb=S)21o|fi`*GiSHb7i^4=c{m`n%am&Q{H=Lx1P# zFk}WCO#B(1IXVGjQKt2^6u3idroDqw10^X zf4iUN5^L}31_Vq7Z+MD#(b#kG*h$~im?ivb$iHy&u1rxA@nNG9ZtLP_M<;5H?|s15Uruh+()(c#9Rt%`?Id|tIwWZJ|@zuYb9;zxmR z^;^rrpSYQp+m9F5-%NNLu3ZuRT$?g=Q0LCWl%}MdtG25cEA{3x+BdpfsyIu3fSTgJ z%yM(vxtQe!K})g&B!5&Bb`VbuV#iJZD&!3(sG`5X*0{L;>Q_nzCEYFP_UTddMIXLh z_Bv3IW7h+A{lZ94lo6UABY<*nT)tPcpDZeZzk^adQ zhV=)6#DC9yQ@@Z21641l*a%9&k>x4`g;{;6XHx%vc+sKxIKYM|v;cNTl}QTn6D z#2$-#*Z@3uOO5}w8a*&{$SSFcxY>Cxv5lHCVH^9ZmlYqC0Ulqz>-^zMqsCnl$Q*6m z@NIv7uCv>8U$~;kqRc1EIw?2T}Jz+wBISIzq=k64{7c;KsJsGzJL3mqfXf#`=N+e{CutU&O$S_ zX;at42AGbEA{Vh~(4_hP@f}SCEvoU0Br)v|+?+LM%Pezp^*nFtd93$Y{1{(+!-w;I zh{fZ(c4kmP<5K6YfM)n~bq_)}**QByrZ=_Rj;dxDN)v(7vJ$+*1KWpeWiAS&J z$&5L92{)x%ifd;g<0SoAG~eVVsp)Qx08Vt~YCKsoJpFTiqM|pIa?ZtJAqk8+-)9=PN=iP!SuqQiJ?rq8>M$I~VS$WwyKVWOC`k;8%l(Z%@W`Hg zoDF{sdR-Ofw1qhv8%R9NiiCXSONxA8?6&q-Uw!!0%SSuwK9XrZvSU$gT##a+F)`Wl zrA^~y%&W*&46T0!-hWgqmp2r?400{XPRP2)Je|ztdi9lW{V_|!Z{V;Epsl9Sspt9AJzRI8|B~xmEAK-Xxn~=;@j< zGbt*OW7vD=x>%e1G^J_5BWPh|E|ERv zqXL8ERykW@1$y5(Lc3&17~LH5p%0^lV#f@WL_hg4h4?1Y&Q~z*6h!Riq+S22CKYoN zKaj!5^lU<==oY^h*H0w?4U-Obg=hm4*ZA z7E|R5U#p8RoePcAU4}OL!*9)iAdI@OL!m#pfAbMo(g||Px{B!xobi=agj@ zIUD-BYsBZCV!(QkJYO9iavW*Pfs#tovWPdzcU)P%GPUq%vXh)>`i#;v(`~NyuJBY{ zSh!`%S2k{F-F<&f!T`j!EIH>Wa95O!d3Y- z5NvZEm=R`AjOA=Hx#s^z-g|~cwQbv?g@7O+2$Hh`l7r-=fJhdRoIxamfJjc3prGU= zsDR|03MA*8qh!fZvQ!aEC~CdIN_)A^-e>Q3?|tX{-g`gln>9zxTyu=kd+V*Y{(&!X zkgE9`fH&)fM6!>#syK zmKBwvJ4U(2HPs@H26l?8nX7NV%t(H;h%n}Edb>;0%7+ZuJE#A;tFzW3HjRrzznIUrl}q8&Cs;J9u8p_r zM5~7VkeBb=YqokRgi>F|_2pgyd@)bv1W~+Kv=)~<_(0y1!4!-V8MFIqQ5=e zKEipQXJmgjTP4c)xpEaoSpaTVmX8<}W$L96grAfCq_5Dj7wW1*5rwLqmd~Ag-&X4@ zaaZ5mQ|giQ{di}4^6*S(^zh6_D8cJB+x3BDS*AvbGI}LJ46D-eNMhN7ZSjdjS`ztF ztsc^dcC$X7oseb0Ibo-_cUcfu2Xpeb+>Z?$HcQLOKiArjDf@q zCU#DE=zgB>BirFT4p!vZUPINr-Ew}Z<_$T_{{w3P%b$34&K|W3!44H?#@vGEEaB6= z@S55vGff7c6JL6T2a~pc@n*L_j0^VF!N@j6qHmlDlAqk}V|y%2fkzRxdU%JkIPQm_ zM6T)A4K>SUhg(|+!M$A?`LWv-%OMZK7kzs({p)DY&NKxhiDwqR8mc_|GBe~NS<->) z&~Yrm&IC6~QT|qV_{pq%fI>P}!A7dAI3r$Fx`!$_j`CAa&|FPjz-4|H^$qm(@x$Pb zY37(ao7Gjd;cNFCR;;a*n8X$dR&&L~j>vKZ!?m)ewu?}z>Bb=M9w^)}f8i9!ud+OA zr+-6-NYa?{Oq3hvK?99V8Ob$~*Yan5vyn#DNv_2-58;jZjHPACL)r&iX3@Squ$#hX z^_4Pffj=U44keuDYdPsi_UGzvpu>n=&pcF;_|ZEb+*e>yBH1+eKCAcM3*0@&6-9b9B(|i;R6U;EtcOUktl1 zKuZHy`?CcSx-$$Nq6E=oe}%6A*h(e_@#srHkKyK-Aj3hc)Y+o?v5nOFYv@UHo=qxl z;{Xl{9*IVh`XvzkSYkMba~QB);w=LMwImbJJH=+OdWH4`5T`F*e8_Ky2Bh0Pf8~;U zVN_4#n9}gDl`bj)|Mns4^&D-VoMs<6oTHwc?jJ5Wkg49lDPD73|OLKB}gn5U9~JzJxjhSm|tlC7B7gNRS7>GBW>^2sKOm*-=) z?R3D9REi6YVi%TUvy|%#>pPrZj;Ovnl+Oeg7E$kSUi(lqK28umW%(9!OBLcZm(uQF ztts3{UomPH+R?h$0VU1t#J3rx$zBZ53J?z4S6knFy37{Z-G*0V)rmu=>ZSDp2f-6x z*~nYr;(Z%cG}?c-qW+tZ2a?-A5_)`U4V%&!me2YzKZ@tTRppQ0Bf+j#c~ z+lS_U$>?t3vX#y5vfjHY4N{1}UWd6YnqZ{xy`F&AHED17Q{n_U&J~RL%H;IrBV9B`Q!Vt1i!uYL@7)XN zl^(=cNm$wQ#dy(32cOy0oHq1~ycZ!D@a(Igs63OO-A`eSXDzzHcYxMv7M;X)A^XyR~*iETg&1^Zfs zi2b`?@yMy=vJ;tO41-0wFO_=XKNE<^nTLImQ}LL^{VugKZF)kqUu?YuL&d5QLMLCF z_SOR2e`Wrn$|7ja`K>t&+WS)cPb+EFmFNc>8aNNnyFjXW!v#}}VPC!89K9Ba4)um? z9MLE=R1hn$1y9_M){lMGTjD6MkW-|4cA@PogRh3dbV7?6xN0dUSt1x~J|3$Yq65fN zke3%w&m%tr{p)LhI@T-+~V zolm4bczyvp{XZRD;jcT4rbn8ol^B0au^0!W8*q0o!RXk<_R|gquIfAN$M|J;M>@Yr zthl+kkru6a*_F0F!bn;VXxG6b&s2 zKL(iQMO~5h*3N0$!pZSg#Qw@I9~p~o@a>A)#TW_5?E!B$1!pV|ospUTwBw|0OGckI=c5P;37m_Cpp3t4pgX~=JCXf7*<#^f30P1ss3ZBEe!v^R zXnEDxnAGoy%u!y_KlaQmCjYj~%g*I^BEUQU+W+K@_YPa;c25Cj`#=FU&{Tk~rhqYj zWC5s$cNx2Zh9=M;^?QR}(z-G5n=iA5HLc164X{^^&_hjig`d`>i4fMP;95}ZVI5>( zI_L_f{J;c+z{z?jC>!>I>kp=gFNhQWt;{ni#`B9*d5VYdlW;-7+R9NOWBi>M;~^Ky zn$K}bUNHiT(Tofoo5qgs5)y1TYU?LrO)U0&6ca?dhbRg9y?w@~f0cZ>6gW|pt#22f z;4ps3H$S(&1Ftj7>@xBmONh_0%=jwwEzO0l;&W6H^BV}sGbv3IV+5zP6A4NQew;$q zW;h46p6zoGPoiAAM@96FgNWSoBT518jx_07wL2DhW@tygp0E75SZ(--6l3Rc00yXB z+I_HkOMI81w|;tWSs0dKWG7CTl_Mt->^oem<&G!`Ib!(U#2iI|<8db!bCfNA+OvV$ zuX@bna>JD5%B48fx#Iw(uCd8$^Na}7;Ci|IubyUyecVz?i97qV^X>cP2j(nV?3C)G zcWf{#D#mtTj^B43ip}GFd$6{bqumhC~1cXTml;e7N|Rcl)y zS6bZ8u^KrdxBql&>!Dr~qJ4L+?$rVUrV11QE>r-1{p9jm3s+K5V8~LYTSHEoZ&m*S z0bJT^>4^FFoC8m}jwo^YXTCF&O1VwmkwHIsDdEkT^hHvIr)at9+Q#0~=X6H70{#uHNh{R@uH_J2 zu{^sMT(2I*b4#`9h`5I+yq8SK3r;ex{h>g+UQ-+q0$KB=%dtDI@;*YBT4N_ohi?r= zwK=b2tCO(KPdAJKb;aMMg7OVo%WU3nh4gn$OOwefNvJ^xE8IgM2Ps!PPwtKS(xgoR zF0^gdKxK*(L9Spy_VhM&06!Bb zq`H_=?e9vF>1dM1a=jvbb3K>`tyghf98Jj(BS1|X(`$xEJmZb=_25Tg?wcH2RVh~P zR*ap(>r4~1{LlDpAnCD8R70fY5Owv6c6R>OQTfkR9Dy%EW%9tqCgQ&3PmnDE!%PAm zt2a+>GvThutWT0+41f6@IQ|1b{%1?wcwZ$1@7qIO>J8cZ9`SSBR<(7XDcccgG;bcJ zJy#fMcnCOGMiA6fcnK5Z=A`cnEyy07XVwtNut6#c;(v=Aw4r9t=)6}Rf_u6zQ@l%a zo0I+t(o@ayyy6tqtFKkJ!;0sRZ^;lp7S@ zca96_Q_}}q=6I6tTiED(2kBt=qr`R%+$o0b+^?BTzvp`xpxD>ht(^SE9-^?!*;cO= zwPnILCaIFWxo?EDepnsomKkx6~-} z5J>A}y+^9&^K-$onc&O@i-WsE_KpYQIVT_3K;mmJTpNlv;A|PHCi?f3dfFNPLjP(^9*Ti(?h1f zHhQl1leb=ZZ}x-5E&Sw~!jtBGHq6O9%BW2L*a(~=%~9Tt@1d4ZHofq6;$C@Ajb1_l zKq{@K5@~5^G_xxYAwSitI9IThx5ExTwZ33jeIqk-V%N^udhaXJj&*| z7Ph8Gzc`F_sI78Uqf1MFx^qjyAFBu}i_l&CNb@_#+;8r)w&9%e&_m(n{bf; zL6Pn`QlEs1-?Enfpw$1XC0UQJEW)oVpk9JJ$Novx=Z3LoxGKb-zHUaKbh6&O-4$}W z5)M_nsYm}cN1HWDnmO-~F^n4WHN8~wMzPYe=@r}B{*poxJ>J zq>pg#ToqO_4-#N5&J7*^ut?PTl1|Pg2Kn)QmpeT}1m$W~ZAr}DF2=r$4g_N=!agGv zuQm=o>iWZRpkkmq40Fw(-6FcCt2YB?aIi<^8RK-#(_mZWGrTJpyM2#v^j5-F#?#y` z)+)$k)|X#zlrk9vMxtrj8IG@l9EUHeau+(ue_e-2b)@Xy+6G`X2sz7D7>z(ziGW9F zg1BJ+a6E9QOQCvdi*~ zh&um@Y-Hh~h%92-P&~ z=4+Nhd9Tgej@psT3JME&ZQB*5vm5!1tWb&W@MA-qQ}EaBvWyAwUf+HowRA)0jLCqX z?2JI#f+KJJ+m@R5C=%x3LJ`EZlUnU#b3`YPFA<#Kkag4L;PLU`ZRErccu@$l+B5!2 zvBRDL(M``~OX68xhs6LTy0M&*_Go0<;->a*P_HQDXsCkF*IUS4|(bMTEC{G;VPXj5-Fe9 zPPN!Fb-w&C9H+h`N;Y7Gx;y+`lGS=?On-NGXU?LkIgaL&D+>PD@hg@U4j@De?-szyQTJg>!9HV%%a@?`;tcV;Fni7 z*|#v=AZOYkY?q4AAIK}Zi#A(J2kdOz{hPr$lJre%l1is#W8M85TC-$|kDf_fayHIV zpD6zzG)I1*TUQd51@ zy0mw`$%ldGjcGDz^bNLHB%=4lbxz#YF-_cF{`mbP@xm|s5Dct5TK^$Rp)YJ4spMSJ zoobrW`2XaY;jkhk&`R&$LbnIDu8{;i55Jln`CPT(sG(jJtM~cD32p`f z?jaMc`vAle#SS8w7xLy)-rGMy#u4w|x~bs{MdZfAhtjwbg}9ig(lnc+6-7CS)I)}l z+G*Rh^!~Z5Z;K}I&pzEV48p%z^0?Wz%~rGrC!}gcl~*6Z(h=|PbX_L>p~sd-fHrmp z!+wXnhh<%oIVpz7>w1M;&;0La>JgMr;BY67majAeHB5U-NsG5UF{uO_g%LsvZ+5x3 z59@4--5egSIXEzVh^92}vFZ}V9LYmU$W2JH=P%fp$f_-sp6!aS_GZqk9^hZ{xJJad zoP{`kmAB<=Ymogmx-W0z=M38b=7C#SYn5V_-{Rux$dhjEkQ0;2PH z>p36i7%;s5{cMCegysfGHL`#Xw3dT@ZO;jka6^5H?R0Y1;OlRh;?mH&MN6`A$GW?F z#(Qt;MXZ1m>_nyLpVI_^yHFYApf9rQ+QZZ90b zRN&}fhoy$gLkeYHCi5dwJi;FWR;?wYAbXqI4!+#xPRlC0-b$>sjdfwNi9tGg<_s;A z5HmH+q}Xj~vid7owza0OcL_f};~Cj}G0MHPR&4P?f9iC-$}w)he%h}o1kEFpX3t?Y zQdmAzsbJNAPnF%T?gLSSzcDnZp4WQG&_2=lu1~D3;4Z&m?^X|nqjJ|vT+=72ItboZ z4w4FPwlAvQC%ETX;RYP{w07rv=XKwBcS;Re0r!7(GnpE>Wa*Td6YU@zE!bqK%W~6U zktFqe*m-oiL8m&D=*_jwg!<_ngKv2*MGcuCztO%nnjwC}eYbhW-+y#l4}n|(bizP& zmYHs%;24*zlxUaZhBJ(rFjtAvutzvKUmodoNeYuAj2~K8D~?;fdnr;WCQ1qF=jDKC zERfYyqk?vaFB6?z1{8vg?DWW=S6?#xH>!|YD3FNwn-uyF!16TI9$-aF14+O=km}5K zY3*Oxz5gri&)F9bq{7{HbXbj{gbA1Yx$EL~6O5**$<@>=g66*ef8R~$a2;I5$!L%a#;eq?L z)Tic@Ekj1@72eptaixH(ixIrS`-AvGH2?8T$~ozn`p>J6F#nssiX$ad>NlZ8DNxATPkD}duI2Qm5&hJ^oQ*!cg-Hvf+?rhnz={F52;|Bvnek7BMDqu!sN5?|Tn zIwv9%Jp-kjEs-jUR#^fazF-<$(o)9g^CiRa?(HU!{(b|XLVz}Kh?{%K(1|)RF&D(> zUz@f^dqrIDk8kI=^^m{I1ELFzzs6A#J)9MoLED29HkKR=~ z;~W&-)=f`lA1mk2^w-N{YhH}Vh&T-RcDdT#EkS@ktlTVU@s8rQG_5$w7`{DsC;#iV z!?O^oWonD8G5XhG7I-`54f5QEssX3C>k;hlf%xo)S(gZ{ea<&uKzevBG8Wz9zA)H8 z_$M;|e``c!zk+v#74ls5xdTF~1+PObxEwCiLT?FiE7uVYULL?{k5Q!$&z~`=XQaa7 z{*-z+@D#wCzmw4Z?UGDWbDYc7;jYTL@|J|XUTLE~A=z_)TuYO06f2;~BmKcFWd-6s zduy)ZI$bF75uWOfUatpZ^(z@=r($T23Ow@M2ZSj$Za>1g2ufT$Pi$2+jfxS$zEAR}>SRUbR!X$Tm#Pb-h>>R| zWyZP+aCEFwj-p5lVTS9rGr#*&j$NZu>ckZS0abXH_P+K9*7wZ$4J(9MP0E{#0fi8T)Gw>wn)NS zZ-#0L$GCoLu*IDqZ^EPPWMX(6EQo6jDh6|tDEnrzMY4a;bf`Y%UaZ+{LReJ`uL^$Z z)^KAu03J43qq336iOZ%204=Kp1s_}hj5IePrm&N=8- zRgZYzPUXu^cQg0k;F7nwbkC^DTG<~?7ms5x&TQbH7GeYFGBQLi_ zNqd#BhOrWth`RMR0@@_n=eSwplIr%h2r0R7N2l=07&Jr7=1I+!J_{0wWM^=Q3=;>0 z$>4+p1L23=G>z%C5*BkzCe{#-HxocF=vD?!@LqPzWZQ6!xMK=wwLp~fw}JArXDKO- zuy$1XJXSR0?wy(6oj!!E47*6vJr$lfUj6Vmgx`BdTHuFP`5$Qh z?@OIAuFf-z48C!zqytO48b_1js-%Nzz?X^U&D!4WJ+R?YpRLjj)7V^c6 zzGFrsC$oY2nJikV7$Ad2IOx|ElfS;ZKaU4q?y)&l?Yo9LV-<-7O*@PXxT{E z#;-x~!thpa4IkY?HaHcBKHbdAtSD1$@QmvDROeEctI=ucw0O^I>$!5 zG6$N|O>VQBVVvrKLl7rlt+#31SE=Vq-47Bw|hoRYnF{dhp~p<=ak9 z3(Y4w9p^38OG91Gk2n->zLdb@^sho@U?xh04iVbxh->cz(^9`iu*Rz}yelD4el;~> zV3%ixQ0sWUF#kiCeA=Nf%rf>0!+cMdj5OUjS+By%oTTRywQFm5SkOAv>x)MlPU4cTEU=Ga^2;w&)H z+Y~2+6=sG0@bDm4XIGr54d@r)`;M&(gmi$Cl&!(hzN%nJPth!;@9dhU-8DXM^E+&q z4gbsnEjBknJeZ~kDZeUis3ra)e-EREOD5tGe+C_&eNM5&@{4Gmg2m802bnYXy;xH=8OPQ{WlW%QLdTO zD7mV3WP15>7{^R1+wB_v?bxH@m8H97&Uv0ZoHR$CN8Gn$&0Rnt0rnG@UmzG=t2UT5 zzdxe?8ROI_0(52;fmW%bT~M%%5AZ5~Z!o=RV6(xz$Sl0D_c3B+>DqY-xn_b%jV=`z6l~BlxqmLHZiqQ}*ywuqK|bmo z*JVu1sR_YGp*PFAVXjy4*vu<}`ZVlotPk20p4~hPU}|~d(%nV2t070*t)ZnWJFaIz z9bhcnY3^uQR^{^aY;lhAuBL16DMJ>-OD#~6E)5ym-gM1koJ+&YV$7m~MC6trlbES& zvdd)19GtwuC)3UadA*(F{TP?7TVNlwZ>|cSa`Nuki{#D*LJ+S5K=>kx5C?S&Eh)Nw zc8>?T8&x=jg#kE-8SS*tK0gJJBx`qg%7)qQfKm{STiV0eJlJ|AJM!IiWMnG<+k*ho zZr7iA%X^*da&+EC*D`szS@rOHPpX3{iE_g1^RmsofNU*FN$9o3lfyg*CU|Grvhn4s5*=t(1WT_m zcWq}N*NdP77{Xe&D~S?6C$7PKU@f2qeOEKBu_-vH@B zTi0jVKhm7v5_wvXhWYKV%U}X2nt|itDNJMAd55^E3`50z* z**LD$xV!W3;E+KG)#~y~@}CTz-wmYbabqx$74pe^XSYcuv~EFTjYyKbFUYFsJ1>_0 zK1oQKOGozXGO5L)uIj&VkhE1bTHI7Qwp!K`uxf(SZErA&2T`&pVPPV*YuPSEglA4~ zhzljhSrl>?Jt0=8UX7ASaTy!u7HK^;&1F6@ z^M&A81#i;^)i01(OceW}wvV98xnt_T{9S!#*ze!|TYnZLuL_71KIu);|FK%Xvtlt!ljIn1*PeA_FfB zVL!h(q8>8+k-&Mv!U+-@C`%FQXc^3G%6nSW7Z{SmfH$kCCpFH?vpg)d?>_WYsk+zzIbJ1gTQ$Bq7cw$I;=UO zL(BL-*|dE96i5c%D9bJIf{WUlo@@>M0tr(;e_sIp>ZKy+LV*Hzs+{Lby=@udpr=OgGTtq|<&$odyZCwuL% ztmj)+^OAHx+uCU`MZdXfk3QE2N_rtuKX2&ciN7s4*ct+m(4T(bq|tJC(bq0c_WA|) z_ygVj2Ql#hqtC7oJ6E01XUr!QT^4%#>dvkIvKt$^Zydae&nBAo#sLw6GHZ8{w{R2~ z=aGN=A>)i!7j(Ba5P@&ffuf#7Izdhc1pdkO_QIF9Ya9J-q0Y;@`j#J+%xH!EJ)em7 z&kz3d!(cmx6a9Lz;h+6<)et6$w!#g5PzN{C zvz3|%2*#t#Yv$opx0Wts*}&7f8~Ztk7$F9x$+Zn(EHkvKP>Eo3Qy{VAXx_W>+jL71so zX60CZz06uy?s|gEBs!yAEOzBgZA*d7ad1#=jWl{Mk5RQnMUMW~1Y5 z)Z>uxfVnyXl0OngAt7^mNe-`&=CiqW>RhbEDJS?#LfR%}QP~-2P9kHX{e*HsalHT9 z`tq-hg`Xsh5RPfEZE=mZbSB$yyzt#i$Lj-|(H_<@%FA)96*2>&}fbyk6eTlhyT&Y#H zK5o$#)@=DoowI(6eySP=LPV$d#?S>~D1t20R#|ytg49yT)s0k__nRHVAvs7xXaoXt zK&SXG+nKalo*Sf%sT(zY%!LjcY@?CCW-0WKNo{bufwUw!kQ1T4bBjSI_rDy6gEO>9 zdHy=vI1gnWjyxhb{vXHzSxZ~N)=2wjx`&VTtS^a{NFjPZ zDk(gZ=;aIo$@026EZOFWV9CQE%rNL8R1S2pwX__VX(H z%m2M_Fo{ z6FNd!+1LXu14dm_VSqG>v2)t_(7zB^21%fuz8W=6<#6u)N`AF#2j+&sj|UH5m(0tH z(P~$j$8dhkbie-+Aiy3%DOuaoR+T5g*>hPfWr^KMsiL}_&j|TmWC*RzQ%1^+HQt=F zrPki#z9fPg!eaba8 zrJpIPZr^aNJdkYSi)bl%<|kS#)X1^vuvp;dj&FObq56YCSJs>J&FVn{KMA)%ar24H z*}dFcuGQ*wQc&Y3MK#@V>{0j^gU{) zvsma(+qrZgXXzy(AQesV9rw$f_pY zcuBu1JTKxd=*nBSi(HI-71Sj)QXVItWf5pTX_MWVY-E@CvGS$J?Vh{eJ{e#b5Ft^u zr!e#@9lLS{#1FS4)AP%PBN5G{x@3s!13nU*K4NZZx&4EcUua%U>!E)|i3=o)38}psb!;X}5s(lfM4a(*0Hf_fM6>rU| zpA4FUGikl_kbP59eR~5~2rorj6+M$%mEUBdQ~ig_kh?~betSd7;{quh4aK-e`JBn{ zhyWg%(y!e$OY6QlXlH%b(Ds?)aVa z6QyNhJAGqX&I7cSo8nd9=VSvs7sJNyuo{iJb$@3bb&Ox){$|5$iG$Jg=nC!8mS}p5 zm9-}mT8>lgbW;Q!?!9533|Byf=%gHtIGs};H8BU1Ixt7%hVb_T!Q~<`$1dLBi1b{W z(}@(A9{w)>qLCqm1qn7INe)gf-P!1qNiHh{*6B^f&Qo0SrBSoyd(Q~HHOk07W*Bug zf&P^98oJ#uiH@yMgCQrS;REeDr&*gJwiKS>H@zrXPu@Oyc4t6+oM(3rrM}o#rxb7R zm09DtDjU`7FmpMNwMJdlehXyH>Q4tXWZQaF>Ug}jr5o63)UMZK`PBF-^+ZxDDsF(9 zqcY9!^2Zz_)4AJ*KAzJ`lmtC9W-V`FFY)%;>J?>-rFKs=s>b6c!euv-niMqj}1JS;98Hfa9|!)NC;N2 zXU9I+V9T_^%lhyVN5d{Cw%E3I)q}*gO66JnfglP0tCuUeSP|Ywv>G=%1Ia$!sxhZ{ zdt-Q7o?^e*kyx0P6bEZP#~@K_oU7bH*g;mkT94bzctr8*q8|?P86H=cmCk##f&&h^ z#u!W|#%|8SlM|=!gHb|$wyYnWoSj^loyx;|wQ)ryLv{-)Bb&WNp-UORAuoOH3`bkg z4;SU-F03G58Q?;NqdMHi^GbdX1!8pBEbNnlyvINqK#G+!IEv1Ic5Akq9+Ntqlabj^ zVylHG8=Hm)l3FKe8r+xHw$`>w5_v z=Prq$9&P^7?wumehXWr}cV~?=w*}HCJ9GoLhX|0I@q) zfes{0$s9UXRbD0&sk*()A2~>h`DyBCbI*8=h548y#s^zxOlqZ5t?frhnRW#c5)lhW zKMD5bVuhE%K>3_O$td8w{(2uUQNY8A1o@zIFol17FSOrznGnJuwjMGuuk`I%w0zcF z;&HRo4waUm0^_Q!^-78X}$Fe!)D>#UuEwG-yP0 zTJ#&``)@XYWiu8mj% z|D7&d5d`|L2^*UKIw2_)|J~ovL$dB|seR20vYd&*8}#r+Wi4y8@|!^Db{DYKr%5p; zFUHD5po!Hj1l|6&0P>_*Sg3#CXR|XycpUcC&fS$GcdtQ*Y(P9V&0uuu0Fgq%oSt!Z zpW$Fy>TE;(_^s4>IrL<>=%Ivwkfg-ppzl3wx#&OE{L+bj>@u7mvWN6LG@-(71N#M7EgQCmLcz8-m6l1T zwx%k47uS&^mD|Zs;;kj%ZHG(;o)^dE%$m;km&_QZ#VAq2^EsPSBMwv%SWxn;FUzXr z-*&4Eu?W#?ql{@Ri66I9>)rnktJ{~=`k{SRLWWE zZzX&J-uPj5#0+p7#_Nj=aBy}S4FW_6XOm@4U=P#NM(Y|{m;Z)>%zRJDk}(JT^fRZ^ z)uwLk^HeU~5A*gEwX9)6UWRI8unJpajZPMMo2ym5hK_^+EhziZ+$_<9r_C$ASSK;U z#XAY_ybVG0&h6U$SFh>`4yr^u+_lsn-d@wWRB>|p8bBgnpd0o&c7I*5F};pKxuk^2 zOqg7;QuThGsSfcXBiHtnzQ7?@f%c}LK52RwZ7^vPe^nI|*^aB$Anox)^T$tVCZ?tv z$0yWr!KL+&eW&oPZE+Z)w+#fb@sEKb@{jE}F_N16iv<%uak!62%H-NV3N9I4P#)}_Vqjy@jR55c zE-F9KHiZhD7cGUkp@(A`kob2zxxZ)O9MevU7*u93ghmvSd?PAXtTR6l)%ZAlowDU< zxnZxvbe{qaq)eHCnvnihW5d!m1M%iBkSC2KxTvxBx33i{(?&Rt)a3Wbdnd(ZXJ~4M zOqBacJVdrFh1RJkE>+i`Y0fIv^oN(_p*B;l2?9O*))%}D^%Dl)JJ*Mbi57C16awzC`?_P$CgHowI2zGjtrysu>udX=T4{rHrt_FZtE3MPLvLjn>Dc0*}f z1Ig2;fVf<-JSQ-as5u-plW0*BxBBQ(?0&phr#FK%XEj_o2p*VUf<=)C+^I?3f=YUZ z&!cWdD%w>CcfaI5J$NOo^^}s5awQ~3!kBd3V)Ce_#q;`uF(HQ6L0t-C5g%P0zY))J ze4ob3iEMW^+6x;ssz5a69E+`ft4<>>)iZ8v1y%%~1jOnzaks=tyW}b9giUz*p*xLXC*HH%(m7 z6I>oFqSyc&-^sMe+b}SAZZ0vL?z#N}$=p>CIwW*GG@W=8gN#*wY2aDur>Q;0jZ!1q zR8iFMx`yu?o~A`tr`R(zQoDD3ll+?J`{vVfo!qJu-BPZDgIZ}O$^mb}y0ovzH;Of* zys~e3NIA&a->BH!N~Y1N zx!({!8?}iMU3E=?slq!wJ=;CkC@G1+VuC0@`tH?He=2NX^~ls$-!t+ekljbD&HM6J~!ox2n9SE;CYs~gJExvMbEa5Bt^oT3*gz}q1$nj=m4Nm ztj+2Vhm=D-~UR8(1wdKNp(61y@+V3*g!HZIY zDe%E{Ae<>)6{?P$)6I2qUM8T%B+OS`#dc!j-@hc}))Y@3nZK=#)j^1~=8Sc{eXY^wz|tOW ziHoOedp;x?=E_>6rS^lFo)pXi6Q-wpx^Cu6ehjLkFu`pVDULzJr<~WyYf8>Rooan4 z$Ig0p`9^nLVOMP3$<@7dDo>81+Y>__)ES$5m{U6Q`Hgyb)_}|u%j_X~d_ef)%B59U zaCvjhA{|`>&N(jRR;OXpSxnxzT9Kvb&{WVQO`%9SEgPl%1;yWiWsxI9a+ zRvM8sfz@~o5j4cBSyneLl)aB`E9l5`h-fFU;J17(PpHS~1hWI#ivPvVzA*RzHlqe! z|49|z^J{FFY2!sl_HQ$&A7(gMGCr+Sv#}<)10k$Fw-=_9=ur9Qm(xyu{YLP(2Df1Q zOh{CYN5WJ~%Z9hEoE*e6uId%a_*g4ac@$ReQ~tU*PqMaHK6|e9api<$G|=!_u48~~ zOx8MP<@JMtG88KDqVb$GiKP-$^1J(&s7eT>w+97`*GMt`Zs2c6AyWXFioJL4YXIGR zpglQ^BgT`s{4K8Egk@BNzdircjSv1))sOlyYqkJR>I9Sd`346L<8I zO4xZuS*7d!Rw61kff{h|tfy25!2CUr4Dsho)ji`}`=%;X3$l}2S~#m=8jH7|K50nD zj^aDA94&@qOSGJgiP zDaH3tfr5Ld>mukL?XAu_U|gw6plb#IbffLPn+B&3qCuJ&u}1jfdoBCvbbfE$qH&vY zfW3+#C~>^%|MK~BA|qW7Ihzr?{%Lg$9czK*$g8vVnD-2C1+X1%Ba3Mj6@QQt2R2Kh zwMW(EB;2T#Bm)15hw-kFoE^ory^$Vr5Xt-pLdQACEwwTsbz=Wz`;TwbXNFKos?hol)oA|p7z)Qaz{I^v-gL4CZ7upD_v+)xe{%q~ z5|_Jc68MHbPS;@3PNarKrgsgnhX#}5n*K{@^g6&JiHev(bWG{O$Hy4>D_m$yUzVwo z*XO_`as~e$t=S3}?o)5#O%QZ~hXb!_=)4MzSp~uz#$%kV*UDHj_(Mj47vl(u)m7=UcNf^cy|Q<$|scC7=j5+Pf5EWCVkW7HYrZYC(|sB;H~B6J#{{y#m&lp@WB*8cu<;PU_UKqLP* zF5=aF;Lr#0{@$}{r`tvjB|^FH8xOgW);+Ol&6M+w6YDWQpPnto8`VsVVSfr<5eq&5 zKVU{78WyHt^#W2~*OPeCSbHuxnWMw`7PUc?`pr3p8w`ahwPHa3cVsfAp(cV>wWgo< zFI$1AavS<(?Iao_1yMB3iJy_MT?)X=D|{gG!BYYLHER@b(qDgo4pV@rVzU5Rs7{VT z#%*fJmuC3}QDD)dP3pR(Z)@C24BMuS_8w>wGawOa47wC#2s!ow7u=xoD7sb^piL&V z)OkYgS_$90DNW(AM5mXwO_7(Oogje!=Or7KYt3DEn`-^}goF0Uj3l0Qts_tiyI!V* z)m?id@_qtQs&5UKYQmvt1vPM%Ngvobt6grz^Yrm!ZfNWO#oT+xHMMT*!YCptO{Gak zx>V^(6_F;=1*EG82qE;|A}B?wfPm7aNN)n66Y0`Bk z_ub$3hmx7h^3FNmGRE^fV~j`AO9Tb5X&`iX$G>VRp9dt%qlY!+gAC$r@OLi8k702iwo8DoU3~lHJAF=cT zJk~8SJ}f8Ph5b->C;d6h8OGTwD+#huJ*wpaHauGd@OArP+uEJz-Iip!6j}IHNVRoW z+`)iAlT&zu&ON2CEYnx^=zd-;3LO+IAqnwk*trG=^N0$bA@zdXAmk&B^j5h95jY!9 z+``n10)|Ho7uMHlc2t4Z`P__M!9KQD+bH^B;BESC>3AvgRs)KMw&rOI6I90U5oBMf`kxMgn_U8@p{kTW3yNy4~{>E7vZvv$U87b`< zG!SV0*iIl}?dgJ((pYVa)=$0&uqY<}%`Zn3*8JQbT4PNf?Esn*lSx4XrWNn_uC&P2t6|5&}c2>u$!Pi@xQGofc!l zpJs~a(T*(uCby(qQ2rpl%oxl5w()r8)wg3^p4{*HY9}&FUpj4Rs+X@TURL*c+Kju8 zGZY`(hTnE7V10on?cs~3Vz1f?a3HJFWSE#Nt_RIza1zt`~?NAbdSkmOVnR|Q=<-L0z` z&^V>TCA1DF6|RlU$m$d1U`W!157`>bg>59;@)AWYwR4V9k;F3ih~h1$cr1F!!e%82 zWPK%^XA0Q$`wCt}I4>j#GB^31uaa-*J*gy5n*C5Kmxs}-d)=seb*$Gl^(N%wWjtn{ zl(_!4Uwf?=UP+pFHXlo67ECN1=)zvsl`tf1c9~F zac!iT5P)6eiAjnYHPKUiHU#aRA`Q=P-f?wqKXxggGr&HK1C_)WfF1P-X#3BwN2=;y z^@f@t-i^rQ_y`!I{q0HyT0bHEE8jrUB&Y;mnsUxWMidkIgW#L7k>6(?d#gZ37(%>6 z8G_3$3n42ZmZBNOJ$aR_D-Oisr<|khG)5+-NzrX$Ef^$&sz+Xs zIHr3iS{0A3DfWK2ET=wUMRGaR_x5AQM88#-_Lih@VENIRswWG(jnjUZ>7lAN?bF$eh;D|=kWkX8 z8i)=&vOj$#mvWnEHEB47sha=5YMDO%pMeQXY4E!8T7eLQ4;DKw_?3QCLc_&!+GN*_ z?WYqur^tv*Rgktt*0vx8>p)ZZ$`&uCeB6cPx+^N~(AGRp=qER=hp#og zd661*mC7OC4+7M`+`U2-Ko&juWq!nI8M;5ety=D!_uW!IgQ?|l*0D04HS<&Gv(-NP z_(J8Hn6-@!Pl}{m%j3Etl+)OlXQg0-z-8Rf;}@=(AsSy2)RP;zJ|u78>BL^062R?? z+%&eBS(nzQHm=A)o3F|S>I~UDfB(Hnpp!h`l=(stO|K9CNHIn^f0dfAS<*$3s$e>b zhGocheMrzi{vKYWT8^Zr#@fc3uV#&<;CEq?t1nU7T{@_MJTr&{Vr4GbS>E<*d^<8o zN^t8fK4Y)r85nFx0#=L}y2=q@BDO{dq)t~0{?*S-jkgtp|r#pq^WSN zOMLIE^-qmdeEVZvuF7jeuK$Dhn6o=~e@CBoN2fUM4XJPbNR zQItc@JgP58#BYb0z@^6MEfNcgD9!KAs#oQS6}6fIZRMX6IV zA;YP3A&8DeE5?kO@%6GHNf6`q1!{`gD8td-ndbh7s4u%*cS60I4|~~*G99Iek;!w9 zYD%}|S|*Y*3&W`V8X0PMN*)QN7I9BbAfnW(D`Uj-l1s86$5KTS#mlJ7r19cEaPk$| zbh-9%hDk2Kz9nSADyNQVbaB(dM3A!qzNhikp3A$IJFIaB?#;m;@I~pfin4^C@9NYQ zZ<)MVgW8U%1ZvdYHZ=Z>`yk=NmKsiqz75^gaWTGC$06Rrg0+H-s_NuBjG;E}tWR^( zhoT2k2c5k|^(sWE&GyUZ8Q1wP_dAY$J_`Xkiw->sBQnQP3DA81&tCl3 z5(z~V7qcF1>nGEH88(L__fv>^)AU!4ec!ZX$qFdZl|fofY-zsrT5GhwAdZR6`usWl z#qs-ihw)}x)6j=o7mtQXa3Ysg>*~t}q45jHD>N_3UYW(;3%6Q&VoI4d*$5mip&gM=}PEydHLCi2GV1uXL05^hy5386;@+d9uyQ0}3AWk)m zD8kfGOgJpH-al}XNsK?|tK-xCu7XH$3p7FWvkiS3l!&Pt^12U?>=xKitdOIB5`HSC zBmXJeD*k;v>(Uo3Zd7gOs~H$z-->{JX>yQq zd?Ss&^bs?SP)}>{?x}8@N0%$=Xh0~$Ac5$zz0>YEnLC4x>eq8O=uaWI%}CeynD^Af z3svqf58>C~5t)`+l`$Js4$2%1x7%C2Z*fSaJsun1KjsPyY}U`eYVcA&`J;#Jn4;z4 z&^4ITmkW0b~5FHo2044GPD!j=?bVa%0jIacwB`YPF#v6DN_+?3GF(jHKzF8gv@ow>eS^}x=G zxD(Jxq2y82P_|hi2Avl;B#uT3mVFw=pnk8ntXgr1UVZW4=Rw}d2Nm%0YEaxbG>4Ij zyps9HQP4@%rV+b8Mk{wkY-`z(RftOaVp^ z9EEJj!oF`Y@^)Y#1?cF#I^bxBsGZ&luEqk@3Rk=>jUHEwm~N0Nv49XHN_y`~$4ISZ z4sWKdMY@dXMeF!uosoED>r`mX`FB?;E<||{s^dG$xj!-?-zQQ~tsLRz7W(dVLg%-D z+E_j1)coH2o-)*NC;G7Fd7b{NyTFeE)F7+{c5p4^4;&?))zm^*6KEI1w#n(6KOHMv zaic+&5g~)T8@4vq<}sJT7f|;jXj~Xyua4EZv!(<~=+08DW9fwkzOTZgVupnk=NnGM zOo`vnZrCXtNqAJewQ=@TIlrQErD+}Y(Wv;riwEZ-wy!r65?5{-J4!d6DkyE829%_A zFS-p3n7Rr#RWeuuq)(6L`n|F6i?9M5kCs|cy8e##ygju)5p%3*_wYb!EI>)(6-PK2qlg4A&4EaLBwj})CsF@A&TML z(lZ1IAu!h7CO&-JCn4YcjAbo;KjnLxwqwvDAHww8 zg?|Fk4+1_6UE1*l*|4_Il)znbO|+gCS`XPTN4`j}l#)NOBBJk}pbW%1WreahJlMYC zc~|j-aIQoExkR@j_81v3w6a+8{#7x(5ux3e!UU`5pNOWIa)ir-k1}qfb-{bTas=>e zr)^s-AwXt=Qi+bfk%y!Ax&?$?eB?#N=jiYsVTH3; zU+r6Qq?O{zo^|BgS*M=Lz|l}J3bUVN_)YXw822^cxmzdpf$de;cacRPfDwPz&iWC!3QqLeXR%zYjD3MTDQlI6wGwoD%-%GkL!--Oyqv%dt5V+_#7@h1_ z)4_oYrn?y}S%(q>D=g>s#hlO04Y92Y^X|}a-z|Zrox(JdE8h*!aJB35yCD^&2GU$y zKQ1+V0|^VzbEpcARAP^*29$S7XrY9frSNdw#L#{556ose=56w3V@Z?K5sQz@O^!Hy zz;G#cKE(Rtfw+s6lpSpe(Cwa%ZJu7ag0q?pgnr!sg`r05@7e-HH!3Q+nFjHt{yy$$ z8c{~*E3ctetJrQ`>{a&4@QdXet1}I7_3sfZJ$EIYk4?A|#OS<9~IAe!CyL+y-Fk_@}}f zo6H*+q`NL~9}JhG2%akd2KImQ)mbv#`s<_};pPJr6AFJFbn*IZaP+3rg zJy0kx=0mCYlU616#OR6w%LzRc$hllFQ6@^!bGxs@w?D8_f;u_4ac!Z7v+(27xI7QI zNT*B5k@B6l4}7JoX{g%D#YS2OnjiA;JgvB16{W^`J7Ij^G(6*YIGCfd&>{K8Ah)pX z1^qEA5uHOjDsw5r?Lfn$q8<_t74eyaKr{DTC3`aqcc`bB3SIt4kZA>~#x8Z<>44l8 zXlrlj!;ngdJqS|v@t8{N{GcV7(>m$HeoXILw?Ic{7Z8G$8svMH42s|D)pv?lO%RF` zP|iY0#DQF==Ke|_733TvK6p}B$*GKrnlqzC8+Sw&wvxG()lxcJwPzn`5-qj_Q#hQ1o&pWozU!}0yfZTRQo^rt!a=+X zMwG0Ts|(#kp;ign^ApM{KC=tRpGD$6GsSA6ZU&zZehkHDu3F=ARP8GZyE*SI_^{-Q zB@QvPU_jujQ%|3uKAHOb=KGl$5hwZfQT1)XvKf|H zD%6&P>qM_^P<$redDY{yz>iFhaR}BWAB|xom{Y;Q!MSFHgMX3(j1~bfTGms66~k3@ zipN!0h>`}2>wj47wwP!>V70$b>8Q^B182CXPnQQ2LsCi!Laer`{V;lDvJ!m?=!Cc2 z8ogBEPX-=HY7c>g7DcBqS+HLJqSGUjtvo07^PIUH+UmI%ad2_2=yRl!NQzM;%u;T% zhUZwbSBRC^iIQ_Raumy4xat!V0^Q0}qVCDJy{Wes(i~`3RTBgD|%C>Db5Le-yGtb0y9ZH(x1r zovC9w(TK;cw3}5W4~>57S`T=W7&DgWycs%DK%O+YVAyk56^teAf3Y>|=A$g{&BG7B zm@FGg>s#ZpK~iUikyWp!Ez|f)?VrtXxTR|Z*c7!R$^zILGxSU}{n%hZPNsrIV_TtV zXoz`aQ>bS3#Ir#DJZo97nW7B~fj>`V_L0fuOJy3pTu&r3qDmeT2|w=-NMIfAZaiXROLisV^5c)P_k&$oSofh>roep! z;p)DR@|K!Zevy3Zszgb>63Ncf=_q-e78fB+&6f{VtnKyjn;>dLy|_=@0-MiY%G_1S zGh&aqxmk6eaG@LnR#4`tFxYBl3O1-_Oxf+eLvQiZFn-R z6ZYScft=GrefCXX4Is^3c5(eoS}kaQrM_uBQeQ~@q8SogaBS6ctbnN%l4pI%JPMD8 zUhL^ysLNh~7+AOd#kS@8SMm5?X*C*tB?+`vn~Ahk+i?#aodJ=l4P8Fn$E?Bwzk1xT zedSCxakjZMnKEV40~f>BQ2F^O?B|!~h7Ht#4wgz9)Hd)w`(_26x?-JiF3<)$``Ltm z4MreN&5>3+*Z!ye>dJ128hStC6t(BE>#1Y%rEM38OT(b=Nc)xQV71a>qD6;+i-wIj zpXAAKBy%i(Zp!atw=hfRUS>=pCK-}wkS9&MsYDECrhQ||e?-r^2)Cl$UeA&>;Ga_e z)O_dte{KH!NoM@{w~9^;`!FjIwnpQ92K=7IhoxX*Oh{ng1rEWq=J!f$1*Ak21oHfU zsw*;ovk>4}C5v-XK-{3Xk?lX>mm3@M2_&T|k1VggM9{>;?T9w4s1 z0`)DrdYL2NJ;gC4*{H`BdsUhkm_M0qSbQ~LApsus5$yuU*-*xHl zO?4j7Q2uyGDN&O+g<T2rdd;4YrK^DMP&Yujuz&w=Be6*m$l8>i)9s z+@;Iw&-;J`uCWtHK;{aZ?E#l8D)EiSW?UQ6|Bd^L6WmbyPcE4Lw?mOia zt-n3b)-Bi9Ww$Ww&$wfwD9x-R9XYVf--TVvd(q6GG`;vKuv_|2DT461g*e_>SmU)b zx@NzUW}p;W)MR#tnw``T^cs)lYTJr7a*e zIqC%O)e%^{efOHrc+wKjbd=k+aoM6wRwws?^L%j(;~yCq2| zUMKAM9{mwuWWCL>{W*4E&;Km`P-r@9f{8vWl!pujYqA?`NCCTq)y^qC{zooSiG?v` zL=93xj;9{s{X7Nrz+GJqm)9%1sEtO3R;U|km|d)B4V-3-4qZ!#^Uc)^_HfVcomK`y{@dv%b4 zF|V4yeK@S!D#fYjrImAkm8aO9fM! zU++;MM}baz7o5It9sD;f3_9P9Lf9-H=B=p7{x;-i<;hP-6{@@PYimvI9ikS%6T7qk z{3cmA`0v$-31}u-K{(}@t~>>E5rk*Wk@xo-&?GKmnB#j-iBY|vNidL_V1{-Cw&XGb z+oA@YG*nkX4##2gDm-~mYg>Yn5`+{~zjv=p(l9E8oSSFL-5z`;rmJQ62TmXuyVA3W zlS?gtaERUS`z!kzzWD4Nrph6w`FNH{kw-4oeu6t5!whTP(pWDj;%EIm%in;Tj>``8 z0d(x=iw?}eoFOGL^|O}Wulwt#!M}+9FbYPD#=|h^G1$*4fzqWyU_@>MC={WLCASLZ zejVe&2}0T}!Kj=AcGy`F!n%zVGrII@y|FyPr$O(peLE=N{0n>$P>zYdQ>gptC;XDE z{^tsy`_VuTRbFO2(if$`l3JwlYuuLP)yoH5YDfEl*wbLC7H}acKp;$&X!RQ&K4&~3 z&%zvo?&H(K@EnYalqayw?>S3pYnOi6Q{j7o^x_D^$pNPaM-8CmQv(>T7R6j=gJ$#$ z<4OzYLjFZdxsti{>Hn^lQWIz?3&3P|f;S0&$6wIFIB!)^MeFjZ{3w4;p3L^1qnUp%3U67mpeuY z{IE=>vv9=AY9=^A5V~nTCP)i367z(Wehi9qz52$vt%5~YZstO5EjrpgUw_nUd%1KJ znyg}cJwIyWTR2nGO@-9Cy`uZrZ^hK@!`Lo_?d>CaA62i~McXX>8l?v#+BRjtnb*GY zZVWW=FaQ6^$Qg*#2EjQ2c&cni>RvV}oYYs^pNKh;p}ypk}L z8+CI|jzVmz?JO*ls6z;*Ux`ktB6{&etm)YLP zH-E`C^cuy*UyfIcGq_?H|NT=s;Bc8n zj(oA+0~!}&^jz~gd3Rw--jPTY*8{55r-CP6nqT;dz2nSJ8Z|6&wUC`b&>B(+S7PK` zGt1#|KVpaSmEDPI(0ADv@CTAS=T5w(Th@tMCMx^R!u$8lwONxwrtu_q#0a&VNN~qR z3i7R_7)u8R=;X=#%T{mela?)dz958q=cWWpQG!WdW|j}$!djPEULOxXFg6eshL{ zD2jwDD`IChLgWQQgA89-l(foeBQacabzaMsu2TMz1!c_Z6j9@MsRx8+R z+d37WRX!gbYmX$jMb?rUUnG4JO4NX3>|y6{E0a8X=xKgWy1}+{=I6NFHcMyxgYK9R zBN~Mt(LcV8OJCD=GNJjLlhGR6rv%GqON`TMvH1{Y6n&F`o-BH-K!2!~0V4GphL=GB z0#T7-u%Dvm)u`TznfRp;m&MS%{WC(DGaTz7Zh=m^VYkV2e0v7DDkUgOirCU zdR%C5_5EFCADV5GpwHV*`fO$KhDDO4hKjLxA>WJJ+If|z78&#rpH^e(>Cnz**(+K% zrtj(g(Cq3mdZmuf65rd8VV|Y)GRazvA00@IYQ=v$Up)Sz*8Ii<+hhAF+FcD;qqy^| zbU?g>4yi2pdd8|Neo_k0N4k2QwmU*o9H&J^eApujedN!MZ=1TtH2}$tYx*>He>|SH zs2Op@mdp$Yy34Q_r#@I)Eo_72has}L=&!H$sF#67F^A$i#NR0(Qu;IcVrFd<>a-d` zIx(q6OR;VBsFype{IR=+v#^dD!_sxjI9+|CS*vK7%OCVnbOU#K#GGj(pVV{pZPv=9 zyTuQx%EitIXMVW1M)ru}&fEKHv?oaEgNUo+7^S+(!7?|YI>V*lXYfiVJM=s6*(_Yf zt9RD~4IP3Vtfv!6qbcp0L#VuuPND8+=gxB}JP|cGlJb7vEk9 zZ5V|YU)*B^D8<6cT0X$N#va&1id){xnSXiwYlY#J1o`=$rUb3*ZH8tQhWlG9D}6^5 zs{$qPC)eQLbPG*xX!L5K&GXPbo9kGDfg%y>6S;}R8o{KRq_f)VBD-qH{fIY652KM- z_;Hhdg1t?YLix9^c#7ip7@b(2C(PEkLnZLi##C!qgJ)al9EfiS_z7?*8F>VA-pX?o zQ_*b^BMx*WFU-nCp?Wzw6u71%SJx!FpRHg;t99SF*(zeczZnxAkCTUSUc15a^zH?U z>Fp=S*qD6FdM@Pt6DPT0>vBnn`z(S?r-`}mWG^mSuC_u*?pGlBeCRAi*qgRjV`#qkT3_w4^FB@3+&2+`4&={fd3l zQ(E&X=}T_R&Glr%`NEOUkpjxxea89{{iGMTM|HzOo_f3^c2DacGAu@FrAEve4H_F5 zD$N|KTv<6;yE>WRyFl54>+aSwpilr=8f<~if-a#ZkweDcZr!tCmy1_DA8lBWloeW3 zs}o-oRLd$n9SqwOqJB?+7jLn7^famm|E{#Z(TETPaaCy>|53?PUJDVw+imB9{D}v= zw?Pn1G_Tgn<<&lXz*FUl)g56&2hQwcn?vLg5lWQ-c(D)O335IqEk(;LPZgMa**d7p zN4GX<4GEpg6l-7+tNU85qui%!Gpzmm-exjCTBDPUHqs!;rD*os-l@gC(h5Zz=)PV? zXwR4J8`?eccZ4Z&=J@suWM^ua3wm9}WvIdq?q|wBDf_^9?ik5G3k$gHND}A-CBC)3 zSL2=q`50wi&$x4QXurqSYRLl9|?I?NKXWk7B znJ-757#4qd2`;Y%&V}-ChO1aOLbGL{5>ij!aTRpwp_;N6>Vj);IY}ztN1Nu!dfOh{ zF&~nh;82G%}Kxe2|g z?eilvZaVvk#oK_DV!uNK^#yJ?grM1nV3^lg+@R)V*@P2Mh@N8g%#X{^$(P?hu$SQ6 znszjxNoqch@h|TZK~lo|4OOtYw^+3;r6(@T1Ba;A%ZqEIfO}yO1YiVD9C?;2BW@e% z`L}h_yPC;MX~eX7A&b@VrNK0EM1n5DHuA%kS?iq0|0Oc+FI)I2u=xL?+w-V@iHZ9M zWIE$-9}h|1wtrDTkvQH!5kZgy*Mj?Czuy|quLDIQ5^J~{a4T~NNkOK$VU1**;9 zy$N1N=Zw*Dswh*IOAAl1t5NBZYmlVjdWP>-@&4-_D`GM6CjNkX{gT@zb@48q60Myb zB&)1EptXla+3E5H4(2cgGh7_aV9PJhC^E*_a#4X8w{|p)U&rb%j$SLSu_L9i=S;wT zXXpN(Sm$4Q?K=9uK9&@HId2uP;Rv-5n!)|hSE)H8^R-97jX_IG8+yN$PEED4q>}Ji zWoYY(D)FVZi&byz^)14SfD)L8SLSp2#Mrm(Zs^?@=%~EiB8>s|yuG$iiD#PI;-lg? z?V-<<{E9YawGMqoQUVf@J(qgRZN|{JYT*z|UfX?w1}N^B0&P#ig-10HXCAw3j${r> zkeFqepo}C>Pp3;X;gNU@I~j<{~I+If8yfv&J0JK$CRgjMX z;w0(nwVp5&ds!6(yl(r-Y4ZIOsL5ST=`cw-C+#p>z^42&wb&V@eEZmyPA1E&FCW81+}W3v&*=ayk;VHlSG$&-a2|85_-*D;X* zCH*C>)axyy!=Qvl7pQHzWiP2vrrtigR#(_qyL3+l_Sds(v*vMXcy&#r71#gzwZfQ} z28arieT?q%MTPY8bplSo)YDNfW==MlP`_0Hv4Qw{M^N)j4}fQ z%cx^H+gcCb%UEZfzCzs&S0NorlmvJU_qu_5cTwQyQ}3Te9AA8iK=XrFbk+_nm*B<| zgPyxJgy43{BukiPSxM?##{+0;foF&^AtXRKvSpoM=8KRy+sKZW_OO{c8p|t+y z>Ys_DV$sn6Hbw&2c!m%A+K9CR(|hU8--$r4~m%h%AlhQYcvP@tsZlzDg)_ja;61t!6Jjs zdcH$b*9}G`g6wIn+;Qn14{xHC0h6361@Ke9c35Mj6?<2&NWO`<7&TWifYGZ-j&WBJ zXzsYP$e~}hIPdDEaEVdR*E7Ms)j`1ZX>3n#a4uq4`(S+xMxgJ$K+G5-A|%v^dO|F& zEPFJHTZx=_bzZ*8PfUD{Im-U)rYgVJw`{px<^d}0p*$Q=gp!#t_yxrl>T9t<6D>bu zV;i-<_VKwVCR=d;+q&oN`UehP6qr&N{m z;4}&_`}}f4T#0=c6*3>*?MI7BBYLdy<2cn`Z*s2k$`Zby4@_>Rw&Lx|ZW&EF(I2bp zFf6+!wC$I3>z%DELz{YAtK?`!O<<4r69>)Dj{+m>N|vW}CGt|W-7b7I)TN5Z!B z_|RZH{kqlB6cdAG0?msCOs5G2;#h;ugU+2Y+)FWTk}MuY$E?5mq7r_IK`m1DnuG-v zGu551`*#Y{u@n;0dz@%+BY0D%G0y4(Vo8-2okXf5!Lx`NQ-F5fFA9cjfwwUVu17%6 za7=zyHqdVq+W}uN>OPtqUw>K1_4`Q}wn~7_@kzHu8qEu)`s@Xxrew)(le0(nn_^L+ zf%@GDZ>@4RPv!&&56Qk})b`_>$k@FsuITU>q!!1YPS|gRuG%(J7b-lFF~-z8+M&JD zbbdYZu|$#r7WNd-F^*?EjTz+z9olun+AuBWNGt+_Z2(C5BA}|$+lOF_2P7DLX^auX z_50*zi9>MK{EGDM2|gA#14JyYv?A898TkKv3V(@h6r{*8A&^&c+8^GcSK^ z%$R-qcJ-IY=24F^ z@@Z(-)*z0KLa>+=EN2AoCf4UU0K3^id|08Udr+yI&dvsiZeHL!*@2;RCH}cZ9!X=gDRyx!z>E7$LnlS87uT%3&{2d9V-xv2?$y2l zffweQT=~Cf3fKZVA^&B{!KGR-&AbDsbJ!s!CW#W;K$&tZ@}Ek};y}0iCFLk8xh*9W z{Z2#9QA%=~92wxn$GZOBzI-TJ!S-6*2D7-Sqvw^1Pwix5U&O2g1cP_-HmT8Zjn(Ve zq})zOZix47qX$svBkLVOf0|e9IM*G+n}n5TbTV<#tIZlp*=nAy?qq^g85NPD#TV*v zIZ)^B;LQ29@ItLE*G#w~*B)!gX+#qJ9`^G$OTe*rR?<*(3$&1&ZRYRQs`M(v*24)O9|n5aohS&-oFH_u=RDKMH=CrLy%cidS(w^wPK0 zfs}iA+%@VAggP6j5Kl_l{DHGx z{gx){JyTkM#JYN_ofz@nIsL_!W8st#o?1bBJJX?!=ey73waNn}Y?hNr$RP&EH3?~= z{ECg03ej5-`evRupOnBl_mofxp?VHmcyDbBZ!%%BUteclinK>~q&a#}|Po@&*ClM40eOpy<}vO_l- zwB?o9ajrO^KP+zZzs#Q?H5rspFhBHA>8q+x(r22{lb=CS;Yxb!WmX`$ZOS{l>2KYb z98FUaAY;UtxVvPj7u?lfXbLQt$U#soWpWkByiq9y=-sAd`XRZ1%xrgE8dMU477}G;p2`^KSs^7dWVT|hL_}xE{UnMX8Eke7UR4y(`-I}Z%L}0{z*K96X!oi7_OKvKy3+M!0=9fVE z%EPBv!L5^=jxtde2J{u(XGUc{xP4%vB%(z$M;x)$1AwsDYwGxNVK5Ra0fd=_5KLYY zIszD0W6iJwl7v5SWQ6*=g2k!oi37u^;Y} z@|!8<_BHe$a|BDQfHEY=WP33c_T8lyaK43~H+E2sqZ_uoOclQT)p~}Up^1dipzzXF zMG^gT_>8M?V8MP}ZgVn}VsRnUkhw!U^o~qj5huzH-@{5w(YHV5{hry~i?8Smk%5SL z%WcFz)S{Lm>-(5kNNy+6628NyMLs9KC&ZQ<^iZjW z@jt^4G~*gur2Uh;K`?^-y#Jrkp6)GK7>_pWx`*VSpG-|&To?kvr;>E>Vr`A6=IL{* zhK&@#?SmMVG%KY1NRw7YvK@i7{Jif_gvMU}x_GP3aaPD;-#WBYmJmEw$ka{^7tZ}q zhjVv&JYK8C_zw2Vk-hr=VTK<06YRsDFd!87M4Zx`DaQe~+^*jb>^~x4cBeo6hpRvF zTiWnjzp2Iwa@O3@y>Ia}hszS^990rjhYCh!2c89ny`|c?OZ$3*a~-J0lES$jY+ayd z*sg(4dtNAkkd)q-kC}g8KkZ8K-0drFfzt>|B*h|ka^;=DT5a7WUY5x=pML-8L_(5# zoTp^@0Sl90W!Ad(KVQi4X=aOvR!Q9g)wZPr7RsqJGgNF@>n0!QUM=Yv(0wN}Pu(+! zxjTEZ86;-=c7Iu@@@aWR{5cCt=919c)AQNmN~?o7#N|s)>iCgvd@Xip&xUKLu`#Jn z`Nmed5R0tl@9re=qt9Xf_%6aLBT$C}d*ONeHiySl*V(yqC%&=}p4xG>2-yZ69wbvn zH&>g*bQcA$?NeL%|~;UI%d$wnER6Cl^G*LDjT( z_as2+Z$2po>9_md#wRUm7utp7ktzWbAN@0oT<3jH8toj!OsRzX(5k%#gJqYj%TGce zc{N_G6;}lrMEDQ&zFefC^Eib#fO=bP0h{>EBino*@xO#nEwB#V+ z<+F8{LDThQ6SH zeNs=5Fh)G3e@y!U0m znNIb1tZTRYmX6*bTz>lYZT=R6q{0yA>($r_iO#RpE_JzXvigHg?a?w$!x=9(nC#gu z*~ar!-oV98#`=Bn(s%YWp+oyUVTo0dy2PEJeVaer@+OXjC7pP5`0&P_=0W^qqqeh6 zTaa5wi!nuT!*PF`dGC z{xyl1_5*X2`$6I4-$ed~$Uej6B}4%U{l9tPIu&oZYn36h>s*hA_T1QNQ=Z5`IO=Zj z7rH#4wEQj}9UX04Q{?8+*6acMUi8LKT9~!#0oV0F=0v9g{UXbGUAR#M|Mo!)NSCJ2 zf~-I_8R<;}abI3I3!e_;KP z5_k6o`G$~su0HB~0mQPT*K)v6r3kIBT;p7kRA58U{KfW)3p-7*;E}GYVJKop{H64M z2QX=69y#~ayQnsfTkg~wArxyW9#(}ZrICTv%pWp^+4-dkYE>;up zp|>i4fu@Yf(|YdlrwS6%BGz&c_spIsX{Oj<^IBcmVJcw*4ZR}!E4S4r-9>IuNZoC= z$P6pj>YWC%yn2YVlx8GPTT+D@08h$U52_XPl&c#LYPhRJV-h zd!6sZaLtN3GyL20nt(h085$Q9{akPCAM4U)M7!Q$zmp@>zvy#LEVC{vWN(W2$vA3? zr=#q;)kBd_=VbIQu`e>SSLROg4|~a_q5F*P2ci6y_(R4*oW~hW1P`N_k}KDb4vH)* zzLb?mx$~?)3s9Q!>ZX16C~T$KS!yBv7#dok=FA#XGcF-d5tF95P6b$mr1}X?@3gu# zg>pk=$pXXmvKlTR*b}L4)=%31?*DYhsPM8<#hs%G1-vThhs$1I31&nP!@k?Jo!OCb zCzKDd-PE=RW+~s#eJkIg?4VuTFr{-}Hr#B!iWmC|=PNZxsk0^aP}lN`K#gM4aINmF z-lmt9eWuY9{ZX{H%Ms~Uqe2=RC7w#?!G57=o6l{Ke3UkPWW%yRlSbJy@X7iVbbNz6 zqU)S`@a5@bwShY676uY0*j2VqPVh!a)fHTZq=o8qk=ywzSwsHT z#FgaDb#sn}16tlDKQZF; zdPoFXcUJuL;Xqr{c2_*KF6p(R#P?Fxe(otT(ywyTZ?VQ-x8zt$4ZbX}-S>C!+2D`p zC|2vs+w(Ic5k`Xj8WB9`gBUV<}YdR}^&YBQgc zr!5B+osFD3IgR;yOx2a9dNuFRmzNi~1uQnH&C!gpgcW8Ctv?OwJiL~Sx>i-^Z)=Gl zl3>j+Zg4lvtQGt^GiO`g)u4F5D}#}fv&ek1>VXIoI+dTW;W4j!C1I`Dd(17_#&{B( z#WK<1s4|s>Ax*B|&U!1sQ)ps|%7>$hB0m#+2rDlMOP04wx4W=)`EJ{@C)jf%FJ3qM zaCs_y9^Qw9yvzz!CTM1Re~TL-5cZ1n74aQfU1%Em3K4E=LH?85%FLz*rnFJDQE?MV zEgzC^Iq*2lAw>& zsAcl?XA`SpD_4;;gGwehQ-|(X;kTP?>>M|^=+y#^DWh6ZU6z})I{Dq_!WXpN4!x&o zT<)hlnnQ&YTz7eb5yMSZmZP4Pwnta$I zO07NYUyu@rw2hR&L(RbFqs9WJZ1tDQp-fG(TO%o1%kXwv2z{0rT)2JJnO=!K%DY=` zHX^WEI*3%g+Ta;|mh-uYo?Ro2gc~}}S^;g9LXPr>PctbtdPTQ4LpFq`2o#<<3yow8 zEfp^Ce?75Tu)B2c>(y+*=qLd@u39+KS$UPq&3X?ZcrRMf94hlxUhn+1if0Wg{Dbun zpTQ=V&FM~C;+5=*+dR1jJ>(x{*Kt4HV2;$EG7DTYF&#aL;OiKTKC}7@HUMNvib2Xe{^gAV7SsL7uvSjG8>~e=xlJ3)7`S3 zUvio_)NXNymU)j+h;%OV{hnv>1H#%Fz@GT7RVm(t>Chd z=Vj+?=fKyZTMzo$cn^*=!WCQXkF^}bCSvEDnOz3PUEfI0EduA4T>4mz$Z5C>EsK@% zQUFjrz<@KhkSa#+&>9nJ)5g9)fk|;DrTlMD$nL$(B$rmj;N`Z?pIWaNkDUf;Kv!y> zP13lsSX9oj)lmlzfF!Es1Hn1~9U0-A z1TqsP$4z0^BC#<0;$9IKhvp(}A~F7qmd5>H)g@*;N+7YKgB0mP zLJg3_w|O(pINW=E_s;vi_s{!-@0@+eIcJ}}c3FG<*7~ix-YLWm>BoDej~+?-7$(Pj z8n@EKJ$D6n48 z!qKG-yR4o=XbY(z`X8r+5?AJ*>e)H!*|SFtj0I^U1@_Q2>SBQV!hkBwd~-jr_P9$G z*r4rzvYA1bFiCj}y1B|v3Klu^`P&Be+tFPPwPv^xvqugru=7J8JWykmbOrMi+-+7! z_U=)O%E+VYDTK z+F-9Iq=C>15mHM$DHuKR#EPKqupca;`#NGRUHs46Sm8EWKsRt)ZI>b)FSAR3vCVj! z@$-+L!Pm~;{%B7})Y+x0!(z-_tEfd`{5d36cl`RixN4**d)$>au%)2d@1uYB&)rh8 zx}p?aD<{#3Q{zuMKOq>#wHD){OTvI{?!1Hzi2pcDlKp6IIy#=B{rva~vNjW@)6{5W zAu%cw{z1BfM3BN8p3UrD^`L#iI=*8472;|KWWf%E^afCza6>Cf;k$Ik02x$;BV6FO z7OXbtUjsaV4)7FyS775evgjn`)I5iejzEc&d2sAc(3Y4lUw|YBp?D_Lv>9rs+K5Pg zs$sQpAME-|!N{U%F{)|~EE}acwQSiL$fUT_>(cobgP}0Y=12HnD>TM^GVY>%f)$7)?fU)4ar(xP!k1OBZ63$4niVnv5 znHm8_%4&ERG&k1;`YP9SR(NX**cnwE`3W?-xb(gKlqM-uMo&v0`!uAMO>gvU;Ek{}eaUjo}dFN&n^Mbf16$ z62tu2@S_l90y6O;TPyKFO(GM7|=>mkoYYnm@z*ybmk>~Urz@^f~q3#PlDgH zj&I}64FJO9j-Hluk&Ddl-F`|?q0084O;EhnA1l&Rlm%}*0w#=} z{U97_BNX95Ri;Y$c(>MMV#aufjDag&rPOL?Ar}mo#dYA81V$mvxu~xYMOiq3*Ak8! zLFLxZ_#14WLi6M3ND3Q_dkn*%R-T}02cDZ+iT9oyE#|C>(SNQnb&17c+qtY&gX>wa zJWxRO>@lA;>HF$BHvYXg62TjL4H_n?df{{X(L^D+JtTQeg20FrVPgRJD0q5*g|yo3 z)~xUw_;<*5{l(zwa`21OVkWmEQWNZNSzH%-5prcI&mwpRLS{3&b`LUF#5zDdmKo-- z`z+_x1UAI3VZF}?pymT}Um?68OwO5ud^cAvj)wAYiCYCDbUeF!xjL5tK@+FQ?Dt+4I z+lw?5{MYplRvR@iUcor)V8U%*O8;(~wZIN2Att*TQ8uZg9h9N& zue3#x8VTMfz1$)1@7`(+BQJFLP>_==6y27qE|3-}`s6nqWl5mE_{Go2O9oGk-|;9N z9(@$3h`7E!O7dH6z_RF)PFA;z8yh?^cw8#pPTMv^*aEn){3?MW9+3Z}_qgt?*Mk{b z3yMcBfn2%GQM>(wXx;BGD|~-L!Rtt+Vu647P544)@_1Tw_eKp9m9s5*&WiejoBEzJ z1EILH`<_IG!sq{pS4w*8>cuLCu=cIeV30`s3c0dzMeH`E#2e7(#pXzUu5c_cxHrap zcV2fWvS+B75xQ0YpKI9$5o}fjk?+pMKZvCNkN;-SZ2Dk+#M9Q-?BTSTH*0CpYx;z% z44S}!EK=OK^{ENjkDbVFkdc7-`mr2difRD= zebA$}_-`~x9u3~w%NiZ?z*h7^l^uZ&JIUs{uhGBSI%$o&L7%j zW6N*$N{Ah6){3+>VR{@AEG=gG8&sS(v^mP+zVqSIW$$6YKQ5NbF-c`PA6{+=E9 zTb5vPy34o#ME(*o8FBlb4hgv@dE8UXXIilTPN1A!-uulxI#*^>JAL`6rAkY?C&KLp zJ~LbS7k_(6j!S_}NlEhJtxI|>A1{y#Lu|po3(*>o@T@zDzVZ5wqhn5Uae_HZc9y0C zqz3dusvnW;D*tb|iM~0iemho4z%_A4Vb!e`AHtsr$Jt(#e4izSUm`Lv7P#9wXRthadG-^ z3i=SsFq8Fpq-S@9V_t`Ajb15YD%^SbT(1A*qfc{Vdejm4@hq%s#3q!|g2VZucUpCC zrpvu#hi9QEgKo_--({=C8^m6d00C?ZV(O)FepxPCbtD%Euvrc>_M*HD&3AZ0rI;Aj za{fFt2?CRf3d?Lj$ewqyT!yKy{H!Fz{>j_^jBudOCqfO*a@9e=?6P`}iVKsss)F-q z^qJ_PyitC-`RWHzN-@qqM_J!9?>brneQveNen%+IUn|$g@af$&`&j!y=F$hTl0%O( z?9atqEf(fSK<@ET6&LpttftKn`WQOIb=ji~!ph4DlNxWx<}>atnOL&+m!hb;aQu43 zA)S4BQa0A~6c8n=7e+;p_#Gq90z7#W=BNJfbYDj?%P&KHqvi^ymWYEQ8k&Pu4cvI$ zOW2CWRaVKUTN3QE~0mDJ@L4yJ2XbdFw1=WsXK zjry$Y*-f4EjM$M6OgZW)Y`WYTMb2z1?i89H=)vE5^4~nZ5Z}316O{Don&>8555Y%= zG!59grvi8g8ZoVg=rd!c`3b2mUYz7Ce8=zcYAs6BC~X#J!4k~ALU@*t?6rs+H%gpaCI$E>te)6G#FS1~@v=kmV zCA2P4;FOod=(YwhJVxpTWxmAAygFRub-<0Mw{+a5(F;Bzcc{~}aKyT9-Z|r%S_n}d zeH7cW<;vyQx0Eax%X{})@Ofva5$Rx&7b@#-6t#D^j_FaS;tNgVo9bEnHy@^&BzIFi zwXq3oesJkDNX3Uo%$qq%!tnRfI7Xs(oqPNx675A{ zgamU#iA&y>FJ(Eg_bnn>{Cs}>w5LaHKgr0_F=Z!fa-#h5k%V+vWaup~iWbhjM6{-* zFE-)Q%WJq}#g(uL^QzmoFHo?YR`+2$u&LC;_~$Nr_(#3pGi>_oJv>}<%%%1Gt{xR* zv(%0-pAcBaoWPn$mR~1NQb|B(Id%)1w?cwmc7Z$NgwmQ+`nqVX~B6KSn0`1iLM^(Y+| ztH;Yk)$Mup{8Se>R@DVwicODkLsk`Zri>NNI@`!>tzDH| zJzk<0dL(ZSlDNg0Uyb8+ZI&}dMcnu2H*Rzs(S0Z(u?m!C`lQz}ca9M~EXV&Phcn-8 zoSIfC+Rxxqm*kGzsD-(C%XB@xg=XcF@knx^w=1^qT{7d~j6|;eIa^I}c*j7oplt{` znz7CcBZwWJEu2EpJ0z?r7unzFIJG3m)X8-YQAu$Wr7tc%<<+!MYoFD6@|hc3Rv~Sv z_Nk1}IQG5XDoCQ&F+Mx;JS=c=j`Fc0`T!k1h4vFccbeDh4U*U?7gVM#Vc?cd$O-3p z6!iH5MBne2psF6Bth%-e*skYh_VJw|q3x%j9|R z#zfZ|U8N}QtA`t7$56b+N#tjbB;!@f2h(&iXJUg&9xz_>eX#G9;J_}D*)Pr{P1emg znDnQ_VLIiHefcV;df4qAni=(OzBg`$ex4o6Sw9Yq5X97tRKX9!T-H_k{WRS?g&2^@ zY3O?opU34Wb2Nc6>ADw!LUUo_0=6@z?Cf!xfCB%R3?IUNQyM>b$ z&wlx7(B;1POMfaeJK4h;&B)2;r<6CX1KpM97468wvFJyBD$_a+=&OOerxu`Vy#6^Y zS}J14V+1Cz9hO*lBRk=^``XTuuV2;o{{UTg(638b<9h5#>bv?s`#aF&Z$6!ZXLOn-H4#OIM#a6 zlZ%V*3K(8&8Iw8%ZG}6ic}hx&%bNualNA8eV}zq0!P~2zg3jA)=O;~Kjh?>bC>uDc z`D;nri8D43P8?!r8h8Yh$sAFgd%FfRdW(FTXWMhDt)lqql^5w;qQU7oGsj4*WaGWY z>;Xb@n6v%&IVH|Apj?l|kMAT21zhrVBVs&IjTQ6cKT3a_LX9s0+%F!aw_x)$U!IH9 zB?K+QmoQKwg#nZu@Ix2Chm_z%_5lxIfQ(g)A^HNNP)a1E>Yz90KPwxP#6EzuF4+3i zG-CGhgkx04db#Ux+lA4`$;oBKfO7*m7Yvv`Kle&GZ88=uR%?qyE~_>8!J0HWDsP)D zz4NrmR5B3WJ zWc|B_Ug02(4D5! z@zoM7*Q1nLTZv)u0&!K#G+CA~{l=x{a&ChAnCXN`i*;9xSY9vcQ!JBvdiiP1e3xEj zu8|dl>l82T_T?IHeawm=D}tHqjZ$dvS^UkYOw%LG_kX$3UBW&!oM6s7&*2IJ7t;bS ztWv||bg&xnk}osgyKk*s>w{-pET#xNO+y|np85li(SshO`{g-;`XuTLQCqck@9tyb zQy1hiS z!`!^lFK2feZdY9P>#Ff`);s$s@&0BWP3>W4k5`U9;WhWFGe#2OcMR;xCF&aN`fayo zDm)e}oLR|!nbcufZK`H9wHVA9i5!h9x`K1^jKLe!+f)X%z;;qzU^4dWPAu8`_P8?^{(upVpx9TC((B@BaRsC8 zuWeKvw+749FXR-a3FXsZWe_g_68SNtA@XAz0st#G{+XEnKVLi8plN~7ufzWZ`KgZu literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-11-$$ _detail.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-11-$$ _detail.jpg new file mode 100644 index 0000000000000000000000000000000000000000..14c848a7c271ff4d7d6fb496b13fbf2e4fe26ba8 GIT binary patch literal 6361 zcmc&&2UJsAx;`OP0g+yWL@9zaH8e#bAVNScRgexMs1z0H0fb1G&;$h(l^_x6q5{$d zq=_gUks?K^1xTm?lbkusojY^AcdfVP&Rg@|$@(2QPc@VC1upkBrrBMHV$?UAx=&q)PB+ZsK0!HjQ}?*5W?UD zhlv9W+%Pyd3~T|A&^no5f4_i#TrdVWBNH3zM-l0%C zz{t(Sv;T-TGq0I5i?}x*Dm>*8tHjCYP5kD)BuS<7J`rrY1O$bIMWhZ$%N&$dR#86zKN`GqfE$t$aC z>l>Stt?eCLFaZ85Ea?6#us`79hHx=3GQt^Ic5uNM0wIBOGcxTz!px&>#^UVFD~<|h z|eM>01h||+B`TnfCe_V zWpZxp`lHCjP;QstE?Ezc7$GVU1n>Qxqn{L|4>5Y6H2r^6U-n(#bjOMe=_l*_aEvdW z@c58nRA&+t(EqG`ZwbRVX_+G#AaQW)YTqllVq?paziaY2M@^n4?-+{k7JGx zQq1t&C|Zwu)xCS&p*!i=c|drxRD~qg5Phblchm{%cm>%?YP+C-Z5^4G;1bsicpg0eernKm^jjwIy(pP%^sr(%zD|RGuuiFMU+uXU+0KMdzL5^m-dTzR^5xz9 z*sy8Sy2|p%j#?+wXUjyN>%|+yb{V z2;AtP^V`vRksP?5_Q$wMKPRdsl605iLXXjq00DRo2z)Mz={qs1Uj_oK^mq`^iszy- zn~;cfZkixHTmVVoAWT^{OdTP$(>Q`LbQWC@Xf*=?w&WP}f=eHoq<)Al8?Ub*-c6O* zK15@XN0Wrnuaha5HK%PxYuqFX7kw54qS~90Ye&8yDPn>kz>lFG&=dlJN6Groau*sX zg$n!1M1o`C=eY{hs!s}}NDSWcUo9BSTz>i-lMuzxF_db}!Z2J1u9tRU zm!8~w8-Hz-KjY|+_Sx>2zBUNvK9^^;WTx3WVa6HxPC*+fizBQb;p$0E_a%eFvtnvG z_QhP(OPBd{Yt~Rk%A_*aK#Pgb>?aYW6F@-XLnR0}rZ;rie%9QtG894e&874kzYOXl z?|vA>;X-ra{z2TD4C7xANEhVB7FoRVV;B#?@~FOr0lw%G#vxHm#xHEg`C37+%s-V!b6+TN@Mfe}lyvA30^Vd5-?)=JG^A<{q7ExyWS zuP|>xXWJVmdm_2ztCLL{$m4xSQq(1_I+}l_-N=F8SEJhly$j;i6zzY)|Lw53QX*S2tzwi{LyGLT4$s3`ZSJl$ z0wKvBgJ*Z+n;!}D0p*rrCN_O*)~OCw=4H>5^hOf>)mwzN8CqOFcu}Rp_CR)2bP5ET zq9Dr#{P?WUkZn#xp%*-Q(In?zt?fVT#XoP$T)>AGth4bf{3XRfS#_@S{W&!)zT1@= zm7XCS;*jjlr% zhd#X3N~;Jgb7u0cauHf|47X1rd`8FCnB1p5ETMP|ylb63{HZbb<7%q*Tur=Tr{@pr zrQ#*=s>(Z8ldSF4kFGObb7SzB^Ecreb01gk^+SFiYbpq;s&&Wg<@JVPG#@;$GO|bF zL0~42vX}O)fyO5Tugk&9z#}WkTnYY4zTSBwZU%0ZXPWj@rFtE{nTuVycBJba`I~v7 zNqf|-{iI%r{MAd-Yxz{=cUhu!e!~6@1E-U56-K6CzPo55Q_<@g+~@}(udKYK=}cMz zNqN()ZO){gj>dOotJ~%v;6~Y`hf>u*04EJ8#6xb(q~L2O`XFcO-7@O2F$2UFd?nAx zeoQUuTv{B@>05DjuY~fwdNKi@uVE*g(_))X1;<_5A@6c(P}QN^Q^i@{bKP`)m5il2 zolvhm>_Ck42s7Z4LcKvPQv7V7*m0C+D85t$y8~$4lwMVa736p#6bIMtE=NC! z6}ptds?<3sc8gHkFYB&S;?g=|oRuLyq=Q3!782xrtLK_ximdHd$js$4drt~HYhdPZ zy+ms&CWz;)t~FT#b^S!21GB@WenmeYZN#6?s4bUjEN}z70&6?rCi;{183eE>-sSj=g&GcUKGW7?=i%vHI zftEg;J7qOsLb}=R`GcUc`K_~yK8ccq^Qo(J$F_oJ2+FviSN|2jQ9UD>kEaLg1{=7y zqgAIxBmxAwxj1pZe)#?wsr67<<~hM2;qZmirNVSvw32?hZr}3BnVZu0_9MSXTtnwV zWQIoYXhB<(M0`(3?u#s-kiG9-S4Le^lBd$stV~>lNz{~8sTPxY#B$o00SJU!3{?@# z2h+pM_evePMDeXk)mUMf)#3^@Y0poXxI`KrVFLjh$VhqZ9NEc?uMU!82qi{W(^tld z1mc9AnLq7pLNCpe3t#$ls75oE;fh19y(gGg=uB}@Q{VH_hH-1YJuF_o@rcdHTx=dC zUY}q%sPG}gW{4v%U*o36&F`Lic!2a?+Hx?A9UHDJ7GR)Al?s^uZuPOFQu;ZaNET}i zM4%}d&Ql5pxX+(49W1>Q8Z#nM8@80>P1Ve*uP>lFlT~A%U>%JR^j;M?ns$Kg$0WNz zEvnDqV3$;^*tZTU(pl|wgOGa$U9FF6czHkj6|8ktnafvo1{Owtc#0O{5ovzv)8NBZ zN^R@Yt5k!))o^L?0Yvl5@80d&jin|}bd;_3Jnnh(LRUNnqjL+Q`2Ga|velpD=_Pv* zNa@r9fzpr$TCRhNulhVS$(Dr{@(BbMW%HLVI8iS5=injYX1#iLYVtxw`q3xfSM(e( zldN?Axp8*#rT99Be9yCqfkc~^_t2@Y=rQ5TdOIIb|WxoYG*_(4S>jFT365c_(H@*F_Q~Ub#0M_3T?x z#h{f!E+upI3^^gwHPdo~;KiSEpQf-!*bXafo*M&nhpi9vpE^xDiI;g|pwon0QS1Z( zXAKKydsBOe7PZbrlh3S?oeRkt2;+ld2JwiTnwrWNFH`51uqBJ19r_IA zqAU=Pvw#ARq%xI~SBZfti^NZ(lCB*LI&=2~d!)&sGH^9IZvzVm^bA36XtE7SmrbS^ zQ;{H`;}{D9kwQ31;WTyoUro)Nb6 zKB+4GKGSSe+w2|Wd9ChQEz#wh5P!RqmT7e$u)uCrB-grWX6>2D^r`*s%Y~s`ee##5 z473v#S%jPNaZ`%D==ntuAR$cX2dd>-tKE`BPr1(Ls4f(nPz=jWwtnEKED)>25$QaO zy!5aM5P;1<)a;3{2{^VT^>H-kIr^tLl7AkR?hu*kkXR$f`}1%&D4?x{d@ccsuo{EV zl+0rnJ|2T+D;$SI$5p+XUIWn-F{|w7^_AB#3t>?$FVwH1s&?lZb3D~G`kaqcE0dZ z#3XK#G>|DZ7e^ekdwX#;;F}Vr+BLwZLE#lPs93GeXfg$YVLy5Jxd&!IU0+!m; zT)tDfY8Sia-8Q)Ew&>RzVz5pUX1YL72vuSVk{di1ZuLwLq1ru zb{;loL|f7k*&A+pa^B`n6ywYd&F)r(Hp1ulya~l{Vycyki_M*@PCD^nx`BMAEFxq| zQ;u6(Y|afUIQ_tiHwdJz;~IlGCt^WU(|KHt@kX#(uAkq(QfvG9 zr0WL$Q%>ymUju?K4{gex#PE6ObBr-x_AN)AURcQHQlJEqHdZur>n=*i}5UDzv%E90R=h`_j8i0Z+&0e;*tj+0l{S_jvz#zV$i1q#NDNSOdEocn z)edXe9AI^1Y=jbG5P|KQNtcruyt5s_tX+b4t@VSV^t*TjD`dt2PCwYbBDZZ+{4BdD z5_H{R+;{1G+!xouUMr=s4I(K!2Y+yftN)aF{9nEMOS*-41Neu){)!)hvSfz;FR6a$ zOUEd{>Gew>u#x6B_|sV`Bvz;4rpk=*GEXDl{Jv%-gZ+84j!!0U30?q!XIa0dXS(*? zz9Y%6Bk(ehO@F7A-?Z~zf1bs}9eS#~sCaDgfn?0rag;vGth=SP?{_254=f3|85Thr z&K9?@mdzm+*iy3*M;1KQmzUMyr(y9zvh7n{%dOoK3Js#<$fWh^&z~@7p1ph)k4Yc) zT3YHGSFXDGrkG0}drohV1GnoDS*`g6Hnu54&OIf6m^ZR1c8|M>M}VsT>E}sbbIutz4!~M z@^2~gfAMMk>R?6IvVM{rHWMS5-erk7e%4iz4u$F#rZ5Y?hvMz0Ui?5jy!y;;EBQbx zef7wBoRU?3HisR%%{2Mqb00FQSzfhsZ%Q-&#Dh>2TD^npn0p}?OVp|vP@F8UT)Eo+ z{^4uI>DhCxb{Yy-%0WQn zy9{yU@b-g{2e?1Bpr=$xrBI#Eiq4cr8FJP|^u`WMkIh|YQVv82tWKr4;vG)Xru~m` hJXt!=s?o$KI%K5FfA2$D8e~rYMmg=*tz!X){|4CJdL{q> literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-12-__details_.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-12-__details_.jpg new file mode 100644 index 0000000000000000000000000000000000000000..4a761ea05845df5673ea5dcf0886ffc12fdd10e8 GIT binary patch literal 10228 zcmd6N2{@E(+y6aw*<}x7Uqjil3`1f__Uv1Q5JL8KM6zU;l*&-{EGhdoNR+Y@k!?tJ zMrdfve0R_LKJQ!4`~LsW^L@whe#dt^PIF!NT=#un=XqV%@A{p;dx#UnIe_uJj-Czx zfdBvm`~!$nfEGYSK|x7DPDM#cNli_4jFyFtmWGCwlleFU3lA5Rmxqg+n@>nuluzKK zAUC(z8S#^+PRq;7Lq(O{(eA6NXf`4D5RaZCnYkvL#fkD9`p)t2&Z^y+aBxYpZ&AOL; z|3S`^rv-&Y#U-U>FJ9Kv*3~yOHg&%0dfnag=56or$mrOI@rjR<^O%Lj&r4sHS5~n* zUw8NRzu^uJfA|FfNdM**`2TN?{lPCL&@U1)GEy?iAAUhdfA|vNNO~I^bLV4Nu zxPa_UDweb9`PCiNf^w$YtXFOf9b*%cpBKjdaP2qG{+eUa|0B=-;n+X@ngnP`A>iPV zG68Vl+o9N#sAK;zt8{wJE7Uevi)5J6^zs&ND-N%Fp}Rhn$@eHqX=5B73;6GvXag+& z$Lt1*Y41CV4kz&4_@!62vaVAm=gzz5_OrMB?9g}oX{6+Aw@>a@D5uxD4I@Gx;J5BR z?#s9_Ls)7(ZZlNgqWN^%-6qMwtFYHI-7jWcb6s<(QHj7M(kCt*Zct%YndS4+`XCj$ zxgQODN1Yja5#0V72=X_DL{a!{y1M~S62<;wR&sB3dL-27xHP7bp6ZQ~;3@;FR0y8H zQHKqBwc)6rqcmwPy+z9`8|{>K;j^w0$yM{JS}HDp!~Y*62ta*)zh!m?T*o@hA%@gddl`I6_!7j1u^?=iQ_WYwK@!%}m2YJu_VD$|wj-qZ!I8eSn)AuPE12b;U|| zr>*61k?#FBt5e^k*RkyjpT;#Eg0^=i^^G`#m#D8hH#)aOQM+_wU~O_yvKk9&Sce1+ zau$=nW|xAs2%7Bhea2=YFtla6hd4iCdA>G-EbR?%i0fO9jv#5o{3Bz^$M5EjbQluH znWp%8xyJ&pop}nCdCc@3y)7bPRMHn1DMdzUd-t)sFt5%>&~nYY(EJ04JB_|+nDcQr#O<&M|q zEEhYUAYK)0%v7emJ&Q3!`*&mg(&XGUQy)~`J0mh$!907hp*Hb^n^&EL;N?jCNxj1y zU{^x&b|TF|;c3KzmpkfFHt+vGoFQ6ePs^>c%i9@-%viVhza zJwPg;lNI7}XQ-XE-J}#>3S=f7p&s$(ag8XBv+Xze&EURR-?qghFdrx)VM%f~38@-9 zS*-1=&sPtF;yue#lWGI%a(Xcsgq!C#isX57(r-NQ;OSQQ3C54tf{yjEC+)qg*|>E1 zB^mTAU7I!#IBH2-_ie*>k>P}g&V{U=OxI|Hkn1{bpw^1e`vJb}`vJW*h62ZNkkU!p zclM)A&k?=K_|f4@BCr^m;z4IRBPi4+2fF7+_qSFa+P=mUt3gx_GMsA{AQDKwt5kv| zd{bDH!d86df%t-#TbnKsFasZB%K_s0tApLqic-nE6G_>s zZp&lHabY-k#M-L;t+&Jg^w3Cyr-HTXl}qU-NqI1W{UfzEZ@*`Zpi(?w8KAmm^Ab1G z6{<$4@GY^mMtBY?book(2M9MNv+zzMt^u2Fe}Vy$55_^SJhr!nzYqZp))7m&DkAV^ ze15iYfDk3|v`MxCujoFc%{mzJ`^2Hp?!l)VG939d6HLavf;}N1Y(wU|ERXt$5NVmpzJ@cj9gI}YX}QYqRV?nqh9B=qB9K-bi(VHa z0*44cGf}s?3)4>~1qp*#Z~V>>e5=&$pG6Oy#Krv)CG~2v#>YqbnAwBRPmr^wL;oy( zNM;FhkNY4IB!{kn*%AR*jnVg28~9Je83%6mFbBc zFXwpC*|o1aJ@90}W2IzWUd-~yt08lMXB>z@l?oJ%A_GFoIgy`RGs^jh0Q(FXLwNeZ z7)OiJ)%!irWY{@J)PXxN9TO9?_%-#KZ7RJ&bL_+JTmcJkt)}rjEi&2+2G^>B4CbDi zoc>q8c#+s*^;gkG3yBZ)EIcYqUEqxw3*TI;JEL^*h!tzOU>i39S^Wh#lLY;el`0XK z;?jjl$zwO~bUO%2P@BTuK*$0G-(sTV_MH9ZS?i!!ELo7jsm(luh~D91WWXRU#qG&( zi|~?e!sGKJC+;)MDSCa+EI; zfqTC8-8*5@p{Pr`L>?Bq{xx`}I>>1{^MWN#}l$@fl}SCF0A4n9dO+-sKk1xx6gU zIYz`4SgwQ;Udg%Pv?w?!oZmv%_!d+0prVv4kiy7wT(#?&j}XGQC%%4)zCno=$XKOb z*q+JFh-sp@SB~EqqMuyHZ1^Eima^Pizx}aTo5fOd@po3j z$qwC7U54*!Ib&mH)p_`LV`V{k(Hr*WfS z>!Snn&KGge{J=!9Dy5z1M6o=QzZ-3`HvzuO@ckG9Bou0a3Oyc3ELcKLn8J@BYt@yL zP{{J-Hju4LI_3Rfa*_jhF=|I%Zgrfjurgmr z?@{W!DLEjWlk)0pEb_xt0t`FtL#2F}A>oDiW|`BGO|bna(aC^~nA7<77W!B?JJ#Q0 z!_4G)oHX0IO=mX}1rr2p+bNVdojseR@3yqAjg7<(757>SOv9({%Mxa25(Bs#s= zKG?8x(nxA=i$%4J(@La5eBvoC6@C;7j5kFG@CtG}*ENnukD^&Q`{u?LVmXlU_zSin}t{&&D?JAC~ z`O-iI)5Ijn-xkjNZew4*Q1(T64_~&0=L`HefEVHwSs*;wAp*&OfA>FWtc@|L;01LnFQj;cFG@u?9Ol6_Ig~d?%c+Y zRi`08rr@r3#Sq{gq*%%*AGay!MfwT%Yz{cZ0_nKa;TP*B?`;&GgKRlo+SxrWfHd$o z$aWY9eUu<%yI&{0q%i~4taii_+-F+m^JvId8rfZP=GMA=fvs-9EQp$ zCal+Zxf4%3)K{j>i7mbGRw%M((C?Ao;BMdbVExxkj!892wFa_wr3WqvZFre!E99dO zD?k#jb}%xfCZUbRqjbTW)`(Uy$TWVdraq!n9EOki3X7O|nv@9a8L2C+dNtc%C z*RscB9MO{d=Dg_M=aF2iGg4Uc$4f6U9GB~=TYL^9N-pSeM}8T^Chdg4Xf2uMB1F9J z*n9<<-iVe|azT-*lsNQ87Xq$?PlboA+;qTz($Oi(YriCWf05V;I+p@-3o$G9lxb3| zOPRr_*MgX|?OQ=jH#+xc-Slt2xOz$P3t321;F|4WOQhWM5s~JaG5L3m{3qU18u-gX zMoENW3&kvzw#6Ryn*n!C#Xm)*CnXJ!@V#jOF?d#!2vo8odsSASW&NQ}jYpKm5}3h6 zHE;QcJS3Qdp$VXa4RuksOLzDkj6x@Kka!|si=6#T1i0F-Uwk*OzI!0K zZ;4_^Y0%JJX!SvP;1oda3Lyf~cTx!gQEHyo0?DFeDiYLyEM#|uW$KiDIP{Ehygc8F ztrHCU2JNq0QvxlY|C9-V8PTtZ{U1bd)OPWw56E>%Ul1BtV`Og0zm=|(W{xnWXh?e} zFtr-Q#P2t{VF|cS7?W|Gyit`Lt}i&%pRT{}EscoHf_gghl7=`vXJL0s~;UJl}BLW?LAi3u*^TDeQZRDlUg!Nq)b22XF4sGo$ z6I%vm+w-WrJwF?$8UUPh1=aF5b@Yy%GE_YrZL*W4N!IzCl_f3u{!$V*IjPdO*Lony zVS7LXG5kfE0-vz$2a=!HgB0eEA7Rua2D?}H{hE7U>v9}C_nTFX zwQfz+4bX2+6<@So$Jb02;O9`;DG&O(dK9LKoIE$Q{xkxw&Z$NT&Ogf{1Iu(-Tk$<#OjN+$o1WSxSazUFeh zC9hX&Z+a+l(g78Su&2m-f7u$0bWQuIRLnu?P)w8qB)?Ade`PjPn)Sx{<=%VOn*yniGNh1th z5P_M8t>2!)F``=4W+2SGIVl9#7ryJ7u&ZW<$b2?gkH!w0b-_xhQ6rK*wN*j$j$7*)#b7rr)mAd z$+w}b*K-xz&2>nsIR&>@ELQvd1QrO02ueZ&I8nsb$|@~>sg{dkd5hR_x*#o$`Hh?Y zJEc+bKGqc<3zH%OmqCU*teUg3_Qx|1@Fx>dq`T6f#_)yL_~$)2y@Ckz5N?9kg5t?8 zWo386#;9QWOK50wSehb})ztBLZzTpQo9 zk$EA;wCUklm6(+?zIO2+37zY??t)5ox2G~w9d;pw+FGu#Y(<{|sg`@gx;Jw;siZgs z^E(6uCq_qq(KCMP^S?d+i=-l1^hhtHg^^B;Bty_f6R1fV!-~(BKMbC zbbEWPEtx#kCtr|DM^J?30Ih%6`V6V~mO3=aLG7ic0DcDWNkRobe3y$-a}bc;qs4p9Q>7BXr%PF0MV4!#nD1hvrEQ(#Cvi zI~S`{dzA7Y1ybB5l}#KJIb1HMAsVWQo7lX#rr5<(vycQ?^=@7If#XD$zwkyt$n-ezTcHv@|xK? z_=$*lX3q_Nr&zk;DcU2YJo>}LxmiQ34Sx{!eQ%y~X0dG)96Nb#1aB@unc5@Ul$+EP z&`3!Hc=G7?6Tf;Hgqyn*IeyY!$js>ELRGI4=5WO$)*!!v z|HOmafk`PwwLpNsCFEIg!bXdxQ?WK?MIg~b|4w$=gof_Uj4)~X$hWwiV#H<417ldE zO4jJE%Sgo=FlBZlZGFJcvI%G;d^M2Tp5D8KLbX3kHd&%{yh$pocUN#G+TMSs-zLI4M;jsO zAj3R?YZ~cnD$Kq3*vRuadNx{<hTRIw>O6v1SuHh7@xA}~+cRNvC}&=9Ns)P?o)yJ?std2?)# zvL{>8BiETIIe*M|n7XC92X*+C-KEU|JPkD2pyX>v#lfteG$TQaDS!A=pia67=|YTjyN`n z2r=|jz$N$?4Jva)l|5+fTrT}s;n|Gf9ut2x^YRT{@J<%ntqcUfVD!JbrlK>_hb)1`P$?VY|0*`ZcHI zo=^4*5rLqA+JX7u8vo(h$;0Q+)msUv22&JUD+aCG277ypmf+1z<+N$j#63Jr6_rh2 z@ml=QG?{|s%wrX`7kISUl{-I9rXaV*fUC9-BKwN#;qEFAxAs=Qqp`Cq_6AO=WUKn8 z?*}*G&xba=s2Lz67N3bxuuK>7=2#UTQMLLw`PpwA+LT9kt#|xPNT3$M`r(^fs0&zB z4=d-k`V%K-b!qg(Kugs`#VK3mfy2tiQlag@3JI3XZk0Rvs}-)-z-;f$MZjH~F!c;V zD6zxmKA31m+q`=r8-Y3}2zT=+4WD&fm`3$L3*CF>rY%^gp-BNOCLcXF5QO2Ps^|cr z2ZfI8j^)W?u~|h%gtK6J5;C?0j^3V@;ep%;rz*O5#h^(JMdv;fxbc1;6MpFIs&K`T zY}G7FF1U$c(0az&h8wvk5r$K>NIwQos%6vRkoXW-z(fQhOCbLUYWl)RU`O+zh4(G+ z%|{9D6`|oHxOGv~2y9<{cFglmw@RY4U+l|TZXz&S);K>)Ege&~YV)Y6a3((_vbwxV zMJI}kjF!xthX|YpR{kC9`R!ael*`V*eXi93GwIHsqk7djLs6nPLX$_LDjx5_aJ%-g zoMGeemJ4C{u9i1U2YwRG&HSEf4dj#BtPd;KX!4eWRLbjy2Y8yigE5lLJlKXPqdE_f z1(POJmpZ$!wCC{3buL?LZrAt@NM{B$G)0WG1?Qh}|R|vQ&l4;?}{{@#xh^-|8?Yl(a7>o!oBM0lyI9}AeeFQ72 z`ey3b%DyrKybHA}RYjmI#+m%mer@^bM7w&GQT3Y;0=P=@y^z~#@XpkKHn()t=Iw^; zg+`=k4PZr&Xh1Fh<_dx3=hjV!V*95Uo{g;g8o6WW^e+yFA91PbN>#C8tPL=2ERwVU zrKpglhc>HkV&V&hTP&VR0Vz=Mb>AOZ%+p~{{W2uD%~6;}OCz&X(}~y+W)Yu)yzao5q`7fhZTa=h3s3vl*=|6$@4F$t z?#U2H^+C1-X6w|q6*Ua14GQB!h?>iLNqxG999)X8Ubt#8wl n==1$P-^|e99{1K}rJMe=Tue^212S)Mw!iwLf#2I9i68$9MjO19 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-13-Transformer Block Architecture.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-13-Transformer Block Architecture.jpg new file mode 100644 index 0000000000000000000000000000000000000000..2a05d074157bfdaff391fec693eb204072197642 GIT binary patch literal 34066 zcmeFYby!?mwl7+^LvWYi5+HbR2*HB~w-6vm1qkjG?oNONg1bv7T!M!X+}+&@FRXaA zyL<26yQ1Oxy80saGcTmZ-b z&{0v*P*Kp)(9ke2&@r)zaImqku*nGs@rbA?XlbY^sHo@|xmf7vpD|ETu?n$0d(Okp z&ri!DEdGL5jEj$-_qT^2U|?WiV_}ow;E?e?rFzQyKYl%a2jHV4yhBVuLZAmA;v*p8 zBRuv1XyJ6CApGqG_{R$Y5eXRu6%8E&6AOL=_z3_J0SO5a83_di85w@JFZ?VC*(s~8<<;IzPGZrv2}KFb#wRd^a=_N2@MO6h)hiSl$?_K`Ab@EUVcGg zQE^FW?YBB`eM4hYb60myZ(sk9fuG|OlT*_(vvc!n>l>R}+dI2^`=@8;7nfJp(3{)e z_(A|6{hclN@4qwl7ryY}d?6wuBO#;x#uoyjJA5GFBcnX!K_!sZKr?eBq~{GlCwiHX zQ`3pTz^4fze&;lXNy5m##&r4{Yk%YHzsFeM|0~Y^VC%HJ5a7x~!UsqKu5Vd$ z{W1SrqtaRXG&8Gc5u8!04DxCvv+H}Ab(Nl7n7dGml$YD+cIpsrA(}UOlxs4`G_1DE zCb`Cb%Kd=&0d#Ux;a8+um-Td9VAMPUI=17WO0CL|0I77^ zJDBxtn`)ajOpJ>h7 zzZ69oTO2bI`j^T7g{-N4H>|v-+(w5#&;74dl}#R!$~((`T=~=7|4Ni@$t>6Z5y`*k z>HmlXZlVAGvl5a&I}R^5_0R|TtG*cVI^Uguo?`JIQ~}1u$(NORwyA>7`J)~3GHJu= zpZpj_Q&@8&`=|BM z^sz-Pw3^;&MFk2Dt~iDoekQ;|S(dw~@5E&$l{9J5g!?jPvP3%T|K+3{k_PRq6Lm9< zH4~;9848F?jOcoV0|OV8TzBgd;KC6g%1#iMS<2CUBXJe!j)U~sU1Y?YXm>1{2w7(X zmT=c0Wet*R7;)poTXw}p@8pl@yU)?Q_H6T*m|PHKiL?N@k*x5s<@yL9mq@PY&9k`` zA1?7b^&Fv6eL`CtL?7>ML@E6Tgl{ixSRn0GAg>^C&^nawsL*MGhLzYTd^_-a(oy?uk@%QI`xCQwJfM6@kFI8U%;-AY zM;A3LcW&QN=IV$2b}aw=orhn{SBhQR$s{)^A`RK+n5^ZTLOoH%&@7!X*^=nJHH=h0 z_ha(CSV<9;f=0@^zuod5qQcK}CP>@KRfWH##e3Qh^`_0bcl)CF26b}``tLjz?d4JQ zXjUhOF>Qb0&d;xlmRTP^eLMCeCpqk+b*16j)?mB;bRy6Dx-V-37ie)ivL53)Zu}zq zthwf~6amYJ90%*=a&9+Lh8!++BM}W&H+7}N8m>A0=pz4cc#s?iCcx>3{T6HJY zM7o(C@_qv&dfyevTKW&Y{HB|K9aZAcs7C+;OAzSr>4Q!iX6yAj_kDCGJbYA#T^bBG zbYA;T5%8Zep%$KsuesCT@ zj>ko~I9mR4%8=fH`3RUb5UDwRd7}OelKSP{c|U#z$-aWbZTIj;Z@2QT)^u;|@25Cf z;XZu117(97B_=Q?TszBG7*CH&n_$VH-b|M1r;wFUR~()mcir>ecAjffryZNCA!m`L z5&!gV(s&OCHPZP{@r2vxjPBvCR_`7AP;IXYeCZ_4Zo`Xv+iyZ){ZahRkXyqzaNGAgqmQ1MYDtUL4vzqBSGY9{ z*rmfd86E*IOU89fe7{HDY0UU&uOAFy$xp_9F)J*MDk~o4PZWj`kEp?F3Ym`ERyzo7 z>M06K=j)S7B$KEn==7f^RWrI$#lgV3Dsqwt9(Q`Fx{3_rU^o4omN(BzH-vTiCt;t4 zv~+^j(IPJlIk6SRLo+^dRaS}B7NhUx(K*BTPo`m)-{mJ;bt^n(Oh@_K6bw3v!uz&H zh2|z4mZTRGK}xBbji<5=(RCd-Hn#(u%o%qFU(C^v37d_Jgg;;-WL|>0B+oQ_EI@nt zlD8PSCopA`tVh7x2Dp4H|>5yL=X+AeB)LvAX2@&Ppfgz77x8NN+pfh|x!59fweLN}}J+!NuLU7S%c_`xQ-5 z(oBQF#o!BzhMI;zn|V+37>zBnpr=SEaWfT;5-7B%I?&Naz{Kq%;BabLr!FP+oz(Er?~L zwUw1TPqkC}Wa6BfP_^Caq>OXH~ zFRL3etw$RYyU7@F7xAXk9VUSOL#n3uP(*K)mm7wudShE(%$j1i03P4QK23*(gELIE zPmMc5>L`0_nepn3myxl5gk`IHU)=YaLtA>>{6s@%x=kpp+B|m@xmfh(G|47+7~|Rd zYJ~2eg7D2}s^~xMk=926nYx}jFwKw+O9Wl!v>iSy0mg%rPeEbebmpKp>BbZ9eN#0S zk#&PXJ_Me(e#ud^;TMkpY>(?cq1)ik>hb6sv|e%xT>*yN8=Vyg^L}5}UN%II+qX4( zm|AGKZ*HUw)IhGjRQaf#8tm(oy->(|vve<-gd7L*n_N+qpSi%(DuCpBU1bkd0);|f z9KR8$&%#H25r6Rv6j1~76)8PmLAE8Op`GRuBP{))pZzP%1#Z}!?+zOFC41beURIQd z_PmD2((3;b0>E<}(zU%C0f*a%q?<@qsP`|g5^9{y&-cB=8*iX<*2nN~2Za1Tg(86& zC*Ip=t>SDv&TqqC)>V{Wr*56+3q9Fs_k&cVecRy1mB^0peXnHP<<7vZ4*XnuN@d%% z@W!-uu?i*phH^!`gW{t4GAWK@F*IhMow2zg*O`{_`@xmuN&p@@6;ddNl9%s|Xibe|#(M z9~ZOH{y#tT33?|v>X#k)M}RmX?3TzA4p>KCJp$T80D>Gq58@P7h8V*WWGTMOB3Yu3 zeGHKTF=(3$+HJ8tY?b)FgraK=DcXYHpCF2@`P~5iP$hCGy4{2Yk37L;7punYn@|6{4B&sI zRYm>XkP$9*HQH1px`3$+o2H*=*X_H{i@61cL~iu&Kh)HC7T;-Wg;hW|JG*wMP9TOlJI*6 zd&CE0d9U4zld$@<%yghHR&Bo{!OzmwGFFBc3^ml~hrp`SYFbE?s^dro9OvZaRr&)& zFopl)WWyb6dT=l&jvrl#u>%-! zu;lrup9`p~It=E8H{3=N`JZR$_3>OIhV%!WfGIzoEtVdIFlhF_U~CwB!$;uJl{PUd zhJg#tp0j;cg`6aZUHzlLfDH{dm_`~K{hp)o^am}Z>Rthk*`BvlJrGSb`y-Q47ym}b z`Tv-X|CQWgLc`Wd#5i&dhl&=4JiWz#(#v9!h`QtSY={wPOH}docV_W;O_bl%kbhPa#n7n zfg|64C_wUeWtl|DQjH&@W&EyzxZjmElRGBQ-x9E6`nyr({XOYyv_dt@t!;Yh~1O+8!*4Vk$(D+KIBFdgc>!8$i({J!oYtmrsM+&gobf`;E+_<=T3Z! z(C7LE4r{VVCFcI(L*_n`kgR(~7(sn|i*D$tVytU|(&j3$**>{ay*$-SeLS9q(IX&0 z3SOimeEDMMkO#C&|C^Vl{h_75`tLuDObTj->KU*5ca`+nJOc1s?&R^9$>MAeI4kJB zh09~2XzXc?X?go=$PxkmHjMD1TBXaOlD!6QCC)DIAq8V3Ga6(`4+JvT7K^CsMeT|w zM{))0Dq2RTM^&Ns0v>erwTK(KZ3^R3L=UsN9~oF?itlj%?vom_03r>e{@7NXWx~-! zZ=j=V@zSsQdY#Qv#)=7v?h1g*g>EKyhy zr>S)RmMCr=NBH7FZd@mD!{tN5Km*Kx`MA{A8mTb1sl_3j2ETQ(FT-J-Q$o3RL7T+v zTqs){F>7E_fWuHxEo;?rwB z8fG+x7bDcD(wj5V#7E^<%HBA?cnxYjqj(!xb!XgPdygVt=n7kIoM7x5#7~sp%M=Z{ z_1pG+!~1;8wyQQ?dEm!7dM`xfP`;m1|5B}y$%oGuO-qVmQimRR>({V0IQW34SLdQ< zYB%3#5fYe)g*$r?$Pqsy@st)5s@r`e2rjS{$(W3Pc20s|84`iV8^RILZ}JIY?~RTj zG)^{3mSS+fKC1Bf&n0sCl*Nspm4w%$_Zm6S1V&%VH}!32FlfZt;3c&z066w{#UqSB zXJE31akodmE!>ICnCDrP&=pl*W8# zwd12N$B2*bhA95pgSldWcKeuvc-(x!7y57|^%CuYrKg%Jf>U%(7-*66kd;{e2j8_| zxwk1F!u6CA`*}1{v|U)R(U{0M%vs3Dv)$a9CgB{R|4U5@UZdK%hZ{fr{v%-KAr6)- zYmDNV8YAXgG+0ysZe|Oqe48fvrAQ&gjA(zeuZe5$48Ox#(~Hp9Ay zHZ6RdF^U~2aIBf29AWi5Uf~Bh5$^De2pp~l{I&K&ZUV3CB&g}bi#z0xfD8RH1JnD! zXB`KDaNli`r10oh1CP6^i16ENT3dh<9Ryi(BIRU)xfxjBZxWOxiU#QTux*HoYS-7+ z&O=_0g|C}4I?40V$Na3K{*}?Oc*t81v_28fQI`s4R~|)d$9art>%x$N zKE#xm1#_y8R9xQg{lzgOS+f!UGH@x^Q)+OkLKb9|U$swyldhDN<={xCEX(pB%Scxo zQh*l-20j81_fKD?CDGO!E-^cL1UnDx>yq`)jJJRx$pNQP9=V>C_T0e|uP`{+Vwb#c z-m=nVIw^{)2=!#*P28)v7A_m9xB|V4RCWxP40y{Uy-S|u6N~q;B6+WIeIgmpt`2k*}YXgid)u_YPEMvzAPM`U{&Pna-I?dwgg$777pK4fTA9#}f2? zwrpl)yy+1Th5O7^5*0{zYHnFpHBWXSHQP;Fh~uTF-UgsmDk`%P6xmLdNG~xg=*>T< zS6JwP(s}_J$@OIrF%#mYH;6YS#7;}0ach-L%wMP=1~SEkS)9XijBy^gKjk48v@v#7 zU0GZUcM{y>9x381lzx3&sR)vdl=d&L`U_m2LzQKMy862{X65+v$YZj1w$VyhDxHVD zw*7@iNTJHz&+ngiG^j|7Qv#KlQ4(ZRUp&Jz>Ce2Jj0<*ik|hq=i4Zs#B4I}1pR1Hlk=iYn8qnA?ev?T&vC`A*)?C5w zJZmUW{=DyvY)zXSU;hVuDJstXu!{&kNyD&-ibx!)k#B>PuLG_^8FJ`ozycu!ZE<}d zf?gZ!kMdj*RIv%g+V7X5SKTY;S_R%ObDI*9Tiw~tQ$9}*Oa10{xIsSb3y-#;4mkRq z>p1c~5scec+x>@CHg#K{~IDTI=t=niNB`^tM{Zaga;d9k?nk z-^NTdn?;@fe)Ivr%Q0prhPHO6hOW34U~Mr`fHrha$<08coC3T0=zJtaCW$Y`y>W*}9t@vlFO=~s1tfq@1_o2#bx9s zH-e?FqA(CoqFr<}(dpc>an)Y8Phgt})H>iOuRZwI&19aR4Qe6>`=Ld!<`~0 zSVgAJx2L)6fuz_DF-{HH%Dl9D*7^nLdrKxwg?P50hBsj=GJ9$0sh6tQ>dkU`~%7Ott zYwO`m;8XG5)?KZifynh*V|G2t)!YpxS-maoz&xnwRmjw8ig?$Ut6rWCt@|&a=Y21& zHa>xP%C~kV=+cQRI=>;1l9pRFl3$2#fwA`6CzQYxG}|*&o&oY9mu0|d*gd157xNd+ zn2P*c`!X7o{Fg75XfPmdu}GGxj{7)HU1{21E4`s#?G(4-v-k2-4PhBv-j+lf#`6B~ zP6Qm<{B!txtCDv~+#O#Es@;*ROPPz4!y&VI?)$p@5VTB_3l>A zb#i`mn9WO#>Ok=A`$bRV7xPXwzi zM_`e*fSaA(`iVJ@e$9p^mFV)!mm5fjYk(++yQflA?r816eCQi4NpSfwJVnb9tfrGg z3u2i)Hqdwk#C}(6j#K&x&Dm(lQ10Qi5uBc;pHOjNR22WJ_2U?A*qOdsBlEe)hVTwj zDHOE9)~ec0UESOe$dSRPgO%(st&0u5BNh_4Tr63gFY}*-4 z!{aAic8s5^WRs3Qh7UXftV=%mVrf^AL&qxad2*n(%Y8pJQ*6Y(Eq<^4JQwj@svEHC zP22(|a7d_g08WdAO08YSvi?AOl%E;#|0MYwIfg_1C&ZuYlJ(+ zc_uQ2Ptc%qN!_rgkgcJZ#Hb_o)4ZnodcCjB-;cuP0cLfAcpuzHsA>x{Ttzba)2y3+hTFZdJ2c;imk547X#8LDTEZ>h)V3P|)Q z;Se>2BOYfhIO0(RT_8zOWkW|+Nyv>Ie_WCX(carPtKmZPy<8y(^(RUGvisTrJM|%Qm+K>jsJrKr9BsZx{1D_usuo(LAjxZJI;NK!pa2H>oKo zD8IbPJiSA86Q4VYV^mO_mIyQF-ESLh2S40zUIZG>0=*q_$~+ zAyFxC?15I}+0zt@07`Q~<5(gyX|daj3d~d5@SwhW3X>$eszsamm`}F_HKc+y{AyUC zKQvnd^Q|&lb&bnCl6t)QW4L^{*W_x!cgP%^mk*2Yr9#^G;Z16qj){2OE@NlaFq4Kc z#>R!>#g`MkxP77tsu)Yrcsas(@-zHLUh}{h-i?t+^5tc@*}({+wXk1kY9`6s6fW)X z^&@1fzyKPP0F}7R<;O?TErUyqzB33;r++DZVSH3-@ByOllRGty^)Ux}mRfsd2H1}d()Z^LlG}u9eNU)GXQAodlcH&p)f_u^ zaNze=j*%NV)UB1np1uq6bXc_bC~N$8NLRN0!3>?iRZn7y6PMcnchFjyHtqWlQP@)vIL`ukqF`A0scU1P*jmZ*^dUpOfNWlL_8}?&^v?qJ7IbfO^nhel1a7EZ`>-VF5 z+0KrUaucQ$jZPYVE#O!Ua-EItET-1ClAZ`DQ3gL_j_!zT z*S40!`s!-JG%+4a)LSm2eUnd`EK~tr*TBFo2z89VA>Xdttr;i*UO6u?24WO}BT3j_ zE`Dl^U7M3^Esn1@EVMup78uTdsA%%D%44l6DH=J0wVyBU=0@&{t^PU5ESIDJ{fJnn zy+egRe7CUT=g(7tuk&Fc0#O{UOX9uJr*iG(Kcr~CWaHsMv(`AdObNuNlf4;i`EnIK zEveftqm%Ep&JYs;qG~L{yTCjMb3op1{W4XON(b;DJ;ttmj-l(F_1uR_M{=2;{up9A zqTfa%`&-%rhtIZek4?EI)Kib#W}G02WdUc$yVTFsOE&HH4BBx5y|YO57RxpJsZw9A zF(hD;L~?kC&wrqC`?lEQe?9N`z0yPRPNzvtu1#<$Xtir!#ID-2bz3|TT^7AM*~Cco=E>KQdf%|+#>n{Kke4cxyjj6ky|BWE`cFsoKL*7!GO z6zZ|kEn-r!HUSdMtvC2;P@hNKe53NT33XfMn{D7CenxjdQf8|&j1UqNQXjT%taPra zC}3zu0qHf{Y$n1XWc$Py&=IH`5YUpD+pOYB)2-jw-29`yzNIx(N78Ms;OlVo%95`9 zYoXVLd|!1`dIn1)9s%Dz|Eif=>^YZBq7)mg5#w*6@DDA!c`pi60tCFfT7K5yC2kYU)}C z-kx$EG<9w0VY=l5_B%r$bDHQc+T`TUele@@2K#%_vfXCW;CmJGwc?zn@xm14Hu`SH za?NP!?4(pAa+<=n$Vk8`h{~hzt}lhcI5TXm!r^di|GA2*cyE#%Ax>AofJFWUK&SU& zrX3L)S9*}Rr#o}Y3??$%t_}Wqb(Y;nF8_j|qs=RX2x0TqB(KyXb%Qyh9Q1ZyN92`( z%V?*-Qq#*FVa7**J4?2>V_ILj^fX=x0Vn@u@ zEpVeU@ZWB;6}wxH&EU)m0gD4)hHm3n$Tn5lahYJ#v%U8eti}u0L_IwB7v}(WrU}hn zL;;Upj$6#o;KMswe645pTA`T7is%YI02^c__wJ7X#_OJrE6iqi?Ow$gu=j7bQqdTn zDl1KvKYQ8?iPjZtj#)ZGSID%~W5@@#mQFM7jqguL>g`&K(SByn+JUKyu% zoMCu|m;YN*T=Vf}O`k&FN8w(b)71L_&ald|BG8_N!00~{qd&zzG^|s~*WM0zwg!Ds z#J{{IUJ)tUiM}dF)BS{(97tz=j_9Y3W~7~kVDoC%jiRy)Wne}uj@_yyqkwJp##2<= z5TV`30zWgk;kTUDwdxk~x182!LUNNB-k&fc{H-#FVEdPv+HbiGF0glAM6Gbi)=g)I zv%DW6cCLciI#Wl|kt`5Fr5{?0s<_^C5;E^|8y9*g0D98vDp-Zj%*OSen`W9^pZ@ic zQQ6zCR!_sIImXJeAl&!i26zu1@a9@JCSdok6;Y1Vq#hw52?+|~q*wJRf-5$Yg^r0> zj#O{2#Pzb2_Bnp}5L%J66yLv6?p?ZSQ#_&-&%>z0`!R}ENY0>l-?dLjzNkiG)_KYEnOYPyjwrasdaeWBuyC3=Ilz5|WwwT_W zg>Ql`MXix$D;?}U#EUQedj9kU&a(h%$ruHun4I(A@OKPukZ!WeAy3JWR>?x>bx%9X zbgQy`bI|-go*B?im|4_gV$928r$SxXaV0}+)up0@G7jE z7FDYk(mL)Dx!%W`a+Cj|lj^fnF<%IRs)&`?j8=&cI|N+hZ%;8lDpM?NF~OBv`~1_7 zCyBk^eu!gtt-`07-2KjEuiCCYT}935z2z!oV#?VO=XX%rrFc97z`T zz5*K7%aJ$8lrtvK(lus(US1wu6TK7o{W2RjF*bQy5?f3UFSw4Ar%uK7OX}dA>Fnpu zLE6s)nVFZn*IB0<@OGrp!i_o-)8rwY-aK%@x+e_XM4Phxi`Z0U9(}9%nrU|x4$uUl zVNxIpdD7lKI;#0S%Yf7&Vv!`bbHA`a4A-}kJqffKF>YY->!Z)~5mj8LaouV-))FYq zlP*BD!v{5`up^S|F=2ueRloNm!Z5JB@%W_!fy!DOv55x=M!@-OFKTBm>bYF34mmIK z3!E9UY&=6?f_QM!>;U$TIoanItcFxhc&}$z;QfF{04`KlzHR1N1>@uiS6+Q4rT$f4 z$ZX6{)kA;433Nty^$OBT2u;Hwz|oR#N&MB;7vW2rQe&oYj_-Mp**_7vQ6(l>U~*3j zk;rj=pJ)u;Bnj12eXYW=<#oZ2a7>ZBUEW+7-~3Kr*^IxE+eG(f3UKop^n75%1jokM z)IgsdyfjI&-;Ze-&#YDt9QIKun6qx@hPoEnOdmMd;B0-5Wnv7W$)bjMxl>~n&illj)$4OClKPRB4a>)| z^WSS@ch6cex4!Lj!NQ9$ApobykC&Ir_U7AfwZ`#auwoqPyodU(twAg=^~+n24DX&9 zKQt{x=Sl^H;(8$YqrJuiTyV;mG>`k(74BWX=PR^L;CDM|z@hFsughFZEo!VxX^rZ0 zPa1?|bs7AMHMycp$LI+wc?Xcp1kPChwy_i`>$K`V@*FhN+W?D;ZN}UYv*Tx5n~u~U z7fkd}Ivx?VloJ=;Eo??Fi1kR+%Pf!CvMWLTsDco=h7aeA4z890t+FF1OXP1HN3;yx8q}8!23Ip(tr0%ULSbPf<6`u5BBhNdAGhS&EY<<{1NcCUQIEy zQemPityGa+e#c;tfK-gck`C|JR2z+lJ-@diU9yE5o7|R|1vW*d$GOvrdX3w^+p${- z`rdK&)SC8k4G-qep{zY+R~Np18r-gq zdgw(lG+kJ9xG?D^0^b2r_vFvr9$mCt=|=$iNiVfipIXX_ag4)}nmIjAZPiTtiqZ)? zo1NWG3Y%ph8)I-v#1aClLMOX7m6tjigAL>sOD%#_g3EB5gO|GAME7T6ZE0Qlx5<#b zxW3fx4!l*!BOn?Kdct|GcD3F|=}a zQ5{M=`R0JoyC!;rGgUj!S@wg}>;twv5(^rG_`xSoSSDxVSXTFp#0&1+8$Wfdce!qH zWauNA8N6NzCh4*ObEvM21O=o%BabrRpoRUk*Q>{Srf4hr{s}Bn+tVR z$&fm3s9YWyv;syUQCMwg|<6%SMYU}l#6NYjpvW9a5hSc)VzId z+3Y+auskM36%E5{?`VVcUKLbgl?{>TGbL(RGu$bf#RgP=rZpIDg3vS+0^M&@gT-E@ z8F@wBH0ZdK`sYv6@oM0G)klE2{ZZ4sV^&`SJWb-6xo{A1K!K95`HenX7Ht_oJbL(r z1V~7V8Tw9MrCOXay~Jg*^)|wuQN4;)irjw}V_#}WSqepc>eAbnW@I@1*`Zr6!NVTq zD5hIQNpHSQlF6^~0kXSNVMYjAgTT<-!hl%6!IXvE*TUUNs(g$Yq!nrvr*)3T4lBF} z7*@v(zHqVpmk5-yA&&j|1^J!YHFGDi78miR4T-`;j4eprU8+kne@u$TKs`(2rRn+Q zmXG6xB+fFFL%Y6yx_m2B;bV9$qTZ%UW#~ws30dpev-Rl@1u5zbdcpoFEYa_p#G@iL z<9b)_A5>*^a#Ka`Vz=4MGZ|A@bYuFqEQOJ^4x&Q$?%CX;l9&@5na60HMulM~rFWpD zd0kh`{ZgCV>D{B~PpQ1KaFDiW_B#&M7T{D`P^$GVE1Hk`j#NhT9d|p+*OMg6Twrmb zweiz8J92@a`u&1nXPR%tb{x_33qi>dzG`KltRgP@4wfc3=QI|>w)orcda-X~X0F<+ zkfu(g(6YDdz+&pm>hRRD{$?Lk5HMRe8n99 z5%3w1EZ(AioabiZUUI_}yE3F;zP`bNg=9?5oEQ6e9#v9g^9qN{($D?@RK*aZT{1KR zrHKK$;%wEqlbk(EP~a_Z+c#Lx)P+=Zn=IVVTU3hO`8@PV_GGQWY^ zIx82&LDpX_^HbovMR(o$l$6s%ap7h2>_6n8pEeD~O|Z`|N!!+{M<+c|SXtY*aUC)d zCwvWTkV~k=f8uUF!BTx)>9>qmi*Z@;ZVhMP<+y^{03zfTN`T)G8G0Io-9QQ`?;4QygMMaPP!p)ZW|4CL;$= zX|lY(CX{J%rx~lx{WBtI&d{29a)+L*PHefQsx-)}Cx`PhGMXi&4x^RWF0 z0N!Xxr@s8UZ(x0XantS{Wj!n>*`BrY2&l+_pAP6|P?_Y6h3oyv0B!0v_HifK{eD<* ztIb*&2xIr^8_>;0#Nbs7r2;RdHCd(P+oPc>A-|59g#II2XJ;5%;oP#}%DlX@9bFaSu;XKCtNNOjZPIBV8u&N_& zavP?Oo4i_HtvG&pc(HO^{r6y~O|7e4KmKf}x08j%R<5-!W)?x0Dp`)bZcvON#;pc)S*Kzb3@JNoMEHa65zpz%Ds7>rAL72IJ1*_RMInpd81`xxCm^&Q%3u3 zBmDf{9`sP9rM=++zN0jDmrKrR>U{olFlLniv-)Cn%(ML9$q2tX)Z}cPiSgNht9?#? zF-j+Gc%{|%=HVK&Ab#j7q;2b$EiFO8BqrsK{jb#-?iy8G_<2sl>&q9u`3qO6lNT|B zhH=^v8wp}jvmTkK`$dypK?ep?JqOZn!Et)Jg6@!FRu3k4B`d{>N~pj!&9&#+z_s=I zIT>>Tl9)VH#bvGRr^-a;L6s{EG%jj0q$>6UEK8)T|A)8yzKZ-e-;tOih1Wd_WIYz` zEY0F&od`R*Ly7>Ysd5R`<(`5)SCWT!kAS2(t$)4Ec_zEGtb(x~G$XCVuAE{>#T{#w za{axzT&Je>OcCuz0^7UASbx_FVEV4hlJNURMl&f)iE}{_gabVigI;iSbJ&&nASiF% zL_GbIropl+rV{qU){Xy2)xoHB+N%b4`~#6Rujeb+jh;VbHhsSKGFFkLUSzxJVDu{U!X*(_ejmbGQmIcR8C$W_ z0y%VCkYOCTT6n)ek@A9|gaxUZYI-`axVK1(y3>>DD3WO_-~K!1_J!#a`rT1m5;kX0 zt}%aJBDqlM~y7fuQ_e%tvk`rP@ZOU1$pL|3LDEtndSu~OVzuYSdS z$H{2r|-|KYa#yHK%Jht~? zt`&J8(mQ*}_(uTCNSpET46)&w{@dA04lJY5&8APb20(N$YZ$17_YvUQkpo{!-O5%> z4fay_OqO3)FT5<+OlU@cdApbIkH;vS{lEpOO=@Wb^i2@=fEY(RmSfr;+77s-8zcxJ zz7>@PjDD7lZ+&tV-s#1Y(pjAA&!ZOjZbV#G*bh(IT$2ce&QG%#@sE`T`m8vq(R{^q zOk0H9^X%UaE#7{r0bhOH&w7iKRhy!|KeVqv7j$|2T(>e)1w1-K;4PK-0P1&(P^ zb9cmP<(O9dN*~vADn+Q{5DxFvp+B7af)6#!tP;0*Qet@qlP^T2I}bf~h}B?(aTr## zkWaf`;9DVrP4BgJPQ|C%D{|-du9A8kX{y|*E1_RS>h!s}8DL-XgHE%txzVlNm$4wY zWdh)0zQx`vDCbk`U*GsSKJ`hq^Q=T>t??VgW~`|l0w2x_f@@_RoQD~u8FEa#%krY# zG?SNSB-bRD`pXUAd);vicH(Y#mmOe}*|>rKQXgc;YFRMEa(47|#;Jqzl8>HuDyEKn zsih^Cq}2ST70CO};UXFEIaJszlUHQg>Yjy$fCbKk_3unbua`Bt(&7|kBzGwZcqaBT zhIAvwf;^;Y5+d-{q<%}A?_ao3({HHj^?bBV=y!WcZu2DZD0D=tCXZT?N!54v{#l%1 zZOio0CF|FISC^Su)YSzeg@m7{2{Q@o8%wm)ZEV^aFN-m=-y)1+ejk4XIB%Fffgj)- z(X-=%wFD=?&rae?QU7b1K*ud-9fwpQp1GiGR$rJL%h@R*34UM*Uy4HD=lx6YomKEw zJ)uGce+jlD8(EsXU3k@Ev6~K^E2i{@K%U+D369BdwSh;NV&?l7dh2a5pC16W6(yja z{yqb3|7{NR6^+lJdg(|yzm@`rQcS=v2Rya4kx<=R)UOo`9rJI&PfC~tGOcDotBBG` zumN#4KazoBJS08+ee~W6dWD=UGJ^g(bDDjpu!mpQmA`EnM7>X=nqE2rWBqX4&0SaN zfbJRmrLJ12{On`T_|F3>Ti##&06a|~ajV@L|GSEHUt+zF(jDRWkAOIJ)FWpTkx0=j z+$fcnWBW>mz%eZ9c4sP%=d>OyyiYuC#LMjawW_C{p4Z(i02`Yf8j>wce-Y|+!mfoR+?umu(^q8zKjli!-(m*?TwegZr_s6;!Qo=! z5^)uE-xlYLUK8w{Ex)@NXF%*PLz6$b)(4KltHeb=(tgpN59%~N0{X+@;#0rNzaj|;Xc^2(6^gmSEtvICF9IW=f^<; zI)P;4;1EhxLmqM(s$-J%% z6mnve#S))uYf~$-__Hg89?y#|upL>F$6C(|(PT%{?nt44m%vBm?mDUXY7X?ritj%D z=$^#yBqk0mDcY}YZFaq^*tFgvKibwutlr1a{ODWu(}$&MVC@kQL~+#m%Woz}+4)`e zV((Sf4hES| z!@}&!FBS%L@ZGIkx1-$c1Eu8E9D4y?oWwiLJ3T6@BYnEHqf^8^vvDD&bM@KNSnddI zdOa2Pj5KdwZ#Ww5ke}#q4EQjVuGn@^3dm1h^i5P{j}wq*^P(9=AS{-Plpy1qfd6mZ z9U^dyDmA5i%VZJu*upCd_72YpniNKDZa15u9d@hjm)@VK@=|f_EO)p;hmZ&-K)A0+j zY3D=NwQ+M$TAFIXtsyzh^`Rc=fr;6ck^!YAv=L(5>mdPHZK35P==c54?DAvu6A&cN ze#Jm(yB2XNks|8S`^TZN2d^4u>Y(=YA@&T87UY)EUoD)_=`M+9P4iC|tlw>Gr)&vu z@GyqA$Oxrn$BSCAoEK0Q8ow!(Zu#IP?hK=Rm@Zj2bhz0mIWf)aB;0}BzQ6yiUatBu z5(}R7R<^Nm5+C4&1G+JwVW!#W@2RgQANy0cr`mFg;(&tkDncGpOdxiC;4>?_*p;W* zxPtH&ZZJ4WZ%S*ZF8-vtHvavKJQ%5^7=QHhIZ5{yWSbeTk^JhVk)BO+0g_9^woU0R;%Oh%BNcwNj?tw!dRzE)mlWN${2)R@8|!FqFP%d>82SLkmEjJZ z*FSy8zIIN(2aSN9A0s_{ean3(YD?iB=#kzNR6XTEYY23{AE;n!598v+{Hd^ikOjFu zzFZ;degtR`#rslBF-Q30ZyGOt4gXe-tG8?B%M>p-}CZbSecz$Jd8?F{hh-yiqcpB5vzt$_wW6Yu?Zt1BYW@lJZrAG z=3Lt|M}P2XW0jQ2SUg&DulDY?FYVRNHsrIdX{(y}4DP6*9cILPx*xNU;db~`YWgnm z?=yVj9&p}>dP*+^BU{1D+wCYi=+o8qaIkX?l-Hpx&m=jK0L6ARq7(s$keW#1887gT z%L4!eLGHI}RbzbAp%s`3;(rQYBE5f}=>L+V#5s&d_bCzA3l!1(rvz??lj=$ILY-7S zQulMJG5mwh%Udo`(&>kMq2&V8LYImmma83tI~s5BVr*x~F!zxgA5S!3iHO;beVWT> zNzPB|qYk3aMePy5&@v9eM>MJ3kZENw0ur~m6GD!Y4&vpjfu}JTc)7Kf7$|k4;tliw#dvjl_e^Rh#eUdY9|gt)o6}!-Un5H%~(GC+|(YpA$|1G z6YhjJ(qxlDqOr<`N44Zu_&?h6c!Iw-dk`8t^$TmuK(OPyWJj+S#SouuEt))fbIhs# z(PZ3r-swm-f^MHl$3H&1VN9GoYBuH)=4wo;kSk4;jJn_m(^I*%=0o?Ej^sdh(7BT6 zT6jA&we#{MBgUi{BvDmr>*In}WaPM|zW)9`JBD;n~~Ru9*j9o+LKIo2jHKl z(JkXq-@mqeh|~_G8x!pnZ-kx<_2G4#TIj=Jw5UmFN#Sx~{*k!*yJe62_rB;X3AA2f z-59U#=m6*cTC?Z8qwBke-aWS%r3~c2t0Qa7rXBYE=v{uBv@$cRGp~cNeG>qOe1ggc zPwR;3Zar80i9ImkCcZSO7PANvnSS z;WknW@|0R=h&DS$C-gITQ*g+W$2lwKGjS`i6VZc_A%M5}YxE6H7_g{TvjLuN>dEXc zEV)4g7L<)1V>pkU4?lEM`O$R&LYYU+iJpZyE|t)8C(v-A5EvkG_W;wTwRkDyg@S-o zkqc=)st)dg6_Yt*%JcaH{R2~P+hqLPsGdj*#-D#(#IsSo07Of^6nldD^Um61dk!!K z4Sn(|s_=QTjkgDBCxe1ei0eqpE(f6F$V~(;*6nQ?`_dJ(;#J4>)~HUdA>Ts{+kokP zy#4@Z!(iB?B2Ys@GBMG5nN?A#DqkS$UIfmEej5n-8nlG9S`0LS6MXG4DQ$qMyvn3F zw*g;lo>2h%gctK;sY9YI{S4KAZTDKBaH&;)Z-<%w8@xMZ zHhKkLms9bEcSp8QErM}%$-^Hn_X#b0;J9Z54r9++l)8R(>!E2XPn9QH3g9;txAt9r zxH=H&>B|%wZXJ9VI)donEqgf;k^eZ4Xs@40$~N4l$Pk$Lx@XGwB}KEN>ZS9#1BN?YVkaA^eeDCZe%o_@VJX<{Bp@?l z#fjTWBwI=ryIk$dzAcC*7`D}AK z2}bp9a%9$`7ds`yfpD^WvI<_j-}VcuE9IXXXg`YRIzkVM_WV6DhGwK}kr)UySP9_z zOIy6e&LYH7oordAL26bP*)eRr3w~fKfA_<4deC8+5;U|P5}kf}1;yQt)$^~*v!+~Q z$aBTn(PMZx0)>Hatrw?zzR&BWdgler-Y`rSb3V|$jEOigG*+q|*=?kSu3pGs@O~Lrh7_;8SvtLb0oIK7dcE;YpyoON zINsgInU@LN=HhpDXs9?yDa86G)8scuw(_FxX+Z%z;0%O23;M zyzmALG5FKh*Lb>txYZ&Y;p&U`0_n7E`I_xf@^;uQcW(LKA1C`>+S%HX)<@?#j1j-( zVV`UUU)R2yz_`!myR=Tj=W{cTu@|lMHXjOE~t z`&KwJMNlS`O28rkQw2&WsASuY#;}Xv1TsJRI*FvN?U{Ylfo}67lwdGyW!_X7b#tPQ zA-;et_?|KsOXqizL{Dmw(;FWpnse$>cTeL%HSUR#SI&`37L8JhsQ!%*7Y7R2LZuSJ zt!k|Z#{y8v3&g9h$q!%X__R{sdf(lAc<#WC{5l>JdXjnmDi3xvvpZk$V$JS_X$kjsWFU$k@e?d4Zi$B>mW!m2JXiS+z|N=XsK_^p0`Cb4@dtY z=IZ9a2QSN!#$1IlS7zjBs~^p%gLy@~UTqwEdG<38g!a2x&^zY|poT%x8POoa=eI=5 z3s-$hZq`@V*IUQEH8A-$Petb>1~@>?UG>*d&qk%bn^Q2F9sRwoTS^3 z3bNJ4;^A&tc7YYnSX>F8;(c8`5EJ%Hp06qJ1y(*~U@F{?5HmFCI{!}Vs33ZrsnJa5 zx$}|ZS*pnDM&w`4L@4<(80=NS@mYuQz!XQq#vxRa z(BYPcg@nBHq-$uUE55dBzt=wN&F=m|xdBnP#~&q1!VifU0bM9Eof5^j)7_qFuKIpP z70mSr!_21QmoRe8Z;NBVi_Ruc(h|dl{}hhaNR5R%I@T zZ<_rNh=Fq7-#zYwUM!c)&|NwtypzgHVEov!s#5g}tGtHsQe+!M=hKWo2IRv`Gks{` zm-g2H(*PTad8}WV1!UL1wI$d;&kr!NQ{df4zfYf6n-^iHtv#w-b86tFe=xu9TfgxG z&KuR-v{U#W2tyd|hE#cfeQ-yzC2DF$!&5@4X$Q^m8(RpX0c;WFY}8a&W2o5O4z>SY z%1^trCUVt#dz+(d_~c&0#X%#GB$Jtw>>6eU*C{~S$CBe4i_@o4#|rttY@VE7STsA= zw75_84R8B!&tP%aMTKVyN4N0gD@`q=x=&tS6Lz!Y+REcdc5NR zUmVxg!Md};-K8(t$^-so?50!s14}H~ak|Fbw@GQRx(uogDwtjPLAhTxqHF(%YGGHu zP%>fB27LIY9N5^&e=BG$%O46l+&g#fnmoIM_yW^Sk6KSzVc&3>BjhWLcATh4>Kv3y zX!_z%k-OX^qx~I4Fc}OhUTRDfb98m^#DBrVduK+5ZHd^I!$EH+A3UWs_zO#(%ZbJ~ zOq|}N+=iyBI~>v2*lLQ;fN*y6Pf;|R6itFle?#Qoe?tSkC#v)AGU0Z&yF ztx8(LRYnw0mxQ;txrhNFQz|`Jm`uZnE(O$77@-Rvc4w*U%f6cvzT(AkAar7lc<%2I z!TS9w%OnH$+ugc$h=&Ir%rCG zu`NYCyvwACedG9EH1D3;IR{2kT9uK|iX=$yi1F!V3P4LD1S~e*DPYmPMr^DM9&PSTn9e+ENIfcs*p}LP}~b zP`)}|pH6hA{foQ+tgVa}6fon^zAMapa-Mz!OKZptZ9W})0h0LPg{fp*ekCx}70w4WU zII{i&TnsnztVA()i(=}Q&yc%`m&K90EpfivB#y(B3{CxBhGf`1TVX2~`$k-zK77?Y zq{Tu!BJqxlJd0FtuTgRSWhFG_@@M4u&+2~fQ|p{XTHY9LeY($tk|r|^G;BCg-;oLE zns$?A`IM%gv?2?+M(341SpHrd_Y-k$gnP+@@p?y+XpBa3uoOAyD zs*~ZqoutcHg5<1;&(Dwbk$7@OA@tI}uuj~^-r(J{U6yb+8ab)GHZAJx*7zhtpAgR* zFf9L~p}zTp!f9Dim2udQY(@fF2+?jS(e$L|JZt)zDb6ZfcqQ@}wsA9@|u#3VHbad943#k-CLh5TH*!jAWcD|^f37EIfm^66{~eVf!B4t;M+}H9x1a*g=M!4w>>;=rj=RL#Mo+Bo5(UofvP zzqaKiJI6!#DCB2diV*I7>4&C061Pov0=tyTtt0Zz)jOIQkMJuLJaI|usE*t?6gz)m zH4Yz2^jmBD(%2~Ou4F6nH|AO7F+7aY;0d=^8TM};xj{5}n`lRyQ(OUT>U%S~`I~?{ zA#>oHYPC=U^jPd2u{)c$SxCEaKU$QCj;9!JCRKm`@V&i*xg_m_eC;#lEq%pEy`cpU z55XKUMGh(Hrg$~v{#Rd~%f8fC1>n*LtQQkOjtn0Fo)f?$NB_nnsv6`K2xaz@7NYqn zS*C8WAP*M%uqRJFew^Z$*d~t(3XCzixI8s#bu-kfyI2Ct?9DNJa^iRrQM?!!H}U9_ zt!!Rf)%!!D))Urodw+d>usZtmSautJ^Ye2R6K$kdF8dVd>X96{@d=uIWV7v9bpEcL9(+BuXl8rHh4NI* zGx3Of?LD#Y?FWZ|+L=KbSF>dGn#m{1xRXMmlL;Y~VAN{{N+v3O=uk?t!x64H(4EP4 z_yb7m?Dri-Gq1mJv|QcP2X9XAd_ob|qk_iw0n_7_Tq=r%kMhl&H>m_=cN#;ORp2vq7ld>;(B6>nA1Ta>W(^ zU+Z^at5vzHjk(WrUoXGrKmg5cD)DnA!TRGvCiS@hcOg+b)rBvv772ED?f7f!>jHWw zY)ijtu|1Swi4$oB1Ra^}wRS28Iw-%dt~MD)33)~@P(*z9&-mc2QJwd3J@Nd5CHMoo zY(17PsznVWz{QPYInk=|N1}rAT00i+%bptOtFdOGIXmW=hZSs_(NHqHTg|%#Rxn?IlNV zFPj@1MA~=D^&Ag0ISX2Nzlr;zF0FzD{ezU*PRuRVw{~0b?O=HIYDKzk&3Tx_tCFEr zYw=T6LlnOpmsd6q^s#s87xmAP^Q01|MVf=Yb)^Ncy1vM`HPy$Hu+3u~lb8PG&~lx8 zScwq+Cnp7|_ADp;8hQn+!?S9?2NN_i`g=AU+BL}_GA*rA7}X8D{>O+p(t%sPPsY0U zBt^&b~$4@uSM zXsTowk!(a0*jq6?plQjpbGEk=j?JU@V!-avO`^;A{s?~c98?I*1P^k82QcmIPlDp0zwOxnd1EU#8pSPOX4 z!$!cX1VA}tgaX_FOyydVp9k~5A@{EAnktn)C#~!~<+;OjiTXITIej|N^^c2feC>Ch ziD?JeoxH>or8i@CRSoJAeV^~#4b64#>CP#M){S|~VcRZhIFO#(nACIs^3tuLHj?h* z{nXvB2QYH7j5TFLLCAoCS<;iRCW8SeLab$vS|FW}%Dny)QPtd>S;zZVJqGxcPPfIR zsc`ShJ(m#l1XzH<)a6n!_`2DDTh|q z%Hj_0QkiVgBg74-B}i%={~-G9sWEZoe82NKnX=Z#MxwhNB8
7RaKeb=zOYKIOR zaz`@CR16i2m)UKlB$Pu7Wx(c-&mU8(N#hcKkjDEA9K48S^~*u+@;#G2(q?Hg>oP8- z(|Ru-R-Eu|Y{Tn&5pk6j8@=M+tpIBK<-cUD|D0G$z9vM;Aba39Gs-4PrLAgl@1eK-dpFNa>m&K3nn^du7%wcCtn-X^ zbR!pJ5gnpDSypZzGatrsk&2!>xqgy03H}a?A(^wuU>ThSRTE!m{nVuSnXtKmb=^(% zLjZe1)yO%^Ro*7$%zK+(&;TD`sy*kD6R2# z#MyDEITj~g)g!{d48WW>kyB9xqBsY4Mlo&9G(lFn`|?0X*gG}l65Cud@Mh|fL}y>f zwDhXdY7dlc4jQ@FZLpF83|A|*_G$+3QvPX@H^F=<@=;I+{-rmgO~Ze+*v+>W=tf*? zqLr=$J#(p|J?3)~@N#-SJ!NQ}$OlsEFRb=Ec>rb~sEE0`X8=aI7UrFMD-K=5M|UUK z&AY$y^DA&BZ`DmDI&@B?lj+zi?+@t1Gk3&+KE&^0D(7+GsJJzzvZLjcobi&$Mk|(h zd0yixP1}NTS?pU24*l;gQFj^uiu+pK2)t9aGD8#=}QeQm2G9f3-XXqn3$)mIC7yo<6VwVUUy~(1YL7 zHx3Rh*pg;Bnyz zOp(Q8yrWbwF)Lm#U%Lq4VMakV#e@dZ_$%KZSd}|JpFD8A!y^DunW0s7>X)_}4hz?m za(j7Dl`$_Xoo#zLczT_J7rMvA)O;v8jC=YT&if%da>x8C-;yrrlp=pF!6hY^ft zcHm>m^6zz0f4;A^@p*q_bNwDT>A$&-Gw^Wepf1%~h;!^T_d2*iQo ztK9jjw}eWhhA#9uZp-OeXqEEp20+^+0{EsD) zJoBV(>$VA$0XTRjXkXgjTsF%^WmMdNV&rmx0fv1^mzK1%v26j+c8}FVx$xeH%-ef{%)G9d3(zIkh0mu3u+$@;J&a+ z9Y?R{3^&T!mR2RMu~dzjFt&WJrHYdRpC%0lOBXwUSIv=7^e12mj)Jkw`Rt_)r^%P>Z0_W9O!bvo*+%hr9hwoi1gpW3920*_r@+YKwUJs zjH#LwOu)InE-TY2)Kj{hMxMPT=C-%+Tzlj7@$kNeH~!t3oba38L7r!UEvt&dQaf-) zG?sSO*d8DJ+~qhb)={dEi)eNR7nu^Wvn5UyT$Z)keglQx13tEk}gUvV#!m` zyTMS%Q10=OMX-<~jm(^7@mo=hz#eN?ecH@gbbP6mwZ+-D#vub4w@SI~EK)c&y}aJm zx<0+n_NgcsABgc;07FjR0|8W%os$tYK#kM@ZNhSj%qy+>ndM?sV^4($%<*d2;OM0=*}?|dtusk4`JxuNxD;C zMJ(W!51F{VjO+YKO%4QYaExufZY~*;lp)s&1c$ zBLYpJws@PGiNv#%#B5jI=0uN<3cFQ1k-bRA1B3Of5ZNbj(es>3x=EZn2K!3PEvhMMKY&-cPWly(sWC8;Y|&;eWY%dK-*sUh^0{8>t= z@Rwdq;hqS(?qyB*xGr&VXv!w3#Jr?bm448x@a&BUJt5WrM`^{mNxqI%N_iATX<2n@ zk=!t?HMeHm9BJIUn||tpKI3~1t$JEOAFW!g&)q>NxrFlovN|pP=s7h(q_ckp8&1-B zpow{|M?+gmd6ENpbs={_HDr}dU4T>aZL2HJ@4Ms^#c)0)734N&yDzCruOe}rMERDu@Kn6qZ5YV!TLxjEt(o{I4 zRDu>X%eRr2xi^UAk2I#>E7Oy=A|1x`P%@x0o4x)m@7XC*@gd;<)!vL>; zAOsjTsyET^*&X+{Zs7NkeD%X-955_eD59X_$;(3>-KXxj{xH=BUyezR!A4qu!4(bx zxN2H}tIkQRXXB`jF5_#k{_GLc;k+HMw=HUTFeMlv9%?USuPrl9?_G?3QtiLoq(|=s ztTfR`O80f5$`1M}bEY8v_{c*A!SP{3@scw!5%PVRW$*=!xC~|wiU~NG&}RMfEdeMh z!_GJgfPrr;{{FdtUR7rXCL{*RU1taNaN8fDmt#|ItT5uBvWQIk=yrHpGGX+56m1y- z-hCo<{BrDaCksm2fDgsn_PjPCMmaWS0(jw5+{~_D+N&&11joR?Mc%7uf@ykpb}K`eJXj>Z^HaE@i3P;GTpmARUSV z#G)+Z{e-c6g8vgnZ_xI z@tT(hA3i~g27WG`k&i3@TWR02@zotbEtTgE(9AV0hkWPSn07Vr7Ja{HtRi7YDOqs? zeHSo)nKk9_OZ!(Otfja`5rs-pTF)5l59eL_Jb2Zl8c>Wc`3c9U(8zb`003H`OB4_d z&&h{FwBK*HmsUqx9CY}JOq9t*lGKnoH^w;tK0fU$IQrUjdv; z8q|N(1djDdUB)-JXz~YAeai&g-3Gtwbfq1@n)r&OuL;~V5%qWx5OH1IHp*zt4HRcjI98e4X$E{wa-ylZY+y}K|?7{ z>;(3-CX_s0H$cl~$Q|B-)D7ek5DuXW50TO+!Mqv4G%b|EgWy-xGdm_?F$ND-uv}Mw z??zi1?xfg7SD5C_8SFMU6WZlWb^Em?uHxepsd4n{cQF#X^4!d%(QS^gG^MYrHIzVD zGiVq$Dy^D6W5J%K8#kud(JRPB#Dt`jM_Avp5uT_N=T+(Mi)^%w>vVb%0BQzCS6D%X zQac%*BJC>*h{z1AXA)zJ=D3P$VV^orM{4GKSZ{(h(1yhxH3k#c%Ei_U^kXjR+_Oq1 zcOs3O<``++!R|A4P@703f?1NTY%v3 zRbzg5Oc(>GTDtR$2h#QHgMzAy3+HqeeQ$3a-sR`uGT^o3QXVcy=3=SEQ1-6Q%z9ixZ(EFu;dI+HblHDC=9`*L%W2c_ zvX2CWbjmlGXsr5)w;_@GBzJbIZZ(DId+ta2_qPuRww(&p@`~O(7n2rwV73xiRno;p zJuHe;l5=68gC%x%&?u*My#J=#K^h~cfrZJb+RUj13H*lbf4_jd!+7W+rQly1H#5y; z&Nv_UaIN;5`V~~VTuFV%l|V|ZcC6~zN^S63V;@9f?jfJ)tyjrNswHz3Z1GjP4LyEU zuhgNUY%Yb(vX(tt7y?AN^Qxl+ovzOqUV(=c$lAGt?z^C$7)s~|H;oB>3>-Jr;_RN< zV?dV{j|XhKi4G~Aai72TtUn7a*G}&IroS;BC((c5`O(+e&|G!+IMtW*bvS(1o5?&y zfzU{vm2=m-f+hZ>bi9kzZF)n+VB~i(7Wd-6VJe|9e8s#u%4JU0q2#5Rq(4x~ogYVB zjF!5;`BptHa1}w@)0udb9WA@{*_X)lV5auS0lEnWlwSO{6s4W+iTUF5O^s7i(Ek0e zg9d~N!*V&%k=q^Pk=mS5qJ^af-R8Lq^RFH0o{CkR-c{DdOxBC@gursVi7K%#6T3jR z+1p-wWBj&4?cK^jou%&#jdbT#wELeZ2I1mVk;dIpc_yY$@KOFX9VxhE)^Go+!z;X36juIY#liJ$>k4{J zeb3OPL_YE7ezx8d@v^b9S^o`>nvSQeJoICeuO*oEJxy6LmkD?)$~4i1Qa`Mhn^e~% z#7|zm@lLeZI>x=yS`4YWv&)*QHntw^tG4`tH^8*QnPg&sT=tlBo=3FQ_y_5%m|@# z?sl5puR8Nh&7?YFHf?=})tf=c|Jga+s$Wj;AEDOuSE#N0JJhmaX!SEuzgWSK_5|y> zRh*`OhuX>5cAKjhz11V{B=INWizMga%bronP1WF84bG0kSX)R5jZmyZj&0btK}HuL5Ucpz zA4siKur>0vj={7+hxe&i!j*U|-;X$B9Bb+yJ-{Pa*s_R3MqOslG1vrxyXAlg?~3Ua zruUr*_aWqP7w)_lzJ7Asb&m-gJbCIxfL_@;CFEX;CT+q?%>J^GYx)e2nwfkL&>_d| zTtX&08}=AwN2-*yW_s^!?jP(Q?7hB6Qz&-JtD>x5y(2{`MBJ|2;5@TkSQMUG3_2}U z+Of6B?0qL`HHQ;BsCYWVh686t+MK7hGdxnJZFltWl`M^o<*l!+ z+!2Bz7re|7#YA++Hn)?=W=P?lZtQ)bIm*RR2Q>)~@@=3s8zGxxt7rVl2L~Z$_+)*3 zMrg|-W7mwvKehAxW zSQg6@y`Ju~NVyM999WtW4w)f&7S>qFnyfNE1>bes(ek6#G1Bqj`c~Okl`*qr`bB_W zP8Dugksvk}JDI2Or3uxdWerEx2S6=bbbs=qn$oo%a+)~i*#}?0D?C>>1zU4hZ`l@4d$=il-g2vtm9)X9@4Uql#7fogrTq7E zLV%j(T5*9kM$sHb$!I^u^j*Za0cT@<(>c-wQX0{xuiRo?7fj!MZ~d78n*TYe>T7FL z!-hQ7S~Ihgf;Q)yV}9n1!#}^mH|L%Xj=jOH(%^8lpCquaA)y{K&m|Pb-WQ!PhlX+P zH>&z(<$SBH3KseLgH}iK1D;4vw*YqKKQ=4C6gsr+ZsI@Nwvmq|8t#ysigRFCm}?on zCYVXFP4hU_)zdwXsl>){`uWY?npE2FMW2`G3*!$nJhBL@!#hUApU)JAp6s#B-i9Y4 zvp!4u^lvp3JnL3Dsr%EaL>`2QA+8B|J8GtHA~h>``+W~j&D+nh4cP+{97eRI-odT` z2zFgUI_BUfzNnFpk8`pTPbcogQJ0%UYhM)|V)Y->Z!1`Y-%ioZd$nRzwu{9W20YVD z$A}F@W5sa4Y0_>b8f7;mn)4&b=NqGk6@e zx+vn#Bh-dfY4<3Ew!}nCU}u(C&S_DNL$}t*{g)j`I}nV7TlHVsX%!CM)X#pE7U!k2 z#63QS)a^g44()RkaFi3RiX`*1_{fqIYiNj*drONNmo9+a3i$GW)ic1~*(EzdqMUAn zzU)!<_V;HvAHMI|xchADBs*V&gc{5;c@U-9%o{j#R$=)*b}e8|5&vnFH83dMCPJ^mk4rT@9b{onr9Uif5!>!E0!%<-ch83kp~@Ki4e(rN9(64tAS3jh6;4fvGc<^x31+SHgWo@BqZtG z70+!=isA(;m3*T82k0rEy*}Y+rP_Z6=~tTWcbCTU%D&a%ZCWY|dwx=dt02e7n?%r$^#acGlT3La_hp|4PzBHWZzHn2{t;U?UPWz61MOeD1C=;ZSuo|DxZrc69B?&Y+TrtI6X5LttR+a|Z3aS?b5b z7)E|dWgeYG-9YD5tB-NwSENit?O9`hpzjVM=8Db)c)rsLR6np|MzbcA+jZ*wxT{!l zB6Z>1v>8<^{DUZAAWSQ%orY}4Q07szX?>&qRtV6L5LgOeJaFud#-nq|!Z*GHJc?BY zHHdZxw35SuvzoNZOz>{i=9cO}^YJTYgXUAQ-l)>HXH*QyrDOk$S?rBWs)qp^9Vss< z)v4@OC+y-B$g307Ua?Bl9%wWwETb9QY0iRp8b!sw>M)Ojy=ODZV%p=2KoR z-X<2KZwbQk%zzYH*H4BtDwkO=PzPgJy+iD5bs98db@_i9EBqVPG4=4rQ zvDQ1k(hTtM0KRH>b)V&%-c=zCY=6>3xy-oij3u8ZKH) zjq8f{VCYu*AAx3{xeTG;iO66{kr0|U4}jB#CiAc}A1rjtok7*g^r3e)lj^U%R!_~g zRSOdz>!$NHB73QBZF>biXOOjArYG^k?kgAkM_wN9AV5(cG?_rfE$Mk<@9ka&pOkCdmqwb z8^jte$L%cJ^cd>7l;^9$1amw;fE>iSIBn5G+tSQa>mQ#sQ%@~ggJ)BGd#CkhWnCaQ zgCGYpl0WL4dQ`Ag(}-^$817kEdENTD+=pYCKYRO~C7%eVVS0>~#XBWNCtH<2&CCG6 zjj(pPqzF&sV{x~G=!F`>HH}$uqt695$@@KK+hY*u$1`1mc;BqJSL#URQj^~g;=~3H z8u@*}9$ZNGZWA2{b96+ALEd$!HtKT}*{?*Bx#QWwHSqCc8V1QjifwA@tHVeaRV-}5 zH2IA$KMKL2hfo}t>xXkO{V_p^$~$J`XH z@t#Wk+3CPRB_N~Dxm_RkLQz`sP>bZuV{!AX1iBk_S*w000H~ z54c+ZWC1KpOb{jp76=5w#>T?IBgMzV#l@o{xkpG!N6kP_M@>t6pP7sG{sRssT3R-t zM;uRh`1$!6SVbg+dBwT-_<4V81O*!#8xI$c5+9$Emywo{_y6_lt_>i@LfJvrMMHT2 zpc12?5u@C70SrhvF;M>U0{-?wK}AEyzyx7ovHz+X&i|J`qJDKJfgqhFGB^VoroPv^y`XLJ|+oQ+)0)j%q zBBHX-ZDZ@|=I-I?sl@{A&-Pzp3^Y&HihO1^;hp_IJhpq1QZshlYY2JTziJ z3b-sY>GClS7ibbX>q}ZJsaB$EN=m`5bMAQ5Ml(}Urf~br7FXD0_x+ySQ=)eJY7$jEMYP=EIzqxB9*ip{L1O$oXQxK+(us#*fBiH?6I7trFZqK$IFJDu(zj&Y0n#f($S6PVdR>*Nab!cPs%Q%$#v>&$yL>bHzocFUYvk7ve*s@XoQrN{z8dEwu z)ABb#_hf81iWGMT-o;O~^XHxm8KNxtv(7bf1^4t}Rv%H)xz z-T9M(ihl0-*-M-$jilx$N-WVHU-KrjxP!C=2x*Kzzd3sjxeB+M&TQkJ-!k|R%JhZ` zE5GH;JOgI|G817hu94IERqspHm(b+(53e`eg)Lw2V`bw`>r6K!f6z(MAzn04Zp=#& z#4DAE)h##agg#7-5mz;L$l;8vYa$K`{H9n+8?+#)%ewe}xgiQ*%l-FDg}=W|fu4%D zzcXEPwX9oG--4FHbfrR^TOCbq)r_U=l|}rsM&wTuruG#ajOCxiit|}IG8pa75hmmqG+(G9ov4bSuaWUf!QClqY^xxBSR9 z%=>~pFuu30*0s&C%sge>-@~4>P4M7o9Y9hS=C)vCT0`NP>glX+u#h^}ym-9|!BkT=!MDBg$?y4*YQSSv)LdK25atZbxYRMphyu`1}VeFspfX5AZM@)1*V$aO3cX36)D5A1gX$o2IakCJ=yPL|v3u&iZyLN=0q!fe9(k~Jm&SYgNp6xNM2uX@tL-doz3-=-H} zwamGTAC}`5n-M=vaFS3gvX^Au)vt>_5*N#R-tggDUOzuX&^+{&RfRk=v>Wi?|7~cr z?BzXcCb*87dSE%8hUa~naxPC`>+36u5R0o6JHCS+;rwLgSw;`lw$S!IR&fh;kfYpd zug0!1ua~lRmvwf<%W`4!Q#5vJ?6i+VLgLE_m3PdcVNK#pcLKgKlys^ zllsYV7~g{h6~dC_iDzvEk6!yNg%ok$W9Uz*@Du%(?1o)aQqAsLwkQFw-~YO@KLe3& z{p{Kz4kd_P|9I3SMqZp zS{h(-fDW{rjh7Xc1>IO@xFIwK?>?>evqTg=L$1R^6^L+kn@mOsOFUhNu2=N4PzNsh zZH?PP8iz#0nad83E)! zG~a_Q2uV{S;tF#xfCo+a^cx6bq$kr>W%4hz%`zJSSKa|-E9Y~kGmg2iLFLNvvhyRK z%NfO~vIn|WwauSiFAnfD2%4eg^vF=Ho`rO~1~AOWy}X_!^K!uxWtgfj*wtU~dx3xMzAVRZ+D*Zw#0zqDDdnYc zxZJ85p^cr@>vf4ooOus}>7Kv%#mb8D`Qt4c>TohL`a$c;x;R2cKQ$%!?cTo7Wu_Ogh3AEzFkQ5-hTUhy1n9QgZcxLu?6i_Mt!4yX`2*}*V@ zN=8ydzy*qIcp+S`AD7+E?3cla@U6TuR15d7r6S0&t!k^zavBxY8^36V-mYp4+FKX~ z4P#3bb0O|~nMsHmM63ELHhfmE_Sn$~V{yG*{ElCEAWa+gzIbpWdIiTu=YHz=RE5AR ziGp_@6V;|Vh1Ooi;y1~z5>n`P-vJCv+Kp+8ak;OY2IITTp@lF89msY%OI7K<&rbM} zfP~rZ2nX@LdTUZh=+pK}HBH51+no@&#@dqPf{sBevGo{Ljlri4kb3yG{M!=$<8zes z?EM2MS`R&aBZEL~VnK9&?`wC~KGK`OQo90(8ax12g6Pyl#{Jk@=Mu7OBVn;?;{we% zGe=3E3Kxo*9Kj}^hC!8WUyVtj3gK_L;7_lNAB^Q0eAZOTVO0;`c^ZMjAXJr&zd{b~ zlkHf!q%yfBmT;@-87y8vx>5aSKH|X#Y7<=gr`y+DuQ`~!#`y!{lw4l$0>(a3heZG~ zM*FRxjYx8KO{cwxUuWw*;kSQ5H5>nHyVnlPqdUKy)FKKcj9HTMrVC3+A&3`v{kNIij#uIy)yd1|?cQ~3Cc(Yp;{~;@<+rm#2lE3q=>B-v~u?PouszUGMH<8io zJgcS&gwkT_Ad>k$XzWbq584yO9DB_Re?0bTD+DIK9;7zJAIhMV+I2k=;ht{p&Yxvg zTiN)e*PgmcqiuiDU7&2$FiCKkKonD`{(tu5884sjjI^-K7272rVgeCw%w&jVPOnm^A|DmF(W0rK26$YmqI+% zdrkI>udZkr^e1LY-B0lK_okh_Qq{8zpc04L6HAI4mS0Z|`k(uKcAhyc!$wfN3VQ}6BOyWi_|(_%D5TVEznFk74j zzmO6IwoTIwM+0aI;~g&+Zici~DA%g1z4;u8-rJV&W~6}EIn-Ec1qf008%#nAX6)fc z^v$cRD}eeVUM9Hq@nogcJ@%`b-+?aeP2dlsx1=O-V1zU|MTyQNdwl_7IAf%%u> zp5Umg*$t}5jhcFmmDkbj007c$w1YDFx)qnHp`K9}M_fDI;5iQ;hg9Ku6$1)yYHx)l z4qZI9Y`hBM*vDYfeU)`bs(qB?wr}I-dX`J5<|U=JFHZgKJ~d9c5#=8 z&`}FDfbn$&&}JT}C0A5TpW+T3=Jz`d>q(0RoB&p7Kyw+4>D8Dxw^YPixL#RHvlU9c zXUc9hNixkR@$w*_Krv@+skMhqQ*a%~={|epYTe`!Yufq}vuT`PWq{={0fBRx03G4F zd_|s6Sb2H1;-RGs`t_GLYNCn_vQ0D;e2H;;D}nYD5;Iqjotp@>js<3*1U8Cx4mLB%AdhPpF9XDc|uPg>U+Ly&)eH*?f>~lrM^QU|i%!8b;vWLXb42 z&vbLe!7>dbv(PVXsy3S6@wU<})bg_hEvoddU=r+&nLUF$fS_Q4LARNjFvh{%=t&3m zM8(p)uQ%TyD$KOGDy}BKXjk&d#M;;~Kb<|d!`mDlqxNDktP?)CMY*D>YdoSxd$fsr zou-wpFxpQ(%_w7ZzNT7x-rI`%{h(qO-bM}JgF&k8(dy0_AMVax#N_=ZYF<8;N%QAV z(p5n8Oq#a0)0Ep@;K(KnTzz4)q;x{fzzIITr*-~HRbhj^*&`~K?r{$dWk(D!{5jiyD{^Gxm44Xxca5CglB`*IcBqh%;;g0P_BWE zs83~Gol$oEAkV&Dyvqxv3eIs7`9g|NsLefI~)I`O8*WRsg}Ar)Je3H6tDRPJ1ek$T!}Bhcth4U z%d;~W2AT+FGI?9$IkeGC$wVfE*XAUi%WK!}PDo%)VM1hY+SJ@=Q&Gm+)Yu%yOXV;( z7xLf^aHGZZQ3p*(nbq6@S=Qi_+2Uj^(#J(wbO0r4y$Uy-O-B!k*J*kVb!3dw=X_{C z{ao=KuqHYNZg#!{0#pIFY8`>!66evk2k|$j*1sjXWh2km+*nt8%H6 zek6`kMsYL4aJ4xKQjv2tK_nXl9t zX%30`M*BkQPirqLo5mS9{HM^fw)C?8aciZxN$DaSak4`-ei16-paw%#w{}`V-B}eQ)>Zl9W~p&6ZcM zNuNd@RC*r!#5hYcDEdcDApDUiWvf~IKdhNC8`jYIVzl(ovApodkJ^~0Rzy$EorJtNe@QpgqI4=0ro+nd*4b1|Db_?YAUv_=# ze4s?*LPbdgCRY0!sQPc<>R;b+1VtsY3YM0a^cFZ?t|&gNcTzf|#IxAE_R-q^<2U~b z$^O@_5xD%(>rj}SYF206pe-Z!D+l^)K(ZHt#U2JK|Z#WB--`nZZ>SeRApRrM(+`TUlWM7+K0d(aGh2>eX*K&n*t${m1W_DE=y!1!RjqSF!A zSmJY8b59PlMfi`}APbssL$S&cI4BAX=hq2oYpa}YJ?S!myX3)4){Mwk+8W#!Y^EoS zjC!|+_^Uu~m^1J)1M%>5EN{cV+ySFp$eaYq1pG0LyhGWizB`~(4km^06KK7GB2z-N zRx2ZTjKz;Kuy_-dUS zA))~1Hwn1|b~=%fP%AMQI&jchd7gERF$Z211&1>HuBqtk%N7Nuf08#IX59)xE*)WY zNRN$l`5VbI;MaRn=eXbQfHt=~07vR~?NP`a`Nb*xSz-prSO6h%fxve}z~%0MUiNZC z=Qwzm!txFX+(VH5u{o}g`Cs%3kOSm6_&v-v(zPW|%SamZ7Z~{XK;Eb#1}&8qYUKGp zQ7PK^KT9q3Cu7ku`h$(K{h^nPKsBc#41XSs58yu;aO2HAC1eW6C>4QL2-D_J!mQwk zTwB4_U6G?9wVUgZvHw9!R95o|;*+c?X2s7b|JB3-IShYnbgy9%#3(s1Lp+jNoo z(`;e;lhj6@|49S%|BesRQC*9VpeKR_AK6*U8huR>Osx!JS<#qeXv@(0z5g&)g0era zhE&ACS<7}S6|7tZ#z?SUJ+9Jz;Uv5%0-97W9f(zFr0`a`jd!#sBMj zeo>#tG%7_WZ`T^$ZHuUj-eMJhefRg=(!b@H{tw<=Q~$VfHXs+yWaJ7kOMi6-DCmz# zWBhs9+(BlkH}kBeF7c3y>{v?8*yvv%@4vv_l5A6~W8$1huzQeZ#OCdp{5RBB$y8MB zAcj~w9T6xRLg^$z>Y&5%7JU>G)S*VVpD8iMed4ajv6i&F`Yd>gN`^*e$5lvtCPya< zG1Pkc=z!tkTIw9dxZD4-;_xQE@?+>q%CgeOgdk|oA*yp4P$lQLAQ+TOc!Zl3!9db< zYv9bolIrtD7q%jKU-vh=S*wUWJa%V~jyNp6n(t>yf^G`QP7;9ViqQ6` z*|%51jj^U$9>93LYZyrP)%6#Kp$jrJDHsrJtNxuL$}*GSWzNY`w|FwWeI4GV_Ph>< z2P7h9jh>b=5^-bQfbS96nP-sSqY}7&ss@y+X59dZ9v}=+ms$?Y#a%-+65C2z@yi7c z9OLxk5=(8bgevUu#;Uw&?2hKq?1xd9K+tT$9JUH<7-O-CG3=yU!^n7Xgw}gDY2YWt zRbe)36pbdbtHW{(xN8@0-y^)d%f2Hd&aPbJs!@(7b;7&`<&uzD=qp~>O0S2HM;Nbt z^$q*_cMJq=j}C2|($0AOO1S>+ID%Ip*UQ(qc~V@=Nsr=)L(!C|)=7GkcoHQIxYmyP z)iJQ51sj1f=>(tXHGCWAUtcy<7m~xTk*4EyQeqXLy-d3b4G@5NO4ma%E-sP=2@}SZ z+p6W_2R4L-MHrs@v!U&YvhOg^d9}d`*R^KC-5E$!vSBXmX(IHTbICL1PcN7ud~K*j zyY!FiAgo=Sq9dJ%N9hKTivwoYUaqyR6fqwnOa%~A(R!d{Mg<-$5t`LyObD%6jxLC^2@y0c{_!^9-~*h!vHO&=MO)|*5uvou4(gB{(=B-y4y5u4p0+9@DzX~U!NIO6H(O--xDl2 z8f25&uz==BCO_tdXjjmJGd(;gtISpZoj%w5oPxS9eo9UCQ2y5zUUz^OT2;>$o`R!m zNEC1-6-_TR*@V5;Xr!PLJZ_WUK=f^ii8u6dZ)|CyxX+Bv+R;-ZgHE3}d7`;P?sl}p zRl%g>>MwZJXlH``+c1K6F|%{8Ka=dYMqfR9KY9ssZatFtnPBj znp{K78@jIaZ>1DYJ;w@E=x!qdER8w6-s?k8!4a)Qoc3id0Yk3~Rz5@VAULGIoFpPy z0woYpzb00%BDQN`$wj*_2qR6F~-jo|J?R@nnf2WoNJGNkPH5eWPp4Z0fXYJ<%(@p zugZ@XLlAwtQZIG0Xuz9^2Vh;Pb3ga`Rg+)j$SAoy9)3W$;RsKXa6AAZ;^Xp)29QW3 zc=PkF$;E!_bVJk@0-1g57aMIom;#r%lM%Wm9wS#j)PEy1FgDezu+u~rqdda1XF{0_ z}X@(-3Y=5spYJwFX> zj8qrdXeDc(^&%f$AK)}-UuK%(_w}vXVb*B<1d=-cc?WPz$Z^KSw_mqZ6zeX@l+WE} zA_hEyHea#RuDZF6m}}=j9mC&DO19J26*kX_JyUV|@_tw3;KvXpMnvw;=;M zSY%U%r|SQ<5v`jMNCL^mtLg7uRDfW%+~Gn`dSBbos_Rji{dk4aiP7hO>!P+J7dF1H zw6Q|b63OzpVDCCJ%f{c^scMjOkljCx7kl|tVsHI=T8ns1C6wTO?$QCZ)_|(*h`ZnS zuPFC>mW5t>S_zk%9b>us^Q}!?=3Al9czP1&cO`X|ju*)WcEu_9QO`|Pb21^wheJS@ zTJi56#vDR#KMPhFnkL4Oh`szqmpHcsdoNq5mv*94_3mjqGofg3jJ!0HgyIevErJ7b zQS(9QyPXHNz0`Fve#aVghW0FKQ(Po>ZwAS>X)rB6&!<{Y9Ng6GOiuU2yUo1_K zWtp#!5ZZcXnO998a&f7lvh8RNom1^VDg?%}lxkom#dG`7Zqo{SDa$l=vf&9QH0D$r z=ofk2DOXkC4kSo41P(0znll(V#~M>bB%MLVODbOACYI=-9b& z(nC~I%fMy;zro7f(jdR~jFT!e{x^#_=^nvzQ&EO8@2swRmN9_?iI<&m2kW>I%yHHa z;=$>Frb^|7p%b{Hkq~9xMthm)oBab5N$i=SB#)cZzJN2K8@&y>D!J-=`i6T*JmJ~L z%*WmaiYjat)>-Ne!Gk|v21j#5mCIbr8_>hb*T<)EX%y`_%@pm|WN4kU+qBRCoHe+@ z401bI*m|f0nmm@Q#dpfYbtzyk}w3W&6E0Hrb~gY&gD3J77MUBikk0LGa%J2?fPSo@WX8_K|=Ul9A`tWV}a^ zYyK_la(s5)n@ww4>tUpskX6AG-`|)or0O8zuL}jKP}F%v%-M@*rRTe%RIeY^Ao6L0 z!E+|clV2r~@ps}xr^!Xf*)KVr5)p^jFV~T?^vloRKos`A03?3dXWqVYam9)VCG)xP zChTebh36%PXxT)f4Y`GMgW+Ai7lP09%&nJXnNUj!oopswioWn|=jzW>wb)GgUY|to zk1}#0FplJ#a9>=X+`h9!xQ|Z<3xHtb$aT%?3>`bvvIl?QCUR!`pa!9VajW3)c;_TC_17Ds;sYW`PJv z!ZQ~5GZkyGBuC#2igIE|)cgg;?d= z+2`uo_@sO`Po><3Yd^zDu0*vJIPOQI3^jsTyscoPt!#v))aiiw-bg}b7Q@ZzM z#w9<>oQ7^%eAM}nPzP(*cy)Z?FI>@IrW8)I^8hQA2wIudfuz{?as8foPBD$vDel+z zdJFZts4bqm2RLQB9ZUy`Hn1m7NHXrFUOSXIbAG-}-hNM3TUNneW0`wh<&FR+ZGt{91b78tqNg+67^xN5X!c zPe5YM0rh5zYVEPABrw*sG~>*~dC2}!4LTiE1k(l+=g&oTEDFtwZ_Pu#U1*z}xNWp} zWfLSto$Vqx^uD%xg4dQj)y5s@v^kZxXR%rw8w}8sK~>#$u<%~(G?qewWV=FJ>Xt%& zW3QUZ7!k#@*q=%O4zOQTj;TgH4@>U|V5^H8zIicO%CSj7o|<@=B;M1l*e(L1vqS5U zy2n%n5lsEwIe(dFXJs4d33ha#7BH23OCs~ulON@vT1QAI&dvJJO>EZKAl=?}jq}JN zR0bto%9q&`kH{y>w!RFOG$LI;zS~NVltDMsmI(PC zUgyMq5m32wHbFsDZ!MJb^GZ~OxCGwI=BW&9{?kx;O^2J3j2#b=9#Nf#P+hfr*<7!0 za~X8>zwkSXg{^5r-^SgxiUQ~Q>@l$93c*y!^{FrG(9mgFS&r9vsA!7B zQg5m3jObib|3s(X{WqEmP)bN^Fye`%G%U|kv1F<|(iYy({JB}&2Dqp?j!82-fuR{d zuL6-_3ym6zZK|EKP@290a$60UhhCPSj&w_kLwgd-cV`0b)sI=cf03}Ts6<+Y`96?{ zFOXM8;P90?Xvl=HoJ*Xiz;KTb&fnSs#Z~TM|HJ{KqNMp0K7ev8TnLv!(u&{uv2Pq0 z9N}sPZ|@JjWxI?MpZ;}vFaG1%g1~t)DatkjYIT`{B7}5${&VQmhwV#wV#AKHDh!>B zCm^x`KPE_R9zn^5us}1{bfQqK})>DB3mpZvOxi3&gNLl)kH$%&2aB7sPzR1Kn^GpZDi zX^)r*UuOi6%_76^)Yh>Yl@eqK)o~(M*C3L#GO!F9J*Mh;!OJUQmk(j+R3StkboZP2 zzOP@ia(&a^gH6>BSZuJqPg2SzWF9MkQgmcRQZFWXy}1Kk*0fmZ4mAkUsT^x-54rQBljPhcCuzxg{buP|~`(vIP>}jsNDV3xU@H zMfc=Yw9i*@lu+VK7)*Et?JX*4AWvNvtB9MXzF2v8a^8q-z?M4~<5c)TOO`kGtB-|q z%*+^abcXlxU=?i)B zDf3cWd*;I(Nz>J^UE^AAr~8ZLlOTtm`F)Ed=e8ZR@jzj23Yd-t#ItLmg0wJnqRZzY z)GB4M&%0`7#TY}~Ge9J4^sOxH&ps%SDdjPM#(bT2%@PGbR-zaP1XPR|I`!kj8Iz88o`&A{+Ata(_(JB? zjCi|As6PM&$BMs(g4$hev;(jw?Sl_}=W4|q<`o8_tM9!JiBm*_R^;qv6J}Mk(Am*5 z!sxQ)Fg}bOP@Ai;&aukUG zu4M;px+N|*P+Vx;Dzv5-rjlm^*{vEZ#%&iEs*SgN_f+fqJ82&aRNu6Vcq|#tx)3*A z*2wiVF*dD@MCES@A@@T@r>96!)>o?O!%OV6^-1>hjgD^Qm$w%PB}#IeB+>7YpiwcA&h5( zQ=4(^pjk(u#c`>*p$ZS>A(kl?I;zY|6M3ZH6qa#ZKOutKkBKW4!>rV+YLj&5NZSr8 zpYdW|_Mq&)aGq|7{|`<+Y>VUG0S~%NF2?SF&#&P~Y(Gh%>=wx|9fMBq1c5hoqY({K zXQd<2mwfvSI5o4J2~E+|%CErg%sP(!uhQc3_*(w3b#3$Gf=b4_uc`{p99=>e=M}b8 z@scns?`P1`cbRTCN*&n?A%e-Xu6}`g@{e8mZ4Gmg@wuH%>p4+1G9@*L;0OP{^6o$X z7{w*3@V2Y~8CpM`K->Y`u(3bXMllNIk_2OabHyKGf(dQcXV$trUL54Y!u7_>=%Q&< zyxU`y4yvDs*?mP!Pk?tBkysfx6>j~93Wl5IUMAjH^$`7;Dp9{H-LdLo9hb36uuqC3 zq}z-zjJyn0SFPgUrS-0ik;pM9HE2w!NBKs_D8|RE#127J4aLl{3b>jNjor$&jPs>>XJ6A0lHgTl^1JSf|K0x+Y98-p zGSrOM*sE!V=`GvKB^Ho8d5jZ-{{ui+jPG7!lt)VD2~@uqv!&p4XG;vi>HW4!AX#4` zH9wC8sX3)ikg!`U!|y97#Uq1|E5-EA8EQUN?=4lYnUW_7*bhErY$;N1EK=uqLk8%S z(Gg=%`WjDn+>(V6j=`H?0+*5)iz`+ZB)a}P;eIdZ=h@?*w|pI!X%62#OKQ4@wr;(qdNoC;dfltIHs=4vzU@mssm1(Ooo{v7xH$IADk?_E?u$o}0_=9Eqp%PmBzcK_u zGLFbE3cjTF?0Z-0#vdmw8oD2R50vmg;)U$>HTA)l(=;VEnz}49%r1<5mQ43U&dYUf z3{;W1v(Um@gbB1%QF`o)2V^Ibqjc4SYg&lWst)gkk~EbbuPJCFCmT-68K8prDMwC1>kJXGP%qTNOGnv2Jlpo_wUsWxB!FJqQm!!lBS`_Xb8kueuY`XemQ_oX%XDZaYo!6U5hM77A zvm7H8FKw7jTd(@_K7mBJ_b18=p{Ubr1e@a=TQ{-@JM!wB1zMyr6se>~0ZSUDg zk%;q^z^YQ_Dyqrq^mcw}ip6Ywc;Askc%X00JjsD1VnrbcYsShSG&EFtmve2^nXa=% z1%)vMyaZl=&2uf8=KDiaW6_Z{>Q;ZdV>A1NpFsmOzQ%0u94^ojZNx+NK{iA0(! znj-ut{a%C!g$f4n>{ISr3d5Oq6^A1ocL3$y`aWUpm-*c;8b+J1%kJ~K#9PaZpmI=h8(JtXgMu#}L^unaoJ2fU9? zzaoabv;eRQeq)>zj+Y?#dHPQgNLR47Hw?U}Dhl1`-;8=DR%6UJ3sbH)v+F==gDM7T z8;}u-FNZnSSx0G`kHuNMt(})bXS`I#Jb8vm-;I=De?0rm#~z*|Y5K=sWY=?`{5DJo zV);odBKRB?jpW?-!sSb(`Ch?TqqdhtRue6veW>tlTRDy^7^OGTWw6xhYw5 z$6XXw6GfFY`f|~m&l(<^8j0ZW`g+}#-xD|Q=OJ`Y;is$0BHHMh3!l+t)mquTE}RgZ z@zN8&)*n_8JSI8&kEgQ=X7>bbYzp<-ZQf>iurXofL6Ct~nsk;%x)RLFwSM^h*Zzmg z;HYs9alK2k0V$l;a(yFNXqLvA-%IZB(2E$B@r}Wl(snQF_5c)620_h1;$8}|E$=?h z?83wIuOuqTkZC+tzcZswLQsYsu$pKim6aiD+d$xlt<`kv-2QWJv26wWWXj zgPx;t=7Po_)4K3|U(C}8naBB~tF# zUu#7z!#;j13yrPG9oSYvC+2v)cwrvYf0a!r2zyz&=fe{I)q1|{s3xVvs_4axjqCl` zbj}3$1!b?~RJU!SQVGnU*Ju0Fz5d*h(sogFW0p9+n;Enp!-Q+bDJDnzUSR$w$!&vNK9ymWJY7cPvFvk?ep<=UFsFw5sF!)9&)x@}YcyYs=zix4Heo z(?G$HUNo{9KQScL+#W%`S74+(y!5ur!?2V2YiX${yzqTy-i*K;cc`Caos^9cb^DuB z{MxM`h2x0!7g?UPW+ketCOC#ec;CHTRXjtV_Nab!0?AAyu%Y#wU)T$^8RJb>$kZZ_ ziM{)G7Tm=D#olnMcRC7wA(B@jqIn?WcYQ2-D&UYOmMauTcJZa6oOH&$NBqS#|qIBGk7xk zo{H+!6{MTKUwWRRJ;*tWy4IykZHkF`7M~s`Qd(ky2Gj1}v2!HYOXT0xd|KOo_GmTk zGhsg9l!9&V3l(c8iRZG6@B9g^WDGO~`rt!v;ZM;zuct#dw9pk^3(YwW#MO>N-y;r8 zky`8SQcI>^np+m=OHWsQ%(e)Xm4CPSHNyhcyW7=R40%3nUCV@Ec5b0J&*<>xP4*dl zH|Sem)dq6{Nb(#>zv5(0RcMCNjT3v8FA1!TXhnZffw6(0<>|)buV>tFZy!71c{=ec zNPH8P3TiK@g~-|2lQprCuq8ZT+a<+A(NqV8dOf|~>7iZ;^%L_dRSP56a|b#@IPAB& z2zTw4Y(JML7wQ=;8qM2W)?Nv*x}C}4!hF?dD+nHJD? zbk2NT3devrlm%Ql;~t6ksqkyN6veZS^43a*pv`}8UCU@fN-HDbh}<2?{q`8LS+?9m zLL?o8&l$}sxt&rmouNH41;lvbB-nT7 z&+H8lF9OY5uNWZW*t zUMUVnPz`pNoZtJ6@OW+lcZi2sabtr=)qW!&j)9>KmpE*VXhvy2s{xzYdK z%}KhzydjRA2(D4Cijxy};rys;14|+$O3mV!dmqHlj2z*7sqb2MS&?k7mEBg@k{z4` znf?({N2aNoN6z%m)9n8%5d{4{mD0Qt6ITwI;=((?e`AvRTCi!Mcjfez zBrSMmXax~O$3IKO@L%Xx<$urQ8{1cK;R36X^E;}+HWhlxw_o=<`$#V-0m8O7jgQ$H z|8gwzGbHu;))Ucz^a`STc4>d8$^3Ud5az%CL_Eq?D`kzU!>|N7J3eQAt!QlHZN4Kp z@2P`uxC39v?QwGJSJ6AS@qpo_$$llNZW~bbo zAT)O*;A>OpuDL*TFGcuBPm&YbM4^8foCQf(+%vw{T9f2>MDvPmLJ=E9DET1>`LGmG zRkR-cvYE(y$V!Bpz=S*DxAY=onnZq=0w?db_92%R45(#cD1yp>JW8ec>Gt)um>8CK z(i*(8?VP@_#uDbJ0fC!Pm2OXPR~*|rRPWu7yGH0)2)%RwVU0|p^itB<@pT{w8cfg& zW8J8%z(h7z69b8SNl8!3+jDUjm{=FnfQvHo?kjzQ_(Fm0#O;HI=wI5`M%8OqlDskr zfdE>GIco7{g`y? zCakY`vfMrQrg|CZXYZTu1T7q>PH>7-6u}zhLZm0%SBMO|pVV{BoQewuc@x~E(w_I~ z1G{Um@O4!)_92DNOd^BZ2hLrJ&hDP~z-Urje=>1&p(8`2K+1qyO&&laL1I)Cxy z=)_#1`-B+tNFcejen-k@H#%C77cEJx^PHSJ?u)CC?k8I92SHKiE8#vVRV7{wFPdkx zLeHb--o%@W9*gN#n91V<&ey|bMK)ksF3jXFErtHhdCblo4~ZWjAW6eJ*_3cEvPm!V z?wRYBnj)C5;W%b$iGIg-`#5KQm##(3mq9BvRtto%R?n47RJB_gxh!~UnF6BC8F}Pj z04UD&6GhTM_|FRW=p|Kr)p#~_hofm$tNORH1m%d)M{QDHvB(j+#L(4pZu8^rJfR8La(^JL-yt+0e$V%gsBMAFGkVbeAD4INs0u-APny?HxG&?lu|<^x-C zPh10eormmHpnNal+dL#$R7AFtesz&w;wXGPOVSw}nWXm@^Z43|1%eykg0GKUkq6GG ze{f1JZ!?4_1I)31Hdy+>yr2Q2tDuq=_y`a%w*v*UrsF zgJq8W7+S%801X73^;Otckil}G6rGv(A$k+lDy>J@J82Z)$AVp0v~H2t2lpe!ekW=K#^ zUIvZHE@&Xb$=UEKtVyl_E&(NsbaTA-^DX{#&coaM$)aVWF2}eSkVagxJR6svsU)M< z`{y<3hty%2oE$2QFWoc+BU2UM$DGouyD^YG3P~+#;^%4qV2*|A)Hwj*Ft})%Fi$ystnkkAbh1-d1*(6m#&=lf>f@64S!=iYnfkNJaN z_ik#3+O?}{ueF}%S-*UDDgk&4m;$EghjeTV;>&8CH`As)e3dGJ{!&0@mI!L$5x2LR z6KFIJrFUkGY?KM926yG#JV@~-!gnX!1|Cv22-eCMh$=9tiND3Mk9GoQ>vJ=xfg0zL zp-T%Fn07|-r)I621N zL{F$fGq}>9!Y_61c!T0Vj2J@Q)L7IuY zJ7uD#GIa4izhFvjlCdzcW$(}Zb`jfaUrLh9 zlVm48d?Z2nDUm9S-qRy>XrTgaZ(Kay}A;t)tiGbl#+=O_5+lJ0lnWWZ4h%?MIExTYz0 zw(fo0w-(FdIMih73UfH8&!6&7l~j1v7RF4?9>^+pG|h<-6Sr#`I&g&($((Y~6#ow~ zp1;YS{dwy%5B>MZNNrkc7Z3s;uBW^92JcvKy}F<`y5}p1(L|UX3b-A;ZtM`ZF~N@T zWC`?7w4Wa{f4ZM>)xz0qDZPJ?4IsrZNqGW*V{#OH%^e>`j#h5W|8=>ji1Z5ph%rvZ z7%}8&cC~4`W>ro_u`p+hVJkm%UU>5n3z~pA@Ba7Xc0aGpYbzC>gddri?z#*_ewm(M zM<$;TI8_zz({CdF2dul=~uX)IxH~i!6-6)RC;YprcJ-k$Zm4|sIMG5-zTWPe! z_qjWKwJ1e!$t5F`11j=}^9h}n9V7K%?K48PH%po>(#4MP5((8c=`(w*sVjMoZwN=` zc0~tz?#Xl|zBb(6-cqfE9M_htbEH$InY8CQ2Mt>7H|k!IECa6$Z-Bb_50y8Apo8JY zYlZFe*S`!=Vvywa=ak?KO>xO9(^XM-K(J)fK}^GaM@i>Zjs zWcOSOn+zFl;XN*mCAj%MH}{m;VDjJ1)-~+d8Gfr4P&%r{tqrtQPc>g1N)+eqBv(g3 zwZVt@4auIRkuzf545O8 zHcj^8vxV*#ZPLXHN4Y__KM>{j-!@q;iz{4ly66-@xeR0s6fX~u!X!QLwLKE+^t@W% z=k>`~u9JuzB#@J@iJFeu+jxQb&uP;l80qPgiV2Q((*FQR+(<3gi92*3&DAQ*dzH&j zx80{Hi&p(2iZi&Oa<3b7IDzkBLF#2{gh{~yQ2~gv&~=T)zq2y`#)R-%W!@}v{%{ya}=;Rq(Zad2%o^3pl>1~p|R-)Q(Sp05}V8|KckF6P=fA1w<53>WCc zD7*`IecyfDr^3n6>qYQ_J9Gl#NuKxIPCxtcR)oSH!%*OEunuI+;SdggUC(w%s>Ns0 zTq&i7m+RGoXj#=(Pd7m}CoL?oH7z>}1vsheb2=popMg>a!(>(_2^|5eFQ7t2+P(qeL_1Z)YIqj5yW1QQcI4vlMVQ3@`bhx9vRVb@PQ^wpux# z*{x$PQB9WzYCe1%!2`w|H|TSjtX-%CaIT$yMw*I(7&iq)WR+H&{AkpN!oEscMi;-PCHg~ElUDl|3 zGIhUX%(MJn`o6lkUjey#5iG0UQWiD?3xbdG>4^4=7^{=y**;jozj6?cD8)a{cMaH! z-#c@Q^*uLj-;)WAsoKAMqtD-b+oYHLYadD2R%-N+s{?LBQ5{AIQAJMcIJ0V^?t$BL zF9NZ=EuYqRXTO>caPmZJw>y*>m$bxjxwR#3P@U3<2b`yrFFOO+*Ogo;qVX${#r{@KC0#Xb0=B-3txcgUO!gL3o3J=rido z+8@scFOl(7m(Vd%9T}ZkZiBxzBlZHjFrB|BhSTtebMznl5lH(0ic@@=fM-Jf7@G_& zO86Jd7Rc(~N!HhT{!g&2fAwSidBTjmi1+>~=6DzEGk5!PcDLZx7)3gRFuU+|Ut@5W zu2veLU3BMmZOCn&LKNN=-ixzw<0mS`W=2IuyHLs-!L6?bhi!`#N9)?Tic+iHN1rwUM|EK?_ zd8R8m1FHrn6fff5tzO*}2*5#wm|EQ&RwNcHlA#4WQfg4&wy3Dus`AMfAvY$S-J_*# zfLVts7dd46#dyE>?zq4H<6LiNG$lwuL{qMwr&50R#*exInb+B3sfum;x2_#Y{@uy_ z<+O{JHS>?WwGoHQT=yK8XKt=hU+MO^8)+|I9`AQOuNb{P#U!~uSrG`7?l~0P%3Q`v z-7@AdNawLZgy8{;m_E4V_V^QZ;gL6dS%rxv{Y8@?T0Bdz=RXU3Ak*MQQmr|dC{#T84}`T%8OGG$B3tDph{0v3?lORUoRm7 z;lHsIN^0G@yMFQmzCsJOUyGV19fLGuyPQ0UFe73Q4jq>V&*lolpG(eqN&#fu?7b=` zW8@FU(mBa#3-i6pcf3z!N7eMM2qy+fh}w2?A1EvpkO&y5q6A`+T*ZHb$Op<+xd4RZ zY+nFhzl|+t)s0V{(pv@)Tz@}>Nks0PT908;F0ySb2BhgXT*eSgl3POO4Vf4vu4XxF(79c-|HsFx4nZ2d+$&&85hCm+?*W@_{hGq9{Cz-j~RedQY1o*Dfc%(K>LFjR3GL)4^hkbH<; z8mqs~d$7d8Fku`RzyB$^w8&9rIMylKBIDwnX3C-Ao?vg_rn}Uh5^Xk|a_ZY#-&>f; zU>orXf6<{f_2dFrmS*xH?esk7)uR`hoRmN$ov(~HT9iSRS8cUrllA?lK-Yv)hta!x zfp!`3UzFdPcGnYd6v`@%gRIf%2-+mL8-?St0LIV7r^4muW8o^^Q5OcTXzBCX&lQ%t zrfQ{rq^su{Ha=eXa+T%H>i$SiiFyX+lL4#$F7ZHSZ**6Y&EHA=m;Odal;*B^?0ca_ zC+R<{AG%3Oh|z}2t7ydAMVrN%>MZfG24)sO zrOTfAv`L~k=q--TRVR!H^aW`jCKN1hN9-5DD*oK^Zf`ba_L)E6^{_ z?FTTy@tqWn)2?^h+Up$u4Ld;}ze%FboVSKOTi1cFQigDOfACJAoy|I0&hhz~wcFCY zAE^Jo08i{#k;-Wn``bL67jp^Hht5J}nSQOo1xs!>hUc&^Dh6@PijOG!krg6y5CY#D zUrF)In;Os6&O$%j+H58e-(`&cfUui)wY=&Lq>BrC2Tl4hPA&5AV`?A!-hLSr#=fyx zh7LSmvC`H)sClw?bZqmpudg+>H>~gN0~sK#jGStKb2&g{grChHj!7&WIJ+~N6dUiEO*c)sDk$JpIQ-hV7pvH`;KSqi85?;;bg3>P}1Of%DBAhuQ4o7cs2BG zQ5}BBBBuzy65y83wrz}6<9rdZbR0ihK+XtorL|{-ybTUM`HvL5*pL5mcqtIXOHa?j zmspPuqA8D8Y$~0vEq~OJcg`ddTR+5E)tI~KA$5V_YxwVvxBs0M&401>4D9pFmlK3g z0_LF`^rP0K=8RzIkc;yR_ZMp$b{~hguJi~T$>S;A7@QPX%+ya_|Cl)#U++7TTE>@| zpCsSUVw_00dA5+_Q8vzfxX_Hd^O#b4!v_!)J^>T%HMa>3fT_N&IvLSz3vhGX9O!#x zB_ypkCX+fm8Ep8_NrAf){S2N{+O|;3ztxs?Ws2xk3!3DV#epW9v9|oU2@1-vbHK9K zFY?S4Hi)>*)|xJa9Hiv>3{`Byj(dvLI~I1J$m3)mX%}v&|B?6TUdh$tC+d`s%x{$V zb&EfPFUE{#mTvO(FF@lM2QLvc$=aWv+B96-MzVLIQ&hxP> z1PiT3sP>gUczQsw`SLgCEPPxyQI{Z_h)`>FdHJe%Bw1F?K9;NrWm%8Vd8&B_nrmuW zvZWrz;y03;^?GYZ&e~A^#HsB#Qi1-*+>V9TJ}PF#){8SE__xD=h)jQ4PC4T zL%X>+B3yW^76jAy-fwn7yV^y$`dwEvdv?E_(M;_zCT#&h9`OU7_K($@vo2fl_;25- z@uble@GOMM1~98KX{`Gw+n^2c6a-@Qkv8ab_0MfxDWmq$U!PHad7c_3y>n-g$3H83 zr(cnF_p>wO$C}(g7g&9Qc3I-!w=|;)hCy=X_-#2MJCEYv8*^t>IyMGoaj(BGMeqLd z{cii>eb``c&|3%GmFn%F9A2Q`sI4T6A~5Vt;-SK3sg9*l<@){_`#g*f&g5da-^)}Z z;C4pb!rA|mS5Knq3@@*IEfXA&NfOvg=gm?l zFOvZeS=ZEJa~qGN06<80m8Kh`JO`CN>mo-#)9TF94Y4#K9rz}vk*X&0khTAEL9o5d zGFY@@`YcJveBSl+Dq-nRsQUQm@je_))_PU_hknNkQAD0s`4pjd1{k?(LgW^D@GEUE zD6_s2vdKi~;%MxdyR0WbYZSW|+CC1J2jQ*p%$pBASmiS@wcU7>yK%^D4H)j8MKDSs zN@E9^g)AR4B+23-f?o*1jsg^z%v?MJs>v$>-HdDTtNW4!;Js8szMH&aOMb^#Sf*i+O=%Ugr>5Lt(*m6_@{I(G7Ow_t zMO^d;cD7foJiK_?XXE(7+ggJ|C2WYrIBy#&ZpKEjyLRJ>tY(wXSw8E9?fl0tk6pYj zl!q?Yz7~+HRkLvPPw%gBE{fSf9agj)GD?L~xzAR+hv*@EjdRsvDa>fQU?GTq{Q2_v z?#fySY;57@4XK_~C4u$$E*tm~@%Z za{SL6*^j!tmRt)!P6O!SSz`91taB<0ElX!loSS}U94)5I7{US}99$TcKxp~l^e5)#^T4vFFN^Q-;M zbd>9Sn*6mUxrcIT0pfoBHd^@uGHK|X!N4#(cmBo>S`$B|_xlHo>9BlL&{-Q_Zy}2N z^Exg@MQRR8<@KV1Dabi)HEm0c^D%)O^Tu27{jUocX@opo?=%0JV9hFv0}pyP*E>`h z%|GY9BaDZT8+j!=E}+2g_6OOC*=jLC6$4I-I?lw}TYN5p)V5*nK}IYHc~>xnQz+$( zx7pVX9nI)M4*^&dy=lP&Au~MgM*D?oDg!gVA^n{TYMhVoRoC_YJl6UJf#=mm9)vg2 z)5_cO3x_{3vc+!#G>Mu$--TrM8Czq^-A+TzTO)eHD+lIAk>a+7-KR6miCFpqvW0(IVSoE1v-z9Nsw{_G z?mqdZ^|p`x+fR)i1NBOtf?Dw34#*9imdv_W?hn=#m6NfXAI)jD{9uvebr3e;ad@ad zXYapG4I`_5j?Nysvl-19>R*hd%@6O|wX6RP>P$f!iQ6D!;6D}~Xp&_LZra{-y}4CB z#ucF9IceFj5dGWrqK-^#&swFt>k$R3w1*zz)AbQ8^na zmkg)JS6Ajcn0xs+fMrb6g%ouAtMX77&<_b9s!j*&=jqh`)bBzz$Wa0H6#yRzI7$e^ z54j1Rnk)maqs)ge_rN#6Q!K!!f%^A>&bz`Dcd2-=H%EVi2J{Pp?y=}%`1Ai9UkV62 zJp}x&y~9YWYf?rlEMARgKC%=>44i8vFj}j>szbs7881of&5?grcf;_z0n;%xT?*dN z0&Gp7dnNLT%ZF50AOE@3TUof&RmL0!Gi_B;DzOCfKC+9&|nJ${Z!!=ph(7-k0Ao1im}u& z5sEt#fPdgdb`tJajl(Xu0up>pYAS>!PzO%!&lL>X0L(AOMqLbf0$?W8vH`kf5#XJf zW1x;HFp+*hVd1aBp9YS);IhaVpaC{4$=%-|L?sw5^Y?O9Byi?T8<1ZYfPcL^9dP0i ziS}5X^j-k1>HL`mK<=;C1I~8I=&g&P2YfSs4Z#hrc*!Vc^>05I0Skdhs0Vy9cm-6_ zMeF~gIPs?lzMOMl{i0p~?sbOi0*lA$-$o5w)Hjp8<=9si!2IAMfPqy_g5j$EqE}hK z^a6`!;m;MJ+L%1rXF~&PaG_Y~{HZ(@mP&{?|JMgGFox{Ux&LXUuaE6ic#jRuLiZa) zW9bF-&C$~dSdVz*SZ-i86eB}?*f6))Fn^uQ9vd-~oEI`M2p1K(1et$mnA6 zhp^(?>&QYF%k`l)Fmk}n_ea2th^}XSeGXWs>|k`Z7`Yt>uoU5Z$en9~QwWA61>*>Y z-$(BK2E9KAMjoCDT;zZ(46hYXd(r|}OgDgG#r-)KLO76q^AmXEAz;)}|Cy3G4q-Uy(CaazB)Rz%1D z8Oa|sgf0ncmhBJ`u8sGaEL!^s>@B@9~!r6XZ&UgJ)Js(*h$MqYe7}$0F z8w9M!*uQ4f2usOAg^gqcWAVz-e{M46qWoV8OTaAyP6|-R@RAdPC%1u>3kJrw>79?9 z#8Lu__0R^u^bG;acXvwuU5W{IZF&+jP_2mMQXDLaG?=ZmyRBSdI?BGAI##W_K^j1g z5xsZ=DF6Ki0l=QWkaxo<#U4@uRpctRs@!QgQPNNk+AH24tCztr^d(+ds$2R%SLxhS zA-t$XC~}Rx3nkPys$HYGY*t|#M^HGA2Ah!vCqsKmv8Di{1I z0Lle_(@XjP`~ANVF$$4yLoUV#%OYyDjghIdwKd{~lorP8vfLv|gQs=`5O?P(fG#M9 zYDfLh2f4e6^Tv`=dRQD=P^Fy?P@15Y{ znQDhMY7}b{LWts**WT0JY}n6lX@kn{$)jPyNiCC`7KhY0U;Ll-Nw(4T0eUw+0`03q zZk(!!mo}rcL!7wD>P}n73QYcSFpiBwks-a$&2wVXW1r3KK8TJmsZE*X+A{k=jDTS* z>LJ&YV0Q8uF>wd(ukh8|Uoh z9C*kokz2*bE4~hdz6qjW90(>w+JnHOjsQEc?9vhs8t6B?$D%*O)K_ttcX+ZVW`l_l zRcS%TWLb1@s zG3lCr;HFjDkukHAOm0?cz~d)#)(@EOSff@HP34&2?UUr%5czL!rRKz?-R}N;|FL4J z`*rvvOT$?$z=+%mc7bcJ`kanGQ)`c(HJd3DSsoEdXNq_*jNZZXqeT0I8S zg6&KkO}!7i&dmq{7!d|nr3qxzfkwsCO3ZZqy0}gX_p=8N)L;=4M`CFU-@o%qu4p{i ze?YGA;jD(78vS5@jcv9@Eay#q=L~1Ei-Mu~ClL}*-&Kjbb4^j0loUUf{eUql-sxm7 zP$#Of*JMQ}Q!Vp=t>=V1LpSuS5DJNV{+TZ?}Ipa~FL&Nb@1+;#BKEM1K&u8Y?ss^cITXJ-ik zvBKu#7NqRu;^W*!(bhvN>j8Ztu>U|Jq4 zXXN4F!?dDeL(`GD$oRc;mC#Rh0ul~ z$4>THE8Y1T%Khmj(MMZh8qzGc3dp6=vPC~!o*9cm$4fjrKb8*nm423Gl}}*9_;#39 zEiEnYHR*jdQ3iM|_bH`{&F`lu-5&r2KvEQ#_E~vR^0kHKYi28^x|C*4^}}O~4w*`u zuGA6!L!7=4jYr0t&sB*&fk3`DKK>(t*WVxoe}oskSZ@MQ=a)j4zz%5w;-AuIY88L# zDAr-QKGEb!Zw*h4QG2$Oax$R*@9^k2ejAf4!T0mDkBKZat9JudxpHc^_1Ae7PdRty zUbd6t2OCZWn|+>NzfycI$7He+qxeHo?b1D@YW5%u)zu3`@JGkmfX#-zQ+aBDYw?ca zAw=yG8M3!_mBrBV(T)Z|h!A4I@O%UZL#lyH zE5{k3r?1J^9)+54{00RCra*L6v<6Yxsm(wf)%dw2o$o?8>3F|hjCHK7d;9xSTzOrn zj%5(a9xEp&3xxLD!3Rn?pIdh=?lU=81r5ViOb~f;shvC|A~K2b!Cm`S*Rz%?8^_d; zHwu&kl^w+v?DL#asaB3SB^(@)Y4WmDWQO*2)8-_dGK-C)qlC~d;*bQ*&X0FA7Y=RH zxZT=2K%0t|d@4pL-{R|bi8y!%y;D}xFvN#5J@#d!-|ABhX`5Z%IxmjeqLwN*wgv~8 z?*m6^=h;mO3_&>r?Gy&5));Gy5wL}L9vgUGPd2M}UgPDVo7pJy#`!f*& zMLscWkNn3d6Ti#O!m{93^Dk$-Gv^?#-n#}ICyzA3m@|Fl2~Ngo3B&@SG{X)GBJHh< zc2mL1z~;)%DSLIEgZOw^3RET?xY@`mw&tXv2pjNIHiUG2tgk0wO(^6ztQmxPN?e$&1 zHqLC9)i?J~2yzw2o;-N)K<$C@r1YL@%oxO`rllof0Y4An0FP5RO?zO#_SA8|=bgw~ zWd=OIT3JLhCSBeh+ggr|ZggVho#W@46|V7b0j@F*RHh`P_n}!*JP2m?hZAs&;W1 zyFZgUXt6298B0!!N7$pruLerL?p+d_A2gpA?NujrKW|F>4WdrFefhF{8fNF`5fW5# z5U4!s6J+M)Sx@*Tb&~BOHSBC7dCpimb#dC9CrU1)(!Vq49)GYjXQ1;9!DCjBCZ3*O zdKARxBe;o+eoEYcMA)l?Kx_tlTpj{-UaGNXA|nQ3zWkaDh*< z23uwbLPv5`kZ|_2Z};$(m)^($+i=I0BLb%*x!Xu*6=Zu^VI|vPh>h&nN$pdKgN$eE ziv6BJiAJT*I+~=vOLE8pIVplAi%j$vcY}Y~YAKC?9&cD85q5uHq?MXaIJvqADoOQ#cmeSNBb@H+DssRM?p|@fo zPntK(c4ovmWuscm*`IuiaLtJkLu~3_n6HhA$?8}f9h$+UJzHX9q?uHqoUtDkN z_plvut3w0lKOWW=JF={1^c&IsBq^x}vW@;w)rO#*)lq7BuYGcE)fOuH`S-D(88TO1 z&}_=2^yM=IOSWe{M+#utqH~}8^8Tf^Gt}k{{siy1DbG;t1Z_Q}=?i{7>*gjW>KyRG z^+IH?dD>EpS7hST6L;A>r~JrHq212AsjtO_uhTz8*7ZsE%6j+K?|NZ)2%5!r8=EBB z(xx_c(9Lk%;*NfqxXkZw-!vHKPY5MbxytOMXl)*O2X!!QtR=ZsJbclbWfbif7jv)k zsgeSEAV><@O!W(K>&^Pfto7=zxkhtlvu|((J^;}MEe9`|E&@K0S`~v??+_jYcT2E- z<$ktBQ^B5xosWwge%{ zgWP8XCU7|KSd-C}IR_1Ze<~APM;>;aYn_%F02>V&rEZWlk*T2Ivq=SR@Q6es0N+Va zye4&)IO%UvU_{SWPF5bpPq@g?FK@Bi#qMzU5`a!8Yx(9KYZ#@hhpXwWx9smGBn_G6 znF)OFO2OV;)-BYSb-Z+WgBlO~^pJt;X(0|ruD$pdbm)3-_LD?!0AY5pyq#XPl$q?T z+MYpE_k_d={}u-e-`Zxplpd!>MPyM$H+6~|-e>+aK)5Dk!45B_un4iW??2b(=@80> z%#RZgt9i&G(R-<#lM@tu@=9ZCb4-r>@#b1}uo`MD3bMPv7@FjDe&M{!mAaYK)LS@6 zOew{gYja@Pr?2jUr)4N*qsqg1A4GUo1Rp6)-QF>PBGvm3U3V;AHPtQH#A`bbKGYK- zq2~0K-|qy(lh`OR4#@{Pt6aH(k5jpN9##pH5%T2~P$gb5p51$y?0t{l@hCtQB96$KIKH*_6!`@S6Q6$*_Kj(h_;( zFaZI>Apipu;2WSgNOzK9g*cs`6jd$l^Byw4R*V-Z=a3(e?PFzVpE)B{H=c*sz~-Sh ziVUMgQU~J|o-1iev6j8BK6-`*u)Um zsDepLdSWCuA^j;m#JJ*n#JxHGY0fa_uzhmHB=(l|XTzAALB&D?E;?erj#<}!_9%>t z9}G`=KmVRfx0Ou(UGJOYqOgCSTTC1kuywTD=c}y3`I%tV3HXFUD%g^ONa`X;QcGU? zJgjdQ^aUccRUeRx(NI|(*9%M3&F|2KDi6l)hy{^)70Q#S)d|h5*;cs@GSR`dSNShc z?Beo)$NF{C$NZUiu+JHr#^0Y^Fhi8owZJg~a*|#M?UfpQ7prGx+JrP`d^u%lH2ma8 zgYc2#9*Bwm%=3G9`3g6xNFvQ+5<=P0e%%Y(gj*&;pfO*GrS-$SVTG}{@PQzH5 z_Z#RJCG9jxc%HJtQn)*>)132B>Vr}oFzg`Gr`TDhI_BjFV`iM$^;>DL)6H)ST%Adq z$B82Rm+xX8B1b7kDewZSe!u`%S3q4y_n~DMT-w9Y2G5M1x#_+L zKU7YyEG-Hs3rBhxqPPp%j{9BI%;It~>NFL0u?G*}>R0!EEjAcCo(EdIGc=^FH#2+m zAoWqW0T{>7q15ZnKd832|J!#LbAJEmRfiCK0vcrF!PPOR7s*`T#PDQ8O1!y1hifRt zd!)=az_u}Nb!n@fmyo+I^xn#-4|2#Vo?7rX=#@bDB9XFUfq3c0%9V>sPoVsK^#~{T z1beIQL+TskEqA5}gy3K*ZTv3OjSw5X_j}70%hc;x6zb5pW&9^nyZ13&v@izJF%OAP z1I!hY#BS_y#Gbec-*6}4>omQlU$WW=2FA5k-(+S{$qd~834zixvtFxuUuI-LyMA17 z-d0MWm^L_%+VoKRdbyJgIuxIza=l6KR(>EzM0M)9IENaKQLTk@br7jX=#iJE^y`H_T4eU8IEWqeuE`i4SC%YX~m z*7*R%^QdF-Y?b_Q5ARFnx$jTThJnT{<}GX1fT2LpdOl(&Vk3$RJ0GC`*?7R2FTk2o z1Z|b|HgFcA!0#{e#ZG-5*3Yi3i$|$R9Tcs$V{nHx(D&F_kyJonc4K^xqChlZ@ZI6n z4!`u-29Cai;S|r0rTzLlMl<%Q(K5|ro`5Q$(U8JD3#$G#@-`XB*T%TD06`0@3r{tU zo0>zGun=b(6?6(bg~BPJ?5mZ`n$#+_5rrJL{(gJds0(p0`qt;_!!rqzLNt+=xi~h|K zblEyxYwXJ1m-DNlKwK13C;_#r=;O`}`@qBFkiz1J8&ML-`125-oOTjLFc_mg_eZlh z7%u^g3oGLfJ$A$ket+#u+#hIJ_yP{ZvO6XqjrL?RwHD(-`;!A|@HVVD5lme1+6fZj=S#I?nDUP+Z5U__pdbQxy)*quy0Wf|uN&yHX8kf!+mRE?ee>YW z3&hTKpvrGhkq*dL1Y?S{l@#t@uA=rV_$C00<`ftvxpF60j0rY8??8N{>yz=ca#|e0 zqok;}rgPNXr7_~_BHHxf<8uaoJ<_-vPRm|Ybx1sj2yZx>6i;>L)JEaLG1-MypPQl* z)UuAQiybDd5QD;i?u>J3uqLvpA;jf{V}(>lix;Drw%c9Jx00SjNa1fN3$iMXzT&kJ zCK~VN)QlLthq;U=`Sd#Fih~fmf!m4u_jZ1*c(kyqG)iu^)9eOLV#1*0m*?rg`%4UG z7_BnUv<&V|;;=voRu}~KjX#$p*z_+?@%vLk6{Z`+hTTsU>^ zCYSo*U^XYbsIkvWcwK!kd@z5Qy-sX6ZBe!9UIJB!PnXmEA-*F_mLr8kCoTH4=2q9 z9o=8oDo3o|IO)q)>|14BaMsA=n_SoSk1CYp@?_{fIH4T?1#(BC$ol}To+tS)0ujjC z(`W=iREd?rH+=P(4}K`VUSF?-03YS^NdMPw_^7f(B5>%3lv7WDV#d3!Ja`@_-es$phtNfF2)W}L-MplxSE%a#d!n}9Sd(xQ zIu1Fnl)_KFvc$RxnG%83bpvi?yOr1P8|t15wL5>_N$Lj4dpwyHZ0w>6c4_IF&AvCQ zF@uQ{RL zmBH*;!6ZUQ-X^Ru&5h9EIsJlY_|G$fxp>gN!re*n<|w?kS}4U87qzXI<@mOEwn5Kz zYMk_f&2%WW6Zf(hHCy6Y68fvS30D(O4|AVlfPhu15_s{f9Gjxmi%zNna)UFVo%^#u z?`*eA;d@fjh>cz-H${{GKLh7VdnS* zv(|ZzDLJxp#__X`)stm2F^%D;2D}VkKbOXJ%We5o$+$*&l?6+p@Jm0D9GF6@FZ`;( zZ{ANV4Sh}|pZK%{?jIcpz43WJJ2c+yTcD^~D&#~AXK(J0WADE*6Uu+UO8Adh3;&t_ z{|E5&KfcO;{&`r616jY;nbc-hmCr5I*#r(-aITxlk`-W=w=J#R6z&AG`6N7zJ?Lb7 z2DN{xo7vIns^ts++z}2byjPoWS41XSjo$9XTlK=CO`RJ30kqK&kSEk7cwF!Y!@F17 zCZQ%amk^I55&45+h$PCdhnQ(Jy$Y?iwVVB!Wk0pv*rcO=%zkeOYP|=h_5lP=_X&%F zAI)+n#SnrSJ^SXu`%-*%ma=_+xG`291&Px<)0yWN&!Io|UTLYVDY^Rm=qsD94SpI2tz+Ii(A@YK3a0i~8!~Q7F;cAdVKo*`9T5Gu6L33_Kx5LWnZBRT(owZqvw4J$9zl;daP>;Vh?`HNURc@yW| z4yy2oS7OuGQQusBMGl~qsF1(uLbB>&Wi3Jyo-9y^kJV{QD{pIp?n=miiO^+By2;}; z4ts`DF`mYF2t_wS$OTJfiaxu%i+NqU+0F7PglDi*a?Z!ujPDqC(#BaXoU3hZz3th2 zD@#=j%A&SD=t@b9k>=FbG~rR_^PhqSV4q2P`G(~V;=Vxgelr2zwGh7nTE8~WqZ8}o zI}9ySJEJV9wGY`Y4If))pi7-x%l>x3j9>n7x}k)~!AjWcjrl$BMf6Tifp z#2Vmoria}G3Fs||cZe#@u3Z_x;uluI$trV~ICJ?g#shZZvBF=!u7Yj-pDrDkHYU}R zXqj@kz4l^1XAYW|Cx?@@BF2@)Cu5frXN13tq1Q72uPg=lGwdo-wf)1FPiEAR&O7J^ z1PhSeF-+`NO^x!3U@^~^FY7 zM&}Gmz?r~CHbQxX-%~aA+-jL`frfcTFtcA$B|KYU=L66Pd%LfkkahivF%a%qg9~l>U9fj8^k{EfZf^5~{NwjQaiZSq6M1q$8 zn2sC6eWE;AWNyCsc3F;iyZy&I4jEvU9d}T>90ItFk-d#j2nePMu^^J{^?n z8eOvsX0mqru99^QO^G*Z$u>NW&nn~;H+0F-uMlQ63Nvc$8FJ07Eg!zWgpqrN^Cs_}x+Ih85`nZMI7(HoIZPv7Wx>WdPf)oQInbIh>J^O?ju<1mt=jBUwdu zM(Mkh0Qwb@EJ3-z`O*m!{XN*FB)r@)&<}IhvDQEI_V;nT$2wy;YPXBI!WCk2 z7ySS{G+Q)`0}m@cbKICh^lg_>;yq`^+Ev_9na?slv8W4ftN4Ma3QIqw^8KS^Ro*_S zVGjI!;LRi11)vZr`~xQ+<~#v5e4nME1|Rx*Lkv#ZyK3}P;JR;KO8gXfv+fx2LS~hz zl4KyJDszN&m9G(q&~0OTF9!1|PRrY-TwkAd+%VyilyoXS6hTs`M|i^U$9*pc6UzM~ zA7?$8cj0Jj%sK|ZJ%xCl`^+a#8r)LbnmISil?NR3m2P?}$WB9(UU)w%HCr(k$hfo^C+**6 zzoX=v80ESnFKc9>3WPpP@v{(28dT;puaCZhT+;8nRa-qz6U8(b@{5ItYOP-9mz$c> z8mK*ZSlJ@9N`U)T)?k_UMq^vZyNkNDQF}M8Z=l$6|93F!Goq#8Nhoh!17mm>?OWnm zws)>__1n}FzDI)|dGjv;AV18p#Nr^7jn-GIMsm64d!NKKl*`aHG;9HSlvTcOsk`#S zr52V&sk+k(`QEZSn%Z+g;~Ec|K;0}^#4^wQIGhtZSstaf&b={KLKfudoPiXIKC4lD z-ZC$~Bv6obrLfSu#>*pes4;iMc}UN>v0=aO+z!yo{%d96|KUUUhg=u$|A|lB_kZ&m z8%b_JRvjg2T`mq?OX6ZS|Dx&t?E(0gGuwdF=FzO2a(PE|jUDM~!})UTMs+cDYt}J!>xJGkGn&FXsZ}O{a)sTWR+UMjdC`$5xj9 zF;V~cUiufdSD5i2QqMdf4IuoO$t<;U`E@&S?D3u6FL?k{fZ|-5`uJPL_>he`MK^^q z!CO@U|KKJtg#uZ?V$ayr0M1uf3|c|_(y7*ygp@Pw)nc2R?khDlrd?U zh@-Gcb1_r5omu^{0++hk&f;gTBrY}={}^4>|>7Vx!%7bBguajb-LZob`wKA)2NyC z97WQ#V%o>@W7e)n?F{sDXSIvnvJ4-eKr4v(XT_E$mI6r-gQ0T%zGoQrO$1@qS~oLt zCLdGR4IoYKQT%h0`u9(-P5{YeRSV!Kw5)b41iTc8uMhJI{S^+!Pt$m2Q?MQ0^8lm# zKbI=_#((0t|6)o<{QoCiyZ<-ohQAY&5)2A;@!5R}Q}H#4aT~h4GSW8?cM)7Sp51C_ z!Q=C+l_@5BV@Q+S>QOw=W*;k3O}6Cb@l#9I_FQr6@`D2zkc-GC6}K4wsmLv9{LGRU z{|^=mNKK7g>x7Lim=cI&`&<{6KE7lvZudz0D$2xW@qu1pUwzXOKuP$=V0hOl$?|Ut zfseoCV|)Q%V@$ldZfsW9#aI`{v!gd&uy+U8d-uLWx$mK9FHQjP;Gc}YL{@_~<^5*s zS4RLy_phUVAX~8%=$VBz%`J4gZ|bAc)x?Id0+RHLeNs_)Op=d_;G+I#bcX-=qCWq$ zWWLmwx?!vWbQNQ_@M>0pb=B0QSbn>r_Y0?TnnAdr$6md0(5u$ppsxUDH*d3d)`2c| zgzLk}p*BgtTMb=~IRetj7M98b>vjO*2-gpO$fCfp2Hwya0-&1enEO_-eZb)I2e-i} zh67xn!0GP_Lvm-Hm_jb7_LmRH;pY}a0{8t93 z_TTnh+(}mFymEDGlXn*hQ3ay0C61S@`7F^{_N1lNU|*>$-2D*U*6rG9)&9RP{=a{x zi9;mmREw2GlRyvGk6fk$)BHI@M1LX5=3&ZqCA=o|qn_W?-RQ=C1bHk2(yBVE%+5*@tJ1c+Bi7*ff|{y( zYg~yXie_}699`@G)EH(|4s$Kc$kH6P4!nJ}Us!5>uspYjGDf$3QLk%}SOo1>x7)I) zi=3QMpwZ3vNsy3C!!~ti*@Nf8C8hYL5)U}PKX>MEy8nl`n$C0NkR4j7VDi`<&==KvtM zdIUPE9*pu69T2XSy*0tl87{7@%dh`Tj%D$OfXjuff1UdRvdsAHuFo(Fz;afN!4fH6 z`LO-##DvM1@@4<)H9V9xoPP~%X|(V+)8ra+f_t)UE304EafQh|EuUTa6Fs%K<3* zQ^0ybip4a6&H4rug>Af>9}}b{=?EE%X9G%Eey>Mkav@dEKFt@L5bFc@XCA;>yc|+B z!{NPnjp5hv+Dg*o|NXZZ{#395PFnL{lpNvhymimU_6;UqzH|2C0hKUIA=Y7X2SMLv z9~tzU`wY+k9>9vDyAj%FI6$CsmkQeS9ta;(^(M=gp0N#2L=cEr%VllVrd_V9V7RAz zkukcUgkRV8J6dpEWPP@vnst=mVVR(j;K0r5DklAA=8u?Pp=0zee7YD!rE^o>^SvHf z->r3@`oqmaVxNA^;pzz!Vgm7^nJZy_dNRcxyGQqiIQYS?o)s%$B=Fg)JCUzBH(y+N zCP{k>+0`~p8r!6jBb_D?oPDg-X$LZ8m(2Fj4U>3&)k`F3a5ulm z0f=cL)2C0H>u$a?+8|LPfRCGn%XHXdOofeHVE-SRGAM10#dL*!ie8=i+lK46Vl>*Ci(Ks&_ z)#DnAmLbMvI0@?f0ZNS~1F|FL?T#ruZvZG$EBW6poi`T8;a@OD?hQaTc>$xF67{nV zg7ps&y`Jou|MU8T7bA-2T&CAW0Er|14GWO9Z$Chr!*Z@&ZFSTLE`JWE(UOdT@8N~* zgoUO|3@-_a)@A16+v6kQD3#q%VDsk5#P?ai+9Koh;fw9cXPQkbV%ZV#UTg`$DB>*S zYpq?s;nc&o^r7}&Ca)O2dMx18eW&wS#o!S(CsPr?ba)o*tLw|TRRA3#aP2*#hQDtgr2!|QHgo|mKv zT*1t0_29Rk2Jb#g$wwYFoowa8v#e9A`PRU9ojVy`UcPa#4fw9cje}5HXtt3WviS!^ zm6hn8@cC(L+);CI)9jx6i8*7fr6&hGt#o%fc&@q!)2O7{pIn}~axgXQGNa>*MblA_ zVI+zwGoy{l)8JVty3d25823mpwmQtAoN@tPE@tlKnC=SJttS$-l^j;1ms>6kY z69k${uF;o8zRd^r3r#F(GX@{|org?66)t<76e_t4VYqQv5bHNZ-3;48U7=5bI&$>DQye%3d+sc~dXBl( z6b6R28Zpt64<-)XF1FkvY;iDK**Y}FUQ*Pr5Q{%}yTeJsy35N4kO z$+wP2ym60yn_oY7?MqoaK26|}m0rlZ&%qR^{XhUUci7v%BS~GQ9R#!`DR^G80i;Vl z-BZkT`zRAzW5K;|TG@-E01XmTwVR6-qRLyYsjdES&(a)QudIe$EmGS2 z#0FT@!!>Wgx9^CX4d;beI}2t+cqP9U-Tl*60J+Ae-=DlL;7VKdzf22l#c5Q{Telg7wa)?t4 zV9ekps(>x{wbqyq{twVG>p1b*Yc4>4pXR-E>IDKroig8`n^aK1nsHdC{8z<^J6G_o zFoA@b4ihgYCl=S04B4oyx2@ynxb%)aekUS8M1(uLuYPV1e+vzg8I7ZR59J5o+gGNi zP?b@4FeztF^IkUs-Be9}^><`$h$z)nrUEFi+i+o@;{_qJ$M>|9_A3uP?3TFO^z~k8@x5@eYNhLAL4yfW0ia-w)7hH|Q~al^!c*%&A>y{p#f; z=y)t3q+~rxRkq6PND?X}O5<%Y(DuIg%VN^ENj z{09NZMX_6Gb`5X-c?J;w)cwAes`R&6Oxe*}OhJ3Tkb#u;kDnzFTAc^u;ue^L%8~Qw z2oq$;>bUF#dH2{G^jJjKsJ^qq;qw&PN$ec-Q*fir|S&CMm?uItqa%44x~J~a2tSdvER zypp7%b^(&Za+_ZqiYULn)#z|LH_gfIMflBTV&qZ2aWnc7!rcEmap{;scb;&(#qE*K zFc)!``R?n-Bj8hsjcB;ua#rH{wXa3fygk>y;{E{1!7gbW+2mdU=x)8L2U@P1SgGIQXANQbh`E9rEN(}p-0$~ zv@xrhf2s1yKV5mnPE+*aj*8YYpSG5?!= zfQ>ZZ_*zhgoIv~NQJ@Kzet;I;C4L(L`He;ABhFZ4hKTY{rYjxnIDqP=kA3kUd&XLH zUiOVz`V^F}^XfpskaQc%DtEVyN&()WAeCFInSdZ>Jkh8}R6R{aYwVlCGsd z?iOL{{oCek%k7R~%Ow{78P9(;Yk%SW-{#U~0~a9oW6_o6ds$YZeq1(etX(`;aq3tx z1CW9MezzXT-*>k=qaMzgSCVa6%hDfpb^qF_nrvx4m!7F>762dt%E_dZyKWf zO%2#GxpP)DII5iMbanS4c6QP0Zs2(%eS~3ydpGXC;d8md1+S6MJB5Q?qy%bR-tES_-46ah|3{w}sIJ zhrV!ax&JUy$9czPHZ1rgrfHY{ZI0#i9R$T;w13hd-{1FfgD45! z`e8Umt@z5gE&GUuQk)u}GGEU>3qF0Y(nGFalhoNJ=r{;;rkWwFUld+QI3)s(4~ans z)O%;t5DKKe5rVgIcC8d_Bdg|-`Aq5l^j!@f9D*^R7g>_@QwtIQtjuwF=A5L?)Q1J7 zPdq*@vsavs-IqMk%KI<)j6wTbFN2Kdl&WutypW_~xm3he zW0-6?Vq@*-R;JN5TU3nkK0eW;vRZEYhEM8I#AJ1mizRKmO5&E<2Q6W?E8lNbGrqR5 z>RcA&s4-8Rnryct+ctYQ=knky?{`MBB)444LxBQE2N%k>+~FAx8)2_w7JXTT@i0!* z13MbCPr|op+X_%D-gH2w`!bNQ&V?o`#?CNx#BywPx)vr)kXnd?Gr9+c?G2s=AFF5T z+#U|#PV$;Bz&B8&b#b|*Zgcm&M$?j1zNGkFhar#IL$WSUuHuqYiMny$CfbdClRX*OyK0!qHeZ2e_)C`+HuJ1VIiK>)Z2M1TQSV`Vi zUg#_A%q>7FqktnHq~{07+ZY9A^^w8tNI$kbYei+c-tG_J9&ODWQ^HfE8@f#=b%zsJ z8(h?Nqq(D13DF@W1nfAKcS)mQ0TdrW>EYp@uT z#T1uoHK-fmn-ZAVq7Zb*b;I97t+0%mjj)m3>v%+5n0DiuFAwVHiLL!BPf0m|ltk5yH;%nu7gi4dcpf@jCj-JD13A`F2y{T=w!ZWnY zB7S{~4Bv)E!Z9JSFYOsXO_#YzU~je$odKThcpbp{Fz65B&u346#f8Z9i-_OHkPh3V zacsw$m$@nzdB+yJlyO9{moytKL))(XB;T?svOK}Yg+{ibpl6R3X8KxV$J01z?MTVS z!Q-vQrTcB&!c^4K8LmP5J%Mf__jbuIEHBnV$NDfk2xZdv2QeeMs@Nks0D=KbN3G39l#`=sDA$`T`*;!L2wB0GN z37*qCtSa@}%5}MF*kbAs$v&a7=$0?rrLxVnq>PDq$2Bv|HXE&R3UTHN455 zXI2rveUqA?NWB6r71IIKcG+pJ-$%_H0UNHSFI6Gbc2NnF#$N+`4JjSzWD`=YI?Y`N z8hKi#Jugqt=wo3wZexhMKdWht^;Wko{p075e5P=<$Rj8xQ%=!JnO)|IYvYEhlhrhc~AyMYDi6;c#iS z*Dcbac%7+y44c*%=}Gs6IG|VXKeYhdL<2Md~dh7C;KeA*Gt%r%(NH=X}{AUcF-W4 zwN-bsaPWJ7mc_uRH8p=_WlZBT5z|8wDke-UT2rBUA5nL3FT;H3xU20-OX!@(hyIi> zX8(CbKdp*qPHJS_pHrCp8RMLLB=+YeiqegA+B1%Y{+Scd=Ya&`GsJQ}5Ae+`7w+J_FIpMC;H|*Q(F|o8fJY|QW#r5SWyMA3r3(pyw zQ?)B0x<5dAH`|<#HsTy5>HVbbtJ+(h6(g|pI3to(a;7H7ZciopKlXjpayBn>O_FL- z#h%+Ruej%SDvyEGVs}TnT)FLN%Zo}nd)FS2guJOgbCkA5sEqE%S&VfhhjK@bsm16U zTz&mM>nUkguA_y!-buy6O)n3oE6>2#lNi=ljjIby_wpyHwV!_?>6a#(mis-(>VN99 zhPD=BYC;|>g>?aEVmd9^zaWvi4zE6Hk0>uc4q+P%nW2i8pR{5D9Y4*n8f!?BCz_H#V94w+# zH2)^wBb?7x1BOW1d8LuSv_USrm!B2B3Nc0XUgw>VrxKb>`J76hQfoRrz8-%kwso2* zK(4N=G+N3=q>XA8EZ*5^(j-yQhY7EB4g&4N$Shq3J;{p)w{kg|s&&|hCS>ucFtn7K zEvOc7BtGVGQR_@iS=SMSPs829XVmwZEf&~s8FTnAGbZ*q8s?E=tkm89R=^Jn^dq?v z1&0zF=*1>c?e&i#FISlZr9UvulUAc*@GL3NLTzce;&pt2XZ&%tLAAyG^8qUhTwbC-zH&?zveKp2d}w<5y8> z+*%Ry*hhODJXt># zwl2P|&AXZ%C5V^-`)P;i^fJ9|qasd1Wlypl%&DtLrMS=RU3IXHPr(+(NceNBLC~w2Y<)^|IW@E(k~Ro zyYJn2V}iTY>N@l4;FQC$5ndHvyfEcHekjHAF~!KDX>OD_5=M(*)CoA@=^ddX0%*F&5pEJr>G0ir{*N}pFQvx|IcS?uEf(aiA%bWWs z_(;r!tT36^x|7#0SQFx_1t=ycOG(^bU?!RF`~iO)~d zBxCAQ?ECM>C=cWZ_)P3&SQshGZAo6YsAE(THAW1>i?ZaZy!eEy*aDgR zY4eHKS&n{<4Rtut>c<=hy$|Xu(e_1OfxS2eP;#fwW9Re-2q?^$H+m3EIWN}&@_Uq6 z#1!WqQ!S16LjSJtvt6R_Zk5lTJdWu4AD}z~;CoH{id( z&@Ww)E8|tvtjkeHnBYgd+XjBdQ~y3FCwO1 z$yo~A5|5W`I9`Akf9By_X{b^FP>9xa|KR z{05El6R-5r)$@@$t~l~o3?iqn6F{EedF=zwjRd4X82&t0aQ1Ub9%bAv#*73GJD7k^ zE&gcc=;wU$^vdYz2MEj0jC8M2lveX{Elkp-nPWBV#_^7i4odXR(S89*W z0Df+0wZ{8Bd1^^l^}tX)MXvSZe!P;U>ixK=ZX(-_MrzIB?ue3Ti{2NS%nc06Y;Cb~ zN`epFs38(+vzM47>=_g$lfx`sFe%h_%VM*Ry&uPLaS832O~joH%ZU-D=}*e?bh)-t z-qivoWCTt-A1ku)ps=be96P|IqS4HR(u>d(`cEx61DKwl9Yni$^w?Z))tNl1zo9T% zgOZSuqR=v-08Fj*$6(|7c1hcd&avWUrl+C zI#hEfecg^me&xZVt74x7UVqQ~%DNJnXJ6Daj$~9QtqgMt-}*}H=k|qoBxs=KTP~N4 zS-EQvID|YjsBonykBt>jU}0mVBwxbuju%kVv;Jhb+%9=kf1(|)x#E@|!D6^~dO`m= zolJ}`S@J&I`(#I2O@8y)tod%vxSij@6t|g9%a@%S3BpIF^pVBVv;*7Ns`lS2nAQhTL9)6{%D1z8uQCw4R40!dohRHF9E{L=997mNkg7D z)2-BV2WSEV4J-VWQugxJSy!S)@M*WAeGjT>N-AWl-bV}R-u@aF@`78t041JbMV@|z z(PHSG%Y#-WD^*#svI@#5`fL4sYgq4fa=t>s)C|tt=F3)-Q~wz@`wL-{|J$*DTiE3v zPJ%n((zG3FI>X<-Yf=47mb#UxiI;d86NJ%_&VDsDkF6z`(Q5?{1EQ|>Z}47j1&uSv z?*IZgBpTWY9?0qOCIVcY1;O>3X4PnUN&3h?K*0ZFAA}S0J42=Y`wRcJ5!pX(^)Fum zF@Xl6OvJ_Hfep-;ku=~ZZqA{s>f)KI&>t|?G=9nF{S_hp-~ImoNm?{O3m3Gwr{NEf z>Ft?4g@f#Udq5dqDEiZa{Ts zH0P@KRRr6rHe4^HoISH~_Q4(hwc%H1;Ch@J40<3vaNl=Wa20*o*c!hVwScNfm{`TkLiSr~Q= zCB9o8Y?o3y!dG89Da{ZCJA7-}v>5kU&|4ymctEkD(`S)$7lL7=75z5oeQ z1T*lo%}eu2KE9tBy+x9NKj`-uqOdO6&;S6DmP&~O=&uipJ>Wm>0rw1Jrp80ZZySGd zDA^*Q2q79C0*+xLSVxEw7lL`nDQJ_%3kJN6fUI2iQ>R2n^fvN`o-(>u^ znG9*t|7|}0;(NfJb1bxYLm0l(e?X_7>oJs`73l+67gzyGdHk70`;UL?Ppsqb?}gE8 z2<=oDfNm0d0Eo0ai{uTof);Q-dF(NBMrSPzq5u#|swE7k*g(i?wM%bxBMvw*g@+~r zxp@GL3+OMYHzNDjY?--rOpP8nOd^ThPB6OE)Q`JLm&!Mp07%{O=1(yg39VnGNC)m} zuJQvk@NpI6{aW4Kst_f|d3;eS1G5Re(YGWJo9!{vomU5>0L0gr6#8uvva1giqdy0C zjn)BKJxLYN*)j+s<&>h8;XsfWMGlnBAAl1FdrGnR^97dPwBo!$!7c5uzqPmEJZ_xS z%OjMC3n}tc1#(X8fgGUc&;^}3x?Swbxyx0rtM}GGzwvq+u6#F%7fo+)LX}8k~kwB=p&Ylyza6Pc*DXR`Ot`c#Z6KRx$P`U<*X%~-RM^--{vm+v z`Ilc@vq&VeYF8gF>2aQs88Iub(rV+l4vH99jW-Y>hA_|N4YFq3uxsdiF^y*BNc;z&9NJ3k$kPQ9}+=j2@^7G6n4qRHE`7 z1uFByep%8Gngc-)PdRXF$Sno2u>313_uRPVTJ=FdtV?si@ zl=6~kZr9}!5iVgv>(3U+!DyKTTYgemSg37!uj4M%g}a@ zBiWLcwMzg;QZ(WlvAQ;pG5|DCf9q1e(+H?^xDTUmUmJCZqLV|eo~6=_N%q;|>y!bH z2Y5Yu)+I3q;i>3WgzOpN&-Ak*01*GOn;bSwl3EwX&}8 zRV4zCC)Lkw*5CGc&CoeE-*DYXaJXdkjqzBFQ_a1&JClQT`c%DlDQHb=L4v=iINwF- zI!CHrToc2uocjqlX3rJ`)>i=9fn8QxCT3dh%vDS7jXL@vzAW#pA!2f`E3^y^vvBi^P%h+@x^i7OA!_v!WnoO7Y zC~mJEVPXaQ^P9@fdfhseErF|{iErogd&GY9|8CP?Nu%%UkRa9L#NXpKEeL?B7>J>> zW-}-&m}uJ%(Eh^Q4vHWa(2Da+O(U03pi@YD^9+&-7)qVe3#8^Ck-&2k6ELp}8lay! zEeyE8v2r||7CMmzsCe&M6?2+EF@14?p?;SWmJMBAY4$b33cH#yTf+{Jtq{lzqqFa~ zU>fa6sB6)H{MuRZ)87(ZQgSV|PNpLzK0DP&-(}!4CkVl1`|5GE;o3HsoFCDejq75e z=X39sBUsUPTU?jS0CVFt?8$(SNDw&Ycl4ycIPXM4$R9Cf&Gek608%gnlQrAbP;~Ufp(zeeV4X zNc-C2Gt@iPE*ln;aks0auy?^QEWv1ukA$trqsayG&~Ko kh5se`@&Dfg4dCOdfWt&25V9Gi2Vx7ddi~O*7(d4T7v>?i{r~^~ literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-15-Transformer Architecture Diagram.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-15-Transformer Architecture Diagram.jpg new file mode 100644 index 0000000000000000000000000000000000000000..4dd886ba0e9f146b28d84bc5f929acec56a831fe GIT binary patch literal 82154 zcmeFZcU)Avr~b1t58*sG2ZXT1-HSS5Qdc_f9bI@bC!m38)DPsRbU;KM?qzzHYw(q<68fF!M1n zSb#gE7?`9Ow_N}edYsr8f4hMH=fb#yiG_`Wdl&B>KDt6J32+Ak6Y~xhCN?$}7P_`S z`aXa~icR)_Uk>NK)=OL#7jl8O@i}){pHzOQ&>lMm30k}f!n;RFMNLD?_K=<95vP!_ zh^Uyjg#1$lMI~jGXFAVy_4Ex4jV!IKZC=^h*}J;Adw6*^aCn>sqXx_f&2`Ul1*Ca0!nW`EAD!q(O|Hn+BScHt+# zPS4IS5SLfK(}e+G{u^27-~UF~U+5x5({%^U4=mi@>B6|B5E z!z2a3z(v&FyuV1t=GCDbMQx3m&^U?uq^2r8kryyc0$&PLd^*!`3sg9a+yZn(Ql7T} z=vP5zz?vJh!#x{10$D~2m0?vFk0+vzM0}C_aKvuOYrX@Vy0s z=Ce?uN=SAiT;zG{E%5zsgD8uQ1h&Ad5~JX)+9MRcBNVa7bev(!^lR?o7Qn6tBORM0 zI55-yHJ*R*=U?*kul4n>{pVl%;lK37|2O(nq>7Zxa^Mj5em7yR`DRK=vqZ)!&-k4eOCM&QR)^_TYeqjyVtB0)Dx-fP^_h+h=iE zqP)xBW>5M#F3RBn>N&>V$p`^no1M1snFkP1n*@$(5w@RgfD(eFZh>CB8;H^J3OF!_I#Hu3Ye}`9=?ae=W4s?u{5}5Whd5 z-M%D?yoHcy2`-Gb*W3r851z9kxQ@?dG8wg@bIoXLOSD)E2|A{SdUkpg-fViH9Ap7R z{fYo|-KrL;dkY{^;{!OoXhs+xRw51JHXmtpEMt=10xZK}D`7}`5ro=`615C%4HBPt z-S*Btm{f{Jyu+(d8TY2S#(Y!3ZXOZ$q`CKJZK=fT)04ya_V_YAr=VMa_%pQ(ldepW ztH3S5UNRULD1(zGV?s3wosrCio-*5FP9!{y_f*^+@Ga^q)Q*J*wDSU_H|-1@&Fu z{OVD$2h$mP7Mj3GfJ`Ju_MQJ<3Rf!pRh*Rc%9mUR`_o?TZy&Q`@cWq=oN+aZzGV|7 zN5Vr9?Xax-m2uxt_Zl0b(|IE{fz^+#M*T}?-V$~ z9yl$9^aRizAHOVyc5<6zt zK#agNE^!bZgHP%m6BaQx`6k8LZvpy(TOdSrO&Xd-fk5rVx82lc9)z9gGF=S&@{nU} z`j~%n!n9lc?L5UOI1bj-@N#5KByzuI1Q+3?_=N8)FgQtTqy*IG6h*)x9rpUMZ0 zdhu!ut_kk7j#6a{JL2T76YbP{zjE{Yv7A%%_~)%ebgozWS8?$^nglT4P`{;C-u z8BqHRRRDo~JPUW6W+IU^B;h{fJ;*oA)0Iq_p#~ijabOF?ZNKE0&N?TEx&;(f6qx?Jx&mc_|A9^_;rj>(rNlTXw+ObW;<+@ip~xk#+7)3THxe+3Q*MdaD`io&h>|xJ31~-AvE3BaWH`LAwj{kn~7|w;h&D*^Yxu8u> zdF3tK=9khCV)EH%jWu#CH;`0U>&cgU*)M>{KtXOK!0~(e+7z2-m%lSZ8-m=yE^VSp z)|cX-=>HBQriHkaF2$@4(n(U$r6p&{+ZB5UY~h}K`eYvNk|gKoV3-s|t(uofbkuI3Hu$B;Vt zWuM&Lp3Rwbf-@Z6bRsZcP`q~2qhE{M`vT-fa{TiW|3QM{<$J}uWnACJWwO3NjOY)$ z=0-M^&;cNfTY&XfNuf@%OO)})ALXDm$~LZyT8fQ1ikm)oi%(60=uctEh|}sJt*6uVDID)G8PahKiPXikhTmYqHY1(X z-j5*_&MPus24C}Bg!~Y7*o(d>L$w_8U5eqyT?6H)m~5gMr76qln!fVfy5@Qn!SKUX ze&xq51DOg+WqZ?R4Ar5#$UfFon56s3jLXs%KEkF!jTw;$BS=jy#*J`vGnK$^k@j=G>H$R1wQ_gsjmQl*a5`TI-!CH|`?3CMb(GkGb z3SVmW{5ro}8yium{8AOjR+F@p+~^x9li=7}YG_9oXzm^PLfzqaoQMY!B@7LWf2KUu z71R5$FF9x_HbRbZIH@1Y7b~Rzlji#FP_V3I?OkN5Vdu#lEke9f2wzKPapfp+JrVNX{XDT>QI740|{kta?U2eQA zoBE0*+#~~)wdT*)5!P*foIEU(huzSereVuUSrX`I`wf$M4J_g+%UG!3ZB_{5BSnOD zr&4CE9z3<1o~w)GU5|&6dDFa3(e$a4(fT1i3|2QM^U5=wtTFJ@^0X|pHRiEu3<6y% ztY5zw>I$wQmXl=+E`x|*NOh_?%I$^&9LD@L5v z`0w3-&*rBua>aHYO)+&xESqe}(w(>vg6PO{*D@m`$h3r1ka?|He!^2-fd;w6#JQFc z+2dOvTBCCqY)}V(eSYCQn_oB0g=I6VSd^rQWtD)-lli%}CX3d5uZ{WmX{GXunyJ^k zHjVWFiIps_wqPgbQlJF?S}FZbf<5NJ!@Flp`_o;9G9Cx5f){UmrU&fn1nkG{RNsVN ziM<=al?=KcxcuA9`<<&+RMblGUX0AjckWC+FVmpl4|cLI`wqDjJC&xwdW}!~6y7{B z=yqP~i@S)YQGX!ioHvP{?fCpXy{6GtkzB2F$-giKy_&)*3_|bwsyqZBHJZ$ zIN|z??F<);MrdBeChn2rM-wvgH{_1e{UvEmO!9dUBa>tqsjM!oMU#B0I!}?b$MkDM z(eLP-ps`r;i30%`ccziaa z+*Xsk4!n+Y%+{I1)#PE<1I}X7wwQ{4*^3CwsTYSWS2BDK7Y8(OGF(8862;uF<&fQc z?@F>KTA1?YjG2}5s>1>BB34!rD66L!@#~Mac#RP8C6!S7)hXK6kR`{+Q}C{xOhlNk zsg;y-XZ_4%t$%!A!=}9qwAMh&JBJyTs$1+_u=%o>O2Juri*t1I~tf$ssY zqGcX$OWf&&6f7F=8s#l;Kg6=o<8LM~3o*mslOXmY{$;>8LAqUiPEmh!)z{y;6RSpLIjZ<6hm7 zLU|&`JfK`ajTw9Qdo~^yS63!;)p6Zh_Y!-8j-|ynx{8MVD;yW>kuVv0`zl)rXgE^3 z#6+@x^>RrLRGKiok6&zs0)PC{Lb__}XS_04s?=2^bPLSu_iB;~l{Ay^pM3kiVwvtc zdPf6J*CXMP+H_vH;>0D#ELr0jInhnCJ@oPBu{7K)vCw?pzR963$BzAzgDdSIolg-B z#a5;uTmuP8p2;LQjt0*eg>+E!mg%6R_eYPrmvh8MFfZYzGhsFFMx`VRcSsxKU#}>8 zX2HI#kK#%fanvB1vxkWXF6OK!MX6E3F?m)a4ZO=49`3ym=!vc$+Wt2)3PX*R^#tDI z$PK&)ktkn!A@e!IOdsyM`nh`TI)hh@bF+igHi0|KK7K|ryho(iM-%reW_W@h0 zp}bv)Xs6HXIdG`ku*hnjU8#d*$eh5lWL;p5ncii`%nlA&)oWZE4j8xvRGTMqBJrKr zLMCHN)L~w8F}*?7z0O7H z{+wYz8!Tr_=s-3DSZ?Rz#sGCV81CJWmu&$#ZSd}E@vKli5_P0)NO zCVow{CHO}xSatHjR85+i;=(4~N4@7VLxLvkZ-o%z11)D^gEBp z>cawPA$2}c&;lFy+UC| zo)Y6i`;u~Qfth#UDLBn_E`Ga?}$Jlm>!|blYH3pn<{JC6jTQdcuCba4D*OE&S**? zsl25!2F=nFVh+a7{6$xnQluaerx&s7SL^BZzizmcz5~1W?VGNPaL!js z40l=0a4*lxstg`v;sn_yD1UUYU6j+7zAKb`er)ulVqTt}pKnolk~1!wBHekRmOOEc zuh*$>Kuq1}tKD|;5=8Z;hcXUQpa6@pfT+$SUp;BJdl$=k+RHRKBL|8gcK}6GkWku4 z0}Ojn$1{vzIRC}7G+p?Qx4db-uT$R8m{G|}h3(2QxID|SeUkR3Y6@jF-lY@raDCD} zW##A8{DQovW8(ZkRmU3V_M5xsUwyjI&O)kGUVfIz;SwQ$aH!=myn!ShkHiS^fIB@X zQ$*FP6&Po>DU0&KJssPFO6?S77I;%p?^ssE1>sKxlb&=UX(7!o*_>}@?*Z$jQcOHyF>a;mQ>(NNq3_)`Vm@Ae zTDU9L!~b>0e9iE5C#DQ>Wz`NMLU zZ>gdwr}@F6Xg?2rQ#k0JuI3#)uHZrU^Kwn}P})Z15Q&9e+DruUG9e(A1>+cn{P!yN zNP|Qy31s_lcT#66j_Fg?8&}0&e-59uP1c!L*(3wSXDAH;)O!iu48Ak23V?SWQfex# zQ_wCKHuwWdatqkwEM5DRKNgWW^nBsgBfV#{*V(ksM3&g)W7XvP)5NIP&YPm&g(bHQ zW;u74ow1>8AQJf1AHc9=f`A%VC!F8q;}x!-Dq)Y~W^qcfIhIB3U=f@h=^d1!9@u-H zYwcF;NLVMeVT6Pk~mQ#ye#T#@+4?JT2Rx(C3d>L>09k%OSaf$B%@M?if9c^3OBh zGt#X%*>^f!CmTT9iEL5K8%lJca)4VlUK`KP{5GtH$z^Ekh)RAt9*47_+`3S_ULPguhR9}2=Lv^{A z%gZ~aI(`a4cMX?2+Bjo5X4qnrzi}{J9dlr`lh+VxbOn%y4yx2$D|9T&KhogWS=UK; z({=AaNzTL*$ql*%ax}WL5S&i%JQO7vbhHpQf}{{djAFobP%6`TM?7eVMt}HTRT*LS zw4Zy?Z0&iHO#JHJqiJ*~n(I93C&>mTGrI+5KW>67ta$)fYk~BJ`;lvC90Up806J~uK&1sQ53Bj&w#c{oRPTI@09)Gw5CIPdY{ORTVVAi`^~)oZZyn92O9GL5oBP&bl0m)h4@~f z8VBBm%?sSg(K~x4NZsRX28ELEHinejp7Egt9~XDGj!VZX7J;4w!6P&~8}Lc)-xQzCkWI z^!bavwQbp}=*aKE-AfO~VaBG=4%&?FaRR!G2ijf_Zvopo1N~Wt@v@XCke#MI3Yf~x z)P46IuQAd_>|{M|Q|HPOUQ+%pS=3?mQ;Wh^*P^{d|F3#Ij{_a#PQQ(;8D#P_Fg{LR zWW&GJ*2gXpV|jfQ%he2zdht;g6ChsF!~zq5C%w{8 zofqaq`dn+!uz&}XNOwUy65@RZr~O-Cu6#3hpTC~njyXM00@J9C5q{6hh(XCS)K8yf z*`Cg%vg%vOtXY-fr9D5bG7=D`~qcrKn|< zv67mNeIhs z3~iD&WK^TQ5}V4HTkQB6o#2rP6o1gu%+H}>yg>eDH>MdJ4DbO7>PJT4(-K%XxBgFhd znZnDTdKyetSLiaR(qouCfX}Gb!imzG(#_Z^Nq02)Rk0r|8PbHVyCZ)) zjk46I+)ycSUOa2>vWCeWh4=O5JZw$*WAipkc>6o%L>ODHDy&$wv&^I=(G=vR>M%SS zv&D?LD@Jbk_AF9jkhU+Hh8jl?mpOPT!z3l-OpWWw@sF$mGq2>1bBNeeur1MjpK|=3 zwJ~Rj^HQgOH>6X&Ka;_9{9&8vlm?w9ZQnx)y{Qo1F8K=`;MGT?;B7(6TL3zPwB2|C z88iHIBWQz$pTi1bDyxgL^VsM ze<=y6U6_dw6mBa3Du39%{$uZ%f`H0nhL1(gg}jt>o;jwITqcZx)>B>e0P(zJMS_o~K>Vr-oY)DFMkyan!D(46+NA2aOb zjn|Y4b~FjHNI7@X9%aJx@YCIG*t1c(Mhd2fbm>Fp#7DyVRE%uXuNqUnL{fH?kH*LK z-5WxLtoUQ9AzaE#j1BAO9@KIt2$d-_$Nucy(b`>RlH#)VaB!RVy)n{sX_{juxxpvV zSm8f1>5Xg?l!x)%+x3wOpqk3@9rTE>vc`JE8l<3_K5Od^C1_zTiz^KjXG_?3b86P(`ry+=GADxFdNLzL-2w$b#GoFy zf%u}RZt8FD8TIa<$Quh6d>GN2+FjyM@HQp*=rjwd%XA!wutdGJM6Y0UVzFBvjr8rf z1<+ZX994rGa7P@%{CxTJ5rnA_p*1YCnzQ|v`dWYP&HENW@?AS!k=MpeAUV*E>n-4b z*80hnSI`jpi9_1IKHb|q4k?&7i!?Y^h-v+(3i;$B;2rJhx8lyudR(gUKc1~m0gDUo zhY{0zId$Q!-sn-ErtpAx7=KF-dhF|#muhNY>syXwmZ3)P;s)sLY#a)I4(-5Pjl05* zyS_n3>?q8jXowDK9u>l4d5Xq^fLEdO(9wdDfUH#|q3h7F>_695`g32pTWIs8{h)nY zdm5~9oc9Ngx@t7o`r!V#_P?I|FK+)k2@8GIn}xqU#}~w1RAiulr)VtCx537|lWN9%@IQhIH zt7PH~MrUH1s&2CVWU6{`it_E7yqGaqdrEK{X>gfj?TY&Gj@qwFuguyKaq~{5bj(4X=_e$diMw&pm?=%lX=LR5OXulpwu808TDP}L1Tm6DUV5f z(a(g&fGQ4@mCpqhZ7HZMaDO{D^8J6PW?)kKFa}E$*&<(u5k+@x%+x-yE8AIDz^mk_ z9#R2J{;g7vvktGXGaXADaO=WQ;bPVBGqyTWJ^gT$7X3E9!25GSO?MbLrtk z$(dvhjNu`&#>t&9E|D$o?R%4l`Y(49Z-Kp~X(_wa(mVS`yG@Hp1>b2-@;^9YPR8cJ z%jk51jS(&-whpggkQ-z5Nxp>P(;KySRlssrT%gx!;m*jz(l3aFy-o?D_Z4*gEUlC7 z5kH5jI657oA9WBv`KF1Rqf%~1YG|}1Q(#+H|C$UhI?pQqPal z3S$Y)4XdkD<&t6zWi6y#CW(Dc_y`7tnw&bF_%mQ?M!S+80iQ|e*?08)1npoMN+xit zbG-oS#*A9=mkkmXNuTeEDRczgao9nO;#M1@BfCxv-7S^F1d%kLm|!l<(IAP)HTSIW zCY!jmQm}05wT7~Fq`W(KVu&q0prs+@IO6lFV-P<^uybbG)!eS0zZc4voaz<}H(n32 zzz&q6hmRN)z!+o(%%j_=f4x3f9Xmvh(I_et$K<|)X`S<|3h*mdrfDP_!A ze=87bRX-`TNH$8Y0%`~x4#Pc+Lq$+?qi;;j2A2JVf@}C zN;qpZO@n+_Vk?Ths`|Kl=%8qj@cY!e(#LWgyyPVJoius^r5zodRO@T&n$omtRee9> zVSU!BdZH++GrE4SFdxy|nR;-VY-Cnz?deL*mK&+5j;k%L)KJct_S6>nD9?`S!pPwu z&#P2tU}D`fltO%wgd5`<2+h|+DbtsBmTT=8gLV+Y=9?-5PX%i8!~rlLL+cmJx@lo! zwfoN+Cixs6)Q7VDxJ>8Vs^~uxqX=7xqy6bs-BD>@?)r+ML#gS6V*|_h#HE=%7RS4E z*k*|?gZ4xm7HcrX>(Q4JDl4`!6RAQ#g2X-UI~k{w9GrKX%#V4J95(#=khNaD24m01 zJjPG3bUBSd(6q9tiRNa~_0Di?!x@sxr==HGmsVhhBCKkfI4C`+#fJ>Kd=_rPcDU{x zXxP0wLa`Ik&{GgfhY2&~De?H9vtoKBc`z9&hZ-GlTAEp~u z=neJMjJ@r+oql+vVJa?$Y(Kjltfl!IBX?S-%-sngU$nF1Rk3Ky%Js1(g2vmzOUh>y z*AAvZ0@3p+C`-(?&2nnV6iM1@c;D@GFF5W~ptyxE5rl{-JT<8<1Lih$YRf!TLrv)! zNf!hEtwf;VwUygxi4!TrQz9jTo43|qwEHo_2qIWDH;tF;>e3SQ(4f!xu^3IShR>G8 z#D&UkzYbHd`!^|`y^Qt#K(57>n;el8%Nz1aeei1^g)w zk3>0RWyxOm@S|hfS9(kBu$w1LByy`qggY+w5_#XxKc#xNd~tyNz_7poCC=)3#t7~0 zl?Q`g)KxX4{@Sk_JokdxBr<g$jmvi2_6DfN9gAG7laH>QxuE!uVR4ZqP^_hA4n+D0`# zXInleuJcbDS`+)RID6w_qW;O{3nC4s5n<*a;xDy;@alBH9>wR-Uca;AKd!DW%^5#) zw7d$+Bnj@6n1^W2(ivJpL`Xl615uyRx5bY%ZBf$c&<;5#n1c+D%#nPAO@qiU@5+|f`9X!iR%~dagOTyPn!IfN zOsV_%@!U3?FX=OPo~Rg5vSdg&k&V_}(>6(~Yel=60%Iko4DyQg({9_tVRN~D44*b; z6Kj%tv9?lyUEVfkMAKkrl~*eC?l&Pzsx(^sg#}L1ljgF87x4Guz_niMQOr?3s&C>{ znp1KSqE%%)9jStBD9vT|`!NeTaRz9F`w&sxJ_}g{d~PZ&?8QdAqz32|IQu)EubgEx z)4#1&;R2q*`Uc0txl3NIU9>;K{EKZ*Y~LIGs8f_@tBK4mPEsUz9;8$Y_%Ma~>H2?x zP^A(zfBu|NV$@T;A89V|2*!>vRzYlbS9oeB)ek%_^b#i7s`2!=U^B%c8Ag5Nl>bpx z@_quPI5jd4Gt20-3E)gFFH*>}y+}>pJ)p?v6Z}e$K>q4Z*%XgynZ8mBMAD$_@F~y6 z{D5;+y8pO8@iSoc^oY)@ZFfRr8g!h0-?qNw<2B*WR2ayzvnu3!ud{?+Y~>oEE>cZ2 z%#SZ$9_kAr8}yW(L_L3ub6xxmRC!9VMWtm#{lEza;nX!p9GUu*$B96o_jPG}`_O`4 zYle?yt^4QAi*K2d)9%vps%JV?QnpbJ zFliCK(Pu>)B=cXxC`{w77!|u9d?q>69?;eOOcS&yDubUt|jmYmmqI>6IWx zud?fBHIubB#@zv2iK@+2_01Cjot926G=>Gb+NvLR1$^!m7un7gBeBN0^CleE+v-#e zK{W<(=^|p>B8;xFlNjyAh-H(p!sDBFne-~PrUt(emgv7Iam=W|yKjr-q7J&APv!_6mx1QGWsd zal#pT8AzmJ!(KM_HWE0UKVp*0MkHMP1}AJXFlK9Ps%rTq$~M5aWb)!=?}hj9YlOL8 z+l-r7E!TnQwbTOH@wahDC?7)A`LXgALvbiv{p}Z|+H^tneaf*%))tNz?7@WVy848Mb@d|eI zMau*pM7K#hOX$w61%MW6WT;bx#eUH|zY!olEu`>#@c`jk*f??MK3JZ*r|r)nCoRA#>J`I`ys_wB$JpI_3;2T#%i#m z2{JyvGk%}C|NFN=%<=5+_CC@SN{W;1wWm>#UwH3^8u^nKYl7rxj`~VQC=j7ro#*{L zrRkzP8yb=kHvEC{iY9CK-l#u*bZ-$?G=j!Yy9qo?-DOR3Q#ed7{%$a);fF|>U}kN* zi6lMZN^q~J9Q9t>ll-wY!z*o6-#9N{OCoRjGB5Jm7P5SU3Fq zc5S&CqQt7`eJJp`2+@vG@Es|EUhzXR7+IS}h3sgj2j~)-9^v z7R}n3GMwp5xl^BJaBiXumY%-F{}4x)Hf#0vGPThl;4wPMp!7TN$$V0qjCY6j<5Y=S zUw!govf%@-0!~9i`XU-_xpfcWfCep?((z%uJEuWf%d)+W*smuJ%K4s_qw8zgWxJt- z*e`4|K@R3RO*Ih8dgb66Z5ny+$f}m7@o_01*xPYF6?sMO4t9<1#Q*GMtE5D|9_H8A z5gaJ4^?2{H;hG>Zt^W?ZE-rfGf<8_?{&Q#6WAtE;sY_yhJ9+vR9hjs0avQB@^;Cg-pRny&hxLsw0#)67&NF_VV7w}md4t~ zaiN;(h=bI*G0?{$`OtU}D}Fr|F;}LN(Y`HCw3=&Iysf9mXluss_fNEXw$Cht^keWD zh6IO-zxD4KyLWl5p`JX3&i)Qre#XOPG2a#E0`qnYexVJrV>QkaVi<~|Bdw<=cqTq# zsoq%^dh=?}DA7*Bc(I_d!3E@!FL!?U%3W^cWimyM3BeLe(~O&4^5y{Di|U;jQtHnF zjw-x?bX9p+-u~6o31mvJVbm-;f1*v38Mvo0_)ZtoU&DR&Fx6C1Trz$vEN z$z$BD2`f)tHbg7%CqCj#5KJ;5jrEzZbzq>dP=P7u!9hO!RgG-xr2hHjC!HTNH5(S2 z4|^8YXu2I{NYU8{RwUCoIU2^N%=_p8Ykh8olqgJTX3qH>XU0lPgVCC}6~s$^j6>Y# zwN+{h@rf9BlRQwcAgiq*=7_Ev?N)S2r`M!>15ioJbI{&&rn6ya zk_e%#n#pizxf3iZZI~B5F!|Rx1KvYbC125Em()w2BaINA&`-DXiLztYYFB;=M|9Pr6lWtJLftfjqJbIRr5L(rc- zWl6ANDH3|6M;dF|0>rT9(14^qq+&~=0Yc&Hg6VuOVR!3nc*z?o&3Au%9Xvo!}F}dT<#r1$-?HS3H;?Ir8Q=cr8 zNjboQ-EPn4syWa|eA)jVDb~^)^mZWVC#Was<<4?p_-1oObc!KNJ{8ZP%2mwK2cnOR zT{@iG=dWvsd{IbX={zeFMvftD=@s13#24wX7G_YX8pm-MDcj8`xLcA>dO00kf9}j- zSGE#kSQm5x;GJ%@svBT)E8oL$oeLC?p;|o(6OvXCrh};1v_+FLbT4)#HW8nDCQu!J z;YHDS>0Rh%X(o4QUhr+2C@>5~kPq+n(b*v@AVu}fb#!%bivZkMR5$BI7Gf2{W5}?cyfAP8dXZA(K1z1 zs~Y8hQTL#`rFr%>(c=Zo;P_oa^_g>s8;{W;hO^I8E4q=@=_+?~a$`F)wy{$~7M_O% zn=rMNyLG&WO09rF%V>$(!Dcx{ox6BROwk8Pzy5ywda=D_nljuDFkX!zlIb&{r)ooe zJ3p!SJ(GUg0eN~S-Per#W*IJ?4tCTDn=lT|be#B(pn%T>dIUKig7U3MmjVTrJB_4b z3KA`%Za&PHB+@3i~qEsMxE7OzX!aN_#&c5~J5z1(SV;wG)3Ymcx5%%`5fUd8pZ- zT~xzz@mAk&GK_ErM0;y(5}(JuDCI;8@8EW~a^KY?9KqFD>1Ev_ z$@np%KpcBT-3@B_QM|K{6nB1K~49_^u8^wH*`>r)|p_o~R0! zc6N+^*}uNKi1=kHDUnNMb1&PfQVe*mci}M;ZZ-{jG)aH|ekgz<$4JN|8fBeuJCKnU zCYw;taOAO6vc_ZCB7Fa*H*rS%NZ+Z(6s+A=(b8Q1i$ z!0SvR!AM}jT3@>a1C5P25jO^ z*~E^y4I>c}`Ns$?p&JoxzOzpW13%xNh~8{6DfAD!W&&`9!VnaRq4{pLiF&{UHz^Ay zsJbye=xgjcEKu60aoDNDQfF#)n%f<6z@-z($Nr;kc97+jq$$I*O#fm@bwm(o+Cga8 zNp`YEygZhBX`f3;5bF6Rv9>jaHwO+Vl`ii1h`cR=XdAre1J&Fmqui_KR@^)%$0%G~ zzGu8mrC)#K>;A>wvpZnS+fCmWdpny05UFp4&UmClEW`n!|D=@kz0NlcNjp3O5c9k7h)};D;yq_rUjKbJ%`XhE7j5&9{McPqHKhzX?~muC+>b)}fdt@+v0F zL6h8is$wwB6B!EU(!R@Zsm%2IvUisWht5fNMBCV>9c#n}U>))#3)#mp%(D9}Zo?ki zQREO>j^~;b&)oozGGkfviKV(C?*e-X-OZRUa~-_2bXEC6hCOH2AQH)G6+{ufv0Bmh z)m97tJ1pOJBWU-T=_&Ctx&@I2*~D_W)qdcTP8h?TUj6!5<<;i!XAuuYIb<42Isfgn z*xyf${pFJ@2$2bep${|*8680a5$G%$^UGVnl=3SR8IJpvC^8UzHvZ~74h<-vSoQJk zUn!4R;yK>G&Na0*+fsMw*_8Z5p;10Gc=nrqlkM`!oe6ZPU=DnPr7wxL!NTS|+$xgkQSKZcH9W>C5lrF1aRV5p0a%*N0Il)PkvN{%!uJnqW zmLc1uT-cZHyKFxy7-^M6-ER*!KUF(|J^~|B2`|mf^Bi3U8-k<+YZl)bBNnG zc2|tR+NKqo?2|1}xubnIGLxmPy0IcfcG~7?B#O)TQ(d*hKy+p37AuWbru`O@9ZCY- zPFq5uk1h&9J3QgIsEF314RoUJzGCi1ibd22bnPJ$CBdWcJ5;Cr2*r5 zY-bQeVoQDcENTyu;oF%aTGh!GT;J*Ookysa55`Ah?&*V9B@w~5z|*CR6LeN0BJX!b zf&b1ZVSLp}eeGtrNa-8LGPhQrA7e)^1dVu2DJO-DkMD1CE5C6T41$QP>A;11!rdxg zmUco2xLKc06nbRcFauRo;ql>bUj|9MO^??5r`*aJ7b_QhN7 zV!C>G(f%O|sfK>r5)y4i=Q{abqVIH#+0d6)>N>TkNx_#*z4Ub;UlP9GODO#DxMpZy z>N9f$HEN6+LHwh z&CSQPL-pglIZakI_>Bw9^)AUEONeEHzHp%=qEK^+rDf7>(a4-)h00_H4h}%wB}{psW0qbChF~Ng9i#I~~gd(jw+wYHB&n}EDkmFh;BW%YD@Ym-MXNUFGNM^q}BH3xl$ONMshD`-O7@F;zh;usqtQM zD`qDZWeS-Zv~u`k?}t?aPLYaUuAVnk(p->#HY0^!j7%e^`O56-doF8u-k_9^$X2li zn%8k=23{CTz5Em>|BWg5bpf0(B&qcSElxF68sR%CDt>Ie2n*Js`F?gz1CRfN5ZW;A zW*UAouN0+5(c;g0q7sEi!JkUE;(xPQ;^QWZ>@)U>-&gs;X>6+hymB&clDDRubk~Uj zDVT*wYU&$2f5Zx1(Y>@vWNbLwGXnopmi_)D^4G$q#_F|^_zp~6B&x2ESB@9_c4Yy> z8+|<{p*o3`Kkn*boqk!`>n_h*nr}-gw>NjH&7=~&SNZOt;>p4zOs3}f__s#p|LJom zoYrS}=T3h5@qN8(WW2+gW{+HQQ~!EzE7m~Qg(`wqds$=h)IkvqGbS59Kp##^DwWQ}^3g2g!BKrgI< z)O{^J-9bdoD~g158r2R>7zvT}t;!KNf)ZV}5<_v}rO-ditX(OLDROT1TIAEp*ym(Grth)VoK$FvTK6@-`-VgvSE3+SDI?lNh>gR(}h4G~a7<{Psr= zl$DJ9y@2eFnd*x9JzF!uZnH89($@PWM-P zG|;6Q9uOt)$~?~6b9i)r=&lf zac#m6r3T-=ajKs@d4kDmr9VHA8eQoJTGmYS)T-&2Ot}}QN!`3RS1xz}dQqW^$rCGJ ziy3#!6eL6lRh}t#kHTt8s+DZW`_<1vK@_lpB2!^!P2CF3DOy6{ew$;N8!j?7Gc{2X z{?bU2_8}t8N>P5)vfj_4wNUnWBP;AR@9X{bjJ1ZwjDG1ROXe&XQ?Ht=ioDW@I(N=F+oQ^Dk-uuiz@(^TPgfO4EA6l`~{I^7VT0ea&Q7rePh|43^L8rcQ6~X!XyLp@STB>+2GjwmK$qznVJLh8}L`0=Y?*sv<0)l|_fCz{PhzKZE1gX+{ji7+kP!tpdl-_&qy-M$$gx(WsfDreq zGtQ_p-<0qCzI|q&eg5D|F7lS=Sx;N#UiZ>!FI=631_^pe9X3B0-67><3s!jb)Gt6) z_~9ZJo=g-nw33+!J~@GSRBIfs<(=ViSkBa{Nn^f zxLC%X6a$V^pmS5}l%o-?E{=oOck}FF#!o9xE<@o7AFOg*gB!J-7DJ-PXkQ1ijd4A3 zwy$KY1=FtA@XT+hUd(OLr1eh~%E@`cV{JlAPs5sc9rwI(!s3`Hl1~;hH-=^~Wp!bj zp6y=lw!T+lP~BnGyd5sxc<|{mBRk^o-F)H|l#T{sCdJ*WwJj<7X0y$1Zi@<$EZ=SU zo3T&;YB_+mp)AisMfXbH!&u$+uqr=aKCvL{CkJY77o{IlP4krU9n&ovIZ(FI7|naJ zcPQh<{PvXu<>gP$3#W$x{3FrR7>%0xL7^F)J08n3E)*eIEbpqzE{f|+KfuGWNnUqf zd3^vToX~4`u}ZdF=JxRD_6>GB=deh%(FVCUji_;Fi(2RwNkh~g`3ZT%xde$YM=Rmc2UxdY97gJV{>@kQx4|zR z)kHa77-#i(=H6pvK5pAt>)>3;F`Qf;Eu47PKVC_D7#!Swkq-gS8|FWv0aaXw@$Mc-4X z-14+Ll*M3F&@BIBr<0YDz?&JUCRXO8sy!D9aUl@MFE@AIkgP3x=NuFMX4=|~N}`f! zh4te;ngb_hR6-~jrP@A=pD0rhtm+Z>!76=MKw=zSWIJ0EmR4;*1L=s!`Yi zWI!}&Nt+gxp{pp?llo;yn&m0Q`@ErQrbVM63K#W)V2#1;{o{#Ix(sb9W2PGfH|)aQ zmS2i0)9F+VdJ@ClmG=wr+*eGUiX8RUBqPbk^T=O&0wSo9^_*oWEV3c8=pmcFPq=pD zSn0##z0;m)M+xVAnIS<`NjzJhvQPZy5+A0}6sE#Q4?cawy6Lm_z)b8~c|ghg{V9ca zFa;MvPTZm|kN2^h*h{hO6DjK#wEzQglg}Jrhr)F!jezqJO}fT5(TPv6hl%7FRjT7- zDP8TrV?u6d?|(~A!QNf@HB~YB5O8Zph649L?Wo^3e)|Bg-}~>P9yI*^K1f2E_~!)4 z{F@a+Pe-M0xCSU!>h;gKIx*XA_R6_GC&yk$z!HPLkx)3VU7~{b+KRCydhJ@{H5cfq zN%CHfGM8wsHv>x6;8Y@pQd_=Q@FuYrTmN-3z_)8=TR$ZBf*sCc%*VVcco0@;V0q?VKivN`JFdW=E7r_O-76=!>*(n zJ3h#5knEs2m*|Wk8Qt&J;qZJ8KeJfgzB}Oft_HW+*z3Ry@!2j?HPxd7i&Pp%(mc?H zz`i7tz>?)`?*WqLlUEA%Prv4goK$tHvhK81*N4tctynI{s0VKkMcRjOB}9+yU$n@k)P`*Bx` zOqJI;4A%PJ*<*G+64-uti?fktn>g3D2&LL4U@ov~+FG`2KIk8sms(^}v)iXJS8pyN zGA(b0BMXC~qxgs+W3gIKGz4>n=M)C51gGSJ)m-iQvyy8{eiDotIYWt40IT2ZOEC20szPj zs(oDB^>-nIKVi2;h(q)_0qe&Ib_VlTE{_2o6aIAf7EWb7Zl6?&CbXFWf+1B6h?U!` zTN?y1d%Mknx)hH@2vK9VkQ+(`l_(dxjA@;KniAL1$%pMn9*ffBH0?}Av1{J6X0+a~ z=9M{LFx`l&er%lwcKyO=66hj)Nu*bd;Ih|wrxG=}tA0NLhyI4w$pIq*gbZ)016>C) zA#PLjYk?P_$7Dj_lV$M!RSn|6OK-h_B;Af$8_bq7MLSv&xY~X0Y;u!psrm}A{9b;i z8T!p(^xyg4vBPicR_%X+cTtNBsiRRhfF>PtcN&f!JGXs)k3~d;7aQ~@Z?a}*;o0>> zqUQ=UnlicbYRFL*Z`-oz|()M`6>9{EfLk=AX1WJ_}ua4RPxeFA;oKqSik z2}^xjiWGJ2`tt}!69N`iP?_o&XlE=|hN272iD3BR(rsvJlT{hJ1|v;5U;Wxyy%FmK z;8rYH)t3S=l9>!*Guyd4&TfIYOT`U?>_=~J&``*q!qfe)^X4%N%=S?fzu)Gy+FXpJ ze(yvuZv=fY#^Yvr6II?Ext;(H(5muvPMDnRQwkC1cxZ4v%sEqEz%^%N?D=pV?Yh`& z?d?F=7Vq8dbLP&`T|@NIbWc9ior3P{09-ZMsP@|ff*w|yyT{%=F6EoE4?ob_H#0H& zM77%Oxt9M(KNnlv|2`?6jqgYbju&8>4XOIsV&F84jg;yw8ah7RadN%Yo* z`*ke_*JDU#KlI^3xP(8^~4;6UC|LZkb_Wp*uAxnF$rsW{qSB<_R@ zpAp*6QND1bvtn-!C9F-kCHtsM2YgLKQMuWc8bfb1m8xO|t3?cq!shL|ei1}l^ zVQpNg1@Np?$ljPd)NKZPIHrj(f_v8QqgA=R)CQ?5+EP{y2?_9&6AWLYZH-?n3@Nd@ zi;uk#R&kUs79c(3r0P3MTjH^y%nNt0VioAV!YQ#JuGWq;O(^xokIQ9g7bEcILd?F6 z8N=P$bW&zB(I6~;XVyz}@t}hod;Ldr@{igG7fN|NbVd#~a`Brwtw@TR=GI-z^hNYu zGK=(+)FSNdrRhy$H!Xuv?w$CkIMo<@uM6c4*Vny~Pv|^t_BrIG5{9t=vqZrY$ghxv zd(#RUpf%Kob39d3w1q}LBWdwnv2i?BkxuMb*ws-{+O&to7lIeOnDrDMDDg-salhjc ztY$XHk(U+~nF$@=8LhO;*eZp-7O-#$PaXN<(9y4}%eKaHVIzcc$+aPkCB=%k-IiFG zm}X+;Zp|BO$9H$<#*2hu2NcOL8_i zTGuwq6R8dYw`>hd2A`HzJzra)VHoxZ4?uLJ+)N!khyxR^j)~=-7U}5ZEl&!xtYAxT zD>7lPTfTqkDLo3;=_^_D2KmcWi7WENu0JJ4?~A{a^&on?6Dap=P|%~Y{e&>D)LaM# zc6vq}QeN;Zty&m-m1QUh3viql)&O`*S?>=3)w~U6O5^FP)*~%}DIL_FY^*C}RF3gI zP1O4N2@370mKYaC5aN}tg~#(@(@61hTv@S;i#$^yS8BrEr=x7{7T1gjTAnfYTrY_i zpdrN%yun^KT`?`aVv2lTY4plxxR7jmBY_>^rj$r}7N9c);f$~sKPMZZ)g1)5b>CP9CRU*+hE+&0JoRJ*BLmSX@)^ z+)@6CtFYmWi9=MO%(qDknTktND$dum+4GEE9BDnR5ojoW@dC1c+ zYb|=xuRiRh|FY0KoYlf}*fccFqXvG2z1i6^hORVFKdaQTi1)U8g&d(fr>T8tqH6F3 zfEEVGmK*zMaB}ezK`@VtUVmw|B&d~n6pp{Ln^UTD_lw2ud?N27)~#pQOs`&rXqoC7 zw4^<8^x{*6#+}w`4_INHHb!eW8?K(?24ZcY&TsvHY%t$GDf&QucWn;cUyi-GxZj_(h-6ce?>lw*ndJ&Ie zRRaX|q~NR%veR|#=ra>PI3(7{+p2n(4yj2q@`HSG54g|7rW6p8qkg1%a*_vI==CKv zI+vI$d-NZ}#h%*f=TiEmHG5_1r)KOOT$8JQ_T|p!M_D0cK8-9dx#fefmgZ4TFeNL> z4TWGuwe%SgH$(Fj!wN?(CDIf~agZ_I3Gz`WxSpRATXK!xATAE;l%1w5@|C6JSg*&_~#RZhmDyOXb62P zEto+{_~PSE)=1$sw(zqRXNh8v_Hfcmrkdii;TLJeH9qWR@s}0JUvYK0ztVlpNQx8= zr7r3B)G1eyd1H0Jebv4{tgGn4Lm(8rLcO6^N{+oKbrpg63}awBZPA(x@K?LyAA1LZ z;N4s?MenbO_j;$53hrQSXnP{(%wy`v%Hphz5~i9nLVd*rIuqoyQGF@RS)Rgm=4tV` z$tmlF1=v%zy0_><_(cxdQ=MZ~;u)bS+U5n_f`!^G1<79;KW2pxgkMm>6@Rn*455cf z^|En-*E*G&ZKaSl8&MQZP2n-n#9Y@6OJ>IztmfiKx$(o0XJNTe(L8V2BLLSzH}aj0w^L%H4_X(ZoxKf#trGP^kGrARwsB~;J3K8$BV)hJ0`0TZBm6r3v|2Md*y zP|8FtUw@p)BW%{0I9rt2FxQtqZ#lx!g$jRb_3CzEm5D;qX6Gb-qN1Lqfq4pg_|)gM-l3WfR$RaIWjzMPQ6 zF>HvvGn}Zsmhs#AT|onZfWJfuxIg7V+CjT)7LbE4kk#8j^q!34ul5`_^sjraMz=Fl zQV{1qV*mJee*2$x(fkQBH0jrG>;2S3>Swb+q- z26c$O<})pgA%zp&yV-PBeKD0IE=E(Lc}1VAOUvs@Vh7jBV`q&yS*-jx4eKpXb+#= zYaCvQ0bi-rMTEFxMxNCHu`eU%9t)YCE8oMIh;BMl@aYZEQ9olc^jUFD`S5jEb;$>| z@((!e%=G^1{Xtf7w{JvBRAVRV7CngYh~L5Kqqa22jJu^@AWzy_n5~ z@mDcNr>7*SS@kmP zymLpWtt+E%{CK>Gv*^x0Zrx{OZ@QNuoDZ+r-03>Mp2aiK0lIT?4f7{KTdUQ$DhyC| zAbVR4-gm+!Hnzm~y_CVvLaBWgq-Atyx9J8(b3rK~18Sr7oDkFj2$3c61X$t<+Gt>< zFN|!Et2s0FO8(8ma`H(%uOnY*6E;H`ENLq*fcKe;NZ5O8=DWOiX4lX{TP13?hibpc zAj8rd3#l5guhjDd!)Z}b5}0~U^Bu*T)yh3pb)Wu&HT_mQJHphG@>`@S5ALKcM)XoV zvR4#ZFg=xBfCPE>Seuj}OyYQ-$85*NFeYx144r;j+?7uIKo-JWCdm%ddQ%D__mnCa zE>(@)j0sGgg5I*cLI?VMbfNlQzM+w2+G81`=jUWB&z%oqrM{e8`Z(p;|GgU956$i0 z`TR*`m>a`I3MmdKGQ=yQ*FPhfOm&hW9r%EMF&_Zx?2~9^gJmL~BMxKN?P*6@0NppJ zKSzJ*XTKczffZM~?%=lIjQk8Q4=VSk}w0OE}wrs_q(_PPQw}A6rt5e0Cn+nJ9*MgHSw-uO`NtMX5 z+vWv|KKkQ*#v6>+7lp;!JVfciW#;QC?inb)UAaH3^ZK$XRzmkW`1L9MSfTI5^YpJ~ z{T*u8VwhZ zm+!u17jo^3fZ#)p{pl1_Mo>zH!0EACq!l#-~(rynF5McqQ?O`7}?x!T>$>gkyLEAuNi3G0RS^>?P zVYHYt1=9J$1A5QoN*W0w$^1J5FC@g1r`EKeS2dbjSF0(guPbgVF2-iVZ1C6WuJk`0 zbgU?A1@T=y9jtdTnob<@>DL?~Shgy62A0dYU;My-bM`v4+h6y_A6BA&uAwko1-O=P zKUWQ?-{k{vG1Yp0_8{Wn{D7@S%(7+v|B-gizq49?T&tTHmP62;|6 z>))74jv$&~w$YG;u3Gj&U?0%}VwSYM;EfarTF!%lT$dQAlHl)v7(2Tllz4g!`ie!O z003$H>sR>!qLR4|z)KZ$$i-2O)Bf#VApimyYKp-Ck%1PquUG^tu>2DOOYqSy7{go} z6vW$roI{y{2_gMPz&TPM0UWp#>W@&DU_2~})x&wExzDY(4Bb+sj3G|)st6CQvaxXQ zes5M<(^pRim_1dql=gfkIXaAv%Nn!FX~%o``i;|Rz3^$h~976r=Qx2 z0ZJrbXM%J5-gW&}Iqt#)Aei<`&!E4(h+^}RnloM)z$7V(GOa88i=6iPE~0MABUIkL z5Ok%HEn72ryB(={VB~X5fQ=fHk)!&%c0Hh@`G1mXG)=kVAoL|ko-s%S^a1=iukX?D zmY~NaKAYzxEw7M9% z5V?$%=bL3d3NYCg-)R%ccD(rC303@8_x$e(Vf-|g|79io;T7sD?hglmuywG67%xf6 zonH4K_IYOJ>vSD;B^02&*L*gHi z)_8qdH^Z617xid&*ah^T8L;;1v+A;VoK`#XAa`=EHsjxQD=g^by-sl6zmEryIuu-!I*0 z020?(F`J1ih~(3As@t6y$YoKMX@W)S_1yjv`EROl{zy)qT1X! zcT@06*Sm%(Isq zp4>xLX(+=_9jo|4m8=VAX?g8`ZdYag>#tb5N)q7j^vyT5Or`1bs@>98wN8X`qUF`I z^nx{>Gr1Z2+IzWct{lpxqxV4XZ09Jf*y`#KUb(fy(UD5ougO}E7P^G*ozRdG^M}v;p!#Y*2yq&yu5`UE%AW( zBgbp|4k1A`3ePE8c5M3_C9dLBcD6lZKgf$xj-2d_k`wz_9@BrD)xmasNPtw@qyZ-% z2y0?R{f`fTx*qR{`R$M|9&C42RN&L0AAr&C7@Drj!&5+5`Q~FTdaVKlR9LC@LEug|BV_ zAFM;(=lo4WxIwMn?8x>&G*5bY} z6$vk_T0(yp+26?YPtvZm^t!EyO7=*nhV81~3wjE_x=dHgkXT-O@K~<7!+JEaa_}ABc zf8751zx=q3d#GevrzuJHRz~U?1iHL$X;PWB$t=aY>oK;kE5k|HyVX*~^ejLt=Y09! z(K-Kxd;U`$!#_Kc^EprgcxdKZJJ1gYI==0dqV%x`?z@Svz`N}Fknx8L$3ZKto;V6g zYH}1SloW&`BekJ9Qh;G}wg=iY=- zF{-e!z4VdBU7Xj@yvRaB{+3vM-L~T~pv1@|SmyXtHrys`!VjoN$M)TIy7u z*WBe~|6*@Tj@>DW7qmAcd;%0wX+6<6b?x8aF(McepBitax4omayr*|}l9-x)S4@MD zzC8_TYw;y{h%tS{uvVhwi4san8&M|&-3%2QZQRm4%Z)Zx>^#3c(-gC^e3%n|Z*mgI z%Nw(fdK~xIf9FAAkVk}Bcq8{EEb+^U@8{<~y4#yQf;X)ctkKTKLWTN#{3Oc|SJ zYdaNG?f=482HbS-=rBwFs)v_JcUx6iEWX*8d@qSW<|R9>4xFMBgmjPj=DSTQNmdjP zS@^6MR&G_w>Kg3e+lBp<=!Pmwx-L=;^aevhnY3_f%@7y#?&8vLSfz6H%SC&6Zm>>a zX+XR8rLwSDAzzBh-G$3HDng8KCSzoL`A{loYmBzFO);m)B9sHE-c4t4GELMqxIK&M z<*rtYv~+tbVSlybsYg(3C7-{9r|3Y#DhuR{FgrJAF=z7I7D=#|AJPD6Vo#TnG9Jn8 zN|;Aa2YuToLgsck9s}?-&AH>)1B-pzdTe6bq#*830FdEt%RBx7|NGCTNB7-598xW_ zjGM}^`lNdy+WZ_xXG5oiQ)c6NWBONr3dXzE#W>`md25FEJ@N@HV)LRT7h^QFVqfT+ zPoj=>omm)8S29{X&KYPcdhkT*NsrlaxG_D!c2w_ww~n1nanm=p7^kZqW(hhWmCysQ zQ9GC}C`ud1k%Ce|=BvMA73FIr6`qvx&m;gIGJIfU`Xk^wXe!7dz-&7T-e+|MuECu< zd*_}@?xp=+FF_U5dJ*0iGp|aKwh&dW+@`t$C@X$Bkkz|L7Yu&=Dc6BE=h?Rle}7yk z@`2bp8ywftw+NY`BH|7GTsgjMb^PMa_7>Eb$s9H+=LL*EjW}klj~sB-{no<7kxajd zMIcSR7(NZ?L>ECpMg!JW{`8c_H*E`y)M_m(2%kCXb>sthXn<;)`ZTVxZoBwF5W_NB92c!5p@!**F zAj^PgF?Ipg8?wE*vvgy+ogI6exF1^!Vn%$-#s?G>2>*(uosMY7D7J+Hsfcqx31Xo~ zF3f4|3a3j4=_%-MS#uF7a_b}dCh|aa?#* zp|No`u%`Kf4awJ19e1Ydng9e^2U)5-Jh=5FYs$FN>EeAB6J-+|`Zbg`p_5ciG#iJ) z75u;{&otRm_E3s4+>Hm4yL5>gHX! zvH@8AD0PW?H-G5mjBZ77l5U3(qDqQt8zjw(t%K(yaVixRg+A7#$^u?Um7FQ*m3LMx zIElK$*N8F(x^nl8fA5j(JfWcsJDfcTi$2nZ4sBk>C{_d29NIci3cX;)H`isbIaM#9 zUkcs$jBW09czGj}6Cl4?Lh! z7luc#a-t&OYBX*1bM&Bg!R|hPt&`A{=$JZjBtuV0GO3(Ur=#7fjEGS0S2n?%8 zxaxUN@ROJ@?lgOv*G6`>*!0BJJ9@_6Y)Ij3+~xGo=2Z@BnneZxW<|% z9m&elpIRqdrgIg9kBfI~iqaEA1tr+f7)-2H?an-zidR>T6-n|Ja8@zKUy4X@W5qoB zQxAdJ-BdpcS^fO2pd=0*+U&PM8#x72^<>-%h@2duRhFMs3ij*#axb`p@w9i=l5MB0 z3KMfJwKv@+Sq8VY>$W4(wry9)q=6l!$Z+>u*kflR_u9^H)XG0iH^DIkWiBlhNT_|E*#I@g&t0=DEd>A? zBH)eKuUK5M*%R%BiDjA7uwPfh@Ap(E0TaA&E6YVh^;hA8ep^)AtH~4X1s)dheIf}9 zEs2B&R&0j`q;r<;LR;`BGcw*THP27P_(d1+!oc*%EscdaO2{HkGN3k@OaY;qe8W8V zh&|7ce$_ELQb9y$uZ6z6w9a!ou%?CXgHH5eAY z@$#mhN<%vNAt&g7lTk?XV|1VY$*l}vOWrrqkhN!X;=HibDe1)}=SRMhnE~;jqKBu1 zd49Zgq4cJU2%!Q-g}uq(V0Es`Vi+Z+fW9ig=T!fZWYwOV511-Zfrywh_glNM`KbKn zNr<0>jFU?;TBaP)z0PZz(BwM!oY%c7Q&jg%DgcLqB<6ZvyNGC0$NoL;4oK+^c(v5r z0L24Eb7015fbDoB6*6B2M&x>xs6ev%swCZmc0pSneJ4EHS{QubWWNml`JTf8;_nwF zSA=w1UAKJM>nHEk>c)3iyK+XE%MIvhfI{L2nmkI}*DlFNQ?_HsJ>Y(pOshRO;z)4$%oWg{3$z81MQ)Xg$At=qA2t-xLp_jdR= z>q5ki?#i#CdN5j}lxAkzk z?K>!;kYz!uoI5ku+9TEme?)LNXg(1A=7cP$`YhiKIDfMNTB*ht5ST)Vf8!>X{*i)4 zlB@9E{rb`6$X%wGORQpcZz4yL)i*yb61m*6kG`WG8*DV_2I~ zn*!k;!w203)<}!;Uwin=WT;KKkTE>Fx|QR3li!9k@6&*rShp`F8*9>gU;0HNhuDUC z#7=0w1nVsITnfHppdnS7*4%Z=&`)V|Z^NpE~8CTlRn~is6 z;4wC(elH^nUJxs9XJHP}P2Tqa1vOthVqNDo(#Ul9U2@A2Dg$!5>)+2O4D7CGm zfAtzi5anG=^MIB_S?LR`gVT{sp@ljk#CAN~hD680Jss0&kVnNCavAZ2a zDnz0Ui$tpT4!PRwmt7PzI7H;X0GsG5>7QL)f8h83J=pqxXfIPx48tb}Eqzm8-|dy& z@rlKX4r)WBzcE5;`M)U5u$r9}Prc6Ud_IplEO|{w@TXAv(?s$+5CNM`_N@eO zSUFwUR(md%&5v-`FZ-}M3fQ+;7XVOro(+()Oq=K358AzNfT_QVDUCK0;j>f`Hxf^h}WcZOf3ioyN{OBn#H$pC-Is@@<3aT$zZD>KU%|8OM8U;oc8 zOa5?#Lx&+DCIB?;>J&ihC5L?DQ+nfSit(Fa|B8hLGzM;I5+s^vE6QMVACcf6&2!6N zuv2QnjUj2XD0s=V_fp4Ns|BzauEG*es2mf~=b!*kD+lrhyzU9unugT|%a(yi!ql{U zL%5{L(EYk$V_(Ix^>VG*+d2Zo)t56nFDo3$tGbVJmXzWQ{gVb)pbh|j+tehK?+L6Y zEITE~<^EPU5NP4oS?{j!rU-P8Z5XP#Wh7KG2F=%1_42SftS>%J3GF2D?!pFfY?S#? zO%nr1*VnSm;Ct~4fdeqxV|RP*O=%1`g2*RRBiAVTo<=D= z6B4R4Gv33RP2S;(KpAUgxSBQ}R#Etvr6gEpe=3^MRpl1HhuY|DI3c+FP=)IV$Zh;h zh=7*)>uYtbV*G)VfD~2T3KS>AczwB{+g8xLQk7!dl2UM4K~i+@z|d76B^ny&%v03dV@v#vY2gBXJX$EZb3Ev0eg4JyK{z42=)_ zfL8mwMs~l-a#h*w(1BZ|EXH~6ZREXe`i05VVNZVC_)8~7->*Ywg)5MYrH;6r& z$Xm@(plz{7?HMm-tH16!j`=HI$3frf@+zhLWdXPOhmq!bH-!HAl+HO$wEaV8v_w9fHz3W`#D)kJEL&JU zRC-a!%#g>DIz*9f#lAon=cCc;-F%aLP82Xe%*%k@WteSq7F(-bOMlavHu4An29|$I zmi}pz`NO{dza2x#ip@v+?9^|}s-B%6sq)_@E`8ExQ*9GR$F^xoL*0XASC0pTXWy+t z4tl<;qpBb}{7{H$F+k^`Xm$Vkg8X3e+dlh=)6HiqPxnP=cJ2>I;4Tmon;o8L#w~Oh zm6V^C4^3jZa)~`NN>W+LZr+A8t<)<2iE9?|R)mEWZ)%(hqi$zS!Brg36vvHTfEfcZ zQ+k-r4piJo*WHK~Mou;JCNiMEu>0P=G{0i|P%(-*3Dfej?D7wZD6qErMvLkg z&7HNOp*3>wc*z9iYw2bsyWTL9#t%c|1>dT5w2lfnjn$avPFR%bbJq+xR#J{mPTdK+ z?NhdtAb^Wsk>aqXt-H+Rlf58RyFZA6lIrS^Tm;eic0dr6E5&A*e!ZGce8 zg)DHOm6BaU1*_i`pI%icpu)<7x9l!Vm0MtKndRBY4oavSWw=$f2f>H8oJkhh-bb&v z7*&X(3-zvj=T~sE*#fpRR$viq?1DA`{|jOA9AZwsT@47OthVJ6fAcE^0|1Og@MtEQ zA7JUQONNZy0-PYB8Vntog#!VRycZW{)59KM+6VnzPe{2c8s5JEYC$_l_EG`+(_c)H z5^wSiG3US(3w?$Wo2Ny?n@*Qr7wFz)VI^zkb%&-OP{d~drGr~gN|;)@V`VK*!`xLV zcI35@ym-Vu;QMT>iHjTHDAzDVV>fg!q=N(K$F8}y-5YckPWRwu1R8jWG7vy=#)EpQsV0YbBCv!7jX1d7gGRs|ebq&#GpS7HzhE<0*03}P$`1c5{$2zv+s zYu z%^0K1F)PRHQy%W7vKTR&LjqET>thbGDB|OE&5H6LcG*3w7AmDTxmS z6ge2M?N19Nu_^lV-w3kGX^aXYvToo_Mxia+>wjvc!r>k8zHH1_EIf8Jr&SYO{HKKDvJGliLgO42fG3WMe{0|J5`i_Yi~has#6{FpJXX(YMkN;*rqr zA#G-@f1$6Y$YkU!7Bv|veAILzSv)mvq3y{pk$;nKilS)KzCBt}F`pCD5rBDP4`^sW z0KpIRQ~OglGrwYqZ1J`i0!o-%5~SZ0@Ip2Kx*Xv2WF7!&p{4^jJ%oh1uKXlLj$apKh4Tt=9J_+z7-jBXG!SKGkV~+^FdGQEs@~sdm|o96_@(Y%O{dChanP-VF&On@$Vfm&OD^M5yQ~ z5u;a)xq4q_`oP1G zMY!iRAjA}kPRoV7dJ1VG;Wk?KIG%vo+lvZC{Jeo{f5o~0=n(Rd6U;Z;CjaIQ4Cu>d z)3!Syxtc!z^bc4W#k)LL-#%&o3zbTNX%$i3zO$1otVLxv{nWO4EXEiXf@f|IZrt7V z+t;>KWS`>A;i*|mI`)1brO9<^UAjk4Wba-}=LA7Lav!gGvb85QES2c9A*jzv^79_c z^_1))U-@3K<)Rf-up2|Zzp)_wTP4?}n>sO?U^m~}9jJR8CT?KADATQTtT%lm>y<~7Q?m1Xl3|SZ_4sZ&N6K9a; zx!oionR6KMcE_o%m3n`I!pftqPc`C2JciN?pGwFXDD~ z%#-yWeU_&jI_c-#Idc(SLIo!)u!iv*a0;SQc2kCvOH9fDtODOEF%1rdjT$m9E`2?5 zBGAUm!`<-7Af1TB;OfMpLx|n*yuG9)v~<|IWDcuy{@4?oshD$}^hwgSZ0hGT;jDX7 z+p|FWTq!J+a!(2Aqw~htI6$_n>{PzFF#hp2b`{yyoy|_IqXp!C$%Fcd2kT$?zeM@rOF%Jv27m#-g&!dVkj0N?bAPrS5+^U1 z#lR~4Qf0XHS~TzpTL>0?9(mJssKe?)@YKI6nE=DQs!^ecq@6~(h9<5V$=wBKh_-!$ zVt4GAk!08f)XJehA`ray&DmX?+MVZl0|+$UsYLWuKE|qTIfF{YA+{O?*Z=-N(I1NP9U&`R@9ZLm2hgqTfFl=3fxQj5QZ@ke0rgj)92cZ1!{CI$ z6PR?bK$x})uLQa;b5r+U{XcmK)YN~E*h+vH^`r=R6BIs&rUC()8r0P=kC0Unur zAfgVO>EpmbeEBnE)d>JM5I|08C4p8*mUMt&CV+-}8ypNl=t6+d=g-eesYY2n2co(9 z639o)=9<(;>yM{blAtk}*+9N56(+y{z+C|9aAERi5M^0G{Lm_$>5yVZi=> zVf=Mwd>2M%wUq1y+E(9uJDe1~`YfVw8l>O&Kgq+;h#LwR+@II5Dej@$fjsrk8=3ngwsDTur}EZBb*_@;=9^?(68XjB;_@+;C0@tp`izUkg=uVkyf7}Xs}COY}abd2JrUD-?9Ko@wfON%#iVd3bf zp%=_{jMKY^rZ&WU^ zGq_TWt!d*Onq=qBd8L2B41IV$v9!=N>0|eCZ%8s5R{~ooAu5?IKF5tVFY@`W4Po74 zgz>W{jFK~=45ucsZz1{8H=RJFsDNhkd~>a%g4vm=wa|>mb37Fx;Y5RZua-JQ))bwx zT?a>pWIr5mlQ`t+wbRNR>ZWEmSwZAmtwL|#GC z3SMX>S9|hT^{+m}V3*fy?vS<&0Iy|_WOqsNlbrb^v3a|UZu#ar993y$k0twKX#Veo zh!QElwO*S5#P#Agz`bwmD+uhb*b zmU+fA%gU^(gwGeEQuWqvDVNQTQpL(*x_ni4)8bzbdL*djqE-|J&@NPddDl~wGs45$LyL*p|D(h<^Bg)BcKKdnbovVX;*(vO!@C=Pfk(T<@Q3$0&G zp8*^M6M-hePA)N zJPxif0^*_`BF;EK%F-{71<-WAymcE0$&jvMXbL@Q^)dLyk4zr9VY%^7kbf*Zq_Dg< zY&a?yEO~E6&#=nRL@pew;YLUArIlF8kK))XKvGbR6@Uwn+5wbgz~|cfP5<0zo7-34 z2Et;G6VZC^PUQbe6ZY3R4sAE_-UIvUFM?9|RgO+;!)lwbabiG^^K)9TIpI8lpsuT8 zcmXt@KETsx?8=0<5TlAJ1mYq*5mUoN(ZI?atpxWIFs2>qu;ddeBGlqnZFNavaz>eB^dBMxi4KE1rNG;U>QQ@bDU$Lfw4#oab zLTjbQ3QkkQ15?V?pS`#oO&M}NJe+6+e`-UA-#|x+FTDIp+{@3gO!Eb3u`wECpc0yY zgq$>`goUMiN3(K=8YO!NQ5N;dB33SP%HqL#be0WyqOkH-0N}rt+>>S?_N!maK+R!* ze8K%Ec~IS9r)o0dLwVn-PoBrGXRk9Fq|f4suIy&f#x(B`T1@7)rMo=0wc!;$E|T+- z&^Zhtcx@}@p;;FCdEAUrzE36W(aXy4Kx@|b+f2kYo2rIwcG5IeXqE>ExwE$Pv}TI& zf)61ong-X(7;%geGcqNJfdD6?572AW|Kjeu+@1W zDJ++9lPkcI+*sKXB;j+K$%aYo_u%W?FDN9(+52>O789>_2rR=rc$QAm9^Zrm(V{#@ zqaPqFAp9^tv`lLGU0oF8tYr}*ALo6AiOx|ko`9VSE{so1>x1Lhnfw7~59GSw; zD1Acm($go@rW%Vh<$MM7ErQc?9UXi>^1w*d9fh6^#;i7J7&hmo2_;_15H>Imd5>rm zju)Eq5IQEn$j}AWt+w_Y@Hp6Ynbu+bdEB>rxUjHg?AV{dCUJTUU)FF zrjqXQGAmUJE#qfIs1JKV=H-x_Cs~aoS4Z_d8*b$o3l8P`7YoViG)uANy>)#d(WKA+ z&EW9q#SoHCV|S6hYVo;`GC=icTMpZqR(%FNrj7UZlp5V)??a-jkxw*hw?PP!}Od6U+^r7?x>jnpK&f>s>bP78SsOY5Yczxgc_D zh0NLL4MRmidBo+Gu1de)nc38*l5>|yhTt#j5;ah;$L4}t(h;N)0#_c0zDmAGBern? zWHNBu_fBt0L5e@rXYjgwk_QWbZZ-J<+MWkCLHB6%9m}++9=RAzO?~0;G)0nmhtJvE zPcO>0yoLT|4HWcn{UM*m+F?JR=CWx|#0 ztoBs3+N3zPam~a?$2NlCo&D^crx(nQ1v{S=n`uAjp4u4oXTxijxkV^wa7{v_zk93^ zE~F{sXr4PyE|nVxuVupKkRX#JQBu;3U*+`Nma6Sub=ZnHu)3f-p z@|trx6{HB*<+Hi&$FFSHrKGJpwCN)#ZgDZMC^>sHEypZncMCtP^htoBIqkB=O^!oo z#>YpfxE=6I6Lfc24Iyga5R&!*3pjf5qSb=qw)Qr0i*)el;zgcV@RoLaHNe zd^ljyOrN{&i_)>7wUEX85K@&@`FHER22y&$^IeZ77U?z(-5LWAp^`&vyEUYp&q@Tm zHqL(li$0H|v+A#PM)EaHax`K15h5_F*u_B$ijq6uW+tixf#P3sV8Pw*HAM}y)a8h% zx4gasu|ZdyKCBt)Z(oYr7@1lbuUHbJodMYn0>ltbu#lY!o`k?!(gMxZSpIKCmD>iT zZV|*((oZKAXRnycD^Wg5O{Vz3P{Ae4uKV10blXd8Kd=9&3Hk~d zA2~-?vUYT#y7z!q$F=7^^+peQTjD2_rHGI&kz0PMXWM^auz*w-MO)CEBK1|_ZUnUJ7?r{T8cwik=YM=UlB)dg+BG9-1K zmsS%7s|jY^UL4A-ONtd#R=%ZLh`OQv`5u@1JJ2zyB~iP5lSYt)j9bE;f_PQ8(G3dG zHjT*~ZTd!9nqKypRq4K*wRD4#esOX|)jWup<}<9t>FXPt!mPKWgL5nySWO$R>VB z6!TH-GG>DxC^!Wd!49QS>^@Q&w$s)cp@CcJNX6cIrEi6sY=bp1ETf;HVXE@oh1Gx2 zU+9s}{qPF8OXy22>UpZOm_+vU*-UoEP+}_=8`_)xkIL26@NwJ)z)RR!e{Mni9zy@k zv-!{Z`yUgM$NDdE%&(zj)iPO*s`b_@g2HoG>WU@F!;)%yrMt1u3F)Wm6e7XsL%SS+ z2rt!#=5u-`D1N8EJXUJ1mI>Pa90h>wb?J^2SVvaB?sjoaH`CtkD;bAJvKL2BpL-zs zNC4;ic|vFVTa_)tdDI3B5WwQ5bj{IPT4K)UM3jZ_I|vy(RB=KYg;bklnG(#(WaF_)6qcJAZ?NCoHmI;$slO(w+nMkFI>LMuJL2)$ z?h0#rZ5vV*<41EJIN-0I;E(qpfpt6l%LynBT?}LPXlDWKSb`1Qj7+K*QdrVuqj)4A zdz{H_ec-M$&HCa(tF=v>(WgA@H7E&m$!Y+C(5VAZ$NDF-{y#Z;o`CI- zSdE$d2?;$dfQI?CC{+>UbV6=f8ldwB`W%65Tp)faM)Laphm0>d0_U=?wy>Bgfi8w}(e}WT`50 z4f-X~4nQ`u!oJ^8M78ImxgB;CBy8huo~*;O5|`LmFyr^R?x8%3>zAb>hc2TvLj^G; z_DPkGBKi%)qKrqdHUqI}Eoau<7iAHtk(RfohjN*K)!K>G)@u9UQ3XXN0{$)VnCD$Ftz-Q84YH?SC_suB#6 zC!JzXHZspKCRz@g=IvA;vkfC_eNe7jE=9W=3!OI`pQKX+fV)>JcHT#*VW!?D6|h`l z58q`Wa9s?Ij7>p3XB^7Gb)>OvI`6q*k!OK!QN%$PUb)%cV|@?IV#Y5lpXm5Lb?Bpk8a8-<)DWlt1Roosr z{K0y24i5ka7~=8CRSfDXmpPg1`}@4;5M>#ib&iX8ea}>3-ZtnFoP&dVRIjpsQa`Ha zEB@E7#2JeG0&K8Gpq%_HjrgM>S6%~&C?O%%$chhY=IRD|BcQnOK_j64r%jWf&v2+Y z`}%X0{=8sCQ@rY{+zf@U<>ah!Xy55P93v~pB@4>UyXf<6{L7+BKpKc{mpQv4V7K9h zn;B2Fog1#E&ZML1%CuDZ$LBJcOky`C#+75(KM`#LjJ(rV%_!|tvt?KqWX!LYj)_NK z<*h1?^`~gPtk9Ecqv}pO%)XI$c(f|f~RYJ=&-|(Wv1-imH0Xd(9?+=9&#v1QYN7kRVy`lA)k1I6|Ibox~z*xj>jw zJB5pVl-`HL*ZJBf$`TEFvhu&2zOm}jT0j*xeK~+6Gq-M=2T1;!1~5cw19fYxmgp`Q zJZm^A6wOx;9C-|E|Czqr({267`gP9&h_Yi|866SHTm1lu0Qzm|Rwm|;?5uQvmzzNT zcV2F6!8gmVIen=Q2Rj;k#nOtIC$sAAbp{W8*s!X_2EgZmNXcVwNz6DrWYHU~KGUUw zvAjIBv}HoJk7u<9mi=&*kZ)>4jDimK#A>R@73U-hJ_kFxNN`-1hEi=jbZBn3(-LlTJ6Aiu>NL^>=a;zx@)7Tj z2OxX_EXv%{kwf$5!Q#v`8W)OTFM-a=I^tkF=ne!RKsZSfeJ3Aip%)k?6A)gSTRPm@ zCU}nDsyYr>lRXdsd^w{8*gm#yRdWvLZGJVwwz)wU8&~YbyOU4indq9_f z-o1VM%iT^+exLpHS^soEnY$2Q2%fD^iCGukJ8iAo#hMlt_}Y~Ob!~DbEguKu@q);`&4B6co`yOUzypESrn5kO`~eb114uN_lhuLe|2(lk z+=*v~?hpk127t4HAL!^jU%NTdab|H~KP}y=>QD0kZHz$>9D^WR*1%H2@;UD9U0X9e zyEc+%SI!QwmvPU?3FNE*3`f6rxn$2(dsf<#&o152&%OU?{Q=PfzY(I){b?5Z&Q6jk z5TiTY|DObUcj5a#yvTo%-$q!P*3yx$#L_Fq2%lrccoTMM6)c`^J`yCu*mM zW_6gg2`xVx0-24Rg3Za}Ue1vppDU>}*$^isCjr5- zjQ@YlPnNWd$IE0XgqJ+O&>s(ok}l=3&N39rBkCSTn9xd7@^uOk%6Ct889TySIhQm- z@*>UBbWAIA`?(9oG%kD*Yk2wc26GQFR)3$CFdboTcVw=X_EJn_?nZFdeiVC#!KFzG zxz|k;U>6m5(cC0N67f}V^==vIKJ8O4eo>A)VIGOrqCMPLH4@euZJ{D9*&=UKm!c25 zldqcH{;E>-VBEV`r~un)?ScwWH6;MYcOIVRyF%WwGBVM?av=&X9=A!c9*m3ut!L!O zUq$(#C5?DS`z?2t9>zXDdgVfC&?@e;hNri3NLw}sqtMcroHo_4^fPDcZG2@R998+D z7WtBT*7=2pPzoHQ2i!67*>yw>@T;{#p=U@Ndi=3-Q2(7ZuvJ)r`oBC9m5%VM-aS`A*V3N<7|tzYZvHB2VL{Bqj%LDL9h?0_y#P+ASzPpNOHP2B zam-XMezXfiMCWtj4EhOlDV4>VQQXVYH%U=RtfOZ8mF^X(I;IYVSN~AV)x>3w|U=k?>vDONjZg7a?h0pMR|oTGgNd_#MCk+hxe`0 zK@QS8w~PkADL)G&`J$($|DAPfo(9j(dD{2xZnlh6aM5U&LBl$`x!T>ZIQ?L+dy?m` zd&?nJ^qg7%8DqQOgkey_@qbYm+V!%ya;W!ipG@^d(HV5TcX+%c&(aiS|EBE`Y3$RX zypk(bxrJX0={x=BH(EcrW*@*U(Kkq=Nh#k6MrCG9Ppoj>y|3M|M5C`1mT_)fA>LV4 zmGaX3Gm1u!i>5{lLt(Tomin2Fuol`oWwFhIsi_}B8gEyt*j(x;@F$dR;NM$r&pGDP zgjY-J&w?p-)z@h;R|4`mZZvk?YFv1b+14R{K3vB;eaU{zA~8#+iJ^-zm8J8Z^(94O z=5f#Z`puZOxAl~kd%Pc?TROOz^TurKMHfgV8Fxmg8ChHsXuY&7D+d|=L|s~xVTvKO zytY)ZN){cCN0YzK&Q8s}w?<_`RewHoo_lrG@|yY>>UCMbE<0*_bjEmFge$@rkL6T-2dqUX1@xN|URnsox~Lk54{G6Sghxl$G7nT!Ifue)iOQp9C;bTX(N*qZ0OZGG zIComJ>vm4pi{2*J7k94qOP1eU4<8KP>(WQbH{D-ig*s$RfJXcm_ z!o2`;O5IKM4MGw$o6T_$o4M14Vhk&bV*2Mhd}n!QyzCP`$dm76k*QEVtEJVP6x7z) z>pyBPHMQiCB-pE|pc+_EDvLANZE^Y9EB-o`F@Qm95&shm9uES6?%{|sW0`dbo6L_h zB5g5b707G7wnxU1`RvIm_q)H|UuYE-x+qwtk;;%U5BGU)y)I(mDP$_-9LJBx??n(0oiP&#;Hq7(3%WiQpw=q9=-y+cPra#3PzSY~O8jc<-kdbY*)Z(bJFJ1Ske5S9D zkTcUYTfesIPZ%~d+qyEsz8DT|Ap#B-7)AgOQ+(=k$$J0F*ekMZBlbiynt|gHT0?yI zptykzJ1Q!+Y15iA?-!evyvkbaR&U=9fPteApbap2Rq-?cOQSq`8x;$bp8}4>zkeRC z9G_@#I@e<^Y}1I&3FkMoOr(BRTwD7C^iBl$*_Bhbp7~jgg5>b@mft@^Bsfs^?$iWi z;tTAhl}Ozkd6bmM572Q@TW5TY(@wrcJ{_upGA$@s_m)p(>CpODE?D&kCVq$q+Q;RB z!|ckDnvF&)$;~fofJ%QQ`Y%E&{09SWe$q4*X^EKTfe~oWN#)Wk=qdIv6QG_U_5hHl zKlZDpal70ay zAE3c|(DfQs#A3i*cZ!}YN3)g(E{vL^MkSS76!YZ){tXwz0M7!AgcEAy@w0UjH@UAu z(knn5qY2tLivW_`dobq!6>w5A>55-{!VVP#bs2DYT-TBOoQ=-k5E7W5*f7qc!jKVo zLY8=;2tvyfI(#jW^MI8NSmX85b*MnFkM08WwM|>aM4uKPnw{<(wD%R@wqTe>OVA<8 zT`vRsUlFj`yoDl(U2y9U3u5#|rUZH==<-u5qGs~(3M)Vr-_c3MQKE;Tj<4%_kGF*b@U?CwCp~ zWhiv>((Y$oB47Jv?I==mp{&``2d*PO%2edt{?f5JmrG)6)R#PuekOdF#zZ5vBnT7f zx3T}|Ww_aFEX?;X6G`~mSN0wYIKO^2(FL1KB?&!7Mgty-r$(4 zsH)sjr!Nb*AK`%214$E%**D!qynbh*@wn~w%edkb!#ddOH*C;9`7c2<#JKK&8J@6r zH3$fDNlVZb>!RP8$sb*m04TT^k~~Dbk<;k6J%m6;Bg-m)FXypn!X!cVCAV@ofW5C> z7|C($l~=c;q_PHCq5lE0g|dL${)ZW~r=8)aKhtWSqGI%oWQzW@j7dAuvMSKi`SPmV z3y=|KtMxs=?g5HI;(cB0Eu~fY_A>Ih{i_;j>a1t@SR`!PUpHRs03O!20e|s54=oH; z8zL!xfD{rQ4Bg!^zu3LJ{>h1r5a6`X$q^66>Um|?#?K>p*5EtiLdbjaaDv{}R{RAA z>Or*|5JewKz+Bb?yn~QgT;%>SfeI9L+Z)Z=Z=A=G^K3&q{QES*$MFc`mkmJ^HeiUM zjU@A%hqM53EkXtdrJB+w31@M*Lf=()`qzM|`G&P^dIziyXy7OUM80$q3}}p0@ zwF505YOlqCCfPm&)V{L}xJl34n+s^V-|4_iVg5j6NAL|(z9c-t?$)l@PQRz+hVkpF z*^KxXok2D#Swr?kB;l%z1;8 zuWU`;!<)jwcWohM4Y*pz*VAKMI^%6=N*nUy11=2if1Yy7fp z`IE@}{4yl3k_h>Pf7czuaH1SCm-F*Y?VY=Qpthxj?KDZ_?H22OpS{mpO@lexNLF_j zAZMHqRm`&8WCouBPu%mer!B>B*hz7Y7p2osed9M`+Zk&*%y44uQzh^ zd8o^r&GSRqgjt5^mdki8ZF5x920Y4p6tNSMj)ujL-#`}Ue}Jd~J9nt!7Pb0<$cZxj zJ&@Pd06QSus9R&QLbsSW+SEITh9E69k-|mwhi(foO_vm1yQZtUCB_m;*+u9 zBEgmY+cC2W+dkuE8Po)^=ZY~6$RLcnF!1DX;8lfvb_Mtr5@44AEC>{l9YUF@#vYtN zqM_9^3%$Aw)C!nxxmTzmPE2Tdo92my3r&^#t9ekaEwBxF1;aI{Ozw@eiv;6JSh{}j zHSX{+w7@3-d3$#m@dhBD=aVA+yt?}HTHd#@BWgWM7fB$eW=3g`hx@bkH7!RzjUijG zDMcKFaCAMS(U^P8}ay@|h5Pt}Lv!{&M*eV0SB0EVYgT440+BBk1wa0vf)5>vxLcmVyx#F7oJgua!Lb%uDx8<%-7Drt( z<*Ri1TwLb+$T;zVwTysPg#4%6F8QW1hx8^|tuTVM>~}G4U)}ZSS?rPR?VEvZXIc|E zy8!yn*h9U>A6EFa@g0r}KN_3sV6_#MLv(zVrYCQsuc5u($8L*v%S9$V;9JC&Q-gm5 zZ&eg_m_Wn#3Ls8!dht%r#~op?Bp@aKovbPCYIxc@460(E5>sNgv(l-YK_=-Y@fz}z zl%85jg_Bmqklp4q4? zTaxr_4K5q?L`gGgPP;k%F>p>7P`0$|o)YxI8S~)H_=QwGEDVcb20x|uT%XpiUq~^V zYIE18<>3G)+o!m@x(M_9u-wO-MT6o zhL>)ppV7@FG?&s>97KSJHcXK3%Z=#hSbap)X-RPX1F3dm=?)GtmYX?St_at7Zb|ji zY-08@BmbhS%f9q}(4&G{5U<3z{FN>6RSJXJ*rve=m}vfjiNvgtBJ1uu0a>V()fkJ%bC zl&dn4pTAzem_hB3$2S*&M=HzFqM`W)iC5ieZh0OvB|emevi)4RsVQjavMe{xBdHrU+=F8YCKp3Zp zdmVT9$&j7)BML^7k~@UbkzZI+RgK3z+puemvE{PcpBr)eN}Jzs0IMgTPiW0i5WOaD zWM;9DSnA72e?I8~Kbpmv`O{VDYEDDDxhm+jiB2n?u3|=3W9{e;KZsML)yK9pY3>+# zpI5fBmP&(bFDrsg?=r*>SRdSy+sF>~1s^o*hVACM zJzRieuTNI<^AAu4?6?n|fwl8biq@*u))ft~X{=y2u4vlc;4b=d{i57O>G@o_OE<2G z;UtnQOrQA6FGEhKY*lya*Aq>~p^Ia}+n6tDk?rqtY!1uN9~pgsp0@(XImUebJfiy? zT7%(T<+J5HT5q;L+B*BZE87~8bh6kF%WyJ>iyBCPFYA1wl$m_o^d!WfuBNp35x+{j zaxl@FbUqb=4jrc&2w9_q>`i@0GFG7lBXt9L@~<~r#tbV)%a?dbTC$VRFMANUVw}Ls zkJ#nDSjPlAeOQGcqEMC?Um7N$K#l${OrC6r>Eo+{KvAJK3{Nv)-4`Q_j<4(`p!QLH zfEhc`kg&rs2A!Xt*Ki@#-_6()vP8+a0^W9U{I>vVt>6a;psG6C3v57%e$s$9egrtt!G3Pj3Ml#^OaW;0<8%xd z1CX7a>eZT>n<+SS978yb2h3zs3g9$)1)Z1txy|YUX+(Zgm&M!)E8fLB7mD6FJ>k0h zsy^&N!R%*AegLtj5)W|GC%OYs4O(C#z(E@XL?o5}!{M0_H0fRhNh6NaaE_+uB@3RP z%nxEDWL%t=TvKEK!n%lj|I=* zaHm7jHnzHShMXG|KW|PSMjrOb02u?!Hk}4czvk?-0fD=0=?ma)p99qfcp=C<{GQo; z&&`}qRo`O_Hx(i*^$b9y9y)fy9qB;6abqziE4x1Chv`dI)7 z{t0ulLq((iWUw;?qEK9Qr}K&vT((Y&NY$*N+G*CDqrsV5hZQkYw`Y->)$191r_n8^YGXT1~O;f{adEpKYIQDQ0XEQ6a36XlLP+W z@>62}@R$#McoO4VO_RflP%%0G_Ea6FZ{qJp96p~u1Wjs;eMw}{9}Lc9k$1iRAc!c=nHFl8+U_*l zbB=RxQ_qA$hmh?XZkElRsR|stagmSN*#kMdbY$8q`-G0(N(F%(>ub5si{IU#0v&vt zUu4+!Zn02DamW=86q_-Bui!Kx`mo+^*vUEBWcB3$_T-uB6mFo~qm`{DwZY;dXB!8a zlt4nIV2h?ZV{RqjZG=z zlV{C7T0Svs9N0}!Du~HF(GS~Ve9(rP*qjGE_ z{n8HJ4)21fBU%&kRw&KaN8ih&cYbeK_>6UcV`e}doWVbzDx&(_rWGT}l(I@$GQEfI z$F`G(ncA}(WW>3OY*<|zk0`LqRoP=>InpQ?<~bTKe|h( zB8=b$5iFO9?0IJEfO6fi$>r3437U{9eIJmEELIJd)nrY=? zQ)~}m&m8McRb_D$v?v=^#G;4U>}=b2?9m!_N<}&2abf0;*7?ZD`0hzu&E?d(7kmvL zL>(6gXPCw}XDF!{n1kCHa|d+ZH$gs5;$dqCMcyS7UhihM(n|~3ybAr2nF1Yt2RQ`g zYUDpRqWzezss8Y5-*|Ju9g7s^`kp$H7sZ+s;6}h|KXK>a;2>cb>8Ml;8gj^K1_usx z^J5(QI9mXaqs`_z;%*2D2CYoZs$2e|dP!X^-c>sCQ$BkS(M&5S2B4O=jp&hg!;*4oVBpegT(?X{S6^;XD zA%~3d!aqQ@4Ec%1$*b0D)agCCBk_RYl4LU9b2RrDQXmZ2w+{!}684)Tci4!5R=nRG ztR4|HLU&|^oqm8^r6o^4w&rbXOc&gHwym~sbn)_Ru%%WocVId!k$YXYb&9$-A1^*N zOA@a!<7F_tfT%O+82)q{i3|~6;T8H;=&47&gT8Q-#{{IqAe#ZKler}|Tw5Egnl2-$ zF`8C3l$qVAfOoK9w>&AP1)9oUO=S|W-Yo_UY5?v3wy!-7n*u>3L+dGn$>hutk zPxo4<$8cUAe}z;vI%+IM9xrCIyrZz#Y~`7U)n;HtD#}|LzF6reTN}8%PN4$jV7Mtz z-}40FY@0auvi6I0a?|N~In&wE5(p*AMzSJi><8#sD678|!8H{l3JPjwCi2a(oSr#1 z!f|n!MyyhqHm@kImPJRcGuA|cMX}pUXa}jJ|EXa!pIRx^5)%wZFOTMx>~S`^J9@>6 z!#G!SuVD)q-ncj8{EWU$hg_eRC-%in&YY0$L5_Vf{`#KfX<|$fWC560fC2cAH9XVT zVWl~aH#!0z?aCpxj~19`L!HgF4C`iYz~SLJ%VWD<(BFCD3>Hu zUYCXIom}I}xb@PGd5uRSx@p93J*NEV{PP|1Sx}>td`+yQk6~OCiH@&T<)@ofaZ*LKTor0K@?m{+Y4A({25GQ_aBKm&N*ludPrrH zC;_SSVc`8^mv#WD6$5HQIpt2FVD&LNxfzw<$wtob@CRPv;k_=nLvUBI6i4S#3!ae2 zV;zkAO5C0fbfSU@jTH&rKR{cLcOZw~e}E1-v-15HeK+U8sZp_sWwx>@H03 z?ge6*q`c0+nz43Is!nbayuSj3xA#J_^5Q91`Y$cYf~5k;vu%5X{LUw6c+1t4Bb5_; zg-IM`w3hcRkMq)xqSCJ<7~N*Z9@tGFF&3efid7xTz1!Tx8DIb^-^I?P-G%J!g}q|S zZ%qZt6n)h9dd_@@o==p(+N|?#E?NePyx-DqvNuIrUg4`YhpoCTlR_n;K>k!zsoQN8 zN4a++7Y%Vr)6rReBm^VUxLDH9;1(5H@260XS1~hV`-#A2fT!lZ22tq;w+mzAUtUBe zx;5wy&<8dm%YxGpc^_ZzRnn$sXkD?7q@5ZmzVyf`{0J*J!E}jxk_qtRk`Blgdi7se z3TwUQCMTHSoH$Q+Kv>dqhyVbgqTE;P zLE9k+0Fmh^4lF6|QpPWix3l=>qLhW5AdnoSZ~kPXQu!lm-E=P+f1OC)aP=>smvfBT zM}qU2=?pA>VEG#Hqi?${($q(xRJrO=>DOl<2mJZO&=GC`S@;UgvN}(2M4nf-r_p-a zQ<72)hc!+Uq5?a01~%nBWSSB{ji8p7!>ORpUfI%v+}-^1gdb{s>KD9O2!0W`osZIv zC%|A50U)5cB$~Yk=LZP-6mVJa<`Ec*v`p-1uy)tY+B-sy4*f7#C3J`?VY=gEV4$Ee zn4w66O`mbTAaEKUo%sRUT%ki<9E3KmA)x!a1ZZqi@Vlc6UGf3bTeqbF>J9tpLjl(4 z!DV^-_ZIVOC2@mKnI)E8Y;W0fi6 zQ~5n)za4n2zZ4&2eCG1vLq1v(5n9c*uBix64VS#$=e07KMk0N$Vp)7Lu*T7t2F5Zp zGs6iu*l0NSc(aQ47qBDYI(Twmi$f0L_Pnl(L4wjeHSeq%Ge#P#G8c-zkb@6Z^BnEF zr<$5#>`HQEh@~xQNghGo7#y1g!R8@ev)To2C1e7PqxXn6PffJ~bNAzkPsS*Fa#(Xd z`4fF5Mm8mTf$@d6Kt@nmkX!+xG&kcWrnPWz6K*u7e#74WdKFeLc6j_~UMz;p2v7iZ z8PFN3r|vwC0-hK0zX)G@H5Q|U(OrRKGilcqBJpc-S7L~RtPCw@kNi2<=jf^;#5))_ zf<@0CDc9)B6_=r&&&o)x&tC&idfJZNYw>4D(E6U!hD>e-fITg0 z_yL+8gdp%Tig${2Jd7vBucO-1D?+ZYhp-QXN^U-b-hv83JqFnp0~_7#ET z`9D_iw8Xr9vu>fVFO5M-{;TNBoL{!)epsC+;7k9vIn*6;uAp5XQu(zA{RRN;f!SLbgaOm?2) z;3r%XCZIeIllnkxt2}5S3h2NbZP$B<)Ik;FyLmUwxH+5d@+<+MYB8+sYTHp;Oy-OmF48X7rbQz!hWlt;Ls@M- zh*kLMCy)hw$X6$bk3H~QzI+s!DDw82!!W!*b%h;>$O$#w)ZK_kF>o_MIMOvmjY*V7 zlE!44m|K@weF|^be)mm?*Sc**)|UGz)>9n~u?%&2BinexYGkoBqtrx-UT~-+378m5 zr?GWhW?+5o?)+A6?)>$Ku|eif?_HFWMw_<$8vVhEI(2ABP2tM(N&oiip+V(L;Wic9 zTH7E{8>~nY(D8xg6kU|@*0j6lQ)L!dbc-^nIaAe0ynN@hcc&_~-BDg6d@qje;ZB?rLw4-^dscz@1*b>LQw-W{9luG7bF!#j>e zDq;Lw2!JxsCoDr}zYg0-&}rvRfP;mo`SvVZ;iHsW?;%#tniPoHEm?U=SFG*Ub%Bp!A2jC9q?#&YjX2%ER<)x{fe z-#`Fz>5;s@Rgen-t5j9g-pYy$U(wYhsWh>Ol$|y?n_%j}oJTiz(SB>o;n0`pXMm4= zstI6o0zw5y?!UTl1S2F27~dh_HJ7`|e>mfL@i^Yaa<{Q7wum}zR7nFCoo4Z^3j z$lQshtL2+CRcZ$Y^}d}@L@;x4fuH-h_SUfF7FiKsqn%yCpH_5m3-x_NrMkSoTEH&D zBubInMmuH6ClG)yun};;ns{#agobZ&k89ICiB)>5Sj?w##GuxS1=?WSciu&F%B+$n ztTgS}b|n1iyjh@Zn%VN$)>KIe)pXUmzD;OWd+x=ZyY#%_Jmo}&VW1pfn}rgduvyUmng!VET!sD(*lLa z@Y{epl<~xU^JhWh|Ly<%CkLBPlPsBHqOZAKk*D|m)m(jchGXeW5?{9GN4Ym1opj{< z=_v>N&tcB*3-@njbtoq+ehPM!Q$7gl(@ z$LmiB(&}dUkkDfscWUSifJj7kKz{lzlB@;>8qL#eBdx6I@(CNbRbKANuAFb4Ya#sM zr5_GUZum3>1w|g@uO9L9g{)BgV(qmzt76$?KnNQ%-pWT7?r18@)wm5XO9N0}yVo|A zy6@xSz0dbcCOP_}!TrVXJ_^g?*e?v%joBE`2Q$5||8xo9X~>*^&@q9;mtl z>|UXO# zO>~HBOKW;9l6u#7mXZb;%VRFRoO%)5sw@+wUr}jfNdg?0es2H2c%;G_CuwPdqswE5 z3b^)?!#?d@cpm&3TVAnUNuc8V#V2Lt-I5C3tSmhe!%V%^^yW4xCB>D>nFD!hu79IDTO1hUaLmevVsw$}?`*%%_igg40ZB@bJ z;_D{_c+f`6IpCy^j|QSc4dH;d_}yP!iG(QRKrFH00NWoBx|s%WvpX~>E1+%`J)p~& z|BFk9A$og=U&#D^dNqE2RbUx#0!wVmc|P=8JQ+--IFL-DQuAUO15Ww@dJSbk%FwWC zqN3d_4f9a2CW4gl$3;CxNm>Vm@}ZuEa%4!f5X^E}CzSMErg)Sjk9y|gB_OD`eE)Bm z#Q*5^f4fQjCqRs|m4H=t+Vqo6R`79t^<6(Z;}Re}CXjzVyw-O@&y;swKJk@ign(s# zij*?GbIcoQ23Iyns6ba!=5S%u7T*;s=N|b6=@=e(bGUza?<= zNir?!-pl9B;sHOl0>!p(V~u_F%XpHp*CL|oRzF$z8Ks_~uA`%AGA7jYvTEbLSxK_4 zVyakMEz9@l*&aX2iT$U`l_D_Pz2Iy}w8fpCrM&=E%z!P=fB`7AEh1GH44P*+vdgBW@uf$y zwi1yKwg)|%47EQnm@5xqn;Xb+Wy?0b1TDxZ5!(H*7bQjW@r^~jE$Ew3sY%MT>hVn) zorF|Zj-glV`LBw*DAm}AOB#-N;r>{0g<<5c!7Nzjh>yV|MZI6%v=;J>w*Kx5CyCCR z9(DB7f$z%VN|YCtALqOAxf+ixxqIXt>?&Mj-IAYDTf4IDvJ?2-Po8}*guJ|JIA8yb%Sq|94WL(<2cb`^fbEp;hIqO0JxLu&a< z=qE2Of`X(Ka#$-E{MA_9*2F%}8seOMKy)yGzL;n!tU?TelzyLqRz|Wd+m3W5oDp0A zCGxw>56egx^{~&_=>FLq)LR=kJf~ldpI#*epamz6<{sQ73^C%H`yIsYJB^o_X<84F z(x$vP7rY|%_-`y8+A`K;@l6woaF$~SZf%A@w`6!Ahsy+*5RIK9C~7b0xB;DAcXix1 zrSS3p*SsBnh6Onj6U;YJGVfhG7aVR*2SHrdcJN2@rK0jB-B?vUx=c6nl(jN1$cde| zQN20v>XXmSzJ*!R8p89ojJ082;n<`mPFf0SIL>%ijJ>}!#A34_j5*n;l_$@Tpw^9p2(#Foo z92+$V4yC>yOz5gEto1K`B)5@rPE24qbTQt;p46SF4}j|sO`!tey^@(z4?lO^(DzEF zA^UWcosen{w|=Z~Um{7k?ucw||M+4xCr}mvz=O>(V_HyvrWVV;3Z2e`paOSsS-|gv z@E4F0 z`JhkwM@EgPWf+S*=2p$4K=%$BADd44UJX?;D4Bu#$b?J5ICB$HgkasqFUd(c zTp-B*>+Z^<+1lH6&{Cz8qKK)~P(#%aZH?6mrN>Oo)X>II(pC{;v}%f}C~BUInyP{# zC`zayNNR|AsCkN_si}AO+;zWozI*R?*E!#CzOMU6va*uj&i-ZZo&CPQ=Y5{%C4KbT zm2BJCJB6@6Is^de4mn$tv#4kXur@XPjnv?u>Jk2P`+^&Jlp08!g*Fi84LV)vG%I{+se*at zrWoB#ipn{)vK^ogPTXOKJ%7YF1ypiK>)BJhOp}P2E-iw&>=2~h8^RpNt{q1u;}=Y&B}sf*YR!cZK*PouC%vE-e z=^J@n_uY25wV|arOx+8o5wA>qT8Z@-I;V+S1KXEh+obS_vl;n!7xsv=^}oy>D`M3i*xQ&k3-i zMjqxR5dS3@)&IYkEZuE}Su_?G4Uuk|{a#xNH>9-H(`)W#(5+AIByPF9g<7EiEf^kRVAYS-ksw z9*!Bf=<n!P>m<`$j0m1LvO^KOr`WKd^fC8qJi$=!vY08L*}Mdd;345dhmDl&hn99P zM~zXk&)v>!RaLxg%~zSYyS^sAQ#2L4zTjkKPsc%5Tl>B{0RKfA&WoKeMmJ zaqocQh07l{8-@7P&K_MDiyVbxyDbLqZ{yq&mRZxw9DR^s+sWv0Enc|^9mEv}+x{23 zcf6@z-AXo1#-_+A7F$=79*|%mYI_;c_+;1?hj`w}g08D$ioJ+!{(@wPqo2#GDCrp@ zlI<)nk$ONVdUynhe=m;p`l{njHV((PRbTEy++T7ka&~#QL_FvZV$s`nI=}u_gzt|3 zY77>$i474ykSdTpErWBH@viZTcbDi6EtWdEk;VGrLZ?Dw2+I=i9Qx8v6B&6f|LCKs zdQwnO?iV=)xA8CQ-$CxEEk=r(u+;X?=w^qbbKk0w&Y77=l0A}i=)DrXv#agkqI|Yr zQT5Go%Kfv<>Z;NAT@$6R9(S;m&dg@59q`Toz{aL8UC;vndE~r&0MxQF%q159`u&?5 zyzwVB8s>;uZtjW7b9?wjs)da+HEk_>!n@r*H*MBT1G;M56c6Oahp&!Cjg;FA0$Fwo zq|iIn>o8h-)CXqUE1LO%Tfw*yza5T_Qk)z!U4CQL@GhBO9L3Hh^IWE*WVFGgttQcs z*f~Yl7q8x5VP!DH#}pKkIuR&Lh66I1*=vBwwQNxU_u&Q`a`;F0`7EfJydV}O%gv0_ zs@i1B+B8?=S5GoVi(lJ07l@m-Kr*7&%-g%nGEMQlnMy~c5TZqLoI&Boz`M_5Ghg)4 z*!)+py}4c(x!!E7e(Ox&aRAPaID$r59AN;iPTwhzO$egQ)+{SQFudKkKS+$xPE>14 zg@N3%(8|lGNay5~FPAG<0B(sbNoAOhyDGrmIhPS6ROc_yR=13(=4C`AX;f9FY|&&4 zxZH{ty`+mNeSU(E2Teijw}#~*_S=qx(Y3XG7%|5_BJRAjQqPjA*ad<6_Hax4_;wwI z+fT>|ER8I02Jhh0ZYk&Uq(5{KPc@&oK5BX<`XL!c3{?0imPry0=e>sMTN@f@dhEOpg0hB~mMNQU;_>5E@M36`_Mcv*^RYiV-)VFGDEGt6=gFH6`LoK(4Z zuQ@h++h#oTwMqRD&ZnuB`|=T+N$9r!ZH?M@gVPPQgp2le`30iACw-2)1vLe{D}UTq z3KxH@prCa9ZpQqC)S2$4tsyu?$LdW`>~Q=w&cDU>Wz8&C|AO!`M4gAlt_d1LImETN zKnf{+fLm{nT7T6|^hR)(P%0wI;R`sc@6x5aC&1)yNG=oO(&6zgl10ncSYq$)8{&9L zWr;(cL*e^-i?nMgYmQlrXUX2PaIyvl;^E?-z5E5cyOLOy@v;AvS>?Hpp95B?G8!s0 zFNahOH*(`%Yv=dE7lww+<8>q55(Flc^%SbFw)r4~53bS#qumQEaOXQ|k{vRp(BR&q6!Txy|?`ABH#8aw#Q`#w^E1(5gBo-qWNpJn~_Y%M@!GvKz7lYSBynN@%wKlE~{?T@IEnIu?{yFX%G=A!~M{VT3tS5w7v+`D8d2M>wYGE278K` zpa0BLk#nH4wsB-F)d~04OIIk7xGlhQ`y{-Oh6y!=&ga< zH*7iIJNvb=_hEF^k~KJ}^`V)jaEPHcTiJII#^qEn9mQBVN872l_Uw(J0D^L|4!*~O zu>uMtXO37xr#;#kCV3F(%XS`=Kuu#?^nz8w>g@2`5eGTOAj8$7rSV*?%!;|Wf-zgm7t*%sZ><;`nkBQeCz?#tR$yGcM~Rgd0|^vaOWd`s9aH+hOEE1h}w z6dK2&Coi8CVjz9uKonowb7iZZ88Bk({7bci7X%0gT}@Mq9LB0Y+9M@rqqAm$|s?o5v1v6*C9pZ77g10}u#l-}-(}0`z!-kJL~d zUU@YNs*tTeHEYC0hcI^j8W zmS$F`@_D;y1dbI`G9EK-ki>HGhA5uO8f&92h9vBC4{p*v6yLAu(0Mo{egfo!>K*XY zC_p65S|*w6OB1mS5VZcwJ*m>}_H2ZqFtFfmTBjz+S zUF;+)Z$wCd5SfcAvT6!<__hq$JRyPiqsg&CxY*oE^*0|}wr3H$zYeF8y6f0+OwJtz z+UBJt=iu5o`975hBuEfV!qu`UJ=0)3AvI#SM*7N{zz9voxcQ?wG;&PYw&OtFB&+Z5 zSr`7%ZGKAhpf)Cq=YkClS^kTmC1iw0I}g$6of^!(5v|D+2($`?lPqq2a&yB#N|*NT z*OfNLc;YRGp@^t}0t~sa*;0?EGPa|nZiDS&fXLDl{9XhvuIY`SOPX(|N zOfr!z9|=GaFr5q}z##AeP{|EWC<4?Y3MBXu=rY;JEZzBm){qkQTSY^_54YLX^4KsZc=43@SDtNGHh*|Ry>_DF5nxPN#!qwJwPOLh>+X>r<E5$hMUzyd z^Wx+-6ryr)x&T10%{jdyn=1P8zmSb*#LhqeJUFVk)yiv#rtj;QN;5IU7nc7B5x`NJ4Y}J7!Ap7RuHC@-y_Yk}+TyBcf*7$1h?`MPYR3qz4NH zdrj~?nG^QRu}`PfXOz)P0@04$k8F*Jv=j`UyS;9=j~X$n42am^dOxm98YOVn); zs1SU4s0k_mbJ}MA#O}So$tfyYKJ6rLJdyi&_~{^9%60i^bsZ&g79&3?=|kpfOw=CR z%Xqt%{s5eBFZB{>Xhb7%GTj&I=Uz-!`9Q=*x!MuzOg&;wpRXY`3*0Fb?6{2^7q$g? z9)oc`*fpc?L_U2=83o$fU|y6joy6i!F87(@gwo*-E0B+6268-bdSBk;hbt-D zgbfh3me+8*oUJUS@GF*vEu7l zwNBH9TOGy-;U%jE?}c#WTqRlZr~v?PwmCul7;^S>s%9CV8T_J~L_VS*WN<*!TsoA# z=Sj={*EN$eKi}zkT}CsPfn`zIu90I8b(q4&+zIw#kz5O}#2BZaNfjfi^-$;ieWqmW zw=-2A%&0-5Nzv$5vJJ~P;DXjCGrT!idxDxKb7hG?4 z&LZY39q1I?)WwrC{V3bsManly8H)O>#Z(d2!6A^LB=&@y{d&o#IyiU4$t!WlC%--4 zZI=$tUrz}ZY?r5prDV3$rH3TXZg96Dec{e)9`wzy=R$@bc`xCdLG?3}VJQL0(-L<3 z+O?h-bmA>pm3y9a4|1<4+Vd>L_>6CchGmgdLe@M^n5#cR#h(q%J)G~epa`55E}X%% zYS)MAldVYnTc2v0AvZd9!+zAv z0<#`*IDb-Ef2Tz0-qs*?)-6JC{Da)$o?1t#*faWe)j-;fHDe4D(l(EbYfnt6>SiP6 z)rDrfnr`8c_;?VkxlE&?1*)CpWp74qR_YD;L5k9fpk_HAF|P9>qN<;D7;mgtL59N7 z~Qsjp5oi0`q(EW6eFuQ+zpXI5CfzfkXJ`(oEP!sF?5jfN=)S6`c z7BPXpUTMqVI|>;)8%$r(=H?H`KDY7_*S`|n6n2~(zX_j@B?f_$~bxBd@N*#V^4@ND7UeX>} zNADaZkCHV$Fq1%fgn`sv?+w*MB*_-DTUw;>}YNxE+{cEKaAd~xd_H5tqu%$=Q+!yPz3W0VMGLupig zQtIw4Q;CBm`nTWopG_H&bPoZ#XldvAl>TrdTgheDyAjLXtN-6r?_K%N1LsEYep(K5yASB z@t4?+T}TBKWE7-@FSUIaGSN;HDPu=>#EO{FvI9Z0V>U(hP*Qh5qNg#{*5xjR-MpCN zZHTx}nh2F5gY!5Gg*xC3GO%M3yu&TmOPW^5aS3C)!LeiYes=Tq9yUZ)@-+LX$$nlhzga}+4|Fe*P?+zF7oJ4Ztg6E`_;w z7UMsZy>870MW5Iy27a4z`Lqq=3SL95G}m;Zi@l{&QM8-pmCr1(4C%1cXqAT?*-aH6 zhP?eI7IuRD_?2(E17-TxSm!Iy@9Lf|yO^GTSA&ZeYp27;2B9<5H7tQ+{hv5otuK=L z27Kll%X=MalGWgLYoz+8M6Md-KQoHB8g)_|Hg{WDlb?@T5->s=dj7Fik6=Wo+ z1MHyBvmy0}K_)i(sKLelwr5Shq&1sJ|7h*?g5x(0BaN4{%a9#3fN-(gNoD(t!BV^8 z0P0h$sNU1GnJVAp#uN4L8a5Ns+Xy(>o+kH>5#f2ImFzP6X`pd~r-}&x&G$UsezU%9 zH|ZcQ=JpJ{+QbKFlI^p8Z%X!OjQhFMz8`%>jNDtfdoa(n>T_eYh41#nC2na|?H%JWKEuAtnNH-EA48xsw z@9%uye)av%-shfs)?IhqJ*+d201^@aAR#`0y9MAWfQgQd zfsTfWfq{XAg?SH$7#9Z{8;64EK0YxGB^@mdB{elY6BjEz0|z5DHJi{w4sITPettSu z5eZ>laV|c7-ajru!otGB!NwuS#UdLCfBALS0T5y$r=x11ATa>Qgh(iaNOwH| z9Rf}?q`$p@|9BxGqoAUpV_;(4!$$l<9RYxhgo1*Mih_oQii-Gcf5iI$Dj^!t10EUl z`|73`49>*7Z{u??8J|>ll4wkvF!7nW1YzAHB_pSxWM*Myd-#Z7Ku}0nMD*!1Svh$H z#pf?IwX}6~_4LgxEUm0w$FceU;nm; z|F(z!f3t^1G{i`wzs7A9E_VR@xnS@e@apRwa7$rtL<113%l?1!=v7Mg*OIvO(}~gCE}k=a%JeZh*QhF;ejQ9YCt*_4tWs zTMo`y2eU$C2h!{6{Xr75;~qi5Wzv)$p6Z6`*y$w{QY<&@OupsD%E&#cZhgBdR2s8k z0$=!v4o_wBiCogU6!r^+i3=+BsF()}d8)Ok^mo8~3M5d%TBS8kV~W%1v$pP$sbZ2r z5e^b?^F!Odm+L}oa@3rx`hjyDfBe zs<%&ZgH(1quu`=e6lde@-2r;lQ%bfG`pbs;WmAIN7K-;OHWpXi zEBXTOv3>r}BmQFz_hRNKq+-z9o&S&)>=HSpU|16P5y%Ym(N0?Fbwbgz!!?%*KYX^g zpwjE9=D#FkGTTx?qq6@T%DAEEK$Dwb=~gO| zov`G$l|W+7!~XStbb>2X4IKt)%R+yDU9&8En_2DMw`Sn-dcT2iT1TSDPyIwYd7N`4 zoHM>K#zMN}$fIh2)ke%kWSeBm2g?^f><+l^C!SE7EUH<cIU4H*^XG@T_~HesVK|u=>Fs$Cn&>X}I@>=wQp|m&6*k+L0?l82 zHzQS(-s6#Fvs43(w7!Ko+}aOvxU*+4UV=`UJY2+01r#~eithj&D$yW}Rn=|d)*L8% z_j~B0DMd=7yyOH(HyUHH(y%D_LgJm1V7!lR zy`4J38S73w%AM{X$8uV&5w=R}b_ncg{-M@?I@;Q;>qcbVq*3nWcHDYB{r4=yt1s0D zVQ6-9X54%}CM5~OQmmvYAAq-T+}uch|`JMKd@uft-*h4NK>ap9E?PMndYh8*vL(%X(570)RIHYrDD4p(|r zQI}QM__EoJc$wmutn~WOpFVO!k+@a1+ifDvDhn9TWK379kNvK%|CC96DCdKS7FCrO z6VArIG|e3l=I6eaMKZ!9%2J6tJrOfPJ1gH`F$=^WJu)GHe(hc+)*S_xbXr+#5`G!& zo_-YMd>pyvr)+Lt0DE&{*&EZMeP-Az`oc`Q?bSW!Pwx9_UihkYr>s3=6#M$MgiA(x zMo~(ZsZTN09H~*Jj$#AaYJoP6)BMoJ3Mj1;o_&COZBu2+oBFda7JlvQl^7=stu8H2 z>yN|uTgBE@g9HrNZ5%4N+({xWBc4QXYlhN$evh8}@@IptE6IIIlM*T}=I-t8Pm>oz zm0vT!*JPAeghJ!O?o$Obv9$GFlTb6#?-q`o)REyZt*?3-#5aC3j)A?lLR%RTaNiD> zt18hel!n*4;@Q)$UZ)M{Kc!KiKAM>X(4@82`yKi?0S``*Ibq42pA?L3MOe}5gXA|f zVN|Bp_y;!v-?iT}Pd6D2Bv?jcIR7{?ppZbbgE5@6#o%`v33^VI7i3;DGWR6(&}1=1 z6Uj?+^GMTNW|ZjKUAy^&W_gl~$46lAJoxU{&>w^$Nbx}Q=21TA!hPc`&RzO@JArLKX(gyybo>n)X9 zp9nk5st4b6jePalMnQOp=leaB%9&xeUnb!|@<(Fd@!e3?mYx3quX11sxr9rxadJTIr~q9g%#>z1|Ga9SwVp8VkJ z!B#j(#xdWmpYvW;j@bSsYCDarG@*_@5MwjG+As<4cf}HYQ2wrqkXt^&* z`InN=znjnpK&-ue)J#(wDP-O8`yp zYKjYWFS4RD&)C2sj=7y;c-vqdD<^W5jc@ijk`4IBQ`Szf4*`XqM>t(jHI#p*dnBHL z{r-&#*i1|be*ghn%}IhLsSqb>bU;`vSH+pct~ zz>%D?v%0vGa;WM0Mul7Z<8%35teu&3)h8?&HvC@|XSZgU8oU{F`1shEsu?qDk*4kd z``vy82R)|;{t{+DdT+h8I93+R9}28crOFmlLh2Qj@x`qDZ9B6W`RtJzyz<>TiQIt*K^Yu!Ju7qnG<- zC{M#{x~{}S?K{BsMhL?2<1~xiYw4$ZHdfJJUHvcoY#R}Ke$Ta$y2NQmpp;xUG533* z`nMp0*H30`UzJEt#@?H0kXndmu{b1PE0>INsCGOL!Jc=D@#re|1; zrb~*bDAR7o1|QTg9u|2KP44o~bJVfl3L;z^n6zqnLG)(~0pRdPvlNeSs{Dy0_3l`n zv995sen$7oaH7d#OX=1U+AbAM1J(j|tk|5nEMWR|OeB+WFg zfz{ymV&)K>38*>j4D;Y3-mWzErB+zyGIK;v4MZKK; z*M)}%1wYDhNR+$DJ%5a~>!UT*_XA?U#LNFIi3x|A2&sR~fFyX0(|Frpvcl0Z$SF>(32YRM zV&N+Kcq4=uX}~rBv;O1`=*}XrHRf!oGobof>`3cn6xkE4V>_rGii-EGMAS1(@`23S zPVIs!`Oe|mF)N{+gm z1@pU$_%Iy0QEY% z#jl*wTyK-?j8ovY$QgQvpu^loMT*%_1{_NI3WFU+l4EBS-$jDcRFDSDQRKh~wshUh&-S>dvJ?h&a&|WQ(EuM(!wB?6($x=&KZXDDBp`O6Yz(NHWN^vqivb0FAB2F zD+^)X5N<(O>6Gx1bmhy>f#RSSQ1|sDsA{Pf*Wr1OE87immri%Z?SWyS^Wj_LEC*8# zNvbdvtEasBnApf3M_6D9QLAFFZk@*QbNqS;DxZJFEZrvyhXgmdTk_cpz?%=oJ-e`a zKubWT54wEk`NW*{BV2oh8*G$1il6T;DRClw8W#GwE>sgd^LpO$OA}KZm&~5@D4taO zp%FsXw?+Fy#hqNzSit5b-+I?MCGV3A<1)qAx=E-aSmwFMqxo3;;YKCZ9;F$yY6lXK zV(J2BVMjjZC~HMGpvUiT_qovja}4oEJW=GhZhNpDqJYcvIy*C!kwkzPsSEnD<8dL3 z#|d-45_Q=o$e(2FAxh;V_t%CYLT~Fv1S%7b4Q9=gDk_RH-nRLw!wZ_!{8S0(4` zTq2{zZshy5{9jN}uPvr8o^LDHH%>boSCClPYdRA(pgwqQ*?9am`V#ee92K;1Ep&@^ zAa%kl)mMvzlk*`6bAYk+jWOEPt2uGHQadutX1}MN?#>C>Nf8mTJYzTa^8D>vVo2F$ z`N_4Aq|cE9Y5t<7^)J_?{*=7_n#}(Cy$8CEWJbMdiMd0Ds@dar$dxnNLzW89X`4DB+psjc0%wpmvbg`#Q47z+dsq^ zPib8W7HK2lPDOY7D6$E1W)jo(V7f)fvT&gO^S)A3!@B`&2Oedd(nqVSw-?&vhhWM6 z2|pp0e5tUq3imf;YMDIz6Ga{2;E~O^D;ezL=yCM&zX*H(GZc0)7P$iq)|_te4eo#) zUL6Btu~Xk(N^q$AAp!OSUvuVZ&*VUA?uKaIkFgj_uXg19j`qKyd@CyP(J=1MgK*_D z4Ek?AY_Pa#Go^57KiC$&>^1!F5&T8jjAXsReXUCBlj69}5@k<(**`81o$>W`(sw<_ zuK{>gK;wbqW2Dn&&Kmy z%ikQ$IoW>N+>)sj72$BD$Wf;E@kl(uzO+JYO&*r8f)ARjgb&+SXCoKeHTl{l^Xg;o zr_09c(Hven(nr%88PNFZe1IvOrbNL=)^7w8u7z88{Ps%4J)#&poPRtV(~vy5pb@1! zayTC&G+K~#cG?g~rce`iVaXuG$+IAEogzl6&fb~Wp%kc3K^y+Dupn3c#0slXT^jH! zY22NTo=tMa{p?i~m)Hl;`SbGK+NaGOz&UKy>XsABZPR+3{YD&LzvFG7c@q5|#O=?9 zGR0QnqmqwTnFrb9M%O)J&_k~3e!!T-SdBp+r>^sxu|fHSh~*PvvhRRzym!DIAg4bk zZ@75}z&26nOat!_esvc#-&7s4^qee z5gKy`FfH_CUE-zR0n+?P@pKU_PE2+i&&I1d2s7kN-~uX$(Tuo$x^vWV(1|%B$)GGv zUOC;xzlnmC==W!2rp(x;uAIa}H{gM3_;4IVO?*QW)FrcM!=wqaR;Cy&^;}YT4Oxz zzeVDwgOMXD?0uzjks{yVr7uQ*1*j|wFN@gAM8W9sixYSV&6h!@KpHs>mji>!qy=Hx z4joll>I2EY_`6&?oMx-RDv`{mE$enVj5bF2rQ zR)O;B9YAPee`;r2G%sWCZt8hyZQvapzS1m@N`&Aot*A$lxYoMAj^b?X2)n}6Ay;8b zsBLOR1tqg=bT*vibz=K@Njp9wy?d~}a)TXr%Nm+&MbzKv zVx@Ml5E!iXA>>3J#uOlY2V}D#lICCv5Ol_{{*xi3TNv$GQ3cXrd|Gm>hg8WFU;8gi zW!0Btn=71RZa;y)+g52d7?8h=Zugf#k!spaz?0DYf1P>OzY25KyB4Kld;QSz!o7OZ zda$T>Sh_7l)0rVb?ci&!TA?N7Ve})AvkP5@s82W@KiqH!r`%T)g*t6 z=;)qrx~(I{9}L!#lkI?%C;phh|4n$r6%l6C-nLrNDnJw*{;bXan@W&+isF^q198#aGo~B22;pLA zK)I1KX^Ezy+w-9&@Q)}r`3zM>nR8L6%1?w*SN zJ{_a_F911~kpa>w_@{X5I5K&aFDtBhFUe{1)00XDCSYvY^L3}42QW-RF2!se>W5Ei zHQwv#;o(1#*BANLR<3f%@~^Np?k|ZsIC}@Iv7MsqmV>fwRNQWQT7xM+ z+(?GOW_yvI4H|qo4LZPbLX0SGmPDspY3)2AXcB|0yMsvWC!hsaauc)Rp8NRJcw=ewx;k`sYYlkB+YVZnuhlC3rby zeyOpc?e*(9?b$4QuMdq;!&vXpRap@}OI&~-k?;IjxPP;jl;~wmN5PgQ94sVF;R=kL z35|zo6S)~d-7oESe)o7wJ{!GljN?vjzK7UP1?QYUt~lnhF^RdzCY z_UIa0(ARgDVruwJP9RU9^QVh)pKX}d9l%X@gQts#dp8RT78nX>38QSzo7g_r3IBJA z<==n)NbFsLfvtI{35)g%CTv(A2 z80%52c%?{^7r4t^_%Zve5ra{JPnxFnkHB2Bee(QPn-DM8(#exR1k0JgeK+=8ck>Wz8>tpyv zHrJ6>6mwxHTQU7h?pGKvvAGX0&QoTD)AUo6>UO%{HWxE?`T8J08{>U;)|zAKar_+M z>+WscH-!rYL(O@t8!N69zIMwi7=d#D?)u9rHX9;g?q2vcM|t$4bge+OKebdPBKWND1t z;>UhHdxd{L!tkBOJpvVMYYfK7gVHfcMo`Z}PbSXEz@}r}1$$qJPz+IAZ zf=X`E!8~*Ca_j0~2TZhNtR#4avL)<6Z?Rp#ZdX(1pe-T5nTv+IwD=XR_Jy~Hcw(Ha z+OYqJS$;_C7I?k66-B8*upZ)O0|{A2vtC~P5Fvz&)IU5d(+|A*6o^+6xn0ShA-8;` zL}_$82}#d(O16EG*GH00-^gT{MEtgo{8TSMXvXr(oFAVlUIoOawaL|)^Qk&3Nw`HV4t|c6;%+jW4MBd?FFp%jk!ZkWz&c?Wo(<6amY(2t!5bA(%8t|a z>tk$0QsjYui~M{qNh@IQ?K#P?SJRk!Tf>BGY!-9Z{2rn4^L2%32epo(#g62kZ_lrC z93QTHs251B!zV&`7~g)nOF!q<#nJQuTV5?<3l9pW?IQAR1~817rOUx_OkLKu{L$QD z>N^6~Fk2BIF$j}BIN!SS z!GBfPI0sWLj<>lIjIwqjBjQ&acSL!Rnh`d{_swD}^BwqsP8?C5F-2sfc_-77j}SR{ z+}hX;#xlZnscBcPiSOJ2bXz9gL;_2UCHUIHMesR?J3xvMv|^U(TR3m0j#*iDc$ACr zMsZySxL%FSy#p`{!=!eQ@5t*v{Jl^tF0=N*!JmBQ6DLVbHQM5MJNC4jXmcliyH0h z4PeUSOQX;*Qzy!iB0J3Zg~)ifrVt9wlN3_P41ug|W8Gak5X;gV|}x-;Z0fnnD>?PUJtG_#n4v}tPSRF-Na zw<8e;>B<6?T6j-vHFEJJH0LwI$G9~(IxyE!ZPvAf=GG&Th3Mi?aYR(!HPYI}-;V#|CPmH2dJK(jI zq}N+Afodq(+vFwbd$pX$DDi{~KMoWT_b6+9enjKh-$pxmn4nstTe>o!&l|Yv@WKLT zH7g$rPs6s>^MejqDYWF~MVMAI5oDOd3NYxtKY-(BAs%Q-m)N=6Pm)t?y}3S-^uZ^?80DO{_Oe5%5G+RzMTQ8=QOTfI4fmgA7XXr_o zNMIK^ITex1&t0w)-;PV-f~CHDoH|V+YA3R-#wC)7L5=u@-gc$Ox1<-ES3JFU0FHHp z)c1m&l}oH9M7CNr8)NO22!HN~C<(nvue_zgeaaf@HqT`3RDWcq*b7^=-l2^+BDcSh z!1dW|V_*OWS75U-TfGaCS6xQ98yi7V9d>ak`yXV5sos2XgCB!`YU}hMF}0BAQK#%y-AqF+dY`&wD*qK<`m1vLJ9(JJ&PKr>dl)&fkqHn2vTTFqxKAr+b zFxA4Rg(k*^{O|okE)AdT-}IPA;~>k7%5lGO)zdy5EQHQp=_a!$S522QXug;J-n08& zD}^eBemj0&_c}ApI=ENA?%+Oi{*6Ju6ES z5%UpWj%63*SO|IISa>^?aCk6K&qQbQFJK;o?4sh)-`~aXBzG`f(7!e}sa)n}_~ zr!o?2a^Lp2VAu2O*2g*#A~E;*A4KzvjN;}P0;S?7kd7aVO4QuhUk4e{v}p4(lX{B7 z)J`44^1rb&Q7R`cS}EwVnV~wLyit*~t6OW1YJ^d_T!G(=Fh#j^aQGU8y;1xbE5FJk z`&RkDy=sJ}=OI5YN4#jL{L}Xx=k!;^*BgF9CB~3%Q2kL;6d8Z(KoSL;Avto_Pt)J>2r9 z>bq@g_Ssl&OdkPhgVTF_apR+&$5_U3w^^VG9ZE^n^IdB>c+ zLB_(E+%HQrmQQ*=P0X&Z0byAq)*hFyz#~d@nq~_NAm3U}^S`B1Y-Op|**65cP&{vYh%SJn~O(hh^e-fGkTl$ zivy|DO0`P%6ApqUKat5-4Ur-_33K-S458YnpMvmFwKTtNclo{8y(+v)`H_#8x5^FM zCxa9G5mStV^FXUOAegcuMu@nWgScyFwym==6^0^XO^lwd8~xkPvU9Q44dVtWOuPCf z)*t}~m5ZT;1v@gw?>LD!Qyo{VfWLi1rUm%zR-#7_g&09K3_#85YU%5i&FV41y;wps zZLDueJF*W?-da|8iZ3**q9@(iacz~x;80j_X>BQ3SZ-hwQ>Nr#r_S(Wx!2P>^5Y=# zi`6Qc`ao5RWA$aPFQk;j`etQ6>@Cu@-@lKHOrKA0N1=O6A#S0mlXeX8d(2O$x0Io6 zZ%*)qo7??E1-A+}G2Qy;qGuV^2-jnr9ZABf27}_GvW#$l5!94>H&|xjGE1~7_%!z> zOhp8$o6?S|g~E?&**2=kJR(T?uJB0LNni7jvzoJwVR)#B>d1iHE~`+811!#jW0_ci z>I;cL3f6CwoBTA^vdRHZMeVBWu4AW(_L3d2-m{6ROR8n#sfcFz9B}y^e;K*-^Io9( zI;Q+><05Y;s%f!%0`~l|TdVQeu9|#-;RwnI`2%L%4(1O;&WrBi!6RZSr{Y9Vd*zur z0a7Plf+SO;NooRvKy=k1Ua$2A-}JbH<9GGOv>~j@FV!BJGl;*qr^F8b;mz1)Lfa}& zJ*QLxSzPe?+D$s^YS4@0F~`&MMcTwhE1G((&#O<=d^schB`9Uxy&lZH40-DAfasLJ zSEQ=sZU*!5`L0gHYF&lsdgN=GwDS@-IJ2Y0#e3JJNi3y497mk1R7bCZ1a+Q` zG3!?2*186f2_G7|=_}9HoPeK%j$*~;6N7wrFT+y zl$y^3rVKe}S>lA%Ye%;ay$Fn33dCad5Pi94k-qtpOaLE2p`JPU$@5O)#J%|1UGY?& zexeG`SFfr?0GzmTt@=5E(ssfXn{W=I_0H`0`S{DlC)*r4iPeDvoueuuy%zhV^~*FM ztc9ay;hTUdyP+)YqCL(hODUpX``QIJf_R_$v)7p z={zC>tO3{QAF^S?Kg(7w%uCvvd=1{1Mk|WPGDct_ig|@qC`>mfdiogQR2soLzB%N| zf=5cO&Xh@t$bfF3Vjr*k%gOrfA80M11UStq9X^izB%4k}Tu~=(w(1d}dajvri+4Fh z96v;G+i%*vRx?}75_BH9Ynxz-oYuE>n z&sbzKE0$-j&nlM;kr&9VmJfY@wR$G1Zdz<0-3TB`9;?4y0s8*I%N)K2x~A3|hnB)$NnKJ`BO;Ek%AD#Mib5DW zK1Rgq2AR5l>OgDO85yWfaM0eoY>u3r!Nlb#yoV)B!kK?qcN?u!Uz0aGnVle_wyCG; zEKSN*HbU1UDj&|6rp%}obo#&{f!sjMV0_$9HaEu0##%klF2obtXho9e)5h)yU_0?2 zW89ja1oG|*#IP13@OAASJ}y=>~sT{n6z@qBCM z2q~0Jut?-#2oE)^IiRs(TZxQ4ZytM&*9eo~#dkpOGNvEp?Wmm+m2mB?BxDFF++kfe zmo_pR>H1z~Alp7=YjQ}wYSGHq&4n)BH@w5DNSdBzZ-66ZGcQ+#*5+xdo@3s8YNy>sp2|h`b#K*zn-%GZ;)lzCPSXLJH=-`K zQZ;Z+gsXJzmV)B4Le(%nYw)3I3`?u=D(-G!>K6lbJ{6+vgiYK|M`@=MY0l|*(70s)GC@lU5?N~UG)J@+fxJ~GvS ztL()z$*`MU6&n|Wryi)`-P%`u4&**B5@4C$Ox_w5!zNE^31`gmPFji*YeUD7z>xBe z%Xj>bMJ!iDU-V49mhJ}K1d&vJ)!*6wi&pP{4X&AawaYi0V!RVDVwqT1ifZhiVfcj-Z)ZD5>kAKc9c+& zXxxC*|HGJ3V_f*knQ{xF5B6%e{T#(eLVdQjNMMQR>UonbDfWOq79}qVnrV;>eKZ`g z2e$y~&yzZOd{%6i1`~CY)jEv}?<(q`wQRowC=fNy5bwvgc#@2FKs&+}!>qnszCM=( z26_QMDv;-)YXy#52GWz16IC*;C%8mrWXw!oYQ3W9@inC>QkIK zs-6or(IT3j{x_PSlS<}0Ag8|vbal3Zn1e^S;Cp8#*VKQ^H-Qi;0v(|p37cK}R>~5w zy^yYV!T!@h*TE)DmMUeu0Z}M{UMiaiCT|VQd$Vyy2k3l2C$r)K#|Bet(LC)`%-;p; zhs?!@%Ozh<@J|Y*I&~E>-2q?eWb`ue#xbH=SPS>xK%0A3t zk7-lF`QmtA{-4l-5x@N9fjmo0p)%jlWsy>qr(s5HcP{VktT>*+7;+X%k~!Dsrr-`- zNKTC9q6_f_jgQ;!q>qE@Bp|HE#=6waPedc9&*r=7MUe+@9X>W>u)sgE4>JM1;r`g` z%^*W}X5BM6-Oz!fmTi{wJm?$SrlzF@8QgcsEHQ_NTi&E*3J=2Rk-!+)gG+OBCUBDP z?dQVKI5-?pyg5yiS~FHctiZ*E-vP&)e+WTteAGVwu`fn9Vp|W>3k>htxC0c|iu@3b z31+(PtV?{vGb_fk%Mo#xm+?wB=~aBUueUGrW@kg1eZ98Jf)*%5)BK*8rX$t(cch7O(^CDFc)v+@oK^S!n*(e8U9u+Gkqz4^8RpjM!#CrL$oItj4}?c$ z&KXaOInKRP3=(zQGMJjU{WJMP?&UdeJ5wj$Dx+Y zl{&f#vA-f~8h;k%!hV`oePJPVqvxf*Aq#ZL0{k7cl9~NIfW-9|5{Zz*>r}34c=Aph z(s#iSr-jF)U1##i3M46^=82&TpR7L-fw&&&wXoxi6^cw(CLJ)_=F<27Xtrkl5>`X_ z_N{)_uY~-K)YQ=?dWGL|=t9d=w5FXl8ymB3p70Q#C~f|6V;BJ>k3NINRivL< zQs|(8!}XQ?6nK;m=+C+*P`Xx$C9_r<^JGnc>gFcuGJ1q_KbP)3LAHYy@M=bmlTUPJ zLqkxZ=ad!pVTlUfDP(djNI6jqlU>~re_E_>Ck081&2qQD-&JXG8Z)q|h~XpTPLlo` zs1Iep&J#Uygqt>CSizqrR-Nms*Cbs0Q8CgTO)=|AsJvyL*-X!_zI>`!*Na60R?YHf&hut+-coAE zWM^2bG%Gz;?jTUO98vxa{NEI1pci3GEc87RNn(0O76zu4mk6vfLU}P{QNF}>|D1X- z&cE}ZA(G9A@e{YZ-ltbQU(vQVQ4_p+WlN#Y7oJQpPrE;+hK4RrGURM}J9HQg6NA3e zX*@dqGjsS|xUXjNDS`Ls>X*BIbvTUpWm5;5iQW2!1+W%AI&hIv{AH^28+4%WNO&XK zxmVnV(q++LY;)KG{H~M2J{1$wCkXr1IgLLWqPO~soBNf2f7sd*|45HEiYboI;Mf^1 z2GYGjJg$6ZlIr(Z=KUzKGlm9FtvALnI+0&T;BKKKd~Y(-g@pafU-YuNzzdy5%D84` zd(J)Pi7yX*s8}NdTl1VfHHV#If2#m`#_*om12*A2_nFN*0Kfga^aN;s5yWSlV@Q(H zwNtw?smRLMy z*JnAbVjL4tUY`!<<;5p2O}pgL!ft~5xPE2BFI@5!WQ`2o8QvdxbfuR$b|X{1T3^?o z6iCzQ zM96vhsL5Wtt(vnIZxyzU!xzfG?PWPOL@F`6$MS1_d%EG_lK127GQyMn63YDf-}Afh z_*-RJy`&@nH#VoMwxkUkOQnl4l%lWzT+MI03DA9F`J%T+`EoDSbh5En(O zaXPlN*iJ}?fVt6kF8WZ`>J@sA@qy1&G^Hm)Jp*6dBYCAiPj!@z?|+zqVdE87jFonm z??K#qd(?X$&9}%HvK$uTMZ5l>y>QXd=FS(uD(3!N@;_&{Kvs>lH6n2PS8Srq)+Vk3x-lU_@e{i(f!gQW{t zhCV*uo5mHq^{821{KR#N0byf*C0ZURY>bkxJsS3`B$p|bvjYVVn;00p9v0)JBe0#L ztPgWHxA*4!G#N;0lVuK&6QR7?peOu9ZJkqsn`5}=dF^*y(|_qN@~+B2{z0N&_A7;> z4vW|~z@BTO_IW!-ciFCXQ&L<^i#%ufu$jUqMW+F&%l_a$w%2|IBsV^|r?WLx<@g!z z{>i@JoUzul$Kh*j1y~Q8uF#7hJn8RrMFJXgUF~LZ>Iwo62U=|nkF~_~joM5*I#}8J z3zY5usnl{`k4O&prgS+y4z0btf1^0d&WNQbp1;_|kdJf-W0b68jp#-_fm$tg8&nOV zP)NGWf`1YG$K77~mqY8vJxv$u^Nam%??;%#C?>#Mht)xH^5p^;E|%fJQHnHm#OnIg zrvEU#jPBT+ZJWtFnKvci>r^{G@RUEmoJB!|x)`IfRkd)pKFv442V$rja&ZDY~E z?rk8iBI}P*H#KzDFmc8l=yUe)S~0pW`Coq~#0jjp`oM;rp&CN(_*qWXPDKy}k%$51 zpzTpmGuTn^XkiKh{)bhae^|FbZY($G)TUfIET6eO%4bSWp|h<=-mvRsH3P^u>RpXE ziIH&2&>`~=_>_1I;b$7yw624I=Cryr^;I}b$2Qw=x4xXY*>KsYA1P+@o_Tc8$b-v! zg{Qp^5{G%uu()!b$VuA*8?%i|jTbsc9o!F`bAb{kS% z$W-STvmmT8hX>peDgb{28xfXNc2ouqt$i(_G(it$~n3mr0wR@ zj>e=e!xmj#XK%QHO<>RKyr`U?YQX71(t)^WNo<)u6+;Cep}8HrV~2k~|NAoQ$@Jy5 z!%`2W|6-z{B4q_~EfIa60&5Bl8%p`-c$H1uAYQ)UX^sDy$oLO#{U6zgKpKONNrUb~ zJ&dTABkXse9Uaw62)gL=IGqlSW`v<5|y6l=I>Z`ICL335_wc$~3;brKdai z^pG}8VCViF(BIO2M7L=;*X{I^C6gM_9LH=wrv&!?+oe)OMS}g)>9Bs?%DZF!LtLb3 zr;fUq#rN^-Cr3SQNJ7-t8^!B?JV0+)660x!t``G4)#j4T9@l}Dai|Ed9)6Av>di69 z?!}-VVooW+a_#u~hsnN?!~gVQFy5;D>=FjrCyS!~?+I;Bn-)z}=Hb_QHW;Tw!@Ae6 zX>%p!f=7$m$i9GQH|C2WmGeiKV(&sE)mE=?vV~ySG^Jwe*097|`CF*exu5&&Z@l@_ z&VVuBX_V3_ZR@mt2i)`s{;MKJ-6y#w2yd@UiWG@Nv)PPfOyT|Gdljzw_SzLz@798`~*cay9%i z&cD^t;)fR=e5|k>sDsUViKt%SXZXIkw)(i@0%80idUFkzO5(3={-&bn=mBnH8$UnO z-qx(wrY6x@>^`9okcX3Egdl2HY*NQPSukaDh?lBttr}@Bv*}F1&p{YIUA_^#A3??) zD(gESfDHD~wR?CBf~dW$z49`F2L;1nh}g0RQ82+q=|Vg;hnVhaT@}t!0y5vtI>$p~ zRPESsk}>5<0}7(Bv;#KAGG5Y-+x zkfWC?Jme5juR-Mloi_PR9Q2Rg(0IZ)@c(EAs)1G_q@ajW!au@1NAD-Qh8cci8LZ1l zjrG{Am_APOXoom}rHSpll|gq_;eaE98eLmgEeQXOvr-z(#|$~J?93lR zMLz{StaDJluWI(3?eqJt$VU~qMUY)>-eu$G452&;svy%n8{`?Q6|Z20DZS($8ubx# zB@{@ecK{Og96ncJC3JAGKK{I%vF73J_uM_HTNC{n>(yJz3AufzTE(Spu^On`?ZY-* z*fWtlN4juYG7R_=1X~CHimkX*N+>AGe`hQ4Kk}Hkacw^FPziKWC5~uI$sm+F08!;y zw?SkwZxQ{1d-yk}e=yZB@A^1+y(s}grCVtoei+Zezt#B7A1ekEV?m_gFL_7`!&qO! zmpkE#DKuHZbUZvvVvZZ~fGnpXwl#GS?EaJ3HPRi9;ep0mix6eKP$#&M!rwV;m3hgg zf(mYW@xCA0OpkKP?gMhSjcb7<)ba!_SImC?L z<-Yg5pWAzX`}yAI`QE?aH!xSMGuFAzbsWcH*~FKm>ZPr#IOn#E{-bgs9Ck_XazcP) zS0Y|G<|}ma^t#@p+iWW^oQv5sWy=oJGSP&^8+qS z$k2nKAv)t}nrf+2f0!#3j`#H6EmwaGdxZc&hL_?1&p`)9U*?V;g4?=%{0c}RC_qV- zl_dP-dfBMrZw@8%dxUSD5305${Xj^k^F#~%L^|tJLrI5Wl9Z$$c)mdROhK+zA?*== zJY6eHladISg8;;FD?S9u1#5O*P1I%DtMG-ikmcX?7K+Z?n88tit=3~8vH5cnthVy<=5cEVpb?1JxTG#OJYk3bn4umTYOA#I*p9$&HU$XWYS}5klP7n;`X|=$}YJ zlPO-77^N!gn0=6^jnMmd)6?kb?5-H!fHrAK$aJ9C7?oZ@?%euhCKlv{#k(c!RI zz*eQsU0!S>7tfkSrV`r4!9`v6yhkY?!j;>sWrWIC><5)01b3n`&n_1!g81lZ)j?lS zRUqveC$|W%^Rn9$AW`#Y%|@nP3WEM-qQBh*Akef1E}vcHK$d-gWHpr*=e){l|8bm5 zuF+&OhF)UeLf2S%kn)HA=q2y(K0Pso;5}I!IE+;P1iCRN<+v^tg*Lzm3JNdy$d{3wMp_)tQ_SyM0dov={jV zL-l9-jTf7s`Sksy9U<;@+(W_mu?ty|p2BW^$@YPx*41k@EpC?t5(Y;HWhpsvDZHUzG4f`g;)#Me=_7T?_H zf9}}ur9JBvhC=|koQu3q)jHfrT6Dz%)yw>%KI`YX3r0=U6S6>h`YrALElLXDkw80t zgy$bGtNfQ2SJ@6URjdIvJbp80{?@Naj}()@(mN&_cR_9!acJKdQL&QBurY6-m@Ub7 zJ4z_Ae}QD|#jX0gNk)Ou_c*S|S&1at(U$Q!R}i?llXSK+Be(0>6tsHBPXZ)yVDb5T z46nDYGQ&IpAf}rJ90s~mPynxike3!%9_j6LtFNk!KB_n;^ByE0J?Kci6!?1-rCz{_HBWVd)mo!- zt1ae4JEmaa_h|w5!g)?#9YB}?;HL; zG;d*v5NHwA4SR&pbB88T{F!#WsC8kCbRcA()RvEH)s({Tw~m#k1Yj; z^3XCalRwqm3a@v@%ac9&N+qZW0N=I&ZpMVhr}_Pxnq_~X$JwoyUFXgx3yLld*S-t7 zsCM^`O;OqT`S`z=)_ph9tb__LChcl(rundz$M9WPd@8|eE0^H86qMy}6P7}9Q~1SW z;nY&W(AOenqKP*ZMk2V`5TUhXHCc5v3=eXTe+1oer9DMeeRcHR{!;wTuuQ&GKjRh) zdvo^tBHxkBo+l!1N==xp&*6}5ZWIkw+1qp2{(7ra-9F*PDS!4+d&tv#x^2gqoN33! z7wx0{P@To>1m0sZLYpd?Y64U{-m1bFLDQyrW-H$>d%r+0 zFYpeWwSPrN|0$|3ck#8?frt^U^!BK8?wGqZ;Z6%!_P{|E_S%MMQPF6n#O-KGJ!B1A zU`;byKYxJik<3U)&Z@#2t}&IVTNQlWsy+uqdhilv21DQH?z98Oh5B8p&>Db`E_yH5W84ndo5{Q5Tnm#_ z!ce{+t3CJ7n=hG}IXC7_942J~DTI8MfZG;oJsLGT6A+JF)+%(`HF0h8gKUh&{)RD5X!atPFR4UBcVM-5SGp#y_ zuX41al@j84(MMLMX7~?hm_&Vq!H=F%hB+a^s;jr_mh0zoNkrrv)gQ`K?@ul5-xo=i zn55KUPK*?TOa;|WMbCi=zLh+i^Enm#_)8Y(-Lw6yug zf0&%~`o`diixb8B$-LH2-2NL0`1oPAWXl~)m9}So#H^=g5#Xt2!nIi;jVE7Eb#yu| zwK^qxQwAo?uusEF1tTVq1Gvsl0>jqE4!Q@)G(xmW4to}B(>Na%;`H{OX{ODuEm#dD z3$z+@HPpR{TS*_QYvrh7vk$0<|1B=6-w7xGl6&pjX0b)LFb<{tqiwT`0$a6|?Ql2fKlJdfV% zLJ=P+0(AyPzDv+aqD$$vOoKBiLjt*t>jMrz1#aE0yVJfx153uPLr4Q+VFot*5I^NK zGXLH0A^B6hx9-SYwq8W_3cNjymx&)GL~?0bs~Hj6_rqjp{PuF*u8$VwLfMMgf>#F) zUl9av`U>~c(=phg|B$f;YaTds7%}(U#AoHYc=2-$bXV2Kc&6D&M~Gn@dXwYzDbF=h z2$(gDvoc2c(M4rdppeI*PDjK1dZJzYWFq>_jY8a&7#b6eJH-j@zp%aT$yV%|prlQqmdtbV z>@S_sls_W)RfNS$6+04$%7+Wg1wL=BoxMslBEsTNBhniM_#9vo^hE{78?@5K%V4Vl zItIMwAG&pU#){vb@&VXZ+a&IGO9UH~q0U-TOP-ELNmf9918u2e!p2F-{jeau^X<$| zerCvz?FSh}yXrSxesc>?w?BK>h0lbvN?}$X1(?AeYt=i{k8nRmBpbF@hPS*szKF>cr&F&`mn0PG`E zl*Z&{RVqt6(wF##JS~3h+px$e8`wR!d2v*j<`tf+o2(mzE zhkI1rpkw}1wUtfB8AY9w^O-IaWf|&_u?%GvOTXuXNg4ZBP7QNX1RuLAx=4FS!mVA&)qzJ@Wtzj z$m$BsZgGA(Pf_hNJ@=kQ=a47;aGrT>?Nw5Xbkl8CFs-2uDiWEm+e-BJ671m*#khDd znb%Nq!Oh5nn=p@g%SYH%-{C>l*=_|Z(~0JkoTDBbo+W$FJ>$3ZUMZYptJ0Znz-w>W z>rXzW8Hgl%7R^{)U@(lmpGf`0bBvTy{Rt^V!aUVh2#WRa5!|Oh#Gdb>_;lQo{DlCA* zsK+x*VJ)6IOr*Cd>*|-g$VajAmHo=P&&lYu!E9p46BHEOt<%IoRQ%eDS%>L;< zQP7*z4@-L4tRw3Upn^Me%N;KoO9i7<*R1f`mNq=InF}~_uKA^Wp2d^eLXJv~Cir?E zTrPW=#bzNSrn15f@B^%S4RF`e3swzwF&WKyOGqp)6+Fj>XmaSyLFf-}-9t1`I8Bax~)S3(PMR3T%L|j)IjE(0{wV|L?6%O%bP>> z4{q;k_n%qYPr!V$l88>}-8177XJ8#0lvMg7{l;76(>E@l)|lO~-AL9g8?_!*}7L zv`a0N&FkG$xe8eNGI42RQqjbYJJYri@zcov|q6odOQh%gDGVF$=B|=fSrwB1w)3X~|bIA&UuElds z3!tS!LzB)XwM`bX;j75My;Lpaymo$~5wQIB zmVIuUljY-l;QSWH#+hFjlH!}mf^5s(e3(;@MhbnU?tJzE2~43Mhtqan469{?)uvfI zcV9)QeOpmZg-#urSQbez;Kh6x9&9?+Qh-z>wP5$eT%WzBK27f4&WNB$$|(mQL)2@?wBI(`Y$K|T7`tREC
    eo#<-jP=PP+d-n8xDtizJ8{$yvpc3iibF@uQuYhMU`J8Vld0s6<_`~bgH#FE?kU` z%snE~)b1?zLQDH@8gNm!iu!XqJPB_Cv$VLzD|P7Ws6HV%W_YI=a`KMgM9@A>fHYQK z20D%RlV6*u&Btr4Wtgt=X!-1{C(q)d#l!>NXKGSVxU+mzX!W=cl2dUkgfj`9ny#WCMSeg z_ft|0VbW7Ui{iY93mSKI33@HDryAtjLPmr>W*N(xU7IfUU3F9O;PBcZ+Qx|I(qfWS zELo`U%!o@ZN^3yXI#kJ;h_NsF5B-6HW<&ZG)#!-aHatC&E3^M}id8&x|B*cFU~}bj z`2-CEdOJ)B66qE3XT9H1-zSaaI93U`=QpdGmbm7f^pZpsgG~9DC@3*kdo)liQO!Qe-F@Rhc(# z;H?Kp0QiK_Yd+~%%i9Q7h_d!Y>>A|yNe%AJhGhnnx@6D*g&*=@7G;i#IlI*0yTK`o zHe{MZPYTRBJMnS(u`zAOrXZ2Qwni};*S?YfhU)c$A22`}hROy;`V&cD9B2ebdDg-0 z3^txAu1Iss14m=LHw&(R-Udkl+wnd*zBUUw8EmB3^wbx90bryFYa}({ zBp0}Z0+OQ7YiC=q)m&kG1;OSEb0HEBY(0RaYwI}mc?lx)*qW757U{cw-yCV3( zh|z5=0~)=TC07|yVFI^)-ZJYKB7Tw|=ovlx-Av<5vNY-A)2++J5~(#}0Lu!+xW^9a zW`@K*2d0#J4_8w~HYfvt=25{Nr=169!7eRelm>TT%as>&(bpPhcELtII)5ix}snG`FRW=EVSz;7=CsWO`oPhxB*m*jG+FEb^*_V6dClKK~wVz!Q5y(l=Cr&F`()p%OqULaqntzBEe8pO+y zIdpn#ut2Rw|IIL-5xIfCV4b(Nu2HKFF&D!@QCU(|^26PozwbV=V{gar;sKj6aj{T$ zxo?14WVFen-P+=+1<+dDY#g3U`9kM5JiVjrq$M{a59T(lII(EbP;Edvq#u)c622eP zWPhB*pBgxH=>9qR<@5!aRQV+Lj@9{ho|^bbi%X?__}FHmZ`!KuMaiway!%7h5>{$i zt5v1wZa-wP118I7JhM0oM?fsmNc>V&RPE8#+C2~mt?AhSk7jtxqWwfNZC>w5KgBHv zAP+SN=@y)2gm@NwZqWl{DXk|B=5h>m4kEueLxL+fI`+Z92x<&b5Z3)>Xkq`OvZ8e`s&0CfdSC?$cw66AZkC1l5Ow<4uGZ?3r&fYnD|0UL0oW`H@%6a(R}Jei$O$|N&EnA_?2 z%g!^j!XhKhUTBR1n+Q5c*)LQ3BWfnzPQa82ogYM4tkwE&rM(QHBQdl0&RUp=i(LC+ zQh)L-3ky%EvYGBzD^ujVy0mGR%4A9P?II6YPnp`CA4*o7_dt#cIbC<1;w(&q%$A|D zt}@waqPvtG^k7Y~^5F-Hy!6F&;_YMW!siPHEr-QV?DoTH=Q`;7brathpRMEclhyJX zmkmUu?#?&~G|3wf>{hr6h=x!#R!OKmR;{^}?1@-$`821YKC;aR zPrmmsBMoUI_O`!0e(LH?O$;BIjzogS2g7^&ABqj0s=w?T5-{c@PdU@m2q|`CwQn2D zovqRy_9X2Qs(kmwh?dhBf=5O89dTuyv6wK(w#61~Q2@T-&Mh)S_j^sm5vVtZ*?WP0 zl`7r~uy`nVyWT%zoz@wv?xlY}mdY%T*Sm2cm^y=U>(CE$8 zuY9#l2#Yi&8TMTEKHjl=Zy)3I)Pj^dEzsl=%=O_#B7M9xJ6)+;Q^n?&o1|7LLb_iH zq+zZ2NFKcTW(b!W9T-;kbR8+ zJe4I7)_(wo)L1pt$Is=_zfByi)znC!$UkfMeRE&l!rcmDM0i-d%UzvM7x9)~U3Ed} zehBjm87NNl8pc+v>;zS_td4QOlcA||E)bsix;`n%#5>Xqz4mG6;V?748+$QQhd=M} zKm7SWAX2$9hE`i}F9*48mb4N@@8Bs~b`%bh&qqae8bQl+3Lj=^GBhw$*TN>~ymh$d z=fAWq5Lw$!e|sTDsaUZ840>;FTRF^hSZ2f|N$P-YE6lf5O=&q*K37W{7N~#X6kI}; zEFap@=Zu{lcx6A>9@CL!rQh8`Gu*EIBt6fP=Sh;A6u3aJWx?|BwtXFY$d2!TBVrOK zXC)Oyee9t6Hko!)^Qn$ZgoYdy(mSh<&JHN8T*zCEPzc)I((~~<)?~2afmE}*3vQnk zhiB7g2RNAWn6gmqeYhI&CV4mavy(aP$w51_3`nSDU1ryX1aKd zZOyg!S~YHMZaKwmuY&|R=%DjC??IoIa2(9j-=rS)(X-B-8WFrq*(7ozfImfvnkWGv z2WvceV8*~|xEyEP~(U1vN(UQ@S*&nU`K6|=WKNZ8H7+f)LjQW zTn&$w>kFzyCKg8C!FCI1SSfkj`MAy26}8Xto~!M`ysi-ECj_uqjy|v$ae2{kd81aE z_#~|eY-Lwxn;(6){jAsN!)Ao1+#r&xO>k6S_ML>-fq%__c*7fvpGa!GSEZ4fCsUED z5F!|x=aY73{;d{w*X$s}T*=4yK)(*J7o9Qbpj3!DL$4=aMmS<)f>Ay>W1H`c?zNRQ zS(p_$77o+lWu#$&p60@0+m?div!vO&@);O&3W{Hwu)P?i;c6<17?d%f25#T^QA(uH z`bcW(- zbuG2u9<@;+P=sEuvk2amBprEBXS4>~(dX{s_g2bVJ(;C&dqP)bG!H0?0WDgw`>p_i z37-34!d$rvn15&mrMX1gZieB`D#s>QMRa0YKwMcfnV5gBBe&WfrgjH4R=Z z5tNT@JMFzJXL-G@6>I4?_&^t{Eey0S^S4J)0Zm$@Kz{p> z4Pcfm%IG(Zu{+3pV5sq7hz{?x%U_UH6I4X?ej|)N)~KhbTsOwmR3d$NsWbm_tk=_f zcw2Y-6N@#~SoQ=r>U@D^O!B~2w~P-~kvlXPl2h*+!_pB_C6b5+(y8L~ z$3KyJYw{t}v~NK8{PA0n-mM#&07Sd=ePhVx9k69g)0GFeC!A2PTuhZgfeIqTmQT15 zj(JAFn&j5zW}~%N-K_KY8U^xw`g$rg^3LzDbsylzQNM!R<0#ZD7R7h*wYtwDl*%K& z-Q1vKtQKeV-~eYKem&;7@V;G@Bb-LqS+Z$7^@pi3e8I2S4&}4$5itr(oxyT?VhAw< zRfJ;#-QSgNTxg*x#rw4j8&jAYU4t$jvw=>0?~Z-iVheNTNm>;{!zzr;nn1=A5A5Fi z6RAOfqEwV4`asHRlJXxCgzG<%eseMN-!5I+UcZdHFW|r33CykuMDzy-Qw#H*Z;_G{ z7^67tkEK4HOpGS120%^yS5rmh6WUJuFUx}lN;8nqK_^U8zg;RDI#rc#!J9@RPZs-N zaj&9{?*Y!auncWvdlM=8g`3sij0~Emd*%A*#(AAD zHdfS)by0Gty_M4Yqo2b7WggS>XA-v4_F+{`(+5tXEIN1^b}ZPvsS5#*Y=ypw$tv@&(!lu-aa*xtu2Sz%YU^4S31^3s-)gYae`Z1 zUww+c0_Yj7Az0N_g! zL=jl;fF`!+ZypEGtD^w*=f69mWlwD0WtX;E6vrFl0DX)hnH9_7 zFh|$(RAfY#UxB{BoECfW@OM!u5~G9M?9u9cE(Ox@*L8gnOV52G)rDtnV`DrAv9`Y<-m?lZ@pOy7kI050MH8Z0W#FM^T`uRYkzhQ29cXwBc%b`HpG zb_cP0q=Q3*+q%`fjZ!|!tbhwj4LTLz{-wKVjQUg(ERG;W8tH2wv)98!r7b@RShC5< zy6-J9Mw5WrI4jHBRP%MBWb(9&j2LQi(Qk1(ej-&JcOSFeESOQC>-&|Lo`#s&OnwD? z^RE`-jxFU4<<{cLRYdPkH)@U8t;!d3zOPPvyL_l&Rk|Q=v3mGw3w@SLy|OnvY}Gdg zcU$Wl>V)+L1Z9`#YMi7r&s!+w5!!m}5IO<+*sW2?Yk@+Li-F{`MVh=CA?%bS< z^<@9TJ}0rUiBJ!4v;;5w>@?Du0JZ#lt#m-2d`dBh>!E_iL5%jpIJOGe?@7%Tqwq)0!#RBq2F2Ds91P@<-Q+ShXwn0XTcRBqAu>Q5w<^jBxtHVBzT z5TYR+p41`&x_EY8vfSQ+y?T;n1Eqy~2EskqFG@3isq60_i0-dKN*oEfMF^zy+RK|| z)-cZ*#>cQZ<#*XjMTUG#HQO6d^XcoiQE}ZCi(Bz5Q1nJhi+QhA8Uq59r=Gn%k^xC0 zO-TdEdRU+}g!qvMAqToIP8K?BQjlTFVallz=D|APP2A&Y-GK6Se|)Y{@gW#MMZ3+N zzCAtaHf#(>&;Vp)r?E zU~8*2CM1P{*j~USUlp+xNWPK9b|ow)%+fo7DE4N)-LV# zf$Jxm{r)t6o3{y7RVS>@!`b$gn>>H<+$%xvUi7q3-PF8D9&MI$&)=VRlR64!{WMih zH^%WRUkCrn*HM@~UvD&|fSCBi!bUY_{Qovr>)mJg_-cCGh`keX%4P{*s$1g}PR90T+ld>mWQ&~Lj~2Bb zuhpZZX(EKXYV(Ov=wcuOby7%x%tpHa=u>8WsAR=0lWuF`(=)v*-m)J(Ab(mHNeLT5 z+{eivSjwjvHjnMwCZ~1l!xrHpyZ>;q7_cJ$|AK=3AH%%IqyS3fo;Wb0g%dr}EwD~}=phzGdLD0Ku=ZOaH6QSBkqbeV(Zn$87 zvS1ni6KQTL>)=$g?a>edXG6)Q@N{NRpAI@4g5BePKIndsaBqr!g8enbEYzI#{f7_M zeS(4fH%jPLLc!#waumu0##Q7D6glM%&}}|E@)3n!2F0wll$F^-#%F0+NT(s5w3gD zoY?qVAs&d7pgt-&p24@xFY>yyAjS&tDyfgb5VF^A$%gbk=|Qvr6V`|n63_wt<9LTM zK3MeWqD7e7L!VgvNL}>KCh2tfGXmr8Hh>&-J_yJQFjxcUnYeFGREL6V_}ToNt8g&x z&NPZriv-ErbO(zd?ZWHRL3g#JH^&Aqo5Ecw=hsPY@>(1w!`H$Y^o%6p$<*=M&zufoYOz z+`;zKVNU^Fwl$_m#W|UV@YPwB>2=0X*1UHyi{dsc4;EZQh)^SSRqmvt8KJm-EVS>Y z6$-PLH1{iHSkLoo;4*DX3>+c_nP59b+?#^+v7@YLk+4GCRV8y?5zG;`p$x;U{V{!k z^+VO&X;qPBu^)LTyRRcmYDRY_G?;&(@&Vj-It5U%-X{XxNip8BU^62;xQFiubRkfJ z(97r}$+dXIJQOW?_49SfM7X5#`Dt{kQy_esq$iVfmpSDmu4JIiwbb%aGw)6HlDiaD zhvcK{_hAv8?2L;^jRLRFz=~p(>W)(uT5|{BAVdhZ&sb~E5+OKZlQSw7pg-Z z6)*LpeH(LZa-m+AR|{6L&!m}_)7u(097Qjd|o>wmDC`tTp)>G;QZ68e|z ziQqsvf-e*v>%_P~CKV|Y9La;vO_U%HOowcD`n+gmV_|`9@p$Fn+({K17RQ;a_#2H3 zjW0X@*_e9&wAnxJL2xg{`NLjO1QZOh0&Cm&$9^%Uxm5;MM|gGpf$*#EH_UIBPs+K< zuA$SSC(1yT()m{8acx7QO$}TEle%Tn^h_vG0~hPzX%`7akinPRimyX0Z!?UCfJJeE z&{_-hVctT;4$XTA?OQKmUm+3P;_$mY5Rn0EWFVWJmunAj3Q(QFICt5K4o}s|9bS2d zRUhSc?A^Y(1&XEY-ZQeJ^c+^|9dmA>_MJmmhE3YZb5N{?j;>i4@ow6)@`+wS+F0Dk zBS5>YN-w6|5i6UJmMY@B@Z6N}$yYHdUu5)r?qL^>p;*^^n2M^;`(^WrVW@Vm91^5{ zK6baZk!h=08M`ZPDs@eem_BX&O;q+tdrlhI;yIlXx{8Pf3OwCykWj)SR*(tz)h1;1 zorS^KdMRWj6G$0BS%@cS0H;AjU>os(O%yoqIO~vhp&KU31oS4bg}*#(Qh4Q z8Xf(LPJTE7s1FUPsz&?YXhek#N8x_MHUl$$Z|u6Y)DYlly}y2P^<;|Z3+CA8=r;9wY9-2X^{QwyXN}~SYW(f~Mfo0Lfwv;IIlJ0F zZv3cq%$Hv$5@rjrZJ8ipT`<?^-kvr@A1vsGExO1{M$t$yWcrJ?augcoB zEGUOGGSp#EKs5QFXE1gqh1^Tb%W-!km&F&Tb#)F2B0|8DgLgc(Yb~CNDl3S*J8((q zb2)e%l<;miUU1^H?XW4#>#LXJ?1b9Vma4YD#nP!tPAh`Yz3g|BArLTWD^eTEB0x8zHIzN@}X7*)_;?o``GM(&B|avwV>3jx*_twEzD_@ zD*|aiifxia@Xe;k@Sz646#J_Y3xx&L-e#5#razLnq9mWB4hBvE%r5-uOpif0?7`IJ zyTAG^f;w6&BdCkUub+>R>V~x9-*TYsjenUpyG>4$0DF*?X}!%7Jih;k@bhh3ptB1lL97>2ZYid6dOtalEh#0{ZovgXV#8tT`sn*b&i9IUD!YHADiSsrkg)F^8x6wQ?HO3w1%~* zsq>yOE zK^kef<3<Mo7r?!jOWvSDkoLiTBJ@oxTSbOE zVG(Mp51`kjc=RUF4wUh0wk!feak}MY_4i**Q`JAks(_db&6AOb`M0$lzdr5&&b6^I zDlZc{IVb=_W4q$V%jK-U*2Ukq7V3KFD^tpB_P4-kpS}R0yL{=q$qg>)6B6CaiU;6Qo6lX;}*hXYwU>JE{aQu*k`E<1(o zCBO?RNVoZFs_JuI8vD1KAioi;byVwd$!V><6VHX2h=3hwl_orevTYW-AxpL?Zqx+; zlLc_i)j#t~elxv=-HARp=@UCnMv=mHvkDd$2X6e+-UEqK&)NJJWol2gUSws4Fe&o& zL#ZFGwuq%)ueAePhZrY!pra}Fh%!_Gn%ROMb#{PQt+y{k7q`1H5`Eq?T_%|0O6kFz zfY?&f$j(k?2==|da@YTV{ElPy6AAH(#}AML1GQiHAWraPIJtjua<{Hp2axm9n?GUtDZ)WT<&UV zFZ(!9K$H+8O};M24_s}8OMXOg1;eMFO0Qm&0t$g!WkSfSO^XaY3IdSW`#!9eR+}zq z_-;{qlEznkTe0#d=z$KN%8q1Y{P;UBM|ux!X|MYiJqA0EMYg%e0Uo~U9pi~ZrsndS z#FBK|C%H!k^q6~UyoQRxjZR%oN}W;SrfIxX9T_;5yBUs z*<4K;uhNh7|Cr1KPK>4>h3wD+O%G3C631i<1vs0`klVcoIj4u5dEt)1)ry$QdfGB{ zb;W54#kOqQ76zNTI541?zX3tKDN%;q-u3}m*khqu?6BV@+rOJNUoW;B{A~O?C%H)5oYOQ z{%;h&+EryWHTiZ=$tNeG+ENX0ls@@Byq1m+Z;=@fB#&y=Vr>6?!D*O(q=(>yc|b0QQbP~m!U5i0KdZZ|ANBX)L4 zZdLJ-EIsxmSwN>ZZlgQsY_8>NSQzQroyKZTb9hcTa_@$!V%Cg;o@NU{#UTE`G5Y zdoi@vo2wqxA8>825)df*{<@6bjwPh?ZKThWP8ZqmN0lqR-f_Occ33`%*=aq;2apxt z@@?ZT$>VcN2!R6UzkJAV&58i>LEoO#5=>9 zjOWYIeDf12ZE|10vAni2Ldbz|2&U0tk=I_jgWPkkajy1;)}go<*yyBAF^k}W)Xr8R z&!3h;+DR{!%&&%Px7g#K*sL2U<|EX0>v(qUA4;!NO((cywU0^hCsFl__g7vJOeSp_ z_w&kFNPQ%A4lAo3p_+UL{xa>u>3IT>?q2Pb+!2d^R=Ut||eXd{ls;x)0 z4E{1MKtm%z3gDK;wZbpte+`0dIth1ZlL1yoj>P~+Q%lqfE-gx3*?##aAZHdM%IhvC zed5WXrqw12sde+*7$HTnXR73duNtFViQh@t>kCd*gCj8wu6g5+KO$ajD*h0Qdn1}y z8Z9KAA+4JOFHEo+^b1Q@y;kl&tWY1B{%^yZ;cwxuN(X%th#zkyO@n@A*Kp#Rfe|8dw7j|p2CyGJQ7b(;b1j{Lunn@eUKEPET5i_SfL3ZyQ>#u6oD*SUk-jq2s8c&NttLau;>p<{eQLYG zYZ`Pc>BHz4!E!AT{q>46iNBgxcQ~Y`gye{e$oF)i?(zrctG8#~5u6ZliO_?$>UkC> zt~eQnHtyE*U+aTS!Yjgxup&n$E}}s}Tu+tgoGa&S;G1;>Ha@co3E`QLqE{hT8al zTHOri#1&FXnmu4BEw;W#d=G=_6eFjMXS`-mqod@~_0gg>`#l~$;~UWKSfjflnRy+P zP+!{Ouv#20D5y8zA-nch=F9xgzMB`L7hJ^+PKTCO@C2_i?-$8##9xf})kJVF3^Lyp z`7DzRkB|ac>AwAl65eG z7QV8%X4uF>IsXRzwq{uaZ$%F*U3;P)Lf>2`?_UwY+t=|?0Fic=Dnl9zJ6oCpcP^bTo4hjAb|BpOm0D$wPM<53z5a6sf zR&ZA7zgKVntqRSbpcC#y~V4DX8!j-BJ=4=`Q{%az-2%srs=^;l!(nyU=vgF&T=eW+}y79 z#BUIoPh~b`?x@Zn!+6&kYQWMGG`N+oR$;);| z5!qEm%bt_SPxQ?v0kL*8sPZ4eaD=7YP?sNj2#xgK(=#pr?FHsdl{Y_xqDzi9V`j__ zB`|1ySeKc9L`&sqp>3F0Fy1_$yi4H@x_(wmG%lAUrzcs*r_$z3_nuF|aZ2Z?bx@ju0lg znb!H|H~zd{!N7_|@IP*Q(fsRr0NyOZYyp?I;)E8B?DNx0!s*>K@|tp_!~6bLVNRJ* z7)Z+A&!3t9vs4aPn`8dcFZ-Sab#3ajCoP3HBqOzZCB)+ zL5qMEh=J?+F^(38suzRd90B`HgsApcFGd*Tw$2PO3yiOF!b^0ZF^R#oeAej&oNSBB zdj|!@IU4-MN|7+FU*0thzX7ek8GjO(0_$k!oORJYTIobm1@T=(%UqNU!^F3M0)+Oo zpxGUY@^Zq1#M9!QE~Z{)_s)0nSD&uV3CAGqLnl{PY^}#5z}~0&m;Z%}0zD-3WQ0p( ze@^K4%J?nnYYq9|d>Sm#)1yz;Vxrb63+$a>^{ZY8c~vKUF*mmOhOXJmxlS23J+m97 z`BihoY^o$HJBWgqGdk++2C1W1`>%Cdk>0GND~T2XX_XX`tw6f3(dR0u2zL!ZULSfa z+ef1QZdQmqqg=_YWxcL@PMpt4Zg4IJ7d&K77T9ad;vp`pd~n;hpBs^T0GWV4MiKO9 z4C-n8#ifDHmvdligCij55eS>qen0ZB*DAl0TP3l|85Ix06;utO?kQ}AX=MF2Y+H9a z!05IlcXSifcYUkC8vgSAT7r*ndqLSE;FIW|xBEW|c_32z|MjGb=sqSaq7LY*dZ52N zX&z(eGENne*O=c#$M`9i-48C3To(qukm{NB8pqmrrP=N$-1iDBzA=tEI$@gROVR~q zKCX#e`3Tm|$gTU8YQ3M8hSJ2n!{FL`ijq7j^vudNUJ3}pgf!Gx+jg{@knx?n0qSdN`;7>K%d&UvZmcpNA0LdtmjN9s z^5J?)l2s6fyz449!^TUD_(@`Nw~)DiUG30|%pV9x@am6sgb)o-;(+sk@BjK1P>f;- z{loJ&PPjUOehnzjZRzo-c{`*^2Do*KCJfIHd*zRSq^K!Za`K(5>9Kr<7z z!0RNICrJ*Yow<8UT6A`!v@x%GvtTVuih@1MXYke5J7fT96)|3&mMt(6T4iRk0Yr|_ zIO3YXl><2Hg9}(lN!fmLj%?k&?SQ~gL;*@>))4gS?kAE9us%mJ1A1)Lb_3-tEh5`` zLdx(o#l^fu(RN|x6qqA{+3Ij$uoU9~2Y~W2Eoe>?b}->xB5NZCfM+BChU`#f)Zr%* z*^V}AS{yhZ=DAWGY#5$X`K1EsopG?(@Tv?;6M9|V!Z|L3==E0mRc}^bB*HZhe##tv z*8EF>V~Ol1()WK{i1PR6i`Vm;A3t)^e{FC!D0&?I)`kn4QAvfZdfyduZ%GpyXe;1= zlmC7V)d|O8(wzdZqpOIIaMfZPYA|+j@2zpPQq=2D_juiuEm)kTho0I4I(VwF`}8~i?i!(WR0!8)7cFyWF* znf7>fB9}QU*OotWQ(e4_9h5mX&XEE-R-QTW#56EF5Abk!<*jEE{ z$ac*!{LTQukhtw~i-0uq0zf99{|}uLItKKTt^kt92^s_;sCJxqYpM4uC;#<3kudcn z(A{MVPznJpp3d>#zax{neteBMSwT|APY2Eigk z)q6~6cCEmFh#XdV*HJkTqgesEG!AY&BRD74at_AYHGgF;BXbVMunpN)=h~M&*t7H6 zVf91?Z+IS`y5w=+LMB#1b>R4ez3ZS91Y*_RCk8amIbzJHGfBhdGFQ-TOv!u0*XJHc z@M)567k5{+<>(4orUIDJRLJ4@##u z?`w=}&?=dVcJjS;(TH66MIsmvq+^oj3G3?fIdA%-Uo#|#Dd=L&#-#Q&`{D0Hnqq>@ zKPbY5391;F2DP4RBykmu4>~4OLG&4|u<(%1aY>PP#~8dn`)ZBj4Zls-${d0CHi5Y) zszE<@^n7oW{qlvU^!Bu5C@j%cE_(P+fF$y)8k%M~pE9Igexv40(h~q4VPf!BP(HG6 zww)zLFPfz`qnXl=5;?tD-ft>yMv0u)xhQ-ql|M`cc}Rhrqk7x>Gd|`4Glu_W#!6A( z($0NoV>|Sdht%dzs8NXHKHvPRl( ztoWVyx3B&u!?(yvEe?w|RL_~lR-@wQUl-u~^NtK{531;4#R6wJ&e24NqSZ&x)qNn+ zLyfX`Z7`ZTWEvFO>P+?2R?x;@>^?#L zi%+78V1H_euJX^Tk+W`{zDD^nvUcbC-fgEBzMBc%^@LNuEJeAzXmEbXUBJvV^DMZb zAZqO_sIYY_Hw0;&#JV^z!KJ5%sk=PZl>$rcPK17U?}o+qmgTz1GBTFTIoYyz6+CvL zJh4ors{2ZZ1w+F$axHc5HF_Xs;;tl}YM&|bJtu+YWwb-nAR+E@L=h@}C}uU8EQ=qB zCbVSWP3m#8Qjc<8p&|)ebQ|YtL}sBo8)=Ua&c|&M#|_NRSEDZT_U0SJj5a@%d*uWT z4Wo{Wjqo$8!(DY04Q}EO!FJ@=Jyc zI!p1^GI&iRH<6)a4>!Xq7;;9xrXw+QPVH$h)3)y>W zrSG0{`(Du9&9n2yzqAus;6MNBdQoNN5h|@m)ibjmAJu^z+V|g7_nzKr`V=dH2r!># z>0Eh>{Ej0u`tjGC1!Ibepdnoc&opW4(8b3kFPo9$aI&V0odFy9w91$&1hav)@b#Dr zFW6)5wnB6sbNM@Wj^t%5cb{6&yOd|9cQ)MG+-BV*Ra8ych#8S*T2-z*9`QJ&Y07*b zb!J{NS0vWU@x<5bXfl_1v{@_-bJ_uBKnv?4ZD}_@|8w9-j)}?OTv?=%@twjOK7#G{ zRWIJ8SLHdTz`~nAk}vU^4@ayx*H|w#t8H6$zM0bjp4dVw9mm7wWlhk)(9o;k@b-Py z&%6e0ZOHlOP}`8YC?c(-t(>)qF|x|6RqGk)dIzi^=sf!Mqt}c)_x!ZZkhoJ|o}gs( zJTcZb&B{><-aYDZ5)VcU+IU0tdF?@SGMqWxs_K@e)x}l*{{sA#{{npI_L4ieRD~Ig ziOV>Q17QcWsn31-B|v=O-w>j|e@YFDx~0p@4IB33-+QzJuR7?(EIBneD}~ussL=k@ zf|>i=$sN6v{o<<3>2WHayI8qkSJCoE%DEHHr?|Z$r9DcKU=~=0C4Qzi&wXeNBx8>_f0TG6fBTSIikzXbRt1GN2OTYlP*C*TT9U=e zrGRh*#xmgm1pR^D?O}m?43(qpTmL$3k-%Mf1L@4>^A>76hF{V!n}D?o;{c6g)IX)i zwrr#b1zgVrbP+H{XDuX%b!qm`uX-MFe_HK*l8#L;VzZSynd|+DmXSH5veb;h`7#Yl zSK3H^j-Bic^1^JlIbK6^7J-I-B|TcHE9R!*@-5sSjor};yZICgW7tqD5z!d?HFFX8 zEVPFcslU!;FQTV%%^Y+cY5prz|8G6>FVTNel_8bVlHSYtDcbbI6W`ChW_P2gQqb4xHS&qme1-oRvULc229S+|yQ z4HbX-je7$84$8pT6{i-qdFRwG619L;vAa5v#~ojC78;4yZ(jRUrBavvi)57N(oU4V z1T_ynAmgIS|LxLHu(#ksPgn6BysmD9fF{C*cj?t#oEm<{2v-~Zvq zYeKL|?{hlUF4R=ZMQrM`tydEja-x*ujKn>%qacxKwClU^!K?=S=t&_EDw#b8m>-%0 zkUWk&$=*+Cz4`?e^!;`4gA0Z&LKpy#+P@PAn+AwoNH^epsu=jrNh>5?No}9ZfkR;| zIOwg@^}(^B@!ix(%lzzY1`NfdEVmcyo%!r=vr+at& zZ!imzbV&Dv=p)hwX`aNC$%ZD?l!op!xhpH0y{8cW5tISuYUkCyajDu@TKhq>3-#$5D999fwFr6Vwf8ZE+7WU04o~;T< z8a6gi#@eRJcD>wwaN(piTg7V?G>bC&xri~sSSxpIWbUOTDqgA2wjrv7%IT;a#J%yg z??vVr9S$AxGr98>=SQVOQ9lCyU9ncwAZ!*^Xr5G^-1EUCZsTweB0~h2r3@|bxDN`L zKC?0TQgnsPs{Iq*eq`#|H=j0#c_kKti&)Wh9`bZM2o>8mR=>|sZo`6mcp5Ci zJoi@Ttvm?EXstdOk^6Z0bA;KKdEg>dmIXRqp1roDs(im^Nq{hKisO>nyVfeYbZxNX z%8~jr>eAto(2x2lK}BIqUF$WPTWYEy^lCUtH(2XgFCCN6E5|oG7&6&sf@%(mGd0y} zCnG-NTtZ0hyu%!>mCIiNn^xp~s!+m?i^}_UhcvMJ!ynI11@T}pj=5MPDRDp!^ywt- zU#o^z22PwU^^n>jv5}yu%3D95c)ACU8wAPx1M>alnPd@mUG zj8v`ByBzU62RWs6hSw|(@kc|*k%QD$_JLC1+9#bcPfWJErd_S&BNT>S1^;J_2L04~mEmi$XSB^tDC z8T!hWnfMUv1O^KoBAtXkZPp;PZ|6~sYQu%22`qhP74Pb8`W;T1pZc8gR#IZct3#lz zWQEqnd+bh$H%hzjad{}i?L^|~lj>6xB$9OFI9=_x6E@nlrr**l1rK?df5n6K|HX!V z4_QCtuu=k-eaQO(LU!1{rVjQm*J3we1^=#2yZ@&hYg4}WrO6^($X;#g(`<$8k8MEpRM46xti@`;#5zNrDY;SjO^STk*jB3QNN%F=Tp9V%v@M8_+3OySsxHgKBR_omK z5?lUAUsK@ZRNkO+b>7Zb`c219n9#z_tjf_Dar~A35E^BdLJ2<-&PSi;lhv<|7Ux`0 z6c3#!UFZIf2tUvE|AfSJ9mHOf!@vqTh?cVdc@O*uG5S7hpgprQ&vI~AH~0@UYB72u z)~zf(yUi4afU&-HRwVv>4L{XplUt{CKI7)uQ!yk@ITi*MQr@JWb2XS`FYDro#Z76o+)45hOPo+sK%Y$Q@Lhv=PoH%jRrU za{#%jjHlT7CiNTAq}ww){YCg&Thc-2Q;Pnp+7DC@|2zNwPmC%1dyO~(9ksy8wFuMf zEJ7ppjSUYDl3Xp&g#w~t9)ylw+N(Q=T;*c0O#TB@Iu;osBVV*qpC|Mr$oBGz_O47J zhQUyHQPz28RzZZwbPJ53N7b3&@4`1f{R~>RKGjG(7dcb6gCOXXUuiqKS%zy#zjZp9 zcs26H5_tWS-bQgQv1#n68#vE8d;BX;9x=ii^K|~ z*nYqKJRB_;K=g&v!pA912=-xkRS=p=hBQF%WWZIk=kC3M&Y&>gHj9_-$FN0!(q(UnCZc*k95D%2B6Bvv-2X z1#${!koYOc!C#?N7(mz|=(_7|N!sG|zzL+&Q%{R4?YsrrIC?}RG{}I++w$P39{8-9afGq$IDDJMsdHz?cJE|P z_eCp@TL~|4{CiKpS;|CU4_2PXgVES&7bn*0?4WEP)H0k9XcG&uH=%TyFm|lMuAQ3p z*0{K)1+-zwPv@93=jWD9?i%&#?aa?z9x#mRYr@q}Rp453H z_Zg}Zd;?#f3cDXHh7J9+^J*8XHb6STp=#V2L05upp4ieQ(S?0gS!;dq45NoIyLCM(wGgO^pV*w%kblq{{dkn_ND@wymtIztSwr{d1hy{9{n_ z%r?%rt}s&C+MJ(!{-V|i==KYI0RRx2|6$P;Dvt*}9;&Bzg?yY|KhfAgJ}wQl__4$M zjhLhZ1b0#j{v+bgAQzlEaFgiyQcyN`_ruCT(0BkpuyZ430`D9^C8a;1(q&K?`;Tc* z4l8xB&})KI+OX{Ct7C7($-fo#$#wC@_bHt!bUA~_Kd_EgSi1I(0s; zdM!KQb8ms(`*|DoaN8&@{iMnqDyyV8;(GG?4}Vsdirxh&O7eH8!l*^JVbn z#811#T7zwhbxCMnd3Q!;L#xt9Je;VaUJB+5kP};t+MnyQW9xA|`j+OmIImB=qgmE5 z;HyflCzfj!YHa1v!u894Sl9+y6H(}o+C%N3Lf1Mh2(1QIe0SdzE|NPKn`Pp6+^Meh zcP`1%!Her6Y>A^yOG%UyZ4>Fv#{4*)06uIJ!YbixH|3h18oPyU%CUFJB=(~jpAiC> zF<;R=)~0UQ$?pXQ&Gd^*--SPsHm}I$MA4u4h6=lC7a#f^Og{d-OMo2m7!Pc5`Cl!B z2hBHSlY9Zds{|97W4%SZ(&u@1^%Vdlq48V zpL;fmI|41YG>>;;i;5{d6`Wsk;G@hi0*s$avDy_}vn$6w*xzFNOs)HLm0`~pIx~jh z?DzbJ{QJMb(F7Znxp$KD&(5b3dJscuz&a?*z8mZK`--gv@}LLd<`58fEdrk_hh*B{ z6DLJ4FRI$yGpA;RKW2f%EluZ&N-qib>8coGY;vkw^+BTZ^p8Zx zDIdrDwQz5R-ZEfs$j^4lkP>p5AeN3VZiuR>-gzTlepU78W<)c0G%fe(wJSW+_Xi6a z&H+~!%_32ivh{bZUNbD%n3YEW-EJAMd+h`~s@I-iimeu?G@*uB)FLvIAUQfTf6}y< z{hkd6ea~UH$!vh1L!&cVM2+~ytw_QIjfNC?WY(&vHKw*hOWMgNYQyXc>vLV~CMq5= zh=nWZm@Q5@6Q3j&_~iqRLA%-WKjGg0s2<(F^G?!#-@tT$5go)yBAp}Pb3A|MbIeKw z|HYW3I~mk4u7Zwb)!6E@D@$pYm`bNBeqJnN^eBw9dnGbj<6k{wT>BY?$X$F!IGa19 z;&kdDCxUgu%BdL7{8VgEQCDBg2gROTyNNrKV=X8lY(4RM0I^cc32?|&8h{FUYJ?C_ z1xg8>1Ore~b#B$tv5a;pFoy5XCi_qP!#)M_{co+wiC<$u$EI+9rPLp~;{X0-g!=_b z#!fg5qI@_UWl9KfyX1;$gl-=Pj>iZ@_@Cp$-w#v2y(@L^f$pE{UP;?scJEjBctG6u z-qW{d8wvfi<_TM2fxHRZ(s6i<^P8?1^B#E}aVQ(LMG-%F~nP5VO2%YnGZLvz*o@vFRp{zAttsUoa@*I z`HUu?Vk2X(om)k$9q|($Mb9hqUO&quy6PDjUIC>5)rBgvN}sz-pc**mLG{ZXdYF?8 z8%ENqr9KFi$}x~tE+T0g~b*sJCQKGY}` zZx`@`xK_o@+ti-L%XJYdWRx;x5{JD_MP4dLnMltX^YlF15NJ)p!;dE!-{?ZrRX1IY zYQB24v&Oj$$tYuRWZTWQzPwDm4s=E}Pab*DJ~FJB7$#$$Id8AaA=jf-q44>BwXP~_ zdBJ4{dZQO0cFS8a-S$=7OGB}?#-HZ<1$*d497uS%pqg2W_2Wm5%y&LZP8~^Y9HPCiUzdQK2nv?Y=usePLzJn74CRyg!1bn^cYUMZ-{7O)AV6-lK8)7&w1%od4W; z2LLH z=E;Q7(n>i9epO>*RLT~A%8+L7xI$VUIS3S=tb*<&Qnjbb=`0}Ltwg=#oH%iA?Z>rU;&*&5b~6W5~Wi!U!758*J5h9 zYdJoa+&+8r7s-*&&5Wm9SnVRXx*WUq_7tbTfE|&lf@JBBY(pBAD196)2F~On{2gA* zRsq@St>4yRL|@WN`dlMSxYg}A{9Re@jwIZ!rjs$2+|SYAe$`!kEE=ywSSu_me*_)B z3E832#+C;7-a+y0HdNw|++ff>`Fx@`a6(i(>HTbxi;sd?F!GH!&5~7FHZbM}SO8AD z6s_lU>MTvD*vDIfUs~ElB2#4*8`$?Q?sY_93)9#5dSvdoftjKzQF(A?0mIE22lR#w*;h%8ObxoRhA9F7N+zo?gSVe}w{CkPYu$C_au`vXiDu@` z^r=QWBFrd<`mlzs8nvu;Dc&~Ocz)9 z!cd(~abi1`N>`1OeoIhDuB`T05VWNwgzZZ0BQW>5_2+)D7T{4@8k*5 z>A8Og^ioED3=EeEm&e>rp)*S<2$GqQ8Rty8-&6ksdMVAN8wl`@9-V^pN0`0MJs&5#pvo)ov0xzO`di(;J8Kzr=a#SW3+3+z8F{8{TMXo&dL{g+KG#z5d)!84ZUiX zF*5Q}fsKJw*Cj{y!xJjo#x2!_r|3Wcb72?Khy(pk&w!{$5oFV`O_zMW)EFN4@>~=* zhbDuqp5uVE&RW9k>1{2Z30tOb4ayT?R6m8XE4&o};%kaS5F{42N6&}9oKV6|g65DR zZGW1*V^+Cw@X$P9Mj8!A;XTd4XTn;9dk`BRwa2H|Fr^KNiU0 z8X)vSx*ON}yXyk5t4&ZT5)45ay}~eh)4qR@sw$=BW=%)>6^iGwml+`0$KPko>6H2T z5o#F#@$VGLN`+2#p542+;x8kRVugX1;SCyDiv`mz|BNOX_r!DyiTxbHb9WDN3ldr8 zBl09f@UIl1=Rgh7#rHNmEhT$=un}X7fYGJ=0X}EFq2nACoj-4rjgc)ZAYtPlA1Sus zWv!!jxAb#?DZjfOu#RXKs4E)5Bkz@&9G|n%SB$M82p? zI2y2rKbN@usZaNGbMyF&pZq&K>rk%|jM=_`AoHbEf1=LKMty}9PuC->&8GVV`oS;6 zP0*OEuZo8JCmdBCP;W-uiOf?NM_5X^Bmkm=QYhEhJ0ZrzGKs}Mt-n`x4j+(q8(wuR zHwiv=^-k(K!az-^o(WM4q19CZt(qDqTcuxGSEDk8d~63E!C4k&2I-FiE$0Y|CDKa2 z03n2Ye5Bp*wH$6A{8a}Xj-j{w7G-#{jC3uu+gy&CLM%bypl(Hz^32h$hmyfire@EKp;7oVe9P;*bo#$Y zG`vK{Shyx6=a0#Y2Q8U(EEyQtZdL18KyMGATu{J}L|qdk_Ujb%R3A56Q`#5EUiCd1 z`tjnh+|{lq$WKnr{>#lzO1e44fQEkH+Or`04oCC(+kO^nd-IE=x7~SSHHa@=#CXJk z++WNf9A1-UwF2(w>D2+X4|SkjIN@0}z5ORQu?@l%qN!?P_ZvKMCIxT*RU^hJ2G4bXI;-{bz!S#ed?cHQQC#@Ykud` z{+7igf9pe(QMpep!dmpUT)Q=;qWU0-c)i&w-h>UipV$h>p6VP}?qR}2#)Ct{O<^*sXT7QiwkZ9cpCF$XwkTkSe(gNno{y(KvBLYJw$R~dhDLSUR;q1A zKbb7f*vBxS-ay&~v|Ej&sg%>rVbek-E|Z*dSO{r`+&A((_3*c`SLah+35xLvGm!`qw_}VAa}y1q0hnLPKV7aAonIu1 zS$%L*hMH!ij^y!C`zskacp;2zMmc)q=DofryOM>G+Hs0ykrMq}ll;8)17utd3B9Ca z_YCu*aO5}oHg1c~5=X4P8XEN zJGLpE-?C*Bw2}WhaC>`uLf}KfTT-NgLmYposFC|UDbFe+gVASenrv-4Ug;VuT+hX6 z(x1%3>1AYXgp&N{r$Rov#H~szbBNMOK6AN`~3Nc}qu% z*P#PHysCD#<2X56;SV=$Vn~K+y35O+J2ANB^zv7?P?RTxX|H~MlQ^SWOOEv%^R>=T zsX{&N3;CX+V0kw`=IRggp`E)&3kXu!XJjYQ9|b|(y-hx0O`G{EC%s7(I)ni}u-syzU$ zbWE)hm+Mk@v?|Y&X0s|@3CvkPlL^LbKTiOupomz|36>8^Xnw@ia959kc>k*KLgL2y z*bco;8kpSvT4v>U=?X)ohQy6HTM(Dt$Q#Y}I8O_dn&G!z!k^;rhT6E~sJll*hQz+P zb@uiYG$weg*7XP~WlY_<0&i1TfA9VM3LFn?`h2vp(^s@9x^ zzA|}*j2T=HI$y*xy@b74XTRofIn&M@@oj=)R3BNouW`jd|fw2%VGI1tQc*XlL7s zbxTih{t;H(7tmQ9v?q&3{=lQonX=De>=uJs37P4a%Hp3MiH|<2a3o76Odb?@{XS1{ z1s~2)t^MOB@zaTkUnF)U*%m2TNq@?VPF5?V;fuB!^%X`#a=yaSt}6{$e{8W-T>L;1 zAY<2ims~z)E&|4CxJiFicbeoX2`4Fe@{W#x1GZeTmt*KbuT%R?cP0Uctd<8)3&==l ztbXGB{HwBEoME=qCXHs5Pdxr|drf!>p*w9gBCtKcj3JrRdL|4YI<~u?D zzepD86))Mw9VvGc)88Up$&wjJtHVM6B|W?IW_W12);$H?&|7h0k>Wav zER>mCM6qa$E%l&F&h^h-`DqE#?jcoi$4Q$`@ErA)c6%^L9b$`KqVBw1;%YqF4ZXyw zdFu4?b+Yv&Dpuh2_L1gXk)}7Z301uzPA5!3Y3J4ZgWTZH z4Ts0vf06w4(qgzjWc4U?qbny1lyHZ8cJA5RPQqgGz~s09%{F3v?ZI1I1JW~Gf1wQj z?w^w2c)WdnkY>+E2-MY@Bfy|(lS9q%f9g#93_Y_TrIuI4^nQZsVi!T$?)a=w-dSO8 zjSmcFKEbPs8HZ(@|Kr*GEIO4H@~K;*`+=W8+j5PT%&eI;nf+xHAPYC+xQQKUiyM$vIM z_W>R9=5@}ix%ihFXIwH>dXnhy_R?Y0x2L<0-78>`2q`$u%u7eIsK8g6Wwc;Iz4UBU zyr5S`GW&XEE>pPwRqn^GS?|6Yn!ak1kc(b&yh|SPlkF0`F;1cwSC)!l`^=_r(;!Ipqt?jD_jq?=b!27bb&@kAB+?Z6zYS&m zM{7R*qD2?xciysI3vN|e^#+z~XxKuQRRtb|xVLABS^ucfgWniYSU48!e zmp(X~g$XtSZ*L@}Hz;sT&b&pP^_IpvF4>~5_MW60Elqf}WDc6I+G>?Q63)VK8IKxL7=idI+_Wrxfc``vT#!UgthNIa&WeEja{jQUSiubqq@(u~7#04KTVS6FmC+&hk-2&;U-YG=9-RcH<5Y}kICNY(3~jKv*63u-*g(yB z;jWa#>^DVuLQV%!2yf?vIA)ud$IBGw+^_mrg07V}@IBmI(=OD$dDfkV(jp%{a%Akb zxP}vBhMBjX+;*_!-I1cCrHMDBJ?R|R2Y&zU0}iKltJl!6J2@*m+xvx?AOxd5rV^J3 zN;6-e0bpE^LhU_=?4EleR_}5k`D-vMx<9ALZ$D{bJ}(D`>uy*J+^LAQ?#Se6wyCXR zw&P!iXftLTi{t3)^@$ZY`dihd)uCP8+F`0XM?-LmJtDPI$IVZzvovRY*;KhG?aFCj zkrv7KdW1~$nf={}7N|<(pQ!WyzSFpvmkvK3ond6ftX6hPH_nn~_o6U7T0d8a*+QwI zk}~c=EY-2J?-`R#&sSIpL!?WQ4D@qxBr2cfk3lWK3#aW`gti@5MDl$M<|X#Lw{Ex+D~2|=PC z@oXuh;&UV|-TAv!g~K%@lRu9050=0K1k%S0hheDh`H1a{kw&fd;?^{Nini6aNzD37 zGjNkFg4U*>bNX7o6TQaX2boeAGpcgOH{Tf@2A<@=cC7n4ipTj4Bqu}r$Ez0$>CW!8 zl9WS)XQw%Dp8Vs><%YFNz}3)UbXoRq5}z9=_wz|uh*^BQOJ2IdB%I~#&w!8w-qOzIIwQ6~bv>Amr}$d=*u!R;=oP{V+V;h3I_yH0JF zwNa$C%2Bn%3-!Vm9(JTwkQL_WBO5%0?o{?HEiKuKv52v0l>9_|F#TT7b&O2bwNr+K z%6^cq#uaUQ406&4M6O+^7~1z#q(rOGy-j$1VZ?F>ZK5)*|wDTRjna0M@>;LzAt_`B4w5> zdohlnTONPg`Vp$$?fcGSc4NaugPR~Z(BpnHI^3FbnX(AapuEiJ?E-%gC{a9 zT9mIPlUvT(jWKr44XURk9+#1SxZ~M9s)8Cds_^p-s~E^*W|3T4xXz)TGofpJc`4Ts zacxbn_c9mrjZu`$A`Xk(Opj zmXVly)o=Mn3-c&i3JC+UpU0;o2vR6cwir)o`F>4l zgnfs``ZjL@xcAlcz0nzS=A#>n!gkj=v(ZadMs8p8A*wY!QQrmFUj&{Wd46Tr;fv%f zOdk}F9;nr8V3@DA?EPSgV z+TY65-pE)Wew#*-BO8hRFarM(kAduB#y}Y+ia&e79`D|`L(q2X2qQ)~MkzBkBHF6e zrntx7_2w&(<3P8H=GJNEc+wRP!0Q~GK-rRQAbcelwCXDn z_{B}u`S1Zmd^zZODQyN(Pgk(O4}*G;nIi)=A`SOIYHfP|5Nb;U`roOxz>P=>=wZXY zDxObOxb(?>FyfpVcS4Bm{5SbwI7+_*TOMS)o`2F}$77TRN2UsU%5b@jor^Au^ou1a zhx98L8RdCmPII95{%+V7H@!VH6lUqIs+1T zgf0H|iib#^UbBM(M^U@*%|c^Eii)AL60Pfbs)sh19My0iHW(Ua=ALI^*SMEy&2?Je zD5s^B>tT@54e#5{qQ)F&CN7ru@%*$%2~wOnd7xNRpq&6d8_#7>t6=h{t}O*Mjlif5 z;_oR?g9k{;>Vb&KezBhZ(tK>4*D?FpQ3+Ke`(2gNA7Y+247j~5FKJ$SRY^|5=Rv}G4G2D|JYavO!!8(PlIik7cpCAOBjBLMfGq0)g))Q_ z_~XL%j#WcJNu>A~K>>nIu!62LVZoqPk})uF-I$|3uKckT{$_(zs27?;NTmYnU%58; zg~vJM;o$A{?@7dHpou5jQF|wA&w`l!B_QS{JFts`%fSDjrYa5}PTFf%tuBHknrImm>WZmz4`Yb+MIs1i{}w)80N4Y(}&w>g^!KiP9BrPq0) zvXYKNm)zwOMn{zyd12B`{4=1Dx>-|+Gur)-mVZR|y_3|gef=L0a3io^F%KgQhmtY2 zaWvijjhd7bKJXhUagV|A=WC%Kh~sErnr z<5Q7Y0qDD`>6E`S4&78)hWP{QA-w$vBYvjYi>!(RrI?*4r(ihE?2EY4ZWXM4(_^Q= zBMojOw^;}sfYdok=fb|vT$j;e1jf75!;ul;pc zdl2p}yTrNMgWqXx8f<*=kg8;jL13tqH>6Khy$-2VX8fSpbe00~<~51T%z7R%qT1G4 zT?fNY2oiy#9uS=>(6-x008XQ>5^`w%=3i%Ppl`T;NDM+wy8nPK05pH1zntTT(;NYef>I-3#`Kl z5F9d)sZ+nJ-ms0b=&J+Fe)Yzv%K4*B>8=zs6RPTI?X5yD9CH3cD$@2hRHWw|u5_qj z797dgiz2qH_wnD}d?cJ%=3fxSBV3Qxx}gMkRbv+nJ)G!FPh!IqzHbqnQSWg`T+`&*h5b z##3WMl2?C@5_nti2^4QjNRAfPQ1cMvhI})j|%w zY~Kv5COrK%D*n>Q;)vCs8o{S~`X8JZw5!kWJc}8Xv!$*GsKDCFxOGx6`V|mMQp*)%to)`X8cdVq#Rr3#y@>MeS1xN9C_yUh;Os zk}zglPHYts2TE~r36)O7XzR@irjs56Grm|>;p?(4UpXwu%F`$jA84OHf9|G3Wr;yD zGZys>{>x#+Z@ymBIXn$$3n9D8U8gJ#0 zh`!fbIDVHr=boWl{sfg(n_iEHtYyYrJHkjm{ykMYI`k;Xgw@CNkHxdc968y0K~s30 zV&8gfkC16}ic9&5QnQlcmv4#bB+vd_L`tIbalcbO{4r!>0M26^{QDAALvSC^3$kbF`F7jy{OjXk9K$-X!4B*`nq`4rlTDv5E}yZljfN5FM{W}R3V)IK#6yXa zy2Y0Ro?=$$%4*O*^QPi99($Xm*0fj<-ZfLxzIiOpvXNbj8aLTtNCcv(fa3dHqYXhQIT@Tj2Ec`z?chh)0cihzbv-x?E3`+U0l;NAkzf^8if1PJ@dKl^ z(IATH?IPEf!hYsYD`pb-D+BgUnf)RmzXsp$Ez@wK*#Y+=Q3LzNPnqxKe?hKa0Cd70 zj{78cU^|>Egk94`)EdwIgTELYgMPna5%IuNkb}E=LG&!xgl$Ls@pW?d?i=;Vs9lTB z%y`XFV5HvvBD&N-XRz>F?(eN!j@iwn$Bdv_ObCGq!^+_b<4$hXGSUPCbr>+ zabAuke_SPR_%ns~s&vy=yE6Z*?qDyX2X>Snj7&+x4CbX$MCa8 zSMyr?-SsY#;E4drqLalJ44W9EYXi5!!8+wsoy*?M3EAhS93`hbUFDB55zPl9(ev< z!m?xAoAv+SQH$^;2a{fcnkiEhX8=c)+KXqSFsrgW2HuvLDZP_27sefo6OE-2VN_!S z^|mQ4#uiS;t4pu8#4&_NA8{0Xv26x9Z?mzeHR4y!Q=-4Ytf zII2xY6|WaB%UwN$At(5iBm!wq3v|QUYq{Vp20+2`qlu3)5*uEvfJ|CS*d^>tppyyJ z?DN?*i>vMDid!;n^(`@6d;Ex+@>M9!O?~9S?>)6Y+V4brsNb=&Vzl1ay^u+I(KyP; zLOCbk+cZmn;T?i;mqyf_&cY+Aa18vsOWDJyjwp=4! z>FF_FaXv*Uzv{-|PhNf@txNq`2eebvxBGG}*Q|qu6lMRtdMebQIq{pj*3^UZyeTsu zFr~@17Bx+N=XvuTdO`4D{Y?kt5Apl8PJfg6eB{CuGCcog)SiJS=>NR))^!Y=%qP

    zN$A32D+BBW!_1`O(I_@rNm+r?JTf#vI}sb<60@MzJypiY&?r0x^0o3O+OcP=r8wVk##;b(JRA4*8r==;}Ov;LWOEb@2*1xz9} zexhM(vz8(8esS-VA38>0mK#k^es+MfwzmKeZ(waI#6W0G=9Dv3ozlt@tsh;cQwksd z*wMDS@r50gcJDf;5ey*fX_qG&)UYx)ouQesZhEy|K{`i0<}(jl8kU~1dy8~hU9LWd zzzHB(uU^a_XqpgW zx~M2_#?wDhwliU~%Pjd>Lgt6Zm@56hj<@(nGZkycIN54(Ca(hm3AZ+*>OBg-o`MNk zF$wXB)REr{X{=~Q7!o`24!=llz91TQgON+i{M{uB@9z{3=(q7Wxt`VL$8o0;>$~Bz z%;q3GJV#_~myt>5>m4KuKKLB8n8*%z0s!_{(L{oCmBykhKhewC1W%oaR@&qQ(0{#3 zh(NA^&E6?Ne)MYmMe>v&30yRA+LtY%_$3Hnjnwd2n_nbtaT-Kv{aUeDMJ&>%RMv2= zjQjRkZd&Uf)7sm29f>%IKTH5^)E|pUBpDddFnxM%s@B>@FG*81Lz168GF_v$d;d51 z)$c;loqJSc+YW$QHCr6q?Gm(LJ#pvQPx!bL+z&F~rm=Tq5sBr1)+P3(6Ben34m=AJ z(bb^PRZUX`YDM2FyKB{Nw|#G+NZp9loSK4#TJdBjgCHJ#9Z0H z!i*6&Tx9=@FmWIOF~teRA^r2G3k^8*PZe|26??uD{S3jt~h~S zOE^)fCCj=>Jb}ED-a`14wrMQXn=Yc7jMp3qY+8tooN7j$Kz5|08BRQ2ycfF(2|%u2 zOC(%)yg!4@GXTA^N0kZ2c<2GKTEOdA@(ms2@>BF@@XyvoC3aXikkJrqE+TV!*Qj^e z26JX<2?frTZy@~@#C)NdhmVbwu&dv?Redn7(+^lKkO!u}XfAqf|I&y5#{+G857FMF zwIrN%UH%79 zY8H%6EFnLa_i+PCn#~bW=NtGum6wy0>*|>}nAKwqtA334 zn%$>daO@I%^I-fcc~tb*&I&|7&bHp%I%P{Cd`E!-N%YQ$?VBPjQmzybWPy^G0OF$x zh!02VB9sO`YYoL~0xsPQyGf)&l8W!-$V)Ds?aJ^V-rILk%;4&7C0{Yi=1&%Mq}xS7 za{FJ!A#N3~U9TQAA&W^j&+cv+!MOf8RcPU)R~cAh}@X$$ZAV z$9V2>k1>`>ls_chzP=l?DoEiH1(F&tBj=~86rLuE?>CSSf=Cj3iyNX}A zxG@}h!xGFE)>mV!JtCJ>%iFD)Jai(=i_LEAbW{=)v7+q!O+f3qkVrdmueBR2&;zQ| z5gW&r6p93hk^x6^b$l{w7XO{Q8IG>j&W~CA zI<9TWK8+S~Fasbx2}NZAfc#+$EZjzVQv&F+$Wy=a6Ay~>wYB<~O5+sU>H!V#J(zG< zJt)5nd-W49Ik30l!VR)aG#=mvrg}fbcpAnZJwaBVsY`2mW+BznS!zr^FH|03K7-C* zgd>5}Gs5HTE_TGS*b6ecy}bXxOWcNGkzu=(mIGuY2Ib6AwU4n?4qI&)FeLK1i@`0P z%O<-t9OM03o}eU70`n2_aJpUmMcf_eYZPe#Bv1}byQv%<_n}!&NaDnZHZx>>p_Y4 zxd&3S>CPlU(c)asHjk-5No^8bA6@qZEwoM)M_x$F)p*HNCJq5dFY($vDNel`-(=f# z+!_5uw4Pi*kh10Wl~E*7A8#b0g@ndN%OSPtwbjiZGS@3HV53SSokD~xbT`SPy?}As z{8iP)UpA_q8Tt=rTKh$z_P!mi1i6>XwBxMThO^Jf2@dt*e+#vrhmcp4Z18fdjU*H$ z>=#e(IxE{DSEBP}%pLQszD;%WCB%CyjZq>9Z#OF%yns$TIoyn_ar0rEZjNbNSe^~&t{ogI(M~_)cZG$_ z!bs`9YHJ^fE=<=~GHS`tVjtXJygDI|@Cd1W!VKQK!_C}m)Z6AbD2C4-h2{vgajH{Q zIv2WaW&EbjI!n{cd;E1rCEZHRB+63>64qHz?y#YW{2;H)*-YWnP)lQSM!U$?k*tgM zPKIXy4-YWLSYiW^d0;|+;@yduI~)Ng+q&uDB;$eV$!;hos(}(?y`b;G+LLend=xoM zEqqqh$9U~09$fbu$}UninXv@>jNX)8>Qb2w#ifJEhThL9AqFqb+Kl$Z*FcA8V3;wt ztH$0DJgbMG%8X{Njn!)zWy?|-GvCs=WW1mHRYY^a%~YR!#;u~=Bl}FV+F;NhQ?rnK zAPGyi{$VZwszq0X4z)_$j_H}N3gkxG7;isK_v3q)K5Swywg`8)f`o7H-Zxo%GIloR zdNkUy_h9uPS8YDTaSjC@u552@sY@%n#8j6rW0a{edYr@u#S zj#0S2f$!kK!Y$*Yii%ILl_99&P|10!7(cl%P>=NFfIz_L{A%iG(J3Iw`xIeYWk1i& zsuXiHXPByD5yEX}vtUV*<=FLd1>5T3nj;`?j!db#b$@|W^}9h$BG>urmCN;05lITC zkKoq3e#|uV&(W4HO3Z1SQe$+gR`m)x_R3Dq^K4fFBqP4OAKe4M+RO0?~G~_*;qpG+h{&XY~{zHH`-fW4g z^{A4ZNJW8qEmxf7E zC?nrycqTt&5aaOujbw{))RGN{m=;3g-hnsY{hRU!n*z>HG_|QIyrtR<;JHmrIyJsP zEd);L7<|~(5REWG24j86&z-C`cmawDDT+a z@GsFG8Qpnu^UC~Z)3REf+3;~_-Y@jWB21){Xmr2{U9R}WKIJK%KWEp-w$=0yJ{pY!WKyEMXSjw%xqp6P=y;TR3EH3g z6EFVHyhbU87#9e^mj6j+J0KR02637$9^1XXIqOR^hT}a)-b2seB2jEaPs{nc=pG-4 z9bT@2U2%goB?<=!RAr7iC0Tblotf3Zw(Wy>qJ8)8I#xP~nZGBOQm-GH=RKtX?@a+N zFH+Tg77vIQ<)Qz-Ll4_;FR{f(;4^#b%+0Ac~L_@7@Qfj9BbtEs1!3P!v?Pm+IjAgskx ziM??b=lm4%;dt<~?4%P5K<0L69S>G7z!b5&-v`uJ#j;`&VX2b(prrH9=R*gEnZ% z@6cIKL2P({?fsVr-?bASMUa?_^)IeG9*K2Xenae~Il@y(pAe!aE|U^$2U^lu5@=Z@ zu1PU^fj->v2&%QXguNZ2)$gP}K7L}S*kF=l$cGhf7@5f8_@1H^8s^~kN`2|v8eRhQ zso%CgPes|zc76!3*nzADBK8|EoSoErbV5$@+a&Iz505ACcZ2c6GSgbWG-aIBkePu8 z%Yn*tI(9Hm z9b4Ol=ofTsW83Qf2g+aTPHR1Syq=inY=$B*whFr-t(qvsQ?}UM$V_&5d%Dh!7Zxw~ zsh1WIalLNim=(Ljg!mnC4m2AuF!OQFRExyD&p5OfkK?S(n-$0^zn0m*H!ltS@G4k& z!d^)vPzSXrv?_5~ZDY?4jM*y_)(S_!)KJo z>l(-fJ59AjQ3A00_x)oGuk^Hpay>TSf>aJUt$p=h00nM&jc4#O+FTKdN+a=Qb~W+y z`0Sv}x7bE2RHcb2W$QrcX?H{K8q{A|`3AK8Rb`S95;H zh|sMLcBaG#`c9>D${Qd^T=hl|a&1G)j1fVRM%&(L#B>GQ8G8bI{dZw)3G^8CayH`HNglEP2 zGZDbu;Ka8$&uEm!nf)D|8!oJV*`WWI(LZ8CgM7OdraDC4AJWpwfj@tVVE{T2SltLI z5W`;xuY$-_)<#hQlgFN0J6=RRxz~v_V;{Wu@A{@GSoBXnAEV-lI%>S_C&|}FZNg^1 zsnF4FpEp&(#orsr)^Q9X@ERDO*(fetYaI?nCfhd*WACBLDzYbSSrbl(juh`}kE6&5 zFm`QP`Httw>2CM!@8cK3b-sTlBe=9sb7>wLLcotpMRG}im&cSeJI)W{3V#=%Js-(z z^*YbMh_)pdJ{q=>i?}hk^4MQNjNZXG_(!9P785Z2)EHyFKo`X0!)k4_xpXu9MBD;% zv3x{HDx(|Hl&`}#hT}mqHC|z>8(ah$@2IJFG}{Fi0DKMGYE{h9gVff0g>q;{Rb^zP z?!D*~xsp#WsZERTN#c2o{jq)n!^3p5nX8YexLV0=)Ml1B`U5J%1rpcQ>z*#d$En<5 z=BPS`Ol?szszz6gX#gVTetpd_H(um-5_cgoarOoP8u~wvf-eAp#6P=3wPk~W)L+1o z9^4GjgRjmK&4VFlo4`Jaje!>;IhsVAcY+;l?7qmxY*L07T=R36rK7#-bp11W6H z29Ata7%3wj0(#3lVDDJ@PPnNJvOrO+Bma$|?MrQEVGAaWLhK&4ry^@&syKwT%F7oy z>PZ4{9~)2~fo##aazoGqxpT3N6}6K)mrh4H{{2^f`D5kRW6&EkBc_*s^ZG&$5mSS^ z(;3wxQ22LfoHju6$!SyqhxYcw>6FnQ_<*#!!=!Ih+mJN;MQq1$f-X-qd*u7UL1DYC zqcY8e9*$d_Rgc>|WRxSn>>%7>AH=A#CNXdHLT#P(CIUjP7+UcwJ^{~gR~#UMn+>m_ z3Ov-Qpv~f6#o8|MDZE8YbmU*Q1UU77k;_uFen#&20Zjw1%Z|i?naXNavQv*uBr|q- zjSbG<)dNTuy(-=6V!)^Ftdx93wL;3((}b@P=VURIYcS808L=O+lZ%h?^X2SmOXfV(1vw zx)-qS+duL6{-x7%Kp=k^(1G1HoD1jYVsv~CZHN8S@3pHJfeCcs`~Js@)+skm-+p^w zzajKt>s?;SeYVp}iB*Qed4)lCn$1k)EvnMc^9-x5I~ER4HJ){9?>>qkO*<|$&FnGe zV0S6_FO@>&Dj^yE|0sp}KOI|suf-I@WM;0vi|$vvj*a{!q;hG=q8^_y4P|p^&f4;(qDM zz?ZDMh^W2`*BUq?2hJID9Dn0EpxE2=uyYcC2A)p@gbR@afTlwhxQsM46mz!(px2#K zD}ZnweCs^ojO2_EaLjhG|G~6_UIB1Nx(EFm(G2`iGi45duWcaQbs!~xnE{~!09F4Z zxu1A`)?l2l+~2O_Nq0ndVFeORz;T2DIOhCG7(xJ?VFgx`9tOrz=>WPOp!MlBpit#C z>?hvUQ7G0Hn9zpvd3eOX#lOY;>#R@@`CA&GEGZ90Iu6PBk?!`zeOFJj6!zV~Md3K( zdAhIk?arnr3uVxY7s}b7m&irdMBY6S{KNK&l}nMBa+dB_jb)9;W|C5?Ir-VTNr{SY zUwn+Ik;?_zMQ+f2=I{m1x@KeG&Gm+ZyM^+FLMXZ$)|1^ozKai0c)N3)j%5!2i_R4( zY(o&c(!Y1E(2BLpXh}G(UP?R;7_Vx#knZiKS^e2l8yRgxK zAG!FAVZRv(0+?cm^Wle>NVHts1i5iBAX1k0O8r~b0083Ow(@tQ*C-Y6-p>AU}J>sWeF%txN&?qyoqH#E&&B0wLs z86{l3ZViyrVBB}lWuCLNn!!E@`WMB)&+dqx^SwddSXPioUHyAsZ4GMoyJ7Wo!?){G z_0i8!8=y&H3$#!T8$R`Oq-+yQ<0QdjIQYVLj)-%~j)=)8RL5?rOzkvOMZPw;J93ow zyi?-MQwZ_!2=K`Lh5U;CcGx_hY)uKyQ&2^)JcTr0ZBOO|0(G}GiQ*58>rQ4XkEc;BrG6Wmw!I-)+wXOyZje&`x zF4Sfqo$jz=KIP}XGn}S5K8u^tsP~sO56OxlC$SoOwiP31&d-|Lq)HJbbyz4i0QaJm z2+rT7O}#i;N%H$QgIaJOYM1NuI^Z+?>Jb9MF$*|*qx?$!}UFqS_!|8L<3_!oZg{craS zr57;?aE})~oD=jn#|4aN#bS-O%*R~==3WFZkj{J!hb!l3Gu>x*wvj=MnC)wuvnLVW zDwIYLHy0VLGe52wB6*PR8&}(0*=@!5P;|`GFz#luz))3^vM7JtzoG9N zF%d|wyYnfArdOu-pjl4Lw^S1ah&%$W9KbnHUzk_fFZMn`dTDih_*FRfx8`cVAO``O zWdJpe4KY)#=uO<6#GbTYEWAj|WT5m_Q=~L(fg|ydeqS$HR25Ad;q@#(ugwxa0@Xqf zeKXZ9|1cv)r$6?G7En<1u*QY+k%NnL{RMygUv&jzsekhYmcZ>ZzvxCRE`|@_-=FEf zJelHk;1f*C|3c7de^DyNpWS965xDSar2#ik`Ycp%f`bRnN5DW5%xd{$q0Sa)K|W9h zA%(|hCJqZT*llN6Bq@|Bf8WakeNv%#HzVVb zfOZb6;XKBC{mn?>?pCc_f#u5SOM(6=0a4)q_0AbOacb90z}|#^X?Dk2!O0B(OOH?( zDp;DdYI(wo(Gtl%zg!ClDg23dah$N$oVX-Dh>$4R{@qRO)IzoW$X=1FaW~l-`>&E z1HjG}V6vnOL!hEifa}8E28tNS>ThGW0Vg}c+}q$p`S#nMc@b$Fw(L-V2cqdtBCk>}Xz3q=-Nlne~o0wadZP zEk~6N(nL;s$I1i;4wUjoSs$g3i)*7Y_S=kNo`ng;$f?WFy)k)Dm|Ms$QAQ*If1>%}CRknbvn; zDw{-Q_PZ*MRKcCt2DsCt*}?|B0epet*#H-MLymPn3r*uUw;t2}BTC(mD-?hQ( zK(|omS9NJ#)^FoT4@9-;@^WUn%L(-q^-;^Qs<@g@vailulh1nk-ZB)azma-C1Sl;? zcpET)phyYI;8dw%A zL=({epyk56l)SUGr>07yCJXEf_JyAjaz4G>N+uEao>$`m1dL-3HU9 zH$B+nb8QxlDBXqMv2_ybEfLT)osYfjKj%EYXd?gG(*A2tR?eSx-!HY;jrL&Kh z^`_cfxVY}WrY!f9=LjqN%g(8p)S~JN60T=rjt%yfQxw|jlZmD|ORC-wmlG_q9SZCV zW!wXs)M7&I*Lo~%F5?`&p!FRB-^?U3wF1XV$2@31SIZpc;jbPM!14a#8~MNqyfv{0 z2-x4i^hPw7P`AA`M=j+>MLpf=O9Z)g!{V#;ClhC_MvtN$beCCwc;Kss{lrToZ_bXX zaH19cKLKr#4Ph&(Xj8@PeLzrZS74JK>~-^HrnT4yH_6uYYu8yLsK$H^{l8*d7BvR9 zor#SYFoE!$FFF}dD$WY{z{?4BU~{8t;?JvCu4FZwuOXl-XbJ=bLNs!}zhPi^(ON;M zAfU8J1QWbp@vD}|Q{=o~f2rZ()9Vw5T=x=Qgp$|oL}mn4$6I_!ncUe>xNS^a%p$M2 zM-6MQF1+X`oU{ERbVq`+-4~clR6J>2gSg_uU%h!65LKKj_Gnj&$XPu$R`W_;0p#X} zb(Wj$&TFXqom&~q1=D~O5e0W&4I3p)aliREgW=Qo{3Sz6S9jq^tPDjcqh)*tWiF0? zI<;``bW9b1$`P!@&}5%9%&kSZ8_H~5#L5zuZ*p0Cf-{EtQWpn};bN;UQ;N?}!Fs<* zVUklgw&zd$Oq>QLOG*?E&xN}~9q%6yvU2SAAr1Wv-52es+=NxXS&X$Nn#LR&-9?%L zI@S1o9JLFkd|+G3V2?PT-JYv ze*B1yJLoE-dqM5KzIb7Hrjpx(13ehyjP#9b3zSKgcq$JGR5`Y_N7Uk~=(I!%e{lVsVGHDY^I(Z`Ybxy>);$JJ^e zBZU>J&@`J;f-%w*B)z0qA})OeiUw!SA-iw~S1~4DKse4j!!Wm=49* z+nd{qWoG}OWxTJ@!78TpWsElq<1-{VYCclvCFSnv%3fJM1AQQQhk*M%jTSJmxLyo} z3x2&Xk#F{3hBc6fq6FmQ#(X%UEJ&SId23dp`phN&>sqvf^vT|t3+U@^x0ZKi%hEc! zu*|I^xKG$A@g5c>IO|bJVr8cDF!N_#erNb)N(C0*L7E9)4r% z!UQo7sSK5|6ABIG+aD61679T>z8lIO>Cc*DW<1RtRGZxnrV*peoe=+KI9(g@5In$U z2%VSIga-NuSh{#T_i#G($#s7KejqzYuom@LL{ULGZ+;CBiziXOIf*n1807x5-2|kNub(HI1Pf? zS~H=q5)3(NTZ4FnG< zbV2h=QUP}@#J|6j8g{n1wE@=k+gHQnW;^C?qSs%YkJQ$dllg>5m|(aa9W;0RR$c$% z2mb28waA%{+djcFt13nWx$>QqaZW;J>M*-U6cyO3^us6-FJ-i(x3nsneKz4sOMJ|V zwwhC#gIAGOG;Cr0u!;K5&r$fl_#BNsd?|1QR*yFk# z7CxX)PFn_7X)7We0w^TkwEL889k$J3B@)z&6Wh<)W?523&88m6%wHNC3MnzjvQy*_ zJ8tYiYiz9JZVlh%E{l)_u*1FnC3YjBrN+cAO*-1Iu^8))aBNFV#j<%U(Q_x+l~s5T ziR{sxnxK}v6%Gnz2Db6D0xTmKHA@DxW1s%-(?kIU`VY&_FEgpm_)03!tSS)HAQ>ub zp8RNhp?m#_8qV8Ia7#mXrI|SE&ND((UPR&*zfw$`1)j1Jq^cq|{Sh+ZunFhvUJ}UxJZ1H(1}S$bdV-0?E7 zs{uq)XZa_;Fo>oC`0MOG<|Y@%SphUp;DQ~wLKgAKe0iL|eq5-(2yagj^TewviBAQo zfu}o>NIRkytJA>NqUjE%(T<2%k^=v375m64JXr2&O7mG`{Cf6u{m98{lcDiZEq^f@ z{8c{zu6T{^a*C?OkxN-U3|sU`si;2%q+lxOTs$zJ6sRm%U9dab8-_OH+nkSJ|FJx^ zRoJT%92bS>nKJ^W-WrNiqEdR|j>bWGEm3{GLG}+5>HlkRY zr1m^CQ0?tBVS(_agC$=;kZSjZ2oe;zfFLp8lz3Kvlw4=@`q{TOE8~FF({-K!*jJ00 z#cj1O7Lz30=L^Rru*SENP%2eI{qJ;&DM7WV3q%&0{B}v2r4==5A7MSrw%#ATfI)2e zI10h~RU7>c-g1XeA%Xm3(+l zDOLxPKL=9Mdy)xCnf58&Wkx#A(b2RsNt5{d@pmq1QK2DHv8~LcHK7PYVkLRAx;PWC z0G#pdOLCi=MDWr*c8&CHn7p4;hqPz%u*9 zpho&W{{_2nXD{e$Hv+woK858P3qS=@pCYJ+p;DQ$-6BgpqiRnOn|5oSoD$dUP<#nX zkDfbGg*KB5VCURwS}WKNue+3QK(9qZHnknCaNx>_^9*&AZBPU_%sj;c$aGVkj6H&8 z@T6&D!WV^--(BzWHI*O2VRZab+mu-ZytI(eU=NG?-cTw-X0e*26Wcc?IOH3addehdu7D&(sH zp_d2PeZh!sU8}o>AUkscD}(XOkXq87)|q$)Uj9r`V6sPjjA@t;z9ka=!|xjL+uhFf ze)|1oKbW?fM4#N;Wo3RL9oRrr<1vC-|7{4178P|)#~J~ zl?#KWM3g@|Otr5sjepX!RFQS9(0vluIA9lE!HdgodC+i(_5$Lfw0w2*1;eUG-*wkq zT5yRF`}?4e*4Pa|+A;?SX9RM+Z18Xc@*rLGSj|7w@)K>JYIS^PsyH2+p?;i zsKz z0Noe3-RC^$r1T1Uxmg-Bk`JP^o(u>P6Gc@#-%HBe`g*UU-ZV@MM~C=A3<|9%$LY~c>&ik3B@`0TQx}KG5?nP+ zoONg8Sij#@_^bv4fp3x~J9bP4qI#dd9%q9d@47>Jto2@CjLDJ+;|N0zJ)KzbAH+;f zK!)Wgz;HbGpwLAi{w13V+4j0TbcclX$D4A5xWwBVP|8$~L})0WBGbFkJZUPvSR(Jg zh^vDV)x4xa&?+U0i3E#1=O{)yK@es%pPp5X4p?JJ*W4X6ORtjtLG z-DB1Tru$1btULthKQ=BYAgsT=AUrRjd<%LH|HJ12!@XxWzhM;HDUVG8SKop0WoRr9 ze&PiQsk+w zPE&td=#*)=n}1EoS05aCD82Bb=6;!h*2|g?begx~nLt}ejn3H?r_U95cpi&6h063i zW%fPUeztKC*}7Uk(4V6L+tJ_vFJ&4;5_?-p_?s9jmtq!$%uoKMzFMx|Gw8(T2dl3L zO|^Thm4savgiv*w7<{^JMj$!Q80XUa=<{mID*Q>=x$YvY!kP;=(q64~5H<}J*-Aj# zR`Cimvj&MB>KQzt1%;QWXUuQc0HUnklDxhRG4w64AAWWZ#V}H7x|iRz;H&7#(zfpm7X0iYkM<|@ZD9A%OmwK83oI6@GBU$CH;iObLuytd8^7Gh!bR>N{~-4aZKC z#o{P_$J3|8DN|vJEi?7^1tvZ1%Soih$ zJ5vIZc?5Vj511=e20HoG4AWb!)-$kA+l(Yj`*P+h|Nc=(s6c301KITY&HeK@+hHlT zyeIOZErHBg4uqq%$2f{@Ih?0HbPnJOn1HCa&l+~N5#CX$-+=OOF&g)fX`6v}N1A&I z{9rE)mQas|uo=frM3Jn1&e?7}$U`*{VGK`K6+UK`_%bbbn-c7lJQM`DvuiGZmPq{9 z!l)uEh*L^r+=a2XCt=tlKt5QXE9WmrFyCVjJI?vFRE37ogPuopg}mwEpk_U8%)O;C zVz^uHzCO2cdMg@`^l?o;Y1*`k3!Fcvs6h6QeoVc;ECY%<9SDI0=jp zqx058$JFSLJ%wF^PCOE_kDPh;G;Upv|7Lsk{_mZy{s(^kSB5Id;QhVz_rL$CzXM<7 zBIQ>VF}+!F`ZY6;SxH%hTyR0Hzlx31ynN+qKcS2akW@;X{WXakdbSUwnmT?RPDb~E zq4WSO&Lu%{;*v~vfrpo+rr{J&OP^jZFbW0)^Dn#vf8xFR0X@H2mg>I9u)dWY>P5Bi zu8-kfx6rwK2lfcTDu+&JB^7S=coUoDB~r^bF)2eh$*Z69I`5L`!|aOVoGX)gS-9Nn zbrtv1$iuSF;JL~;6Qk}#lsH5cGm*F2Gia9vP%(L>rh#BV33D7+JNJv1vY(^9&HVVC zYVh7Z`#1gX;>W$cP+h!Zel(SQ6}yw3GDP3!(-534qwn0=2JrMSRfb(AD`}xqU}mE9 z%dIkL>q@CW)NOzKa^d!Mps@eR5xfDT zGFdltb6CuFjPA)DYY_M_Mp}|dV;k|gN`ea$%L-fYW9Tfn!#Z?LI<+lQE^8A^d|>Ii z)<31Mzj|(y{ISBJ7$_a})lO|4m8od7fOm2&~F=vTMc)(W35 zdTquh)B=|(?9uesI(BNG6xMqH?U>&-eC|#C{>@h6-tk2SRtL5a)+)HHBqGDOX-{`lNS$w%)hiKz*sFqBZGc4~)Er$|~ARa@417|4>=!!l3lw{eNFOH4M>LKsMGH$=C}Z5!{8=E zn)07`YDhRU|3>+vD?s$rPZ-s+NPev=uBriTo3)XkyCt;YlUKsAmxp#sXu6|(gu6P# zV~RW6`BK4wZvJ7oOBVZe3Ys&teW&C2wg&Dg0uQed=s6cY={CE^0xquQ&Y1l0!_L_A zHt4wRoBzp314hZq(Ybkj=&86dY8nvnE$Kx0#7%Gb2ox6$FAZ@&C}Yx73ArLzpG^3N zF=2FFsJf~*vH4^^wAtDnkl@RO{>0nu0kTcZX=KvdPzs*<1x$$fS)y5B11=&Ct#i5) zyZA&5oU7nY7}EWERE;CGKv&k+g>m@~jLp+LaTidy2?C)YFk>2+pR@wdpD+zl$2Q%d z^fW8Ymu?)b%}U`quT}e--s@bWYt1~~BE@si74{dC%CYe9ip(k~7B77pwBJUXB+vP1 zPxk%2qwT@M!WJf}Kls-S$@}0o_cj zZV#s_0`;hh1WLme-{e8Pvy56l0UPjk^RhQzfC8|benN9cBvpFT>fD9|{(uP%(;p4h zN=VEi%gs>~vS7YFd?-ES5{c|I3CH2S3z?t=+a@fSCHzd@rQEXSigB=|QV!Yyh>-osTJxt7^ zT!8nYy%=8??D$ToI&{vNErq$iAcM8Ffksd?h?Sgj^yZ`xbJC^0%@+_~kGDNFs_2`A zUXH1=Cy<|b8I(O%6J89co=}4Ohf=VURTpDI+&O+4+3PWRjHn%#f(gwwcFX55$*@=Z z++=>Zq-|as_5-!Anp`7@wZ4W`ZcT%Iud%~u`cT;&PI`@TG)&rMCxDNK5=-n*XHBD( zd$qiorcT@z5l}u-Jogf73A(Zcazf0FwuU1?C8?vm`nGKJyh4~#%`3|PeV z$Z;0U5$pxc^EBG{VH=F1x_8bsnv+J(S%E5!JZ?f|D9oLoE2)0a#QG?ZZ|aT48w!-` zYp=~x4$&4xE<1qnpV3T>iRQa~Z+SkP^t__1VMLHkWlwj#-#zk1O5k`@Zc1a(-S?bO zbGM_#bGwjIC5W}n{kcQ;hAa>&0+V9MBj)NsTUkD|y?Cxz473%}>pQ(62HapD8zBFF z2>K1ruYe6u&r(q^D?4x)R>47YRIPZIhNfP+=~{U7whG3mk)9wO$xXj;U?mI@jJQF` zll@Npjo@P<5T4@@jq6La&Ef1P*SGc8(GM-t-I^XwEwal4!O%#TC5{&p_!Xn3G$c?M zoN!l9>7GsZam;(V=*zLo5saoD{e+IREk^jQzwl&e$xdfsAA%E4EOyMs2)9ICN4zcN z=DR&GdoLXne7V+dE2Xfb< zR7t@Mq`LVRJyj+A&KZ>|i!QE!tcjgyvIn%Y}y|7Qt={Nob`Yzu2Igp?fifLwV zCfjMp>LW`Ya@;c$$y-`QYs9e&4WTy-Knvx{PemVx986G>KyM11vl;;%Zx}HijS%7E zgKZwmCFq$2#VT5GQpe5p*^vDa69`h)0l70lbwV_lMg}x_SK9ml5WWO%K`!DXXc++F zKU~A+;6~6rV>fZMLvI@<>qdy7i^c8GWam;9LiQtGkO5k#EZuVRC*F(E8!zUPItkb1 zuy=H4`mh0F_63h|+n-hR$Oo5!V2WvC<`#H2YThT#s;^Ok zvRD+370wFAolS97oF6nMuC1kAODp#q1@p{aLqc}iedU%ob(faVho38Mv}Ib-x0gm$ zDR2yJ4>#Xpq#@hNYlVJO=D^;qNIhV4OT}TQ0Hw+X-lla7Ca^vHEBIItF5a+beDC)N zLR*4&?9jmcS=j7QlOd~4MqPO&v#H8_$ch~6^2d}gPZh$f0YfeKMMy}3D*+ee>WpRi zDoi|@XX!LxpH;;ta|FtR12nu8mQV`8=jUtcQ;*qtI$P#QrY{%Q1+?+qe@yD%!do39zIKIwOG0D>ytt`7 zB%lq$mJDt8v{oMMvp4%6C^}$81~}~Rmhhw_6v?E9cM{b>^6jjk;~roK@jXR}#YLv| z#wVq$udOpu2E4?Oakf#bT>3TAOJ1EhO?5z$bZ_VkXqE`B=3+HKOC%$KqI}Up3G12| znY!cxS)H$}h!Lh>hj)ZqAm?<`p6fh z9^ofC*#GC9_>Tc+{}aCx{c?+`%bfnN2cc?#P~X|0S0^{#UaA>kR+`GS?~W0T z?dyL5rM^{B+F-6lwo@> zPnuK-J@7U*VL)V@N{$iv?vl7SyQvZpXdrPvI8l2shmP}6uROx2(PL(}b=%!b+WArG zP-&cY#t<^~+^k7OR}OElVTKTEEw^|x`YS9ru?;f>$6wv)n-q)N=dY8SZTCa{;H@K#Uz`@+@^^fAde@MO02nCQh1|5 zfv!C1$Zfu(x9$};qZKi?7(QJB_ok!euwlA==WcyJ@xDyPHmB}!SU``+0i2bA<5>^L zQ2*mDstc?B@nH3nkmIcjrk96DJS$}fok4f7fbIo z7FUF4qe$6YtG5kS$5B+z@`!b7H1Vw?dgT>ujHJlniK8Ta1(b5<&G0Tv82eU*u$Cz%&w2|W^94m?%*$sA5${rzsPU^-@CsaIYb%^E%3f09?qWy$M z`W3C#@;hEv@_SpT#;3{-OVhjtlqqH-vWvgto_#nR9ur^OJ>l?S2igo6^x?v+@Fu@~ zLcFzEeRW0QM%9(!h~YyuAo`0EPdNAN00Qo?3dz6E{rdAKa1bYXEsGM2&unCoh? zu3S-Id#wPLdM}PgGp@ysYQZTa1UGCBK;d}Kr7!ifVWmnvYEoy02{{TTd6pg9LW7}- ztTam*`+k{Z2bV_1h;zK8t3`Ttl8|#GM%LL7n*>K0P8mkK8TE=;<25eA|D!<`R}b1b z!8zyMn3?W!B_#!?IC|{y@fR)EtG?J;oP5vi25HkYJ{BY_UVLEvSfE5v8izFEH4t0Z+_5ZO|%gjV3^ ztBY>p;!q+xzuj47+LxH{!VMCWn#im$V3?$grEMMcy+!v9@%;`so-Jm#!$ksrFr&IM z5Mg`QyJ(8l=FQ>jckc*1&cZfKpnPmzCJz2#C|+u7dfAeB13RPo(b6u#x@XVgNAJ5c z)MZu1$w%CKGVQbPdCoOm8u|2x+7L@sQH|fm<@9Go@@%SdS4|mJ@Fy3sb~p5lFy@*e z__h98e&h0lckhQr_McHmo%$LCIv7d-rwe65T+1l<6lfdx_nDJ_hh#z!(jw-#otb;Y z{L;?rT4jm5tmpdoN$g5J%raklzYymtLMSys_imzL=RrVZpfHape^10EYBDD5)(OGM%*0je1UH1ZcG?~Kw=wc5N9VsNU_PIU6^9H7a%1jiHPm0m15A$9X zf0!)ps4a5rk5$TbJ7UsO7}-9^f9$mTjSZ)g@=D3fFW)Oe%t0X&HCwj84eD@k1Ggri zt^BOjTupq6LXtl+@>Lho0o|&6@WhESQm!Fh#av%hFpgFfv*jxQMpf=`D&D&ii*e{p zp*o`y6|p?7(0o|Jn96fMnF93XJ(y^dE9O_PDi#9>u0=N(_fg!r^dAuU57X0;U_~=0 zBMs;c^weFIib*P_6su`#Yi?<-npnPfz?xhp;*K_KHy@#)rGw_p^@6~<$>F!N+IInGOr_a~lj z1qhI4h%SJz7&pM{&#y-m+`cVeyB`nM~ zEX!YZk)@h)+%Y=32N6#VGznwp*%=$dA@8DUZfLjNG0kCHwN&MM@hCC27XvtH31(U#_dW%~2}%-1dI{%Y3bfZ>PB+=5 zm!==`CI<}NTary9d6klxmu&?3aBv_{f+e&F=*-lG6Jf9i4{UjPY{a;k>S}EsAnXti^D@`~9`FEQv(}hzVq%d~ zkxy5q-JUM&oVp}>)_mg8)#^ezPgm1*|>XW{8 z?VI%3FhOyVN!(~(bDy2#jUFoEsSg@KQa|xD)~q@%8WIr3t>9QXqrIM>Sw-;)l`B;s z4kj$*tx=p67ys~!1QVJt#1`HchS;KdQ4#ls-u5$N!A>@(Su{#Vif)qnvShElB$I4n zBy)nr2cmDy@BPQOX{>OSz+u zf#c87y7e|%|hjoePIGTuY|I2|26P63zuDXf&B3hZUGW+qSP z@m2T~&PSxOHkn061@e^@zxOro+BgGu-Mf5|UE9?BSf3uRV&0C(5aTW=x^1{8T4$p1 zS>K?GD4IU%VP4fL9#xNXS(p2-v1_X>%>b{#07O0QIlo(s3PEZFySx_-qKg$@W@wcy z7U**@`4u&7M;llgvZRWLSINJ>aZRf2l`?Xd5)&_vM9a)3lf^OQ*Qb8nwJVKsxEdot zHa)*h-1|gu;^IL>KcRJ2g4qf_arH&VfTlVd4v)&fM%vjL|M2r@!lQ!n{q}($3{Gao ze&oaScpoVmDWnLaWVr2Js^ucE;zZ+?wAGr~*|I`Yk;<7L+|<6&N;WkzUX}#wrvCmI zQEnPiH=sEp3V56qY+%8q$W`e#5pAoLTZ5PJnvdwfaH(`R&P}PYLu-rrHhUzu5?|8w zJ>Bb&S)we(j}ieu>sgzV-|kahP;ntO zve+?b8p4=?CTsjn@kJ}|ujV=wwidZ1T+A-+PU`3)bkjK-N?r!O&{>&|5Tt!!hM9n| z3Bc3;v7Gqt1M|yX)K9D$c>*HtQ+IyWI6+M8?|`KJ&$SH=2FezCkAMyqJ2#IDc22dl zThnq5RGyyY1~Ej}uy~diIvGiY$#MP^U@pkc=}(QLirw_#feB&3QKA8?k0;nC5kd6ayWP!;yo}M8-FNSDC-6&;Q|j z%r8wqgdPq!^51!WVEn+4)q~3!to$6FwZTDu14_adJQKAdzwy}LdF|hREwVB^sEkWf zv?)E)P7(_iX1q+E8$^7ChIwhebw#*&Jc$09{0#EEsV6^Xw^ zT-SxneFLVdh5-O7xp2rj9+1z0;QYX#Q0%R}>^XHrYoAfWNNrA3q`$=_yotKr?~(N| z+pdm~l0{acMpc7or7^=4MOiss{sZHhp3Y`SA!RsMLFN<1^4kwSK4eA4gkBeQn8MDt zOlCcZwxGyo2{+AoVc#Y+ZzGHr$zCufK{n2{P;&?%#1p^Ky~m^B4J!>EyY2%~@T3 zFdX^sNQXY5V7;Tx6CJsRLmgO*fSPBC3Vj#dGrNNx!m zP-`!(S!%V9Jj!JR@*Ay3t6Uln$AufBFQHG0q!$gdvU8C$Duo9r-BkJtD&%r&}%e3z^92#pvrjo@BP!0uuQ z0#Cp=PhZ(bjc2(YYu;ql{W`HD(;m>$lhi4zzAk6k^j!WQ!HEI7VTq1iQbJ>d4=$H+ zE39g1!{Fc``@`d7=@{%`O9*H?{+b3TjNUPBe{3P{}3 zw@iTg7abQeUUT#j3yNvnetb(g3^LD{bM{gPa-;A>pZFR=*OE~To3FMsUKpE7_A2Nq z(`NGYUer?CET5-^X1P8i{w;``IB2c_asvr%|LVpn6F4eg9yKc}6f05b-p~h)I8v^OCF9by@idg~8@k-p6ND#927uHWz`(Lqs5jJ~H>m zBkCXgl-=PUou+Go0Nj=Ku~PuDNe`hJYB<7Ldb2rR!WmV9Dg(z};2iNZElB*4dBmKBiA6 zvjaiPzv#VR+3Y9ZLnjB%ji((M@c4jfCrCTuV)MD2*XSbdGUi$eYHw_|cMO64}fP~QBHS_`x*r%8D9Led=JSBO;Pk=$Q*`+GsGebCz-W-zC#700 zf4dBsy6gIts|r(5U@s5+iL{WFPhXNIw`JMbx*nvJiAm=*2BjEnIv|u>P7mU4wU0Ml zwB}0qj?lPhzXp?S=9* z#>Ls1&e4e)-Hl6~*-oV#%i2y9vx9l9K$3+cuO9zYLx1-NP3eH zqHDpJ6UoZ6c;o77C0tIq)13WHs*Ot*K89PyK;9>s6O66~3XTG-&fDtb0D8Nk*qUd*BFgJOtB-u|7P7!8E zy>)@L zZ(ICoqqYx*fRW4xoQL$#Er3;K950Gj^iTLr7e8sMSK%puyaUM84A$JVTq+6>>n;oy zHM84&Oo6x^8!R7>pBgQX*VXhFemNr`jHXOCBYDcR_bw?<94uuJ7lfY5nVNFz+RB>#Ie=jsYNaITQa3h2Sgu-RMLAb_cvX6TJtieJ-3&SXoTo=h~7KB)Ji?LT9%-+ zd18ovrk3z(t>EoPri9Whn}+Z5SX%M-(uJ0I98eyFcg^?%I>;%H%a4a+@y;>Ci|X{PnmWN zsr!XZCN$(&2R*>zCtPGhw&L=YP;L~q)$WlTo%=0+8Shqn2!)CLMY_h#?oG8M^g`=AWt(Up$D$#Mgn(k4aAsm6dO zVk2t1y2E_9$19gA-A5;Q5*92z@~k-wOS6?ZX*HRwtS+9n{QSzFgs>H5x*>exo!u%& zrprnA{3vCOlnLjF^`a`-tPuT;(3iC`?nSH;nib{JJm9@g=XYN%Aig_5V;o%^3WPBXo!5HD-w7fVCY?ww6dZ^RXF z_@Svfo78K8I|Lihmz=ZfP3TmtyS>SShwEEtQX$d^e^n{nXFAoE96=jNchEF5KF4H!+U0-{g2!4BwnwT z#U>u;JLeu$z7|I7*#5R`^mNFEYt67=vo@)+B+EiiC7@!we^dFqy81C2Zp+%of@iKi zF-E7C?H<-o^e?%I^g5nl4(68vT~)7gtO2DY23hD%1UHnj-#9#Xb$D-b*xH4W%PS>QsXesn1{Cu^pXGC$JlTuBFNBeN`%`utpl9^vLJXt>XG6v~G5Bm5(QeWf3$LvA%!fgYG2y8XsC0NEckO9 zTG8xlH!$J@Ds(y_&wy7$rSTNr9yPQ8gNH-_z(Hpm_Lg*F>x5dxKw)l&NmW%?*t^KM zS8kg0*&BL|tP`d~Vj#zyc57RDMhqjDqGy5v!uY&VaxZlmUNizl*4`;AhMnGi+2M@3 zajjK8N;dl(VpJLCN1UUy)3L>l#Td%N!pXAKtw<}aL;u@|Ieo!6S2OOpcUQzY;?hL8 zg?dM!I?g!z9r4ASQRZhNE(rn74|gNeYQi2|jpa$$6n;NZA07G*mtdaWpuUlnCdQe^ zYan@_MP`ajcYAC3U0NyyxO!$K1>$n&j@ zEvM4@kk9JMliEf1AjaFM&2&2J-a`E|uOS9rGii2@`Ht-u2Nfnh&)MR(JpCxSmXXD0D(phf-70yzgvQFA!B(i|$GcEYgs z@$rUHI$0>6)xBFl8l1evsxsd_b@oXsYY!0r63K0{t2a{sg-t8fZ6#SkuS;6uQF4bk zKHD>TmW21mlS!InYElDI?yltF8CATGbe5j?fnTuqF)E;5=L!_&+FoK=YfXGVW&&y% z#_HSbu8XgazNLhkGE*&U2pw--vFWpGl$Q8SVi1ttNPE5ZCM?nJ=;W4w287YA;Et`gjQhZ*5~@##2gtWVUv{?4w^P%Qblvqg%vQ= zJ=fcnF6~6shR`A0BfVr+zl*&Eq4Lqf`?WRIWqx4hiS1&WN$?tXX7c+xFTl}7%0xAZBZBLX`p7hLtC-}NeG!rHq?^P}n z)$gcJyDU|>_~0GTA_@yCse^iRCX=Gp!k2)QB?G$Q#m3yBxLXstk+bPIo{bbhB7u-t zn?YCfyFwHq)?oX3Iigk9H(L{;a^bm~WX~V=_-`pGul$!A7XFv(zJ>qV?BhHD)C9X_ zhalTj5TFm0(<&XP+yAWH1OKY^r!GJH@aRolrP-Czx2^6nzwMJ$eLQv%vQ~JUBm*&vcWTX^AvntJ zfxSM|@8{4tD5Dx&(9D=HZfzp7S%WDz)Vz0dpkey*+!gLSyT3m zoO3J^vcu%UD(b|!b`b;(4>_#UYJ=N3dmnP>zg#A26}Pw;0Py^PgxVbumv@EOG?!q(^qhT^zE{>qGe2+IW~1WW2X8HC)oNwkR4QE4%0^nIQA_P0WN@?YR;m)SyruW5o!b7@jZi%YQb>rK2 zVVX^TJyd+IPorG%GwWA`y7xjLVJP(lI}$fJ)e##p^>pSl{Je1fA^b_C8Q0r3AX#BlR9m>MZ95tVPERXm?8R?YE!#RuB+~dxC3fqlAX<1-|H8 z^EkdMkbMHaHq(#EQqqxTQy1AcR0tt8D{+wYb2}3zVx=Z2l-dR=rMjRQugq&{~F1)H`5q$VqG-uB5@= z@SN`U(}uA%pvBjUKwj=Fn3lM0-iObfY9br|qeF7%W9xedPMohZq#aBav|=)ABC9gu z4j)gi*EPO|GVh8HNG|bDwi0UXXFTdCX{Qc&D&?EvgyBLC{5_yO+@Z+LXLC?Ba!#!z zseY8Pm8XUS?`+^8;yd}e6QVLuM^80d<*hTjT=rzU=lyN15;;f4DdwD1-OkIW&L}IK zHj$XvNMOVTe|A$)O%IU)znc|PnHPn);vX1H5*2`b{W&nluSu67_Nx39R3$BhZQxq;AHrk%*>P#d7+phr z72BX_cfx&K%rOFtQ+SX6z}Wf>C}oSQcn|6!H1o5pkOxK5@SpluI5b~>kN{9rdK|>! zVf{^Nqb8f3^}|Gin@#N}d3Rj)k)<=7PRQ3~`(Cyq);60Lp|Ko}b*nm6x4&J^DXxow z3*{E(Zdy{81AkvG_!E}*n%D!B&O3v_7Q0*hg~+G+v+{f1uT~!_!`wfiy5^K=A9dPyVqIzoa|)ga~*{K?y4LRQFJC#*NyGnEw8*3-=unDlZQw0;!VzG^5crH2OJk2?-mQM9}?s|7cQ`~ z>w8J%A!cey=-<`u5*EBfl-D}_Ch^n_vK9t8>ndatWv8vMuRnZSv*xcqwBvPmi6#-K z(wW0WCMs-r`2-yOLcC(1hV%9^_$aD;7dJ0Co{keC5TkWdt?kMM61u$&D^U6(nL_<gYm4mAA9EfL^QyGjt;)Ut3*Q3pu&D(DdF9@VCR;7NSZY~@3diN(N}mK*)4{P)6s_yq^$fUA(O z#CO%s%^RY|(u67xG_uCpNBQ6f#NT&qfh?qCDZy2Imco~6wSZXb!)Fqk$w^>-I9zz% z(4k&x{*s_OlqLgwZ&|1nPkqHYqHPHj?&PG0APNvt_fuupvp993e^FK$GA%!q^hz~P zAC4Jz374ovN2j1v>b>>VDoYid+^}rK-eHpP)nJ13HZ2ONh0lle>C^cZw|6!5ybe~r z&(}+t>}P9y9ZP_#4n?ciC=fa}EMO=$FGS$&I*ugfZYGnIA{}iOt|+udU)Kr>==x18 zxkpKf$#mgF4<+pR{Ou~paPMJ}kX{$Ll~#aF-N~V{)(=nSG=kCvxx#(jA zrt#q(-bqO4j!TKMiOq1Fq@rM6a>`{$>f+$}ZRf#DoWY;1yp<19wk+3|_r#1-joo;9 zrc@5HlCmvT+X*^$>@lWOUejVGWv7zICO@b+z^=(OL;^Vps}{D`02$&-z z+{JjYC(MaEZ}kXx0)pbc3T!V?4FFJ`_-h3ymBr)a7H{>I$gCvKd)XIsy(;24w&85s znDiO+Y=Q^SU}qa?H86p>0K8JqsUNF~!u1W4Sm;^QBeLU-=ynWOjet*&p38!zk}Fyi z4(@Z6bFwVSk22ScS18OPXz)z&R^;Q41S^TIFBJ(dmeK%Q>U=oe^rnnPQp&1BW_P0W zb^A&bGjX0sGKRgy-F<*L784Vw;U_MYSRY+if|ckcsR!b);WC}15RHT!i8clMuvYB% z{oKg$UTb95Ud7iFlS-{_ezeFK|_T{IDQ5sGs+RuRsQYfBX@!G8^ znebgHq3-7Sw+hUJ_!S0Qoj`h%B7A4t=Mly_}Q z=z)O4UeiYTmA>H#_AtiSeUn+M`xumKX^Rp z5_SZR(Sg};S)l{%P^>--fD3X4LvmFEtt(ED7FWO}Io|_@P5ic?XhFjnvKlobgpV=2 zZ?(MeRjYnEu(TB-a^I2WH5LBXP5iG|3_1DZ@A80y;hu#rG#ST%;Q~{G>v4)8m$`lZ z8$NjzO{(MJ8-wvTn7l`iGc)XJeLjWq`uysC!cZJQ+dIeH$z;>VV#%UOl=A+Ny78{y zKfO`z^M@4wgEtzox(L8h*xEAd%m)y3?F?%WMnUk`;GK>1&V*GI))}TG*#@`dt}fG! z!?q9;umz`8JO|gDrHWcXT*#akaf;<02F43}HZrHN+;K~-4=sNsgqnd{XRDw?f$L4z zd+hlEHUjhEJ#UG`8rv{Wt~Xq-;ihmaQyp?PWKbUE%c&qJRLRiL1%#?)~i@`ei3tR@^uDKwf^{rtIO;7|v z&KOiqcsbr5d+4yh{vgHdx&nm;p{8SZ$E;5bd}uL%x>T|EPza2Ch5`YT@q3_sL%^hU zY{|(VV8<7oFM+EqsIOz?9Ee%RQ&DMS%wNDY4a`E9(Crh}V(JXHxHu}f!S|<@KmNd| zUOP0A4q1O*$i7ulBSWTTCa(>ehn!QL6knX%UE$@Y6hQ!o^pJN8*e5mz_@T!xk?rA= z^ocTLnqM#h6t3rhV|v)njh34MyiO?Kbw2xVz0Q;dU@Lt=R()N}tEzVd-AqoJbif4n zY(4{uS%UTJ{w8W7_R{Wz;Yl9E8R*G&?~p$Rd#osKpO|s!OyfYsBxIkSy@z}k`Ps-d z{8x&(I%yuJ!=}K!7w5?kP?#65^UC?2D`SMzJ?v3iTtX3AckkLO_X2KgH|UcQl<2Ff zC!6fc@7o_zdo9NhwOxPmVoA)y^de!aUUP9qN%jCcCT&CSsb0WDN|eY4F0YSu!eTq( zxf6Pw;hBwoqJ#LlQq}`lrA1k^X=is`j*hc?3c!_dK+nt*=>I~nR*sWgpTU{pk1zeX zqb`}X!qZKKuwLt27m|oc*D+#&UgES(NZbR*j_lb4rmyn=h?*CVAAr-LiZi)UB@Mzy zoUVWipHW7!(20UrHl6H43O&T#J6oYh^f+V-a7guzJ+@(OZczKYvpJx_Y@|Tba}RkG zz44r>&kHa+3?745UOzCjin40#a3P(uWk75J8qd!y$oH~7v7c*^5m(nN9WDr8O$9|` zyjI@fP`d4lff1#!eZvIkZ~9$bfGePAQV^Xf3gPmu`#9F)ztu7Z<5RpU~l3i2wM-5cTE+$uso|EZZZ%X=`l_<0el?mon7|8EXdkv9osf7X%RX&m2KcPaEt5NZLIlgu$LQp z4EssI5;4zN$b9WjwwcI^%yX+(UTwEWbzMPCqqd zwiz@L9g#pmU33~b1nhmmk-D*max;LySE66ibR*6of0era0<>@Un|JX?t($9M zvy&E7iz*?b$yt|Bm5Wo zr8RY)c8p+ulGO+Bx8m2x4RU&?z+Fe=6J7#h2FBgh=zes+P7q3I4PZ(Z?l0B@+o-7q z_W`SHz^?DL{=hhzJ^F!h1I~c5O*C1A(prjmH|$jT7bFk_Pv~@yOWv37T6*mwqH%Uc zZT)=H#lp<3`TV0C_h*V^32VXwmG;}zr5oq8L4UnC{^GD>_TKLti?#7)NtI7o$`cs- z{p_j62u&4BlLtGqMV78?XHd%XQJ+4eX01{_{K~AyFWAlP0V#`jw;TLsL zAhzW6!CDc|Ypww&mRu$L{RKM&^xqn!V36zlv*7TNJ}*PO_g|M$DTGeU-xBpm(%>( z`<0ni3sa&t&sVtkAIuYott)6p(|)uUy;Q}HA@4%omOrqKw()A@7@r?|9ra>jhamR* z=k>RtR~8KN2M)-OwFz6dhzW5k6Y196Z9WeOnCv+M2pq*}f9XX351;=RO&ftb)cV3p zQnBWfk|)gCiEaW*9Y%NdlOjU7-*FYukLxPu=IZ%QzVe_rD+Z_e3RM^Ihf8-RurxVl zihp{%QP%1}Gf4B<)a-elOY&Rnw><3Z#{06lMV3bVnlgc$inBwy&vxD3<-4W_P1-_#Js-Q!XV*>FdRrR&5QQ6$JiQLBVQj=#fe>S zyal)6sHdYn6{UGL&Q-(}^d+b_9Mzv4xF#Be$(n#TzzGDP{i(sOBEjYHPIv{GwLA}S zx6jv~c{qMCnXL)8g#zIL;f=>B z=T(oI*CMRGJ7t(3dosU}=B$75bK)R?xU~si%8_<>gNdYjHG{>S@WJ`OQ<+Lk*r`>{ zKXT#ND(Fmf%*resz_`(E3@6_q0>x^7y3FO6E69AK57lozKBAulEk5{LDpt0mDz81w z9IhAEqx(`hk6V{=KTo`yk+@K2KQNRlG$qaG79xDKVtfm_7=QlqnD)V`nnC|=YPMyv zXyvFK_s)y5VZT`ej!`!ExSOzt_Fz$2gUg+wPBf`FG1_0m6@7yhIq_Ia&QNz&YujR2 z7q%a{4mqY8ZQ6eb>VP8Uk3koN8*@l9*pi0KaI;%De>QT|obO>lT0OcG#J|YPGWTdL zfN?F;xt33YPen#wzkc|BDZ-UAgNGx2$FD79?PNz{PVK;2;_ir-bCyNiV}y5{l^of* zVOw;cuTPAXa`l*vGv%8xiTf7Xm3fZ$tIPfFEzt?$;gNOeC(V$@1L}|-@Y1A`fPmt9 zG|K#E(*YpxDFaEsM3MrUFSPb{n;xGLtn%#%bP^6N3HFZXexjt*@%FSRx{MQHvkAtKd?OLhCPTtwdliL zroi}MlrK7FP8%3|cQXg_y)+xovYMtrKl8o6c-Yd;jyGd6;M%0P^8tiW!s~h13aI=f zoE~KKTN)@2X4PCak~*keJ68K*AaNlyv5RwOml(0vSNa}%2Djy{+#k+#NZz4QP-;*d z=be5SrcU1kg~-+RYwn7m+n;PpB(d*8_l>}H)SFw6+(MsVV(O0i0!tP%fPQz-0uzt~ zSm(`#%s+>O-{Z1?$D?EorRp8$+eb30B9OffEBrmNVRY#^w8uqjzw!(G;o_7ys7)j* zGS>plGwRe^$e&?89lS=Ub8V=auHj-RfKx3q``gmW7^6t|S&$&e5c<)M7P0|HzMkF`h`Z`cLg_^-E()h|-){n{?)yMeBuEOS+Lcv>X! z!9&V>d5{)4K+6xAJp%@%{`D|73~A>AxVV@qYKHcfxgi+}r)Dl&9BzHjAuo#O_PRrJQG{#3_^{H!nGm;~-97pGIav$&LirAGnG zs#<;Gkt`Dwef=|o!31$Ix3x3(?Ttpdy80@=0S@E2^N~Cp4Nx64dSGoCDs%Q1;U>|e zt1xrdFUgj9^WRfVEor#LZO2&jcxT+s ziHbw&YI918txHyv)M293Oa1lhBC89dvMe{a$|%?IapbMiP9YlHU%4kdNzx;RW%>d z9Okf`>6ZY*-%ZmC^vpt?u51U@ zOJiwvYQ0eo&k8)E39E6+NHZ>kjx)xGRbd=O&Y#p|OkF6Oh(Jr1nt+7(Us(-r^`oiL zhu(^>FQNAG5>0FT8|0Iq;QiSJ&Aq0j0dS_y-+cTG;>Y_zDDN%M$X34$8THOBx}B>60X@>7Nxq)O z*HaNhc+2;0BbzGBG>y{5snu6fB8posSWmIv*_rOGPJiij7Ic3D_>2e4OHT?rym4v4!<1L+k(*`I*bg2g)r&I(vjl|5a8xEh4VX zn+-ke*f;}ZTdQcjzID?c843c}+6-6NPJ1YP@Ii`2f^+y{(ygrvtWyV2j7xRGh~4=L z4_(*X=e6$%iNC(j&q(t(t$&jcO1#;sKm4NM@wqx}#(r{%VzyQ4#UuV~gM>{Hm$;?1 zzyb7x6{oXu>fLz#-dagPd*W+Dm?vP5Ep}mc-qJ7UX;Pj5TonbwNq19Zjs)coj4m8B zLhvZpnL_8dH`^pY#{kwfDna$o@R|aH`*-L-W#&FFl%`fR;Xzu?qS)LbLTzsWm=~ z_ndR)VvF^w9fR7V1)ST3jcgn2SR@4wxuf$!s*i+u0K{P{ErD7m)lH)ZqjtM+(Ot~W zTRv{*(zk>b1C>=S*m?ik3*}m0-D3i5xohX$&faXR8E>X~nJd8*)aYz2e&2z3N5JCE z>x=EqGa~E}_bU^36h3FE*U%Kd!QAhcTSx+$F;es`G1z*VGuE^3-d7JY?xXd5F{a=w zn9f~l1?Mh;bj<(^o4IWRY_&|F2e;WoL(rUo7cXFStY7`t#Ox>aBPmyF^Q003o%16=@DODUmPBA}$OkkTq=i2Pu(`KY0Gs<1NQMI>#%qLbfvzrrTIWxZ{l~+|sir8!j`^7k zH>JKCq}7LPj32yh^GJd0sdVB~;mB&KFvo6&KucKb_i_eOH?EdL6(*IaRsKURyoxZ` zV>jl5u`odteg+&Rw{WGd;$R=G_(Cg|*&MCL_?$d=SJv%V?Bsf$8^~yg4@#_auGULd zrYCh8Adl9HG{LA82RqJPhm|E&@}7P=-*Ywr%5(o;&Oz;|Cq4{8lN92heG}}^*IAM3 zi~HzMtZqILSr~1}S5v1-$sTVhOn=>`{ls-NkkFZHLN_q{@;!Y|_KVYU&-C;n*Y?_! z%CDDLILf5*bi zB&r`#L{?+W7skG5(%Xw!ylQ6Oy>=ItNlhwO&dA+{fW4&>AC53 zgylDz+lp)3ABJVYmO<#--V76v%QFDL%n#O9*t64vcAh}ZYt=C7p(AUAQjd zlTh_Wn5o{Ki&QH@f?3^uyRJbV2dW%tsEuN|00EP(XM2D%pfZ~1anE8bEA3N4&2R6o z24vzV{DDa)6+vKJ@HUDq=+|G9`}IL-Qe)e=B7FuJ?O&iCdP{uDlGq74*n}m;dun60bC5#&2U_yC0i$Jq?^wLHF3DMEH~nPN z1C^*SwqQTz1cuSVo&sIPC*;es4tV0TZA-kHKEgKA|` ztlhEBq2>oup-xOJ0D;w|>V)C8o2O>EK->qu%I%&(lE1UnQj>K13pXbHA*UTq+pxEC zZhQ7oa(CB# zL2V$}l%p{%Bkd!-l%Wo?`YTiD zmLS)ZUSDmyJ{GiBW!M5Z^e>ua>Zh>*$n=YL(lO$x91;^=;FMFR7WiD+PNL&z9>gQq z^Ti}t9>uHOv>-TqWd_XCLasb|;a)NeS6a=5Ju2K$)H;#r=U&Yj>?CtkU1w`KWNhl& z%6TJYWzu5T?z~$x#{NFr7@5`3!yV~}4yBTKwXd+0OKFEBcbfb4^?~{xw$%l8%tC`$ zCvAxeXHpa5y4+!TDCLckP0wT`IS_gOBJ;;0j*Y>0wP0*@+T6eoemAqA-^5Gx*R!wx z_5c`Na)1u9U#N+oHX>lVjPEi$v>B08_ex=}aL7{)CJ!1TaC8T+028$}PE$-}VCa7* z+H@fiDxfNa67eZ# zBF9#^R>Eo+-_8@CL;R9+^Pjk>9~v;(16%(Qi~f3C&b5 zS!Wy5`qBWr@+rfz&ZD}L;M_suI6L~!lttZMHW@cC8oo7enrx|z z!k(Jpf7(4y;QRgB|I;`8uha^;%f!G$2l2fD1dUrMI1V4nAUXT`SHKkgQwkygbgjH< zKbm;hS9}b#6L41lspon#bj+olSYVDKQa-9TtiJ5A`}5Jp2iCQ@TDP*ham9nq(atcA zgwBv`oeXI^voGgLbDKb*mCQbZ5=`yP)J=Pj@ouCd%$*{>J-B#>FQQ{JJP#)&^XFX7 zf0zjW`xN7U_4R*|DF3%u1AoE37-$$Sw3C|k2PuAoeY-Qz=F46T0d6 zsdW(B4gf?eHsM8rB|@M20CTxAvex5a-t3)y4GMaaVKc@zO1QKKThs~1;!j3*E^Nk= zuVCf@i*sQ+)3^g5YN#yWSeH79gH%Y}(?$?NV*S-KB~RAq%0e(!nijD9hkP_3EqFL7 zsQk)*elQ!7bei7(uxKI#W6ipr)y(=^h@Xs|G?qQ1hUyK4f8vyrD;_CTHausRf@K;4 zjfU({EJRmZ4!JStV#ku{r zPX=~^$&2Hx7*>#71IlrKQsm1%w|{E=eyMV^*^aBl=w%ft@rF^Et_1BcPS7#*D<10L z^uy|0ncYRVk+KPE7RqzcR~WDgf4Ms%)<6HWq8Oen@t;(26!5uK{_foiBKDN3ijc|c z9Vr@Ps(%M%yM7Dz8?Bb%X@G@iRM)i&twOSEol5hbX4bwSw$Oil%RGX)>&+atxS01? zc6JPRcy!vn)|<|Y#0*aOinV_8x#@S-d_mrg$@9Hs4XQz2^LIi+TNC6?aRQHLmsG$Nd&V|Bfo{#qT)30o*uJx}TI(La0H?o|fjz6hgTmnUjJa^|z{1 zjxFb_nkS_X`1t{BCqv|u5f_dW!U^~N`!zpl3!LQdv;{n(_oP3mVFQE?0wwe7FAtgm zUtqj)s>6XkRX1cnF@E$Q?y0>g{geXDI9y>2dO51x**gj2?Ne!R><~Ho_ssA&xdG$& zai+hJCx;J9Ql09F7k;F;vF5SE(7>*EMO-M8wy9mb&#KxOb-GYAf!Hr>RqT+ImcX@M z3nNs;fjAU=M{H?%D3_AByG4mTEdTiJab2>1PQL5yP8%D}wrrBNfbUB}7^d)#-6`$4 z$f;&H1;qs{(LIx|wA84p+B8JsvzS>^vBxp=j>W;I5BZqQO6=nOL%K65cG*eQXFS54 zv=PVGAWCC9&PiNT&r9E)Rz3c3?^ULP`lx?bzVo7oGQ}MqoPF8Qm$`F@>|j&l3Il>b zZfXwCkf-43fyeDm;V*#G6%!*YT;U@Yh;Ok3p`!V5U)b4fNguwI7l&G6Yy*hT=D0ye-V+FG=4d;}+Vx=sI$2 z!*;l&J4#+JS$R4-ttc@v(HCalu7DLG3oa>&qT>P+pk%5>%DtZ)>u-u^wiVN@j1Huk zvR8t?L=pG*V|ZwiFW#|NSvdPzeZY={DU6S-h+UcNa5}$sh`lVpyp=ceNaIyQ<8K2N z_=j{Xm+cTT{+n8>t!;KG<}TVQHiRx;0~1~~=zcyF_3;*k7nR^=e)h%aX032nh7s}8 zs65-#N4hSqe2qAiX09yA=dWvY|3-z`^h&M}gM(t0d!He!Jc(O-JmUa6D6i|K(g(&V z&iXGV%+9nC+3-w{fbp4%m^uT)X&F0w?~VKRArct}Mp|O-OQ0K@u>#3rqa0bIMJu;h zaZ47iRImpEIy@r%vrAN_QP*7X4dnk^Ckx=#I6#|Ljh;s zpH*ZN^zB@N5Yc^uS<`F$9*e64^34EnzN94b0cqH&r%0>u^A-;#AGJg+48*R`P5x z0aqnU;>&UNFg+4~gx6ijD23tw(SyXuI2*+YEFQnE>wHt(j-Mv+E+2fPgeM-~>~2|y z=(SGrPAI9R3(!y87)eO+_J04)+^IdxwJ&n-C4BwyaSE|rXd5*LyrYBDs70yS-ezQl z5Gc{reHrhy%J=N@+R|N2z9&C-JCKUboqX-g7zf2f9FROP^-SLG-urUb%{JxIcOU;a zi3a34;M(Ii!X3IVmCq%bgj0wW{*9A$ohHA|iruBUt~vx)eNTZsK<$JzBtLr#AyKfM zU|VS7vB}eGV;^-({50BtyDguXu)lEqESVFIwOU-! zj;M9%2L{fpn|doE@FVrrtJswt>e+{59|dwio_>x$n{+Q((q9usb#kuMialJFFJ$uz zy;|gLKFcY;vQiueQao=qU9{@}WYo!QU4bi?=#WE0=~fq?2UjWlG@MFau_PNn1XcYu z|1l+khE9@I=#*TbVPS?Hh#kmL6u|a3zESiHN9Eb$Z!roKm^mAB0FL37AR;^>-tIK~ z$>Hn>?NT4G_{88!;d(WQ!L8C$ZNZ`*l>c<%thV0esN1@mtt%*R4L0(%fG^ZI72?tQ)cRnsDb#KM)=VuL(jHYkj|4P z9Sl5r@1gf12!sFufjo@3ac1AnTi<@z+1*e3>CW6Q_s-liXYM)Ye|`s;?$+Q^98b38lKbj4DrR573PBX@^U<5BMDG1rF6>d4>#$f`L#W?NE zvn@x=6&_|Y>5t0}f*{T|P_qw}l6|pno+qg5cbLV(`{6j}hUy07D)p9_G>r){UbHK} ziWpKDkYz*gcoLpe9DhPl-ZpyeWvhQZa2K zY#w{<$oQVj)>aJQ$d#m#Gdo*4(N%t*q3Wht1~l1JfQ%!E8*e+NZ<5{JDRV%BIHjTV z(Iyeo@eBdOWZR%5mqL(bvigrleObR+SF~Pj6?>JfsBB&U5lFIHV#M>0kQ~Wq(&^m< zz1oNC&nP=y^a~Hk5JNS?mDHa%Zsf1>QnVJi)#>{@!S*4w4Qa`2jBqO7<5N`1OfuXS zK5$>*m5xYl`(9*caBnto<9N7*#g?l|UMVClZIC6uq(tili;jRuma#OX+>+oDv4~v} z9Ex17_Z(oJ28EwMU%|!aD@dn?G#)fB!J_YnE8#e+M^gf8z42S@hi99LrFVbd>TaiR z;%5NCMkO(z_dEK}X;t6Y_47=sD)2o@aq}c}U>jwwCQA-ywGXCBt z!tdDK@#{WG>#sc38Gcxzn>$<|Qu@wabJV$!z3#R&GCiz9<}mcCH>P~WKOEfpZgWw1 zf(5$Zz6{~ZfN);CX3Z}W8AsJ^h?w}V;X3~+Ed4h_Qc;lgK0fP(0M>RFGO~5z8pxX5 z^4NC>=)|LlPbfu@|H@Xd2t@AUcAejW-k6&?uWPz)OO*RN&w3yb*o=rI+z8K!;XVHl zCRz-2WGob`oYyvZO}*~7Xk$v%&iqWL(CUao2$(Xa!h983n=$n6{kBpk>1H+Tz_xLf zO4aYzB6iim2hjbpJ}6>P0ns?{RJLC^Q%*5itE3|-vd(Y*Eq+L5N_KJd7FW|r#EwVU z2X0vg?V&ZD*xowyq?x(XECsCZl?I(mxM&{wTtjUtt+rzdoz&4e`|)H$8g|}x8(|f> z49;z(k`WD&K^JgR`I`s*C&2}DQZ&tOL#X9|CwOOH)NR)<#kRCq59M2OIQKV0@e--wr#1Q{nNgjsvf4;8gPXB zI;QT)f@fr5dcNO>6M%17WengGK9tv0uEL^xRx+R5>U1`Wcqzyr*g*tIU&6%kL~>KP zE)gHJ8&BC!l3HByNEOA5RfRza7rvp zi&8WbWA!qM=n66q{1cr22i$Fj*%_7Vm11w2V2J1iXUcaiU3l;j`=yjw<<82?b=T4k zfBGvUkGJ!9ehXJ?;1->Fn9U?G0WN0?zK6k<)}UxrRI_KT#XEA~DXD>dUjP){Cgc8l zW7R5VE5pp2i%JF@MB_1WO^7|q$*E5dNV;gV&wzb7)P@l;wwcYuKmO%AfQ@AE8+ND0 z9eBn}25egm)v@v;q7vaq+ZaF>&B`8n0vAqY2xtKU$OAF{h@1YrGS7n`?jl0{(idr( zMSH>J$fubntyx6v8UYrLYI6gh$CeXqY6qr?d2tuL@EJ#^LcqSXPTnh*p=Uj6=V23? zfwzsmIs~$JsSJeH4--u!nmMMFD(orV)(b>&ei|7$KOJx5igsW%qEDvepoZ*94zcvMA=V*E4M1yk;JUq3Jr`*C*b zo^6w5DBu2B)+*}Kn(p|TOU>I0cm?NcpQKb@fGIxvECic=9C04qdxDbg3{K)YotmA9 zPNf;Zu_-RZ9eoqkS$5KvjGDCn?22~CJK-D#{~pTWHf0rdtiO~X`?}iuBgb5SoNn(o ztb{ysylDMc^+#3KyS3OSL@l}09uSJgUj218Jy|jG7fxrOOB^`tlB?b2M;Pb4hmqE` zQ1vR+iWH+&MRn=VxM`jv)M}5~HDWwjwLZ1f7E4u|^0czMKyVCm;4d*a!U*U#$~f*5 z0!CoL3^t)sZaneGE@+e z)cUSHC`ZncZQpMX3?3*-hU&xjbwQm<)V{@acS$bTuH2A%ZHeR46X#jS3 zg}S#zAU?NDxbBQ=!iCs-zg)-tJK$1Jb%XfgvhHtU&ugjYcXt)LL@{KAs9oF5fTfik z^HVNcy0fXnGW5gpRgCVW{sMg>6E3D{ITO^yVbJ03jk6gu);&sx>0oM;0}U&S+W>*8 zerN3ER^%L7Sd<#?go^C)M#Tg^bkN)5dYVgm`N40u zR36%aSB6ewnJYfzR@qq$;Lrm|WOOE)NJf$L9ryx2rT6k2d4GH((ozfpJyXP6p2DID zEW<5zp*-;(l}lSmzyiD9d;0y)-#B?BeV4FJbAPt!K@8x4JBW;%jln4Utl$!&HAQ37 zNresQV>zVExBZ)QMXhU#O{{HJVrZ`Y)U@1wsSGI87xjsob}zx>?#y2wKswmS*{g_X z0;{Y}7^ttWPE|;{_ShkhBB|x`WP>sj(Av$8S{x(z{>F@?S|4+>{h4V*op4_wlfouq zgq*s(fhB1WWHiS>BOCk1+w!%We36Fn1T6_bTNMz$D}y)QBKwY-5c!qpm*O{^Fa)03 zN__<#3cWamTiw7o^iv<(!aD$m0A+*5F${a|v~B0Q4SZIItGHfnl8%D?TqF^;@b}R> eKmA1hNT~g%>;D#<{|_AJPhR~0aG&D)%-;aO_`&7? literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-17-Attention Mechanism Matrix Multiplication.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-17-Attention Mechanism Matrix Multiplication.jpg new file mode 100644 index 0000000000000000000000000000000000000000..09d76992b63f9a455d9bbd9e44a637d3b5a4d1be GIT binary patch literal 20207 zcmeHu2UJtrw)UoXPz32jih$CaN)3v1k)l#UQ;;qoy&D7p=}kaEsnWZ2q$o&LdXvyQ z(u0%$A^ErG-g}P6bKbe$``#P>|BrFw*pi*By|d<8bA9uh-&`B}89M`(Doh=_=UgqW0^hJu`ojGTe`3>6IvBZQTOk(v1%rvUdkc0LYf zW}eFz_=JST#Ka)n5;9jrqybwT-ijtDC!rr&qv}z@Xr#A)&Ex&*Bpj zpT9`T$b6lZos*lF|E9d6vZ}hKwywG5Lu*@mM`zc-;Lz~M=;txS?A-jq;?nZUmsRA> z*WJDS1JvQsH@k2EyuX64ZPkNS9Z1z0zn`ysT_H%~$|67**YS?dfApmkb9B}gR zP6JS2k3vclY5%4=@lxDM>3F-ZvmK+F9k1-?Y>)Sx%7iyR?QZExr!r!J-u=f|z|R?u zL`a{(0_SX}kI8MIN4oqjE;v1&vf@}EbpsCz083b4zT$`T%Km;i=H<9~lS51&beqxv zBc=5Hi4hCUe{S*jf%(r^{N^P88H@jn#eeYPKiK(eT=`EN`7Or&Z7lrZ3l!d1;9i6h z7EoO*j;!s)#{wOU2bdPdC`|lEBNpHxD=VA<6{e)FC9QtXCbnCokXz;JCH(kkpDC-} z$=%{6<*K2L2M8Ghtsu-yExTTkj~^W#p8e9Y#InrVnb{`v(<=Dh)Eanu&f;P}Ql541Kgz|UPFqbul#_Kq>vc@8Tg+)DcdUi)wJ9ybT*C9G zJ+B^x0&N5JCbN@n>j-PuaKz#}dA3lJfJ=RH+<2?JHnqIFw^3PX8=9h7Q@fDefvGwK zz>^XB!|_7+*;TeD7ZR;MGi9$krrH3$v^^diA4=v*D6&pIkU6-;b`pFdRXN=eew7(R zgv>_Ce@JBIW-RKjj+7!6G)4?4aK4DYA^>N7zMDpw@S_obdVY~N>^VQ1b+JVxed3mM zIIYr95|9w8VD3u(S|ld=Y6pA5uQMp|qqjJxkYIsm@vV~}amWd8LY*YAD9Z%z2s6h5 zM#b>m4As@OXDB-?zy6?s>k=8+FPkZ~huCm+F{IyyN#bp)g?C0mJsit?pIZ zD>JkFHM+CycJBr@$UlR`Ume$HMb7ztZJ*VzR;@~9wAbkcjIH?3&y`&3LMh6k=KQ2sI6li%7nsf}eUbD6s{@R)G|F=}JbtNs#VT_EF3mKMhVD&}rheEyK91 z8~3gek_a1|@L=p&$(n2B?sZHw<$OqVlx0dNHP;wXd1=F%q;jXbR?c(>S#5u%P0G~o zipdkm4)-J$_^iE!1r!;#)}TuLVjG0r8)6^f>_qWP=2P8?W(g9e8eA1TKt|{v4!mbW z@P#}p$Ua#k7UV^ z?6AAYvPu8+g;L{Ok#X283x$uwtjuLqlLiOe5UPWhJMBnQ>!YT?3L`HppwT$At~0l+ z5%;OmXyuAT*>K`DKb;_iu95_AP1w*O-wc1`^7lG z*wu3TX#B&YbED!V3|7c)&I4Rtdh_VZiS_}cj}+fcI62BOLdh!`a<)Hyp({#0IqN>S z+^c;G9f2x_Ufke)+=pfnd?I_UQy?IR&zNP1uX*C4GT`sRZH>4&AMMy?+xRO?Q98ee z?hrO$fzq~2ERf|is${tQBi#E*Xar5~uzkh?qXtnYkn^stT|SqvKn!IIrp^$)a>!fo z>+MJ-+hQzBtm!0;)9I3JodP)xXXb}kKe*DqTZK^8Yx(AkTT{#nxg6`CRr5rA3ahC) znNTjqIp-M5JIIqpY56XNrFOd{C{Cr$)|&N#VaUwni^mdE$4%h}3!z_f|HYud8#*G( zW$i^%jf(}8kob;Qxf4bqVRjlzYB|QR!~QmtE&9(Q@k9er_-NT7 zY3^JbuC-fHwktTMxgPGA_tn!!EDf$MfR+Ny|I~rgk>aa`pY?s)Gz1Gqraj3KT!W?|I(D~ z*;4Qq>Fh6_v(0;QimsabTJ_7kqH)9(fJYo zjHTw`5pPiPt-3Iw0qU7+IMfWpJqq5mK7q(`jA~F8W)*%B*o4n!abbby~ z+pUqseNWZ9%02Hc)0;xG9*jb^%P}h1Ay~jL6$}I7ZvT=`{;c1De)gql7j!ejoEDSA zQ)1JF(?=3(zk%2vXxyJNpx>w-rAZqrb3GFO2tOfKF5u?Se0^fK>1FnS=o%vK6x7j> ztnl`$sS@{`H7}ZNgzb^biy-}w(;pd*`#b$^4#-5U8M)E`9pOQoBKd{MirW#Ym|knxjvk`5?xokBtxj@ z2R(Dg##SxwLCLwg$D!TiVeCTd?-n!(hnoV#Js)|;SC>}NMVOuMUTjukcIM&Cljx;i z#i2E8PoW8Zk63JcMX}OAt>604!qD#gV_^EG|L}({9DxA*@H6dMfb%zld{ck$EZTN7 zAGkoQGvK4#AjOO@3jEdR?+4(&djAIwyZi^evdae?=oq{kN8MN;(-<7;Z@AD29l!z` zC71}W)t}=DuPttCe|mr0N#zP|LhTkynF;RNjo1GChW=8Tgrnm{(B)OCV-(M^YYzJ4 zDi(kf_&`q5!CVdM4msGy0uc}dUDMYCNAMdeR3Fh4#J&QQP;;SMGnpGBsbBYqPSCX5 z@cqCo^vMK#x&i37>wi^p76$&VXiKFxY z^ON2Ft^dFEup|HrR0zVCH|H@J2r~2d+5zU#CUgsAbd11p(?7aT)AXxJB_S@nYgI`o z@uB6jqA@S1jDp_4f4e~%RlV&0e!t#LA;K+|SiK%?YwfRT93BN=Fr>)Tpe_Mvh#BXM z>eA^1`16ycP1}-VG04|8=<$N?9ckW|w~N(c#!r47>!@lNymttkK;6^(R?y{e_(4bJ zdLO(O3^Q9Xx_jIc5|Put+F#&lvrcP&&w?8u!cUsTbQ7VGQJ6!TAmUhC3%%&m&Qxql z=v|n0^Oe@O!NSx13bbvS|7NTQ;bVH|R=qG5FlG$?T3uXRt8Ri(bqK@)YwentKQL?) zHlGLJ3QNy}*QItfHk8CY5>d*j{?KrJVq#K>i*7M=($I`Gt9>VpN+hE~`V|BNFKODQ zfTA{7Z$^RcGdhhTOhvo-iLMk;af8A`>e?0>Pve`Y4-(bhy#owR4i;cZBR}Gm@9f6{ z%SRxj0fSf|vVUO(3w%8R@yri&!*(k?P*w;`-BB0QI9k90N5TcA#Y(e}`nRVQiB(Ma z_hW^WzcMk?@Ar2z4_qI}uagtyxWK^%N9-P?5%ZMT4`Y~7ObXjOtD|nl7YS~wI9dqUMYO&- z6XLb(_7ZdZC_cyJr6Y@o>b*IQ~#quyQNeyLkL z&7scR@FA7i;hdkeYG~5h`qHME73bOU@hynJrlyHT#GL;N;K+wK6PF%z*{2 z*{rXHqwawcg;CUrBjo6euBCAm1f?8Yqh+WR<)dEc;id*(&#P$n)FrbNb5e2SwM5Z+ zn~fBqpxQ{vGHR3Qi7(Wjm7nlbsOTs}v&s{GiM&etsNdNsv|V4_VJ~5N?<*B19Of$C zG-3cB{RBM{wkp_$${D3t&2>^q8%T&DT&mn*P0pbbNR>*YuBq9uJ~BXN>dZaD8i{xpJkBR<|knT zW?!-y*ck0DKbs z0Y&j|d9E0UV4knxVAKME$UV0uLuv{u2mE+HRoi0sD$r51ZJ8r#wfJM z7wF!~^pPa2w}?UsBkhuUwCixf!Vg%E=Az76+#a!^4EfE6itS34hdCpgMFsd5pItAc z9nDt8C){iGbcHd-uYS%7oxH)X?I2@H=~no>ZAQj%%1=sz$-`6%LiR;+e0;U)atZfm zdwy4r1d9lwUIyv)jYHEK#-_YAui^@1V_u%oRE1iVsI#a_g%^3s#X!K;3bIxLB40E7 z1Q=2O{Tc88r8riMzLz87| zlkicxO%*JDIHT*l9+$_5u>R&)5!ssjAuTzG3R(!Hg97cCny3?!g7RX~*pmDgwZEIaEm+5E|)maA6hLszE;4dXs4FUL-Qy6@NY$;*tzV@*f;%sfBz1GV92 z#nPf(i3FW$ZT4jD7j)-N#|^PQsvMrwA2)8xb0$>{U0r3iB&WQJrnmvH0QrlFGFxeZ z+0{oM9gYrzLC@xwOqyQfZ(3V|oX}USqN)1MF1stx=!RQ8yfx|Z+EDpS+atQ`6Kf4s-CtEPY_9w)L@(BtJ;{kC_(o6hL&XY3lIlOvvVaJ26S)^k?W!9HrGQ`57EId}j zd1rW1u%(Xc>5gptYdyVsKO$QLm+(>+90Z{i-pzlKN&Q*oMR`jPT!SxW9gcQ}p!eDAW_1AO4w@ldLWr$-DEd7pjllQDgU0~$`l>ZLzZ@07!^}? z(cGFZ-PP(y%;a4yos$W}o=674BsW@KB z%Ik>kV(N9o6auvy;eJXowl7d&d0*)E_k~uHPO^GlJ@}M>FfR)4JO3ud{cN<)x_3(D zyT6I!5X&Ev8b>TZ23<(9O5N)Or4}>Yd%p_l1T=Rjtc&ARUKN(uJtzj4xa-vWb2xH4 zFfKTcLcd~xN>n)}*nt!a%z)Y9{P5_CgWJ&!eqUc}--ph1-q**NoZATJ(JOe0e=(;& zt1hPQXoy7}ftxzMfo$=u*&K9S0|ZlBjR_k5bJ_NLF80g&-%*~z6L&SnsX*^+h8x$G zI2ay^R*T_m2h#3pphb-mPsr&Y2aDg-4BI!LwBbLGK^Fl&8w$Q>umJfeSc8x!Xg@iG zL1&ZJn^x|AOILX1AW}L(AnRn6)Hw0FjEPKWSjI3Ef5ZflMIG-{f+3-F}yB4T&O zpv_F1joE*F0>fv8w@k{rl8Ul;D|kdN0Nv^WN$Pj^2M!UTN8G>NrUKp~J_hf54^G%4 z9Y1MNIj|T42Ynpu)e7Mo+b_I=V{-$Y+4o1Y&G-aKK5_W9;14_ecJn@>o*~_S=ytc8 z$*zzgd^$e!yIbRYt^6zFVZUcGDrsoh)E2~TM&IHr=9mwU-PK4s7g9dS*lJ)-H^^h) zMNL~RaO9m8tq<3PwC-1ITJ*zAwj)*{Ur)dq$s;B#5SsbBhB+g7Vev02Ksu@qHZTEE zaHX|K1Esh}b%+52TN4EqXw}XMDk1NyNGzpM(Y|(;ngsE_?4TjvG9-I$0RP3HmICck z!he;C|Fh4Xl6uk4nm|7uO-Tn6N?y>vKwc3WKWlQd`wO7_Ui;MTg3xaio=3&!R<>o_ zjU(IfBzWr;V4^$PVZsDm$h9%q%-Kn!R9ctRJt3sR;8vsh_gQ!`0Yq2;#{g|B5DE+( zQ^sR~t6=i#SA-Gs4OYbhY45SXJCGGIe$4O_mu<;{_uU1T6)!vja&Mq%MF_r4+tl`E zC5@`0kv_kqF{zSPM5Y%u`16coGHZ zfKs|U<`jeo-dB(TpTRdlt3UYsWgljg0khtZh8b|Vwcy`7?d0^3h}Pn-i&Q^~8m>Pm zyDNClP+*S5xuyt`G<9$_RsMrBxn_U^eioD>5p{L{%XPB9HsO!e(okSgJoWH`F)bD- zF~93S(dE)}|q1Q*t{xZ*6TPDewUCZ(R`NlZs%g^V|vAKmY)6J(!9PAX6J8{4+2;EuN2xDl-cmW z4}KilKdI#Xp;Goso6O_u>@x<3{fCS0PY)=@y}+bwOZ)FBJ1uR)aKnVLT?JB|Re72z zh=D~>0Z>y=V)-EQkyERisq-}MbshX{ulsI!f2bS&`LF%4zW8VD@Scq{Exh))g=qi@ z=QD9I!!Lq&DsShUc!P>N&yU%{ANau)A61lN2`aNaDKaxVtG1et<76TErrdIMgIMxs z>&s=cySgA?H76k5(O=6!oNPz)fI8cCnl`*VX7yC?f}k%r(7}79 zK%E}m9EH-J62Jl}Ztw#Xcm;FpI%|(_lypV|6^5odtZDgG>InuqtMvUPkt%Lfma^^r zslO)+hD|MRFH~&dSi89cB3}iK6-orM==Ou z0a~Sfg_46!EU+zNg1K0tWqi3Z)y>*8>ibWl*27G;CP4_k--NznoDpR!yUwSD>8Qk< zOk)Vuz;L8>5@UOE+=M1i1fyBOkk*FfWMC+C>}=EYTM)@R;i&w$>m8-80-Mxx_q8|m zW|ZHNr)bdHAboC9PAuf}a>d04#VPbYt$Lc!$g(Z9gcT`BvP80{$ zaI$9j3GW&-Sj^8WhZ&VsSJek!H=nj;^T@u%!Rc7TD??Wc5pF7w4&-tkCbjI!bhcUNVYR`+*(yy`+%QdGCbT3mhXG0-&E^Q zyw~ks?W|4_UT%PB_^`%szgp*x<%5T{oH%=2^=%L$RHk}OHS#6&i|mCIJ4A%v3mpmm zalB6Ln|+$V)n0BHE4-m;*;|&yi0!5z_e568s?w&K_++_Jd-EXE_ArBVj*FMy)BwH; z-YNRCo!1IK9^4e6%3;=}n$;lem?qZJiVVyRrNB$CgJNed3VQ(5G_lH3|6s3zF3cYH=QBKNTV-=aWbazY*v7JqS8j1dSQEXt$cc!kaPrg^Gk?}0BJ!G<@k>E6G6x+{v2{!e11ZmYRRa<8?Q$sQ zv^w{3Vk!y(#R7CnX!azH13~DvEF2?Hm0JQ$G3pgXXw-&*n~<{OPCoXcsX(184CGVa zU1V@ut5;xlgL`5gtu;F-+ujW8o-AvK^x(6My-fFYQQypzYI={>>ti#*FyU;^+;;i( z*Ku$tH{MBIYMaRpherTd^o1T>C{P34G}^A@#8G46V(p?`d#`}er!AC%Sn8%Re8vY0 zpp78gq1MoedeG$$u>h(Lk&%;P)C&GExZmc;s5BK9X z9c9n-@i|l|C4DI|k2W?^WGGqg8G1FWG`$W$0`1|=7eKwx=_?k%TgC!&tJVI{1>-nO z(-wR;TXp42ms8{Xv&Bz)!thy|0eCGqUys42NS7ZjTeAr@+}RYgmR%D3ek{wxqNwyV z=BQ4!TxhV`*kn_FHr=}ufGZ4v55*Z3IUhmL;TRB2OLC#RDUvo$qgr`gz1j|cZG58i z+pwRWo96yPwt;$2^0QG+rQO*#x0n+}l(Q1_{H|DBIVZe4_10^l2&}-TkAVvE#d$C# ze{W>xHsJ1GyP|3L{t^hERyIya)97b#Qurtcf|vl>ZzKi2qlM-MF%N{fUHTv6ORW|b zfQ({+y`AX}-8+W4SRiT*OsEbW7I9errEY>#lTOlxt& z!a$||1;n1F`k1@sAj{F{fo;D6O`8?U86)=)j5~suaGT-65=wXnC73vUtChhjmp(do zB&>rsy!Hp_tp$1G>w9zC1$9m?wdlhBDiS98d8j6@Q!N$%`C$92s1x6)le2pwV~+#& z!+MX1jp5VtN(e#d3L7}b@CcF6h2=y?s$+%a6MRAF7BX{#E_L@5=z~pdFBF(}Ht{h1 zV7seiyMs6n9Nep@*p6v51KDUn zV{xc@&8kpFD6i&E34#ouAb7d2Y3(yy)X2ywl^mS3WGm?2D)hy-xb$5V(h0O&YudpD zC##qroqJaw{Dl@Vi7h&hSM}nS7Iy3EdwY9bFvR$wztHAXQGg$?TWkyuy`|q?2wRoB zuEUu1taMzSX^JRP|F!m2{>p1Yn~D|z;{ zI6q8AaerDf;Zh9q>GFuzIQ=&%c^e8fC{VgLwhJ%ESm(Yonh^gG=)Z6lbGm}+nHx*F zf$JIT&I;m-#ULcHI>SeYzUAoo!*}}O^F0nnM<6?%m?p+9pPLhty55#GO@w z38xi7>5HQDTWp}zTXuJcD#=W!d~=k`-X1di z>k$nfRm=0>kRP@vOY!E(Ocv1cA|lb~i9 zC5@Va&2x%%_dgdiw7CgMhg)uo7i4?n=pZ$M9b6$)u(9hQY>@-rW6wjboh;qgir>dy zIB0e72$QQoUW(vK^!mz4Em9^F&9r#8C6}W*OFqj3)~+}t&RcN7{F=XJ*Sv{h?hc2TC8_5wYaU@iPJHZxMs1^fq|os}j$o2d5n0A%aFwYK^>GqQZD5gkkD{*)daB}1DC@r{ zeF*GlYalP_4wl#!VY>56ii6Y1D8bxFyTyZvyBowwSHU<38Sc5@J_an%D|9d|ca z)^5UqP9beyatIEyKQTwuMF$cTr*#f2O1~;-Pe7a~WlAlrT#5~_=ZvJEi*nDM=BgI; zVVO_RAX&-Lv~7Ux@9e1Qb5(0`C%C_45PGB){)pD31tam6Rb;qc5^s81Ex(IhWqD^{ zYErHfKOhCADdVN8$k@NfqJec-b5Zg2os`8K(t zPmhdeQ9;uwh1n4Fz5I`48jF+T`PYL9Cexns^xA8{5+Fec!^_dt<9l?a{%adC*4-Bm zNd%l<+D(SPoEIxfA45eRw=o)R!`W?fMf$54l>OO>Wz#|h&(eiq>btaklxqe`9QdW> zGbm)$Y)`w_NtN-$@zHAfWX9eU@$!`@j3PsA?&HiQPnxbBt8c&2ov`!Au;0~sTgaBSOY^4HD@&$IbD0lUe(9@+x z-iB+rdd=bY06buigFA^SRBnR0GLi{~@;l;=oVTPCBv^_ksT_O_=VO>eO{Sch&y%^3 zddkNY{s3BDaXjuTyZWj5V7(l7{8jK~W(Pqj$@#KD*2*zkS%Ut(4b{yfupsU!P>5;k zVsOh3nqctgxa+;lI@B%y4knP7p_Dzv+qKF;xEBkpkZl&O-FNQ8YgeU7&m`cnTDR8g zibL%;>%F=ku^TdGbWR=sIuon+P`K-V-hM*VeBMFXo36;OYqh#f^1VXyWyB zv=|7uHV=NGC=MiG+D5i@LU<9}RyYb>$n?;_)J(%ygn!(|=ah+omVlL+#ir=u{O{R{ z&XBBVCIvF~4vVFFm|nkw%a`CzXEH_KH&>vqj5FR+U0wCq51#F%wH=`wFf6bh?K4ow zY96)v(s*7X=5gC2dqSlK7bs;#k(tV{&sEyJSCy4{H70!WXEMr20pZQn=k!OTX=6TN7zKvVEMTgqdckvihtW%R{>}46++{DZEv5O#blj!6~W8E8CZKS9ecI8bT z^!0vl0IG(THF^v{jEt-~=Og+YBM(8v`o3GL36OYG<-m^XUrTG+yQ{|sDsEnUXNS0v zcXxF@^b;4UE1>&cpD&z;rP~RVbIT=N>;3!q%{kUy&^beaJ(kOq))Q#{k4RnV|-?r_8$^_?V3&=uO znnpn7j~8kLCKU=qNC z=}N}3*P4d%6lYxfivkDk&wlp<2^r9_ugz2_{<3Dx-`ZN$nfZdN z`XUPpy%5Jkj@(TOuHwQAyxF0lT{z!#B{`Q()mDy#yTmP7{o%((nT0PO4K>c+-L+lISGB9=>}^CjsPc=f+Kr z=2AQE81&*+rJFgLJI_TpZftT`W!)I3=xV{gFTY}=jAjti;+e^9iGZ{Ff#3|z5QtD=xFzhCW!e@gO{u#q{fW^!1az{?4i{>~Fn zj9cykR~Zf`79f@F19wy{f_L_dg$KGm(`nL+sUFF8f?L54cfe)&P6gT=RRKjGn!sl` zQ)i9LX?#tu?&ZBqML;k<;HEBWjqS`0Dl8zYJB=dYM(OG+K=h)0Z7eK<=VN8g(k5Jv zI8YupQjI1nV<7rG@`^@>>ej2bZv(*`!vDwhRr~uavhNPdEw!*a)sYv>69fgWvR7QU zyuwk#kuLVSxR8iJ!3DY0kexO`}P*Y9(rWYj@W*TGbL zg~Ru-9SeNW0{tq!=>*@Q=?W-$gvx~bE$0>oe}5efy!pHXn1M2F^*}+eF*_$@_j^3m zpO-b8zH)imPrmb654jrupn0Q`%_cK+MZ#X*m4M&%L*&Y88qmmlj zwARV{*IQ728jgM(*8j%&-&`*QbiL(ou6O>MP|*Y`Z0A9tVykHv|9|*-zMsh2XI(TX zJO<_Z`F4|`r>u(XtWSlC3cO%5CCWQPX|c49=3i)g`uW8KQ{2a~8dnV` zQzLiz)mp=5HzYTu##75Hw)o!`wA15LWIbN%y=ugU^UNQI$e+sV>p^LGDXKp!>(%SC zs|e}EC&Ewg55qcJ0WS)g@bsYq@;rL>dPVPo0uy>@OY6vj5vljEvuDGnGC~4O`;r@H zB&BlzxNGBR!G{TncM}=klVy!BXy<-URB^lGU81J)RO1-Fx2`^oHMt6nnGJiO)(U|~ z(Bs#W?`MVQCeMl-RmJ5^(S^VlS3h^TfAGC2!McmHihw_zYeMRjr1QUhon&)u-9X+wr|tJW8KAjtu@{9REjE+G6cOQA0iav-20E z`rGBpef6QbEL9@W&am5>L=d}lkCe4mN#lYd6Kg8UlsJOOuBqD=B0S6qZp?Re@BQ$& zfeuAtxR^IsPnq_xx$xY^O_f!7@*({x%G(ERind<$#>;YbL-ls^G6qV`wsq=@m95F} zUX4i~B*cQ**HMnOK(@i$VNOb&QG19`c*t|c# z%eJT-6XVib%h_X|;;nH13*3>DZUXZS<2D$EqKn8a7L?iTeivT>=Q&>u?Z8U4tF9Es z9R5>xTa*F`BsOA&d<#BX{aFq0yU+Yb_D2qSj`=@0ds06)|2+12p5fMtq7n^`%r9ct z9d)JKZwBlJypNywbH|wKJeYaQ%!5`33^M*!ZTuO!{VTqwRh3tkQ->K?-KFj@kLwB- zIFQks&%46Bm$#BedXVIskQS4*k=4-EHg{b=lajVwk%9U2nu>GIn6bKp`0%UxopD(d zt6qE(*Pgagmv5_nh948BHJ=^x)8a_dH>Dlt_Ss7}>QB!aT(+#<-b_3r8hqYP(~0ux zmo_T#1_i)FH)Q++bUPk3e|%BIr2{dbexrqJJot&?Wlo=ms+*igr2AA9Eay=6?<+V% z#Hs>elcw}I8hd+NFNa1Dmr^Zrr&muI1lBc;?I-fJe?nX;!JlzlA-Js2S20bcFP;)_ zZ?}=8nAk-v%3~pMh5Tm!x{yrnU+y#dbzJ_PIQAxM`{eZ$Pc;%xMd+>n*v1_^D!u|C z&Y1A!DA+JKFzgKMF1l79aq)^p8s{Y#ighuxOHX|+Tn?U?=%g7)4iP|H-476!`075- z^mQ5n(j!fLozlw6LQS!qp8m93d^J-jmYV>>5srQ%%l1Z2kjog~gM}pJ6i;q03;a#D z2_?WW4z{aFBtF}OXN931-m_Cu``TZ5nTn6^T{{DQYBWI1z?~;P@DtMKHyM;RAk)L) znEL~4zi7Yd@-a;L^-g_5qt1u%@P*A@WB=;hG|G#;v%|%&kn9>~$EL(o_-?+@J2Vk| zj+&S)Z+j>rJlbIAQE;^e$rW>w$eJ;t^Hnk|BWBan6vOkrV?5=>a^!He0^qcb5@C^Q z)j^CzI8~h{bF_1}DBX;lrMX9?Rhc+S6v-?90mD*-ycr-H@5&iNLR#;vbdcdHp~E^8 z`Z>c-dRc3mDpVY`es2VALer-=(dx)}HRmaMAe{2P)Zvwm`S6!tVRi1I_I9q*G+z+o zZFD!ZJndKCy2IYY_2%K9PFlBpf*gHZ;-j9gH)%ck(12I*!5&1YR0NTrbFqeKsYAwb zsE=?fK(JEgR&xR@vZTzuc&NMVjMa6sXC}oHI(wI{4DiHW8M=6mG+z?9F!YZHq<;L1 zBd<4*s8?#MRNqjc{q6DwGqZ-1Q^yvz!WDw059Vvjz{5! zTTd#OiU`Ex>tg9H*S#K%IlTUoW8<9!UJ6z2K^j>KCu8qEZOaPvwYq0!J}M&i`ohCh z*WJ!%4 z*;>NA6y5H$9I5D~o9^+kCAd|VB~Bx9etndeUBgaRgd?7Ek<-iaaFnCi3UbQljzbQ@ zhT~ef|G*AY$fI7daC=oDX>(9feCgaHsY#UGZ`czzmt8ZAR>-l@Y`5czo^9c!=6sy! z)C84J%`(`IEO(uQP{6<^5DFgHSF13qs(kR2BI$J!f-?f+AXNTKF!}Gm!T;meP)Q)c zJssjj8ZiH@s}ZWJXWP|>#y{O2!?`1^fpf zM*(R78xs=?69XFy3kwGa8yBDS96lZ%J|!^;At@~tJsm9-4GjYeH!A}ZCo>HVo6u#> zE4%^%0`#mR62g4q-24K3-zGu9!NI}D!>2fRj)ISohLP`2|B&wiB5YJ9RAV$0CIFQP z1&s&=*#OXkcw(UZcX!Ow7#MI2(&fdY%$=SvA=`$Z+zvnOf!@^%hL`J=ij!8;R zNqv`=o{?EtR9sS8R$ftA|FNO5skx=Kt+%g#U~uU3@W|Bk%`Gv)$&8_X7-M#&T z!=rC>p#W%qB@6ufSHgaviwL9(6&)Q79qSuiD5y?gK_f!PVC2OlmQukoupwdMdxA}R z?PX%Y2OMU8)eW)-wmrBPSOlgnZhj-}FO>Z;!o2?>%Dxl!D_tW1J{k)6^3aF?D6oIT zmgI%|50~tCLTCT>hC%-E&p6i~k10HwF!vGb{xTEZ2w14dkOKc{Mcq<%PeSNsW$hb? zdNB`&VUHB+p?XYCx|U3aS(ygENn6E>BzM;<3`jr}*tRAdbzW|)WbtIvCqn|pt@ik( zuD06D^#!x51WK%s8d>wn4;r!6nukns7E?elZ$K(CmN!2X51Z&mlPSvzvHkm0ZQSp*krt4Y{|q zUSq2C&xr4&L#UQqhrA(NtUhVgPNi<|isEP+nwn#C6Lv&CbR{MbW|$`V18hmv$@zq@OV1d>(X>EogDtmug$JT$kVGje;}u}A=tbr`>ywWAr*#XT3v279nLxxG{u zjRdHm1@`NQyZRkSAV@zc*u@Et^i&W1=SnB|QQjP^DX&LPyvoP_*N%|y05lhp0f(#P6EZ6 z#P1!Zh{nFU`2iR|LeSG+di&}^+;I8ET2Yh8O6-zf-(P_hadXPN46 zp%)_oyQ3jvf?H*dH-kVmQ1#*;*9Ct-k8Ht4$Po(Fu-y~#E%ROf`Ra#XR-w&QOpDrY zhg2jlYL)5mWf`Gjai$o?-vGWs&|s~cAclD_>UP+-+~N;K29Q90J?zF2bNxNCf(_#* zbU|r~HeAJ4pQxrH;yGi9i+^htO(fnd*{ zEdjNhCoq2NW(ob#A{HV@d)>TdHRxTk`C!>#zMf72MpE3Bg~zshgtVrpjV2cQn=|Am z9y&;1p+AQzEn`&C$>i&MW%`mQ7c`h-)?J0AnK3+iR`!`!sRX!m>fa^)22ry_deL`{d|rd za0>5*DjC($Y*-fu{qYeJfZw8a2s!*}hy-BzNZ>{C(tsOAkE4HT88~U226TTH2^_>+ zJq1y2Aaf5dX_ZXT3dBr28@9fT1U3auVI7jCzCtcD!X9}BwbKykHL^Z)2NbJwou;qn z3`9tdX1*S|X{OE@B+%XK z#Xh!lp{y*%-m&*EnA_<*>4tdOe5&>k|4D#@*KHDEG}atb6-i1YFdAbGKjAsqg4O3X zvfm~@9X75< znd=xo!Q1i~ZtLNrHH1UOHAwmDRg>D6-CZ)SlQ-kpmqa@Fq}g8Tm|8uS#XcfYw-Oj# zPS>U^VK97Tg=dAr45>)KffWcKfy3#^Z_1-7w~_Smwa|t{=Adht!CnMFTF<}VD)}!~ zkh<7$lNZZ{C6=l;Oe_*uSwhFma4c7#E9TRfMS5gtLzB|PJ=qpq42G$ixA7e&5*YV9 z$Zyz1_F3ZWZjnRVQVWZ8*l;oVa;%xDMm}8jh<1DXTO<^2HYmz?wn= zZj3fpPUousxq@b8-n?}$b&7vt z=%Rb!h@(hSstBIIqt4NeIc~6Y5~KUs(N+^H2rtJC5Tu&#MH1^OICF3 z&&O8jKgKOcQl#j!R8Nr-tQUMAm_H_6I&FglP;*3q#FWL&#XUD-C5;jM(3TGye|Bk1s!+4?8QT@*KK+) z@MtXdV?$)<=O>c#i<}k5#zObQ4Lb~aE0T)9n(vZmyKR%%j?)bl4G|CX4n47ichQxR>)5CzktWl z)D}37!t8Bs&W6Tx4wYg>L+EA93I^1iMl_#-+5=b&e*MWVT;tI?i`lK%FKSg%agdd) zldp*6ve{cpIW8u&D^hn0MQMB5;$5%q($B}eLNPAX(z$NCfM=F%yC8CPz8_YwB<3M{ zx2XDM>Q$#Fs;ijJ$Ax+OQAfd(FNM-k$RhgrLfd4eu)Hr@u--bENVc&yb~0Nau-i8b z7tf@d$(56I-lOXOiNO7q$7LJGy%`usU){Hp!+*=;j!wU#M_i8)QM&rlS~I9$7LBV) zOT#GU%l%oSORpCS-(hxTlTgaf$ZD#V$z6_$ySi~Ze)=4`S(!p6j$tJA69K;5F&Y|F zY?j9!7Q?1Ml0Vx&T2Ldt`o%I-l!docOZkN2`_hjajKR1+X9n)aL?4ru_G&lPZEhV- zHSZf*ec6}rqgA3NRYp^}AfP6L>whH^6QFyw^pm%FO4MU-3ID%fLes+@r{oN zm}TAi66hDuk2unAA6tS}mnPg4b%0Fs`1Y?iAsnMxlSZ0XT9ClF0*U~N#DCFWzv0=r z#0PO})=!cMKrc)-Gxl_*M3R#vUAWM3Tm3abvvA(cl!uzC*@>?L&&x_B5Ylc!gY2zI z4z@pNRB}#Q`cK{=Kp*NvR z_EV5RkEQyp#w8L9|_FiC$5fu>g zws=V>5=5e=$G%3V1yzSS6eR-*IAHMLHc|`r=OvV=XK*K7+e)vQ+iR4$qXqIvwjEKM2%5r!nZ)kxckpkqn+J~g5ASSA-1cY7-JbRSMeSAO z@F1jmJfR&!in@bQNPuCXj(*M}o@gUEifkikoK!{E5Bv8#7c=Mk-SrjB7X+Ad4)qtPKVFk6kwAz@-SQh4d*rD)u=O0)FNat_s2v$Wmza@&J&fJM zCC_d(aU&szhX5qAGYtvUup$D?JAY_b>QjjnOs;M3NYMS2;Oe^RD8HPYmagEPoNHXn zI{84f&a*20!?~FX39LLNnV3pm6MlzZmdV{)7o}+K6VTc!ERMSx=>>Q^zY@rE zU&!vI?!5obaeMp=I^0U%UKh)pxY9AY{ou;WjE@_*RL?hbm`(cP}wME2#eYUY2E0Q*`N3t!80A}R!O0W(&LKQ=;LyMO0VUy&+ly5nBho(@eL9B>AsHy5_skY2H+Bm;`jG>5bQN{{u0m7 zjx~wKJ(}Ob>LqeOg~+FLd`yl6$mHP6BaVu%oiVKtEC*yu2npa|X2^ z@CJDhVgMip8l3cP=Fifr@&h+qB}m<{vG#JAE005je?aXA%(HqHl0QpAOKfO|#$1}u zqQ;fxx9d|^;?y@z#2PdlSnw8jAO6uKA@5pJOfu=(&29GWfB^HZ7Vwq!fB7Dg{na|t z z3)kK4rq7`;S8b9%^K5+=r53p~a>BY*_!vhyVLH?Owp$D6_+!?_*yokwSkbJBH;=di;Kh{pV&gp}Kg?k0ld0jgtJJ*t;skNwwXz;q`HQf9b z=JGVnSM5_>2lAe*7OpalyJOndFF1Bytmb>QEw9VQ;<$8Mbn|Y(n3ly+X|_Wiq4OKf zvZ4wLF}C83CTjbTue2<>6-x?T=DKkY)Zkrun*Q#Hgh8dbqs3{S1TK0|G0A>fl- zU)JU=4`T^6!AGIo&%+<&n z8OFF=in{%@BW6a$ogh2yF&d?~;i6hKOa3B^I3>Tgc>K{C32n)fjU;WkM3&%kD|PownfA(teaaK3tH-sqdrs z(SF2yR%{!7u;{U-be6N##}Bn^8#Qy!hm;jv((!Z(_jwq~W$%QIzL>8ez@4!l*q4C# zZwl3xZbE8ubS)<6Hm?@Y*368I$fad$aSaHsEN{o9LHd!vsy>nQl)U|ASV{3t^&oN6 z$kL&sE}q6b`Yik-#V504-eVsAK6jmIk|`KrYbKt7t*U%qJ5Q0=Yi_%z%R2YJO6`)( z(x0!!N_BRthdt_r4k zkkwQh@^EI8skw#s#*TJ7SxaAP|5#_Z>6ww|kp&x_^54Xkx`{( zdoRa_$7F1*8QK&{d^PAAY?PuyVszdX9gwWjo$D76tGPEly{v49Ye?Tb9xr`N!g&(YM?CkcPdHKw>x%kw5Iu$t=Q_mv4@1ix zTD}DOdo(s&joi}JZ43QC73soqKQqCBCC)q87~(?H^`iQ+N2Z0VXGQW-mO+LVa|3?9 zOF^ZOvaBqmZCDj3~xzH)xuA$tJZJ_n086h5_IE*Gj-#L z$^+`_F-B$R+l5lE>on4LpakZj?sv*dW7_O{ICInbbLLE2f@UMsEYAfJ^YuH=e2ht* z53x)vEQ-_P*E$wmB2F<%@l>4WlziXz#ZJe}&HYm>PkSG1c~*PlWG^qp%vJxHyU0Vc ztEtvTQ9^waRh-I{kVoh1cpoTag*%sS?kEwP#WD2eU}n4Rb6O}}Fd+1pAie=7k(Y+bX`-o;gn zqP9Yhq{Zj0sC-K8p}bAEE-0yJ^V~^_Z!8PFd*G(fEzoWzs3+TLBA;TLuXCN$s8po+ zKE}nU0C5G&89kwo+G}s_hd9uLon-dPnU1U5UiFD2Qf zl`CjB1t)rj0{x1#x4bx!O3;IS-Gj2J_80bVIz@Z9)-*_KB*R5Uw|Q^ICMQttry6mq zA4a&7yG(yH%lR<1d`_T|Aj@_glF0mo`NHvYl0I66Wbb(@RRirCahZLzSu+l~{x2a>Q~i4qwI0)y0TP+-a<3?0eW_uU}xMkZaB)J1MZn zm8`z_Nr;!3Mq|j&frqFQ=9G5erIuYQlC>Z-B#eb*;dp>nBq6D$Mw2OilE?^XfI0(<^4r0My2#j1e&)qT+8u@tF*vk3RdXF*k9(ey^RDopN4&I2G3&E z6%}9kk+l~rzEY7fdhpHFqEdqUg4qbmfzVGkMzZxzuV+mUo+osv^iv6-cuPzE;^LgW zCeszAwddSQPHl_5MiFVo@S!bH7)!VjxR;(=H}lE*Xl}AC_1CJGcoO|C5fZ3`MnIx} zS>fe8^v=`!A4hyjf@2u@UU!63k~pF)h#pgI_v@w_c|2Rb7IaZ$HF4pGg|D0DQrWgWE9lSX0%%~CoF?C#XJsSGFo)fG|xBRlBN z#&852WTj6m>JzS8?Hp%V=BmnHYbPPQkMB>&JW_Sz3xt0S3HTVKuQaXm(U`n*Z%iPF zQrUz=B(r(<@{Q(a}$4ajvEF*_T49})c)1Z);$REp()Svxv-GrKP!CgREPtg%k86L_PDmb z&XSO}W1?BPw2hV2mt74F^eadGqR?lE4>^#M^ra}#+;=@Lmj#xd7hJNv#^ir+Bczw2JuEI$kh=o%nwH3l9mo{7LzExKdxvLS(UU~Z5$)o&&Uxu_$7 z=k}9$@OzsVGBU@oEoXWXnjIY{>(keiQeAUg9FxBUC=tVT2fhv#N1OD0dP(zGSNst( zF|pLhI3D1cNJP*3S@b(!~gvi)-<23=ojS^kBRMF?wvYWYUW-pEZ%i;Ni}FIr7p*S7^|Y1_H5 zkMsogpNoa-$^^cKoxCvVNc&3p#XXG{8fzny{%&}uWJ)!A3Ok3mwu%ypTf%FZzC=ZR zBR3`5;Gqb<92ULG(*w;SXD5WLCG=zwx}&2RI_b=#+~aQcFyp{lY24_|&F7a?%8CeM z3vD@u+{KUp^VN97PW%y9W^Sy^#s0QFUxL)jd45|GOkm)tXukfGAfW!)db))uH+Z7L zcS}sI>{uqnM=N=vH7tdwE!6eNL|%_1rS_zsKs=F_j7jDQ&XliAq_})b%W}7`DmFC8 zoDDZo_A#c2NGh0Y^Q)Bp)Ei{{y;FWL_}gNP1Q6vP;j=lJjb{YVsrP_rNd-X#ifOUH zcOUe}R6+k`g!mj7ZL)-CYX4eKM{8pQBgOQ`$%l6qLy|A@)oYyTNu-DA#OB1cdi96q?WH-9X868bzw zgV$@u;Lt9ZI%aQavo6HduGu1Kx!aqi}fYQ8n??KMkU`V4$f5XAlwB9u+ z*~nt0!)gbfnPsgCdUhQ@&Gyp;ta&|X5M2NA!DYkr(J^r+ZGY+WVM2^g9o&h!HezP0 zIgCA-ouisW8w$b-KNl9=17j-!1g-aTHa% z5KubbqkqnlJ4YNP_08~H%$zfP-{ozQ`#giu@q@~?U~JWq2{vYmU;*spZj>jgnjB6} z!YON`AX`Jd|0cR(ws-t-kL9vE1nvZ%OEbVRmXkYFM6bsrMfAbOs{O<|2&V7@q`Ui!{l4i0{4*w%BbUoPY}yy;jUEIhh(~2Lmhc$ z9AzG$I!R|Zy&VdiJ|MbVdD?!+J<>;&xB>kr1x9mypZsintLVX+-%jp!)~74KCkb)9MOfykZQVYf#Co7 z9HLW)6REitieNO=UO!}8uG;U~kOO?|A%@v(Nkj%Ff8!M4cfhZC?R z=oDcutO|O91k4K_PUT#!-x5n#)xRpf%EfV>?~3yg$!$^^qu7oD%2d(fR1ETrJaDxt zO#j}&#&4RFKOTk{!|I{$0;eyl5#PQrEh(~hPTO9^K+ul=7)=Q=3#OZr@O+EF?_tKD zv?76xOZo`*;)PTrnyiB8lPBPbo-GWH1D-|tC!ZM)&hhUVqP!Hxy|MPfIWi`$%)spN z0n_C)^+$BGB=7eHL%ysN-Lvpy-qb7JC+ajaaH4VaA$v>|E+IE@)bkCyn*Rg>{>CYd zu3IJfW>gbbHNvTfQ%qPXn>m#-;7bFfcPM^8+@e<5zrtzai&r)3eDO7J(8bWpkCQni zB_4}_S)o?`Y46NOJd4csYWTLOt1KJEKkM>^i3a*t{Qpb*Vh}_>TJ-5YojxZO!L!JHxd5+R_ScXe=ja)zN7d$@(G>4Pj6(rnRe)(&XcTyzGLuD)66r_9B<;_Pfb zY&GV?!a@IUXW#%WK#KC(;q$HWnnLY~qEDYy?_Yj0{`?UQp{_=%5v+wrEoORt-h0uD z94lKcL3T%X2a*q3^oadJPJG)@21j^o+%E3D18dn3*@HGp%auf5vltf@z>GSJUK`KuP|T!;2fA2ch|$GNrQzuhB(Zx ztDpr&St&?(Z5=}UpJ-puoHUyxEjScXf#p1ZJJbDiT*>gnIirD>b5%I#dUm4LW-D4k z*R(m_C{OW9BZNU z=-`{Ig}ggE&?hVA0}<;ZPL1p?nx}!7cf3p~D7Iy@H}KV$5#sNSxn>M3DK}B~0#FPWxGY;@9q&b`OiW>cl!O6@Kp{*&ZBDL1_AU8*>k*^dZ1lJXYv34c8^2uCY zvqwHaeG;Z5dD((p>ENxd0GJlRaG*n62H&%Rqn*5*?B86LB(t{4<0(FlCjNXip7YUj zHF_b6X=Sm%9oonIT{i~%a((vRSlUCs7K1_gYcg7kzwNaA-b(pr?yJ=l9^Y<#0ApX` ze((x^rk`kB$>zkgu;%#oL8=n_63=b{$y?X+ZW$}E9G%RlN_VF!>W(nVa7N>2Q|~2Y zJqW@4hICRC?KRo{2|dW?X`)w^@szk?Nk?UQX@kW{K?X0Fq@Ky(CX0NzW4StA@`T|D zs_awTq?DiLv*X9!ji@TM$cGh?&xHm{hC{iLz#f+V%^7ixd-)FbQFd(63q%T)1GSjWvfG)L+O2Eh}g zzAQLYKxT3R8EtHeqsBDFmJ8>nX3iDY0?50xpW=l2XG14{XuS8@-e8kp*pj^u#VW=p zJ`d=P3^1>T;3b&Chz(uO1bUN-U{j2}&;>@UlU+GDi>TZQI{3j~ZN~7ER2eWBQZJnX z>%K;KycBXEZy$1mErkR;RE#kr$ljcu2Ks#pnb%Zm8rHwg^DSY)PaaHIct*cJaN)k0 zidNl5=py~j`KO4ZWCSrSGs62kY*J0@e`51||8 zIDMuG1%I*U{iQ8UTKS?VV?)$aUH%hfo3h0M0eI29JH8yW4VNSu9MB|AYO+f6Vh}4e%D>jU2s_7B8t@dQqg5l&@|+upLf8{QUoDfmO0 zeqjc8);{Wd1^C@#ZZ{^DKV&W*T;EvV9PI~#LuZ;psZSBqyG%Nu$*yQ+$A0ZB& zpE;(b^T1}>^!Lq8Hu5FF<&{VJ(@dn_@}>W1UrxIIsF|xrB+xsqyQ1r{s+{@%Ebxy-yFu9AI13thxdf~J+a(X*(QuCZ=S#wYbqpQ6Z&%! z=RBkg-7GWiesD8k@3a)fVScFuey~7Q@K=p(qknWgM5J?KOZ&P)>xGOZs1Hu{%bj_| z0~};qMC?Bbu6S*pBOHC$EDwSyTOIh-+Z_65G+N*p;uw7s<*%7q3`0OBGj z-Ipq3=i*a7e-5b1D`@}PDppXRJJq6cFG3gkqM9d{)qfSPIi;1|^i2*PR+dzAGM5L4 z(Bv^b0Y9Wppc#TQN6}FqG*7XSYt+{E#Zi}fiw(OE`BW#DV~R&WxltB7T~}RdLVU#^ zrn^h`RIB85rZ!`|vXj~4khA;hdkqCWB>IK;&U@4vrhnppA@~Dd9{;ez{9(oVZ+H%h G9R7bEtGNgO literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-19-Prefix matching and copying.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-19-Prefix matching and copying.jpg new file mode 100644 index 0000000000000000000000000000000000000000..9e4210f29174d612060f68d0bddb123247182ed3 GIT binary patch literal 15070 zcmeHu1yoeq_y46kC8R?Hl#p&1!XZSY1ZhM;x*NqoK~y9JbPxmv3F(eOM3InEy1P51 zbC~&Gzxm*;-+SNs#ajP2v(Bu0@7y`}?0xqB?9blk3}z5B2~b{B)ldbnumAuHd;^#f zKm{Pg$0xwYBP1XoAR;0pCZ!=GB_SbYpgu=I!@|hM%EHLZ%+4jq!_LWnfti_C@)Ez0 zu$Y(_8;`WSl!%<5sF=vlPOylGh)79D&y$gz7vW&$5c#KHm@fboArOu84hM@9z^1~& zp~Av60Bqnm@v#2%0{-&C!p6bH!zUmlA|?TED5nIlv2bv(adGhQaB;!A1HtbBTq-E{FDm|+ZSv&RJeE?!uZrmx&&74=Qu?k5z;6} zzbmdKx*)2zL3_ueikleDh!x>gVp+jbI!&c3PrCTKFbO zsSn2F5Jx2$k!$I9qfB$=`@btz<(iLc;pNXDhx*mKa9pMIAk|XLGvh(BReaAX4xZBP zz2{%c&mYwigKw~vBSgRJ42rO8M-%BuVmDa38-HfpWX*6*9~m_tZu9rrBMTI~GE~nK zA6oYc5Mp4bxNe&O>?H80el-h8%2{3QpktHro}0q+1uE<-rqZ~(7xB=&JN$#5bok zrDnP}g-hepOJla3uI7>oxn)YTDCiC(>eDxIeED47^Zw1~@L+Sad^@oHZSpQIwr~pz zj2KwYDp2>U+bp<|P<36+kUrmjHZbqd>}o0fwuCK{kvmd)X1p9FoA>!+#@iMFvb4Nz zte2lOa2((@o^NIGf2`d@5<>*QKwJ; zxZ|JJhv3)}<+vjqV|ZlkYRIE=IAqm;=b_%%lCJr3g7>ld@*cNxm~iO$#~Th&C~ihc z)J`=9s10nw090yc&K2>VhPEepzY&*xXU*kr*g;Yxx*#{hWgQ!q{u#eqQ(XXlG*OMT|bb8?*CM zd4s^x*vO$Ax`*e=OJ8VSt&9&vI15zSvI)6~XxD%Q{-dbYDHa1y=H`#Sk+RwBk%@gZzuhlgN4=vX zjCK_{ASp>i6D|5J>aBl^oY&G9W&;vStRa(HSL(4*DqM+|kCnFszR}9du1n`mY`n3X z*zVrO0C@XsC@93;qKKXCYeh{cMO;`4oaaXY$JZZ8v7XO#x)(Ck3bkHR)-PM=|Ygj$`xa{BC#@BfeK zht~7@fY^zJMp|k7usU`;wYr6b&QNVV;O2v94HC0U4Fh=pi*_j%mKA%_gUa4%sp-`~ zFQIQ@ZLrYjDzdp)Vt^LaPS`2~3?z(D)sw%O4&v@GCASTbtsYuYde@{dWnI)jVRRf0Q!^Q zdcUh&n_xY%F^FG{B4rF1r<&<`I-{a>I(AQdw*wA_+&gzV1P zVt{v}hHF;Fu;$G?3@{LZKtFX?WGeeDrTujlj;tq^xiEki2Dl6ykb~i!0(qAB4M>AC zbCF320~FbSQxjfZ5E&%$K+|6y9@gQ2HrwjJD^U zyD;_*u8>KJcjX0amdq6cSjj^$fF?)&Fi;07U-l{nK!8zoL&Oy=7|Z9KI@Z2y3u@wG6(*Dee|=<^%{D1c}+X+425_Y8QLouhK! zDWNep(EdKx6RwKN3d>TB#HqX93B9Uy1V zP3gW>Y=h%_;XTy57wgH{;n=R2$RnSYwRi^#tg|O{5*cgNS2d?>r3=eLz004>Kn~A| zyB_U30K>0QmF7(w+2Y?8hgm9+X~PaxX`^PNl*TWfSvqRTyYtmM%IAp3WI zXzJIv{>@OvC({Pk1Io$Y4ZUPd!sQ`mZB3rfpubD2Z21hymzFUaU?R3cQ_@ECD&mnj zt$)*|YsaJ=PTKEH?B)HhN0EsG@O{Py`19lHN`7O*Vj5VoBP5BWwv)C_S@s3&nx~IP($x+9&7xlqn8S zy|9%v43KekPey2E+}d);2QdmeKAM%53eD7F<>*`1Cb;!wy>t@=qV9#+Bm|&Tr9N!dg3haanIIX7Lqty0j zldorXigOWt?7y}IMTY&D4^&=DEjN|Qn@ZsFCw|#6mp|7-)Rd8$~U&RZ= znh$ccWhV~9sbX0d-JqWEBAP4q|4IQkz4BHaM z03)dwVA0vMM{1cXM? z3`{78_r*D8#%4xCLrTm@-^kX-hL^)=HHEoIk0=^T#G~je?1c#6?PE5Pqa9)be4_5j z^mYnKH}L3b+aJqwxAOK&P4R|d=j0XvujzaQpzDK^>j z_a~&9q!ou64ONud+M9WymYbdn3SK~#lvTv1#2l?Thv)7`Un`}S->5`6lp|12*B~pe zYK{f8rXdH%osd=Tnxo4m?f0VvzW70)O>JcLNFT{qYlP(|H5A7%59+uB13Zd_ZF+x% zYH-t^x&+pLSyEvgAQe7fki-D=re!+#|A9LQ__FNX{M?3?WgJJZDh#>0iV4UzmvEKj znV(Nqa)S|%_Wu~s?DS?)x+!Wh{7p)KGzZ?abW;u_(d>?}W1=OnfM84pX@qQXIbwi4 zGLTPqIV;kxp?GZMt2DDUJbis8#bVyZpW`~hyAABnUO6R`cRBshD=;4EFXE8=7A)rv z&!IF6Mi-CJPnhjJ#XmJAbBB(zuO1s*+gyuPq{9HGKs@bqSMxPfa>>IBMsOoRY=OQ{ z<@uk!Ly75?H8_-<+>>{E4jj#6&B`kMoxh7H2|s&TK&=#{A~;Dd-E~HVDUGV&``ZO)?fn7ST)pD z<0{e+dpSx^E9fRmnysj10mag1I;-GfQ2aBCg>iPJKj zv2UaOhoZgwPkPT8ndX1})azFLXzblpzx}6o=r0_9k{QrVx2QrJSvGSKtL(A*=Irt9`@3s4z+$;X*X!9Dq&1+zVcp${~#MzMkDP-n{ zCCGB#!WOf#u$d~!k<*S5GSl{kt?k2?On4JB6T0b{qucS@lXoRg57Y<)H2=tl4ZKF6 zpUAs({@Y$?AdWa^F=XK&JUIAOF2JVfLzRg7&DZQn*-bpe0evu%vGHXK{(<%Pao(wi zBu&*HeCaH&QA8P*j3oC8CYufv@>6Wr@e3+KRn7o@UF4hhw#v0 zMV>^jg0P|}R8%fV$U&S>g92BcaV483iCz!&edK!28?N4dp7Y%uthL0uVhE6dB(1Fz z?H>OmW{fqc@Xuh7Eda-}2>%CI-*RJ3u0pg8Nj+i0{QR`69o1WhIL{!BRiFjMhj3+X zv+?zI&NhZJnQ%4QhAFoiU!dkSi(COy4~f&9?TowE#x@~e*x%gT%6cp5cMKO@fw<>`x;R6hSM z^DDDtzyoB_vkd+i2;mK@DxH`iXIKfcr9AXz8BHtIBvJTt$`wQK6s0U9wYK#M~r>M?e~-Xyw@xA ze@P!8C2J56miKpS8sqdoBzI_5eW(c064NbBx1Mpwkt*={(Xis_?Evp&5J)i9voXF# z*`4^PNe7>ZARihrQrw9FY=&X(7~sbWEJDbxpM>uk2d5gAadl%0jugv%YqWP;6upD? z`+6TI=NZ$iOreJxF^u3xlS-I#(~BQoYuv5(-{f<}=k?FK6&$jIXpbi~^WjUIEJ+fJ z_TR`>*GZiH2`F?;K3cqe+^=d&s?cSPpA+8RZ8YVvMeFfBr{Pg#kW}hKV}xhk#WJ6_ zYYB_Pa|Yjn6I~&KV;}ZYYuh7V{GC79$C_q@P@4~zMYj;gO{+>bNlA-8%vrYFBKZhf zCCQAY}GyQD)^oO@ZZHkOl=9;aw zg2~6UycM@oHRjk8+oHxr)_CMUCtX5VKV6+B%lt+lH5zc@nUD}x5_;&kaCmSsFR=iA ze;HCzKPx7y4R3O1in1&;X{zWi{BdPri$B#_fg}{^f4$P}rpI8TaI#Fl%)zqFxkGr7 zG+m)5<@3SdvZz-vDXgDval(Uq9_lf`ogOTIRO)cekX-HZ*MpX=zFfDe+)Wyt;pWkBH&xY07sP_%s*yMbDd(^X5URZN)0xafrQd|Q|j(G@Fl-il+ z0GeqBfnFZ(yycabd1t6pXdA-xxL5$9y(S%Rn_kbCG+$#9OvwE)~o9txbe9x#OcX=l^=_a7cSu_Gr8zjz^h!_ zphsXgLdKLYe_C32o1^{`0ekRZu<^uc?FAygLUwSvYc?EtFu;XQJkuzgK?@^`k#UQO zQ7yksIi$BUbS`7*9JOZuAG3dgU|ongG`2Hp*h49!E)joN1Od8hm8 zI-_1iit3F~$y@vMU6vCjX6+a7*2YxFf*#dh!evej#Wl!hQwzRY;%LVOH{UAj*0{u! zG&ma@1RMc)ddIF|utkCl*kr#pS}kqu^Ci%!Szlv-U}MN3>z|79r)R<`_MpIT_+n34 zu5vvqY$Vw^`B*NaV?mL8!xt4+0W*f|+%+JC%xvpofI=ktb*FtmfW3(|oo=a2L*?k0 zQEUv1TI-(q5`FZzmV8_+=k<|cX|wlCp9u*LZ)%w{b;g7zH9HEfD3(?aIE(v}rcU== z%gH5+N`4Y5eSwF0Up#w{#TFgN4@QGaH3wk|^o zvhMsPA?GJV*M0W6->0q0bfAvFdfTKvSON}w3LYUeg@5e5ha6v|ySotO0R!Q;37w7J zfgO>lfz^p-5cXYOfK{z>yCoUVK{dYb%iG^;6SKaYpTyE#POu)3q#837A$sjY`!M`N zhecnxyreV_*ZeTA?u}z|(2o=&P!e<0868xrQ)#LOHe|qAJ`U`~+zt=o+$sy!pd>N;@wXUej z)0K;GbS|eU{@SvJ0f3AYySt`-E+nR!Z!UcPMB!fQHPvTE17HiK_$*w&&s#E)GtsT| zPQL$q-At}jt4x$qNXo!~Ie-{>y9)j;bHt+F-i5RBxsBogE@d7T!=cQ!;V*|CfumHW z|M?fPL=`PhY_EXX)g#aupJjT{D}yR+e%OQlposxWddDwdfT&M6=sNI@)Qs%1HIZq) zn)0zX^*OJvhR4vbR_`xfHh0yE)h~^>{YJ$TH&>2YWR}2emoMEyr}tQNip-F34JvJhOR}j9SP)A!%fDEKS#P#Cj8m#KlZabKO)#+Qg zvg8ac#jQ|>7|&MwF?^Y%(>)hHQ$Th$7$`E8tFByo{KKHAm&+fw-@+E2quK*5UNs@h zb$%Og=@$zuf^5; z1l0(lJKdDV&!xF0u?9M^Zl0N}(?cd3uG?5Z1+M(0L#`*Zv}}%>`VZ1nyYF^vVgUQ* zgZ-u$(Vgbp0KvE1Z*AEt-)MQeUrL@N z0NTA){p7hBfyo-yAu4k8q`GF8E~?)qCmg-cS~hXol7O+)qR+pjQ{2AlOZK_u+OifF zBP4IG((8^?J*f=0opZ{*kj22RD}?hR*16A9h9hNPJ*7=XX$I zjrPeI5^$CW0WSSiAY0mnt)7SN*qx2->D@akQ`uTM4p9`t>ESBOI@Ef++lkgeMzIq( zONvy>7+_1}z!&{d8^l8sq$8h6kL@QqB%MdPFFfY3C>~$8gz)m<0xo0WUI#5Xq}U6- zkM4bGyRk(MkpVvCE(tkGC=dI>IZL>TzY$^0cuO1VDKX&s;u^}K@q1iWQEzkKwxm0j zSHq{quEtb_TNS=9hf89uy}K?muuTa#nK$=&C_0K7!eFqyQPOGf6^Ae zI1%D>#rslWdcv%xST-M?%`ovhrKq^~6bm;_8O~W{1hH3dJPEJz&I7b02Uj!7LM31E zMwy!6Kj0qQBA)yEb3y+c zGO%d5cP?x*nZeS0)D?pPDg9X zOph6y6DQS$5h}fG?L`=%wES6P7g>7cjDJ!rB(w*$BAw}(6EdB4uYq0U^#%5D#@e!t z&?MV?6~!8Ex~_DqqCz|#iBFdH)vkLt1x%bGJZx6|u!U~W8L+{HoM+jk67#{!;n0op zAaKX>&!~{96K@5l#%7YcpZq2#Vlr1PsBSniRd)@VUL$w=8)IylFi|bjv*yP8^Hr`z zE0FCaTQK5kVSonRQ_PM3tDmnLLDqrZ;R3w1)BLURIDO2$4^iKI6jMGHbk-K7SUuE# zl&~a#lBdnsOeR_B4f<|-%-NhOu#kQHvAEs+BWqvrsSZz5LDiH*dYzoAIMib>?O6#1 zODWv>f8kIN6tOFsGdSJB*TJe%(~iT1IX$DrV)8*6jq$4-B}dPc3XljG!A33T%*G04 z9O({1w90GbX@Xi*5c8`U0vg2?i}{2)tibnCPI7dG>n?t;bKv?N!O;qujX7wlA` z(Pc=jFzx20!m^r+|{-AOwvOc-mL(9i_wBJinbE=O4zf)-RLHz@&~y9=La4-}4C(L#6BY#ANu_y?hK z=M$+`lByrxG$&tt99DXgL=RW{K5|Gfco*?qA6Du*EO~fat^M2rKjE$%MeB={ zsw<0VqaOWMAMb6h7FYAN=mX}cy8JeDs`kG2*836YtZDiKYjQUsq8~>+Am{#b9_;GD z@N;!rU2O#1PPe67#&(Pk78V!3nJlg(*8F`xtGJ|EUj<0Y|WW5kXdIAPKa z77m-hgL zmNiKc1+ur^fjQb;@NY96E3ip91WHTit#NU`oOVinNTG0;akBPk)e}}t(ilY+^%QdW zcQX=R=#YvVB@wZgK%~2V*Izah{hNDWhN2Ejn@qL? zSt-c0r$)ZjP|D%xX~usZ;*WSf?anbwfAEya=g~m;y&!qcu10)LimxnLf3lFdV6Pf^uvhozE)oByabyNWJ}=1 zh|;l?x*J`WqSaEItPLN=PQ7K}6`#(+H=pm8&d9rwc(pm}Dl*7e1iM<@ z75`Lg+En?b!=0-*{Mzh-Y=HZts0>RwiAOg{i~wV~K5Sud!O!phmj6ndHWctQ@eScD z8NkoH*m`ZTTg=KiM&jG9;J5vDDck`~C=Hk literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-2-Transformer Architecture Diagram.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-2-Transformer Architecture Diagram.jpg new file mode 100644 index 0000000000000000000000000000000000000000..0d0ec77a9a5a6377b1d4068bbddde964a00a2692 GIT binary patch literal 84211 zcmeFZbyQqywlBJ{K+r&N4-g1Wkl+*qhY%csC%6Q+LMwRi009C7cXzko0fIx}?(R@Q zidTDgzjJ!`?Q_q$Z;W^U>c#k=s9Iyynsa{hH+?7ILQ2PWxMnXYBMnyqGLq$cL?SuFnK*dKR zc=l4_(NonA=yZ;R+LsLuposO=FshPQjrIod_i>sSE*uyg*Feo_Wb7)v>+}HSo#BWK-S=l+c zdHDr}MOD=`wRQCkjZK|h-95d1Kl=y9Cnl$$(=)Skt842Un_JsEyL+c+=NFe(*RbC= zf6#>lp!|(2#P7cm_D^)-Bj`d#MMXhH|AQ_hWOu|u!AC`V_VN*dgev+6$ES4Mei($W zV=^l{9@F!voe&v2jbRcq@UA{T{e!eWDf^!h=KueRvi~6LKk1qSuu+f@!b8CaK)`kQ z{=Cnt&aImx38LB>CEoETZzok0DRIF7^vUVcCmE^fh6kX+cH{w|C=_vj02t1@+pjR2 z9)K|=y^05*Ao_mnt^g(d0SMd|e*n^^+hOdD55T$$q|-G6GGeoQN4)R=oHp(3WZ*i( zpO&J)y$m0KmAdT*Aad*hcntM>07T0nCwk5gK<|Eb)+fOc*xMSs--ligfd70toL}~i zNf-U@s`UZr`12;pqQZb}(5lcVXuI|pj%5ddEz+E%TGE`)T|WS5^`JYuCZQK7-~W9- z|K`uX_s_re>)-hEZ#?|Bz4*UlPhVZH)%#!U2J+8!=_zPC$d_!tmMz0{Ss62=x|Mza z`ekYmo`k8<4Y$GzlpPQi;vJDw&U^o|_6v2b1JEA*1K#9R_TP^SfE;pCg=%d>I-R1u2Ga3%x zG02%}Dd>QV1RfC^2C{7b{Q{bPg&qC?$jD0rKIrT1znRjWBesp^G(60DG2~tfaumJ! z&QSOPi1k=AxJ_N9Ll}pFhH(J zMw9y9=4Nr@0l*fK%-lPUw3o-17Iryb=jP}Ssm5*lUAEopi8KAV=H17VjUWcac`tu+ z2As;Dq)|xQC}BOY=oQv<;(2MYIA@%^oD50&0Xg{$Hqz|(Pdk7pp?gatS9^*@iITx= z%iWycP?g?m?!CmN*LEMidFEGPkjPt;(@(Ml0;(z}4Rz z*;Ve&M-#z{yOW~>OI-~(de{!OCs*}6s1Mf?-nH~# zbfc~6Z_|U^rv=JoxUC^GOYiII;1g%n8^8Mo&R_U!y2xIY=T*fQZwm_$6J6yz0QE5>^)#k76uh@x`lHB>Fuxz#}?H#H7?J z9WS$G@(}HQb7)`JYKLQXqah~8F~r>HD~HhekHhgMj25L*J?r8uaH3LmUGmMG4m$^% z9l6{>)@Droi)Wv0<}T0+qYE#F{mPYD2<{t99t)8; zrEXlqQqt>Y&evs3$1b$7wcr_SlP9rqH_k?(0;qOSC{~Zd=GMF5G$-&#LihR@-ICEge-P9#u9JIW+B zuj6o%zeF?YX%jA;*^8$Im($V$5hX|(A1oN{Gf)%x304L z8sS^!?X6AfE=r0dOg||;){K2M=T`Sf)-yK~X1FhZ`sOBl@FR32Z+^BQRHV9%yC>=g za?$q_G`*9A5a*1U_z;agD3*heyWiYeVfPWnne*E1| z(H58a^CecmJu9d)4JT!d*gx&X>8r+KisG|zJxMDU4nIV1HR415d2=)UC0|Tl7RTvh;l3I*l&IcB}O* z6+YOFIs&}CuL-Ax5h$aBsc*GT9)Qn$1zVAtb_q7UjzS4Qx;bh<%&~UfiC959_8s#_ zi@FJK#cYM=wTemFCjAZRoU~t@UQY1EpYp4mLQ7XzyP+o#X7v&HJ8of({|h2h`*?Vqq8m)t zQuq*kyinpEP-YQ4K1|wZ5}lP*O1mTpy0a-Y+!139*VJR2szIS6JM=rteMa_meXr#K zP*^*>9KqQmKWa-SaJ7LQvmYA@`|xe1325P8E98I|l-muNMX0+mTobkCfCGA(IHNzd@U10X+`?R6FeE_t#_7pqBw|ZMU#dzv6K(euD;B#iV=qw= z$js2QnEn(Hx!~grYu~7WF~J)Q9)KM}B)-kp7%HY8ua`v&z>MWzF%)Yjbf5-(MOD7Z~AmeZ>h!Y z#%55&p@jY@UhcFZ-M2+k8Ea|RVVCBq%f6!aL?CoXwDfC%W;gPyj*i{yujeDYNYHmS zpr_uFtM9yJOTIrb0gf5w+yS+Ov?TwPGviW zep(@Nf^l!1kh?}h4aPf|OAj?4zqBQm*J9193C)h3cl_~TCc1oA?M_xN60gUQ^faS3 zVXI4tThf!xV2IG}Wjl31`gm1~&w3<%pC>X{z0>(sVS%-i3|{=xAb+loa6wv=T)+mL z(kDh=dwSAJPa5Lh*tnr-mvwe;y9ryAx;_8|d^mcFCy$Cf zve0i`9Q{klk4*JH{X#Dw<6FP!neG~V08khBB5tBTd&{#Yy`SiJ!C5e zdAuhx{T}&>TauJ#^OX;P6=mTaF1P6C{C!8G1pRW~$S$ z^W^|B@|F{5dSNf85;TQxVW((-u!oIDL~6b5Qd4Z+X9uP6lrxxjnP?aCesWwXZ0%ic z=#~7_sEE>;uLEd9oejO+o(`&Mq?Z`x2F1wOBm!fg)5 z#TcQl1zrg&TBa@ugJW`TH&u|AjdZ{Mk@83Xp!lz&M5(K~vtQNnsLCup)2689qZ4Mt zbxXDf#e`-mNsRB2FaYV&sn;)zCA|_DXLwA`Lc3GMg(mc$F5Z7A-o9q3NmZ2m__(l(i$gWWS7ZvL zikV}7ymfU&7D=qZXPtBj#*e_1(_g%~nooti#ZTFq4g)%H^SLaPR)sd|sPP*7$v(FQHZ~7K+h<+h zl(-b2i3F6au8iny)4$Tmns&d#BqbnqTzDc;xZe-QtpCim0(H6fBwl=eF#VY8{f}P^ z%t8bLs9b8I?gYZqB`QeLg#Y7(KA2u(bobRp`SuYa|B%g)6+OoZSK!OP51>QTFJe{2 zX(V9BaQW>}$N>=oVX~6zKRf`cG`HEmwMe2KfQ>Y8BD_P}=!Uu2jph>n0cZln17iP* z`2X@fmbBa=9V_BUS09Hj-NI=(H%JSkXalVNaXoWaHIOywyxi8xU007g$w|Biz(DHb zodN#{f2h<)NbrkC=x2H3dR6-4qL^XlBL~NXkkJXIdl=2S?3H^MXvZoJeusCOy?DfF z^>lPU;RO9IeA?)WcoK9`q}u}@K9#-4;f5^x!TAgayKdm=D3Vrd{9&(RHd!j73}&jS zTW8{6V@KC`R!_%w%0>R+d|rqJI2#_E1Umll0Fa9=tsc^OQW0&po1A_G@rpes;DVFA zCj|}XC=tu8q}j{5{zRpeQhIHd2YFEvwMVvo)&Pvi=o67Mn$xP~ioA zWpPc|-(afD+3++zDw3-vl=@o5j(84N4s=MpJTjN8bbu}^?Qv@ao7VWEvg7J`X>E)1rkYp>~IzHszv(& zphpjAKVW+RZWKjb`L+_pXr_<0YLg-OEwQ>k&HN(hH+*uP1mXK!ch-G>Wh?bKQ|6eV z$9?Kr^bPsnzxf~8Ga)eSY|{p<4YdI0+w^;txhX8ipiF65Cu*Pd1v_Q0f@os zHoC$_)ZWIF^hf@ImIAvA(354I5i934?Sv)p9J5L%*FYDl>k-bMToPVnf{Q~zm&Y`4 zf0j*;Nl)d_5219?Jp?RtK`OMANE5BX2=3Ed@b*6dzUY77Go$~bJ(KBeVR0@|s<-Gj z+`onb`b+~Ye+NFyXjBB@zpG2*{NGE!46K~(i3 z1b2$2r>~l9ix-yOe(6lr8Ts9xfyJwh&Kb$)^I-t&{stL){m|W;`39jnM&p^U+Bh48 z^H0g8mKTHZ{cbSxVi#k#Vs(ZGtA&k-zLdn%Sin4?t#~^6d$zH`=|yOD-)$ z+GBghTZd+dCUlT%hZ-q*UoAhhML8>Vt}dPEzTkF3*I__Z4iO{cl|E*wCQr8Tk+!hB zIZ7{^!wP@g{mExTgq!BIuvR2viOQcTP)%1Hg%O1so8sAVQtjfG`v;=RwT`jiK0WBW zXuREUqBaMPBN+O^AE3fpUXY^b+?v&LF$Zpfq}da&z4;Sec9{Qi)6~aKJ_uH1wVeM& z+<)2s*Ipqtp|vZ@ztOE9fGY|F)@1G%mh)Dhz+2zP=9Yh6dY)jB+Sw^);l?&J-N-N% z{g0^Cyde6&UVr>Pyro4~R?;L?=e<+@+J|-BEyJ8-$&X@JF7Q5_8N0Tk2 zx&Lrk)-DhI$k#04{K26r{}eYOe>AE$xan)M7s*amnB;%w28rq4XCYucTNg6J&3RAu zoba#dN8kpp4|1j~qKJr7ip%c_r<(C@K0g49Gy@ea|5fE1>EfSqu|znvo3#hF&r7>` zvKnP9zjI*=(8eryYD{%`vm%t#9syDnJ=^dIUZjOFeVT9)$-I;jb;0x9$(!kcPv736wL+_j`sNrC7)>Mk?yMWpD~08c1T*lI zvs%XO$!%^}_HLxtb@s?>TQ^Hf<9~eaq6}U!Q!HSjaBvP7GE`FRN#=~;HK5mQub5=# zLv3*8)ygB$iV~f7Y>2R+ythPY&1%4dy{T1(X-M`^*ggO?nA;S%BbW0+786U^r^=zV zHdI<$gr~%j**(}SS0{>ib7dUgBEDz5S9Siek_;fgaG0&dqzsmfhdX1XWc9(PV0R0C zwDs;yF(xtzMiSZcR%PVkk@X?xuDYcc2KER1B;*-OqnDZ}k}z!{zT- zb?{B3rBsTt0o+fj%uanuv+?qJ5YJPIhA9B<=(Prs>nXag9KIEx=WYo9 zb=IeiDM@rEPJddq)p@n7HgR}d%0QsEAXC-74vP<#C=Pp-m$1$M0AS8{(VQPb9)JM7 za=5VmvcI?J>M^lpIYcbR-F$Bc)p0-3$-f-isj%N{o8Xg-P(Xias^pq^N)9aeTA{5b z6`Pyw)gc+~G&c<^OW{1V)1_#n$FN`V$cMb0vH3kG-J%5@MIl*^(VxZ>Ccpb!=gx4! z`@32;CKI_Z=bYZ=i{>FcrA8Vz%i70gcjH=S9}c9|$Lx=lrjDYE5z0`@d*&}sN2>UF z#_YN3yZ~)fk6c-JR~u8ssR%#!4BIOE=GnJF>cs_QG9L|pE}QQSF`T}P%ZiUM-|bfF z4zI&KtCNu7SOE_J3K6$1*$}J#Y{n-UJi0Je9+ddWk4GXe!LEyhwx3N7-Lj9WZv5tN zTrbNd#xGfY^pu0oX0K&gj;NQWq6W%x-yxT9NgT0a{Wz~nEVcH;dik=4DG3+Sw7lW# zT4Odu4PVc6=qW*#pa15Q8)sFs(6VghMv`!4MhD`7sEpLyeP031)mR^7r zA!W^RNo6z%#-8W9I&&ygM&0CPjhsWb6ZYu0ETEg2yCIj0B88|1hB zZ+4-LBiKte$u{4zZOq+MgMT)2vBp-t!;RK24N&vwWt9-4*m*9|9H-Ut{eCxZiUZ~4 z4SiAc49;v1Z@^^OE+Fuq8k2uI&C& z9JnyNiq4d;4~@~KdZN&Wvj)Ngdr3lFY)EUun*-au73pmfEsc$YFsia07n}tI7##AG zcR{OqU=z5{@dI#TgMhC-n^35TRSqn_n;d5Bzfom&+mc>1$l$~x zDr=&IoXQ-2b+fj2MikC8WC@j{HC6_(*nfe0Ym%5KzZ*Zg!^gYIO}$(^3L&HUSkn?W z-8l|_h~FfkG($zDrLY@&S~udG4*5U^Y3=rwO}=|JXyy|6|aQ%Ao9#y zJ$Z}lS>0a}2^8RUCtRo3pCIuT3?`}!dPPmasWwHrwXF-Ur9|;u#WJq=ti!$@Rwf0l zmg??{Pfd3;6GP#6mJhp0mwRFSwJ#cy-;(qpnA8~3;_8h3a4v^O9 z^C6R+WMr$%vNF9x@`!ww>UJik0_x@A!dXG+oqeIT+BsF?jR^hddEDItAhcVk5ueEN zvXK2FQEI7marW-kdcRj_npE)85;uwxYidz|D7Ug<7gMbzl(?oJJ+y_{)$swKj@e_2 zQkITcjiG0PKl*{^MwGnpxwx6Q)SINpsJov0_;Y!7!-w>0tBSW2aspPhfH=1dIvZvGIh}*xiG?*wb7-DDx|JolZ!n=}@W-ZunM zICLg5#d1Bq(Jr)_3#u#5@A9#r7j5Z!@hc*vOi&95h#3uI3y7!ejV_Wb*>zYJFAHas zl%?&G*;A;hxWvW)DYU+$CNWJm@ZCd#hdLGypWu&unv3X~X==P4w(+g<`aZ7ZhUtGa zyab;uJ<8HCwkztoqz=x{@}oQoDbNQi>V(U*B;}5m$=|Y-Ltl+7g0>sPYIia=Otl}c z`G{Bx;S88u$8e^Q9g%t5%gCQ<2{n1G?CvYtw-sPiq~Hy zY|X30DU52#gfYL@QAvZoGFqe|_&|+~i}^=Lrg_t<2l=b-`foK6dNy{)(!gsXWwB z`J@d{pK+zt{+8-_HVF<_9V=(7@kntR+yGXkhjU=ebl;sap2*0G;uu7JjFyA|Xll?l z#_82>PEps2ECEMlIrxW~y!(ry73K^IY7Y#Mc;2p5T2Cx zty-)WqoWot!twjHEtsnJCS9&c=qluMnK4-!8V%dva`gG{X~!%Z;+w5_oXf`F`rdaiKf^i1CV4;-F?-22Kp+0P#-8rptxE)Gt~+k&eaZ|nUc6geo34Vl1Z~9 z);_&!pirQRnSq7^;QFUv%cC-5D-iP{9E}RXn*paS3*OWjb;e|2FhjETV?t|#nG2Oi z^7&A_$c*Im4xwKv^uo2pci~f?$Y_0Jq~*ReX(g6`+ot%0OJ(+D$Mca6ZP}4a&SYdX zH@*bai@Ft%lJ*ZvTRs4tZFLVo2BL7)e)1mD+2lo$VlyN*w>K1DY94ADl$iM%N#B>W z%JitBniq_rnxLr75OWWAnfztUZ8+Mo?&bIRq!B)IqF>g@KD~Z!u7}tdvx1FvG)@{vj&L8} zR#8-B40_s$i0=m7sY#_ddkHtgNjR6)=7QP+NBIq6D^qhTA|%Y?-tt8|g7>}B-6*4! zroP!{W$@bA>e1b|J1_p5XygA%q|veKi*+g%oTy@D$X*i+^(wTJa@wUfVRmDaIp(!E zD?4kXQtxxvHG0B(jHPTkZ~p<8Q)DEbGyyZ$aiNSS?E*>3n{18M!1`LX>hyVu)1InS zVwG5GV?EvBBFRY5Nah04W$wxLCk|fiQQdVdMZau{_|I5Ac^Zj%>&^ax+myGOBh=t6 z8k8B`w@@W7zR4xs*Hs@oqgyssP6Q5@n!LiBZKUv=9SK?bfS;=pV&6y6an_bG=NW{q zDxKNyvQ7`pBHrnRB#*7G$RU9Q$offl(`qBL`n<}#rN+ur^wRZ$JC`6rEAJ@YJ;P1W z6bE_q3yPeeOBfcN^-;=byT1F;@8ken4w>K)bTEaL^1j1oR7?@%2>RvZpKXIS|4DFv zTC@NqHf6P~mNO*xWb~9*btn88_{lps4Cl5}__#yG2Vkr*J29+&R9|dhrF%zQ*Ho$1 zNTA~p_Y8yN+P(qc5)KGEh&`H9OsrS@oa{no8Q?#7NhR$JHDqm=^c}l0V1>+Hx=rK` z4U|n8R-HZaydN6^9H%`|U}DoW^t8{5aA`=>qg7@e%Lq6 z(=FY|rF&3kQ}V_#TtVOH@&J@-rRct6rR+QrBB60dN91kKSUcc|_N*P5CYG>t3c=h4HCEP$;zwk? z5<~1Bb8WTecI(QA0t^H{uka;5>I$aje@W*$6d;|S^b3elVJLGG8ji6qlEyCQO-R%# z-=8a)N_d~y(6rAOg2N>#MC$ zhy7j*(Z{4k+R}N@U*Rfa&&H1~ZiP$HJ6IVrU-}c>Vz<#WeI}HdDNpSxWh3OzSlr2WXbD z^)+!!Mr59NY&_Z|WstyzMJwemJG)+9Prj?nf{p@MxT@ao(wWxQ8m`}3qo8a5jf4-g z-@co!U@l~ZB^;U0dj%niePK-vl!4iHQG$&70?p}SZ5B{L$?B9j(?D$4eP`%<=-UhH}&v98%uBOxtIk_udP{&XVme=MSGyrRnGAM^^q2wbF0KEYMo;m zrs*I=&V7$H@r3El?+B3%JVm4I31c-uLmK4DSaF72g5 zf=jwiy79#8i8$>QU4y%i@eqfJ@}aO3_Lk9f9>ylXt?AcjLuS4 z2AS*JY8YI`>d2ebH(CJmzYU`RxAZx#*^JeyFq3w=olT(J2lYv)EkqiF4%z^FP_y9^%bTW%4YEq}*P` zQ~^_AO~l2<1ga_fd+Xv)$=j2{rCC1ft_T)Ch0s9K=T-;c+hN2C#(qg<3Gcft^D z7_^Aw`hGR!neGJ4=qEG={d39?EF_2Csfg0=H~@~^q_D*|@|&65;Cuq^i(dwm9^NI#=3=_R1*W!pA7kXTA+ZSNL0f%oW?TwDsJ7!ck5=4HN8fG z;NAokH7Q#sK2cP|BO*O!cf2vl`(WwhDzXUn{!JUr$#7_H%>aIGqE=%QCQlmvS9eh) z{TnDeE}Jj#ntwXc-lxbQ<|@42pb_OLi&IJ>4Y*|3NS?~d?XRMoN2GjXko1RU&)Vp% zI9iD7&BPu6&nD~5OT6DNbV2uy_V>v(xs496Xa)35^tqO0Er{@%*(bZ#uQ%|#`8>sM z19lCq5V^MJuSVbv?Pu0wt{3LxGkca|h?-&H0nxF~R1)9ZIzA(e2fv ztY*s83#w|McY>`V{ElXhc{~XQLKjsf=nY|-{hA{u@wCnnTCWN-3~hu;l2fk(F8W$x z;^*Y2}^(2~c;FA+`!ZccbCxaht|ha^J`j&5T3;K*+lM5GO+!8b?}&ly^uNfy z3P&^~M_M5Y2S3{{Ouwq2kNW=07G=$#f8Eh6j`t7EnSWO9%UMr@V;F1K?-iDo!+&0& zHq_ef^Ew|XPnw*TYZQp}S#v7w)P)?HgITZfbYl@*!%jAJF+|-vG^G&Y zk0t}G#C+iyM&A^t7R<9!=LlaNQkye+Jp0Cu*0OxXBl*mVH1g*LGsjRYDW%_G*u(>{ zb=3N5xigKrW=c<^Zj2XJJ`-z4uqe54DcGmJuk`>_@s8Y}>q5$Jm2VKX7Xk93LOzJv zR?{eYf)3b1X40Eyy;bJ+t^d}ceR-hMFTZ5=Z2Au%_cU-1W59h(5iAye9k1$bVWv*2 z|DNz0GaZoU0FD2AWgN8i1MfcRN`t?!uuJf2QhubdN>##bPp}|Uo98rNIn4OxYH#^K zWk*}rh3oVB#HBDAc;&;`x)~Yed)wVU(hTh7K>uz_-O&jO*U!*0w8nngyr*`L52Mni zc8G#jmGSCBc$3&S8}!0c!b2JxV(4drWkrGv9Wnsms&o!Ogr~rH%JOgKU4#x<{H}Fz z9Y#!D_A4vjoP&mnvkQxxF1QLCw3+hv+nADvS0jzr4M#I4%h>7bcz8a(+vH}=Fh=p$prkZ)91XLNw9Ad{otLQjlDky- ztbU=@BoO--VjR^$6p=Dr!YH-Ps`o>6==Q@W$toZPd#%ePn9B;KXL=5%e zZoi}qkzCox?LqLr>-W_7;tkov!o}uo_O8y&F~YY4Er14UNRFa&E%w{b&=IbQ!vnyQ z?Hqmrx(hpjv?2Nx@XNv)mr)*opXKPch!lP%`A!jw_s6Y-Et*i*;zMF#6bm!7{-!29 zbE_k4dHj!u3_*5SA3R;@lD5GU@K2una2y827%pFSiIu~?t|&BBkWTpXfSqB17uPj9 zs#louzQENjU*J=v&#_yrhXsPoF4RwTjcUOrKo+R3=&F3jLH% z?Q?2WzauAT_Q1El?^tF3atW(toYeHy;lSaZZ}Zr(PRfi0Fr>xHamJP+4r0W) zjPov14TjiJbaxO0R*Xg(OoBEj5S>E$M!(5(PbT2fFnH-7vYDenVL|xz28wWz=Qo-~ z#f6Z*z;*>3WwF0F@}-Aw*VaZ9n;UEDZZ*5Ps{0-twwWM(Y3%6r)WY)vZQvoQG)B<~ z-?7w{h{*Sqv!ZjtgsVN#-ke!ord77(h`lqxAO*x)el+6zHmyd|VolGhzA2=0i8^ZS z&2c>EjIc$STmWMU%&NVz_FlFtvZpz&zzXiqZ@+{{3A#z9=lD0x$-2Cl_-YrW{1*vW zrk}CO$d$$+B_CcF*_(v0m;F$;S0Fd_pfQ+iIjTYp)kEGRxr{vdTki zZ@I8~D}~`3h!`yRq2+I_nvYR$;xxM%3ngRvEXhNcM32*Kg8y^GoBEQSP~$t&J0q<2 z4HD43oBal>TSXfQ2Yf3?~6Uzt2uW+1w$7(*WhX7Iwrd<0|HU(w&-I4c^1-o#j zxAe94)kY?*HG#-|_q}t=N>j0z-evNNA1tUjn*3H8POgI3-g5Ob<{XrWcKD~*kc1s% zKAD3#OZd~uKPJxFpr#nJw0_@`37lc8O9s;07O7tgF+$>5RT1- zm5^}iYtyg!S6>6nXJSPx`nQiP>U9(Q2!z1p9nTCt*~$k1o_8UoIaz2#!H;X4Ux)PUm(L)9GBntEI`Ekp)!^kILIqW7AUA8vw|uTd z|M;H)`>$uG)L+F`ZnTq)5JpX@`cFkoZe{c7xG3jsM>t|B7JsTJpJmexg`!DeRv7E| zEkQ*p_wKpe;cTL`&^)|K1JUY|oAB&Y|588v)vv|4UK9D%r|eU|-1(^yfLGlY8Qbss zYcL?j|NYOcyjJ@I8ZsU0fX(VDqy59yS_-{;g@ZS|UC=_;d z&_{g}-8avytqmR+r%(iUJG>63v=n0T`vhW1&N-j`r4Q*>m_UTZ6F04%atQNgXp5+B zK4(=%SV85z%UybeiL<*h43QGUN5oXpf6f}JP4Y(aXq`c7YW5XJp4azWeIr(i+-vsh zxY3+@fn0l)ZdRnrmzQZ_@BXoyIC$-O46Y!V8b&@At90cpMs1)j#$7qBj7aOZ-o@6J zQ7_fU#ED9H_o-N>Y+d$Qy@bseKHILn8+|&tDo$kn*KSyp&5^(Mf6D&Vi_gHzxI;kj zzk0$&G1h<48T5(fU-hFJ`V_&cro~}pqt-VAvsR_q#SG3vbv&HO9w>PSxV1Zc{SUy0 zB~XtbUnsj~&3aDw2WB%+%elLWo%2N7OL5W|*u2stoKY8t-!N)jAw)h+FFx42yQC!F z8L3x%$XfSKoM6yQ6dF)4yGwG$rqvYDD;lMDL@eI(h2L1#al9BD*U9fW9f4XYUTJl< zQ~yd$pY_EoL!im#url@3FDqZnIC8sGem_IuW)FH2{_{P%t7esmCYHg5%HG$9q)f@c z{-rX_V1DZBEgGCXZC%6EXHk2sqYAzPL{Q-KC^vsz($SW3jO}Vds#!*oOD%`GFASJ~d6h2gSA%2MoZSkNvt=ns zbd~{(q&KF10cq$&^V(D^R3M!iTi;{73O(lIOX9|;=kcaUm@g;`_g)4Zu~z;^w=$l< z!$H!x~%g=;~wO?LR7cbSbb7ZyFgl#>-Aq*o^VhNY} z#lmi??a&F&W+fvkbC*Q-D523wn zAqf9+76QBE8I58j)?Ub7yY;%~=_Th3vvs+Rd<}&%+m@}?8k`k@o=ri*5dugrvsMRR zoy$(}rn0IId*fiY#h6avhhVTQYk}_=!Uz`#}BqoqA z1)s`_e0h4G!l_}*18>eGrJo6NC#lO{INm8!9z+a6pk09OhzAiJBSoymAaa8_rH^w* zF_;5E*k5+X(ZyxBbAE}_*As@_42i8~@n0=}=A*&qAF|7tv4W?m@gGw{T4_Q4>@M+o z%9ZsKy!$p-8ngEHXJ-vK?*@_-VGa(CiyX5B<_|8`pok zN*Ag4QY?NDMos3*ZoaRXBxkk3hMsW%8p*X$vYgd`){7#TcB4sUhMlaxCtOIAYgAyD zeQuj7D$V0O;*Gi5@~ZLyAXFFPx)2<>EK6uud}M!M|8&_(ovKp5J_Xm)?g()yH%&h6 zR^M&j_;1=mL-zsepY6mBqZ+c7jHH{y-`G9?$l}_2#RfzcHcbjPo+4L_>e!JX!LDR4 z<#M|BBl9@Bf_UO*_@Qo`dZUnhde5^ zHB+^s3R9&_Q{_tsdZ49HT;$*j5tkyNyxO%$y4yF^P0$OzMAh#d81SLQV_PPs>qTzh zrlJ(Y+1q|WGK%Q<(c?s%Itd+s~HoCW6Yi@-s>t?Nex1I>A4Z{5|Vh| zp0xJaA5$$7K6r@7=Tspc)wAc8n!P8hZ)>F2R@dApCpKc206ip-%B>gd5+)c5A{&t= zEMs0X8TjqacgX&%qRM}K?j_|DnabV(L^OnQR#=b~v;FnbN{fp6N=5(+CH>B25`>e@ zyt-2*bOcHv?>{Ze+}u6g{ZePgNs5h|$9~80~emFh!3Mv0R=KJ z&wf=exVyXEd{<)7=4?>ucc)8g}H`>~x2UmII(I1f*W z9HO*(qB$mLQ*%zTu^-#kmg1~5gqlURS3lo;XG(Lq6rBIh--l$(T$V3gxJP{RHqo`f zpl#60GFb<&f^2;Wk<7-NSUX}EJNXr#$fW@xXJp>cDI&#RaA#Gp_*W@B3;+hpWB#W@6tcK zLkv}o1xM4zbp7mxaPF$Wy3}upBd*52FD%0M_^2FwPwyJ;Y%)#^Q}5HWB+2&c)5|^Q zFh?(6u)&qL0$nPk4)1b9z*5oED{>?Qu3spwsxF&yuj6p6v|CVB8hyveMI2bhtZnq>6=l zJo8$lR<7o$*P-h*jyrvWa#U)-=hT!CU65)mY9!rOSJTWz;Q?J+7>HwZ=MXoeEnEyp zU*lc=C(Uvm4W<(*PI1Dc*`14MpGh80vAnL@CCOS-w1nTq#01ap`4m(17}6&8Hlemr zjky>Y9rBlivL$NSW4rA)5OWOmf_Hcf&=c2{jr4*d5IeDpwWos{K zbup{O+;{b{RYwvT`E~)g zm+5t~+)RgY@&%(0n$Hr;o*(99I6p3pd*YX94Hirhc@&DoitM+q#cgQg?g} zT4PfxWUR>)QQ*24IrYxVABlP`u`{`*sW*T!@jXYT-H@$k0L)x6_zn+S^qc11_7Sg- z;j~`!qzk*u@Qh*6|HIsS$2HkyYr|Mk6j13kDkw@70g)O&X(A#;N@yaT2uP6*0g);Y z0RaI8={*Qiq(-`k^d>cu(0eD45Wnp+Gv^g&=FH4F^UnAF@%-e@6E^$atL(MbwTgEp zOP+QN?DkP|L)6DhPpAp03F22C-1k^~T3U~<9sfagcP;P*HqlTTcfYhWR3HA)R>vzB z`T40awr>8I(LXrnw;gf8nR~Crr!z#6_QKvm$~~o5mLAqd{JfX@-zoKr4U3vOL1c|$ z42}l@DmT&((tKO+V9BWY8runq8;5+>SrMIf&mSU9VLk`$4H@pufyD)$h#nS#Q{!Gl zHAbYMc1gOSE4O?l9kOzBBw4AYfUl`A+>S3Jh~TOi^i?j&4EETsmxIZmJuITXDARfX z$UZ!481;=WRkzFa6JzVnigLUTDud;Axjb!aX2dvog5^n88@<=tha;PvZtwmeqv5M} znjH2WAC3M&_JLwErAsFPe_2e#&|?P{sYD~qL0%k@{O-tVp$aO*I;EG+3}D&HvVY4E z{_|)5e-~n%9)sUgy;FxZCqKVcC|HcHSoJ0hjd8SQR-HS+BnVMoBKg$T=~fN~Eo~_! zUJ+%s-t;_~{*|E8=DF}RX}~P@^6VEx2e-!rKk-8LoDSuUWZ^E+e6z4V=rB94?_-u^J%-k&OZ)kLS-S_qM0I;GTnAgKOIeR zn$^JsAs)LGk~|ebIZhB>TQu;g+BwCHuVw4aXU4~9UivB`|!jeAg$cvUFbOas!9etqPsu%)B; z*759FY2Jt#+t_URnss7TVn8%T3b{)+cdjv7LA_ZP#nC8zB6P#~WtMc`rp?!Qd8I*6 z$zn~q2;RWe+Ecu_=4{{;#1H#wp2TBxdJVi*B#BPC?Zfr$$D7PNo#--29T3w`1C?X9 zh*y3gO2%S!m*)*le87ZCfol_tuEE9;HjP`{mJ=cW;y@b}f6wFXXMgq=D=isv=GPP~ z9_ZZ(pSUOU*a>N~9uC)9`Q?*Gr6 z|GU$b+2O`JDOV3{`R8DWJ%Zm&1K6L{z!+m0V`v{93N17>p?rSv=Bv{WDl?ZK$2<$Y zl1@&ILr{6w9+A97RsBqYxw*?r`;Z}2&z^hA=Ry2OJ@Sf0?}gsUaC5nc>y!>Y7gn`& zQj$2C7D7ElEbgYpr=K`~#*td?D!da~{B)o3RdH3)&AJKdujokZ8xW-|Jnq)HUj&)Z zTqFnySaiywZefjL9P-`xw~tm_({G>|9kpe;9^6%&^`4hDn$%fi@q_GFKDG}Qv8Pd< zs8Q#N6{3ONVLjoy>%zLdnKo3~Pbr(}VsxS>t6P!}p4Ilm?qaaTxB_8Z->Xi4OqnC% zlk==N^9G#FxQQogZKz)w=&5~b$_9D+EDH6y*EhN)M~;n!iZc#()}Ro1bLGy zz+|Q{YHKG4!Z2V|RQPT7{WMm!_w2gdn-2)z9h7{=+5-@Z-$8{UW}=Sm%N#G+G(OzM zS?~GsYvbgHbybmjJT4+?75cY(SE^MC!td-pAwSy^an_rszhFZ5yl&saE4h0DIn8#3OsV^$plp3(@_!pLpM2keXor(xQdohJ!oTrRMda#;m zyI>y# z629I~)pkou{7UA0VJJ|FWh4>(s~VvTPoGcXRCO}8$_P{15oAI*i%H*p=(2e@`Ga>? zA14y_kB(XQ?BjfvKz|h^J`r18s2Hbai46mQCG4IA2K#{gWiuCcXk(^W4Rq~otO(evUsZ#rDx7=%6w`ACjmB2mmSn!*PhPIg zP7!z%*WYJOskj7ci_f+6cpEip{@n9b?)gc}R!p3PHqPtC#~E?cMz!-&Zksmi#*}~* z5j}WWUXy!i&h0<%gMzB^g2xA>O>^6J>(-NKSE=Kn8!HEe09sP5cWs#efz#NdpQ<0o z1RzfJ$BJh?#}lwxp(67gPk4g*Mc_y}LrMuXs@Ow8@fR4cLsrwvS5Y~6qr3^vH|}}b za~-ZBzqQ(58^;{w|Aalp;f{ELo zmx1uXkO*`gi2B;q72+NU4=PEwK1MP71O#DGN9em& z?>+yg(~LrCt%^73Le~{HSvHxXVF`xgIS{jLNCkQb!phh5K zaEhFs0r5TQ9u`ju4@rOJo>ye1j+EKcVf10gJBOXJR{R)`ZcdS=s_7Nv_d^-#S8fJQ zls9``2qq8=)Gow8ZRd25;@1&8hW=FiWK}>~`9(>-fY7u`u zMp+XWPvcl?&HLSRizJ6Ea$6wz4;X`5@HIN4_MIarQaflqn8dTIzMly2JV~8(l;WZh zc-43NaiFH2s0w+s*11onKiEQN zLV-H4g#GdRkpzYa#4HrV<$TD2t*viTg~w$dw5#sY*kB{@y%q_ zKJpzV7sw>KqrjRZ9<-!liqccbFf1o)BNfz!lN4*HPE%Ee%G@H8J<9%EZ_c=ACU zDsYFikKW(rVQ3bd^RHvxRRl1iZS&~04zIFY%xYmZ$cT|HMqs8vj^pPwH9uz-;}dY5 zv-(P$q4lnE=BCd$9!Ye6t_p}Bv=*F0d>asR6WVs$*g!6aR{R&2I>Ib{y`=bKEIDAG z_1?)LyvFVzJxUL9Y?M^}*I8l>4`KV~&a-&MP2ZpB65Kq^lWK2&E$7JY@=dLzYJN=%dhtp}vx|eb82$A9B~VN}{`_@*(ietgVc$iRz$- zVC_Xk{p9lY6p1gW9PMlGT&07F2ol5(>6Ff2)7$ z^(0BRKuqGjIFjQoX%n*3mG-$8T0_3S`TGk2llWK3{do>^Ug1K7E(AudBAKcoE1BNy zlVFflt-qbFGl3I25cjKa8*B%B52rw!!)G$_rC+sor%;d__|9VWdS~l$g#0LPr^32$ z=L#rK^rn#f%bJA;bPePniN!cZc{|>k^!iPtbQ0flN5wd6<87GX#JiJY-+1B+zrLi>$f;Z9opf@+il-6CQ>eQyu( z#?LSr7DQhTvZSt(X~KjZ0`J?@rsU|&X@@`ceDbm$-Plt2o~^Ugk*k?`p!nPBa_C*i z#+Lhf&faO{v`@?(KEJ`lQx*&7NLtB5Wyu>&2<}Orc3g#z#egtgnRN8Uc^~*eiS+6G zN>xR;(MxhGoKoDj%_6QDf2n{Jxy_LLxBt*E!rm`hpqmg7JVk?qjgYjVT|ZI3e7@v+ z{V#53G56ykvQK_h9vs%V<3)Y7Z3Qrev0l=~z=h{bzJz4U<|5Th7gr;~=S7Y-KnxO1ktEEnKAzeRZ~!FC67h# z(eXi7A4Bo|u(h2b^u&q2fZ@@SOPVh%)GseBlv~?LZ;KB?wjm%O=h}$I(`171I_+B} zMXXt8@*a$wluBv+vV4DYD_k)_zV`~DWqjPAY zJj7(JaK}pp0MFv~APlFul)Xa?E=6#Fb)UazLTqxw5Z}Kc)^gjr_=!5H*6YK`lzQ#8 z%UIl#{=>NTnDmyt54|oz7`nvYw&eZmA7LLxB0lpOf1X26TwF8Wr-)6)-+&Npka(q^ z{Z!;KvPhHwtYSu4E=r4#RzeV~T&m#8yh5N3M^s^ri6?sz_)w`*c8~)PLr3-^4jhwO zDn41gc8xYcIx98fd1~*H(%SgR2LAnY@BX^!3W4|dK|<{%kkIi~6x6U4P?I;fQio?T z*U#fL#9d59i$}!WjUIMLz7!jaDCvhe_6VpqJwVsM*^wiX+b+p@kpIdWI=RxcAUe$cWy@I2-Z zU7!Proy>S@O3g=oeK8AD-`sofv#jXQEkcnim4Diqr!0llKgc$kB$DlZkj3>BSh2Hw z=(1fqw_AP>>gkA^cEZd`<*|I~5?L_D2Y$EkY`xQxJRv)>hxjD#@}RWhL14t4ohFZB zd?4t^*t2<}kp8p$v?||f(51mAhg?Ku0U+}``8yY$0GJP>Cr+{8^F&vAz}BKfx9#!n zlgk=DE3lvg1mI?asVGa`P0z)Ql}03z&Ith5wkrqNI{qdgF)sTz4w?38z!@*&;{ev> zj3He_9z6MspNth?LsEz+aK*N(-0wyy2HKvn-m&;zxcyFIgEkVkZwx_*&18T!AP4B{ z1U<;1ir>88`xwy7`TM=ZA7qW{Ki7TMPyFBRFKtA9W%pR#F559oC{f+gQmQYBru6D* zTaIJ>flrR#gFao-uV;bK;(Z3-RrkBbWTLr6`Piq({lA5672fRQt%eDYq`A~yx>3&{ zbjT@l|0~-Vjg(MNjZgzgk;vg;JaqdRX3@c2?ylX?d()e^n+p$^U&eiEdzL&Rw$Hq` zQV;8cQKbnc-UDfSNS?s?EJ6@*hYiWLD#k?j1&_YNTo}ZEpBf`q>#RC@eWRt%%tyqt z9#%9UDFJt!xxP3IEPg8L^;CF{!qg*tnxPu?Z1LE4;jGCJX!T4Mou1dMbKd;BgoZ8j9@5q*%Gn@21nRWI|&D3pDRX)w-Mc+&Mzz=*9dXE6~74H;Te| z_sk0-63TaU)~l_M4wJb7a+9&cp44|RC@$<_$#NJ4j6e0Mkb}UGPZQ@A#pOgF+!dbi zio92q?_J9RewXpI-`OZH{(zWEA7G+CsKWHuLa#nhDgct$sM0c#r-l{mmdt#i+u0JX zIR>~kriBH775#HO^Q=>U&9LAVJpeafz(ix1Ggp=#3i*Ltm)x9#nyMoCAUCHSkdrIh zeV!y)8TEB^#)Ub)qt=WF+T7>KE`(MB69d{Ji zjYWQ=4!S1jUQyEeL|GFBgXtnq$m-Pd$}NIhvbJg`2O>poG+#-a)nm`cdma%SM=ezv zh@?3)ICz!Cv~q^1gi2yGEq`|N(S1<4PN(?rI9YCJ@yCli3n`{qqhrN!XKwezkTVUR zY?(8NiF@IthA^>rXL!E!-1RJPHj6gdgf9oQatD>ByN)OlAa$|UZCgur3b2$di8Xs- ztIJhBtZ7=)E7W=2;r)8zA;h{D-M_@*k?CPtzy++}N6FA{5|H2Sn17A`vVLj>rs2Q3 zbj+>ee0bmYZ{^`85*3-l8=a1l`i@w9(~z`m`y*ejNhATaA)mG3CrECla$SF6wqt~x3hj&2v zFmDY}`8FMUxi+xy+L)|%cSpjb5LX(|Y-+yo9%s0LD_w(I>xkYOeayrZ=Pi!iUR)*0 zZJ%YW$&$|1 ztB(2}cW8n2-PZb3FP=lb$DGC`2?E6Pd*~7Qy7F~t6FV*CK2vKup*K$pSlB`l{~Kyp z{=9|4Z2wIbPTIRM{0GIds$jfDk8z2yf#D3{K0oHVf;W1;^DYkhr4(_%jQUS2rw@v2 z(ixsup)?y0PM7Njk6Q8ScwbhzYi}t?`7{`YMxz+jqpTE|Zg_b}2UbQ&aDJ1#zC7|LRZq zd*usw{?BFV{l(|e!d6y#YkOK2o0Wwhxzy?qa&nQp92#SAu!rhRO7(It&7MO`r<3=N z44NBQ0AT5oCNSJk*AcVShHq`VDOm@Rr1l?VC!{m&CMzP&WV`As6rQ(TD7Pshd;9Hs z!X%=!+GGK%Mx-4=tnDojy=aevnML)HLur+pHvQaj8uFo$lln<%O-|I1(a0G#xxUUu^1&g>@Fh4QdqxjD0Mn1%Fedt>4=k@NN;>41Azht&?vzbWU)fZw z)ct%{prcqoy+iQ+4>DVPA<4t&2ic#9SSVhcEK!X7+QUshfJ6VvKhR$v?6)yLao&uT zznnf1_-j%Eh(FP8z$J%R#N9_I!u(z0i4yDqP~Ecw3?u8GgdOqu2^Gho|I7-dYJU)4 zf}r2bUJDMQP8GNhfWrSxT`Lu(J(t}-`(p>(s7m{%XIBHk%$W;hp_K4b=( z-0s@WC2V$52a^YG7TI(RV0#3vrZYaOI)3s&f^IrMMLaBxBS zTm-M&zk#=$5mawfvKG!s#&x>wB_^WF>R9=Uch#TJ-!N|;tDYIC;>P)~iNh_IA3)AqFdrsq3+?y$@dGSz=7i%FSSbNHA3)6r_igmote!b(J z=ew3n3`+v#VW?unVhpB%lp*~JG=a-Hg3ZF<5Y#!dT+nuS2}n8srCd^0&@Ieffy?vPqD($BC-x~$6H`q;QJ+@fqP_+-!HV$?Lq@)K5k;5-P@s zGEftV<5uhYxW1pda)5mQ3Na^vn1;Ayjr~7B&EGUd)|{wEz+m|11(Zrb2wm2V7fvR& zw2w)ec5R9rkg7Kh>WG7yBS-i}6#TKYIA7N~s}$6UUZ2 zclD>nf8NKJW|=+6u?8W@8|)cyZJ{bFAi;ow+{vJY>OjrSWwUMZCBB`x?jq8XgNtG( zKFg)EGset?;E2Y;--EFEBm;y{+B|2xt|bBiC6asRw4HYzN!ZMIxo;ceWlIrdHTC+z z*W2yZ@%KNNzb!u_ z7v?#bHt5Z=&NDJ;vE)s~(UjG?7U6e&b;culiKJX8`+PZoI=fJEEsGdNg8U$J2Nk@q zO%{9nHgXQ8H7g6|edziAQosFcE%(o#{}c7xf5XcCiv@h7a|^P|ouIbw04xLW1Ls%Tni(gFN|vI@00r)}r$S)j0VbR)1nSP0COQcl&>v)UhL`9d+8N*Yo!r5& zwe8iJ{oUjrWTE+EG=^*T!MsKrYE`R~FVoj$)S{u|3fu@}O0!}SUGKtujKOSq?B3%% z%E@jF}&3>H@e9v(s7gZ6#cB_+VZj zXDnAwC9J?Tdq|gmp{>pH=~lu6*}a79t}F&;-wy`%jRN?s9>l^HA~LUsD6{%TgK^NX zfs;<4nePPL%(PLt)#}{fqv=36%$TwqO<0CgM?uLbt$CIUz>sj)+l@? zT(>F9Ka~F_d`aO7npJ^QBmk%E^>CQ7t~Rb=Tb*%)hbh*jS%T)hPfXOPqRIsIE^P^B z4-1Vis9aw8;5A5w z)p(jhM@rU$hfeTQDeVUJZj&?P<y>7lvB}k)YRdHHgX-bsaILG-NXH0>}0f z=ZkEXT`<7nm)*?I&Y{T*B0ZJq5@R5*$Vf2?ZB5cV?1HiL4Gm=j<$U%fpN73hmJA|X zo@?%NKU^fw2L@yPYQ+ySog87zR{elAVv|;+(=EDDHD`QIbdM61b*VX-H{$5F(OOe4lxT&f>-Vc2+H@D=^UxxB`h3)CaPyj!n9V>h?kfh7{|aEZop=_`F+P*;1~G<*uZQfItK`+lzi#}U2?X2!l} zhgg#~JtbW=;d|jX=@)r zTZ)b`A|%_G`*syS=U}EEv(Oh4$$qk9pFV9CY_1|hEm`rRibzuJ7CZ#;DiE31LsB1I zIqh>(J=$U7lhj=yfGa+LsdUmckjk}IQxVmdAOtB87hZh^k9;{O&d~5NX#a*v$O!;- z1?L0>a)cq$ZJ^Y`Sc$rHg&E+(y;R@wKzv~kPo^}U+hc{jNoWZXzEN3Q3ol#er#pM- zwoIB64Uvfmzo*8AR}esBy&56XKK_Q|c zbbHGFoR8c!xOr;v*$-GWGu~#ysLQ-`fG;b8iS={$xe2H8E>GsAi};X<)wwE_(#i8XSD2*7pXHauWEqB+ksx^gY=@mf z+pY`5cQ1RQYo5EXUtSvs9p~@ie`Uk2buQL>faXH8qRf|*1tGP<_T-0RFw?mrv}MZ6mmC>= zJ_1(5v2m`}*6DDk_z5@vTc__GpW={IUNI<*Ys4^G!&Ea0i5!mU^sDJAd>wCjO`iE) zSB_@OGc1v%pzYgCK08+Ekn5Hy0tr@z^KGJ!4}UhZKZWRU5EXhVJvjzr@>u8@9!CCZ+3&BHBk3mPqYU4jld0nu+EI0EHFpFcgCtu9 zR|>MOgL;NH&mbu=72iUlr=03Dg516?m*2$~UU|$fM0Zqf`N2u6-O0!ev%PY;#IZy3 zNNmj=fvqH-Rc=Jh699ADJaAp>|BCQzA{~b=e_0|XWnT`!eITK_9};;vNcUT{HS5u5 z%w`CBw)^N34wAAJx&xWl3mD#&$hBj+j8Cjr1J&K2DZykvt@IU@&a-pkfq4bio%UHJ zsB;bXs(7bxSNon_6G})Xok6gwkjk6Jbyo<4K)_deJl6^JMZ-t-Kdi}Kkm@hAy!A6G z1ev-gYyzx?W%LBJOMl;ZJ3WgOKPRdPMk5Uh;yM|IjPQ?}sNNP(V=iF$<0Rpjsw;F~ z1%f1b#CE+*v6Od{C*GGt;as~wPQie2NHt4ZUAp^rB_w1DA*h&)V5`20Y}JPb*d>09 zpO1x{Em<%WJG)t6dBG!56+{o&O$?r9HQMs`wzyV4k9GCY9C=L9BlDcK(6GmEWFS7n ziz{$?$oxfJ+1YbK)8h-#lK>sRXn?@=-@1#p-7aJrv+jAz-KS`ee(X|wFD+(nTFHGT zhnUOZ^GtkKTF9zxhu6#=mG%hwYPteLN7U=xDx}0%TU+prW2N$V#^omN$xd>xZf>EX zApYXi$LFzn`aLZAR^-!>$3YAr4>mYq@4xiwNQOvD;jo?gD@a`;Bttt*@)PlCe~|YH z??)e@pA({}!$?r606Yo8%eFuNHV`{gFT)lBFLx&ZDKd>j(Sv@S!$D zJL{WA5~B}?mqwzq^>>Ryfuab@(L&O5c}Ef*-hWN3)ZMvVnn@%i@=Hz4SC&S_o4aq$ zXj%3b8LJMMHy)AiLIu9ax;1lALe}=&QN2r2hg$BIUH3`v_cltHC6~xFa_iJ8*+{Ke z9p7{bwQ6)M3Vm?X`KIG_D#ha{cWGdECXs+ujQ&Gke(_@KEp2BZ*bHhpR|I+vz3F7) zJoU~StrRMui{A^Ie0R>ua9I+o;b9{s(mA0po40^UG zAVAp^$S)x9c~F)UJry@ov*eUP(frR~`F|BL|Nb*|+;1~;AqNK@`AYl$oq5Xn0^lV6 z)&OQQ@4uR6rhOTWeW7|-U&XS#s;0J@mv3U zlFmhze1{CiBPu(v+*W3N=Q z?iAfMlsQ@)$l^ZvkZ}Ka%%8W+jsS@cty;kF5BnLQxBOcHdeDPF$wZ#+*kq5s?x-3q zpKxSzN0ETed9y2fo9|98bUZGUw4PWHecvm|w4cVJCe<=*)|cu}$=N}0VZSG8DmGn- zI03H=3N8`!&Y6v4mTF08-PR8uc`q2a8>0rj6VxEh*)LL~qFshw-5Q9adsS;dqgutl z6U!3Eee*l}$ps}UZEt-XW%or)L&Bw~r1jC0j6N6E_O!j5!`G5PZUEfFHunda1Dlo) zoPK4XD%(pjR~gq@LchAzW8K{_&7bS)e(xK()viO55@9SaBeh;q!#D_WP5_GA(dn_1 z%ILFyjlhVF>}lu|x!2OEYe5W`Vg$N_>*a_9t7ip+fJkvDFL{jOcSx|0Zg{mQiN=qE1iQ`jF4ldJfiFV_Dnhbwah0{GVPu*aYXl)&3tM$#OKndKz%3f zYQr7a;S-(yu?z`pZztC1OOL3qV`F>LA9_o}Fv#!QZ~Q5-2-?q5EcY3K=H|&=4I+lZ z?DriYUkApldOmi~1bxQ@8Sg(`E)?B&t|b3KW)4En^)FXt#iG7U-b zG3!BbLK^6kdS=!}kONgxPEF#R?ALlFd9vqv3=NG1;!l+*tqB-ah>>WzG>H$eV;rBJ z=!&$5We8ke=g?Q`os!<=2$hrUCtdK8+>or{nIO7S4vnaW(2}?rNv{Uh-mI7Jkv9-- z(r&IhTid<%VR-~8@GPa(rKT{{ZxmI8e3QN$P7OCae(D;6UQdZQw?tolZRQg*DP?YZ z$?<_PVVh37<_n;FxGoGdr=ZO8;_rTJT=;tKR5`Z!!Q~wMb>rZbW1o5+7KKYrV=B%( zTcNDn*=#z zPXJz%=e>2->MnH2s#A13kq!^85#6jX7j%aPY0oF%UPgG>*LK8xP3QApl9{V01u`lRn{+~flLg}0>6%YEJ& zkMdpK9L|;D7%m~2<5;MV&YhiyJ!cS-+c+UNk{jE4bcZ8?d%6gF%9f@<2h8Abk}5e9 z63;ET||88nj(!~Oc;b?;1Pzi(HL=2-iVB)v9@IbA7T z;5=e0eXf~?2%5pv35e0q2q0R5B7aF4cn*itBV(6DDQ zC?g^;(saYg@Gb44P@AMq%KSp1)m2w)-tO0^o7yUF9D@7fuW#+mWVl4S@ui2{9lyH9 z`NpN@EXYr!D2>KJcbOzbLB~9^A7n#3=%D>O-rSZTkdDOr*NZ#*2zD1tRrK>~=)IrD zX}s;YF3v{up&JC$s#*nZk6v1f=)P;@6cFs!1Bei?GmkFB0(uL!G#|BdzuTYsndDV_ zBu@iOE)uh-9=(e#dDy+#_Sl{I>6@Rf?i21GWEXU--aR5C`^3opEH*SJXH3;J!*Fb5 zM5?c=9!x`g+jEVH;@Zj9xgKA>LB1U8%6Ye3TOGyF#*#9J;SWYib!0j&pD9ju7!XWr zj8-#Cy$fg0+Av2TrK24e>k7EWDr9bwPd|{{ntR}N39G8vNS(uX^_U|Svn?6UC5OKLtN`iVcxukK;VL@l zWMDbZ7F?hrKsD=OC*6=XxBYF_r0h0Hu4MMz;a_9imGhH69H2ri7qL)@fMqW~-xe=` zkYs@qQ)OTm1CXm*>OXsX0~LMA1qLNP!-0liAIEq|cLIS8YGeuWI`4EOEh^f|(>|aD z#k^#++Mpn>3PEKhnou+9*m^A2{%P>xMoU&RKg>W{x1=n*(9H&8SQctlmc+rG?n_Q) zi8{%I8}Q{=9f)yAAI7ve-Y?k;@BgwWw#yk4@3iua$EE^v9RH=Ed04A5BKg=?L3zRF z1@+mFWXL6HT7=uln}-p>UW$3Mv1L2zqt>=%mN8l!tdZ;Aqeni{9_j&oKmUl+HvBC+ zsZjcF#eMJmzhXT(7;*O&h#1pb-yd3Vnv>>yHm+xTN%wZ}+PNcnJYEGOVgBP)u+~8@ zp>%xI6-~|34d&yR(>WZ9oT~Gv^lQY~BY`#1TBIQam0K>K^fG)*+_JP^`1|(8G306W zRmXR4Ag{g;Xh&O{Tdxk*dw6u_w_eWHzGZIyWZb;_WUT7q!j!^IuA>y2wj6cC{2{`F z^aJZ{PlbK7;utim*RLa7ouXv6Ssz@+J4_%KzvW*nPGU59-FZab{hFIW*tzg3ZozM? zU@G_9vYcGJ#b9%6nf235&a*J#GpZhT;VChp9|AdX5f0Iz%Y4-W>3I7GJt#hrwgvYY zK9(j2w+%{}MXr~NGA2949}8Fnf=DdAGnKI{45qI3+ny=B>q=Y(GyIHi+ezKiZ`0jq zWV!{SwtL21lQ;!G4KirlJ%6j;l&bN{-hJ}dZkPAA=e$HR;3`sqejJ_KHv$b#jRZsn z>z$a=K{4QiXL<_mRt|=gLeGpzN9K&RT6^-9rf=_k8RmO=*Y=`RLo)V~UkAK5xt^Ie z;V6tn7tvVDbhJou`s0;}^5fKRKTRe%JrGeN!+YwWW7o066PMl``?$_O6=FxmcvvR? zh{J*mtINx{r0Z*xNpV{^rJFjxg#~vHLrCEF8xcWFA_c3{?VuUg8MiMz z5*4T@HaTIv%P*^NUm#aaTRAe-Bz11(x(UejM}1YbT9t}&cj+0vyDF*NYh`wRX?spv znEKQ#iBaWw51(`gzk%dX+IcoR%%DT?%lbbY zpmk&jNf_N#Z9LItD?L7M#dn!txD(0r@XLed4vF1IpsO!=sYkAOD|6*FvxgM##g2M( zKVkbzdXIr+0~EgnysN?>l3bNIPMQh0w*@o%UeI+~Xh2k_%mD)h-lg_80RJng+3$Y7 zdbR-n_xHs~>+`Z~HPyDfC=c%JR_M!|g&=$pm`?qjhakm!aW}Fg&avV==h+gG@OS?BM4H=ogAx-(dck8b3K=8$g2ZZ7#UsIF0 z^_S@$IjE#R#x?w$Yk0E|Z25ER1bT0Qi_~gJgwAWu8IXhY*GaQg%>L8pA7o}I5@hS@ zW`^|hElI)w0K3sGnWR`iPOADr_5d}W<8{Rf6kYylTB1#5#lw6S>xPFj_gvNMqWy>FF=n}I z6`M@WYMEAtwqXING}dC+y{LRmNh+=VK?UwF-gR^q)gQi(#}&*zvDx-`mE&Z8X|hq3 ziOaUpk#UM#^5r*ea<&L~->l4EMC-}VK6;VK4Zur+lG=>?Z9#OVd!MKCZ+Yw*IbBh( zE2$s;0^$9wfB)|%vFZAp@7;P)s^^S8qmq%Y)YQ%QDDJD2B0m+BVl*2i3)k~K8ugq^ z`qiQ7B>FY>4BH7?j;5q5YmdXtU7AJ9ZU}PQU5au1lH$wQ?p&}&El*$eL6Is;xZLwA zeUYwG@8{4b#unEZMzJ3WhEycA0O>QmGrjy_UGp!cnyNmHL>~RBY~C!>A|_Ibqm+k2aZU=d42a3oSbOGfEzc9{0VW)(D}>452E#)K@gz=4dPJ#wvbS z=#^QsEUvdF*#uWWzq3@YP*vo~=+iCfx?OlA7+rhhbPTT4y*VVyxXl*+cvw519YS9) z`fQ!t@=S3Q+X$nSnArD0eXOcD7Qu|U=WJcdBf7fzvgFp4Xwk%E0iSD($Yx)W$3Ew0 z%XdzZ1!C4Ptw3L*BYwS);+@-_DXx$ zHee=Esy(gV%>CkU(Z%=2BHV8><}8=MOYyefM;J@yh$Eg=|Jp)AjuqehNZV~;hRwsGtZ6a}TX{`6=3|-wA-c!p^dgQ*R?IXs=)v}(# zZI1%%Zfl&Q*X{HFNd3_9WD}VOt+Z6vWY5;AIivYO!p*By<&|Y_nZbhfQDH*<^EV6F zPBm(?a|F6?_6pU}Ch5jcm?=wjMQ#Xu#8&6s?~cezbxT-LEnAa**JHgloY9<_<@|%J z=CsDmFq;`hsAvUap+#wkNl755OlV|oz!63==XWPT$c$`bUy*bt6X-+(;GT4`1BEYg z5Mj3qw#vw}8Gy0J-^j@BoE(-r6r~#Wt(EM_yXe*j3eAP7=4vG1iRMd9Wbl!m(ep>AjrIG3J`XV?kw%wQzEN* zZxKXZC?!CAK$ZoOZVa(j37s&(+mk#ope6>j0E)f_FMwuQR0|{zH+xV91}yY-Dah}O zx$OweH_h-22txkgLC*9Na&n^aJg zlYUDr0)YL0wEFT6l?S>F|L@xB-&qWuhX0^2Sa|tE@3Oh)YtwVpjw323PbiZS=dWKs zf&B=<>$)0m^fn;yf?&&$3z~JX=!IIrnm)6t9zCEU4jqeWjC#**V%y?u?bmU$odKy} za7i8QuDf{$=W_wkf&#=g&Lf8Zmh0&-+P&zqZM-%E>3J^+2zjx8aNp5>O8>7v(k*D@ zdJdh`=tV39lPt;>XzC&BSFO=ojR+zAr+K4RGg}ZAD`xGi8*iUJg$_?Vu;vKtur52! zWbjH(jD1U=L=OSPp8*+mB#LD;e#~)ixq>SQBza2ecyR0&Jah@2U)G&`VPiYVdsl2; zEE}-F7@|`Ru*Uah1W7UF=9Yy<%Rr3C_5<3TTz6hpW) z`^PeZwTicVR)tBXL72eMl*~QTnW(g|rG7g7{=gL3Wk))Fbm}J=Hv%sHjiVLT0 zi(TgvsdPl~lyg%j)db%0G8l3DVCgSOpMiJ*CUm`e+6tmry zu`8;TWpE^=0>F{@pPuX(eI`Uot+~~(`of;l`u<7N*NT?sPI8Fp)E|23P01%Nq;$s~ zt1S!vapx*AXwFN~`hO z3`5iW`TJq5JYU7r)c3YB0HH%_2eaS|)ARKcs4dZ$fUts#x4p9COn7r$y(;ELbu5*} zjE3Ln{0-Y0b)*@we%;NUR0+JTVjTqa-m5*0WZ9?x(B3XYAZ#<{b{KYwg9~iatj%9O zbIE4nIuec7k(9pZ_BbqLWev;@sVXFi>PW&L3rL5DYp!0Iz;wM(1{iIFTKdkLlZ z9*w2@EPvRTl9dWQ{KrkfCtl2TwJ^Y75%KSkz_+;3&ZNS15|{A-a_Rv2eeWfRlR8&d zn&vz>w;irw4W=A~n759GF^~u2XWb+RT#39Ag}ZYTXd+`L5T{MR!GM#KerDR%fpa7x z6{|+Y{?JH3n0~)p|IjTH=-m|rs0#Rb1_nX@XgPV@-(PNiItIr^4eL`#Hg^)M6f8*< zE4UoQoqkx>ne6-fxdj8RgNtvu*6%goU$khKYmCiK6!6-eHTh5(rQ^NAaMP-n?D5XR zEr;1JVY#9!inn?eesyTzD3l@=N({{+Rb_2uv32MD%ifJv-E^H{?rmm8O4$tTp;F8@>baX@tT~FdzK7lK+|5sI8+UPRt79ws<{NJD^ zQ`i2qmP9*8)86HZY^eZX>N-cdH(xC~Oy1_KA$S)KvHE`4ZJR3MB0JterA9YZ>Ab=aA!)%-w+?@YX(2SO3qqFTruK11 zcD3c)t-mAAC~iAtjY7`ow_T1YhdiR>_0)0^J0fNzhV_-|37tTcN@`+NEp=bW)V}Y{ zFUe`o7NyU$sCZ4L554v+?Yu*@maUglK0cI<;cib_X;J?Q7`)k>C6o)+u?!OspR;x>z9-hLFZndWDj7-gMRKZRg$0&bD$l z=btlsbV`ikrsf~tKIwtz`T)i~FM=SK!=q?h&o`Ls*yyou#&{SH=yp^>O}z3Q0B+C; zt3u*>7j!kvpLi8IA)vNL{|G4Pw?V)V0=2fA03NUqATog#67y42h>f}ZsddF&V4wt~ zqBq(=`eJX{ndk8njzuHM& z5=}ocm}u#m8`Mk3(c6RhQqq@J`FJ;=*u|L@myiL{&VM-;gVk3(^GPDr{m0+Z#Z=C< z+6PpX$H!MU+eJz|e7I9ED-WVnRG91Lw6PI0-oM5&QC~h{f{oV}(m}l!xWPD#!d!*^ zO~Fk`jMm!O)xz-7EQ{_*{r>W&`r$@A)-*P@Q;e2L?|KmR-Nw69?+_D;ztC{a6~J9M;@mV|zZo`0B!v^1*j?FHPOWEU(Ns={EDhFCc!Q`jvyk1n=~` z20E(;tiaf085qadI(L8J*p1qG?nJ3$Z-_+b>JhN6PBPz0oxAV_Zl0!r__h|)oj-lUgE@4Y3| z2uZxlU)$o@=j?O#yZ65L{@?>slbMyZ#vF6ZQNE)QPxJGYsUw;UQ}(%Afn$LrC0<$y zdt(g~Wi1U+Ir&2xkw`!DbMWllhdQ=mH( zz;*D$9%91z%4%yBpp*qr(iVD5FxlBIMK^b3nu`~K_1q`k$vT^-WkIBE6EW|-O;{GD z>-N;qq7y;|-8xud64z9h;fkCm-YTyJPiEZqA(RuH7QlPP4eCuqA{9?`s9t+BQ8gCV zaa!VcZc!^+ZJVD8e;4;w_4y;v8?L}%lM?Ia*x>i>wpbBT|B|bj&uQHf-URpNi8G5-s7W1|Hz}} zyu0ommx4(^2gk= z*x&dZMc--lqHQ#ZW2k3q(6vjw(yfjM?O2&`W3$3(RXFp9X-iuvJ0&UgEhEkwi*2nA zF;b?r7Y=S=xl|bhga2b`Sj=aaKD^?OLpo9ya;UNc#N5?WR2){JK+{$zpbAEPe{UI= zkF?cS3i>&U9oOU^%Ezcze&i$Etr<%nTA`jB&bOt)i0~OU>e`{NOt?8N#1&kw@l`IH6hwEpF@Tr4`1F)rfU{%h zq&W)mh40_I45kviC!|jN*QvZ+X!{_GhMd_Awu{IKjAPZ=r=!I}EX15pq9YkPxIJ$PF3*}M3 z790%pUoyQ)kp#UAeHw0jdM}MQy}F^NlyUAas=6-u59JK+1TWM5KbR-{k7<}9@lXEk zjUxAVV`1)p+m$@+^B2xW^8tlTWds%~vBxH_U7g}Ho$srvUNL#abf&T-@A9hfNHB&1r$jWn@cx2<)S`1)b)n#hU%9V%p3rGO&lrWv^t#N#Os#f!$Kp_WDd62lo5>aWUg?GlBkA6^u{Y*m zU?OuW^YIsz)Dw!EbHxn0yY1{TAg`F}KT;n4DpmWxRH8-@#6p2S7u!&1sRs$#Dq5^?v!n#d+DT z2{y&o_p}zih`+v9%?y=d;i?9vag{pnnG2I#KSo#A#~x2_E5SV9r#IQN-aGP0-HVR0 zoI?6!;YUYygg%*i3h-!MWDh1>G6OrAScNh@SCp~bFi)ir=xbtZ!LtoHCLwgo>O6(@ zVIeT^wZsowH@?=Ig*Oujj|~9XEo;djG~Y4! z(|db8a1o9>?74yussOZzPq!_ew>FrI@s07rsn9w6SO#K1WEg?c1;UE01OMS=eEIUV zq=}FTA=~4d;}LjLo(4=!I5YH4l^O~wQ6DKPu&pNXH8k&PM8oh!r8xuBYFCFRY18^+ zu=3XVHibICK;pR}b@lK{VCSX)ybn)`upV30jyEl(;?HXNFg_VD$%b2CE7%IUnCvRK zdYmcMwA_B`vt&+PAJR?%{3yVQP7G0DKC`b@4#v{L5a<^>6yrf2whnoEE;b8_AMm~R zNa9K6_t^GGpHqdqNW&{~bEvwEi-^p*xLGffp{R`~0K zBi}()ypajJ#KNtZg*G>lG~4VV->IoLwOTT2Z(68)2P$S$9>de5?;c7UcH^ZMjp6+xo_8e@Ac<~#3+JeExyolBbv!oy6aH#ab5YP3d03; zvd_jRxv#v+fs#ChpK}fNY$<0jejcJvi%KwN_g^VAMv2DC3eE^z#c^7c(o(qfv;0jEBIrrV%NC^_AhPP zHRuQ*D`h$_ntLQ$+_aVmU9L-yFmZE|_o9}Y8>yzoeF(Ql`%^=yiR5`vZu?@#$Y;bs zQmokvn=W z3>@~*MO%=*=NNfH|>b(Owzy{R8n3DPg+z(V=ei6wFHDip4+ zBKgYTR*qb>yakQ@H9RfJ8E}$oXan_)HlwzS>}4$U)p(l&HK3VVbO3rZl`>OLtqyT4 zD4Zo>m-C(ldCKiyx4kg$$>Y|S-5|4R7>xc{*=as-eDca)yP4 zouIz#N4Z6!pqTr}$CPZ^5lN;TyCxN=q>{7%rPoxLZtELthS!EjpOmYb8jn*7P+?r1 z1_rp4b)2P{@z=1D^1gGc=l69A`*x!;qUV*Gzs5PZ=%UxTdHQzU7N3k1vgz~2FmEGP z3Pq8+SjNW?m4OPR>Bq)0ROm%x%odiZ1#c}@j3@w9Q=F{~2ZnRGSF<0>X4b&e7PV&g zL9}#y3~cD-pM~mcu_fOaxCfqF%+^s{)9dQJH6Lbo^e%iV1K7ip$X)>Pi4B$u8L*2{@Mc})V8WROM4 z@~xxZ>mOXM8Dx;>YYvn35g-||t_GxK&1)bFF(5h&LBW0~RbTVEJr7pF-$4p6_2V+Bc839(Q*#ZR z7Tdb9gOp2Z_yT>0gZg>|j_zWv4yJ7-J4+~I08$C~xSzFxwF2JiA5xytRkpA#>U#0+ zMXpV`QZlK9@!e}=4|$#Pu0D$?C@M)vkV9*>Pcv4tpHv*b{0N%nfb!Ixqn zPKozdc%h_V!unOpywO8x?nBF&<<8FUAa8BUn&cF+Tgx*rc5o|a#D)=fhpSy+=^Y z^$zehq@bobh9Ln&vY`21z#?ZarpJVj4sV4R^e?cM*T3}zCjssxiDU>Z+@&YB+5?|r z@&Pzi1*&vq32WF*ZVt^yH0RyFj>>8Yz=iSB;?=aDqAOn&D;-0cr=D5?-X&=Y5qx=5 zPGFtqixzq3c~@!5LzB1IfgTmeoW*@F=++AH$&95pzC2|lZJAD3XeY4Z{QRvK2C8%Y z@sC%YPFRF6_fDGFR9fx%rhGE^!ss9LyznW;(Q5(*f0+Y&1}moQYh1K5d}uod29J|= zcN99--b@hG?>@IZbc^9efVM&^g|3s?5Psr0CF}j?fU%Z6CaX=o`); zs9MRE7;DrX3!6r35`4^N(c@iL7)$b7_7wfiV~O?AC=kGb!TsKh-62!$|6djE%xO&`=7 zgfi>Q=wZCnXh)~p=EXCUIi67T4qOJaGT}xy^gRPMPuMuL_B4jJ&qLf)UEh|xpWvkI zE$&2dDV9M6cC8D}9 z#<+M>9M6BDWW(TmZBX)1w#!7*Ym3|um;^;d(jz_mRRqsS+gF#q=Gz{38|tDBhhoQy zR0uTWue~O$ zoV@yJcsWudW`*-Sd-A(>Y0GO~KGHJKpy~4wJ1$W`f?Ul~!|P&LJLHnol{}2#qxWT_ zPpYfpdGe@%|ATy?5YT2OcFeX^;dtH$r~~F~M5uhSorRpAkl9BaCJrLJbZ8Z^6E~~t3s@GU8u8p7;#RF-AhRa( ziCs1P4q>KKlBg)gqDl}wi5=M|{}fY{ZdG34e{L=C z^;H}fg*2Ti$|fw=VAV~^+R4~LB=;sPPn@%Gh1lZB-MLm>wuhkolgK9puJhm zX_o_g{(07S4|Ru-YMs;Q)TSMcf}gl`kw0zL4~)LCIg~l)OK{shI*m1rj3&aMCEWj$ zdQ?;M7YgDnMs3_BPkA5#pkfAV05%_NSWQ?5V|Xe6^20_wP&D^s>SZIwtzUIvAa~rw za7C$1P2e7zYtekj7YA;u43c&!F2~~CU@n(CckENeY#9+9eRBZN9mo!OR&ocJB*GjY zv+_?<_HGTYD*yi<`N7&UKH=itd?XRCmyi?KRsec^^U(PbfgLf^`z>8C06;AG8%L=n zbCuL#Het@b@;AS9qGB_HF@+-lFCr{Vj-#Z&+Yygwk;r{BkwWodMA-)QhULYZA(l?5Us& z`|@t_B!f5%bJT7V!~V7I6czZF4|2DJ^Z;~USE{~zk0;goO3&stj9##%j06_K6Zpck z)%G?-0#hb~GN2<5z@wF3t*1c7r7k}sqyZ!?-nNAY*0IPd#f+BQ!^@OJ8JP{Qu;RbX zL=2`50F5is_gVxkoGaa=lbT>WT^-{lbdqmMjq18z+jCW%(EY@;9zLf@S+mK7TC8@A3mLGRiWXwA=7Z8$aQ0j8B*^Eyr8;JnAK{ys8 zT4yMwdDrL^0?C%GFIdLdblfuK*_!q$ejZmNB`8YM*V9mG`Z!{<}OCM~=}4KCV^Sd2ls$m11wMDbM-%W#`A z3w?9c!0hVw&H+7l*(a^T>8`ackBKn%Zr(1U`!gJ~4jU>n42VoCb@@iw8GX+Mqq1j`XCfi?S2F0 zRojxlEPROeb(HWMh{Ie@?lC6V7)Y=H4v44Uo3Uy5GjsfKSQ3ACyfQKfF?*6Z!fj)P zXVER=SK{nEKN>=8I#C|rl*x zd;SFIXy4<)K9?ymOh;}vZ+o$g+SK5^fgkx$bc__T;uP;LdvNJi3}qvjq6%U@>==@A zyd|WTf(_75%4sSg6OFhz#B#;ILe=*6Q?(p%=3z>`BZpt28*eUuCG(wwyh?4GfCKxQ z9Yg!B-6JXGxfDu_+YFS}jqKR3b?a++_$^_Jfjb=A%1?X7fF4bj3trcGPk1grfgRRM z#N#41`KTM-EvXK7U0AUB4jN4OFcLOB&a{GM&GkBJ4ikYMuN49bFJe71(nR$5ou(N| z@KN!C?FzZ*X@NMkU{ikRC}YcMg^o>=5S&pnm?Gz-lAjWnalckg-CAOK91L1Tm6b|8 ziTXw~c!%LT=n-ecztT+7-$7T)5!Vd_00E+9REUYG^exLZ948Fy7D+9t%YRvWq0Wq9 z-C`FGx!j4Hj-JZ|Zz;1PHk%4gDPf@5%+DNurHTyf%xk^MNb*)>8v}k}HFoBp6R;Ng zgYwfUL;6~dTVU>iUcD)KfujNc>H;A>nBWoYE7Z1ckxQ_(%7iu=8pK^(c=KMSDr~i) zVo{s$a-fsV+oC><=Ak;G9U0`I;UvvIq7%`70nxl~zJiR}ay*k4l8Z+kBK?A7RIC~U z&$UfzTR$p1P+UJ8M)%Qp2%$!@tJ7s4*Ho##V^Uq2R$20%m~m!?s%&%PsW8+eZ*l@2 zvxD2Y!?0gRjDXwvOVNS~{0WPEv5R1{h@rrM_0~R*^phwn3Hr;vq2Z|q#m8V$jMohA zzF)Rd#I~Ly+Bp6zv>T4pd|wg3nL5 zYEyCN)oOvZ9+UKKf7&P*@hI60nSr}FgdlsE?tik3mVcQQJ))BX>)NoYmJ%&y?y9U6@O?$FTmu?}4hU@P<|Tuhr8{hIDe zO+l}b1hKNd>O6V3g+%Ai{30U1Joi6A`a#8x(BL7F>!tlX(Kvro+b2S^RRZS)3USZz^0s6R|;bpQ_4^_nzZF}z9H zYg%~LA#P#5fbmCQTMKH*>}ptr*1fTB6z>*qtq6lpU!BJOTVXK|cxsZX@Kl>}Yg%=D zPX@IsU+r^0dC#e@N3ASBfvTgepO~4{TJv0e>3Kbr1IP95r;3qN306hMHNx)9TEJ#{ zwdy-)s#sJ7NX<+Gq9RD(l-VCu%k!U1^Mz%FG;?W!t4J={1}ZRsFuxPvzk+ zo1MGmAIj{{?jis9QV;-Gt^If3kLr|vu?7D@)7Q;#Nm2iVW*8`!rs8IUE7pYH+aYZ9 zsmUX>)(5}UerV`0$5HYGJ;2ztLGVDr%Kwr(PXWOBZLejy$Fx0xoc_d^r>v^831@-3 zPb7+lK^DzW015-hCH#O)oj|mgAyoEu{@D@W&7)PqvWE($KPOB5`uG2> zW$#$MCt%po4>(*22Ew#|0f84<_6+PZajFq~<9|RMset73p9s=_4~GzyQb1JjZ*=eg zuKH(CWOWh1i#5nQGX4Wa4EY0P{RfKp&w=gfPlo(59{{+#_&2c@h^_vPg8la&B(ZxT*Eqcr#?8F6>|xm zIQ$Gaw!jFuC-gBG>RUQHlY0s6EBF}P+WZ{yh5h;DR&BDRRvFE=gNNTP@v54;UAAku z*sw664Br2v){!qXn=bjrz>5}f{?cSwNqc^5#EM}skQekHD;%{ItZJ<@SVsI7#D2qr zs}E=N4SN1?0IN+x6Q9Sh+HI1ResL;#&0rh zRFjy=Dd#)L254gN6Ifa#Jw>wYspT7_{H`%?bE2md_;_`%6QjOrgS z>C=z*G(Q;pGg$o3f#~T+2mWQcKiSj|kdDmuo{aN6mPXV{*%brNXU4RQNvoVMQA5rc zb{Yy7Tk4H$huK*dOdC$D?ON+D0YP=sN8_~tlA-NjeeDlO!z=N9f8e4!PMOKYHOr3W zrkIIe7?^9hnefMu+Q=+vnGQE88Dbt8A}&AU2? zXIneE@6`G{n@DCw40kUtr;549u7Qe4;aPe&Q(UK8e0&15^bS++psPAcOHs3J3WwwR z094j;xoYWgqbIx1P3py$zYp##4vlhE539ZWX2HTZ@h}}hTB`bG<=nFzSY3Vu5m<2ZG-{IzeFlPT6C7f0r5x>lMiZT1Q{`muF|M&1;sG~nA zvu1Ztcn#^c2A(f*Te;=K2ur#Obkn0i4I4tM<|}r+@bY5zxxKQX4Wy{3hAM(**mPOk zim|`tE(eX3sd1OLL-N6dcE){d@bZ0R|9qVpkln-Q-;NWG{=2TWL}E0HF`@-Vs?)pl zA;jTC&;eV4%uwWKjIhkxXWQ$$1X#KUsenC#STCQ{CSKSb)k{vv&4sKE6(qWufLFPy zgYU+(#fp%Hr^N(3cVmPDRB6vgcVC1IND4GPj&^+*^#@d1Gt)Bph8k@nH@GFtqyimc z{~|%^?P1s_)4t!d^VbWmtkD$+`HO{86F(O$vlbWIG1SOhpT1r>Q269S^{3cDNtZze zO8*;@{TvtU*;iOE_XI6}SW-TYV=*c-JgqJMXG>H+7T}EY{>}LQ_t)qjROr9=us?9; ze-8nO&uJ;kIKA-yWnm)}ZUXg*zR~9itq~D@AAr6#EwIorUcAPijkB zl}?LcO6IgB9bTj#Qem?lnI7d-5wtJPqouv8f6*ZG@$Jk+<3z1DZ@!HTd&ybS4tyN` zvy8a}SN8)gU4;mAKKmqVO~roi{*7nMZ|rc;CBrw)0`AP&|EY0ma7)CZG~y5*@DksS zR`Y#PQ(*#r*oW`_kwUZ$PeQMZA?Aoi@Z(2R6aNj&be1c9)1u7^lWE@O!(VR1AIRvx zZp*(|z$Lr@CHiK?U##%ImG%EjF#dPx?*A=v(T%6Bs;NOw$Q1FG93BiY$;HJ!kW1E% zPYC3%HCeGW99bP%uBZQ06W8vvw?RjTfXBDN;O2`1(*yKGO2*ZhR*_{uBbxpEoir~6 zcaPAilC*!09FasnO0huqMp)LpH5L%90Q~)Dz9G{;^Pi=(9L$kB50uyce*oJC5Fq{o z%I%MTV`>i*vdg#NP4)O}0+!eHx`3ve;PH?4I1wg&@}3B2`Wh%fwGJeFGj5xNnaG6- zkh}H;d$4m(hD_WvUy`pj6!an2e`Zb3Ocz|{9_HK}C*henApd;pQrm8x$vHM{1_EMY z8wg|Hu7@LFM*fY?E1IH^+6v7+EwAS({Nm=0qxUFF#^aUD*Q&%H{2PT z`=Mq9+RPo;6lTOvq*STX!%`p*un#D?T)o+Fv<4K1#6#x90T2Qq5|4ubO?4k0=mA7i zu5U6DfYtF2=wCwGI-Uznn{bGe*lYd(`}X1PmClrWehsVvI&xD%4=H7wIkoB>TjNy9 zBC%FtwP=_Hr_K-s_T)Me?N~zQ+mr)L$q2SX(d2HKx6YT<9Z6S}m=t zAZg{xA8Kf#liXo%H~N`A{EtB)qR9bZIGiuR9V3ALotXO`qJWc67=RK&fd+X)$3>7O z(wJYqxl>pM++yD1AjxzAY6=%F50@01QY47&oP-vscsS0z7sO>!4l)4EhP&!9k)|}+ zaX>@7oD8wAzQ;fZ@7%csx)(kx8U^d4HM}V|mL|3o=Co@oA4_k9pu)07%Zs5!(|1VSviV~k5&|EcQ9j_#V5QAG-?;l_vrpYHGC)Cr9icQ zxc8l7hH%l+!W2h!u=++7nQwXNc%kbrPxf5)$N3_!Yc~_%8EgHb*s|(sQFTfTq00lr ztfMiWimQ9tvELeY5x>N{Zcn8k%gd!P7p$k!L${zhXqFHb98?b#6TiUPxA!0= zl@BP*V=cx)(f4EE6?9!Gp6Q!S^9?~H%bWTl=zA8&9GB?!OEYW7PFWN$v)3hF z2|&*khHzCqYIA<_{ygDp@wxlY6}i%XT{mH9cE9qFhEPVsa+b`}Nmu($W82Clomu{8 z3=L~J34BA(HwHi1_yuHrrGRy!k^kUmXPg=pXGgx}jC~4})8-sl8hFn;CQ-JzZAaT> z>rbunH$FexftDAU6*%7pW0!?F(2fV=IJ{k-?W-W`mm$Tbr?}lvcxj^zog0(Ym zs*-f^4r}_!A_W479Mde2U#AnQb1(Sjq?|sUgmkVVBSy6+I~=shyekf9s(C(C=JJ$) z{sC7vR+=}*HCHB!FF?3QZd;?OI#^75FA_y&n9Ov#UW7|&cb1aLO4YdqEN?CQ$=g`HyWlJ(Q9sApukVDx zjFv$MducJTJm^C>z zqcg5=ag@HMcOnC$da-q_Ki(5 z%WAQ8;O5ox}XO;Uue_sn5`|a-Ya2<~G;j$9|bgs|R2!OcqzHrmr33W`m zKBV!n8npqNlKZImo8StWpzogJK|wM!ykVsw0P#mp0@s;lTjsTCI@Bvh>*4sAWABgh zOS-prD|7i+?w>hxba8A^wA_&r?bI30QwJ%4)MeAR<3ri^v)>QTjFBdso{Qzw%#RUG z{WWnotuQn2r{2_Nu+QBQLZG4D&Vu^jx(t2Up;`4CV23!bG-B}`#G$`|70y`K^PDd7 z%X^d!WJ`MK#sp#g%wX*$E>SpJKHfTIcXnfVRz66YrXx+kGAj7J{(j2YwKzID4a>M{ zct`;6V(`8t`s=2!ySW^Fojmmd-UusiYYL$vVbjzt+i>b^KuQVPKs7w);oVI#sh#w> zHrw;k1nrC-&U&zqA9CB(?erUbJsh)i$15TWEi0UMtXl+nB+@^)nIxJZ__7?72b)+pgv4GSmB*m6jm}Y^?ocM#Kj3g>-#BU_s zCAf23Ct2!qo0SPw;!RKZ%aD0v7kAaNq3@uIeGA%gnwQE$wr3vagV+)&M*1wguJv1I zoIPryi$8W$4Kj`aawLRaBRGcw+K0IPgc{E4uG%3;VPBPm-U9uJg;a7-mKxbPHC5Y& zMj>9;s@e~|-qwV>c}D#9ZV%u)aDJL;ouSu*#N70IMzfMhNAmOy?qYWp{*m&0foid1 z?!lCjv~k_Y^@>;DK@(ym^vBC(0GucAm^TC*m$f?0Fr-ZnLGyD!ktT#Gy&3DnC+EJn z^M$fV4LfEIY-0J$cvG8YlwvqPL}*Pmn=jfw7r#7h!H5hMhT)X%!y?Fw4ig8mhEECfsj+3#L^$ zTDK7bMDbKuRpajh9&(u_Esi@{-xTl7MsKP9;Li6s{Gz^de;F6_v0;Krlv}iqeZh|g zi_*Xt>?T4b0#DsXt&bX?S|`A*ZKbv2&{iu4`N!peJ&+u+h}poqW1R8WJ|NHS(L?k4 z8+B;f$Xv@KJ?Pl}+KtyYGS!KlT;gR4TR@_wb?XMI8eBCSZ3CYhw$|y$0WR1b=Es&2 z!-O<|yQ2#>hNz{f421|f0M6B+FcO8SyS$#4fco@oU1euyj&S-mWk`ox{GPPTTHT#_ z=aCYFYY~!9wI83ox`xE=!n@N4GDnxsP3YG#%su(=6ta_d2VCT&XVz4;^ok$lRKPFI zxO+VM6|s2x3xso^Tb1O{cixCnS1wPcHmkMFp?+H!Z(29hhUY|Wx`h>IWn+AQ18%zP zth{|mctZ=luE^=`TdwxlC1XC|m6XEVCwY~j1FoQ(RS}jFI$fWcd<1Tb z%~xBiHdPps)KeQ8_F8pzw;!fi!d8t*kG1cZxdSd|jopkPDPi{_gWDb9YwzBGovS=^ zP(6l46wA^p?c|D`rEIkA{og@mT@A-H>9LfdhCKD4H9V8(yS2ry_r3uElV`#+y5CtdzFe_*_?uLIq zCDYMDk-K3+`kI@NjMlpxy|ElSLe6)`vkoew7SC*5ZzjPnOsSCWPb@2k@-AS`Y~I4@ zr92kfx}=M?sXbgOuR+blHmw!$K2i`T<$PInR>#C4bfL7!=w}PI8sb-fdDug8p6}v4 zM|syXH6qBswX(7%v877{s~$-DBVzW-{Qmd*wP(72G0PtPGS(a(fUXbMp6Q%0uV<}B z^CskN&`D@VKmB-4Rcw*0b(rr)5Iucy{Py;hE6O&-Ol zB8XxI0;KbOUT2KWT7Y%T-AkUiuhrF>Y&6x-2QGq>C|>Bkd$YTqUb!4N1x#)TUOTJZ z@))R9rSRQ7`$<`GgkZ1~3(3fiGYO_4YPA|xo|?R)K%@Rw!VVdi>RAvSsSO9;L5T05EQg0q zM0|g}B?EtX{5$C5caR4%wYJdQ;46T+GZh1Ai#smNkT*)gPpe$-0sPZyGI%^Ra$>wX^qD91ADXRPE4<2}Tc9O|n7rym8-D+rluH2RO`= zWP8TqL3T_OmF4sKqYm zyk)W)+?t;C;>~!B)zM~gv zRiHn+p#IUD5%h~Coq ziK(1uqB!bt_95pUcDxK%M9Mt2P?U9*HfSA5Cu;po0R+YznDn4usI(~0pO{k~aqlZ? zd}e>L7`qQO>Mdul)TbeGp(%||9aG7zHJT3He#=@KF52uma`y|3 z_%eFQlF`LJ(&gPYJ%YTjM;u5=I8&r58Fv4xZ#I^Ha0_Z`a8uj1XHf8Xw99gan)l+3 z`q0BJ32rxi5{Nh-BuiVW_GMSz@bdB?vIPm+=Yq36#tT^IDMl}dR|hB$zh?1Y#$DQ1 zZ>vGD&Od9Pfv+~M)57QxM!a$yz&>Kst_|7XwV}h;y8M6^-JRf(adjr}t3|DZMg*1R zK8vU4eda&g_nd7zD%;e0H{wD8;_LYwNaDwM{oQS1I=eI_+muR| z(K*k%HTi82IWA5$_9k^1wo#rEdc;s;HRMFtO>HNjP)a0q2GgK87=bF(Q%K}SOlgrp z=i>dIefp$Ub3bRn-~oax5(d+mz&UqP+u^s(mmk~3>fH|}p{>vh#jNs_z3afXxZvO= zgV)0Qed%7+$%}7=+zo*op+|G4ENUX3I@*F8Q(LFlkS6iASjV&n2%_xf9IUviy)fj` z2L48$0g_H5O{VEu#FzN{YTLvLo@kC$6Il1$K}B48qg~!a5ES<@EU7^yGv$(prsq{! zFRG6}Y@Y76v^>`f(uW;^Gp~h%ZSj|Nvbzz)LHzv&-)={=Z&ByrI6_)wF85A}0%-EE zhEHq*_mt)c1k) zZSv;S1HzM#bCqL7SdDqJkTMC=pv5%o^>T3U#{+WK=0bm6YpfmMb8(xO415H90PGxK zBRzpT8=?17U4l2ka}8+7o1FFc{NOBAN*_L{Mot%S6F=l=r+BH(4pU|{>Mb?6iMdI9 z3m4ZZ45 zFAn~8S!qoR@!ms|{dk8_0)eb}cLA~LzDm$pR0g5FZ)b+u=!EX{xHBQr9>A47&My?a zSy1ceIkSGVDoP$;-WNmk_7NEQ;-%%4HkRvej?N45Q(}SO-YOvQ#46EvnVs_P28)A` zL@$+Djaq^(4=#Y0AK&omMQz7{wM5sns)_mG7y}cp*rcv^kE#P|X|C@uuH}mPr{}L5 zBd~5YfSb^TvI#j&GH6rsLKT)b4o)gQ6V1lYOdmO3gTeC5S&Im=xlu$|1`6ow-Su{l z4(9!g%JaB1+Y(J@L?tr_BtEEqUBZDZqbLb6H*`Fc_jGD)W}XeLJVnri93x9pu3+>A z70ZXr==Ms$W`jKAi!1%#9>2syY8KaNmPpw6%9jtONri+t;jS$pyCdwd=k;Wt#~RZ1 zUj)*PNM=HJK&$l7i`eA5MOTto2?&v~spxPbN-tXWfS2~yk8+^AVHFCuU8Kt}@!dyVE9(ih}% zcLuhQy+$V(9&7{FtHwZ54L$N%E2No)qJKeJ%RF9}Zk1IMp|?gHSSNq`Y9lXi(mTt5(5BfJEWsCfsR@7%LfqU6`7hVr zJcrI(04Ii5T25?LNZXd2Xl{|wJXTNr7JWPq3?LcuRxPizq8gJ6jI+`gR!uA*F<_LR z&ESD}FgxKlI0zEeJU@v_`XARSis}?Z)5P@~R+`evj0p*|AzExA4_{ynf z*dB4yZ6A$VX?o>fHK?%Q*t5B-zk>G%JpA2&I*bQ}5+-+avAZWTiPrh%m)dz7|4x;yo*3W*@0)KHt|B%O;8OUzw2+i3qaZC4>lJ;Po@5DqjA380J zF7Ml6u8Zy!eiHpW%~p$MDYrfhlNs0X@a64be7JY5ZSwu5G}o%iQkkL5HJ4ns=BG25 zP2XccO8WwuJ{dwWm{A)jkQ@obmiv`)7R__iqQ^(IRgnGBbvejY*%I?Ya{0FDD?{uj zLdRv4j=;fYbS&BSvV|T;B$1vnJqK7FE&~l@JQJBV^sH)>T_b`|r4s&HrufG$uiiLINy*)9=WcPc zz&_`?wYq7|)no6)d^|{7ndV_LV>>msU2ao;C`j551oSEzPv#$PSZHoOjMrDntSK5j zYRB1!u13MmKphJ@i_;m`A|;qvF@?Xd<)o%a3W9LrFHi>LDRXYhlpc zbD!+=-aM+9o(0NZul^43{R`y#Pkzhj704i`m4fg&!C^rH(fXzL# z|7Yp#kMi69&d=q@d2%=D3*+P!gVTmnDp$+O0U~xq|PR>kxHVDjO0vdfF(_tbK-A7PoU9{I#o2l5 z<4N^lS(BEcUS?xkLmvKO-R})o2%>{QO+UR%0!-$i_i>0KAqVbt<$dKPHIAsuk&7}W zVb^;1Sqom@mN4Eh*y;-vQU$bIy=bzPL_)r9;Dg%VVA(WeI8RBWI9jwNnX`YNTuzNr z$>=(VXhk;io>b%vRkAY8nWkwhGv;OfZ_kFYuxD;il!5I#;sJwuVDCKQ#y+OEcPST$ zE>gU1cILK=NGEJxGel;o?5!SZsI+EAO^cXHcRo-qcyVS{&$)4nQQC`Kw3V+w0bx%+ zZZI*O(P;mOx{LBnVK}jvw5%BW<;k2zC~;uZZ#6MZZmOO)DHj#2U1lk{WIa4Oi}kFf z+jR19=fPo)kIpo~uvbvzd0*Vlu8b6$+Abetftzj5wT8OD?|UHnbbx(xRkt`h7}Ofj zI86@}W;~qhxz2aQodNEID>6&8P2dz_>U01%tsPJ&a81knk6uRA^d1T&n3tIe&3c}J zLsVTi4l}rNtFF$z9iU07?)!4DSB!qS((%X~E1oaGuLV=S$C@iiu8j({pz{?H3X@A; zLxDiEUR&k_5E*Iw4iE?0$pZ|NP|VRuJy4HV52U(WScl)Fq_&90_;1Kv60j5*2ui+n zv3L>El8I^t`a*~^n<-(!JPX!pViN1=g5BBP(R~q&3_9y$7oK75FrMRG&$aO~_N47F z=nA4+QXN$+Eu+8y)<+r?Va8P(Nc)`U=rikU+*{xK8xy=8>3)*Aw#T{GG5^7K1#RRb zY<&SfLtRm6k!P3{(^@uK+NPP3(e}uUfSDX*KiY5(Yhd0wi)Jn_%h^#iyY_Noy_*); zua3ay9VN}v`8$YQ2CToPHSQmm-N{ql zL>xB0?RDos9SeRKJQXsx;+tmO+^lC3hFdEdmH$>@2{{9bSInUfRbu&=%gM2tu_f?a zx*$K5Y{Qmvb=2jAH}VOQgsWCZLx-048CejBSoYSR4+vncli6N`Z=@?((}+YFu8T&w zYQ@@J%+?}ci3tQF{a>2m09ugOvXzAm^%r*eNJx;-L{9`c{O~EuO2g+*06nDOyNY~iDZf9^Jot(9|(?!bM z9^iSgC4!lDeZOv#XL?O;i{Iw$1Hps+9a$#k5V&%s@ird~dE zE#Tk+6GGRLxyjfg?HJayf*V;Wuh@i00;=NwZ zWgd!TBP;Vu^!$PX6;TE-;xp#?NUtQRWyH^u=2_PdQCq9V$iKYGEPl8U3|dOX0EgB& z9Wzgvb>WlJo_r%h@LhGf{URRXyZu&#*jncG;wN$PAxhl86X%6l)zx zj?P*e?GC6=XpK`&N4_<`6gVF{ZP3p6qNURN!_k*r29l%9E)gTPmY5bd}?J6cjGoVfR4 zcXw6+Xseb%5`fl#jeEt+_gtrBRU4{fHKRM?8PsWCJg1346mJ|i1bV2_R(WKQen#nm z_-(@de4-~87>;_J_s?`d=fM|UTzL9|gf@r#_7R z9w*qUgzve4+5q+ldCnG@CWu+VN}T<*Ge zeu3*nfci`})G)^b?NXVP546^E;%JiJ*_O4eqaJwioe4{4&5 zQ%R)G12qs+FnqIH0RD}%qi`QaRIPTXP6|@SN@X}T33Zo}UA((@Z;+Lb3axs8}lwL3P!TE^r49E{g2<*i7_$CLF)lRRn5G;q(gOl$ zTK&aEUwsCkcBEWBBQ6j=E#_Suo8fwfz|KDBk2MjmD*0~46^>8nD?6l*t@9DZ35CLLdyA2|6Ku)lI(SsMsA{}CL_Ib_tZ#lv%KqrS`Jm9!LK%pGAhbP%XF@WTHi~r3DQ(b11PWnN07v#$8!(QJ#~d&&VL;vv{fiVVc?1?tw4zagjtBx`>x_3CbJm=Vtn2DS z-d5&IWZs)%2(n5vJL9=ZQ1PQ%%wIo1p*c{;1S$j6R|RnIvmR<|=*S2R@g5kF^#9^VWX7-(UgGwm|IqL~g`GoZ?;X>gzV%jNs+D4BeX$yKtICj#X)|n0`ID7uov{Czq4+QYXe8 zP6BPN-CZf^q7cu~g-=rs@o0@UYxNiL3`sumnWyhJi}t4AAP7d?Xb(!6w+#ZIUED&e zG2@gKZ*3StP3}5bb$t8S<`Al#KT2qL73KHo%hB!iVq%B~K9z#E${rWEeObp#zYOCQ2M_dBD68?HaO1vo|bt zYGOUU70W|-QLp31`Krr%Y@E*nt2^15Z;Kb_{Pw{>RORD)Mc#^jwrdg3^NA2Dpz@!h zEW|v3hC2k}d7Glo02Ox*SWGT`(AxU`C<7TjiQTsKD_@^$rkDAV*3 zthpZcbEmQ2xLOr-Mka?iT9#S*31$d?*XSZ3`vNNfAw^jE7d2W)NrQ=lj+TX#P?*!j z9>?b8$&Gf!7T?gJYSfBV)bP_JOun!DnV|>d2(+4e$qm^Uhu{b`$Aibum6!WZo#&$F zJnJ?~_NeSHw%PB$wfPrF`{?w@CGk{hBBMthn7R}BGfVC5th{vx2gS+3pceGA|3%ze zhQ+ll?V?Q}Aqf`T-63dj4Z$GI>!Y5f^JCsSY^R!EDod7V{(`a&Cyr z?y|@{!$5)ne*pY&R-&(QAwaq$L20v85FZaA2j zkAp@BJ**Cn4FB$J1Lczc|GOb9rIgCOUf2orvQCAxpCC~*MuFQ-lkog_Z1wF5O)-w0wsTl~ zRbiX?A@(D)1J5Z7Wq_`<6P&nsGJPyr7IlGr(VEuI*&Gg`bQ7fC6^ZUGpYm7# zC##+ov4BoPA>Qsh=@dm!BKKG7-o)L*<-rH(yB=#{UM&#>I3uvs=ujm2`-`%vc=Mtt zroO#!TaNt)D}R6Xod2!0-y+^<$J;R>M`WxG3-j%1Pnx6OsM;43boW_!M+q%7xR1>i z1JD^jxb_gh$G%zs5XPyoORpbEMh_WQ=hp*#qeb${-`7;fui0_!4qd8!y4*DthiL>oIb?2f^J_n9R4yFxYv{mV za<)r{s?@n2Wb=mVrBPN*qdbXdve*1BEzXH;+{Z0c zBCy!kPL_AcwpGmWmljbrsqV8|=n}9$jlt;Op6P$j7#PERk!2j&hEH>Kl^8Jx+ zIUQMg6ycCaXdYceV7i+qVnhWwpHgEsJBYYLhjzjS{1`o)1=(swX{ftnzW6ofXD@#R z_zHg0S;OZe=ZioW42XZu@sK!Ml4;C67rZJ6lEou2JGTc+Um{9tM4Z z!NVBoONJ=@DEamg;YeMk*go2?7WlUY(fbIFe0F?i)>y^vJM<9}ov;mf_zx~YO zlll#S0S(Cga*QoZuX&YNP`4#1|MO`2PWZ}g{3ssan6Lo{nk>IV7cyJ&QvM_p{htG3 zfAc8*W$XX_s`l!E3Ss1Wq*U!F&ax40SE+sN%|`T0P$5NmbJ;Q?T02Q@qQLs;YoTVC z^bqx7^y%J0X+0THT7EAPhQ*>t|Aq^7Hvqs@tbwJ#nTB6{PNRjtpEEZ3eJ)EG$z8hA z=_?4yA?Z;(?EZR2@Y?IK!K1HCRU%54Np4NPUmB9)as`fyY#3bpvtcVCm+uxm-q4_I zzyfT15R5P4WSit5iv_U>ujOY-^j8eQNY&7D3z#nx>D*Jk8*dwCM;=8gyFIch%!fjn zo#$d_MQ{9kWC^_TzBH(Lgg;`|O1GELJJ<@cu#oc5qi!bh0U-Dd{Qk&H2yv;gz<83sX+Fezi=WcGFv)fQ%Js2-uR_Ur^ ziT%J?xC>__<$AxGRrk2OWzd^HE*2mc(r!n6m1K|+wiO|y$dv`bqm={AcTFpxQJZT; zkhK={_%fL8ad@K5xT_FgfWqiHRA0DLd1WY#OX;npv)QW9v+)I=>L$3Ric8DPy#8Rs zrb_&&nwG&8Eh;no4Cj$Y(v92j226_Z4mKxdg3S3oE9nc&O63QwQNPY_@^voR8F^w6)(DNcbZ)HeB7(;VQntj>r z%mQBj5=O(DzNs^cl(?Ua1V1i##Y~ImZ01)&(%%XRP5DEaclx5ycjd%d?rIkSn z`+R~peW=R@aonnnkmzo*u0Wf01*E&OF6wx}CfoknKT0P`(fskd@S!pMh!dewF0`j= zh#bQ}nZ|S|woGiPm12tC%m%YZmzQQ%VL<#te%}j9&=Jv>3;}<>L z;M((#{LW~bqg7ADvApW)x^>;)qZlDkvY4eJPr%>ANF6Cxy3**D)j$8Jl6A}?Y&~AS zP~@|*v$AW=9e#2RLFSk9ZWj09hqf3ts) zQJa?DQzrR*()Ih-;5jm>8T|*FlF^QxwUy%cIS9Y~SZEg4>yg!`*D12)}CYd zTNnW@t9$Ioz4ue`n!2P|2kl`bdm`CJl-LtHSq+nG^>H?g$&Vf=juXV7P_yA~tE#b* ztn=VzVOyylBazE{UF}xJJ5Fd5wK0Q1lsbx(A0xW~-Hd!0d~ zJ~8Ri%lAyeiAkue)PeaiN*^y;7nLYBJG4WW)pMO1->AEqO<|@qTW?-~8g^KULLFKh z0jc3HGfkuzqw)0u3MLV{CYIt-)f$p#Z2gbyp{1W)_b8R6XXrmH_F9Du>JxBlj>S$z zkG5*VUwuYWS4W*}o@&K+*DRI>YAZ4sDxkz4!bW&e04KispE|0QiGFzk;{XB^R-!+!p=Qu=4dMob{dP z&GrN@vtJg*dVd036r|v+*loa)YhI^#mMb;tQPZ9CyjuG`TEk>ojO?zjUjWRI3;Sx~ zuH7(Z#k8GclOo$2W%61>EJPO(2;mEHb;G?M;fCpr%x`*}67O`Wz7&+5vK7-EF7)ZC zQ!&ea&b_qv+pN~*E{!jN{qqJBJ-Jo!m^Whoo|lzmb28(*zfQM=njn0ymcLc*a6`ZD zcqWx=&f$pGW*OdxD^q~=Gw?EZ>-|E}Wl!T&Er!j$7(ZoH?)=kA)r&$?rA973{~|n# zzSdNRxUM#wJmRXY^{VhspXkc3(`NoOZMw7tdq>%e5($N^OJ_FVi)O=w2De6Qm&h8q z2S49=41l<|03dF9}b+vJWeL`QOxZUm)Z&`Df`?Ponwg?$6 zMODo80%ZdpF*d*&!C^BGv+JY1n`0C}z9_{R$1y-j{EN1BAlxzk0~B7m*Xlg0JMAM@ zf4tmlKAp{cAmH4IZJv|`A-!cUf^+6LAErgt^9W_}&)%R+!I}c!voz5v(u-tVwc; z$vkiT5>+_1oGjtTH5J)9N$j~r1sB=~bcx(|P=opW>8R(WTm*=4NgSItYgK_=&FY8k zEi0>%qXtVAf>XI74c!`XE<|Y$w7ngLfWee8`DM-9TJ*z=gj3<8&%Ni`Yav%yKle_n zMl^V~(N%YD_uPcc!rc1E=%NNQANvxHvckkwO$avbTW0|_goG*}v66r3)&8!I!Gq6x z_b)QbJ9pJjq{%j%)ckJ09OvwM5Ta-}6%e!9MGQrnb&js>*~&-*2-A!>0H+c#Q?SVhwCHcHAGVLC{i_8r_@hVas}bumyL~q=n2L{mHuC zvBOEl=D3rK8D*=cs?Q7k4l+vbG^{Ud!Bcz+SHb1o7jWj8gi1P&o-z3I&H zGWv9D=FbDXQWQGyBX>6;(Bt)|Z5#%>ZS*4c>B(M+a8}QJZ%>B!0F-+y9rT4)VI*hU z*!6i82DY# zTY?SBzt}N1O^dOX$)YKlCINYKt4ov=8vP9CFsxI~o9 zOQOPaODA_h%oK54RXQKcX@<9dL0n&&QCD0$qJ4#XiV9=Jn*?*4|9H4B{`Dh9erPSMJIv z3#;#Tarm;=nlh585SdteIOMIpHE%3^Gk3{cMPeK)7e~fpS+$n}io3JuW3b*z!}$+O z9U$zg7`Br;iOyo~x=Lo`0z8=7$`0>917FKjyye?kW?Gn?93LP`In$LBmve?RF<69{ z(-kPJ8Z#c*`Y}Xz2&Pzu1n0IA6BFBj-I8!im7CiW-y zo}YL&dF|qk@Kohk&7+rt48ax%2N@;u5mWmOUsqlsc%0LQ%l|U2p#(803}@g7(lae1(4G? z!__U00>UvG4oPHd4j})sCgnChFE6r=Gqc0ll!R~g$cKs5{=9Gkst zdSxXSOgBIlkxZRkNR<~#9qF|4sLp$D1$($)w}iX3COPWL*=Nelypam=Zm&Q@4mQMc3viEj0h7SNph#0!rkHNJ&}*gYMB)pEoT*b z;j}Ie`U5nn3vl{^^UL1?NlqSX$#Wq7zA$KdbcK z<$396G7H)6rYQpFI=cmnI;K%t4gTjnYI=9!MkoU}fOoSW5TQNyE_jJgeI6qiNzH9XA;qJ=$*~&<9 z;smBaty#1TAjl*5iSV7i;PUpprr(K)hScO@rDy=+B497)^(55q&;%(L7_s(U8U@$vjTg~xV{DAm|QBjn45oLoI z8~5YhN3_wJ5vURI-}72{49i-a=)RUznLoZ4)XM{i1yH5+>svg# zQl5KBvm>hHQjCy9^1W8PRGqOt3?+h{n>3{qza;ZK8Mw_wl|lD#W6$G*=e{qZ&{x?p zRNz#RS%~KDvE8N!H+$iF>)TGbwV$tJ{OYFEU-8z!@(fHS#G0wzy00r9o;|wdk_a-s^7G#4c{AI4$O4l_I`tB4v}q$3bV|?=$t4dCDJLoLw}s z+sk*{5~l$hMts`P9aFs%hiX@-1b)WfF)?3vyutu<C)O}GkfYq|wSe?k+ z>p-y)VNd#Uiy4(BKvd}zOS8uvNI{d>Wh;9JPtb+V4esXVZ(1x~ZQpB_$-U?xn1Bt0 zXDQ`<899|5H=CgSlx(tjvDPwF0`t2)+Vur(6TG|s>iSg%Q+CQn{Z1}5d*a8<77KS1 zU2)>kgwEme8sSO<)-LH!TN)9`*-&qODsr)(sTx6n1fh7nJ( z#oYQ$AIUbT;`wy@4=g~8O-`WOOy%0*hREl$E9Xbd$N-yD7d{bJ-%lnDFqgHt(V((E zAg4MO390M96#*J>4OZZPew-fX3QO0Eyi}cN8T_`@H@|WdcAMDB-?nqDo2(D*zbOM| zv}6p}nsyEYIf)*5X;}hY*1h{`(`O%zZMzqYf;C9QG6V-)bR`hok{KAs6XUTm?y5a8VY8K7e$JjvbqMbz z)7b}7NtE}ZK)2vvIC1zGdf-N!%Sl0ihdJfGa)JY-%tieN-^slIrXkk1q<+Tjs2sfz-k&mot7WE*j3wb;e$? zRmHWpV&+zvVSAE{5~i+E>ReAMA8UUSyib!V+ngu^dXvUedkv3&P6dZ03t8FS| z{JEH0$#Ayu{y9~ktTRxXdCtCM`ZdM!BFE|CBTOdUq7l^zkDF55zdYOj&my-UZM1-+ zSLJj@{Y}@ z!OcO{AulIa9CC^VE)3l$&Cx7>0P!`A)%C^iA3UXZFapeTU1LnoeT3-#0 zw=csl7(6cVDhM*Xm^HLxTRYIScRSN15%RaT+Y*LYF1*ay#@V-B$Ory{)8sP^wq!LT ztV1b+k2igQ1*Rtge&5&>m{w4p0?WXK_V)mO!|b2_qLL)_$`$E)`v<@(6cltIvm_E2 z<D^l$ z6{?=MCk%}#`Wg>8HfZ=JcsdJ5sjkF9OttJd%F#F8y)GsV!{*tW55=i=LXOChpj@u$ z)CrhIv$$>CozIqQzs1k6CP83BN=lC?f60G0wg6@nW54-lbo8lg|&)U zv!O4chxDv4udLlt@eWY5(?52xLe|2=Q9b<+kn+jVb7Q#?xw;nhd9Bsh{Vq)L&pNGA zQz^FF?DT6}w;v#m*Vm3)2IsNIr^pu~4R^kRr)7ovBVrEY;&|aKB|_M`eJdv%K$5%> z%-*$f!lehDB`hdS&ivQt1NN)?WtnGgyIXTw&!{iKITL)%zQhW3llD?i z7UbwPyuLt~7r(?Gf?3M9j13}1>63q7Cwfzl*k1CEN$!3j!7(bSek7N{0edha$sKaP zE%1JQp3CbbuiSSk{IzX{b~IZlSQua@Ct0_% z0{WNPH;rD#OlXf0BtKjt2%Lo9c7!YNq5XzZK&)D^TemX?^g`!(0#EjXo*v`@RRHO3#!oh>T^&a8)*uA!nV-tIsvRS4#u$cAq18DA99_vD%kkQ2JTpVNb&v36AO9|wQpAAPhlpPsc5>BZh0 z+Rz_X#OCiT4=SQd38&L$tQ-0GrwwBPLi%4eObhzE+4bd|wyW*um^oR8t9vvy#wZ+( zIP0OL!2Y2k8>My41=F)Hr}>g=7sB4yWS%l95~<}r02PTv6!m456_;#Ka<1Pf7!IK= z741}GG(Q)CskzIaL7ujUp5?r1Zvg5kz&AQe_q)wxB)Z)%i{h&4SYg&f+NwIh8N+{h z=RE+NDJd|mbH^)GQ|N3<#`U9Y(HuperwzxzunI7=53o4=v%2)ZebQ=U@(2)Il2kfn z=enYXv|r+1ad?4!C(kN)s{!Bj(h!(4AQ?B+9+V0Ma^X?I)pfpa&Tc-6iYk*j9v#ko z(1AB)&F4J*`Pm<|>Z?Fl3?zKb5fWSnhI>Z@3cN{@+cTsGb|FVNmn$4+kBs2D;%qfP z4IJcu->d((Ih1wovA41mqJ4zs%LY>qsm*X1Ib4PF_qG?9R(xQ1vLO2|PR{Z5)wfhh zz*t*s?HR1hQ!%2Vb^xK=p9alE0%J0@n~Jbz>YR&wZ)|C}-N$KHDV)0-^Iq=Ry;d&y z8bd5za@TfPyY>JvF>e^l35#4_O1J)@ zue9O(^HB$A3%wA1qM5A|?hq?bmQ7Usb*gmHSf2-I!73V8t-&n1Fjr0>Djt9^Wf$3q zd=lx6w=XpOAusY^eo5<6^AKC=lF+S2B)%=r`{)XN5zsUyq+Bk-9*B^0sgER)!M39R z`Fn77($~9Jm=17J0KF(&|I-V3G1$2H_3hNe?6Y;inCztUk-eKLX#Tj5?+N!!o_1dJ zJs9-ZsAoaVdrDvtgI<(FPukyCX6{kZMP$N#e0$BA#nG)Z$y>wU%+@&X%&kD*kf%%? zazKAvc)T6wSdF1tb@WthIMr=SsMqNLq4ps2br+iyRd!Yh5w{C?0^W}c)Wo;CB z?;FXW;VrUnF;82&q|wGcU~h3mbgOFa8J7{DgjlY5aPWJpXei(8H4lLWYVeFi@abxT z87p%(`%xn>-HMaaQ70Vj&lU0#_XV;%8hW*2*gJUr33F(iB6G^UXH{iE&N^)<)iw_K zGj$jT4&EQFYPr1n@4fIU{-BL2W^ulrQ;qO-yuTEYG{0O{G{1AmP*dBSfa%1&rZ}Go zqinzr-v>seiX5#Sw9L8>zb)IH_f5~-R32j2e`MJCvRzbeNKO0g`uO6qivG+qBG5rM$%^z#fG#*30w;)aR0;gr(- z0`LJ=TJtx(uTaE=DZI2JgS^ z&d%E^1G`QZYMR(TCd1fzRFA!)hX@WsFPz)@JY}6F=Oy_50A+BX0vv5tz^Tmw9LWW@ zSJu=Be4u~}S82OV-;KXTHAeuJr2@tXEguUjW0X7@u@H@IOD2)%p79GQ=B&wCSiZ7i zB_T|qpJ639aFtklPaPzEm$RXIbuuP-%;boqE8Zq^tFbAyo%IU%F6-zsz#Tjx|GIhu z-0&@j++Xt??XoB#g)cWZuWAg?d9EvxTI%avfXSc3b$ZBF_2P(pXQt>BLwgyquX<%V zQXJj$u18ekZIRTUPB1TSsG7Itp5_qhL`$C*P$d)(27o%lD}FDCr+Ya0jc9L^?)z83KNoKj8Ib%#U^2gEu5i9PRaIjFupx`-Q1GGr3nmAQ+hUgw{Zy%u z#=ozip4jkdzPn)Trd5azosPPbm9#yWVZV-J;Q`^uQyNv?$h|Sh# z9hZJYu*rPY$NU9z^zoM;?Siy~DSOd{ddzlBnBj7994bG^(4e-B%6jb+R#&-xfrs8? zt-tL<+Bj|j!-dE$d+hH|DSr_iDY-TD9B}HB+Y#ZqvL{1TR1MS@#2?Ebmg_^>3vWHQ zYjDmv{aUhsm`-QOMP$m4W0nux%DXu#g}W%H>AX1nNn9EyzK{m@klp2#GLZbxnxYbH z)%)}c{uV4DVfJi+AFry8%F=b*wps$950VFt`igucq`*ncZRSLkwcDx4N}{u-jja2~ z(Zy}!foC#6BEDxZfiM(?SFFirsml_+eKd%;0cKTYTw+-uypghdIM?fSts9~)E}5fb zQTh+dpUI`4+;s3twXnz0BNOi)9g1ocl}fzT!JT&;Yt8&c>zq1l-v$@7PIafy-4;cFjqxUEOc(P6Ylccle|~1kHCe>l#o8TIV2z96a9q+MdAffBaIUhb^stWBV&v&wsn^oZ&x`-XN)`$#uuq z8&f3N4ltx4IJYOs*t4eJD7?+;*uRXs-Z|jl+Ij_NHn|GS_D77(D;(YYyi&d$oPCbUs((j(^AO=AJg19`GE!|-9Ebw>V7}SE_cL> z!1*Sf82O}7_u{^(e|q9CKU)z~O_8J<|HBh+X`d;8Gu^#1l*BTudBo2oX=5qYXWMwH zTEG8+>gYh*j(T50xUC`nC$S<-97uL@s78%;Uq~=coh$3rethqY$AZ;8Q6l>|iWub~ zsaf~wC{+TXglKvY(Gt1L5nUv07+n}4z3gM4>K0|}G30+TiJGny)%&g@eYfwq{s8UG zs{-V~ksbm^YZiV_CUp&s8FpWAokfe>Gj&tdjDD$`{kv%|7NlZ<%hp%C%^OA!nK^_r zcxWCx^XcF=M2M_z9ZP(glu^Q4?$gg&v`I4HC~bNsGw(KBNfJzKy#ewg%rc7pT>}5j z;%hW<#TiF6HKvmMij*|Hv#AqDE+Kaz*>O-`l4~Ap^qd>*=-&<=!lY6IM6;eZx(A;S zQ%#fNMv7>dGRyW}z?Wc+uhCFjmHZGwmgeufyXgR;O9MW-zKs9HJM3KTqvBTYN8M~C zWb>puXR3jEi21lEpe0Vk>VRmj(68w2ZwE7g z`}Ag<{>+4~zFfJL70vdv{C)~WGrr}H?sT(RzdsR`k8!o^!vp?i0r*8~^q+9V+r+duxv*|0GnH$^7lsYQy=x4WH-Uc2*_u`@S8a@Jm z)1-m(Yl5aUzr1O)A;;rG9U&&(kI2(^l4)R%2XkdU&+H<&l@955UGBiPP}Fl_p6F$T z-Cq!O#D9XO|MlJf9fv;eESsQ%w24W4TS}z@cP9@0NbR1R37r+!ZXmz)5-q2WH z-lbhVm4gF*7v0DHh5I_HZhZHevaJ9YGgGka9a%`%mcH_fTrWbQZ1srC#pOggt?34N zby+9$4zC!1484K+a*6sl*w*xPW&?Rsc%goTAis0JDNrx#@^;EGHryGhTKX2u^3x&q zaJBsw4bSPo$h|X->Qr`YE36#t+UHa(#^Oy5#4WO8E9+q?&+bZO211&j>uS#lKsslD z(S)o}$l5kjlwxOXyg5!EUM2F=T}oXA0uQb@ptKUUOQ4Gj$bwZKz@_dWdDJAihB?)~ zRn~WRsXQm&zN$M5ZE`!q0}Tifc0)T!i=d;1(Km+)J~&X;1^RD8+rvFngFt!{bbWNG zVsG9fP|xzF*0QNZjE#1%=#u_Qs||f@3!F5?c`~WAkQ}Ku?t@O*Tr!xc3$eGSEPWv~ z1!9{}*9jZesVx>)3_%6zNghZUK_kz&4%mqxMwO(aS>>H6+tsu>qz^vqqsLpPGZAIE z9vFT85&ctf;jr>jqV0S}p20a2=n}z^(~Pl6ISN>9TZ9+?I1T^7c+9xS=@ZWeZTgSP~f|g-+3J=$h6!r78~WN zC6Jnn#-C&GBw3tx%xmB{Q54i8fN2E{uzyk!Elrt{!W_X=zV z*GJ!-TGEr>h17YOhVb=F@_paHh~fz^%0d{dywoy))_4 zj1VvDj46>_q03cxE9t~)odh>wdw*<_EBP}I5FoW3`5aLq{sk?!b3-*j^H;DN zkhF*Sn+HIx4@LB+jMttL)4(bGmrzCjm)K7H(gK{l_9Ye{t}(}2^yzHyXq|!hRBSVM zhklNIUe)$0l3c#bi{BF(0?7Bfl=J~+Dw7XRZz}W7;elj&Ty@81&@8L-ylG#z# z>A%!0=daatZT;kpyzdJroZ}bFl+Ni}^|u9C;Y*2|`^Qzi&M1HkLrN`=WpCS`MvHRg z2ojEOr{9ZOno-ok0qVZK5FP#)C1oljG;0rVRSt8sm^yNv9CYUqZ!MWXMUypWXDiBZbUwQ^DgNE2E0grS5=`al+a)~GvGXN2JT_$M;KiwInheD zIWJlm;mkrai&5r*#YvgY_#pUGLM~XTw5T~p;fb77zjzOa6UVCec9`D0rKzkPKoj+s z&QE!>T;5`5l>Y&8J=psurjofQn1S!wszY8IrOjwbXA?EC)M{Jsau$um}w1P?gO_8l2u7jg9juX{n z$J@N2OYT8v-ft3|pJDp-^ZF-4x0P=P+yB{eCH=a8#h?4VTXvZDwm`If0^wHJs*kKpIE$!TX9=_cb~kffYf;`Yci+Lo01n{_)KXSmfB6QbN)R6 zU{xoCLf&Y(Mm}Cq4!y75JX5#3upOnSIIt81*o&BdwVnTm>;DBlTf(#ou+B2&*x|W% zPeR#&rgy0SQ@Q7>c;dd+sB+;JyrpA(Y+K;MH|dRZ8UKLe2$d#vs#~YHO6@mST+I3# z9@Rz$B{`qS;ZS`t*_|Vx%*?cb?_if})tH)B*eC1skS%h7C zWV6wJ)E|8{ji#RRFA?VJmA8~Q(&Sn8esYAr81=uTGk`J+GC;l_dT?f)w01G)?D#s6vQ!xT4V8zy12Q~h+1gghaR^Y7){WUVH0?_P|eR zl6j~PidB`c9nn;xaaBy@)V0;N$$2XmauU8G>7sdu)5dbo_t)|W054QbUb3f5sMFjO zu%x&4vKkHT#wtC&i+*2GJ=NMi3t+~%R{O_`f2Az?{moEGbLYj;e=5s@n0|5M{n{CA z9dX@PiQDsiP}@t;YIhAQbV#@ai%*dzGFR3OE$Ar<4P@P)$LCp|kqx-yW3JRi+AHtR zl~RMkB-g&|F29)ZO6GTobW8VaJROB!QOL5@|9ab-)FCiS%R+PL>u8}B_8`*h6j0Ky zzm5F$4jT|vf@LW~Cu zVM6pi-nyvm{JGcVX_RAi-Z$Ef%v72$vrf-OY`caiET#L4jn)9MlL+vq{EJ^u_oI4s z`p<)`mWDXjn{kF>eZNeK@J%K-Qxz1)0b-KrvH>ZNj7t% z7d1~g_{?-jwVn!6XBJva?b9_8Rht?zcI@CQWsR2c^z3$UVI7Av6YRS+kbl`JhCq0 zdJqokpDIKDAf|ekoMXIp{V39iDG0V=D#8AYrQ6_Pt0aFd~!_7FT;po03 zbFNLd!C}&OzvmJk`(@TN2O$UqLXC}e4T*-SA5L=957)M03N8hO8MrD-Z=k_`L=7Iv zb_JHf+KJ1;oIfqz2Rcw0BzB&N`iiJMkPPG_7nKZM=1ga9=0Y1Qe#DvkPP!*P#t=O{ zd1ry3{mGfAigJ(16VK(Hm$Sndp1K-x0-;NdkL9qvqnvu&a1oA*AG>+?#us>PH8K?$ z^}nElURyq^{o>wh0~y8MF`;h=rquZUEk<#(TSqrCwkjg{#lzL~GnCF95e6z=7@Y8; zs&My_c*1SHTM^cN%tyWmtkQdx_oJ5?Yf?5tLdBFl#HfpGW=ab8RPtAPBkAws)FsG- z%+gBrIfEzpB}0%D(-pZbu-Y`nILXgV0oXWp?@f{f18&8gfE1r%0c(ib z7b7E@*sr|N8zz2omquCw>8QtdWqIeYy7X1ko*za-E|90yV$j9qZ+bYdfwWSb4Qf{d z5^L(f@f^Liy*UbFj;T3zX3d0&ndQ}Zzyt0F>||iWNSdLtJ$da$*;d|)(AqN1kx27X zAszop6wsBzWAKcFtBS8O+cC!brHB-LCea^6Lh<--Tsr7IZBMG;@?mdUCd3r);@8t8 zo;ox@+;OP_6i%>(_l}mEcUNyUSC=erHE}+QF+=UaC8ZLw-Nd#p1$m<0SM^KYIebD1 zDcozvYq@VdAJy1o8p`LPa3eD__$HlyM9>@sN&|fyq62melf>s6Cfe0E?HaSODN23a zbH7WN+}*qq&9AKvsx^kB_ZaD@wYNDFr&s>`-u1%nWSA~6tt1^JRMJLsOq690Bv`is z9qbPv;=JGWuZUlNE~h~aG80B$xia(T7JIXHAi6dJTLaQF9%j`0e}F`gqU=M~U#?tw zk;SU&;TL453b(B5D?}b2UrvcxT-%pt z0_@`3k@d;y^k_`J*jPr0kkvHu4iidKLJx!QHU^1FPwmLi?UaH7KD|~I9(DlIIyeyW z3M_>6&g`*)$H(VkX!tP(%(Txf#opTV8TnC6Q%VCZd+?6B$ri0^xs2={wjn2~^pO)| zLSfX+?e?t{(6{L`Z&MN<>&eXaQ;9BtfF%5doCLhcA0NzwdliH(N*pt7(y$XmkKI&E4ahN%x<^mT>^IG; zml@8~yuM7{;$|**fHxf4^Yxlj`6k9Q!xkIjN_XAvd46Ts*wdW1p6^KtKi`WEX}XAJ za4~wTK3PW)Pjq@|$vP@_<7->dH3XE$aC;;gE-Bc$+nAFrFZMQ@>nFeeo>Qq!a?67H z{a@JGzpm^l@Dsy<)z%lHt?#z$N3twiF&c}1SfIo>dpV8rPs}pIi93@dxR0wQ_sh)# z4p}GG@wpjEKb_gR31{L^z#}sP4_fdpB}k5=s<3(*k{CpW3UKocTUKES-ctRvc;%p-Yn)jfC*0*K(LLRc3ou*iMv6UyKU!ut@1beZ1n2rtH!So0#kKLLO5-kq2t_Lr<1eox%!TRvx%vM9VOk?G z5P_>D?bvhW>6b~4;CGc(3r{WO(eRIy^Q_%WpCB0!*?ZkisY5DMkm{P6*w9EJ(a-*# z%+wVV&D2B)lS`A?TP`S=qe(W;BsHwNv?KZ517gq*_zGcUv4?8+HFX4+kK*c_9iy?L z?!mk-+)~GBd^)N950GiaYiuYIR}|;5kIV6*mkz?t!M>cqP~&GJ@w~MCz$z6s`yBWi z-^v3-@7}sZ^7_h?Wt$IDn6dn2B@*8WiXIcwZwfQSsd)K<{@yVq?GDB||1y)r!VqA= z)BhnBrqDs*ts8U8fBZHi6Asj#Y(}fd#w)hVR7sd4e*j{g4yBl_)!;+pX#k~lO z_;P+hn6s* zmXCe?_FSGG^aIK2%kc2{g-!{?m*aycg zMLxJL(&4~9t-XF`L}5G^6weyl4|@9r6UpywYMAEoH3MTK<-{U=cb8;&&G0giH9+dT z&ufSHIU=bL@fu-vXne18z`pIsW>5JsjFr&zv)vYV`_Kebg>(8u4z@y& zyz|C3^A__7w<+|pj`mMRj9|i!7thgB4%=uw%r1My(q5Tqi-ooqzV(EqQ+5z=>lf281PtEql7wux^puRVvR{L z$HH9^#veNK0#z6sca5<*al)ZA%0wvSuu=7?X$)V<_41_r`i~S%*hL_nC~>(9eV>Vt zmcX<=^c8p}_E5))1r_e3PM@}~&h#4$%8iT^S;}XJB3^LAgkc1J=gM&~LTOSC{N9Pd zGF>0^m9vi-b)re>FJ{tT2BX5)(y>Ty)YAP);{5tEib6&5PZ79%{2NrDx%JE+T#5#!zee{0-x`R#UIbcA z)Nd-Rq$OYUot9_3JYM;Kc0``#pjrRJl_jPB?DucTa-nKf|wGq22Xu zyIO!Jz{>XgY7vi9yuRtXx&PGe`cw&)w=&<}Uuw{)vG`y%clpa*>79CiZttync2V-( z(y9AgQzcwu-ZJR1`(1re>g>F6y3YlZ*PJt@fH${3ENYpm8p7Q3?a_M7nh@!rH{@E9 z{q+1T@qead`~PMGuG3wtvZ()-{y#6%<$s%j`}HqQ0v-Y8_yhYc$1w3e?Asz4?2vmx zRA%!fNBpRMIJY<`*X!NENk5NW-?44SB=*98*N#d4$@X^qaBt$nZsGLBH;!j>n|09O3}?EnA( literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-3-Self-Attention Visualization.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-3-Self-Attention Visualization.jpg new file mode 100644 index 0000000000000000000000000000000000000000..408e1112b3d515773977bdc0d01d84c013b7dfc8 GIT binary patch literal 37010 zcmeFZ1y~*1mM&UIa1ZVQ0whRqcY+5N65NBky9NRY1PKrz!QCYg+}(oP3JZ6ag*$I$ zpMAQ|-rc9q?Y{TD*Z1DD_$E~!Rm@eh=9pvrV~l^yho28i0FJ!0oHPIj2LN!eAK+mY zkOWYWkdTocp&%n8qoSgqVc=t8prd1u;6291ry!-Gq#z|Hr>0|Npr&D_B`0U(ea6hj z&c($=#lSDj$05YZ$;I)j6F5{iIv=gX~9no>VaSlR7BP1dwA$>~E!1(Mr7dHCR^n@J=Ph17r*y%A1$6FM9 ziI~iac2rtU)gyv8&SPkVbX?0%p}&Uq+sOXM1{U;R8ri=L>|e(P0Wc8YV3UV{3y1>e zsfJzdq)&&cs|O4ega;bC33794G1SpQ>7;W4dCM853j;cA@*jXvUc(2V{psQZK$i6Y z+{^O>KLAEQ9)LR%8+{7ECH*LA`vEw`1U~>ravl#r&(;I*4h(t#GVEI)fFOm|dtQAC ze7NC%(^>ZYV|~&4w+*m9QCS~=%XwEj~H&_Mlb_@K8FDuDHePHhJ3CVA=r1p zFJP$0tkl(Q*zf`9mmUq+BY{DdY!;jB4ES)y|Fhd)UtIdDGIh;7jyp9R^9aH4dbE6^2f<0JBJ9?h32AuM6MOkS9#alk zm0q8gig-MXZdUPc7K77p)5Y~g2HKhJvsKo1a1l1MqgPl`(lax#&aJX)L-LYdM$wU2 zxC4wA^u=1%*XD(-H=-(4jWZW5BLe-kyFpF{hAa94AUt20qr;$TpMHDSANr#EiIPu6 zW@v6RYOrep9xLC<9JRzR*$4}Ew~$mvziO#$4Cg0Yd>+@+`$*fH>H*LP-1=8oO2}`3_wUA2bVU8e<&I!_W)Gkmq*x-;y(au z;;O!nAAlIj&HD}y(QBLxZ1Eej3hm^5oc;GARtxNuqRrcvQ&!bOG&9R`X9|o-ju_$x zxUT|Fnc_y44HWVM`p8xe?|kP1wrcV%7qNx--Ht47q|dwUt4XSA>Udr7#%U3sN~ABN z2LhCz;dV8>hZFOM$ax#$Q>!Zdx{}tT81 z;_%j&e>okfDa^L2nAstGEO9h1E&Dlj32$hs;i6a+$5mMV0q9D?YIe%_vBHA4Fj;c# z`r?E+a^%b7%SNYtG0JSSLaiMq65AJc2dl`Bmf}r;JMm)fwHD+6M7hp-!h;*dS~ziN)Vw_R#?nA#C5^%ob3dsc~gf_ zEwysTakAXoiQdd9H&sT-w0jmblO9?XB{{%k!Ph2qlLVYS0=vF%xk1~hp+Dif;j062 ziX^)v)v^rY2-$Fr^^Dj2?ELmX%O7x|ye0<2h&ne|I|LngR>RY9-M3`Qlt}0HRt_)VIOxSn% zg6C#Ib%f)hHopM*DQ3}B9}2;3kF|nlyjAo>sxZ#_vHSphw=~4?)y~_g*&!XiV@B(% z#0vSs18Oz2o`bT+m)ktPRO?C_2`zJLM6rcmw4!CAQ;Yd=WkD9LBwT>-)3%z4LW!0o zWWRaNXsHRWB#^3r8~=Jb-Ub2WR5&AvS1(ByS5#m;X`0qDwc-B{ge&82i&yR??r;kALFP)YYvAbGu zqr3cRzej6KahrSPC9Neq=fxoiXpC^rz}DjI`bGkIjT16w%FMG07Kv9{H5l6`eOfX#&-ooENX5r(%ia;g0z4^tD8Ar}ZZY)`H8x(lLFY{q?- zn^RdN#Ua}_kQ4O!@#1uo8@Imn$w-w|LK8>>W`($qF78|R9)Pc<`kI{y zBJ`IxPF97*3TBis(m z4(}Jp$9+Ao7sGuOtSECmxNmEqGEOJVxb$pA&B!{ySzS4cMdCpC1_|9yDr-Ilt1YMP zLj@Nnf)Z@5@f^$S1a~x1P6NdyVJvWGI;pQ>p4NB2C_$7Y;+Y>_@`I({EWXjLGz;pC zGjq>Bx4~4JVYM>_sz51t{1n37fUze+k7SPy_N=+CC2P|*PCI^b)RZk!`)1NY)AVVo z$g)xM6mw~wLnApDCBi>21^nrj{o8~qb|`6i_`WSIX?4SVL3&EOs(cw|gd>wCY^8O! zV%r#?*G#+J5TEITG$fszT$~knv{D}NUpWdhB22YoWPT1p>7#9a<5_?9?&!2M(_m(1 z@lY|&0EZUAv5y})R=7C&QjHE~Kv&XHlfxLkj~(J`S4Gz4*rU#*+Xe_!KpXGnA`kC8 zV8JG7G~iGKj3`Buwn32mW8rlj;@hc;?L!rUPVs%k7oueb4(bD=W;Vps6o=F#`ZP)+Nd9&Tws()+NFUD5U&W5T&wq>VS0Me4OT` zGZx#dH$3OHGIB|;6UGRQ6tn$UY8k=X=YH6^916R#Nm8*p_?CejNT1*XPZ`FWZkEe$ zBNs#;fN!m*+)5(~71J#oo+F@1yXlp2JCRAb#}=58`L%T!SjlbvT!utH^?$I{1g((u z8_8zJJ&j=xZQo1*tNmG$Kt~Fl-ar~%CN zktw&jamv>2fGkIKj)wmhb;xt1^F{$1^OkPfSJ%XZbQgzZ(#ok{P|`P>qz9QQ-qj!N z(rh_Zp`FJAGMtu_kX_iuYHz>!W}MI|jZBo46A~zJNV!XtMD&*1PE#pHb!Pz+%CFo; z9@QH>UFGU0GknyOUa+W0N@7QoWJMmLtS!zk>TJ`!T;EW8sRl*vbeZdS!3q=9ot1mi zfBez0{=9eh)wr7nw5E1B9s4pCy_vT&ub;GMzGrGOCj#!I#{Up8z1B8^3 z8-@zCbi*PpYQ|6iPNI0tA0vu0LjGJeI*|#v$f?F&4sL`_8{*F!Hg_<+SfWF!79acC zLM^A3ZxnNjbnB{}eQTWqk&G7+_(n=~$GLnAF@0@Y#;X{od&V!MPZm(_mp+Lgz|B4Y zWtIrztZw_116w-t)-`N;lraj9;V_`cUkZo=!JmAz~l zOQ;?jg_FWp{$JIQc~{d>dskmzD&%o(K-}DtAdZ~t1I^eIwX1p|=O3YxT>^4P=Y>w0 zIy3T!9y%f4WygO|`4xF0UTj}#7LJKg2CRVIU!<9b6?rNhlL;G2z0>VH|MbeGGs4!_ zWp||hdfKlyM96B@PQJ~v`?Q#aa=2bD>7 zI47KV^d*AREBRuRuZKN4prbyOsZ$h3h>H9E>+79(f)dRr)3jA-6yiH z3B^n|ll4PJP5vIIIM`4?u+TXr(b4)w*wOJs=LLt@g(HLu16skJz`rGnOn$A&K-H;l z=KE;b4rT7AU$NYz>Vd}zl?k;u-Lu(Pf8$BCGXX|~#LM@~jU$AY9EURB5>O84EeERh zBrsvZ6bKUE)?Ror>Z^}bYFu(@sP;A|=%?~%Wjdun#; zY(`cs;$4<(yMN5>*|oFDPcSz(4$@Dq{Z=lSMtoqm7Nm|z9u$xYseIYCXI@FFOtg zP9g<6TAqW2?D8%XyDb!|^jUdzZS;k{b8=$f?aeobI8y@E(DIdaq$ViH z#9v!_$(e%0=OhR_*QeUd(w}0YNswU9+g6cuUQc+iF{Q@TK0hs0KpB~7+2N6OzTu>= z;>yiOg-F$N?8@Bg0ujX~;mODS47XnVB`_6hh$DoJ4*`Xmjh#`D7<@E)>>i2MK}UI`0P7Q6MPiKQP)P-cbCRm5Dy>f5FXTMV((KW_M`bnP zz*@Bf`oiL^?5M94e$C1Vs5a@D%GahIRz3)Ogr95~=+zi^BZY@<^73?RwxWsfEZGkV zDLR_zfR))Qi<8ye@Ux_KZin|GB-7mdZqG_`gU%h!zU%1q!^Pe0N5Ea5WneKax1_t- zK~1_j^sNmc9?xv>4D^fkLBdh;jC=SPCXi;o&~@qy;*ufc&Lo*oW|8jf85tWmyFL5_ z3kQuiatu2*s)0g>?IP$4roSKUIH)b;d|W0gFsh}TIY6GFmJs`dnj zEKh$sBGqyA!!wX%Nqu7n&$3Y5yt<*tJoo3C8H#;lr&2 z{pfPOwS%f(gxZ(l9YDw&&%tk-ir<-%AeK`H!67v$@^<($f}4y@Rzyh@@;j1@94f5N zpMI50D;u1D0NPv+XS6Rp6u*JbFzpxUb$8FC>YUd?QP|i7(55o5ITJe#xR1aet5ZQ8 z*>X)4({$Rop?CA7BtSc@gGzs~0s4`7Yu4q2Hu&tvjcz?Rljl?tMr0ZGbNncL47gz> zrcu}s5!x2;p`!1qb#LVCtG?cfI=OI<6X2m|97rQ8)~%*hFS5<$P_E)vDI+IeO144n zKj!ZkkWJ-}+c$ALGyHsW@@Uvp_?YKy#P!MR(|Pn2hY+sAJql7kY~_jU0UiAXB5h;I zQI04qLAU$(ozJjNx{tNP z1$f0Lq(VWR(?hx1bCwN5Mc=)MAnz)#+T~UU&w5wS%A%Jq-sK)Sghr*iI42UR=?PCB zlI|-WMuboX!#5N$0$Jhzam!P@&jMY5j`Jtz9{?4GoBS(fB%-&XYiXV~_w5?3*Uv1o z+tQCgLNGVMsMzohHa`e|MjT-NfCu0ynK%rP!{Et%M0HtIAR1%V|Jm(dF`FNU!x>&c zLqI&tt^6Ze6GD9e=uSEf&(W%3*i1nRfQ5ShL>F*@g?f)ws;F*|3}I2?A39rG|8U0s z(Tx6gzy42&`>(Bx|IH+>9t^dX^Ke2+&9-hxn(cd&VDmacn%qs<)NM{iG9o-t0$f(@ zr=VI>H_lqMUUgi99^1I+Nq5&{wn`BR*pEr!11kTOrtLjlO&Lwm@Zo7gV?*3PP0dHL zTwBv)PXLf($S6q$k#IF|N9DWZ6Vs$!mE@!IJOEGLMH1UC**^-TsAvcjDEdut73qYQ zn&e&u*rgAgaDHDiA7)PrybdquQFRY*q{)?4`gw?W~I1w8*{?fJizj35LZ-Dj3;tL>a?+!If-p02fiEzQ?9ij3M| zUMlWGB@65HK!()1V!EhIx!Gw<-^j;-6vVMV$HrUOEyc=7n+YRJo|QC?5bdxOwz=Nu zI_#D2znm~ldegn`?6?lJ!O7Ax@KZx|lg!J7hwF>2G*x;kXDa0$(>os33CfK)s=vKf zM(V~_5qtEj5v1l@kL|$HHkHy_(?sydEMs9(zR@?kr$t9QV5+0Uc}p-VRs8wOh8j0S zJYE9Fxie%$6LL~u=0Bls5t5+Hyo3!c(PiCsEQ35O%pSNBxofz684l{?6{^|PEno`8UW}mBO5=Qpnoxl|S0>HQ+KbSY|Q z#v)2z;&2E>n(l@hjzJgINiU3@=T57bLD8f2n zs2F>wJ`M)Oo--bpS69T)NinmDSZUdrgfP_XjUqe!dd5Gl5~h!^$lz^f>wfK|6N^6>P*}EistB9Cw*QaQ*fI*jJO9pM5fHgH<)JmSTpIm zg=}c>*yD|3u1SHUf7AW%UsrL~4C52LPU^OE)BT97I{=X zY^yLNCf)2JU5j*JOrq!P{p>Uv{5^bp%r*Ir_26b-pnB-8q(!@yh?!)yO|Vi70de)X z!;`>fR*9F*_`Qp1IK}j586k_CnV}9mSxmzbS%eWzjm1Yk)tF(`f^U8i}#YK?wJqMmWL^nhJW6O-Umk-0ZX@OMm1-e(L)r@ReqS&G)IlIWX1O zZ43mKJc=PoA%#C}E_lI7`@`9YHT`&WTOU+{+366~4c@s#j3SHwz8;aEm0YszG{kom z7IHU7o=AYtvv7J$a0m8%Z@&Tv?R^wQD^{eR7dY>yn-Zo_$`Hjhz5d6{s_9G4D9A@#aOO$F z1+O^)ttHblk@!upH*1mGKCfggRHHIZxW1t&W$-;p4l9oB&Y*Y75>{wx(H7^{J?-cV zm_1cf`F3O7eeP5E`pSjln+zgj-&zoMnM)5N!l>o%U*9_oHi-2huF|$Mikxqm67CXl zNXYcpV{{t^QP_xVb|y`i(9}+I?|PMJypCV4{&uV*5kU;!Bh%JL(l*kp1x3m&kSW|w zl1`*!(=#NYD5H<)qm^xAK#xhFdkdbfoA6uFoqsGnby!{3`rT|Y#-0!z#4`1AR{3Cl z^qX?$ShSupa^WxG&?H)gQLf$x%!2L`8s-0B#qIU)PpHz@MyOyJnHWD%;{E5S=i!Xy z`J#Mi=Sfg&SaXT4IJv_vUj|+IJplF(z<#?FzGk>z;NJt={aJELu_Q<;lf~F3+M|&g z(~oMjsMC!gaaFmVd)iLm?Q8X$!tUyviI^BtqTVtj8_JP=LrB7+WkbwRcX8hq8A?7= zVD_xA_b5_Yy45+8%45;5Pu?7hu-Aw6b6Xc!p2)GpXzMk0GIckd*_DVf92KAbUEr_w z*ARu;_V3!O00I4v(Fv75=`grI``5cEusq41h*y76DSt(Re|z)_=z+azV2kRU1eSL- zFIO&Sy4)9XUmzK?_*&CVSdQLk!UwQ(!R@3Le}pP-MwH!3+GSmuGnUv5(%VGpY&@oU zHVjX_rQ6D_a8KmXxp=I!btB$6U5Rxli)V&n(6K0s`QsiRSXYWzi zhimra>k(zcij#mMhgXe!FzKMAp4@X2t(xKPB65_V@%*qiz_))}3`=W-onbmps8@o8 za;Hvj+I(&UU*(WeE@7I2|WkB^Quz zS3_wO9sp#|GCxhZki~mEUD|w({G)jcl-c_$hT{xOU57BuZ5_LUaNVA`WZ?!YN?IXjjaaTPm))bq$@rdwsU6PWq~DQSJllrtllVOaw^;n=V1Aqh*3m`wb1Q2 z&Us1)V9l~j>a1~a%<_>~p+yGP8WErDVY)9ou7z^9d}X+O%3I^1Ir7Z(hQ>~I0%3g% zwS9zpGf{(m`=0aC*gDB*LY=lw^v%hu7V&4XjnNE<;BTYIUoICm&|iFPL2Ic0`Xq?a zenF0WSzW?eu9OPPGfod$*5F3LX_okI=}j1O3^hUUZkim=DtY&&wL!0aMv;A1Sd|rkfuuyYIR|YA^?7$(Pl`4=xc0o0DjtD>S8Xs-P!mTuotZ2SseBJX{GyPMFM+TPeD)L*` z?OUc&&rq9rRoySI74ToU55e<@T`3dO+wQ@(P>WQ5fnh8OBTdTaxAyccRUj8!>WlZ~`!CAuU0L**RksJf&(7&wLA~A=7hvSd9rh!hgx6f=0+!}QY$>7M7 zXFYymMGHi&^{r5cFv^wTgCEwAklF@Q=2qI1W6l?%|6*S%tLM>J{1+e#UK>%wWrP-5 zDdkB3trc|;vS=<~S{qzu3MI@fF8@$X#@onPOzN>n{G>p^)SoX^s_r;brYS!z-BD?P&%Wj<>Iqm5lR_ z#3~&K0?!ZMbLom=dY&jx>?6iLoeH1JiIYDT6605++@Oo^?;8>J%D`Ir9mrt$ygu?z zE!qja_BvSAWKaI+nFG%hM>60QOtN&1n|4*0IhsfLJWjW6(qP#)_)&93m_%DnkfRJ; z3dSa+QZb}9?AD}M*9NzADB;Vu972wWu5PeX%%`K?OKE2$OqSdfN-}73UUDwud(b3{ zd4d8gJFSV{qlxu#3~oLMnH&w+`xjzr8Fsg-V!#GgQq2wnP{X#HQ z8*JZut3agw!IH(}ytOjfWfP%wkUfmo#Qfg>z^TXTo(0Bjxs_XPX1x}kc2?lp?^_0qjws4Uqn_(>eL+|D9UeNy=9)`uPs(@7lt$SqZ zxbTjnrVHGs0Mg8J7Y{f8NEgS7nXF!P)Gs)*9)v>2UND}ae|TxW(uIIs)v@0djfO)Z zf%#tome13Qou1~!y9gQX;-ac;`B^t?X&}v+vsuuJT8|vM-0j~5K&nx{&?ym1C2ztjIKu;Ud|Kc3N_( zO&zf8)9*=~3<;z>j~;-=UY)-2>Un-!+E9gL&otciPsDb72Q3++NY%L|?i&4wUeB8q z$P*Z1LG|3i7F%FNul%bahOb{u(H1iJ5l&0d{@XY}Fqb1M4TAm|Rfv5cV+|Fb?ao?< z5uT^^?V#CmWP}WRY{uTTwKOcb%C!J8MYfdKXOZZXM5-tNh7#r{zmR7-|k|6LeDI-h$Zo!H$P5Z6xV$adi z9Iio-v7cXH6n=;_S-AlteCpwQk~4lcmKUmZ6Rexc-@nArMcF!{)57z@^+;n0Y($bN zRZMUz)Isib=gmFGb-$LOoM3o85krG73T_0j4NFy{PP$1GDvBk|+;Sm0Gjk{=L_Bx{ zSHF8tC!LK+`7erVXc*!km;Hsw2%|SAd*FfMJ_nVU-jvb!Ox#N z_q6GIL>CF9-Jcob^dLo_R4O%iv2w}4#C;U>5^D!yOM4f&=@n9(%|K zMFnU@s9VSe^=0lmH!2mE@4a1xDL1Y#a09G*=)<|`g4Y`yj695RWF*)}$ouUAe0-ox zAl@XYo)pbjbrAXlhQ*t{23zyqRWXX@-;nfcUJv(a!q|BO5J7X3LW7){QBm+@Tw4Z~ zPxrNbj1%6xp$${k9C6yb{a`=ksqu+)`@IGQbdpk>Aa41q5VuK*)%3T^818wMeF$V` z+A>AA>tif@Z$}WmX-ZsFwUQx`r>w8!joCtNuAc7Dn{Ts<=U?WH+{{_AMOl|L({7x$ zb-!J{Z-=CZaTPGXV0Lh$T1a;P{?%~cO61KfNP=&T6xjW%G3%c;AFG;SIrCE)e@2ao z!9md-b!$c-!>TmagSt{5iL5kgU#-PrS8eyI)8t8G&Df!ljT0_^=ktEjz5!IewYTZH zmOF<8wOtg1tIAL8NT)6BQtwyhVw)^7-UPnA3>SJ0RV?>(p8i>EUFb`%-0IB-AM;Ek zT)elQCQxAD_k<~jkb*Q4+%xTyL`PF-ghQwD> ztlayO>ezx(Mh^F{TjBE$(72q5is5{J6lN9?$wmzR`l znlH%1UIejlZGGX%bSdkzGokW%<~cZWs*V_U4|8H(!)zZx_a!VzX}ow2D;_p08?S4n zdH@Cm?;!zu)w78KO}(2}oV_EhYldp;4?r$$Xk99cU2fgy7;;(t4gi&w^*tEm{Tq4nzea7>+23($N9JuGIu*CGG~q7eu08=Ejh?hn zq(A|5^T<+T%P{uOsyWvt;R!|Cm0S?BlBi7WEY;T-aNRn;)i)tEFvnE#VMMUCozm3r zpTh~gEn4`*8y}%w5kK?P@s=?{}#$YJPdy(lHrg4bSU~<_KT3jpiI=P z#A^O&H=q-S#~91^nb{gpZU*wFFmdS=NoUM3h+n`pgp2_9#J{`9Gq|Ra*~alk;@70I zZ)B;XhxLa!5Mw_ z8P2{#&~rb0mf!QXm!CnD-6ZB}qg2D-ZJm{;Gy2u|T0t9w*S^l@|3>ck-@4}i=-kLN z=zK)BWx`?hRb9*EQ}@amU3cY0zJ<lsNa9X=$e)SZITgTluVKdvX3O(tSDZfA&_U?~+JGKL{YtIz2E zY`5{ub|NICuC5Y#km-U-hx#PEe=_FaNw;W+fckyBOF8;gejVrHRmI#Z9IoPZY0h^5 zmN*5nJh1&Mvh}>`oUoSHy;kidrH_Vr$gy718})6g-N%3ONo>z8ZnKwaKF&G@P^#!los=R-93oL4?! zpAd%UTHu78(mumY{W2&T_SI!NTN+1-E)-p9z8xspwIBNS*6;v?EgShSP?=|vbk^%+ z&IcqAKN(nEnAWY>L{A2@Xk~S$$M?ZwO(o}cZ+mK2%8i_(!*C-gWutFGl35qKPFF?0 zlM7rph|_7%6gTE=@a2>~lgjteCdG1H8zNjCy#;;rk6vZuG$UBu023vuDhku2+o7uE z1>4TOK+BYC51v)OX+q^(@1V;dc}&%*UGXQ``*T5MWc}Y`mq`5J(!_3LNd>|D2DY%A zR9YlDwG?~XHA*a`=PxIj*PvWUz6sqgvT9G>d>!>arl;oV+3kdj>9y{(xe37VtktNW zTHHI=`4lSaKu+~c`>|EzY*rcvXP~o{-3uBhLuAPr1>0BGeUWL)g`;P<1IGj%PwuJV z*J66*f!(cN^ICluQj>4O62B&)>lF%?knOvUvL2$v!~et~w&XpENOLf|pV##&-Abz~ zkNFrU7Vg>HTf8f|!ZMiMNrYs$ zR|g)r(c?$os2xiEdETo7X0s94{J@o*+)R7J16GRJWO>>ld(1BJR_xW1Z&xhHSd7B# z_ccR&9XTJlP>TKDeox~_U$^PFq@a|vCkOiruU;$?kgvVnfE_v@V`gEJUp*JZeVqN8 zqXiX#wy;e!-uSWx^UY?LH`{2bk?xkNT*!J?R7DZ|+bp{xXJX0}qcR!!;!@EMInzR1 z+=9X!f=)%x=51W)BGXxwGP6~pdjbXjV&`(qQ9`~fzcx(*`k<7Vc{5g76w;FmrL$hR zGt3gwbtc23`N0$Xdg^spzFYmx_s|;n?ab6n@*rkV#)) z)50-ErZvX>Lom$Y&37w2t(BV3v9XMlZ+z?MjGX2khZ{cjY?T3PDvdF3B)iS8VSw=< zKS(7ArHp?tUWLJlQH=0qU`zdqd6d2AeE?3d9)Ozk8y5HYKbXR~#=8ws(f%|MSlpE6 zI{n4nGjer;WVb9xU*fSio{WbVy7ckMur!i2qY!7dj034i{F~t-1l}2NTvRjD_6tGC zV)QM-$r&0?W86*aQbJYn^d#5MgH(9V1Tq;V;^=LIeFOANZMl5QBb~nRND_y5Z$K;N z8rR^=fXB*T-gkEQno74*{(0FGNt(~Jr}5Nbx$!|+7~d%a)wQ|q zgai%zLWw?O-Xw6(fohvIpcoVMajr45TpvbjHnB2ZP`-}?MZws?{mKom-|Z6*_D64u z5tWwFHr7O~(8$YXhI&s{tG%lBVG$svqizV&b$Au4guJQo8wU!GBp&BG`SS=^!trH{ zX_#rmfi&6YsFlK<*dc6w zHHx`VRESdwiK#a~$&|ThJWoipW2w$Y?FciCSvOPr)EMQI)cE&9zkg+QiTt&;H;}FA zPkb)QKb80f{wjF>BN49t&(+bcf7hJ;4&DEUuD^C}$Lz38qKk1!5eO31fw8A! z)%e)IQ_B#M`MDS)h)w(u#%z9?aQDzZONrP-dRvJ-CsNH1kOIJR>5Y$Hi0!Ytn%yIIjszq#2W2U<3 z*RFi@{r#e!m0J}V{=7E4L$G}S4m%!z5dCrnvX4=;2?Nvhr&l8CLdc?AaT~SE0VgoR zLmjrQX@v}z*;6T3K03C()g-@vEx-Vzr7`~&sZjh5^I`w?@3X-Ew_@=8*SLyN@no+) zE{e)q=_%&cKP6CFRSsX^IBs|ue3@FIcXkm`N;A+#x!l~MehVvupQ>i&)=5X>gOY?& zVU?@O1?blb)Wi@Kqvd*)X#Xs;_)2?j(tnMWSI-^S1zEiQ5LtqJ*|O9-c^rV&Y_a4i z>6H{RF}xXD*mf(ob{{`>X3YiL3Q06}Dn%5+S0epIM`?aN>jB`A@me(+TG!HTD)U5) zKsZH+yH0r zV8ymfU@(z-E8|w{P|8qrkn4>v_eArLYpVEwCnLK$C&DesKq!Q9l_4uD5V={apDZlr)O!Ld%lw=KShFq-Qk$aAQMy4FOiR zRIY)1>qm70$uIcJv0}s7?t7uy>+$ddSdd-AeEE0y1NUD!{&o8T&|gWK85l4Auy29P zAE?0@zgVk(*y^Fo`QK1n|KlR4e^ey(KYLwG=cPf(b6TgF*XgC)g#9kz3^~k1h42}7 zCotBN=<^@V)2F4PG&U7VdIqTC-EAlZl@kB`4eL4OeK2(Bxu>Dll$Knv;b((J&bsI}-|VBZ+;_^; zRi6WhE@%h6N{R)IjW#!{{Ld{7J<}Dd`%Z=igD;KO{QX3krx^!whfMs(C(aAncH!Dd&1Yw(e zflm2m1CaFeW$CBNXMUc&7L2|-4d`efqrg_2_-41S%=Z8!M4s?&c>5D%OBrhtY2?Lv zyU$R3@%@he-gh*#d7=E!7=|kN7YN8ta+{>)?vM-f=9Z<*vRZPTnk&DRJ8WyB zENWIn>XeD>`O^z55H)|916?4C@9v+L%1bF15l90|XgUlc<^r$dxi=@D!#x00GPmc2 zJHMagQs#p43x)AjRJHyy=I+n;{(qkc{#}RTf?OUIIn*Wb#pP|2Gu~Nk)5yPFb#oO8 zRr_Sq-a_Z;q9clztT)-j?Y828jUpjsW5<}zJ$DcO~z^mf^hMDNo4 zM71`J7^6SOP+D|Mt7g7$MzEp|A%zmj5@8W?FvChXUq+*ye9)tZBq}~#2MVVYn)Jou zQp=y!89BcR#!t7zaDQ8yhq9S;v@a_4Ui;|dBJaCf?SjjoP1lNDoMIufW2Xo=a%bOs=anw3)2^%Ajss+n5BBedif8Q4nD!!AEf38=`|<#ep=1Tds@ z$6;tNSw)-T(BnWRX@o<@#-fm5J`EuknvJ8>b>yRwVHOt z_bH-%BCCzo;XPBjO@8vXZQULXn_b1_sMeL1JO(IJ<)yaqG~p|l-6RqbMti~nZm?Y? zk`5{|J&cNwdl2=R@Fm4c#!>M@p{T(6x%0FkLecj4w zk6<;QLV~=#k-Fls{ZXkFM5%}dF6(-T)Ca>Lx-%|aPwdXh`%xkT3LioN9q~5{Ya@vf>)a^_>2D<*1**)LdiBGtdA1L2al`9tP74jmB>8@<5T8?7 zdr|_gQL1<{;M*_=*lhMaTY~s? zXpV@2EEE^nJ!*^sRBjrxbP>8T|wbS^-WHe{@l>l_?faPCG>DK>Nyz|`p?VS)x+?r zeZ_npYOcyo{9K_qTY#N>l|eKx13P0mjqFR;Ck66~v`8`}sxO619W4Y9Y)oHMC@8eV zEb)ee?+2aIOgl!Ms4{T`*bbVk+dSGrW(dmfgMv77D6PryHH1F-II>@kA}vNVz48@Z zrqEUYW@GcQk+}9sX*N(WuoG|=0COTo+?o}8GQo&KVOc-*^$fq?Rg;`E%aZ|Ro5&Rh zHbGPVgj?Pg?ZGQ#x*y}G6l3)i92o@7u!EG#U%8z15rnOPu3(v1%z)h$!#ff&Dy0#Z zMa)@78dfy^S_ush>O14oaHwTZJrfo`RaFR!cj)i)ZJ~`PA9Prb|oYOQ(Gq*a}C; z(fhX^Ir=;3rnVXjF1XtC*eQTZ3Wi%*`M&#&iKyjAJcv|u1h+Sk`Pv@cr`eJ1K`KkYVH>T76kjV{Qlf(M zn5>Tn7GK@vGzptohp_CQu19h8^QlLPA%W-vgPcL(7*rUaesq{9AFWq+zWxij=EVa9_ z41dj^8J_Nu?vEzX^D)K!RqCHf*IXT@{EFdWHLg0PJKKp^0l>vS1Q*(Yn^$!R~#51QyG1%xX5`o)mF4_%3l@|B(~$EQnl55a+-%i5~W`x7`9hJyXeB@6s=8G zKI)N|m0UUbRWWz))WGES!X7uElv1z!A{oAZc99J7(zlVAa`_rb6vp8UyDdr(QZGBB zABJT?YD^x0aBeX9)0g%6mgSs3(rl!D#76EYRB5F9_sV${ST$Ha016FK4hI%2S6DVk zt39D|LaFraYc8X8{h~*>BfA>RO}G}g*gjl|FD47FG;afRwn|I(OR{I|5S?CRv2vc} zt%v3>I2=g4J|VrSmKQNNZMlmQykHT#G<};9dlGsi48#-Q5b;1TP#4Dt9N{ zeNLY~ea5-({kZqM@qR#6QF}n`&G+rK)|_+Awd-(;Vq>}`PL(+pm%{7QC3xr~Ad|Dy zqz^dGf2AoW?cj?0em!8&HL!(WFb*^Faw-$h=gJo!oFUa#HEgLoGw>5h z5taFCcf8oI4zICE(>+8`9p{$zfUDWAsNi6b8EWd(v-*g0aG)#4(;_yvwa%rHTm~?i zQQVNUzgs8mL;yd-kF)!BvN_vu8)#Y&6v11ZJ10`W(RFFH{PY4LXUhh4CAf!0jee1g z=jjfv3KAjHI`@?uk!Xam`+v!7jnuLrWM(#_RTIf%Na0*)zykXFX{nc0k>?p>?R1Y3 zY_;nXP>})h==BW3Zsru%O%R^jMkUvqfW_(WAi@&}ImE(sarS}oPLnBoeXSU=$B%6E zEp!@xjZK^hoJ;q4Wd3;_Qh5M19FjZZHB{9%xZwQjF%EQ02Pbhs9-_{f7HEw_Z>!mw zTyL&DIm4;|?x0lkH!3d5d50ehIosV2@xD%(v98AcymwIpI;!Hk1m|QPIBba!9ZdQ= zF0y@%61-S6n%}gso`%<3U;X@tZffqgXeKW z7E}$taRs!keXv)W;0{$&^;A4EtE3|~3JCT7lwTOkL2E=qUscblvZlNSdM>+69NtPvfo)qH@_ z=dkaas&`No=zqk*p6{hf zSv_>K=5h>*N&60H#$tIW^nmsi-PeZAwda9Z+lS%3vDRIPr=qg`jR1@^)7 zBd-Qg&30R2OU*Kh<3yt1F)v=i^VAqMRUvls{Mo*`;iG&Lby+3A%013F{|c{V7RRhV zibK*jpC5$ZK1q5Oa@?m~aAPkC@>UpqCPM6*YoslBzA4~49vDi|~BtB}263i7sz#LS8p4I(lm z1?7o%_#hdxU6iO(j>Y(mq7jLAvP#kQ$unEZxwq}seM%33z?4<;kLuJ+PbcLwLVJd{ zc0a%|Rf-zHdn_s{tNq>kHkxWkE}E#f?GTnKORLHey?0&v^qOJ4!&*52n6~99{CQqw zt+0XAH+$;r(!RYC5&J#Y8^(w`)K&~C=j`pbgT}wZD2jW-roW~TT z92vzDBwLT5-TX#pM(A@XI-e!{$Sb`nbdQuv$EF-7`t96_0jRcOaQ=LO0|ibrJL-19 zz4{5Zg2%st^A8F|F~SK$TlFDTOFNr0$r*PT*cNu$G|aN4#qSUfcto!5vha zyc~(r)aNoYUC7Mn^U2t7pYm-h?ml5av{G?cye8UNc}fCwCXzXO#$6;e>2w;1&cigE zQ&RocA6^+dyZ#a&JvKjmYUW=P!MsM{Vj-du+{!MC@hS%)MtU(to`ka z$h#<#*EeLOCRY{CpU4e4(a+cy7IEaKMx7a+vG=pm-!ccQn|z#;77YT_6ApzPsuH(@ zNTtNEK-n2J7OKfou5~G~xZG5P>)9j?=s^z(27uh_FRXA`aPng-gPEbC4$3lBMlZ86 zih$+YQERPhWaB%w5N~Lvd^UQ-@Q(Ax5?7qj;wnCkzhI8VzL<jVM&*mO_v7z@Dnr2N`e6pJE}K{) z4J5ucCDU!;3|tVpmWl%RS0p0_MNd?_2(y;pp3iNfF7Gs*_Jmtd#^Vm+R-A7YlhskJ8G+`+wkr6Vta>!|GOO z&34DE`U&Y*F+|T1tA{Y_D{|rS6sJP3L;kHiIeH{88^X`wTImNV!Br0}IBKe3Q5(aS z=^H?PxpHQI<1*X2$ZAPF*Hk?JrvK;?44D;DE`q?Gn_@r+uBW@kuK=H&l&%+J0n6LOQ@KVC(%EhW*FaJThbt z`f6H(Fy|`_nUmR9fDsnnC*03wv43L3oXscSLEFlT^h(tW7X$#1tob=zg(U>78NCux z2i8jJ5jfm{+=lmSKFIhY;Pb~#l17g>h74Jx{Pvi4a?ev(1`t?8{qQG2*R7(HRQq$VY!ax#iSNvi zn0*)#Wq_HK;Au24R$HT{6u*0%-;-foDuq#-WQzIlUi~(TXx<(LX*wi=__>jN`3|}Z z^!;3(Mt=v1bFVq>8au!SiR^>iPNcdK_wP4Qm8j&bX3kVhb&uB}$nrg{0LL_~6%U&? z?2>Qo$8Qgmj|&CE$gn>S(BZI37O4sI>Dg&FrW)of4tZX~a5)#5+5(7|gL1rHr5)>3 zf7{n|Gfs&!b#+O2Cq)yWpTO5i>LTdbU^$wTX({=T`P$lWGr!kkP&p}^3)*yG>u#C8 zy+AbnsDhnErfB_$*5m=+a5c5cs<-efTeYV^dl&kv@1QsAK%+{b!^zR7>b$8a+1|~s zlq%~&wEVB;E5-pe=Z5fqgPH7XOi`(i1gpJ%+H4?ryZU+ZbZ3RJtk3{8Zi`A84O$O( z%+6(OXGd>q!*@`jviXMIbJ;bzv@cmQMj`r)j&$!o7JXE-n}W5ek$u+S&Q(qi;|@f> z+ZyC(s6NEHAX&&-!oT!IIMixNSembnv(`^liC4Bof<%S-PSlGao+v352}%<~Z)_O1 z<{brv+AvD?{lc6uCsrIOj*^IwwTIYIPu$$pL{~npHX!H&xZuOouTnHpq~Uw7?%g+( zr0rU;a%R=;*Jn6~m>Ga{avB#6ai$=%(|V9*XHz(!<>dNFoV|*Oq`c>C`1v>bCCAow zpMEep8D46?t;uBgQ1+C4MAfM_0T2JShqwfZYCPHUcaTyAlNA1CEQ}SqW{TdAq$2zR9iTO*%2O|I z6*ClTPl~SH0(2cfc>Q|6KCEFbb14pnNK(GC@CrYAa=-jw0xC_KC()cj@h3BO^EZY((7u-oT9Y3lw}zxAL1I% zHwN)@_qx3Zz9-0$pHIT4hxgWPX~M;kTIK{N@#gx#M2RGnBa8G-a&BZdpHR5$vWMWx zNBzlq*-hEW)PW{28rKCy@tz@tei%M2tEmhxRdFcuq?$r~S(zMQaxkZ>{1SQgHaF?P z9;88xR_Bkxhd)2p-=arcGcRBH*?{TN0hVyoi@qCt4*9s!1-at}coBnx9u&3pvIFH1 z;+lT-Dik;@6*z1JCE7i?Znq-BFwqP^;Xq?v1mH{Ye%5IFTZV!Wbaw4@fms^VgRj+L(2h1cpoS37`^T{P z-}kxsUtm4exK|Y@q?Wd+spHvZzv?V7<*7#7jLuxt(j2GZLH`(~e?H%Id_&rmectxs z0IA6p{+X?y47)j84!n<%a9-!ovGYoC=oLNmilN67bl&K6p>0 za#$OQY}~chYgWKMD#^)H68#PHYCwop6IKKIlQ9!fjaopc)%>>o2{2W7e;>%=p}-t5 zaIM?LjxDnNTA{vIbk5xE28o*RD)k$pd9$AHpqqv34VBp>efMp=);^;3(&1}|BS?GQ zB-eOKw0gKA+GuPA_hVzl9!+B~k1wH}t5giQD61oA51>=+Df9x?b1e9&xP6SokUAe} zk*=d6=-=c<9D71i*_swn=4JD)&Wd6-M&QuXtmmRnJ*+W&7b_$(&of_6-8ZwQbBvL^ zuTov*@-i|dEWLTIais>IM14!abA_FUR%Fj<%Bw#@yNl3l^r>Hj;PJ~lT`6_d16H0H z3&@pMp%pnx4t11Jr>CXGgRfc8B>{7GTD~;Af6CZOPVmoB(Fdq!Rjl~0fduE9DE=C& zhiJQ?K`P~ixaByf}ha1@X<)2~^+t47chZY@FluOZz)%x~W1 zLz@r>2o_D-!!Xq`IYTw!$=6n-o_;kTvx zoJ(0-T@GEmqQ&I#z4vF%xOnn*L7+Y0Gk_jOeFx=pT^coQ4RBYdbhtgpD1UPM3;E?2 z)vfk7TOBVeA5n(`0Q;S1|Fm^}O;K;FQDuS8B;h^PP_gvxpnKEUd??A!;e?vy3M}x2 zax-ZYEvLJN#oS(dWKok#&;8<8S&A3=_%`yUx7<9ajAR%`w8|e4t~lN97F#ncJ|Dg@ zAh26p5Vf4~be}O#7_c*J*PHk}IKtp^d{xS2aVXds@h@4K;QCqE?`K9AJ622Ar5}_$J+Bk71`n?!k>}>TLN==t= zrOCqG(D1gK!j^zKSEdP`5>qENhP26>%$fnm!6C zF26DXM(d?2D3riujEs?N^)^HjNm$!c<6a>0`iF(u(^v9z6y=DRMuyS7f)q)p23V~s z!ifkkmwvETG(cc4V^w$f1lgIYI8rpDY5o?cc`lQ_O*Gr?)VT~lSQG0DzVH5VGMSt! z+pc%n&5lpkXA;X;3SMp8wks7ynVIKM`0EkS(UpYNELSfc-iqYZ6&k}q8v0{;i%ZCR z$9+l#{@BOG1C8OfyoBxXECrs`0R3C15jPt%YZEg&k=k+o*s)r9(4D#FBmAY*eeMuC zVeI1~c&8VJB~bqjpF7Kh@uVuKXH|wGDka17_*XYJ5!}5zO&PCp66F*=j6zx^mGvs) zUW7a=bt_x)h~E_QqBTQ@5V|b3Q)Lgmw-MszjeJyvnVxR{sVxF)gSM4OVOckHSr}hw zxV-1?gyVLo$(F`Tf7dvhId7i4o@T&~k^}jCd<7BL)+L}~SD0jfU5pf&F082Iz{BDo zG3D#oXpr<4%3rx(Xqp7JSw0eAc)oZiB;<4&G%T)>J` z(2%^3c&!2cQOg)L4+4=bzIM;5Fd@D*JP`*3a~ClA7kE@h-4K z`-9(p2f;L>GJl35-?{DZ)*#KUtM`wAB6e2c8gnXgs>qW55VHxmm79wm6YEOp?kOI~ zjh_!bL~)BsUBsVKg;Y&B+Xd@D52Ot>uKEKh9Gkl=Jd+XCwRM$H(MOifAnzI6fLQ&L zO$1iiu1WL@^Ig;OnkcgPb$_#a=Hj@{dVW2aMD7N;qbn)a&yJK*$h_rYG>W1NH*3}d z3}l4V(kZKqr(nw}*_hdk<)u|Ai`o0rC%}}UZX(;u3)|}(mbpcS{}_BW5Ka5UC!%7% z1sg@}+$=rq^Ur)I)WOuWtV-kSmA{x;z6dw6l5~PbjuZRfVcDgabEOjMY`I+zriui* z=5~>mzASc7wvXz6!o!Jcx1DOyF}CmM$=r^*>(kd3H+AU{_e`Io2iNFjcbzxhc8#5H zO4q%L<4}n5xR8uv?tOK|BlA{b=o)zjIeUFk&&MyF$tQl`$`|Js7ga= z3Qign++)Pb4nYnLq36AY0o;mZ6oahDEyrpNmFpZ4m zZ*~{p6P~+k)HP0873{Q6YOZRN1PZxJjA^21vOT(tL>NytsLe zrfkN_i`JKiBoxXi-j6RjK#nSZ5*(5aY@a>ru(j#8zU#wqzhi4!d&zxDM%^vo3Lp(Z^Ht zm1mk78=(`!51CY8$ay@0C91N5TMqj~8kQytmciUr?&eo9j}F#jeRpXhJMs9xsqQ~k zZlCp!$4VsGzfmzdui$i0ZuTCn6{g4gl$pkXa=^iIZb`=0umZq7S1(1oE~w@Tghf%J znOBg=!rskx4=E??bK9vO#)1WSW2~AJT=UF3Wee@2b6kGxhCvu?{p~b6uWx-zN|mn| zs%qDg3n&SeALD*#+)zt~h%eOT;G81U z-Wk$W!4tL&C7hQ^({fQmiku;Jr<67B#%R?9_Yq&DI46dtSuqkOZ8DIceH;IAd3zue z37$W)<%Br}1E5P(lj6pY4Eghk!Q&Hy94+!YT_X~tSODhqmsmSGC#F)r*lP}%T@CPM zlX!`4AvCUbODdD)EubES4s~RA?<08-s$bq9M(K zSB*EJ!*9n&qrm2r?ftX8&~X7Hyd@sd+ndY>XCEE<8d&ad%nqKm$=61?+q3rsj_A<1 zS|29)J#L{5ko|Ui=bkkOD{S8cs5@usk#Qje-bvj|Xw?069+>iyN z)~{TUEx1vP;Urey&M0Gunx4-#^yi=k!FwLIB=3+G9`o@^I`+~&;2<4vCqKb*DJ zLg#l-y~6l4{)6uzyX>!|u|+v$><#5^YzHl@+7D7R4Dn0ZXd<#gl2$6rij+tB|us=3XY?wbxh!$wXu~=u)>L<8N$N)Paf4N zw3{37=m5+mq+vqstk;+Ht+O*XiHe5n*ef4b>#Wosm+WSx-sxXrygBfmc-jZ0N{v=v zO%`#_8+Oap=#w45#nwk}zMm5}z}jGrdW#Lh*93k2k?sEY+IVDhB5d2PmH%9BOOX1p27;pCIuH)2zL()e26}~TvZMsbx_5z({&Xy3e)Lbob_CB#QBH?yF zBEIqWDQJwqm7Xjqtk@^iopX8T{%NORhZ`GDmy$Br1R`e7`UG>hcjKu$?(3t z&m$he!oNj|w!g=ZA`hu{=RXuy{KZ7`TTMmh+VDm|2Q?F&9Fi zAK83&mtibIOgLY}lUa8vr!rVez_BP5vq!MSl%aPDH-LpQUd>_+!}}KBwixZJ{{wsl zrl4FKHQ2HK;o+y!Ed5S#12JIqQ1pU@(ww#ghL(SM=*ti3_oNsErG#*uHw(=cHF4mQ zA8MXS8{18&)A(KlWnU+{-bo(z-?(32V?8SS^;y`{X3SNP+wcvfoL^m;|FSr`3V`oL zDyx7?a4e5j!lJKet1%r$Z%JS-TWU2osdDpH_Wp`I$?^Zrsi;nXty04vGU}ZiZqak&KV8z$NQV;Gv!56aPCJ?A?AX2j79AWaKz>9Cx0!l; z=8Knx+$ebIFKr}YogL%7kaCh}yNu-1f4yKYD|SSqEzMiW&%BX_<3w(9I^FyF5u-yh zKe9mTIA2vxuGCtjTdNEszzLJcg-v?x+G6)#cX9mgANe0R#^z^$%aKPDiG2qEP1=)R za<`mo{Q&19W%J30)Pa80;r-4`PTxTVk`x%uxfVaNe=i1R2_4(^1Z}rE7k_us8nzL* zCcG8Qge!(w`;@&)D9jz48{pu+%XGr&na7Q&sXfY!J)pMf&?fjyLIeByRE5oh;uPJ` zcqQ3}JLNg+=j#}l{rLN}B~w@@|3!frYuX&_Q>Z%G&dNHgE4wK-Bpu$6>7^G>JwH@ilEksl+CzL@LXPF zyZO-L3)hdyk(K3T_CJ>iXU6rkTVEA95sA+kFm`trsw&lCL-Ld-I^Rk2O-{o@#eQw& zsNIl9H61A2E?Dkky2YwtZtbk8yBK=%ZxOKmp9omvuSJQeMZ#Wcri@wKn2@#ek0d_dF6yOc;@`&C-HS=35T=v5i~iHu)mqP;*S0Ui#cktrr!!o?Wuv8W-AK0$OQwm@155u_;sgeTP%q1PxFn2m7SDPZRJ$h_HELXi|b34B)_clwG z6Et`gC%B^y3sGPgSXlQN+khUy7~bm5^LwCMXK12&6JD%8z9-w2_ndy3cCFpAgch;N zcPBMbxYLv6XxnDtY3XQM6eW9hX@WD$yO!8tJo_sHY+zk!iS=$Yf<@gi$R!G;7w1tTND!@yoMtwt}5zw0hFp zJV!13!g=9s3GH6@M)x-mqyr}vu(*Ez6~1V}XfuXBi6N1kjJOt}fjsyYkxQ`-(&%P) zR9=O0`HAvJ5&cDM2BC4-vio_xMD`Yki$Cg;`uU@Cn|6l%`@?E|p{_%)G`x%-AMfA5 zct9^xHdfTNtU}U#K7dg)(NR)Dc0t9Cm?@OwT;B>GKUyyZI^#5U zi;Czx$Ooq+->DNwG4S0zE);$vgR~wfoG)A|8!tS&Y0O&Bky;;8Lmjh3cawo~KpZa^ z;+O|t*j5{!^NfkBs;czvplW>DP};9iBF_&p&2{@`K6f+*7V!MbfM+Gu?nXhs-1O#$ z!l*=0-nrd5F3KW7KEU|hWT@Au(9{F0bR$rMHW-9o5~J^Q*r<@ovCb+ zis#70l=Fp+CVPm&!i98KI(pq5dkO2;nsZt@p3$#MBl@iE%3PErDK*~lKBOUxJfSGE zZ9|dNxugduPg)hTSwE)RwJtEcD^#%F5*bY6$Lt4shsHRJ?ffHw_`xXTjsa*l*u1d) zlz*Ra%b7}>${oJjfBAuIzuI)ospZN%<*94>L0GPxtvyru=>g?xEDa1d^$GU$DyB2m z=kwqeO~}VbcdH|_!;Gqf)bzPN89(p!!v!%qqi56IJUDPP{Sx)MfP!&~qdO~1QuUMX zdC+jzbDL3<>QWyQokGCg!SfbnhwdM0G}2#dwCq>1-*$FFB3a@oV~L$#egP5h4j8?> zY=`A_(eAvoZl)m6t5jpAK}-eBGzHz-0Y={lRrk<~h_!TgF)xy8Zpv5la=tQkh`3IF z)=MWm&xgxT!FM@)BOgm`Y&V{!JBO*Q3cL!-CnYVvWob!T?>a+;NqJ1l{xcYUlcozjofBu|DEc}tTXKR`))AIb za{bBT;hLeDJ^uc)OH;m~#s}zw@j!!<17|nL3G)dp2Ogd%OO>Kl3vFTj+ZT0)p-y?Q zpWrBZ@u^>44Ro&CV>;SBG}K$9x+RZXtJDF{XE)YgoXnm;>iS4+O#;ds+PICLOXr`6 z$!d%FHpfXaT}{Nt`{CCre8*hZ!T$EXyEoorqYYx=%dM4nt*inOnsH8CZRAdbuw??^5W z-PPOy>b1ySX9TYlY9csewOL+GPAEsjN&uHV){~Ot&f<9jLr_9BGtHr`vu(|@g;}7e zd+*s9UMncS=V7tEZE;pxemPU`?}ACbpz>Ot*h6kTy=7WjD9SCRqDrq+it3^vZT zN|wVRRHc<&zDnhc*{qB4Og3;YaOpMv%Rf;@8$ZU%{Jo@Xw53KNM3s+eGOs)1(!k)l zbctd(Tx&>Zq>Z$-f15b=NDR}Okw9$oxKNWd<~Uqn#DhL_YZg_;)#Cu22fih1(j!7)Qj3LS}{@Px>6spSq$geqiXmmH;Vy~YPr)> zUPlf<2e|rb#@RL)Wf&4^(7T&Mo9!Zf!)jp1o>=TGTEV$BDW|AyI?wqQ>Q8t<=-#A~ z@*rAZIP>Ih3kM@&tiZY$#c0WU_B6agT?&(@jZvQxhrH{Z41?&-_!OEIig7y8i~4m* z+^)1}aG%li=A4c3P*a7uMG9@zcrim`*v4KfitDgrBGK<%{NN<+?UErtjWQc2b8u1_ zYtOx$vT=Cnh{H4w_m9Z?nh9bl_IQBq&7Jq7Ok~EmAM_04AtpLLCfzYwUrrMtV(l~( zCVt8^rX-ePv3rCYa%-pvl;^0`r{Q6@TvbvVq?N8l-N&OoqN+v1SH0RDXD37&VjOL_PfU&BEP%ACQ&3@2-3k!YS1e zW;R(XUu?@2{RR`tm-N)^lwZ_etSEAaeQi8_obpz+KTjYvGn%>&AfEINGCobjN@q`s zCck6Omm-`CU}c+dT+5vPqw&n_k;2^Xpsr&`_#t9#Gn;B@bNtSImQ+5 zHVV*6%#bE9t6CR`gD$C>clm8eJF6}ie6w_)>O4Bsc-3l(?)0Mf#X=WgacQS2q>yD^ zi^(HU6l!X#pO-yHAjxh>c%81(TKD3K({QbEv;-&hi))MNU-z+y*oItZc#tLs(aYtQfDyLv97y1q!WpUho4@EE3-Rj zoxzOsy-@9Lng}WgA|#cPpddB*iCg|26W!r_#k;z*9-rny-Y;T{j@I;q$GJ@&NM8!% zTWgp^o;D#rdnrtbObl=Fgt{9YFy`Dh$l{}9z@-27nkUru$-Oz~5yNGdJ5~ z@nWo1H5xyT_UQoAX|i=*GeUvrsWA(^8C%UUb2d))w=EDEc+UAS?&<+L`<>)tEv2>8 zI8aKqc(W*!^_e#?`0dxZD5U!)>g-I0_cAdJvXbnBd|FQ*a=}d#RycNv;@DQB!z?-g zlaDnC=0F-fZw+zQ)R*G~UoSnLNRdx3SqG8ZzsO2*7J7}Td22A?S@H;5DVx*;294ot z`8U#Md8(iXOCUay@_%V~^1q;W`1Pa#5yHRrh57BhJB|x}_95eM0RGVS81^+MYqGtX zxdwZt34Z+m$%}+!D@ceqJvOClQ*^qky-Bk50LIQrpsRIV)VTt88WP|Mcj62QF(vJR zw4woX*F>-010!_R%q484h`~V?i51s)Y+xP#jtUk;ovyXlOMtUj27L$RrJ}^XgLd+O z?hg!92IK+*@GvME4T2-3h_4oXfIS6|;goY-Q? zWF~izkhRMzWHf;oLB=3g_5c?qbt-ku3*1aZFHdqM-c+XAKmM$^dl6%OynoLzZeYGj z$hyn6P&XqeP^M=-VJw%*Vg$Igs6hYr*Fx%ILRNq|x53@tL6*mvKflAwa{GJMUg%08 f-(*YYSBMRoo26HFWX<(Z{Hyzp{1gyqotyuVPN5AV_@QBrlDaMy2tr| zm!F@Xo=rqT_@OvAAOFK&FG0h_#l^pgPeDLH@sN>*@!|jR2h|J^;{f$Hn$gjifE&bU z=)`EKc7PsSCnnk-AK+gfv>WIcm{{02xOg|g8>(&rH_*`0Z(yKfVq#!`cYA~X4`2{u z-e%;L#=4{N9Gl63#4SmuAaVu*-P_R7M51lPR=f_ZtfnQejoe;0zU=?N5_1Mjf?-BkeHs4 znU$TBo0nf+QCU@8^R2e7wXMCQ6V}z;Gdwan1|OgJIk~X7w7jyqw!X1>aCmfla(Z@t zaruibGywfS$O8ZU55oRI7cofJ4Gau)4D4TYq1|u;2RbnZCL=G_ZD|ec=MHz69=^jN zk%>wxYr$pa)7&S0;W&gx#=^hAdhm<1KPda(5%&K75oLc7_HVi-0eo~cPdaj=1>dI@|n&x01t_ZXrENuiE9+GTNWO=xQmCnTssXo%Fuw zlqc{!R}|p57gn{*?pVM8tT+w8H4FKE26_S$>A~#4U!ykJU>61Gj#IZyYLEPc;j0K{ zcu`<3=?rvb3`n`QuztLd7b|U+q<4QBI&HZ>tAx-pAxn{$%u0&3M9)sI4cy1}W0rT@ zU;UJu!P$Sq7dx-mV0dkQ<}^3aw$c+L{>c^bBY;Eb>kGlu0bdE3;MJzT6BBW+`<(~{ zq}e=10nIL73FCrSBiiP0md-pelem7)tKEYQ#bfwjrBy|rwO)>8C%AbC>_+*la;Ss^ z%y2a{lM#!ZxICk_p_8UySyso^l_OC{`)fGOGK`jdQLp^r>_NsQvqPnCr!RX`kHpv1 zYQBu+zF()K8lId?DiK0>$>lm|jGJio+vrQ1K(nnQ8lrSxiCpxY^xBf&mRGE~xHUgh zoIUB>y>L>iJJHsKsoi)JtN#O0 z+VGIk_}}-ys1O|yl_qrwZqKCcH1t8$5yj+S#Jqbqqtl4ve1y*@w6P)mNwN~M`V7NO zV)-Et-ylY4BT2Aw9lgslg#`nCKtqn?KZasoZq{di^@mF0Xhtq!Nl4+3)^)KReFiKI z6C3+ILmoq?aj!HfYg}-X0C{N%Kb}RM|D0O0 z^mwKbvySypt@JD5WbhFRAlhiU*`T$G0$x~MaxKaEF|mHb<@EmS(F_z~6QD*9QE@ z2DLfA`N-dGgIFoUm{8MOq)!jB!#3@(-5d;TtQYbST93P>z8-Or3v)3Rfs6E+;XW+w zQO+oRo+5Uut;+7lzg70{>ib_@8;!RUzQxVYyDFi|Akyak8qt2on9IWEQgbVXzG?Vp zAJqEY;%&m*_BV6yS`h5$^1*HUv)={0bZY9gYv5c!_~nn(L?Prp?ND8TGsESw&I;hc z*1=`#JV;WGR|>KZL&5qOq3A`@P38PgHO-!r3h-zTkf;KWBM) zie6LXQq^&;g;W@%MgimfHOL^+Np&CCnFciJaM@ZjC$Rlguifl{!7Lj^XIQ@FOC@Ni)M2<6Y_6Hl@u)&-lx# z1+$|k?l<)po>$BECL7(8Y=zIEpA^T{-s9G4o#m-dJo#XtQ$18SU4u(7oEut@;wzeO z%T^aP8hTQR7P5F1P`U3~BB~d1s-jQyQ|~tGQW|=aZvSLw6;3GND*E>>MD6Wl=>t_uELg~-FfexTv^EOHaa@pD4J(&3C+RR z$u3ERXK_l8H__xGeOJ`^xQbC+fpZGW0YO~Ii&jH#$qnw4d-0#bGQ=_hC2q#!wY%ON z)R3aJCDD+={MVp?ryd~`#}jJrWl!W^-8f@GRt=BeE9yr9IBUU-2PDlX;FZx7S1iKU z0SnGWQiSX@ku#feOz|Dpq6y-(VX-Tr_U-9$#F4<^2YPcXcka4`_ViK{Q*6R$AygNp zI^#>DKGd6nig=QHXeJatKtu6tS= z(}hoCC0dx7O5Z9isu2w4T1!$Ne6o6OMU`1S-2Y}h-cmO@%i%qt{B&a3e(kPdOUuin zuxO*6M5ib9#znF|+bP$g7S-(NmBntb9_od;AU=esZ5z=G<`+b(n9kbW?$sLtsppxm zKdZQSRhqsQA*;?rY$N^S?`Pz3rTG59*1kjLPjPY|H)m)#7+_ETkFnZdb_9LKxMHYZ z({<9hN}r3pUmRB^uA4n$ZNPX1XhE&`-CXlwXqjPtm~5rb@Yb&LE7EP^3{HIODx6}) z?)rEBWQcT8Fv)K|MSeodGGSdzCFV-)2$P?*DIg;JRH2^uDbb@gW&4)oqd~xRW2({1 zRv<&({FNMMBQ8xaaH!^6Lu}}NQ#;P%kwaVV<6-_(?jNK|EUv}%E#USzkSr-`6F62$4BqiJE)E*vT-Ggt5ccqU~*uc77E?0_`1xF;IhOS$V>-w~D<1CHCo)RR zWD`VhzWNVd3Do|juN2=OxJI;y9)DD>+Ze}G+Z_i%DR(j z;P2n@Y3afAqak71v3J^EPgPs&x5t)&~{+h ze%Uexts;}4%^?q%kgBkqbj%fV2t9)t97AtfBRS!8Jzo8?ZFkAT%)eDug_~B&Q#yse zWS~>fFht%zRPYyoK%RWj@auZcnu)KyqP7TxJB1#X2&9-|n_7P-rNamqa|DKhLrTl@mx>&TQIyQv9_dsxs#(rpV zX0O_c_rea6>|E7*ZqYYZWvkFW{OG`N21Z-;zAH;_s#EXc1D*V-D+B$)O=CHLMB{%y z%&li>hFxI6DpwDV+IdOa?S@n+j`eQ^ps(9Nkhi`NUE;Y1q5y0+Q-ouk0cfRBz=yjq zL9LTJM;l0BN;bC}4hb3k8H^ zfJ>nN*AjG15ysAzS!UX=xSIjJ(3h7Zx#%|rTM}8bmUSZtI(9}CoVBdUHD~xO#@=?? z=1jk+S^xbI!Thb}{qC%a*JratT#^D9Q-BV3F2lZ@Q$};5xU6mFWYH(%w8}9%rciHU zQ}lJ9I58Li;V*6(3iz7y5xVc;TTUe7GvQzs-8i{yimpbfh7DUCE}&tLNx7;Bjl zIROv-;l-upM6u1+Yf{bdJp`j51wyxtz?!O}!%`s|#-Kb@ zd?{3k}+?~EnV%^Z@0(?7pV%#mcNYPOYN-TO9uO0|sb_1OL7|u? zl@CGFvHW@?nEkJTVUC}rcO&ze-;Z&@x43B;75X=KMD&(nIh%K;lz@*ia3(dvZr4tkkr)70X+?c<>*B zHh(L;znYLGooi}!x@CPhXELQ6h0mPn=J@eY;_2;LJdkGwGI$0xbD%sa@6k7&r3n^?OPHPNw@GCcty4 zRy*WevvZ+try$hV6Eptjq`ukHrgh61`i{ntd<(+y^x-(8amHr6tR zOp>?W3)za7QUDg__D(r-j*2_PakJ(7x^uUB1{^zFzC74)Xf9G2p#8$_?8W-a6r@0?q z=#2>&rhnVu5hya+<4YEuxSm)-0eT9xJ4pB3rn0Apm;P5mj<2>Zb@D?$7l%&z+kA(w^X zqXV`;@}2lV@AkI4CJMVW2e2x_5c-D_n*ZI~B|`oC!e{^cJNy!88%f)tFBDJ!WPI-I z3I!A?oU%ZVuDwwJwCQrxixT^xm?y_aWP1b>TGF(4h5`&CwiQvp+T|4zyyYbK^daF( z*JEAc>mYgr3NVGv?IEk^C!#>LJvqAuZ`sn=eo51{vtG>ncgla~$wZe()D&Dd=!?C- zK)`X?XB4m$P&@bcX$tt->ty5>Y)#avO_JlZ1o3}j<9LH}(?1j|!gM`%du*zr7>E?+ zkZC{xPloPXagSUJ9{a2a-+`7(mt3BrtF zRX+Y53Rouakq`(2WvIMT9M@3Nnkb``JsacU{%v$m_a?Wfks{wnksWx*{hO3!GVV(L zk8RG!fQKC1zx#XF0nb6pZ&GJ)*mQdXmwLTvBR}EG{!n z&eW#y(QtNYg#i(b0_u^ZP5S)s(C2cb?GYsBAqJx^!f*MqtcDFXY6xsu#?RC!6SaZz z^!y~Wq?d_OuA@5i0fm^SLEn8kr(v_|v2SD~H8#~*i- z!X)k3oRy@q2U?~SErWGS^D zNk6SNWJ~MBZUpQ#^h9+-_;(E3sj?$5g4QM;-grAEX}2=%>gxoPHE`IhbFLXQqrEPe zWt*+h|J2o_Js<}4Gi|LoEhLnbI4v5Bq-Ir|lh{r=YalBlY zT`hh27+zA$u4)Gs!N6>Kh4Zz{qF0!z-C|7uwf?F6hUoX@Ny1h1T{<S7K#+iFunsmRu4ds>Y^lquzePY=)Ly~| zH8%=NOLL)b8RQhzeCGC4R}2N<7RRFi$w?RzkGtu7hJHVB1A4Sm2%S#{L4MfCH(JvF zVktV#b0k)E#c8J$n9FW=?|x?O|5R!xcOMsX6#aeg#n2l;%YPlt+0&yEG$R?hHmD+obBe8b zC^xKb->ZfTt3#HFolrpM#y=B{RI{`+e|+s9WNUAwPOK3ECd%o_#env~Jkm-As4 zrzqlF)w${76b{+f08d4o8(hBYxy5e3&kR`*7zWYClpc6elp99@)Hxu(?^mZnZS$@6 zP4&B8)+D2Mb|^2?n{^fZEq1K$S@V&TU*C4IUWkCJ*@?vUbog-}IdwR(5@1K4tO1@s z$5lX3W^0B5Fv(33b|LSEG>rECp|<}Nqg=Biv|IKIvITs8eZ2VnH|+z7&j+D(?<(Mu zeUqM2^jG)S_JSxscJ>`JRsW}jXZ~-g;}B9BFLY_DyvZl*<$CkJnmr3=W@togvv~0E z-;N%6d7C}d+Q{Pm z<$iOx)4RmuOVGaDTi9`2uoFjh-^&M`7N$H{f7pp%(Nc+!D5m)+(U@5> z>=Q~eoRt`g>LfW2ZjU^-IVGr<>|HL2tr@Z74KB!Ww~!g;%Bdcy^rF_|M!UnJem@91 z!=Y;}?JJS+Fu(W1r4Gp(akiqvN-RGn72Mt_y^;&DxvjwBDp63aQbICetmaYJomL!n zqnjlFjr}tNMnr?_rNu!}*YOliR!y83$IIwRJ`(&gc3#F#?I)5N&GVj5YXsqE)x(Cy z(j@!J^)G{;;ulNch!acE@z^h*0Lt@-mc*iEox}&mu4cKJG1JO5#Pfc)07i8wZJ5O7 z%Ghq$;+Mg|y4ehi7lV#Ra_-~Aj2FX9c)ZVl2n%C1`UnqGM6*X}@}p^!JY%N?X#Fiz z78>zvOv`)IH--%_vZ42i)N?+2e&-E?jTZqPV^v{ssf(!}RkQSe$`;7kg}q=7B9k*i zOceGD=N(6%i5q2cC&%cQQ9}w5SqB zb_l`?Z3IvLdeP;A}d5`!+ zoB1Xf=Zvxmtw+exPLT|o9~j~0(ad{b@pqYBPw(_Qi-r}4fJB9f4N zF3ZWUH8E3tV^`;%dDXieW!XG6J4%%sCAB0pmw6aoW4-oxMpGM?Vdc~>ubw2umHAsR z8etEpm{+d^I$GURDqR@w0oTiuxz$aEX3 z?EPkzbi2tNJ@mGi?-LF>H9~71fmW=lL-Um3H%K!(EUtF}f=lU1G{_t2fo+LpRiz`W zWHuH#UBe1*qIsiDDA+k_Zl2YOh`<_YM$n;Z2UxNBYRf-OScXZC03fV)6PH551E%}} z+ATfOv`y{?a^H3{J2lPdXKxx`gK%kumxoMD?=3$LnA)e44nwsIc2T$M_q$;7*0%9p zO?NJ>F^Ptr@2`aGQz7WV3gp_!ensowmr^t(0p{WK80&{Un^D=x=zP4*fPC%UwEBaT z2>WcK*$$D@;ZzW7&k=eJT^Ln6yL9|Ex?m8lY(gb5{*=-zx4$0;=gp#r{53R zx)b)OtNP$e0gM0J-VF=Ju`@?~cWjA3x_}xxC2n;gE`(v$pqDq=mtGacVcpmk?fY%h z-jcO>O$+Qpa^I5X7_`_vp~(|8QBKuR4v2J$6ozTFC8lJtUeQ;`b4Slgm#<3c=!Ls3 zvx}*ny~FU-w3?-;57Dk3y&FuSY~&GMqo|4B0yxm?A`#v(Y zTQKOBfyQ}^LhRbprqItCnNEIT1!o_H+tB7H3q$OQHY4S->FSHybgokS)T00(CW@LQl=;;#gO;WI~QMatCF+|2v}JQJ+`qGUESqK zg>>O{km`})zrpERc3F1Nfi5tYh8n3jIFP)_NSBv><;I!jI~BT?;wP@-Q?yXuJNikM zwJvT%XH1qC-Tlo>;LK8qsdcb1wzl!JEncoKMnQ0m*CT?h_)LP-r*EVwvF%q9dBolo zBiV*BY66^m{Mlw!c644$xj!SpU$%$}h5Hwcl2?~aPbrtT-QQKvY9Sri@S$t0fisQn zCQgV+w3oxEZxd%u0Gd(*HyquUkBhvq)#6o#zE&hhWk$Fv-&WvMdm~|;zC8O?ZQlOk zHh;wB89lrXZw*0!0+{5ZP=LwO<+!v_!80%x`}tQwmgajcD4}fdRAEw+IzZ)`QEXMG z!ZbVmG2v6GdLhhmvmJONF-S7wgj7()MmE|*k_6%xCFr$BOQ^TA?^Z>RwpGrEW75PN z(FUu8($^`{C);3+8v7c8INB=i;U;nNFtwO-Dm(6NtN}py(rHh3s=e;fD>T+k=EwP) zIO16IWJ@dECmk=KtpfQHVg0pNOxHQe;r$Vbu$Lbxh!$3!>^2auTT}1hZ3^kSJ*<4t z#gh^GICavE7)Go;DL;zGYD9RYW?gI*5GV`oC&94O3TvLoKH5m%BH!MmVu!G4I?*v( zz7CzlLf4T;X5o|=9f`zI4u++Y|NzilC-tK130oRpHH+r71=(`J`wfaCPoU# zB8j{z=iTwvf+VwNA|)A5Ft_ZE%qeO(D*1{yk8qm%9|9b6!m%RP>4khJ7yhF+G%?c3`e zLF^A6B;nIl4H@dG5Y|FH`%<*q)zg6|(yAAO*T?sNNs^8h1w^H`o9=;3U9a0HNZ8*) zgd&+xfae?vkQ-Mu+H(bKJv4vSUZE!+j3a`DOi$oM*EF$QD?8Px*5g7>oa)67`!}Be zXCgZq?psJX>XOSXY-G4{Ny((C$ws<_#E-|U3eN_#9Q~rveU)Q2F>O>QPj4j6@(we{ zbu)kN;?A~oBz_n6et8$;rR&mcviaim{kixdw^!4eu5GW@X(!d-a%i7j%IAf zZg@E>TTA0gPre~qMkD>Oi@QhZ`FP6r(Vq=mc`vYU+@pf8E(|d>B~EwK6xFj1N$X&T zIuaG>@;A?>wqJ5JIEcp|>?A@2C49?X4&XLpzndFr)c(=|XWBi4V<|+>;;a#F>j>dP z8_;(A_7Y1n&tEOJ+6(yC*_B7%4&sb{J}ZL)JPsAtqSe?+^9L7f)wVSaGA2pmRjaEk z9zNh;QAi?a$m?EXxKf{Pll=-FRQ!~Y!^pv_*HEjum7O)}gXH?Mu-Tvef`jb_s@Ic~ zle3vIf;(T)1ExlA$m7z&3_8JQpid|G`wXBKAt<2QyP(frh|VIR=GC-?xZTW1mElO` z%`d{vkK8_dq0Ejp&P23~HN)t42*2&AD!WbGj88oja(6=389~?Mipy)o1}~@^L{NY+ zH{@7FS6^>%@foQ~%iO^Q7-t!GEvWi+ef7t6vF8;b8(6iad+(3ZEhFuf{#L}x5w1sv zA+AHKt{+7v^b|#yv%R;fitn`RB%b878$(P-_$gfsEh{&;GKr1ah;=M5poPyT)Unk{ zwjdJ*`t^n!Uw#_M1#@`0^eS$@!BPV}W15St8lc`5+-WxXAa93}@b)>_4P%y8%TEiX zy+0yk?w5QlZZIph?4_Zj8s%Rg9aVF{_Y7q+%oUma2p;_IZA`y(E<^!@>*UpQx4w>! zwDMWiX?J{xE2p%J)_1cbs54DUx&@vkFBg5q#7f8v7rAFAZBbm8K?87R9`8%c8nlkIrvjxLwbppd5^R#@liJG@cn zgq}iMQ?r<_4MEpacfX;^-D69Oz^ThxkAlx>+3ZvgGro{j{Qm5@uZOw_nBkl;a=-Y} zUT$gdB*ZCiy6{c^_j{zwhnbYz`riNtm{tl$MKPq!hpvA3?el^eB^w$Y&7b-$b_b=C zkF8sLDLH|qk}YHZzmnLRV@=DHkgKY$a|@(20QEsl07@kDqkZG#ZhJdxw$E^~bE5%BU^Gig_s@JW#YN3P#zWXl2h zHNNZIITs4p`2>A-9)b)IvqzAjfYY9h^&GcpLbEIEcGF|tDd>C+sE9g}Ron$>vm1gn zKCIP5dB)y}Jm>L*gu_nAifq>s1dcA$h$y0g!=l*wlD1=eWT7|?E)Xn}qJEE=-1M~O zz7w)J?^9-6C~?WP-y+W?#ZJ>cXmh66e_KZEa69b_xu~{p>Xg2nYEI#!n+4zJo#-bd zL5q|YJ_Uw5ny$;UM#Lqcwk3D-a;>}Fho+S5X&fdmq_AJUP?~l^7V65$-JP2G0~Y>* zhFYwlDmOcypywG~TOF%oGny?JyFD4%Dg1gsL#NuHf@e4KPV6bbcQq)j11WtghXQ12 zYmhggSLXJnn}VLk8OEkFa<(^4qIL23Hg>%W?pzt_!@nDo^=`Z7AWcswx>!76MEE{M zmEM-CHY)E@^M7u6K1t=8fdyBtYhz%`D(|BNyj6&8~k)9O;=7BOvzP z^HWvRFRFdJU56B4$-DzM>6SR3h(&&nedts*Ld9*kA-XeDsQbV@acm(L<71oZ!lzI_ zWW{jS?@Z2F*kQgxe^fK<-xp78XwDdIWPf<5oCl-1LB5V@t$k1BTlnvyf2vDI2Iq-m zFL^7CAGam0CCch6Giww)Z{Y#EFQ%(6CpfkFFE8+h&B6(lE^0r=Yuu>{=PM6asHp3$m#!3Ea#I|Vp6f8P3 z{A8NQx<>8uAsV~r?&@Xu8skD<74(kYw9%w{+-Qhb0#{~d5)Nlz?)93?0|j_B^fzUE zi4fTDXbQ|2o$T}ot2b7)J5Zh~e-Rr?<9Ry=dmKHRqcMk4sYoxO@Aqx|yGA=7Ya5@T zLu9~+WEnpP7vBKibs{6>r{bAqH<6fR{F+@izwbqlgYR zCtGT?_Ex4`E@?DZx53Ni||2X&0hp6wY;PY%n~+lN

    D_iunL%uxH{d@<@!#`p|m=qW@ZN~nQ%UyJ3e}KL~$+8UR%1K6+1T(-3q<62?EvL zimq{azKQCim3GN`Z-1I$LrcETP-+uZ#t@ZJdlx0?sq*nh7=F9>X@M(7O&`VWmx}l_ z0V5o}dFtmRPnIIFmZ^hZMm|tv6WD&Uo>-i9J3GfQ(Kf zPFD4{S;O1&7G_$aDi_n=Oc){tsmT*nEwyCyZ6mz9xDT9ym2CI5OCGZfMY&$L_6y!o0+Lc1t z7|e?TxOlj27fi45Q-{m44_1aEVKlR}L9qS1X{^(Jn^6ka)*jL5jp+?nS8g3bO$R%t zIoLTrodrR(LsuKCu%U2jAGY`XdTf~i%*uGJKZhZ(D8K;v(y@~VW`uOE9rsQVxl4b$ zdP!k8iBHu^O`TkWlZPlQ^eb2ry*5c{$88R7di*}3>#J?nn_Elv8&VzduPUx zAISR!JV;S1C&?DGf&;;(g=8)H+7Y6lLdH1NO!0@Deg1k6#GW2?6>Hj4PiC&-ynR-~ z2d+ZcS%mNM8)*f0+s{{%ud0S#e{#No-So+uC20u-&=lwB||iUBD|s0dJEuJTCDUxqXV)*8Jc7Q$`s3xo0dI{fz3p0blUvU7yKtOKW} zRY()Bod!vH%4cFBp7R?Va8CXiIdn`p> z1hW(`%s2GTrcc$z+&|Rot=IgN#0#+e>`p(YVDH0;DQWu;dT0tSmLeJt8PLDp{IM1x zRim#t$&w&k zPFUgFRRppI;TyK%tx*cEn7YlUqTT2^>#W%`(VjL?qX7CuuqGh8UvthlD>f;nmY#jy zGo>|SpP!S}~H>eN)}<^Jg)Nie<)l`E*nr zlZ!*K^vCh}Hh-_)PLWx+-pJ^ol7!Nkfa~A8NrKtYD6lk%WKA3vy_uHOnyy?*pYbJ? zSRV%mEKYLdhC1e&OIETy6Q&jG5IkZ~>4PzBHov5A%PrC>|LzSCBHb2 zXeEyu+6W!+@s`JdN&GCkwh|p*r^#{p_aqTJDnu6uKjalD$jyIE5^*lW)3lc1<9yu| z>G%Z&Ovce4U0`3IvLi>qyh&;oWUow*e*c3hbWk4gE62igh4d6!Q~1Zy=>L3-1O3kh z{q@i(Qo09}zO^27aGCyWBIDl#IZj3dG?l-e+*5A~hR=9*^M8@|7TMd;S9gb+!fS_% z%&RLUcMs{eaz9F6D zT<*T&NjOaaL}yYWOtbpjlwXUhzrEaeCOl@qQTKXi2#RI^`4>eSUax2}Nkbrpf-40{PKB^JIHvp+cVmpTuh_KC<}hwWaBBA$zdALxCmQtJb4@xf)vh{AsCr&jm* z^5}VJuUFs^m~a0+9d~Ubq`r^vd@OIQ@U*%@8liiq(CkD2_Pv5GO~g-%Z`SvXUEHkx zE&5YU0dCm7#-2j8vEQeY1eav)*}KukGH~Hh^Mn-u?W7P_{FMMXue%R>wn@O{0ipoR z$2xm&KmhD3ZCJ|!fzF{^qUs%k(N(Qjua59p7RYwzE)KW?rL&@}eYa4H4V&kdO~=D4 zEB4(_iQs|8f%i(tVzQ=t;zdu36B}ETd!kDX-3PB8_(gQo(0i_9<>dGR4ZdhU>`6O` z0wr1EuYJC0Nyrt|^i2=8DMvSRN~oC{fG}aM9=$!H*!Ybnce;x*h5ad3WK=3|8~Wg5ox#)HMki!3 zEn&B}#doVO$cv|-AR}~>H|jdWY){XbzbBq9?NpEk_pp8O+!cP-Y`eClGi*w{i7x6E zAn;OlXm@8>_QDnqx>u)ML8w*Rndbo7Wl7#@U%CRY3Bl|jvioto%8y5{RMpzu#@R)` zglVL!b(pa?WjQJ>5K)^PaiydOLqDismLnsWk?A=h3$EXXaZOc+=`k9hlRn zRsm zac^YOf^!M7eDD^MJVkuWrsZqViN-BitEf%zIC2qdl6ZGQXGY^gQ5Z59a7w0ek(j)p zjXc>o0$n@ZYh4j1Y=8bUUaJQ!1t#yJ2xDJ!Qy-XP5DiLpM%XUX+gqYpr{)CHSVUYR+}~f)w3JSIffO?aeOn8G2nkH5Ef= zvhIk}<=CjfAM6O5S0l~Hd%>6igUTbv=L&LO>Fv;F4PH=eTBqA0igWr0gbmA^WPAoc zV>*%>5f1|BcX}wge1h}N;r;%zIp)^YlP>WAmGv&`?+=y{ih83K2l;+r>&MEfx3zG^ z;3-k~ z#3l%qrR_3pEfS`=&kT2GKr(rEudUs`lNdg_f_L6P0dvl?a|zuATl0opBi7Ypw1I%e zFQG2ZdgCp1M14NKc6u6CsP|%b^AT%9$J*I2H#v4rlk#+RWsUrd*@D{5iPVcNmp3JB z%TY<-%e14QnEvsh^Y2a+S7Npw2B^FqWUvQYq`MA(il6QnC}nKbq#hFeaKSPeo-Lo^ zS#EDaiYPrjqz(JlnZHhoP|MP1v)tzVWOfiW0bQ!x+fEj>KSv%s$#YBjI(GpeZ+$4t z+Q$yz7Wp=6TR47IU6w$Nyz+q@OhOU&ss=V49j){!9bokQHfrd75%aj6yX>>x3CS5x^NxJymvh@%8u|? z{mbXZS055;G+D#5X9;Cjm|YHJT!QUy~QgVn~vbt>q@>nY?5gQVBBbY(4VLSsEdMFt}ctY%j3iDAbv z)^+K}4ULSGkjvGtn4!=*?C&;e>SL8NO!Q)!qg<~Sk=O0L6o~Gx`2Aw7I)d>*xpPW_ z1^GNGy~DII$l6@AUkGb2ajw_`dB^K!3V#T#maatNz|aQ^1mPJ`<=z8UUdeyS z+TcxHYDg}A38D;%0x&s20633nsgF5d-zO5&XsWIVxjMUDXm)hB*5-<$sKx%y`L6qT zgNBhp4(y_24-6jVJonx|ZG}RV%BcLut#&wHHO4mW8e?6q^e8773ja-fz0N9dWL7qLO;PzaL%%;_{Su zHG(?i;}sCoOb#|2+^xq_omu;Etc?P48l`idpHPH?O@C>i;(X8Y7${jxR(?K9eXH65 zX_Wvjtn)zkgEIj~+v4Jgim!3ib+C7~y7}jTx{vg;l7?v#n`s9mLKb)4%6Y(CepHUV zy+W@xyP#@65cjklNVW-qP;9=w3`^Yo&euWhCu%OEsPV9k;cZ5=2(2YLoTtj0PsMmL zAj8VC(aAxy6QapMbAHtpO`|`nk}LE)XR_BNwNrrkq|k;|q;Se7W`AH{=<*_7YwFS3 z^k+nyp;&|cgGG9xpM=fb0t1kPiFE+PzZsgfOKSUZW$ei7?tq`rgR>bF5I(ea4&VAz z_T%_O0)PDBbm?V&J`ut*-&$kK)8&hajigY;$PP@mW@MN#Rm$Asp?5tTXBZ*3$Q&l5 zbAaI*Hm)(cO4brs_(rCz_OeW^(^qJ3`y@m~sD51hO&nRl+3fHDwF});))CQ3vHU@y z3umE$|3bBd3q7&*@Kmcpz9C0i3=)Ib8@D(GwqOy+q_;e#vuD(x(La#qctq!g$!19( z;N5Sfl6}t8x*wEX{-d&zCeS@S@Py%&uW<6s$LFRQMqiq$^=%A-M64EljR5FGIO~1M zZ(N(Iu6~{Bia_~~s;gKeZJkXk(MbPKMRD7lfg9_WK6&}4SOfiWkX9u@*o6SkB*Gku z%y>L}oI7|hYb4O;hx`evf*!I0Jf~+Lb3m7Ek3HQ)FO`P!3F@>W^CA6Tkk!@HFw*U@>b<3 zoX=Dlx>F1XRT$I$TuA@~=;sFq_?JPDWAq<{gRNG{8Q@d7V2rmj@2rYIW{61l`=}Wa z2q<^VG{7Ci0oue5U!HzjY-<&*h}fGE7u>qw19RsvEAc&QW5>8evZS1x1o4{3l3zK^ zPQDU)@_cw>%~9ece@%1DmCCl*_u>B=|ufI%!yl3=5Q)(Fp&5d z&1JC2A32rx=4&WHyTy(ANbnm=e8tqULG+BPdJUt3emKc%n)*g;$g{L6Y>#k_-;opm zYj}0D*Xuk_`DVJ`%mwGOgDWk_0<&Emrw%jSIUDM#h1QC1PEXxAh{3o+G<`HAP2?EM zmkHT%ATfyWJ};Fb(jKW?UxNMUd0Fg(oRb%d$jLhU_*ONKv29X1x9c`Ix$0wcCbbNw zbpo(Y;h7E9UjL$mLU$G92!Dh&3g;$W3yYRL17m}*->i5HN#zWX`3^{sf zDH_)Jwwq$z{sUzeIX}}xuE@U7QCulGDu8_?auX6^Zk|5L=2`k#DXCTjMdjWTX;Q%u zG<_ixq5YN09lN?DH#g>@#^9<^-%HX^UivUukP~FJUY)TEhF^0sCzCd;Q_s5Q*jCeh zMXhC|Fk*Z~V~oa6?L+LrfZCC1g%yNWV)`} zL+D)j+p=>mK+SE*o=|_lcA$Wuqif6l71gTg)#8Kx%cu1FQCmU{Rwks4_dQkfl;eUf z-VQ^TVwN8@DDVrEPGBd8EQ|T6yjm zy56XpT&<#$WM0P}S8Xyy_)LIpBpa4sz8W@$P|LPFVl5nQI6K6LnOO(u&rLHd6P<=* z=Y1sz?ixd=X04ZJHMaM2ZD;W68#2auNPcWsnw}?I*0KMIP+KIl`?YU;TAEh;D>!TH zAd@klBG%I)GVsQS`3ZPrf=1LT6%%C>%vl=BQ*~W?dCypWU@62dM%;py#2|^cu)qV{ zo7lJmSp8v?H@O4Y;-*s_wQ$WNqlD?7pIM&NFIML9O0tC6{PVS{)zDuJ!GF|EjDNg1 zfPBEdestF!@hJ&vuwj!pesSRTAuwyK4Z2+E2SS729{98S{eS#T>R;d>)_|Q9w6oYSB5g(_l58^~2<9;B7t_gL`GSEC+Kz zF`o$SQ`fJAB^smK&LKh#qjn2bHBZVPFDQ(pAji5lB_GzN(4M{;mm2{ANaB2B zZOX6CS@=uCXfmnpuQ&oJy%sh=qtmpoteT^rFRP8Y)u?1Y+d`Fszs4ZS$${PknkQu>^qXZ;s3I+YhymKI0H z&_MWUMN0e~{ay!=%LXJ^p9L%UY5A3i|L{9ASZPrtq_gRnuVP~jc~FB7L03R9%rZ5u zBE3X#cii$!jB~-SW5Ei7C?D6cqx+SfyGaDi5N2fSLo_E{12I&=fR)lJ{XiA?ogYM2 z(51ch?~Ju49rJ!D;8DJ=${!7nzV2WgW@$;yYtXcMy4NAK5`M(-D@2IBIt&Imk z8I_`-AgDwo6bXTmE*+&s!JtU*g9;%;q$4d90jZ%#mw*Ze1VRhFDP8F$AicK$LR26K ziT617j(5g&*SEeuvXYavl6Cfb_Brpn_xn7*L#$dmn6o%cxuRk!9lSo84#wgn_l|*9 zRYBH|plKAmj5nZ<@G&W+3F^~?e@&3|LNCuY@5zD_jZdXe4}#pPVvMD#WuZkE4hDZ$ z{1yQsm+RWz>PIV+@&Gj@+aq4hw)vG;`CyxF7w^m{!Xnm9!i4JIT4S}Q!=c4jVitd9 zpBqYFC1}pij6^s|OXf~eci^AitJKSM6)=!rB&ix;MwyP)+HQH6@$l)5F?~@H7ii5H z-~1Nr8oZ~+6x^!J((iQmAl?d<_6C~-G+Z3j5Hv|-lyla15CKzY3J{IL^2yJUNl9CI zD}7@VWbtj<7{r!NKWQrCEmk;tpL3k**_<4U7fUn1e_UoHUFR}VZcSD$A6Yfcd0mdz zz}O0wuX=&50c|LvW&NXEDB`xl_J_Zl_~uNg0@Vs4zjGbYM2w zs4+795|O_XaET6@Al`%fGNuAX&cE@MrNq2WegyOw(X2Uzlg;McPAKaR!bVh(Z&Rxm zBT^3k(HGWOS{cEe*}PlQy34WCG-p4u%2V~RF(?9*Hg8-;dTkHn{mH^!LgehUgbjVY z=!AsI;DVJN%A@`E&>S(Qx4D5u#_;JCo`xg^?e;45sqdh)i1=w;O2~64EAZ3Mm*M{F z&a3*Dupj(YMA8OZWQV<<@j5+S|D!x1+fwx7<1T)R;Sp|maRQ`-En<0CiKu^(yt37y^FGaL{+hYG?KOE#2%ZU}G zf!4&zh~JE|oi7-G1M+hJaxD|A=p3v@W;o*mNjt~kxU13S3d1&>#d4P*R%xp2hf7}t z2iY^`0&J)MIrf;JV}Eykjy;UMFy}MySPrI%EEUd0OKMk{X+0KmXXc4%>t`t|-X2p1 z=2>L1(pu_>BL6(hh1|1Phj=}D)1g$P9G5A=Wi(#J@c`o2{~^|oJe)9F#9{BT!Ithh zoA2&gqseQ&xZt#@a5lQj@4aAvm|sH6*msa@CHSKZGu@)_9NcZ5mgi6QJHbgOIU;$v zQeRx7Acv9cDV}g47z9t6SWMZEM~Sab$pScm#<0dMBj5w%t_xWi+OId^U}z{`nDzyV zjG*YA`P3u?t7BIk{mPsJa+kWxWmGYi7{y)T3K<^E@>;ka`@x$R_czZs!JHawFia4L zY?krpMRBG~->LUFQN|c_Nn;aW7Py;9FBuM=p37j)4Id4rF*2^O-vE*A z-|WZ^*JJ>}=|gt@46d2c@NclQe{NO%7lQWR?-^6JB#sh&G7Q=txgWku=~$VLcKgcEOC1~jyxe58j38y+s122r z^1QKsEs>`gQ|it8W%>zDc*NJIVA7!mSHK^P2)V%P%Nhuc>mn%a9r!1%&_#erF_Ne? z;3O`NF1%-bf1m#iF#vN^00F+yO{BWNL^cRijz|r<+KL)Bkv29))g{}G9N89u%e7yK z2?_u%eT5XDJZfCGaSPM?tlVb}is$cxNbhT{PQI{`9#^u}z5L$ZpN-unK*?b!ra5gf!d;!L=POsk*f)%HR12bGo-LVr7N# zDHSh>x0yZao*`t6mS4VQqkR02t0VkZiZ^l7$=)4Q{UWmE-_A#kyy#;vZ+jNh+&;Dub^U~e znHl*kvqoZ6T=jvvy?^ze$#m1%kHTnC_jKvFnHlQP{EKHi5NGO9{sVXx8}4sd4K`an zH4KFvUY_zWsO&9S9-j1K8vFY*ru&y-kGM8$)$wtI(&CC)NKHFJ~zAv8L zP*+XL;pl#>`1C?4$a=gHS8hZP_Q)9`Z-xE4GAJIToB@;wxU*R zGLlO0tr)y{yDRGV4>x%_RF(&-YyCKsu6&9K^l|c=6Gj6BKiymNKBB1j+7|1Rv95h~ zr_E-O#&rrlwxVdh7pUBx-bEVnR@#q~?5qi||B@ z5-EUx)pR>yF8Ufwyy;uULfwgRFD$z8hjHOdv~7&^Ig!8s6_Hj-7bjw-Tm9WxI7B4*k~EpFU+{$8 zh4XYJ%Kh+0wyej1Xt7Qt#N~3wef|sBrU1br8ER#|!_=wLL7=xzZn!hnQ__p?01VN! zh8x&THgAt_M%g2>1TP_73inf~Ni+R4vJIZ2ZK`x7>D8!Q&mp`jU zAlKAdO*q_oL4skrDJaPtY)eJzgsw<00hM#ZN%L|}F zHCbJq>qp;rtbJn$91BqL2hx(wwrNGwy^#ZXo5k$9SMrIh6`Y&Rq*euO11N3?xN0nm zT>VI$ZeEKk8a8L9KL2X(%A0Dw@y?j$k>aDBn8(Qv;j7c<^o}q$Pd=FSJ**Px z3n!#!*T7r_CHluSkM5mT?a}h5<-h7bK)pciPTLL=#O`I<_`>fiUT37?RaH!vq{YKd zAT4=F#71_q)dO8qvh2RnLGwE0{A`NmX-}6UGRtZ8PI**pGhKXu7>&{+4y|wBkF!ZS zYjtrZj@#q+GeOy&05`JxhxoreRL^*gE-)~6rXpfh%tozWae-yV$BAuvi{C=Ym5

  1. <*C@26zaj+?5` z*#K2rT<$*)^MQq%oJSeurdNja7vLVRzNWgi^P;Q?gtwb`yCVv z^f(tBR_(LcEY&Hv+yTaho?Y-TQBeW!mEJ6mEGykwN*CPYMdC9pJmZmP^Go(+ zAtRnYO6V~@Nkw)x`ws}$f(})n+u;o(vTErKp|`;t7VoZ9U?j$x`n4;=_Z7JIvYkIZ zXBc(ub|d}`?Fz6l%}ZD0+k!>InvJ-yHvHW5u7-({mX?LRJHi5|a2|;^y=PV}bbp3a z|5r#uzl1~rLOM6{YucHVn0mbt=P*Y_&TdFeA{>*t!pzWt29bxDN6Vg(ZYg+Q7hWyXVNL1>)m0FtSR(ID~7ya?I9``wY!uUIy(hGt`|(U7P$Y4b(&%gXp&bGZXz~7wc>HrK>lcN2AY38x%=NVQ|9lZ{{qN!PA32W literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-5-Self-Attention Mechanism Process.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-5-Self-Attention Mechanism Process.jpg new file mode 100644 index 0000000000000000000000000000000000000000..d530bb0525c81dff3639fde2cedcb4b19b883698 GIT binary patch literal 78423 zcmeFZ1yCIAwl3UQAUFgF4nYFJ-8BTu1WSTj2oM~C6KsG4w*UbW+$Avh;O-LK9R?qC z2r@zMeEXa__uhT>srvuFZk@mGuH92>s%ECEr&o8s>s{-4p7q|(-LC<}FO^i405miJ zKtsKN`$a$jzO*bP_ZS614knfB|)$SZIISfPdf6&@nKv9$@1<#KlAH03`;{(J(O3F)^^PFfmbk z`=NdZFiEgTpYY2+c%)^9&E!lb5Ez$@!~DFmgZ%aQ35(!+m!OBZ6ptyXs9B$~u|MMw z5*85^6PHkUp{S&+qWbd9TWuX(J$(an3(F5y);6}TZtfnQUZ1=_e+do=4GWKmPxzLY zl>GfiN=|NGenDYT@z1L28fa}@eM4hsS9ecuU;n`1#N^cU%1mciAkISy?c*rdH`h@(w%Q!9ti_khN;xE_!=Gp%<$AbPpdG;TU{ZGFZ0DKHI z)Xl>n0c3&8$o)k>k zdw}J4ciRPS<2^90W>|3#6vf_--xOhF-UDCul2OiD@-UHI*;1fgFd!TneHzz=R6scZA^ykp$ z9tc{@yc1QvVKczKxoEiuI{tPNWib)Jj_kU`nCuSp_zuqzj9g|oNw;SBy>NLCVAaXq zI5tYKW2F7-eExNP{`LI)Ykd7H{`@N*{%gMYZ<$Z=AiuxsK3iEAe#m(*kRvOgwUtt+~7u8r66Ci#M(A@(AQV8DLpt82JH#!HhaOQi! z4b0JrSiMOYCA|mcIQH)WMPbLTyByg|<z~`uM)be~O0#VTRH!SK%TtnD z=F)ZdQUsL)y&EFBrE)i>16=PU?}7MFFw<)+WAJVPWDY!~e7Ae29lX)@hb@B@b&a)W z?;>oMLvLSzk7Bpvh;ky>f&YG6W1DT(EkhO>Ox#HGB%3?sep98tV2j)FFkS+>Y%fNi z21s6Urk^N>Z6S1nAy81HTabrU#M`L&)VPTFybH1ZvM*n=;y+@ujG>R4xEI*Ua3J=V z;1WvJ^ugJEf?d`F)(OL|HERv4on_vDpzawF@jf*;;=@UK_*M{X=U^K?su?27Wu!z% zDd46<3YHX|@71I-5q+CE&y!=r zW9hJ%GZ+%daQq=a63+Q!s`ZJNhDv?P%C$kzvvn|!N^P}78pfp@8zpl_qMxr;c5Cc| zvXu0uz2A*}wD5VI4`O(IzK955HmBIsc81!bcHCr{h1YFcOaV*-sA8O8&OTdb2&cA+ z@Kr3a(K24BmKWjrVpyj^tf@cZ_mCM1c#>)R>|d)13>psYe}_PFj3+>xSlaZPGdz(S zZTMaxFpwXwUeKp;yzeLdJ<6{!0F_IsEW(J0-%fbQGh_*jBHPNYT%SN>>aeydX4Q3> z=hqLZPOmQd?}x^NrrTdUGinL)cFdFtIONk?nrI!G2QJiXhnw%tH^l#7byv#nS)@1I`gBFT+5# z333{W_nm@{Pp+1vzn*}(d$=ZUV1T^xWZ$4+_J$`VdIswB6U2$$iw9$WaYFcth_+ z^DRJ*E#@2A@_JCDb@jUL7LCkD^#g3xrr(9){D;;hk6tNOEeiI-

    +w?UN8BN#AKQ(fDMB=<1cSD^-UJwI+`o`8nhRa z>|R^ZRi#89ZM^~ZoO-Go{S5-0yFXI=>rFb9Rf*Z665kF<(r3$)vfM7EsDb06GS3bf zc2s%P*2M>hSUa%tAV3ctI*bVkiJXne0WvMU4Bpwk;G1Vf`}C$M7l=E!>~ViL2QTaw z765`EGZjvjKu@cGF3dmJ($NP#T=Bj^;_kGqJ9**Vg^0->Pi5YyeG-8}mXWS{DTCdM z%U-#*t&pd zp{5a6e#u>ivg93(78#es9Un~NWS=C)<|Nvuxd3}{CePdvhJBu%A9k*N7fD?z7Tbe_ zFFG(YyHh6KM0tv{um7BBW174C`{OF#a37OItmxUMWPWivroX(FX0ESjimUp) zpqkzk>yk&#)<(82Hf+K&ohu=*l~OI~&0Z>bOgvx*OO+<1lP>(UW*7eSw>AZP7Zw|K zxmNp60a5l+0?}Rr`byZdGc|z=$)WGuH-9Vw6F0Q#d6wI6kGyqmQBR2?K&JDc9OiUz z_;%0Z%D2ul^Sxb<7E{hQ9b6?1e^;dTA-0a1r+nqKd3X+}UPp_DHWAb9X@HP*4yt_T zlHJU;TfM+X;RoZ%+T;HhCkDFFeYw^vyok(*9fU{;Yw&X3szt;KO`N@Ag;7c6TC2+^ zaq@%2*q_KZT~~*l8b5byMH3i12c&R+VHc>3NEczmA)R1&c0}kqZ`$f&)uCTs4vp zb2GM?8>nrRFwH_S0|$CM7TSW(==|c&9q0$o$IMK;w7(O{sf=(OhNN5OFR&J5+F<23 z`psoUF~zoPV#n>Jto*RKmK?>r9Aklvz?M2KXl^=3&=2=B6iw|S?KDVz1ykU9*^$>g z+Ps`tM&BP{JZ;KdTg8Xo%=XDms!gqbiBGlQx~{iJtg^{KU5=km@E$@A%Y{lfzjGrt zm4zE6dIAG$Ncu3Kz(;zb)^P`aHm@y5tJ+X4pYMZ0rDQeF`P1_*Mol{#=$?aIbC6{h z?L}ju;oK}CFYW>DuS45a7mH~wV&zJeCs>KSP#9-ASbg&Nu>ANnY}z79#T=clP@Lge z|DmzL9<{9vc;3#;R_HcWW{}C0)2U#|oD8kgYe;XhT*_h14rDmTmU^vN!1*Yo^OHTA z*j^d}IUS3!WT?CWrZK1)gACce93Xi|TmG#%aF%>h&r7*0#WcxYNLps8k1Zvdt}&=Q zNE9P#2Q#*}jDX>lu_pYpdJT1;alwIHs|_808&%(i9Ia`O6SuvutnkT5EJZn<>89li z-?h4H$;uSShR?9bJ<#2XAFGnF;-~LU<~7{(z_ND1w^L>Jdmq$W=;CBq?!-uecC25d z*@fNGY*&6QVtB%rH)~g5fF~h$ZXU!&cp_KS?QKDrF@-jaq?1F0#(c!4{^r-V7eBsn z?I5$Eeezqi3~X+T&|u!XByDF2PIdZKMu^Q^$atk|g_W3JIyvlQhyQws?^A|YXALCE zG7RqAtEO0NBjHhP?I0{k<1^&qx;0AaK3YcP4V-ctzMXbtLd1N%xY8b}cLR&Ty_MFt zdkIBU^3r4kV{G%!TYBz+fSS1-L_t7`{;A9WU0KSj0Y+wPxiUj9DO&vVVITU0EE{Tk zUc!U|1AV0(9&Xsf`s6?O~KCmpp)fQF@HO)O-(6_{!B8^u!dpxP8&}5qserg6}UUDxyL+@U(?f zXRIdOd->gT&A#E>*CXxvFquiX;6zT%NpG;3yOXO`?%d+2162eSQ=n#VVG$)Ap~ki~ z#z1yGH1)zJgS{YlD+K>#M0<_Z=-rdWSc6GKOLrYTgbwRnGD#aA{bon=pUUW8d)_H>tf zOmSA$m2!U=_~bC0KYGLBJv@CJUUR6RNsZMxN)Q|uMS0@!gO;BPJ4*^yIL!rt+yfux zUb8dHadB_0gQKcQSB(;ez136KAwgQ_c&Or8{> zc`gT&c6NE&vEoSUT@iCiO4|1UVm4(CkX8X3ZdW0#PmOpSRFpiI0#Rrg1@+4>h>!~B z`aEQ{KNzBx&w&TZU#VQ17R82r6j!42;g1Z5eG2B6hU{H2hS8f|_`QugJ)`4LIa|be zmw{6XiKhv*N)Zl!J~(_CiSMG};ys3)aU<6H?j=?|Hz+02ew0HjGXWTDpYxkJ@zrHAETYVQmag}j~ zwv&N1?chcfEJJH7vZAI0+r`NVVMzA<1(MM_$)cNQn{08uJVLYVb=mq2&+Tq_skU?=m#<%+xewX)C!zsezobB#k71rX$KHa4vh+UQnG zt@cNxdSPM7R8=_7^W}>>tzP^ef5KT9T1NO&fc^b2Wc8^RI?Ur9(8XFVonUyOKgo7d zwo+m%;a6Kyu(Wqs`Ou;4y>-6A&C9L_0W!u2x|&*g2?Q&oy&F&@ zkgXKf29v#tQH{{0$h4ahcE#t)?_O#Ai8a{b+)Ev$@p9E1-<)u4Mnxr|%cvI)Ba!TC zpR)i${=g`=_MT71t3fwRv#Aj9x+t{PwVNQfDGp8^u~S%PSKwA|N6+(^@9m$4nWL*u zX_o;tpj&XyV~PcrZI!d8RmEBS1wy}^@M9`f+XY$dIrJP*c* zeP1>RQlrD$b-{+72J!p{j!j|m%f$R|F~O+}S?#ol;rDh3q@DKE_O#F<1C{mcWbNL< ztjgCuLC+7-$uup!veeq_uJVe~!63e^a*iqkLBZ}>l0~G;^)H&JN+JF%DxL$8pMgu8 zv+nDX;t(0;#f6s&Y6VF3k&)fcU#tE^IsZYoOB@&4U7vd2GTAH(0~7B%r!A_LE=Ed9 zez?0Me4uQCCi7hwPEUL?wSWa#9?5m%6^1C8)|f68byl0>hDkETVBZ6#LSW;s&fE7z!SZTez{+g**aF9k4TJZviamhtzbV>t-!^#HJ--Lp@2^> z`Pq>6gT0nAQO^ZQZx~^4Jh|NH@qX6q!nhF4OlwP0@7xQ}QqL3R*d>xx*-UlYoq6uo z`HRu&o{Mtc!QZt6*MjLyjT=(1d`lhUQZ(#JGSoPEH0_>x4-gx&g%>u|#Jyuv3+UT~ zz&(@X9^P*AUsqqcrg{@jc0@3LU+YCUD_x+Qw`VxocMn z$XbSGvZmnstC_D{N^(sK@7(*uvIt(Xqgb>dqCe~lm$#p}WX~+8-Q52kP|6|%_km9} zq(CT0sigdtY`Tf)D(oIuW*Dq!{;xEFMyqyE(Sa)PU})(E?n;^m}8 z#7V8=_0mm5H3tV>DCxgYyT6?{L!PDcENTQVuh+_xDJdl3h-qs0;^UvgAY*8XWSalT zu>TVz+VH>MI5)IWn5M1e@O#Zk<7lYmR9<$K+|QL?+jH;sfA-iqWKbrQijtD{zhd)`U;+_hL8wW)Yi zy&gxNnuc^Cws+s$hCd7)inLh+{K_$^bS9#7+96&{`DPZ&Wud>k)5&*a=RMla7E$%O zEOKTxsI-EYcRKIX63jtv#f4k0vw0~ym;-tEM%VsP!;F@ek+I>z3(JNKc5`f+&5yaV zZ#0xuNsK84`1&oyF}-J)=@gkH`m^IcKiie>&;R$k8Dkg5J=XkNnKVF^r$Y(*%`mYB z!7olB(!5I1Wv@Ab%xp3=Rx&L)+O5!=Zw@|p9Z3vV0`NwOzpu8(ux0--MyzM@4;Qs$){O;EpB*vc@<|H8LF-w2iY@mWpdr@W~lCz@g+!{T)cidClyzU?Zg_jkk z9G}*z`L;K%BRpvj=`@OaoV-?zoCv*mwWrPR0b_hj6dYqrvt>FJA@w+yEwq`TP_Ob9 z-pN|;I=$ueWA2V+3|rm^JM5g7WZ74_`eZvAqFf;VG;P+2S-*6%!mqxAxvs?{#j)6| zR6nvs;)tf?eb0-EdP^2sk~kA*m;;^rXtaTQ?kb!@&bsW*A$Ulhjnr*P(nk~PkFW;0 zmu`=v|&N4-9CnAxzy}s^{8W^3LII6I(m+?4OpTguxkj5ix1E0w2 z*h4(O=lxQlT>c+ed*7Dv_wvipmzN^zi z5^139vlnrGd!zhQ_#OypHp^etw^6#|jC|VIIx%qSSf`TZh|MBKr=$jWXZ+Y$#k0Mt zm^g4Ru{l=v=xsnl%-WHJ+&s*ikzw@Gm2b171L+6RgGCcVub&(0O=G^Y@9JFp+-6lKmqy@xdO`5V(m+QR zkwPoNDfDaz1Zo;Kn4z)j;dK&}9bm+8W<1+_HRfx%zWPXwOG2iE;MX|8C^3EvcG(ON zcRbPTJ-G8SvMUqUa@yj{ys|T&$JgkHxj&vBl#<@F4lu(K$>RumVrr?e;bP?-=umq{ z#Nxr7fTZJI3a}?E%Yb~NJWZGihAb+05#2tTB!eG1}K>N!Z`ykwM&RV6{|{k}zG{k9%@LLZB}0zI&E#PlS1 z^uk!c;|9{dL1&G@+bK&bG)jhy)S1X>s`(IPua8sbn8BD;opwd4Y&(_!6a5UER!_OQ z&1xjtic6@Fy##Go%SO92zAd~~h8ZV*8!sznIdI6q$NX}e_O57ureyCb`OD;*?EH4h zujfJ+G1gSczpm9+efmR~@2@C9UNg#iboXSaw9 z`*`lO+r|@U+%id?5ye7}Yt)%oKg3|8l?sre3*f^#UjgocvSWP)(!A!Xsk8lZ0-?h0 zL$abL;R+AAgW|V37<~(|G%o8=6@Q$zu-bEJPSxfqRwb`Pvrj+X6yplW-IVZl*yYFK z*c|2#{x)d#xtS+4Ub=+SWT zzl*&-z%URWy!eBbpADapq0Q8uJZ~zpcceZw{N@viQ=OX@RO)5#)@QEXqrvd<^RHtv}E zXcyEI$coBo_9+4rL*(hLo^#Y)tcD949vUzZ8tn1gEW9n4beQwIB;^Pi3|7H;Acy(+ ztOQKl6-}e|!z@Mnvod3y)B3WvBn`Lmm)~8^?=8HR zzo8?xl5H3XDQOu78+z4F>Z^T4#z8zY{0*GDzP+%8}>ZeP?&Tx=klzG6n|g$tyep*ak+u))a%_Ua<6Qu%T?sZ=D^4Ijb0u#V?}?A zV)W(v_b0`(2z@|^pZf+;67ET<&KLcoH?MM2Hz~wM@ts57`|r{zi=&r$)tRPtUA=HM zFwE~c9Pai=s+4qi!<>fR?LnJxR%Gu>USQ*?qwDQ6>g4|5#bn9WQODMo?$}JWv~%nN zj@9rQOay)0g=|erDPl00;u0r2>UH*DEs<&AM{A6zhDeiCm~sfjSCuf?`H|rYS^R`3 z&b<6L3zucF9jTqHvd`;3e|iyiyv5SuZU^uIsXrXwPcAA^z%=h8WT;8Ro?E}Ayn|I? zRWV^b7x3Zzs+?c2wXbA^UNgaaR87WX;xW%b_6~nU>1}{ACcQOA5sGKVNJD}9@zLer zIf1TwK*?BF+34@`5@6@%zo;}y@cxV1B!J<+$fN&JZvB7yJ1n0e8a1VPNIB$PW8Hf^ zMJcnXuPy>abhduEj!Ojfy^^oT<;Y9juDN;U3~eqvE|S)gNDR9xi`e2n0%Qf9!{sx# z5Ju5Rs&Geq^Is3s4hd$d4t3bGaAytt68GCLJv=Un9Pcc+xy*h!-EQ@-T}O;gHM6cS ze?T}1b3f8(YA|}Q51CGl35XYGFqqE?5QK7LD9_k$)J)Y3fttMTf$2UWjt)uR23mh| zy1I}WPBs3TML+6JNs4!ej@*;xeBzj8eCl$WS2_?fG_0P){Ke<9A{D>heFZ&?=UXw1 ziE7yO>-jLW&OoS90=6B_-kUJp@m!!@1|{sLxnJ%Z3Ta|L`iaDya4_#0F0{Gwl`u~1 z>7MoI<)hFLkZ$JTbMfA8bSOzL945~ z-hRV#eaHEAI_5DySX_}aokiNXS-=B;uGl&YS3!C5@mqTSfdQ=j4fnT6H1L6&dcOj?cGAF)NsC)l%+C8m6AbsH>f zC6Rk4PI*aQIKy#C9tUrmTVcEhGJSZjZ@`^#HKxRgPu^9%R>mG-W5{=Jbibi4_+eLM zP`mF_^xAs#X9cEzNa^@_L`!FXZtmVtzKza>tpb^8dRuc}JXmTP&R(kY7dW8cY~$d$ zj$Ks6ADUS97^y0~~T$1Yn%iC3ptW`0+8zZyrZ(<|N&P$FAg_(D2mo&FJ&#a1|brdvu{FMbF!Hd(3rG!fFcH2!%FiqS021;kM7Gt&D)cm zp2^gqUxa*yit+0n&G)dx$YFgiVnOJekj3vwslxnZe6CCCx!JxbT!b!Z3i*U#(+Swp z(HLmpBW)*NPC=|e#Me-d_+8M%%eFa=Mz?AX)*ViVphhbTSjsFhT z5dOW&%QcDNR1121{Xh@AJM?pR7bEk(P{oz<{ueSj!0#VLoByWo$e?PbYU5nPE5&|J zR*r;H1DiO4yMj~KTKW6bj&&b+&YJd7q@nY+JmW{A>^_A<6OZOdn z`t9IQ@KqV#aYW(efqX9jd{kGnM^eCM?@0ap7yQU}8`nudo_4e_6;>nSs=fehshtSm zxCGl6kr~=7Eah=koA-4|r74L_IqA8yM;04B5(57nxkyBcH0(2I3v*j05Z>oj7yDf5 zTIxx0{eTbSooc*$i7@VjKqD;3WTFZ&8Hyk;PwS;Xxu0RnPOqxl zCVVyM(<~b*+af=BJX>Tmi5{gI6>fs@K7t!KbzU2LFK;|PRkZ!pgJ#s1&s&UaW6bQwuk{1~e=3w|vp-o8Ice;$_ymywxu8-5o{tfxPe zv!rHk0RpxWlW7%fIQ)U9u?E1o?H3b6p~>fUBtjJd6TP$)a-T3`a)lGt4sD%CykuWa zACZvyl(@$$YBbVh_tlYiT2n}z`syGIH6R^a&K(#=>{YlQo<~>8FP|o!@5oG{?WKGu zv@G2lH{JVPdQpvT9w(1h@zMMGv(A3E6xW?V1yYrw!3?6mu%M^FRjCL4o-nHD!RGsA z^U;V%j2PsY_(sXK$i~1(VmdbTZnCfs?Cj@G`K=6b+D9FaCauZ_1-zrl+i#>1c_|mU zKdoma@-raZVX911+>hRm(ttmCbE5M(_t?U(w(b%)mI()yH4f6%pS|zUN=#SU`qP5D zX}<86-fHMfHf(mHJi-ymPYi_yftCer*P3yU8Dt7wh(@xen?Duro$OKz-_@Dr!gZ6F zPOr5Zz6f*^y-+ztvqy}Y$|>QOyT$Xdu#ZUk*>+0wN9?+wSb!|o@qmE{oU|nE5{vq0GJQK?8g&*Q^=oO-^sO(k_vZ})P^Dnu)EBND=_*H*Z7V3{7xexM#7#+fePRm0uxZNvsrqN z=t3@=tJa%-8Vf;evb`>AjBmg3Q( z`(0ZFRIsUfUC+-(alBfj6lpJXYf+rbtbIjYEzhSn^JDh_DSh8f`V$$;Vu<)}Zcm91 znsDs^4kaAvp4Ss+4x%sW?EH9oS4Jot4!$zVX_R4^%svlitO#`&~!;2IqPke#TA=nGq<2RBpmGbj85;*DJK>4a6o{tI> z&&&n9frF|tX^1wBDBqJz6HA)imNVZVS|75gNfKKlTBnCopR;SOd*;6 zB)SSD$cL3G)2@ReD1xQ@N|8#dU7pGcwRJV8x}y>LRAn#h`S^4jOvkZ! z&9sa!0pOQ|^mMb_GK{tcJ$Wgvdnsuo${(u*!*kj52m;UxSz-#BIM^{FCVd%-^lzRf zbXKX^IxUP?V@3>bIl9b=FQVug=PfgNU@u>C8z*gryu`{~9Td zI9!~37j=I zZVfdl@1N%u@35eMlxNO&JJZBYfv@5pl^x#j+gZ6g(X~Y3*%V`Q;h%&kKlv!m`t`ie zU=E=i;x67SZwZqI^i=(vFbsz6d~ssD!uz{`mle`tWrI1QITdq+X zWr8aKQdjptr#}Y*J5+d1n5w2k^>|&y!5&h**%g9WJy7jz9@9lOY%s9{LTrlCGfbZq zxi{2oM!>sXlXAW3EyUf0%bbW)Qo&lD7TKoICqR1R;Ek5)r)=yz@AO4(1yI3{bwWtm zu=`S$_;69Hx~aJ}h~|wu#a6BU^R%1{^=VA}Q@a*~YzVJ=`9qs4wVK*J>6(iwNtIt@ zVLw3c0@S@WVh0;RoOseA(+>9BjL!F(?8kK zL6=;s!!@6ne`{ylk?t=AY2bM^b!i&yNzk=^_AFH8inVd|*0-!BRPADa&48KadS3oA zCqU|Mqf>@?!d+U@c%0Y%7^+eloOt8;c>t6@vNovcn@4-m%VA3B<=MTrvDo&=?sTu+ zNDt4VuEx`ejp(aDyVX?Yn~A2z{|GbxOW(op%(-ffg9vE%Y(MfWj3S=%ed{}|Wd&lI zy|k+j0j$&KJ_LB6`Y9;qx4N1@PgUlN3;Mh>z`b0is;UjqTZ|nhCGU7~WkIlQ_}%*LiU0zR_y<9V?Q#&)b56!1bq1N_xwnhQ*{lm59-^&X5<}^`y=*-m{kLrev*^ z+cu5))p~X_MbxnIsj@9;7P&cAqBh18JL-{_612#0Rpn)d4e&!luhFhIdkGGzeaGAd zvZsUcW4A;K2Y${*$(>G*jfQFOh-O_Ws`Ebxk-EnVZ%`fS7B;T^4ze3RCwM;ja&o7w zHJuR}h&RA|rYSqBem>yk)7AABA=N52b%wC{B=nw>)5A!WOpj5j?QA-*S2p;gxU_h} zrGXWTY5nBlxu`EFC*4Ucggpx6?%;{F+#GL(gXpJ`@-8w%PCJnzZO&C0r-hm_E9@@x zMUvlrRQB3r=nkzvobp+ng$9qEO*fu?nNXDyK|D9LiQA0R1U~w?PF5BRr+?^^o7kRm zr|K3NCsf_6Jjn87tH?`BEPZ8^m_oNs^tdoAH)HT(^?>>?iQ=MXl*31Y0j4l_4Iij`&Lx1VhQ#y(Ek7LhQd!8|T8u33?94*(7D?9hG^^R+SE}^C-^S>ty z|7jHoF0BHyyf>0E;owJw7v?Jv=v9@)mxSPtK4z7FrXCyIRmw&|rDauPs2t9QC4a?> zcK%7`O`v`(>xM-oYp{giiDU?C7A>+7_?iPJjp#!f{;XU3XqQlR@)RYm!@V?^C_kr^ zt#kJ~boPAsKDB~eNd?<+8bB`0V>NS44hc4mE0*Q){SW8t{DkiJ*2Bfta`@785$dg zTh?4=DoLe|sY`Llbu6X{9J|-fUAkGdyaV@z`3LhyESBmAfxR1xQr?(Hu=ceD@J&JL09@9zO32Bzk^-6d%btCPN zx4Hn2sTy|hAEr@cr{pRn+a!J!!NE%k93R=-f=RE-t{cWl_f!k}UO(V-df0F6@94=VXy zoDm|Z+X~Z((d3-9Mi#VGcYz9hn&;bhOSL9_2HL0qaXL6BH+Hvo&vXX@*Cxe3oi#6M z)ok(}@Ur0HGfw|T5;#WG8z5M-=g$Pc!lY-dnRYw(NqkJa3$d1}rrLb=J-s91axbOO z*$PJGFETHC(MM^DhXv=mY`L~yhvAU`j zhQVLx`RVx>XsyH`e#%|mOa3Ag?#oYop1ltmIuPabJ)$i*>6t1Lca($$2ttGzD3Bwl zWT$>yIMtZuRYjwZ3Wk38%KGH7AO*ASPGF23#}14kEK_b`7z(Z02QMnCHia@YdIF~p zH=jvrlpD-N$|1nPP&zGN0io1kQO$b0yqGKH=)-()%ccBM!wxNV#O(U?EhHv{;IR3! z!p?MIA!;;{lpp(}9NIZxMaPaIX)=4Gy}p0Tt2*PWVNej}UY z&78^+g_2-4=`bm6Q*^J_*K!Rr?ga8&+?G?@sxn=LJN*IdeN-4+-WgbchueSXqFyEG zt%c_$DaHs>eQ*tsDL|1HF1n}=6{#m?vh%%_=obYXbg@<~pnAIEeefZJA<^yW#y@TLe9Yj|Ky-h$w7 z|J!jjEsIq1RV+qiAsJp<$eylhCvuJX9E!Wh;=<1knbjtqoVqQ`%#6q$Ep}GMj=iPc z2|EK5%-@#lO)*wSB7J-Rq+M~u=v@QW7^!#t$lwxiAF4S(siN`5oFm1#WP$Mjpv;hN}HF!m_xtiAs{izuDp)r_xr z>D&rZ0yLjLwbbS0&XY__@0PAwr!&Oy?i@{ye){m@9v~0~l@l2*!CM&{l6y`vB?v$s zI?HW4Urj;;$E#;g$#)3~76yJQ7fZ<><%IP-s*^EDNLt)W!@pY7ZUhiM-`z8*;Yn8x z+YpN2lHFgd^^+$3>Io`DLlUHGQht1V9vLkqo_8qq6%DF>yu`kHh?1MY_VW?oFYgg7 ze8vVN_U=tgCcU@V!&G%)3mQ^o=@;hpR=QvTjV9#_g5&U?91$>6k((E!ox8MF3BK!q4RL#=_&yCrgB$|Ed|~z zzCLtJYWvlzut#K@JnVSlEP2zAaAHt$&`&mWf9zx@l($*+r#vi^Z1KAea~o*FG|pZU z#D4tj?2e&a<2l5pb79EU`|ZMvVKAANLF#`ZC{U2$KbNrR@AH%Y{JChSK4^v_1!lb- z?*F^L3&cuf%oe<$Yn(Ot-Y&`MdX3~yN;zsZ6Q@XFbs-H*qb?e^ z`WZ)xIo=uJrJb`ZLRfB!7X)TwmB#c>c7$?SP05-VAT$uAMZO^(LVHWko+R z6fcXW5xQ;_lA@id4O(_wSAWq6SwtWQ*?jN1 zLV4S^{D$@J*|gZ$4bdlBt3c;c^`vaSo63gKJ!AKnRPzEk#>oQ+yTn2}=| z_=xJOHj$bBGx1Bz%6C zxPYQJxrAC|Uw$5#EafhR-EB34C~pEv+;mV`xlOj0WjxLG@N7KzXI0%B?(V`0z1a*u zb?g?vX+|_mNCJADz%Z(DP->JGS$F%^Ze8tJTa4z@);Pc9 zBgIVLI71nR6Pe)k`IMuTZ<8k;*w~*Ky-_a=2w=acOi*rCO3`!UP~+9i zJ1Dd+r1#lcU0w2U@6J2D25pu!w_YrFiKJY+yg87c&y$!1M*JA!+VM1tHF{|cTL7x4 zNj;Kr^^MxsQL4||~=mFD}Tj<&!N2WM)DkC=g7 zF)WA`ep;{15MCyDM^a7=Z{DXDFr-X9Hr%a^q?d^wXlQ$xw7BjR9B)}a#qB5@nSBp% zpBpAOu$`FKon>Jg-)NnrJxaAxlRpQQjoEx{8Dhwm(>*VeUFSfAl=Mcf6+7fQH940q zb5UM(H)a0OHMg;}DQst~f`%-ROo*A|;vosOTPm zpY3A`8u6o6fACu^d0!=4*=z-COEsEO4y~EF2fq6!KivIA2a5Z zfqETGGB&J^S_k4}V`I_iwD2To#Z)X|J9ueCTirPWn+$D!T-%*nX;!r&>FPp(u11cE z_K5?EI&GKo&X^{ial4$#n%rnq9V;30`6Oh&GCOekk{Nw=^2ut>aHu8wC1( z#FnzHMbHAR@!9$@DyH>S0^LOgz2OkA$5Udq8w8Q+@UwJ!@V!bpaq|fstAM$OJmZ#g zKZ_=|v!Y9efKLf1{lU@96&!sbzojNl2^sUk^Ur5|S7YP9$@Wphpy}== zph@vf^|XEY(b1uU73t82b+!e9VhrO{P%p+`qx-Z2?_Z0kR@em}Ls&YU70ONwRsHOhimbsr>E=g~&S z#XWJi2jS&nOFT^fO&vOaBG)J`_-26=f2 z-uA?fv54d+o_UX69#xX46VN-_00HOKLUNA?V8fSnZ9f*F0~F0tusjO@%AsIZ^`kBd{rs&*=EJf)qo`;2GO5I@w}U$J+OmWcxKve`T3Aq zz^LCRCp(;dmeVh$zfyp8E+=6-A~HLF&Kb972SPj5bs?+*y{A}&i$6LmF=(njN%>5- zk;A{j+0{cvsTuB z`jsiRdlp5f&78T_;G}Tb$++DzR!{Lul2V7qy`l`>QSKja{3n{P%faW1q|?^}K9k zWoBiBJY)eUE8Lq!goX4gG~nbM3_ z=7vg)39I1=sAym0T6b197y*RLOf2-GVdXdLoRq7;F`tMCUZ;WSztUPcd z1oUG7dT7szuTpJeV}iy>L8fe2`PiWc^oc``?BR&N7814EiDP1)ttpAyT)FA1F^4iK z`Jd$v7B$e~1@8ehR#_&Z?zlZv<%&0Rqjo(?ZKrscnt2e-okHX5t#S|eYy84Ja9ifo z7%z<tgfJXA(2W_TCmK_9)rFMzew;eHf5Kl(7fu~?XxEzIZMWC zP-~LZJ|9(?T&oUn*Gs^PEuk;<8&R_!|62w{8{K!4*u-#+K8#uw=QSjn=l30j{B(2X z8N)$l^{1xLt5H#pB`az15q^VY@2AONumjlB%t?5;?kYSl>Tzt^O<|p|f50=_&ObE` zvuZIbpjzX~$oLoLQlxX(jQDAudp)iuq$loupUd+iX1p9rc_#h3`U6>G-tnJ0Q1WMd zZ0lNtc%YbzXBz~fuleTWzn5DDam6Al&AT_lQ);p+}rDlX4H}2VVCJro;Q^G2A2ze-`>2~x4t9PiK9*LaT8L59}tRj%_{ zD_JdQ*$7h5@-CW}M*n+A=Per5)vUYG#gDagJhi&2sCmfJjJie8_M!Hc4x9LLV1Ikx zn6ARVC7I4LJg86*xF&mb-Ek66-??Z}&#=eG^{&w^3OfPrb4G@iyiUjS8)R&6qOj#xsFShH@Lgng z(Q#%i%}t8l)|A5fGJ|pQvFeu&y8I!BrWE;xjVRY^C^Jp?uEz5I00hpCk$I94>A5~c z<@SS`%Kn}2f_n2yxQohHw4nBT;4#8oYDOpDp=8%YtbX>uEgWhG+7~)(~*<8@s~U8ctIZ8;!!{=J8yC`uGR9yYeIdgS@u@itAhR zMVnv&0t9zR@C0|a5IkrCG!6lRyESgX9RdUh?$$Vs6C6TtcMnb)Z!~wGnLF>!`Old- zw_eq|QV*)DyB4+Sz1H4)t#5tbPo!RKHQf>@*=;<~j;TM6!=@_p+s}H9EWzKM13+kA zTOhv=NQM7rEEhrE!+IKdo4-hvT+ivTvPUyff08g8u3GL$I<@dW<3XsQ)T^s48R)G_ z|5{v@5!IpdQ6V*RBO^rOBf_rSsXUU^Q~GL8_v}NP3gr{w=-7+KYprX`F+0)sv^BDp z%bd@IgJZWQ0~qr1nwoxCeH6R7i6u|&ot7U;csD8kx1De) z63P=CsXfd8QQ8TTL`e#G&Eg%-zOHjz6lGDLy){(=_xum2Jg%DKgmtpt@tr!awV zbwRqXv6P1O5GYq3oXc+%p2(ZwG~Y&M{TD1FC~hA$bH_BFg6OLa1w!i=xUd7#73!?8 zn;+gB_C-lnEH(Uzts*~IhHV`<9!Zz!wXajXvO}1Ru~&LLlfd#`sSXe1^I7Ndn1bOor7+jf06bwj8ps$T0{iqUU!{iPAwg-OZrI`Wj zfr+0y){h;J(O9=xWf9d*=1aKy8lF4RJvR3x;?`K_#cr0ZS!kae^^Eb(`q#4dnGs3y5A@^?U-Yi=pQlc8y3cXfvX!GhP3tO&fvkX*Z3ihtcGk&7 zJcW+}dVZkMFbsRRlD`jeyHG=M)a$Xq>~VPJ*xTtwr&cB&@Zrlpz^f8!;YQwTT5K5b zF{KMMw>J5|9{oXW*8Jr{lRB%o`W_X;5h-0T$&?(XPZA$wvzzjn5^#)KLrz+{ zsKlnzuAXx(_f>t@tQ4g%=Pchjc#Ah4JlR(Dt|PXmmqfTYa@ISJc%)IpMt*C#l)zQV zRW9dFGMQ&dCI zSuqvQjfTM8rW8SwXnRk3|EP_de$5c>?ff@|y+WzM9t zJ}$BwTv8zsg)YHg?cnFKQiFU>iHR-j+#1?aGKBA${|%e+@8AE*LHu8*_=B_~&H)1u zceQnhH(R}0SLq)soQPy143_J-JfG^XBLyanxEUgKj&txveZ3H4x_eo9^31~>`)SNZ zDzZqx`f059v==mg^O&I5gP^ORqd9L`w&R`F7^U;#px}#>{vG?~p+2>7*ERfoVsZLY zihVAH%7eO!2`eS%9NH4hvPs{S`5~r4kjLfrqF;#*P0_vdRM?MM0c;P;jybF^$d9Oo z#g^wv7AUEl;TD!N1>u}_(Q+&abLi1?nUKydug1z#0;P`H5Iy;3aH^)=0q4u4pDGwb zV=72F&pOeP;0I&9i(cdJj!+(7e_*Ily&I?$&?}1jvuk+jAvcTF)p#$A#_c!*pFA| zEGtti)G> z7J=i>tInP30X-=n)y;N2eKbJ^i%Aw4O^3Mf~CC{E+&22W45qj;@& zm0~OuC$}c~D8eh>NbxlEq7Dq|C)D2s>VG3Q?A#QQh#>xLgLMFVuUi5ORZ}Mg`LrJg zC_<6M8}`I`CitlI6vuOGVCQBH)5rFV6f^GqVZTHZzIowTA{S0yV)-&DwM^SD*|paz zvwrq>lU4Q{zLz=0F3KPYhaR~~<%!RE%U59;_7PB1#uVT z|Hk?y!l4Q9Af8ZTMhh))Ue=(q@u+x^G zQz^CqCBG$MUQvH~mrwaDa4|qc$>4pz)?&1~;fqP$8$0$qAA@~6I<{8SQFV2OFJB6sFdx4tF^ylr zO8c0ol*$q!v-Az*xzcY9P2Q+9Tv{*apiZB%FPr7*esTW%{z%Vq8E18?9prxNyJ;GOgpae^G}HdRFl5*ei%(KZc!2N@zx=d@=P;`eNSv0h!KR4|ReevXWu3k6;VK`#Wk zsv*w%b;H7VMKVRY;qAXlDp^M8gclbF5(!(~hLHJ;WDp^xHZVU%cz=VHaq6rV{vbQ= zepL6C16`KOIqqC!`6*Ucdf!0!y@Xg+_~P#{R<>S_!piZ_Xk07~4V7+W?~==c2aUiaDfWo#v_ zl!c$svR%!dq`ybJ9KwAZ-12>7;Mbw1^`Y{7GM^GZnKpA}1s7YN3m|MZ_ryu}IQxo9 zXD%|8Pe#%*FbG6(^!;(pub@0+9NJ zaQq4+MUUEj8U`=6sxF}CU52cUr^TUc4H=f{G@M^j zh=sVC4tUn?u_4>p*q5r<*0&YN-`)LQHMFD^*0y@r;Dt$auV+c#i#Hs0JQ=ru{|+&( zZ>_7TYjL;w6x1glL3%~T!~`x~}HD8xZS4A>SS7 z^&S!xD_-d{vAa6{cml4UFZwz_lUVS0B`b!8<|w@Vu>cXaQR$-ASvTD#fZwoOtyHv1 zKwGrrDG`^*Q@_3LG9u8Qg?6Wsy|K@G%((PH`T3EsU6-a_xMIi);&smNbiANO|K}XM z?TTBfr;7WYdO{p8S|*FQNOeF z=f`?V)Sv6LVZftk$Nj?tro)HvO>C+dyPCN>8jev;qHEu5dirKfr*R7VA*BCX%KOo} z0kVJy0g~6f#H2?JX$1qDRZkdm^A`}Glt2AYDz&I?4*Dhd9mm5q9^hqU<@16Lt0D!k z^DRkQ++#mAPeZqBQ-;=XDGzW5DT-Ht*e?BqS72s)GSAnYLf-9(0_2!1osVqYyqSq~ z#Pcd+-OfhU2%~LcK&)4|{x-H?q9r7Ju=4v5Y5+YBk741zY+=a^85(Eo?*&Tq_B<49 zN{29r&jkmy9>1AcG`ftk4Dpr+;{@6iBCKc6hWm&;80?RB8N$-#epBzXaC2@;yzgIA zyv97srdZLj_J4k{!qekYVaOHqiryyhvE*%;-cU9TM1b_9dCc*d3gzHvflxwzku`CP ze^7DjDJAS_Y3K1D&?3vs*t0TWLNP*iDHM;2xKMwb`f-lSN~NBwLn{yFB&CMVXeGWj zK?!en$uSfT_2u*__(@2k2v!22bU zK}h1F`oxz%Ag*PtV`D^X-=ak@?<+oK73lzXY0QV4Y|7Vhhl>HChsrf~{^3yGF(PFr zyUiuvDk~3)5V;8d*tKZm^LBZrq%1VF%AFm+lHpRJO>E;X3^v8KG(5uxxs3yxh+$*t+*M#Eb;gqP6pqzxG|OETcw-+1_%D zXK`gwXg5yqu7x}jp87R7@SWs~PO$w*bL3DjumUSe#DOR+I(QQLEQGCfFS$z40(}NPMbx3jXP9fiyJ?SX5%K|{4XGr6ik&~ON2S2c) zw6z;k_#kQGhcC%cygwF|M}N>vaEgqq1!~c0KrHJtqK5fbRh7ap?Zp( zH+H@)`*oQIg#xEq8NN72N9wcJi!p^0eu z=~DEr=*K)*Yh#+BKGr-&1~wlj=+z)x-YHSN9zOapyJ*;BL*ZOD*s2+RBFRL(D<=6J z{p;hHAA>i5Trsi~u#9S`( zF5#P5(!3g4CgB3Te7+UE+gtS%0JY?mq^4s{tow!9H)f@MT4*2*C2&qW4K09m4&#(l zcG0{crhtpe1^mUTF{$}eaREW|4QnH zg{9Z)+oDRC*hsw7Ks&Z5fuG9^H7`5vur*F8)>%paYV{x*229-BgV}U_{^DK zviUjDB;OIlx=2#z%6tL{vbpmMItRLKc0%gpBOcuOqYs0l+zLl5GApLC9>yj43}4p) z*$te$@&%)X12xYp0Y?Mr#11ojbp2a1@-UqlyLvX_Yt1&h#;f@Hu>(o6@7H4JvjKw2 z+VWFvI;VK9G&S9evPpRD(-Ue)V$)?>eEu4Qb-T;jET;nBJqEgni6g{;by3(KLKl?3j!k^tlSO4fYm@J3im_(mIIzpv`4exTEKybt1uxo_O7kN`c=xyi zW8B^)i9H)RyEMi{&-EXvW&Ew}|2EJL=R)b+d*?dp(Xc{(w%dchSTU>JlG_Mb z+`M{`o7{2rBrdg4Q1JVxO>r5?B6V_^zKqyohLs!diwXzc*E>=NXmX065jZrrcnx8y z0DB95*T_6<&KtOm7YIw33G3jrj_cW$KWgZW(Zq8r&mocp-Gn#{<$;yUWJ-qYV?@>X zTFgn1NTXxZ=wd%_E!MP2@h7dTVp*qmWEL(gy?cJ7k1}BN9fOJ`!mC@xSkl0kDhc5o-iQmCf{_xco$hJe-6XDA9<~#SnRp3;Ufzpb_I-kAEzaAtj27cX zl1#sIW=J>YmJPP_K+fkHy6hME)o{|rS75ZZY_K}7w;~F!z^8}jm(yl`e=aAwe*Vog z`orR<;QYNJ=|U@herkKsvDzhiz0)ZX$uV6}@fNnP56_0OwANAw*t8dKOWdr&J#T9z zya>V~>3|JU98gJyt%Pu|$g3!g?z$v;-}gR(w5Z$vfQUGHV`!v|(ud{L%htTeJi*WJ zc3rl{F;Di8V-WzNoF3LIaS1qT023aYXMNa1-=pRaA5EZSWUxtddnxf#ThnUf3(e(- ze=(SK+4@HvL4vyoazUfbJ&{t4QhI2u}Sa z0)7Tq*ijmjFW0qCHTcpBjwPMz_0gx)aej*CM3t^WT$V!X*=TQXIG`893MUEftdrdp zT`uw5a_`R7l{ZFSADrB79{ivPZI5`FX&$54!XYUuv^yu(&Ww!JjW=Sk>42}ZvLu=6 zK(EcgRw}NP|4I`{{sxS)3w8N+Rn`H>Sw+V6E;0oZ{DLKP);Dl19i0jchAnIZC98lr z69=o}y;%(`8*gtBGt{+V$w&}oQk=^7j_nnv)R*xim3>_Daq^dP?W`95nHnmW?4TlpSE?+?A zPh7a+xKWIYN8oY9p#49Mi2uS7Kl~56E0UA|e()bl1)4zqv9_QZ^4{OYX?1SL_QO3b zY@?-qJH~-7A!HU}$#TxBPXc&LDvr^$qm0IF-7G&s@>)a8KlqqS9XV-le*$MF2sELT z?NY=QSt3nqXC45a266=}^SY}De8{iPjKM~uV|fbiEy2P40VoS}+>J_49hQ7)hkcPo z1W#y~cKJIswa+*}1Znu;t?K#+N89K8Ve(1e*oTLm4lq4xB>hIf>HD?At z`EYerraN0@m$L`+s$7p>((N~om*t+P4hrCyP}fsQ)vD2#%S_C18`H;2(D;iYBT;tw z0g13H@tN-alo4+#1_8Ys8lP#lj8K2$K@{E(qlKT%~{ zR5xQ$D-eTZo?xoBB+p)qvO5_dVEON_6%?yC>&48a|MTGu2W)+|d_CHZ;?ggxc0JE3 zoPxXJW1S($t91avKnG&)uxhQ~Te?is?&R&u((3E@-R}*Qce#OROc_dG2s{I+o3e2^ zRU_v4B{Nm+&)O=U`kptR?x8xSHm;%DveO4yz2KI+ZYz(9Af$QoqZb)5b&n4J-J?4Y zI;Yb)CEbzBGltLh^J^&Wy!;u(H>9<1NvaF23!Kd-uG(n;!LAZ+@qqLPbY|`F$R3B# z?0EJElq2z|to#SmH+990mA>}ZHKI*m;~J0>!a7)eTHW3;CAt(rZaj7sp)Hs;?MR9f zpbZ>}3G2$fKnE6^(oRGR5#vnvZjrj7#b4SbHnafQE&v5}!zTdLpyg`f^)G&~`~=42 z@7wQRuc7&m)BFK376EYXJ>Z%<+X(tTaV0k=wr zJ|>pX6X_s%QW{;A41+hjl|KCSSq9{Xy|9Tdou{Z>?Zzt_)bYjRFVG)RY#OR~Xs|>{ z(a>W4d)NQIli9!}T0s;8i&rPGcEH+tTJ|sKAdpZ6h{%un@=$c64n_|iM9L%k$HBn4(3U z8anw0G^iAEcPi)NM8|a3MG7c4N*Bx?v3tam!#;;XIeLEn&3+3V zi`P(dwCbTP){e{7ee1MsK=_pfH`+&IG0-?{g{nQh#E z*1BodgG$AXwsFGE(g1ZfLR*vwKK9|m$t~4g(qpwuB|b;VehL{z9S}o6BebnUh6G86 z4kamZdpT~%?}1s>1faF#kk=I{ze4m?y)kpa-*^960zO7}aYl8p5vQ!1y~Jj``lT?b z%)WKZ89omhi8iUlk((mQghCj_^bC}($H3F|nbJddqO}Xcfvlc`&&cElk1PRK1tm|9 zhnQ28r4{*Fxyr=?wzkadgIjNl1{=sg_XNVbN7t@BHrEf99tyL$D&KATl)Su6Ot$U| z>T`9x=bXr$xMbrrIAFucccGe|6#X&j84;Y>r>fB`H1>Y71>2+ND(hwFC`&V=OnqzY ziNvb_hVgCWc*y0WD^kzc#vv{`P%D95F)bP!$S+Gn;#}y19BWRVEk)eat->Cehv_A- z-I{6L<-oFJDo=0b+U=~xY3gVAGMPSPDWv#vZ9_CkQ9mBmpA6EU?#YPsl2gGVcv;$M zPk$+S7d}14K>9u?ivRR@Uy|CZg{U*)k!fwgI!S-1xG#Tka5h3DPjR~%wJ}~rXzQK@ z`TZnY=LYKOWBhLyhO996vEuw?p)(g!5tr;VSygn}cIV+9cX#oZO|y9^E-qpfS``&N z#`5PuXgpb-zse+n>YMHErBULXPY=gB_DE=6Es|`1-wF`kO$Ti@i1CZ)8KENH^lu)F zy_?A_Y%^iWZ^|7^voRR;FG0o+Sa_CPyB_j)hJxt$^!))8dyl$s^-SQH?l^ z0Z;(d7WhtMoLi|i4BLu1efCjdVA0rQ$|eLE!U(ym=uoL*?Vu&Lrc@5_uc*qD z4^dT%%mJyo%m#2kYOfBaWWg1m&e*HiY3n0xEq-uwYJB$!OcXoSCvMnq8*j+uk)7as zP!u_KZ6irEFyyT;!D=o5J?u_?(V%4iH&HU(IK+!<8&^^-$JD+?hK@8L0h z+rP_=mUUV2c``TbzmRlY7}RQ)L8`iLEu!t%>@FEIwHBR(d7lwQAfu{C-N*~lMs=d? zdvtKrobnbqd7I-!@@hTdr(Cb@SD=U$0_09@N+TKop1`faZWy0|hw{WEg$?jC@yuxl ziM_`w#dQ8W;8pbPL)_Tpio2izCVLeo8`ciDTLadVL2 zIXDh-GX58C21=imxY zUfj_m{v@P>Bjcl8PMq20gCxOV*ZJ`(aE4WL{>4+2CJ*K}G>wZnWE zHg;3qcBbqNlpcR1)0lodXL`sZI4>ZX=*vd{vXD`&Pc!s5=^K9-8ou#6;rh1I;xxs# zV&#W5SszDaZTPa0C!_T=OwQ3XPLNg2^hFIQOI5Mh7yetmbLvu0yQ$sim|~-Zx0E(4 z)Fhg8bQLti#`XvyJ?7O1P}ElVRExi(PWzlTS9m zE;fEUep}+d$e0qom-`XG3&Y}kBqYKw@1lkpl!ktt6)Kg|-96WST~4Y2>gO4Rrc7;Y zvgew*l|jP5Gh5a1N;VFjw6=1{NEy~az3%i=>>cDwd?OoEZpRjibsa7YGs)k6(78Lb zX}?yq=)eYfo)qFE%7K{jlFIy7b9)pxPSqw*q&R~#`AZF^P+b!f`>1AyPr}au&J1nJ%JRQ4=`lWE_3osbDg(smY-Wh{Q zY^+j53myY4-(;b3H98>nXpGoFwqn*y5nE~$eC^D$9|<~-Vd+j0e>ov)QgTpdsLI!v z@5P3`XrQ%R_43OTG?(Sw93sVY=}_Fyz$`Mda-T)_WYGcEz8unFp$;Qkf}xZ4dG%*( z*idmO&#)S&Y~)QBr+Y@#rf8-9M*U7ooi+wgec)OdPU|J2ug_Pao@(&wAXlZo;WSBt zQOSv_q0?L*vtxsZO^s2V(r}$Z1i`VWMbQGZ6;4RzCrB{Zwm+!0k!;TYlegvAo=EVJ3 zP&(kG$@)larPYXU6{a!0GntMs%wS4Z43 zvP7#VQ_Q}j=lUA=>xV6d^LTF@mZ?L$d$x~0nz-S)DVvLzFZj34Z?)ShCDe1|2|1 z-7E8ampe+ZY2oc;8_h1)M}I$E5NjrP%V{s3oWI$Qv!y$g6fy&5!#n*5R&| z){{!vwM-b;y|x0}v5{n8N0G<|lHR2#aLSq%+^>_k2<@3xN;jmy;ffxe^mdDgI>NNv zI*Y#PRIAgz(E}TO!PXWzjDP%OS&4JKmC}L=)Mu)~@zwSKNCzCyZ*>s+4+&36tTh1^ zduQ6F`X1e5H%z(~ch@wx;)+LBavRb>k@0KLe_<1WquQ-&-U+}sp@`rQ5%&Id92v?} z>tF1Z{MkOV20)g3FAqS_*4|a(DN)VBA&U~pBIV-2CiMurwGvnRoFJqvs1Q`BziPb` zZ|z|5QwDo!zO^kwTU{gFSdcM@TecuzBY@1X88i)B6`g6gldFcq!W!h+i_rKxr;i`U zup3pdjK8(wK#;3;^b@0-Bbnfgu-f?CkZwVG_Q6@ZxhNra8j_QjJiB`w*8uAiU1) zOq;dKz01yG|3U7l$EM;lLf;(L2^qrP|EdD*;wL(Na?$y8MZx&dVEZ>!1@|A&7=J=H zfLgGCL-TvG$fiWu3gt|TmdZd9-dtpfpH5Czz)4vRahqcJG!t?!NTwqH_KoCi zj((uzeAL|M)$F$yL5)jqGdfiSi_`AGuuMvI7We6+TJ^r*{a7{bF_*@>=G11YD)cfw z?LEp(?eb%LarP7t@VWOZKYS#d$U}r|DCyYdH5n)>E9(79lcor@l)7mTGg(3=+9!8Q zv}$o{k9HOra3GB!mrhwkLsNB$c3Dj?rDvf;zh}IrxnO_*#1hnDAWsM6k1*N?9xx2q z|A4xj|A0>2_6gr#)nZZ8z_0sU<{{MG3qCBHO4V^(=asg>{R?Sh=#Ha2!6(>mH1LdN z|MBF_-}fXOKn_mV=&#AX#Bzol@VP%i`9xw*@zmoHEA?+_vqu8&xLf=-gFm1k0Iafk zILTamY(^})_LufOz}+y3HiT`@DKceL#lKF3EVm}iP}%$7!V}tx_^y`bGx#WQ)!*zv(q|Jir`QKzhQU{mAw!89@w_q$!2(0 zvDfJJkT`#U@oUkuoW7c-m3=C{@}LzUa79z|=u7nYVs<+;+R)-rJv7$%WBv?A-Q;dG zb$dKFr_3EVOM1}w1s>B0|##QYy ztE2z_)Hi!}{$+^&(e14M|MUB+N(fPSN_*sQuA^(^1xq1+FI>H4u=z^q6)=I9F6YDg&p~n zRP`Nd(AQ&PSxD~jbJM0Wpn#buPgNHipcR5s@fAEaexXEzv)9<7l1Y+kW1sPXAW>DL z_nLk?>A>bqBn#I!xfqFwd((BB-zU}@OBRnU-?whH9rWdT$5y})uk(hYPIDF-N&Vah z>ej!xX#Wq5h=2e2fB!o5pTQ>nNxLzkyid!IXe!Yv;e>;o;zwO=Sem`rQJI*Ss>7ff znn1*YTptHgkR;QcG_u6ScWnnl3AYWiXz?C4`9KYL_t~40hhJx zO1CWhQ@iQwk&P<+GCf2P%5rR*oG&TBx39Uot#a7GD+4O>s`3j$ zqBG>y-d%jf$wzdwn@mT#^CW{yo$KaGDjo>*rtW15q)PWwvBWBRUiXxUO~>}S($>Eo z!O`2)zE4uc`vaQ$+Fl79)nm2Hs$V|mJZCxP z5`9l`_$c8}10K4o6j8RqjFP-ZcNYE$`2+H}c~_LzKh}z z+y_Euz}%tl<7K6w9-FAJC2=Z5B&AuVusiaKTEcDQIDBPn$%muBU{~Dt3%A9?Hu~*2 z0Q>q!9I^i;l0m-!@<&u9Es&m^5rZdw-fF!5EEe>u#k#k1hIe*u zbKERN$yeiGQsG6~bNdhI=r+$~H|9urk)(U5U$@Wj#Yjufx;9K*+7K$}e9Rt+1h3_@ z7*nr2g_gC=b9 zbPC|x!m$kuG4)NRJko-kP}0D#QDM_g-JeaTpOrU9kl#%e*0t45C$6{mYM?Kgs!5`% zh@2am=SZQIn#0%RXEcrEGmaDMx&6!2&E4lKgsBVG=afqJ()nXf^Ly9|7WEeOm2F(7 zr0jDEH#jRkIAUyDbO7Gy6ZUQ~+@mKVN|mycSZU(h+@od37aGpi#}KBldOOozL-Oe= zdr^Un5m}|yh{qXUf7fBhXwqm4lZASu0Nf|-7T>dNsL?02ZqKk5W#UA<=v;hjU{>m2kCZ>gRqtJ_+z$ksVr;szt542l4GhCEw1^j zcRW=5#F6GiCfhBtM4q*C0DeP83V#Ybr%vp9;U*EjS-xnKu9E9Gd!<{L{PXZSFLZ+{ zY#KTJ{P%OI^(vhGj@OMao09v>K!^Hcv9<;+m(LjeGi+;F$#JFoN&8u^;qN-(F_tZu znF=3cY=5me9#(G)eNc}x)Ju}2j@szN8>v`@1hIV03l-yjI9C6Xo%SZUxo$0ergrus zQ_&){=Y4HUvXEKO236_-(q>zpUZyZM#ekEwlYuhoSh!3}m-PmlAvr}ZaZcb`&r zRkD>4Ve8ElJuQr%>6=pcV!Q3mg>Nig4$6@3AC7r}F0+r71r;d$9U9i?L+Oo#oJWOA z$hPAYv%JIYj6F{nV@cJ_3lf6L;+(y-(|ma0vuQ5AP@7cuAtztr2G^9|%q!cVg&U-G zdjtkPP2XameE@q8qI^lH+o zF08Q`FM8J)D-@k{aI9T-LX>U2)DM5rTDXFEn6k0l&8;*9ZC%mR0^hiCecA784(_{- z=?d4uw{zExR|y0G=Rayn#nTwS7%gPH^kQGXdqqZY6Kp~ye&uB7P&jXuMTaOZE2_b_ z#+%>2T3e_wWIWwz*(q4;Nns)rgzELC!1vv0@Gk7=@b)0H$Rp{8Wz3n{6zmdR_%|N$i1cq{(zp;DrhBw4$ z=Xx~7}Myl}bb#oUTRiE{u4f&^B(Rdwe{#5RP$67%P zZL`||2$XJ%UTX@WtsHu$%D$*vR;*+JXn-?xd50I^*7xTGV1D!X)lznL_(1S>qJs}&Qs1iQtx(+1Y_$%PjFf!G zVn_MJlk9h#^`Yp?P4~GrwuhP$EN`_sWs3MEbxq0=`lJCx(7t7n46DoA+TKHYdjXc* zL6ZY)ai9WSp^9}lw#g>EiTGaK3v-dez|P!@{vfa^m4n^F!ak)R^t+clHa0=N4m>~@ zyGsF<#~}UPzi8mCLr(m4E~xwJ<{CxGoB3U$2=zC2>H;`366}#_oEL^sVv}x{|82vf z^(_+>gs}~#eTtJn(o5!}Q)t8#&brlrXE^YrzN$!8@Z9u;-MoLe+kvdClj?xq(^G*M z*2Jsxcas-OhiQ{6A`}oU1y3Q1rlx*L&1a5nt_KkD%=K(ktB$AeGb?v_`h?KKj8Fgo zFhGCjj$D1Opv+j&$6U(76Fxm!$~T(f+`(1t?fYiyk)gmcN>D?JC2Vbgrms(nUpO_k zfyIEB<)VdLWU38b5G$^1PNV}8o!sOPRh3#qZ=|0 z{?)5ypL8^k7m6P5_gj76FsZ zY&?0UP9O_pKKoVk=Eg`j6xMJ`!;ckiLZ-pA^gI&$&2$Q8v|Q3LIs ztHE$BkHUY9_IMsg2VE?H29>;h3?gqsK9GkDXN+>d&;^5s(|g-#2M3 zPk3&%3n3ZZi~fl|Lm|Qzmg+_Pp`-SDQId5zbrSX$jc9Lb>jtfUfPC zuFimch)KxS$C#JzXEkJ9#@&yj4cJ`gcayP(m7_XwMMX*$Rt{Ear$0W9`kKsYPMS)v zf^et+6_au|!GTUOksr*J8jYd*Iy#gCK!%rbmF^Fy+Gp(+qcR3J6B)pRRe?J%kaTG9 z{I@d@NsnDq{(EQ*_0&EhMgDUIMz^3Mo;Uw#{(DDMV6(mzzJ0se;AvrDd7Z@2R@)RI zTIgJwI6x+%Q0^rW`C_(Bx^(}l2&2;OWf1rz_lxlGFSPww!KFh0%d01bDdi6hvR&*6bTvZAoB4VEr95 zDF#k#Y=bYrmRIt%g2}NnRu(4T^%IvkIdOTB&?D~^@(@XV!m@=M<)>Av`>)SBV&ez* z_E$gHwZ4s7V#)&TlBH@*@6KF=%-Q*9Tc*>WW}I~j!ajDf%q zEGa$ko8gjo$c6@{X%ypL%)7{^2Q~!#;uK924KlN$d=FEviZwh$N_f+y|3xL!W|v#qR$-c+(u-btDHQpBpBJa*?-QPriqjcinqC4 zj_^;^KXA{EKtDBGz63{~Q?_MDQhdE;BIA?p5htsLLlaqTrA2ti*x4CG3Y+2|cHK{u zZD-zymi0{q1yuuENSS0lC5Og+fySn>Y~+Es>o0MK`qw=a*$wyH@T0WRCHlBW#Et)z z?jcJ!LiMidiJY0&g5k;%ja^-b#2aR0*uQF!{0Bw)|3wBE;MEuK#<)$0iL%D@Ae9>B4<2 zK%TlTOkCf4;WSZUCqhi&w=SU?x=0m-qNjslnznPs)-3&i`Mw~5y)gXqsM8teiM29T zmmd{;`b^(S*vwI1p7Xq}z7}u4eaP1P6Rac$lbD=6yGX&8p|B~Ys)%Pf@o3dF9!WGQ z7PM~sL6|1AJ`Q_c$+SR+t+KwB7aF)wJ2DM?ZMhbZuZN1_ZY*Wvbma^eBNyw{KnQ;N zPVg%%S7mC3^mz7MiED5`Y7y|dLQb6aqktHF zm}Y9v`eBy11Jf@c(U76sotXe5KiHHsHinKave;NAGSjLh^ao>>34$@t%dqJ|Rh# zJ{ES-`P?ddVX-i8SYN!i^e!)^F!sXM8dx%P#~{YVMPvk^0EPl| zix5{rHMCEy8^;{Ijjznh7o#%Fl0V_qJ;&mBw|*wHKUwQq73N1i(IDSIVIyR?36g1U zYECo{pdO?O*FCVz!6CMzP(P*V)uU3?nCQ#O%#H9H;Z4kMn2re`MHSKnS!B5w%mBCk zmwQS0r%KOFZ!1?Y*8c6!c&YNd3n{|D6xF+JBECK0-(kGj_*RFeWs9uva`bmxS}|d= zS{j(yxs|%i$f=VZ)CdQe>5UmS+h2BXo_E<7H7DxeYy@F^5+dCb(T=lp@Jcs~5eUzf zg=-un^`7?croF4IRp=Lr79id4SN0_UPe8QaM)geXu^1P>%l1#RAN@(;$-11BRV)TM zmL%UJr2PzfLBqcK^{QCYn}V75kd(DDIpyhdIr6%?`1}9?F8W8t3N>`Ta>#&k9r;fB zFzuREin8n8X>Aav?}yRtx;48fr5O>LwKL$bCL43gww2nUs7JI}dBF@ZxCUXjeh8 zJ>=o~V`*jhr_L?C7(C)%v(w=I7v~Z33t!7$h^}KV4A>_h(%3bSe znQ~ukx`vW$5{DH5r4^h@zDeFg77`cxv`|z@r`N}$T=CrgJaR#^2IzfW(5T9mWg7wi zea8bC_l$Wjq67CMQ9DZGU`gs=fKP%xDuM4KS2=R(HJ~wBabnRhzRe{`_vY;}>6#-+ ztRv3sfF#Ralv0U9MNTWs5w1!;Oc5Jh!Hv6Z6M2gAeDPihPRQAxkRu#{nNRsuis_tw z&H)8a?@e_V`2aq=Bp*pGhw5Jme1h4Zff7M|)$Kk|<}Jk6w$oSbmbhWm)qZeo26WLP zSQ-IcE9U?3!dwiwC;yZ!b~%=|{XJxDI3}~^s0@Rz&ssRMGXjv9M=tKnw0@bbN$J*Y z`!06CCHqI~!R9^3x&?eUiqGzy0c#3R(J+Odi+#n;ieE;o49N3%HErW??Xz%pxZA6N zdwP{Rtf-=I5h=dgp5QT+sJbE%?Gj7s7w4B)?Tp|==K4blop(Era98@>aqI@?-EL}% zw$8-cU8$*0tzYoyCw23AEP?^AQSEd@vD^&Dz4@lm(PW~|!V-AIsR)(sxTG~Mxd|^Y z0CM6etcG5-q`$NZ7z$UMs-gtHqE{gUdj$#dv03ctX2#a!F0Ac`_zmz%_7j>~?!&0U z_haWPF+ts@DNT=*@RZ!r9ttHFN^iD2ndLE__?iHb`=5eU^a#`vYcIpxGKjYpES;~# zlifJsMk}2k?`K9UY(?-w|C;C|v5RnoK#9CL;<{)fvoeTV46eGXui7I|n;|=$B|nqJ z=5@36QAqk1=itS9!Ubc1NEtx56sii(K2s3|1nGjHnwA>rREYY}x_(pkZx+d6CVi?2orHuDu)!*2Ml*;k|M?Nox@WOV-=U{XOrp z$%U?>{s3VuA`IlXhxhL&)W^9+E4KvVy&hSXr61cB_d|52z0kj9-@4QPpSxzLq8dhG6Rso79q+fP@7D6js z1R~A3hc$}UFqXcLtjPS@V*Rl2gEuNZgyE}af5X|UE@XLH?1TG({cmD)e!e@bwQyr9 z+U-&KtK1-%q}@$4s%!)oe0S~}Q{*xObW8C!k^13P+=m76T84*d!|O^{$XA$R+yB3# zu>VCl+y7hS{$GwW$a;WMhzoVE^73RdEO6$6kFx-i4t@ zd{-PIW^uFPgMm$;m3pSBr{6(*yZPCRK9uXbaiJLJpvQ35Ii}TO;gC{@oA&gzRSr5) z6-67RQXgZuzw$=!b;67ZW4;)Rx(`VFgc8<}-j~Gz8+Yel&%D2G)eHx;4oRw>K_f<7 z`>~5Ti=>_m47P8%X_s(vJ43&p+Wl({_e-I2U)5=xp#rlW-+OAXjlhFAuN>E-s?!d! zLwHNy3TLOrT}XwY)7utP4}rwSw|v7a=I5Ruyqn`}9Sx*R5De`nMUw3e@QCCaVDv`g z*jRyUiA^FenSImz=_Vg7thc5FrrG=%;G6Gq-`93fZD5|s3y>UaRn#lE~64Ae39Z0(Bf7*dLE1n zw3j)NMrULG00HY(?-n4H+>niy6(+p|OKSZAg0B1lS_N1KSWuv~%}ANU>wmNs3jNDJ zp~(Axw#I#R2>eI~D3u`ldux<8=QpHTTmYx$^PjC%RzkP@Q2H30o_{o^Kb?Idg|7Sq zbSw8SjkUD^S8><_MN5CV*zu99Ap{H>!q&X-IrK#64# z3Sh1LIhCv!%E`qODM;5JAnet@3|vzRCOn%4y5#r!VkCUHKLABt9(*^X(vJbUOE54g zGRvl_K=UKfGRRli|2W2t^M;h^AIF#s0LSzJ$GrL1V@NQ+kBR)<0hwi!_M3nD4BFqS z06GM4Y(@`(ZqxDsOf}%zQh#S@9&oyL6%+}(K?$}Z^W{_mf} z{dver06U-@nGy!f^_y=16Yx_uvV!!6r0sXhp8*C9cM@tbXgJ&9OZ_#w8AszWtXYiT zEr5Yzv-k8uk--zdV1C*D?nwvG@-)!0Ao%waRUxzEjta*R%UluswWobI>RG?f2|Ps> zFRx{6ymI_>(iJfCQ>XL~g*0G$lTIxw)!tVEVl+zuwOCso-3Ea0?6ut#fU0B7&y)WKNBy5Def>=Z$n#!~d={%>p#*Iz z`%ZO9a)QPF8}FpdEQW+AjI%4lc+5I~D6MO6th4f*5GmT92r9S|sNqCBQWvO*{vC6P zkN#Uh?OzE7`8F@*4?12FC8&c6`-HT+*UwqOyc+Rq)Y^$hE^TjlY@@8uSQ;lZ9$cX02tdopRLj z2L&mxCH#8+@GZc@t8GN0{!;NV&0kVP5l4k$+96TTbCB4Af0dQ6NHsDDp z1jO$HXrTS1`asTb{u~n{gf;tHfMNu4j!{{RfbLNPi$-o*1;ip7e{1n{6%P|-RI&YMH~Cej|Z?0j+0-TV?60<8AsAcw!#@+#c>k7NdY3KSnkvPBelMkvEE(Z7|M zBcQ1D<@PI@GMN)Akmql{0A=?dprLm%XFPE?4}TwVQK5wCsRFhzK>zf~Ru<>{o+8<2 zuG(#z>o_q@vdTA_fAUzI=Vjhk;xQ?GaFmmdTpk4$L3v<>`1hIq>*xQ^%hrH715vK~ zw1>D8OFVrRl}VdJ>HUqS&F>*tg2kxk*FQM4G>L zJgT;Pr(8X|{h@(Od6Bn&7(&$T8p|ys2c<$&YtUh%-#NbP{gqVJQqvHl5}on2zd{bj zhIU(X4g&6Gx{n0bXI_|{1q?3yMXPojNo!q$hne&s`Vx_oqkqc4&MUcK8{b?$UX~9YI z#?*Pq0=ZOC;Je5EuCfe0?QtzO#c1u-ZciVMMB>aqR}adbu2c@&fn&+mJdlfPvv5fn zWYdtYGTi+J?|EHgigC55vXdp(H)PQxC1jwwivXn17DGF=A3*6~^a!5rL6}{4p?UGa z%R9P9`jv}d{N#-TsSe#j)GGK^L2O377d-7z{D`&nOTb)0ywrdG`_Yq*J|obMdnM(r zXB#V&6tFt0HJ;%~+ZlFzlA70`d)1HqjwS#W^A)n#Ys8TV{vfQzsR)~3Qo!w0S)-1O z{&gUdyeU=!et}YoY*D=2YjUua?^DlORFCzJU-w?4S(sHJz@zLpgFH2CL0YD{gGrNw zyPt`Q7tH_ER^jHp)hjaa#+d0lpnu=0-)qL=Ogj&;ksasZbs|S_f6CfWR9@8^UO}&& zfk?cileAjARv5JoorGwFz*UCNN8_QM13VK&r0YSQ(QCKn9rU{{^$I1uO8|tM)WN~4 z(5O3oO}`OkYo*NL(ysTwTa{HKY$Q+ZuDv1K* zf|fG_7P~QwHE^EzJk|FDF|p4di&@FY->I8~Ih)XJ$B`8a0R2l9;O1XXcXo1aOwYHuVaCrAz{`KN2;Mr>O?3|Zy z&=OW55Q2XPcs2X%h{roA9}0d2K;bLxN@(~_XFAT_-k~$^ZcuIgZpo21VvaWaGqnSsuDjz>sm1htJ~pMbRx39ACh&cKi@FtL zG2>_EEloWS$vx`5*B-lFwolW_i`Y@Sg7Suuhid6%0IId*K!@1`3MrJHEUIPgG!e@H)1Z$HTOz!*|?SmNK zu9B{kad(Bk|sQ_|skKr0c)un(cPKd3+H$Ti(?iwv=J> zE}8+H9Qef4SbM4hNJxusDomWrazFGCJL}q1Zniqk;5w0E>0sNGJSk5vt)x&w*|C1B zn#i9j9JQdRt!}6bCr?n9o$Hds3W8_&ajwuvQv!OxQEKEnP1ZRn?+h=t3X-fmzx(YE z(|A?HsZDs;wXuTlnr;u-If_Jo6sYZA=Qt+3ccCH1*vE8&Wz&ISorL*jn!kXT^W;pP zl)xXxl-4NRZ#vL3or{&A+i^%j9>hUkugXxF&vjJ8F%F(}{yTo(tM;9+`a4d0JYN1& z)WgCe#};^Mm!TYOISonz}c@DKQd~@AOG-=e>Zd5<7i9AUKB1w&vSpk z^>&#YmXyJ{#eurYx$qB&ESL%c>|lFqZoT~|92Q+vlu&e@Nb2=E&}ZpQGveU>(;JB{@n?vvBGD8aKM<~wq7<2u zr&Op5lt>s}lXff%FS8zpNz68$avM0^=^j&_MdrIUy8So>XVK<#h(;YaUOQH{Sl76~D>7qmBB+goIwX!;r&1W<6e5 zR@Boh-F+UaD7PuwO0iMQeOPE=Wf3HkJzaaWGb$>b&@+{nEKC;}xc3H+{fwXqh=kSj z{J2L#PBi`+$GWsVM91ZtqB)JKVi~Op($%vlz1;^isrbJQGUkwIa0YVqg-f$}<17NT=J8IzT{7p>)l|NR0^s`hy{`1~#U@O{Rd%iIKSI4Ys5lQ16l2-U*zB zPUhxT3$WAu{qe6^n$R!#&{58AbM%As>j%IN{YyTuYX)~i|NQ)yvlw3L-(a->spa5^ z-)Awr1e9kaft`DB1oZD`$pqeIqom!7=m#l(o+h&hI*JFx;%H9b^IuQX1ibRa5BbjAuL+EJzO64`hpe!SxdPWZ9 z6$ZPYXZ!}xtslR?1)7Adv;$w86vbY9ITjT) zf%1o+0*bvaw1wUmYD3v&dSUsLO>aXtqUz(wFTTc=K=CH-LD(Xr3S&x$hnK%%FT!Cw zC4SHe)$;wBx`@RiLJOMR9uFTdARh0gm-1Z1Cpj}9-g_4}&;GR?>#!)VPc-lXRB+X~ z)*OA6I=gq7pXUDVLfbxmrpAcX22a2TyI1ypvk+hd56GWAzHPt|OzH#jpj>U@i~Y$Y ziK?h9+a>SIv3pxLDr@rTSTHW$x9S&Ee6KQq-B#QtoCnq_=gh;>EC#DjW6+Rk%3)ib zTLnz1e(CukctU#{9m)pqMd4MF{3AnXD^C;lfdl3M*HREth`8u2Ka< zjVdkQ1Ppk+uXpCM$t0zygY!@pBV5}Fj3q#t2Qt)9aqDu84z#mX0oFp7 zdix-Gx)3unscfNlaHwf7Ak3?h`cIcDDeM9sjpa=VrAF87_1v-I>hJpc4bbT@i#!@q z?FNd{dIlYdR9EOF=nX==J(COc?EHQ@eN1{1&fv79F9$^lD%7+z2h*z=VYoRSdU;q_ zj(1^`eEIC-)oY7gliTR8u|gW|UMeMWnuO*zx z!Klb>T*OKRP1%^D$3xk^`k3B6|j|b*7N{TH|P6>#LlENWDW0UQw zh$3pjZ!}fe{2CtTWl}obbbTLc=kG$ac*|L)tY35F@l>d_a0#|NT%y`)F2-)iZ6KyQ zHW_VW;SA$DO%+a;D5YwSHC$0hn=8c%;{tKH^(uIsfOL6`mA?UC)gxhvmyNX_gHh5h zzn(Bk?Fx1?J)XFh@^ZFfZ#_M|=}d_o?wlKXC^z+i0qa0bCwzq?f-fA#V+B82kP`0p zN7)FGm7mA*kXUp zZEy~|yh2AqB9UgQ9`?e9RFebQ2|Qor1HG}clEzFRi5j&i_PP)ktH^^Lj^z1Feujpf zuZs(euCjC|`ru-8#jYL_Q@|Orzt-Fmwr1ETp>$P#6=-9U{h5>PY_;`uk5>iSs~5qP z?5BN+o@5+9gBw^7gLFsIL_oT3y~H3NtADOIaFx01T;a4@jSNeDsP@#vf+2y=*jEeF zq~Rp3wPTEErceLycHgS8A^uJ9JptuC7fRUyPGkk&cadWUQ7?>GdrX3(KG`qD)z6PU zU5ET?ycn6(#Ity9;aF&Q2B&hpQ(nGmkSjN)s|YInr&byM50g?#S|JMFDLn(i7eZH# zqAGAqpLE`X`noDuVCuoIo4vsXZ`-^C*~=_fSz0gsN8-?%izhL$M{%-v5dFGq4W~2F*?~pv4{jOmil>QZHJbugvH4G}c#zp>Y zqJDo$AJyP#YPzUGPZ6tqUz7A|_h+Tjc&!@W(_jPlQbJ#`f{#CTZ`~%^EDM?|v|B8kY2q@G-6PyeKEbOAOg&hY$5l2=E#^F`5hz z-uxZ_lyQ8s8Ub=^(2xtHdJq5x??P{=YO!GdJM7Lxqq`k_gj~Fd)&pM(G}|w?ajG)t zNz)4`174{2tFHPO*#>SNU^o8efdR0M0}zT;y@hV8c~1RPz72Sb*7!YjJX<8%srx*V zZ6gE}*LdH-zzURQU&+zp1wX(}NBg>+wTGKAq(70F87-0C3DVf%E&#HTEZoRhwR($V zM9bO!ezlF(C6+1dKT0shXtpIA8~O|?V&oF_tPFPUvf+s$~Eha*-{^sF*#;NLuQ7q zKE`8E;@5o#;}~4Ayan_4s(XFO#91W1riIUyJVik9ZgG!sUMyGjLN+O@ zUq=*>fBx8yi=9LZj*$+OR;nIz-jfu3jiU3q(k+4e4ZO`)>(}?UURe`W zDdhEJxqOA8J*|YGI6_=2Y`Zk7rVeV|S}26GNo#G^nF2RYimzFg86`T>m?zv1`8(X$ zCVsp-YTWs0i{0aATD%p~D}Gi=Qu7s})Lj>t6Z5IXt3#0*O9|^+s|q6tH(Z+9#C!pQ z+j*$-Eo}J6@#pTnTj0kwSU&>z7Y3(d2u0MbJ(hRB1j3 z?(XhuseBDPH{=4b;$-)Tm%j&NOUUP>Aa@*3){uR(xu0aJ{8+%smvgUH6EAFMfVXnR z5W~8Sc%$)UJLOI#F3Ij`oQrRuV6cu1A48x6+_`e!Lj!kyappyhv5CO|+dDo*?XWt4 zp(Xl#`s~uY)07Fb9ceskSNbR#i;b&71H0TEp;zKiZl{Mxib{bsMypUzay`$=U{cU< zBSkn`h5St5<0m+bUQS^xmaZKnr1uxpsIP*lB_o+*g@;}6+I>Jr=3u)TG=MF*$>kV7 z<+ddQ!_>P&o8CS3Ph^gc74XL|C$?E>eJ@QrX8xj~ZlXe%)0Xz&)2E3utPLl*0V^?( zSO30Q0dS+A2cCj)r3+lr)Z1@rD&{#;Jpj$8M_=0m3GyT#tTYW_cN;lJl(DsruiL$f zwGK-C_=TiZkrWQIPfjVA-4+=kZj6sks2x{fb+Yk%!j^qJs)>gMfVbo~A$=&jX`gN; zh8hv3H#>?mOoqb;7Wy76vb)jxxaV*kw8hHH$2-*eG2F9i3@LdOYD_P^KO@>7a>nA} z=#N%Ez4sm8LJ72ntLb9*%$7|6M|15}-4oUcA*}D$^15n-oj-CwO+z0+o@;RDaG)9y z&DovB^$o={zI{$@6tXBMv=rJ98rVA#tK7?PR`4Q$reJnMF|Kyq>w%bbZlL7xGS8a$ z`+$cnb)n0_{`Q8fN^1EJby->Kd@w>I5dcEhX$-Yknt$|QN|bPb&l{l9M7zHYCfV4+ zEAWv8*?@?S>}&yom^qDV4w`gt)V9C=5(kCYN#@7?yfQK*fHina+IhIxJ}o6oIF1?} zu=%|lCvC2NkY^1}_6hgbTE7i5$gOW-Q+SV(!9@Yz;+}GPtg(Gi;-M`0Kfn>4DVVA$ zg>v)cVgAv+;~P6m=U3vSZli!H;Oez2D>}|#;+oigk9^O+QrBIKAHj0fxM%s6ZkQ#%&Bgz0@=3vDdtlT=)R0N9xQg2ec5;;manf3jCXD0fl* z0NYUVcZ7g|)TJgKa4`VjM44W>`xxyj-#k^pnxI^s=*tF~Qw!*`<=%2K2Xq|u;k-@L0$mPwzqyT#RkJ0%PI>OS>n|ga9 z#j&X~$zkBglP93O`Hd+n{{*`cNQ?+RZiSA6r}~93%vQ?i6~nFVIEjmcB)X zcU&IXcE6Jdic)PWdW-XXFRNYyyd@M;a?C!t8&s znZN}{Um*ED$n4SoiF>7Vw4lKNtb9gX52Oejl%l`kPL!6W9?lN^$)$aP*x--2yv>L{b>Xdk;YK5=@*OeRBpP)s0MLq@EGS4A3p znRlP1G5T&oZvU62rReBUz`7Ff`_NU>(TsaO?ktS2l_g#bW;lrhUg-lx01@Z_Gojhbd(l{a?S*FS)Ih*hxg+keyc1=y(i zA6gWOAm+-6+LHy`bRSFTh+2jNcJ7~BY~OGPhj&h(f7p3s(;IWwj@5elnC%rrt1T|* z^cvT{wTOuWc%qt(leHuF?D(T*=b1&XRSTb}e7sr7p8w=2g}m*yz&y^3u3Wwm*;Jl( z5<01G;Jc3_k<@qF^c_Ld*=4o~oHBnZxqq|rAvaHoo4)Myp&v`hEZU*Fo=@ui{`$62 zOZegCQ~tfx8Ynfq_14Y-qIc$i_uzhg5U!b_$^GJU&LA_36+1SRdo`L~N%_DeH%9%^ zij*+{YN#%T{be)SRVETcxZTEC1B-zTy^)AAprToD2oYz@A?~4X?2~wYD(pV%+RPhW z@L)m7JO|ybj8f?NVVf}E?fka)iPN|M3yDl1fA4B+fJwAPmLl*>Z8UX7@lMK{Ry++a{4Ee>WT`sfHrXzAExZ<` zCKa@t!yigu@N|&6Feka%-MLAjh#k(N`-p)HtMDT;5&;rNZFI79W~O`{Rku#GEE> z0cCh0I!twu++O_ArSGh&4c&`sTHWsAoBFbSpMGaC|8MdksKD#aMj1fhBb~0- z2;yscK-%&h`e+#+s;y zhmc8=DE0TW;%R&55T0&GvrL^Dt2t|jD7CkyD9*Jb>Yryh*p<0953n719=*h_sf77> zFe`upbFc?i%R$;o_+P>&)*^fXT6O9(6(&^lBKUDMs66Ua26P*@FK0KN(k%}b&9)>UMD%#y(Q4Rt2{C? zn{`^VtVRMQ-Rd3Q*pP_GS}|NuUVYFQKltjF%J-3Skkdb}@cN{yJle zFK}wHY0-6~(Fv!_VnC$SaUX7-r1^>^vNx6^lER~SuA%;0jd`o&(7Dp62j@3RaEPD4 z0^@{L^4V*Tvc-Y37{x=3&vWQ|pZl@*eEe;(EuN3>otSdXs!c5aetD4fM*Sk)@J*bX z`tc9GGm$y<@lv@jrehj>$x}+e6oA+92k6ZXbX%=-=^vgS#mey7mAwb+w@6&--b3XHY(XD#;e(qP0sn63+HD{q`ltNCSHOZq zxILU@+{3fB75`39u&mw_R&0s-T*co(p0@%e4|{y-YF+~z zmTO^G5BC>bhzxm|CvWwmRcP1(j-51;1v6}zTI<5#FQ%peVIJR?fodt8B_El>_Axxb+7WE%q{#2mhX-OV+vDmD01&z6m-<30nj0U>h2 zJY5E{n}p~nUxZU>VvlIV+oc4MBlj^mh|q=5A`M8FgBnk@;N#vWY;HZTccf`rq}gJ* z2_cD!jYrmvYqw*Sv3@lR;1hqi^o$MUfpFelF^K*az%p-p0=KP;-BSUKS$jBNYDaY{%Cp z)>j@G)>;8^Rzpdrmd6MhO|9^iG(_j|^x#}Av?{JFh9#UzyjIuZXXTFUSBvfCEOMYq#Ps*!5c*}=VJOv99S z>nrXC};AjsQosSbzBX zmG8@Utw0$USV8`lKEgagIO-2jo$-KG@sApBX;pl(PA>obU#x^T3pWn-^r{ZtMQQYq zykt*e!CcVz-N%GvldG||fYC+bRXU^ip@x?^nUZ^HLB3yV#?9L6_)pxK3!jHHkn`mm z$Uzf)t3;h-FY(mu2&zpn57uvtpzUaV9rmRB`gk)H2f}%If*wId!W=32dQDP4Z8-}& zM^~wCs&5Ly9mGQMy_5yauc%lPTz3JZ7pBx)NJy_4m-D?(3BLHO#xJwp1TQQO;{8-t z6qv2qI?~(lEky8R=6;7~DV?K6t(i1U;>)t^2i4QWA9@Qjr6FlDF z!vRx&Hy^&RgW5%M^PR-GSxFt^9tUTfyVc&Vy~a)|7KvefG~44%&;P{D{;v|Uaa#~V zQ9vBJ66THjHti{@XAL?QjKi0y!DbrYgs^^#_)%te+wS#^>VAT;;*ZeN~}@{_uHnWNFT!-2_5B7(yQAP|V=*1yUXtp~8}@D%z< z!q=mn(v004$<=UX$D66ZU7aed4GlU1CP)NWw~ zUKbQ)fF3{WN$^9MWwv!G;pJtl*!2NrT$k9gjC`SfQh1@)PwWchdle(R)CB#v3!F?x z8rjlj{;Iyj{eEJ@RAW`Kz{4)p@#Kc_krO-E!^W;Zd)W7VHR7; z_d!_yu|cm7Cs?Q)^MgcV*+`lixnCbzOCvOBZD&BicHF+0rMJFjF_(CQXAc6Jc!2!@ z1wO!VL)fPeRY|_k1BS$`ikMdc9<$SAU!nQemdI<My?AV*8)A?JTEE8!8nx$i z=sq;$;Ou!;%CgNdTI6yjQR>+M4Xet3jLsE{*P`yGmIHFy^g4mu4LL;MdyucwT~udR zOIb&k2O3=CKjyI64z~oqV>e216F`}E=^loyc4NrXUrbnf#8k4hswTa!`hH9)%OyBx zodc(}Krb(EG=C3085p0e*O2h~iPwfm>ANh7sP*U5X$Mq}=C&R`W6W){dy(LN@q8cQ zy*&chVn=p-68I=bi{n*LNspo`@m>Mn+fI9lU_2`*@Z?h06g@`P>F0|n!S0#51hY~LXGerv_D3a4GGk4EDnmz{ZL-4jnCFfLre>ilKk3f z@(QP7bkGZ1nAxD?$%uXjpu5wmtjeuNAdWyOKHbTjv6J-H=W9=b;T&mdAv`ceb-Z%v z&LbWr^}4nYuOwIUAL(Hva9Y~y-EPO)$tI*%HzV)S_|f?E3#jyPuQB}NECs&UT-iH` z{Q7@%#c?KN~}7&O$d+->zBdzb%9Rq~t^yzG6=K3-X2#TFZ7W>ZP%c zm@z86sGx7qJz)eA8sKX-jJV}7_40`WEiIE%J^44LC6>NTS;ih0ap!A2=-mTh&x=|p zsi2V|)YG9_#+S1}E0pQte(54QWQD2p%bZdHT6%oJ`1A7)NZOqC`~Kzc1cEzHzFu;) zmq75k>h#2?gAdAte3d(;P74sGds0R!l(Mhz*&{%wIi%{jxpj@fM{w?RJ8fG)d#OJr z$x1Sq)YWj|K8h)G*%*&;D$-;Yd_Hd012co_%Fp0k;_)7>`3?iA$ZLa_vbQAYs0s~cMdY~ zDwA&|NGDC(`4l9>3XONfJvvQ{OnA|2PdmV)XRK8@grBrbJo8pd0G;h|N#QSm402t0 z-TG?k>-wPfhVgp(_8HJ9o`nft6m-RVwUP+p7-0J<9v0QH74KJty+afOcs5ODBNG5N za>QvnhSUO}4tRtXfthB|H3lf01^oJkG>q8m2$?zy^O_zSnFkUi@4Bl1lYt_IcRqYE zJsaQBnCkQAwtTuH&1iTT%_IWFFSIL3l(fZ!;s3^!rVLpzr!qiW@#Lj(M>Sv7KXkCL zP2b3fZD0&M04$zFiz{f*d42MN57AHpMclo?if8$x(2qeEXdo3Lq5)3v^s}6Yv&YZ*I4v>0f!kv!r%3!m{H zSvZaL6QdN5&O}l3wHA{(1TGTs8Jc(=3c3?8dMRV|5KTYLY^Prt^nP8rSqDdIOvt{Cywa{!cmL_p{D2QlU>kE3u`P4Pn zfzEU|qU@7^lH52?fZ0>Tjh`Qbm3z>A$K%J&@Eo0?Kt>O%^g(Mt)FhFgANqM^)(9YlACyw910D3Sz)k< zH!=Pon$5|}v}0B5)n+i2eHME=Qr=9%l}_$fPpLw>vBz5Ot#*E%V~Y61Sl|snhtds? z*e`O>&05$2P@q!o{2AfqOYy5S)(~0xPuhZ9tPOtuB(;`)KJpu?$*W8Ck& z6ns1U%9?-u&biIT%&@8tkwvS+(CU!0f_W`EyQ5y*GlReFXndcK3VfFN5?A*e;Kd*O zE-0aIhE(vZtGrHn+IL?>_fs^mylVg3>H(zpo&Z53J3ZhS_RR_?YiO#7Z5$-Mcz`&+ z33&vRHB7m?V~wEyv(KC!b?kpOrUOa;huZ90k%QxsO8E}sW=aJM1%rkI>n68xyq!nz z0N3GyKfzY3ttR4%_rh2XY$Ip81AJD;%3@xiyN~g}4|Wp|RlaO#ZE2Y*F`l0OYHV-_ zcIG%Hpu$ity;gc{R12tJez@i@L>Kn(z;NP?ILSL)u|Fl_G1Ahw3lE!IKdzrDN6ea| zm>NrG9q)@daVk7xd&N(X=SGr~ag1pG@c=t{%8$?{py_R0Qz$00*lv^%!slqI#S#q? zX<#lT5sdcuI*43-;ZLpmt+jp9nOE2}+o3&8LGZCWLD12@CjSKGjFv1d-E{b)(^+Fe z-xAJe)lHE*f)pUtzEOsdzRKtQlh=5Z@G7^jc0LOwc_yZ-`)__I%(Xn5W!U*C&DiwHN z8eGdoBR-9_kQ9zRpH1@aOSsd4f7SlUB9_Y-U~5seQyA=|6~^QnD2ohnB(NtePdAk+ zgI>fgQOJ_}J!J7?cf%nv*QlR7Sf2QGobLK?He}8N94h+qWdnNTi-JEUmzZnC_wGD# z@2P%Y2^mIr+Rn%{E2_ZQuF3B{y{c`utKyUIOgAH~){vr3Bsv*BS3X%M9+Q#fk#r*2 zLOO~X`xHx%bti{-Aft~c_N!r%3V{S_yanE805_WGY-&cwuQ|TCU~Q=4S6Ki0u{-y* z7jXk<{@%pi>s=&DY6&WcysOsZQc_uWl zUsSRZA^QdaO_YF7SUabk*HSM-^ieSd+1DY=aQc4igZ$`7E8P>Ufz>+Ov^X}e5^X~vJ9Y{(I@77e=WCs zqquX}kroqaWZG|?{(Lh2Cn3ufP?-HQb+;bJI#9$<=c~z9_Ds{qZ0k(M{@9!6!IJg# zPc5Tk%3dOUD0I+98<$sdKnbPhyYcxN4LVA~_3rM^(?K(?(`lQ$x#)ObM|Rjk`qIGu z8ok*=s|JtPTXHYx;>}lln4i@keZEG6czI>9G3XrJYlD{>__L-xy+w{CMr|2N3xc#S zU-_)7x|&Y8`as{G#wZ!X;4QC{qC+x=^#?$RU= za9wHrRMv(cf@rB}BcnksC%7bp@9T5G3H%H?WX-wi!~>RHliCO8oJ ztbgwP@t*fRtg3QoKcOa|l4U1zLSbUw6=Sa1X7C6`er9ex;W$(tdoLpQfsCJdg1Vwe zwBuod{UEe+#5P)c79#Dz>OhsehR1l+tYNNhk5hvVXGlKch+hg#m?3fz7!c!U>DARp zIf})@9uJh5v52)bd#Gh5s$EegK*N-CrB98_96+*Z_3pE0hf-VO~n%D=^xy9SE z4E{R>eZBJ{62oum$Jo+(jEgeH_dmv+1v_37dwNZ@!a|3-T`%J9z1YcKU?#t90X!yS z=vaEgFD)q=Q;#eKX3RSUMyr;BNs01ZUY?c#8UN5cf(r;TY19m7_?%2IPWy{8-j*%4 zg!+`QM`9=SI3)ET494rkL-!!q;)iuNMDzPY5JQ5Gn^~fnCgv_GXR6#o!W4KmyfjVd z!xH8T6+b0eb&)Z-9#fo_50nM4+mFEthQBP|FfGVFOA&{p*?m7TUDnlT*ODbf&S=w0 z>e|`;0b-u@cd~?G9~8=`=$?@npH|cuCTuX0Y9{wnxF?SbF%u;$t1n156rV7X^mX|N zw9n^)GTQIB0y3Mi0cGLHFDt*8j+f=J<)JzbU|gE*b7nB&@0-Rj(U%DrurCX+-_ z|7c9y?+5;-)16_v|Mu<&ZC&%Y`TOT|L5rW*In5Ns5QHABym{g9eXn76$Zhdn66=_I zR71n$5>PAlU1`CNPL?1}Q?pLVPbaC3GZ#ly2JdU}_ggEQAm}0`Z_2Wi(Ya~;w4Z5- z@q($5<;;v>?||dmzTj2Y!8$RV6O1h5f7Bm#S)_c7p`R$Pz&91&)UOTKlxiEcq_6fR z3I)e_39IUlzlF_nY^qMarZ6)3qH;IG(wC{5Ne&?2y6S4l7Zc>X0zC#*^21MKo~oh_ z%IdDl-`ZRed8H+Uy4mN8(dg{Bh~Aao6ca9TF7SIc`o@@1^yQdG)Pvq%R#VTaOt{YU zY&%GUu9a{m-_2Y<&E98nieHM59uMMNWZN9l`iSk9{df%tpf&Ag_lP!Qi&|Z#xd6Wp z`bWgs8EC)R0>C_tw#pf+dW)y5uNN|4;gRdr$P4z-nPbDzBtsHb?K)8G-N8RL8FH(<%Cm z&qjDH6g?0L3W~ixDX+v^oZHDF8Rx6yIIeKw(NzG`GohK5l+~;sNf-S)}*ac7f76GV!r9fGl9nG~e9d(hl zfiHjR>&)lobVgho2k1k+YhyrW4#k#Bw^!PmBYf7zF7E5b4Mrk(+e}mvgR=6EppvIA zu{yUSQ^0h>n}9$xSKmzl;XjQfA;n`l z9Y$w&I+x8O?12iKOS48xf(}SYt0Cz^?6YB7(k*Yw4Dd=Ms1SVTxf1>LZqoc=Oc6K} zB+=Q%Jb60*9I;?nmnl7#_);_Kc;5K|p1Sm^2(C1;V)&Vgh3%TI`zzg~Fgd*9`why> z=okmKBjfD=!Z4M^pQ(At%P(T0o!(F+20jro*;>3S`IOMW9#I&VzkNhpBLYMe-qcND zA3>>%-#G%=(VB}Wg<35MsrS_9S_X|z=AEj)y#S;_uDKh&rdg-k zpFjeLXxy{4Dp1?hMlHB9AL7GWN^TLdM+wM?Lx>PjBK%slV19@kW8Cll#?Y zRY7^VH`wGkK_kpIt!7gI6qsm^42xhYfEIztZ4(aa8wOQ~w)lfM*6}2XkUqYnxEkSY zY!2Z)@o4+aZX><=n>_5B1k4ZSwr6Mw-4qa6{PvuidOaFes(`G~o#cDL+MVpx{u+Nh zP&Bb)_JsHkkl(b@awQSETvCpyazzHMwD)M=Lc&p~RPL&`Z^7DXj(-ozb?mDq_#9Q# zmn=>>VQOprY|`m0kIVam_^Y6)IEkAd3{*J3H_4<6cC9j#|BXWbXO#FGWGg7wpANkP823L@ZwPOy|;V}1SFtwjjp`1neJUUK(GAB#nN-@Tn zY37yU$y*!Nxh#K&f^iJF{W%4?8AC25Kde>JTVM^F|MW>@Z=d-Ub~I5&(2I!{?A7kz zLB)o|&NGC6iLRqO;p20GUWwO}6|RfeJGNMti!pvyUMRU1!R3gu>+~p^jnIlb`qw|g z1junU81YI;a ztDGut3D7Co_GDt9>h<*sp0P-!ZJCmMtl8e}vxfRjeZ%g~LBqb`3h=vde>rFHhNX2fRy6KM&S6tZP%TGBb$>2&-zNCA@u6 zFC0+w2;LC(Ij=-SiZg${*3Epgy1MVVFtkS+8wGfFlUtyyqtk>h(_OJ(Q(lfAf?@^( z1KNQaFF8rCuaLa&(o*B6+=Hgi?2x1!m<7H^FAAzM+ru1`o}vJmiC=z?$M*6g-kypD zo0KZj#5a(TI>DRT>^2b?+S%)z@u{Hlm=CCFcxE5G6`%5J@6ga#G2(|%anrp5x#~fpA zLsG#}Yz(KZVk1LA8P&=V!WJJx?K;56ZpuSBm{$)hP%Ic0hR-wUs&l)WVufYwy;(lbEOi=lAc*c(iWGF6?)jT!jFe zKL^-)RFc*fS0Ha;PH zh@1K39rXn0a29CHN}8v}sH1QH=(aTZV^=X-KRq^}E}5l7a>zMZUkU9`$MwQGZh`KgYDsPP=D5cx=9IV$gk7>_v3 z>26}wrkwVVea=tWRv`nr;1E{~^+bj1qvQANIt{fR*DrKOhPD)kFOpupNy5$PeHyqB zz=^1cC~wiqi|RMCe$47Zu$-T@&->N1C0a6W>#5OExd&Xu0e)vCEN6c|f5GOQL`6L6 z;St^tt1#~Hn4QsNi*~^Z%0o{|XpG4w|1PQSw>XR`6Ju{4#yk6`cdTTzDM7^>Bu*iG zlyp(l$LB_d8PMu?Rqrk-#4Y538i0V}`rW=*=uR#p`auUqNKJjxSGV9rf~l{H;)|GC zZ0m0iwqYP#PQ+Ie&pzE+#kV*d4b@W|G!ajKgNDlN^O)S(&1JL9bjpuveOjGONo394 zXcgI-s_KI%Ubv~Nher1U$-BjU|tM&{^ut&7UafoLB0^@A*KP{o@LY^aTonO)G4y zFx+R$@k^;z-2~#-hU@3;mN>>``Q{8|dCnLB#PQ$T!{b(h!3q@Z)@9ieL_13Y6O>@s zYFWKf!oS=y1OUO54^X-Iy1D09-fFzvW*9?Zp!iAq?X+)d+a|dhZD4pe#as<DdQuFi!bW#xPGle`h&TK6kSiezFaetU<6hD#x8xwBoYklH_&l(N15mIbFM}GcX zmMj8(a9jOFLB^t7z#Z&ji3tLtUXEQmh3*1@8cPkT@$7+Dfz1`K;x%jxFy&>7yH}6D zGGDv({HR>bjy5~M=%CH&Uc|%vYpd4@IJK%pL})s(oWqmU>DTLJh@bfr-HG6&_g2T{ zh2KGr1bXEf_H8)%xD?xxEb=x^ynR*3%ld%dvE<&f3QZWr5Qd8}N_Vz!s~T}wR1-F~ zG#vUd&N1-y4W#Msx{odUMp-=`jHKXi2WVBnG};&mqXn$O9;tJkMV8IR{(gs?^9nf_ zBizX0Z6RR0P_Jn<=0LH29&50SQThb^t_ocf1RbN~B>*mh95K*ynM8yvvj4N73~PHv zyFW*=_io$KfN)fCp_*J?k9HK*))lolnva9`1H8;EftU$hlio-WPXDriT%yM;n;3? z6{HZ1)YPdTX*YMizK#1w!EU`iM-zQzl8YwaZ0)RncnwlC#&YE<@VE!k*YGJf9YwZY zk*)_^%-hJ4?OuZ#t5*@J90?11ucu>2FD=GtilnjU zyR`CAoc;6fun1~?o_IS#d^*l&-LH6Jd~9bDN8Kb2-z!R~?%rFtyLR={X&B_3I*&~L zK&d+I;A_&95q2KZLg^@+LI!S%@1$623F7A-J9e|0NM6+4Q%2xhp{Fd7b-YNvqNv3V{_a9UI;O$RbAh%jMS)5B$FD_Pr~ zff@?|=Bs4_L`Uox&ARYv7!&w@!PjMoM>Zd3gf3_b5I5U;w}v?gHFVlOI0z0SxNbiUWV+88Vu*4`U{j*9*#o#QVW!=F2M0_t-SL`)&b{*0G!|6rB^ zmY~cOV!=ZHU5dP+9~T@5tIKJw4ZRxl5&uMV8(E{tjA4Hp1ZI_{Gm>IJmUoj3Q+7Kd z6?4Wzqj0E5K6(R%(rC|ZOe8Z04aW9sAZ?0}pyvM# za>PIJE4c{sl_{}VVM^@G!s!LvPg(%O`8pIo% z$hUqrnO}H9K`o3q=R2yCSL9?fcfZxdM~<>w?GaO$H;*qr%5f6PSc3X5SX{%}r##58 z6yP|OMK%odA+8ZK1TU|yR(cn7qMymff9JGWLpPUXp;;M30ZpoKC_)(*HTyXA<|0DE zVjWBX(OHqNEHD{%E6R;qZvRYj@>O=(s7;|zXu&hFTf~5{==*1gzUyv&;L8(x;%NNq zP&yy=JNG2SH>}@zS)gJc2;NCalm8Q|IfWEdYmtx3j*=M3%reGsER1RJ4<%`DpAc2N zG;a!RgxhhgO7r2Khlj51vD#0&M@(x&=Lw-NmuviVwUD1Prbj}g(jYbvE}p6(+7#u3 zOU38z0WuPq*0<{a_S>hA!>)$U>(>=Gr!Dl|xFHdTk2BPn_~@b=zppuB!E#&tC4JP# zB-G&TcXu4UnpW~k@JjiIt{Wa$$LFVvUc3(Ob5;Okf@%+;I@2xk61?lmUk;D5ZyBp; zT7M|E3_Hr!>QY<@%ii9Bc0S$6x_TG}^yq(s#u(O((RQrhvC!=@aIzahq=)l`CdsGU|xvyDBb=g(!JB|sQh!V zPd@mkZ(z-@SvLQ7Uz^5!tt@PWru(}6K6cS6v1U-<1Kcj}tNU(sA!VawdXfvQ^Ts@s z3&BApcYfH!G<&aQ@gSaFw|ghT%}wE<#WB}eYv!_x?s30tte%L3^R%iyZx^4h2FRZ89)hEe*!wK5UWRyRP`VDZT424)~SS{emxNsqj~c7{HNmcwbxUK(OSQtTuMUR2A!U_#BSKoS?(#EXc-J z%yzAgTcTkiXMxl?0u${MN2Sr5y{>KyT^Ko9c9(Uvd?s&!yUDc8S%a# zmd8zxQ~YtBFobB{`I#$on`2?+&q&ZgPE&oOjm|yzr|wO*yX$pN_6WroZ3Aq?tfvUG z1+Qhi70;l^1cYjQj6tB=aEfA?RfDKv! z4Y_Vw=zVDCn=4r69y4WJ9^MmT*_g!_jmymrQYqO)>bWh0+dY6~5N_0)311;6I~#~s zzO4V_$IK6C9e+4WV>l#0MD!4`{hlUtCiB$kPzB}g3PV$s6MaGLVPr@vmclW9tTa}f z1waXDmD8MHZ^=@uNmkdRK*F`2)ccJe^@*!F?dn?pJDu-OW9&a%x46@+D|jty8>A6G zV(7xX`Qw2YY~r$3PP6)YlGimxM+so5d=R3nS1H3yquOHWqAPy>flyS&B-w~}8?CiU zaZrIfWCiEevJuH{4BLv%at`?fS-dhOR^7ABwz0p>rCo>BGY@dFn z&H(#EEZpAK8cyI5MKevx*h%d9FI&ds$n-bPJW{wt6p$H8)m$ z#EMJdaG1TVs{m&t()UbOHKg78{7zp}{oR}L&yxHyYtg|6DlTonfn5e!7;_pM$(P%% z<|H9q$%tSRS{%c9B}HEu-2Dr(V{vUX17Cuze*<|3#imeOTy zuz#S>?h6DSd28>j`!h2hMh)Y3-s_*X@170DWW4QakV=c~kfv;-v(egZBdAL~g?9I8 zdQ9;VK6zD@(ao66)gI#Z$@TAH>VDAG2!VX(R_~_B;?j$8#kP{J+y3%}Ks)1Vhrw%_ zt%yQB+mn}WeD_RSbLPI^lW0G)2&5}kO)C zv)c)3CQ2}SHB@06p91G{HIlb5X(h`xz?|`Yui^!_@id6sBI||B9B$?Ha(ltl&G6r# zkFF=q6^Ep^vac9?e(kEP9dg;Y4P2A1VHtycqc#7;eUQHd3Q>X?en!hB4rzS+G z&d@TVSNr4RC6ZGZxT`){>i#SX!o5OZ=!KNNj}5| z_lR7W@XoLyRHFpsZWW0wqoF6A9Dqe97j!!>uHOn#jmYQ+v^*D{-GWf2j;P~286O#{ zUcGh7EXc8O8zlz;m=A5tGt7wJps-p3lMM#rbfQPMZ{WCisC1&5yH9to*t1ex2l7?6pAR$fa$9l! zcy<~1jFal;{R$+fO6GOPFB84!J+apj#RiO5HY7_5@aC`$V~WILnX;X8?~LLfl#clW zr*DY1he1Yv$i@RL%3p1pwtFDJza(1X5z7jzX~2KE>g>hlVMYpcy$!@4s6vV>6BSVL z`MXUYq9jOdF5W5GdED{jRHb=@MK5qVij$3?yt1#~;59#Mn4y_gE0K<(y772z@@SGz zLZGZZ#+8c3;nU;Fw~Xl4=itrUxXYd3Uvz%bVh|$2d7Fi!f--3f5wXxwL4jMPHD;Ss z82rr&elnI@Ev5^17&f>YTjGQ%!m5lqj;zz1g*%CaE=F;#XB% zl4#y%s2k%oSBOH}(#&y)VA`^WF0dRXyPpW^@-wSE?o_*U3b{@!<12)*JT{a-Vm1&=bL8tY< z|9UdI^1Yu?(-RG$TwjToN?aR~Zv_^13=(qP%X)kG(Td9v79L6+SNk9O;r>X$|2I;f z{#WZ3A)W?;e0jC|S#9{y8)FJ3mZp|kw&Mgi`&(M9XtR&1YKW zJ^I#~(5ZrJl%!wJ`m9mj4V$WW3Eyql3LGZIq_$qU{>a|-ZQdn+?h1G$yK%0LyF(RK%8Vk`1M91?exZ_UQoAx#h-tF(mzAq+Rj8mVhx0lrS z#Q1w6=eM`%j7}33x_Ag6XqhcTR*3y^65TCoa4~+CL)i=s?6U~iUcj@s2=K0xr;9w$ zM`#Vv*=M=xq2_PL1aQ~GSoMI@Wmqo}!F+70eU_Vl*YaRz`*3M}HvTBFHH8$%xV>Qc zbN;cxXBN+_=lo{1<^zAyte2yqF6-jl<#^r50XI19lVY>YMnN&f*PD| zjm`%|SlZD>-8gw?q_(`jo`GfR+pKe}4mpO(>{RYXP3rUXtSC@$AcJ7>al!?6f@z-(mYT7Ht<*GvM~*MtD@?d%7^D%o;R6q{@fhYEt zAR$&C`sL(yBO*FABywAQd`so!>~gN?Th!O}IPcw>;pYT~FKedk4Bc9%9tm3AB!S|# zb@a8TNOjUVa7poImPyYGeG4F_ydtyS^;zfv0z04-MOb z#mP5?E(I<+3I0Dec~6AOU%O?z>yw^%?56V4iJ+j#!2_I)-N+dro+wR+4<Ce94vsEfqRPV`_F`p?_;){B=!>VxekdGex*uSo@W9ML_oeU4v9_&_mjnS&; z45MS74~!1q&but$D{KJFHFIDre5ho!ZHCl=H!Sjh3~@X##0`Ijm>T$in(Uxym!q86 zJwD#z4J%A-RL4#XclOzs^yg@RvC*MtkoE>Nk5Rrg?@<}40zfC@a#!fsR2Jb z-e1oX2si*ap2>%?=B!hK9{^JYp=BMB4>fFv*$}~DdPQD4kUkZZEG*ffCjuPv&wc)X zxpUkirXrSRLdd`v*T=o78AJTjJu&P{%{f-|%h`?TBl55i!TE^E z^|yggv2?Eka6=)pf=l9|>MUn6oUmCWq>Rmc?0q_c?U6v7XdlZo5hheYrf(WJ24;P4 z`YA;`&F&=_l>q&AY%bb&BQASu$)rpn{!rDyXIj5FrId719F{+i&$SJ zxaDPbU~k+z=v?^Gr#N+UN$5NfFaHt0`1ALFD~9nGmhoS?JGKE7#t7VQB0|p?e?y;6 zSIFeSwkz8W9Z43zS1eT5b!A)U1FJT;WZHf)LB_yPX47d8bD2I+Ab zx^IaKIz)J1*}6OeBCN7Xny|)`4UG&{i5dD0tVu-591`XYvCp?qQ=lqSM6gB zf2?*WjWDAZmHQh{mRj!7+I@XAE^HXS%+Hx9-~7g{`C+d+mi9ypvZFUt|j2VTYx z?i!ZgBS|@+;hu}jiIj7>!Z!9LWgl+Q>UN#B3*Ohr2*4P_387ss|mG(A^pQ-1yzb`b6NQ_^&1dc_%6u1H8Hj1Dqx) zLt|~iICmchumtXY1ON6)bIG=^Th0KS5ob?W_>wXREH)jcokDoZI=Kb%7sMLZC0Tbk z1-3klz*O7G^YaB?mE9`IBk*zI4L7v2>8C#%0+v?BfxD2c>&Do5Ll4Ed?N1(vsJP+m ztT*V`FWvLdT@6;$h9alPuoebbWJ!m(OXMT+fD^FBE5c`N)PJ31va{yl)HyFCsOGi$7?~{ z5YM$B!APo4EG+n3MN}Zd&Hy$Tch3U-O;Q14w1f!7jls0A@d6=Ej=fmUZ_s-;EtE@- zJ3yj$4~FS^#>v-ByUzO#+03#^@MrFIRbzLM5BuuKtKAwdmd_t9STW|f9)+; zK>al4lmp2K>}SF(azk=khELYY`aPI&fwla{`H=5}+cdDAPlt^_3|k-z4KzO?FogKv z`<D$r8+9Ld@o1NSr<-n#HY4rgNm)$UZ;Az~ZXa3RloE4KY=)}YjXpW1y zD%{ILK0)fxHo1OmZx%bi5Nj3JbN1Z3KhPbB~2SPI78n& zKMNZcESwxykfRR)SzFVPhlQRR`f;m9S+w9qXX?+LSxU}jDt+~g>N=O=9Os-kyUp3`rYzxJb5?xPb@7l}quvesTa+9c*l;@-#w) z4&p^oWBSVHz!`-rSU((=4X3FbyC!#45n++I5rUVRTSS$6=W0J~DPDPn$}D-8x@j!| zm?8Ii))hwk&G$&KCm9B0zyPr&*~RVtMQKaFk#5+2VU#hL_n9 z{YgeCTC~hPjh|0Us>eW=2}rYsbyDC^V4gfj9Ea&{)mM&GdxMw6-!eueIAMlC4k++* z1{y?cO8z##VW-5_h{sERD{vHrQ2oC)^!UF;B&OnNum%w2Ikk`@Nqy!l3o72aKJ$sP&lXs_AN?{eM*+ zb)Tqsc?A2awoy%*i!j_nwyuHFv`=djfP1>jKdNPZ^UB zrh~5K0e%K~e}7b(ctOBdU^Y=TB@WFZmOfYgIgZBlz-9C4AuYq}a|uR+&N!VcX#`)C zptwMcG$hfc$IZwW;sko@P2R?2ia`d5r zV`w}bjN4F&>_?>*<#PG@SWeh`b)l50a%x<>-x0>@nKEDx&T^1XXheQTjoE9(iMg{Qntw{|kx79!%B*M(!~q<4!>xduvNmTRIVAbBgL~?>nDJ0OiBN zE;p#%dMh@NZ-ud))Ot+jh+Nn{n_TA-Y5Wbci(pnTap}V;MLQ$5ii>uI4Kl;5Ts+)A zEEsjzk+>%<_xloyGV$}N9lw?$MrwMRy>d(LkP>a4I$9lVZ?$9G%G8=gp)|%7-U1=9 zaL0f?EKCq$$E^}IR2S$dFkrU)<$~=*#d;R&KY#c1H{{=;ofE%7#YBCKC?H2V*a7JE z1;0U>jIm$@Ffr%HaDCWp1l7pGD`I)WgHW}@DCwGm~4%;|~u*v58zo17Gd*N1+ zVZ~4sNF^c`jS$08cI{#ZSL~4&46bRxK5yB~vTV=FxmAJ{CU{O4wYw=k=0Az)-dq=Ctvq@XZ0ZZWhhBe!bxZ2K-P0_Gbrb-;U_?5|-Y-zA#(Dw0?GE{Ew>+ z{2$(_E7t0`5$@mC9H75$dd7^}_zk+lc#{G0IO{>R%e2=`u^>G{HoqEH?U-bD;;44I40gcA+Y$4LCkIFswL>LKW!KlldU~7T z7e8$;LdBm$w`b`37$o^+MK(-SHl7K^xERN7bV^YmO;Q~mODukYolcS|i)t~Zm)%=< z*|p*&MfAhVid9<|hVImq}Qs$BoR*3y9keL_qHbdJC4H%Q)C zSHWmA4E@Ot&`3%GBxU!S#6a6<4?sBo*S`zr8v;)h#odgfOY!=kCu)Br<=>Q6+>lK- z6J3S>1}!rV1~C3-sloQl<;*2z?=GX2jVCk{P^Emo;XQv;FE;?zD zqvR_J#kP~ybsf2g^C;%3swxwblxJQ>P*RFfPBet{K%<96&2renjrxZ!Mcbkqmv@V5c7Dmm zNGijdmxsQc9JG0XYAk}X-$AcOYRnkkYJH!|&{c-md%s{Lt}XR3JtWZT>{NNkxMFI3 z6=V{Aqs$WED3%})c_er!KdL*q);@W}+E9}r+s9+Y9CmrQ(B{`z2fYvA|MwWfpkuK2 zzd=3{#2J;wFGS7Vb|)n^f{%kFxjP*JzSLj)s04)1>9}Dc(MwPvaqHDodr2470+Yl57y6H@qe8Ibeb$VU;(J z5*|@5mDC)M7F#J%TE5RpEL^X}YF05*BRXoITROME9uiFv!8TE*NGx$eO<%eEm+N|QD8l2B7TO+~Ac#TDf-l#*F~Jm?QMF1FRxH9vRwB41Q+g&utXBKk zFM(?8pyHUaqRG`0g=#&|$)2kmr$Q7|ruoqjo}VAv;#nu1_VHnRxald!ZWap24G!IY zeW})CGmOIZp>4Q~gwI{ths{On=Ai1~utGn?kZk0PW!B^c6MdJc<&o z)MYzxR&C)Xdi_{6(>u?4RfKp`#UonW;AxQ2%F$d;ckZ&lp9ApQx&u4HpN6f0-4qQ} zoO4880oh33ZxE#76K3=t?$ctz5sgOXCiE9>2Monb)KkweW~>p6;`@n z5F@LHXmiclm=cWo*{0a%JeVzv>nsrYE6R@V+9I?X&eRlJ?3$}YAl)Kc2%C_36gSV}f znc`9UV-n82&SgYSI@|#WZrQJ%)!hc<5xt-Gaf;2Ts7QF$j@88phCYg$&lo-Wd>Gq<{xlNlsXs@rT=%_EY)ybILcNI=VvK%Fzwy(d!tJYmV%gsK4K2vYmE(P_lYG9~ z&38rW`fuUnvfCy}=DuGN3z6nf0+gY42a4a~1`Z|FojAL<-8gx8V9{Q!PgN~%#whqw zO^)?8VQ`j~(Ug-$>EqWv)SZbce(snP|MYZBwoCItf4-%r&BnR z5od{5jXID}oTb<$?@21Km2R3mCN&?o%ISMvSdAaK!`^zYAr|_-Qr< z5=*_pIbJ`*q@904)D_qeraAX)j=PDQ*61HeWCSy1bD|XJrX>YB+XTifZE9Kt*3Ukb zJu$EUmMP3S$+XqT^=4i4m(dj+Ckgz?oSTfCxOLW9V<#$k-M!obQL50S?2)EZnF?uM zw8p%ncu#pBSF@QkaA*2wkY|d=JxYeHw$gc|0>%dg#-Nt~-4*R9x3yIYQwvMH^_n{XTRXR8$^giiLTH3BO!K<1ph~Y9`8swq?F)mC0`mqMEV6K z6L5O*)8|s%4rdU2#Q}e3JNCk|x+um0s zyJ~rlGH=j@WGM-mNxu{e0%iHeo5ZipF*hR5>pk8n?=Bd})J5Sp_Q!7wfhEs>VjpYh zdvf=vS-bAKNs>xW8ey*Z2=%_?fAXMR$|zjIpLNQ!o?&2|usa>*U$#4BGA@d`O7KbW zVmm7-P|?zn*Gqp@NxBt3jM~sb-i~otKPJ#Oidw|5PetazvSgy49Ajb{TAnZc=NCET*<-;q|~FbR)r=cGeQjC->Cnart|+~2L5#-a>^oOO2d%cnR*jZ zV8?O3s$o54UM0t5iB1MOLGB`TQY9qeco8D@6GEZ}7b>lEeLT*-iyq!V-i&4|ZN(P4 znq^p;Ph{vwO#Ivs*Xom+o)S-s=3$@3Dg7vQYOm#T6P(?hsd1nm%w6=S^Go_ZXEflu zy&|_T$U=)rfR^(VKN@1YxIDrmBSs&yG6Y~5&qlRT?xk9F(cj;DXyT`!Dxgv%OZnsZ z509rHw>fD#pVVW)Q9RV7+~Mty^{NXOVT3?&&JAw`K8um~l_|)|bK> zP03KwfL1ye+No#2M;b5hwl<9W(<{3gI;E-zTRQtGahBWhn39n^FIyqJj{_n$S|hsN zgOPnQ9(gY><(AVw9ryU#v|I$He`avb_B=pAg7jlrIMqr%V$eJ2(c>jliPL%pLC^H& z)&m=s%Y+3Me-D>2r;oTs@g`v<)(qk^8!x8cqj7zUoO8DZy@YT0`^XGDIeSR;GlaKa zkDp3DiaWMLi<-anQgCXG;LcJP&@U5w-58Z*r7>~tJ_|`iJz{A^_C>gH^VFq5$mS~iv_I(O^2~U2jF7P1T{_DL)T23pLJ4X4Je0=VHUdw6IZ_!rq-@G^2d9Djhrqb zi@I0o=PBK+X}K+iIjdVa2j#vp!4bP;WVC83fB4{*V^K-Iy;+bqg1fYW4@udrOq)ew z>Iua>P`1OF82+HEYqizz89Dgk=?Cl_m&DBAR*;)NKJw0avjvo)@HWZn?reho$V~U8 zCjZ)z(Fv|3fopBl`ZHHWU(xpx61yC!iz)0nJQkvtUn(~vL2N+IX?L^7&pxM~ z(>Q?{{$li*5PxKXlVuG=HAZndba$DN|8b5b(kp0-H?uhacClU&#ECSpy~-F=!bua( z;tk0!d%D$w++cd(2jq) z0GgBw$4C>4Gb?QY$*KD%uqt2W!Q+aCf z+~=ZEyfi10oXV3G%${yp5B9&U3g!7-`zSijznu%wK-Gywp2wJJV)`fS`Pv5D+{Za| zR%R%Yl!)Gt%YfJQ0VPmhG5Y1+VlO|b((>hpN@sWh*miUWa&#e0mrjso_bbNiH{Hh8xnm+`Jk>YWpws18{>pN2$KDvW%J_ z<#gD^^$bEnWEeCXVXb4-dZg$YBriEe7&9#RgQ43v``@V`4bM`1vMv zencVPr@M!lP6^M`bqmM-F=6@tgzNM#`8t0cNB{qJkN?E~`3<16!&=Tu2AbPY{K#fq z6r5Z1oow%)2}p#_U6_}`0MYEt5YY{_n=$=A^xN-XXCYV7iK7pGgXXyQfxa>!`z}l_ z_(I{NjE^3WqLo1M-2}nfPqlOozy7qt3oe?r-)A?kMSvXyq_X+v-;j;X$&`8L(WCZbK_0wgsfHzXMC=#wMShE_Od z^^x$UT!HF4%M_~oYMPMhe~n}OZQ}d_WdG(@m9qC{y|ufE_gwZo9!{=s|7fPr0of6X z5q>T}EbL)^byN-&uSjmrH8m)4IQ!dGecKpw`hM^b8 z2qHda{B8*5Cg~r=rFYTXS?H$OjA*Tmbgi@ECQJCbDyo1i+OIk1q~4dN5!7Y(f*-zA zvHuNH9jn1EhGo>+Z$WAUa9Q2E@2GC|imkcKG*l4wUnH&Eoi6ans$H^8eyTVz*b{9w z!c1jF=X9x#Tlt?Ip#S!rk!7xGF-`(32zm&py)Tg3dcBmtIn9w`-H2nY`>EsC)pw$6QO)7VhJcmIqlN-yp+E4`EONLSlJ@{74$N^gXKTQbkz)jXS5O;$esp0Y5ptx-sVSxMMK&;T;HPeZhBTdN z@G);ii`fSC;bZ284@-uDY>(-chy&ym=x7cfMUXjJ4aI0lWaLr*^H%atTMH#(_Vdvf5yUX#BA+IQu5_5botaDbX07(BMTKrY_n+UKg)G zE5K%3(DdtPBS-IqJVW)a1+hXZbj7&2;N!8!Oj%nv=6tE~)X2=g(y6&(|IsE^HA@b^=An47$p642 z$x7!oz8RI)O5MuCpB7dzdKJ&*oz$}LFGTCQ*!=2w>zFjF+7N3M=PA2i&ngs{gA(-> zDT+89+HHyG5c{A@{c#C^p(SNT2QT+y?;iHJ{UI{Lm+xO|s|a7bOu9W)GZgP~YsqRT zag(dR+dMkR3k0vsu9|&^e4WXKRCu O{deF050I?i^ZyScM$@?f literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-6-Multi-Head Attention Mechanism.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-6-Multi-Head Attention Mechanism.jpg new file mode 100644 index 0000000000000000000000000000000000000000..05a0cc858c1a98334fb912583c2d55db2ddcf87a GIT binary patch literal 69112 zcmeFZ2UwKLmM+{#&WhwLL6E5A98{7>&RHdcJKjGu!qhk>k7vd4);^pJ{`4SXtY-~JSJaT+|a-Ij&4|x9J&utsv9@ZUgG+k5_ z2Ed(rD5&>PZo2?<$bO=u{PqF-<%4nu6%8E&6AK#$7kNX?UBDd_RMb0YsOacuXvn*L zk$(rE-9smO@I)HpzS?U{1}9=3|F|qH#-|nSBFOuTQL1F&&O$;c@vnIE#SK4RnJ z7Z4N@7J2quMpjN<;f02#mbQ+rp1!G>xrL>bwT+9bo4W_l(<|_OP;f|SSa^KGhs31h zk146yIk|cH1%*XlDyyn%YQc5&4IQ0b-95d1{R86@lT*_(-)AAvmDRQNjm@p?9oX^7 z>Dl=O{POCjTqpq4e-R7$|Gx{`v4-Ngn6AU70HO$ve z_ZfKnv523>WmU9eGxDk*k-Tvp!y#qjgEGT@iuRjif1h9h|67v%FTwsM7X*NZih`Ux z)O!F)zCF3Z~itYq*aSK2POJ3W@z?G{BFZaD~0ReLv2tgwn;y?TI=lJ}&e*V<2zwB0j z?mvI-&wrj5f1aoR|G^t!1#qzJ6hGS6nz~Q>%25n*oFx_ZT0QbynbcNi4f0Te)w z?gzJk3O8GCuyj(Cf)46rF~I3C{G=5RjtjkbU!9rG)-qb|=FVL9u&Lo`{fLYX4RczT{}j>^j^6>!B9&YF4W|3D#d`DHd`3vL0-sGUn^ z>%P~^*?#ys2)Pjg&`k5@ad?BsCcnugofy)rF}}zcVB6XPWQViKHPdx?-xgw5q7CtFdP60jBjv9Z zjncLqn^9_Sy##Otay<~quP@rs{u--3HfLr@6Jsp#v9y)IhiSa&$*Mxrv)8_^h3f-) zudqgIeMS5e{2_jVKAt(YwDd<&iz!(P-PYd1QV-*0rM{5g56Jus3vY(nv8_HkYWb9J zY)BRH2}ttrlc(L`^+_`F0}r(3|Ca#K~3 zr_p|I{7C;?QvW4O7PV^$!Nv`7lUodttr_|H00b|Rd~rT-GJ*;u)EpJFT~WDbb22;l zu;9UCTn`J@poqJ3A57t1mU&(@0Ae+n|MudNnsNPskPtMJcySUC&M$e%N=u96F4zw@ z(TM@SoOUz9Uv2??m(Y=`izbAcj+_)avD!aT(WZ-HNu6m+e^3WbTuJ2=SYM5!!i!{h zgThhMu@;i?u#phNd5+=6A^$wZujlgw}5Lc+5zd|82Rj`CB$8J!Yf>O$SnY40?7tx zf;!M&-vXpTcaba(1cYFVF{j^@%T+UW_*ZR7{-Kw0m9Z{4@W;mfAm3gt_10-P`_L`mk8S-yzH@Z~S1%coja$GUn);)7 z-j6WgUQbNee)@fEcG&>%@o66q-g7isqjR{!eKS^~NtJaMb^; zu+{Ju5F@#k4opV0>zG_UD$e}U@cFzJqO{()`2$XX*4{ZR5$<`(!VD3-H%C zhif3hs*pOCKZn$ag0lwS(dPG2+;9a~x-0zcRyY|Qd<@CLas?uFu^FjRcUEr!ltXh+ zm8*wHsx{z~1A;B8h4>bb{T8YAk7|GF_|<~MmiV9V_;Wn|+t){U)lTZW9GGN#$H3L} zT4oCeeTs?wYw}@2lI^K(D?tg02oo^4>L_fSJ74bcFvRDH87q~a8B;XAdv*Qf&6)gBdIy;ChrhE#tV^%(Y;lwd%r6VmZB)&&{WEjOEW}mXQ~k z1g22@;_f5F;<;K_&7<|^*sDXAm{FYVCZ*T5MQ}^Cs-}#+^W(Pmf(7uv>Vv8|G|Qu6 zN{LUjG1;@$v1wxvsn}vu3J73W5@!G=7qkXb)Pv0LaFWJRr<*Ut%MDEnYgsOIY!>!P zRHJMz5wc85Jm@ulSnf`ck!csP2df`iTlT5!AKjk|z;sD4PFz`VK-_D^-tlUaIU37z0)X6zhN=Xu6!#wRcYCw zYN6|mZvh(WymvIRNk<={340&btHaQPLoWVHPNn(Oum)FweX!#_zo&jJANLLSOLKdd zmqMy%)6x&SQmY+!1s5HU$=-mL+1S4{z&%%nRHPC*uR@~Q$GeO*E!Y8-4gNty0tZs7 z^rqrNkMSi=RtVqo;R_t&HrinfTjxEldsX&@LG?QertpfNIKL)a5Gnu*Ot9EfeUCXew|0jI`x{MR|rBzgq48?SS+j0lg$KDN1m8{0yBuZf@6|5Ep`W9 z(n$X1@!w&aqVkwrQlAp$jUfEkE6PNES1Gqezb1B^hR_TNfv!lf>-Y z`c#g@CY-&bk2O0S@UyU$)~39+evSY%DqYdn_PDE-f4=>J!cfG&dL77!ltIJ!!se!up9oFqzz%hQ!Euietb&&3tE6{bI9cVWk z^gx?^V{tPVxq47>ShM~9qr%IKY-&X63fb7!okq&lykDy4UrpNO@ExLph}DcNelC*# zDw6O29fC|-GD31qc}8-NP7)De^SuN_3hGNM1AUhO-H=YEw}p%RZDQD3ey@a=;QH;K zmSe}?K``L;gVtO|C6Xnt0A)RO+rO(cmyjK)bsDU`mLUmQ86jc`sy1&GDEDAW=yf1z zdQ04Q@Evf!t*naCIefTkG9S@ndt*HEgEL3l^G&>Y!k}8ZA~y|&ElYd4@Wi!)gExOr z^S3+Gs*zwE;;*mkbpa}9s+cM=BvE9c`}r`p-VnxM4JQXHu)1PTH{N&L8R5fmKLM-1 zs5NVLy^&mx#gf*P>yV^EU2Ee6eY9s=B7;(SLew`eKVB&sEs_|h9T1q&sjWW%Y5roA z!qd6NTWgTEKc?}HwYjBZhp)P6 zLM=*PjGwY$%;2ulkz{B!dwKXfMkAu*N7Zvt64;YHd0H=@y(Xu3Mbpt+j=`SadTg?| z%PZ|sW}}1QUBA*VeKV;pB`S5EZ<9{c&$cCk=~IlpHczJXkPbcK3ViDc+Zf zhfmO}2ajr~Ad9tS;$Ysy)}T|UX5j^{p_j9wALDsOo3@Sh~V7yr&nc z*3Wb%rgPun2-YN7wj~l1yhx51h;b;q$@%5&!ycm*=d^ItyD%|;8d6cX+4@w^3`x#(*R-6VH{ zXl)yUW*{}Vjlo|H{OV5JQE)yJs`oVb@=C)aa;$8j5mZrkbvvR+TJncjK3AN?EUZ%X zzTBkKA8tKbgOI)VAz5w=V@%!H)skzl>Su#eC#UomOxP78XOTx2Vs?`+Kn zYs7*m0dBvMl92=9dZAjVObLGIOwLRdC2r3N7J*Hh|`P(-`i6kdgRzVCu)KS5%M+T3$AH6)?8MrC^oFx0rjsn9yT1zci4 zZUHmIh^HkwL(L=d$Z5j(E4(!W26i==1nuQCP8C?aikBv;i^|ww6x+w7!MDjN2`vj@ zYrE?!<4i|fov{xJNzHxHnCp+;;j6laNGSlybv+q6YU$NWIp*9`_>iI7j)|l#N8SwO{j8HN@kacIm{55tbTLg=RYi>xL|_>?Rnko9oXXbU%Y4*KZZa zUlr(EGD1!qb=w8W2IK76M*mWSw1d#8`C9}?!o`>~%R`l$_`bK+kRJ#N+jXrvl!-Wv z*A|vl(f_n2%3OLI-Zr%Q)ijbbcv>WlU02Ht@}k57Xq2VOqQe2pEwtA9QG+8ZXOk*Pf?)|3{CWEn@hBBkcbhx{83DgB!NCq zk#ysd+33XpwjxUs8QZtq&$O+H&oY7Bor|3Cy_rRa<$UVY&Pbrd5caaIr)9V3CbV{rWz`+hdU=~&9z6pLFF~y-x~Hhxl#Ayz3$a|y5tjMQ*e|%tDLe@8;PE3WV5Y! zu}XejrS*=;7NQ&h6^$a)Y_6R+6BpTe?LqRbn2NeW6!C`V$r8km&kaNY^zOO^;MSzc z9t<2PiaM>E=sr0vi0b4X@Fj3`(TB7~mWqkL&efZ>p3x7^@x{4`Q31C*`^$0k2C0f9 z80-16?W7aI^U7yYZxVgQ6yL%-2Gsm4X}i*Ea#jo;Myy~4mmQq_D8(5*lu-maK+RPv z30{!C_K%eTcwvTIyFJV)OT*M3J*XDf%i-z7JLU)zA;_ZvGI@|ZbqFK#Xoy;G4Hjef zEi;&ve^b{O?HVk;`X&FcMXtsr*6PakT52p4sg^U^Gc~&EdReMtiqyDZw7$U$>6Nl; zrkT?yy7mdHtF*Jy;JgB(mtZHufr?s>gvR*aglC5@t=!*1%FfMQEs8HgF(~cA@HOi6 z@?P?T%MsyTl0~?ncGI+&sDt=6)~+9Q3nzqN|sROJ zpCCfH^q^eyI2Zc1E#DjVQL!gQx;yV^Dv~;m7kBsq_$&gXpt`JmLR-Fts3Qng;3u64 zF%@XBEhoDgP(wB9x{nzLY!>(3W+VV!g`XOTa|wJG=7^r=dRO0XRZmHeQdMEfeC&bcC)x&98{hvMLc(zGU?5Ca%!h&d%1iiCkN0 z_Qwr|9LNi>G}c1j@3cbhN_W?a%!?q$Ij{8jcuB=bbe0itVc*E@rmHs9xBHvGxRD~U z`7~A^{sYv5$P50BcfwN~Vlt}^7~7PblY+_TSAIx3z@^*4;Hg%X5=DAPI29CNycKM*DT8_I78NXi(F}i7F0m3rhdjA zdAfZfAf0~c^2jC1wJg$}PCG>S8=eIOAAvjQ+wPqsLSUH$I}7mxSFc zxBeh}b{%Cow8B1Q*_(=p_ffQkelilu_NkEBTre|hx#R)RcGhL$TZvhuvmSxNOIaqk zT!&PYYZAFiSssKRM1?*GL8I1==_I^-DDl(p$X_mW6g3pMBS|zqqLZvHSBW9|Ddm}h z1cT!@EL3_r>Bu+DbkLCN1UjV0GIbPGEy8Hgw8s zvuGrfpJb*>H1xqlDdW-9DT1Ln^qcGZxMx!Sp$m+k(J|>t)tZ8ZC9OEc7oPEXF;?FH ze#j^p$3F}YZ_kELcscl}ctW?%sK}%8CY1{-6QMaAov7(fQZ!2;KF$rzXB&w{)`B7u zRn}ym&pq&{d_aqG)Q?9<{PTTuTVeQNjmXU|poUXcD2xn$GuP%lT|@58NhCGOV%4b; zucArlryQ7l@e023gz|?fa5B>q*|w;X-OHj=p{L=%ndJD^7y8; zA@*Ymxd3KJ7a238N5w=v%1j_Vm4=}k@}{AJY`B8zayHA?iuqgPa9@cfn4r#7sRIp> zsHo?IRE3XU=iZ+SNlaB0euvwf^dvjZ0VC9pX2Li*R}}2=iYSfp$Ep)eLwYH;rD!}t zn%vAww!PA(_Dil_UTKiDf*X2Xw+%asM0%7wgco?#g62@$Q!9=(7ZGENpPE`Px7U65 zn@_e4n}kC^KXu`9ji*rg zEmKo&`5;}r%)D)qXD}q}$nFNb0ZMj5)|a|D-{D5#YUx_qx>=chB>76E@~u+{r2%Gy z-`i#Iqg3{4snhs=%k=Ws51me~U~C-Ez(Y3s^@@E>AK}7TJs)#eV)NAe-4SO_w0jEl z+8;u%SS5is?2ON0)GaKPlQlGoJzwEA!&F74B5tXj#D!m+>eJfa^nO=GzXcFqOC{*S zRlL7!K`gS%*#;d%8PUASjPs(sy)}7v&#yu{({Modfgm+a#W|s@6Bd-b1gkj#>2;TM zLjAU*uIUpWR9K|@Np-hH@;#vmG?u5>&+txpn_iR1bW{{ zOt$Z0bf!`?K22FlViAvUvzSeY9 z=`*@iJ1+9aagt5ZLaz;{(+c(A4y%@-$@=M{Ci@0fOcix$0XbhHc`(E*#hZ)lYj9PF ze4Iw-WKu?zVBYkb*R8$ zFz%a>veu88gwKB=#ilC$$Zr;j`x7G_5q~ohw0tEI(zUH;-IB4zBqt6o<4A1kkK`bK z^Aq2jmr52GZ>Rs}oFz+}8A*A0Fr%6A0&crPh!repXQZjy)jZJn6SxX)nmpI}Xmrip zSP%b_&P+y{^Wr3aLmpS~qnXy}Tuk$b@kWE=Nu+(=?uo1x!$OHZJP2B&^1juy)@0H8 zi44bcHw>{mAdYFA9mUm*&~$t)QE1gf_w{Na(Y+-Yu5qKMEzSOi^UBCxK07MHH;-*c zBq)4@T91qnIUYrZkP)W73aMA_c7@n1q907bFDDAL*z6Crz~7k3zLi#N9fUFPh!iv8 z@qF!L9GxaM#yYAd3pn5=+%6Up4a9X@zVeU6S<0^ekUE>n_YrBT1^?Av`gyfLG7^mG z{-;Pjcof~8zxUVb{;Sr2Xv!rj}Hy;p@{*C)5^k_F*?dYIv1G6?oc>23M(Od?SsK zRNY9Z1!t(5rqO-#^L}BCoxIGdB)RST>r~p%+OZ4;Y6?Qc04wUmgH-m=uGoQiSudW) z^azFwWA9}eStIK0#e4N&`Dq$bcBrtuUP>zv>-=GZBn~hDbYFjA=g!pY3B>5wtU_H< zI{$K?+$&|LJyz6rA1y+a>3K3gltF2sbuorsI*diY9K9wO$#4+VNMd-yn_0WLu_ZAc zOD|XpV^r^Bv~Mw8d$Lz_D~!%ylHZoNx?y~uaUxY{ZSm6}Z`Gr=<){K!CZR(1C&m)B z>X_IsxYT$@;z2HSM-^iHw*X1ORyYT80#{v8CMTFuS=~J=85~|k)cei%X42-@i<|VU zfh_dk?tMEV>G>t+u%QODXnrAUrg=fOz@8@wkov}mk-Ixf_nHJV;Q<}P6da36D#I0t z>5sHjzZ|vQeMXCzMA;s3LG!pbhhz!H15BenV>pT|y9s66e&ig$&X4A0v{Tx{Sl!~C zoMFj|D}`_B&X8}+sF>^xKJ_y5nRtwh7I+O>=5L30Z@?_lihD?r{EOJ^K74bzjj-R~`c$gSd834)yC=kPx$vfHP zNx~~+G11&GX?9+~RPccz-NqDf2aJYPPXF<5`5UYIWO!VbCuaSwssjEecxG#>LS+-$ zu=E^(X09;kps#QlXO05bfg5`uu^K6bCW^GhTbC_XRB(3FqC*&16{x&(R$Z5&S2L-h z_+Y4q*Y>tcSUt%$&0bu?3i(aqD3sMo|xO{1Xa`7(< zMh}ah1dB3@q5I%}>)Hem1oELQYI)E75Eq%BH!*6r1<+Hl#4@a%Qi$n@LkWo&Y?CR| zKNT4mNEdBaG7Oz7H1(P3Q0P3B5)wP|dQ%8rxUT^deD5l#75~V=_}J`qtU57E+;$Q% z%Iw_yGY|V@(g`^}9{hf$w-mZw0YZF5eCqPB+|Y5ks2y^m`t3MKovx3{3a5oZIsDnF z9e?#F3Q835y&>{?b7&H?j1T~p&@BkH>uTt(ZCp{^gUOj3Jg$(NW$(eGdqz?o8EmO7V`J_S3eLW%@;8UtnkapjF5cCFfZkWnVP{Qb4AeJI zK6H)B>v?H)mq`A=OXn`+GbB9_hT7GT{yElYUx`Q5gF%`)v{88?`khA~AIiSK#a*Mh zyzc6;A-q6m1u6_}+RSKt1-(VOi5Z2XgHHvlZMC)0++C~~)mA@)Yf(S{;>cC|B@{ao z^;cT2e(0Bg-QOIPzh?>lk?*Ar1-cN!dBZt9gHG{5jC9=FW{VEw^6yph% z%86;2%4%?}o*7TygGhQ0aouHSl~RkH#k)rvjm}_^{IJ&^B1gPl>gnPcFAJ$ZD~0@i-cB}BE#kk9#&x)pZLSw7e-vZJi|)jQd%`u#66j= z9jmCcQ}4n717p{`S#Y4xINv8xMwTTj1{{MOuhc0dFH=O2TrkHxAZ=<#WW+1pk;|5? zEw1>=r|aTjj?sMu`H~eQu71->yXivCQ7pnQcs>Q&;jJJ5sYH;0SdFJ;+ zzsf7pJP?o(JLKslq>dYAZ9;IBvLiP_c@kS&N^ zX}o*^JKAeOYNvP+1b;7(9Xgzzjy>_VNelXeq6XI@LBg3SGPCa+2J~|zZA!Upw%?=v zTDEf|v`*9QeOJ;^h< zY3Y=jC@rgs<8a@hc=y=}u6flPbjp4AgPn;AXK=vlFAR6Yw`S=jr|RcxAmO66BHMvSFCrc8+<=p#2PDE zIxtVwF;JvGHWoW_ub`kf&&yD$Y-j-2*Hwd)R=y{oRmxJVumotvC63hSX8Wd43%d!Zz%EX&f zA>2AATI{#Z*YB9=6_jB>9TvMeG%h&Fmw`sz!8NJu%-W8dyUYeX;$+0$Ac*mMQs;G6 zs?UllsXbbq<~47y192o3W4c5VD2yc(3PE`;~3 zJ7d63uRZ(jjtl7<~4_qhijZbT@)&aFeCqibm>kdQQ+b@_?vSUccqBUS?HI)Elrrnp5_v7Aims zpQTe@In2$?&GhY{i>nLY_>1F4msmkzCIT-4c6~pwgIO^JDEB?vm5EO0opVU(5@i42 zL4CA%q~@J?dwRbola}EY5fXXJ4%oKu7-0NKyFW^0%D5Q7aRyl53O@?>0NT&Gy~JG> zMJ95mns7_?o>3|CCB>{~<0U!rWD;!BtD-fOhh|qzRS(f(t)*V56KD+f9+;{nuv>oZ zquoG;Qg?{2#9YxFNFEs=E#oJo$ z4pOw9h1Tf|Nctt>ZPPRS+!jVm6#+pBy=aOLm!CR=gbdy3y)*@q5L^jelo#O+r_yM< z`JAigjpzt*~eHj3Fta0TWZYRGJ&S5Gk3YGWdBKK#okpiyqf2Ty72Eamf)fbSyxqv+Bbs%B>7NYV6;fBS&ue-z z!qksv%khx_DrM38G96yTiW&(s3E}3x7S6U3@oo8`d&rrzb*P1nW=7=6C{{>L!9Tdu za?hKNY*z=Ijvmo9ef6POyr!f7F3`XVLx4aQL)MvndMlIg@kA3zk&Ix_*r}aJ50=B) zB=>lyP49g)1}3rf5wX}O-L4|RL7<{*8E|V7QL1@)+d-k-6MM&t=qPH ziBUSYGLBM#CY|cQH@zJj8#}11wUh6rDBE_veWURg1-o20&D52(wn6xNL$2@VThCZW z)Ik>i&icwbLWX7a5MGqe;{N2(#eqUFvX~#Sn8Mp49R*?j$rw#a96w3dUR|jcq|K~i!e|-L*@HF{j7ylD=;i2D$=~Mnb z^#<^3&fMrx^ApGkN8FC`)Wy_7GC zcw}_#W~MI}=oG~$%4z)fHE&(+<(yz}Pe(ZGUwq_9(Dz@F5&A!7^!_6T^51jsA3~Sk z{u*DY@=Av0cigba?-8}n-27<(Qe$8L9dW7{X06AU8uBBIK1Us(88uE)J6YOX)Q91U!*rL~FXqSlGKJK?xo5Mk6 z4?kd>NNjz}l5Bc2?Bz0D3_RsTCjAlK+4gwm7MEljA9bPs+_22o>Be?KzOQ{-isFhE zTmX-e15fewcTFdhX;lP>OVfmk7e8$}>4HRO9Q|k)G&u^Bwg(I9WMk#%;e{p>Q4>kC zkoP!EK&sp_@ZgP|#x$S01X=_p58V`nhPu0JhNo*j66k3?Ifi`(y#So)v-h1z-m_hN z2$OrydCIA-#IvL~zUX=GhtS=T#|-F<#XcJ{t}-%hZzZaZqC}>WinOE@8+`kg8W93a zfAYB)AatHdsGQxD^EAX5p8PPURAI$VMQ(hxA%3xAM#(%f@mbVC7tO4Au^l~ZBW<}< zdED-4hn@DL7kWDyk5ymwU9OS0z2e=UKoWe~=EJQ85ZiNort!+9#hzm+e{8&>i~240 z9{zndG1h6aRJpFfAqUMLn$}KcC%H#`xzIYb_jFM=05#V;1NGPItb= zm`tq0s3CV=^r{8V8#Z|ppqVXGpYB4M;%VNAP8Ko2MIR{s|Sc`l)Y1(FdJimDi~q7o41`x zuOKbuMD^uRO-*yc+az~mN?G56$Rl*uR@^2Nt}2fbD;Hbrk|^Q0(i#$@eBh^i=*_Y$ zb|%4v#$?H5D2tx3-1nRT{U)6wNvizb9;S-zF8(n!sh1!t~pzZ}sP`4;(bP1LZ@C8r)M z%6X);z+M?-jQa_mEjV{}rap({7qH7%W2^2J>PE(5qNe9+ojNhFm$S9qJ1s|r;o2-R zz|k#B7j2U*o-?VC5|{Q^;^fsKsGY*Me00UsphGMjUiOJAqD6;?HFT`=^K8c@P3$P` zTD#$ZGYEVCwNe2v8G4@7!de5fFDPa%33b>z9H@5g(Y>RJa_qK$IlZ6W?CAmX*u!>v zFu^oBzD+Ob;R$L!Lo(lPDS5kB@2cDi=k#8TaHR=@>(O!ceJoMM-yy@fRa7T8;5D>C z1Ucf(pRbtW?PBPdJ{?h8)|F{M4-bwWln&RwnlQY6*)rZ_NCCE3ULtr8eqolZ9pf*A5SBaAA();W4lZgIF(3T_k7SJaDGfVne z-VVUe$NVk!7Je==vtxBCb#>gup=5~uA5p7epRvR zMZU(YQY&>`Bu2x3uIlaSAzJ>w+p(gX7a7Ny=f?*&AhR9GQy~Ln9w-y&T8nO0B<6;I z6xl^=I;0mw5A=&Pw}2qG<-;p7q>J{)ncO5&vDpY6xHbyTm2++Z&w^SJcjS=AgE~ID z>KiFoVnus(ndn~>NiF?jP*DfJltcF-uT5^`zog)h>@#mIQb9jsPGSKWboJFzGg4S1dTQp z+ya)qfYb}F3&oonQ!{>@FS%c5Lfho`4@p9efTD+8G;YAaHK5z9!#_7i16F zAv1oUxKX2k=Gc3Og$uwGtFV2&TU^c((di}N4|~?ZnY&$kZeiX`O-i@?8;ol{` zNi8_Y+;E=y%^R-~s54<-ZdM7Nw0qbXAz^m!Tgi4bfhKi$g;>9X1IIBnla*1-SBf&! zwg#D@B|pYVj)NKH$ukA>ChsY*0u#-U# zojkv>%^`F<*`e(Xs&f|UeIp) zvzNy1un)S2Z#jIaD|I-}jN_n`h~{xud9zjm#}9)#raZs@` z?=bDhDw#~tHeJq4V@wqyP4=klkNRGD-SrM%L9m6c9r1S;h0%KVXIBS02WaIw*3*Vi zl=VQ+YHETPpS%wvA@)Y`mm(UIaik-0)2^7PRun>vk9dE1T1r#+P~jfIxfW`fBj&0U z%?(M<_M(`@G*|V4w@lp_bHKMH_7O4X;->`I%-HR;i5dvfmt&m-1#VE^7jvB_eTAtX zVrZpIgWMxT6BQUUya0z*Ws z@-8afY?<(3FErV9PazYGQ=%+nzqqb>gXix_L4%AZ+b1zlZUs-d&e=U3yO^8rsqFdQ zqZt{#c5UbDQe->}6n&Al(ljX@e*he&giNV?o88mB&e^d-t;w!)2+{A-z6B7{+v#5k zwN`B;4O<`N;h-qsM1AX>AZZe_guZbUQ$%i9%dRHZS^&E-U1NF}*7$0`bk76NV&nC< z`+<&&WxbW$>L8T60DX=VL}6*4BW7SMX|shq(b_`YA@OR^D4f z4d9C#`NQa%EhGt_EZH%b2!C_MX50hWZa|C+%863$3SX_(f+#IYi>{x-xB|IDB%yaH z-KHkr$xP>PuJB=n1e>EK#H9=Oc+Y%6CT)1zEzd2m_tFw24Wm9x%(HfeC^{O+l1CCI z3k#omf#@7lj&_K>5Q`PtR-vt6CmCF;Xw!J-%^y>C4$kpPX7=<3JgJvoY3~8$KO0KO@&< z?ox0}zsZ60?Cldy|5*70(V2|FWe^brj@xm!KRfR_Rc1cHc$-k@RN zM~t$XdMfc`n!8D~Xb#Av``ub_SMqz(ill^b_t(c?atg{N5=Tw8TbL%^v4c9E6wp}Ef@6bz2tU?ZSRuWS=_byzp ztf@o2(Ja1DnEc3TVI?ZBY*m{kCdCa+oYihHbo_@^Yor%@eh}x+E(KZHGMo>Rut#B(V9|UdMVeY z*6G@<*XzY5ntVu7e(Wo%Q#CCCh8A;rT_`j*#=9jbHjt@eeFPZdy{;sXn?Slc;GyuD zUWhZ?#spU&%$6_g{KxFUxC~Y<5hL!PC5eae_}LNx@OZGR;bAzWJNevdVxuG7hVC(&>NdkskGzYUkz2R{CTXFD5q;S2O3i@VdQ1J}>gf6B2H>b-QWih-_chr20|1d;#ry(mW%jmfZR#@1w zJfY@UTQAf&XnBbCuoFQrUR}K1bs_K-Ev}u(a`gUp*K2n7zvM;aGO;p;0^A-}I-D(t zFj@7JF`da!&}F=+)A+EEU1qKPVMkQPaHfQix>&OF7C^-C$ij|m4S=B_p`pX;!lT5i zfG6=1w1B59d0)RlAYjJap37fXV2HJgl%r=!JbOJc-P{qboI~*3v*4F^(H&L&1nknew}4h6q;bGM zXQRJg6RhUMcMF);XuC=+&zzs%;T`dCM8e(c(7+&?5xPYC9F4e1o4zewVQ)cgeMY2F z5tN2R#`r8Z@ML5{M-GD218GL^Wn8S7ARabzh4(_LGUEP~P=+&s*S^pQlyclsEmPiI zGTz>gW}-Q71i-k9$i5nng)oRs{PL!lU*$b*c=Y4n7zRnGB@*ZbDj*FIWKQ>r?f7o$ zR5>@Fuc8l;1$OXq%)n;r#0;$Tk&9L`+dL$iYEChUTjM=Uu*(E{0j{H48>l0fYWpZB z$|nV5!;GFlXL7QEu*(4(g5)8><`TAf&a$gNAp zz^B_D_|XRtK4r7qf7Ga;oBGjH!>3AHZ06sWgUz>?cTFx8)G>VIaf6Y7xJCvtc8&dE z4HzfO&`u%8Bz2|>A{QpX0@V1yaJX+`VbJrjL^%ufIbc|7N5-oMx+5tPd@xM)9gjDE zU7XMn8KkJ6pCIswd+4O2cRiMH^onviignuExk}Wr&PwaTv*bo{Ic@O_adf?H8#qz^ zW?~(mKmLVR@Y^+C-flPT)cv4oz^6f%YaXYo)9hMO*Yhb76Lb-k1*NhLl`ezxi59wD zo*pNTqzx?C%{CVD$91x_#!OxE;PkrIUoVrjuRsZCuLBMoKcX`E37-{2tTy5$6Ij~+{jMi(akzKxoBfI*2-hvmU~l*IbVGBx80Txc zw3{+Qq-O4(BAjZTL?wdy>?hV??iuu;J;fbI}1G$K8u zP{0y)%~-VS?w(HRt4;G1p*UzzcVHTYacyos@BAb>lR7$a>r@r<^Og=7;?9xgRY4fX zYz^+dT3(~Pk@5tGzvddzV~oYK=|y-rl?u@|gxFER328O4zh%7pytQYs>quNl6&_Zg z@&$;gQ(GIXP$m|~q}G#{l0yC%K%lTv45w0Rjx`1*>wz`8F)%t8 zse8cA8_6D8!f{OY^Lp-FWX^I39sC4m-_R$kp=Pn<|oH;$FUL!Q#Xcj@bXjL$`L~7#*Blr}7 zxY;`lw!&w?Mfpt~Ro$djPc-USWH60%7QC2IpoQR?!~7=$0!EI&s_~I~qqzilMO?7! zqR5lKo?#jnEB_6N^7q)U92njNo8q>X&{}^GbO}Ruc89ETgi*HAQ~hbhC+AL%&v+Rh z(+yI03;$iG-v7l8sJHP<^2TzSYLgXhnVe3O+MG*aCCdYClpg0$kVwCsqVCkkyvoXg z+-ImT2Sl@LC${>F<1x>)4@X*N^?3*7y!j22b{rvxn_7|!J0P~sMwl8?>1R5*gllwx znU1ne-!=FMvqr*Y>O%Dd$6m#nR90(?r^9ghO?vW0#080_T<6h9djnHX3Vn=j7lZ0?FSpm?;Aa|1UxjsszLlj}+WXzFKNeL!lvtz1|1sQMqs*bt1*2lj1N{0A+dm~Ts(|*S5zbEd06Aes# zq$={b9V0^q`}_Uc!W@DW90T7}GrcTzpXOC&>HXB<#-p016kBUb#+&KO z=$tpu&UbHWuZBc2x?_fD==2LT;G>tN9W&}x&Ao9VRUp@+rMP0ZZwo-p$QmT*kx96g zgKJ59ARD-kiEQAMD)|K2z}T6UCW5%-r9{#HN8MXTwcT#p!cb^w(bD4DqAgn7p+NEC z?zCtK?oNRM#fz8Vu7RRK3&o{)2oPL~OG2??@ArGRoU`{H=kD*GalbRh{rDppqXR-p z)>_Yc)|_ijwNjMriGOw+X$WXN6Z8tk)1Z30_KW)x0VjWDQgX?xgmcX*T*aoDEfqyG z^FpT>T?xz?NE1K(`?d0aEAR1t$IbcgG4;RotfRfy)w_Iy!oygSHe22XWWML=Ow)hg*a1HBI!bQ-Vo9nwq zzv=F>W^hXk{gFb~&3Mx?F8-J1X_Lwd*${*S~RqNj%2s z)eHgFTX1=NzGs^0yy{BtawB}!6jY&k{_1yqJ+bg)0vF%gyX`)wd7a-}h|9-mH zONinLPslBO>9$dl9L=DXj=nbh)nHXjC;q&8+Q$ZhyK(m4RMSX}a`ZszZ>nn(0rPnA znpQz#gdBb>rMe8Ebg4ewm3&PJkH5aHxj}+?#l@{XIQZ#e_p<3GCi?p`$9F)Cl=;Zd z>q(zY z*jgT~m++=NH1>j!hL=2y`Qdt_<$mP>Wn&lpQjt7$ZYHnmnraX7=lWAYZCgaxg$aD_pnzke%LEzk}_X#*t3bm>hzxM}8bch!&XC zU!-S@B7LBQWJmW(uhAz`jK*HG=diF>YVpBsS@a`N1@EGHGpmt%$5Q?6RJ|>zhT80N z_oIR5=3&31E8p!o9(8tAzOz3(lIoz9BORzNuG)P|wq}}q2Z>Gi-g$La4qEqD`U-nE zAyl($l(ELf?6cSYDd$0z?#3}UpxqTux!$ab8cgD@$bzR1&%i{e!1BO<_ul^w>+t_h zsvRyb*hmKu#M$!#G0;{)MMH`5DNFvqaJB>F?Z|2Yz!Ye{sD9Bu^f7-u!*Qkpm6e-5 zs;1WOcwl%qITylbnZ%wC2s2Q32cL3+q%LvjFZH4`?Kef{_Jd&!!B=<&z@T3_IF5b+ zDRq0lDj3Bq)!)K>Dc#NrV6=#201$QsW;gVxFdy~1a47JP7#l#~1r0@timn_U(%*$5 zq%NQ2Zs-WmQ|vTR>zE@)8S%rFrs<BQCR{tFlY9=@2tS!?K z>urQj)u!EM{0wPI$aic)0v%A%F{g64V!8kD&OW{VOrtyxvXCKXl4xt?t|Dil`v z^})6KknZmT?~e$kuXSm~P763IKG{f-E8}8sB$X$L35^1D(B5PZVKR{zWoJN;r33XT2N^Z{{6H(LlPHsUg zm@b(0S~oUF`}B}-K8?T(`A$ju{ZLz<88xCJs?ZbnWIlok?8>)Yz5RRqSEZTY>~TBK z+OH}@q=F|wdzsT@eY1=jWjgJEQIcAG^kzc>x%#^Klc5KZ_6{6RISftJt0T$M-a0i-hA*MjAFEH#;n=&lQHBFz}Q7b^Qrs#&zmdM)1B= zI~I)%y|);i6BcG%)X(rK?EM4dv4<3{nLV;8v5+(NYl9;}(G+#e`wjav)}Ibf_@a-* zNlSt)eOTqE2YX~e6s~uZl+JJn#&T7*ckdN-m>5tkSz>pp)O>v;^Au zmv{WZ{~Z3Coof$Jxwm*m>UWp}cO=386nN%VgWC$mw;ceaHDNRW2sLZpfU^KXO~Y^F z12SY>xtVys;q2Ba7NimoP#5|L{1&6BU=*Pgs_^cy^CRGg{=5aO(gd#=X_<|gjG7=K z82C%aHvFlLI1w?W#@1)XSYoj)*UGMkh$hZPRml@TBk2n1%adLEw#s$?132zq(oe^T zF1g@F2L(Gf%LG|(Etvb}+J;-MgeOpTZBt5R%jv!p65+KqoXJdq*_P{?Drh!LdHk+9 z-d5zaFr8AqCDE-|3u3EnMjb}PMsCkEviq9^*LS)H#TbOl*dezv*YOyAi%p@U&2BmG zLmwzOy#}tavtF~cl#WgeIj{Qg-mHDFaFJJV@z-ZWQ+2Oi>Rp{n1ZzxlavW^hMvT0t zRHV80Q-!c6*0cPM5@f4*=~`g?syEJwCE#rF4-Anz4@|(P1lWKRfXeoKqimiz-B&z< zz^jtq{2q#t5;5_{%I&T{?W7!8cP1VHukE~Ro_ZE5ag@*E6=8wfMO1vHRP58tdYT-2PU?r)!`oMMf)+iI=6%HII++q1W)j>N z6PeIO?v+TtV*jMinu7L6JmxK*X?S*>;k-1MqFa`#zi&nCGMO@bW^Yh95+T883hq@l&I9-{VwcR_bn1*~Ap@tdR3JO!My> zGbbyzaB;31^gKpy2-D1#A{-6pijJz^@6&Niv`cy4udXbYb95D6B5`=c!551n@`Jau zH`T#}$n}Y(A3mdGc8Pu@Oq1)k?vZerar1umSahJoK-wyDGAo1nN?&NBlgQNMa=*JN z-RL|&fZLPtRekdN?=HZ(NnW63w{PYFU?26k-m?USDW20LTVRGLhvA(92pTZHum)t_{K5>qgXq4t3>p03 z@*n8BY1M<0x|F{V>zixV%@*Bs#v6xq* zw93Y$hZ_G=9R@Usa8SV!HR1#XCW|{%R?{Q4)JjZgWj}i3m6+VX!T(Y~4O+dujp#h~ z6f))wX`Ro{0k)VlnEvLGI=b-MUje|W8WZlYOD|^uROa6nVwHIw_d@7Dy%YYOGbkH~ zOt?`@C^4Q?&yy+lHJ?f3&9mjF0Nelupq?&apdV>jp#a*Xse^_QsFd9Fp%b}Y{XGND zNeV39?awbo9oa*($ISsYGjG2tzZXi0f9%N%cw6mG9)7Z+AAQuR`IDlW!j%E9&#Pk3 zEFSoIVolWrv8*&@qxKDb#+V`W^#ey!kHdhzshq1_LexjHN3{`A!~dxYnrfXisqN2J z@^AzTR^ld-M;K4t29C0c$NN;A^kAN?FoLIvF9Iunz*1X3>*ne7-N-6^6QE4FNA7Nt z!R$D;;Wc4!A#Ad}+-B;xtx-SW?9@=kkaI;5n<&fW)k1T)kz&y@~<{{)zKtdVVZ$4_Io)>Qe+aV39ComxxFkPqsHT!Qa z#)j5t@21avui`>Z7Cp6h0xQR7`^=W2-GIbu zd|G@XfjH7wmvxwAy&gXyy2bq<)Nm+2tv|ii`8C_NG22^#Z5v-g-hRgV60An05W(Le zj2{R=2#>>GjPDzz{LkH^NmHRdnR2N0Ze)L{vlR=Iyp3a9?JWHlsL_dY3=9&KyWWYl zvwzBUv<2 zz!nlUR$;`JyBmL(qy~(6^o=0LpW0P{l%qc<3U4q5Sw&;=U$lma8s|xndd1dWdE^mJ zMX$B=W$QQ`J3vJx27!TVvcUB$oIrsoDP#38uWNnF`Tm#ELm(+#r_Sdj=qBNR5PscSWE?tDYELE6V|2kuPEzUcqb~G`lfO&yW-~B;? ztQ^>5r>Yo32x*!;WVW;B$vSYIQE#qHT@S*6w;OPYt6Q$cv2~k5MthS@?fG~Il6>TF zdDSV$r4wJ%&C9un>i3o{e1}xmL+7bgMp@RNPS66!>tbw-$Ja@0sVlMV-h~ESomDElMJ3wayQmUw&WYW`%v_CNHudNq&jpZnJ zlK@aYG!L~BD*!{AZbt-Ym(DGi_1JMUHrJ*?SQCWBg&01FRg_v+%lS-CDiZSDT^B!+fB zwT8hSwv~7968FPK-xgCG^(Xt&kVo7Y7y0;cYOBxC9pM$vI4$PYX5Z;$d=k-NOguI= zN7V%Y>;Nh6+`kzCgtN}KnK3NTo(rvCw+n4wU${12q zr$m>kz-q(}5fo^piGd5N)EYlP_RW(79zH1C*j@eb&pTJIg_TsYq}#<88@_ujP4VoI z)V^g<+_rSc8~Z8JUN9&ilqqd*uE=_PNoxs{H6*7f1RlM!r<2n7wgj@cdZrHa3N{mm zq65Xd(`K(uK@a1C-@;c&9c#|$AJDR^H8un5kbBg=;$P;^I=J&)i2dU#z%} z4&!G=eJbx*%%9xQ7kjE$N*2EO1H;uT|7O(TlD!AI)=4;Uzvp*rTn}~TZrhO9$-s^% z(L0S@x+g7l=(l&z;Em;;?69V54r%Dra}s};h2#EfTDiYZsa=CfCs)>*N`DS8gigiKg%u$J#Gm;Fm%!&~~f zM&sl<5=Vo(hftk2x68qQwKE59h?->*3Kv=AmXjgYfemU5Z%#xM{3XqjJ=?dONBY0t;uUZzdDu%W&YPVt8U_p52BH z9rlsf?*LHz+$f~=?Rki$dnr=M@5-x?;22}ZnOXqwcVovJ31Of zuTVO%tiQXv)nt{Tft6+X?$8cS5P0@Q9=^w{$F(zY&)NrO7C=jj#0K9LR`b^Cm+9ZV_zaB(J~=cPK8vu<@z0Fw7xTA^O(HdK zpQ9zQ!q3?xJk36 ze{^6ZKowev-e1$CMMCF&V%Dp(-AkyWEMQYdPYmf02BB#_C`nJchz`Xp{G3!Eo&WLt zni(OmneNp~McVsn{Qkfd=5tD!8%n>)J*Z&A_mK0Dr#WV3yP@pp2 zJ278S=*2}b1sM#nrv9u`_uGB>Vb+z#VUq{<(Q}>TOU$7f1~g05x)iO^l27b{2{1CX z>dY(v)yqb)SXmhUl;KOm3pY9ggkI7Tg06BBK!Kc9=XL5&+P;fh(+Li+NpD)r-mh31 zUn7dTK;9mumzjYeaj_cAeJKwtvGRW!DmSksBI7a;1147K-?w@&b8s~PFUrY;hP;KbJyyFqeN|q^OZORnmqq$D(_U)ph02x)n zd2Ojs6c6eKj+r$Ruart1U*8O;kHhf03xfhunet(}_3<7}&Dee0EC_4Bg;o~X;vD%0 z#_rWctU%9T3w*S}@vfJR{CBdc>UpLcQq*AQVr5>HLO0%E)ll%o-DBv08XGKJT!cQn z;bBAb5#45Gdy?87X_5E)-p`T^n6SSumFhN<+D;wKNZY(>2IxmfdjfOA`bkc*^@EmQ zOvG{|ie^eCr@ER$66#;EbM9OJ^Jfo~c(K}|<$>HwsR2>K)YU^q^$y4hd!@flJn+}A zKO#h6C4T-$1uUBUOORKm|6{w7@Gk-07zLF3y8ohm`Twfi8+>CzJ0$|rz8z93{w+tD zx(zN7{T^#JT55(EyT_)0uj9Wvj`(v}IE{8lSz2j1Nn?tAat3t2ftjQC5IYK6RYERb z=BI_3-Dhzq2i_)V8tWvhni3*}r9Q3fUPXrUKOgLvNirly2#2`evfB2Ah9R|$Q_f^j2zC5c-Q0Sd+YdJs4y;L1-_zS1cvtn-*Y^Meb|RBPfi z;tM4m8PHAE2D~f=wRvTN4L080O=#xM3dr_dmDD>Ed`kz^mx|#v3-6aG48GfSYNP&UG~qW-y-V!qNM<`5sb+^#nQN_#1LC0LXgVLP z{eJeEY3>7YR@=2;d=rjJS9hmpk@f*?g|CfS8$AWKJ4u|LPcfsub-FA(n?BR!?Yz*O z47t<6M5wN_Qv|3wuA#Svg2f43n*@djD1;Zco7C^M)I{a(j;xJ7biXaE?56P>pfm0K zTz-RDvZz9O9Nj29Y;QCjB_m|OIdBE^B3$Y%m2qqAmARBg`&SW-%y2JxZIJD`hg_u3 z_urA6A%E+7{-->a%rw!qT6L2*=BuW%nkllj&c5s;&$YhmdeYf=A7D+*PO-E%jCZ(;THEPw#9=y_qHfvyDh1F^Bjp9tcc!D8zFHrYKRz65h|;dA;a<+mK_p=e`{ii-;|v)@}ld-uNRA6@xlY@exsiI(Gj}2dET2WoA%}58 zdBte-i-#Ps>zI;csapdP*4Q-@=jHb*l3rtpXAw4FH2g2!e@)Uqc|P(4Z=;T~+2x}# z;nju0JkKeKUmhHY=B{by6x-VhJZw>U*C!&z(D-xwhPgGZS7gGW_=ek7dv=+h@eaFZ z(s!=fsHd0CJiZi_LdGKtvb9K--*FA;!X?kc(9NatiB`wGG}_jWUgDyA@pY$G!52d<4U%}SknH48eH^OT6Ffis+S)rNK7e&y z2}gJ(2GS=4FoG(IoJ+^XtQ!;Q{TBwSWWEYhUeG>dq6srKu&*mT7C5SBOVpk&-#OJ| zjTGpg$xB;45daR+=Ul7W{pt#D&*b~1ufhfeU}nRh%EG8hfE@Dn-k;QiipAX~av0eK zxt`wisKdf)l0hnjq~->=f*o?M1$ z&qA4+6y-B~98*)He&LAcs-^yXzu%f3D`=eUkv~$ol+8&xdRYp5tKWqzEPkGaDs>N8 zYr|+EF#z>Q?OLBwZ|S%KLR1m(K^wmu>D2-T;&)TgORUb4!cBaR^n92Q*O=tGr(XJ^ zxWh9X)oJUr=E_I`F%6kY8qOcKY2oAfdqk)vqPJlcgs%nq_aucn?iO380>swl&&sR5 za&3}7!gj6QwgaR(XFOWqE%~h^P%z&_L!8lq@$~y|4V<+nRq?OT&B9ji=0gS*A*0sQhRW`v8%W6|fdF){vJjXd|Cf++2u9@UXkP@XkIt*w z>Q#XhGEyJ5h&@LTFYY0uW*dbeFL6`q8iSf@={AGM<-0g{UUfu?k&e&-8%gzgs82SAt44LWM#todX^29E%4jd6Z&3 zj*6^a9X%qgSvzzA3%WfjAiD~$_tj6EA5Xj7Bl37RoU?irdf(PFj&doqAk;9b3k%WA zN!{nTZtAQ#6Chzw4}bn-~^ed^IfnU61z5V?gRez^%hQ_V0vAw#M^p%L?PbOpQ8NTbPS8+d6Zi zn<&j=dHYckj1L&Vm+0S}OaBL^{`$n|myWqAwF#PN%E^YN7J4vo+LI+NshNE9%$&;y zz`a?`ky`#Tc`XGMD|(ondYUBbdES@(yqVff@OOKX=(X!>x)$4YFnERB-i%g}#W?e! zB}9o(@Kw+U_pHg1OUQyYTCg_14|_;5je&P?4tYJx=qz%x|D%BjYu9pvtIdH+fHqC< z{DPRi@tcYe6VKiA1JE))$|eJhw(ZSl?POE|gPtvUY`a=99p9nO`7saR`r~9*N8#)H zPz7zYV`r4J6*z;jKBD_&K#T!=&FkP9*%>Q846;=()kA&F-Cz$or?pvT#~YbLisXmWryVXXI=N8rCdbm8V(68AV#oxL z!l#8O=*-d&D$cYVSo$mDVM`&d0Sq3RLrM}%gilQG$qpc!xRsVR#I|=qgxALN;8kT- z$a1R@I?lHHv5)q;eki$WTkG8q$cbu```Ai+>}X=BQq8QVDYnCdk)+ z-zJ;Hcd5HmbEb#ZIVUQ)Kb16Z6@MMPl9ENTa3C`3E7i-5Id^F^{-$M;%lTQoqtQGU z$IljDORuGqW_&dY_i03{A+)pcjt2~E>i2cekLJtc_Pvy8Nv_z)zsMt?B^oe_Zzt&4j|8NpVaNx*cRb&guca?w3+v%TkqTb_uAG12y5(Xx-+ zJYOYX)k395)v&a{5h^MRgB!%lt>q7e7cn=la!y-44p(zSnvA|ByZS&P>fXaZcnzU6 z_4g)AEn`oQqcJX!?Axj`flrm|R28~E`}A`YcMY$vZ#|lKiVw0*c_ROCT-ubY5LD*V zvtH%r1Lx*npzgq>Jbr@nl4WUrj_*Q=>7HxWy%$*XWi|H!NthILz#YkqzFB5ZD{Z@` ztgS_z2t=MbMtLaIH4C?GjZ?SkEYMkDa$^=J(q@`D=RQ0?QMGEAP}i{YW4zy49+6{} zI;jt2q1*4}4A)wk-N(tkIu56irU_q0M0@9VU;+K0T@^JoZh$x=P~N&d%!U?`I~B~m zeC|~yJWk29EmGx7Y@GWf;PFJ`t=is^smx2prj_ki`sQLz2Hp;Yr4WoWz;?o?z?^80 zj=t&7a}Iii;#-ICs8 zP8=YYXSS&5a5R0bH&pB#;`&FXjK87DBDPRJ2tyd1N^eH$}Qa|t8kZ5q~{d72y z`j&A1yTI~4c)OBjRPSr-_<1Ag1R)>_IcKNH?Huh3jbL6sj7Q?Q)J;{mK}?9PI${vS zW?<$R^CS$v>Y*?wE7-#}S7rJ=I7@4ZA-Rg;1vTR_FJ`_S2g+4K#Yf&g)d`{ExM z)&uS4^&b}!r)@>X*lKzs(bGP3r(liCKnEd(Eoz}JLl^x#&6ZGavSq4{$aT~0C@KSj ztaEmUGK$|JAvjEy6dQ;V^mgRYjpiD42g|UcWpl>oO10h>J@?#Rv%-pnXQH#bb_ltC zY>2zoIKoefh!&N~21-_inyi_+p9YNil8;=Q=Qcs!Ee?1!8GF!sIg3q{j zBikx#_PXnL0Dm16vPThRQ|Mb_Zcn_YY1F3ba-r6}>G(Spo@EhxUOoF=>V(lQ_-upx zE|fhPej5Dg>O4d;(pNCb`Dw3Z(h=+D_6-_*OS~N3pyuk89!LOEJ>qmpbTuD>oM&*0 zWK+bT@z)bzTI6K)c3bZ_BxWfuoi{TYVIIwHKdODDm&W{F_%P9AyxaLV%au98{FY*7 zZk#f0*n zNnbKRv?;yOQ{}bH17YJ&i zSCGbe*r3jft`4)DnHxS3D9=1lDWiD$>}XFkVO=vmA6$bIwHc5!*WW01J$SfN2514d zHAQbN>!dx>5O;e%T$lJ+`iMtMzse10OaICV{PX$wvwKQiv`WR*Vy5lR#O5P;$QCLO z9H&yzEZ@kQdH8iKxkShUJWEcgKg*=(Pm20u>ag`CWif!C`b~jX3dujI``*@cyv*EjHqcp+EXlzNS2>js@lpyHIbd19Bk3lT&c1(6Ac1%mjLb z`-pBOEc;F*{V(Is_LfdnzleZ^oOziPO7M4=VrEAxR z#_E@G>-&dJ>{2k}T_yJ73`rtDJ=6@WJ&` zPoU=Gk^TU1X1kpqm+lFI92oS(E-`3gEouEBonW$`9-Spz`c}SxJau^WuB8_G-etkI zQOl3eJ;X~Ho1nVIQdXj zF?sZM_(6P3p=5%8>GO9WSFf=onWD`H81HIx;P%Up&{(^E*ymSOMiKSJCITY@zn!_> zFDbZT>_`=ZUViVm+PyWF(ee_pTacf;v0W<{e@WKpNHDq@x}}nuVUex4d%axRc@GBV zFS-ohpbc!Sj7gv7uM7b>T9b@YIbm&iA*ESzt1g2FAV0T~D;BeX@Tss-$HbXIKtylKf%gkB-qHbKRcK*3E{cG@PIvWwrs0; zg~-Ra_v_`k`BU($9xvrV>;v}qbDxf|**4u?iEfuyr}ueL<&>^v?$grwK#!{#Po@kF zD#}>5Ht$LI1KCHr)p%I$s;4fSHhb1MbNc&+c8xl?>BswPjtj;+iGyiswe6$113uHV zsA7deJ3%8Op)YA)ImA9)&ryuJLL`&XL^4yW(^Z@|w5089q_2fUB#Y2P5~JQ;Qs8TR zdgl|#`LDjq@&O%7=i6CvJoi5$>1e5ccI?YEjxvj{R4yJ(cSJH zkvg!1aIl}z0SF5@l+?FBuo>d=4P9x$#f(n$-+@+jmb66&?Q^1P%L{jg=nz_kVP~ z-J;wn6bz6pQ>j`}T3eFTQ`8_j6#FuU`;zi;`CrFLd)?JN_<9ox*u+(?^7+J%E!g5Q zu%_B?Y*pe}{mkOmg247PB-T23U%Np0kVYYfz7KHf0m88LCqC@Sj~2YV=n0gYl*77$5TMu|qq@6~WJe z^9WdtPdH=Zf@U50eKP5!hf*#tu5#GFh4p-6LFCUAp~;E~S~}{%1{_t{v9C4*&&QXY zcMbJ|=+HZ;E>xRjB*-dlYvr((97M?zPmXcVKUt@gdifR*cI$$hR3mC-w)JKN4sp7W z#BkP-#xmIQV*gxXB01B8Gl@M|uN0QSrF+lkrJY%F7uoY>y)C3lM3RWx20;WXKA{NR zSv`4(=Wf9B{9fDo`q|T$kLpt?+SRgO-Pq<_l9GRF7uqt$)EmY_FTz zbk-?!t?OPhHtHfqoV#LA%g16j9paTQ(js_KwKJi(;(y<-yxPeA0*;WCq---@@rhJ} z7kZ%YEj>D3aj{Bw`EB^>0Xc2Qe$_OfJ@gcwhLDZJJvWhAsw%a-kSKKXx!b4lco$oP za|vL8N*(aCY^OcyI7=;SzQ8PE66HeNm>hNZth&47{llsz7!Su+<#49Be%Y6+Va9BC zV>SF<`?o=FL<1)3MMc?DP)DVbxqjbh;aeb@myVKqvdONG_hS7A!)h+l31+=SThHGK z2$w5ZM*U@zl~m?GmrSa?djakfuRdzKhsN& z*6-}>`nLz3kag|Hkn+Xn{RgqP+3MvqEeEBv#&Hif6blVDE%`m}>DM{VUe#XZj`lzaA;;8gKwy#<{5_tob&0`K*4z|LNSzniE z8w)xX+10p_P97-jLgvlnq4wqu%=wY+{coKp7Md-qMQ5!71pIhS6UQn063fvtfxAL1lj94Pm+}+Uaq%v$h!}QyYI|f~c0bJBx@69%)(bXW z>;+C6z#O`xw=Whx3B33NBjRaH%kUmq5Jfwpk#C(qTWhwFPwXM|v@binA{ZBwewf00 zs6vpAW#&+lxYc-R5xUu0KJB>GERsgT*{@zkNw*U2VZ<64;JW5h-H`oaCuwff<(GoL zi}{b&1D7d8UlLC2Ir-KQ6?abo7t%_Tg+u=yZ|lHE%s1WcE$J}nt}_7>R8#fd{?*s2 zD+Z~j0fJwr#3{>`LFvekekpp@(09h+Ot4nu9_3MQ|02&px{uxV;i%*r25!oZ=M ziBl57Q(9VVuJM247Q}~skn0QWTT{u_n^kXo!2&PI@%+>fSfHgiEN+ zhTGB4>fs#9CQ43CQ6$4}qH9+VFiswb>3e#w-I8lG3z4_Vh5tCf;8k^qAD|Us6bM;6 z#x~yh#)k0Jwl3um;+03pP<*h%FcCt{T5zhM`U$@d=uY| zJpCH4(S;@t-FGkEwMQTScv4WRO*a|py8nib0!^2JX4RPW*gY-h&(!Q4XJ(V$DiYHHR{HRr153BE3~JPRvg{pVwgU+?>{=k_OSp;)6D6L2Kb z3ah?*SDM&snhjccKbEPOKUt?>!Vq=gGfM?jw173A03qM?&WGbgb0gT^PgI^pWPdVY zj}ZT$F1>4*!-C3b5?8Wbufa1hVyV2}Iq$?-een>W=*PxO<6P-`%VGHX`)ChP9mfSo ztjor5Wm8+S^w#F1y|!ZAg6$=* zFq@cup5|+s$Q~f-OfU7-Ivr*FeYwx5o7%D+b*gx3??`a6v;&;iIAx7mr_P|bN7Sb7 z%88JrQtSb?uRV+#SyTshc|@$UH=Jmqe$^7esw?KWj<`*Idh4bn`CrO8hF`)aJm_wK zUFNmQ!5~(1pVZ>hj$pC?M0>Le?ZKh^hYYt1? zmx`*epG&FO^d?xBGX~%2R9#U8Bi-J!u6=>_(UXhc(+qyWb(s$@zkk0xw%S(`DKDeQ@l#1 zU}Mv?!=}44ne^|u9O!X;>H_c@vmXN_yz_h>j7Ya8;k1S>71%~a$tX!Y!q~`F{<*N$ z4@5W4;SGy3$Fx?cG}|y9Kl8`pE>?^Ux|!Vu4IG?uAKPB!iI35SZnBcf61Nt)zRXM! zaf9VE?8TzAt5tu5LcbW6-yOZG;Pw#W$O+Q_IICWkEj3-T%VMiTt@arzD8zAHsa=jN z@#d$H)*TNvuM8O{o4{4X9D5$@ea8%F`C!>qntmcWNkro zRK`hF`fL(}e92w3;Bk%Q8OHpE_>$;t&rS-1OXs)o8@dYdha31DYd>O@g!^9)nDM%> z=gD(Z)oK-eaPil21(}9V?6DQ>rEyk#OM_`w24Y~I^=S4xclc7TS;O0Jc!KEyUbM1(W&TVl6=?DD(=|SjXE%;b24?Z7HmRg zyvbh>6K{G3h-;A;>*GP=hS`%_(T5S)g+^xP3*!V!$U@oo6-F|&Ut5H`x8oZTgkg*% z!p%pIN=S57B((bP{L47G3L(b&C);mCcwqS#0E3~aVf--RAOR+OTUh_&TJSj6x%x#t zUcoJCZuvgMCYnJs^;GbiD}xkIZF5L+bQeH%9MMSYYczJoUsfr){|kG|%FvAG228ve z>&7e1*eLKwLL)0l;D9T5?FsxNp=;jq$O#=nyLR`GI(gv2;7W+L(f7P%J%I>$2|jH7 zSOCTwaoeScZOOnrd4-dItlv%#9mPf3T_vVpprSl>w?sge{cqKPHtX(6Gbg`xZ;ic$ zOX+yxClot-avqS*ShV@T1vwDn*IA0^4C8S4WA1;@2C;u# zEM8ZAzs8dAX37e$e6eIZlalQ7_5&HIc3)f~?QuE&QOWL=x3$q@Gq`;e8?4Mxo3#>3 zOi4<6N>wZP)FKBq?Jzha+{uZBr=YA0Zw~XUGOZPn;a|46dk&iz^#$L3 zBXMX5HbT4V7VGI5Gd87=kOR@zH+!R9eChWnZl;^M2i6Amj$E(a1EyFpJ8V=1$ zH_Gnjs555#!1+1eNPeeqy5)d9TSnEaAEqEJ=a!x$3TsH()0sLx;$IL6$gI2!^>r0= zalw0F@n|!o&jUv8b>8G=I?guWOhrj2FU}T{q)Pb2I24qusbREiykgkr?Z~U*%oeFc za#QuKAaCj{TgWIvsbOT!X(R3iW|#Tgcq;XaNr#+;`DL40-K@ju$3)iFoHdJ7;(iN` z;*_r!eJpM33!D$G*g+e*&vTxG@qJ>mTQN^TU}A^I8%oG)>Oa*K=X^wYx9~ zz!r1mXYqG+{~D!XDdigfaBxsq9MJ$Njx{PWvRsmfP30#EKl+`apoLS+O4dL-=@gw` z4mPq$*d&m@qPBFoGcf~O`upMLTwTqo)}@`C`>Zy7vi|uQSPPuAA?MH~Fn(*QHm_?9 z_AfJ0adZpYijba+EZYnfeXJ+}Cck^~+mpKdJ!&iIs=>uV#eKW#RRp7$ZoG=%!?6cA zP$l{}2el~X8^o;UqMYlw5{Fkr=J*)41y)$rCd~%TOqm*|KyzlyWVU*wXaCG$am1cL zm`($OSeuIv!TjtDg&Pr^PMWmQ%#sZ1UT%LZk(2FO>J#%e4%jN=b!-KUf$*L`3su0m zG{BmEcJ&~!(rDk$K_3y`Bq_@nEAB){N#SMT_VB#MKHM2Z>n&U%oBEr)tRQ;vFWMpO zxo8%cL`ICt)Ok~lS5d!#wc`d2|Cj{MYpJNOqE<1G^G34B(%*;F>aOe$q0rB3r75SH zyAw&kj9G&H%Tx|>Tqg~AcSm_6=NG!Qi505ja*uGb=Cx$eiBh$LYOpn}5Pzw9#(r&a z31|YO_Y(YI3nR^!=I5rH+t@(c(p&$9K0bawrMBA7>K5yyzOJL*>=V4)z zn`^X^xW4P-2H0J9d$LYT>=*|_k_5V!P)LjixASd|eMeRsUIwiIX?Fg;;MMt;bOd_j zG1p|f+Xp%!E!~vW#Lo(r+J<{+Lcf-zL7p^I$j)@p=R($IvMoT|+%}d3I|N$= zw$dNc=Ge}|0S{yp&4DuHy=ppeIok}l;_TT8FM1J(AhJkf z9`Vq1VdO)Sn-j4Odm=YgTN9rncI%wgUGk|pqrvBG9iRFwqu+k$y7xWnr7#n2v6{LS zD^q+^boAC1wV0ukla)bUl*-_N3#oz@s0o&Kfp9flc$>5^oIfQe#m-$;2GdNQ)bZ%B zRYl|{Ir5%UH*W-UB3TCYPOA=$`$gqbIhwxtw!(>%cr2ZNDUYpqzCO%pKd;b8_WEk- z1hR)X7(o+56?}^Hp2&&^i1QNa!mq-9pAccs924xj?eUCSma)lI&aavPgm z?J*-Mjm2XDeadxsNGW_jH0$xhG4^^qEsvkW=QaZGi>hG~ZqZ|t0$&711)SFO;(92y zLe)F5!;Tq=G=CF^3Nti-XBRjAdf%;pq!ojUK+`MBLq8X~+rDftonsIdC4a!!OuCPu zgMQEiVUDhQ-are6ho-U6Rd`OiHS^EgA71B^ ztL=4yq`rMVm#s&IKjT+2=9P2V+7|2OWF5VoeQjx%+J~6i%V?RjPwH}hSBV8v?&`NT zczL-GNOaB@wWJ+d(V5uCl5F0(|Neh?d+VsUx-DO{@FZApcMTTYHMl#$-3oUN8X&k! zu)^KlA-KC2?iSoNw@&vv-#OiV`}XViUcbMp#()}Q@3rQdd+s@Z@)bYAo5DITH@C(0 zvZ75>wm1SY&(KWMW*1=J;l@p@P3Yi_F5z`k*2)#y)$qBp{yB>GK(VV)J#I52&y0Xi zla&S50odv!^ILalCAU<*RCDE+^Xc27a>3q^c?h-;Tfia=IJMMT+b#mr>Dav*oL{w< zC$qG=9+>p=Og}fft-YzHF8S=(G-RD>{feij_?Ehkodg|hedI^gyn>a#|CzZ_ETW#- z@wcc_F1T6G##UL=%RTJW-sMw6ODy>4@Qfl^^sNEvmxB;k06}%ILOTB;dRFZMd@@!A zG#|M#{{~&0A2m2z2yB*iLCiA_4_vrff103g1IFuks_{w{?U~(T!KaiTwa&517wY%A z6WZ^!i7kIvPb}}*I}W^>*ifoVn>nURexI0^=R&$5Sg}JyA(O0Qz`thoZWy9N4V zzBSuzQr}zc<3sM==hVocZze#2))=pX_jXKXH7#4?8r$IW zH?+TDw~pCATl@=ltLV+^GgOC%-7ZRQ`_3SKVO9}vVbqaw)c3k*3Aq_wb8iT0z~k52 zchEaZ;^xY_-n5PjIN(8GPY5|!lOE{>}Ik_nkb)jz?Y{piEcl zbhJk6Hb1U!qUUtZ-D~ldFm2QPK6xh#P3P7`$9hX--VyIlM|89Q032>#=rr41w;tO` z-)MDKfuSh-W@JeL12Y}&7Li}c@`3o7RsG&N>ut_#I!-@@-27i=;3c{XbZ4Y?{D znnHTtFr?cRu_Dhmac|Xe=&GKJS^(d<<(vXSpWYIoR$hc3;$0dpE~?*V8m!ZcX)*{z zuU8r&3ars} zn^q~(&7+YKQgz?yQT0dovJ9;l`4pRGew)-as!#}fnj{-5$g-UE#Y>;bI~rdKjynyy z*-K=xwJmMcE7Hh6v~0(^i2`O%_co2x)l4n`U)j)Q{kA7jJ!UqUZ56*-+9q`cE>Y0= zdeB^!e6t|h8|>Gr*ZDZ|w4R0B9{>5JE|iCrQEn$!sJ$@*nfaDZ9&k6v=KK4y z#WVpnRcVGSwq0-^{9HvHe4G>XEc{)x>SK^8o7#`L&K6T_ydGmFgrEZS#7*ehJGYk# z5RdP96bbwaqA;WhU)xk7JZ_`;hc=$xRaS=6BhcqZF;+|f0#FC#2Lm`GUh*qtbr~zI zraZ=nIj61N7Bq-KGBqeWulsN6IhWMdC-$Mdy4uDGo+U>!nqs{Vh}@&TfCL?qkL)LY zs4ooR>&rsUh0-=72&xT>G{>GVC$+pzf>ok86)!&?;{3Mg9)b%zTz##Hn32D{xG~PK z3r!wc3&}MlXe@G&yoWV1vMwF<{^r;#(Z}xy8Atq#_eoiAdLufzj-o!$@A|@lY-Qq3 zDxkvlRv`rU=DMt*QZjtTl$GT)XJ0bn(d<*Dda)kMtLRJPO0pU(`v)M}-9*XNF+XjGk`sZUbZ95DzYxJ167_pW@c1Y$yjT`-BYeLo-^ zC?9L2;1~60TKBThBihQy>O>3@Wk<1&LjHqF-0qu3XZn})UhT|8s{KGuqhL%wi~)7Y zgYeU5X~x|y#cP%$-;I4U575ZTAn#;ol`YzS<1}MbX6a`^o4)T%0t%~D+w?i}om3CL zi(=u3?}O*Z$B;3Z#lE^Y8Nx|pW)#ujQ4oHW+&qQ#aPGxUMsnA+y=Ee58(FP+`2$F@ zRrRpp?MnqxO$I+xm+Raeq%9DzIX?wmtPTJ$15RbNcw-J)bc^}9 zD#m^5fKRbfx8BUK$9af!S#5NFwM+bNq|wWK^Mh94Z`buWPc9KCCFZg^LmRH$&(mF{ zoB+WfXddXM8v%1D)=4!umCTS$FH0FWv=0=-H~BwYjwM zch5Y;wkDGW)lZC%79_#u9@+yw7K30=_+{4bH7?C`<>pCAC*d$Z==U#l5Iyads)`rK zBAuhs4eR;(gV`OU7lM2|`)=v{`Q5D-aNx4wfcgvTL!wiq@N7mSJN;WO{A z_S=?X>lpJ}Fd?;xP!N(gM}>OByO%Z2W1;h&lgIB$`{^1}`So+xI(ph3z9qQH{Kn(nuSDfmQ~vuMypj4->1Ijb3^^!;>nVN{iA$ zEp3<=6*la+KLB#p$9@&nv~`H`3FFjb)t9cY0>X}QbE%vh}%^YI_%~< zZ4VvuczB5Arvv)H~| zFHQJ4Y5_W>I@#d7%Wk10;8+V!ifqAyUvGk~?P}511{RPQ`J(R*zdaXOVq)KbcP_Se zwDpTnv^x}IdvZJY(W+2bL^MF*P+z-@u;=5 z{DAt1qQS15S<0%YqTnQN&;tC%QzBidP0^9j{5IPt5-*I4k^S}*o$a$nTuN^q{!=57 zW)iKLhJelL{r2C4yeOrOl0@@z{uQx9C$n9DlJy{VOI07c^Q9x@W0Uo#pRbdat01A) zK_Xm}hAO8dsks<-1w8TodeG!h=uUm>SaKJUI9m*B%Kyk zct)Uvqvl8zTYH=pwuS^kKsIQj+sjyYYd_zqYxa-b;XId(T-4PTIS7n-}Ky`1wFrN8Pq%a!<)^mHsD9Plb z`h@Rx*4Lt)H|-Df`l`_vTv=5szyx`WhLj(hS0<5>o{NT9G;h$&Fl#Xd)D0EDPU@x^ zzq+&Cw;Au|pjO-jiXUSj7ehBZ9&W-9H&2#PFe&eHR?47;%R6fK_iMhTX^vg;>-z=!O!JR`LGWnQJJ&i2N1 zbi2t}v0KRaLf4nAiN+R`;!x{6SCxngNfg#3UuqL1I;LJMDoe!XESj6!QWKM!SZP0= z-6*lLt+-NGSvZz`Q=4yR#AmLbGjC{XcbpxLV*Pr_C+-?N7$0gEU41qfH=ux?2_o>h1aLDUtwh5OJ#A4*Cg~ zWZzRzD?5^3Eu(>54&6<2tlucWuT?1K?eXo71MmUD9pDCh&5Z0@xp#cbWCy+$Iv8Mo z#fMK4dbymhSh*$7xtt-vJW=*AW|fB=$Bm7hPAD|8rzW9bt%r^)4`|OJnxW=EeSr-1 zENI~MXfDKY@~H6-K*!nX*el=EnEeIkagox1kwb9t6AUOO!d6FKMk0Bk`)&b25_$)B=;f^_br3osQfbfKx9@Cw!dVz}x$* z4DI~!IyxbTvu!ozFF4R0X{7~{G72mkPp@#e*YtJajh6#i)CmRBB?Aoco5S{weAQxg z%_4#2iffby@G(&N_QeoT+amhLvNSJaXAYmer@uie1Sp37i5u2&xN2SEmd&)X@1MQ@iaF@FGrl_uNdP2mIvU9jn>sAAQ<(NRu% zazEan+cgIJ+m7nJH;}9vr}6AaTa(B|UKECdNkH^p{s4NrmNpPL7&#`M6U-ZVVkunVeeHi54647i zR8_;iYRmo=vLspi1=ZLsnB-i&DtO`gc-V1!bdH{CP8qCC?+1(K57A`|V5YgKk|<7m zP2bfMVksCeIS^#|p*^$J+!`TYn<74I>u848#`0+2L@;8*_tO*(5BDh}D)ixYCy4(5rb<8aq_PwozQpHi~l*jj(_LUi$# zX}EF8I&UmXVB@MAkJkkUJJxN|#n2FOFNa)aMzY-r;UpM7|#CJ`prU&O}smHK#7a<@RYt!a7lK!1N^grejZCWGA{~4*6 zmV4Kh5iDN;sJ#+%IlOo*{dN+|h8bKzT3Ebq;kPN|~H zc{H=WEfsI_Hd6v~)$dlX1kX4bd$7@=y-+W} z*T-ve1TN?TGZH1G3H(l_bK+4!Ro2E-f+q=WC(8DyTwh1FSJ26CqU?Vjl#sja{e)0` zu5<#q?|X=b9iuA&^7i5nkDC@9|U)U9;1Fv&BW#2(g?)gAQ>H3>Mwduwv{ z=y(MNqSU9qyrJneKBF9cd~kn;=EDt1+WqM}KCJhnc+y~yXOg2t9N@eth;XrY5R!dy zI+1w*RQhNfBE4(;bAoxV3CZ;Su0oGq7V>Os^Wjnq>IK@h8ArBpPG{U2OP{RHkNjIN z`0}ie5ZgMj%IthnK?qde2PMET091CfMSEShFeLt<V-~c3>`a?=t zh<^p!RkTXV>Z*;VhZr^=Pi#@4OA5X zuX*EZdUwmHh1xw1mi+V+bYCbguB}R3ffhhmJgO~gQwP+>O$mW0lI4wch^5Ytg>TW( z;U%7%F(`xh1VZl)ojEUv^ug>s_oS!%)9zcJ;g&)p-|%LB#q~h9IyJt(?YWe^bHDFG zeIUEdB45Mo-86Uy4RtXQlQ8=V(dGS{KG}b^ME~{L3)OXXJot5^mV2V){#QHu*Hy%j z&BuI8;T6n#?I`SY2owuDSAx{Sptwgt$ z=G>6t)sqq0^S>2_)J-Lie(m&S)w92E+Gu~@JWDa!noJBCMv^m0(*PV;rG=tQy8>RQ z&nDjRXYhU|8{6y*L$tH>k|;h&OuMl}NU!Ax0}PR%#JT>f3x)6=|1~=0-)eSw!|nh5 zjl}$i;Qsq2^Z!s*Pv{|j7Y`(tNB9DyAmv503{&wkH3=_a3f_E2<576=mViMS$lsY! zib(wHwEKnjuQ;vSwNlocITXYQGfjE>=*?q0UtK3+h7>T!V-)ScB1eoqvmac^_zi38 zAUK_cvayHf!}WcD0;(0YWdR~(v!SlO=LQv#nrE&&qx~f zP|f7TN(i+z2edV@M_qKuN|SMh@aYY5QR8CqPy7U90y|4K?HY@ zJms(GrO^Iwt{?>S_-}*z-zT5HAIibl<4;}xF0Tk}!Cu6?BH|0ZK}kYnur2(&FCao_ zB$`)5|MkAd{39nZRb+QGBrohkrhH2Iam_801@=-^mq`THUgiHJXjnTVmptJ zU2NCRwmC2__LUbt{(jM{fT~F_g)i8TfxG4Qz;TQHQ@s2^OlKnfUzPa;9=^W{C6$i+ znlplQ;=r24{wLKhrtmhAc8Z^41t3kcMM!-5bd=X|Mie6I+S^{zOJutPE`&Y|EZDx# zyvFe_faO$uGL;a2=JiT;)k2=hU zA5G(Ln&CUA0gYS3TpHBlZ zy)mX8Y{WM=J}ggS&FK@*7$xJNOVWWQ3UCQS5C}hd&m4CWTvYV;wMJLu9ig)SMAu!@ z8PKX`W*y_%gnIbu6*q9k z2zCbjMD_x?>}S^&&g@>+7+R>RoYL4}^=dZBL_`H*PjyQ;36#z_=S8?sor$`=&(bfL zxtYH

    7vqA>E{pJv}|WV@vy83OouHVB9n`oQ!J$5=0Zr5l8&qze}>h2^im zTP*T&_|HF9)^rC!&vrVs3I(5LuL^m&%rlg2RN)o73jKJZWM<#)RlL$#L<1jAwe3t1 zl1%)S#snqBm38$*ECbG{rD8L4GBzzvG}B^w);@U1E*~ZPTJvVcz$%wQo)vEVZ&>En zTc9A@)g!mqB|UT&O5iV|0Rg^O;lkRps~4CPVu^C6L3U%QOg&Ty)mT4#R}t$?A(Vd0 z>b0Xr@OKos6h{V#9rLdjy8jR>pZ}3m|2u`=|5z3CfB)dr{I>hwz&ZZ_piB-|t^NM_ zynRCTe<77X4Pkc@Ku+}A(oyb_No+}k6mOIhwP|&dq7eT%p#Q%J#fW4656C$DrtBFvz`( zSxeI1Rp3{=xo7Aq@t1DLWl3S$cciQ zJ_Jf$%Z2l@nVRsMrxw0Dt+ZoMJ~v~D-Ire6)eQ&|^=C#f(&!vO$^MLV9gkrc;deIJ zxnba}VF_K3nUP$ppuIUv&}b4WQCfp465XY$@e5K`e)zk{F>UsTdkxN}LJDVSDIMU8kN8TND9=hZ;v}=>((j5p zUZG2hUV$?1UrN0#&k7m_avRs(wj~pi)o+BDPlP|p7wc)uQYadq+qY9vududRv?vKG zKg`-u&7EokXNTH?s#l1=Reh3hl67u73U2~hF5HE@uQY3=;RD&29y9?Q73yjmtDAJZ zrBrZJVoSNRk?Jwlx~X#ttW<_-8&B&bTfS5}6&KmDdZ*#YJnN$CmAm8H{vPk>nT+qV zurx+pI}Zq7DP*5!D8rBmlzcAMG<}5wSPpVBro`_UfR2WUl z`B)l{zzYO1MFT995X&8Me(QgtpJV4Nz~R zEsN^XU*~M|hVXdAg^K(Z{At3sZ1b9$I@`XV+5wI+#so*?d1aq9t8!~S{aX^Tr7oO= z&!5zRTeWi>*o`f-_!42Fh+$*|Uf+Z5S`L~zr0+%>=KITl=XV+^8b`Wl%}AS7b}v(~ zwfYQ@6?x=Bnt}e-ADU=T=7K5h$ZH@Dt<~1EnUcX9tY~8aadnVigZTN)Dhkvt(V0oi z;3+zdDX=Uyh!e(zq^sdaOq!dF3wZJ&D|20{tEg8=LcX1!U^%1x{PhZ-oB`=(y_Qwa!R(iML9rB>)qjLV1N;JFhk7mKlP`7}C+w2bz zG9ufFAFz3L8ydC`N6=35pm-&k7j#R$BInA(WWhO(UP;zx(5|RkBcz{1uHCdGILa|L zQ}d;Q^9#@9w@PN;#9n&~+ag5SAMk5Gl6|MCWe;j7uZ003_8k2;i)XeY2PP22JAw*| zP-<}wzSsNPP@)3#LTxd(Sk{=cN~zN&wRq_wAg5z5g@~<?Ic!vVmK-fR6yaw(+x)qT@~~vJRpC&|T9a&^?$h-Gtuw0X>*TpX2c@KtuP_MJ zln{x!TU#`pVUW)Rh7DQ@{h8yGn)9&?=4-T26Ymg39BCXI$@4P55^UMsF#8`s`f$ia z&QXfE-Fi?sHbL_2OKPQ~gVJ|)DIA13n6&bzoMv>U6him-{?93_sr=QM&cg@-pQs=+ zrZu9#T$uz<+q*uG`5!#XgX6;3$+NtE%hVvC$d{;3ww3FTx`wZ`I((MpfPg?Qd zBRdLdn%B8jM;>*O>$$jJyDm4mS2IuB>Vd~>?zWWA-$ls24=K;OkEfih%pM67K#ajR zRjHdpmboU`9jP5#B|fk{EXc1XIXklsIv91;d$teAW~B_nWDA#F6JpHWl~B)6S|=Gf zPzUT^Q-z=MZwkPAaw`-qHURY7I~!)VF!Uy32Q+CdxM81EnQjSmXxoSN!b9GPS?UJ* zBw{G}nLxyBy>-G2Q9q~h!nDiy2-=(y`JVC4Mn_pPC2xPcJKy(=e@9@L+WPYFy!P~z z2GDBowww_Kcn^m4NMjYF^J0MV_+137@NSn7Qi*a2%0l(<+Nf|E$Fdh9*4 zGlxGlR9ADl*H(&(bFVczq~Y2*<&h?tqiHe`2XS=C&gmri~=)(TUi!%NC&1Nz%r z7mxzi=|$lm0}J{L<)Gae@*#cRj>C3&H!WC}pYb?9jlJEVc49g=Hv<*3Zz@V@VX&_i z=Fcdv26DU1Jw-Wx8OO5|eP?pVIFC-$w{*vew@pw@J=In^N36Pt4lgf|dT|Cf7H1<_ zSE1;vji%aFf+%R?HsgBpOtO1n0u~E-=i(on<&t+Mn43*VlafpgD|-v~Cf?mus5r2C zsY&dAuH)$8O2sd+>AyZ8a*Qql`_pws6bg*2rsMMO`7v_~qnxWo7oBK;AX>-Rl7ZB& zd`aQZBDe5)(fhI$LZ{dlyh9^0k`71~0!{9i5kzxOekqRRy-tUs0xd(T$8A5ZF`>*- zm1WuBlvt&?p>xvCXl=lUcFD|bKjI5UQ4M)#7&FZKv=y}X1==(xLq6Z$mgGfqn`mx} zOU)TU;3&lYN^qXk%_)l4(*tcRAf_|KoQAo{~R+Vd0DWS5E^! zZrVK6KY(}$7x(F^^UXn77`_ZLB|7*abh3W^D6U83oL2P`10wtbsPdlN!*X!}{9ncP zfB#boPcs%5o!8$>-COq1Uap6P?Os7SLSlqffR*Qk7J3{@dq$^4tnhf5v^O@uY@@|T z5r#yDVY4}aPu0x z*v*-GDfePjq)W^|xX`Hu(ae2xDTb_yMxYxgto%hT?1F;^&iHs?35SOp8bV)8?QW+= zbNbOZ{))^Qa?fwXzhDLzBGOYyli%&sr}=T*)lswH-rA(mgdEHMBqqqgq?Jw>`LuV=pnx-lw5c9q zCI)$xD1}59K;xTNMZ*h9tJz+86BVLF<@ygiib^(Nocv^N6CjnVm~yf?H_WjWVoPNZ zT0;p3ke=LgM)2*Iq;AvOw`Lvyd=KS7xo2S@?7l`+6TuV00cXeUIB+f_- zRxxQwkq|25EwdQVQr*~|{GN$oeM<1*Y7nZ!aq5l3yzEm4*BKEhwvn#{l3Z}}F^fO% zeU4Ep&XwGvQAPNc;ngp&$AP=*#&tF#yQu}?_Ko|e}mR)^jgri3#xPp!b@4-6RQ1Bj(b05LJ@*t%z#dTr9v9up! zcDTcIvGG3n$D2*AFP??NlmkPnU@(l+@w;%<>btGb36@vdtW(xy25Ekcrpi17Vnfvp zz>dHn)bymo;!MwiP4aPe`Mx;WB%$rG5Q&WNz2myvbw}+|Fu}W_ZUm(dN#vpOSj zNAA1}#-~jIX-Yj)q9>c;4!lu^8oA^U>DUq**J;v`|Lo_WqaC0oP;Ju4W@(z{ zw6__*IKu|j$I4#sUPq$)DzQy@DZ$3~150%T@>V@_-l0FL+8Rol)JOV!B89xcg-hu# zh+=-O!(ooajoxuLsGvKF9sJhB+#U-RKh+-`{l)p3875(}LYgYLy=hvXw+2K-_hZZu zAoTkHp{fADILkaNz| zH?<*dUo1;FqHOb$D2g|^?xY|AG24v z`Jhs@5fUoFFLl=2ij)Tp*(GuP98~J;Duz%OtdN0s-u~)bo`tLwg57Ot1NCiPb9wPI z;A845gVD7658rPhRj5U@RC@?q^oobSFbqG!aG#G~s+zA!rLb#V0pu)bu}E zHHD}7zPOME`Z``3T1nB8CJ<)t$T|?epZ5cJ@&mzS-pm3wAO*Uy;wP~*o!rjE{OS1>(m^C~a69+{G+_E3`-Qm!qKG_~N z$JQi`*amsI?ObO=dlFATG?!EDnaon|B|OL<1BCjoBAEm%K`a(U{ickSRXYjy3bG-# zMJ2|4j^k)6<3>&;)-aQpS2$@}mC`J)P!qo2F(r#SqDKAz%zLdd(TDU;e7Ar5d~G@q z#v%s-4H(iCjypydzd+`8I;* zwN>M+bPCLKAJoZBg?q>+N@-$t3&qxc5&9c}k{hT9=l5yb^gh}fr{UM<{@pOP5k!a? z&$D}t8M=d5s|c#$(h+6+j=_yYtyf;RC)zhk@P-X-jjhUP*3D*#?G?^k%rRn6zV&5A z<3JbU0Wr;j*p8u0kc9oc(~|xlKmkx$7_V1G$GaeTK)hZecGw{dJ&XT2Vd#8zp&&s% ztkO8`6IIjF-ZCi}w(8pkAp{@*-OwN0G2e$VQmGHWMoei2c=mVja?b}5ygxU*o)m1-x?hS{toARasHtH7JK+$yZ!2Cqf{9Sb z4fQ7sGkOXM31pVSk)IPg{Ld6i%3S_QweDZkL{!l=cP-T7@M5%CCTrqLD3fN_9UA?378`SegF=4K*K_svd+ zsD$Xf3HI^v26yTdpMB^-zi zkNf=l1Z~5!X>ySMqw>V^AApkd2R=1#f7@> z&)@Vd*W@{`0U%S?chxRS#~wCHYAEw6E)KW7syhkY%llIbZk@Gs@HO6RE)KlX%bp+( z3uGu_<;{kBE9&4M3KYR}U6=TB5D$qFxHkQU{z0%wFd1^F3=c!^d@VemOD-5JMGcZ> z+Ru<#Nix8V^kPD|zRBJIN8Hx1eFlfyqusBZD6bwnYU{E)_w$Oyr4nxzoh5o<-`-`v zuPCpu#n)&GE=puL_I&CKuxXz1Fs-+JK_gRI)IywnjKfPe*=@QL4-;@LDJoYydn2;We+9 z8^FPBoUY6uXDDewPqq=W(@hYzQ-Mw-$mm6WO0%*T8l>BzmsdsN;*$N|9s#V+XW0N_ z%S^UxLZ_GL);)pHL0Jd9$B)+<31gfS9T)eAu(?OTMYzmQmbG+M8;?i@v9Mwi+|UwX8;(xdbb8u0kQW0zSBt?V+W#Mxz*WhuYzi;!|N;hfdufB!>? z)-Y5Xb|__xFT%@qP;sCP@uVyOYd(=>)Y!zFtch~0m-|CG;5Toadgpw(uE79#6K72* zBCA69i0OglI)Px&rXMe025s#Rz>+;=Td3CNSCHhj-R)Y|k$MFlQ-f^k4(Q&-poK-M z&3NeeD+T4{B8QshDkFH@wPF-`_fSoQpeiZDgf^1Zaq&cy;K-$5s&WePLHut5#(O(d=+&Ptm>b1OWGBaUJv&onWV7n!HvJ&p1TZ zD@~4i7BEam3`)XGP7^3tOGMX(Pd5sOHU=_xblVWR3MhItF;uVU>xea~v~knb4t}J^ zy@xm_i&EF%YUYK!2c20)#_6=%1U;im;1qdLb#$18uQBX~A6wpg27enewwExJ@cR(& zBPCw4vNZTHSsKcsuNcYtSS=&&26MRTOri=a5u54Kc8#NVPgydXyO+df3RDeh)}I zLH&9GR@wSK<*xRZ1Xyx=n|z3S;8L=L7BORhH1aVdu_C3bSKf)N9bH@abMFw93bC&5 z=)5P+qsg>Mji1J~J!nI+;E#Sw`kyDnQ+?r*ZkGrIsOn=$5^%ndoHQV}Yk1GAkVySpOC zz2jVw!hFnIqd)poN~rKk;~>IbN1!Aigs_s(7Mmwcs&DF;7Tv?&7?VlgCEq9 zFpk~loCeT4Q}mR~#=@ax|EE}H-uwKP_Gf|qcJg&kx$&s!VlrN*)eD|d4X`VCcz@$K zUSL{7>|sIe@4CIGW(Jh zO-cN5h87EGbya*nk3c8yo15aZ2KVTvceYzvq;#gBt%!F!B6A^tKo5RXmqHz#Bd}sh zL3^FtJzrzkHJRBNTO`zPA1hfYh8v9t#oF3tG`apJQVTJmQx-^jcemtUZl8f>yQY<44~? zv}uT8Oq!2nZjAY#ctlXgLQ+%XG!nRB0=4=$WBH)KYo3I={T z8#b9gg^lkSqlniMi#PlOz}|#Moh^j3l(#t0>vp`02)e3!u1|nq0nbrx3WPpVS#}iy3$cA(qv=cwZrA9!yMUcIGWk z(Tl7;FH2n(TVz?AnkaT}`zc#ZpLJ*}^dfW(v5b9$VPIdaUNNCGq?mkX5p~8fH8v9%d^KcWQ9NFX z(`KYUlY#h3$kd)Mb1SKChAB9#!T}R{_-miv782IVa9DBsdw)|xT)8rQF~bV$7rTIS z{9)@jj8-GTa)x*Oo1lz^Bupp?=#)D4CVa@4)IR+Oy0JPWRO;~PZstHA%awxbowjO-i&D@$bB++JKeA_Q_nF4n)%?i`%{9N z$ej!&=xOzG&}E*ZH>NcC(3hVpWGOXwPu}Wje{r1Y;?e|EdW=lGn`U$IS@NQ5ry^9MI9IBA{#c0zgR`!_p=()0S^t)@NLob z-v-xm*Lm0NSuFfBTsEt0F}g}Hhoz$)6~1@Ic=zOXthhrlhXgt7WF1^r2vTrAmG`QR zO&JHAApIW5N*JbbCgo~<9Li$|2ob)=LUM~Dx>=+?c1H)mGD!42A-NU}0_~Bm6baEN zA*{LbI3;-dl5=6Cz)Oo(5L}2Wb1Y)JnR3iC(*iT$DZvAtJq&S2Lvzy%%3i_byo0@K8mBoHSvhUINk~tLhQi)S!m(^&G-NB zL>k`Non_b5r=9JK5Pr<@Bn)&b#-jgP8WNy*p2Pob3IG?A`}KrZRFkbYeW)G=&Z~bIe&u7)cy{6_C22_7@D8ag7qv ztJZmC-LtdRjjb&v(;;$<8`m#yI~o7n<4Y8OY}nXp79<)v8zk}zfZ9`WbU$U~<=Svt zV*=M_&0v;TLMI`K02xQ-hVNDh(mYW2Dp?a`p=nk0G%kG5N>a=*{17wY&kLrbhs5^j zPIF@#Pj;LrWn8tHp?4r}_&}Sd-xh1NvJpevBa-}-#WLPw^RGwe*(&~Uz=HLOTBj^^ zCfsP{ZG;dDaaT@07aOvOZGV0UBgB^StEJ%23FilO#8eLTOYvM$Y?p_tVe&lYQ-an%aeML$E&pxLL2oIzRA76W#@6}u z&MqsM@%AmD(3-hp?lKGpdauUNvg6sjPWVIc#E*)r~`2yt;WB9Q-4 z)iKqG&w&qTfX^jA{xxrYtA>So`D1nLSoQ%3+QRW@p@!-6O-m;^7`S0=R4QU=eg%t1 ztTi<3yXK{_3bt{EP*vfjv8B|C)e-BzgW4^+Dgfe)FL5VRT6h(GPGXu8>XAX$LqaYn zeA5hdOdYVsI#pD;JYxR1b(0JaLxLMJKXf~%Zwx+vUEn}7!{kwCRZzwStvFwk{93S! z*senGXi}GEuTWUHdRtTIAEggI`_jnDa}o}iYcY%H<2;)la~InX0t<@|`0at523J3n z+&z0|>Vvnzh-3`YtONIad6_jCa4Ib;=eHH971=sTyujWuabg}{)L~^ zZCc2aV^Ef`OppO1q_6Q|l4Gl_p|-OtnIL1&@v=fT^@&E<0c`9#D(;tfAdXcqKTzoWoZLIJ<*NWI)VN%8k z%32iQQ9e&Tz3U8HI*C<3nV2tajqOw9?SXgu`|Ikg@9p510XL)o2?3V01Q|}hWu4#o zpzYN&vI3^D|5e>}2Q{^=`ye7x1kuo&Nm7>XF0fPjE>M4BKS!U2)q1r(4X zRe_+A&>_?yp!D7eCA0vMVrcPh@0)jgo;&xwx$ii4=Ki&3_RL=2x7J>>)>nVODGX5Y zp`G`l@1$i-;+ks_01SLQ!v!*B)4rjiU!K6}y|y0ZF|>C~#8#VWX985bCmqe&%)=o& zto=%?)Ag8qNV8;O;1vlq>4N;CXmRzCW~((iQDJ56r8}UklYEi4hs4YWv!8c4etBQ} zSffbChT8G~ z8NArH!&E%d&e!&YLuFNZ{8qSdmkT%sW;#IO#C@Y{_X|}TJVWG@q>50*-gAb7XE*RX zr~}{eJsVxOyh|u$LsaSG=rW$>!-(Jmo)!63W$yVtLu^bL|5lmQQ+#Qu>0*+fydy)7 zt>Rl{dX|r!3*N@zlj$?~nEK}w%*k&=hnYHE-z;}T#psWX*7gfZZk?D(H{Yaw7h@2vyEfW_r_`pLY0k5MSJm$-Z96@yimodvEn+HC1Wd3wRcI zMU=46FI4_6MPDV^+HGF+(ZUFRk$xepjmv#*D!E^+ywB-Uao>IA$ej76ybJtz(edk` zxj_FAcRj5WU?b^}kV({)Wu4nH*Eo~S$HDk>^YeJ23ZKg4OpA=Esm?p)#+teE&&G7? zL2YtMJ2?Pnn0t+JAW-RB#-j4zx8TM8p?WoNFK`d;;;B6HRWIecvMngs_bOCM|=AFLI(G6A`<; zgo&n^N5YV?ZuL~TGloN|llR%IB+`SFP}fAmKCZAEnitUrT@)${8}{8;kj;wP2`x}JmwPb_SC?{lff6-c4!Z=E=v-(#c_kTC(c@WYjk5o~N%WT9Qyx73zDd?4 z5Z8Q-jO+SLP)t%wGR1im);xhdlsrTovOHYS!xl(rVsAf`TbZPo%cFNjpuiWEcNH@w zS(I!x9z0~LQy^~n#lJeoMyQS*GO-wDkWw z`uBgOoc~&MkQr;*Jr_?D91_t`HEyz|&J-74KZU{eqrjA;KZ#sMfsunAguPI%C)-8aTGZS#G>6Vo9H9U; z@4LoxNdX;v;;-6+)7LBsLp2=#RoKV?Z{y&>;EYwJq9t~x+B9FiFND|CZt=QDQMDHCwI*@!eI1Mlx)YE-zd!sP3E8a0nTeb3#{014^3uZSBOI@e%IT+E zI5OBN@%L=-$tWv^`!hzP2}*LV5F754H2BS9#bm{iyNQvYM(eY}q>@ zb4558o*jAY3p_O7_-6a&=WP$xYf&DJ{N)Q}a;|xcnAy>E>+j%qbnbUQlm=mnu3w$q$=ClHsZ9y5_N) zMF*ilX=*eByiSiEq4n&LbPBIo6)C%CZ zpID?0)a?|dDmPEc!TFBK-4y`ET^rgttdK8RLYHLgpYpBztD4Xl@7 zJUmW*Qk8qCFyKzmpZYS_#&ZtyOmJeT;_RXvEQbLXcwwNHYf6w5egadZVd*=&N@r1? zT+>NDAiW&s^!LpOszR`y0{TyIuwYn-Md*Q!NU(Ij5foKYJtx0OokNkiYRqax&cQia zY&B<`u0;2`r?Jx;R3ir`%t#5$Gq!`6{n3?Lfmuz;CgR6bOPUSaJ>+G<=V}W0!L)q_ zVM>powaGE>Rc8eiJkzH%-dM}N$a}~F)E3|$_ow3o01oy zv3F{RrE$?`Td$QzpGYiNDSzLl@B&j82Ttke7U2-)<9AJft=@kErofvDkZc5cq#_Y? zwJbi*;-BEh+T||hp4}<}n}#myAG#K`&R{5;&i4_-tg0$wMjtH-;5^+DhZC+UI#icR zkN`UIv(8}w)IHR~Vi{xrN5V!t{t#dwS1e{4H);dBw9>x)1}X}_QK+~{rF7Req?ap4 zB94ompoGRc)Kdba4HmvjevYN~#LnpQ@(+aQE1;PGbT0{^mZs8+dk+wq4&>fwMIxu& zV-qvr73v@&oRcJ`M;v}Cdf5;9)wBJGvaq+ccO2F+NYYX;)t*Oe~-*nT*kaHehJ zX7=)n-7}DHAOMj~edpRA?)!iD75~bo$Y&+~Wm8EY(29As9eL9_`*t&7ED_Fg-?#Ak z_%k<#Wd*KZ{otfC0 z4kCp58>sc>fr?j(K19mMUGunYPZaEkqwd2m^Uwb`-~QM6?%!E;=^JMWqFzy?;57_U z>WY76w}g0$ch@nz9ll=D2*|{-?+=CM1IWCIcLs~+YEX%ZoV$u)d~yD*k5l2_o3xiX zo_hA$$13!4wTRhz>r1ZX&TW16bpKI}rDb+cxp1W0+C7A@&Sy7&z18zKHokK~SGV6R zWzqWRr&zzX_tYBo*+n|=R{yzT_NOS6)l;VdR=vM_P{)No4y6oIhFP0nm8Jn4l{@|wl z6*stzSEOztbQ4?hww0q@ASr$LWNdH5d2_}JFhv2>yEVCj$g$rvMc~_fTTB$qR;~O< zPqFj0h<6Nf22wK%@4yu$+z(%bA)qclUd0l4!+9q+3l>~q{58}P^7|rz|aPxqQo@5Z*ou9I}XROTGOaImj_zwMc5w~IA?iXlhLKu`hv z2+yX>q}Ctf$EbNQSb6T4l9xr@SJT=V&YVouSL&c4C$;<-{Kt1(32R&*7csWuur%24 zljYEjCv-Sjp4nUQmBqS*O2!|${!k;~e(+1~Bj$#GI^4z*lJbYHM>#~kgL?2GQg}Ya z{4cGADWbqDt+kvi-n;GZaf+#AMiNh;xg4+m zotI97z}dX&JZl_%StUYvblqA;xxFd&1xU|nE@8A76YOFF^fFVjShugpXS@jLXgJ@SE zUOoy!VzzY+l=aa5GM8+jm_EFxsP<0Z&ZT%^bTQ&7zq>tH=4-}xSq3uNm zOwVGo9UPf!m97r62*+Iz`^W&Mr|8hj+$`OJ@AII$4{(g+3l+PCo$+bU^?46Pj&#YP zC2nL=@HbFc4w;z@l(^m;0yOYb2Eg~U30jXOA4dQNWC-FdqGtz*m_Cglf)P!aLCBKN zFk~YPLW)p)%W%s{+Ys@SgjkZne!&r=5&nn>+;dDYwDQN#9U_eXAOOH!_yz)y-?CSM zJzNhUC*p*63+33Wpbvm8{XeBZX3bVIw-p^SzqarmLU{(M`DBHQA%Bwcr|FDk&D$(6 z19s0p$oSKQeD%1ZCSC&x!=Hrk{yd|H!M}9L^alXy!ve3%%K%0DYp4&mYpPiwPiKqA zwZV6Qqcd?_bMf-*8`^#iAv+G(>rA zW4Xs}oZTN9euf(CJrvJd?hxN~Fr0&Eav0`?0c+E1B3!0ee+m}Skdw3L-$y7jrR z&AxW>NA`ww2kP1T>89WTX$)1@o8474=qFDK5A*>2i6O{(hGJ2FVHU$#+(PYFo)b0^ z>_S0wm4ej~;>mV*8w2F`f&PCI@SpXAi@Yx4)i)P<2E@=*8jo&rI{t1F(wT6D@hT-xs);G8ovPn_p@esSOTbI_CMMbI@B1!V;g1_lU(0sI4@XFzfw zJRBTc9Be#XTwHv7JOUz$t3-r^L^NdNBoqv^OpFY)^z=7aZnNLG#luWb&mqdmbBA9@ zNQjADTv|*(>b9Vez|TT3@bU492#Ky=y?R~XCjCu;fBS`Q1d-xh!NC~9#JB~zLW+S& zih*tgF#+ww#`xm}`s)kh3MLjd4lW)(0U_{(@@t?g7?_w>urRT)v9N$|`v9MVut>4V zZt~0GkZU}{z2!h5;2)KS$9%7(iBfY2#v=IKF#w-{iuyVYEh`&42Pc=1u!yLbxP;t& zc?Cr!WfiT*+B&*=`cF(>ynJP5Zei)H{C;$Na*FtI z_Oo9YAk4pX3;6q&j{VjzQlMW~u&^+(aDVm-F1Q(2VJ|P-pPj>+qmSR5hE?jYiS4Y+<4A>nt3)f9z$x(H zE=hVQ2k0M;>O!y9cv`biZ|U?WO%%}F9`=NOHPg4$tih5K^7JL$_!08<#Q0jt<`_r6 z*Zm)y+ow<$v6z4i_3k=24AQQ8yTQ;PIoL+f(NsZ3g8H?(;t$9vzVZI~IZk1k zQ5W!e)*M|1ZtS;y>t`OF?Gi^q7+D_qdKQ&bB9-yV3~jr~+BMglxc#GXxQ-o8YpSXd z8K&uUbTsK@#a>&tU%hiK=C{c&DLve1pd4eJD=Q+Ce3KM;w+KI#;7R41BO_e1lZwKZ z&tIGeEW6xWxT`MgI=nxP8kfwMd{e}`N~8STe!FEd;;IApk`2@SQQ-p$M71j}@+NE@^|8hZ?z$Wf5Ct=SxENQ6&`cWtC*JcM;jdYcxw zhZ+pXJiI5Cu2frCfA7ZD=cFmABMOXcyZlKb%XrTh)SRqqsXECwh)szEs`vVgiUgZz zS3L*Ek39;7&hOhOsG4Vo-ifr2g3&6luS;7&mD8(q4qqwks@xGv#j_KplHP_b4aLWv-*zCtUV)R0qoX@(p>a4l6vA^2@by`=#B_w4A}jxBc_n19g*6 z_sA^Wt}I2ymv+k=a=C~%4oLZMVCwYf7LOJ|Z3^}^EebLZizuDWE8FdTw)Y%Oh|5JL zyalFb6MSM57YuvYc%MzJb3D&G=$dUMe&|vL(vYY4Zx4*dO$niz>`@f6T|`SCGK-YO>FJS^>otYAEwp^WH{uwm)kC|*KT5r0^nH#a*i~rBff)!xrFc2P zOx_S6?*Wi75(|X30^Q^@6N4nMd-1AyDvYJUzW!0 z)w8mv#OhDx3zFzsb!gY-^KD+TZH(OR%S3|?g0Q0Yq7}y;f)tmoa%?7(3=0is8P5+3 z*(=~#zicoZAN%{d`=9mq@BdG!d9L+!#CPDN=fmq+)zxM?!~%9jq&Fux2xO3+iw0Rv zK?dxQE)C6*dj@Jg&qB9o54r_Cr0JgCplaHS#$D01AmT_51^sapAddS4^vy!K%cAzy?sSv0 zoBH8vMIb|C&UJcFUXYbS6vb|vGf>(;Z->S!`p#;3O^Nb1G=hte8c1_iNqzZf!PC00 zN$eF#m~CA2!CHf%cN1QQa-{nQQE1%7rqn@$G)jl}u8tiy%Ys}U?z_+SQYtPhQ7rtu zcN))(ZbiN;Mh-cEFJYvgd3@;=!KbYwcn13b~^kLV{aBr5s6KjJp)l>JqH_zAMV`0`EdWc5$SXFk^l*4>Db0@_hykVz7gKxoV{*T5$rgl67+U))oGh zC{aO8Z~`-dr+KqMo-C|5Hj6O6j#&Sba!8bOyU+yI8VqrW>nX-eT>Uuj4S2d&I3d2>*+3XBRmx=m0 zs0$6sBAv+4{k*wws5^pc6o72qci9|^nVpuzrqEDI1spqOPpv2#bW@!Y4VrXaL4#x? zDyHoV9tXBPceg{`OXfy{iuXEDfed!BTB9W0B z-`RQag!~~YUHw%3Fc$^Pv@U+w!G#B_&inmxr|pm*+9^S3(BT7bNi@htpZbO(IF5s{hyHuj##CyMt@QO2gQ2zlyx zE_MVI4a%R$xjy(g7HcBjoB8g+a`3?k3Jl*a%89iS9Ni(?K&hw)LUyt?P%-6}GbJ3d zd-}TILrrfwG{}e#e3X={`ZTiD^Cg9be&Qb!@#kdxPY(MOGVsbyv&jjK$ptyDn@h8| z5*ic<#fMD3gd9C68(;RK%1O@`6dEXZe=^$mDNCL{r8_;0=H-xSKwYn$pVfYH~#X#I8zQF zE{&8;A6t|E=y<*XW-74J1!A3Ti+$|1;Itdj>szK%Cl=C5T^CaW^C(87qhvVw zeIriRWDnnwoY|+dbEM&un|a_M9Cwq})o3-h;PQ8cW{=^#Ne6K-eG@wK_^0QI7){69 zR&jd4Ywo)dy7~)qlhp)!VNAwWw|<==JTUO<7oa!@Np*9{>SW9$8Z^Quhh!k>8R2d< zW*(LTuS60oY`YCL*;(5rYxM{Ozs^g_kLkMMT$Cj&Hb*xS9aAG65PFy-x$-J@nZa7$ zqR%2*uOm1a4brPta!2ht?DvrGL`7s5OGAPksshf z2FKDNi{2(kZ6TyA%p48+l&^r~Gev`FL0cR}FOc%Er@*yrCFJc9q(#WGdO!%WkJe<# zz6@h@$S@6&xaf}t3CbiQ9&k3m%A`nu$XF%?4T5lB7N%D-A)d(K0@bFdqd^THAk9NC z@CDHsJ%XF51#(CMH@OJ01s`4C@%eD4iH5Sk^-kO}uFj^sM5@S2 ziO`!px?SH7ZO`=^MwL-pC~a^v8uWurh5;}+EP3A{IWZR_$h?kUF6KmK$z&VIQ#_{q zX(Z@%+f06t&Y&6!(4eDqiGfk;+f;C2AKoZo4y_)p4SWYir?rM`@tZZd(oHuue_t`t z4~2U3Xf!Q+nczpHM)PvcY!$tlcW=K4TJ+5_WG41>>SV^Er(Hp^F7lBq>=<(dFB7K6 zB}B&J8BBc$nj~Qn1#9)WOibJK>bB?%Gcb>my4Hxdx^aMk(Ki&;X0Kwoe!JXjfV`cT zcdzK^Ldv@m%Jp&0^?<<3)r_7o)HAAkvtM8>qIZ2Y+V$PYA(=X#ZaFRShXKYw+j{x7 z{=+_dTpX6?j*73m&;0w9){&|}iq$f`5mqaiUH-Au;JagtsyUXm>pgw-djfb2@^0fp zEmzCzXWC2JXoHjnpK>sOC}W(^AXfIKhU4MyXpr)Bjp@UCA_}7u++W3QBo`U#-xY?` z2au3jkS*yy_-bHy7gC$1#~t}dNhUd@3=IlmhJ8o*0dp183=LY|88pvyX7m9}gyPc1 z1&9QN1K6Fmk&q^ENFeIv5@H^h&utSBd62#7VnfRk8YHmoE2;&7J1@bUmJq;lapY_F zfFNr2b5e8IkrcWw|D2!tGkz)cJZ=vTVG5or0+6H#gMpZ1;nE3k`Zo0|sXPk!9Du4MRXEN9LoS z&S7mH@UHK(L(!mJNe472wZjP83H940gZ?F)hn5w0;#TtL{6sJpgchU+=2S$KxnCkE zTHkMomsz7h_Gl1SIjjMt92o-H>Efzwp*kcvx$LRZUwt@BvZ2s;Wd0M{a4?~StSZ`_ zp-O*x%Waj&Whn9u5E`3|L(Vi9ly%9uf@~Dd82BF4G*pWPPqW2#9%;lFo52FxO08n7 zbg27j-9AAxur}75zWeVIFPv}Jpg}Yi@iI4uN8|&vuGaPDpHzovzg1I#p5I^eGzH78 zr4UJh9bNFt>-ayantF~OHmj4}ZNc(U)Yu8NJtcd>xI*2-uQ6PD8?u^-5JlA(L$)M! z$~3H3kA;6*9G@%jkcwY40jx70Wi+UL z6Ih0IE08m`_=VHa6r^n(S*UdAKGTzU(|$DQayc2lRdb}mal&2YV`&r#1bUYW$+0pa zm$E6?l!%C1Am&>(n6cpP4eVtV%{5#x6YKL#qVS@el_|g75$?|2J}(5b3*r^xKrvZT ztJ)gA{q*AA%;(C7Arq8?Xi)P!yy1de_27J49VKe~R-UoVjigYuWCH@LtOeHb{o!-e z84yN=51~PXUhRO1e#qBwkPAl4d9wab9l+YZREi>!_W+ovLqhdvP(6_9t+;G{(Ad?r zKO6nV==OUtOuyALk*=XwV$`yux$7~6Xjriq@TD~|-T8Twi&g-O)Jn(1xOGYVB%d_g~Q!+K$x1r2frmUqe$1R28w1VR(h zkIDb)a3PlxgeU)Ng5dnMM6Zh?<<}*@wfpS@w9t?!&-P8KY^BGxRi7^q)pi8e0ofEL zGJj}nn!G$%-lMYV6Q#AwqPoF%WM1;}=745_YRV5z8K7fD@hN9&V`X^EH&%6c2aq39 zaDb+%4qUKXYEC|yJ?iF2FBO$?Isr^ulqwLa7bgHKs}gz_3cg5hSk)YA*kx-tp|9(4K@!GjlThLA;5qHJNZmr;AA=kN9MfHnP5KX36 z{AY(jqL=|`T*C>@#REcf&y&jKsd6KM`F*8h3rbq;F3cjXg;o>dX$wXyy(6oaU&{he z+=>3SpQck6>q!jlAp}|Jb#LOnu2q*!~ zrxhWZClneY9|m`yYK9yM3etB;(7x?HN`sEYG>Uyxfh} zp10Tz*1{lM&f{)CK1N>tkrjTwhCcq0ENg8z{R_ieMR?eVPU#?-SGSc8P-Y$(?menw zyLPtp>ZYv0zEgJY$%gM8J{KQ#I2azEwkE0;QtdA}8(H#-=ovAOCeM(P$zoKA6DA{y zOtgk&W^9(%i{aSg)j37UQL_?_?`SXrZHPWoff)IXUeEKuX1s+gf{t_j3#8KQYVGDq zZTsq+lKB+{pLNCoCPWx~&B(x;3)UAPmoGXal_Ie|MNC6j%pcP4_Ss%?!j3vrD3Z{; z#we&hDZj3-yVSrUoW?6vz|oMr@hoAD4M=ws7aPt11O=Ge)0IWGdu1kJeNK6u?`3_y z6zDj0+xy4XzwBCMf@c;rV05u(=QgAw`g;sU*+y7P>qcC}VtTDqCr>2VPvfxyxCKbw zo*5nGQl)&coQU3c5RkK_>m*R>c|S?h)B%pw7`3x;fxh!3*b-EU1mh>g_ap!#w9r@d%MjnjzM-gV(!)>qXhMuJ4!&N(_gR`ibw`y)R5cZ6rcHN zik-{!ZObG^aG4NfCWO2df?id@k8IJJ?YZqqFER;HV25S(wr_PnwmU5H5X*!N#x}QG zlt+)PdR(sBS7_kqIdu}XMmS6-GPN??MDn^oKk|{$DSIcra|`1clDXT2wNiDUcW1{< z0{Z3{@q9uE4I-9(>98hq6fI)vdA(KRbb4{ZJ<;y0k&PRs z(BTj^f&_^FR*AR5(EK>+VzK_+gECB;-S?ynE#%=POv~a$1|2ez)b)gQmGO$>W!cq_ z&v_;uS?d!oI=ojK@l(5M>3zS_08C^~eT#l;+U2;m67Svn5{$5wK?euD9BCE%T6F+n@VNjIHK_qx56_^x# zZsR8KCBVNJ#{VE#^aa3@E+hEHg)`F`O>W^=K0>(QG+-t-B=3cCq+a>;s8WB+1V3Kl zJ12so-o|~axop3!9Hiep)=*NyWLYi~Z?Cmn$(AW+s0It1P$=JFwschY^yc3+_Dt}> z9qj+_LNN?bgjEV`0U0Ev3mCX&;*Oj!Ua>!e$E_bS4ZfVhIc$UjCfBMK3 z!Sz0hCtIZ7=MiA6OMZ#PGN}5cuU)>zVxgWpJ|s_x$d;x^> zS)`+}4gxthL1Cl`hX;Wg`IA;tOou2)tY@GER1@37ttpq1M|c+v8u)?IHc~ba91eaO z!|-_A=8{Doruz~j)*RYb4J_K2#f7si4>Txq?|ucnFUuc*4jrTTxjdvZD+4n55OQQN zk*Z_X#)s9n0>$?U7=96YHvi1Q)^JJR??$fl*X!SRLKjyW5EFK^%_#B__7`*y3U9rH z&n`a|Ke# z3Kq-#=zrL^QI)3iuwh9jmZVl1O-;S zT)GA5gp%Ygs&ho94Ljw;C%3%v>0t9S&8VosoOTSx^mvWovJ2yx-Ram5)b<;aY++)! zqh@JP8Pq6Z1W1@{;{mLJ8o<7WE^^`7sMm9RC=?%n0Ax%z%0M3-l+{tI4&5+!?vt7B z7ZHC(^vsRP+a;eWPN(A94-wbFnO41uPE)*UsTlv+iA32#cKY^fq+a;)8Ue5Fq2^3)@EKzHH|y1fxA23i8dG{ymXm{^9aF zla{TygHGz{q4xyJE~k&*lju+?XOKx8bpYtCqb>DH?0A|U%1QAvsq=6P27pP<<%VdCn1<%_MKB5V-L1q z`)dfAm%PfJA$=EwOZnF65;#(E8LH;AzV_E>^B>~=6%>biJc`sL5{DuzlT@|WxoN=F9y`BO?GWl1vY`(oD?`<&J)hUYtRe0n^I?+VIbxPP>g z!M5Sj<#o#JT63Cb>$y2_anvABxqZCcfa%S9@{mv&F8w;|;i}V&jVtZF=eD)j{2@s!ZaYxfwCNhNlNA`=rXk zl*MAEJ9nKqE|V7jn4gIJi~Qs;Ef}w}faG%4QWQ}3lfS^U6+EoqtajIXv1ZvPYu~tp z%Py>>9Qkp5{+?tSO561B*@?CV4zHo-O_a9T->*G>Fr1=Qm>$Y?^{1mJ=DvhBNcky{ zsF+;>2&0SGwDE|DvN^`q!MHTT9Y)+EdmG?b&q5DKQY z^!d2-C5JasdA=7B`|0w;d2229SKy#nmu?vx>U#}Vkv{F4mG>MjgM8FFOo0Y@nK#PJ zS$&0KA;dE*5@rzbT4UC{P$-keOM%%|mK*dX?TZf49yz%(O5d+YG9gSDuK=J;_yFpP zGKvDyjDS&JVA(=oWa_Y@5MbT`$=V<$V0RnEfaEz6A^(%fxqB5b@gRURBG@>a-qqzC zxb(9O6m^GR!zMG(aKC6skTVz-3eXIGv?eHwj-NGf0|W!e0)oJ0LTDV9Pl~H~UH$Mh;$-PkkXSKD?ld&$wQftp5oj00e8h{&%L8voyc5gf!NN@rtBTJ zSE9!){vIqsI-;6f8qkZU$=oB}Oj4D`c!Rgu91nZ44ZeM2)DN>%28Y|Jq~{d99t+Ro zNH4qgJF)tw$WqBpX8Pl!oZ2pveJGR4U8%1A{eXh;FLacxJR1VEpFuddDo;2FKRvk& zdU*7Q0|wd$yo`k-hWt}oq}Q4G`{`qLmweY;F9eB_F@!n}0p)#_)Cy(g_r$a@~q zU8gwM5@N1NeOJ_qhynB0mw%bN_($-+2jg%m1ddm9G_`T z?>Qv6y;2l)r-ubhTr|(gC`v9DEEjm%=0=4E!S|@o1AhR+^>JmXd`Vhr6LQ3FM9;_r zo2(3Dx@K|rHN8&-SiR7Gs*2kK739LXQ!iREbo?yy$yP!{VXnt>-O7?Tq8~!i9F;JI zHJRQ%(b|-V%jFHuCbWp=ajS5=mu~Lp&O+71_@lrADJPS^QfQ!FGG5k)L#OxL)ct9M z@_x&0S)vWwMv^pus|gvJv|_vg?+e@x&ueiu4KHHlu5l=d{fG@Rl_DZ?&Roks9@ ztLF+=d@{|E{wiO)ASD*)#`NuKvYy#zPulN^JKP#pD^@v3Ri`53j?5pTf*yD!#M`1b zo1O(4=PsVR7uM3Gw-I2;z9a>aXk($id|06<5{yRlF!Z7rh6@XAji(L5HLu4#eohca zB-mz+U~YAz?T>VNmPf*8IC6bvzTGnpGla2M9JwR9T#0lC?0AdoLhrEqcK1rlbS9(N z@Uoee^P!6pIm%!rWf=|f@Vt=uz%9n%ue)^chF1t2WTrE=|LJMyx=(*=Df#*}3gXO@ zD`k5qc730`F=1(P{bAmsS=PF_<0`$6$$szEX3r-Qr#>TZ!BAh|WeGmXbtj4CgrDWt zMOi$FLig%9+ymuDk=VRJv3upm>*+&TAMPkI%=HA|VJ1sku3@cJ>?-Wn!O4|is1lJF z2Rk3>!{+FYt7SK*9b5f!FuJ73+%Xxp%NEFXpViaaorsjO<=L!!)>pJ8dM(y3bBgV? z0rDERje3RK?s2s{q^ff8CS9bM%_Ljl7FYI)$s0ZYDvP*9CN)d(=k-fL4Q-NiO{{fa z0t*N7n%B%mq9;9ySC|Cp7m3|B*oSNQ2odQWZJy5Yt`4(z2fGIc>>LwWLS@JFoD2^F zMqNmZ=4;#EezjC|4qYL zM82$u^l?{3wQvJpeOa`BIcQ6mRev_j%zU7t;V7A#yOl+8tnuWdp=EEF!)G5$mA8vJ z^hC5u^oJoisRXqDx%qT`Jpb`vy7+hXNe#i6^)ue?F8(kZd-r^Trgqu1pCoSH*2nCb z$q}m6k$-|0Kl&%ill$SE#}WW2lREelzh17t1VoH{w!Md&C3-1z`-3QQe}P9(Nk5T@ zq-NfgfOQVc3>-oqnnwXtXzjPfRfApOWs+Ne*wL!E-eZ;4%OJ#);!b6bIr*1{06L@$ zv%7Gv1qdnEEZ!idH-Ev931H9w&!8kf`0W%Am#cNtmOSk=hTNuUi)Yv5y$bD(*%06* zCW`II-1#R1*t6LJT~@aT?wq}sr8&h+KBl}m!1zd^ZMTLh+j6|}&|<=MSJpiHMHI2b z3r22s6zKxsxanFT2tr(h4m(nCBA}rm53;BNSYaVZi?loU`B^p!3ku-4hC75dnc98l z7t~D&8O^8TPk8pOybh~rcg@+%nDlf`U*;uwaITP_F}1~;&)X1tUElPb`}Jn3T`{KE zr9;YLImZL_NS!fOBDkD=QE_VqUgEd>@W8TEU{`;_dt| z#tI=^>@RGxXQa8~(4b&XnC|JZIuhRm#TJP$gJMGBbnktOn;E@Qng*FX#@$U2uD_xBrM3whJVKBz z@(QX|H`F9PR`oJZurS!;Zq8vz`L@aFqkG;`J!jOHW>R3~!Iory%Yo(jE5*r`45B0E z@h>qNBdp;W(T9CQ{yI9bN;i(NUe!XI7I2aH%Ln+AR`y8`Wk#3|n5k7<#aIL`79K%} zbPg`nEmWVoPTyftd;RU#;7>>$Jt7!Lp<1^V4wmOw6@H^^pT55mcLbb>06?0Qb5>#OKw*n&2Y^;(Fqxxn) zF{Y^ry_(A4(|fhLO1HK4POw9WF+Cg?PgnUE+v~QgOoCAKk!lE+rOet&ChGu^YK3RP zVcq&~023lik^_)kfv|5Q4`Ixx%hBwB#TDpEz~QMYW~`HMHm+CLao-S4wLtGQ7v zVWB)!fPbwiPN;J$%NJ2tO)bv0DY z5{_!0W{9kH7-K*U8miu@mU-o_Pm!&BQGV`D09uW92RT0UNC`xe+I%rIe1gcoAGqiC zm1Y+;7`hZgpL65=brrB%E28I6)Xlh=xsngGzJo2i^Y^P%AozPsac4&b`3Rzr1mIuZ zgE9cf@m(X+5q}FBW+X98Z&-*pGxrm zrBB(h}hH*sClpZ zdIbi^HuhW-5W|eN8a+C|MwlpnR)4m1Au|6`T%T-8G%~GS-gfZkwyZn2KV_>Jf9Cj2 z^lUVDS;!<@mE4xcNfQtj({|4mAIvrWg^K+K>3~Wh_1X8zEUX^B>y5b?VG7dc37Dp$ z|4)s#T`4vV`<0*Qp@SvoAjN>gLs5EUeu9Ms+aeJ`K0)(7uLYYICsURo!UPRs>1b&< zUX9hCc$pu0qJEeU{yYBsFO^H&=jT2mefve3Q)Ex=1@w}y_=&S5gXm*4FQd-|KH7{O z2Bi~hZ=pOkkMbLFp3J8Funm=gV+Kg9?0Y>-a>o#NLzBRS3Y07q0Ff1rXf`i&$OSi0t+oB3;Y zMBY+x*SAoa6iB|MFNIYqm1K_#zz`;rW`;LvZIYOnP5|1WB~8Q%D)bBVM;}mzMG||7 zhB1J662x~f9cc?g`t4AXibyUKddSe@F*HcL0kH@2qNH4DfaQ7vsG+=;E^-~v5tRXZ zAmAMrln+h5CjwhF266wNl_Ii0h`Oa8DbccQv-=%7ry9f7a@^B!n7s+8tYnu`uE{Mm zyrpuO=@M_N$iH?)T=Hef;fi!Sf7E-4LEPe@|7A~+?#;!-6xsoYyhv(&SsSu$?6+^< zA3XRRQyzIP?O^0XfzfF^ix~60HPvMC64U)r`_;Cl7xPS42dU;J>@4Fu;u-lWhZgN{ z^jlnf&92=4n0iHuKIPK~mk!VEoK(ER2kpb;j%jB6(2uxxCpV6B&HuWy@iHU+&!xi* z_P#F}@?0LJf;eLo>)f|FqXzuTzqeUF8%z#;-W_I7B6Rj&u}k09l~2eU3nGC@`K!q3 z)BE7>)KJfH$?M>^FAWlolDnswaKuzxaBcZlu|1>cBw`KHS49O3nmXJ%NfxhW{ z{mLH<6aPp*tZcx=g03%9scpaT(aTi!7#`D+Z#T{&908|kRXlQ2!tV7j&28x9yii(i z)IFBf-qn~1#ADK!O^jQJs%se@9hNy=rjxp)u#VPoiS3=J83S4CF7NML9oHYitA8kW z=qv^zbez(txj#xMe@pt3_@(6wpl&Vg;t6|^;6ZF`CaoU{-2{Q2Vm!X2Xq}^c4eO!#x94549=FuFT6rQa;v=EnLO)Gog5q>SYrU z^99l``F516u5n6(9TFk;Z$85e?0}0dJ85WIk~;bjgn1-IR9?o{e`5v=P^1t5`xqpA z0A!3xf1xLkV1S${jI2!bk z>Jva!L;(Q?(=;jS(b>0s`}!-0xuqQ@?5%N`TXj`N3R64Wcfq$u^NB1^fK{I7C)PAdwB`!+f~x?O1+fI!dHvW+f!tZ{u@rTC;*t}mo401 z2vLVGLeh^P3&^h_x26joJBi^bIz4_JTzXLMv|ur?rc)2-I{vzyOAGW)S#S;rX;OX- zY2N?&Ekdk@5Ax_cBCoE_)5bp-x%J)Us?tT=%_;5rbMi@S+!P06d02Zlu(t?Kkq~Cu zTdW$GJx{rvSicf=G1vIw1eyvH{7iHP?D@Lk@)&-{R<=RP^Z3hn`n#dzSULgawMvy< zpo_oAf}S`8vV0qmy`0A8_WaN0FFI`Zh<9iZ`_1KpGq}kbwwKH0^Ji~SMB$v?=Enrz z?ZO@z&Rcl0#%8rN@%E>eaA%NqX+2z(*kKZ8ZF98EbFM9RsbxyifyVN-dn+`EWM$O(pzL0tb|_{8!IZwe zCwZIX(tLH61taz7@~AMOm4%H9VT_PtCuRIPi@nEswb%=CZ|cxTcqB!^pfE zCsBfYAF{h8y=xuCc9M?`wCc)>5d+!E9!^(l?J^~}A1q<=$N1qN>}eM?tiD1-0vbEO zW)Jbo#}>twnS3-_*f^s#x$0sd3;V7}Eed;E`(KGQHmD8-5rvj}Hq9s;KW)+2X)d%->Wo{rkREBJ##|#v- zgT;bDt|oAy=@6m43rqZO_cu2XpPmO>>QmAK!;xguDtWUGGM zr$2IxJ{OS!-}>*B*L_+=4&Mh?VW857F*`& zYE_T-NaK+-8Fq=bfl2o8EIVFJQRX#yJkWr*qo+<~3xk!)aP5t=mltbwNf)ob^R8-j zk_OtcyGbkX50pR1Rp3sTm^r46@uc#-ZmFD4LS`%{*kgA)pZUIt6z8M6O=i z`LXgNU$&ak>lKN*QOhednix2BWM7L4HDXC#A6_`2)I^(^j@K8uJx{jb43|z%>=&Be zcId7=19-)C;Lqi(@K0oaaZmGOhEo?_wik!+kUu8xz>dlN$=K7OF@A|qfXKNGO zk9M^fz{ z@i`b^I4w=;s7Gxqa3-8*NNpcO8UW2Wvx1;g4o{F=IO!bk#l z-yJR5*!FN!-p!k}8~&zA_&{ zEq6d2p;q?IJ;KytQJ;XxI)Y1nq(H~ulTQDvvNNd4fo%OL4$k+;#0c@+1ihg#nlb(Q zO?Ipc&SO#2lB6lq^k>@a?5;P<6JTjGHrjLMJ~LxDiqEyUGS{8Pl4N>CQDcy5 zTV5r)%N!2 zR+-bd)0puH1`Rnj7^`g}e7l zzyETFeCC_D`6MgqOWKdrLLC63h1NeCgQ$2r;fgWiWaY5ShOY!ZFo^u?&T1W*zlT!V zujb5Gf!%In&|hqZhTy-sGoCt--I1R5JIU(^_$(p=lX(w`vN6s#Hi3npz}|Y`EgS`{Z7X#`l9EA_{LY}}bdSm=~kL=buy@y4cPY9-r7w&z1cRtCJCeUdrcNm^g zVz>6#`GAy%iHf-f)j3u+!)uFS!!gSH;AXaZjczmvy|3`>mhbg*+|Y6#-{tRnqSQ^y z;ogPmcoPWOu7M)=j5ZXt38lrO8Q!VV>dGqBH@EM5_1b2SFoTnv+cU`?ZP{Er`24cE zWs{Krc8cU>x;BFbnW47(IavI?w`3GtBqQfHWR8-FVDsUbh{U#ulkw|`%2l^UPK#Xm z;*{x=oemOqLS92t;faWV0mD12-S^`wz6=>YxMmv6&Y5C8Jc&>^P=nj1hlEk9T#Mzm=aI$!L+tK#r*k!ny39%-fJ94$}5bq~}Vg{msKTz}Ji==!GK z!LLDWEX~aZcwmi*f7ylGUPZi_M%>(7|5O5R6;_I7?W>wiZ^~r`hF4!<2cc90%CS?? z_K#M{b{1y`o=TwOQS$Gbp5+0}qt4%6>k{ zTB_r2mXUyE)=-vwJ+~g7F-%u#d-?Daoj*IfFiCJNQ|b&-Z$gLrYI+mzdd0;~3hyMD z`us+U;ek$+bxZ1LTL38d~x-7OJKxSaW|=ta&&$rmWG9o;=suUrEo@3!u} zJpTPMgp>)KS&=s6BX5xqWwJYI_glxQ=~kyo7J4t8xS%Y3gz0kYnNAc<#mIhHi34j* zHRn3+Ow|?l&<=ajo5km3GkA#h-N0|)V1C3%WQgF33!HtrM(-kK7iDs8 zP-#1`6y_^ToM$;P+_JW93+#jFwe;`>))+h863N+jyEW;bImP93uSvVb*2R2yF+iF* z-T9p~`zr$|>i}P#F;Rvr8<;xsWZXbDm~|=I zbvP9xrMXG+hESRp1?NeXNLD5)@TOb{>I{rDVuF3})&poKVy5#=fH`v6JMQt8hqEJZ z`C^R;KejgHG1jh^_yIl4MdHsL;Y0$3JUS`e_sFUY^W`+YTRP)@Yg*4l(unEC;!4q( zLF9FB&9aKP^_9jGxGWhBOVn(TJSIy_gCDlxMH2FX3ki|&CV91>#qf`)7ssLZ)NqFa z%`TRnq%hX1Uc6f*&%S?d>!L5^ACb=5A`W0~x+rYLF@W9T@X|lM6B6pnU5w>UMn-ZY z7-%@jazp4)h9rUtTbqVQ&ky(F)F`_N;LRY-4`JKQ%Jk)9@n;l2P=^?0=W7I?&(_lr zL(5uA3n8NU6p6RXukb7#x0rbVR4%5|wmg$|1~P5WY<3lR_yXme9H0LUE(KX4d&P@W zRUl7RY>Pa*27o;EK=Md!6w=fHJa?j58MfY(!vWqniL0=);7%9f?qp;eiZ>kX#%A z&lWg2<2Wm1S|hQN*?4ogjG^CKX1wsXe!!u)2(322Vzye@(YV*NGw}1=SsNlYnqA!l3=G}ll|$V>W=7R5k|Q>?WYp;~)^wy80OxrM&D2Wbp5ib+xhr&1$;j0~tTVfJ-r@yD?;eU8jX-ZJ@!SenW9Vd1!{}4= zT2Uw5s*HYOj0jP!tPWdRj@zUBaCS|dQtlqYY!Se*ETgtP)xzRVqXKd{6BN1dni^l- zuvbD_GH5L+1^|kJ6^(|$XoGJdd^`cmk7qk_n~Q0X9GGbms4yXi0tTHlFaaA>IWRXYt>VSRj|o^fwN%{SJ%=4!&#PJz490>G1|zS?0TY-)L=1e|jk zN}4(==4TmZO20{#CjaR3I0%(tbk}XJsSJV&8L`@!#pvkGA%8G&uPIHo@H0GSY6cW_ zrlybKR6*IZbXIfE+o#t+Y@D`~dYB_O65?h|M%YIZE`$knOCRi4=|4_iDKZeFOv-o1 zY(87hU`xx{QWYNc!+o!M%eJwxQwKAZ(sH4h7W+$($UEXh+N}|?7a>jztt-*W{q79s z1S47uJiL|fYQ;Sq80zZ3WvdE>Q8F9E^0Xd-fOZL!T-&U9uzXcf`)m29c>4&8vFnkB zbClZ6gcL8mu<>(#W~QIwL3@wY z-AbE+nLwwwS_Y1km4dipTbm0)9_l2_7APKU8pi|(!MJnmKv%FjQPoBp8!*hf%hUsv z{#zyHasK->pP^-`e66&bhFv$SHXgl9Z3GVg^l%HxCQrdL%8=u!qG3O%*Em^|%71HD zsFq(}Fux~RATC(1x3^;*x)x!bxg^J=o7PeIc|z|I*{o%VhjmXmaXC-7bUo_y)<>M1 zadpz28fDw;a1Po?^CvGT3R`%Be5md5FWeqZ4+E8a10v z-es`Y+HIFhe30f_elP3n`XaIA%%I^I4i4PAUZ)n#ygk$5L|v(En01EE9jZ8yIG41; z3bssLoJN`b4S^Y}o{vH++?jMG$(EDnrr0t%o5Ez(7uj#1Z9-&amOk3qMEhPvUQlLD zqx-PG^H5lAVd(gHOf8wra-wFJ2?xus&e%$K)0^JrCy1Fam7Bs|7ZpPh+ zrHTOQ!-|`^#dcnd@G0>$ zmB7^Hda>sxhXG3QH=0vR5!);OL|=4+X|a#JOpR-YC`720OSUGCmoY&I`LsJ#NGR%_GsNefoJNcwGiU8;*KmTz-Ug0aYh%6T8|4Xx1N~f zWm?-Ms8wnP)y=LkldcUt4lmtEo-wXJoQW!%h_!)jh$a}jx=yz*AyG@AZOhmki1JVbc#7CYi z`;i$w5viRM;PbLdWo4IczPUcH(CQN&+0Yp~D4@6!P4( zHe+{*{~(mOB*qgiaw^!;3`bSq`~#>*XnI;dk?|o--syq6P)SS z(JwkU9Ahfh3@>Gd)tE4j@FTQzs%rwAzdY0sEsBrCBlmmqq2*x!wn(-7<3^GXcYAYF zyu*_j^y7NtXE=`cO*c58sc65J1JEU`L{L3V{`Ih&2p-Baym`FWRaEAXuF^i*6n8Vk zhh@FUtI)fYoAw^ysz!sZ8&$+#y?#*`prp=dT&;p##6a#3;V;AUH%qBb=ktgf<7BrL z}CNgeAZ2P2gh4V#K6pOC~`RXl}o<@4&c{|T){SDou_`$<)^Kc8hu#R}!?j<=*QNPUrf5{$) z3`X@+6^|3aH`aAEn}?}&td_GhWZKRaGN%;M^0#nR`M4X8O}I{^PqfFMhN-b#;5_Gg zm#Av3YMn|e+i(`9h#8XvNn^Sg-zs~p8P4ORy)YDJ zz?3sKY{Lty+}O& zHdB`~cAdd(varIExml>VVWn5q$P6?$uaU47wPqWSFkQe}7A_@?0f10+&cAgGO zWEPx33k)y3RIA1W)s?A>3qJOFL$0|dpUYh{(`c_lf2KgkV+&DqkV}GudS4MExhuV= z(Q{f0Z>Gp&Tb|DleIY&A-1K;dH~i)iSfV*T;yAuEL?MN->D;qTk%+Fa;)AeTyaRnE zm+g_h7;1K?&q(UCT(6~9u8MZX$J5ZjCsP2PoQcN|^4dRUF z%t||-f@oZ|pkF*U*1pd{is%W$yH!9lj@3e^tspY`>TFl`U#V91fXXy{q-~Gg-^7{7 zkIxOLi(gkez5d?ldYq<|)S`!koM6awXuILst1ydWuCl)xprP+;-edNuJ0GTS6&#zjKw~xO)Q=9 zi$!EZ%mKQkc>#rkd+~bS1I-1sQHRxwMs|m?3LE_zQB>0+e^yuDLC`YRNBVYFOxL;h zKjIr^k{%Y?fBAjRaet5nTCz`|%flSLKyM1_J-!=D`e6+It=VTCf zEhhc-a1Fa$t<=ralA#M%?gut=2veka;WzU%gci1U;fGM&ojbHhH2L9!sf>ow;!A~^ ztu^{Wv1=JDW{m2^+mkS^T>WW*`wO-&E=X6hSyIoPhE5biJ&%^2X3!0zM`0JU?G2ym z775xko!jY;M!s)Yt0VIX`cv%B-3FQ2f@|2|P*lT?`|f$W2hhY`F+1$rGOkrFSz*%c zj_#dPP5yLDJ)Y;t^lEKP7l;_Mospvdru=dgNg-(f7&OMA0^vX3D-hMnFaYIPRCDQMZH|Z!~jNM$i0Wu41 zWosag$bC$aZ?L$ut0m;yUT z*cvuE9W5szD}heGzI3ef%W>cIodLXUhK$K}Y|6Qf zHFxOFiVBmn58glagK!X-k^r2sU{tp_X#0d-ic|0t$#Axc8Hai2xo)Xh z#d%TxUDoCdO{rx#+S?v}pd}G=veAZ{(m!N)u)J*Ni?c&C?HlE}-x*oVL*fQb5% zpX9yKgXdAJs8e>0{lhsg9k#^%8<_3^MV`{fZu?4b9HV>i-K*+=ww9*K%Y2kLKcU z)wBkdvAOfX3Xzs=hWp@PvprGVEPPA>zV14i=a|H`x5xi0fwBJ+F1NcQ)Cik-vSV1- z$ZzeF|Fb$U-yFq>)vy14F1A?xiIUqW@ECrb%|^tDQ{4#+5VQiO2%o3yEj%wO#jLp& zkyTg;C~S5{d=YdbD4(hmS^zoy5S2H)9sc#&CctYtDr9K$MX9oAK*X{HKhR>#HtUmo*vm5u zrwqwBAY+!ECrY&Ol|tmhLn)}2OCW$Dcg8xmq&6c{a5bjLF9oO6m=njTvKEsTI0!o^ zD6X$GJ~-x2!m38tjF!2_Y&{KFI?`>7W_`lTeN%XLK!J1dh8zU6{7(MLq({2SAWGEs zFe$~o^-jHExP^6s-zViqozZW^c_oWLfy&;1OPNlB5>5g!465^t-SiRPAp4awIXK*$ zMCi;d&N!*^_|AF)&$y|hl77wG>d``F2tTx+WentYd84<7B%hj!zp}BoWb;@z?6A0a z{++#3-d^^;C`o+*9Q4KCF1J2FE#JJVJAXRHoLsF!99NsKSZ<#|JF=sGro19A>W$`^ z_Z}SgYSG)SJ6cqbj7t-#1Rg`>n%HKJug;99w+}XBBYz|Mva#n|oYgpIHxpHWC`7@Z zNDo4U=h$AR2@lAOm+ZyR;Bw&9cwS6?Yr1WHN?k4?#)hPhzHYT&=d4bTt~&&ODIuH_7y zTIJx9d*->-H)Uk}=ms|Lrz6VCukEv6d*yy6YvsJh)uZy<_ExS&tZFFP_7~(cQ)Stt z(Y!ke^P!~W-tYJnyM$aaYv}ke45^7^$qyq*qE#&8CAR9iQuWoOxS29Lk#2_pK*NUH z0k0M>pcFMu-12P0uF;&jnSC;uaz@7;aRWu3EXAK=&0oivcO|@e*5{;dA*fy0Ny@2| z9n?sD9$Z~^Vv4gHx~$yy+3tWt3H(z~`S-Bl&T&M28-{|B+zqv)!$gu5q#lJVLp~`_ zi^h2?L2@S?uA!yTJ709l~^zmW#l{#7Ix!>0wx1jz`z zdmg+W2Xp}OdWF>`UpV!WVZyPG-p$Jl`eI{FzDmd!C05G`$@bhF=>Czdnr7Z7v_GV4 z^@HZcb?R@%G%kw}pa6G;kPpZX-U8hTIkIgLjF0yKc+Y+CL&Gm7nsYk`rTw(NOC{-5TLgJqn&UU2_a|NcDn%I5kL`z zodv8#q-IA;fVHS$%nEWV!Bx_c1X^zZ5*w13(}OK;#43@-b*wB0>pW!gDu<|}5h0dX z-S8q?w~eaV*_xbbbQgrOh*&4|{asKZaEFv$?U@I{K)v3d8&C(?M^|B0JC2T@)=kKZ zT=GLLTm9-8{YmQJX~*Eb3cpgI-hnPQk8SS?rK}DRCMaGzI^B1m$0*`AeNpl3B>@lQpIxI)uz|iQ>f1*Oan2#8oR0jRqIMQL(AE!jY)+GO z-Q^87_tQW~>hUec_qr1dT%=bbeOv&ys^G#e@SzT-Q(W>lFX@k`98=RYVeZ1V1b|y2xtL@m*UbClxjK;F4mn z`SEN(rRK+nXM4|n#^DlMdTStJ8hWwg(!0837s11jt^JLyU@P0HSM7Xj&zP6zE1)m+ z8gWtV(J%7Wn-EAU54OpAuZeE@;0tTEgSv+dKB{XYwIVt(zS*^ixM%lap5nJZYoA|? z7(-=!iDJr$94*Ru^_O|s{NK;Zs)S6<-z-((fjs+%e_p*Zgj(^Q! zNedBu*BgD;XZ8w;>vmyJ|8xNS=iJ-BFaL4B|LJ0MLCp6HF1W{MKkfZ|ebGj4s4MW< z?8?3=1k*u}>+dm-T*6xt2pXoT^w|l&JX7{AL~(p&NBV$2Uj`Vb*GhcjhcmJx-G3y$ zjMwY`;P(A4Ipe?h2;&@5TUzZ=Y-{K@-oY%|@033{wtTH_J&W3mz|a=ZwyrnQhG79J z!NJap6@ZKztsdD|{?8}# zC&OX>j_SIjqZ(C_sqCURwyADgb{6wIRm<3N?E$?OCmzFrJLk!XNm=zSO* z8#%TwF7QmPlK^q)ySoLWjd1*c8x%+?iIbXug;oB&-Awl|p`qe3nKc_epiDcLNbkz_ zl0_eq$cUMr+QuJk<)8EzXxBEyzjAr>z%cET8)f13>B4tH1e$pUKF;&;5eSYCQcp%+ z>^%Bu`fK3d<-VJc*#gk_6~|vh2DS3<^bV!Jhn7mPb)6aT9X~5_)%>tr6<9asd&x*K zk$>OsirR(X|5nug%j!6u7y2vDASf7knO(Aja1$GP7&8MNk)SxcQn1^6_01sHNoUJE zvJ^dzyolTH@rQ}@)!~(k!3{Vr{Dr?q&~Nbya^hWFD(%Y?=oI9x%{TBAT9`aS z=LikHvkZ=52X{V~JtABpY5$FAe^eXzvNNIO#J_0>4@XbyFMiP%VgD(O*E}7xXT7j^inYw@{v)=`y*@g!3 z9W}dnGT^#E+5}Mq?~{(gF$}Zhj`oC@LfYHoDH2C&|-|bd> z-;7e&?+r~vE?`V6lHxZa@FW~}=1VWMr)L){PlHoJ(i9qhIGm=fxWOIMudJZB*1*Br zcWdK0HDvm4z%uvWfo0dfCee2-Lf`wb0)MT8ea?<#Zs zE1DgSPv7fM6(x_u!E@mEh2J@s>BI}iHf?hv0u=yt8cD`tD3f#s*orGx0;AU*ngeh6ZlL<1OvTruWCBiN z{DJyw(+_OaiI!eW@18csO;`n<(A~vP+-CC12zBzYA-Pq49C6(=Ki}#b8ROgXV8m-6 z&}fC_7*cc4PP(jVY){2dl%)7^2Pk*Gj(fD+EY_LV`H0HnK!;C1l*-Aa)Dcz)F}&}0 zc&X~Ui)y9GrAYBPlt+pZJMwXBVJ^j^TaXiN)Z%A^F2}LNeIQZ2l4vErcu-ytk+gHG zyWm-Uz9!o^>EmC3L}HR(n4#o_u<~`2{nDFa;SV`qi=3~% zpy~ao?Vi)^73IvTIrt#5?Wm&+k}khbV5A8uJh@7^|Ih?bgxlC#lauh&X)&z3YkzvC zUEKgMs<;6G@jth;$)6pk>cUrFm-K|S|5LWJkTB&<@@Jw7AJhyoI48?wC33ZjKZ|Y! zT<(lERFPv)b=6W-U=Cc851Oa1#r{HB7_nn^g;F_^HJp2{zW;2R6CP3;{A*=i6l<4( zA}?;XwTCqFlnPoOD83=?96mG$4b@avo#aZ|RS)#Gs!m92GRLr$pXJ&xt@3ay;|VJh z+%CJ~$G03?zLTPG=_bjFj`7OMm!;>9*3)j-pbMXzM7ZKCPif91sQupHrA$Qx8? z$3c4J^Q)f(k?RQz;i=iM*x)0UZxOiQ-8W{yne7@tvlGsg`V3Pf@{w<2RfLQ55CA~! zcKNQ1Bf#jhmG#m4y5kpjs$?83D6ZLX6^Qf{7qwnE&7h$+)mLNuU_Y>T*F(iqVDCK* zwc1L6NCeTrb$)*3#tA1w{O1W;!|NuZhNQ2g;%JXdDmBj@AXwHSX6Dt_Jfe9Mnw#0k zv_;jj&hnO%Pc9Y2BcP+2n8W9V+80B8Uskekh3lw&b(OcwM9nvU{;&J`7GwLc*V>%X zQPh5>dz5bwSsky8imY{bSS#-?+N6|vEK`Y6>8X!$8tgiKxw>x=J+9>Nl4YMQ?w5)- z|Mqb$_RWc7&-l}Xq+IQ>FWM6mNgVrQu>nTIL{irpor9#^{Xwl0cyRIcx;E?K-39^u z4-SvLeZ_ipHe&LLeDWC*t=!7`E9|xBrj*H(FT`xQg%f*+bI2>m5xQ8Rmce@uWYtAbib6RBN3_O!o_22Afz9E&JXX=xk8O9 z>98@4#H!Y_pI&D2!G<#D<|3D8Vz(T)S{xrB1m7)nY0r79uR5@8fBQXRTtjz2L+LK^ zZ3w)&&xNw9+HtP3lIvFQfY@hVl1SX$C>M0;w*CFx@ml_uWtlrWGi^DOeg^)m!(5oP pkb~1pe6r)!!4~^Bx93%LyER2!eJ1k%w8;NQvH$-Fi1bdd(3`*!jn*u_r7=EcyEk%!`M56u#&8At+~EAzquCpA^ZezN?Ack0U#g% z00Qs_z<&j90HnmkB*Z64Nk~Y@$Vka4>8U6wC@7ifPSen{vp_i5S=iXla$n{>%Xx{5 zjg3$I!lf%$MMOj(yppmKLNb?yMTCw%gn*2UjFN(qiHeFz=p5TQq5tv&{|TTaC6FM9 zBP8GiPS6q%(h}es0SMU7lLX%%z@HC-6NE%3iAhMw$SJ@EC8vNB1cZbqhzL)fBq9P^ z`+$E3h-gpJox6IC__W4-5>5wtp{G%ANV%>T)-l}eS?3mh;P{M;obe13GYijoUcL(# zMMTBKB_yS8+>}>Pyrrb9sim!>tEX>ZW^VD&(#qP#$=Su#&E3QE`3wJmz@Xre=$PML z#m4>qIzBb+ZF)v#R`$COMa3nhW#tu>^$m?p%`L5MpL_fI2L_Qt!y}VZ(=)Sk^9zeh z=#9;-?VVlB-u_Ws1OVY5VS)EQ0{bN{S}?8?L_~x{Bu8-(oNxm#LRzAe=dKddUDF`B z?{Jz^=qV}v^{6+6b!1$^ch?ynIQEb;a*IszppQcP9@&2`uxI~Ek^LRmkGMtvNLix(4!`|>~ZXVroN|QSz;rENH5^b`;Za%cUAgD3OTR3u`NdUHE z1MGzGrTQ^*7uFL*%ggEUU2=$aCq$_gh+w|mTK9b&xld^bUmhrwV^3me*5#inpsjhz zsB@R#iGsO9+qEGYCk@Fg{-PX`7dfH1*>^{A9FMj~_~%EZ8%JKf>TTosQbubzAE7G{ zwUQiAb8%#jzshR$>=SgW&|9tEJeCA@{a9^7013_BHjVmrSNP=JFLHo+%t-l28(NLy z)L<>D9)8F;kUg6m5fR<|LE-blCk6{iBG2b{^3^6Iu`zz!WuD2 zI;+83NhV^@#7eCAd4@9|Xf~Zh}ZxI?WAPV5ERoP&N-`4IEq)^eksNGP; z!tj8}JRZ1Vvy2MEsDQmIXut#8tMDzc{7ye|aAIUE>tGvKFw?k(UC2Izt4p&xFHQl% z1|RYTQE~mxMn?_F=S1Lpf^B%<0U8hVZ1*9rrs07$U$uiXk+{ed8iUjjKEgJvHXh*l zs)PsL{|T0g0tR@SYcG767!T}U%}Gx=5yJP!1)>BGEO`s#0ke~txGv*72UA)+(2udi zHB+b^Ci}^PADBWTf&s(k^nHy7?uXhR#+UL#rs3G&eapiDf!ZCLAzs4senKUO79w7R zyu@yM4{m(hN8C3ty;LAma4FYBk>_>fU7NXr?DiZRI-}NGCQ${uy6@jH%ns(P@XYwW zikS^1R2;+1*6w*ft28A;jSug_%*x?g`f5aYfY9@h%1KM-Ny76=X|%h6`@75Gs))%s zFL`rJ4o;rR7FW#+j)R>R7V=IBo*^Sl*K2s{O{9w~EGj4YA~-WQ48WNo1J}KwZXVeM zc_m2PgY~`!omb4O;o||Yjoa1I;M6D?TR6fg;WMw@-Eh^>m7miRKcmI5A@MDT?j3j_ zg=I4`5VD~rV!A&8-{LZfoU-3z5RKeNC*T1&(Rm0pY$n;NcKZ{!E*1Qw;*Tcj&+E!Z za4cK@0#({11nZ9lv3Wlhrw{@lrsJ?Nb$GyZ2b>cITOG9)!9CvRxY|FH)Za5yw7s@> z#s#*&6?gP?Z@@H#4Xs2RdV(-YVF9{-g;ROgFNj+Jjxf>eeHssN(L}(OVPFs5=vo?C zK7-HPN4FdXuEMr1E`M)z)Di&m5eRiGVn+~i2r-s*Vu4}fX7E6vI~dI}`t%xSm z4dY_O0YNP;%x2Lx9<7Uwv*X2m0vAgfOQpe&)e(Y;#APU=O_euy?Ah-^_-jKw8{iV6`*OO0T z^3+`HJGfH1owa%%g&AVoj8>!2eroq*vt_`Q#TifnVg?+}h0;8mHZb%M7UZG38Xia; zsnKha*N42;$&=b7V^FlmRHhMKJ7SC zU8f1Yppu%nBB_EIGF(Z6?pt>uLYlyyRy<(H1>3^CVzTfIw4~FT`6RJx?OXe%TSqOV zc1Ha#*2~j}{*tHO`no>GUIo{n#pL+@hBAsj24eO+ z5I`n+OMopsQbEEY1lKjo7W`kpN`#=l2b3&FvIV>d{MRFK*Ky$c-?ziL)&F zzf@BRNNwIcV2^bt!K}9n+h^6alsieUA;%#+f!Jg6KFW>vblZA1YCGem&#y_yB61tm5s0g^hCY(y|u-2&Tx=<8C z6!#oV22V}WjU{d^%y{oIJkU(!yBuHJ=KJPX1c7qp7&#gTw*HGuMDXfg@L3I{aeU*V zEo6II9Oq}-31tISQ4bvT6of=L;3D!T1dgC^-a6iVQ8!8P5hUjjtRPO+`e4gf?Z8*>x+iF}Jd8eIJ>BtJWWOik6I|J?c|c&2>Ixq> z9+Iql zV$oPul=eZmnLK$2pe_Ym7(iVor`1rLb^oGOs6A_7Y&zM^&kK@0pRMWkBn%g7Gnk8D zfVaNMgCk#n3s~S?r>5HviLR?}glK^5oi||zp~{{A!B41K739l|LXayYmG7Xb}lUFsvRX+sLMCH-a5V^-2FL?L_wojWvo$;4I1)nJeU%)rcfKnyE zaF2x#1Im+ga8Mds)JCQH@nUAd+_(-$9@gUg4F+P+?Kq_g@Y@gIhvX$$e{m#`d+3Fe zA63KEfVroe%1Xz#2JF{rCfKiN5KY8>B2YSzocY)Bz@Zo7kSeDrUwBdw8(m8S=G6-D zEhAM@-$e}?aU4%b=3mm*59xC(DR3b+n`Wq(2#BiCy;%<)yfvbiQ>K?mn@__-x&BV~4c zK^mEWGNoU8tex6x*Oba3Gkj$YL;ZX^pKfVsYG*S4a%Qx3!kSE&Ui?gUh{~-c9|e*C zTP+*CxX)DrN5uv2wW&3xZtY74FX>4!#st@#8 z7RIvSxpL@=_*v2Eh@_il-Cp~M4bO|K4L!5F=@n?^oyUbjJM^aIWj^D3f~XV?d4_)P zq5ptE)rr@BMqvdk;3uIESh)XF*cB-LbJ;J)u6o2he_l+cKeMIAF(L)jj)5?T_qfpi zgLIC_=eU&bK8{hq>Q|mLviyZgzZ(z)(@?Oyi3$WOS4cUC)FbE~t0lj;`ktzEf36kt zlPkg@IF$|fP&ItZ2k;!Af8kFQ4X{Pq2BJwC^Q2V;?K84IRf>yS-q(P5AAK;9Rc<7H)mKyOL=3VfO&ZxkD{K_zrk{T=p_F>fFsuyV&)}+F{u^tc_r3rEx^e zgz@E(`Vk3EmZmatN#41ctGDn6N%*PUxK^m@6a{r0lb)A^a)pCyvt8 z2XAoM{)?)>Pu!R3hxC8d_OrO-Bjo?*4C@K|S5gmHvij46I4Yy+kLAV4pEKqO_%XuO zhaFdTKXS}5={D&9Gap+K-uN{kPXC%SzdH&v%#ss`s9Wyv=R${pH!@wF$- z?#H>!74N3#MIz>VW<5^czN>BP^&V@w$CZv&j1a`itr5xR72}>U$dqq8jW7101}hBs z^2wCiMiCBy4R;&rG}Bo*kJfLSRteZkXT&~(7rO@};L)lb^uiEp5O-)tAGh;Z`M0W| z)}5&u#~LYEgW}w9gBrVb{-iv;PG-aR4qxFQzCQbCsaWyWWCRiVK9J+W2!*DE~jMHD|DzQ<|2Q^-X9=upr~Z z1F;~-9PI3Gb)&_~4x`LUIiUk$H{r_>D<)Q7^h2-m1h;W-s5(C2cRVQkwx9T%Pa834 zTj27jYHwe#@w$>EFL#gpBaDi28M4}#I@EcisdAuzdf=cZfT9li7GV-lo<<^W?i0bw zphBVA|86;M`;D4PuYE~EQ_1@>YjatL9>r`~f6v_j8G_7(`2S}%sg%axfqS5V(i3hh;0%EtLG?Rao4i5Yq&bN=II@F1S8RWFX0`}cj@takSo~4+ z#em;UFIup_j=%%*C7^rtQ%&1XQ1uh~e)e#6_2rmN{)Mw)+k^e*g0Y)K^RJQ(d?r5g zbBS@R_>epOi(A`vOeFUIt=oTJen`PU-k`+&H29kbF=d@`aw*&2k}#J@#yC%sq+ zSl?e&sO-2K|+8rpf=Dsa>Ym0?1^s|)A$AUmdt6KXunlZA=WFH z_IfJD*dX>u^JxF!8~T4snjVMRA3F~+ea9T*TzL3#)9(if{ZE$rFYXLXdwX4_9hRdZs*2kB z1o?urFgg>-mUL%8n$&482FDPi4+#{zPq9^nW&R9muQ|wDj;i#=bG?tI30d!7 z+b3`M*+D@n(VS9%4MRRx)cg>LUi>$Guj$;;bo9|bs*Q#I2OvMJ{QuW2*9jRnuw?dv z?@z$C&gbK*@j#0rj#r&#-wm>VS{KTTv8-u;uc3MpYv{FN*JV2zpD2)dF!2%gBjbGo zMnbA&11@u3!?otygADTb#=sROz*AQ^?WjKdWAmxN3hPr;)J#jF%+%Y*Z~ICuCo3Mcx-y|!*7IgQfIwX|u@|CUt)<%jL*WVI4cI|kur zi5~iWFPgOK2eq?TBNNklN|f?_yEu}Z`z%>yVq5poO`hU`GE^?I(E+#@;H( zlF0g~ddf0{%gEeYt5}+2KJ+$*na5p3S=M5YkCD@eePD?#JghyZ{UItd{8Zii48x|( z@(K#chD_@*CVBQ<;dBMbz}}5v@AbJDdWn0H76`y z8Pn(DJFTYWP_YkQ5FiJuz-_|Bbt?bCxCwUZ<{id>Io6x{T#w&5bRQrULH`m_VA$?9 zxpHTOXk274x;X?E9CC_e-U#+oMl#2GjI`HI@KQW$&}2jNa!NLLB8LyE7d&L~0~;d$ zEul!z1dK3drrG!k#Hbg0Iadv^=uyxfG`5T}t#mMq_xx?p{AJX9e`e6th+UG&u5}`N!s;fm5*&_2~7}(NS|o9xN_yXMfB*>ZPcSQZKA6hF)>9Uz}7zhnTE2F z!>!Y$HEGu}ZsiiGPz@zYv~a&9=2yne3|k|0uZjQ8=^-}HJVMs8D96Z`Yfpv>pE9Me z9%(lWb#goPvJv7s=4nYNTJ&efbZEl7l|Rz2FIT=S6W3h3mACTTPWV;}c?#%RM#-`E z>(D%}R!+XOKKe21?14yb2iwkU8PnUlH!jHNZ2v93kC6PwWmFp^YR@<{{##GAsBM;h zPy_R9KM}8+Lvc2CfiNqz8rfhXi;fHt!HBX#Nc0Z_l*pM`wSe%v0BL2{xtfc9+UAu&9Yvct__h&4$kS2>~Z$eOVn%OS14-Zl}1<-`GK~?`Ru$m z>y}|9f|S)z8X8tOqKbeu%>kE=drJlbhdBm5DRNv9W1P^Xv(W18IX{_}de5eoAdxjv zjHZ@h2X8~~C+&xjbN%Tz5Q6TKTJyM*Aqz4VUJT=r$Oc_cv9c^HDVJxrL{`|{qlPA! zuY7j(yF%=VG!}12utfywm@p*og$?p$j2v=~^eAUuoOEjGgN3tXK6BAuyCj+?-gu&~ zQEaeF8)g})dT7*@u%`O-aw`pg~MLm_07j)80N6uWSP~mG~F;5Fh z1e7~+2d1C52Mm+KTxGu{CPv>;wQiq;2~W;rIT3a(`#{lw?yH3=w?|!zh-N>LEsAV= zZ^`u0&aWBCy%*T_i_Y?Hi_YPs(Uf<+Z}*7IzwftSE_T>C$SOY2NvfVBy2(WC=Q2N7 zBduJB`_i+VE7K*sT#9a}eLVVAi*mZuK@Z2WUa&)*p0}hqc38TrUAdX8#{aO9kc^30 zhG2Xvg&H!s8^jejrBL%WFXGL8%XfI7vA&c4;Nes3mj~+f+av`XuTU;uFmd&Y#)HKI zCSwQ7y78)YQKoI1GDEEm3cmOKd_4AbRsVe!QG{W=+HsROZ5}cCan|z=<+Skt8bX7L zOXfICn)}!{-d-h6C(nl1{8IhZ&6hdP$7}LqLW;Z4ym6nP{Un+yyrL4+F=JWD(&7 z{uj;>tvroA5kbA?`;HDBfRpEZNC^Yl^!fHzf%C>UG^@JJtyUheo8O~WON~&qTJVh4 z_ZAxT_F$ihj|@6r=9;V@0PQ+E{aI__T_&4bvu&9@>Gh8Sy`oiKPbsDBVuo9+I9o3c z)r}@!jrd*k4g@g#ii-b+HlDsUn9A>w!Sqq{0+SH`<2@yI1wO*hO}`0ae5DUy6quE2 ztsb&I{h>R1L7niy>2}^QyG3KR*fUT;OypYwTRS&vwlNN+g5|WR)J;}6-_Y5oG6P=~YF$5Rnq&5R?<;&!u!i_h<tb+E=zaX8O8} zl67^)GBXcwbX>#(^!e2?=Sl};#Ga`<$^Jm5y{?&gDq zuO9SDox8m8*>R(atw_8=grMjrZ{xc){!f2rm@Sps4fnn=l?c5A-%$Q)up{R@*?>mwdScrnQUFYk}3UQ&EF`w<}InwZ3~onY+yFX7+x9 zCHi10AT7(?zQIkb%r*Yvl4iTTPj=fG1Vd3vGyUnv-#5r4TIS_AYPoVXiX)!9qd2{M zF*Y=J2=iR&xn`e*(8wLK@XrjLWF*hK)NNzeJH~^F%ElidMaEAp!YqXhmvq`$uEbOd zojV)m6hHq6>yocJ0i%%5%q!U0sgilA$aRyEPg}nXTGv9zrcUw}#WII@BX4*9&DG$` zkLgA@PnDY6S4yz?*eX<;wTMA4_t6A$4vOaZJghfaB!D~~F-GZUs029}E#vn6%<&xc0O?mH(zsP|rKNC+M zuM?e+a2Ich<4A3Yx5}hKU;dhm!&{p;x>0$C&T#f(Wv#0Xb`}rD1+&4OXLXY7hJ5Qy z)UFh_(`mRJdDkT`QUcaDomXrX%Bgs9DuF@pWd^&WV;E~c&Qtv$wP{}-Ow8T&+CV$O ztr8mP@dmQ6Ea-zmNsZ(eTOEL_mhLNC4sc5_fb&D&c9N4~sMet6`3tkx*h`b>!fLt0 z=U)=5oli=2A6!#IM(x@mTP`To3B?htn3xgPN=n*NllH`vE= zjI}#!a&Jt-NEUayzh2|n&Rz3j$@0n1qWOF%*Od;qxg@2d;(dz*wp@joIhrT)@(He2 znX@Z$CD+J5@r0&LtFil(>Gg`X8?!2hOimO%&DC)&Cw*d|z|8<|oq~lZvcswpSC)cd zZ84}gRhKKD2f1=_Rc5IXRVJDD~*agWp%nueCX{f2O(*(nK3f;TU;cfsA6@J z826VLWV4f1Q1a_xS8merV)bskr7zi6=C>@&XAGq$Ob@nQ zhx{IqoMc;&>L+0u1U*wdU@zk1q}rF~BAs+uaXj94Lc|^_OTULbTXHKRXt-r2uh%6!SEub7@-&JR7ge$11V-Lft+`s@@W-H1|Pu-^WY=^3XI< zLtQlK;71mBm|jImXp>_Ym_@sqQCei>!5nS_A}jLMOKA53heJi5gvPdLr01m)SOgSOsSq&7 z-IEc^O#3|FZQ=@7#N7M*v;8TU+9`>+riO+OqfJDmD?GVEp}*Hp_*+ZaNlw6(fh}W5 zbLHidG!q*KHJ)M}Y5y5rjo)WV&zc8n_JuU4g&Ld_+f++Ol~%emypzv1(xWf6YXMd} zpZGW!g}Lce(wm4QP4ze~m4}Ql6)>tWO_#cM($gXM?{T=l67#+w%$|rz1f&O@ zVT{fyZ{1g%M;)g%!iRbKINv|42xV%s=mch*;`7oyG`;PsUyDiLKD|f_Gyq$W zg>4x8LvrLJd^v#Qw2psR;E?`xailrV&4!@DSm?`_v#qV)907?HJ*X(=H6@%nqG+;B z&>I!=;lgyuH-TZ}8<*ZZCH3Hne>OEzuf}4FTm$ zd75ji*?5L!x6YKFHNlG^wsW)9D}m_i$u_!kLW0PO6zXNHQqQB}aIsx26%==k5NAS; z88flyx5n&wl2Q_pX*92bG4)r?Xr!h>)$2DUA8c>zBs`OTYZfg%GgYz1pp#1E1R3L^SCxabX6g;!O)%nTB=zN^{)lS3Y+r zP7F4X5Uq7Sq4Cz0puL2BTYTu@?g zwtjz&gpyNZFJ%0tCVE5XqBs@Ny=h#0H_463NG9ts{xjCr45K9j2`?&&YXU#Vb9!2M zptkd$mD5m!c^jciVe?4{@c3Wtk5d9aodo-S+~ObnPAQ;R=gTxu5We~*#TqHD7#nNo za%o&M0?yJ~-dl{Sk+I$tV=U4KfU;-?t5592y;qy))d>Z*>@N4}@s##Gnv5Rfq7-B! z&yd(m3Xz+18gkHB!a;hrF?R(_Zs?GX-J;dpgPjL~&hbM%@uk?t8LVdW@)k0)2eY1=++gU31q9$i`vYwqo( zaFPkozpe3NlWqH9i{`2g*#Cc2B@jakCSf$TeYyLkp=KUJVPxl@>icbV8F^{97AHxu z3m=H_Q)s>aTD06w9<;M>*L)T9^%!acw+!V=4ehwp{Oye2meGFY4AQUsaL@bgE@wu+ zb32tuN1iCZ#O0NQ7)sO8|7$_wW8@A|$rzLl zuI`~=2YCMf?7{uW&*B_@V^1dsz!eR literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-9-N x N Grid Structure.jpg b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/images/image-9-N x N Grid Structure.jpg new file mode 100644 index 0000000000000000000000000000000000000000..4827aa34ef5d40b2e6f954d2fb156f36bce2feeb GIT binary patch literal 13705 zcmcI~2UJtt*6pDPA|RrG^dKP8RUkBhh)5HuQltxrbWkAD4T$vKKS4q1Eg-#vbPxgQ zz4x9_BPDsq@4oS#`2KtEf8Q8?7@LuO*qpuBZfnlD7H$+b3tV|5rzi*D;Q;_1_ygc3 z02$yS5fL%bg^R?*#3UpaFOgA^laZ2=(OtbpLB&K5VP>LdWMpCIzs17J%f`saDZ<5j z=dO^D5agDaq^N)dzo3x7ub1GFkdTm(lF^cr(+b>Rydm%(e{f#`%8PhZcozxqSOI)W zJOWBQTnhjJ$9Vzo?+@_5KX~{Agcpd2FOpm$1sjxI0r2q%2=ECBE?gia1Y3K7e+LLD zFI>HGSDNUWx(PAsD=Gn>sLvPK9uzfFYxJSm1x+2_l3b#>PD@9Blj9aA7q<{pSVUAz zT;`#yoVZazF*0%PJ&Tst#gG0k3qhsT9^9zeh%PXsE>*(FR{e#0J z%+KRrcHsd8|1b;u{SU+bu!|D33!jjXfROl?U3mB|U?HF+yl~?#(N$@6Vv|?bSOt78 zQayL%i0V!b=H~?=M2h6YTotoeP7(E0_(2WCH2Nt7JhOLgo6M5p#|DkUM(~Ny11Un^8 z!T}*^(b5NrUOkz?OSL$F7SY=lfCIeu5NPK=-@X{>^&-V7C$-Y+x-R;4$>&k)z+-kL zH+&VgJC7dutG)VodXR@jgu+%9WN^S&w+2kL7_EdXba36WK))9qaC&Nqz)*B#x8vsDB0V+^HXP3m8P z{||Pbc7ATzeK&+oJ}Ogp-~by;c)NE0=fTTyd}UiBV-uTbh46+N7s9J8@W*e2G^{5Q zo2;$T;_IBLQPQ$hYIp__euzbC7aU+>D31e_I0~w(AeV4J2lo!PsX781udMfEam zbZ{ayn8&Q=ANw`c`;G)T4q&|wZktUeVyrjMn=gMph#qys{G~2S)k38%3z%Zsu@A!# z*v1OTF<+j>09N7?4k)lW&AZ(O+hw=G0f*$3IH1o?FFpmmM11CCyrh(k?SxXPKmNxc zcfxSMqD~C9ehaprD!!Wc5#4Yayo(r_%EHFVD{g8^u{&$3%L+UQ=KO>Y7BV{r?1Op+ z9PkC~rtg}%Tm=reVGNFVR1L8jl%3CQ@9wHP_tBh6UCr5#O8x$Ym&a0{iQ?wGbkwTZ zrR}n<9Xsx?;BTA0GHiW0uJVznaL!O)rK&f^Dcky$)O4kY(`D|pZO8eoj>5?mB$mV4 zyj1bFC->p7`%qJ>6>LJQUVByd!6M_cg$AVHiB7BfscHZImd5-rpSX=jfd6#&WPsXb zdgbQ$S?5`8m8{v`t5z<|SIWFp?2Ot~2h%9ih4m!@1JDXzKg_;LzTk3HfBnQFT&VY| zSBb@_-JZ>S4@%CYW%QEFGl78f z>&i%f=_%6!PsFte#mfnnB@<&4><-xmD)*xoo_+IcWVgtB{s!7Gj|5zknnjN9RMKQ_ z;lHn~J9#PHBOo|1T(l3tEA+2@2xa0wdsY@d#8OkN@5$cdi+^9nGXN!dX0_qya9Ig1 z;CU_AHF{+p61XWL<3u1?d!<-^{2nb^k7`%VhlQL%`2r7RRKp=6^dREOS;=>7Uohus z1P`-OW9W9T*zu)Rt5j~GslJxAdX{ojLle_2ooe}LQK1P>xsQgyoR_512Zg01h8EE@ z?^CVc&|-V522I>PM@|Q+T`&lpb3bGY^%$&11bjm4G@jj(uZ&C8-JkF({?sXVF`dz6 zQJeh@BbqOzrGd;~-KBs1$HYW0gF&mslKjO9Z`!wS%+{4vGJ~Y2PuBHN$vN|G^;|ae zE`j!ZXqx0`dAk>z&mMniWT=`7T^vS)Vp(MKFc6)}K8`_Kj)Ag^$c)%!8W>EJ+bNua z81iUq%k%c31Cuc(X)CFI#4>)yVz;Qmium>B-asZNI>j%cb($$gLFB=pM9z(vZ+lW3 z7PK832dO^Ix94r{>AiW)0U(==>1EWm@a2b+ zA_A|iX&bRLjFe{Q?|DfMCR;ZZ)<7ymv+J73Q^!xm&70Q0;Crj&x;A@Cp&swH@NDX4cM2_d{G$rN z!y>$Z17ypdpdrE6n(CR0D`_I!;Ew1CeSNkI0t(GOF+V(pdjDvwj=t>|5L^p0BXkKf zcd=pD5gy@AI<2C56}P~ev?~}^x;)x5aB72G7vDMxc#b;NdsB2FVf;yN%^vFuc0eHS zW9`AlplOaLY`Cnm>^3c;N#3zU+2rN>@W~Q++Dk2gjVp{4NWtc6jyiMni`^sbME#{0 zCCwA-2{*OvX(`z@dBA-I31^0YdyS@|&>CjDhXcwAXb`u*O)Nd53|-vb7{QOOa(tzj zF~)c+=nbp>x}K`v!I{)L4p12gFI<`;Aqg$!>>|3PV1BqdAF*h zB!1vAnsGR?)$4FdjF9ex$3>*cKa+Rj|-jJ!>aQSov=dOWzlrF3#n zbHFf>B$QQ)yO}J%iaH2|1LTKFd#}4zE35V1O}2bwq(~fJa+-Bh{xR7C@Ih_orNM`U z4IDri@YRUBwgTgpXOn!RGI>}3-b92@``!nBDNSNV3(+;*WEM=2o!vZ)WMjsrzP!)g zwtypgEXeqdj~|)$?1z0Ud3`~ps)@Z4lJ)XkKj zTG`BXoqo0n-9}zMOVsvNV%na~({%W%mF{Q;r&IZ%AY7Hf`d0Js-BvU5@TxrDLs8~g zg&QGsf(kEzuPcmuKH5@L>n`qvK73gX^&VYCym{=!Qp=)_M2Jg5h5psg#f8M}s}^l+ zY^r#dDl5iOEh-NzUQI<9@yL9|l9i#fgNLsu#C+Sm9zSQg`~AbmPoc@)3rMI(unHrR zb^E*zO4G*yk%j)C|Kx=2KOEH761n#>rCIzNXX>Z5zn$!&Cl3x_$!j>dv<^Ggg~EqA zFgjRi9mFB^AY#ARx<7e^LvYq8bBan`pZO)|X92L_`OG@}yiyCp0nb)p`*`{Xf7JKm zn-f}^f;lr;QiC}OuKzRNtNZ3I4v1ni2K}d9J2u36uoMSGaKbhcaR9fW{GnoYYOv|w zV_flc08v0~%E6Hp4#=P!ku^B)s8vlrquEkZgSn!Afq-Ad%m%SbjhZ?0BgO_M216Tu zUcxwFn|yJ=LuFZ$Cr$%R2(k^%)L`Fq@E5?-?QAG$Sos#K52N&oo+>kxc@8mE6b8HM zPWkJ@hYu-jb=-fT6&+Ql1e&c2>sr`WH0+GL;iq!MspW6qZ8B2Ji&pkvYm6g@w}WuN zsvQpaW{X&mKUXv4qk1ZqvL9}9a=iry?1Mcn3B%B7=V}G zurd0TGqOFzEF)}A{!p$82gpdwV2CL&+6PomlV3c;0ncqx_655^fc_E(_(6Vau0N>$ zY}tZtK1@(IphpS+c zk@;r}|7bk!`_IP|o5tW!e@x(j*ijtN7ID5>t;aQL0NM>b3zVmdX1SWxU<4U@`(O?ls8MJ}L5hhT zQ!(9@vSy_(o{RPXhdUYLLK)-us8Q$c9&?DSTYwMiklh@Kh0@S4r9_lfd>7;MS;1mXy2H*~#d!yM znR(^g&Bv*heVw`E-MlWsdAGLzy1`v&>i+mH4S&(^Kc(CQPMUqzCfJfo zoh)^Vj-nqQ((`=Zgl_52Q}{uvY@wul%wHiT%{=yd3tEjXk=sp6PCnshuDQk(`(2i` z4$u)l*-_Ns>tyAYJ%#0aHQsM_?Ilj4+Ku(Zf)2G+b~IhE|NI{8LPwki(c^$KUwhcT zzP53Qr!su5kYX)m>=3P*ZQM#BOS$BZGMaw16YS5A&OfA9<(nx@K)*wFlOj3VOjTE1(I z=~D{*65R6qw9$7|#IG49gu)h$AqOWG;Eunm?(x}ix0}s9cu~S+-tNicp;9{v8UMZm zKJ^*f1dmX4RlpAlZ_aBg)3dDCE^2ob%4}RT6`*+!G^oX5tM}(X{A9o1ygv@)% zKdLH=eQHqiv|>h^4NsMTsfsPjTC9|3w1S6ig0ql!#M8+G;_sYyZ%S3Uco1IsJ;l~L z&VP7@G4i|0O5@iyF{z=?C?hOsA<#NwnTdAP|rF+xxr`LBs& zuD#0~%pV@BE-T&t%e_&zS_(&6Gcyh5Z<}3}mdeO!XJuY@;2_ko`IBcOxwah_q%N$# z98YNGsbS~^$?5Z%b)Ca<`nL+EHC4*xU&2-Te3GX#%XsUr=~|U2-;4;BP`7;%+s)nV z8OUzx#`kWy4nKGxj!|m(_2hRnQGngN@}(VxS~rd9wFfm1?!Uw%ek;vX$0Od}sAr!v z&smlywmny=uT>Tgh3~Ft(P}I>@ZJ$4bS{VvetknR(?a>tyD{!CgQns|u#t3n8*%;;)_92V{F)BO*@(QGb=9rLGFK2mF zvn7m-_f!m8^3*k^CB(sjGgc3}{*&J+nhu}Lce)yK&DFlQRZq~$rC*hioK-=O2(LJc zEt_p08O$|yb%kSlI_r9_r$9ec>_C_1M59Dmi_(6sERLEnhfw$uvx?Y3XPy)zir0vt zCitm@%sgqfX=qz~f{V{A0WE|@eV`2Ptp1$2U%TwP@gH1928nsm-fF~LUR39nI=0kJ z_Cson70xJ-?3Pn&2KsWj1`uaWM4ZtK<$&;qSq(7)#cJZ)vG*;r&3&9|4=IzaZx~$Z z-zjaah*9&16|C7IEL0&PlG?s&C=wunN9=FEwN!?t2;^s}&FgT>cCGNmc!cA-t1x7<$5Fy7_%Tx0RRc3eQq? z$DIiuIth15{aAHVw~yOS;}pI^bp%S{wsQJ;&o=w`raQ@QRB;DbHmwwOKSlNW+e&by z^CxA!*JIgSS{gSDV&A>bT>DC2Nd1|kgA>Dfw5^T=iIPq;;*bh*%xcX;pOhSb4P&9X z_|8<^+vXdV1~Z{uALTrAwR^eKYq>Xb>JCA-reh)9R|3Zcn2b@*t>cPg`dx=h7`s|| zl!vOmu0)td&$3MtX*l|OaXpOS#|gb}`i8`<+!K!tC7EbFn%Cp0mjJoDA3%K`>~G|CI~$XionAj;GZ?n3 zSg#J)wkvqhR$24cOW~BqXUc|>JqCtV)F&V7sJjtOEHnFpuAnkUa2*E(rC}+FK;)Fv z0>jWYoGRPptc+yh4{+)Eg>I*almgqJPxF&zf3f@36B7q3hT?8f(O3J^YB#56c=rtAev+fI+sROCfKVl=u;LoGPn_ye@0ciQ7-5W2YeGZN>R<= zPqT8|b{5j6IAsN?t2ELEoJ&=VI=E)lATGAi1E*sEOVwGt?V$b?h6*V>BPuIAB?saA zx%O%_hTZFp=gt^x?z-_Yxs5S~9?|s0(`;^rt=Unby621jmhdP08eSrcPzyphUB}Vq zO-+quy^l-(ZZ+xuyVLUTXBb@ZrS9|2?&%0KMBE`zLPTsBZqtByZ@0T$K#+ADRP|N8 z`xhK$yp7tIqzUHC7Ju+dcm0Ej8NwZHI4*nfL_ylGe|uPYF_wLs_g;$>jkWby??#vooBcKbpG z?{D86X_?>A190lAQ~GsX6YOYci#6T+Wl~N_5~ll5KH%9OPDY#JEn-myb#&%y2VQe} z!v{CNETQkD=IGnd8!k=8$5+RFIJ%3-_vae=D3yCzR>=)vk)@v$vQf$RjF2nZ4WC~X zc#eLoZ63^(zn*RQ7a_sTpjisNJd?b&>P(QlpFQZ`>k>tjd@r_>!rY>|5G-{bGL8hC zJw$9rEa@`;*Uv$jyW^OP?rqKs7C=C8ON$roC8#Mt4pT3!+3kUB?&T+`)|0f}eOe`} zQ`JyX?5eT7eUaXUvgu|dR}>DYu98SGQLU_e+ds8Zl_+W=oe(W|DOXZa;u3yKSr29g zE9(Z^*d74yb2elW--9lViW6DCrMf}(h3{{*;Jt&4+9F}U3+t+;X@B$e=L52Kq}>uZ zR0<;ZYA$ForzgjDLS(gy2J@i`WXE1xWs~=3I~}+?s`fiLKHfb0Ao69b%LU(1g(R3L zk~s2mhK`u%;V0fMDvhfEdnB-~ENkOp=K&=5&+`N-vM62)coJS4417Pwuc*!m15KGL zS?{oP!|Jls#0tN}4V@HyYlk#fh{51O#CH$NP6vX?b_jic0QH z7oRF;tX98H$*>WZcx7m1mz#bh{^R9QvWZhHBfpLm$hqOi0ZkvpndD`wf7}Qj+0@x^ zx|8SzIWmhb@{JT*VbmT!6-Fgmr>WDc?LDk_4gH2yi#eG#dcA_mD-o+pr^#NHfV<6G zokA8A;9s>_H7=WA{}FNi{3ngdEyf? z14w9Iw-Kpb!clOJM9t_9>5&}C*V7A@1CLr%40=|Q4KQz4njr-HQyQI`lkJ&$>E>-( zO5FVOk|l}XDu+B}QZY#i{qP2A(mFx4FUcP7qsK*)@365)YBCe6AwIsAq8r$P8vDER z`R$WP<{ukYD>CexEFn#0QT6RA=r>V0R<9l=dX);wI#KN@@^mo0_7<^N0`@y{N}!>s z1u5^lm~K7(;b}hmzO+lpwbbe+mraiT7mtv3_^eVBI%XuLDgRWu9t~}8o-zjoKz%)4 zy;#=o`t=LE9>*zv?F|g984f3)cAAcbvzR}dXKp25eSB}(<*Cz%X~~raZ)BLo!+~1q ztR!-cnJt>up>Gb)msF&`MY{KOeG<=mFV&v7c3fI7f+`HsZ_%4fl+9PUk=SUeWg|N1 z8*eId*Vb$x8$TZdkx^=(rMS9AFMlZYHXdf~^XTIA{>qt^^be0@Rif%C)vr=?!~6x? z4yFCOQ>pDQHRhkT1gBGXXai9`y?QSKPKR(&o_qH9ES&eUVFLrMBc$Ads_BIg#e4mMSG-gql62x@DCWP zoV%*$5LjxGwO*e0%63lQiv5cGgNu#ClTTi}veGT--acSht(DEk=(N^zS~+VC(8yPZ zQ@xe08g$fBp%C3~%--+1@I#7MU)m~T{Aqf%W=_`4B>V&uH}bm+!~{RiY(5QOX-7w0 zi;q`j2orBkf8b-D)@4>?I>Oftv5*vBk+?MVqn31T=dw{!w{Q!xQ<99<$SevQA4*s;jCRnka(a`p3q-_@2HBH*bnqGMdFRWvjJ{cxEUnE*8r( zA~8}sh=4!KEs~4Qxo0=AbTgn}p_h8KIyoTAMqo?{VYea~J3l|UV?M^R>dFR_iPF`9 zQJmg>U9rI*pKko<)(4VwphH~TdsKUIw=yj*NHZ+DEL>-<)_4pniZ)7-j1`CX?V6uZ zCzDrmh;qw`i7h_UF+3;2=dUtT_tgr-kaG`wbL+6maHCgbXUo(Lg6g;7fL)$W9ANx9 ztFhvc6>C9T5!-ZX9-?% zM))0jWy}O3ceapOeFt=`9%qPYR`gYnbd51e940?u8Koh?!3z@oUtRJCTZNT?Kp~hROI>peTZa=XP=0rw_{?e*{l}426Y2X{?U$S`vHbf6(RD>YU2$53SX$F5?vqW;Urn$;1T5ulPD__r{AF8*yu?Nl!U zs;5S($0wjt-ygUp8640O9t`UIfBVi~eT5-^XFQzeTrahsa<3}M>2K0zA+*o)MyhqB z@wAcNx4q8Xb%yu$=J#K@++pMaLl&g5(u`zOoDUOU9V{p5RcVo*-sHR(Wa& zL3xAvz9)#3zJkynQ@v)1%#VO28}`^>uRzFgKyT~}h5-y$T$ev(kzxSzTDTfe!U!yX zwH{~jWj7bG!hid|)DE$w$yraf*$9d$S3#M|#2)Q<{fkT;HGknh!grRxUUeR(mc!45kIW={h^gPfEcBke;_}}RqL$e9*N44RHbs1~baq%Br zJQ#Hn^0lY0MfX$guN^=auB_|OM7nYx_O{W-9W3X#Fc|AeqT#j&42XC5uNOX_a%Y?# zPrecU0sm;xJRUfdIT!@Qj++sXP#KGubl#%lC|wq#CWl+ptXCuBgECcftOebKBl*|Z z$Mr3H#5vgy_VZ}=+TnK8Eqv8nYU08dzePp572ZV;Wd+Tm5Meimp``j@0-7>e%~wt4 z4Hx`HnWws?QS;ax?V-ZbovQ|hjh^M|eC5d{alV`R-VSP6WRtt@q3F$@Wkph5lgm|W zFKxu(!$^2WqjR*;cy2AKz|p7r`F`He(lkL|zB5Je8=isv)kGf4t*vi)3O$Z;s=SI4 zi3jvnlR>SrW3*Bv;1xsBJab461~U6>^oN$KlaYwjQ=%{R1BPi;h5|zma|O)EvPmgE zL3_RGcOIg{1MEIwV$?_C?W8>XOA}AJuRdbC>9nb8Hb7OCZ?!pDaHgtf=b@zRCM4(q zIZSFZr&+jJ?w;I?&KQQ*-D^H}YZ&5}>a}H3t8cS^I!nfR3+YHMCR~_aPzS3A=FL4VS1MYMsGhIiCg9%a~LvlfE*Af{Z}(A3SV$8~#Al zHcxlymPVJx4^K@AFAp!Kq1uo`q@)F_=zPcBz(_{M4`fFN%VnO|_zHt0%ppO~Yx=k7 zcHd}g1h6~a~%=1h*sTzr&va- zjR)U5k%TW3Q$`Ob^k^eA7%=YeT>BO|iD7v!-QnuP1dn_}E8&r%E`;UhEI-G`&-Odl zFgnpVz-mKJClp#I*i7Gwt-sMfI3Ww2Nm=UX`Sf9hO!KPSu_l5dt!AK}b-nzKmCkU5 zpjFqhlB?GHn~e^_kFP~`uO*jBw4N|ARBaq5S7ll5j>#gN7V9?WrzdS+%-mQNC5All zQ5`A^gFP5XAB5h^-m)`mTZ7De0vuJp6tbbyjra!kRY}71Jd{<93@x0c)Gmg;+U1&) zU*&G|L#~Rvp({&itniq1z)FFBlY#Lbo7gaLX2``BVt~mbOX6V@10T9h36AZY)cl^ zpCFvTGg@E|f*NhfVyo1Ow~CojX2DS2?RLET>&lw_&@h(cqUdK z{}Uv|qls}zo0_poaEI4r{&ncm<#uaqOOvhxdsXi4@flBc<(HD0M3MqKmPGy6Go+og zFM{eclLIETf_e0rJz_i}r^fY-$2)D@9H?7seJQ9qm}otG8?zC`&NF3=E@vrLhe^)6 zb4i_s=&U2HqDVATMGc$v?6zZDo-a<7F)K+>u@0sdTnBIOPN7fZ*$5yr}5}H9|bWZ<> zk>=d|Y_F_1T5w3?Xe{q}7zv%URKIoCmo;ad08TI-GC?XSH#sq*Ck@BD5$e>&W`Y2Jq%jUT_##T^O10-dz%&IKPD$Syd|Enz#X!oer9m2*WM5p_0^j2S7rT${A6>(vRv&e<#qC!*U~%t1%AA|3-!@I`B?O;G}&YFpDC=1 z<4-9t$C(E6URK`Lss2(@=gPd@PwMZ}68YZjvKeBBU!HhUOSe`sbCxP5Zc# z{tZF1Ostfx$8a1~>uZex=P(hu3gKgOzp@YPiy5`jcv4wR$`-D_Lk|aSir=G;8XnZw z-Pn{g`b*DDM>faf!s-_3rj~kXUs4oMxVNEJe78%yhy4%%!vM?y*_7SK(R=yBCepPX`^eeOPh2_zi$B!GH159|cn0A|GeR?$JsG5!E440#q}mK)&*L4{i) z_K5}>P}H!d2usM5J+9w7`wx2NJoGt$3C5$0keThXc6boAG95$lNrBRD7p3%2ZphxP zkGHXJwGl^Wh_*t=<<)m!7)&B&Fi8`{mZ}(eA=oj2)ePoR%A9;Dv2RV?H{%I$<1@m1 z9I$hY1HvHZY9J1L=LE(IJ_&<>+B*kJg=iA{4C?s|Dj+-g^+L%{UaH?S&Pp=ChmU%( zpmzZiTYj$%&7AA|09Z`pp%DosU0MrGdRpX}* zke5p`eo}%Fif_H!5kKwgQ}p_nnfj>S3VBITcCZeKo^ip>O)v8gsjQ3w4eVGyIOR9N zRYi<|Nld@jtar|Jer|2I2-e@2-~MglB7XZ`0H$l4xBCC>dii|MHWa_CI=)Dftygqn z{`u>BRs3ZERpUqN=l29B>M$^iO@5%UO|<;ie2_oe+;NQfeG~l$JL-1M&)9}?=Z0ev zT^ulTfCK1Z=V~<|1^RshbN;!u9`}z^-YcD0D1qx8mvRQO7chP6_gY@VxjlStVM{4; zety`_%O0KFkYWfY?Q%0dd2-Jz?kRXfCVdgoQFTaN(~ikM95;^@D4dEGrGEN))<^c? z$Elt~&9e_~N`vaMirK$J>;K_`w?v7V(dmD~Wpl;)37v!xzezFKeK3*8g*p1NQ9wU_+fB|eTR6{Y282Sf@Vwe<-lDZ8}7D4n}jTjECqC)zd+Lb8vT wWKCQj8EQj94y*Z#>9+Tg9Qw{xlKiH7RG_qF@Y1Ul@ZkHEFaJMSAh@yr1sKT}YybcN literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/manifest.json b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/manifest.json new file mode 100644 index 000000000..c0de27e80 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/manifest.json @@ -0,0 +1,80 @@ +{ + "version": "2.0", + "job_id": "job_0e0ec2025607", + "data_id": null, + "source_file_name": "8.pdf", + "processing_date": "2026-06-16T03:39:28.853867Z", + "processing": { + "page_count": 27, + "billing_status": "charged", + "cost": { + "micro_dollars": 40500, + "credits": 0.0405 + }, + "timing": { + "started_at": "2026-06-16T03:37:55.446046+00:00", + "completed_at": "2026-06-16T03:39:28.840141+00:00", + "duration_ms": 93394 + }, + "stages": { + "timing_ms": { + "document.profile": 8189, + "pdf.extract.standard": 65917, + "toc.detect_range_llm": 1402, + "md.detect_toc": 1427, + "heading.merge_pre_pass_llm": 1162, + "heading.hierarchy_llm_call": 3194, + "heading.hierarchy_llm": 3203, + "md.predict_headings": 5078, + "md.deferred_summaries": 12336, + "md.build_dataframe": 1, + "pdf.parse_md": 19209, + "document.cleanup_unreferenced_images": 1, + "document.compress_images": 0, + "worker.parse.document": 93322 + }, + "token_usage": { + "prompt_tokens": 5575, + "completion_tokens": 409, + "total_tokens": 5984 + } + } + }, + "statistics": { + "total_chunks": 46, + "text_chunks": 23, + "image_chunks": 19, + "table_chunks": 4, + "total_pages": null + }, + "HIERARCHY": { + "CHAPTER 8 Transformers": { + "8.1 Attention": {}, + "(8.3) The chicken didn't cross the road because it": {}, + "8.2 Transformer Blocks": {}, + "8.3 Parallelizing computation using a single matrix X": {}, + "8.4 The input: embeddings for token and position": {}, + "8.5 The Language Modeling Head": {}, + "8.6 More on Sampling": { + "8.6.1 Top-k sampling": {}, + "8.6.2 Nucleus or top-p sampling": {} + }, + "8.7 Training": {}, + "8.8 Dealing with Scale": { + "8.8.1 Scaling laws": {}, + "8.8.2 KV Cache": { + "1 x dk": {}, + "1 x N": {}, + "1 x dv": {} + }, + "8.8.3 Parameter Efficient Fine Tuning": {} + }, + "8.9 Interpreting the Transformer": { + "8.9.1 In-Context Learning and Induction Heads": {}, + "8.9.2 Logit Lens": {} + }, + "8.10 Summary": {}, + "Historical Notes": {} + } + } +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-0 Self-Attention.html b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-0 Self-Attention.html new file mode 100644 index 000000000..52a9acaff --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-0 Self-Attention.html @@ -0,0 +1,106 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    LayerTokenToken Description
    Layer k+1TheThe
    Layer k+1chickenchicken
    Layer k+1didn'tdidn't
    Layer k+1crosscross
    Layer k+1thethe
    Layer k+1roadroad
    Layer k+1becausebecause
    Layer k+1itit
    self-attention distributionthethe
    self-attention distributionchickenchicken
    self-attention distributiondidn'tdidn't
    self-attention distributioncrosscross
    self-attention distributionthethe
    self-attention distributionroadroad
    self-attention distributionbecausebecause
    self-attention distributionitit
    self-attention distributionwaswas
    self-attention distributiontootoo
    self-attention distributiontiredtired
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-1 Causal Attention Mask.html b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-1 Causal Attention Mask.html new file mode 100644 index 000000000..cebe8ccc6 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-1 Causal Attention Mask.html @@ -0,0 +1 @@ +
    q1·k1-∞-∞-∞
    q2·k1q2·k2-∞-∞
    q3·k1q3·k2q3·k3-∞
    q4·k1q4·k2q4·k3q4·k4
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-3 .html b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-3 .html new file mode 100644 index 000000000..5171deb4b --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-3 .html @@ -0,0 +1,8 @@ + + + + + + + +
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-4 .html b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-4 .html new file mode 100644 index 000000000..5171deb4b --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/tables/table-4 .html @@ -0,0 +1,8 @@ + + + + + + + +
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/chunks.json b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/chunks.json new file mode 100644 index 000000000..188c1802f --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/chunks.json @@ -0,0 +1,6120 @@ +{ + "chunks": [ + { + "chunk_id": "918d50be-5ab4-5a44-bd38-606c92c9d288", + "type": "text", + "content": "Probability Theory Lecture Notes\nPhanuel Mariano", + "path": "Notes.pdf", + "metadata": { + "length": 48, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Probability", + "Theory", + "Lecture", + "Notes", + "Phanuel", + "Mariano" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "abda7ab8-8c77-5a4b-a28c-0e0a196b44c2", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 1", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "11c6d758-e49c-5063-86d0-2b746869ba05", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 1/Combinatorics", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3e3673f4-4062-516a-a05a-f23828096404", + "type": "text", + "content": "\\- We need a way to help us count faster rather than counting by hand one by one.\nFACT. (Basic Counting Principle) Suppose 2 experiments are to be performed.\nIf one experiment can result in m possibilities\nSecond experiment can result in $n$ possibilities\nThen together there are mn possibilities\n- I like to use the box method. For example. Each box represent the number of possibilities in that experiment.\n- Example1: There are 20 teachers and 100 students in a school. How many ways can we pick a teacher and student of the year?\n\\- Solution: Use the box Method: $20 \\times 100 = 2000$ .\n- The counting principle can be generalized to any amount of experiments: $n_1 \\cdots n_r$ possibilities\n- Example2:\n- A college planning committee consists of 3 freshmen, 4 sophomores, 5 juniors, and 2 seniors.\n- A subcommittee of 4 consists 1 person from each class. How many?\n- Solution: Box method $3 \\times 4 \\times 5 \\times 2 = 120$ .\n\\- Example3: How many differen 6-place license plates are possible if the first 3 places are to be occupied by letters and the finals 3 by numbers?\n- Solution: $26 \\cdot 26 \\cdot 26 \\cdot 10 \\cdot 10 \\cdot 10 = ?$\n- Question: What if no repetition is allowed?\n- Solution: $26 \\cdot 25 \\cdot 24 \\cdot 10 \\cdot 9 \\cdot 8$\n\\- Example4: How many functions defined on $n$ points are possible if each functional value is either 0 or 1.\n\\- Solution: Box method on the $1, \\ldots, n$ points gives us $2^n$ possible functions.", + "path": "Notes.pdf/CHAPTER 1/Combinatorics/1.1. Counting Principle", + "metadata": { + "length": 1450, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "We", + "count", + "faster", + "counting", + "hand", + "FACT", + "Basic", + "Counting", + "Principle", + "Suppose", + "experiments", + "performed", + "If", + "experiment", + "result", + "possibilities", + "Second", + "Then", + "mn", + "box", + "method", + "For", + "Each", + "represent", + "number", + "Example1", + "There", + "20", + "teachers", + "100", + "students", + "school", + "How", + "ways", + "pick", + "teacher", + "student", + "year", + "Solution", + "Use", + "Method", + "times", + "2000", + "The", + "principle", + "generalized", + "amount", + "cdots", + "Example2", + "college", + "planning", + "committee", + "consists", + "freshmen", + "sophomores", + "juniors", + "seniors", + "subcommittee", + "person", + "class", + "Box", + "120", + "Example3", + "differen", + "place", + "license", + "plates", + "places", + "occupied", + "letters", + "finals", + "numbers", + "26", + "cdot", + "10", + "Question", + "What", + "repetition", + "allowed", + "25", + "24", + "Example4", + "functions", + "defined", + "points", + "functional", + "ldots" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "214e915c-d178-5586-9a00-c1c50b266b7c", + "type": "text", + "content": "1.2. Permutations\n\\- How many different ordered arrangements of the letters $a, b, c$ are possible?\n\\- $abc, acb, bac, bca, cab$ Each arrangement is a permutation\n\\- Can also use the box method to figure this out: $3 \\cdot 2 \\cdot 1 = 6$ .\nFACT. With $n$ objects. There are\n$$\nn (n - 1) \\dots 3 \\cdot 2 \\cdot 1 = n!\n$$\ndifferent permutations of the $n$ objects.\n(★) Note that ORDER matters when it comes to Permutations\n\\- Example1: What is the numnber of possible batting order with 9 players?\n\\- Answer: 9!(Box Method or permutations)\n\\- Example2: How many ways can one arrange 4 math books, 3 chemistry books, 2 physics books, and 1 biology book on a bookshelf so that all the math books are together, all the chemistry books are together, and all the physics books are together.\n\\- Answer: We can arrange the math books in 4! ways, the chemistry in 3! ways, the physics in 2! ways, and B in 1! = 1 way.\n\\- But we also have to decide which set of books go on the left, which next, and so on. That is the same as the number of ways of arranging the letters M,C, P,B, and there are 4! ways of doing that. MCPB, PBPB ect..\n\\- So 4! (4!3!2!1!) ways.\n\\- Example3: Repetitions: How many ways can one arrange the letters $a, a, b, c$ ?\n\\- Let us label them $A, a, b, c$ . There are 4!, or 24, ways to arrange these letters. But we have repeats: we could have $Aa$ or $aA$ . So we have a repeat for each possibility (so divide!!!), and so the answer should be $4! / 2! = 12$ .\n\\- If there were 3 $a's$ , 4 $b's$ , and 2 $c's$ , we would have\n$$\n\\frac {9 !}{3 ! 4 ! 2 !}\n$$\n\\- Example4: How many different letter arrangements can be formed from the word PEPPER?\n\\- Answer: There 3 $P$ 's 2 $E$ 's and one $R$ . So $\\frac{6!}{3!2!1!} = 30$ .\nFACT. There are\n$$\n\\frac {n !}{n _ {1} ! \\cdots n _ {r} !}\n$$\ndifferent permutations of $n$ objects of which $n_1$ are alike, $n_2$ are alike, $n_r$ are alike.\n\\- Example4: Suppose there are 4 Czech tennis players, 4 U.S. players, and 3 Russian players, in how many ways could they be arranged?\n\\- Answer: $\\frac{11!}{4!4!3!}$ .", + "path": "Notes.pdf/CHAPTER 1/Combinatorics/1.2. PERMUTATIONS", + "metadata": { + "length": 2063, + "summary": "The text introduces permutations as ordered arrangements of objects, using examples like arranging letters a, b, c (6 ways via 3!). It explains the factorial formula n! for n distinct objects, with order mattering. Examples include batting orders (9!), grouping books (4! * 4! * 3! * 2! * 1!), and handling repetitions (e.g., arranging a, a, b, c as 4!/2! = 12). The general formula for permutations with alike objects is n!/(n1!...nr!), illustrated with PEPPER (6!/(3!2!1!) = 30) and tennis players (11!/(4!4!3!)).", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "1.2", + "Permutations", + "How", + "ordered", + "arrangements", + "letters", + "abc", + "acb", + "bac", + "bca", + "cab", + "Each", + "arrangement", + "permutation", + "Can", + "box", + "method", + "figure", + "cdot", + "FACT", + "With", + "objects", + "There", + "dots", + "permutations", + "Note", + "ORDER", + "matters", + "Example1", + "What", + "numnber", + "batting", + "order", + "players", + "Answer", + "Box", + "Method", + "Example2", + "ways", + "arrange", + "math", + "books", + "chemistry", + "physics", + "biology", + "book", + "bookshelf", + "We", + "But", + "decide", + "set", + "left", + "That", + "number", + "arranging", + "MCPB", + "PBPB", + "ect", + "So", + "Example3", + "Repetitions", + "Let", + "label", + "24", + "repeats", + "Aa", + "aA", + "repeat", + "possibility", + "divide", + "answer", + "12", + "If", + "frac", + "Example4", + "letter", + "formed", + "word", + "PEPPER", + "30", + "cdots", + "alike", + "Suppose", + "Czech", + "tennis", + "Russian", + "arranged", + "11" + ], + "keywords": [ + "Permutations", + "Factorial", + "Arrangements" + ], + "connect_to": [] + } + }, + { + "chunk_id": "f385f7bd-44b5-5c7e-b746-cc7f4aeaba44", + "type": "text", + "content": "\\- We are often interested in selecting $r$ objects from a total of $n$ objects.\n\\- How many ways can we choose 3 letters out of 5? (Does order matter here? NO) If the letters are $a, b, c, d, e$ then there would be 5 for the first position, 4 for the second, and 3 for the third, for a total of $5 \\times 4 \\times 3$ . But order doesn't matter here. So we're over counting here....\n\\- But suppose the letters selected were a, b, c. If order doesn't matter, we will have the letters $a, b, c \\ 3! = 6$ times, because there are $3!$ ways of arranging a group of 3. The same is true for any choice of three letters. So we should have\n$$\n\\frac {5 \\cdot 4 \\cdot 3}{3 !} = \\frac {5 !}{3 ! 2 !} = 1 0.\n$$\nOr what we did was $5 \\cdot 4$ , or $n(n-1) \\cdots (n-r+1)$ then divided by the repeats 3!.\n\\- This is often written $\\left( \\begin{array}{l}5\\\\ 3 \\end{array} \\right)$ , read \"5 choose 3\". More generally..\nFACT. If $r \\leq n$ , then\n$$\n\\binom{n}{r} = \\frac {n !}{(n - r) ! r !}\n$$\nand say $n$ choose $r$ , represents the number of possible combinations of objects taken $r$ at a time.\n(★) Order DOES NOT Matter here\n\\- Recall in Permutations order did matter.\n\\- Example1: How many ways can one choose a committee of 3 out of 10 people?\n\\- Answer: $\\left( \\begin{array}{l}10\\\\ 3 \\end{array} \\right) = \\frac{10!}{3!7!} = \\frac{10\\cdot 9\\cdot 8}{3\\cdot 2} = 10\\cdot 3\\cdot 4 = 120.$\n\\- Example2: Suppose there are 9 men and 8 women. How many ways can we choose a committee that has 2 men and 3 women?\n\\- $\\underline{\\text{Answer:}}$ We can choose 2 men in $\\binom{9}{2}$ ways and 3 women in $\\binom{8}{3}$ ways. The number of committees is then the product $\\binom{9}{2} \\cdot \\binom{8}{3}$ .\n\\- Example3:A person has 8 friends, of whom 5 will be invited to a party. (We've all been through this)\n\\- (a) How many choices are there if 2 of the friends are feuding and will not attend together?\n\\* Box it: [none] + [one of them] [others]\n\\* $\\binom{6}{5}+\\binom{2}{1}\\cdot\\binom{6}{4}$ (recall that when we have OR, use +)\n\\- (b) How many choices if 2 of the friends will only attend together?\n\\* Box it: [none] + [with both]\n\\* $\\binom{6}{5}+1\\cdot1\\cdot\\binom{6}{3}$\n\\- The value of $\\left( \\begin{array}{l}n\\\\ r \\end{array} \\right)$ are called binomials coefficients because of their prominence in the binomial theorem.\nTHEOREM. (The Binomial Theorem)\n$$\n(x + y) ^ {n} = \\sum_ {k = 0} ^ {n} \\binom{n}{k} x ^ {k} y ^ {n - k}.\n$$\nPROOF. To see this, the left hand side is $(x+y)(x+y)\\cdots(x+y)$ . This will be the sum of $2^{n}$ terms, and each term will have n factors. How many terms have k x's and n-k y's? This is the same as asking in a sequence of n positions, how many ways can one choose k of them in which to put x's? (Box it) The answer is $\\binom{n}{k}$ , so the coefficient of $x^{k}y^{n-k}$ should be $\\binom{n}{k}$ . ☐\n• Example: Expand $(x + y)^{3}$ .\n$$\n- \\underline {{\\text { Solution: }}} (x + y) ^ {3} = y ^ {3} + 3 x y ^ {2} + 3 x ^ {2} y + x ^ {3}.\n$$\n\\- Problem: Using Combinatorics: Let's prove\n$$\n\\binom{1 0}{4} = \\binom{9}{3} + \\binom{9}{4}\n$$\nwith no algebra:\n\\- The LHS represents the number of committees having 4 people out of the 10.\n\\- Let's say the President of the university will be in one of these committees and he's special, so we want to know when he'll be there or not.\n\\- When he's there, then there are $1 \\cdot \\begin{pmatrix} 9 \\\\ 3 \\end{pmatrix}$ is the number of ways that contain the President while $\\begin{pmatrix} 9 \\\\ 4 \\end{pmatrix}$ is the number of committees that do not contain the President and contain 4 out of the remaining people.\n• The more general equation is\n$$\n\\binom{n}{r} = \\binom{n - 1}{r - 1} + \\binom{n - 1}{r}\n$$", + "path": "Notes.pdf/CHAPTER 1/Combinatorics/1.3. Combinations", + "metadata": { + "length": 3678, + "summary": "The passage explains the concept of combinations (order does not matter) versus permutations (order matters). It derives the formula for combinations: n choose r equals n!/((n-r)!r!). Examples include choosing a committee of 3 from 10 people, selecting 2 men and 3 women from groups, and inviting friends with constraints (feuding friends or those attending together). The Binomial Theorem is introduced: (x+y)^n = sum_{k=0}^n binom(n,k) x^k y^{n-k}, with a combinatorial proof. A combinatorial identity binom(n,r)=binom(n-1,r-1)+binom(n-1,r) is demonstrated using a committee selection analogy with a special president.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "We", + "interested", + "selecting", + "objects", + "total", + "How", + "ways", + "choose", + "letters", + "Does", + "order", + "matter", + "NO", + "If", + "position", + "times", + "But", + "doesn", + "So", + "counting", + "suppose", + "selected", + "arranging", + "group", + "The", + "true", + "choice", + "frac", + "cdot", + "Or", + "cdots", + "divided", + "repeats", + "This", + "written", + "left", + "begin", + "array", + "end", + "read", + "More", + "generally", + "FACT", + "leq", + "binom", + "represents", + "number", + "combinations", + "time", + "Order", + "DOES", + "NOT", + "Matter", + "Recall", + "Permutations", + "Example1", + "committee", + "10", + "people", + "Answer", + "120", + "Example2", + "Suppose", + "men", + "women", + "underline", + "text", + "committees", + "product", + "Example3", + "person", + "friends", + "invited", + "party", + "ve", + "choices", + "feuding", + "attend", + "Box", + "recall", + "OR", + "cdot1", + "called", + "binomials", + "coefficients", + "prominence", + "binomial", + "theorem", + "THEOREM", + "Binomial", + "Theorem", + "sum", + "PROOF", + "To", + "hand", + "side", + "terms", + "term", + "factors", + "sequence", + "positions", + "put", + "answer", + "coefficient", + "Example", + "Expand", + "Solution", + "Problem", + "Using", + "Combinatorics", + "Let", + "prove", + "algebra", + "LHS", + "President", + "university", + "special", + "ll", + "When", + "pmatrix", + "remaining", + "general", + "equation" + ], + "keywords": [ + "Combinations", + "Binomial Theorem", + "Combinatorics" + ], + "connect_to": [] + } + }, + { + "chunk_id": "361672af-1ad3-5905-900a-149ae3860d02", + "type": "text", + "content": "\\- Example: Suppose one has 9 people and one wants to divide them into one committee of 3, one of 4, and a last of 2. How many different ways are there?\n\\- Solution: (Box it) There are $\\left( \\begin{array}{l}9\\\\ 3 \\end{array} \\right)$ ways of choosing the first committee. Once that is done, there are 6 people left and there are $\\left( \\begin{array}{l}6\\\\ 4 \\end{array} \\right)$ ways of choosing the second committee. Once that is done, the remainder must go in the third committee. So there is 1 one to choose that. So the answer is\n$$\n\\frac {9 !}{3 ! 6 !} \\frac {6 !}{4 ! 2 !} = \\frac {9 !}{3 ! 4 ! 2 !}.\n$$\n\\- In general: Divide $n$ objects into one group of $n_1$ , one group of $n_2$ , ... and a $k$ th group of $n_k$ , where $n = n_1 + \\cdots + n_k$ , the answer is there are\n$$\n\\frac {n !}{n _ {1} ! n _ {2} ! \\cdots n _ {k} !} \\text { ways. }\n$$\n• These are known as multinomial coefficients. We write them as\n$$\n\\binom{n}{n _ {1}, n _ {2}, \\ldots , n _ {k}} = \\frac {n !}{n _ {1} ! n _ {2} ! \\cdots n _ {k} !}.\n$$\n\\- Example: Suppose we are to assign Police officers their duties. Out of 10 officers: 6 patrols, 2 in station, 2 in schools.\n$$\n- \\underline {{\\text { Answer }}}: \\frac {1 0 !}{6 ! 2 ! 2 !}.\n$$\n\\- Example: There are 10 flags:5 indistinguishable Blue flags, 3 indistinguishable Red flags, and 2 indistinguishable Yellow flags. How may different ways can we order them on a flag pole?\n$$\n- \\underline {{\\text {Answer:}}} \\frac {1 0 !}{5 ! 3 ! 2 !}.\n$$\n\\- Example: Suppose one has 8 indistinguishable balls. How many ways can one put them in 3 boxes?\n\\- Solution1: Let us make sequences of o's and |'s; any such sequence that has | at each side, 2 other |'s, and 8 o's represents a way of arranging balls into boxes. For example, if one has\n$$\n\\left| \\begin{array}{c c c} o o & 0 0 0 & 0 0 0 \\end{array} \\right|.\n$$\n\\- How many different ways can we arrange this where we have start with $|$ and end with $|$ . In between, we are only arranging $8 + 2 = 10$ symbols, of which only 8 are $o$ 's\n\\- So the question is: How many ways out of 10 spaces can one pick 8 of them into which to put an $o$ ?\n$$\n- \\binom{1 0}{8}.\n$$\n\\- Solution2: Look at spaces between. There are 9 spaces. So $\\left( \\begin{array}{l}9\\\\ 2 \\end{array} \\right) + 9$ .", + "path": "Notes.pdf/CHAPTER 1/Combinatorics/1.4. Multinomial Coefficients", + "metadata": { + "length": 2263, + "summary": "The passage explains multinomial coefficients, which count ways to divide n objects into k groups of specified sizes. It provides examples: dividing 9 people into committees of sizes 3,4,2 yields 9!/(3!4!2!); assigning 10 police officers to duties (6 patrol, 2 station, 2 schools) gives 10!/(6!2!2!); ordering 10 flags (5 blue, 3 red, 2 yellow) gives 10!/(5!3!2!). It also covers distributing 8 indistinguishable balls into 3 boxes using stars and bars, resulting in C(10,8) or C(9,2)+9 ways.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Example", + "Suppose", + "people", + "divide", + "committee", + "How", + "ways", + "Solution", + "Box", + "There", + "left", + "begin", + "array", + "end", + "choosing", + "Once", + "remainder", + "So", + "choose", + "answer", + "frac", + "In", + "general", + "Divide", + "objects", + "group", + "cdots", + "text", + "These", + "multinomial", + "coefficients", + "We", + "write", + "binom", + "ldots", + "assign", + "Police", + "officers", + "duties", + "Out", + "10", + "patrols", + "station", + "schools", + "underline", + "Answer", + "flags", + "indistinguishable", + "Blue", + "Red", + "Yellow", + "order", + "flag", + "pole", + "balls", + "put", + "boxes", + "Solution1", + "Let", + "make", + "sequences", + "sequence", + "side", + "represents", + "arranging", + "For", + "arrange", + "start", + "symbols", + "question", + "spaces", + "pick", + "Solution2", + "Look" + ], + "keywords": [ + "multinomial coefficients", + "combinations", + "permutations" + ], + "connect_to": [] + } + }, + { + "chunk_id": "eb7a7665-d841-5367-89fe-bb21aac028e9", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 2", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "4ddba2cd-0b97-56c9-a7bf-fef3654617b1", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a9c9adb1-1810-5983-bc0a-8e4e01a08383", + "type": "text", + "content": "\\- We will have a sample space, denoted $S$ (sometimes $\\Omega$ , or $\\mathcal{U}$ ) that consists of all possible outcomes from an experiment.", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.1. Sample Space and Events", + "metadata": { + "length": 143, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "We", + "sample", + "space", + "denoted", + "Omega", + "mathcal", + "consists", + "outcomes", + "experiment" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "59862297-994e-5146-afee-5bf3117b3577", + "type": "text", + "content": "\\* Experiment: Roll two dice,\n\\* Sample Space: $S =$ would be all possible pairs made up of the numbers one through six. List it here. $\\{(i,j):i,j = 1,\\dots 6\\}$ . 36 points.", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.1. Sample Space and Events/- Example1:", + "metadata": { + "length": 175, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Experiment", + "Roll", + "dice", + "Sample", + "Space", + "pairs", + "made", + "numbers", + "List", + "dots", + "36", + "points" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c0170e45-65d5-584e-a2f1-15d56710a254", + "type": "text", + "content": "\\* Experiment: Toss a coin twice\n\\* $S = \\{HH,HT,TH,TT\\}$", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.1. Sample Space and Events/- Example 2:", + "metadata": { + "length": 57, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Experiment", + "Toss", + "coin", + "HH", + "HT", + "TH", + "TT" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a6f8aee3-f1db-5fe2-995e-3568ddb78184", + "type": "text", + "content": "\\* Experiment: Measuring the number of accidents of a random person before they had turn 18.\n$$\n\\cdot S = \\{0, 1, 2, \\dots \\}\n$$", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.1. Sample Space and Events/- Example3:", + "metadata": { + "length": 128, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Experiment", + "Measuring", + "number", + "accidents", + "random", + "person", + "turn", + "18", + "cdot", + "dots" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c2dbc4ac-13be-5675-a67c-1efaeb039215", + "type": "image", + "content": "\n\\- Events: An event $A$ is a subset of $S$ . In this case we use the notation $A \\subset S$ , to mean $A$ is a subset of $S$ .0\n[images/image-1-_- Events_.jpg]\n", + "path": "images/image-1-_- Events_.jpg", + "metadata": { + "length": 161, + "summary": "image-1\n\\- Events: An event $A$ is a subset of $S$ . In this case we use the notation $A \\subset S$ , to mean $A$ is a subset of $S$ .0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-1-_- Events_.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "1796fd14-ca8d-5488-aa69-2713383c86b0", + "type": "image", + "content": "\n- $A \\cup B$ : points in $S$ such that is in $A$ OR $B$ OR BOTH.0\n[images/image-2-- $A _cup .jpg]\n", + "path": "images/image-2-- $A _cup .jpg", + "metadata": { + "length": 99, + "summary": "image-2\n- $A \\cup B$ : points in $S$ such that is in $A$ OR $B$ OR BOTH.0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-2-- $A _cup .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "338e7883-9cbe-56a3-8b36-6d3c84a22bff", + "type": "image", + "content": "\n- $A \\cap B$ , points in $A$ AND $B$ . (you may also see $AB$ )0\n[images/image-3-- $A _cap .jpg]\n", + "path": "images/image-3-- $A _cap .jpg", + "metadata": { + "length": 98, + "summary": "image-3\n- $A \\cap B$ , points in $A$ AND $B$ . (you may also see $AB$ )0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-3-- $A _cap .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "554d29a6-9e8f-5cef-811e-264704ce176a", + "type": "image", + "content": "\n- $A^c$ is the compliment of $A$ , the points NOT in $A$ . (you may also see $A'$ )0\n[images/image-4-- $A^c$ is.jpg]\n", + "path": "images/image-4-- $A^c$ is.jpg", + "metadata": { + "length": 118, + "summary": "image-4\n- $A^c$ is the compliment of $A$ , the points NOT in $A$ . (you may also see $A'$ )0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-4-- $A^c$ is.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f9016cc9-12ff-5808-ac41-a7af4683b73f", + "type": "image", + "content": "\n- Can extend to $A_1, \\ldots, A_n$ events. $\\bigcup_{i=1}^{n} A_i$ and $\\bigcap_{i=1}^{n} A_i$ . • Example1: Roll two dice.0\n[images/image-5-- Can exte.jpg]\n", + "path": "images/image-5-- Can exte.jpg", + "metadata": { + "length": 158, + "summary": "image-5\n- Can extend to $A_1, \\ldots, A_n$ events. $\\bigcup_{i=1}^{n} A_i$ and $\\bigcap_{i=1}^{n} A_i$ . • Example1: Roll two dice.0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-5-- Can exte.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "9c41be8f-575c-5ef3-9d0d-bc465abf8215", + "type": "image", + "content": "\n• Example1: Roll two dice.0\n[images/image-6-• Example1.jpg]\n", + "path": "images/image-6-• Example1.jpg", + "metadata": { + "length": 61, + "summary": "image-6\n• Example1: Roll two dice.0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-6-• Example1.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ac14ea33-4b24-51b0-8fb9-9c6151d1a078", + "type": "image", + "content": "\n• Example1: Roll two dice.0\n[images/image-7-• Example1.jpg]\n", + "path": "images/image-7-• Example1.jpg", + "metadata": { + "length": 61, + "summary": "image-7\n• Example1: Roll two dice.0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-7-• Example1.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "edaa4d87-b809-5c00-9243-7984182f85d8", + "type": "image", + "content": "\n• Example1: Roll two dice.0\n[images/image-8-• Example1.jpg]\n", + "path": "images/image-8-• Example1.jpg", + "metadata": { + "length": 61, + "summary": "image-8\n• Example1: Roll two dice.0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-8-• Example1.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "467f4d98-881d-5e69-a8d9-8446aa815ca0", + "type": "image", + "content": "\n• Example1: Roll two dice.0\n[images/image-9-• Example1.jpg]\n", + "path": "images/image-9-• Example1.jpg", + "metadata": { + "length": 61, + "summary": "image-9\n• Example1: Roll two dice.0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-9-• Example1.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "8926ab66-74ee-55ef-9fe3-983c6634a1a6", + "type": "text", + "content": "\\* Let $S$ be the possible orders in which 5 horses finish in a horse race;\n\\* Let $S$ be the possible price of some stock at closing time today; or $S = [0, \\infty)$ ;\n\\* The age at which someone dies, $S = [0, \\infty)$ .\n\\- Events: An event $A$ is a subset of $S$ . In this case we use the notation $A \\subset S$ , to mean $A$ is a subset of $S$ .\n- $A \\cup B$ : points in $S$ such that is in $A$ OR $B$ OR BOTH.\n- $A \\cap B$ , points in $A$ AND $B$ . (you may also see $AB$ )\n- $A^c$ is the compliment of $A$ , the points NOT in $A$ . (you may also see $A'$ )\n- Can extend to $A_1, \\ldots, A_n$ events. $\\bigcup_{i=1}^{n} A_i$ and $\\bigcap_{i=1}^{n} A_i$ .\n\n\\- Events: An event $A$ is a subset of $S$ . In this case we use the notation $A \\subset S$ , to mean $A$ is a subset of $S$ .0\n[images/image-1-_- Events_.jpg]\n\n\n- $A \\cup B$ : points in $S$ such that is in $A$ OR $B$ OR BOTH.0\n[images/image-2-- $A _cup .jpg]\n\n\n- $A \\cap B$ , points in $A$ AND $B$ . (you may also see $AB$ )0\n[images/image-3-- $A _cap .jpg]\n\n\n- $A^c$ is the compliment of $A$ , the points NOT in $A$ . (you may also see $A'$ )0\n[images/image-4-- $A^c$ is.jpg]\n\n\n- Can extend to $A_1, \\ldots, A_n$ events. $\\bigcup_{i=1}^{n} A_i$ and $\\bigcap_{i=1}^{n} A_i$ . • Example1: Roll two dice.0\n[images/image-5-- Can exte.jpg]\n\n\n• Example1: Roll two dice.0\n[images/image-6-• Example1.jpg]\n\n\n• Example1: Roll two dice.0\n[images/image-7-• Example1.jpg]\n\n\n• Example1: Roll two dice.0\n[images/image-8-• Example1.jpg]\n\n\n• Example1: Roll two dice.0\n[images/image-9-• Example1.jpg]\n\n• Example1: Roll two dice.\n- Example of an Events\n- $E =$ the two dies come up even and equal $\\{(2,2),(4,4),(6,6)\\}$\n- $F =$ the sum of the two dice is 8. $\\{(2,6),(3,5),(4,4),(5,3),(6,2)\\}$ .\n- $E \\cup F = \\{(2, 2), (2, 6), (3, 5), (4, 4), (5, 3), (6, 2), (6, 6)\\}$\n- $E \\cap F = \\{(4, 4)\\}$ .\n- $F^c$ all the 31 other ways that does not include $\\{(2,6), (3,5), (4,4), (5,3), (6,2)\\}$ .\n\\- Example2: $S = [0, \\infty)$ age someone dies.\n\\- Event $A =$ person dies before they reached 30.\n$$\n* A = [ 0, 3 0).\n$$\n\\- Interpret $A^c = [30,\\infty)$\n\\* The person dies after they turned 30.\n\\- $B = (15, 45)$ . Do $A \\cup B, A \\cap B$ and so on.\n- Properties: Events also have commutative and associate and Distributive laws.\n• What is $A \\cup A^{c}$ ? = S.\n- DeMorgan's Law:\n- $(A\\cup B)^{c} = A^{c}\\cap B^{c}$ . Try to draw a picture\n$-\\left(A\\cap B\\right)^{c}=A^{c}\\cup B^{c}.$\n- This works for general $A_{1},\\ldots ,A_{n}$ : $(\\cup_{i = 1}^{n}A_{i})^{c} = \\cap_{i = 1}^{n}A_{i}^{c}$ and $(\\cap_{i = 1}^{n}A_{i})^{c} = \\cup_{i = 1}^{n}A_{i}^{c}$ .\n\\- The empty set $\\emptyset = \\{\\}$ is the set that has nothing in it.\n\\- $A$ and $B$ are disjoint if $A \\cap B = \\emptyset$ .\n- In Probability we may say that events $A$ and $B$ are \"mututally exclusive\" if they are disjoint.\n- mutually exclusive means the same thing as disjoint", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.1. Sample Space and Events/- Others:", + "metadata": { + "length": 2875, + "summary": "The passage defines events as subsets of a sample space S, and explains basic set operations: union (A ∪ B), intersection (A ∩ B), and complement (A^c). It provides examples from rolling two dice and age at death. It also covers properties like commutativity, associativity, distributivity, DeMorgan's Laws, and the concept of disjoint (mutually exclusive) events.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Let", + "orders", + "horses", + "finish", + "horse", + "race", + "price", + "stock", + "closing", + "time", + "today", + "infty", + "The", + "age", + "dies", + "Events", + "An", + "event", + "subset", + "In", + "case", + "notation", + "cup", + "points", + "OR", + "BOTH", + "cap", + "AND", + "AB", + "compliment", + "NOT", + "Can", + "extend", + "ldots", + "events", + "bigcup", + "bigcap", + "BOTH.0", + "Example1", + "Roll", + "dice.0", + "dice", + "Example", + "equal", + "sum", + "31", + "ways", + "include", + "Example2", + "Event", + "person", + "reached", + "30", + "Interpret", + "turned", + "15", + "45", + "Do", + "Properties", + "commutative", + "associate", + "Distributive", + "laws", + "What", + "DeMorgan", + "Law", + "Try", + "draw", + "picture", + "left", + "This", + "works", + "general", + "empty", + "set", + "emptyset", + "disjoint", + "Probability", + "mututally", + "exclusive", + "mutually", + "means", + "thing" + ], + "keywords": [ + "Events", + "Set Operations", + "DeMorgan's Law" + ], + "connect_to": [ + { + "target": "c2dbc4ac-13be-5675-a67c-1efaeb039215", + "relation": "embeds", + "ref": "[images/image-1-_- Events_.jpg]", + "position": { + "start": 789, + "end": 820 + } + }, + { + "target": "1796fd14-ca8d-5488-aa69-2713383c86b0", + "relation": "embeds", + "ref": "[images/image-2-- $A _cup .jpg]", + "position": { + "start": 889, + "end": 920 + } + }, + { + "target": "338e7883-9cbe-56a3-8b36-6d3c84a22bff", + "relation": "embeds", + "ref": "[images/image-3-- $A _cap .jpg]", + "position": { + "start": 988, + "end": 1019 + } + }, + { + "target": "554d29a6-9e8f-5cef-811e-264704ce176a", + "relation": "embeds", + "ref": "[images/image-4-- $A^c$ is.jpg]", + "position": { + "start": 1107, + "end": 1138 + } + }, + { + "target": "f9016cc9-12ff-5808-ac41-a7af4683b73f", + "relation": "embeds", + "ref": "[images/image-5-- Can exte.jpg]", + "position": { + "start": 1266, + "end": 1297 + } + }, + { + "target": "9c41be8f-575c-5ef3-9d0d-bc465abf8215", + "relation": "embeds", + "ref": "[images/image-6-• Example1.jpg]", + "position": { + "start": 1328, + "end": 1359 + } + }, + { + "target": "ac14ea33-4b24-51b0-8fb9-9c6151d1a078", + "relation": "embeds", + "ref": "[images/image-7-• Example1.jpg]", + "position": { + "start": 1390, + "end": 1421 + } + }, + { + "target": "edaa4d87-b809-5c00-9243-7984182f85d8", + "relation": "embeds", + "ref": "[images/image-8-• Example1.jpg]", + "position": { + "start": 1452, + "end": 1483 + } + }, + { + "target": "467f4d98-881d-5e69-a8d9-8446aa815ca0", + "relation": "embeds", + "ref": "[images/image-9-• Example1.jpg]", + "position": { + "start": 1514, + "end": 1545 + } + } + ] + } + }, + { + "chunk_id": "ffe1da75-c671-5234-9084-4f1942ea493f", + "type": "text", + "content": "\\- Let $E$ be an event. How do we defined the probability of an event?\n- We can attempt to define a probability by the relative frequency,\n- Perform an experiment (e.g. Flipping a coin)\n- Perform that experiment $n$ times and let $n(E) =$ the number of times the event occurred in $n$ repetitions\n\\* (e.g. Flip a coin $n = 1000$ times, and let's say that $n(\\{Tails\\}) = 551$ ) Then it's reasonable to think $\\mathbb{P}(\\{Tails\\}) \\approx \\frac{551}{1000}$\n- So maybe we can define the probability of an event as $\\mathbb{P}(E) = \\lim_{n\\to \\infty}\\frac{n(E)}{n}$ . But we don't know if this limit exists, or if $n(E)$ is even well defined!!!\n- So we need a new approach.\n\\- Probability will be a rule given by the following Axioms (Laws that we all agree on)\n- A probability will be a function $\\mathbb{P}(E)$ where the input is a set/event such that\n- Axiom 1: $0 \\leq \\mathbb{P}(E) \\leq 1$ for all events $E$ .\n- Axiom 2: $\\mathbb{P}(S) = 1$ .\n- $\\underline{\\mathbf{Axiom 3}}$ : (disjoint property) If the events $E_1, E_2, \\ldots$ are pairwise disjoint/mutually exclusive then\n$$\n\\mathbb {P} \\left(\\bigcup_ {i = 1} ^ {\\infty} E _ {i}\\right) = \\sum_ {i = 1} ^ {\\infty} \\mathbb {P} \\left(E _ {i}\\right).\n$$\n\\* Mutually exclusive means that $E_{i} \\cap E_{j} = \\emptyset$ when $i \\neq j$ .\n\\- Remark: Note that you take a probability of a subset of $S$ , not of points of $S$ . However it is common to write $P(x)$ for $P(\\{x\\})$ .\n\\- Say if the experiment is tossing a xoin. Then $S = \\{H, T\\}$ . The probability of heads should be written as $\\mathbb{P}(\\{H\\})$ , but it is common to see $\\mathbb{P}(H)$ .\n\\- Example1:\n$-$ (a) Suppose we toss a coin and they are equally likely then $S = \\{H,T\\}$ and\n\\* $\\mathbb{P}\\left(\\{H\\}\\right) = \\mathbb{P}\\left(\\{T\\}\\right) = \\frac{1}{2}$ . We may write $\\mathbb{P}(H) = \\mathbb{P}(T) = \\frac{1}{2}$ .\n\\- (b) If biased coin is tosse then one could have a different assignment of probability $\\mathbb{P}(H) = \\frac{2}{3}, \\mathbb{P}(T) = \\frac{1}{3}$ .\n\\- Example2:\n- Rolling a fair die, the probability space consists of $S = 1,2,3,4,5,6$ , each point having probability $\\frac{1}{6}$ .\n- We can compute the probability of rolling an even number by\n$$\n\\mathbb {P} (\\{\\text { even } \\}) = \\mathbb {P} (\\{2, 4, 6 \\})\n$$\n$$\n= \\mathbb {P} (2) + \\mathbb {P} (4) + \\mathbb {P} (6) = \\frac {1}{2}\n$$\nwhere we used the rules of probability by breaking it down into a sum.\nPROPOSITION 1. (a) $\\mathbb{P}(\\emptyset) = 0$\n(b) If $A_{1},\\ldots ,A_{n}$ are pairwise disjoint, $\\mathbb{P}\\left(\\cup_{i = 1}^{n}A_{i}\\right) = \\sum_{i = 1}^{n}\\mathbb{P}\\left(A_{i}\\right)$ .\n(c) $\\mathbb{P}(E^c) = 1 - \\mathbb{P}(E)$ .\n(d) If $E \\subset F$ , then $\\mathbb{P}(E) \\leq \\mathbb{P}(F)$ .\n(e) $\\mathbb{P}(E\\cup F) = \\mathbb{P}(E) + \\mathbb{P}(F) - \\mathbb{P}(E\\cap F)$ .\n\\- It helps to draw diagrams to prove these.\n\\- Try to prove at least some of these yourself.\nPROOF. (a) Let $A_{i} = \\emptyset$ for each $i$ which are disjoint. So\n$$\n\\mathbb {P} (\\emptyset) = \\mathbb {P} \\left(\\bigcup_ {i = 1} ^ {\\infty} A _ {i}\\right) = \\sum_ {i = 1} ^ {\\infty} \\mathbb {P} (A _ {i}) = \\sum_ {i = 1} ^ {\\infty} \\mathbb {P} (\\emptyset),\n$$\nsince this would be infinite sum so that $\\mathbb{P}(\\emptyset) = 0$ since $0 \\leq \\mathbb{P}(\\emptyset) \\leq 1$ .\n(b) Let $A_{n+1} = A_{n+2} = \\cdots = \\emptyset$ so that $\\cup_{i=1}^{\\infty} A_{i} = \\cup_{i=1}^{n} A_{i}$ hence\n$$\n\\begin{array}{l} \\mathbb {P} \\left(\\cup_ {i = 1} ^ {n} A _ {i}\\right) = \\mathbb {P} \\left(\\cup_ {i = 1} ^ {\\infty} A _ {i}\\right) \\\\ = \\sum_ {i = 1} ^ {n} \\mathbb {P} \\left(A _ {i}\\right) + \\sum_ {n = 1} ^ {\\infty} \\mathbb {P} (\\emptyset) \\\\ = = \\sum_ {i = 1} ^ {n} \\mathbb {P} \\left(A _ {i}\\right) + \\sum_ {n = 1} ^ {\\infty} 0 \\\\ = \\sum_ {i = 1} ^ {n} \\mathbb {P} \\left(A _ {i}\\right) \\\\ \\end{array}\n$$\n(c) Use $S = E \\cup E^{c}$ . By Axiom (2) we have\n$$\n1 = \\mathbb {P} (S) = \\mathbb {P} (E) + \\mathbb {P} (E ^ {c}),\n$$\nhence $\\mathbb{P}(E^c) = 1 - \\mathbb{P}(E)$ .\n(d) If $E \\subset F$ , then write $F = E \\cup (F \\cap E^c)$ thus since this is disjoint\n$$\n\\mathbb {P} (F) = \\mathbb {P} (E \\cup (F \\cap E ^ {c})) = \\mathbb {P} (E) + \\mathbb {P} (F \\cap E ^ {c}) \\geq \\mathbb {P} (E) + 0 = \\mathbb {P} (E).\n$$\n(e) Write $E \\cup F = E \\cup (E^c \\cap F)$ , (Picture of venn diagram of both ) hence by disjointness again\n$$\n\\mathbb {P} (E \\cup F) = \\mathbb {P} (E) + \\mathbb {P} (E ^ {c} \\cap F).\n$$\nNow write $F$ (with picture) as $F = (E \\cap F) \\cup (E^c \\cap F)$ and using disjointness\n$$\n\\mathbb {P} (F) = \\mathbb {P} (E \\cap F) + \\mathbb {P} (E ^ {c} \\cap F) \\implies \\mathbb {P} (E ^ {c} \\cap F) = \\mathbb {P} (F) - \\mathbb {P} (E \\cap F),\n$$\nsubstitute into first equation to get\n$$\n\\begin{array}{l} \\mathbb {P} (E \\cup F) = \\mathbb {P} (E) + \\mathbb {P} (E ^ {c} \\cap F) \\\\ = \\mathbb {P} (E) + \\mathbb {P} (F) - \\mathbb {P} (E \\cap F), \\\\ \\end{array}\n$$\nas needed.\n• Example: Uconn Basketball is playing Kentucky this year.\n- Home game has .5 chance of winning\n- Away game has .4 chance of winning.\n- .3 that uconn wins both games.\n- What's the probability that Uconn loses both games?\n- Answer.\n\\* Let $\\mathbb{P}(A_1) = .5$ , $\\mathbb{P}(A_2) = .4$ and $\\mathbb{P}(A_1 \\cap A_2) = .3$ .\n\\* We want to find $\\mathbb{P}(A_1^c\\cap A_2^c)$ . Simplify as much as we can:\n$$\n\\begin{array}{l} \\mathbb {P} \\left(A _ {1} ^ {c} \\cap A _ {2} ^ {c}\\right) = \\mathbb {P} \\left(\\left(A _ {1} \\cup A _ {2}\\right) ^ {c}\\right) \\text { by DeMorgan's Law } \\\\ = 1 - \\mathbb {P} \\left(A _ {1} \\cup A _ {2}\\right), \\text { by Proposition 1c } \\\\ \\end{array}\n$$\n\\* Using Proposition 1e, we have\n$$\n\\mathbb {P} \\left(A _ {1} \\cup A _ {2}\\right) = . 5 +. 4 -. 3 = . 6,\n$$\nHence $\\mathbb{P}(A_1^c\\cap A_2^c) = 1 - .6 = .4$ as needed.", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.2. Axioms of Probability", + "metadata": { + "length": 5714, + "summary": "The text defines probability through axioms: non-negativity, total probability of sample space equals 1, and countable additivity for disjoint events. It introduces relative frequency as motivation but notes its limitations. Key properties derived include probability of empty set is 0, complement rule, monotonicity, and inclusion-exclusion for two events. Examples illustrate fair and biased coin tosses, die rolling, and a basketball game probability calculation using DeMorgan's law and the inclusion-exclusion formula.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Let", + "event", + "How", + "defined", + "probability", + "We", + "attempt", + "define", + "relative", + "frequency", + "Perform", + "experiment", + "Flipping", + "coin", + "times", + "number", + "occurred", + "repetitions", + "Flip", + "1000", + "Tails", + "551", + "Then", + "reasonable", + "mathbb", + "approx", + "frac", + "So", + "lim", + "infty", + "But", + "don", + "limit", + "exists", + "approach", + "Probability", + "rule", + "Axioms", + "Laws", + "agree", + "function", + "input", + "set", + "Axiom", + "leq", + "events", + "underline", + "mathbf", + "disjoint", + "property", + "If", + "ldots", + "pairwise", + "mutually", + "exclusive", + "left", + "bigcup", + "sum", + "Mutually", + "means", + "cap", + "emptyset", + "neq", + "Remark", + "Note", + "subset", + "points", + "However", + "common", + "write", + "Say", + "tossing", + "xoin", + "The", + "heads", + "written", + "Example1", + "Suppose", + "toss", + "equally", + "biased", + "tosse", + "assignment", + "Example2", + "Rolling", + "fair", + "die", + "space", + "consists", + "point", + "compute", + "rolling", + "text", + "rules", + "breaking", + "PROPOSITION", + "cup", + "It", + "helps", + "draw", + "diagrams", + "prove", + "Try", + "PROOF", + "infinite", + "cdots", + "begin", + "array", + "end", + "Use", + "By", + "geq", + "Write", + "Picture", + "venn", + "diagram", + "disjointness", + "Now", + "picture", + "implies", + "substitute", + "equation", + "needed", + "Example", + "Uconn", + "Basketball", + "playing", + "Kentucky", + "year", + "Home", + "game", + "chance", + "winning", + "Away", + "uconn", + "wins", + "games", + "What", + "loses", + "Answer", + "find", + "Simplify", + "DeMorgan", + "Law", + "Proposition", + "1c", + "Using", + "1e", + "Hence" + ], + "keywords": [ + "Probability", + "Axioms", + "Events" + ], + "connect_to": [] + } + }, + { + "chunk_id": "a3071b11-d554-5c48-a450-29a22f16f903", + "type": "text", + "content": "\\- In many experiments, a probability space consists of finitely many points, all with equally likely probabilities.\n- Basic example was a tossing a coin $P(H) = P(T) = \\frac{1}{2}$\n- Fair die: $P(i) = \\frac{1}{6}$ for $i = 1, \\ldots, 6$ .\n\\- In this case from Axiom 3 we have that\n$$\nP (E) = \\frac {\\text { number of outcomes in } E}{\\text { number of outcomes in } S}.\n$$\n\\- Example1: What is the probability that if we roll 2 dice, the sum is 7?\n\\- Answer: There are 36 total outcomes, of which 6 have a sum of 7:\n\\* $E =$ \"sum is 7\" $= \\{(1,6),(2,5),(3,4),(4,3),(5,2),(6,1)\\}$ . Since they are all equally likely, the probability is $\\mathbb{P}(E) = \\frac{6}{6\\cdot6} = \\frac{1}{6}$ .\n\\- Example 2: If 3 balls are “randomly drawn” from a bowl containing 6 white and 5 black balls, what is the probability that one ball is white and the other two are black?\n\\- Method 1: (regard as a ordered selection)\n$$\n\\begin{array}{l} P (E) = \\frac {W B B + B W B + B B W}{1 1 \\cdot 1 0 \\cdot 9} \\\\ = \\frac {6 \\cdot 5 \\cdot 4 + 5 \\cdot 6 \\cdot 4 + 5 \\cdot 4 \\cdot 6}{9 9 0} = \\frac {1 2 0 + 1 2 0 + 1 2 0}{9 9 0} = \\frac {4}{1 1}. \\\\ \\end{array}\n$$\n\\- Method2: (Regard as unordered set of drawn balls)\n$$\nP \\left(E\\right) = \\frac {\\left(1 \\text {white}\\right) \\left(2 \\text {black}\\right)}{\\binom{1 1}{3}} = \\frac {\\binom{6}{1} \\binom{5}{2}}{\\binom{1 1}{3}} = \\frac {4}{1 1}.\n$$\n\\- We can always choose which way to regard our experiments.\n\\- Example 3 A committee of 5 is to selected from a group of 6 men and 9 women. What is probability consists of 3 men and 2 women\n\\- Answer: Easy $\\frac{men\\cdot women}{all}=\\frac{\\left(\\begin{array}{c}6\\\\3\\end{array}\\right)\\left(\\begin{array}{c}9\\\\2\\end{array}\\right)}{\\left(\\begin{array}{c}15\\\\5\\end{array}\\right)}=\\frac{240}{1001}.$\n\\- Example 4: Seven balls are randomly withdrawn from an urn that contains 12 red, 16 blue, and 18 green.\n- (b) Find probability that “at least 2 red balls are withdrawn;”\n- $\\underline{\\mathbf{Ans}}$ : Let $E$ be this event then $P(E) = 1 - P(E^c)$ , $P$ (at least 2 red) = 1- $\\mathbb{P}$ (drawing 0 or 1 balls).\n$$\n\\mathbb {P} \\left(\\text { drawing 0 or 1 red balls }\\right) = \\frac {\\binom{1 6 + 1 8 = 3 4}{7}}{\\binom{4 6}{7}} + \\frac {\\binom{1 2}{1} \\binom{3 4}{6}}{\\binom{4 6}{7}}.\n$$\n\\- Explanation of Poker/Playing cards: Ranks and suits,etc!\n\\- There are 52 cards in a standard deck of playing cards. The poker hand is consists of five cards. There are 4 suits: heats, spades, diamonds, and clubs ( $\\heartsuit$ ♠♦♣). The suits diamonds and hearts are red while clubs and spades are black. In each suit there are 13 ranks: the numbers 2, 3, ..., 10, the face cards, Jack, Queen, King, and the Ace (not a face card).\n\\- Example 5: What is the probability that in a poker hand (5 cards out of 52) we get exactly 4 of a kind?\n\\- Answer: Consider 4 aces and 1 king: $AAAK = \\left( \\begin{array}{l}4\\\\ 4 \\end{array} \\right)\\left( \\begin{array}{l}4\\\\ 1 \\end{array} \\right)$ . But $JJJJ3$ is the same probability.\n\\* Thus there are 13 ways to pick the first rank, and 12 ways to pick the second rank\n$$\n\\begin{array}{l} P (4 \\text { of a kind }) = \\frac {[ \\text { choice of ranks } ] [ \\text { given rank how to choose a hand } ]}{\\binom{5 2}{5}} \\\\ = \\frac {\\left(1 3 \\cdot 1 2 \\cdot \\binom{4}{4} \\binom{4}{1}\\right)}{\\binom{5 2}{5}} \\approx . 0 0 0 0 1 3 9 \\\\ \\end{array}\n$$\n\\- Example 6: What is the probability that in a poker hand (5 cards out of 52) we get a straight. (no straight flushes, can't be of the same suit)\n\\- Answer: Consider: A-2-3-4-5-6-7-8-9-10-J-Q-K-A- There are 10 possible straights.\n\\* Given a straight Say A2345: There are $4 \\cdot 4 \\cdot 4 \\cdot 4 \\cdot 4 - (\\text{of the same suit}) = 4^5 - 4$ .\n$$\n\\begin{array}{l} P (\\text { Straight }) = \\frac {[ \\text { choice of straight } ] [ \\text { given striaght how to choose a hand } ]}{\\binom{5 2}{5}} \\\\ = \\frac {1 0 \\cdot (4 ^ {5} - 4)}{\\binom{5 2}{5}} \\approx . 0 0 3 9 \\\\ \\end{array}\n$$\n\\- Example 7: What is the probability that in a poker hand (5 cards out of 52) we get a Full House. (3 and a 2 of a kind)\n- Answer: It would be [3 of a kind][2 of a kind]. AAAKK or KKAAA are different! Choose suit: $13 \\cdot 12$ .\n- Then once we choose within each group there\n$$\nP (\\text {Full House}) = \\frac {[ \\text {choice of rank} ] [ 3 \\text {of a kind} ] [ 2 \\text {of a kind} ]}{\\binom{5 2}{5}}\n$$\n$$\n= \\frac {1 3 \\cdot 1 2 \\binom{4}{3} \\binom{4}{2}}{\\binom{5 2}{5}} \\approx . 0 0 1 4.\n$$\n\\- Example 8: (Birthday Problem) In a class of 32 people, what is the probability that at least two people have the same birthdays? (We assume each day is equally likely.)\n\\- Answer: Let the first person have a birthday on some day. The probability that the second person has a different birthday will be $\\frac{364}{365}$ . The probability that the third person has a different birthday from the first two people is $\\frac{363}{365}$ . So the answer is\n$$\n\\mathbb {P} \\left(\\text {at least 2 people}\\right) = 1 - \\mathbb {P} \\left(\\text {Everyone different birthday}\\right)\n$$\n$$\n= 1 - \\frac {3 6 5}{3 6 5} \\cdot \\frac {3 6 4}{3 6 5} \\cdot \\frac {3 6 3}{3 6 5} \\dots \\frac {(3 6 5 - 3 1)}{3 6 5}\n$$\n$$\n= 1 - 1 \\cdot \\frac {3 6 4}{3 6 5} \\cdot \\frac {3 6 3}{3 6 5} \\dots \\frac {3 3 4}{3 6 5} \\approx 0. 7 5 2 3 7 4.\n$$\n\\- Really High!!!", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.3. Equally Likely Outcomes", + "metadata": { + "length": 5340, + "summary": "The passage covers basic probability concepts with equally likely outcomes, including examples like rolling dice, drawing balls from an urn, selecting committees, and poker hands. It explains how to compute probabilities using combinatorial formulas, such as combinations and permutations, and includes the birthday problem as an illustration of probability calculations.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "In", + "experiments", + "probability", + "space", + "consists", + "finitely", + "points", + "equally", + "probabilities", + "Basic", + "tossing", + "coin", + "frac", + "Fair", + "die", + "ldots", + "case", + "Axiom", + "text", + "number", + "outcomes", + "Example1", + "What", + "roll", + "dice", + "sum", + "Answer", + "There", + "36", + "total", + "Since", + "mathbb", + "cdot6", + "Example", + "If", + "balls", + "randomly", + "drawn", + "bowl", + "white", + "black", + "ball", + "Method", + "regard", + "ordered", + "selection", + "begin", + "array", + "cdot", + "end", + "Method2", + "Regard", + "unordered", + "set", + "left", + "binom", + "We", + "choose", + "committee", + "selected", + "group", + "men", + "women", + "Easy", + "15", + "240", + "1001", + "Seven", + "withdrawn", + "urn", + "12", + "red", + "16", + "blue", + "18", + "green", + "Find", + "underline", + "mathbf", + "Ans", + "Let", + "event", + "drawing", + "Explanation", + "Poker", + "Playing", + "cards", + "Ranks", + "suits", + "52", + "standard", + "deck", + "playing", + "The", + "poker", + "hand", + "heats", + "spades", + "diamonds", + "clubs", + "heartsuit", + "hearts", + "suit", + "13", + "ranks", + "numbers", + "10", + "face", + "Jack", + "Queen", + "King", + "Ace", + "card", + "kind", + "Consider", + "aces", + "king", + "AAAK", + "But", + "JJJJ3", + "Thus", + "ways", + "pick", + "rank", + "choice", + "approx", + "straight", + "flushes", + "straights", + "Given", + "Say", + "A2345", + "Straight", + "striaght", + "Full", + "House", + "It", + "AAAKK", + "KKAAA", + "Choose", + "Then", + "Birthday", + "Problem", + "class", + "32", + "people", + "birthdays", + "assume", + "day", + "person", + "birthday", + "364", + "365", + "363", + "So", + "answer", + "Everyone", + "dots", + "Really", + "High" + ], + "keywords": [ + "Probability", + "Combinatorics", + "Poker Hands" + ], + "connect_to": [] + } + }, + { + "chunk_id": "d82ad301-7d5f-5e8c-a0cb-238b7ccc249d", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 3", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "44ed73ed-4948-5b67-b757-7c31fa97bdde", + "type": "image", + "content": "\nPROPOSITION 2. If $E$ and $F$ are independent, then $E$ and $F^c$ are independent. \\- Remark: Independence and mutually exclusive, are two different things!0\n[images/image-10-PROPOSITIO.jpg]\n", + "path": "images/image-10-PROPOSITIO.jpg", + "metadata": { + "length": 192, + "summary": "image-10\nPROPOSITION 2. If $E$ and $F$ are independent, then $E$ and $F^c$ are independent. \\- Remark: Independence and mutually exclusive, are two different things!0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-10-PROPOSITIO.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "24361958-8eaf-5c7e-8aeb-24e9059545d1", + "type": "text", + "content": "3.1. Independent Events\nDEFINITION. We say $E$ and $F$ are independent events if\n$$\n\\mathbb {P} (E \\cap F) = \\mathbb {P} (E) \\mathbb {P} (F).\n$$\n• Example1: Suppose you flip two coins.\n- The event that you get heads on the second coin is independent of the event that you get tails on the first.\n- This is why: Let $A_{t}$ be the event of getting is tails for the first coin and $B_{h}$ is the event of getting heads for the second coin, and we assume we have fair coins (although this is not necessary), then\n$$\n\\mathbb {P} \\left(A _ {t} \\cap B _ {h}\\right) = \\frac {1}{4}, \\text { list out all outcomes }\n$$\n$$\n\\mathbb {P} \\left(A _ {t}\\right) \\mathbb {P} \\left(B _ {h}\\right) = \\frac {1}{2} \\frac {1}{2} = \\frac {1}{4}.\n$$\n\\- Example2: Experiment: Draw a card from an ordinary deck of cards\n\\- Let $A =$ draw ace, $S =$ draw a spade.\n\\* These are independent events since you're taking one at a time, so one doesn't effect the other. To see this using the definition we have compute\n\\* $\\mathbb{P}(A)\\mathbb{P}(S) = \\frac{1}{13}\\frac{1}{4}.$\n\\* White $\\mathbb{P}(A\\cap S) = \\frac{1}{52}$ since there is only 1 Ace of spades.\nPROPOSITION 2. If $E$ and $F$ are independent, then $E$ and $F^c$ are independent.\nPROOF. Draw a Venn Diagram to help with the computation, but note that\n$$\n\\begin{array}{l} \\mathbb {P} (E \\cap F ^ {c}) = \\mathbb {P} (E) - \\mathbb {P} (E \\cap F) \\\\ = \\mathbb {P} (E) - \\mathbb {P} (E) \\mathbb {P} (F) \\\\ = \\mathbb {P} (E) (1 - \\mathbb {P} (F)) \\\\ = \\mathbb {P} (E) \\mathbb {P} \\left(F ^ {c}\\right). \\\\ \\end{array}\n$$\n\nPROPOSITION 2. If $E$ and $F$ are independent, then $E$ and $F^c$ are independent. \\- Remark: Independence and mutually exclusive, are two different things!0\n[images/image-10-PROPOSITIO.jpg]\n\n\\- Remark: Independence and mutually exclusive, are two different things!\nDEFINITION. We say $E, F, G$ are independent if $E, F$ are independent, $E, G$ are independent, $F, G$ are independent, and $\\mathbb{P}(E \\cap F \\cap G) = \\mathbb{P}(E)\\mathbb{P}(F)\\mathbb{P}(G)$ .\n• Example: Experiment is you roll two dice:\n\\- Define the following events:\n- $S_{7} = \\{\\text{sum is 7}\\}$\n- $A_{4} = \\{\\text{first die is a 4}\\}$\n- $B_{3} = \\{$ second die is a 3\\}\n- Are the events $S_{7}, A_{4}, B_{3}$ independent?\n\\* Compute\n$$\n\\mathbb {P} \\left(S _ {7} \\cap A _ {4} \\cap B _ {3}\\right) = \\mathbb {P} \\left(\\{(4, 3) \\}\\right) = \\frac {1}{3 6}\n$$\nbut\n$$\n\\mathbb {P} \\left(S _ {7}\\right) \\mathbb {P} \\left(A _ {4}\\right) \\mathbb {P} \\left(B _ {3}\\right) = \\frac {6}{3 6} \\frac {1}{6} \\frac {1}{6} = \\frac {1}{3 6 \\cdot 6}.\n$$\n\\- Remark: This generalizes to events $A_1, \\ldots, A_n$ . We say events $A_1, \\ldots, A_n$ are independent if for all subcollections $i_1, \\ldots, i_r \\in \\{1, \\ldots, n\\}$ we have that $\\mathbb{P}\\left(\\bigcap_{j=1}^{r} A_{i_j}\\right) = \\prod_{j=1}^{r} \\mathbb{P}\\left(A_{i_j}\\right)$ .\n• Example:\n- An urn contains 10 balls: 4 red and 6 blue.\n- A second urn contains 16 red balls and an unknown number of blue balls.\n- A single ball is drawn from each urn. The probability that both balls are the same color is 0.44.\n- Question: Calculate the number of blue balls in the second urn.\n- Solution: Let $R_{i} =$ even that a red ball is drawn from urn $i$ and let $B_{i} =$ event that a blue ball is drawn from urn $i$ .\n\\* Let $x$ be the number of blue balls in urn 2,\n\\* Note that drawing from urn 1 and independent from drawing from urn 2. They are completely different urns! They shouldn't effect the other.\n\\* Then\n$$\n\\begin{array}{l} . 4 4 = \\mathbb {P} \\left(\\left(R _ {1} \\cap R _ {2}\\right) \\bigcup \\left(B _ {1} \\cap B _ {2}\\right)\\right) = \\mathbb {P} \\left(R _ {1} \\cap R _ {2}\\right) + \\mathbb {P} \\left(B _ {1} \\cap B _ {2}\\right) \\\\ = \\mathbb {P} (R _ {1}) \\mathbb {P} (R _ {2}) + \\mathbb {P} (B _ {1}) \\mathbb {P} (B _ {2}), \\text { by independence } \\\\ = \\frac {4}{1 0} \\frac {1 6}{x + 1 6} + \\frac {6}{1 0} \\frac {x}{x + 1 6}. \\\\ \\end{array}\n$$\n\\* Solve for $x!$ You will get $x = 4$ .\n\\- Example (Gambler's Ruin) (Used in Finance or Actuarial Science)\n\\- Experiment: Suppose you toss a fair coin repeatedly and independently. If it comes up heads, you win a dollar, and if it comes up tails, you lose a dollar. Suppose you start with \\$50. What's the probability you will get to \\$200 before you go broke?\n\\- Answer: It's actually easier if we generalize the problem.\n\\* Let $p(x)$ be the probability you get 200 before 0 if you start with $x$ dollars.\n\\* We know $p(0) = 0$ and $p(200) = 1$ . So by the law of total probability\n$$\n\\begin{array}{l} p (x) = \\mathbb {P} (\\text { Win 200 before 0 }) \\\\ = \\mathbb {P} (H) \\mathbb {P} (\\text { Win } 2 0 0 \\text { before } 0 \\mid H) + \\mathbb {P} \\left(H ^ {c}\\right) \\mathbb {P} (\\text { Win } 2 0 0 \\text { before } 0 \\mid H ^ {c}) \\\\ = \\frac {1}{2} p (x + 1) + \\frac {1}{2} p (x - 1). \\\\ \\end{array}\n$$\n\\* Rearrange the function to get\n$$\n\\begin{array}{l} 2 p (x) = p (x - 1) + p (x + 1) \\quad \\Longleftrightarrow \\quad p (x) + p (x) = p (x - 1) + p (x + 1) \\\\ \\Longleftrightarrow \\quad p (x) - p (x - 1) = p (x + 1) - p (x) \\\\ \\Longleftrightarrow \\quad \\frac {p (x) - p (x - 1)}{x - (x - 1)} = \\frac {p (x + 1) - p (x)}{(x + 1) - x}. \\\\ \\end{array}\n$$\n\\* This tells you that the slows are constant. What does that tell you about $p(x)$ ? It's a line!\n\\- Thus we must have $p(x) = \\frac{x}{200}$ .\n\\* Thus $p(50) = \\frac{1}{4}$ .\n• Example (A variation of Gambler's ruin)\n\\- Problem: Suppose we are in the same situation, but you are allowed to go arbitrarily far in debt. Let $p(x)$ be the probability you ever get to \\$200. What is a formula for $p(x)$ ?\n\\* Answer: Just as before $p(x) = \\frac{1}{2} p(x + 1) + \\frac{1}{2} p(x - 1)$ . So that $p(x)$ is linear.\n\\* But now all we have is that $p(200) = 1$ and linear and domain is $(-\\infty, 200)$ .\n\\* Draw a graph: Now the slope, or $p'(x)$ can't be negative, or else we would have it that $p(x) > 1$ for $x \\in (-\\infty, 200)$ .\n\\- The slope can't be positive or else we would get $p(x) < 0$ for $x \\in (-\\infty, 200)$ .\n\\* Thus we must have that $p(x) \\equiv \\text{constant}$ . Hence $p(x) = 1$ for all $x \\in (-\\infty.200)$ .\n\\* Sol: So we are certain to get \\$200 if we cna get into debt.\n\\- Method2:\n\\* Just compute There is nothing special about the figure 200. Another way of seeing this is to compute as above the probability of getting to 200 before $-M$ and then letting $M \\to \\infty$ .\n\\- We would get $p(x)$ is a line with $p(-M) = 0$ and $p(200) = 1$ so that\n$$\np (x) - 0 = \\frac {1 - 0}{2 0 0 - (- M)} (x - (- M))\n$$\nand letting $M \\to \\infty$ wee see that $p(x) = \\frac{x + M}{200 + M} \\to 1$ .\n• Example: Experiment: Roll 10 dice.\n- What is the probability that exactly 4 twos will show if you roll 10 dice?\n- Answer: These are independent. The probability that the 1st, 2nd, 3rd, and 10th dice will show a three and the other 6 will not is $\\left(\\frac{1}{6}\\right)^3\\left(\\frac{5}{6}\\right)^7$ .\n- Independence is used here: the probability is $\\frac{1}{6}\\frac{1}{6}\\frac{1}{6}\\frac{5}{6}\\frac{5}{6}\\frac{5}{6}\\frac{5}{6}\\frac{5}{6}\\frac{5}{6}\\frac{1}{6}$ . Note that the probability that the 10th, 9th, 8th, and 7th dice will show a two and the other 6 will not have the same probability.\n- So to answer our original question, we take $\\left(\\frac{1}{6}\\right)^4\\left(\\frac{5}{6}\\right)^6$ and multiply it by the number of ways of choosing 4 dice out of 10 to be the ones showing the twos. There are $\\binom{10}{3}$ ways to do this $\\binom{10}{4}\\left(\\frac{1}{6}\\right)^4\\left(\\frac{5}{6}\\right)^6$ .\n\\- This is an example of Bernoulli trials, or the Binomial distribution.\n\\- If we have $n$ independent trials, where the probability of success if $p$ . The probability that there are $k$ successes in $n$ trials is\n$$\n\\binom{n}{k} p ^ {k} (1 - p) ^ {n - k}.\n$$", + "path": "Notes.pdf/CHAPTER 3/Independence", + "metadata": { + "length": 7777, + "summary": "The text defines independent events as those where the probability of their intersection equals the product of their individual probabilities. It provides examples including coin flips, card draws, and dice rolls. It proves that if events E and F are independent, then E and F^c are also independent. The definition extends to three or more events requiring pairwise and joint independence. Applications include calculating probabilities for drawing balls from urns and the Gambler's Ruin problem, where the probability of reaching a target before ruin is linear. The text also introduces Bernoulli trials and the binomial distribution for independent trials with constant success probability.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "3.1", + "Independent", + "Events", + "DEFINITION", + "We", + "independent", + "events", + "mathbb", + "cap", + "Example1", + "Suppose", + "flip", + "coins", + "The", + "event", + "heads", + "coin", + "tails", + "This", + "Let", + "assume", + "fair", + "left", + "frac", + "text", + "list", + "outcomes", + "Example2", + "Experiment", + "Draw", + "card", + "ordinary", + "deck", + "cards", + "draw", + "ace", + "spade", + "These", + "taking", + "time", + "doesn", + "effect", + "To", + "definition", + "compute", + "13", + "White", + "52", + "Ace", + "spades", + "PROPOSITION", + "If", + "PROOF", + "Venn", + "Diagram", + "computation", + "note", + "begin", + "array", + "end", + "Remark", + "Independence", + "mutually", + "exclusive", + "things", + "Example", + "roll", + "dice", + "Define", + "sum", + "die", + "Are", + "Compute", + "cdot", + "generalizes", + "ldots", + "subcollections", + "bigcap", + "prod", + "An", + "urn", + "10", + "balls", + "red", + "blue", + "16", + "unknown", + "number", + "single", + "ball", + "drawn", + "probability", + "color", + "0.44", + "Question", + "Calculate", + "Solution", + "Note", + "drawing", + "They", + "completely", + "urns", + "shouldn", + "Then", + "bigcup", + "independence", + "Solve", + "You", + "Gambler", + "Ruin", + "Used", + "Finance", + "Actuarial", + "Science", + "toss", + "repeatedly", + "independently", + "win", + "dollar", + "lose", + "start", + "50", + "What", + "200", + "broke", + "Answer", + "It", + "easier", + "generalize", + "problem", + "dollars", + "So", + "law", + "total", + "Win", + "mid", + "Rearrange", + "function", + "quad", + "Longleftrightarrow", + "tells", + "slows", + "constant", + "line", + "Thus", + "variation", + "ruin", + "Problem", + "situation", + "allowed", + "arbitrarily", + "debt", + "formula", + "Just", + "linear", + "But", + "domain", + "infty", + "graph", + "Now", + "slope", + "negative", + "positive", + "equiv", + "Hence", + "infty.200", + "Sol", + "cna", + "Method2", + "There", + "special", + "figure", + "Another", + "letting", + "wee", + "Roll", + "twos", + "show", + "1st", + "2nd", + "3rd", + "10th", + "9th", + "8th", + "7th", + "answer", + "original", + "question", + "multiply", + "ways", + "choosing", + "showing", + "binom", + "Bernoulli", + "trials", + "Binomial", + "distribution", + "success", + "successes" + ], + "keywords": [ + "Independence", + "Probability", + "Gambler's Ruin" + ], + "connect_to": [ + { + "target": "44ed73ed-4948-5b67-b757-7c31fa97bdde", + "relation": "embeds", + "ref": "[images/image-10-PROPOSITIO.jpg]", + "position": { + "start": 1712, + "end": 1744 + } + } + ] + } + }, + { + "chunk_id": "991e2590-bc40-5145-bde3-5c782f1247c9", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 4", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "39e95be7-e389-550d-adca-409859d3ac06", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 4/Conditional Probability and Independence", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f8dccbfb-46d0-5cef-bd97-5901cd46ec6d", + "type": "table", + "content": "
    WifeHusband makes less than 25,000Husband makes more than 25,000
    Less than $25,000212198
    More than $25,0003654
    ", + "path": "tables/table-0 Income by Gender.html", + "metadata": { + "length": 231, + "summary": "table-1\nA table shows counts of couples based on wife's and husband's income levels, with 212 couples both under $25k and 54 both over $25k.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-0 Income by Gender.html", + "keywords": [ + "income", + "husband", + "wife" + ], + "tokens": [] + } + }, + { + "chunk_id": "a1d19f26-7ddc-5cff-b1d1-00765b6e9ad4", + "type": "text", + "content": "\\- Suppose there are\n- 200 men, of which 100 are smokers,\n- 100 women, of which 20 are smokers.\n- Question1: What is the probability that a person chosen at random will be a smoker? $\\frac{120}{300}$\n- Question2: Now, let us ask, what is the probability that a person chosen at random is a smoker given that the person is a women? $\\frac{20}{100}$ right?\n\\* Note this is\n$$\n\\frac {\\# (\\text { women smokers })}{\\# (\\text { women })} = \\frac {P (\\text { women and a smoker })}{P (\\text { woman })}.\n$$\n\\- Thus we make the following definition:\nDEFINITION. If $\\mathbb{P}(F) > 0$ , we define\n$$\n\\mathbb {P} (E \\mid F) = \\frac {\\mathbb {P} (E \\cap F)}{\\mathbb {P} (F)}.\n$$\nNow $\\mathbb{P}(E\\mid F)$ is read \"the probability of $E$ given $F$ .\"\n- Note that $\\mathbb{P}(E\\cap F) = \\mathbb{P}(E|F)\\mathbb{P}(F)!$\n- This is the conditional probability that $E$ occurs given that $F$ has already occurred!\n- Remark: Suppose $\\mathbb{P}(E \\mid F) = \\mathbb{P}(E)$ , i.e. knowing $F$ doesn't help predict $E$ . Then this implies that $E$ and $F$ are independent of each other. Rearranging $\\mathbb{P}(E \\mid F) = \\frac{\\mathbb{P}(E \\cap F)}{\\mathbb{P}(F)} = \\mathbb{P}(E)$ we see that $\\mathbb{P}(E \\cap F) = \\mathbb{P}(E)\\mathbb{P}(F)$ .\n- Example1: Experiment: Roll two dice.\n- (a) What is the probability the sum is 8?\n\\* Solution: Note that $A = \\{(2,6), (3,5), (4,4), (5,3), (6,2)\\}$ so we know $\\mathbb{P}(A) = \\frac{5}{36}$ .\n- (b) What is the probability that the sum is 8 given that the first die shows a 3? (In other words, find $\\mathbb{P}(A \\mid B)$ )\n\\* Solution: Let $B = \\{\\text{first die shows three}\\}$ .\n\\* $\\mathbb{P}(A\\cap B) = \\mathbb{P}\\left(\\{(3,5)\\}\\right) = \\frac{1}{36}$ is probability that the first die shows a 3 and the sum is 8\n\\* Finally we can compute\n$$\n\\mathbb {P} (A \\mid B) = \\mathbb {P} (\\text { sum is } 8 \\mid 1 \\text { st is a } 3) = \\frac {1 / 3 6}{1 / 6} = \\frac {1}{6}.\n$$\n\\- Remark: When computing $\\mathbb{P}(E \\mid F)$ , Sometime its easier to work with the reduced sample space $F \\subset S$ .\n\\- Note in the previous example when we computed\n$$\n\\mathbb {P} (\\text { sum is } 8 \\mid 1 \\text { st is a } 3)\n$$\nwe could have worked in the smaller sample space of $\\{1st\\ is\\ a\\ 3\\}=\\{(3,1),(3,2),(3,3),(3,4),(3,5),(3,6)\\}$ . Since only $(3,5)$ begins with a 3 and has the sum of 8, then the probability is\n$$\n\\frac {\\text { total number of outcomes in the event }}{\\text { total number of outcomes in new sample space }} = \\frac {1}{6}.\n$$\n\\- Example2: Experiment: Suppose a box has 3 red marbles and 2 black ones. We select 2 marbles.\n- Question: What is the probability that second marble is red given that the first one is red?\n\\* Answer:\n$\\overline{\\cdot R_1} = \\{\\text{First marble is is red}\\}$ ,\n$R_{2} = \\{\\text{Second marble is red}\\}$ , then\n$$\n\\begin{array}{l} \\mathbb {P} \\left(R _ {2} \\mid R _ {1}\\right) = \\frac {\\mathbb {P} \\left(R _ {1} \\cap R _ {2}\\right)}{\\mathbb {P} \\left(R _ {1}\\right)} \\\\ = \\frac {(2 \\text { red }) (0 \\text { black }) / \\binom{5}{2}}{3 / 5} \\\\ = \\frac {\\binom{3}{2} \\binom{2}{0} / \\binom{5}{2}}{3 / 5} \\\\ = \\frac {3 / 1 0}{3 / 5} = \\frac {1}{2}. \\\\ \\end{array}\n$$\n\\* Solution 2:\n\\- We could have done the same example more easily if we look at the new sample space $S' = \\{R, R, B, B\\}$ thus $\\mathbb{P}(R_2 \\mid R_1) = \\mathbb{P}'(\\{\\text{drawing red}\\}) = \\frac{2}{4} = \\frac{1}{2}$ .\n\\- Example3: Landon is 80% sure he forgot his textbook at the Union or Monteith buildings. 40% sure that it is at the union, and 40% sure that it is at Monteith. Given that Landon already went to Monteith and noticed his textbook not there, what is the probability that it's at the Union?\n\\- Solution:\n$$\n\\begin{array}{l} \\mathbb {P} \\left(U n i o n \\mid N o t \\text { Monteith }\\right) = \\frac {\\mathbb {P} \\left(U \\cap M ^ {c}\\right)}{\\mathbb {P} \\left(M ^ {c}\\right)} \\\\ = \\frac {\\mathbb {P} (U)}{1 - \\mathbb {P} (M)}, \\text { since } U \\subset M ^ {c} \\\\ = \\frac {4 / 1 0}{6 / 1 0} = \\frac {2}{3}. \\\\ \\end{array}\n$$\n\\- Example4: Suppose that Annabelle and Bobby each draw 13 cards from a standard deck of 52. Given that Sarah has exactly two aces, what is the probability that Bobby has exactly one ace?\n\\- Solution: Let $A$ be the event \"Annabelle has two aces,\" and let $B$ be the event \"Bobby has exactly one ace.\" Again, we want $\\mathbb{P}(B \\mid A)$ , so we calculate $\\mathbb{P}(A)$ and $\\mathbb{P}(A \\cap B)$ . Annabelle could have any of $\\left( \\begin{array}{c}52\\\\ 13 \\end{array} \\right)$ possible hands. Of these hands, $\\left( \\begin{array}{c}4\\\\ 2 \\end{array} \\right) \\cdot \\left( \\begin{array}{c}48\\\\ 11 \\end{array} \\right)$ will have exactly\ntwo aces, so\n$$\n\\mathbb {P} \\left(A\\right) = \\frac {\\binom{4}{2} \\cdot \\binom{4 8}{1 1}}{\\binom{5 2}{1 3}}.\n$$\nNow the number of ways in which Annabelle can have a certain hand and Bobby can have a certain hand is $\\binom{52}{13}\\cdot\\binom{39}{13}$ , and the number of ways in which A and B can both occur is $\\binom{4}{2}\\cdot\\binom{48}{11}\\cdot\\binom{2}{1}\\cdot\\binom{37}{12}$ . so\n$$\n\\mathbb {P} (A \\cap B) = \\frac {\\binom{4}{2} \\cdot \\binom{4 8}{1 1} \\cdot \\binom{2}{1} \\cdot \\binom{3 7}{1 2}}{\\binom{5 2}{1 3} \\cdot \\binom{3 9}{1 3}}.\n$$\nTherefore,\n$$\n\\begin{array}{l} \\mathbb {P} (B \\mid A) = \\frac {\\mathbb {P} (A \\cap B)}{\\mathbb {P} (A)} = \\frac {\\frac {\\left( \\begin{array}{c} 4 \\\\ 2 \\end{array} \\right) \\cdot \\left( \\begin{array}{c} 4 8 \\\\ 1 1 \\end{array} \\right) \\cdot \\left( \\begin{array}{c} 2 \\\\ 1 \\end{array} \\right) \\cdot \\left( \\begin{array}{c} 3 7 \\\\ 1 2 \\end{array} \\right)}{\\left( \\begin{array}{c} 5 2 \\\\ 1 3 \\end{array} \\right) \\cdot \\left( \\begin{array}{c} 3 9 \\\\ 1 3 \\end{array} \\right)}}{\\frac {\\left( \\begin{array}{c} 4 \\\\ 2 \\end{array} \\right) \\cdot \\left( \\begin{array}{c} 4 8 \\\\ 1 1 \\end{array} \\right)}{\\left( \\begin{array}{c} 5 2 \\\\ 1 3 \\end{array} \\right)}} \\\\ = \\frac {\\binom{2}{1} \\cdot \\binom{3 7}{1 2}}{\\binom{3 9}{1 3}}. \\\\ \\end{array}\n$$\n\\- Note that since $\\mathbb{P}(B \\mid A) = \\frac{\\mathbb{P}(A \\cap B)}{\\mathbb{P}(A)}$ then $\\mathbb{P}(A \\cap B) = \\mathbb{P}(A)\\mathbb{P}(B \\mid A)$ .\n\\- In general: If $E_1, \\ldots, E_n$ are events then\n$$\n\\mathbb {P} \\left(E _ {1} \\cap E _ {2} \\cap \\dots \\cap E _ {n}\\right) = \\mathbb {P} \\left(E _ {1}\\right) \\mathbb {P} \\left(E _ {2} \\mid E _ {1}\\right) \\mathbb {P} \\left(E _ {3} \\mid E _ {1} \\cap E _ {2}\\right) \\dots \\mathbb {P} \\left(E _ {n} \\mid E _ {1} \\cap E _ {2} \\cap \\dots \\cap E _ {n - 1}\\right).\n$$\n\\- Example5:\n\\- Experiment: Suppose an urn has 5 White balls and 7 Black balls. Each ball that is selected is returned to the urn along with an additional ball of the same color. Suppose draw 3 balls.\n\\- Part (a): What is the probability that you get 3 white balls.\n\\* Then\n$$\n\\mathbb {P} (3 \\text {white balls}) = \\mathbb {P} (1 \\text {st W}) \\mathbb {P} (2 \\text {nd W} | 1 \\text {st W}) \\mathbb {P} (3 \\text {nd W} | 1 \\text {st} \\& 2 \\text {ndW})\n$$\n$$\n= \\frac {5}{1 2} \\frac {6}{1 3} \\frac {7}{1 4}\n$$\n\\- Part (b): What is the probability of getting 1 white ball.\n$$\n\\begin{array}{l} \\mathbb {P} (1 \\text { white ball }) = \\mathbb {P} (W B B) + \\mathbb {P} (B W B) + \\mathbb {P} (B B W) \\\\ = 3 \\frac {5 \\cdot 7 \\cdot 8}{1 2 \\cdot 1 3 \\cdot 1 4}. \\\\ \\end{array}\n$$\n\\- Note that\n$$\n\\mathbb {P} (E \\cap F) = \\mathbb {P} (E \\mid F) \\mathbb {P} (F)\n$$\n\\- Example 6: Phan wants to take a Biology course or a Chemistry course. Given that the students take Biology, the probability that they get an $A$ is is $\\frac{4}{5}$ . While the probability of getting an $A$ given that the student took Chemistry is $\\frac{1}{7}$ . If Phan makes a decision on the course to take randomly, what's probability of \"getting an $A$ in Chem\"?\n\\- Solution: Let $B = \\{\\text{Takes Biology}\\}$ and $C = \\{\\text{Takes Chemistry}\\}$ and $A = \\{\"\\text{gets an A\"}\\}$ , then\n$$\n\\begin{array}{l} \\mathbb {P} (A \\cap C) = \\mathbb {P} (C) \\mathbb {P} (A \\mid C) \\\\ = \\frac {1}{2} \\cdot \\frac {1}{7} = \\frac {1}{1 4}. \\\\ \\end{array}\n$$\n• Example 7: A total of 500 married couples are poled about salaries:\n\n[tables/table-0 Income by Gender.html]\n\n\\- Part (a): Find the probability that a Husband earns less than 25,000?\n$$\n* \\underline {{\\text {Answer:}}} \\frac {2 1 2 + 3 6}{5 0 0}\n$$\n\\- Part (b): Find $\\mathbb{P}$ (wife makes $>25,000$ | Husband makes $>25,000$ )\n$$\n* \\underline {{\\text {Answer:}}} \\frac {5 4 / 5 0 0}{(1 9 8 + 5 4) / 5 0 0} = \\frac {5 4}{2 5 2} = . 2 1 4\n$$\n\\- Part (c): Find $\\mathbb{P}$ (wife makes $>25,000$ | Husband makes $< 25,000$ )\n$$\n* \\underline {{\\text { Answer: }}} \\frac {3 6 / 5 0 0}{(2 4 8) / 5 0 0} = . 1 4 5.\n$$", + "path": "Notes.pdf/CHAPTER 4/Conditional Probability and Independence/4.1. Conditional Probabilities", + "metadata": { + "length": 8628, + "summary": "The text defines conditional probability P(E|F)=P(E∩F)/P(F) for P(F)>0, and explains that if P(E|F)=P(E) then E and F are independent with P(E∩F)=P(E)P(F). It provides several examples: rolling two dice to find P(sum=8|first die=3)=1/6; drawing marbles from a box (P(second red|first red)=1/2); Landon's textbook problem (P(Union|not Monteith)=2/3); card drawing with Annabelle and Bobby (P(Bobby has one ace|Annabelle has two aces)=C(2,1)C(37,12)/C(39,13)); urn with replacement and additional balls (P(3 white)=5/12*6/13*7/14, P(1 white)=3*5*7*8/(12*13*14)); Phan's course selection (P(A∩C)=1/2*1/7=1/14); and a table of 500 married couples' salaries (P(husband<25k)=248/500, P(wife>25k|husband>25k)=54/252≈0.214, P(wife>25k|husband<25k)=36/248≈0.145). The general multiplication rule for n events is also given.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Suppose", + "200", + "men", + "100", + "smokers", + "women", + "20", + "Question1", + "What", + "probability", + "person", + "chosen", + "random", + "smoker", + "frac", + "120", + "300", + "Question2", + "Now", + "Note", + "text", + "woman", + "Thus", + "make", + "definition", + "DEFINITION", + "If", + "mathbb", + "define", + "mid", + "cap", + "read", + "This", + "conditional", + "occurs", + "occurred", + "Remark", + "knowing", + "doesn", + "predict", + "Then", + "implies", + "independent", + "Rearranging", + "Example1", + "Experiment", + "Roll", + "dice", + "sum", + "Solution", + "36", + "die", + "shows", + "In", + "words", + "find", + "Let", + "left", + "Finally", + "compute", + "st", + "When", + "computing", + "Sometime", + "easier", + "work", + "reduced", + "sample", + "space", + "subset", + "previous", + "computed", + "worked", + "smaller", + "1st", + "Since", + "begins", + "total", + "number", + "outcomes", + "event", + "Example2", + "box", + "red", + "marbles", + "black", + "We", + "select", + "Question", + "marble", + "Answer", + "overline", + "cdot", + "First", + "Second", + "begin", + "array", + "binom", + "end", + "easily", + "drawing", + "Example3", + "Landon", + "80%", + "forgot", + "textbook", + "Union", + "Monteith", + "buildings", + "40%", + "union", + "Given", + "noticed", + "Example4", + "Annabelle", + "Bobby", + "draw", + "13", + "cards", + "standard", + "deck", + "52", + "Sarah", + "aces", + "ace", + "Again", + "calculate", + "hands", + "Of", + "48", + "11", + "ways", + "hand", + "39", + "occur", + "37", + "12", + "Therefore", + "general", + "ldots", + "events", + "dots", + "Example5", + "urn", + "White", + "balls", + "Black", + "Each", + "ball", + "selected", + "returned", + "additional", + "color", + "Part", + "white", + "ndW", + "Example", + "Phan", + "Biology", + "Chemistry", + "students", + "While", + "student", + "makes", + "decision", + "randomly", + "Chem", + "Takes", + "500", + "married", + "couples", + "poled", + "salaries", + "Find", + "Husband", + "earns", + "25", + "000", + "underline", + "wife" + ], + "keywords": [ + "Conditional Probability", + "Independence", + "Bayes" + ], + "connect_to": [ + { + "target": "f8dccbfb-46d0-5cef-bd97-5901cd46ec6d", + "relation": "embeds", + "ref": "[tables/table-0 Income by Gender.html]", + "position": { + "start": 8049, + "end": 8087 + } + } + ] + } + }, + { + "chunk_id": "eb7d7e65-56c6-5ea6-b89c-fd2053108647", + "type": "text", + "content": "- Sometimes it's easier to compute a probability once we know something has or has not happened.\n- Note that we can compute,\n$$\n\\mathbb {P} (E) = \\mathbb {P} (E \\cap F) + \\mathbb {P} (E \\cap F ^ {c})\n$$\n$$\n= \\mathbb {P} (E \\mid F) \\mathbb {P} (F) + \\mathbb {P} (E \\mid F ^ {c}) \\mathbb {P} (F ^ {c})\n$$\n$$\n= \\mathbb {P} (E \\mid F) \\mathbb {P} (F) + \\mathbb {P} (E \\mid F ^ {c}) (1 - \\mathbb {P} (F)).\n$$\n\\- This formula is called: The Law of Total Probability:\n$$\n\\mathbb {P} (E) = \\mathbb {P} (E \\mid F) \\mathbb {P} (F) + \\mathbb {P} (E \\mid F ^ {c}) (1 - \\mathbb {P} (F))\n$$\n\\- The following problem will describe the types of problems of this section.\n• Example1: Insurance company believes\n- The probability that “an accident prone person” has an accident within a year is .4.\n- The probability that “Non-accident prone person” has an accident with year is .2.\n- $30\\%$ of the population is \"accident prone\".\n- Part (a): Find $\\mathbb{P}(A_1)$ where $A_{1} =$ new policy holder will have an accident within a year?\n\\* Let $A = \\{\\text{Policy holder IS accident prone.}\\}$\n$$\n\\mathbb {P} \\left(A _ {1}\\right) = \\mathbb {P} \\left(A _ {1} \\mid A\\right) \\mathbb {P} (A) + \\mathbb {P} \\left(A _ {1} \\mid A ^ {c}\\right) \\left(1 - \\mathbb {P} (A)\\right)\n$$\n$$\n= . 4 (. 3) +. 2 (1 -. 3)\n$$\n$$\n= . 2 6\n$$\n\\- Part (b): Suppose new policyholder has accident with one year. What's probability that he or she is accident prone?\n$$\n\\begin{array}{l} \\mathbb {P} (A \\mid A _ {1}) = \\frac {\\mathbb {P} (A \\cap A _ {1})}{\\mathbb {P} (A _ {1})} \\\\ = \\frac {\\mathbb {P} (A) \\mathbb {P} \\left(A _ {1} \\mid A\\right)}{. 2 6} \\\\ = \\frac {(. 3) (. 4)}{. 2 6} = \\frac {6}{1 3}. \\\\ \\end{array}\n$$\n\\- In general:\n\\- So in Part (a) we had to break a probability into two cases: If $F_1, \\ldots, F_n$ are mutually exclusive events such that they make up everywhere $S = \\bigcup_{i=1}^{n} F_i$ then\n$$\n\\mathbb {P} (E) = \\sum_ {i = 1} ^ {n} \\mathbb {P} (E \\mid F _ {i}) \\mathbb {P} (F _ {i}).\n$$\n\\* This is called Law of Total Probability.\n\\- In Part (b), we wanted to find a probability of a separate conditional event: then\n$$\n\\mathbb {P} \\left(F _ {j} \\mid E\\right) = \\frac {\\mathbb {P} \\left(E \\mid F _ {j}\\right) \\mathbb {P} \\left(F _ {j}\\right)}{\\sum_ {i = 1} ^ {n} \\mathbb {P} \\left(E \\mid F _ {i}\\right) \\mathbb {P} \\left(F _ {i}\\right)}.\n$$\n\\* This is known as Baye's Formula\n\\* Note that the denominator of the Bayes's formula is the Law of total probability.\n• Example2: Suppose the test for HIV is\n- $98\\%$ accurate in both directions\n- $0.5\\%$ of the population is HIV positive.\n- Question: If someone tests positive, what is the probability they actually are HIV positive?\n- Solution: Let $T_{+} = \\{\\text{tests positive}\\}$ , $T_{-} = \\{\\text{tests negative}\\}$ , while $+ = \\{\\text{actually HIV positive},\\}$ $- = \\{\\text{actually negative}\\}$ .\n\\* Want\n$$\n\\begin{array}{l} \\mathbb {P} (+ \\mid T _ {+}) = \\frac {\\mathbb {P} (+ \\cap T _ {+})}{\\mathbb {P} (T _ {+})} \\\\ = \\frac {\\mathbb {P} (T _ {+} \\mid +) \\mathbb {P} (+)}{\\mathbb {P} (T _ {+} \\mid +) \\mathbb {P} (+) + \\mathbb {P} (T _ {+} \\mid -) \\mathbb {P} (-)} \\\\ = \\frac {(. 9 8) (. 0 0 5)}{(. 9 8) (. 0 0 5) + . 0 2 (. 9 9 5)} \\\\ = 19.8 \\% . \\\\ \\end{array}\n$$\n• Example3: Suppose\n- $30\\%$ of the women in a class received an A on the test\n- $25\\%$ of the men/or else received an A.\n- $60\\%$ of the class are women.\n- Question: Given that a person chosen at random received an A, what is the probability this person is a women?\n\\* Solution: Let A the event that a students receives an A. Let W = being a women, M = not a women. Want\n$$\n\\mathbb {P} (W \\mid A) = \\frac {\\mathbb {P} (A \\mid W) \\mathbb {P} (W)}{\\mathbb {P} (A \\mid W) \\mathbb {P} (W) + \\mathbb {P} (A \\mid M) \\mathbb {P} (M)}, \\text { by Bayes's }\n$$\n$$\n= \\frac {. 3 (. 6)}{. 3 (. 6) + . 2 5 (. 4)} = \\frac {. 1 8}{. 2 8} \\approx . 6 4.\n$$\n- (General Baye's Theorem) Here's one with more than 3 possibilities:\n- Example4: Suppose in Factory with Machines I, II, III producing Iphones\n- Machines I,II,III produce 2%,1%, and 3% defective iphones, respectively.\n- Out of total production, Machines $I$ makes $35\\%$ of all Iphones, $II - 25\\%$ , $III - 40\\%$ .\n- If one Iphone is selected at random from the factory,\n- Part (a): what is probability that one Iphone selected is defective?\n$$\n\\mathbb {P} (D) = P (I) \\mathbb {P} (D \\mid I) + P (I I) \\mathbb {P} (D \\mid I I) + P (I I I) \\mathbb {P} (D \\mid I I I)\n$$\n$$\n= (. 3 5) (. 0 2) + (. 2 5) (. 0 1) + (. 4) (. 0 3)\n$$\n$$\n= \\frac {2 1 5}{1 0 , 0 0 0}.\n$$\n\\- Part (b): What is the conditional prob that if an Iphone is defective, that it was produced by machine III?\n$$\n\\begin{array}{l} \\mathbb {P} (I I I \\mid D) = \\frac {\\mathbb {P} (I I I) \\mathbb {P} (D \\mid I I I)}{\\mathbb {P} (D)} \\\\ = \\frac {(. 4) (. 0 3)}{2 1 5 / 1 0 , 0 0 0} = \\frac {1 2 0}{2 1 5}. \\\\ \\end{array}\n$$\n\\- Example5: In a Multiple Choice Test, students either knows the answer or randomly guesses the answer to a question.\n\\- Let $m =$ number of choices in a question.\n- Let $p =$ the probability that the students knows the answer to a question.\n- Question: What is the probability that the student actually knew the answer, given that the student answers correctly.\n- Solution:\n- Let $K = \\{\\text{Knows the answer}\\}$ and $C = \\{\\text{Answer's correctly}\\}$ . Then\n$$\n\\mathbb {P} (K \\mid C) = \\frac {\\mathbb {P} (C \\mid K) \\mathbb {P} (K)}{\\mathbb {P} (C \\mid K) \\mathbb {P} (K) + \\mathbb {P} (C \\mid K ^ {c}) \\mathbb {P} (K ^ {c})}\n$$\n$$\n= \\frac {1 \\cdot p}{1 \\cdot p + \\frac {1}{m} (1 - p)} = \\frac {m p}{1 + (m - 1) p}.\n$$", + "path": "Notes.pdf/CHAPTER 4/Conditional Probability and Independence/4.2. Bayes's Formula", + "metadata": { + "length": 5559, + "summary": "The text explains the Law of Total Probability and Bayes' Theorem with examples. The Law of Total Probability computes P(E) by summing over mutually exclusive events: P(E) = Σ P(E|Fi)P(Fi). Bayes' Theorem finds conditional probabilities: P(Fj|E) = P(E|Fj)P(Fj) / Σ P(E|Fi)P(Fi). Examples include insurance accident risk, HIV test accuracy (98% accurate, 0.5% prevalence yields 19.8% positive predictive value), class grade probabilities, factory defect rates (machines I, II, III with 2%, 1%, 3% defects and 35%, 25%, 40% production), and multiple-choice test knowledge (probability student knew answer given correct response is mp/(1+(m-1)p)).", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Sometimes", + "easier", + "compute", + "probability", + "happened", + "Note", + "mathbb", + "cap", + "mid", + "This", + "formula", + "called", + "The", + "Law", + "Total", + "Probability", + "problem", + "describe", + "types", + "problems", + "section", + "Example1", + "Insurance", + "company", + "believes", + "accident", + "prone", + "person", + "year", + "Non", + "30", + "population", + "Part", + "Find", + "policy", + "holder", + "Let", + "text", + "Policy", + "IS", + "left", + "Suppose", + "policyholder", + "What", + "begin", + "array", + "frac", + "end", + "In", + "general", + "So", + "break", + "cases", + "If", + "ldots", + "mutually", + "exclusive", + "events", + "make", + "bigcup", + "sum", + "wanted", + "find", + "separate", + "conditional", + "event", + "Baye", + "Formula", + "denominator", + "Bayes", + "total", + "Example2", + "test", + "HIV", + "98", + "accurate", + "directions", + "0.5", + "positive", + "Question", + "tests", + "Solution", + "negative", + "Want", + "19.8", + "Example3", + "women", + "class", + "received", + "25", + "men", + "60", + "Given", + "chosen", + "random", + "students", + "receives", + "approx", + "General", + "Theorem", + "Here", + "possibilities", + "Example4", + "Factory", + "Machines", + "II", + "III", + "producing", + "Iphones", + "produce", + "2%", + "1%", + "3%", + "defective", + "iphones", + "Out", + "production", + "makes", + "35", + "40", + "Iphone", + "selected", + "factory", + "prob", + "produced", + "machine", + "Example5", + "Multiple", + "Choice", + "Test", + "answer", + "randomly", + "guesses", + "question", + "number", + "choices", + "student", + "knew", + "answers", + "correctly", + "Knows", + "Answer", + "Then", + "cdot" + ], + "keywords": [ + "Bayes' Theorem", + "Law of Total Probability", + "Conditional Probability" + ], + "connect_to": [] + } + }, + { + "chunk_id": "586cf404-6458-557c-b97b-809a1ff8f71d", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 5", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d55e64e1-a8ec-5779-835a-2cc7a4017f12", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 5/Random Variables", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "993e58bf-faaa-563c-946f-cee49e14ef01", + "type": "text", + "content": "\\- When we perform an experiment, we are interested in some function of the outcomes, instead of the actual outcome.\n\\- We want to attach for each outcome, a numerical value.\n\\- Definition: A random variable is a function $X: S \\to \\mathbb{R}$ or write $X: \\Omega \\to \\mathbb{R}$ . (Use capital letters to denote r.v)\n\\- We can think of $X$ as a numerical value that is random, like as if $X$ is a random number.\n• Example: Toss a coin\n\\- Let $X$ be 1 if heads and $X = 0$ if tails\n\\- Then $X(H) = 1$ and $X(T) = 0$ .\n\\- We can do calculus on real numbers but not on $\\Omega = S = \\{H, T\\}$ .\n• Example: Roll a die\n\\- Let $X$ denote the outcome, so $X = 1,2,3,4,5,6$ (its random)\n\\- That is $X(1) = 1, X(2) = 2, \\ldots$ .\n• Example: Roll a die, define\n$$\nY = \\left\\{ \\begin{array}{l l} 1 & \\text { outomce } = \\text { odd } \\\\ 0 & \\text { outomce } = \\text { even } \\end{array} \\right.\n$$\n\\- Can be thought of as\n$$\nY (s) = \\left\\{ \\begin{array}{l l} 1 & s = \\text {odd} \\\\ 0 & s = \\text {even} \\end{array} \\right..\n$$\n\\- A common question we'll have is \"What values can $X$ attain?\"\n\\- In other words, what is the range of $X$ ? Since $X: S \\to ?$\n• Example: Toss a coin 10 times\n\\- Let $X$ be the number of heads showing\n\\- What random values can $X$ be? 0, 1, 2, ..., 10.\n\\- Example: In general in $n$ trials, $X$ is the number of successes\n\\- Example1: Let $X$ be the amount of liability(damages) a driver incurs in a year.\n\\- $X:S\\to [0,\\infty)$ .\n• Example2: Toss a coin 3 times\n\\- Let $X$ be the number of heads that appear, so $X = 0,1,2,3$ .\n\\- In other words, $X:S\\to \\{0,1,2,3\\}$\n\\- We may assign probabilities to the different values of the random variable:\n$$\n\\begin{array}{l} \\mathbb {P} (X = 0) = \\mathbb {P} ((T, T, T)) = \\frac {1}{2 ^ {3}} = \\frac {1}{8} \\\\ \\mathbb {P} (X = 1) = \\mathbb {P} ((T, T, H), (T, H, T), (H, T, T)) = \\frac {3}{8} \\\\ \\mathbb {P} (X = 2) = \\mathbb {P} ((T, H, H), (H, H, T), (H, T, H)) = \\frac {3}{8} \\\\ \\mathbb {P} (X = 3) = \\mathbb {P} ((H, H, H)) = \\frac {1}{8}. \\\\ \\end{array}\n$$\n\\- Note that since $X$ must take the values of 0 through 3 then\n$$\n1 = \\mathbb {P} \\left(\\bigcup_ {i = 0} ^ {3} \\{X = i \\}\\right) = \\sum_ {i = 0} ^ {3} \\mathbb {P} (X = i),\n$$\nwhich makes sense from our previous calculation.", + "path": "Notes.pdf/CHAPTER 5/Random Variables/5.1. Random Variables", + "metadata": { + "length": 2267, + "summary": "The passage introduces the concept of a random variable as a function mapping outcomes of an experiment to real numbers, enabling calculus on outcomes. Examples include coin tosses (X=1 for heads, 0 for tails), die rolls (X equals the outcome), and defining Y based on odd/even outcomes. It explains determining the range of X, such as number of heads in 10 tosses (0-10) or liability amounts. A detailed example with 3 coin tosses shows assigning probabilities to X values (0-3), summing to 1.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "When", + "perform", + "experiment", + "interested", + "function", + "outcomes", + "actual", + "outcome", + "We", + "attach", + "numerical", + "Definition", + "random", + "variable", + "mathbb", + "write", + "Omega", + "Use", + "capital", + "letters", + "denote", + "number", + "Example", + "Toss", + "coin", + "Let", + "heads", + "tails", + "Then", + "calculus", + "real", + "numbers", + "Roll", + "die", + "That", + "ldots", + "define", + "left", + "begin", + "array", + "text", + "outomce", + "odd", + "end", + "Can", + "thought", + "common", + "question", + "ll", + "What", + "values", + "attain", + "In", + "words", + "range", + "Since", + "10", + "times", + "showing", + "general", + "trials", + "successes", + "Example1", + "amount", + "liability", + "damages", + "driver", + "incurs", + "year", + "infty", + "Example2", + "assign", + "probabilities", + "frac", + "Note", + "bigcup", + "sum", + "makes", + "sense", + "previous", + "calculation" + ], + "keywords": [ + "Random Variable", + "Probability", + "Outcomes" + ], + "connect_to": [] + } + }, + { + "chunk_id": "5b0fb9b1-364c-5e51-8bcd-504298364622", + "type": "text", + "content": "DEFINITION. A random variable that can take on at most countable number of possible values is said to be a discrete r.v.\nDEFINITION. For a discrete random variable, we can define the probability mass function (pmf), or the density function of $X$ by $p(x) = \\mathbb{P}(X = x)$ . Note that $p: \\mathbb{R} \\to [0,1]$ .\n- Note that $(X = x) = (\\omega \\in \\Omega \\mid X(\\omega) = x)$ is an abbreviation.\n- Let $X$ assume only the values $x_{1}, x_{2}, x_{3} \\ldots$\n\\- In other words, $X: S \\to \\{x_1, x_2, \\ldots\\}$\n\\- Properties of a pmf $p(x)$ :\n\\* Note that we must have $0 < p(x_i) \\leq 1$ for, $i = 1,2,\\ldots$ and $p(x) = 0$ for all other values of $x$ can't attain.\n\\* Also must have\n$$\n\\sum_ {i = 1} ^ {\\infty} p (x _ {i}) = 1.\n$$\n- We often draw bar graphs for discrete r.v.\n• Example: If we toss a coin\n\\- $X = 1$ if we have $H$ and $X = 0$ if we have $T$ .\n\\- Then draw a BAR graph\n$$\np _ {X} (x) = \\left\\{ \\begin{array}{l l} \\frac {1}{2} & x = 0 \\\\ \\frac {1}{2} & x = 1, \\\\ 0 & \\text {otherwise} \\end{array} \\right.\n$$\n- Oftentimes someone has already found the $pmf$ for you, and you can use to compute probabilities.\n- Example: The $pmf$ of $X$ is given by $p(i) = e^{-\\lambda \\frac{\\lambda^i}{i!}}$ for $i = 0,1,2,\\ldots$ where $\\lambda$ is a parameter (what is this?) that is any positive number\n\\- Part (a) What values can the random variable $X$ attain? In other words, what is the range of $X$ ?\n\\* Sol: By definition we have $\\mathbb{P}(X = 0) = p(0) = e^{-\\lambda}\\frac{\\lambda^0}{0!} = e^{-\\lambda}$\n\\- Part (b) Find $\\mathbb{P}(X = 0)$\n\\* Sol: By definition we have $\\mathbb{P}(X = 0) = p(0) = e^{-\\lambda}\\frac{\\lambda^0}{0!} = e^{-\\lambda}$\n\\- Part (c) Find $\\mathbb{P}(X > 2)$\n\\* Sol: Note that\n$$\n\\mathbb {P} (X > 2) = 1 - \\mathbb {P} (X \\leq 2)\n$$\n$$\n= 1 - \\mathbb {P} (X = 0) - \\mathbb {P} (X = 1) - \\mathbb {P} (X = 2)\n$$\n$$\n= 1 - p (0) - p (1) - p (2)\n$$\n$$\n= 1 - e ^ {- \\lambda} - \\lambda e ^ {- \\lambda} - \\frac {\\lambda^ {2} e ^ {- \\lambda}}{2}.\n$$", + "path": "Notes.pdf/CHAPTER 5/Random Variables/5.2. Discrete Random Variables", + "metadata": { + "length": 1979, + "summary": "The text defines a discrete random variable as one with at most a countable number of possible values. It introduces the probability mass function (pmf) p(x) = P(X=x), which maps real numbers to [0,1]. Key properties include p(x_i) > 0 for attainable values and sum of all p(x_i) = 1. Examples illustrate a coin toss with a bar graph and a Poisson pmf p(i)=e^{-λ} λ^i / i! for i=0,1,2,... The Poisson example computes P(X=0)=e^{-λ} and P(X>2)=1 - e^{-λ} - λ e^{-λ} - (λ^2 e^{-λ})/2.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "DEFINITION", + "random", + "variable", + "countable", + "number", + "values", + "discrete", + "For", + "define", + "probability", + "mass", + "function", + "pmf", + "density", + "mathbb", + "Note", + "omega", + "Omega", + "mid", + "abbreviation", + "Let", + "assume", + "ldots", + "In", + "words", + "Properties", + "leq", + "attain", + "Also", + "sum", + "infty", + "We", + "draw", + "bar", + "graphs", + "Example", + "If", + "toss", + "coin", + "Then", + "BAR", + "graph", + "left", + "begin", + "array", + "frac", + "text", + "end", + "Oftentimes", + "found", + "compute", + "probabilities", + "The", + "lambda", + "parameter", + "positive", + "Part", + "What", + "range", + "Sol", + "By", + "definition", + "Find" + ], + "keywords": [ + "discrete random variable", + "probability mass function", + "pmf properties" + ], + "connect_to": [] + } + }, + { + "chunk_id": "054e6bd1-a9e0-5539-a95b-bea0d622b364", + "type": "image", + "content": "\n\\- Solution: Summing up the probabilities up to that value of $x$ we get the following:

    0\n[images/image-11-Step Function Graph.jpg]\n", + "path": "images/image-11-Step Function Graph.jpg", + "metadata": { + "length": 133, + "summary": "image-11\n- Chart Type: Step function plotted on a Cartesian coordinate system\n- X-axis Range: -1 to 4 (major grid lines at intervals of 0.5)\n- Y-axis Range: -1 to 1 (major grid lines at intervals of 0.5)\n- Data Segments:\n - y = 0 for x < 0\n - y = 0.2 for 0 ≤ x < 1\n - y = 0.5 for 1 ≤ x < 2\n - y = 0.8 for 2 ≤ x < 3\n - y = 1 for x ≥ 3", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-11-Step Function Graph.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ef46425d-3167-5027-bfc6-60cbd43e058a", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
    xy
    -0.50.0
    0.00.2
    1.00.5
    2.00.9
    3.01.0
    3.51.0
    ", + "path": "tables/table-1 Data Table.html", + "metadata": { + "length": 510, + "summary": "table-2\nThe table shows six data points with x values from -0.5 to 3.5 and corresponding y values from 0.0 to 1.0, indicating a non-linear relationship.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-1 Data Table.html", + "keywords": [ + "values" + ], + "tokens": [] + } + }, + { + "chunk_id": "46291678-9c41-5a63-934c-531be5199a95", + "type": "text", + "content": "\\- One of the most important concepts in probability is that of expectation. If $X$ is a random variable that what is the average value of $X$ , that is what is the expected value of $X$ .\nDEFINITION. Let X have a pmf $p(x)$ . We define the expectation, or expected value of X to be\n$$\n\\mathbb {E} [ X ] = \\sum_ {x: p (x) > 0} x p (x).\n$$\n- Notation $\\mathbb{E}X$ , or $EX$ .\n- Example1: Let $X(H) = 0$ and $X(T) = 1$ . What is $\\mathbb{E}X$ ?\n$$\n\\mathbb {E} X = 0 \\cdot p (0) + 1 \\cdot p (1)\n$$\n$$\n= 0 \\frac {1}{2} + 1 \\cdot \\frac {1}{2} = \\frac {1}{2}.\n$$\n\\- Example2: Let $X$ be the outcome when we roll a fair die. What is $\\mathbb{E}X$ ?\n$$\n\\begin{array}{l} \\mathbb {E} X = 1 \\left(\\frac {1}{6}\\right) + 2 \\left(\\frac {1}{6}\\right) + \\dots + 6 \\frac {1}{6} \\\\ = \\frac {1}{6} (1 + 2 + 3 + 4 + 5 + 6) = \\frac {2 1}{6} = \\frac {7}{2} = 3. 5 \\\\ \\end{array}\n$$\n\\- Note that $X$ can never be 3.5, so expectation is to give you an idea, what an exact.\n\\- Recall infinite series: If $0 \\leq x < 1$ then a geometric series is\n$$\n\\begin{array}{l} \\sum_ {n = 0} ^ {\\infty} x ^ {n} = 1 + x + x ^ {2} + x ^ {3} + \\dots \\\\ = \\frac {1}{1 - x}. \\\\ \\end{array}\n$$\n\\- One thing you can do with series is differentiate them and integrate them: So if\n$$\n1 + x + x ^ {2} + x ^ {3} + \\dots + = \\frac {1}{1 - x}\n$$\nthen\n$$\n0 + 1 + 2 x + 3 x ^ {2} + \\dots + = \\frac {1}{(1 - x) ^ {2}}\n$$\n\\- Example3: Let $X$ be the number or tornadoes in Connecticut per year. Meaning that the random variable $X$ can be any number $X = 0,1,2,3,\\ldots$ . Suppose the state of Connecticut did some analysis and found out that\n$$\n\\mathbb {P} (X = i) = \\frac {1}{2 ^ {i + 1}}.\n$$\n- Question: What is $\\mathbb{E}X$ ? That is, what is the expected number of tornadoes per year in Connecticut.\n- Solution: Note that $X$ is infinite, but still countable, hence still discrete.\n\\- Note that\n$$\np (i) = \\left\\{ \\begin{array}{l l} \\frac {1}{2} & i = 0, \\\\ \\frac {1}{4} & i = 1, \\\\ \\frac {1}{8} & i = 2, \\\\ \\vdots & \\vdots \\\\ \\frac {1}{2 ^ {n + 1}} & i = n. \\end{array} \\right.\n$$\n\\- We have that\n$$\n\\begin{array}{l} \\mathbb {E} X = 0 \\cdot p (0) + 1 \\cdot p (1) + 2 \\cdot p (2) + \\dots \\\\ = 0 \\cdot \\frac {1}{2} + 1 \\frac {1}{2 ^ {2}} + 2 \\frac {1}{2 ^ {3}} + 3 \\frac {1}{2 ^ {4}} + \\dots \\\\ = \\frac {1}{2 ^ {2}} \\left(1 + 2 \\frac {1}{2} + 3 \\frac {1}{2 ^ {2}} + \\dots\\right) \\\\ = \\frac {1}{4} (1 + 2 x + 3 x ^ {2} + \\dots), \\text {with} x = \\frac {1}{2} \\\\ = \\frac {1}{4} \\frac {1}{(1 - x) ^ {2}} = \\frac {1}{4 (1 - \\frac {1}{2}) ^ {2}} = 1. \\\\ \\end{array}\n$$\n5.4. The C.D.F.\nDEFINITION. Define $F: R \\to [0,1]$ to be the function\n$$\nF (x) = \\mathbb {P} (X \\leq x), \\quad \\text { for any } - \\infty < x < \\infty\n$$\nto be the cumulative distribution function, or the distribution function of X., or CDF of X, or c.d.f\n\\- Note that when $X$ is discrete,\n$$\nF (x _ {0}) = \\mathbb {P} (X \\leq x _ {0}) = \\sum_ {x \\leq x _ {0}} p (x).\n$$\n- We sometimes use the notation $F_{X}(x)$ to highlight that $F_{X}$ is the CDF of the random variable $X$ .\n- Example: Suppose $X$ is equals to the number of heads in 3 coin flips. From Section 5.1, we calculated the p.m.f to be.:\n$$\np (0) = \\mathbb {P} (X = 0) = \\frac {1}{8}\n$$\n$$\np (1) = \\mathbb {P} (X = 1) = \\frac {3}{8}\n$$\n$$\np (2) = \\mathbb {P} (X = 2) = \\frac {3}{8}\n$$\n$$\np (3) = \\mathbb {P} (X = 3) = \\frac {1}{8}.\n$$\nQuestion: Find the c.d.f of $X$ . Plot the graph of the c.d.f.\n\\- Solution: Summing up the probabilities up to that value of $x$ we get the following:\n$$\nF (x) = \\left\\{ \\begin{array}{l l} 0 & - \\infty < x < 0 \\\\ \\frac {1}{8} & 0 \\leq x < 1 \\\\ \\frac {4}{8} & 1 \\leq x < 2 \\\\ \\frac {7}{8} & 2 \\leq x < 3 \\\\ 1 & 3 \\leq x < \\infty \\end{array} \\right..\n$$\n\\- The graph is given by\n\n\\- Solution: Summing up the probabilities up to that value of $x$ we get the following:
    0\n[images/image-11-Step Function Graph.jpg]\n\n
    \nline chart\n\n[tables/table-1 Data Table.html]\n\n
    \n- Note that this is a step function.\n- This function has jumps, and not continuous everywhere.\n- But it looks like it never decreases.\n• Properties of the CDF:\n- 1. $F$ is nondecreasing, that is\n$$\n* \\text { if } x < y \\text { then } F (x) \\leq F (y)\n$$\n\\- 2. $\\lim_{x\\to \\infty}F(x) = 1$\n\\- 3. $\\lim_{x\\to -\\infty}F(x) = 0$\n\\- 4. $F$ is right continuous. That are two ways that you can think of right continuity:\n\\* $\\lim_{x\\to 1^{+}}F(x_n) = F(x)$ , meaning \"the limit from the right equals where the function is defined\"\n\\* If $x_{n} \\downarrow x$ is a decreasing sequence then $\\lim_{n \\to \\infty} F(x_{n}) = F(x)$ .\n\\- We take these properties as facts, though one would normally have to prove these.\n\\- The following proposition does not have to be proved in class, and can be highlighted with the following example. But we include it here for completeness.\nPROPOSITION 3. Let $F_{X}(x)$ be the CDF for some random variable $X$ . Then the following holds:\n(a) For any $a \\in \\mathbb{R}$ , we have $\\mathbb{P}(X < a) = \\lim_{x \\to a^{-}} F_X(x)$\n(b) For any $a \\in \\mathbb{R}$ , we have $\\mathbb{P}(X = a) = F_X(a) - \\lim_{x \\to a^-} F_X(x)$\nPROOF. For part (a).\nWe first write\n$$\n\\begin{array}{l} (X < a) = \\bigcup_ {n = 1} ^ {\\infty} \\left(X \\leq a - \\frac {1}{n}\\right) \\\\ = (X \\leq a - 1) \\bigcup \\left[ \\bigcup_ {n = 1} ^ {\\infty} \\left(a - \\frac {1}{n} < X \\leq a - \\frac {1}{n + 1}\\right) \\right] \\\\ \\end{array}\n$$\nand since the events $E_{n} = \\left(a - \\frac{1}{n}\\leq X\\leq a - \\frac{1}{n + 1}\\right)$ are disjoint then we can use Axiom 3 so prove that\n$$\n\\begin{array}{l} \\mathbb {P} (X < a) = \\mathbb {P} (X \\leq a - 1) + \\sum_ {n = 1} ^ {\\infty} \\mathbb {P} \\left(a - \\frac {1}{n} < X \\leq a - \\frac {1}{n + 1}\\right) \\\\ = \\mathbb {P} (X \\leq a - 1) + \\lim _ {k \\rightarrow \\infty} \\sum_ {n = 1} ^ {k} \\left[ \\mathbb {P} \\left(X \\leq a - \\frac {1}{n + 1}\\right) - \\mathbb {P} \\left(X \\leq a - \\frac {1}{n}\\right)\\right] \\\\ = \\mathbb {P} (X \\leq a - 1) + \\lim _ {k \\rightarrow \\infty} \\left[ \\mathbb {P} \\left(X \\leq a - \\frac {1}{k + 1}\\right) - \\mathbb {P} (X \\leq a - 1) \\right], \\text { by telescoping } \\\\ = \\lim _ {k \\rightarrow \\infty} \\mathbb {P} \\left(X \\leq a - \\frac {1}{k + 1}\\right) + \\mathbb {P} (X \\leq a - 1) - \\mathbb {P} (X \\leq a - 1) \\\\ = \\lim _ {n \\to \\infty} F _ {X} \\left(a - \\frac {1}{n}\\right). \\\\ \\end{array}\n$$\nNow you can replace the sequence $a_{n} = a - \\frac{1}{n}$ with any sequence $a_{n}$ that is increasing towards $a$ , and we get the similar result,\n$$\n\\lim _ {n \\to \\infty} F _ {X} (a _ {n}) = \\mathbb {P} (X < a),\n$$\nsince this holds for all increasing sequences $a_{n}$ towards $a$ , then we've shown that\n$$\n\\lim _ {x \\to a ^ {-}} F _ {X} (x) = \\mathbb {P} \\left(X < a\\right).\n$$\nFor part (b). We use part (a) and get\n$$\n\\begin{array}{l} \\mathbb {P} (X = a) = \\mathbb {P} (X \\leq a) - \\mathbb {P} (X < a) \\\\ = F _ {X} (a) - \\lim _ {x \\to a ^ {-}} F _ {X} (x). \\\\ \\end{array}\n$$\n• Example: Let X have distribution\n$$\nF (x) = \\left\\{ \\begin{array}{l l} 0 & x < 0 \\\\ \\frac {x}{2} & 0 \\leq x < 1 \\\\ \\frac {2}{3} & 1 \\leq x < 2 \\\\ \\frac {1 1}{1 2} & 2 \\leq x < 3 \\\\ 1 & 3 \\leq x. \\end{array} \\right.\n$$\nGraph this and answer the following:\n\\- Part (a): Compute $\\mathbb{P}(2 < X\\leq 4)$ . We have that\n$$\n\\begin{array}{l} \\mathbb {P} (2 < X \\leq 4) = \\mathbb {P} (X \\leq 4) - \\mathbb {P} (X \\leq 2) \\\\ = F (4) - F (2) \\\\ = \\frac {1}{1 2}. \\\\ \\end{array}\n$$\n\\- Part (b): Compute $\\mathbb{P}(X < 3)$ .\n\\* We have that\n$$\n\\begin{array}{l} \\mathbb {P} (X < 3) = \\lim _ {n \\to \\infty} \\mathbb {P} \\left(X \\leq 3 - \\frac {1}{n}\\right) \\\\ = \\lim _ {x \\to 3 ^ {-}} F _ {X} (x) \\\\ = \\frac {1 1}{1 2} \\\\ \\end{array}\n$$\n\\- Part (c): Compute $\\mathbb{P}(X = 1)$ .\n\\* We have that\n$$\n\\begin{array}{l} \\mathbb {P} (X = 1) = \\mathbb {P} (X \\leq 1) - \\mathbb {P} (X < 1) \\\\ = F _ {X} (1) - \\lim _ {x \\rightarrow 1 ^ {-}} F _ {X} (x) \\\\ = \\frac {2}{3} - \\lim _ {x \\rightarrow 1} \\frac {x}{2} \\\\ = \\frac {2}{3} - \\frac {1}{2} = \\frac {1}{6}. \\\\ \\end{array}\n$$", + "path": "Notes.pdf/CHAPTER 5/Random Variables/5.3. Expected Value", + "metadata": { + "length": 7957, + "summary": "The passage introduces the concept of expectation (expected value) for discrete random variables, defined as the sum of outcomes weighted by their probabilities. Examples include a coin flip (E[X]=0.5), a fair die (E[X]=3.5), and a geometric distribution for tornado counts (E[X]=1). It then defines the cumulative distribution function (CDF) as F(x)=P(X≤x), which is nondecreasing, right-continuous, and has limits 0 and 1 at ±∞. Properties and examples show how to compute probabilities like P(X 0} x ^ {n} p (x).\n$$", + "path": "Notes.pdf/CHAPTER 5/Random Variables/5.6. Expectation of a Function of a Random Variable", + "metadata": { + "length": 3149, + "summary": "The text explains how to compute the expected value of a function of a discrete random variable. It provides an example where X takes values -1, 0, 1 with given probabilities, and Y = X^2. The expectation E[Y] is found to be 0.5, and it is noted that E[X^2] is not equal to (E[X])^2. A theorem states that for a discrete random variable X with probability mass function p(x_i), the expectation of g(X) is the sum of g(x_i)p(x_i). The proof groups values of g(X) and uses the definition of expectation. The text also defines the mean μ = E[X] and the nth moment E[X^n].", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Let", + "random", + "variable", + "Can", + "find", + "expected", + "things", + "sin", + "Example1", + "denote", + "mathbb", + "Find", + "Solution", + "Note", + "left", + "Thus", + "cdot", + "IMPORTANT", + "While", + "01", + "Not", + "equal", + "Since", + "neq", + "In", + "general", + "formula", + "function", + "That", + "fact", + "THEOREM", + "If", + "discrete", + "takes", + "values", + "ldots", + "respective", + "probability", + "mass", + "real", + "valued", + "sum", + "infty", + "NOTE", + "No", + "prove", + "lecture", + "But", + "completeness", + "PROOF", + "The", + "group", + "fashion", + "Using", + "definition", + "expectation", + "begin", + "array", + "star", + "end", + "Now", + "bigcup", + "plugging", + "back", + "needed", + "Remark", + "Revisted", + "underline", + "text", + "Sol", + "We", + "DEFINITION", + "call", + "mu", + "moment", + "quantity", + "EX", + "geq", + "called", + "nth", + "From", + "theorem", + "moments", + "calculated" + ], + "keywords": [ + "Expectation", + "Discrete Random Variable", + "Moment" + ], + "connect_to": [ + { + "target": "ffbbcfd1-7178-549c-ac23-8bafc67c1572", + "relation": "embeds", + "ref": "[images/image-13-Thus plugg.jpg]", + "position": { + "start": 2386, + "end": 2418 + } + } + ] + } + }, + { + "chunk_id": "55925058-6a9d-5c99-9cda-442261ff19f0", + "type": "image", + "content": "\nDEFINITION. We define # CHAPTER 60\n[images/image-14-DEFINITION.jpg]\n", + "path": "images/image-14-DEFINITION.jpg", + "metadata": { + "length": 69, + "summary": "image-14\nDEFINITION. We define # CHAPTER 60", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-14-DEFINITION.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "9c679369-ba01-5294-9f78-a47df727a7c4", + "type": "text", + "content": "- The variance of a r.v. is a measure of how spread out the values of $X$ are.\n- The expectation of a r.v. is quantity that help us differentiate different r.v.'s, but it doesn't tell us how spread out values are.\n\\- For example, take\n$$\nX = 0 \\text { with probability } 1\n$$\n$$\nY = \\left\\{ \\begin{array}{l l} - 1 & p = \\frac {1}{2} \\\\ 1 & p = \\frac {1}{2} \\end{array} \\right.\n$$\n$$\nZ = \\left\\{ \\begin{array}{l l} - 1 0 0 & p = \\frac {1}{2} \\\\ 1 0 0 & p = \\frac {1}{2} \\end{array} \\right..\n$$\n- What are the expected values? 0, 0, 0.\n- But there is much greater spread in $Z$ than $Y$ and $Y$ than $X$ . Thus expectation is not enough to detect spread, or variation.\nDEFINITION. If $X$ is a r.v with mean $\\mu = \\mathbb{E}X$ , then the variance of $X$ , denoted by $\\operatorname{Var}(X)$ , is defined by\n$$\n\\operatorname{Var} (X) = \\mathbb {E} \\left[ (X - \\mu) ^ {2} \\right].\n$$\n- Remark: $\\mathbb{E}c = c$ .\n- We prove an alternate formula for the variance. (The technique of using linearity is important here!!! Hint Hint)\n$$\n\\operatorname{Var} (X) = \\mathbb {E} \\left[ (X - \\mu) ^ {2} \\right]\n$$\n$$\n= \\mathbb {E} \\left[ X ^ {2} - 2 \\mu X + \\mu^ {2} \\right]\n$$\n$$\n= \\mathbb {E} [ X ^ {2} ] - 2 \\mu \\mathbb {E} [ X ] + \\mathbb {E} [ \\mu^ {2} ]\n$$\n$$\n= \\mathbb {E} \\left[ X ^ {2} \\right] - 2 \\mu^ {2} + \\mu^ {2}\n$$\n$$\n= \\mathbb {E} \\left[ X ^ {2} \\right] - \\mu^ {2}.\n$$\nTHEOREM. We have that\n$$\n\\operatorname{Var} (X) = \\mathbb {E} [ X ^ {2} ] - (\\mathbb {E} [ X ]) ^ {2}.\n$$\n\\- Example1: Calculate $\\operatorname{Var}(X)$ if $X$ represents the outcome when a fair die is rolled.\n\\- Solution: Previously we calculated that $\\mathbb{E}X = \\frac{7}{2}$ .\n\\- Thus we only need to calculate the second moment:\n$$\n\\mathbb {E} X ^ {2} = 1 ^ {2} \\left(\\frac {1}{6}\\right) + \\dots + 6 ^ {2} \\frac {1}{6}\n$$\n$$\n= \\frac {9 1}{6}.\n$$\n\\- Using our formula we have that\n$$\n\\begin{array}{l} \\operatorname{Var} (X) = \\mathbb {E} [ X ^ {2} ] - (\\mathbb {E} [ X ]) ^ {2} \\\\ = \\frac {9 1}{6} - \\left(\\frac {7}{2}\\right) ^ {2} \\\\ = \\frac {3 5}{1 2}. \\\\ \\end{array}\n$$\n• Here is a useful formula:\nPROPOSITION 7. For constants $a, b$ we have that $\\operatorname{Var}(aX + b) = a^2\\operatorname{Var}(X)$ .\nPROOF. We compute\n$$\n\\begin{array}{l} \\operatorname{Var} (a X + b) = \\mathbb {E} \\left[ (a X + b - \\mathbb {E} [ a X + b ]) ^ {2} \\right] \\\\ = \\mathbb {E} \\left[ (a X + b - a \\mu - b) ^ {2} \\right] \\\\ = \\mathbb {E} \\left[ a ^ {2} (X - \\mu) ^ {2} \\right] \\\\ = a ^ {2} \\mathbb {E} \\left[ (X - \\mu) ^ {2} \\right] \\\\ = a ^ {2} \\operatorname{Var} (X). \\\\ \\end{array}\n$$\nDEFINITION. We define\n$$\nS D (X) = \\sqrt {\\operatorname{Var} (X)}\n$$\nto be the standard deviation of $X$ .\n\nDEFINITION. We define # CHAPTER 60\n[images/image-14-DEFINITION.jpg]", + "path": "Notes.pdf/CHAPTER 5/Random Variables/5.7. Variance", + "metadata": { + "length": 2727, + "summary": "The passage introduces variance as a measure of spread for a random variable, contrasting it with expectation which does not capture variation. It defines variance as E[(X-μ)^2] and derives the alternate formula Var(X)=E[X^2]-(E[X])^2. An example calculates variance for a fair die roll. It also proves Var(aX+b)=a^2Var(X) and defines standard deviation as the square root of variance.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "The", + "variance", + "measure", + "spread", + "values", + "expectation", + "quantity", + "differentiate", + "doesn", + "For", + "text", + "probability", + "left", + "begin", + "array", + "frac", + "end", + "What", + "expected", + "But", + "greater", + "Thus", + "detect", + "variation", + "DEFINITION", + "If", + "mu", + "mathbb", + "denoted", + "operatorname", + "Var", + "defined", + "Remark", + "We", + "prove", + "alternate", + "formula", + "technique", + "linearity", + "important", + "Hint", + "THEOREM", + "Example1", + "Calculate", + "represents", + "outcome", + "fair", + "die", + "rolled", + "Solution", + "Previously", + "calculated", + "calculate", + "moment", + "dots", + "Using", + "Here", + "PROPOSITION", + "constants", + "aX", + "PROOF", + "compute", + "define", + "sqrt", + "standard", + "deviation", + "CHAPTER", + "60" + ], + "keywords": [ + "Variance", + "Standard Deviation", + "Expectation" + ], + "connect_to": [ + { + "target": "55925058-6a9d-5c99-9cda-442261ff19f0", + "relation": "embeds", + "ref": "[images/image-14-DEFINITION.jpg]", + "position": { + "start": 2695, + "end": 2727 + } + } + ] + } + }, + { + "chunk_id": "9b0575dd-5f13-504a-904d-2060a866e733", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 6", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f2bcc211-9639-5674-bbae-408a2edadef2", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 6/Some Discrete Distributions", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "dd6fa397-16fa-520b-b7ba-868eebf2c6f2", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 6/Some Discrete Distributions/6.1. Bernouli and Binomial Random Variables", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a6cc3b56-2872-5eda-acf9-caa177a9f41a", + "type": "text", + "content": "- Suppose that a trial or experiment takes place, whose outcome is either success or failure.\n- Let $X = 1$ when the outcome is a success and $X = 0$ if it is a failure.\n- The pmf of $X$ is given by\n$$\np (0) = \\mathbb {P} (X = 0) = 1 - p\n$$\n$$\np (1) = \\mathbb {P} (X = 1) = p\n$$\nwhere $0 \\leq p \\leq 1$ .\n\\- For this $X$ , $X$ is said to be a Bernoulli random variable with parameter $p$ ,\n\\* We wrtie this as $X \\sim \\text{Bernoulli}(p)$ ,\n\\* Properties:\n$\\cdot \\mathbb{E}X = p\\cdot 1 + (1 - p)\\cdot 0 = p$\n$\\cdot\\mathbb{E}X^{2}=1^{2}\\cdot p+0^{2}(1-p)=p.$\n- So $\\operatorname{Var} X = p - p^2 = p(1 - p)$ .", + "path": "Notes.pdf/CHAPTER 6/Some Discrete Distributions/6.1. Bernouli and Binomial Random Variables/• Bernoulli Distribution", + "metadata": { + "length": 608, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Suppose", + "trial", + "experiment", + "takes", + "place", + "outcome", + "success", + "failure", + "Let", + "The", + "pmf", + "mathbb", + "leq", + "For", + "Bernoulli", + "random", + "variable", + "parameter", + "We", + "wrtie", + "sim", + "text", + "Properties", + "cdot", + "So", + "operatorname", + "Var" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "25cd33db-722d-5a20-81ce-fa7e59125604", + "type": "text", + "content": "\\- We say $X$ has a binomial distribution with parameters $n$ and $p$ if\n$$\np _ {X} (k) = \\mathbb {P} \\left(X = k\\right) = \\binom{n}{k} p ^ {k} \\left(1 - p\\right) ^ {n - k}.\n$$\n\\- Interpret: $X =$ the number of successes in $n$ independent trials.\n\\* Let's take this as given.\n\\- We say $X \\sim \\text{Binomial}(n, p)$ or $X \\sim \\text{bin}(n, p)$ .", + "path": "Notes.pdf/CHAPTER 6/Some Discrete Distributions/6.1. Bernouli and Binomial Random Variables/• Binomial Distribution:", + "metadata": { + "length": 348, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "We", + "binomial", + "distribution", + "parameters", + "mathbb", + "left", + "binom", + "Interpret", + "number", + "successes", + "independent", + "trials", + "Let", + "sim", + "text", + "Binomial", + "bin" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "42769436-83b8-5863-9ec7-802bd486b7cb", + "type": "text", + "content": "\\- Check that probabilities sums to 1: Not really a property but more of a check that $X$ is indeed a random variable:\n\\* We need to check two things:\n(1) That $p_X(k) \\geq 0$ , and this is obvious from the formula\n(2) Need to check that $\\sum_{k=0}^{n} p_X(k) = 1$ .\n\\* First recall the Binomial Theorem: $\\sum_{k=0}^{n}\\binom{n}{k}x^k y^{n-k} = (x+y)^n$ .\n\\- Then\n$$\n\\sum_ {k = 0} ^ {n} p _ {X} (k) = \\sum_ {k = 0} ^ {n} \\binom{n}{k} p ^ {k} (1 - p) ^ {n - k} = (p + (1 - p)) ^ {n} = 1 ^ {n} = 1.\n$$\n\\- Mean: Easiest way to compute $\\mathbb{E}X$ is by recognizing that $X = Y_{1} + \\dots + Y_{n}$ where $Y_{i}$ are independent Bernoulli's.\n\\* Thus $EX = EY_{1} + \\cdots + EY_{n} = p + \\cdots + p = np.$\n\\* We can do this directly too, but this would involve proving that\n$$\n\\mathbb {E} X = \\sum_ {k = 0} ^ {n} k p (k) = n p,\n$$\nmeaning we would have to prove\n$$\n\\sum_ {k = 0} ^ {n} k \\binom{n}{k} p ^ {k} (1 - p) ^ {n - k} = n p.\n$$\n\\- Variance: We first compute the second moment. As before write $X = Y_{1} + \\cdots + Y_{n}$ where $Y_{i}$ are bernoulli's.\n$$\n\\begin{array}{l} \\mathbb {E} X ^ {2} = \\mathbb {E} \\left(Y _ {1} + \\dots + Y _ {n}\\right) ^ {2} \\\\ = \\sum_ {k = 1} ^ {n} \\mathbb {E} Y _ {k} ^ {2} + \\sum_ {i \\neq j} \\mathbb {E} \\left[ Y _ {i} Y _ {j} \\right] \\\\ = \\sum_ {k = 1} ^ {n} p + \\sum_ {i \\neq j} \\mathbb {E} \\left[ Y _ {i} Y _ {j} \\right] \\\\ = n p + \\sum_ {i \\neq j} \\mathbb {E} \\left[ Y _ {i} Y _ {j} \\right] \\\\ = (\\star) \\\\ \\end{array}\n$$\n\\* Now each term $\\mathbb{E}[Y_i Y_j]$ for fixed $i, j$ can be computed as =\n$$\n\\begin{array}{l} \\mathbb {E} \\left[ Y _ {i} Y _ {j} \\right] = 1 \\cdot \\mathbb {P} \\left(Y _ {i} Y _ {j} = 1\\right) + 0 \\cdot \\mathbb {P} \\left(Y _ {i} Y _ {j} = 0\\right) \\\\ = \\mathbb {P} ((Y _ {i} = 1) \\cap (Y _ {j} = 1)) \\\\ = \\mathbb {P} (Y _ {i} = 1) \\mathbb {P} (Y _ {j} = 1), \\text { by independence } \\\\ = p ^ {2}. \\\\ \\end{array}\n$$\n\\* Now there are a total of $n^2$ terms is $(Y_1 + \\dots + Y_n)^2$ , $n$ of which are of form $Y_k^2$ . Thus there are $n^2 - n$ terms of the form $Y_iY_j$ with $i \\neq j$ .\n\\* Hence using $(\\star)$ we have $\\mathbb{E}X^2 = np + (n^2 - n)p^2$ .\n\\* Thus\n$$\n\\operatorname{Var} X = \\mathbb {E} X ^ {2} - (\\mathbb {E} X) ^ {2} = n p + (n ^ {2} - n) p ^ {2} - (n p) ^ {2} = n p (1 - p).\n$$\n\\- Sumarize: $\\mathbb{E}X = np$ and $\\operatorname{Var}X = np(1 - p)$ .\n\\- $\\underline{\\text{Moments}}$ : We can also prove $\\mathbb{E}X^k = np\\mathbb{E}\\left[(Y + 1)^{k - 1}\\right]$ .\n\\- Calculator(TI-84):\n\\- 2ndDistri>binomialpdf(n,p,x)=P(X=x).\n\\- same with cdf.\n\\- Example1: A company prices its hurricane insurance using the following assumptions:\n\\- (i) In any calendar year, there can be at most one hurricane.\n\\- (ii) In any calendar year, the probability of a hurricane is 0.05.\n\\- (iii) The numbers of hurricanes in different calendar years are mutually independent. Using the company's assumptions, calculate the probability that there are fewer than 3 hurricanes in a 20-year period\n\\- Solution: We have that $X \\sim \\text{bin}(20, .05)$ then\n$$\n\\begin{array}{l} \\mathbb {P} (X < 3) = \\mathbb {P} (X \\leq 2) \\\\ = \\binom{2 0}{0} (. 0 5) ^ {0} (. 9 5) ^ {2 0} + \\binom{2 0}{1} (. 0 5) ^ {1} (. 9 5) ^ {1 9} + \\binom{2 0}{2} (. 0 5) ^ {2} (. 9 5) ^ {1 2} \\\\ = . 9 2 4 5. \\\\ \\end{array}\n$$\n\\- Example2: Phan has a .6 probability of making a free throw. Suppose each free throw is independent of the other. If he attempts 10 free throws, what is the probability that he makes at least 2 of them?\n\\- Solution: Let $X \\sim \\text{bin}(10, .6)$ then\n$$\n\\begin{array}{l} \\mathbb {P} (X \\geq 2) = 1 - \\mathbb {P} (X = 0) - \\mathbb {P} (X = 1) \\\\ = 1 - \\binom{1 0}{0} (. 6) ^ {0} (. 4) ^ {1 0} - \\binom{1 0}{1} (. 6) ^ {1} (. 4) ^ {9} \\\\ = . 9 9 8. \\\\ \\end{array}\n$$", + "path": "Notes.pdf/CHAPTER 6/Some Discrete Distributions/6.1. Bernouli and Binomial Random Variables/• Properties of the Binomial", + "metadata": { + "length": 3737, + "summary": "The passage covers the binomial distribution, including checking that probabilities sum to 1 using the binomial theorem, deriving the mean as np by summing independent Bernoulli variables, and computing the variance as np(1-p) via the second moment. It also mentions moments and provides TI-84 calculator functions. Two examples are given: one calculating the probability of fewer than 3 hurricanes in 20 years with p=0.05, and another finding the probability of making at least 2 free throws out of 10 with p=0.6.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Check", + "probabilities", + "sums", + "Not", + "property", + "check", + "random", + "variable", + "We", + "things", + "That", + "geq", + "obvious", + "formula", + "Need", + "sum", + "First", + "recall", + "Binomial", + "Theorem", + "binom", + "Then", + "Mean", + "Easiest", + "compute", + "mathbb", + "recognizing", + "dots", + "independent", + "Bernoulli", + "Thus", + "EX", + "EY", + "cdots", + "np", + "directly", + "involve", + "proving", + "meaning", + "prove", + "Variance", + "moment", + "As", + "write", + "bernoulli", + "begin", + "array", + "left", + "neq", + "star", + "end", + "Now", + "term", + "fixed", + "computed", + "cdot", + "cap", + "text", + "independence", + "total", + "terms", + "form", + "iY", + "Hence", + "operatorname", + "Var", + "Sumarize", + "underline", + "Moments", + "Calculator", + "TI", + "84", + "2ndDistri", + "binomialpdf", + "cdf", + "Example1", + "company", + "prices", + "hurricane", + "insurance", + "assumptions", + "In", + "calendar", + "year", + "ii", + "probability", + "0.05", + "iii", + "The", + "numbers", + "hurricanes", + "years", + "mutually", + "Using", + "calculate", + "fewer", + "20", + "period", + "Solution", + "sim", + "bin", + "05", + "leq", + "Example2", + "Phan", + "making", + "free", + "throw", + "Suppose", + "If", + "attempts", + "10", + "throws", + "makes", + "Let" + ], + "keywords": [ + "Binomial", + "Mean", + "Variance" + ], + "connect_to": [] + } + }, + { + "chunk_id": "2a02afef-b752-5e43-9048-df4d031e2a63", + "type": "image", + "content": "\n$$ \\- Summary of Theorem: This theorem says that suppose $n$ is large and $p$ is small, Thus0\n[images/image-15-$$ _- Summ.jpg]\n", + "path": "images/image-15-$$ _- Summ.jpg", + "metadata": { + "length": 128, + "summary": "image-15\n$$ \\- Summary of Theorem: This theorem says that suppose $n$ is large and $p$ is small, Thus0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-15-$$ _- Summ.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "4ac0c4ac-21e4-54ca-9cdc-8f83c81bc9f1", + "type": "text", + "content": "\\- We say that $X = 0,1,2,\\ldots$ is Poisson with parameter $\\lambda > 0$ if\n$$\np _ {X} (i) = \\mathbb {P} (X = i) = e ^ {- \\lambda} \\frac {\\lambda^ {i}}{i !} \\text { for } i = 0, 1, 2, 3, \\dots .\n$$\n\\- Or $X \\sim \\mathrm{Poisson}(\\lambda)$ .\n\\- In general Poisson random variables are of the following form\n\\- Suppose success happens $\\lambda$ times on average in a given period (per year, per month etc). Then $X =$ number of times success happens in that given period.\n\\- Possion is like binomial, expect, $X$ is infinitely countable!\n• Examples that obey Poisson R.V\n\\- 1. The number of misprints on a page ogf a book\n\\- 2. # of people in community that survive to age 100\n\\- 3. # of telephone numbers that are dialed in a day.\n\\- 4. # of customers entering post office on a day.\n\\- Calc2: Recall that $\\sum_{n=0}^{\\infty} \\frac{x^n}{n!} = e^x$ .\n\\- Properties of Poisson: Let $X \\sim \\text{Poisson}(\\lambda)$\n\\- First we check that $p_X(i)$ is indeed a pmf: First it is obvious that $p_X(i) \\geq 0$ since $\\lambda > 0$ . We to need to check that all the probabilities add up to one:\n$$\n\\sum_ {i = 0} ^ {\\infty} p _ {X} (i) = \\sum_ {i = 0} ^ {\\infty} e ^ {- \\lambda} \\frac {\\lambda^ {i}}{i !} = e ^ {- \\lambda} \\sum_ {i = 0} ^ {\\infty} \\frac {\\lambda^ {i}}{i !} = e ^ {- \\lambda} e ^ {\\lambda} = 1.\n$$\n\\- Mean: We have\n$$\n\\mathbb {E} X = \\sum_ {i = 0} ^ {\\infty} i e ^ {- \\lambda} \\frac {\\lambda^ {i}}{i !} = e ^ {- \\lambda} \\lambda \\sum_ {i = 1} ^ {\\infty} \\frac {\\lambda^ {i - 1}}{(i - 1) !}\n$$\n$$\n= e ^ {- \\lambda} \\lambda e ^ {\\lambda} = \\lambda .\n$$\n\\- Variance: We first have\n$$\n\\begin{array}{l} \\mathbb {E} X ^ {2} = \\sum_ {i = 0} ^ {\\infty} i ^ {2} \\frac {e ^ {- \\lambda} \\lambda^ {i}}{i !} \\\\ = \\lambda \\sum_ {i = 0} ^ {\\infty} i \\frac {e ^ {- \\lambda} \\lambda^ {i - 1}}{(i - 1) !} \\\\ = \\lambda \\sum_ {j = 0} ^ {\\infty} (j + 1) \\frac {e ^ {- \\lambda} \\lambda^ {j}}{j !}, \\text { let } j = i - 1 \\\\ = \\lambda \\left[ \\sum_ {j = 0} ^ {\\infty} j \\frac {e ^ {- \\lambda} \\lambda^ {j}}{j !} + \\sum_ {j = 0} ^ {\\infty} \\frac {e ^ {- \\lambda} \\lambda^ {j}}{j !} \\right] \\\\ = \\lambda [ \\lambda + e ^ {- \\lambda} e ^ {\\lambda} ] \\\\ = \\lambda (\\lambda + 1). \\\\ \\end{array}\n$$\nThus\n$$\n\\operatorname{Var} X = \\lambda (\\lambda + 1) - \\lambda^ {2} = \\lambda .\n$$\n\\- Example1: Suppose on average there are 5 homicides per month in Hartford, CT. What is the probability there will be at most 1 in a certain month?\n\\- Answer: If $X$ is the number of homicides, we are given that $EX = 5$ . Since the expectation for a Poisson is $\\lambda = 5$ . Therefore $P(X = 0) + P(X = 1) = e^{-5} + 5e^{-5}$ .\n\\- Example2: Suppose on average there is one large earthquake per year in Mexico. What's the probability that next year there will be exactly 2 large earthquakes?\n\\- $\\underline{\\text{Answer:}}\\lambda = EX = 1$ , so $P(X = 2) = \\frac{e^{-1}}{2}$ .\n\\- Example3: Phan receives texts on the average of two every 3 minutes. Assume Poisson.\n– Question: What is the probability of five or more texts arriving in a 9-minute period.\n\\- Answer: Let $X$ number of calls in a 9-minute period. Let $n =$ number of periods, $\\lambda_1 = 2$ . Thus $\\lambda = 3 \\cdot 2 = 6$ . Thus\n$$\n\\begin{array}{l} \\mathbb {P} (X \\geq 5) = 1 - \\mathbb {P} (X \\leq 4) \\\\ = 1 - \\sum_ {n = 0} ^ {4} \\frac {e ^ {- 6} 6 ^ {n}}{n !} \\\\ = 1 -. 2 8 5 = . 7 1 5. \\\\ \\end{array}\n$$\n\\- Important: Poisson is similar to Binomial in the following way\n\\- FACT: Poisson approximates $Bin(n,p)$ when $n$ is large and $p$ is small enough so that $np$ is of moderate size.\nTHEOREM 8. If $X_{n}$ is binomial with parameters $n$ and $p_n$ and $np_n \\to \\lambda$ , then\n$$\n\\mathbb {P} \\left(X _ {n} = i\\right)\\rightarrow \\mathbb {P} (Y = i)\n$$\nwhere $Y \\sim \\text{Poisson}(\\lambda)$ .\nPROOF. See class textbook.\n\n$$ \\- Summary of Theorem: This theorem says that suppose $n$ is large and $p$ is small, Thus0\n[images/image-15-$$ _- Summ.jpg]\n\n\\- Summary of Theorem: This theorem says that suppose $n$ is large and $p$ is small, Thus\n- If $X \\sim Bin(n, p)$ then we approximate $X$ with a possession by letting let $\\lambda = np$ so that\n$$\n\\mathbb {P} (X = i) \\approx e ^ {- n p} \\frac {(n p) ^ {i}}{i !}.\n$$\n\\- When can we assume $X$ is Poisson: Another consequence of this theorem says that when $Y =$ \"the number of successes in a given period\". And if the number possible of trials $n$ is large, and if the probability $p$ of success is small, then $Y$ can be treated as a Poisson random variable.\nNOTE:\n\\- (1)Why is number of misprints on a page will be approximately Poisson with $\\lambda = np$\n\\* Let $X =$ number of misprints on a page of a book.\n\\* Since prob of error, say $p = .01$ is usually small, and number of letters on a page is usually large, say $n = 1000$ . Then the average is $\\lambda = np$ .\n\\* Then because $p$ is small and $n$ is large, then $X$ can be approximated by a Poisson.\n\\- (2) Let $X$ number of accidents in a year\n\\* $X$ is Poisson because the probability of an accident $p$ in a given period is usually small and while the number $n$ of times someone drives in a given period is high.\n• Example: Here is an example showing this.\n\\- If $X$ is number of times you get heads on a biased coin where $\\mathbb{P}(H) = \\frac{1}{100}$ . Suppose you you toss 1000 times. Then $np = 10$\n$$\n\\mathbb {P} (X = 5) \\approx e ^ {- 1 0} \\frac {1 0 ^ {5}}{5 !} = . 0 3 7 8\n$$\nwhile the actual value is\n$$\n\\begin{array}{l} \\mathbb {P} \\left(X = 5\\right) = \\binom{1 0 0 0}{5} (. 0 1) ^ {5} (. 9 9) ^ {9 9 5} \\\\ = \\frac {1 0 0 0 !}{9 9 5 ! 5 !} (. 0 1) ^ {5} (. 9 9) ^ {9 9 5} \\\\ = . 0 3 7 5. \\\\ \\end{array}\n$$", + "path": "Notes.pdf/CHAPTER 6/Some Discrete Distributions/6.2. The Poisson Distribution", + "metadata": { + "length": 5570, + "summary": "The Poisson distribution models the number of events occurring in a fixed interval, given a constant average rate λ. Its probability mass function is P(X=i)=e^{-λ}λ^i/i!. The mean and variance both equal λ. It approximates the binomial distribution when n is large and p is small, with λ=np. Examples include misprints per page, homicides per month, and earthquake occurrences.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "We", + "ldots", + "Poisson", + "parameter", + "lambda", + "mathbb", + "frac", + "text", + "dots", + "Or", + "sim", + "mathrm", + "In", + "general", + "random", + "variables", + "form", + "Suppose", + "success", + "times", + "average", + "period", + "year", + "month", + "Then", + "number", + "Possion", + "binomial", + "expect", + "infinitely", + "countable", + "Examples", + "obey", + "The", + "misprints", + "page", + "ogf", + "book", + "people", + "community", + "survive", + "age", + "100", + "telephone", + "numbers", + "dialed", + "day", + "customers", + "entering", + "post", + "office", + "Calc2", + "Recall", + "sum", + "infty", + "Properties", + "Let", + "First", + "check", + "pmf", + "obvious", + "geq", + "probabilities", + "add", + "Mean", + "Variance", + "begin", + "array", + "left", + "end", + "Thus", + "operatorname", + "Var", + "Example1", + "homicides", + "Hartford", + "CT", + "What", + "probability", + "Answer", + "If", + "EX", + "Since", + "expectation", + "Therefore", + "5e", + "Example2", + "large", + "earthquake", + "Mexico", + "earthquakes", + "underline", + "Example3", + "Phan", + "receives", + "texts", + "minutes", + "Assume", + "Question", + "arriving", + "minute", + "calls", + "periods", + "cdot", + "leq", + "Important", + "similar", + "Binomial", + "FACT", + "approximates", + "Bin", + "small", + "np", + "moderate", + "size", + "THEOREM", + "parameters", + "rightarrow", + "PROOF", + "See", + "class", + "textbook", + "Summary", + "Theorem", + "This", + "theorem", + "suppose", + "Thus0", + "approximate", + "possession", + "letting", + "approx", + "When", + "assume", + "Another", + "consequence", + "successes", + "And", + "trials", + "treated", + "variable", + "NOTE", + "Why", + "approximately", + "prob", + "error", + "01", + "letters", + "1000", + "approximated", + "accidents", + "accident", + "drives", + "high", + "Example", + "Here", + "showing", + "heads", + "biased", + "coin", + "toss", + "10", + "actual", + "binom" + ], + "keywords": [ + "Poisson", + "mean", + "variance" + ], + "connect_to": [ + { + "target": "2a02afef-b752-5e43-9048-df4d031e2a63", + "relation": "embeds", + "ref": "[images/image-15-$$ _- Summ.jpg]", + "position": { + "start": 3853, + "end": 3885 + } + } + ] + } + }, + { + "chunk_id": "cc9e8a27-8286-5afd-9539-797b90b01d86", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 6/Some Discrete Distributions/6.3. Other Discrete Distributions", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a8928d64-8f39-55f8-8dd1-a4f55127f548", + "type": "text", + "content": "\\- We say $X$ is uniform, and write this as $X \\sim \\text{uniform}(n)$ , if $X \\in \\{1, 2, \\ldots, n\\}$ and\n$$\np _ {X} (i) = \\mathbb {P} (X = i) = \\frac {1}{n} \\text { for } i = 1, 2, \\dots , n.\n$$\n\\- $\\underline{\\text{Exercise:}}$ $\\mathbb{E}X = \\sum_{i=1}^{n} i\\frac{1}{n} = \\frac{1}{n}\\sum_{i=1}^{n} i = \\frac{1}{n}\\frac{n(n-1)}{2} = \\frac{n-1}{2}$ and find $\\operatorname{Var}X$ .", + "path": "Notes.pdf/CHAPTER 6/Some Discrete Distributions/6.3. Other Discrete Distributions/• Uniform Distribution:", + "metadata": { + "length": 384, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "We", + "uniform", + "write", + "sim", + "text", + "ldots", + "mathbb", + "frac", + "dots", + "underline", + "Exercise", + "sum", + "find", + "operatorname", + "Var" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9d0b3bc7-4b1d-5ba1-a99a-bb62a618b7f4", + "type": "text", + "content": "- Experiment: Suppose that independent trials are held until success occurs. Trials are stopped once success happens. Let $p$ be the probability of having a success in each trial.\n- Let $X =$ \"number of trials required until first success occurs\". Thus $X \\in \\{1,2,3,4,\\ldots\\}$ Here we have\n$$\np _ {X} (i) = \\mathbb {P} (X = i) = (1 - p) ^ {i - 1} p \\text {for} i = 1, 2, 3, 4 \\dots .\n$$\n- We say $X \\sim \\operatorname{geometric}(p)$ .\n- Properties:\n\\* We first double check is indeed a discrete random variable: This follows from what we know about geometric series:\n$$\n\\sum_ {i = 1} ^ {\\infty} \\mathbb {P} (X = i) = \\sum_ {i = 1} ^ {\\infty} (1 - p) ^ {i - 1} p = \\frac {p}{1 - (1 - p)} = 1.\n$$\n\\* Mean: Recall that by differentiation of the geometric series, we came up with the formula $\\sum_{n=0}^{\\infty} nx^{n-1} = \\frac{1}{(1-x)^2}$ , so that\n$$\n\\begin{array}{l} \\mathbb {E} X = \\sum_ {i = 1} ^ {\\infty} i \\mathbb {P} (X = i) \\\\ = \\sum_ {i = 1} ^ {\\infty} i (1 - p) ^ {i - 1} p \\\\ = \\frac {p}{(1 - (1 - p)) ^ {2}} = \\frac {1}{p}. \\\\ \\end{array}\n$$\n\\* Variance:(Leave as Exercise for student) Note that\n$$\n\\mathbb {E} X ^ {2} = \\sum_ {i = 1} ^ {\\infty} i ^ {2} (1 - p) ^ {i - 1} p. (\\star)\n$$\nThus we can differentiate $\\sum_{n=1}^{\\infty} nx^{n-1} = \\frac{1}{(1-x)^2}$ again to get $\\sum_{n=2}^{\\infty} n(n-1)x^{n-2} = \\frac{2}{(1-x)^3}$ .\n\\* From this we will attempt to get $EX^{2}$ in ( $\\star$ ) by splitting the sum up:\n$$\n\\sum_ {n = 2} ^ {\\infty} n (n - 1) (1 - p) ^ {n - 2} = \\frac {2}{(1 - (1 - p)) ^ {3}} = \\frac {2}{p ^ {3}},\n$$\n$$\n\\sum_ {n = 2} ^ {\\infty} n (n - 1) (1 - p) ^ {n - 2} p = \\frac {2}{p ^ {2}}, \\text { now split },\n$$\n$$\n\\sum_ {n = 2} ^ {\\infty} n ^ {2} (1 - p) ^ {n - 2} p = \\frac {2}{p ^ {2}} + \\sum_ {n = 2} ^ {\\infty} n (1 - p) ^ {n - 2} p\n$$\n$$\n(1 - p) ^ {- 1} \\sum_ {n = 1} ^ {\\infty} n ^ {2} (1 - p) ^ {n - 1} p = \\frac {2}{p ^ {2}} + \\sum_ {n = 2} ^ {\\infty} n (1 - p) ^ {n - 2} p + (1 - p) ^ {- 1} p\n$$\n$$\n\\begin{array}{l} (1 - p) ^ {- 1} \\mathbb {E} X ^ {2} = \\frac {2}{p ^ {2}} + (1 - p) ^ {- 1} \\sum_ {n = 1} ^ {\\infty} n (1 - p) ^ {n - 1} p \\\\ = \\frac {2}{p ^ {2}} + (1 - p) ^ {- 1} \\frac {1}{p} \\\\ \\end{array}\n$$\nThus\n$$\n\\begin{array}{l} \\mathbb {E} X ^ {2} = \\frac {2 (1 - p)}{p ^ {2}} + \\frac {1}{p} \\\\ = \\frac {2 - 2 p + p}{p ^ {2}} = \\frac {2 - p}{p ^ {2}} \\\\ \\end{array}\n$$\n\\* So Thus\n$$\n\\begin{array}{l} \\operatorname{Var} X = \\mathbb {E} X ^ {2} - (\\mathbb {E} X) ^ {2} = \\frac {2 - p}{p ^ {2}} - \\frac {1}{p ^ {2}} \\\\ = \\frac {(1 - p)}{p ^ {2}} \\\\ \\end{array}\n$$\n\\- Example1: An urn contains 10 white balls and 15 black balls. Balls are randomly selected, one at a time, until a black one is obtained. If we assume that each ball selected is replaced before the next one is drawn, what is the probability.\n\\- Part (a): Exactly 6 draws are needed?\n\\* X = number of draws needed to select a black ball, the probability of success is\n$$\np = \\frac {1 5}{1 0 + 1 5} = \\frac {1 5}{2 5} = . 6.\n$$\n\\* Thus\n$$\n\\mathbb {P} (X = 6) = (. 4) ^ {6 - 1} (. 6) = . 0 0 6 1 4 4\n$$\n\\- Part (a): What is the expected number of draws in this game?\n\\* Since $X \\sim$ geometric(.6) then\n$$\n\\mathbb {E} X = \\frac {1}{p} = \\frac {1 0}{6} = 1. \\bar {6}\n$$\n\\- Part (c)(Extra Problem to be done at home) Find exactly that probability at least $k$ draws are needed?\n\\* We have that\n$$\n\\begin{array}{l} \\mathbb {P} (X \\geq k) = \\sum_ {n = k} ^ {\\infty} \\mathbb {P} (X = k) \\\\ = \\sum_ {n = k} ^ {\\infty} (. 4) ^ {n - 1} (. 6) \\\\ = (. 6) (. 4) ^ {- 1} \\sum_ {n = k} ^ {\\infty} (. 4) ^ {n} \\\\ = (. 6) (. 4) ^ {- 1} (. 4) ^ {k} \\sum_ {n = 0} ^ {\\infty} (. 4) ^ {n} \\\\ = (. 6) (. 4) ^ {k - 1} \\frac {1}{1 - . 4} \\\\ = (. 4) ^ {k - 1}. \\\\ \\end{array}\n$$\n\\- Note: This could have been done for a general $p$ . Thus\n$$\n\\mathbb {P} (X \\geq k) = (1 - p) ^ {k - 1}.\n$$\n\\- Negative Binomial(Need to know for Actuarial Exam):\n\\- Experiment: Suppose that independent trials are held with probability $p$ of having a success. The trials are perfomed until a total of $r$ successes are accumulated.\n\\* Let $X$ equal the number of trials required to obtain $r$ successes. Here we have\n$$\n\\mathbb {P} \\left(X = n\\right) = \\binom{n - 1}{r - 1} p ^ {r} \\left(1 - p\\right) ^ {n - r} \\text { for } n = r, r + 1, \\ldots .\n$$\n- We say $X \\sim \\text{NegativeBinomial}(r, p)$ .\n- Properties:\n\\* This is a probability mass function. Can check that $\\sum_{n=r}^{\\infty}\\mathbb{P}(X=n)=1$ .\n\\* Mean:\n$$\n\\mathbb {E} X = \\frac {r}{p}.\n$$\n\\* Variance:\n$$\n\\operatorname{Var} (X) = \\frac {r (1 - p)}{p ^ {2}}.\n$$\n\\- Note that $\\operatorname{Geometric}(p) = \\text{NegativeBinomial}(1, p)$ .\n\\- Example: Find the expected value of the number of times one must throw a die until the outcome 1 has occurred 4 times.\n\\- Solution: $X \\sim \\text{NegativeBinomial}\\left(4, \\frac{1}{6}\\right)$ . So\n$$\n\\mathbb {E} X = \\frac {4}{\\frac {1}{6}} = 2 4.\n$$\n• Hypergeometric Distribution(Need to know for Actuarial Exam):\n\\- Experiment: Suppose that a sample of size $n$ is to be chosen randomly (without replacement) from an urn containing $N$ balls, of which $m$ are white and $N - m$ are black.\n\\* Let $X$ equal the number of white balls selected. Then\n$$\n\\mathbb {P} \\left(X = i\\right) = \\frac {\\binom{m}{i} \\binom{N - m}{n - i}}{\\binom{N}{n}} \\text { for } n = 0, 1, \\ldots , n.\n$$\n\\- We say $X \\sim \\text{Hypergeometric}(n, N, m)$ .\n\\- Properties:\n\\* Mean:\n$$\n\\mathbb {E} X = \\frac {n m}{N}.\n$$\n\\* Variance:\n$$\n\\operatorname{Var} (X) = n \\frac {m}{N} \\left(1 - \\frac {m}{N}\\right) \\left(1 - \\frac {n - 1}{N - 1}\\right).\n$$", + "path": "Notes.pdf/CHAPTER 6/Some Discrete Distributions/6.3. Other Discrete Distributions/- Geometric Distribution:", + "metadata": { + "length": 5515, + "summary": "The passage covers three discrete probability distributions: Geometric, Negative Binomial, and Hypergeometric. For the Geometric distribution (number of trials until first success), it provides the probability mass function, mean (1/p), and variance ((1-p)/p^2), with an example involving drawing balls from an urn. The Negative Binomial distribution (trials until r successes) is introduced with its mean (r/p) and variance (r(1-p)/p^2), noting that Geometric is a special case with r=1. The Hypergeometric distribution (sampling without replacement) is defined with its mean (nm/N) and variance formula. Examples include calculating probabilities and expected values for specific scenarios.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Experiment", + "Suppose", + "independent", + "trials", + "held", + "success", + "occurs", + "Trials", + "stopped", + "Let", + "probability", + "trial", + "number", + "required", + "Thus", + "ldots", + "Here", + "mathbb", + "text", + "dots", + "We", + "sim", + "operatorname", + "geometric", + "Properties", + "double", + "check", + "discrete", + "random", + "variable", + "This", + "series", + "sum", + "infty", + "frac", + "Mean", + "Recall", + "differentiation", + "formula", + "nx", + "begin", + "array", + "end", + "Variance", + "Leave", + "Exercise", + "student", + "Note", + "star", + "differentiate", + "From", + "attempt", + "EX", + "splitting", + "split", + "So", + "Var", + "Example1", + "An", + "urn", + "10", + "white", + "balls", + "15", + "black", + "Balls", + "randomly", + "selected", + "time", + "obtained", + "If", + "assume", + "ball", + "replaced", + "drawn", + "Part", + "Exactly", + "draws", + "needed", + "select", + "What", + "expected", + "game", + "Since", + "bar", + "Extra", + "Problem", + "home", + "Find", + "geq", + "general", + "Negative", + "Binomial", + "Need", + "Actuarial", + "Exam", + "The", + "perfomed", + "total", + "successes", + "accumulated", + "equal", + "obtain", + "left", + "binom", + "NegativeBinomial", + "mass", + "function", + "Can", + "Geometric", + "Example", + "times", + "throw", + "die", + "outcome", + "occurred", + "Solution", + "Hypergeometric", + "Distribution", + "sample", + "size", + "chosen", + "replacement", + "Then" + ], + "keywords": [ + "Geometric Distribution", + "Negative Binomial", + "Hypergeometric" + ], + "connect_to": [] + } + }, + { + "chunk_id": "82f234b4-a734-54c1-8efc-d9bfc79d351f", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 7", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "cf49bf4d-a522-5943-bb72-a1fbaa007e10", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 7/Continuous Random Variables", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c0ec30dc-b519-5f65-935a-a22278e58874", + "type": "text", + "content": "DEFINITION. A random variable $X$ is said to have a continuous distribution if there exists a non-negative function $f$ such that\n$$\n\\mathbb {P} (a \\leq X \\leq b) = \\int_ {a} ^ {b} f (x) d x\n$$\nfor every $a$ and $b$ . [Sometimes we write that for nice sets $B \\subset \\mathbb{R}$ we have $\\mathbb{P}(X \\in B) = \\int_{B} f(x) dx$ .]\nWe call $f$ the pdf (probability density function) for $X$ . Sometime we we the notation $f_{X}$ to signify $f_{X}$ corresponds to the pdf of $X$ . We sometimes call $f_{X}$ the density of $X$ .\n\\- In fact, any function $f$ satisfying the following two properties is called a density, and could be considered a pdf of some random variable $X$ :\n(1) $f(x)\\geq 0$ for all $x$\n(2) $\\int_{-\\infty}^{\\infty} f(x) dx = 1$ .\n\\- Important Note!\n- (1) In this case $X: S \\to \\mathbb{R}$ and the could attain uncountably many values (doesn't have to discrete)\n- (2) $\\int_{-\\infty}^{\\infty} f(x) dx = \\mathbb{P}(-\\infty < X < \\infty) = 1$ .\n- (3) $\\mathbb{P}(X = a) = \\int_{a}^{a} f(x) dx = 0$ .\n- (4) $\\mathbb{P}(X < a) = \\mathbb{P}(X \\leq a) = F(a) = \\int_{-\\infty}^{a} f(x) dx$ .\n\\* Recall that $F$ is the cdf of $X$ .\n\\- (5) Draw a pdf of $X$\n\\* Note that $\\mathbb{P}(a < X < b)$ is just the area under the curve.\n\\- Remark: What are some random variables that are considered continuous?\n- Let $X$ be the time it takes it take for a student to finish a probability exam. $X \\in (0, \\infty)$ .\n- Let $X$ be the value of a Apple's stock price at the end of the day. Again $X \\in [0, \\infty)$ .\n- Let $X$ be the height of a college student.\n- Any sort of continuous measurement can be considered a continuous random variable.\n• Example1: Suppose we are given\n$$\nf (x) = \\left\\{ \\begin{array}{l l} \\frac {c}{x ^ {3}} & x \\geq 1 \\\\ 0 & x < 1 \\end{array} \\right.\n$$\nis the pdf of $X$ . What must the value of $c$ be?\n\\- Solution: We would need\n$$\n1 = \\int_ {- \\infty} ^ {\\infty} f (x) d x = c \\int_ {1} ^ {\\infty} \\frac {1}{x ^ {3}} d x = \\frac {c}{2},\n$$\nthus c = 2.\n• Example2: Suppose we are given\n$$\nf _ {X} (x) = \\left\\{ \\begin{array}{l l} \\frac {2}{x ^ {3}} & x \\geq 1 \\\\ 0 & x < 1 \\end{array} \\right.\n$$\nis the pdf of $X$ from Example1.\n\\- Part (a): Find the c.d.f, $F_{X}(x)$ .\n\\* Solution: First we check that if x < 1 then\n$$\nF _ {x} (x) = \\mathbb {P} (X \\leq x) = \\int_ {- \\infty} ^ {x} f _ {X} (y) d y = \\int_ {- \\infty} ^ {x} 0 d y = 0.\n$$\nNow when $x \\geq 1$ we have\n$$\n\\begin{array}{l} F _ {X} (x) = \\mathbb {P} (X \\leq x) = \\int_ {- \\infty} ^ {x} f _ {X} (y) d y \\\\ = \\int_ {- \\infty} ^ {1} 0 d y + \\int_ {1} ^ {x} \\frac {2}{y ^ {3}} d y \\\\ = \\int_ {1} ^ {x} \\frac {2}{y ^ {3}} d y \\\\ = 1 - \\frac {1}{x ^ {2}}. \\\\ \\end{array}\n$$\nthus\n$$\nF _ {X} (x) = \\left\\{ \\begin{array}{l l} 1 - \\frac {1}{x ^ {2}} & x \\geq 1 \\\\ x & x < 1 \\end{array} \\right.\n$$\n\\- Part (b): Use the cdf in Part (a) to help you find $\\mathbb{P}(3\\leq X\\leq 4)$ .\n\\* Solution: We have\n$$\n\\begin{array}{l} \\mathbb {P} (3 \\leq X \\leq 4) = \\mathbb {P} (X \\leq 4) - \\mathbb {P} (X < 3) \\\\ = F _ {X} (4) - F _ {X} (3) \\\\ = \\left(1 - \\frac {1}{4 ^ {2}}\\right) - \\left(1 - \\frac {1}{3 ^ {2}}\\right) = \\frac {7}{1 4 4}. \\\\ \\end{array}\n$$\n\\- Fact: For continuous R.V we have the following useful relationship\n\\- Since $F(x) = \\int_{-\\infty}^{x} f(y) dy$ then by the fundament at theorem of calculus (Do you remember this form Calculus 1 or 2?)\n$$\nF ^ {\\prime} (x) = f (x).\n$$\n\\- This means that for continuous random variables, the derivative of the CDF is the PDF!\n\\- Example3: Let\n$$\nf (x) = \\left\\{ \\begin{array}{l l} c e ^ {- 2 x} & x \\geq 0 \\\\ 0 & x < 0 \\end{array} \\right.\n$$\nFind c.\n\\- Solution: $c = 2$ .", + "path": "Notes.pdf/CHAPTER 7/Continuous Random Variables/7.1. Intro to continuous R.V", + "metadata": { + "length": 3617, + "summary": "This text defines continuous random variables and their probability density functions (PDFs). It states that a random variable X has a continuous distribution if there exists a non-negative function f such that probabilities are integrals of f. The PDF must satisfy f(x) ≥ 0 and integrate to 1 over all real numbers. Key properties include that P(X = a) = 0 and the cumulative distribution function (CDF) F(x) is the integral of the PDF from negative infinity to x. The derivative of the CDF equals the PDF. Examples demonstrate finding a constant c in a PDF and computing probabilities using the CDF.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "DEFINITION", + "random", + "variable", + "continuous", + "distribution", + "exists", + "negative", + "function", + "mathbb", + "leq", + "int", + "Sometimes", + "write", + "nice", + "sets", + "subset", + "dx", + "We", + "call", + "pdf", + "probability", + "density", + "Sometime", + "notation", + "signify", + "corresponds", + "In", + "fact", + "satisfying", + "properties", + "called", + "considered", + "geq", + "infty", + "Important", + "Note", + "case", + "attain", + "uncountably", + "values", + "doesn", + "discrete", + "Recall", + "cdf", + "Draw", + "area", + "curve", + "Remark", + "What", + "variables", + "Let", + "time", + "takes", + "student", + "finish", + "exam", + "Apple", + "stock", + "price", + "end", + "day", + "Again", + "height", + "college", + "Any", + "sort", + "measurement", + "Example1", + "Suppose", + "left", + "begin", + "array", + "frac", + "Solution", + "Example2", + "Part", + "Find", + "First", + "check", + "Now", + "Use", + "find", + "Fact", + "For", + "relationship", + "Since", + "dy", + "fundament", + "theorem", + "calculus", + "Do", + "remember", + "form", + "Calculus", + "prime", + "This", + "means", + "derivative", + "CDF", + "PDF", + "Example3" + ], + "keywords": [ + "PDF", + "CDF", + "continuous" + ], + "connect_to": [] + } + }, + { + "chunk_id": "518ec1c0-44c8-5851-b7e8-86a65a042fdf", + "type": "image", + "content": "\n$$ PROPOSITION. If $X$ is a continuous $R.V.$ with pdf $f(x)$ , then for any real valued function $g$ ,0\n[images/image-16-$$ PROPOSI.jpg]\n", + "path": "images/image-16-$$ PROPOSI.jpg", + "metadata": { + "length": 139, + "summary": "image-16\n$$ PROPOSITION. If $X$ is a continuous $R.V.$ with pdf $f(x)$ , then for any real valued function $g$ ,0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-16-$$ PROPOSI.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "ed52bb50-398e-5ecb-9c04-c710dc961e9e", + "type": "text", + "content": "\\- Recall that if $p(x)$ is the pmf (density) of a discrete random variable, we had\n$$\n\\mathbb {E} X = \\sum_ {i = 1} ^ {\\infty} x _ {i} p (x _ {i}).\n$$\nDEFINITION. If $X$ is continuous with density $f(x)$ then\n$$\n\\mathbb {E} X = \\int_ {- \\infty} ^ {\\infty} x f (x) d x.\n$$\n• Example1: Suppose X has density\n$$\nf (x) = \\left\\{ \\begin{array}{l l} 2 x & 0 \\leq x \\leq 1 \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\nFind $\\mathbb{E}X$ .\n\\- Solution: We have that\n$$\n\\begin{array}{l} \\mathbb {E} [ X ] = \\int_ {- \\infty} ^ {\\infty} x f (x) d x \\\\ = \\int_ {0} ^ {1} x \\cdot 2 x d x \\\\ = \\frac {2}{3}. \\\\ \\end{array}\n$$\nTHEOREM 9. If $X$ and $Y$ are continuous random variable then\n(a) $\\mathbb{E}[X + Y] = \\mathbb{E}X + \\mathbb{E}Y.$\n(b) $E[aX]=aE X$ where $a\\in R$ .\nPROOF. See textbook. It will be shown later.\n\n$$ PROPOSITION. If $X$ is a continuous $R.V.$ with pdf $f(x)$ , then for any real valued function $g$ ,0\n[images/image-16-$$ PROPOSI.jpg]\n\nPROPOSITION. If $X$ is a continuous $R.V.$ with pdf $f(x)$ , then for any real valued function $g$ ,\n$$\n\\mathbb {E} \\left[ g (X) \\right] = \\int_ {- \\infty} ^ {\\infty} g (x) f (x) d x.\n$$\n• Example2: The density of X is given by\n$$\nf (x) = \\left\\{ \\begin{array}{l l} \\frac {1}{2} & \\text { if } 0 \\leq x \\leq 2 \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\nFind $\\mathbb{E}\\left[e^{X}\\right]$ .\n\\- Solution: From the previous proposition we have that $g(x) = e^{x}$ in this case thus\n$$\n\\mathbb {E} e ^ {X} = \\int_ {0} ^ {2} e ^ {x} \\cdot \\frac {1}{2} d x = \\frac {1}{2} \\left[ e ^ {2} - 1 \\right].\n$$\nLEMMA 10. For nonnegative random variable $Y \\geq 0$ we have\n$$\n\\mathbb {E} Y = \\int_ {0} ^ {\\infty} \\mathbb {P} (Y > y) d y.\n$$\n\\- Bonus:\n\\- This proof is a good practice with interchanging order of integrals in Multivariable Calculus.\nPROOF. Recall that dxdy means Right-Left and dydx means Top-Bottom.\n$$\n\\begin{array}{l} \\int_ {0} ^ {\\infty} \\mathbb {P} (Y > y) d y = \\int_ {0} ^ {\\infty} \\int_ {y} ^ {\\infty} f _ {Y} (x) d x d y \\\\ = \\int \\int_ {D} f _ {Y} (x) d y d x, \\text { interchange order in Calc III } \\\\ = \\int_ {0} ^ {\\infty} \\int_ {0} ^ {x} f _ {Y} (x) d y d x \\text { draw the region to do this } \\\\ = \\int_ {0} ^ {\\infty} x f _ {Y} (x) d x \\\\ = \\mathbb {E} X. \\\\ \\end{array}\n$$\n\\- Variance:\n\\- Will be define in the same way as we did with discrete random variable:\n$$\n\\operatorname{Var} (X) = \\mathbb {E} \\left[ (X - \\mu) ^ {2} \\right]\n$$\n$$\n\\operatorname{Var} (X) = \\mathbb {E} X ^ {2} - (\\mathbb {E} X) ^ {2}.\n$$\n\\- As before\n$$\n\\operatorname{Var} (a X + b) = a ^ {2} \\operatorname{Var} (X).\n$$\n\\- Example3: (Example 1 continued) Suppose $X$ has density\n$$\nf (x) = \\left\\{ \\begin{array}{l l} 2 x & 0 \\leq x \\leq 1 \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\nFind $\\operatorname{Var}(X)$ .\n\\- Solution: From Example 1 we found $\\mathbb{E}[X] = \\frac{2}{3}$ . Now\n$$\n\\begin{array}{l} \\mathbb {E} \\left[ X ^ {2} \\right] = \\int_ {0} ^ {1} x ^ {2} \\cdot 2 x d x = 2 \\int_ {0} ^ {1} x ^ {3} d x \\\\ = \\frac {1}{2}. \\\\ \\end{array}\n$$\nThus\n$$\n\\operatorname{Var} (X) = \\frac {1}{2} - \\left(\\frac {2}{3}\\right) ^ {2} = \\frac {1}{1 8}.\n$$\n• Example4: Suppose X has density\n$$\nf (x) = \\left\\{ \\begin{array}{l l} a x + b & 0 \\leq x \\leq 1 \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\nand that $\\mathbb{E}\\left[X^2\\right] = \\frac{1}{6}$ . Find the values of $a$ and $b$ .\n\\- Solution: We need to use the fact that $\\int_{-\\infty}^{\\infty}f(x)dx = 1$ and $\\mathbb{E}\\left[X^2\\right] = \\frac{1}{6}$ . The first one gives us,\n$$\n1 = \\int_ {0} ^ {1} (a x + b) d x = \\frac {a}{2} + b\n$$\nand the second one give us\n$$\n\\frac {1}{6} = \\int_ {0} ^ {1} x ^ {2} (a x + b) d x = \\frac {a}{4} + \\frac {b}{3}.\n$$\nSolving these equations gives us\n$$\na = - 2, \\text { and } b = 2.\n$$", + "path": "Notes.pdf/CHAPTER 7/Continuous Random Variables/7.2. Expectation and Variance", + "metadata": { + "length": 3771, + "summary": "This passage defines expectation for continuous random variables using integrals, presents properties like linearity, and introduces variance. Examples compute E[X] and Var(X) for given densities, and solve for parameters using moment conditions.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Recall", + "pmf", + "density", + "discrete", + "random", + "variable", + "mathbb", + "sum", + "infty", + "DEFINITION", + "If", + "continuous", + "int", + "Example1", + "Suppose", + "left", + "begin", + "array", + "leq", + "text", + "end", + "Find", + "Solution", + "We", + "cdot", + "frac", + "THEOREM", + "aX", + "aE", + "PROOF", + "See", + "textbook", + "It", + "shown", + "PROPOSITION", + "pdf", + "real", + "valued", + "function", + "Example2", + "The", + "From", + "previous", + "proposition", + "case", + "LEMMA", + "10", + "For", + "nonnegative", + "geq", + "Bonus", + "This", + "proof", + "good", + "practice", + "interchanging", + "order", + "integrals", + "Multivariable", + "Calculus", + "dxdy", + "means", + "Right", + "Left", + "dydx", + "Top", + "Bottom", + "interchange", + "Calc", + "III", + "draw", + "region", + "Variance", + "Will", + "define", + "operatorname", + "Var", + "mu", + "As", + "Example3", + "Example", + "continued", + "found", + "Now", + "Thus", + "Example4", + "values", + "fact", + "dx", + "give", + "Solving", + "equations" + ], + "keywords": [ + "Expectation", + "Variance", + "Continuous" + ], + "connect_to": [ + { + "target": "518ec1c0-44c8-5851-b7e8-86a65a042fdf", + "relation": "embeds", + "ref": "[images/image-16-$$ PROPOSI.jpg]", + "position": { + "start": 919, + "end": 951 + } + } + ] + } + }, + { + "chunk_id": "43b72c77-8861-5334-97f9-e0cdd65b87c3", + "type": "text", + "content": "\\- A continuous random variable is said to be uniformly distributed on the interval $[a, b]$ if\n$$\nf _ {X} (x) = \\left\\{ \\begin{array}{l l} \\frac {1}{b - a} & a \\leq x \\leq b \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\n- So $X$ can only attain values in $X \\in [a, b]$ .\n- We say $X \\sim \\text{Uniform}(a, b)$ .\n- The cdf is\n$$\nF _ {X} (x) = \\left\\{ \\begin{array}{l l} 0 & x < a \\\\ \\frac {x - a}{b - a} & a \\leq x \\leq b \\\\ 1 & x > b \\end{array} \\right..\n$$\n\\- Example1: Suppose $X \\sim \\text{Uniform}(a, b)$ Part (a) Find the mean of $X$ . Part (b) Find the variance of $X$ .\n\\- Part (a): We compute\n$$\n\\begin{array}{l} \\mathbb {E} X = \\int_ {- \\infty} ^ {\\infty} x f _ {X} (x) d x = \\int_ {a} ^ {b} x \\frac {1}{b - a} d x \\\\ = \\frac {1}{b - a} \\left(\\frac {b ^ {2}}{2} - \\frac {a ^ {2}}{2}\\right) = \\frac {a + b}{2}. \\\\ \\end{array}\n$$\n\\* Which makes sense right? It should be the midpoint of the interval $[a, b]$ .\n\\- Part(b): We compute first the second moment\n$$\n\\begin{array}{l} \\mathbb {E} X ^ {2} = \\int_ {a} ^ {b} x ^ {2} \\frac {1}{b - a} d x = \\frac {1}{b - a} \\left(\\frac {b ^ {3}}{3} - \\frac {a ^ {3}}{3}\\right) \\\\ = \\frac {1}{3} \\frac {1}{b - a} (b - a) (a ^ {2} + a b + b ^ {2}) \\\\ = \\frac {a ^ {2} + a b + b ^ {2}}{3}. \\\\ \\end{array}\n$$\nThus after some algebra\n$$\n\\operatorname{Var} X = \\frac {a ^ {2} + a b + b ^ {2}}{3} - \\left(\\frac {a + b}{2}\\right) ^ {2} = \\frac {(b - a) ^ {2}}{1 2}.\n$$", + "path": "Notes.pdf/CHAPTER 7/Continuous Random Variables/7.3. The uniform Random Variable", + "metadata": { + "length": 1416, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "continuous", + "random", + "variable", + "uniformly", + "distributed", + "interval", + "left", + "begin", + "array", + "frac", + "leq", + "text", + "end", + "So", + "attain", + "values", + "We", + "sim", + "Uniform", + "The", + "cdf", + "Example1", + "Suppose", + "Part", + "Find", + "variance", + "compute", + "mathbb", + "int", + "infty", + "Which", + "makes", + "sense", + "It", + "midpoint", + "moment", + "Thus", + "algebra", + "operatorname", + "Var" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "078f595d-28b9-51b0-94cb-61f578b8b059", + "type": "text", + "content": "\\- Suppose we are given the pd.f.\n$$\nf (x) = \\left\\{ \\begin{array}{l l} 9 e ^ {- 9 x} & x \\geq 0 \\\\ 0 & x < 0 \\end{array} \\right.\n$$\n\\- Part (a): Set up integral to find $F_{X}(x)$ :\n\\* We have for $x > 0$ , that\n$$\nF _ {X} (x) = \\int_ {0} ^ {x} 9 e ^ {- 9 y} d y = 1 - e ^ {- 9 x},\n$$\nso that\n$$\nF _ {X} (x) = \\left\\{ \\begin{array}{l l} 1 - e ^ {- 9 x} & x \\geq 0 \\\\ 0 & x < 0 \\end{array} \\right. /\n$$\n\\- Part (b): Set up integral to find $\\mathbb{P}(1 < X < 5)$\n$$\n* \\int_ {1} ^ {5} 9 e ^ {- 9 x} d x\n$$\n\\- Part (c): Set up integral to find $\\mathbb{P}(X > 3)$\n$$\n* \\int_ {3} ^ {\\infty} 9 e ^ {- 9 x} d x.\n$$\n\\- Part (s): Set up integral to find $\\mathbb{P}(X < 2)$\n$$\n* \\int_ {0} ^ {2} 9 e ^ {- 9 x} d x.\n$$", + "path": "Notes.pdf/CHAPTER 7/Continuous Random Variables/7.4. More practice", + "metadata": { + "length": 714, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Suppose", + "pd", + "left", + "begin", + "array", + "geq", + "end", + "Part", + "Set", + "integral", + "find", + "We", + "int", + "mathbb", + "infty" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "68362ba5-02d4-59ca-bba4-a806a10b903f", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 8", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8584178e-8809-5ec0-8f25-810a91a8b5f1", + "type": "image", + "content": "\n8.1. The normal distribution
    0\n[images/image-17-Standard Normal Distribution Curve.jpg]\n", + "path": "images/image-17-Standard Normal Distribution Curve.jpg", + "metadata": { + "length": 74, + "summary": "image-17\n- Chart Type: Bell curve representing a standard normal distribution (Gaussian distribution)\n- X-axis Range: -3 to 3\n- Center Point: 0 (representing the mean and median)\n- Key Features: Symmetrical shape with peak at x=0, tails extending towards negative and positive infinity", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-17-Standard Normal Distribution Curve.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "95079a7c-b5e5-5a10-841a-919c0a64caef", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
    xy
    -3.00.000
    -2.50.047
    -2.00.169
    -1.50.362
    -1.00.540
    -0.50.728
    0.01.000
    0.50.728
    1.00.540
    1.50.362
    2.00.169
    2.50.047
    3.00.000
    ", + "path": "tables/table-2 Symmetric Data.html", + "metadata": { + "length": 940, + "summary": "table-3\nThe table shows x values from -3.0 to 3.0 with corresponding y values peaking at 1.000 when x=0.0, decreasing symmetrically to 0.000 at the extremes.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-2 Symmetric Data.html", + "keywords": [ + "x values", + "y values", + "symmetry" + ], + "tokens": [] + } + }, + { + "chunk_id": "44ed73ed-4948-5b67-b757-7c31fa97bdde", + "type": "image", + "content": "\nPROPOSITION 2. If $E$ and $F$ are independent, then $E$ and $F^c$ are independent. \\- Remark: Independence and mutually exclusive, are two different things!0\n[images/image-10-PROPOSITIO.jpg]\n", + "path": "images/image-10-PROPOSITIO.jpg_2", + "metadata": { + "length": 192, + "summary": "image-10\nPROPOSITION 2. If $E$ and $F$ are independent, then $E$ and $F^c$ are independent. \\- Remark: Independence and mutually exclusive, are two different things!0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-10-PROPOSITIO.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "515e5017-d557-5d09-99fc-376c8b833ec6", + "type": "image", + "content": "\n$$
    0\n[images/image-18-Standard Normal Distribution Graph.jpg]\n", + "path": "images/image-18-Standard Normal Distribution Graph.jpg", + "metadata": { + "length": 48, + "summary": "image-18\n- Chart Title: φ(Z) (PDF function of standard normal distribution)\n- X-Axis Label: Z (Standard Score / Z-score)\n- Y-Axis Label: φ(Z) (Probability Density Function)\n- Key Features:\n - Bell-shaped curve centered at zero.\n - Blue shaded area under the curve represents the Cumulative Distribution Function (CDF).\n - An arrow labeled \"CDF\" points to the shaded region, indicating the probability accumulated up to a specific Z value.\n - A vertical line marks a specific point on the X-axis labeled \"Z\", defining the upper bound of the shaded CDF area.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-18-Standard Normal Distribution Graph.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "3e1da8fa-2773-5f1b-a652-01c160815d8a", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
    ZCDF
    00
    zPeak
    ", + "path": "tables/table-3 Z and CDF Table.html", + "metadata": { + "length": 278, + "summary": "table-4\nThe table shows Z values and their corresponding CDF, with entries for 0 and z mapping to 0 and Peak respectively.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-3 Z and CDF Table.html", + "keywords": [ + "CDF", + "Peak" + ], + "tokens": [] + } + }, + { + "chunk_id": "1c2741ec-e4eb-51e6-b48a-0b57076c9932", + "type": "image", + "content": "\n• FACT: The 68 - 95 - 99.7 Rule
    0\n[images/image-19-Normal Distribution Curve.jpg]\n", + "path": "images/image-19-Normal Distribution Curve.jpg", + "metadata": { + "length": 77, + "summary": "image-19\n- Chart Title: Normal Distribution (Bell Curve)\n- Axis Labels: Horizontal axis represents standard deviations from the mean ($\\mu$), marked as $-3\\sigma, -2\\sigma, -1\\sigma, \\mu, 1\\sigma, 2\\sigma, 3\\sigma$. Vertical axis is unlabeled but represents probability density.\n- Key Data Points & Percentages:\n - Central Region ($\\mu \\pm 1\\sigma$): Two sections of 34.1% each (Total 68.2%)\n - Second Region ($\\pm 1\\sigma$ to $\\pm 2\\sigma$): Two sections of 13.6% each (Total 27.2%)\n - Third Region ($\\pm 2\\sigma$ to $\\pm 3\\sigma$): Two sections of 2.1% each (Total 4.2%)\n - Outer Regions (Beyond $\\pm 3\\sigma$): Two sections of 0.1% each (Total 0.2%)", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-19-Normal Distribution Curve.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "a8c29f1b-5148-5953-b396-9a5e56ae50cb", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
    Standard DeviationPercentage
    -3σ0.1%
    -2σ2.1%
    -1σ13.6%
    μ34.1%
    34.1%
    13.6%
    2.1%
    >3σ0.1%
    ", + "path": "tables/table-4 Normal Distribution.html", + "metadata": { + "length": 659, + "summary": "table-5\nThe table shows the percentage of data within each standard deviation from the mean in a normal distribution, with 34.1% at the mean and decreasing percentages outward.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-4 Normal Distribution.html", + "keywords": [ + "Standard Deviation", + "Percentage", + "Normal Distribution" + ], + "tokens": [] + } + }, + { + "chunk_id": "c1989fd8-8e37-58e8-a6f2-facd32cb6e30", + "type": "text", + "content": "8.1. The normal distribution\n\\- We say that $X$ is a normal (Gaussian) random variable, or $X$ is normally distributed with parameters $\\mu$ and $\\sigma^2$ if the density of $X$ is given by\n$$\nf (x) = \\frac {1}{\\sqrt {2 \\pi} \\sigma} e ^ {- (x - \\mu) ^ {2} / \\left(2 \\sigma^ {2}\\right)}.\n$$\n\n8.1. The normal distribution
    0\n[images/image-17-Standard Normal Distribution Curve.jpg]\n\n
    \nline chart\n\n[tables/table-2 Symmetric Data.html]\n\n
    \n\\- We'll usually write $X \\sim \\mathcal{N}(\\mu, \\sigma^2)$ .\n\\- Turns out that in practice, many random variable overy the normal distribution\n\\* Grades\n\\* Height of a man or a women\n\\- Note the following:\n\\- If $X \\sim \\mathcal{N}(0,1)$ then\n$$\n\\int_ {- \\infty} ^ {\\infty} \\frac {1}{\\sqrt {2 \\pi}} e ^ {- x ^ {2} / 2} d x = 1.\n$$\n\\- To show this we use polar coordinates. Let $I = \\int_{-\\infty}^{\\infty} e^{-x^2/2} dx = 2 \\int_0^\\infty e^{-x^2/2} dx$ The trick is to write\n$$\n\\begin{array}{l} I ^ {2} = 4 \\int_ {0} ^ {\\infty} \\int_ {0} ^ {\\infty} e ^ {- x ^ {2} / 2} e ^ {- y ^ {2} / 2} d x y \\\\ = 4 \\int_ {0} ^ {\\pi / 2} \\int_ {0} ^ {\\infty} r e ^ {- r ^ {2} / 2} d r = 4 \\cdot \\frac {\\pi}{2} = 2 \\pi , \\\\ \\end{array}\n$$\nThus $I = \\sqrt{2\\pi}$ as needed.\nTHEOREM 11. To help us compute the mean and variance of $X$ its not too hard to show $X \\sim \\mathcal{N}\\left(\\mu, \\sigma^2\\right)$ if and only if\n$$\n\\frac {X - \\mu}{\\sigma} = Z \\text { where } Z \\sim \\mathcal {N} (0, 1).\n$$\nPROOF. We only show the ( $\\Longleftrightarrow$ ) direction. Note that\n$$\n\\begin{array}{l} F _ {X} (x) = \\mathbb {P} (X \\leq x) = \\mathbb {P} (\\sigma Z + \\mu \\leq x) \\\\ = \\mathbb {P} \\left(Z \\leq \\frac {x - \\mu}{\\sigma}\\right) \\\\ { = } { F _ { Y } \\left( \\frac { x - \\mu } { \\sigma } \\right) } \\\\ \\end{array}\n$$\nfor $\\sigma > 0$ . Similar for $\\sigma < 0$ . By the chain rule\n$$\n\\begin{array}{l} f _ {X} (x) = F _ {X} ^ {\\prime} (x) \\\\ = F _ {Y} ^ {\\prime} \\left(\\frac {x - \\mu}{\\sigma}\\right) \\frac {1}{\\sigma} \\\\ = \\frac {f _ {Z} \\left(\\frac {x - \\mu}{\\sigma}\\right)}{\\sigma} \\\\ { = } { \\frac { 1 } { \\sigma } \\frac { 1 } { \\sqrt { 2 \\pi } } e ^ { - ( x - \\mu ) ^ { 2 } / ( 2 \\sigma ^ { 2 } ) } . } \\\\ \\end{array}\n$$\n\nPROPOSITION 2. If $E$ and $F$ are independent, then $E$ and $F^c$ are independent. \\- Remark: Independence and mutually exclusive, are two different things!0\n[images/image-10-PROPOSITIO.jpg]\n\n• Summary of the normal distribution:\n\\- If $X \\sim \\mathcal{N}\\left(\\mu, \\sigma^2\\right)$ then $X$ is normally distributed with\n$$\n\\mathbb {E} X = \\mu ,\n$$\n$$\n\\operatorname{Var} (X) = \\sigma^ {2}.\n$$\n\\- If $X \\sim \\mathcal{N}(\\mu, \\sigma^2)$ then $X = \\sigma Z + \\mu$ where $Z \\sim \\mathcal{N}(0,1)$ . We call $Z$ a standard normal random variable.\n\\* A Table of probabilities for Z will be given!!!\n\\* This will be called a z-score table.\n\\- Z scores:\n\\- Because $Z \\sim \\mathcal{N}(0,1)$ is so important we give it's cumulative distribution function (cdf) a name. The distribution $F_Z(x)$ of $Z$ is\n$$\n\\Phi (x) = \\mathbb {P} (Z \\leq x) = \\frac {1}{\\sqrt {2 \\pi}} \\int_ {- \\infty} ^ {x} e ^ {- y ^ {2} / 2} d y.\n$$\n\\- NOTE: A table of $\\Phi(x)$ will be given but only for values of $x > 0$\n\\- Note this is symmetric[ $\\underline{DRAW this}$ ] thus here is an important fact: $\\Phi(-x)=1-\\Phi(x)$\n\n$$
    0\n[images/image-18-Standard Normal Distribution Graph.jpg]\n\n
    \narea chart\n\n[tables/table-3 Z and CDF Table.html]\n\n
    \nTHEOREM 12. If $X \\sim \\mathcal{N}(\\mu, \\sigma)$ then\n$$\n\\mathbb {P} (a < X < b) = \\mathbb {P} \\left(\\frac {a - \\mu}{\\sigma} < Z < \\frac {b - \\mu}{\\sigma}\\right).\n$$\n\\- Example1: Find $\\mathbb{P}(1 \\leq X \\leq 4)$ if $X \\sim \\mathcal{N}(2, 25)$ .\n\\- $\\underline{\\mathbf{Answer}}$ : Then $\\mu = 2$ and $\\sigma^2 = 25$ thus $\\frac{X - 2}{5} = Z$ so that\n$$\n\\begin{array}{l} \\mathbb {P} (1 \\leq X \\leq 4) = \\mathbb {P} \\left(\\frac {1 - 2}{5} \\leq \\frac {X - 2}{5} \\leq \\frac {4 - 2}{5}\\right) \\\\ = \\mathbb {P} (-. 2 \\leq Z \\leq . 4) \\\\ = \\mathbb {P} (X \\leq . 4) - \\mathbb {P} (X \\leq -. 2) \\\\ = \\Phi (. 4) - \\Phi (- 2) \\\\ = . 6 5 5 4 - (1 - \\Phi (. 2)) \\\\ = . 6 5 5 4 - (1 -. 5 7 9 3). \\\\ \\end{array}\n$$\n\\- Example2: Suppose $X$ is normal with mean 6. If $\\mathbb{P}(X > 16) = .0228$ , then what is the standard deviation of $X$ ?\n\\- $\\underline{\\text{Answer}}$ : We apply our Theorem that says $\\frac{X - \\mu}{\\sigma} = Z$ is $\\mathcal{N}(0,1)$ and get\n$$\n\\begin{array}{l} \\mathbb {P} (X > 1 6) = . 0 2 2 8 \\quad \\Longleftrightarrow \\quad \\mathbb {P} \\left(\\frac {X - 6}{\\sigma} > \\frac {1 6 - 6}{\\sigma}\\right) = . 0 2 2 8 \\\\ \\Longleftrightarrow \\quad \\mathbb {P} \\left(Z > \\frac {1 0}{\\sigma}\\right) = . 0 2 2 8 \\\\ \\Longleftrightarrow \\quad 1 - \\mathbb {P} \\left(Z \\leq \\frac {1 0}{\\sigma}\\right) = . 0 2 2 8 \\\\ \\Longleftrightarrow \\quad 1 - \\Phi \\left(\\frac {1 0}{\\sigma}\\right) = . 0 2 2 8 \\\\ \\Longleftrightarrow \\quad \\Phi \\left(\\frac {1 0}{\\sigma}\\right) = . 9 7 7 2. \\\\ \\end{array}\n$$\nUsing the standard normal table we see that $\\Phi(2) = .9772$ , thus we must have that\n$$\n2 = \\frac {1 0}{\\sigma}\n$$\nand hence $\\sigma = 5$ .\n\\- Example (Extra): Suppose $X \\sim \\mathcal{N}(3,9)$ find $\\mathbb{P}\\left(|X - 3| > 6\\right)$ .\n\\- Answer: Get\n$$\n\\begin{array}{l} \\mathbb {P} (| X - 3 | > 6) = \\mathbb {P} (X - 3 > 6) + \\mathbb {P} (- (X - 3) > 6) \\\\ = \\mathbb {P} (X > 9) + \\mathbb {P} (X < - 3) \\\\ = \\mathbb {P} (Z > 2) + \\mathbb {P} (Z < - 2) \\\\ = 1 - \\Phi (2) + \\Phi (- 2) \\\\ = 2 (1 - \\Phi (2)) \\\\ \\approx . 0 4 5 6. \\\\ \\end{array}\n$$\n• FACT: The 68 - 95 - 99.7 Rule\n- About $68\\%$ of all area is contained within 1 standard deviation of the mean\n- About $95\\%$ of all area is contained within 2 standard deviation of the mean\n- About $99.7\\%$ of all area is contained within 3 standard deviation of the mean\n- This can be explained by the following graph:\n\n• FACT: The 68 - 95 - 99.7 Rule
    0\n[images/image-19-Normal Distribution Curve.jpg]\n\n
    \narea chart\n\n[tables/table-4 Normal Distribution.html]\n\n
    ", + "path": "Notes.pdf/CHAPTER 8/Normal Distributions", + "metadata": { + "length": 5924, + "summary": "The text covers the normal (Gaussian) distribution, its density function, properties, and the standard normal random variable Z. It explains that any normal variable X with mean μ and variance σ² can be standardized as Z = (X - μ)/σ, which follows N(0,1). Key results include the cumulative distribution function Φ(x), symmetry property Φ(-x)=1-Φ(x), and the 68-95-99.7 rule stating that about 68%, 95%, and 99.7% of data lies within 1, 2, and 3 standard deviations of the mean, respectively. Examples demonstrate calculating probabilities using Z-scores and finding standard deviation from a given probability.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "8.1", + "The", + "normal", + "distribution", + "We", + "Gaussian", + "random", + "variable", + "distributed", + "parameters", + "mu", + "sigma", + "density", + "frac", + "sqrt", + "pi", + "left", + "details", + "summary", + "line", + "chart", + "ll", + "write", + "sim", + "mathcal", + "Turns", + "practice", + "overy", + "Grades", + "Height", + "man", + "women", + "Note", + "If", + "int", + "infty", + "To", + "show", + "polar", + "coordinates", + "Let", + "dx", + "trick", + "begin", + "array", + "cdot", + "end", + "Thus", + "needed", + "THEOREM", + "11", + "compute", + "variance", + "hard", + "text", + "PROOF", + "Longleftrightarrow", + "direction", + "mathbb", + "leq", + "Similar", + "By", + "chain", + "rule", + "prime", + "PROPOSITION", + "independent", + "Remark", + "Independence", + "mutually", + "exclusive", + "things", + "Summary", + "operatorname", + "Var", + "call", + "standard", + "Table", + "probabilities", + "This", + "called", + "score", + "table", + "scores", + "Because", + "important", + "give", + "cumulative", + "function", + "cdf", + "Phi", + "NOTE", + "values", + "symmetric", + "underline", + "DRAW", + "fact", + "area", + "12", + "Example1", + "Find", + "25", + "mathbf", + "Answer", + "Then", + "Example2", + "Suppose", + "16", + "0228", + "deviation", + "apply", + "Theorem", + "quad", + "Using", + "9772", + "Example", + "Extra", + "find", + "Get", + "approx", + "FACT", + "68", + "95", + "99.7", + "Rule", + "About", + "contained", + "explained", + "graph" + ], + "keywords": [ + "Normal Distribution", + "Z-score", + "68-95-99.7 Rule" + ], + "connect_to": [ + { + "target": "8584178e-8809-5ec0-8f25-810a91a8b5f1", + "relation": "embeds", + "ref": "[images/image-17-Standard Normal Distribution Curve.jpg]", + "position": { + "start": 331, + "end": 387 + } + }, + { + "target": "95079a7c-b5e5-5a10-841a-919c0a64caef", + "relation": "embeds", + "ref": "[tables/table-2 Symmetric Data.html]", + "position": { + "start": 430, + "end": 466 + } + }, + { + "target": "44ed73ed-4948-5b67-b757-7c31fa97bdde", + "relation": "embeds", + "ref": "[images/image-10-PROPOSITIO.jpg]", + "position": { + "start": 2344, + "end": 2376 + } + }, + { + "target": "515e5017-d557-5d09-99fc-376c8b833ec6", + "relation": "embeds", + "ref": "[images/image-18-Standard Normal Distribution Graph.jpg]", + "position": { + "start": 3292, + "end": 3348 + } + }, + { + "target": "3e1da8fa-2773-5f1b-a652-01c160815d8a", + "relation": "embeds", + "ref": "[tables/table-3 Z and CDF Table.html]", + "position": { + "start": 3391, + "end": 3428 + } + }, + { + "target": "1c2741ec-e4eb-51e6-b48a-0b57076c9932", + "relation": "embeds", + "ref": "[images/image-19-Normal Distribution Curve.jpg]", + "position": { + "start": 5860, + "end": 5907 + } + }, + { + "target": "a8c29f1b-5148-5953-b396-9a5e56ae50cb", + "relation": "embeds", + "ref": "[tables/table-4 Normal Distribution.html]", + "position": { + "start": 5950, + "end": 5991 + } + } + ] + } + }, + { + "chunk_id": "d97d704e-f265-58f9-a2b1-6472332b6f1d", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 9", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "199b7ba8-7cab-5e3f-84e2-87e9795c9e70", + "type": "text", + "content": "9.1. The normal approximates Binomial\nTHEOREM 13. If $S_{n}$ is a binomial with parameter $n$ and $p$ , then\n$$\n\\mathbb {P} \\left(a \\leq \\frac {S _ {n} - n p}{\\sqrt {n p (1 - p)}} \\leq b\\right)\\rightarrow \\mathbb {P} (a \\leq Z \\leq b)\n$$\nas $n\\to \\infty$ where $Z$ is a $\\mathcal{N}(0,1)$ .\n\\- Recall that if $S_{n} \\sim Bin(n,p)$ then its mean is $\\mu = np$ and standard deviation is $\\sigma = \\sqrt{np(1 - p)}$ .\n- So what this theorem says is that if you want to compute $\\mathbb{P}(c \\leq S_n \\leq d)$ then using the fact that\n$$\n\\frac {S _ {n} - n p}{\\sqrt {n p (1 - p)}} \\approx Z\n$$\nor\n$$\n\\frac {S _ {n} - \\mu}{\\sigma} \\approx Z,\n$$\nthen\n$$\n\\begin{array}{l} \\mathbb {P} \\left(c \\leq S _ {n} \\leq d\\right) = \\mathbb {P} \\left(\\frac {c - \\mu}{\\sigma} \\leq \\frac {S _ {n} - \\mu}{\\sigma} \\leq \\frac {d - \\mu}{\\sigma}\\right) \\\\ \\approx \\mathbb {P} \\left(\\frac {c - \\mu}{\\sigma} \\leq Z \\leq \\frac {d - \\mu}{\\sigma}\\right). \\\\ \\end{array}\n$$\n\\- Note that $S_{n}$ is really discrete. In fact $S_{n} \\in \\{0,1,2,\\ldots,100\\}$ , while the normal distribution is continuous!\n\\- Note that if I tried to estimate an equality: The wrong way to do it would be:\n$$\n\\begin{array}{l} \\mathbb {P} \\left(S _ {n} = i\\right) = \\mathbb {P} \\left(\\frac {S _ {n} - \\mu}{\\sigma} = \\frac {i - \\mu}{\\sigma}\\right) \\\\ \\approx \\mathbb {P} \\left(Z = \\frac {i - \\mu}{\\sigma}\\right) = 0 \\\\ \\end{array}\n$$\nas we know that for continuous random variables $X$ we always have $\\mathbb{P}(X = a) = 0!$ .\n\\- Hence we need inequalities if we want to estimate a discrete random variable using a continuous random variable.\n\\* So we use the following convention. $\\mathbb{P}(S_n = i) = \\mathbb{P}\\left(i - \\frac{1}{2} < S_n < i + \\frac{1}{2}\\right)$ .\n\\* We have no problem here, because $S_{n}$ can only be integers, so we're not hurting anything by saying “ $i - \\frac{1}{2} < S_{n} < i + \\frac{1}{2}$ ” as we know that $S_{n}$ can only be $i$ in that interval anyways.\n\\- Example: Suppose a fair coin is tossed 100 times.\n- (a) What is the probability there will be more than 60 heads?\n- $\\underline{\\text{Answer:}}$ Let $S_{100} \\sim Bin\\left(100, \\frac{1}{2}\\right)$ so that $S_{100}$ represents the numbers of heads in 100 coin tosses.\n\\* The actual answer would be\n$$\n\\begin{array}{l} \\mathbb {P} \\left(S _ {1 0 0} > 6 0\\right) = \\sum_ {i = 6 1} ^ {1 0 0} \\mathbb {P} \\left(S _ {1 0 0} = i\\right) \\\\ = \\sum_ {i = 6 1} ^ {1 0 0} \\binom{1 0 0}{i} \\left(\\frac {1}{2}\\right) ^ {i} \\left(\\frac {1}{2}\\right) ^ {1 0 0 - i} \\\\ = \\sum_ {i = 6 1} ^ {1 0 0} \\binom{1 0 0}{i} \\left(\\frac {1}{2}\\right) ^ {1 0 0}. \\\\ \\end{array}\n$$\n\\* But this would be almost impossible to do this the long way by hand.\n\\* So we will give an approximate answer using the normal distribution:\n\\- So here take $\\mu = np = 50$ and $\\sigma = \\sqrt{np(1 - p)} = \\sqrt{50\\frac{1}{2}} = 5$ . We want more than 60, so approximate using $60 + \\frac{1}{5}$ :\n$$\n\\begin{array}{l} \\mathbb {P} \\left(S _ {1 0 0} > 6 0\\right) = \\mathbb {P} \\left(S _ {1 0 0} \\geq 6 0. 5\\right) = \\mathbb {P} \\left(\\frac {S _ {1 0 0} - 5 0}{5} \\geq \\frac {6 0 . 5 - 5 0}{5}\\right) \\\\ \\approx \\mathbb {P} (Z \\geq 2. 1) \\\\ \\approx 1 - \\Phi (2. 1) \\\\ = . 0 1 7 9 \\\\ \\end{array}\n$$\n– (b) Estimate the probability of getting exactly 60 heads?\n$$\n\\begin{array}{l} \\mathbb {P} \\left(S _ {n} = 6 0\\right) = \\mathbb {P} \\left(5 9. 5 \\leq S _ {n} \\leq 6 0. 5\\right) \\\\ \\approx \\mathbb {P} (1. 9 \\leq Z \\leq 2. 1) \\\\ \\approx \\Phi (2. 1) - \\Phi (1. 9). \\\\ \\end{array}\n$$", + "path": "Notes.pdf/CHAPTER 9/Normal approximations to the binomial", + "metadata": { + "length": 3470, + "summary": "This passage explains the normal approximation to the binomial distribution. It states Theorem 13, which says that as n grows large, the standardized binomial variable (Sn - np)/sqrt(np(1-p)) converges in distribution to a standard normal Z. It notes that Sn is discrete while the normal is continuous, so a continuity correction is needed: P(Sn = i) is approximated by P(i-0.5 < Sn < i+0.5). An example with 100 fair coin tosses is given: (a) P(more than 60 heads) is approximated using 60.5, yielding about 0.0179; (b) P(exactly 60 heads) is approximated by P(59.5 ≤ Sn ≤ 60.5), giving Φ(2.1)-Φ(1.9).", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "9.1", + "The", + "normal", + "approximates", + "Binomial", + "THEOREM", + "13", + "If", + "binomial", + "parameter", + "mathbb", + "left", + "leq", + "frac", + "sqrt", + "rightarrow", + "infty", + "mathcal", + "Recall", + "sim", + "Bin", + "mu", + "np", + "standard", + "deviation", + "sigma", + "So", + "theorem", + "compute", + "fact", + "approx", + "begin", + "array", + "end", + "Note", + "discrete", + "In", + "ldots", + "100", + "distribution", + "continuous", + "estimate", + "equality", + "wrong", + "random", + "variables", + "Hence", + "inequalities", + "variable", + "convention", + "We", + "problem", + "integers", + "hurting", + "interval", + "Example", + "Suppose", + "fair", + "coin", + "tossed", + "times", + "What", + "probability", + "60", + "heads", + "underline", + "text", + "Answer", + "Let", + "represents", + "numbers", + "tosses", + "actual", + "answer", + "sum", + "binom", + "But", + "impossible", + "long", + "hand", + "give", + "approximate", + "50", + "geq", + "Phi", + "Estimate" + ], + "keywords": [ + "Normal approximation", + "Binomial", + "Central Limit Theorem" + ], + "connect_to": [] + } + }, + { + "chunk_id": "6e96aa26-0dd6-545c-a2be-f0a58135a9b7", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 10", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9e7060e2-623f-55ff-9974-745ffeaf8fcd", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3b39a1bc-8b29-5d9f-b2d0-8b4d662a2384", + "type": "image", + "content": "\nPROOF. To show this we have • Example3:(Exam P Q29)0\n[images/image-20-PROOF. To .jpg]\n", + "path": "images/image-20-PROOF. To .jpg", + "metadata": { + "length": 87, + "summary": "image-20\nPROOF. To show this we have • Example3:(Exam P Q29)0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-20-PROOF. To .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e417464e-06da-5f91-b295-00b2ee701b20", + "type": "text", + "content": "\\- A continuous R.V. is said to be exponential with parameter $\\lambda$ if its pdf is\n$$\nf (x) = \\left\\{ \\begin{array}{l l} \\lambda e ^ {- \\lambda x} & \\text { if } x \\geq 0 \\\\ 0 & \\text { if } x < 0 \\end{array} \\right.\n$$\n\\- We write $X \\sim$ exponential $(\\lambda)$ .\n\\- Summary:\n\\- CDF: Let $a > 0$ . Note that the cdf is\n$$\nF _ {X} (a) = \\mathbb {P} (X \\leq a) = \\int_ {0} ^ {a} \\lambda e ^ {- \\lambda y} d y = - e ^ {\\lambda y} | _ {0} ^ {a} = 1 - e ^ {- \\lambda a}.\n$$\n\\* Thus\n$$\n\\mathbb {P} (X > a) = 1 - \\mathbb {P} (X \\leq a) = e ^ {- \\lambda a}.\n$$\n\\- Mean: $\\mathbb{E}X = \\frac{1}{\\lambda}$ Thus $\\lambda = \\frac{1}{\\mu}$ .\n\\- Variance: We have $\\operatorname{Var}(X) = \\frac{1}{\\lambda^2}$ .\n\\- How to interpret $X$\n- $X =$ The amount of time until some specific event occurs.\n- Example:\n\\* Time until earthquake occurs\n\\* Length of a phone call\n\\* Time until an accident happens\n\\- Example1: Suppose that the length of a phone call in minutes is an exponential r.v with average length 10 minutes.\n\\- Part (a) What's probability of your phone call being more than 10 minutes?\n\\* Answer: Here $\\lambda = \\frac{1}{10}$ thus\n$$\n\\mathbb {P} (X > 1 0) = e ^ {- \\left(\\frac {1}{1 0}\\right) 1 0} = e ^ {- 1} \\approx . 3 6 8.\n$$\n\\- Part (b) Between 10 and 20 minutes?\n\\* Answer: We have that\n$$\n\\mathbb {P} (1 0 < X < 2 0) = F (2 0) - F (1 0) = e ^ {- 1} - e ^ {- 2} \\approx . 2 3 3.\n$$\n- Exponential distribution is Memoryless (Markov)\n- Example2: Suppose the life of an iphone has exponential distribution with mean life of 4 years.\n\\- Part(a): What is the probability the phone lasts more than 5 years?\n\\- $\\underline{\\text{Answer:}}$ Let $X$ denote the life of an iphone (or time until it dies). Note that $X \\sim \\overline{\\text{exponential}}\\left(\\frac{1}{4}\\right)$ since $\\lambda = \\frac{1}{\\mu} = \\frac{1}{4}$ . Then\n$$\n\\mathbb {P} (X > 5) = e ^ {- \\frac {1}{4} \\cdot 5}.\n$$\n\\- Part(b): Given that the iphone has already lasted 3 years, what is the probability that it will last another 5 more years?\n\\- Answer: We compute\n$$\n\\begin{array}{l} \\mathbb {P} (X > 5 + 3 \\mid X > 3) = \\frac {\\mathbb {P} ((X > 8) \\cap (X > 3))}{\\mathbb {P} (X > 3)} \\\\ = \\frac {\\mathbb {P} (X > 8)}{\\mathbb {P} (X > 3)} \\\\ = \\frac {e ^ {- \\frac {1}{4} \\cdot 8}}{e ^ {- \\frac {1}{4} \\cdot 3}} \\\\ = e ^ {- \\frac {1}{4} \\cdot 5}. \\\\ \\end{array}\n$$\n\\- Memoryless: Note that the probability of lasting 5 more years, is the same as if it started 5 years from anew!!\n\\- In general the memoryless property says that if $t, s > 0$ then\n$$\n\\mathbb {P} (X > t + s \\mid X > t) = \\mathbb {P} (X > s).\n$$\nTHEOREM 14. If $X$ is an exponential random variable, then $X$ is memoryless.\nPROOF. To show this we have\n$$\n\\begin{array}{l} \\mathbb {P} (X > t + s \\mid X > t) = \\frac {\\mathbb {P} ((X > t + s) \\cap (X > t))}{\\mathbb {P} (X > t)} \\\\ = \\frac {\\mathbb {P} (X > t + s)}{\\mathbb {P} (X > t)} \\\\ = \\frac {e ^ {- \\lambda (t + s)}}{e ^ {- \\lambda t}} \\\\ = e ^ {- \\lambda s} \\\\ = \\mathbb {P} (X > s), \\\\ \\end{array}\n$$\nas needed.\n\nPROOF. To show this we have • Example3:(Exam P Q29)0\n[images/image-20-PROOF. To .jpg]\n\n• Example3:(Exam P Q29)\n- The $\\#$ of days from beginning of a calendar year until accident for a BAD driver is exponentially distributed\n- An insurance company expects $30\\%$ of BAD drivers will have an accident during first 50 days.\n- Q: Whats prob that a BAD driver will have Accident during first 80 days?\n- Answer:\n- Step1: Let $X \\sim \\exp(\\lambda)$ number of days until accident. We know\n$$\n. 3 = \\mathbb {P} (X \\leq 5 0) = \\int_ {0} ^ {5 0} \\lambda e ^ {- \\lambda x} d x = - e ^ {- \\lambda t} | _ {0} ^ {5 0} = 1 - e ^ {- 5 0 \\lambda}.\n$$\n\\- Solve for $\\lambda$ and get $\\lambda = -\\frac{1}{50}\\ln .7$ .\n\\- Step2: Then compute\n$$\n\\begin{array}{l} \\mathbb {P} (X \\leq 8 0) = \\int_ {0} ^ {8 0} \\lambda e ^ {- \\lambda x} d x = - e ^ {- \\lambda t} | _ {0} ^ {8 0} = 1 - e ^ {- 8 0 \\lambda} \\\\ = 1 - e ^ {\\left(\\frac {8 0}{5 0}\\right) \\ln . 7} = . 4 3 5. \\\\ \\end{array}\n$$", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.1. Exponential Random Variables", + "metadata": { + "length": 3974, + "summary": "The exponential distribution is a continuous probability distribution with parameter λ, defined by its probability density function f(x)=λe^{-λx} for x≥0. Its cumulative distribution function is F(a)=1-e^{-λa}, mean is 1/λ, and variance is 1/λ². It models the time until a specific event occurs, such as phone call length, earthquake occurrence, or device lifetime. The distribution is memoryless, meaning P(X>t+s|X>t)=P(X>s). Examples include calculating probabilities for phone calls with average length 10 minutes, iPhone lifetime with mean 4 years, and accident occurrence for bad drivers where 30% have accidents in the first 50 days.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "continuous", + "exponential", + "parameter", + "lambda", + "pdf", + "left", + "begin", + "array", + "text", + "geq", + "end", + "We", + "write", + "sim", + "Summary", + "CDF", + "Let", + "Note", + "cdf", + "mathbb", + "leq", + "int", + "Thus", + "Mean", + "frac", + "mu", + "Variance", + "operatorname", + "Var", + "How", + "interpret", + "The", + "amount", + "time", + "specific", + "event", + "occurs", + "Example", + "Time", + "earthquake", + "Length", + "phone", + "call", + "accident", + "Example1", + "Suppose", + "length", + "minutes", + "average", + "10", + "Part", + "What", + "probability", + "Answer", + "Here", + "approx", + "Between", + "20", + "Exponential", + "distribution", + "Memoryless", + "Markov", + "Example2", + "life", + "iphone", + "years", + "lasts", + "underline", + "denote", + "dies", + "overline", + "Then", + "cdot", + "Given", + "lasted", + "compute", + "mid", + "cap", + "lasting", + "started", + "anew", + "In", + "general", + "memoryless", + "property", + "THEOREM", + "14", + "If", + "random", + "variable", + "PROOF", + "To", + "show", + "needed", + "Example3", + "Exam", + "Q29", + "days", + "beginning", + "calendar", + "year", + "BAD", + "driver", + "exponentially", + "distributed", + "An", + "insurance", + "company", + "expects", + "30", + "drivers", + "50", + "Whats", + "prob", + "Accident", + "80", + "Step1", + "exp", + "number", + "Solve", + "ln", + "Step2" + ], + "keywords": [ + "Exponential", + "Memoryless", + "Probability" + ], + "connect_to": [ + { + "target": "3b39a1bc-8b29-5d9f-b2d0-8b4d662a2384", + "relation": "embeds", + "ref": "[images/image-20-PROOF. To .jpg]", + "position": { + "start": 3065, + "end": 3097 + } + } + ] + } + }, + { + "chunk_id": "2da53499-a394-5b1e-9615-8e51bb756e44", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.2. Other Continuous Distributions", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "2f36b0f0-6363-59cf-80ee-78d29a69fb98", + "type": "text", + "content": "\\- We say $X \\sim Gamma(\\alpha, \\lambda)$ has density\n$$\nf (x) = \\left\\{ \\begin{array}{l l} \\frac {\\lambda e ^ {- \\lambda x} (\\lambda x) ^ {\\alpha - 1}}{\\Gamma (\\alpha)} & x \\geq 0 \\\\ 0 & x < 0 \\end{array} \\right.\n$$\nwhere $\\Gamma (\\alpha)$ is the Gamma function\n$$\n\\Gamma (\\alpha) = \\int_ {0} ^ {\\infty} e ^ {- y} y ^ {\\alpha - 1} d y.\n$$\n- If $Y \\sim Gamma\\left(\\frac{n}{2}, \\frac{1}{2}\\right) = \\chi_n^2$ , this is called the Chi-Squared distribution.\n- The chi-square distribution is used a lot in statistics.\n\\* Its mean is $\\mathbb{E}X = \\frac{\\alpha}{\\lambda}$ and $\\operatorname{Var}X = \\frac{\\alpha}{\\lambda^2}$ .", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.2. Other Continuous Distributions/• Gamma Distribution:", + "metadata": { + "length": 624, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "We", + "sim", + "Gamma", + "alpha", + "lambda", + "density", + "left", + "begin", + "array", + "frac", + "geq", + "end", + "function", + "int", + "infty", + "If", + "chi", + "called", + "Chi", + "Squared", + "distribution", + "The", + "square", + "lot", + "statistics", + "Its", + "mathbb", + "operatorname", + "Var" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c56bd0b1-6fad-5a4d-861d-22627c875837", + "type": "text", + "content": "- Usefull in engineering: Look in the book for its pdf.\n- $X =$ . If there is an object consisting many parts, and suppose that the object experiences death once any of its parts fails. $X =$ lifetime of this object.", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.2. Other Continuous Distributions/• Weibull Distribution:", + "metadata": { + "length": 216, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Usefull", + "engineering", + "Look", + "book", + "pdf", + "If", + "object", + "consisting", + "parts", + "suppose", + "experiences", + "death", + "fails", + "lifetime" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "50a5cb2f-f47c-5d8b-95d5-08af1588ba8b", + "type": "text", + "content": "\\- We say $X$ is cauchy with parameter $-\\infty < \\theta < \\infty$ if\n$$\nf (x) = \\frac {1}{\\pi} \\frac {1}{1 + (x - \\theta) ^ {2}}.\n$$\n- Importance: It does not have finite mean: That is $\\mathbb{E}X = \\infty$ .\n- To see this, We compute for $\\theta = 0$\n$$\n\\mathbb {E} X = \\frac {1}{\\pi} \\int_ {- \\infty} ^ {\\infty} \\frac {x}{1 + x ^ {2}} d x\n$$\n$$\n\\sim \\frac {1}{\\pi} \\int_ {- \\infty} ^ {\\infty} \\frac {1}{x} d x\n$$\n$$\n\\sim \\lim _ {x \\to \\infty} \\ln | x | - \\ln \\lim _ {x \\to - \\infty} | x |\n$$\nwhich is not defined.", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.2. Other Continuous Distributions/• Cauchy Distribution:", + "metadata": { + "length": 517, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "We", + "cauchy", + "parameter", + "infty", + "theta", + "frac", + "pi", + "Importance", + "It", + "finite", + "That", + "mathbb", + "To", + "compute", + "int", + "sim", + "lim", + "ln", + "defined" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "d6eca75d-ec0f-54b2-9409-609277af66b9", + "type": "text", + "content": "\\- Fact: For continuous R.V we have the following usefull relationship\n\\- Since $F(x) = \\mathbb{P}(X \\leq x) = \\int_{-\\infty}^{x} f(y) dy$ then by the fundament at theorem of calculus we have\n$$\nF ^ {\\prime} (x) = f (x).\n$$\n\\- Example1: If $X$ is continuous with distribution function $F_{X}$ and density function $f_{X}$ , find a formula for the density function of the random variable $Y = 2X$ .\n\\- Solution: First you start with the distribution of $Y$ :\n\\- Step1: First start by writing the cdf of $Y$ and in terms of $F_{X}$ :\n$$\nF _ {Y} (x) = \\mathbb {P} (Y \\leq x)\n$$\n$$\n= \\mathbb {P} (2 X \\leq x)\n$$\n$$\n= \\mathbb {P} \\left(X \\leq \\frac {x}{2}\\right)\n$$\n$$\n= F _ {X} \\left(\\frac {x}{2}\\right).\n$$\n\\- Step2: Then use the relation $f_{Y}(y) = F_{Y}'(y)$ and take a derivative of both sides to get\n$$\nF _ {Y} ^ {\\prime} (x) = \\frac {d}{d x} \\left[ F _ {X} \\left(\\frac {x}{2}\\right) \\right],\n$$\n$$\nF _ {Y} ^ {\\prime} (x) = F _ {X} ^ {\\prime} \\left(\\frac {x}{2}\\right) \\cdot \\left(\\frac {x}{2}\\right) ^ {\\prime}, \\text { by chain rule on RHS }\n$$\n$$\nf _ {Y} (x) = f _ {X} \\left(\\frac {x}{2}\\right) \\frac {1}{2}.\n$$\n\\- Goal: To be able to compute the cdf and pdf of $Y = g(X)$ where $g: \\mathbb{R} \\to \\mathbb{R}$ is a function given that we know the cdf and pdf of $X$ .\n\\- Why is this useful?\n\\* For example suppose $X$ represent the income for a random US worker. And let $Y = g(X)$ be the amount of taxes a US worker pays per year. Note that taxes $Y$ is dependent on the random variable $X$ . So if we only care about the random variable $Y$ then finding its PDF and CDF can help us find out everything we need to know about $Y$ given we can find the PDF. Recall that any probability and expected value can be found using the pdf.\n\\- Example2: Let $X \\sim \\text{Uniform}((0,10))$ and $Y = e^{3X}$ . Find the pdf $f_{Y}$ of $Y$ .\n\\- Solution: Recall that since $X \\sim \\text{Uniform}((0,1))$ then\n$$\nf _ {X} (x) = \\left\\{ \\begin{array}{l l} \\frac {1}{1 0} & 0 < x < 1 0 \\\\ 0 & \\text {otherwise} \\end{array} \\right..\n$$\n\\- Step1: First start by writing the cdf of $Y$ and in terms of $F_{X}$ :\n$$\n\\begin{array}{l} F _ {Y} (y) = \\mathbb {P} (Y \\leq y) \\\\ = \\mathbb {P} \\left(e ^ {3 X} \\leq y\\right), \\text {then solve for } X \\\\ = \\mathbb {P} (3 X \\leq \\ln y) \\\\ = \\mathbb {P} \\left(X \\leq \\frac {1}{3} \\ln y\\right) \\\\ = F _ {X} \\left(\\frac {1}{3} \\ln y\\right). \\\\ \\end{array}\n$$\n\\- Step2: Then use the relation $f_{Y}(y) = F_{Y}'(y)$ and take a derivative\n$$\nf _ {Y} (y) = F _ {Y} ^ {\\prime} (y)\n$$\n$$\n= \\frac {d}{d y} \\left[ F _ {X} \\left(\\frac {1}{3} \\ln y\\right) \\right], \\text {use chain rule}\n$$\n$$\n= F _ {X} ^ {\\prime} \\left(\\frac {1}{3} \\ln y\\right) \\frac {1}{3 y}\n$$\n$$\n= f _ {X} \\left(\\frac {1}{3} \\ln y\\right) \\frac {1}{3 y}, \\text { since } F _ {X} ^ {\\prime} = f _ {X}\n$$\n$$\n= \\left\\{ \\begin{array}{l l} \\frac {1}{1 0} \\cdot \\frac {1}{3 y} & 0 < \\frac {1}{3} \\ln y < 1 0 \\\\ 0 & \\text { otherwise } \\end{array} \\right.\n$$\n\\- but since\n$$\n0 < \\frac {1}{3} \\ln y < 1 0 \\quad \\Longleftrightarrow \\quad 0 < \\ln y < 3 0\n$$\n$$\n\\Longleftrightarrow e ^ {0} < y < e ^ {3 0}\n$$\n$$\n\\Longleftrightarrow 1 < y < e ^ {3 0}.\n$$\n\\- then\n$$\nf _ {Y} (y) = \\left\\{ \\begin{array}{l l} \\frac {1}{3 0 y} & 1 < y < e ^ {3 0} \\\\ 0 & \\text {otherwise} \\end{array} \\right..\n$$\n\\- Example3: Let $X \\sim \\text{Uniform}((0,1])$ and $Y = -\\ln X$ . Find the pdf of $Y$ ? What distribution is it?\n\\- Solution: Recall that\n$$\nf _ {X} (x) = \\left\\{ \\begin{array}{l l} 1 & 0 < x < 1 \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\n\\- Step1: First start with the cdf and write it terms of $F_{X}$\n$$\nF _ {Y} (x) = \\mathbb {P} (Y \\leq x)\n$$\n$$\n= \\mathbb {P} (- \\ln X \\leq x)\n$$\n$$\n= \\mathbb {P} (\\ln X > - x)\n$$\n$$\n= \\mathbb {P} (X > e ^ {- x})\n$$\n$$\n= 1 - \\mathbb {P} (X \\leq e ^ {- x})\n$$\n$$\n= 1 - F _ {X} \\left(e ^ {- x}\\right).\n$$\n\\- Step2: Then take a derivative\n$$\n\\begin{array}{l} f _ {Y} (x) = F _ {Y} ^ {\\prime} (x) \\\\ = 1 - \\frac {d}{d x} F _ {X} \\left(e ^ {- x}\\right) \\\\ = - F _ {X} ^ {\\prime} \\left(e ^ {- x}\\right) \\cdot \\left(- e ^ {- x}\\right) \\\\ = - f _ {X} \\left(e ^ {- x}\\right) \\cdot \\left(- e ^ {- x}\\right) \\\\ = f _ {X} \\left(e ^ {- x}\\right) \\cdot e ^ {- x} \\\\ = \\left\\{ \\begin{array}{l l} 1 \\cdot e ^ {- x} & 0 < e ^ {- x} < 1 \\\\ 0 & \\text { otherwise } \\end{array} \\right. \\\\ = \\left\\{ \\begin{array}{l l} e ^ {- x} & - \\infty < - x < 0 \\\\ 0 & \\text { otherwise } \\end{array} \\right. \\\\ = \\left\\{ \\begin{array}{l l} e ^ {- x} & 0 < x < \\infty \\\\ 0 & \\text { otherwise } \\end{array} \\right. \\\\ \\end{array}\n$$\n\\- Thus $Y \\sim$ exponential (1).\n\\- Example4: Suppose $X$ is uniform on $\\left(-\\frac{\\pi}{2}, \\frac{\\pi}{2}\\right)$ and $Y = \\tan X$ . Find the density of $Y$ and what known distribution is it?\n\\- Solution:\n\\- Step1: Find the cdf and write in terms of $F_{X}$\n$$\n\\begin{array}{l} F _ {Y} (x) = \\mathbb {P} (\\tan X \\leq x) \\\\ = \\mathbb {P} (X \\leq \\tan^ {- 1} x) \\\\ = F _ {X} (\\tan^ {- 1} x) \\\\ \\end{array}\n$$\n\\- Step2: Take a derivative and recall that since $\\frac{1}{\\frac{\\pi}{2} + \\frac{\\pi}{2}} = \\frac{1}{\\pi}$ then\n$$\nf _ {X} (x) = \\left\\{ \\begin{array}{l l} \\frac {1}{\\pi} & - \\frac {\\pi}{2} < x < \\frac {\\pi}{2} \\\\ 0 & \\text { otherwise. } \\end{array} \\right.\n$$\nThus\n$$\n\\begin{array}{l} f _ {Y} (x) = F _ {Y} ^ {\\prime} (x) \\\\ = \\frac {d}{d x} F _ {X} (\\tan^ {- 1} x) \\\\ = F _ {X} ^ {\\prime} \\left(\\tan^ {- 1} x\\right) \\left(\\tan^ {- 1} x\\right) ^ {\\prime} \\\\ = F _ {X} ^ {\\prime} \\left(\\tan^ {- 1} x\\right) \\frac {1}{1 + x ^ {2}} \\\\ = \\left\\{ \\begin{array}{l l} \\frac {1}{\\pi} \\frac {1}{1 + x ^ {2}} & - \\frac {\\pi}{2} < \\tan^ {- 1} x < \\frac {\\pi}{2} \\\\ 0 & \\text { otherwise. } \\end{array} \\right. \\\\ = \\left\\{ \\begin{array}{l l} \\frac {1}{\\pi} \\frac {1}{1 + x ^ {2}} & - \\infty < x < \\infty \\\\ 0 & \\text { otherwise. } \\end{array} \\right. \\\\ \\end{array}\n$$\n\\- Thus $Y$ is Cauchy(0).\n- Exercise: Show that if $Z \\sim \\mathcal{N}(0,1)$ then $Y = Z^2$ is a Gamma with parameter $\\frac{1}{2}$ and $\\frac{1}{2}$ .\n- Example5:(Actuarial Exam type question) The time, $T$ , that a manufacturing system is out of operation has cumulative distribution function\n$$\nF (t) = \\left\\{ \\begin{array}{l l} 1 - \\left(\\frac {2}{t}\\right) ^ {2} & , t > 2 \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\nThe resulting cost to the company is $Y = T^2$ . Let $f_{Y}$ be the density function for $Y$ . Determine $f_{Y}(y)$ , for $y > 4$ .\n\\- Answer:\n\\- Step1: Find the cdf of $Y$ is and\n$$\n\\begin{array}{l} F _ {Y} (y) = \\mathbb {P} \\left(T ^ {2} \\leq y\\right) \\\\ = \\mathbb {P} (T \\leq \\sqrt {y}) \\\\ = F (\\sqrt {y}) \\\\ = 1 - \\frac {4}{y} \\\\ \\end{array}\n$$\nfor $y > 4$ .\n\\- Step2: Take a derivative\n$$\n\\begin{array}{l} f _ {Y} (y) = F _ {Y} ^ {\\prime} (y) \\\\ = \\frac {4}{y ^ {2}}. \\\\ \\end{array}\n$$\n\\- One thing to note, is that we’ve been using the following useful property:\nPROPOSITION 15. Suppose $g: \\mathbb{R} \\to \\mathbb{R}$ is a strictly increasing function, then the inverse $g^{-1}$ exists and\n$$\ng (x) \\leq y \\text { implies } x \\leq g ^ {- 1} (y).\n$$", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.3. The distribution function of a Random variable", + "metadata": { + "length": 7038, + "summary": "The passage explains how to derive the probability density function (PDF) of a transformed continuous random variable Y = g(X) using the cumulative distribution function (CDF) method. It begins with the fundamental relationship F'(x)=f(x) and demonstrates through multiple examples: Y=2X, Y=e^{3X} with uniform X, Y=-ln X yielding exponential(1), Y=tan X yielding Cauchy(0), and an actuarial example with Y=T^2. The key steps are: write the CDF of Y in terms of X's CDF, then differentiate using the chain rule to obtain the PDF. The method relies on g being strictly increasing so the inverse exists.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Fact", + "For", + "continuous", + "usefull", + "relationship", + "Since", + "mathbb", + "leq", + "int", + "infty", + "dy", + "fundament", + "theorem", + "calculus", + "prime", + "Example1", + "If", + "distribution", + "function", + "density", + "find", + "formula", + "random", + "variable", + "2X", + "Solution", + "First", + "start", + "Step1", + "writing", + "cdf", + "terms", + "left", + "frac", + "Step2", + "Then", + "relation", + "derivative", + "sides", + "cdot", + "text", + "chain", + "rule", + "RHS", + "Goal", + "To", + "compute", + "pdf", + "Why", + "suppose", + "represent", + "income", + "US", + "worker", + "And", + "amount", + "taxes", + "pays", + "year", + "Note", + "dependent", + "So", + "care", + "finding", + "PDF", + "CDF", + "Recall", + "probability", + "expected", + "found", + "Example2", + "Let", + "sim", + "Uniform", + "10", + "3X", + "Find", + "begin", + "array", + "end", + "solve", + "ln", + "quad", + "Longleftrightarrow", + "Example3", + "What", + "write", + "Thus", + "exponential", + "Example4", + "Suppose", + "uniform", + "pi", + "tan", + "Take", + "recall", + "Cauchy", + "Exercise", + "Show", + "mathcal", + "Gamma", + "parameter", + "Example5", + "Actuarial", + "Exam", + "type", + "question", + "The", + "time", + "manufacturing", + "system", + "operation", + "cumulative", + "resulting", + "cost", + "company", + "Determine", + "Answer", + "sqrt", + "One", + "thing", + "note", + "ve", + "property", + "PROPOSITION", + "15", + "strictly", + "increasing", + "inverse", + "exists", + "implies" + ], + "keywords": [ + "CDF", + "PDF", + "transformation" + ], + "connect_to": [] + } + }, + { + "chunk_id": "693bebe8-3dc0-5b4e-a8a8-ca3e6f3f6ae8", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 11", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b914d5fd-dbc7-52c3-a6f4-2cd6d63a8df1", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 11/Multivariate distributions", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3db06442-0c54-564c-816e-05a3a49c15aa", + "type": "table", + "content": "
    outcome123
    1 $(X = 1, Y = 2) = (1, 2)$ $(2, 3)$ $(3, 4)$
    2 $(2, 3)$ $(2, 4)$ $(3, 5)$
    3 $(3, 4)$ $(3, 5)$ $(3, 6)$
    ", + "path": "tables/table-5 Table of Pairs.html", + "metadata": { + "length": 315, + "summary": "table-6\nThe table shows pairs (X,Y) for outcomes 1-3, with values like (1,2), (2,3), (3,4), (2,4), (3,5), and (3,6).", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-5 Table of Pairs.html", + "keywords": [ + "outcomes", + "pairs", + "values" + ], + "tokens": [] + } + }, + { + "chunk_id": "f7bae61c-dd09-52ce-a494-28cc9293a01c", + "type": "table", + "content": "
    X\\Y23456
    1 $\\mathbb{P}(X=1,Y=2)=\\frac{1}{9}$ 0000
    20 $\\frac{2}{9}$ $\\frac{1}{9}$ 00
    300 $\\frac{2}{9}$ $\\frac{2}{9}$ $\\frac{1}{9}$
    ", + "path": "tables/table-6 Joint Probability Table.html", + "metadata": { + "length": 397, + "summary": "table-7\nA joint probability table for X and Y with values 1-3 and 2-6, showing probabilities like P(X=1,Y=2)=1/9, P(X=2,Y=3)=2/9, etc.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-6 Joint Probability Table.html", + "keywords": [ + "probability", + "joint distribution", + "random variables" + ], + "tokens": [] + } + }, + { + "chunk_id": "8cee8921-358e-5d1c-9669-7b5c0c1c4ac5", + "type": "image", + "content": "\n$$
    0\n[images/image-21-Graph of Inequality.jpg]\n", + "path": "images/image-21-Graph of Inequality.jpg", + "metadata": { + "length": 48, + "summary": "image-21\n- Chart Type: Cartesian coordinate system showing a shaded region\n- X-axis range: -1 to 2.5 (major grid lines at intervals of 0.5)\n- Y-axis range: -0.5 to 1.5 (major grid lines at intervals of 0.5)\n- Shaded Region: The area where x ≥ 0 and y ≥ 0 (First Quadrant)\n- Boundary Line: Vertical line along the Y-axis (x = 0)\n- Color: Light orange shading fills the positive quadrant", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-21-Graph of Inequality.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "c8441181-ded3-5af9-9d55-9b7b63494ccb", + "type": "image", + "content": "\n\\begin{array}{l} 1 = \\int_ {0} ^ {\\infty} \\int_ {0} ^ {\\infty} c e ^ {- x} e ^ {- 2 y} d x d y = c \\int_ {0} ^ {\\infty} e ^ {- 2 y} \\left[ - e ^ {- x} \\right] _ {x = 0} ^ {x = \\infty} d y \\\\ = c \\int_ {0} ^ {\\infty} e ^ {- 2 y} d y = c \\left[ - \\frac {1}{2} e ^ {- 2 y} \\right] _ {0} ^ {\\infty} = c \\frac {1}{2}. \\\\ \\end{array}
    0\n[images/image-22-Linear Inequality Region.jpg]\n", + "path": "images/image-22-Linear Inequality Region.jpg", + "metadata": { + "length": 373, + "summary": "image-22\n- Chart Type: 2D Cartesian coordinate plot with a shaded region\n- X-axis range: -0.4 to 1.2 (major grid lines at intervals of 0.2)\n- Y-axis range: -0.2 to 1.0 (major grid lines at intervals of 0.2)\n- Shaded Region: A triangular area filled in orange\n- Boundary Lines:\n - Vertical line at x = 0 (y-axis)\n - Horizontal line at y = 1\n - Diagonal line connecting origin (0,0) to point (1,1), representing the equation y = x\n- Region Definition: The shaded area represents the inequality where x ≥ 0, y ≤ 1, and y ≥ x", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-22-Linear Inequality Region.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "97165dff-b50d-5b8a-b445-486734c346b5", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
    xy
    0.00.0
    0.20.2
    0.40.4
    0.60.6
    0.80.8
    1.01.0
    ", + "path": "tables/table-7 Linear Data Table.html", + "metadata": { + "length": 509, + "summary": "table-8\nThe table shows six pairs of x and y values from 0.0 to 1.0, where y equals x, indicating a perfect linear relationship.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-7 Linear Data Table.html", + "keywords": [ + "linear", + "data", + "table" + ], + "tokens": [] + } + }, + { + "chunk_id": "2030b2ac-81ec-5b8f-8fe5-f89550865d07", + "type": "image", + "content": "\n$$
    0\n[images/image-23-Graph of Inequality.jpg]\n", + "path": "images/image-23-Graph of Inequality.jpg", + "metadata": { + "length": 48, + "summary": "image-23\n- Chart Type: Cartesian coordinate system showing a shaded region representing an inequality.\n- X-axis: Labeled with integers from -2 to 6.\n- Y-axis: Labeled with integers from -1 to 2.\n- Shaded Region: An orange rectangular area covering the first quadrant.\n- Boundaries: The region is bounded on the left by the vertical line x = 1 and below by the horizontal line y = 0 (the x-axis).\n- Mathematical Representation: The graph visually represents the inequality $x \\ge 1$ for $y \\ge 0$.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-23-Graph of Inequality.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "c99e46c1-571b-5dd8-8514-d07d65020a6e", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n
    XY
    11
    ", + "path": "tables/table-8 Data Table.html", + "metadata": { + "length": 220, + "summary": "table-9\nThe table contains a single row with X=1 and Y=1, showing a simple data pair.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-8 Data Table.html", + "keywords": [ + "value" + ], + "tokens": [] + } + }, + { + "chunk_id": "0b844d7e-560d-52c4-9bd6-40d889832de9", + "type": "text", + "content": "- We discuss the collection of random variables $(X_{1},\\ldots ,X_{n})$ .\n- Discrete:\n\\- For random variables $X, Y$ we let $p(x, y)$ be the joint probability mass(discrete density) function\n$$\np (x, y) = \\mathbb {P} (X = x, Y = y).\n$$\n\\* Properties of joint pmf:\n· 1) $0 \\leq p \\leq 1$\n2) $\\sum_{i}\\sum_{j}p(x_{i},y_{j}) = 1$\n\\- We also have the multivariate cdf: (★★) defined by\n$$\nF _ {X, Y} (x, y) = \\mathbb {P} (X \\leq x, Y \\leq y).\n$$\n\\- Example1: Experiment: Suppose you roll two 3-sided die.\n- Let $X$ be the largest value obtained on any of the two dice. Let $Y$ be the sum of the two dice. Find the joint pmf of $X$ and $Y$ .\n- Solution: First need to find the values of $X = 1,2,3$ and $Y = 2,3,4,5,6$ .\n- The table for possible outcomes and their associated values $(X,Y)$ :\n\n[tables/table-5 Table of Pairs.html]\n\n\\- Using this table we have that the p.m.f. is given by:\n\n[tables/table-6 Joint Probability Table.html]\n\n\\- Question: Find $\\mathbb{P}(X = 2\\mid Y = 4)$ ?\n\\* Answer: $\\mathbb{P}(X = 2\\mid Y = 4) = \\frac{1 / 9}{3 / 9} = \\frac{1}{3}.$\n\\- Continuous\n\\- For random variables $X, Y$ we let $f(x, y)$ be the joint probability density function, if\n$$\n\\mathbb {P} (a \\leq X \\leq b, c \\leq Y \\leq d) = \\int_ {a} ^ {b} \\int_ {c} ^ {d} f (x, y) d y d x.\n$$\nwhich is equivalent to saying that for any set $D \\subset \\mathbb{R}^2$ then\n$$\n\\mathbb {P} \\left((X, Y) \\in D\\right) = \\int \\int_ {D} f (x, y) d A.\n$$\n\\* Properties:\n· 1) $f(x, y) \\geq 0$\n· 2) $\\int_{-\\infty}^{\\infty}\\int_{-\\infty}^{\\infty}f(x,y)dxdy=1.$\n\\- We also have the multivariate cdf: (★★) defined by\n$$\nF _ {X, Y} (x, y) = \\mathbb {P} (X \\leq x, Y \\leq y).\n$$\n\\* Note that $F_{X,Y}(a,b) = \\int_{-\\infty}^{a}\\int_{-\\infty}^{b}f(x,y)dydx$ .\n\\- Thus note that\n$$\nf (x, y) = \\frac {\\partial^ {2} F (x , y)}{\\partial x \\partial y}.\n$$\n\\- Marginal Density: If $f_{X,Y}$ is the joint density of $X, Y$ . We recover the marginal densities of $X, Y$ respectively by the following\n$$\nf _ {X} (x) = \\int_ {- \\infty} ^ {\\infty} f _ {X, Y} (x, y) d y,\n$$\n$$\nf _ {Y} (y) = \\int_ {- \\infty} ^ {\\infty} f _ {X, Y} (x, y) d x.\n$$\n• Example2: Let X, Y have joint pdf\n$$\nf (x, y) = \\left\\{ \\begin{array}{l l} c e ^ {- x} e ^ {- 2 y} & , 0 < x < \\infty , 0 < y < \\infty \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\n\\- $\\underline{\\mathbf{Part}(\\mathbf{a})}$ : Find $c$ that makes this a joint pdf:\n\\* Sol: Step1: Draw region of Domain first!!!\n\n$$
    0\n[images/image-21-Graph of Inequality.jpg]\n\n
    \nnatural_image\nBlank Cartesian coordinate grid with no plotted data or labels\n
    \n\\* Thus\n$$\n\\begin{array}{l} 1 = \\int_ {0} ^ {\\infty} \\int_ {0} ^ {\\infty} c e ^ {- x} e ^ {- 2 y} d x d y = c \\int_ {0} ^ {\\infty} e ^ {- 2 y} \\left[ - e ^ {- x} \\right] _ {x = 0} ^ {x = \\infty} d y \\\\ = c \\int_ {0} ^ {\\infty} e ^ {- 2 y} d y = c \\left[ - \\frac {1}{2} e ^ {- 2 y} \\right] _ {0} ^ {\\infty} = c \\frac {1}{2}. \\\\ \\end{array}\n$$\nThen c = 2.\n\\- $\\underline{\\text{Part(b)}}$ : Find $\\mathbb{P}(X < Y)$ .\n\\* Sol: Need to draw the region (Recall Calc III!!) Let $D = \\{(x,y) \\mid 0 < x < y, 0 < y < \\infty\\}$\n\n\\begin{array}{l} 1 = \\int_ {0} ^ {\\infty} \\int_ {0} ^ {\\infty} c e ^ {- x} e ^ {- 2 y} d x d y = c \\int_ {0} ^ {\\infty} e ^ {- 2 y} \\left[ - e ^ {- x} \\right] _ {x = 0} ^ {x = \\infty} d y \\\\ = c \\int_ {0} ^ {\\infty} e ^ {- 2 y} d y = c \\left[ - \\frac {1}{2} e ^ {- 2 y} \\right] _ {0} ^ {\\infty} = c \\frac {1}{2}. \\\\ \\end{array}
    0\n[images/image-22-Linear Inequality Region.jpg]\n\n
    \narea chart\n\n[tables/table-7 Linear Data Table.html]\n\n
    \n- There are two ways to set up this integral:\n- Method1: To set up $dA = dydx$ . We use the Top-Bottom Method:\n- Where the region is bounded by\nTop Function: $y = \\infty$\nBottom Function $y = x$\nRange of Values $0 \\leq x \\leq \\infty$\n\\- Hence we use this information to set up\n$$\n\\begin{array}{l} \\mathbb {P} (X < Y) = \\int \\int_ {D} f (x, y) d A \\\\ = \\int_ {0} ^ {\\infty} \\int_ {x} ^ {\\infty} 2 e ^ {- x} e ^ {- 2 y} d y d x \\\\ = \\int_ {0} ^ {\\infty} 2 e ^ {- x} \\frac {1}{2} \\left[ - e ^ {- 2 y} \\right] _ {y = x} ^ {y = \\infty} d x \\\\ = \\int_ {0} ^ {\\infty} e ^ {- x} e ^ {- 2 x} d x = \\int_ {0} ^ {\\infty} e ^ {- 3 x} x \\\\ = \\frac {1}{3}. \\\\ \\end{array}\n$$\n\\- Method2: To set up $dA = dxdy$ . We use the Right-Left Method:\n\\- Where the region is bounded by\nRight Function: $x = y$\nLeft Functionx = 0\nRange of Values $0 \\leq y \\leq \\infty$\n\\- Hence we use this information to set up\n$$\n\\begin{array}{l} \\mathbb {P} (X < Y) = \\int \\int_ {D} f (x, y) d A \\\\ = \\int_ {0} ^ {\\infty} \\int_ {0} ^ {y} 2 e ^ {- x} e ^ {- 2 y} d x d y \\\\ = \\quad \\text { do some work } \\\\ = \\frac {1}{3}, \\\\ \\end{array}\n$$\nwhich matches the answer from before.\n\\- $\\underline{\\mathbf{Part}(\\mathbf{c})}$ : Set up $\\mathbb{P}(X > 1, Y < 1)$\n\n$$
    0\n[images/image-23-Graph of Inequality.jpg]\n\n
    \nbar chart\n\n[tables/table-8 Data Table.html]\n\n
    \n\\* The region is given by\n\\* Setting this up we have\n$$\n\\mathbb {P} (X > 1, Y < 1) = \\int_ {0} ^ {1} \\int_ {1} ^ {\\infty} 2 e ^ {- x} e ^ {- 2 y} d x d y.\n$$\n\\- $\\underline{\\mathbf{Part}(\\mathbf{d})}$ : Find the marginal $f_{X}(x)$ :\n\\* Sol:\n\\* Then\n$$\n\\begin{array}{l} f _ {X} (x) = \\int_ {0} ^ {\\infty} f (x, y) d y = \\int_ {0} ^ {\\infty} 2 e ^ {- x} e ^ {- 2 y} d y \\\\ = 2 e ^ {- x} \\left[ \\frac {- 1}{2} e ^ {- 2 y} \\right] _ {0} ^ {\\infty} = 2 e ^ {- x} \\left[ 0 + \\frac {1}{2} \\right] \\\\ = e ^ {- x}. \\\\ \\end{array}\n$$\n\\- Part(e): Find $\\mathbb{E}X$ We have\n$$\n\\mathbb {E} X = \\int_ {0} ^ {\\infty} x e ^ {- x} d x = 1\n$$", + "path": "Notes.pdf/CHAPTER 11/Multivariate distributions/11.1. Joint distribution functions", + "metadata": { + "length": 5505, + "summary": "The passage discusses joint probability distributions for discrete and continuous random variables. For discrete variables, it defines the joint probability mass function (pmf) p(x,y) = P(X=x, Y=y) with properties 0≤p≤1 and sum=1, and the multivariate cdf F(x,y)=P(X≤x,Y≤y). An example with two 3-sided dice finds the joint pmf of X (largest value) and Y (sum), and computes a conditional probability. For continuous variables, it defines the joint probability density function (pdf) f(x,y) with properties f≥0 and double integral=1, and the cdf as the double integral of f. Marginal densities are obtained by integrating out the other variable. A detailed example with joint pdf f(x,y)=2e^{-x}e^{-2y} for x,y>0 finds the normalizing constant c=2, computes P(X1,Y<1), finds marginal f_X(x)=e^{-x}, and computes E[X]=1.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "We", + "discuss", + "collection", + "random", + "variables", + "ldots", + "Discrete", + "For", + "joint", + "probability", + "mass", + "discrete", + "density", + "function", + "mathbb", + "Properties", + "pmf", + "leq", + "sum", + "multivariate", + "cdf", + "defined", + "Example1", + "Experiment", + "Suppose", + "roll", + "sided", + "die", + "Let", + "largest", + "obtained", + "dice", + "Find", + "Solution", + "First", + "find", + "values", + "The", + "table", + "outcomes", + "Using", + "Question", + "mid", + "Answer", + "frac", + "Continuous", + "int", + "equivalent", + "set", + "subset", + "left", + "geq", + "infty", + "dxdy", + "Note", + "dydx", + "Thus", + "note", + "partial", + "Marginal", + "Density", + "If", + "recover", + "marginal", + "densities", + "Example2", + "pdf", + "begin", + "array", + "text", + "end", + "underline", + "mathbf", + "Part", + "makes", + "Sol", + "Step1", + "Draw", + "region", + "Domain", + "details", + "summary", + "natural", + "image", + "Blank", + "Cartesian", + "coordinate", + "grid", + "plotted", + "data", + "labels", + "Then", + "Need", + "draw", + "Recall", + "Calc", + "III", + "area", + "chart", + "There", + "ways", + "integral", + "Method1", + "To", + "dA", + "Top", + "Bottom", + "Method", + "Where", + "bounded", + "Function", + "Range", + "Values", + "Hence", + "information", + "Method2", + "Right", + "Left", + "Functionx", + "quad", + "work", + "matches", + "answer", + "Set", + "bar", + "Setting" + ], + "keywords": [ + "joint pmf", + "joint pdf", + "marginal density" + ], + "connect_to": [ + { + "target": "3db06442-0c54-564c-816e-05a3a49c15aa", + "relation": "embeds", + "ref": "[tables/table-5 Table of Pairs.html]", + "position": { + "start": 788, + "end": 824 + } + }, + { + "target": "f7bae61c-dd09-52ce-a494-28cc9293a01c", + "relation": "embeds", + "ref": "[tables/table-6 Joint Probability Table.html]", + "position": { + "start": 884, + "end": 929 + } + }, + { + "target": "8cee8921-358e-5d1c-9669-7b5c0c1c4ac5", + "relation": "embeds", + "ref": "[images/image-21-Graph of Inequality.jpg]", + "position": { + "start": 2435, + "end": 2476 + } + }, + { + "target": "c8441181-ded3-5af9-9d55-9b7b63494ccb", + "relation": "embeds", + "ref": "[images/image-22-Linear Inequality Region.jpg]", + "position": { + "start": 3453, + "end": 3499 + } + }, + { + "target": "97165dff-b50d-5b8a-b445-486734c346b5", + "relation": "embeds", + "ref": "[tables/table-7 Linear Data Table.html]", + "position": { + "start": 3542, + "end": 3581 + } + }, + { + "target": "2030b2ac-81ec-5b8f-8fe5-f89550865d07", + "relation": "embeds", + "ref": "[images/image-23-Graph of Inequality.jpg]", + "position": { + "start": 4835, + "end": 4876 + } + }, + { + "target": "c99e46c1-571b-5dd8-8514-d07d65020a6e", + "relation": "embeds", + "ref": "[tables/table-8 Data Table.html]", + "position": { + "start": 4918, + "end": 4950 + } + } + ] + } + }, + { + "chunk_id": "4971827f-b924-59b2-a999-b254b0d71cda", + "type": "image", + "content": "\n$$ \\- Example1: Let $X, Y$ be r.v. with joint pdf0\n[images/image-24-$$ _- Exam.jpg]\n", + "path": "images/image-24-$$ _- Exam.jpg", + "metadata": { + "length": 85, + "summary": "image-24\n$$ \\- Example1: Let $X, Y$ be r.v. with joint pdf0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-24-$$ _- Exam.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "c525b3f0-fd7d-5540-8b8c-60997eebf13b", + "type": "image", + "content": "\n\\- Solution: Since $X, Y$ are independent then using the Theorem form this section we have
    0\n[images/image-25-Linear Function Graph.jpg]\n", + "path": "images/image-25-Linear Function Graph.jpg", + "metadata": { + "length": 136, + "summary": "image-25\nChart Title: None visible\nAxes: X-axis ranges from -0.2 to 1.2; Y-axis ranges from -0.2 to 1.0\nData Representation: A linear function plotted as a straight line starting at the origin (0,0) and ending at approximately (1,1).\nShaded Region: The area under the curve (between the line and the x-axis) is shaded in orange.\nGrid: Background grid lines are present for both axes with increments of 0.2.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-25-Linear Function Graph.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "75d5bd34-e5fc-5ced-af4c-8a414c8b8f2c", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
    xy
    00
    0.20.2
    0.40.4
    0.60.6
    0.80.8
    11
    ", + "path": "tables/table-9 Linear Data Table.html", + "metadata": { + "length": 501, + "summary": "table-10\nThe table shows six data points where x and y values are equal, ranging from 0 to 1, indicating a perfect linear relationship.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-9 Linear Data Table.html", + "keywords": [ + "linear" + ], + "tokens": [] + } + }, + { + "chunk_id": "d56fd4b6-7847-559f-a64b-72af7530b4bd", + "type": "text", + "content": "\\- Discrete: We say discrete r.v. $X, Y$ are independent if\n$$\n\\mathbb {P} (X = x, Y = y) = \\mathbb {P} (X = x) \\mathbb {P} (Y = y),\n$$\nfor every $x, y$ in the range of $X$ and $Y$ .\n\\- This is the same as saying that $X, Y$ are independent if the joint pmf splits into the marginal pmfs: $p_{X,Y}(x,y) = p_X(x) \\cdot p_Y(y)$\n\\- Continuous: We say continuous r.v. $X, Y$ are independent if\n$$\n\\mathbb {P} (X \\in A, Y \\in B) = \\mathbb {P} (X \\in A) \\mathbb {P} (Y \\in B)\n$$\nfor any set $A, B$\n\\- This equivalent: $\\mathbb{P}(X\\leq a,Y\\leq b) = \\mathbb{P}(X\\leq a)\\mathbb{P}(Y\\leq b)$ .\n\\- Equivalent to $F_{X,Y}(x,y) = F_X(x)F_Y(y)$ .\n\\- Random variables that are not independent, are said to be dependent.\n• How can we check independence?\nTHEOREM 16. Continuous (discrete) r.v. $X, Y$ are independent if and only if their joint pdf (pmf) can be expressed as\n$$\nf _ {X, Y} (x, y) = f _ {X} (x) f _ {Y} (y). \\text {(Continuous Case)},\n$$\n$$\np _ {X, Y} (x, y) = p _ {X} (x) p _ {Y} (y) \\text {(Discrete Case)}.\n$$\nPROOF. See textbook.\n\n$$ \\- Example1: Let $X, Y$ be r.v. with joint pdf0\n[images/image-24-$$ _- Exam.jpg]\n\n\\- Example1: Let $X, Y$ be r.v. with joint pdf\n$$\nf (x, y) = 6 e ^ {- 2 x} e ^ {- 3 y} 0 < x < \\infty , 0 < y < \\infty .\n$$\nAre $X, Y$ independent?\n\\- Solution: Find the marginals $f_{X}$ and $f_{Y}$ and see if $f = f_{X}f_{Y}$ . First\n$$\nf _ {X} (x) = \\int_ {0} ^ {\\infty} 6 e ^ {- 2 x} e ^ {- 3 y} d y = 2 e ^ {- 2 x},\n$$\n$$\nf _ {Y} (y) = \\int_ {0} ^ {\\infty} 6 e ^ {- 2 x} e ^ {- 3 y} d x = 3 e ^ {- 2 y}.\n$$\nwhich are both exponential. Since $f = f_{X}f_{y}$ then yes they are independent!\n• Example2: Let X, Y have\n$$\nf _ {X, Y} (x, y) = x + y, 0 < x < 1, 0 < y < 1\n$$\nAre $X, Y$ independent?\n\\- Solution: Note that there is no way to factor $x + y = f_X(x)f_Y(y)$ , hence they can't be independent.\n• Example3: Let X, Y have\n$$\nf _ {X, Y} (x, y) = 2, 0 < x < y < 1\n$$\n\\- Are $X, Y$ independent?\n\\- Solution:\n\\* We cannot use the previous argument to claim $f_{X,Y}$ can't split, because for example, maybe hypothetically speaking $2 = 1 \\cdot 1$ , so hypothetically it could split.\n\\* So we must find the marginal pdfs and then check if $f_{X,Y} = f_X \\cdot f_Y$ .\n\\* Important! But whenever the domain of $f$ is not a rectangle, you MUST draw the region of domain for $f_{X,Y}$ . And here the region is $D = \\{(x,y) \\mid 0 < x < y < 1\\}$ . (Please try drawing this region on your own. If you struggle with this region, go to https://www.wolframalpha.com/ and type in $0 < x < y < 1$ )\n\\* Note that $f_{X}(x) = \\int_{x}^{1} 2dy = 2(1 - x)$ for $0 < x < 1$\n\\* Then $f_{Y}(y) = \\int_{0}^{y} 2dx = 2y$ for $0 < y < 1$ .\n\\* But $f_{X,Y}(x,y) = 2 \\neq f_X(x)f_Y(y) = 2(1 - x)2y!!$ Therefore $X,Y$ are NOT independent.\n\\- Example4: Suppose $X, Y$ are independent uniformly distributed over (0, 1). Find $\\mathbb{P}(Y < X)$ .\n\\- Solution: Since $X, Y$ are independent then using the Theorem form this section we have\n$$\nf _ {X, Y} (x, y) = f _ {X} (x) f _ {Y} (y) = 1 \\cdot 1,\n$$\nfor $0 \\leq x \\leq 1$ and $0 \\leq y \\leq 1$ . Draw region ('What do you think probability will be by looking at the region?)\n\n\\- Solution: Since $X, Y$ are independent then using the Theorem form this section we have
    0\n[images/image-25-Linear Function Graph.jpg]\n\n
    \narea chart\n\n[tables/table-9 Linear Data Table.html]\n\n
    \n\\* and get\n$$\n\\begin{array}{l} \\mathbb {P} (Y < X) = \\int_ {0} ^ {1} \\int_ {0} ^ {x} f (x, y) d y d x \\\\ = \\int_ {0} ^ {1} \\int_ {0} ^ {x} 1 d y d x = \\int_ {0} ^ {1} x d x \\\\ = \\frac {1}{2}. \\\\ \\end{array}\n$$\n11.3. Sums of independent Random Variables( $\\star$ )\n\\- Fact: If $X, Y$ are independent, its not too hard to show that the cdf of $Z = X + Y$ is\n$$\n\\begin{array}{l} F _ {X + Y} (a) = \\mathbb {P} (X + Y \\leq a) \\\\ = \\int \\int_ {\\{x + y \\leq a \\}} f _ {X} (x) f _ {Y} (y) d x d y \\\\ = \\int_ {- \\infty} ^ {\\infty} \\int_ {- \\infty} ^ {a - y} f _ {X} (x) f _ {Y} (y) d x d y \\\\ = \\int_ {- \\infty} ^ {\\infty} \\int_ {- \\infty} ^ {a - y} f _ {X} (x) d x f _ {Y} (y) d y \\\\ = \\int_ {- \\infty} ^ {\\infty} F _ {X} (a - y) f _ {Y} (y) d y. \\\\ \\end{array}\n$$\n\\- By differentiating we have that\n$$\nf _ {X + Y} (a) = \\int_ {- \\infty} ^ {\\infty} f _ {X} (a - y) f _ {Y} (y) d y.\n$$\n• (★)Here are some interesting cases:\n\\- Fact 1(Only thing I'll test you on): If $X_{i} \\sim \\mathcal{N}\\left(\\mu_{i}, \\sigma_{i}^{2}\\right)$ for $1 \\leq i \\leq n$ and are all independent then $Y = X_{1} + \\dots + X_{n} \\sim \\mathcal{N}\\left(\\mu_{1} + \\dots \\mu_{n}, \\sigma_{1}^{2} + \\dots + \\sigma_{n}^{2}\\right)$ .\n- In particular if $X \\sim \\mathcal{N}\\left(\\mu_x, \\sigma_x^2\\right)$ and $Y \\sim \\mathcal{N}(\\mu_y, \\sigma_y^2)$ are independent then $X + Y \\sim \\mathcal{N}\\left(\\mu_x + \\mu_y, \\sigma_x^2 + \\sigma_y^2\\right)$ and $X - Y \\sim \\mathcal{N}\\left(\\mu_x - \\mu_y, \\sigma_x^2 + \\sigma_y^2\\right)$ .\n- In general $aX \\pm bY \\sim \\mathcal{N}\\left(a\\mu_x \\pm b\\mu_y, a^2\\sigma_x^2 + b^2\\sigma_y^2\\right)$ .\n\\- Example1: Suppose $T \\sim \\mathcal{N}(95,25)$ and $H \\sim \\mathcal{N}(65,36)$ represents the grades of Tyler and Habib. Assume their grades are independent.\n\\- Part(a): What is the probability that their average grades will be less than 90?\n\\- Solution: $T + H \\sim \\mathcal{N}(160, 61)$ . Thus\n$$\n\\begin{array}{l} \\mathbb {P} \\left(\\frac {T + H}{2} \\leq 9 0\\right) = \\mathbb {P} (T + H \\leq 1 8 0) \\\\ = \\mathbb {P} \\left(Z \\leq \\frac {1 8 0 - 1 6 0}{\\sqrt {6 1}}\\right) = \\Phi \\left(\\frac {1 8 0 - 1 6 0}{\\sqrt {6 1}}\\right) \\\\ = \\Phi (2. 5 6) = . 9 9 6 1 \\\\ \\end{array}\n$$\n\\- Part (b): What is the probability that Habib will have scored higher than Tyler?\n\\- Solution: Using $H - T \\sim \\mathcal{N}(-30,61)$ we compute\n$$\n\\begin{array}{l} \\mathbb {P} (H > T) = \\mathbb {P} (H - T > 0) \\\\ = 1 - \\mathbb {P} (H - T < 0) \\\\ = 1 - \\mathbb {P} \\left(Z \\leq \\frac {0 - (- 3 0)}{\\sqrt {6 1}}\\right) \\\\ = 1 - \\Phi (3. 8 4) = 1 - 1 = 0. \\\\ \\end{array}\n$$\n\\- Other facts.\n\\- Fact 2: Let $Z \\sim \\mathcal{N}(0,1)$ then $Z^2 \\sim \\chi_1^2$ .\n- If $Z_1, \\ldots, Z_n$ are independent $\\mathcal{N}(0,1)$ then $Y = Z_1^2 + \\cdots + Z_n^2 \\sim \\chi_n^2$ .\n- Fact 3: If $X \\sim \\text{Poisson}(\\lambda)$ and $Y \\sim \\text{Poisson}(\\mu)$ , and they are independent, then $X + Y \\sim \\overline{\\text{Poisson}}(\\lambda + \\mu)$ .\n- List out stuff and then stop.", + "path": "Notes.pdf/CHAPTER 11/Multivariate distributions/11.2. Independent Random Variables", + "metadata": { + "length": 6298, + "summary": "This text defines independence for discrete and continuous random variables (RVs) via joint pmf/pdf factorization and joint CDF. It provides examples checking independence by verifying if joint pdf equals product of marginals, highlighting cases where domain restrictions (non-rectangular) prevent factorization. It then covers sums of independent RVs, deriving the convolution formula for pdf/cdf. Key results include: sum of independent normals is normal with summed means and variances; sum of independent Poissons is Poisson with summed rates; sum of squared standard normals follows chi-squared distribution. Examples compute probabilities for average grades and difference of grades using normal distribution properties.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Discrete", + "We", + "discrete", + "independent", + "mathbb", + "range", + "This", + "joint", + "pmf", + "splits", + "marginal", + "pmfs", + "cdot", + "Continuous", + "continuous", + "set", + "equivalent", + "leq", + "Equivalent", + "Random", + "variables", + "dependent", + "How", + "check", + "independence", + "THEOREM", + "16", + "pdf", + "expressed", + "text", + "Case", + "PROOF", + "See", + "textbook", + "Example1", + "Let", + "pdf0", + "infty", + "Are", + "Solution", + "Find", + "marginals", + "First", + "int", + "exponential", + "Since", + "Example2", + "Note", + "factor", + "Example3", + "previous", + "argument", + "claim", + "split", + "hypothetically", + "speaking", + "So", + "find", + "pdfs", + "Important", + "But", + "domain", + "rectangle", + "MUST", + "draw", + "region", + "And", + "mid", + "Please", + "drawing", + "If", + "struggle", + "https", + "www", + "wolframalpha", + "type", + "2dy", + "Then", + "2dx", + "2y", + "neq", + "Therefore", + "NOT", + "Example4", + "Suppose", + "uniformly", + "distributed", + "Theorem", + "form", + "section", + "Draw", + "What", + "probability", + "details", + "summary", + "area", + "chart", + "begin", + "array", + "frac", + "end", + "11.3", + "Sums", + "Variables", + "star", + "Fact", + "hard", + "show", + "cdf", + "By", + "differentiating", + "Here", + "interesting", + "cases", + "Only", + "thing", + "ll", + "test", + "sim", + "mathcal", + "left", + "mu", + "sigma", + "dots", + "In", + "general", + "aX", + "pm", + "bY", + "95", + "25", + "65", + "36", + "represents", + "grades", + "Tyler", + "Habib", + "Assume", + "Part", + "average", + "90", + "160", + "61", + "Thus", + "sqrt", + "Phi", + "scored", + "higher", + "Using", + "30", + "compute", + "Other", + "facts", + "chi", + "ldots", + "cdots", + "Poisson", + "lambda", + "overline", + "List", + "stuff", + "stop" + ], + "keywords": [ + "Independence", + "Joint PDF", + "Normal Sum" + ], + "connect_to": [ + { + "target": "4971827f-b924-59b2-a999-b254b0d71cda", + "relation": "embeds", + "ref": "[images/image-24-$$ _- Exam.jpg]", + "position": { + "start": 1084, + "end": 1116 + } + }, + { + "target": "c525b3f0-fd7d-5540-8b8c-60997eebf13b", + "relation": "embeds", + "ref": "[images/image-25-Linear Function Graph.jpg]", + "position": { + "start": 3247, + "end": 3290 + } + }, + { + "target": "75d5bd34-e5fc-5ced-af4c-8a414c8b8f2c", + "relation": "embeds", + "ref": "[tables/table-9 Linear Data Table.html]", + "position": { + "start": 3333, + "end": 3372 + } + } + ] + } + }, + { + "chunk_id": "4d24fdda-7a90-5453-a68e-fc807a0040ed", + "type": "table", + "content": "
    x\\y01
    0.4.2
    1.1.3
    ", + "path": "tables/table-10 Joint Probability.html", + "metadata": { + "length": 138, + "summary": "table-11\nThe table shows joint probabilities for two binary variables X and Y, with values: P(0,0)=0.4, P(0,1)=0.2, P(1,0)=0.1, P(1,1)=0.3.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-10 Joint Probability.html", + "keywords": [ + "probability", + "distribution", + "table" + ], + "tokens": [] + } + }, + { + "chunk_id": "12002c24-537b-5130-ad36-8dc97eea6d3f", + "type": "text", + "content": "\\- The conditional pmf for a discrete R.V. is\n$$\np _ {X \\mid Y} (x \\mid y) = \\mathbb {P} (X = x \\mid Y = y)\n$$\n$$\n= \\frac {p (x , y)}{p _ {Y} (y)}.\n$$\n\\- We also have the condition cdf: $F_{X|Y}(x \\mid y) = \\mathbb{P}(X \\leq x \\mid Y = u)$\n\\- Fact:\n\\- If $X, Y$ are independent then\n$$\np _ {X | Y} (x \\mid y) = p _ {X} (x)\n$$\n\\- Example1: Suppose the joint pmf of $(X,Y)$ is\n\n[tables/table-10 Joint Probability.html]\n\n\\- Compute some conditional pmf: Then the second column is\n$$\np _ {X | Y} (0 \\mid 1) = \\frac {. 2}{. 5} = \\frac {2}{5} \\text { and } p _ {X | Y} (1 \\mid 1) = \\frac {. 3}{. 5} = \\frac {3}{5}.\n$$\n\\- Are they independent? Note that $p_X(0) = .4 + .2 = .6 \\neq p_{X|Y}(0 \\mid 1)$ , so no!\n11.5. Conditional Distributions- Continuous( $\\star$ )\n\\- Def: If $X, Y$ are continuous with joint pdf $f(x, y)$ then the conditional pdf of $X$ given $Y = y$ is defined as\n$$\nf _ {X \\mid Y} (x \\mid y) = \\frac {f (x , y)}{f _ {Y} (y)}.\n$$\n\\- defined only when $f_{Y}(y) > 0$ .\n\\- Def: The conditional cdf of $X$ given $Y = y$ is\n$$\n\\begin{array}{l} F _ {X \\mid Y} (a \\mid y) = \\mathbb {P} (X \\leq a \\mid Y = y) \\\\ = \\int_ {- \\infty} ^ {a} f _ {X | Y} (x \\mid y) d x. \\\\ \\end{array}\n$$\n• Fact: If X, Y are independent then\n$$\nf _ {X \\mid Y} (x \\mid y) = f _ {X} (x).\n$$\n\\- Example1: The joint pdf of $X, Y$ is given by\n$$\nf x (x, y) = \\left\\{ \\begin{array}{l l} \\frac {1 2}{5} x (2 - x - y) & 0 < x < 1, 0 < y < 1 \\\\ 0 & \\text {otherwise} \\end{array} \\right..\n$$\nCompute the conditional pdf of X give that Y = y where 0 < y < 1.\n\\- Solution: We have\n$$\n\\begin{array}{l} f _ {X \\mid Y} (x \\mid y) = \\frac {f (x , y)}{f _ {Y} (y)} = \\frac {x (2 - x - y)}{\\int_ {0} ^ {1} x (2 - x - y) d y} \\\\ = \\frac {x (2 - x - y)}{\\frac {2}{3} - \\frac {y}{2}}. \\\\ \\end{array}\n$$", + "path": "Notes.pdf/CHAPTER 11/Multivariate distributions/11.4. Conditional Distributions- Discrete(★)", + "metadata": { + "length": 1758, + "summary": "The text defines conditional probability mass functions (PMF) for discrete random variables and conditional probability density functions (PDF) for continuous random variables. It provides formulas: p_{X|Y}(x|y) = p(x,y)/p_Y(y) for discrete, and f_{X|Y}(x|y) = f(x,y)/f_Y(y) for continuous, defined only when the marginal is positive. It notes that if X and Y are independent, the conditional distribution equals the marginal. Examples include computing conditional PMF from a joint table and checking independence, and computing conditional PDF from a given joint PDF.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "The", + "conditional", + "pmf", + "discrete", + "mid", + "mathbb", + "frac", + "We", + "condition", + "cdf", + "leq", + "Fact", + "If", + "independent", + "Example1", + "Suppose", + "joint", + "Compute", + "Then", + "column", + "text", + "Are", + "Note", + "neq", + "11.5", + "Conditional", + "Distributions", + "Continuous", + "star", + "Def", + "continuous", + "pdf", + "defined", + "begin", + "array", + "int", + "infty", + "end", + "left", + "give", + "Solution" + ], + "keywords": [ + "Conditional PMF", + "Conditional PDF", + "Independence" + ], + "connect_to": [ + { + "target": "4d24fdda-7a90-5453-a68e-fc807a0040ed", + "relation": "embeds", + "ref": "[tables/table-10 Joint Probability.html]", + "position": { + "start": 376, + "end": 416 + } + } + ] + } + }, + { + "chunk_id": "85c24b5e-a968-500b-97ef-51d8be32212b", + "type": "text", + "content": "- Goal:\n- Recall that from section 5.7 we can find the pdf of a new random variable $Y = g(X)$ .\n- Suppose we know the distributions of $X_{1}, X_{2}$ then what is the distribution of $g_{1}(X_{1}, X_{2})$ and $g_{2}(X_{1}, Y_{1})$\n\\* For example if we know $X_{1}, X_{2}$ what is the distribution of $Y_{1} = X_{1} + X_{2}$ and $Y_{2} = X_{1}^{2} - e^{X_{1}X_{2}}$ .\n\\- Steps to finding the joint cdf of new R.V. made from old ones.:\n- Suppose $X_{1}, X_{2}$ are jointly distributed with pdf $f_{X_1, X_2}$ . Let $g_{2}(x_{1}, x_{2}), g_{2}(x_{2}, x_{2})$ be multivariable functions.\n- Goal: Find the joint pdf of $Y_{1} = g_{1}(X_{1}, X_{2})$ and $Y_{2} = g_{1}(X_{2}, X_{2})$\n- Step1: Find the Jacobian:\n$$\nJ \\left(x _ {1}, x _ {2}\\right) = \\left| \\begin{array}{c} \\nabla g _ {1} \\\\ \\nabla g _ {2} \\end{array} \\right| = \\left| \\begin{array}{c c} \\frac {\\partial g _ {1}}{\\partial x _ {1}} & \\frac {\\partial g _ {1}}{\\partial x _ {2}} \\\\ \\frac {\\partial g _ {2}}{\\partial x _ {1}} & \\frac {\\partial g _ {2}}{\\partial x _ {2}} \\end{array} \\right| = \\frac {\\partial g _ {1}}{\\partial x _ {1}} \\frac {\\partial g _ {2}}{\\partial x _ {2}} - \\frac {\\partial g _ {1}}{\\partial x _ {2}} \\frac {\\partial g _ {2}}{\\partial x _ {1}} \\neq 0.\n$$\nat all points $(x_{1},x_{2})$\n\\- Step2: Find the unique solutions of equationf $y_{1} = g_{1}(x_{1}, x_{2})$ and $y_{2} = g_{2}(x_{1}, x_{2})$ in terms of\n$$\nx _ {1} = h _ {1} \\left(y _ {1}, y _ {2}\\right),\n$$\n$$\nx _ {2} = h _ {2} \\left(y _ {1}, y _ {2}\\right).\n$$\n\\- Step3: The joint pdf of $Y_{1}, Y_{2}$ is\n$$\n\\begin{array}{l} f _ {Y _ {1}, Y _ {2}} \\left(y _ {1}, y _ {2}\\right) = f _ {X _ {1}, X _ {2}} \\left(x _ {1}, x _ {2}\\right) | J \\left(x _ {1}, x _ {2}\\right) | ^ {- 1} \\\\ = f _ {X _ {1}, X _ {2}} \\left(h _ {1} \\left(y _ {1}, y _ {2}\\right), h _ {2} \\left(y _ {1}, y _ {2}\\right)\\right) | J \\left(x _ {1}, x _ {2}\\right) | ^ {- 1}. \\\\ \\end{array}\n$$\n\\- Example1: Suppose $X_{1}, X_{2}$ have joint distribution\n$$\nf _ {X _ {1}, X _ {2}} (x _ {1}, x _ {2}) = \\left\\{ \\begin{array}{l l} 2 x _ {1} x _ {2} & 0 \\leq x _ {1}, x _ {2} \\leq 1 \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\nQuestion: Find the joint pdf of $Y_{1} = X_{1} + X_{2}$ and $Y_{2} = X_{1} - X_{2}$ .\n\\- Step1: Find the Jacobian: Note that\n$$\ny _ {1} = g _ {1} \\left(x _ {1}, x _ {2}\\right) = x _ {1} + x _ {2},\n$$\n$$\ny _ {2} = g _ {2} \\left(x _ {1}, x _ {2}\\right) = x _ {1} - x _ {2}.\n$$\nSo\n$$\nJ \\left(x _ {1}, x _ {2}\\right) = \\left| \\begin{array}{c c} 1 & 1 \\\\ 1 & - 1 \\end{array} \\right| = - 2.\n$$\n\\- Step2: Solve for $x_{1}, x_{2}$ and get\n$$\nx _ {1} = \\frac {1}{2} \\left(y _ {1} + y _ {2}\\right),\n$$\n$$\nx _ {2} = \\frac {1}{2} (y _ {1} - y _ {2}).\n$$\n\\- Step3: The joint pdf of $Y_{1}, Y_{2}$ is given by the formula:\n$$\n\\begin{array}{l} f _ {Y _ {1}, Y _ {2}} \\left(y _ {1}, y _ {2}\\right) = f _ {X _ {1}, X _ {2}} \\left(x _ {1}, x _ {2}\\right) | J \\left(x _ {1}, x _ {2}\\right) | ^ {- 1} \\\\ = f _ {X _ {1}, X _ {2}} \\left(\\frac {1}{2} \\left(y _ {1} + y _ {2}\\right), \\frac {1}{2} \\left(y _ {1} - y _ {2}\\right)\\right) \\frac {1}{| - 2 |} \\\\ = \\left\\{ \\begin{array}{l l} \\frac {1}{2} \\left(y _ {1} + y _ {2}\\right) \\left(y _ {1} - y _ {2}\\right) & 0 \\leq \\frac {1}{2} \\left(y _ {1} + y _ {2}\\right) \\leq 1, \\\\ & 0 \\leq \\frac {1}{2} \\left(y _ {1} - y _ {2}\\right) \\leq 1 \\\\ 0 & \\text {otherwise} \\end{array} \\right. \\\\ \\end{array}\n$$\n\\- Example2: Suppose $X_{1} \\sim \\mathcal{N}(0,1)$ and $X_{2} \\sim \\mathcal{N}(0,4)$ and independent.\n\\- Let $Y_{1} = 2X_{1} + X_{2}$ and $Y_{2} = X_{1} - 3X_{2}$ .\n\\- Question: Find the joint pdf $f_{Y_1, Y_2}(y_1, y_2)$ of $Y_1$ and $Y_2$ .\n\\- Step1: Find the Jacobian: Note that\n$$\ny _ {1} = g _ {1} \\left(x _ {1}, x _ {2}\\right) = 2 x _ {1} + x _ {2},\n$$\n$$\ny _ {2} = g _ {2} \\left(x _ {1}, x _ {2}\\right) = x _ {1} - 3 x _ {2}.\n$$\nSo\n$$\nJ \\left(x _ {1}, x _ {2}\\right) = \\left| \\begin{array}{c c} 2 & 1 \\\\ 1 & - 3 \\end{array} \\right| = - 7.\n$$\n\\- Step2: Solve for $x_{1}, x_{2}$ and get\n$$\nx _ {1} = \\frac {3}{7} y _ {1} + \\frac {1}{y} y _ {2}\n$$\n$$\nx _ {2} = \\frac {1}{7} y _ {1} - \\frac {2}{7} y _ {2}\n$$\n\\- Step3: The joint pdf of $Y_{1}, Y_{2}$ is given by the formula:\n$$\n\\begin{array}{l} f _ {Y _ {1}, Y _ {2}} \\left(y _ {1}, y _ {2}\\right) = f _ {X _ {1}, X _ {2}} \\left(x _ {1}, x _ {2}\\right) \\left| J \\left(x _ {1}, x _ {2}\\right) \\right| ^ {- 1} \\\\ = f _ {X _ {1}, X _ {2}} \\left(\\frac {3}{7} y _ {1} + \\frac {1}{y} y _ {2}, \\frac {1}{7} y _ {1} - \\frac {2}{7} y _ {2}\\right) \\frac {1}{7}. \\\\ \\end{array}\n$$\nSo we need to find the joint pdf of $X_{1}$ and $X_{2}$ .\n\\* But since $X_{1} \\sim \\mathcal{N}(0,1)$ and $X_{2} \\sim \\mathcal{N}(0,4)$ and independent Then\n$$\nf _ {X _ {1}} (x _ {1}) = \\frac {1}{\\sqrt {2 \\pi}} e ^ {- x ^ {2} / 2} \\text { and } f _ {X _ {2}} (x _ {2}) = \\frac {1}{\\sqrt {2 \\cdot 4 \\pi}} e ^ {- x ^ {2} / (2 \\cdot 4)}.\n$$\nThus by impedance\n$$\n\\begin{array}{l} f _ {X _ {1}, X _ {2}} (x _ {1}, x _ {2}) = f _ {X _ {1}} (x _ {1}) f _ {X _ {2}} (x _ {2}) \\\\ { = } { \\frac { 1 } { \\sqrt { 2 \\pi } } e ^ { - x ^ { 2 } / 2 } \\frac { 1 } { \\sqrt { 2 \\cdot 4 \\pi } } e ^ { - x ^ { 2 } / ( 2 \\cdot 4 ) } . } \\\\ \\end{array}\n$$\n\\* Thus we have\n$$\nf _ {Y _ {1}, Y _ {2}} \\left(y _ {1}, y _ {2}\\right) = \\frac {1}{\\sqrt {2 \\pi}} e ^ {- \\left(\\frac {3}{7} y _ {1} + \\frac {1}{y} y _ {2}\\right) ^ {2} / 2} \\frac {1}{\\sqrt {8 \\pi}} e ^ {- \\left(\\frac {1}{7} y _ {1} - \\frac {2}{7} y _ {2}\\right) ^ {2} / 8} \\frac {1}{7}.\n$$\n• Example3(if time): Suppose $X_{1}, X_{2}$ have joint distribution\n$$\nf _ {X _ {1}, X _ {2}} (x _ {1}, x _ {2}) = \\left\\{ \\begin{array}{l l} x _ {1} + \\frac {3}{2} \\left(x _ {2}\\right) ^ {2} & 0 \\leq x _ {1} \\leq 1, 0 \\leq x _ {2} \\leq 1 \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\nQuestion: Find the joint pdf of $Y_{1} = X_{1} + X_{2}$ and $Y_{2} = X_{1}^{2}$ .\n\\- Step1: Find the Jacobian: Note that\n$$\ny _ {1} = g _ {1} \\left(x _ {1}, x _ {2}\\right) = x _ {1} + x _ {2},\n$$\n$$\ny _ {2} = g _ {2} \\left(x _ {1}, x _ {2}\\right) = x _ {1} ^ {2}.\n$$\nSo\n$$\nJ \\left(x _ {1}, x _ {2}\\right) = \\left| \\begin{array}{c c} 1 & 2 \\\\ 2 x _ {1} & 0 \\end{array} \\right| = - 4 x _ {2}\n$$\n\\- Step2: Solve for $x_{1}, x_{2}$ and get\n$$\nx _ {1} = \\sqrt {y _ {2}},\n$$\n$$\nx _ {2} = y _ {1} - \\sqrt {y _ {2}}.\n$$\n\\- Step3: The joint pdf of $Y_{1}, Y_{2}$ is given by the formula:\n$$\n\\begin{array}{l} f _ {Y _ {1}, Y _ {2}} \\left(y _ {1}, y _ {2}\\right) = f _ {X _ {1}, X _ {2}} \\left(x _ {1}, x _ {2}\\right) \\left| J \\left(x _ {1}, x _ {2}\\right) \\right| ^ {- 1} \\\\ = f _ {X _ {1}, X _ {2}} \\left(\\sqrt {y _ {2}}, y _ {1} - \\sqrt {y _ {2}}\\right) \\frac {1}{| 4 x _ {2} |} \\\\ = \\left\\{ \\begin{array}{l l} \\frac {1}{| 4 x _ {2} |} \\left[ \\sqrt {y _ {2}} + \\frac {3}{2} \\left(y _ {1} - \\sqrt {y _ {2}}\\right) ^ {2} \\right] & 0 \\leq \\sqrt {y _ {2}} \\leq 1, \\\\ & 0 \\leq y _ {1} - \\sqrt {y _ {2}} \\leq 1 \\\\ 0 & \\text {otherwise} \\end{array} \\right. \\\\ \\end{array}\n$$", + "path": "Notes.pdf/CHAPTER 11/Multivariate distributions/11.6. Joint PDF of functions", + "metadata": { + "length": 6854, + "summary": "This passage explains how to find the joint probability density function (pdf) of new random variables that are functions of two original random variables. It outlines a three-step method: first, compute the Jacobian determinant of the transformation; second, solve for the original variables in terms of the new ones; third, apply the formula using the original joint pdf and the absolute inverse Jacobian. Three examples are provided: one with a uniform-like joint pdf and linear transformations, one with independent normal variables and linear combinations, and one with a polynomial joint pdf and a nonlinear transformation. The method is general for invertible transformations.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Goal", + "Recall", + "section", + "5.7", + "find", + "pdf", + "random", + "variable", + "Suppose", + "distributions", + "distribution", + "For", + "Steps", + "finding", + "joint", + "cdf", + "made", + "jointly", + "distributed", + "Let", + "multivariable", + "functions", + "Find", + "Step1", + "Jacobian", + "left", + "begin", + "array", + "nabla", + "end", + "frac", + "partial", + "neq", + "points", + "Step2", + "unique", + "solutions", + "equationf", + "terms", + "Step3", + "The", + "Example1", + "leq", + "text", + "Question", + "Note", + "So", + "Solve", + "formula", + "Example2", + "sim", + "mathcal", + "independent", + "2X", + "3X", + "But", + "Then", + "sqrt", + "pi", + "cdot", + "Thus", + "impedance", + "Example3", + "time" + ], + "keywords": [ + "Jacobian", + "joint pdf", + "transformation" + ], + "connect_to": [] + } + }, + { + "chunk_id": "b5ba609a-4b00-5f63-adc5-4e06362c83c1", + "type": "image", + "content": "\n(b) If $X \\leq Y$ then $\\mathbb{E}X \\leq \\mathbb{E}Y$ . \\- Example4: Let $X_{1}, \\ldots, X_{n}$ be independent and identically distributed random (i.i.d.) random variables. Suppose $\\mathbb{E}X_{i} = \\mu$ . We call the quantity0\n[images/image-26-(b) If $X .jpg]\n", + "path": "images/image-26-(b) If $X .jpg", + "metadata": { + "length": 263, + "summary": "image-26\n(b) If $X \\leq Y$ then $\\mathbb{E}X \\leq \\mathbb{E}Y$ . \\- Example4: Let $X_{1}, \\ldots, X_{n}$ be independent and identically distributed random (i.i.d.) random variables. Suppose $\\mathbb{E}X_{i} = \\mu$ . We call the quantity0", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-26-(b) If $X .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "571e60e8-f360-5c03-910a-f37318902e72", + "type": "image", + "content": "\n- Covariance is a measure of how much two random variables change together.
    0\n[images/image-27-Scatter Plot with Trend Line.jpg]\n", + "path": "images/image-27-Scatter Plot with Trend Line.jpg", + "metadata": { + "length": 121, + "summary": "image-27\nChart Type: Scatter plot with linear regression line (red) and data points (blue).\nX-Axis Range: -20 to 60.\nY-Axis Range: 5 to 15.\nKey Data Points:\n- X-axis ticks at intervals of 10 (-20, -10, 0, 10, 20, 30, 40, 50, 60).\n- Y-axis ticks at intervals of 5 (5, 10, 15).\nTrend Analysis: The blue dots show a positive correlation, indicating that as the x-value increases, the y-value generally increases. The red line represents the best-fit trend through these points.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "images/image-27-Scatter Plot with Trend Line.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "637b41b5-ec03-5097-9006-af70977b8d48", + "type": "table", + "content": "\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
    xy
    -203
    -154
    -105
    -56
    07
    58
    109
    1510
    2011
    2512
    3013
    3514
    4015
    4516
    5017
    5518
    6019
    ", + "path": "tables/table-11 Linear Data Table.html", + "metadata": { + "length": 1096, + "summary": "table-12\nThe table shows a linear relationship between x and y, where y increases by 1 for every 5-unit increase in x, ranging from x=-20 to x=60.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "file_path": "tables/table-11 Linear Data Table.html", + "keywords": [ + "linear" + ], + "tokens": [] + } + }, + { + "chunk_id": "64f88945-538d-5a9b-bdff-5b02a0b57039", + "type": "text", + "content": "Expectations\n12.1. Expectation of Sums of R.V.\nTHEOREM 17. Let $g: \\mathbb{R}^2 \\to \\mathbb{R}$ . If $X, Y$ have joint pmf $p(x, y)$ then\n$$\n\\mathbb {E} \\left[ g (X, Y) \\right] = \\sum_ {y} \\sum_ {x} g (x, y) p (x, y).\n$$\nIf $X, Y$ have joint pdf $f(x, y)$ then\n$$\n\\mathbb {E} \\left[ g (X, Y) \\right] = \\int_ {- \\infty} ^ {\\infty} \\int_ {- \\infty} ^ {\\infty} g (x, y) f (x, y) d x d y.\n$$\n\\- Example1: Suppose the joint p.m.f of $X$ and $Y$ is given by $\\begin{array}{c|cc}X\\setminus Y & 0 & 2\\\\ \\hline 0 & .2 & .7\\\\ \\hline 1 & 0 & .1 \\end{array}$ . Find $\\mathbb{E}[XY]$ .\n\\- Solution: Using the formula we have with the function $g(x,y) = xy$ :\n$$\n\\begin{array}{l} \\mathbb {E} [ X Y ] = \\sum_ {i, j} x _ {i} y _ {j} p (x _ {i}, y _ {j}) \\\\ = 0 \\cdot 0 p (0, 0) + 1 \\cdot 0 p (1, 0) + 0 \\cdot 2 p (0, 2) + 1 \\cdot 2 p (1, 2) \\\\ = 0 \\cdot 0 \\cdot . 2 + 1 \\cdot 0 \\cdot 0 + 0 \\cdot 2 \\cdot . 7 + 1 \\cdot 2 \\cdot . 1 \\\\ = . 2 \\\\ \\end{array}\n$$\n\\- Example2: Suppose $X, Y$ are independent exponential r.v. with parameter $\\lambda = 1$ . Set up a double integral that represents\n$$\n\\mathbb {E} \\left[ X ^ {2} Y \\right].\n$$\n\\- Solution: Since $X, Y$ are independent then\n$$\nf _ {X, Y} (x, y) = e ^ {- 1 x} e ^ {- 1 y} = e ^ {- (x + y)}. \\quad 0 < x, y < \\infty .\n$$\n\\- Then DRAW FIRST then\n$$\n\\mathbb {E} \\left[ X ^ {2} Y \\right] = \\int_ {0} ^ {\\infty} \\int_ {0} ^ {\\infty} x ^ {2} y e ^ {- (x + y)} d y d x.\n$$\n• Example3: Suppose the joint pdf of X, Y is\n$$\nf (x, y) = \\left\\{ \\begin{array}{l l} 1 0 x y ^ {2} & 0 < x < y, 0 < y < 1 \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\nFind $\\mathbb{E}XY$ and $\\operatorname{Var}(Y)$ .\n\\- Solution:\n\\- We First DRAW and then set up\n$$\n\\begin{array}{l} \\mathbb {E} X Y = \\int_ {0} ^ {1} \\int_ {0} ^ {y} x y (1 0 x y ^ {2}) d x d y = 1 0 \\int_ {0} ^ {1} \\int_ {0} ^ {y} x ^ {2} y ^ {3} d x d y \\\\ = \\frac {1 0}{3} \\int_ {0} ^ {1} y ^ {3} y ^ {3} d y = \\frac {1 0}{3} \\frac {1}{7} = \\frac {1 0}{2 1}. \\\\ \\end{array}\n$$\n\\- First note that $\\operatorname{Var}(Y) = \\mathbb{E}Y^2 - (\\mathbb{E}Y)^2$ .\n\\- Then\n$$\n\\begin{array}{l} \\mathbb {E} Y ^ {2} = \\int_ {0} ^ {1} \\int_ {0} ^ {y} y ^ {2} (1 0 x y ^ {2}) d x d y = 1 0 \\int_ {0} ^ {1} \\int_ {0} ^ {y} y ^ {4} x d x d y \\\\ = 5 \\int_ {0} ^ {1} y ^ {4} y ^ {2} d y = \\frac {5}{7}. \\\\ \\end{array}\n$$\nand\n$$\n\\begin{array}{l} \\mathbb {E} Y = \\int_ {0} ^ {1} \\int_ {0} ^ {y} y (1 0 x y ^ {2}) d x d y = 1 0 \\int_ {0} ^ {1} \\int_ {0} ^ {y} y ^ {3} x d x d y \\\\ = 5 \\int_ {0} ^ {1} y ^ {3} y ^ {2} d y = \\frac {5}{6}. \\\\ \\end{array}\n$$\nSo that $\\operatorname{Var}(Y) = \\frac{5}{7} - \\left(\\frac{5}{6}\\right)^2 = \\frac{5}{252}$ .\nTHEOREM 18. (Properties of Expectation)\n(a) $\\mathbb{E}[X + Y] = \\mathbb{E}X + \\mathbb{E}Y$\n(b) If $X \\leq Y$ then $\\mathbb{E}X \\leq \\mathbb{E}Y$ .\nPROOF. Part (a) was proved for the discrete case. So we only need to show the continuous case:\n$$\n\\begin{array}{l} \\mathbb {E} \\left[ X + Y \\right] = \\int \\int (x + y) f (x, y) d y d x \\\\ = \\int \\int x f (x, y) d y d x + \\int \\int y f (x, y) d y d x \\\\ = \\int x f _ {X} (x) d x + \\int y f _ {Y} (y) d y \\\\ = \\mathbb {E} X + \\mathbb {E} Y. \\\\ \\end{array}\n$$\n\n(b) If $X \\leq Y$ then $\\mathbb{E}X \\leq \\mathbb{E}Y$ . \\- Example4: Let $X_{1}, \\ldots, X_{n}$ be independent and identically distributed random (i.i.d.) random variables. Suppose $\\mathbb{E}X_{i} = \\mu$ . We call the quantity0\n[images/image-26-(b) If $X .jpg]\n\n\\- Example4: Let $X_{1}, \\ldots, X_{n}$ be independent and identically distributed random (i.i.d.) random variables. Suppose $\\mathbb{E}X_{i} = \\mu$ . We call the quantity\n$$\n\\bar {X} = \\sum_ {i = 1} ^ {n} \\frac {X _ {i}}{n}\n$$\nthe sample mean. Compute $\\mathbb{E}\\left[\\bar{X}\\right]$ .\n\\- Solution: We use the properties of expectation\n$$\n\\begin{array}{l} \\mathbb {E} \\left[ \\bar {X} \\right] = \\mathbb {E} \\left[ \\sum_ {i = 1} ^ {n} \\frac {X _ {i}}{n} \\right] \\\\ = \\frac {1}{n} \\mathbb {E} \\left[ X _ {1} + \\dots + X _ {n} \\right] \\\\ = \\frac {1}{n} \\left(\\mathbb {E} \\left[ X _ {1} \\right] + \\dots + \\mathbb {E} \\left[ X _ {n} \\right]\\right) \\\\ = \\frac {1}{n} (\\mu + \\dots + \\mu) = \\frac {n \\mu}{n} \\\\ = \\mu . \\\\ \\end{array}\n$$\n\\- In statistics, the sample mean is used to estimate the actual mean of a distribution.\nTHEOREM 19. If $X, Y$ are independent then\n$$\n\\mathbb {E} [ X Y ] = (\\mathbb {E} X) (\\mathbb {E} Y).\n$$\nPROOF. In the continuosuc case we have\n$$\n\\begin{array}{l} \\mathbb {E} \\left[ X Y \\right] = \\int \\int x y f _ {X, Y} (x, y) d y d x \\\\ = \\int \\int x y f _ {X} (x) f _ {Y} (y) d y d x \\\\ = \\left(\\int x f _ {X} (x) d x\\right) \\left(\\int y f _ {Y} (y) d y\\right) \\\\ = \\left(\\mathbb {E} X\\right) (\\mathbb {E} Y). \\\\ \\end{array}\n$$\nThe discrete case is the same, except replace integrals with summations.\n\\- In general, the following is true:\nTHEOREM 20. If $X, Y$ are independent and $g, h: \\mathbb{R} \\to \\mathbb{R}$ then\n$$\n\\mathbb {E} \\left[ g (X) h (Y) \\right] = \\mathbb {E} \\left[ g (X) \\right] \\mathbb {E} \\left[ h (Y) \\right].\n$$\n12.2. Covariance and Correlations.\n- Note that $\\mathbb{E}X$ and $\\operatorname{Var}X$ give information about a single random variable.\n- What statistic can give us information about how $X$ effects $Y$ , or vice versa?\nDEFINITION. The covariance between $X$ and $Y$ , is defined by\n$$\n\\operatorname{Cov} (X, Y) = \\mathbb {E} \\left[ \\left(X - \\mu_ {X}\\right) \\left(Y - \\mu_ {Y}\\right) \\right].\n$$\n• After some algebra one can show that\n$$\n\\operatorname{Cov} (X, Y) = \\mathbb {E} [ X Y ] - \\mathbb {E} X \\mathbb {E} Y.\n$$\n\\- The covariance between two random variables give us information about relationship between the random variables.\n- Covariance is a measure of how much two random variables change together.\n- If the greater values of one variable mainly correspond with the greater values of the other variable, and the same holds for the lesser values, i.e., the variables tend to show similar behavior, the covariance is positive.\n\\* Thus covariance measures if there is a linear relationship between $X$ and $Y$ .\n- The sign of the covariance therefore shows the tendency in the linear relationship between the variables.\n- For example, the following plots shows a positive linear relationship between $X$ and $Y$ :\n\n- Covariance is a measure of how much two random variables change together.
    0\n[images/image-27-Scatter Plot with Trend Line.jpg]\n\n
    \nscatterplot\n\n[tables/table-11 Linear Data Table.html]\n\n
    \nIn this case $\\operatorname{Cov}(X,Y) > 0$ .\n\\- Note: If $X, Y$ are independent then $\\operatorname{Cov}(X, Y) = 0$ . (This is not true in the other direction. Meaning $\\operatorname{Cov}(X, Y) = 0$ does not imply that $X, Y$ are independent!)\n- So $\\operatorname{Cov}(X, Y) = 0$ means they are uncorrelated.\n\\- Properties:\n- (i) $\\operatorname{Cov}(X,Y) = \\operatorname{Cov}(Y,X)$\n- (ii) $\\operatorname{Cov}(X, X) = \\operatorname{Var}(X)$\n- (iii) $\\operatorname{Cov}(aX,Y) = a\\operatorname{Cov}(X,Y)$\n- (iv) $\\operatorname{Cov}\\left(\\sum_{i} X_{i}, \\sum_{j} Y_{j}\\right) = \\sum_{i} \\sum_{j} \\operatorname{Cov}(X_{i}, Y_{j})$ .\nTHEOREM 21. (★) Formula for Sum of Variation:\n$$\n\\operatorname{Var} (X + Y) = \\operatorname{Var} (X) + \\operatorname{Var} (Y) + 2 \\operatorname{Cov} (X, Y).\n$$\nGives us a formula for variation of $X_{1},\\ldots ,X_{n}$ :\n$$\n\\operatorname{Var} \\left(\\sum_ {i = 1} ^ {n} X _ {i}\\right) = \\sum_ {i = 1} ^ {n} \\operatorname{Var} \\left(X _ {i}\\right) + 2 \\sum \\sum_ {i < j} \\operatorname{Cov} \\left(X _ {i}, X _ {j}\\right).\n$$\n• Fact: Note that if X, Y are independent then\n$$\n\\operatorname{Var} (X + Y) = \\operatorname{Var} (X) + \\operatorname{Var} (Y).\n$$\n\\- Finally we have the following: Its standardized way to know how correlated two random variables are:\nDEFINITION. The correlation coefficient of two random variables $X$ and $Y$ , denoted by $\\rho(X, Y)$ is defined by\n$$\n\\rho (X, Y) = \\frac {\\operatorname{Cov} (X , Y)}{\\sqrt {\\operatorname{Var} (X) \\operatorname{Var} (Y)}}.\n$$\n\\- Fact:\n$-(1) - 1 \\leq \\rho(X, Y) \\leq 1$\n- (2) If $\\rho(X,Y)=1$ then $Y=a+bX$ where $b=\\frac{\\sigma_y}{\\sigma_x}>0$ (Straight positive sloped line)\n- (3) If $\\rho(X,Y) = -1$ then $Y = a + bX$ where $b = -\\frac{\\sigma_y}{\\sigma_x} < 0$ (Straight negatively sloped line)\n- (4) This $\\rho$ is a measure of linearity between $Y$ and $X$ .\n\\* $\\rho > 0$ positive linearity: Meaning that if you were to draw a line of best fit, then it must be a positive sloped line\n\\- The closer $\\rho$ gets to 1, the more $(X,Y)$ seems to be in a positive sloped straight line\n\\* $\\rho < 0$ negative linearity: Meaning that if you were to draw a line of best fit, then it must be a negative sloped line\n\\- The closer $\\rho$ gets to $-1$ , the more $(X,Y)$ seems to be in a negative sloped straight line\n\\- (5) If $\\rho(X,Y)=0$ , then $X$ and $Y$ are uncorrelated.\n\\- Warning:\n$-\\rho (X,Y)$ does not pick up any other relationship, such as quadratic, or cubic\n\\- $\\rho(X, Y)$ is not the slope of the line of best fit. It is simply tell us if it's positive, or negative relationship, and the strength of relationship.\n\\- Example1: Suppose $X, Y$ are random variables whose joint pdf is given by\n$$\nf (x, y) = \\left\\{ \\begin{array}{l l} \\frac {1}{y} & 0 < y < 1, 0 < x < y \\\\ 0 & \\text { otherwise } \\end{array} \\right..\n$$\n- Part (a): Find the covariance of $X$ and $Y$ .\n- Part (b) Compute $\\operatorname{Var}(X)$ and $\\operatorname{Var}(Y)$ .\n- $\\underline{\\text{Part (c)}}$ Calculate $\\rho(X,Y)$ .\n- Solution:\n- $\\underline{\\text{Part (a):}}$ Find the covariance of $X$ and $Y$ .\n\\- Recall that $\\operatorname{Cov}(X,Y) = \\mathbb{E}XY - \\mathbb{E}X\\mathbb{E}Y$ . So\n$$\n\\mathbb {E} X Y = \\int_ {0} ^ {1} \\int_ {0} ^ {y} x y \\frac {1}{y} d x d y = \\int_ {0} ^ {1} \\frac {y ^ {2}}{2} d y = \\frac {1}{6}\n$$\n$$\n\\mathbb {E} X = \\int_ {0} ^ {1} \\int_ {0} ^ {y} x \\frac {1}{y} d x d y = \\int_ {0} ^ {1} \\frac {y}{2} d y = \\frac {1}{4}.\n$$\n$$\n\\mathbb {E} Y = \\int_ {0} ^ {1} \\int_ {0} ^ {y} y \\frac {1}{y} d x d y = \\int_ {0} ^ {1} y d y = \\frac {1}{2}.\n$$\nThus\n$$\n\\operatorname{Cov} (X, Y) = \\mathbb {E} X Y - \\mathbb {E} X \\mathbb {E} Y\n$$\n$$\n= \\frac {1}{6} - \\frac {1}{4} \\frac {1}{2}\n$$\n$$\n= \\frac {1}{2 4}.\n$$\n\\- Part (b): Compute $\\operatorname{Var}(X)$ and $\\operatorname{Var}(Y)$ .\n\\- We have that\n$$\n\\mathbb {E} X ^ {2} = \\int_ {0} ^ {1} \\int_ {0} ^ {y} x ^ {2} \\frac {1}{y} d x d y = \\int_ {0} ^ {1} \\frac {y ^ {2}}{3} d y = \\frac {1}{9}.\n$$\n$$\n\\mathbb {E} Y ^ {2} = \\int_ {0} ^ {1} \\int_ {0} ^ {y} y ^ {2} \\frac {1}{y} d x d y = \\int_ {0} ^ {1} y ^ {2} d y = \\frac {1}{3}.\n$$\n\\- Thus recall that\n$$\n\\operatorname{Var} (X) = \\mathbb {E} X ^ {2} - (\\mathbb {E} X) ^ {2}\n$$\n$$\n= \\frac {1}{9} - \\left(\\frac {1}{4}\\right) ^ {2} = \\frac {7}{1 4 4}\n$$\nAlso\n$$\n\\operatorname{Var} (Y) = \\mathbb {E} Y ^ {2} - (\\mathbb {E} Y) ^ {2}\n$$\n$$\n= \\frac {1}{3} - \\left(\\frac {1}{2}\\right) ^ {2} = \\frac {1}{1 2}.\n$$\n\\- Part (c): Calculate $\\rho(X,Y)$ .\n\\- We now use\n$$\n\\rho (X, Y) = \\frac {\\operatorname{Cov} (X , Y)}{\\sqrt {\\operatorname{Var} (X) \\operatorname{Var} (Y)}}\n$$\n$$\n= \\frac {\\frac {1}{2 4}}{\\sqrt {\\left(\\frac {7}{1 4 4}\\right) \\left(\\frac {1}{1 2}\\right)}} \\approx . 6 5 4 7.\n$$", + "path": "Notes.pdf/CHAPTER 12", + "metadata": { + "length": 11053, + "summary": "The text covers expectation properties for sums of random variables, including formulas for discrete and continuous cases. It defines covariance and correlation, explaining their roles in measuring linear relationships between variables. Examples illustrate computing expectations, variances, covariance, and correlation for given joint distributions.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "Expectations", + "12.1", + "Expectation", + "Sums", + "THEOREM", + "17", + "Let", + "mathbb", + "If", + "joint", + "pmf", + "left", + "sum", + "pdf", + "int", + "infty", + "Example1", + "Suppose", + "begin", + "array", + "cc", + "setminus", + "hline", + "end", + "Find", + "XY", + "Solution", + "Using", + "formula", + "function", + "xy", + "cdot", + "Example2", + "independent", + "exponential", + "parameter", + "lambda", + "Set", + "double", + "integral", + "represents", + "Since", + "quad", + "Then", + "DRAW", + "FIRST", + "Example3", + "text", + "operatorname", + "Var", + "We", + "First", + "set", + "frac", + "note", + "So", + "252", + "18", + "Properties", + "leq", + "PROOF", + "Part", + "proved", + "discrete", + "case", + "show", + "continuous", + "Example4", + "ldots", + "identically", + "distributed", + "random", + "variables", + "mu", + "call", + "quantity0", + "quantity", + "bar", + "sample", + "Compute", + "properties", + "expectation", + "dots", + "In", + "statistics", + "estimate", + "actual", + "distribution", + "19", + "continuosuc", + "The", + "replace", + "integrals", + "summations", + "general", + "true", + "20", + "12.2", + "Covariance", + "Correlations", + "Note", + "give", + "information", + "single", + "variable", + "What", + "statistic", + "effects", + "vice", + "versa", + "DEFINITION", + "covariance", + "defined", + "Cov", + "After", + "algebra", + "relationship", + "measure", + "change", + "greater", + "values", + "correspond", + "holds", + "lesser", + "tend", + "similar", + "behavior", + "positive", + "Thus", + "measures", + "linear", + "sign", + "shows", + "tendency", + "For", + "plots", + "details", + "summary", + "scatterplot", + "This", + "direction", + "Meaning", + "imply", + "means", + "uncorrelated", + "ii", + "iii", + "aX", + "iv", + "21", + "Formula", + "Sum", + "Variation", + "Gives", + "variation", + "Fact", + "Finally", + "Its", + "standardized", + "correlated", + "correlation", + "coefficient", + "denoted", + "rho", + "sqrt", + "bX", + "sigma", + "Straight", + "sloped", + "line", + "negatively", + "linearity", + "draw", + "fit", + "closer", + "straight", + "negative", + "Warning", + "pick", + "quadratic", + "cubic", + "slope", + "It", + "simply", + "strength", + "underline", + "Calculate", + "Recall", + "recall", + "Also", + "approx" + ], + "keywords": [ + "Expectation", + "Covariance", + "Correlation" + ], + "connect_to": [ + { + "target": "b5ba609a-4b00-5f63-adc5-4e06362c83c1", + "relation": "embeds", + "ref": "[images/image-26-(b) If $X .jpg]", + "position": { + "start": 3357, + "end": 3389 + } + }, + { + "target": "571e60e8-f360-5c03-910a-f37318902e72", + "relation": "embeds", + "ref": "[images/image-27-Scatter Plot with Trend Line.jpg]", + "position": { + "start": 6260, + "end": 6310 + } + }, + { + "target": "637b41b5-ec03-5097-9006-af70977b8d48", + "relation": "embeds", + "ref": "[tables/table-11 Linear Data Table.html]", + "position": { + "start": 6354, + "end": 6394 + } + } + ] + } + }, + { + "chunk_id": "af115d75-90e7-5d5d-b6fb-e917efa6994c", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 13", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3886005f-9dea-5164-91ef-cd67d20d61b8", + "type": "text", + "content": "13.1. Moment Generating Functions\n\\- For each random variable $X$ , we can define its moment generating function $m_X(t)$ by\n$$\n\\begin{array}{l} {m _ {X} (t)} = {\\mathbb {E} \\left[ e ^ {t X} \\right]} \\\\ = \\left\\{ \\begin{array}{l l} \\sum_ {x _ {i}} e ^ {t x _ {i}} p (x _ {i}) & , \\text {if X is discrete} \\\\ \\int_ {- \\infty} ^ {\\infty} e ^ {t x} f (x) d s & , \\text {if X is continuous} \\end{array} \\right.. \\\\ \\end{array}\n$$\n\\- $m_X(t)$ is called the moment generating function (m.g.f.) because we can find all the moments of $X$ by differentiating $m(t)$ and then evaluating at $t = 0$ .\n\\- Note that\n$$\n\\begin{array}{l} {m ^ {\\prime} (t)} = {\\frac {d}{d t} \\mathbb {E} \\left[ e ^ {t X} \\right]} \\\\ = \\mathbb {E} \\left[ \\frac {d}{d t} e ^ {t X} \\right] \\\\ = \\mathbb {E} \\left[ X e ^ {t X} \\right]. \\\\ \\end{array}\n$$\nNow evaluate at t = 0 and get\n$$\nm ^ {\\prime} (0) = \\mathbb {E} \\left[ X e ^ {0 \\cdot X} \\right] = \\mathbb {E} [ X ].\n$$\n\\- Similarly,\n$$\n\\begin{array}{l} {m ^ {\\prime \\prime} (t)} = {\\frac {d}{d t} \\mathbb {E} \\left[ X e ^ {t X} \\right]} \\\\ = \\mathbb {E} \\left[ X ^ {2} e ^ {t X} \\right] \\\\ \\end{array}\n$$\nso that\n$$\nm ^ {\\prime \\prime} (0) = \\mathbb {E} \\left[ X ^ {2} e ^ {0} \\right] = \\mathbb {E} \\left[ X ^ {2} \\right].\n$$\nTHEOREM 22. For all $n \\geq 0$ we have\n$$\n\\mathbb {E} \\left[ X ^ {n} \\right] = m ^ {(n)} (0).\n$$\n• Examples of Moment generating Functions\n- Bernoulli: Recall that $p(1) = p$ and $p(0) = 1 - p$ . Thus\n$$\n\\begin{array}{l} m _ {X} (t) = \\mathbb {E} e ^ {t X} = e ^ {t \\cdot 0} p (0) + e ^ {t \\cdot 1} p (1) \\\\ = p e ^ {t} + (1 - p). \\\\ \\end{array}\n$$\n\\- Binomial: Recall that $X \\sim \\text{Bin}(n, p)$ if $X = \\sum_{i=0}^{n} Y_i$ where $Y_i \\sim \\text{Bern}(p)$ thus\n$$\n\\begin{array}{l} m _ {X} (t) = \\mathbb {E} e ^ {t X} = \\mathbb {E} e ^ {t X \\sum_ {i = 0} ^ {n} Y _ {i}} \\\\ = \\mathbb {E} \\left[ \\left(e ^ {t Y _ {1}}\\right) \\dots \\left(e ^ {t Y _ {n}}\\right) \\right] \\\\ = \\mathbb {E} \\left[ e ^ {t Y _ {1}} \\right] \\dots \\mathbb {E} \\left[ e ^ {t Y _ {n}} \\right], \\text {by independence} \\\\ = \\left(p e ^ {t} + (1 - p)\\right) ^ {n} \\\\ \\end{array}\n$$\n\\- Poisson: If $X \\sim \\text{Poisson}(\\lambda)$ then\n$$\n\\begin{array}{l} m _ {X} (t) = \\mathbb {E} e ^ {t X} = \\sum_ {n = 0} ^ {\\infty} e ^ {t n} e ^ {- \\lambda} \\frac {\\lambda^ {n}}{n !} \\\\ = e ^ {- \\lambda} \\sum_ {n = 0} ^ {\\infty} e ^ {t n} \\frac {\\lambda^ {n}}{n !} \\\\ = e ^ {- \\lambda} \\sum_ {n = 0} ^ {\\infty} \\frac {(e ^ {t} \\lambda) ^ {n}}{n !} \\\\ \\end{array}\n$$\nnow recall from Calculus 2 that $e^x = \\sum_{n=0}^{\\infty} \\frac{x^n}{n!}$ so that\n$$\n\\begin{array}{l} m _ {X} (t) = e ^ {- \\lambda} \\sum_ {n = 0} ^ {\\infty} \\frac {x ^ {n}}{n !}, \\text { with } x = e ^ {t} \\lambda \\\\ = e ^ {- \\lambda} e ^ {e ^ {t} \\lambda} \\\\ = e ^ {e ^ {t} \\lambda - \\lambda} \\\\ = \\exp (\\lambda (e ^ {t} - 1)) \\\\ \\end{array}\n$$\n\\- Exponential: If $X \\sim \\exp(\\lambda)$ then\n$$\n\\begin{array}{l} m _ {X} (t) = \\mathbb {E} e ^ {t X} \\\\ = \\int_ {0} ^ {\\infty} e ^ {t x} \\lambda e ^ {- \\lambda x} d x \\\\ = \\frac {\\lambda}{\\lambda - t}, \\\\ \\end{array}\n$$\nwhich is valued whenever $t > \\lambda$ .\n\\- Standard Normal: If $X \\sim \\mathcal{N}(0,1)$ then\n$$\n\\begin{array}{l} m _ {X} (t) = \\mathbb {E} e ^ {t X} = \\frac {1}{\\sqrt {2 \\pi}} \\int_ {- \\infty} ^ {\\infty} e ^ {t x} e ^ {- x ^ {2} / 2} \\\\ = e ^ {t ^ {2} / 2}. \\\\ \\end{array}\n$$\n\\- Normal: If $X \\sim \\mathcal{N}(\\mu, \\sigma^2)$ then $X = \\mu + \\sigma Z$ so that\n$$\n\\begin{array}{l} m _ {X} (t) = \\mathbb {E} e ^ {t X} \\\\ = \\mathbb {E} e ^ {t \\mu} e ^ {t \\sigma Z} = e ^ {t \\mu} \\mathbb {E} e ^ {(t \\sigma) Z} \\\\ = e ^ {t \\mu} m _ {X} (t \\sigma) = e ^ {t \\mu} e ^ {(t \\sigma) ^ {2} / 2} \\\\ = \\exp \\left(t \\mu + \\frac {t ^ {2} \\sigma^ {2}}{2}\\right). \\\\ \\end{array}\n$$\n\\- Property: Suppose $X, Y$ are independent then what is that m.g.f. of $X + Y$ ?\n\\- Let's try to figure out:\n$$\n\\begin{array}{l} m _ {X + Y} (t) = \\mathbb {E} e ^ {t (X + Y)} = \\mathbb {E} \\left(e ^ {t X} e ^ {t Y}\\right) \\\\ = \\mathbb {E} \\left(e ^ {t X}\\right) \\mathbb {E} \\left(e ^ {t Y}\\right), \\text { by independence } \\\\ = m _ {X} (t) m _ {Y} (t). \\\\ \\end{array}\n$$\n\\- Thus we know that\n$$\nm _ {X + Y} (t) = m _ {X} (t) m _ {Y} (t).\n$$\n\\- Note: Also note that of $f_{X}(x)$ is the pdf of a r.v. then it's m.g.f is\n$$\n\\mathbb {E} e ^ {t X} = \\int e ^ {t x} f _ {X} (x) d x.\n$$\nThis is similar to the $\\underline{\\text{laplace transform}}$ of $f_X(x)$ . $[\\mathcal{L}[f](s) = \\int e^{-sx} f_X(x) dx]$ .\n\\- Recall that there is one to one correspondence of laplace transforms. That completely determines a function.\nTHEOREM 23. If $m_X(t) = m_Y(t) < \\infty$ for all $t$ in an interval, then $X$ and $Y$ have the same distribution. That is, m.g.f's completely determines the distribution.\n\\- Example1: Suppose that m.g.f of $X$ is given by $m(t) = e^{3(e^t - 1)}$ . Find $\\mathbb{P}(X = 0)$ .\n\\- Solution: (We want to work backwards). Match this m.g.f to a known m.g.f in our table. Looks like\n$$\nm (t) = e ^ {3 \\left(e ^ {t} - 1\\right)} = e ^ {\\lambda \\left(e ^ {t} - 1\\right)} \\quad \\text { where } \\lambda = 3.\n$$\nThus $X\\sim \\text{Poisson}(3)$ . Thus\n$$\n\\mathbb {P} (X = 0) = e ^ {- \\lambda} \\frac {\\lambda^ {0}}{0 !} = e ^ {- 3}.\n$$\n\\- Summary:\n(1) $m(t) = \\mathbb{E}e^{tX}$ . We have a table of mgf of distributions:\n(2) The m.g.f helps us find moments: $\\mathbb{E}[X^n] = m^{(n)}(0)$\n(3) If $X, Y$ are independent then $m_{X + Y}(t) = m_X(t)m_Y(t)$ .\n(4) The m.g.f. helps us determine the distribution of random variables. If $m_X(t) = m_Y(t)$ then $X$ and $Y$ have the same distribution.\n\\- Recall we had a section on sums of independent random variables.\n\\- Example2: Recall $X \\sim \\mathcal{N}\\left(\\mu_x, \\sigma_x^2\\right)$ and $Y \\sim \\mathcal{N}(\\mu_y, \\sigma_y^2)$ , independent. Then what is\n$$\nX + Y \\sim \\mathcal {N} (\\mathrm{?},?)\n$$\n\\- Sol: Note that\n$$\n\\begin{array}{l} m _ {X + Y} (t) = m _ {X} (t) m _ {Y} (t) \\\\ = \\exp \\left(t \\mu_ {x} + \\frac {t ^ {2} \\sigma_ {x} ^ {2}}{2}\\right) \\exp \\left(t \\mu_ {y} + \\frac {t ^ {2} \\sigma_ {y} ^ {2}}{2}\\right) \\\\ = \\exp \\left(t \\left(\\mu_ {x} + \\mu_ {y}\\right) + \\frac {t ^ {2} \\left(\\sigma_ {x} ^ {2} + \\sigma_ {y} ^ {2}\\right)}{2}\\right). \\\\ \\end{array}\n$$\nSo then you look at our table and check which distribution has this mg.f. with $\\mu = \\mu_x + \\mu_y$ and $\\sigma^2 = \\sigma_x^2 +\\sigma_y^2$ . so that $X + Y\\sim \\mathcal{N}\\left(\\mu_x + \\mu_y,\\sigma_x^2 +\\sigma_y^2\\right)$\n\\- Example3: Suppose $X \\sim \\text{bin}(n, p)$ and $Y \\sim \\text{bin}(m, p)$ , independent, then what is the distribution of $X + Y$ ?\n\\- Solution: We use\n$$\n\\begin{array}{l} {m _ {X + Y} (t)} = {m _ {X} (t) m _ {Y} (t)} \\\\ = \\left(p e ^ {t} + (1 - p)\\right) ^ {n} \\left(p e ^ {t} + (1 - p)\\right) ^ {m} \\\\ = \\left(p e ^ {t} + (1 - p)\\right) ^ {n + m}. \\\\ \\end{array}\n$$\nLook at the table and see what distribution has this m.g.f. Thus\n$$\nX + Y \\sim b i n (n + m, p).\n$$\n\\- Example4: Suppose $X$ is a discrete random variable and has the m.g.f.\n$$\nm _ {X} (t) = \\frac {1}{7} e ^ {2 t} + \\frac {3}{7} e ^ {3 t} + \\frac {2}{7} e ^ {5 t} + \\frac {1}{7} e ^ {8 t}.\n$$\nQuestion: What is the p.m.f of $X$ ? Find $\\mathbb{E}X$ .\n\\- Solution(a): This doesn't match any of the known mg.f.s. Thus we can read off from the mgf that since\n$$\n\\frac {1}{7} e ^ {2 t} + \\frac {3}{7} e ^ {3 t} + \\frac {2}{7} e ^ {5 t} + \\frac {1}{7} e ^ {8 t} = \\sum_ {i = 1} ^ {4} e ^ {t x _ {i}} p (x _ {i})\n$$\nthen $p(2) = \\frac{1}{7}, p(3) = \\frac{3}{7}, p(5) = \\frac{2}{7}$ and $p(8) = \\frac{1}{7}$ .\n\\- Solution(b): First\n$$\nm ^ {\\prime} (t) = \\frac {2}{7} e ^ {2 t} + \\frac {9}{7} e ^ {3 t} + \\frac {1 0}{7} e ^ {5 t} + \\frac {8}{7} e ^ {8 t},\n$$\nso that\n$$\n\\mathbb {E} [ X ] = m ^ {\\prime} (0) = \\frac {2}{7} + \\frac {9}{7} + \\frac {1 0}{7} + \\frac {8}{7} = \\frac {2 9}{7}.\n$$\n• Example5: Suppose X has m.g.f\n$$\nm _ {X} (t) = (1 - 2 t) ^ {- \\frac {1}{2}} \\text { for } t < \\frac {1}{2}.\n$$\nFind the first and second moments of $X$ .\n\\- Solution: We have\n$$\nm _ {X} ^ {\\prime} (t) = - \\frac {1}{2} (1 - 2 t) ^ {- \\frac {3}{2}} (- 2) = (1 - 2 t) ^ {- \\frac {3}{2}},\n$$\n$$\nm _ {X} ^ {\\prime \\prime} (t) = - \\frac {3}{2} (1 - 2 t) ^ {- \\frac {5}{2}} (- 2) = 3 (1 - 2 t) ^ {- \\frac {5}{2}}.\n$$\nSo that\n$$\n\\begin{array}{l} \\mathbb {E} X = m _ {X} ^ {\\prime} (0) = (1 - 2 \\cdot 0) ^ {- \\frac {3}{2}} = 1, \\\\ \\mathbb {E} X ^ {2} = m _ {X} ^ {\\prime \\prime} (0) = 3 (1 - 2 \\cdot 0) ^ {- \\frac {5}{2}} = 3. \\\\ \\end{array}\n$$", + "path": "Notes.pdf/CHAPTER 13/Moment generating functions", + "metadata": { + "length": 8355, + "summary": "This text defines moment generating functions (m.g.f.) for random variables, showing how to compute them for discrete and continuous cases. It explains that moments can be found by differentiating the m.g.f. and evaluating at zero. Examples are given for Bernoulli, Binomial, Poisson, Exponential, and Normal distributions. Properties include that the m.g.f. of a sum of independent variables is the product of their m.g.f.s, and that the m.g.f. uniquely determines the distribution. Several examples demonstrate using m.g.f.s to find distributions and moments.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "13.1", + "Moment", + "Generating", + "Functions", + "For", + "random", + "variable", + "define", + "moment", + "generating", + "function", + "begin", + "array", + "mathbb", + "left", + "sum", + "text", + "discrete", + "int", + "infty", + "continuous", + "end", + "called", + "find", + "moments", + "differentiating", + "evaluating", + "Note", + "prime", + "frac", + "Now", + "evaluate", + "cdot", + "Similarly", + "THEOREM", + "22", + "geq", + "Examples", + "Bernoulli", + "Recall", + "Thus", + "Binomial", + "sim", + "Bin", + "Bern", + "dots", + "independence", + "Poisson", + "If", + "lambda", + "recall", + "Calculus", + "exp", + "Exponential", + "valued", + "Standard", + "Normal", + "mathcal", + "sqrt", + "pi", + "mu", + "sigma", + "Property", + "Suppose", + "independent", + "Let", + "figure", + "Also", + "note", + "pdf", + "This", + "similar", + "underline", + "laplace", + "transform", + "sx", + "dx", + "correspondence", + "transforms", + "That", + "completely", + "determines", + "23", + "interval", + "distribution", + "Example1", + "Find", + "Solution", + "We", + "work", + "backwards", + "Match", + "table", + "Looks", + "quad", + "Summary", + "tX", + "mgf", + "distributions", + "The", + "helps", + "determine", + "variables", + "section", + "sums", + "Example2", + "Then", + "mathrm", + "Sol", + "So", + "check", + "mg", + "Example3", + "bin", + "Look", + "Example4", + "Question", + "What", + "doesn", + "match", + "read", + "First", + "Example5" + ], + "keywords": [ + "moment generating function", + "mgf", + "distribution" + ], + "connect_to": [] + } + }, + { + "chunk_id": "83c27522-7d2b-54ae-ae21-58663fa59ec4", + "type": "text", + "content": "", + "path": "Notes.pdf/CHAPTER 14", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "438540a7-c1db-5b11-8c1e-9d220ad7c89b", + "type": "text", + "content": "14.1. The Central Limit Theorem\n\\- The CLT is one of the most remarkable theorems in Probability.\n\\- It helps us understand why the empirical frequencies of so many natural populations exhibit bell-shaped (normal) curves.\n\\- Recall that $i.i.d.$ means independent and identically distributed random variables.\nTHEOREM 24. (CLT) Let $X_{1}, X_{2}, X_{3} \\ldots$ be i.i.d. each with mean $\\mu$ and variance $\\sigma^{2}$ . Then the distribution of\n$$\n\\frac {X _ {1} + \\cdots + X _ {n} - n \\mu}{\\sigma \\sqrt {n}}\n$$\ntends to the standard normal $Z$ as $n\\to \\infty$ . That is,\n$$\n\\mathbb {P} \\left(\\frac {X _ {1} + \\cdots + X _ {n} - n \\mu}{\\sigma \\sqrt {n}} \\leq b\\right) \\approx \\mathbb {P} (Z \\leq b) = \\Phi (b).\n$$\nwhen $n$ is large.\n\\- The CLT helps us approximate the probability of anything involving $X_{1} + \\cdots + X_{n}$ where $X_{i}$ are independent and identically distributed.\n\\- When approximating discrete distributions: USE the $\\pm .5$ continuity correction:\n\\- Example1: If 10 fair dice are rolled, find the approximate probability that the sum obtained is between 30 and 40, inclusive.\n\\- Solution: Let $X_{i}$ denote the value of the ith die. Recall that\n$$\n\\mathbb {E} \\left(X _ {i}\\right) = \\frac {7}{2} \\operatorname{Var} (X _ {i}) = \\frac {3 5}{1 2}.\n$$\nTake\n$$\nX = X _ {1} + \\dots + X _ {n}\n$$\nto be their sum.\n\\- Using the CLT we need\n$$\nn \\mu = 1 0 \\cdot \\frac {7}{2} = 3 5\n$$\n$$\n\\sigma \\sqrt {n} = \\sqrt {\\frac {3 5 0}{1 2}}\n$$\nthus using the continuity correction, then\n$$\n\\mathbb {P} (2 9. 5 \\leq X \\leq 4 0. 5) = \\mathbb {P} \\left(\\frac {2 9 . 5 - 3 5}{\\sqrt {\\frac {3 5 0}{1 2}}} \\leq \\frac {X - 3 5}{\\sqrt {\\frac {3 5 0}{1 2}}} \\leq \\frac {4 0 . 5 - 3 5}{\\sqrt {\\frac {3 5 0}{1 2}}}\\right)\n$$\n$$\n\\approx \\mathbb {P} (- 1. 0 1 8 4 \\leq Z \\leq 1. 0 1 8 4)\n$$\n$$\n= \\Phi (1. 0 1 8 4) - \\Phi (- 1. 0 1 8 4)\n$$\n$$\n= 2 \\Phi (1. 0 1 8 4) - 1 = . 6 9 2.\n$$\n\\- Example2: An instructor has 1000 exams that will be graded in sequence.\n- The times required to grade exam exam are i.i.d. with $\\mu = 20$ minutes and SD $\\sigma = 4$ minutes.\n- Approximate prob that the instructor will grade at least 25 exams in the first 450 minutes of work.\n- Solution:\n- Let $X_{i}$ be the time it takes to grade exam $i$ . Then\n$$\nX = X _ {1} + \\dots + X _ {2 5}\n$$\nis the time it takes to grade the first 25 exams. We want $\\mathbb{P}(X\\leq 450)$ .\n\\- Use CLT,\n$$\nn \\mu = 2 5 \\cdot 2 0 = 5 0 0\n$$\n$$\n\\sigma \\sqrt {n} = 4 \\sqrt {2 5} = 2 0.\n$$\n\\- Thus\n$$\n\\begin{array}{l} \\mathbb {P} (X \\leq 4 5 0) = \\mathbb {P} \\left(\\frac {X - 5 0 0}{2 0} \\leq \\frac {4 5 0 - 5 0 0}{2 0}\\right) \\\\ \\approx \\mathbb {P} (Z \\leq - 2. 5) \\\\ = 1 - \\Phi (2. 5) \\\\ = . 0 0 6. \\\\ \\end{array}\n$$", + "path": "Notes.pdf/CHAPTER 14/Limit Laws", + "metadata": { + "length": 2676, + "summary": "The Central Limit Theorem (CLT) states that for i.i.d. random variables with mean μ and variance σ², the distribution of the standardized sum tends to the standard normal as n→∞. It explains why many natural populations exhibit bell-shaped curves. Two examples illustrate its application: rolling 10 fair dice to approximate the probability the sum is between 30 and 40 (using continuity correction, result ≈0.692), and grading 25 exams with mean 20 min and SD 4 min, finding P(time ≤450 min)≈0.006.", + "page_nums": [], + "document_top_summary": "This document includes: CHAPTER 1, CHAPTER 2, CHAPTER 3, CHAPTER 4, CHAPTER 5, CHAPTER 6, CHAPTER 7, CHAPTER 8, CHAPTER 9, CHAPTER 10, CHAPTER 11, CHAPTER 12, CHAPTER 13, CHAPTER 14", + "tokens": [ + "14.1", + "The", + "Central", + "Limit", + "Theorem", + "CLT", + "remarkable", + "theorems", + "Probability", + "It", + "helps", + "understand", + "empirical", + "frequencies", + "natural", + "populations", + "exhibit", + "bell", + "shaped", + "normal", + "curves", + "Recall", + "means", + "independent", + "identically", + "distributed", + "random", + "variables", + "THEOREM", + "24", + "Let", + "ldots", + "mu", + "variance", + "sigma", + "Then", + "distribution", + "frac", + "cdots", + "sqrt", + "standard", + "infty", + "That", + "mathbb", + "left", + "leq", + "approx", + "Phi", + "large", + "approximate", + "probability", + "involving", + "When", + "approximating", + "discrete", + "distributions", + "USE", + "pm", + "continuity", + "correction", + "Example1", + "If", + "10", + "fair", + "dice", + "rolled", + "find", + "sum", + "obtained", + "30", + "40", + "inclusive", + "Solution", + "denote", + "ith", + "die", + "operatorname", + "Var", + "Take", + "dots", + "Using", + "cdot", + "Example2", + "An", + "instructor", + "1000", + "exams", + "graded", + "sequence", + "times", + "required", + "grade", + "exam", + "20", + "minutes", + "SD", + "Approximate", + "prob", + "25", + "450", + "work", + "time", + "takes", + "We", + "Use", + "Thus", + "begin", + "array", + "end" + ], + "keywords": [ + "CLT", + "normal approximation", + "continuity correction" + ], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/doc_nav.json b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/doc_nav.json new file mode 100644 index 000000000..17d450ff1 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/doc_nav.json @@ -0,0 +1,756 @@ +{ + "version": "1.0", + "file_name": "Probability Theory Lecture Notes.pdf", + "stats": { + "total_chunks": 109, + "text_chunks": 69, + "image_chunks": 28, + "table_chunks": 12, + "max_depth": 4 + }, + "sections": [ + { + "title": "Root", + "path": "Notes.pdf", + "level": 1, + "summary": "Probability Theory Lecture Notes Phanuel Mariano", + "chunk_count": 1, + "children": [] + }, + { + "title": "CHAPTER 1", + "path": "Notes.pdf/CHAPTER 1", + "level": 1, + "summary": "", + "chunk_count": 6, + "children": [ + { + "title": "Combinatorics", + "path": "Notes.pdf/CHAPTER 1/Combinatorics", + "level": 2, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "1.1. Counting Principle", + "path": "Notes.pdf/CHAPTER 1/Combinatorics/1.1. Counting Principle", + "level": 3, + "summary": "\\- We need a way to help us count faster rather than counting by hand one by one. FACT. (Basic Counting Principle) Suppose 2 experiments are to be performed. If one experiment can result in m possibilities Second experiment can result in $n$ possibilities Then together there are mn possibilities - I like to use the box method. For example. Each box represent the number of possibilities in that experiment. - Example1: There are 20 teachers and 100 students in a school. How many ways can we pick a teacher and student of the year? \\- Solution: Use the box Method: $20 \\times 100 = 2000$ . - The counting principle can be generalized to any amount of experiments: $n_1 \\cdots n_r$ possibilities - Example2: - A college planning committee consists of 3 freshmen, 4 sophomores, 5 juniors, and 2 seniors. - A subcommittee of 4 consists 1 person from each class. How many? - Solution: Box method $3 \\times 4 \\times 5 \\times 2 = 120$ . \\- Example3: How many differen 6-place license plates are possible if the first 3 places are to be occupied by letters and the finals 3 by numbers? - Solution: $26 \\cdot 26 \\cdot 26 \\cdot 10 \\cdot 10 \\cdot 10 = ?$ - Question...repetition is allowed? - Solution: $26 \\cdot 25 \\cdot 24 \\cdot 10 \\cdot 9 \\cdot 8$ \\- Example4: How many functions defined on $n$ points are possible if each functional value is either 0 or 1. \\- Solution: Box method on the $1, \\ldots, n$ points gives us $2^n$ possible functions.", + "chunk_count": 1, + "children": [] + }, + { + "title": "1.2. PERMUTATIONS", + "path": "Notes.pdf/CHAPTER 1/Combinatorics/1.2. PERMUTATIONS", + "level": 3, + "summary": "The text introduces permutations as ordered arrangements of objects, using examples like arranging letters a, b, c (6 ways via 3!). It explains the factorial formula n! for n distinct objects, with order mattering. Examples include batting orders (9!), grouping books (4! * 4! * 3! * 2! * 1!), and handling repetitions (e.g., arranging a, a, b, c as 4!/2! = 12). The general formula for permutations with alike objects is n!/(n1!...nr!), illustrated with PEPPER (6!/(3!2!1!) = 30) and tennis players (11!/(4!4!3!)).", + "chunk_count": 1, + "children": [] + }, + { + "title": "1.3. Combinations", + "path": "Notes.pdf/CHAPTER 1/Combinatorics/1.3. Combinations", + "level": 3, + "summary": "The passage explains the concept of combinations (order does not matter) versus permutations (order matters). It derives the formula for combinations: n choose r equals n!/((n-r)!r!). Examples include choosing a committee of 3 from 10 people, selecting 2 men and 3 women from groups, and inviting friends with constraints (feuding friends or those attending together). The Binomial Theorem is introduced: (x+y)^n = sum_{k=0}^n binom(n,k) x^k y^{n-k}, with a combinatorial proof. A combinatorial identity binom(n,r)=binom(n-1,r-1)+binom(n-1,r) is demonstrated using a committee selection analogy with a special president.", + "chunk_count": 1, + "children": [] + }, + { + "title": "1.4. Multinomial Coefficients", + "path": "Notes.pdf/CHAPTER 1/Combinatorics/1.4. Multinomial Coefficients", + "level": 3, + "summary": "The passage explains multinomial coefficients, which count ways to divide n objects into k groups of specified sizes. It provides examples: dividing 9 people into committees of sizes 3,4,2 yields 9!/(3!4!2!); assigning 10 police officers to duties (6 patrol, 2 station, 2 schools) gives 10!/(6!2!2!); ordering 10 flags (5 blue, 3 red, 2 yellow) gives 10!/(5!3!2!). It also covers distributing 8 indistinguishable balls into 3 boxes using stars and bars, resulting in C(10,8) or C(9,2)+9 ways.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "CHAPTER 2", + "path": "Notes.pdf/CHAPTER 2", + "level": 1, + "summary": "", + "chunk_count": 9, + "children": [ + { + "title": "Axioms of Probability", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability", + "level": 2, + "summary": "", + "chunk_count": 8, + "children": [ + { + "title": "2.1. Sample Space and Events", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.1. Sample Space and Events", + "level": 3, + "summary": "\\- We will have a sample space, denoted $S$ (sometimes $\\Omega$ , or $\\mathcal{U}$ ) that consists of all possible outcomes from an experiment.", + "chunk_count": 5, + "children": [ + { + "title": "- Example1:", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.1. Sample Space and Events/- Example1:", + "level": 4, + "summary": "\\* Experiment: Roll two dice, \\* Sample Space: $S =$ would be all possible pairs made up of the numbers one through six. List it here. $\\{(i,j):i,j = 1,\\dots 6\\}$ . 36 points.", + "chunk_count": 1, + "children": [] + }, + { + "title": "- Example 2:", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.1. Sample Space and Events/- Example 2:", + "level": 4, + "summary": "\\* Experiment: Toss a coin twice \\* $S = \\{HH,HT,TH,TT\\}$", + "chunk_count": 1, + "children": [] + }, + { + "title": "- Example3:", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.1. Sample Space and Events/- Example3:", + "level": 4, + "summary": "\\* Experiment: Measuring the number of accidents of a random person before they had turn 18. $$ \\cdot S = \\{0, 1, 2, \\dots \\} $$", + "chunk_count": 1, + "children": [] + }, + { + "title": "- Others:", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.1. Sample Space and Events/- Others:", + "level": 4, + "summary": "The passage defines events as subsets of a sample space S, and explains basic set operations: union (A ∪ B), intersection (A ∩ B), and complement (A^c). It provides examples from rolling two dice and age at death. It also covers properties like commutativity, associativity, distributivity, DeMorgan's Laws, and the concept of disjoint (mutually exclusive) events.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "2.2. Axioms of Probability", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.2. Axioms of Probability", + "level": 3, + "summary": "The text defines probability through axioms: non-negativity, total probability of sample space equals 1, and countable additivity for disjoint events. It introduces relative frequency as motivation but notes its limitations. Key properties derived include probability of empty set is 0, complement rule, monotonicity, and inclusion-exclusion for two events. Examples illustrate fair and biased coin tosses, die rolling, and a basketball game probability calculation using DeMorgan's law and the inclusion-exclusion formula.", + "chunk_count": 1, + "children": [] + }, + { + "title": "2.3. Equally Likely Outcomes", + "path": "Notes.pdf/CHAPTER 2/Axioms of Probability/2.3. Equally Likely Outcomes", + "level": 3, + "summary": "The passage covers basic probability concepts with equally likely outcomes, including examples like rolling dice, drawing balls from an urn, selecting committees, and poker hands. It explains how to compute probabilities using combinatorial formulas, such as combinations and permutations, and includes the birthday problem as an illustration of probability calculations.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "CHAPTER 3", + "path": "Notes.pdf/CHAPTER 3", + "level": 1, + "summary": "", + "chunk_count": 2, + "children": [ + { + "title": "Independence", + "path": "Notes.pdf/CHAPTER 3/Independence", + "level": 2, + "summary": "The text defines independent events as those where the probability of their intersection equals the product of their individual probabilities. It provides examples including coin flips, card draws, and dice rolls. It proves that if events E and F are independent, then E and F^c are also independent. The definition extends to three or more events requiring pairwise and joint independence. Applications include calculating probabilities for drawing balls from urns and the Gambler's Ruin problem, where the probability of reaching a target before ruin is linear. The text also introduces Bernoulli trials and the binomial distribution for independent trials with constant success probability.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "CHAPTER 4", + "path": "Notes.pdf/CHAPTER 4", + "level": 1, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "Conditional Probability and Independence", + "path": "Notes.pdf/CHAPTER 4/Conditional Probability and Independence", + "level": 2, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "4.1. Conditional Probabilities", + "path": "Notes.pdf/CHAPTER 4/Conditional Probability and Independence/4.1. Conditional Probabilities", + "level": 3, + "summary": "The text defines conditional probability P(E|F)=P(E∩F)/P(F) for P(F)>0, and explains that if P(E|F)=P(E) then E and F are independent with P(E∩F)=P(E)P(F). It provides several examples: rolling two dice to find P(sum=8|first die=3)=1/6; drawing marbles from a box (P(second red|first red)=1/2); Landon's textbook problem (P(Union|not Monteith)=2/3); card drawing with Annabelle and Bobby (P(Bobby has one ace|Annabelle has two aces)=C(2,1)C(37,12)/C(39,13)); urn with replacement and additional balls (P(3 white)=5/12*6/13*7/14, P(1 white)=3*5*7*8/(12*13*14)); Phan's course selection (P(A∩C)=1/2*1/7=1/14); and a table of 500 married couples' salaries (P(husband<25k)=248/500, P(wife>25k|husband>25k)=54/252≈0.214, P(wife>25k|husband<25k)=36/248≈0.145). The general multiplication rule for n events is also given.", + "chunk_count": 1, + "children": [] + }, + { + "title": "4.2. Bayes's Formula", + "path": "Notes.pdf/CHAPTER 4/Conditional Probability and Independence/4.2. Bayes's Formula", + "level": 3, + "summary": "The text explains the Law of Total Probability and Bayes' Theorem with examples. The Law of Total Probability computes P(E) by summing over mutually exclusive events: P(E) = Σ P(E|Fi)P(Fi). Bayes' Theorem finds conditional probabilities: P(Fj|E) = P(E|Fj)P(Fj) / Σ P(E|Fi)P(Fi). Examples include insurance accident risk, HIV test accuracy (98% accurate, 0.5% prevalence yields 19.8% positive predictive value), class grade probabilities, factory defect rates (machines I, II, III with 2%, 1%, 3% defects and 35%, 25%, 40% production), and multiple-choice test knowledge (probability student knew answer given correct response is mp/(1+(m-1)p)).", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "CHAPTER 5", + "path": "Notes.pdf/CHAPTER 5", + "level": 1, + "summary": "", + "chunk_count": 8, + "children": [ + { + "title": "Random Variables", + "path": "Notes.pdf/CHAPTER 5/Random Variables", + "level": 2, + "summary": "", + "chunk_count": 7, + "children": [ + { + "title": "5.1. Random Variables", + "path": "Notes.pdf/CHAPTER 5/Random Variables/5.1. Random Variables", + "level": 3, + "summary": "The passage introduces the concept of a random variable as a function mapping outcomes of an experiment to real numbers, enabling calculus on outcomes. Examples include coin tosses (X=1 for heads, 0 for tails), die rolls (X equals the outcome), and defining Y based on odd/even outcomes. It explains determining the range of X, such as number of heads in 10 tosses (0-10) or liability amounts. A detailed example with 3 coin tosses shows assigning probabilities to X values (0-3), summing to 1.", + "chunk_count": 1, + "children": [] + }, + { + "title": "5.2. Discrete Random Variables", + "path": "Notes.pdf/CHAPTER 5/Random Variables/5.2. Discrete Random Variables", + "level": 3, + "summary": "The text defines a discrete random variable as one with at most a countable number of possible values. It introduces the probability mass function (pmf) p(x) = P(X=x), which maps real numbers to [0,1]. Key properties include p(x_i) > 0 for attainable values and sum of all p(x_i) = 1. Examples illustrate a coin toss with a bar graph and a Poisson pmf p(i)=e^{-λ} λ^i / i! for i=0,1,2,... The Poisson example computes P(X=0)=e^{-λ} and P(X>2)=1 - e^{-λ} - λ e^{-λ} - (λ^2 e^{-λ})/2.", + "chunk_count": 1, + "children": [] + }, + { + "title": "5.3. Expected Value", + "path": "Notes.pdf/CHAPTER 5/Random Variables/5.3. Expected Value", + "level": 3, + "summary": "The passage introduces the concept of expectation (expected value) for discrete random variables, defined as the sum of outcomes weighted by their probabilities. Examples include a coin flip (E[X]=0.5), a fair die (E[X]=3.5), and a geometric distribution for tornado counts (E[X]=1). It then defines the cumulative distribution function (CDF) as F(x)=P(X≤x), which is nondecreasing, right-continuous, and has limits 0 and 1 at ±∞. Properties and examples show how to compute probabilities like P(X b \\end{array} \\right.. $$ \\- Example1: Suppose $X \\sim \\text{Uniform}(a, b)$ Part (a) Find the mean of $X$ . Part (b) Find the variance of $X$ . \\- Part (a): We compute $$ \\begin{array}{l} \\mathbb {E} X = \\int_ {- \\infty} ^ {\\infty} x f _ {X} (x) d x = \\int_ {a} ^ {b} x \\frac {1}{b - a} d x \\\\ = \\frac {1}{b - a} \\left(\\frac {b ^ {2}}{2} - \\frac {a ^ {2}}{2}\\right) = \\frac {a + b}{2}. \\\\ \\end{array} $$ \\* Which makes sense right? It should be the midpoint of the interval $[a, b]$ . \\- Part(b): We compute first the second moment $$ \\begin{array}{l} \\mathbb {E} X ^ {2} = \\int_ {a} ^ {b} x ^ {2} \\frac {1}{b - a} d x...frac {1}{b - a} (b - a) (a ^ {2} + a b + b ^ {2}) \\\\ = \\frac {a ^ {2} + a b + b ^ {2}}{3}. \\\\ \\end{array} $$ Thus after some algebra $$ \\operatorname{Var} X = \\frac {a ^ {2} + a b + b ^ {2}}{3} - \\left(\\frac {a + b}{2}\\right) ^ {2} = \\frac {(b - a) ^ {2}}{1 2}. $$", + "chunk_count": 1, + "children": [] + }, + { + "title": "7.4. More practice", + "path": "Notes.pdf/CHAPTER 7/Continuous Random Variables/7.4. More practice", + "level": 3, + "summary": "\\- Suppose we are given the pd.f. $$ f (x) = \\left\\{ \\begin{array}{l l} 9 e ^ {- 9 x} & x \\geq 0 \\\\ 0 & x < 0 \\end{array} \\right. $$ \\- Part (a): Set up integral to find $F_{X}(x)$ : \\* We have for $x > 0$ , that $$ F _ {X} (x) = \\int_ {0} ^ {x} 9 e ^ {- 9 y} d y = 1 - e ^ {- 9 x}, $$ so that $$ F _ {X} (x) = \\left\\{ \\begin{array}{l l} 1 - e ^ {- 9 x} & x \\geq 0 \\\\ 0 & x < 0 \\end{array} \\right. / $$ \\- Part (b): Set up integral to find $\\mathbb{P}(1 < X < 5)$ $$ * \\int_ {1} ^ {5} 9 e ^ {- 9 x} d x $$ \\- Part (c): Set up integral to find $\\mathbb{P}(X > 3)$ $$ * \\int_ {3} ^ {\\infty} 9 e ^ {- 9 x} d x. $$ \\- Part (s): Set up integral to find $\\mathbb{P}(X < 2)$ $$ * \\int_ {0} ^ {2} 9 e ^ {- 9 x} d x. $$", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "CHAPTER 8", + "path": "Notes.pdf/CHAPTER 8", + "level": 1, + "summary": "", + "chunk_count": 2, + "children": [ + { + "title": "Normal Distributions", + "path": "Notes.pdf/CHAPTER 8/Normal Distributions", + "level": 2, + "summary": "The text covers the normal (Gaussian) distribution, its density function, properties, and the standard normal random variable Z. It explains that any normal variable X with mean μ and variance σ² can be standardized as Z = (X - μ)/σ, which follows N(0,1). Key results include the cumulative distribution function Φ(x), symmetry property Φ(-x)=1-Φ(x), and the 68-95-99.7 rule stating that about 68%, 95%, and 99.7% of data lies within 1, 2, and 3 standard deviations of the mean, respectively. Examples demonstrate calculating probabilities using Z-scores and finding standard deviation from a given probability.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "CHAPTER 9", + "path": "Notes.pdf/CHAPTER 9", + "level": 1, + "summary": "", + "chunk_count": 2, + "children": [ + { + "title": "Normal approximations to the binomial", + "path": "Notes.pdf/CHAPTER 9/Normal approximations to the binomial", + "level": 2, + "summary": "This passage explains the normal approximation to the binomial distribution. It states Theorem 13, which says that as n grows large, the standardized binomial variable (Sn - np)/sqrt(np(1-p)) converges in distribution to a standard normal Z. It notes that Sn is discrete while the normal is continuous, so a continuity correction is needed: P(Sn = i) is approximated by P(i-0.5 < Sn < i+0.5). An example with 100 fair coin tosses is given: (a) P(more than 60 heads) is approximated using 60.5, yielding about 0.0179; (b) P(exactly 60 heads) is approximated by P(59.5 ≤ Sn ≤ 60.5), giving Φ(2.1)-Φ(1.9).", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "CHAPTER 10", + "path": "Notes.pdf/CHAPTER 10", + "level": 1, + "summary": "", + "chunk_count": 8, + "children": [ + { + "title": "Some continuous distributions", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions", + "level": 2, + "summary": "", + "chunk_count": 7, + "children": [ + { + "title": "10.1. Exponential Random Variables", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.1. Exponential Random Variables", + "level": 3, + "summary": "The exponential distribution is a continuous probability distribution with parameter λ, defined by its probability density function f(x)=λe^{-λx} for x≥0. Its cumulative distribution function is F(a)=1-e^{-λa}, mean is 1/λ, and variance is 1/λ². It models the time until a specific event occurs, such as phone call length, earthquake occurrence, or device lifetime. The distribution is memoryless, meaning P(X>t+s|X>t)=P(X>s). Examples include calculating probabilities for phone calls with average length 10 minutes, iPhone lifetime with mean 4 years, and accident occurrence for bad drivers where 30% have accidents in the first 50 days.", + "chunk_count": 1, + "children": [] + }, + { + "title": "10.2. Other Continuous Distributions", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.2. Other Continuous Distributions", + "level": 3, + "summary": "", + "chunk_count": 4, + "children": [ + { + "title": "• Gamma Distribution:", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.2. Other Continuous Distributions/• Gamma Distribution:", + "level": 4, + "summary": "\\- We say $X \\sim Gamma(\\alpha, \\lambda)$ has density $$ f (x) = \\left\\{ \\begin{array}{l l} \\frac {\\lambda e ^ {- \\lambda x} (\\lambda x) ^ {\\alpha - 1}}{\\Gamma (\\alpha)} & x \\geq 0 \\\\ 0 & x < 0 \\end{array} \\right. $$ where $\\Gamma (\\alpha)$ is the Gamma function $$ \\Gamma (\\alpha) = \\int_ {0} ^ {\\infty} e ^ {- y} y ^ {\\alpha - 1} d y. $$ - If $Y \\sim Gamma\\left(\\frac{n}{2}, \\frac{1}{2}\\right) = \\chi_n^2$ , this is called the Chi-Squared distribution. - The chi-square distribution is used a lot in statistics. \\* Its mean is $\\mathbb{E}X = \\frac{\\alpha}{\\lambda}$ and $\\operatorname{Var}X = \\frac{\\alpha}{\\lambda^2}$ .", + "chunk_count": 1, + "children": [] + }, + { + "title": "• Weibull Distribution:", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.2. Other Continuous Distributions/• Weibull Distribution:", + "level": 4, + "summary": "- Usefull in engineering: Look in the book for its pdf. - $X =$ . If there is an object consisting many parts, and suppose that the object experiences death once any of its parts fails. $X =$ lifetime of this object.", + "chunk_count": 1, + "children": [] + }, + { + "title": "• Cauchy Distribution:", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.2. Other Continuous Distributions/• Cauchy Distribution:", + "level": 4, + "summary": "\\- We say $X$ is cauchy with parameter $-\\infty < \\theta < \\infty$ if $$ f (x) = \\frac {1}{\\pi} \\frac {1}{1 + (x - \\theta) ^ {2}}. $$ - Importance: It does not have finite mean: That is $\\mathbb{E}X = \\infty$ . - To see this, We compute for $\\theta = 0$ $$ \\mathbb {E} X = \\frac {1}{\\pi} \\int_ {- \\infty} ^ {\\infty} \\frac {x}{1 + x ^ {2}} d x $$ $$ \\sim \\frac {1}{\\pi} \\int_ {- \\infty} ^ {\\infty} \\frac {1}{x} d x $$ $$ \\sim \\lim _ {x \\to \\infty} \\ln | x | - \\ln \\lim _ {x \\to - \\infty} | x | $$ which is not defined.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "10.3. The distribution function of a Random variable", + "path": "Notes.pdf/CHAPTER 10/Some continuous distributions/10.3. The distribution function of a Random variable", + "level": 3, + "summary": "The passage explains how to derive the probability density function (PDF) of a transformed continuous random variable Y = g(X) using the cumulative distribution function (CDF) method. It begins with the fundamental relationship F'(x)=f(x) and demonstrates through multiple examples: Y=2X, Y=e^{3X} with uniform X, Y=-ln X yielding exponential(1), Y=tan X yielding Cauchy(0), and an actuarial example with Y=T^2. The key steps are: write the CDF of Y in terms of X's CDF, then differentiate using the chain rule to obtain the PDF. The method relies on g being strictly increasing so the inverse exists.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "CHAPTER 11", + "path": "Notes.pdf/CHAPTER 11", + "level": 1, + "summary": "", + "chunk_count": 6, + "children": [ + { + "title": "Multivariate distributions", + "path": "Notes.pdf/CHAPTER 11/Multivariate distributions", + "level": 2, + "summary": "", + "chunk_count": 5, + "children": [ + { + "title": "11.1. Joint distribution functions", + "path": "Notes.pdf/CHAPTER 11/Multivariate distributions/11.1. Joint distribution functions", + "level": 3, + "summary": "The passage discusses joint probability distributions for discrete and continuous random variables. For discrete variables, it defines the joint probability mass function (pmf) p(x,y) = P(X=x, Y=y) with properties 0≤p≤1 and sum=1, and the multivariate cdf F(x,y)=P(X≤x,Y≤y). An example with two 3-sided dice finds the joint pmf of X (largest value) and Y (sum), and computes a conditional probability. For continuous variables, it defines the joint probability density function (pdf) f(x,y) with properties f≥0 and double integral=1, and the cdf as the double integral of f. Marginal densities are obtained by integrating out the other variable. A detailed example with joint pdf f(x,y)=2e^{-x}e^{-2y} for x,y>0 finds the normalizing constant c=2, computes P(X1,Y<1), finds marginal f_X(x)=e^{-x}, and computes E[X]=1.", + "chunk_count": 1, + "children": [] + }, + { + "title": "11.2. Independent Random Variables", + "path": "Notes.pdf/CHAPTER 11/Multivariate distributions/11.2. Independent Random Variables", + "level": 3, + "summary": "This text defines independence for discrete and continuous random variables (RVs) via joint pmf/pdf factorization and joint CDF. It provides examples checking independence by verifying if joint pdf equals product of marginals, highlighting cases where domain restrictions (non-rectangular) prevent factorization. It then covers sums of independent RVs, deriving the convolution formula for pdf/cdf. Key results include: sum of independent normals is normal with summed means and variances; sum of independent Poissons is Poisson with summed rates; sum of squared standard normals follows chi-squared distribution. Examples compute probabilities for average grades and difference of grades using normal distribution properties.", + "chunk_count": 1, + "children": [] + }, + { + "title": "11.4. Conditional Distributions- Discrete(★)", + "path": "Notes.pdf/CHAPTER 11/Multivariate distributions/11.4. Conditional Distributions- Discrete(★)", + "level": 3, + "summary": "The text defines conditional probability mass functions (PMF) for discrete random variables and conditional probability density functions (PDF) for continuous random variables. It provides formulas: p_{X|Y}(x|y) = p(x,y)/p_Y(y) for discrete, and f_{X|Y}(x|y) = f(x,y)/f_Y(y) for continuous, defined only when the marginal is positive. It notes that if X and Y are independent, the conditional distribution equals the marginal. Examples include computing conditional PMF from a joint table and checking independence, and computing conditional PDF from a given joint PDF.", + "chunk_count": 1, + "children": [] + }, + { + "title": "11.6. Joint PDF of functions", + "path": "Notes.pdf/CHAPTER 11/Multivariate distributions/11.6. Joint PDF of functions", + "level": 3, + "summary": "This passage explains how to find the joint probability density function (pdf) of new random variables that are functions of two original random variables. It outlines a three-step method: first, compute the Jacobian determinant of the transformation; second, solve for the original variables in terms of the new ones; third, apply the formula using the original joint pdf and the absolute inverse Jacobian. Three examples are provided: one with a uniform-like joint pdf and linear transformations, one with independent normal variables and linear combinations, and one with a polynomial joint pdf and a nonlinear transformation. The method is general for invertible transformations.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "CHAPTER 12", + "path": "Notes.pdf/CHAPTER 12", + "level": 1, + "summary": "The text covers expectation properties for sums of random variables, including formulas for discrete and continuous cases. It defines covariance and correlation, explaining their roles in measuring linear relationships between variables. Examples illustrate computing expectations, variances, covariance, and correlation for given joint distributions.", + "chunk_count": 1, + "children": [] + }, + { + "title": "CHAPTER 13", + "path": "Notes.pdf/CHAPTER 13", + "level": 1, + "summary": "", + "chunk_count": 2, + "children": [ + { + "title": "Moment generating functions", + "path": "Notes.pdf/CHAPTER 13/Moment generating functions", + "level": 2, + "summary": "This text defines moment generating functions (m.g.f.) for random variables, showing how to compute them for discrete and continuous cases. It explains that moments can be found by differentiating the m.g.f. and evaluating at zero. Examples are given for Bernoulli, Binomial, Poisson, Exponential, and Normal distributions. Properties include that the m.g.f. of a sum of independent variables is the product of their m.g.f.s, and that the m.g.f. uniquely determines the distribution. Several examples demonstrate using m.g.f.s to find distributions and moments.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "CHAPTER 14", + "path": "Notes.pdf/CHAPTER 14", + "level": 1, + "summary": "", + "chunk_count": 2, + "children": [ + { + "title": "Limit Laws", + "path": "Notes.pdf/CHAPTER 14/Limit Laws", + "level": 2, + "summary": "The Central Limit Theorem (CLT) states that for i.i.d. random variables with mean μ and variance σ², the distribution of the standardized sum tends to the standard normal as n→∞. It explains why many natural populations exhibit bell-shaped curves. Two examples illustrate its application: rolling 10 fair dice to approximate the probability the sum is between 30 and 40 (using continuity correction, result ≈0.692), and grading 25 exams with mean 20 min and SD 4 min, finding P(time ≤450 min)≈0.006.", + "chunk_count": 1, + "children": [] + } + ] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-_- Events_.jpg", + "summary": "image-1 \\- Events: An event $A$ is a subset of $S$ . In this case we use the notation $A \\subset S$ , to mean $A$ is a subset of $S$ .0" + }, + { + "path": "images/image-2-- $A _cup .jpg", + "summary": "image-2 - $A \\cup B$ : points in $S$ such that is in $A$ OR $B$ OR BOTH.0" + }, + { + "path": "images/image-3-- $A _cap .jpg", + "summary": "image-3 - $A \\cap B$ , points in $A$ AND $B$ . (you may also see $AB$ )0" + }, + { + "path": "images/image-4-- $A^c$ is.jpg", + "summary": "image-4 - $A^c$ is the compliment of $A$ , the points NOT in $A$ . (you may also see $A'$ )0" + }, + { + "path": "images/image-5-- Can exte.jpg", + "summary": "image-5 - Can extend to $A_1, \\ldots, A_n$ events. $\\bigcup_{i=1}^{n} A_i$ and $\\bigcap_{i=1}^{n} A_i$ . • Example1: Roll two dice.0" + }, + { + "path": "images/image-6-• Example1.jpg", + "summary": "image-6 • Example1: Roll two dice.0" + }, + { + "path": "images/image-7-• Example1.jpg", + "summary": "image-7 • Example1: Roll two dice.0" + }, + { + "path": "images/image-8-• Example1.jpg", + "summary": "image-8 • Example1: Roll two dice.0" + }, + { + "path": "images/image-9-• Example1.jpg", + "summary": "image-9 • Example1: Roll two dice.0" + }, + { + "path": "images/image-10-PROPOSITIO.jpg", + "summary": "image-10 PROPOSITION 2. If $E$ and $F$ are independent, then $E$ and $F^c$ are independent. \\- Remark: Independence and mutually exclusive, are two different things!0" + }, + { + "path": "images/image-11-Step Function Graph.jpg", + "summary": "image-11 - Chart Type: Step function plotted on a Cartesian coordinate system - X-axis Range: -1 to 4 (major grid lines at intervals of 0.5) - Y-axis Range: -1 to 1 (major grid lines at intervals of 0.5) - Data Segments: - y = 0 for x < 0 - y = 0.2 for 0 ≤ x < 1 - y = 0.5 for 1 ≤ x < 2 - y = 0.8 for 2 ≤ x < 3 - y = 1 for x ≥ 3" + }, + { + "path": "images/image-12-PROOF. We .jpg", + "summary": "image-12 PROOF. We start with the first definition. Let $X = x_{1}, x_{2}, \\ldots$ \\- Using this definition, we can prove linearity of the expectation.0" + }, + { + "path": "images/image-13-Thus plugg.jpg", + "summary": "image-13 Thus plugging this back into $(\\star)$ we have that - Remark: $\\mathbb{E}X^{2} = \\sum x_{i}^{2}p(x_{i})$ .0" + }, + { + "path": "images/image-14-DEFINITION.jpg", + "summary": "image-14 DEFINITION. We define # CHAPTER 60" + }, + { + "path": "images/image-15-$$ _- Summ.jpg", + "summary": "image-15 $$ \\- Summary of Theorem: This theorem says that suppose $n$ is large and $p$ is small, Thus0" + }, + { + "path": "images/image-16-$$ PROPOSI.jpg", + "summary": "image-16 $$ PROPOSITION. If $X$ is a continuous $R.V.$ with pdf $f(x)$ , then for any real valued function $g$ ,0" + }, + { + "path": "images/image-17-Standard Normal Distribution Curve.jpg", + "summary": "image-17 - Chart Type: Bell curve representing a standard normal distribution (Gaussian distribution) - X-axis Range: -3 to 3 - Center Point: 0 (representing the mean and median) - Key Features: Symmetrical shape with peak at x=0, tails extending towards negative and positive infinity" + }, + { + "path": "images/image-10-PROPOSITIO.jpg_2", + "summary": "image-10 PROPOSITION 2. If $E$ and $F$ are independent, then $E$ and $F^c$ are independent. \\- Remark: Independence and mutually exclusive, are two different things!0" + }, + { + "path": "images/image-18-Standard Normal Distribution Graph.jpg", + "summary": "image-18 - Chart Title: φ(Z) (PDF function of standard normal distribution) - X-Axis Label: Z (Standard Score / Z-score) - Y-Axis Label: φ(Z) (Probability Density Function) - Key Features: - Bell-shaped curve centered at zero. - Blue shaded area under the curve represents the Cumulative Distribution Function (CDF). - An arrow labeled \"CDF\" points to the shaded region, indicating the probability accumulated up to a specific Z value. - A vertical line marks a specific point on the X-axis labeled \"Z\", defining the upper bound of the shaded CDF area." + }, + { + "path": "images/image-19-Normal Distribution Curve.jpg", + "summary": "image-19 - Chart Title: Normal Distribution (Bell Curve) - Axis Labels: Horizontal axis represents standard deviations from the mean ($\\mu$), marked as $-3\\sigma, -2\\sigma, -1\\sigma, \\mu, 1\\sigma, 2\\sigma, 3\\sigma$. Vertical axis is unlabeled but represents probability density. - Key Data Points & Percentages: - Central Region ($\\mu \\pm 1\\sigma$): Two sections of 34.1% each (Total 68.2%) - Second Region ($\\pm 1\\sigma$ to $\\pm 2\\sigma$): Two sections of 13.6% each (Total 27.2%) - Third Region ($\\pm 2\\sigma$ to $\\pm 3\\sigma$): Two sections of 2.1% each (Total 4.2%) - Outer Regions (Beyond $\\pm 3\\sigma$): Two sections of 0.1% each (Total 0.2%)" + }, + { + "path": "images/image-20-PROOF. To .jpg", + "summary": "image-20 PROOF. To show this we have • Example3:(Exam P Q29)0" + }, + { + "path": "images/image-21-Graph of Inequality.jpg", + "summary": "image-21 - Chart Type: Cartesian coordinate system showing a shaded region - X-axis range: -1 to 2.5 (major grid lines at intervals of 0.5) - Y-axis range: -0.5 to 1.5 (major grid lines at intervals of 0.5) - Shaded Region: The area where x ≥ 0 and y ≥ 0 (First Quadrant) - Boundary Line: Vertical line along the Y-axis (x = 0) - Color: Light orange shading fills the positive quadrant" + }, + { + "path": "images/image-22-Linear Inequality Region.jpg", + "summary": "image-22 - Chart Type: 2D Cartesian coordinate plot with a shaded region - X-axis range: -0.4 to 1.2 (major grid lines at intervals of 0.2) - Y-axis range: -0.2 to 1.0 (major grid lines at intervals of 0.2) - Shaded Region: A triangular area filled in orange - Boundary Lines: - Vertical line at x = 0 (y-axis) - Horizontal line at y = 1 - Diagonal line connecting origin (0,0) to point (1,1), representing the equation y = x - Region Definition: The shaded area represents the inequality where x ≥ 0, y ≤ 1, and y ≥ x" + }, + { + "path": "images/image-23-Graph of Inequality.jpg", + "summary": "image-23 - Chart Type: Cartesian coordinate system showing a shaded region representing an inequality. - X-axis: Labeled with integers from -2 to 6. - Y-axis: Labeled with integers from -1 to 2. - Shaded Region: An orange rectangular area covering the first quadrant. - Boundaries: The region is bounded on the left by the vertical line x = 1 and below by the horizontal line y = 0 (the x-axis). - Mathematical Representation: The graph visually represents the inequality $x \\ge 1$ for $y \\ge 0$." + }, + { + "path": "images/image-24-$$ _- Exam.jpg", + "summary": "image-24 $$ \\- Example1: Let $X, Y$ be r.v. with joint pdf0" + }, + { + "path": "images/image-25-Linear Function Graph.jpg", + "summary": "image-25 Chart Title: None visible Axes: X-axis ranges from -0.2 to 1.2; Y-axis ranges from -0.2 to 1.0 Data Representation: A linear function plotted as a straight line starting at the origin (0,0) and ending at approximately (1,1). Shaded Region: The area under the curve (between the line and the x-axis) is shaded in orange. Grid: Background grid lines are present for both axes with increments of 0.2." + }, + { + "path": "images/image-26-(b) If $X .jpg", + "summary": "image-26 (b) If $X \\leq Y$ then $\\mathbb{E}X \\leq \\mathbb{E}Y$ . \\- Example4: Let $X_{1}, \\ldots, X_{n}$ be independent and identically distributed random (i.i.d.) random variables. Suppose $\\mathbb{E}X_{i} = \\mu$ . We call the quantity0" + }, + { + "path": "images/image-27-Scatter Plot with Trend Line.jpg", + "summary": "image-27 Chart Type: Scatter plot with linear regression line (red) and data points (blue). X-Axis Range: -20 to 60. Y-Axis Range: 5 to 15. Key Data Points: - X-axis ticks at intervals of 10 (-20, -10, 0, 10, 20, 30, 40, 50, 60). - Y-axis ticks at intervals of 5 (5, 10, 15). Trend Analysis: The blue dots show a positive correlation, indicating that as the x-value increases, the y-value generally increases. The red line represents the best-fit trend through these points." + } + ], + "tables": [ + { + "path": "tables/table-0 Income by Gender.html", + "summary": "table-1 A table shows counts of couples based on wife's and husband's income levels, with 212 couples both under $25k and 54 both over $25k." + }, + { + "path": "tables/table-1 Data Table.html", + "summary": "table-2 The table shows six data points with x values from -0.5 to 3.5 and corresponding y values from 0.0 to 1.0, indicating a non-linear relationship." + }, + { + "path": "tables/table-2 Symmetric Data.html", + "summary": "table-3 The table shows x values from -3.0 to 3.0 with corresponding y values peaking at 1.000 when x=0.0, decreasing symmetrically to 0.000 at the extremes." + }, + { + "path": "tables/table-3 Z and CDF Table.html", + "summary": "table-4 The table shows Z values and their corresponding CDF, with entries for 0 and z mapping to 0 and Peak respectively." + }, + { + "path": "tables/table-4 Normal Distribution.html", + "summary": "table-5 The table shows the percentage of data within each standard deviation from the mean in a normal distribution, with 34.1% at the mean and decreasing percentages outward." + }, + { + "path": "tables/table-5 Table of Pairs.html", + "summary": "table-6 The table shows pairs (X,Y) for outcomes 1-3, with values like (1,2), (2,3), (3,4), (2,4), (3,5), and (3,6)." + }, + { + "path": "tables/table-6 Joint Probability Table.html", + "summary": "table-7 A joint probability table for X and Y with values 1-3 and 2-6, showing probabilities like P(X=1,Y=2)=1/9, P(X=2,Y=3)=2/9, etc." + }, + { + "path": "tables/table-7 Linear Data Table.html", + "summary": "table-8 The table shows six pairs of x and y values from 0.0 to 1.0, where y equals x, indicating a perfect linear relationship." + }, + { + "path": "tables/table-8 Data Table.html", + "summary": "table-9 The table contains a single row with X=1 and Y=1, showing a simple data pair." + }, + { + "path": "tables/table-9 Linear Data Table.html", + "summary": "table-10 The table shows six data points where x and y values are equal, ranging from 0 to 1, indicating a perfect linear relationship." + }, + { + "path": "tables/table-10 Joint Probability.html", + "summary": "table-11 The table shows joint probabilities for two binary variables X and Y, with values: P(0,0)=0.4, P(0,1)=0.2, P(1,0)=0.1, P(1,1)=0.3." + }, + { + "path": "tables/table-11 Linear Data Table.html", + "summary": "table-12 The table shows a linear relationship between x and y, where y increases by 1 for every 5-unit increase in x, ranging from x=-20 to x=60." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/full.md b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/full.md new file mode 100644 index 000000000..03c341575 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/full.md @@ -0,0 +1,4426 @@ +## Probability Theory Lecture Notes + +Phanuel Mariano + +## Contents + +## Chapter 1. Combinatorics 5 + +1.1. Counting Principle 5 +1.2. Permutations 6 +1.3. Combinations 7 +1.4. Multinomial Coefficients 9 + +## Chapter 2. Axioms of Probability 10 + +2.1. Sample Space and Events 10 +2.2. Axioms of Probability 12 +2.3. Equally Likely Outcomes 15 + +## Chapter 3. Independence 18 + +3.1. Independent Events 18 + +## Chapter 4. Conditional Probability and Independence 22 + +4.1. Conditional Probabilities 22 +4.2. Bayes's Formula 26 + +## Chapter 5. Random Variables 29 + +5.1. Random Variables 29 +5.2. Discrete Random Variables 31 +5.3. Expected Value 32 +5.4. The C.D.F. 34 +5.5. Expected Value of Sums of Random Variables 37 +5.6. Expectation of a Function of a Random Variable 39 +5.7. Variance 41 + +## Chapter 6. Some Discrete Distributions 43 + +6.1. Bernouli and Binomial Random Variables 43 +6.2. The Poisson Distribution 46 +6.3. Other Discrete Distributions 49 + +## Chapter 7. Continuous Random Variables 53 + +7.1. Intro to continuous R.V 53 +7.2. Expectation and Variance 55 +7.3. The uniform Random Variable 58 +7.4. More practice 59 + +## Chapter 8. Normal Distributions 60 + +## CONTENTS + +8.1. The normal distribution 60 + +Chapter 9. Normal approximations to the binomial 64 + +9.1. The normal approximates Binomial 64 + +Chapter 10. Some continuous distributions 66 + +10.1. Exponential Random Variables 66 + +10.2. Other Continuous Distributions 69 + +10.3. The distribution function of a Random variable 70 + +Chapter 11. Multivariate distributions 74 + +11.1. Joint distribution functions 74 + +11.2. Independent Random Variables 78 + +11.3. Sums of independent Random Variables( $\star$ ) 80 + +11.4. Conditional Distributions- Discrete( $\star$ ) 82 + +11.5. Conditional Distributions- Continuous( $\star$ ) 83 + +11.6. Joint PDF of functions 84 + +Chapter 12. Expectations 87 + +12.1. Expectation of Sums of R.V. 87 + +12.2. Covariance and Correlations. 90 + +Chapter 13. Moment generating functions 93 + +13.1. Moment Generating Functions 93 + +Chapter 14. Limit Laws 97 + +14.1. The Central Limit Theorem 97 + +## CHAPTER 1 + +## Combinatorics + +## 1.1. Counting Principle + +\- We need a way to help us count faster rather than counting by hand one by one. + +FACT. (Basic Counting Principle) Suppose 2 experiments are to be performed. + +If one experiment can result in m possibilities + +Second experiment can result in $n$ possibilities + +Then together there are mn possibilities + +- I like to use the box method. For example. Each box represent the number of possibilities in that experiment. +- Example1: There are 20 teachers and 100 students in a school. How many ways can we pick a teacher and student of the year? + +\- Solution: Use the box Method: $20 \times 100 = 2000$ . + +- The counting principle can be generalized to any amount of experiments: $n_1 \cdots n_r$ possibilities +- Example2: + +- A college planning committee consists of 3 freshmen, 4 sophomores, 5 juniors, and 2 seniors. +- A subcommittee of 4 consists 1 person from each class. How many? +- Solution: Box method $3 \times 4 \times 5 \times 2 = 120$ . + +\- Example3: How many differen 6-place license plates are possible if the first 3 places are to be occupied by letters and the finals 3 by numbers? + +- Solution: $26 \cdot 26 \cdot 26 \cdot 10 \cdot 10 \cdot 10 = ?$ +- Question: What if no repetition is allowed? +- Solution: $26 \cdot 25 \cdot 24 \cdot 10 \cdot 9 \cdot 8$ + +\- Example4: How many functions defined on $n$ points are possible if each functional value is either 0 or 1. + +\- Solution: Box method on the $1, \ldots, n$ points gives us $2^n$ possible functions. + +## 1.2. PERMUTATIONS + +## 1.2. Permutations + +\- How many different ordered arrangements of the letters $a, b, c$ are possible? + +\- $abc, acb, bac, bca, cab$ Each arrangement is a permutation + +\- Can also use the box method to figure this out: $3 \cdot 2 \cdot 1 = 6$ . + +FACT. With $n$ objects. There are + +$$ +n (n - 1) \dots 3 \cdot 2 \cdot 1 = n! +$$ + +different permutations of the $n$ objects. + +(★) Note that ORDER matters when it comes to Permutations + +\- Example1: What is the numnber of possible batting order with 9 players? + +\- Answer: 9!(Box Method or permutations) + +\- Example2: How many ways can one arrange 4 math books, 3 chemistry books, 2 physics books, and 1 biology book on a bookshelf so that all the math books are together, all the chemistry books are together, and all the physics books are together. + +\- Answer: We can arrange the math books in 4! ways, the chemistry in 3! ways, the physics in 2! ways, and B in 1! = 1 way. + +\- But we also have to decide which set of books go on the left, which next, and so on. That is the same as the number of ways of arranging the letters M,C, P,B, and there are 4! ways of doing that. MCPB, PBPB ect.. + +\- So 4! (4!3!2!1!) ways. + +\- Example3: Repetitions: How many ways can one arrange the letters $a, a, b, c$ ? + +\- Let us label them $A, a, b, c$ . There are 4!, or 24, ways to arrange these letters. But we have repeats: we could have $Aa$ or $aA$ . So we have a repeat for each possibility (so divide!!!), and so the answer should be $4! / 2! = 12$ . + +\- If there were 3 $a's$ , 4 $b's$ , and 2 $c's$ , we would have + +$$ +\frac {9 !}{3 ! 4 ! 2 !} +$$ + +\- Example4: How many different letter arrangements can be formed from the word PEPPER? + +\- Answer: There 3 $P$ 's 2 $E$ 's and one $R$ . So $\frac{6!}{3!2!1!} = 30$ . + +FACT. There are + +$$ +\frac {n !}{n _ {1} ! \cdots n _ {r} !} +$$ + +different permutations of $n$ objects of which $n_1$ are alike, $n_2$ are alike, $n_r$ are alike. + +\- Example4: Suppose there are 4 Czech tennis players, 4 U.S. players, and 3 Russian players, in how many ways could they be arranged? + +\- Answer: $\frac{11!}{4!4!3!}$ . + +## 1.3. Combinations + +\- We are often interested in selecting $r$ objects from a total of $n$ objects. + +\- How many ways can we choose 3 letters out of 5? (Does order matter here? NO) If the letters are $a, b, c, d, e$ then there would be 5 for the first position, 4 for the second, and 3 for the third, for a total of $5 \times 4 \times 3$ . But order doesn't matter here. So we're over counting here.... + +\- But suppose the letters selected were a, b, c. If order doesn't matter, we will have the letters $a, b, c \ 3! = 6$ times, because there are $3!$ ways of arranging a group of 3. The same is true for any choice of three letters. So we should have + +$$ +\frac {5 \cdot 4 \cdot 3}{3 !} = \frac {5 !}{3 ! 2 !} = 1 0. +$$ + +Or what we did was $5 \cdot 4$ , or $n(n-1) \cdots (n-r+1)$ then divided by the repeats 3!. + +\- This is often written $\left( \begin{array}{l}5\\ 3 \end{array} \right)$ , read "5 choose 3". More generally.. + +FACT. If $r \leq n$ , then + +$$ +\binom{n}{r} = \frac {n !}{(n - r) ! r !} +$$ + +and say $n$ choose $r$ , represents the number of possible combinations of objects taken $r$ at a time. + +(★) Order DOES NOT Matter here + +\- Recall in Permutations order did matter. + +\- Example1: How many ways can one choose a committee of 3 out of 10 people? + +\- Answer: $\left( \begin{array}{l}10\\ 3 \end{array} \right) = \frac{10!}{3!7!} = \frac{10\cdot 9\cdot 8}{3\cdot 2} = 10\cdot 3\cdot 4 = 120.$ + +\- Example2: Suppose there are 9 men and 8 women. How many ways can we choose a committee that has 2 men and 3 women? + +\- $\underline{\text{Answer:}}$ We can choose 2 men in $\binom{9}{2}$ ways and 3 women in $\binom{8}{3}$ ways. The number of committees is then the product $\binom{9}{2} \cdot \binom{8}{3}$ . + +\- Example3:A person has 8 friends, of whom 5 will be invited to a party. (We've all been through this) + +\- (a) How many choices are there if 2 of the friends are feuding and will not attend together? + +\* Box it: [none] + [one of them] [others] + +\* $\binom{6}{5}+\binom{2}{1}\cdot\binom{6}{4}$ (recall that when we have OR, use +) + +\- (b) How many choices if 2 of the friends will only attend together? + +\* Box it: [none] + [with both] + +\* $\binom{6}{5}+1\cdot1\cdot\binom{6}{3}$ + +\- The value of $\left( \begin{array}{l}n\\ r \end{array} \right)$ are called binomials coefficients because of their prominence in the binomial theorem. + +THEOREM. (The Binomial Theorem) + +$$ +(x + y) ^ {n} = \sum_ {k = 0} ^ {n} \binom{n}{k} x ^ {k} y ^ {n - k}. +$$ + +PROOF. To see this, the left hand side is $(x+y)(x+y)\cdots(x+y)$ . This will be the sum of $2^{n}$ terms, and each term will have n factors. How many terms have k x's and n-k y's? This is the same as asking in a sequence of n positions, how many ways can one choose k of them in which to put x's? (Box it) The answer is $\binom{n}{k}$ , so the coefficient of $x^{k}y^{n-k}$ should be $\binom{n}{k}$ . ☐ + +• Example: Expand $(x + y)^{3}$ . + +$$ +- \underline {{\text { Solution: }}} (x + y) ^ {3} = y ^ {3} + 3 x y ^ {2} + 3 x ^ {2} y + x ^ {3}. +$$ + +\- Problem: Using Combinatorics: Let's prove + +$$ +\binom{1 0}{4} = \binom{9}{3} + \binom{9}{4} +$$ + +with no algebra: + +\- The LHS represents the number of committees having 4 people out of the 10. + +\- Let's say the President of the university will be in one of these committees and he's special, so we want to know when he'll be there or not. + +\- When he's there, then there are $1 \cdot \begin{pmatrix} 9 \\ 3 \end{pmatrix}$ is the number of ways that contain the President while $\begin{pmatrix} 9 \\ 4 \end{pmatrix}$ is the number of committees that do not contain the President and contain 4 out of the remaining people. + +• The more general equation is + +$$ +\binom{n}{r} = \binom{n - 1}{r - 1} + \binom{n - 1}{r} +$$ + +## 1.4. Multinomial Coefficients + +\- Example: Suppose one has 9 people and one wants to divide them into one committee of 3, one of 4, and a last of 2. How many different ways are there? + +\- Solution: (Box it) There are $\left( \begin{array}{l}9\\ 3 \end{array} \right)$ ways of choosing the first committee. Once that is done, there are 6 people left and there are $\left( \begin{array}{l}6\\ 4 \end{array} \right)$ ways of choosing the second committee. Once that is done, the remainder must go in the third committee. So there is 1 one to choose that. So the answer is + +$$ +\frac {9 !}{3 ! 6 !} \frac {6 !}{4 ! 2 !} = \frac {9 !}{3 ! 4 ! 2 !}. +$$ + +\- In general: Divide $n$ objects into one group of $n_1$ , one group of $n_2$ , ... and a $k$ th group of $n_k$ , where $n = n_1 + \cdots + n_k$ , the answer is there are + +$$ +\frac {n !}{n _ {1} ! n _ {2} ! \cdots n _ {k} !} \text { ways. } +$$ + +• These are known as multinomial coefficients. We write them as + +$$ +\binom{n}{n _ {1}, n _ {2}, \ldots , n _ {k}} = \frac {n !}{n _ {1} ! n _ {2} ! \cdots n _ {k} !}. +$$ + +\- Example: Suppose we are to assign Police officers their duties. Out of 10 officers: 6 patrols, 2 in station, 2 in schools. + +$$ +- \underline {{\text { Answer }}}: \frac {1 0 !}{6 ! 2 ! 2 !}. +$$ + +\- Example: There are 10 flags:5 indistinguishable Blue flags, 3 indistinguishable Red flags, and 2 indistinguishable Yellow flags. How may different ways can we order them on a flag pole? + +$$ +- \underline {{\text {Answer:}}} \frac {1 0 !}{5 ! 3 ! 2 !}. +$$ + +\- Example: Suppose one has 8 indistinguishable balls. How many ways can one put them in 3 boxes? + +\- Solution1: Let us make sequences of o's and |'s; any such sequence that has | at each side, 2 other |'s, and 8 o's represents a way of arranging balls into boxes. For example, if one has + +$$ +\left| \begin{array}{c c c} o o & 0 0 0 & 0 0 0 \end{array} \right|. +$$ + +\- How many different ways can we arrange this where we have start with $|$ and end with $|$ . In between, we are only arranging $8 + 2 = 10$ symbols, of which only 8 are $o$ 's + +\- So the question is: How many ways out of 10 spaces can one pick 8 of them into which to put an $o$ ? + +$$ +- \binom{1 0}{8}. +$$ + +\- Solution2: Look at spaces between. There are 9 spaces. So $\left( \begin{array}{l}9\\ 2 \end{array} \right) + 9$ . + +# CHAPTER 2 + +# Axioms of Probability + +## 2.1. Sample Space and Events + +\- We will have a sample space, denoted $S$ (sometimes $\Omega$ , or $\mathcal{U}$ ) that consists of all possible outcomes from an experiment. + +## - Example1: + +\* Experiment: Roll two dice, +\* Sample Space: $S =$ would be all possible pairs made up of the numbers one through six. List it here. $\{(i,j):i,j = 1,\dots 6\}$ . 36 points. + +## - Example 2: + +\* Experiment: Toss a coin twice +\* $S = \{HH,HT,TH,TT\}$ + +## - Example3: + +\* Experiment: Measuring the number of accidents of a random person before they had turn 18. + +$$ +\cdot S = \{0, 1, 2, \dots \} +$$ + +## - Others: + +\* Let $S$ be the possible orders in which 5 horses finish in a horse race; +\* Let $S$ be the possible price of some stock at closing time today; or $S = [0, \infty)$ ; +\* The age at which someone dies, $S = [0, \infty)$ . + +\- Events: An event $A$ is a subset of $S$ . In this case we use the notation $A \subset S$ , to mean $A$ is a subset of $S$ . + +- $A \cup B$ : points in $S$ such that is in $A$ OR $B$ OR BOTH. +- $A \cap B$ , points in $A$ AND $B$ . (you may also see $AB$ ) +- $A^c$ is the compliment of $A$ , the points NOT in $A$ . (you may also see $A'$ ) +- Can extend to $A_1, \ldots, A_n$ events. $\bigcup_{i=1}^{n} A_i$ and $\bigcap_{i=1}^{n} A_i$ . + +![](images/6b3f0ded4cd051c630042135580e78684c6922459b11c4e8aa2caecb8c45b6d5.jpg) + +![](images/4d7245c4b763b1a984167fda64161a505bddabac420c7708955314935e85d98a.jpg) + +![](images/2b501a36ebecf8e6510d38ab34de1a9cde321b71f7e70a0f69c6dbbc4e4e4ddf.jpg) + +![](images/0980c5188cef806986a7a7a0baa01ffe43d9bce1849cadc0e8426f19083351b8.jpg) + +![](images/9b8442f2c186cef8f527574b1fb1244613aa87219c600d472348076690325230.jpg) + +![](images/40109a50e86248f3043f070378bf98a7dc21d41ad5243decfde80c27bba87595.jpg) + +![](images/c24f97de20802b2c7ccfffd988145c2a2c047e18011082bc8f78149a6ac1335f.jpg) + +![](images/42f649bc847ee0170d19273dc0242abcdd147d9b5644d17c5cf30a219a7a71a7.jpg) + +![](images/42fe4758aade9c313ae1a02151578611ce8f7d2345aec6a8469c30d7820f8050.jpg) + +• Example1: Roll two dice. + +- Example of an Events +- $E =$ the two dies come up even and equal $\{(2,2),(4,4),(6,6)\}$ +- $F =$ the sum of the two dice is 8. $\{(2,6),(3,5),(4,4),(5,3),(6,2)\}$ . +- $E \cup F = \{(2, 2), (2, 6), (3, 5), (4, 4), (5, 3), (6, 2), (6, 6)\}$ +- $E \cap F = \{(4, 4)\}$ . +- $F^c$ all the 31 other ways that does not include $\{(2,6), (3,5), (4,4), (5,3), (6,2)\}$ . + +\- Example2: $S = [0, \infty)$ age someone dies. + +\- Event $A =$ person dies before they reached 30. + +$$ +* A = [ 0, 3 0). +$$ + +\- Interpret $A^c = [30,\infty)$ + +\* The person dies after they turned 30. + +\- $B = (15, 45)$ . Do $A \cup B, A \cap B$ and so on. + +- Properties: Events also have commutative and associate and Distributive laws. +• What is $A \cup A^{c}$ ? = S. +- DeMorgan's Law: + +- $(A\cup B)^{c} = A^{c}\cap B^{c}$ . Try to draw a picture +$-\left(A\cap B\right)^{c}=A^{c}\cup B^{c}.$ +- This works for general $A_{1},\ldots ,A_{n}$ : $(\cup_{i = 1}^{n}A_{i})^{c} = \cap_{i = 1}^{n}A_{i}^{c}$ and $(\cap_{i = 1}^{n}A_{i})^{c} = \cup_{i = 1}^{n}A_{i}^{c}$ . + +\- The empty set $\emptyset = \{\}$ is the set that has nothing in it. + +\- $A$ and $B$ are disjoint if $A \cap B = \emptyset$ . + +- In Probability we may say that events $A$ and $B$ are "mututally exclusive" if they are disjoint. +- mutually exclusive means the same thing as disjoint + +## 2.2. Axioms of Probability + +\- Let $E$ be an event. How do we defined the probability of an event? + +- We can attempt to define a probability by the relative frequency, +- Perform an experiment (e.g. Flipping a coin) +- Perform that experiment $n$ times and let $n(E) =$ the number of times the event occurred in $n$ repetitions + +\* (e.g. Flip a coin $n = 1000$ times, and let's say that $n(\{Tails\}) = 551$ ) Then it's reasonable to think $\mathbb{P}(\{Tails\}) \approx \frac{551}{1000}$ + +- So maybe we can define the probability of an event as $\mathbb{P}(E) = \lim_{n\to \infty}\frac{n(E)}{n}$ . But we don't know if this limit exists, or if $n(E)$ is even well defined!!! +- So we need a new approach. + +\- Probability will be a rule given by the following Axioms (Laws that we all agree on) + +- A probability will be a function $\mathbb{P}(E)$ where the input is a set/event such that +- Axiom 1: $0 \leq \mathbb{P}(E) \leq 1$ for all events $E$ . +- Axiom 2: $\mathbb{P}(S) = 1$ . +- $\underline{\mathbf{Axiom 3}}$ : (disjoint property) If the events $E_1, E_2, \ldots$ are pairwise disjoint/mutually exclusive then + +$$ +\mathbb {P} \left(\bigcup_ {i = 1} ^ {\infty} E _ {i}\right) = \sum_ {i = 1} ^ {\infty} \mathbb {P} \left(E _ {i}\right). +$$ + +\* Mutually exclusive means that $E_{i} \cap E_{j} = \emptyset$ when $i \neq j$ . + +\- Remark: Note that you take a probability of a subset of $S$ , not of points of $S$ . However it is common to write $P(x)$ for $P(\{x\})$ . + +\- Say if the experiment is tossing a xoin. Then $S = \{H, T\}$ . The probability of heads should be written as $\mathbb{P}(\{H\})$ , but it is common to see $\mathbb{P}(H)$ . + +\- Example1: + +$-$ (a) Suppose we toss a coin and they are equally likely then $S = \{H,T\}$ and + +\* $\mathbb{P}\left(\{H\}\right) = \mathbb{P}\left(\{T\}\right) = \frac{1}{2}$ . We may write $\mathbb{P}(H) = \mathbb{P}(T) = \frac{1}{2}$ . + +\- (b) If biased coin is tosse then one could have a different assignment of probability $\mathbb{P}(H) = \frac{2}{3}, \mathbb{P}(T) = \frac{1}{3}$ . + +\- Example2: + +- Rolling a fair die, the probability space consists of $S = 1,2,3,4,5,6$ , each point having probability $\frac{1}{6}$ . +- We can compute the probability of rolling an even number by + +$$ +\mathbb {P} (\{\text { even } \}) = \mathbb {P} (\{2, 4, 6 \}) +$$ + +$$ += \mathbb {P} (2) + \mathbb {P} (4) + \mathbb {P} (6) = \frac {1}{2} +$$ + +where we used the rules of probability by breaking it down into a sum. + +PROPOSITION 1. (a) $\mathbb{P}(\emptyset) = 0$ + +(b) If $A_{1},\ldots ,A_{n}$ are pairwise disjoint, $\mathbb{P}\left(\cup_{i = 1}^{n}A_{i}\right) = \sum_{i = 1}^{n}\mathbb{P}\left(A_{i}\right)$ . +(c) $\mathbb{P}(E^c) = 1 - \mathbb{P}(E)$ . +(d) If $E \subset F$ , then $\mathbb{P}(E) \leq \mathbb{P}(F)$ . +(e) $\mathbb{P}(E\cup F) = \mathbb{P}(E) + \mathbb{P}(F) - \mathbb{P}(E\cap F)$ . + +\- It helps to draw diagrams to prove these. + +\- Try to prove at least some of these yourself. + +PROOF. (a) Let $A_{i} = \emptyset$ for each $i$ which are disjoint. So + +$$ +\mathbb {P} (\emptyset) = \mathbb {P} \left(\bigcup_ {i = 1} ^ {\infty} A _ {i}\right) = \sum_ {i = 1} ^ {\infty} \mathbb {P} (A _ {i}) = \sum_ {i = 1} ^ {\infty} \mathbb {P} (\emptyset), +$$ + +since this would be infinite sum so that $\mathbb{P}(\emptyset) = 0$ since $0 \leq \mathbb{P}(\emptyset) \leq 1$ . + +(b) Let $A_{n+1} = A_{n+2} = \cdots = \emptyset$ so that $\cup_{i=1}^{\infty} A_{i} = \cup_{i=1}^{n} A_{i}$ hence + +$$ +\begin{array}{l} \mathbb {P} \left(\cup_ {i = 1} ^ {n} A _ {i}\right) = \mathbb {P} \left(\cup_ {i = 1} ^ {\infty} A _ {i}\right) \\ = \sum_ {i = 1} ^ {n} \mathbb {P} \left(A _ {i}\right) + \sum_ {n = 1} ^ {\infty} \mathbb {P} (\emptyset) \\ = = \sum_ {i = 1} ^ {n} \mathbb {P} \left(A _ {i}\right) + \sum_ {n = 1} ^ {\infty} 0 \\ = \sum_ {i = 1} ^ {n} \mathbb {P} \left(A _ {i}\right) \\ \end{array} +$$ + +(c) Use $S = E \cup E^{c}$ . By Axiom (2) we have + +$$ +1 = \mathbb {P} (S) = \mathbb {P} (E) + \mathbb {P} (E ^ {c}), +$$ + +hence $\mathbb{P}(E^c) = 1 - \mathbb{P}(E)$ . + +(d) If $E \subset F$ , then write $F = E \cup (F \cap E^c)$ thus since this is disjoint + +$$ +\mathbb {P} (F) = \mathbb {P} (E \cup (F \cap E ^ {c})) = \mathbb {P} (E) + \mathbb {P} (F \cap E ^ {c}) \geq \mathbb {P} (E) + 0 = \mathbb {P} (E). +$$ + +(e) Write $E \cup F = E \cup (E^c \cap F)$ , (Picture of venn diagram of both ) hence by disjointness again + +$$ +\mathbb {P} (E \cup F) = \mathbb {P} (E) + \mathbb {P} (E ^ {c} \cap F). +$$ + +Now write $F$ (with picture) as $F = (E \cap F) \cup (E^c \cap F)$ and using disjointness + +$$ +\mathbb {P} (F) = \mathbb {P} (E \cap F) + \mathbb {P} (E ^ {c} \cap F) \implies \mathbb {P} (E ^ {c} \cap F) = \mathbb {P} (F) - \mathbb {P} (E \cap F), +$$ + +substitute into first equation to get + +$$ +\begin{array}{l} \mathbb {P} (E \cup F) = \mathbb {P} (E) + \mathbb {P} (E ^ {c} \cap F) \\ = \mathbb {P} (E) + \mathbb {P} (F) - \mathbb {P} (E \cap F), \\ \end{array} +$$ + +as needed. + +• Example: Uconn Basketball is playing Kentucky this year. + +- Home game has .5 chance of winning +- Away game has .4 chance of winning. +- .3 that uconn wins both games. +- What's the probability that Uconn loses both games? +- Answer. + +\* Let $\mathbb{P}(A_1) = .5$ , $\mathbb{P}(A_2) = .4$ and $\mathbb{P}(A_1 \cap A_2) = .3$ . +\* We want to find $\mathbb{P}(A_1^c\cap A_2^c)$ . Simplify as much as we can: + +$$ +\begin{array}{l} \mathbb {P} \left(A _ {1} ^ {c} \cap A _ {2} ^ {c}\right) = \mathbb {P} \left(\left(A _ {1} \cup A _ {2}\right) ^ {c}\right) \text { by DeMorgan's Law } \\ = 1 - \mathbb {P} \left(A _ {1} \cup A _ {2}\right), \text { by Proposition 1c } \\ \end{array} +$$ + +\* Using Proposition 1e, we have + +$$ +\mathbb {P} \left(A _ {1} \cup A _ {2}\right) = . 5 +. 4 -. 3 = . 6, +$$ + +Hence $\mathbb{P}(A_1^c\cap A_2^c) = 1 - .6 = .4$ as needed. + +## 2.3. Equally Likely Outcomes + +\- In many experiments, a probability space consists of finitely many points, all with equally likely probabilities. + +- Basic example was a tossing a coin $P(H) = P(T) = \frac{1}{2}$ +- Fair die: $P(i) = \frac{1}{6}$ for $i = 1, \ldots, 6$ . + +\- In this case from Axiom 3 we have that + +$$ +P (E) = \frac {\text { number of outcomes in } E}{\text { number of outcomes in } S}. +$$ + +\- Example1: What is the probability that if we roll 2 dice, the sum is 7? + +\- Answer: There are 36 total outcomes, of which 6 have a sum of 7: + +\* $E =$ "sum is 7" $= \{(1,6),(2,5),(3,4),(4,3),(5,2),(6,1)\}$ . Since they are all equally likely, the probability is $\mathbb{P}(E) = \frac{6}{6\cdot6} = \frac{1}{6}$ . + +\- Example 2: If 3 balls are “randomly drawn” from a bowl containing 6 white and 5 black balls, what is the probability that one ball is white and the other two are black? + +\- Method 1: (regard as a ordered selection) + +$$ +\begin{array}{l} P (E) = \frac {W B B + B W B + B B W}{1 1 \cdot 1 0 \cdot 9} \\ = \frac {6 \cdot 5 \cdot 4 + 5 \cdot 6 \cdot 4 + 5 \cdot 4 \cdot 6}{9 9 0} = \frac {1 2 0 + 1 2 0 + 1 2 0}{9 9 0} = \frac {4}{1 1}. \\ \end{array} +$$ + +\- Method2: (Regard as unordered set of drawn balls) + +$$ +P \left(E\right) = \frac {\left(1 \text {white}\right) \left(2 \text {black}\right)}{\binom{1 1}{3}} = \frac {\binom{6}{1} \binom{5}{2}}{\binom{1 1}{3}} = \frac {4}{1 1}. +$$ + +\- We can always choose which way to regard our experiments. + +\- Example 3 A committee of 5 is to selected from a group of 6 men and 9 women. What is probability consists of 3 men and 2 women + +\- Answer: Easy $\frac{men\cdot women}{all}=\frac{\left(\begin{array}{c}6\\3\end{array}\right)\left(\begin{array}{c}9\\2\end{array}\right)}{\left(\begin{array}{c}15\\5\end{array}\right)}=\frac{240}{1001}.$ + +\- Example 4: Seven balls are randomly withdrawn from an urn that contains 12 red, 16 blue, and 18 green. + +- (b) Find probability that “at least 2 red balls are withdrawn;” +- $\underline{\mathbf{Ans}}$ : Let $E$ be this event then $P(E) = 1 - P(E^c)$ , $P$ (at least 2 red) = 1- $\mathbb{P}$ (drawing 0 or 1 balls). + +$$ +\mathbb {P} \left(\text { drawing 0 or 1 red balls }\right) = \frac {\binom{1 6 + 1 8 = 3 4}{7}}{\binom{4 6}{7}} + \frac {\binom{1 2}{1} \binom{3 4}{6}}{\binom{4 6}{7}}. +$$ + +\- Explanation of Poker/Playing cards: Ranks and suits,etc! + +\- There are 52 cards in a standard deck of playing cards. The poker hand is consists of five cards. There are 4 suits: heats, spades, diamonds, and clubs ( $\heartsuit$ ♠♦♣). The suits diamonds and hearts are red while clubs and spades are black. In each suit there are 13 ranks: the numbers 2, 3, ..., 10, the face cards, Jack, Queen, King, and the Ace (not a face card). + +\- Example 5: What is the probability that in a poker hand (5 cards out of 52) we get exactly 4 of a kind? + +\- Answer: Consider 4 aces and 1 king: $AAAK = \left( \begin{array}{l}4\\ 4 \end{array} \right)\left( \begin{array}{l}4\\ 1 \end{array} \right)$ . But $JJJJ3$ is the same probability. + +\* Thus there are 13 ways to pick the first rank, and 12 ways to pick the second rank + +$$ +\begin{array}{l} P (4 \text { of a kind }) = \frac {[ \text { choice of ranks } ] [ \text { given rank how to choose a hand } ]}{\binom{5 2}{5}} \\ = \frac {\left(1 3 \cdot 1 2 \cdot \binom{4}{4} \binom{4}{1}\right)}{\binom{5 2}{5}} \approx . 0 0 0 0 1 3 9 \\ \end{array} +$$ + +\- Example 6: What is the probability that in a poker hand (5 cards out of 52) we get a straight. (no straight flushes, can't be of the same suit) + +\- Answer: Consider: A-2-3-4-5-6-7-8-9-10-J-Q-K-A- There are 10 possible straights. + +\* Given a straight Say A2345: There are $4 \cdot 4 \cdot 4 \cdot 4 \cdot 4 - (\text{of the same suit}) = 4^5 - 4$ . + +$$ +\begin{array}{l} P (\text { Straight }) = \frac {[ \text { choice of straight } ] [ \text { given striaght how to choose a hand } ]}{\binom{5 2}{5}} \\ = \frac {1 0 \cdot (4 ^ {5} - 4)}{\binom{5 2}{5}} \approx . 0 0 3 9 \\ \end{array} +$$ + +\- Example 7: What is the probability that in a poker hand (5 cards out of 52) we get a Full House. (3 and a 2 of a kind) + +- Answer: It would be [3 of a kind][2 of a kind]. AAAKK or KKAAA are different! Choose suit: $13 \cdot 12$ . +- Then once we choose within each group there + +$$ +P (\text {Full House}) = \frac {[ \text {choice of rank} ] [ 3 \text {of a kind} ] [ 2 \text {of a kind} ]}{\binom{5 2}{5}} +$$ + +$$ += \frac {1 3 \cdot 1 2 \binom{4}{3} \binom{4}{2}}{\binom{5 2}{5}} \approx . 0 0 1 4. +$$ + +\- Example 8: (Birthday Problem) In a class of 32 people, what is the probability that at least two people have the same birthdays? (We assume each day is equally likely.) + +\- Answer: Let the first person have a birthday on some day. The probability that the second person has a different birthday will be $\frac{364}{365}$ . The probability that the third person has a different birthday from the first two people is $\frac{363}{365}$ . So the answer is + +$$ +\mathbb {P} \left(\text {at least 2 people}\right) = 1 - \mathbb {P} \left(\text {Everyone different birthday}\right) +$$ + +$$ += 1 - \frac {3 6 5}{3 6 5} \cdot \frac {3 6 4}{3 6 5} \cdot \frac {3 6 3}{3 6 5} \dots \frac {(3 6 5 - 3 1)}{3 6 5} +$$ + +$$ += 1 - 1 \cdot \frac {3 6 4}{3 6 5} \cdot \frac {3 6 3}{3 6 5} \dots \frac {3 3 4}{3 6 5} \approx 0. 7 5 2 3 7 4. +$$ + +\- Really High!!! + +## CHAPTER 3 + +## Independence + +## 3.1. Independent Events + +DEFINITION. We say $E$ and $F$ are independent events if + +$$ +\mathbb {P} (E \cap F) = \mathbb {P} (E) \mathbb {P} (F). +$$ + +• Example1: Suppose you flip two coins. + +- The event that you get heads on the second coin is independent of the event that you get tails on the first. +- This is why: Let $A_{t}$ be the event of getting is tails for the first coin and $B_{h}$ is the event of getting heads for the second coin, and we assume we have fair coins (although this is not necessary), then + +$$ +\mathbb {P} \left(A _ {t} \cap B _ {h}\right) = \frac {1}{4}, \text { list out all outcomes } +$$ + +$$ +\mathbb {P} \left(A _ {t}\right) \mathbb {P} \left(B _ {h}\right) = \frac {1}{2} \frac {1}{2} = \frac {1}{4}. +$$ + +\- Example2: Experiment: Draw a card from an ordinary deck of cards + +\- Let $A =$ draw ace, $S =$ draw a spade. + +\* These are independent events since you're taking one at a time, so one doesn't effect the other. To see this using the definition we have compute +\* $\mathbb{P}(A)\mathbb{P}(S) = \frac{1}{13}\frac{1}{4}.$ +\* White $\mathbb{P}(A\cap S) = \frac{1}{52}$ since there is only 1 Ace of spades. + +PROPOSITION 2. If $E$ and $F$ are independent, then $E$ and $F^c$ are independent. + +PROOF. Draw a Venn Diagram to help with the computation, but note that + +$$ +\begin{array}{l} \mathbb {P} (E \cap F ^ {c}) = \mathbb {P} (E) - \mathbb {P} (E \cap F) \\ = \mathbb {P} (E) - \mathbb {P} (E) \mathbb {P} (F) \\ = \mathbb {P} (E) (1 - \mathbb {P} (F)) \\ = \mathbb {P} (E) \mathbb {P} \left(F ^ {c}\right). \\ \end{array} +$$ + +![](images/2dce10c023707461abdcde66f4a4f1023cba8c47f7e93817c621289c12bd2eeb.jpg) + +\- Remark: Independence and mutually exclusive, are two different things! + +DEFINITION. We say $E, F, G$ are independent if $E, F$ are independent, $E, G$ are independent, $F, G$ are independent, and $\mathbb{P}(E \cap F \cap G) = \mathbb{P}(E)\mathbb{P}(F)\mathbb{P}(G)$ . + +• Example: Experiment is you roll two dice: + +\- Define the following events: + +- $S_{7} = \{\text{sum is 7}\}$ +- $A_{4} = \{\text{first die is a 4}\}$ +- $B_{3} = \{$ second die is a 3\} +- Are the events $S_{7}, A_{4}, B_{3}$ independent? + +\* Compute + +$$ +\mathbb {P} \left(S _ {7} \cap A _ {4} \cap B _ {3}\right) = \mathbb {P} \left(\{(4, 3) \}\right) = \frac {1}{3 6} +$$ + +but + +$$ +\mathbb {P} \left(S _ {7}\right) \mathbb {P} \left(A _ {4}\right) \mathbb {P} \left(B _ {3}\right) = \frac {6}{3 6} \frac {1}{6} \frac {1}{6} = \frac {1}{3 6 \cdot 6}. +$$ + +\- Remark: This generalizes to events $A_1, \ldots, A_n$ . We say events $A_1, \ldots, A_n$ are independent if for all subcollections $i_1, \ldots, i_r \in \{1, \ldots, n\}$ we have that $\mathbb{P}\left(\bigcap_{j=1}^{r} A_{i_j}\right) = \prod_{j=1}^{r} \mathbb{P}\left(A_{i_j}\right)$ . + +• Example: + +- An urn contains 10 balls: 4 red and 6 blue. +- A second urn contains 16 red balls and an unknown number of blue balls. +- A single ball is drawn from each urn. The probability that both balls are the same color is 0.44. +- Question: Calculate the number of blue balls in the second urn. +- Solution: Let $R_{i} =$ even that a red ball is drawn from urn $i$ and let $B_{i} =$ event that a blue ball is drawn from urn $i$ . + +\* Let $x$ be the number of blue balls in urn 2, +\* Note that drawing from urn 1 and independent from drawing from urn 2. They are completely different urns! They shouldn't effect the other. +\* Then + +$$ +\begin{array}{l} . 4 4 = \mathbb {P} \left(\left(R _ {1} \cap R _ {2}\right) \bigcup \left(B _ {1} \cap B _ {2}\right)\right) = \mathbb {P} \left(R _ {1} \cap R _ {2}\right) + \mathbb {P} \left(B _ {1} \cap B _ {2}\right) \\ = \mathbb {P} (R _ {1}) \mathbb {P} (R _ {2}) + \mathbb {P} (B _ {1}) \mathbb {P} (B _ {2}), \text { by independence } \\ = \frac {4}{1 0} \frac {1 6}{x + 1 6} + \frac {6}{1 0} \frac {x}{x + 1 6}. \\ \end{array} +$$ + +\* Solve for $x!$ You will get $x = 4$ . + +\- Example (Gambler's Ruin) (Used in Finance or Actuarial Science) + +\- Experiment: Suppose you toss a fair coin repeatedly and independently. If it comes up heads, you win a dollar, and if it comes up tails, you lose a dollar. Suppose you start with \$50. What's the probability you will get to \$200 before you go broke? + +\- Answer: It's actually easier if we generalize the problem. + +\* Let $p(x)$ be the probability you get 200 before 0 if you start with $x$ dollars. +\* We know $p(0) = 0$ and $p(200) = 1$ . So by the law of total probability + +$$ +\begin{array}{l} p (x) = \mathbb {P} (\text { Win 200 before 0 }) \\ = \mathbb {P} (H) \mathbb {P} (\text { Win } 2 0 0 \text { before } 0 \mid H) + \mathbb {P} \left(H ^ {c}\right) \mathbb {P} (\text { Win } 2 0 0 \text { before } 0 \mid H ^ {c}) \\ = \frac {1}{2} p (x + 1) + \frac {1}{2} p (x - 1). \\ \end{array} +$$ + +\* Rearrange the function to get + +$$ +\begin{array}{l} 2 p (x) = p (x - 1) + p (x + 1) \quad \Longleftrightarrow \quad p (x) + p (x) = p (x - 1) + p (x + 1) \\ \Longleftrightarrow \quad p (x) - p (x - 1) = p (x + 1) - p (x) \\ \Longleftrightarrow \quad \frac {p (x) - p (x - 1)}{x - (x - 1)} = \frac {p (x + 1) - p (x)}{(x + 1) - x}. \\ \end{array} +$$ + +\* This tells you that the slows are constant. What does that tell you about $p(x)$ ? It's a line! + +\- Thus we must have $p(x) = \frac{x}{200}$ . + +\* Thus $p(50) = \frac{1}{4}$ . + +• Example (A variation of Gambler's ruin) + +\- Problem: Suppose we are in the same situation, but you are allowed to go arbitrarily far in debt. Let $p(x)$ be the probability you ever get to \$200. What is a formula for $p(x)$ ? + +\* Answer: Just as before $p(x) = \frac{1}{2} p(x + 1) + \frac{1}{2} p(x - 1)$ . So that $p(x)$ is linear. + +\* But now all we have is that $p(200) = 1$ and linear and domain is $(-\infty, 200)$ . + +\* Draw a graph: Now the slope, or $p'(x)$ can't be negative, or else we would have it that $p(x) > 1$ for $x \in (-\infty, 200)$ . + +\- The slope can't be positive or else we would get $p(x) < 0$ for $x \in (-\infty, 200)$ . + +\* Thus we must have that $p(x) \equiv \text{constant}$ . Hence $p(x) = 1$ for all $x \in (-\infty.200)$ . + +\* Sol: So we are certain to get \$200 if we cna get into debt. + +\- Method2: + +\* Just compute There is nothing special about the figure 200. Another way of seeing this is to compute as above the probability of getting to 200 before $-M$ and then letting $M \to \infty$ . + +\- We would get $p(x)$ is a line with $p(-M) = 0$ and $p(200) = 1$ so that + +$$ +p (x) - 0 = \frac {1 - 0}{2 0 0 - (- M)} (x - (- M)) +$$ + +and letting $M \to \infty$ wee see that $p(x) = \frac{x + M}{200 + M} \to 1$ . + +• Example: Experiment: Roll 10 dice. + +- What is the probability that exactly 4 twos will show if you roll 10 dice? +- Answer: These are independent. The probability that the 1st, 2nd, 3rd, and 10th dice will show a three and the other 6 will not is $\left(\frac{1}{6}\right)^3\left(\frac{5}{6}\right)^7$ . +- Independence is used here: the probability is $\frac{1}{6}\frac{1}{6}\frac{1}{6}\frac{5}{6}\frac{5}{6}\frac{5}{6}\frac{5}{6}\frac{5}{6}\frac{5}{6}\frac{1}{6}$ . Note that the probability that the 10th, 9th, 8th, and 7th dice will show a two and the other 6 will not have the same probability. +- So to answer our original question, we take $\left(\frac{1}{6}\right)^4\left(\frac{5}{6}\right)^6$ and multiply it by the number of ways of choosing 4 dice out of 10 to be the ones showing the twos. There are $\binom{10}{3}$ ways to do this $\binom{10}{4}\left(\frac{1}{6}\right)^4\left(\frac{5}{6}\right)^6$ . + +\- This is an example of Bernoulli trials, or the Binomial distribution. + +\- If we have $n$ independent trials, where the probability of success if $p$ . The probability that there are $k$ successes in $n$ trials is + +$$ +\binom{n}{k} p ^ {k} (1 - p) ^ {n - k}. +$$ + +## CHAPTER 4 + +## Conditional Probability and Independence + +## 4.1. Conditional Probabilities + +\- Suppose there are + +- 200 men, of which 100 are smokers, +- 100 women, of which 20 are smokers. +- Question1: What is the probability that a person chosen at random will be a smoker? $\frac{120}{300}$ +- Question2: Now, let us ask, what is the probability that a person chosen at random is a smoker given that the person is a women? $\frac{20}{100}$ right? + +\* Note this is + +$$ +\frac {\# (\text { women smokers })}{\# (\text { women })} = \frac {P (\text { women and a smoker })}{P (\text { woman })}. +$$ + +\- Thus we make the following definition: + +DEFINITION. If $\mathbb{P}(F) > 0$ , we define + +$$ +\mathbb {P} (E \mid F) = \frac {\mathbb {P} (E \cap F)}{\mathbb {P} (F)}. +$$ + +Now $\mathbb{P}(E\mid F)$ is read "the probability of $E$ given $F$ ." + +- Note that $\mathbb{P}(E\cap F) = \mathbb{P}(E|F)\mathbb{P}(F)!$ +- This is the conditional probability that $E$ occurs given that $F$ has already occurred! +- Remark: Suppose $\mathbb{P}(E \mid F) = \mathbb{P}(E)$ , i.e. knowing $F$ doesn't help predict $E$ . Then this implies that $E$ and $F$ are independent of each other. Rearranging $\mathbb{P}(E \mid F) = \frac{\mathbb{P}(E \cap F)}{\mathbb{P}(F)} = \mathbb{P}(E)$ we see that $\mathbb{P}(E \cap F) = \mathbb{P}(E)\mathbb{P}(F)$ . +- Example1: Experiment: Roll two dice. +- (a) What is the probability the sum is 8? +\* Solution: Note that $A = \{(2,6), (3,5), (4,4), (5,3), (6,2)\}$ so we know $\mathbb{P}(A) = \frac{5}{36}$ . +- (b) What is the probability that the sum is 8 given that the first die shows a 3? (In other words, find $\mathbb{P}(A \mid B)$ ) +\* Solution: Let $B = \{\text{first die shows three}\}$ . +\* $\mathbb{P}(A\cap B) = \mathbb{P}\left(\{(3,5)\}\right) = \frac{1}{36}$ is probability that the first die shows a 3 and the sum is 8 +\* Finally we can compute + +$$ +\mathbb {P} (A \mid B) = \mathbb {P} (\text { sum is } 8 \mid 1 \text { st is a } 3) = \frac {1 / 3 6}{1 / 6} = \frac {1}{6}. +$$ + +\- Remark: When computing $\mathbb{P}(E \mid F)$ , Sometime its easier to work with the reduced sample space $F \subset S$ . + +\- Note in the previous example when we computed + +$$ +\mathbb {P} (\text { sum is } 8 \mid 1 \text { st is a } 3) +$$ + +we could have worked in the smaller sample space of $\{1st\ is\ a\ 3\}=\{(3,1),(3,2),(3,3),(3,4),(3,5),(3,6)\}$ . Since only $(3,5)$ begins with a 3 and has the sum of 8, then the probability is + +$$ +\frac {\text { total number of outcomes in the event }}{\text { total number of outcomes in new sample space }} = \frac {1}{6}. +$$ + +\- Example2: Experiment: Suppose a box has 3 red marbles and 2 black ones. We select 2 marbles. + - Question: What is the probability that second marble is red given that the first one is red? + +\* Answer: + +$\overline{\cdot R_1} = \{\text{First marble is is red}\}$ , + +$R_{2} = \{\text{Second marble is red}\}$ , then + +$$ +\begin{array}{l} \mathbb {P} \left(R _ {2} \mid R _ {1}\right) = \frac {\mathbb {P} \left(R _ {1} \cap R _ {2}\right)}{\mathbb {P} \left(R _ {1}\right)} \\ = \frac {(2 \text { red }) (0 \text { black }) / \binom{5}{2}}{3 / 5} \\ = \frac {\binom{3}{2} \binom{2}{0} / \binom{5}{2}}{3 / 5} \\ = \frac {3 / 1 0}{3 / 5} = \frac {1}{2}. \\ \end{array} +$$ + +\* Solution 2: + +\- We could have done the same example more easily if we look at the new sample space $S' = \{R, R, B, B\}$ thus $\mathbb{P}(R_2 \mid R_1) = \mathbb{P}'(\{\text{drawing red}\}) = \frac{2}{4} = \frac{1}{2}$ . + +\- Example3: Landon is 80% sure he forgot his textbook at the Union or Monteith buildings. 40% sure that it is at the union, and 40% sure that it is at Monteith. Given that Landon already went to Monteith and noticed his textbook not there, what is the probability that it's at the Union? + +\- Solution: + +$$ +\begin{array}{l} \mathbb {P} \left(U n i o n \mid N o t \text { Monteith }\right) = \frac {\mathbb {P} \left(U \cap M ^ {c}\right)}{\mathbb {P} \left(M ^ {c}\right)} \\ = \frac {\mathbb {P} (U)}{1 - \mathbb {P} (M)}, \text { since } U \subset M ^ {c} \\ = \frac {4 / 1 0}{6 / 1 0} = \frac {2}{3}. \\ \end{array} +$$ + +\- Example4: Suppose that Annabelle and Bobby each draw 13 cards from a standard deck of 52. Given that Sarah has exactly two aces, what is the probability that Bobby has exactly one ace? + +\- Solution: Let $A$ be the event "Annabelle has two aces," and let $B$ be the event "Bobby has exactly one ace." Again, we want $\mathbb{P}(B \mid A)$ , so we calculate $\mathbb{P}(A)$ and $\mathbb{P}(A \cap B)$ . Annabelle could have any of $\left( \begin{array}{c}52\\ 13 \end{array} \right)$ possible hands. Of these hands, $\left( \begin{array}{c}4\\ 2 \end{array} \right) \cdot \left( \begin{array}{c}48\\ 11 \end{array} \right)$ will have exactly + +two aces, so + +$$ +\mathbb {P} \left(A\right) = \frac {\binom{4}{2} \cdot \binom{4 8}{1 1}}{\binom{5 2}{1 3}}. +$$ + +Now the number of ways in which Annabelle can have a certain hand and Bobby can have a certain hand is $\binom{52}{13}\cdot\binom{39}{13}$ , and the number of ways in which A and B can both occur is $\binom{4}{2}\cdot\binom{48}{11}\cdot\binom{2}{1}\cdot\binom{37}{12}$ . so + +$$ +\mathbb {P} (A \cap B) = \frac {\binom{4}{2} \cdot \binom{4 8}{1 1} \cdot \binom{2}{1} \cdot \binom{3 7}{1 2}}{\binom{5 2}{1 3} \cdot \binom{3 9}{1 3}}. +$$ + +Therefore, + +$$ +\begin{array}{l} \mathbb {P} (B \mid A) = \frac {\mathbb {P} (A \cap B)}{\mathbb {P} (A)} = \frac {\frac {\left( \begin{array}{c} 4 \\ 2 \end{array} \right) \cdot \left( \begin{array}{c} 4 8 \\ 1 1 \end{array} \right) \cdot \left( \begin{array}{c} 2 \\ 1 \end{array} \right) \cdot \left( \begin{array}{c} 3 7 \\ 1 2 \end{array} \right)}{\left( \begin{array}{c} 5 2 \\ 1 3 \end{array} \right) \cdot \left( \begin{array}{c} 3 9 \\ 1 3 \end{array} \right)}}{\frac {\left( \begin{array}{c} 4 \\ 2 \end{array} \right) \cdot \left( \begin{array}{c} 4 8 \\ 1 1 \end{array} \right)}{\left( \begin{array}{c} 5 2 \\ 1 3 \end{array} \right)}} \\ = \frac {\binom{2}{1} \cdot \binom{3 7}{1 2}}{\binom{3 9}{1 3}}. \\ \end{array} +$$ + +\- Note that since $\mathbb{P}(B \mid A) = \frac{\mathbb{P}(A \cap B)}{\mathbb{P}(A)}$ then $\mathbb{P}(A \cap B) = \mathbb{P}(A)\mathbb{P}(B \mid A)$ . + +\- In general: If $E_1, \ldots, E_n$ are events then + +$$ +\mathbb {P} \left(E _ {1} \cap E _ {2} \cap \dots \cap E _ {n}\right) = \mathbb {P} \left(E _ {1}\right) \mathbb {P} \left(E _ {2} \mid E _ {1}\right) \mathbb {P} \left(E _ {3} \mid E _ {1} \cap E _ {2}\right) \dots \mathbb {P} \left(E _ {n} \mid E _ {1} \cap E _ {2} \cap \dots \cap E _ {n - 1}\right). +$$ + +\- Example5: + +\- Experiment: Suppose an urn has 5 White balls and 7 Black balls. Each ball that is selected is returned to the urn along with an additional ball of the same color. Suppose draw 3 balls. + +\- Part (a): What is the probability that you get 3 white balls. + +\* Then + +$$ +\mathbb {P} (3 \text {white balls}) = \mathbb {P} (1 \text {st W}) \mathbb {P} (2 \text {nd W} | 1 \text {st W}) \mathbb {P} (3 \text {nd W} | 1 \text {st} \& 2 \text {ndW}) +$$ + +$$ += \frac {5}{1 2} \frac {6}{1 3} \frac {7}{1 4} +$$ + +\- Part (b): What is the probability of getting 1 white ball. + +$$ +\begin{array}{l} \mathbb {P} (1 \text { white ball }) = \mathbb {P} (W B B) + \mathbb {P} (B W B) + \mathbb {P} (B B W) \\ = 3 \frac {5 \cdot 7 \cdot 8}{1 2 \cdot 1 3 \cdot 1 4}. \\ \end{array} +$$ + +\- Note that + +$$ +\mathbb {P} (E \cap F) = \mathbb {P} (E \mid F) \mathbb {P} (F) +$$ + +\- Example 6: Phan wants to take a Biology course or a Chemistry course. Given that the students take Biology, the probability that they get an $A$ is is $\frac{4}{5}$ . While the probability of getting an $A$ given that the student took Chemistry is $\frac{1}{7}$ . If Phan makes a decision on the course to take randomly, what's probability of "getting an $A$ in Chem"? + +\- Solution: Let $B = \{\text{Takes Biology}\}$ and $C = \{\text{Takes Chemistry}\}$ and $A = \{"\text{gets an A"}\}$ , then + +$$ +\begin{array}{l} \mathbb {P} (A \cap C) = \mathbb {P} (C) \mathbb {P} (A \mid C) \\ = \frac {1}{2} \cdot \frac {1}{7} = \frac {1}{1 4}. \\ \end{array} +$$ + +• Example 7: A total of 500 married couples are poled about salaries: + +
    WifeHusband makes less than 25,000Husband makes more than 25,000
    Less than $25,000212198
    More than $25,0003654
    + +\- Part (a): Find the probability that a Husband earns less than 25,000? + +$$ +* \underline {{\text {Answer:}}} \frac {2 1 2 + 3 6}{5 0 0} +$$ + +\- Part (b): Find $\mathbb{P}$ (wife makes $>25,000$ | Husband makes $>25,000$ ) + +$$ +* \underline {{\text {Answer:}}} \frac {5 4 / 5 0 0}{(1 9 8 + 5 4) / 5 0 0} = \frac {5 4}{2 5 2} = . 2 1 4 +$$ + +\- Part (c): Find $\mathbb{P}$ (wife makes $>25,000$ | Husband makes $< 25,000$ ) + +$$ +* \underline {{\text { Answer: }}} \frac {3 6 / 5 0 0}{(2 4 8) / 5 0 0} = . 1 4 5. +$$ + +## 4.2. Bayes's Formula + +- Sometimes it's easier to compute a probability once we know something has or has not happened. +- Note that we can compute, + +$$ +\mathbb {P} (E) = \mathbb {P} (E \cap F) + \mathbb {P} (E \cap F ^ {c}) +$$ + +$$ += \mathbb {P} (E \mid F) \mathbb {P} (F) + \mathbb {P} (E \mid F ^ {c}) \mathbb {P} (F ^ {c}) +$$ + +$$ += \mathbb {P} (E \mid F) \mathbb {P} (F) + \mathbb {P} (E \mid F ^ {c}) (1 - \mathbb {P} (F)). +$$ + +\- This formula is called: The Law of Total Probability: + +$$ +\mathbb {P} (E) = \mathbb {P} (E \mid F) \mathbb {P} (F) + \mathbb {P} (E \mid F ^ {c}) (1 - \mathbb {P} (F)) +$$ + +\- The following problem will describe the types of problems of this section. + +• Example1: Insurance company believes + +- The probability that “an accident prone person” has an accident within a year is .4. +- The probability that “Non-accident prone person” has an accident with year is .2. +- $30\%$ of the population is "accident prone". +- Part (a): Find $\mathbb{P}(A_1)$ where $A_{1} =$ new policy holder will have an accident within a year? + +\* Let $A = \{\text{Policy holder IS accident prone.}\}$ + +$$ +\mathbb {P} \left(A _ {1}\right) = \mathbb {P} \left(A _ {1} \mid A\right) \mathbb {P} (A) + \mathbb {P} \left(A _ {1} \mid A ^ {c}\right) \left(1 - \mathbb {P} (A)\right) +$$ + +$$ += . 4 (. 3) +. 2 (1 -. 3) +$$ + +$$ += . 2 6 +$$ + +\- Part (b): Suppose new policyholder has accident with one year. What's probability that he or she is accident prone? + +$$ +\begin{array}{l} \mathbb {P} (A \mid A _ {1}) = \frac {\mathbb {P} (A \cap A _ {1})}{\mathbb {P} (A _ {1})} \\ = \frac {\mathbb {P} (A) \mathbb {P} \left(A _ {1} \mid A\right)}{. 2 6} \\ = \frac {(. 3) (. 4)}{. 2 6} = \frac {6}{1 3}. \\ \end{array} +$$ + +\- In general: + +\- So in Part (a) we had to break a probability into two cases: If $F_1, \ldots, F_n$ are mutually exclusive events such that they make up everywhere $S = \bigcup_{i=1}^{n} F_i$ then + +$$ +\mathbb {P} (E) = \sum_ {i = 1} ^ {n} \mathbb {P} (E \mid F _ {i}) \mathbb {P} (F _ {i}). +$$ + +\* This is called Law of Total Probability. + +\- In Part (b), we wanted to find a probability of a separate conditional event: then + +$$ +\mathbb {P} \left(F _ {j} \mid E\right) = \frac {\mathbb {P} \left(E \mid F _ {j}\right) \mathbb {P} \left(F _ {j}\right)}{\sum_ {i = 1} ^ {n} \mathbb {P} \left(E \mid F _ {i}\right) \mathbb {P} \left(F _ {i}\right)}. +$$ + +\* This is known as Baye's Formula +\* Note that the denominator of the Bayes's formula is the Law of total probability. + +• Example2: Suppose the test for HIV is + +- $98\%$ accurate in both directions +- $0.5\%$ of the population is HIV positive. + +- Question: If someone tests positive, what is the probability they actually are HIV positive? +- Solution: Let $T_{+} = \{\text{tests positive}\}$ , $T_{-} = \{\text{tests negative}\}$ , while $+ = \{\text{actually HIV positive},\}$ $- = \{\text{actually negative}\}$ . + +\* Want + +$$ +\begin{array}{l} \mathbb {P} (+ \mid T _ {+}) = \frac {\mathbb {P} (+ \cap T _ {+})}{\mathbb {P} (T _ {+})} \\ = \frac {\mathbb {P} (T _ {+} \mid +) \mathbb {P} (+)}{\mathbb {P} (T _ {+} \mid +) \mathbb {P} (+) + \mathbb {P} (T _ {+} \mid -) \mathbb {P} (-)} \\ = \frac {(. 9 8) (. 0 0 5)}{(. 9 8) (. 0 0 5) + . 0 2 (. 9 9 5)} \\ = 19.8 \% . \\ \end{array} +$$ + +• Example3: Suppose + +- $30\%$ of the women in a class received an A on the test +- $25\%$ of the men/or else received an A. +- $60\%$ of the class are women. +- Question: Given that a person chosen at random received an A, what is the probability this person is a women? + +\* Solution: Let A the event that a students receives an A. Let W = being a women, M = not a women. Want + +$$ +\mathbb {P} (W \mid A) = \frac {\mathbb {P} (A \mid W) \mathbb {P} (W)}{\mathbb {P} (A \mid W) \mathbb {P} (W) + \mathbb {P} (A \mid M) \mathbb {P} (M)}, \text { by Bayes's } +$$ + +$$ += \frac {. 3 (. 6)}{. 3 (. 6) + . 2 5 (. 4)} = \frac {. 1 8}{. 2 8} \approx . 6 4. +$$ + +- (General Baye's Theorem) Here's one with more than 3 possibilities: +- Example4: Suppose in Factory with Machines I, II, III producing Iphones + +- Machines I,II,III produce 2%,1%, and 3% defective iphones, respectively. +- Out of total production, Machines $I$ makes $35\%$ of all Iphones, $II - 25\%$ , $III - 40\%$ . +- If one Iphone is selected at random from the factory, +- Part (a): what is probability that one Iphone selected is defective? + +$$ +\mathbb {P} (D) = P (I) \mathbb {P} (D \mid I) + P (I I) \mathbb {P} (D \mid I I) + P (I I I) \mathbb {P} (D \mid I I I) +$$ + +$$ += (. 3 5) (. 0 2) + (. 2 5) (. 0 1) + (. 4) (. 0 3) +$$ + +$$ += \frac {2 1 5}{1 0 , 0 0 0}. +$$ + +\- Part (b): What is the conditional prob that if an Iphone is defective, that it was produced by machine III? + +$$ +\begin{array}{l} \mathbb {P} (I I I \mid D) = \frac {\mathbb {P} (I I I) \mathbb {P} (D \mid I I I)}{\mathbb {P} (D)} \\ = \frac {(. 4) (. 0 3)}{2 1 5 / 1 0 , 0 0 0} = \frac {1 2 0}{2 1 5}. \\ \end{array} +$$ + +\- Example5: In a Multiple Choice Test, students either knows the answer or randomly guesses the answer to a question. + +\- Let $m =$ number of choices in a question. + +- Let $p =$ the probability that the students knows the answer to a question. +- Question: What is the probability that the student actually knew the answer, given that the student answers correctly. +- Solution: +- Let $K = \{\text{Knows the answer}\}$ and $C = \{\text{Answer's correctly}\}$ . Then + +$$ +\mathbb {P} (K \mid C) = \frac {\mathbb {P} (C \mid K) \mathbb {P} (K)}{\mathbb {P} (C \mid K) \mathbb {P} (K) + \mathbb {P} (C \mid K ^ {c}) \mathbb {P} (K ^ {c})} +$$ + +$$ += \frac {1 \cdot p}{1 \cdot p + \frac {1}{m} (1 - p)} = \frac {m p}{1 + (m - 1) p}. +$$ + +## CHAPTER 5 + +## Random Variables + +## 5.1. Random Variables + +\- When we perform an experiment, we are interested in some function of the outcomes, instead of the actual outcome. + +\- We want to attach for each outcome, a numerical value. + +\- Definition: A random variable is a function $X: S \to \mathbb{R}$ or write $X: \Omega \to \mathbb{R}$ . (Use capital letters to denote r.v) + +\- We can think of $X$ as a numerical value that is random, like as if $X$ is a random number. + +• Example: Toss a coin + +\- Let $X$ be 1 if heads and $X = 0$ if tails + +\- Then $X(H) = 1$ and $X(T) = 0$ . + +\- We can do calculus on real numbers but not on $\Omega = S = \{H, T\}$ . + +• Example: Roll a die + +\- Let $X$ denote the outcome, so $X = 1,2,3,4,5,6$ (its random) + +\- That is $X(1) = 1, X(2) = 2, \ldots$ . + +• Example: Roll a die, define + +$$ +Y = \left\{ \begin{array}{l l} 1 & \text { outomce } = \text { odd } \\ 0 & \text { outomce } = \text { even } \end{array} \right. +$$ + +\- Can be thought of as + +$$ +Y (s) = \left\{ \begin{array}{l l} 1 & s = \text {odd} \\ 0 & s = \text {even} \end{array} \right.. +$$ + +\- A common question we'll have is "What values can $X$ attain?" + +\- In other words, what is the range of $X$ ? Since $X: S \to ?$ + +• Example: Toss a coin 10 times + +\- Let $X$ be the number of heads showing + +\- What random values can $X$ be? 0, 1, 2, ..., 10. + +\- Example: In general in $n$ trials, $X$ is the number of successes + +\- Example1: Let $X$ be the amount of liability(damages) a driver incurs in a year. + +\- $X:S\to [0,\infty)$ . + +• Example2: Toss a coin 3 times + +\- Let $X$ be the number of heads that appear, so $X = 0,1,2,3$ . + +\- In other words, $X:S\to \{0,1,2,3\}$ + +\- We may assign probabilities to the different values of the random variable: + +$$ +\begin{array}{l} \mathbb {P} (X = 0) = \mathbb {P} ((T, T, T)) = \frac {1}{2 ^ {3}} = \frac {1}{8} \\ \mathbb {P} (X = 1) = \mathbb {P} ((T, T, H), (T, H, T), (H, T, T)) = \frac {3}{8} \\ \mathbb {P} (X = 2) = \mathbb {P} ((T, H, H), (H, H, T), (H, T, H)) = \frac {3}{8} \\ \mathbb {P} (X = 3) = \mathbb {P} ((H, H, H)) = \frac {1}{8}. \\ \end{array} +$$ + +\- Note that since $X$ must take the values of 0 through 3 then + +$$ +1 = \mathbb {P} \left(\bigcup_ {i = 0} ^ {3} \{X = i \}\right) = \sum_ {i = 0} ^ {3} \mathbb {P} (X = i), +$$ + +which makes sense from our previous calculation. + +## 5.2. Discrete Random Variables + +DEFINITION. A random variable that can take on at most countable number of possible values is said to be a discrete r.v. + +DEFINITION. For a discrete random variable, we can define the probability mass function (pmf), or the density function of $X$ by $p(x) = \mathbb{P}(X = x)$ . Note that $p: \mathbb{R} \to [0,1]$ . + +- Note that $(X = x) = (\omega \in \Omega \mid X(\omega) = x)$ is an abbreviation. +- Let $X$ assume only the values $x_{1}, x_{2}, x_{3} \ldots$ + +\- In other words, $X: S \to \{x_1, x_2, \ldots\}$ + +\- Properties of a pmf $p(x)$ : + +\* Note that we must have $0 < p(x_i) \leq 1$ for, $i = 1,2,\ldots$ and $p(x) = 0$ for all other values of $x$ can't attain. +\* Also must have + +$$ +\sum_ {i = 1} ^ {\infty} p (x _ {i}) = 1. +$$ + +- We often draw bar graphs for discrete r.v. +• Example: If we toss a coin + +\- $X = 1$ if we have $H$ and $X = 0$ if we have $T$ . + +\- Then draw a BAR graph + +$$ +p _ {X} (x) = \left\{ \begin{array}{l l} \frac {1}{2} & x = 0 \\ \frac {1}{2} & x = 1, \\ 0 & \text {otherwise} \end{array} \right. +$$ + +- Oftentimes someone has already found the $pmf$ for you, and you can use to compute probabilities. +- Example: The $pmf$ of $X$ is given by $p(i) = e^{-\lambda \frac{\lambda^i}{i!}}$ for $i = 0,1,2,\ldots$ where $\lambda$ is a parameter (what is this?) that is any positive number + +\- Part (a) What values can the random variable $X$ attain? In other words, what is the range of $X$ ? + +\* Sol: By definition we have $\mathbb{P}(X = 0) = p(0) = e^{-\lambda}\frac{\lambda^0}{0!} = e^{-\lambda}$ + +\- Part (b) Find $\mathbb{P}(X = 0)$ + +\* Sol: By definition we have $\mathbb{P}(X = 0) = p(0) = e^{-\lambda}\frac{\lambda^0}{0!} = e^{-\lambda}$ + +\- Part (c) Find $\mathbb{P}(X > 2)$ + +\* Sol: Note that + +$$ +\mathbb {P} (X > 2) = 1 - \mathbb {P} (X \leq 2) +$$ + +$$ += 1 - \mathbb {P} (X = 0) - \mathbb {P} (X = 1) - \mathbb {P} (X = 2) +$$ + +$$ += 1 - p (0) - p (1) - p (2) +$$ + +$$ += 1 - e ^ {- \lambda} - \lambda e ^ {- \lambda} - \frac {\lambda^ {2} e ^ {- \lambda}}{2}. +$$ + +## 5.3. Expected Value + +\- One of the most important concepts in probability is that of expectation. If $X$ is a random variable that what is the average value of $X$ , that is what is the expected value of $X$ . + +DEFINITION. Let X have a pmf $p(x)$ . We define the expectation, or expected value of X to be + +$$ +\mathbb {E} [ X ] = \sum_ {x: p (x) > 0} x p (x). +$$ + +- Notation $\mathbb{E}X$ , or $EX$ . +- Example1: Let $X(H) = 0$ and $X(T) = 1$ . What is $\mathbb{E}X$ ? + +$$ +\mathbb {E} X = 0 \cdot p (0) + 1 \cdot p (1) +$$ + +$$ += 0 \frac {1}{2} + 1 \cdot \frac {1}{2} = \frac {1}{2}. +$$ + +\- Example2: Let $X$ be the outcome when we roll a fair die. What is $\mathbb{E}X$ ? + +$$ +\begin{array}{l} \mathbb {E} X = 1 \left(\frac {1}{6}\right) + 2 \left(\frac {1}{6}\right) + \dots + 6 \frac {1}{6} \\ = \frac {1}{6} (1 + 2 + 3 + 4 + 5 + 6) = \frac {2 1}{6} = \frac {7}{2} = 3. 5 \\ \end{array} +$$ + +\- Note that $X$ can never be 3.5, so expectation is to give you an idea, what an exact. + +\- Recall infinite series: If $0 \leq x < 1$ then a geometric series is + +$$ +\begin{array}{l} \sum_ {n = 0} ^ {\infty} x ^ {n} = 1 + x + x ^ {2} + x ^ {3} + \dots \\ = \frac {1}{1 - x}. \\ \end{array} +$$ + +\- One thing you can do with series is differentiate them and integrate them: So if + +$$ +1 + x + x ^ {2} + x ^ {3} + \dots + = \frac {1}{1 - x} +$$ + +then + +$$ +0 + 1 + 2 x + 3 x ^ {2} + \dots + = \frac {1}{(1 - x) ^ {2}} +$$ + +\- Example3: Let $X$ be the number or tornadoes in Connecticut per year. Meaning that the random variable $X$ can be any number $X = 0,1,2,3,\ldots$ . Suppose the state of Connecticut did some analysis and found out that + +$$ +\mathbb {P} (X = i) = \frac {1}{2 ^ {i + 1}}. +$$ + +- Question: What is $\mathbb{E}X$ ? That is, what is the expected number of tornadoes per year in Connecticut. +- Solution: Note that $X$ is infinite, but still countable, hence still discrete. + +\- Note that + +$$ +p (i) = \left\{ \begin{array}{l l} \frac {1}{2} & i = 0, \\ \frac {1}{4} & i = 1, \\ \frac {1}{8} & i = 2, \\ \vdots & \vdots \\ \frac {1}{2 ^ {n + 1}} & i = n. \end{array} \right. +$$ + +\- We have that + +$$ +\begin{array}{l} \mathbb {E} X = 0 \cdot p (0) + 1 \cdot p (1) + 2 \cdot p (2) + \dots \\ = 0 \cdot \frac {1}{2} + 1 \frac {1}{2 ^ {2}} + 2 \frac {1}{2 ^ {3}} + 3 \frac {1}{2 ^ {4}} + \dots \\ = \frac {1}{2 ^ {2}} \left(1 + 2 \frac {1}{2} + 3 \frac {1}{2 ^ {2}} + \dots\right) \\ = \frac {1}{4} (1 + 2 x + 3 x ^ {2} + \dots), \text {with} x = \frac {1}{2} \\ = \frac {1}{4} \frac {1}{(1 - x) ^ {2}} = \frac {1}{4 (1 - \frac {1}{2}) ^ {2}} = 1. \\ \end{array} +$$ + +## 5.4. The C.D.F. + +DEFINITION. Define $F: R \to [0,1]$ to be the function + +$$ +F (x) = \mathbb {P} (X \leq x), \quad \text { for any } - \infty < x < \infty +$$ + +to be the cumulative distribution function, or the distribution function of X., or CDF of X, or c.d.f + +\- Note that when $X$ is discrete, + +$$ +F (x _ {0}) = \mathbb {P} (X \leq x _ {0}) = \sum_ {x \leq x _ {0}} p (x). +$$ + +- We sometimes use the notation $F_{X}(x)$ to highlight that $F_{X}$ is the CDF of the random variable $X$ . +- Example: Suppose $X$ is equals to the number of heads in 3 coin flips. From Section 5.1, we calculated the p.m.f to be.: + +$$ +p (0) = \mathbb {P} (X = 0) = \frac {1}{8} +$$ + +$$ +p (1) = \mathbb {P} (X = 1) = \frac {3}{8} +$$ + +$$ +p (2) = \mathbb {P} (X = 2) = \frac {3}{8} +$$ + +$$ +p (3) = \mathbb {P} (X = 3) = \frac {1}{8}. +$$ + +Question: Find the c.d.f of $X$ . Plot the graph of the c.d.f. + +\- Solution: Summing up the probabilities up to that value of $x$ we get the following: + +$$ +F (x) = \left\{ \begin{array}{l l} 0 & - \infty < x < 0 \\ \frac {1}{8} & 0 \leq x < 1 \\ \frac {4}{8} & 1 \leq x < 2 \\ \frac {7}{8} & 2 \leq x < 3 \\ 1 & 3 \leq x < \infty \end{array} \right.. +$$ + +\- The graph is given by + +![](images/82976a735b008c82fde988fc1ac55be22d046f0fa0a3790f3a35aa022acc2a7e.jpg) + +
    +line chart + +| x | y | +| ---- | --- | +| -0.5 | 0.0 | +| 0.0 | 0.2 | +| 1.0 | 0.5 | +| 2.0 | 0.9 | +| 3.0 | 1.0 | +| 3.5 | 1.0 | +
    + +- Note that this is a step function. +- This function has jumps, and not continuous everywhere. +- But it looks like it never decreases. + +## • Properties of the CDF: + +- 1. $F$ is nondecreasing, that is + +$$ +* \text { if } x < y \text { then } F (x) \leq F (y) +$$ + +\- 2. $\lim_{x\to \infty}F(x) = 1$ + +\- 3. $\lim_{x\to -\infty}F(x) = 0$ + +\- 4. $F$ is right continuous. That are two ways that you can think of right continuity: + +\* $\lim_{x\to 1^{+}}F(x_n) = F(x)$ , meaning "the limit from the right equals where the function is defined" +\* If $x_{n} \downarrow x$ is a decreasing sequence then $\lim_{n \to \infty} F(x_{n}) = F(x)$ . + +\- We take these properties as facts, though one would normally have to prove these. + +\- The following proposition does not have to be proved in class, and can be highlighted with the following example. But we include it here for completeness. + +PROPOSITION 3. Let $F_{X}(x)$ be the CDF for some random variable $X$ . Then the following holds: + +(a) For any $a \in \mathbb{R}$ , we have $\mathbb{P}(X < a) = \lim_{x \to a^{-}} F_X(x)$ +(b) For any $a \in \mathbb{R}$ , we have $\mathbb{P}(X = a) = F_X(a) - \lim_{x \to a^-} F_X(x)$ + +PROOF. For part (a). + +We first write + +$$ +\begin{array}{l} (X < a) = \bigcup_ {n = 1} ^ {\infty} \left(X \leq a - \frac {1}{n}\right) \\ = (X \leq a - 1) \bigcup \left[ \bigcup_ {n = 1} ^ {\infty} \left(a - \frac {1}{n} < X \leq a - \frac {1}{n + 1}\right) \right] \\ \end{array} +$$ + +and since the events $E_{n} = \left(a - \frac{1}{n}\leq X\leq a - \frac{1}{n + 1}\right)$ are disjoint then we can use Axiom 3 so prove that + +$$ +\begin{array}{l} \mathbb {P} (X < a) = \mathbb {P} (X \leq a - 1) + \sum_ {n = 1} ^ {\infty} \mathbb {P} \left(a - \frac {1}{n} < X \leq a - \frac {1}{n + 1}\right) \\ = \mathbb {P} (X \leq a - 1) + \lim _ {k \rightarrow \infty} \sum_ {n = 1} ^ {k} \left[ \mathbb {P} \left(X \leq a - \frac {1}{n + 1}\right) - \mathbb {P} \left(X \leq a - \frac {1}{n}\right)\right] \\ = \mathbb {P} (X \leq a - 1) + \lim _ {k \rightarrow \infty} \left[ \mathbb {P} \left(X \leq a - \frac {1}{k + 1}\right) - \mathbb {P} (X \leq a - 1) \right], \text { by telescoping } \\ = \lim _ {k \rightarrow \infty} \mathbb {P} \left(X \leq a - \frac {1}{k + 1}\right) + \mathbb {P} (X \leq a - 1) - \mathbb {P} (X \leq a - 1) \\ = \lim _ {n \to \infty} F _ {X} \left(a - \frac {1}{n}\right). \\ \end{array} +$$ + +Now you can replace the sequence $a_{n} = a - \frac{1}{n}$ with any sequence $a_{n}$ that is increasing towards $a$ , and we get the similar result, + +$$ +\lim _ {n \to \infty} F _ {X} (a _ {n}) = \mathbb {P} (X < a), +$$ + +since this holds for all increasing sequences $a_{n}$ towards $a$ , then we've shown that + +$$ +\lim _ {x \to a ^ {-}} F _ {X} (x) = \mathbb {P} \left(X < a\right). +$$ + +For part (b). We use part (a) and get + +$$ +\begin{array}{l} \mathbb {P} (X = a) = \mathbb {P} (X \leq a) - \mathbb {P} (X < a) \\ = F _ {X} (a) - \lim _ {x \to a ^ {-}} F _ {X} (x). \\ \end{array} +$$ + +• Example: Let X have distribution + +$$ +F (x) = \left\{ \begin{array}{l l} 0 & x < 0 \\ \frac {x}{2} & 0 \leq x < 1 \\ \frac {2}{3} & 1 \leq x < 2 \\ \frac {1 1}{1 2} & 2 \leq x < 3 \\ 1 & 3 \leq x. \end{array} \right. +$$ + +Graph this and answer the following: + +\- Part (a): Compute $\mathbb{P}(2 < X\leq 4)$ . We have that + +$$ +\begin{array}{l} \mathbb {P} (2 < X \leq 4) = \mathbb {P} (X \leq 4) - \mathbb {P} (X \leq 2) \\ = F (4) - F (2) \\ = \frac {1}{1 2}. \\ \end{array} +$$ + +\- Part (b): Compute $\mathbb{P}(X < 3)$ . +\* We have that + +$$ +\begin{array}{l} \mathbb {P} (X < 3) = \lim _ {n \to \infty} \mathbb {P} \left(X \leq 3 - \frac {1}{n}\right) \\ = \lim _ {x \to 3 ^ {-}} F _ {X} (x) \\ = \frac {1 1}{1 2} \\ \end{array} +$$ + +\- Part (c): Compute $\mathbb{P}(X = 1)$ . +\* We have that + +$$ +\begin{array}{l} \mathbb {P} (X = 1) = \mathbb {P} (X \leq 1) - \mathbb {P} (X < 1) \\ = F _ {X} (1) - \lim _ {x \rightarrow 1 ^ {-}} F _ {X} (x) \\ = \frac {2}{3} - \lim _ {x \rightarrow 1} \frac {x}{2} \\ = \frac {2}{3} - \frac {1}{2} = \frac {1}{6}. \\ \end{array} +$$ + +## 5.5. Expected Value of Sums of Random Variables + +\- Recall our current definition of $\mathbb{E}X$ + +\- List out $X = x_{1}, x_{2}, \ldots$ and let $p(x_{i})$ be the density of $X$ + +$$ +- \text { Then } \mathbb {E} X = \sum_ {i = 1} ^ {\infty} x _ {i} p (x _ {i}). +$$ + +\- We need a new definition that will help the linearity of expectation. + +\- Goal: If $Z = X + Y$ then $\mathbb{E}[X + Y] = \mathbb{E}X + \mathbb{E}Y$ . + +\- Definition 2: Let $S$ (or $\Omega$ ) be the sample space then define + +$$ +\mathbb {E} X = \sum_ {\omega \in S} X (\omega) \mathbb {P} \left(\{\omega \}\right). +$$ + +\- Example: Let $S = \{1,2,3,4,5,6\}$ and $X(1) = X(2) = 1$ and $X(3) = X(4) = 3$ and $\overline{X(5)} = X(6) = 5$ + +\* Def1: We know $X = 1,3,5$ with $p(1) = p(3) = p(5) = \frac{1}{3}$ +\* Then $E X = 1 \cdot \frac{1}{3} + 3 \frac{1}{3} + 5 \frac{1}{3} = \frac{9}{3} = 3.$ +\* Def2: We list all of $S = \{1, 2, 3, 4, 5, 6\}$ and +\* Then + +$$ +\mathbb {E} X = X (1) \mathbb {P} (\{1 \}) + \dots + X (6) \cdot \mathbb {P} (\{6 \}) +$$ + +$$ += 1 \frac {1}{6} + 1 \frac {1}{6} + 3 \frac {1}{6} + 3 \frac {1}{6} + 5 \frac {1}{6} + 5 \frac {1}{6} = 3. +$$ + +\- Difference + +- Def1: We list all the values that $X$ can attain and only care about those. (Range) +- Def2: List all possible outcomes. (Domain) + +PROPOSITION 4. If $X$ is a discrete random variable and $S$ is countable, then the two definitions are equivalent + +\- NOTE: No need to prove in lecture. But here for completeness. + +PROOF. We start with the first definition. Let $X = x_{1}, x_{2}, \ldots$ + +$$ +\begin{array}{l} \mathbb {E} X = \sum_ {x _ {i}} x _ {i} p (x _ {i}) \\ = \sum_ {x _ {i}} x _ {i} \mathbb {P} (X = x _ {i}) \\ = \sum_ {x _ {i}} x _ {i} \sum_ {\omega \in \{\omega : X (\omega) = x _ {i} \}} \mathbb {P} (\omega) \\ = \sum_ {x _ {i}} \sum_ {\omega \in \{\omega : X (\omega) = x _ {i} \}} x _ {i} \mathbb {P} (\omega) \\ = \sum_ {x _ {i}} \sum_ {\omega \in \{\omega : X (\omega) = x _ {i} \}} X (\omega) \mathbb {P} (\omega) \\ = \sum_ {\omega \in S} X (\omega) \mathbb {P} (\omega), \\ \end{array} +$$ + +where I used that each $S_{i}=\{\omega:X(\omega)=x_{i}\}$ are mutually exclusive events that union up to S. + +![](images/6f437c0ac8008a25805533cace1b47a72e8d8d5117b198b302ef248424b007c1.jpg) + +\- Using this definition, we can prove linearity of the expectation. + +THEOREM 5. (Linearity) If $X$ and $Y$ are discrete random variables and $a \in \mathbb{R}$ then + +(a) $\mathbb{E}[X + Y] = \mathbb{E}X + \mathbb{E}Y.$ +(b) $\mathbb{E}[aX] = a\mathbb{E}X$ . + +PROOF. We have that + +$$ +\begin{array}{l} \mathbb {E} [ X + Y ] = \sum_ {\omega \in S} (X (\omega) + Y (\omega)) \mathbb {P} (\omega) \\ = \sum_ {\omega \in S} (X (\omega) \mathbb {P} (\omega) + Y (\omega) \mathbb {P} (\omega)) \\ = \sum_ {\omega \in S} X (\omega) \mathbb {P} (\omega) + \sum_ {\omega \in S} Y (\omega) \mathbb {P} (\omega) \\ = \mathbb {E} X + \mathbb {E} Y. \\ \end{array} +$$ + +If $a\in \mathbb{R}$ then + +$$ +\begin{array}{l} \mathbb {E} [ a X ] = \sum_ {\omega \in S} (a X (\omega)) \mathbb {P} (\omega) \\ = a \sum_ {\omega \in S} X (\omega) \mathbb {P} (\omega) \\ = a \mathbb {E} X. \\ \end{array} +$$ + +□ + +\- Generality: Linearity is true for general random variable $X_{1}, X_{2}, \ldots, X_{n}$ . + +## 5.6. Expectation of a Function of a Random Variable + +\- Let $X$ be a random variable. + +\- Can we find the expected value of things like $X^2, e^X, \sin X$ etc? + +\- Example1: Let $X$ denote a random variable such that + +$$ +\mathbb {P} (X = - 1) = . 2, +$$ + +$$ +\mathbb {P} (X = 0) = . 5 +$$ + +$$ +\mathbb {P} (X = 1) = . 3 +$$ + +Let $Y = X^{2}$ . Find $\mathbb{E}Y$ . + +- Solution: Note that $Y = \left\{0^2, (-1)^2, (1)^2\right\} = \{0, 1\}$ . +- Note that $p_Y(1) = .2 + .3 = .5$ and $p_Y(0) = .5$ . +- Thus $\mathbb{E}Y = 0\cdot .5 + 1\cdot .5 = .5.$ + +\- IMPORTANT: + +- Note that $\mathbb{E}X^2 = .5$ +- While $(\mathbb{E}X)^2 = .01$ . Not equal! + +\* Since $\mathbb{E}X = .3 - .2 = .1$ . Thus + +$$ +\mathbb {E} X ^ {2} \neq (\mathbb {E} X) ^ {2}. +$$ + +\- In general, there is a formula for $g(X)$ where $g$ is function. That use the fact that $g(X)$ will be $g(x)$ for some $x$ such that $X = x$ . + +THEOREM 6. If $X$ is a discrete random variable that takes values $X \in \{x_1, x_2, x_3, \ldots\}$ with respective probability mass function $p(x_i)$ , then for any real valued function $g: \mathbb{R} \to \mathbb{R}$ we have that + +$$ +\mathbb {E} \left[ g (X) \right] = \sum_ {i = 1} ^ {\infty} g \left(x _ {i}\right) p \left(x _ {i}\right). +$$ + +\- NOTE: No need to prove in lecture. But here for completeness. + +PROOF. The random variable $Y = g(X)$ can take on values, say $Y = y_{1}, y_{2}, \ldots$ . But we know that + +$$ +y _ {j} = g (x _ {i}) +$$ + +and as we see there could be more than one value $x_{i}$ such that $y_{j} = g(x_{i})$ . Thus we will group this sum into this fashion: Using the definition of expectation we have that + +$$ +\begin{array}{l} \mathbb {E} [ Y ] = \sum_ {j} y _ {j} \mathbb {P} (Y = y _ {j}) \\ = \sum_ {j} y _ {j} \mathbb {P} (g (X) = y _ {j}) \\ = (\star). \\ \end{array} +$$ + +Now + +$$ +\mathbb {P} (g (X) = y _ {j}) = \mathbb {P} \left(\bigcup_ {i: g (x _ {i}) = y _ {j}} (g (x _ {i}) = y _ {j})\right) +$$ + +$$ += \sum_ {i: g (x _ {i}) = y _ {j}} p (x _ {i}). +$$ + +Thus plugging this back into $(\star)$ we have that + +$$ +\begin{array}{l} \mathbb {E} [ Y ] = \sum_ {j} y _ {j} \sum_ {i: g (x _ {i}) = y _ {j}} p (x _ {i}) \\ = \sum_ {j} \sum_ {i: g (x _ {i}) = y _ {j}} y _ {j} p (x _ {i}) \\ = \sum_ {j} \sum_ {i: g (x _ {i}) = y _ {j}} g (x _ {i}) p (x _ {i}) \\ = \sum_ {i = 1} ^ {\infty} g (x _ {i}) p (x _ {i}), \\ \end{array} +$$ + +as needed. + +![](images/e90273795db24c64a57790f3b937347d00a2b759d188e78546fae7464a083998.jpg) + +- Remark: $\mathbb{E}X^{2} = \sum x_{i}^{2}p(x_{i})$ . +- Example1(Revisted): Let $X$ denote a random variable such that + +$$ +\mathbb {P} (X = - 1) = . 2, +$$ + +$$ +\mathbb {P} (X = 0) = . 5 +$$ + +$$ +\mathbb {P} (X = 1) = . 3 +$$ + +Let $Y = X^{2}$ . Find $\mathbb{E}Y$ . + +$$ +- \underline {{\text { Sol: }}} \text { We have that } \mathbb {E} X ^ {2} = \sum x _ {i} ^ {2} p (x _ {i}) = (- 1) ^ {2} (. 2) + 0 ^ {2} (. 5) + 1 ^ {2} (. 3) = . 5. +$$ + +DEFINITION. We call $\mu = \mathbb{E}X$ to be the mean, or the first moment of $X$ . + +The quantity $EX^{n}$ for $n \geq 1$ , is called the nth moment of X. + +\- From out theorem we know that the $n$ th moments can be calculated a + +$$ +\mathbb {E} X ^ {n} = \sum_ {x: p (x) > 0} x ^ {n} p (x). +$$ + +## 5.7. Variance + +- The variance of a r.v. is a measure of how spread out the values of $X$ are. +- The expectation of a r.v. is quantity that help us differentiate different r.v.'s, but it doesn't tell us how spread out values are. + +\- For example, take + +$$ +X = 0 \text { with probability } 1 +$$ + +$$ +Y = \left\{ \begin{array}{l l} - 1 & p = \frac {1}{2} \\ 1 & p = \frac {1}{2} \end{array} \right. +$$ + +$$ +Z = \left\{ \begin{array}{l l} - 1 0 0 & p = \frac {1}{2} \\ 1 0 0 & p = \frac {1}{2} \end{array} \right.. +$$ + +- What are the expected values? 0, 0, 0. +- But there is much greater spread in $Z$ than $Y$ and $Y$ than $X$ . Thus expectation is not enough to detect spread, or variation. + +DEFINITION. If $X$ is a r.v with mean $\mu = \mathbb{E}X$ , then the variance of $X$ , denoted by $\operatorname{Var}(X)$ , is defined by + +$$ +\operatorname{Var} (X) = \mathbb {E} \left[ (X - \mu) ^ {2} \right]. +$$ + +- Remark: $\mathbb{E}c = c$ . +- We prove an alternate formula for the variance. (The technique of using linearity is important here!!! Hint Hint) + +$$ +\operatorname{Var} (X) = \mathbb {E} \left[ (X - \mu) ^ {2} \right] +$$ + +$$ += \mathbb {E} \left[ X ^ {2} - 2 \mu X + \mu^ {2} \right] +$$ + +$$ += \mathbb {E} [ X ^ {2} ] - 2 \mu \mathbb {E} [ X ] + \mathbb {E} [ \mu^ {2} ] +$$ + +$$ += \mathbb {E} \left[ X ^ {2} \right] - 2 \mu^ {2} + \mu^ {2} +$$ + +$$ += \mathbb {E} \left[ X ^ {2} \right] - \mu^ {2}. +$$ + +THEOREM. We have that + +$$ +\operatorname{Var} (X) = \mathbb {E} [ X ^ {2} ] - (\mathbb {E} [ X ]) ^ {2}. +$$ + +\- Example1: Calculate $\operatorname{Var}(X)$ if $X$ represents the outcome when a fair die is rolled. + +\- Solution: Previously we calculated that $\mathbb{E}X = \frac{7}{2}$ . + +\- Thus we only need to calculate the second moment: + +$$ +\mathbb {E} X ^ {2} = 1 ^ {2} \left(\frac {1}{6}\right) + \dots + 6 ^ {2} \frac {1}{6} +$$ + +$$ += \frac {9 1}{6}. +$$ + +\- Using our formula we have that + +$$ +\begin{array}{l} \operatorname{Var} (X) = \mathbb {E} [ X ^ {2} ] - (\mathbb {E} [ X ]) ^ {2} \\ = \frac {9 1}{6} - \left(\frac {7}{2}\right) ^ {2} \\ = \frac {3 5}{1 2}. \\ \end{array} +$$ + +• Here is a useful formula: + +PROPOSITION 7. For constants $a, b$ we have that $\operatorname{Var}(aX + b) = a^2\operatorname{Var}(X)$ . + +PROOF. We compute + +$$ +\begin{array}{l} \operatorname{Var} (a X + b) = \mathbb {E} \left[ (a X + b - \mathbb {E} [ a X + b ]) ^ {2} \right] \\ = \mathbb {E} \left[ (a X + b - a \mu - b) ^ {2} \right] \\ = \mathbb {E} \left[ a ^ {2} (X - \mu) ^ {2} \right] \\ = a ^ {2} \mathbb {E} \left[ (X - \mu) ^ {2} \right] \\ = a ^ {2} \operatorname{Var} (X). \\ \end{array} +$$ + +DEFINITION. We define + +$$ +S D (X) = \sqrt {\operatorname{Var} (X)} +$$ + +to be the standard deviation of $X$ . + +![](images/275c260817f53b0c5226c9641d495030dbd3e5584cdba4cdfdfb77829dd4dcd2.jpg) + +## CHAPTER 6 + +# Some Discrete Distributions + +## 6.1. Bernouli and Binomial Random Variables + +## • Bernoulli Distribution + +- Suppose that a trial or experiment takes place, whose outcome is either success or failure. +- Let $X = 1$ when the outcome is a success and $X = 0$ if it is a failure. +- The pmf of $X$ is given by + +$$ +p (0) = \mathbb {P} (X = 0) = 1 - p +$$ + +$$ +p (1) = \mathbb {P} (X = 1) = p +$$ + +where $0 \leq p \leq 1$ . + +\- For this $X$ , $X$ is said to be a Bernoulli random variable with parameter $p$ , + +\* We wrtie this as $X \sim \text{Bernoulli}(p)$ , +\* Properties: + +$\cdot \mathbb{E}X = p\cdot 1 + (1 - p)\cdot 0 = p$ +$\cdot\mathbb{E}X^{2}=1^{2}\cdot p+0^{2}(1-p)=p.$ +- So $\operatorname{Var} X = p - p^2 = p(1 - p)$ . + +## • Binomial Distribution: + +\- We say $X$ has a binomial distribution with parameters $n$ and $p$ if + +$$ +p _ {X} (k) = \mathbb {P} \left(X = k\right) = \binom{n}{k} p ^ {k} \left(1 - p\right) ^ {n - k}. +$$ + +\- Interpret: $X =$ the number of successes in $n$ independent trials. + +\* Let's take this as given. + +\- We say $X \sim \text{Binomial}(n, p)$ or $X \sim \text{bin}(n, p)$ . + +## • Properties of the Binomial + +\- Check that probabilities sums to 1: Not really a property but more of a check that $X$ is indeed a random variable: + +\* We need to check two things: + +(1) That $p_X(k) \geq 0$ , and this is obvious from the formula +(2) Need to check that $\sum_{k=0}^{n} p_X(k) = 1$ . + +\* First recall the Binomial Theorem: $\sum_{k=0}^{n}\binom{n}{k}x^k y^{n-k} = (x+y)^n$ . + +\- Then + +$$ +\sum_ {k = 0} ^ {n} p _ {X} (k) = \sum_ {k = 0} ^ {n} \binom{n}{k} p ^ {k} (1 - p) ^ {n - k} = (p + (1 - p)) ^ {n} = 1 ^ {n} = 1. +$$ + +\- Mean: Easiest way to compute $\mathbb{E}X$ is by recognizing that $X = Y_{1} + \dots + Y_{n}$ where $Y_{i}$ are independent Bernoulli's. + +\* Thus $EX = EY_{1} + \cdots + EY_{n} = p + \cdots + p = np.$ +\* We can do this directly too, but this would involve proving that + +$$ +\mathbb {E} X = \sum_ {k = 0} ^ {n} k p (k) = n p, +$$ + +meaning we would have to prove + +$$ +\sum_ {k = 0} ^ {n} k \binom{n}{k} p ^ {k} (1 - p) ^ {n - k} = n p. +$$ + +\- Variance: We first compute the second moment. As before write $X = Y_{1} + \cdots + Y_{n}$ where $Y_{i}$ are bernoulli's. + +$$ +\begin{array}{l} \mathbb {E} X ^ {2} = \mathbb {E} \left(Y _ {1} + \dots + Y _ {n}\right) ^ {2} \\ = \sum_ {k = 1} ^ {n} \mathbb {E} Y _ {k} ^ {2} + \sum_ {i \neq j} \mathbb {E} \left[ Y _ {i} Y _ {j} \right] \\ = \sum_ {k = 1} ^ {n} p + \sum_ {i \neq j} \mathbb {E} \left[ Y _ {i} Y _ {j} \right] \\ = n p + \sum_ {i \neq j} \mathbb {E} \left[ Y _ {i} Y _ {j} \right] \\ = (\star) \\ \end{array} +$$ + +\* Now each term $\mathbb{E}[Y_i Y_j]$ for fixed $i, j$ can be computed as = + +$$ +\begin{array}{l} \mathbb {E} \left[ Y _ {i} Y _ {j} \right] = 1 \cdot \mathbb {P} \left(Y _ {i} Y _ {j} = 1\right) + 0 \cdot \mathbb {P} \left(Y _ {i} Y _ {j} = 0\right) \\ = \mathbb {P} ((Y _ {i} = 1) \cap (Y _ {j} = 1)) \\ = \mathbb {P} (Y _ {i} = 1) \mathbb {P} (Y _ {j} = 1), \text { by independence } \\ = p ^ {2}. \\ \end{array} +$$ + +\* Now there are a total of $n^2$ terms is $(Y_1 + \dots + Y_n)^2$ , $n$ of which are of form $Y_k^2$ . Thus there are $n^2 - n$ terms of the form $Y_iY_j$ with $i \neq j$ . + +\* Hence using $(\star)$ we have $\mathbb{E}X^2 = np + (n^2 - n)p^2$ . + +\* Thus + +$$ +\operatorname{Var} X = \mathbb {E} X ^ {2} - (\mathbb {E} X) ^ {2} = n p + (n ^ {2} - n) p ^ {2} - (n p) ^ {2} = n p (1 - p). +$$ + +\- Sumarize: $\mathbb{E}X = np$ and $\operatorname{Var}X = np(1 - p)$ . + +\- $\underline{\text{Moments}}$ : We can also prove $\mathbb{E}X^k = np\mathbb{E}\left[(Y + 1)^{k - 1}\right]$ . + +\- Calculator(TI-84): + +\- 2ndDistri>binomialpdf(n,p,x)=P(X=x). + +\- same with cdf. + +\- Example1: A company prices its hurricane insurance using the following assumptions: + +\- (i) In any calendar year, there can be at most one hurricane. + +\- (ii) In any calendar year, the probability of a hurricane is 0.05. + +\- (iii) The numbers of hurricanes in different calendar years are mutually independent. Using the company's assumptions, calculate the probability that there are fewer than 3 hurricanes in a 20-year period + +\- Solution: We have that $X \sim \text{bin}(20, .05)$ then + +$$ +\begin{array}{l} \mathbb {P} (X < 3) = \mathbb {P} (X \leq 2) \\ = \binom{2 0}{0} (. 0 5) ^ {0} (. 9 5) ^ {2 0} + \binom{2 0}{1} (. 0 5) ^ {1} (. 9 5) ^ {1 9} + \binom{2 0}{2} (. 0 5) ^ {2} (. 9 5) ^ {1 2} \\ = . 9 2 4 5. \\ \end{array} +$$ + +\- Example2: Phan has a .6 probability of making a free throw. Suppose each free throw is independent of the other. If he attempts 10 free throws, what is the probability that he makes at least 2 of them? + +\- Solution: Let $X \sim \text{bin}(10, .6)$ then + +$$ +\begin{array}{l} \mathbb {P} (X \geq 2) = 1 - \mathbb {P} (X = 0) - \mathbb {P} (X = 1) \\ = 1 - \binom{1 0}{0} (. 6) ^ {0} (. 4) ^ {1 0} - \binom{1 0}{1} (. 6) ^ {1} (. 4) ^ {9} \\ = . 9 9 8. \\ \end{array} +$$ + +## 6.2. The Poisson Distribution + +\- We say that $X = 0,1,2,\ldots$ is Poisson with parameter $\lambda > 0$ if + +$$ +p _ {X} (i) = \mathbb {P} (X = i) = e ^ {- \lambda} \frac {\lambda^ {i}}{i !} \text { for } i = 0, 1, 2, 3, \dots . +$$ + +\- Or $X \sim \mathrm{Poisson}(\lambda)$ . + +\- In general Poisson random variables are of the following form + +\- Suppose success happens $\lambda$ times on average in a given period (per year, per month etc). Then $X =$ number of times success happens in that given period. + +\- Possion is like binomial, expect, $X$ is infinitely countable! + +• Examples that obey Poisson R.V + +\- 1. The number of misprints on a page ogf a book + +\- 2. # of people in community that survive to age 100 + +\- 3. # of telephone numbers that are dialed in a day. + +\- 4. # of customers entering post office on a day. + +\- Calc2: Recall that $\sum_{n=0}^{\infty} \frac{x^n}{n!} = e^x$ . + +\- Properties of Poisson: Let $X \sim \text{Poisson}(\lambda)$ + +\- First we check that $p_X(i)$ is indeed a pmf: First it is obvious that $p_X(i) \geq 0$ since $\lambda > 0$ . We to need to check that all the probabilities add up to one: + +$$ +\sum_ {i = 0} ^ {\infty} p _ {X} (i) = \sum_ {i = 0} ^ {\infty} e ^ {- \lambda} \frac {\lambda^ {i}}{i !} = e ^ {- \lambda} \sum_ {i = 0} ^ {\infty} \frac {\lambda^ {i}}{i !} = e ^ {- \lambda} e ^ {\lambda} = 1. +$$ + +\- Mean: We have + +$$ +\mathbb {E} X = \sum_ {i = 0} ^ {\infty} i e ^ {- \lambda} \frac {\lambda^ {i}}{i !} = e ^ {- \lambda} \lambda \sum_ {i = 1} ^ {\infty} \frac {\lambda^ {i - 1}}{(i - 1) !} +$$ + +$$ += e ^ {- \lambda} \lambda e ^ {\lambda} = \lambda . +$$ + +\- Variance: We first have + +$$ +\begin{array}{l} \mathbb {E} X ^ {2} = \sum_ {i = 0} ^ {\infty} i ^ {2} \frac {e ^ {- \lambda} \lambda^ {i}}{i !} \\ = \lambda \sum_ {i = 0} ^ {\infty} i \frac {e ^ {- \lambda} \lambda^ {i - 1}}{(i - 1) !} \\ = \lambda \sum_ {j = 0} ^ {\infty} (j + 1) \frac {e ^ {- \lambda} \lambda^ {j}}{j !}, \text { let } j = i - 1 \\ = \lambda \left[ \sum_ {j = 0} ^ {\infty} j \frac {e ^ {- \lambda} \lambda^ {j}}{j !} + \sum_ {j = 0} ^ {\infty} \frac {e ^ {- \lambda} \lambda^ {j}}{j !} \right] \\ = \lambda [ \lambda + e ^ {- \lambda} e ^ {\lambda} ] \\ = \lambda (\lambda + 1). \\ \end{array} +$$ + +Thus + +$$ +\operatorname{Var} X = \lambda (\lambda + 1) - \lambda^ {2} = \lambda . +$$ + +\- Example1: Suppose on average there are 5 homicides per month in Hartford, CT. What is the probability there will be at most 1 in a certain month? + +\- Answer: If $X$ is the number of homicides, we are given that $EX = 5$ . Since the expectation for a Poisson is $\lambda = 5$ . Therefore $P(X = 0) + P(X = 1) = e^{-5} + 5e^{-5}$ . + +\- Example2: Suppose on average there is one large earthquake per year in Mexico. What's the probability that next year there will be exactly 2 large earthquakes? + +\- $\underline{\text{Answer:}}\lambda = EX = 1$ , so $P(X = 2) = \frac{e^{-1}}{2}$ . + +\- Example3: Phan receives texts on the average of two every 3 minutes. Assume Poisson. + +– Question: What is the probability of five or more texts arriving in a 9-minute period. + +\- Answer: Let $X$ number of calls in a 9-minute period. Let $n =$ number of periods, $\lambda_1 = 2$ . Thus $\lambda = 3 \cdot 2 = 6$ . Thus + +$$ +\begin{array}{l} \mathbb {P} (X \geq 5) = 1 - \mathbb {P} (X \leq 4) \\ = 1 - \sum_ {n = 0} ^ {4} \frac {e ^ {- 6} 6 ^ {n}}{n !} \\ = 1 -. 2 8 5 = . 7 1 5. \\ \end{array} +$$ + +\- Important: Poisson is similar to Binomial in the following way + +\- FACT: Poisson approximates $Bin(n,p)$ when $n$ is large and $p$ is small enough so that $np$ is of moderate size. + +THEOREM 8. If $X_{n}$ is binomial with parameters $n$ and $p_n$ and $np_n \to \lambda$ , then + +$$ +\mathbb {P} \left(X _ {n} = i\right)\rightarrow \mathbb {P} (Y = i) +$$ + +where $Y \sim \text{Poisson}(\lambda)$ . + +PROOF. See class textbook. + +![](images/cd064e0ec628e2ab5741778dfd20bd09eb428c877936240cb08730f39a214f00.jpg) + +\- Summary of Theorem: This theorem says that suppose $n$ is large and $p$ is small, Thus + - If $X \sim Bin(n, p)$ then we approximate $X$ with a possession by letting let $\lambda = np$ so that + +$$ +\mathbb {P} (X = i) \approx e ^ {- n p} \frac {(n p) ^ {i}}{i !}. +$$ + +\- When can we assume $X$ is Poisson: Another consequence of this theorem says that when $Y =$ "the number of successes in a given period". And if the number possible of trials $n$ is large, and if the probability $p$ of success is small, then $Y$ can be treated as a Poisson random variable. + +NOTE: + +\- (1)Why is number of misprints on a page will be approximately Poisson with $\lambda = np$ + +\* Let $X =$ number of misprints on a page of a book. + +\* Since prob of error, say $p = .01$ is usually small, and number of letters on a page is usually large, say $n = 1000$ . Then the average is $\lambda = np$ . + +\* Then because $p$ is small and $n$ is large, then $X$ can be approximated by a Poisson. + +\- (2) Let $X$ number of accidents in a year + +\* $X$ is Poisson because the probability of an accident $p$ in a given period is usually small and while the number $n$ of times someone drives in a given period is high. + +• Example: Here is an example showing this. + +\- If $X$ is number of times you get heads on a biased coin where $\mathbb{P}(H) = \frac{1}{100}$ . Suppose you you toss 1000 times. Then $np = 10$ + +$$ +\mathbb {P} (X = 5) \approx e ^ {- 1 0} \frac {1 0 ^ {5}}{5 !} = . 0 3 7 8 +$$ + +while the actual value is + +$$ +\begin{array}{l} \mathbb {P} \left(X = 5\right) = \binom{1 0 0 0}{5} (. 0 1) ^ {5} (. 9 9) ^ {9 9 5} \\ = \frac {1 0 0 0 !}{9 9 5 ! 5 !} (. 0 1) ^ {5} (. 9 9) ^ {9 9 5} \\ = . 0 3 7 5. \\ \end{array} +$$ + +## 6.3. Other Discrete Distributions + +## • Uniform Distribution: + +\- We say $X$ is uniform, and write this as $X \sim \text{uniform}(n)$ , if $X \in \{1, 2, \ldots, n\}$ and + +$$ +p _ {X} (i) = \mathbb {P} (X = i) = \frac {1}{n} \text { for } i = 1, 2, \dots , n. +$$ + +\- $\underline{\text{Exercise:}}$ $\mathbb{E}X = \sum_{i=1}^{n} i\frac{1}{n} = \frac{1}{n}\sum_{i=1}^{n} i = \frac{1}{n}\frac{n(n-1)}{2} = \frac{n-1}{2}$ and find $\operatorname{Var}X$ . + +## - Geometric Distribution: + +- Experiment: Suppose that independent trials are held until success occurs. Trials are stopped once success happens. Let $p$ be the probability of having a success in each trial. +- Let $X =$ "number of trials required until first success occurs". Thus $X \in \{1,2,3,4,\ldots\}$ Here we have + +$$ +p _ {X} (i) = \mathbb {P} (X = i) = (1 - p) ^ {i - 1} p \text {for} i = 1, 2, 3, 4 \dots . +$$ + +- We say $X \sim \operatorname{geometric}(p)$ . +- Properties: + +\* We first double check is indeed a discrete random variable: This follows from what we know about geometric series: + +$$ +\sum_ {i = 1} ^ {\infty} \mathbb {P} (X = i) = \sum_ {i = 1} ^ {\infty} (1 - p) ^ {i - 1} p = \frac {p}{1 - (1 - p)} = 1. +$$ + +\* Mean: Recall that by differentiation of the geometric series, we came up with the formula $\sum_{n=0}^{\infty} nx^{n-1} = \frac{1}{(1-x)^2}$ , so that + +$$ +\begin{array}{l} \mathbb {E} X = \sum_ {i = 1} ^ {\infty} i \mathbb {P} (X = i) \\ = \sum_ {i = 1} ^ {\infty} i (1 - p) ^ {i - 1} p \\ = \frac {p}{(1 - (1 - p)) ^ {2}} = \frac {1}{p}. \\ \end{array} +$$ + +\* Variance:(Leave as Exercise for student) Note that + +$$ +\mathbb {E} X ^ {2} = \sum_ {i = 1} ^ {\infty} i ^ {2} (1 - p) ^ {i - 1} p. (\star) +$$ + +Thus we can differentiate $\sum_{n=1}^{\infty} nx^{n-1} = \frac{1}{(1-x)^2}$ again to get $\sum_{n=2}^{\infty} n(n-1)x^{n-2} = \frac{2}{(1-x)^3}$ . + +\* From this we will attempt to get $EX^{2}$ in ( $\star$ ) by splitting the sum up: + +$$ +\sum_ {n = 2} ^ {\infty} n (n - 1) (1 - p) ^ {n - 2} = \frac {2}{(1 - (1 - p)) ^ {3}} = \frac {2}{p ^ {3}}, +$$ + +$$ +\sum_ {n = 2} ^ {\infty} n (n - 1) (1 - p) ^ {n - 2} p = \frac {2}{p ^ {2}}, \text { now split }, +$$ + +$$ +\sum_ {n = 2} ^ {\infty} n ^ {2} (1 - p) ^ {n - 2} p = \frac {2}{p ^ {2}} + \sum_ {n = 2} ^ {\infty} n (1 - p) ^ {n - 2} p +$$ + +$$ +(1 - p) ^ {- 1} \sum_ {n = 1} ^ {\infty} n ^ {2} (1 - p) ^ {n - 1} p = \frac {2}{p ^ {2}} + \sum_ {n = 2} ^ {\infty} n (1 - p) ^ {n - 2} p + (1 - p) ^ {- 1} p +$$ + +$$ +\begin{array}{l} (1 - p) ^ {- 1} \mathbb {E} X ^ {2} = \frac {2}{p ^ {2}} + (1 - p) ^ {- 1} \sum_ {n = 1} ^ {\infty} n (1 - p) ^ {n - 1} p \\ = \frac {2}{p ^ {2}} + (1 - p) ^ {- 1} \frac {1}{p} \\ \end{array} +$$ + +Thus + +$$ +\begin{array}{l} \mathbb {E} X ^ {2} = \frac {2 (1 - p)}{p ^ {2}} + \frac {1}{p} \\ = \frac {2 - 2 p + p}{p ^ {2}} = \frac {2 - p}{p ^ {2}} \\ \end{array} +$$ + +\* So Thus + +$$ +\begin{array}{l} \operatorname{Var} X = \mathbb {E} X ^ {2} - (\mathbb {E} X) ^ {2} = \frac {2 - p}{p ^ {2}} - \frac {1}{p ^ {2}} \\ = \frac {(1 - p)}{p ^ {2}} \\ \end{array} +$$ + +\- Example1: An urn contains 10 white balls and 15 black balls. Balls are randomly selected, one at a time, until a black one is obtained. If we assume that each ball selected is replaced before the next one is drawn, what is the probability. + +\- Part (a): Exactly 6 draws are needed? + +\* X = number of draws needed to select a black ball, the probability of success is + +$$ +p = \frac {1 5}{1 0 + 1 5} = \frac {1 5}{2 5} = . 6. +$$ + +\* Thus + +$$ +\mathbb {P} (X = 6) = (. 4) ^ {6 - 1} (. 6) = . 0 0 6 1 4 4 +$$ + +\- Part (a): What is the expected number of draws in this game? + +\* Since $X \sim$ geometric(.6) then + +$$ +\mathbb {E} X = \frac {1}{p} = \frac {1 0}{6} = 1. \bar {6} +$$ + +\- Part (c)(Extra Problem to be done at home) Find exactly that probability at least $k$ draws are needed? + +\* We have that + +$$ +\begin{array}{l} \mathbb {P} (X \geq k) = \sum_ {n = k} ^ {\infty} \mathbb {P} (X = k) \\ = \sum_ {n = k} ^ {\infty} (. 4) ^ {n - 1} (. 6) \\ = (. 6) (. 4) ^ {- 1} \sum_ {n = k} ^ {\infty} (. 4) ^ {n} \\ = (. 6) (. 4) ^ {- 1} (. 4) ^ {k} \sum_ {n = 0} ^ {\infty} (. 4) ^ {n} \\ = (. 6) (. 4) ^ {k - 1} \frac {1}{1 - . 4} \\ = (. 4) ^ {k - 1}. \\ \end{array} +$$ + +\- Note: This could have been done for a general $p$ . Thus + +$$ +\mathbb {P} (X \geq k) = (1 - p) ^ {k - 1}. +$$ + +\- Negative Binomial(Need to know for Actuarial Exam): + +\- Experiment: Suppose that independent trials are held with probability $p$ of having a success. The trials are perfomed until a total of $r$ successes are accumulated. + +\* Let $X$ equal the number of trials required to obtain $r$ successes. Here we have + +$$ +\mathbb {P} \left(X = n\right) = \binom{n - 1}{r - 1} p ^ {r} \left(1 - p\right) ^ {n - r} \text { for } n = r, r + 1, \ldots . +$$ + +- We say $X \sim \text{NegativeBinomial}(r, p)$ . +- Properties: + +\* This is a probability mass function. Can check that $\sum_{n=r}^{\infty}\mathbb{P}(X=n)=1$ . + +\* Mean: + +$$ +\mathbb {E} X = \frac {r}{p}. +$$ + +\* Variance: + +$$ +\operatorname{Var} (X) = \frac {r (1 - p)}{p ^ {2}}. +$$ + +\- Note that $\operatorname{Geometric}(p) = \text{NegativeBinomial}(1, p)$ . + +\- Example: Find the expected value of the number of times one must throw a die until the outcome 1 has occurred 4 times. + +\- Solution: $X \sim \text{NegativeBinomial}\left(4, \frac{1}{6}\right)$ . So + +$$ +\mathbb {E} X = \frac {4}{\frac {1}{6}} = 2 4. +$$ + +• Hypergeometric Distribution(Need to know for Actuarial Exam): + +\- Experiment: Suppose that a sample of size $n$ is to be chosen randomly (without replacement) from an urn containing $N$ balls, of which $m$ are white and $N - m$ are black. + +\* Let $X$ equal the number of white balls selected. Then + +$$ +\mathbb {P} \left(X = i\right) = \frac {\binom{m}{i} \binom{N - m}{n - i}}{\binom{N}{n}} \text { for } n = 0, 1, \ldots , n. +$$ + +\- We say $X \sim \text{Hypergeometric}(n, N, m)$ . + +\- Properties: + +\* Mean: + +$$ +\mathbb {E} X = \frac {n m}{N}. +$$ + +\* Variance: + +$$ +\operatorname{Var} (X) = n \frac {m}{N} \left(1 - \frac {m}{N}\right) \left(1 - \frac {n - 1}{N - 1}\right). +$$ + +## CHAPTER 7 + +# Continuous Random Variables + +## 7.1. Intro to continuous R.V + +DEFINITION. A random variable $X$ is said to have a continuous distribution if there exists a non-negative function $f$ such that + +$$ +\mathbb {P} (a \leq X \leq b) = \int_ {a} ^ {b} f (x) d x +$$ + +for every $a$ and $b$ . [Sometimes we write that for nice sets $B \subset \mathbb{R}$ we have $\mathbb{P}(X \in B) = \int_{B} f(x) dx$ .] + +We call $f$ the pdf (probability density function) for $X$ . Sometime we we the notation $f_{X}$ to signify $f_{X}$ corresponds to the pdf of $X$ . We sometimes call $f_{X}$ the density of $X$ . + +\- In fact, any function $f$ satisfying the following two properties is called a density, and could be considered a pdf of some random variable $X$ : + +(1) $f(x)\geq 0$ for all $x$ +(2) $\int_{-\infty}^{\infty} f(x) dx = 1$ . + +\- Important Note! + +- (1) In this case $X: S \to \mathbb{R}$ and the could attain uncountably many values (doesn't have to discrete) +- (2) $\int_{-\infty}^{\infty} f(x) dx = \mathbb{P}(-\infty < X < \infty) = 1$ . +- (3) $\mathbb{P}(X = a) = \int_{a}^{a} f(x) dx = 0$ . +- (4) $\mathbb{P}(X < a) = \mathbb{P}(X \leq a) = F(a) = \int_{-\infty}^{a} f(x) dx$ . + +\* Recall that $F$ is the cdf of $X$ . + +\- (5) Draw a pdf of $X$ + +\* Note that $\mathbb{P}(a < X < b)$ is just the area under the curve. + +\- Remark: What are some random variables that are considered continuous? + +- Let $X$ be the time it takes it take for a student to finish a probability exam. $X \in (0, \infty)$ . +- Let $X$ be the value of a Apple's stock price at the end of the day. Again $X \in [0, \infty)$ . +- Let $X$ be the height of a college student. +- Any sort of continuous measurement can be considered a continuous random variable. + +• Example1: Suppose we are given + +$$ +f (x) = \left\{ \begin{array}{l l} \frac {c}{x ^ {3}} & x \geq 1 \\ 0 & x < 1 \end{array} \right. +$$ + +is the pdf of $X$ . What must the value of $c$ be? + +\- Solution: We would need + +$$ +1 = \int_ {- \infty} ^ {\infty} f (x) d x = c \int_ {1} ^ {\infty} \frac {1}{x ^ {3}} d x = \frac {c}{2}, +$$ + +thus c = 2. + +• Example2: Suppose we are given + +$$ +f _ {X} (x) = \left\{ \begin{array}{l l} \frac {2}{x ^ {3}} & x \geq 1 \\ 0 & x < 1 \end{array} \right. +$$ + +is the pdf of $X$ from Example1. + +\- Part (a): Find the c.d.f, $F_{X}(x)$ . + +\* Solution: First we check that if x < 1 then + +$$ +F _ {x} (x) = \mathbb {P} (X \leq x) = \int_ {- \infty} ^ {x} f _ {X} (y) d y = \int_ {- \infty} ^ {x} 0 d y = 0. +$$ + +Now when $x \geq 1$ we have + +$$ +\begin{array}{l} F _ {X} (x) = \mathbb {P} (X \leq x) = \int_ {- \infty} ^ {x} f _ {X} (y) d y \\ = \int_ {- \infty} ^ {1} 0 d y + \int_ {1} ^ {x} \frac {2}{y ^ {3}} d y \\ = \int_ {1} ^ {x} \frac {2}{y ^ {3}} d y \\ = 1 - \frac {1}{x ^ {2}}. \\ \end{array} +$$ + +thus + +$$ +F _ {X} (x) = \left\{ \begin{array}{l l} 1 - \frac {1}{x ^ {2}} & x \geq 1 \\ x & x < 1 \end{array} \right. +$$ + +\- Part (b): Use the cdf in Part (a) to help you find $\mathbb{P}(3\leq X\leq 4)$ . + +\* Solution: We have + +$$ +\begin{array}{l} \mathbb {P} (3 \leq X \leq 4) = \mathbb {P} (X \leq 4) - \mathbb {P} (X < 3) \\ = F _ {X} (4) - F _ {X} (3) \\ = \left(1 - \frac {1}{4 ^ {2}}\right) - \left(1 - \frac {1}{3 ^ {2}}\right) = \frac {7}{1 4 4}. \\ \end{array} +$$ + +\- Fact: For continuous R.V we have the following useful relationship + +\- Since $F(x) = \int_{-\infty}^{x} f(y) dy$ then by the fundament at theorem of calculus (Do you remember this form Calculus 1 or 2?) + +$$ +F ^ {\prime} (x) = f (x). +$$ + +\- This means that for continuous random variables, the derivative of the CDF is the PDF! + +\- Example3: Let + +$$ +f (x) = \left\{ \begin{array}{l l} c e ^ {- 2 x} & x \geq 0 \\ 0 & x < 0 \end{array} \right. +$$ + +Find c. + +\- Solution: $c = 2$ . + +## 7.2. Expectation and Variance + +\- Recall that if $p(x)$ is the pmf (density) of a discrete random variable, we had + +$$ +\mathbb {E} X = \sum_ {i = 1} ^ {\infty} x _ {i} p (x _ {i}). +$$ + +DEFINITION. If $X$ is continuous with density $f(x)$ then + +$$ +\mathbb {E} X = \int_ {- \infty} ^ {\infty} x f (x) d x. +$$ + +• Example1: Suppose X has density + +$$ +f (x) = \left\{ \begin{array}{l l} 2 x & 0 \leq x \leq 1 \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +Find $\mathbb{E}X$ . + +\- Solution: We have that + +$$ +\begin{array}{l} \mathbb {E} [ X ] = \int_ {- \infty} ^ {\infty} x f (x) d x \\ = \int_ {0} ^ {1} x \cdot 2 x d x \\ = \frac {2}{3}. \\ \end{array} +$$ + +THEOREM 9. If $X$ and $Y$ are continuous random variable then + +(a) $\mathbb{E}[X + Y] = \mathbb{E}X + \mathbb{E}Y.$ +(b) $E[aX]=aE X$ where $a\in R$ . + +PROOF. See textbook. It will be shown later. + +![](images/1e610fd0cd4de09f82ce873bfc6d251bb82d73d0948434e96e152f0298f14cc0.jpg) + +PROPOSITION. If $X$ is a continuous $R.V.$ with pdf $f(x)$ , then for any real valued function $g$ , + +$$ +\mathbb {E} \left[ g (X) \right] = \int_ {- \infty} ^ {\infty} g (x) f (x) d x. +$$ + +• Example2: The density of X is given by + +$$ +f (x) = \left\{ \begin{array}{l l} \frac {1}{2} & \text { if } 0 \leq x \leq 2 \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +Find $\mathbb{E}\left[e^{X}\right]$ . + +\- Solution: From the previous proposition we have that $g(x) = e^{x}$ in this case thus + +$$ +\mathbb {E} e ^ {X} = \int_ {0} ^ {2} e ^ {x} \cdot \frac {1}{2} d x = \frac {1}{2} \left[ e ^ {2} - 1 \right]. +$$ + +LEMMA 10. For nonnegative random variable $Y \geq 0$ we have + +$$ +\mathbb {E} Y = \int_ {0} ^ {\infty} \mathbb {P} (Y > y) d y. +$$ + +\- Bonus: + +\- This proof is a good practice with interchanging order of integrals in Multivariable Calculus. + +PROOF. Recall that dxdy means Right-Left and dydx means Top-Bottom. + +$$ +\begin{array}{l} \int_ {0} ^ {\infty} \mathbb {P} (Y > y) d y = \int_ {0} ^ {\infty} \int_ {y} ^ {\infty} f _ {Y} (x) d x d y \\ = \int \int_ {D} f _ {Y} (x) d y d x, \text { interchange order in Calc III } \\ = \int_ {0} ^ {\infty} \int_ {0} ^ {x} f _ {Y} (x) d y d x \text { draw the region to do this } \\ = \int_ {0} ^ {\infty} x f _ {Y} (x) d x \\ = \mathbb {E} X. \\ \end{array} +$$ + +\- Variance: + +\- Will be define in the same way as we did with discrete random variable: + +$$ +\operatorname{Var} (X) = \mathbb {E} \left[ (X - \mu) ^ {2} \right] +$$ + +$$ +\operatorname{Var} (X) = \mathbb {E} X ^ {2} - (\mathbb {E} X) ^ {2}. +$$ + +\- As before + +$$ +\operatorname{Var} (a X + b) = a ^ {2} \operatorname{Var} (X). +$$ + +\- Example3: (Example 1 continued) Suppose $X$ has density + +$$ +f (x) = \left\{ \begin{array}{l l} 2 x & 0 \leq x \leq 1 \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +Find $\operatorname{Var}(X)$ . + +\- Solution: From Example 1 we found $\mathbb{E}[X] = \frac{2}{3}$ . Now + +$$ +\begin{array}{l} \mathbb {E} \left[ X ^ {2} \right] = \int_ {0} ^ {1} x ^ {2} \cdot 2 x d x = 2 \int_ {0} ^ {1} x ^ {3} d x \\ = \frac {1}{2}. \\ \end{array} +$$ + +Thus + +$$ +\operatorname{Var} (X) = \frac {1}{2} - \left(\frac {2}{3}\right) ^ {2} = \frac {1}{1 8}. +$$ + +• Example4: Suppose X has density + +$$ +f (x) = \left\{ \begin{array}{l l} a x + b & 0 \leq x \leq 1 \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +and that $\mathbb{E}\left[X^2\right] = \frac{1}{6}$ . Find the values of $a$ and $b$ . + +\- Solution: We need to use the fact that $\int_{-\infty}^{\infty}f(x)dx = 1$ and $\mathbb{E}\left[X^2\right] = \frac{1}{6}$ . The first one gives us, + +$$ +1 = \int_ {0} ^ {1} (a x + b) d x = \frac {a}{2} + b +$$ + +and the second one give us + +$$ +\frac {1}{6} = \int_ {0} ^ {1} x ^ {2} (a x + b) d x = \frac {a}{4} + \frac {b}{3}. +$$ + +Solving these equations gives us + +$$ +a = - 2, \text { and } b = 2. +$$ + +## 7.3. The uniform Random Variable + +\- A continuous random variable is said to be uniformly distributed on the interval $[a, b]$ if + +$$ +f _ {X} (x) = \left\{ \begin{array}{l l} \frac {1}{b - a} & a \leq x \leq b \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +- So $X$ can only attain values in $X \in [a, b]$ . +- We say $X \sim \text{Uniform}(a, b)$ . +- The cdf is + +$$ +F _ {X} (x) = \left\{ \begin{array}{l l} 0 & x < a \\ \frac {x - a}{b - a} & a \leq x \leq b \\ 1 & x > b \end{array} \right.. +$$ + +\- Example1: Suppose $X \sim \text{Uniform}(a, b)$ Part (a) Find the mean of $X$ . Part (b) Find the variance of $X$ . + +\- Part (a): We compute + +$$ +\begin{array}{l} \mathbb {E} X = \int_ {- \infty} ^ {\infty} x f _ {X} (x) d x = \int_ {a} ^ {b} x \frac {1}{b - a} d x \\ = \frac {1}{b - a} \left(\frac {b ^ {2}}{2} - \frac {a ^ {2}}{2}\right) = \frac {a + b}{2}. \\ \end{array} +$$ + +\* Which makes sense right? It should be the midpoint of the interval $[a, b]$ . + +\- Part(b): We compute first the second moment + +$$ +\begin{array}{l} \mathbb {E} X ^ {2} = \int_ {a} ^ {b} x ^ {2} \frac {1}{b - a} d x = \frac {1}{b - a} \left(\frac {b ^ {3}}{3} - \frac {a ^ {3}}{3}\right) \\ = \frac {1}{3} \frac {1}{b - a} (b - a) (a ^ {2} + a b + b ^ {2}) \\ = \frac {a ^ {2} + a b + b ^ {2}}{3}. \\ \end{array} +$$ + +Thus after some algebra + +$$ +\operatorname{Var} X = \frac {a ^ {2} + a b + b ^ {2}}{3} - \left(\frac {a + b}{2}\right) ^ {2} = \frac {(b - a) ^ {2}}{1 2}. +$$ + +## 7.4. More practice + +\- Suppose we are given the pd.f. + +$$ +f (x) = \left\{ \begin{array}{l l} 9 e ^ {- 9 x} & x \geq 0 \\ 0 & x < 0 \end{array} \right. +$$ + +\- Part (a): Set up integral to find $F_{X}(x)$ : + +\* We have for $x > 0$ , that + +$$ +F _ {X} (x) = \int_ {0} ^ {x} 9 e ^ {- 9 y} d y = 1 - e ^ {- 9 x}, +$$ + +so that + +$$ +F _ {X} (x) = \left\{ \begin{array}{l l} 1 - e ^ {- 9 x} & x \geq 0 \\ 0 & x < 0 \end{array} \right. / +$$ + +\- Part (b): Set up integral to find $\mathbb{P}(1 < X < 5)$ + +$$ +* \int_ {1} ^ {5} 9 e ^ {- 9 x} d x +$$ + +\- Part (c): Set up integral to find $\mathbb{P}(X > 3)$ + +$$ +* \int_ {3} ^ {\infty} 9 e ^ {- 9 x} d x. +$$ + +\- Part (s): Set up integral to find $\mathbb{P}(X < 2)$ + +$$ +* \int_ {0} ^ {2} 9 e ^ {- 9 x} d x. +$$ + +## CHAPTER 8 + +## Normal Distributions + +## 8.1. The normal distribution + +\- We say that $X$ is a normal (Gaussian) random variable, or $X$ is normally distributed with parameters $\mu$ and $\sigma^2$ if the density of $X$ is given by + +$$ +f (x) = \frac {1}{\sqrt {2 \pi} \sigma} e ^ {- (x - \mu) ^ {2} / \left(2 \sigma^ {2}\right)}. +$$ + +![](images/dbef8516a5d35088c39e1b33dc1177989c6a686e6ba9e1ff195e6eb14639e048.jpg) + +
    +line chart + +| x | y | +| ---- | ----- | +| -3.0 | 0.000 | +| -2.5 | 0.047 | +| -2.0 | 0.169 | +| -1.5 | 0.362 | +| -1.0 | 0.540 | +| -0.5 | 0.728 | +| 0.0 | 1.000 | +| 0.5 | 0.728 | +| 1.0 | 0.540 | +| 1.5 | 0.362 | +| 2.0 | 0.169 | +| 2.5 | 0.047 | +| 3.0 | 0.000 | +
    + +\- We'll usually write $X \sim \mathcal{N}(\mu, \sigma^2)$ . + +\- Turns out that in practice, many random variable overy the normal distribution + +\* Grades +\* Height of a man or a women + +\- Note the following: + +\- If $X \sim \mathcal{N}(0,1)$ then + +$$ +\int_ {- \infty} ^ {\infty} \frac {1}{\sqrt {2 \pi}} e ^ {- x ^ {2} / 2} d x = 1. +$$ + +\- To show this we use polar coordinates. Let $I = \int_{-\infty}^{\infty} e^{-x^2/2} dx = 2 \int_0^\infty e^{-x^2/2} dx$ The trick is to write + +$$ +\begin{array}{l} I ^ {2} = 4 \int_ {0} ^ {\infty} \int_ {0} ^ {\infty} e ^ {- x ^ {2} / 2} e ^ {- y ^ {2} / 2} d x y \\ = 4 \int_ {0} ^ {\pi / 2} \int_ {0} ^ {\infty} r e ^ {- r ^ {2} / 2} d r = 4 \cdot \frac {\pi}{2} = 2 \pi , \\ \end{array} +$$ + +Thus $I = \sqrt{2\pi}$ as needed. + +THEOREM 11. To help us compute the mean and variance of $X$ its not too hard to show $X \sim \mathcal{N}\left(\mu, \sigma^2\right)$ if and only if + +$$ +\frac {X - \mu}{\sigma} = Z \text { where } Z \sim \mathcal {N} (0, 1). +$$ + +PROOF. We only show the ( $\Longleftrightarrow$ ) direction. Note that + +$$ +\begin{array}{l} F _ {X} (x) = \mathbb {P} (X \leq x) = \mathbb {P} (\sigma Z + \mu \leq x) \\ = \mathbb {P} \left(Z \leq \frac {x - \mu}{\sigma}\right) \\ { = } { F _ { Y } \left( \frac { x - \mu } { \sigma } \right) } \\ \end{array} +$$ + +for $\sigma > 0$ . Similar for $\sigma < 0$ . By the chain rule + +$$ +\begin{array}{l} f _ {X} (x) = F _ {X} ^ {\prime} (x) \\ = F _ {Y} ^ {\prime} \left(\frac {x - \mu}{\sigma}\right) \frac {1}{\sigma} \\ = \frac {f _ {Z} \left(\frac {x - \mu}{\sigma}\right)}{\sigma} \\ { = } { \frac { 1 } { \sigma } \frac { 1 } { \sqrt { 2 \pi } } e ^ { - ( x - \mu ) ^ { 2 } / ( 2 \sigma ^ { 2 } ) } . } \\ \end{array} +$$ + +![](images/200a852e9f1c0e839c0f986a964ffb0886610e3a88cb96ca19bbb8990c4c437f.jpg) + +• Summary of the normal distribution: + +\- If $X \sim \mathcal{N}\left(\mu, \sigma^2\right)$ then $X$ is normally distributed with + +$$ +\mathbb {E} X = \mu , +$$ + +$$ +\operatorname{Var} (X) = \sigma^ {2}. +$$ + +\- If $X \sim \mathcal{N}(\mu, \sigma^2)$ then $X = \sigma Z + \mu$ where $Z \sim \mathcal{N}(0,1)$ . We call $Z$ a standard normal random variable. + +\* A Table of probabilities for Z will be given!!! + +\* This will be called a z-score table. + +\- Z scores: + +\- Because $Z \sim \mathcal{N}(0,1)$ is so important we give it's cumulative distribution function (cdf) a name. The distribution $F_Z(x)$ of $Z$ is + +$$ +\Phi (x) = \mathbb {P} (Z \leq x) = \frac {1}{\sqrt {2 \pi}} \int_ {- \infty} ^ {x} e ^ {- y ^ {2} / 2} d y. +$$ + +\- NOTE: A table of $\Phi(x)$ will be given but only for values of $x > 0$ + +\- Note this is symmetric[ $\underline{DRAW this}$ ] thus here is an important fact: $\Phi(-x)=1-\Phi(x)$ + +![](images/64fe743adecf18b4e700e521caee6cf1fac6e5cc8a04297483424dc3f759a58b.jpg) + +
    +area chart + +| Z | CDF | +| ---- | ------- | +| 0 | 0 | +| z | Peak | +
    + +THEOREM 12. If $X \sim \mathcal{N}(\mu, \sigma)$ then + +$$ +\mathbb {P} (a < X < b) = \mathbb {P} \left(\frac {a - \mu}{\sigma} < Z < \frac {b - \mu}{\sigma}\right). +$$ + +\- Example1: Find $\mathbb{P}(1 \leq X \leq 4)$ if $X \sim \mathcal{N}(2, 25)$ . + +\- $\underline{\mathbf{Answer}}$ : Then $\mu = 2$ and $\sigma^2 = 25$ thus $\frac{X - 2}{5} = Z$ so that + +$$ +\begin{array}{l} \mathbb {P} (1 \leq X \leq 4) = \mathbb {P} \left(\frac {1 - 2}{5} \leq \frac {X - 2}{5} \leq \frac {4 - 2}{5}\right) \\ = \mathbb {P} (-. 2 \leq Z \leq . 4) \\ = \mathbb {P} (X \leq . 4) - \mathbb {P} (X \leq -. 2) \\ = \Phi (. 4) - \Phi (- 2) \\ = . 6 5 5 4 - (1 - \Phi (. 2)) \\ = . 6 5 5 4 - (1 -. 5 7 9 3). \\ \end{array} +$$ + +\- Example2: Suppose $X$ is normal with mean 6. If $\mathbb{P}(X > 16) = .0228$ , then what is the standard deviation of $X$ ? + +\- $\underline{\text{Answer}}$ : We apply our Theorem that says $\frac{X - \mu}{\sigma} = Z$ is $\mathcal{N}(0,1)$ and get + +$$ +\begin{array}{l} \mathbb {P} (X > 1 6) = . 0 2 2 8 \quad \Longleftrightarrow \quad \mathbb {P} \left(\frac {X - 6}{\sigma} > \frac {1 6 - 6}{\sigma}\right) = . 0 2 2 8 \\ \Longleftrightarrow \quad \mathbb {P} \left(Z > \frac {1 0}{\sigma}\right) = . 0 2 2 8 \\ \Longleftrightarrow \quad 1 - \mathbb {P} \left(Z \leq \frac {1 0}{\sigma}\right) = . 0 2 2 8 \\ \Longleftrightarrow \quad 1 - \Phi \left(\frac {1 0}{\sigma}\right) = . 0 2 2 8 \\ \Longleftrightarrow \quad \Phi \left(\frac {1 0}{\sigma}\right) = . 9 7 7 2. \\ \end{array} +$$ + +Using the standard normal table we see that $\Phi(2) = .9772$ , thus we must have that + +$$ +2 = \frac {1 0}{\sigma} +$$ + +and hence $\sigma = 5$ . + +\- Example (Extra): Suppose $X \sim \mathcal{N}(3,9)$ find $\mathbb{P}\left(|X - 3| > 6\right)$ . + +\- Answer: Get + +$$ +\begin{array}{l} \mathbb {P} (| X - 3 | > 6) = \mathbb {P} (X - 3 > 6) + \mathbb {P} (- (X - 3) > 6) \\ = \mathbb {P} (X > 9) + \mathbb {P} (X < - 3) \\ = \mathbb {P} (Z > 2) + \mathbb {P} (Z < - 2) \\ = 1 - \Phi (2) + \Phi (- 2) \\ = 2 (1 - \Phi (2)) \\ \approx . 0 4 5 6. \\ \end{array} +$$ + +• FACT: The 68 - 95 - 99.7 Rule + +- About $68\%$ of all area is contained within 1 standard deviation of the mean +- About $95\%$ of all area is contained within 2 standard deviation of the mean +- About $99.7\%$ of all area is contained within 3 standard deviation of the mean +- This can be explained by the following graph: + +![](images/1e6b1e4b08c6161b53359d9295bf76cf66512da21abf00a5a7a76887318c6619.jpg) + +
    +area chart + +| Standard Deviation | Percentage | +| ------------------ | ---------- | +| -3σ | 0.1% | +| -2σ | 2.1% | +| -1σ | 13.6% | +| μ | 34.1% | +| 1σ | 34.1% | +| 2σ | 13.6% | +| 3σ | 2.1% | +| >3σ | 0.1% | +
    + +## CHAPTER 9 + +# Normal approximations to the binomial + +## 9.1. The normal approximates Binomial + +THEOREM 13. If $S_{n}$ is a binomial with parameter $n$ and $p$ , then + +$$ +\mathbb {P} \left(a \leq \frac {S _ {n} - n p}{\sqrt {n p (1 - p)}} \leq b\right)\rightarrow \mathbb {P} (a \leq Z \leq b) +$$ + +as $n\to \infty$ where $Z$ is a $\mathcal{N}(0,1)$ . + +\- Recall that if $S_{n} \sim Bin(n,p)$ then its mean is $\mu = np$ and standard deviation is $\sigma = \sqrt{np(1 - p)}$ . +- So what this theorem says is that if you want to compute $\mathbb{P}(c \leq S_n \leq d)$ then using the fact that + +$$ +\frac {S _ {n} - n p}{\sqrt {n p (1 - p)}} \approx Z +$$ + +or + +$$ +\frac {S _ {n} - \mu}{\sigma} \approx Z, +$$ + +then + +$$ +\begin{array}{l} \mathbb {P} \left(c \leq S _ {n} \leq d\right) = \mathbb {P} \left(\frac {c - \mu}{\sigma} \leq \frac {S _ {n} - \mu}{\sigma} \leq \frac {d - \mu}{\sigma}\right) \\ \approx \mathbb {P} \left(\frac {c - \mu}{\sigma} \leq Z \leq \frac {d - \mu}{\sigma}\right). \\ \end{array} +$$ + +\- Note that $S_{n}$ is really discrete. In fact $S_{n} \in \{0,1,2,\ldots,100\}$ , while the normal distribution is continuous! + +\- Note that if I tried to estimate an equality: The wrong way to do it would be: + +$$ +\begin{array}{l} \mathbb {P} \left(S _ {n} = i\right) = \mathbb {P} \left(\frac {S _ {n} - \mu}{\sigma} = \frac {i - \mu}{\sigma}\right) \\ \approx \mathbb {P} \left(Z = \frac {i - \mu}{\sigma}\right) = 0 \\ \end{array} +$$ + +as we know that for continuous random variables $X$ we always have $\mathbb{P}(X = a) = 0!$ . + +\- Hence we need inequalities if we want to estimate a discrete random variable using a continuous random variable. + +\* So we use the following convention. $\mathbb{P}(S_n = i) = \mathbb{P}\left(i - \frac{1}{2} < S_n < i + \frac{1}{2}\right)$ . +\* We have no problem here, because $S_{n}$ can only be integers, so we're not hurting anything by saying “ $i - \frac{1}{2} < S_{n} < i + \frac{1}{2}$ ” as we know that $S_{n}$ can only be $i$ in that interval anyways. + +\- Example: Suppose a fair coin is tossed 100 times. + +- (a) What is the probability there will be more than 60 heads? +- $\underline{\text{Answer:}}$ Let $S_{100} \sim Bin\left(100, \frac{1}{2}\right)$ so that $S_{100}$ represents the numbers of heads in 100 coin tosses. + +\* The actual answer would be + +$$ +\begin{array}{l} \mathbb {P} \left(S _ {1 0 0} > 6 0\right) = \sum_ {i = 6 1} ^ {1 0 0} \mathbb {P} \left(S _ {1 0 0} = i\right) \\ = \sum_ {i = 6 1} ^ {1 0 0} \binom{1 0 0}{i} \left(\frac {1}{2}\right) ^ {i} \left(\frac {1}{2}\right) ^ {1 0 0 - i} \\ = \sum_ {i = 6 1} ^ {1 0 0} \binom{1 0 0}{i} \left(\frac {1}{2}\right) ^ {1 0 0}. \\ \end{array} +$$ + +\* But this would be almost impossible to do this the long way by hand. +\* So we will give an approximate answer using the normal distribution: + +\- So here take $\mu = np = 50$ and $\sigma = \sqrt{np(1 - p)} = \sqrt{50\frac{1}{2}} = 5$ . We want more than 60, so approximate using $60 + \frac{1}{5}$ : + +$$ +\begin{array}{l} \mathbb {P} \left(S _ {1 0 0} > 6 0\right) = \mathbb {P} \left(S _ {1 0 0} \geq 6 0. 5\right) = \mathbb {P} \left(\frac {S _ {1 0 0} - 5 0}{5} \geq \frac {6 0 . 5 - 5 0}{5}\right) \\ \approx \mathbb {P} (Z \geq 2. 1) \\ \approx 1 - \Phi (2. 1) \\ = . 0 1 7 9 \\ \end{array} +$$ + +– (b) Estimate the probability of getting exactly 60 heads? + +$$ +\begin{array}{l} \mathbb {P} \left(S _ {n} = 6 0\right) = \mathbb {P} \left(5 9. 5 \leq S _ {n} \leq 6 0. 5\right) \\ \approx \mathbb {P} (1. 9 \leq Z \leq 2. 1) \\ \approx \Phi (2. 1) - \Phi (1. 9). \\ \end{array} +$$ + +## CHAPTER 10 + +# Some continuous distributions + +## 10.1. Exponential Random Variables + +\- A continuous R.V. is said to be exponential with parameter $\lambda$ if its pdf is + +$$ +f (x) = \left\{ \begin{array}{l l} \lambda e ^ {- \lambda x} & \text { if } x \geq 0 \\ 0 & \text { if } x < 0 \end{array} \right. +$$ + +\- We write $X \sim$ exponential $(\lambda)$ . + +\- Summary: + +\- CDF: Let $a > 0$ . Note that the cdf is + +$$ +F _ {X} (a) = \mathbb {P} (X \leq a) = \int_ {0} ^ {a} \lambda e ^ {- \lambda y} d y = - e ^ {\lambda y} | _ {0} ^ {a} = 1 - e ^ {- \lambda a}. +$$ + +\* Thus + +$$ +\mathbb {P} (X > a) = 1 - \mathbb {P} (X \leq a) = e ^ {- \lambda a}. +$$ + +\- Mean: $\mathbb{E}X = \frac{1}{\lambda}$ Thus $\lambda = \frac{1}{\mu}$ . + +\- Variance: We have $\operatorname{Var}(X) = \frac{1}{\lambda^2}$ . + +\- How to interpret $X$ + +- $X =$ The amount of time until some specific event occurs. +- Example: + +\* Time until earthquake occurs +\* Length of a phone call +\* Time until an accident happens + +\- Example1: Suppose that the length of a phone call in minutes is an exponential r.v with average length 10 minutes. + +\- Part (a) What's probability of your phone call being more than 10 minutes? + +\* Answer: Here $\lambda = \frac{1}{10}$ thus + +$$ +\mathbb {P} (X > 1 0) = e ^ {- \left(\frac {1}{1 0}\right) 1 0} = e ^ {- 1} \approx . 3 6 8. +$$ + +\- Part (b) Between 10 and 20 minutes? + +\* Answer: We have that + +$$ +\mathbb {P} (1 0 < X < 2 0) = F (2 0) - F (1 0) = e ^ {- 1} - e ^ {- 2} \approx . 2 3 3. +$$ + +- Exponential distribution is Memoryless (Markov) +- Example2: Suppose the life of an iphone has exponential distribution with mean life of 4 years. + +\- Part(a): What is the probability the phone lasts more than 5 years? + +\- $\underline{\text{Answer:}}$ Let $X$ denote the life of an iphone (or time until it dies). Note that $X \sim \overline{\text{exponential}}\left(\frac{1}{4}\right)$ since $\lambda = \frac{1}{\mu} = \frac{1}{4}$ . Then + +$$ +\mathbb {P} (X > 5) = e ^ {- \frac {1}{4} \cdot 5}. +$$ + +\- Part(b): Given that the iphone has already lasted 3 years, what is the probability that it will last another 5 more years? + +\- Answer: We compute + +$$ +\begin{array}{l} \mathbb {P} (X > 5 + 3 \mid X > 3) = \frac {\mathbb {P} ((X > 8) \cap (X > 3))}{\mathbb {P} (X > 3)} \\ = \frac {\mathbb {P} (X > 8)}{\mathbb {P} (X > 3)} \\ = \frac {e ^ {- \frac {1}{4} \cdot 8}}{e ^ {- \frac {1}{4} \cdot 3}} \\ = e ^ {- \frac {1}{4} \cdot 5}. \\ \end{array} +$$ + +\- Memoryless: Note that the probability of lasting 5 more years, is the same as if it started 5 years from anew!! + +\- In general the memoryless property says that if $t, s > 0$ then + +$$ +\mathbb {P} (X > t + s \mid X > t) = \mathbb {P} (X > s). +$$ + +THEOREM 14. If $X$ is an exponential random variable, then $X$ is memoryless. + +PROOF. To show this we have + +$$ +\begin{array}{l} \mathbb {P} (X > t + s \mid X > t) = \frac {\mathbb {P} ((X > t + s) \cap (X > t))}{\mathbb {P} (X > t)} \\ = \frac {\mathbb {P} (X > t + s)}{\mathbb {P} (X > t)} \\ = \frac {e ^ {- \lambda (t + s)}}{e ^ {- \lambda t}} \\ = e ^ {- \lambda s} \\ = \mathbb {P} (X > s), \\ \end{array} +$$ + +as needed. + +![](images/a7a320d4d8322e285795673461440b6b31ca85f35738acc41b6b9f5563f054b2.jpg) + +• Example3:(Exam P Q29) + +- The $\#$ of days from beginning of a calendar year until accident for a BAD driver is exponentially distributed +- An insurance company expects $30\%$ of BAD drivers will have an accident during first 50 days. +- Q: Whats prob that a BAD driver will have Accident during first 80 days? +- Answer: +- Step1: Let $X \sim \exp(\lambda)$ number of days until accident. We know + +$$ +. 3 = \mathbb {P} (X \leq 5 0) = \int_ {0} ^ {5 0} \lambda e ^ {- \lambda x} d x = - e ^ {- \lambda t} | _ {0} ^ {5 0} = 1 - e ^ {- 5 0 \lambda}. +$$ + +\- Solve for $\lambda$ and get $\lambda = -\frac{1}{50}\ln .7$ . + +\- Step2: Then compute + +$$ +\begin{array}{l} \mathbb {P} (X \leq 8 0) = \int_ {0} ^ {8 0} \lambda e ^ {- \lambda x} d x = - e ^ {- \lambda t} | _ {0} ^ {8 0} = 1 - e ^ {- 8 0 \lambda} \\ = 1 - e ^ {\left(\frac {8 0}{5 0}\right) \ln . 7} = . 4 3 5. \\ \end{array} +$$ + +## 10.2. Other Continuous Distributions + +## • Gamma Distribution: + +\- We say $X \sim Gamma(\alpha, \lambda)$ has density + +$$ +f (x) = \left\{ \begin{array}{l l} \frac {\lambda e ^ {- \lambda x} (\lambda x) ^ {\alpha - 1}}{\Gamma (\alpha)} & x \geq 0 \\ 0 & x < 0 \end{array} \right. +$$ + +where $\Gamma (\alpha)$ is the Gamma function + +$$ +\Gamma (\alpha) = \int_ {0} ^ {\infty} e ^ {- y} y ^ {\alpha - 1} d y. +$$ + +- If $Y \sim Gamma\left(\frac{n}{2}, \frac{1}{2}\right) = \chi_n^2$ , this is called the Chi-Squared distribution. +- The chi-square distribution is used a lot in statistics. + +\* Its mean is $\mathbb{E}X = \frac{\alpha}{\lambda}$ and $\operatorname{Var}X = \frac{\alpha}{\lambda^2}$ . + +## • Weibull Distribution: + +- Usefull in engineering: Look in the book for its pdf. +- $X =$ . If there is an object consisting many parts, and suppose that the object experiences death once any of its parts fails. $X =$ lifetime of this object. + +## • Cauchy Distribution: + +\- We say $X$ is cauchy with parameter $-\infty < \theta < \infty$ if + +$$ +f (x) = \frac {1}{\pi} \frac {1}{1 + (x - \theta) ^ {2}}. +$$ + +- Importance: It does not have finite mean: That is $\mathbb{E}X = \infty$ . +- To see this, We compute for $\theta = 0$ + +$$ +\mathbb {E} X = \frac {1}{\pi} \int_ {- \infty} ^ {\infty} \frac {x}{1 + x ^ {2}} d x +$$ + +$$ +\sim \frac {1}{\pi} \int_ {- \infty} ^ {\infty} \frac {1}{x} d x +$$ + +$$ +\sim \lim _ {x \to \infty} \ln | x | - \ln \lim _ {x \to - \infty} | x | +$$ + +which is not defined. + +## 10.3. The distribution function of a Random variable + +\- Fact: For continuous R.V we have the following usefull relationship + +\- Since $F(x) = \mathbb{P}(X \leq x) = \int_{-\infty}^{x} f(y) dy$ then by the fundament at theorem of calculus we have + +$$ +F ^ {\prime} (x) = f (x). +$$ + +\- Example1: If $X$ is continuous with distribution function $F_{X}$ and density function $f_{X}$ , find a formula for the density function of the random variable $Y = 2X$ . + +\- Solution: First you start with the distribution of $Y$ : + +\- Step1: First start by writing the cdf of $Y$ and in terms of $F_{X}$ : + +$$ +F _ {Y} (x) = \mathbb {P} (Y \leq x) +$$ + +$$ += \mathbb {P} (2 X \leq x) +$$ + +$$ += \mathbb {P} \left(X \leq \frac {x}{2}\right) +$$ + +$$ += F _ {X} \left(\frac {x}{2}\right). +$$ + +\- Step2: Then use the relation $f_{Y}(y) = F_{Y}'(y)$ and take a derivative of both sides to get + +$$ +F _ {Y} ^ {\prime} (x) = \frac {d}{d x} \left[ F _ {X} \left(\frac {x}{2}\right) \right], +$$ + +$$ +F _ {Y} ^ {\prime} (x) = F _ {X} ^ {\prime} \left(\frac {x}{2}\right) \cdot \left(\frac {x}{2}\right) ^ {\prime}, \text { by chain rule on RHS } +$$ + +$$ +f _ {Y} (x) = f _ {X} \left(\frac {x}{2}\right) \frac {1}{2}. +$$ + +\- Goal: To be able to compute the cdf and pdf of $Y = g(X)$ where $g: \mathbb{R} \to \mathbb{R}$ is a function given that we know the cdf and pdf of $X$ . + +\- Why is this useful? + +\* For example suppose $X$ represent the income for a random US worker. And let $Y = g(X)$ be the amount of taxes a US worker pays per year. Note that taxes $Y$ is dependent on the random variable $X$ . So if we only care about the random variable $Y$ then finding its PDF and CDF can help us find out everything we need to know about $Y$ given we can find the PDF. Recall that any probability and expected value can be found using the pdf. + +\- Example2: Let $X \sim \text{Uniform}((0,10))$ and $Y = e^{3X}$ . Find the pdf $f_{Y}$ of $Y$ . + +\- Solution: Recall that since $X \sim \text{Uniform}((0,1))$ then + +$$ +f _ {X} (x) = \left\{ \begin{array}{l l} \frac {1}{1 0} & 0 < x < 1 0 \\ 0 & \text {otherwise} \end{array} \right.. +$$ + +\- Step1: First start by writing the cdf of $Y$ and in terms of $F_{X}$ : + +$$ +\begin{array}{l} F _ {Y} (y) = \mathbb {P} (Y \leq y) \\ = \mathbb {P} \left(e ^ {3 X} \leq y\right), \text {then solve for } X \\ = \mathbb {P} (3 X \leq \ln y) \\ = \mathbb {P} \left(X \leq \frac {1}{3} \ln y\right) \\ = F _ {X} \left(\frac {1}{3} \ln y\right). \\ \end{array} +$$ + +\- Step2: Then use the relation $f_{Y}(y) = F_{Y}'(y)$ and take a derivative + +$$ +f _ {Y} (y) = F _ {Y} ^ {\prime} (y) +$$ + +$$ += \frac {d}{d y} \left[ F _ {X} \left(\frac {1}{3} \ln y\right) \right], \text {use chain rule} +$$ + +$$ += F _ {X} ^ {\prime} \left(\frac {1}{3} \ln y\right) \frac {1}{3 y} +$$ + +$$ += f _ {X} \left(\frac {1}{3} \ln y\right) \frac {1}{3 y}, \text { since } F _ {X} ^ {\prime} = f _ {X} +$$ + +$$ += \left\{ \begin{array}{l l} \frac {1}{1 0} \cdot \frac {1}{3 y} & 0 < \frac {1}{3} \ln y < 1 0 \\ 0 & \text { otherwise } \end{array} \right. +$$ + +\- but since + +$$ +0 < \frac {1}{3} \ln y < 1 0 \quad \Longleftrightarrow \quad 0 < \ln y < 3 0 +$$ + +$$ +\Longleftrightarrow e ^ {0} < y < e ^ {3 0} +$$ + +$$ +\Longleftrightarrow 1 < y < e ^ {3 0}. +$$ + +\- then + +$$ +f _ {Y} (y) = \left\{ \begin{array}{l l} \frac {1}{3 0 y} & 1 < y < e ^ {3 0} \\ 0 & \text {otherwise} \end{array} \right.. +$$ + +\- Example3: Let $X \sim \text{Uniform}((0,1])$ and $Y = -\ln X$ . Find the pdf of $Y$ ? What distribution is it? + +\- Solution: Recall that + +$$ +f _ {X} (x) = \left\{ \begin{array}{l l} 1 & 0 < x < 1 \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +\- Step1: First start with the cdf and write it terms of $F_{X}$ + +$$ +F _ {Y} (x) = \mathbb {P} (Y \leq x) +$$ + +$$ += \mathbb {P} (- \ln X \leq x) +$$ + +$$ += \mathbb {P} (\ln X > - x) +$$ + +$$ += \mathbb {P} (X > e ^ {- x}) +$$ + +$$ += 1 - \mathbb {P} (X \leq e ^ {- x}) +$$ + +$$ += 1 - F _ {X} \left(e ^ {- x}\right). +$$ + +\- Step2: Then take a derivative + +$$ +\begin{array}{l} f _ {Y} (x) = F _ {Y} ^ {\prime} (x) \\ = 1 - \frac {d}{d x} F _ {X} \left(e ^ {- x}\right) \\ = - F _ {X} ^ {\prime} \left(e ^ {- x}\right) \cdot \left(- e ^ {- x}\right) \\ = - f _ {X} \left(e ^ {- x}\right) \cdot \left(- e ^ {- x}\right) \\ = f _ {X} \left(e ^ {- x}\right) \cdot e ^ {- x} \\ = \left\{ \begin{array}{l l} 1 \cdot e ^ {- x} & 0 < e ^ {- x} < 1 \\ 0 & \text { otherwise } \end{array} \right. \\ = \left\{ \begin{array}{l l} e ^ {- x} & - \infty < - x < 0 \\ 0 & \text { otherwise } \end{array} \right. \\ = \left\{ \begin{array}{l l} e ^ {- x} & 0 < x < \infty \\ 0 & \text { otherwise } \end{array} \right. \\ \end{array} +$$ + +\- Thus $Y \sim$ exponential (1). + +\- Example4: Suppose $X$ is uniform on $\left(-\frac{\pi}{2}, \frac{\pi}{2}\right)$ and $Y = \tan X$ . Find the density of $Y$ and what known distribution is it? + +\- Solution: + +\- Step1: Find the cdf and write in terms of $F_{X}$ + +$$ +\begin{array}{l} F _ {Y} (x) = \mathbb {P} (\tan X \leq x) \\ = \mathbb {P} (X \leq \tan^ {- 1} x) \\ = F _ {X} (\tan^ {- 1} x) \\ \end{array} +$$ + +\- Step2: Take a derivative and recall that since $\frac{1}{\frac{\pi}{2} + \frac{\pi}{2}} = \frac{1}{\pi}$ then + +$$ +f _ {X} (x) = \left\{ \begin{array}{l l} \frac {1}{\pi} & - \frac {\pi}{2} < x < \frac {\pi}{2} \\ 0 & \text { otherwise. } \end{array} \right. +$$ + +Thus + +$$ +\begin{array}{l} f _ {Y} (x) = F _ {Y} ^ {\prime} (x) \\ = \frac {d}{d x} F _ {X} (\tan^ {- 1} x) \\ = F _ {X} ^ {\prime} \left(\tan^ {- 1} x\right) \left(\tan^ {- 1} x\right) ^ {\prime} \\ = F _ {X} ^ {\prime} \left(\tan^ {- 1} x\right) \frac {1}{1 + x ^ {2}} \\ = \left\{ \begin{array}{l l} \frac {1}{\pi} \frac {1}{1 + x ^ {2}} & - \frac {\pi}{2} < \tan^ {- 1} x < \frac {\pi}{2} \\ 0 & \text { otherwise. } \end{array} \right. \\ = \left\{ \begin{array}{l l} \frac {1}{\pi} \frac {1}{1 + x ^ {2}} & - \infty < x < \infty \\ 0 & \text { otherwise. } \end{array} \right. \\ \end{array} +$$ + +\- Thus $Y$ is Cauchy(0). + +- Exercise: Show that if $Z \sim \mathcal{N}(0,1)$ then $Y = Z^2$ is a Gamma with parameter $\frac{1}{2}$ and $\frac{1}{2}$ . +- Example5:(Actuarial Exam type question) The time, $T$ , that a manufacturing system is out of operation has cumulative distribution function + +$$ +F (t) = \left\{ \begin{array}{l l} 1 - \left(\frac {2}{t}\right) ^ {2} & , t > 2 \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +The resulting cost to the company is $Y = T^2$ . Let $f_{Y}$ be the density function for $Y$ . Determine $f_{Y}(y)$ , for $y > 4$ . + +\- Answer: + +\- Step1: Find the cdf of $Y$ is and + +$$ +\begin{array}{l} F _ {Y} (y) = \mathbb {P} \left(T ^ {2} \leq y\right) \\ = \mathbb {P} (T \leq \sqrt {y}) \\ = F (\sqrt {y}) \\ = 1 - \frac {4}{y} \\ \end{array} +$$ + +for $y > 4$ . + +\- Step2: Take a derivative + +$$ +\begin{array}{l} f _ {Y} (y) = F _ {Y} ^ {\prime} (y) \\ = \frac {4}{y ^ {2}}. \\ \end{array} +$$ + +\- One thing to note, is that we’ve been using the following useful property: + +PROPOSITION 15. Suppose $g: \mathbb{R} \to \mathbb{R}$ is a strictly increasing function, then the inverse $g^{-1}$ exists and + +$$ +g (x) \leq y \text { implies } x \leq g ^ {- 1} (y). +$$ + +## CHAPTER 11 + +# Multivariate distributions + +## 11.1. Joint distribution functions + +- We discuss the collection of random variables $(X_{1},\ldots ,X_{n})$ . +- Discrete: + +\- For random variables $X, Y$ we let $p(x, y)$ be the joint probability mass(discrete density) function + +$$ +p (x, y) = \mathbb {P} (X = x, Y = y). +$$ + +\* Properties of joint pmf: + +· 1) $0 \leq p \leq 1$ +2) $\sum_{i}\sum_{j}p(x_{i},y_{j}) = 1$ + +\- We also have the multivariate cdf: (★★) defined by + +$$ +F _ {X, Y} (x, y) = \mathbb {P} (X \leq x, Y \leq y). +$$ + +\- Example1: Experiment: Suppose you roll two 3-sided die. + +- Let $X$ be the largest value obtained on any of the two dice. Let $Y$ be the sum of the two dice. Find the joint pmf of $X$ and $Y$ . +- Solution: First need to find the values of $X = 1,2,3$ and $Y = 2,3,4,5,6$ . +- The table for possible outcomes and their associated values $(X,Y)$ : + +
    outcome123
    1 $(X = 1, Y = 2) = (1, 2)$ $(2, 3)$ $(3, 4)$
    2 $(2, 3)$ $(2, 4)$ $(3, 5)$
    3 $(3, 4)$ $(3, 5)$ $(3, 6)$
    + +\- Using this table we have that the p.m.f. is given by: + +
    X\Y23456
    1 $\mathbb{P}(X=1,Y=2)=\frac{1}{9}$ 0000
    20 $\frac{2}{9}$ $\frac{1}{9}$ 00
    300 $\frac{2}{9}$ $\frac{2}{9}$ $\frac{1}{9}$
    + +\- Question: Find $\mathbb{P}(X = 2\mid Y = 4)$ ? +\* Answer: $\mathbb{P}(X = 2\mid Y = 4) = \frac{1 / 9}{3 / 9} = \frac{1}{3}.$ + +\- Continuous + +\- For random variables $X, Y$ we let $f(x, y)$ be the joint probability density function, if + +$$ +\mathbb {P} (a \leq X \leq b, c \leq Y \leq d) = \int_ {a} ^ {b} \int_ {c} ^ {d} f (x, y) d y d x. +$$ + +which is equivalent to saying that for any set $D \subset \mathbb{R}^2$ then + +$$ +\mathbb {P} \left((X, Y) \in D\right) = \int \int_ {D} f (x, y) d A. +$$ + +\* Properties: + +· 1) $f(x, y) \geq 0$ +· 2) $\int_{-\infty}^{\infty}\int_{-\infty}^{\infty}f(x,y)dxdy=1.$ + +\- We also have the multivariate cdf: (★★) defined by + +$$ +F _ {X, Y} (x, y) = \mathbb {P} (X \leq x, Y \leq y). +$$ + +\* Note that $F_{X,Y}(a,b) = \int_{-\infty}^{a}\int_{-\infty}^{b}f(x,y)dydx$ . + +\- Thus note that + +$$ +f (x, y) = \frac {\partial^ {2} F (x , y)}{\partial x \partial y}. +$$ + +\- Marginal Density: If $f_{X,Y}$ is the joint density of $X, Y$ . We recover the marginal densities of $X, Y$ respectively by the following + +$$ +f _ {X} (x) = \int_ {- \infty} ^ {\infty} f _ {X, Y} (x, y) d y, +$$ + +$$ +f _ {Y} (y) = \int_ {- \infty} ^ {\infty} f _ {X, Y} (x, y) d x. +$$ + +• Example2: Let X, Y have joint pdf + +$$ +f (x, y) = \left\{ \begin{array}{l l} c e ^ {- x} e ^ {- 2 y} & , 0 < x < \infty , 0 < y < \infty \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +\- $\underline{\mathbf{Part}(\mathbf{a})}$ : Find $c$ that makes this a joint pdf: + +\* Sol: Step1: Draw region of Domain first!!! + +![](images/071c3f64aeb248121917c187cea61972b8ea8b96e39a431e53bfe0ab1633c1bf.jpg) + +
    +natural_image + +Blank Cartesian coordinate grid with no plotted data or labels +
    + +\* Thus + +$$ +\begin{array}{l} 1 = \int_ {0} ^ {\infty} \int_ {0} ^ {\infty} c e ^ {- x} e ^ {- 2 y} d x d y = c \int_ {0} ^ {\infty} e ^ {- 2 y} \left[ - e ^ {- x} \right] _ {x = 0} ^ {x = \infty} d y \\ = c \int_ {0} ^ {\infty} e ^ {- 2 y} d y = c \left[ - \frac {1}{2} e ^ {- 2 y} \right] _ {0} ^ {\infty} = c \frac {1}{2}. \\ \end{array} +$$ + +Then c = 2. + +\- $\underline{\text{Part(b)}}$ : Find $\mathbb{P}(X < Y)$ . + +\* Sol: Need to draw the region (Recall Calc III!!) Let $D = \{(x,y) \mid 0 < x < y, 0 < y < \infty\}$ + +![](images/aacef5faf74b558104e4f0faa7e4fb79374d3d04106ab6ba63e846f5eb6c3982.jpg) + +
    +area chart + +| x | y | +| ---- | ---- | +| 0.0 | 0.0 | +| 0.2 | 0.2 | +| 0.4 | 0.4 | +| 0.6 | 0.6 | +| 0.8 | 0.8 | +| 1.0 | 1.0 | +
    + +- There are two ways to set up this integral: +- Method1: To set up $dA = dydx$ . We use the Top-Bottom Method: +- Where the region is bounded by + +Top Function: $y = \infty$ + +Bottom Function $y = x$ + +Range of Values $0 \leq x \leq \infty$ + +\- Hence we use this information to set up + +$$ +\begin{array}{l} \mathbb {P} (X < Y) = \int \int_ {D} f (x, y) d A \\ = \int_ {0} ^ {\infty} \int_ {x} ^ {\infty} 2 e ^ {- x} e ^ {- 2 y} d y d x \\ = \int_ {0} ^ {\infty} 2 e ^ {- x} \frac {1}{2} \left[ - e ^ {- 2 y} \right] _ {y = x} ^ {y = \infty} d x \\ = \int_ {0} ^ {\infty} e ^ {- x} e ^ {- 2 x} d x = \int_ {0} ^ {\infty} e ^ {- 3 x} x \\ = \frac {1}{3}. \\ \end{array} +$$ + +\- Method2: To set up $dA = dxdy$ . We use the Right-Left Method: + +\- Where the region is bounded by + +Right Function: $x = y$ + +Left Functionx = 0 + +Range of Values $0 \leq y \leq \infty$ + +\- Hence we use this information to set up + +$$ +\begin{array}{l} \mathbb {P} (X < Y) = \int \int_ {D} f (x, y) d A \\ = \int_ {0} ^ {\infty} \int_ {0} ^ {y} 2 e ^ {- x} e ^ {- 2 y} d x d y \\ = \quad \text { do some work } \\ = \frac {1}{3}, \\ \end{array} +$$ + +which matches the answer from before. + +\- $\underline{\mathbf{Part}(\mathbf{c})}$ : Set up $\mathbb{P}(X > 1, Y < 1)$ + +![](images/b9bbef76d2bbbb33fea36f6544c0e7ff5ef26a71ca7a7984369a1b87a5af6582.jpg) + +
    +bar chart + +| X | Y | +|---|---| +| 1 | 1 | +
    + +\* The region is given by +\* Setting this up we have + +$$ +\mathbb {P} (X > 1, Y < 1) = \int_ {0} ^ {1} \int_ {1} ^ {\infty} 2 e ^ {- x} e ^ {- 2 y} d x d y. +$$ + +\- $\underline{\mathbf{Part}(\mathbf{d})}$ : Find the marginal $f_{X}(x)$ : + +\* Sol: +\* Then + +$$ +\begin{array}{l} f _ {X} (x) = \int_ {0} ^ {\infty} f (x, y) d y = \int_ {0} ^ {\infty} 2 e ^ {- x} e ^ {- 2 y} d y \\ = 2 e ^ {- x} \left[ \frac {- 1}{2} e ^ {- 2 y} \right] _ {0} ^ {\infty} = 2 e ^ {- x} \left[ 0 + \frac {1}{2} \right] \\ = e ^ {- x}. \\ \end{array} +$$ + +\- Part(e): Find $\mathbb{E}X$ We have + +$$ +\mathbb {E} X = \int_ {0} ^ {\infty} x e ^ {- x} d x = 1 +$$ + +## 11.2. Independent Random Variables + +\- Discrete: We say discrete r.v. $X, Y$ are independent if + +$$ +\mathbb {P} (X = x, Y = y) = \mathbb {P} (X = x) \mathbb {P} (Y = y), +$$ + +for every $x, y$ in the range of $X$ and $Y$ . + +\- This is the same as saying that $X, Y$ are independent if the joint pmf splits into the marginal pmfs: $p_{X,Y}(x,y) = p_X(x) \cdot p_Y(y)$ + +\- Continuous: We say continuous r.v. $X, Y$ are independent if + +$$ +\mathbb {P} (X \in A, Y \in B) = \mathbb {P} (X \in A) \mathbb {P} (Y \in B) +$$ + +for any set $A, B$ + +\- This equivalent: $\mathbb{P}(X\leq a,Y\leq b) = \mathbb{P}(X\leq a)\mathbb{P}(Y\leq b)$ . + +\- Equivalent to $F_{X,Y}(x,y) = F_X(x)F_Y(y)$ . + +\- Random variables that are not independent, are said to be dependent. + +• How can we check independence? + +THEOREM 16. Continuous (discrete) r.v. $X, Y$ are independent if and only if their joint pdf (pmf) can be expressed as + +$$ +f _ {X, Y} (x, y) = f _ {X} (x) f _ {Y} (y). \text {(Continuous Case)}, +$$ + +$$ +p _ {X, Y} (x, y) = p _ {X} (x) p _ {Y} (y) \text {(Discrete Case)}. +$$ + +PROOF. See textbook. + +![](images/a30768ec5d0cef10efe409d93934fd345af55ca29e0ff4aaaf38416b8964f267.jpg) + +\- Example1: Let $X, Y$ be r.v. with joint pdf + +$$ +f (x, y) = 6 e ^ {- 2 x} e ^ {- 3 y} 0 < x < \infty , 0 < y < \infty . +$$ + +Are $X, Y$ independent? + +\- Solution: Find the marginals $f_{X}$ and $f_{Y}$ and see if $f = f_{X}f_{Y}$ . First + +$$ +f _ {X} (x) = \int_ {0} ^ {\infty} 6 e ^ {- 2 x} e ^ {- 3 y} d y = 2 e ^ {- 2 x}, +$$ + +$$ +f _ {Y} (y) = \int_ {0} ^ {\infty} 6 e ^ {- 2 x} e ^ {- 3 y} d x = 3 e ^ {- 2 y}. +$$ + +which are both exponential. Since $f = f_{X}f_{y}$ then yes they are independent! + +• Example2: Let X, Y have + +$$ +f _ {X, Y} (x, y) = x + y, 0 < x < 1, 0 < y < 1 +$$ + +Are $X, Y$ independent? + +\- Solution: Note that there is no way to factor $x + y = f_X(x)f_Y(y)$ , hence they can't be independent. + +• Example3: Let X, Y have + +$$ +f _ {X, Y} (x, y) = 2, 0 < x < y < 1 +$$ + +\- Are $X, Y$ independent? + +\- Solution: + +\* We cannot use the previous argument to claim $f_{X,Y}$ can't split, because for example, maybe hypothetically speaking $2 = 1 \cdot 1$ , so hypothetically it could split. +\* So we must find the marginal pdfs and then check if $f_{X,Y} = f_X \cdot f_Y$ . + +\* Important! But whenever the domain of $f$ is not a rectangle, you MUST draw the region of domain for $f_{X,Y}$ . And here the region is $D = \{(x,y) \mid 0 < x < y < 1\}$ . (Please try drawing this region on your own. If you struggle with this region, go to https://www.wolframalpha.com/ and type in $0 < x < y < 1$ ) + +\* Note that $f_{X}(x) = \int_{x}^{1} 2dy = 2(1 - x)$ for $0 < x < 1$ +\* Then $f_{Y}(y) = \int_{0}^{y} 2dx = 2y$ for $0 < y < 1$ . +\* But $f_{X,Y}(x,y) = 2 \neq f_X(x)f_Y(y) = 2(1 - x)2y!!$ Therefore $X,Y$ are NOT independent. + +\- Example4: Suppose $X, Y$ are independent uniformly distributed over (0, 1). Find $\mathbb{P}(Y < X)$ . + +\- Solution: Since $X, Y$ are independent then using the Theorem form this section we have + +$$ +f _ {X, Y} (x, y) = f _ {X} (x) f _ {Y} (y) = 1 \cdot 1, +$$ + +for $0 \leq x \leq 1$ and $0 \leq y \leq 1$ . Draw region ('What do you think probability will be by looking at the region?) + +![](images/93acb122105318bb84c87f86bfb3d45f21cba400a8bd06fcc38be8144657aaea.jpg) + +
    +area chart + +| x | y | +|---|---| +| 0 | 0 | +| 0.2 | 0.2 | +| 0.4 | 0.4 | +| 0.6 | 0.6 | +| 0.8 | 0.8 | +| 1 | 1 | +
    + +\* and get + +$$ +\begin{array}{l} \mathbb {P} (Y < X) = \int_ {0} ^ {1} \int_ {0} ^ {x} f (x, y) d y d x \\ = \int_ {0} ^ {1} \int_ {0} ^ {x} 1 d y d x = \int_ {0} ^ {1} x d x \\ = \frac {1}{2}. \\ \end{array} +$$ + +## 11.3. Sums of independent Random Variables( $\star$ ) + +\- Fact: If $X, Y$ are independent, its not too hard to show that the cdf of $Z = X + Y$ is + +$$ +\begin{array}{l} F _ {X + Y} (a) = \mathbb {P} (X + Y \leq a) \\ = \int \int_ {\{x + y \leq a \}} f _ {X} (x) f _ {Y} (y) d x d y \\ = \int_ {- \infty} ^ {\infty} \int_ {- \infty} ^ {a - y} f _ {X} (x) f _ {Y} (y) d x d y \\ = \int_ {- \infty} ^ {\infty} \int_ {- \infty} ^ {a - y} f _ {X} (x) d x f _ {Y} (y) d y \\ = \int_ {- \infty} ^ {\infty} F _ {X} (a - y) f _ {Y} (y) d y. \\ \end{array} +$$ + +\- By differentiating we have that + +$$ +f _ {X + Y} (a) = \int_ {- \infty} ^ {\infty} f _ {X} (a - y) f _ {Y} (y) d y. +$$ + +• (★)Here are some interesting cases: + +\- Fact 1(Only thing I'll test you on): If $X_{i} \sim \mathcal{N}\left(\mu_{i}, \sigma_{i}^{2}\right)$ for $1 \leq i \leq n$ and are all independent then $Y = X_{1} + \dots + X_{n} \sim \mathcal{N}\left(\mu_{1} + \dots \mu_{n}, \sigma_{1}^{2} + \dots + \sigma_{n}^{2}\right)$ . + +- In particular if $X \sim \mathcal{N}\left(\mu_x, \sigma_x^2\right)$ and $Y \sim \mathcal{N}(\mu_y, \sigma_y^2)$ are independent then $X + Y \sim \mathcal{N}\left(\mu_x + \mu_y, \sigma_x^2 + \sigma_y^2\right)$ and $X - Y \sim \mathcal{N}\left(\mu_x - \mu_y, \sigma_x^2 + \sigma_y^2\right)$ . +- In general $aX \pm bY \sim \mathcal{N}\left(a\mu_x \pm b\mu_y, a^2\sigma_x^2 + b^2\sigma_y^2\right)$ . + +\- Example1: Suppose $T \sim \mathcal{N}(95,25)$ and $H \sim \mathcal{N}(65,36)$ represents the grades of Tyler and Habib. Assume their grades are independent. + +\- Part(a): What is the probability that their average grades will be less than 90? + +\- Solution: $T + H \sim \mathcal{N}(160, 61)$ . Thus + +$$ +\begin{array}{l} \mathbb {P} \left(\frac {T + H}{2} \leq 9 0\right) = \mathbb {P} (T + H \leq 1 8 0) \\ = \mathbb {P} \left(Z \leq \frac {1 8 0 - 1 6 0}{\sqrt {6 1}}\right) = \Phi \left(\frac {1 8 0 - 1 6 0}{\sqrt {6 1}}\right) \\ = \Phi (2. 5 6) = . 9 9 6 1 \\ \end{array} +$$ + +\- Part (b): What is the probability that Habib will have scored higher than Tyler? + +\- Solution: Using $H - T \sim \mathcal{N}(-30,61)$ we compute + +$$ +\begin{array}{l} \mathbb {P} (H > T) = \mathbb {P} (H - T > 0) \\ = 1 - \mathbb {P} (H - T < 0) \\ = 1 - \mathbb {P} \left(Z \leq \frac {0 - (- 3 0)}{\sqrt {6 1}}\right) \\ = 1 - \Phi (3. 8 4) = 1 - 1 = 0. \\ \end{array} +$$ + +\- Other facts. + +\- Fact 2: Let $Z \sim \mathcal{N}(0,1)$ then $Z^2 \sim \chi_1^2$ . + - If $Z_1, \ldots, Z_n$ are independent $\mathcal{N}(0,1)$ then $Y = Z_1^2 + \cdots + Z_n^2 \sim \chi_n^2$ . + +- Fact 3: If $X \sim \text{Poisson}(\lambda)$ and $Y \sim \text{Poisson}(\mu)$ , and they are independent, then $X + Y \sim \overline{\text{Poisson}}(\lambda + \mu)$ . +- List out stuff and then stop. + +## 11.4. Conditional Distributions- Discrete(★) + +\- The conditional pmf for a discrete R.V. is + +$$ +p _ {X \mid Y} (x \mid y) = \mathbb {P} (X = x \mid Y = y) +$$ + +$$ += \frac {p (x , y)}{p _ {Y} (y)}. +$$ + +\- We also have the condition cdf: $F_{X|Y}(x \mid y) = \mathbb{P}(X \leq x \mid Y = u)$ + +\- Fact: + +\- If $X, Y$ are independent then + +$$ +p _ {X | Y} (x \mid y) = p _ {X} (x) +$$ + +\- Example1: Suppose the joint pmf of $(X,Y)$ is + +
    x\y01
    0.4.2
    1.1.3
    + +\- Compute some conditional pmf: Then the second column is + +$$ +p _ {X | Y} (0 \mid 1) = \frac {. 2}{. 5} = \frac {2}{5} \text { and } p _ {X | Y} (1 \mid 1) = \frac {. 3}{. 5} = \frac {3}{5}. +$$ + +\- Are they independent? Note that $p_X(0) = .4 + .2 = .6 \neq p_{X|Y}(0 \mid 1)$ , so no! + +## 11.5. Conditional Distributions- Continuous( $\star$ ) + +\- Def: If $X, Y$ are continuous with joint pdf $f(x, y)$ then the conditional pdf of $X$ given $Y = y$ is defined as + +$$ +f _ {X \mid Y} (x \mid y) = \frac {f (x , y)}{f _ {Y} (y)}. +$$ + +\- defined only when $f_{Y}(y) > 0$ . + +\- Def: The conditional cdf of $X$ given $Y = y$ is + +$$ +\begin{array}{l} F _ {X \mid Y} (a \mid y) = \mathbb {P} (X \leq a \mid Y = y) \\ = \int_ {- \infty} ^ {a} f _ {X | Y} (x \mid y) d x. \\ \end{array} +$$ + +• Fact: If X, Y are independent then + +$$ +f _ {X \mid Y} (x \mid y) = f _ {X} (x). +$$ + +\- Example1: The joint pdf of $X, Y$ is given by + +$$ +f x (x, y) = \left\{ \begin{array}{l l} \frac {1 2}{5} x (2 - x - y) & 0 < x < 1, 0 < y < 1 \\ 0 & \text {otherwise} \end{array} \right.. +$$ + +Compute the conditional pdf of X give that Y = y where 0 < y < 1. + +\- Solution: We have + +$$ +\begin{array}{l} f _ {X \mid Y} (x \mid y) = \frac {f (x , y)}{f _ {Y} (y)} = \frac {x (2 - x - y)}{\int_ {0} ^ {1} x (2 - x - y) d y} \\ = \frac {x (2 - x - y)}{\frac {2}{3} - \frac {y}{2}}. \\ \end{array} +$$ + +## 11.6. Joint PDF of functions + +## - Goal: + +- Recall that from section 5.7 we can find the pdf of a new random variable $Y = g(X)$ . +- Suppose we know the distributions of $X_{1}, X_{2}$ then what is the distribution of $g_{1}(X_{1}, X_{2})$ and $g_{2}(X_{1}, Y_{1})$ +\* For example if we know $X_{1}, X_{2}$ what is the distribution of $Y_{1} = X_{1} + X_{2}$ and $Y_{2} = X_{1}^{2} - e^{X_{1}X_{2}}$ . + +\- Steps to finding the joint cdf of new R.V. made from old ones.: + +- Suppose $X_{1}, X_{2}$ are jointly distributed with pdf $f_{X_1, X_2}$ . Let $g_{2}(x_{1}, x_{2}), g_{2}(x_{2}, x_{2})$ be multivariable functions. +- Goal: Find the joint pdf of $Y_{1} = g_{1}(X_{1}, X_{2})$ and $Y_{2} = g_{1}(X_{2}, X_{2})$ +- Step1: Find the Jacobian: + +$$ +J \left(x _ {1}, x _ {2}\right) = \left| \begin{array}{c} \nabla g _ {1} \\ \nabla g _ {2} \end{array} \right| = \left| \begin{array}{c c} \frac {\partial g _ {1}}{\partial x _ {1}} & \frac {\partial g _ {1}}{\partial x _ {2}} \\ \frac {\partial g _ {2}}{\partial x _ {1}} & \frac {\partial g _ {2}}{\partial x _ {2}} \end{array} \right| = \frac {\partial g _ {1}}{\partial x _ {1}} \frac {\partial g _ {2}}{\partial x _ {2}} - \frac {\partial g _ {1}}{\partial x _ {2}} \frac {\partial g _ {2}}{\partial x _ {1}} \neq 0. +$$ + +at all points $(x_{1},x_{2})$ + +\- Step2: Find the unique solutions of equationf $y_{1} = g_{1}(x_{1}, x_{2})$ and $y_{2} = g_{2}(x_{1}, x_{2})$ in terms of + +$$ +x _ {1} = h _ {1} \left(y _ {1}, y _ {2}\right), +$$ + +$$ +x _ {2} = h _ {2} \left(y _ {1}, y _ {2}\right). +$$ + +\- Step3: The joint pdf of $Y_{1}, Y_{2}$ is + +$$ +\begin{array}{l} f _ {Y _ {1}, Y _ {2}} \left(y _ {1}, y _ {2}\right) = f _ {X _ {1}, X _ {2}} \left(x _ {1}, x _ {2}\right) | J \left(x _ {1}, x _ {2}\right) | ^ {- 1} \\ = f _ {X _ {1}, X _ {2}} \left(h _ {1} \left(y _ {1}, y _ {2}\right), h _ {2} \left(y _ {1}, y _ {2}\right)\right) | J \left(x _ {1}, x _ {2}\right) | ^ {- 1}. \\ \end{array} +$$ + +\- Example1: Suppose $X_{1}, X_{2}$ have joint distribution + +$$ +f _ {X _ {1}, X _ {2}} (x _ {1}, x _ {2}) = \left\{ \begin{array}{l l} 2 x _ {1} x _ {2} & 0 \leq x _ {1}, x _ {2} \leq 1 \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +Question: Find the joint pdf of $Y_{1} = X_{1} + X_{2}$ and $Y_{2} = X_{1} - X_{2}$ . + +\- Step1: Find the Jacobian: Note that + +$$ +y _ {1} = g _ {1} \left(x _ {1}, x _ {2}\right) = x _ {1} + x _ {2}, +$$ + +$$ +y _ {2} = g _ {2} \left(x _ {1}, x _ {2}\right) = x _ {1} - x _ {2}. +$$ + +So + +$$ +J \left(x _ {1}, x _ {2}\right) = \left| \begin{array}{c c} 1 & 1 \\ 1 & - 1 \end{array} \right| = - 2. +$$ + +\- Step2: Solve for $x_{1}, x_{2}$ and get + +$$ +x _ {1} = \frac {1}{2} \left(y _ {1} + y _ {2}\right), +$$ + +$$ +x _ {2} = \frac {1}{2} (y _ {1} - y _ {2}). +$$ + +\- Step3: The joint pdf of $Y_{1}, Y_{2}$ is given by the formula: + +$$ +\begin{array}{l} f _ {Y _ {1}, Y _ {2}} \left(y _ {1}, y _ {2}\right) = f _ {X _ {1}, X _ {2}} \left(x _ {1}, x _ {2}\right) | J \left(x _ {1}, x _ {2}\right) | ^ {- 1} \\ = f _ {X _ {1}, X _ {2}} \left(\frac {1}{2} \left(y _ {1} + y _ {2}\right), \frac {1}{2} \left(y _ {1} - y _ {2}\right)\right) \frac {1}{| - 2 |} \\ = \left\{ \begin{array}{l l} \frac {1}{2} \left(y _ {1} + y _ {2}\right) \left(y _ {1} - y _ {2}\right) & 0 \leq \frac {1}{2} \left(y _ {1} + y _ {2}\right) \leq 1, \\ & 0 \leq \frac {1}{2} \left(y _ {1} - y _ {2}\right) \leq 1 \\ 0 & \text {otherwise} \end{array} \right. \\ \end{array} +$$ + +\- Example2: Suppose $X_{1} \sim \mathcal{N}(0,1)$ and $X_{2} \sim \mathcal{N}(0,4)$ and independent. + +\- Let $Y_{1} = 2X_{1} + X_{2}$ and $Y_{2} = X_{1} - 3X_{2}$ . + +\- Question: Find the joint pdf $f_{Y_1, Y_2}(y_1, y_2)$ of $Y_1$ and $Y_2$ . + +\- Step1: Find the Jacobian: Note that + +$$ +y _ {1} = g _ {1} \left(x _ {1}, x _ {2}\right) = 2 x _ {1} + x _ {2}, +$$ + +$$ +y _ {2} = g _ {2} \left(x _ {1}, x _ {2}\right) = x _ {1} - 3 x _ {2}. +$$ + +So + +$$ +J \left(x _ {1}, x _ {2}\right) = \left| \begin{array}{c c} 2 & 1 \\ 1 & - 3 \end{array} \right| = - 7. +$$ + +\- Step2: Solve for $x_{1}, x_{2}$ and get + +$$ +x _ {1} = \frac {3}{7} y _ {1} + \frac {1}{y} y _ {2} +$$ + +$$ +x _ {2} = \frac {1}{7} y _ {1} - \frac {2}{7} y _ {2} +$$ + +\- Step3: The joint pdf of $Y_{1}, Y_{2}$ is given by the formula: + +$$ +\begin{array}{l} f _ {Y _ {1}, Y _ {2}} \left(y _ {1}, y _ {2}\right) = f _ {X _ {1}, X _ {2}} \left(x _ {1}, x _ {2}\right) \left| J \left(x _ {1}, x _ {2}\right) \right| ^ {- 1} \\ = f _ {X _ {1}, X _ {2}} \left(\frac {3}{7} y _ {1} + \frac {1}{y} y _ {2}, \frac {1}{7} y _ {1} - \frac {2}{7} y _ {2}\right) \frac {1}{7}. \\ \end{array} +$$ + +So we need to find the joint pdf of $X_{1}$ and $X_{2}$ . + +\* But since $X_{1} \sim \mathcal{N}(0,1)$ and $X_{2} \sim \mathcal{N}(0,4)$ and independent Then + +$$ +f _ {X _ {1}} (x _ {1}) = \frac {1}{\sqrt {2 \pi}} e ^ {- x ^ {2} / 2} \text { and } f _ {X _ {2}} (x _ {2}) = \frac {1}{\sqrt {2 \cdot 4 \pi}} e ^ {- x ^ {2} / (2 \cdot 4)}. +$$ + +Thus by impedance + +$$ +\begin{array}{l} f _ {X _ {1}, X _ {2}} (x _ {1}, x _ {2}) = f _ {X _ {1}} (x _ {1}) f _ {X _ {2}} (x _ {2}) \\ { = } { \frac { 1 } { \sqrt { 2 \pi } } e ^ { - x ^ { 2 } / 2 } \frac { 1 } { \sqrt { 2 \cdot 4 \pi } } e ^ { - x ^ { 2 } / ( 2 \cdot 4 ) } . } \\ \end{array} +$$ + +\* Thus we have + +$$ +f _ {Y _ {1}, Y _ {2}} \left(y _ {1}, y _ {2}\right) = \frac {1}{\sqrt {2 \pi}} e ^ {- \left(\frac {3}{7} y _ {1} + \frac {1}{y} y _ {2}\right) ^ {2} / 2} \frac {1}{\sqrt {8 \pi}} e ^ {- \left(\frac {1}{7} y _ {1} - \frac {2}{7} y _ {2}\right) ^ {2} / 8} \frac {1}{7}. +$$ + +• Example3(if time): Suppose $X_{1}, X_{2}$ have joint distribution + +$$ +f _ {X _ {1}, X _ {2}} (x _ {1}, x _ {2}) = \left\{ \begin{array}{l l} x _ {1} + \frac {3}{2} \left(x _ {2}\right) ^ {2} & 0 \leq x _ {1} \leq 1, 0 \leq x _ {2} \leq 1 \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +Question: Find the joint pdf of $Y_{1} = X_{1} + X_{2}$ and $Y_{2} = X_{1}^{2}$ . + +\- Step1: Find the Jacobian: Note that + +$$ +y _ {1} = g _ {1} \left(x _ {1}, x _ {2}\right) = x _ {1} + x _ {2}, +$$ + +$$ +y _ {2} = g _ {2} \left(x _ {1}, x _ {2}\right) = x _ {1} ^ {2}. +$$ + +So + +$$ +J \left(x _ {1}, x _ {2}\right) = \left| \begin{array}{c c} 1 & 2 \\ 2 x _ {1} & 0 \end{array} \right| = - 4 x _ {2} +$$ + +\- Step2: Solve for $x_{1}, x_{2}$ and get + +$$ +x _ {1} = \sqrt {y _ {2}}, +$$ + +$$ +x _ {2} = y _ {1} - \sqrt {y _ {2}}. +$$ + +\- Step3: The joint pdf of $Y_{1}, Y_{2}$ is given by the formula: + +$$ +\begin{array}{l} f _ {Y _ {1}, Y _ {2}} \left(y _ {1}, y _ {2}\right) = f _ {X _ {1}, X _ {2}} \left(x _ {1}, x _ {2}\right) \left| J \left(x _ {1}, x _ {2}\right) \right| ^ {- 1} \\ = f _ {X _ {1}, X _ {2}} \left(\sqrt {y _ {2}}, y _ {1} - \sqrt {y _ {2}}\right) \frac {1}{| 4 x _ {2} |} \\ = \left\{ \begin{array}{l l} \frac {1}{| 4 x _ {2} |} \left[ \sqrt {y _ {2}} + \frac {3}{2} \left(y _ {1} - \sqrt {y _ {2}}\right) ^ {2} \right] & 0 \leq \sqrt {y _ {2}} \leq 1, \\ & 0 \leq y _ {1} - \sqrt {y _ {2}} \leq 1 \\ 0 & \text {otherwise} \end{array} \right. \\ \end{array} +$$ + +## CHAPTER 12 + +## Expectations + +## 12.1. Expectation of Sums of R.V. + +THEOREM 17. Let $g: \mathbb{R}^2 \to \mathbb{R}$ . If $X, Y$ have joint pmf $p(x, y)$ then + +$$ +\mathbb {E} \left[ g (X, Y) \right] = \sum_ {y} \sum_ {x} g (x, y) p (x, y). +$$ + +If $X, Y$ have joint pdf $f(x, y)$ then + +$$ +\mathbb {E} \left[ g (X, Y) \right] = \int_ {- \infty} ^ {\infty} \int_ {- \infty} ^ {\infty} g (x, y) f (x, y) d x d y. +$$ + +\- Example1: Suppose the joint p.m.f of $X$ and $Y$ is given by $\begin{array}{c|cc}X\setminus Y & 0 & 2\\ \hline 0 & .2 & .7\\ \hline 1 & 0 & .1 \end{array}$ . Find $\mathbb{E}[XY]$ . + +\- Solution: Using the formula we have with the function $g(x,y) = xy$ : + +$$ +\begin{array}{l} \mathbb {E} [ X Y ] = \sum_ {i, j} x _ {i} y _ {j} p (x _ {i}, y _ {j}) \\ = 0 \cdot 0 p (0, 0) + 1 \cdot 0 p (1, 0) + 0 \cdot 2 p (0, 2) + 1 \cdot 2 p (1, 2) \\ = 0 \cdot 0 \cdot . 2 + 1 \cdot 0 \cdot 0 + 0 \cdot 2 \cdot . 7 + 1 \cdot 2 \cdot . 1 \\ = . 2 \\ \end{array} +$$ + +\- Example2: Suppose $X, Y$ are independent exponential r.v. with parameter $\lambda = 1$ . Set up a double integral that represents + +$$ +\mathbb {E} \left[ X ^ {2} Y \right]. +$$ + +\- Solution: Since $X, Y$ are independent then + +$$ +f _ {X, Y} (x, y) = e ^ {- 1 x} e ^ {- 1 y} = e ^ {- (x + y)}. \quad 0 < x, y < \infty . +$$ + +\- Then DRAW FIRST then + +$$ +\mathbb {E} \left[ X ^ {2} Y \right] = \int_ {0} ^ {\infty} \int_ {0} ^ {\infty} x ^ {2} y e ^ {- (x + y)} d y d x. +$$ + +• Example3: Suppose the joint pdf of X, Y is + +$$ +f (x, y) = \left\{ \begin{array}{l l} 1 0 x y ^ {2} & 0 < x < y, 0 < y < 1 \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +Find $\mathbb{E}XY$ and $\operatorname{Var}(Y)$ . + +\- Solution: + +\- We First DRAW and then set up + +$$ +\begin{array}{l} \mathbb {E} X Y = \int_ {0} ^ {1} \int_ {0} ^ {y} x y (1 0 x y ^ {2}) d x d y = 1 0 \int_ {0} ^ {1} \int_ {0} ^ {y} x ^ {2} y ^ {3} d x d y \\ = \frac {1 0}{3} \int_ {0} ^ {1} y ^ {3} y ^ {3} d y = \frac {1 0}{3} \frac {1}{7} = \frac {1 0}{2 1}. \\ \end{array} +$$ + +\- First note that $\operatorname{Var}(Y) = \mathbb{E}Y^2 - (\mathbb{E}Y)^2$ . + +\- Then + +$$ +\begin{array}{l} \mathbb {E} Y ^ {2} = \int_ {0} ^ {1} \int_ {0} ^ {y} y ^ {2} (1 0 x y ^ {2}) d x d y = 1 0 \int_ {0} ^ {1} \int_ {0} ^ {y} y ^ {4} x d x d y \\ = 5 \int_ {0} ^ {1} y ^ {4} y ^ {2} d y = \frac {5}{7}. \\ \end{array} +$$ + +and + +$$ +\begin{array}{l} \mathbb {E} Y = \int_ {0} ^ {1} \int_ {0} ^ {y} y (1 0 x y ^ {2}) d x d y = 1 0 \int_ {0} ^ {1} \int_ {0} ^ {y} y ^ {3} x d x d y \\ = 5 \int_ {0} ^ {1} y ^ {3} y ^ {2} d y = \frac {5}{6}. \\ \end{array} +$$ + +So that $\operatorname{Var}(Y) = \frac{5}{7} - \left(\frac{5}{6}\right)^2 = \frac{5}{252}$ . + +THEOREM 18. (Properties of Expectation) + +(a) $\mathbb{E}[X + Y] = \mathbb{E}X + \mathbb{E}Y$ +(b) If $X \leq Y$ then $\mathbb{E}X \leq \mathbb{E}Y$ . + +PROOF. Part (a) was proved for the discrete case. So we only need to show the continuous case: + +$$ +\begin{array}{l} \mathbb {E} \left[ X + Y \right] = \int \int (x + y) f (x, y) d y d x \\ = \int \int x f (x, y) d y d x + \int \int y f (x, y) d y d x \\ = \int x f _ {X} (x) d x + \int y f _ {Y} (y) d y \\ = \mathbb {E} X + \mathbb {E} Y. \\ \end{array} +$$ + +![](images/4b7f05d06fa0582aa56da7f71b704c3cc3f861ba3822814985957db373b01781.jpg) + +\- Example4: Let $X_{1}, \ldots, X_{n}$ be independent and identically distributed random (i.i.d.) random variables. Suppose $\mathbb{E}X_{i} = \mu$ . We call the quantity + +$$ +\bar {X} = \sum_ {i = 1} ^ {n} \frac {X _ {i}}{n} +$$ + +the sample mean. Compute $\mathbb{E}\left[\bar{X}\right]$ . + +\- Solution: We use the properties of expectation + +$$ +\begin{array}{l} \mathbb {E} \left[ \bar {X} \right] = \mathbb {E} \left[ \sum_ {i = 1} ^ {n} \frac {X _ {i}}{n} \right] \\ = \frac {1}{n} \mathbb {E} \left[ X _ {1} + \dots + X _ {n} \right] \\ = \frac {1}{n} \left(\mathbb {E} \left[ X _ {1} \right] + \dots + \mathbb {E} \left[ X _ {n} \right]\right) \\ = \frac {1}{n} (\mu + \dots + \mu) = \frac {n \mu}{n} \\ = \mu . \\ \end{array} +$$ + +\- In statistics, the sample mean is used to estimate the actual mean of a distribution. + +THEOREM 19. If $X, Y$ are independent then + +$$ +\mathbb {E} [ X Y ] = (\mathbb {E} X) (\mathbb {E} Y). +$$ + +PROOF. In the continuosuc case we have + +$$ +\begin{array}{l} \mathbb {E} \left[ X Y \right] = \int \int x y f _ {X, Y} (x, y) d y d x \\ = \int \int x y f _ {X} (x) f _ {Y} (y) d y d x \\ = \left(\int x f _ {X} (x) d x\right) \left(\int y f _ {Y} (y) d y\right) \\ = \left(\mathbb {E} X\right) (\mathbb {E} Y). \\ \end{array} +$$ + +The discrete case is the same, except replace integrals with summations. + +\- In general, the following is true: + +THEOREM 20. If $X, Y$ are independent and $g, h: \mathbb{R} \to \mathbb{R}$ then + +$$ +\mathbb {E} \left[ g (X) h (Y) \right] = \mathbb {E} \left[ g (X) \right] \mathbb {E} \left[ h (Y) \right]. +$$ + +## 12.2. Covariance and Correlations. + +- Note that $\mathbb{E}X$ and $\operatorname{Var}X$ give information about a single random variable. +- What statistic can give us information about how $X$ effects $Y$ , or vice versa? + +DEFINITION. The covariance between $X$ and $Y$ , is defined by + +$$ +\operatorname{Cov} (X, Y) = \mathbb {E} \left[ \left(X - \mu_ {X}\right) \left(Y - \mu_ {Y}\right) \right]. +$$ + +• After some algebra one can show that + +$$ +\operatorname{Cov} (X, Y) = \mathbb {E} [ X Y ] - \mathbb {E} X \mathbb {E} Y. +$$ + +\- The covariance between two random variables give us information about relationship between the random variables. + +- Covariance is a measure of how much two random variables change together. +- If the greater values of one variable mainly correspond with the greater values of the other variable, and the same holds for the lesser values, i.e., the variables tend to show similar behavior, the covariance is positive. +\* Thus covariance measures if there is a linear relationship between $X$ and $Y$ . +- The sign of the covariance therefore shows the tendency in the linear relationship between the variables. +- For example, the following plots shows a positive linear relationship between $X$ and $Y$ : + +![](images/64908a431074cbe7cf4daf03811bccb1153513c8a97bfecd51f33c0910385b5e.jpg) + +
    +scatterplot + +| x | y | +| ---- | ---- | +| -20 | 3 | +| -15 | 4 | +| -10 | 5 | +| -5 | 6 | +| 0 | 7 | +| 5 | 8 | +| 10 | 9 | +| 15 | 10 | +| 20 | 11 | +| 25 | 12 | +| 30 | 13 | +| 35 | 14 | +| 40 | 15 | +| 45 | 16 | +| 50 | 17 | +| 55 | 18 | +| 60 | 19 | +
    + +In this case $\operatorname{Cov}(X,Y) > 0$ . + +\- Note: If $X, Y$ are independent then $\operatorname{Cov}(X, Y) = 0$ . (This is not true in the other direction. Meaning $\operatorname{Cov}(X, Y) = 0$ does not imply that $X, Y$ are independent!) +- So $\operatorname{Cov}(X, Y) = 0$ means they are uncorrelated. + +\- Properties: + +- (i) $\operatorname{Cov}(X,Y) = \operatorname{Cov}(Y,X)$ +- (ii) $\operatorname{Cov}(X, X) = \operatorname{Var}(X)$ +- (iii) $\operatorname{Cov}(aX,Y) = a\operatorname{Cov}(X,Y)$ +- (iv) $\operatorname{Cov}\left(\sum_{i} X_{i}, \sum_{j} Y_{j}\right) = \sum_{i} \sum_{j} \operatorname{Cov}(X_{i}, Y_{j})$ . + +THEOREM 21. (★) Formula for Sum of Variation: + +$$ +\operatorname{Var} (X + Y) = \operatorname{Var} (X) + \operatorname{Var} (Y) + 2 \operatorname{Cov} (X, Y). +$$ + +Gives us a formula for variation of $X_{1},\ldots ,X_{n}$ : + +$$ +\operatorname{Var} \left(\sum_ {i = 1} ^ {n} X _ {i}\right) = \sum_ {i = 1} ^ {n} \operatorname{Var} \left(X _ {i}\right) + 2 \sum \sum_ {i < j} \operatorname{Cov} \left(X _ {i}, X _ {j}\right). +$$ + +• Fact: Note that if X, Y are independent then + +$$ +\operatorname{Var} (X + Y) = \operatorname{Var} (X) + \operatorname{Var} (Y). +$$ + +\- Finally we have the following: Its standardized way to know how correlated two random variables are: + +DEFINITION. The correlation coefficient of two random variables $X$ and $Y$ , denoted by $\rho(X, Y)$ is defined by + +$$ +\rho (X, Y) = \frac {\operatorname{Cov} (X , Y)}{\sqrt {\operatorname{Var} (X) \operatorname{Var} (Y)}}. +$$ + +\- Fact: + +$-(1) - 1 \leq \rho(X, Y) \leq 1$ +- (2) If $\rho(X,Y)=1$ then $Y=a+bX$ where $b=\frac{\sigma_y}{\sigma_x}>0$ (Straight positive sloped line) +- (3) If $\rho(X,Y) = -1$ then $Y = a + bX$ where $b = -\frac{\sigma_y}{\sigma_x} < 0$ (Straight negatively sloped line) +- (4) This $\rho$ is a measure of linearity between $Y$ and $X$ . + +\* $\rho > 0$ positive linearity: Meaning that if you were to draw a line of best fit, then it must be a positive sloped line + +\- The closer $\rho$ gets to 1, the more $(X,Y)$ seems to be in a positive sloped straight line + +\* $\rho < 0$ negative linearity: Meaning that if you were to draw a line of best fit, then it must be a negative sloped line + +\- The closer $\rho$ gets to $-1$ , the more $(X,Y)$ seems to be in a negative sloped straight line + +\- (5) If $\rho(X,Y)=0$ , then $X$ and $Y$ are uncorrelated. + +\- Warning: + +$-\rho (X,Y)$ does not pick up any other relationship, such as quadratic, or cubic + +\- $\rho(X, Y)$ is not the slope of the line of best fit. It is simply tell us if it's positive, or negative relationship, and the strength of relationship. + +\- Example1: Suppose $X, Y$ are random variables whose joint pdf is given by + +$$ +f (x, y) = \left\{ \begin{array}{l l} \frac {1}{y} & 0 < y < 1, 0 < x < y \\ 0 & \text { otherwise } \end{array} \right.. +$$ + +- Part (a): Find the covariance of $X$ and $Y$ . +- Part (b) Compute $\operatorname{Var}(X)$ and $\operatorname{Var}(Y)$ . +- $\underline{\text{Part (c)}}$ Calculate $\rho(X,Y)$ . +- Solution: +- $\underline{\text{Part (a):}}$ Find the covariance of $X$ and $Y$ . + +\- Recall that $\operatorname{Cov}(X,Y) = \mathbb{E}XY - \mathbb{E}X\mathbb{E}Y$ . So + +$$ +\mathbb {E} X Y = \int_ {0} ^ {1} \int_ {0} ^ {y} x y \frac {1}{y} d x d y = \int_ {0} ^ {1} \frac {y ^ {2}}{2} d y = \frac {1}{6} +$$ + +$$ +\mathbb {E} X = \int_ {0} ^ {1} \int_ {0} ^ {y} x \frac {1}{y} d x d y = \int_ {0} ^ {1} \frac {y}{2} d y = \frac {1}{4}. +$$ + +$$ +\mathbb {E} Y = \int_ {0} ^ {1} \int_ {0} ^ {y} y \frac {1}{y} d x d y = \int_ {0} ^ {1} y d y = \frac {1}{2}. +$$ + +Thus + +$$ +\operatorname{Cov} (X, Y) = \mathbb {E} X Y - \mathbb {E} X \mathbb {E} Y +$$ + +$$ += \frac {1}{6} - \frac {1}{4} \frac {1}{2} +$$ + +$$ += \frac {1}{2 4}. +$$ + +\- Part (b): Compute $\operatorname{Var}(X)$ and $\operatorname{Var}(Y)$ . + +\- We have that + +$$ +\mathbb {E} X ^ {2} = \int_ {0} ^ {1} \int_ {0} ^ {y} x ^ {2} \frac {1}{y} d x d y = \int_ {0} ^ {1} \frac {y ^ {2}}{3} d y = \frac {1}{9}. +$$ + +$$ +\mathbb {E} Y ^ {2} = \int_ {0} ^ {1} \int_ {0} ^ {y} y ^ {2} \frac {1}{y} d x d y = \int_ {0} ^ {1} y ^ {2} d y = \frac {1}{3}. +$$ + +\- Thus recall that + +$$ +\operatorname{Var} (X) = \mathbb {E} X ^ {2} - (\mathbb {E} X) ^ {2} +$$ + +$$ += \frac {1}{9} - \left(\frac {1}{4}\right) ^ {2} = \frac {7}{1 4 4} +$$ + +Also + +$$ +\operatorname{Var} (Y) = \mathbb {E} Y ^ {2} - (\mathbb {E} Y) ^ {2} +$$ + +$$ += \frac {1}{3} - \left(\frac {1}{2}\right) ^ {2} = \frac {1}{1 2}. +$$ + +\- Part (c): Calculate $\rho(X,Y)$ . + +\- We now use + +$$ +\rho (X, Y) = \frac {\operatorname{Cov} (X , Y)}{\sqrt {\operatorname{Var} (X) \operatorname{Var} (Y)}} +$$ + +$$ += \frac {\frac {1}{2 4}}{\sqrt {\left(\frac {7}{1 4 4}\right) \left(\frac {1}{1 2}\right)}} \approx . 6 5 4 7. +$$ + +## CHAPTER 13 + +## Moment generating functions + +## 13.1. Moment Generating Functions + +\- For each random variable $X$ , we can define its moment generating function $m_X(t)$ by + +$$ +\begin{array}{l} {m _ {X} (t)} = {\mathbb {E} \left[ e ^ {t X} \right]} \\ = \left\{ \begin{array}{l l} \sum_ {x _ {i}} e ^ {t x _ {i}} p (x _ {i}) & , \text {if X is discrete} \\ \int_ {- \infty} ^ {\infty} e ^ {t x} f (x) d s & , \text {if X is continuous} \end{array} \right.. \\ \end{array} +$$ + +\- $m_X(t)$ is called the moment generating function (m.g.f.) because we can find all the moments of $X$ by differentiating $m(t)$ and then evaluating at $t = 0$ . + +\- Note that + +$$ +\begin{array}{l} {m ^ {\prime} (t)} = {\frac {d}{d t} \mathbb {E} \left[ e ^ {t X} \right]} \\ = \mathbb {E} \left[ \frac {d}{d t} e ^ {t X} \right] \\ = \mathbb {E} \left[ X e ^ {t X} \right]. \\ \end{array} +$$ + +Now evaluate at t = 0 and get + +$$ +m ^ {\prime} (0) = \mathbb {E} \left[ X e ^ {0 \cdot X} \right] = \mathbb {E} [ X ]. +$$ + +\- Similarly, + +$$ +\begin{array}{l} {m ^ {\prime \prime} (t)} = {\frac {d}{d t} \mathbb {E} \left[ X e ^ {t X} \right]} \\ = \mathbb {E} \left[ X ^ {2} e ^ {t X} \right] \\ \end{array} +$$ + +so that + +$$ +m ^ {\prime \prime} (0) = \mathbb {E} \left[ X ^ {2} e ^ {0} \right] = \mathbb {E} \left[ X ^ {2} \right]. +$$ + +THEOREM 22. For all $n \geq 0$ we have + +$$ +\mathbb {E} \left[ X ^ {n} \right] = m ^ {(n)} (0). +$$ + +• Examples of Moment generating Functions +- Bernoulli: Recall that $p(1) = p$ and $p(0) = 1 - p$ . Thus + +$$ +\begin{array}{l} m _ {X} (t) = \mathbb {E} e ^ {t X} = e ^ {t \cdot 0} p (0) + e ^ {t \cdot 1} p (1) \\ = p e ^ {t} + (1 - p). \\ \end{array} +$$ + +\- Binomial: Recall that $X \sim \text{Bin}(n, p)$ if $X = \sum_{i=0}^{n} Y_i$ where $Y_i \sim \text{Bern}(p)$ thus + +$$ +\begin{array}{l} m _ {X} (t) = \mathbb {E} e ^ {t X} = \mathbb {E} e ^ {t X \sum_ {i = 0} ^ {n} Y _ {i}} \\ = \mathbb {E} \left[ \left(e ^ {t Y _ {1}}\right) \dots \left(e ^ {t Y _ {n}}\right) \right] \\ = \mathbb {E} \left[ e ^ {t Y _ {1}} \right] \dots \mathbb {E} \left[ e ^ {t Y _ {n}} \right], \text {by independence} \\ = \left(p e ^ {t} + (1 - p)\right) ^ {n} \\ \end{array} +$$ + +\- Poisson: If $X \sim \text{Poisson}(\lambda)$ then + +$$ +\begin{array}{l} m _ {X} (t) = \mathbb {E} e ^ {t X} = \sum_ {n = 0} ^ {\infty} e ^ {t n} e ^ {- \lambda} \frac {\lambda^ {n}}{n !} \\ = e ^ {- \lambda} \sum_ {n = 0} ^ {\infty} e ^ {t n} \frac {\lambda^ {n}}{n !} \\ = e ^ {- \lambda} \sum_ {n = 0} ^ {\infty} \frac {(e ^ {t} \lambda) ^ {n}}{n !} \\ \end{array} +$$ + +now recall from Calculus 2 that $e^x = \sum_{n=0}^{\infty} \frac{x^n}{n!}$ so that + +$$ +\begin{array}{l} m _ {X} (t) = e ^ {- \lambda} \sum_ {n = 0} ^ {\infty} \frac {x ^ {n}}{n !}, \text { with } x = e ^ {t} \lambda \\ = e ^ {- \lambda} e ^ {e ^ {t} \lambda} \\ = e ^ {e ^ {t} \lambda - \lambda} \\ = \exp (\lambda (e ^ {t} - 1)) \\ \end{array} +$$ + +\- Exponential: If $X \sim \exp(\lambda)$ then + +$$ +\begin{array}{l} m _ {X} (t) = \mathbb {E} e ^ {t X} \\ = \int_ {0} ^ {\infty} e ^ {t x} \lambda e ^ {- \lambda x} d x \\ = \frac {\lambda}{\lambda - t}, \\ \end{array} +$$ + +which is valued whenever $t > \lambda$ . + +\- Standard Normal: If $X \sim \mathcal{N}(0,1)$ then + +$$ +\begin{array}{l} m _ {X} (t) = \mathbb {E} e ^ {t X} = \frac {1}{\sqrt {2 \pi}} \int_ {- \infty} ^ {\infty} e ^ {t x} e ^ {- x ^ {2} / 2} \\ = e ^ {t ^ {2} / 2}. \\ \end{array} +$$ + +\- Normal: If $X \sim \mathcal{N}(\mu, \sigma^2)$ then $X = \mu + \sigma Z$ so that + +$$ +\begin{array}{l} m _ {X} (t) = \mathbb {E} e ^ {t X} \\ = \mathbb {E} e ^ {t \mu} e ^ {t \sigma Z} = e ^ {t \mu} \mathbb {E} e ^ {(t \sigma) Z} \\ = e ^ {t \mu} m _ {X} (t \sigma) = e ^ {t \mu} e ^ {(t \sigma) ^ {2} / 2} \\ = \exp \left(t \mu + \frac {t ^ {2} \sigma^ {2}}{2}\right). \\ \end{array} +$$ + +\- Property: Suppose $X, Y$ are independent then what is that m.g.f. of $X + Y$ ? + +\- Let's try to figure out: + +$$ +\begin{array}{l} m _ {X + Y} (t) = \mathbb {E} e ^ {t (X + Y)} = \mathbb {E} \left(e ^ {t X} e ^ {t Y}\right) \\ = \mathbb {E} \left(e ^ {t X}\right) \mathbb {E} \left(e ^ {t Y}\right), \text { by independence } \\ = m _ {X} (t) m _ {Y} (t). \\ \end{array} +$$ + +\- Thus we know that + +$$ +m _ {X + Y} (t) = m _ {X} (t) m _ {Y} (t). +$$ + +\- Note: Also note that of $f_{X}(x)$ is the pdf of a r.v. then it's m.g.f is + +$$ +\mathbb {E} e ^ {t X} = \int e ^ {t x} f _ {X} (x) d x. +$$ + +This is similar to the $\underline{\text{laplace transform}}$ of $f_X(x)$ . $[\mathcal{L}[f](s) = \int e^{-sx} f_X(x) dx]$ . + +\- Recall that there is one to one correspondence of laplace transforms. That completely determines a function. + +THEOREM 23. If $m_X(t) = m_Y(t) < \infty$ for all $t$ in an interval, then $X$ and $Y$ have the same distribution. That is, m.g.f's completely determines the distribution. + +\- Example1: Suppose that m.g.f of $X$ is given by $m(t) = e^{3(e^t - 1)}$ . Find $\mathbb{P}(X = 0)$ . + +\- Solution: (We want to work backwards). Match this m.g.f to a known m.g.f in our table. Looks like + +$$ +m (t) = e ^ {3 \left(e ^ {t} - 1\right)} = e ^ {\lambda \left(e ^ {t} - 1\right)} \quad \text { where } \lambda = 3. +$$ + +Thus $X\sim \text{Poisson}(3)$ . Thus + +$$ +\mathbb {P} (X = 0) = e ^ {- \lambda} \frac {\lambda^ {0}}{0 !} = e ^ {- 3}. +$$ + +\- Summary: + +(1) $m(t) = \mathbb{E}e^{tX}$ . We have a table of mgf of distributions: +(2) The m.g.f helps us find moments: $\mathbb{E}[X^n] = m^{(n)}(0)$ +(3) If $X, Y$ are independent then $m_{X + Y}(t) = m_X(t)m_Y(t)$ . +(4) The m.g.f. helps us determine the distribution of random variables. If $m_X(t) = m_Y(t)$ then $X$ and $Y$ have the same distribution. + +\- Recall we had a section on sums of independent random variables. + +\- Example2: Recall $X \sim \mathcal{N}\left(\mu_x, \sigma_x^2\right)$ and $Y \sim \mathcal{N}(\mu_y, \sigma_y^2)$ , independent. Then what is + +$$ +X + Y \sim \mathcal {N} (\mathrm{?},?) +$$ + +\- Sol: Note that + +$$ +\begin{array}{l} m _ {X + Y} (t) = m _ {X} (t) m _ {Y} (t) \\ = \exp \left(t \mu_ {x} + \frac {t ^ {2} \sigma_ {x} ^ {2}}{2}\right) \exp \left(t \mu_ {y} + \frac {t ^ {2} \sigma_ {y} ^ {2}}{2}\right) \\ = \exp \left(t \left(\mu_ {x} + \mu_ {y}\right) + \frac {t ^ {2} \left(\sigma_ {x} ^ {2} + \sigma_ {y} ^ {2}\right)}{2}\right). \\ \end{array} +$$ + +So then you look at our table and check which distribution has this mg.f. with $\mu = \mu_x + \mu_y$ and $\sigma^2 = \sigma_x^2 +\sigma_y^2$ . so that $X + Y\sim \mathcal{N}\left(\mu_x + \mu_y,\sigma_x^2 +\sigma_y^2\right)$ + +\- Example3: Suppose $X \sim \text{bin}(n, p)$ and $Y \sim \text{bin}(m, p)$ , independent, then what is the distribution of $X + Y$ ? + +\- Solution: We use + +$$ +\begin{array}{l} {m _ {X + Y} (t)} = {m _ {X} (t) m _ {Y} (t)} \\ = \left(p e ^ {t} + (1 - p)\right) ^ {n} \left(p e ^ {t} + (1 - p)\right) ^ {m} \\ = \left(p e ^ {t} + (1 - p)\right) ^ {n + m}. \\ \end{array} +$$ + +Look at the table and see what distribution has this m.g.f. Thus + +$$ +X + Y \sim b i n (n + m, p). +$$ + +\- Example4: Suppose $X$ is a discrete random variable and has the m.g.f. + +$$ +m _ {X} (t) = \frac {1}{7} e ^ {2 t} + \frac {3}{7} e ^ {3 t} + \frac {2}{7} e ^ {5 t} + \frac {1}{7} e ^ {8 t}. +$$ + +Question: What is the p.m.f of $X$ ? Find $\mathbb{E}X$ . + +\- Solution(a): This doesn't match any of the known mg.f.s. Thus we can read off from the mgf that since + +$$ +\frac {1}{7} e ^ {2 t} + \frac {3}{7} e ^ {3 t} + \frac {2}{7} e ^ {5 t} + \frac {1}{7} e ^ {8 t} = \sum_ {i = 1} ^ {4} e ^ {t x _ {i}} p (x _ {i}) +$$ + +then $p(2) = \frac{1}{7}, p(3) = \frac{3}{7}, p(5) = \frac{2}{7}$ and $p(8) = \frac{1}{7}$ . + +\- Solution(b): First + +$$ +m ^ {\prime} (t) = \frac {2}{7} e ^ {2 t} + \frac {9}{7} e ^ {3 t} + \frac {1 0}{7} e ^ {5 t} + \frac {8}{7} e ^ {8 t}, +$$ + +so that + +$$ +\mathbb {E} [ X ] = m ^ {\prime} (0) = \frac {2}{7} + \frac {9}{7} + \frac {1 0}{7} + \frac {8}{7} = \frac {2 9}{7}. +$$ + +• Example5: Suppose X has m.g.f + +$$ +m _ {X} (t) = (1 - 2 t) ^ {- \frac {1}{2}} \text { for } t < \frac {1}{2}. +$$ + +Find the first and second moments of $X$ . + +\- Solution: We have + +$$ +m _ {X} ^ {\prime} (t) = - \frac {1}{2} (1 - 2 t) ^ {- \frac {3}{2}} (- 2) = (1 - 2 t) ^ {- \frac {3}{2}}, +$$ + +$$ +m _ {X} ^ {\prime \prime} (t) = - \frac {3}{2} (1 - 2 t) ^ {- \frac {5}{2}} (- 2) = 3 (1 - 2 t) ^ {- \frac {5}{2}}. +$$ + +So that + +$$ +\begin{array}{l} \mathbb {E} X = m _ {X} ^ {\prime} (0) = (1 - 2 \cdot 0) ^ {- \frac {3}{2}} = 1, \\ \mathbb {E} X ^ {2} = m _ {X} ^ {\prime \prime} (0) = 3 (1 - 2 \cdot 0) ^ {- \frac {5}{2}} = 3. \\ \end{array} +$$ + +## CHAPTER 14 + +## Limit Laws + +## 14.1. The Central Limit Theorem + +\- The CLT is one of the most remarkable theorems in Probability. + +\- It helps us understand why the empirical frequencies of so many natural populations exhibit bell-shaped (normal) curves. + +\- Recall that $i.i.d.$ means independent and identically distributed random variables. + +THEOREM 24. (CLT) Let $X_{1}, X_{2}, X_{3} \ldots$ be i.i.d. each with mean $\mu$ and variance $\sigma^{2}$ . Then the distribution of + +$$ +\frac {X _ {1} + \cdots + X _ {n} - n \mu}{\sigma \sqrt {n}} +$$ + +tends to the standard normal $Z$ as $n\to \infty$ . That is, + +$$ +\mathbb {P} \left(\frac {X _ {1} + \cdots + X _ {n} - n \mu}{\sigma \sqrt {n}} \leq b\right) \approx \mathbb {P} (Z \leq b) = \Phi (b). +$$ + +when $n$ is large. + +\- The CLT helps us approximate the probability of anything involving $X_{1} + \cdots + X_{n}$ where $X_{i}$ are independent and identically distributed. + +\- When approximating discrete distributions: USE the $\pm .5$ continuity correction: + +\- Example1: If 10 fair dice are rolled, find the approximate probability that the sum obtained is between 30 and 40, inclusive. + +\- Solution: Let $X_{i}$ denote the value of the ith die. Recall that + +$$ +\mathbb {E} \left(X _ {i}\right) = \frac {7}{2} \operatorname{Var} (X _ {i}) = \frac {3 5}{1 2}. +$$ + +Take + +$$ +X = X _ {1} + \dots + X _ {n} +$$ + +to be their sum. + +\- Using the CLT we need + +$$ +n \mu = 1 0 \cdot \frac {7}{2} = 3 5 +$$ + +$$ +\sigma \sqrt {n} = \sqrt {\frac {3 5 0}{1 2}} +$$ + +thus using the continuity correction, then + +$$ +\mathbb {P} (2 9. 5 \leq X \leq 4 0. 5) = \mathbb {P} \left(\frac {2 9 . 5 - 3 5}{\sqrt {\frac {3 5 0}{1 2}}} \leq \frac {X - 3 5}{\sqrt {\frac {3 5 0}{1 2}}} \leq \frac {4 0 . 5 - 3 5}{\sqrt {\frac {3 5 0}{1 2}}}\right) +$$ + +$$ +\approx \mathbb {P} (- 1. 0 1 8 4 \leq Z \leq 1. 0 1 8 4) +$$ + +$$ += \Phi (1. 0 1 8 4) - \Phi (- 1. 0 1 8 4) +$$ + +$$ += 2 \Phi (1. 0 1 8 4) - 1 = . 6 9 2. +$$ + +\- Example2: An instructor has 1000 exams that will be graded in sequence. + +- The times required to grade exam exam are i.i.d. with $\mu = 20$ minutes and SD $\sigma = 4$ minutes. +- Approximate prob that the instructor will grade at least 25 exams in the first 450 minutes of work. +- Solution: +- Let $X_{i}$ be the time it takes to grade exam $i$ . Then + +$$ +X = X _ {1} + \dots + X _ {2 5} +$$ + +is the time it takes to grade the first 25 exams. We want $\mathbb{P}(X\leq 450)$ . + +\- Use CLT, + +$$ +n \mu = 2 5 \cdot 2 0 = 5 0 0 +$$ + +$$ +\sigma \sqrt {n} = 4 \sqrt {2 5} = 2 0. +$$ + +\- Thus + +$$ +\begin{array}{l} \mathbb {P} (X \leq 4 5 0) = \mathbb {P} \left(\frac {X - 5 0 0}{2 0} \leq \frac {4 5 0 - 5 0 0}{2 0}\right) \\ \approx \mathbb {P} (Z \leq - 2. 5) \\ = 1 - \Phi (2. 5) \\ = . 0 0 6. \\ \end{array} +$$ \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-1-_- Events_.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-1-_- Events_.jpg new file mode 100644 index 0000000000000000000000000000000000000000..8e83d4f8cfcd2d0c5d4ff77dffc37be45a176375 GIT binary patch literal 5062 zcmd6qc|4R~`^V4N#@bMdd@bFKsRLt}tdS5rq5z+eEt zC>MYRfjXe4ql44Y(!=3!1_pXYW_A{4CMM>i2M;0GdAN|gJY3w|#{^G^9OD-k;O0ij zi;AC=l9iQ3iu|dfAgz2tMppXAATS0724*H^P8JqUX+CZ~>Hl+s-UBvz@CbBM!}tLe z8;qI_1~mdCB~Dt{&j$Y4U{usJv~+NK21X`IhcZ?`1*4{>qM@dxrJ_52JC|p!DG}JWkA6zghABs`4(a`cq(H&GXf?sw!#4ml9p8ZTr z#;ZC80U6`f!<#j7J1z$A#8@p#3EK-+=}HS7iSK_HV8Mz)TIJOdd5GKm+2$nTW)w zWhv_5$@QF8qS&V;*+@2JWS&dkfU*yf&DOS>@Fefjn-=gn08mHiel(7kize|VZ6?&> z(2A;E9K_~XYnzn^%A!|~q|EpRN|xH`#%R&4M{BospzC2BU&_$qRI9anu=L%*T_@Xs z2KDqEeY;U)3exG(&g6PAhRiC-1c6R2M0;&h)Yj>Dk`OrUh&}}YE=kq#jLv`;k}FZ4 zqe?&Qt!?+AAwbV@9|ChM@R`v`t=(g`91uvCMJ8b)Inr)*&h3YzHE@NzwY!)GfqP8# z5ZIs#S$AE-{_5vZ$6H6w{Q$7bQW*b0uMP;*(~e_E)N0!mTlnAHXv_}DDk6RcfZwN{ zg}}d=9Dj3@xlZ8Xf&UC(ZlBWw0;_j@?Z`o$fj7vqm|wiyCbAp|`RU$spRy7Hza`+8 z%sIZ>34NgUThbx$D?joVC;y+aOf?~Ze=w`zi-=rG$G@}jyPm^WUSaeH{iJ=6?g>A< zUSdFV#s+hm`@2EI#{^eqr+Jp-K5wjTrQh4}x*LzX4VBF8;8N_64~7i7lJ+efA# z@bdl&`AMt6E{4zZcyUUe&93^Z?vk-5R%%z}qv@D$RZB-8H^W2X2(;epM*39YsPugk zJ=&vb5kyJ^me@k)-l`KRLd!$BF~wl0&PQsDIW?KMeBsDhOx41udg7;ejb}G3Rt8P0 zAz*HdI@M%+P;104UI~}gNsFL$zMxnh<%aQckK2xSe>jY2eaX#&8}eE)=Q!W%+C4O1 zULM@GjvTNFX;aapW9R)!+A*!-4XWaODx6wTA#YS&y^<_Eb}=?^;cj>0m(p@QC50PU z!7Obyoy8!j)2#Cqi@5{0N31mXU+M7M$39lze^+HyP${u;V!X1qh?i(81RrW!IqF3e@j%6Gi0sQA0y z&Y+4yfz_bFbpB1ds`gjxuXv|A;U=6{Bh|}i09AjJfnxae6kfL(;Q)C?5WsG9n)Vmt zWmyv=Yf&;nyaEBdiw5#w?UY|F1iU5pj+7ePtIQZyT6Y9NV9F|vT+dNUG7<9p67X@x z9}WTjp|nlb8VG!Oj*4&o#-aWqGh8&tNaK93oSzNioZqK%B{lkoSP6mDaBQV4%z0vl zNzW&B#vzpb=yh)ZBN%wtt;RZA;>E^`l-erh_pp8m<7k592&Nbivr0RQzi)S$tR0U` zCzOAwYdin=0|amsU071;GX-Xh=|BN^ryscFYjqyCT^7Z*aB#Npl)PuRb` zUnR|(+UTv?RMZ#dJJ{$J#d;3M>Rnc8fFHFzGiaZ@fr6iD*xAz%5q$bpIySsoMcri0 zEc@wbzxa5Pad%5gX(@P0^hO^?xH~7jTr)3Pw{yNVe==?R@|8>-_^X&uG~fAMl^$6G zEp@T1K_;DvV>W}8ZQ--k^5z}aR%`;4Ho1JdXW&5_7J)A70lBr_?`mS|AP_`gBAKSG z2Knl7Qj{#PVR71HyH87l<7gWaHi!4fsUausMW!r0gzv(IFuLmY6Ez|;Z@Mq_j3d7_ zeLriQmK83w9ln6d9J}lkDEg+hYKOMX#AzS_DVNeY&^%1PZS9vQS8YKme{{KMARv_l zI|fIMjZdoPH$(?~xMJz-Gx7)SfnJ}m{3(-@dMf%62Mx8Sm;S(4A&Y>Qr;!F=rARH-QsZyN5<+d^Po|7n*FQSA9D za`%xtrvD5P0=f>FcE}c+1Bv0N7t}(}6!>muduikKWTf7oSB$eR4{;TFSax_#Wy8TL zNHd-?)F*|G-7AsarD^P`@05SjQJ09(~sxA zr%q=FrT+|<-N>wrY0k4A%)*S@lh(FE=xMQK|liTo%~%TP}FZ zSlHITv1-e<;k>0`ks0)7)B#%?C(=C}SNDw5^}EW0N)MEW)RNu(Uaum?{Q3;7XCwZ~ zzHv&zvK`a4k*C_lz+-WGyKC-K%=hN(n-(@7F!r=lELEbD7ObRTnbU{x8tkyK(O)IR z=y!P$mKkx_dD^yxAZt{jm-zf3kS%?|KxC-+`g_bbiBQipq889z-v5>3nqLk&XMPwM5ByZarYvjaY--#yHCyNfO zJRR-XBcE;_8G}p2966s>P73fTE_TDvG)eCxmGs zV#rH&Bne0C#D(1`9|#aAl9<^Wu7CXHbX!j$ATSJpomZH-vdpM3yKgF^5UA!L zKe7~+N?rfk&sgR3Hzj(aH@VG}(PgB;{(jou_*_ihJWj}#zg&1=o%~i?s|_=J2{CJ< zeJn_=MTJFrVnnRqw#00tL#P>Qu7U4K1glHJ>T-9L5OJ5E6cN!RCy`_QCj{*1$=pQG zxd2VXFDiN1#I$Ryr?j9(aoR`C*d>3|+-*4))Z zOd6JGZl#S|I1EnMDDtn@CrMsWYkPGdLd61Ol(_CYrDJa)aH*x?*yHcB<~gmB0v}8C zVr|t6UTs-mlOVumy%;M=5H}rs?|1Z%m6#3gK;cRzE3tMv;}G4h3trQ8JN*-agyki) z2}O<5=gM(Nyo6`9T;qA+;pT!0Z`G-(-nY(}8%dX)Tt2sRp)#uvtWy46kQp9IAdJ+1 zu`0-m<>WW+5J;w_t^rqk>&X{o?{_T??x9VhEv@c$-$csn9d>uIH+=obpsT1boilty zE>i0yXECOsm&k?76A4?HaJ(0jP`V|^>6k7e^l+&oC?JTcD)HwUTXp?)9#Ox~aoo`r z>Aw{%VlVzyeV)akqpNu2eiWE5mF-=x%Re|FRIQ^o7^MimH(?s1sixi!0sWbZMFHb4 z1UVV^*jCPa_4xO&OJ)hIdiF)Y!4Zc_9bJ0HE5O9kxKMkohMINQ!GRe9Or*SBvEsr4 z8!qA1DK5=CZjOi1;>Mx+1!MLKa!+I>T+cutnl3&56gE3Fv++J}80r(Ekqa#37F+wa zSnMW_MCLgaD>x~as3PkoJCNs(ciYOXc4NhgEK1(Au#+_(!D=)vw@ym3oMpIivw1ZE zIcKD2AaQECcYfvj_^RSl31r?xI-e%gUo1JkIgm$l zkY%kp?zQFFb;=u@TJE`56DoGoUp(!WwhVV`oO4_P%!LHgw%9Di=?FRbUmbfTZ)0vn zpH99SCAbNEL^>}BQ@y<@XKkLMny$<7hdQr+&#V!2OsRj$bBK>|n zm+2OX#_1ShzQd%GO46>F?Q73=emxc4ZzKx=7WZ77zF4^&Q>xqWnY1ssW*U|CveG{`wYAwQ;y)M9n;a# zJzC(1w-|L!+Mz6S?0!!e)@=Ian}iuFZEjNOr5@KZgFVXm-e@zgwwA;QQdLgX3nVNR zMMn`Du_Uu-zzJZ>77+N4AvCV++n=JlD6C~?cxbk~=_{=xBeFs|q#H7{? k|GbZi+>bFG!STCfWNZa5i$uCUe|%2h=ebDaZ3p!A9|{ZBL;wH) literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-10-PROPOSITIO.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-10-PROPOSITIO.jpg new file mode 100644 index 0000000000000000000000000000000000000000..d293d5bf73625e88ec1cd3fddacc01c296889e11 GIT binary patch literal 853 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}niCwprZwe|*^SpJAQvpWZ?{!4LMW`Zn^5 z@7(#3?>Vb7=TW7~q_dK9yLu*hJh{MLJ@K_`@%#_#_elR|a9Y>mG~XaS5>&t;PnUg!9iGWJ8PNxc)Ots9XM@Au)UZ zU#ZOchvHZ2UzB`g|NA-pKf{Tyix*~W%&@ACI$m5@5|=*r*$(Z!cW)O9Zw)E6ZCJqn G|0V#svOv55 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-11-Step Function Graph.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-11-Step Function Graph.jpg new file mode 100644 index 0000000000000000000000000000000000000000..a005f2424497c4293a8766453358895513d1ad9d GIT binary patch literal 59546 zcmd>nd0b5U|L|#(YzgTK%~gn8HxyZ_$xV&2Or>O>OK7oV$wskkJP z=He1Udx#n-+MD)iwlmNB%*;76;ro4l&+~fzd)$}hHs^dk@BRJxytk3OmwZNsJK8(g zBPuEgq5}URl8=ZjGH}2E%>fz%H8nN0v<40uGHU3M!Gnj488K3O)OdZ0!Fc^~<9?m^ zyYa7vzZs1iH^pM=Z!>0^o10UN=UL62W%;|A`7C@96)i2TA%lnfGIZ!KvnGt2Fzf&L zm*gKrXP`=-`fxQBLqt_aMNLOV@-IRGI5kwHzYzH^6;(BLjRBeiwFV7_2eO7Esw!$~ zs_JSQ8tUrsbTE95sOxBqm@sqUfRWA{H4S%(q!W)Q%%fg&#{<0Z@%p!JA2w<2genzZY$katzP4?dCOK$FK?f{e*5+xIOrd6 zTiU*CJXkIg^14m7Y+5Zf& z!~a*9k%%eqszipUsQ~lTbPy`S6HK`tsuQz*(EMX7fBE&);c3<%p9-d5ee!8d}kTaj^2S&x|ii zkfn8^RP*z^5}->}vg$KkitJ~}U1;MKRt)mn_6;-gQ!`o6WJwJM&f}Y9#$;09#Ci;5?DVsP4_Ci@o$%Stpv%>P7=BukVB#}Y1U*@{o746 zAG#Db*Ii45z>=$0Q&gYLe4k=NJ#h;)B6nLI;>A8-%<1km|4|%9&FoseKtq!+LH@E% zJAsyr1(+Xy{8UC@wP1zErVb;6#DzC59>gW$tMg-1fhGDdx-GK0Ol4DA;eu!Twv9M+ z&vVQ6o?l!Jic)0&xlMbCmUeTiUSe57v~*6~P<<1s6jEn72@uy!&$>R`rl)o+dChtl z&5bRW^B`Ge^QJTB#@kt^6`HoKwZV(dFH0k!HSC+i0yYD9z| zE8J?xa;rHX267twFpxlJAlEZ+o9|4-_;rMTx91?#cIC7MZs*17A#L=YqF&IGaQ7gY zhK^XFHgcNrpK+MK1Q{eJ@-|a}Er@K+qepL$RSA}=-ZCO3c!#VIKA0ciq*lABBh4T| zZF%y9^E@5=$pwKcM6&e;8;u$q*`i^c_Qq5TbW_kvYWIK99lN~~j#aF7RekZ-v4CAu zWso^-D8LTw6+WPrTljj|jQr=@+Jkl5x164te?P?}J%Qb@3&&;K$I2UWy_z<2#P!}^ zOPd}p5%x)t1ID%=8%edfVgKy0@f{t$J|RT~o!M=5!q$(~p}(yFHY^x+&EW=LYB%Cz z8QuOe%ag2{?_)C4l+py&N508ak8i2_P3WBxL_gSlmIO&Kg=NCB?-Lemgj?X)um@Qq z?mK;~3fXXFrW#+aHmPftvO7#E57C~e{SstHvYP!xBi)ZcC{iO6YhYK-_@7u3liRcA z_pV#fl(R(G2+!=SCPhfqar5>Kvpvd6WcDHcFyj*d)Yu4$82QXmMYQY)UrvMgbBs%- zRWEsBtB4=OXkcr-|IXiQ|3r@2ghzh9o=rV5d()kpU>cW|p$CiwJ*{BtN-|{R2_3z1 z>d0G7Dwd^9xj=>Zm%!u3Rt2F)Tkl%_%mI^Qja%qx@I*~(8z}7e7Sd$qCVtC%u#k!0 z=u*)BA~YFU{Y%1@FlTydSAp;xNOlbQK~&laj&RY(&)bKG6$aJo64X0=9y5*rS*d*w z)b%PF5b^z9zb)M`$)hZ?OpvM0z7M;VmJs%M+U=DxGeS@8EaMn_ zCH=mqH^tdz*!7cB5csW{#jOg*a!^}3rFoM;Kaw{$>`;V=^-tLm@pgu-%}8gvQ7Yr8 z``8@cdKcGS;tB~8%)aViGvu;(5vz^$b@_OWIlet3&>S>ouHPsmYm5Y8o|PcG*%wkw z%p?e9Dprfmv>Ab~vcv&C(-v+s#ZLJ=biK>!YnLGR{+1vO&(rRzjH?Z(Ji=VH~AtWkT|23F4A9389&;k{}%HSVHpGW{OaP6qTJt z+{e~%_){9Hh+{LThp?ZU>;%EBX+OPLmLg6rr@Z+H3ncG*$6N05jU`9 zPz#GuwsqQHFYGv+p-(fKie-628l zQ)3VZ*Gy`y1X(UYGBPE|^~#L$X@y`aO4-e6K+St`gYP=qAvC?!RL<;4^1l!oRd6oF zWS#^m%W|;Mu$LgswIUzK2nq5ozpqIBEdGLa70Ua_FN*mqT=a2ogt%CCM^UPY@kQ&dC5ML(lnUc z*Zu(TTqcS@rYDK~7fO(I@A$6eLlyrnQyD8kGMNpWl7wWAI0&`JVsNH?FR}>m3g5 z(8WWORa-;0a}IFtZd#UO z0Y=R-6|}Jh(@k0tnY^MoK8}&Km{lE+5#-$U2SEkGfeh+As~|)*=OrdVx}EH=M;q)ze=i0F9nY3Zs_d_`D&yE8rXcs%Kx{%p43K+Upa%#m z@mphU`_kRWuAs+-pS`q&1G_SKCMXjU@YpT@?Ko0tilp4CtjfB!G}r9 zu21T%%`VePXa44L-|=HzyI6OU2+N zwf?pQ$+O#CJ)Q<>T8@0F6?JyL#Y62@;!+@pZ>JOm2jA!R&PdiNm=L zxHOR@cHNs0Ya7znjWwj{j*}owlt!BY1W03vB_@qhxe6#{Q!7|~z1@`Rw1>neYZc~E zr$r_quDd42{sAlEyB!h#5j+JU=>FBg!VZ`t*1QILu>+(c8K<_H)%6~%lwkxx%BYh_ z#t8ad3oi51_^=cO^(h^loHp%mK90z#gbQFjKnk7zi3*OTaws_qv>k5+xb6ZcfVOFW zRiNOK>tM}E3h9<0=^;-s|17WX4eSCrCZpuy;(FFe<6zX%mjGvB`^jhmxZ|LNd|a#> zjiWgIg@Pqs;(AuWb^Au+s}wF1tpW7Wo(|0XC&CPxGXo@q8vIaujbe+CX80NBZ9?3T zRyRtE5id#XX9cpphLrUz-@Xn?PbEeCOV)mcG?z~l+JZQiw^%J(mHZB~W{96*TNNU0 z?=XkOF<@2cPRH$e+`x?vmd8PwLK<4ErV0t60izOI(a@M&H7ODw{C}O)-^2 zYMHXt1nZNCTOXqm!fIZlzFX3VnkZSQ&kmLVRin6grO8}snQ@7Nosq85evCXSS`?sv z+qr;uU4ozjeG>U8a#c&25;vrIB6_ch`iR>64!9fZ?Y^I|ZCU3awN6vqe zgAR6x`=N=QBYF8raSCb8mLN_n$EeV0ucCH_^YXa@ZClbAm~z6wa5sPrAw~I0+@Y?S z9!+-hPa;#U%YG*%7mS;cKQSZu@WOK*p>wU8Ok3A4cpWabes@iRoGZX)TeWK#s&Do` z!q-1=E_OS^u-htGg;kkzKqNsb!InWNL!X-E$K^{9IIOba1=$h5#d!_78#O@^=cFb1k3f?u#ibHt z`uWmkx_-vhLl|otu>QQr$2*>GNIw(O*xE|hxuN-j9fP=I<)!R;>>b#7)4npATWMYi ztFf28oBj1~0W;vy*Cd`E1ggAy&+_%!EPQv{=%7B0oGu9>ThL4S?Q+(1>V0;MfeezD zC0WZXSAFJG9{E;Ku?2)Gg6odbj^t9e(%eRPIS?&?9rYi|pf54+O5p2d`yk;$cQ`z=gbbmGT}$iD zxGF)Yy=)#IeSVYgEq*3S?kp9C^PM!N%wE6XC8JyRoJ&81eO4LKCxzHS_(X~;;&R@t z^bSQ3oXG(U3mSr(=&cLp46Xb0Y(ePa$f>#)^N!5^?e+~1z8+T)sDIk<$muvrBC@yA z_;{OXMvr;<^!QSj>P4$uR=fKTNf7-z&NDKUmG03O=zN=TJa)$Z-0Xpc+pI{cYm$K8Vvr98cN)rRq)7Jo5ob+SUlnZANQ&#IJ35M>tm%VT`U5FgT3CdPSXGOd(8?CLM2kLsCnKki*qK=%{1^FuK0 zQ9of^>fIXy>lsvkNstrL7c>DwTTw3D7$Rt+(~@>ZM+$yxGPw zYhiNLeR;kJCG5&((Ed6az|Et9@t~)#pLxR?XT46+hgx729vakB#xrGo(5IIq ze`4+k!7|0};I65bYFeVdelUAEUr)RZkH77(8PQKK{)-_<1io*jgEr-%U;hR<=qEFq zBFv0mu!DM2EOD81RdEbbpFJ7dlyl%Ju8Syt zJG)jcpqV!)V1cS0HOWoD=BAt*vD-uSDiZj` zeizaD>%wx5GLvYj&siHOhH#q=%NCojkr}19L7e^~8?kTw0=QxWr`)l%rYXF%JSo*9 zJm`1XFWCD6126SAUyCaFUD3RPX0cJ`WbI<1fP>FA`RSNuQ6)i?DvYx6@Q`qqW6L z_0f&Aq``Al*x(aavccF`XIUsqD$83xI8yrzTAu0ab6F-1qb`H_2e(n$8bIN1&#Ue+ zR(>JTtF(|=w;?RM+Nv`83Fhx0 z=r{&aSBuS-qx?pHqcz;TvJg8Z*BbwyeL|EEsGTvWoiDC---(%Pgk@uhUF*lnpA&rm zNa?hnm{rsNU{;y`Z_MgAk-@~`^q!%1Tgv`af1U7WJ@w(jkDv7KND!^9N8qYH7iF|3 zt!=!w!bd-RiTB{_*xbL>B*F8!}PFJlj#2KH_bGJW=$YOu?eDmdzR@~{Kj^kpi z*)=Zclhtt$lw3}ECo=h0V1H$Lo#As%8;ifygx(Ud01Mh7LAdYZtH(F&M-S1l5pOiH z!@DqcjWbKPf#J16oH%UG8BST?%CKI0IGtF}ks!PlSI@YOkn0t|0v;(ZKs0&Wz*b6uuKIdTKlUAP}4J&Ln`_{4H=%znsPuUAH^B0`M z74%9Q*{Hwkuj9AjK+WFt!;4cTh!SDqI_yt~Z*9hk!#qDcsgHfg%!1>jg0w62SF*%+ z3wo!_5nmbS?~zeU%hkK~0aH=lE?z-9ck18fTz-diW2Q2Z2N$(S5H)LxZo`s6YALp- ztpJGc@HLN^`_Iw$A8{l4BhY4bfbAMfAE)*YR>9g?{m~uat-s5QO_1u?f5&**ANfvq z%i_Fm)~}9V`F?faI5lROE2?1A%3tptE+4dT#G<2hM}HY}ZSTIx8(vknwk@$6-{Yi-$c;)ys_?Vf2aZU=`{ zJBs80FD0KsmQ+xA(ims%jqUlJ#vF{_ZC$QJk*NEeVG`set>e6~mo1nuE#o!k6B83s zYt|hITgq;~o$GOQDL1R|HClf-5E9g6M!}w>LF8TwI9?kGlllsiNNRv@Gw>?{he~8XN9>TLM(Dfog<|(RdD2Oc#7rjv^;%p!$wis&yqLiZ@y8Z zxCKH+N_p(nR zZMGIyL1BGssWmpEO_cIBRFY)wkIaG(*VoZ|mWez;HqxUIFdc>fvr zwFb877Px*OWfg&FsCX>o&(@!+4nlGB7W_zAP9{Ox$`{gB#iU9ceUk+OB}4&S5Rgc; zj~5N9luQV{^IMQ07D$kECnt?VB`1%Mxh*$SE~M3rL!g4=73AAcNMt?$x`r&qYH$PB zZ*!%9p4%0|sOZuh%ur9}8ES=t$Wzz~qUR<}5h$HtKREuZ8LmMvT6kP-+I*o<(S zpd9zRL#&>nX3<+4bOOS~Px}ga%q+yK_@~ZsG(;6gHcB1EA3&Qc{Q!;y!8K zHnsC&wp}L;oJJmg6d-m9od|Ya?i!Z6{2TOHt!3|j%CT?hmvKifotzRkP#%zuiVthc z7PjJ1^Ua*5iIDA+ASSySrrtr5qs677l&5iV8l(r7W~u4CvHX)r6s$354PA5){kR(K z_He6AOaKmqiL=f_>UQ?}xPjjqSX?G#@b}8x>l7rpRP4w4%K_r>jnqNX_Xo-n>yWXG zftfP6UobwdW3$g{_(e&-AS&aimO?sOJy!Ni%^aYv*jbpf-FcOKwOl_o18Y=DE0 zL}!iKe?8O)@^tJr$~O}w;Ya~Ez6eXCkzVI#Jb)P$05I_^v8~v|F%s?3?vB&KNNjyn*DmPNXyMF9}Lg;&pW?FhK5rBPciRY#~EG$QjrrwHFO z1+jv+Ypba|&CV2U$4Koy9uso#$*PY1yvqKR4QM_Fhw$|b31XLptjCKc=1Ma=CbGYP zSkk;Y6ayrtIwB?(5`;mib#e8@@~F>2F!1MzG!0}PGV#nqROnP6PRAE&^IfW7rcbL& zIlJ~>^qm*ly{R0KnhnPR;M{H$Jfz~^P_j0*jg=@%$qaakU8nMBok^u2_RbVxJby)l z>b$%rG>`o@A^Dpy1vlBS%^<0nc8$?TGo#z%v`lWK=v_@*PkV5FriF=r<4WZWyV~-~ zE8^_GoYOb?wl1hStErNogclYbT}Ew~8Vufm*`8iN@9R8*J;>}#N%c3}(R1n1TkqT{ zhc{Cy4}3Zz@({u4zWqd!Si_Xfy^ZPS;#CF$i|&*ub;pCZX6^+VDy?zRaO21t)j{9AYTpU*6!lt_j@v48PgmSIk7G=22D7(P@EL} zoO%lVi>mxgcC^@!9oG3D$ZA`&$m^^G@#bjv{Q`(J`bcv@G8)WK!fI;NL1z^vHu=D( zd)9q$=q)Lem!|yq=yxPNDM#bqBEOoXbxn(ZpM073%51_*2+$-6Dym!{xW{Z=f<4T{ z!W|zjn>NOA`SKRBcp+wdz9{e|)|fIiN*9i1h-0U{BkBh@aj z3Dfbij4)BySnN?4b`*4nENPHezWBmpp2oq0X<&wT^R!&L4){HFr_2@1yh_UY{0{pkVZPDUdajMW_M85M3X-! zFIhIa-ws)xu;x`!{|%n{WnjBCPuz_@ziJ^kL9L>`x-x$d;(3N5WblgN5PAF`mN)$o zZ)_hkIc+vh7T_S%P{pZ{%Ni_#_&J9LMLSc}7C0WVek@`%Z^7VpWu>3NC&5dtRPq5x zrn%NpAtwmc=}E!Uw-5}@WA+s*N*g-i^=(j;S?ipjRqw!ii0Y)t%C3ICuWIgaC~gSE z1hb7vz<-q$W4WJGe~ku0p!=XeUqOlyJxrX5c;1Y}1L_<8(*D=?7v^P^Ruq6LRL`jB z?^mX%WhekER1J;=!q7V?jWq6i9r&$@$xG#uz@#g=tGIz8lJ;f0Y-0ZiZevx?CaN41m=>t+4CjKG63i7gl*P&g*GpFNKp_{4_)MQqeZ=XDSS797ia-bv_ab4F z)(ah)@e>_;2il0&A&#lMsnx5)W52T%(XfE&>&aK$`qw>>KX-=b!-SlD=yAcN+vFXJ z+}m4Uy)uMH=TbUSym4oXEF7X$N0ZS{Nv0CR#EZ~2=2to(YI6YuG{2t>Aa74LIttxP zE>Z7F5LU5|m!FjSWQ9bYbO7^}T{1GOL|r(J5{O@nbh=Y=Nl7Z*L9KnI zE%?TMbt7(2tEzv8xGk)YnMKM1lH5@N#6*tLRP7-yWd7q&M zf|u8bi;7YWg!*tA^NZAUlqi-;(S%~B2ZyYS*nw@7FV@!+NCGcZGU`t1A%xjF7p*$p zGML*Fh3foJ3c(^YYFyFek=)=8k^N2SLG>94TYr%t2q|zSg$trW808fR{&&hI{(qos z1NNQO>aC)2(M-k{YS$YRB|YB)Dpz@4e1IVy9kV}l8fmSRjG!K+gKI3j(I2#}>Ow~` z_dW=ra%F@+X~+A!nUm_TSu*1~NMT%c2oH&5T58p0R|AJ4NcO{Z2QTzg;(h8L>QE%{ zNO}!o{-jERFe1vUiMpSI5kHXh6}$O4{1iF6;Fe@=z3?2-&#Y!vcqx}Wp2MxL3{z{i(eozSqsI1 zP*w#8JXx8be+L@mjDHN)WlW`ONAM9Yvj@&BP=}ls#NgXgBW$NmO;KuxkS z4vsuL368|tlzTG5IQ(sN{rJ>pQ1F=`nu2tF!TREH7!$6CPJ3**!BO-j0sq-`JOt^a z;qMtfQ&bCH($+-^1bMR>DC%S9O^a-ADP|t8p1I)p;L2J|^pm>312_mY%MS2~SAyZ1 z)vx5wAZV5F3lj%R>I)0?D)s6qyYMnrd^0Da!1b&OMY_tUj-Ao<6u58QR@}TS0SuFQ z_2DmkulSLbeVX{=1~_GKhX;gv0GYu+w59w2j%gbD8uoaXjlFUAk`X0R>r3az`teYY zh*v&4sYp3dV7c}>KOYL8ZxR*Iq1rg8(jCE++>bTAYw^|^cV0Nb;>b&ys+lFA6@U0( zULobJbpz)1XE$B}&V&H?C(0|?B8mQ%4Byhc-58wFy4OOT72G_S;ssa^8w>YA?%)8B zvUKv`9bhLv1>bl|_)Qk~64d+@T5ebpBxjZMfTeCe)rwc3M&G-dgL(OyZ>i;QyUT^wr`{vub50ujV@<88 zSNy2ZZUF4LhSQ_lx6uU*JWi-_X3w(nc0(Ah{A8r8m{+!0d@Qg{s-bS1;#t!2We^q< zM7_g9&iaKVP`BwB-a5F8iHF*xTp_Pw2or}m#oO$eIM#YM_BYgBr`h)_!Y-7GvU$fw z>)pI(t7La9M7jtLCKR*{YP8(o$C}#;g*rGRM#^ta3Q5)bKa^Cp`L7PE5u`ju?E6Ijd8xevnVPpyKAewtJjaHAxu#Umllh^lj$l8}-mlTo~U+Ze}B+0u9T zPwzS@<27O8L0;S@STa6FbTRwz!tQka1Zv$}PEQN9+sI~g5H9BgG1}>Rr&%CZQtKnX zAGx19m~Gd^gjNIx9K1zD)wkTl?a8_S>dMJ|itExEUfVMWhET61F?((Y52c~3->2U% zyf9xpRPAjS7;*O^L6JtP36+DEQY)C!KSKw~O?GNKA4=F59ACzoXdb7#uqRGK(MPh- z;%c1(mf6N)I3Db}MJ>nfs!YcTME&IZZHkyDHWT}p)dwe1%f-)odZdk#(k0?$3k_1f z`MfR3<=#RIs6Bs!Y82Y1vGbr*1x-iD+y)=I4DX5k zBV^5NINvXYH*0ejKwI6D=N*xAJcBbpkP6k~hE>E%B?X1cp&Sm@TL@J*VU0cAaf8Pv zC*vU)J_!s$81$HRe2y2R2i+wQoRm{li)>Y)N%3=jRu2_d3=Xdw=q8P>$q717#GSw{ z@8;nNB2-2sikwy}CCG&usG{^0U&^RBz-4Dsk6>1Cet6`L%CUfGo3vD0#+$j8+zC|a zQi`OIlr_m;yRz-DOs)`0PQN24LaqJE{zQ3Q7PHQuI4Ip#toKvDvYMSA z*9XT^dr-fCA5%Zlol@On4}9$A_Mpv-cTl#CM|v?Kb`iT9BK=buU4G->_ND8Nv4EWy zOmIm{s!4A3UAgB#dnmmMU4rCp^x@-Ykv4O&c ztR&Pu!IIkwW!3KZ72=)nd8oM;ZjofNLXZ=g4!UGql91Z&U;PJ~*2pDxS;B)WBov)T zr4`geQ6z4PW)tG!EZ5SnRZP~H-Z)Kpa8BhrPcy5(<3X0;jdO%#IXFM%trhp;orAL! z{fHmPdIl%eGU%CF0YwU%637IKG}NduU8ynw5_+{yvC0DI2-VLhQr{~tMwpjlEUe|@ zjaO1PkswQ+$Rd@y(3CZ}A;g<}#y#?oVfjGp&bWV%eXhD-ncm@I_3Y)P{E$^fzfP$< zFm7{gPQLoGy{0)EB-T6~c zr*Q@t%1sHvEtw(_q?Aw%yKYyXJ zTSwfzejHk8m&PVBO9W_7G|dJxC(HlbP;3^zwtxZIqNUV*=&689zvDpR34T{9`EtUx zg7rz_$p);hA{^@z0Hd=sGt_j`;#}S$v~dP>>v)u~dJ7nQei()gIsm8#$r~}<95Scx zV#6aHB9Dq6$RlmeR3eWzkjDYY<6Ao_1BXP(gA|j?AMWied%sR)@35d7*vtm6`+M@9 zg)7|1;92_5RRjk<)(HA(4+L z%%ZIOcpBvFHzmZWD%}}UTS3PaT{xA}hGS@P1P%Wp%hj0^xp#=t$vSKXNew>q|4O@R zCMJKCweC77T4|*D*OiP8aSHlwA9NafyZglSYnb^l{{g8_D5c5-$9FOc4^H5@QK<33 zbCvkD6siwhT*)j5BF%f?UB@XtkkeCGXLp~|66AC(inr?&MaYtq3s7)RDJHVNAUR2N zZ-Rf4<*?&aAze)T9u#oa!BPh%c-WubOu3dk;LOHvcG#(_+UaW^32&qC=ndYr(oXQ< zQ1d83x|&$&ln?r-NRZU*?iDwr!7qGhI(Cv3{0L4NEfG{Kctuaz45H4=W=yH~XoE$p zh7=w>GttspKSWbt5|Xu3VWCy%LL)k8wX4|2C}9-Qk8Iht$@t+Cr=-{FG-Khs^bbUe9oeJr zmda_KK_AAPb9slxou~7c=$)Hn{EHLy_z(jhW>|N@^o8LC#VP$R5 z0tqsS@|?k~glp*0xMAl>MCNseYu|YZJn<64H_ ze+@8dqcEIZNSs$IuHlK|Gf7}gQ$w=Gt+6|X6A2FUgNljNJ+hKqH!|4p=}X^k*GBG| z*>JGK8$)&31eu^FQzjpGod%fpE(`0ewtjs)QiYshW?y;H8`$pz*U>ELJ3z2*(G}2FM>+x04bj})bJ@B zt#z#%db-m`mc+yDrxx16p|>iXB**#GgXq!+xwmNdRF<9^Rzho^iNzH4Zx+3l7CCG8 z0efH3giFREp@q2D3zENPi$LVJLd5osZ}Ezk2G3)wGrP>LFdyekHNCvQ<~HvQgdhiQ zJpr=%298*lKFZ=$k&An^eckyN-!A)HrMqd<^p8~TrD$yUn~_-XWlT6b8*lNSQpfzW zw+Fp=fzp1=i`CV_;eAN{_&8@ujOK@rn$+jp;$^v(ac5c#@I6P8ubZP!##~1W@1UHm z*$wegR!2UTV_Wy0IcQ-w;m%IZK_-4(>a#Vc^Lu@TEft@L8;|o#gU79K88(}@p7uoF za^<0@szY6yC_&x7L)vYAc0V*ql#3kZ8NIi9#Q3&VlO&AK%+FPNg}ZzA|Kk+G=Ve63 zGy3=_Pmh0zc+{AD)LZ+A`LZPO7&iCLdJViol?1pL0G@GePHR+r-y154-0TNr%#1cM zP%}GC^tygh`_7Z9U#jxjDM4`MBdL}CKz@B9mi$ven0Us3h8yk_&+S9dF#w;`C+?x$ z3JhSseyt!*nMH_pan#!bYxlo;{%=ryNzR(XLd^1av^g3zB6g4867SvYu~P%W&S@JG z1a5?Cq_mQ0ozz)7zwHI<8H6^Z6bXc`P|Pj0x+@t(XRC`s?Zz8E-G&=E2&@#7-cli_ z;{&MBs1 zDE3ao4e>uCw_og<*aMlJa%K0bLTbvm9J^y9L7Dn`()Fb-7aCvW2tjGO+$vX2QJV+W zbw0h@wY6DEJBj4ZAcp)E=f?@FpkLtRRjX!uSquW&wB)Ol?P(l6HXcfK=n#n6W=|9Z znX9ZNETk7nw}*DfrVovC2JoB*e3*N2Hv|n5*OMRq*6)&5G{8?_7Yd~py4H=#rXL?6 z!fJyc@>J4t+WUoTLvx%-Fx*V1*(-cI31Trxmi;G@x8aL#y~aJH7}{OQ54Qd;?|&Ur z+Bn)^3Gh~#AhJ}d!grl6*-pLpKKSCy-d$i#4_1^b)kLWM6m8FiKtU~>f=#5`o+GP( za~j}x$?jMK7)d;26HUDC^I8!ZJWmk2{P^^1S*H?oj=Joz=}GZY94fNwb^hH&=W41o zc#UAqF|Bea5}*;&`xm|)xP;c9WhM0$e`E2#8aHCIA-}L;%<>9x85}39Lb8mB&t)Z^ z_z(qe`WfT41YuGsG_;PmqFv^p51$NiZTJ#kw{ifEgfdK9&{A zS%L$!Ob}C_Gl?hhuIE;h6)+}(jzMc2l%w|?cth=aMK2Zk+>3_GE~U4EA^xQt(tB^F zJ_%`}=UFMkrl_XqqKi7CyP=pPVv@BVDg?ywhVjmN=%`Mw2}9r>bmdO+*8h<2{% zEII~d!v*B}h|u|50@LLBv*K8Kl>?8SlmrRj1CW?B6(;ckx+`d^6mpU}39SI&EY@8B zP#9KP23@vrwu*HLVbv~FioihhpZKx!v#C`$XRzuc)i+&t2RBkV*x^uZrLFlxg(>p` zQONe!L#wp}afGVUpYZ|*cPE;P_VxB+yowCIoMOu1$ReeEA5Vo&ZWq5^p!;oEr8+_o znf6KiziFmleIoK=5`21h4|4<^1?+99O?l-ZN&R?2o*~heJL?J+_|)16 zE!89?Af*b25!n=pl`DF6gelZ_wvBMC)~i-9K-qgEOk&B|XT{fWS*>EwS8!J65t5=#8(xJx-V1n?TO?iMrR~$;i4LxV80BQlK6H%tRH&)Q zi59OB1O)r|63gAFMDx#5dfuQh%WQNjSJdZ&yi(}C3Tcp+qC*0ep!u#>`?KZ5gmkHF z6LmQSWg?9z$Ozw^)Fr`ehO3Al3=m~ zaHzteQQXT**T?I9Us$5=9H`xM;s#q2g5^w@ELLTS8+iM%FKpC`7Ja-oTcdtWBiTCAEUw6;HZXxUgWz1TY{# zfEJoViz@+=TmQglk%s?Qs%JD1r4+maX$W_NQSQViA}OTMY${BeXrkW_wxT^+Mc%7_ zZea#>IPN7(mBRGmbxUxChM_F~dtED^FcoSksvPi0XsIy2`fmSA_-c|Uq=HH70Th;s zq%jC;>w_dvBJy)uwSlREE?#wytaa?s_ zm!?sJmkt|dGO@0;P^JA~d|yy!Rwtu9z5wO7K&24}fY$i+`YvWC=D(DMQ!E!2+pu(7||t1D>~D81X(pR_6be zE$T&IqEDC*8?L`2eXC!I#p+5|U+}(#*u4!;u>gy#^~6dx>hhs*ndZJnVe5|f0b&!n zVikPM7F1mjk2@#$f;p&feUJWyj|L=uOoOCO*9W*P;|*>y2kU;=qSKvM$k%RH!c~Rv zuc@8S`Aq17f{8=Wj!x1-J*NK@ru7Gu7Ou^mA0a^ok9}_R#+SYAttRz8-3Mv)A?o3i z31FNH$ASyyP+`KAycw*MLuvk=E#xQ}x7WTfXr>FFpu+TtR$`n3S+hX`=(^ipm;le> z6Q&f+U5jXd*T9CLC`_NBNc%nrm=pqmXcJTK+;!NM04Gm#BlLqQ@w#6>*Z_-s zLA~g_$Yib&dM6)bE&kR|y1^9Wa>BK%oO=W&OF% zd7FWfT_8yg~^@7ur>>{u8$R(!agSBRxN)V6K48KLmIZaa)nnDqn)#FdXgL2eK* z+o%yE{!&}V@;flTz~Cv~N-j4!kiIAKKn82!0^li?4Ugb-5RFza8qkhM13`WX(O_49 z!VV_aM_`4}8czfVQhAfKwBV(E#j$ukpm|~oH4kQN#LEUekuN8Ii)d%%p?M3mr7fBv1n{Iy zvr&~#6m(zAfzeroA3@MvW68&$B~+S+Zma|?Vk~P01IxDddtG~<=npsv}s*narcW}0$Mdg+~ zFo4mmMAGy6*;Dz2pa-@%v?}X-!_d0;z5=n3As!y{VROK{u{9;s8jrpLRxRZ5ei+7j zrrCL&POtwwY$UW`Z76fu?2-OEz($lvxyArm(+b<6Y-$~3>&XHam(2lX&N?~EQ6_Yb zos)Hr0n8`&SccVSAxrGW7~I_)P^4kXkE95CCCG==R_7PkCrb%ZfR`-(ZRLgF<%Dnu zi#Dpb9?G0B4-66!Lr!?nlbmSew+@PYyUj^r>X0K^7Y!6H*(*}xUpjA85vryXBK9{L z?|oXQ_aGnB`JtcefKf$qwN9^i0gLY=tFP}NxQ^Ilv;iO&V5UK z(G#aC%~H;uOiN!>=iku)2?4vS$vvKuTG)p zUKl>ib%S&@BEX8H?qXdc0-FJmT$>TnvO}P| z@HHal`<>i0CyDx;{Zj%JSEMf7JuQ08x1EGs-56&xjpCk_P(TMcI zj_j^DEuVSByuOrKRzCAy2raL`ga{ZS^lPKtufPd-QPm81D9Z*WpP+beKtu7mSX@k% z2fX5}X`G09MxtM3Gy0lj*+E!!>hULgMVk~l05>sNgDxmXVtpy!6*K{XQIe3EZ~#fR zp`E_ci$fM7>*K(s)a6Xh?;xf6$hniCMZkbT{S$@}XDB<%9ze1i?&8-=GC^w9@emDO zJD!{|P==I@2t?ixr^3YGM1M?8^o$gBbuwvq($LJ9nlIM)nIxrX^M_f81GU8i%Re#K zTpx&i+KGOZj)FCHmdliU9IWmGgFp6G;>;cRRem3FcC_z%fKeNKa7~;NZ&aZef|nL_ zE&linb^jwry3Gd~c5M=u_)(us^$k&;T2Z8ydz@V#g>np;hNk>F77R;;Le2KaqbtNE z!M&JHGc;PDh2Tu{?1qz>#No@-H!Nc1&~wnE0W1F1`7@-E%008BUo(}SsFNH zuLK^v&v{~HC%p?w$5lWp0i)|ByhA%9e-6Uj`(uVM{*)lDENG{g}qoq64*})ky8sm>z2GbfTFnSZ34rKVmFE)fBu_gO4os}lIguvixP5E}t z#9Me3^@f3lZ~=k)W6;Uy7vY${vAB~M6T8arHJU?bi;JO%P)?oeM1dh*&o2iL+Ne&N z!IvZa`uu|SwnKMOX&cotBj>xf`%ZkR&$(*2+Ti_!oqc~T6x&TM`IP+JrO)jxN4}NG zywBk+ogs}-IIq*_DnV)ti{ZS7QC^oCcMLoI=jp+Hdt7<(7Y|JP;EK8P^g|Y{!b!w? z5ifs#8WkHc-3UW#L~khVXCR*i=K}3-UfS^}11GV1~d2sDUguhSGOG zGyr<;W1*bMDm6WLlX9|}d(i$Fx}Ei3qpPhO-%)6{kRL6Trf zXFtkM0Lo8r@~wGmMbZ%nFOyj30`;}mK{a67Hq*A>TNB}}HI|I0PXmPd1f0)VKS&&; zS4aXf`xf6t;U*N;WloSU{4i(?2Q-FyG~f~;kJDxx3|}m?rDQ*ih=0%n!u;cV5o+pk z$+)plodHo%2Oh1R>6j2?m*9a>K{{qpbi7puK4ynapzUoW*<7k1d7}RiDBNfJv6>lu zG*Oq2r-JO6lkX(0Zwm0+wLag7pZ*WEWd;ya({WqN9giPb7j<&$mVys7{`0q)R0`{(K-VK~sHe434EMJ8@)7 zf(}evt1=d3f%#ZLzA~p2vxHkm#x_x^4@@wzgZODe19~(NdbDIjvL!salD#}O@Lo0m znhk*7mRIokDWI5j&=3lsWj5cnXB<%|>V*6nPHDmoFuA#OD(~x`*&XG+7rA1%O`rS= z`%Zpaw0JY?2{o@gUg-=_rs7GEXC7hl!HY86&_MrT&j?`)5lC(pMzB8-@HB_-7c|Qv^QV>7jdfby_=Y|1Ol@pY8$G`l#nSX@&Pw_YEWxckw z*V=wOroe3H@Q_W+S#llGaY1lzo0Sh;_EV+d^(78<4#IQ7r2E}TCno+Q zE(qDwui5*ev8coMwX3~@bqVW%@BOO$?^?ZIM`H)GzfI|agHP;qC<=qra7hAs3r+Fu z`I|@SDo*9fR&Y}urG>a(a+m?j3BcMK8U}m|`^#$WvXkj@k6<>4-VhoM0>K ztaP%y+wn!Fw{^=kakK|~tJQlLC)p7%oScsxx{a}F*90eDe_@p%yfd+JCDqGWe3?>Z z9UEWGQhGS8o|wdKcRJ94x7Aazp+U^1rL2zWOfCd911gC6rO^vpjjPU5|K5%^jHNVU z<}dH2=?vDnkze{zv~7*7TjFo+{|u3z^JasR;15T{=eLOOWDDob5wqsj3OQ@Cm0(Iy zOxn7;TJtVz|5kJ}U{*ATS}O8B?Y(lcuJN){9gtT^m*p^}&p2T(Z$sE)JDSv2IZ!?G)2Uih_@CaXqtYSmA8@IBu%kI{neC(WCQRg9ySR2!? zdmN|SSC&jy;$`Au;yCrR`$-u24Pq4wU3YSUCp*Vq8b@&+fTqO4u=i)|nTwz-0U!4* zl4)bS|3pZ17941U_9T%QQIhC+iLJ}G}XmSa$h z_hAsP|Ino;*Vt_nSzhR;?5Yj5WsuM(n>EO7?W2|@@uruSx!#T7(Mv*qJtiHNzlY;@ zz#zt0wuTEatp!qA3GihT=Y+NQA^$3aSE*}Hj-mtpT@wZSaTHn2b$DZ`oJNOx9iUTO z2vIQCs{Cu;r!^48kg%Q8~;@8OH0TT?$}+kOI{EF-P5OO|ha48EQN_>#T5 z8K%g~>Tg4}W$a6?kHUxKOD&wVJ5*jpq^bp9HgQh$%v4UVBZ)AI;iWNTGpF;kcSCcA z>~%OCD9vGz{Va9<5_a!FaRGWW`aTp7#V8jKX~?A}`CHUdTvQ^zkSRXeq zKA-+6duJ{5C(22sX!Vmr|2c|20o_xDUkX1t4QVB z4@S~+^lzU!f-{8TtWbC_Cd{?t>l1 z$1MDJWl7PZ2zRsha5`ERFuD50skWGH1dLI7Dnb4*SMikl3xZzW0yGDv_2=T2j>LD_qm%FUmayF%V@_$TEx?0+97DS%g;qf5qL*2fAi^X5_ zEkjmi>^f*54D!H6T&0~397bKUEbn-U&XEn>>75HsP78-y*=QklEPY8QrIsf8HJj5t z+lO%)%01lAZxUan){Dlxx0Sbk?AGbi=O00ztw?5Gh&=MH8NX$c@7jHRZzX$Nk<`4O z5;QdEHiVCkCOOY_*?kBzulc;TMY}IGi}C`E&^!NUuOHk|+daMUZ!eG*5Y@tUSp#@+ zEzLy~)>o>1h@1PB(dW3&!1c=Kz`$bv{o0;%xLpm{-|emk7XTpG&^71lpxmf$p)b3^ zr;I%TGvBcO$cIu4eSFnATB*F~{xofk4R5yP9%MPus~68eTIx{8JIYIh7u*V*j~4cs z-I6=_b&Fy(^I*nMO4;&N!;Xl&VffNv4q8;65@IXz%6+JLCkgJqM;16roxVFod94e1 z1M+5p2m{`=<31~+>T98O1 z_wMlMbLq~lxP0OGUE{14MT@4W{Y{)G_E6jZYNm8AXr@M?f6bx}*rxtcg6Q7+dq(vM z$jz!{5hYM{{WjiQJ)}Z?DLo=9y7efYzG?N!u)u~t^1#A^Jv-^`@k;-OdT;VObk^yK zCRDO}?-lkfn|+Eurs`;6bz=!0h0wc+7IAN44v?n*AiH$*hl?^`+W=jo=anyNZ;S-! zvxvTi{!h^dY*UGsASXYrp69!$8^%gj7O#V9W|E>brGMLuhoBD=ywfh!%*B0_<(QlP zE!3Jh6(J4vT1~+zAfO7eAp2>F=0}#}-#Hf1k7P4*0qz@>=*sy;V~}BaMr^mnKqY>Lal;y+Rbi`5`!kRZ@7IIsN*XW zq4&1(fphP4^^VLwGKkW*EbB~X_dVxGZV z(h|yB2L&Inl;F%k32fR_qs*&we=(kEs8h0DvEx+tprxF%p4HELLB5=U89m%=)>BUZ zsvgcD{=2UpEHXs0B0&vY0Z{9ny4(>K&+W_kiEmi^MqGa~@Q6lY5aZVZI-d4U` z+F!J3WyVTh1%5-B?+U!*NpgHa69CTB8fr_2?W?ZXn$Ygs5%p)}F3YteyL?*LAN66LhQzryt zgDwX8Zeink|2+c1YJqm`Ca9xJf+D#jNO^(#8=`$IL0w`^s6F}ZLU{XO70nbk=j;LAsb0e?9r~V3U1((& zXUnY`vV`pg#k)O4Z@9W&bRC=3hWvVPWOu?M1NKI~{!xCj!j##m6Q+&5G+;K><&oS0 zVTGvIApUmRn{Mw_ZPVg&)Knk(34gxKzG`vn&L#ot;N>~j(|sG8SfNQ?=JgHB%j}nx zbN3evO}3B!>^(%b=uLl)cHO0Wn_l>6_lEGJU&$a7xOdryW>mcW!`i8|!#;m(X6i8iwAkUBl-%{z6l z7P{Q^Kt$Kvy%*zlG8mR#PZ*ZNBdF53jv7PP+8La?rr2ByWB~~jh*iV$IsB?yF>gEt zO@^-PUtrwX9RLE?t>uOPU|@T^MA?pk`CUF|BD!iteT{n=!FP~$SRglts)pLTj&;Zx zedW@QW!sBw2ljB52&%Fgy~i%N>#DDz7^ZbV=dYk;Kl&~Y8LO9L(xDPGDOlEW&fADH zH9LOtdRS)sK<|OYhJ5YG^csPCk&USbE@_$_C>{67W4^`wY+&!P^YBPqFBf))HSYDv zkGdDudur@^2e}m~xC}#9yU!T_n}LF!T!Cs54woV5?*Ln1;}#(Ujvhp?dfFb=N1Pl& z`sE$rH3$2u?`(4ChT9ffv7;@pUc@qP#7c)yRCU}^4(Bs3R_VGL8%>&j^DWb|7_#my z@dm5B4VSr`&deQR^19{qs>HlIjDhu@`37rTaan5n#l72_9UB!HQBj`hwxe^-6PJhS=Ieq_^zuA?$91Cpa$2sdJ#%2oR-4inByh=JV`rX~D zBjtad`@C|*j6Y6h9W6;pI`>qz==MX)HIK%s$4;yZ4ePBB`>RNMsQ11_@i8k8D_u3` z1uM+IH{Q)|rI>e;uZ3t`@ZdIw;l$1cWA zC5(pI9?H6}n=2Ta)T2%K+eGpCkmVi@^H{g`bcL*5PmycQ3=VLnCpbL65pZuuiOcpG z_dR%fPw#qoDaCN=(;%A{%Vh=6I-0iPx7n;GOxtqvZhF|m3CkY_teIn)_hKBoF3MLY zCPvT?z$1go`uGn9epD3KVWF~ffOnyn9EL1*QWV#f&%9iv3;9!0x7Zd9*Q`lbRs*~1 zD9)@-pWYB=Dqt_BbVY5=J~ZPFXWuCEWz5+J_(Pe8PyW`gymk*qULwLw&$#3r>#+BT ze_S?rZS3#7Umvcq`f)q^@P!@&?^IjwBbLQ>Td#Sg_B`Moh)vx$!sWH&tl5u(%qEDc zwc}lOr)~{u)O=e1u{$qnZQk1tIlk-!c7v6(j&kJl+^1#DmqyPIlTFH>rfIU|R%a^D)%)6IQ;vX{Xg7KK3BQaWA$&7f9O_>*GKjuA^mnq zJ)l7h0G<$@H+Pi}ySc9#bjcLfiUKeek7y?*@xQ z@sCibXtdG~SAoAOCoizR8~Ayi@7XuLS$9(GxSI#BKIit``(?6C`=8@Y+GCz?#e&8@ zh3Dh9H7lE*T%U2KxUE8SU4Zcq&AfTCyM5)yZ2ho4_+s{vIo&#)1BbT0j8V3*^k78G zUGSanlXPF{(fy*5X%CX#g?X*YQ(ZHMBsZCj(kMC)6QF`a<`65<&Y;}K_8 z9Ne7Gc=o2`3S8a+B#{^j9=nP2@&*nN^i(}P#$3nvQ_srkY;Bf!ZJ|}*b_j3zxKuO&Q$o;hIVg2o*`{^F}o39Rj>@$0)$;zE&pHqLwtH!47i6GwO z5Azttc2#o_%rWiG)Yl%@%SoVIz-gy7&4U6t?w@Y};}1PEO1MGcEH{ zr%vB(H(&T9-^FUUHg5CZTQdKUL0!|r*5c5e`N~Q^RWlS-4lAO8xi^rS}-s2u+T@Llj5<}Do6j=?ExOA0LAxNTmInY z!-WBwKVdg_N?dT@!sC}OUzn^^lzH`TYjcUu+z0nFig!M|TCcP3+LrNGzl4tZ-C;+t z!LK77Uw9@weYC=SQRgwfd$IP~1eg2ob=N%5QP6aGvUJHZRi~EWhqn^ZTwHY6mDTHI z=vrgoPc3g##gU#E-}PDm+^TwV3|P-6seLb$pmqO2+ve52=IV9~G9kM~EP*&Qihjqy zLh%wfRPI@8>}u}b^HvZ4`s>WnXYb1gicA-IT1QUV-}6K+G+Z&&rMquD`cgGH-G5`hoza8?#_JT% z{%TsM;=3oAvD?$tGR=Ga%(#kuNoEJ{>^Lzvf7c1);!~Zao!YZ!V2R=}uIo`=R@lW?p)a`$|rQ!xEsdxtoqxCJD3J%Iri zt0ogj;nqiqPu+YckzkBUb0~l>4EbQ56RQJ?p^}Izv5rGlF3%*29r4-9} zprivIF|6Jg{72Y`2CSYI;|m9H3T)9NA+T_QfEwbU0+8ByI7vaIBw>jeybg0Fz&ZFW zP~dqxsMjN@g(Ilnd}m|q@OZ9d8dZV*6NRWdf&#A(Cl3-BXf>6CYK=AE?_>b=_Usr9 zUhUHc(IeeXZ37x%eAqDh`~NNaKYVAAiw0F$2nHl&u^=?+4IWb+K?6Jz&~9gdz{tWo z3a|qb(Ng8#34k{xT3IVfHHC9QU<<%Y2QG)B#!J9^5CInj5yD{H|7Qb;1iui&&BjV4 z9Y&yrixTi+C!bhP@~H0uq7ak>(CdIg41&!P$plFQpyvv-J;e2b1;ffvUx(%iv~ww( z_~|<kczN9LOrRJx5hZTgpg3X}HIgxqP`z&>@oZ4&Eq&gW_}LVnc*r};5wx`>PT6`Nl4;Pe7r>}q=k)p zj9(nnpx_+}gN;t=j#ola_j5#ml*`z$-yIdC2`*1-VCI;ANw;kBHK@n;USo=nwSdGH z1d4L}I;&@MD^!<2bCGc)0K1Y=jzI`{`Ljyn!%PblxExLrAl!|?XLbSq0SQ^nB;DQy z8;B0B#=_>u_*34xii@%tZ2WJ<2S=uzsy7S;oLIa}W(pw-wpxhW&slv!`CE{BB*WrR ze5T>}541p|+PwEnJ?r5L<_7B>uAt6YA}y-7h+(q&9VYN_UvsF<=ClQ5y}3hsb0Li4 z?fk!W%e&hg0jE*Ofugj+I6E$E2%?cvJBVs7Mw1Q^nkOv)&6|x6S$ddR1=|op*{Dwe z;-}po=5;AcL?#iqpa=|fQl<|>@;KM{z0gbvu?aZe$q^Gjv3~doI!tK?z$?9>)GTf; ze?&d2r=XYBRp{J?E6SD*IOsczP>{BQ_br%kqGPiJ)-3tJ3v^%K%xz-oqrFE>Qpn_+j&{8T+FBOOi@h`H)kSoDMpC`}B6!@&(L zZp+#bzW%RVXzM|~DIX)8uU=vMa!z{y)pkohgZzos+yLUK>XP0TZ*?LNK-LE@DrXQ6 zim6uc8vWYaU@VogCT%XjN~iG=r+rX@EP$9U>Dp+g?JCy?%^>mgV@woTkQjam9jI#W zkhG7k_p)CjZoS%bae0Yz)%v&thyUn$KXD1PAxglNlh#A37cFd_sk5`5#cT8)Mj8jX zxQyyLn|8IH^^yv13OJS(X`vStkOR7~ugUte1xedyLSJwGkPasEpUmx4xHfoHIS`)^ zm87O(1QngheXh}jk3={Cl)*nP@HfPWu_`1sw$3X$QkR`sE39q=O2wHf`yI4y%6Hsg zB(GruV;8LH-Z%6w38D7<9*X*jEk(AT+o##rLlI*KG*4$UD+5J*Eryt|DQ4~|uZhRs z09^N;H=#49FtxaS@1Zk56PEIMqFQ7L`!XwCRoLsSIRdcC-*}@eig4(p#Kix+7pSHp zP@o`nc!5j27?2Yb!^ycbAAxGnJ09kt-M6v&XBJn^|Bp2FQa1ba#%s9O+hPs9YQ zB4)uWd|(R94q#conH{=^j{CAoKWG4ZzA1SDV-C5k-nPlaaUg;Z;mGg|?IDtlKc!x? z_-_DI2%svSp}Zgd`I&%N1$)HMk&tGfp;M`>6?KDXVj50)hg#f#I{yoR@)}TVm5C(y1P2#M@u5|^rd?>AV21ZF1_eGr${>G4 z=d}93=)PR6WtfmPF%*EP5o1y1b+UUKDgl)OI@fx((nN}}BNBrPKQWluKa$0aeNo?ukK2GR81JgR$pC2op7Hyr{J_VB~JPeN5oB$JD z6Z#h=cl$tahuXv2`7hZ3hS;}DazrDE zhSyVG3%fAGCyiTX4pDeTO2Om{M~`)5{a3~dI1c*Oa>olx!Eb=A!?xzsb{H@pxcR7w zNde?;V!vbOsseO6?9~7m5t#f?qkw(6HtZdOF-oj(TyomVPYvNHON$F|h!jq4D1Lp! zUX!9OD7LNlrMI|_02FmVo#@YxWRY$vZf~7%3O@V%jk};HUbw&0-Nn)Y$vT}c{MEJ2 zNko{f#&+f}k5E^!0j|EW3^ONqb+trFwMDktw)am#!89mgAB2D?QfLf$Q3fja zD&putJJ6a1VC$wplEVd(9P)8-TWEAv_efWdjCjhrm-QO$UlnNzd&9=!&`qQQrWafQ zV+V@X?`nOdDAgfz0EioBs7fT#>7O2ej)*R>VzOb1jEc** zPrD6@!D0-Sq>!}UZ~9b60Bd2FsVE(@A1%s$XkfaYY@jRz2%-L5aY&=y=hR4vnrvq} zIoz;7$qLVXH$0-qEI^)cIsW;G*IX?wN_R@lZwT64VCz+*Vs;!2eYDs}13rpkO3HB7g`4hr z+YyULvNqgcxC4docWH!MCM`Rgr~h}so}{q#q5FSk84Kiq#PqEaGc=R~T}rzBRt?eN zpoFo&9ZFVP*`0AK`@pJlt#*aBb@9jamG7qh*=0y7@>@D2-}HWg+K|3Xz#lN9s;Tt5 zyAgg3@9AyL=&}udz$z?Pe9Us#4Rw}*0aMiP3Rx-(D0R2&g)sUES>|JnlF$6orf+KOKo4BS2FYo1B%6lb+X`gF@bKJU}k6 zvKnY=#D}PI^e9XCrK-kA7GDL|hpzjAe?pt68ATm@FUqk7-hKPzQNRddO1{$A?`gyb zun@iqM^jnM;A9ajX8H<{l;48lgMPyK63IJ+hq}S8l~uud;teG*j>({mYtrMBOl#?R z-||Erg1Fhq;&H+$F8R9FRa6K$jx&(m4wzTu_6L*@)*h7yNCvBX(Jj#xFLL&5{iZry z`69kQ6b_+DAW{e7uQMQFOz|<^X5n$QWiP6fXvVJ3QC zGV!k^e(Zv#)J054FCI#l$NIv>&m-^POJ613_{Q+SgP>wA9RHJp9AZrC!!_1(B&_zO zD$rnM0{rdRCddvd(My!$@=0^Bou3`pI(S4LCvl`w9z2n-+X<0(fm+C4KKUz7DyVr-=miR?Cwih0b z)-GQs$EE16=c3QDXw&Vt9D0jF5l+*LzQmM=CNJG+sI#Svi7}UCn~cxZ+rA7KD4$dU zN`zP?Xayor1(P>dQeGbl)(3$DCapw6X5vy;Z0mIc?0I=nEYEikKtpcOdW+(n^t=Pc zF;PY=83045BwZC;DUs7y1=!=QdR`r0gz{gQu5`OtN1RGiF-3jZ9`n1XNQX-@1&b#Z z;8oPp2{z^i?D0pv)rLH2TC-a9CpRo)1yj>wQ16Ecdo1p2$oI*A(T$-80?Eadnj_ov zI;^hpKxkh~$?7#OPYy(#$z zwro?K2I<%T;(rHGk;5wiIZ{ROEjI#SrI_1sbIAtPa(5LbN`(TR8S2z3cK*?Db`F%a zunaap(Re11!j2UwsnSf6o~I%IY_r>qk49RMxeQQr{!{cp)4xQ{rf4Z#i}3vlpr)O) z1X&^toEK^aAEHYRJsOIk5#tSsZ`hp57C!s{5lp_CRS6Z2um*o;?&4gE#mr?s+z%37 zPu4=)dhIbU#D?VG2mm(J$^&aL;k17wOy9YuLl)8dr`5PfL$X_1wiqb02%2k136KY% zbT|leF^dX=;!Gikb}DE`~zM*Y>9mz zV+KaIZg5&TA%|G~=Id@Kxu2_Z%+nQ;mX|laTHt@8f_HcC)CrDDC&@7K+TMl9*-fjd z8c-80{mSrx7lGAtT$`XFSKF&{tf6SyS;winE;I~{{KfN#XdUYUuL?r%H_#ANDX-B) zkJ|yoma=5EJjQ_R&?n!eG>4%y-QlPOyyk@gPIpE(2A~bNFoz-8jYOK_E1b+6*#9Vs zTXB-js3KaUVOXxy2F)EQ0Bixt9d6%ekioV*0?Px z0#D9htGwvKmd9`pUPmW%4r_F2&!K+RlEE5DBe6^j;~j~-w^Mw|$0iQ-f~Hrwv#K26 zd|L+3|6j(>^knAH(k`(Ql!XhU1zsxw@Bx4v-i{UjUQB~@Kyynbq99ahggH|HJ`lC? z;nk@V_Ax=UUm57xT+X!6sGF8OW|Vu{!0MJ#T<7DrQJ4j&1^) zMN6WLCIY*J43t0xe-YWDnZrl{D+nb0zi5BhdZ?g@R;e)*K9G%8oDGtBZA>9!Pd=WN{x|sV}{wwK)9}r zJK$l+Z7+JJd{u7HohKT~6fL?-O9#&zP!5Jp1^cgzYB&J=*gR`oWbc))4D%E|?3NZ~ z6(cEIY*B5)-%w86QI#f<00vM3)uv-$^!om0A_VrMU#38RusZ2uh|?Cpl8ebt9&3=;}Y{) z&Yu~Dmou(+hmCVwXO^pJeXq!+=dAYVW9M})YBfe=rW`2Qddp|SAHUS`OAx=08LaVw zZIjx@*R&o{koiGWv*N(VRN;y>Z@N#yo5W{(W6cABFVAl9>RC3uR97!UJ6OtOBqqZF0$wwkdRsb9hBA08a__^AU!KGSaIJNx2k@ zWoFg!9rS*fs`*s9vK(j+se}Yb5HJmB%Il%%Xd$f zS_FON>2p$%P#wT=?^!c0@;VC!I)DxoWFe*9dqiSwkHxr1U@UUMbs_Dp*vX}ll$57s z#(KdVqRJuiVz(~wC(bedRj-=2{H!xUIVk$sqO4NkV4#FmZQM>2MffgpU+d0H8dLd% zeRDEc_CO=A^=%b>vdZ*zS!!^~nV_Jr44628>4RCbT48kL$_?1%63<*);B0RJkrcrA zA?$3!R&g2b7`@<2Opr(gv$MsURVAfiX9~s1S2N1b`YEQ;j_*%tH~Bv2J5SQEnzj6_ zQ8p~|lvebcupZo1#@_9uOm*cku!GSTD*xup<9&KJ$n5=WO(TdY7`9@qyGGLA=idR> zBo+gjB@Eqyc6Q3xqoo(jMWg0p8-fi#dRK%FOu zDWyynb@d<7sE`WyM}r7Y5*~keB&VNR3ExNtr{+f3kL;P^t^ru@F1!!0mU^@K?y9s9 zaR<5viQTE>=-w_=pTTY!I{8nm$#hW`zg1M>Vk=QNBIy|co7@+m3r!X17S=q1#EfxQ zGQda}3Chk=C#^0iqu50<+(_$r+05d!AXLvyXspEajFk_66DT*iFT{KbDgy?34gG{L zG=d$6u~9-H-)zeqd&R6lF}D$L$9Q5!<{6WX-hqmhzCG5h393EE-v?-KxQYs5aO2-Z5?5wwn;?eq#=2iE*(-|J5YUl|CGU< z#rMTKO8cTy=UH6^$gK5aGWI(xZah}x>N{`z4-@Lg$rrr&Jse=a%DMrA_C=5Ai=eTR zSINAo`-&s-O8mqQVhXw(UinTNolr91S9L(tGZ($4dCCsXKA%<}zDHlIehh$7-GD-s z_cxE9HQe3@-D#fgPX_!Nta257R&yUaq#irA;h9ynPb*NdcwpcD@pJE$0lyCiL?5}I z*l?m++8OiON*KxV;2nB=>>MKmjwd{Q+*}>nrcBwF~ z7rWn?Kt<#iz#!#)-UV^aTsdLD3QhyH1u-)B^lD&8S;)~q?4Isp3fZTGWx|f~%81Tz z*lF$?upF9_*aw3R*SM+rdc~1wFZ`=${c9&A9aJez5iVZGXz9bo`$10_YH`iGsH*HHDZ_?yENJ&n z=0~RhkYf5Fn3WlJd)FWp2;7t|mACr5;VtI-@gBzNqPMcV77qBHtv)raZR!^4tbux7 z5Mu={uIcL^M}XRUdKKI857>%UZGnCJ`h*6j8Rpo{13VmNW^6Aip^suF&m-MHZHNmh zmO^3(m>qG5+i&>5EBC8scC>yLz1@b%#F)A z_VvtzNBXpJ!=9jsIJs6NewdM-Yf8v9Y}lfQ=y#1iL;7}m9~*H#bWR_McX)MO)=C%( z3ayxt9y4SsI@!0=UO1#QRV_wKkwA?hw)yykt<%o2;O~zedJ@Y;*OM6z|2Cu@?ytFv z`3>o&nB>l=g}(+T%7F7S8>A~2d}x15hS~AVE<$3mDsa#RxGxQ|ug<5MASy>r21hkB;>V}9{t^rD#U=k!@^rR^yIs!#e~ z&qjwoIoE&^l_%8Nl8z1WR@`&uJ9b;C(9M9?bo(uFPtt}arV1lLY<9dy(S-Elw9m2b z^V&+=5?v1)DPV4X>81zwnp+R)I@VfNt*pn}La#6`rVdOHWVQu}-s1XE+rfFs0Rv>a zIwOP+P1Q_LfI%N42fh28KG;TkqIl4ROfF=&aTm9WW%5Q%(8Tgce@uuZuJ1JuQo??N zQSB|xuDD=U9)rm8NLcoj5qF^Sg)eh&PVX|q?SR<9x5301*L#T#+^QbD=F*g8bJAdj zYie+ruk}=bG$R;yq|s}bTxMJ8tyV||knyq=iy8-0TV1bt_qMmMCv4K@;B%D{h*7<* z?Syc`ZtAB^y!T6mkF7wvJ?%h&`X&*d z^yW3pa(8A$;$ROf96!AgFB~c3czXitnQncQn;}4Ku&>X;ne+tQs8&QT9qFWw~t$;27vI>`u4%$TxJU2#8;yoT~GZ{qboX2*dVdk#mHRL8yML& zE;M-&ww_M;5xNg55-rw|a~&I0dK}g!Sji%=@B%9tfGMNPfWOpWd{;IeCTF@K$ks%W z0V~DAe@bIhm|g`c7C<`W!Fq-!u~i0G4Z$LPwJKEsoP=hbrgx5d0oo1EpZELG;*dI6 zza9F&NKle%W&@;GVh!t=!MvKfC&E&=3%z|b!(srEwwzJFr*WMuDv+Vb_#Q=>-RC&R zxm~aTt0=k#ioB58r@r$g`Wp7FSm$Nu8khpXH=C9}SE0}01&cl` zGI?!+&qiI#3+M{nn*^)S%ym2dApcGzUqj$&Nf4FGt6t`%qBAxn2TCj6vI|@y`rQAO zkp}Q-)W?@4+acyBkV6B!Q7-^*_RQ)#-kw~01l>5$=ncrL*ekMoy;$x7)qX!{`)ggp z&iDz5LkBR86fOh)^$Oq(k!!K)kGz0mEET|Ojf-t4=|gP@ae7ZqbXlnR8y|e6v;r>O zY}n}WI%=%)I=#m76RIk)T=h4C7x!80TXCFQjx_T=mGd+CZ;y~yl1)ye2m<5+!B#y) zXljzIGY%&%R+^qT7f4P7NTmvNPJSbG8k@O3cdu&&!Vt}Db~Yz;BI_4P!d=*$uFD!s z6nJj&39~Q_|!gx%tx07 zwoGQB<2M%=7Zk`o;C6hH6CQT>w87S%9oTZv=d);YVu$>?5K@>77QtDYkgp2s@?+*l zmA7+=r~~~x_{oFViZlR3u(TtkLmpVi#n}$hM`9YlDpbKLKz3jjLnD>Odo2;gqw(bN z|AiY#;zHaZB1Ei{hAIGVl?v_$_WSh#1imxId`kolo5T9oAdXED%|tup`Uss7Qd47) zoq~;>Uf_{M;N)TxAi9@GYhn-02CAUPh$mPR+0Rr++G~JaAP{qjSncxANd4_|h9uLe zbMyZ4GAxe~eHPh+$k8HZO6Wfm)O+!Holr7bfs7Q*v?e32n0J5*6v2V5^ZAL@5X;w~ z#h&CyNwJ3wA;1}IKY9T)9goA$ETEYeUefhFKcb~-RuKA$TUmz^1R*(pie?IHjg+-) zjuD@bMxcnMJ5aE2+iEZUQ^Hc$cCl;HNCsYY0_v7SmawK#97nEDH;bOP!*Exbx8^?< zcK!dAN>*QfGz$*5p@oQ{B*nE8B*~9%DYuKt30sSa>|}4oghPsecD7?J&cH#AkEnxE zUl*W}j#UsqWelLo0!ytmktoI);b~+6klDIc1C=GN!W#(%GFU-H&>Z}99==QTp+Lfx zzNkQKD{DD<~vr6FzdpNuBj&d-O1xQO=| zoV@Wtd@{W{2@;sF+rbisKn;L11Q5Ic7$-F3Wq~jo`j9T;O(b}JF|*GB3R~RgK-mD& zSB~~(NVVoUU%d-snI(D*$Ex9z(5WR7{p6<9Zw%A5Rc^w07%~uz1>ic>vu89Rm!O9s z)5VbT#5(jM2tGDbaiy&DQ85jw(P$@lI#C{LO#7XIElPuc_OGr>5`u%`2m(m5_yN!B zqr98UPnhQLUyUY?yJ87x;tABTD~Z}-bA=&sx(YdRTD>ddG1gKFop4z7JYMWTriLNw zje*^OoY88U|5m<4TG$#=km7n7qX{HiD4z0SRiLK+U310D4Jau803tR?YMxe2+A5Ah zL|uU^K3e<-L)?@X;J}*<=a>b>R~V2AC;yY#C{UPgG@ZHlaKG{p)B5N9B3O|}L`sm0 z8I60dD(GPW5In&B6+<=Xk0{dS)5|WdmD~&fKz23rsQpa-T@fGnM11GJzU(#8 zryHt&X1xiW_>+7kmtVIbir%J2+%o2sY3~(b6O@3?l5Eub%VMAzX}@FwK#Q>$wlJKz zJDIA!W0i1wK)54dcj0g6VG7{=769)pQ6-8#eLL%@Pd`d2us=|A6drCxHc0r+BA;A{ zf^9an3Vyr;wwnswrDMry=#aNXc@sy%W;z`vA4~p~MC`YLjdV90&j2_(^%RXwARf^f zXhQBK?lC{$OEU{?-4k`)>`kl20sE`q6)T(F5zO!94bD^>nm`Vz!4!4??G_aP%M9Ii zhrn0h#U&L9VoU?1%_3C7Qf_{(5z1BK%T^{6VLRcsjex%2-07v z2E#Uyh$nDQkYTiFKxulD>cy&b0|3tlx+K6nv6T6q@b?Gu4-1{ZtX6&#pHI2Dp6xMG>QecuR%k<7K79i z@n{qT`l%^Ub2dRg1UptNd?v-%eIeo+oR5KxSUtJy9m=ZYrUH(=?MYf0D1(w#&_`2X zUg?V#!X^|o9hWWIlEF5Qv@=R7Xjjv0K;Y~ab_srrAT)Fv{1Fi37>RnObt&zrpUGX+x4 zA}H;w0U&7SHu92YR`ddVKVo+`1H+2#~)%K+OT;vLkBG8EYt& z18)q&ZXOC?8{NjNQMpY%6bdrPmshnR2BhUjMnrJm*Md!b7fqhh$@SS!!G=1}FoP3u zl8>X?ftxA3)E_;DjOZ3yI>p0Dh7!!FkpDNQnp$ECrJKx3ZqH}IER^<|Fzh(#-atxd zYe>zDy8|f^T!`R0IaG2VBBzd?!^oD=tf%7leEB25KI2ZWL>W5D(m-Iy|ED|IoG=wW ztV(Bfw6IPgo+bYDWp4#(xpc=ONi5#>o!ugDr6L-#@JN{DtqSQ zSIvv$ybSC++h+l9oCj~{%U7xjyXVcd?VMfhT~uBcs&vrMYQS;`TH27m8qD~#$u15M za6W^j6wX!3?fBuHF|QKKHvHB%E#+aDocKBwVrc8jpkyFu8hT;aC}o=@;Yr~RHV?8}A_#9EJf=)s#;78`4hgWoOblmv zK5PGZa&i`{`|tVpu=&Fa24w0WXZEu^L7crpp7=^k!&Nj9_6#D2U!*xUt$-Rks0=r3 zYHc+d*9B+f&rrdO1^N(V8BUCE1A+nypqfTPREN~t0zWa4qIZ@<(?~N358Wgk{8{caJ%c@>|kam5?J`yEjU zk+Y_({3a$w9n=|&#-DZMo59zs(c>~o&8HpirwnZyS6NTUM=Mu_Rf5T94yUC<)B$~u zGn+Oj{p=04bWmLwYAA>1%_wWKNu0xhUXFGJ;Xc^#M7h}YqR+BPRPw2cKFBKJxeo3w zmT@Iyt|lI#2WH#2N{khg>Mbf&t60y(*7q#_x}jBHzB6ThomsF0KwSim&u|ytW7}!} zi5VpA0IuKwNE&oWk{)m%z9FkKKqwybKL!by?KeT9R3b>6iC=(hXQpAaGm{~cM)O=- zTmFb@NU)`#f0lB}eG3pLq!P%xKdMAUVUsqA<2m`rRcxz--bu=-{k4u+v@1mSoo3;- zaUuS`ARMx#>@i6|8;YyKLM+3zw#R?(XS#XxMz^At=R5PprY~U88N6pen?cx2)a&?U z5M=596ezMK0)^$a9Ok=4h$pu>q~GsW{0c0@X#@e^_Y)KsKi!`(Cn8B00D%%HYQRRs zA&5L zs)MhBCl5VNjyBe5X%-}oJpqpZ&9w>9k8NknvE%E5q2q}*GA>%>F7QMxnn~K=Vf7l& zLY6b*2w}Zel+s(#`&bcTngeieeP?0X(CSC535vzGkb+}F%1v5Ge=qP^CF;;aJc;F} z+O|{S(+mSO$~moa796`R{0QLzjsQ1c;VC@G&J99uQ_EiJs_(kP=_^S>pd$T6Pyi-?LdbN-`3c(eW#IgQkqJO)`UaY z`n*Y#e|QWEN&a0%XXZJ+CgV0`<#@P&@&+Orobhr5wo+^b2qc2fynAs23p+Wlq}PFq zS6M)o^ARP>(dhMez`WCX6tCPLtemsGTo`UNNkGn>^PNyHyV#6*5msDor8qRXyghzu zfDMT_@#2}P0rnvJ(ydL!uU)oPlZSq9?Tns3e@y%k1AlC<5Xl&KQha={rLXNd^j5NN zMRR(RyvG{HPMtL3_1Xl{FHi5e$epYj9@dqtu~egb{sBMHHVf$gH&S-9)mQNeb1)jtvGKgYO7p7`7xQ+5?9y;nWpLqb(UwN&^jFb_(wUzLAl48iU* ziFTT>xywA(Ya(*X-W%SLHB^g)uRKGnL0YSzTKGvI1YXNw%@hWZttc5>3$EFeC}|#AB9bpW&*rybCG%Mc2>dN3^=k24*V?oWGeBHu+hmFJZoGfT^#;%i_@CmOOKJ9 zY$!)o8*1|JW1I8~Ryn?BrA}mB;-FHBNR})5F#%>Vo(7zoo(L zyICz8?Cef7_iojh8@eKS!&Sak$m9%B zX4&pKZ+`VFJSg`k!&XB+$H#DHf7;!MVL4@O!>fH+2XBtc+dDL!LRgcWWq?;QS{Q!X z1q#E{xE6hCL)pnnYze8GNeN4pWcu0s!o{q0H2Ph6_^y@c`8Q%EOTDF@IWKwP8`NN+ zd~v|=r0^W0ki_P{EAgzDh7jvcd*P_^Xn_EP3H;%4+vqCNZ#5GgzsvgqFBu(*3|iK4 zGE4x3Ui+$wwWp~!P}tt5-hLCR$b)dL_%15OaE>-gXvgLt81hJ%PLaq5_ZDsPtXXRO zo0At7QsehS$A3H4t1MVBV2@OrqP8M&U&~>aD42;mH9C>KXE4_~2|gM7m9azGM2cu< zC~>=slfdE$2i>XQylpiC?&5~p%vZ**kVvA#%~?qZFV60;Gw}R%?Q-R6Cy6dYM_P;J z@V82461rv6rLq7&=-ymt9=qG3d%kN1@~I9a3xSL#x&>H8WnQR@S+*-&gN2cA^ga!3 z3+<4Unty$98bnTTD5$mm-#m(}^1gQ~1ggQoV7k5^j)qGdtCuW_1tHH2nSW5f5#kC8 zX#rwR7@#X?NGnz6t!*zM5vx08!DURaxwb?w>J6gjjJ+5$Z?uGPBb<^KMvDz|A;~mU z;-)F3BWd+s&uB(6czF?bnxcyTFKHLw zMxfPhw`0yGFF>*gcMXD2Z5jQG=x^JctrOW`3h|^I|C+T# zh=3M3xxrI|lKN%KmIYr1qo`mtpq>Z%r2uWDBIBnJ>k1~UGSC&p4Qxc-p%n_jG(VM? zDr1sEhcFf?g<^z4C%b)<@*r-f1}wPS)vyS8VnQ3&@Pgz332Ys8K^h4&k)w z7IYZ`(jtk@C^f#XutG4c!|+C42#r|q?K zyF7laRsZ3Q`*K7xwey5q-J**(XS_D74E`Zx*iVN}eDG@$CC=g(%QRJ=v0V9b){j>o zG3B&cy<-rkj63cJ74pq)op;~nF(Bh(p6-?Hy5)1skmUA?^P*J5*n;_a$Dr~qiGML} zpg5%j7S3e~u9LB|s(Ch!mT|Cgo)~c?Y#3?;7~eD28jMjkVc4akAmQhUly6459RUh8 z0uFNhgvr6sFNOt?M-jpMDR`Vu)2wHmi@E>F3BYzdm_{zAnn?UF#ox&ov;D8IJ`KN?VS zD11g4RcE!YmuxdZ?J}_Sw2Tg%@jjMPn+ma#9SzFWKsAMGw|KeJ$)8IS86{1w0 z=)5!IF?6D0l;mC3HfTB%=y+uhV}1iYb!L#5wQ&U;q!$rHa>Nw9{2&P!1rZ(3V@p!m zM@pd1!mPw4Mb@<6jV4;_RG2ury8I2|JWur(Qn*!Ox-o!Q2b-g?JIH>Q3zX!Rj6Rs~~5ZGlKq|Jxp0y9JEp1MBDUc5#0F zBOCIn5BA>nol&O>wHH-;1baIbpwO=Uo=D5eXsrF-;_;(9P5BK1r^;1FdBR@g#W^(@ z5g8-0s#vvV)YoqN%Fi@#5b;^H<~oJJ^1&}K;2y$KeOZc5X*Bed0sO(H3ni(_VN&@Q1Jq6}4DOr4RgzV>M2Sa5>3|zpF2ngsj8% z^Q{hhUx)PWf6;YekG96e&_r-#ZU?{&QJM!h@^9yw_mz&{p|Tg)4MLRSvMh=We_)} z3Fqwt|HH)8+r=WI&z!m?eMVa4CY$XwF8!-$hJsk-h7HU{#VWqHjMDQ9c1yet>>T=Q z%DAKPLs&uIwcJUCehY)M8K~uU({*x2cGgtVI2jKL9FZ~>a5|q{l`rp)`?A5U z=`JHjDHnBFzgNaHx4!=HsLfP#-^ys>0ClH*_9pMLea88qQu)V6thvAa`doP6Nunajq#35Qta=Pm>P3gaGE^D>vx#GRGQLA?Uk{hf){+6eJArVuymFb#TJIMZv zL*fGX?Vo)b%=XAHJk@Vhtlb*!u5Jb!ByhExv>+r;ScjGO8I9y=7qZi$Ub`-PaTpLR pD2ELv!vO>`aM;y^!_x1fOnB_9hJR2@w7Tq+sRC1oBFwMv{|{0Xr)&TK literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-12-PROOF. We .jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-12-PROOF. We .jpg new file mode 100644 index 0000000000000000000000000000000000000000..4e6a91f796e8581c3008c615f6e376a5b3db484c GIT binary patch literal 853 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}niCwprZwe|*^SpJAQvpWZ?{!4LMW`Zn^5 z@7(#3?>Vb7=TW7~q_dK9yLu*hJh{MLJ@K_`@%#_#_elR|a9Y>mG~XaS5>&t;PnUg!9iGWJ8PNxc)Ots9XM@Au)UZ zU#ZOchvHZ2UzB`g|NA-pKf{Tyix*~W%&@ACI$m5@5|=*r*$(Z!cW)O9Zw)E6ZCJqn G|0V#kr9im= literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-13-Thus plugg.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-13-Thus plugg.jpg new file mode 100644 index 0000000000000000000000000000000000000000..ccfa2bb313b5d03765598522164df29bacd667a8 GIT binary patch literal 835 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}mu{D^(iURpwo7s40KQ;9;kA;h_9if$skd zU2A9G|M6kxe};9we|nRCXi4-pelbdR^6jyCD}PsnVgHAD754wy_VWK{U=ytW7`i@j z#m%f~-Q~;gMc=(`sF=2WYKn8;4uy$9KLzX0tNdr!=UTjvlOglg3})uvVekJlxXog? zzxO{wKll7!Pwt<%=H68Q#O=cV*T>}lGKB4Y|ED76Kf`ske{MU@bNpvGF?VtO^*Qms tI2Q1|+PR*!{)Bz-e})ZP@BiyHRSo+f{-+UW-W$#G{|t+L8S4Mv1OR^}IlBM= literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-14-DEFINITION.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-14-DEFINITION.jpg new file mode 100644 index 0000000000000000000000000000000000000000..5cafd5e0bcb5144e12b898921c47cbb5e8e2b0e9 GIT binary patch literal 854 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!*^Tn@XI9`Vzt2mTQ;`N&N^9abY#gdrX$6X z^&jR{*#B$W%m1H&O|brBD1YY^^9~y=2`0Py>g@k^c*crP6#w(d|M!ysHWNFu4@-}n zp2^=D_w2g-M}FDe$e6Ra1=Am=Y}n}@v$(m*QuOowKWcTF{}~oekNnSY z;Aj1hRSds=AD{n8YyaQcsQ(NvzO0+JPw&I`&aZ!+kNk37`)=pTJGW0(b5tzbJ@Mkc Pf=ZSL*Ke6H{=W$Tu6jo? literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-15-$$ _- Summ.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-15-$$ _- Summ.jpg new file mode 100644 index 0000000000000000000000000000000000000000..e43849aef511ef7079995097f8f6a10b8ff9f102 GIT binary patch literal 873 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}mu{D^(iURpwo7s40KQ;9;kA;h_9if$skd zU2A9G|M6kxe};9we|nRCXi4-pelbdR^6jyCD}PsnVgHAD754wy_VWK{U=ytW7`i@j z#m%f~-Q~;gMc=(`sF=2WYKn8;4uy$9KLzX0tNdr!=UTjvlOglg3})uvVekJlxXog? zzxO{wKll7!Pwt<%=H68Q#O=cV*T>}lGK3wyZ&v)D;kwzs(2s&2^E-7roA?E?Habh* zns~fgX;tF|27aanyT|LY55|Auzw)2qjc)mWhQ+=$&HouV=Xce=NWE77!7l!v*1iQl hRBg7~^Q?KkW9rjcp3l-Jx#^too-@f`^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}niW(%Upstu+-}avWv*w0k}rZ(Unz&z%|m zy=vN?@8V(`Piyn$y)8QGsLtfWf2hHFmiFxb3@ysO{~6~0t^d!!zJ30mvdIUY1unkl zta|jwBVM;>689dZO+BR3qM>tSa^tHE=Bn@iZ1n11Pp(&vl;ix*Fj3m+Kf`*7`Y(*p z#qQ4?RU(gl)t>Lu_0;4(|Bv;amvu{i=L==Jac4}^I5W$7+O5URDlvTZTnw*Q&-xJl zr}?Y>Z|>Ot3@|Chv<{69h`|1+F4Ek8WV(gdl3H0go_5RsyE3>HL+fHa8^2-1sm0WYX@1ref1QL54f zDbk{VU;*zCc=bxE7^L;1J&Ph&I&U4mYWxwlPoBWRa5jb!XW&{I3 zAOHYSUI2Lv&;#gbXlQAu>1b(b>FMbhnAn+_7#W#(SlJ-#0=!T`0bYLogQBwH2Zg0Y z`1vJN4oe@AS5#DliXYQdRnU+>hGl1C?^`}7ru#aq1S(ThG|^bDL_+&sKuhr}fgODZZU ztEe7R(>tLLGcYtddFHI8m9>qn-6f~X&R1Mq-F*H0uLmG*1cpaMMn%WO#-*m+PS41^ z^FvmCL1EE@hs7n2DxOtVRo6U!QQO?o+ScCD`MPUhaAlr@i3K zCaiFij$Jn;@98Ugk)sxPj*C8T88}51v0{X;s{N?h-=|pEKcv~uiv6P3D8K{;QH%#> z0knb5ZHfF)hJWD@<~qWcbns-O=h0^xxFR%rMcs4lO8p@kSB*T+$wO0#@uLmp|I!Ee zS1R}Oy(M{;a;tS$Po)&|h}~gm9;)}pHXODrU$;4+a*Ce}NCK;Ta$Bl9=8d5A z)r5z)6cvfH-epQF%$j+UhhDW`Yr?``Um)-`*QR%dVXPyk58mZZn$;1o$5?L>sJe}3cs zTB+$)rd`qC`~+lhV(HRKtar*fw^GUjgkvii;E?6Bz|3ZJoEE(r8UTKWd5Q>Xwlj@- zTu}YATIVd_B-I_%Pn+6;U+92K?ajtil?cTqJ4>(LTGO!V*gd)&-IcJp8L>GJZ_)>l z&phy$Q<=D-2`<-}9pMuW+I-J$l9cB9cC-*kksAS+3;63fndncpBaSQHn~v*Clds*! zUBEAAi6&DJNTH95+SK!ec)C?GF!Io57fuk5l1uGw8?}D~x!s*~C+AF0(6F-`NpN6& z%sJewGSTV6xbId!wJ9^A1xBz$-+OSbFd}%3HeD;KrgWAJuc^XHx{?fD*zdqFiTcALCE&mY~s1A)8*?Kl!mu<~apCnDrs z{BuXYuJGb@lK)~lKfb6H!W74#mVR)jpOyzd*^z?4=(DQDEDLX4KDD{dF^1{W62;FH z;h(j|;$#kg@({K?swROo$lI8QPhl4WyITV_cq4D+8%~=cG?@Xnw}0`aum1EOIjp0Q zaDwH?vl2b507Ri%L@tXUFY2R;t4oR#R7`4a=a$U#QW#-QXvzvX5lAi``v@Pm&#|IRq z4kd>6M(?j618GUMWB`4gWDK9up=LM#dGOROCAgQ*b8&K|#0sa=y+YIAl^cqY+JrCn zxrPE{a}>ue$*Kw*<5)X)!>e5mEPyiPtqNl)eqSPK_Av!h69}wxoS)-4eVm8AYDSw* z!&BpU{yk+vn4@iubhOtGO}$cs#s9H+@A00yb#?z*$@0=2I-JEjah}pIx+NE{(TZZi zc3~q+^*3ehuG0lQ`FURAMP@H*ku>6*FKV3Yp zea*AHJKw~B*C&&9jl|>Y_4s1?XGRr{X!j{Cea<^8!41Bt!}zQ4 z`Dyb1Kp6cPQa8lg(bPSq^WF<;J~sodtfs1}3Xp-5>$Of#MF|=f5>D67hL7_CtLO+V z#iZ5{p11*=zgpyz(a)xDbUyN~#cg%Bj`}Y*PsVsJDD!2S)F*p^5Dxcq%7UpMd_Lfo zg;dEHkZO~RqCV|H_2E|`{OfT3vpzGY>c=Uk^m4$AHqbN#=Ec4?IcP`TdwDg@Y-el6 zT|KmRq`xEQ?q%8RE1cm)86GfV#2G%HT*98a7Zx3!)_fm&Mlz>noO!j*UvzbMRY$=2 z-osZ#@AR%*y(t{4Tg>71ZTJNK4N}1TChdI_dGGgapWusb6XdCVl6qu7Vdm&K?}}-J zpvVjSWQ^>tNcqXmxv`5AKJ(;Y# z()G$5W-J(1&t+5FKzuQECg0wW?Zh)==Ntlw+U@~tiH`XWh`QFz5V;YLitLyg$Y3Mg zwl{o!px9;Gq&&8$<$?>vB0Aj5%hq$>OYDExDpaKt?^{rpWggE79=3Krylf?r3%Tpp zr-Jor3wLQvkX^^Wm;-evtvOgcek&T|Qx2kCa4@Bm8PpxPzWqj@sA`g#2?~%qDWflz z!%2!krLIWzM=2cZ*&aHxGu|X`9l@$^MEAenLL6_1vu)QWSn+F^icZdp49$gCWOVqi zv`upN?UN_E8%S>_VBdW%y({DI2+ifY$A4`4W?P`1WNrF+hLkW!;MPtDugPQ!ioXR{rX=wA zO4{bzn=46m`H>-rdAEo_1z1V7r_Lm_I z^2X;ehCE^xv;q*fk^E2%M(o53|Bp~E2ZVvm#f%E=ZS@+oX?NUao6zkeb1?lI9c`vm z;xT^J!_sC6R&N7Nbb04^jg%rq32P+bsx-r)m*;f*K8fd@SI32F!v2ilNlpfdeq^v) zaf}z#t{oWFhRoeaGq2?{hTSkFH2@cp=7h`3vHpd4&uE^MbD>%K(;=vYKvbB9&}A$$ zgOsrgL8{o*W?hElYct7=q^FHmtZJ-{X)N(k=@v=4bEPWauYmIJf%Dhz|F~HwW}-Sk zcW_PVBkF*kIy#ypS3|g-SR_4-bP$HXCyRm>&4$g|OCL1w5;%3p| z`uE~zij(71T4U<@n68iD_X^CiBa#LuLPdxv1m3Y0PS$ttyDHoUWJPWX!rZfow)pM6 zWio)3B(T*@n5=g2hpTQXk^w5j{$==ZQhdEdcDa~3_GUDLX2%q~^HFi$S%4N`0L+fa z|K8KT^ZoxLv;72N`6mSsGBY~uO1GEL7auXEW|k^yN~ zsJdwb04FrHCdf5EG4f2kv7N@G&*1xdK5KfH`J?;t0kU)$`M2?@QG#6C=FY-NZ5OULBEcM zAdiMtYctXQ=ZNvkHiG$~V*g6xJ*a!^(mPFUo@$~W5I~(!naRZ}?Jb<)dp(|SUMm&9)RwRC;H1EEiQunJh zcG>q{-q&@mm80I(7(j}-^K~4^AflJWJ&ly!^@9ffV19@Ej^%+1TqXnJP=7K2U);@T zX#Ycv8`ViDk*cFT>kMqJBrtsQ%%x=a-gJCbxeH=P2oAz#JjL?O zE1_u|6v_jPJ$u?UqXRb?==PoOr_DlW<9c%*ZK;Knqc4?Axfd3nWh3Gm4%?p7JG_r2 zS^pYbg=k+yd>T1q@FjCj%P1`|m@x8uOpc3~jcpjTL1auAe^5PQvy}E^lTtim^)Da- z+C~jx%aDs|^U$5n({yE{{{+UVAj@SDr*IO>68LmZg6gDuXG+PH^N-Y0-Fr0538gyy z)5;sC<;nv*##lPbtyBoht@BP~z_!rUU@4`YKn&Lr=!+vs&A|n%qImQdmya`viz_uV zvvp{*$-+H!UHuw_VO`A~%W~MPMWi@BUJp++T1s2Q+&fjV8omgRDlX2m)B)$RgWeDw zUuHogFPL9;<<5aBwj$eW1OM?B;6F8_f9`)iEOEVVS5;s)X;mS2J4S7UHU`O;kI49t zKO|jnT_n0gI8^DXzm(?bVcDxotS)yi+nSnNm9{9!MTf}Xk=+4U6N?**1;=*OFR8Ba zTQw#T9!xMkUt4CgHF`LHv+6@T4T?=i@P+=F!75`JyZ4^bo~(!@m>%WPzDF_q)%D_X zS>dXivp+P$LZ8Whp*FUG%TZEagz1^B=-r^#lY418yw%p#+mVkPMmsBXrzE?4m~G0% zu??TfYC`?;G~dXx9cbU8+7PU2Lg9kBOk#|@OYEt|!K@{uN)THAMs*RPra42wguzsL zprc>ZDN5UJrd;c&dsbm7XC!vv)}rVS2UeFmb0MBRPHy+b-&Q)w1Q#FYREzS?T2T*% z;-ce|`4*-J7kO$9CoxQog)#dDqFx2dyh?MZuN^vcp8ms805W(eRA@LvY5vR1q)x^2 zp+mjnF}I&dQs>%nD^V9{rK^N4(o^p(}N ztDwZ&pxH)l;&JRU#86g%`BNtu$Vg3^&sxUCshWuz78m52-!%Pc7r}$3ajzmkK`Q2o zv~W7q_0_0vxCJJAS0=;A(|=h3hJ<&~Ox;D!Gzfop?GO%%xs-nKhPjhhZQ+WW-KrO7 zw=mKS|5^GLOMa##gmwv2_+3`)AB2b?RURdNl>_64U7TzIRk719IeU70QlUfx>D0M^ zb|aq7x@4*j?e-zioZokfR}R*3-Bm>?Ngl38;hiP!J@x9+2?SqwLgB^GYlV0%%Up=j zS17UIKn7xvTof+{3+7$nW0h}!L1v_KDP`PARAi$#zM(Z!i*h>hHg6- z)WTHVC_5lFPwq;waLTZl zkWC0r`TfHMSxTwm)-WEXhX*m)?kMU?zL^`9vsum8MKcn0LAx|kw^7ke2kMnC<>-$j zU@{DCib8;nJdJ+Zy0>sJErkmcAO?$^bBEo)gz!woqo<_`2emX^j%Jvg9L06%=rEXP zYd>TMEu6KM_aJ~udLi*P!Wq*8L%8L8~W762zw(5SFUTO!DccSk`7<%wvh!zKktLKg;oj@tlg@ z2W@YBKH3<*QsA}?=?c|SLG3557$eF?C)U~yG#^^xOdV?uk=!}u6OrsKlJpV%_2cWTVt%FHKSfBQnBm&y=S2N~I zamQV6>}xqV6jGO6l4aGENH4m*GeQP-rJJB+puZZmoI+Astov#a569zMX0)-W=59)3 zeUI@9X$v}K@V{QJ*Y06Vy>|&QwZ#SnseSGI_)(t0v5uxHKj?3q0aLw5nI7}1VA@YQ zf?~h*jQ>lL(8Kn`(pFc(Ne^4U3V-F8lA4zv&!a-oI%FU+bBYqbA2`&}b(*sw`o}QVFeY(2CXsNtBE0nDiYNQsj_ixVu9s)HBaspmk-L4fo#Da(CCe? z&h<6aav`ZrxH&S==V3}BQX*9g zwFRNK1t%^~TLfrWmJbA&6`dUBvksGsGM zDC-T4SM-vcY^Ayp8Y_QJ@zR?!3vp7Xf7z%ddCWPE<4t+7rcsGh*hoE>N{mwMQdQ}e z(BMldiTpbcwF<)|z;y<;>DpaIe0Y+$`9l;z@7sKtreg}NI~;g3 z>QSvB{PEb8_n!mxKMkroh1e39tGR>75vD+hT`(fd9WFGiCF+vNdspVNmvEPGsQet=bfVPQ_gaCo z^{zNJEP8)m5O36Tk*kt{T#GsaG{e2uZ*Ai;>H|$7h+d4opfG7v4$;))u1nL?;GOe8 z-MWBi?O=Lfj&pONg{9*7+*H(_uMXeY@1SaSYpE$HXU_gY<&lll@rm(UC*lf|S`*XL z@!3{QBO03C-o?jUBfcC#MbBuM(W1V?8-w(H;L1b(F~ZNDzD;S}^x1MgpB-_DHqGHd zX)ZConb)ayEkrnDdJO%8r~>SWb;+`h;8Xo8bAd6#e(v`D@-yv^L!Os0e49{syyt2E z&hLjwQ=LtF)OE-yO!j9rNW5a1uosIl4!rR$xohMT&718jWT5QpT`v&zta2RiJw~iAlJO@Z~sK1{Cu|bfZ4|=ZhfkX@407o`_z>8TB)KLZH^5wseRQ=Uf&lxoJf;Q zo7Q_Kvd?a`w*Kv|gX|jESHVP(sg#fVT`&ZDPSd+l7%wqY#Zoe_f_^ZWQZ*AZh1hHZ zd}}Y~;>8oo6g{*I?hl&v5=6ZG4zi`11X2(=AO8G+tjR_JVmm4q!Pez-HpscQ>#vkP zykGaOe-Wi#+1P2luNt-OvRJQLp<5P#ItJ7CSFu}hwd0+13)Pc-d*|kBnK?~&=jV>R z#Qm#n0UH<=x4x^%K2O8E@3)7t84k$qniiMWl$DS9AFHa;1irKeu9eRLG{8%=S`8i= zo0oOSi*vj}gz8Mt`OrY?yw)ONOXqQ&>Fs{-+&8vTPH7mFBg#80Rx0nYcrFr1Pr6F` zrtIPT4a0J4lPcB~oS8h7h6=aJN?Dc!_|2|^Ji)XDwch2pm(if`qQSnow1k**dgRoY zNmfO~darhK?a9AV{cGz*s!!_(6;l4Hjw;u|wE0(Py~ilFIU}4V&BCec^s_={WZ;8V zCD6U;NYoSaXz_ZtJ)2ZVDfya}q(0tEhIu{|(O%A)&3pwvmhoEAIKhprC<=5}>PDpI z(KYnx0mV+mU9!9>mnenw#qQt0? zttGqDGr{zij;6j-3p0addItF?-rcu2a(P4WgS#G%GaknraMDvJFBWtJK-)?R(>v+1)%;G&kArWybO0RRa9 z2b|3UssI@Y2`LFN87V30g$rcllnhjq6cm)Kbo4X~oNU}&oNOE%m#&IjyL9C`4+jUo zw7_*yF-b{D?rXA&GU5s%5|ZNQKtLBRT%e?&WTB#B5x>lFS^OWq&fWvGWFTXLG(yl7 zfPfZ6NDDga0l4w&BnJI<0e`tb1cXGyB&1{)$SLpx>MjBVAVNX{B0^$fA|m|g5PUyC zL`zI}Sxkk59(tejiYJ43c+zV!o?F$OjE1A=s}iFw+PFz|6~d}4A6IsIj33AOxfWp(ZQ`UYnA=idIo zA@ju(i)58nuBiHI+YknYr4rbVFsC2CF`pN9yoHe(IVGO&$Pm?oe zKv+`vp(u|4t~1VNUDnSjt8kEafxqJU4SMIm$m1_}tVmx`lf$P50bxdl23pmK>oH-6 z4g>|qA;nwlFJV6x^C%=uU=5aQ1gqS~SbpaEo<*zM`gxC>2znX5e)X>yRPV8LPgU@= z(YxYI&8+yH&=elWr%7u~W`93XW!sQ2DHGt&hyQXK>9KkWORjzfT;MNNyAr%~7<8&)CaS zhpkpMHpnSuqx#E zX6_qEhfxyA%AO=>OPS3d0YW4DtZjx(we{&|Up|cnC3sA=jSU?LZE95Ixs84hrvfac zx=(a4T3E+!`cLd2Z~p0s&HnwW1*xBizDi@4oa^JYjoGOd#&YqmAb?K?ajW80%izqF zb%(}g`EbQ(%l(-F^jG+LGh$kDYcDGEtBV!sSX2c<)aF~zSOr6z|ER|;K`o@W4CKbmtPW?!F4g}_Ev zQsWGe9pGX7ewS6jeLA7L`A&L{A>&7`jH;|9^(QIafmi*E`iLh!Qqalf(WqFj(=mOE z>_jMFJC{}`PGo#2yi8FO?pC2b>k3?7U5ZPcIagmv(v{fzT#8WUf}taN;}A~V4;n*| zqQS8=dCNy;#qj_l1IKMMuyhyBZ0Vq%?kQx40iOmQ*+0hfGyng;mY9_7S5zF1oC=dr zfOrW@vJVXOO%F?9)XII#^j%GnHRVzGX@#qY^!KxLg3<>&N%J;s;?Q!d0>M{DMWl=M zhy&*9eg=C#eLE98AA!lE9$z+#x|iOKaS^?fuSvt3K33mOwn@JcN@L&XHAoR@b>wZ@ zE^Mvpx_Dtq$Nlk)ggF?14o8b{V|60%nx&eAarfXg=|MtYsa?wTa#0>MUwa-Q{8uOr zw)G>fMj^tbwv~Qg614ZV<73&&MgCs%61O=8{^i%KjC@PVAeq%BeNO*kZgMBOuks~H z;ba#!T#K5UD>pgfva)1aJXOrX5sYSwh=CmgWCF(8!Q$;I|$3!A=j|}6YJb#zcb)V31Qi|B&R$??GA+}nfH(&0$zI?LlWWhrd?Y6 zq{GT|S!74fZ;)bOJz`Ut`v%smr{OL|E5zrGizxWZs~A$w8dqsUjNU77FTYQZ_kqD| zy!upo9ELf3PoH*G_tqqrm7TB9>b81U8WWzRDYs*3YSR0_a=5EJ)lv;Pbz3zKe@2+d57uGpas zH)8o=c^uFiK-1|KX!JlNGLY4w{-H?F0m9~69T33Xx4<}D;8gm+EfLAVM4Gp56M$am zYi5qvE^2Hfxzj|g?Y57AAx~Hfut|v<2ESUAw(rBba?s(Q*a7doGk_degDnMr&rV^_ z#UoQ`WFG7}`orUXoyM&A&H3C@m#=3)q~a3c86f&&1;=m7ys#?)v{816INBmW7D4g58JBK}Z#G845KIXhdD zM>?9^xdG4P@m ziZ}ZIOf$3(93dE0a~9_ZITg++{^mqLlg@#j+wc^gsW2wEPJ*T7e`CyJM?jit>XqzK zo;gkc0zCW&OO}sZKq%p#5{`ZG3|eaC+w@(g@DeSYSVZ}UPExFRA!x$3RoV~>s~E;; zDc-jQJ6K4&_p1yBW&A1?TR1Whdn;f*s&7zt<#cY0^6-Zq=XXF+@TS3r$@@;0QW`Dy zx$I1!w+7vwYF8$;UrUxw>1CnL*WjrG9YEzOlkmU)ix;A||GWh2p*JZ~ka@)WqR;Df zq#M_Zxr-rApqS@uf|q8>M3qwc*L5zf5qWMKQojJ85QcxEU6x@0YC9A4wv3+Jmb$P- zR%N&xnt8=6LkIZ6g!k97#}=1m5tk=%a?X#e6__q6?6ggIo|F}(_m@j~NwLe?a&5r6 z!W456I&4<4_aZE8xRI3#szvH07Hgjp9l26_{j-)yvXc*X^d*i6c?$6V#XTZK$Dao>8^;8%qU&-ov1G*NzvWL62iFQk7#{DfJGna5<5iA$S;(N+IT-x0%=Q!|FF;zNC{#ES z>{jZOT=q%G>K3du(A2AIz~#A#?NQDlyM2&Riz!JsDvQ&dkyeAKaX@G4yVTHR+;s)R zF>=dW|2kLyPqsoeqE}e*bV>O_VtL9;yC1d#RbJ_6ozOk=Fm5G{{5{v3wxH9!cHO6A z^iN7Oq;`kGw#$PpX~gE4ma20PO9-cfY0&Q=Fx7tSY|uB<>ugcXv>2jMiws zxh<|_N@48|$UwLC0MBh@mK;{}9v9nq+Q+XQemTU!>&vs|JE5Wq5@pR2A}S|B8@5XK z;lG0V7IrNIa7QpL;J{yr8Ed%O#S|ibWti^UHm9eJK8F20>3gK8s~o$e6Z_R`E=MB$ zIF*C>9K$m}a8-%dqSt$|e0PK8ioIhUV1%_pEnqcIq&ZSJhFQ8D5GRmMvO7IkurW|s@b z!Z%1iny2g@Fb??NL2;#^S{j~oc8eb)?*V(8%H(b9&NKwE-vSK?Ms>apu(B2ihaZg) zelI2+3K-ywl;ObZ&vSL52?2x|UKkJ<+QB7;C3BkA)w=4Yy8;oLR=Lj=>^$;jY9Gl2 z<;xfZnk2CJ=iH!i9x@4Uy1$DEL~piyiqrCIY4o!9PRi+O=7)Nx6;77qJ)O8wRi|B? zgZUi0b~%VV<|I!rp1*Ij-)IWHzdmZ9o?+OU^%fx28sy#;O*jNsI)z7AbmP=$aclc; z#8iVa9baGDIb;ATo{Iho5fBJkErrM=Z%AEnliS_YY-DEgV!Y>_;BwG$!M|(4`klt= zy#_(i_lF3Yes4nyo8h9b!h_{C;@KE0nQTEVjQk^PyxY|f(aNN4ea25%^`5tzM_v!Y zGSpZCk$Tkf$d}`m5_Mm;ZBPad=*;m1Bw2 zPwu!FkJwYsXd9RyaOBe4fx7!p(f00&Irt){Q72m(N;ba!VbSgJ%I6wkWa-J>W32_t z(y?3Hds1T+Z!GvSV(wLD0xMw0V`u6*Z)tP_2Jrw3hoRy+geETIO!`k~(BEb}qN!H8 z7?SN89nn?YDDIPI4cIdv#}V&FM4#}Y)lyLP?QReXmcnJuBV;M8D;fRla^yQ|>;OJk z`v!KLOP}c2e1k={pu%MJse-{HCjyVh*WbQy+|fiek2n_^hQk3P=%?gdVQsi82RVTH zvByoP>;H3Rt|=fP4T*&Xy4I~)TVTy zcR$JGQExsXWnE2?>i^y@oIcDJTF3llCmIy9)WNoC62X7ZkGh8qNQ3Ld3pg=$!Nt@` z!jq~$ELS?XW`35tnZ^70Gs)=tYn9eH4;muBB|RyNO%4(2KJh?V_zdb+F2eS1%N$L; z$Xo5KWBmNt&Cx9V&ZK?It1R2xBz>*CCO_e9LhbMdPETdo?O3ySTDe#C`RdG|VH?f@ z6Q*A&dwy6H{4Oq)a(~E>qcA)wDvqJn1o8Blu={1cF#m6cquECB<IfESFGdIjOkOR!jzmNR4{AdYe7NW;{M>LH}u!^8XK`GD6+WP zQEhUuKG*=F@I(K6r(&t{unds_snDV@+aUyfa&h54d-x)^GG&_ zc)cQk$nRWZ_f#ju2N%+u%qZ&~p@-L!j97*h!E$rxhR!>s9?54w4>O6w(Ds|DwrLuz zThxJ>K-A2byuAPRqQTp~Msi|1|DnDR^{zwTw<7DD8jK_4hW@B9h3AXN$hsTU1Hm2z zf>*uNS@sb&q>|(P#r!ty6Tr?M-U+^-!djFcM$Z9-Tk?Zi0)_)`#y7IQfhyzGQ}nja zON}Ju?2s=(mEM7-)d5T8XTa8}owsuTLo?3mnU+4@AXA&Vf?3xoT@SCkDkZg@N@X^v z7N7F>&cxU!0<(>zQ;oVbX1U!nF(dnivln{ka-N!b73l#h`jtoWy*GBFQcl^vxQsmXv9#R8^hDf1?f0&D_@ygicu8$_+*=1qLEs&bZ2M5s6Fk# z*6Uj;$!l_G;BFui5mn%(FVA|%#+apF%Ti0I#_$e@v$-XIetZP^;N4#=>c_cLwdGQ< zZ}+)(V3tF7*wpQ`;QW=$U8>|-mU^|qWP#5*kNcjmgujd82;Wo#Z7Mt1)dp97Db*AZ zVlymR|F~8eEQ?W}6&{S7$ZgYO=4-ncg#3KKJhmPfYnEiM;3yowzef|jA5+&RC=@ft zEJ#74YQx8uaJ?#s;mt66*5&qxMn9qPcO;M2^p=_6#Q!>e{f+N?84L*SLBj;_jQhTT zq1J%Fjp2M};iBhn$VpwSpb&uN&vvB%ANi0-=+1zK1SX=309*g(s?p}vp0yVe9i(@t z?6LAzjg559a#5y7$dyN5H_I;}6x)PiF;N+w&Osgs=Vkg^u)y*Y*B_dZRledf8lPlm zm{V+cmWOFyc^JEzzx*ZkNNijR>EF^a2NC~5Sc>aU_1Cp`YO_aE^fZq zr;NX^vmuPe#mYz2T~vD3`}g)m(_4^N`o52tZ`D*!o!;{Zf#7MV z{fKgG93uDfpNka#(J0d)MG~luN|zjS0u{gWdEVk0B52`mPWF+D%BlLYHx3d&)6+rA zo>j89sR{OdbNG~t%39{9R`;s5X~3&NjGOPHo?n#T-i%cxOYH8I$-QddyoZSue=Mc= z;*+OCR_HEza?3FK%ET^)z7;W24Izz)(9F?&EUf1%j3@oMBE84C`G=^CRZN}%Nl#%- zr>?S@zX?tG-3yfXF&D`B3txvBuy`pS)9ljWH>`0WZsLPI6VUJ?q(@R8-m2D_NH1 zw!zh&ATx97rJ3OJP~+b{#D6NN{GSD+22S~5bZ{#}u38|Y@JFU<_JqV7Z|jY#PX$Yc zbScbiEAC@EdT5>GZdaHGREXxc(tZniR+MR|{o%`M!-+HERgeXLg(V9OVV(D{YbjH0 z3cOceeHnOGlDwi4u~Jii({IpQ-K=4(YDM{r*=-Rd-NKuerlvJ zukgdPD2jV{foT4-yjb8BH4oVt;C(b@I6WrMW(2BArPJ36ZkWs<*_+cX2@-sUQOd$O#ew> zOYrY_m$+AyH$v#G9%$*x;jK)jryJ1GlS@zG-%+!j9Z)5E33(sI>B32`orW@Huim8z z{DnK(b<;OUJMjIL@$7DqTXeyoP?BkIN!9B_+R5R0rP{Xll5J6PheZf4+7*S&6n8U9 z$H>Jg?wCD9{TOx=1l6-g_I`Mq-vLYFvI_D%LQWUDkI`3{0>&r^@nz^EpI910@b6d{RcP=#Z z)tk+6FZ>QXh#~~L8HfD?cdV6mb@{g29}L9gYHiFWtKSM_TRqV&ZY}3LecN+zlIw|s z_}`pbd&gPvIHFF|!cwWQIS#H-na&}zCsnL;5M;!s)&0FNKb!)>U;z4eHZ^A|2B4|% z8eD*7(IJf9Ur|tY#k@e|n_zgdt~|T_hjp7E3`MVy>%%Fg;o||&0BMo2-RCXl`Zt_- z8>IBEG|TDbX^zCW)JL~p0N9_6R*$0P!SrfcTecLniElvN=GSR_VdixahoK?TI5UW& zOLSOq-Mg2;kf7?dOL~1AE8!)36DndG;$%{z$nUOq<@J<56q1$Arg}ENdt}wyoM|4y z05zQ|z7JtAV)6UierdPijXymhSv3(xuM5# zst!#UaJR#RlEjynj)Ac_#n6%VqUImZkb`xh4=lB?1?8C zDW-PsAiTl=Ucdi)T2xx%2q2puu98QxQ|N$-jWyxaH;mA!N#0m`l25rCIO^`0H`eqD zl|T#0TX$24C8`pAKHti9bmXRYpq(*rU7uTP1amEh7sFI}b@yG!D?J{|epXHmgb{_> z%)9tD`*cChcyDz}euS!B_!m6yYb55w1;VIFAS~qm-|Dwu7x#Gp=47mJmzbaTC{iTj3_8?O?Z@;JrLAkH?@T1-Q+hCQ#(<0&t z@O6wASyhCB@dFj21I1EiW;=G3NwwF`)lni>mR{^vzFmx`@?smzo>4??GF{rfcKAV2 zDRp!m`F3H3x995pXr7S~x`~$N964kE!#D9S2L4-}n0RbB<=*C}n3m+}>A@O6djT*Y ziP8%xRZ&X)8PeZ-m$hHAB-5nR#0#{mj~IKoz@+dHukfGg_MiR&FDY|tI_A_m=G6V< z9PWcLSbi*3%(}gXwSEMh0kLD4-3x%RKl0@ zh=%wFMeFqhS=~`jm@?JLpHF0xr1ms5#?Rjj(9n2cpJF4h3}YI=7h{X6dn+8NaFf=U zhPVYIzv{PMS4b(}?^iMU#sXg5rB!oMBwx3@k(Dgve1fr*hTFXil*n?H-EuDDtdz!D zD8%4AD?;)o*5y05VgJZx6L6)@21}j+iAzUF2G>6%aQGRH{pM>-&X_<~roiMxt$rwvt z6&?WgndlqO=f$$jCkxPh8JDlVr8YhTac0}W1Sa2vY|An2q#2QTAm3FxlQya1zg9d{p@N)`;JR z--;V}hTsK+`J9TYdFq`&Uy2*FoUc`R2}zhEw`%V=q-~f`l3vKsB^TZzZh0Q#X6NjX zcnO-4tA$x$^`8l`UD>KdnNCu3#P6&2;3QiyM1yo<@zakOJTq4l-B=jEMt4hNSWK&4 zzYV`7=F}JBtwnY{$%8*qsYCYe0WhA@89#$Ndx~H0k%y2HKfQMuo7rhV663cBT&MYb%Y6Mk-54ZSWy^uV|yTbj(TyM0?^dVdLDLJq>cD#v$G|V;<=@XcCkO*p=BIG-#XL+TwA{ zL5Y&y3n+3{OtjdSk}%NqWFWIoh8S~tdD?_j1Vd{Y;!^aCB)gBl3Jv?{GrF=3EeVgp z#RHUG10VAWe@N9(_YIVOrgZg=`HzNi%+;?T4)|`Hwn_VxduGspdA+|CS}?Qt^YZGj z=^L}x>J*>KWnEBdqX6mh{EHF(UO%a#q^_N>v*eaqg2t9}w?;J~@-ECwa4ms&6#h5J z)2T2&tg+S*s%&K`z@H~_p2KEldZ+%o2Vg0c3I2Bf3<$&L8$#kPyRtntG@PugxcvnG z{j?b#3O{4-AS)gMQ;SDb?k>Aj&tE+Pp~Ht$2|YmuMa6JSw9!<5MLm8Navy`9{4_Y= zCK(>m+7cDj?NXmTwjx*1Tz1@DsCg-x5uYxr?5#X-CaeB2P))H0fg6m4PD~jH5Vq2Zj z%vd^o7b|>br$F-a*a$Ocri5-jzPQE9=R^@-kwCYuerKfWA>&8M5@nus%cjDs^vF!$ z*J&t$jp|M84-3|J$?V)yc*csPFI|46c?SG=hm-ozQF&Psw5YzQSFC>WG#7UaUbZP+ z^q?GEio0qtETN_D*AS9if3xAum8OE% zn~--0KPC*U3|)=N=M|mi6NDV@Z0^4BeKx^@ZJ8V$sc#2YM2e7R9w## zo*LLt+k`D$4;-w2m7m)DP_A-$DsFkINqKRyp^v9p!j<;Qz-&to&^C3K`VR{5Cq#77 zYo=N9bpf)6dBN54zofFrqKCb?*Nm@WLgU=U51oG@a|h7LIMuYt6>ywb`$enm&WY($ zUC<}`(pGGG{vCXQQ$C2Try1Ks_vBLDe2@tU?P92M92ENEC)zhuEYx$4u{IzuW1y%y zx+GY6YM;bI#5YRfR^GnNJ?y+V&!xwtaS_LCN z-vhGm2SpgN-wpv>Of&JG;3`4K9(cl=i7Ji<< zw7^UH+GhZ`f3NIlOVeD*Izf>es*VyG>GLj2n# z)-fR_IU)U(H}L&`oAE#PbKn<#2K;5^c;^0Nk-vQpOWL%^3d7Khdg6-|e;GBr4yAeW z`_B)o5Zl%9(#S_Si+3yxdcywHaSo3892|ZSi7&k>Q8#7D zm;DN%+zVecg>(sD7AZUtk~p6UhM!0uoriVpF~|FvrKjMoB+WFElqw6Q?}!uYm= z(t%&*SyjthZ*K65h)p#ypmFsbKQEp_6uxIJoDz|)^1t=!*=R_xTvf1JmIss0^eTS& zWiP|XeZW_D2B?63!FLzpnnPc^=3CZHe*Dl9O>{hxm}yOo)W?o5l~-JFoYwb(JofFl zs304VaKqqv?m_voWR#|;tvV05+tSohsqXQjrHbdUUhRP6jPK0UtH-1%e(dxkgl0q- zvG+;(%XbBwLy0h(5B(#%ZdB<8-i%n(XX~%gj=NvLIr$P6Nuc=laem4710++R&PD4_ zt3zQsFdI#ru>tY&SQgxe0hoj8<%h9D=RfY-e|oUA)VdB*<1Z?@ zN6%oGsdbN~fZzo_lm6#i2i%HAOUn|_fvWuJuz?o-oR1sdWS^K`FALuJ`D>i5oJUkF zFKYUJpbw$}5zW4#=-mtWcf_28EDKB9G}c#-vuPNhgIn?EP(I&%cK0GdDX!19|1<&b zgSFi08q$1rK~a;j_e2@iIpX5R50smAs9MN}r?6bD!wD5OViOQ^7(Zw+M`G%gCfqZR z2=C4QR&kch9vd4WeB{(C2M65wf4h&aYu4vN2;R;?6I^^4DBbz-RUxW;{MovjPbBap zKA{d*}o}A;g=u32gE_MLCG?lWvWn(jilPR z6Cm4w#HIsb>BOeffYf8ANG#k(5RE;Qy74?zR4#Jq!T3E`s=kg1%`+v@lpZBPxyL_W zKD?6R>OIqTtqZ4sHjAp&5UyM`kyk^BB|8z0?|ovC{KT>?JaXN$gpko&s_Ra$8M=33 zol^Qv#t`tKYKAr0KUrhQBuWK;sJ!qiKHmIqgR4J+tpC^l#dQ3WZ` zKe>z+De&W+oQhu6z5GdfW{nlBj4c5#X`LwH)iDTP{NGH#$>N9dc61JdW``0yf literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-19-Normal Distribution Curve.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-19-Normal Distribution Curve.jpg new file mode 100644 index 0000000000000000000000000000000000000000..e25446b463b073c629ebf73c77e47355fccd12f4 GIT binary patch literal 16019 zcmd731z40_*EW372$BvdokJ=zARTfKDV>rE3IhlTNVkCK(5W;6Lx(6`qBM%gP)aG? zog(pFgWmUZ$Me3=`@hHk9mfa9HFNdsz4lt?I@j83pM5_229T>Nswe_jSO9k57P)M@T?GaNzpc8d zg(9x{7c$*2-ut<*MeeS5VGH>cuS?zzsVtN>GJ3T5j71h9X%%(HiJ$Z_Zi%&>QdiFdyB|9e< zm6u;oSoG?3WmR=eZC(Al*0%PJ&aUnr^uXZI@W|&cqf^r}vvczci%ZK}+dI2^`v-^L zk1+kh0&sqI3;g`qvA^|;0_+zyE-nr(0j6JA*dE}9LxGFODuhoduT5~r`4Zc;AVSEE z*cTOVFR%;iY*Jae^b=8Yz@|92FkL(E+5fC#kN#IZ`=ew3>enc65eEw#JRAx@7C1P9 zW(N}8M5CD!up$Ybu9k^>-@S|cX4=w_jEoFhbj_GdtV-D_6+TtX$Ug&2JNL?eW5fdl zAn8j8AR7t&%a#Mv(ZcSBLjY}MTLf@^G#}BA6q3V(X#c;zaSCt(0t6@sD2N4Po8v@j z&C|N*74s(&L_1A?>ptU>tt9#r7967WmrX@Y1qm2~@+3$1!vP2pGz%N~KizmpK%pk@ z<-mYS9qWis*!?OgrvLnfzyIR@jbr&GH~|Xq8csd~I5l<8fHJG2k!;dOjT<-5fVlj^ zGhjsM#J-an@G5_8ThwUBLp^i{|D>?xEuW8k*2;pW(?mR8weaWlXK1=K&=j1wI4Q z_p;V@Zc%Eb)91)hPed`%0-r;Mok|4^(wgO=Vnd^BdnWdGX5mXrznY5mn&nKWFnQ)eMbfyKqh^iB^Oqy zB?Luf2VyK3843HpvMvF6N~LuPL!-46;xc+DSw< zr`mG1trrytcX!yC6c?0y^P0O|ki*T@?0FbDQ&?3hzxzX+EdDZ;rPy>4g{!;V7n*2a zKiBffru4KB6Fz^-+S=O`<&4scHNicmZk?{f`8Z8JM!pir2I2c+iwH3(Q11;5+C=#S z{9ISeJdpCD)VBlrC`{u5N-Rp>PrySR+6;Hz}v-Qoen4LgtSsvHDJnA)P zY0vnWFkc}-a8%+3GfAUwN^jAvi7vdBZ|>g=!;OQ3VK?`%BIeUmo-lmd-_PHpE%Pz5 zD#*k6j4pGPBRv;od=ohG0<0LB@j$T<7-43iK`*8v_e<6whEOK`6nFwqiLp-|dho_v z)GyRi=0RZb*z&l74i3cazxac5bf){04v?urrJ*<$#QJy0o9^_FVbxePT zGx#7^N^h0QeE_-EbT;(PTO^TCD>8aOs)}4;SfG z`<|?jY$unSnQiJox2vStM>Q3mi9Wkj+Jka?;xKJPYxOV=$KwJVcITBPd8S>gkM{M` z)JSMa=9Phxj4TgRrq8QAh=;y@=n!@v7(X36Y% zH-59)=I|TWDKi(og3l*ueskOrBLkDn!&xJ+&+B&{sAxAZMtZsx)++Z4Ta`ldcIonQ z+AW0;IWR434-gsEx#b0P>kfi4cz#;77@Sr9 zR}|2~noPpzBIMiyzyWS=U6sj%PXH7~L;|p62f_bn%zs0sl(A#b1Dd5*u{M;wAWh3n zb06HRH(*gzdQ9p9EnQohPXvx-bJbkg1AD~qn{N;8Nk)^~ZI6#rLm5`P*UV_zWNFl= z1@4Z%n>7*Cf9c6#c!ko-K5A%j3TieH^LdVzb{CbM)Z~yD@mX8RjwctR>oDGP(>^^n zZCTK?W1>$iPwdOL439l;fmDIQXmH}<5$zL#0)g14iLqGs>Jz;8C$ zcQ;cVj$4dI$}Da1d&0-PL{@lRdp%!8X5>8$SA#=>Kuv`FYXtriJ^9~U|6h$N<01xN zU|I+)JW$NEycX-rCA~UF9eEQS+-Psos9sOou;b zSZfM@eAgKujGcw^QQf!Q%|qsOp-O>S_5Apag#jspT^?~EQ6*fX_nD0g&PRi@TSz>r z^&7mz3X-9qrrQk-wSh{zbNz=~K|TJ%5MsR0=9VX{2PO5~+$uK)w45(Xe8;pQG~Uu~f#Dr452ZBx8WmWBmvB zzC5FwZ?<@dM|3ozwz<=;d+2InVw)=iZw);n+ve-;(0W}jAiT6JtiM(kw?1L>G$^tT zmp;u!c(if4Q)dA5(R4X5Z8>~IFQ`92FC8c;kcedmUhGeSX5qlBK?D5#B?&D

    3$rCLXT)cp6?fGU!Vp-q4BpMRD=W?!1;9JF`;OVEA{#qvR;buicudI%lN z4kw5t^s2^yJch^p_u2)b3@0Qx3mEcgNT{xkyG09y3v?HDw{$O=>u+6Yze)Enoclw` zlm$~5Qtg3FsxN%&W(LRbx_;z&$Qo!BS25#-fe&KlqxPc_i|O*i&DS8oXRE_K1B^h_K0Zfni@i1j z5vBrCv3*o7tm@Tgj2ps6s`3Qp#Y;^`RWbz-Y0BX-q@7Zk&vV^r@{IM;HPD0&ENP4J zC~4KLW?pDS2(CU;X^cOJDZr8L_5Ox~!^k%`A@RSG-8rgJ;)u$%xSAv=%;mQb6E@ypJbpaC55Y^3R*f(S5Rpvwj$ zi21l@pxm{=umU7p;E03#8Dwlc&Wh*9v>;#0MYV%xk*a|4IabGkLE0ZXzlR1S5PCrM zmw(0#wee5?e9u|)+^Qwt+x%5@Go zM^Ns)+!&B~s2rsviRaQ;2KS@B{7szKmJjrI2D70MZN+b(!DEyG`!9okhB0{V zKUCutMm1g)G|dt)_~@YNjWLRl{Q(na7~Ez7KM_ptv}|ojz4u?N@-cw))6@M^6Fe@NO1kL(A0|Lf|!1xRbcPr}gGofafx_``jXMn{z9!)k)-i$WvmQ+H4cA+kUTDbvH~#;k%;40@$tfQ>nZ&1*MGc zeIF~IJRwrmWuGk4eOcmMc)5AHNSV==ma22RTn_y()kJ{a^;$_>R~_@}2#K$#xpLM6 z5*t6eXjl+bXi`~=fF!mDV#C}1lY*CV&=Yr8lGF9O3iN$|pOae)FEB{@pG5b$s8?&GLE0V`jv?XE5x-lQzH662~ zq-u4E?e(661|wI8TH!dJe1^JdoL(ex1D^I^JaRMhiHyKAi$T)(+aFlwICez@qJ-wH z1*?io4;w5zEdBc~WNBsL>g3zQ@teYsj{cZF|ar7R*YhDl~P& z-7Aw>AC~;FQupFsl!!QyJtN}`2Tu!Zx7KZ3c(@3*N&|ygvNylIoyD}f=B|BG!4dm% z?GYbTGVCM5{hIQvNK)*&fu#cnyts*87Hij+QHP~sITbDnctWoh%DtyPe6*s7Z<$A# zEX6VNm>^Zt(_8#TvnLayX(u`t%&s8c%raY($<|SIvkLIWkBAH(7KihJp29CoDb9twGrE2mTH{MCiprELrnL|@WqX3$mrVDWL`+O5Pf*SQ}6puQeC)h z^cHmz+Bhx8tyg>N{;(o5C*bFWtj$eB5W-nFd z(mbusN7q&*&X{ElOf@8p%Jod{QRWx)22d#2U%t_gRU74OWI8Y#J8$hF!fe;i$98MU z|MD3SLw*LhyjauqbUKw&IO#z)d(dpX9&Ghx(ZI7=9(A#!Za2byzk{>Rry!P-Z#;K% zqBGgf_E>l;q&hy5Su19p-~IIifRXRZjN8MD`79B^79ccJ<~}<8u$C@CzgHgCd@cG` zg~_e2G!#=ZRuRL6X(D%ecBS!=w`C(GOSPkB3ZXewVCa|7Q~>mHfn1jwsfx)yvgr#TSIKpYj2*;Z;}B3cad)BzdZUw1H0?dRFwi3vzf|6sk+&|GdL@b@hDf40SQ zBE^FMUF;1679wvx@?9V!oK-9rOuZrHmo>X z*S8{XJ%4ocFH&E*NJ=vHfTkQj-cqI7^7!UfP3&b`YO>U)hAibQ0cj%_a&0GXTz)E^ zcxuf(T)gCdZG1xdWJ!Rp;$C+jeoeydZx2hxB$srnr}wgB_D}`O4|DNLz%Zs0?FTm$ zvFtpR0nIaKi>}Y=+$fbmANca84`c>?V0QB}l1(PK<`Z*u0u-Yr%nHB$0;#h+BMEJH zP7^h-mvs?osvQM=Nr+hL6~tOmiFhEPeOn=(_3c4fL%W4&x=rTyX}(y^z3{!VH|-YN zWan(UqM%AaY%&UWbjw-NP)Qc?g zN8|2U-<*}C3tDg{1Eu*|1Gcf5*Y)(TKRyFyab?E~ltgws-+-Yl^AyodM{#MyM(psj zVw)xx#*6l3SFNHuUYe~-i4^7)B9wod3u}6ad#_?LI|H{iH2VP|Z172<%w6w|kVbW< za@$ym+8yl3mZK3BbhBN5mCqXjdtw&u)%#_}vF(V7B1S+jl)A~~YH%Kz77n8=;`h)fSodto% zV3rjE=zy8ge`(&YSc`{D)mQ-?DE0R#26KQ3n=&?4b}#$WD+9xzzfsozkS z=uqn_nTncUccfEZ5GkJ$5Vcn(Y1?|90>(gwUl2GZy2Jdt1@tC`?zi>46>FS|Y?+-x zTj71ZyG4`G0jnl--Zkecm6ETxqhXtf5Z^j}u`J>k&+3l(Z&?ZV3Fr<=nb$lmw4y-Y zoa{j~ZT-nYEi_5a;M(j{vQm=)rMKtPGW@mjCmS_5 z8b{pN3+*mf#8J7%MhzjpEc>(6Ip}H|8mY7>bnTb^U=&oz!H0b(U@&m8(TB1X6)5iq z1xo0?BKoFZTzsK4>_ZrpIfvI>3bwhFtd-_qZ!&LyKR2T)>!H{1wI7=AhhL~mOl^-5 zG8S?`y9%1F`sk-#I!(ybRu%)y+%-5gggl4e$ngs<5ZFT^jn#pq5jOq zciQM5kLu(t)WQzQxSS>#;SfW4P*HS2;NZcKckDk0A?I0#zXu_O^>m#~gOy=b0(RBp zzf*Sb`yl}4@k7385uiLDoB?f5$n*;8G={E(5gaS)odFdZHLp`>FMVM*^{37|c z0cGC~_*#m2OaHYfCF9a5r%UFCl19`A7u9piBPtf9i|^YW$k>{N(XHrJnUVK0Cu|+} zy{mE7OONud(2?#`8!X*ZgVopgb-PHl^?HPrU!0Mr{83_s2i$HgLBh*GUp_fg}Z<3a8QV}+qmnv4(q z=l;F-e&+16wKkf~(tOYB<2BIhg4AEw>+2F<&&7U4mmkWiW}wxI3$27wgBxt_paZY( z3FwTdR{4Y{ix$xJ8ltPl%ycz=Jfw6CEL)o#B17_19>D7+ovX;GV!yq=>dRkU^nm_a z9Rw>YL*TW2pw!zrq0!Q1d$|>g%UiaWaRhO!Mr+@{A!|y&Cg@TgbLn38<8=@{5#YSk zr7%HCl+oaa!1*NKMYn>4(~4+n46+WEHhwk|B_)-ludUJk$4cIo_{0iX5ttyUg@F{N?ICDowk2gg?|5nB_26lUYY9YD-0V6@eDA|${T5zVIWm$?4O>^0$#l6##S`p} z7Yo{i7U4#Dv2uOaqThM!AGP;>P1~kpelUY_qu#F%OL&2bv8=yo8RINh^5J`FZY?^4 z#;PtMA|<%h9Y+nFH1c8c?bA}eY#p?L?jy(u=T!QQ`5B88I@1K=7_u3;-bh3BXWF$C zNo*9sQA3eMo{kEh@d|NNH@*?AZhGUWMJveB9^t2q`V|*b-iq>1tZ1mnfrUf~%dvjR z6@*ZF6O}u=kyKjWx~m7DjW;)(x$^Y6$4hjOhwPPPx+Mh$wz-N=o_*7Jc~5bZHm|*R z`C^u}P2zQt*r#lnTWc`+i%QqiA39%_;=h_+?o~PdlA2ayIy_wfANy?Jkgm5l=9*(` z5%@;L8MBJKzKbeaX{mX z)F-7YOL~>G)Fwmx&cVZZ^>tB~t*k??`n&KpRW8}OUE;y@XlEH5sg*;kK6Q&D6-Aaj zBTiI|VTT{fPUd=4hh|LOHLza5sk(cC+GmE391s`ZxG5qs&l&z9{6p|%CA(hau#uy7 zvL-+8xI{>RT`b{f<~uu^`Wsf^p6(ZA?r2E9P@58yW!{>;bPj{MGt){>Ffe%k_Iz_Z zNIFbUS;5(+MUFpMST^$ZpOfjBd9s+_nZ?!=7}}e{@8~`xV6X*KAa>Ol>Z;PRJ22J>2X{eXIQA-61N4Kge=PJOMapju1aTFW>boj^1$6M2_QXuroHK( z%l7O;N>@w{cj!SGTCM9inYNa-lX%IrX|BZZiOz4|HRXJCZ$8z%xnD}$CThHvt?qD5 zh<&}VG;;ih;fH<0l*|wjEP=GB{VzE0HHKCvAzKqsZ?HL{UiMAU?ih19*EVrie zXh@)W#<2*HrUM*1oSR86@Jn4?=ev30ZoKcDD9igeqGfVjF-h+8w-Kd?%mj3B=DQfW zUVDzw%VVmuUEP)O`i;_HQH-^*K!ZwMf@QTuD0h>y1b&X2^UR6C@U3tK`beaYTOR&? z`5nP5t7tE{0ZFED=U}76O2Osl@+^5Dh*qzvyq!-6v-2f&u`RdOL7_iG5mQ& z*~=WksD#40^94+Kr(m~e@P?k2+U6Gse<;i}ZuB<9k>V@y%wFzUm`AcF>dG;uC*?a7 zaIJXhpC%$|wbU3aYt1M$GG`=<38Ow!UcC{OzEU#8k~H4|e`FS2xpSz68*68<(>vG6 z_L@vJ%E7v0@($#6cEzgwlg-Gg-s*e~k4xVp)b11sn4_m9s|_o?lArQH!NMQ!%NBA^ zXlvinG3`mYT!36%wC`IjGxm8x@Ac>{d23&bY4YTX0}wxv(ThD>~wk*J)!j+dv)Rcv}oVn$kqgD}VZ`iy^umhd$%o>%I4R z75xjf8U=L_pdNOGG}hrV1C=TLyCDagVUVGq;B`F}3}K4@3#T|I_O%V&uM5sZ()%AJ zu0)sONJe{WYrS{A$d}9kXIeupxb-nC8FSo3tG*(aP;5arJ+)C?FSyo;GdLnL%}l@K zP12ye!%>3o&hBciIn5mIF`9P>d$h$fBbZmKgq%Kgm0k^<<(Bp=aJ^diG!o`Ma(Zp( z_@4i8lCTtY)YvTsH!7;;U?=z0YvmV+lR#|)}RNE_bm3VcO(J!hi z8Q9^f#Mq95Vyw`09ePPB;WsMASim4CsV!%urE#GT7MM6AjmW9OehJjIEc>9DNXd+a z)Y4`~gFauNeM?tN%}&D;t*yTkZ@}+auUSy^co7*sA>POAI(^O<-m;2Yi^9dc>!TL(!T+u3&jr0(TMi6T5TD>fPpAp5(_Db zm9|~roVoyvA;JSys`Cl+v~_wQSf&*Ya!)Qqb)Wz%Xkre)ZaK-$3$g=V&Kd4u)aJ;~ z6q~#Sn5X$6eN*B9j>GUt`QxhgC~{7MpO(FeKzwXP$+ut5NN7O1U}HF{@?n-c`2wtN zZyqUh!?HT(mj0(=QcdGN3sO_KP%|23#I2n^Ih5;F>L3=7-6i`)wx?*GHdT>^&r^50 zu^N36d~1g@*2L+#CoO|vneFIM6pFH{ajWi58vaMVxgS4%%b32uH|gzXJ8JN{_|RRO zD^q5WfP=}l!#jgfO4a4HOt~rPqi9pNa^}s24mFQ2PKEq=x*e;x4fAY|PBRPIUcsHo z$Aurwh~!_BNEob!MURV@%Yuo-gi@Q6{uBE*HUXgU0B zlRPGOZ=>CoG_tPvh3uoiiWn?7m#wf|84phB1MeQq>=nvo=Xcvrqkp4gYxr0hN&oUK za015>Gb54=FEySqP2$#%lOp`=+7ZxEA5Oee@I~?o6Z@m9)0H81_yEpk4gv?j+RY-> zR34ZOzjf_eOxk{Fx}B}%OJ}0Ei3+0;@7+z$QE47!3Ybp?a=C>xhb>U-@&xBvq9Z+* zkNaiRQzd@8g6~}Zogrz`CiKHRIO6RIvCpr+>1Td=t1jcS$LMrixS~vx{y^M?h z<&-NAk6nY=yAr*wtl;2YKvUUUkJ%yk)ceYUd{Z#<3e8RyDY3fj)KdYNlC-!C6?n@BQ`OY`*lB37DIE z!wW3C8cp(@0kQ5I(xSumzBfLR+pGzEH&CqG!jXT2b`;Q=s*(8SEJuHOfp*|FV@mK`Qh8f07!YaB9I!`o(Z5*#-eD`n(-g}50X{jVx)>hfpmDifn9(wZuodTR|x|P6Bhgn5yX0Ug=~kC{11YPMH5==#yRM zHbv@+=|(7r{-7>5vxt+OE6Jq|68IZIK!TsKd6s+_|~grnNMj^*+r7s zHk>`KlQObf+L&}wE9&8?l$#TI9I3y&X|rc}POAzuL+^kQ&b55q;CL|zkvpNdEdGfY zbFBTbA+R#RF?T{q0$`UKQd2=C8IraIJ`gYFcOfh?xXT|Mu~$;0Zie+Dtp-v#F*#UQ z>NWIO0Fqh&KC#Y<$Xnmf@u*^0sU^+L?W+=mbHlLYVb(FRn(0Pl=hFq?3;{*Hr;9#4 z?W|u zc|rEu4?k}#WwP`B6yr(;7?h<6J8=KV|85|gl2KKhuSr`SKhON|4B({~uh2|MmD*ll zTedjp+bQUp;mUmQZT$V|_ft3Osg&OF?>sjQ@9K=^Htx!d>zKUv%lp7tnQ=2XjYh0+ zo6I~Ra3;3vATJdbsv7fo8CH(0pL;x554BMutsRl;-b!_)(ogu#3<KOb}n04G8t7w6Z(l>soI#$UnlbO&W9t$dY{5BGB>HAx#?*IESJ$AoKBYfyEPgVo|(Sn~#kC(Qi~JD?zZP z3?{K_8n#YOB5HPOk^|tZbr3af?$-9J0O0DDl8MC`5UIPCp+dZl&>m6aXt{dRj1RWAv*(n2g5gA zRkfq5RH9Mdca97tr-@C*s=qk#wKcfbA#f6y@psc0#=B&RT#lL0Eg%8{MI$GO~ zH1ekS*U^egLB?(5ezupoqOUU(!rleUY@43&eBI37{rdgKv+Z(azy>L_q1t5{yXy7k znCImBssG$@!FiwEewe;V^j72%3Xb#4zsvMdI$9JwFP~R)-1PptPG$)G?am|29bz9z zdHl1(S@xY*y;=^{tmsG>SKqJmbWXuj89TLWIgUo;r=feqACwSgLT~wuGAM_gqKe(} z&GBWFE*?qHFcwf11KV)y@>tVR+fN6L`?vo!wK2CIt*Ik}C zY|OYzhf?z;r*$6lny`5ea_5{7&HZ4}B6~_te3OIr$4q0aJRn zqEbS>MeQwyiI|_>VyTQ%9&qK2N7@nyOC4E+c)PW(#~s-O=&GCnL>_8m;B20bzkbb> zpUaGX#OTY^$9+aPH&|QDfnYWXW3bf`qwze`MtUhucKzq0Mz^M+Tahls)9CJ7r?t)58aPj zCrAkZU-xEV-J&P@P&dW!_~iX}LsY*~<+BG;1E-MrM9{`vPWAq{sp(qG3FbL!sSeHH zdM!n^bl6WJ*ZTZ^58D|6<_L25ymrdP~cZ05ttG0KSo$3qFuAdms5%|$`^RA!$!r~H&fc8zY}W8AFAIC zm^<}eE@?E$ws=1~`jsPkkw7*fcMw*iKZ2LjwNIBI{eHP)h_8YA0|f`a@$PI_<$lO3ILaMoU31;7;0<- z9^)gnFEw@wIZghlXqlS@u%Do!(%FQo`x46GRfBo=PjnJ#CbhLWHNSgYjNZ~Rb24$~ zd9TYDyTAtmBFXl+%DS`iSNzI6sqwC_+6HbOgS+9owYN~Qn!R)J?d}vS)SI>s%*>Tp zCsU(*w1h)I#o)Pnrzrl0AA``_K@j^XV6w2Vl@V!Ixz~N$pE98j(eI9eh6=^yq!Z zARra&o&o1T!3?SWc}eN#C9)8AU}0@%Bxf24d}BYn$Dm3rZN8@Gj-L+7>T|#+>z&gh zk;#5J2IgbFJ0Pm0_WfAID8wI^z(pg*93wT$}yh{fL{LPUb;Mx|fw|_dpyjdD{ zg6|}|kz=+FZ3iA${Bf-GueXo*fGcJgX#R9UeQJUnA>o^py{s0woqWB^MNy9SgwFyj zZ^X>Fsi=Xq-nr5|2uI8!Ca2>>lA8y={3651b%&{%;C2t=x`g}5a>Pp`$M!j<>mNOL zvk0(-HS+dg%fL--HRYdGsdswj>VFtY29(%QAMn3i+E#AP0jqh$~f;*WLeX3ABEZmYgzpyYf0Omhgm`?8TsI5U9=*clqW^01F=xu9VB zUP2JJ|6YE+rz)eBeBAuBzE`f<*ppfM)l2V^wmNeqh0&<)dE@F;kCd+9Y}0Z!89eEb=3 zL49qoW_z7Knc~f6ighsRmhGfeEm|tXiV~j;)UZ%Dl_|B}>-rhXJ+n#-IaqQ_29*Is zBGnCorEa|4QsI>~0Xw~b8#G8;j zl;$;3WlDCKq;?uWKUCFNS3h8F`wbD*+Um#WSZr})XSAEap(csk&7Xo^t`;V|OI$M< zDAr`yx=+s?zIJk>0F37_w9|zJNLOs{a{c?LWkx1&69WtuLYYgnncH80V zG6#BF{LiQjEtz?55bkBS*dFR54^wq7zq$xIA=TXnp3}7?*Wb}6MwK{*F19u=(5{%x zs_I-VaZ&OH{1Mr~a(J}a;KGauoqlE9vVacXL*p~xw2pp}3_~gy?*>iQ7iCts zRXs`Fh|Yk&s-+mK9gPh-K;UF--V064M9vTRf6P-J@JRkXQp~$1Cov*C1rqIa(UbCZ zQ#F^BM<*)+_yv7`&=hQ>N5Cn#jJa&V}`Xd#2ZMQ_~#@=CC58p{<^K%#3DD2^(bupC}rZlA8q z?0&u*pbE@L*^0kA6d}7?bzAVv!uadm(ZN%=KUXO~!k!M8}7Mk;H z5m#l~wU$pwZ_QQKzIH>ZIn^U@BDV}5->;Nb)8M0&=;b2BSuxLi(F&2ce}jscNlbgR zuHg-_h24Ru%cL1Tb#Lt>i4iMY9$enoGKhQDg=koIig4O~{wEB$oiBGgUz-7;2iL`5 z>b6zSCkuaHoAqlp1sm*-fzMSF7cbm`JMC)X5c%43#wj-#QM55567ZvQJATpjbqYT$Esy zzgLI;{Kn@w|ItxtD46>Kf4v3umzQ$kyQ&3&1NIz3h3kU3P6V#T1Ri46uie&}{#WaP j988J4de(kc0SBc7*uV>wOoLgVEPsmVfA)=a_T~Qob!w{x literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-2-- $A _cup .jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-2-- $A _cup .jpg new file mode 100644 index 0000000000000000000000000000000000000000..b31e232b36b65819df2a30266b57da86823395ab GIT binary patch literal 5538 zcmdT|XH-+$w%(zIUZhDAq!$rs76gJIB_d)Bh!O!6ktU!bT?nFphF(mwzu>v!z5`>t_wIdToHxc>87qJGx8|DPoO93ptpyH&Uje~m z7H|sy0s#OB^9O*R05gD-g9FOJ&IyG=xwtsFc}4hmd3bmwb_ffI$VkfWl97~_mQ&oT zA}7B`L0Ve%fZCpY+Io6=vMPs+4(b^0)z#D4Gy>t`;^O7u-O0zdQ)jpIZk>O7fNudI zP9P5GV}-~AEJ6@gAqYqUWSM@lLw*n79|OX|%ErzC<>cb#VFJnp0Tu`=D+?PdJ3AX2 z6CKWc4zLNa@7S$ zuBX5Mz`;WXW=G8}EUnDqO$6BbweYmskx=K?OjiAU;n`1(8uAi@oy88Q{Tzc zGYgAL)a8}cwO{L-ydVJUpR}0&e-hiqONhyfg^i7s4Z6t-!V<)stU_$;yR|uX96>-Y zToabpxy31Bnv(zG4=x4WQxwsQ*SopJ6!pfG7B;Evd^kbigR3nFynO-!l71NU1abg2`_7Lp*8f2> z6YH8E?13U1*pu$Zu^m^64wBB>WBI-VP)wUXTQPr0mk0uP?wf%?UNaU1ZmBW!&<)o4 zAkg*OU^>%aYih446O$z)DI;JLxF|V3_ND_{b8gH+yDytzUkn1ApQAy*#0LffkTjZ_ zv(sCbHTVtKIE3*&2uYUmXbhrp>p(s07U}U5cj@{XB*yQduXs*RwS)iP_@=sH)-->$IK^3 z*tg@ihtf4T=*T(WZt7Uttw)z87!K$)`*p@)m+M3AvE%2l* z>4a~`Z%^p|;{t#z7)Ke*8U%wt<>srT(NP$(EMd=P6F z4G9ABxi3J#8HpxWlteW5+Iv6ecR>xeT=Nu5R50yfU%Y%Ct1ZJoBpT7Z^A>Drn9vq_ zhz$tfhOuMwzr=1lUteO@rilpb-mR3eI_`O!1+f}*C~s0=_m!$W{1owIC?ygOghHHb zL{0%XjOw2=Ir*_-q>+t#=X3JFhtuJX&B`)nBXY2gNIg%9D7DLSsuGGT&(`$ml9N$? z6rFk3l4pHK+VJP2WuMwslha~o1+19*zMso}GGuL@u!2{Uxp2AU3@+oL z4i`wJt4e(2;nkA*V|{!xBTw*0DJQzC`MDz_&Lv&DjW`fPjj4H6z2C~c>}2MA)zm%% zNV|1-{&aQvd;L+IAyK<_$CD(}BN8+Zpm)A~;(q6A$d5P zoH*D7TU48sU7kF^i1R|n)=!-6sh{nTr2|OAE!iimg)z)?G#IsG%)ZHPz4oA8T@@F} zty<#R?5tBk!KVUPk3RR-iVCPy((0pbOMfeyCnfDpel9(vFu^icb)}DqQ1KU8zD&>4{GQ{rq+eyBSqv>c^<;%n!3vG{&GK`~9wR zudIa9S)(d0oupHBQk169n7DC1Ez*9Nch9^+uGTnfj2Zb|2z1s3mI^3lL)!F{Q=LEC z=r`XFP`YzGW`EXxbzf+H3uR-OT9OqUyV5~CLs;p&x}ZX$>%z0FW%lY4y|@~m+1$m! ze$*e|;LTd9vu(;%8F)Y^IHdLngTS6Wup2P(jg}?Rj_FFyR-brzuCcLk8RN%&vW|6# zQ?;L7c3~botaQt_n;uAw6;#W=v#~UTexPh=Ff1pmJk|t9j>N) zcroPFQ`ZqsQa}FvMrJ*d24Ba_;+?GFzJcAO!myVkEAP}^w5A<>X))IQ*3In@+b-%1 z*`R@0rhjXB>ITq#2OF!E1_}ew0W1@$GjqY*Ms07O164IO{pA`7D1^^orHW!;0xVj5 z2yK`(%TV)%A{JyS?j}6)W)~NRbmK2XL&-X5p>yHa#=XMtT)jvfU6jLNG?K@;f0uuO zmajQJyDTLqF4@TTf{jLkM*7cteSUnHRLyUHR4J_03Dd%h;CAZKCs}SqLcNLzC@GY_ z!=a4D#mb^xHYkEhH@_`T%}Xpv6%}_=v(GS}LHS7->Kh1xhg?aMQ!uhyMvh44mYocb zJLAzJI2!de)XIS2Dx0BCEFB;=XNVw9njp+YfK!4y1x@5ANAS4F!+tTwDn16wH?_ld z@9&v>CX+N`bll~{w*}GJjHRQqYUZhFr%Ocnc9kM zYko>-c#Xg?H*#gvJ_UhpA6v>QEzZD!9@5B^LEP?t(QADmAb@Ty!=neYVe3F4T37;q zM2NIn?WFe1D&g3QHGsb5HSG(_9CIR zQl|bY_-PlU=YhkdnwFKPylnHeM1_diHX1VS3?$h1b;0pY`SVPfq+l=8jTjXxD7Z4x z=H~mo$bvvEEaYVFWBNPwRF~U>gt_64^p?1==o~1n6}E==S|ZNA5otv>ucw=gx%FnW zE^8>mR$QHSx_b^yH`Lg@m`g{P2jxEey8>xf9xOR)USH|#hzb!5KbGR)ktgurRm&H= z;{m$%Nwj+9iJDRai*lmpQBkqV6ADm9$cqtdK3%PRvfa^Um%jOXjd}OJqRDVL5;@-CS&QDueqt%+birh+L88dAdvJvCk1O4?;~62MQ+sS__T!7axwF&b}K2vk4z7y#8y_B zCK}f<%9j{;ymPR>!+mDRF4<;r-5n3E*|$1pS$zGXC;ipc@j~y8yUfD-cF=V53yrI; zL=~NtLs!pO_TX_3-Z{E8+IUaCCsD9_5kC(z6rc-5<&VOohR-Efu?+Z(s zbQsS(G=BEt`PRnWv7y(UPEJ(kFF*2)7$vs(4@h0Tm_&cdP@u(rKT-cWCHu@UmU<8R zVHD>)hsL8SD-#Ds<>D_0o=1KZxWvqYoggy}fC$1S^(bb4ss%Sj2ma$YN9%pzHoly! z`P%AaGZqkd6-I|5sn7FYY0Aq$&B}UybhuXBRsPnWbUlKgTqO`=ESwIbDGp~VDW)77 zYWVixo6)(%4`1@u2ML4u7FflAk>!2V*9s+b%(5CH8xD?((_+7Zz;k=1kjTXt??$&q z2&^x=(7DJ&)59(zAka#9h8|4+vaVUk*aZS9?#GzmgJ>$`_ zO90uVpP+@Hv{POVT;!@9mwF;wjU|U2*ig=Uj@vt@S*`&s>yyJwNr>;Tly^f>^rzNa z8Y~mCT=>0ML}xU$bQb8W0uRxS6NNEvN)LLC8N6LuS6`lwt%c)vXi=r=9y1(=$8W$^ zJ7Eu=W!5I{|DvM7vR*&duqxByd z_S5UI2Dp;jeNyP%jL=8chQx4*&JXBg)EX%s8=a2B)>0nDKfdj>8b6umU}I=ho-%07 zJy2%s-YXw<5c@tbBrbyfWM??Nk$;H#!=Y&SruA1%Dbh=+hYH?KW;?|LK|lk$TC;C` zFvknce-nvM9cc3_`ji+WCUo4G_(hE-For-1XnZz!*w@g0xCQwLmpZn0?m_X-hXszM zGfEXVBhCx0_oFL6fj|f&7v=uzG=KJ>Sx#>j`7ADV)OY@jX6B;;CvCEPh5vdo`;Acd z#(GgNykl@NHO{}Ea80pl#r$i8GEvWcZ?r`;{7T}*u)D=wdI|B_s;o*$c{1&RZzL$v zi>AY?_jWH97bmy-5ngYQcw_FbYkhaBk9i|Y{w$sBq^>`eA0J-weTvy(ZrpHbs93Hu zwsC;xHXkDGx<2QwQ8W2dqa=f!&-r_Rv`))Zh$wlrmfRodpWiJin-~is@Y6)AI%5IBj>8@h)|4SOLx2UYE#+nPt Z#r{OnlGD!%fW%_gHRe&BXk`h0{0|i?)J6aR literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-20-PROOF. To .jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-20-PROOF. To .jpg new file mode 100644 index 0000000000000000000000000000000000000000..e634b3a418cbd34398cc6729d294141533b32021 GIT binary patch literal 853 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}niW(%Upstu+-}avWv*w0k}rZ(Unz&z%|m zy=vN?@8V(`Piyn$y)8QGsLtfWf2hHFmiFxb3@ysO{~6~0t^d!!zJ30mvOU7FSHG^7 z7kCvm?VsXxF$uk;U8YuZo^!2pkkp@iU;+O!{rbm;p8wlv{>gOWgZV!Nx9I=+*#4iv zL3CqGa~?4>4R6FZWP9G-TqIB=5B@==4`n)Op_0{=58 z=WYJa5P$hU!;51p_di^;C-pyr-7NEe%qRad9M|sN^U?O((%QmWBL~%;6PFcqIq)1; Mm-+In!T$eE08(*1U;qFB literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-21-Graph of Inequality.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-21-Graph of Inequality.jpg new file mode 100644 index 0000000000000000000000000000000000000000..fb74107b04bef98a0b94885c59dad7c36567375e GIT binary patch literal 15988 zcmbVz3p|wh_y2=xsx8xXRbndbO13PCB8Ih0X)`YA!e%P9i;6BH!+4b4Y>8PRmE00T zB@_`c+s!D-{nB8javft#=3>n4ng8b*x_tNdef@8GRfFeqKIis1?{m)QJgQFBJ7}J> z(>5naLj!^|z#mA}1vx-7XUxFNm_8GO!OWU9bM~A?bLV{X&77qR7Gf9aEyL;SEnB{P z#p?BjD^~qruzdL%OQRotG&MIj#~E(6*<`k5{RVS0l!(TxS+nMRGiS-%xl7DeE?;T( z|NN!;2bw=q1FMmzrLhXqoUfrZUqkgA!T~;~YpB1VsV@ypt!dL|U}nym{SA1abRMLs zp{1ocO>6q}Y16>d6!3d!+WhGYR+`$+Sm<^Dv+Bemv$F|VGYx(!ezMrT>Er4R2Tz`x zHTzqgB}lYL|DdGwVOja3il@(>zo=oA8~)GI2;04{)ud>lEcSg6w%7}L_liqhYzb=@o8$vU$4w0=6> zE2LLhqr}%lEt*GmOSlo5YoJH#3#7GlaP55uun847;#ele`u3dSe`;|THZ2-12w}V= zCD7Xm9OW3_V-(*V)S~dyu%C8lYH`MiN*NrdLA}a02-Z(WM3BIayWOo)FS{==8xVEr zhRfNsiAK$HGiz@*2Urw!mp;Dy!PB)rBt2#mZzRC4a|CA)nbG@e`%R)^b)tfX*kp?k zVf)5ms}mI#Svz(nU#n8)_^Tirc&r6kL}%{G5O$~_J^cG$g$o(4$a42S6%>pl*NOAT z3A@`4^rZH>ml)UvAy)8HD$n?r-q$LKL7doadus$|#Nj!)2nFfU&C~+oeeCULl~!1u zF;U7@LGc+MbCBi|6_ie&;HaS4gWEQflEmY*L%pLJpP#nu@OW@|&iAzj2L1T=gz;pW zLOe-)vj+b6*na$RGTGzF<{+?6l+E=Ri5@N*eG@HrZb8T@sF$WNf=deUgM}z(kD6Wj zr9_VNHbxIV2j`V>2&O%6^-GQ*<74LVuBhLqB;C{}ZHl1fi$t_yYwRm>xEs)ny?9UP?+#z- zF@w1)2Jup}Wvpc=r-^d&wksD!?iq_8Y!}m|>M3L6pZ-dP-SV*lPQ=M%~ZEJ1s zXpaohKWpvpfd9etCcM!wWsR&=zCw1_`au>j;hQaNwbth5zs-A@dg-i$2-H1-lD5fz zu&|$+q!f#gkDu+6_?3NTQ+hsR^!{+>ZZ<jvGqwk`dQw#mXlWjrexUM%e4@54;eVQDUQl z`mbj0WlG}!W56aqFsdLIpkN9?n6rBA<*m@kZ9Grbn+`veVx9`R;@h6SZe7y3hhv&n z^Ug&X&-Hp?{sbO8|IpAcVR5Qq2n$-eMYhW#^T}gJ<2mn`&%>4=h6NT)O(Tk{JsPZr zimjf4?cYAZ13}S`In)RF{;_FN>je zGdkoqQ${`l8J4IZe!{PqGh%r(hcK3rnb*$JiX?dJ0^zMuVgbCV43Gc@k`M z=9!hko61&-3i^ZW!DQ*hOg=lyUx(g)ZTt*8cnOx&-ts;ZHn~*OLGtHZuq+$wimRX^ z(nT8|TVqFkIm#xx1+B-G?J9^uPs9BlgPc(_62D|6lU+gOmwlvqK^YT_#>O7}d^csf zUN6`O|8&ifGbfN-+w2KPX#3VGY-@{c16DDHl8p`024sifKU}9})YvAyq_+6R0Z zVf;*JMmM7oo|#qfo*}y`qrP9Hv2K;J=@A_P!H;3P&8h z2n`>j_cr4sI#Rc70nmdejVOa(HAFY7&M4Lhz8f|>I(hLO+&aWb1$iMEEL*XW?|?~d50+L|1uZZ>CRZ|o$RX{B3gYZYh-a7g5NL(% zY2_;D1RWsyjGGTzgc6&op|!02Qn`NXe*yiO_u3vOOTlrvMWJ^swpCMd-B&Ew-4`TI zkiik*t#YbFSEPcLK>@}`w?qKUT2GV^IpA2eVfITa$!^h)6+JNlo`(CSL}Gwh zXhxDqKMDF{dEskza|#9*F1oe4KhmRvnU)A3${C7>A{%sGE40c=H)&k~gu^}d8;*=c z&_=OD=QD=uK)hvv)p~Wp$fR4#Dg5TY6CXta0K5!>J;zQnx+J2f$_UyVluDsi|c-7W*TLh48X*73Ak}Ka>hK%dIC~vGtI2w8)=Etw#6V;Ti%M zPD@m_077YWfML*nKHl;G^E^;%Igh_R@v(xIDayLQ=&>Qt3ZV5^3EepWsED0oJ(*%y z6J7z8`(bz{LRfQLVm^SOrqXlF7mlY$iQOckhs(h;QX(k;3YlD4gevzfD&wleBDHFR z+v#H*tnw78-|^eepU7hqZhhN794c{hbo=heZsI!RFgPMg6~x?V6fD#k@QI-}Ukkuo zQNVS(0mQ^P+X;2z88TO`L&05=lsrD)g_Xe~y}4{>AsWS};kBBh3Ii1q9v-%qx0KN&DZ;FOn47@H*67A@ zE&HV@!h1*DfCztFK2;774uoKEE_${0e|VI(6@IZDuM7uKJr^R?GD$Sw30O! z_TZc`YB+-8-bM^P(v!&81nmL`&zLAR%MrH^owN_lJPa{OsJJ7~~OyStKMki~Z} z>&+XsZdO4{WXqg56BnU#S;eg+CiIkkwDEu?wu9&OZ^)YAFG47b+M9AtxNktOjtY&FTu*1MGi1NCzkq5X&D z+Xc2CBog!S!@WR{*Xd&d6*OuhG#}uJGpBdcfr?h$A6XusUNY;uvZY@; zq;8JrN92PFT5H@veM((LTKL_~i)$|!d*ut*gF)g#j`% zza@#OitWl?zWdUCp^oxfpO}e6P|Wzz>HFwoToX6NHm;Z! z2H0x2OP4gmagfa-9^U%2Z93g~cE*npfH?f;&xv(l+08nTiY*{3{WjDMU`2b$xDNm) zGf1l$Qjs0HhJ&t*i*TkZYn8ftZhVw6>Lp0$7-> znPFNTGGt?;yuD^HWxon?Oreh%PBa>()CU3b*bhXQ3?;YnaMoxU>KxLd99QC>W`^}M z@lA(^@fU0&OZc`VsW=6Zp(SVyZNIb&B|A_$4}QNoV*o$K!_g$^rKi8?s@%71c!1UY zL($V_qpNX`=cu4rFO{}8i^a2+PsIE7(G}r@n%(jS{uk7k{m+;{3Advl5Qd`oaoSCZMN^U^=)BZ(iP22be@9 z1vwwAl@mruwA(5uPLog<*)Nk7PBcRAK%vJ#3%w$1Y&|)Ar-^~IDY9QKArT8~Zc$<; z`A1{-7>M>*yeXHrZ2+$rgI~in35O!#&zw9Pz~>li6)70B(jRN(F=|$_{zYKzCVM8^ zRLr3Ie%bZTdk4w^1VVSO>S!m8W6l%vPH6pb{&} zobhtiF*&z2R6+~8pcE^H5=RjOX-l4_d4feo8X2@=mM*NvSpoEVDItQRZiSUS5t(aD z@4nDi_=e6k;3J*+isy(7IlY)I0|nQ;Vih!Wkd`^K`RBKi4mx?-g#pVqRT#Ly3SW+2 zhLP3Cuw1*jVGI4F9VliB7s|O(pa`_mqPb4<i|A7ha$Nvdi5zlA^x+!7On6GgwC&gH(U0uSfqoNRZTj?L} zI?dHgnQn5G@`TMyUwV=aDDiegBQ_ww9*5VS8D^6>c5{EIOd0ruT}Eo`((+94lg1Fe zwAOnM=us&7yGN-xT6ZWgKt_I zfL}AN?0W`?8=H;bf8=nTBR^(+c`p9jeFZ{nb&qn`D;`(D=9dXGF#wka^sr93}hL8JO8uRnS?N~ zDhMEsZ9*gu`o5nPd7v>g>)HuPj2oMkGSSfJ3Mwka5Tr(>je9{ji42xG%5O#)%%S_k zS1B*}qyF`EU?W+xi6Vk7k5#<3sigQ#J`Yd~2c&YjMJl&V(r9LWZW%1{hC%aS)I{*P zT=AMLS2ETiAt38gL9+Sz&N$P^kRdq7F@RJDC&M6TgG(s4-AVPNWPmcVa9530dUQ$! zboIBh!is%^-S-%oPlt19AHxN^;R ziYJ0X2L<}fMILq%tJF!?Gv#>R`Z}qxpiow;f?}1$Jdb`AR6$3pL$9c1e}z!S`F0M# zrF$^qtZzo-ci0njS*s@E8*!%**6CnIVH`>#c&)Hi2%wbvh?3p!lRZ&|B#qD|yU|*b z>#F7JP!z!fjY|5MJ5h)Og$bxTLB(QH6mbWjFf6`vy>&bMbQ@l4%KO)wyz`;$TG$&I zJzFHDdWEpZ;|v0O&w>POK!i&>uGi}sX)n}IIwozXC(rimDLM`^Het{f?dU3KrO)mG z8-JkD5`XQHLao9A>j8$nv~k*T6WJD$DcS-!Ct#a$cSMK2{qRnG1}qPk1BcGjn60Q2 zE^~iO>D?L{=z|$_4)x6hUf?Mu)RWz(_gDM2Cx;$euAN>o_iZ4C1aj@T4hcXJx`1|H z5i?+8R9N{Y_H3(Uoc53xbGD^DN+(F}bp1{TV1{j(so>nHOvjMCpxKIN1LkJvf;HA< z+F+Nihx!i8%&byv6ej-gNlWE*QW=rdH4!Ti@nXGh`%!!7oJqFk)Mb{lnqzNDaNR zX)?E4qqzg@7*JRziL*v^x-st|+}SW%dqFDW;z6 zQ5|G2ZC=ldMs;I#gft-$LGg@JY!ig!nMA_rwi^1s4(Z$gMRCERs2$@I~PHO0T_OY00e zQd>!|0qug&tJU-ztb_o({xn2b9$R!uwtHh>qV*PH zG3Qq#)8S0~Z|*ZKisupFYupr_k0u^>W{uH*vN*u7;dI3m%XL#)zrhECcYAl*#+n|I02cNv@$GTH&nLFs<>5swX_~b2U zlL`b`-IVL@&-DC>XiF<<*x!J{o_;2)D74lPALFNWX(>TU^HK|log|K zV(6zRTsgVL^E?aSguDdmK^2g7)wl43g2S$fJa?WmK^;VQ-my!UU_KR*ZJQM;NK*xE zaQSWdDOLj93qlv_j=V1J)T0=id{L|f(;Q{x{n^!GgMj-fUm!bo7vlcZ|GD~m(8%>% z;9mMr-y0bZ0+qw>@|s%~&+196*8r%Oj^AvhUjeBHNG)IhlYa7@x9C^tYeo1MAkmz8 z>!B{({<;-jDyC7c!VP=r{0GFptJxJ*1Q2$O-om4WO0f!R&z!lS^BBG=H6Pi)CA^Kf ziVwGRQ2dtjUjuZ(>7!ecl_PAf0kT0k=JAQ_j!Kih!dD;zK}^OOrNl2m|LnT+4dMr( z$OgNI`o59lfzcBURnC8oQ`-1wyw`wDt38MCv0R?re48RB*cjBb?bl8p7?0ijesBNW z)kmYQte$gj!>;AkO@cxlV(6xK*Ao&sR$$Nz>~0 zP2&u5!=th@;@tSXtZm=Enb&Y5aP!gYT!a5}SpUOmKBK{q#CMk($OSNa=RgkwL~4`Q zAOyX$u5F)R6d?#k#QIxVkzrQY543ygs6kB)dl#%Xq`6_{D}~@RBDQ{sV<#fA9P9g! zL2AKI)x!+h&9k2g8oOIEj>9^oTYfTQZ%Koze%7sS(9s&zMq0-z*Zl*s9lQBgo%*HV z_NzMo7Xw@*Q9gX*tB9`=JCkV}W(AYa&a8_-_p$lUu>a?K{b16TG=P-M^v1ZD!9`8D7TI4VB^@dz`Qhh8_w}wN?X0pS(M5Go+Eem8F&~ z)!bmev`@}eka)9G23VW)Z2TVt3K&}aLU{a7$mBB#HcHA}!HO2<-s1ro?B)ubgu<7N z@Zz~M8SKP7WG(#q@kc9dk;O=`9s!i8L7$g4310$i-$|bYKfia8!5Z`apkyIow%`cq zJffK!P}5!LtQ_6yW(BsqO8I(o*$H4ByIFlyKDyHV2b4vQayStcf7NIwVA4-3r*z4K zD6&&kWRg4cb8fYm%Xm*nFu8EK$#L<|$pwc<>p(auh-AoXa?DJwR*`Y$uq{3zeDFY+Ca=Q78vhb#;KZx~Y8@4UyC-tQbKXiiUp+76VQ=5V$yIb0zYjnhoN zG_WpohBciY4iEQud;VDmEyv(LDuD8EWM|uf2A}Grsr+9zhV<(Efg&-7B-_-zXBC{|BhWtY{==YGRL9F z+$i1hARw~GZwLD0>6CD|WG%55=X{7g%#Z_hX3d5o-}9v&09aH|vO4LvJkBDE^zwdN z!fy?c&_3U00JK60>k>$>GPcr#U8fz0VUh~FeOLoG>Njyan*R+iYNAhoynyICnJZ>H zz%>28fr&OEJKaEPhJG%mzxrIJ`l-v*ye(yNHkUL(mBcx_hQ&%Ma35^j7U#i}-B%U# z3~7?&Z-q@G4-dgH9KPMOzbwFkaZ?c5(~4k_0#TM7Jv*-KAI8V9xh~ICP#Db5K&qgP z^N!)e^AQ8Cz5M{T3b^e2EG-`1OpJd9YOWX|roIaM5hPDTn#Lvi&PqPolA4c9HKWheKC`rSrvqp`?6?X$Xb}|Vpx2IQ&e<)P*o#5Uy`LYBClglz%mg8N`SgX7~8>$ z!q}HkMzbfV1=gT_bp(-_cVJ5*>RMqeXyC>mpsB{jE6LDNKMQOXD!@SLV&LDz4v_AT z!jHwxH8{|`MZgsbw_{`8c{GcPB!E^_my&kxSJFRKE_V%UlrzlYt;Aw;UN&239##$|_b8$-?SRPOH!7G)2x!t^i!&`;UM9fbw!o8nc_wOarv z;;WMYP!Rg<=KqsDGIm-XO)uTFrtUO%m||2EDBtocwcI+~qh_Scpb%U~taN)sMt)1m zW5D&zdR-|AQunmZVfq*`?57>kKbaYFJ;eUf!|JVXJtS2&yE| z2mEhm`vEDYySwEYK@Z66*E62dM|r^;^?$9lAr#WYmtGIZ>$DPy zL1obzxieO6z)bDyz&logIC4ahck|1~LLdC>Ef*qwxpHyM?bZK09CJ#VFOZP9%K;W7 zv03<$oiQ)0c3-92bU3J>^&a7H@h;KJK>cGsZC22GeofwS(T?fneIqgP%9gfXq&PB0 z<~n1;1kJfiGFm~Eo=ojfz6w=*XSG{0O8+<;Mpl*f$?1Qgk@-1s5yzX@a2jAjT@s5v z>IM+OC3ygw&dpLmDMs{(i*MC6MziU&!D;V_no&QFK6cAll7y)MP!oe}$wk1~JD3YR zo92utI}i<`3VQAt+B*7u+w*Bs{{chUmWx_zp@IS2#J9U!#_o767D+M!7S`TRiZ~y4 z*7HpyvEtT`j1_&C;{7zu$6S)j_&xBuMQnwm7Ah{=rURt^KwK46>}~p0xl#iIwh4f3 zSk1Ku{k>5$oqk#cMR};8vBmd3HKAs#Izw~rv<)v*Xr`*5zwYYxgggzJw`28h{bwIu z+V8bJ^zoUV-y+c#t^s1Ky~_R80GfMJ^O zh_}ikRwZqYzxg|7=Z;?TE$})a=mAM^#%B7&brSCmakNJTy<5&!>As|zwJ1IlahCrAVw_F6s6vdIzOb*s%QJhvz+qMeoPrc z{ClQ3#ogu^_vFYa*CQ4gU=XQN9*z7u0D9*k=^O9-Qudkbky9?!H@G-of4BF0bOH>{ zdCaE$U@>^vW#&O_PM`@3E)*@AqI7)FtE8L0j*V!b95<&0wBSMj#zQD`N6La8XOK89 zHgj=1smEpAmCM(1ccl{QLR%)8_gmJM>)l@oAj=lzzT2>yl$jMNRKl6rKtDL|6yHB6 zzqPypYyyhwPm$qD;<1311JXbl+;9Wv^7L*&w@X?u$2&2ein`;0cVS3R<|e;x!Gwj!lvZf!#0M8@`_LLi|Z7#|QV8blZB< z6v)ws z0b0tCp86zZunKw%r+qR5e*Cq1hbad=#C)CmM9_m)Z7G}SkMVEsjexNpoP*NXl?F8-Cx*$hn*%5cV|fO z3l&4MTj4S7{hjYs(8B`Dj4S+1^_~;gb-hX$J(a`w{x?C$_myXn;_7G_5EJbg|5t@K zuYcfQ67k8b``UJuU-jk)e>uGM=MxW>E(kdqA9P3 zAF;R*QAU9W(cMb4?;H=#3R}#pKD74cZwb5CeABzyc#ZEroJ%)O-@9yS_GHeVd$&U` z_UykDAH(`2_lAGV4HO8@k@~=o=7R?L9pWc1K4@&XV2^k0BW9D4Zc=ouo_(L2N7Fu| zy?X`%N8h6ZRX%U+32(qC<|AY*f6kEs6HFFglvm&&4HBBjIuP7AwaVE+8Y?t$P~eGV zXmwnO$Oe=ip;xM7gt@i2AQospQjn4nyks(^Xw-7w$UI{MF3|unZL9<94{V77y}|>@ z$vGT)O^pZXPDkd<(WIG4m#C~2!|>y^LF*{k!T(KGw5enNJ8(-K7g6(0q=%HOpk#!F z1B#$oABC$Ax<v(esoC#`n2xyBSpty ztJ5OB?yMY1Jzbz(hTvrWmQ~P^$48Rii9rQ% zWC*r867xz?*CD4X$sss-9Vj5sJWr#zrrPTxSz9!-qdpL8&*uJoKv619$wY7|be^!ouPyN|;v>fKoMsl@`Nbk>rCasL*a6bIp-WvtE9`arI@&TF1Ch zJb1n+@9s0@c{aGrKccW@dnG%J!q>Dl?)xIyzm?j+@k1VbuAlt2{&fyuL_ddEWHrS~nKF?UU`i z>{RW2oUAWil!deaKiGxff(o{;fY(){Nixc$wxFXC32u#x48(c>=0D|}0T^lZ;o-?g zG~z-FF?90kxQK`EL^eY!C581M8%0@kWc! zIvxJTRlZu9qVxrpek949^~;oVX7%o?!h*{K;7z1!pxTs$H@^Rrp?_v9BpvI){twQW zktkhi0HOda(e-He*O=4O~OnL&d%c`tl`8uiFR_rpR`Spjs@^klB zsGx*)Ir5|(+~$}CKy9J@k-NAY7j)~LcuyI)V%)@*_zTMHnNc7D1SMB!v%P}h;g+E_ zpZ3g1l6n`?ifKW;qh1@y`RTOlzbb2I?nl!r7 z!w`C->$C-Q*$(=v9`CZ!?FI-B*)_Wu!q;a^?6O?8sBf_}OJR1Z_RIEWvheIGSf1&x&T?;uEtuX>0<>${Z29vR|j^v*04$$`0dVcC+*Ss($t&rY_ml-hDn+f)rZAB9hi;on3p z!%5+S&z8nt4#3UWyS{^{S?*72) zq~}029GH6vm4r^;RweDX={krN_LuCQ}pqo!wqcyP|?tp$#c?BC!VC zO2+~p_dMcRAz(`Yrr(97hv3IBGk~R% zY%w}2TnlpADwi&h#oJ`5AUZ(>UChMX@-I=uwt^dvn)Llroh+W)s-9!`>X>|R(Lv}? z4g=EBl17k1fpavr6WUjk?yQJK?%T3?7c=UJqbB{H>#uNMlGqL)$N+At@yz^(J#d2` z%m;VRAM!Lf%Z`E~u2j=psN3fgUDkfnvEnUKp|64_2~MvN>H_kBHvnFxb_*VZTqG84 zt%Uh2Unp*4ybH1=pILYFxtBe@c~|FC!TldS5|?^JpR*lzIl-~}HapG&n*gt^J!cN; zq-J>^Ns93vS3$Xno*qGz@8ec)ZuNCaG8;(@9=>s+{2@8IhKzf9IR15s9(I5YYxv1p2i8Qymb`B@;h-a=JDi~p@Y#KLUL$iECYh8-iuPo#kU-oP@$Dc))F)(>TSfMq8)&`-#HA=Sy` z{Zx~hfv$L(?O4%Cua2Q91bBt#s)F`GrN{;zOc3FCt{Ol1kEg;3oAl~>LT-z|#G{LK?I9NlcTho*sO5J!RxrU*Q2$Oo zYZSNTiNM66i{tk7q4;uMHmPo=B5$5pE*?9r1trJWt|=AE{Viu}|-`XHF# zA3~39XFGSlQZ)4q1gy!4bOeZyR>jOY(_$$*| z3ETl9_S1m7=6-Va|I|xd;Nw!@EEvUo^wGUFGo~+!P0Kzpt@rYbHJQ_IUR=xhc1er0I{XW9sxZlxh091qQe*v9)sxB-Wpufr?Jyt;^FjIP$ zaWW#VqpLZo_=!|kl%sAAIFpw_2TeQa;FYi>(PPX0=agDP7f!2Yj`mSG!1YBo01N;OF-Ou6uMbNOFH&W8uyPu*@Ua+ z)(d5$CT?IFp%`CIPu`gD7vpM=J&!A*9mBK1>wjxB@-9r_^v$qnK&o_Ywg%O-BC)H3 z<5a`KUhOgMQmoO`Eg^mv2{zq}3mShxRFfiXFH$+QS@Sap z6MSYA7&)u0!UPqp&7NL^PKUSkq`U?O%Ol6iH}jB1l|+ep1QoRW)4f2x6Oy@V3Rs=y zg+T!cZkG7kDvHgY7BhW}&t&^+yJ_C=Es}XW{qcv0n!dp@S~DPMX^1|O|6(DgkM!#A z>~B4)$^jiC2P!(4>#Dixdo^xu&6~o_WVdr?tIGA+@9mtXQ?GX47#GQ0rN)~y%%%WA8z0q&Rf=i#{@freUsLH;*RvgW2V-?pFDf z4&#^jY}|inOQq$K(Ni2d?KM*gr0Y;>tjeCwxT;p?G@@y>}~QIqb}to(p-+iaSVwb$F@s&^1y5S>~}z zN5{nQ>hap!Rw0I6tn^w2&ieQ%AgD$N>dCn^7y1i<*MHG>@}i}>XXAkpu^dOeyiBa> G?f(J^K!;xd literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-22-Linear Inequality Region.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-22-Linear Inequality Region.jpg new file mode 100644 index 0000000000000000000000000000000000000000..15af7e8e1932b853a84c54b7507f0b779e8a740e GIT binary patch literal 17366 zcmbWec|4Ts{|9_qN{gBnPDPoDBGr)+WtkyG>1fCnPEDo7E)_E2R!(WtB$0?AG1ewo zrfhShkR@czHkI9kFpSymd9Hitdph6m?~mtsoL8sD+}G#&T%YCoZ0}ptCi(&`Gc!44 z0!c_fkOcS-5_Lev(1Q8%W#-RYAR{BQaN&YQvT{pg7cZ7wvtlJiZk+;FX`RB_wd+-W zS6#nxoATPVYC2oC?a};T^Z)aY z=x=EG0*Or$AyN_>A<5+uQp+Vo?;tGTbDo6w2b%klkd%_1H(zGK!bOY019{6JNeL+_ zNolEh^Q5K0(}Cc7NP79a6&o~-=C8CoC9~06Zs)b|#0ARxvPBqThz67YwPUUt82XffQjkBLuOXiHnw)h>>ZrWoIQ8m*~RtJWglO^EB>UA zzi!+Ny>hQJ{`AF$MZc+Necd#U9!-3*|91zw z{{QP_|2fz{=W2&!r6hpzq?SW?h{IDeP%rf2pE=yGE$ex(r`cKZy}aVh<=IO~STo-i zge$lK>D~#GS{8JviDzN#AM9m-q2@t{4%ofvetBKxq+dc0wp(8Np9kgp>a2!XCW_&&$l6`8k6TA{{e=U$HJ5nSZj6zKwCJ|wiBe?SgvD~;v8|KwB~2q)!b$ET z$VXmbzGm9*B#fDy_5fJ!Wn{R*(x$(A6J62-sx%s^t7_h2Tgt8LRg+*_J7-!3U23ei zNTlOGlK7-W(8D|TZgw;FjtyCMjXCc9eCFhSccaxkccAP8I&N?PvM=V}PbQ(~yYNN{rzVUT*EbakkKL8_$LmO@AVz=T|r zthHbujXjmHP%T{q;X5R`VdQO3y;(>1oJ<-pFr{Vz8PDx|?T+jlAu39JohM-f)I8Hm z#ix&|oD6uk)`|Z%c-!38DdnrnpT8Lz0lZiScRMW^NMLgkZpu>jvPK4-CC${H1xtO( z*oCDZn9$+T*;5WLYvd1WLJ|)Nt|_fi49~m5&Dbd~ffF!*ntK8HT*X8-)f3<8^(o=u zL~FGdJDP;$I&WdR=ZPRnG(7etK_+P}HoSmhwiv4!WHK0NkgigcS1N)Ig{Fz1sOusA zZv7g(KMXK}T%{)Hc=m)$C6l$Xk^h$33BP70TNf1NO_=ijz_V7)RB&#XPgfJU+)xBr z1O>SJ3d@HB4VU+VQ;fl$Iw4ij!v&;Zt+BxM0gf)>!jdbj&&X9#c z--`w$0%IwHw##JgG`Ibv7kzT)nc70PYmZj1+3Ok1x0r`ntckqED#I1EbyNX|)G8ZL z&Q%1G$|T4hcTBk@Kf2LsIyU6`7oR7!Zi_nznv740gZRD#&a*FO--C$Z1$?WytwyNT zNF;{H9o)k7$OSS<70aZ&;v`p*of=8*NaiXE9Z00S<=yS>*dYrM#4NSkK0@nAMnxoa zfUAv$nM7WDjBicnpN0bFMhr0O)_7!ZcMQ7y*{NAuDT64 z!q3Z)mm;XzNLZd0Hj;wjS>riBETnH$oc&_$^W7o89hb}FXUO<-3u!7 z1{RdhW+&1-bZ9(l0}<4hkV@0=Fb}H$3LWl19580ko7W=9n_=df-6?{UeWuxRb@}aZ zT@%>U%67W(k_bvRZbe8K;Gj`Blm(upGIo0jup@Qlt1dq+NAA#7mp(e!0$7op1g=L| zE>vG^i?ad;%ogH{T>g=K>UJ-WfE2N4oy4c?G5qZ=c1Hjpe07#GFeJqe!VWmuxE_6A z1e`P;l}aN5q(<>ZdV(G`Ipqq%ns}r}T!#yANwLn%4|kYTfk3F>R1VN$EFBZhXr>S> zj~1>=>h(-&Y{>|G)jIOGFajuo9z#UYRw5X`@3NOGANNeu*axvZ0?A^K+HNi9V#zo0sEJ3w6FS>;p4 z=tGE{MEMYNd_V+woDo6w?OE^|-l8iKL&;iKS{RmB}XHV>M`xae!O>}fjM-r%Z_#UkCPV} znyir{rO3EY4I;kAJzAPIg*>{mv~*G_z6rX>1YZsrE{PDH!H)VD}XZs3xc z)5s9@fj5UdWVncNIZ$n#BTKkNZe3D8hzU?%iBoIGr(ii@U&^0wMJ{6$+h=Cf$# z_UgSdbB8Zjf_pQNLHLnUPCdyLThlXBxZ0SV`a%RX(S$x^5o9bUq}Jf|E>8nfyE8<1 z$k2l`$&LJ;_TX)ez|n0PX4u zVvp1c>sn$}ZuikY7Gp?$(G(`}W3a-#HoWSk?(4Pv8Xwv?Jd1~$F&~QwB-z`*?u#*S zGIYSVwigqDIH%B7wx*7a9hqFoXL0Lg)=*bY6AD4NJ2QrK z*gTCJ==-kM?6LRpcghtwD}urIu%-V#Ik_~X_)7L?r6&4mP5ppkq047{m(V4%x}rvx z*GSjkgbw^+Y#o=>Cv-`mP>$@cr1Rvqa4iyS_2jkq^ii z%lzD*wEwI%TyoSG^O&9mJ*8JOTiYY;eKFCOrxJGdm7HpAZ4X|v0e1$kG0qb~uUvYc z1~OZoC~G{y)Zj!AjlVhG-W%+0C>fC(P+aUHvDfP{gO;cMY6QDuh-G-TI-rb3%Y!zk zUKB90w7*bt4o`?6iLsF*+X^4|W8<{HKK}kmP&TTBG)ZhGa9{^#)$J3UscEVwaSZeE z!A}O3MZ69uDIPrSiv>*1dwhW#8pO`5ekmve8P@OUvimE*S2G=qH)%SkEG*xmUyjk? zONL87bmLBR=iWOL{+Fg{f$dsZwi2~HtX733w_YoA&zdh+DXv|)x!21-k8do>+&QuG z1ODs3TqASafqSNX%&iy2U)hGox^VU45%!oSzJw9CTg99@fa+T86u0}$jvV=ZD<|-E z0A-?v!;#;b`5-(Vb&&N*WP{^ z08B#SO;Oo|?^NoG&qa}3eBbjx5!4~C5gmVu+e*zk;Iq+x>m-d=$rnNSYLbCs+g8+x zVyJrkL#B`RvDm;FlI?a1hzo_YWF0RCFy6DiJFifUX6mxnA-TH}&qzETBk>hL#;leg znE=lqcab;vNrxvl`;=AAIE7fZHwSgC8rux~Qv^-a^Rghv2%)2JiEhRr*AmB|Y0pr^ zLBa2{9YU%FA#e#~7Kdr7hg&p2U^ct9-A;xR;e(&3ulKO4A$HfI(b{;#a4E2I3q0!? zJh+dVZfBd(OYSyDvhjV1$d9y=D#m1s7k|cC@)$Dq4UUr$I*Xvov&6i39||a3=Rjxe zjDE0Wj@pVp53kh>joKgPr&j$^O6ATa^A%O$TDMd#m?fTyl6oOSs&3Bx;F)mGuGzfcAePfYnr$HvoeT4P38-F1|n#jT8)Qz zhN^JdV>e4QssOk1&{8&?`7IR49UkkD5Z0?H(>K!NwC9 z!bxG{(bXn60oWixhSr5Old%PEY8G}%(X>vLFcu@NynXURE5AQj1@G_j7&|_q!4Ktv z6@6UshC8pAeg^;KhtAX`L&u(r16uq+1SR);**$ulDse0ZdLaBdQ_s90RNu9r6X&ld zg4QG^Dq_Usp{QrU`eD5w=A3t5`nRV(#Z`Jc2s${Ys;>Ey!3KmpcNd30eOtZM{x~k+6 zw=?{8KsMyoc+c7v=7fSEDz}MvmdIu42nAYAdB>@o9Ms-;%`5hZpo$+LT`}d} zi9`l@B84R6JAs25eKa0QG@gmY;kNVsBc1B&;Z20{*12T_*FqW8 z<{3AGRnaGgERTzzG$N0MMQ(0koF(oOp6S95B&LpL*?k}a<9&taI>RMuk|WyqDK~cN zPa}58Zl^^tiejk(ww457=or&T=~9&GFevY6JUXJ`Go7J&Ey5;dKmRgOR|I7QAuPicmBebVND}s&2x3{i5fj5c7nWZw*4eF_h9p_YWlldJkpgwB-2H>unXnWR~Dhivq} zJ4rKU4jC>#{&+&BiUA5ZM!Qloj>yjd(Hx#iU;JieB^GR%c?7t^BFY@idnL?*V#4@U zq1q-v!vxJ(KqR;-hieO`hOhGL&k(c%SoI)~LVeTZND}!QhxaRg{~Pr@`Km@pKDt4k z1D75zW7M5KRKqwkjGxW)?7Iu=YIK)QeXBTg^=Ma!XV10Pmbgh0=3~sqA`n!v^hh9y z3cCDmx1Yya@JpS*jGVvIthZI%<4#P zDAB=p3i%-zlvw3k`ZtJDT~S|rrlz0U9S`YWXW+c_jaM}Im@xlCVNK{Yju8H=HCeb$ zX5f$T$2Fn-Sh!UfLHPYOzicu=0&KlBK6lAyPZ=W7RlXI3TY`fjD+=sDbcL3VRR1j) zqAQlvYDl30?i=e(&?#=C=bE5ejT+q7F;L_ObcaknSA5mEw4|h|uoaL@3H_4}qB7qC z^7CA~R5+wq8(tJ`p_RFG{uQYvzU7N_T@XnMH6(54f@%F0_&(`K?>JS2-2-peA9NSIQ3}y?Um{y zCaUEipacbWkG%XD|A!F~g_|}BO6Sz8qR~gMD#jP1c2pUElG~aHGMqyztG(h#q~;ZR z8p399Z!vgHO&Ym^j3z*x@hPrfmD4(z^89wLfYrp|BrLcM(=m~7Gvymcsd?+;cAU*H z{w{OKa?Q$-GLWgAXS_~3vA-SXGpxt2m-@w`@09>1gi~N=s{d`_lwc^kGFIjzBMw>v z?+|1(@|i-{JA;ge4B)((hwxB;ank6c&7I~yG~*ovtj3mL zh*$&M3iA!f{yk^2O%R8cJ)Lkc-U-;WAFux7jR*6wS6kaaSIhDMBrgT-7&dVVB%U!% z1K~DLWICSpfhubpYm=}L+^ZmB8XrmRHR_{p2eE27)m++Nn8|5Kki-|xPzH=dkORI8|0z~sNwH%ZqRcWf z>c2lrU@2UHOX4=B>lo!;>@pq4I{3~mT&e?HjEv{IsUkO<-Dc|QZOr$Yeg?PrY;Kt! zHb67#Z2hI4T4TD7#EbE57~{sAp2vR{%@_wM+o;B*R(g^%iyPx?IsKT$S@mX0OG4R) zn6+K=(OJ^el-5WRb5aY^==>Rh4p_eQ85(w8w6lxEqnJVN6xIfiCy1c4yEhD39!#nU zr7@@QU0M@K>qOA5U&<+YlZIAmT7p`&4{JD3O@h$vaJE8l^)T`ku}(JW1N(ctCFdM& z9aS3m6uc=V^`U)Hv|T{vm36^WxF%`;v&YsSouC^hH%7rkoIYPr{TJTkF%DHum4r&+jY1e%n1%uC87= zxk**k_mG8y2ukcFuv>|B`@7Dx5mb$^$2TUTI#Mikm$G;Iz{ANpc*&U7PdP|b7m88UHYA-D| zSR!%Lkwyfq>}4MYyh3jwAa_BA(iX<9Qx?>!+|#clf&9@}*htu>$(+!F-tY2QYueaU zFZcW{z&yOZ+*nvhG$`cxs&r8#fsWJFN3@V zm?v4F$}bO=^eig_pHMt(~0KV(x93);oT zies3~+FGM!c_V75fzFcIIz&(GHvDkiIM#CH=-z^M+%b^!fF~D|=}Q%b_Q=X%bvy6e z3tgr^C=SrMpAHebrl1XX3^iqW-1s1q^C6|Y=j2x)3d7~+ST+2YY-Akx&wXJ1TPM(N zAt*YOn$KUE#Gt%AEAZ~6Edjs%if76}-@9O6EPqf0-O%3U(?^>-5QhvJkR|5YI!X>d zy6^?glIuN-%DJZo@wfQ=qY;<-P2Zd@AH7gf&$qr-5#6a{Bv9?XdBJ{qtXDlF(5E?fx~sh3V&WO^uICH1eMXOAiL6gdo{Lp**Xj zTBTf5E2A*|_uyNhn#5yxSVnl6AY5QQ18Js>;@I>7rW!yQBBn$&2Vr8t_2wQL4GjG5+;+X38QyhO=1aE zx|D(c30?Dd^Dl5sc{zgjX5%Rh6Zez2AnLUtU&O22yX4)q*U~)@#d5MD{DH+-WXL=9tA=bXcZEh15+hyVg>nq zB&DV?N;9l}{cx%vpL=c)?gQp=LvO@e1o?XLLDuut5Mw^0PqS(dVbq+kOB`XRJ{;D^ zMtM?^V~iOK&m>S-#U_F}QDbDpa9+0vT7MCgQgr@4DpzPXk^b1ulB?Vo3qn}+Dl~?b zv|>5uKy9gk$U!rlk>O&37ip;o(z;lT@k-wR*m8uA*0Aq|L5~zuL476enyi|hPFP3` zYN(o8S3JbTrz1A8X|6Y05YUEi@+z9PsG%>d8$cl)oKXWf`$>CymF>qCMi!s+P@+(j?_L-laA zHn@Anon9HFc~)qT?p+6JrqtB-Gd;IKkx++zN8D=wY>VN6$h}h^ZQm=`P&lGd$-gV8 zUA)s{?I95q)FIe%{Vd;jT(y(PJ=aZb58mYQ`Gl#1U+u8E_LH4OhguPT`pi-8hapuV zAM|EIoI&m+fkq8?ub}*I3?*|@1 zTWrIB9Ixyn&)6h0-*8Wq#J^2C{CZlW8gv)Heem@l-EtI;@?=cD7s}-Q<7W@^g{&&h zh~)_{{*sEVNLJxhi&o1qVw5N7ih=6-ZBkJI`MVBw8xzdTjxng8QXewwh8dB&{YHcA zXw;V}fiLi*phFf!6l}@xQ5xW$1J$}dvHq%sG!>36q8Yc}*5P_ou?^CdeX!AtZCw~a zHvHq+zYG#0O!8=b77k+*v5_F%6f)Yi#_}qJpBZD-z+8>pe6S$)F+q#s7y|@9$X-uS z3J8`wI>e~4arj5VLIk!Eq@^)j`bhEJ?7Dx5 z?*2WgJ;)w3-A*G6GYA+deW{($eq>bLj+uM0*wl-Z>6rvv9ayt1a2g)jHGDLw9cN!a zo?Sp0m?BoiDhWd(1!)vFGCv1qf{Dx`cR%m%KvJAOAh+O6TtQhC8_Wsl+^tn;2}kFj zCg)Z32YxPq>sFyLaCjKMDafADVv2H(jtjO6kPNPs6SI%Dwei!F>DMBtP)Epxf236S zV55NS@B^s-LS{PLuqyVtl3Z=!hlh--jcrZ)M)8^@_5~Qg9nzH2ul;bxZ+}KrXzut9 zZi4r6mWK1dd}~~|N_Y|O&)d6Z^=|tl0_eeLpqEIlgg#}MIq1^7Hk=y8+{iXa>vn*> zVf9~&xr2=|y(g%7^R*^d^_~c&e+pjXha|wQ`@sn72Epb<4`F@4T_jr%&P9XXYepor zT*!PUf^;#w24o0%)YfCp@J1D|y0I1ub|`ryv92$xxO1akv{R&%R9>s{UTUoKnud0{ z?TWwcvx`*KC)XgCX$l_Ljp(f~$4}4Yv6)HG5`F~_de}1V34xDC27Ta|@;>Xq`OR*w zV|cYeyvK=#Z%i(h}`PG$zQei;HXL#%5># zq*dvHNga)2+Gi*!EM-48Qpw>i`+evbvIx{$+I!mV4z9nam1(fYayT#pCNw7Igf&4E zD<5Cn>vWBWwkODyRarLX{`anz>#cN8_dEFM;V3D)tvP|DK+|^tGw-{|%V`mGD&0xp zq~I>sj=DNEL%H_R)3~-Y&Z!c!V2G^{$hgIB{nGjOXih?1bi9cVh4D(f=+ z6_{I2N2m2qwGK*(r)#s^Rw%e*HzL;17}q%{Ql7~J*T9Qx5j1$3(npOBIxvI=&v-Mz zUE~#xIdTWeF((ThglZr$y9?YUOh1piiu!+rPYt6Emhr~<`{aW-)a;#<9f3t+R;3W@pT=pgHQMRB<6=xlrhW^MRnt(DV{ zxr9z4!OC#yNf76VW)?Yzca}m+wqCuE)Wo-f{k1+wFL1D(8k9+UakIR3+o|@{TIGwQ z-ck5Cv=1)0m)>mM8r_8}b1T$BV?bHr2+)^C zB|s9)G|Ztx@a4hlPyl+C(Q6MxQzt~Zk(p{IaBc}9XBg-*)U=qfSIR4{#@ zGfJ{J#toS@q}!{x9soPxHJ%`IN(2FgxyrN$oobrq%xT{xwZes~2*6pqdg=Z6j!|eR=hzbk!Oh>VKM(ioKp8jm++wPH%7~{O<0X!79)0s6nY%1pf>; zJVDTsa#l9^nB;;|nPMw9JtF-85{3bD0*T>ViP;9$aMq?cS`>4zo{gWVu_~!>6G1qT z#51!ES5S?ow6Ifm+4xThhxPu%vYkm3?m0DW-;&cJC`k;b(fpqQoo~rO5oCh)U?6lt z_ba&y4!lw@h-aDdsXXh2=iTUtLE?TGI`(j=u;Nl)mUH@M&(L20U&!BgQ1iZBmc&c?2Jg7bswQE}3n z8>1f}6NIl=Fshrr;z#MMsD~HOe$!KKkNO^780g}=z#9D+ci-w=nahk6&B_YysZHUU=}ynv+s##itS{(AJ@58L1T`e0iDyMZ z!orR*T94%k(7_I?5MK1^Gk4QF&H@|4+;i%B)q2F2op6#ds5OI07W3L&XX#l6-fu+$ z#rwqM`xa7wei@TB0CV*eByPmSX9Una-7S%mH@Rw7^LJEo@+Law{te~)(mQqjM;69z z=iT~r(6eA?WH@1vZZ zE2o^iHu1ASp#R4J&QT<@doYMWD{qALHH?12V_J=r6-xY5+f5ueJl+Yow8HMI{#8yad7+hD4dY z{0x_bIeIY-K(JjzkHKB4pXOOAu48x?Pzc1VdE^)z3K}gQgnT?uysH;U4E0TWQQ^#0 zL{VKo`H?7h`CAw<23OnT+1eR<^y+HceAC;#@n`r{p*uQg&jhF!1DWJGgv$>7fG*({ zn@uriuVZ3tr@xw}Jr{m&+_r3ki(~6Eveofx|lO*n(EeUXXUGg^4fof5%Y? zx#4af`bW4$iTQo^A2u(8E^)1QWlv_Isyh~lf89N$=@c>>Om1am2VJW7O|(Iw%e=RpkiS2G?w-I&kcgzyK3<(prOB6+c~4TNJUZHw&*`=HdLk6y!kVBWtps&v zV*Te5?gF1WyEHsVhe`O1_o-|0ziZWFyFEL zV$?B}K2tB1Pv_M2nIG_=4xrZKi{d`< z)usdFj!?LXOEf(!~euZmU3FP)@R{nSXxEJ_y*rP}mp00RH?tk8DC4j8oZ z2-2432WG%wjU58lttmJ&K^SmS!WfwZQN-$<)g{4fj;UBh?Dim&p%LIW^Sm@mfKtR! zxmIu@zC3sif(eT;n2KXpNwbI)+iX2&O*Ez zRC#osf)8Z!-w?CCy4W-Tngll}@m`s2Fjn8@wkAV9DapIuKNc&8H3@zOX*T z?cm=&`f5|ViGSE{{>w;mH1c@_?SNuO#BuW*_k7hW8M|K2>zs0`_1D^>8y}i=BAq&{ z{i$W~Y8#s}y&b>zPL#h8PhLCKQ;_heDt3?_CKd#FME}0I+&v4RTGEUI17pMG6&wBCCTT|9H^O!BABzJ6Z|LHu zT-b6N-7t+UCm}td3UXZs0Yxa0?o?kdwPJr81K^ql?N?qB zuv%N&tN_lY-a=gPSx7nT?`G(7#w<@9q=4MERp{t9o%UZr&~OD;JYvq+=n2P>c>F~V zXRAdEsXu1%fFzZlD#f!_4LXaVP>|ATI1tvo-m-rp77Vhtf=RIf&P1KHqi0yphZ+1u58Vd^6}0VE&CEJ2VSxsh6I=AGB6lh4y2h-4d=%jw9Vi*kI0qEkoX5 z`_=8lYMtBL{o>|QAdRC|0LBum*smHLdDn{&=z{=_`$J-7S;X&~{t8`TG+ zxaH%5^%+ovc7H0=la8)nH(%%@mx=`s@Iocd@ije0bod!L{D?*%euTb}~Eh(Fw9=8Z<7z z-;$dPC|~oz^g?Uh3Qi4n{OB+k?Q~p)8x>kj+_4X)4Bl_pb_TnFPi1@x-dmkv>Ubju zr`BB4*Vw0Nbgy@<=Y;oDN$;m$&ZXqNddRL^KW_4}eAmq-a*Nr@JMP`bDBsE6{P4k_ z+us(w|LuUcVhN~&oKBv8_wb6rt~Q%2-6}U_y~Dfy)UPWlRFfHMO#lGyXU>lc;oh@i zuBJc=|1Y-S@tD@ISg%e4f;Y?P+@xtGNeT40c4GGzx8u>1kiX`mYjz)XbGx-$I@fEN zn|sZwo$3xF=I?IEaO`{21~DaDpZR(u^%|IZvuu)I-gIjl?bexGCJ{U%v!>iM;7kw8 zaP3%Vu6=)_!#4vjOS*=BRrQL}kbGNRw5PN5OH`X5?rF-=?g)awRkVA5#mnNx zD@pdtNenf;9MI{i>)Hj(g79FH1WMEX?*y!5^>QP2asZe0bqG zkKo%EKbnP83L_eS^aroK>{!0Q^tR9Gg_U|9#dua2p1+iX+&u1xa8X0poqBjl$f+-U zk`+gXgXve3QZQJ-sV2^(`FMuRpsX%_c48DpKQg-V*;saYUeCFS)?Nb{?|nI*@Pnx- zrR%|a^K!4ozPpK)bNxPEWqeKz;%6?DE(@ttzSn%$L)WbX)@8oe%?uws+xZ()I&D3` zI$|wcot{6pimThY?R}5og84g2Kfg2$h%bJfLGrjhabYW> ze%z7qqXQUHa;k&A#K@U(6WW0O{u>hs$`8CBZutA-(w!z3ypJpI@*+OqjBeGjY<%~r zxM$amW92ii%f5Ho#dl`&#tr4yXGit>pSt^wSFL3#8xSoy@LKj%X+^-wJ5Mej?y3{> z)ymKB4nd-urgiuG>i)&?9<3>Sm7vxa>#Z`S&#p5xcm5Y=XifS4aK>eP87N)Kog?-a zPjsNXCL)T!SRDFdYlsin2Jctw0&O;>(ANpFdRaI(wSdn8LgqXbwZ(Ev4;ch6d+7Rn z*~zk1f=v$jtefY^?iS08oURihu2BNc=Xo(5ihB>r$81r)F!)6((MtPl=L;}>$%QFD{pvKq^jKaZt^z7=x!Q|7JC%-Yd)yGQ?zimulCVV`4UD^;k9n-KFP%x zcY>EpbfTCWhKALow*JeAw3-?;d?~1rP-i@fVOE&?sMjdA#>x_afiYH{qN!w z{-*~Ut|7MhG1xl++^3cfK$a;fx39j0fgEQqzkEN`fVX7lxMMd6-a}{L(@-O$wu$)1^RZso88~kvdiaV zftmI&lc95;KOYXbH;t&atlm=ECAWZdK55_59i*o5!N(;Ivsy$(P;N3v#~&30EPYtj zx#;EgZeN9J1FaB!NTtWpsm5s^#q}uTe4bU8pLyiSr03Q%Re^R(v z5N;N6rZfk;7I$l@7Sb*oPofYr08BeAK%s|L5&wCQ#c$D%}(^@0qNDY88 zhTulMEq;+DSo!|`4o+*@*qBkceuEeedF0n=bi(|fQ0-Z$*x)R_DphLb+fS;~%ebuw z&hx4WYV^zU-sm3&k{2e+{={!0=)p1oiw7@rVK=#v#V;`gD@>?XPtu5U=J< zq zGH>_kd}Oe&48BQb({GI7Sfl{J=n@qm|8RYf$^cO6{)ehj0Hi)e#ZH9)5CBlNSMprC z<9aZb0AA97NpsWagaxviN~ zhOh&L{XHuP*N+5NawZUk*yH~jP4zP^%&we834+BajtXk{sc~c7-wfBFSV8KqFRTQw zJ0NEPfufzj9xOp;WI|y}rnYZf!pi#+d(z!oiJo^kJ3x92b6o@NgfURg#-NdsbAwkK z^Wj`(oRRWze-A$Ts-`PE6XVOGaeFXAAzok(8LSMM)UoQmPAMWjDNq7Xv)c+R8P1L= z`~*N=j%)o^Fu97e1GN5{_;8Nly!N0k0kvT4B!Y6)11Tdq12(_Mus&elp#8jRE8$n9 z*1Uj)TmhLSOb|gU4KZOjVo?_m?xG-eKrL`5>w~S1a8nteU2VleM&`XJYM5m!knuTT z;c=Yhq>d%CJ1$3R+*8Qb!Qb$pM%1n8R2673QwX>yW(LUwz*o zWvlOO#5G=O0lA<7Qy9~-#HU|p6rfr9(!kT_f;U~}Gj{n*QQbw53q$u>JF8DvV*n4N zft%H(;{{{j9-aeA>TxY-rm+{hE68UBm3Nu33q~uG{T@CqgTPC7JRs#T2iC|Zq#rxw z0^Y+RWB~E~X1|y4q=3h_66?gU)+ch7Frele_g`|u`!WE-gVtP1zh_Cw+h<`fit@nI zGJ-sywMDIt6rdv!6i=-Obft&*EUzbw7N9=RA+Q1p9k4m?M$_#;Eh1B_NpHsIw7bVo z0(Qi_K_pdBko`hG$)uo-@dotvrU3bkc`XBToMFG$OJ`Da1+~oL16L@6I#_T{Po!zr zaW7(L8541!Ae)nl+-NU5!KtS}%-Bw#@^dA;&hyQ=jW)RR+*oiKq<9tpP5~>-MgVJU zw0S|tkDdaZLjyvBSlnCq@42ywWBYIO^U=R8|FJn#$_HCdI0p5D-Y>dWuTZPa?ujw*y-Pn5K zBsd*LkcVS_ZR0?1~bndxtK|A<`CHk2Lq=KuPj-5;?r>y6Kzy2)~ zC%3E+R9RvP)k_!RuLPY#i!SP@UBA~XG53dF6ani2oe2OH zq4Ilx$J0@#8Oo-Pw)T6*=l;UFy9YpWjh-4pC5}4fiob+Cz!?Dqx%wvINeQ@VQbUgW zhk?^!!0ANL=j81G(p8`q-1T8a1i;Jy!$om0>N4QE9%Qg00Ave%j6xf34HZyOCNxoG2-s2kDSGPog0i^X*{ccKm?h7aOe>_14-|+QTp%vT;+KI~Gu`R&5aG^X)RbD<1aIQNGRRXJS{r9^d z=u-5h#RCv{=~}1`1h5qp>wwSt_i`4}Y#=}YIs!8nEK6|^pS6S&!4d*CrBVSc(dYjM DA^#2| literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-23-Graph of Inequality.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-23-Graph of Inequality.jpg new file mode 100644 index 0000000000000000000000000000000000000000..72ad399c99912dcfa8a899b2e3a32561fb65d0d4 GIT binary patch literal 16580 zcmbWe3tWup`#=6nNzSv#M%1k0(*c`PBI>CPIc%m>D7(!%Cp4m@;}e!vWJC(7nS|*e zIg}yQEG@MarSo0}*!J&J+8 z-kdo<%=^*!2g8L%bLN;>nl4GD6mg0lKKxa@`t3MIYw)&_PFsACc=R1 z?+^`=rl!UiO)af4V_vGBzlT#dOF8+ZN2&D+U$?%rdkq^3Q7l9m0G^Xz%ftAf8?zbP#G zySSpVs=B7O?w|UWkFB5D+B-VCx(5V8k$6xtB$W|-kr2&cw&3q!#{Pq^slb=Um@%4T zv@{= zyK?c59fh%5EGf>m-%-^WPWB@?EULn$ivNxDlGOP*Hmy_Yt*O#PJ(EI-SxofO-N{ULV^IP|n)|F)w)O&ho8HL~M&yqeR4X=fG@ zV_|kmF?sPMj=jR6KEn394}=snHkrQAPRDrbZ~ieMac3%S)MGo%*=#TR;GKe=lcF{*D#xZ0?7W)%wgk3z&hGgtA8xbFIoRXP z?=){Q|GMdq*ab5m{<0@CmREeYM1}ZcD4uN}_HwyAcGaIzIT}=|-R8k(D&&QwQi}Dn zv{OUN1erA2F20?4yLoXQ8DnR!V?e9_{YiJ5n-XVtk3C4?4Arp`w-TE-M?qIpn%#g+}5oF4q z@EU`i;+*;D!X~QX5k|Ya-X~dpOz{LuM+L8y4%vwN<@^FcW`_!?VeeBRIb0^J!@AOK z9nlt#E4V{0`Lz_ueigzs$8`{qd6pu*N+OXt#|dq^MPe1wQl?NrD@n!-faSsEpHTaK0JN(NaP$8dj6F}H_| z4bHA-k1pVv*o|5>w8blTsgPK6<(t9OZ)G-AB0VI?wBxCe3|sTOd^bMvf6fm6lvwWQ zrhj$F%XfbPOAs2ocA=MgEl4J(cyBX*s7SA@h2c)wRl!%=;xiQkm_2_}t?y^x7wx#9 zLe3(a?hZ1p1qFd*(q>MS)}m%A^-yC>Z*A11Gr%av>w0-ju4^LMs+=5v+a2kKk#uhiS}-E&k(nOa0KaYEfT z&j@a1{rQQXQ8VI1{L$plZeC((PbjUS{``;0b--*NXy`0L9XT}KqesJ(`Srkf|LTyX zc;Zi-g|HZ8AKn<|rb1j-!})P1NDubGxkv+q@e4t}I|%*uh*AxDiPOOu1{@%%iB<+m za{Nh7EzYB2*xhB{fXs1o${;h=hYdo^tO%d4rGUex27`TEy^UKpv66YVLpvqb=Zsp; z--7ka)=!sm>yF^Cvu%0G{FuJ%Lghch%-J6RDIUSyxz(Al8|q!eT}$ysx}DW09!x6v z5RzUMKpfGZz@C8kK=n9YWnFc0Xn2=3&Kzrf(~3X06r^@fd>=nC9P5b3%;I*- z@%`ai@-i`g`V9UkEJEhurh8o6(;cd%rCaKmM?TKCopM#VGt{(t&L8##-mF#yqS#v7^dqj3BgNOUkEpbwT+4gh!z& zWSOl-Xr&l^$B-Tza&gu@*i$93$vRA1#-+O0PQh|)lBLxFmHbcK%^x36{QcKcvD=)s zKbUal%uwWw_a`ppIUMiVQkSnnZm=dXCSXNR4u~-25THW}S5^P&_-{jT2$jyNHx3M$wnTgQ};e3yQe&0 z`Iuk%wVdO&Q|P%&TIa!?y_CC4q63UQ^V#JUkAEzvWv8;V&6_Bj&XL8nsJII2oVeBO z$M^8-l;73~AAqU4uU{tlKHOM6-3t>R5iPG-*O(9_W7^x_mL921>h;pFtGLMx#JS~8 zQ++}B)$1CS@U@-o6tqXT6T7=EP=y${V9&5?Yi>&~z=Y7W1%J^VIo9?XbX0^dtaM2a z^->G&UfPd4Gie~};d#?S3mi@IcZ)lNncnl%S_+?ck4-4Hb9ngeA1iB9dqcfQOo0OH zxs3ZSWFEz5jD0Ql?ZHz5Y_b>r@q+Bx^cLCZNX-djOW&(??6~$!2y^5$o!$#QJ3!$VMl3tT2kJJVtBV`Wci`^}QE&7pDi$2xe|VZejQ7 zEIbRf6YG>;Xt(q{^IVej{wPc&R$?eULbYZS7#x*9(jsZ-Pho%6TUGysyqv*aZ`1c7 zJwD)`jw|1=CN8g2YtDIikE3dzLIXgZoh-T9qnvEv9KZo?CS7=LB^HkcG||j=3AW#P zc6`>I+0(`mq#jiv%``d02bs>Ew|4wOk?n^dhliJhxgyo%=lH%JC2v=s9l71SG>_b` zwC>z4Ww!a`C%!tZLUL)>atwHzka^@bPH7(cLWR__!AF9r2YcxmZ-n#nw{>-G875Xm z&Qg>>7(wRN+@6a!a+j)*bhMz?y4w|tQDjsNN>acCvCWHl-DPxN+i69VUD7yQ3g)d6 z%*@CR9iwr30jKT1DPBZxNaENMOeQi?u-kmtoV?j8Bv7)GeHWQR@sh@+;1n!lHCYL+ zEFT8dVlg2ZRVO5Dx(eCHHACL?v#_D{_wjw5N^S!^sFtKC5SrU5P-Re2f3gaBZE--r zap2<(+n!;~lxsOts*shT$?cxWL!o%qXH+t>&tGw80mr$aIf1K;Riuy}MF?u!J(;-*mRs0CK;v zQ5N`#QhvjDzbG>BhbL6XE-#I3UufD#2pZZwBM8QpDgG94Xw*IS+}?&ZpNMQ^rFP+{ zuAJn|c2928pn>RvBD6-4ng-@lkx^S&=Y-fxF9^-;6(}_*vLA&gA+DFwGAOIMt3pn4 z7x-D{Uz^dYLXsQ>oJQ-s6pfHC5_YxONsuH!!BHWPdDkpxvC^HHcabsZb?#CTS|F0# zww04hhXME%q4TiIiVTq6H%PLTQ%m!x{c`2O1{LxTz4#_Wz`T}|*eoJ_PvIAWEbKwK z*$YpB_<}Jq!1*>0zZJHP2p)o~9upD-IqOH@;9+@y6L0PMuzlTF1g~cpoK+*BW+B=_ z@~Fa&${kAc9AHf^vb=ymLBt3SS=wKZ4eJSca5i=WtIq6~hkApL`OIl^a_7n>bKz*z zad+AH`1j!(5tpS*{Djay^D`)q+wzcBm1Q`E5; zl*Vz?8~ROtRjqy-(E>HGs!^=iP9Z({d#21KHR-W0ctRq%0&8zc5&zA-Ms@NGKl~#} zg(MzPq|p|Iq8^BDA6hG$K$g9hQMw#=)E*z;)z&%fp)EQ2Iap~zs=(6ecde4ISO^*} zHD~pyXH4&9rsDonRR{x)`vV7QWYGE-RQ)n6=#MvCA5XOHMqjUqMxad#+x6J`7i3Dt z>Jl2YCVY&X-Hjzs1eN{w?i|V*Xp~5jOK8j3bPwdX^1YbyF2OnXP?m3FfP@`*s7qEp z*5NGr$^D58#m`kgKx-(h`}7Hh5GY$5C_JD-%ImZLI@E5kdWs0?kSBMe30BD`YrR^Z z_4REL)K-dF8mRc7NQW$r5>aa6=NxLw)Q<=nbEw@?KLX1@mSW->o)}DJ=VX`agRL~H z28PrNWc08ajm>9fAi!BVmBk7jRG2<`lkYK57rcx5Own>w_Ee^KX#I<;m+<^I z5m6yeTOak=PMZ=_t&XWq>9w+ZzUV=dq1KDDCsr84>!q|&mt^7NSQPY6&GPvvD9bSgfk^Hjsesj=yF3BCqmqMj>|>ckQ1XFZ7uE1(xD zo_OL_B-q~$o}JPw>`b@mZ42S-O9kATh^V70dso8-N4#!ZCWzsZg`gJbf;Sjcn3;n6 z+lke>s1TOL8nr^iIs=&rM+y~L){wAyBTt51y_&bWR@$PIpI4V_#`SSq>_b1vRERxD zHQ2YS`CL?hvA!}i3fZ`Bx#5}1rd1?Q#gfY26W`4)@!3$+wHd;BlMlwTuf+o?c6haxPws1_$eOHpY{TQbj+kHBgo*e-*ft2VkadPl zfk)(>L0B0?y{}!KHeqF6Sl@dU@-bC_681cX57s9evAy7*h06@sxI2X=Nn6UH`OtXw z_ieF9ustelkK5<`w5Jr}9Wqt=35H#xw4n?9(RWecCFL0v;fDS>^=K&B z8L~e{ro7steWEaB$KUz|hqw8frL`=1_#ygMqHlBNYDIFJPqI|)8>&cWTil+jL+ELL zjiuaokasB4Y8w38DLi|~_6d%+(a+AtP1Dh9IqE}9kwmoloc@oss=nELFt^xd8=Pgr zirRj+p5y5OQaJ-s!Og-so_oTE|A7sr?HG3QwpgP;sbh7{|a!rJKp8Ry1%OF8Yap@ zs}_O<`D70vMNoh%nYdZB|ISj5>=IX4D#P+zIldCdAAEMUSz~t% zn%PGOb>zC2(gSlTjee-0RE}jkH+IKG9WrAYX;J$@&}qHW!r~ykeFa$Z-vVZi*PwWH z#6xyn1zwX2WyWDXTrZ?G@8eeak8UWK`_J-fVu{hwx&Axfe!yFk2r)J&(BmEG(Y%Bzms zT6+^ATi|yGkatlZ9g=Sq^7nCl6Ryi6U_`Wd=PT|A?@uJy9RqSBp4Co|>%BQD92N2v z4C|`t2$0sNDxScxaI@xpzUj@O_z^iK-Q`*EgwPATdQN=FPEgn6!-4hdYWzQgcv zecmb%`mASiHdwvjpo9WI1npbdxXpN)dh+Sz6v5)D4vl~iw@_H%QWN+6Z7H>pQarLo z6e+ElZY5G@O-97eRc1iO_A}b|4`Do}yXB5c zmgu7$H{f4siG72P7CS#tOuI(q*#gArVtsC#`ukP>n+xV?OXjRXk3waYTc4JczejNM zaFo0#%tfL?21nKKhB27kpCe9j%wD#5^Z zr2|dc9*Q=n&CTC0b<`S0>4=ODxp(U>Yr1nRaP^ks+2gEjvNL|cZyrxs_p(nBGA4qtYQ} zVfxUbq5RPNG2HlSuO2b=q2hwnC6`#yb+~K+W;T|2*22h<{mBz@9|Mm-rf*P?lO|K` z{ko%C9g{zKCUf#umr#4n+iYhU=%|qB{bevsHhbGM@OELRALW|rQ6{@9F1(jxPpcmo zQ5VCGPq4Iyv)$_1Z{i(_@*L&eefpWo%(0ILGbIjfo<7Dy5qHnwGtC<*o3xvk zN2WgvNoLMs#mE5-h2kN_hPZCcJx&kadV;{W{2O}>U{B4=kKn>Xa`>{(xpcS8E2o&t0R z7)eR_U3cnV6c$?u<DwVxp+stpT{VCExL6p2z);(0Uog(QdvH2bYHCaNVJ@igR506tS9GaK^jn>%XDg-Y+_);zEpQE>zMyQ2%)plA1 zp^O!TdN!oJ(1sm2X)toD)FLfUf1d9w-Fj_A8(MEIarC1Ci_VaBh^jT6E+7H;`kV=( z#avDC8pZZEdE?9jA3p1#yr8#j<E=WeU zv)d-r?XV{A;{WsrYR^XggCd>pZFVc7%0cZULryGjRad8J!%lgJKVPpgO9X5AM zJ&%W8zE@PZOFCP`_CXuz72$f34b*o*i3K%L8SSCM%5tZRCz#PWOsa<^WmzvQh3N0} zl9oBw1yw%fNXkB$J}C5(&Q1}DN2V{yHOWxlWxNYQcPR%mj%Kt|Zi)2I?mg+uP#A5) zjQlWaU-qGMHn1d1#3EHcW!w_$M`Wx%goWTg-TZul5uL-JdWH3L6iQM3B==xTmnR>*}+eb@TJ#nwp{AzALTsChz|Za^{-E+|nZ%{ZV)x7JdJ1 zq5NKMBXwX_*Dmh4b1AE}OM{$9jbVF!y7~~TBk5AB;O3u=<5tnG!#TO5&u+Ii9lr`3 z_T71W)Rcv2o)YTq{Y`o`q_7WSzuSr9|CBl|T^jOtzS z(~faXdp7+AGrMfZPO%xdgS6!n-}bJQ-@80LR!a3i76&k{i=*?j_5nDP=IxR={ApWdZL%9C|M$}{xZ zMlcP2RVO|vZ<@!z=}RHitu8AKnlN(?`MOwVZy##fX|nsMjY}1j7O>odX;WSb4Pu5@ z`3w%v7SZ;VMy;?7DGJ-8c4b>81yEut_Zq{ALop~lEhqhpEQzwPtFzftkw7d}>6&n+ z&DsP(dg<58?TQtJnIhKo=lL#n#tT3$q~ng-@}JDFCvuBw?Z81%2Y;uT>SNIlPUlTB z|5HgtXAETiG?6zn0HhwkK|4}V<)-gF&c`qKy3Zok^Yd}-mp+7FFWWZnJEW&vbYE^h z`Q(Y-;@Ri=o8x9hYwE1<9bgVk_Ph1wkw4jrvP_NU$ua!d{a4QV%F*A+Gi=OM7HbkV{Ke)!Yg}aDLOu))+VJ%xA4&PjW ziOjobL)w%fr?F{^>jU+v!jubSj=2$$F+6d36*jvbBW5tA#~HO7g|raYAwBrK z;*tXtcnLlZ^fS*5GKScq2g&#^tzy}>Uz;x^*`>ceoJe^e@}#ZJAFG7YoTotOwKr1KoZ_vwqfgR_j=ZK~5@qs()P zf?*kAB_JhRz1{+^PySG97-h{%&006w_70`pU5!sQpA4l1?p^G|9AfmNhl0xJa;KSr zQAM3-#$E@|$@N--Hx|Jg2;&r?CCH%&Ee)?Q?NHg%dfuc}w=^NuC=$>3F}VdQvUM4` z(L?ktM}>E&k89>5s~WBr8ZDf%s%xmaZ)n|`)hfhR-&}Poh*nBuUBG@D_VmE4~h@$(@SGELu{Sbc+fb=s6Z`q1UBP;OfV+J7HUZ9gdH8ZD2$ z_R0V5yd!HO-_s`y1sucvir5U9giSQuNNxivCz!+(wBsGxpWL0eiR8dN+QE7ZM zafqBf3SK4Z*P|cp?}}6ili)MO!%J^cPD0z0A^Ik#-xu#1bEMFAT!Y~1#a81?Z=2M1 z%1(jNynTWCUb|N$I&7-=uVpE(^7ZR@19K~66saFWVIpH+52Y8qD=Ku}Fs|g7RSm13 z9RV@l4_~Ct8d?#;%ndN&PT}$fky)p0lpR^~2c+y72DVy;3!_&7LRc$q$wdPXxvBHApIe8CH zcb*LXnQEcQUlE;f=TgFI*OC_6lr}CfE(XR45i`;PBWe^E8>3pI%gg@)kyhBu=@y*{+f?VGzrLe4_R6${Cqv*{IF>FhAO|>uKpz!!Q8Ft= z$lqsVB|3n09WP54qq|sZ#*I>H&n>ZYIv?@2ym0hB%EI;53H1l-h7U?f*nyb17!Y%xWf}Vt7=>bz_tB={d)ivrRw|Qi?(5DzK z!+tv$p}F#V(A`<5_kQ|g2mhRvYjMwR`Cg+pH7TXJ8cR1#Iclfy8rJvz^}-@B=ZjbA zuRBO4RHw$SHmI(TPptWjLWq^|ZwL|VM6Q2X)aAkKk31QGhg;Eb64;PB#e$YfJrx4T zhMQHu{udxW6PR9^+HPMYS0QrK3}?{iw@ndpjav_+@CPl@<(6`BmF1i}vQS$rcVhw3#2Jz?v;jE)~OA;5oV zW5|M*CcawSDue_DkO{Lv-;GRyD=_PK!XoO275;}`g$+;Fh%2z*y4U1W-XHea`mu7u zLr%i7ryDY&-H$CkYFqZxIkLZe1SE`ON-eS4h0r*q7UActy&E8NGXbzvufOhOj4 zf^n(+_ln*=Q~(|RTV;VX67e>z4?EEvY8@7H#O46cCz03k)3PJz-0on8ho$8W37P6_ ztC?DFbE4ZChRk%k<4*qZdHw74>w`XgjM!gqV_y|ukTpW;P5*5sajSHq6KC?gYbS&h z2sw6L_>Ab5^>pI*_V48kA~K{eRz$FRO%VUp4)2;A*qIeyw)SGhdx1gnGYB>5y_8b8 z{c&l_eAlGS9&*xwI{+Zo$6tH6*+V^jQuCe=AYY;51^s~tZ1 z%@Ot2Vm_n8x>&!zkvX@g+j{bXWrtBvh6&gnRCEGVT zX`a3nO6dLO7rd-owg<%F;ue2g5s{f9(m`fGw!nvMA=oYSC#{>W_Fazs_4&&}=OXl% z>eFxA&5MEzTI+A;?uSm5x$=l-BG!0A&W5=gZqGNe%XTG&qh@X{vC@W2&X5boQbKMw zO?w+lpwO1&(r!*{ijzHU1Tx|z16Gh?}L+dPvOajWLc335x-0YRPs|25&l@z7_A2&Gf6wYNJ#6XZ>5 zph_Bu@4jMj7*XD%NJKQ4i-BR#)fBduv;mS4nbWM?0zzp?|2$<}7V8&gAPXGwHa(ek z=eN^i$z5)KB|G?^=YH3wX|m_yu?;im>NCiCvR^)#3kM_9x65|gS(`l}Kr>kG)P0n| zH~K34ZM$Vr(3IBZi2Z*`$Ylvp>+#% zhiv#>cg!vFBJQ=#?Z%Q?VM^wgc9BjEi)c&q9W+Os9lI7%PiQr&xHa64fYq2oK)VS! ze8o7HX9Y=K;CZE3(W!12sUqoZh^M}T7NDR%il8c!02T%R$X0g>R5{>i9R8frZbA}l zh$p|}Caf-{_GNagWge|e(O4e6f|(EmR~dol+e5`&JzIvm3slG=bOLgyDBj^+o&y#X zaGlAcy^9#WFA86>#?+7u1(}}r;4GjbGY={<(428U^FKO8Y$pd!_mbA8z}1p)y1YjU z+xH7~5VD=d_!cF45K~wi68D6_Pg!L$RG*;BV&r;5WBezoRYM0&9Kn@Oc99Z{c7J<` zLA9M)Xc%}%m}%F1%8*V^2o7L7;2AcjPeFGAYi!o}=QjR5J)aMHtgQ?4@9Npa3$)>u zS8y89O^COeC2Q}Q^QcxGumEwrv`x2{ao1j3*29!_qarPq_9y|b!09FQgdhXa@aRx$7KTA`sX~&Jz4@zrOVVH{xaRKRvyf+Mt z=ae{fnE<~*Ecs(h#FVj6IV_DyaJM0xY6^(tM{`W!~SPuvQOONaD#c&CHelH ztCb)h%?{7xyp>;@NQP&3JsMGPs_CQQE5Whnrw51hG~}&YbG~C|2+`&PSNCbY|JhAK zGFKxle$IJtCwbaISK`$TM1?4^BS@LaVK-{N{4X7V$SGokBhQb)3fNCa=() z)2c4zu;qLicGf;@$3e;b?Y@GBi>=w_@u&hI+xjuw;ETon;^C{OmDqHs5QjeHK&Q#l zvUYPBx<|36K79Uvxp~$Vcg0f}$yW|+5kkqAi77&yG~aKj6mnd;dPFIaOpf_GXcl3o z3?|KI9)yv-*u|i#{qE3P(y2lOaGf+03P1bTttP+Yv4>oSP1lSuSIQo$5I6{}9~0W4 zo)7iyp&^#`7xTTovORRLi;vHkw%Miu( zj4QiZ(%0<#1o7=fETo1Lk6y>#%gL?(^Ez2rvV`)+`e5^66(WTC3?mW=@eawBHn0e9 zp}|~|f1=acTA60j#w?+U!U~37Yq%@kfMhahw4*QG0;Vy9jar@7TcAaiRRGlERe~CfyA@AZSoh9^p%Dr)#ZLgV4gY z7fZ;I2a&yhiw(gFfeF*}g>t5PJc~7c6LU$}iOx{#ytDgHZijf2zAwD z_|DCnp%n$#!!;GV!xs(9$25(-NXVhl`|phk253s@qA=+C&>@-rSA1~$E2PiQXM$E8 zR7m0LuRiVz%zhq)w98iaL8d6IV+cITA;f)(gvl6 zEGX3UE7H!rbH@wlgnky*4i!Qp>V%Lt1n?w-NmU<0_*D>#tewv+1<@1?73Pr_4pBOl zg#(cY^B^t{W41U(QDY&-2he}zd`~R4XawyhE_wMU_z9*K|+e}(^nQL__2=ZW+X&Zr~6>9%E1+Q5ORvN*t) z$Ti5>Yaf^85tIf7>^Iwd0tFG4$V=ns9ZEj+(QB*Q8|NP&CDmI8Xl^YZqkCuK#M4`} ziLQUOH>ukG_3}qV(kqH*c6jFJzdBXLvNvh|Dko73^yYpD>kYfTTo^0U$_kS z+lXi6$8;l6^Bmb7o@{R3+Pze;VBGF9%-2QTqxfy}DI%Tc7|}s{m!OVaZd0Ef{3iNX z^Ow*@2C<2{WfhPCcRz^5C&y&)65PDuxWj(9XT^4AqE58-x%;?3aUL7IYo{4s4+E!C zv;9B1S6E|n&u4}7R1xhh>WD|(yb7Vq%1_iq#EkM$cXSYhN3@>|dn3{qfpNkSVWzBr zr#zRcAQ<)6fyRmqZV%-2<$7$^xYp=3+0t58pKY`y=Uv{5(I7<5A9B+4Slh^i-b83q zAzZ6Sd{^Bws%|+&EcOx{C1y{WW(@WsQ}&XlJexjI;!u%vM(JOs^ry%)$~Egp39LB} z2}B+_==g9ig*|MNp2R{nT!)Ydq5!bt#5yy6=yDuU66xM(=yy9e+UB+8JMK}*U~1XF z_p{heQ-?)e>4y9AAq)Z64xOi8XPzHc(yfuGai08KIC%-0&rPrxRq{w9Y55p~-4N#! zO{S96z0oTu5KP^}N(P`#YcH*O99l5sL{e+|-ktMViXM1^kviJcf_O!*DIj97i0BoJ zgrG~fs0Dwh)FP`Jfsr+8iEG_D4?SZLk7i=tPrf#y+LA%t5XI`_S!>Rp5@tF#-x|KK zqri31@KnFL5&c$;G`2Wtu05ZZ@6T_(owycS!r&5J-2q+=Hg?DrdRrcVPkz9LlAt#d zHyrm|S__T>!km##Rzh!lVWy90`DlPxBH{HXF;jmmDT;Yq_zYujG%2RhbyUthbrNEsY0~02?BL3^dxtgpEVw1s(xJ47fkjZ zk1c^mBiMy5zU`Rv@~6^xcE;)haz|<9Uh_6iaUPoY<;}|ot@8rrjhsYWJe3NKD)CD9 zzFsp|^@|r3qRXwgJs%fJLfoqb&1>=&ym`@@5UivS(LNYYh4`AXH#82u_ z+5}0%iL$|X;l1I9IK;}I;hhBz^_(847EqfqOPjQVG{Oe*_Y;BS@EZUMG&Iz(bcM6q zpgA=^zlx;O1zmnOpvw<#d@QI>l=l?f<4dG9JdnkG^fg`(>?*~2Gu7O)#=c>`pxwxs zA{rxYX!mLA{5O)#JhMUE9M483F_(tCg#fbRaY#fv2<9T~SV(%Uyt%Oc+68!ZA;N(l zk8mw&Ovu~Hnxs=`=y9Q?`h2Ix5YKz4eb_gzT_~^dS8}H@_eW@!cM+;i>a*8vWSWU_ z$s>3P0e6C#_6nNP8#%hgMi}N-O}&BdhpfSrc+>)}8on8M)`CkO4zdt-pKiMkBz?DY zsUp*Sc%#D29(mHDfE(k;elGN+Zttahgd6)Ya<$sdmjFXG#51Q+awP;uJXmr<*g)9+ zPH0Aqwy_slc8t9F8VI#c`HC8rHhA%aTq|$m;;{Yje5>-S@<*PP3;+R*Ma-_kcDmS} zo447JkX%#uLT&@2nA;rzUyr<4Q44q&=gu}RLQ5B3^Eca4zk$In@Zj!=+Dsu7WfhX# z?gM@SQ9rRlF96<#pQfP1s44&8m0|u`H!)U&;w?0*%fIS6Wnqc(9IuTgKBVx3Chr*I zzS-?QXP^^qi#8xHh=eW4ck4ti^ziND{*-P{`DS0Ez&k;cr$dnq6Q@3k6VwuLZL=#7ifZ63kSx>XpwwSt^80|S9=wDp6z0HSTNQms~b5jI<@QQ#yRF7?6x>TJsx2>(Wf#QMFOT)lglvqF? z;NbC)>lbP^6RrIPS}L8PpK=jmJxavBC=oK-atYh=YqKgc<~)ONx_ceVh=q1-9c@+( z*Be!!*@C~Pe|8D!k&rR(Tf9++S1}JMCD(j)kx-rpM z*7y*0)cW7?!M3%{Jg9=$Z-~E2guicYaYG^doZV+!E7LI(&eQuwk$k8h})hGh&RO_1B)eExjr`&P%^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}niW(%Upstu+-}avWv*w0k}rZ(Unz&z%|m zy=vN?@8V(`Piyn$y)8QGsLtfWf2hHFmiFxb3@ysO{~6~0t^d!!zJ30mvdIUY1unkl zta|jwBVM;>689dZO+BR3qM>tSa^tHE=Bn@iZ1n11Pp(&vl;ix*Fj3m+Kf`*7`Y(*p z#qQ4?RU(gl)t>Lu_0;4(|Bv;amvu{i=L==Jac4}^I5W$7+O5URDlvTZTnw*Q&-xJl zr}?Y>Z|>Ot3@|Chv<{69h`|1+F4Ek8WV~`3j{>u2x^XE0TfVa5=b`6Q4|D3P(*|T z5JC|n(ou+tQlv^%T2y+CKtf2{xhtUOyzl#c_l|q-828_W5e7;2T6@j4=A6%*&zxKM zS@;$D?V!a03rI{1g2cctBGR6%a(k|j&mE?5_%bnxWn#kj5E`sgN^JfI`uQU!E+Hu;Ewe~=@e*)B z!Ecbbn1qD5q=b}|q$IeS0KSJLmq{((pl2?J^1;*Q1jcsqM*qyYk4VAL`rB zT)iQ?c-87PYZW(c($Lh>);BOT+Oc!j-hKNmEUgY4v_I-_%<+%oPA+H9oxkAf=I-s| z>*s$h03Y(#&0C?j?}XitiGA=e?os^Xl+>qb>6DDjte1KD1%*Y$C8h5w-hZg9s;>Fi z+|t_C-tqZM=eO^D{R4wT!y}_p(=$xgEPIZ_70D$ANi2v3{w@giA95`Na*0byN=V9x z~F#Tk*gD0Dj^0APhuH_f!JKll;FkSwRSU z_#>@i>pXWCd@F^0n$`tGRQRCZ5k@`gO#7|Kt8+> zU5WvhuJkzJNmQb;vE(eSCI9**%vdi*AnB0CYh)oef)!5$gje`r-VrjQs~&b6?d5NJ z0=Mfn3891PLg<(f+G{2~HB4vMWaxBzl5Pp1Kj6Q(i}P%q1tUUeEECxitn$wUl@%2* z{h#r?TfuUDpJgwm}_0%y^uP7k_0hI=f3kepnE zP!$u$su4nuTd8b#8GjgB*zM1vE3n{-`w8eCK6ZvI9jfATxku&)2_H64srrr9uKVNh z6Zn}I87WxVXhj>#65S;38q zn0BKk*a@8_a9y~WfmA}&i>@E^(T+K6+VPAnw7yC@QldLemdTE2tnfufm&8J-&@lvS>Ij( zvD3|w(jlZYN+9NL&+eEaOIP>RdMYI&D{K}-K7iyPw-U{k{krMO6~#QC6Gq%3WW{K~ zF2N922$|bq`0#!qbb`*wflMt`oGt2b6sZbDE@}=r-LLP z0d^)OSb4vKtr&Po+P}M>#W?Zxu6N}3gjp+0i4clXynnLGLa)`A+sqe2<4uywI?V^D zHC?>Q>uxfS=o+W_mc40F@I#t^ngsla?oUSY#he1Ic8__*AWrD-m~Qt6IqN%#HuPTo z7{*LaIfGjvgvbVmyJktRnuArHtqf5yMeK<6`5}QpNu>Ag=jbm8y|E$0SzcL&_@?}j zt3ioE=ruv^!YKEWY6H>_VVV*RuXjLC1`C{t{k$@u@Q=&E!hqI*9aNcyhc^YQHS4V3 zrE}+t5ZXE=gq9)0*@xVuhd9PU=o)u%s1;ohewqyfO8_cgKrcZUxM59kH6RU5!-a3z zc}HjcbXWg4e7ltbKce|3Be4f)weYK^CJECT88&9B&$g@6ycC`n&-6x)XjfWS@z_{4 z+>Y+ogbeQmgirfn-V-wJH_L_g1MQ7C=+6mfb^p9<}HEd^@D|O z2wwZ^lrH(*^WY^mPNAJhLqh0$f{K7vF>h73oy-S~YCglu*cp=^>)+!v`1!c+24}ry zlqQ+z_LLI(ZEz+Z$TEccs7bi5GC}#Nk_@>LMEFe_cWvQ%55sUXW57c=ka5M z(i$&g>o{*+4GmtlWBpU5j(!%R=AiE(gf1jZ3L&tIUykL=3&z-i?XBgKE{=U?k2skR z6xDnN>!$qN4ALecq=PW_+Hkl51=YmEy zUSxcxLa1mz8*Zb`I6><9A+=s2%6N?MuxdjLpJ0^eOCskB1aP3SLdljMpiIv zq6AI?P4t-3FeV2wH;&rXwcwG3(3Pz7SMq-V>v!KNggTtz4*Pk-qx%Rn#j^VT6aiGT zj8fOF3I3)e+GT>44BhKpuwsLe2JnECrZ(%7RkJ!mNN(P!WRa2V9Cpl@K<;Rp5b9q7 z3!zi#LvAxxndQ83g{R3Ny32Nj(K!)YH=!of8LAfC{U#}9RPFDxj z-im*DvP(Ut2~XJkxQvMPGfvdk^yW2x_B@_A@F0Ewe*K;p(sU$SE|rSiYhxm(D)_{t z?q3+MKtPt_r4k{8f*$`22yB3Dv{eo*ge8mbbpH#EClY4_Fgb&SU{b)0Ui|i8v|5x- zu_viVe_}oo*hd=e8nGcu?|Tz+mt|tEJe-CtB~mz_%;ZzTG$HU3yPt7xt#Z{&+H^;A z{EpZxj%RpvL%&XUV3=nYl{0}7a4~!b$aF_X?I~khHJo1YI3{PmVo$(D-Y_0+R1-oU zY*&rddC~Va!L@1(qzzh8S?*heo(6IZ%h~H+%%AGaYA8D8Tf|5f#WDlYNFQ8qyNM};k}QQ#*wZDxWi$fHi*pv9%Yq)Ho~)^Y zhuzbuibCiN9e~ePk(FeQUU8S^wpKR3ko5Td{%!ehiG8~xO2(1|W2Unv*a^k_=gr}M zc`-`t0GIv~e|lD4Rl0lD`~Fo3>TXuUjXTR@1{Sa?!XKjyY&c7CPY>+!$9LB&v)6W9430L?uU$*5d6#jU#paYAlhpB1mLdlRKit*kj45meZhO9{9*Vrx}&%9+Sj2?ghXXea*b);q0@ z`lG{piq2hprw>=Ozlxs+59a@yR$7et&VB~g!YvZrpbhRMFWDy2&(6dFYMS{^tz0l! z`@$Z$(G_NZ#OZs6xK*sCy8IC5)8Xsq*UqR%+`=yY%b)vLQ{4SmBBu2&QnufR+sqI` zr!(ME(^O7NhWHLT`{K0GK?!84+ybEEE)44;lRTdf-wmDdlI*p1hJtD*D>i2AA@j^zdOnF2Fs2z?swV202I-(Bb*bG;_ zCUIr|XXuFGAGl-+a;7L?smx`^+K>wzB`R|z&PoW~(+L#>Rsnks%clh(v?g1LL&y}- zVi*`LW)j@=)_=*mgwU)axd?s?ix%FmU2pEirT}H9a27)WaIbE44~m&g@5iXwM{z%J zLAXh}+$2U5+)7X7Q~EtAoYvS@9z_&jnl&PctE=gwAVoOHm=;uyViNU)P#;&93&PcU zcSj!WHsS+$fxXPkrPZ>PG7;a6hL~4#q#-jo6t9v40*``~wNqe(_I$7izQ@j*io<3#x%St4WLX)t~Q90BUOTnWl`jFSV*i&HuwYGEXHUhI}f0>Mt-6m8DJjm(;&xL;DvlF zAmfH`(Xaj&cn9|sRrkml!#p_vQ}kiCw&Zw4TAdJ5^F?l?_zNMM@P02A05uBkwGhG$ zYiod$9;S^Y2q6Qq4DctfyrN6(ftI{z1*uzk>?m~B`X2GfP7`AO5lRono1_rE^4XsiuFOa->%?X zoYBRCP?j|;#;@eWkr6bt4BEfjZOAz&RphRl6w_T>+&Gp8Z(T-wi|S~AZ~!amWo?(~ zB%T`)(2+7|iQ*-+=7^P=^~7{!YZm%XP8@{|bD6=Wvz)jha5ZZfa&5U9luBh9*gHmG za=j3{pGj8E3qs9|E^d;C#%E@S0vdu7hnVyu=vizTq`1!wq*&8Yb8O&}%~yK3ST{~b z(wZEs*JEaNLhEQ#3^bVgrO9Y#C`g31u@J@!HW{i+I>S#DWm}2?9@CMxZlLRM@p$w% zZs4c&uul28VK`I>QO)G1TX-qtq6j1GbQCL;vzUYu3;=wBxh$z-LJPkZc_D;QU^>~u zZiBiXccDl2Rwkf6YHr|v;KU^(XiC{~na{*?^COrLLkCl;m8s}~(FA5;{A(s=D=ME~ z3z8=ktvy)PwPoTH*iVpO8f2c*?_yip@kckIW{1ncB*3sMCd*lyL%Y~<83-EAfsAXh zlVO*xq|-}xk%nwQsv7_^NJyFC?>yo(3sb>hks*sDm!MRU1=YFTF1JH+-EGlzUh#O; zH!iQ$al_PVa8Rq~z##zmB=xMcn4DgsX|Gl8$cv7o&_CN3Ea`Lk4YTvrY?_zbJ(9dPq4(jt9m`K1*uFH6n-tO%cx2miOXU%w$53>W-&2n- zrz~*?XBy4?+L##=ZN&Q3=r@=L2WIe5A5)8ZW?f6rm;GkRGEygsl!vuTP8l~xeDSoz zVcq4&scKY80VMScWn}zNYSYR79_u2LV#TN;Z~@W)T|ej4q?gX(SmGN$g$R-^Z3DR$ zh?dR~#>_%{Y1`Nc8Y#o)mrDjX5Dk}GBmrGbrUbYNK3U(xJfv@9V!)ntP=NNnrm-7A zmXW_*F*kjPV<|GO2#^+k#E5dM2YV;scmT8hp95QM(BOb-MQL$p4`EpRxEfZc+^#us zfWqEyXoffk)wb|x0=yrZR|_K6d23O1u(etPT-`3sa6fsg-AN(6VGu#V&P)otmSH#S zKViYRZ#uc7AOF~^QwbrYgJd19O#1~Z`TlYPRIYvx2Jj1tQWXIaNCfb2bQquXuK;r- z@YYNY?GqUL2mDG<$5YF1SFfr-%KCs`+3&RA?li?E zF87`hLaEkG;3&F!?`&17^#U@Yfm4dG`47%;86lZA)ak|d=esKf-s-`C8;zmZ1evmv z89E415Ed}F4Q6sia)Y{HBfvLh!}q@xqM2^ADf(1qr>R&M|@|2j4^YKO}0IbV`fWdu>9#5uvLnYve+-*1+ApZcs#-V zgC#&}l}~Un<^i-mc7W!qf%BRof4oX7jiDCJwT5I9;53;m0h|U|oZEpZou6QpeLiv2 zCW+p6B{V>XS(Iz2e8H=zBF{S0Kpl(B!Xb)*@xH8)nxNIW<$O`~vf}-Op z%balHyR&FY>+gSi4HVfTAsJ!JwUjr}XC{mgW5i&~FR(oBsRUS|RO=1_$(&(VRzWD# z1qa$jV-p;EbEv&A2!&QG@VoVZ)oPMBTkNF+jAjIB{Ez8GDdM6(ACT5ps9p`Ygy$)0 z@0*Wh6rj;xF`#y~X0=RoYhW|SP4I!AxluBsVp;VFE3K1G$K3k~pSswCl5T@W)nzhX zIm@yGF3uz+-3ZPZG-|-hSM&y}Cg}r)5|F2V1i~-cwp%1%tIJHZ*S5VL^LiqLo|%NQ z71PgoP=-uQq~?)RmC1a%EPa--*2J!FD9 zJ+UvW*Ewh?d0bPh*9pD?Yv1Sv4yzi30A8o>Z7YhyC|GSZpCN(VklOGz2Qx> zTsw8Q7yNLJt;6dNoW{MM1MO4`D`rU5wi2)Nk?+XS)(AsVG(M@9Iguln;VgFF?19-i z8i;a&?xbRi_D}0_izX%uHdiJPzq6FfMmHVf`QU(b#{{8St_70c zp0ba%yQY{Km25b=^}t~GSZnUxbMgJ2%k}P(TGIY3JeN3h{>&uy)MXQy0&hoZ&198# z`b~Kl?<|}Dnz7~S?ugbtGE=4X%Y(}#OxY`cc`I16EnD zEI(qjPoNo!?jiX>mBFjs85qrBV}OT_1Ih%=x4lG{CfhgtK~(^_-9br(=%8Fm{gT45 z>{c$zIt+$IcBAw5&!9!7wOf0j+^(ag^G5aqfJRt%GkNDh6W8?vM3?qpg{-gq)ahgg zzo=LrCOITzMo~dIbC7$>?Ez%=@5R?|l5MWwA-IVFg9 zM-zVWowo9x)eV4sJUyB(wk4YKkPcK-gs#9IUTzNX5UG1&0j(U~U?#gf>&BkzHNz&u zDeKL4o(aG-{DXl@WACL7XhgS+9Rlg%71}}cWK;&?GYUl-nmX|#S~`QP>OfqJ!s314 z(T))_U@vrYfvdReXHGj^4Sr@y>UNKRG=Wj`nauzIauRsdlUO9UZ9AcbNtVu54(azy zYP}vg%TX#yGa-KE9zm7UZ$bu-Cb7|_JUpv_toxJRZI`cb5KMR_wS?$#QCmJH(X3lGxx0W0kJ-i zDCx$t)>T3X4}%KbJ+jP0(=&W9B+yPxIkexa`R_zcRNkuMMR4uyB3q`7>_^6q*xOH% zG7%p!2|Zv7YH&7;b;u}CV&w&|P}0A9`A$_G{4U0}bk)_Wt?woViXD_2j^&_(+_~gzpNZwt3z`=EyRWm>&H$m;9h~5`Y<7iL3skSf+yBPPEpT4YHl7 zVC2AU5(6jg)9TkRc&r;Mc+1anilieZ07Ngt+K{b{^MHw~`T)_s{mHJP4z4Yp3jT;M zzf82>6O$iANz)p3#g&@Iishv0@pc0Z$eBe(+poKn5iJ-aM}2@hOF&9Nl_c9qhC%H+ zm7{xI#Q_%)5gqlAuI6JJJ0aRJI}L6TSJBt?yu~wrOmz>TOlsdav1dL>ZE17&oiC}S zz;j4AQX-<(gYE_ua+D|q##fL&V(hQnYTy4Pw-QtM@nIV^UeR<><$%n(Ft5)O@K7h3 zH>-WXT>JO;S9cNI{78(V5kx^u($BBl2>{mK79=1{Dv|n#KHS7aPuB&u=4MWJ40$He zznR$sxag6g&2u+LxI6jZF*_veKUe z#1KM5UUTf279OP&l*G17*hHMM&fm$~4Zp#ZYBzuBoP?RKBC1MnSX+4m7G!1(qR@}R zwk&XV0F<3)d2dK(u&SY7xhTZj3qrhFy@`Df{mPx;pYM)50Cq3|>;T{dNvGUzn1Fc) zDpJNcj)*r4yVcv!0qkcOjC_W9V!H!llr-nr3T?XR$hylQ9W^U3ddO9l^iJ^6MN zIw07bR8h0Je#_0LX9AdO{|lo~y>wB>nFB6&t4H5|Gn0sZvvRX@MHEYqg?^{`C$X8N z$BP&l)vA*>of_L@Jb zyQ&FZlx{7#1KVTCuKE#{Fpo$hKOu=pmg4Qk%v54{{YP|O8(U7gfDCm1fe=9-_lTyk zWgUDFy71*&FYJvr=d0c-3RPpqEz< zh@+Td%wPDUbC_OCnx!WS;8s+(N>VJeobV$GA+iG@ zBm*+}4C!wbNz4~Q1RC~-sG!!+A&L)0v~`rZU0EWv3+Ng!ClI(pET@-QfZheZK&D{d zC~yW1fU#<6BV<7L8#+Lq;R@J&JsX}fmCxG?*xwJP`+*~$XzAi>bz#{H`VzV1+L>C* zrLI>k9LyN|P3M8mxVMU&$$yV<5nrj-cS_+zxLpN# z7rb#x?-_n*@}x=D^W=<;Pu9O&6&<-uY^@8TfE!(QT)SjL&-F#Yg1!%@8;wsGCFO<3-4*;mWZ#z6zQHh(2Mp`9T#k zWXbtaw(mud7k8C``F+^9dHYt!L1lug*cZ@gqg;YHA3lHq;21MM&R6-BE-=5)Ze>A+ z*qbLsAAwsvD7zDokw|QEoX(7Vd3o!q6V$q0T%Ij-;4Q~`vCXIHq}Wv_(2s~do401LRnFgi>8_l-b0-Hg|_#y0C9rqbr5eA#i51 zouG#nUOMG_!QK`bu~CCsFRhZCgT`9g69+Ze9biLl-wE8ALEU1Z#wfrYB;;{2;j0j;5CJ0WCle)cT-H+X{rU9f$_|}jOJ;0O(s@XWi4CGoYh6ZxOJ(5D{t|wen z^?anIboLDS{a$gbXV#*=q#oHU|8{Rio2N?ewWD9GQV5?&M%t+Q1`EAl!>>VAyPKqY zqlM6VXB2mV*mRq(Z9bb&mBr*fgPYUQ!&9@^$*5WP=C_+b;Cm>6nu14xQolEnOAP%P z;sn3dcq9_uB&JL7Sp5c8ht@ne9H=3LR+z1F|0`}!^OAVKI4`5csU>gHO5vB4hfdYJ z4#|~&19p&Z(4bnsCGU_yMf~xSH)}?^!a$)YrpI}KU6|>b*`nW4_l*+cz3z#8c*}Jw zx=$<7nSm~kNgk(t+iStYGM@6|_?sJ9Ca675`$qtse|VaX1lDkL=Zw@%3K)OVdT_JuoEz3-NSqO|2I$_en)umi5Y#7A9NgB&08A<=RS|vJ z^7;;DXws_&GxxZu?!kXnz#qRjP$Gny2gin!LIjT^zQ#=C!fDn>4o?GgsO6ch0?gzw z`a$1g_hokj)jyO#Nr9*I8LC(gFJT&(Oz_5q0=eB=rSwBswYy1cbg3^u?^dhbend&%L)#WYi8HW}uS-P@*fLe{mhe;ScqHh>Bj>4-&?=IkUP;$ig z%_t3>3K4PMi3tsb92B>r8wO1z=R1kOFC5)oo#b2802l;#Rx#aeg_~#yw7C97P$(ti z>C&#Qev?;W4$+3FL_EG(c12*&SXUXew7$f2Z%3;rr^|!S(D|Vtr<()~DP2|xF+-=y z%PX_or!_=fCJlgK(tu~JA;rB$|JR7-SY964jP4C7nR8g9-i&kq?ThxS-zLo$g$1~L zIDO)yM$(~BHSeX6jbD_xh$|1A=gLXPFUdC_x;#@G7Y%qz5QMz9Im^M)ksD9vamaGT z*0Y>YI%{_2P*U)CV&8y?+)|sC>BlE)Uk6nrGpy=fVe2rvAU`bI+TcJ{j|p~$;)!FR z8lsN_T(YP42CsQ9NHeu(D{;cSR{(qqySNR!Kh0L66;rlvHT^Ma+$8Z5E_KVYp_k$x z`wbvJ_#;`hMF9KCptZ)0AMuONzW)Zse>hO$PkQ8|JQB0*8yH@Nhhj8%>5rdIptXM8 z&@T#r2^2p+Xw4JqM||_-$)Q1I8N9{0kNgTMpg6#Y)FO@XPeGJP)WDJXb*2`bpU`Xeo|Ia?cN9|te`7oci!!S+z=dNgH0$uj};HA29&|E-To`}GK@BFF!%B4cCu z_07TSei8Q3_WkJXK2K$5it2hY#OnW>=BZ~dgKB1saG+h}IcaJmjgtVZ1F-AX5a!>i z!_Kv>xp#W5pFjU@{b-eVmXnRn?wuF3dfz)&-ITcx+T*rijkh;yo!0?QziL6Tv%OPp zs=xu`yN*9Ci9h#5H$Fcq_OdkI-gy$8X5`gi>eAFyJB@zCq?)qZlERLCLS=j()8K(y z+s_K2eD@u3@un^yY$TIqF?R`aa%}6r2az5%NsO$fe1{X2;7LcJw z>g>BH-?a^K%vg|!cnxaRy(aXv+JfsRkxdb_2}O5mEj!BIOkTt})Rb!QzQ=hr00Pc_ z)2S7Mm>}T505Hr?i;D$ZoihKzu%>4=D}92~SchM{HUReskcX{)ph;t5H7|s-uvTM0 zN?U{vib65yp05{`NoL`w%=DY0e1~i6_<$3I8!yRT(+QqN${l#)CFK`oF%%^TTdkm`?Ymy44I! z2k3NIs3w4Bsd~3j(sq8m=(85VF5P%Qas!?s3Cr?eQjT8LnCu$#jGZ7&7(lWM-OgI7 zCnRQ8)QquGg3&*_Z>5wEdhA&lK5l5+-OoqPy$$zLTf=cS5$tln7S4qt!G-YL?-;&2 zr7X<**=L(#-flPTkB8GPAF!tBS_}tF%(aI;BGGXT!Fs7>Q!)=}QycR}K>1Zd2vugR zE?+;Rgs2x1WccqZ~zyFAh=R#(1mGyc4-x~9??-yV58 z!Ygg!v2NbZ^IqyNzwe$s`SdPqY9s9`zOigd`jZe^?rch8{ssiOcWN#q5Ptu#u?Q>* zR5fyL{s(-V1&cA#038%upqpk~H20i?f$|~8`yXd}t+#!!DsowWa<6yFZ$yuBT*uhS z6~92`g@MxQD4xx5E$aTC@vXNwHLk6HMm|#7-cdXfX{LGn_{AfCIDgb3$Nf&;)1gE> zvGhv*_h&khc{6ab^N@*Y%LFpvF|r)MQm6=((D3Yj=i@89`C}a&pB;7EHuNOooJnWB zRjZC~^gr$_{r2k#5bB*He;#v2`#yLfVY1b`N-%R$FM9U}Q7q0(Ik2MED-o~4L^9>a zaZX?)?sR+6vRNgMa+({G;J)Ko!To8emqls;-PR3yZ$NkI8Ww`B8PiuXf*_`a?E^T}D4$EMlx`IekJOr8yRH*)vRpe{)9K#Pf00;++Qh&~$XhZc!*rs8;& z;juSd0!0YrrdPS-DAc?xl+lRFAENOJS_4Gc@Ti8zZ1a|@StUyqhLVRA%7HwvXej1P zJFzvZ{n(rFOYqYnwarMq$ur+)U%xne^pWz}jp?0#TwN6GTOgRFDbBv`^-3(;7+tyo7;YUO>UuD2@i|q z``!Cv%z0bku{!VGAv90NSLF;$fOlQGx`|$1hK2D98(>;JXdRhXl6&rwjSSkYiptZk zxl|CG`!KIaviR$X3nboPX|!rU#0=G(KK6&s#mlAd^L|`^^xb1L=#Ag_i3^ymN;7>s z^?%7`R=DdE+92YEX$L4{d~sV|<)nt}&IukTwse5{V-8B*xzWW%-O=2x=!WdM+Qd}v zb<20|9Li+WU$j)cgx-btAmCKMp-0yVLY-!;cyfim@8n@S3OiM*oTT*5UAh_|gbq|< zraDLdx&v6nG8s5WPW2qocWlhE?Q~CQJEh;-+9gM9`it|5VNU=GU)+rYb^h#pUwgsR z+|0sLuS?L`(ylXD@~YsPnXiq6EHq8JnmDJCBq6^oplXkMVZy~LD#e~r{J{JTd*e>5 z{2VMV?|rXIS^DbEO>sbzqsF|)Z!fPpzO(+pi=F)hiV3YJb$r(+HK4~k`&RD+hTv2C z3v?Ctb(40nShM=`dMrA3pvN!2JT|6AvCxio@+av7hcn5 zAba>jYHWH>^WUj>P7l*gqOPDqoGcJ8D*Tn5^=I~j>e;XNWXLt&sq^+UEP=5pZrk|aAsbs%6fPMW zIb40{Lb@S}=-nAcTd`SOLZW7v02+jTwS;%@QbD8Eu6WIlkyfH!w{aolM%x9cf$)fS zP=;H{MU7cUVEky?eOP1-hSH#?D3o9O?~fHZCSnL4Ge!%%RJ^JEj1krex+-|fe2ON} z7VpR23#xBvzo*ea-|ojSllOW}5=AFeQO+xbxa=4JCkE$Y2C?y_Tpoz&Df|Oo$ocv5 zCBRWvvH$ukQKp-$Xm?QY;K3u%8b2OZm^p)~>k$!AI{>qU6Sez@o@se(&kmn4+DjkV zcZXNVNl0dxEkC;ISG^_E56-72ntPJ;h0u2RnTg;g^mlUs@1LQ2Db>XD{9*Bf?t~T= zrXhkfVKijDkAto20p0;}Stv7u5boz=M5EGJ=5RS!{V{={J&sByV(PKt&*91x!$q}N{YrSd|^DQsXz_@Eanyf=7D3+Q)SxNC%|`Haz! znfyX;9IB0&-}Lu@wL5WO`#j}kL>|*t%)cY@@fc{T&TrIZx!oL!s2oiAZ}#9w}R_K#SeF+ zm_NL4u{g`5X5@(DBmTa&t*VNJB)5#`PxG8n`EI7R^}ya5vHD9l!KH#gTN#Q(baZoA z|7&zsW7bKW!Q?Y)fJ$4;fj@RrbLFojRs_Zfgiz^~H_r=$*KVCig;Yd|Z5%QR$Pt{t z{p^hT!B<^t)z)1oP8U2!0kXO;Js7h1B9mMQm&F zN$_L~v0@Iit*wXyM6LrkX*gVPYc}&ydCN2v%Ya!~&)Am(+UXd6KIR=<+_w4T+kC!& zR_ZG5p!*^yyM;#=xSQbugR;BkXxui%p0_W8z_)Hv-AY|UOk{=G@?1jAG!@4PM7w8Z z$B9atj-pOE-C1mbrGNGZsa8AfcnkCB2ceU1H(08an3#xVbw1C?@yKw>-;zDM^Q3x% zKScH!m3F?%ddnm)DK&@3oly@&&o6xk)NY_dfo=)nFeYG}o;a!>UMwAP#t2kuUc>KT zP~_=(TY?aUP><=2;EfwW3!GyzqDv`QzIwWA;8dt+xyxRK)Zs|E$cl-5ahE|FHj8=Z zDjq~*J0c+F8aujbfQ_97G&cWKMQ7q{F@`3DA|K!Cos?_g3n5Vly)08f>}=f3YSDD% zNv&46(?GiUp%!3ciNvZoY{BxAZ(o4R=SIk~(_k$)9*DYgtoQJ+AIq>)fdVbk#SoBu z%#Ea{bt_>TyG4)xk;THnvrZ2{C3`B6T5xb+R?s$z38#-ZJf1{>+z$w7Bzojct>%M^B00g8ePS0adln~@Tl)x-cd*JK&!!c(`OXC1*_rvG)c6U4{;o+?6S~;eFF$+*#lh)b7dW!H~(3%1^Y5+1tKBG zX+KB~pbLCnAxXVJRa^wB2*z9HEv7edQcpKmTIcBirvy^57QKX+IjK` zds*ENx#6yt&m1+V>(L4KP&-F#Pq;R%VIkNQ#qKe;>0B54zLIOZ|Y(W(PDGUda(~S(o4|xq-OA3i`tM()3UvC)o=XQq=Z3Ek(wXZ%2w6!ZkgTy z54`jTedUU4o-VlzTANlyTS%MK2wYQw*L;e(djHBz3;VeTfYS_IdN&ZPQR=B(!DWKA zf-NM|fXxt(*FSIesOsalNJ2(RusZn8OyAB7Lhm#sFJ5z&np>}6E8gli^(ZGL;>E%B z_ED#S;6Hx=$|lPvx~AJim^ z-Rl3-`0kNU`=Rw~wh(OKgDw`<9}iTnT#Nl~IS1HBlUx1k{T1@<%oe3esPrTG;3HW0 G<$nPGl5^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!*^Tn@XI9`Vzt2mTQ;`N&N^9abY#gdrX$6X z^&jR{*#B$W%m1H&O|brBD1YY^^9~y=2`0Py>g@k^c*crP6#w(d|M!ysHi?h-TR0DV z3~TfM+oJTJ;n9jU>$?iw_VayJQg$eJJt6<)YQ_RyexdpkM<2%j`p+==KZA>BtZt?H zTjt8r()d4V!03Tf8_VxD+SFU0NuV0>QnwI|O$rMT>hVuEn8v zp-{N#mY#j&-_LjUx!?Kk4bM!%%F4_fbBsCW81H!3y7_do1h^|NEhh~?K>+|zkbi)i zS-=y(o!hrDZlmA9z`(%7yn}^9jEjSfjYCdIgilOOK}$nTK}B_sfs^?j9XmZ06^j5X zI~O-EFE1^#keDEkDCa|7o*$i{U}9q8VB?VC;*#;)r@GJcZ$EC@00ehX_HLb`qR;_u z5ul(FpxkrMC+ z6cT>&R7zS#R!&}BLlX?q($+CCH8Z!cw6b<`c5!uc_we+86Yw@LC>RzU6B`$w@Gdba zD?2AQFTbF$sIscMrnauWp|QQAv#Y!3V{hNs_{1b)>eKYh^2+Mk=k<-vt?k33bxLzQEhL*@U74T)C+LM=JzMl}k%&Tv~zm_5QnG;9&ig^AT{rx+? z{paTvRc?b-S-=VE@&Y zS+I>r@)Ve1mK_qDb1?0OYXpX{AdYaCw7i<9;~96{+J#~%VAv*0H+k#!A=#WV&~a9?swryq zt^0iJlJ9iMBI&Gwp1bWjwV0n0WR0o0V|hQUYz7)5Hh#G7eSwh{*Cm6YOeCF0r!EH% z5aJyZi%~5PRI7+bxXe?9RKwP*X#-~-`MZwK%(8Kx!0Roj#-`9k6?N>LmziSbNmi2U zU9(^ylz@L*Yv4!ki~4hQ$e*j|lCP=U?g9>+LpsaK$m1nYyM+LnVisKj@!vwGwb*Uw zsP0anObuRV-T>zLx;A}|1pBX-uWRaZ4Xh#HP`gzroBVwQQ{WANKJ@)Q?T!R>%fCIC zzh-iOG}Akpx?fntRV4tbtN$dJvZ9`L~WoiGd4ssp9dNWp0(oh~TKn!D9ediIbKx4|(7B6gVAH{1DAN$-w zb@Wuo!h^w>bwjM!M6-Glcend?WDhecMlL}0V86zm4_Zp{S%RyliMh<@pyU#oDo;zM za0W7d(7Sk8`W?H2St%NNv1U;pg;l#>qmEo%QSF9b@lP}GVif8p^7R&V?}U4SQ;Ahj zr2hZeP?>H3DmQ@rq?Hr#3o5f4fc*_XXYcoe%g$Z_5#5f3*Bj>}7+zBMgvfISK8W?v1J#?90PzS~GL%$-k7p6RdgXlOtLt(y%)>zaY4}oxE4oG)Y-d%^!Z`FH6`n> zpB9KW!~^t3vm)>I-Px43vbsyGDh2p=8_%!KpH#L6W=P&Q)F2|>-BkEbd#4eAjJWzg zcL0BM2LI#tK;k;C_J<4w@4s*X`*vsqYAb!%$a13^&H32H<%1_efwRIa#?a@C{YnDD z4NT#{f$$=+L8ou|A$#mk*d>Nmt1bA3pI`(vOI*LGWSf3Jug28|K8smu5mAmAX6ZJ+ z>`a06r#YPiqZUu~I4RGM)?X%U4-j~jxyRsIezC#4CMQO8CoUyx<>(f#5zc>g&{umJ z1qVvdw5MxQfIO@Rw%66`7=gXC9M^}y#|5e*I9NI?r4d|zWzRHS+@}3Lt_KckbMMxR zLb?$+;8JwON8EP#ewco`wHZnDOC1+gvmOmo3w0gZ3)PGIsE}}Z3 zpZ3+gBjr}-8l(XH=^L-*lzld*s^+tNS+wqj$=4oz)vsV@a3+1hr+=sdTX+= z;GT|-QW>;lCL}~=A+W`UT}*3SQ_m2z4J!WhUWyn@kHwNr=ZnP_a8_ZF-#sY$4t1=7 z@;#Zswvav{W1?KQ*7K@VtsbW;{MV(1TE(1Sp1Qi}1hiz4#7mKVh0AaBs4j>|=7AC= zdj?X>i*?T*Uuro#DrGOP`)c9@+j+q&!tsbtYvr-bA&{Iebz9knV5--z9B@ zBJLHiXw&VXoVO{tk+fFINtq#T!ZHn^X1@5#7vayOEey)`5mCFfV*08o#u_M_=)y3N z@9Lk2l7Ai&lM8%9Ztu=hq{tB9#Q)UABn37zLkY;x8A7!}y|l4ms0vQ8$KwfhKy@3rv%z@zZ%a z!5rzY+0rKS^W6DNqy)a+-u-%v?us4`V=PKIq4=9D}G9{hTjL1v=spUu_53 z${34tvS1sfnv(_FVxPt;e@GtL{&!IYYqWet_QSik{4D1%M)mXz395TtqVfwVDyOJQ0AtHO0T3V$GbmH4a~W|RiB$w z&{fp!gfU*aUyAr(mgK6=&O^l5JYGa^z%NMC*4Oh_b)lxl&$EssyC*~utP%Q$%-9SF ztc9ARNS1(3a4nDo)8g1IY0rZ>rOtZgKrEG|rF*ux!cMvBXWm)!WGhQ?3=0tG28(g_{t*E;%z^&{aTmaOK=zt_nBEatnd_>URN%x_0J zH{r#r9MzHG|iA_y;` zX48ar19SUHo0t@HGZ9u4lX+Dr@zWE&4j%@sAZb@T2X8Z%!}w)uNOAAS*mqcjLq`g_ z$8zPHGo|%Lb-6-p^pUqxgQPY(mmaFXIwOD7P4U;_vF(aFeg&d%9a!Y8a&x6j%{ zWBd)}v`G=7%*2*h+@UYOT~QSXvL!c^Ms;$ z_p$s9za@F)_T+&})0;$mmD}Q_uJJmGxqfIc?$_H7A-<>?aG+H0 z!ixrT-YuCJY;2Df!z~)*xoBjtC_U)OmAs!;L}mR zWQy{{ENuXM>odl80{&W3#L?Rxn#{sg8-+}Z@2XX$JXZd3kNJP&|KImf)Ldo& zchsp7h^A>|NIWt{{&Q~~!FW+zBXN@Z9{?k4Ip4x+n*D?rRXSJEVAY-u2KcMrnLFzq z30akTe}J~Q?m1%UoH))AbGzs<^HGAfpMLIxd$pGv-MQz#r4V{r;)5&YLJ~nZ@5czR zVCI9%s>p2$!6_owl3{KYvU<;98=nu~Mgh;Op`miaWJRtEk4E=vG|R(RdAQriO5^eG zPEOD|IMJ76MUe^LHVDk+a3XeU<(jnRO(n4^O?Xfq)#b6#flmdc*sl?*IWdwhjCB1D zoqAyGw=zAp#QIoh(7~?A-KRq>%JQzBKp#Njp+Ov#D@iz^bJ61?uSZ>D?T9koLQv|g zY*2$aYWRA&ix26RrzrbZ!jqHr=8o=G)~T^iiVSohun}-1CFTNbo4ib|(B+BJ;uibS zkkXCnc|(EG_LN^y%O?S7)tHtZKds)g{YwrXnm{Ni>{OXpL@Ec(Qm$M1cFhZKoM^q= z`IG2G8O;UxYfb!)_jmf2o0#P~>>e~FeSzXA5ylA*aVPNT%X*~5de|SX8AdE<)K=6~ zsT4$&wE~Jw$X0XR7rXGOwmpI0o53HOWW{!>4@hAcLA>rE<;QWim&BkW#c|-=fN;hu z_G4-T5A7DEc9>~&&J)fyEcz%$iI0xB9thX^o>3`c-+%PH|LOlA(oTxfg#`@>Y6aO%H;O=_w{%Xl3|hYKeO#)N7+^JX1G#MZDQi-p*eGbH7{! zYyhhe?^Jy82y=F11mj8e&W1WnmRXW|tdN6jPg{rfDxBvAQ14T!G8V$z$)QNP??Ht> zlvP{b7}(8oakuLFOWmpjO@fr!mK&)$-1ffZkQXBNFQ*r0_U% ze0M^T35dZk#aG}yGavFKk7w?gAn1#B`MD1Sv7x-_crbY0V$Q`W0u7v{7DCRM$YUk!VTabL?pU3W%zM(VR}dqw_N z4F`2KZmgU%U-L$xTwC2uk70aT0>aBD1#@T{c%_Cnwgp&>*n+E5@|1 zorRuLN<}4QDJ~y|bG@DtT!0T{Qg$WYN0te2XZq)oz0`r&SnE-x?#oFLL*lhmlLn{k z7|KUC8s75k``?%GJQBzCmYC$j2PVtBJ|lN6AD%CJf|;P%IhLwrL&(TSYVPGi43i_L zC$8Cu|Fb;!fBG3Pa76)o@bO3oR4AzY16>_U1|Cs|0mYLO;$$4We4ORG7gQ;PiO z?J}>b2&V$zHG*2ejLOt7OyC}lO^Nwx&^}C1-8VlH+Q&Ogblw4(NV@aIR6{I?FUOc% zvOiJZS)}|NYqI9mu2nL&myKf>^RDnq|s9Ebrh~5Aonhtiw}Kn|*#mmpR+&N$L@lI0|s1&-5uD^N_)AgfSb;c46Nb5EZ{ zoJLO-(x9xWg?L3Y#%^b@kNs3myN-mWrb)^QtIE%svlsRwkRSL3nEcC```;taKM~z! z-jLXkZpUJm>1)3E-%1A`w)mq4SB12zQqxx@s1J~MZVswt+y5EV1ss98UvawYY@dHG znp7Y$ZX+)<&yypR=h`e0>xR}n=pIJb2S+@%pfNKUye6=!Yjg|jbh}ru;BM@1knk*= zA;_4R<&3N-DXr8dS2q!^dq-^BLGyEHr+k)9?=FE}QP(!;q5Ks4X;A-6y%C-;BGpNW z(EoFE728)N9z>7G`4wztV};YbR*>E1-nOJr+zhsgmXE?|D*CSI*q9#9UTSg4u0TKF z3DkpiN7Z~c3$2u+a!oyQO#yROwqzHJQE1!9(0C3R#p`*HL*QBSbIZ?*8Qi)GA<#{&OYG^;`qAE$n1|*Q6@r%hu8U{Bh3ha4!LK6&ecF*hyIoJB!-w*-$ zaBG|IkzFiid&+uCT=b-U?jtj+#7imHCYq9HhIEI?*0-q!`UlAn1fPOe-o$GZ+1WU{ z7U_EKdPqRYX756mv|it5g)sTC$@W1}%g&|SMj+a*N3?Z^2GQk&B`A!tqUO!xDP*%!9h2qCln!3D%-01jg&(Y|BATp?v)^PZ6UpbUeRE}0NUX1Z?oQjreGYcA zw>P^N>Oa%uktsKl6B5VA2>cM#rP--*Rzp2)%?hF`g{Rt>rPFaW2=Q6NK$vp9^Pxy1 zm-wGL-oJX@Z&3GxA4Qt(A4MAY*stg+=;`lAGVoUZ4rllmS#7;fP$PTNLgd-m(O&E- z>MKyK_wEaq4_)r0ozMqF$}8`}8WQQ-mZ^M^aY*!tivw$6K@B|5g*wa45Z9a4D?_Kg z(VQ`zo?X$C!_?h|(`1w*{JY|OsH`hmikqortL8Vf z>3N$>cGm%O?sgkn-HKLEd8N67XzYBD&p0u3TZW!*UAPpUUTP2$hPNQ3$G;rVWn-0g zccQFsxKNIxil%~|I;(2H;gj&_8++$n#74dhyYqM&r+0CkVg1B}w-bJvP4$6+?X&?~ zD1}nf%I(FX2&RYQea{8qjUAGe6+tY))PaYU$2!)T(#Yalao?F*e3f|wwX|O?3CYU^ zWFZn&_8{(UY&xgebEm{dCf^W-os`gy<>_%H{j$2SW*@da)-nqDT>U4gpQg-7p!jMJ z6-)-pbaGwi<@D1O4vH3agZF)^1#CWcYo*Jd_Q62z#=~)hc|MKYYKZe)LqxG~ZOwoQ zji%8LGz969Sx`|8uY2AE*8n5#xv*gt$EQz!iwlyC76>H!psI)8MT@y3yxadMVVzsE zd-6F->0;mvO?I}GwAd5($kn$(Wd^xe3g{GlrJrXjlI)TY!u?Pd4&tEaPcbfw(6zE) zN4hriNjR3-PVr${m!adnSX=97&%uA?Gw>NaGYGLg{zAO5-!R9TfA446mX^E zl06Ip6k)EXJYU=Ot&?&{!srCyx~W11`#?A+eAB3r_@z)oeDIqF>tb%EwDu?@C|5zI zDn$(V@3x0Sp|$g?_IpX}DF0N?n6mioP@2_D`n0zV!>cE)>g2b4#b`2`YjvB<0|If_ zG;I1P$vzGWC)tQ(h}J2d>&3rf6pnJHAK(au;b?{?FtcU>|@ zImpG)fF-)oy+|0u}CVTvPY_<nClPte7Dpu6Xv)MK=^?f?yNd5+tVyoR3{!61cd@yL z$A#p2)&2>tT^GzO>1CtJC}{-WtN3DqP$fp4G6bs*yFJneO(D`UU`QU#=k%xecmY8O+9kzC7Xm-#D%UP1+3WMpKU@K`fK9-lO~asyp_A3m87jZxn%6f z!Wn0>GYVOXoMQA!H*e9rZ8ll4fHh0z?L^INmGU}7Vw!Euv}9HB;p1=3q`cnG!0Tvn zGpjGsvR)V~5lVQ@OfngjxI7PDx5FzP9IJ?A8S+d6`PH81I*5n7O|n_TUI*?VT5Ltg z=<1FLxM{dy>qN_EVKmi^?`jQ&bX~VM+gI%6H5qj%_tschtxSq)6NYNTA9L4R06F;X zO1~yH!i(KLqAttK0K3-Y!3*j|=)FTBr?Jar2>cp;$9u2obuZ0S=jf$Em15>cD`GWwbLERQc}mRP0>57V(&qmh1pHzQc&sc zf`leRoh@u>QFI#DVU_!5A|j>oh(QGf9vs&>(A!B-Ig0|lcO`p_mlA^7SY}Qa8i($} z8<372LP6%T_%(=j{${+VGl)vnkJyNWoR2%z{QIH{wO&0t&)1fVSF{I8kH$NahKe$* z)iUeEfbN1nfTf}IJn3Q_i_h1GgH|o`)tm)m-8l!hm)MM+;P~A+E7H(#@k*II#+rn% zxki{}0Wxp^=p4l0K+KR8uci<&8abP)b%4X`w8_t}P!7#od{(4un+ZL1XQSV^l|kbf z60ro)XmxU}vbCq>P2V!&C^=O_@&md@a|2UZ+xK=a30P{!8$2*sFi-N4(-nX$di=+w z$3K=K|I&M4ddaq2uGNuV1ua|j!#x9^WpNm@;Z#z)Mi{XL8pI3--SZ?Dj6EN|*1)hk ztdbR_^ciEqJifl~_Z{$eke?u{uXvSJQgjz-a2~&~UYgI^Cj>(FnU2X`1l3~HTI#&s zOr)uiNfd)<#vt&!Q0-1YSebwL) zuqUtMOqJfL*r%_aFT2G%?MS!5q|b3hL1?eKQ!}C0Y={$QEsku)aIzTxh`vS}fj01n z@qV7i;`_uw7qI8cme9w7bQ-85&l|T_7s2 z?u=V8h7oU}y68hJ*QS+qt&lRxAvkqK+uMw?+w|_m;ys8UJ#f!}M&aFwCCTQi6a3_) z=Ck-CLAL<%z{gVZ099Wu(0KRy10s3i2DQ(0T%T=1kiP1j)}O`_if$Te1NP{WxZhhk z;=+!k_(NG$&sQbi!uelIo;=Q`?ZH5F+~3Gd62^Bu7N3aa*f`z#~T=0lxp-ox(F z{9&vh9Xiu)lpTZI2X9V<%C6^`(uwr;=hq2_taTUhwuD0hn<~S>>XEvL)`&|8`mo|9 zHC@TobOb}q#PESZu~V@nx87W*>TChumwnWGb^v`cJr!mbSyAt)4Cez zzrjy_w^X)m67Fr12s;*ZZP|@lXw<>i4lfQ=4d_~=2-RVP&OQqqrqwu+$z~rHGpEGb zQ~_v19gBy7Qhr5@fEzRE;hdR=z@bDN)1zwU==>W%nFRI8Z#<*W%w~O_w6{F?JGzaE zr<40>rVVwLT7}ESN#k~VC0E+)2}JDB@}tEt(%S4$z0@qY_1rnWe8Z&F&IfY(bJ;KY zlybNM1q&bOB_nbjJJosfT4ay~akfxtdF-c4E;9}Wz%ol3riAW0BYZuPj&7)JYmjDj9Paoz4GTmI z&h1B9Lh`CGH}FJo$GgHDDR$u(nB0u*04GQRm8u?30&hwPQZv87PX7WO_zgY45eIg; zaR%GhdNq#)QhKg8r!T`(SCW11_RYMjKXN^6K3r%SC@}UD%wr(vb}T|EQ^tMQNaiu} zSvk4vIU{y;PuV>AkK*Dn_V0G^zxB!`s=8t+D)9U(A$%8Sl~29>6<_t?Q(Ghb&{Sz_ zL(+3?Q3ay|>ZZOWBJFF;0|pwPoGg}sQVyi>&Qfd6T>6(gUmKdk5wI8G^pATo%J?R) z*{7{q8HDBh)Rx=hsX+D+=Zxq~p!&=vV1=VE7L=6I(XCZ?eImM?S^qz5L1ljK(wm9+DyK_sPA zlfjjKE^ifrCWoI->f8W2!Tj9&JUM}t#g5Pfq3v_nr-!Zq6V!Ec6?3^1gtz3EBWqere2E#V=YtCc#gz`za)n-snZz$odL(*M4AmX%XP3DLOzwNXPhix7WSkqehb+&FUL@{#rx}U3kwI8iZUc5%hUcLd~NoBQZALu(OfK3+(PwHLb z#@`}D{g&&|AU13sQNlhRvSl<4=R%@sjF;YOowd7NiTG-AWl&JIUjk2Oe}W=DXNyh^ zv+Aeup8I3rQ zvnQnWL$G&VB#WLie1OJKmn3KF!J9-y8K)T*xFPtkJ2m^_`RW($+0Y}9)3b?;$rp(2 zd4K8oJ3?@e<)U^h$FP`fk^IbF_sedZWiPcIN^UeqwjN(m!y5pmZR}SAU8bqo<5zsS zD-W=&uIU5&Tiys3iV7`Pwo_^uEob*a0rm^=9D?fg&r?SDCV zMfv%A_wV5(!RyWjSwrQMt3qvwCv-ny^Zrg`%l0Uq{!}lw5hk~*ht=qDyIEny=S|HX zL??+isO06vTJGq^ae0vhIh0a!M7O;xlKu`i!+-G(yJr>3i>qNV`5$Ylq+Pc2%5xE z+hl}Dke8=%KP^5|q>@)?4E0P}uQ4Q_o)Pln*b^xK2=}qw>Yp(f*vt+n);}5#8@K`B zc#DsVa(mlFqQw&V>4s9wOkQc40r^zgi=Xu%`jN9cV2SqbNf6EwNYYMYh3u&Xwc8yiZ*oyju*I(+)!zU6g|JYnDcU- zupKSu4iXKh$DXFG{?1wdV`=Rtk$uUzX|wdoimv`e?CztQjnEScR$v_qeao*DTOKWh z)(k&6cxj5|6_q81mT-Rbq7_Gz&??`QnyE!WY2kCW(f3s`kcDu+Dl#NH?*HG~q5e<(3ei%Z7t!Kbg_9qG;3|-y z^T(_DQ&*-BIC7$vh-@S61aGxWbzS4t9qN}Q@CMw`yM_K_Y4O(dxjZWxri6RwoCL z^z~!f179O~80F`7FJBh88(X|rtBl`w>hk1soE~qNA)cw)e4P5Z&y^SvM*D@XZbH4F zjJE;+K$0AE@_hzm z3dtXB?l!{#zqlCEH6B%w6%n1Q_WILJD)A@iTV>!Y&n-V;~6FvO)@uB{$ ztgJm1zzK1UYBurysX6eNt%klA-&HKKGnihF4jsvHmjbsohj&lH8Epz$6z#}rcO%4F zORA*ob;LW1an%j32egR>9@I#D(ccsTw$fum@EIwJ27~3q{E1bioI-ykv>T`wp2Ma( zgx3p{@uWDb`*KvP$EEdJH9+xC+ znLZFIQ2HoJ0-4>gLVL_#OQRqw^`NxF_-vOvcgOAZQB?VmNL*(=+`(#yO0`;ssciW>wNpeY-;`DyObCjJ$yn0?gOf4rdCd zs}m!0gm<=cfVhz54BztqVo~9b#S7G=L;TPZQPGcegcyPx;0bm^o_Mq+J9tm=<10_`iMGzPrP73$XpaVHpb6>vlcx;>;MywPZg)bw zxHV-WvtF+YnlPNkQq6%Ww*3=f8@J9g9gZrIzC6h*ruwJnAuc70xR3f~AKYJ#J=Wzu z$Ab1tQ-p2IFqlMYIiD3aKRCFizI8R9NFsw9C%6f+<_eQnbD;%o&22h;ST?M0$!S9J zym_b}whMga;hhOMfs<|Yzw)1HoTVmg|KQ_fL-GEQiyA5oYbVa84%d@@RO1so?W%JR zfbtrdyaggra;F8JiGAl%+>M&xoR~b)QY-10V!k{3o%3-+zFcB#@cieEVIi~JZMLUt zK6u`SrPc=b`Q4(PjgpwOG~#(es=}pQOzd5__+lAO7bT_Tp7nW z_|mmV;9#+o9J=>?DrD7kgGz*IxFS7egZ%H%#=p1@03~-;kvYuF^j*gE_;CjQ@J|{FmZC8pHp^ z_iv#E98)}fFtH&P3dJTmYt!hSy943Go=Kni3@JWX%jYqs`U1gIQf9e5EFZ6}+A{;TCO8(RR_;b#A~HsB7HGH}x;BNm;zMfv-biOXV*wV|dbOcT$_}Q)DGiL^ zZF+wsy0iQl5wlXPbH}3J>Ko6)ph$vkB$D&q_Z;a%f8p5yfA|mkcX|4KUK$hXiuumBm6Dg| zMHFItp#T|%6c{24RyYy6$ea3!(FmFA{&KDEZ`TqT{JiEMa>xV8^VhtrM`GH0J;8aYfCr(822VLRu{{4>!7v1=(ff~jd z_!+rbDbKS?f{Jx=wVf2_kr1VZ`R`NrPp_t)qXxRGWkGHRH1A<#8h7P}z>VeZ!u5@5 zL-A{U@@%`OY#^fcZ%t4g#+KDk-jJAXkZpOPlE(Tn$H;hn@2~sT<+E0jYyLQGEx*)2 zM9b}8E3C{Kud*ehyNa+NVp}cpH(2p|fRK zY6UKo)|;4rm!#~Nmzv*^r`=<7=i;uB19&s5DmEo`Vf9a(4nfw(9}uf}n@{OIks=lh z-v4TC;&18Qe_j*$Nt`z_*D^q1OSb*Sa zuL+448u*Icj!INj;0EC3Ox-&=f6KuBRh2D#BVtF>{H2ddRXmxU(en=42jYm-R5*rx zcqNOeQJ*J)5}a1K?vNul)}C*`v|uCRxi}xdO#C81|sRWi(=mc zC{xJSvF-Ybxiy%8(*#{hK>51G;g#6?L)%pIVyfh}d)`5c&c^m%2sG98rI^i}W}}}* zIfjyX!k(~0?}gUmdE%X}U8A)ex-7)%tQg_#3*4uA<&@j-gR4^;@)#`E5L=&LN@HNNB5T5-rX+0`84L`G1n zO#9vQ{m*^@+epPlc!%xzP(LC2VAADlpV6&O&0H zNBHH-P1NZE2E3Y{!bAQ?(4ZRtQcDc}o$xo7h&V0b&$9Fr5PUD26=*D09d|^>4?$+SR&>2ggXU!7 zkTE`TMy#IkWTlV*zCl=jjy&7JY{e>{5c}P?Z$77+sLdHCd>`@8hp6!w25~v0fJ^PUt*#lv50q@_yPbYanGX3O^VJpe=NrIO23~c!YP3B% zU1NP;xw!o5hBBcdYBwi+Kh6&|P4I&k7;8gW>m$UGHV1GcMz~1E8)39Kdi?C5XsY&_ zu`%dgedyg^`j()stgP>gl&Rrx07st3FHna$kpK9y@=qB6+XDsgyX82ZSTF6>r`?nE zl|Y$JG!ta>f3KsRVm2|^)=UO6L3LR7PrI_f74Pi#91jUP=n)hDc7n1vMhNI#!dk0Z zQNB1=?nesuEY4ewfivAw`5nOzStY2w9T|!n$hc04x0g$TxiXHK(70llGRo0f=KF*U=`br8V=Ve#wEm-g*s& z{8D^Rr@sdDeYK2V{T5O$#TP$kI0LWoU33qVtKbWJhnWT*Z&^uN%U>kb;O(a+(yIOp zH+f`YOZB76)*N%JoQgK;o#>wv_QZuW*p?$FNKYo7>_^8Lc)is6Pgv1tR#Nq(-q0}C zRLSbB@Ms{7XM5-mgDdeM0q!Ixi%LszYAEO(w{6U+h`nev)8E6DvyF>Xnod)j#?179=k=nvrd3p?kwmaEG_-J=jhKV2Do%C znT~^>`ZTtE2SquoNJ|@5GygIqwZ7rb+Klye9;YAQ0PJsJ0!&l>XsoJAX^hcySu3Ky zI5=0f?Hhp0>z-WUdMKqGkuJu(6ZzaPs|tMl(}X0bb^C%O%mHbMUzFh(DwT&~b0flf z-J(^$7~BulWP1FzUR{C7PTt1NM-5%Z8$X?7fwJ+6HoW;uwM@h>+m}d}kjhIb4oB2$ zLWzu*nSwOPk$xKWscz1@^oH%Fafzth{^f2cOenjoP+FBk#_(kO`>82HCwe1j!o$J4 z)w;QLX{_2Ghd(Cfj9zKx@9=oogNiN49Q?uidF4o~-8NM}DB)Jx@h)^?Wgnxeb@T6@}rlhnrdwJU86FG$_{BsUk5@`^7v!pdA2hfio{oV6I z>ua8IXO9K)Xc^~idOd`lf7{z3dtcnbHsF&E`yP)~8^|6O9Lw1uQux~lhGFK;na*kD zp!FS$b3LPL%~IrWmv{|N?1CYPD2H3^cT~q)JkbG6qnj)im0LD6 zeqlcbEkQ7l{qYsXk9*k|`Lr89KtD^^?j*oAauTFR zH-Fqu2H6><3^Ez;B|>(F>?s}wK+eD~lf{B^M+LKOb*7WU{3L)nMyW0D7( z%}?JFaB`m%eT^sSSY92Vmt>Fh8zQz4*8%L=*wQuE%B_5mAqVo0Zh{gIo<{QrV{tRE z*(*Aq{3}5*k&^?B3&SR8JIf#Ad83YiOVeL`-98J zoOH8IXpZP+?nUbz%Hz3#eZg=G!^jsHOcfk97L^CFM<21O500`kzNy?UJonGRMxOC* zUl8Z+>M^`%!4hsv*yER1Z$r{Icl8MNgn~DO{QgJstsR|f)YSEZHK|@^2jrcB9{rhG zaPLbnqefV7@ZpnNeRG+eRlwv;K<(wa-R~BM$gw0CPK88uthT3p$* z^%>Lt++BYm94_|~xAWIIIo;x%CU0jWQ_LF{@t+;IJEXR1Pru^{e^&~k#5P&>2!6_L z#!>$|?F3~`R~a~g&F((aMxc&VXF~*3)XT>6qQB{78jot=iwJ9Pw^3$) z7lIc(ET71(EKxJ;f+*zYnweND1JP@dLFM?2o|G4qx>JJKAxe)I0sgY28!XTK?$EUM zwvX`qfnLcB6+xON5nu#E2B}x%_!Ol6Pj=pr`UGi;sm?ZQ;b41vS0-qFnfRo0i*4Vx zl^w-uKOgQoT3*zM0Tm4Gd?>gvtVxlArzkXa2iDbacpYO?P?HEVcM)VSkiT|GwpgJI z`HrtyTMrY2?=6cUxBwUt@`kvxlHwh!ZtRxh)y(V~MB~Wosui$5@lk5l`=e9l>L6E( z11%ndQ#Ku;{4IMq_{zUrHor}=Rdj_`VxS5ODv|b~QhHWx@8o<{SV03HXTl1lBADl( zht3!CVhXr{?^ZnTEj2;deJ9@~lsY=}_0$AJF>mMVBeU*=f)6+XStQRZSWw9xO`y2$ z^fW4#_6B6f1qt@$NjbZcj#!hdSU%c<<9|4_S6ptIW|VR}PRHqBE-XeCOE;nwL?)lx z*P-F(DFS|Bf&sxck7iaBj~<6EQ2dP6V+XAfKHdSEQ0R@{RW zgD>|FXoZFoDFLBB$0vdywHuFe3x zAW}6iSpaykDC@2>c0n&nSH}p)U9!AejYU_Q28k5)A1ErA?G0MmkPiv{{k255TV1JJ z9J(B4QWt|>O|swJ+G4=rBD(>6E3V36%UOc~)|{2`6{$m#b67#QFGQTQ45!oX)diev zmbyudkt`^SLC0D7LpP{KZrcXth4FXFgg!K?LsW16;a({+Dx1i36sE(18ZjY?4`t`M z;NhvNc)?QOY>4qPM!u;!@m;VkJ>RUV1dWwL?)mGCbOl}I6YWsk(Ysw*^X_(^Qt#?_m{3lc}fEn)Q&Wev$4&@a* z-?^NWe8`8+SS04^=y=4l^J6eql*yqICrb)-0iL@XN>MPyP*)5s$qAMA{ee{y85&A~ z9CyGZugw$1L$q(*3AT*!W}h@pdc;?iM|nZBsY*7wsv&*#@$}@9U$L31O86hWneN|R_pAGPUa#l*{a(*CbLNkk>wC_;Kj&QMI*dWaB(Tp&&rlBl zfdBx+d;yFRKnLL1vj@C~odXO8b8>QU@j!TaxVd=_2ng~)#6)1?Vj`lV64LUr5|T%x zL`C5zEB+AI)I-8hy%J= zL6QIqKZund#ApIw%>T&_`f~w)Ef5PU8~YwG2PYRdb3@rafCa?L%EHFV&d$ch+#SLE z9bn^U7dWJ-wMP(p5iIEoQHo5==8)1Zu7{d;Q>2wI`Q7B?+Ak!0Kt$%SEL;wuqI&#< z+DUaCU5uW-fuWIwrIoeKIa|BSSDb%yadmU|54aW>6dV#7bu0RI%$-<#Qu4i&`wvpn z(up}wa-ZhqKPz}uQd;)9yrQzIp|PpCrS)A~drxoQ$NquAq2Z}%^312tvvc!bR#w;6 zzin(%x3<4?fdJNjVKG1d1@;dvekKdVZj%D1tpat zIUw4J*~RsoQp)BO=q0~yuKm&~Q!*>x(f%a+zk%KSkH~(4{mnHD@UViI$z$aQ(7?u+ zag&H90iplQJuvX9#AuwaQuf6!N+AVKl#E~-`fUv)17IOD0FgCP9HNkX^q5i<0}$9! zCv3Zx&GxT7$6QlcX^@N_L>5p^i_$A$(SDXBcw5fn-0OyZgWCt`f&?xG(AAoYo3zJ~ z^=Y!=PzI1qFk%386%3%Inx)gL`H8ZOwFN#`>y|OxGV|fgRau zT!ZhF)3)$BiN;N!xiWxTf;p#z}Xs&o4efemLiW{$@$flBj3+pAwT4T zosWoGh}$v^{#DTKUmcOaf5$?lzMW0SO-K?7YwN!?)AIh^#OU`=T-ctbTa5pb2$f&b z>rqep5#@)*|L^I9ue^;;y@A5yCYb<2Zk-E`yg$@|o;SlQUOvR1kMaY)LHgy&J@%@V z4qIzJ+&SYk0eb{Z#@Ljnja>;LHNR|a9`@0>P49RBp7>z5W*D&0wys)w#md=?3%oOn zO2R2SK3Z|2>uV^}>sN7e%|*wHNrwwgi91T3wd{R|0a*gafEEooS7PN|Ga?=i{AoE! z!Py7yTWZmJL1s9&K^K@;dFO5bWN~oueSn2jy4>x{iK4hyb-nTzJ?@a@ZTtA+k z)wL(g-O05-tL&!pU~Y9?UUDSu6~@Wn$@w7}CBcJmuCz)1jPzS>bq6fX)xs(u+XvW>1obCn=wIW8ou2G15lg#- zSC7Zbalnaq;JYD;au|TP03G(hs^ED(18BYVHf9ZKMosjgCa#M?u;(97&BB-mea62R zy$F+~$J>!TUQtO5KVEl)@n;oGKBBbzyw;7S0@{707Xx*KcK#UM z;YnXECEZel3V2k#*yJ7O*1i~j9UV-+R3p;TGH|x!ZcqET+EKEw0ofivGzV$NONOdY zC^2ar=cwEbp+4gJF^k-M+WNgc6sHv$rQqOAez`T(Nc!GsZS=pk(t| zXI=LrX6bf1`(j&ZoIW1OAFuNxkcL@bO0?^)Xa$G&X^zxVc0Kyw&wS0t1mSw;b7sYi zTFtwPnv2S>^!pVz`MCs`Gk_{Gp>Fb{@$+{_$h^vA6e9BeoLIa0bRoEAfuBI3gu<54dLsbQB{!tHqLtSReOxEC=wf}f*KCD}< zU@B>o=wDxk9Jf4rPygwphz>OXRtWF*I=580CjYQM=m=UqGuQPSv9dDr>|PU>UhS%L zwk^l*{|?`XH#f|)H9EEt`1M=7RXW1EInYH+gk!5T$-{a*wqs&vX2pP69+egG4%+H% z9F3%pVRUFN#zuJ|e=E}*8J6lz6ws2a5+|1r^yC%@9p`!V@cChW`%PHAYzg0jg{|Y! z8L!pUPdcv+MhZmEUOPQtX-mW-47vWU9`}MT=#?pjDO$6ornJ4LbOaa!v7W66X>G=B ziLQ6hd4D_qvnH;um2DGQ~YCgSm*ua=A$$}3s=XqaBy z>X&oUd4!zv3btuHJG~n`Y{0kS{B)q^BJIlVV9da?jqQON48*4?Q5T<>#Z{04Iz0yz z;p*rxaZyTPVGd{`4V>vWHf%jZx97_$x?7{}W0pX0Oi?vC&IKpF==y#Pkf-eXCAvyu zmuGo)_DIEEUoQ%_bn-KR=GZ?W?}W9+dQPZ5z7wZ>_YWWP42%MurL)0VW^V`Bz7X)z zw?G~)j)FCs(@}%);)uaQKIXOJH!TG8(5YXCYY`BK(t!ry} ze*Ho~+vP*_<>1aL1|W%4ugRMR#}MwIkiMb_-mm4`m}kp1A+6I&XpxnJA^o3+Pxdao z%Z~P$mRqfL?pyO!*>b*CB16O@Q$H~gZg6$H*>LDS808Ds;L_#B&7F5o319#`(*iVY7b#7X zj8749>V$ctn>D(6Vva`yy_!W8&6*cAdCA?-c)1&&K2k)q+Y&<4qZn1@xpCbvib-cjGOp>P`pkl1;K z0lW;)+*b5Rc-DmDe{6!JMmnGy<#ciT(l}cm<*A8(axnc?bcqz~M|IOw?N%7N**yC& z#BgrE6#~8TwKV^frS8SC0;xG8II%{s|Ie~u@++hCIV&g~&$!|rl>H8303ekK-ZSzj zh1@fHb%yo)nO{KjB#$JPdMdr(kQy9Nojl1Ee64hdZ@Kh+n&+r^7E&xT9g0o>0@kq@ z06I0nFD*dRtpaA-OQZRUWfH;?!oNPBNI2iD-cKy7I_aIR_HGWVd0Kzw?BzRwPdmab z%9P3=SV!Ma(GcuucBVm8a}o~w?k>V<2G9^a2%nbicHXk{t5VUl-*_5#Q)cKDv<1Cq zG!*I!1wSt-EMM$in1E&!v-JH(QRL~jw7a1$yHPU?pgYdqE70lJo66l6*5s0m zr%PDc9NQ%hk-=_KpRVfk!8&q@#U{fU^CK3c;eykkX;W@H$6}T#{bZlYomV6tG>PoM z8!UUP+9+I#bTBqG0YNEn1oa#jJyNmDSKd?e`J}+PeuFUwVP{(tE%CDX-t-J_W~ zar|k&qT0zDuB=A*dP+yRd^S(N2>yj#b1eUuuKI~5arF1sU@2avDC;gm)ansmz1&BG zlW;M^tiArNwGx_Ql)|R2mpjg_+Zy?uv{G+nDo(_VLcJj0R!gl&w@hDxo z{0ewvz^At`@@^a73hU+*slJU3@o#DNYWR%KlmOJIm5G~QQ6vUpquBl(0fM#Nj&8cU zaM69R;$HX%^r!TotE60&qL!4w5#@~@ieKv*nNoUOSF`1HtuF6fv)=`_Rd7RK9MQWgQBu?&BC$N@{e((X!PaW_? zYP&?BbY*8kt)Nkt$xb|95WLLUfLV>4_vB~<)!bdu)30N@*=_` zH64q+0PF$kAl9z*o>yd+4gRJ5TDyGnCJw)Ko^p06$1Qz`MsI(7-@fhcyIOlY8p<+m zSJ6JIt6zfGuPu#tUb=)`nodc5@g8Cw=5f5~Yq_nr)5~bi-O%f$z58>*=#|kKvQ#Or z9@z5$ShS|ifYxel&JtD)V;F}9x{Y0~JKCQZBnw2FmhwmefUHY`klj~WXYfI#<-v0Yd* z$qFE5efS?oT*%alZ0&eB5Xxm|C&ytV h=bFVvypIQr{vNTse_k`p5sUebVE;a1RpA&z{{=J33jqKC literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-4-- $A^c$ is.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-4-- $A^c$ is.jpg new file mode 100644 index 0000000000000000000000000000000000000000..f9ed60f892be89fc1035ee2e2318453ed93c6d5b GIT binary patch literal 5225 zcmcgw2UJtpx;`Ob=m-RrqCvVuP`V&h2t}kLAP5ZNAVmZz(ha?e3>HvO&;U}Th!9FZ z0dW99kw7TY37`R#9ufoO+&FjUzR|VreRJP>_wKWHvi^0>KKuXvea`oN`+y_hG_d!y zp@|^?fdBx6aRA^LpbxO_*#q0d!U}`I*w|RvIpJKK92}g2eEdA{gF=Wy2ZcmLL?z^< zMa7Pai-^dm9XWbjL0MTDA+4@;LQzv*Nm=pRO(1M+Y@8gN2e`NnC>|C$toUy?uod8C z1#SQYC`1fk;)OtYA>eBO!N`*Z@>>IbZxALZGs_+rD;qlp;{xnnfC&PHGBHD0SeTg^ zS4T0<1I)ZEe1{eE_V8P|!Njh@6=O0Vvx=XpZWFM^uSzIg2#954-?#sOppc}Lw9FA% zWffJm6Y3iJ1}H-#W0TW1wrB0^9nLwrU;NqQlBbtS$y*faSK2pR5CHmjSd8Dl1N#9NF9R17Gc%MK_6-+=DTL9Xyv!_z754DyS;5?{ z@{1|Pu)fPw&?0)k5z8%JFkSqdY{^CFI5xB^x?_njej62?VIf>>U{p zh$oKGqj4ZWXP*ZF_M9lM#~)1B$8?A7Uvo|)vXp7JZ*RJWle~P!x!wQu@C8y) zmA`rh)m*NH#8CwT)0M8P1>X%p7Gu&QV&0-(UG)V4U1u~1umpfW0%G-n6S{A|D+sim zZ<%q{g8!Ad3S6cGzaMC{BfW2GlVunLR_%{b?TuJotI4>q?K9xCeg5D^T&(PAm1UR$ z2(bNHzY7TlfgOP!I{e2}E_;U$w zyIYnf2ut{Cl+19$0mfv1PX31+EKoue=)E87tTS_>jQi0!k z`8%7}j;g6jrQX--inpJ4LlNy2f$KZAt9LG+a_v!Twp(uB_?5|5uD7>wBEm^;>!D%? zZB%%KsMyr)e+Xv&Hiey`V_7WFd4^fbd4(vLFtqOjkn!@MmFzMPvV5Q0?vAa1dzPHG{jNt6VAQQcSeuAPCIy zM1a6~1PJJvOU#e_hfupFF>x9xIdmde1!E zN|yz0FF4yGoWa$ar3#T6lHadT5Kw)LGBws0qMZA@iQfxZWBQt|)E_Cwo>uR5We(9; z8-L}NMf)>~ru`isZIv~tmv5`C1@-9nbEvjw(g_dIN9_(tWD%X3O7n>8?znSv>K;0M z;SlpmjeCikDW3*^J?&^@peJy*R^Iu7&Efe%zxyNG)pLYRtMl*b8@tCM5*QG$?`b6Q z01dw8jK^PW@Tabg@Yi)@Wf1duL&Kt<6I>R@(LEJQg8L7pPKBWz$;Y@N_4`$*qesTJ zwdu-oa@clA>tgqWAo`8KtfFzIVgE@Ml%T|xz!249reRI1WKd;3OIX@w%QX0;$lOoH zoh4-G#%5)=fVA|b>aoO9YxDQEAxW--nz|o#IL9#O;t~ht3CZtw1I(+)(^1|Ik-Ww4 z#NJn!YS;LyzgxPJVzI3d)%MjU$jp4{Op68g5=wKog=FvDWW4~p9K;Q?nkuQp;BFFS zxo=JL6$J7;DZtE=NH$|-d<@#TSJ;=d&##HC`qNR-8fU^k~0HklX`!sgy6Zo zb3)bUpA-~0d#YLcXdg`TcgJm*x$sSgTtu`>lOkw+Fng3uKiSmml6fMn+e(7(MqFRq zIPXYYn?OJ`y9_2;X{BYjYP*GiNBuz{GEB~{bX`gdxeWw1`Oi&qfoW(6vN6(cz^ul`!&EEY{?dWj3`HW9a*o9Z@ywA15{Vg+Ol~le69Sk|+ zbM7GKuJJ9l;T3(+?Q-qAuI(DHaYb?M;jvogZ{EF_bAPh7qraPVz>Qpio5w2brp=n2 zv~kTEvthi3xJ^F5QEHc+E>_;R==ldrGXCfM?CHD~yz<2eA;V&sn`jErIT?{|9RVmG0&UH)>qD6Sw%w zy2MqUtzm_F5+5%7lu{qsk> z_B&<06sklwsNknJV;DM%;-f~7mwd8NopbY?Cv9&RYOCB$ix>4DRZ z!9Y#Fr&My3KhI{lkb~dN%rdXh`jeL{tw zm|9f>&?G5b`23DCx;9d;YDxc|OYrkLKihb>a1#8axJ`NR0G`F)U!-MGy$R zMi%HvE)ITs^XOURHmMhricFsVMw0A4m^Z@WH=Z)FlHrR-VwMmb&3e1VGqXkGCb<(z zd+Wk$(XCUzbQ0`?UDX#xA5O-U>y&QChPBXzKHmkQsF?2GfA=kGe!`ZQ1#Jj%hGl2bb?;)f znwsg%7ok%sg2PI^&1uFN_BvkyxEAwdD`kIj;?xvxU9DEq3%Zza4^@92ARp0`W9{w1=cJW5|T!zPxFyx+51s)@aj zdIm{qwuGgHGRQhRzUvlArT5N=f@|4sUpO1ET|D);eDJ4kL1WQ%))iMn{ma!KB+HZD zp>7z4#B6Z7XP%d;U(GkeUCTFUjFR$e!S5uiFQ);M5nvx){YMrPv;#&9(^g=;N_`P z&+LO0Pd`Na)dM57na+VN#A*4cx| zr{xvy?YL_V;yw+=4Qab(Z_ijev`n~l(SpjWeTWkXrlG}Q$2ylx=X^1Lt%rJcI+ohl z{@{mZ0*t+&n`BE*X=vcNS~tplQU2m-{O$bi45~$tdAv736K1Pnyv+EYG{TYJwo(1+ z4bS?pO&wb7qhUTE5M^v@W3rZ5(a|HvK2APKA5hO|v4n8uQ6D<#suS z;l(`PD$L3Tm`uE}$_QUV%8wb`lsm<-;1O0-AKwOu%61eBwjS)of3+(D;QzLKV5?^N zM$Q|4Xz%F0x{T~DU&i`sbPFSW1HOvJFkh`4F^BuZ#xWL?&>>QlEPZGgP10dl{YZTf z=pcO`E}Z8%i)ntk?c~g;?e5C%8sqbUaKZN>W;(n=VBU3fbY&N|f?i2RQzsCA4)#L_ zJ^zY5nFW_xOzh@h((eHWox$&SeG7A$s(#>3chPVEJOHJr}(p**Y% z?{apxqLLw#*C})=L$T8FKZO2TcB$)CUwlL8ag-LyYn`1d8GjDaVis%d@1lKtS~d}k ruO%-JGd5X4KZO6alBE41V;f#7T;7H~KEs8xhyX@uQ2TY!;D^5f5r{&? literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-5-- Can exte.jpg b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-5-- Can exte.jpg new file mode 100644 index 0000000000000000000000000000000000000000..03ee0fec496d586e7787ee86d6abd74599b375e3 GIT binary patch literal 5168 zcmd5=2UHVVx1JCnG>P1^XU^<7-|lC?A#fbnZ>Vda3qT+M z0HGfMI09$`EX>R>W+oOG493dJ!p6b7kAt0^Lx`IP&WjXAi6Dg$h=bzt5(f{-iXjk^ zDpIn?6qJ;dP!gxqPAZ;}Kdz+sQwW5Wm6d~?LvY_dK}Au7sN#QnfNudV7T_U3ghCDh z3|tT>7X)kuQ1o>&LH-=TzYPcjl#z)U#=^?RPM=V*A7Fq$p$v>rCMHHk`s@JubAXYH ziCa`blbHu=13Pq^SMgqK8jF}#$veJFz02aqZQbs(vhfQD3JD)RA|WXyt)zTH<>aZ; z+Glli_4Ex4jW3&+;>@m?+g-bU!`{Ks$=$=#%iG7-FDN)9G%Wl<#2<0-kDok!_BRU%-CD#YM-(z{m(?g#E+?VeqCCl#7u`RDqdW6AQDs&2vcc9t*El zY+A`XRf2C>MYR))&2T zk(xbA6$a_Pc2(yW9&Efwyc*~k3j(aVB_N>Xl?MU@8xSDRY(4;iL(Ih>U>QaWmfDUq z@}88F?;}lJ=)pC{zq@=1O|r#M8OmHf+o2~UKcQzX44}U+ps9G&a=bHYO~NT< zV;x14kJBC)vs$@OC|CE=V-f9=ewA8k7-(-Z|M1?{AjZgQ+#S7Vdi18Da`ZM&1!iXl z4gvyrWnTOS^f?FwXPALNn(Ac`c=HMbRy+ue*q4N!q2q>j(tAog(~;wbnXGzFvYI|GPT=_I=Nu5AGa_ zt-OEae~{ObI#bDfK}zoCczu;cq@$^sg(}_;fxvoy<|OQ0Lhs)s&u72%kLI0uQX4eX z9T@vLL%LM)6`R?F+wFUC{R#3r_w@dA0qKzq{|}Xz;xxv`qy3t{HabjM@`fZr%Avd6 zKxm@=5nmhC5(Hu!QQNDSUDlcW!Rly(J>Diw{DN5Z9H~ER-4**q8*%(?s?B|S!J8T1 z5#)U8S^BF`90(L?)7_P?|C0a1_MSEUHRXE!Yoo}{4ofBy>}le@4CGyaZ6G~DgBPm< z@M85p@);>WG(LzgOF|rq{H9w)Q1@@#SDo-ay=+&en-n&9^<*w#Lcx9t4tI7dwp(|p zp|+*J^`PJ`5?z!Eq>diVF>{#|vUh5vyFsTI2t+5bo8KeUxv)o5Jrb&HCXrBVqD4!d z4)Vh+<5mP_=i4iXSBTFluKp`B82{HVE{Sv*$dPqbV+G|Q(~QX&RA1FoBLRWJte1l) zNS=v?@8qT&5=`A%pK1w0qV7JNPMFC}9kJ2PArj9cw9HwC~s`UTy*9} zjC|npc@2TwB9UowYsiGSrrhW-a=g+OPQE3bLzUkraAvkPSENA$1cZ~X)&4Qf_galJ zf?OCV<74W@u1s7H-l(}318)^9Drzz=8IWsf3Hd63MNNikG_B=S_hW1ac(Lbfba*d9 z=mZPD^Q(O`$kWuQ{U9)yK$yYkm~+jdxzJxV20&mw1_XLSJO)1gZqv(Z@BcEJWp)<9 zmyEpN0@q7Qt+=~LJUTfSM)TDryot)$bf4GYSBCQ#$j#3NqAG+#QxgrpRlaip0X%AO zH#Tl9*Z3U>kYtan$gS08PWod{_MO6e#{8e#%UijO~M63prF;S@yZBlj2iQjYsi<(F{)<%$>WhPt-Idq3D3eyw-1}w{EahKx?Z?2&KCt`T3N`?x-4;e!@uk5al)(AQ%$={ z3P>{v|9Dz;&Z~%zOV?cQjI`uZ^D>iLna@F=hSBeJrUw??r1)db4)L@39>92yUrRaX zkqr|ozGL`F`H7Fmbr<>OJ&aQcjnSVk}FO9-yRX10bv?$hIcq!v?CSAinC2;$fqcEOSEJv1=Dwms8oy&FO`SEHHSu!|{oH z&uqJ&{$$t-lx6-7ga4p~N%@YTQs?O}?}xRTT!r7?^@Er?K$BOVhKjCS2s)g0NVPip zg|mv!S0ozR6RmB<$72u{<@3>->r|9O_j2`k#I|{zHTgzNEiN*RD2lJ#&b+1%(>(1> z#i2GjZNkHpT!+537*9njjv@+qu~xoQA-r@4rV|p=6wQTX#|tYh=_cjY9E0{rb=%Aw z*xGd_IZs7COE6>6lah{Esh*4Cyd0gzP`}aWH%1j#7kEYNVk~{n*p3d>Y=VCjUX#Qm zoCttp*p2Czefg5X@~|o6o27;rU(}S{R!L#)HEC`Zq6Zs5&07PoI=p`pP9Yawbw|5T&f?jE z>{PwjZsj56UG_nYWoLBviWtuBlbZUlUQc;wFzo!#jAePW<%I(JK4}4-+)h6AZhf?z z_6R@VZoNOoV0pvJnAAD)Lx>h%7{E9%5Y3$)djX5d%_Tw6Wv*WT(LD;Ws(%Cj4ENRL=j6KhtNC$vq# zcYSkc!E4wRcbXmxVZIZ+CXrsqvX&B_^5dWr2vFJxGg`GyQ@<%OUu2T4aV4n?zbSJi zgxPGkO=E)%vsOrQPVYFNJFa+4`t1huU76XZGX7atMn&5@8T8ddG-(x1H7N{jy$*7j z?FK1^&lg72yz+L3sTT`9h`;DgU`hSR=QI#_S}27iEG7n{b~$hu%Qq45b+5A2`6RLX z6NT~z`lLb?1A@Y5G(3MMYQ7ItqeRW+RWr&HF1Fn(_d&pXUZzw1=$y3dT0PQYx+bMN zmM|2ZOc>X3G#Ivg-tj~(`4%%E5|7pKm5BH&)zhVn-Z>atSxjaxo?vGbn}aI+tPZy_ z@3N{f`L#AaI#qZEwLUBW&%Lm5hBDks(2&XDsGsVp6L&q)H5VaSG}(Jbdrzf73825; z3ZvE0<6p~9y63kZLngb(o-9EP7k;zrhT7g=l#sS`EImRJ8sFf$saJQMGd&QGHf49Y zHJXt~D6m^-lA(;?@s|gbMqI4VRlj8$t~_{Y(e0709ul0wqAiXEdQvCGyxj<4!QF6!l4=xt;N6S$+6>SE^<)nw92 z8Y+n|)9cEf->7=*EtWSP0X!eIeU$0N7Qt%k2eDRnCiSkUJSbMa6Xyy7BI`kYht&w< zXXCbTHY+mk85TMw9?2Vb3kS~x&VfMp!2k?FlTyTFg4N1}XuO41xRIqD+oEJm@Tc~d zV|8vLe%5w~4j-alR;oJGR^P-#=!RVn>ixF3K!C$;E5&bd zzBh|ge~98>ynegvGd<%HhgRT`DL+oqyFDvMdInMB-P2qYXsJo;F=Nedp%Yf!_%l$e zK>+a*haRr%*m)gYc{Y8PQ@rwvK1a@=S~R1*@{lQVI+WZ{!PV(5`V24Eol@fc%}g;W zi5b|wcsI+QY?LHTR#R<}LDqfL(hdyb-+)!2-;OP$G}gvw3c(H%RZhIJvQVoaDs*y0 zR!qif%fN3rC(>I*a}Y2%T&Gf?T&|teDs)aHar#KX5- zvObp4H(qxuKcQl?=5DYjz~0zt^&rDslQwqJ<-=v%|}*e2baD z4_rc8w+P3KWN!w@()+89fJ?Vnw)I5F1a8S!NZGy?#-p-*;yyo;W?TV@F z%KiEIk|uzpAi&84-05_()mWCUbJuM);!S(XU_kuKoAicND8}IwpeO%YNZ+-8r_5JmoIcH{_HGB5#-`>Bo&pzM)_!T&0psTM7 zz+eCXgDwCZ22KNv^z;n$bc_rP3`|Un2xblz<^u2E%^7HZV@(PH^i3b9h7AU1gMp0z60(yH_D2K$r@?68v~=_gj7*3FkU;4nfCdJK)6l}{=xAvn>2T;h zK+8sVSWsSzp554iLCB9o;YM5rqp)^K9j8g>qKKm7<(o{1BV61^c|^s;B_ySklvPwu zo>Dt~Mn@N`r*B|tW^Q3=g|~Kcc5!uc_we)&2n-6oay2C4*6m2*ohVX#!h^(zNsk^U zXJ%#ROG%>=#^Y5H1>8S~xAkPh2pXU}(VEXz2vy=?`leGdTFM3n|=S z!V1Tt6 zw#`xf#kW6>B!Pg3Lfg@}Biz=w2fT;j|;AHjAoT6(*=x)Fb^w4CP@C2wcZucV&wY z653@AK>$dvzSi8@=chzbP?-0OUK%` zrt%42**ihtfL8-~<)aA*)QT62XU0B1i>{FS@QhGDv`E=q?x|U1n%@6Ge?lj=FBfT( z{{++gF zO1iqcH9p1jxCN>}q^2}pHO|#GKYFma35kCc?qlB?%DUe*v76`ZctIfDI*AHHT}5$( zdpd4UR`Hko1!vZNH`(EIkk*g>&pZ>o6x^VflcvmR+#=Ri%Q?kqX$k`OUE25jeW#2Z zU4M@(@vgGkW~ClZQ{QO^fv{w0VrChI?BDTH@tSNA2ylIGiC6XLx)(EAZM~A$#cw1p zE0QX&gCP0faz@ke=?t*j4@+9R+v9+HfikoM$7e$u3XYP zy*1gFSvE)ID_Bi$wAoR+B3cK__p-Q`{Z=U7_d>y`=AO+_jdOUOpIL=siNScmh*Cyo|TJtq&5OLwwFqiQ)4 zjQFaLD-|79Ix5l^rn2m*Alo~Wc#F5eCzSAdByYIX#1^j~=y*4!?KD-RzxfSB-{oRi z>I?{&nqaf73!}tY{Eu%?)Jsa?wQ$fh)C`WjH^OVTe7W_ z6f}8;Om{EwRcyyZ7Izw{l<|09bPdm4!GS<*)5;tOP^t}EL7=*($0CtrGs~s#mNl%r z_9;Tp2_qG|pcr6mCO4n+kWhk;`!=D_tb4|gD*obm4WpOom4iQ8+@sNH^2m$Vmnqa_ zN$TS2w3oYtO2w*59yfGXPvaUBxmc)rXCZy6U94~Lo*%2@wUuj$qaMje60)>~x&wyK zpMfJQ?nq)xd#K82GQq3B>U6k^LYub&F}i0_t5A6)p7auR^1Zx;9uyjvqm_BY*Wo_G z{CT<7ze{p|ilb?m4vDQVXYn1AjI#D+8C7!jUpFk1BMx9j?EBv8k{;{m;}axw4Z5&x zuSDAtX9ql0UgAgcMk4JddiBPL<(Y3M8q>MD!%Xv1&Tqb${iK+8_{>NFA@8&uvfz90 z2w(50K~Js1jY_5-2OsvGhK`ttQop{8j|*6*_G04XMplU(NutQ=feTxhRmc8gwn?%A$ZI%yv-?e8q&wq z64ZJM)-CG0KA*&jVn2oB{I_~Zm&1wT1CX!cwMi0BXbL>Jbo)j(yeh9AY)&^N4bc|N6Mo_veLzE{+J_q(BoKuv5rq5-Om){PnBP+MoV>!bj0y}Ayk)= zku#Hg7BLPbgBhyW^lZ+RqEVBmpt=GV@?Ko@Y^xe(K zI?SrxZ}mHzZQA zOTKgk>nl&RkCt@b3d(otpUxyLpiX3xBmii@?`)Ph%Z0Hh*CJ65mcHb#*42T)2dn;g zcwF)Zl20I^HhC!3>Br^3ZOhx5{~6?vX+G7F^5rd z5Y@ZCNQ7B}KwSG-5O@(r-j?@@F8cWEUNerl(=N@qtNHBvJEM@^Dt+dt&6_OleL+Rc zep*);RJw?m`SfF}m>d_r9}av!+&lFc1rA+ik<}N};W!6_2Cb#!H|tjNh&IL`!hoQt zU+TN<$}rVr^K4FfLpIR)%86Dy;fwDhf&vDSv-mQpzKusBGy8tg#2AO@T3gWUro~AI zr2XUA70!lupyjJ&;jUPBa(}%TRcC1X-rOa{ez)Io9d1iSsS4F8YKw8z^T7}qgZb|m z>u>ipLaPIinhIHN5?g+h%KyhF`jcY5Dt zsJf{45#sweGQ36(o%@bdT1NWZb903S=pa?TYi5P9@Q?Q?r%Co43f1;ob zg6wd3(ZpuABQyL&yG|Qt{MW;jiiv2pBS~Ynbdyj7C<5-%M zi6gHk7vmK;^(uTObo-!`izsK%*~uqGpRu-m?NQbG!G$-CO7@zbU{+di_M2=iy9H@N zPt6fe0D6Eg->WS(H7DpQdrE5RqkXyCx}cK08zD!SO(94bZ{CL%K^Nq_IaEFIm#qEc z-7i%DT6pu`a*2H-R7kuybxTA-Bo)@8Oyp^o%Lw#yZGI_fIz_G$iW}|N>HE~l^gz-T zE|40WoAVMIDDt)JJ3#9eqh{VTG}SZ^*=E}Rz$L857e1?u&_`CJr1iGe>bHB|*tcrAnC!MQ+s)Cf#+pN2<%AlHg6q|f1 zlz7}PQ2nmOy&?Uh<&U?loOT*UF%4_aL4cV2FlB7c!Kd2zn8vX7upK#B-S=DrFLE6O z5?!nZhX*sPgb!3bL>VIeNAiS)>0jUe4V$TS+un5*+u31rI*5~s z+xi=|_GhC;VKbMb1RLGM@B86ftA+c%nKpLE45h`AsG_*S?BrnL(XX$4AKGf(Q;gNS z@twSYtgjx?kIr6hVpllLtUl}+sUC&W-*)bMZsv^^SM<5c_~@G!`R#l8)uzh^g((|l zC-d+<*VY@?=Y-F{@0wdd+#C*Z8_LdxPRZQx>ylo=hxT!@)I1QgMQ;NF{m@P_u_X9m z9Glar!#Naso0S9t%cs7R*I5%lKv&kU^aoXX(KTt@R(oV-Yd+rHQNB=1uKhvkeFB@6 z{bIo?a%-yxI+aR5XF4})veNL$LWdj^o0fDhoe;}iskaQ&Yk`@wOLe-zs?|i=o~a9xzjEY1Q5kq zAV5xoN>rNIP2*qZKkP4UKiK*A_F+rH8~z`f*ZZ&HGbC?+yo`Krqxl1+R%XuT-DK&k k+mkG?2rlnC=rJs^Pgq!3OjJyE>sDE9MR`T-|8wEK z0VIWhdq6jcPYKu}$p@0;<2C^h-Z};N{wTmd3f~qGzkncENLWOaH=#lT*un<_ZQ%zA z2=MdsX2)*Nf0MOsDcOpP z@VUV4O4`?jq))_WmNp10AF)}KIUht3*`Wd(*+>75_6OO226p5BMfNAy53T_~48+F^ z9!L^^1FLA6p^j{ptzBJ3Ok#uIlG&$r*==gg@DU^W2_gR1imFRrp&R-^$N>D@H9am6 zztfou6zjq9(KA%DzFtaOgM*OiC&SjSXs59z%RDuC*1a~}aWBW{z|lx$7g zTS3K_Bhl>r>6%>N*)002KKfRW{)N`yJa?D2!*+#QIiaVt(Nf51r;1k9#NYb#_PG(R zlFGP%aL)buDg9yAwu}+;zgS;DS${^7;En}@{aoNCwRw1D&t*0kxk%(}NA5&-hSo3c zfj5@s5?6sf+%(ANrRC6oNvxFdoJ%W-K1EO2zpa(N-BzPvDEG;jcKcMO@(b zX^u3y@zUtR`nD<#D2L^sf-N;`?;ju@05Zl;IlF&$t^9Ox@apXb8bxSOaLs{G13ly$zCmOR$bcc?a)3n*z;aRIj= zwCZ9woHhh!f|m-lo)91HdMRH@*rAGfj&;1=y+O?-RJThIFb4q?LK~#Rpp;XK0uIfa zI~F~BJ{MXTOV)#tPFR4q_5zLW`;Np$|n`g@?G=X6rY9DT%j zABecYnH<6JZ{I>?$?Ofw(hjyjeG9JOqi?l=4x;GAl^O#}!`S(~Rp}oe3QZa)_;jD9 zH;6O`)_ED1XLngcF1z%U9mI?m1)NLmosMh>+!e)Y%#MTFyMvT^(I>))J$2TlBUB}CZ3_Iw)6LwgUppvm|o zY+cq9rq8vK9BV^~b04R4#~z!~!u@&zP=0umA7 z<<~PhX6<`yLuG74c^OtUC$>tMAhZ^}umDIU}^GZi70vzMAD zC5=Mt6@e0dKUOS*e&>Sj+sE<582lXfZXz2N*^R!9%a8t-ku46Vcd+l)3~xrXvI`o% zasga87Z`-Jm9zw4qdngxV4yFQZyro&duZBa(O?$}oA5xWHwl1L(t8Hk`ip0~(+^3` zEKcw(J33p`Mp?b>qf(1}t)4tt(s^##jmKQu!3vHjCb-XfI&0{Wu$15WJr!P{Z$gYa zy6SvSb``>+ba=8+s=OYlVg;;c*q(+KpU^@%nCA(ZiBNc4xg`V)6m-OCk@IMDw~2rL z8S<5kM#t!PF?I~_w3{$3&LI5AXAM4Un-U^mV-Hw5_)^c=t;LKI;iEq+hnc& zPckuXbCM$Irq)Ad3b9pr#qwB(?cbhrgteRO@A^0+}w=c;W5 zbyU5|wsn*v`Oi7xaDzniV1?yN7ODrmYZ>$HndgG3qAMX8?KEQXs!x$=CC*6 z_ao1#V!wG&2c6p@4&uJlcVAt{q_(qMj+m&*2G}k%ks09Xf>4|Fcdn{6M;`@3wA#R` zG2*$a1$8SPL+jdKWNaE2znH#PS__j*dr6*(sf%;Y^J#CY@ZW-gW+`Lnr`fm2n?ddq z-o$42rJWE)EBY~^&IveXkO9mc{9woDv-5=K*@UglA z+U9Jc^XHaDi>zp_J({5fC8kyWNhVxCeH0Ri)6$i7U$egbn``brCf@!*@_^o5<_DM_I3U))f zvh!;mv12vsXA{w?W7tXQFr(PTv=k~^IcwlLPP6y$wO7VEWp*nGyUk10IzN+f8)Nou z45S|~sKEECMmAD&*%j@PJ-`LTc4!jGXJ=C@`p|WM%D7amZgZ* zxQ|uI@C8N4b4x_vQ}b%p4z`aQm@$0&@-_Z43Rwh3{T#IHlBD5{GHyslpncHg3qG=Xz!< zTB4DZVY|iXJ_*IBG{fdWAN>;dT5wOW?XhJ9LDf8UtAa5?S8s=7uc>1E?|ZSMX>Mm| z0$#r6p_n&DD@$2tV;r}ygPe3yC%-ujdh1~+)>RvGhvC0+LTGhuSSGB7m-(_D@-iRr zy%v5u&aA45BYUQp<@DQu)+dVG2Jc(zdBGcN#ce-b3sra$;2**wT&|C6k8ymO(cPn= z)NgJ2J_LoZJRtK!nC6)JF-`bv0v9-fh)m)g2u`>~%3tpSe~NKI;pb~;xB-N|0NIFL zbCCT=>3GmDg+jO^`qL>j}(bePeUO4Qg4xez^^%>O+^z{ z!`~E!jj4;d8fTN2#xK9y-LgmhewTR6hVL|{zJ#qx$(v0@y2v6+CKYg*g=GE&qpbityLGapjz?(ch& z*5%lSOeT{|*x>1gV*5@b6>quE>QWt>8PSoST_4`eSal{|6RZt-tgiLjlYMcnF|wYO zId_vnBA-y3+T;vTD})jcu`G#5No)6GXIy=p+}W|AQ5(PLs{~8cGv8lysJ(WS{y2?zcb4M_UvBAw!5@FT z;ARY#-n#SGLzlCUt`;rQJV+nB(F2Lk z*2Pkn&=Rihte11Hv!?~JkSr!v%3gyDd|?<(q{L@gNWtt;ozOQqrw!;*$@fi?v-jIl zydN}#Mx0w3GNU%a$7`(<;6v^+s&hLXn3k`F6L=4Zixb)T!P8g6#Ony62bi?u&dV}* z>R{vaybGbMzAQkQ3)toj=B`|M0Y{~}m(Cc%D+J-QUxL}wXTn4^CvKf4y{;_ZmF{5^ zsTqXq4ApVo&^`n2n>lK;Z6<5)rU24E$H^&;aw2nsG>#kfEut=Csgd^7x@ZZ zcy11Uhdy!t)48%Pw*f#|iZd6PM&?f(Dxb>vV01yBV&bD+4G_$7u|&}s^Lif&0h2s$ z9mbQ_L#*158wKviJ(9tDcWqUu;GJhP#k=z^gJ!LsCHRfK0VOs^Z*0r-D=`zhDe(;Q za6LwC`h5nAiGF_3Aw216!i79zd#kxM=7w1xCz1=$a@bG~rGg8j*R!U1Z^3mD%*1M- z_IgrOh7lL=<{eHQ?rh#8@fKu2w|=>g3+#c9xErI)%kIWl8Fpl#%l*-l#T0%RoXbf+C^ZWLmJ3!WVQVMhG1B(*73 z=xU&No;hiG;fEP&ljzF@4AFEdM`lGvI_lV+ZsOp-BQbnxGJ~3r-uPiBjs6KKMgXDE zuE>BgjbeUw0sK#<*Ea|_FNyz^FUFA)S&vdvk_QfYAl0|F-PtucfLVWmnTm6>!(&ry KQ5X!I+y6HcnX4@T literal 0 HcmV?d00001 diff --git "a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-8-\342\200\242 Example1.jpg" "b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-8-\342\200\242 Example1.jpg" new file mode 100644 index 0000000000000000000000000000000000000000..c8f136b717fde19caa753a48bae170ba2fff388f GIT binary patch literal 4983 zcmcJR2{hDg+ra8^L6Su-+SKgJMVYTxn|DwpX=QB{kyjR#rVMZ2<$gBKpFrb z5CDLfAAm6q=m8vTZ0u~T9PI4uoSYn7JdnLS+}u1O`~tiXaZyPLaZxd`Lo$l8hYlZ? z788@xIClJmvYMKjq^zdSNtIKIs%k2`n}9euIeEBwg!k?hR*@2uQu$vm#%q9&1Be9% z!Jxyy9zGD555#x_NHXuq3i@*a|6HIwU=~(3b`DN1Zl*!?eqav>4Bo>6W@TkzVOobV z{|8w3Sox)t;cNmZXZFKB5S3f^vpA&Bls5~S^{>jPUh<9L7CU#K%7Gwnp;>}S=-p!U3R(R>gMi&_Vd3U5Ev935_vo7PIOExHu1s3q~w&; zM`_tPxp~j>3kvZSl~vU>FJ9KxwY0uzYbSJczI{J1I5a%+VRY=%%DYNmzQS8n>0*6&@ zaX`-8&nj=`lvXuc6};rz&m|*G0AZ#uJExFRl z0CLP;DbFPs+aaAm-U1Sz(-}Z9`}~EVSapScXtFF@t?f7iXn9MtqGbd;sTU#K8zs=l zMht*kx-D~k#DoDf?PUOM)_Zf+QWUNS%P)Bxkt?DNIVx{`saj07bbV6lcgome$?q8h zh`K-1yS13l0H*ISfSH0{9~3>xPN_+4thTNlz_fG_-KgXN2C&kKAzJ*bTE6Q$2R!IQ zEj^R8E1LM$^vNZtk{tpj!gf#TB^P$9(ELK+Og$|SN@Kff6arNG@_p;~`x$|m6$ZDTr zeb+L-zF$uNMV-4`yY@XNjW`<>bFfSNsbqWd!7lrV7Vgw5@+xOedhxcTRHZhS+t&^= z9saKz(5@HI?Xpqob8t`pGcBd7Z~C82A+MvmweZFVAKpitMlb+-2rZXD#!mM1(uELr zVe_S#3?Td<1Hk6|qL-?nmBlntPQ;KN=68|Fga&=Qqmc*2H6iG50`fQ2x3%E7dF?2! zJcIAte0G%D-BkEjH)|Q2nwW(9TFzYNrY5#G$>fM^Y+plnmkdm5ox|()5ZT$q<|tW% zX$!o3w%4^KH*F=3Gk}ab3~lo&T^{$dTAL;xR@J#ZSuPnNlD7BT>tbyKF3C268bvy+QF z_8a$`<3h4eUc|CO`B}`2_Uo9~C1RoXX1V0D??n_%v>yMC6=%_;&8Q^Wgt)0+Dx5$k z_aQ&CdAKqif>NYVeFlhl>&7VPG`#NmU7OeuWv|3D3+03=r5ARAqovNrYX;iO%Vj;Y zymQIz90iR7g~oJz+hHf$-1#+#*HACDrqanf4pUU77E3 zA;a($yN-IbaIer)1QY+0Y8@r7N=7q0z&w`>y0y4d|H$Eo%b$y`dpuYzTEa2penla! z*F*5Z{_-=e0dU&PzSKi;5}8Tz9tw~VM~i0fWyqe}LYnQWiD>G^{&54B42yX7fRlL4 zm~SrwkO(k0Y{eyo+lndiEepBtJi8z+yRQdes6V8d#go50j zQd;p0MLV4;H(G?-@^+mX7sl8j?F*8h7__;m=QvCV@%Q?k9vAhteEnJnqiXG~&{p#J zw~uoMS$>O_x&wuJ^qRIK63;t(=g{ZWjXdV~gxq(7V6He!RUKJu>X40vs_ena zJ>3;v$4zG2klF5{*D>#!7smagve3Jkh(}}KjdQlTG3{&JtRwtoG7R~zg zl1S{(uMm_zQc53U%6c$f<(P|VOZ(I;Jp|Dno|A>o9J8J#F3riNe+hO%&v*8yGJvql zo1S_7YTCV!MWejYa_kY%^73g0V3rX{%NbuU^OWH0BQX2-bKT|xl<%>aPxi~2I}0$ypk#fL^ZUH))OVHQQ;*iqI4%rc zk4W@MXUs&l(L44qr}B?v*}UE!rfFIo~ql87N`{^p01 zWMi+yM@sWVXk){WAbp4#c%<>?py7V1`7)$esLv?x-$M2wTprVTY7jOZ%>WGYf;K`3 zB1GwEkvcB9+yIWa@7M4uCe^-T2Hqpv;n~SfYb|0(jUB|M^)_8F-*#I|xpK!S4(Z!| z`8zLs>r$0`-gvp$cV=|pyU_umE*SQoF_a4CnI(QW?fMeo}hq(R;^P;JBzhGT^%8G@KcIlf4_bR zqBN;Wb44@=4TA&)c*1Yj<2Gt4W>e?HVc@KeNSxBVbzFj<1vm zbvFIfV871=_F;xn0XO~m!8P#hV^mL`vq}dkn&qaig6oCVQ2N&-+)?@vfGMCyUb@W% z)tS`inJRkpE(|{daO$aAgxMJE92nalzBYQhYndIj88~6E5Qz&*FoZdu05T+TByG-wkX{o z*~w+s7Ez|(gar}RW6*8+0mNcDkw84)82fIph}Oq)1?cJ!j2ZCy|R5`8KnBo zJsZ2$E5^)jU{^c~{9UA_Zm|`*7uzFmAv;fBhOG+VzF9BA3>W}Th_s$dppnoiEpMH; z)tL5)|7kZRl#4t-)^e%5O(zWc(d5Zepj#|15_^4R-TdOUwd1kyf`LcR5L|1emF-u7$b*S$--?4UX4!5U+{O%>oOnQsFTZX01tAlO~)O8=k`^t zhP$R2-H~|vlI{f^ENZ^tfsgkiXpHdHEQZPmncnZa7pTb$WCHA?dXk54bLz~IF5^@Z zNe?Obp$& zgw+%e;Q>ytTJ=NS?=oQ$LpEB4B5z!$UEeOLvr5(zQm(OYglc@m z=-C~*V+v=HfJ0QG&5A*0C*Ui#J{|tCR#UFs>5ui82PC`gnK71(O=3tfi`PTD<${R4^Y2wkFg}!gqwP6#Jgo*q8gvxywWrr&_ z=JV0kb!=rqWzX)p#d#<5U#lcc@@A2A;zKq!<3Fu`!Y%by?SC{%xw_^3g#q;HG&a%e zNFEmE);f0TN$0i_h#Oe;J5=47$%hMpnA$2_lG4?>pKQssbCI2wLKTJv2sAJ zjGWA4Q0LL)3fGh!EAy)Y-j(LDdA-}Mv^M&)Lg>S!&f}F~LEDkMlfdN{>~$Mg-_Rqf z3JdTB_?xU?U|(v!>Ews>dggw)h_V~<%+W0D3IlL(V*veS`jT8tbD6f75nBd8^6Ym^ z{6LwlkR+PZkzcatRM@sUlu$qN{O5UYnVg_HgV?9)&&(j7$EVM;8M`>DOgM~bN4bmY z3A0293u?|(Zw~|<1;$?a&*oV;VDFZvN!1U)W+&9U%cyIz^Kg2n4x+rsYAmXH>Z~v5 z&`$r$u`V)uTD9e}mv<>~zNNS|$6xxZ$D~fnw`H5`%BISrcj-FS2V)2uPW-$hnZqd^ zL(xWyt{a|y%n>r>TE()vnW}u@@zda1^-PK}a|RwppNphFvcHKj^RPzTit&7Ov{srP zs*Cvi{;uMuD-(#JQ(u%li0O~C4u}~vS*?%+c`0fcZsWS|!>1pw5IX70(ZtKtQ0?z&iI}3n`vjW$O@VDbqd7DoGR&)S@efRq z*=L8=q(C+KlAbJYk6XthgLldcCb~)Bw8$M556tX5>g7Gmb!B0iCJo(P9bf(I_Y?T? zg-IKliJbyO2zxBGl&rM5xz`ho`FszijibbT)rnG=IV)RH1?QJ>>9B|cAQySx;oeq# z%i~KhEYzLJJh6ohvFWbtfJ?X(fB333d+S>}nhddc5$J1Q|C0Hly8Z9f_jz6M1sM`A zC(K0Mvkg{N+%&(SwDZ-v-;r6H_rwW(7-dI{c}&E^YLz0mWu@1;(-Kr^Yx^#J8-nq$ z#qwoPlB5~VuN(8%{-_^z4f0r> zIK2hba&Q?XDTPYYXO3lDKQ5AoLX^jZ8S3<^)88}4&dh?Ql`O;$LlRz&qRd-}sQuD3 zX63X_K0qVM{V;S)(wAtLBFk-@8rd=IqK9Yiz859`GK9V_$ocR_H&0?djRE`xlk>wc z#1axJ96gvb@jpQ0F7N;UW$s>yl97zAk?~hKk9fImhn-s8px!z}#XJ#wU!%=mQO+3s EFOd-EU;qFB literal 0 HcmV?d00001 diff --git "a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-9-\342\200\242 Example1.jpg" "b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/images/image-9-\342\200\242 Example1.jpg" new file mode 100644 index 0000000000000000000000000000000000000000..a5d13a9c07bff22965d12d6bc2818171d32ac74a GIT binary patch literal 5379 zcmb_g2{=^$*S}*h)-*y0BkM$wwGuPY&ssudWQlB{DUw|=QIbhSMTlW2OPCY}C1pv; zz8h=yH9ON7bKmK=yuV+)&-*_A|NnWv^PD@+x%YeT`Fzef_nh-wuooNwcA6L%8vqap z06>^O0PF{j103w^Fm^T$7!1bA$-%`Vz{|tU%_G9Ui%&pY6d@rlDkdf=qa-gWwO?9H z40&kZ{(~y&>govjBRYpwwUyM=RewAL!pX_W!_6bi%PXw9S8T89zrDcs0GtDe0@|Pu zDS!nIfx;nREr4L+WP|*^fIk<61fpimZ8C>t9qD|2@U^E$u^ zXXD?iqRYO^!VxCrEub0}m%$;eSM)*fblZ}Qn$xv#PA;L{!Xl!wa`MQ13hEk~hYlan zI{ur!fuWJH$(gg37^`#EHqI`t7v0=FF8TQS`3D3B1>d?IaVPR_6h8jpqsIw}Pm+>} znOV=CXXoVRy?R|-Qc5l>uc)r6t@~L2>2pJSM`u@ePj4UP>*(0{#N^cU%e~9o zCiOe*2QLT!{eu?s_y@6Hc)^*xSXf!1tgs)vAS?k)fx=nY_NuV+>sr7Zy?05ehH(h! z#bp$I;FMN7y(H*#t&K}aM*XYo@(*giGyB(wh5tv){+rm}yeI$<6v7N16b_()HIhi+ ziUaqh?ozl1;j+hw2U+Cp%k+c7(>~qhAV9V4sajJaG1y}nayfKs$-sJ3nGs&ClY}kz zl1Z#@siG0@-s2aWlh7K_|>9PbY5XC*(gVK@__hX@gv;qPRnaM@ghv1sl}wh898KtLG31 zv{qD6h0)Yx^iSIA=Zi*qiSqlert`#mh>psX8HpZ~y=szu4fgx(CFW~p|Ah7nPCdPF zGhM@OIfQNQcrc89pnsUQ+X@7rFc8>f__J#7Ch@{mQ6WFe^9tTtMAz$NTy*C!_V)h# zrsCt9n>%jx%$W6eI&>hqFKy`T`PuKkgn@)+-N`ZS;Su8G^e@KZCNc33jK8G83GDKv z$t!xzl=G~=CGLiQczd3`zcPGkPtcrwz^PIYcn3`!QE;@Ij}ozkz0k_eSm&nH^PbhqzcC9T^!XlTZ2Nj3iCZ|#d^6~%*Vt!TNJ=Q z1tDCLrJ=?}{n*SSj;*ULryf7fwaeeMk~sS+N-FkClr(J0r!z#0Yr*sFzR?$g2hfpI zaWG6toa?q4p}i=g*2%{U8K?w1=XPA|7K5c%z1)r0CCdA$M*JSH%W*O?205Q48kLV7 z{TreNC(E50bv z(+rsqv&?BC{+rIjPbftY_%!t#*NZaS7B2bN4}-z+S8E&Dr|i z3nAF{;%%iG<5D3?NS0=6>_`+#Y)64D@lB?npi8jDZKnbO0SkQ#U29rt#Iw1CkG!Wo zazCn5BvrI%V*uu~|$*?)T!GAOoALo}B!as>vM zFTYk#;6G59y5~y5H{FDm^Es(4St}9T^Rk+KXX0|N8B$X&jH}H}|aeU={V`*C^jE^(Mn(RYF>o zWq1s_B58+H%AgskXfmCYUXm(Qpq=T(48RS~6rhMq|GENYI zU{7SWyRM<4k23t~Z6cK`ceRirCpI&SOz_CUGY|H453dNfqvxfW36+US`K5?mrN%u| z9yRQwp_rS;`m}HTQkwi71VLcjrvp8YkY_~OjA0Yubc3Kk(kL72h)bL4*8Yn7O@%2i z+FW1zpxN`WRm2BfbcCz$s(RdQ=f-wJ;tj>6(T3R3>&0KbJjp*BTQ1COj~H~iM9Him zTWXf3khzU*x{;%^2IM%W>R%3pyvPmbq8RyHafJHo~Q>VmApv^qgb9n1l!CDEe3| ze;um=9B#zd*4C7bN6ABx0og|= z7o%^7PH$UAmL1mMi&!`8);{7Iq`FYHZ=$(s@bRf0#Ad5HgrqNE4*4PH?3I=VP{`Jc zMc1Uu9ENO7W1JuOqEFRR$nNQU$xa>|zG-TLwWpku)#y92SwsC~eTU&|8P5V+?LKB) z=-Rt_cxLiBH1t9)T``HIntw#5Z5$IRL*eox)9%@%tE&VL7w=C_fp2o2ln1cKwz^F*&`!BV~9MR+_;^^&rv z8)Ijsvv_C&2sD(a({UFmW7+Xf=n;T55sw54ovSnFDcN`PVVr{BlVIs_#}Pdm9X)D9 zW*J);JsC1!J2m&R6~>I$AO2kY z)*bwP$$4(FAV}QXiEK!SwqJ4lIoFFR|MX}6-eu{u#AWH(5QM&fmG^v_?$0&zjqDoZS?9;7q@n~sqPVu?@Zy&$=yAh*2V)MTS)4nxZjfuRcL z&bB4)?@-HBvWX%Xw^`T>)CGC)Hu6a9GuJx!jpI`;wq57y4}UNPiauYw6&jXf19_R+}CYKhkoA(a9e_w&FBLn?LgLBIMo((#m-_tLVB78rq z$MW`326lSllRSk39)ETNf%>R)-sB&{$Dg9TIlA3y#(S<0x_aK>qsnX!H({HZzZ2`7 zbdxUj>@np0*yWmM)n~`DQ{*`5k08{Z19rkm_WtuhWx33wz)x+-ewAl*BO1YwnO_6fZ7}{5? zVk9(tHq4N6;57)uD(5Ayxpz)suBSbhAZuUV)XGB=OSM>jbmH<@ff~(JKkO*hh+~$vQ-oTc{1<5Ke zs_~68JCb%cWqWwO9x9u#_bjTThKzv#q_FVy+Igwr;~mnk$ot>(w@7c#h^;i%cP6){ z{|@&D3V(o9b<*wz4dZdY;9u4%A zEv<~`Hk!Qu@{`Fb)w^}sM&GiiT(@3Y>SmtsOLNXy@9sj?iQcj=iOy{|be1uPyAIqi zL-Y+?RVJQx^74m1VpvXZB$kqw_HQJ@?S89KSU!9K2Y1l&dv3cwVV6kpR?0mQKG}k- zeT1B8oWjf<`R{Q(X1m+nX&C%cikEocyC%It5^qCH4<@gE_4M-s5U6fk4r5l%sDmJY zoo{EpO$L~ELj)hcsAmHk71=qn6`>C=_h?0o3%$(w9At;w#?}_T#g5KRC@K}d+&g>V zK)2fWr^(d`%EbdA#BbJ~Q*)DcM&~0tbkJN$PLidaQBEVJo!{tY%bKPb&db_5?EM}B znL`8hn_+WBd=4){Et>-kYso{LZulX{^=y;=XAv)DVZse$Tn1gm+84tIKRPHz+GWu& zRpE|OZGCG%nBJWuj|inW#GggY2GI=#D;?BNT;Z9|pXMeFde+0w>2we+%c|jU~>haC0Zn8-)A z$yBdJ-VAvVknpj&9vsqydVQjNq49(6{$tQwjrzdI(Nj^x#L|+7TPMsBD~p!{*w71A z`i#)7aWXBY;Klf*;JStOw4%gb!e+yWrF%vUD<{7Kaz0b=QPI8|a%bvDukKYe7v)mp zsrHvv@0@FPEQuy4Ag&%IZyBx?+Fv%^$Teuc`1H4kP%0FvxcD@A=y|?-wgR5-WucdQ z|C_^u27}{T%yGN1`rjF~>rqNg;#JY=s3nOTY8D8XA)B> zAmCw*Yw2{@Y|Y#<`N9ZgHp6!j!TJK%1rVoWeg4K@-6JWrbu!jCHy-Yw<0pYU6PteOn&{Lb%X>VYJG!Ff7tt zM{a5l|BVLmmkq^sWM*$V?a}o#ufMs({$)dOM!;g;Jj6t&<m8Y8SPBJb1g{=CgL)eY?X EFLgnQWifeHusband makes less than 25,000Husband makes more than 25,000Less than $25,000212198More than $25,0003654 \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-1 Data Table.html b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-1 Data Table.html new file mode 100644 index 000000000..5893b0bea --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-1 Data Table.html @@ -0,0 +1,34 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    xy
    -0.50.0
    0.00.2
    1.00.5
    2.00.9
    3.01.0
    3.51.0
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-10 Joint Probability.html b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-10 Joint Probability.html new file mode 100644 index 000000000..b2fd735fc --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-10 Joint Probability.html @@ -0,0 +1 @@ +
    x\y01
    0.4.2
    1.1.3
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-11 Linear Data Table.html b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-11 Linear Data Table.html new file mode 100644 index 000000000..1591544a9 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-11 Linear Data Table.html @@ -0,0 +1,78 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    xy
    -203
    -154
    -105
    -56
    07
    58
    109
    1510
    2011
    2512
    3013
    3514
    4015
    4516
    5017
    5518
    6019
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-2 Symmetric Data.html b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-2 Symmetric Data.html new file mode 100644 index 000000000..45213e4b4 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-2 Symmetric Data.html @@ -0,0 +1,62 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    xy
    -3.00.000
    -2.50.047
    -2.00.169
    -1.50.362
    -1.00.540
    -0.50.728
    0.01.000
    0.50.728
    1.00.540
    1.50.362
    2.00.169
    2.50.047
    3.00.000
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-3 Z and CDF Table.html b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-3 Z and CDF Table.html new file mode 100644 index 000000000..080cb3fe8 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-3 Z and CDF Table.html @@ -0,0 +1,18 @@ + + + + + + + + + + + + + + + + + +
    ZCDF
    00
    zPeak
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-4 Normal Distribution.html b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-4 Normal Distribution.html new file mode 100644 index 000000000..9b14b90a2 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-4 Normal Distribution.html @@ -0,0 +1,42 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    Standard DeviationPercentage
    -3σ0.1%
    -2σ2.1%
    -1σ13.6%
    μ34.1%
    34.1%
    13.6%
    2.1%
    >3σ0.1%
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-5 Table of Pairs.html b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-5 Table of Pairs.html new file mode 100644 index 000000000..474fee68b --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-5 Table of Pairs.html @@ -0,0 +1 @@ +
    outcome123
    1 $(X = 1, Y = 2) = (1, 2)$ $(2, 3)$ $(3, 4)$
    2 $(2, 3)$ $(2, 4)$ $(3, 5)$
    3 $(3, 4)$ $(3, 5)$ $(3, 6)$
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-6 Joint Probability Table.html b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-6 Joint Probability Table.html new file mode 100644 index 000000000..d27928b5e --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-6 Joint Probability Table.html @@ -0,0 +1 @@ +
    X\Y23456
    1 $\mathbb{P}(X=1,Y=2)=\frac{1}{9}$ 0000
    20 $\frac{2}{9}$ $\frac{1}{9}$ 00
    300 $\frac{2}{9}$ $\frac{2}{9}$ $\frac{1}{9}$
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-7 Linear Data Table.html b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-7 Linear Data Table.html new file mode 100644 index 000000000..1e699e76c --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-7 Linear Data Table.html @@ -0,0 +1,34 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    xy
    0.00.0
    0.20.2
    0.40.4
    0.60.6
    0.80.8
    1.01.0
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-8 Data Table.html b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-8 Data Table.html new file mode 100644 index 000000000..b5298d82a --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-8 Data Table.html @@ -0,0 +1,14 @@ + + + + + + + + + + + + + +
    XY
    11
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-9 Linear Data Table.html b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-9 Linear Data Table.html new file mode 100644 index 000000000..c5060018a --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/tables/table-9 Linear Data Table.html @@ -0,0 +1,34 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    xy
    00
    0.20.2
    0.40.4
    0.60.6
    0.80.8
    11
    \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/toc_hierarchies.json b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/toc_hierarchies.json new file mode 100644 index 000000000..9b227f21e --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-probability-theory-lecture-notes/toc_hierarchies.json @@ -0,0 +1,54 @@ +[ + { + "toc_range": [ + 2, + 56 + ], + "scan_range": [ + 2, + 228 + ], + "toc_with_level": "| id | heading | level |\n|----|-----------------------------------------------------------|-------|\n| 1 | ## Chapter 1. Combinatorics 5 | 1 |\n| 2 | 1.1. Counting Principle 5 | 2 |\n| 3 | 1.2. Permutations 6 | 2 |\n| 4 | 1.3. Combinations 7 | 2 |\n| 5 | 1.4. Multinomial Coefficients 9 | 2 |\n| 6 | ## Chapter 2. Axioms of Probability 10 | 1 |\n| 7 | 2.1. Sample Space and Events 10 | 2 |\n| 8 | 2.2. Axioms of Probability 12 | 2 |\n| 9 | 2.3. Equally Likely Outcomes 15 | 2 |\n| 10 | ## Chapter 3. Independence 18 | 1 |\n| 11 | 3.1. Independent Events 18 | 2 |\n| 12 | ## Chapter 4. Conditional Probability and Independence 22 | 1 |\n| 13 | 4.1. Conditional Probabilities 22 | 2 |\n| 14 | 4.2. Bayes's Formula 26 | 2 |\n| 15 | ## Chapter 5. Random Variables 29 | 1 |\n| 16 | 5.1. Random Variables 29 | 2 |\n| 17 | 5.2. Discrete Random Variables 31 | 2 |\n| 18 | 5.3. Expected Value 32 | 2 |\n| 19 | 5.4. The C.D.F. 34 | 2 |\n| 20 | 5.5. Expected Value of Sums of Random Variables 37 | 2 |\n| 21 | 5.6. Expectation of a Function of a Random Variable 39 | 2 |\n| 22 | 5.7. Variance 41 | 2 |\n| 23 | ## Chapter 6. Some Discrete Distributions 43 | 1 |\n| 24 | 6.1. Bernouli and Binomial Random Variables 43 | 2 |\n| 25 | 6.2. The Poisson Distribution 46 | 2 |\n| 26 | 6.3. Other Discrete Distributions 49 | 2 |\n| 27 | ## Chapter 7. Continuous Random Variables 53 | 1 |\n| 28 | 7.1. Intro to continuous R.V 53 | 2 |\n| 29 | 7.2. Expectation and Variance 55 | 2 |\n| 30 | 7.3. The uniform Random Variable 58 | 2 |\n| 31 | 7.4. More practice 59 | 2 |\n| 32 | ## Chapter 8. Normal Distributions 60 | 1 |", + "toc_tree": { + "## Chapter 1. Combinatorics 5": { + "1.1. Counting Principle 5": {}, + "1.2. Permutations 6": {}, + "1.3. Combinations 7": {}, + "1.4. Multinomial Coefficients 9": {} + }, + "## Chapter 2. Axioms of Probability 10": { + "2.1. Sample Space and Events 10": {}, + "2.2. Axioms of Probability 12": {}, + "2.3. Equally Likely Outcomes 15": {} + }, + "## Chapter 3. Independence 18": { + "3.1. Independent Events 18": {} + }, + "## Chapter 4. Conditional Probability and Independence 22": { + "4.1. Conditional Probabilities 22": {}, + "4.2. Bayes's Formula 26": {} + }, + "## Chapter 5. Random Variables 29": { + "5.1. Random Variables 29": {}, + "5.2. Discrete Random Variables 31": {}, + "5.3. Expected Value 32": {}, + "5.4. The C.D.F. 34": {}, + "5.5. Expected Value of Sums of Random Variables 37": {}, + "5.6. Expectation of a Function of a Random Variable 39": {}, + "5.7. Variance 41": {} + }, + "## Chapter 6. Some Discrete Distributions 43": { + "6.1. Bernouli and Binomial Random Variables 43": {}, + "6.2. The Poisson Distribution 46": {}, + "6.3. Other Discrete Distributions 49": {} + }, + "## Chapter 7. Continuous Random Variables 53": { + "7.1. Intro to continuous R.V 53": {}, + "7.2. Expectation and Variance 55": {}, + "7.3. The uniform Random Variable 58": {}, + "7.4. More practice 59": {} + }, + "## Chapter 8. Normal Distributions 60": {} + } + } +] \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-statistical-learning/chunks.json b/apps/api/app/data/demo_documents/stem-statistical-learning/chunks.json new file mode 100644 index 000000000..7f462381b --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-statistical-learning/chunks.json @@ -0,0 +1,6257 @@ +{ + "chunks": [ + { + "chunk_id": "7186ef3a-49f0-5c7d-8289-c7062b796d1b", + "type": "text", + "content": "Statistical Learning\nRajen D. Shah\nr.shah@statslab.cam.ac.uk\nCourse webpage: http://www.statslab.cam.ac.uk/\\~rds37/stat\\_learning.html\nIn this course we will study a selection of important modern statistical methods. This selection is heavily biased towards my own interests, but I hope it will nevertheless give you a flavour of some of the most important recent methodological developments in statistics.\nOver the last 25 years, the sorts of datasets that statisticians have been challenged to study have changed greatly. Where in the past, we were used to datasets with many observations with a few carefully chosen variables, we are now seeing datasets where the number of variables can run into the thousands and greatly exceed the number of observations. For example, with microarray data, we typically have gene expression values measured for several thousands of genes, but only for a few hundred tissue samples. The classical statistical methods are often simply not applicable in these “high-dimensional” situations.\nThe course is divided into 5 chapters (of unequal size). Module I encompasses the first two chapters, and Module II covers the last three. Our first chapter will start by introducing ridge regression, a simple generalisation of ordinary least squares. Our study of this will lead us to some beautiful connections with functional analysis and ultimately one of the most successful and flexible classes of learning algorithms: kernel machines.\nThe second chapter will give an introduction to the Lasso, a method which has been at the centre of much of the developments that have occurred in high-dimensional statistics, and will allow us to perform regression in the seemingly hopeless situation when the number of parameters we are trying to estimate is larger than the number of observations.\nChapter three gives and more in depth study of the Lasso, and some of its extensions. In the fourth chapter we will study graphical modelling and provide an introduction to the exciting field of causal inference. Where the previous chapters consider methods for relating a particular response to a large collection of (explanatory) variables, graphical modelling will give us a way of understanding relationships between the variables themselves. Ultimately we would like to infer causal relationships between variables based on (observational) data. This may seem like a fundamentally impossible task, yet we will show how by developing the graphical modelling framework further, we can begin to answer such causal questions.\nStatistics is not only about developing methods that can predict well in the presence of noise, but also about assessing the uncertainty in our predictions and estimates. In the final chapter we will tackle the problem of how to quantify uncertainty in high-dimensional regression settings.\nBefore we begin the main material of the course, we will briefly review two key classical statistical methods: ordinary least squares and maximum likelihood estimation. This will help to set the scene and provide a warm-up for the modern methods to come later.\nClassical statistics\nOrdinary least squares\nImagine data are available in the form of observations $(Y_{i}, x_{i}) \\in \\mathbb{R} \\times \\mathbb{R}^{p}, i = 1, \\ldots, n$ , and the aim is to infer a simple regression function relating the average value of a response, $Y_{i}$ , and a collection of predictors or variables, $x_{i}$ . This is an example of regression analysis, one of the most important tasks in statistics.\nA linear model for the data assumes that it is generated according to\n$$\nY = X \\beta^ {0} + \\varepsilon , \\tag {0.0.1}\n$$\nwhere $Y \\in R^{n}$ is the vector of responses; $X \\in R^{n \\times p}$ is the predictor matrix (or design matrix) with ith row $x_{i}^{T}$ ; $\\varepsilon \\in R^{n}$ represents random error; and $\\beta^{0} \\in R^{p}$ is the unknown vector of coefficients.\nProvided $p \\ll n$ , a sensible way to estimate $\\beta$ is by ordinary least squares (OLS). This yields an estimator $\\hat{\\beta}^{OLS}$ with\n$$\n\\hat {\\beta} ^ {\\mathrm{OLS}} := \\underset {\\beta \\in \\mathbb {R} ^ {p}} {\\arg \\min} \\| Y - X \\beta \\| _ {2} ^ {2} = (X ^ {T} X) ^ {- 1} X ^ {T} Y, \\tag {0.0.2}\n$$\nprovided $X$ has full column rank.\nUnder the assumptions that (i) $\\mathbb{E}(\\varepsilon_i) = 0$ and (ii) $\\operatorname{Var}(\\varepsilon) = \\sigma^2 I$ , we have that:\n- $\\mathbb{E}_{\\beta^0,\\sigma^2}(\\hat{\\beta}^{\\mathrm{OLS}}) = \\mathbb{E}\\{(X^T X)^{-1}X^T(X\\beta^0 + \\varepsilon)\\} = \\beta^0.$\n- $\\mathrm{Var}_{\\beta^0,\\sigma^2}(\\hat{\\beta}^{\\mathrm{OLS}}) = (X^TX)^{-1}X^TX\\mathrm{Var}(\\varepsilon)X(X^TX)^{-1} = \\sigma^2 (X^TX)^{-1}.$\nThe Gauss–Markov theorem states that OLS is the best linear unbiased estimator in our setting: for any other estimator $\\tilde{\\beta}$ that is linear in Y (so $\\tilde{\\beta} = AY$ for some fixed matrix A), we have\n$$\n\\mathrm{Var} _ {\\beta^ {0}, \\sigma^ {2}} (\\tilde {\\beta}) - \\mathrm{Var} _ {\\beta^ {0}, \\sigma^ {2}} (\\hat {\\beta} ^ {\\mathrm{OLS}})\n$$\nis positive semi-definite.\nMaximum likelihood estimation\nThe method of least squares is just one way to construct as estimator. A more general technique is that of maximum likelihood estimation. Here given data $y \\in R^{n}$ that we take as a realisation of a random variable Y, we specify its density $f(y; \\theta)$ up to some unknown vector of parameters $\\theta \\in \\Theta \\subseteq R^{d}$ , where $\\Theta$ is the parameter space. The likelihood function is a function of $\\theta$ for each fixed y given by\n$$\nL (\\theta) := L (\\theta ; y) = c (y) f (y; \\theta),\n$$\nwhere $c(y)$ is an arbitrary constant of proportionality. The maximum likelihood estimate of $\\theta$ maximises the likelihood, or equivalently it maximises the log-likelihood\n$$\n\\ell (\\theta) := \\ell (\\theta ; y) = \\log f (y; \\theta) + \\log (c (y)).\n$$\nA very useful quantity in the context of maximum likelihood estimation is the Fisher information matrix with jkth $(1 \\leq j, k \\leq d)$ entry\n$$\ni _ {j k} (\\theta) := - \\mathbb {E} _ {\\theta} \\bigg \\{\\frac {\\partial^ {2}}{\\partial \\theta_ {j} \\partial \\theta_ {k}} \\ell (\\theta) \\bigg \\}.\n$$\nIt can be thought of as a measure of how hard it is to estimate $\\theta$ when it is the true parameter value. The Cramér–Rao lower bound states that if $\\tilde{\\theta}$ is an unbiased estimator of $\\theta$ , then under regularity conditions,\n$$\n\\mathrm{Var} _ {\\theta} (\\tilde {\\theta}) - i ^ {- 1} (\\theta)\n$$\nis positive semi-definite.\nA remarkable fact about maximum likelihood estimators (MLEs) is that (under quite general conditions) they are asymptotically normally distributed, asymptotically unbiased and asymptotically achieve the Cramér–Rao lower bound.\nAssume that the Fisher information matrix when there are n observations, $i^{(n)}(\\theta)$ (where we have made the dependence on n explicit) satisfies $i^{(n)}(\\theta)/n \\to I(\\theta)$ for some positive definite matrix I. Then denoting the maximum likelihood estimator of $\\theta$ when there are n observations by $\\hat{\\theta}^{(n)}$ , under regularity conditions, as the number of observations $n \\to \\infty$ we have\n$$\n\\sqrt {n} (\\hat {\\theta} ^ {(n)} - \\theta) \\xrightarrow {d} N _ {d} (0, I ^ {- 1} (\\theta)).\n$$\nReturning to our linear model, if we assume in addition that $\\varepsilon_{i} \\sim N(0, \\sigma^{2})$ , then the log-likelihood for $(\\beta, \\sigma^{2})$ is\n$$\n\\ell (\\beta , \\sigma^ {2}) = - \\frac {n}{2} \\log (\\sigma^ {2}) - \\frac {1}{2 \\sigma^ {2}} \\sum_ {i = 1} ^ {n} (y _ {i} - x _ {i} ^ {T} \\beta) ^ {2}.\n$$\nWe see that the maximum likelihood estimate of $\\beta$ and OLS coincide. It is easy to check that\n$$\ni (\\beta , \\sigma^ {2}) = \\left( \\begin{array}{c c} \\sigma^ {- 2} X ^ {T} X & 0 \\\\ 0 & n \\sigma^ {- 4} / 2 \\end{array} \\right).\n$$\nThe general theory for MLEs would suggest that approximately $\\sqrt{n}(\\hat{\\beta}-\\beta)\\sim N_{p}(0,n\\sigma^{2}(X^{T}X)^{-1})$ ; in fact it is straight-forward to show that this distributional result is exact.", + "path": "notes.pdf", + "metadata": { + "length": 7955, + "summary": "This course covers modern statistical methods for high-dimensional data, where variables exceed observations. It introduces ridge regression, kernel machines, the Lasso for regression with more parameters than observations, graphical modeling for causal inference, and uncertainty quantification. It also reviews classical methods like ordinary least squares and maximum likelihood estimation.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Statistical", + "Learning", + "Rajen", + "Shah", + "shah", + "statslab", + "cam", + "ac", + "uk", + "Course", + "webpage", + "http", + "www", + "rds37", + "stat", + "learning", + "html", + "In", + "study", + "selection", + "important", + "modern", + "statistical", + "methods", + "This", + "heavily", + "biased", + "interests", + "hope", + "give", + "flavour", + "recent", + "methodological", + "developments", + "statistics", + "Over", + "25", + "years", + "sorts", + "datasets", + "statisticians", + "challenged", + "changed", + "greatly", + "Where", + "past", + "observations", + "carefully", + "chosen", + "variables", + "number", + "run", + "thousands", + "exceed", + "For", + "microarray", + "data", + "typically", + "gene", + "expression", + "values", + "measured", + "genes", + "hundred", + "tissue", + "samples", + "The", + "classical", + "simply", + "applicable", + "high", + "dimensional", + "situations", + "divided", + "chapters", + "unequal", + "size", + "Module", + "encompasses", + "II", + "covers", + "Our", + "chapter", + "start", + "introducing", + "ridge", + "regression", + "simple", + "generalisation", + "ordinary", + "squares", + "lead", + "beautiful", + "connections", + "functional", + "analysis", + "ultimately", + "successful", + "flexible", + "classes", + "algorithms", + "kernel", + "machines", + "introduction", + "Lasso", + "method", + "centre", + "occurred", + "perform", + "seemingly", + "hopeless", + "situation", + "parameters", + "estimate", + "larger", + "Chapter", + "depth", + "extensions", + "fourth", + "graphical", + "modelling", + "provide", + "exciting", + "field", + "causal", + "inference", + "previous", + "relating", + "response", + "large", + "collection", + "explanatory", + "understanding", + "relationships", + "Ultimately", + "infer", + "based", + "observational", + "fundamentally", + "impossible", + "task", + "show", + "developing", + "framework", + "begin", + "answer", + "questions", + "Statistics", + "predict", + "presence", + "noise", + "assessing", + "uncertainty", + "predictions", + "estimates", + "final", + "tackle", + "problem", + "quantify", + "settings", + "Before", + "main", + "material", + "briefly", + "review", + "key", + "maximum", + "likelihood", + "estimation", + "set", + "scene", + "warm", + "Classical", + "Ordinary", + "Imagine", + "form", + "mathbb", + "times", + "ldots", + "aim", + "function", + "average", + "predictors", + "tasks", + "linear", + "model", + "assumes", + "generated", + "beta", + "varepsilon", + "tag", + "0.0", + "vector", + "responses", + "predictor", + "matrix", + "design", + "ith", + "row", + "represents", + "random", + "error", + "unknown", + "coefficients", + "Provided", + "ll", + "OLS", + "yields", + "estimator", + "hat", + "mathrm", + "underset", + "arg", + "min", + "provided", + "full", + "column", + "rank", + "Under", + "assumptions", + "ii", + "operatorname", + "Var", + "sigma", + "TX", + "Gauss", + "Markov", + "theorem", + "states", + "unbiased", + "setting", + "tilde", + "AY", + "fixed", + "positive", + "semi", + "definite", + "Maximum", + "construct", + "general", + "technique", + "Here", + "realisation", + "variable", + "density", + "theta", + "Theta", + "subseteq", + "parameter", + "space", + "arbitrary", + "constant", + "proportionality", + "maximises", + "equivalently", + "log", + "ell", + "quantity", + "context", + "Fisher", + "information", + "jkth", + "leq", + "entry", + "bigg", + "frac", + "partial", + "It", + "thought", + "measure", + "hard", + "true", + "Cram", + "Rao", + "lower", + "bound", + "regularity", + "conditions", + "remarkable", + "fact", + "estimators", + "MLEs", + "asymptotically", + "distributed", + "achieve", + "Assume", + "made", + "dependence", + "explicit", + "satisfies", + "Then", + "denoting", + "infty", + "sqrt", + "xrightarrow", + "Returning", + "assume", + "addition", + "sim", + "sum", + "We", + "coincide", + "easy", + "check", + "left", + "array", + "end", + "theory", + "suggest", + "approximately", + "straight", + "forward", + "distributional", + "result", + "exact" + ], + "keywords": [ + "High-dimensional statistics", + "Ridge regression", + "Lasso" + ], + "connect_to": [] + } + }, + { + "chunk_id": "13477b10-0e00-51f8-bfdb-1242f161ec3c", + "type": "text", + "content": "Kernel machines\nLet us revisit the linear model with\n$$\nY _ {i} = x _ {i} ^ {T} \\beta^ {0} + \\varepsilon_ {i}.\n$$\nFor unbiased estimators of $\\beta^{0}$ , their variance gives a way of comparing their quality in terms of squared error loss. For a potentially biased estimator, $\\tilde{\\beta}$ , the relevant quantity is\n$$\n\\begin{array}{l} \\mathbb {E} _ {\\beta^ {0}, \\sigma^ {2}} \\{(\\tilde {\\beta} - \\beta^ {0}) (\\tilde {\\beta} - \\beta^ {0}) ^ {T} \\} = \\mathbb {E} [ \\{\\tilde {\\beta} - \\mathbb {E} (\\tilde {\\beta}) + \\mathbb {E} (\\tilde {\\beta}) - \\beta^ {0} \\} \\{\\tilde {\\beta} - \\mathbb {E} (\\tilde {\\beta}) + \\mathbb {E} (\\tilde {\\beta}) - \\beta^ {0} \\} ^ {T} ] \\\\ = \\mathrm{Var} (\\tilde {\\beta}) + \\{\\mathbb {E} (\\tilde {\\beta} - \\beta^ {0}) \\} \\{\\mathbb {E} (\\tilde {\\beta} - \\beta^ {0}) \\} ^ {T}, \\\\ \\end{array}\n$$\na sum of squared bias and variance terms. A crucial part of the optimality arguments for OLS and MLEs was unbiasedness. Do there exist biased methods whose variance is reduced compared to OLS such that their overall prediction error is lower? Yes!—in fact the use of biased estimators is essential in dealing with settings where the number of parameters to be estimated is large compared to the number of observations. In the first two chapters we’ll explore two important methods for variance reduction based on different forms of penalisation: rather than forming estimators via optimising a least squares or log-likelihood term, we will introduce an additional penalty term that encourages estimates to be shrunk towards 0 in some sense. This will allow us to produce reliable estimators that work well when classical MLEs are infeasible, and in other situations can greatly out-perform the classical approaches.", + "path": "notes.pdf/Chapter 1 Kernel machines", + "metadata": { + "length": 1736, + "summary": "The passage discusses linear models and compares unbiased estimators with biased ones. It shows that the mean squared error of an estimator can be decomposed into variance plus squared bias. While OLS and MLEs are unbiased, biased methods can have lower variance, leading to better prediction error, especially when the number of parameters is large relative to observations. The text introduces penalisation techniques that shrink estimates toward zero, enabling reliable estimation when classical MLEs fail and often outperforming traditional approaches.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Kernel", + "machines", + "Let", + "revisit", + "linear", + "model", + "beta", + "varepsilon", + "For", + "unbiased", + "estimators", + "variance", + "comparing", + "quality", + "terms", + "squared", + "error", + "loss", + "potentially", + "biased", + "estimator", + "tilde", + "relevant", + "quantity", + "begin", + "array", + "mathbb", + "sigma", + "mathrm", + "Var", + "end", + "sum", + "bias", + "crucial", + "part", + "optimality", + "arguments", + "OLS", + "MLEs", + "unbiasedness", + "Do", + "exist", + "methods", + "reduced", + "compared", + "prediction", + "lower", + "Yes", + "fact", + "essential", + "dealing", + "settings", + "number", + "parameters", + "estimated", + "large", + "observations", + "In", + "chapters", + "ll", + "explore", + "important", + "reduction", + "based", + "forms", + "penalisation", + "forming", + "optimising", + "squares", + "log", + "likelihood", + "term", + "introduce", + "additional", + "penalty", + "encourages", + "estimates", + "shrunk", + "sense", + "This", + "produce", + "reliable", + "work", + "classical", + "infeasible", + "situations", + "greatly", + "perform", + "approaches" + ], + "keywords": [ + "biased estimators", + "variance reduction", + "penalisation" + ], + "connect_to": [] + } + }, + { + "chunk_id": "53607235-6e4b-53e8-a948-215f0f6d226f", + "type": "text", + "content": "One way to reduce the variance of $\\hat{\\beta}^{\\mathrm{OLS}}$ is to shrink the estimated coefficients towards 0. Ridge regression [Hoerl and Kennard, 1970] does this by solving the following optimisation problem\n$$\n(\\hat {\\mu} _ {\\lambda} ^ {\\mathrm{R}}, \\hat {\\beta} _ {\\lambda} ^ {\\mathrm{R}}) = \\underset {(\\mu , \\beta) \\in \\mathbb {R} \\times \\mathbb {R} ^ {p}} {\\arg \\min} \\{\\| Y - \\mu \\mathbf {1} - X \\beta \\| _ {2} ^ {2} + \\lambda \\| \\beta \\| _ {2} ^ {2} \\}.\n$$\nHere 1 is an n-vector of 1's. We see that the usual OLS objective is penalised by an additional term proportional to $\\|\\beta\\|_{2}^{2}$ . The parameter $\\lambda \\geq 0$ , which controls the severity of the penalty and therefore the degree of the shrinkage towards 0, is known as a regularisation parameter or tuning parameter. We have explicitly included an intercept term which is not penalised. The reason for this is that were the variables to have their origins shifted so e.g. a variable representing temperature is given in units of Kelvin rather than Celsius, the fitted values would not change. However, $X\\hat{\\beta}$ is not invariant under scale transformations of the variables so it is standard practice to centre each column of X (hence making them orthogonal to the intercept term) and then scale them to have $\\ell_{2}$ -norm $\\sqrt{n}$ .\nIt is straightforward to show that after this standardisation of X, $\\hat{\\mu}_{\\lambda}^{R} = \\bar{Y} := \\sum_{i=1}^{n} Y_{i}/n$ , so we may assume that $\\sum_{i=1}^{n} Y_{i} = 0$ by replacing $Y_{i}$ by $Y_{i} - \\bar{Y}$ and then we can remove $\\mu$ from our objective function. In this case\n$$\n\\hat {\\beta} _ {\\lambda} ^ {\\mathrm{R}} = (X ^ {T} X + \\lambda I) ^ {- 1} X ^ {T} Y.\n$$\nIn this form, we can see how the addition of the $\\lambda I$ term helps to stabilise the estimator. Note that when X does not have full column rank (such as in high-dimensional situations), we can still compute this estimator. On the other hand, when X does have full column rank, we have the following theorem.\nTheorem 1. For $\\lambda$ sufficiently small (depending on $\\beta^{0}$ and $\\sigma^{2}$ ),\n$$\n\\mathbb {E} (\\hat {\\beta} ^ {\\mathrm{OLS}} - \\beta^ {0}) (\\hat {\\beta} ^ {\\mathrm{OLS}} - \\beta^ {0}) ^ {T} - \\mathbb {E} (\\hat {\\beta} _ {\\lambda} ^ {\\mathrm{R}} - \\beta^ {0}) (\\hat {\\beta} _ {\\lambda} ^ {\\mathrm{R}} - \\beta^ {0}) ^ {T}\n$$\nis positive definite.\nProof. First we compute the bias of $\\hat{\\beta}_{\\lambda}^{\\mathrm{R}}$ . We drop the subscript $\\lambda$ and superscript $R$ for convenience.\n$$\n\\begin{array}{l} \\mathbb {E} (\\hat {\\beta}) - \\beta^ {0} = (X ^ {T} X + \\lambda I) ^ {- 1} X ^ {T} X \\beta^ {0} - \\beta^ {0} \\\\ = (X ^ {T} X + \\lambda I) ^ {- 1} (X ^ {T} X + \\lambda I - \\lambda I) \\beta^ {0} - \\beta^ {0} \\\\ = - \\lambda (X ^ {T} X + \\lambda I) ^ {- 1} \\beta^ {0}. \\\\ \\end{array}\n$$\nNow we look at the variance of $\\hat{\\beta}$ .\n$$\n\\begin{array}{l} \\operatorname{Var} (\\hat {\\beta}) = \\mathbb {E} \\{(X ^ {T} X + \\lambda I) ^ {- 1} X ^ {T} \\varepsilon \\} \\{(X ^ {T} X + \\lambda I) ^ {- 1} X ^ {T} \\varepsilon \\} ^ {T} \\\\ = \\sigma^ {2} (X ^ {T} X + \\lambda I) ^ {- 1} X ^ {T} X (X ^ {T} X + \\lambda I) ^ {- 1}. \\\\ \\end{array}\n$$\nThus $\\mathbb{E}(\\hat{\\beta}^{\\mathrm{OLS}} - \\beta^{0})(\\hat{\\beta}^{\\mathrm{OLS}} - \\beta^{0})^{T} - \\mathbb{E}(\\hat{\\beta} -\\beta^{0})(\\hat{\\beta} -\\beta^{0})^{T}$ is equal to\n$$\n\\sigma^ {2} (X ^ {T} X) ^ {- 1} - \\sigma^ {2} (X ^ {T} X + \\lambda I) ^ {- 1} X ^ {T} X (X ^ {T} X + \\lambda I) ^ {- 1} - \\lambda^ {2} (X ^ {T} X + \\lambda I) ^ {- 1} \\beta^ {0} \\beta^ {0 ^ {T}} (X ^ {T} X + \\lambda I) ^ {- 1}.\n$$\nAfter some simplification, we see that this is equal to\n$$\n\\lambda (X ^ {T} X + \\lambda I) ^ {- 1} [ \\sigma^ {2} \\{2 I + \\lambda (X ^ {T} X) ^ {- 1} \\} - \\lambda \\beta^ {0} \\beta^ {0 ^ {T}} ] (X ^ {T} X + \\lambda I) ^ {- 1}.\n$$\nThus $\\mathbb{E}(\\hat{\\beta}^{\\mathrm{OLS}} - \\beta^{0})(\\hat{\\beta}^{\\mathrm{OLS}} - \\beta^{0})^{T} - \\mathbb{E}(\\hat{\\beta} -\\beta^{0})(\\hat{\\beta} -\\beta^{0})^{T}$ is positive definite for $\\lambda >0$ if and only if\n$$\n\\sigma^ {2} \\{2 I + \\lambda (X ^ {T} X) ^ {- 1} \\} - \\lambda \\beta^ {0} \\beta^ {0 T}\n$$\nis positive definite, which is true for $\\lambda > 0$ sufficiently small (we can take $0 < \\lambda < 2\\sigma^2 / \\| \\beta^0 \\|_2^2$ ).\nThe theorem says that $\\hat{\\beta}_{\\lambda}^{\\mathrm{R}}$ outperforms $\\hat{\\beta}^{\\mathrm{OLS}}$ provided $\\lambda$ is chosen appropriately. To be able to use ridge regression effectively, we need a way of selecting a good $\\lambda$ —we will come to this very shortly. What the theorem doesn't really tell us is in what situations we expect ridge regression to perform well. To understand that, we will turn to one of the key matrix decompositions used in statistics, the singular value decomposition (SVD).\n1.1.1 The singular value decomposition and principal components analysis\nThe singular value decomposition (SVD) is a generalisation of an eigendecomposition of a square matrix. We can factorise any $X \\in R^{n \\times p}$ into its SVD\n$$\nX = U D V ^ {T}.\n$$\nHere the $U \\in R^{n \\times n}$ and $V \\in R^{p \\times p}$ are orthogonal matrices and $D \\in R^{n \\times p}$ has $D_{11} \\geq D_{22} \\geq \\cdots \\geq D_{mm} \\geq 0$ where $m := \\min(n, p)$ and all other entries of D are zero. To compute such a decomposition requires $O(np \\min(n, p))$ operations. The rth columns of U and V are known as the rth left and right singular vectors of X respectively, and $D_{rr}$ is the rth singular value.\nWhen n > p, we can replace U by its first p columns and D by its first p rows to produce another version of the SVD (sometimes known as the thin SVD). Then $X = UDV^{T}$ where $U \\in R^{n \\times p}$ has orthonormal columns (but is no longer square) and D is square and diagonal. There is an equivalent version for when p > n.\nLet us take $X \\in \\mathbb{R}^{n \\times p}$ as our matrix of predictors and suppose $n \\geq p$ . Using the (thin) SVD we may write the fitted values from ridge regression as follows.\n$$\n\\begin{array}{l} X \\hat {\\beta} _ {\\lambda} ^ {R} = X (X ^ {T} X + \\lambda I) ^ {- 1} X ^ {T} Y \\\\ = U D V ^ {T} \\left(V D ^ {2} V ^ {T} + \\lambda I\\right) ^ {- 1} V D U ^ {T} Y \\\\ = U D \\left(D ^ {2} + \\lambda I\\right) ^ {- 1} D U ^ {T} Y \\\\ = \\sum_ {j = 1} ^ {p} U _ {j} \\frac {D _ {j j} ^ {2}}{D _ {j j} ^ {2} + \\lambda} U _ {j} ^ {T} Y. \\\\ \\end{array}\n$$\nHere we have used the notation (that we shall use throughout the course) that $U_{j}$ is the jth column of U. For comparison, the fitted values from OLS (when X has full column rank) are\n$$\nX \\hat {\\beta} ^ {\\mathrm{OLS}} = X (X ^ {T} X) ^ {- 1} X ^ {T} Y = U U ^ {T} Y.\n$$\nBoth OLS and ridge regression compute the coordinates of Y with respect to the columns of U. Ridge regression then shrinks these coordinates by the factors $D_{jj}^{2}/(D_{jj}^{2}+\\lambda)$ ; if $D_{jj}$ is small, the amount of shrinkage will be larger.\nTo interpret this further, note that the SVD is intimately connected with Principal Components Analysis (PCA). Consider $v \\in R^{p}$ with $\\|v\\|_{2} = 1$ . Since the columns of X have had their means subtracted, the sample variance of $Xv \\in R^{n}$ , is\n$$\n\\frac {1}{n} v ^ {T} X ^ {T} X v = \\frac {1}{n} v ^ {T} V D ^ {2} V ^ {T} v.\n$$\nWriting $a = V^{T}v$ , so $\\| a\\| _2 = 1$ , we have\n$$\n\\frac {1}{n} v ^ {T} V D ^ {2} V ^ {T} v = \\frac {1}{n} a ^ {T} D ^ {2} a = \\frac {1}{n} \\sum_ {j} a _ {j} ^ {2} D _ {j j} ^ {2} \\leq \\frac {1}{n} D _ {1 1} \\sum_ {j} a _ {j} ^ {2} = \\frac {1}{n} D _ {1 1} ^ {2}.\n$$\nAs $\\|XV_{1}\\|_{2}^{2}/n = D_{11}^{2}/n$ , $V_{1}$ determines the linear combination of the columns of X which has the largest sample variance, when the coefficients of the linear combination are constrained to have $\\ell_{2}$ -norm 1. $XV_{1} = D_{11}U_{1}$ is known as the first principal component of X. Subsequent principal components $D_{22}U_{2}, \\ldots, D_{pp}U_{p}$ have maximum variance $D_{jj}^{2}/n$ , subject to being orthogonal to all earlier ones—see example sheet 1 for details.\nReturning to ridge regression, we see that it shrinks Y most in the smaller principal components of X. Thus it will work well when most of the signal is in the large principal components of X. We now turn to the problem of choosing $\\lambda$ .", + "path": "notes.pdf/Chapter 1 Kernel machines/1.1 Ridge regression", + "metadata": { + "length": 8207, + "summary": "The passage introduces ridge regression as a method to reduce the variance of OLS estimates by shrinking coefficients towards zero via a penalty term λ||β||₂². It explains the optimization problem, the closed-form solution after centering and scaling, and the bias-variance trade-off. A theorem states that for sufficiently small λ, ridge regression outperforms OLS in terms of mean squared error. The singular value decomposition (SVD) is used to show that ridge regression shrinks coordinates of Y by factors D_jj²/(D_jj²+λ), with more shrinkage on smaller principal components. This makes ridge effective when signal is concentrated in large principal components. The passage also discusses the need for selecting λ and the connection to principal components analysis.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "One", + "reduce", + "variance", + "hat", + "beta", + "mathrm", + "OLS", + "shrink", + "estimated", + "coefficients", + "Ridge", + "regression", + "Hoerl", + "Kennard", + "1970", + "solving", + "optimisation", + "problem", + "mu", + "lambda", + "underset", + "mathbb", + "times", + "arg", + "min", + "mathbf", + "Here", + "vector", + "We", + "usual", + "objective", + "penalised", + "additional", + "term", + "proportional", + "The", + "parameter", + "geq", + "controls", + "severity", + "penalty", + "degree", + "shrinkage", + "regularisation", + "tuning", + "explicitly", + "included", + "intercept", + "reason", + "variables", + "origins", + "shifted", + "variable", + "representing", + "temperature", + "units", + "Kelvin", + "Celsius", + "fitted", + "values", + "change", + "However", + "invariant", + "scale", + "transformations", + "standard", + "practice", + "centre", + "column", + "making", + "orthogonal", + "ell", + "norm", + "sqrt", + "It", + "straightforward", + "show", + "standardisation", + "bar", + "sum", + "assume", + "replacing", + "remove", + "function", + "In", + "case", + "form", + "addition", + "helps", + "stabilise", + "estimator", + "Note", + "full", + "rank", + "high", + "dimensional", + "situations", + "compute", + "On", + "hand", + "theorem", + "Theorem", + "For", + "sufficiently", + "small", + "depending", + "sigma", + "positive", + "definite", + "Proof", + "First", + "bias", + "drop", + "subscript", + "superscript", + "convenience", + "begin", + "array", + "end", + "Now", + "operatorname", + "Var", + "varepsilon", + "Thus", + "equal", + "After", + "simplification", + "true", + "outperforms", + "provided", + "chosen", + "appropriately", + "To", + "ridge", + "effectively", + "selecting", + "good", + "shortly", + "What", + "doesn", + "expect", + "perform", + "understand", + "turn", + "key", + "matrix", + "decompositions", + "statistics", + "singular", + "decomposition", + "SVD", + "1.1", + "principal", + "components", + "analysis", + "generalisation", + "eigendecomposition", + "square", + "factorise", + "matrices", + "11", + "22", + "cdots", + "mm", + "entries", + "requires", + "np", + "operations", + "rth", + "columns", + "left", + "vectors", + "rr", + "When", + "replace", + "rows", + "produce", + "version", + "thin", + "Then", + "UDV", + "orthonormal", + "longer", + "diagonal", + "There", + "equivalent", + "Let", + "predictors", + "suppose", + "Using", + "write", + "frac", + "notation", + "jth", + "comparison", + "Both", + "coordinates", + "respect", + "shrinks", + "factors", + "jj", + "amount", + "larger", + "interpret", + "note", + "intimately", + "connected", + "Principal", + "Components", + "Analysis", + "PCA", + "Consider", + "Since", + "means", + "subtracted", + "sample", + "Xv", + "Writing", + "leq", + "As", + "XV", + "determines", + "linear", + "combination", + "largest", + "constrained", + "component", + "Subsequent", + "ldots", + "pp", + "maximum", + "subject", + "earlier", + "sheet", + "details", + "Returning", + "smaller", + "work", + "signal", + "large", + "choosing" + ], + "keywords": [ + "Ridge Regression", + "Shrinkage", + "SVD" + ], + "connect_to": [] + } + }, + { + "chunk_id": "5c39958c-bb3c-5b09-addc-a5233bc70034", + "type": "text", + "content": "Cross-validation is a general technique for selecting a good regression method from among several competing regression methods. We illustrate the principle with ridge regression, where we have a family of regression methods given by different $\\lambda$ values.\nSo far, we have considered the matrix of predictors X as fixed and non-random. However, in many cases, it makes sense to think of it as random. Let us assume that our data are i.i.d. pairs $(x_{i}, Y_{i})$ , $i = 1, \\ldots, n$ . Then ideally, we might want to pick a $\\lambda$ value such that\n$$\n\\mathbb {E} \\{(Y ^ {*} - x ^ {* T} \\hat {\\beta} _ {\\lambda} ^ {\\mathrm{R}} (X, Y)) ^ {2} | X, Y \\} \\tag {1.2.1}\n$$\nis minimised. Here $(x^{*}, Y^{*}) \\in \\mathbb{R}^{p} \\times \\mathbb{R}$ is independent of $(X, Y)$ and has the same distribution as $(x_{1}, Y_{1})$ , and we have made the dependence of $\\hat{\\beta}_{\\lambda}^{R}$ on the training data $(X, Y)$ explicit. This $\\lambda$ is such that conditional on the original training data, it minimises the expected prediction error on a new observation drawn from the same distribution as the training data.\nA less ambitious goal is to find a $\\lambda$ value to minimise the expected prediction error,\n$$\n\\mathbb {E} [ \\mathbb {E} \\{(Y ^ {*} - x ^ {* T} \\hat {\\beta} _ {\\lambda} ^ {\\mathrm{R}} (X, Y)) ^ {2} | X, Y \\} ] \\tag {1.2.2}\n$$\nwhere compared with $(1.2.1)$ , we have taken a further expectation over the training set.\nWe still have no way of computing (1.2.2) directly, but we can attempt to estimate it. The idea of v-fold cross-validation is to split the data into v groups or folds of roughly equal size: $(X^{(1)}, Y^{(1)}), \\ldots, (X^{(v)}, Y^{(v)})$ . Let $(X^{(-k)}, Y^{(-k)})$ be all the data except that in the kth fold. For each $\\lambda$ on a grid of values, we compute $\\hat{\\beta}_{\\lambda}^{\\mathrm{R}}(X^{(-k)}, Y^{(-k)})$ : the ridge regression estimate based on all the data except the kth fold. Writing $\\kappa(i)$ for the fold to which $(x_{i}, Y_{i})$ belongs, we choose the value of $\\lambda$ that minimises\n$$\n\\mathrm{CV} (\\lambda) = \\frac {1}{n} \\sum_ {i = 1} ^ {n} \\left\\{Y _ {i} - x _ {i} ^ {T} \\hat {\\beta} _ {\\lambda} ^ {\\mathrm{R}} \\left(X ^ {(- \\kappa (i))}, Y ^ {(- \\kappa (i))}\\right) \\right\\} ^ {2}. \\tag {1.2.3}\n$$\nWriting $\\lambda_{\\mathrm{CV}}$ for the minimiser, our final estimate of $\\beta^0$ can then be $\\hat{\\beta}_{\\lambda_{\\mathrm{CV}}}^{R}(X,Y)$ .\nNote that for each i,\n$$\n\\mathbb {E} \\left\\{Y _ {i} - x _ {i} ^ {T} \\hat {\\beta} _ {\\lambda} ^ {\\mathrm{R}} \\left(X ^ {(- \\kappa (i))}, Y ^ {(- \\kappa (i))}\\right) \\right\\} ^ {2} = \\mathbb {E} \\left[ \\mathbb {E} \\left\\{Y _ {i} - x _ {i} ^ {T} \\hat {\\beta} _ {\\lambda} ^ {\\mathrm{R}} \\left(X ^ {(- \\kappa (i))}, Y ^ {(- \\kappa (i))}\\right) \\right\\} ^ {2} \\mid X ^ {(- \\kappa (i))}, Y ^ {(- \\kappa (i))} \\right]. \\tag {1.2.4}\n$$\nThis is precisely the expected prediction error in $(1.2.2)$ but with the training data X, Y replaced with a training data set of smaller size. If all the folds have the same size, then $\\mathrm{CV}(\\lambda)$ is an average of n identically distributed quantities, each with expected value as in $(1.2.4)$ . However, the quantities being averaged are not independent as they share the same data.\nThus cross-validation gives a biased estimate of the expected prediction error. The amount of the bias depends on the size of the folds, the case when the v = n giving the least bias—this is known as leave-one-out cross-validation. The quality of the estimate, though, may be worse as the quantities being averaged in $(1.2.3)$ will be highly positively correlated. Typical choices of v are 5 or 10.\nCross-validation aims to allow us to choose the single best $\\lambda$ (or more generally regression procedure); we could instead aim to find the best weighted combination of regression procedures. Returning to our ridge regression example, suppose $\\lambda$ is restricted to a grid of values $\\lambda_{1} > \\lambda_{2} > \\cdots > \\lambda_{L}$ . We can then minimise\n$$\n\\frac {1}{n} \\sum_ {i = 1} ^ {n} \\left\\{Y _ {i} - \\sum_ {l = 1} ^ {L} w _ {l} x _ {i} ^ {T} \\hat {\\beta} _ {\\lambda_ {l}} ^ {\\mathrm{R}} (X ^ {(- \\kappa (i))}, Y ^ {(- \\kappa (i))}) \\right\\} ^ {2}\n$$\nover $w \\in R^{L}$ subject to $w_{l} \\geq 0$ for all l. This is a non-negative least-squares optimisation, for which efficient algorithms are available. This is sometimes known as stacking [Wolpert, 1992, Breiman, 1996] and it can often outperform cross-validation.", + "path": "notes.pdf/Chapter 1 Kernel machines/1.2 v-fold cross-validation", + "metadata": { + "length": 4467, + "summary": "The text discusses cross-validation as a technique for selecting a regression method, using ridge regression as an example. It explains the goal of minimizing expected prediction error for new observations, introduces v-fold cross-validation to estimate this error, and describes how to choose the optimal λ by minimizing the CV(λ) criterion. It also notes that cross-validation provides a biased estimate of prediction error, with bias depending on fold size, and mentions stacking as an alternative that finds optimal weighted combinations of regression procedures.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Cross", + "validation", + "general", + "technique", + "selecting", + "good", + "regression", + "method", + "competing", + "methods", + "We", + "illustrate", + "principle", + "ridge", + "family", + "lambda", + "values", + "So", + "considered", + "matrix", + "predictors", + "fixed", + "random", + "However", + "cases", + "makes", + "sense", + "Let", + "assume", + "data", + "pairs", + "ldots", + "Then", + "ideally", + "pick", + "mathbb", + "hat", + "beta", + "mathrm", + "tag", + "1.2", + "minimised", + "Here", + "times", + "independent", + "distribution", + "made", + "dependence", + "training", + "explicit", + "This", + "conditional", + "original", + "minimises", + "expected", + "prediction", + "error", + "observation", + "drawn", + "ambitious", + "goal", + "find", + "minimise", + "compared", + "expectation", + "set", + "computing", + "directly", + "attempt", + "estimate", + "The", + "idea", + "fold", + "cross", + "split", + "groups", + "folds", + "roughly", + "equal", + "size", + "kth", + "For", + "grid", + "compute", + "based", + "Writing", + "kappa", + "belongs", + "choose", + "CV", + "frac", + "sum", + "left", + "minimiser", + "final", + "Note", + "mid", + "precisely", + "replaced", + "smaller", + "If", + "average", + "identically", + "distributed", + "quantities", + "averaged", + "share", + "Thus", + "biased", + "amount", + "bias", + "depends", + "case", + "giving", + "leave", + "quality", + "worse", + "highly", + "positively", + "correlated", + "Typical", + "choices", + "10", + "aims", + "single", + "generally", + "procedure", + "aim", + "weighted", + "combination", + "procedures", + "Returning", + "suppose", + "restricted", + "cdots", + "subject", + "geq", + "negative", + "squares", + "optimisation", + "efficient", + "algorithms", + "stacking", + "Wolpert", + "1992", + "Breiman", + "1996", + "outperform" + ], + "keywords": [ + "Cross-validation", + "Ridge Regression", + "Model Selection" + ], + "connect_to": [] + } + }, + { + "chunk_id": "d2727c87-7c48-5848-8484-317c63c94852", + "type": "text", + "content": "The fitted values from ridge regression are\n$$\nX (X ^ {T} X + \\lambda I) ^ {- 1} X ^ {T} Y. \\tag {1.3.1}\n$$\nAn alternative way of writing this is suggested by the following\n$$\nX ^ {T} \\left(X X ^ {T} + \\lambda I\\right) = \\left(X ^ {T} X + \\lambda I\\right) X ^ {T}\n$$\n$$\n(X ^ {T} X + \\lambda I) ^ {- 1} X ^ {T} = X ^ {T} (X X ^ {T} + \\lambda I) ^ {- 1}\n$$\n$$\nX (X ^ {T} X + \\lambda I) ^ {- 1} X ^ {T} Y = X X ^ {T} (X X ^ {T} + \\lambda I) ^ {- 1} Y. \\tag {1.3.2}\n$$\nTwo remarks are in order:\n- Note while $X^T X$ is $p \\times p$ , $XX^T$ is $n \\times n$ . Computing fitted values using (1.3.1) would require roughly $O(np^2 + p^3)$ operations. If $p \\gg n$ this could be extremely costly. However, our alternative formulation would only require roughly $O(n^2 p + n^3)$ operations, which could be substantially smaller.\n- We see that the fitted values of ridge regression depend only on inner products $K = XX^{T}$ between observations (note $K_{ij} = x_{i}^{T}x_{j}$ ).\nNow suppose that we believe the signal depends quadratically on the predictors:\n$$\nY _ {i} = x _ {i} ^ {T} \\beta + \\sum_ {k, l} x _ {i k} x _ {i l} \\theta_ {k l} + \\varepsilon_ {i}.\n$$\nWe can still use ridge regression provided we work with an enlarged set of predictors\n$$\nx _ {i 1}, \\ldots , x _ {i p}, x _ {i 1} x _ {i 1}, \\ldots , x _ {i 1} x _ {i p}, x _ {i 2} x _ {i 1}, \\ldots , x _ {i 2} x _ {i p}, \\ldots , x _ {i p} x _ {i p}.\n$$\nThis will give us $O(p^{2})$ predictors. Our new approach to computing fitted values would therefore have complexity $O(n^{2}p^{2} + n^{3})$ , which could be rather costly if p is large.\nHowever, rather than first creating all the additional predictors and then computing the new K matrix, we can attempt to directly compute K. To this end consider\n$$\n\\begin{array}{l} (1 + x _ {i} ^ {T} x _ {j}) ^ {2} = \\left(1 + \\sum_ {k} x _ {i k} x _ {j k}\\right) ^ {2} \\\\ = 1 + 2 \\sum_ {k} x _ {i k} x _ {j k} + \\sum_ {k, l} x _ {i k} x _ {i l} x _ {j k} x _ {j l}. \\\\ \\end{array}\n$$\nObserve this amounts to an inner product between vectors of the form\n$$\n(1, \\sqrt {2} x _ {i 1}, \\dots , \\sqrt {2} x _ {i p}, x _ {i 1} x _ {i 1}, \\dots , x _ {i 1} x _ {i p}, x _ {i 2} x _ {i 1}, \\dots , x _ {i 2} x _ {i p}, \\dots , x _ {i p} x _ {i p}) ^ {T}. \\tag {1.3.3}\n$$\nThus if we set\n$$\nK _ {i j} = (1 + x _ {i} ^ {T} x _ {j}) ^ {2} \\tag {1.3.4}\n$$\nand plug this into the formula for the fitted values, it is exactly as if we had performed ridge regression on an enlarged set of variables given by (1.3.3). Now computing K using (1.3.4) would require only p operations per entry, so $O(n^{2}p)$ operations in total. It thus seems we have improved things by a factor of p using our new approach. This is a nice computational trick, but more importantly for us it serves to illustrate some general points.\n- Since ridge regression only depends on inner products between observations, rather than fitting non-linear models by first mapping the original data $x_{i} \\in \\mathbb{R}^{p}$ to $\\phi(x_{i}) \\in \\mathbb{R}^{d}$ (say) using some feature map $\\phi$ (which could, for example introduce quadratic effects), we can instead try to directly compute $k(x_{i}, x_{j}) = \\langle \\phi(x_{i}), \\phi(x_{j}) \\rangle$ .\n- In fact rather than thinking in terms of feature maps, we can instead try to think about an appropriate measure of similarity $k(x_{i}, x_{j})$ between observations. Modelling in this fashion is sometimes much easier.\nWe will now formalise and extend what we have learnt with this example.", + "path": "notes.pdf/Chapter 1 Kernel machines/1.3 The kernel trick", + "metadata": { + "length": 3495, + "summary": "The passage discusses ridge regression, presenting the fitted values formula and an alternative computation using XX^T instead of X^TX to reduce complexity when p >> n. It introduces a quadratic signal model, showing how to compute the kernel matrix K directly using (1 + x_i^T x_j)^2, which requires O(n^2 p) operations instead of O(n^2 p^2 + n^3). This illustrates the kernel trick: ridge regression depends only on inner products, so non-linear feature maps can be handled by directly computing similarity measures k(x_i, x_j).", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "The", + "fitted", + "values", + "ridge", + "regression", + "lambda", + "tag", + "1.3", + "An", + "alternative", + "writing", + "suggested", + "left", + "Two", + "remarks", + "order", + "Note", + "times", + "XX", + "Computing", + "require", + "roughly", + "np", + "operations", + "If", + "gg", + "extremely", + "costly", + "However", + "formulation", + "substantially", + "smaller", + "We", + "depend", + "products", + "observations", + "note", + "ij", + "Now", + "suppose", + "signal", + "depends", + "quadratically", + "predictors", + "beta", + "sum", + "theta", + "varepsilon", + "provided", + "work", + "enlarged", + "set", + "ldots", + "This", + "give", + "Our", + "approach", + "computing", + "complexity", + "large", + "creating", + "additional", + "matrix", + "attempt", + "directly", + "compute", + "To", + "end", + "begin", + "array", + "Observe", + "amounts", + "product", + "vectors", + "form", + "sqrt", + "dots", + "Thus", + "plug", + "formula", + "performed", + "variables", + "entry", + "total", + "It", + "improved", + "things", + "factor", + "nice", + "computational", + "trick", + "importantly", + "serves", + "illustrate", + "general", + "points", + "Since", + "fitting", + "linear", + "models", + "mapping", + "original", + "data", + "mathbb", + "phi", + "feature", + "map", + "introduce", + "quadratic", + "effects", + "langle", + "rangle", + "In", + "fact", + "thinking", + "terms", + "maps", + "measure", + "similarity", + "Modelling", + "fashion", + "easier", + "formalise", + "extend", + "learnt" + ], + "keywords": [ + "Ridge Regression", + "Kernel Trick", + "Computational Complexity" + ], + "connect_to": [] + } + }, + { + "chunk_id": "2c0ce7d4-68e6-54fe-9486-f543876634f9", + "type": "text", + "content": "We have seen how a model with quadratic effects can be fitted very efficiently by replacing the inner product matrix (known as the Gram matrix) $XX^{T}$ in (1.3.2) with the matrix in (1.3.4). It is then natural to ask what other non-linear models can be fitted efficiently using this sort of approach.\nWe won't answer this question directly, but instead we will try to understand the sorts of similarity measures $k$ that can be represented as inner products between transformations of the original data.\nThat is, we will study the similarity measures $k : X \\times X \\to R$ from the input space X to R for which there exists a feature map $\\phi : X \\to H$ where H is some (real) inner product space with\n$$\nk (x, x ^ {\\prime}) = \\langle \\phi (x), \\phi (x ^ {\\prime}) \\rangle . \\tag {1.4.1}\n$$\nRecall that an inner product space is a real vector space H endowed with a map $\\langle\\cdot,\\cdot\\rangle:H\\times H\\to R$ that obeys the following properties.\n(i) Symmetry: $\\langle u, v \\rangle = \\langle v, u \\rangle$ .\n(ii) Linearity: for $a, b \\in R \\langle au + bw, v \\rangle = a \\langle u, v \\rangle + b \\langle w, v \\rangle$ .\n(iii) Positive-definiteness: $\\langle u, u \\rangle \\geq 0$ with equality if and only if $u = 0$ .\nDefinition 1. A positive definite kernel or more simply a kernel (for brevity) k is a symmetric map $k : X \\times X \\to R$ for which for all $n \\in N$ and all $x_{1}, \\ldots, x_{n} \\in X$ , the matrix K with entries\n$$\nK _ {i j} = k (x _ {i}, x _ {j})\n$$\nis positive semi-definite.\nA kernel is a little like an inner product, but need not be bilinear in general. However, a form of the Cauchy–Schwarz inequality does hold for kernels.\nProposition 2.\n$$\nk (x, x ^ {\\prime}) ^ {2} \\leq k (x, x) k (x ^ {\\prime}, x ^ {\\prime}).\n$$\nProof. The matrix\n$$\n\\left( \\begin{array}{c c} k (x, x) & k (x, x ^ {\\prime}) \\\\ k (x ^ {\\prime}, x) & k (x ^ {\\prime}, x ^ {\\prime}) \\end{array} \\right)\n$$\nmust be positive semi-definite so in particular its determinant must be non-negative. □\nFirst we show that any inner product of feature maps will give rise to a kernel.\nProposition 3. k defined by $k(x, x') = \\langle \\phi(x), \\phi(x') \\rangle$ is a kernel.\nProof. Let $x_{1},\\ldots ,x_{n}\\in \\mathcal{X}$ , $\\alpha_{1},\\ldots ,\\alpha_{n}\\in \\mathbb{R}$ and consider\n$$\n\\begin{array}{l} \\sum_ {i, j} \\alpha_ {i} k (x _ {i}, x _ {j}) \\alpha_ {j} = \\sum_ {i, j} \\alpha_ {i} \\langle \\phi (x _ {i}), \\phi (x _ {j}) \\rangle \\alpha_ {j} \\\\ = \\left\\langle \\sum_ {i} \\alpha_ {i} \\phi (x _ {i}), \\sum_ {j} \\alpha_ {j} \\phi (x _ {j}) \\right\\rangle \\geq 0. \\\\ \\end{array}\n$$\nShowing that every kernel admits a representation of the form (1.4.1) is slightly more involved, and we delay this until after we have studied some examples.", + "path": "notes.pdf/Chapter 1 Kernel machines/1.4 Kernels", + "metadata": { + "length": 2730, + "summary": "The passage discusses how non-linear models can be efficiently fitted by replacing the Gram matrix with a kernel matrix. It introduces the concept of a positive definite kernel as a symmetric map that yields positive semi-definite matrices. It shows that any inner product of feature maps defines a kernel, and notes that the converse also holds. A Cauchy-Schwarz inequality for kernels is proved via the determinant of a 2x2 kernel matrix.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "We", + "model", + "quadratic", + "effects", + "fitted", + "efficiently", + "replacing", + "product", + "matrix", + "Gram", + "XX", + "1.3", + "It", + "natural", + "linear", + "models", + "sort", + "approach", + "won", + "answer", + "question", + "directly", + "understand", + "sorts", + "similarity", + "measures", + "represented", + "products", + "transformations", + "original", + "data", + "That", + "study", + "times", + "input", + "space", + "exists", + "feature", + "map", + "phi", + "real", + "prime", + "langle", + "rangle", + "tag", + "1.4", + "Recall", + "vector", + "endowed", + "cdot", + "obeys", + "properties", + "Symmetry", + "ii", + "Linearity", + "au", + "bw", + "iii", + "Positive", + "definiteness", + "geq", + "equality", + "Definition", + "positive", + "definite", + "kernel", + "simply", + "brevity", + "symmetric", + "ldots", + "entries", + "semi", + "bilinear", + "general", + "However", + "form", + "Cauchy", + "Schwarz", + "inequality", + "hold", + "kernels", + "Proposition", + "leq", + "Proof", + "The", + "left", + "begin", + "array", + "end", + "determinant", + "negative", + "First", + "show", + "maps", + "give", + "rise", + "defined", + "Let", + "mathcal", + "alpha", + "mathbb", + "sum", + "Showing", + "admits", + "representation", + "slightly", + "involved", + "delay", + "studied", + "examples" + ], + "keywords": [ + "kernel", + "positive definite", + "feature map" + ], + "connect_to": [] + } + }, + { + "chunk_id": "34f4b574-f926-5e3b-a1ee-23df1f89771c", + "type": "text", + "content": "Proposition 4. Suppose $k_{1}, k_{2}, \\ldots$ are kernels.\n(i) If $\\alpha_{1},\\alpha_{2}\\geq 0$ then $\\alpha_{1}k_{1} + \\alpha_{2}k_{2}$ is a kernel. If $\\lim_{m\\to \\infty}k_m(x,x') =:k(x,x')$ exists for all $x,x^{\\prime}\\in \\mathcal{X}$ , then $k$ is a kernel.\n(ii) The pointwise product $k = k_{1}k_{2}$ is a kernel.\nLinear kernel. $k(x,x^{\\prime}) = x^{T}x^{\\prime}$\nPolynomial kernel. $k(x, x') = (1 + x^{T}x')^{d}$ . To show this is a kernel, we can simply note that $1 + x^{T}x'$ gives a kernel owing to the fact that 1 is a kernel and (i) of Proposition 4. Next (ii) and induction shows that k as defined above is a kernel.\nGaussian kernel. The highly popular Gaussian kernel is defined by\n$$\nk (x, x ^ {\\prime}) = \\exp \\bigg (- \\frac {\\| x - x ^ {\\prime} \\| _ {2} ^ {2}}{2 \\sigma^ {2}} \\bigg).\n$$\nFor x close to $x'$ it is large whilst for x far from $x'$ the kernel quickly decays towards 0. The additional parameter $\\sigma^{2}$ known as the bandwidth controls the speed of the decay to zero. Note it is less clear how one might find a corresponding feature map and indeed any feature map that represents this must be infinite dimensional.\nTo show that it is a kernel first decompose $\\|x-x'\\|_{2}^{2}=\\|x\\|_{2}^{2}+\\|x'\\|_{2}^{2}-2x^{T}x'$ . Note that by Proposition 3,\n$$\nk _ {1} (x, x ^ {\\prime}) = \\exp \\left(- \\frac {\\| x \\| _ {2} ^ {2}}{2 \\sigma^ {2}}\\right) \\exp \\left(- \\frac {\\| x ^ {\\prime} \\| _ {2} ^ {2}}{2 \\sigma^ {2}}\\right)\n$$\nis a kernel. Next writing\n$$\nk _ {2} (x, x ^ {\\prime}) = \\exp (x ^ {T} x ^ {\\prime} / \\sigma^ {2}) = \\sum_ {r = 0} ^ {\\infty} \\frac {(x ^ {T} x ^ {\\prime} / \\sigma^ {2}) ^ {r}}{r !}\n$$\nand using (i) of Proposition 4 shows that $k_{2}$ is a kernel. Finally observing that $k = k_{1}k_{2}$ and using (ii) shows that the Gaussian kernel is indeed a kernel.\nSobolev kernel. Take X to be $[0,1]$ and let $k(x,x') = \\min(x,x')$ . Note this is the covariance function of Brownian motion so it must be positive definite.\nJaccard similarity kernel. Take X to be the set of all subsets of $\\{1,\\ldots,p\\}$ . For $x, x' \\in X$ with $x \\cup x' \\neq \\emptyset$ define\n$$\nk (x, x ^ {\\prime}) = \\frac {| x \\cap x ^ {\\prime} |}{| x \\cup x ^ {\\prime} |}\n$$\nand if $x \\cup x' = \\emptyset$ then set $k(x, x') = 1$ . Showing that this is a kernel is left to the example sheet.", + "path": "notes.pdf/Chapter 1 Kernel machines/1.4 Kernels/1.4.1 Examples of kernels", + "metadata": { + "length": 2308, + "summary": "The passage defines several kernels: linear, polynomial, Gaussian, Sobolev, and Jaccard similarity. It explains how to prove they are valid kernels using properties like sum, product, and limit of kernels. The Gaussian kernel is shown to be a kernel by decomposing it into a product of two kernels. The Sobolev kernel is the covariance of Brownian motion. The Jaccard kernel is defined for sets.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Proposition", + "Suppose", + "ldots", + "kernels", + "If", + "alpha", + "geq", + "kernel", + "lim", + "infty", + "exists", + "prime", + "mathcal", + "ii", + "The", + "pointwise", + "product", + "Linear", + "Polynomial", + "To", + "show", + "simply", + "note", + "owing", + "fact", + "Next", + "induction", + "shows", + "defined", + "Gaussian", + "highly", + "popular", + "exp", + "bigg", + "frac", + "sigma", + "For", + "close", + "large", + "whilst", + "quickly", + "decays", + "additional", + "parameter", + "bandwidth", + "controls", + "speed", + "decay", + "Note", + "clear", + "find", + "feature", + "map", + "represents", + "infinite", + "dimensional", + "decompose", + "2x", + "left", + "writing", + "sum", + "Finally", + "observing", + "Sobolev", + "Take", + "min", + "covariance", + "function", + "Brownian", + "motion", + "positive", + "definite", + "Jaccard", + "similarity", + "set", + "subsets", + "cup", + "neq", + "emptyset", + "define", + "cap", + "Showing", + "sheet" + ], + "keywords": [ + "kernels", + "Gaussian", + "positive definite" + ], + "connect_to": [] + } + }, + { + "chunk_id": "dfc77b88-af2b-5d7c-9f67-238ba0e82f0a", + "type": "text", + "content": "Theorem 5. For every kernel k there exists a feature map $\\phi$ taking values in some inner product space H such that\n$$\nk (x, x ^ {\\prime}) = \\langle \\phi (x), \\phi (x ^ {\\prime}) \\rangle . \\tag {1.4.2}\n$$\nProof. We will take H to be the vector space of functions of the form\n$$\nf (\\cdot) = \\sum_ {i = 1} ^ {n} \\alpha_ {i} k (\\cdot , x _ {i}), \\tag {1.4.3}\n$$\nwhere $n \\in N$ , $x_{i} \\in X$ and $\\alpha_{i} \\in R$ . Our feature map $\\phi : X \\to H$ will be\n$$\n\\phi (x) = k (\\cdot , x). \\tag {1.4.4}\n$$\nWe now define an inner product on $\\mathcal{H}$ . If $f$ is given by (1.4.3) and\n$$\ng (\\cdot) = \\sum_ {j = 1} ^ {m} \\beta_ {j} k (\\cdot , x _ {j} ^ {\\prime}) \\tag {1.4.5}\n$$\nwe define their inner product to be\n$$\n\\langle f, g \\rangle = \\sum_ {i = 1} ^ {n} \\sum_ {j = 1} ^ {m} \\alpha_ {i} \\beta_ {j} k (x _ {i}, x _ {j} ^ {\\prime}). \\tag {1.4.6}\n$$\nWe need to check this is well-defined as the representations of $f$ and $g$ in (1.4.3) and (1.4.5) need not be unique. To this end, note that\n$$\n\\sum_ {i = 1} ^ {n} \\sum_ {j = 1} ^ {m} \\alpha_ {i} \\beta_ {j} k (x _ {i}, x _ {j} ^ {\\prime}) = \\sum_ {i = 1} ^ {n} \\alpha_ {i} g (x _ {i}) = \\sum_ {j = 1} ^ {m} \\beta_ {j} f (x _ {j} ^ {\\prime}). \\tag {1.4.7}\n$$\nThe first equality shows that the inner product does not depend on the particular expansion of g whilst the second equality shows that it also does not depend on the expansion of f. Thus the inner product is well-defined.\nFirst we check that with $\\phi$ defined as in (1.4.4) we do have relationship (1.4.2). Observe that\n$$\n\\langle k (\\cdot , x), f \\rangle = \\sum_ {i = 1} ^ {n} \\alpha_ {i} k (x _ {i}, x) = f (x), \\tag {1.4.8}\n$$\nso in particular we have\n$$\n\\langle \\phi (x), \\phi (x ^ {\\prime}) \\rangle = \\langle k (\\cdot , x), k (\\cdot , x ^ {\\prime}) \\rangle = k (x, x ^ {\\prime}).\n$$\nIt remains to show that it is indeed an inner product. It is clearly symmetric and $(1.4.7)$ shows linearity. We now need to show positive definiteness.\nFirst note that\n$$\n\\langle f, f \\rangle = \\sum_ {i, j} \\alpha_ {i} k (x _ {i}, x _ {j}) \\alpha_ {j} \\geq 0 \\tag {1.4.9}\n$$\nby positive definiteness of the kernel. Now from (1.4.8),\n$$\nf (x) ^ {2} = (\\langle k (\\cdot , x), f \\rangle) ^ {2}.\n$$\nIf we could use the Cauchy–Schwarz inequality on the right-hand side, we would have\n$$\nf (x) ^ {2} \\leq \\langle k (\\cdot , x), k (\\cdot , x) \\rangle \\langle f, f \\rangle , \\tag {1.4.10}\n$$\nwhich would show that if $\\langle f, f \\rangle = 0$ then necessarily $f = 0$ ; the final property we need to show that $\\langle \\cdot, \\cdot \\rangle$ is an inner product. However, in order to use the traditional Cauchy-Schwarz inequality we need to first know we're dealing with an inner product, which is precisely what we're trying to show!\nAlthough we haven't shown that $\\langle \\cdot, \\cdot \\rangle$ is an inner product, we do have enough information to show that it is itself a kernel. We may then appeal to Proposition 2 to obtain (1.4.10). With this in mind, we argue as follows. Given functions $f_1, \\ldots, f_m$ and coefficients $\\gamma_1, \\ldots, \\gamma_m \\in \\mathbb{R}$ , we have\n$$\n\\sum_ {i, j} \\gamma_ {i} \\langle f _ {i}, f _ {j} \\rangle \\gamma_ {j} = \\left\\langle \\sum_ {i} \\gamma_ {i} f _ {i}, \\sum_ {j} \\gamma_ {j} f _ {j} \\right\\rangle \\geq 0\n$$\nwhere we have used linearity and (1.4.9), showing that it is a kernel.\nTo further discuss the space H we recall some facts from analysis. Any inner product space B is also a normed space: for $f \\in B$ we may define $\\|f\\|_{B}^{2} := \\langle f, f \\rangle_{B}$ . Recall that a Cauchy sequence $(f_{m})_{m=1}^{\\infty}$ in B has $\\|f_{m} - f_{n}\\|_{B} \\to 0$ as $n, m \\to \\infty$ . A normed space where every Cauchy sequence has a limit (in the space) is called complete, and a complete inner product space is called a Hilbert space.\nHilbert spaces may be thought of as the (potentially) infinite-dimensional analogues of finite-dimensional Euclidean spaces. For later use we note that if V is a closed subspace of a Hilbert space B, then any $f \\in B$ has a decomposition $f = u + v$ with $u \\in V$ and\n$$\nv \\in V ^ {\\perp} := \\{v \\in \\mathcal {B}: \\langle v, u \\rangle_ {\\mathcal {B}} = 0 \\text {for all} u \\in V \\}.\n$$\nBy adding the limits of Cauchy sequences to $\\mathcal{H}$ (from Theorem 5) we can make $\\mathcal{H}$ a Hilbert space. Indeed, note that if $(f_m)_{m=1}^{\\infty} \\in \\mathcal{H}$ is Cauchy then since by (1.4.10) we have\n$$\n| f _ {m} (x) - f _ {n} (x) | \\leq \\sqrt {k (x , x)} \\| f _ {m} - f _ {n} \\| _ {\\mathcal {H}},\n$$\nwe may define function $f^{*}: X \\to R$ by $f^{*}(x) = \\lim_{m \\to \\infty} f_{m}(x)$ . We can check that all such $f^{*}$ can be added to H to create a Hilbert space.\nIn fact, the completion of H is a special type of Hilbert space known as a reproducing kernel Hilbert space (RKHS).\nDefinition 2. A Hilbert space $\\mathcal{B}$ of functions $f:\\mathcal{X}\\to \\mathbb{R}$ is a reproducing kernel Hilbert space (RKHS) if for all $x\\in \\mathcal{X}$ , there exists $k_{x}\\in \\mathcal{B}$ such that\n$$\nf (x) = \\left\\langle k _ {x}, f \\right\\rangle \\quad \\text { for all } f \\in \\mathcal {B}.\n$$\nThe function\n$$\nk: \\mathcal {X} \\times \\mathcal {X} \\to \\mathbb {R}\n$$\n$$\n(x, x ^ {\\prime}) \\mapsto \\langle k _ {x}, k _ {x ^ {\\prime}} \\rangle = k _ {x ^ {\\prime}} (x)\n$$\nis known as the reproducing kernel of $\\mathcal{B}$ .\nBy Proposition 3 the reproducing kernel of any RKHS is a (positive definite) kernel, and one may show (as Theorem 5 indicates) that to any kernel k is associated a (unique) RKHS that has reproducing kernel k.", + "path": "notes.pdf/Chapter 1 Kernel machines/1.4 Kernels/1.4.2 Reproducing kernel Hilbert spaces", + "metadata": { + "length": 5521, + "summary": "Theorem 5 proves that for every kernel k, there exists a feature map φ into an inner product space H such that k(x,x') = ⟨φ(x), φ(x')⟩. The proof constructs H as the vector space of functions of the form f(·)=∑α_i k(·,x_i), with φ(x)=k(·,x). An inner product is defined on H, and its well-definedness is shown using the kernel's properties. Positive definiteness is established by noting that ⟨f,f⟩≥0 due to the kernel's positive definiteness, and Cauchy-Schwarz inequality is derived by treating the inner product as a kernel itself. The space H can be completed to a Hilbert space, specifically a reproducing kernel Hilbert space (RKHS), where the reproducing property f(x)=⟨k_x,f⟩ holds. The reproducing kernel of an RKHS is a positive definite kernel, and every kernel corresponds to a unique RKHS.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Theorem", + "For", + "kernel", + "exists", + "feature", + "map", + "phi", + "taking", + "values", + "product", + "space", + "prime", + "langle", + "rangle", + "tag", + "1.4", + "Proof", + "We", + "vector", + "functions", + "form", + "cdot", + "sum", + "alpha", + "Our", + "define", + "mathcal", + "If", + "beta", + "check", + "defined", + "representations", + "unique", + "To", + "end", + "note", + "The", + "equality", + "shows", + "depend", + "expansion", + "whilst", + "Thus", + "First", + "relationship", + "Observe", + "It", + "remains", + "show", + "symmetric", + "linearity", + "positive", + "definiteness", + "geq", + "Now", + "Cauchy", + "Schwarz", + "inequality", + "hand", + "side", + "leq", + "10", + "necessarily", + "final", + "property", + "However", + "order", + "traditional", + "dealing", + "precisely", + "Although", + "haven", + "shown", + "information", + "appeal", + "Proposition", + "obtain", + "With", + "mind", + "argue", + "Given", + "ldots", + "coefficients", + "gamma", + "mathbb", + "left", + "showing", + "discuss", + "recall", + "facts", + "analysis", + "Any", + "normed", + "Recall", + "sequence", + "infty", + "limit", + "called", + "complete", + "Hilbert", + "spaces", + "thought", + "potentially", + "infinite", + "dimensional", + "analogues", + "finite", + "Euclidean", + "closed", + "subspace", + "decomposition", + "perp", + "text", + "By", + "adding", + "limits", + "sequences", + "make", + "Indeed", + "sqrt", + "function", + "lim", + "added", + "create", + "In", + "fact", + "completion", + "special", + "type", + "reproducing", + "RKHS", + "Definition", + "quad", + "times", + "mapsto", + "definite" + ], + "keywords": [ + "kernel", + "feature map", + "RKHS" + ], + "connect_to": [] + } + }, + { + "chunk_id": "c4a536ec-096f-5f97-86df-cafd678903e5", + "type": "text", + "content": "Linear kernel. Here $\\mathcal{H}=\\{f:f(x)=\\beta^{T}x,\\beta\\in\\mathbb{R}^{p}\\}$ and if $f(x)=\\beta^{T}x$ then $\\|f\\|_{H}^{2}=||\\beta||_{2}^{2}$ .\nSobolev kernel. It can be shown that $\\mathcal{H}$ is roughly the space of continuous functions $f:[0,1]\\to \\mathbb{R}$ with $f(0) = 0$ that are differentiable almost everywhere, and for which $\\int_0^1 f'(x)^2 dx < \\infty$ . It contains the class of Lipschitz functions (functions $f:[0,1]\\to \\mathbb{R}$ for which there exists some $L$ with $|f(x) - f(y)|\\leq L|x - y|$ for all $x,y\\in [0,1]$ ) that are 0 at the origin. The norm is\n$$\n\\left(\\int_ {0} ^ {1} f ^ {\\prime} (x) ^ {2} d x\\right) ^ {1 / 2}.\n$$\nThough the construction of the RKHS from a kernel is explicit, it can be challenging to understand precisely the space and the form of the norm.", + "path": "notes.pdf/Chapter 1 Kernel machines/1.4 Kernels/Examples", + "metadata": { + "length": 797, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Linear", + "kernel", + "Here", + "mathcal", + "beta", + "mathbb", + "Sobolev", + "It", + "shown", + "roughly", + "space", + "continuous", + "functions", + "differentiable", + "int", + "dx", + "infty", + "class", + "Lipschitz", + "exists", + "leq", + "origin", + "The", + "norm", + "left", + "prime", + "Though", + "construction", + "RKHS", + "explicit", + "challenging", + "understand", + "precisely", + "form" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a9c40994-0f11-500d-89f0-a65befdedb76", + "type": "text", + "content": "To recap, what we have shown so far is that replacing the matrix $XX^{T}$ in the definition of an algorithm by K derived from a positive definite kernel is essentially equivalent to running the same algorithm on some mapping of the original data, though with the modification that instances of $x_{i}^{T}x_{j}$ become $\\langle\\phi(x_{i}),\\phi(x_{j})\\rangle$ .\nBut what exactly is the optimisation problem we are solving when performing kernel ridge regression? Clearly it is determined by the kernel or equivalently by the RKHS. Note we know that an alternative way of writing the usual ridge regression optimisation is\n$$\n\\underset {f \\in \\mathcal {H}} {\\arg \\min} \\left\\{\\sum_ {i = 1} ^ {n} \\left\\{Y _ {i} - f \\left(x _ {i}\\right) \\right\\} ^ {2} + \\lambda \\| f \\| _ {\\mathcal {H}} ^ {2} \\right\\} \\tag {1.4.11}\n$$\nwhere H is the RKHS corresponding to the linear kernel. The following theorem shows in particular that kernel ridge regression (i.e. ridge regression replacing $XX^{T}$ with K) with kernel k is equivalent to the above with H now being the RKHS corresponding to k.\nTheorem 6 (Representer theorem, [Kimeldorf and Wahba, 1970, Schölkopf et al., 2001]). Let $c: \\mathbb{R}^n \\times \\mathcal{X}^n \\times \\mathbb{R}^n \\to \\mathbb{R}$ be an arbitrary loss function, and let $J: [0, \\infty) \\to \\mathbb{R}$ be strictly increasing. Let $x_1, \\ldots, x_n \\in \\mathcal{X}$ , $Y \\in \\mathbb{R}^n$ . Finally, let $f \\in \\mathcal{H}$ where $\\mathcal{H}$ is an RKHS with reproducing kernel $k$ , and let $K_{ij} = k(x_i, x_j)$ $i, j = 1, \\ldots, n$ . Then $\\hat{f}$ minimises\n$$\nQ _ {1} (f) := c \\left(Y, x _ {1}, \\dots , x _ {n}, f \\left(x _ {1}\\right), \\dots , f \\left(x _ {n}\\right)\\right) + J \\left(\\| f \\| _ {\\mathcal {H}} ^ {2}\\right)\n$$\nover $f\\in \\mathcal{H}$ iff. $\\hat{f} (\\cdot) = \\sum_{i = 1}^{n}\\hat{\\alpha}_{i}k(\\cdot ,x_{i})$ and $\\hat{\\alpha}\\in \\mathbb{R}^n$ minimises $Q_{2}$ over $\\alpha \\in \\mathbb{R}^n$ where\n$$\nQ _ {2} (\\alpha) = c (Y, x _ {1}, \\ldots , x _ {n}, K \\alpha) + J (\\alpha^ {T} K \\alpha).\n$$\nProof. Suppose $\\hat{f}$ minimises $Q_{1}$ . We may write $\\hat{f} = u + v$ where $u\\in V:= \\operatorname {span}\\{k(\\cdot ,x_1),\\ldots ,k(\\cdot ,x_n)\\}$ and $v\\in V^{\\perp}$ . Then\n$$\n\\hat {f} (x _ {i}) = \\langle k (\\cdot , x _ {i}), u + v \\rangle = \\langle k (\\cdot , x _ {i}), u \\rangle = u (x _ {i}).\n$$\nMeanwhile, by Pythagoras' theorem we have $J(\\| \\hat{f}\\|_{\\mathcal{H}}^2) = J(\\| u\\|_{\\mathcal{H}}^2 + \\| v\\|_{\\mathcal{H}}^2) \\geq J(\\| u\\|_{\\mathcal{H}}^2)$ with equality iff. $v = 0$ . Thus by optimality of $\\hat{f}$ , $v = 0$ , so $\\hat{f}(\\cdot) = \\sum_{i=1}^{n} \\alpha_i k(\\cdot, x_i)$ for $\\alpha \\in \\mathbb{R}^n$ . Now observe that if $\\hat{f}$ takes this form, then $\\| \\hat{f}\\|_{\\mathcal{H}}^2 = \\alpha^T K\\alpha$ , so $Q_1(\\hat{f}) = Q_2(\\alpha)$ . Then by optimality of $\\hat{f}$ , we have that $\\alpha$ must minimise $Q_2$ .\nNow suppose $\\hat{\\alpha}$ minimises $Q_{2}$ and $\\hat{f} (\\cdot) = \\sum_{i = 1}^{n}\\hat{\\alpha}_{i}k(\\cdot ,x_{i})$ . Note that $Q_{1}(\\hat{f}) = Q_{2}(\\hat{\\alpha})$ . If $\\tilde{f}\\in \\mathcal{H}$ has $Q_{1}(\\tilde{f})\\leq Q_{1}(\\hat{f})$ , by the argument above, writing $\\tilde{f} = u + v$ with $u\\in V$ , $v\\in V^{\\perp}$ , we know that $Q_{1}(u)\\leq Q_{1}(\\tilde{f})$ . But by optimality of $\\hat{\\alpha}$ we have $Q_{1}(\\hat{f})\\leq Q_{1}(u)$ , so $Q_{1}(\\hat{f}) = Q_{1}(\\tilde{f})$ .\nConsider the result specialised the ridge regression objective. We see that $(1.4.11)$ is essentially equivalent to minimising\n$$\n\\| Y - K \\alpha \\| _ {2} ^ {2} + \\lambda \\alpha^ {T} K \\alpha ,\n$$\nand you may check (see example sheet 1) that the minimiser $\\hat{\\alpha}$ satisfies $K\\hat{\\alpha} = K(K + \\lambda I)^{-1}Y$ . Thus (1.4.11) is indeed an alternative way of expressing kernel ridge regression.\nViewing the result in the opposite direction gives a more “sensational” perspective. If you had set out trying to minimise $Q_{1}$ , it might appear completely hopeless as H could be infinite-dimensional. However, somewhat remarkably we see that this reduces to finding the coefficients $\\hat{\\alpha}_{i}$ which solve the simple(r) optimisation problem $Q_{2}$ .\nThe result also tells us how to form predictions: given a new observation x, our prediction for $f(x)$ is\n$$\n\\hat {f} (x) = \\sum_ {i = 1} ^ {n} \\hat {\\alpha} _ {i} k (x, x _ {i}).\n$$", + "path": "notes.pdf/Chapter 1 Kernel machines/1.4 Kernels/1.4.3 The representer theorem", + "metadata": { + "length": 4320, + "summary": "The passage explains the Representer Theorem, which shows that the minimizer of a regularized empirical risk functional in a Reproducing Kernel Hilbert Space (RKHS) can be expressed as a linear combination of kernel evaluations at the training points. This result is key to kernel ridge regression, where the optimization problem reduces to finding coefficients α that minimize ||Y - Kα||² + λαᵀKα, leading to predictions f̂(x) = Σ αᵢ k(x, xᵢ). The theorem bridges infinite-dimensional optimization to a finite-dimensional problem.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "To", + "recap", + "shown", + "replacing", + "matrix", + "XX", + "definition", + "algorithm", + "derived", + "positive", + "definite", + "kernel", + "essentially", + "equivalent", + "running", + "mapping", + "original", + "data", + "modification", + "instances", + "langle", + "phi", + "rangle", + "But", + "optimisation", + "problem", + "solving", + "performing", + "ridge", + "regression", + "Clearly", + "determined", + "equivalently", + "RKHS", + "Note", + "alternative", + "writing", + "usual", + "underset", + "mathcal", + "arg", + "min", + "left", + "sum", + "lambda", + "tag", + "1.4", + "11", + "linear", + "The", + "theorem", + "shows", + "Theorem", + "Representer", + "Kimeldorf", + "Wahba", + "1970", + "Sch", + "lkopf", + "al", + "2001", + "Let", + "mathbb", + "times", + "arbitrary", + "loss", + "function", + "infty", + "strictly", + "increasing", + "ldots", + "Finally", + "reproducing", + "ij", + "Then", + "hat", + "minimises", + "dots", + "iff", + "cdot", + "alpha", + "Proof", + "Suppose", + "We", + "write", + "operatorname", + "span", + "perp", + "Meanwhile", + "Pythagoras", + "geq", + "equality", + "Thus", + "optimality", + "Now", + "observe", + "takes", + "form", + "minimise", + "suppose", + "If", + "tilde", + "leq", + "argument", + "Consider", + "result", + "specialised", + "objective", + "minimising", + "check", + "sheet", + "minimiser", + "satisfies", + "expressing", + "Viewing", + "opposite", + "direction", + "sensational", + "perspective", + "set", + "completely", + "hopeless", + "infinite", + "dimensional", + "However", + "remarkably", + "reduces", + "finding", + "coefficients", + "solve", + "simple", + "tells", + "predictions", + "observation", + "prediction" + ], + "keywords": [ + "Representer Theorem", + "Kernel Ridge Regression", + "RKHS" + ], + "connect_to": [] + } + }, + { + "chunk_id": "9ffb9480-0872-59d5-8e9c-1c884dfc121c", + "type": "text", + "content": "We have seen how the kernel trick allows us to solve a potentially infinite-dimensional version of ridge regression. This may seem impressive, but ultimately we should judge kernel ridge regression on its statistical properties e.g. predictive performance. Consider a setting where\n$$\nY _ {i} = f ^ {0} (x _ {i}) + \\varepsilon_ {i}, \\qquad \\mathbb {E} (\\varepsilon) = 0, \\mathrm{Var} (\\varepsilon) = \\sigma^ {2} I.\n$$\nWe shall assume that $f^{0} \\in H$ where H is an RKHS with reproducing kernel k. By scaling $\\sigma^{2}$ , we may assume $\\|f^{0}\\|_{H} \\leq 1$ . Let K be the kernel matrix $K_{ij} = k(x_{i}, x_{j})$ with eigenvalues $d_{1} \\geq d_{2} \\geq \\cdots \\geq d_{n} \\geq 0$ . We will see that the predictive performance depends delicately on these eigenvalues.\nLet $\\hat{f}_{\\lambda}$ be the estimated regression function from kernel ridge regression with kernel k:\n$$\n\\hat {f} _ {\\lambda} = \\underset {f \\in \\mathcal {H}} {\\arg \\min} \\left\\{\\sum_ {i = 1} ^ {n} \\{Y _ {i} - f (x _ {i}) \\} ^ {2} + \\lambda \\| f \\| _ {\\mathcal {H}} ^ {2} \\right\\}.\n$$\nTheorem 7. The mean squared prediction error (MSPE) may be bounded above in the following way:\n$$\n\\frac {1}{n} \\mathbb {E} \\left\\{\\sum_ {i = 1} ^ {n} \\left\\{f ^ {0} (x _ {i}) - \\hat {f} _ {\\lambda} (x _ {i}) \\right\\} ^ {2} \\right\\} \\leq \\frac {\\sigma^ {2}}{n} \\sum_ {i = 1} ^ {n} \\frac {d _ {i} ^ {2}}{(d _ {i} + \\lambda) ^ {2}} + \\frac {\\lambda}{4 n} \\tag {1.5.1}\n$$\n$$\n\\leq \\frac {\\sigma^ {2}}{n} \\frac {1}{\\lambda} \\sum_ {i = 1} ^ {n} \\min (d _ {i} / 4, \\lambda) + \\frac {\\lambda}{4 n}.\n$$\nProof. We know from the representer theorem that\n$$\n\\left(\\hat {f} _ {\\lambda} (x _ {1}), \\dots , \\hat {f} _ {\\lambda} (x _ {n})\\right) ^ {T} = K (K + \\lambda I) ^ {- 1} Y.\n$$\nYou will show on the example sheet that\n$$\n\\left(f ^ {0} (x _ {1}), \\ldots , f ^ {0} (x _ {n})\\right) ^ {T} = K \\alpha ,\n$$\nfor some $\\alpha \\in \\mathbb{R}^n$ , and moreover that $\\| f^0\\|_{\\mathcal{H}}^2\\geq \\alpha^T K\\alpha$ . Let the eigendecomposition of $K$ be given by $K = UDU^{T}$ with $D_{ii} = d_i$ and define $\\theta = U^{T}K\\alpha$ . We see that $n$ times the LHS of (1.5.1) is\n$$\n\\begin{array}{l} \\mathbb {E} \\| K (K + \\lambda I) ^ {- 1} (U \\theta + \\varepsilon) - U \\theta \\| _ {2} ^ {2} = \\mathbb {E} \\| D U ^ {T} (U D U ^ {T} + \\lambda I) ^ {- 1} (U \\theta + \\varepsilon) - \\theta \\| _ {2} ^ {2} \\\\ = \\mathbb {E} \\| D (D + \\lambda I) ^ {- 1} (\\theta + U ^ {T} \\varepsilon) - \\theta \\| _ {2} ^ {2} \\\\ = \\| \\{D (D + \\lambda I) ^ {- 1} - I \\} \\theta \\| _ {2} ^ {2} + \\mathbb {E} \\| D (D + \\lambda I) ^ {- 1} U ^ {T} \\varepsilon \\| _ {2} ^ {2}. \\\\ \\end{array}\n$$\nTo compute the second term, we use the ‘trace trick’:\n$$\n\\begin{array}{l} \\mathbb {E} \\| D (D + \\lambda I) ^ {- 1} U ^ {T} \\varepsilon \\| _ {2} ^ {2} = \\mathbb {E} [ \\{D (D + \\lambda I) ^ {- 1} U ^ {T} \\varepsilon \\} ^ {T} D (D + \\lambda I) ^ {- 1} U ^ {T} \\varepsilon ] \\\\ = \\mathbb {E} [ \\mathrm{tr} \\{D (D + \\lambda I) ^ {- 1} U ^ {T} \\varepsilon \\varepsilon^ {T} U D (D + \\lambda I) ^ {- 1} \\} ] \\\\ = \\sigma^ {2} \\mathrm{tr} \\{D (D + \\lambda I) ^ {- 1} D (D + \\lambda I) ^ {- 1} \\} \\\\ = \\sigma^ {2} \\sum_ {i = 1} ^ {n} \\frac {d _ {i} ^ {2}}{(d _ {i} + \\lambda) ^ {2}}. \\\\ \\end{array}\n$$\nFor the first term, we have\n$$\n\\| \\{D (D + \\lambda I) ^ {- 1} - I \\} \\theta \\| _ {2} ^ {2} = \\sum_ {i = 1} ^ {n} \\frac {\\lambda^ {2} \\theta_ {i} ^ {2}}{(d _ {i} + \\lambda) ^ {2}}.\n$$\nNow as $\\theta = DU^{T}\\alpha$ note that $\\theta_{i} = 0$ when $d_{i} = 0$ . Let $D^{+}$ be the diagonal matrix with $i$ th diagonal entry equal to $D_{ii}^{-1}$ if $D_{ii} > 0$ and 0 otherwise. Then\n$$\n\\sum_ {i: d _ {i} > 0} \\frac {\\theta_ {i} ^ {2}}{d _ {i}} = \\| \\sqrt {D ^ {+}} \\theta \\| _ {2} ^ {2} = \\alpha^ {T} K U D ^ {+} U ^ {T} K \\alpha = \\alpha^ {T} U D D ^ {+} D U ^ {T} \\alpha = \\alpha^ {T} K \\alpha \\leq 1.\n$$\nBy Hölder's inequality we have\n$$\n\\sum_ {i = 1} ^ {n} \\frac {\\theta_ {i} ^ {2}}{d _ {i}} \\frac {d _ {i} \\lambda^ {2}}{(d _ {i} + \\lambda) ^ {2}} \\leq \\max _ {i = 1, \\ldots , n} \\frac {d _ {i} \\lambda^ {2}}{(d _ {i} + \\lambda) ^ {2}} \\leq \\lambda / 4,\n$$\nusing the inequality $(a+b)^{2}\\geq4ab$ in the final line. Finally note that\n$$\n\\frac {d _ {i} ^ {2}}{(d _ {i} + \\lambda) ^ {2}} \\leq \\min \\{1, d _ {i} ^ {2} / (4 d _ {i} \\lambda) \\} = \\min (\\lambda , d _ {i} / 4) / \\lambda .\n$$\nTo interpret this result further, it will be helpful to express it in terms of $\\hat{\\mu}_{i} := d_{i}/n$ (the eigenvalues of K/n) and $\\lambda_{n} := \\lambda/n$ . We have\n$$\n\\frac {1}{n} \\mathbb {E} \\bigg \\{\\sum_ {i = 1} ^ {n} \\{f ^ {0} (x _ {i}) - \\hat {f} _ {\\lambda} (x _ {i}) \\} ^ {2} \\bigg \\} \\leq \\frac {\\sigma^ {2}}{\\lambda_ {n}} \\frac {1}{n} \\sum_ {i = 1} ^ {n} \\min (\\hat {\\mu} _ {i} / 4, \\lambda_ {n}) + \\lambda_ {n} / 4 =: \\delta_ {n} (\\lambda_ {n}). \\tag {1.5.2}\n$$\nHere we have treated the $x_{i}$ as fixed, but we could equally well think of them as random. Consider a setup where the $x_{i}$ are i.i.d. and independent of $\\varepsilon$ . If we take a further expectation on the RHS of (1.5.2), our result still holds true (the $\\hat{\\mu}_{i}$ are random in this setting). Ideally we would like to then replace $\\mathbb{E}\\min(\\hat{\\mu}_{i}/4,\\lambda_{n})$ with a quantity more directly related to the kernel k.\nMercer's theorem is helpful in this regard. This guarantees (under some mild conditions) an eigendecomposition for kernels, which are somewhat like infinite-dimensional analogues of symmetric positive semi-definite matrices. Under certain technical conditions, we may write\n$$\nk (x, x ^ {\\prime}) = \\sum_ {j = 1} ^ {\\infty} \\mu_ {j} e _ {j} (x) e _ {j} (x ^ {\\prime})\n$$\nwhere given some density $p(x)$ on X, the eigenfunctions $e_{j}$ and corresponding eigenvalues $\\mu_{j}$ obey the integral equation\n$$\n\\mu_ {j} e _ {j} (x ^ {\\prime}) = \\int_ {\\mathcal {X}} k (x, x ^ {\\prime}) e _ {j} (x) p (x) d x,\n$$\nand the $e_{j}$ form an orthonormal basis of H in the sense that\n$$\n\\int_ {\\mathcal {X}} e _ {i} (x) e _ {j} (x) p (x) d x = \\mathbb {1} _ {\\{i = j \\}}.\n$$\nOne can further show that (ignoring a multiplicative constant)\n$$\n\\mathbb {E} \\bigg (\\frac {1}{n} \\sum_ {i = 1} ^ {n} \\min (\\hat {\\mu} _ {i} / 4, \\lambda_ {n}) \\bigg) \\leq \\frac {1}{n} \\sum_ {i = 1} ^ {\\infty} \\min (\\mu_ {i} / 4, \\lambda_ {n}).\n$$\nWhen $k$ is the Sobolev kernel and $p(x)$ is the uniform density on $[0,1]$ , we find the eigenvalues satisfy\n$$\n\\mu_ {j} / 4 = \\frac {1}{\\pi^ {2} (2 j - 1) ^ {2}}.\n$$\nThus\n$$\n\\begin{array}{l} \\sum_ {i = 1} ^ {\\infty} \\min (\\mu_ {i} / 4, \\lambda_ {n}) \\leq \\frac {\\lambda_ {n}}{2} \\bigg (\\frac {1}{\\sqrt {\\pi^ {2} \\lambda_ {n}}} + 1 \\bigg) + \\frac {1}{\\pi^ {2}} \\int_ {\\{(\\pi^ {2} \\lambda_ {n}) ^ {- 1 / 2} + 1 \\} / 2} ^ {\\infty} \\frac {1}{(2 x - 1) ^ {2}} d x \\\\ = \\sqrt {\\lambda_ {n}} / \\pi + \\lambda_ {n} / 2 = O (\\sqrt {\\lambda_ {n}}) \\\\ \\end{array}\n$$\nas $\\lambda_{n}\\to 0$ . Putting things together, we see that\n$$\n\\mathbb {E} (\\delta_ {n} (\\lambda_ {n})) = O \\bigg (\\frac {\\sigma^ {2}}{n \\lambda_ {n} ^ {1 / 2}} + \\lambda_ {n} \\bigg).\n$$\nThus an optimal $\\lambda_{n}\\sim (\\sigma^{2} / n)^{2 / 3}$ gives an error rate of order $(\\sigma^2 /n)^{2 / 3}$ .\nIn fact, one can show that this is the best error rate one can achieve with any estimator in this problem. More generally Yang et al. [2015] shows that for essentially any RKHS H we have\n$$\n\\inf _ {\\hat {f}} \\sup _ {f ^ {0}: \\| f ^ {0} \\| _ {\\mathcal {H}} \\leq 1} \\mathbb {E} \\bigg \\{\\frac {1}{n} \\sum_ {i = 1} ^ {n} \\{f ^ {0} (x _ {i}) - \\hat {f} (x _ {i}) \\} ^ {2} \\bigg \\} \\geq c \\inf _ {\\lambda_ {n}} \\delta_ {n} (\\lambda_ {n})\n$$\nwhere c > 0 is a constant and $\\hat{f}$ is allowed to range over all (measurable) functions of the data Y, X. The conclusion is that kernel ridge regression is the optimal regression procedure up to a constant factor in terms of MSPE when the true signal $f^{0}$ is from an RKHS.", + "path": "notes.pdf/Chapter 1 Kernel machines/1.5 Kernel ridge regression", + "metadata": { + "length": 7795, + "summary": "The text discusses kernel ridge regression, focusing on its predictive performance. It presents Theorem 7, which bounds the mean squared prediction error (MSPE) in terms of kernel matrix eigenvalues and regularization parameter λ. The proof uses the representer theorem, eigendecomposition, and a trace trick. The bound is expressed as δ_n(λ_n) = σ²/(λ_n) * (1/n) Σ min(μ̂_i/4, λ_n) + λ_n/4. Using Mercer's theorem, the bound is related to kernel eigenvalues μ_j. For the Sobolev kernel, eigenvalues decay as 1/(π²(2j-1)²), leading to an optimal λ_n ~ (σ²/n)^{2/3} and error rate O((σ²/n)^{2/3}). The text concludes that kernel ridge regression is optimal up to a constant factor for MSPE when the true function lies in the RKHS.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "We", + "kernel", + "trick", + "solve", + "potentially", + "infinite", + "dimensional", + "version", + "ridge", + "regression", + "This", + "impressive", + "ultimately", + "judge", + "statistical", + "properties", + "predictive", + "performance", + "Consider", + "setting", + "varepsilon", + "qquad", + "mathbb", + "mathrm", + "Var", + "sigma", + "assume", + "RKHS", + "reproducing", + "By", + "scaling", + "leq", + "Let", + "matrix", + "ij", + "eigenvalues", + "geq", + "cdots", + "depends", + "delicately", + "hat", + "lambda", + "estimated", + "function", + "underset", + "mathcal", + "arg", + "min", + "left", + "sum", + "Theorem", + "The", + "squared", + "prediction", + "error", + "MSPE", + "bounded", + "frac", + "tag", + "1.5", + "Proof", + "representer", + "theorem", + "dots", + "You", + "show", + "sheet", + "ldots", + "alpha", + "eigendecomposition", + "UDU", + "ii", + "define", + "theta", + "times", + "LHS", + "begin", + "array", + "end", + "To", + "compute", + "term", + "trace", + "tr", + "For", + "Now", + "DU", + "note", + "diagonal", + "entry", + "equal", + "Then", + "sqrt", + "lder", + "inequality", + "max", + "geq4ab", + "final", + "line", + "Finally", + "interpret", + "result", + "helpful", + "express", + "terms", + "mu", + "bigg", + "delta", + "Here", + "treated", + "fixed", + "equally", + "random", + "setup", + "independent", + "If", + "expectation", + "RHS", + "holds", + "true", + "Ideally", + "replace", + "quantity", + "directly", + "related", + "Mercer", + "regard", + "guarantees", + "mild", + "conditions", + "kernels", + "analogues", + "symmetric", + "positive", + "semi", + "definite", + "matrices", + "Under", + "technical", + "write", + "prime", + "infty", + "density", + "eigenfunctions", + "obey", + "integral", + "equation", + "int", + "form", + "orthonormal", + "basis", + "sense", + "One", + "ignoring", + "multiplicative", + "constant", + "When", + "Sobolev", + "uniform", + "find", + "satisfy", + "pi", + "Thus", + "Putting", + "things", + "optimal", + "sim", + "rate", + "order", + "In", + "fact", + "achieve", + "estimator", + "problem", + "More", + "generally", + "Yang", + "al", + "2015", + "shows", + "essentially", + "inf", + "allowed", + "range", + "measurable", + "functions", + "data", + "conclusion", + "procedure", + "factor", + "signal" + ], + "keywords": [ + "Kernel Ridge Regression", + "MSPE", + "Eigenvalues" + ], + "connect_to": [] + } + }, + { + "chunk_id": "b8559518-b978-5ef5-8b4e-e481eec6222d", + "type": "text", + "content": "Thus far we have we have only considered applying the kernel trick to ridge regression, which as we have seen has attractive theoretical properties as a regression method. However the kernel trick and the representer theorem are much more generally applicable. In settings where the $Y_{i}$ are not continuous but are in $\\{-1,1\\}$ (e.g. labels for spam and ham, fraud and not fraud etc.), popular approaches include kernel logistic regression and the support vector machine (SVM) [Cortes and Vapnik, 1995].", + "path": "notes.pdf/Chapter 1 Kernel machines/1.6 Other kernel machines", + "metadata": { + "length": 507, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Thus", + "considered", + "applying", + "kernel", + "trick", + "ridge", + "regression", + "attractive", + "theoretical", + "properties", + "method", + "However", + "representer", + "theorem", + "generally", + "applicable", + "In", + "settings", + "continuous", + "labels", + "spam", + "ham", + "fraud", + "popular", + "approaches", + "include", + "logistic", + "support", + "vector", + "machine", + "SVM", + "Cortes", + "Vapnik", + "1995" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a276368e-44ce-5a4e-8744-777a506e79b9", + "type": "image", + "content": "\n$$ Consider first the simple case where the data in the two classes $\\{x_{i}\\}_{i:Y_{i}=1}$ and $\\{x_{i}\\}_{i:Y_{i}=-1}$ are separable by a hyperplane through the origin, so there exists $\\beta\\in R^{p}$ with $\\|\\beta\\|_{2}=1$ such that $Y_{i}\\beta^{T}x_{i}>0$ for all i. Note $\\beta$ would then be a unit normal vector to a plane that separates the two classes.0\n[images/image-1-Support Vector Margin Visualization.jpg]\n", + "path": "images/image-1-Support Vector Margin Visualization.jpg", + "metadata": { + "length": 397, + "summary": "image-1\n- Diagram illustrating the geometric margin in a Support Vector Machine (SVM) classification problem.\n- Two classes of data points are plotted: red circles ($y_i = 1$) and blue crosses ($y_i = -1$).\n- A solid line represents the decision boundary (hyperplane), and dashed lines represent the margins.\n- An arrow labeled $\\beta$ indicates the normal vector to the hyperplane.\n- The distance from a point $x_i$ to the hyperplane is annotated as $\\frac{y_i x_i^T \\beta}{\\|\\beta\\|_2}$.\n- The total margin width between the two dashed lines is labeled $M$.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "file_path": "images/image-1-Support Vector Margin Visualization.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "b05a8426-67d1-5214-8f71-79921aa879c1", + "type": "text", + "content": "$$ Consider first the simple case where the data in the two classes $\\{x_{i}\\}_{i:Y_{i}=1}$ and $\\{x_{i}\\}_{i:Y_{i}=-1}$ are separable by a hyperplane through the origin, so there exists $\\beta\\in R^{p}$ with $\\|\\beta\\|_{2}=1$ such that $Y_{i}\\beta^{T}x_{i}>0$ for all i. Note $\\beta$ would then be a unit normal vector to a plane that separates the two classes.0\n[images/image-1-Support Vector Margin Visualization.jpg]\n\nConsider first the simple case where the data in the two classes $\\{x_{i}\\}_{i:Y_{i}=1}$ and $\\{x_{i}\\}_{i:Y_{i}=-1}$ are separable by a hyperplane through the origin, so there exists $\\beta\\in R^{p}$ with $\\|\\beta\\|_{2}=1$ such that $Y_{i}\\beta^{T}x_{i}>0$ for all i. Note $\\beta$ would then be a unit normal vector to a plane that separates the two classes.\nThere may be an infinite number of planes that separate the classes, in which case it seems sensible to use the plane that maximises the margin between the two classes. Consider therefore the following optimisation problem.\n$$\n\\max _ {\\beta \\in \\mathbb {R} ^ {p}, M \\geq 0} M\n$$\nsubject to $Y_{i}x_{i}^{T}\\beta/\\|\\beta\\|_{2}\\geq M, i=1,\\ldots,n.$\nNote that by normalising $\\beta$ above we need not impose the constraint that $\\|\\beta\\|_{2}=1$ .\nSuppose now that the classes are not separable. One way to handle this is to replace the constraint $Y_{i}x_{i}^{T}\\beta/\\|\\beta\\|_{2}\\geq M$ with a penalty for how far over the margin boundary $x_{i}$ is. This penalty should be zero if $x_{i}$ is on the correct side of the boundary (i.e. when $Y_{i}x_{i}^{T}\\beta/\\|\\beta\\|_{2}\\geq M$ ), and should be equal to the distance over the boundary, $M-Y_{i}x_{i}^{T}\\beta/\\|\\beta\\|_{2}$ otherwise. It will in fact be more convenient to penalise according to $1-Y_{i}x_{i}^{T}\\beta/(\\|\\beta\\|_{2}M)$ in the latter case, which is the distance measured in units of M. This penalty is invariant to $\\beta$ undergoing any positive scaling, so we may set $\\|\\beta\\|_{2}=1/M$ , thus eliminating M from the objective function. Switching $\\max1/\\|\\beta\\|_{2}$ with $\\min\\|\\beta\\|_{2}^{2}$ and adding the penalty we arrive at\n$$\n\\underset {\\beta \\in \\mathbb {R} ^ {p}} {\\arg \\min} \\| \\beta \\| _ {2} ^ {2} + \\lambda \\sum_ {i = 1} ^ {n} (1 - Y _ {i} x _ {i} ^ {T} \\beta) _ {+},\n$$\nwhere $(\\cdot)_{+}$ denotes the positive part. Replacing $\\lambda$ with $1/\\lambda$ we can write the objective in the more familiar-looking form\n$$\n\\underset {\\beta \\in \\mathbb {R} ^ {p}} {\\arg \\min} \\sum_ {i = 1} ^ {n} (1 - Y _ {i} x _ {i} ^ {T} \\beta) _ {+} + \\lambda \\| \\beta \\| _ {2} ^ {2}.\n$$\nThus far we have restricted ourselves to hyperplanes through the origin but we would more generally want to consider any translate of these i.e. any hyperplane. This can be achieved by allowing ourselves to translate the $x_{i}$ by an arbitrary vector b, giving\n$$\n\\underset {\\beta \\in \\mathbb {R} ^ {p}, b \\in \\mathbb {R} ^ {p}} {\\arg \\min} \\sum_ {i = 1} ^ {n} (1 - Y _ {i} (x _ {i} - b) ^ {T} \\beta) _ {+} + \\lambda \\| \\beta \\| _ {2} ^ {2},\n$$\nor equivalently\n$$\n(\\hat {\\mu}, \\hat {\\beta}) = \\underset {(\\mu , \\beta) \\in \\mathbb {R} \\times \\mathbb {R} ^ {p}} {\\arg \\min} \\sum_ {i = 1} ^ {n} \\{1 - Y _ {i} (x _ {i} ^ {T} \\beta + \\mu) \\} _ {+} + \\lambda \\| \\beta \\| _ {2} ^ {2}. \\tag {1.6.1}\n$$\nThis final objective defines the support vector classifier; given a new observation $x$ predictions are obtained via $\\mathrm{sgn}(\\hat{\\mu} + x^T\\hat{\\beta})$ .\nNote that the objective in $(1.6.1)$ may be re-written as\n$$\n(\\hat {\\mu}, \\hat {f}) = \\underset {(\\mu , f) \\in \\mathbb {R} \\times \\mathcal {H}} {\\arg \\min} \\sum_ {i = 1} ^ {n} [ 1 - Y _ {i} \\{f (x _ {i}) + \\mu \\} ] _ {+} + \\lambda \\| f \\| _ {\\mathcal {H}} ^ {2}, \\tag {1.6.2}\n$$\nwhere H is the RKHS corresponding to the linear kernel. The representer theorem (more specifically the variant in question 10 of example sheet 1) shows that $(1.6.2)$ for an arbitrary RKHS with kernel k and kernel matrix K is equivalent to the support vector machine\n$$\n(\\hat {\\mu}, \\hat {\\alpha}) = \\underset {(\\mu , \\alpha) \\in \\mathbb {R} \\times \\mathbb {R} ^ {n}} {\\arg \\min} \\sum_ {i = 1} ^ {n} [ 1 - Y _ {i} \\{K _ {i} ^ {T} \\alpha + \\mu \\} ] _ {+} + \\lambda \\alpha^ {T} K \\alpha .\n$$\nPredictions at a new $x$ are given by\n$$\n\\mathrm{sgn} \\bigg (\\hat {\\mu} + \\sum_ {i = 1} ^ {n} \\hat {\\alpha} _ {i} k (x, x _ {i}) \\bigg).\n$$", + "path": "notes.pdf/Chapter 1 Kernel machines/1.6 Other kernel machines/1.6.1 The support vector machine", + "metadata": { + "length": 4280, + "summary": "The text derives the support vector classifier from the separable hyperplane case to the non-separable case using hinge loss and L2 regularization. It starts with maximizing the margin M subject to constraints, then introduces a penalty for misclassification, leading to the primal objective with hinge loss and ridge penalty. It extends to include an intercept term and generalizes to reproducing kernel Hilbert spaces (RKHS) via the representer theorem, resulting in the kernel SVM formulation with predictions based on kernel evaluations.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Consider", + "simple", + "case", + "data", + "classes", + "separable", + "hyperplane", + "origin", + "exists", + "beta", + "Note", + "unit", + "normal", + "vector", + "plane", + "separates", + "classes.0", + "There", + "infinite", + "number", + "planes", + "separate", + "maximises", + "margin", + "optimisation", + "problem", + "max", + "mathbb", + "geq", + "subject", + "ldots", + "normalising", + "impose", + "constraint", + "Suppose", + "One", + "handle", + "replace", + "penalty", + "boundary", + "This", + "correct", + "side", + "equal", + "distance", + "It", + "fact", + "convenient", + "penalise", + "measured", + "units", + "invariant", + "undergoing", + "positive", + "scaling", + "set", + "eliminating", + "objective", + "function", + "Switching", + "max1", + "min", + "adding", + "arrive", + "underset", + "arg", + "lambda", + "sum", + "cdot", + "denotes", + "part", + "Replacing", + "write", + "familiar", + "form", + "Thus", + "restricted", + "hyperplanes", + "generally", + "translate", + "achieved", + "allowing", + "arbitrary", + "giving", + "equivalently", + "hat", + "mu", + "times", + "tag", + "1.6", + "final", + "defines", + "support", + "classifier", + "observation", + "predictions", + "obtained", + "mathrm", + "sgn", + "written", + "mathcal", + "RKHS", + "linear", + "kernel", + "The", + "representer", + "theorem", + "specifically", + "variant", + "question", + "10", + "sheet", + "shows", + "matrix", + "equivalent", + "machine", + "alpha", + "Predictions", + "bigg" + ], + "keywords": [ + "Support Vector Machine", + "Margin", + "RKHS" + ], + "connect_to": [ + { + "target": "a276368e-44ce-5a4e-8744-777a506e79b9", + "relation": "embeds", + "ref": "[images/image-1-Support Vector Margin Visualization.jpg]", + "position": { + "start": 364, + "end": 420 + } + } + ] + } + }, + { + "chunk_id": "e09d5257-3cef-5dcb-95ac-20c01fbcfe3a", + "type": "text", + "content": "Recall that standard logistic regression may be motivated by assuming\n$$\n\\log \\left(\\frac {\\mathbb {P} (Y _ {i} = 1)}{\\mathbb {P} (Y _ {i} = - 1)}\\right) = x _ {i} ^ {T} \\beta^ {0}\n$$\nand picking $\\hat{\\beta}$ to maximise the log-likelihood. This leads to (see example sheet) the following optimisation problem:\n$$\n\\underset {\\beta \\in \\mathbb {R} ^ {p}} {\\arg \\min} \\sum_ {i = 1} ^ {n} \\log \\{1 + \\exp (- Y _ {i} x _ {i} ^ {T} \\beta) \\}.\n$$\nThe ‘kernelised’ version is given by\n$$\n\\underset {f \\in \\mathcal {H}} {\\arg \\min} \\left\\{\\sum_ {i = 1} ^ {n} \\log [ 1 + \\exp \\{- Y _ {i} f (x _ {i}) \\} ] + \\lambda \\| f \\| _ {\\mathcal {H}} ^ {2} \\right\\},\n$$\nwhere H is an RKHS. As in the case of the SVM, the representer theorem gives a finite-dimensional optimisation that is equivalent to the above.", + "path": "notes.pdf/Chapter 1 Kernel machines/1.6 Other kernel machines/1.6.2 Logistic regression", + "metadata": { + "length": 794, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Recall", + "standard", + "logistic", + "regression", + "motivated", + "assuming", + "log", + "left", + "frac", + "mathbb", + "beta", + "picking", + "hat", + "maximise", + "likelihood", + "This", + "leads", + "sheet", + "optimisation", + "problem", + "underset", + "arg", + "min", + "sum", + "exp", + "The", + "kernelised", + "version", + "mathcal", + "lambda", + "RKHS", + "As", + "case", + "SVM", + "representer", + "theorem", + "finite", + "dimensional", + "equivalent" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "3de58cf2-eaa5-53f6-ab19-2f7a83eb9424", + "type": "text", + "content": "We introduced the kernel trick as a computational device that avoided performing calculations in a high or infinite dimensional feature space and, in the case of kernel ridge regression reduced computation down to forming the $n \\times n$ matrix $K$ and then inverting $K + \\lambda I$ . This can be a huge saving, but when $n$ is very large, this can present serious computational difficulties. Even if $p$ is small, the $O(n^3)$ cost of inverting $K + \\lambda I$ may cause problems. What's worse, the fitted regression function is a sum over $n$ terms:\n$$\n\\hat {f} (\\cdot) = \\sum_ {i = 1} ^ {n} \\hat {\\alpha} _ {i} k (x _ {i}, \\cdot).\n$$\nEven to evaluate a prediction at a single new observation requires $O(n)$ computations unless $\\hat{\\alpha}$ is sparse.\nIn recent years, there has been great interest in speeding up computations for kernel machines. We will discuss one exciting approach based on random feature expansions. Given a kernel k, the key idea is to develop a random map\n$$\n\\hat {\\phi}: \\mathcal {X} \\to \\mathbb {R} ^ {b}\n$$\nwith b small such that $\\mathbb{E}\\{\\hat{\\phi}(x)^{T}\\hat{\\phi}(x')\\}=k(x,x')$ . In a sense we are trying to reverse the kernel trick by approximating the kernel using a random feature map. To increase the quality of the approximation of the kernel, we can consider\n$$\nx \\mapsto \\frac {1}{\\sqrt {L}} (\\hat {\\phi} _ {1} (x), \\ldots , \\hat {\\phi} _ {L} (x)) \\in \\mathbb {R} ^ {L b}\n$$\nwith each $(\\hat{\\phi}_{l}(x))_{l=1}^{L}$ being i.i.d. for each x. Let $\\Phi$ be the matrix with ith row given by $(\\hat{\\phi}_{1}(x_{i}),\\ldots,\\hat{\\phi}_{L}(x_{i}))/\\sqrt{L}$ . We may then run our learning algorithm replacing the initial matrix of predictors X with $\\Phi$ . For example, when performing ridge regression, we can compute\n$$\n(\\Phi^ {T} \\Phi + \\lambda I) ^ {- 1} \\Phi^ {T} Y,\n$$\nwhich would require $O(nL^2 b^2 + L^3 b^3)$ operations: a cost linear in $n$ . Predicting a new observation would cost $O(Lb)$ .\nThe work of Rahimi and Recht [2007] proposes a construction of such a random mapping $\\hat{\\phi}$ for shift-invariant kernels, that is kernels for which there exists a function $h$ with $k(x,x^{\\prime}) = h(x - x^{\\prime})$ for all $x,x^{\\prime}\\in \\mathcal{X} = \\mathbb{R}^{p}$ . A useful property of such kernels is given by Bochner's theorem.\nTheorem 8 (Bochner's theorem). Let $k: \\mathbb{R}^p \\times \\mathbb{R}^p \\to \\mathbb{R}$ be a continuous kernel. Then $k$ is shift-invariant if and only if there exists some $c > 0$ and distribution $F$ on $\\mathbb{R}^p$ such that when $W \\sim F$\n$$\nk (x, x ^ {\\prime}) = c \\mathbb {E} e ^ {i (x - x ^ {\\prime}) ^ {T} W} = c \\mathbb {E} \\cos ((x - x ^ {\\prime}) ^ {T} W).\n$$\nTo make use of this theorem, first observe the following. Let $u \\sim U[-\\pi, \\pi]$ , $x, y \\in R$ . Then\n$$\n2 \\mathbb {E} \\cos (x + u) \\cos (y + u) = 2 \\mathbb {E} \\{(\\cos x \\cos u - \\sin x \\sin u) (\\cos y \\cos u - \\sin y \\sin u) \\}.\n$$\nNow as $u \\stackrel{d}{=} -u$ , $\\mathbb{E}\\cos u\\sin u = \\mathbb{E}\\cos(-u)\\sin(-u) = -\\mathbb{E}\\cos u\\sin u = 0$ . Also of course $\\cos^{2}u + \\sin^{2}u = 1$ so $E\\cos^{2}u = E\\sin^{2}u = 1/2$ . Thus\n$$\n2 \\mathbb {E} \\cos (x + u) \\cos (y + u) = \\cos x \\cos y + \\sin x \\sin y = \\cos (x - y).\n$$\nGiven a shift-invariant kernel $k$ with associated distribution $F$ , suppose $W \\sim F$ and let $u \\sim U[-\\pi, \\pi]$ independently. Define\n$$\n\\hat {\\phi} (x) = \\sqrt {2 c} \\cos (W ^ {T} x + u).\n$$\nThen\n$$\n\\begin{array}{l} \\mathbb {E} \\hat {\\phi} (x) \\hat {\\phi} (x ^ {\\prime}) = 2 c \\mathbb {E} [ \\mathbb {E} \\{\\cos (W ^ {T} x + u) \\cos (W ^ {T} x ^ {\\prime} + u) | W \\} ] \\\\ = c \\mathbb {E} \\cos ((x - x ^ {\\prime}) ^ {T} W) = k (x, x ^ {\\prime}). \\\\ \\end{array}\n$$\nAs a concrete example of this approach, let us take the Gaussian kernel $k(x,x') = \\exp \\{-\\| x - x'\\|_2^2 /(2\\sigma^2)\\}$ . Note that if $W\\sim N(0,\\sigma^{-2}I)$ , it has characteristic function $\\mathbb{E}(e^{it^T W}) = e^{-\\| t\\|_2^2 /(2\\sigma^2)}$ so we may take $\\hat{\\phi} (x) = \\sqrt{2}\\cos (W^T x + u)$ .", + "path": "notes.pdf/Chapter 1 Kernel machines/1.7 Large-scale kernel machines", + "metadata": { + "length": 3986, + "summary": "The passage discusses the computational challenges of kernel methods for large datasets, particularly the O(n^3) cost of inverting the kernel matrix and O(n) prediction cost. It introduces random feature expansions as a solution, approximating shift-invariant kernels via random mappings based on Bochner's theorem. The approach uses random cosine features to achieve linear-time computation in n, with applications to kernel ridge regression.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "We", + "introduced", + "kernel", + "trick", + "computational", + "device", + "avoided", + "performing", + "calculations", + "high", + "infinite", + "dimensional", + "feature", + "space", + "case", + "ridge", + "regression", + "reduced", + "computation", + "forming", + "times", + "matrix", + "inverting", + "lambda", + "This", + "huge", + "saving", + "large", + "present", + "difficulties", + "Even", + "small", + "cost", + "problems", + "What", + "worse", + "fitted", + "function", + "sum", + "terms", + "hat", + "cdot", + "alpha", + "evaluate", + "prediction", + "single", + "observation", + "requires", + "computations", + "sparse", + "In", + "recent", + "years", + "great", + "interest", + "speeding", + "machines", + "discuss", + "exciting", + "approach", + "based", + "random", + "expansions", + "Given", + "key", + "idea", + "develop", + "map", + "phi", + "mathcal", + "mathbb", + "sense", + "reverse", + "approximating", + "To", + "increase", + "quality", + "approximation", + "mapsto", + "frac", + "sqrt", + "ldots", + "Let", + "Phi", + "ith", + "row", + "run", + "learning", + "algorithm", + "replacing", + "initial", + "predictors", + "For", + "compute", + "require", + "nL", + "operations", + "linear", + "Predicting", + "Lb", + "The", + "work", + "Rahimi", + "Recht", + "2007", + "proposes", + "construction", + "mapping", + "shift", + "invariant", + "kernels", + "exists", + "prime", + "property", + "Bochner", + "theorem", + "Theorem", + "continuous", + "Then", + "distribution", + "sim", + "cos", + "make", + "observe", + "pi", + "sin", + "Now", + "stackrel", + "Also", + "Thus", + "suppose", + "independently", + "Define", + "begin", + "array", + "end", + "As", + "concrete", + "Gaussian", + "exp", + "sigma", + "Note", + "characteristic" + ], + "keywords": [ + "kernel trick", + "random features", + "ridge regression" + ], + "connect_to": [] + } + }, + { + "chunk_id": "7cd594d9-d5dc-51c2-9620-5d9823201a9d", + "type": "text", + "content": "", + "path": "notes.pdf/Chapter 2", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "af777472-f092-5f6a-b8b0-353117754573", + "type": "text", + "content": "", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "ed2cb7be-e420-5875-a7df-c42d0216c355", + "type": "text", + "content": "Let us revisit the linear model $Y = X\\beta^{0} + \\varepsilon$ where $\\mathbb{E}(\\varepsilon) = 0$ , $\\operatorname{Var}(\\varepsilon) = \\sigma^{2}I$ . In many modern datasets, there are reasons to believe there are many more variables present than are necessary to explain the response. Let S be the set $S = \\{k : \\beta_{k}^{0} \\neq 0\\}$ and suppose $s := |S| \\ll p$ .\nThe MSPE of OLS is\n$$\n\\begin{array}{l} \\frac {1}{n} \\mathbb {E} \\| X \\beta^ {0} - X \\hat {\\beta} ^ {\\mathrm{OLS}} \\| _ {2} ^ {2} = \\frac {1}{n} \\mathbb {E} \\{(\\beta^ {0} - \\hat {\\beta} ^ {\\mathrm{OLS}}) ^ {T} X ^ {T} X (\\beta^ {0} - \\hat {\\beta} ^ {\\mathrm{OLS}}) \\} \\\\ = \\frac {1}{n} \\mathbb {E} [ \\operatorname{tr} \\left\\{\\left(\\beta^ {0} - \\hat {\\beta} ^ {\\mathrm{OLS}}\\right) \\left(\\beta^ {0} - \\hat {\\beta} ^ {\\mathrm{OLS}}\\right) ^ {T} X ^ {T} X \\right\\} ] \\\\ = \\frac {1}{n} \\mathrm{tr} [ \\mathbb {E} \\{(\\beta^ {0} - \\hat {\\beta} ^ {\\mathrm{OLS}}) (\\beta^ {0} - \\hat {\\beta} ^ {\\mathrm{OLS}}) ^ {T} \\} X ^ {T} X ] \\\\ = \\frac {1}{n} \\mathrm{tr} (\\mathrm{Var} (\\hat {\\beta} ^ {\\mathrm{OLS}}) X ^ {T} X) = \\frac {p}{n} \\sigma^ {2}. \\\\ \\end{array}\n$$\nIf we could identify S and then fit a linear model using just these variables, we'd obtain an MSPE of $\\sigma^{2}s/n$ which could be substantially smaller than $\\sigma^{2}p/n$ . Furthermore, it can be shown that parameter estimates from the reduced model are more accurate. The smaller model would also be easier to interpret.\nWe now briefly review some classical model selection strategies.", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.1 Model selection", + "metadata": { + "length": 1514, + "summary": "The passage revisits the linear model Y = Xβ⁰ + ε with E(ε)=0 and Var(ε)=σ²I. It introduces the concept of sparsity, where the true set S of nonzero coefficients has size s much smaller than the total number of variables p. The mean squared prediction error (MSPE) of ordinary least squares (OLS) is derived as (p/n)σ². If the relevant variables in S were known and used in a reduced model, the MSPE would be (s/n)σ², which can be much smaller. Additionally, parameter estimates from the reduced model are more accurate and easier to interpret. The passage then mentions that classical model selection strategies will be reviewed.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Let", + "revisit", + "linear", + "model", + "beta", + "varepsilon", + "mathbb", + "operatorname", + "Var", + "sigma", + "In", + "modern", + "datasets", + "reasons", + "variables", + "present", + "explain", + "response", + "set", + "neq", + "suppose", + "ll", + "The", + "MSPE", + "OLS", + "begin", + "array", + "frac", + "hat", + "mathrm", + "tr", + "left", + "end", + "If", + "identify", + "fit", + "obtain", + "substantially", + "smaller", + "Furthermore", + "shown", + "parameter", + "estimates", + "reduced", + "accurate", + "easier", + "interpret", + "We", + "briefly", + "review", + "classical", + "selection", + "strategies" + ], + "keywords": [ + "MSPE", + "OLS", + "Model Selection" + ], + "connect_to": [] + } + }, + { + "chunk_id": "5a987c0e-9e46-59fe-853e-2b8ce09a51f9", + "type": "text", + "content": "A natural approach to finding S is to consider all $2^{p}$ possible regression procedures each involving regressing the response on a different sets of explanatory variables $X_{M}$ where M is a subset of $\\{1,\\ldots,p\\}$ . We can then pick the best regression procedure using cross-validation (say). For general design matrices, this involves an exhaustive search over all subsets, so this is not really feasible for p > 50.", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.1 Model selection/Best subset regression", + "metadata": { + "length": 425, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "natural", + "approach", + "finding", + "regression", + "procedures", + "involving", + "regressing", + "response", + "sets", + "explanatory", + "variables", + "subset", + "ldots", + "We", + "pick", + "procedure", + "cross", + "validation", + "For", + "general", + "design", + "matrices", + "involves", + "exhaustive", + "search", + "subsets", + "feasible", + "50" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "566a597b-fba9-5dff-86c3-3c471738ad57", + "type": "text", + "content": "This can be seen as a greedy way of performing best subsets regression. Given a target model size m (the tuning parameter), this works as follows.\n1. Start by fitting an intercept only model.\n2. Add to the current model the predictor variable that reduces the residual sum of squares the most.\n3. Continue step 2 until m predictor variables have been selected.", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.1 Model selection/Forward selection", + "metadata": { + "length": 360, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "This", + "greedy", + "performing", + "subsets", + "regression", + "Given", + "target", + "model", + "size", + "tuning", + "parameter", + "works", + "Start", + "fitting", + "intercept", + "Add", + "current", + "predictor", + "variable", + "reduces", + "residual", + "sum", + "squares", + "Continue", + "step", + "variables", + "selected" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9958a626-5198-5531-be05-c8a208646beb", + "type": "text", + "content": "The Least absolute shrinkage and selection operator (Lasso) [Tibshirani, 1996] estimates $\\beta^0$ by $\\hat{\\beta}_{\\lambda}^{\\mathrm{L}}$ , where $(\\hat{\\mu}^{\\mathrm{L}},\\hat{\\beta}_{\\lambda}^{\\mathrm{L}})$ minimise\n$$\n\\frac {1}{2 n} \\| Y - \\mu \\mathbf {1} - X \\beta \\| _ {2} ^ {2} + \\lambda \\| \\beta \\| _ {1} \\tag {2.2.1}\n$$\nover $(\\mu, \\beta) \\in \\mathbb{R} \\times \\mathbb{R}^p$ . Here $\\| \\beta \\|_1$ is the $\\ell_1$ -norm of $\\beta$ : $\\| \\beta \\|_1 = \\sum_{k=1}^{p} |\\beta_k|$ .\nLike ridge regression, $\\hat{\\beta}_{\\lambda}^{L}$ shrinks the OLS estimate towards the origin, but there is an important difference. The $\\ell_{1}$ penalty can force some of the estimated coefficients to be exactly 0. In this way the Lasso can perform simultaneous variable selection and parameter estimation. As we did with ridge regression, we can centre and scale the X matrix, and also centre Y and thus remove $\\mu$ from the objective. Define\n$$\nQ _ {\\lambda} (\\beta) = \\frac {1}{2 n} \\| Y - X \\beta \\| _ {2} ^ {2} + \\lambda \\| \\beta \\| _ {1}. \\tag {2.2.2}\n$$\nNow the minimiser(s) of $Q_{\\lambda}(\\beta)$ will also be the minimiser(s) of\n$$\n\\| Y - X \\beta \\| _ {2} ^ {2} \\text {subject to} \\| \\beta \\| _ {1} \\leq \\| \\hat {\\beta} _ {\\lambda} ^ {\\mathrm{L}} \\| _ {1}.\n$$\nSimilarly, with the Ridge regression objective, we know that $\\hat{\\beta}_{\\lambda}^{\\mathrm{R}}$ minimises $\\| Y - X\\beta \\|_2^2$ subject to $\\| \\beta \\|_2 \\leq \\| \\hat{\\beta}_{\\lambda}^{\\mathrm{R}} \\|_2$ .\nNow the contours of the OLS objective $\\|Y - X\\beta\\|_{2}^{2}$ are ellipsoids centred at $\\hat{\\beta}^{OLS}$ , while the contours of $\\|\\beta\\|_{2}^{2}$ are spheres centred at the origin, and the contours of $\\|\\beta\\|_{1}$ are ‘diamonds’ centred at 0.\nThe important point to note is that the $\\ell_{1}$ ball $\\{\\beta\\in R^{p}:\\|\\beta\\|_{1}\\leq\\|\\hat{\\beta}_{\\lambda}^{\\mathrm{L}}\\|_{1}\\}$ has corners where some of the components are zero, and it is likely that the OLS contours will intersect the $\\ell_{1}$ ball at such a corner.", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.2 The Lasso estimator", + "metadata": { + "length": 2001, + "summary": "The passage introduces the Least absolute shrinkage and selection operator (Lasso), which estimates coefficients by minimizing a penalized least squares objective with an ℓ1 penalty. This penalty shrinks estimates toward zero and can force some coefficients to exactly zero, enabling simultaneous variable selection and parameter estimation. The Lasso's ℓ1 constraint creates a diamond-shaped feasible region with corners at axes, making it likely that the OLS solution contours intersect at these corners, leading to sparse solutions. This contrasts with ridge regression's ℓ2 penalty, which yields spherical constraints and non-zero shrinkage.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "The", + "Least", + "absolute", + "shrinkage", + "selection", + "operator", + "Lasso", + "Tibshirani", + "1996", + "estimates", + "beta", + "hat", + "lambda", + "mathrm", + "mu", + "minimise", + "frac", + "mathbf", + "tag", + "2.2", + "mathbb", + "times", + "Here", + "ell", + "norm", + "sum", + "Like", + "ridge", + "regression", + "shrinks", + "OLS", + "estimate", + "origin", + "important", + "difference", + "penalty", + "force", + "estimated", + "coefficients", + "In", + "perform", + "simultaneous", + "variable", + "parameter", + "estimation", + "As", + "centre", + "scale", + "matrix", + "remove", + "objective", + "Define", + "Now", + "minimiser", + "text", + "subject", + "leq", + "Similarly", + "Ridge", + "minimises", + "contours", + "ellipsoids", + "centred", + "spheres", + "diamonds", + "point", + "note", + "ball", + "corners", + "components", + "intersect", + "corner" + ], + "keywords": [ + "Lasso", + "ℓ1 penalty", + "variable selection" + ], + "connect_to": [] + } + }, + { + "chunk_id": "42216c6b-4a6c-5920-b88a-5496461581c6", + "type": "text", + "content": "One important difference between the Lasso and Best subsets regression is that the former involves a convex optimisation, which makes it computationally tractable for large-scale problems, in contrast to the latter which typically only feasible for problems with p < 50 or so.\nRecall that a set $A \\subseteq \\mathbb{R}^d$ is convex if\n$$\nx, y \\in A \\Rightarrow (1 - t) x + t y \\in A \\quad \\text { for all } t \\in (0, 1).\n$$\nA function $f:\\mathbb{R}^d\\to \\mathbb{R}$ is convex if\n$$\nf \\big ((1 - t) x + t y \\big) \\leq (1 - t) f (x) + t f (y)\n$$\nfor all $x, y \\in \\mathbb{R}^d$ and $t \\in (0,1)$ .\nProposition 9. (i) Let $f_1, \\ldots, f_m: \\mathbb{R}^d \\to \\mathbb{R}$ be convex functions. Then if $c_1, \\ldots, c_m \\geq 0$ , $c_1f_1 + \\cdots c_mf_m$ is a convex function.\n(ii) If $f: \\mathbb{R}^d \\to \\mathbb{R}$ is twice continuously differentiable then\n(a) $f$ is convex iff. its Hessian $H(x)$ is positive semi-definite for all $x$ ,\n(b) $f$ is strictly convex if $H(x)$ is positive definite for all $x$ .\nLasso solutions need not be unique (e.g. if X has duplicate columns), though for most reasonable design matrices, Lasso solutions will be unique. We will often tacitly assume the Lasso solutions is unique in the statement of our theoretical results. It is however straightforward to show that the Lasso fitted values are unique.\nProposition 10. $X\\hat{\\beta}_{\\lambda}^{\\mathrm{L}}$ is unique.\nProof. Fix $\\lambda$ and suppose $\\hat{\\beta}^{(1)}$ and $\\hat{\\beta}^{(2)}$ are two Lasso solutions giving an optimal objective value of $c^*$ . Now for $t \\in (0,1)$ , by strict convexity of $\\| \\cdot \\|_2^2$ ,\n$$\n\\| Y - t X \\hat {\\beta} ^ {(1)} - (1 - t) X \\hat {\\beta} ^ {(2)} \\| _ {2} ^ {2} \\leq t \\| Y - X \\hat {\\beta} ^ {(1)} \\| _ {2} ^ {2} + (1 - t) \\| Y - X \\hat {\\beta} ^ {(2)} \\| _ {2} ^ {2},\n$$\nwith equality if and only if $X\\hat{\\beta}^{(1)} = X\\hat{\\beta}^{(2)}$ . Since $\\| \\cdot \\|_1$ is also convex, we see that\n$$\n\\begin{array}{l} c ^ {*} \\leq Q _ {\\lambda} (t \\hat {\\beta} ^ {(1)} + (1 - t) \\hat {\\beta} ^ {(2)}) \\\\ = \\| Y - t X \\hat {\\beta} ^ {(1)} - (1 - t) X \\hat {\\beta} ^ {(2)} \\| _ {2} ^ {2} / (2 n) + \\lambda \\| t \\hat {\\beta} ^ {(1)} + (1 - t) \\hat {\\beta} ^ {(2)} \\| _ {1} \\\\ \\leq t \\| Y - X \\hat {\\beta} ^ {(1)} \\| _ {2} ^ {2} / (2 n) + (1 - t) \\| Y - X \\hat {\\beta} ^ {(2)} \\| _ {2} ^ {2} / (2 n) + \\lambda \\| t \\hat {\\beta} ^ {(1)} + (1 - t) \\hat {\\beta} ^ {(2)} \\| _ {1} \\\\ \\leq t \\{\\| Y - X \\hat {\\beta} ^ {(1)} \\| _ {2} ^ {2} / (2 n) + \\lambda \\| \\hat {\\beta} ^ {(1)} \\| _ {1} \\} + (1 - t) \\{\\| Y - X \\hat {\\beta} ^ {(2)} \\| _ {2} ^ {2} / (2 n) + \\lambda \\| \\hat {\\beta} ^ {(2)} \\| _ {1} \\} \\\\ = t Q (\\hat {\\beta} ^ {(1)}) + (1 - t) Q (\\hat {\\beta} ^ {(2)}) \\leq c ^ {*}. \\\\ \\end{array}\n$$\nEquality must prevail throughout this chain of inequalities, so $X\\hat{\\beta}^{(1)} = X\\hat{\\beta}^{(2)}$ .\n□\n2.2.2 Prediction error of the Lasso with no assumptions on the design\nA remarkable property of the Lasso is that even when $p \\gg n$ , it can still perform well in terms of prediction error. Suppose the columns of X have been centred and scaled (as we will always assume from now on unless stated otherwise) and assume the normal linear model (where we have already centred Y),\n$$\nY = X \\beta^ {0} + \\varepsilon - \\bar {\\varepsilon} \\mathbf {1} \\tag {2.2.3}\n$$\nwhere $\\varepsilon \\sim N_{n}(0,\\sigma^{2}I)$ .\nTheorem 11. Let $\\hat{\\beta}$ be the Lasso solution when\n$$\n\\lambda = A \\sigma \\sqrt {\\frac {\\log (p)}{n}}.\n$$\nWith probability at least $1 - 2p^{-(A^2 /2 - 1)}$\n$$\n\\frac {1}{n} \\| X (\\beta^ {0} - \\hat {\\beta}) \\| _ {2} ^ {2} \\leq 4 A \\sigma \\sqrt {\\frac {\\log (p)}{n}} \\| \\beta^ {0} \\| _ {1}.\n$$\nProof. From the definition of $\\hat{\\beta}$ we have\n$$\n\\frac {1}{2 n} \\| Y - X \\hat {\\beta} \\| _ {2} ^ {2} + \\lambda \\| \\hat {\\beta} \\| _ {1} \\leq \\frac {1}{2 n} \\| Y - X \\beta^ {0} \\| _ {2} ^ {2} + \\lambda \\| \\beta^ {0} \\| _ {1}.\n$$\nRearranging,\n$$\n\\frac {1}{2 n} \\| X (\\beta^ {0} - \\hat {\\beta}) \\| _ {2} ^ {2} \\leq \\frac {1}{n} \\varepsilon^ {T} X (\\hat {\\beta} - \\beta^ {0}) + \\lambda \\| \\beta^ {0} \\| _ {1} - \\lambda \\| \\hat {\\beta} \\| _ {1}.\n$$\nNow $|\\varepsilon^T X(\\hat{\\beta} -\\beta^0)|\\leq \\| X^T\\varepsilon \\|_\\infty \\| \\hat{\\beta} -\\beta^0\\| _1$ . Let $\\Omega = \\{\\| X^T\\varepsilon \\|_{\\infty} / n\\leq \\lambda \\}$ . Lemma 15 below shows that $\\mathbb{P}(\\Omega)\\geq 1 - 2p^{-(A^2 /2 - 1)}$ . Working on the event $\\Omega$ , we obtain\n$$\n\\frac {1}{2 n} \\| X (\\beta^ {0} - \\hat {\\beta}) \\| _ {2} ^ {2} \\leq \\lambda \\| \\beta^ {0} - \\hat {\\beta} \\| _ {1} + \\lambda \\| \\beta^ {0} \\| _ {1} - \\lambda \\| \\hat {\\beta} \\| _ {1},\n$$\n$$\n\\frac {1}{n} \\| X (\\beta^ {0} - \\hat {\\beta}) \\| _ {2} ^ {2} \\leq 4 \\lambda \\| \\beta^ {0} \\| _ {1}, \\qquad \\mathrm{bythetriangleinequality.}\n$$", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.2 The Lasso estimator/2.2.1 Convexity and the Lasso", + "metadata": { + "length": 4725, + "summary": "The passage discusses key properties of the Lasso regression method. It contrasts Lasso with Best Subset Regression, noting Lasso's convex optimization makes it computationally tractable for large-scale problems. It defines convex sets and functions, and provides Proposition 9 on convex function properties. The text addresses Lasso solution uniqueness, stating fitted values are unique (Proposition 10) even if coefficient estimates may not be. Section 2.2.2 covers prediction error of Lasso with no design assumptions, presenting Theorem 11 which gives a bound on prediction error with high probability when lambda is chosen appropriately, assuming a normal linear model with centered and scaled columns.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "One", + "important", + "difference", + "Lasso", + "Best", + "subsets", + "regression", + "involves", + "convex", + "optimisation", + "makes", + "computationally", + "tractable", + "large", + "scale", + "problems", + "contrast", + "typically", + "feasible", + "50", + "Recall", + "set", + "subseteq", + "mathbb", + "Rightarrow", + "quad", + "text", + "function", + "big", + "leq", + "Proposition", + "Let", + "ldots", + "functions", + "Then", + "geq", + "1f", + "cdots", + "mf", + "ii", + "If", + "continuously", + "differentiable", + "iff", + "Hessian", + "positive", + "semi", + "definite", + "strictly", + "solutions", + "unique", + "duplicate", + "columns", + "reasonable", + "design", + "matrices", + "We", + "tacitly", + "assume", + "statement", + "theoretical", + "results", + "It", + "straightforward", + "show", + "fitted", + "values", + "10", + "hat", + "beta", + "lambda", + "mathrm", + "Proof", + "Fix", + "suppose", + "giving", + "optimal", + "objective", + "Now", + "strict", + "convexity", + "cdot", + "equality", + "Since", + "begin", + "array", + "end", + "Equality", + "prevail", + "chain", + "inequalities", + "2.2", + "Prediction", + "error", + "assumptions", + "remarkable", + "property", + "gg", + "perform", + "terms", + "prediction", + "Suppose", + "centred", + "scaled", + "stated", + "normal", + "linear", + "model", + "varepsilon", + "bar", + "mathbf", + "tag", + "sim", + "sigma", + "Theorem", + "11", + "solution", + "sqrt", + "frac", + "log", + "With", + "probability", + "2p", + "From", + "definition", + "Rearranging", + "infty", + "Omega", + "Lemma", + "15", + "shows", + "Working", + "event", + "obtain", + "qquad", + "bythetriangleinequality" + ], + "keywords": [ + "Lasso", + "Convexity", + "Prediction Error" + ], + "connect_to": [] + } + }, + { + "chunk_id": "74485141-0115-54d8-bafe-f106a5321e4a", + "type": "image", + "content": "\nProposition 14. Let $(W_{i})_{i=1}^{n}$ be a sequence of independent mean-zero sub-Gaussian random variables with parameters $(\\sigma_{i})_{i=1}^{n}$ and let $\\gamma \\in R^{n}$ . Then $\\gamma^{T}W$ is sub-Gaussian with parameter $\\left(\\sum_{i}\\gamma_{i}^{2}\\sigma_{i}^{2}\\right)^{1/2}$ . We can now prove a more general version of the probability bound required for Theorem 11.0\n[images/image-2-Propositio.jpg]\n", + "path": "images/image-2-Propositio.jpg", + "metadata": { + "length": 413, + "summary": "image-2\nProposition 14. Let $(W_{i})_{i=1}^{n}$ be a sequence of independent mean-zero sub-Gaussian random variables with parameters $(\\sigma_{i})_{i=1}^{n}$ and let $\\gamma \\in R^{n}$ . Then $\\gamma^{T}W$ is sub-Gaussian with parameter $\\left(\\sum_{i}\\gamma_{i}^{2}\\sigma_{i}^{2}\\right)^{1/2}$ . We can now prove a more general version of the probability bound required for Theorem 11.0", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "file_path": "images/image-2-Propositio.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "e5f80a54-9ab4-560a-a894-0b20f9641032", + "type": "image", + "content": "\n$$ # Chapter 3 The Lasso and beyond0\n[images/image-3-$$ # Chapt.jpg]\n", + "path": "images/image-3-$$ # Chapt.jpg", + "metadata": { + "length": 70, + "summary": "image-3\n$$ # Chapter 3 The Lasso and beyond0", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "file_path": "images/image-3-$$ # Chapt.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "94c48f8b-53b4-5787-aa21-2fc330c9f5ce", + "type": "text", + "content": "The proof of Theorem 11 relies on a lower bound for the probability of the event $\\Omega$ . A union bound gives\n$$\n\\begin{array}{l} \\mathbb {P} (\\| X ^ {T} \\varepsilon \\| _ {\\infty} / n > \\lambda) = \\mathbb {P} (\\cup_ {j = 1} ^ {p} | X _ {j} ^ {T} \\varepsilon | / n > \\lambda) \\\\ \\leq \\sum_ {j = 1} ^ {p} \\mathbb {P} (| X _ {j} ^ {T} \\varepsilon | / n > \\lambda). \\\\ \\end{array}\n$$\nNow $X_{j}^{T}\\varepsilon/n \\sim N(0,\\sigma^{2}/n)$ , so if we obtain a bound on the tail probabilities of normal distributions, the argument above will give a bound for $\\mathbb{P}(\\Omega)$ .\nMotivated by the need to bound normal tail probabilities, we will briefly discuss the topic of concentration inequalities that provide such bounds for much wider classes of random variables. Concentration inequalities are vital for the study of many modern algorithms and in our case here, they will reveal that the attractive properties of the Lasso presented in Theorem 11 hold true for a variety of non-normal errors.\nWe begin our discussion with the simplest tail bound, Markov's inequality, which states that given a non-negative random variable $W$ ,\n$$\n\\mathbb {P} (W \\geq t) \\leq \\frac {\\mathbb {E} (W)}{t}.\n$$\nThis immediately implies that given a strictly increasing function $\\varphi: R \\to [0, \\infty)$ and any random variable W,\n$$\n\\mathbb {P} (W \\geq t) = \\mathbb {P} \\{\\varphi (W) \\geq \\varphi (t) \\} \\leq \\frac {\\mathbb {E} (\\varphi (W))}{\\varphi (t)}.\n$$\nApplying this with $\\varphi(t)=e^{\\alpha t}$ ( $\\alpha>0$ ) yields the so-called Chernoff bound:\n$$\n\\mathbb {P} (W \\geq t) \\leq \\inf _ {\\alpha > 0} e ^ {- \\alpha t} \\mathbb {E} e ^ {\\alpha W}.\n$$\nConsider the case when $W \\sim N(0, \\sigma^{2})$ . Recall that\n$$\n\\mathbb {E} e ^ {\\alpha W} = e ^ {\\alpha^ {2} \\sigma^ {2} / 2}. \\tag {2.2.4}\n$$\nThus\n$$\n\\mathbb {P} (W \\geq t) \\leq \\inf _ {\\alpha > 0} e ^ {\\alpha^ {2} \\sigma^ {2} / 2 - \\alpha t} = e ^ {- t ^ {2} / (2 \\sigma^ {2})}.\n$$\nNote that to arrive at this bound, all we required was (an upper bound on) the moment generating function (mgf) of W (2.2.4).\nSub-Gaussian variables\nDefinition 3. We say a random variable $W$ with mean $\\mu = \\mathbb{E}(W)$ is sub-Gaussian if there exists $\\sigma > 0$ such that\n$$\n\\mathbb {E} e ^ {\\alpha (W - \\mu)} \\leq e ^ {\\alpha^ {2} \\sigma^ {2} / 2}\n$$\nfor all $\\alpha \\in \\mathbb{R}$ . We then say that $W$ is sub-Gaussian with parameter $\\sigma$ .\nProposition 12 (Sub-Gaussian tail bound). If W is sub-Gaussian with parameter $\\sigma$ and $\\mathbb{E}(W)=\\mu$ then\n$$\n\\mathbb {P} (W - \\mu \\geq t) \\leq e ^ {- t ^ {2} / (2 \\sigma^ {2})}.\n$$\nAs well as Gaussian random variables, the sub-Gaussian class includes bounded random variables.\nLemma 13 (Hoeffding's lemma). If $W$ is mean-zero and takes values in $[a, b]$ , then $W$ is sub-Gaussian with parameter $(b - a)/2$ .\nThe following proposition shows that analogously to how a linear combination of jointly Gaussian random variables is Gaussian, a linear combination of sub-Gaussian random variables is also sub-Gaussian.\nProposition 14. Let $(W_{i})_{i=1}^{n}$ be a sequence of independent mean-zero sub-Gaussian random variables with parameters $(\\sigma_{i})_{i=1}^{n}$ and let $\\gamma \\in R^{n}$ . Then $\\gamma^{T}W$ is sub-Gaussian with parameter $\\left(\\sum_{i}\\gamma_{i}^{2}\\sigma_{i}^{2}\\right)^{1/2}$ .\nProof.\n$$\n\\begin{array}{l} \\mathbb {E} \\exp \\left(\\alpha \\sum_ {i = 1} ^ {n} \\gamma_ {i} W _ {i}\\right) = \\prod_ {i = 1} ^ {n} \\mathbb {E} \\exp (\\alpha \\gamma_ {i} W _ {i}) \\\\ \\leq \\prod_ {i = 1} ^ {n} \\exp (\\alpha^ {2} \\gamma_ {i} ^ {2} \\sigma_ {i} ^ {2} / 2) \\\\ = \\exp \\left(\\alpha^ {2} \\sum_ {i = 1} ^ {n} \\gamma_ {i} ^ {2} \\sigma_ {i} ^ {2} / 2\\right). \\\\ \\end{array}\n$$\n\nProposition 14. Let $(W_{i})_{i=1}^{n}$ be a sequence of independent mean-zero sub-Gaussian random variables with parameters $(\\sigma_{i})_{i=1}^{n}$ and let $\\gamma \\in R^{n}$ . Then $\\gamma^{T}W$ is sub-Gaussian with parameter $\\left(\\sum_{i}\\gamma_{i}^{2}\\sigma_{i}^{2}\\right)^{1/2}$ . We can now prove a more general version of the probability bound required for Theorem 11.0\n[images/image-2-Propositio.jpg]\n\nWe can now prove a more general version of the probability bound required for Theorem 11.\nLemma 15. Suppose $(\\varepsilon_{i})_{i=1}^{n}$ are independent, mean-zero and sub-Gaussian with common parameter $\\sigma$ . Note that this includes $\\varepsilon \\sim N_{n}(0, \\sigma^{2}I)$ . Let $\\lambda = A\\sigma\\sqrt{\\log(p)/n}$ . Then\n$$\n\\mathbb {P} (\\| X ^ {T} \\varepsilon \\| _ {\\infty} / n \\leq \\lambda) \\geq 1 - 2 p ^ {- (A ^ {2} / 2 - 1)}.\n$$\nProof.\n$$\n\\mathbb {P} (\\| X ^ {T} \\varepsilon \\| _ {\\infty} / n > \\lambda) \\leq \\sum_ {j = 1} ^ {p} \\mathbb {P} (| X _ {j} ^ {T} \\varepsilon | / n > \\lambda).\n$$\nBut $\\pm X_j^T\\varepsilon /n$ are both sub-Gaussian with parameter $(\\sigma^2\\| X_j\\| _2^2 /n^2)^{1 / 2} = \\sigma /\\sqrt{n}$ . Thus the RHS is at most\n$$\n2 p \\exp (- A ^ {2} \\log (p) / 2) = 2 p ^ {1 - A ^ {2} / 2}.\n$$\n\n$$ # Chapter 3 The Lasso and beyond0\n[images/image-3-$$ # Chapt.jpg]", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.2 The Lasso estimator/2.2.3 Concentration inequalities I", + "metadata": { + "length": 4979, + "summary": "The text discusses concentration inequalities, starting with Markov's inequality and the Chernoff bound, then introduces sub-Gaussian random variables. It defines sub-Gaussian variables, provides a tail bound (Proposition 12), and shows that bounded variables are sub-Gaussian (Hoeffding's lemma). Proposition 14 states that linear combinations of independent sub-Gaussian variables remain sub-Gaussian. Lemma 15 applies these results to bound the probability that the Lasso's regularization parameter exceeds a threshold, assuming sub-Gaussian errors, yielding a high-probability bound for the event that the Lasso's condition holds.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "The", + "proof", + "Theorem", + "11", + "relies", + "lower", + "bound", + "probability", + "event", + "Omega", + "union", + "begin", + "array", + "mathbb", + "varepsilon", + "infty", + "lambda", + "cup", + "leq", + "sum", + "end", + "Now", + "sim", + "sigma", + "obtain", + "tail", + "probabilities", + "normal", + "distributions", + "argument", + "give", + "Motivated", + "briefly", + "discuss", + "topic", + "concentration", + "inequalities", + "provide", + "bounds", + "wider", + "classes", + "random", + "variables", + "Concentration", + "vital", + "study", + "modern", + "algorithms", + "case", + "reveal", + "attractive", + "properties", + "Lasso", + "presented", + "hold", + "true", + "variety", + "errors", + "We", + "discussion", + "simplest", + "Markov", + "inequality", + "states", + "negative", + "variable", + "geq", + "frac", + "This", + "immediately", + "implies", + "strictly", + "increasing", + "function", + "varphi", + "Applying", + "alpha", + "yields", + "called", + "Chernoff", + "inf", + "Consider", + "Recall", + "tag", + "2.2", + "Thus", + "Note", + "arrive", + "required", + "upper", + "moment", + "generating", + "mgf", + "Sub", + "Gaussian", + "Definition", + "mu", + "exists", + "parameter", + "Proposition", + "12", + "If", + "As", + "class", + "includes", + "bounded", + "Lemma", + "13", + "Hoeffding", + "lemma", + "takes", + "values", + "proposition", + "shows", + "analogously", + "linear", + "combination", + "jointly", + "14", + "Let", + "sequence", + "independent", + "parameters", + "gamma", + "Then", + "left", + "Proof", + "exp", + "prod", + "prove", + "general", + "version", + "11.0", + "15", + "Suppose", + "common", + "sqrt", + "log", + "But", + "pm", + "RHS", + "Chapter", + "beyond0" + ], + "keywords": [ + "sub-Gaussian", + "concentration inequality", + "Lasso" + ], + "connect_to": [ + { + "target": "74485141-0115-54d8-bafe-f106a5321e4a", + "relation": "embeds", + "ref": "[images/image-2-Propositio.jpg]", + "position": { + "start": 4056, + "end": 4087 + } + }, + { + "target": "e5f80a54-9ab4-560a-a894-0b20f9641032", + "relation": "embeds", + "ref": "[images/image-3-$$ # Chapt.jpg]", + "position": { + "start": 4948, + "end": 4979 + } + } + ] + } + }, + { + "chunk_id": "f0d56da7-07f3-5ce3-af47-cfb010068323", + "type": "text", + "content": "The Lasso and beyond\nIn this chapter, we will study the Lasso in more depth and present more precise theory about its behaviour. We will first study the variable selection properties of the Lasso, which will require some basic results in convex analysis that we present below.", + "path": "notes.pdf/Chapter 3 The Lasso and beyond", + "metadata": { + "length": 276, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "The", + "Lasso", + "In", + "chapter", + "study", + "depth", + "present", + "precise", + "theory", + "behaviour", + "We", + "variable", + "selection", + "properties", + "require", + "basic", + "results", + "convex", + "analysis" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f3a172ee-749e-5509-b8e2-731a4fa70046", + "type": "text", + "content": "", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.1 Some facts from optimisation theory and convex analysis", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "53804534-c935-5aa7-98b9-5bfec2c08005", + "type": "text", + "content": "Definition 4. A vector $v \\in \\mathbb{R}^d$ is a subgradient of a convex function $f: \\mathbb{R}^d \\to \\mathbb{R}$ at $x$ if\n$$\nf (y) \\geq f (x) + v ^ {T} (y - x) \\quad \\text { for all } y \\in \\mathbb {R} ^ {d}.\n$$\nThe set of subgradients of f at x is called the subdifferential of f at x and denoted $\\partial f(x)$ .\nIn order to make use of subgradients, we will require the following two facts:\nProposition 16. Let $f: \\mathbb{R}^d \\to \\mathbb{R}$ be convex, and suppose $f$ is differentiable at $x$ . Then $\\partial f(x) = \\{\\nabla f(x)\\}$ .\nProposition 17. Let $f, g: \\mathbb{R}^d \\to \\mathbb{R}$ be convex, and let $\\alpha > 0$ . Then\n$$\n\\partial (\\alpha f) (x) = \\alpha \\partial f (x) = \\{\\alpha v: v \\in \\partial f (x) \\},\n$$\n$$\n\\partial (f + g) (x) = \\partial f (x) + \\partial g (x) = \\{v + w: v \\in \\partial f (x), w \\in \\partial g (x) \\}.\n$$\nThe following easy (but key) result is often referred to in the statistical literature as the Karush–Kuhn–Tucker (KKT) conditions, though it is actually a much simplified version of them.\nProposition 18. $x^{*}\\in \\arg \\min_{x\\in \\mathbb{R}^{d}}f(x)$ if and only if $0\\in \\partial f(x^{*})$\nProof.\n$$\nf (y) \\geq f (x ^ {*}) \\quad \\text { for all } y \\in \\mathbb {R} ^ {d} \\Leftrightarrow f (y) \\geq f (x ^ {*}) + 0 ^ {T} (y - x) \\quad \\text { for all } y \\in \\mathbb {R} ^ {d}\n$$\n$$\n\\Leftrightarrow 0 \\in \\partial f (x ^ {*}).\n$$\nLet us now compute the subdifferential of the $\\ell_{1}$ -norm. First note that $\\|\\cdot\\|_{1}:R^{d}\\to R$ is convex. Indeed it is a norm so the triangle inequality gives $\\|tx+(1-t)y\\|_{1}\\leq t\\|x\\|_{1}+(1-t)\\|y\\|_{1}$ . We introduce some notation that will be helpful here and throughout the rest of the course.\nFor $x \\in R^{d}$ and $A = \\{k_{1}, \\ldots, k_{m}\\} \\subseteq \\{1, \\ldots, d\\}$ with $k_{1} < \\cdots < k_{m}$ , by $x_{A}$ we will mean $(x_{k_{1}}, \\ldots, x_{k_{m}})^{T}$ . Similarly if X has d columns we will write $X_{A}$ for the matrix\n$$\nX _ {A} = (X _ {k _ {1}} \\dots X _ {k _ {m}}).\n$$\nFurther in this context, by $A^{c}$ , we will mean $\\{1,\\ldots,d\\}\\backslash A$ . Additionally, when in subscripts we will use the shorthand $-j=\\{j\\}^{c}$ and $-jk=\\{j,k\\}^{c}$ . Note these column and component extraction operations will always be considered to have taken place first before any further operations on the matrix, so for example $X_{A}^{T}=(X_{A})^{T}$ . Finally, define\n$$\n\\operatorname{sgn} (x _ {1}) = \\left\\{ \\begin{array}{l l} - 1 & \\text {if x_{1} < 0} \\\\ 0 & \\text {if x_{1} = 0} \\\\ 1 & \\text {if x_{1} >0}, \\end{array} \\right.\n$$\nand\n$$\n\\operatorname{sgn} (x) = (\\operatorname{sgn} (x _ {1}), \\dots , \\operatorname{sgn} (x _ {d})) ^ {T}.\n$$\nProposition 19. For $x \\in R^{d}$ let $A = \\{j : x_{j} \\neq 0\\}$ . Then\n$$\n\\partial \\| x \\| _ {1} = \\{v \\in \\mathbb {R} ^ {d}: \\| v \\| _ {\\infty} \\leq 1 a n d v _ {A} = \\operatorname{sgn} (x _ {A}) \\}\n$$\nProof. For $j = 1, \\ldots, d$ , let\n$$\ng _ {j}: \\mathbb {R} ^ {d} \\to \\mathbb {R}\n$$\n$$\nx \\mapsto | x _ {j} |.\n$$\nThen $\\| \\cdot \\| = \\sum_{j} g_{j}(\\cdot)$ so by Proposition 17, $\\partial \\| x \\|_{1} = \\sum_{j} \\partial g_{j}(x)$ . When $x$ has $x_{j} \\neq 0$ , $g_{j}$ is differentiable at $x$ so by Proposition 16 $\\partial g_{j}(x) = \\{\\text{sgn}(x_{j})e_{j}\\}$ where $e_{j}$ is the $j$ th unit vector. When $x_{j} = 0$ , if $v \\in \\partial g_{j}(x)$ we must have\n$$\ng _ {j} (y) \\geq g _ {j} (x) + v ^ {T} (y - x) \\quad \\text {for all} y \\in \\mathbb {R} ^ {d},\n$$\nSO\n$$\n\\left| y _ {j} \\right| \\geq v ^ {T} (y - x) \\quad \\text { for all } y \\in \\mathbb {R} ^ {d}. \\tag {3.1.1}\n$$\nwe claim that the above holds iff. $v_{j} \\in [-1,1]$ and $v_{-j} = 0$ . For the ‘if’ direction, note that $v^{T}(y - x) = v_{j}y_{j} \\leq |y_{j}|$ . Conversely, set $y_{-j} = x_{-j} + v_{-j}$ and $y_{j} = 0$ in (3.1.1) to get $0 \\geq \\|v_{-j}\\|_{2}^{2}$ , so $v_{-j} = 0$ . Then take y with $y_{-j} = x_{-j}$ to get $|y_{j}| \\geq v_{j}y_{j}$ for all $y_{j} \\in R$ , so $|v_{j}| \\leq 1$ . Forming the set sum of the subdifferentials then gives the result.", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.1 Some facts from optimisation theory and convex analysis/3.1.1 Subgradients", + "metadata": { + "length": 4013, + "summary": "The passage defines subgradients and subdifferentials for convex functions, states basic properties (differentiability, scaling, sum rules), and presents the KKT condition for optimality. It then computes the subdifferential of the ℓ1-norm, showing it consists of vectors with infinity-norm ≤1 and whose nonzero entries equal the sign of the corresponding coordinate.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Definition", + "vector", + "mathbb", + "subgradient", + "convex", + "function", + "geq", + "quad", + "text", + "The", + "set", + "subgradients", + "called", + "subdifferential", + "denoted", + "partial", + "In", + "order", + "make", + "require", + "facts", + "Proposition", + "16", + "Let", + "suppose", + "differentiable", + "Then", + "nabla", + "17", + "alpha", + "easy", + "key", + "result", + "referred", + "statistical", + "literature", + "Karush", + "Kuhn", + "Tucker", + "KKT", + "conditions", + "simplified", + "version", + "18", + "arg", + "min", + "Proof", + "Leftrightarrow", + "compute", + "ell", + "norm", + "First", + "note", + "cdot", + "Indeed", + "triangle", + "inequality", + "tx", + "leq", + "We", + "introduce", + "notation", + "helpful", + "rest", + "For", + "ldots", + "subseteq", + "cdots", + "Similarly", + "columns", + "write", + "matrix", + "dots", + "Further", + "context", + "backslash", + "Additionally", + "subscripts", + "shorthand", + "jk", + "Note", + "column", + "component", + "extraction", + "operations", + "considered", + "place", + "Finally", + "define", + "operatorname", + "sgn", + "left", + "begin", + "array", + "end", + "19", + "neq", + "infty", + "mapsto", + "sum", + "When", + "unit", + "SO", + "tag", + "3.1", + "claim", + "holds", + "iff", + "direction", + "Conversely", + "Forming", + "subdifferentials" + ], + "keywords": [ + "subgradient", + "subdifferential", + "ℓ1-norm" + ], + "connect_to": [] + } + }, + { + "chunk_id": "91574af7-541a-596d-a7fe-87641dc847f9", + "type": "text", + "content": "Consider an optimisation problem of the form\n$$\n\\text { minimise } f (x), \\text { subject to } g (x) = 0 \\tag {3.1.2}\n$$\nwhere $g : R^{d} \\to R^{b}$ . Suppose the optimal value is $c^{*} \\in R$ . The Lagrangian for this problem is defined as\n$$\nL (x, \\theta) = f (x) + \\theta^ {T} g (x)\n$$\nwhere $\\theta \\in \\mathbb{R}^b$ . Note that\n$$\n\\inf _ {x \\in \\mathbb {R} ^ {d}} L (x, \\theta) \\leq \\inf _ {x \\in \\mathbb {R} ^ {d}: g (x) = 0} L (x, \\theta) = c ^ {*}\n$$\nfor all $\\theta$ . The Lagrangian method involves finding a $\\theta^{*}$ such that the minimising $x^{*}$ on the LHS satisfies $g(x^{*}) = 0$ . This $x^{*}$ must then be a minimiser in the original problem (3.1.2).", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.1 Some facts from optimisation theory and convex analysis/3.1.2 The Lagrangian method", + "metadata": { + "length": 676, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Consider", + "optimisation", + "problem", + "form", + "text", + "minimise", + "subject", + "tag", + "3.1", + "Suppose", + "optimal", + "The", + "Lagrangian", + "defined", + "theta", + "mathbb", + "Note", + "inf", + "leq", + "method", + "involves", + "finding", + "minimising", + "LHS", + "satisfies", + "This", + "minimiser", + "original" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e20f442e-30ca-5ef8-8aa2-e9b9555ac756", + "type": "text", + "content": "Equipped with these tools from convex analysis, we can now fully characterise the solutions to the Lasso. We have that $\\hat{\\beta}_{\\lambda}^{L}$ is a Lasso solution if and only if $0 \\in \\partial Q_{\\lambda}(\\hat{\\beta}_{\\lambda}^{L})$ , which is equivalent to\n$$\n\\frac {1}{n} X ^ {T} (Y - X \\hat {\\beta} _ {\\lambda} ^ {\\mathrm{L}}) = \\lambda \\hat {\\nu},\n$$\nfor $\\hat{\\nu}$ with $\\| \\hat{\\nu}\\|_{\\infty}\\leq 1$ and writing $\\hat{S}_{\\lambda} = \\{k:\\hat{\\beta}_{\\lambda ,k}^{\\mathrm{L}}\\neq 0\\}$ , $\\hat{\\nu}_{\\hat{S}_{\\lambda}} = \\mathrm{sgn}(\\hat{\\beta}_{\\lambda ,\\hat{S}_{\\lambda}}^{\\mathrm{L}})$ . These are known as the KKT conditions for the Lasso.", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.2 Lasso solutions", + "metadata": { + "length": 655, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Equipped", + "tools", + "convex", + "analysis", + "fully", + "characterise", + "solutions", + "Lasso", + "We", + "hat", + "beta", + "lambda", + "solution", + "partial", + "equivalent", + "frac", + "mathrm", + "nu", + "infty", + "leq", + "writing", + "neq", + "sgn", + "These", + "KKT", + "conditions" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "dffcc427-9d64-5b64-bb61-609762d5d855", + "type": "text", + "content": "Consider now the “noiseless” version of the high-dimensional linear model (2.2.3), $Y = X\\beta^{0}$ . The case with noise can dealt with by similar arguments to those we will use below when we work on an event that $\\|X^{T}\\varepsilon\\|_{\\infty}/n$ is small (see example sheet).\nLet $S = \\{k : \\beta_k^0 \\neq 0\\}$ , $N = \\{1, \\ldots, p\\} \\setminus S$ and assume wlog that $S = \\{1, \\ldots, s\\}$ , and also that $\\text{rank}(X_S) = s$ .\nTheorem 20. Let $\\lambda > 0$ and define $\\Delta = X_N^T X_S(X_S^T X_S)^{-1} \\mathrm{sgn}(\\beta_S^0)$ . If $\\| \\Delta \\|_{\\infty} \\leq 1$ and for $k \\in S$ ,\n$$\n| \\beta_ {k} ^ {0} | > \\lambda | \\mathrm{sgn} (\\beta_ {S} ^ {0}) ^ {T} [ \\{\\frac {1}{n} X _ {S} ^ {T} X _ {S} \\} ^ {- 1} ] _ {k} |, (3. 3. 1)\n$$\nthen there exists a Lasso solution $\\hat{\\beta}_{\\lambda}^{\\mathrm{L}}$ with $\\mathrm{sgn}(\\hat{\\beta}_{\\lambda}^{\\mathrm{L}}) = \\mathrm{sgn}(\\beta^{0})$ . As a partial converse, if there exists a Lasso solution $\\hat{\\beta}_{\\lambda}^{\\mathrm{L}}$ with $\\mathrm{sgn}(\\hat{\\beta}_{\\lambda}^{\\mathrm{L}}) = \\mathrm{sgn}(\\beta^{0})$ , then $\\| \\Delta \\|_{\\infty} \\leq 1$ .\nRemark 1. We can interpret $\\|\\Delta\\|_{\\infty}$ as the maximum in absolute value over $k \\in N$ of the dot product of $\\operatorname{sgn}(\\beta_{S}^{0})$ and $(X_{S}^{T}X_{S})^{-1}X_{S}^{T}X_{k}$ , the coefficient vector obtained by regressing $X_{k}$ on $X_{S}$ . The condition $\\|\\Delta\\|_{\\infty} \\leq 1$ is known as the irrepresentable condition.\nProof. Fix $\\lambda > 0$ and write $\\hat{\\beta} = \\hat{\\beta}_{\\lambda}^{\\mathrm{L}}$ and $\\hat{S} = \\{k : \\hat{\\beta}_k \\neq 0\\}$ for convenience. The KKT conditions for the Lasso give\n$$\n\\frac {1}{n} X ^ {T} X (\\beta^ {0} - \\hat {\\beta}) = \\lambda \\hat {\\nu}\n$$\nwhere $\\| \\hat{\\nu}\\|_{\\infty}\\leq 1$ and $\\hat{\\nu}_{\\hat{S}} = \\mathrm{sgn}(\\hat{\\beta}_{\\hat{S}})$ . We can expand this into\n$$\n\\frac {1}{n} \\left( \\begin{array}{c c} X _ {S} ^ {T} X _ {S} & X _ {S} ^ {T} X _ {N} \\\\ X _ {N} ^ {T} X _ {S} & X _ {N} ^ {T} X _ {N} \\end{array} \\right) \\binom{\\beta_ {S} ^ {0} - \\hat {\\beta} _ {S}}{- \\hat {\\beta} _ {N}} = \\lambda \\binom{\\hat {\\nu} _ {S}}{\\hat {\\nu} _ {N}}. \\tag {3.3.2}\n$$\nWe prove the converse first. If $\\operatorname{sgn}(\\hat{\\beta}) = \\operatorname{sgn}(\\beta^{0})$ then $\\hat{\\nu}_{S} = \\operatorname{sgn}(\\beta_{S}^{0})$ and $\\hat{\\beta}_{N} = 0$ . The top block of (3.3.2) gives\n$$\n\\beta_ {S} ^ {0} - \\hat {\\beta} _ {S} = \\lambda (\\frac {1}{n} X _ {S} ^ {T} X _ {S}) ^ {- 1} \\mathrm{sgn} (\\beta_ {S} ^ {0}).\n$$\nSubstituting this into the bottom block, we get\n$$\n\\lambda \\frac {1}{n} X _ {N} ^ {T} X _ {S} (\\frac {1}{n} X _ {S} ^ {T} X _ {S}) ^ {- 1} \\mathrm{sgn} (\\beta_ {S} ^ {0}) = \\lambda \\hat {\\nu} _ {N}.\n$$\nThus as $\\| \\hat{\\nu}_N\\|_{\\infty}\\leq 1$ , we have $\\| \\Delta \\|_{\\infty}\\leq 1$ .\nFor the positive statement, we need to find a $\\hat{\\beta}$ and $\\hat{\\nu}$ such that $\\mathrm{sgn}(\\hat{\\beta}_S) = \\mathrm{sgn}(\\beta_S^0)$ and $\\hat{\\beta}_N = 0$ , for which the KKT conditions hold. We claim that taking\n$$\n(\\hat {\\beta} _ {S}, \\hat {\\beta} _ {N}) = (\\beta_ {S} ^ {0} - \\lambda (\\frac {1}{n} X _ {S} ^ {T} X _ {S}) ^ {- 1} \\mathrm{sgn} (\\beta_ {S} ^ {0}), 0)\n$$\n$$\n(\\hat {\\nu} _ {S}, \\hat {\\nu} _ {N}) = (\\mathrm{sgn} (\\beta_ {S} ^ {0}), \\Delta)\n$$\nsatisfies (3.3.2). We only need to check that $\\operatorname{sgn}(\\beta_{S}^{0}) = \\operatorname{sgn}(\\hat{\\beta}_{S})$ , but this follows from (3.3.1). ☐", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.3 Variable selection", + "metadata": { + "length": 3407, + "summary": "Theorem 20 provides conditions for Lasso to recover the true sign pattern in a noiseless high-dimensional linear model. The irrepresentable condition requires that the maximum absolute dot product between the sign vector and regression coefficients of inactive variables on active ones is at most 1. Additionally, the true coefficients must exceed a threshold involving the inverse Gram matrix. Under these conditions, a Lasso solution exists with the correct sign pattern. The converse also holds: if a Lasso solution recovers the true signs, the irrepresentable condition must be satisfied.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Consider", + "noiseless", + "version", + "high", + "dimensional", + "linear", + "model", + "2.2", + "beta", + "The", + "case", + "noise", + "dealt", + "similar", + "arguments", + "work", + "event", + "varepsilon", + "infty", + "small", + "sheet", + "Let", + "neq", + "ldots", + "setminus", + "assume", + "wlog", + "text", + "rank", + "Theorem", + "20", + "lambda", + "define", + "Delta", + "mathrm", + "sgn", + "If", + "leq", + "frac", + "exists", + "Lasso", + "solution", + "hat", + "As", + "partial", + "converse", + "Remark", + "We", + "interpret", + "maximum", + "absolute", + "dot", + "product", + "operatorname", + "coefficient", + "vector", + "obtained", + "regressing", + "condition", + "irrepresentable", + "Proof", + "Fix", + "write", + "convenience", + "KKT", + "conditions", + "give", + "nu", + "expand", + "left", + "begin", + "array", + "end", + "binom", + "tag", + "3.3", + "prove", + "top", + "block", + "Substituting", + "bottom", + "Thus", + "For", + "positive", + "statement", + "find", + "hold", + "claim", + "taking", + "satisfies", + "check" + ], + "keywords": [ + "Lasso", + "irrepresentable condition", + "sign consistency" + ], + "connect_to": [] + } + }, + { + "chunk_id": "65d174e0-7900-5de6-9ed3-2aff6bb27166", + "type": "image", + "content": "\nFrom this we get ## 3.5 The compatibility condition0\n[images/image-4-From this .jpg]\n", + "path": "images/image-4-From this .jpg", + "metadata": { + "length": 86, + "summary": "image-4\nFrom this we get ## 3.5 The compatibility condition0", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "file_path": "images/image-4-From this .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "9026bf27-6510-5888-baa9-58575bd31b43", + "type": "text", + "content": "Consider again the model $Y = X\\beta^{0} + \\varepsilon - \\bar{\\varepsilon}1$ where the components of $\\varepsilon$ are independent mean-zero sub-Gaussian random variables with common parameter $\\sigma$ . Let S, s and N be defined as in the previous section. As we have noted before, in an artificial situation where S is known, we could apply OLS on $X_{S}$ and have an MSPE of $\\sigma^{2}s/n$ . Under a so-called compatibility condition on the design matrix, we can obtain a similar MSPE for the Lasso.\nDefinition 5. Given a matrix of predictors $X \\in R^{n \\times p}$ and support set S, define\n$$\n\\phi^ {2} = \\inf _ {\\beta \\in \\mathbb {R} ^ {p}: \\beta_ {S} \\neq 0, \\| \\beta_ {N} \\| _ {1} \\leq 3 \\| \\beta_ {S} \\| _ {1}} \\frac {\\frac {1}{n} \\| X \\beta \\| _ {2} ^ {2}}{\\frac {1}{s} \\| \\beta_ {S} \\| _ {1} ^ {2}},\n$$\nwhere $s = |S|$ and we take $\\phi \\geq 0$ . The compatibility condition is that $\\phi^{2} > 0$ .\nNote that if $X^{T}X/n$ has minimum eigenvalue $c_{\\min} > 0$ (so necessarily $p \\leq n$ ), then $\\phi^{2} > c_{\\min}$ . Indeed by the Cauchy–Schwarz inequality,\n$$\n\\| \\beta_ {S} \\| _ {1} = \\mathrm{sgn} (\\beta_ {S}) ^ {T} \\beta_ {S} \\leq \\sqrt {s} \\| \\beta_ {S} \\| _ {2} \\leq \\sqrt {s} \\| \\beta \\| _ {2}.\n$$\nThus\n$$\n\\phi^ {2} \\geq \\inf _ {\\beta \\neq 0} \\frac {\\frac {1}{n} \\| X \\beta \\| _ {2} ^ {2}}{\\| \\beta \\| _ {2} ^ {2}} = c _ {\\mathrm{min}}.\n$$\nAlthough in the high-dimensional setting we would have $c_{min} = 0$ , the fact that the infimum in the definition of $\\phi^{2}$ is over a restricted set of $\\beta$ can still allow $\\phi^{2}$ to be positive even in this case, as we discuss after the presentation of the theorem.\nTheorem 21. Suppose the compatibility condition holds and let $\\hat{\\beta}$ be the Lasso solution with $\\lambda = A\\sigma \\sqrt{\\log(p) / n}$ for $A > 0$ . Then with probability at least $1 - 2p^{-(A^2 /8 - 1)}$ , we have\n$$\n\\frac {1}{n} \\| X (\\beta^ {0} - \\hat {\\beta}) \\| _ {2} ^ {2} + \\lambda \\| \\hat {\\beta} - \\beta^ {0} \\| _ {1} \\leq \\frac {1 6 \\lambda^ {2} s}{\\phi^ {2}} = \\frac {1 6 A ^ {2} \\log (p)}{\\phi^ {2}} \\frac {\\sigma^ {2} s}{n}.\n$$\nProof. As in Theorem 11 we start with the “basic inequality”:\n$$\n\\frac {1}{2 n} \\| X (\\hat {\\beta} - \\beta^ {0}) \\| _ {2} ^ {2} + \\lambda \\| \\hat {\\beta} \\| _ {1} \\leq \\frac {1}{n} \\varepsilon^ {T} X (\\hat {\\beta} - \\beta^ {0}) + \\lambda \\| \\beta^ {0} \\| _ {1}.\n$$\nWe work on the event $\\Omega = \\{2\\| X^T\\varepsilon \\|_{\\infty} / n\\leq \\lambda \\}$ where after applying Hölder's inequality, we get\n$$\n\\frac {1}{n} \\| X (\\hat {\\beta} - \\beta^ {0}) \\| _ {2} ^ {2} + 2 \\lambda \\| \\hat {\\beta} \\| _ {1} \\leq \\lambda \\| \\hat {\\beta} - \\beta^ {0} \\| _ {1} + 2 \\lambda \\| \\beta^ {0} \\| _ {1}. \\tag {3.4.1}\n$$\nLemma 15 shows that $\\mathbb{P}(\\Omega) \\geq 1 - 2p^{-(A^2/8-1)}$ .\nTo motivate the rest of the proof, consider the following idea. We know\n$$\n\\frac {1}{n} \\| X (\\hat {\\beta} - \\beta^ {0}) \\| _ {2} ^ {2} \\leq 3 \\lambda \\| \\hat {\\beta} - \\beta^ {0} \\| _ {1}.\n$$\nIf we could get\n$$\n3 \\lambda \\| \\hat {\\beta} - \\beta^ {0} \\| _ {1} \\leq \\frac {c \\lambda}{\\sqrt {n}} \\| X (\\hat {\\beta} - \\beta^ {0}) \\| _ {2}\n$$\nfor some constant $c > 0$ , then we would have that $\\| X(\\hat{\\beta} - \\beta^0)\\| _2^2 /n\\leq c^2\\lambda^2$ and also $3\\lambda \\| \\beta^0 -\\hat{\\beta}\\| _1\\leq c^2\\lambda^2$ .\nReturning to the actual proof, write $a = \\| X(\\hat{\\beta} - \\beta^0)\\| _2^2 /(n\\lambda)$ . Then from (3.4.1) we can derive the following string of inequalities:\n$$\na + 2 (\\| \\hat {\\beta} _ {N} \\| _ {1} + \\| \\hat {\\beta} _ {S} \\| _ {1}) \\leq \\| \\hat {\\beta} _ {S} - \\beta_ {S} ^ {0} \\| _ {1} + \\| \\hat {\\beta} _ {N} \\| _ {1} + 2 \\| \\beta_ {S} ^ {0} \\| _ {1}\n$$\n$$\na + \\| \\hat {\\beta} _ {N} \\| _ {1} \\leq \\| \\hat {\\beta} _ {S} - \\beta_ {S} ^ {0} \\| _ {1} + 2 \\| \\beta_ {S} ^ {0} \\| _ {1} - 2 \\| \\hat {\\beta} _ {S} \\| _ {1}\n$$\n$$\na + \\| \\hat {\\beta} _ {N} - \\beta_ {N} ^ {0} \\| _ {1} \\leq 3 \\| \\beta_ {S} ^ {0} - \\hat {\\beta} _ {S} \\| _ {1}\n$$\n$$\na + \\| \\hat {\\beta} - \\beta^ {0} \\| _ {1} \\leq 4 \\| \\beta_ {S} ^ {0} - \\hat {\\beta} _ {S} \\| _ {1},\n$$\nthe final inequality coming from adding $\\| \\beta_S^0 -\\hat{\\beta}_S\\| _1$ to both sides.\nNow using the compatibility condition with $\\beta = \\hat{\\beta} -\\beta^0$ we have\n$$\n\\frac {1}{n} \\| X (\\hat {\\beta} - \\beta^ {0}) \\| _ {2} ^ {2} + \\lambda \\| \\beta^ {0} - \\hat {\\beta} \\| _ {1} \\leq 4 \\lambda \\| \\beta_ {S} ^ {0} - \\hat {\\beta} _ {S} \\| _ {1}\n$$\n$$\n\\leq \\frac {4 \\lambda}{\\phi} \\sqrt {\\frac {s}{n}} \\| X (\\hat {\\beta} - \\beta^ {0}) \\| _ {2}. \\tag {3.4.2}\n$$\nFrom this we get\n$$\n\\frac {1}{\\sqrt {n}} \\| X (\\hat {\\beta} - \\beta^ {0}) \\| _ {2} \\leq \\frac {4 \\lambda \\sqrt {s}}{\\phi},\n$$\nand substituting this into the RHS of (3.4.2) gives the result.\n\nFrom this we get ## 3.5 The compatibility condition0\n[images/image-4-From this .jpg]", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.4 Prediction and estimation", + "metadata": { + "length": 4762, + "summary": "The text discusses the compatibility condition for the Lasso estimator in high-dimensional linear regression. It defines the compatibility constant φ², which measures the restricted eigenvalue of the design matrix. Under this condition, the Lasso with tuning parameter λ = Aσ√(log(p)/n) achieves an MSPE bound of O(σ²s log(p)/(n φ²)) with high probability. The proof uses the basic inequality, Hölder's inequality, and the compatibility condition to derive the oracle-like bound.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Consider", + "model", + "beta", + "varepsilon", + "bar", + "components", + "independent", + "Gaussian", + "random", + "variables", + "common", + "parameter", + "sigma", + "Let", + "defined", + "previous", + "section", + "As", + "noted", + "artificial", + "situation", + "apply", + "OLS", + "MSPE", + "Under", + "called", + "compatibility", + "condition", + "design", + "matrix", + "obtain", + "similar", + "Lasso", + "Definition", + "Given", + "predictors", + "times", + "support", + "set", + "define", + "phi", + "inf", + "mathbb", + "neq", + "leq", + "frac", + "geq", + "The", + "Note", + "minimum", + "eigenvalue", + "min", + "necessarily", + "Indeed", + "Cauchy", + "Schwarz", + "inequality", + "mathrm", + "sgn", + "sqrt", + "Thus", + "Although", + "high", + "dimensional", + "setting", + "fact", + "infimum", + "definition", + "restricted", + "positive", + "case", + "discuss", + "presentation", + "theorem", + "Theorem", + "21", + "Suppose", + "holds", + "hat", + "solution", + "lambda", + "log", + "Then", + "probability", + "2p", + "Proof", + "11", + "start", + "basic", + "We", + "work", + "event", + "Omega", + "infty", + "applying", + "lder", + "tag", + "3.4", + "Lemma", + "15", + "shows", + "To", + "motivate", + "rest", + "proof", + "idea", + "If", + "constant", + "Returning", + "actual", + "write", + "derive", + "string", + "inequalities", + "final", + "coming", + "adding", + "sides", + "Now", + "From", + "substituting", + "RHS", + "result", + "3.5", + "condition0" + ], + "keywords": [ + "Lasso", + "MSPE", + "compatibility" + ], + "connect_to": [ + { + "target": "65d174e0-7900-5de6-9ed3-2aff6bb27166", + "relation": "embeds", + "ref": "[images/image-4-From this .jpg]", + "position": { + "start": 4731, + "end": 4762 + } + } + ] + } + }, + { + "chunk_id": "e9a6ccec-c4f6-5db0-bc9f-c70b6f59c75b", + "type": "image", + "content": "\n$$ We would like to apply the result above with $\\check{\\Sigma} = \\Sigma^{0}$ , and use to argue that if $\\Sigma^{0}$ satisfies the compatibility condition, then so will $\\hat{\\Sigma}$ with high probability. In order to do this, we need to argue that the event that $\\max_{jk} |\\hat{\\Sigma}_{jk} - \\Sigma_{jk}^{0}|$ is small occurs with high probability. We can obtain such a result with the aid of concentration inequalities.0\n[images/image-5-$$ We woul.jpg]\n", + "path": "images/image-5-$$ We woul.jpg", + "metadata": { + "length": 461, + "summary": "image-5\n$$ We would like to apply the result above with $\\check{\\Sigma} = \\Sigma^{0}$ , and use to argue that if $\\Sigma^{0}$ satisfies the compatibility condition, then so will $\\hat{\\Sigma}$ with high probability. In order to do this, we need to argue that the event that $\\max_{jk} |\\hat{\\Sigma}_{jk} - \\Sigma_{jk}^{0}|$ is small occurs with high probability. We can obtain such a result with the aid of concentration inequalities.0", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "file_path": "images/image-5-$$ We woul.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "5d683ec4-e2ca-56d3-a1ab-b83f8db32b24", + "type": "text", + "content": "How strong is the compatibility condition? In order to answer this question, we shall think of X as random and try to understand what conditions on the population covariance matrix $\\Sigma^{0} := \\mathbb{E}(X^{T}X/n)$ imply that X satisfies a compatibility condition with high probability. To this end let us define\n$$\n\\phi_ {\\Sigma} ^ {2} (S) = \\inf _ {\\beta : \\| \\beta_ {S} \\| _ {1} \\neq 0, \\| \\beta_ {N} \\| _ {1} \\leq 3 \\| \\beta_ {S} \\| _ {1}} \\frac {\\beta^ {T} \\Sigma \\beta}{\\| \\beta_ {S} \\| _ {1} ^ {2} / | S |}\n$$\nwhere $\\Sigma \\in \\mathbb{R}^{p\\times p}$ . Note then our $\\phi^2 = \\phi_{\\hat{\\Sigma}}^2 (S)$ where $\\hat{\\Sigma} := X^T X / n$ and $S$ is the support set of $\\beta^0$ . The following result shows that if $\\hat{\\Sigma}$ is close to a matrix $\\check{\\Sigma}$ for which $\\phi_{\\tilde{\\Sigma}}^2 (S) > 0$ , then also $\\phi_{\\hat{\\Sigma}}^2 (S) > 0$ .\nLemma 22. Suppose $\\phi_{\\tilde{\\Sigma}}^{2}(S) > 0$ and $\\max_{jk}|\\hat{\\Sigma}_{jk} - \\check{\\Sigma}_{jk}|\\leq \\phi_{\\tilde{\\Sigma}}^{2}(S) / (32|S|)$ . Then $\\phi_{\\hat{\\Sigma}}^{2}(S)\\geq \\phi_{\\tilde{\\Sigma}}^{2}(S) / 2$ .\nProof. In the following we suppress dependence on $S$ . Let $s = |S|$ and let $t = \\phi_{\\Sigma}^2 / (32s)$ . We have\n$$\n| \\beta^ {T} (\\check {\\Sigma} - \\hat {\\Sigma}) \\beta | \\leq \\| \\beta \\| _ {1} \\| (\\check {\\Sigma} - \\hat {\\Sigma}) \\beta \\| _ {\\infty} (\\text { Hölder })\n$$\n$$\n\\leq t \\| \\beta \\| _ {1} ^ {2} (\\text { Hölder again }).\n$$\nIf $\\| \\beta_N\\| _1\\leq 3\\| \\beta_S\\| _1$ then\n$$\n\\| \\beta \\| _ {1} = \\| \\beta_ {N} \\| _ {1} + \\| \\beta_ {S} \\| _ {1} \\leq 4 \\| \\beta_ {S} \\| _ {1} \\leq 4 \\frac {\\sqrt {\\beta^ {T} \\check {\\Sigma} \\beta}}{\\phi_ {\\check {\\Sigma}} / \\sqrt {s}}.\n$$\nThus if $\\| \\beta_N\\| _1\\leq 3\\| \\beta_S\\| _1$\n$$\n\\beta^ {T} \\check {\\Sigma} \\beta - \\frac {\\phi_ {\\check {\\Sigma}} ^ {2}}{3 2 s} \\frac {1 6 \\beta^ {T} \\check {\\Sigma} \\beta}{\\phi_ {\\check {\\Sigma}} ^ {2} / s} = \\frac {1}{2} \\beta^ {T} \\check {\\Sigma} \\beta \\leq \\beta^ {T} \\hat {\\Sigma} \\beta .\n$$\n\n$$ We would like to apply the result above with $\\check{\\Sigma} = \\Sigma^{0}$ , and use to argue that if $\\Sigma^{0}$ satisfies the compatibility condition, then so will $\\hat{\\Sigma}$ with high probability. In order to do this, we need to argue that the event that $\\max_{jk} |\\hat{\\Sigma}_{jk} - \\Sigma_{jk}^{0}|$ is small occurs with high probability. We can obtain such a result with the aid of concentration inequalities.0\n[images/image-5-$$ We woul.jpg]\n\nWe would like to apply the result above with $\\check{\\Sigma} = \\Sigma^{0}$ , and use to argue that if $\\Sigma^{0}$ satisfies the compatibility condition, then so will $\\hat{\\Sigma}$ with high probability. In order to do this, we need to argue that the event that $\\max_{jk} |\\hat{\\Sigma}_{jk} - \\Sigma_{jk}^{0}|$ is small occurs with high probability. We can obtain such a result with the aid of concentration inequalities.", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.5 The compatibility condition", + "metadata": { + "length": 2869, + "summary": "The passage discusses the compatibility condition for a covariance matrix in high-dimensional statistics. It defines a quantity φ²(S) that measures compatibility, and presents Lemma 22 showing that if an estimated covariance matrix Σ̂ is close to a matrix Σ̃ with positive φ²(S), then Σ̂ also has positive φ²(S). The proof uses Hölder's inequality and bounds on the difference between matrices. The goal is to apply this with Σ̃ = Σ⁰ (population covariance) and use concentration inequalities to show that Σ̂ satisfies the compatibility condition with high probability when Σ⁰ does.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "How", + "strong", + "compatibility", + "condition", + "In", + "order", + "answer", + "question", + "random", + "understand", + "conditions", + "population", + "covariance", + "matrix", + "Sigma", + "mathbb", + "imply", + "satisfies", + "high", + "probability", + "To", + "end", + "define", + "phi", + "inf", + "beta", + "neq", + "leq", + "frac", + "times", + "Note", + "hat", + "support", + "set", + "The", + "result", + "shows", + "close", + "check", + "tilde", + "Lemma", + "22", + "Suppose", + "max", + "jk", + "32", + "Then", + "geq", + "Proof", + "suppress", + "dependence", + "Let", + "32s", + "We", + "infty", + "text", + "lder", + "If", + "sqrt", + "Thus", + "apply", + "argue", + "event", + "small", + "occurs", + "obtain", + "aid", + "concentration", + "inequalities.0", + "inequalities" + ], + "keywords": [ + "compatibility condition", + "covariance matrix", + "concentration inequalities" + ], + "connect_to": [ + { + "target": "e9a6ccec-c4f6-5db0-bc9f-c70b6f59c75b", + "relation": "embeds", + "ref": "[images/image-5-$$ We woul.jpg]", + "position": { + "start": 2413, + "end": 2444 + } + } + ] + } + }, + { + "chunk_id": "b7b2c630-d897-52e0-9d61-11c77c774ab6", + "type": "text", + "content": "When trying to understand the concentration properties of $\\hat{\\Sigma}_{jk}$ , it will be helpful to have a tail bound for a product of sub-Gaussian random variables. Bernstein's inequality, which applies to random variables satisfying the condition below, is helpful in this regard.\nDefinition 6 (Bernstein's condition). We say that the random variable $W$ with $\\mathbb{E}W = \\mu$ satisfies Bernstein's condition with parameter $(\\sigma, b)$ where $\\sigma, b > 0$ if\n$$\n\\mathbb {E} (| W - \\mu | ^ {k}) \\leq \\frac {1}{2} k! \\sigma^ {2} b ^ {k - 2} \\quad \\text { for } k = 2, 3, \\ldots .\n$$\nProposition 23 (Bernstein's inequality). Let $W_{1}, W_{2}, \\ldots$ be independent random variables with $\\mathbb{E}(W_{i}) = \\mu$ . Suppose each $W_{i}$ satisfies Bernstein's condition with parameter $(\\sigma, b)$ . Then\n$$\n\\mathbb {E} \\big (e ^ {\\alpha (W _ {i} - \\mu)} \\big) \\leq \\exp \\left(\\frac {\\alpha^ {2} \\sigma^ {2} / 2}{1 - b | \\alpha |}\\right) \\quad f o r a l l | \\alpha | < 1 / b\n$$\n$$\n\\mathbb {P} \\bigg (\\frac {1}{n} \\sum_ {i = 1} ^ {n} W _ {i} - \\mu \\geq t \\bigg) \\leq \\exp \\bigg (- \\frac {n t ^ {2}}{2 (\\sigma^ {2} + b t)} \\bigg) \\qquad f o r a l l t \\geq 0.\n$$\nProof. Fix $i$ and let $W = W_{i}$ . We have\n$$\n\\mathbb {E} \\left(e ^ {\\alpha (W - \\mu)}\\right) = 1 + \\alpha \\mathbb {E} (W - \\mu) + \\sum_ {k = 2} ^ {\\infty} \\alpha^ {k} \\frac {\\mathbb {E} \\left\\{\\left(W - \\mu\\right) ^ {k} \\right\\}}{k !}\n$$\n$$\n\\leq 1 + \\frac {\\sigma^ {2} \\alpha^ {2}}{2} \\sum_ {k = 2} ^ {\\infty} | \\alpha | ^ {k - 2} b ^ {k - 2}\n$$\n$$\n= 1 + \\frac {\\sigma^ {2} \\alpha^ {2}}{2} \\frac {1}{1 - | \\alpha | b} \\leq \\exp \\left(\\frac {\\alpha^ {2} \\sigma^ {2} / 2}{1 - b | \\alpha |}\\right),\n$$\nprovided $|\\alpha| < 1/b$ and using the inequality $e^{u} \\geq 1 + u$ in the final line. For the probability bound, first note that\n$$\n\\mathbb {E} \\exp \\left(\\sum_ {i = 1} ^ {n} \\alpha (W _ {i} - \\mu) / n\\right) = \\prod_ {i = 1} ^ {n} \\mathbb {E} \\exp \\{\\alpha (W _ {i} - \\mu) / n \\}\n$$\n$$\n\\leq \\exp \\left(n \\frac {(\\alpha / n) ^ {2} \\sigma^ {2} / 2}{1 - b | \\alpha / n |}\\right)\n$$\nfor $|\\alpha| / n < 1 / b$ . Then we use the Chernoff method and set $\\alpha / n = t / (bt + \\sigma^2) \\in [0, 1 / b)$ .\nLemma 24. Let W, Z be mean-zero and sub-Gaussian with parameters $\\sigma_{W}$ and $\\sigma_{Z}$ respectively. Then the product WZ satisfies Bernstein's condition with parameter $(8\\sigma_{W}\\sigma_{Z}, 4\\sigma_{W}\\sigma_{Z})$ .\nProof. In order to use Bernstein's inequality (Proposition 23) we first obtain bounds on the moments of $W$ and $Z$ . Note that $W^{2k} = \\int_0^\\infty \\mathbb{1}_{\\{x < W^{2k}\\}} dx$ . Thus by Fubini's theorem\n$$\n\\begin{array}{l} \\mathbb {E} (W ^ {2 k}) = \\int_ {0} ^ {\\infty} \\mathbb {P} (W ^ {2 k} > x) d x \\\\ = 2 k \\int_ {0} ^ {\\infty} t ^ {2 k - 1} \\mathbb {P} (| W | > t) d t \\quad \\text { substituting } t ^ {2 k} = x \\\\ \\leq 4 k \\int_ {0} ^ {\\infty} t ^ {2 k - 1} \\exp \\{- t ^ {2} / (2 \\sigma_ {W} ^ {2}) \\} d t \\qquad \\text { by Proposition 12 } \\\\ = 4 k \\sigma_ {W} ^ {2} \\int_ {0} ^ {\\infty} (2 \\sigma_ {W} ^ {2} x) ^ {k - 1} e ^ {- x} d x \\qquad \\text {substituting} t ^ {2} / (2 \\sigma_ {W} ^ {2}) = x \\\\ = 2 ^ {k + 1} \\sigma_ {W} ^ {2 k} k!. \\\\ \\end{array}\n$$\nNext note that for any random variable Y,\n$$\n\\begin{array}{l} \\left| \\mathbb {E} (Y - \\mathbb {E} Y) ^ {k} \\right| \\leq \\mathbb {E} | Y - \\mathbb {E} Y | ^ {k} \\\\ = 2 ^ {k} \\mathbb {E} | Y / 2 - \\mathbb {E} Y / 2 | ^ {k} \\\\ \\leq 2 ^ {k - 1} (\\mathbb {E} | Y | ^ {k} + | \\mathbb {E} Y | ^ {k}) \\quad \\text { by Jensen's inequality applied to } t \\mapsto | t | ^ {k}, \\\\ \\leq 2 ^ {k} \\mathbb {E} | Y | ^ {k}. \\\\ \\end{array}\n$$\nTherefore\n$$\n\\begin{array}{l} \\mathbb {E} (| W Z - \\mathbb {E} W Z | ^ {k}) \\leq 2 ^ {k} \\mathbb {E} | W Z | ^ {k} \\\\ \\leq 2 ^ {k} (\\mathbb {E} W ^ {2 k}) ^ {1 / 2} (\\mathbb {E} Z ^ {2 k}) ^ {1 / 2} \\quad \\text { by Cauchy - Schwarz } \\\\ \\leq 2 ^ {k} 2 ^ {k + 1} \\sigma_ {W} ^ {k} \\sigma_ {Z} ^ {k} k! \\\\ = \\frac {k !}{2} (8 \\sigma_ {W} \\sigma_ {Z}) ^ {2} (4 \\sigma_ {W} \\sigma_ {Z}) ^ {k - 2}. \\\\ \\end{array}\n$$", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.5 The compatibility condition/3.5.1 Concentration inequalities II", + "metadata": { + "length": 4029, + "summary": "The text introduces Bernstein's inequality for independent random variables satisfying Bernstein's condition with parameters (σ, b). It provides the moment condition, the moment generating function bound, and the tail probability bound. A lemma shows that the product of two mean-zero sub-Gaussian random variables satisfies Bernstein's condition with parameters (8σ_Wσ_Z, 4σ_Wσ_Z), enabling concentration results for such products.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "When", + "understand", + "concentration", + "properties", + "hat", + "Sigma", + "jk", + "helpful", + "tail", + "bound", + "product", + "Gaussian", + "random", + "variables", + "Bernstein", + "inequality", + "applies", + "satisfying", + "condition", + "regard", + "Definition", + "We", + "variable", + "mathbb", + "mu", + "satisfies", + "parameter", + "sigma", + "leq", + "frac", + "quad", + "text", + "ldots", + "Proposition", + "23", + "Let", + "independent", + "Suppose", + "Then", + "big", + "alpha", + "exp", + "left", + "bigg", + "sum", + "geq", + "qquad", + "Proof", + "Fix", + "infty", + "provided", + "final", + "line", + "For", + "probability", + "note", + "prod", + "Chernoff", + "method", + "set", + "bt", + "Lemma", + "24", + "parameters", + "WZ", + "In", + "order", + "obtain", + "bounds", + "moments", + "Note", + "2k", + "int", + "dx", + "Thus", + "Fubini", + "theorem", + "begin", + "array", + "substituting", + "12", + "end", + "Next", + "Jensen", + "applied", + "mapsto", + "Therefore", + "Cauchy", + "Schwarz" + ], + "keywords": [ + "Bernstein's condition", + "sub-Gaussian", + "tail bound" + ], + "connect_to": [] + } + }, + { + "chunk_id": "fafe08cb-421d-5087-bb9c-8568798a007f", + "type": "text", + "content": "We may now apply this with $\\check{\\Sigma} = \\Sigma^{0}$ . To make the result more readily interpretable, we shall state it in an asymptotic framework. Imagine a sequence of design matrices with n and p growing, each with their own compatibility condition. We will however suppress the asymptotic regime in the notation.\nTheorem 25. Suppose the rows of X are i.i.d. and each entry of X is mean-zero sub-Gaussian with parameter v. Suppose $s\\sqrt{\\log(p)/n} \\to 0$ (and $s, p, n > 1$ ) as $n \\to \\infty$ . Let\n$$\n\\phi_ {\\hat {\\Sigma}, s} ^ {2} = \\min _ {S: | S | = s} \\phi_ {\\hat {\\Sigma}} ^ {2} (S)\n$$\n$$\n\\phi_ {\\Sigma^ {0}, s} ^ {2} = \\min _ {S: | S | = s} \\phi_ {\\Sigma^ {0}} ^ {2} (S),\n$$\nand suppose the latter is bounded away from 0. Then $\\mathbb{P}(\\phi_{\\hat{\\Sigma},s}^2\\geq \\phi_{\\Sigma^0,s}^2 /2)\\to 1$ as $n\\to \\infty$ .\nProof. In view of Lemma 22, we need only show that\n$$\n\\mathbb {P} (\\max _ {j k} | \\hat {\\Sigma} _ {j k} - \\Sigma_ {j k} ^ {0} | \\geq \\phi_ {\\Sigma^ {0}, s} ^ {2} / (3 2 s)) \\to 0.\n$$\nLet $t = \\phi_{\\Sigma^0,s}^2 / (32s)$ . By a union bound and then Lemma 24 we have\n$$\n\\begin{array}{l} \\mathbb {P} (\\max _ {j k} | \\hat {\\Sigma} _ {j k} - \\Sigma_ {j k} ^ {0} | \\geq \\phi_ {\\Sigma^ {0}, s} ^ {2} / (3 2 s)) < p ^ {2} \\max _ {j k} \\mathbb {P} \\Big (\\Big | \\sum_ {i = 1} ^ {n} X _ {i j} X _ {i k} / n - \\Sigma_ {j k} ^ {0} \\Big | \\geq t \\Big) \\\\ \\leq 2 \\exp \\left(- \\frac {n t ^ {2}}{2 (6 4 v ^ {4} + 4 v ^ {2} t))} + 2 \\log (p)\\right) \\\\ \\leq c _ {1} \\exp (- c _ {2} n / s ^ {2} + c _ {3} \\log (p)) \\to 0. \\\\ \\end{array}\n$$\nCorollary 26. Suppose the rows of X are independent with distribution $N_{p}(0,\\Sigma^{0})$ . Suppose the diagonal entries of $\\Sigma^{0}$ are bounded above and the minimum eigenvalue of $\\Sigma^{0}$ , $c_{min}$ is bounded away from 0. Then $\\mathbb{P}(\\phi_{\\Sigma,s}^{2}\\geq c_{\\min}/2)\\to1$ provided $s\\sqrt{\\log(p)/n}\\to0$ .", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.5 The compatibility condition/3.5.2 The compatibility condition and random design", + "metadata": { + "length": 1884, + "summary": "The text presents Theorem 25 and Corollary 26 on the compatibility condition for sparse estimation. Theorem 25 shows that under i.i.d. sub-Gaussian rows with certain growth conditions, the empirical compatibility constant is at least half the population version with probability tending to 1. Corollary 26 extends this to normal distributions with bounded eigenvalues. The proof uses concentration inequalities and union bounds.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "We", + "apply", + "check", + "Sigma", + "To", + "make", + "result", + "readily", + "interpretable", + "state", + "asymptotic", + "framework", + "Imagine", + "sequence", + "design", + "matrices", + "growing", + "compatibility", + "condition", + "suppress", + "regime", + "notation", + "Theorem", + "25", + "Suppose", + "rows", + "entry", + "Gaussian", + "parameter", + "sqrt", + "log", + "infty", + "Let", + "phi", + "hat", + "min", + "suppose", + "bounded", + "Then", + "mathbb", + "geq", + "Proof", + "In", + "view", + "Lemma", + "22", + "show", + "max", + "32s", + "By", + "union", + "bound", + "24", + "begin", + "array", + "Big", + "sum", + "leq", + "exp", + "left", + "frac", + "end", + "Corollary", + "26", + "independent", + "distribution", + "diagonal", + "entries", + "minimum", + "eigenvalue", + "to1", + "provided", + "to0" + ], + "keywords": [ + "compatibility", + "sub-Gaussian", + "asymptotic" + ], + "connect_to": [] + } + }, + { + "chunk_id": "898aeb06-044b-5808-a09a-eb0a2ab79732", + "type": "text", + "content": "One of the most efficient ways of computing Lasso solutions is to use a optimisation technique called coordinate descent. This is a quite general way of minimising a function $f : R^{d} \\to R$ and works particularly well for functions of the form\n$$\nf (x) = g (x) + \\sum_ {j = 1} ^ {d} h _ {j} (x _ {j})\n$$\nwhere g is convex and differentiable and each $h_{j}: R \\to R$ is convex (and so continuous). We start with an initial guess of the minimiser $x^{(0)}$ (e.g. $x^{(0)} = 0$ ) and repeat for $m = 1, 2, \\ldots$ .\n$$\nx _ {1} ^ {(m)} = \\underset {x _ {1} \\in \\mathbb {R}} {\\arg \\min} f (x _ {1}, x _ {2} ^ {(m - 1)}, \\ldots , x _ {d} ^ {(m - 1)})\n$$\n$$\nx _ {2} ^ {(m)} = \\underset {x _ {2} \\in \\mathbb {R}} {\\arg \\min} f (x _ {1} ^ {(m)}, x _ {2}, x _ {3} ^ {(m - 1)}, \\ldots , x _ {d} ^ {(m - 1)})\n$$\n$$\n\\begin{array}{c} \\vdots \\\\ x _ {d} ^ {(m)} = \\underset {x _ {d} \\in \\mathbb {R}} {\\arg \\min} f (x _ {1} ^ {(m)}, x _ {2} ^ {(m)}, \\ldots , x _ {d - 1} ^ {(m)}, x _ {d}). \\end{array}\n$$\nTseng [2001] proves that provided $A_{0} = \\{x : f(x) \\leq f(x^{(0)})\\}$ is compact, then every converging subsequence of $x^{(m)}$ will converge to a minimiser of f.\nCorollary 27. Suppose $A^{0}$ is compact. Then\n(i) There exists a minimiser of $f$ , $x^{*}$ and $f(x^{(m)}) \\to f(x^{*})$ .\n(ii) If $x^{*}$ is the unique minimiser of $f$ then $x^{(m)} \\to x^{*}$ .\nProof. $f$ is a continuous function so it attains its infimum on the compact set $A_0$ . Suppose $f(x^{(m)}) \\nrightarrow f(x^*)$ . Then there exists $\\epsilon > 0$ and a subsequence $(x^{(m_j)})_{j=0}^{\\infty}$ such that $f(x^{(m_j)}) \\geq f(x^*) + \\epsilon$ for all $j$ . Note that since $f(x^{(m)}) \\leq f(x^{(m-1)})$ , we know that $x^{(m)} \\in A_0$ for all $m$ . Thus if $A_0$ is compact then any subsequence of $(x^{(m)})_{m=0}^{\\infty}$ has a further subsequence that converges by the Bolzano–Weierstrass theorem. Let $\\tilde{x}$ be the limit of the converging subsequence of $(x^{(m_j)})_{j=0}^{\\infty}$ . Then $f(\\tilde{x}) \\geq f(x^*) + \\epsilon$ , contradicting the result of Tseng [2001]. Thus (i) holds. The proof of (ii) is similar.\nWe often want to solve the Lasso on a grid of $\\lambda$ values $\\lambda_{0} > \\cdots > \\lambda_{L}$ (for the purposes of cross-validation for example). To do this, we can first solve for $\\lambda_{0}$ , and then solve at subsequent grid points by using the solution at the previous grid points as an initial guess (known as a warm start). An active set strategy can further speed up computation. This works as follows: For $l = 1, \\ldots, L$\n1. Initialise $A = \\{k : \\hat{\\beta}_{\\lambda_{l-1}, k}^{\\mathrm{L}} \\neq 0\\}$ .\n2. Perform coordinate descent only on coordinates in $A$ obtaining a solution $\\hat{\\beta}$ (all components $\\hat{\\beta}_k$ with $k \\notin A$ are set to zero).\n3. Let $V = \\{k : |X_{k}^{T}(Y - X\\hat{\\beta})|/n > \\lambda_{l}\\}$ , the set of coordinates which violate the KKT conditions when $\\hat{\\beta}$ is taken as a candidate solution.\n4. If $V$ is empty, we set $\\hat{\\beta}_{\\lambda_l}^{\\mathrm{L}} = \\hat{\\beta}$ . Else we update $A \\gets A \\cup V$ and return to 2.", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.6 Computation", + "metadata": { + "length": 3098, + "summary": "The text describes coordinate descent as an efficient optimization technique for Lasso solutions, minimizing functions of the form f(x)=g(x)+Σh_j(x_j) where g is convex and differentiable and each h_j is convex. It iteratively updates each coordinate by minimizing f with respect to that coordinate while holding others fixed. Tseng [2001] proves convergence to a minimizer under compactness of the sublevel set. Corollary 27 states existence of a minimizer and convergence of function values, with uniqueness ensuring convergence of iterates. The text also discusses solving Lasso over a grid of λ values using warm starts and an active set strategy: initializing active set A with non-zero coefficients from previous λ, performing coordinate descent only on A, checking KKT violations, and expanding A if violations exist.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "One", + "efficient", + "ways", + "computing", + "Lasso", + "solutions", + "optimisation", + "technique", + "called", + "coordinate", + "descent", + "This", + "general", + "minimising", + "function", + "works", + "functions", + "form", + "sum", + "convex", + "differentiable", + "continuous", + "We", + "start", + "initial", + "guess", + "minimiser", + "repeat", + "ldots", + "underset", + "mathbb", + "arg", + "min", + "begin", + "array", + "vdots", + "end", + "Tseng", + "2001", + "proves", + "provided", + "leq", + "compact", + "converging", + "subsequence", + "converge", + "Corollary", + "27", + "Suppose", + "Then", + "There", + "exists", + "ii", + "If", + "unique", + "Proof", + "attains", + "infimum", + "set", + "nrightarrow", + "epsilon", + "infty", + "geq", + "Note", + "Thus", + "converges", + "Bolzano", + "Weierstrass", + "theorem", + "Let", + "tilde", + "limit", + "contradicting", + "result", + "holds", + "The", + "proof", + "similar", + "solve", + "grid", + "lambda", + "values", + "cdots", + "purposes", + "cross", + "validation", + "To", + "subsequent", + "points", + "solution", + "previous", + "warm", + "An", + "active", + "strategy", + "speed", + "computation", + "For", + "Initialise", + "hat", + "beta", + "mathrm", + "neq", + "Perform", + "coordinates", + "obtaining", + "components", + "notin", + "violate", + "KKT", + "conditions", + "candidate", + "empty", + "Else", + "update", + "cup", + "return" + ], + "keywords": [ + "coordinate descent", + "Lasso", + "active set" + ], + "connect_to": [] + } + }, + { + "chunk_id": "ff30e85d-8864-5f93-a367-b51bc1d62134", + "type": "text", + "content": "We can add an $\\ell_{1}$ penalty to many other log-likelihoods, or more generally other loss functions besides the squared-error loss that arises from the normal linear model. For Lasso-penalised generalised linear models, such as logistic regression, similar theoretical results to those we have obtained are available and computations can proceed in a similar fashion to above.", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.7 Extensions of the Lasso", + "metadata": { + "length": 379, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "We", + "add", + "ell", + "penalty", + "log", + "likelihoods", + "generally", + "loss", + "functions", + "squared", + "error", + "arises", + "normal", + "linear", + "model", + "For", + "Lasso", + "penalised", + "generalised", + "models", + "logistic", + "regression", + "similar", + "theoretical", + "results", + "obtained", + "computations", + "proceed", + "fashion" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "773ba558-4492-5ec8-b31a-662fd47e6be0", + "type": "text", + "content": "The Lasso penalty encourages the estimated coefficients to be shrunk towards 0 and sometimes exactly to 0. Other penalty functions can be constructed to encourage different types of sparsity.", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.7 Extensions of the Lasso/3.7.1 Structural penalties", + "metadata": { + "length": 191, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "The", + "Lasso", + "penalty", + "encourages", + "estimated", + "coefficients", + "shrunk", + "Other", + "functions", + "constructed", + "encourage", + "types", + "sparsity" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0d73e37d-6a74-5b36-b9a6-4f88da80967f", + "type": "text", + "content": "Suppose we have a partition $G_{1},\\ldots ,G_{q}$ of $\\{1,\\dots ,p\\}$ (so $\\cup_{k = 1}^{q}G_{k} = \\{1,\\dots ,p\\}$ , $G_{j}\\cap G_{k} = \\emptyset$ for $j\\neq k$ ). The group Lasso penalty [Yuan and Lin, 2006] is given by\n$$\n\\lambda \\sum_ {j = 1} ^ {q} m _ {j} \\| \\beta_ {G _ {j}} \\| _ {2}.\n$$\nThe multipliers $m_{j} > 0$ serve to balance cases where the groups are of very different sizes; typically we choose $m_{j} = \\sqrt{|G_{j}|}$ . This penalty encourages either an entire group G to have $\\hat{\\beta}_{G} = 0$ or $\\hat{\\beta}_{k} \\neq 0$ for all $k \\in G$ . Such a property is useful when groups occur through coding for categorical predictors or when expanding predictors using basis functions.", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.7 Extensions of the Lasso/3.7.1 Structural penalties/Group Lasso", + "metadata": { + "length": 701, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Suppose", + "partition", + "ldots", + "dots", + "cup", + "cap", + "emptyset", + "neq", + "The", + "group", + "Lasso", + "penalty", + "Yuan", + "Lin", + "2006", + "lambda", + "sum", + "beta", + "multipliers", + "serve", + "balance", + "cases", + "groups", + "sizes", + "typically", + "choose", + "sqrt", + "This", + "encourages", + "entire", + "hat", + "Such", + "property", + "occur", + "coding", + "categorical", + "predictors", + "expanding", + "basis", + "functions" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "0bd4f4a5-1f23-5278-9bc0-ca33517d626c", + "type": "text", + "content": "If there is a sense in which the coefficients are ordered, so $\\beta_{j}^{0}$ is expected to be close to $\\beta_{j+1}^{0}$ , a fused Lasso penalty [Tibshirani et al., 2005] may be appropriate. This takes the form\n$$\n\\lambda_ {1} \\sum_ {j = 1} ^ {p - 1} | \\beta_ {j} - \\beta_ {j + 1} | + \\lambda_ {2} \\| \\beta \\| _ {1},\n$$\nwhere the second term may be omitted depending on whether shrinkage towards 0 is desired. As an example, consider the simple setting where $Y_{i} = \\mu_{i}^{0} + \\varepsilon_{i}$ , and it is thought that the $(\\mu_{i}^{0})_{i=1}^{n}$ form a piecewise constant sequence. Then one option is to minimise over $\\mu \\in R^{n}$ , the following objective\n$$\n\\frac {1}{n} \\| Y - \\mu \\| _ {2} ^ {2} + \\lambda \\sum_ {i = 1} ^ {n - 1} | \\mu_ {i} - \\mu_ {i + 1} |.\n$$", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.7 Extensions of the Lasso/3.7.1 Structural penalties/Fused Lasso", + "metadata": { + "length": 777, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "If", + "sense", + "coefficients", + "ordered", + "beta", + "expected", + "close", + "fused", + "Lasso", + "penalty", + "Tibshirani", + "al", + "2005", + "This", + "takes", + "form", + "lambda", + "sum", + "term", + "omitted", + "depending", + "shrinkage", + "desired", + "As", + "simple", + "setting", + "mu", + "varepsilon", + "thought", + "piecewise", + "constant", + "sequence", + "Then", + "option", + "minimise", + "objective", + "frac" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8c0077ed-dba5-51da-b797-99172712ec6b", + "type": "text", + "content": "One potential drawback of the Lasso is that the same shrinkage effect that sets many estimated coefficients exactly to zero also shrinks all non-zero estimated coefficients towards zero. One possible solution is to take $\\hat{S}_{\\lambda} = \\{k : \\hat{\\beta}_{\\lambda,k}^{\\mathrm{L}} \\neq 0\\}$ and then re-estimate $\\beta_{\\hat{S}_{\\lambda}}^{0}$ by OLS regression on $X_{\\hat{S}_{\\lambda}}$ .\nAnother option is to re-estimate using the Lasso on $X_{\\hat{S}_{\\lambda}}$ ; this procedure is known as the relaxed Lasso [Meinshausen, 2007]. The adaptive Lasso [Zou, 2006] takes an initial estimate of $\\beta^{0}$ , $\\hat{\\beta}^{\\mathrm{init}}$ (e.g. from the Lasso) and then performs weighted Lasso regression:\n$$\n\\hat {\\beta} _ {\\lambda} ^ {\\mathrm{adapt}} = \\underset {\\beta \\in \\mathbb {R} ^ {p}: \\beta_ {\\hat {S} _ {\\mathrm{init}} ^ {c}} = 0} {\\arg \\min} \\left\\{\\frac {1}{2 n} \\| Y - X \\beta \\| _ {2} ^ {2} + \\lambda \\sum_ {k \\in \\hat {S} _ {\\mathrm{init}}} \\frac {| \\beta_ {k} |}{| \\hat {\\beta} _ {k} ^ {\\mathrm{init}} |} \\right\\},\n$$\nwhere $\\hat{S}_{\\mathrm{init}} = \\{k:\\hat{\\beta}_{k}^{\\mathrm{init}}\\neq 0\\}$ .\nYet another approach involves using a family of non-convex penalty functions $p_{\\lambda,\\gamma} : [0, \\infty) \\to [0, \\infty)$ and attempting to minimise\n$$\n\\frac {1}{2 n} \\| Y - X \\beta \\| _ {2} ^ {2} + \\sum_ {k = 1} ^ {p} p _ {\\lambda , \\gamma} (| \\beta_ {k} |).\n$$\nA prominent example is the minimax concave penalty (MCP) [Zhang, 2010] which takes\n$$\np _ {\\lambda , \\gamma} ^ {\\prime} (u) = \\left(\\lambda - \\frac {u}{\\gamma}\\right) _ {+}.\n$$\nOne disadvantage of using a non-convex penalty is that there may be multiple local minima which can make optimisation problematic. However, typically if the non-convexity is not too severe, coordinate descent can produce reasonable results.", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.7 Extensions of the Lasso/3.7.2 Reducing the bias of the Lasso", + "metadata": { + "length": 1804, + "summary": "The passage discusses drawbacks of the Lasso, which shrinks non-zero coefficients toward zero. Solutions include re-estimating non-zero coefficients via OLS or the relaxed Lasso, using the adaptive Lasso with weighted penalties, and employing non-convex penalties like MCP. Non-convex penalties may cause multiple local minima, but coordinate descent can work if non-convexity is mild.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "One", + "potential", + "drawback", + "Lasso", + "shrinkage", + "effect", + "sets", + "estimated", + "coefficients", + "shrinks", + "solution", + "hat", + "lambda", + "beta", + "mathrm", + "neq", + "estimate", + "OLS", + "regression", + "Another", + "option", + "procedure", + "relaxed", + "Meinshausen", + "2007", + "The", + "adaptive", + "Zou", + "2006", + "takes", + "initial", + "init", + "performs", + "weighted", + "adapt", + "underset", + "mathbb", + "arg", + "min", + "left", + "frac", + "sum", + "Yet", + "approach", + "involves", + "family", + "convex", + "penalty", + "functions", + "gamma", + "infty", + "attempting", + "minimise", + "prominent", + "minimax", + "concave", + "MCP", + "Zhang", + "2010", + "prime", + "disadvantage", + "multiple", + "local", + "minima", + "make", + "optimisation", + "problematic", + "However", + "typically", + "convexity", + "severe", + "coordinate", + "descent", + "produce", + "reasonable", + "results" + ], + "keywords": [ + "Lasso", + "Relaxed Lasso", + "Adaptive Lasso" + ], + "connect_to": [] + } + }, + { + "chunk_id": "2e4a2c6f-7a22-594c-a219-7dedfd91b81c", + "type": "text", + "content": "Graphical modelling and causal inference\nSo far we have considered the problem of relating a particular response to a large collection of explanatory variables.\nIn some settings however, we do not have a distinguished response variable and instead we would like to better understand relationships between all the variables. In other situations, rather than being able to predict variables, we would like to understand causal relationships between them. Representing relationships between random variables through graphs will be an important tool in tackling these problems.", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference", + "metadata": { + "length": 573, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Graphical", + "modelling", + "causal", + "inference", + "So", + "considered", + "problem", + "relating", + "response", + "large", + "collection", + "explanatory", + "variables", + "In", + "settings", + "distinguished", + "variable", + "understand", + "relationships", + "situations", + "predict", + "Representing", + "random", + "graphs", + "important", + "tool", + "tackling", + "problems" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f888b102-0866-5746-aac9-60cea775f239", + "type": "image", + "content": "\nLet $Z = (Z_{1},\\ldots ,Z_{p})^{T}$ be a collection of random variables. The graphs we will consider will always have $V = \\{1,\\dots ,p\\}$ so $V$ indexes the random variables.

    0\n[images/image-6-Let $Z = (.jpg]\n", + "path": "images/image-6-Let $Z = (.jpg", + "metadata": { + "length": 220, + "summary": "image-6\nLet $Z = (Z_{1},\\ldots ,Z_{p})^{T}$ be a collection of random variables. The graphs we will consider will always have $V = \\{1,\\dots ,p\\}$ so $V$ indexes the random variables.
    0", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "file_path": "images/image-6-Let $Z = (.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "0c02156c-3322-5b54-ac09-f66a9231c59d", + "type": "image", + "content": "\nZ2 --> Z3
    0\n[images/image-7-Z2 --_ Z3 .jpg]\n", + "path": "images/image-7-Z2 --_ Z3 .jpg", + "metadata": { + "length": 54, + "summary": "image-7\nZ2 --> Z3
    0", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "file_path": "images/image-7-Z2 --_ Z3 .jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "0d1020da-cba0-5605-aaa8-61a20162792b", + "type": "text", + "content": "Definition 7. A graph is a pair $\\mathcal{G} = (V, E)$ where $V$ is a set of vertices or nodes and $E \\subseteq V \\times V$ with $(v, v) \\notin E$ for any $v \\in V$ is a set of edges.\nLet $Z = (Z_{1},\\ldots ,Z_{p})^{T}$ be a collection of random variables. The graphs we will consider will always have $V = \\{1,\\dots ,p\\}$ so $V$ indexes the random variables.\nLet $j, k \\in V$ .\n- We say there is an edge between $j$ and $k$ and that $j$ and $k$ are adjacent if either $(j, k) \\in E$ or $(k, j) \\in E$ .\n- An edge $(j,k)$ is undirected if also $(k,j) \\in E$ . We then sometimes write $\\{j,k\\} \\in E$ and $j - k$ to represent this. Otherwise if only $(j,k) \\in E$ the edge is directed and we may write $j \\to k$ to represent this.\n- If all edges in the graph are (un)directed we call it an (un)directed graph. We can represent graphs as pictures: for example, we can draw the graph when $p = 4$ and $E = \\{(2,1), (3,4), (2,3)\\}$ as\n\nLet $Z = (Z_{1},\\ldots ,Z_{p})^{T}$ be a collection of random variables. The graphs we will consider will always have $V = \\{1,\\dots ,p\\}$ so $V$ indexes the random variables.
    0\n[images/image-6-Let $Z = (.jpg]\n\n
    \nflowchart\n```mermaid\ngraph TD\nZ1 --> Z2\nZ2 --> Z3\nZ3 --> Z4\n```\n
    \nIf instead we have $E = \\{\\{1,2\\}, \\{2,4\\}\\}$ we get the undirected graph\n\nZ2 --> Z3
    0\n[images/image-7-Z2 --_ Z3 .jpg]\n\n
    \nflowchart\n```mermaid\ngraph TD\nZ1[\"Z₁\"] --> Z2[\"Z₂\"]\nZ3[\"Z₃\"] --> Z4[\"Z₄\"]\n```\n
    \n- A graph $\\mathcal{G}_1 = (V_1, E_1)$ is a subgraph of $\\mathcal{G} = (V, E)$ if $V_1 \\subseteq V$ and $E_1 \\subseteq E$ and a proper subgraph if either of these are proper inclusions.\n- Say $j$ is a parent of $k$ and $k$ is a child of $j$ if $j \\to k$ . The sets of parents and children of $k$ will be denoted $\\mathrm{pa}(k)$ and $\\mathrm{ch}(k)$ respectively.\n- A set of three nodes is called a v-structure if one node is a child of the two other nodes, and these two nodes are not adjacent.\n- The skeleton of $\\mathcal{G}$ is a copy of $\\mathcal{G}$ with every edge replaced by an undirected edge.\n- A path from $j$ to $k$ is a sequence $j = j_1, j_2, \\ldots, j_m = k$ of (at least two) distinct vertices such that $j_l$ and $j_{l+1}$ are adjacent. Such a path is a directed path if $j_l \\to j_{l+1}$ for all $l$ . We then call $k$ a descendant of $j$ . The set of descendants of $j$ will be denoted $\\mathrm{de}(j)$ . If $j_{l-1} \\to j_l \\leftarrow j_{l+1}$ , $j_l$ is called a collider (relative to the path).\n- A directed cycle is (almost) a directed path but with the start and end points the same. A partially directed acyclic graph (PDAG) is a graph containing no directed cycles. A directed acyclic graph (DAG) is a directed graph containing no directed cycles.\n- In a DAG, a path between $j_1$ and $j_m$ ( $j_1, j_2, \\ldots, j_m$ ) is blocked by a set $S$ with neither $j_1$ nor $j_m$ in $S$ whenever there is a node $j_l$ such that one of the following two possibilities hold:\n1. $j_l \\in S$ and we don't have $j_{l-1} \\to j_l \\leftarrow j_{l+1}$\n2. $j_{l-1} \\rightarrow j_l \\leftarrow j_{l+1}$ and neither $j_l$ nor any of its descendants are in S.\n\\- If $\\mathcal{G}$ is a DAG, given a triple of subsets of nodes $A, B, S$ , we say $S$ $d$ -separates $A$ from $B$ if $S$ blocks every path from $A$ to $B$ .\n\\- The moralised graph of a DAG $\\mathcal{G}$ is the undirected graph obtained by adding edges between (marrying) the parents of each node and removing all edge directions.\nDefinition 8. Given a DAG $\\mathcal{G}$ with $V = \\{1, \\ldots, p\\}$ , we say that a permutation $\\pi$ of $V$ is a topological (or causal) ordering of the variables if it satisfies\n$$\n\\pi (j) < \\pi (k) \\qquad \\mathrm{whenever} k \\in \\mathrm{de} (j).\n$$\nProposition 28. Every DAG has a topological ordering.\nProof. We use induction on the number of nodes p. Clearly the result is true when p = 1.\nNow we show that in any DAG, we can find a node with no parents. Pick any node and move to one of its parents, if possible. Then move to one of the new node's parents, and continue in this fashion. This procedure must terminate since no node can be visited twice, or we would have found a cycle. The final node we visit must therefore have no parents, which we call a source node.\nSuppose then that $p \\geq 2$ , and we know that all DAGs with p-1 nodes have a topological ordering. Find a source s (wlog s = p) and form a new DAG $\\tilde{G}$ with p-1 nodes by removing the source (and all edges emanating from it). Note we keep the labelling of the nodes in this new DAG the same. This smaller DAG must have a topological order $\\tilde{\\pi}$ . A topological ordering $\\pi$ for our original DAG is then given by $\\pi(s) = 1$ and $\\pi(k) = \\tilde{\\pi}(k) + 1$ for $k \\neq s$ . ☐", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.1 Graphs", + "metadata": { + "length": 4772, + "summary": "The text defines fundamental graph concepts including directed/undirected edges, subgraphs, parents/children, v-structures, skeletons, paths, directed cycles, DAGs, d-separation, and moralised graphs. It also defines topological ordering and proves every DAG has one via induction and source node removal.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Definition", + "graph", + "pair", + "mathcal", + "set", + "vertices", + "nodes", + "subseteq", + "times", + "notin", + "edges", + "Let", + "ldots", + "collection", + "random", + "variables", + "The", + "graphs", + "dots", + "indexes", + "We", + "edge", + "adjacent", + "An", + "undirected", + "write", + "represent", + "Otherwise", + "directed", + "If", + "call", + "pictures", + "draw", + "details", + "summary", + "flowchart", + "mermaid", + "TD", + "Z1", + "Z2", + "Z3", + "Z4", + "subgraph", + "proper", + "inclusions", + "Say", + "parent", + "child", + "sets", + "parents", + "children", + "denoted", + "mathrm", + "pa", + "ch", + "called", + "structure", + "node", + "skeleton", + "copy", + "replaced", + "path", + "sequence", + "distinct", + "Such", + "descendant", + "descendants", + "de", + "leftarrow", + "collider", + "relative", + "cycle", + "start", + "end", + "points", + "partially", + "acyclic", + "PDAG", + "cycles", + "DAG", + "In", + "blocked", + "possibilities", + "hold", + "don", + "rightarrow", + "triple", + "subsets", + "separates", + "blocks", + "moralised", + "obtained", + "adding", + "marrying", + "removing", + "directions", + "Given", + "permutation", + "pi", + "topological", + "causal", + "ordering", + "satisfies", + "qquad", + "Proposition", + "28", + "Every", + "Proof", + "induction", + "number", + "Clearly", + "result", + "true", + "Now", + "show", + "find", + "Pick", + "move", + "Then", + "continue", + "fashion", + "This", + "procedure", + "terminate", + "visited", + "found", + "final", + "visit", + "source", + "Suppose", + "geq", + "DAGs", + "Find", + "wlog", + "form", + "tilde", + "emanating", + "Note", + "labelling", + "smaller", + "order", + "original", + "neq" + ], + "keywords": [ + "DAG", + "graph theory", + "topological ordering" + ], + "connect_to": [ + { + "target": "f888b102-0866-5746-aac9-60cea775f239", + "relation": "embeds", + "ref": "[images/image-6-Let $Z = (.jpg]", + "position": { + "start": 1119, + "end": 1150 + } + }, + { + "target": "0c02156c-3322-5b54-ac09-f66a9231c59d", + "relation": "embeds", + "ref": "[images/image-7-Z2 --_ Z3 .jpg]", + "position": { + "start": 1352, + "end": 1383 + } + } + ] + } + }, + { + "chunk_id": "ff423b86-f1f0-5195-9418-c213481131f4", + "type": "text", + "content": "We would like to understand which variables may be ‘related’ to each other. Trying to find pairs of variables that are independent and so unlikely to be related to each other is not necessarily a good way to proceed as each variable may be correlated with a large number of variables without being directly related to them. A better approach is to use conditional independence.\nDefinition 9. If X, Y and Z are random vectors with a joint density $f_{XYZ}$ (w.r.t. a product measure $\\mu$ ) then we say X is conditionally independent of Y given Z, and write\n$$\nX \\perp Y | Z\n$$\nif\n$$\nf _ {X Y | Z} (x, y | z) = f _ {X | Z} (x | z) f _ {Y | Z} (y | z).\n$$\nEquivalently\n$$\nX \\perp \\perp Y | Z \\Longleftrightarrow f _ {X | Y Z} (x | y, z) = f _ {X | Z} (x | z).\n$$\nWe will first look at how undirected graphs can be used to visualise conditional independencies between random variables; thus in the next few subsections by graph we will mean undirected graph.\nDefinition 10. Let $Z = (Z_{1}, \\ldots, Z_{p})^{T}$ be a collection of random variables with joint law P. The conditional independence graph (CIG) for P is the graph with $V = \\{1, \\ldots, p\\}$ and an edge $\\{j, k\\}$ between j and k if and only if\n$$\nZ _ {j} \\nparallel Z _ {k} \\mid Z _ {- j k}.\n$$", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.2 Conditional independence graphs", + "metadata": { + "length": 1254, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "We", + "understand", + "variables", + "related", + "Trying", + "find", + "pairs", + "independent", + "necessarily", + "good", + "proceed", + "variable", + "correlated", + "large", + "number", + "directly", + "approach", + "conditional", + "independence", + "Definition", + "If", + "random", + "vectors", + "joint", + "density", + "XYZ", + "product", + "measure", + "mu", + "conditionally", + "write", + "perp", + "Equivalently", + "Longleftrightarrow", + "undirected", + "graphs", + "visualise", + "independencies", + "subsections", + "graph", + "10", + "Let", + "ldots", + "collection", + "law", + "The", + "CIG", + "edge", + "nparallel", + "mid" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "8a4ef36d-8ee9-5ca0-9ab3-dc1b154dab6c", + "type": "text", + "content": "Estimating the CIG given samples from P is a difficult task in general. However, in the case where P is multivariate Gaussian, things simplify considerably as we shall see. We begin with some notation. For a matrix $M \\in R^{p \\times p}$ , and sets $A, B \\subseteq \\{1, \\ldots, p\\}$ , let $M_{A,B}$ be the $|A| \\times |B|$ submatrix of M consisting of those rows and columns of M indexed by the sets A and B respectively. The submatrix extraction operation is always performed first (so e.g. $M_{k,-k}^{T} = (M_{k,-k})^{T}$ ).", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.3 Gaussian graphical models", + "metadata": { + "length": 526, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Estimating", + "CIG", + "samples", + "difficult", + "task", + "general", + "However", + "case", + "multivariate", + "Gaussian", + "things", + "simplify", + "considerably", + "We", + "begin", + "notation", + "For", + "matrix", + "times", + "sets", + "subseteq", + "ldots", + "submatrix", + "consisting", + "rows", + "columns", + "indexed", + "The", + "extraction", + "operation", + "performed" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5bbb46e4-5a8d-5439-bc4b-6476acc723aa", + "type": "text", + "content": "Now let $Z \\sim N_p(\\mu, \\Sigma)$ with $\\Sigma$ positive definite. Note $\\Sigma_{A,A}$ is also positive definite for any $A$ .\nProposition 29.\n$$\nZ _ {A} | Z _ {B} = z _ {B} \\sim N _ {| A |} (\\mu_ {A} + \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} (z _ {B} - \\mu_ {B}), \\Sigma_ {A, A} - \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} \\Sigma_ {B, A})\n$$\nProof. Idea: write $Z_{A}=MZ_{B}+(Z_{A}-MZ_{B})$ with matrix $M\\in R^{|A|\\times|B|}$ such that $Z_{A}-MZ_{B}$ and $Z_{B}$ are independent, i.e. such that\n$$\n\\operatorname{Cov} (Z _ {B}, Z _ {A} - M Z _ {B}) = \\Sigma_ {B, A} - \\Sigma_ {B, B} M ^ {T} = 0.\n$$\nThis occurs when we take $M^{T} = \\Sigma_{B,B}^{-1}\\Sigma_{B,A}$ . Because $Z_{A} - MZ_{B}$ and $Z_{B}$ are independent, the distribution of $Z_{A} - MZ_{B}$ conditional on $Z_{B} = z_{B}$ is equal to its unconditional distribution. Now\n$$\n\\begin{array}{l} \\mathbb {E} (Z _ {A} - M Z _ {B}) = \\mu_ {A} - \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} \\mu_ {B} \\\\ \\mathrm{Var} (Z _ {A} - M Z _ {B}) = \\Sigma_ {A, A} + \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} \\Sigma_ {B, B} \\Sigma_ {B, B} ^ {- 1} \\Sigma_ {B, A} - 2 \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} \\Sigma_ {B, A} \\\\ = \\Sigma_ {A, A} - \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} \\Sigma_ {B, A}. \\\\ \\end{array}\n$$\nSince $MZ_{B}$ is a function of $Z_{B}$ and $Z_{A}-MZ_{B}$ is normally distributed, we have the result. ☐", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.3 Gaussian graphical models/4.3.1 Normal conditionals", + "metadata": { + "length": 1349, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Now", + "sim", + "mu", + "Sigma", + "positive", + "definite", + "Note", + "Proposition", + "29", + "Proof", + "Idea", + "write", + "MZ", + "matrix", + "times", + "independent", + "operatorname", + "Cov", + "This", + "occurs", + "Because", + "distribution", + "conditional", + "equal", + "unconditional", + "begin", + "array", + "mathbb", + "mathrm", + "Var", + "end", + "Since", + "function", + "distributed", + "result" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "7bf71d7a-6246-53a9-b8ee-d0890013e3e8", + "type": "text", + "content": "Specialising to the case where $A = \\{k\\}$ and $B = A^{c}$ we see that when conditioning on $Z_{-k} = z_{-k}$ , we may write\n$$\nZ _ {k} = m _ {k} + z _ {- k} ^ {T} \\Sigma_ {- k, - k} ^ {- 1} \\Sigma_ {- k, k} + \\varepsilon_ {k},\n$$\nwhere\n$$\n\\begin{array}{l} m _ {k} = \\mu_ {k} - \\Sigma_ {k, - k} \\Sigma_ {- k, - k} ^ {- 1} \\mu_ {- k} \\\\ \\varepsilon_ {k} | Z _ {- k} = z _ {- k} \\sim N (0, \\Sigma_ {k, k} - \\Sigma_ {k, - k} \\Sigma_ {- k, - k} ^ {- 1} \\Sigma_ {- k, k}). \\\\ \\end{array}\n$$\nNote that if the $j$ th element of the vector of coefficients $\\Sigma_{-k,-k}^{-1}\\Sigma_{-k,k}$ is zero, then the distribution of $Z_{k}$ conditional on $Z_{-k}$ will not depend at all on the $j$ th component of $Z_{-k}$ . Then if that $j$ th component was $Z_{j'}$ , we would have that $Z_{k}|Z_{-k}=z_{-k}$ has the same distribution as $Z_{k}|Z_{-j'k}=z_{-j'k}$ , so $Z_{k}\\perp Z_{j}|Z_{-j'k}$ .\nThus given $x_{1},\\ldots,x_{n}\\stackrel{\\mathrm{i.i.d.}}{\\sim}Z$ and writing\n$$\nX = \\left( \\begin{array}{c} x _ {1} ^ {T} \\\\ \\vdots \\\\ x _ {n} ^ {T} \\end{array} \\right),\n$$\nwe may estimate the coefficient vector $\\Sigma_{-k,-k}^{-1}\\Sigma_{-k,k}$ by regressing $X_{k}$ on $X_{\\{k\\}^{c}}$ and including an intercept term.\nThe technique of neighbourhood selection [Meinshausen and Bühlmann, 2006] involves performing such a regression for each variable, using the Lasso. There are two options for populating our estimate of the CIG with edges based on the Lasso estimates. Writing $\\hat{S}_k$ for the selected set of variables when regressing $X_{k}$ on $X_{\\{k\\}^c}$ , we can use the \"OR\" rule and put an edge between vertices $j$ and $k$ if and only if $k \\in \\hat{S}_j$ or $j \\in \\hat{S}_k$ . An alternative is the \"AND\" rule where we put an edge between $j$ and $k$ if and only if $k \\in \\hat{S}_j$ and $j \\in \\hat{S}_k$ .\nAnother popular approach to estimating the CIG works by first directly estimating $\\Omega$ , as we'll now see.", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.3 Gaussian graphical models/4.3.2 Nodewise regression", + "metadata": { + "length": 1921, + "summary": "The passage discusses neighborhood selection for estimating a conditional independence graph (CIG). It derives the conditional distribution of Z_k given Z_{-k} in a Gaussian setting, showing that zero coefficients in the regression imply conditional independence. It then describes using Lasso regression of each variable on all others to select neighbors, with OR and AND rules for edge inclusion. Finally, it mentions an alternative approach of directly estimating the precision matrix Omega.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Specialising", + "case", + "conditioning", + "write", + "Sigma", + "varepsilon", + "begin", + "array", + "mu", + "sim", + "end", + "Note", + "element", + "vector", + "coefficients", + "distribution", + "conditional", + "depend", + "component", + "Then", + "perp", + "Thus", + "ldots", + "stackrel", + "mathrm", + "writing", + "left", + "vdots", + "estimate", + "coefficient", + "regressing", + "including", + "intercept", + "term", + "The", + "technique", + "neighbourhood", + "selection", + "Meinshausen", + "hlmann", + "2006", + "involves", + "performing", + "regression", + "variable", + "Lasso", + "There", + "options", + "populating", + "CIG", + "edges", + "based", + "estimates", + "Writing", + "hat", + "selected", + "set", + "variables", + "OR", + "rule", + "put", + "edge", + "vertices", + "An", + "alternative", + "AND", + "Another", + "popular", + "approach", + "estimating", + "works", + "directly", + "Omega", + "ll" + ], + "keywords": [ + "CIG", + "Lasso", + "Conditional Independence" + ], + "connect_to": [] + } + }, + { + "chunk_id": "c98331ce-c03a-5862-8514-a15c38c6f191", + "type": "text", + "content": "The following facts about blockwise inversion of matrices will help us to interpret the mean and variance in Proposition 29.\nProposition 30. Let $M \\in R^{p \\times p}$ be a symmetric positive definite matrix and suppose\n$$\nM = \\left( \\begin{array}{c c} P & Q ^ {T} \\\\ Q & R \\end{array} \\right)\n$$\nwith P and R square matrices. The Schur complement of R is $P - Q^{T}R^{-1}Q =: S$ . We have that S is positive definite and\n$$\nM ^ {- 1} = \\left( \\begin{array}{c c} S ^ {- 1} & - S ^ {- 1} Q ^ {T} R ^ {- 1} \\\\ - R ^ {- 1} Q S ^ {- 1} & R ^ {- 1} + R ^ {- 1} Q S ^ {- 1} Q ^ {T} R ^ {- 1} \\end{array} \\right).\n$$\nFurthermore $\\det(M) = \\det(S)\\det(R)$ .\nLet $\\Omega = \\Sigma^{-1}$ be the precision matrix. Note that $\\Sigma_{k,k} - \\Sigma_{k,-k}\\Sigma_{-k,-k}^{-1}\\Sigma_{-k,k} = \\Omega_{kk}^{-1}$ , and more generally that $\\mathrm{Var}(Z_A|Z_{A^c}) = \\Omega_{A,A}^{-1}$ . Also, we see that $\\Sigma_{-k,-k}^{-1}\\Sigma_{-k,k} = -\\Omega_{kk}^{-1}\\Omega_{-k,k}$ , so\n$$\n(\\Sigma_ {- k, - k} ^ {- 1} \\Sigma_ {- k, k}) _ {j} = 0 \\Leftrightarrow \\left\\{ \\begin{array}{l l} \\Omega_ {j, k} = 0 \\text {for} j < k \\\\ \\Omega_ {j + 1, k} = 0 \\text {for} j \\geq k. \\end{array} \\right.\n$$\nThus\n$$\nZ _ {k} \\perp \\perp Z _ {j} | Z _ {- j k} \\Leftrightarrow \\Omega_ {j k} = 0.\n$$\nThis motivates another approach to estimating the CIG.", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.3 Gaussian graphical models/4.3.3 The precision matrix and conditional independence", + "metadata": { + "length": 1316, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "The", + "facts", + "blockwise", + "inversion", + "matrices", + "interpret", + "variance", + "Proposition", + "29", + "30", + "Let", + "times", + "symmetric", + "positive", + "definite", + "matrix", + "suppose", + "left", + "begin", + "array", + "end", + "square", + "Schur", + "complement", + "We", + "Furthermore", + "det", + "Omega", + "Sigma", + "precision", + "Note", + "kk", + "generally", + "mathrm", + "Var", + "Also", + "Leftrightarrow", + "text", + "geq", + "Thus", + "perp", + "This", + "motivates", + "approach", + "estimating", + "CIG" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a2502a8d-7d21-5228-b04e-2c7c52bcb6c3", + "type": "text", + "content": "Recall that the density of $N_{p}(\\mu ,\\Sigma)$ is\n$$\nf (z) = \\frac {1}{(2 \\pi) ^ {p / 2} \\mathrm{det} (\\Sigma) ^ {1 / 2}} \\exp \\bigg (- \\frac {1}{2} (z - \\mu) ^ {T} \\Sigma^ {- 1} (z - \\mu) \\bigg).\n$$\nThe log-likelihood of $(\\mu, \\Sigma)$ based on an i.i.d. sample $x_{1}, \\ldots, x_{n}$ is\n$$\n\\ell (\\mu , \\Omega) = \\frac {n}{2} \\log \\det (\\Omega) - \\frac {1}{2} \\sum_ {i = 1} ^ {n} (x _ {i} - \\mu) ^ {T} \\Omega (x _ {i} - \\mu).\n$$\nWrite\n$$\n\\bar {X} = \\frac {1}{n} \\sum_ {i = 1} ^ {n} x _ {i}, \\qquad S = \\frac {1}{n} \\sum_ {i = 1} ^ {n} (x _ {i} - \\bar {X}) (x _ {i} - \\bar {X}) ^ {T}.\n$$\nThen\n$$\n\\begin{array}{l} \\sum_ {i = 1} ^ {n} (x _ {i} - \\mu) ^ {T} \\Omega (x _ {i} - \\mu) = \\sum_ {i = 1} ^ {n} (x _ {i} - \\bar {X} + \\bar {X} - \\mu) ^ {T} \\Omega (x _ {i} - \\bar {X} + \\bar {X} - \\mu) \\\\ = \\sum_ {i = 1} ^ {n} (x _ {i} - \\bar {X}) ^ {T} \\Omega (x _ {i} - \\bar {X}) + n (\\bar {X} - \\mu) ^ {T} \\Omega (\\bar {X} - \\mu) \\\\ + 2 \\sum_ {i = 1} ^ {n} (x _ {i} - \\bar {X}) ^ {T} \\Omega (\\bar {X} - \\mu). \\\\ \\end{array}\n$$\nAlso,\n$$\n\\begin{array}{l} \\sum_ {i = 1} ^ {n} (x _ {i} - \\bar {X}) ^ {T} \\Omega (x _ {i} - \\bar {X}) = \\sum_ {i = 1} ^ {n} \\mathrm{tr} \\{(x _ {i} - \\bar {X}) ^ {T} \\Omega (x _ {i} - \\bar {X}) \\} \\\\ = \\sum_ {i = 1} ^ {n} \\operatorname{tr} \\left\\{\\left(x _ {i} - \\bar {X}\\right) \\left(x _ {i} - \\bar {X}\\right) ^ {T} \\Omega \\right\\} \\\\ = n \\operatorname{tr} (S \\Omega). \\\\ \\end{array}\n$$\nThus\n$$\n\\ell (\\mu , \\Omega) = - \\frac {n}{2} \\{\\mathrm{tr} (S \\Omega) - \\log \\det (\\Omega) + (\\bar {X} - \\mu) ^ {T} \\Omega (\\bar {X} - \\mu) \\}\n$$\nand\n$$\n\\max _ {\\mu \\in \\mathbb {R} ^ {p}} \\ell (\\mu , \\Omega) = - \\frac {n}{2} \\{\\mathrm{tr} (S \\Omega) - \\log \\det (\\Omega) \\}.\n$$\nHence the maximum likelihood estimate of $\\Omega$ , $\\hat{\\Omega}^{ML}$ can be obtained by solving\n$$\n\\min _ {\\Omega : \\Omega \\succ 0} \\{- \\log \\det (\\Omega) + \\operatorname{tr} (S \\Omega) \\},\n$$\nwhere $\\Omega \\succ 0$ means $\\Omega$ is positive definite. One can show that the objective is convex and we are minimising over a convex set. As\n$$\n\\begin{array}{l} \\frac {\\partial}{\\partial \\Omega_ {j k}} \\log \\det (\\Omega) = (\\Omega^ {- 1}) _ {k j} = (\\Omega^ {- 1}) _ {j k}, \\\\ \\frac {\\partial}{\\partial \\Omega_ {j k}} \\mathrm{tr} (S \\Omega) = S _ {k j} = S _ {j k}, \\\\ \\end{array}\n$$\nif $X$ has full column rank so $S$ is positive definite, $\\hat{\\Omega}^{ML} = S^{-1}$ .\nThe graphical Lasso [Yuan and Lin, 2007] penalises the log-likelihood for $\\Omega$ and solves\n$$\n\\min _ {\\Omega : \\Omega \\succ 0} \\{- \\log \\det (\\Omega) + \\operatorname{tr} (S \\Omega) + \\lambda \\| \\Omega \\| _ {1} \\},\n$$\nwhere $\\|\\Omega\\|_{1}=\\sum_{j,k}|\\Omega_{jk}|$ ; this results in a sparse estimate of the precision matrix from which an estimate of the CIG can be constructed. Often the $\\|\\Omega\\|_{1}$ is modified such that the diagonal elements are not penalised.", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.3 Gaussian graphical models/4.3.4 The Graphical Lasso", + "metadata": { + "length": 2826, + "summary": "The passage derives the maximum likelihood estimate of the precision matrix Ω for a multivariate normal distribution, showing that the unpenalized estimate is S^{-1} when S is positive definite. It then introduces the graphical Lasso, which adds an L1 penalty λ‖Ω‖₁ to the negative log-likelihood to promote sparsity in Ω, enabling estimation of a conditional independence graph. The objective is convex, and diagonal elements are often not penalized.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Recall", + "density", + "mu", + "Sigma", + "frac", + "pi", + "mathrm", + "det", + "exp", + "bigg", + "The", + "log", + "likelihood", + "based", + "sample", + "ldots", + "ell", + "Omega", + "sum", + "Write", + "bar", + "qquad", + "Then", + "begin", + "array", + "end", + "Also", + "tr", + "operatorname", + "left", + "Thus", + "max", + "mathbb", + "Hence", + "maximum", + "estimate", + "hat", + "ML", + "obtained", + "solving", + "min", + "succ", + "means", + "positive", + "definite", + "One", + "show", + "objective", + "convex", + "minimising", + "set", + "As", + "partial", + "full", + "column", + "rank", + "graphical", + "Lasso", + "Yuan", + "Lin", + "2007", + "penalises", + "solves", + "lambda", + "jk", + "results", + "sparse", + "precision", + "matrix", + "CIG", + "constructed", + "Often", + "modified", + "diagonal", + "elements", + "penalised" + ], + "keywords": [ + "Precision matrix", + "Sparsity", + "Log-likelihood" + ], + "connect_to": [] + } + }, + { + "chunk_id": "32181875-bd81-5d8d-a62e-100d18993cf0", + "type": "image", + "content": "\nExample 4.4.1. Consider the following (totally artificial) SEM which has whether you are taking this course $(Z_{1}=1)$ depending on whether you went to the statistical modelling course $(Z_{2}=1)$ and whether you have heard about machine learning $(Z_{3}=1)$ . Suppose
    0\n[images/image-8-Example 4..jpg]\n", + "path": "images/image-8-Example 4..jpg", + "metadata": { + "length": 314, + "summary": "image-8\nExample 4.4.1. Consider the following (totally artificial) SEM which has whether you are taking this course $(Z_{1}=1)$ depending on whether you went to the statistical modelling course $(Z_{2}=1)$ and whether you have heard about machine learning $(Z_{3}=1)$ . Suppose
    0", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "file_path": "images/image-8-Example 4..jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "7779d1ba-ce37-5fb6-913d-c23af50d0fdb", + "type": "text", + "content": "Conditional independence graphs give us some understanding of the relationships between variables. However they do not tell us how, if we were to set the kth variable to a particular value, say 0.5, then how the distribution of the other values would be altered. Yet this is often the sort of question that we would like to answer.\nIn order to reach this more ambitious goal, we introduce the notion of structural equation models (SEMs). These give a way of representing the data generating process. We will now have to make use of not just undirected graphs but other sorts of graphs (and particularly DAGs), so by graph we will now mean any sort of graph satisfying definition 7.\nDefinition 11. A structural equation model S for a random vector $Z \\in R^{p}$ is a collection of p equations\n$$\nZ _ {k} = h _ {k} (Z _ {P _ {k}}, \\varepsilon_ {k}), \\qquad k = 1, \\ldots , p\n$$\nwhere\n- $\\varepsilon_1, \\ldots, \\varepsilon_p$ are all independent random variables;\n- $P_k \\subseteq \\{1, \\ldots, p\\} \\setminus \\{k\\}$ are such that the graph with edges given by $P_k$ being $\\text{pa}(k)$ is a DAG.\nExample 4.4.1. Consider the following (totally artificial) SEM which has whether you are taking this course $(Z_{1}=1)$ depending on whether you went to the statistical modelling course $(Z_{2}=1)$ and whether you have heard about machine learning $(Z_{3}=1)$ . Suppose\n$$\n\\begin{array}{l} Z _ {3} = \\varepsilon_ {3} \\sim \\mathrm{Bern} (1 / 4) \\\\ Z _ {2} = \\mathbb {1} _ {\\{\\varepsilon_ {2} (1 + Z _ {3}) > 1 / 2 \\}} \\qquad \\varepsilon_ {2} \\sim U [ 0, 1 ] \\\\ Z _ {1} = \\mathbb {1} _ {\\{\\varepsilon_ {1} (Z _ {2} + Z _ {3}) > 1 / 2 \\}} \\qquad \\varepsilon_ {1} \\sim U [ 0, 1 ]. \\\\ \\end{array}\n$$\nThe corresponding DAG is\n\nExample 4.4.1. Consider the following (totally artificial) SEM which has whether you are taking this course $(Z_{1}=1)$ depending on whether you went to the statistical modelling course $(Z_{2}=1)$ and whether you have heard about machine learning $(Z_{3}=1)$ . Suppose
    0\n[images/image-8-Example 4..jpg]\n\n
    \nflowchart\n```mermaid\ngraph TD\nZ2 --> Z1\nZ2 <--> Z3\nZ3 --> Z1\n```\n
    \nNote that an SEM for Z determines its law. Indeed using a topological ordering $\\pi$ for the associated DAG, we can write each $Z_{k}$ as a function of $\\varepsilon_{\\pi^{-1}(1)}, \\varepsilon_{\\pi^{-1}(2)}, \\ldots, \\varepsilon_{\\pi^{-1}(\\pi(k))}$ . Importantly, though, we can use it to tell us much more than simply the law of Z: for example we can query properties of the distribution of Z after having set a particular component to any given value. This is what we study next.", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.4 Structural equation models", + "metadata": { + "length": 2612, + "summary": "The passage introduces structural equation models (SEMs) as a way to represent data generating processes, going beyond conditional independence graphs to answer interventional questions. It defines an SEM as a collection of equations where each variable is a function of its parents and independent noise, with the graph being a DAG. An artificial example illustrates how variables like course attendance depend on other factors. The SEM determines the joint distribution and allows querying distributions after setting variables to specific values.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Conditional", + "independence", + "graphs", + "give", + "understanding", + "relationships", + "variables", + "However", + "set", + "kth", + "variable", + "0.5", + "distribution", + "values", + "altered", + "Yet", + "sort", + "question", + "answer", + "In", + "order", + "reach", + "ambitious", + "goal", + "introduce", + "notion", + "structural", + "equation", + "models", + "SEMs", + "These", + "representing", + "data", + "generating", + "process", + "We", + "make", + "undirected", + "sorts", + "DAGs", + "graph", + "satisfying", + "definition", + "Definition", + "11", + "model", + "random", + "vector", + "collection", + "equations", + "varepsilon", + "qquad", + "ldots", + "independent", + "subseteq", + "setminus", + "edges", + "text", + "pa", + "DAG", + "Example", + "4.4", + "Consider", + "totally", + "artificial", + "SEM", + "taking", + "depending", + "statistical", + "modelling", + "heard", + "machine", + "learning", + "Suppose", + "begin", + "array", + "sim", + "mathrm", + "Bern", + "mathbb", + "end", + "The", + "details", + "summary", + "flowchart", + "mermaid", + "TD", + "Z2", + "Z1", + "Z3", + "Note", + "determines", + "law", + "Indeed", + "topological", + "ordering", + "pi", + "write", + "function", + "Importantly", + "simply", + "query", + "properties", + "component", + "This", + "study" + ], + "keywords": [ + "SEM", + "DAG", + "causal inference" + ], + "connect_to": [ + { + "target": "32181875-bd81-5d8d-a62e-100d18993cf0", + "relation": "embeds", + "ref": "[images/image-8-Example 4..jpg]", + "position": { + "start": 1995, + "end": 2026 + } + } + ] + } + }, + { + "chunk_id": "35687f81-7e72-5e10-b0e0-17f4359adc48", + "type": "text", + "content": "Given an SEM S, we can replace one (or more) of the structural equations by a new structural equation, for example for a chosen variable k we could replace the structural equation $Z_{k} = h_{k}(Z_{P_{k}}, \\varepsilon_{k})$ by $Z_{k} = \\tilde{h}_{k}(\\tilde{Z}_{\\tilde{P}_{k}}, \\tilde{\\varepsilon}_{k})$ . This gives us a new structural equation model $\\tilde{S}$ which in turn determines a new joint law for Z.\nWhen we have $\\tilde{h}_{k}(\\tilde{Z}_{\\tilde{P}_{k}},\\tilde{\\varepsilon}_{k})=a$ for some $a\\in R$ , so we are setting the value of $Z_{k}$ to be a, we call this a (perfect) intervention. Expectations and probabilities under this new law for Z are written by adding $|do(Z_{k}=a)$ e.g. $\\mathbb{E}(Z_{j}|do(Z_{k}=a))$ . Note that this will in general be different from the conditional expectation $\\mathbb{E}(Z_{j}|Z_{k}=a)$ .\nExample 4.4.1 continued. After the intervention $do(Z_2 = 1)$ (everyone is forced to go to statistical modelling), we have a new SEM $\\tilde{\\mathcal{S}}$ :\n$$\nZ _ {3} = \\varepsilon_ {3} \\sim \\mathrm{Bern} (1 / 4)\n$$\n$$\nZ _ {2} = 1\n$$\n$$\nZ _ {1} = \\mathbb {1} _ {\\{\\varepsilon_ {1} (1 + Z _ {3}) > 1 / 2 \\}} \\quad \\varepsilon_ {1} \\sim U [ 0, 1 ].\n$$\nThus $\\mathbb{P}(Z_1 = 1|do(Z_2 = 1)) = \\frac{1}{4}\\frac{3}{4} +\\frac{3}{4}\\frac{1}{2} = \\frac{9}{16}$ . On the other hand,\n$$\n\\begin{array}{l} \\mathbb {P} (Z _ {1} = 1 | Z _ {2} = 1) = \\sum_ {j \\in \\{0, 1 \\}} \\mathbb {P} (Z _ {1} = 1 | Z _ {2} = 1, Z _ {3} = j) \\mathbb {P} (Z _ {3} = j | Z _ {2} = 1) \\\\ = \\frac {1}{\\mathbb {P} (Z _ {2} = 1)} \\sum_ {j \\in \\{0, 1 \\}} \\mathbb {P} (Z _ {1} = 1 | Z _ {2} = 1, Z _ {3} = j) \\mathbb {P} (Z _ {2} = 1 | Z _ {3} = j) \\mathbb {P} (Z _ {3} = j) \\\\ = \\frac {1}{\\frac {1}{4} \\frac {3}{4} + \\frac {3}{4} \\frac {1}{2}} \\left(\\frac {3}{4} \\frac {3}{4} \\frac {1}{4} + \\frac {1}{2} \\frac {1}{2} \\frac {3}{4}\\right) \\\\ = \\frac {7}{1 2} \\neq \\frac {9}{1 6}. \\\\ \\end{array}\n$$", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.5 Interventions", + "metadata": { + "length": 1899, + "summary": "The text explains how replacing a structural equation in an SEM creates a new model, with a perfect intervention setting a variable to a constant (e.g., do(Z_k=a)). It contrasts interventional expectations (e.g., E(Z_j|do(Z_k=a))) with conditional expectations, using an example where do(Z_2=1) yields P(Z_1=1|do(Z_2=1))=9/16, while P(Z_1=1|Z_2=1)=7/12, showing they differ.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Given", + "SEM", + "replace", + "structural", + "equations", + "equation", + "chosen", + "variable", + "varepsilon", + "tilde", + "This", + "model", + "turn", + "determines", + "joint", + "law", + "When", + "setting", + "call", + "perfect", + "intervention", + "Expectations", + "probabilities", + "written", + "adding", + "mathbb", + "Note", + "general", + "conditional", + "expectation", + "Example", + "4.4", + "continued", + "After", + "forced", + "statistical", + "modelling", + "mathcal", + "sim", + "mathrm", + "Bern", + "quad", + "Thus", + "frac", + "16", + "On", + "hand", + "begin", + "array", + "sum", + "left", + "neq", + "end" + ], + "keywords": [ + "Structural Equation Model", + "Intervention", + "Do-calculus" + ], + "connect_to": [] + } + }, + { + "chunk_id": "2123956b-b229-541b-93d5-2e41d3e83276", + "type": "text", + "content": "The DAG of an SEM can encode a number of conditional independencies present in the law of the random vector Z. To understand this, we first introduce some so-called Markov properties on DAGs.\nLet $P$ be the joint law of $Z$ and suppose it has a density $f$ .\nDefinition 12. Given a DAG G, we say P satisfies the\n(i) Markov factorisation property w.r.t. the DAG $\\mathcal{G}$ if\n$$\nf (z _ {1}, \\ldots , z _ {p}) = \\prod_ {k = 1} ^ {p} f (z _ {k} | z _ {\\mathrm{pa} (k)}).\n$$\n(ii) global Markov property w.r.t. the DAG G if for all disjoint A, B, S ⊆ {1, . . . , p},\n$$\nA, B d \\text {-separated by} S \\Rightarrow Z _ {A} \\perp \\perp Z _ {B} | Z _ {S}.\n$$\nTheorem 31. If P has a density f (with respect to a product measure), then all Markov properties in definition 12 are equivalent.\nIn view of this, we will henceforth use the term Markov to mean global Markov.\nProposition 32. Let P be the law of an SEM with DAG G. Then P obeys the Markov factorisation property w.r.t. G.\nThus we can read off from the DAG of an SEM a great deal of information concerning the distribution it generates. We can use this to help us calculate the effects of interventions.\nWe have seen now how an SEM can be used to not only query properties of the joint distribution, but also to determine the effects of certain perturbations to the system. In many settings, we may not have a prespecified SEM to work with, but instead we would like to learn the DAG from observational data. This is the problem we turn to next.", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.6 The Markov properties on DAGs", + "metadata": { + "length": 1496, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "The", + "DAG", + "SEM", + "encode", + "number", + "conditional", + "independencies", + "present", + "law", + "random", + "vector", + "To", + "understand", + "introduce", + "called", + "Markov", + "properties", + "DAGs", + "Let", + "joint", + "suppose", + "density", + "Definition", + "12", + "Given", + "satisfies", + "factorisation", + "property", + "mathcal", + "ldots", + "prod", + "mathrm", + "pa", + "ii", + "global", + "disjoint", + "text", + "separated", + "Rightarrow", + "perp", + "Theorem", + "31", + "If", + "respect", + "product", + "measure", + "definition", + "equivalent", + "In", + "view", + "henceforth", + "term", + "Proposition", + "32", + "Then", + "obeys", + "Thus", + "read", + "great", + "deal", + "information", + "distribution", + "generates", + "We", + "calculate", + "effects", + "interventions", + "query", + "determine", + "perturbations", + "system", + "settings", + "prespecified", + "work", + "learn", + "observational", + "data", + "This", + "problem", + "turn" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "9e892b83-1fe1-52b9-a8c6-1b5c36effb2f", + "type": "text", + "content": "Given a sample of observations from P, we would like to determine the DAG which generated it. We can think of this task in terms of two subtasks: firstly we need to understand how to extract information concerning P from a sample, which is a traditional statistical question of the sort we are used to; secondly, given P itself, we need to relate this to the DAG which generated it. The latter problem is unique to casual inference and we discuss this first.", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning", + "metadata": { + "length": 458, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Given", + "sample", + "observations", + "determine", + "DAG", + "generated", + "We", + "task", + "terms", + "subtasks", + "firstly", + "understand", + "extract", + "information", + "traditional", + "statistical", + "question", + "sort", + "relate", + "The", + "problem", + "unique", + "casual", + "inference", + "discuss" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1b6c3135-8d31-5e21-a7b7-5ff8c8e6a56a", + "type": "text", + "content": "There are three obstacles to causal structure learning. The first two are more immediate but the last is somewhat subtle.", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.1 Three obstacles", + "metadata": { + "length": 121, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "There", + "obstacles", + "causal", + "structure", + "learning", + "The", + "subtle" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f2bab9d7-1bdb-5091-a58a-2ef886e596e9", + "type": "text", + "content": "We know that if P is generated by an SEM with DAG G, then P will be Markov w.r.t. G. Conversely, one can show that if P is Markov w.r.t. a DAG G, then there is also an SEM with DAG G that could have generated P. But P will be Markov w.r.t. a great number of DAGs, e.g. $Z_{1}$ and $Z_{2}$ being independent can be represented by\n$$\nZ _ {1} = 0 \\times Z _ {2} + \\varepsilon_ {1} = \\varepsilon_ {1}, \\qquad Z _ {2} = \\varepsilon_ {2}.\n$$\nThis motivates the following definition.\nDefinition 13. P satisfies causal minimality with respect to G if it is (global) Markov w.r.t. G but not to a proper subgraph of G with the same nodes.", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.1 Three obstacles/Causal minimality", + "metadata": { + "length": 628, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "We", + "generated", + "SEM", + "DAG", + "Markov", + "Conversely", + "show", + "But", + "great", + "number", + "DAGs", + "independent", + "represented", + "times", + "varepsilon", + "qquad", + "This", + "motivates", + "definition", + "Definition", + "13", + "satisfies", + "causal", + "minimality", + "respect", + "global", + "proper", + "subgraph", + "nodes" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "58583ac8-96dc-5c52-ab81-896969cb127a", + "type": "image", + "content": "\n$$ For a DAG $\\mathcal{G}$ , let0\n[images/image-9-$$ For a D.jpg]\n", + "path": "images/image-9-$$ For a D.jpg", + "metadata": { + "length": 67, + "summary": "image-9\n$$ For a DAG $\\mathcal{G}$ , let0", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "file_path": "images/image-9-$$ For a D.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "c2d82480-df61-51e4-9d87-886d11b1c1e1", + "type": "image", + "content": "\nThis motivates the following definition. For a DAG $\\mathcal{G}$ , let0\n[images/image-10-This motiv.jpg]\n", + "path": "images/image-10-This motiv.jpg", + "metadata": { + "length": 106, + "summary": "image-10\nThis motivates the following definition. For a DAG $\\mathcal{G}$ , let0", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "file_path": "images/image-10-This motiv.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "780ecf56-216e-56c6-9c64-8def1971d000", + "type": "text", + "content": "It is possible for two different DAGs to satisfy the same collection of d-separations e.g.\n\n$$ For a DAG $\\mathcal{G}$ , let0\n[images/image-9-$$ For a D.jpg]\n\n\nThis motivates the following definition. For a DAG $\\mathcal{G}$ , let0\n[images/image-10-This motiv.jpg]\n\nFor a DAG $\\mathcal{G}$ , let\n$\\mathcal{M}(\\mathcal{G}) = \\{\\text{distributions } P : P \\text{ satisfies the global Markov property w.r.t. } \\mathcal{G}\\}$ .\nDefinition 14. We say two DAGs $\\mathcal{G}_1$ and $\\mathcal{G}_2$ are Markov equivalent if $\\mathcal{M}(\\mathcal{G}_1) = \\mathcal{M}(\\mathcal{G}_2)$ .\nProposition 33. Two DAGs are Markov equivalent if and only if they have the same skeleton and v-structures.\nThe set of all DAGs that are Markov equivalent to a DAG can be represented by a completed PDAG (CPDAG) which contains an edge $(j,k)$ if and only if one member of the Markov equivalence class does. We can only ever hope to obtain the Markov equivalence class i.e. the CPDAG of a DAG with which P satisfies causal minimality (unless we place restrictions on the functional forms of the SEM equations).", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.1 Three obstacles/Markov equivalent DAGs", + "metadata": { + "length": 1084, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "It", + "DAGs", + "satisfy", + "collection", + "separations", + "For", + "DAG", + "mathcal", + "let0", + "This", + "motivates", + "definition", + "text", + "distributions", + "satisfies", + "global", + "Markov", + "property", + "Definition", + "14", + "We", + "equivalent", + "Proposition", + "33", + "Two", + "skeleton", + "structures", + "The", + "set", + "represented", + "completed", + "PDAG", + "CPDAG", + "edge", + "member", + "equivalence", + "class", + "hope", + "obtain", + "causal", + "minimality", + "place", + "restrictions", + "functional", + "forms", + "SEM", + "equations" + ], + "keywords": [], + "connect_to": [ + { + "target": "58583ac8-96dc-5c52-ab81-896969cb127a", + "relation": "embeds", + "ref": "[images/image-9-$$ For a D.jpg]", + "position": { + "start": 126, + "end": 157 + } + }, + { + "target": "c2d82480-df61-51e4-9d87-886d11b1c1e1", + "relation": "embeds", + "ref": "[images/image-10-This motiv.jpg]", + "position": { + "start": 232, + "end": 264 + } + } + ] + } + }, + { + "chunk_id": "a31b88e5-f467-5e79-b1ab-9a8c9e7fecf2", + "type": "text", + "content": "Consider the following SEM.\n$$\nZ _ {1} = \\varepsilon_ {1}\n$$\n$$\nZ _ {2} = \\alpha Z _ {1} + \\varepsilon_ {2}\n$$\n$$\nZ _ {3} = \\beta Z _ {1} + \\gamma Z _ {2} + \\varepsilon_ {3},\n$$\nwhere $\\varepsilon \\sim N_3(0,I)$ . Then $(Z_{1},Z_{2},Z_{3})\\sim N_{3}(0,\\Sigma) = P^{0}$ with\n$$\n\\Sigma = \\left( \\begin{array}{c c c} 1 & \\alpha & \\beta + \\alpha \\gamma \\\\ \\alpha & \\alpha^ {2} + 1 & \\alpha \\beta + \\gamma (\\alpha^ {2} + 1) \\\\ \\beta + \\alpha \\gamma & \\alpha \\beta + \\gamma (\\alpha^ {2} + 1) & \\beta^ {2} + \\gamma^ {2} (\\alpha^ {2} + 1) + 2 \\beta \\gamma \\alpha + 1 \\end{array} \\right).\n$$\nIf $\\beta + \\alpha \\gamma = 0$ e.g. if $\\beta = -1$ , $\\alpha, \\gamma = 1$ , then $Z_1 \\perp Z_3$ . We claim that in this case $P^0$ can also be generated by the SEM\n$$\n\\tilde {Z} _ {1} = \\tilde {\\varepsilon} _ {1}\n$$\n$$\n\\tilde {Z} _ {2} = \\tilde {Z} _ {1} + \\tilde {\\alpha} \\tilde {Z} _ {3} + \\tilde {\\varepsilon} _ {2}\n$$\n$$\n\\tilde {Z} _ {3} = \\tilde {\\varepsilon} _ {3}.\n$$\nHere the $\\tilde{\\varepsilon}_{j}$ are independent with $\\tilde{\\varepsilon}_{1}\\sim N(0,1)$ , $\\tilde{\\varepsilon}_{3}\\sim N(0,2)$ , $\\tilde{\\alpha}=1/2$ and $\\tilde{\\varepsilon}_{3}\\sim N(0,1/2)$ . Writing the DAGs for the two SEMs above as G and $\\tilde{G}$ , note that $P^{0}$ satisfies causal minimality w.r.t. both G and $\\tilde{G}$ .\nDefinition 15. We say $P$ is faithful to the DAG $\\mathcal{G}$ if it is Markov w.r.t. $\\mathcal{G}$ and for all disjoint $A, B, S \\subseteq \\{1, \\ldots, p\\}$ ,\n$$\nA, B d \\text {-separated by} S \\Leftarrow Z _ {A} \\perp \\perp Z _ {B} | Z _ {S}.\n$$\nFaithfulness demands that all conditional independencies in P are represented in the DAG. In our example $P^{0}$ is not faithful to G, but it is faithful to $\\tilde{G}$ .", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.1 Three obstacles/Faithfulness", + "metadata": { + "length": 1718, + "summary": "The text discusses two structural equation models (SEMs) generating the same multivariate normal distribution P^0. The first SEM has variables Z1, Z2, Z3 with parameters α, β, γ, and under the condition β+αγ=0 (e.g., β=-1, α=γ=1), Z1 is independent of Z3. A second SEM with different structure and error distributions also generates P^0. The concept of faithfulness is introduced: a distribution is faithful to a DAG if all conditional independencies are represented in the DAG. P^0 is not faithful to the first DAG (G) but is faithful to the second (tilde G).", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Consider", + "SEM", + "varepsilon", + "alpha", + "beta", + "gamma", + "sim", + "Then", + "Sigma", + "left", + "begin", + "array", + "end", + "If", + "perp", + "We", + "claim", + "case", + "generated", + "tilde", + "Here", + "independent", + "Writing", + "DAGs", + "SEMs", + "note", + "satisfies", + "causal", + "minimality", + "Definition", + "15", + "faithful", + "DAG", + "mathcal", + "Markov", + "disjoint", + "subseteq", + "ldots", + "text", + "separated", + "Leftarrow", + "Faithfulness", + "demands", + "conditional", + "independencies", + "represented", + "In" + ], + "keywords": [ + "SEM", + "faithfulness", + "DAG" + ], + "connect_to": [] + } + }, + { + "chunk_id": "dc289d00-b46c-5c6c-b810-10b13e28e216", + "type": "text", + "content": "Proposition 34. If nodes j and k in a DAG G are adjacent, then no set can d-separate them. If they are not adjacent and $\\pi$ is a topological order with $\\pi(j) < \\pi(k)$ , then they are d-separated by $\\text{pa}(k)$ .\nProof. Consider a path $j = j_{1}, \\ldots, j_{m} = k$ . We may assume we don't have $j_{m-1} \\to k$ as otherwise the path would be blocked since $j_{m-1} \\in \\text{pa}(k)$ . Let l be the largest $l'$ with $j_{l'-1} \\to j_{l'} \\leftarrow j_{l'+1}$ ; this must exist as otherwise we would have a directed path from k to j contradicting the topological ordering. In order for the path to be active, $j_{l'}$ must have a descendant in $\\text{pa}(k)$ , but this would introduce a cycle. □\nThis shows in particular that any non-adjacent nodes must have a d-separating set. If we assume that P is faithful w.r.t. a DAG G, we can check whether nodes j and k are adjacent in G by testing whether there is a set S with $Z_{j} \\perp \\perp Z_{k}|Z_{S}$ . If there is no such set S, j and k must be adjacent. This allows us to recover the skeleton of G.\nProposition 35. Suppose we have a triple of nodes j, k, l in a DAG and the only non-adjacent pair is j, k (i.e. in the skeleton j - l - k).\n(i) If the nodes are in a v-structure $(j \\rightarrow l \\leftarrow k)$ then no S that d-separates j and k can contain l.\n(ii) If there exists an $S$ that $d$ -separates $j$ and $k$ and $l \\notin S$ , then we must have $j \\to l \\leftarrow k$ .\nProof. For (i) note that any set containing l cannot block the path j, l, k. For (ii) note we know that the path j, l, k is blocked by S, so we must have $j \\rightarrow l \\leftarrow k$ . □\nThis last result then allows us to find the v-structures given the skeleton and a d-separating set $S(j, k)$ corresponding to each absent edge. Given a skeleton and v-structures, it may be possible to orient further edges by making use of the acyclicity of DAGs; we do not cover this here.", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.2 The PC algorithm", + "metadata": { + "length": 1922, + "summary": "The text presents two propositions about d-separation in directed acyclic graphs (DAGs). Proposition 34 states that adjacent nodes in a DAG cannot be d-separated by any set, while non-adjacent nodes can be d-separated by the parent set of the later node in a topological order. This allows recovery of the DAG's skeleton by testing conditional independence. Proposition 35 addresses v-structures: in a triple where only j and k are non-adjacent, if they are in a v-structure (j→l←k), no d-separating set for j and k can contain l; conversely, if a d-separating set exists without l, the structure must be a v-structure. These results help identify v-structures from the skeleton and d-separating sets.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Proposition", + "34", + "If", + "nodes", + "DAG", + "adjacent", + "set", + "separate", + "pi", + "topological", + "order", + "separated", + "text", + "pa", + "Proof", + "Consider", + "path", + "ldots", + "We", + "assume", + "don", + "blocked", + "Let", + "largest", + "leftarrow", + "exist", + "directed", + "contradicting", + "ordering", + "In", + "active", + "descendant", + "introduce", + "cycle", + "This", + "shows", + "separating", + "faithful", + "check", + "testing", + "perp", + "recover", + "skeleton", + "35", + "Suppose", + "triple", + "pair", + "structure", + "rightarrow", + "separates", + "ii", + "exists", + "notin", + "For", + "note", + "block", + "result", + "find", + "structures", + "absent", + "edge", + "Given", + "orient", + "edges", + "making", + "acyclicity", + "DAGs", + "cover" + ], + "keywords": [ + "d-separation", + "DAG", + "skeleton" + ], + "connect_to": [] + } + }, + { + "chunk_id": "2d538e6d-e3ac-51f6-93ef-2d50623a17ff", + "type": "text", + "content": "The PC-algorithm, named after its inventors Peter Spirtes and Clarke Glymour [Spirtes et al., 2000], exploits the fact that we need not search over all sets S but only subsets of either $\\mathrm{pa}(j)$ or $\\mathrm{pa}(k)$ for efficiency. The version assumes P is known and so conditional independencies can be queried directly. A sample version that is applicable in practice is given in the following subsection. We denote the set of nodes that are adjacent to a node j in graph G by $\\mathrm{adj}(\\mathcal{G}, j)$ .\nAlgorithm 1 First part of the PC algorithm: finding the skeleton.\nSet $\\hat{\\mathcal{G}}$ to be the complete undirected graph. Set $\\ell = -1$ .\nrepeat\nIncrement $\\ell \\to \\ell + 1$ .\nrepeat\nSelect a (new) ordered pair of nodes $j, k$ that are adjacent in $\\hat{\\mathcal{G}}$ and such that $|\\operatorname{adj}(\\hat{\\mathcal{G}}, j) \\setminus \\{k\\}| \\geq \\ell$ .\nrepeat\nChoose new $S \\subseteq \\operatorname{adj}(\\hat{\\mathcal{G}}, j) \\setminus \\{k\\}$ with $|S| = \\ell$ .\nIf $Z_j \\perp Z_k |Z_S$ then delete edges $(j, k)$ and $(k, j)$ and set $S(j, k) = S(k, j) = S$ .\nuntil edges $(j, k)$ , $(k, j)$ are deleted or all relevant subsets have been chosen.\nuntil all relevant ordered pairs have been chosen.\nuntil for every ordered pair $j, k$ that are adjacent in $\\hat{\\mathcal{G}}$ we have $|\\operatorname{adj}(\\hat{\\mathcal{G}}, j) \\setminus \\{k\\}| < \\ell$ .\nSuppose P is faithful to DAG $G^{0}$ . At each stage of the Algorithm 1 we must have that the skeleton is a subgraph of $\\hat{G}$ . By the end of the algorithm, for each pair j, k adjacent in $\\hat{G}$ , we would have searched through $\\operatorname{adj}(\\hat{\\mathcal{G}}, j)$ and $\\operatorname{adj}(\\hat{\\mathcal{G}}, k)$ for sets S such that $Z_{j} \\perp Z_{k}|Z_{S}$ . If P were faithful to $G^{0}$ then, we would know that j and k must be adjacent in $G^{0}$ . That is the output of Algorithm 1 would be the skeleton of $G^{0}$ .\nAlgorithm 2 Second part of the PC algorithm: finding the v-structures\nfor all pairs of non-adjacent variables $j, k$ (in skeleton $\\hat{\\mathcal{G}}$ ) with common neighbour $l$ do\nIf $l \\notin S(j, k)$ then orient $j \\to l \\leftarrow k$ .\nend for", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.2 The PC algorithm/Population version", + "metadata": { + "length": 2164, + "summary": "The text describes the PC-algorithm for causal structure learning, which efficiently finds the skeleton of a directed acyclic graph (DAG) by testing conditional independencies. Algorithm 1 iteratively removes edges between nodes if they are conditionally independent given some subset of neighbors, assuming faithfulness to the true DAG. Algorithm 2 then orients v-structures for non-adjacent nodes sharing a common neighbor, using stored separating sets.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "The", + "PC", + "algorithm", + "named", + "inventors", + "Peter", + "Spirtes", + "Clarke", + "Glymour", + "al", + "2000", + "exploits", + "fact", + "search", + "sets", + "subsets", + "mathrm", + "pa", + "efficiency", + "version", + "assumes", + "conditional", + "independencies", + "queried", + "directly", + "sample", + "applicable", + "practice", + "subsection", + "We", + "denote", + "set", + "nodes", + "adjacent", + "node", + "graph", + "adj", + "mathcal", + "Algorithm", + "First", + "part", + "finding", + "skeleton", + "Set", + "hat", + "complete", + "undirected", + "ell", + "repeat", + "Increment", + "Select", + "ordered", + "pair", + "operatorname", + "setminus", + "geq", + "Choose", + "subseteq", + "If", + "perp", + "delete", + "edges", + "deleted", + "relevant", + "chosen", + "pairs", + "Suppose", + "faithful", + "DAG", + "At", + "stage", + "subgraph", + "By", + "end", + "searched", + "That", + "output", + "Second", + "structures", + "variables", + "common", + "neighbour", + "notin", + "orient", + "leftarrow" + ], + "keywords": [ + "PC-algorithm", + "skeleton", + "v-structures" + ], + "connect_to": [] + } + }, + { + "chunk_id": "c54a6bc9-3f1b-5b3e-914c-6d88fb257274", + "type": "text", + "content": "The sample version of the PC algorithm replaces the querying of conditional independence with a conditional independence test applied to data $x_{1},\\ldots,x_{n}$ . The level of the test $\\alpha$ will be a tuning parameter of the method. If the data are assumed to be multivariate normal, the (sample) partial correlation can be used to test conditional independence since if $Z_{j} \\perp \\perp Z_{k}|Z_{S}$ then\n$$\n\\mathrm{Corr} (Z _ {j}, Z _ {k} | Z _ {S}) := \\rho_ {j k \\cdot S} = 0.\n$$\nTo compute the sample partial correlation, we regress $X_{j}$ and $X_{k}$ on $X_{S}$ and compute the correlation between the resulting residuals.", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.2 The PC algorithm/Sample version", + "metadata": { + "length": 635, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "The", + "sample", + "version", + "PC", + "algorithm", + "replaces", + "querying", + "conditional", + "independence", + "test", + "applied", + "data", + "ldots", + "level", + "alpha", + "tuning", + "parameter", + "method", + "If", + "assumed", + "multivariate", + "normal", + "partial", + "correlation", + "perp", + "mathrm", + "Corr", + "rho", + "cdot", + "To", + "compute", + "regress", + "resulting", + "residuals" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "b5f3d362-0c22-592e-a395-ac451df1c1bd", + "type": "image", + "content": "\n$$ 5.1 Using the debiased Lasso in practice0\n[images/image-11-$$ 5.1 Usi.jpg]\n", + "path": "images/image-11-$$ 5.1 Usi.jpg", + "metadata": { + "length": 79, + "summary": "image-11\n$$ 5.1 Using the debiased Lasso in practice0", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "file_path": "images/image-11-$$ 5.1 Usi.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "8722a030-69a1-58b0-9aea-62048ffe5b2a", + "type": "text", + "content": "High-dimensional inference\nConsider the normal linear model $Y = X\\beta^{0} + \\varepsilon$ where $\\varepsilon \\sim N_{n}(0, \\sigma^{2}I)$ . In the low-dimensional setting, the fact that $\\hat{\\beta}^{\\mathrm{OLS}} - \\beta^{0} \\sim N_{p}(0, \\sigma^{2}(X^{T}X)^{-1})$ allows us to form confidence intervals for components of $\\beta_{j}^{0}$ and perform hypothesis tests with $H_{0}: \\beta_{j}^{0} = 0$ , for example.\nOne might hope that studying the distribution of $\\hat{\\beta}_{\\lambda}^{L}-\\beta^{0}$ would enable us to perform these tasks in the high-dimensional setting when $p\\gg n$ . However, the distribution of $\\hat{\\beta}_{\\lambda}^{L}-\\beta^{0}$ is intractable and depends delicately on the unknown $\\beta^{0}$ , making it unsuitable as a basis for establishing confidence intervals.\nWhilst several methods have been proposed over the years, typically they have involved placing conditions on the unknown $\\beta^{0}$ , other than the usual assumption of sparsity. Given that the task is to perform inference for $\\beta^{0}$ , such conditions are undesirable. In the last couple of years, there has been a breakthrough on this front [Zhang and Zhang, 2014, Van de Geer et al., 2014], and here we will aim to cover the main ideas in this exciting development. Our treatment follows Van de Geer et al. [2014].\nWe begin our investigation by considering the KKT conditions of the Lasso. Fix $\\lambda > 0$ and let $\\hat{\\beta}$ be the Lasso estimator with tuning parameter $\\lambda$ . Recall that the KKT conditions give\n$$\n\\frac {1}{n} X ^ {T} (Y - X \\hat {\\beta}) = \\lambda \\hat {\\nu}\n$$\nwhere $\\|\\hat{\\nu}\\|_{\\infty} \\leq 1$ and writing $\\hat{S} = \\{k : \\hat{\\beta}_{k} \\neq 0\\}$ , $\\hat{\\nu}_{\\hat{S}} = \\operatorname{sgn}(\\hat{\\beta}_{\\hat{S}})$ . Setting $\\hat{\\Sigma} = X^{T}X/n$ and rearranging we have\n$$\n\\hat {\\Sigma} (\\hat {\\beta} - \\beta^ {0}) + \\lambda \\hat {\\nu} = \\frac {1}{n} X ^ {T} \\varepsilon .\n$$\nThe key idea is now to form an approximate inverse $\\hat{\\Theta}$ of $\\hat{\\Sigma}$ . Then we have\n$$\n\\hat {\\beta} + \\lambda \\hat {\\Theta} \\hat {\\nu} - \\beta^ {0} = \\frac {1}{n} \\hat {\\Theta} X ^ {T} \\varepsilon + \\frac {1}{\\sqrt {n}} \\Delta\n$$\nwhere $\\Delta = \\sqrt{n} (\\hat{\\Theta}\\hat{\\Sigma} -I)(\\beta^0 -\\hat{\\beta})$ . Define\n$$\n\\hat {b} = \\hat {\\beta} + \\lambda \\hat {\\Theta} \\hat {\\nu} = \\hat {\\beta} + \\hat {\\Theta} X ^ {T} (Y - X \\hat {\\beta}) / n,\n$$\nwhich we shall refer to as the debiased Lasso. If we choose $\\hat{\\Theta}$ such that $\\Delta$ is small, we will have $\\hat{b}-\\beta^{0}\\approx\\hat{\\Theta}X^{T}\\varepsilon/n$ , which can be used as a basis for performing inference.\nWe already know that under a compatibility condition on the design matrix $X$ , $\\| \\hat{\\beta} - \\beta^0 \\|_1$ is small (Theorem 21) with high probability. If we can also show that the $\\ell_{\\infty}$ -norms of rows of $\\hat{\\Theta} \\hat{\\Sigma} - I$ are small, we can leverage this fact using Hölder's inequality to show that $\\| \\Delta \\|_{\\infty}$ is small. Let $\\hat{\\theta}_j$ be the $j$ th row of $\\hat{\\Theta}$ . Then $\\| (\\hat{\\Sigma} \\hat{\\Theta}^T - I)_j \\|_{\\infty} \\leq \\eta$ is equivalent to\n$$\n\\frac {1}{n} \\| X _ {- j} ^ {T} X \\hat {\\theta} _ {j} \\| _ {\\infty} \\leq \\eta \\mathrm{and} | X _ {j} ^ {T} X \\hat {\\theta} _ {j} / n - 1 | \\leq \\eta .\n$$\nThe first of these inequalities is somewhat reminiscent of the KKT conditions for the Lasso. Let\n$$\n\\hat {\\gamma} ^ {(j)} = \\underset {\\gamma \\in \\mathbb {R} ^ {p - 1}} {\\arg \\min} \\left\\{\\frac {1}{2 n} \\| X _ {j} - X _ {- j} \\gamma \\| _ {2} ^ {2} + \\lambda_ {j} \\| \\gamma \\| _ {1} \\right\\}. \\tag {5.0.1}\n$$\nFurther let\n$$\n\\hat {\\tau} _ {j} ^ {2} = X _ {j} ^ {T} (X _ {j} - X _ {- j} \\hat {\\gamma} ^ {(j)}) / n = \\frac {1}{n} \\| X _ {j} - X _ {- j} \\hat {\\gamma} ^ {(j)} \\| _ {2} ^ {2} + \\lambda_ {j} \\| \\hat {\\gamma} ^ {(j)} \\| _ {1};\n$$\nsee the example sheet for the final equality. Then set\n$$\n\\begin{array}{l} \\hat {\\theta} _ {j} = - \\frac {1}{\\hat {\\tau} _ {j} ^ {2}} (\\hat {\\gamma} _ {1} ^ {(j)}, \\dots , \\hat {\\gamma} _ {j - 1} ^ {(j)}, - 1, \\hat {\\gamma} _ {j} ^ {(j)}, \\dots , \\hat {\\gamma} _ {p - 1} ^ {(j)}) ^ {T}. \\\\ \\begin{array}{c} \\uparrow \\\\ j \\text {th position} \\end{array} \\\\ \\end{array}\n$$\nNote that by construction,\n$$\nX \\hat {\\theta} _ {j} = \\frac {X _ {j} - X _ {- j} \\hat {\\gamma} ^ {(j)}}{X _ {j} ^ {T} (X - X _ {- j} \\hat {\\gamma} ^ {(j)}) / n}.\n$$\nThus $X_{j}^{T}X\\hat{\\theta}_{j} / n = 1$ and by the KKT conditions of the Lasso optimisation (5.0.1), we have $\\hat{\\tau}_j^2\\| X_{-j}^T X\\hat{\\theta}_j\\|_\\infty /n\\leq \\lambda_j$ .\nThus with the choice of $\\hat{\\Theta}$ defined as above, we have\n$$\n\\| \\Delta \\| _ {\\infty} \\leq \\sqrt {n} \\| \\hat {\\beta} - \\beta^ {0} \\| _ {1} \\max _ {j} \\frac {\\lambda_ {j}}{\\hat {\\tau} _ {j} ^ {2}}\n$$\nWhen can we expect $\\lambda_{j}/\\hat{\\tau}_{j}^{2}$ to be small? One way of answering this is to consider a random design setting. Let us assume that each row of X is independent and distributed as $N_{p}(0,\\Sigma)$ where $\\Sigma$ is positive definite. Write $\\Omega = \\Sigma^{-1}$ . From Proposition 29 and our study of the neighbourhood selection procedure (see also Section 4.3.3), we know that for each j, we can write\n$$\nX _ {j} = X _ {- j} \\gamma^ {(j)} + \\varepsilon^ {(j)}, \\tag {5.0.2}\n$$\nwhere $\\varepsilon_i^{(j)}|X_{-j}\\stackrel {\\mathrm{i.i.d.}}{\\sim}N(0,\\Omega_{jj}^{-1})$ and $\\gamma^{(j)} = -\\Omega_{jj}^{-1}\\Omega_{-j,j}$ . Theorem 21 can therefore be used to understand properties of $\\hat{\\gamma}^{(j)}$ and hence the $\\hat{\\tau}_j^2$ . In order to apply this result however, we need $\\gamma^{(j)}$ to be sparse. Let us therefore define\n$$\ns _ {j} = \\sum_ {k \\neq j} \\mathbb {1} _ {\\{\\Omega_ {k j} \\neq 0 \\}}\n$$\nand $s_{\\max} = \\max(\\max_j s_j, s)$ . In order to make the following result more easily interpretable, we will consider an asymptotic regime where X, s, $s_{\\max}$ etc. are all allowed to change as $n \\to \\infty$ , though we suppress this in the notation. We will consider $\\sigma$ as constant.\nTheorem 36. Suppose the minimum eigenvalue of $\\Sigma$ is always at least $c_{\\mathrm{min}} > 0$ and $\\max_j \\Sigma_{jj} \\leq 1$ . Suppose further that $s_{\\mathrm{max}} \\sqrt{\\log(p)/n} \\to 0$ . Then there exists constants $A_1, A_2$ such that setting $\\lambda = \\lambda_j = A_1 \\sqrt{\\log(p)/n}$ , we have\n$$\n\\sqrt {n} (\\hat {b} - \\beta^ {0}) = W + \\Delta\n$$\n$$\nW | X \\sim N _ {p} (0, \\sigma^ {2} \\hat {\\Theta} \\hat {\\Sigma} \\hat {\\Theta} ^ {T}),\n$$\nand as $n,p\\to \\infty$\n$$\n\\mathbb {P} (\\| \\Delta \\| _ {\\infty} > A _ {2} s \\log (p) / \\sqrt {n}) \\to 0.\n$$\nProof. Consider the sequence of events $\\Lambda_{n}$ described by the following properties:\n• $\\phi_{\\hat{\\Sigma},s}^{2}\\geq c_{\\min}/2$ and $\\phi_{\\hat{\\Sigma}_{-j,-j},s_{j}}^{2}\\geq c_{\\min}/2$ for all j,\n- $2\\| X^T\\varepsilon \\|_{\\infty} / n \\leq \\lambda$ and $2\\| X_{-j}^{T}\\varepsilon^{(j)}\\|_{\\infty} / n \\leq \\lambda$ for all $j$ ,\n- $\\| \\varepsilon^{(j)}\\| _2^2 /n\\geq \\Omega_{jj}^{-1}(1 - 4\\sqrt{\\log(p) / n})$ for all $j$ .\nYou will show on the example sheet that then $\\mathbb{P}(\\Lambda_{n})\\to1$ for $A_{1}$ sufficiently large. In the following we work on $\\Lambda_{n}$ , and $c_{1},c_{2},\\ldots$ will be constants.\nBy Theorem 21, we have\n$$\n\\| \\hat {\\beta} - \\beta^ {0} \\| _ {1} \\leq c _ {1} s \\sqrt {\\log (p) / n}.\n$$\nWe now seek a lower bound for the $\\hat{\\tau}_j^2$ . Consider the linear models in (5.0.2). Note that the maximum eigenvalue of $\\Omega$ is at most $c_{\\mathrm{min}}^{-1}$ so $\\Omega_{jj} \\leq c_{\\mathrm{min}}^{-1}$ . Also, $\\Omega_{jj}^{-1} = \\mathrm{Var}(X_{ij}|X_{i,-j}) \\leq \\mathrm{Var}(X_{ij}) = \\Sigma_{jj} \\leq 1$ . Thus applying Theorem 21 to the linear models (5.0.2), we know that\n$$\n\\| \\gamma^ {(j)} - \\hat {\\gamma} ^ {(j)} \\| _ {1} \\leq c _ {2} s _ {j} \\sqrt {\\log (p) / n}.\n$$\nThen\n$$\n\\begin{array}{l} \\hat {\\tau} _ {j} ^ {2} \\geq \\frac {1}{n} \\| X _ {j} - X _ {- j} \\hat {\\gamma} ^ {(j)} \\| _ {2} ^ {2} \\geq \\frac {1}{n} \\| \\varepsilon^ {(j)} \\| _ {2} ^ {2} - \\frac {2}{n} \\| X _ {- j} ^ {T} \\varepsilon^ {(j)} \\| _ {\\infty} \\| \\gamma^ {(j)} - \\hat {\\gamma} ^ {(j)} \\| _ {1} \\\\ \\geq \\Omega_ {j j} ^ {- 1} (1 - 4 \\sqrt {\\log (p) / n}) - \\frac {c _ {4} s _ {\\max} \\log (p)}{n} \\\\ \\geq c _ {\\mathrm{min}} / 2 \\\\ \\end{array}\n$$\nfor all j when n is sufficiently large. Putting things together we see that on $\\Lambda_{n}$ ,\n$$\n\\begin{array}{l} \\| \\Delta \\| _ {\\infty} \\leq \\lambda \\sqrt {n} \\| \\hat {\\beta} - \\beta^ {0} \\| _ {1} \\max _ {j} \\hat {\\tau} _ {j} ^ {- 2} \\\\ \\leq 2 A _ {1} \\sqrt {\\log (p)} (c _ {1} s \\sqrt {\\log (p) / n}) / c _ {\\min} \\leq A _ {2} s \\log (p) / \\sqrt {n} \\\\ \\end{array}\n$$\nwhere $A_{2} = 2c_{1}A_{1} / c_{\\mathrm{min}}$ . Thus\n$$\n\\mathbb {P} (\\| \\Delta \\| _ {\\infty} > A _ {2} s \\log (p) / \\sqrt {n}) \\leq \\mathbb {P} (\\Lambda_ {n} ^ {c}) \\to 0.\n$$\n\n$$ 5.1 Using the debiased Lasso in practice0\n[images/image-11-$$ 5.1 Usi.jpg]\n\n5.1 Using the debiased Lasso in practice\nTheorem 36 shows in particular that\n$$\n\\sqrt {n} (\\hat {b} _ {j} - \\beta_ {j} ^ {0}) \\approx W _ {j}\n$$\nwhere $W_{j} \\sim N(0, \\sigma^{2}(\\hat{\\Theta}\\hat{\\Sigma}\\hat{\\Theta}^{T})_{jj})$ . Let $(\\hat{\\Theta}\\hat{\\Sigma}\\hat{\\Theta}^{T})_{jj} = d_{j}$ . The approximate equality above suggests constructing $(1-\\alpha)$ -level confidence intervals of the form\n$$\n\\left[ \\hat {b} _ {j} - z _ {\\alpha / 2} \\sigma \\sqrt {d _ {j}} / \\sqrt {n}, \\hat {b} _ {j} + z _ {\\alpha / 2} \\sigma \\sqrt {d _ {j}} / \\sqrt {n} \\right],\n$$\nwhere $z_{\\alpha}$ is the upper $\\alpha$ point of a standard normal. The only unknown quantity in the confidence interval above is $\\sigma$ : this can be estimated using e.g. the square-root / scaled Lasso [Sun and Zhang, 2012].", + "path": "notes.pdf/Chapter 5 High-dimensional inference", + "metadata": { + "length": 9615, + "summary": "The text discusses high-dimensional inference using the debiased Lasso. It explains how to form confidence intervals for regression coefficients when p >> n by constructing an approximate inverse of the sample covariance matrix. The key idea is to debias the Lasso estimator so that its distribution is approximately normal, enabling inference. The method relies on sparsity conditions and uses node-wise Lasso regressions to estimate the inverse covariance matrix. Theorem 36 provides asymptotic normality results, showing that the debiased estimator's error term is small under certain conditions. Practical construction of confidence intervals uses the estimated variance from the debiased Lasso.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "High", + "dimensional", + "inference", + "Consider", + "normal", + "linear", + "model", + "beta", + "varepsilon", + "sim", + "sigma", + "In", + "low", + "setting", + "fact", + "hat", + "mathrm", + "OLS", + "form", + "confidence", + "intervals", + "components", + "perform", + "hypothesis", + "tests", + "One", + "hope", + "studying", + "distribution", + "lambda", + "enable", + "tasks", + "high", + "gg", + "However", + "intractable", + "depends", + "delicately", + "unknown", + "making", + "unsuitable", + "basis", + "establishing", + "Whilst", + "methods", + "proposed", + "years", + "typically", + "involved", + "placing", + "conditions", + "usual", + "assumption", + "sparsity", + "Given", + "task", + "undesirable", + "couple", + "breakthrough", + "front", + "Zhang", + "2014", + "Van", + "de", + "Geer", + "al", + "aim", + "cover", + "main", + "ideas", + "exciting", + "development", + "Our", + "treatment", + "We", + "begin", + "investigation", + "KKT", + "Lasso", + "Fix", + "estimator", + "tuning", + "parameter", + "Recall", + "give", + "frac", + "nu", + "infty", + "leq", + "writing", + "neq", + "operatorname", + "sgn", + "Setting", + "Sigma", + "rearranging", + "The", + "key", + "idea", + "approximate", + "inverse", + "Theta", + "Then", + "sqrt", + "Delta", + "Define", + "refer", + "debiased", + "If", + "choose", + "small", + "approx", + "performing", + "compatibility", + "condition", + "design", + "matrix", + "Theorem", + "21", + "probability", + "show", + "ell", + "norms", + "rows", + "leverage", + "lder", + "inequality", + "Let", + "theta", + "row", + "eta", + "equivalent", + "inequalities", + "reminiscent", + "gamma", + "underset", + "mathbb", + "arg", + "min", + "left", + "tag", + "5.0", + "Further", + "tau", + "sheet", + "final", + "equality", + "set", + "array", + "dots", + "uparrow", + "text", + "position", + "end", + "Note", + "construction", + "Thus", + "optimisation", + "choice", + "defined", + "max", + "When", + "expect", + "answering", + "random", + "assume", + "independent", + "distributed", + "positive", + "definite", + "Write", + "Omega", + "From", + "Proposition", + "29", + "study", + "neighbourhood", + "selection", + "procedure", + "Section", + "4.3", + "write", + "stackrel", + "jj", + "understand", + "properties", + "order", + "apply", + "result", + "sparse", + "define", + "sum", + "make", + "easily", + "interpretable", + "asymptotic", + "regime", + "allowed", + "change", + "suppress", + "notation", + "constant", + "36", + "Suppose", + "minimum", + "eigenvalue", + "log", + "exists", + "constants", + "Proof", + "sequence", + "events", + "Lambda", + "phi", + "geq", + "You", + "to1", + "sufficiently", + "large", + "work", + "ldots", + "By", + "seek", + "lower", + "bound", + "models", + "maximum", + "Also", + "Var", + "ij", + "applying", + "Putting", + "things", + "2c", + "5.1", + "Using", + "practice0", + "practice", + "shows", + "suggests", + "constructing", + "alpha", + "level", + "upper", + "point", + "standard", + "quantity", + "interval", + "estimated", + "square", + "root", + "scaled", + "Sun", + "2012" + ], + "keywords": [ + "Debiased Lasso", + "High-dimensional inference", + "Confidence intervals" + ], + "connect_to": [ + { + "target": "b5f3d362-0c22-592e-a395-ac451df1c1bd", + "relation": "embeds", + "ref": "[images/image-11-$$ 5.1 Usi.jpg]", + "position": { + "start": 8792, + "end": 8824 + } + } + ] + } + }, + { + "chunk_id": "274c19ee-6fee-5491-b765-6cccf4846ff0", + "type": "text", + "content": "L. Breiman. Stacked regressions. Machine Learning, 24:49–64, 1996.\nC. Cortes and V. Vapnik. Support-vector networks. Machine learning, 20(3):273–297, 1995.\nA. Hoerl and R. Kennard. Ridge regression: Biased estimation for nonorthogonal problems. Technometrics, pages 55–67, 1970.\nG. S. Kimeldorf and G. Wahba. A correspondence between bayesian estimation on stochastic processes and smoothing by splines. The Annals of Mathematical Statistics, 41(2): 495-502, 1970.\nN. Meinshausen. Relaxed lasso. Computational Statistics and Data Analysis, 52:374–393, 2007.\nN. Meinshausen and P. Bühlmann. High dimensional graphs and variable selection with the lasso. Annals of Statistics, 34:1436-1462, 2006.\nA. Rahimi and B. Recht. Random features for large-scale kernel machines. In Advances in Neural Information Processing Systems, pages 1177-1184, 2007.\nB. Schölkopf, R. Herbrich, and A. J. Smola. A generalized representer theorem. In International Conference on Computational Learning Theory, pages 416–426. Springer, 2001.\nP. Spirtes, C. N. Glymour, and R. Scheines. Causation, prediction, and search. MIT press, 2000.\nT. Sun and C.-H. Zhang. Scaled sparse linear regression. Biometrika, page ass043, 2012.\nR. Tibshirani. Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society, Series B, 58:267–288, 1996.\nR. Tibshirani, M. Saunders, S. Rosset, J. Zhu, and K. Knight. Sparsity and smoothness via the fused lasso. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 67(1):91–108, 2005.\nP. Tseng. Convergence of a block coordinate descent method for nondifferentiable minimization. Journal of optimization theory and applications, 109(3):475-494, 2001.\nS. Van de Geer, P. Bühlmann, Y. Ritov, R. Dezeure, et al. On asymptotically optimal confidence regions and tests for high-dimensional models. The Annals of Statistics, 42(3):1166–1202, 2014.\nD. Wolpert. Stacked generalization. Neural Networks, 5:241-259, 1992.\nY. Yang, M. Pilanci, and M. J. Wainwright. Randomized sketches for kernels: Fast and optimal non-parametric regression. arXiv preprint arXiv:1501.06195, 2015.\nM. Yuan and Y. Lin. Model selection and estimation in regression with grouped variables. Journal of the Royal Statistical Society: Series B, 68:49–67, 2006.\nM. Yuan and Y. Lin. Model selection and estimation in the gaussian graphical model. Biometrika, 94(1):19–35, 2007.\nC.-H. Zhang. Nearly unbiased variable selection under minimax concave penalty. The Annals of statistics, pages 894-942, 2010.\nC.-H. Zhang and S. S. Zhang. Confidence intervals for low dimensional parameters in high dimensional linear models. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 76(1):217–242, 2014.\nH. Zou. The adaptive lasso and its oracle properties. Journal of the American Statistical Association, 101:1418–1429, 2006.", + "path": "notes.pdf/Bibliography", + "metadata": { + "length": 2863, + "summary": "This passage is a bibliography of 22 references on regression methods, including stacked regressions, support-vector networks, ridge regression, lasso, relaxed lasso, fused lasso, adaptive lasso, scaled sparse linear regression, grouped variables, Gaussian graphical models, minimax concave penalty, kernel methods, random features, representer theorem, and related topics in high-dimensional statistics and machine learning.", + "page_nums": [], + "document_top_summary": "This document includes: Chapter 1 Kernel machines, Chapter 2, Chapter 3 The Lasso and beyond, Chapter 4 Graphical modelling and causal inference, Chapter 5 High-dimensional inference, Bibliography", + "tokens": [ + "Breiman", + "Stacked", + "regressions", + "Machine", + "Learning", + "24", + "49", + "64", + "1996", + "Cortes", + "Vapnik", + "Support", + "vector", + "networks", + "learning", + "20", + "273", + "297", + "1995", + "Hoerl", + "Kennard", + "Ridge", + "regression", + "Biased", + "estimation", + "nonorthogonal", + "problems", + "Technometrics", + "pages", + "55", + "67", + "1970", + "Kimeldorf", + "Wahba", + "correspondence", + "bayesian", + "stochastic", + "processes", + "smoothing", + "splines", + "The", + "Annals", + "Mathematical", + "Statistics", + "41", + "495", + "502", + "Meinshausen", + "Relaxed", + "lasso", + "Computational", + "Data", + "Analysis", + "52", + "374", + "393", + "2007", + "hlmann", + "High", + "dimensional", + "graphs", + "variable", + "selection", + "34", + "1436", + "1462", + "2006", + "Rahimi", + "Recht", + "Random", + "features", + "large", + "scale", + "kernel", + "machines", + "In", + "Advances", + "Neural", + "Information", + "Processing", + "Systems", + "1177", + "1184", + "Sch", + "lkopf", + "Herbrich", + "Smola", + "generalized", + "representer", + "theorem", + "International", + "Conference", + "Theory", + "416", + "426", + "Springer", + "2001", + "Spirtes", + "Glymour", + "Scheines", + "Causation", + "prediction", + "search", + "MIT", + "press", + "2000", + "Sun", + "Zhang", + "Scaled", + "sparse", + "linear", + "Biometrika", + "page", + "ass043", + "2012", + "Tibshirani", + "Regression", + "shrinkage", + "Journal", + "Royal", + "Statistical", + "Society", + "Series", + "58", + "267", + "288", + "Saunders", + "Rosset", + "Zhu", + "Knight", + "Sparsity", + "smoothness", + "fused", + "Methodology", + "91", + "108", + "2005", + "Tseng", + "Convergence", + "block", + "coordinate", + "descent", + "method", + "nondifferentiable", + "minimization", + "optimization", + "theory", + "applications", + "109", + "475", + "494", + "Van", + "de", + "Geer", + "Ritov", + "Dezeure", + "al", + "On", + "asymptotically", + "optimal", + "confidence", + "regions", + "tests", + "high", + "models", + "42", + "1166", + "1202", + "2014", + "Wolpert", + "generalization", + "Networks", + "241", + "259", + "1992", + "Yang", + "Pilanci", + "Wainwright", + "Randomized", + "sketches", + "kernels", + "Fast", + "parametric", + "arXiv", + "preprint", + "1501.06195", + "2015", + "Yuan", + "Lin", + "Model", + "grouped", + "variables", + "68", + "gaussian", + "graphical", + "model", + "94", + "19", + "35", + "Nearly", + "unbiased", + "minimax", + "concave", + "penalty", + "statistics", + "894", + "942", + "2010", + "Confidence", + "intervals", + "low", + "parameters", + "76", + "217", + "242", + "Zou", + "adaptive", + "oracle", + "properties", + "American", + "Association", + "101", + "1418", + "1429" + ], + "keywords": [ + "Lasso", + "Ridge", + "Kernel" + ], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-statistical-learning/doc_nav.json b/apps/api/app/data/demo_documents/stem-statistical-learning/doc_nav.json new file mode 100644 index 000000000..6372734df --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-statistical-learning/doc_nav.json @@ -0,0 +1,559 @@ +{ + "version": "1.0", + "file_name": "Statistical Learning Notes.pdf", + "stats": { + "total_chunks": 71, + "text_chunks": 60, + "image_chunks": 11, + "table_chunks": 0, + "max_depth": 4 + }, + "sections": [ + { + "title": "Root", + "path": "notes.pdf", + "level": 1, + "summary": "This course covers modern statistical methods for high-dimensional data, where variables exceed observations. It introduces ridge regression, kernel machines, the Lasso for regression with more parameters than observations, graphical modeling for causal inference, and uncertainty quantification. It also reviews classical methods like ordinary least squares and maximum likelihood estimation.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Chapter 1 Kernel machines", + "path": "notes.pdf/Chapter 1 Kernel machines", + "level": 1, + "summary": "The passage discusses linear models and compares unbiased estimators with biased ones. It shows that the mean squared error of an estimator can be decomposed into variance plus squared bias. While OLS and MLEs are unbiased, biased methods can have lower variance, leading to better prediction error, especially when the number of parameters is large relative to observations. The text introduces penalisation techniques that shrink estimates toward zero, enabling reliable estimation when classical MLEs fail and often outperforming traditional approaches.", + "chunk_count": 14, + "children": [ + { + "title": "1.1 Ridge regression", + "path": "notes.pdf/Chapter 1 Kernel machines/1.1 Ridge regression", + "level": 2, + "summary": "The passage introduces ridge regression as a method to reduce the variance of OLS estimates by shrinking coefficients towards zero via a penalty term λ||β||₂². It explains the optimization problem, the closed-form solution after centering and scaling, and the bias-variance trade-off. A theorem states that for sufficiently small λ, ridge regression outperforms OLS in terms of mean squared error. The singular value decomposition (SVD) is used to show that ridge regression shrinks coordinates of Y by factors D_jj²/(D_jj²+λ), with more shrinkage on smaller principal components. This makes ridge effective when signal is concentrated in large principal components. The passage also discusses the need for selecting λ and the connection to principal components analysis.", + "chunk_count": 1, + "children": [] + }, + { + "title": "1.2 v-fold cross-validation", + "path": "notes.pdf/Chapter 1 Kernel machines/1.2 v-fold cross-validation", + "level": 2, + "summary": "The text discusses cross-validation as a technique for selecting a regression method, using ridge regression as an example. It explains the goal of minimizing expected prediction error for new observations, introduces v-fold cross-validation to estimate this error, and describes how to choose the optimal λ by minimizing the CV(λ) criterion. It also notes that cross-validation provides a biased estimate of prediction error, with bias depending on fold size, and mentions stacking as an alternative that finds optimal weighted combinations of regression procedures.", + "chunk_count": 1, + "children": [] + }, + { + "title": "1.3 The kernel trick", + "path": "notes.pdf/Chapter 1 Kernel machines/1.3 The kernel trick", + "level": 2, + "summary": "The passage discusses ridge regression, presenting the fitted values formula and an alternative computation using XX^T instead of X^TX to reduce complexity when p >> n. It introduces a quadratic signal model, showing how to compute the kernel matrix K directly using (1 + x_i^T x_j)^2, which requires O(n^2 p) operations instead of O(n^2 p^2 + n^3). This illustrates the kernel trick: ridge regression depends only on inner products, so non-linear feature maps can be handled by directly computing similarity measures k(x_i, x_j).", + "chunk_count": 1, + "children": [] + }, + { + "title": "1.4 Kernels", + "path": "notes.pdf/Chapter 1 Kernel machines/1.4 Kernels", + "level": 2, + "summary": "The passage discusses how non-linear models can be efficiently fitted by replacing the Gram matrix with a kernel matrix. It introduces the concept of a positive definite kernel as a symmetric map that yields positive semi-definite matrices. It shows that any inner product of feature maps defines a kernel, and notes that the converse also holds. A Cauchy-Schwarz inequality for kernels is proved via the determinant of a 2x2 kernel matrix.", + "chunk_count": 5, + "children": [ + { + "title": "1.4.1 Examples of kernels", + "path": "notes.pdf/Chapter 1 Kernel machines/1.4 Kernels/1.4.1 Examples of kernels", + "level": 3, + "summary": "The passage defines several kernels: linear, polynomial, Gaussian, Sobolev, and Jaccard similarity. It explains how to prove they are valid kernels using properties like sum, product, and limit of kernels. The Gaussian kernel is shown to be a kernel by decomposing it into a product of two kernels. The Sobolev kernel is the covariance of Brownian motion. The Jaccard kernel is defined for sets.", + "chunk_count": 1, + "children": [] + }, + { + "title": "1.4.2 Reproducing kernel Hilbert spaces", + "path": "notes.pdf/Chapter 1 Kernel machines/1.4 Kernels/1.4.2 Reproducing kernel Hilbert spaces", + "level": 3, + "summary": "Theorem 5 proves that for every kernel k, there exists a feature map φ into an inner product space H such that k(x,x') = ⟨φ(x), φ(x')⟩. The proof constructs H as the vector space of functions of the form f(·)=∑α_i k(·,x_i), with φ(x)=k(·,x). An inner product is defined on H, and its well-definedness is shown using the kernel's properties. Positive definiteness is established by noting that ⟨f,f⟩≥0 due to the kernel's positive definiteness, and Cauchy-Schwarz inequality is derived by treating the inner product as a kernel itself. The space H can be completed to a Hilbert space, specifically a reproducing kernel Hilbert space (RKHS), where the reproducing property f(x)=⟨k_x,f⟩ holds. The reproducing kernel of an RKHS is a positive definite kernel, and every kernel corresponds to a unique RKHS.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Examples", + "path": "notes.pdf/Chapter 1 Kernel machines/1.4 Kernels/Examples", + "level": 3, + "summary": "Linear kernel. Here $\\mathcal{H}=\\{f:f(x)=\\beta^{T}x,\\beta\\in\\mathbb{R}^{p}\\}$ and if $f(x)=\\beta^{T}x$ then $\\|f\\|_{H}^{2}=||\\beta||_{2}^{2}$ . Sobolev kernel. It can be shown that $\\mathcal{H}$ is roughly the space of continuous functions $f:[0,1]\\to \\mathbb{R}$ with $f(0) = 0$ that are differentiable almost everywhere, and for which $\\int_0^1 f'(x)^2 dx < \\infty$ . It contains the class of Lipschitz functions (functions $f:[0,1]\\to \\mathbb{R}$ for which there exists some $L$ with $|f(x) - f(y)|\\leq L|x - y|$ for all $x,y\\in [0,1]$ ) that are 0 at the origin. The norm is $$ \\left(\\int_ {0} ^ {1} f ^ {\\prime} (x) ^ {2} d x\\right) ^ {1 / 2}. $$ Though the construction of the RKHS from a kernel is explicit, it can be challenging to understand precisely the space and the form of the norm.", + "chunk_count": 1, + "children": [] + }, + { + "title": "1.4.3 The representer theorem", + "path": "notes.pdf/Chapter 1 Kernel machines/1.4 Kernels/1.4.3 The representer theorem", + "level": 3, + "summary": "The passage explains the Representer Theorem, which shows that the minimizer of a regularized empirical risk functional in a Reproducing Kernel Hilbert Space (RKHS) can be expressed as a linear combination of kernel evaluations at the training points. This result is key to kernel ridge regression, where the optimization problem reduces to finding coefficients α that minimize ||Y - Kα||² + λαᵀKα, leading to predictions f̂(x) = Σ αᵢ k(x, xᵢ). The theorem bridges infinite-dimensional optimization to a finite-dimensional problem.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "1.5 Kernel ridge regression", + "path": "notes.pdf/Chapter 1 Kernel machines/1.5 Kernel ridge regression", + "level": 2, + "summary": "The text discusses kernel ridge regression, focusing on its predictive performance. It presents Theorem 7, which bounds the mean squared prediction error (MSPE) in terms of kernel matrix eigenvalues and regularization parameter λ. The proof uses the representer theorem, eigendecomposition, and a trace trick. The bound is expressed as δ_n(λ_n) = σ²/(λ_n) * (1/n) Σ min(μ̂_i/4, λ_n) + λ_n/4. Using Mercer's theorem, the bound is related to kernel eigenvalues μ_j. For the Sobolev kernel, eigenvalues decay as 1/(π²(2j-1)²), leading to an optimal λ_n ~ (σ²/n)^{2/3} and error rate O((σ²/n)^{2/3}). The text concludes that kernel ridge regression is optimal up to a constant factor for MSPE when the true function lies in the RKHS.", + "chunk_count": 1, + "children": [] + }, + { + "title": "1.6 Other kernel machines", + "path": "notes.pdf/Chapter 1 Kernel machines/1.6 Other kernel machines", + "level": 2, + "summary": "Thus far we have we have only considered applying the kernel trick to ridge regression, which as we have seen has attractive theoretical properties as a regression method. However the kernel trick and the representer theorem are much more generally applicable. In settings where the $Y_{i}$ are not continuous but are in $\\{-1,1\\}$ (e.g. labels for spam and ham, fraud and not fraud etc.), popular approaches include kernel logistic regression and the support vector machine (SVM) [Cortes and Vapnik, 1995].", + "chunk_count": 3, + "children": [ + { + "title": "1.6.1 The support vector machine", + "path": "notes.pdf/Chapter 1 Kernel machines/1.6 Other kernel machines/1.6.1 The support vector machine", + "level": 3, + "summary": "The text derives the support vector classifier from the separable hyperplane case to the non-separable case using hinge loss and L2 regularization. It starts with maximizing the margin M subject to constraints, then introduces a penalty for misclassification, leading to the primal objective with hinge loss and ridge penalty. It extends to include an intercept term and generalizes to reproducing kernel Hilbert spaces (RKHS) via the representer theorem, resulting in the kernel SVM formulation with predictions based on kernel evaluations.", + "chunk_count": 1, + "children": [] + }, + { + "title": "1.6.2 Logistic regression", + "path": "notes.pdf/Chapter 1 Kernel machines/1.6 Other kernel machines/1.6.2 Logistic regression", + "level": 3, + "summary": "Recall that standard logistic regression may be motivated by assuming $$ \\log \\left(\\frac {\\mathbb {P} (Y _ {i} = 1)}{\\mathbb {P} (Y _ {i} = - 1)}\\right) = x _ {i} ^ {T} \\beta^ {0} $$ and picking $\\hat{\\beta}$ to maximise the log-likelihood. This leads to (see example sheet) the following optimisation problem: $$ \\underset {\\beta \\in \\mathbb {R} ^ {p}} {\\arg \\min} \\sum_ {i = 1} ^ {n} \\log \\{1 + \\exp (- Y _ {i} x _ {i} ^ {T} \\beta) \\}. $$ The ‘kernelised’ version is given by $$ \\underset {f \\in \\mathcal {H}} {\\arg \\min} \\left\\{\\sum_ {i = 1} ^ {n} \\log [ 1 + \\exp \\{- Y _ {i} f (x _ {i}) \\} ] + \\lambda \\| f \\| _ {\\mathcal {H}} ^ {2} \\right\\}, $$ where H is an RKHS. As in the case of the SVM, the representer theorem gives a finite-dimensional optimisation that is equivalent to the above.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "1.7 Large-scale kernel machines", + "path": "notes.pdf/Chapter 1 Kernel machines/1.7 Large-scale kernel machines", + "level": 2, + "summary": "The passage discusses the computational challenges of kernel methods for large datasets, particularly the O(n^3) cost of inverting the kernel matrix and O(n) prediction cost. It introduces random feature expansions as a solution, approximating shift-invariant kernels via random mappings based on Bochner's theorem. The approach uses random cosine features to achieve linear-time computation in n, with applications to kernel ridge regression.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Chapter 2", + "path": "notes.pdf/Chapter 2", + "level": 1, + "summary": "", + "chunk_count": 8, + "children": [ + { + "title": "Introduction to the Lasso", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso", + "level": 2, + "summary": "", + "chunk_count": 7, + "children": [ + { + "title": "2.1 Model selection", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.1 Model selection", + "level": 3, + "summary": "The passage revisits the linear model Y = Xβ⁰ + ε with E(ε)=0 and Var(ε)=σ²I. It introduces the concept of sparsity, where the true set S of nonzero coefficients has size s much smaller than the total number of variables p. The mean squared prediction error (MSPE) of ordinary least squares (OLS) is derived as (p/n)σ². If the relevant variables in S were known and used in a reduced model, the MSPE would be (s/n)σ², which can be much smaller. Additionally, parameter estimates from the reduced model are more accurate and easier to interpret. The passage then mentions that classical model selection strategies will be reviewed.", + "chunk_count": 3, + "children": [ + { + "title": "Best subset regression", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.1 Model selection/Best subset regression", + "level": 4, + "summary": "A natural approach to finding S is to consider all $2^{p}$ possible regression procedures each involving regressing the response on a different sets of explanatory variables $X_{M}$ where M is a subset of $\\{1,\\ldots,p\\}$ . We can then pick the best regression procedure using cross-validation (say). For general design matrices, this involves an exhaustive search over all subsets, so this is not really feasible for p > 50.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Forward selection", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.1 Model selection/Forward selection", + "level": 4, + "summary": "This can be seen as a greedy way of performing best subsets regression. Given a target model size m (the tuning parameter), this works as follows. 1. Start by fitting an intercept only model. 2. Add to the current model the predictor variable that reduces the residual sum of squares the most. 3. Continue step 2 until m predictor variables have been selected.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "2.2 The Lasso estimator", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.2 The Lasso estimator", + "level": 3, + "summary": "The passage introduces the Least absolute shrinkage and selection operator (Lasso), which estimates coefficients by minimizing a penalized least squares objective with an ℓ1 penalty. This penalty shrinks estimates toward zero and can force some coefficients to exactly zero, enabling simultaneous variable selection and parameter estimation. The Lasso's ℓ1 constraint creates a diamond-shaped feasible region with corners at axes, making it likely that the OLS solution contours intersect at these corners, leading to sparse solutions. This contrasts with ridge regression's ℓ2 penalty, which yields spherical constraints and non-zero shrinkage.", + "chunk_count": 3, + "children": [ + { + "title": "2.2.1 Convexity and the Lasso", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.2 The Lasso estimator/2.2.1 Convexity and the Lasso", + "level": 4, + "summary": "The passage discusses key properties of the Lasso regression method. It contrasts Lasso with Best Subset Regression, noting Lasso's convex optimization makes it computationally tractable for large-scale problems. It defines convex sets and functions, and provides Proposition 9 on convex function properties. The text addresses Lasso solution uniqueness, stating fitted values are unique (Proposition 10) even if coefficient estimates may not be. Section 2.2.2 covers prediction error of Lasso with no design assumptions, presenting Theorem 11 which gives a bound on prediction error with high probability when lambda is chosen appropriately, assuming a normal linear model with centered and scaled columns.", + "chunk_count": 1, + "children": [] + }, + { + "title": "2.2.3 Concentration inequalities I", + "path": "notes.pdf/Chapter 2/Introduction to the Lasso/2.2 The Lasso estimator/2.2.3 Concentration inequalities I", + "level": 4, + "summary": "The text discusses concentration inequalities, starting with Markov's inequality and the Chernoff bound, then introduces sub-Gaussian random variables. It defines sub-Gaussian variables, provides a tail bound (Proposition 12), and shows that bounded variables are sub-Gaussian (Hoeffding's lemma). Proposition 14 states that linear combinations of independent sub-Gaussian variables remain sub-Gaussian. Lemma 15 applies these results to bound the probability that the Lasso's regularization parameter exceeds a threshold, assuming sub-Gaussian errors, yielding a high-probability bound for the event that the Lasso's condition holds.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + } + ] + }, + { + "title": "Chapter 3 The Lasso and beyond", + "path": "notes.pdf/Chapter 3 The Lasso and beyond", + "level": 1, + "summary": "The Lasso and beyond In this chapter, we will study the Lasso in more depth and present more precise theory about its behaviour. We will first study the variable selection properties of the Lasso, which will require some basic results in convex analysis that we present below.", + "chunk_count": 16, + "children": [ + { + "title": "3.1 Some facts from optimisation theory and convex analysis", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.1 Some facts from optimisation theory and convex analysis", + "level": 2, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "3.1.1 Subgradients", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.1 Some facts from optimisation theory and convex analysis/3.1.1 Subgradients", + "level": 3, + "summary": "The passage defines subgradients and subdifferentials for convex functions, states basic properties (differentiability, scaling, sum rules), and presents the KKT condition for optimality. It then computes the subdifferential of the ℓ1-norm, showing it consists of vectors with infinity-norm ≤1 and whose nonzero entries equal the sign of the corresponding coordinate.", + "chunk_count": 1, + "children": [] + }, + { + "title": "3.1.2 The Lagrangian method", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.1 Some facts from optimisation theory and convex analysis/3.1.2 The Lagrangian method", + "level": 3, + "summary": "Consider an optimisation problem of the form $$ \\text { minimise } f (x), \\text { subject to } g (x) = 0 \\tag {3.1.2} $$ where $g : R^{d} \\to R^{b}$ . Suppose the optimal value is $c^{*} \\in R$ . The Lagrangian for this problem is defined as $$ L (x, \\theta) = f (x) + \\theta^ {T} g (x) $$ where $\\theta \\in \\mathbb{R}^b$ . Note that $$ \\inf _ {x \\in \\mathbb {R} ^ {d}} L (x, \\theta) \\leq \\inf _ {x \\in \\mathbb {R} ^ {d}: g (x) = 0} L (x, \\theta) = c ^ {*} $$ for all $\\theta$ . The Lagrangian method involves finding a $\\theta^{*}$ such that the minimising $x^{*}$ on the LHS satisfies $g(x^{*}) = 0$ . This $x^{*}$ must then be a minimiser in the original problem (3.1.2).", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "3.2 Lasso solutions", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.2 Lasso solutions", + "level": 2, + "summary": "Equipped with these tools from convex analysis, we can now fully characterise the solutions to the Lasso. We have that $\\hat{\\beta}_{\\lambda}^{L}$ is a Lasso solution if and only if $0 \\in \\partial Q_{\\lambda}(\\hat{\\beta}_{\\lambda}^{L})$ , which is equivalent to $$ \\frac {1}{n} X ^ {T} (Y - X \\hat {\\beta} _ {\\lambda} ^ {\\mathrm{L}}) = \\lambda \\hat {\\nu}, $$ for $\\hat{\\nu}$ with $\\| \\hat{\\nu}\\|_{\\infty}\\leq 1$ and writing $\\hat{S}_{\\lambda} = \\{k:\\hat{\\beta}_{\\lambda ,k}^{\\mathrm{L}}\\neq 0\\}$ , $\\hat{\\nu}_{\\hat{S}_{\\lambda}} = \\mathrm{sgn}(\\hat{\\beta}_{\\lambda ,\\hat{S}_{\\lambda}}^{\\mathrm{L}})$ . These are known as the KKT conditions for the Lasso.", + "chunk_count": 1, + "children": [] + }, + { + "title": "3.3 Variable selection", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.3 Variable selection", + "level": 2, + "summary": "Theorem 20 provides conditions for Lasso to recover the true sign pattern in a noiseless high-dimensional linear model. The irrepresentable condition requires that the maximum absolute dot product between the sign vector and regression coefficients of inactive variables on active ones is at most 1. Additionally, the true coefficients must exceed a threshold involving the inverse Gram matrix. Under these conditions, a Lasso solution exists with the correct sign pattern. The converse also holds: if a Lasso solution recovers the true signs, the irrepresentable condition must be satisfied.", + "chunk_count": 1, + "children": [] + }, + { + "title": "3.4 Prediction and estimation", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.4 Prediction and estimation", + "level": 2, + "summary": "The text discusses the compatibility condition for the Lasso estimator in high-dimensional linear regression. It defines the compatibility constant φ², which measures the restricted eigenvalue of the design matrix. Under this condition, the Lasso with tuning parameter λ = Aσ√(log(p)/n) achieves an MSPE bound of O(σ²s log(p)/(n φ²)) with high probability. The proof uses the basic inequality, Hölder's inequality, and the compatibility condition to derive the oracle-like bound.", + "chunk_count": 1, + "children": [] + }, + { + "title": "3.5 The compatibility condition", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.5 The compatibility condition", + "level": 2, + "summary": "The passage discusses the compatibility condition for a covariance matrix in high-dimensional statistics. It defines a quantity φ²(S) that measures compatibility, and presents Lemma 22 showing that if an estimated covariance matrix Σ̂ is close to a matrix Σ̃ with positive φ²(S), then Σ̂ also has positive φ²(S). The proof uses Hölder's inequality and bounds on the difference between matrices. The goal is to apply this with Σ̃ = Σ⁰ (population covariance) and use concentration inequalities to show that Σ̂ satisfies the compatibility condition with high probability when Σ⁰ does.", + "chunk_count": 3, + "children": [ + { + "title": "3.5.1 Concentration inequalities II", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.5 The compatibility condition/3.5.1 Concentration inequalities II", + "level": 3, + "summary": "The text introduces Bernstein's inequality for independent random variables satisfying Bernstein's condition with parameters (σ, b). It provides the moment condition, the moment generating function bound, and the tail probability bound. A lemma shows that the product of two mean-zero sub-Gaussian random variables satisfies Bernstein's condition with parameters (8σ_Wσ_Z, 4σ_Wσ_Z), enabling concentration results for such products.", + "chunk_count": 1, + "children": [] + }, + { + "title": "3.5.2 The compatibility condition and random design", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.5 The compatibility condition/3.5.2 The compatibility condition and random design", + "level": 3, + "summary": "The text presents Theorem 25 and Corollary 26 on the compatibility condition for sparse estimation. Theorem 25 shows that under i.i.d. sub-Gaussian rows with certain growth conditions, the empirical compatibility constant is at least half the population version with probability tending to 1. Corollary 26 extends this to normal distributions with bounded eigenvalues. The proof uses concentration inequalities and union bounds.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "3.6 Computation", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.6 Computation", + "level": 2, + "summary": "The text describes coordinate descent as an efficient optimization technique for Lasso solutions, minimizing functions of the form f(x)=g(x)+Σh_j(x_j) where g is convex and differentiable and each h_j is convex. It iteratively updates each coordinate by minimizing f with respect to that coordinate while holding others fixed. Tseng [2001] proves convergence to a minimizer under compactness of the sublevel set. Corollary 27 states existence of a minimizer and convergence of function values, with uniqueness ensuring convergence of iterates. The text also discusses solving Lasso over a grid of λ values using warm starts and an active set strategy: initializing active set A with non-zero coefficients from previous λ, performing coordinate descent only on A, checking KKT violations, and expanding A if violations exist.", + "chunk_count": 1, + "children": [] + }, + { + "title": "3.7 Extensions of the Lasso", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.7 Extensions of the Lasso", + "level": 2, + "summary": "We can add an $\\ell_{1}$ penalty to many other log-likelihoods, or more generally other loss functions besides the squared-error loss that arises from the normal linear model. For Lasso-penalised generalised linear models, such as logistic regression, similar theoretical results to those we have obtained are available and computations can proceed in a similar fashion to above.", + "chunk_count": 5, + "children": [ + { + "title": "3.7.1 Structural penalties", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.7 Extensions of the Lasso/3.7.1 Structural penalties", + "level": 3, + "summary": "The Lasso penalty encourages the estimated coefficients to be shrunk towards 0 and sometimes exactly to 0. Other penalty functions can be constructed to encourage different types of sparsity.", + "chunk_count": 3, + "children": [ + { + "title": "Group Lasso", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.7 Extensions of the Lasso/3.7.1 Structural penalties/Group Lasso", + "level": 4, + "summary": "Suppose we have a partition $G_{1},\\ldots ,G_{q}$ of $\\{1,\\dots ,p\\}$ (so $\\cup_{k = 1}^{q}G_{k} = \\{1,\\dots ,p\\}$ , $G_{j}\\cap G_{k} = \\emptyset$ for $j\\neq k$ ). The group Lasso penalty [Yuan and Lin, 2006] is given by $$ \\lambda \\sum_ {j = 1} ^ {q} m _ {j} \\| \\beta_ {G _ {j}} \\| _ {2}. $$ The multipliers $m_{j} > 0$ serve to balance cases where the groups are of very different sizes; typically we choose $m_{j} = \\sqrt{|G_{j}|}$ . This penalty encourages either an entire group G to have $\\hat{\\beta}_{G} = 0$ or $\\hat{\\beta}_{k} \\neq 0$ for all $k \\in G$ . Such a property is useful when groups occur through coding for categorical predictors or when expanding predictors using basis functions.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Fused Lasso", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.7 Extensions of the Lasso/3.7.1 Structural penalties/Fused Lasso", + "level": 4, + "summary": "If there is a sense in which the coefficients are ordered, so $\\beta_{j}^{0}$ is expected to be close to $\\beta_{j+1}^{0}$ , a fused Lasso penalty [Tibshirani et al., 2005] may be appropriate. This takes the form $$ \\lambda_ {1} \\sum_ {j = 1} ^ {p - 1} | \\beta_ {j} - \\beta_ {j + 1} | + \\lambda_ {2} \\| \\beta \\| _ {1}, $$ where the second term may be omitted depending on whether shrinkage towards 0 is desired. As an example, consider the simple setting where $Y_{i} = \\mu_{i}^{0} + \\varepsilon_{i}$ , and it is thought that the $(\\mu_{i}^{0})_{i=1}^{n}$ form a piecewise constant sequence. Then one option is to minimise over $\\mu \\in R^{n}$ , the following objective $$ \\frac {1}{n} \\| Y - \\mu \\| _ {2} ^ {2} + \\lambda \\sum_ {i = 1} ^ {n - 1} | \\mu_ {i} - \\mu_ {i + 1} |. $$", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "3.7.2 Reducing the bias of the Lasso", + "path": "notes.pdf/Chapter 3 The Lasso and beyond/3.7 Extensions of the Lasso/3.7.2 Reducing the bias of the Lasso", + "level": 3, + "summary": "The passage discusses drawbacks of the Lasso, which shrinks non-zero coefficients toward zero. Solutions include re-estimating non-zero coefficients via OLS or the relaxed Lasso, using the adaptive Lasso with weighted penalties, and employing non-convex penalties like MCP. Non-convex penalties may cause multiple local minima, but coordinate descent can work if non-convexity is mild.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + }, + { + "title": "Chapter 4 Graphical modelling and causal inference", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference", + "level": 1, + "summary": "Graphical modelling and causal inference So far we have considered the problem of relating a particular response to a large collection of explanatory variables. In some settings however, we do not have a distinguished response variable and instead we would like to better understand relationships between all the variables. In other situations, rather than being able to predict variables, we would like to understand causal relationships between them. Representing relationships between random variables through graphs will be an important tool in tackling these problems.", + "chunk_count": 19, + "children": [ + { + "title": "4.1 Graphs", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.1 Graphs", + "level": 2, + "summary": "The text defines fundamental graph concepts including directed/undirected edges, subgraphs, parents/children, v-structures, skeletons, paths, directed cycles, DAGs, d-separation, and moralised graphs. It also defines topological ordering and proves every DAG has one via induction and source node removal.", + "chunk_count": 1, + "children": [] + }, + { + "title": "4.2 Conditional independence graphs", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.2 Conditional independence graphs", + "level": 2, + "summary": "We would like to understand which variables may be ‘related’ to each other. Trying to find pairs of variables that are independent and so unlikely to be related to each other is not necessarily a good way to proceed as each variable may be correlated with a large number of variables without being directly related to them. A better approach is to use conditional independence. Definition 9. If X, Y and Z are random vectors with a joint density $f_{XYZ}$ (w.r.t. a product measure $\\mu$ ) then we say X is conditionally independent of Y given Z, and write $$ X \\perp Y | Z $$ if $$ f _ {X Y | Z} (x, y | z) = f _ {X | Z} (x | z) f _ {Y | Z} (y | z). $$ Equivalently $$ X \\perp \\perp Y | Z \\Longleftrightarrow f _ {X | Y Z} (x | y, z) = f _ {X | Z} (x | z). $$ We will first look at how undirected graphs can be used to visualise conditional independencies between random variables; thus in the next few subsections by graph we will mean undirected graph. Definition 10. Let $Z = (Z_{1}, \\ldots, Z_{p})^{T}$ be a collection of random variables with joint law P. The conditional independence graph (CIG) for P is the graph with $V = \\{1, \\ldots, p\\}$ and an edge $\\{j, k\\}$ between j and k if and only if $$ Z _ {j} \\nparallel Z _ {k} \\mid Z _ {- j k}. $$", + "chunk_count": 1, + "children": [] + }, + { + "title": "4.3 Gaussian graphical models", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.3 Gaussian graphical models", + "level": 2, + "summary": "Estimating the CIG given samples from P is a difficult task in general. However, in the case where P is multivariate Gaussian, things simplify considerably as we shall see. We begin with some notation. For a matrix $M \\in R^{p \\times p}$ , and sets $A, B \\subseteq \\{1, \\ldots, p\\}$ , let $M_{A,B}$ be the $|A| \\times |B|$ submatrix of M consisting of those rows and columns of M indexed by the sets A and B respectively. The submatrix extraction operation is always performed first (so e.g. $M_{k,-k}^{T} = (M_{k,-k})^{T}$ ).", + "chunk_count": 5, + "children": [ + { + "title": "4.3.1 Normal conditionals", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.3 Gaussian graphical models/4.3.1 Normal conditionals", + "level": 3, + "summary": "Now let $Z \\sim N_p(\\mu, \\Sigma)$ with $\\Sigma$ positive definite. Note $\\Sigma_{A,A}$ is also positive definite for any $A$ . Proposition 29. $$ Z _ {A} | Z _ {B} = z _ {B} \\sim N _ {| A |} (\\mu_ {A} + \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} (z _ {B} - \\mu_ {B}), \\Sigma_ {A, A} - \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} \\Sigma_ {B, A}) $$ Proof. Idea: write $Z_{A}=MZ_{B}+(Z_{A}-MZ_{B})$ with matrix $M\\in R^{|A|\\times|B|}$ such that $Z_{A}-MZ_{B}$ and $Z_{B}$ are independent, i.e. such that $$ \\operatorname{Cov} (Z _ {B}, Z _ {A} - M Z _ {B}) = \\Sigma_ {B, A} - \\Sigma_ {B, B} M ^ {T} = 0. $$ This occurs when we take $M^{T} = \\Sigma_{B,B}^{-1}\\Sigma_{B,A}$ . Because $Z_{A} - MZ_{B}$ and $Z_{B}$ are independent, the distribution of $Z_{A} - MZ_{B}$ conditional on $Z_{B} = z_{B}$ is equal to its unconditional distribution. Now $$ \\begin{array}{l} \\mathbb {E} (Z _ {A} - M Z _ {B}) = \\mu_ {A} - \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} \\mu_ {B} \\\\ \\mathrm{Var} (Z _ {A} - M Z _ {B}) = \\Sigma_ {A, A} + \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} \\Sigma_ {B, B} \\Sigma_ {B, B...Sigma_ {B, A} - 2 \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} \\Sigma_ {B, A} \\\\ = \\Sigma_ {A, A} - \\Sigma_ {A, B} \\Sigma_ {B, B} ^ {- 1} \\Sigma_ {B, A}. \\\\ \\end{array} $$ Since $MZ_{B}$ is a function of $Z_{B}$ and $Z_{A}-MZ_{B}$ is normally distributed, we have the result. ☐", + "chunk_count": 1, + "children": [] + }, + { + "title": "4.3.2 Nodewise regression", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.3 Gaussian graphical models/4.3.2 Nodewise regression", + "level": 3, + "summary": "The passage discusses neighborhood selection for estimating a conditional independence graph (CIG). It derives the conditional distribution of Z_k given Z_{-k} in a Gaussian setting, showing that zero coefficients in the regression imply conditional independence. It then describes using Lasso regression of each variable on all others to select neighbors, with OR and AND rules for edge inclusion. Finally, it mentions an alternative approach of directly estimating the precision matrix Omega.", + "chunk_count": 1, + "children": [] + }, + { + "title": "4.3.3 The precision matrix and conditional independence", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.3 Gaussian graphical models/4.3.3 The precision matrix and conditional independence", + "level": 3, + "summary": "The following facts about blockwise inversion of matrices will help us to interpret the mean and variance in Proposition 29. Proposition 30. Let $M \\in R^{p \\times p}$ be a symmetric positive definite matrix and suppose $$ M = \\left( \\begin{array}{c c} P & Q ^ {T} \\\\ Q & R \\end{array} \\right) $$ with P and R square matrices. The Schur complement of R is $P - Q^{T}R^{-1}Q =: S$ . We have that S is positive definite and $$ M ^ {- 1} = \\left( \\begin{array}{c c} S ^ {- 1} & - S ^ {- 1} Q ^ {T} R ^ {- 1} \\\\ - R ^ {- 1} Q S ^ {- 1} & R ^ {- 1} + R ^ {- 1} Q S ^ {- 1} Q ^ {T} R ^ {- 1} \\end{array} \\right). $$ Furthermore $\\det(M) = \\det(S)\\det(R)$ . Let $\\Omega = \\Sigma^{-1}$ be the precision matrix. Note that $\\Sigma_{k,k} - \\Sigma_{k,-k}\\Sigma_{-k,-k}^{-1}\\Sigma_{-k,k} = \\Omega_{kk}^{-1}$ , and more generally that $\\mathrm{Var}(Z_A|Z_{A^c}) = \\Omega_{A,A}^{-1}$ . Also, we see that $\\Sigma_{-k,-k}^{-1}\\Sigma_{-k,k} = -\\Omega_{kk}^{-1}\\Omega_{-k,k}$ , so $$ (\\Sigma_ {- k, - k} ^ {- 1} \\Sigma_ {- k, k}) _ {j} = 0 \\Leftrightarrow \\left\\{ \\begin{array}{l l} \\Omega_ {j, k} = 0 \\text {for} j < k \\\\ \\Omega_ {j + 1, k} = 0 \\text {for} j \\geq k. \\end{array} \\right. $$ Thus $$ Z _ {k} \\perp \\perp Z _ {j} | Z _ {- j k} \\Leftrightarrow \\Omega_ {j k} = 0. $$ This motivates another approach to estimating the CIG.", + "chunk_count": 1, + "children": [] + }, + { + "title": "4.3.4 The Graphical Lasso", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.3 Gaussian graphical models/4.3.4 The Graphical Lasso", + "level": 3, + "summary": "The passage derives the maximum likelihood estimate of the precision matrix Ω for a multivariate normal distribution, showing that the unpenalized estimate is S^{-1} when S is positive definite. It then introduces the graphical Lasso, which adds an L1 penalty λ‖Ω‖₁ to the negative log-likelihood to promote sparsity in Ω, enabling estimation of a conditional independence graph. The objective is convex, and diagonal elements are often not penalized.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "4.4 Structural equation models", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.4 Structural equation models", + "level": 2, + "summary": "The passage introduces structural equation models (SEMs) as a way to represent data generating processes, going beyond conditional independence graphs to answer interventional questions. It defines an SEM as a collection of equations where each variable is a function of its parents and independent noise, with the graph being a DAG. An artificial example illustrates how variables like course attendance depend on other factors. The SEM determines the joint distribution and allows querying distributions after setting variables to specific values.", + "chunk_count": 1, + "children": [] + }, + { + "title": "4.5 Interventions", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.5 Interventions", + "level": 2, + "summary": "The text explains how replacing a structural equation in an SEM creates a new model, with a perfect intervention setting a variable to a constant (e.g., do(Z_k=a)). It contrasts interventional expectations (e.g., E(Z_j|do(Z_k=a))) with conditional expectations, using an example where do(Z_2=1) yields P(Z_1=1|do(Z_2=1))=9/16, while P(Z_1=1|Z_2=1)=7/12, showing they differ.", + "chunk_count": 1, + "children": [] + }, + { + "title": "4.6 The Markov properties on DAGs", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.6 The Markov properties on DAGs", + "level": 2, + "summary": "The DAG of an SEM can encode a number of conditional independencies present in the law of the random vector Z. To understand this, we first introduce some so-called Markov properties on DAGs. Let $P$ be the joint law of $Z$ and suppose it has a density $f$ . Definition 12. Given a DAG G, we say P satisfies the (i) Markov factorisation property w.r.t. the DAG $\\mathcal{G}$ if $$ f (z _ {1}, \\ldots , z _ {p}) = \\prod_ {k = 1} ^ {p} f (z _ {k} | z _ {\\mathrm{pa} (k)}). $$ (ii) global Markov property w.r.t. the DAG G if for all disjoint A, B, S ⊆ {1, . . . , p}, $$ A, B d \\text {-separated by} S \\Rightarrow Z _ {A} \\perp \\perp Z _ {B} | Z _ {S}. $$ Theorem 31. If P has a density f (with respect to a product measure), then all Markov properties in definition 12 are equivalent. In view of this, we will henceforth use the term Markov to mean global Markov. Proposition 32. Let P be the law of an SEM with DAG G. Then P obeys the Markov factorisation property w.r.t. G. Thus we can read off from the DAG of an SEM a great deal...properties of the joint distribution, but also to determine the effects of certain perturbations to the system. In many settings, we may not have a prespecified SEM to work with, but instead we would like to learn the DAG from observational data. This is the problem we turn to next.", + "chunk_count": 1, + "children": [] + }, + { + "title": "4.7 Causal structure learning", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning", + "level": 2, + "summary": "Given a sample of observations from P, we would like to determine the DAG which generated it. We can think of this task in terms of two subtasks: firstly we need to understand how to extract information concerning P from a sample, which is a traditional statistical question of the sort we are used to; secondly, given P itself, we need to relate this to the DAG which generated it. The latter problem is unique to casual inference and we discuss this first.", + "chunk_count": 8, + "children": [ + { + "title": "4.7.1 Three obstacles", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.1 Three obstacles", + "level": 3, + "summary": "There are three obstacles to causal structure learning. The first two are more immediate but the last is somewhat subtle.", + "chunk_count": 4, + "children": [ + { + "title": "Causal minimality", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.1 Three obstacles/Causal minimality", + "level": 4, + "summary": "We know that if P is generated by an SEM with DAG G, then P will be Markov w.r.t. G. Conversely, one can show that if P is Markov w.r.t. a DAG G, then there is also an SEM with DAG G that could have generated P. But P will be Markov w.r.t. a great number of DAGs, e.g. $Z_{1}$ and $Z_{2}$ being independent can be represented by $$ Z _ {1} = 0 \\times Z _ {2} + \\varepsilon_ {1} = \\varepsilon_ {1}, \\qquad Z _ {2} = \\varepsilon_ {2}. $$ This motivates the following definition. Definition 13. P satisfies causal minimality with respect to G if it is (global) Markov w.r.t. G but not to a proper subgraph of G with the same nodes.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Markov equivalent DAGs", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.1 Three obstacles/Markov equivalent DAGs", + "level": 4, + "summary": "It is possible for two different DAGs to satisfy the same collection of d-separations e.g. $$ For a DAG $\\mathcal{G}$ , let0 [images/image-9-$$ For a D.jpg] This motivates the following definition. For a DAG $\\mathcal{G}$ , let0 [images/image-10-This motiv.jpg] For a DAG $\\mathcal{G}$ , let $\\mathcal{M}(\\mathcal{G}) = \\{\\text{distributions } P : P \\text{ satisfies the global Markov property w.r.t. } \\mathcal{G}\\}$ . Definition 14. We say two DAGs $\\mathcal{G}_1$ and $\\mathcal{G}_2$ are Markov equivalent if $\\mathcal{M}(\\mathcal{G}_1) = \\mathcal{M}(\\mathcal{G}_2)$ . Proposition 33. Two DAGs are Markov equivalent if and only if they have the same skeleton and v-structures. The set of all DAGs that are Markov equivalent to a DAG can be represented by a completed PDAG (CPDAG) which contains an edge $(j,k)$ if and only if one member of the Markov equivalence class does. We can only ever hope to obtain the Markov equivalence class i.e. the CPDAG of a DAG with which P satisfies causal minimality (unless we place restrictions on the functional forms of the SEM equations).", + "chunk_count": 1, + "children": [] + }, + { + "title": "Faithfulness", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.1 Three obstacles/Faithfulness", + "level": 4, + "summary": "The text discusses two structural equation models (SEMs) generating the same multivariate normal distribution P^0. The first SEM has variables Z1, Z2, Z3 with parameters α, β, γ, and under the condition β+αγ=0 (e.g., β=-1, α=γ=1), Z1 is independent of Z3. A second SEM with different structure and error distributions also generates P^0. The concept of faithfulness is introduced: a distribution is faithful to a DAG if all conditional independencies are represented in the DAG. P^0 is not faithful to the first DAG (G) but is faithful to the second (tilde G).", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "4.7.2 The PC algorithm", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.2 The PC algorithm", + "level": 3, + "summary": "The text presents two propositions about d-separation in directed acyclic graphs (DAGs). Proposition 34 states that adjacent nodes in a DAG cannot be d-separated by any set, while non-adjacent nodes can be d-separated by the parent set of the later node in a topological order. This allows recovery of the DAG's skeleton by testing conditional independence. Proposition 35 addresses v-structures: in a triple where only j and k are non-adjacent, if they are in a v-structure (j→l←k), no d-separating set for j and k can contain l; conversely, if a d-separating set exists without l, the structure must be a v-structure. These results help identify v-structures from the skeleton and d-separating sets.", + "chunk_count": 3, + "children": [ + { + "title": "Population version", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.2 The PC algorithm/Population version", + "level": 4, + "summary": "The text describes the PC-algorithm for causal structure learning, which efficiently finds the skeleton of a directed acyclic graph (DAG) by testing conditional independencies. Algorithm 1 iteratively removes edges between nodes if they are conditionally independent given some subset of neighbors, assuming faithfulness to the true DAG. Algorithm 2 then orients v-structures for non-adjacent nodes sharing a common neighbor, using stored separating sets.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Sample version", + "path": "notes.pdf/Chapter 4 Graphical modelling and causal inference/4.7 Causal structure learning/4.7.2 The PC algorithm/Sample version", + "level": 4, + "summary": "The sample version of the PC algorithm replaces the querying of conditional independence with a conditional independence test applied to data $x_{1},\\ldots,x_{n}$ . The level of the test $\\alpha$ will be a tuning parameter of the method. If the data are assumed to be multivariate normal, the (sample) partial correlation can be used to test conditional independence since if $Z_{j} \\perp \\perp Z_{k}|Z_{S}$ then $$ \\mathrm{Corr} (Z _ {j}, Z _ {k} | Z _ {S}) := \\rho_ {j k \\cdot S} = 0. $$ To compute the sample partial correlation, we regress $X_{j}$ and $X_{k}$ on $X_{S}$ and compute the correlation between the resulting residuals.", + "chunk_count": 1, + "children": [] + } + ] + } + ] + } + ] + }, + { + "title": "Chapter 5 High-dimensional inference", + "path": "notes.pdf/Chapter 5 High-dimensional inference", + "level": 1, + "summary": "The text discusses high-dimensional inference using the debiased Lasso. It explains how to form confidence intervals for regression coefficients when p >> n by constructing an approximate inverse of the sample covariance matrix. The key idea is to debias the Lasso estimator so that its distribution is approximately normal, enabling inference. The method relies on sparsity conditions and uses node-wise Lasso regressions to estimate the inverse covariance matrix. Theorem 36 provides asymptotic normality results, showing that the debiased estimator's error term is small under certain conditions. Practical construction of confidence intervals uses the estimated variance from the debiased Lasso.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Bibliography", + "path": "notes.pdf/Bibliography", + "level": 1, + "summary": "This passage is a bibliography of 22 references on regression methods, including stacked regressions, support-vector networks, ridge regression, lasso, relaxed lasso, fused lasso, adaptive lasso, scaled sparse linear regression, grouped variables, Gaussian graphical models, minimax concave penalty, kernel methods, random features, representer theorem, and related topics in high-dimensional statistics and machine learning.", + "chunk_count": 1, + "children": [] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-Support Vector Margin Visualization.jpg", + "summary": "image-1 - Diagram illustrating the geometric margin in a Support Vector Machine (SVM) classification problem. - Two classes of data points are plotted: red circles ($y_i = 1$) and blue crosses ($y_i = -1$). - A solid line represents the decision boundary (hyperplane), and dashed lines represent the margins. - An arrow labeled $\\beta$ indicates the normal vector to the hyperplane. - The distance from a point $x_i$ to the hyperplane is annotated as $\\frac{y_i x_i^T \\beta}{\\|\\beta\\|_2}$. - The total margin width between the two dashed lines is labeled $M$." + }, + { + "path": "images/image-2-Propositio.jpg", + "summary": "image-2 Proposition 14. Let $(W_{i})_{i=1}^{n}$ be a sequence of independent mean-zero sub-Gaussian random variables with parameters $(\\sigma_{i})_{i=1}^{n}$ and let $\\gamma \\in R^{n}$ . Then $\\gamma^{T}W$ is sub-Gaussian with parameter $\\left(\\sum_{i}\\gamma_{i}^{2}\\sigma_{i}^{2}\\right)^{1/2}$ . We can now prove a more general version of the probability bound required for Theorem 11.0" + }, + { + "path": "images/image-3-$$ # Chapt.jpg", + "summary": "image-3 $$ # Chapter 3 The Lasso and beyond0" + }, + { + "path": "images/image-4-From this .jpg", + "summary": "image-4 From this we get ## 3.5 The compatibility condition0" + }, + { + "path": "images/image-5-$$ We woul.jpg", + "summary": "image-5 $$ We would like to apply the result above with $\\check{\\Sigma} = \\Sigma^{0}$ , and use to argue that if $\\Sigma^{0}$ satisfies the compatibility condition, then so will $\\hat{\\Sigma}$ with high probability. In order to do this, we need to argue that the event that $\\max_{jk} |\\hat{\\Sigma}_{jk} - \\Sigma_{jk}^{0}|$ is small occurs with high probability. We can obtain such a result with the aid of concentration inequalities.0" + }, + { + "path": "images/image-6-Let $Z = (.jpg", + "summary": "image-6 Let $Z = (Z_{1},\\ldots ,Z_{p})^{T}$ be a collection of random variables. The graphs we will consider will always have $V = \\{1,\\dots ,p\\}$ so $V$ indexes the random variables.
    0" + }, + { + "path": "images/image-7-Z2 --_ Z3 .jpg", + "summary": "image-7 Z2 --> Z3
    0" + }, + { + "path": "images/image-8-Example 4..jpg", + "summary": "image-8 Example 4.4.1. Consider the following (totally artificial) SEM which has whether you are taking this course $(Z_{1}=1)$ depending on whether you went to the statistical modelling course $(Z_{2}=1)$ and whether you have heard about machine learning $(Z_{3}=1)$ . Suppose
    0" + }, + { + "path": "images/image-9-$$ For a D.jpg", + "summary": "image-9 $$ For a DAG $\\mathcal{G}$ , let0" + }, + { + "path": "images/image-10-This motiv.jpg", + "summary": "image-10 This motivates the following definition. For a DAG $\\mathcal{G}$ , let0" + }, + { + "path": "images/image-11-$$ 5.1 Usi.jpg", + "summary": "image-11 $$ 5.1 Using the debiased Lasso in practice0" + } + ], + "tables": [] + } +} diff --git a/apps/api/app/data/demo_documents/stem-statistical-learning/full.md b/apps/api/app/data/demo_documents/stem-statistical-learning/full.md new file mode 100644 index 000000000..8f03b6b76 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-statistical-learning/full.md @@ -0,0 +1,2427 @@ +# Statistical Learning + +Rajen D. Shah + +r.shah@statslab.cam.ac.uk + +Course webpage: http://www.statslab.cam.ac.uk/\~rds37/stat\_learning.html + +In this course we will study a selection of important modern statistical methods. This selection is heavily biased towards my own interests, but I hope it will nevertheless give you a flavour of some of the most important recent methodological developments in statistics. + +Over the last 25 years, the sorts of datasets that statisticians have been challenged to study have changed greatly. Where in the past, we were used to datasets with many observations with a few carefully chosen variables, we are now seeing datasets where the number of variables can run into the thousands and greatly exceed the number of observations. For example, with microarray data, we typically have gene expression values measured for several thousands of genes, but only for a few hundred tissue samples. The classical statistical methods are often simply not applicable in these “high-dimensional” situations. + +The course is divided into 5 chapters (of unequal size). Module I encompasses the first two chapters, and Module II covers the last three. Our first chapter will start by introducing ridge regression, a simple generalisation of ordinary least squares. Our study of this will lead us to some beautiful connections with functional analysis and ultimately one of the most successful and flexible classes of learning algorithms: kernel machines. + +The second chapter will give an introduction to the Lasso, a method which has been at the centre of much of the developments that have occurred in high-dimensional statistics, and will allow us to perform regression in the seemingly hopeless situation when the number of parameters we are trying to estimate is larger than the number of observations. + +Chapter three gives and more in depth study of the Lasso, and some of its extensions. In the fourth chapter we will study graphical modelling and provide an introduction to the exciting field of causal inference. Where the previous chapters consider methods for relating a particular response to a large collection of (explanatory) variables, graphical modelling will give us a way of understanding relationships between the variables themselves. Ultimately we would like to infer causal relationships between variables based on (observational) data. This may seem like a fundamentally impossible task, yet we will show how by developing the graphical modelling framework further, we can begin to answer such causal questions. + +Statistics is not only about developing methods that can predict well in the presence of noise, but also about assessing the uncertainty in our predictions and estimates. In the final chapter we will tackle the problem of how to quantify uncertainty in high-dimensional regression settings. + +Before we begin the main material of the course, we will briefly review two key classical statistical methods: ordinary least squares and maximum likelihood estimation. This will help to set the scene and provide a warm-up for the modern methods to come later. + +## Classical statistics + +## Ordinary least squares + +Imagine data are available in the form of observations $(Y_{i}, x_{i}) \in \mathbb{R} \times \mathbb{R}^{p}, i = 1, \ldots, n$ , and the aim is to infer a simple regression function relating the average value of a response, $Y_{i}$ , and a collection of predictors or variables, $x_{i}$ . This is an example of regression analysis, one of the most important tasks in statistics. + +A linear model for the data assumes that it is generated according to + +$$ +Y = X \beta^ {0} + \varepsilon , \tag {0.0.1} +$$ + +where $Y \in R^{n}$ is the vector of responses; $X \in R^{n \times p}$ is the predictor matrix (or design matrix) with ith row $x_{i}^{T}$ ; $\varepsilon \in R^{n}$ represents random error; and $\beta^{0} \in R^{p}$ is the unknown vector of coefficients. + +Provided $p \ll n$ , a sensible way to estimate $\beta$ is by ordinary least squares (OLS). This yields an estimator $\hat{\beta}^{OLS}$ with + +$$ +\hat {\beta} ^ {\mathrm{OLS}} := \underset {\beta \in \mathbb {R} ^ {p}} {\arg \min} \| Y - X \beta \| _ {2} ^ {2} = (X ^ {T} X) ^ {- 1} X ^ {T} Y, \tag {0.0.2} +$$ + +provided $X$ has full column rank. + +Under the assumptions that (i) $\mathbb{E}(\varepsilon_i) = 0$ and (ii) $\operatorname{Var}(\varepsilon) = \sigma^2 I$ , we have that: + +- $\mathbb{E}_{\beta^0,\sigma^2}(\hat{\beta}^{\mathrm{OLS}}) = \mathbb{E}\{(X^T X)^{-1}X^T(X\beta^0 + \varepsilon)\} = \beta^0.$ +- $\mathrm{Var}_{\beta^0,\sigma^2}(\hat{\beta}^{\mathrm{OLS}}) = (X^TX)^{-1}X^TX\mathrm{Var}(\varepsilon)X(X^TX)^{-1} = \sigma^2 (X^TX)^{-1}.$ + +The Gauss–Markov theorem states that OLS is the best linear unbiased estimator in our setting: for any other estimator $\tilde{\beta}$ that is linear in Y (so $\tilde{\beta} = AY$ for some fixed matrix A), we have + +$$ +\mathrm{Var} _ {\beta^ {0}, \sigma^ {2}} (\tilde {\beta}) - \mathrm{Var} _ {\beta^ {0}, \sigma^ {2}} (\hat {\beta} ^ {\mathrm{OLS}}) +$$ + +is positive semi-definite. + +## Maximum likelihood estimation + +The method of least squares is just one way to construct as estimator. A more general technique is that of maximum likelihood estimation. Here given data $y \in R^{n}$ that we take as a realisation of a random variable Y, we specify its density $f(y; \theta)$ up to some unknown vector of parameters $\theta \in \Theta \subseteq R^{d}$ , where $\Theta$ is the parameter space. The likelihood function is a function of $\theta$ for each fixed y given by + +$$ +L (\theta) := L (\theta ; y) = c (y) f (y; \theta), +$$ + +where $c(y)$ is an arbitrary constant of proportionality. The maximum likelihood estimate of $\theta$ maximises the likelihood, or equivalently it maximises the log-likelihood + +$$ +\ell (\theta) := \ell (\theta ; y) = \log f (y; \theta) + \log (c (y)). +$$ + +A very useful quantity in the context of maximum likelihood estimation is the Fisher information matrix with jkth $(1 \leq j, k \leq d)$ entry + +$$ +i _ {j k} (\theta) := - \mathbb {E} _ {\theta} \bigg \{\frac {\partial^ {2}}{\partial \theta_ {j} \partial \theta_ {k}} \ell (\theta) \bigg \}. +$$ + +It can be thought of as a measure of how hard it is to estimate $\theta$ when it is the true parameter value. The Cramér–Rao lower bound states that if $\tilde{\theta}$ is an unbiased estimator of $\theta$ , then under regularity conditions, + +$$ +\mathrm{Var} _ {\theta} (\tilde {\theta}) - i ^ {- 1} (\theta) +$$ + +is positive semi-definite. + +A remarkable fact about maximum likelihood estimators (MLEs) is that (under quite general conditions) they are asymptotically normally distributed, asymptotically unbiased and asymptotically achieve the Cramér–Rao lower bound. + +Assume that the Fisher information matrix when there are n observations, $i^{(n)}(\theta)$ (where we have made the dependence on n explicit) satisfies $i^{(n)}(\theta)/n \to I(\theta)$ for some positive definite matrix I. Then denoting the maximum likelihood estimator of $\theta$ when there are n observations by $\hat{\theta}^{(n)}$ , under regularity conditions, as the number of observations $n \to \infty$ we have + +$$ +\sqrt {n} (\hat {\theta} ^ {(n)} - \theta) \xrightarrow {d} N _ {d} (0, I ^ {- 1} (\theta)). +$$ + +Returning to our linear model, if we assume in addition that $\varepsilon_{i} \sim N(0, \sigma^{2})$ , then the log-likelihood for $(\beta, \sigma^{2})$ is + +$$ +\ell (\beta , \sigma^ {2}) = - \frac {n}{2} \log (\sigma^ {2}) - \frac {1}{2 \sigma^ {2}} \sum_ {i = 1} ^ {n} (y _ {i} - x _ {i} ^ {T} \beta) ^ {2}. +$$ + +We see that the maximum likelihood estimate of $\beta$ and OLS coincide. It is easy to check that + +$$ +i (\beta , \sigma^ {2}) = \left( \begin{array}{c c} \sigma^ {- 2} X ^ {T} X & 0 \\ 0 & n \sigma^ {- 4} / 2 \end{array} \right). +$$ + +The general theory for MLEs would suggest that approximately $\sqrt{n}(\hat{\beta}-\beta)\sim N_{p}(0,n\sigma^{2}(X^{T}X)^{-1})$ ; in fact it is straight-forward to show that this distributional result is exact. + +## Contents + +## 1 Kernel machines 1 + +1.1 Ridge regression 1 + +1.1.1 The singular value decomposition and principal components analysis 3 + +1.2 v-fold cross-validation ..... 4 + +1.3 The kernel trick 5 + +1.4 Kernels 7 + +1.4.1 Examples of kernels 8 + +1.4.2 Reproducing kernel Hilbert spaces . . . . . . . . . . . . . . . . . . 9 + +1.4.3 The representer theorem 12 + +1.5 Kernel ridge regression 13 + +1.6 Other kernel machines 16 + +1.6.1 The support vector machine 16 + +1.6.2 Logistic regression 18 + +1.7 Large-scale kernel machines 19 + +## 2 Introduction to the Lasso 21 + +2.1 Model selection 21 + +2.2 The Lasso estimator 22 + +2.2.1 Convexity and the Lasso 23 + +2.2.2 Prediction error of the Lasso with no assumptions on the design . . 24 + +2.2.3 Concentration inequalities I 24 + +## 3 The Lasso and beyond 27 + +3.1 Some facts from optimisation theory and convex analysis ..... 27 + +3.1.1 Subgradients 27 + +3.1.2 The Lagrangian method 29 + +3.2 Lasso solutions 29 + +3.3 Variable selection 29 + +3.4 Prediction and estimation 30 + +3.5 The compatibility condition 32 + +3.5.1 Concentration inequalities II 33 + +3.5.2 The compatibility condition and random design ..... 34 + +3.6 Computation 35 + +3.7 Extensions of the Lasso 36 + +3.7.1 Structural penalties 36 +3.7.2 Reducing the bias of the Lasso 37 + +## 4 Graphical modelling and causal inference 39 + +4.1 Graphs 39 +4.2 Conditional independence graphs 41 +4.3 Gaussian graphical models 42 + +4.3.1 Normal conditionals 42 +4.3.2 Nodewise regression 42 +4.3.3 The precision matrix and conditional independence ..... 43 +4.3.4 The Graphical Lasso 44 + +4.4 Structural equation models 45 +4.5 Interventions 46 +4.6 The Markov properties on DAGs 47 +4.7 Causal structure learning 47 + +4.7.1 Three obstacles 47 +4.7.2 The PC algorithm 49 + +## 5 High-dimensional inference 52 + +5.1 Using the debiased Lasso in practice 55 + +## Chapter 1 + +## Kernel machines + +Let us revisit the linear model with + +$$ +Y _ {i} = x _ {i} ^ {T} \beta^ {0} + \varepsilon_ {i}. +$$ + +For unbiased estimators of $\beta^{0}$ , their variance gives a way of comparing their quality in terms of squared error loss. For a potentially biased estimator, $\tilde{\beta}$ , the relevant quantity is + +$$ +\begin{array}{l} \mathbb {E} _ {\beta^ {0}, \sigma^ {2}} \{(\tilde {\beta} - \beta^ {0}) (\tilde {\beta} - \beta^ {0}) ^ {T} \} = \mathbb {E} [ \{\tilde {\beta} - \mathbb {E} (\tilde {\beta}) + \mathbb {E} (\tilde {\beta}) - \beta^ {0} \} \{\tilde {\beta} - \mathbb {E} (\tilde {\beta}) + \mathbb {E} (\tilde {\beta}) - \beta^ {0} \} ^ {T} ] \\ = \mathrm{Var} (\tilde {\beta}) + \{\mathbb {E} (\tilde {\beta} - \beta^ {0}) \} \{\mathbb {E} (\tilde {\beta} - \beta^ {0}) \} ^ {T}, \\ \end{array} +$$ + +a sum of squared bias and variance terms. A crucial part of the optimality arguments for OLS and MLEs was unbiasedness. Do there exist biased methods whose variance is reduced compared to OLS such that their overall prediction error is lower? Yes!—in fact the use of biased estimators is essential in dealing with settings where the number of parameters to be estimated is large compared to the number of observations. In the first two chapters we’ll explore two important methods for variance reduction based on different forms of penalisation: rather than forming estimators via optimising a least squares or log-likelihood term, we will introduce an additional penalty term that encourages estimates to be shrunk towards 0 in some sense. This will allow us to produce reliable estimators that work well when classical MLEs are infeasible, and in other situations can greatly out-perform the classical approaches. + +## 1.1 Ridge regression + +One way to reduce the variance of $\hat{\beta}^{\mathrm{OLS}}$ is to shrink the estimated coefficients towards 0. Ridge regression [Hoerl and Kennard, 1970] does this by solving the following optimisation problem + +$$ +(\hat {\mu} _ {\lambda} ^ {\mathrm{R}}, \hat {\beta} _ {\lambda} ^ {\mathrm{R}}) = \underset {(\mu , \beta) \in \mathbb {R} \times \mathbb {R} ^ {p}} {\arg \min} \{\| Y - \mu \mathbf {1} - X \beta \| _ {2} ^ {2} + \lambda \| \beta \| _ {2} ^ {2} \}. +$$ + +Here 1 is an n-vector of 1's. We see that the usual OLS objective is penalised by an additional term proportional to $\|\beta\|_{2}^{2}$ . The parameter $\lambda \geq 0$ , which controls the severity of the penalty and therefore the degree of the shrinkage towards 0, is known as a regularisation parameter or tuning parameter. We have explicitly included an intercept term which is not penalised. The reason for this is that were the variables to have their origins shifted so e.g. a variable representing temperature is given in units of Kelvin rather than Celsius, the fitted values would not change. However, $X\hat{\beta}$ is not invariant under scale transformations of the variables so it is standard practice to centre each column of X (hence making them orthogonal to the intercept term) and then scale them to have $\ell_{2}$ -norm $\sqrt{n}$ . + +It is straightforward to show that after this standardisation of X, $\hat{\mu}_{\lambda}^{R} = \bar{Y} := \sum_{i=1}^{n} Y_{i}/n$ , so we may assume that $\sum_{i=1}^{n} Y_{i} = 0$ by replacing $Y_{i}$ by $Y_{i} - \bar{Y}$ and then we can remove $\mu$ from our objective function. In this case + +$$ +\hat {\beta} _ {\lambda} ^ {\mathrm{R}} = (X ^ {T} X + \lambda I) ^ {- 1} X ^ {T} Y. +$$ + +In this form, we can see how the addition of the $\lambda I$ term helps to stabilise the estimator. Note that when X does not have full column rank (such as in high-dimensional situations), we can still compute this estimator. On the other hand, when X does have full column rank, we have the following theorem. + +Theorem 1. For $\lambda$ sufficiently small (depending on $\beta^{0}$ and $\sigma^{2}$ ), + +$$ +\mathbb {E} (\hat {\beta} ^ {\mathrm{OLS}} - \beta^ {0}) (\hat {\beta} ^ {\mathrm{OLS}} - \beta^ {0}) ^ {T} - \mathbb {E} (\hat {\beta} _ {\lambda} ^ {\mathrm{R}} - \beta^ {0}) (\hat {\beta} _ {\lambda} ^ {\mathrm{R}} - \beta^ {0}) ^ {T} +$$ + +is positive definite. + +Proof. First we compute the bias of $\hat{\beta}_{\lambda}^{\mathrm{R}}$ . We drop the subscript $\lambda$ and superscript $R$ for convenience. + +$$ +\begin{array}{l} \mathbb {E} (\hat {\beta}) - \beta^ {0} = (X ^ {T} X + \lambda I) ^ {- 1} X ^ {T} X \beta^ {0} - \beta^ {0} \\ = (X ^ {T} X + \lambda I) ^ {- 1} (X ^ {T} X + \lambda I - \lambda I) \beta^ {0} - \beta^ {0} \\ = - \lambda (X ^ {T} X + \lambda I) ^ {- 1} \beta^ {0}. \\ \end{array} +$$ + +Now we look at the variance of $\hat{\beta}$ . + +$$ +\begin{array}{l} \operatorname{Var} (\hat {\beta}) = \mathbb {E} \{(X ^ {T} X + \lambda I) ^ {- 1} X ^ {T} \varepsilon \} \{(X ^ {T} X + \lambda I) ^ {- 1} X ^ {T} \varepsilon \} ^ {T} \\ = \sigma^ {2} (X ^ {T} X + \lambda I) ^ {- 1} X ^ {T} X (X ^ {T} X + \lambda I) ^ {- 1}. \\ \end{array} +$$ + +Thus $\mathbb{E}(\hat{\beta}^{\mathrm{OLS}} - \beta^{0})(\hat{\beta}^{\mathrm{OLS}} - \beta^{0})^{T} - \mathbb{E}(\hat{\beta} -\beta^{0})(\hat{\beta} -\beta^{0})^{T}$ is equal to + +$$ +\sigma^ {2} (X ^ {T} X) ^ {- 1} - \sigma^ {2} (X ^ {T} X + \lambda I) ^ {- 1} X ^ {T} X (X ^ {T} X + \lambda I) ^ {- 1} - \lambda^ {2} (X ^ {T} X + \lambda I) ^ {- 1} \beta^ {0} \beta^ {0 ^ {T}} (X ^ {T} X + \lambda I) ^ {- 1}. +$$ + +After some simplification, we see that this is equal to + +$$ +\lambda (X ^ {T} X + \lambda I) ^ {- 1} [ \sigma^ {2} \{2 I + \lambda (X ^ {T} X) ^ {- 1} \} - \lambda \beta^ {0} \beta^ {0 ^ {T}} ] (X ^ {T} X + \lambda I) ^ {- 1}. +$$ + +Thus $\mathbb{E}(\hat{\beta}^{\mathrm{OLS}} - \beta^{0})(\hat{\beta}^{\mathrm{OLS}} - \beta^{0})^{T} - \mathbb{E}(\hat{\beta} -\beta^{0})(\hat{\beta} -\beta^{0})^{T}$ is positive definite for $\lambda >0$ if and only if + +$$ +\sigma^ {2} \{2 I + \lambda (X ^ {T} X) ^ {- 1} \} - \lambda \beta^ {0} \beta^ {0 T} +$$ + +is positive definite, which is true for $\lambda > 0$ sufficiently small (we can take $0 < \lambda < 2\sigma^2 / \| \beta^0 \|_2^2$ ). + +The theorem says that $\hat{\beta}_{\lambda}^{\mathrm{R}}$ outperforms $\hat{\beta}^{\mathrm{OLS}}$ provided $\lambda$ is chosen appropriately. To be able to use ridge regression effectively, we need a way of selecting a good $\lambda$ —we will come to this very shortly. What the theorem doesn't really tell us is in what situations we expect ridge regression to perform well. To understand that, we will turn to one of the key matrix decompositions used in statistics, the singular value decomposition (SVD). + +## 1.1.1 The singular value decomposition and principal components analysis + +The singular value decomposition (SVD) is a generalisation of an eigendecomposition of a square matrix. We can factorise any $X \in R^{n \times p}$ into its SVD + +$$ +X = U D V ^ {T}. +$$ + +Here the $U \in R^{n \times n}$ and $V \in R^{p \times p}$ are orthogonal matrices and $D \in R^{n \times p}$ has $D_{11} \geq D_{22} \geq \cdots \geq D_{mm} \geq 0$ where $m := \min(n, p)$ and all other entries of D are zero. To compute such a decomposition requires $O(np \min(n, p))$ operations. The rth columns of U and V are known as the rth left and right singular vectors of X respectively, and $D_{rr}$ is the rth singular value. + +When n > p, we can replace U by its first p columns and D by its first p rows to produce another version of the SVD (sometimes known as the thin SVD). Then $X = UDV^{T}$ where $U \in R^{n \times p}$ has orthonormal columns (but is no longer square) and D is square and diagonal. There is an equivalent version for when p > n. + +Let us take $X \in \mathbb{R}^{n \times p}$ as our matrix of predictors and suppose $n \geq p$ . Using the (thin) SVD we may write the fitted values from ridge regression as follows. + +$$ +\begin{array}{l} X \hat {\beta} _ {\lambda} ^ {R} = X (X ^ {T} X + \lambda I) ^ {- 1} X ^ {T} Y \\ = U D V ^ {T} \left(V D ^ {2} V ^ {T} + \lambda I\right) ^ {- 1} V D U ^ {T} Y \\ = U D \left(D ^ {2} + \lambda I\right) ^ {- 1} D U ^ {T} Y \\ = \sum_ {j = 1} ^ {p} U _ {j} \frac {D _ {j j} ^ {2}}{D _ {j j} ^ {2} + \lambda} U _ {j} ^ {T} Y. \\ \end{array} +$$ + +Here we have used the notation (that we shall use throughout the course) that $U_{j}$ is the jth column of U. For comparison, the fitted values from OLS (when X has full column rank) are + +$$ +X \hat {\beta} ^ {\mathrm{OLS}} = X (X ^ {T} X) ^ {- 1} X ^ {T} Y = U U ^ {T} Y. +$$ + +Both OLS and ridge regression compute the coordinates of Y with respect to the columns of U. Ridge regression then shrinks these coordinates by the factors $D_{jj}^{2}/(D_{jj}^{2}+\lambda)$ ; if $D_{jj}$ is small, the amount of shrinkage will be larger. + +To interpret this further, note that the SVD is intimately connected with Principal Components Analysis (PCA). Consider $v \in R^{p}$ with $\|v\|_{2} = 1$ . Since the columns of X have had their means subtracted, the sample variance of $Xv \in R^{n}$ , is + +$$ +\frac {1}{n} v ^ {T} X ^ {T} X v = \frac {1}{n} v ^ {T} V D ^ {2} V ^ {T} v. +$$ + +Writing $a = V^{T}v$ , so $\| a\| _2 = 1$ , we have + +$$ +\frac {1}{n} v ^ {T} V D ^ {2} V ^ {T} v = \frac {1}{n} a ^ {T} D ^ {2} a = \frac {1}{n} \sum_ {j} a _ {j} ^ {2} D _ {j j} ^ {2} \leq \frac {1}{n} D _ {1 1} \sum_ {j} a _ {j} ^ {2} = \frac {1}{n} D _ {1 1} ^ {2}. +$$ + +As $\|XV_{1}\|_{2}^{2}/n = D_{11}^{2}/n$ , $V_{1}$ determines the linear combination of the columns of X which has the largest sample variance, when the coefficients of the linear combination are constrained to have $\ell_{2}$ -norm 1. $XV_{1} = D_{11}U_{1}$ is known as the first principal component of X. Subsequent principal components $D_{22}U_{2}, \ldots, D_{pp}U_{p}$ have maximum variance $D_{jj}^{2}/n$ , subject to being orthogonal to all earlier ones—see example sheet 1 for details. + +Returning to ridge regression, we see that it shrinks Y most in the smaller principal components of X. Thus it will work well when most of the signal is in the large principal components of X. We now turn to the problem of choosing $\lambda$ . + +## 1.2 v-fold cross-validation + +Cross-validation is a general technique for selecting a good regression method from among several competing regression methods. We illustrate the principle with ridge regression, where we have a family of regression methods given by different $\lambda$ values. + +So far, we have considered the matrix of predictors X as fixed and non-random. However, in many cases, it makes sense to think of it as random. Let us assume that our data are i.i.d. pairs $(x_{i}, Y_{i})$ , $i = 1, \ldots, n$ . Then ideally, we might want to pick a $\lambda$ value such that + +$$ +\mathbb {E} \{(Y ^ {*} - x ^ {* T} \hat {\beta} _ {\lambda} ^ {\mathrm{R}} (X, Y)) ^ {2} | X, Y \} \tag {1.2.1} +$$ + +is minimised. Here $(x^{*}, Y^{*}) \in \mathbb{R}^{p} \times \mathbb{R}$ is independent of $(X, Y)$ and has the same distribution as $(x_{1}, Y_{1})$ , and we have made the dependence of $\hat{\beta}_{\lambda}^{R}$ on the training data $(X, Y)$ explicit. This $\lambda$ is such that conditional on the original training data, it minimises the expected prediction error on a new observation drawn from the same distribution as the training data. + +A less ambitious goal is to find a $\lambda$ value to minimise the expected prediction error, + +$$ +\mathbb {E} [ \mathbb {E} \{(Y ^ {*} - x ^ {* T} \hat {\beta} _ {\lambda} ^ {\mathrm{R}} (X, Y)) ^ {2} | X, Y \} ] \tag {1.2.2} +$$ + +where compared with $(1.2.1)$ , we have taken a further expectation over the training set. + +We still have no way of computing (1.2.2) directly, but we can attempt to estimate it. The idea of v-fold cross-validation is to split the data into v groups or folds of roughly equal size: $(X^{(1)}, Y^{(1)}), \ldots, (X^{(v)}, Y^{(v)})$ . Let $(X^{(-k)}, Y^{(-k)})$ be all the data except that in the kth fold. For each $\lambda$ on a grid of values, we compute $\hat{\beta}_{\lambda}^{\mathrm{R}}(X^{(-k)}, Y^{(-k)})$ : the ridge regression estimate based on all the data except the kth fold. Writing $\kappa(i)$ for the fold to which $(x_{i}, Y_{i})$ belongs, we choose the value of $\lambda$ that minimises + +$$ +\mathrm{CV} (\lambda) = \frac {1}{n} \sum_ {i = 1} ^ {n} \left\{Y _ {i} - x _ {i} ^ {T} \hat {\beta} _ {\lambda} ^ {\mathrm{R}} \left(X ^ {(- \kappa (i))}, Y ^ {(- \kappa (i))}\right) \right\} ^ {2}. \tag {1.2.3} +$$ + +Writing $\lambda_{\mathrm{CV}}$ for the minimiser, our final estimate of $\beta^0$ can then be $\hat{\beta}_{\lambda_{\mathrm{CV}}}^{R}(X,Y)$ . + +Note that for each i, + +$$ +\mathbb {E} \left\{Y _ {i} - x _ {i} ^ {T} \hat {\beta} _ {\lambda} ^ {\mathrm{R}} \left(X ^ {(- \kappa (i))}, Y ^ {(- \kappa (i))}\right) \right\} ^ {2} = \mathbb {E} \left[ \mathbb {E} \left\{Y _ {i} - x _ {i} ^ {T} \hat {\beta} _ {\lambda} ^ {\mathrm{R}} \left(X ^ {(- \kappa (i))}, Y ^ {(- \kappa (i))}\right) \right\} ^ {2} \mid X ^ {(- \kappa (i))}, Y ^ {(- \kappa (i))} \right]. \tag {1.2.4} +$$ + +This is precisely the expected prediction error in $(1.2.2)$ but with the training data X, Y replaced with a training data set of smaller size. If all the folds have the same size, then $\mathrm{CV}(\lambda)$ is an average of n identically distributed quantities, each with expected value as in $(1.2.4)$ . However, the quantities being averaged are not independent as they share the same data. + +Thus cross-validation gives a biased estimate of the expected prediction error. The amount of the bias depends on the size of the folds, the case when the v = n giving the least bias—this is known as leave-one-out cross-validation. The quality of the estimate, though, may be worse as the quantities being averaged in $(1.2.3)$ will be highly positively correlated. Typical choices of v are 5 or 10. + +Cross-validation aims to allow us to choose the single best $\lambda$ (or more generally regression procedure); we could instead aim to find the best weighted combination of regression procedures. Returning to our ridge regression example, suppose $\lambda$ is restricted to a grid of values $\lambda_{1} > \lambda_{2} > \cdots > \lambda_{L}$ . We can then minimise + +$$ +\frac {1}{n} \sum_ {i = 1} ^ {n} \left\{Y _ {i} - \sum_ {l = 1} ^ {L} w _ {l} x _ {i} ^ {T} \hat {\beta} _ {\lambda_ {l}} ^ {\mathrm{R}} (X ^ {(- \kappa (i))}, Y ^ {(- \kappa (i))}) \right\} ^ {2} +$$ + +over $w \in R^{L}$ subject to $w_{l} \geq 0$ for all l. This is a non-negative least-squares optimisation, for which efficient algorithms are available. This is sometimes known as stacking [Wolpert, 1992, Breiman, 1996] and it can often outperform cross-validation. + +## 1.3 The kernel trick + +The fitted values from ridge regression are + +$$ +X (X ^ {T} X + \lambda I) ^ {- 1} X ^ {T} Y. \tag {1.3.1} +$$ + +An alternative way of writing this is suggested by the following + +$$ +X ^ {T} \left(X X ^ {T} + \lambda I\right) = \left(X ^ {T} X + \lambda I\right) X ^ {T} +$$ + +$$ +(X ^ {T} X + \lambda I) ^ {- 1} X ^ {T} = X ^ {T} (X X ^ {T} + \lambda I) ^ {- 1} +$$ + +$$ +X (X ^ {T} X + \lambda I) ^ {- 1} X ^ {T} Y = X X ^ {T} (X X ^ {T} + \lambda I) ^ {- 1} Y. \tag {1.3.2} +$$ + +Two remarks are in order: + +- Note while $X^T X$ is $p \times p$ , $XX^T$ is $n \times n$ . Computing fitted values using (1.3.1) would require roughly $O(np^2 + p^3)$ operations. If $p \gg n$ this could be extremely costly. However, our alternative formulation would only require roughly $O(n^2 p + n^3)$ operations, which could be substantially smaller. +- We see that the fitted values of ridge regression depend only on inner products $K = XX^{T}$ between observations (note $K_{ij} = x_{i}^{T}x_{j}$ ). + +Now suppose that we believe the signal depends quadratically on the predictors: + +$$ +Y _ {i} = x _ {i} ^ {T} \beta + \sum_ {k, l} x _ {i k} x _ {i l} \theta_ {k l} + \varepsilon_ {i}. +$$ + +We can still use ridge regression provided we work with an enlarged set of predictors + +$$ +x _ {i 1}, \ldots , x _ {i p}, x _ {i 1} x _ {i 1}, \ldots , x _ {i 1} x _ {i p}, x _ {i 2} x _ {i 1}, \ldots , x _ {i 2} x _ {i p}, \ldots , x _ {i p} x _ {i p}. +$$ + +This will give us $O(p^{2})$ predictors. Our new approach to computing fitted values would therefore have complexity $O(n^{2}p^{2} + n^{3})$ , which could be rather costly if p is large. + +However, rather than first creating all the additional predictors and then computing the new K matrix, we can attempt to directly compute K. To this end consider + +$$ +\begin{array}{l} (1 + x _ {i} ^ {T} x _ {j}) ^ {2} = \left(1 + \sum_ {k} x _ {i k} x _ {j k}\right) ^ {2} \\ = 1 + 2 \sum_ {k} x _ {i k} x _ {j k} + \sum_ {k, l} x _ {i k} x _ {i l} x _ {j k} x _ {j l}. \\ \end{array} +$$ + +Observe this amounts to an inner product between vectors of the form + +$$ +(1, \sqrt {2} x _ {i 1}, \dots , \sqrt {2} x _ {i p}, x _ {i 1} x _ {i 1}, \dots , x _ {i 1} x _ {i p}, x _ {i 2} x _ {i 1}, \dots , x _ {i 2} x _ {i p}, \dots , x _ {i p} x _ {i p}) ^ {T}. \tag {1.3.3} +$$ + +Thus if we set + +$$ +K _ {i j} = (1 + x _ {i} ^ {T} x _ {j}) ^ {2} \tag {1.3.4} +$$ + +and plug this into the formula for the fitted values, it is exactly as if we had performed ridge regression on an enlarged set of variables given by (1.3.3). Now computing K using (1.3.4) would require only p operations per entry, so $O(n^{2}p)$ operations in total. It thus seems we have improved things by a factor of p using our new approach. This is a nice computational trick, but more importantly for us it serves to illustrate some general points. + +- Since ridge regression only depends on inner products between observations, rather than fitting non-linear models by first mapping the original data $x_{i} \in \mathbb{R}^{p}$ to $\phi(x_{i}) \in \mathbb{R}^{d}$ (say) using some feature map $\phi$ (which could, for example introduce quadratic effects), we can instead try to directly compute $k(x_{i}, x_{j}) = \langle \phi(x_{i}), \phi(x_{j}) \rangle$ . +- In fact rather than thinking in terms of feature maps, we can instead try to think about an appropriate measure of similarity $k(x_{i}, x_{j})$ between observations. Modelling in this fashion is sometimes much easier. + +We will now formalise and extend what we have learnt with this example. + +## 1.4 Kernels + +We have seen how a model with quadratic effects can be fitted very efficiently by replacing the inner product matrix (known as the Gram matrix) $XX^{T}$ in (1.3.2) with the matrix in (1.3.4). It is then natural to ask what other non-linear models can be fitted efficiently using this sort of approach. + +We won't answer this question directly, but instead we will try to understand the sorts of similarity measures $k$ that can be represented as inner products between transformations of the original data. + +That is, we will study the similarity measures $k : X \times X \to R$ from the input space X to R for which there exists a feature map $\phi : X \to H$ where H is some (real) inner product space with + +$$ +k (x, x ^ {\prime}) = \langle \phi (x), \phi (x ^ {\prime}) \rangle . \tag {1.4.1} +$$ + +Recall that an inner product space is a real vector space H endowed with a map $\langle\cdot,\cdot\rangle:H\times H\to R$ that obeys the following properties. + +(i) Symmetry: $\langle u, v \rangle = \langle v, u \rangle$ . +(ii) Linearity: for $a, b \in R \langle au + bw, v \rangle = a \langle u, v \rangle + b \langle w, v \rangle$ . +(iii) Positive-definiteness: $\langle u, u \rangle \geq 0$ with equality if and only if $u = 0$ . + +Definition 1. A positive definite kernel or more simply a kernel (for brevity) k is a symmetric map $k : X \times X \to R$ for which for all $n \in N$ and all $x_{1}, \ldots, x_{n} \in X$ , the matrix K with entries + +$$ +K _ {i j} = k (x _ {i}, x _ {j}) +$$ + +is positive semi-definite. + +A kernel is a little like an inner product, but need not be bilinear in general. However, a form of the Cauchy–Schwarz inequality does hold for kernels. + +## Proposition 2. + +$$ +k (x, x ^ {\prime}) ^ {2} \leq k (x, x) k (x ^ {\prime}, x ^ {\prime}). +$$ + +Proof. The matrix + +$$ +\left( \begin{array}{c c} k (x, x) & k (x, x ^ {\prime}) \\ k (x ^ {\prime}, x) & k (x ^ {\prime}, x ^ {\prime}) \end{array} \right) +$$ + +must be positive semi-definite so in particular its determinant must be non-negative. □ + +First we show that any inner product of feature maps will give rise to a kernel. + +Proposition 3. k defined by $k(x, x') = \langle \phi(x), \phi(x') \rangle$ is a kernel. + +Proof. Let $x_{1},\ldots ,x_{n}\in \mathcal{X}$ , $\alpha_{1},\ldots ,\alpha_{n}\in \mathbb{R}$ and consider + +$$ +\begin{array}{l} \sum_ {i, j} \alpha_ {i} k (x _ {i}, x _ {j}) \alpha_ {j} = \sum_ {i, j} \alpha_ {i} \langle \phi (x _ {i}), \phi (x _ {j}) \rangle \alpha_ {j} \\ = \left\langle \sum_ {i} \alpha_ {i} \phi (x _ {i}), \sum_ {j} \alpha_ {j} \phi (x _ {j}) \right\rangle \geq 0. \\ \end{array} +$$ + +Showing that every kernel admits a representation of the form (1.4.1) is slightly more involved, and we delay this until after we have studied some examples. + +## 1.4.1 Examples of kernels + +Proposition 4. Suppose $k_{1}, k_{2}, \ldots$ are kernels. + +(i) If $\alpha_{1},\alpha_{2}\geq 0$ then $\alpha_{1}k_{1} + \alpha_{2}k_{2}$ is a kernel. If $\lim_{m\to \infty}k_m(x,x') =:k(x,x')$ exists for all $x,x^{\prime}\in \mathcal{X}$ , then $k$ is a kernel. + +(ii) The pointwise product $k = k_{1}k_{2}$ is a kernel. + +Linear kernel. $k(x,x^{\prime}) = x^{T}x^{\prime}$ + +Polynomial kernel. $k(x, x') = (1 + x^{T}x')^{d}$ . To show this is a kernel, we can simply note that $1 + x^{T}x'$ gives a kernel owing to the fact that 1 is a kernel and (i) of Proposition 4. Next (ii) and induction shows that k as defined above is a kernel. + +Gaussian kernel. The highly popular Gaussian kernel is defined by + +$$ +k (x, x ^ {\prime}) = \exp \bigg (- \frac {\| x - x ^ {\prime} \| _ {2} ^ {2}}{2 \sigma^ {2}} \bigg). +$$ + +For x close to $x'$ it is large whilst for x far from $x'$ the kernel quickly decays towards 0. The additional parameter $\sigma^{2}$ known as the bandwidth controls the speed of the decay to zero. Note it is less clear how one might find a corresponding feature map and indeed any feature map that represents this must be infinite dimensional. + +To show that it is a kernel first decompose $\|x-x'\|_{2}^{2}=\|x\|_{2}^{2}+\|x'\|_{2}^{2}-2x^{T}x'$ . Note that by Proposition 3, + +$$ +k _ {1} (x, x ^ {\prime}) = \exp \left(- \frac {\| x \| _ {2} ^ {2}}{2 \sigma^ {2}}\right) \exp \left(- \frac {\| x ^ {\prime} \| _ {2} ^ {2}}{2 \sigma^ {2}}\right) +$$ + +is a kernel. Next writing + +$$ +k _ {2} (x, x ^ {\prime}) = \exp (x ^ {T} x ^ {\prime} / \sigma^ {2}) = \sum_ {r = 0} ^ {\infty} \frac {(x ^ {T} x ^ {\prime} / \sigma^ {2}) ^ {r}}{r !} +$$ + +and using (i) of Proposition 4 shows that $k_{2}$ is a kernel. Finally observing that $k = k_{1}k_{2}$ and using (ii) shows that the Gaussian kernel is indeed a kernel. + +Sobolev kernel. Take X to be $[0,1]$ and let $k(x,x') = \min(x,x')$ . Note this is the covariance function of Brownian motion so it must be positive definite. + +Jaccard similarity kernel. Take X to be the set of all subsets of $\{1,\ldots,p\}$ . For $x, x' \in X$ with $x \cup x' \neq \emptyset$ define + +$$ +k (x, x ^ {\prime}) = \frac {| x \cap x ^ {\prime} |}{| x \cup x ^ {\prime} |} +$$ + +and if $x \cup x' = \emptyset$ then set $k(x, x') = 1$ . Showing that this is a kernel is left to the example sheet. + +## 1.4.2 Reproducing kernel Hilbert spaces + +Theorem 5. For every kernel k there exists a feature map $\phi$ taking values in some inner product space H such that + +$$ +k (x, x ^ {\prime}) = \langle \phi (x), \phi (x ^ {\prime}) \rangle . \tag {1.4.2} +$$ + +Proof. We will take H to be the vector space of functions of the form + +$$ +f (\cdot) = \sum_ {i = 1} ^ {n} \alpha_ {i} k (\cdot , x _ {i}), \tag {1.4.3} +$$ + +where $n \in N$ , $x_{i} \in X$ and $\alpha_{i} \in R$ . Our feature map $\phi : X \to H$ will be + +$$ +\phi (x) = k (\cdot , x). \tag {1.4.4} +$$ + +We now define an inner product on $\mathcal{H}$ . If $f$ is given by (1.4.3) and + +$$ +g (\cdot) = \sum_ {j = 1} ^ {m} \beta_ {j} k (\cdot , x _ {j} ^ {\prime}) \tag {1.4.5} +$$ + +we define their inner product to be + +$$ +\langle f, g \rangle = \sum_ {i = 1} ^ {n} \sum_ {j = 1} ^ {m} \alpha_ {i} \beta_ {j} k (x _ {i}, x _ {j} ^ {\prime}). \tag {1.4.6} +$$ + +We need to check this is well-defined as the representations of $f$ and $g$ in (1.4.3) and (1.4.5) need not be unique. To this end, note that + +$$ +\sum_ {i = 1} ^ {n} \sum_ {j = 1} ^ {m} \alpha_ {i} \beta_ {j} k (x _ {i}, x _ {j} ^ {\prime}) = \sum_ {i = 1} ^ {n} \alpha_ {i} g (x _ {i}) = \sum_ {j = 1} ^ {m} \beta_ {j} f (x _ {j} ^ {\prime}). \tag {1.4.7} +$$ + +The first equality shows that the inner product does not depend on the particular expansion of g whilst the second equality shows that it also does not depend on the expansion of f. Thus the inner product is well-defined. + +First we check that with $\phi$ defined as in (1.4.4) we do have relationship (1.4.2). Observe that + +$$ +\langle k (\cdot , x), f \rangle = \sum_ {i = 1} ^ {n} \alpha_ {i} k (x _ {i}, x) = f (x), \tag {1.4.8} +$$ + +so in particular we have + +$$ +\langle \phi (x), \phi (x ^ {\prime}) \rangle = \langle k (\cdot , x), k (\cdot , x ^ {\prime}) \rangle = k (x, x ^ {\prime}). +$$ + +It remains to show that it is indeed an inner product. It is clearly symmetric and $(1.4.7)$ shows linearity. We now need to show positive definiteness. + +First note that + +$$ +\langle f, f \rangle = \sum_ {i, j} \alpha_ {i} k (x _ {i}, x _ {j}) \alpha_ {j} \geq 0 \tag {1.4.9} +$$ + +by positive definiteness of the kernel. Now from (1.4.8), + +$$ +f (x) ^ {2} = (\langle k (\cdot , x), f \rangle) ^ {2}. +$$ + +If we could use the Cauchy–Schwarz inequality on the right-hand side, we would have + +$$ +f (x) ^ {2} \leq \langle k (\cdot , x), k (\cdot , x) \rangle \langle f, f \rangle , \tag {1.4.10} +$$ + +which would show that if $\langle f, f \rangle = 0$ then necessarily $f = 0$ ; the final property we need to show that $\langle \cdot, \cdot \rangle$ is an inner product. However, in order to use the traditional Cauchy-Schwarz inequality we need to first know we're dealing with an inner product, which is precisely what we're trying to show! + +Although we haven't shown that $\langle \cdot, \cdot \rangle$ is an inner product, we do have enough information to show that it is itself a kernel. We may then appeal to Proposition 2 to obtain (1.4.10). With this in mind, we argue as follows. Given functions $f_1, \ldots, f_m$ and coefficients $\gamma_1, \ldots, \gamma_m \in \mathbb{R}$ , we have + +$$ +\sum_ {i, j} \gamma_ {i} \langle f _ {i}, f _ {j} \rangle \gamma_ {j} = \left\langle \sum_ {i} \gamma_ {i} f _ {i}, \sum_ {j} \gamma_ {j} f _ {j} \right\rangle \geq 0 +$$ + +where we have used linearity and (1.4.9), showing that it is a kernel. + +To further discuss the space H we recall some facts from analysis. Any inner product space B is also a normed space: for $f \in B$ we may define $\|f\|_{B}^{2} := \langle f, f \rangle_{B}$ . Recall that a Cauchy sequence $(f_{m})_{m=1}^{\infty}$ in B has $\|f_{m} - f_{n}\|_{B} \to 0$ as $n, m \to \infty$ . A normed space where every Cauchy sequence has a limit (in the space) is called complete, and a complete inner product space is called a Hilbert space. + +Hilbert spaces may be thought of as the (potentially) infinite-dimensional analogues of finite-dimensional Euclidean spaces. For later use we note that if V is a closed subspace of a Hilbert space B, then any $f \in B$ has a decomposition $f = u + v$ with $u \in V$ and + +$$ +v \in V ^ {\perp} := \{v \in \mathcal {B}: \langle v, u \rangle_ {\mathcal {B}} = 0 \text {for all} u \in V \}. +$$ + +By adding the limits of Cauchy sequences to $\mathcal{H}$ (from Theorem 5) we can make $\mathcal{H}$ a Hilbert space. Indeed, note that if $(f_m)_{m=1}^{\infty} \in \mathcal{H}$ is Cauchy then since by (1.4.10) we have + +$$ +| f _ {m} (x) - f _ {n} (x) | \leq \sqrt {k (x , x)} \| f _ {m} - f _ {n} \| _ {\mathcal {H}}, +$$ + +we may define function $f^{*}: X \to R$ by $f^{*}(x) = \lim_{m \to \infty} f_{m}(x)$ . We can check that all such $f^{*}$ can be added to H to create a Hilbert space. + +In fact, the completion of H is a special type of Hilbert space known as a reproducing kernel Hilbert space (RKHS). + +Definition 2. A Hilbert space $\mathcal{B}$ of functions $f:\mathcal{X}\to \mathbb{R}$ is a reproducing kernel Hilbert space (RKHS) if for all $x\in \mathcal{X}$ , there exists $k_{x}\in \mathcal{B}$ such that + +$$ +f (x) = \left\langle k _ {x}, f \right\rangle \quad \text { for all } f \in \mathcal {B}. +$$ + +The function + +$$ +k: \mathcal {X} \times \mathcal {X} \to \mathbb {R} +$$ + +$$ +(x, x ^ {\prime}) \mapsto \langle k _ {x}, k _ {x ^ {\prime}} \rangle = k _ {x ^ {\prime}} (x) +$$ + +is known as the reproducing kernel of $\mathcal{B}$ . + +By Proposition 3 the reproducing kernel of any RKHS is a (positive definite) kernel, and one may show (as Theorem 5 indicates) that to any kernel k is associated a (unique) RKHS that has reproducing kernel k. + +## Examples + +Linear kernel. Here $\mathcal{H}=\{f:f(x)=\beta^{T}x,\beta\in\mathbb{R}^{p}\}$ and if $f(x)=\beta^{T}x$ then $\|f\|_{H}^{2}=||\beta||_{2}^{2}$ . + +Sobolev kernel. It can be shown that $\mathcal{H}$ is roughly the space of continuous functions $f:[0,1]\to \mathbb{R}$ with $f(0) = 0$ that are differentiable almost everywhere, and for which $\int_0^1 f'(x)^2 dx < \infty$ . It contains the class of Lipschitz functions (functions $f:[0,1]\to \mathbb{R}$ for which there exists some $L$ with $|f(x) - f(y)|\leq L|x - y|$ for all $x,y\in [0,1]$ ) that are 0 at the origin. The norm is + +$$ +\left(\int_ {0} ^ {1} f ^ {\prime} (x) ^ {2} d x\right) ^ {1 / 2}. +$$ + +Though the construction of the RKHS from a kernel is explicit, it can be challenging to understand precisely the space and the form of the norm. + +## 1.4.3 The representer theorem + +To recap, what we have shown so far is that replacing the matrix $XX^{T}$ in the definition of an algorithm by K derived from a positive definite kernel is essentially equivalent to running the same algorithm on some mapping of the original data, though with the modification that instances of $x_{i}^{T}x_{j}$ become $\langle\phi(x_{i}),\phi(x_{j})\rangle$ . + +But what exactly is the optimisation problem we are solving when performing kernel ridge regression? Clearly it is determined by the kernel or equivalently by the RKHS. Note we know that an alternative way of writing the usual ridge regression optimisation is + +$$ +\underset {f \in \mathcal {H}} {\arg \min} \left\{\sum_ {i = 1} ^ {n} \left\{Y _ {i} - f \left(x _ {i}\right) \right\} ^ {2} + \lambda \| f \| _ {\mathcal {H}} ^ {2} \right\} \tag {1.4.11} +$$ + +where H is the RKHS corresponding to the linear kernel. The following theorem shows in particular that kernel ridge regression (i.e. ridge regression replacing $XX^{T}$ with K) with kernel k is equivalent to the above with H now being the RKHS corresponding to k. + +Theorem 6 (Representer theorem, [Kimeldorf and Wahba, 1970, Schölkopf et al., 2001]). Let $c: \mathbb{R}^n \times \mathcal{X}^n \times \mathbb{R}^n \to \mathbb{R}$ be an arbitrary loss function, and let $J: [0, \infty) \to \mathbb{R}$ be strictly increasing. Let $x_1, \ldots, x_n \in \mathcal{X}$ , $Y \in \mathbb{R}^n$ . Finally, let $f \in \mathcal{H}$ where $\mathcal{H}$ is an RKHS with reproducing kernel $k$ , and let $K_{ij} = k(x_i, x_j)$ $i, j = 1, \ldots, n$ . Then $\hat{f}$ minimises + +$$ +Q _ {1} (f) := c \left(Y, x _ {1}, \dots , x _ {n}, f \left(x _ {1}\right), \dots , f \left(x _ {n}\right)\right) + J \left(\| f \| _ {\mathcal {H}} ^ {2}\right) +$$ + +over $f\in \mathcal{H}$ iff. $\hat{f} (\cdot) = \sum_{i = 1}^{n}\hat{\alpha}_{i}k(\cdot ,x_{i})$ and $\hat{\alpha}\in \mathbb{R}^n$ minimises $Q_{2}$ over $\alpha \in \mathbb{R}^n$ where + +$$ +Q _ {2} (\alpha) = c (Y, x _ {1}, \ldots , x _ {n}, K \alpha) + J (\alpha^ {T} K \alpha). +$$ + +Proof. Suppose $\hat{f}$ minimises $Q_{1}$ . We may write $\hat{f} = u + v$ where $u\in V:= \operatorname {span}\{k(\cdot ,x_1),\ldots ,k(\cdot ,x_n)\}$ and $v\in V^{\perp}$ . Then + +$$ +\hat {f} (x _ {i}) = \langle k (\cdot , x _ {i}), u + v \rangle = \langle k (\cdot , x _ {i}), u \rangle = u (x _ {i}). +$$ + +Meanwhile, by Pythagoras' theorem we have $J(\| \hat{f}\|_{\mathcal{H}}^2) = J(\| u\|_{\mathcal{H}}^2 + \| v\|_{\mathcal{H}}^2) \geq J(\| u\|_{\mathcal{H}}^2)$ with equality iff. $v = 0$ . Thus by optimality of $\hat{f}$ , $v = 0$ , so $\hat{f}(\cdot) = \sum_{i=1}^{n} \alpha_i k(\cdot, x_i)$ for $\alpha \in \mathbb{R}^n$ . Now observe that if $\hat{f}$ takes this form, then $\| \hat{f}\|_{\mathcal{H}}^2 = \alpha^T K\alpha$ , so $Q_1(\hat{f}) = Q_2(\alpha)$ . Then by optimality of $\hat{f}$ , we have that $\alpha$ must minimise $Q_2$ . + +Now suppose $\hat{\alpha}$ minimises $Q_{2}$ and $\hat{f} (\cdot) = \sum_{i = 1}^{n}\hat{\alpha}_{i}k(\cdot ,x_{i})$ . Note that $Q_{1}(\hat{f}) = Q_{2}(\hat{\alpha})$ . If $\tilde{f}\in \mathcal{H}$ has $Q_{1}(\tilde{f})\leq Q_{1}(\hat{f})$ , by the argument above, writing $\tilde{f} = u + v$ with $u\in V$ , $v\in V^{\perp}$ , we know that $Q_{1}(u)\leq Q_{1}(\tilde{f})$ . But by optimality of $\hat{\alpha}$ we have $Q_{1}(\hat{f})\leq Q_{1}(u)$ , so $Q_{1}(\hat{f}) = Q_{1}(\tilde{f})$ . + +Consider the result specialised the ridge regression objective. We see that $(1.4.11)$ is essentially equivalent to minimising + +$$ +\| Y - K \alpha \| _ {2} ^ {2} + \lambda \alpha^ {T} K \alpha , +$$ + +and you may check (see example sheet 1) that the minimiser $\hat{\alpha}$ satisfies $K\hat{\alpha} = K(K + \lambda I)^{-1}Y$ . Thus (1.4.11) is indeed an alternative way of expressing kernel ridge regression. + +Viewing the result in the opposite direction gives a more “sensational” perspective. If you had set out trying to minimise $Q_{1}$ , it might appear completely hopeless as H could be infinite-dimensional. However, somewhat remarkably we see that this reduces to finding the coefficients $\hat{\alpha}_{i}$ which solve the simple(r) optimisation problem $Q_{2}$ . + +The result also tells us how to form predictions: given a new observation x, our prediction for $f(x)$ is + +$$ +\hat {f} (x) = \sum_ {i = 1} ^ {n} \hat {\alpha} _ {i} k (x, x _ {i}). +$$ + +## 1.5 Kernel ridge regression + +We have seen how the kernel trick allows us to solve a potentially infinite-dimensional version of ridge regression. This may seem impressive, but ultimately we should judge kernel ridge regression on its statistical properties e.g. predictive performance. Consider a setting where + +$$ +Y _ {i} = f ^ {0} (x _ {i}) + \varepsilon_ {i}, \qquad \mathbb {E} (\varepsilon) = 0, \mathrm{Var} (\varepsilon) = \sigma^ {2} I. +$$ + +We shall assume that $f^{0} \in H$ where H is an RKHS with reproducing kernel k. By scaling $\sigma^{2}$ , we may assume $\|f^{0}\|_{H} \leq 1$ . Let K be the kernel matrix $K_{ij} = k(x_{i}, x_{j})$ with eigenvalues $d_{1} \geq d_{2} \geq \cdots \geq d_{n} \geq 0$ . We will see that the predictive performance depends delicately on these eigenvalues. + +Let $\hat{f}_{\lambda}$ be the estimated regression function from kernel ridge regression with kernel k: + +$$ +\hat {f} _ {\lambda} = \underset {f \in \mathcal {H}} {\arg \min} \left\{\sum_ {i = 1} ^ {n} \{Y _ {i} - f (x _ {i}) \} ^ {2} + \lambda \| f \| _ {\mathcal {H}} ^ {2} \right\}. +$$ + +Theorem 7. The mean squared prediction error (MSPE) may be bounded above in the following way: + +$$ +\frac {1}{n} \mathbb {E} \left\{\sum_ {i = 1} ^ {n} \left\{f ^ {0} (x _ {i}) - \hat {f} _ {\lambda} (x _ {i}) \right\} ^ {2} \right\} \leq \frac {\sigma^ {2}}{n} \sum_ {i = 1} ^ {n} \frac {d _ {i} ^ {2}}{(d _ {i} + \lambda) ^ {2}} + \frac {\lambda}{4 n} \tag {1.5.1} +$$ + +$$ +\leq \frac {\sigma^ {2}}{n} \frac {1}{\lambda} \sum_ {i = 1} ^ {n} \min (d _ {i} / 4, \lambda) + \frac {\lambda}{4 n}. +$$ + +Proof. We know from the representer theorem that + +$$ +\left(\hat {f} _ {\lambda} (x _ {1}), \dots , \hat {f} _ {\lambda} (x _ {n})\right) ^ {T} = K (K + \lambda I) ^ {- 1} Y. +$$ + +You will show on the example sheet that + +$$ +\left(f ^ {0} (x _ {1}), \ldots , f ^ {0} (x _ {n})\right) ^ {T} = K \alpha , +$$ + +for some $\alpha \in \mathbb{R}^n$ , and moreover that $\| f^0\|_{\mathcal{H}}^2\geq \alpha^T K\alpha$ . Let the eigendecomposition of $K$ be given by $K = UDU^{T}$ with $D_{ii} = d_i$ and define $\theta = U^{T}K\alpha$ . We see that $n$ times the LHS of (1.5.1) is + +$$ +\begin{array}{l} \mathbb {E} \| K (K + \lambda I) ^ {- 1} (U \theta + \varepsilon) - U \theta \| _ {2} ^ {2} = \mathbb {E} \| D U ^ {T} (U D U ^ {T} + \lambda I) ^ {- 1} (U \theta + \varepsilon) - \theta \| _ {2} ^ {2} \\ = \mathbb {E} \| D (D + \lambda I) ^ {- 1} (\theta + U ^ {T} \varepsilon) - \theta \| _ {2} ^ {2} \\ = \| \{D (D + \lambda I) ^ {- 1} - I \} \theta \| _ {2} ^ {2} + \mathbb {E} \| D (D + \lambda I) ^ {- 1} U ^ {T} \varepsilon \| _ {2} ^ {2}. \\ \end{array} +$$ + +To compute the second term, we use the ‘trace trick’: + +$$ +\begin{array}{l} \mathbb {E} \| D (D + \lambda I) ^ {- 1} U ^ {T} \varepsilon \| _ {2} ^ {2} = \mathbb {E} [ \{D (D + \lambda I) ^ {- 1} U ^ {T} \varepsilon \} ^ {T} D (D + \lambda I) ^ {- 1} U ^ {T} \varepsilon ] \\ = \mathbb {E} [ \mathrm{tr} \{D (D + \lambda I) ^ {- 1} U ^ {T} \varepsilon \varepsilon^ {T} U D (D + \lambda I) ^ {- 1} \} ] \\ = \sigma^ {2} \mathrm{tr} \{D (D + \lambda I) ^ {- 1} D (D + \lambda I) ^ {- 1} \} \\ = \sigma^ {2} \sum_ {i = 1} ^ {n} \frac {d _ {i} ^ {2}}{(d _ {i} + \lambda) ^ {2}}. \\ \end{array} +$$ + +For the first term, we have + +$$ +\| \{D (D + \lambda I) ^ {- 1} - I \} \theta \| _ {2} ^ {2} = \sum_ {i = 1} ^ {n} \frac {\lambda^ {2} \theta_ {i} ^ {2}}{(d _ {i} + \lambda) ^ {2}}. +$$ + +Now as $\theta = DU^{T}\alpha$ note that $\theta_{i} = 0$ when $d_{i} = 0$ . Let $D^{+}$ be the diagonal matrix with $i$ th diagonal entry equal to $D_{ii}^{-1}$ if $D_{ii} > 0$ and 0 otherwise. Then + +$$ +\sum_ {i: d _ {i} > 0} \frac {\theta_ {i} ^ {2}}{d _ {i}} = \| \sqrt {D ^ {+}} \theta \| _ {2} ^ {2} = \alpha^ {T} K U D ^ {+} U ^ {T} K \alpha = \alpha^ {T} U D D ^ {+} D U ^ {T} \alpha = \alpha^ {T} K \alpha \leq 1. +$$ + +By Hölder's inequality we have + +$$ +\sum_ {i = 1} ^ {n} \frac {\theta_ {i} ^ {2}}{d _ {i}} \frac {d _ {i} \lambda^ {2}}{(d _ {i} + \lambda) ^ {2}} \leq \max _ {i = 1, \ldots , n} \frac {d _ {i} \lambda^ {2}}{(d _ {i} + \lambda) ^ {2}} \leq \lambda / 4, +$$ + +using the inequality $(a+b)^{2}\geq4ab$ in the final line. Finally note that + +$$ +\frac {d _ {i} ^ {2}}{(d _ {i} + \lambda) ^ {2}} \leq \min \{1, d _ {i} ^ {2} / (4 d _ {i} \lambda) \} = \min (\lambda , d _ {i} / 4) / \lambda . +$$ + +To interpret this result further, it will be helpful to express it in terms of $\hat{\mu}_{i} := d_{i}/n$ (the eigenvalues of K/n) and $\lambda_{n} := \lambda/n$ . We have + +$$ +\frac {1}{n} \mathbb {E} \bigg \{\sum_ {i = 1} ^ {n} \{f ^ {0} (x _ {i}) - \hat {f} _ {\lambda} (x _ {i}) \} ^ {2} \bigg \} \leq \frac {\sigma^ {2}}{\lambda_ {n}} \frac {1}{n} \sum_ {i = 1} ^ {n} \min (\hat {\mu} _ {i} / 4, \lambda_ {n}) + \lambda_ {n} / 4 =: \delta_ {n} (\lambda_ {n}). \tag {1.5.2} +$$ + +Here we have treated the $x_{i}$ as fixed, but we could equally well think of them as random. Consider a setup where the $x_{i}$ are i.i.d. and independent of $\varepsilon$ . If we take a further expectation on the RHS of (1.5.2), our result still holds true (the $\hat{\mu}_{i}$ are random in this setting). Ideally we would like to then replace $\mathbb{E}\min(\hat{\mu}_{i}/4,\lambda_{n})$ with a quantity more directly related to the kernel k. + +Mercer's theorem is helpful in this regard. This guarantees (under some mild conditions) an eigendecomposition for kernels, which are somewhat like infinite-dimensional analogues of symmetric positive semi-definite matrices. Under certain technical conditions, we may write + +$$ +k (x, x ^ {\prime}) = \sum_ {j = 1} ^ {\infty} \mu_ {j} e _ {j} (x) e _ {j} (x ^ {\prime}) +$$ + +where given some density $p(x)$ on X, the eigenfunctions $e_{j}$ and corresponding eigenvalues $\mu_{j}$ obey the integral equation + +$$ +\mu_ {j} e _ {j} (x ^ {\prime}) = \int_ {\mathcal {X}} k (x, x ^ {\prime}) e _ {j} (x) p (x) d x, +$$ + +and the $e_{j}$ form an orthonormal basis of H in the sense that + +$$ +\int_ {\mathcal {X}} e _ {i} (x) e _ {j} (x) p (x) d x = \mathbb {1} _ {\{i = j \}}. +$$ + +One can further show that (ignoring a multiplicative constant) + +$$ +\mathbb {E} \bigg (\frac {1}{n} \sum_ {i = 1} ^ {n} \min (\hat {\mu} _ {i} / 4, \lambda_ {n}) \bigg) \leq \frac {1}{n} \sum_ {i = 1} ^ {\infty} \min (\mu_ {i} / 4, \lambda_ {n}). +$$ + +When $k$ is the Sobolev kernel and $p(x)$ is the uniform density on $[0,1]$ , we find the eigenvalues satisfy + +$$ +\mu_ {j} / 4 = \frac {1}{\pi^ {2} (2 j - 1) ^ {2}}. +$$ + +Thus + +$$ +\begin{array}{l} \sum_ {i = 1} ^ {\infty} \min (\mu_ {i} / 4, \lambda_ {n}) \leq \frac {\lambda_ {n}}{2} \bigg (\frac {1}{\sqrt {\pi^ {2} \lambda_ {n}}} + 1 \bigg) + \frac {1}{\pi^ {2}} \int_ {\{(\pi^ {2} \lambda_ {n}) ^ {- 1 / 2} + 1 \} / 2} ^ {\infty} \frac {1}{(2 x - 1) ^ {2}} d x \\ = \sqrt {\lambda_ {n}} / \pi + \lambda_ {n} / 2 = O (\sqrt {\lambda_ {n}}) \\ \end{array} +$$ + +as $\lambda_{n}\to 0$ . Putting things together, we see that + +$$ +\mathbb {E} (\delta_ {n} (\lambda_ {n})) = O \bigg (\frac {\sigma^ {2}}{n \lambda_ {n} ^ {1 / 2}} + \lambda_ {n} \bigg). +$$ + +Thus an optimal $\lambda_{n}\sim (\sigma^{2} / n)^{2 / 3}$ gives an error rate of order $(\sigma^2 /n)^{2 / 3}$ . + +In fact, one can show that this is the best error rate one can achieve with any estimator in this problem. More generally Yang et al. [2015] shows that for essentially any RKHS H we have + +$$ +\inf _ {\hat {f}} \sup _ {f ^ {0}: \| f ^ {0} \| _ {\mathcal {H}} \leq 1} \mathbb {E} \bigg \{\frac {1}{n} \sum_ {i = 1} ^ {n} \{f ^ {0} (x _ {i}) - \hat {f} (x _ {i}) \} ^ {2} \bigg \} \geq c \inf _ {\lambda_ {n}} \delta_ {n} (\lambda_ {n}) +$$ + +where c > 0 is a constant and $\hat{f}$ is allowed to range over all (measurable) functions of the data Y, X. The conclusion is that kernel ridge regression is the optimal regression procedure up to a constant factor in terms of MSPE when the true signal $f^{0}$ is from an RKHS. + +## 1.6 Other kernel machines + +Thus far we have we have only considered applying the kernel trick to ridge regression, which as we have seen has attractive theoretical properties as a regression method. However the kernel trick and the representer theorem are much more generally applicable. In settings where the $Y_{i}$ are not continuous but are in $\{-1,1\}$ (e.g. labels for spam and ham, fraud and not fraud etc.), popular approaches include kernel logistic regression and the support vector machine (SVM) [Cortes and Vapnik, 1995]. + +## 1.6.1 The support vector machine + +![](images/268cebd69b94c534f035f9a0b86c698368208e4d22331695d1b51ce7a442639d.jpg) + +Consider first the simple case where the data in the two classes $\{x_{i}\}_{i:Y_{i}=1}$ and $\{x_{i}\}_{i:Y_{i}=-1}$ are separable by a hyperplane through the origin, so there exists $\beta\in R^{p}$ with $\|\beta\|_{2}=1$ such that $Y_{i}\beta^{T}x_{i}>0$ for all i. Note $\beta$ would then be a unit normal vector to a plane that separates the two classes. + +There may be an infinite number of planes that separate the classes, in which case it seems sensible to use the plane that maximises the margin between the two classes. Consider therefore the following optimisation problem. + +$$ +\max _ {\beta \in \mathbb {R} ^ {p}, M \geq 0} M +$$ + +subject to $Y_{i}x_{i}^{T}\beta/\|\beta\|_{2}\geq M, i=1,\ldots,n.$ + +Note that by normalising $\beta$ above we need not impose the constraint that $\|\beta\|_{2}=1$ . + +Suppose now that the classes are not separable. One way to handle this is to replace the constraint $Y_{i}x_{i}^{T}\beta/\|\beta\|_{2}\geq M$ with a penalty for how far over the margin boundary $x_{i}$ is. This penalty should be zero if $x_{i}$ is on the correct side of the boundary (i.e. when $Y_{i}x_{i}^{T}\beta/\|\beta\|_{2}\geq M$ ), and should be equal to the distance over the boundary, $M-Y_{i}x_{i}^{T}\beta/\|\beta\|_{2}$ otherwise. It will in fact be more convenient to penalise according to $1-Y_{i}x_{i}^{T}\beta/(\|\beta\|_{2}M)$ in the latter case, which is the distance measured in units of M. This penalty is invariant to $\beta$ undergoing any positive scaling, so we may set $\|\beta\|_{2}=1/M$ , thus eliminating M from the objective function. Switching $\max1/\|\beta\|_{2}$ with $\min\|\beta\|_{2}^{2}$ and adding the penalty we arrive at + +$$ +\underset {\beta \in \mathbb {R} ^ {p}} {\arg \min} \| \beta \| _ {2} ^ {2} + \lambda \sum_ {i = 1} ^ {n} (1 - Y _ {i} x _ {i} ^ {T} \beta) _ {+}, +$$ + +where $(\cdot)_{+}$ denotes the positive part. Replacing $\lambda$ with $1/\lambda$ we can write the objective in the more familiar-looking form + +$$ +\underset {\beta \in \mathbb {R} ^ {p}} {\arg \min} \sum_ {i = 1} ^ {n} (1 - Y _ {i} x _ {i} ^ {T} \beta) _ {+} + \lambda \| \beta \| _ {2} ^ {2}. +$$ + +Thus far we have restricted ourselves to hyperplanes through the origin but we would more generally want to consider any translate of these i.e. any hyperplane. This can be achieved by allowing ourselves to translate the $x_{i}$ by an arbitrary vector b, giving + +$$ +\underset {\beta \in \mathbb {R} ^ {p}, b \in \mathbb {R} ^ {p}} {\arg \min} \sum_ {i = 1} ^ {n} (1 - Y _ {i} (x _ {i} - b) ^ {T} \beta) _ {+} + \lambda \| \beta \| _ {2} ^ {2}, +$$ + +or equivalently + +$$ +(\hat {\mu}, \hat {\beta}) = \underset {(\mu , \beta) \in \mathbb {R} \times \mathbb {R} ^ {p}} {\arg \min} \sum_ {i = 1} ^ {n} \{1 - Y _ {i} (x _ {i} ^ {T} \beta + \mu) \} _ {+} + \lambda \| \beta \| _ {2} ^ {2}. \tag {1.6.1} +$$ + +This final objective defines the support vector classifier; given a new observation $x$ predictions are obtained via $\mathrm{sgn}(\hat{\mu} + x^T\hat{\beta})$ . + +Note that the objective in $(1.6.1)$ may be re-written as + +$$ +(\hat {\mu}, \hat {f}) = \underset {(\mu , f) \in \mathbb {R} \times \mathcal {H}} {\arg \min} \sum_ {i = 1} ^ {n} [ 1 - Y _ {i} \{f (x _ {i}) + \mu \} ] _ {+} + \lambda \| f \| _ {\mathcal {H}} ^ {2}, \tag {1.6.2} +$$ + +where H is the RKHS corresponding to the linear kernel. The representer theorem (more specifically the variant in question 10 of example sheet 1) shows that $(1.6.2)$ for an arbitrary RKHS with kernel k and kernel matrix K is equivalent to the support vector machine + +$$ +(\hat {\mu}, \hat {\alpha}) = \underset {(\mu , \alpha) \in \mathbb {R} \times \mathbb {R} ^ {n}} {\arg \min} \sum_ {i = 1} ^ {n} [ 1 - Y _ {i} \{K _ {i} ^ {T} \alpha + \mu \} ] _ {+} + \lambda \alpha^ {T} K \alpha . +$$ + +Predictions at a new $x$ are given by + +$$ +\mathrm{sgn} \bigg (\hat {\mu} + \sum_ {i = 1} ^ {n} \hat {\alpha} _ {i} k (x, x _ {i}) \bigg). +$$ + +## 1.6.2 Logistic regression + +Recall that standard logistic regression may be motivated by assuming + +$$ +\log \left(\frac {\mathbb {P} (Y _ {i} = 1)}{\mathbb {P} (Y _ {i} = - 1)}\right) = x _ {i} ^ {T} \beta^ {0} +$$ + +and picking $\hat{\beta}$ to maximise the log-likelihood. This leads to (see example sheet) the following optimisation problem: + +$$ +\underset {\beta \in \mathbb {R} ^ {p}} {\arg \min} \sum_ {i = 1} ^ {n} \log \{1 + \exp (- Y _ {i} x _ {i} ^ {T} \beta) \}. +$$ + +The ‘kernelised’ version is given by + +$$ +\underset {f \in \mathcal {H}} {\arg \min} \left\{\sum_ {i = 1} ^ {n} \log [ 1 + \exp \{- Y _ {i} f (x _ {i}) \} ] + \lambda \| f \| _ {\mathcal {H}} ^ {2} \right\}, +$$ + +where H is an RKHS. As in the case of the SVM, the representer theorem gives a finite-dimensional optimisation that is equivalent to the above. + +## 1.7 Large-scale kernel machines + +We introduced the kernel trick as a computational device that avoided performing calculations in a high or infinite dimensional feature space and, in the case of kernel ridge regression reduced computation down to forming the $n \times n$ matrix $K$ and then inverting $K + \lambda I$ . This can be a huge saving, but when $n$ is very large, this can present serious computational difficulties. Even if $p$ is small, the $O(n^3)$ cost of inverting $K + \lambda I$ may cause problems. What's worse, the fitted regression function is a sum over $n$ terms: + +$$ +\hat {f} (\cdot) = \sum_ {i = 1} ^ {n} \hat {\alpha} _ {i} k (x _ {i}, \cdot). +$$ + +Even to evaluate a prediction at a single new observation requires $O(n)$ computations unless $\hat{\alpha}$ is sparse. + +In recent years, there has been great interest in speeding up computations for kernel machines. We will discuss one exciting approach based on random feature expansions. Given a kernel k, the key idea is to develop a random map + +$$ +\hat {\phi}: \mathcal {X} \to \mathbb {R} ^ {b} +$$ + +with b small such that $\mathbb{E}\{\hat{\phi}(x)^{T}\hat{\phi}(x')\}=k(x,x')$ . In a sense we are trying to reverse the kernel trick by approximating the kernel using a random feature map. To increase the quality of the approximation of the kernel, we can consider + +$$ +x \mapsto \frac {1}{\sqrt {L}} (\hat {\phi} _ {1} (x), \ldots , \hat {\phi} _ {L} (x)) \in \mathbb {R} ^ {L b} +$$ + +with each $(\hat{\phi}_{l}(x))_{l=1}^{L}$ being i.i.d. for each x. Let $\Phi$ be the matrix with ith row given by $(\hat{\phi}_{1}(x_{i}),\ldots,\hat{\phi}_{L}(x_{i}))/\sqrt{L}$ . We may then run our learning algorithm replacing the initial matrix of predictors X with $\Phi$ . For example, when performing ridge regression, we can compute + +$$ +(\Phi^ {T} \Phi + \lambda I) ^ {- 1} \Phi^ {T} Y, +$$ + +which would require $O(nL^2 b^2 + L^3 b^3)$ operations: a cost linear in $n$ . Predicting a new observation would cost $O(Lb)$ . + +The work of Rahimi and Recht [2007] proposes a construction of such a random mapping $\hat{\phi}$ for shift-invariant kernels, that is kernels for which there exists a function $h$ with $k(x,x^{\prime}) = h(x - x^{\prime})$ for all $x,x^{\prime}\in \mathcal{X} = \mathbb{R}^{p}$ . A useful property of such kernels is given by Bochner's theorem. + +Theorem 8 (Bochner's theorem). Let $k: \mathbb{R}^p \times \mathbb{R}^p \to \mathbb{R}$ be a continuous kernel. Then $k$ is shift-invariant if and only if there exists some $c > 0$ and distribution $F$ on $\mathbb{R}^p$ such that when $W \sim F$ + +$$ +k (x, x ^ {\prime}) = c \mathbb {E} e ^ {i (x - x ^ {\prime}) ^ {T} W} = c \mathbb {E} \cos ((x - x ^ {\prime}) ^ {T} W). +$$ + +To make use of this theorem, first observe the following. Let $u \sim U[-\pi, \pi]$ , $x, y \in R$ . Then + +$$ +2 \mathbb {E} \cos (x + u) \cos (y + u) = 2 \mathbb {E} \{(\cos x \cos u - \sin x \sin u) (\cos y \cos u - \sin y \sin u) \}. +$$ + +Now as $u \stackrel{d}{=} -u$ , $\mathbb{E}\cos u\sin u = \mathbb{E}\cos(-u)\sin(-u) = -\mathbb{E}\cos u\sin u = 0$ . Also of course $\cos^{2}u + \sin^{2}u = 1$ so $E\cos^{2}u = E\sin^{2}u = 1/2$ . Thus + +$$ +2 \mathbb {E} \cos (x + u) \cos (y + u) = \cos x \cos y + \sin x \sin y = \cos (x - y). +$$ + +Given a shift-invariant kernel $k$ with associated distribution $F$ , suppose $W \sim F$ and let $u \sim U[-\pi, \pi]$ independently. Define + +$$ +\hat {\phi} (x) = \sqrt {2 c} \cos (W ^ {T} x + u). +$$ + +Then + +$$ +\begin{array}{l} \mathbb {E} \hat {\phi} (x) \hat {\phi} (x ^ {\prime}) = 2 c \mathbb {E} [ \mathbb {E} \{\cos (W ^ {T} x + u) \cos (W ^ {T} x ^ {\prime} + u) | W \} ] \\ = c \mathbb {E} \cos ((x - x ^ {\prime}) ^ {T} W) = k (x, x ^ {\prime}). \\ \end{array} +$$ + +As a concrete example of this approach, let us take the Gaussian kernel $k(x,x') = \exp \{-\| x - x'\|_2^2 /(2\sigma^2)\}$ . Note that if $W\sim N(0,\sigma^{-2}I)$ , it has characteristic function $\mathbb{E}(e^{it^T W}) = e^{-\| t\|_2^2 /(2\sigma^2)}$ so we may take $\hat{\phi} (x) = \sqrt{2}\cos (W^T x + u)$ . + +## Chapter 2 + +# Introduction to the Lasso + +## 2.1 Model selection + +Let us revisit the linear model $Y = X\beta^{0} + \varepsilon$ where $\mathbb{E}(\varepsilon) = 0$ , $\operatorname{Var}(\varepsilon) = \sigma^{2}I$ . In many modern datasets, there are reasons to believe there are many more variables present than are necessary to explain the response. Let S be the set $S = \{k : \beta_{k}^{0} \neq 0\}$ and suppose $s := |S| \ll p$ . + +The MSPE of OLS is + +$$ +\begin{array}{l} \frac {1}{n} \mathbb {E} \| X \beta^ {0} - X \hat {\beta} ^ {\mathrm{OLS}} \| _ {2} ^ {2} = \frac {1}{n} \mathbb {E} \{(\beta^ {0} - \hat {\beta} ^ {\mathrm{OLS}}) ^ {T} X ^ {T} X (\beta^ {0} - \hat {\beta} ^ {\mathrm{OLS}}) \} \\ = \frac {1}{n} \mathbb {E} [ \operatorname{tr} \left\{\left(\beta^ {0} - \hat {\beta} ^ {\mathrm{OLS}}\right) \left(\beta^ {0} - \hat {\beta} ^ {\mathrm{OLS}}\right) ^ {T} X ^ {T} X \right\} ] \\ = \frac {1}{n} \mathrm{tr} [ \mathbb {E} \{(\beta^ {0} - \hat {\beta} ^ {\mathrm{OLS}}) (\beta^ {0} - \hat {\beta} ^ {\mathrm{OLS}}) ^ {T} \} X ^ {T} X ] \\ = \frac {1}{n} \mathrm{tr} (\mathrm{Var} (\hat {\beta} ^ {\mathrm{OLS}}) X ^ {T} X) = \frac {p}{n} \sigma^ {2}. \\ \end{array} +$$ + +If we could identify S and then fit a linear model using just these variables, we'd obtain an MSPE of $\sigma^{2}s/n$ which could be substantially smaller than $\sigma^{2}p/n$ . Furthermore, it can be shown that parameter estimates from the reduced model are more accurate. The smaller model would also be easier to interpret. + +We now briefly review some classical model selection strategies. + +## Best subset regression + +A natural approach to finding S is to consider all $2^{p}$ possible regression procedures each involving regressing the response on a different sets of explanatory variables $X_{M}$ where M is a subset of $\{1,\ldots,p\}$ . We can then pick the best regression procedure using cross-validation (say). For general design matrices, this involves an exhaustive search over all subsets, so this is not really feasible for p > 50. + +## Forward selection + +This can be seen as a greedy way of performing best subsets regression. Given a target model size m (the tuning parameter), this works as follows. + +1. Start by fitting an intercept only model. +2. Add to the current model the predictor variable that reduces the residual sum of squares the most. +3. Continue step 2 until m predictor variables have been selected. + +## 2.2 The Lasso estimator + +The Least absolute shrinkage and selection operator (Lasso) [Tibshirani, 1996] estimates $\beta^0$ by $\hat{\beta}_{\lambda}^{\mathrm{L}}$ , where $(\hat{\mu}^{\mathrm{L}},\hat{\beta}_{\lambda}^{\mathrm{L}})$ minimise + +$$ +\frac {1}{2 n} \| Y - \mu \mathbf {1} - X \beta \| _ {2} ^ {2} + \lambda \| \beta \| _ {1} \tag {2.2.1} +$$ + +over $(\mu, \beta) \in \mathbb{R} \times \mathbb{R}^p$ . Here $\| \beta \|_1$ is the $\ell_1$ -norm of $\beta$ : $\| \beta \|_1 = \sum_{k=1}^{p} |\beta_k|$ . + +Like ridge regression, $\hat{\beta}_{\lambda}^{L}$ shrinks the OLS estimate towards the origin, but there is an important difference. The $\ell_{1}$ penalty can force some of the estimated coefficients to be exactly 0. In this way the Lasso can perform simultaneous variable selection and parameter estimation. As we did with ridge regression, we can centre and scale the X matrix, and also centre Y and thus remove $\mu$ from the objective. Define + +$$ +Q _ {\lambda} (\beta) = \frac {1}{2 n} \| Y - X \beta \| _ {2} ^ {2} + \lambda \| \beta \| _ {1}. \tag {2.2.2} +$$ + +Now the minimiser(s) of $Q_{\lambda}(\beta)$ will also be the minimiser(s) of + +$$ +\| Y - X \beta \| _ {2} ^ {2} \text {subject to} \| \beta \| _ {1} \leq \| \hat {\beta} _ {\lambda} ^ {\mathrm{L}} \| _ {1}. +$$ + +Similarly, with the Ridge regression objective, we know that $\hat{\beta}_{\lambda}^{\mathrm{R}}$ minimises $\| Y - X\beta \|_2^2$ subject to $\| \beta \|_2 \leq \| \hat{\beta}_{\lambda}^{\mathrm{R}} \|_2$ . + +Now the contours of the OLS objective $\|Y - X\beta\|_{2}^{2}$ are ellipsoids centred at $\hat{\beta}^{OLS}$ , while the contours of $\|\beta\|_{2}^{2}$ are spheres centred at the origin, and the contours of $\|\beta\|_{1}$ are ‘diamonds’ centred at 0. + +The important point to note is that the $\ell_{1}$ ball $\{\beta\in R^{p}:\|\beta\|_{1}\leq\|\hat{\beta}_{\lambda}^{\mathrm{L}}\|_{1}\}$ has corners where some of the components are zero, and it is likely that the OLS contours will intersect the $\ell_{1}$ ball at such a corner. + +## 2.2.1 Convexity and the Lasso + +One important difference between the Lasso and Best subsets regression is that the former involves a convex optimisation, which makes it computationally tractable for large-scale problems, in contrast to the latter which typically only feasible for problems with p < 50 or so. + +Recall that a set $A \subseteq \mathbb{R}^d$ is convex if + +$$ +x, y \in A \Rightarrow (1 - t) x + t y \in A \quad \text { for all } t \in (0, 1). +$$ + +A function $f:\mathbb{R}^d\to \mathbb{R}$ is convex if + +$$ +f \big ((1 - t) x + t y \big) \leq (1 - t) f (x) + t f (y) +$$ + +for all $x, y \in \mathbb{R}^d$ and $t \in (0,1)$ . + +Proposition 9. (i) Let $f_1, \ldots, f_m: \mathbb{R}^d \to \mathbb{R}$ be convex functions. Then if $c_1, \ldots, c_m \geq 0$ , $c_1f_1 + \cdots c_mf_m$ is a convex function. + +(ii) If $f: \mathbb{R}^d \to \mathbb{R}$ is twice continuously differentiable then + +(a) $f$ is convex iff. its Hessian $H(x)$ is positive semi-definite for all $x$ , +(b) $f$ is strictly convex if $H(x)$ is positive definite for all $x$ . + +Lasso solutions need not be unique (e.g. if X has duplicate columns), though for most reasonable design matrices, Lasso solutions will be unique. We will often tacitly assume the Lasso solutions is unique in the statement of our theoretical results. It is however straightforward to show that the Lasso fitted values are unique. + +Proposition 10. $X\hat{\beta}_{\lambda}^{\mathrm{L}}$ is unique. + +Proof. Fix $\lambda$ and suppose $\hat{\beta}^{(1)}$ and $\hat{\beta}^{(2)}$ are two Lasso solutions giving an optimal objective value of $c^*$ . Now for $t \in (0,1)$ , by strict convexity of $\| \cdot \|_2^2$ , + +$$ +\| Y - t X \hat {\beta} ^ {(1)} - (1 - t) X \hat {\beta} ^ {(2)} \| _ {2} ^ {2} \leq t \| Y - X \hat {\beta} ^ {(1)} \| _ {2} ^ {2} + (1 - t) \| Y - X \hat {\beta} ^ {(2)} \| _ {2} ^ {2}, +$$ + +with equality if and only if $X\hat{\beta}^{(1)} = X\hat{\beta}^{(2)}$ . Since $\| \cdot \|_1$ is also convex, we see that + +$$ +\begin{array}{l} c ^ {*} \leq Q _ {\lambda} (t \hat {\beta} ^ {(1)} + (1 - t) \hat {\beta} ^ {(2)}) \\ = \| Y - t X \hat {\beta} ^ {(1)} - (1 - t) X \hat {\beta} ^ {(2)} \| _ {2} ^ {2} / (2 n) + \lambda \| t \hat {\beta} ^ {(1)} + (1 - t) \hat {\beta} ^ {(2)} \| _ {1} \\ \leq t \| Y - X \hat {\beta} ^ {(1)} \| _ {2} ^ {2} / (2 n) + (1 - t) \| Y - X \hat {\beta} ^ {(2)} \| _ {2} ^ {2} / (2 n) + \lambda \| t \hat {\beta} ^ {(1)} + (1 - t) \hat {\beta} ^ {(2)} \| _ {1} \\ \leq t \{\| Y - X \hat {\beta} ^ {(1)} \| _ {2} ^ {2} / (2 n) + \lambda \| \hat {\beta} ^ {(1)} \| _ {1} \} + (1 - t) \{\| Y - X \hat {\beta} ^ {(2)} \| _ {2} ^ {2} / (2 n) + \lambda \| \hat {\beta} ^ {(2)} \| _ {1} \} \\ = t Q (\hat {\beta} ^ {(1)}) + (1 - t) Q (\hat {\beta} ^ {(2)}) \leq c ^ {*}. \\ \end{array} +$$ + +Equality must prevail throughout this chain of inequalities, so $X\hat{\beta}^{(1)} = X\hat{\beta}^{(2)}$ . + +□ + +## 2.2.2 Prediction error of the Lasso with no assumptions on the design + +A remarkable property of the Lasso is that even when $p \gg n$ , it can still perform well in terms of prediction error. Suppose the columns of X have been centred and scaled (as we will always assume from now on unless stated otherwise) and assume the normal linear model (where we have already centred Y), + +$$ +Y = X \beta^ {0} + \varepsilon - \bar {\varepsilon} \mathbf {1} \tag {2.2.3} +$$ + +where $\varepsilon \sim N_{n}(0,\sigma^{2}I)$ . + +Theorem 11. Let $\hat{\beta}$ be the Lasso solution when + +$$ +\lambda = A \sigma \sqrt {\frac {\log (p)}{n}}. +$$ + +With probability at least $1 - 2p^{-(A^2 /2 - 1)}$ + +$$ +\frac {1}{n} \| X (\beta^ {0} - \hat {\beta}) \| _ {2} ^ {2} \leq 4 A \sigma \sqrt {\frac {\log (p)}{n}} \| \beta^ {0} \| _ {1}. +$$ + +Proof. From the definition of $\hat{\beta}$ we have + +$$ +\frac {1}{2 n} \| Y - X \hat {\beta} \| _ {2} ^ {2} + \lambda \| \hat {\beta} \| _ {1} \leq \frac {1}{2 n} \| Y - X \beta^ {0} \| _ {2} ^ {2} + \lambda \| \beta^ {0} \| _ {1}. +$$ + +Rearranging, + +$$ +\frac {1}{2 n} \| X (\beta^ {0} - \hat {\beta}) \| _ {2} ^ {2} \leq \frac {1}{n} \varepsilon^ {T} X (\hat {\beta} - \beta^ {0}) + \lambda \| \beta^ {0} \| _ {1} - \lambda \| \hat {\beta} \| _ {1}. +$$ + +Now $|\varepsilon^T X(\hat{\beta} -\beta^0)|\leq \| X^T\varepsilon \|_\infty \| \hat{\beta} -\beta^0\| _1$ . Let $\Omega = \{\| X^T\varepsilon \|_{\infty} / n\leq \lambda \}$ . Lemma 15 below shows that $\mathbb{P}(\Omega)\geq 1 - 2p^{-(A^2 /2 - 1)}$ . Working on the event $\Omega$ , we obtain + +$$ +\frac {1}{2 n} \| X (\beta^ {0} - \hat {\beta}) \| _ {2} ^ {2} \leq \lambda \| \beta^ {0} - \hat {\beta} \| _ {1} + \lambda \| \beta^ {0} \| _ {1} - \lambda \| \hat {\beta} \| _ {1}, +$$ + +$$ +\frac {1}{n} \| X (\beta^ {0} - \hat {\beta}) \| _ {2} ^ {2} \leq 4 \lambda \| \beta^ {0} \| _ {1}, \qquad \mathrm{bythetriangleinequality.} +$$ + +## 2.2.3 Concentration inequalities I + +The proof of Theorem 11 relies on a lower bound for the probability of the event $\Omega$ . A union bound gives + +$$ +\begin{array}{l} \mathbb {P} (\| X ^ {T} \varepsilon \| _ {\infty} / n > \lambda) = \mathbb {P} (\cup_ {j = 1} ^ {p} | X _ {j} ^ {T} \varepsilon | / n > \lambda) \\ \leq \sum_ {j = 1} ^ {p} \mathbb {P} (| X _ {j} ^ {T} \varepsilon | / n > \lambda). \\ \end{array} +$$ + +Now $X_{j}^{T}\varepsilon/n \sim N(0,\sigma^{2}/n)$ , so if we obtain a bound on the tail probabilities of normal distributions, the argument above will give a bound for $\mathbb{P}(\Omega)$ . + +Motivated by the need to bound normal tail probabilities, we will briefly discuss the topic of concentration inequalities that provide such bounds for much wider classes of random variables. Concentration inequalities are vital for the study of many modern algorithms and in our case here, they will reveal that the attractive properties of the Lasso presented in Theorem 11 hold true for a variety of non-normal errors. + +We begin our discussion with the simplest tail bound, Markov's inequality, which states that given a non-negative random variable $W$ , + +$$ +\mathbb {P} (W \geq t) \leq \frac {\mathbb {E} (W)}{t}. +$$ + +This immediately implies that given a strictly increasing function $\varphi: R \to [0, \infty)$ and any random variable W, + +$$ +\mathbb {P} (W \geq t) = \mathbb {P} \{\varphi (W) \geq \varphi (t) \} \leq \frac {\mathbb {E} (\varphi (W))}{\varphi (t)}. +$$ + +Applying this with $\varphi(t)=e^{\alpha t}$ ( $\alpha>0$ ) yields the so-called Chernoff bound: + +$$ +\mathbb {P} (W \geq t) \leq \inf _ {\alpha > 0} e ^ {- \alpha t} \mathbb {E} e ^ {\alpha W}. +$$ + +Consider the case when $W \sim N(0, \sigma^{2})$ . Recall that + +$$ +\mathbb {E} e ^ {\alpha W} = e ^ {\alpha^ {2} \sigma^ {2} / 2}. \tag {2.2.4} +$$ + +Thus + +$$ +\mathbb {P} (W \geq t) \leq \inf _ {\alpha > 0} e ^ {\alpha^ {2} \sigma^ {2} / 2 - \alpha t} = e ^ {- t ^ {2} / (2 \sigma^ {2})}. +$$ + +Note that to arrive at this bound, all we required was (an upper bound on) the moment generating function (mgf) of W (2.2.4). + +## Sub-Gaussian variables + +Definition 3. We say a random variable $W$ with mean $\mu = \mathbb{E}(W)$ is sub-Gaussian if there exists $\sigma > 0$ such that + +$$ +\mathbb {E} e ^ {\alpha (W - \mu)} \leq e ^ {\alpha^ {2} \sigma^ {2} / 2} +$$ + +for all $\alpha \in \mathbb{R}$ . We then say that $W$ is sub-Gaussian with parameter $\sigma$ . + +Proposition 12 (Sub-Gaussian tail bound). If W is sub-Gaussian with parameter $\sigma$ and $\mathbb{E}(W)=\mu$ then + +$$ +\mathbb {P} (W - \mu \geq t) \leq e ^ {- t ^ {2} / (2 \sigma^ {2})}. +$$ + +As well as Gaussian random variables, the sub-Gaussian class includes bounded random variables. + +Lemma 13 (Hoeffding's lemma). If $W$ is mean-zero and takes values in $[a, b]$ , then $W$ is sub-Gaussian with parameter $(b - a)/2$ . + +The following proposition shows that analogously to how a linear combination of jointly Gaussian random variables is Gaussian, a linear combination of sub-Gaussian random variables is also sub-Gaussian. + +Proposition 14. Let $(W_{i})_{i=1}^{n}$ be a sequence of independent mean-zero sub-Gaussian random variables with parameters $(\sigma_{i})_{i=1}^{n}$ and let $\gamma \in R^{n}$ . Then $\gamma^{T}W$ is sub-Gaussian with parameter $\left(\sum_{i}\gamma_{i}^{2}\sigma_{i}^{2}\right)^{1/2}$ . + +Proof. + +$$ +\begin{array}{l} \mathbb {E} \exp \left(\alpha \sum_ {i = 1} ^ {n} \gamma_ {i} W _ {i}\right) = \prod_ {i = 1} ^ {n} \mathbb {E} \exp (\alpha \gamma_ {i} W _ {i}) \\ \leq \prod_ {i = 1} ^ {n} \exp (\alpha^ {2} \gamma_ {i} ^ {2} \sigma_ {i} ^ {2} / 2) \\ = \exp \left(\alpha^ {2} \sum_ {i = 1} ^ {n} \gamma_ {i} ^ {2} \sigma_ {i} ^ {2} / 2\right). \\ \end{array} +$$ + +![](images/bee210c900bbc2669fab5522f561a51ed5691176ecfca0bde291ff9aff7d1a33.jpg) + +We can now prove a more general version of the probability bound required for Theorem 11. + +Lemma 15. Suppose $(\varepsilon_{i})_{i=1}^{n}$ are independent, mean-zero and sub-Gaussian with common parameter $\sigma$ . Note that this includes $\varepsilon \sim N_{n}(0, \sigma^{2}I)$ . Let $\lambda = A\sigma\sqrt{\log(p)/n}$ . Then + +$$ +\mathbb {P} (\| X ^ {T} \varepsilon \| _ {\infty} / n \leq \lambda) \geq 1 - 2 p ^ {- (A ^ {2} / 2 - 1)}. +$$ + +Proof. + +$$ +\mathbb {P} (\| X ^ {T} \varepsilon \| _ {\infty} / n > \lambda) \leq \sum_ {j = 1} ^ {p} \mathbb {P} (| X _ {j} ^ {T} \varepsilon | / n > \lambda). +$$ + +But $\pm X_j^T\varepsilon /n$ are both sub-Gaussian with parameter $(\sigma^2\| X_j\| _2^2 /n^2)^{1 / 2} = \sigma /\sqrt{n}$ . Thus the RHS is at most + +$$ +2 p \exp (- A ^ {2} \log (p) / 2) = 2 p ^ {1 - A ^ {2} / 2}. +$$ + +![](images/56624c7cc85aa5e00a902e43d582ad6d502424bfc0c47ac138533e7e828c694f.jpg) + +## Chapter 3 + +## The Lasso and beyond + +In this chapter, we will study the Lasso in more depth and present more precise theory about its behaviour. We will first study the variable selection properties of the Lasso, which will require some basic results in convex analysis that we present below. + +## 3.1 Some facts from optimisation theory and convex analysis + +## 3.1.1 Subgradients + +Definition 4. A vector $v \in \mathbb{R}^d$ is a subgradient of a convex function $f: \mathbb{R}^d \to \mathbb{R}$ at $x$ if + +$$ +f (y) \geq f (x) + v ^ {T} (y - x) \quad \text { for all } y \in \mathbb {R} ^ {d}. +$$ + +The set of subgradients of f at x is called the subdifferential of f at x and denoted $\partial f(x)$ . + +In order to make use of subgradients, we will require the following two facts: + +Proposition 16. Let $f: \mathbb{R}^d \to \mathbb{R}$ be convex, and suppose $f$ is differentiable at $x$ . Then $\partial f(x) = \{\nabla f(x)\}$ . + +Proposition 17. Let $f, g: \mathbb{R}^d \to \mathbb{R}$ be convex, and let $\alpha > 0$ . Then + +$$ +\partial (\alpha f) (x) = \alpha \partial f (x) = \{\alpha v: v \in \partial f (x) \}, +$$ + +$$ +\partial (f + g) (x) = \partial f (x) + \partial g (x) = \{v + w: v \in \partial f (x), w \in \partial g (x) \}. +$$ + +The following easy (but key) result is often referred to in the statistical literature as the Karush–Kuhn–Tucker (KKT) conditions, though it is actually a much simplified version of them. + +Proposition 18. $x^{*}\in \arg \min_{x\in \mathbb{R}^{d}}f(x)$ if and only if $0\in \partial f(x^{*})$ + +Proof. + +$$ +f (y) \geq f (x ^ {*}) \quad \text { for all } y \in \mathbb {R} ^ {d} \Leftrightarrow f (y) \geq f (x ^ {*}) + 0 ^ {T} (y - x) \quad \text { for all } y \in \mathbb {R} ^ {d} +$$ + +$$ +\Leftrightarrow 0 \in \partial f (x ^ {*}). +$$ + +Let us now compute the subdifferential of the $\ell_{1}$ -norm. First note that $\|\cdot\|_{1}:R^{d}\to R$ is convex. Indeed it is a norm so the triangle inequality gives $\|tx+(1-t)y\|_{1}\leq t\|x\|_{1}+(1-t)\|y\|_{1}$ . We introduce some notation that will be helpful here and throughout the rest of the course. + +For $x \in R^{d}$ and $A = \{k_{1}, \ldots, k_{m}\} \subseteq \{1, \ldots, d\}$ with $k_{1} < \cdots < k_{m}$ , by $x_{A}$ we will mean $(x_{k_{1}}, \ldots, x_{k_{m}})^{T}$ . Similarly if X has d columns we will write $X_{A}$ for the matrix + +$$ +X _ {A} = (X _ {k _ {1}} \dots X _ {k _ {m}}). +$$ + +Further in this context, by $A^{c}$ , we will mean $\{1,\ldots,d\}\backslash A$ . Additionally, when in subscripts we will use the shorthand $-j=\{j\}^{c}$ and $-jk=\{j,k\}^{c}$ . Note these column and component extraction operations will always be considered to have taken place first before any further operations on the matrix, so for example $X_{A}^{T}=(X_{A})^{T}$ . Finally, define + +$$ +\operatorname{sgn} (x _ {1}) = \left\{ \begin{array}{l l} - 1 & \text {if x_{1} < 0} \\ 0 & \text {if x_{1} = 0} \\ 1 & \text {if x_{1} >0}, \end{array} \right. +$$ + +and + +$$ +\operatorname{sgn} (x) = (\operatorname{sgn} (x _ {1}), \dots , \operatorname{sgn} (x _ {d})) ^ {T}. +$$ + +Proposition 19. For $x \in R^{d}$ let $A = \{j : x_{j} \neq 0\}$ . Then + +$$ +\partial \| x \| _ {1} = \{v \in \mathbb {R} ^ {d}: \| v \| _ {\infty} \leq 1 a n d v _ {A} = \operatorname{sgn} (x _ {A}) \} +$$ + +Proof. For $j = 1, \ldots, d$ , let + +$$ +g _ {j}: \mathbb {R} ^ {d} \to \mathbb {R} +$$ + +$$ +x \mapsto | x _ {j} |. +$$ + +Then $\| \cdot \| = \sum_{j} g_{j}(\cdot)$ so by Proposition 17, $\partial \| x \|_{1} = \sum_{j} \partial g_{j}(x)$ . When $x$ has $x_{j} \neq 0$ , $g_{j}$ is differentiable at $x$ so by Proposition 16 $\partial g_{j}(x) = \{\text{sgn}(x_{j})e_{j}\}$ where $e_{j}$ is the $j$ th unit vector. When $x_{j} = 0$ , if $v \in \partial g_{j}(x)$ we must have + +$$ +g _ {j} (y) \geq g _ {j} (x) + v ^ {T} (y - x) \quad \text {for all} y \in \mathbb {R} ^ {d}, +$$ + +SO + +$$ +\left| y _ {j} \right| \geq v ^ {T} (y - x) \quad \text { for all } y \in \mathbb {R} ^ {d}. \tag {3.1.1} +$$ + +we claim that the above holds iff. $v_{j} \in [-1,1]$ and $v_{-j} = 0$ . For the ‘if’ direction, note that $v^{T}(y - x) = v_{j}y_{j} \leq |y_{j}|$ . Conversely, set $y_{-j} = x_{-j} + v_{-j}$ and $y_{j} = 0$ in (3.1.1) to get $0 \geq \|v_{-j}\|_{2}^{2}$ , so $v_{-j} = 0$ . Then take y with $y_{-j} = x_{-j}$ to get $|y_{j}| \geq v_{j}y_{j}$ for all $y_{j} \in R$ , so $|v_{j}| \leq 1$ . Forming the set sum of the subdifferentials then gives the result. + +## 3.1.2 The Lagrangian method + +Consider an optimisation problem of the form + +$$ +\text { minimise } f (x), \text { subject to } g (x) = 0 \tag {3.1.2} +$$ + +where $g : R^{d} \to R^{b}$ . Suppose the optimal value is $c^{*} \in R$ . The Lagrangian for this problem is defined as + +$$ +L (x, \theta) = f (x) + \theta^ {T} g (x) +$$ + +where $\theta \in \mathbb{R}^b$ . Note that + +$$ +\inf _ {x \in \mathbb {R} ^ {d}} L (x, \theta) \leq \inf _ {x \in \mathbb {R} ^ {d}: g (x) = 0} L (x, \theta) = c ^ {*} +$$ + +for all $\theta$ . The Lagrangian method involves finding a $\theta^{*}$ such that the minimising $x^{*}$ on the LHS satisfies $g(x^{*}) = 0$ . This $x^{*}$ must then be a minimiser in the original problem (3.1.2). + +## 3.2 Lasso solutions + +Equipped with these tools from convex analysis, we can now fully characterise the solutions to the Lasso. We have that $\hat{\beta}_{\lambda}^{L}$ is a Lasso solution if and only if $0 \in \partial Q_{\lambda}(\hat{\beta}_{\lambda}^{L})$ , which is equivalent to + +$$ +\frac {1}{n} X ^ {T} (Y - X \hat {\beta} _ {\lambda} ^ {\mathrm{L}}) = \lambda \hat {\nu}, +$$ + +for $\hat{\nu}$ with $\| \hat{\nu}\|_{\infty}\leq 1$ and writing $\hat{S}_{\lambda} = \{k:\hat{\beta}_{\lambda ,k}^{\mathrm{L}}\neq 0\}$ , $\hat{\nu}_{\hat{S}_{\lambda}} = \mathrm{sgn}(\hat{\beta}_{\lambda ,\hat{S}_{\lambda}}^{\mathrm{L}})$ . These are known as the KKT conditions for the Lasso. + +## 3.3 Variable selection + +Consider now the “noiseless” version of the high-dimensional linear model (2.2.3), $Y = X\beta^{0}$ . The case with noise can dealt with by similar arguments to those we will use below when we work on an event that $\|X^{T}\varepsilon\|_{\infty}/n$ is small (see example sheet). + +Let $S = \{k : \beta_k^0 \neq 0\}$ , $N = \{1, \ldots, p\} \setminus S$ and assume wlog that $S = \{1, \ldots, s\}$ , and also that $\text{rank}(X_S) = s$ . + +Theorem 20. Let $\lambda > 0$ and define $\Delta = X_N^T X_S(X_S^T X_S)^{-1} \mathrm{sgn}(\beta_S^0)$ . If $\| \Delta \|_{\infty} \leq 1$ and for $k \in S$ , + +$$ +| \beta_ {k} ^ {0} | > \lambda | \mathrm{sgn} (\beta_ {S} ^ {0}) ^ {T} [ \{\frac {1}{n} X _ {S} ^ {T} X _ {S} \} ^ {- 1} ] _ {k} |, (3. 3. 1) +$$ + +then there exists a Lasso solution $\hat{\beta}_{\lambda}^{\mathrm{L}}$ with $\mathrm{sgn}(\hat{\beta}_{\lambda}^{\mathrm{L}}) = \mathrm{sgn}(\beta^{0})$ . As a partial converse, if there exists a Lasso solution $\hat{\beta}_{\lambda}^{\mathrm{L}}$ with $\mathrm{sgn}(\hat{\beta}_{\lambda}^{\mathrm{L}}) = \mathrm{sgn}(\beta^{0})$ , then $\| \Delta \|_{\infty} \leq 1$ . + +Remark 1. We can interpret $\|\Delta\|_{\infty}$ as the maximum in absolute value over $k \in N$ of the dot product of $\operatorname{sgn}(\beta_{S}^{0})$ and $(X_{S}^{T}X_{S})^{-1}X_{S}^{T}X_{k}$ , the coefficient vector obtained by regressing $X_{k}$ on $X_{S}$ . The condition $\|\Delta\|_{\infty} \leq 1$ is known as the irrepresentable condition. + +Proof. Fix $\lambda > 0$ and write $\hat{\beta} = \hat{\beta}_{\lambda}^{\mathrm{L}}$ and $\hat{S} = \{k : \hat{\beta}_k \neq 0\}$ for convenience. The KKT conditions for the Lasso give + +$$ +\frac {1}{n} X ^ {T} X (\beta^ {0} - \hat {\beta}) = \lambda \hat {\nu} +$$ + +where $\| \hat{\nu}\|_{\infty}\leq 1$ and $\hat{\nu}_{\hat{S}} = \mathrm{sgn}(\hat{\beta}_{\hat{S}})$ . We can expand this into + +$$ +\frac {1}{n} \left( \begin{array}{c c} X _ {S} ^ {T} X _ {S} & X _ {S} ^ {T} X _ {N} \\ X _ {N} ^ {T} X _ {S} & X _ {N} ^ {T} X _ {N} \end{array} \right) \binom{\beta_ {S} ^ {0} - \hat {\beta} _ {S}}{- \hat {\beta} _ {N}} = \lambda \binom{\hat {\nu} _ {S}}{\hat {\nu} _ {N}}. \tag {3.3.2} +$$ + +We prove the converse first. If $\operatorname{sgn}(\hat{\beta}) = \operatorname{sgn}(\beta^{0})$ then $\hat{\nu}_{S} = \operatorname{sgn}(\beta_{S}^{0})$ and $\hat{\beta}_{N} = 0$ . The top block of (3.3.2) gives + +$$ +\beta_ {S} ^ {0} - \hat {\beta} _ {S} = \lambda (\frac {1}{n} X _ {S} ^ {T} X _ {S}) ^ {- 1} \mathrm{sgn} (\beta_ {S} ^ {0}). +$$ + +Substituting this into the bottom block, we get + +$$ +\lambda \frac {1}{n} X _ {N} ^ {T} X _ {S} (\frac {1}{n} X _ {S} ^ {T} X _ {S}) ^ {- 1} \mathrm{sgn} (\beta_ {S} ^ {0}) = \lambda \hat {\nu} _ {N}. +$$ + +Thus as $\| \hat{\nu}_N\|_{\infty}\leq 1$ , we have $\| \Delta \|_{\infty}\leq 1$ . + +For the positive statement, we need to find a $\hat{\beta}$ and $\hat{\nu}$ such that $\mathrm{sgn}(\hat{\beta}_S) = \mathrm{sgn}(\beta_S^0)$ and $\hat{\beta}_N = 0$ , for which the KKT conditions hold. We claim that taking + +$$ +(\hat {\beta} _ {S}, \hat {\beta} _ {N}) = (\beta_ {S} ^ {0} - \lambda (\frac {1}{n} X _ {S} ^ {T} X _ {S}) ^ {- 1} \mathrm{sgn} (\beta_ {S} ^ {0}), 0) +$$ + +$$ +(\hat {\nu} _ {S}, \hat {\nu} _ {N}) = (\mathrm{sgn} (\beta_ {S} ^ {0}), \Delta) +$$ + +satisfies (3.3.2). We only need to check that $\operatorname{sgn}(\beta_{S}^{0}) = \operatorname{sgn}(\hat{\beta}_{S})$ , but this follows from (3.3.1). ☐ + +## 3.4 Prediction and estimation + +Consider again the model $Y = X\beta^{0} + \varepsilon - \bar{\varepsilon}1$ where the components of $\varepsilon$ are independent mean-zero sub-Gaussian random variables with common parameter $\sigma$ . Let S, s and N be defined as in the previous section. As we have noted before, in an artificial situation where S is known, we could apply OLS on $X_{S}$ and have an MSPE of $\sigma^{2}s/n$ . Under a so-called compatibility condition on the design matrix, we can obtain a similar MSPE for the Lasso. + +Definition 5. Given a matrix of predictors $X \in R^{n \times p}$ and support set S, define + +$$ +\phi^ {2} = \inf _ {\beta \in \mathbb {R} ^ {p}: \beta_ {S} \neq 0, \| \beta_ {N} \| _ {1} \leq 3 \| \beta_ {S} \| _ {1}} \frac {\frac {1}{n} \| X \beta \| _ {2} ^ {2}}{\frac {1}{s} \| \beta_ {S} \| _ {1} ^ {2}}, +$$ + +where $s = |S|$ and we take $\phi \geq 0$ . The compatibility condition is that $\phi^{2} > 0$ . + +Note that if $X^{T}X/n$ has minimum eigenvalue $c_{\min} > 0$ (so necessarily $p \leq n$ ), then $\phi^{2} > c_{\min}$ . Indeed by the Cauchy–Schwarz inequality, + +$$ +\| \beta_ {S} \| _ {1} = \mathrm{sgn} (\beta_ {S}) ^ {T} \beta_ {S} \leq \sqrt {s} \| \beta_ {S} \| _ {2} \leq \sqrt {s} \| \beta \| _ {2}. +$$ + +Thus + +$$ +\phi^ {2} \geq \inf _ {\beta \neq 0} \frac {\frac {1}{n} \| X \beta \| _ {2} ^ {2}}{\| \beta \| _ {2} ^ {2}} = c _ {\mathrm{min}}. +$$ + +Although in the high-dimensional setting we would have $c_{min} = 0$ , the fact that the infimum in the definition of $\phi^{2}$ is over a restricted set of $\beta$ can still allow $\phi^{2}$ to be positive even in this case, as we discuss after the presentation of the theorem. + +Theorem 21. Suppose the compatibility condition holds and let $\hat{\beta}$ be the Lasso solution with $\lambda = A\sigma \sqrt{\log(p) / n}$ for $A > 0$ . Then with probability at least $1 - 2p^{-(A^2 /8 - 1)}$ , we have + +$$ +\frac {1}{n} \| X (\beta^ {0} - \hat {\beta}) \| _ {2} ^ {2} + \lambda \| \hat {\beta} - \beta^ {0} \| _ {1} \leq \frac {1 6 \lambda^ {2} s}{\phi^ {2}} = \frac {1 6 A ^ {2} \log (p)}{\phi^ {2}} \frac {\sigma^ {2} s}{n}. +$$ + +Proof. As in Theorem 11 we start with the “basic inequality”: + +$$ +\frac {1}{2 n} \| X (\hat {\beta} - \beta^ {0}) \| _ {2} ^ {2} + \lambda \| \hat {\beta} \| _ {1} \leq \frac {1}{n} \varepsilon^ {T} X (\hat {\beta} - \beta^ {0}) + \lambda \| \beta^ {0} \| _ {1}. +$$ + +We work on the event $\Omega = \{2\| X^T\varepsilon \|_{\infty} / n\leq \lambda \}$ where after applying Hölder's inequality, we get + +$$ +\frac {1}{n} \| X (\hat {\beta} - \beta^ {0}) \| _ {2} ^ {2} + 2 \lambda \| \hat {\beta} \| _ {1} \leq \lambda \| \hat {\beta} - \beta^ {0} \| _ {1} + 2 \lambda \| \beta^ {0} \| _ {1}. \tag {3.4.1} +$$ + +Lemma 15 shows that $\mathbb{P}(\Omega) \geq 1 - 2p^{-(A^2/8-1)}$ . + +To motivate the rest of the proof, consider the following idea. We know + +$$ +\frac {1}{n} \| X (\hat {\beta} - \beta^ {0}) \| _ {2} ^ {2} \leq 3 \lambda \| \hat {\beta} - \beta^ {0} \| _ {1}. +$$ + +If we could get + +$$ +3 \lambda \| \hat {\beta} - \beta^ {0} \| _ {1} \leq \frac {c \lambda}{\sqrt {n}} \| X (\hat {\beta} - \beta^ {0}) \| _ {2} +$$ + +for some constant $c > 0$ , then we would have that $\| X(\hat{\beta} - \beta^0)\| _2^2 /n\leq c^2\lambda^2$ and also $3\lambda \| \beta^0 -\hat{\beta}\| _1\leq c^2\lambda^2$ . + +Returning to the actual proof, write $a = \| X(\hat{\beta} - \beta^0)\| _2^2 /(n\lambda)$ . Then from (3.4.1) we can derive the following string of inequalities: + +$$ +a + 2 (\| \hat {\beta} _ {N} \| _ {1} + \| \hat {\beta} _ {S} \| _ {1}) \leq \| \hat {\beta} _ {S} - \beta_ {S} ^ {0} \| _ {1} + \| \hat {\beta} _ {N} \| _ {1} + 2 \| \beta_ {S} ^ {0} \| _ {1} +$$ + +$$ +a + \| \hat {\beta} _ {N} \| _ {1} \leq \| \hat {\beta} _ {S} - \beta_ {S} ^ {0} \| _ {1} + 2 \| \beta_ {S} ^ {0} \| _ {1} - 2 \| \hat {\beta} _ {S} \| _ {1} +$$ + +$$ +a + \| \hat {\beta} _ {N} - \beta_ {N} ^ {0} \| _ {1} \leq 3 \| \beta_ {S} ^ {0} - \hat {\beta} _ {S} \| _ {1} +$$ + +$$ +a + \| \hat {\beta} - \beta^ {0} \| _ {1} \leq 4 \| \beta_ {S} ^ {0} - \hat {\beta} _ {S} \| _ {1}, +$$ + +the final inequality coming from adding $\| \beta_S^0 -\hat{\beta}_S\| _1$ to both sides. + +Now using the compatibility condition with $\beta = \hat{\beta} -\beta^0$ we have + +$$ +\frac {1}{n} \| X (\hat {\beta} - \beta^ {0}) \| _ {2} ^ {2} + \lambda \| \beta^ {0} - \hat {\beta} \| _ {1} \leq 4 \lambda \| \beta_ {S} ^ {0} - \hat {\beta} _ {S} \| _ {1} +$$ + +$$ +\leq \frac {4 \lambda}{\phi} \sqrt {\frac {s}{n}} \| X (\hat {\beta} - \beta^ {0}) \| _ {2}. \tag {3.4.2} +$$ + +From this we get + +$$ +\frac {1}{\sqrt {n}} \| X (\hat {\beta} - \beta^ {0}) \| _ {2} \leq \frac {4 \lambda \sqrt {s}}{\phi}, +$$ + +and substituting this into the RHS of (3.4.2) gives the result. + +![](images/2a35f2543cc17134ea9176787cb0ed853069c1d4e336e0c17d05a348f7cee302.jpg) + +## 3.5 The compatibility condition + +How strong is the compatibility condition? In order to answer this question, we shall think of X as random and try to understand what conditions on the population covariance matrix $\Sigma^{0} := \mathbb{E}(X^{T}X/n)$ imply that X satisfies a compatibility condition with high probability. To this end let us define + +$$ +\phi_ {\Sigma} ^ {2} (S) = \inf _ {\beta : \| \beta_ {S} \| _ {1} \neq 0, \| \beta_ {N} \| _ {1} \leq 3 \| \beta_ {S} \| _ {1}} \frac {\beta^ {T} \Sigma \beta}{\| \beta_ {S} \| _ {1} ^ {2} / | S |} +$$ + +where $\Sigma \in \mathbb{R}^{p\times p}$ . Note then our $\phi^2 = \phi_{\hat{\Sigma}}^2 (S)$ where $\hat{\Sigma} := X^T X / n$ and $S$ is the support set of $\beta^0$ . The following result shows that if $\hat{\Sigma}$ is close to a matrix $\check{\Sigma}$ for which $\phi_{\tilde{\Sigma}}^2 (S) > 0$ , then also $\phi_{\hat{\Sigma}}^2 (S) > 0$ . + +Lemma 22. Suppose $\phi_{\tilde{\Sigma}}^{2}(S) > 0$ and $\max_{jk}|\hat{\Sigma}_{jk} - \check{\Sigma}_{jk}|\leq \phi_{\tilde{\Sigma}}^{2}(S) / (32|S|)$ . Then $\phi_{\hat{\Sigma}}^{2}(S)\geq \phi_{\tilde{\Sigma}}^{2}(S) / 2$ . + +Proof. In the following we suppress dependence on $S$ . Let $s = |S|$ and let $t = \phi_{\Sigma}^2 / (32s)$ . We have + +$$ +| \beta^ {T} (\check {\Sigma} - \hat {\Sigma}) \beta | \leq \| \beta \| _ {1} \| (\check {\Sigma} - \hat {\Sigma}) \beta \| _ {\infty} (\text { Hölder }) +$$ + +$$ +\leq t \| \beta \| _ {1} ^ {2} (\text { Hölder again }). +$$ + +If $\| \beta_N\| _1\leq 3\| \beta_S\| _1$ then + +$$ +\| \beta \| _ {1} = \| \beta_ {N} \| _ {1} + \| \beta_ {S} \| _ {1} \leq 4 \| \beta_ {S} \| _ {1} \leq 4 \frac {\sqrt {\beta^ {T} \check {\Sigma} \beta}}{\phi_ {\check {\Sigma}} / \sqrt {s}}. +$$ + +Thus if $\| \beta_N\| _1\leq 3\| \beta_S\| _1$ + +$$ +\beta^ {T} \check {\Sigma} \beta - \frac {\phi_ {\check {\Sigma}} ^ {2}}{3 2 s} \frac {1 6 \beta^ {T} \check {\Sigma} \beta}{\phi_ {\check {\Sigma}} ^ {2} / s} = \frac {1}{2} \beta^ {T} \check {\Sigma} \beta \leq \beta^ {T} \hat {\Sigma} \beta . +$$ + +![](images/3a6c50ce0525db8181a7e8185ef7fe022ce516b3b33f5f68c57c241d91890925.jpg) + +We would like to apply the result above with $\check{\Sigma} = \Sigma^{0}$ , and use to argue that if $\Sigma^{0}$ satisfies the compatibility condition, then so will $\hat{\Sigma}$ with high probability. In order to do this, we need to argue that the event that $\max_{jk} |\hat{\Sigma}_{jk} - \Sigma_{jk}^{0}|$ is small occurs with high probability. We can obtain such a result with the aid of concentration inequalities. + +## 3.5.1 Concentration inequalities II + +When trying to understand the concentration properties of $\hat{\Sigma}_{jk}$ , it will be helpful to have a tail bound for a product of sub-Gaussian random variables. Bernstein's inequality, which applies to random variables satisfying the condition below, is helpful in this regard. + +Definition 6 (Bernstein's condition). We say that the random variable $W$ with $\mathbb{E}W = \mu$ satisfies Bernstein's condition with parameter $(\sigma, b)$ where $\sigma, b > 0$ if + +$$ +\mathbb {E} (| W - \mu | ^ {k}) \leq \frac {1}{2} k! \sigma^ {2} b ^ {k - 2} \quad \text { for } k = 2, 3, \ldots . +$$ + +Proposition 23 (Bernstein's inequality). Let $W_{1}, W_{2}, \ldots$ be independent random variables with $\mathbb{E}(W_{i}) = \mu$ . Suppose each $W_{i}$ satisfies Bernstein's condition with parameter $(\sigma, b)$ . Then + +$$ +\mathbb {E} \big (e ^ {\alpha (W _ {i} - \mu)} \big) \leq \exp \left(\frac {\alpha^ {2} \sigma^ {2} / 2}{1 - b | \alpha |}\right) \quad f o r a l l | \alpha | < 1 / b +$$ + +$$ +\mathbb {P} \bigg (\frac {1}{n} \sum_ {i = 1} ^ {n} W _ {i} - \mu \geq t \bigg) \leq \exp \bigg (- \frac {n t ^ {2}}{2 (\sigma^ {2} + b t)} \bigg) \qquad f o r a l l t \geq 0. +$$ + +Proof. Fix $i$ and let $W = W_{i}$ . We have + +$$ +\mathbb {E} \left(e ^ {\alpha (W - \mu)}\right) = 1 + \alpha \mathbb {E} (W - \mu) + \sum_ {k = 2} ^ {\infty} \alpha^ {k} \frac {\mathbb {E} \left\{\left(W - \mu\right) ^ {k} \right\}}{k !} +$$ + +$$ +\leq 1 + \frac {\sigma^ {2} \alpha^ {2}}{2} \sum_ {k = 2} ^ {\infty} | \alpha | ^ {k - 2} b ^ {k - 2} +$$ + +$$ += 1 + \frac {\sigma^ {2} \alpha^ {2}}{2} \frac {1}{1 - | \alpha | b} \leq \exp \left(\frac {\alpha^ {2} \sigma^ {2} / 2}{1 - b | \alpha |}\right), +$$ + +provided $|\alpha| < 1/b$ and using the inequality $e^{u} \geq 1 + u$ in the final line. For the probability bound, first note that + +$$ +\mathbb {E} \exp \left(\sum_ {i = 1} ^ {n} \alpha (W _ {i} - \mu) / n\right) = \prod_ {i = 1} ^ {n} \mathbb {E} \exp \{\alpha (W _ {i} - \mu) / n \} +$$ + +$$ +\leq \exp \left(n \frac {(\alpha / n) ^ {2} \sigma^ {2} / 2}{1 - b | \alpha / n |}\right) +$$ + +for $|\alpha| / n < 1 / b$ . Then we use the Chernoff method and set $\alpha / n = t / (bt + \sigma^2) \in [0, 1 / b)$ . + +Lemma 24. Let W, Z be mean-zero and sub-Gaussian with parameters $\sigma_{W}$ and $\sigma_{Z}$ respectively. Then the product WZ satisfies Bernstein's condition with parameter $(8\sigma_{W}\sigma_{Z}, 4\sigma_{W}\sigma_{Z})$ . + +Proof. In order to use Bernstein's inequality (Proposition 23) we first obtain bounds on the moments of $W$ and $Z$ . Note that $W^{2k} = \int_0^\infty \mathbb{1}_{\{x < W^{2k}\}} dx$ . Thus by Fubini's theorem + +$$ +\begin{array}{l} \mathbb {E} (W ^ {2 k}) = \int_ {0} ^ {\infty} \mathbb {P} (W ^ {2 k} > x) d x \\ = 2 k \int_ {0} ^ {\infty} t ^ {2 k - 1} \mathbb {P} (| W | > t) d t \quad \text { substituting } t ^ {2 k} = x \\ \leq 4 k \int_ {0} ^ {\infty} t ^ {2 k - 1} \exp \{- t ^ {2} / (2 \sigma_ {W} ^ {2}) \} d t \qquad \text { by Proposition 12 } \\ = 4 k \sigma_ {W} ^ {2} \int_ {0} ^ {\infty} (2 \sigma_ {W} ^ {2} x) ^ {k - 1} e ^ {- x} d x \qquad \text {substituting} t ^ {2} / (2 \sigma_ {W} ^ {2}) = x \\ = 2 ^ {k + 1} \sigma_ {W} ^ {2 k} k!. \\ \end{array} +$$ + +Next note that for any random variable Y, + +$$ +\begin{array}{l} \left| \mathbb {E} (Y - \mathbb {E} Y) ^ {k} \right| \leq \mathbb {E} | Y - \mathbb {E} Y | ^ {k} \\ = 2 ^ {k} \mathbb {E} | Y / 2 - \mathbb {E} Y / 2 | ^ {k} \\ \leq 2 ^ {k - 1} (\mathbb {E} | Y | ^ {k} + | \mathbb {E} Y | ^ {k}) \quad \text { by Jensen's inequality applied to } t \mapsto | t | ^ {k}, \\ \leq 2 ^ {k} \mathbb {E} | Y | ^ {k}. \\ \end{array} +$$ + +Therefore + +$$ +\begin{array}{l} \mathbb {E} (| W Z - \mathbb {E} W Z | ^ {k}) \leq 2 ^ {k} \mathbb {E} | W Z | ^ {k} \\ \leq 2 ^ {k} (\mathbb {E} W ^ {2 k}) ^ {1 / 2} (\mathbb {E} Z ^ {2 k}) ^ {1 / 2} \quad \text { by Cauchy - Schwarz } \\ \leq 2 ^ {k} 2 ^ {k + 1} \sigma_ {W} ^ {k} \sigma_ {Z} ^ {k} k! \\ = \frac {k !}{2} (8 \sigma_ {W} \sigma_ {Z}) ^ {2} (4 \sigma_ {W} \sigma_ {Z}) ^ {k - 2}. \\ \end{array} +$$ + +## 3.5.2 The compatibility condition and random design + +We may now apply this with $\check{\Sigma} = \Sigma^{0}$ . To make the result more readily interpretable, we shall state it in an asymptotic framework. Imagine a sequence of design matrices with n and p growing, each with their own compatibility condition. We will however suppress the asymptotic regime in the notation. + +Theorem 25. Suppose the rows of X are i.i.d. and each entry of X is mean-zero sub-Gaussian with parameter v. Suppose $s\sqrt{\log(p)/n} \to 0$ (and $s, p, n > 1$ ) as $n \to \infty$ . Let + +$$ +\phi_ {\hat {\Sigma}, s} ^ {2} = \min _ {S: | S | = s} \phi_ {\hat {\Sigma}} ^ {2} (S) +$$ + +$$ +\phi_ {\Sigma^ {0}, s} ^ {2} = \min _ {S: | S | = s} \phi_ {\Sigma^ {0}} ^ {2} (S), +$$ + +and suppose the latter is bounded away from 0. Then $\mathbb{P}(\phi_{\hat{\Sigma},s}^2\geq \phi_{\Sigma^0,s}^2 /2)\to 1$ as $n\to \infty$ . + +Proof. In view of Lemma 22, we need only show that + +$$ +\mathbb {P} (\max _ {j k} | \hat {\Sigma} _ {j k} - \Sigma_ {j k} ^ {0} | \geq \phi_ {\Sigma^ {0}, s} ^ {2} / (3 2 s)) \to 0. +$$ + +Let $t = \phi_{\Sigma^0,s}^2 / (32s)$ . By a union bound and then Lemma 24 we have + +$$ +\begin{array}{l} \mathbb {P} (\max _ {j k} | \hat {\Sigma} _ {j k} - \Sigma_ {j k} ^ {0} | \geq \phi_ {\Sigma^ {0}, s} ^ {2} / (3 2 s)) < p ^ {2} \max _ {j k} \mathbb {P} \Big (\Big | \sum_ {i = 1} ^ {n} X _ {i j} X _ {i k} / n - \Sigma_ {j k} ^ {0} \Big | \geq t \Big) \\ \leq 2 \exp \left(- \frac {n t ^ {2}}{2 (6 4 v ^ {4} + 4 v ^ {2} t))} + 2 \log (p)\right) \\ \leq c _ {1} \exp (- c _ {2} n / s ^ {2} + c _ {3} \log (p)) \to 0. \\ \end{array} +$$ + +Corollary 26. Suppose the rows of X are independent with distribution $N_{p}(0,\Sigma^{0})$ . Suppose the diagonal entries of $\Sigma^{0}$ are bounded above and the minimum eigenvalue of $\Sigma^{0}$ , $c_{min}$ is bounded away from 0. Then $\mathbb{P}(\phi_{\Sigma,s}^{2}\geq c_{\min}/2)\to1$ provided $s\sqrt{\log(p)/n}\to0$ . + +## 3.6 Computation + +One of the most efficient ways of computing Lasso solutions is to use a optimisation technique called coordinate descent. This is a quite general way of minimising a function $f : R^{d} \to R$ and works particularly well for functions of the form + +$$ +f (x) = g (x) + \sum_ {j = 1} ^ {d} h _ {j} (x _ {j}) +$$ + +where g is convex and differentiable and each $h_{j}: R \to R$ is convex (and so continuous). We start with an initial guess of the minimiser $x^{(0)}$ (e.g. $x^{(0)} = 0$ ) and repeat for $m = 1, 2, \ldots$ . + +$$ +x _ {1} ^ {(m)} = \underset {x _ {1} \in \mathbb {R}} {\arg \min} f (x _ {1}, x _ {2} ^ {(m - 1)}, \ldots , x _ {d} ^ {(m - 1)}) +$$ + +$$ +x _ {2} ^ {(m)} = \underset {x _ {2} \in \mathbb {R}} {\arg \min} f (x _ {1} ^ {(m)}, x _ {2}, x _ {3} ^ {(m - 1)}, \ldots , x _ {d} ^ {(m - 1)}) +$$ + +$$ +\begin{array}{c} \vdots \\ x _ {d} ^ {(m)} = \underset {x _ {d} \in \mathbb {R}} {\arg \min} f (x _ {1} ^ {(m)}, x _ {2} ^ {(m)}, \ldots , x _ {d - 1} ^ {(m)}, x _ {d}). \end{array} +$$ + +Tseng [2001] proves that provided $A_{0} = \{x : f(x) \leq f(x^{(0)})\}$ is compact, then every converging subsequence of $x^{(m)}$ will converge to a minimiser of f. + +Corollary 27. Suppose $A^{0}$ is compact. Then + +(i) There exists a minimiser of $f$ , $x^{*}$ and $f(x^{(m)}) \to f(x^{*})$ . + +(ii) If $x^{*}$ is the unique minimiser of $f$ then $x^{(m)} \to x^{*}$ . + +Proof. $f$ is a continuous function so it attains its infimum on the compact set $A_0$ . Suppose $f(x^{(m)}) \nrightarrow f(x^*)$ . Then there exists $\epsilon > 0$ and a subsequence $(x^{(m_j)})_{j=0}^{\infty}$ such that $f(x^{(m_j)}) \geq f(x^*) + \epsilon$ for all $j$ . Note that since $f(x^{(m)}) \leq f(x^{(m-1)})$ , we know that $x^{(m)} \in A_0$ for all $m$ . Thus if $A_0$ is compact then any subsequence of $(x^{(m)})_{m=0}^{\infty}$ has a further subsequence that converges by the Bolzano–Weierstrass theorem. Let $\tilde{x}$ be the limit of the converging subsequence of $(x^{(m_j)})_{j=0}^{\infty}$ . Then $f(\tilde{x}) \geq f(x^*) + \epsilon$ , contradicting the result of Tseng [2001]. Thus (i) holds. The proof of (ii) is similar. + +We often want to solve the Lasso on a grid of $\lambda$ values $\lambda_{0} > \cdots > \lambda_{L}$ (for the purposes of cross-validation for example). To do this, we can first solve for $\lambda_{0}$ , and then solve at subsequent grid points by using the solution at the previous grid points as an initial guess (known as a warm start). An active set strategy can further speed up computation. This works as follows: For $l = 1, \ldots, L$ + +1. Initialise $A = \{k : \hat{\beta}_{\lambda_{l-1}, k}^{\mathrm{L}} \neq 0\}$ . +2. Perform coordinate descent only on coordinates in $A$ obtaining a solution $\hat{\beta}$ (all components $\hat{\beta}_k$ with $k \notin A$ are set to zero). +3. Let $V = \{k : |X_{k}^{T}(Y - X\hat{\beta})|/n > \lambda_{l}\}$ , the set of coordinates which violate the KKT conditions when $\hat{\beta}$ is taken as a candidate solution. +4. If $V$ is empty, we set $\hat{\beta}_{\lambda_l}^{\mathrm{L}} = \hat{\beta}$ . Else we update $A \gets A \cup V$ and return to 2. + +## 3.7 Extensions of the Lasso + +We can add an $\ell_{1}$ penalty to many other log-likelihoods, or more generally other loss functions besides the squared-error loss that arises from the normal linear model. For Lasso-penalised generalised linear models, such as logistic regression, similar theoretical results to those we have obtained are available and computations can proceed in a similar fashion to above. + +## 3.7.1 Structural penalties + +The Lasso penalty encourages the estimated coefficients to be shrunk towards 0 and sometimes exactly to 0. Other penalty functions can be constructed to encourage different types of sparsity. + +## Group Lasso + +Suppose we have a partition $G_{1},\ldots ,G_{q}$ of $\{1,\dots ,p\}$ (so $\cup_{k = 1}^{q}G_{k} = \{1,\dots ,p\}$ , $G_{j}\cap G_{k} = \emptyset$ for $j\neq k$ ). The group Lasso penalty [Yuan and Lin, 2006] is given by + +$$ +\lambda \sum_ {j = 1} ^ {q} m _ {j} \| \beta_ {G _ {j}} \| _ {2}. +$$ + +The multipliers $m_{j} > 0$ serve to balance cases where the groups are of very different sizes; typically we choose $m_{j} = \sqrt{|G_{j}|}$ . This penalty encourages either an entire group G to have $\hat{\beta}_{G} = 0$ or $\hat{\beta}_{k} \neq 0$ for all $k \in G$ . Such a property is useful when groups occur through coding for categorical predictors or when expanding predictors using basis functions. + +## Fused Lasso + +If there is a sense in which the coefficients are ordered, so $\beta_{j}^{0}$ is expected to be close to $\beta_{j+1}^{0}$ , a fused Lasso penalty [Tibshirani et al., 2005] may be appropriate. This takes the form + +$$ +\lambda_ {1} \sum_ {j = 1} ^ {p - 1} | \beta_ {j} - \beta_ {j + 1} | + \lambda_ {2} \| \beta \| _ {1}, +$$ + +where the second term may be omitted depending on whether shrinkage towards 0 is desired. As an example, consider the simple setting where $Y_{i} = \mu_{i}^{0} + \varepsilon_{i}$ , and it is thought that the $(\mu_{i}^{0})_{i=1}^{n}$ form a piecewise constant sequence. Then one option is to minimise over $\mu \in R^{n}$ , the following objective + +$$ +\frac {1}{n} \| Y - \mu \| _ {2} ^ {2} + \lambda \sum_ {i = 1} ^ {n - 1} | \mu_ {i} - \mu_ {i + 1} |. +$$ + +## 3.7.2 Reducing the bias of the Lasso + +One potential drawback of the Lasso is that the same shrinkage effect that sets many estimated coefficients exactly to zero also shrinks all non-zero estimated coefficients towards zero. One possible solution is to take $\hat{S}_{\lambda} = \{k : \hat{\beta}_{\lambda,k}^{\mathrm{L}} \neq 0\}$ and then re-estimate $\beta_{\hat{S}_{\lambda}}^{0}$ by OLS regression on $X_{\hat{S}_{\lambda}}$ . + +Another option is to re-estimate using the Lasso on $X_{\hat{S}_{\lambda}}$ ; this procedure is known as the relaxed Lasso [Meinshausen, 2007]. The adaptive Lasso [Zou, 2006] takes an initial estimate of $\beta^{0}$ , $\hat{\beta}^{\mathrm{init}}$ (e.g. from the Lasso) and then performs weighted Lasso regression: + +$$ +\hat {\beta} _ {\lambda} ^ {\mathrm{adapt}} = \underset {\beta \in \mathbb {R} ^ {p}: \beta_ {\hat {S} _ {\mathrm{init}} ^ {c}} = 0} {\arg \min} \left\{\frac {1}{2 n} \| Y - X \beta \| _ {2} ^ {2} + \lambda \sum_ {k \in \hat {S} _ {\mathrm{init}}} \frac {| \beta_ {k} |}{| \hat {\beta} _ {k} ^ {\mathrm{init}} |} \right\}, +$$ + +where $\hat{S}_{\mathrm{init}} = \{k:\hat{\beta}_{k}^{\mathrm{init}}\neq 0\}$ . + +Yet another approach involves using a family of non-convex penalty functions $p_{\lambda,\gamma} : [0, \infty) \to [0, \infty)$ and attempting to minimise + +$$ +\frac {1}{2 n} \| Y - X \beta \| _ {2} ^ {2} + \sum_ {k = 1} ^ {p} p _ {\lambda , \gamma} (| \beta_ {k} |). +$$ + +A prominent example is the minimax concave penalty (MCP) [Zhang, 2010] which takes + +$$ +p _ {\lambda , \gamma} ^ {\prime} (u) = \left(\lambda - \frac {u}{\gamma}\right) _ {+}. +$$ + +One disadvantage of using a non-convex penalty is that there may be multiple local minima which can make optimisation problematic. However, typically if the non-convexity is not too severe, coordinate descent can produce reasonable results. + +## Chapter 4 + +# Graphical modelling and causal inference + +So far we have considered the problem of relating a particular response to a large collection of explanatory variables. + +In some settings however, we do not have a distinguished response variable and instead we would like to better understand relationships between all the variables. In other situations, rather than being able to predict variables, we would like to understand causal relationships between them. Representing relationships between random variables through graphs will be an important tool in tackling these problems. + +## 4.1 Graphs + +Definition 7. A graph is a pair $\mathcal{G} = (V, E)$ where $V$ is a set of vertices or nodes and $E \subseteq V \times V$ with $(v, v) \notin E$ for any $v \in V$ is a set of edges. + +Let $Z = (Z_{1},\ldots ,Z_{p})^{T}$ be a collection of random variables. The graphs we will consider will always have $V = \{1,\dots ,p\}$ so $V$ indexes the random variables. + +Let $j, k \in V$ . + +- We say there is an edge between $j$ and $k$ and that $j$ and $k$ are adjacent if either $(j, k) \in E$ or $(k, j) \in E$ . +- An edge $(j,k)$ is undirected if also $(k,j) \in E$ . We then sometimes write $\{j,k\} \in E$ and $j - k$ to represent this. Otherwise if only $(j,k) \in E$ the edge is directed and we may write $j \to k$ to represent this. +- If all edges in the graph are (un)directed we call it an (un)directed graph. We can represent graphs as pictures: for example, we can draw the graph when $p = 4$ and $E = \{(2,1), (3,4), (2,3)\}$ as + +![](images/f2196179f022b603ca647361198c41d590437ca4b0b378e7ec80ca0f0374f571.jpg) + +
    +flowchart + +```mermaid +graph TD + Z1 --> Z2 + Z2 --> Z3 + Z3 --> Z4 +``` +
    + +If instead we have $E = \{\{1,2\}, \{2,4\}\}$ we get the undirected graph + +![](images/0b524b3eb1a572a3d5dfe20617e5c2199ec7bf9ddab29f6a3999eab0e707db4e.jpg) + +
    +flowchart + +```mermaid +graph TD + Z1["Z₁"] --> Z2["Z₂"] + Z3["Z₃"] --> Z4["Z₄"] +``` +
    + +- A graph $\mathcal{G}_1 = (V_1, E_1)$ is a subgraph of $\mathcal{G} = (V, E)$ if $V_1 \subseteq V$ and $E_1 \subseteq E$ and a proper subgraph if either of these are proper inclusions. +- Say $j$ is a parent of $k$ and $k$ is a child of $j$ if $j \to k$ . The sets of parents and children of $k$ will be denoted $\mathrm{pa}(k)$ and $\mathrm{ch}(k)$ respectively. +- A set of three nodes is called a v-structure if one node is a child of the two other nodes, and these two nodes are not adjacent. +- The skeleton of $\mathcal{G}$ is a copy of $\mathcal{G}$ with every edge replaced by an undirected edge. +- A path from $j$ to $k$ is a sequence $j = j_1, j_2, \ldots, j_m = k$ of (at least two) distinct vertices such that $j_l$ and $j_{l+1}$ are adjacent. Such a path is a directed path if $j_l \to j_{l+1}$ for all $l$ . We then call $k$ a descendant of $j$ . The set of descendants of $j$ will be denoted $\mathrm{de}(j)$ . If $j_{l-1} \to j_l \leftarrow j_{l+1}$ , $j_l$ is called a collider (relative to the path). +- A directed cycle is (almost) a directed path but with the start and end points the same. A partially directed acyclic graph (PDAG) is a graph containing no directed cycles. A directed acyclic graph (DAG) is a directed graph containing no directed cycles. +- In a DAG, a path between $j_1$ and $j_m$ ( $j_1, j_2, \ldots, j_m$ ) is blocked by a set $S$ with neither $j_1$ nor $j_m$ in $S$ whenever there is a node $j_l$ such that one of the following two possibilities hold: + +1. $j_l \in S$ and we don't have $j_{l-1} \to j_l \leftarrow j_{l+1}$ +2. $j_{l-1} \rightarrow j_l \leftarrow j_{l+1}$ and neither $j_l$ nor any of its descendants are in S. + +\- If $\mathcal{G}$ is a DAG, given a triple of subsets of nodes $A, B, S$ , we say $S$ $d$ -separates $A$ from $B$ if $S$ blocks every path from $A$ to $B$ . + +\- The moralised graph of a DAG $\mathcal{G}$ is the undirected graph obtained by adding edges between (marrying) the parents of each node and removing all edge directions. + +Definition 8. Given a DAG $\mathcal{G}$ with $V = \{1, \ldots, p\}$ , we say that a permutation $\pi$ of $V$ is a topological (or causal) ordering of the variables if it satisfies + +$$ +\pi (j) < \pi (k) \qquad \mathrm{whenever} k \in \mathrm{de} (j). +$$ + +Proposition 28. Every DAG has a topological ordering. + +Proof. We use induction on the number of nodes p. Clearly the result is true when p = 1. + +Now we show that in any DAG, we can find a node with no parents. Pick any node and move to one of its parents, if possible. Then move to one of the new node's parents, and continue in this fashion. This procedure must terminate since no node can be visited twice, or we would have found a cycle. The final node we visit must therefore have no parents, which we call a source node. + +Suppose then that $p \geq 2$ , and we know that all DAGs with p-1 nodes have a topological ordering. Find a source s (wlog s = p) and form a new DAG $\tilde{G}$ with p-1 nodes by removing the source (and all edges emanating from it). Note we keep the labelling of the nodes in this new DAG the same. This smaller DAG must have a topological order $\tilde{\pi}$ . A topological ordering $\pi$ for our original DAG is then given by $\pi(s) = 1$ and $\pi(k) = \tilde{\pi}(k) + 1$ for $k \neq s$ . ☐ + +## 4.2 Conditional independence graphs + +We would like to understand which variables may be ‘related’ to each other. Trying to find pairs of variables that are independent and so unlikely to be related to each other is not necessarily a good way to proceed as each variable may be correlated with a large number of variables without being directly related to them. A better approach is to use conditional independence. + +Definition 9. If X, Y and Z are random vectors with a joint density $f_{XYZ}$ (w.r.t. a product measure $\mu$ ) then we say X is conditionally independent of Y given Z, and write + +$$ +X \perp Y | Z +$$ + +if + +$$ +f _ {X Y | Z} (x, y | z) = f _ {X | Z} (x | z) f _ {Y | Z} (y | z). +$$ + +Equivalently + +$$ +X \perp \perp Y | Z \Longleftrightarrow f _ {X | Y Z} (x | y, z) = f _ {X | Z} (x | z). +$$ + +We will first look at how undirected graphs can be used to visualise conditional independencies between random variables; thus in the next few subsections by graph we will mean undirected graph. + +Definition 10. Let $Z = (Z_{1}, \ldots, Z_{p})^{T}$ be a collection of random variables with joint law P. The conditional independence graph (CIG) for P is the graph with $V = \{1, \ldots, p\}$ and an edge $\{j, k\}$ between j and k if and only if + +$$ +Z _ {j} \nparallel Z _ {k} \mid Z _ {- j k}. +$$ + +## 4.3 Gaussian graphical models + +Estimating the CIG given samples from P is a difficult task in general. However, in the case where P is multivariate Gaussian, things simplify considerably as we shall see. We begin with some notation. For a matrix $M \in R^{p \times p}$ , and sets $A, B \subseteq \{1, \ldots, p\}$ , let $M_{A,B}$ be the $|A| \times |B|$ submatrix of M consisting of those rows and columns of M indexed by the sets A and B respectively. The submatrix extraction operation is always performed first (so e.g. $M_{k,-k}^{T} = (M_{k,-k})^{T}$ ). + +## 4.3.1 Normal conditionals + +Now let $Z \sim N_p(\mu, \Sigma)$ with $\Sigma$ positive definite. Note $\Sigma_{A,A}$ is also positive definite for any $A$ . + +Proposition 29. + +$$ +Z _ {A} | Z _ {B} = z _ {B} \sim N _ {| A |} (\mu_ {A} + \Sigma_ {A, B} \Sigma_ {B, B} ^ {- 1} (z _ {B} - \mu_ {B}), \Sigma_ {A, A} - \Sigma_ {A, B} \Sigma_ {B, B} ^ {- 1} \Sigma_ {B, A}) +$$ + +Proof. Idea: write $Z_{A}=MZ_{B}+(Z_{A}-MZ_{B})$ with matrix $M\in R^{|A|\times|B|}$ such that $Z_{A}-MZ_{B}$ and $Z_{B}$ are independent, i.e. such that + +$$ +\operatorname{Cov} (Z _ {B}, Z _ {A} - M Z _ {B}) = \Sigma_ {B, A} - \Sigma_ {B, B} M ^ {T} = 0. +$$ + +This occurs when we take $M^{T} = \Sigma_{B,B}^{-1}\Sigma_{B,A}$ . Because $Z_{A} - MZ_{B}$ and $Z_{B}$ are independent, the distribution of $Z_{A} - MZ_{B}$ conditional on $Z_{B} = z_{B}$ is equal to its unconditional distribution. Now + +$$ +\begin{array}{l} \mathbb {E} (Z _ {A} - M Z _ {B}) = \mu_ {A} - \Sigma_ {A, B} \Sigma_ {B, B} ^ {- 1} \mu_ {B} \\ \mathrm{Var} (Z _ {A} - M Z _ {B}) = \Sigma_ {A, A} + \Sigma_ {A, B} \Sigma_ {B, B} ^ {- 1} \Sigma_ {B, B} \Sigma_ {B, B} ^ {- 1} \Sigma_ {B, A} - 2 \Sigma_ {A, B} \Sigma_ {B, B} ^ {- 1} \Sigma_ {B, A} \\ = \Sigma_ {A, A} - \Sigma_ {A, B} \Sigma_ {B, B} ^ {- 1} \Sigma_ {B, A}. \\ \end{array} +$$ + +Since $MZ_{B}$ is a function of $Z_{B}$ and $Z_{A}-MZ_{B}$ is normally distributed, we have the result. ☐ + +## 4.3.2 Nodewise regression + +Specialising to the case where $A = \{k\}$ and $B = A^{c}$ we see that when conditioning on $Z_{-k} = z_{-k}$ , we may write + +$$ +Z _ {k} = m _ {k} + z _ {- k} ^ {T} \Sigma_ {- k, - k} ^ {- 1} \Sigma_ {- k, k} + \varepsilon_ {k}, +$$ + +where + +$$ +\begin{array}{l} m _ {k} = \mu_ {k} - \Sigma_ {k, - k} \Sigma_ {- k, - k} ^ {- 1} \mu_ {- k} \\ \varepsilon_ {k} | Z _ {- k} = z _ {- k} \sim N (0, \Sigma_ {k, k} - \Sigma_ {k, - k} \Sigma_ {- k, - k} ^ {- 1} \Sigma_ {- k, k}). \\ \end{array} +$$ + +Note that if the $j$ th element of the vector of coefficients $\Sigma_{-k,-k}^{-1}\Sigma_{-k,k}$ is zero, then the distribution of $Z_{k}$ conditional on $Z_{-k}$ will not depend at all on the $j$ th component of $Z_{-k}$ . Then if that $j$ th component was $Z_{j'}$ , we would have that $Z_{k}|Z_{-k}=z_{-k}$ has the same distribution as $Z_{k}|Z_{-j'k}=z_{-j'k}$ , so $Z_{k}\perp Z_{j}|Z_{-j'k}$ . + +Thus given $x_{1},\ldots,x_{n}\stackrel{\mathrm{i.i.d.}}{\sim}Z$ and writing + +$$ +X = \left( \begin{array}{c} x _ {1} ^ {T} \\ \vdots \\ x _ {n} ^ {T} \end{array} \right), +$$ + +we may estimate the coefficient vector $\Sigma_{-k,-k}^{-1}\Sigma_{-k,k}$ by regressing $X_{k}$ on $X_{\{k\}^{c}}$ and including an intercept term. + +The technique of neighbourhood selection [Meinshausen and Bühlmann, 2006] involves performing such a regression for each variable, using the Lasso. There are two options for populating our estimate of the CIG with edges based on the Lasso estimates. Writing $\hat{S}_k$ for the selected set of variables when regressing $X_{k}$ on $X_{\{k\}^c}$ , we can use the "OR" rule and put an edge between vertices $j$ and $k$ if and only if $k \in \hat{S}_j$ or $j \in \hat{S}_k$ . An alternative is the "AND" rule where we put an edge between $j$ and $k$ if and only if $k \in \hat{S}_j$ and $j \in \hat{S}_k$ . + +Another popular approach to estimating the CIG works by first directly estimating $\Omega$ , as we'll now see. + +## 4.3.3 The precision matrix and conditional independence + +The following facts about blockwise inversion of matrices will help us to interpret the mean and variance in Proposition 29. + +Proposition 30. Let $M \in R^{p \times p}$ be a symmetric positive definite matrix and suppose + +$$ +M = \left( \begin{array}{c c} P & Q ^ {T} \\ Q & R \end{array} \right) +$$ + +with P and R square matrices. The Schur complement of R is $P - Q^{T}R^{-1}Q =: S$ . We have that S is positive definite and + +$$ +M ^ {- 1} = \left( \begin{array}{c c} S ^ {- 1} & - S ^ {- 1} Q ^ {T} R ^ {- 1} \\ - R ^ {- 1} Q S ^ {- 1} & R ^ {- 1} + R ^ {- 1} Q S ^ {- 1} Q ^ {T} R ^ {- 1} \end{array} \right). +$$ + +Furthermore $\det(M) = \det(S)\det(R)$ . + +Let $\Omega = \Sigma^{-1}$ be the precision matrix. Note that $\Sigma_{k,k} - \Sigma_{k,-k}\Sigma_{-k,-k}^{-1}\Sigma_{-k,k} = \Omega_{kk}^{-1}$ , and more generally that $\mathrm{Var}(Z_A|Z_{A^c}) = \Omega_{A,A}^{-1}$ . Also, we see that $\Sigma_{-k,-k}^{-1}\Sigma_{-k,k} = -\Omega_{kk}^{-1}\Omega_{-k,k}$ , so + +$$ +(\Sigma_ {- k, - k} ^ {- 1} \Sigma_ {- k, k}) _ {j} = 0 \Leftrightarrow \left\{ \begin{array}{l l} \Omega_ {j, k} = 0 \text {for} j < k \\ \Omega_ {j + 1, k} = 0 \text {for} j \geq k. \end{array} \right. +$$ + +Thus + +$$ +Z _ {k} \perp \perp Z _ {j} | Z _ {- j k} \Leftrightarrow \Omega_ {j k} = 0. +$$ + +This motivates another approach to estimating the CIG. + +## 4.3.4 The Graphical Lasso + +Recall that the density of $N_{p}(\mu ,\Sigma)$ is + +$$ +f (z) = \frac {1}{(2 \pi) ^ {p / 2} \mathrm{det} (\Sigma) ^ {1 / 2}} \exp \bigg (- \frac {1}{2} (z - \mu) ^ {T} \Sigma^ {- 1} (z - \mu) \bigg). +$$ + +The log-likelihood of $(\mu, \Sigma)$ based on an i.i.d. sample $x_{1}, \ldots, x_{n}$ is + +$$ +\ell (\mu , \Omega) = \frac {n}{2} \log \det (\Omega) - \frac {1}{2} \sum_ {i = 1} ^ {n} (x _ {i} - \mu) ^ {T} \Omega (x _ {i} - \mu). +$$ + +Write + +$$ +\bar {X} = \frac {1}{n} \sum_ {i = 1} ^ {n} x _ {i}, \qquad S = \frac {1}{n} \sum_ {i = 1} ^ {n} (x _ {i} - \bar {X}) (x _ {i} - \bar {X}) ^ {T}. +$$ + +Then + +$$ +\begin{array}{l} \sum_ {i = 1} ^ {n} (x _ {i} - \mu) ^ {T} \Omega (x _ {i} - \mu) = \sum_ {i = 1} ^ {n} (x _ {i} - \bar {X} + \bar {X} - \mu) ^ {T} \Omega (x _ {i} - \bar {X} + \bar {X} - \mu) \\ = \sum_ {i = 1} ^ {n} (x _ {i} - \bar {X}) ^ {T} \Omega (x _ {i} - \bar {X}) + n (\bar {X} - \mu) ^ {T} \Omega (\bar {X} - \mu) \\ + 2 \sum_ {i = 1} ^ {n} (x _ {i} - \bar {X}) ^ {T} \Omega (\bar {X} - \mu). \\ \end{array} +$$ + +Also, + +$$ +\begin{array}{l} \sum_ {i = 1} ^ {n} (x _ {i} - \bar {X}) ^ {T} \Omega (x _ {i} - \bar {X}) = \sum_ {i = 1} ^ {n} \mathrm{tr} \{(x _ {i} - \bar {X}) ^ {T} \Omega (x _ {i} - \bar {X}) \} \\ = \sum_ {i = 1} ^ {n} \operatorname{tr} \left\{\left(x _ {i} - \bar {X}\right) \left(x _ {i} - \bar {X}\right) ^ {T} \Omega \right\} \\ = n \operatorname{tr} (S \Omega). \\ \end{array} +$$ + +Thus + +$$ +\ell (\mu , \Omega) = - \frac {n}{2} \{\mathrm{tr} (S \Omega) - \log \det (\Omega) + (\bar {X} - \mu) ^ {T} \Omega (\bar {X} - \mu) \} +$$ + +and + +$$ +\max _ {\mu \in \mathbb {R} ^ {p}} \ell (\mu , \Omega) = - \frac {n}{2} \{\mathrm{tr} (S \Omega) - \log \det (\Omega) \}. +$$ + +Hence the maximum likelihood estimate of $\Omega$ , $\hat{\Omega}^{ML}$ can be obtained by solving + +$$ +\min _ {\Omega : \Omega \succ 0} \{- \log \det (\Omega) + \operatorname{tr} (S \Omega) \}, +$$ + +where $\Omega \succ 0$ means $\Omega$ is positive definite. One can show that the objective is convex and we are minimising over a convex set. As + +$$ +\begin{array}{l} \frac {\partial}{\partial \Omega_ {j k}} \log \det (\Omega) = (\Omega^ {- 1}) _ {k j} = (\Omega^ {- 1}) _ {j k}, \\ \frac {\partial}{\partial \Omega_ {j k}} \mathrm{tr} (S \Omega) = S _ {k j} = S _ {j k}, \\ \end{array} +$$ + +if $X$ has full column rank so $S$ is positive definite, $\hat{\Omega}^{ML} = S^{-1}$ . + +The graphical Lasso [Yuan and Lin, 2007] penalises the log-likelihood for $\Omega$ and solves + +$$ +\min _ {\Omega : \Omega \succ 0} \{- \log \det (\Omega) + \operatorname{tr} (S \Omega) + \lambda \| \Omega \| _ {1} \}, +$$ + +where $\|\Omega\|_{1}=\sum_{j,k}|\Omega_{jk}|$ ; this results in a sparse estimate of the precision matrix from which an estimate of the CIG can be constructed. Often the $\|\Omega\|_{1}$ is modified such that the diagonal elements are not penalised. + +## 4.4 Structural equation models + +Conditional independence graphs give us some understanding of the relationships between variables. However they do not tell us how, if we were to set the kth variable to a particular value, say 0.5, then how the distribution of the other values would be altered. Yet this is often the sort of question that we would like to answer. + +In order to reach this more ambitious goal, we introduce the notion of structural equation models (SEMs). These give a way of representing the data generating process. We will now have to make use of not just undirected graphs but other sorts of graphs (and particularly DAGs), so by graph we will now mean any sort of graph satisfying definition 7. + +Definition 11. A structural equation model S for a random vector $Z \in R^{p}$ is a collection of p equations + +$$ +Z _ {k} = h _ {k} (Z _ {P _ {k}}, \varepsilon_ {k}), \qquad k = 1, \ldots , p +$$ + +where + +- $\varepsilon_1, \ldots, \varepsilon_p$ are all independent random variables; +- $P_k \subseteq \{1, \ldots, p\} \setminus \{k\}$ are such that the graph with edges given by $P_k$ being $\text{pa}(k)$ is a DAG. + +Example 4.4.1. Consider the following (totally artificial) SEM which has whether you are taking this course $(Z_{1}=1)$ depending on whether you went to the statistical modelling course $(Z_{2}=1)$ and whether you have heard about machine learning $(Z_{3}=1)$ . Suppose + +$$ +\begin{array}{l} Z _ {3} = \varepsilon_ {3} \sim \mathrm{Bern} (1 / 4) \\ Z _ {2} = \mathbb {1} _ {\{\varepsilon_ {2} (1 + Z _ {3}) > 1 / 2 \}} \qquad \varepsilon_ {2} \sim U [ 0, 1 ] \\ Z _ {1} = \mathbb {1} _ {\{\varepsilon_ {1} (Z _ {2} + Z _ {3}) > 1 / 2 \}} \qquad \varepsilon_ {1} \sim U [ 0, 1 ]. \\ \end{array} +$$ + +The corresponding DAG is + +![](images/7d319b73dd945edc024d35d275066b7bbeb778ab499403585d8c2d4138978ef3.jpg) + +
    +flowchart + +```mermaid +graph TD + Z2 --> Z1 + Z2 <--> Z3 + Z3 --> Z1 +``` +
    + +Note that an SEM for Z determines its law. Indeed using a topological ordering $\pi$ for the associated DAG, we can write each $Z_{k}$ as a function of $\varepsilon_{\pi^{-1}(1)}, \varepsilon_{\pi^{-1}(2)}, \ldots, \varepsilon_{\pi^{-1}(\pi(k))}$ . Importantly, though, we can use it to tell us much more than simply the law of Z: for example we can query properties of the distribution of Z after having set a particular component to any given value. This is what we study next. + +## 4.5 Interventions + +Given an SEM S, we can replace one (or more) of the structural equations by a new structural equation, for example for a chosen variable k we could replace the structural equation $Z_{k} = h_{k}(Z_{P_{k}}, \varepsilon_{k})$ by $Z_{k} = \tilde{h}_{k}(\tilde{Z}_{\tilde{P}_{k}}, \tilde{\varepsilon}_{k})$ . This gives us a new structural equation model $\tilde{S}$ which in turn determines a new joint law for Z. + +When we have $\tilde{h}_{k}(\tilde{Z}_{\tilde{P}_{k}},\tilde{\varepsilon}_{k})=a$ for some $a\in R$ , so we are setting the value of $Z_{k}$ to be a, we call this a (perfect) intervention. Expectations and probabilities under this new law for Z are written by adding $|do(Z_{k}=a)$ e.g. $\mathbb{E}(Z_{j}|do(Z_{k}=a))$ . Note that this will in general be different from the conditional expectation $\mathbb{E}(Z_{j}|Z_{k}=a)$ . + +Example 4.4.1 continued. After the intervention $do(Z_2 = 1)$ (everyone is forced to go to statistical modelling), we have a new SEM $\tilde{\mathcal{S}}$ : + +$$ +Z _ {3} = \varepsilon_ {3} \sim \mathrm{Bern} (1 / 4) +$$ + +$$ +Z _ {2} = 1 +$$ + +$$ +Z _ {1} = \mathbb {1} _ {\{\varepsilon_ {1} (1 + Z _ {3}) > 1 / 2 \}} \quad \varepsilon_ {1} \sim U [ 0, 1 ]. +$$ + +Thus $\mathbb{P}(Z_1 = 1|do(Z_2 = 1)) = \frac{1}{4}\frac{3}{4} +\frac{3}{4}\frac{1}{2} = \frac{9}{16}$ . On the other hand, + +$$ +\begin{array}{l} \mathbb {P} (Z _ {1} = 1 | Z _ {2} = 1) = \sum_ {j \in \{0, 1 \}} \mathbb {P} (Z _ {1} = 1 | Z _ {2} = 1, Z _ {3} = j) \mathbb {P} (Z _ {3} = j | Z _ {2} = 1) \\ = \frac {1}{\mathbb {P} (Z _ {2} = 1)} \sum_ {j \in \{0, 1 \}} \mathbb {P} (Z _ {1} = 1 | Z _ {2} = 1, Z _ {3} = j) \mathbb {P} (Z _ {2} = 1 | Z _ {3} = j) \mathbb {P} (Z _ {3} = j) \\ = \frac {1}{\frac {1}{4} \frac {3}{4} + \frac {3}{4} \frac {1}{2}} \left(\frac {3}{4} \frac {3}{4} \frac {1}{4} + \frac {1}{2} \frac {1}{2} \frac {3}{4}\right) \\ = \frac {7}{1 2} \neq \frac {9}{1 6}. \\ \end{array} +$$ + +## 4.6 The Markov properties on DAGs + +The DAG of an SEM can encode a number of conditional independencies present in the law of the random vector Z. To understand this, we first introduce some so-called Markov properties on DAGs. + +Let $P$ be the joint law of $Z$ and suppose it has a density $f$ . + +Definition 12. Given a DAG G, we say P satisfies the + +(i) Markov factorisation property w.r.t. the DAG $\mathcal{G}$ if + +$$ +f (z _ {1}, \ldots , z _ {p}) = \prod_ {k = 1} ^ {p} f (z _ {k} | z _ {\mathrm{pa} (k)}). +$$ + +(ii) global Markov property w.r.t. the DAG G if for all disjoint A, B, S ⊆ {1, . . . , p}, + +$$ +A, B d \text {-separated by} S \Rightarrow Z _ {A} \perp \perp Z _ {B} | Z _ {S}. +$$ + +Theorem 31. If P has a density f (with respect to a product measure), then all Markov properties in definition 12 are equivalent. + +In view of this, we will henceforth use the term Markov to mean global Markov. + +Proposition 32. Let P be the law of an SEM with DAG G. Then P obeys the Markov factorisation property w.r.t. G. + +Thus we can read off from the DAG of an SEM a great deal of information concerning the distribution it generates. We can use this to help us calculate the effects of interventions. + +We have seen now how an SEM can be used to not only query properties of the joint distribution, but also to determine the effects of certain perturbations to the system. In many settings, we may not have a prespecified SEM to work with, but instead we would like to learn the DAG from observational data. This is the problem we turn to next. + +## 4.7 Causal structure learning + +Given a sample of observations from P, we would like to determine the DAG which generated it. We can think of this task in terms of two subtasks: firstly we need to understand how to extract information concerning P from a sample, which is a traditional statistical question of the sort we are used to; secondly, given P itself, we need to relate this to the DAG which generated it. The latter problem is unique to casual inference and we discuss this first. + +## 4.7.1 Three obstacles + +There are three obstacles to causal structure learning. The first two are more immediate but the last is somewhat subtle. + +## Causal minimality + +We know that if P is generated by an SEM with DAG G, then P will be Markov w.r.t. G. Conversely, one can show that if P is Markov w.r.t. a DAG G, then there is also an SEM with DAG G that could have generated P. But P will be Markov w.r.t. a great number of DAGs, e.g. $Z_{1}$ and $Z_{2}$ being independent can be represented by + +$$ +Z _ {1} = 0 \times Z _ {2} + \varepsilon_ {1} = \varepsilon_ {1}, \qquad Z _ {2} = \varepsilon_ {2}. +$$ + +This motivates the following definition. + +Definition 13. P satisfies causal minimality with respect to G if it is (global) Markov w.r.t. G but not to a proper subgraph of G with the same nodes. + +## Markov equivalent DAGs + +It is possible for two different DAGs to satisfy the same collection of d-separations e.g. + +![](images/a6f5a470d138e1e27857e6707f11d231913d097ff50c0ba6af11f8fdbee59332.jpg) + +![](images/ec355d366ee52eacc83b1433ca0f13272259605d708744de8ee54a5bc3eeeda3.jpg) + +For a DAG $\mathcal{G}$ , let + +$\mathcal{M}(\mathcal{G}) = \{\text{distributions } P : P \text{ satisfies the global Markov property w.r.t. } \mathcal{G}\}$ . + +Definition 14. We say two DAGs $\mathcal{G}_1$ and $\mathcal{G}_2$ are Markov equivalent if $\mathcal{M}(\mathcal{G}_1) = \mathcal{M}(\mathcal{G}_2)$ . + +Proposition 33. Two DAGs are Markov equivalent if and only if they have the same skeleton and v-structures. + +The set of all DAGs that are Markov equivalent to a DAG can be represented by a completed PDAG (CPDAG) which contains an edge $(j,k)$ if and only if one member of the Markov equivalence class does. We can only ever hope to obtain the Markov equivalence class i.e. the CPDAG of a DAG with which P satisfies causal minimality (unless we place restrictions on the functional forms of the SEM equations). + +## Faithfulness + +Consider the following SEM. + +$$ +Z _ {1} = \varepsilon_ {1} +$$ + +$$ +Z _ {2} = \alpha Z _ {1} + \varepsilon_ {2} +$$ + +$$ +Z _ {3} = \beta Z _ {1} + \gamma Z _ {2} + \varepsilon_ {3}, +$$ + +where $\varepsilon \sim N_3(0,I)$ . Then $(Z_{1},Z_{2},Z_{3})\sim N_{3}(0,\Sigma) = P^{0}$ with + +$$ +\Sigma = \left( \begin{array}{c c c} 1 & \alpha & \beta + \alpha \gamma \\ \alpha & \alpha^ {2} + 1 & \alpha \beta + \gamma (\alpha^ {2} + 1) \\ \beta + \alpha \gamma & \alpha \beta + \gamma (\alpha^ {2} + 1) & \beta^ {2} + \gamma^ {2} (\alpha^ {2} + 1) + 2 \beta \gamma \alpha + 1 \end{array} \right). +$$ + +If $\beta + \alpha \gamma = 0$ e.g. if $\beta = -1$ , $\alpha, \gamma = 1$ , then $Z_1 \perp Z_3$ . We claim that in this case $P^0$ can also be generated by the SEM + +$$ +\tilde {Z} _ {1} = \tilde {\varepsilon} _ {1} +$$ + +$$ +\tilde {Z} _ {2} = \tilde {Z} _ {1} + \tilde {\alpha} \tilde {Z} _ {3} + \tilde {\varepsilon} _ {2} +$$ + +$$ +\tilde {Z} _ {3} = \tilde {\varepsilon} _ {3}. +$$ + +Here the $\tilde{\varepsilon}_{j}$ are independent with $\tilde{\varepsilon}_{1}\sim N(0,1)$ , $\tilde{\varepsilon}_{3}\sim N(0,2)$ , $\tilde{\alpha}=1/2$ and $\tilde{\varepsilon}_{3}\sim N(0,1/2)$ . Writing the DAGs for the two SEMs above as G and $\tilde{G}$ , note that $P^{0}$ satisfies causal minimality w.r.t. both G and $\tilde{G}$ . + +Definition 15. We say $P$ is faithful to the DAG $\mathcal{G}$ if it is Markov w.r.t. $\mathcal{G}$ and for all disjoint $A, B, S \subseteq \{1, \ldots, p\}$ , + +$$ +A, B d \text {-separated by} S \Leftarrow Z _ {A} \perp \perp Z _ {B} | Z _ {S}. +$$ + +Faithfulness demands that all conditional independencies in P are represented in the DAG. In our example $P^{0}$ is not faithful to G, but it is faithful to $\tilde{G}$ . + +## 4.7.2 The PC algorithm + +Proposition 34. If nodes j and k in a DAG G are adjacent, then no set can d-separate them. If they are not adjacent and $\pi$ is a topological order with $\pi(j) < \pi(k)$ , then they are d-separated by $\text{pa}(k)$ . + +Proof. Consider a path $j = j_{1}, \ldots, j_{m} = k$ . We may assume we don't have $j_{m-1} \to k$ as otherwise the path would be blocked since $j_{m-1} \in \text{pa}(k)$ . Let l be the largest $l'$ with $j_{l'-1} \to j_{l'} \leftarrow j_{l'+1}$ ; this must exist as otherwise we would have a directed path from k to j contradicting the topological ordering. In order for the path to be active, $j_{l'}$ must have a descendant in $\text{pa}(k)$ , but this would introduce a cycle. □ + +This shows in particular that any non-adjacent nodes must have a d-separating set. If we assume that P is faithful w.r.t. a DAG G, we can check whether nodes j and k are adjacent in G by testing whether there is a set S with $Z_{j} \perp \perp Z_{k}|Z_{S}$ . If there is no such set S, j and k must be adjacent. This allows us to recover the skeleton of G. + +Proposition 35. Suppose we have a triple of nodes j, k, l in a DAG and the only non-adjacent pair is j, k (i.e. in the skeleton j - l - k). + +(i) If the nodes are in a v-structure $(j \rightarrow l \leftarrow k)$ then no S that d-separates j and k can contain l. +(ii) If there exists an $S$ that $d$ -separates $j$ and $k$ and $l \notin S$ , then we must have $j \to l \leftarrow k$ . + +Proof. For (i) note that any set containing l cannot block the path j, l, k. For (ii) note we know that the path j, l, k is blocked by S, so we must have $j \rightarrow l \leftarrow k$ . □ + +This last result then allows us to find the v-structures given the skeleton and a d-separating set $S(j, k)$ corresponding to each absent edge. Given a skeleton and v-structures, it may be possible to orient further edges by making use of the acyclicity of DAGs; we do not cover this here. + +## Population version + +The PC-algorithm, named after its inventors Peter Spirtes and Clarke Glymour [Spirtes et al., 2000], exploits the fact that we need not search over all sets S but only subsets of either $\mathrm{pa}(j)$ or $\mathrm{pa}(k)$ for efficiency. The version assumes P is known and so conditional independencies can be queried directly. A sample version that is applicable in practice is given in the following subsection. We denote the set of nodes that are adjacent to a node j in graph G by $\mathrm{adj}(\mathcal{G}, j)$ . + +Algorithm 1 First part of the PC algorithm: finding the skeleton. +Set $\hat{\mathcal{G}}$ to be the complete undirected graph. Set $\ell = -1$ . +repeat + Increment $\ell \to \ell + 1$ . + repeat + Select a (new) ordered pair of nodes $j, k$ that are adjacent in $\hat{\mathcal{G}}$ and such that $|\operatorname{adj}(\hat{\mathcal{G}}, j) \setminus \{k\}| \geq \ell$ . + repeat + Choose new $S \subseteq \operatorname{adj}(\hat{\mathcal{G}}, j) \setminus \{k\}$ with $|S| = \ell$ . + If $Z_j \perp Z_k |Z_S$ then delete edges $(j, k)$ and $(k, j)$ and set $S(j, k) = S(k, j) = S$ . + until edges $(j, k)$ , $(k, j)$ are deleted or all relevant subsets have been chosen. + until all relevant ordered pairs have been chosen. +until for every ordered pair $j, k$ that are adjacent in $\hat{\mathcal{G}}$ we have $|\operatorname{adj}(\hat{\mathcal{G}}, j) \setminus \{k\}| < \ell$ . + +Suppose P is faithful to DAG $G^{0}$ . At each stage of the Algorithm 1 we must have that the skeleton is a subgraph of $\hat{G}$ . By the end of the algorithm, for each pair j, k adjacent in $\hat{G}$ , we would have searched through $\operatorname{adj}(\hat{\mathcal{G}}, j)$ and $\operatorname{adj}(\hat{\mathcal{G}}, k)$ for sets S such that $Z_{j} \perp Z_{k}|Z_{S}$ . If P were faithful to $G^{0}$ then, we would know that j and k must be adjacent in $G^{0}$ . That is the output of Algorithm 1 would be the skeleton of $G^{0}$ . + +Algorithm 2 Second part of the PC algorithm: finding the v-structures +for all pairs of non-adjacent variables $j, k$ (in skeleton $\hat{\mathcal{G}}$ ) with common neighbour $l$ do +If $l \notin S(j, k)$ then orient $j \to l \leftarrow k$ . +end for + +## Sample version + +The sample version of the PC algorithm replaces the querying of conditional independence with a conditional independence test applied to data $x_{1},\ldots,x_{n}$ . The level of the test $\alpha$ will be a tuning parameter of the method. If the data are assumed to be multivariate normal, the (sample) partial correlation can be used to test conditional independence since if $Z_{j} \perp \perp Z_{k}|Z_{S}$ then + +$$ +\mathrm{Corr} (Z _ {j}, Z _ {k} | Z _ {S}) := \rho_ {j k \cdot S} = 0. +$$ + +To compute the sample partial correlation, we regress $X_{j}$ and $X_{k}$ on $X_{S}$ and compute the correlation between the resulting residuals. + +## Chapter 5 + +# High-dimensional inference + +Consider the normal linear model $Y = X\beta^{0} + \varepsilon$ where $\varepsilon \sim N_{n}(0, \sigma^{2}I)$ . In the low-dimensional setting, the fact that $\hat{\beta}^{\mathrm{OLS}} - \beta^{0} \sim N_{p}(0, \sigma^{2}(X^{T}X)^{-1})$ allows us to form confidence intervals for components of $\beta_{j}^{0}$ and perform hypothesis tests with $H_{0}: \beta_{j}^{0} = 0$ , for example. + +One might hope that studying the distribution of $\hat{\beta}_{\lambda}^{L}-\beta^{0}$ would enable us to perform these tasks in the high-dimensional setting when $p\gg n$ . However, the distribution of $\hat{\beta}_{\lambda}^{L}-\beta^{0}$ is intractable and depends delicately on the unknown $\beta^{0}$ , making it unsuitable as a basis for establishing confidence intervals. + +Whilst several methods have been proposed over the years, typically they have involved placing conditions on the unknown $\beta^{0}$ , other than the usual assumption of sparsity. Given that the task is to perform inference for $\beta^{0}$ , such conditions are undesirable. In the last couple of years, there has been a breakthrough on this front [Zhang and Zhang, 2014, Van de Geer et al., 2014], and here we will aim to cover the main ideas in this exciting development. Our treatment follows Van de Geer et al. [2014]. + +We begin our investigation by considering the KKT conditions of the Lasso. Fix $\lambda > 0$ and let $\hat{\beta}$ be the Lasso estimator with tuning parameter $\lambda$ . Recall that the KKT conditions give + +$$ +\frac {1}{n} X ^ {T} (Y - X \hat {\beta}) = \lambda \hat {\nu} +$$ + +where $\|\hat{\nu}\|_{\infty} \leq 1$ and writing $\hat{S} = \{k : \hat{\beta}_{k} \neq 0\}$ , $\hat{\nu}_{\hat{S}} = \operatorname{sgn}(\hat{\beta}_{\hat{S}})$ . Setting $\hat{\Sigma} = X^{T}X/n$ and rearranging we have + +$$ +\hat {\Sigma} (\hat {\beta} - \beta^ {0}) + \lambda \hat {\nu} = \frac {1}{n} X ^ {T} \varepsilon . +$$ + +The key idea is now to form an approximate inverse $\hat{\Theta}$ of $\hat{\Sigma}$ . Then we have + +$$ +\hat {\beta} + \lambda \hat {\Theta} \hat {\nu} - \beta^ {0} = \frac {1}{n} \hat {\Theta} X ^ {T} \varepsilon + \frac {1}{\sqrt {n}} \Delta +$$ + +where $\Delta = \sqrt{n} (\hat{\Theta}\hat{\Sigma} -I)(\beta^0 -\hat{\beta})$ . Define + +$$ +\hat {b} = \hat {\beta} + \lambda \hat {\Theta} \hat {\nu} = \hat {\beta} + \hat {\Theta} X ^ {T} (Y - X \hat {\beta}) / n, +$$ + +which we shall refer to as the debiased Lasso. If we choose $\hat{\Theta}$ such that $\Delta$ is small, we will have $\hat{b}-\beta^{0}\approx\hat{\Theta}X^{T}\varepsilon/n$ , which can be used as a basis for performing inference. + +We already know that under a compatibility condition on the design matrix $X$ , $\| \hat{\beta} - \beta^0 \|_1$ is small (Theorem 21) with high probability. If we can also show that the $\ell_{\infty}$ -norms of rows of $\hat{\Theta} \hat{\Sigma} - I$ are small, we can leverage this fact using Hölder's inequality to show that $\| \Delta \|_{\infty}$ is small. Let $\hat{\theta}_j$ be the $j$ th row of $\hat{\Theta}$ . Then $\| (\hat{\Sigma} \hat{\Theta}^T - I)_j \|_{\infty} \leq \eta$ is equivalent to + +$$ +\frac {1}{n} \| X _ {- j} ^ {T} X \hat {\theta} _ {j} \| _ {\infty} \leq \eta \mathrm{and} | X _ {j} ^ {T} X \hat {\theta} _ {j} / n - 1 | \leq \eta . +$$ + +The first of these inequalities is somewhat reminiscent of the KKT conditions for the Lasso. Let + +$$ +\hat {\gamma} ^ {(j)} = \underset {\gamma \in \mathbb {R} ^ {p - 1}} {\arg \min} \left\{\frac {1}{2 n} \| X _ {j} - X _ {- j} \gamma \| _ {2} ^ {2} + \lambda_ {j} \| \gamma \| _ {1} \right\}. \tag {5.0.1} +$$ + +Further let + +$$ +\hat {\tau} _ {j} ^ {2} = X _ {j} ^ {T} (X _ {j} - X _ {- j} \hat {\gamma} ^ {(j)}) / n = \frac {1}{n} \| X _ {j} - X _ {- j} \hat {\gamma} ^ {(j)} \| _ {2} ^ {2} + \lambda_ {j} \| \hat {\gamma} ^ {(j)} \| _ {1}; +$$ + +see the example sheet for the final equality. Then set + +$$ +\begin{array}{l} \hat {\theta} _ {j} = - \frac {1}{\hat {\tau} _ {j} ^ {2}} (\hat {\gamma} _ {1} ^ {(j)}, \dots , \hat {\gamma} _ {j - 1} ^ {(j)}, - 1, \hat {\gamma} _ {j} ^ {(j)}, \dots , \hat {\gamma} _ {p - 1} ^ {(j)}) ^ {T}. \\ \begin{array}{c} \uparrow \\ j \text {th position} \end{array} \\ \end{array} +$$ + +Note that by construction, + +$$ +X \hat {\theta} _ {j} = \frac {X _ {j} - X _ {- j} \hat {\gamma} ^ {(j)}}{X _ {j} ^ {T} (X - X _ {- j} \hat {\gamma} ^ {(j)}) / n}. +$$ + +Thus $X_{j}^{T}X\hat{\theta}_{j} / n = 1$ and by the KKT conditions of the Lasso optimisation (5.0.1), we have $\hat{\tau}_j^2\| X_{-j}^T X\hat{\theta}_j\|_\infty /n\leq \lambda_j$ . + +Thus with the choice of $\hat{\Theta}$ defined as above, we have + +$$ +\| \Delta \| _ {\infty} \leq \sqrt {n} \| \hat {\beta} - \beta^ {0} \| _ {1} \max _ {j} \frac {\lambda_ {j}}{\hat {\tau} _ {j} ^ {2}} +$$ + +When can we expect $\lambda_{j}/\hat{\tau}_{j}^{2}$ to be small? One way of answering this is to consider a random design setting. Let us assume that each row of X is independent and distributed as $N_{p}(0,\Sigma)$ where $\Sigma$ is positive definite. Write $\Omega = \Sigma^{-1}$ . From Proposition 29 and our study of the neighbourhood selection procedure (see also Section 4.3.3), we know that for each j, we can write + +$$ +X _ {j} = X _ {- j} \gamma^ {(j)} + \varepsilon^ {(j)}, \tag {5.0.2} +$$ + +where $\varepsilon_i^{(j)}|X_{-j}\stackrel {\mathrm{i.i.d.}}{\sim}N(0,\Omega_{jj}^{-1})$ and $\gamma^{(j)} = -\Omega_{jj}^{-1}\Omega_{-j,j}$ . Theorem 21 can therefore be used to understand properties of $\hat{\gamma}^{(j)}$ and hence the $\hat{\tau}_j^2$ . In order to apply this result however, we need $\gamma^{(j)}$ to be sparse. Let us therefore define + +$$ +s _ {j} = \sum_ {k \neq j} \mathbb {1} _ {\{\Omega_ {k j} \neq 0 \}} +$$ + +and $s_{\max} = \max(\max_j s_j, s)$ . In order to make the following result more easily interpretable, we will consider an asymptotic regime where X, s, $s_{\max}$ etc. are all allowed to change as $n \to \infty$ , though we suppress this in the notation. We will consider $\sigma$ as constant. + +Theorem 36. Suppose the minimum eigenvalue of $\Sigma$ is always at least $c_{\mathrm{min}} > 0$ and $\max_j \Sigma_{jj} \leq 1$ . Suppose further that $s_{\mathrm{max}} \sqrt{\log(p)/n} \to 0$ . Then there exists constants $A_1, A_2$ such that setting $\lambda = \lambda_j = A_1 \sqrt{\log(p)/n}$ , we have + +$$ +\sqrt {n} (\hat {b} - \beta^ {0}) = W + \Delta +$$ + +$$ +W | X \sim N _ {p} (0, \sigma^ {2} \hat {\Theta} \hat {\Sigma} \hat {\Theta} ^ {T}), +$$ + +and as $n,p\to \infty$ + +$$ +\mathbb {P} (\| \Delta \| _ {\infty} > A _ {2} s \log (p) / \sqrt {n}) \to 0. +$$ + +Proof. Consider the sequence of events $\Lambda_{n}$ described by the following properties: + +• $\phi_{\hat{\Sigma},s}^{2}\geq c_{\min}/2$ and $\phi_{\hat{\Sigma}_{-j,-j},s_{j}}^{2}\geq c_{\min}/2$ for all j, +- $2\| X^T\varepsilon \|_{\infty} / n \leq \lambda$ and $2\| X_{-j}^{T}\varepsilon^{(j)}\|_{\infty} / n \leq \lambda$ for all $j$ , +- $\| \varepsilon^{(j)}\| _2^2 /n\geq \Omega_{jj}^{-1}(1 - 4\sqrt{\log(p) / n})$ for all $j$ . + +You will show on the example sheet that then $\mathbb{P}(\Lambda_{n})\to1$ for $A_{1}$ sufficiently large. In the following we work on $\Lambda_{n}$ , and $c_{1},c_{2},\ldots$ will be constants. + +By Theorem 21, we have + +$$ +\| \hat {\beta} - \beta^ {0} \| _ {1} \leq c _ {1} s \sqrt {\log (p) / n}. +$$ + +We now seek a lower bound for the $\hat{\tau}_j^2$ . Consider the linear models in (5.0.2). Note that the maximum eigenvalue of $\Omega$ is at most $c_{\mathrm{min}}^{-1}$ so $\Omega_{jj} \leq c_{\mathrm{min}}^{-1}$ . Also, $\Omega_{jj}^{-1} = \mathrm{Var}(X_{ij}|X_{i,-j}) \leq \mathrm{Var}(X_{ij}) = \Sigma_{jj} \leq 1$ . Thus applying Theorem 21 to the linear models (5.0.2), we know that + +$$ +\| \gamma^ {(j)} - \hat {\gamma} ^ {(j)} \| _ {1} \leq c _ {2} s _ {j} \sqrt {\log (p) / n}. +$$ + +Then + +$$ +\begin{array}{l} \hat {\tau} _ {j} ^ {2} \geq \frac {1}{n} \| X _ {j} - X _ {- j} \hat {\gamma} ^ {(j)} \| _ {2} ^ {2} \geq \frac {1}{n} \| \varepsilon^ {(j)} \| _ {2} ^ {2} - \frac {2}{n} \| X _ {- j} ^ {T} \varepsilon^ {(j)} \| _ {\infty} \| \gamma^ {(j)} - \hat {\gamma} ^ {(j)} \| _ {1} \\ \geq \Omega_ {j j} ^ {- 1} (1 - 4 \sqrt {\log (p) / n}) - \frac {c _ {4} s _ {\max} \log (p)}{n} \\ \geq c _ {\mathrm{min}} / 2 \\ \end{array} +$$ + +for all j when n is sufficiently large. Putting things together we see that on $\Lambda_{n}$ , + +$$ +\begin{array}{l} \| \Delta \| _ {\infty} \leq \lambda \sqrt {n} \| \hat {\beta} - \beta^ {0} \| _ {1} \max _ {j} \hat {\tau} _ {j} ^ {- 2} \\ \leq 2 A _ {1} \sqrt {\log (p)} (c _ {1} s \sqrt {\log (p) / n}) / c _ {\min} \leq A _ {2} s \log (p) / \sqrt {n} \\ \end{array} +$$ + +where $A_{2} = 2c_{1}A_{1} / c_{\mathrm{min}}$ . Thus + +$$ +\mathbb {P} (\| \Delta \| _ {\infty} > A _ {2} s \log (p) / \sqrt {n}) \leq \mathbb {P} (\Lambda_ {n} ^ {c}) \to 0. +$$ + +![](images/24d45084cc85684f1c64573b740c676d4429c9d6ece6380d25afafb7933b9240.jpg) + +## 5.1 Using the debiased Lasso in practice + +Theorem 36 shows in particular that + +$$ +\sqrt {n} (\hat {b} _ {j} - \beta_ {j} ^ {0}) \approx W _ {j} +$$ + +where $W_{j} \sim N(0, \sigma^{2}(\hat{\Theta}\hat{\Sigma}\hat{\Theta}^{T})_{jj})$ . Let $(\hat{\Theta}\hat{\Sigma}\hat{\Theta}^{T})_{jj} = d_{j}$ . The approximate equality above suggests constructing $(1-\alpha)$ -level confidence intervals of the form + +$$ +\left[ \hat {b} _ {j} - z _ {\alpha / 2} \sigma \sqrt {d _ {j}} / \sqrt {n}, \hat {b} _ {j} + z _ {\alpha / 2} \sigma \sqrt {d _ {j}} / \sqrt {n} \right], +$$ + +where $z_{\alpha}$ is the upper $\alpha$ point of a standard normal. The only unknown quantity in the confidence interval above is $\sigma$ : this can be estimated using e.g. the square-root / scaled Lasso [Sun and Zhang, 2012]. + +## Bibliography + +L. Breiman. Stacked regressions. Machine Learning, 24:49–64, 1996. +C. Cortes and V. Vapnik. Support-vector networks. Machine learning, 20(3):273–297, 1995. +A. Hoerl and R. Kennard. Ridge regression: Biased estimation for nonorthogonal problems. Technometrics, pages 55–67, 1970. +G. S. Kimeldorf and G. Wahba. A correspondence between bayesian estimation on stochastic processes and smoothing by splines. The Annals of Mathematical Statistics, 41(2): 495-502, 1970. +N. Meinshausen. Relaxed lasso. Computational Statistics and Data Analysis, 52:374–393, 2007. +N. Meinshausen and P. Bühlmann. High dimensional graphs and variable selection with the lasso. Annals of Statistics, 34:1436-1462, 2006. +A. Rahimi and B. Recht. Random features for large-scale kernel machines. In Advances in Neural Information Processing Systems, pages 1177-1184, 2007. +B. Schölkopf, R. Herbrich, and A. J. Smola. A generalized representer theorem. In International Conference on Computational Learning Theory, pages 416–426. Springer, 2001. +P. Spirtes, C. N. Glymour, and R. Scheines. Causation, prediction, and search. MIT press, 2000. +T. Sun and C.-H. Zhang. Scaled sparse linear regression. Biometrika, page ass043, 2012. +R. Tibshirani. Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society, Series B, 58:267–288, 1996. +R. Tibshirani, M. Saunders, S. Rosset, J. Zhu, and K. Knight. Sparsity and smoothness via the fused lasso. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 67(1):91–108, 2005. +P. Tseng. Convergence of a block coordinate descent method for nondifferentiable minimization. Journal of optimization theory and applications, 109(3):475-494, 2001. +S. Van de Geer, P. Bühlmann, Y. Ritov, R. Dezeure, et al. On asymptotically optimal confidence regions and tests for high-dimensional models. The Annals of Statistics, 42(3):1166–1202, 2014. +D. Wolpert. Stacked generalization. Neural Networks, 5:241-259, 1992. +Y. Yang, M. Pilanci, and M. J. Wainwright. Randomized sketches for kernels: Fast and optimal non-parametric regression. arXiv preprint arXiv:1501.06195, 2015. +M. Yuan and Y. Lin. Model selection and estimation in regression with grouped variables. Journal of the Royal Statistical Society: Series B, 68:49–67, 2006. +M. Yuan and Y. Lin. Model selection and estimation in the gaussian graphical model. Biometrika, 94(1):19–35, 2007. +C.-H. Zhang. Nearly unbiased variable selection under minimax concave penalty. The Annals of statistics, pages 894-942, 2010. +C.-H. Zhang and S. S. Zhang. Confidence intervals for low dimensional parameters in high dimensional linear models. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 76(1):217–242, 2014. +H. Zou. The adaptive lasso and its oracle properties. Journal of the American Statistical Association, 101:1418–1429, 2006. \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-statistical-learning/images/image-1-Support Vector Margin Visualization.jpg b/apps/api/app/data/demo_documents/stem-statistical-learning/images/image-1-Support Vector Margin Visualization.jpg new file mode 100644 index 0000000000000000000000000000000000000000..1e7a24d598f2189d2035f4cf693b93a0bce663d5 GIT binary patch literal 37312 zcmeFZ1z42p+BW>qozg8JAR*ESNJ@!xH%c?oT~bm*gMff^x5Ut)q;z-5Al;3C_>ZpT zT1)rd@7nKvzxO--<3Il4c(~_b<`|y)zOOj1^Sqw0SAwOh=h!SiiVB>djkX)fQLgsfJa0?LPA7@z1ti1I)I3W zgnyeu4EdJwQxsZz0#3h}OjJ7YiuZ&nBM0G(20miNXhOnFf!d`=H}t$<9{F^ z@kmliT1NJbqm#3XtDCz=z{|j(;E+&oY}~8(gv8fL$ywPs zxq0~og+-NB)sULny84E;_Kwc3?hie^qhsR}lT*_(v&+zxPpfO|8=G5)N5?0pXXh86 zFTc$T4nX*ATCm^0P3)I>;lbtwkBEqXi1KY-aPY3M0|5^a={5&4zL+w~Q~O)AoPMYT z;xU;O@6qVER1OHAIgFqa(Q_}~Is7)Y?=$=NCg%Si&Fqhf{W-5$022WYws{D6fCzAY z$&}@b{{P~GOm={jo2vR1P|s~Lsw|2TCnbM};Qym1*j6@*r4I_L0h7y-xNS*-|9|{t z9hOHEVmwra;Uo!^ejEsjI*(N+V}G+@|D&@d#Tw6)LJcO>eL*L}umZP8IT51GSx(WC zTABVqMV6???=G@k>1%j(xDh|sjulb(xC#xUcoI|$;mh3$j_6YN`v~^dtf(P!Av$X( z0rMz44Ajbi2Q&SC^d3@W%^C!_x+ z{kY2i?GYh4MZwA%AW)Y4en_)*+^=M;j&*B;YC_-*N8oFtzi8da{=K zJiDmOfL@Lgp}3zfGLZYnvS5DE)TtVYC2Dk`a`NX!&xw_{K#0b&7iL;=Tv5cC1+>-s zb6lW}Rsd5L`2XtRaFUy2@TwY6hPl#@V{9_^)oIT6YsV(;-2IXYUBJ$L|yn1}lmHT-|(&(CR#Q5zfLkr$ZP&ezt zLk_0Q$dT||7dpf%_1aw6$MdXGEtISF%| zqY4+Fiy5#yJ0{PEI}3zk`D6%JWLut!QmzvCenOBFrO5AKpgstbBlXT*PUgaW*gawty-rSNm`l^ zQM5ihd1I2sOpQ3hlOFi|y0c@9J%zDeXc)OPYl&@Oq#>5~W!-%w66U&3)LmsT8#mf| z!8y$B*;U;#j(>ODBoi@NpRbFIq0}j1($#Oxr=d}1!0e({bIeIQQ+JmA6&`%rsgK79E@jnO4)dft+yRF~7vnYt8c z{Km?;3+Q^g`VVveGaCGe3jaGJ3j1Rn$xXSRHWcfjzNwwY z)*#Q=O=eSAOdl*8icOEIucFZ5CrTA2X<|V(wp}x-Uz*)+rHd<=W@@Ns44%rs*6peU z?V4L#WIQ%igFPF*0=R}BuR)a@+58S*Ehl;h7B0Pmy(B2P zbFgwiKdM7`PVwC{x)@QSU1xH+8FTQgbNol6O6!gMF&?XeO#%6|Frx$)4tKlq7Fpq( z^Wt$h9c$rH=o(9C>56;SS3qN@p5>&N`!|oLa$a6zgibbK+zKHMI4zefqhgWe-UnCe zx%S4v&~41zy`;kQv8w7tQ-{or011*-+8ygYXyB0Fn1AuiD~Pjo0N{(0wR zn7-bV))|el7z(Yci_P@WsyDk?3#o2;oCT?}jpBk|f!Jenv1~W+Lj#TQ**=@;d4I(f zLuFlczh{NcP{FX*trMdDcz#{p#V=BMXUE^DJte&Y&$0=^;*^Hte=k_6>o!Xbb}q{m zzlK_-n>-!F;LFIsQ2RaS{m6a4jt@LkMq!@xQ{u7QmVN9o`WuFjcN-N8R%#y(A1lZ} zR7VGt0LnnNKk)?$+ln{-ie?2Qc9^IRl4>eRc>u}_cJ~*#|MV=PeKGyFg!}71B~(?Z zj~!D?w#c;9*L-P^S2`>ug~eE*Y$?lIX19nnx|`<*X)Gj@D;+pbc}ZBCh2UMFZxvpL z?N#?>T$wTG_5$a7d)E6;=5j)DkZY_FJ_q*VFy>(W{$sGR=1_fGjZ6j{`2h^v>OF*?+kXe_fOR*>iQ>(#L~? zT3K~}k?`l}_9L7BA0O2e?GYX6*_m0(&z0UHwfGmf1!$4oi@!4* zx*E6F1+Jg(&M09=yKp$Ym##l$!nvh-6r$J%{}pgT2X-v}P4xITsPBI-eY4PYyr8Q% z=z4X2z9WtRQFG94L9>+Kb8I1jM69aea4@jB@i)uqsA@SN{|o3bN@j@eq8DE&3L?!P#?tyfiRW2}I~Ez<9+bVB zI`4Pcn+?gnlk%1jVqgok4W3Pd6n48^#?(tTYvZF~6Y)2OyK=+Da?sj%1 zJ|!7|G~PyEnTYU!1Z9GXbnZ;N>0`w?*Y_Y>ZQ`-FF-dH{Y@?q#>wogX%jk1x8u}NO zaxh6oxHP6E88$coP(l=lfNsuOPiz;O)`sRX3q&J>P}->Y#;%_tMp(s4vTCtY1f=<- z>KvxF>0Iz>#$0;GPaV=xPu5YA%YOx4>bAE~9#VoIgug5H2zXO`+<3s?D=0#6k+3Zr z!&ZHP4~h-az_ueqQPsI5DrguC0~Iju9!GN+O6c7kw}xmn$(12Apf2Bk^m&D3op8d9 zh$rg3j+R)vIGL{i+24>e6wP-niF4WhsQtdBBJR_i+=rGwB9KX(ByN8tEyBWUbmaIm z%9uB13H9uF+LHMaUm#C~V&+)_j_OV07&8gS=RS8ZV-8~4&JNs-4gsU6Le2n4IU8yIvbM#^#X*sKv#6O0jmDzcM}&dUa%{?HsCzrp5{0^6!~Ogs zd+Ejs2dyDf+O?jZua0=gjVs->j$`IMz`$Zshn-*=I4Akg7RLhqSZqQ_%t+6$y`YEL z_)XE9mDBQP<9-7A!<1fm-K%7BwX_2KJZ?%WbNpuT}P)htquR$tA9UCA*)8(vfHWtZBwqOZ6?a0uyr zgNWN{z5?$G5Jvyb*?#fY;sryU875HDD2-2(Rj1?SZqAZzO>5-vBE3=ZJApUngk07v z+0P&HkEgSyf(>YEt91z`oz6N3PdR$SaIi0{HCkDcFtg)7Hi>UQPRWgMa2|9SH!no? zB7!wg*EZ$uY&>Ud^IeB4)T0K!+` zLrjj(RKw)LK$EM>s2Eef!(9!kjIrk>%VRrP)wTL>I~tWd(1A2COZuT=Scwpad5hzk zzHA5YX5WKLB@{79`y0%#jzpEA;_a%>mJTA0CMbC2wq3AP_OC4P?-FjkBCaI9)xnSt zW}Mb5W^v4qm(Yd=@yrI_7Z9!}&WaYQ+KP0JKCxRMPF*pqs`fnHw@;glS_Gdp6o~>P z2n5wqLYW4&5UbuE7&=lj`qR9WC#_w&hfZ!oDKn8@o!-<+y#9}bDQJAIb*R{d%e$x{!gx)zVPmCM`pT2 zO*TkBia{3$QXR+BB8y+>vr^vFvDc!=TZ?8F(;j!w<*Gs>bFhBQ_rj)!Jt3#THiT+x z*V>lVv?{4fNkpm?RQ!6E&9v^-GM&o631QFNfOo+*QG%=N)%F97rs8V}h|3)bcoKLV zC5wE9$;^mcG6%h)DP2bAFI+=Y3XK>32MbG8a~1`kBPZOrj83igcCBeDYvepF68NK^ zs!IWdctqT&U_xn8B!Oyv82ZhCuP?{D@Gc|R643QdR}EX!CXTq+@do8)`EoSzl`%Bt zhLT&&ceyB%+z?TM4ERUy%#ksdXONV!=vKzHlg(ZPy`!$|by<2OV108?);4;ycsX;A zEmxb)KDyTe-lK2yM9`c0O1a@>h36(>EW)%OcpN4wNEsDrhX4+;d5S_B!%br{6rJJ`vreX>!j+VgB&?~yAq+RSk9Qp)5Z5| zVQ$w@%tooBW-e`Eii5%Q<G$9dkvyyKt=R7bJSPZO z7D+FbLBf6B9@A~>QNs5$4+*y}Uu2sX&Ezgbv?(H-I0-2s6@5$GEaZlVcgp-I44yj- zaSwznzcfu5zabP3Ez<^_QG+TRK9FP=V{+c#Le0Qp{+mSqU8U(fq@16ew3oJ-GjK=r zZ06Ec*EwrjQVQ;^lkdfr!bauck}sySx)siRS6-hky1ue5^2Cy8%Gw_51G2hfT^D{{^5tG*(c||X(68S zsb#(GT2c4e)+pQ@rP|j=V2k_9<1w|GTQh^P-EZ*oGf#LdeBT8D0pny)o<_@S#^Suj zJaL^i)vzlQ3Y@cd8Q4%TG_eCk?Qk*G`l$<5$87a}MV)Y0@kM&q6uu!PLfc_jv+|AR zD~D(t6$|+DuqIpVAWg-$;fq$shlEXDEAyA?A{cJ8rp-5thxF;7iJZ^zy02b(z-czo zeGK!#32JVPUbGs!9@FdT8Ok^zocu{AfKxYa#wn1sTzyPRz(337u-< z%R9S;q{mN9t~&v*!{w;3TL#&&bI=yjIN_f z&n?&>terPb7W1~(Ts?7A+*SQ~+4KRHgD`&B&=k*txVVl4f1vEm1_4|WWnfL%A>R|o zSa%{pLgXD$kB@eW%^rky31ZGTVVYH#yOmXH)cdz?kQ*ERWw8G<3B;r-%LyVWamwzB zVrZQ=uvTq}pUrNx-EoQbkWZIn7BeQd) zT8ouPba_BOvBi{0U;Sv%A!RMD5oY2^w|}QH5KJQOL}bTE&|%_AiIq`ny-j(4j-WlN zK0CzU073C>T|}5|Ec$dUUvLrGyV@BUg`JuIy2aG2-Mp1@176;#l%{ezG%HISQz=dCScM9%SxYtk-ht}IfIYw<7#OGS_dhon-akfz>8r#lcurO^_IS_He1YyvNKW{@a1+%a~*vMP3F-hx}? ztfx1_>RBj9_@()<=Gdimd|LWbEJ(O4Wo5zitvu$M+Kd@cvQUF{zeKNd+eh!QY4C~{ zap}T_K-Gnyf#bULKt#J!f^fTVHQdIyvV4sHqka!SHUnEkUaZLNSh>}KjD_hX-X*e7 zoO%o8U2lm_Vx~sxobkuqkx6=cbjO6Y@6Oibxl~`Te__}YFWnMA?;qY@dS@q_yH(># zZ3aqoYEgc5JEUvh12_@5Tk#?8QKi)Tm>{)CXzK6zdCEBgqw%|isnJY1H5PW@8nhiz z!3V9yS=#7X4$9W$jdELLfldB_H9`DnGNNZ5&Tb3AQW$ERHTepdC+t2+S@XLQS`|HqGSN zaipe`w@EC&B+mLipvKTVIR8{hZ*k z`sOORF>Awx3^^^q1((&n;FDh;wF*e;FISW&&#BvXEHkGJXU6z816f~xZbsa!K8ZG* z!p6~0#h(NoTkI3r@qz$G6y^51YU3g|i&SR`qaPOzR5+^{703YYqrk%;($5;k-)Rqj zMu9&m{JW_0s1G+SGBBH2+Gowe&3H9>$e*k(RP->^84_8#8sJQ6c6tchvgszC+#X51 znl4u;SoarvO+^|lWP@HA^PT`sQ)a^2xs;Rq>H$J0vp>h&Y`NyExu&lGW|R7vwj5uu zGfMw+57G77(Potg_*#>XjC$K?B2NGr`&-n;dgXqvdx(nGDHi!2Z}8}5n&#^3#l%pU zR$OS}d|aC(Z&O~dbU7ro+0&ozpGL3$cr}b1RGe!yE7fXWW&yJgr&ORW(l|2peW@Y3 zTeMqjmt{CIW01tH@K5*;u^g-1q&ul-;#CbNM+IqKSg^q%b$Q&<&x_C1-k8Lj(dwWJ zL;+mSx#3*CuOmy+Az!FZW$&zR_*i3t_Il~S>USIzpgjS&}rk<5D7@d z&GoT4xND)p+lhbZ3Pb;@>=Vly6Jws3=rTa_j1A*JxDSFE@dbJFUK+z#+a^pgs?S5>$4Y6ICJK6qQ4psFfe%hwjWr4@Zq`^l4s z)WNqp?B?F{^vArX@LS~Bq&eZKz~q<{jc)U!lSD5P0GQw(|0R}w!%F`~JpG1ptXoz4SHPmyJMYn5+H0TU>NUgJMkmou#AV($> z#So>%k`f7Be=>UWW-=9!uza525ph%pk19&C(cw;Vv%nid(chNruSY>YZc@h@{jvWp zPPI*Up4H=@U;g&ik}u{u6~j)S*=^Ugvc%?(6f=qp7!TEhgIK+`6PLK`2{8H|oDqK5 zxFH>HdRmaz2+hz~P}sH>5n+z(37d_6!raJN{GNRpBfy)Bp-U1}r9kfk+bC+K?Z-pW3T`lpe5Q zEdybrKg!78mY2+9R0W0wfBMvaq9nj2m0P%^-5zBe+sEa0z{YWY7*fXg-eX}Y_|*I$ zsA0W7>PBeYCYJGO@=5Zlko~gcAyM+3T49ZB>*Kjz0e?XyWo`kbL8|P#pO)TJ$PDE zc{Cb81^`c$5t3;Zjk5$Vu2oh!u2$U zL&(BjhL_+n|IJuIYaiRYt7zYzK&=_`OS zjo~SCOy|ft&<88$lv?G>D127H<|WUeXpT(Cz~{lEmI6D85So)VSIi&PqDlEJNcZSg@h_2 zsICZ(PVnobd(bcujKnik$;e6@i;VVdT=XiG|(U{>B z3(^pN#&hSdi}w3-z5K8yknQrSuAm8@NAdNd(8rM+B^JA08R4Zd9Pea1oSDz%3NYGBEPMHdp9sraqQb82j_ zCp;Y?5+zO)xvhGJ6j5rW^dF85PmC;N|b535#shh;y3c>+5;tp3k{l4Xg zUb_>lUAJReSE0fkRBMfwD^0G;(v#ND#;aF8Z4)u}`>TJfc`)l;fKZWDABE&rDkPgn z_}X13qJG8^?PCz-q5i{vN*|cy2Rj{>$QH?x%R5}Wf3=4ewHG;23R94OPAig*j##;k z^LjEB9RaL6u)vB&8c$G#OMn^csrhERg>_s=vY<8int_kkW!}&SQKAK>RK0luVi=M* zly4Cvr;kVe{QUL%#2HZYuoDIj_GXW{F{&iOn$3>4YzDe8;Xzx9jAD&#LzkRn*ypC?3QaVx_x3dC?#%$ zh@v87Ys?;oewQ3YiLce6S2!3m`>#LODt2hV$n!EyiG#FkXsN7_7<7II#TgQOkg+3E z+(LBlsO$L3v%~hRR6#LG`5Q`h{b*#QgrNMLr4@Y#*xeof3k)HY(3pJgq$^e5lKM>Y zR46^8z=_g8-fChmdF12E+C4dE=hF-<9askY)sFvq{%?Zc--5w+r$HDLdYY|&;i+jz zv=QRUc>O}OnO;$6xB8*kl)-3bYqj0ThX<77=g8$fWOC$ETcjjL0rQ%PgH>KUha$6d zn2S`cK8sA72x(t|$G|BwGrw6RlpvWAhupXK-GStf?gtevBp4}=P{<8bYbc~b$=NiO z>Z+^Qm;!?1rBsEJPD`!+3Gr0nB@ML}X&3O+TkPedq6_g}7xq;`TF}&}9C=RP3|plj zj##bJGs1Pe_Tp>B=39Db&%JoaP3bnK`1ibvtl8Te@*|y5Ps?v12P;T4Oeqe+QHE*e zb`sciRCIttveAp~0~x5#xv%s&2L*NcHJFS^)Rn9z(oTOdSL44<|V=t~w{fK_cRge^+w_%Cfw%eI(e zNfP+SnVH@pWDs0_U+;esl>hiUT;Ja`I{(e?f479?YtLnEXVjI+>hy)JNkak7Q#+BD zQ})t2z*-Y2w4g~bCzQn9_3d>c9#v6WgU&HE(Wa&W?i6un1FOk91@_&kZ1(fwVQz1o zDc#YJtyR%NoMlpio@7At?*tv3q%`-)T-#CaR!l}87mVz9J?f=u52duv9^I|ARKma0 z5RFSO9J(z^ssGOc1)?vfhpPT0x48>?5vQyW`mwxir%qHD596@V-E>yUTE3;WfEFyk z;7~DQ-E(P&8cX_{|HP-dYXukb__PV51c(ou9)BrUPgzTFT}K&)iC-9QQtfSLx-^EI ztDlpXKb!Lm|4adEUoOJBw`1&GD!9wSjROH!?SL5p&PxoLo`{7AN z-)mKSkID^!;iJX>t%K6OeGB@|oO?j!uMZ7@+aebGuFD)JPKdzV(Kc7)ctmN+|r zW3O;`U!M=m4N7%sBWJqj^;*Y?kH$Hc9ctKBeey+k?nxzhaPSTowQEA10FEdoM|&dy zZ)QduCK0{C*1Akr)%d2N4x*u=?mJ+hdvBiT%|M+z(t72yZ(*6kj%BH|UiXTOEC=)k;}& zH;hi})AqS_lCrCCE_sFMKuToWlD+4d476rlRH;@>(2X>y+hR3$WS}pJg zvPR^CS?46CmN5!v&o>kV1d?~G*(OIv$(_thL*e1O>HW4W;I00pgXZ_QC%gf@c;gA>K-B9~P=KN2A!r%2T{^5;(8X};8!V_l?wI-WLGHVUhRzUr9e4%zZ zz6;9M?+?a|;h>H;a`S9UR<*pZH$Ru#Kbp5@iR>}y_^Q7CnalFV@R>rX%dRJV>&H}{ z)YGy|AbHd@fonltt4y<7FDOxq9XeX&s)!$S)n}D~hv`rAQH-Xj2V9ADgZA+movIOBJ6qk9-f=lC`2sZZ#6pKAdcYtsXAGZwGtv8!a@ z9)!{mYI>}-DFx$#yQY)ZNi3U#As$r08Jjm$W3hdTe#Q~f^f3F6TP)R@p0k{cnVl7f zG0AO3;cyxjKj(`Wt^2*v(gi=uRq(&CxPLuI;o*HVJ`2mPN!&fQ7=3isk4PE(a^@H1dU~^lTDyEi{Vz25G4l%wg!>A&iS1zwn0U=u#eG{g-8%zLPHF;{jj76= zgZGsy8)i<$VfWNI5mf)$ux+;}crehhOB~{4yRda;cz^CIz^3M`sz0D&t?}NOXN+u3 zq(*=kcYUq9j0u6SO_;IvP&hL2?#nvG1h*(Po6}YUQ>Z@g`g;Rd=giA&neddmDa2>n zW7LcA%@|L{OCi?6AGqmsvdQJsMPjC@qLW}Y0W*H zS@zX6P~z(AuRx%|!FkZhR{+bh&(au|a#7;^&1Ia<37~?Dqbw&0IJVpwxzoU5mGrc1 zK0KT(JI~^bOo*K@qM@y9b&j$p1MNB6r|P%7QLxnu^3`GwNxB0rD zz1EVqiAEZkQ5Bz|sJ<^dRy@cL-mxCm+2jS+ePAOo{+wokpfS0HG3QiQ6JrwEr9S?N zY~@;}@*=pC_XV{!iB?yT=evmTsAX0CSN4+DKqS*7|FB)reI)xdk~;PD4v#PI&6K&2 z_}BZy#y$@t2qEvA7K0iX&fM7Uabd5d7I~oW6@xE4EymglKu1iP6O5c$myA}szc|)! zHR`_xU{w%r1xn8T9wswZmm+o*OWxSVd$_MYWfK;&`{PI?>wMhh!NV>$nY=Y$Idogr zdiB99sHyPf=Yc~lO|hKCZ)Dm(xDTQx@20agJw9d*SZ#_AJFaYICPH{|kIwcD8tUhG zP+?Y{lFImE0=LNT-Z3$KP~X$5`!)-FH}(hE z_%=GU@Ev4bL_NV5^(vK`Td#d(Cq`M+`i(y9EZ&?A`A#3I*ZaZ1GPjTU~)q5sFuwkWfwv@DD_}MCT17Uch#1``+XsV+@NV8^#4=?mq1D_qp z>>taK_lF{2n7Lz2lkTNE4Rk1#v#t-L_Z@b3VBnzXyl!Wm+7od_efV%Fql~>Rk&3#t z73QwN9kaWT!HgwtI}X98)K8o=_2 z$c7Vr2cHD8M_MRxoRg^(Ut#JS1}Fb?F$nJl+;57@Va=iwDbfccN4vCwHnxqPMN=76 zQp}Kqbj$n><7&U#QqMYQx@0quHxpR$Hiu{FszRp7Uw=xj{YxS6b8Y@>txf>^-eICT zVQp|OS68dpv>`g6mlda}!HjPg8@u5UZIMV^Idf;BM5+BbGV}o_-Y~V*)<)JVqwI=O zuSdl^1H-VYGyCUA|1)Xw`)g~b&8q64!(2-j+cg3IDaCVKp9nwN9{zIZkPpK6Qi$$LIQ=+;6l=1|`B+hqVt44EJcBSZ%gt&^hRtp!?Mx!+9{pX$ z0AW;P#(t-|Gq4DpV+cnASTOws!OreO{-=%27#cn(+htxGN6+(ZV)OI7K_dl!Os^<8 zA>TY_y2ssHpdHcp&g%&gT(`*f_GC|egI<(#+;3%`CRYANa()NK$%L0E)sn2#g01lz z+Y>fXytv|PrKWes&xN?iy{XozN1N%MjIo}Ck?iakX*~@$Vg^?r;^~S6FCx!ck6X$WtX#SQ)Eml@lQn{o}Y|QHC zPjeURK8ahEDw2Khkfq;?bh+4Fm}ZL)#9cMFx!xPz>v-&elk4HVmT7B5%T+-IdZ~wb z1nci7?Ahk?Xr2N;9ob<7X|?p5Teg|g^KvAw+cr77enb3|p8`sBht9Z}g|Dc>nGjkl zC!h0>H->@aUMA=|hKhd$(27s{F38(q-419%L&--U(HI@aD~i)tCiYJBDduN%lX5wk zKi=ETNK;^fuZ{2$;R~sr6Lgb7CjYQ*Ojk4%9Y-4wM&L^iZFCJ;7g{J_2gZK#Qs#VjJ#@=)c$ zK;;o6wZ3PJLjYFuXBns_-;M&@5^wN!Wv4eaF zfIeTK5Vh%Ym}0d1RIDRU`%%9I!iOf=j}KUR^zr32w&V&=(ashg2suc@H=G{S#4kTo zH)kEaka^Indgw4F$0JkRtP7qG{aXj9)?WB3@$KhIKL zE^^on)0@k)56xaSI%C+L9+u0FwRppFGE<2UmK!W5Ckts#4){yaOH7d0@xc0A9bd2> zMkEHYiP|nL$#oF*z=R{HoCg=lB@GwR7Po@+MtX<1N zKb-J^2PbFcb`@UG)`Z2K3j1drw8WRP9&I&z8MHMc(RG_Uw`WHoIXy4c)kp3f({7kh z)-Ed(#T<03#)k3wFvYmpRib8!HJl5+l6`;zZ1-Uygr^X8hi)3$)Sg*=^DvJ&^{fV4 zAY%ZlcgkX|mweH^h~>WRGouzYX-knsA)Ik=zUOi#dpIl4whLYlI=5r^ibqWMRPy_a zINO44bA@5n9h+p*J$%_cr5C$Ox1xR0S4GLGfUy_vm)#hmU)=1UAMq91fbvWXz5lT(*zUb{;bN51y5RR)kYh|_)Yr*A=N!=&pQGT|@*x9uDvO$;fxN2j)jc!)w46gd%G!3_00tf$LaQk|Sb|xL zN0~ly)JJa2nalUzufB3i>60@*8pc-;{qV&C!ONJ!DXYF5Fr2p%Mjtb87)o8;dFJuP zerQ!!2PQ9FI-OPM^*leitIX}=?%fF(yY;olT_ipU8YPW=SjAFrvxP%j%laJFHl7Lb z6Ys4oy<-?WqU1~zBv)fX_V03 zA#hj!&c#1IX$Dw^JUf77c;dvC&p*Br-2cwa0bTSC04Dtr^sVY;%ArQyo`;TO&;hb_S@71 z>@TZWYpBR0Dhu2^qi!G`W~q6zmKw^l+Ts=B0dQ(cNQyWYE-)zNF+1~!7-7vp*i1NQ z67m~Egbd^5_^-t>hfKObF%5h-+t^}7!q`)DDiyqZzTJh|H(O3NXx(tFWSiRwusa;~ z%o83OOW882)vhD>ah_{rVCR854wgcUr>x1ntdTx3K0n}pF9m~+)A5)w0?5o-crmd4 zHg>#=sU1FZ*^%9}EA^)x7Mz89-4M$QW5w~Eon&jBjRE%ui<{RRSLm+6UVOG0VoVyx zdu9$tdiOu4@YMcA8&xxN(=cS}^;c+=!yHC@rNUKo+rlVdqE&NG;vOwf zlGEI9!?FFKnPGR|j_-1ZF*dclmVozOr+M>2y;J|%@L;zo#Rz=4V-A<)>#KN;K>)Ww zpedMjQL@d4MtpX!(1HS=zCPetR5wp<>DI7C@{{x9PbHve6ctZ&U^uI3+1ZvN1!d|w zg~PnoV}{~$iZJZP>J73coxNaC;iEML+|yptq*&QUF+v>=`=Ujq*@;7%F__5Ah#o)m3p&3t{&Mj zuC5A&5m)?_-UEPs)-Pp4jwxoYmI)OqcXf|xBxVz?M|9WM6PuR$;S#<%LdPnE&P2X)ft#fZhOe0WG6h2X_0?Xx0bZLk7c8Op(kr?%MVLlQrR{j2zgp4>8!Z=6bljY0-HfNc6?s27`2a#KT1tqYN9V|*-SyPz?}Ghaar!8vd7k$cvr+E4A{ zP4~5Lm?l0AH}QEEN|40Ls4|xh$J&1Zy}7V>Vh7^&JD0Ha(BPQH5!0c@JBHnSB@=!*g;D$=(csR3 ze$2WB7v5I@uKT21>d&qr7&`q=`aXZ8e`bunId9=fxo_XTV@zsh{I2j4{dg{nyqx?Z z)?!b`!U;q?_bFO)z<&Np2}M5(G@1ORSZ{{4MaH^luMTVPVX;b9Z7p~dF$i^nPRtj^ z>-_xH7vGzF-J`-zQd-fri=kie(8Avs>Xu6z?muf5sAng=XlWO(;tk?(^|a%awrWZk zDu)2F?kl_aL+u#>>g3h$X`NKm>3l_0N4WkJ zaxIYTdR0@DNd)dfMKA=gc3a%p6h<010qaL5tkOMmXCxsQ3Qn*!{pij4fEasZ-?lq> z*~=KJc^FifyTPk5CT-5s`#Fj^{+MHs!yvW5inoPwbj3VKBM^|(in9mMf$^-u&E%IJ zxi4K~=tE*L17qx@mo;Ym3yp^{hv7#ITv0k$;+w8a2O4~y^|?*P^zxfLT`2kr1U0Gq zKy7brQJWrp1>Wwqx7hamo%WSK*(k17;YbktSxP3%~xPR439e{yxzG+Y{)e!2vEn&VkePPAh z{zQsuohog7Tb@zlBF={@vkZ+*NQB(RzXKn0F|fIK%z8Jox>kk_s+ARPttRG+1Za#p zd9vZR2?~;EFhv{t^_=r~{G2%uSPYuSOG^AkcZ8=Lb|GR$gx*7tryDN_TD%pn%sUc6 zuHJ7t;eCn%z6^UFPBxPLp<(|)^OQLn4(BDT@^~|qc)URfKZ*Bg0A*+Sa&oi1nA7p| z{W9^1(YmwIdt+|iL!;TTVLxzXoSyH}Ysj=HgZv&X|57Bly)Mtqvm;*_4DI@iC{X&E zgs$@ZB6RzXE+d&xHLmCj4x}8txlR+H@bo+Ui)lp-aKbm|hj+9EKh=$-1o( zquC6`kg)@tyd4-8_mbwXJy^@W%ir1f@d{e$Bqbo-UTlf(NAHsIf$pLrIlbHPh+&J4 zzZWk30=IbXc~FF{u_G<5B5H(sB%g^B0bJO@R{PE(scx5^yw{rRwY=gX=E}Q;@Wal+1wBt_08y&m|aoA z48!Vc@q^)yudCbEXPcRu65k(Je+BTY2|wT8L>0M@<$)5)Blm^3N2Kcc?An$lY7>58 z+HT}H97NK=fM^P$TWSF76Pz4{^L$P?k%7o^_i{@258U-X>D*C4QXd3JVco(%aj@Sz z67G`v7n*$&SB}&1E(Wi(#>v}T@C|uaTeSDS0{(bQ={i6wbcrH`+%Mb0{!5is;wn0< zG#y}n>4ozpm1E+xP|A=Stle|74gct8pPD3A z)W+7m1s_8aU@m->#Xq$w?qTz=Xg~1ei^eRP&1qOR+iK7JA;OX52Jt~@!7dv#X-!7i zBzkq4ogps+zM@GkqqBTd@%a*K_F8DS`qSab0Rp(53+WCV`&P;E7I}rS)!TK)1Wr%& zO&u254z`!18q(f)rK(J=gC#${rEarY4-|ToMmtL60xwuF*aO`Q7Z^s~h2a}iQ948u zG7B)z23Q2$n*e}yDPwZ_KisrLd&l4_v;}=Me#0?46Jk%5{@j#Rz~&*#h_-qV3H-xt zzXP_I#LgEC6QAgl?g_mbI+x~|-nwkg3pJXmE$c&%MpI6ceV2iefkXnp?LHQzLr$Rx zW0j?)CCdcJaBS&x!j&>`8F$mOZ z5SWPbumG+24!cI!t@cyo2ZxUitdjkxA9l_idHTAQm8)1cn2wx;w#GYr1#qYY4G|=q zZ@0xBJ?O5_z-q;Ha~bN;03l7TGR?(4;x`q}s{5kAj(b6NyRf}LI6JT7-Mj=JYvC4_ zUdu&_Md_za9gR#MJ6$V3G|8dyN%zrf?UjNK9cW%$o(I6H5aR;`OL*~& z;jB~}SW4U>{o*aq;f@J}Ke2DbmU<*9j=p}lK(R=R<5#&Ag-o003_uk$2exCQA_Ya>l=Zp_?=6BAw zT;J=u4prZ7dE0o0WIxAOCbdJRia#Pll>^`nisxd~{odYH*Q!GqOuN#N0~PFba_5q! zI?d=eCu}qoHtxpqV3OLXDwxl`uuqvE0TdsxO~{>XWu_(?!a}}}m^n~aL`k0IAZ2RNZMZKp zb(8I?6qWfYKNCtj+deFFh=s}0Nhq=qv}dQ4>LRz*D-yjyQ&}9l zSLRs3``)=I$Rbf{bJMC*>|KSB>TUe$DPH0>B$P6Ac!bui(l|_pRpf$*X7yaSNv7Uy z@&2NxEf}M;*E@Q(%&}fZDQ>1CiQ-c{VA!!l1%efEYh2kOGDRVGXr%59<5lG{-qQ>w z$C!6Y2gD25tr2=OV|L&+Ug;h@)aLEXO& z&HiufkSIPc9dj#LJrHLG2d=sXZmGJ;itARWkG{A5_GRuTrP?nX2f*HqSOOaZ*0xdw>UXPfB0g>SS z;lPeqzb9Pf8}q(E4kdyvTbW)Wba5dWvs(f~dWvGjX&Ob9*foWL3TYtkC-@2Dbi=Pl z5+LCAfp|%-K87a#z70=Qs`ns6@wl2kMOO|;J?MftW@K;<1m(ws|e1Oz>G5 zS0%wXwgtZh%_A$Js2gi#rHoshM@jl=KD#w(X7v3<&!=r)%YbxFn)+1c%O|+L+WTG} z`~CY@o)_$gc9n6)A;TOlQpDLJ@8||q;IvEn^@4X{o0N<2$IM6l+)P#l22vSaWr^pltsYjEMxLM6f@`3b zM>4?IEK`gAjQDR!)EP$4=;TOS<++xL!?39?T}UBLMf6>&Dg7* zFRfQ{`)n>63m&MCkw?sKH2A%}B%gooNgXaTLJJu4j@PBG9`^*sl-f**Q)(OyB-w=Q z0Vyjc>hQ(dhvjfj&bebcFj3Qs)szR98g=w-yJiNy!Dt{!I0s*{E=g8V#2r;mB8W$_ z^--UwZ(YG#-dDkyx=m7gr=5WFIVi>;%iZNvU(~?rO-!ej ziy9MVh+hGL6HwsSAHc(MF8xQk;HhQBCNfh4Gv15YoQl-i%~r^>1w|Jx76jsBDk?CskPmw&w;mx%&wZHmws6 zsC-)FZx7pf4#_r7w<0XISe>ufbe@th`>iOcg($2RPg;w9#)TFdhJG zRsqA+VU{A8ADm}0mEaI|)GI~zsAt?=u1_r113gCf;+_KQ6KnYu@tm$zo-6squTW!r zbgG#es$U-H3=MzLZL+c=Z6Y@Th_l$rKcDC+2nSEf^m35*KzVE@=EAnFN6f&O=v`=} z&OKP0m)#2wu8d^lyV?7SwysG6F@-dtskXTmiJV>_bQ(TYBG#gkj7;NitMg~uKc3zF zZtiz#34ggu^%4TbZaZP6ho;EQEjiJV$PZT1EPbK8gK;^u&oNh@tS%HvMWv-I7Nw;{ zQl9>X=%Xv4V2`8MF6@UWZ;ZX$i|X;g zT1))a;Q`8K;Mt4|@j?7+03!qpU{-f1I-K1;p?p7N@%mR8xVP9tAVWLX#C5AhP$*ha`xj$^uLJ;UvKOa}N;c5iJ~?H<`%4teT8c?S zh8kU+oynn74GL4%Efh$;WT1I2o5SRJc!SF-GQwpVVQ%4?$&n(}w1Q5(ICpW>2V}U> zu$DJvrBEEOhg$TPF9}R~?HaS16)Hn_kz181nH^xCfjD-~$P&mwgEFD`!r1bkaLOQ+ zV7X6Py%s|F)&fxC#Z|wtkp9TZR+!JDD<3)KZYAQQtW&JAzsR^iFFU7FpY!CLuX4#X zFsE@y&#-KtZF-W!kMUY%SFo)GIRuR8o;fCX+BhVSwa>Og>7K`r>AjXWs1xiuyjS@` z5=&o>nP~5-iN8r$y|)y-L(#GwjC^_TkxJcWmc=W02t(kbSI>;_y%W+5fdA@~_~Ctg zUvu}@Of6Tsi)&mVev>oZ5dZI>$txdBMoS%#b8yhq^uC>GrTsyC3deU4at6PUs;(V_ zg;asC@cMU9d;yuhzR6NNb3u;-SX4RA6uLgUmHed~1Xz)RkYZ$$T@4=!CR^6*)G}V~ zuy<(PD`nikZ!6pUe9^gN{H~Im^Q;}2p6^$Uj~(OBk}FW4W*dTqpj*9G-$Co}9Jc3G zH?OoopUe+4rEW69SHuL#t)7O1S^Nmd@u`o?+V#2xE5$v*uMS$?HXd3T`Kke9k;fre z&bG4SD+?dR3!sY=A;hvz1!-6YeB&V6ygt(0Ql{ek1(W<)=-3{js8400=1x(~e~S#S16r3`*63K{a^M{t419R$P6nH74*tA~Mkw0E z!2D+BQGk$A>0>9%-l6k?(ut%mmbbNGg{$%pig%G98X1)od=SHnGt=GQs(TJRLOgLI z9zY~l(OIYMS=^YPQPHEtLQAAfdt%W%rO73TiP2E5tYfX@bOV)Qx8j?waJ<6cQ3@V zOMTcNkIH&kubBBte_G>|r&E~LDIq2$ytRik*Vh@h`%=H9sFu+j2Jg;W&gurNYkHhN z8F)^!*-nq+vwusUPUxw|teS2d#F6x`xe!7YTpa~wr>1iu+OgW?0y<-v_<*EN<`qVXT!30H_J<>f zE@!QK`pk8aj@c4c&)*&7&vfC6Zc3j|UEH=hmp0?vX&IA~`S4J?hlFu+^@y{nL5m=9 z$gFzCvm>#um8}6xxLKaaypu?BH~CdWIBtNwTO7D(Auk!0Cd?6k{00Niko|D;{&eGl zo#Y$!UxYj`$LJ$_9(ZS&`W5$ zP!xVI>yB#iz^sQ`gBFd?VwV`dCTxBR7W&z7d1QVNRZ=Ag&^d^PE=}?9e9o}nBYnDc z^C?(4*_={@C?DQ)nY!cA>!W#K2#Mgoa|R{FeDAJ2RV`VIW`$*(sKL*-H3Ml=f@d?A z1|JcZ?2)4$>yPLSBgx7gC*0?sI#HQMoQR1(hu-AE+rjkWFh9Q3S7nmfl;`}OqExM7 z@ILOW>X-Msc?x&6yO`L4gbA8Z^8!7zKZ8isZ&C#y#3Z3LQfB3zkzD3qovcSRM<1tq zXUOF!WgW$VN6Okt#c_lI`Zz>>qXAu=$yzy;D*(5%GS1k05~8-$KMEoT^wJAJ$NW=z z{G;&q(=i>?Sk_TUx`L}QfA+-)hL_WV+!b-Dlmsr^t?vtbiQ>x<6dYw)S5vz5TG7{3 z^#GDk=F-jlKroL_sHMPGNpFY>S4uL15V*ENq5dem3x7(hKE^MgxXvt*kyp9Ib9}5|hL5qfrdW1uPknPO z@xURh#Vo1uOZ;?-vUj&~5^^2f%OemAaPTIzlY$-smJD6ZSZu73KUGROnHH24`tPE& z!nsuvTOJEwh!r%Q19KxY;ru$UKzaQAhYH!wgQ^}Rb=WC|YA0Ol_w==WH}sYn$#{dS zlfv5S7QVQM2mrJ>iD1*D#m%8FwSYYG{Y-+I6gg0<=}Xp8VvhCSR2*r$@wCH30fG8$ zrt<5ye4Kp0V6ZN7={dOJu*s6Ld#l5myETVz9W7NEY3UW*db9=EpJ+K_M1ey<)WsuZr)Cl{(>&-6#2a zSJ`|c8+2ft$l{}|PoGRXnJPKi0m+`KsJ^@QMnov>UE4N3Mh(eBFl>L!P-tDcSJ77~ z=RM>Y^9Hz2jL*q#Qxd$1zbaIM_GP7>5rTXN!Ly$(GKM6Xj65U9 z=!T&kU5mZV3+`o!<-BcrA}6i)IVLKwip}_OpGk2|z8}60@gPA@NtJ4{MO2hyMP#4@ z9aYIkFBbjQj{{`Rh7DtvP6Qa8nLTj$6Yr5uRk{ZILTG@9ve&d!UL)F0}_%4@B1kV+nX8xBL`MrXloCgNA zdaKmZ_XF?sekiJXWB&QvQI8xUJ66|X7L#ISii7E{TFkZ^J#S+?R~aPOdS}tqO>Akcf?>YA_Ir?m2h)UyHb>_Q$`ysDgP zm#%jW-}kIFvBN5v`?k+Szu^H(-^hDu>CR_Xa%FC`7 z9wTnHvvhX!E^q$0fEZFR}*>dSnnG5k2MhNverEj0pU=6uGl>-l_5yg zcA0$Fqj0?)k&?=6Wx5i|P#I%A(>ci-*IX`pa7De}#M@N%9~jmO6IFs)ivdkBcLB*W z-)cb$4a`IgsR{gLgxr#2u>6?vn&I5%X-QbND{(TMt5fK74E&?lOyw`QVxd>!r0*_C zV=Wp;(@)DGu7_G*tGa1&ORDZQ;z$bl9n^S`8)M@aFKT&&T=l8C>@YI&7-=p-PNVEM z(Hl*%kxmpMQMi(*I)Q@FPo{%aP<0e zf@^zU5s#eVz?k?Qm6*F%b%m#e`4iAtX!Kr1M!;lsdC~RTmVS@d(t;f*u3L@Gm1W+c z%N8AmH~~{yrjioh9tvN)0pY%foI&bEs=9PPih(FPoiiE+BrX+>-*AcItykgYo?n!r zav1$}l&MG|^;JtOe}`fo?fN)xlk&)cNLw5k6#Ukzf?B*?yQ_*Lb3a2B?Af3*6gPl2 zto1IDaaFfXRn3-70(>p8(m*n1X4pOjkm0*Pz{gNG&iLV){nzzt-7s2dKDrEd_^}Sc zkh$&TgdJR|r`Mcc|G_0TF{4gQ>q74P<~&+LU$xiJtx7w{ZNsFi6zxl{{q6mGXw6{X zGREb{V!6_X7Gr@adW!{ZJRK>5z49Q`5!E83)sN*?`pUtz4;$yzDhp)W$$;T1)^^}{<-L_?>9Y@s|qXq3qmnIX@^Fu*31#|BSIz4h)w&^ zOIy8K-rpeO^Sxbg<>)a8n!8(uVlMXd37xWtS7V70cKjN~X)P){9h9Kxr}Ck+*V44I z{K@P3w6_V0Q$;zV>(mzSmt&rr+m}Hc;@%679}lbGdYmauF=Qy@H2YVIAm@2`GXV8m zn8NZ3J7#-(pOE9aczqz9Jpbn=5dOC4DnxTX(EY#~-U&@gv5Mcm#Q%-ASV&k9z;u&g z&#y-Lnb?F$1WOL0s#)ul0Sk-i;vyeF6z-MHj*0Y1gdWrPhPBpG+*RGadE)1WP&`!* zq?{nfqw~?Vvfg%lE1Zg!p{e-=BEEpj&R#mgLn^~27rLiwXISS%rkfoJz4-*RyMv_q z_ewDMX&)OJl5kZ&YR(-a$d&lUet##KGv#Q}aJj&8Q@(%tqOsby60Go0z3(6| zK^_~~jlpP2e`~92EYOH)%ZnX+A+^-n(r4T(w#756cn5>4-J$OJ_G5zOiu38cIc(Rh-Vgl0$_`0K{o6G#= z-Ru$+oB!cW4t>r;3rgIy;$`}l<7JXr{kI?xU=0l0T`G+d%Bmb%`Y_=hfL856X=i09 zJJ^~R7(fVILbU7Fnx-`ulFM0oofI4WfxIC7!^b?stY$Y@q(u3w#`R4UqVmlL#U35N zKgdwre+p$qlPUle)Pw7%a_$Mh_UEds6M08=&l2Tp0~sh!El>T&x6^Eq4gW&S&-@Wk z@{1>2orh}EZUMX}r28sQi+;vg7p*rZtb4%L!D!mwo~Od&g-4-Q>@0jtLY}RBCmV3e z5Cf2Iz<889!J?n#YvhgAx)of6;fCk)iH&{OatqNM;wJxyLd&^btS&7hz_<9)50Iq> zy`g4lxBVh7@}=*kV{}7Hww44}56=$n02o!AjeJ@a=E1kDwUW`&X(d!v99^9UC{TNN=2W$v_Klqx~cJNl04lV*KumR-t5H&E7R@e z{ZNH+x#(TsYcB_5o@uo3uB}XbC=Nr-0d^WV)Q}{=o{8Oeul9lrgSWk-{>?X$R*{$E zmklEHh>AP`%_UiSxiQVKVu*2&CjcYi?-VSSl`HDja<1BYd?3SLTD)4xz0drh$I5c> z9`v4W;+%aMt^V8H<$+4whCz3(H0KxG29OW|lE_D2^*t}@plRd+1ia}kSmclCa(txX z*!KAja(2i>tk@Fu1tHifJX4M?O}CRc5K7H0eB0T zHM^;95?7lMZe=gnx^Lel;V*sYk(2GqKHs4Q6G@=`)!2<}fgN+t{}606nS)bS`PxQI?!pO#jJ%<=fzFurxGk7K>I?sLe;G!& zA!pogQx{KI)yF$n&%o&^v0Gz~2;sn6Z9=%>VMu`<;)5)&H=P5_c(Q1MI5R(DCj6{o z!w8*9b=pYJ4t=4AY9=QxM_KUV>ICxQE~-2Xg5+HAfXqpDlr$6KF<eVcV=G zGeZy>ljdN zw6~6x`U{qe9X+s(V64MWn8b|itsRVdcCBncu$(zwc)#!^U9Jf*iu-bAC4edc9CjBj z}_2^Z|*6YSkp2nn!?rm~_OR{~JCiAL*O{jBk>U z88^~3FFTHmPrs!9C^)dMRV&0mnZJb?^r+kBUTwRzC+H;IWho|^krwxoi-vt>&twU9 z|Ht_Z5EW@zpq%8P(i{^n_=#KwU%6AH*QM$&MQ$rtbb0eeX3RWZZ5pU%D7W!{(qcQ2 zvoNjsSkzgybBtoxD~o;brCNa%+g`LL{E1C|o3pG9>p?x*YxKcdY{Hej3pLTY zjY0GZd19L+f$I*j=nX(jt@XK#&2rCVtzL8+2;1&j(~bpYG}nBn0AsQm4UJjjJqyoY z+#cEiw};r#$2QTk&+fSKrWtM|c;_D8urxgK*r{rNdRWM~fjq18N7^;gAkYa$gfx~e zm6fmW1j(I`pY!6}4{5ObJME@5i+bIKv;5^hyk6j5Lu9bw)Fl zH9OGfD5@W&_VL2t$`N{~Nb|%yTg{m%m2~(_F&xvR(oD^MDICptd_BAJhf3L*n_Nm3 z9k@fQ9hyS8HEuCGv0Pkk1nQ;5K`dO|GWN#rKWG*`I$Ez=UoyL&A(3O;04(<7A1nE@ z)$@=0{{+^ih(o{}MTq0~qPBBwGbhN}7^OI)0uHr|CIiU?+t%aC;MlY8j<{ZRSMg=- zH>rZ{8fYU&=m*f^ZyZ@v&Rx*4_zognk6b%cFim8BEG!~W*Eg8{^C@hD zk+WZ?kkR{r{tcX0bH@!SY{3+{%LOlPH#aZdKLeVcDdEnZ4gS=j9ZQAJTIl4mS>0#{ zO9E~v>6YE@aRf>eV*W%jKn?QPLo?QLN*r25+GRIT-_SQA) zN5Gx)&l1spNH9R*`P)pp#|wX^N`D&Wfq|WteUzaV1=ua#_W-MGn!RFyuYV(`TSQmu z8opTpf(>i#zS76V)bhS!1&P5NrERgij&UO?axuP%oY))yv5r3OcK;3%cv8xB;Xo_V z3j`v>09YKq-|@fl$PKPk_qi(|S|;1CzqOifLe0}liTgO9Btl`y(C#6W?8OUD-NKhF zoGlTFWsy_t5dduXp04=x6KYTOHh*>zm)nppS7i8&@|u*svJ#hNq5ir(Owl(;GR2YJ ze+W%qu(h+gu)QWLjK-9g_qFRg_ZBGwPO9)OdqP0Nunn&Lvyc4;$IlR~IU^C70Fh7GgGn@%T|1jF zTnkFA(Ff3Oc5%GK>Xf^7mHnv}bDVt)PL$+UM(?b5cnqR{)rkWHx05<4KAYlBCWshB zUUUZ1e|mgnlQvPrItaTlk43C$W~GYroDSx0wIZ2Q7>85p9b(!^ci{- zuB{o&9nAQZMTDrm@%?fk=A?XPadQ}6%vD8P4V ziGa(gtSrJHEv6Z@7v;u0rvMHWNCn8cmSVh6VEXRi!gfN} zoj>GCJqzBXKE_Xb>xB!5Wvs=tJdiM})O}lAz@E%wn=pC1*-%Z2zM9-0PwLXl^OoJe z?|k#`U|qgSdpoAB-nAs_$M*|bFdz*>L! z>5avhAD^Q78}9$Vy*=|}vr~Bh3-Noh{w-<$Ypd$5-HWuD$$znS>zWC*^I#a~CiNvG zu5QmRBfc`=_+KEqA@r#HgLM7x zL1-KC@u;^3E8R3}0@7MihXnY?l$@B$9mCW*mOKTtOLO)Pk7|-eq+ERx@-cbH)>*{SQ#atbS$=5 zFX{)({MHTH8c>aK&(^NSD2w1YPGltJ66sEnP#r4w?lh-v^VmnGo|D(*!$gnQwi{i2V+cJ>iUX9DV{H?O;itaj-{R*#(FSrw zY-+*3g9=KZD0@$6zuA;h$=$ef+U1W@bJiS*Uo;zET zA#_u4F2`|u{GP-^<<^?o3X4ChW;kEv6 z&@-zCYWMbRopT6o0SSMan95USFH>lQB!UI)fs0$d_76_)x6?ecWy4|x+I2uT`t;8} z%a6y;lq7#=rMZGPtEgfshs-0B>g}3ooaix$`-dJpG!vdHsT)8GQrtxjjHTWe+nnV? zit(n6pim#tlh4{qTkR*K7$hmtCbmT>dTeq`ymR(~*fackI&Fjx-p2K1mn~%sc(FiDq?AOwL}vPxJPeg9D`60~ z%-LuyTlHHXk>V260fr98m6>{2dw$VHjI9;HWOXrSeiK`{JhcpKE+)GI7zjO*mydLw zU3N}-&XU8qy{f~+Q}Ozz$6Aa5i6m0VKj-`Z3`hUXYR-4(wXaJjEg!YEdbLxFQ8%BE)CkDd0me5 zgPqPenSC2xLs5xE_wvnaOo6Q|*3>hv0ig#(b0ccVUR?3Bk`UC4F zs&fpDBIH_Ijy^zz?2v%kx!qc+IF@^sO>QKGb%NR^+2dC&hvWj;qTF}`j(nDhVSb~(VJfhlGufKhF zJg|m!@x2n1MN{ZQwT5jAL#ksUiTpxQzoYucCPAKZ*^y!AOQyue|hZ6Cn7C-#Snw2t^t~ z_q15%T9x^5vk=*!S+UzWV^9XJu0G{sZ|BZ)p|ANWGjAPnWo?r=gO{s$?=QNe( zX~K-MTqRE&XV(EE0TGP~dE(mykUkX<4i2v3Z@t6zH-b>B%@CF1M(pM(#!9NcU~)5;0v->B!cez_bm&pGfR!I3O5E;Gfj96H*dQ z9yW{i5!AXpxZ4)2VCfJ?kBE`s4r_F}+++;PMoOw|&bT@v#NSj1B_FGPjAU2h@?gh$ zpPacU641}BIls~asOU*meGmGVKn?weTFDD`cn$;$Q*VM+mTY%&IV4!2rBnn~DZ_e8 zWM$Kj9(kHs33%Mu7#&^Q084VoBJ!?Tt+56PM6OowyU~>XjJKQ=y8l!{@DHxA z-|Fo@+v3I_>XjKyep9-7r*>&pWx7{z1RA*vO+yVQ);5u59%hvu=%;7Q9A#f&3;Wh9 z9;}B2!jPgftAB5_vEusT26v>>`O8LM%o2J^eXC7~f`j(Re74?3c3 zGe!yPL?|p~?0Nk<)0f`sjaiTy+_u0g2>|rAUZOS0F{%nJs$bPI6G0%6-)>?}Gk+uK zp|~e0Vd0&RB(+70VToaE(%pbrYU3eK_04Da&*_f0&eI4nycI+ZVsD^Yidu@7xfwZl z&)PwF9Aa}+y9ky;5{2oAC$Uri1MCLS`P<|*chINkSA?EfFUiQ)R73+Kl| z@vojO`BM{iHv9dT`*IXjDur*SHLg9aCrHloBPXIRyUf_VM+@VEsd%hJ{B&MnxyQNK8sjNqzY$=gr&P zy!>|sg=HVhD=G<9)iup6t!mL{#8XoyRIx{=>V}4bJXP2_K z|Az|%fd9s#KmP{%7Z*33i-D05%n13z1!C}{JD8i1NkWmC=d3B@wkNNo(i0Zwx%lkT zCRQost0cZVUcGGm(ke5@$bZoOBKyz4LjGT5{{!~FT%!O77({=0U~T{f?Cv9Sg4qzo z0@o?|Zkf>_+bE$D=1Fk4;FIeK0ruX|#M3tKKFT>*Z8VIv!EHT@4!b zaAambk6Kq_-S=F51#G>N>-_wTY-_%^u4Vq#xiDxxBhztl?xzDaJmnA1@os_EgA(Q6 z^-5}XL?6Ikn{VM;CV%K`ZZ5z3M{(5lA z2Rd3DH1Z-_?B(Z2Xo#Oa6b<_0#t9c@qUMtZrSwmBwS#nj8vQykxR8OWttHz<;vb8- zT&*)gvZvpfN4DMP8i%i{!&8451J^RI4pL9vGaGUCOMy9su8y=neAlW*44CMX^jBzH z%t%@i6C(}9&$uai|Au|lbcUQNcTA%J*w^pj_#6FO2gY(Hgb$@p#48AAMNG3kp=CP( zsHr}{CqmMD2yXJ^xx^7|J{PYi2Oc=;ADs{lY;>A6BTHJwqI_ro!^^wx-5oNSiktR~Yncq@RBnDnXmyzzu5-Uj z=p*bU!x&1<&8Y&fMLvpEQT66a&;&ww-=yXvXZvu$0a-&>m^nBL@a8)P1x)p!fW^(m zna<;FF#F~GwnRUR)ea6F2c5!+YX{K-Il)>va;tVT@^6^(O0k}W7&Tw+ra;7mTE;~3 zIb?q0JT78TGFdb7sF&qu6XtLuti8kyBayljD`?9A2$+5{cVTk5ytFER#`@%pSAR+1Nc(J3xVa}blEr0A?+ECHv1lESu^balRsP!}K*<<8 z+$@JxJ%utS=g!iAJWR;yJlemVfhRIEoNFd}Nk{soYFtau(XOZO1H;cV`i`s70Gp4~ zk5u~)Az3y#Pd=JNXi&+zR#b-`=HF0LT@ix<^=GOjLd5(fWE$-53ws{++K5W)S4OWF zMzZ$h$M>D;(d6X!g%mv_BjGmn6#pb5*22Qnx~T!nrOVs}DJaftMXtq9ryp6rRs-Vu zaZ>8aO{{IsyR|Ej9VOkhfxwtySF(h#2jOMJy z;*@(|;7UcwDUw0%?U`=g0KE|EhOnk^ps76J+l=YuuU6Y@zBsuLq^TJ1b!fZxP#XLI zslG8+^LqD=uc+^dMd#%gILt3+a86FnfpO#E?lbBEG%#<9`lyWt1V#hN(a%=t?cSVJ zIQV6{^Ezm6UEuc)g26^sp?r)6bYmLVOASZeXux8f-R?-%kxR45wK`9xInUW@{jOYb zORwb8{GsWk{b}5=3ZKbJc?@UASYGM9)x`fH%oqN7F~y@T3Ki=)#q%}}O;!)y`miE> zb%kMOKqPdYW0RgAwp9q6D7#Q<(07I{l3h_dIRDziCegvVGwduOc64^f zchf^bdVcMgwvkRoNHCHmDM8xup;Ql(Uk5L zW1B37y2aV;qDmW`^f`73YUa#Jd6ghl!zwIZCqjzETn{VNvcFps5mG@kM4zwZMRXUp zgzn7-tKy5LYAeDQ_$QzpWfI$r4yHGGXQrb3d~XW%;Tbz4PtL^U@1nVu^#T8gR^r`S ziqkCOIgfr~!mXe(^`MK5)3v$vB%M&Nr`$4`AdO!ErSW?U6~*{BbssZ^Zr7Z6<2>$P zAc*R!;tC!18l$*xHV@~+A|lp)lqUvZ+4&8{+NHT1v$*sNk)KQLvzFCi8M(+NSNGcV zu|2M<{vgZ3yzJcN{d&Up3OQlhvB0Lf_o){89ZL^B#q>ZjF!mSg$PKjuu>qI3!@I=1 z-PH72jc9;&S`cO%PAwcCJ~Yn%3G0jqQ&|~VY^Y6D{rYL9E|qs7u+{sxu;=gsUV(B} z3r+*h!`MR$vJaMkl(52oeSxb_l|mYE#aoGdqJ5YWu`{4mnHcjW_mm4HQ||>Q;Epm#RK)Hj&B$pmdQIDJ>(ampp(a z(u#};;UpOIeE!pyQQ>iUriZW6*JIoR42F3q46X-qM#^_$3&#wy;`Z)+H6r>ut;Icd zt?F?YnD52U4@n%g`%*psREdz3EG;uXJ$=im0Rc8BJ+$_&)HyP=mKI!Ub>;9t0w2BE z9i!gY7|OXX^T({uDe~1_;^Er42B4ClpgVOfpQBJu0$e=i#;it-<(GMgwo%L%i*=16s-_uQTR#Jj%T8o1OS5VER9Iou4G#ETV)K*g#@<(B6K2f@ zQ)z(2*rqf%>%3L3DOC2|C-i9`_pj7{(*NxS=&$C%vIZ>LE5te}-g5{QsBD@ zeOK3`M53IVDK!wmfX%~A3dy|E#~}u4Yekp{AVYPU1@tSKgk|hk|n5=Hf&t zg`89&QhIuBEs0PQyHRz0cZB|Lf+RSRYK*V!h#I!T=G!CX?cJ804O3i%9~mI2FV{Gs gXh@DBOdr+`la&Mt3r_>00Q4s!iy<13#YOx6UpG!hMF0Q* literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-statistical-learning/images/image-11-$$ 5.1 Usi.jpg b/apps/api/app/data/demo_documents/stem-statistical-learning/images/image-11-$$ 5.1 Usi.jpg new file mode 100644 index 0000000000000000000000000000000000000000..b48a9f9bfc49e254dbd60db7a654d72745dbaabb GIT binary patch literal 869 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}n(iT@hcH)plmvORxPqMSR*mnd?ip@90hK za^2p~*JH%-^LLZ>^4Wh~=D(Z&G$M}WKf|QThwFc{M*U}av2+1j{h>(nzuZ#)8U8UJ z|Ict-%iCVmy?9l0=I57IOQ&qN-m`31XJ~ZRJDWD!*8J<1YyUHxc>1vZ`@8xtp$sL$ z^(!X-z5Hi2_ksDJ{B2DCGbF5z|0}X}&*PGu^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}njBlO`z^@o@xqJ+je#c>irYN5-z+YdhDQ zp1s>5d@4L>0@GWmjgM`Ld>P6^!Y%(ZaFob@k-Gk$;X}IqkE_DV?K&Ub<>5^*@u(^VZ+c0_t literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-statistical-learning/images/image-3-$$ # Chapt.jpg b/apps/api/app/data/demo_documents/stem-statistical-learning/images/image-3-$$ # Chapt.jpg new file mode 100644 index 0000000000000000000000000000000000000000..ecc506ca2914e67af177cfd09d31f81fb5c55df5 GIT binary patch literal 847 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}njBlQu@oZ4X$|_2`fJ^6XpZTetKE)l0rQ zY5I1GxsXoTwQ6T8Hia~v$7&7MRoZj@Gqe;({%2Tt8%P|;*8gW1>Xh}%zRM&#a*uQM zire>QwoJTpIB!8t&xV72rna@e1K2~&e|6g5wLi6C59fadmCuLczj1B-&v4N@dUO4X z$$u~ZS>600z0XEF^FcoU?U*0^-fa~nw=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<c1}I=;VrF4wW9Q)H;sz?% zD!{d!pzFb!U9xX3zTPI5o8roG<0MW4oqZMDikqloVbuf*=gfJ(V&YTRE(2~ znmD<{#3dx9RMpfqG__1j&CD$#!}njBlVuzgWy|J2jWKVjQC|GYj{Cy;b(g+H z&z@^_FEhoe^?=S>sYw-ePi8T^*H~BlPr#S?UqxK~gJX~XGxUX(%+j6K`k0@;*Z!^H z+QTuNq3@PXFc1{7D6@URym9tlm-+AJKaGfE`Oh$^^5Od5tWp0NUMyXE(=j84aZ*TQ zjDPDMN4?1pWyO7WefYOWPgh5wDPWsQ!%L}^2hZDdF}%y3bL~IFlcEp%e}A|CrPcq# z`})5Iss9ZB7|efMRjxPN{^%@U#+JE$_oSVzy;ZMn>~(u_LS}vy!yAo7T?`=n|0V!& CI652v literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-statistical-learning/images/image-5-$$ We woul.jpg b/apps/api/app/data/demo_documents/stem-statistical-learning/images/image-5-$$ We woul.jpg new file mode 100644 index 0000000000000000000000000000000000000000..882731033d2be2a145ca13c44b100d23c9e9c21b GIT binary patch literal 802 zcmex=^(PF6}rMnOeST|r4lSw=>~TvNxu(8R<|EI`$@KzRlhK~^C}Lq|5@z(jVXLJ_0Ji3>TDoi-j64Z8S2#W<;` ziIYoATtZSxRZU$(Q_IBE%-q7#%Gt%$&E3P(D>x)HEIcAIDmf)JEj=SMtGJ}Jth}PK zs=1}Lt-YhOYtrN?Q>RUzF>}_U#Y>hhTfSoDs!f}>Y~8kf$Ie}c4j(ys?D&b3r!HN- za`oEv8#iw~eDwIq(`V0LynOZX)8{W=zkUDl^B2fpj10^WZvjb&$I$#G2=o^d3kx#~ zJIG&*Oyxk#EXcyDXviky7|5PjD6C}E$RXl1apA^;oXW;QA4HRiE^>*fm^@Vd2=W@( zXT*7|i7cPNJ%;etEe0NDMquPI3o_Uo)w=G% zoAx&SIK|O+|+@(fll}8y$KG(09{P*&o)!YZ>fAY66{m+oFI{vT7*4^g6I_>Y; zpW3j8^FM>i=fm;exVHXhxajR5%UV3^{-3@$@&62U&;K*HhP%f++y0+Hd3x)AhWk5# KND5uQbk092#A0{C_#EP0s@LkkRnZqAYGbN z73n?nP^6Pc3y?Co@7+89H}898W_G@novd?mvVME7zA59BIpCzBu7NH90s#Ps`T{7E zfHuIyz`)2r&&0^c$jr>d!p6hS#>&ceii?|rM_3RnA}o0N^citE$upudVy90_sY=Tz zC_*3*u;gX+OG;{T7a>Z2JOsqd%*@8hCcw@vpmg^1S*8EyL1_UvnSf1TlNKZj&~Soi zIYE?m08EvW9`v^X{5e20v~=_gj7-cdtkfH-P69L_T3Q-9T6%gqI_lk_)aw8pCq36$ zMJ)zyGiOFoA0DNrDesuXu9bh~H6OrGKd}71c|ZuV`P_ z(bdy8FtoU3X=QD5+t%ft>wPzv`vYG;|0e-~LGZ|^=$P2J_=ME77wH+9FJEQl=H(Z> zFDxo9srX!3Rb5kCSKr#!-qG3h?fZ|xq2ZCyvGEDaJa%DmY5CX6>Kgv{&hFm+0palI z4_qLC_V2K$?|%pOPq;X#xM=9;Xz3XLzy+cSpb{-79sOBF1}-f#MrR*xQKhF$Jl9g* zm49UxyJ(K%z5958g-;wZFMJUKcD@obQLp_HNBcSdXTX>s4sp(BUYHB%07oCLF!-7`NmCB_G{RI z^_VGUvQSoLkk0Zbtpl*hZCFDjVmwEFX3Pi-_#-$CW7#Fr>rNK$%E3c@5@2}M_G5JR zY>SmuQ4r0#h3-Q{ix2WgxVlKfP9%0aZWOHX4u_X>1_yoc=eZBoSA7YQ&>sT4u zi$gc4hrzW=HJ#aay?kb%fF@RR_Ly+w4ML)m}!sUttLH>8I>&VrHCfyr;Kod z?%%%zoXF1t-j%J_2gK|K?KxP@->HgneYxi{p}A$^?C8soj&({X$;{eFgN5;=Eyv!7 zE?37qQL1rO_c`)=05NhR*k;S5wbv&-B=ZkRD7vdugm%BWMI5R)IJJ8Cb_GOUEy4(@T_zuAreIe#Uk(zD+{jHotF0_4Kz54A z8saoOy9#41v_JvqDL|`nVZnYD1#rdlp^m3D@FzyuE{}KBXTRC)4b=iRiprc?!sJZ& z3I1OB$c>})&#!}z%@z88HYjx6fg`(%l!r}0Iz0bCyz$ss?g@f+TYzoH_Rn%Qos&>V z1-L$P3ia*Yeh%BVAOSnunqr5sn^egwb5iM+y&|^r4bd<=ZMIGU#>5zn3En|iq2bXc zQHQd^Czm5{#HF}D;CP5QwI_*mn11l9t&eF66CYxURo-bZDMl(;ArAeMNaN4JU!EqB z9ZH*Wm^?jYKX|5M24^k=iV5iFY9{23pP)C^;Q=g#t#yZ#yDz$zjN8rFq2{s`TEJ%6 zu~U|7QncN2A(B{dthrGUK}4^4pyFO_znp#b!0=0QbW7PnYt#85HRoKt;?>mD z*rg1~`T4MwwsJ7qb`mXz&(2hRs=SBakEA7~zd85Cm47-*6fs7>T4(Sbiae+QtwAh< zg(f`>TW5}XU9&J)MR9X^U2`RNn;e=33V@FGG7 zJhZOKYlmgX%~q_!eZ}oQo;7T`H|y5S#R{VUjgsFKch?Y{4oi<`m~eiMJVOOR%c@x& z3qdsp=x(F48_)^iXnzP`ro#hT4oIW`OaYy*?b?MPeZ1QHLE$SkKBP{f7VX+kkD&{_ z8p`i#aF*3o0;GLpv7Mu?UQuZCu>!HQp=m@0(#;)=2x71L)V1yUsI*V3&pcF$W}0Vw z%1H3ryNFM0s?rvHMQRED0FRl@--ve^j-l$kGd|IHK=F$)Zc*Fx3uB(Widq3wD~4#> zsMIh!f9E*tng#qIjir{}$F z!qkY|^Sz`Nb*QV?Y`F0tJH226R`1&Mwy@=s?PQMrf@9ke3SOs4ucyXoKlz;w|Ka(M zXkmeSN+}T>^X6unBA&Ri+kQrG16pdlejdqw57zZUd&vJpWdu=xVByk1kDfgcDHBft z^s_bitp_ElpTjHfHHFmz*7>F&$497IQf@v~FnJN7Lnw0>1>o`+Snmmew_PJ7&oy&H zzds!O{@y_!4mv}c39Yap^MF=wQJ39ULm4049OYN(!&yIE_WDP8#ZAW`4lShC)CdP{ zpvHoMM{=;!3kO1`*9PlHlOD7DU~cvR>qs-^rH;=VThHZlWl~gT zh&NK_9I%z;)@2k;7w?@5y@E2>m;JD;kG==3xUV%mH^9nxW=C;rBuze!17LkFoUg3 z`_M0ih*oJ|>OVL(QvifB1z4~g7|=ZZizu22-BqS~)PsF6Sy}J@{OTag1M2 zItPI{hr=*8;Th~_zk%QnJimFaKECepm81wI^$vilb4d+)MIM4(#IqtinSx2BZ?7E& z*$J39BAWEH$a)3^t8mqDHSzwhaWW6FN<9})S=!HTBcDL$E{I(xc!OiP=Kavz>(7N= z-S}L`m#X$G_%nnK)th;X;CbSU@)r*gNj&|o!jB44mccuH`{Hj?#axk^d}wdAYex(T z&Q)1Pi4zO_AH6I^X3l*V_gZ+y0N>m0adUKXBzR+0BJ(jbQ`7E8mA=fLW%raSB^a(J z^ke|LXHqZZ;q?+AeM~TP@dIBO9~v&mwpexow2!71%|iIndW24Rmpy5Lw8o3PdpXs0 zIxge+`HTx9OJJ#%z9*6zEwn=`xzq@baWn8;JwI#nNB4-Gz zA0J5XU{>I8_qBzbSChUjQz1V*28)0)L?>C$gpU9<8+Z-NfbG!M;0zny5nKAIJ#uox zFKx$)oJVj427GI_O0*ji+Q>@c3SCMMoZfxtww{9dBDjM-73ga{s7wI}H$x-@GR^ec z;m_V;jyPAz~}$UUyICZ0}{aq zCOdt8v*8zaW)v>}&fQ;4BkZ=^VQm5o8tkx}c3AqKav6^c$07dDeHWzt z10O8wLgo#K+hYy3P(FCt(|Yno3m%MzIN~IZ zBAU)b@U56HiHH`?KBD;o?``Ubcp4`r2?3eJvQAF_}3o$N_#?&sdnfvYD+zklV* zo1R>0H98o87MpNnf0syY=aEEtG=}7 z6$iz%FuW^rAjFV8xv#&StBJ0DaOFklUG@36gUpfoGZerB@LoL%eV^{ko}9!?UwNTQ z%rP8cF_rcm0uruHDYSKGEzovj`eZ|F{+vElU}CDZp6#1mw_MAL1U$02n=(xaK}Fi$ zbomWOviSSezH6~1;GO4}&^L0!)kdL-E_zTmd||A=I?H^Zdxn_$m*7(LZ-C_ScoLiW zw5(q(?A#|)T?Bmj;-LaV{U9w36P*Z?GSX)r%@)zBlt5 zehybgx^yE(P9ncq6#zl|iMWX^Wy>+5XPEM|^G)ZtGFi{iJngYHy{3yjaQ1)9@c+YS z`=?NHxWy@r3xr2OxuKFBD#Zu6())HX zBIAd?ha@T5!J;zxHUu{$)YnW7U|~UJm^{t5b_ckq!9EsZy~IKD#W;3og5!;d2V;@i@EX&Njv@Ra1d6;0(c1Yc}0GV2&?RXHfq>>7W^o&rGV#!^d1EyG4`jPMjf90Uk%})y;*-M*qBfHzJ4I!w%+vgB%W{+)rxq3+f!5d`OLy zC&CMtvSzd@Y>;`d_v4!r2)tD~1Sa9>Jour}#k^>1O95!mbpq5_V(u|e1mq@2{Xw*>9U>e9Ae~J0$ywD^T zn7iJFIn7sZ7*xMps_=PeVfV?}uL&*Uw>}u5s8D+fTL;4P_hIaysRmA+15t!cYay$? zwy8z6&77&>6Ok+zsI5_xp;id$E_;k2&H);O^_gG^u*sLFQLSHC6K%W(%mdpK$|?Z3m&J1g4G67Ss; ztK5|SUQ@LC=_AG`>J!Mv!C9Y_uQD_o3p7=$-tJ13aG;i2y}p`!MSrCBCW6wqvGzXJ z!1o*(ve%q8Jm&iq!dxU*w4gXn$4tine7$Ixfa7!_*Mz;@)b;6+sFFJH!9ZCPlnaIw zygahO^NG>D?gnx+Tp9LZmY`p0a>TYu&DnWS3wlIxTS*dw9|a(?nBa5R&tFg_bIEd( zBd;LH^pb=D3c$=4PXV;vLa8>DMugECCiGB%sKo0OAg32a0iGI=At+o547Gb|s2L{# z?F@&k_sEeL2b*J-i-*Y(h{;=T}_=5eWLgT>@ zMHz*)fX`e*Kc5Z?n7WQfZrut~rr&Eqzk_{#aHa2=gudrcNX5q}rsnW7Z?fBzhwT$D z=%!iJDYiRD#lG#O(baJ2bMEx$k;y?}dmFL}HoW;GA4>GR{hSoqWJUpG1rU#(fJty* zg5VR1{piJ->}FT*^Zq6R>*OiJlv?2yf4WHE+Y0n(WXAk_v}p+d+yP-;%AHkZm@)=u za|aI1f8SCL5PJ*%DH%nF|Xzbl65;eQ(FGP<3D{`0=A}HTj=|=p562q z*z!B5>391yKqnZINGpNPJjvNQepTO8F+BKgEsE3D19Q;UHXkn|x8A&yia9r24Qa1011H9B%N34x2J=o6=_ADt?p3K+PU0b@vM!r;f3-k*i zR3J7Vy_xGPyV=p$U_@G0Ds`a%!ph@a^V5QtQlJ$d>|nI#rj-ze^F-LM_Ga!7*?E`- ztnv_1?Ts{jcPR;6ERJZ2hC2Xmh&!;^U8YYIpyeI~V1a&ucpuT(Am)@1(Xau{Q?uiO zsw5g2DDD)r6}iDKM1IN#Ca|IAZ6T%%0s9Qe)C4+_gUDcl%R8pkr5cn@ESXl4V3o`R z>TqvnC8~RV?D@FHSN=fo4Z0RmP9MWL;Sv{8!g%^k_Tlp9u#*%Z&z4YXK zSkh~VFgKPrmSt0ptLb9b9dyd*>)`wD<24S`lh5VHes$!jek&Q- zb_HVI`8D<%dFe14)`blUsYR> zbn2Re+>C+#WlM?KB_|zPAoSw!m}?0Pw}$oHfZR>~JkoVSdr^>8_^%?nJhxwM$|{D;@)-Y53-d zvv`L}eybXG=dy8>Qky9!i!O zd#l;uwEeI@v*`^<(C28F2ve1f==vy!q=p{ ziPmU25}PX?iQ>)8YZ{- zOThQZcia*-U^ypzziPpA*hk<4rLd=2D_ zxLeL>6#1I1qmS@WfY_kcm^!(Ry`NyzWviNW*8g+L=Jp(RxpnbjHrJ)g7Q=3}Bn}?X z&|ejgH8sa*KY)i8)Wkw$46w4Wm$J9Ods1PIk?((W!aamzF22(J*Vx|iQv^16{&iyHkJjd$A5b0POY2BOU>b*t zm2HrmPZZ|r-!g*X200_e(lT83#Iq*;+7z*NM}YyN_50WC_dBl^!Hd-(ZX-623s-G- zO&MRIBOAQuRTx_xC5(03{f4T3HVv#)$U|dk*UdRj2X32?W=e6(*!#k+L)+$--nnxr zgsfd^q@x~~x)8FwGlu5hSNE-|C9;t33A%t2$YLU9NCR9|DQT6KRI;R3tzojXhc=`( z>-yMfU>&)VeIlEo$;RB2^Oue?P1E>&)V>nfE)PaDFBhTl%8u zm0S79U2y&I!+XEqJKabb!xI%9T&bO%cz3)>+vSA8^&5EOR8n8n^2{QK@dPF;Kh2yY zF8h4Vp?8y@2g^d_{_Ro!zG0SKI8}BIWQa%bZ+8?sI=xS?>5IN)=r=J`Z=N?h@jCwY zBo=(qZpAj}cr;5&!vm?C-n2cZzmGP`4Oa+}pPT0Ry6?%5a$n5-#QnFuhprkQy3%%6 OH|>W1-kAeaCjJ+!T_ooK literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-statistical-learning/images/image-7-Z2 --_ Z3 .jpg b/apps/api/app/data/demo_documents/stem-statistical-learning/images/image-7-Z2 --_ Z3 .jpg new file mode 100644 index 0000000000000000000000000000000000000000..c9f57d8f8f1fd2196ad6d339f739b790046f3258 GIT binary patch literal 7701 zcmd6MXH-+|y6u7h0Rck?DG5!AG(kW?AfN*BAp#=OK?S7=D1!7R(jpKM0RidKM2JWU zJ@klvNC%}Sbg7|+5Xg;tf8W{X-n+**=iFcS&3M)tBU$Smna{i4=b3XZ$`8sc!1jlp zfgV6Z0{}GC4?vj!t^$k<3}6O&Mlcx6#Kd@tg^QJinVIDb2PcF}h#v+O;ujPY7MGV4 z7L^qf6qLFsEqh*3Sy>q-sit{JNkjgEveNHKXqcFoSeRM(SXud$LOvT3a~Q* zzkqcRjVM6NP6J}6p>zT;>i?vt`NshM8Z@*ZI(h~$Bhx8n>IF4y04)s&L`w&vr>CQ% zUL8XH9iU^U=MYiUVc;~q2Nr$8r4*5r$0&BCvW@%Jz?S%h``*u)PVw;ao#8(#At@z& zPFY3u;w3fptJie(^sgKIVRqa6j)kR_wWHI6ht4jpZclxDpZWO*1V+AyijIkmL%dFY zlaiYDHa#Q1;C*3HaY^Zi&sEhmwdlI~hW3umu5Rp?p09&L!y}_(KYotm@N@GEi%ZKZ zzX;nqyLdapRlOMe**hIaIsTy(bCa@=)k{m(a`!)1H?{8FQUl6p<@cZ z_k>eaDT0yfN>W~B83z$o>P^zi^EMEFc={ z;(^!!IB-Cc$`3#FJ7nU=TBQALt~BsCJW4hlW@@b(GV4p$diNMosrE5%jsl!<6I`YM ztiC+4$793tgTp81I!OKC_ontwC;(Gf=gY9{CEVv8SeTea>?g8XuR1KA-|iz}f|#r~ z8muXnm2bFeHlMsUhZ^I{h3Mv&S-SC97v|fo96YwO!AzwxflA4`R=KnYsecpaWHQTw zF8kFrWJ?LhKbTKx|A4$@OdEbSziEGGj^Typd{zG=Ut_k(VmjBc5y_D-|j8eq7sw%jLfbhxl=Z--<8AS zW`Viel)9vgnL8G^6SKA!#<_h`pL=={p?ml1nd|5Dbc9&t04`HqYFytaiQi~5*E8e{ zrw2?&eKT1>26e-Gm*1ePBm6F-mw^pqZLsx@e|fMe!Hx%iV$cM~F=@H1-+yH1AbEeC zd@cljwd0m+Y6l+s^101_jUqYqpLK>kG1KWWzCz%IUb(&4C@Rz3#lJy^SjYQUZ!!c)0 z&cciIO>{I9o;a{+@JtI=Bkv4Mj{9fcgg!_Zc6Pn5wmmUoES822K#ArxlH~Q+zEIThn2r!M_OR-QRpxFLr3sp zO5c>uhB0wJmxTiG`;QN`tA5fG4qCS+G&Rbpa~}0f6wkI4)i)}{1g0T2hAfCLedON> zjgnNVQfg~Sv*dN+iMV-RObb^7hrt|-Q#!3PBX%0J`}xK3k5{lpto?EG(`98EYpikA zkQ~Qg`yjK8&rzigpDrxCOxaYM{^6=UbWlUrb>GhMCb~K1Q#y^zM)8N7Cy2rPaE%JD z)Bb6=0)rqNZ?eVK_QK+TykNSBkU<3L1XcjeT8a)fD)L8Pi!mQ8-VsTTzS_}wJP>mq zPMSmfK*^*9PnNO_RxG$^B$IyhmXB<>g?W{n71&ftu;u$|{s2`kokpd*~O*DX=?x z)J%=-%d!*T92iYkx__x{U!|A0)UD(q?N#9N=_Q8!7wl}-r{%~dZBQS zB=)JI=pRK% zz4oq$hDRUJmkqi1p727}S-rYRJKhJ@L*u0spy+YD!tzWZ5U9&#sx7z~uW>tA&e2s| z{Shf;VWAaG-RhsV73q@SK>=<-^)ZAKn&TVJCMZZSfORL7K^!h~h zG;gA~Sd^^ZB59W9kTjO#kG$PXj6@mDZbH8IuiRPU|6X`zodRf>97tj)07_V3nwijf zqS%08UtVOE(N_$q@znUT_&9)RJ02OD@?%6(#_;oi&$p1v#u0aG(Ar7e%`tmgd*0!v z5R2mi->k|j#H{Y7vv`ZKMBh5ELq2CM1?oNFHdaj}%|xJtGCWStXFl{qd5Fly?7@v% zE#6yX&Op{M;~>NIOLmU>sna^Y22{z%j{(&6T-qi)eDY-A)+Pl=J>eICGhrzJhx@>4 zZ0Q${G!hzF{{hOd71I5BjDH^^)ku&p*s<*$zS5j;tgYF6&So>FcC1BSOM-s`#gNbW z?WOb2DPf0eR-zuwhpRn{g_x}ei-m6c2T)~#a6-30PmoI->~=V8G+enl!FV$H0Yjrwba7Iyk};~gU+bIbomT=`dU?#3tV?cN_Ce+ko)Ut^ovVX28doA z`}REj9@Rl0X*3!Pk)d5P5Kk2bZW+Onn1FMFJ(J9E5}%7dM$?=A3Z94ej}AlmMtB&9iCEh!T&ypJig*(v&p)j1 zLwr>6xY{1diCjuH9pMP*_YUpjl{ZGKQGoEVLK86c%=o>S0$eR>rU0R96yOkwB@lh28m=A>G&J0t4WuGz3=2Q*^5F69i1)k!r6=h!pGQCYa9oLAZbHUkSXo@~;!>z18END`|EC0Tdv z7*wT9m-9XCHqqmBqe`PYRFXiyd;aU^;|znm>$as|CQCUYwRuop8dpdRh=RlJ$gx4skrxQt$YeqgkCSb9x%`W{e zD8K15e!2Ut$H}GB_YF)_c()q^v|)(do1#2;PC@ z0^TM?uLdjPvn0F5H3CzUy+7-9&dC1wla5DBxz*2m9bVBNAIZw zc+Ene-pbO6E%3P%ZA)g!xEsl%HqTM1*ZnEqYiZtwoSngPi`N};Gi_6r6Shq98mvKT z=l`U6I3(_PU^A)|_Wfz&h{wj=rrN~CahJvCx-75t+EOBxbp+Y9IsR@Wcfj#vGxu2v zz%u!33>1veX!O^&Dunxj=f-|;TI@AHHQL%pQ7!y1h4pY}>+#bx&K2re1FI}e3k~_g zW9`MN38@&*(F~Pzh5VJzko33}jqGDa*v$@En*QJmTT^{88lpMl+F1@JDVttD8`5o; zz>_SX&5Sjdzli#5@YStJgAeG~;IkpeQK&t?_)z>(JxB4?9NVB*Mt;L4;B-`HNKl8Y z2w^O0LvBtm%ys2%pXcS2gZsxFO=?Xbt9`ic zgz5MBmJYGV7&wDbx}eRqM2wo9@R>f59{56Sg?>r(wVour|@R(xJ)_yPK!?5 z{maZM zV?O`=+g|*(UGJB}H%_{T7S1m0XXp$EfexYW13ptqSp6aVr=jxuH8rvKMB^TVDS*RF zv&~;Q_*=C6brxx;LATVTj&3k6p;P|o;rejK650u^S={y)=(J~Ue7%D^UNgQYq97kr zqNeCPD)FoMmwvsb%uq=#i=v@5T?$P%5z|0`h^VVUoLMECCaaZ zu=p%-+6!@>uP+=bGBg;785FoI==$7y{!s~nvxOlwV7Rwrp^pywxOHjy{8~SQ)bHop zD#wIP7*aMnH|bmnX_=iQ_lR+(7uf9^8LeRmi)mr6Dk5ixMMD`>YwB6i^oZv6W1E$> zik7~%7QUf(2F|#tb4fj|D8c=|8CtM7wRn@L0k1sKbu0Oyuh-w`MPIhKF(H~_Pqdphyo z3bOxV&foX^e>1O)f<1vJ8JD0^!+M`2WB6e9LbfI8T%zUCDPTCisW;dUzsY?b%Djb1 zei-N$_)}4@W5hJ{A#_d){~pRlOei6f(mMIpb|`?X&#|T|vf$mtC2eSuZUB}ntfd^Y zSWV0<0mF8tyVY)=(3d9o5^Opmd+TZkHH=*_!8Q-4=aD9i6GXrW2sG!$5}olkWAKJU zMX+kujP~PWxC!!WnkFGzU~6f-@zj4*yT7a6KhO1ETx6YojkDs<#;SUrlo+0TXdlq{ zIWx2%bYVj;$OLR=yM(0ne(kZa)m{pa@|5WV*@U3#G^CtxvE!NC0nAKJt_|7)?@ zDo^CZ*ei~dJH=QK6Q0UGm9`RzxtYKgIqax`Y!BB|LhMA&Vq#U{T3LiMOJeUW#=Vu$ zBcHF6>Gh`N2%AwigUxVneGO?2P}QPVuL7%$dPrMTlDGWPn3bx32)=QR1E|-y+xY8W zCd8cw-fY8@)@^+LV&7&XzLz}&3>t?w9UjztG*V8Oaf>HxYV&ES5^B&d9_1xovOc1U z0QnB%mucxFr@8#Z#meA|Tfb~FiA@G`FOlmW{s{xEcB_5| zFgD-i%CJ)TDJ_PjwbYz;aO3Hx%VFc9&7%KwHmF5HZIhcQqf1E0=qQ~?(y(pwW13I~ zkJzmKAgtSFN8vqe+A_%YT!g}bI&_^!!=2hW=1UYH+dI2&BTR6fbESr=S6MBI=nk~C zR&ILYgF|tU94E42bpx^|v3j@2ziUthfOA2&vxziXvEay!Dk?93!jwj7!FuaHKtc zoSd`qwV4mP1hbvvz8FD%JbpCk(lCPF4~zS#Wz(oS`#UK~q|&*Sq+vE=S6kjrB&)rh zjypaJpd;`gZuGx`&*2s3;+l2Nw2~5cX{HCPpu7ZD6rxGRvE1r;rJa6Vo|;rE?#(fy zXw5Q!x=8%Ey~^qD*Y>zxr^0@5;mx)jzN79${{&>8WdG!`*3{8etM197oLr4B5a)nr z9Pn9LDbRo@;*!Yq`&6-$XL|dtlpr9zYpQ*w;pDwf2-IvKdcIaix$+tTf+Q1eZ21D zF$iDu7Cfp*u?3FzC@B^lyfDH=0fKiafF!`}li<}kP5}~gj0aQ><*PN-DoY#;-A|8& z9WFpe&S+5?dwI^Sk5eF>A+kz=&5E@Xt&9sV;O98?_CT_}3}p7ny|41+s-y3_08#T9 z>a0u3G;Y*#2~j3I&tVBR-NMGr9^ML~y=A&})3Q=WmvMng0#0nqx-#ky&5z!=eT$$rvs$I9P&*>}Q?&IMhrOllsH<%9`GrpUS%p0Ii^!^R`z!S?e<38WKBDT%G)r zcfQ?0{d8}rhFm~#_#0}0g&;Hpo2(T1%MlddiFm6ov@I629O6E>`dtOL4@W{d>u$21 zSArUP>mrD9REm!7(j-X~;*k9L$vjB^ z`;FsjY&U2X$?752;hGW%H}x-w<+h~obFK62t2c8`Gh?B$ExZJO{q_YX0{yRrg%bh3 z#xK}PtPTYj{azID-UWV|tTL~;#+Lgh+9FMf-gREnE)M>xoEgLD?`+cW`qx26gKlqU zjDqs0X=CM{c!d%5PDaT{s-nByPXV^OG#Y4gAycPQDJT~Pt z_)+E4Lkr0{Atf0IqiZ(gc~FkBT_>+ab39fM9J=0Z3)t@R%`Y}XmyCl#7=l>Ybc(-x zQnEgLLL(>_Ssex91J}yL5x9$*QNbEoe6QO#mMrVzM6x%05-+`prY}5e8ULi!HpdOv zy+-e7E^8UiFgKlL?lYnHX|P;&+Du00)HNs13UlqgIYf;%)Q`|`ON-ZGyjW@4*o)46 zDzo=aD+OpV3S?1ayqhTO&3*;hdcSD~dpg*F0Bg_Yid-^O!_;~;zPL_|hm}&fX2hx8 zhX1V}%6!*igH`?XFEH{vTQ#=v1{G5537LzuN7}md1h2ss?7&ZHZp)bZ<6*BA?%>Y!a4+HhIuxUKe+Sr-ogy)k5aE-=W$tn<`doz9~Zs=xDyh)QzdQbDcTonV6*e?i$ws zfkxE61AT($+{YwNL(uj{DG5{GB7WNaK^w;YJ9V>ZT4#xwcqDhR>Yc9*x|w^^Y|*cf z9?Ay0Tppa28o*bYty({gy=2tQ$!w z>iN~yhMMI>j81si=|j}DimO;CqXQ?_t7K(N(wM*+)8`ocjxDV2q$KP{cD?YxR&nR; z=4Jeq&(x!`;t~Lc22NIdRO;1#&X4V?4FTaFcU z)1EI?S3w+fPjdW0#<-%FCFh4V?o|}^I@&)-?!l6SCL)gNsLRetbbjTI`A^#OHw%-A zCnn)Cd;N6Ln1P9hjRk)y$O^zhVl4ZrXbvOB8`Y&iIa|(%a;gR17;f{q`A@0Aa!xI=g+06I7>pn>$R%3?UUDji*0Q|P?XfQ{+0E5z` zDb;Wwa9}&C5+seI_Pp3{C^BR{x_JK5>=*qh`;)r;#k?*3=M~kfJxxY6w%AcC(md#L zV27NW_+v;z!SmR;?ycR;g5wmJd}yvkA=Ql4k7+sMnsKM7;d^<2AIkk|;bQJY6V(R0 zF_eGgD^|%pG4T5>&Ho)#9-f2iM&b<@M#Nq({W0nYlPD8%++Ej%0)kX{3#AYD*EQEI3nRXPaL zQ9!B$gS60lLJc8=8{hAifByNrZRYMX&p9*u>^WyWd#}Crv)0k*(GMBw0%2l0#>~QXg5~&emQ$Q3*|>!Gp~6D^f`TG2X$g_j=fwmC zB~{Lzmz7ggRD?=g)Krz%kXBHX|NRjV6B855aTdN4C-~$=1x4lm$L;6~z;O&{1f2$h zP6KotATS5$s11P9>|_A_-GF~>AUZHT10&=Z6Z3J}fJ%0N4g?0%(SsQn=;>*r18MI8 zdJYCoQMpTuCr#`jr`@>ZL*g@!iCr#h=Ds;VfGOO4@|200hnMdZzxWvm$+PDam6TOf zFRJNY(YvZ|V0g{+)@?I$3rj2Hz55RyIygGHdw6kRa910*VNXvw6?W(bas9HHaIl=V+1?;6E}^YnVp+oSX^2rZf81?z>M0n0a7|)8fS6)c$7n-y`<)f5q(IiTwvJ9KZqw(H0NP0cZnz zWXY^xX31=rBm9RErDEWbwqG?JoJ ziIDBF=hRuXl)|7#4m5L0RE~h>6&_V4_$H9er2m4)L)I7r(Vh?CF8Sl=0C&gB%a^f&m5e31oujfF8T@2TN=pk zU(=p;@+pU(bdvi-vis|Gd$v`VmeE@-%24^UY(@fa137hjp}eiKFly z0ZdBIyveA;f^bkCH$n^kjhLCC#2s4#KQ2sq4qSOP37& zh$Ccf9|8R}N{jrE=N>Z0$9(yv-?}`=wtvejeFSXC?Y|=`6@TEqoDYD4Ax>6*FDNRj zD>^naK8(4v>vYWt_i5+6G3}}VsM`^6Y`x~Klz2_sN%SXJplo3HF#k5cBsh!^9$V~E z=@=k~k5U}cFr-`OC^-CqLkEdXHVIOnvCh}p3HeA*wv!DwntoRL>&`FjbzN(akJT4T z%??TL%jf28o#g4p2l5yw@lj&DjbE+eVru!ML}pxtW1a@p3gv@KvLYpc5^Yj-fpr|K z70LbjBPT7!PGppQDnVA;ixIhpmdW@0I?_f=FS2(n-UA-`h0z!6sy6Q`^^z7kks zU>r)LnmCrB;;E&Jp$&117;>NkU15OEuxW>KBcGOVEMm#L=iTZa{*p6vFMFR+D;18H zYA8lD3$sSGqZ27p{H3%L(X2I58mSQ10sSK ztKANVNTG30+OWCi?*p7nou}$;oa7GHpf{@;E^F$pby797&uz%;7@d&rnwXr>b$%pw z@Y7myA>-6CJO@Bzy%UHKv*eOx0SUYg?aSAQ^aQv}^!_Q_pxmDIF;dM5mP=EqaPvfu z6+PSf*{O7wWDPRDPUeJz8I_+fk&Z!t=5UwGlE4CEq<72L;WO3`qr!L7J7nmWoT<4# zq-B2jm0F8ZPNA)~BY|^Lg>OEtovC|U)y}LKI}yy1l2qV?d*Vd49^tGt$OGN`viw#c znRp5s13XD#i>Bz~rMMa^2=`uE8TQGw2%n+E6r?4{tBxZF7sM;$P&yO;uy6Yg_m=#X>$?@>el< zMl-X9bRX@0s%=-k)%!Tx^nF)_eVz*g)Gm-MKw(T1PlhkLw0R zVaARRLDDzU=MX5mw$xB40GnHQxy`pN)O&-V61uLKRD-SD*PD#+gr>|t&f;F<^aHIe z-JA^2tolrJBg$eueU#YKtPxt@m*~i+zn*m7=(B|@T{r?lWt61Mt80jP5%iB`Cyd6f z7?868Y6Zw?(yFeBI#Dwe|M+fs4^LV2UCRN($LI;spbw}dqR{tGSrm2JsaLp$-Sr!0 z5*$KU_Xw-p{lXWGoC6kOGKAk_k(my4E};QRJgLr7MTsd2iddH890@%=I^f~$5|v>s zagvx_Bd=O+c^nccnlNO{-W4=(?hyNOt<$dVGsLtH0FQiph$=je3XksCY%A{aD-aR9 zwZmOer1??4Aqb%+#r54QK_Pm$0<_R1Fv~1cq+9RM7 zHJA^68`rWYf3<`b^x-eeT_Gr=V`8J!^>ms;BCu?sjkwU4|)+3;!5R(;rv5%xDjb20% zA%}K?)lsvPVwyzSpEI;$6$#9x39`74Aqw0Dp4k7v|hyq z2;o~dTl1V4CzdZ!*n5;eLG45LvDrHrbhVRwq`JOlvI=g&EN%J0X3+()xpi?(F1C48 zRoL<7hARB?Ib3?YB?-?Is{4gjSfv;Df~`DZz6i5&1Q6bnuE|Mf z#J|`PQm@<=8#*k72-om!KgP7ANe>1fv&E^&s*iybM%4sH>PiWcEIbUVr-dodZpD7L=xrL)hm|~atLCp~i9e2f7OWyHuYVT|I8(F@8xZ($1fHQsI_Z^2*&yeot`K++ zX{|PVx#hEbW<|x*j_(3S7D)m=?D~QZvfQ)@Ar#~TzeU3z0G4PYa|xdze8*s{?}+~r z2Xmyx{6G-yuuWS}wz6^b<6NEH4#Oj>D@8F}ha;Ybr-BXh3I-Wjd@L;?7TgtB;ziD}Rkg zcnh+(7Ay4ck!M+B|321555%i+K|{Mf4HRVWs7g?vlhfH#i5#h;=kmYabp!_Xv;oh0_7v2-fa@D(f&z$p?7)V>J@VAn}`(j$*LLW}JBR-ST3;L$* zj$89p@Wu8bEUxRu659M1YN_e|gWJ8zL;8bj6=TQD1P~5wZX&CYEv>^vx^XU=a{dk1 zMsbqMiQT-zrs?{T{)Wp9hq<7G+QXddG3j=FO5-Lv!VelXh<4B4=+3~88!3Y0)-<_J zWofO3Cz9U>N}09d_d;Zbxrp@a)t8eciQHNT!Q5jbp1UC!r$<&tfMIDpa`+bl_KbQr zSWbeJJ{9FLL6sPKaRlT(+0agy9pP%?^ElLtX?j*+W7?*2W%@5!#T#*6UQkJjRB>qY z1aDO!!N)~E;xn==0=`&T{-P}7%8d^g(5*Ynz?SG#2^VR?XP!8>tz2jIawW^L_;T&q z{MWRg;Pksv{h5P`D}gXa6N|Cj^tm^UJ!Nq+_RGrK&4m>=AmT1=DxuPMOPdns`^D{r zQdeV=Dv)2o%rs5QYj@37;Rq9JM60*+g~}d7Ma}|4G0$)1X~#r%xVy)G8BcY;`6x5z zTftTafGa$2b!8+f$Gcs@2)$RY%K7mwD&_D z>&AkCB|#(!Y1`Uln&8l1ab_q5b+F$)U*-Y9J+7Jp)b#nNeAv>B;n@^i2CG!N9qXf= z`ni#cr~*Yh4!s_~=lLC5OQ*M8{rnE~xnF8LoPjgN^F~n=anG}ehXq+|vdABJjX3u& z+uTd9Z$M7i6GOjL&p^k@#tc~b)$=D#=k?DbcM69Bj0snmYANvCcxqd7->d>68F%JY z!TV)?o5Y_T3T}@R(6UaiZCMBvY>m(p4#eucgboq3UIyz;baaa5SY^}m?Oxil`?eJ2 ze1iEN0gj^qm`TP$?dKZzXFB1ezcx+}dRVWkX}uGbk;zDc!|xc7g;7JuNL8c-Uhj%Y z@CRAQG(3aZ&o@9O1!uXhxvpc6T{fC3 zJw2wWXQT)CCcg+Uk!TG(L+%;R8P9(TU+whO?RF6{tvbc{r27!NiyM#F5XCC)2WYd_ z?3w81bw92a_9*7Q%}2*1k!RIKM27{47SJ*^s`$|P2|h7-BaiNKy&Kw%H7>163u&wb zk2K|~vUJnfWj6Xz9<{5WT`MyO@>*oo{TY!BV{PA$!z6(vkiaxk@1QKPH zbdzOoG~KssrLQKhq@}*s+BfYyF{!ek>(}Sc6-^JT$@D^vZ%VE{J?4BT_P9@iB=Saq zpcD5k)?iTot=FIhB88I_$bVs?bC8!0nU|FZ9Fxk3DuO@n3=&;#TsYXxZDR-V&mrL!uHA@*{e-fI;4BbTcD6O9`8jb@5F zMQN!_Pf!kKayr`~^ETHVNg+TuaX}kmk6L3hCE9eI*|2R&deS?5<>PJMMcGj@Hk+nG zxLZ+;+)*8Aa+eAWT@_=_h|qI%4MZ%U-^@k}o?Bb`)emvk^M%K(X{cuwRfT4df(EzR zQo_HYKnx8z8&U~ACL36&)e7bK6sbXAdb@mUVt!qB-sXB6Wn)!ca*+Y>*ruh)zSy=y z;rR}A7oo+_$0b=OHuIO zizb|kw>`^+DP#$wzY}WfO^BF;mt!eXVxQ3V%L%QFGhHFEvlJ?l)DOY@`Skd68Oe+{ z<^jru&L_;{U|^~Ge3OZQ+1dkASbHU$Wx39erSix7e$cWd^ATVLbdntD#wZE|yo4@N zfO=j9I*h$-4R2B38P?)iyb0dYV+sjbW|p&?W$+pX%w=X=_*$nmm9?c$dKcPMC)(e8 zUoj~B9_`_+R93X?j9J90EZz6?Bjn!1jflY#RPF%uI~%x$?&hJhBN9ZHs7IA~*maPB zGxkcuy|SosQ4XY5HriB2NF_AjT!4y;FJ6Yjx~QYcCl|4jJ7=$>FqfAVtnS%4SZ8c* zH&lU=shs$>!V%r2e0lv^NP=rdT4JZugm@!Ss}MOouF^^#R^~l45o5c0>G&amxsR`!tz?~VW+MGz*}iQNfjQ=I8s-&GrNKhCuo*D7|p$P5zFG^^XM zy$WTc21^A6GjGM6Kw~Jb#EP>maMVKuHrHUZscoFC%hanw3=z?Z`>@6GWXvM-2yjvq z?VN`!=m^LEn*#nHJ|S;OOmWW}Jn?Xbzr31Z;kSp}C(g18=^x~Q2~OMJrnO)`PGs`L zP+J|O8!yBiJu=0%Lp%Zm^|4~AP?L%31~L3+?G811wVcqnH_mCCK|U}l%zjzhq!)Pn z+kwpBFDkmnAL`c*70EcS@yttvnJ10*c$Ic`hg%E7tw$=kKrz#)XZCaliYmGbSNv<% z><#Yv6h%ug%Wd5slgRtate5Ksqj`j9mxj`Gk*;NJXR_mQ!;1HhA4fJtv1{`$1aU1U z^XosoTiXO%R;wr}ut|{h?tTiX?yior#%kyHG;udLrDzuJr*RJ4FV#i;+cpO&%(4l3 z_Wndd-aw~z__aCnFIuwwVUsMR&Dk*-a=oz`Lk-~Po2o-*2DU#)^++_4D)VZ>EH#7_ zyq;8xP~Q}U#Ssx^H>4>hS&p)$U)4M@e~lgiRl&%u4Aza?BObl1UrUq_EB@-9)Qj&W zwOTddrSlEA6>}n_{SFx&j(=xBz*KdoTHP^2Rjikm&}!3Lo#(t1o1WDPnqy3`>(Uhc z0(4)X@m|)!`9QfN1 zP)a0#NKu+dFVcIH5}Jt+l8N_yXJ*a2-VZag&)Vn9*=w)o_kZ?z&i~NI=(E7d-;7O- z0VXB@U}9VVeFC@wu(Pp&*;v`ZU@!*>?)r5(IgM*ukTbPGOSmm_XX_fzT&_4pa?0_Ax z1Y(i`n0c8%yiD{q0LEA+E7RWr{4>3yx8Xg%P`!+r|zp%Kpyt2BszO%cx zfAE7$`FZ$@iwOYz1B-G02iU)G@iMrWSy(_U;9p!!%>Il7@v^X|z*Q|t0`3p{IR&NF=45t$q5VzvzXOZ>f06wc*ne}41Kc1c#^!-|0X=|x zD3=$`Dfbt=4F>gdh9l8%^O*0BBPKAn!F&|~y-f`x^16wyzIda+au*#Ce7v)KZZ%7U z5NhM9p}~nb%soCkLyDwqxn_ku9d&Wr6wDAQrh+WM`TZ_FelT2+Yp=%wkCGjTcGFv( z`DB>zZkU62h}lqx_PGhlUaP;B!;0t7@6B1mjh(0D_ZR!Xg-Dy~W8+?%w>Yvn5w|>o ztdn_)m4}qy%wfjK{hP$i&vRW1R8+j1C=9pkm3}5*eWD<>&o0X;#0Kp$x6#0pGmY9S z86PzaRDYZ8@6%+aI@Sj+K)U#-CE*SUJ^u5~qJ~%KfWn3XbypW}zUmb=$Feg(?cCAb z{z3;(*Pm4^5yKG@xV^jB`RO>WPrZD{;`{G_CwmF<2KZro`+D#$MGjSp3` zw63K!^nF6_5uYD?azTx025I6T2jt=8fP%Z$mS38!xl_ylXw5Pm;C(E*8Ejic3C`>2 zck)*LIiALlx4SXt()~}w(Ckn8(Cv$_UFTyebd%@ezxi!*J1Qq0*~PhSIW|1VHn*ky z$umVkwGcI;s{1|iZx-+#_^i3q?+2*vK(++Y zSyl}oji7+lZJcIaK%r5V3Hq3P2kCSSpA6|#Zsg-NAH z!Ze|FeLcul)?iDP{Gg6oo=M4C3SYryy3pL5@31fieo)@E+bi&q4xDK4fF*3e;YLHJ z&jQe*=s=u`^VfQ+{9NZ33**W1rLm(gCXZcaV##qiY+f|+ zVVd0a7o+hvK4rwl$5!_B;{waZP2(0yHotgvD0zK6wYJ9~3i^g5=vJAlQ^OncCCT^mA+)Z=RI88;L zm7Sf9M9BAqoI2kgq2s0$Ug5krsED2Nd{p<9KZTUKuBn%~B?lT@&BFngFPFIMgHp`B z_wGk4*VgGaF1$?6tckD1P3SeFdB<}?;n`%P$64wPC{2PEhqpjl_@ z6VYb(U#DCeW|7=C&nP*#*f*fcYWYPUxScwlq6vQ?lP+tQ@bS)LO>Tri4%jxdzQFPO zA>MD!Ac`!i2x#88E)tJAbvp=;%fhl^jnyf-;>K5E@6*ft3bV@>u?+ zIvv=XRyx|SV5zot^&-%LtevnK_oV6xI?x-?wEb3}uOImYKSK-tLHwg>j*z$3sRL^tZ+BJx{##Lna!tjMK(qURJv}v7@sQ~Gb3)8c$rx@vvK3tb9G(W zS9-7X1eF8OsA!hdaLqI6nF?NN=1{EMx=m~`!V?tPm+|=n{Kbk*s%z5qfxDRr3Dp>6 zC;ajWfy|e96U5<fi-&2_tHA z0;}aI_T_nrfemXmKp-Z1+PCxL?7e})Qi4`i_C1{k&aRFIYS|Ap@7UZuWDYE|2JEs! zTn6#FLfAIUC6SESEz3i;0}K7fx^1_z$*4Y*wo8?>Nh|SneDR(4)#*}@2OTB{IUM^F zt2H+i2Q9G2H-^rtgb}-hLt}7@A5OYl+-`ZzY$u%-J-^Z~eH+F&XhDwcn4Ulq9ca5C zDKlf1?J3@IS)aNNi?zEJNSbL)$z|cu75<7qUd|rA;rC9eThw^{l(4|9V62NTDdyh5 z#>Fr;XFkP!(NQBXW%iE@6POcWnW6H4H@>Lk=ISB(b8eWH0m>(OF}WA8A(n>~*{|2) zN0ILc4Fq?Nn0c9G4)GJhy!sON=Z*JPMU-1Kwn8N9hfD;XlKg3XhNhP zseX#pQ(V`p-UG$y;{%swz{OqRp3Jd{eiPcgVDE}YJ}G}H;62F?p&L32MQ&CBqBKJU zh}O)|W0m>!{8Hn4(nzj{8-CKI*B`BLPAXvN>>WuT#qaHiDkMcw`@1=p*El8@KY7#q z9tx_o+;Cky?E&UFW+9Q1b0P+J4qs9{u_j_OU5Gh4zv<>xptn)nN%aKvG-&3pg-TP( z7rr-MyHWKa@)TO~nbXI2rNlOyx9`PZ&qE(igtb8Clnw=|&8b3+x2SDYn-tlytjFG4 z?t<@b#QbnJZC)4X?b7}86l~ir$AS$)m89{}fm}yJI?x>AO1%LiVJX*W(P(};z`7H* zbFt9~xJ3v4XVZQ) zdx%vZeh1vzXgNN_@6Q=utUMZWTOC{nL>!bSaiB<){FWC!RUE>^DAi(`keC*Yo1H2L zx?7hd8t@sy;T2~I_b;6yqn?+5qe za60|9mZpmvkSjF?E61fx3~^+!j5)CGqXQ&o2tyu+AfIHsXP=sg%WE$We)Z+V5!R=s z2dDLy!!D+5{}6;7i7`+q`|DfF)I#Fnm5#Vh!m%k9^3B zdsDPx-z^@|*fRg_Vt12}bHLNg2C#ZQ-of;xA=+|U#visj{5&=FxG(7O=OB!WM25}# z{)KFZTEUC~?J7qO0cx58=C7r>%J<9R7tbYD`PFJk>@xZ`yF*VybZ7ehj>FkqGWRky{ zE^^yYc=P;1gqc>fMcuJn4p+@`C?fKEi^SO~N^DWvz=X7W?@fY1_eXO+$m2`Y*xyI- z){mNbh8t?hyaknA(b5T@ZJs|nY4X4bQ2)9$P3z+7Hcdwb_L)NC9-g||A%(Ay`BoAl zJ|yj#OB@Q}(H4>m?^H~0+dFv1_>hE>@m^Y=Usrhbdy6URu!XC9*H?n0#vu|OS(xC%^iF#(=$|PhcSl{BA&bcIJ$)Sq` z?K2HX;`FB=!Mm$h_gAv5cC+G}LwTx5nCB!T_>`*?;bm~Jl8Te*Wa?VyNAi}bG$=Qc zy}W!O5^6vH^svD6J!t-BhnNfNzt!TALJk!95DGafCi%10!a*%(#( zZF|XxEUw`^p3-_H~H9um>mx8xn@ae z4Df4&PL2g=iV6{vEt8&LX@4ou0Z0k;7*e!GL}w-AC)R$Plom{1<&#Uzi=YLObGWRL zMgWwNrkDF3T4pnjuagb( zK8|fJ!j(hi9&;7dN5k>(QIEaskG8jt^FvfJ9TVX8rY5X5?ruM~p-6qfH@eE+E@#4I zTS6gOyH}`oc5g2W&79Jtt}g^=f1XzA-0_c#$t-GeRx#4m%ArjuV$E~6`Lu_Srnh7< z+}<`tU&nk-FZ<)Pw1;Oy0^IgX dict[str, Any]: + """Return Official Library metadata with ready demo source links.""" + return { + "categories": [ + { + "category_id": category.category_id, + "label": category.label, + "description": category.description, + } + for category in _OFFICIAL_LIBRARY_CATEGORIES + ], + "sources": [ + _source_payload( + source=source, + source_payload_by_demo_source_id=source_payload_by_demo_source_id, + ) + for source in _OFFICIAL_LIBRARY_SOURCES + ], + } + + +def get_official_library_source_payload_by_demo_source_id() -> dict[str, dict[str, Any]]: + """Return ready Official Library metadata keyed by demo source id.""" + return { + source.demo_source_id: _source_payload( + source=source, + source_payload_by_demo_source_id={}, + ) + for source in _OFFICIAL_LIBRARY_SOURCES + if source.demo_source_id is not None and source.status == "ready" + } + + +def iter_official_library_sources() -> tuple[OfficialLibrarySourceDefinition, ...]: + return _OFFICIAL_LIBRARY_SOURCES + + +def iter_official_library_categories() -> tuple[OfficialLibraryCategoryDefinition, ...]: + return _OFFICIAL_LIBRARY_CATEGORIES + + +def _source_payload( + *, + source: OfficialLibrarySourceDefinition, + source_payload_by_demo_source_id: dict[str, dict[str, Any]], +) -> dict[str, Any]: + ready_source = ( + source_payload_by_demo_source_id.get(source.demo_source_id) + if source.demo_source_id + else None + ) + payload: dict[str, Any] = { + "library_source_id": source.library_source_id, + "category_id": source.category_id, + "title": source.title, + "source_url": source.source_url, + "mime_type": source.mime_type, + "status": "ready" if ready_source else source.status, + } + if source.demo_source_id is not None: + payload["demo_source_id"] = source.demo_source_id + if ready_source is not None: + payload["canonical_document_id"] = ready_source["canonical_document_id"] + payload["size_bytes"] = ready_source["size_bytes"] + payload["chunk_count"] = ready_source["chunk_count"] + payload["original_file"] = ready_source["original_file"] + return payload diff --git a/apps/api/app/services/demo/source_catalog.py b/apps/api/app/services/demo/source_catalog.py index cf2511f80..181638909 100644 --- a/apps/api/app/services/demo/source_catalog.py +++ b/apps/api/app/services/demo/source_catalog.py @@ -9,6 +9,10 @@ from pathlib import Path from typing import Any +from app.services.demo.official_library_catalog import ( + get_official_library_catalog, + get_official_library_source_payload_by_demo_source_id, +) from app.services.demo.source_projection import DemoSourceProjection @@ -205,6 +209,196 @@ class DemoSourceDefinition: ), original_file_name=None, ), + DemoSourceDefinition( + demo_source_id="demo-financial-nvda-q1-fy27-earnings-call", + canonical_document_id="demo-doc-financial-nvda-q1-fy27-earnings-call", + title="NVDA Q1 FY27 Earnings Call Transcript.pdf", + mime_type="application/pdf", + size_bytes=288_661, + asset_directory="financial-nvda-q1-fy27-earnings-call", + chunk_count=44, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-financial-nvda-q4-fy26-earnings-call", + canonical_document_id="demo-doc-financial-nvda-q4-fy26-earnings-call", + title="NVDA Q4 FY26 Earnings Call Transcript.pdf", + mime_type="application/pdf", + size_bytes=306_317, + asset_directory="financial-nvda-q4-fy26-earnings-call", + chunk_count=63, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-financial-nvda-q1-fy27-cfo-commentary", + canonical_document_id="demo-doc-financial-nvda-q1-fy27-cfo-commentary", + title="NVIDIA Q1 FY27 CFO Commentary.pdf", + mime_type="application/pdf", + size_bytes=75_796, + asset_directory="financial-nvda-q1-fy27-cfo-commentary", + chunk_count=19, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-financial-nvda-q1-fy27-results", + canonical_document_id="demo-doc-financial-nvda-q1-fy27-results", + title="NVIDIA Q1 FY27 Results.pdf", + mime_type="application/pdf", + size_bytes=379_572, + asset_directory="financial-nvda-q1-fy27-results", + chunk_count=120, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-financial-nvda-q1-fy27-presentation", + canonical_document_id="demo-doc-financial-nvda-q1-fy27-presentation", + title="NVIDIA Q1 FY27 Quarterly Presentation.pdf", + mime_type="application/pdf", + size_bytes=3_371_774, + asset_directory="financial-nvda-q1-fy27-presentation", + chunk_count=39, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-financial-nvda-fy26-annual-report", + canonical_document_id="demo-doc-financial-nvda-fy26-annual-report", + title="NVIDIA FY26 Annual Report.pdf", + mime_type="application/pdf", + size_bytes=15_850_437, + asset_directory="financial-nvda-fy26-annual-report", + chunk_count=551, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-financial-goog-10-k-2025", + canonical_document_id="demo-doc-financial-goog-10-k-2025", + title="GOOG 10-K 2025.pdf", + mime_type="application/pdf", + size_bytes=940_068, + asset_directory="financial-goog-10-k-2025", + chunk_count=300, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-financial-meta-q1-2026-results", + canonical_document_id="demo-doc-financial-meta-q1-2026-results", + title="Meta Q1 2026 Exhibit 99.1.pdf", + mime_type="application/pdf", + size_bytes=154_270, + asset_directory="financial-meta-q1-2026-results", + chunk_count=17, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-financial-meta-q1-2026-presentation", + canonical_document_id="demo-doc-financial-meta-q1-2026-presentation", + title="Meta Q1 2026 Earnings Presentation.pdf", + mime_type="application/pdf", + size_bytes=239_814, + asset_directory="financial-meta-q1-2026-presentation", + chunk_count=55, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-financial-meta-q1-2026-earnings-call", + canonical_document_id="demo-doc-financial-meta-q1-2026-earnings-call", + title="Meta Q1 2026 Earnings Call Transcript.pdf", + mime_type="application/pdf", + size_bytes=131_758, + asset_directory="financial-meta-q1-2026-earnings-call", + chunk_count=13, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-financial-microsoft-2025-annual-report", + canonical_document_id="demo-doc-financial-microsoft-2025-annual-report", + title="Microsoft 2025 Annual Report.docx", + mime_type=( + "application/vnd.openxmlformats-officedocument." + "wordprocessingml.document" + ), + size_bytes=1_044_155, + asset_directory="financial-microsoft-2025-annual-report", + chunk_count=122, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-financial-alibaba-fy2026-interim-report", + canonical_document_id="demo-doc-financial-alibaba-fy2026-interim-report", + title="Alibaba Fiscal Year 2026 Interim Report.pdf", + mime_type="application/pdf", + size_bytes=346_418, + asset_directory="financial-alibaba-fy2026-interim-report", + chunk_count=216, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-stem-jurafsky-transformers", + canonical_document_id="demo-doc-stem-jurafsky-transformers", + title="Speech and Language Processing, Chapter 8: Transformers.pdf", + mime_type="application/pdf", + size_bytes=1_363_617, + asset_directory="stem-jurafsky-transformers", + chunk_count=46, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-stem-deep-learning-transformer-network", + canonical_document_id="demo-doc-stem-deep-learning-transformer-network", + title="Deep Learning - Transformer Network.pdf", + mime_type="application/pdf", + size_bytes=2_619_505, + asset_directory="stem-deep-learning-transformer-network", + chunk_count=25, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-stem-probability-theory-lecture-notes", + canonical_document_id="demo-doc-stem-probability-theory-lecture-notes", + title="Probability Theory Lecture Notes.pdf", + mime_type="application/pdf", + size_bytes=708_671, + asset_directory="stem-probability-theory-lecture-notes", + chunk_count=109, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-stem-introduction-statistical-learning-theory", + canonical_document_id="demo-doc-stem-introduction-statistical-learning-theory", + title="Introduction to Statistical Learning Theory.pdf", + mime_type="application/pdf", + size_bytes=1_208_674, + asset_directory="stem-introduction-statistical-learning-theory", + chunk_count=62, + examples=(), + original_file_name=None, + ), + DemoSourceDefinition( + demo_source_id="demo-stem-statistical-learning", + canonical_document_id="demo-doc-stem-statistical-learning", + title="Statistical Learning Notes.pdf", + mime_type="application/pdf", + size_bytes=614_119, + asset_directory="stem-statistical-learning", + chunk_count=71, + examples=(), + original_file_name=None, + ), ) @@ -216,14 +410,32 @@ def list_sources(self) -> tuple[DemoSourceDefinition, ...]: return _DEMO_SOURCE_DEFINITIONS def get_catalog(self) -> dict[str, Any]: + source_payloads = [ + self._projection.source_catalog_payload( + source=source, + chunks=_load_source_chunks(source), + ) + for source in _DEMO_SOURCE_DEFINITIONS + ] + library_source_payload_by_demo_source_id = ( + get_official_library_source_payload_by_demo_source_id() + ) + for payload in source_payloads: + demo_source_id = str(payload["demo_source_id"]) + library_payload = library_source_payload_by_demo_source_id.get( + demo_source_id + ) + if library_payload is not None: + payload["official_library"] = library_payload + return { - "sources": [ - self._projection.source_catalog_payload( - source=source, - chunks=_load_source_chunks(source), - ) - for source in _DEMO_SOURCE_DEFINITIONS - ], + "sources": source_payloads, + "official_library": get_official_library_catalog( + { + str(payload["demo_source_id"]): payload + for payload in source_payloads + } + ), } def list_chunks( diff --git a/apps/api/scripts/validate_demo_documents.py b/apps/api/scripts/validate_demo_documents.py index fea9b6f2b..e68a39e83 100644 --- a/apps/api/scripts/validate_demo_documents.py +++ b/apps/api/scripts/validate_demo_documents.py @@ -24,6 +24,10 @@ def _bootstrap_python_path() -> None: _bootstrap_python_path() +from app.services.demo.official_library_catalog import ( + iter_official_library_categories, + iter_official_library_sources, +) from app.services.demo.source_catalog import DemoSourceCatalog, DemoSourceDefinition from shared.services.storage.zip_result_schema import ZipResultSchemaBuilder @@ -39,6 +43,7 @@ def validate_demo_documents(*, write: bool = False) -> list[DemoValidationIssue] issues: list[DemoValidationIssue] = [] for source in catalog.list_sources(): issues.extend(_validate_source(catalog=catalog, source=source, write=write)) + issues.extend(_validate_official_library_catalog(catalog=catalog)) return issues @@ -120,6 +125,81 @@ def _validate_catalog_projection( ) +def _validate_official_library_catalog( + *, + catalog: DemoSourceCatalog, +) -> list[DemoValidationIssue]: + issues: list[DemoValidationIssue] = [] + category_ids = [category.category_id for category in iter_official_library_categories()] + source_ids = [source.library_source_id for source in iter_official_library_sources()] + demo_source_ids = { + source.demo_source_id for source in catalog.list_sources() + } + + _append_duplicate_issues( + source_id="official-library", + field_name="category_id", + values=category_ids, + issues=issues, + ) + _append_duplicate_issues( + source_id="official-library", + field_name="library_source_id", + values=source_ids, + issues=issues, + ) + + known_category_ids = set(category_ids) + for source in iter_official_library_sources(): + if source.category_id not in known_category_ids: + issues.append( + DemoValidationIssue( + source_id=source.library_source_id, + message=f"unknown Official Library category: {source.category_id}", + ) + ) + if source.status == "ready" and not source.demo_source_id: + issues.append( + DemoValidationIssue( + source_id=source.library_source_id, + message="ready Official Library source is missing demo_source_id", + ) + ) + if source.demo_source_id and source.demo_source_id not in demo_source_ids: + issues.append( + DemoValidationIssue( + source_id=source.library_source_id, + message=f"unknown ready demo source: {source.demo_source_id}", + ) + ) + return issues + + +def _append_duplicate_issues( + *, + source_id: str, + field_name: str, + values: list[str], + issues: list[DemoValidationIssue], +) -> None: + seen: set[str] = set() + duplicates: set[str] = set() + for value in values: + if value in seen: + duplicates.add(value) + seen.add(value) + if duplicates: + issues.append( + DemoValidationIssue( + source_id=source_id, + message=( + f"duplicate {field_name} values: " + f"{', '.join(sorted(duplicates))}" + ), + ) + ) + + def _validate_doc_nav( *, source: DemoSourceDefinition, diff --git a/apps/api/tests/contract/test_demo_documents_contract.py b/apps/api/tests/contract/test_demo_documents_contract.py index e07704a4f..c6b6267ea 100644 --- a/apps/api/tests/contract/test_demo_documents_contract.py +++ b/apps/api/tests/contract/test_demo_documents_contract.py @@ -82,6 +82,11 @@ async def test_should_return_demo_catalog_with_resolvable_canonical_citations( sources_by_id = {str(source["demo_source_id"]): source for source in sources} source = sources_by_id[DEMO_SOURCE_ID] spacex_source = sources_by_id[SPACEX_DEMO_SOURCE_ID] + official_library = cast(dict[str, Any], catalog["official_library"]) + library_sources = cast(list[dict[str, Any]], official_library["sources"]) + library_sources_by_id = { + str(source["library_source_id"]): source for source in library_sources + } examples = cast(list[dict[str, Any]], source["examples"]) citations = cast(list[dict[str, Any]], examples[0]["citations"]) citation = citations[0] @@ -96,11 +101,37 @@ async def test_should_return_demo_catalog_with_resolvable_canonical_citations( assert source["original_file"]["can_download"] is False assert citation["canonical_document_id"] == "demo-doc-tsla-q4-2025" assert citation["canonical_chunk_id"].startswith(f"{DEMO_SOURCE_ID}:") + assert spacex_source["official_library"]["library_source_id"] == ( + "financial-spacex-s1" + ) assert spacex_source["canonical_document_id"] == "demo-doc-spacex-s1" assert spacex_source["chunk_count"] == 922 assert spacex_citations[0]["canonical_chunk_id"].startswith( f"{SPACEX_DEMO_SOURCE_ID}:" ) + assert [ + category["category_id"] + for category in cast(list[dict[str, Any]], official_library["categories"]) + ] == ["financial-reports", "research-papers", "stem-books"] + assert library_sources_by_id["financial-spacex-s1"]["status"] == "ready" + assert library_sources_by_id["financial-spacex-s1"]["demo_source_id"] == ( + SPACEX_DEMO_SOURCE_ID + ) + assert library_sources_by_id["financial-spacex-s1"]["chunk_count"] == 922 + assert library_sources_by_id["stem-statistical-learning"]["status"] == "ready" + assert library_sources_by_id["stem-statistical-learning"]["demo_source_id"] == ( + "demo-stem-statistical-learning" + ) + assert library_sources_by_id["stem-statistical-learning"]["chunk_count"] == 71 + assert library_sources_by_id["financial-nvda-q1-fy27-earnings-call"][ + "demo_source_id" + ] == "demo-financial-nvda-q1-fy27-earnings-call" + assert library_sources_by_id["financial-micron-report-530bd7ed"][ + "status" + ] == "planned" + assert "demo_source_id" not in library_sources_by_id[ + "financial-micron-report-530bd7ed" + ] async with api_client_factory() as api_client: chunks_response = await api_client.get( From acf60a4d701d195c170813bedb3564369dd97a29 Mon Sep 17 00:00:00 2001 From: suguanYang Date: Tue, 16 Jun 2026 09:47:24 +0000 Subject: [PATCH 2/4] Promote parsed official library docs --- .../chunks.json | 3596 +++ .../doc_nav.json | 518 + .../financial-micron-report-530bd7ed/full.md | 508 + .../images/image-1-Abstract Grid Pattern.jpg | Bin 0 -> 201734 bytes .../images/image-10-Concert Photography.jpg | Bin 0 -> 70971 bytes .../image-11-Futuristic Car Interior.jpg | Bin 0 -> 145941 bytes .../image-12-Business Meeting Discussion.jpg | Bin 0 -> 84296 bytes ...mage-13-Silicon Wafer in Manufacturing.jpg | Bin 0 -> 181024 bytes .../image-14-Semiconductor Cleanroom.jpg | Bin 0 -> 204111 bytes .../image-15-Laboratory Safety Work.jpg | Bin 0 -> 132435 bytes .../image-16-Corporate Executive Portrait.jpg | Bin 0 -> 132532 bytes .../images/image-17-Abstract Digital Art.jpg | Bin 0 -> 92300 bytes .../images/image-18-Abstract Digital Art.jpg | Bin 0 -> 87349 bytes .../image-19-Micron Presentation Slide.jpg | Bin 0 -> 121273 bytes .../image-2-Micron Executive Portrait.jpg | Bin 0 -> 249370 bytes .../image-20-Silicon Wafer Close-up.jpg | Bin 0 -> 89306 bytes .../image-3-Automated RAM Testing Machine.jpg | Bin 0 -> 176695 bytes ...image-4-Automated Industrial Machinery.jpg | Bin 0 -> 133194 bytes .../image-5-Cleanroom Manufacturing Floor.jpg | Bin 0 -> 160019 bytes .../image-6-Micron Semiconductor Wafer.jpg | Bin 0 -> 108122 bytes .../image-7-Server Rack Infrastructure.jpg | Bin 0 -> 140113 bytes .../images/image-8-Micron Data Center.jpg | Bin 0 -> 117432 bytes .../image-9-Laptop Keyboard and Screen.jpg | Bin 0 -> 115983 bytes .../manifest.json | 97 + .../tables/table-0 Segment Financials.html | 1 + .../tables/table-1 FQ1-26 Cash Flow.html | 1 + .../tables/table-10 Memory Revenue.html | 1 + .../tables/table-11 Revenue by Segment.html | 1 + .../tables/table-12 Gross Margin & OpEx.html | 1 + .../tables/table-13 Non-GAAP Financials.html | 1 + .../tables/table-14 Non-GAAP Net Income.html | 1 + .../tables/table-15 Adjusted EBITDA.html | 1 + .../tables/table-16 Financial Highlights.html | 1 + .../tables/table-17 Cash Position.html | 1 + .../tables/table-18 GAAP vs Non-GAAP.html | 1 + .../tables/table-19 Stock Compensation.html | 1 + .../tables/table-2 Financial Data.html | 1 + .../tables/table-3 Financial Results.html | 1 + .../tables/table-4 Revenue Breakdown.html | 1 + .../tables/table-5 Sales Volume Change.html | 1 + .../tables/table-6 ASP Changes Q_Q.html | 1 + .../tables/table-7 Non-GAAP Guidance.html | 1 + .../tables/table-8 Non-GAAP Estimates.html | 1 + .../tables/table-9 Memory Revenue.html | 1 + .../chunks.json | 3847 ++++ .../doc_nav.json | 542 + .../financial-micron-report-9c0becf5/full.md | 555 + .../images/image-1-Abstract Grid Pattern.jpg | Bin 0 -> 201734 bytes .../image-10-Laptop Keyboard and Screen.jpg | Bin 0 -> 106761 bytes .../images/image-11-Concert Photography.jpg | Bin 0 -> 70961 bytes ...-12-Micron Automotive Interior Concept.jpg | Bin 0 -> 110528 bytes ...ot GR-12 Carrying Package in Warehouse.jpg | Bin 0 -> 73965 bytes .../images/image-14-Business Meeting.jpg | Bin 0 -> 84075 bytes ...ge-15-Semiconductor Cleanroom Facility.jpg | Bin 0 -> 203611 bytes .../image-16-Laboratory Safety Work.jpg | Bin 0 -> 132252 bytes .../image-17-Construction Site Progress.jpg | Bin 0 -> 210159 bytes .../image-18-Micron Executive Portrait.jpg | Bin 0 -> 132945 bytes .../image-19-Abstract Digital Background.jpg | Bin 0 -> 92050 bytes .../image-2-Micron Executive Portrait.jpg | Bin 0 -> 249370 bytes .../images/image-20-Abstract Digital Wave.jpg | Bin 0 -> 87124 bytes .../image-21-Micron Presentation Slide.jpg | Bin 0 -> 121261 bytes .../images/image-22-Microchip Wafer.jpg | Bin 0 -> 89074 bytes .../image-23-Abstract Digital Background.jpg | Bin 0 -> 128825 bytes .../image-3-Automated RAM Testing Machine.jpg | Bin 0 -> 176502 bytes ...mage-4-Industrial Automation Machinery.jpg | Bin 0 -> 132958 bytes .../images/image-5-Cleanroom Operations.jpg | Bin 0 -> 160877 bytes .../images/image-6-Silicon Wafer Close-up.jpg | Bin 0 -> 73613 bytes .../image-7-Micron Semiconductor Wafer.jpg | Bin 0 -> 108122 bytes .../images/image-8-Micron Data Center.jpg | Bin 0 -> 117432 bytes .../images/image-9-Server Room Interior.jpg | Bin 0 -> 124701 bytes .../manifest.json | 98 + .../tables/table-0 Segment Financials.html | 1 + .../tables/table-1 Cash Flow Data.html | 1 + .../tables/table-10 Memory Revenue.html | 1 + .../tables/table-11 Revenue Growth.html | 1 + .../tables/table-12 Financial Results.html | 1 + .../tables/table-13 Non-GAAP Financials.html | 1 + .../tables/table-14 Non-GAAP Net Income.html | 1 + .../tables/table-15 Adjusted EBITDA.html | 1 + .../tables/table-16 Financial Data.html | 1 + .../tables/table-17 Cash Position.html | 1 + .../tables/table-18 GAAP vs Non-GAAP.html | 1 + .../tables/table-19 Stock Compensation.html | 1 + .../tables/table-2 Financial Data.html | 1 + .../tables/table-3 Quarterly Financials.html | 1 + .../tables/table-4 Revenue Breakdown.html | 1 + .../tables/table-5 Sales Volume Change.html | 1 + .../tables/table-6 ASP Changes Q_Q.html | 1 + .../tables/table-7 FQ2-26 vs FQ3-26.html | 1 + .../tables/table-8 Non-GAAP Estimates.html | 1 + .../tables/table-9 Memory Revenue.html | 1 + .../stem-transformers-tutorial/chunks.json | 18070 ++++++++++++++++ .../stem-transformers-tutorial/doc_nav.json | 3062 +++ .../stem-transformers-tutorial/full.md | 6089 ++++++ .../images/image-1-Profile Photo of a Man.jpg | Bin 0 -> 12322 bytes .../images/image-10-Model cat.jpg | Bin 0 -> 741 bytes .../images/image-100-__details_.jpg | Bin 0 -> 10173 bytes .../images/image-101-__details_.jpg | Bin 0 -> 10058 bytes .../images/image-102-__details_.jpg | Bin 0 -> 9280 bytes .../image-103-RL Algorithm Architecture.jpg | Bin 0 -> 28537 bytes .../images/image-104-RLHF Process Diagram.jpg | Bin 0 -> 30725 bytes ...age-105-Direct Preference Optimization.jpg | Bin 0 -> 18513 bytes .../images/image-106-Robot Head Close-Up.jpg | Bin 0 -> 20135 bytes .../images/image-107-Architecture Diagram.jpg | Bin 0 -> 41899 bytes .../images/image-108-ChatGPT Overview.jpg | Bin 0 -> 32589 bytes .../image-109-Avocado Armchair Designs.jpg | Bin 0 -> 45090 bytes .../images/image-11-wolf fox.jpg | Bin 0 -> 725 bytes .../images/image-110-Millennium Falcon.jpg | Bin 0 -> 39724 bytes .../images/image-111-Millennium Falcon.jpg | Bin 0 -> 39665 bytes .../images/image-112-Three Devils.jpg | Bin 0 -> 210151 bytes .../image-113-Call Phrase Logic Flow.jpg | Bin 0 -> 27728 bytes .../images/image-114-E --__YES_.jpg | Bin 0 -> 10112 bytes .../image-115-Sentence Structure Diagram.jpg | Bin 0 -> 14749 bytes ...e-116-Linguistic Relationships Diagram.jpg | Bin 0 -> 26646 bytes .../image-117-LSTM Cell Architecture.jpg | Bin 0 -> 23559 bytes ...118-Recurrent Neural Network Unfolding.jpg | Bin 0 -> 29014 bytes ...ge-119-Attention Visualization Layer 5.jpg | Bin 0 -> 35107 bytes .../images/image-12-dog squirr.jpg | Bin 0 -> 707 bytes ...e-120-Transformer Architecture Diagram.jpg | Bin 0 -> 43107 bytes .../images/image-121-__details_.jpg | Bin 0 -> 9452 bytes ...22-RL Algorithm Reward Prediction Flow.jpg | Bin 0 -> 28652 bytes .../images/image-123-ChatGPT Interface.jpg | Bin 0 -> 38432 bytes .../images/image-124-GPT Model Evolution.jpg | Bin 0 -> 53593 bytes .../images/image-125-Architecture Diagram.jpg | Bin 0 -> 41685 bytes .../images/image-126-ChatGPT Overview.jpg | Bin 0 -> 31702 bytes .../images/image-127-ChatGPT Dialogue.jpg | Bin 0 -> 64529 bytes .../images/image-128-OpenAI Whisper.jpg | Bin 0 -> 107875 bytes ...-Object and Person Detection on Bridge.jpg | Bin 0 -> 169430 bytes ...id_ Faces, Landscapes, Birds, and Bags.jpg | Bin 0 -> 104693 bytes .../image-130-Collage of Diverse Images.jpg | Bin 0 -> 222772 bytes .../images/image-131-Collage of.jpg | Bin 0 -> 5896 bytes .../images/image-132-_details_ .jpg | Bin 0 -> 10022 bytes .../images/image-133-_details_ .jpg | Bin 0 -> 4978 bytes .../images/image-134-_details_ .jpg | Bin 0 -> 7968 bytes .../images/image-135-_details_ .jpg | Bin 0 -> 9128 bytes .../images/image-136-_details_ .jpg | Bin 0 -> 7951 bytes .../images/image-137-_details_ .jpg | Bin 0 -> 7735 bytes .../images/image-138-_details_ .jpg | Bin 0 -> 9226 bytes .../images/image-139-_details_ .jpg | Bin 0 -> 5457 bytes .../images/image-14-Wildlife in Savanna.jpg | Bin 0 -> 21879 bytes ...e-140-Person using robotic exoskeleton.jpg | Bin 0 -> 27688 bytes .../images/image-141-_details_ .jpg | Bin 0 -> 6408 bytes .../images/image-142-_details_ .jpg | Bin 0 -> 7358 bytes .../images/image-143-_details_ .jpg | Bin 0 -> 6920 bytes .../images/image-144-_details_ .jpg | Bin 0 -> 7184 bytes .../images/image-145-_details_ .jpg | Bin 0 -> 5932 bytes .../images/image-146-_details_ .jpg | Bin 0 -> 6461 bytes .../images/image-147-Alphago Documentary.jpg | Bin 0 -> 28079 bytes .../image-148-DeepMind StarCraft II.jpg | Bin 0 -> 56987 bytes .../image-149-OpenAI Five Dota 2 Gameplay.jpg | Bin 0 -> 83962 bytes .../images/image-15-Steampunk Sky City.jpg | Bin 0 -> 23882 bytes .../images/image-150-Med-PaLM M Diagram.jpg | Bin 0 -> 43331 bytes ...age-151-Google Med-Palm 2 Introduction.jpg | Bin 0 -> 84932 bytes ...-152-Google DeepMind Protein Structure.jpg | Bin 0 -> 31496 bytes ...image-153-Comparison of Input Capacity.jpg | Bin 0 -> 24101 bytes ...er Architecture and Corruption Process.jpg | Bin 0 -> 46591 bytes .../images/image-155-style Patc.jpg | Bin 0 -> 8094 bytes .../images/image-156-__details_.jpg | Bin 0 -> 7635 bytes .../images/image-157-__details_.jpg | Bin 0 -> 7222 bytes ...e-158-Architecture of a Gating Network.jpg | Bin 0 -> 21214 bytes .../image-159-ReAct Framework Diagram.jpg | Bin 0 -> 18811 bytes ...e-16-Cartesian Space (2d) Scatter Plot.jpg | Bin 0 -> 37628 bytes ...age-160-Performance on SQuAD Benchmark.jpg | Bin 0 -> 24041 bytes .../images/image-161-Q_ Roger h.jpg | Bin 0 -> 1495 bytes ...image-162-Prompting Methods Comparison.jpg | Bin 0 -> 44035 bytes .../image-163-Mind Map_ Balloon Science.jpg | Bin 0 -> 38665 bytes ...e-164-Self-Questioning Process Diagram.jpg | Bin 0 -> 58130 bytes .../image-165-Cute Robot Illustration.jpg | Bin 0 -> 13239 bytes .../images/image-166-OpenAI GPT-4 Chip.jpg | Bin 0 -> 100227 bytes .../image-167-Agent Architecture Diagram.jpg | Bin 0 -> 87119 bytes .../image-168-Multi-On Search Interface.jpg | Bin 0 -> 110802 bytes ...image-169-Online Driver's License Test.jpg | Bin 0 -> 82747 bytes .../image-17-Feature Space Scatter Plot.jpg | Bin 0 -> 32082 bytes .../image-170-Automated Driving Levels.jpg | Bin 0 -> 109236 bytes .../images/image-171-Gorilla Tools Logo.jpg | Bin 0 -> 13758 bytes .../images/image-172-_details_ .jpg | Bin 0 -> 5387 bytes .../images/image-173-#### Direc.jpg | Bin 0 -> 6227 bytes ...-174-LangChain and Multion API Example.jpg | Bin 0 -> 65376 bytes .../images/image-175-OpenAI GPT-4 Chip.jpg | Bin 0 -> 48639 bytes ...e-176-Transformer Architecture Diagram.jpg | Bin 0 -> 62404 bytes .../images/image-177-Memory Component.jpg | Bin 0 -> 10786 bytes .../image-178-Cute Robot Illustration.jpg | Bin 0 -> 13251 bytes .../image-179-Cute Robot Illustration.jpg | Bin 0 -> 13256 bytes .../image-18-Classification Diagram.jpg | Bin 0 -> 39313 bytes .../image-180-Cute Robot Illustration.jpg | Bin 0 -> 13198 bytes .../images/image-181-Robot Icon.jpg | Bin 0 -> 13141 bytes .../image-182-Cute Robot Illustration.jpg | Bin 0 -> 13081 bytes .../image-183-AI Agent Architecture.jpg | Bin 0 -> 50984 bytes .../image-184-Cute Robot Illustration.jpg | Bin 0 -> 11688 bytes .../images/image-185-_details_ .jpg | Bin 0 -> 4070 bytes .../image-186-Robot Mascot Illustration.jpg | Bin 0 -> 11729 bytes .../images/image-187-_details_ .jpg | Bin 0 -> 4098 bytes .../images/image-188-Logo.jpg | Bin 0 -> 11106 bytes .../images/image-189-_details_ .jpg | Bin 0 -> 4275 bytes ...19-Diagram_ Cat and Dog Classification.jpg | Bin 0 -> 36967 bytes .../image-190-Cute Robot Illustration.jpg | Bin 0 -> 10853 bytes .../images/image-191-_details_ .jpg | Bin 0 -> 4276 bytes .../images/image-192-_details_ .jpg | Bin 0 -> 4286 bytes .../images/image-193-Simple car.jpg | Bin 0 -> 1232 bytes .../image-194-Task Verification Flow.jpg | Bin 0 -> 75083 bytes .../images/image-195-Simple car.jpg | Bin 0 -> 1263 bytes .../image-196-Cute Robot Illustration.jpg | Bin 0 -> 10786 bytes .../images/image-197-_details_ .jpg | Bin 0 -> 4259 bytes ...ge-198-Diagram of Actual vs Ideal Path.jpg | Bin 0 -> 24329 bytes ...ge-199-LLM System Architecture Diagram.jpg | Bin 0 -> 73019 bytes .../images/image-2-__details_.jpg | Bin 0 -> 2216 bytes .../images/image-20-2. Feed th.jpg | Bin 0 -> 8655 bytes .../images/image-200-M[_Etherne.jpg | Bin 0 -> 6432 bytes ...image-201-Neural Computer Architecture.jpg | Bin 0 -> 74343 bytes .../images/image-21-Sound Signal Waveform.jpg | Bin 0 -> 13833 bytes .../image-22-Alignment Visualization.jpg | Bin 0 -> 44515 bytes .../image-23-Word Embedding Visualization.jpg | Bin 0 -> 10765 bytes .../image-24-Word Embedding Visualization.jpg | Bin 0 -> 11860 bytes ...image-25-Diagram of Country-City Pairs.jpg | Bin 0 -> 30510 bytes .../images/image-26-1. Tokeniz.jpg | Bin 0 -> 1770 bytes .../images/image-27-Tokenizati.jpg | Bin 0 -> 1752 bytes ...ge-28-Transformer Encoder Architecture.jpg | Bin 0 -> 27963 bytes .../image-29-Sentence Structure Diagram.jpg | Bin 0 -> 29106 bytes .../image-3-Timeline of AI Evolution.jpg | Bin 0 -> 64277 bytes .../image-30-Sentence Structure Analysis.jpg | Bin 0 -> 28630 bytes ...ge-31-Transformer Architecture Diagram.jpg | Bin 0 -> 102780 bytes .../image-32-BERT Architecture Diagram.jpg | Bin 0 -> 75148 bytes .../image-33-Book Cover_ Deep Learning.jpg | Bin 0 -> 21866 bytes .../image-34-BERT Masked Language Model.jpg | Bin 0 -> 61192 bytes .../image-35-BERT Architecture Diagram.jpg | Bin 0 -> 49131 bytes ...ge-36-Transformer Architecture Diagram.jpg | Bin 0 -> 47918 bytes ...-37-Architecture of Vision Transformer.jpg | Bin 0 -> 70520 bytes .../images/image-38-__details_.jpg | Bin 0 -> 2871 bytes .../images/image-39-_details_ .jpg | Bin 0 -> 9758 bytes .../images/image-4-Composite_.jpg | Bin 0 -> 8545 bytes .../image-40-BERT Architecture Diagram.jpg | Bin 0 -> 47145 bytes .../image-41-GPT Decoder Architecture.jpg | Bin 0 -> 43934 bytes ...ok Cover_ Build a Large Language Model.jpg | Bin 0 -> 17136 bytes .../image-43-Neural Network Architecture.jpg | Bin 0 -> 60639 bytes .../image-44-Book Cover_ Deep Learning.jpg | Bin 0 -> 20816 bytes .../image-45-Ambiguous Pronoun Resolution.jpg | Bin 0 -> 61841 bytes ...mage-46-Attention Matrix Visualization.jpg | Bin 0 -> 53215 bytes ...mage-47-Attention Matrix Visualization.jpg | Bin 0 -> 48360 bytes .../image-48-Autoregressive Process.jpg | Bin 0 -> 77292 bytes ...ok Cover_ Build a Large Language Model.jpg | Bin 0 -> 17146 bytes .../images/image-5-Sound Signal Waveform.jpg | Bin 0 -> 14568 bytes .../images/image-50-__details_.jpg | Bin 0 -> 2939 bytes ...ok Cover_ Build a Large Language Model.jpg | Bin 0 -> 17136 bytes .../image-52-LLM Word Prediction Process.jpg | Bin 0 -> 104245 bytes .../images/image-53-_details_ .jpg | Bin 0 -> 2887 bytes ...54-Encoder-Decoder Translation Process.jpg | Bin 0 -> 88407 bytes .../images/image-55-Raschka, S.jpg | Bin 0 -> 7056 bytes ...56-Encoder-Decoder Translation Process.jpg | Bin 0 -> 93345 bytes .../images/image-57-GPT Decoder Process.jpg | Bin 0 -> 25071 bytes .../images/image-58-Decoder Process Flow.jpg | Bin 0 -> 33575 bytes ...ok Cover_ Build a Large Language Model.jpg | Bin 0 -> 16327 bytes .../images/image-6-A Cat in the Snow.jpg | Bin 0 -> 11384 bytes ...ge-60-Transformer Architecture Diagram.jpg | Bin 0 -> 84944 bytes .../images/image-61-_details_ .jpg | Bin 0 -> 2832 bytes .../image-62-Matrix Structure Diagram.jpg | Bin 0 -> 13083 bytes ...mage-63-Transformer Block Architecture.jpg | Bin 0 -> 42901 bytes .../image-64-Attention Mechanism Diagram.jpg | Bin 0 -> 24218 bytes .../image-65-Softmax Function Example.jpg | Bin 0 -> 49644 bytes .../image-66-Diagram of Input Processing.jpg | Bin 0 -> 19152 bytes .../images/image-67-__details_.jpg | Bin 0 -> 9019 bytes ...e-68-Attention Mechanism Visualization.jpg | Bin 0 -> 23872 bytes ...ge-69-Attention Mechanism Architecture.jpg | Bin 0 -> 23656 bytes .../image-7-Text Classification Model.jpg | Bin 0 -> 40382 bytes .../image-70-Attention Visualization.jpg | Bin 0 -> 48538 bytes ...image-71-Specialized Information Types.jpg | Bin 0 -> 73069 bytes ...e-72-Multi-head Attention Architecture.jpg | Bin 0 -> 60326 bytes .../images/image-73-#### 2) We.jpg | Bin 0 -> 2200 bytes .../images/image-74-#### 4) Ca.jpg | Bin 0 -> 2548 bytes .../images/image-75-5) Concate.jpg | Bin 0 -> 5588 bytes .../images/image-76-_summary_t.jpg | Bin 0 -> 3558 bytes .../image-77-W\342\202\200^Q _deta.jpg" | Bin 0 -> 1978 bytes .../image-78-W\342\202\200^K _deta.jpg" | Bin 0 -> 5320 bytes .../images/image-79-_summary_t.jpg | Bin 0 -> 3353 bytes .../image-8-Text Generation Model Process.jpg | Bin 0 -> 55090 bytes .../image-80-W\342\202\201^Q _deta.jpg" | Bin 0 -> 1797 bytes .../image-81-W\342\202\201^K _deta.jpg" | Bin 0 -> 5648 bytes .../images/image-82-_summary_t.jpg | Bin 0 -> 3894 bytes .../images/image-83-_summary_t.jpg | Bin 0 -> 1725 bytes .../image-84-Q\342\202\207 _detail.jpg" | Bin 0 -> 7740 bytes .../image-85-Transformer Decoder Block.jpg | Bin 0 -> 30861 bytes ...mage-86-Attention Matrix Visualization.jpg | Bin 0 -> 53010 bytes ...mage-87-Attention Matrix Visualization.jpg | Bin 0 -> 50727 bytes ...ge-88-Transformer Architecture Diagram.jpg | Bin 0 -> 62064 bytes ...89-Neural Network Architecture Diagram.jpg | Bin 0 -> 56036 bytes .../images/image-9-Fill- dog.jpg | Bin 0 -> 751 bytes .../image-90-Attention Mechanism Diagram.jpg | Bin 0 -> 56930 bytes ...mage-91-Transformer Architecture Block.jpg | Bin 0 -> 42132 bytes ...ge-92-Transformer Architecture Diagram.jpg | Bin 0 -> 53874 bytes .../images/image-93-Diagram.jpg | Bin 0 -> 16529 bytes .../images/image-94-B --_ C[_L.jpg | Bin 0 -> 5608 bytes .../images/image-95-C --_ D[_O.jpg | Bin 0 -> 9601 bytes .../images/image-96-__details_.jpg | Bin 0 -> 9265 bytes .../images/image-97-__details_.jpg | Bin 0 -> 9175 bytes .../images/image-98-__details_.jpg | Bin 0 -> 9465 bytes .../images/image-99-__details_.jpg | Bin 0 -> 8980 bytes .../stem-transformers-tutorial/manifest.json | 317 + .../tables/table-0 Transformer Era.html | 38 + .../tables/table-1 Frequency Data.html | 410 + .../tables/table-10 Animal Street.html | 66 + .../table-11 Word Transition Matrix.html | 69 + .../table-12 Word Transition Matrix.html | 69 + .../tables/table-13 RNNs vs Transformers.html | 1 + .../tables/table-14 Accuracy vs x.html | 26 + .../tables/table-15 BLEU vs X.html | 26 + .../tables/table-16 Exact match rates.html | 26 + .../tables/table-17 Exact match rates.html | 26 + .../tables/table-18 Accuracy Table.html | 22 + .../tables/table-19 Accuracy vs x.html | 26 + .../tables/table-2 Grouped Data.html | 21 + .../tables/table-20 Accuracy vs x.html | 22 + .../tables/table-21 Accuracy by Value.html | 22 + .../tables/table-22 Table Data.html | 1 + .../tables/table-23 Table Data.html | 31 + .../tables/table-24 RNNs vs Transformers.html | 1 + .../tables/table-25 Accuracy vs x.html | 26 + .../tables/table-26 BLEU vs X.html | 26 + .../tables/table-27 Exact match rates.html | 26 + .../tables/table-28 Exact match rates.html | 26 + .../tables/table-29 Accuracy Table.html | 22 + .../tables/table-3 Data Table Summary.html | 546 + .../tables/table-30 Accuracy vs x.html | 26 + .../tables/table-31 Accuracy vs x.html | 22 + .../tables/table-32 Accuracy vs. Scale.html | 22 + .../tables/table-33 Age vs Sleep Bounds.html | 31 + .../tables/table-34 Probability Table.html | 36 + .../tables/table-35 MLP Layer Prob.html | 46 + .../table-36 Attn Layers & Seattle.html | 38 + .../tables/table-37 Reflexion Trials.html | 48 + .../tables/table-38 Automation Levels.html | 60 + .../tables/table-39 Instruction Format.html | 1 + .../tables/table-4 Frequency Data.html | 334 + .../tables/table-40 Path Values.html | 410 + .../tables/table-5 Token Sequence.html | 1 + .../tables/table-6 Numerical Data.html | 1 + .../table-7 Word Transition Matrix.html | 69 + .../table-8 Word Transition Matrix.html | 69 + .../tables/table-9 Table Data.html | 31 + .../services/demo/official_library_catalog.py | 6 + apps/api/app/services/demo/source_catalog.py | 33 + .../contract/test_demo_documents_contract.py | 24 +- 340 files changed, 40217 insertions(+), 5 deletions(-) create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/full.md create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-1-Abstract Grid Pattern.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-10-Concert Photography.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-11-Futuristic Car Interior.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-12-Business Meeting Discussion.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-13-Silicon Wafer in Manufacturing.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-14-Semiconductor Cleanroom.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-15-Laboratory Safety Work.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-16-Corporate Executive Portrait.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-17-Abstract Digital Art.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-18-Abstract Digital Art.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-19-Micron Presentation Slide.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-2-Micron Executive Portrait.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-20-Silicon Wafer Close-up.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-3-Automated RAM Testing Machine.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-4-Automated Industrial Machinery.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-5-Cleanroom Manufacturing Floor.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-6-Micron Semiconductor Wafer.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-7-Server Rack Infrastructure.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-8-Micron Data Center.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-9-Laptop Keyboard and Screen.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-0 Segment Financials.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-1 FQ1-26 Cash Flow.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-10 Memory Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-11 Revenue by Segment.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-12 Gross Margin & OpEx.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-13 Non-GAAP Financials.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-14 Non-GAAP Net Income.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-15 Adjusted EBITDA.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-16 Financial Highlights.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-17 Cash Position.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-18 GAAP vs Non-GAAP.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-19 Stock Compensation.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-2 Financial Data.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-3 Financial Results.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-4 Revenue Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-5 Sales Volume Change.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-6 ASP Changes Q_Q.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-7 Non-GAAP Guidance.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-8 Non-GAAP Estimates.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/tables/table-9 Memory Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/chunks.json create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/full.md create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-1-Abstract Grid Pattern.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-10-Laptop Keyboard and Screen.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-11-Concert Photography.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-12-Micron Automotive Interior Concept.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-13-Robot GR-12 Carrying Package in Warehouse.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-14-Business Meeting.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-15-Semiconductor Cleanroom Facility.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-16-Laboratory Safety Work.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-17-Construction Site Progress.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-18-Micron Executive Portrait.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-19-Abstract Digital Background.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-2-Micron Executive Portrait.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-20-Abstract Digital Wave.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-21-Micron Presentation Slide.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-22-Microchip Wafer.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-23-Abstract Digital Background.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-3-Automated RAM Testing Machine.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-4-Industrial Automation Machinery.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-5-Cleanroom Operations.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-6-Silicon Wafer Close-up.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-7-Micron Semiconductor Wafer.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-8-Micron Data Center.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/images/image-9-Server Room Interior.jpg create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/manifest.json create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-0 Segment Financials.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-1 Cash Flow Data.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-10 Memory Revenue.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-11 Revenue Growth.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-12 Financial Results.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-13 Non-GAAP Financials.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-14 Non-GAAP Net Income.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-15 Adjusted EBITDA.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-16 Financial Data.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-17 Cash Position.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-18 GAAP vs Non-GAAP.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-19 Stock Compensation.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-2 Financial Data.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-3 Quarterly Financials.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-4 Revenue Breakdown.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-5 Sales Volume Change.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-6 ASP Changes Q_Q.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-7 FQ2-26 vs FQ3-26.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-8 Non-GAAP Estimates.html create mode 100644 apps/api/app/data/demo_documents/financial-micron-report-9c0becf5/tables/table-9 Memory Revenue.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/chunks.json create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/doc_nav.json create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/full.md create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-1-Profile Photo of a Man.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-10-Model cat.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-100-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-101-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-102-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-103-RL Algorithm Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-104-RLHF Process Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-105-Direct Preference Optimization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-106-Robot Head Close-Up.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-107-Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-108-ChatGPT Overview.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-109-Avocado Armchair Designs.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-11-wolf fox.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-110-Millennium Falcon.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-111-Millennium Falcon.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-112-Three Devils.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-113-Call Phrase Logic Flow.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-114-E --__YES_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-115-Sentence Structure Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-116-Linguistic Relationships Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-117-LSTM Cell Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-118-Recurrent Neural Network Unfolding.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-119-Attention Visualization Layer 5.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-12-dog squirr.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-120-Transformer Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-121-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-122-RL Algorithm Reward Prediction Flow.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-123-ChatGPT Interface.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-124-GPT Model Evolution.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-125-Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-126-ChatGPT Overview.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-127-ChatGPT Dialogue.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-128-OpenAI Whisper.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-129-Object and Person Detection on Bridge.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-13-Image Grid_ Faces, Landscapes, Birds, and Bags.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-130-Collage of Diverse Images.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-131-Collage of.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-132-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-133-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-134-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-135-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-136-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-137-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-138-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-139-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-14-Wildlife in Savanna.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-140-Person using robotic exoskeleton.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-141-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-142-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-143-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-144-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-145-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-146-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-147-Alphago Documentary.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-148-DeepMind StarCraft II.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-149-OpenAI Five Dota 2 Gameplay.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-15-Steampunk Sky City.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-150-Med-PaLM M Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-151-Google Med-Palm 2 Introduction.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-152-Google DeepMind Protein Structure.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-153-Comparison of Input Capacity.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-154-Diagram of Transformer Architecture and Corruption Process.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-155-style Patc.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-156-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-157-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-158-Architecture of a Gating Network.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-159-ReAct Framework Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-16-Cartesian Space (2d) Scatter Plot.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-160-Performance on SQuAD Benchmark.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-161-Q_ Roger h.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-162-Prompting Methods Comparison.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-163-Mind Map_ Balloon Science.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-164-Self-Questioning Process Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-165-Cute Robot Illustration.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-166-OpenAI GPT-4 Chip.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-167-Agent Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-168-Multi-On Search Interface.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-169-Online Driver's License Test.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-17-Feature Space Scatter Plot.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-170-Automated Driving Levels.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-171-Gorilla Tools Logo.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-172-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-173-#### Direc.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-174-LangChain and Multion API Example.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-175-OpenAI GPT-4 Chip.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-176-Transformer Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-177-Memory Component.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-178-Cute Robot Illustration.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-179-Cute Robot Illustration.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-18-Classification Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-180-Cute Robot Illustration.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-181-Robot Icon.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-182-Cute Robot Illustration.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-183-AI Agent Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-184-Cute Robot Illustration.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-185-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-186-Robot Mascot Illustration.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-187-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-188-Logo.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-189-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-19-Diagram_ Cat and Dog Classification.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-190-Cute Robot Illustration.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-191-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-192-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-193-Simple car.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-194-Task Verification Flow.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-195-Simple car.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-196-Cute Robot Illustration.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-197-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-198-Diagram of Actual vs Ideal Path.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-199-LLM System Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-2-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-20-2. Feed th.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-200-M[_Etherne.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-201-Neural Computer Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-21-Sound Signal Waveform.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-22-Alignment Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-23-Word Embedding Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-24-Word Embedding Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-25-Diagram of Country-City Pairs.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-26-1. Tokeniz.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-27-Tokenizati.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-28-Transformer Encoder Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-29-Sentence Structure Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-3-Timeline of AI Evolution.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-30-Sentence Structure Analysis.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-31-Transformer Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-32-BERT Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-33-Book Cover_ Deep Learning.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-34-BERT Masked Language Model.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-35-BERT Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-36-Transformer Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-37-Architecture of Vision Transformer.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-38-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-39-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-4-Composite_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-40-BERT Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-41-GPT Decoder Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-42-Book Cover_ Build a Large Language Model.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-43-Neural Network Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-44-Book Cover_ Deep Learning.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-45-Ambiguous Pronoun Resolution.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-46-Attention Matrix Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-47-Attention Matrix Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-48-Autoregressive Process.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-49-Book Cover_ Build a Large Language Model.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-5-Sound Signal Waveform.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-50-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-51-Book Cover_ Build a Large Language Model.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-52-LLM Word Prediction Process.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-53-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-54-Encoder-Decoder Translation Process.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-55-Raschka, S.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-56-Encoder-Decoder Translation Process.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-57-GPT Decoder Process.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-58-Decoder Process Flow.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-59-Book Cover_ Build a Large Language Model.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-6-A Cat in the Snow.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-60-Transformer Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-61-_details_ .jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-62-Matrix Structure Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-63-Transformer Block Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-64-Attention Mechanism Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-65-Softmax Function Example.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-66-Diagram of Input Processing.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-67-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-68-Attention Mechanism Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-69-Attention Mechanism Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-7-Text Classification Model.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-70-Attention Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-71-Specialized Information Types.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-72-Multi-head Attention Architecture.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-73-#### 2) We.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-74-#### 4) Ca.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-75-5) Concate.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-76-_summary_t.jpg create mode 100644 "apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-77-W\342\202\200^Q _deta.jpg" create mode 100644 "apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-78-W\342\202\200^K _deta.jpg" create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-79-_summary_t.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-8-Text Generation Model Process.jpg create mode 100644 "apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-80-W\342\202\201^Q _deta.jpg" create mode 100644 "apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-81-W\342\202\201^K _deta.jpg" create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-82-_summary_t.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-83-_summary_t.jpg create mode 100644 "apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-84-Q\342\202\207 _detail.jpg" create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-85-Transformer Decoder Block.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-86-Attention Matrix Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-87-Attention Matrix Visualization.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-88-Transformer Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-89-Neural Network Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-9-Fill- dog.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-90-Attention Mechanism Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-91-Transformer Architecture Block.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-92-Transformer Architecture Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-93-Diagram.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-94-B --_ C[_L.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-95-C --_ D[_O.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-96-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-97-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-98-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/images/image-99-__details_.jpg create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/manifest.json create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-0 Transformer Era.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-1 Frequency Data.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-10 Animal Street.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-11 Word Transition Matrix.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-12 Word Transition Matrix.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-13 RNNs vs Transformers.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-14 Accuracy vs x.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-15 BLEU vs X.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-16 Exact match rates.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-17 Exact match rates.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-18 Accuracy Table.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-19 Accuracy vs x.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-2 Grouped Data.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-20 Accuracy vs x.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-21 Accuracy by Value.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-22 Table Data.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-23 Table Data.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-24 RNNs vs Transformers.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-25 Accuracy vs x.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-26 BLEU vs X.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-27 Exact match rates.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-28 Exact match rates.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-29 Accuracy Table.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-3 Data Table Summary.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-30 Accuracy vs x.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-31 Accuracy vs x.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-32 Accuracy vs. Scale.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-33 Age vs Sleep Bounds.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-34 Probability Table.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-35 MLP Layer Prob.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-36 Attn Layers & Seattle.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-37 Reflexion Trials.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-38 Automation Levels.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-39 Instruction Format.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-4 Frequency Data.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-40 Path Values.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-5 Token Sequence.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-6 Numerical Data.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-7 Word Transition Matrix.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-8 Word Transition Matrix.html create mode 100644 apps/api/app/data/demo_documents/stem-transformers-tutorial/tables/table-9 Table Data.html diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/chunks.json b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/chunks.json new file mode 100644 index 000000000..3fb5197a2 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/chunks.json @@ -0,0 +1,3596 @@ +{ + "chunks": [ + { + "chunk_id": "e7206d28-6d5a-563d-a6fc-b6070a6a8bb9", + "type": "image", + "content": "\n# Financial results
    0\n[images/image-1-Abstract Grid Pattern.jpg]\n", + "path": "images/image-1-Abstract Grid Pattern.jpg", + "metadata": { + "length": 64, + "summary": "image-1\nA close-up view of a circular object featuring a precise grid pattern. The image displays vibrant, iridescent colors including magenta, cyan, green, and blue, which shift across the surface. The grid consists of intersecting horizontal and vertical lines creating small rectangular cells. The lighting creates a gradient effect, with the left side appearing darker and the right side brighter.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-1-Abstract Grid Pattern.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "7c7d3528-69e3-5333-ad98-4cfe628e6380", + "type": "text", + "content": "FQ1 2026\n\n# Financial results
    0\n[images/image-1-Abstract Grid Pattern.jpg]\n\n
    \nnatural_image\nClose-up of a semiconductor wafer with green, purple, and blue photovoltaic bands (no text or symbols visible)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results", + "metadata": { + "length": 238, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "FQ1", + "2026", + "Financial", + "results", + "details", + "summary", + "natural", + "image", + "Close", + "semiconductor", + "wafer", + "green", + "purple", + "blue", + "photovoltaic", + "bands", + "text", + "symbols", + "visible" + ], + "keywords": [], + "connect_to": [ + { + "target": "e7206d28-6d5a-563d-a6fc-b6070a6a8bb9", + "relation": "embeds", + "ref": "[images/image-1-Abstract Grid Pattern.jpg]", + "position": { + "start": 41, + "end": 83 + } + } + ] + } + }, + { + "chunk_id": "7e1a55c0-3a66-5053-a86e-44f555639f30", + "type": "text", + "content": "During the course of this meeting, we may make projections or other forward-looking statements regarding market demand and supply, market and pricing trends and drivers, the impact of technologies such as AI, contractual terms, cost reductions, our manufacturing projects and related investments, expected product volume production, our market position, expected product announcements, capabilities of our future products and technologies, future events, such as the end of life of some products, and future financial and operating performance including financial projections of the company and the industry. We wish to caution you that such statements are predictions, and that actual events or results may differ materially. We refer you to the documents the company files from time to time with the Securities and Exchange Commission, including the company's Form 10-K, Forms 10-Q and other reports and filings. These documents contain and identify important factors that could cause the actual results for the company to differ materially from those contained in our projections or forward-looking statements. These certain factors can be found at investors.micron.com/risk-factor. Although we believe that the expectations reflected in the forward-looking statements are reasonable, we cannot guarantee future results, levels of activity, performance or achievements. We are under no duty to update any of the forward-looking statements to conform these statements except as required by applicable law.\nThis presentation includes non-GAAP financial measures. Non-GAAP financial measures represent GAAP measures, excluding the impact of certain activities, which management excludes in analyzing our operating results and understanding trends in our earnings, adjusted free cash flow and business outlook. Further information regarding Micron's use of non-GAAP measures and reconciliations between GAAP and non-GAAP measures are included in the Appendix.", + "path": "micron-530bd7ed.pdf/Financial results/Safe harbor statement", + "metadata": { + "length": 1958, + "summary": "This passage contains standard legal disclaimers regarding forward-looking statements and non-GAAP financial measures. It cautions that projections about market demand, pricing, technology impact, manufacturing investments, and financial performance involve risks and uncertainties. It directs readers to SEC filings and a risk factor website for details. It also explains that non-GAAP measures exclude certain activities to help analyze operating results and trends, with reconciliations provided in the appendix.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "During", + "meeting", + "make", + "projections", + "forward", + "statements", + "market", + "demand", + "supply", + "pricing", + "trends", + "drivers", + "impact", + "technologies", + "AI", + "contractual", + "terms", + "cost", + "reductions", + "manufacturing", + "projects", + "related", + "investments", + "expected", + "product", + "volume", + "production", + "position", + "announcements", + "capabilities", + "future", + "products", + "events", + "end", + "life", + "financial", + "operating", + "performance", + "including", + "company", + "industry", + "We", + "caution", + "predictions", + "actual", + "results", + "differ", + "materially", + "refer", + "documents", + "files", + "time", + "Securities", + "Exchange", + "Commission", + "Form", + "10", + "Forms", + "reports", + "filings", + "These", + "identify", + "important", + "factors", + "contained", + "found", + "investors", + "micron", + "risk", + "factor", + "Although", + "expectations", + "reflected", + "reasonable", + "guarantee", + "levels", + "activity", + "achievements", + "duty", + "update", + "conform", + "required", + "applicable", + "law", + "This", + "presentation", + "includes", + "GAAP", + "measures", + "Non", + "represent", + "excluding", + "activities", + "management", + "excludes", + "analyzing", + "understanding", + "earnings", + "adjusted", + "free", + "cash", + "flow", + "business", + "outlook", + "Further", + "information", + "Micron", + "reconciliations", + "included", + "Appendix" + ], + "keywords": [ + "forward-looking", + "non-GAAP", + "disclaimer" + ], + "connect_to": [] + } + }, + { + "chunk_id": "671a6e23-08c1-5690-a692-03c5c9780718", + "type": "image", + "content": "\n## Safe harbor statement
    0\n[images/image-2-Micron Executive Portrait.jpg]\n", + "path": "images/image-2-Micron Executive Portrait.jpg", + "metadata": { + "length": 69, + "summary": "image-2\n- Subject: Professional headshot of a middle-aged man with graying hair and beard, smiling at the camera.\n- Attire: Wearing a textured blue blazer over a patterned light purple button-down shirt.\n- Background: Solid black studio background.\n- Visible Text/Logos: \"micron\" logo in the bottom right corner; page number \"3\".", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-2-Micron Executive Portrait.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "69038dff-0fdd-5511-af76-9c3e43f06242", + "type": "image", + "content": "\n
    0\n[images/image-3-Automated RAM Testing Machine.jpg]\n", + "path": "images/image-3-Automated RAM Testing Machine.jpg", + "metadata": { + "length": 54, + "summary": "image-3\n- Industrial automation equipment featuring a vertical array of mechanical grippers or probes.\n- Multiple green RAM memory modules (SODIMM type) are held in place by the machine's fixtures.\n- The machinery is illuminated with purple and red lighting, highlighting the precision components.\n- A conveyor system with metal rollers is visible at the bottom for transporting parts.\n- Date stamp \"December 17, 2025\" is visible at the bottom left.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-3-Automated RAM Testing Machine.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "4c9b8a05-10a4-5dfb-809f-4359ff99f6f2", + "type": "text", + "content": "Chairman, President and Chief Executive Officer\n\n## Safe harbor statement
    0\n[images/image-2-Micron Executive Portrait.jpg]\n\n
    \nnatural_image\nPortrait of a smiling man with gray hair and beard, wearing a blue blazer over a patterned shirt (no text or symbols visible)\n
    \n\n
    0\n[images/image-3-Automated RAM Testing Machine.jpg]\n\n
    \nnatural_image\nIndustrial automation scene with robotic arms and green memory chips under purple lighting (no readable text or symbols)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Sanjay Mehrotra", + "metadata": { + "length": 527, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Chairman", + "President", + "Chief", + "Executive", + "Officer", + "Safe", + "harbor", + "statement", + "details", + "summary", + "natural", + "image", + "Portrait", + "smiling", + "man", + "gray", + "hair", + "beard", + "wearing", + "blue", + "blazer", + "patterned", + "shirt", + "text", + "symbols", + "visible", + "Industrial", + "automation", + "scene", + "robotic", + "arms", + "green", + "memory", + "chips", + "purple", + "lighting", + "readable" + ], + "keywords": [], + "connect_to": [ + { + "target": "671a6e23-08c1-5690-a692-03c5c9780718", + "relation": "embeds", + "ref": "[images/image-2-Micron Executive Portrait.jpg]", + "position": { + "start": 85, + "end": 131 + } + }, + { + "target": "69038dff-0fdd-5511-af76-9c3e43f06242", + "relation": "embeds", + "ref": "[images/image-3-Automated RAM Testing Machine.jpg]", + "position": { + "start": 335, + "end": 385 + } + } + ] + } + }, + { + "chunk_id": "5dac29b5-e66d-5872-8dca-f5cbb3e1989a", + "type": "image", + "content": "\n- This financial performance was driven by our strong execution across end markets and products in a tight supply environment. We achieved a number of records in fiscal Q1.
    0\n[images/image-4-Automated Industrial Machinery.jpg]\n", + "path": "images/image-4-Automated Industrial Machinery.jpg", + "metadata": { + "length": 217, + "summary": "image-4\n- Primary Subject: Close-up view of an automated industrial machine, likely a conveyor or sorting system.\n- Key Features:\n - Central rotating carousel mechanism with metallic arms and platforms.\n - Transparent safety shielding (likely polycarbonate) surrounding the machinery.\n - Vertical LED light strips visible in the background.\n - Warning labels and sensors attached to the central column.\n- Context: The scene is illuminated with strong blue lighting, suggesting a clean room or high-tech manufacturing environment.\n- Visible Text: \"December 17, 2025\" located at the bottom left corner.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-4-Automated Industrial Machinery.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "bb6920d7-549a-5b61-a924-863e155d13e6", + "type": "text", + "content": "- Micron had an outstanding start to fiscal 2026, delivering fiscal Q1 revenue, gross margin and EPS (earnings per share) well above the high end of our guidance.\n- This financial performance was driven by our strong execution across end markets and products in a tight supply environment. We achieved a number of records in fiscal Q1.\n- Total company revenue, DRAM and NAND revenue, as well as HBM and data center revenue and revenue in each of our business units, also reached new records.\n• We have completed agreements on price and volume for our entire calendar 2026 HBM supply, including Micron's industry-leading HBM4.\n- We forecast an HBM TAM CAGR of approximately 40% through calendar 2028, from approximately \\$35 billion in 2025 to around \\$100 billion in 2028.\n- This \\$100 billion HBM TAM milestone is now projected to arrive two years earlier than in our prior outlook. Remarkably, this 2028 HBM TAM projection is larger than the size of the entire DRAM market in calendar 2024.\n\n- This financial performance was driven by our strong execution across end markets and products in a tight supply environment. We achieved a number of records in fiscal Q1.
    0\n[images/image-4-Automated Industrial Machinery.jpg]\n\n
    \nnatural_image\nInterior view of a modern industrial facility with blue-tinted machinery and control panels (no visible text or symbols)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Overview (1 of 3)", + "metadata": { + "length": 1385, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Micron", + "outstanding", + "start", + "fiscal", + "2026", + "delivering", + "Q1", + "revenue", + "gross", + "margin", + "EPS", + "earnings", + "share", + "high", + "end", + "guidance", + "This", + "financial", + "performance", + "driven", + "strong", + "execution", + "markets", + "products", + "tight", + "supply", + "environment", + "We", + "achieved", + "number", + "records", + "Total", + "company", + "DRAM", + "NAND", + "HBM", + "data", + "center", + "business", + "units", + "reached", + "completed", + "agreements", + "price", + "volume", + "entire", + "calendar", + "including", + "industry", + "leading", + "HBM4", + "forecast", + "TAM", + "CAGR", + "approximately", + "40%", + "2028", + "35", + "billion", + "2025", + "100", + "milestone", + "projected", + "arrive", + "years", + "earlier", + "prior", + "outlook", + "Remarkably", + "projection", + "larger", + "size", + "market", + "2024", + "details", + "summary", + "natural", + "image", + "Interior", + "view", + "modern", + "industrial", + "facility", + "blue", + "tinted", + "machinery", + "control", + "panels", + "visible", + "text", + "symbols" + ], + "keywords": [], + "connect_to": [ + { + "target": "5dac29b5-e66d-5872-8dca-f5cbb3e1989a", + "relation": "embeds", + "ref": "[images/image-4-Automated Industrial Machinery.jpg]", + "position": { + "start": 1178, + "end": 1229 + } + } + ] + } + }, + { + "chunk_id": "9d174ae5-31f3-55c3-b754-48c519c49f65", + "type": "image", + "content": "\n- We are excited about our customized HBM4E customer engagements, which offer further differentiation opportunities to us, and we continue to make excellent progress on our HBM roadmap.
    0\n[images/image-5-Cleanroom Manufacturing Floor.jpg]\n", + "path": "images/image-5-Cleanroom Manufacturing Floor.jpg", + "metadata": { + "length": 230, + "summary": "image-5\n- Workers wearing full-body white protective suits (bunny suits), hoods, and face masks.\n- Environment appears to be a high-tech manufacturing facility or cleanroom with specialized workstations.\n- Equipment includes computer monitors, keyboards, and various tools on metal tables.\n- A yellow label with the letters \"DE\" is visible on the back of an office chair in the foreground.\n- The floor has a textured, grid-like pattern typical of industrial settings.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-5-Cleanroom Manufacturing Floor.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "b5208ad5-2cd9-5f48-b2de-86be2d94b676", + "type": "text", + "content": "- We are excited about our customized HBM4E customer engagements, which offer further differentiation opportunities to us, and we continue to make excellent progress on our HBM roadmap.\n- Memory is now essential to AI's cognitive functions, fundamentally altering its role from a system component to a strategic asset that dictates product performance from data center to the edge.\n- This structural shift means that system capabilities heavily rely on advanced memory for real-time contextual processing, which is vital for achieving autonomous and intelligent behaviors in AI data centers as well as in applications ranging from self-driving cars to advanced medical diagnostics.\n- With our technology leadership, differentiated product portfolio, strong operational execution and solid balance sheet, Micron is in the best competitive position in its history and is one of the semiconductor industry's biggest enablers of AI.\n• We anticipate substantial new records in revenue, gross margin, EPS and free cash flow for both the second quarter and the full fiscal year 2026, and we expect our business performance to continue to strengthen through the year.\n\n- We are excited about our customized HBM4E customer engagements, which offer further differentiation opportunities to us, and we continue to make excellent progress on our HBM roadmap.
    0\n[images/image-5-Cleanroom Manufacturing Floor.jpg]\n\n
    \nnatural_image\nInterior of a cleanroom manufacturing facility with technicians in full protective suits working at workstations (no visible text or symbols)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Overview (2 of 3)", + "metadata": { + "length": 1586, + "summary": "Micron highlights its customized HBM4E customer engagements and progress on its HBM roadmap, emphasizing memory's shift from a system component to a strategic asset for AI. The company's technology leadership and product portfolio position it as a key AI enabler, with anticipated record revenue, gross margin, EPS, and free cash flow for Q2 and full fiscal year 2026.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "We", + "excited", + "customized", + "HBM4E", + "customer", + "engagements", + "offer", + "differentiation", + "opportunities", + "continue", + "make", + "excellent", + "progress", + "HBM", + "roadmap", + "Memory", + "essential", + "AI", + "cognitive", + "functions", + "fundamentally", + "altering", + "role", + "system", + "component", + "strategic", + "asset", + "dictates", + "product", + "performance", + "data", + "center", + "edge", + "This", + "structural", + "shift", + "means", + "capabilities", + "heavily", + "rely", + "advanced", + "memory", + "real", + "time", + "contextual", + "processing", + "vital", + "achieving", + "autonomous", + "intelligent", + "behaviors", + "centers", + "applications", + "ranging", + "driving", + "cars", + "medical", + "diagnostics", + "With", + "technology", + "leadership", + "differentiated", + "portfolio", + "strong", + "operational", + "execution", + "solid", + "balance", + "sheet", + "Micron", + "competitive", + "position", + "history", + "semiconductor", + "industry", + "biggest", + "enablers", + "anticipate", + "substantial", + "records", + "revenue", + "gross", + "margin", + "EPS", + "free", + "cash", + "flow", + "quarter", + "full", + "fiscal", + "year", + "2026", + "expect", + "business", + "strengthen", + "details", + "summary", + "natural", + "image", + "Interior", + "cleanroom", + "manufacturing", + "facility", + "technicians", + "protective", + "suits", + "working", + "workstations", + "visible", + "text", + "symbols" + ], + "keywords": [ + "HBM4E", + "AI memory", + "Micron" + ], + "connect_to": [ + { + "target": "9d174ae5-31f3-55c3-b754-48c519c49f65", + "relation": "embeds", + "ref": "[images/image-5-Cleanroom Manufacturing Floor.jpg]", + "position": { + "start": 1358, + "end": 1408 + } + } + ] + } + }, + { + "chunk_id": "ca6013a0-9e2b-5d4e-8e18-fcd9bdf3c486", + "type": "text", + "content": "- Sustained and strong industry demand, along with supply constraints, are contributing to tight market conditions, and we expect these conditions to persist beyond calendar 2026.\n• We are making progress with customers in our discussions for multiyear contracts with specific commitments.\n- Simultaneously, we are focused on maximizing our production output from our current footprint, ramping our industry-leading technology nodes and investing in new cleanroom space to add to our supply capability.", + "path": "micron-530bd7ed.pdf/Financial results/Overview (3 of 3)", + "metadata": { + "length": 502, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Sustained", + "strong", + "industry", + "demand", + "supply", + "constraints", + "contributing", + "tight", + "market", + "conditions", + "expect", + "persist", + "calendar", + "2026", + "We", + "making", + "progress", + "customers", + "discussions", + "multiyear", + "contracts", + "specific", + "commitments", + "Simultaneously", + "focused", + "maximizing", + "production", + "output", + "current", + "footprint", + "ramping", + "leading", + "technology", + "nodes", + "investing", + "cleanroom", + "space", + "add", + "capability" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "a3b29ebc-ae52-5371-98bc-7027feb10ab9", + "type": "text", + "content": "- Micron’s technology leadership is foundational to our strong competitive position.\n- Micron has led the industry for four consecutive technology nodes in DRAM and three nodes in NAND, with progressively faster yield ramps in every node.\n- Our 1-gamma DRAM node is ramping well. 1-gamma will be the primary driver of our DRAM bit growth in calendar 2026, and will be the majority of our bit output in the second half of the calendar year.\n- Looking beyond 1-gamma, development is underway for 1-delta and 1-epsilon nodes, which will feature innovations that we expect to extend our differentiation and technology leadership.", + "path": "micron-530bd7ed.pdf/Financial results/Technology and operations (1 of 2)", + "metadata": { + "length": 625, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Micron", + "technology", + "leadership", + "foundational", + "strong", + "competitive", + "position", + "led", + "industry", + "consecutive", + "nodes", + "DRAM", + "NAND", + "progressively", + "faster", + "yield", + "ramps", + "node", + "Our", + "gamma", + "ramping", + "primary", + "driver", + "bit", + "growth", + "calendar", + "2026", + "majority", + "output", + "half", + "year", + "Looking", + "development", + "underway", + "delta", + "epsilon", + "feature", + "innovations", + "expect", + "extend", + "differentiation" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "1ac76b45-020b-54d9-8b95-763e3b1d1bb5", + "type": "image", + "content": "\n## Technology and operations (2 of 2)
    0\n[images/image-6-Micron Semiconductor Wafer.jpg]\n", + "path": "images/image-6-Micron Semiconductor Wafer.jpg", + "metadata": { + "length": 82, + "summary": "image-6\n- Subject: Close-up photograph of a silicon semiconductor wafer featuring a grid pattern of microchips.\n- Visual Details: The image displays iridescent colors (red, orange, green, blue) caused by light interference on the wafer's surface layers.\n- Text/Logo: \"micron\" logo visible in the bottom right corner.\n- Page Number: \"8\" located next to the logo.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-6-Micron Semiconductor Wafer.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "76038959-82c6-5fbf-859c-ae99b3375cf1", + "type": "text", + "content": "- In NAND, we are ramping our G9 node with robust yield ramps across both data center and client SSDs. Our QLC NAND mix, including G9 QLC, reached a record high during the quarter.\n- Technology transitions to G9 will be the primary driver of our NAND bit growth in calendar 2026, and we expect it to become our largest NAND node later in fiscal 2026.\n- Calendar 2025 is a record year for Micron in terms of both internal and customer quality measures, positioning us well to deliver for our customers as the memory industry's quality leader.\n- As our products are increasingly integrated into higher value applications, our leadership in quality is becoming a more important differentiator.\n\n## Technology and operations (2 of 2)
    0\n[images/image-6-Micron Semiconductor Wafer.jpg]\n\n
    \ntext_image\nmicron | 8\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Technology and operations (2 of 2)", + "metadata": { + "length": 835, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "In", + "NAND", + "ramping", + "G9", + "node", + "robust", + "yield", + "ramps", + "data", + "center", + "client", + "SSDs", + "Our", + "QLC", + "mix", + "including", + "reached", + "record", + "high", + "quarter", + "Technology", + "transitions", + "primary", + "driver", + "bit", + "growth", + "calendar", + "2026", + "expect", + "largest", + "fiscal", + "Calendar", + "2025", + "year", + "Micron", + "terms", + "internal", + "customer", + "quality", + "measures", + "positioning", + "deliver", + "customers", + "memory", + "industry", + "leader", + "As", + "products", + "increasingly", + "integrated", + "higher", + "applications", + "leadership", + "important", + "differentiator", + "operations", + "details", + "summary", + "text", + "image", + "micron" + ], + "keywords": [], + "connect_to": [ + { + "target": "1ac76b45-020b-54d9-8b95-763e3b1d1bb5", + "relation": "embeds", + "ref": "[images/image-6-Micron Semiconductor Wafer.jpg]", + "position": { + "start": 741, + "end": 788 + } + } + ] + } + }, + { + "chunk_id": "b878aa87-2a0f-5ac9-9cee-0e082385dfd3", + "type": "image", + "content": "\n- As the world’s leading technology companies advance toward artificial general intelligence and transform the global economy, our customers are committing to an extraordinary, multiyear data center buildout.
    0\n[images/image-7-Server Rack Infrastructure.jpg]\n", + "path": "images/image-7-Server Rack Infrastructure.jpg", + "metadata": { + "length": 253, + "summary": "image-7\n- Multiple server racks with perforated mesh doors\n- Green and orange status indicator lights visible on equipment\n- Raised floor tiles typical of data center environments\n- Cable management tray visible at the top left\n- \"micron\" logo and page number \"9\" at the bottom", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-7-Server Rack Infrastructure.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "f141ce1f-2bb8-5ee0-ad19-181fc0c5f222", + "type": "text", + "content": "- As the world’s leading technology companies advance toward artificial general intelligence and transform the global economy, our customers are committing to an extraordinary, multiyear data center buildout.\n- This growth in AI data center capacity is driving a significant increase in demand for high-performance and high-capacity memory and storage.\n- Server unit demand has strengthened significantly, and we now expect calendar 2025 server unit growth in the high-teens percentage range, higher than our last earnings call outlook of $10\\%$ .\n• We expect server demand strength to continue in 2026. Server memory and storage content and performance requirements continue to increase generation to generation.\n- Micron has a differentiated portfolio of high-value data center solutions to address these requirements, including our HBM, high-capacity server memory solutions and data center SSDs.\n\n- As the world’s leading technology companies advance toward artificial general intelligence and transform the global economy, our customers are committing to an extraordinary, multiyear data center buildout.
    0\n[images/image-7-Server Rack Infrastructure.jpg]\n\n
    \nnatural_image\nInterior view of a server room with rows of server racks and illuminated green indicator lights, labeled 'micron' in the corner (no readable text on equipment)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Data center server demand", + "metadata": { + "length": 1367, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "As", + "world", + "leading", + "technology", + "companies", + "advance", + "artificial", + "general", + "intelligence", + "transform", + "global", + "economy", + "customers", + "committing", + "extraordinary", + "multiyear", + "data", + "center", + "buildout", + "This", + "growth", + "AI", + "capacity", + "driving", + "significant", + "increase", + "demand", + "high", + "performance", + "memory", + "storage", + "Server", + "unit", + "strengthened", + "significantly", + "expect", + "calendar", + "2025", + "server", + "teens", + "percentage", + "range", + "higher", + "earnings", + "call", + "outlook", + "10", + "We", + "strength", + "continue", + "2026", + "content", + "requirements", + "generation", + "Micron", + "differentiated", + "portfolio", + "solutions", + "address", + "including", + "HBM", + "SSDs", + "details", + "summary", + "natural", + "image", + "Interior", + "view", + "room", + "rows", + "racks", + "illuminated", + "green", + "indicator", + "lights", + "labeled", + "micron", + "corner", + "readable", + "text", + "equipment" + ], + "keywords": [], + "connect_to": [ + { + "target": "b878aa87-2a0f-5ac9-9cee-0e082385dfd3", + "relation": "embeds", + "ref": "[images/image-7-Server Rack Infrastructure.jpg]", + "position": { + "start": 1121, + "end": 1168 + } + } + ] + } + }, + { + "chunk_id": "07c78111-5577-517c-b221-cb28a67f4d01", + "type": "image", + "content": "\n- Micron’s HBM4, with industry leading speed over 11 Gbps, is on track to ramp with high yields in the second calendar quarter of 2026, consistent with our customers’ product ramp plans.
    0\n[images/image-8-Micron Data Center.jpg]\n", + "path": "images/image-8-Micron Data Center.jpg", + "metadata": { + "length": 231, + "summary": "image-8\n- Visual: A futuristic server room or data center aisle featuring rows of glass-fronted server racks. The scene is illuminated by vibrant pink and blue neon lighting, with glowing light strips running along the floor and ceiling infrastructure.\n- Text visible at bottom right: \"micron\" logo and page number \"10\".", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-8-Micron Data Center.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "1cf461ea-3915-52d3-856e-e6e1c67e1719", + "type": "text", + "content": "- Micron’s HBM4, with industry leading speed over 11 Gbps, is on track to ramp with high yields in the second calendar quarter of 2026, consistent with our customers’ product ramp plans.\n- Our HBM4 uses advanced CMOS and advanced metallization process technologies on the base logic die and DRAM core dies, which are designed and manufactured in-house.\n- This, along with our unique HBM design, packaging and test capability, enables Micron’s industry-leading performance and low-power leadership.\n- Micron pioneered the adoption of LP DRAM in the data center. Micron's Low Power (LP) DRAM server modules consume one-third the power of DDR DRAM server modules.\n- Building on this leadership, we have sampled our 192GB LP SOCAMM2 product, which enables a $50\\%$ increase in capacity per module and a rack-scale LP DRAM density of over 50TB.\n\n- Micron’s HBM4, with industry leading speed over 11 Gbps, is on track to ramp with high yields in the second calendar quarter of 2026, consistent with our customers’ product ramp plans.
    0\n[images/image-8-Micron Data Center.jpg]\n\n
    \nnatural_image\nInterior view of a modern data center with server racks illuminated by purple and blue neon lights (no visible text or symbols)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Data center HBM and LP DRAM", + "metadata": { + "length": 1253, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Micron", + "HBM4", + "industry", + "leading", + "speed", + "11", + "Gbps", + "track", + "ramp", + "high", + "yields", + "calendar", + "quarter", + "2026", + "consistent", + "customers", + "product", + "plans", + "Our", + "advanced", + "CMOS", + "metallization", + "process", + "technologies", + "base", + "logic", + "die", + "DRAM", + "core", + "dies", + "designed", + "manufactured", + "house", + "This", + "unique", + "HBM", + "design", + "packaging", + "test", + "capability", + "enables", + "performance", + "low", + "power", + "leadership", + "pioneered", + "adoption", + "LP", + "data", + "center", + "Low", + "Power", + "server", + "modules", + "consume", + "DDR", + "Building", + "sampled", + "192GB", + "SOCAMM2", + "50", + "increase", + "capacity", + "module", + "rack", + "scale", + "density", + "50TB", + "details", + "summary", + "natural", + "image", + "Interior", + "view", + "modern", + "racks", + "illuminated", + "purple", + "blue", + "neon", + "lights", + "visible", + "text", + "symbols" + ], + "keywords": [], + "connect_to": [ + { + "target": "07c78111-5577-517c-b221-cb28a67f4d01", + "relation": "embeds", + "ref": "[images/image-8-Micron Data Center.jpg]", + "position": { + "start": 1039, + "end": 1078 + } + } + ] + } + }, + { + "chunk_id": "b531cc06-0764-5f75-885a-acb68553988a", + "type": "text", + "content": "- Our data center NAND portfolio revenue exceeded a billion dollars in fiscal Q1, and we are seeing strong momentum across our data center SSD portfolio, enabled by our leadership NAND technology.\n- In the performance SSD category, Micron has introduced the world's first PCIe Gen6 SSD, leveraging our G9 NAND.\n• We are seeing rapidly increasing qualification commitments for this product, including at hyperscalers.\n- In mainstream storage, our SSDs based on G9 NAND are already seeing robust demand in the first quarter of calendar 2026.\n- In capacity storage, our QLC-based 122 and 245TB G9 SSDs are entering qualification at multiple hyperscale customers.", + "path": "micron-530bd7ed.pdf/Financial results/Data center NAND", + "metadata": { + "length": 659, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Our", + "data", + "center", + "NAND", + "portfolio", + "revenue", + "exceeded", + "billion", + "dollars", + "fiscal", + "Q1", + "strong", + "momentum", + "SSD", + "enabled", + "leadership", + "technology", + "In", + "performance", + "category", + "Micron", + "introduced", + "world", + "PCIe", + "Gen6", + "leveraging", + "G9", + "We", + "rapidly", + "increasing", + "qualification", + "commitments", + "product", + "including", + "hyperscalers", + "mainstream", + "storage", + "SSDs", + "based", + "robust", + "demand", + "quarter", + "calendar", + "2026", + "capacity", + "QLC", + "122", + "245TB", + "entering", + "multiple", + "hyperscale", + "customers" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5936dde7-425d-5d66-929d-043189dc3148", + "type": "image", + "content": "\n## PC
    0\n[images/image-9-Laptop Keyboard and Screen.jpg]\n", + "path": "images/image-9-Laptop Keyboard and Screen.jpg", + "metadata": { + "length": 50, + "summary": "image-9\n- Close-up view of a laptop keyboard with illuminated keys in purple and pink hues.\n- The screen displays a colorful abstract wallpaper with wavy lines in blue, white, yellow, and pink.\n- Visible text on the keyboard includes standard QWERTY layout characters and function keys like \"Ctrl\", \"Page Down\", \"Enter\".\n- Bottom right corner shows the logo \"micron\" and page number \"12\".\n- The image appears to be from a presentation or promotional material showcasing technology products.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-9-Laptop Keyboard and Screen.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "3459eb0b-049f-52fd-88d6-9d51b7648594", + "type": "text", + "content": "- PC demand continues to be driven by Windows 10 end of life and AI PCs.\n• We forecast PC unit sales to grow high-single-digit percentage range in calendar 2025, above our prior expectations provided in our last earnings call of mid-single digits.\n- As we look ahead into 2026, we expect these demand drivers to continue, while memory supply constraints may affect some PC unit shipments.\n- Micron has completed multiple OEM qualifications of our 16Gb 1-gamma-based DDR5 and our G9-based PCIe Gen4 QLC SSDs.\n\n## PC
    0\n[images/image-9-Laptop Keyboard and Screen.jpg]\n\n
    \ntext_image\nmicron | 12\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/PC", + "metadata": { + "length": 621, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "PC", + "demand", + "continues", + "driven", + "Windows", + "10", + "end", + "life", + "AI", + "PCs", + "We", + "forecast", + "unit", + "sales", + "grow", + "high", + "single", + "digit", + "percentage", + "range", + "calendar", + "2025", + "prior", + "expectations", + "provided", + "earnings", + "call", + "mid", + "digits", + "As", + "ahead", + "2026", + "expect", + "drivers", + "continue", + "memory", + "supply", + "constraints", + "affect", + "shipments", + "Micron", + "completed", + "multiple", + "OEM", + "qualifications", + "16Gb", + "gamma", + "based", + "DDR5", + "G9", + "PCIe", + "Gen4", + "QLC", + "SSDs", + "details", + "summary", + "text", + "image", + "micron", + "12" + ], + "keywords": [], + "connect_to": [ + { + "target": "5936dde7-425d-5d66-929d-043189dc3148", + "relation": "embeds", + "ref": "[images/image-9-Laptop Keyboard and Screen.jpg]", + "position": { + "start": 526, + "end": 573 + } + } + ] + } + }, + { + "chunk_id": "54647d0d-8d16-54bb-95d5-7999ba06e14f", + "type": "image", + "content": "\n- AI is driving memory content growth. The shipment mix of flagship smartphones with 12GB of DRAM increased to $59\\%$ in calendar Q3, more than twice the level from a year ago.
    0\n[images/image-10-Concert Photography.jpg]\n", + "path": "images/image-10-Concert Photography.jpg", + "metadata": { + "length": 222, + "summary": "image-10\n- Primary subject: A pair of hands holding a smartphone horizontally to capture a live concert performance.\n- Scene details: The phone screen displays a stage illuminated by bright spotlights and laser beams in blue and red hues, with the silhouettes of a crowd visible below.\n- Notable features: The person holding the phone has dark nail polish. The background is blurred with pink lighting.\n- Visible text: \"micron\" logo and page number \"13\" at the bottom right corner.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-10-Concert Photography.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "7e06ce90-27f5-5ef3-bb3e-0bdff6a91e38", + "type": "text", + "content": "- Smartphone unit volumes in calendar 2025 are on track to grow in the low-single-digit percentage range.\n- AI is driving memory content growth. The shipment mix of flagship smartphones with 12GB of DRAM increased to $59\\%$ in calendar Q3, more than twice the level from a year ago.\n• Micron is accelerating innovation across our mobile DRAM portfolio.\n- In fiscal Q1, we began sampling our breakthrough 1-gamma 16Gb LPDDR6 product to leading OEM and ecosystem partners, marking a major milestone in next-generation memory technology.\n- LPDDR6 will power AI at the edge, delivering over 50% higher performance and improved power efficiency for flagship smartphones and AI PCs.\n- Micron also sampled our 1-gamma LP5x 24Gb product and began volume shipments of the previously announced 1-gamma LP5x 16Gb product to multiple OEMs.\n\n- AI is driving memory content growth. The shipment mix of flagship smartphones with 12GB of DRAM increased to $59\\%$ in calendar Q3, more than twice the level from a year ago.
    0\n[images/image-10-Concert Photography.jpg]\n\n
    \nnatural_image\nHand holding smartphone displaying a live music show with stage lights against a red and pink background (no text or symbols on screen)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Mobile", + "metadata": { + "length": 1240, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Smartphone", + "unit", + "volumes", + "calendar", + "2025", + "track", + "grow", + "low", + "single", + "digit", + "percentage", + "range", + "AI", + "driving", + "memory", + "content", + "growth", + "The", + "shipment", + "mix", + "flagship", + "smartphones", + "12GB", + "DRAM", + "increased", + "59", + "Q3", + "level", + "year", + "ago", + "Micron", + "accelerating", + "innovation", + "mobile", + "portfolio", + "In", + "fiscal", + "Q1", + "began", + "sampling", + "breakthrough", + "gamma", + "16Gb", + "LPDDR6", + "product", + "leading", + "OEM", + "ecosystem", + "partners", + "marking", + "major", + "milestone", + "generation", + "technology", + "power", + "edge", + "delivering", + "50%", + "higher", + "performance", + "improved", + "efficiency", + "PCs", + "sampled", + "LP5x", + "24Gb", + "volume", + "shipments", + "previously", + "announced", + "multiple", + "OEMs", + "details", + "summary", + "natural", + "image", + "Hand", + "holding", + "smartphone", + "displaying", + "live", + "music", + "show", + "stage", + "lights", + "red", + "pink", + "background", + "text", + "symbols", + "screen" + ], + "keywords": [], + "connect_to": [ + { + "target": "54647d0d-8d16-54bb-95d5-7999ba06e14f", + "relation": "embeds", + "ref": "[images/image-10-Concert Photography.jpg]", + "position": { + "start": 1017, + "end": 1058 + } + } + ] + } + }, + { + "chunk_id": "b7564c4d-e963-54f9-b31c-72df1111e666", + "type": "image", + "content": "\n- Micron is uniquely positioned for growth with our differentiated product portfolio and automotive market share leadership.
    0\n[images/image-11-Futuristic Car Interior.jpg]\n", + "path": "images/image-11-Futuristic Car Interior.jpg", + "metadata": { + "length": 170, + "summary": "image-11\n- Steering wheel with integrated controls and central emblem\n- Dashboard featuring four circular digital instrument clusters with blue glowing displays\n- Large vertical touchscreen infotainment system displaying vehicle schematics and interface icons\n- Center console with a secondary tablet device, gear selector knob, and control buttons\n- Ambient lighting accents in blue and red tones\n- \"micron\" logo visible at the bottom right corner alongside page number \"14\"", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-11-Futuristic Car Interior.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "a34ea9d5-ccaf-5ddc-af70-575e36a3f9cf", + "type": "image", + "content": "\n
    0\n[images/image-12-Business Meeting Discussion.jpg]\n", + "path": "images/image-12-Business Meeting Discussion.jpg", + "metadata": { + "length": 55, + "summary": "image-12\n- Scene: A group of four professionals gathered around a conference table in a glass-walled meeting room.\n- Key Action: A woman in a purple top is standing and pointing at documents on the table, appearing to lead the discussion or explain data.\n- Participants: Three other individuals (two women, one man) are seated and listening attentively.\n- Background: A large screen displays mathematical graphs or charts with curves plotted on axes.\n- Visible Text: \"December 17, 2025\" located at the bottom left corner.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-12-Business Meeting Discussion.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "69e91ae0-c7c5-508a-883a-948bebab2ba0", + "type": "text", + "content": "- In automotive, L2+ and L3 adoption is driving robust demand today, and our customers' roadmaps indicate a significantly higher memory content in fully automated vehicles.\n- Micron is uniquely positioned for growth with our differentiated product portfolio and automotive market share leadership.\n- Our ASIL-rated LPDDR5x and UFS4.1 NAND products, optimized for automotive and advanced robotics that include bandwidth-enhancing features, are seeing strong demand and have already secured billions of dollars in design wins.\n- In industrial, demand continues to strengthen, driven by the growing adoption of autonomous systems across various applications.\n- Long-term demand trend trajectory remains robust for memory and storage in industrial applications such as in factory automation, aerospace and defense, humanoid robotics, edge networking and video surveillance.\n- Across both auto and industrial markets, LPDDR4x and DDR4 are also experiencing strong demand, and we are making investments to provide long-term supply from our Manassas, Virginia fab.\n\n- Micron is uniquely positioned for growth with our differentiated product portfolio and automotive market share leadership.
    0\n[images/image-11-Futuristic Car Interior.jpg]\n\n
    \nnatural_image\nInterior view of a modern car dashboard with digital dashboard and smart lock display (no readable text or symbols)\n
    \n\n
    0\n[images/image-12-Business Meeting Discussion.jpg]\n\n
    \nnatural_image\nThree business professionals in a meeting setting with a large projection screen showing mathematical graphs (no readable text or symbols)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Automotive and embedded", + "metadata": { + "length": 1647, + "summary": "The passage discusses strong demand for memory and storage in automotive (L2+ and L3 adoption, ASIL-rated LPDDR5x and UFS4.1 NAND) and industrial markets (autonomous systems, factory automation, robotics). Micron highlights its differentiated product portfolio and market leadership, with billions in design wins and investments in long-term supply from its Virginia fab.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "In", + "automotive", + "L2", + "L3", + "adoption", + "driving", + "robust", + "demand", + "today", + "customers", + "roadmaps", + "significantly", + "higher", + "memory", + "content", + "fully", + "automated", + "vehicles", + "Micron", + "uniquely", + "positioned", + "growth", + "differentiated", + "product", + "portfolio", + "market", + "share", + "leadership", + "Our", + "ASIL", + "rated", + "LPDDR5x", + "UFS4.1", + "NAND", + "products", + "optimized", + "advanced", + "robotics", + "include", + "bandwidth", + "enhancing", + "features", + "strong", + "secured", + "billions", + "dollars", + "design", + "wins", + "industrial", + "continues", + "strengthen", + "driven", + "growing", + "autonomous", + "systems", + "applications", + "Long", + "term", + "trend", + "trajectory", + "remains", + "storage", + "factory", + "automation", + "aerospace", + "defense", + "humanoid", + "edge", + "networking", + "video", + "surveillance", + "Across", + "auto", + "markets", + "LPDDR4x", + "DDR4", + "experiencing", + "making", + "investments", + "provide", + "long", + "supply", + "Manassas", + "Virginia", + "fab", + "details", + "summary", + "natural", + "image", + "Interior", + "view", + "modern", + "car", + "dashboard", + "digital", + "smart", + "lock", + "display", + "readable", + "text", + "symbols", + "Three", + "business", + "professionals", + "meeting", + "setting", + "large", + "projection", + "screen", + "showing", + "mathematical", + "graphs" + ], + "keywords": [ + "Automotive", + "Industrial", + "Memory" + ], + "connect_to": [ + { + "target": "b7564c4d-e963-54f9-b31c-72df1111e666", + "relation": "embeds", + "ref": "[images/image-11-Futuristic Car Interior.jpg]", + "position": { + "start": 1195, + "end": 1240 + } + }, + { + "target": "a34ea9d5-ccaf-5ddc-af70-575e36a3f9cf", + "relation": "embeds", + "ref": "[images/image-12-Business Meeting Discussion.jpg]", + "position": { + "start": 1434, + "end": 1483 + } + } + ] + } + }, + { + "chunk_id": "b457542c-a5ba-53c6-a9db-7f5fb6ec9dd5", + "type": "image", + "content": "\n• Over the last few months, our customers' AI data center build-out plans have driven a sharp increase in demand forecasts for memory and storage.
    0\n[images/image-13-Silicon Wafer in Manufacturing.jpg]\n", + "path": "images/image-13-Silicon Wafer in Manufacturing.jpg", + "metadata": { + "length": 192, + "summary": "image-13\n- Primary Subject: A close-up photograph of a silicon wafer held within an automated manufacturing or inspection machine.\n- Visual Details: The circular wafer displays a distinct rainbow iridescence caused by thin-film interference, overlaid with a precise grid pattern representing individual microchips (dies).\n- Context: The background is blurred, emphasizing the high-tech nature of semiconductor fabrication.\n- Visible Text: \"December 17, 2025\" located at the bottom left corner.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-13-Silicon Wafer in Manufacturing.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "5c41e7aa-803e-5bee-89d5-004d2e5d6b0e", + "type": "text", + "content": "• Over the last few months, our customers' AI data center build-out plans have driven a sharp increase in demand forecasts for memory and storage.\n• We believe that the aggregate industry supply will remain substantially short of the demand for the foreseeable future.\n- The dramatic increase in HBM demand is further challenging the supply environment due to the 3-to-1 trade ratio with DDR5, and this trade ratio only increases with future generations of HBM.\n- Additional cleanroom space is necessary to address this increased demand, and lead times for cleanroom build-out are lengthening across geographies.\n• Together, these demand and supply factors are driving tight industry conditions across DRAM and NAND, and we expect tightness to persist through and beyond calendar 2026.\n\n• Over the last few months, our customers' AI data center build-out plans have driven a sharp increase in demand forecasts for memory and storage.
    0\n[images/image-13-Silicon Wafer in Manufacturing.jpg]\n\n
    \nnatural_image\nClose-up of a semiconductor wafer with colorful grid patterns, captured in a lab setting (no text or symbols on the wafer itself)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Market outlook (1 of 2)", + "metadata": { + "length": 1162, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Over", + "months", + "customers", + "AI", + "data", + "center", + "build", + "plans", + "driven", + "sharp", + "increase", + "demand", + "forecasts", + "memory", + "storage", + "We", + "aggregate", + "industry", + "supply", + "remain", + "substantially", + "short", + "foreseeable", + "future", + "The", + "dramatic", + "HBM", + "challenging", + "environment", + "due", + "trade", + "ratio", + "DDR5", + "increases", + "generations", + "Additional", + "cleanroom", + "space", + "address", + "increased", + "lead", + "times", + "lengthening", + "geographies", + "Together", + "factors", + "driving", + "tight", + "conditions", + "DRAM", + "NAND", + "expect", + "tightness", + "persist", + "calendar", + "2026", + "details", + "summary", + "natural", + "image", + "Close", + "semiconductor", + "wafer", + "colorful", + "grid", + "patterns", + "captured", + "lab", + "setting", + "text", + "symbols" + ], + "keywords": [], + "connect_to": [ + { + "target": "b457542c-a5ba-53c6-a9db-7f5fb6ec9dd5", + "relation": "embeds", + "ref": "[images/image-13-Silicon Wafer in Manufacturing.jpg]", + "position": { + "start": 945, + "end": 997 + } + } + ] + } + }, + { + "chunk_id": "de721a47-f8e6-508c-a448-0ddf63d7ed78", + "type": "image", + "content": "\n- We expect 2025 NAND bit demand growth to be in the high-teens percentage range versus low to mid-teens previously.
    0\n[images/image-14-Semiconductor Cleanroom.jpg]\n", + "path": "images/image-14-Semiconductor Cleanroom.jpg", + "metadata": { + "length": 162, + "summary": "image-14\n- Environment: High-tech manufacturing facility (cleanroom) with rows of large, stainless steel industrial machinery.\n- Personnel: Worker wearing a full-body white protective suit (bunny suit), face mask, and hood walking down the aisle.\n- Lighting: Distinctive yellow lighting in the background, typical for photoresist processing areas in semiconductor fabrication.\n- Infrastructure: Raised perforated flooring for air circulation; overhead cable trays and utility lines.\n- Date Stamp: December 17, 2025", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-14-Semiconductor Cleanroom.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "6c929fbd-b9e3-5665-bd3b-33715a547c46", + "type": "text", + "content": "- Calendar 2025 DRAM and NAND industry bit demand growth expectations are higher than in our last earnings call outlook.\n- We now expect calendar 2025 DRAM bit demand growth in the low $20\\%$ range versus high teens previously.\n- We expect 2025 NAND bit demand growth to be in the high-teens percentage range versus low to mid-teens previously.\n- We expect calendar 2026 industry DRAM and NAND bit shipment growth to be constrained by industry supply.\n- We expect both DRAM and NAND calendar 2026 industry bit shipments to increase around 20% from 2025 levels.\n- Micron is working hard to support our customers' demand during this time, and we expect to grow our DRAM and NAND bit shipments approximately 20% in calendar 2026.\n- Despite significant efforts, we are disappointed to be unable to meet demand from our customers, across all market segments.\n\n- We expect 2025 NAND bit demand growth to be in the high-teens percentage range versus low to mid-teens previously.
    0\n[images/image-14-Semiconductor Cleanroom.jpg]\n\n
    \nnatural_image\nInterior of a high-tech semiconductor fabrication facility with workers in full cleanroom suits walking along a central aisle (no visible text or symbols)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Market outlook (2 of 2)", + "metadata": { + "length": 1225, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Calendar", + "2025", + "DRAM", + "NAND", + "industry", + "bit", + "demand", + "growth", + "expectations", + "higher", + "earnings", + "call", + "outlook", + "We", + "expect", + "calendar", + "low", + "20", + "range", + "versus", + "high", + "teens", + "previously", + "percentage", + "mid", + "2026", + "shipment", + "constrained", + "supply", + "shipments", + "increase", + "20%", + "levels", + "Micron", + "working", + "hard", + "support", + "customers", + "time", + "grow", + "approximately", + "Despite", + "significant", + "efforts", + "disappointed", + "unable", + "meet", + "market", + "segments", + "details", + "summary", + "natural", + "image", + "Interior", + "tech", + "semiconductor", + "fabrication", + "facility", + "workers", + "full", + "cleanroom", + "suits", + "walking", + "central", + "aisle", + "visible", + "text", + "symbols" + ], + "keywords": [], + "connect_to": [ + { + "target": "de721a47-f8e6-508c-a448-0ddf63d7ed78", + "relation": "embeds", + "ref": "[images/image-14-Semiconductor Cleanroom.jpg]", + "position": { + "start": 983, + "end": 1028 + } + } + ] + } + }, + { + "chunk_id": "aea91de8-8634-5878-993d-42ba3450bca5", + "type": "image", + "content": "\n- Micron plans to increase our fiscal 2026 capex to approximately \\$20 billion, versus our prior estimate of \\$18 billion. This increase will primarily support our HBM supply capability, and also our 1-gamma supply, in calendar 2026.
    0\n[images/image-15-Laboratory Safety Work.jpg]\n", + "path": "images/image-15-Laboratory Safety Work.jpg", + "metadata": { + "length": 279, + "summary": "image-15\n- Subject: Scientist or technician wearing full PPE (face shield, safety glasses, protective gown, orange chemical-resistant gloves) working inside a fume hood or glove box.\n- Environment: Blue-lit laboratory setting with visible ventilation piping on the left wall.\n- Text/Signage: \"SCRUBBER EXHAUST\" label with directional arrow on vertical pipe; date stamp at bottom reads \"December 17, 2025\".\n- Equipment: Fume hood/glove box with sliding sash and internal lighting; overhead exhaust fixture visible above work area.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-15-Laboratory Safety Work.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d6c73d81-de26-5433-9e39-27f27835ee6e", + "type": "text", + "content": "- Micron plans to increase our fiscal 2026 capex to approximately \\$20 billion, versus our prior estimate of \\$18 billion. This increase will primarily support our HBM supply capability, and also our 1-gamma supply, in calendar 2026.\n- We are pulling in equipment orders and accelerating installation timelines to maximize output capability. Micron is also investing across our global manufacturing footprint to add supply to support longer-term demand. We are seeing an enthusiastic customer response to our planned U.S. supply.\n- We are pulling in our first Idaho fab timeline, and we now expect first wafer output in mid-calendar 2027, earlier than our prior expectations of second-half calendar 2027. Earlier this year, we announced our plans for a second Idaho fab, which will begin construction in 2026 and be operational by the end of 2028.\n- We are making good progress on securing necessary permits for our New York site and appreciate the partnership with the state of New York and the Trump administration.\n- We plan to break ground on our first New York fab in early calendar 2026, which we expect will provide supply in 2030 and beyond.\n\n- Micron plans to increase our fiscal 2026 capex to approximately \\$20 billion, versus our prior estimate of \\$18 billion. This increase will primarily support our HBM supply capability, and also our 1-gamma supply, in calendar 2026.
    0\n[images/image-15-Laboratory Safety Work.jpg]\n\n
    \ntext_image\nDecember 17, 2025\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Supply efforts (1 of 2)", + "metadata": { + "length": 1498, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Micron", + "plans", + "increase", + "fiscal", + "2026", + "capex", + "approximately", + "20", + "billion", + "versus", + "prior", + "estimate", + "18", + "This", + "primarily", + "support", + "HBM", + "supply", + "capability", + "gamma", + "calendar", + "We", + "pulling", + "equipment", + "orders", + "accelerating", + "installation", + "timelines", + "maximize", + "output", + "investing", + "global", + "manufacturing", + "footprint", + "add", + "longer", + "term", + "demand", + "enthusiastic", + "customer", + "response", + "planned", + "Idaho", + "fab", + "timeline", + "expect", + "wafer", + "mid", + "2027", + "earlier", + "expectations", + "half", + "Earlier", + "year", + "announced", + "begin", + "construction", + "operational", + "end", + "2028", + "making", + "good", + "progress", + "securing", + "permits", + "New", + "York", + "site", + "partnership", + "state", + "Trump", + "administration", + "plan", + "break", + "ground", + "early", + "provide", + "2030", + "details", + "summary", + "text", + "image", + "December", + "17", + "2025" + ], + "keywords": [], + "connect_to": [ + { + "target": "aea91de8-8634-5878-993d-42ba3450bca5", + "relation": "embeds", + "ref": "[images/image-15-Laboratory Safety Work.jpg]", + "position": { + "start": 1396, + "end": 1440 + } + } + ] + } + }, + { + "chunk_id": "f2df1635-69e5-539a-bf75-1e9e7c001a06", + "type": "text", + "content": "- In Japan, with the support of METI (Japan’s Ministry of Economy, Trade and Industry), we are making technology and manufacturing investments.\n• We are enabling future DRAM technology transitions in coordination with our Boise R&D team.\n- We are also adding cleanroom space in our Hiroshima fab to support these advanced nodes, which will increase production scale and optimize fab economics.\n- In Singapore, our HBM advanced packaging facility is on track to contribute meaningfully to our HBM supply in calendar 2027. As HBM becomes a part of our Singapore manufacturing footprint, we expect opportunities for synergies between NAND and DRAM production.\n• We are pleased with the progress on our assembly and test facility in India, which has initiated pilot production and will ramp in 2026.\n- As we make progress on our strategic manufacturing initiatives, we will continue to be responsive to the market environment and disciplined with our capex plans", + "path": "micron-530bd7ed.pdf/Financial results/Supply efforts (2 of 2)", + "metadata": { + "length": 958, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "In", + "Japan", + "support", + "METI", + "Ministry", + "Economy", + "Trade", + "Industry", + "making", + "technology", + "manufacturing", + "investments", + "We", + "enabling", + "future", + "DRAM", + "transitions", + "coordination", + "Boise", + "team", + "adding", + "cleanroom", + "space", + "Hiroshima", + "fab", + "advanced", + "nodes", + "increase", + "production", + "scale", + "optimize", + "economics", + "Singapore", + "HBM", + "packaging", + "facility", + "track", + "contribute", + "meaningfully", + "supply", + "calendar", + "2027", + "As", + "part", + "footprint", + "expect", + "opportunities", + "synergies", + "NAND", + "pleased", + "progress", + "assembly", + "test", + "India", + "initiated", + "pilot", + "ramp", + "2026", + "make", + "strategic", + "initiatives", + "continue", + "responsive", + "market", + "environment", + "disciplined", + "capex", + "plans" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "f5620e25-f721-5ce0-acd1-9764cabacabe", + "type": "image", + "content": "\n- In Singapore, our HBM advanced packaging facility is on track to contribute meaningfully to our HBM supply in calendar 2027. As HBM becomes a part of our Singapore manufacturing footprint, we expect opportunities for synergies between NAND and DRAM production.
    0\n[images/image-16-Corporate Executive Portrait.jpg]\n", + "path": "images/image-16-Corporate Executive Portrait.jpg", + "metadata": { + "length": 308, + "summary": "image-16\nSubject: Male executive with short brown hair, smiling at the camera.\nAttire: Grey suit jacket over a light blue patterned button-down shirt.\nBackground: Solid black studio background.\nVisible Text/Logo: \"micron\" logo visible in the bottom right corner.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-16-Corporate Executive Portrait.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "81aa3508-5c73-514a-b8d8-f4cbf6d2f054", + "type": "text", + "content": "Chief Financial Officer\n\n- In Singapore, our HBM advanced packaging facility is on track to contribute meaningfully to our HBM supply in calendar 2027. As HBM becomes a part of our Singapore manufacturing footprint, we expect opportunities for synergies between NAND and DRAM production.
    0\n[images/image-16-Corporate Executive Portrait.jpg]\n\n
    \nnatural_image\nPortrait of a man in a gray suit against a black background, no visible text or symbols on the subject or background.\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Mark Murphy", + "metadata": { + "length": 504, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Chief", + "Financial", + "Officer", + "In", + "Singapore", + "HBM", + "advanced", + "packaging", + "facility", + "track", + "contribute", + "meaningfully", + "supply", + "calendar", + "2027", + "As", + "part", + "manufacturing", + "footprint", + "expect", + "opportunities", + "synergies", + "NAND", + "DRAM", + "production", + "details", + "summary", + "natural", + "image", + "Portrait", + "man", + "gray", + "suit", + "black", + "background", + "visible", + "text", + "symbols", + "subject" + ], + "keywords": [], + "connect_to": [ + { + "target": "f5620e25-f721-5ce0-acd1-9764cabacabe", + "relation": "embeds", + "ref": "[images/image-16-Corporate Executive Portrait.jpg]", + "position": { + "start": 299, + "end": 349 + } + } + ] + } + }, + { + "chunk_id": "9be41014-954a-506c-a35c-5b4acd1c4f11", + "type": "text", + "content": "\\$13.6B\nRevenue up 21% Q/Q\nand up 57% Y/Y", + "path": "micron-530bd7ed.pdf/Financial results/FQ1-26 revenue \\$13.6B", + "metadata": { + "length": 41, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "13.6", + "Revenue", + "21%", + "57%" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e4f539ba-8e6d-50a6-9e56-f842973248ca", + "type": "text", + "content": "", + "path": "micron-530bd7ed.pdf/Financial results/Performance by technology", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "c668840a-e766-53f0-aed3-8dbd70dd8916", + "type": "text", + "content": "- \\$10.8 billion, representing 79% of total revenue in FQ1-26\n• Revenue increased 69% Y/Y\n• Revenue increased 20% Q/Q\n- Bit shipments up slightly Q/Q\n• ASPs increased approximately 20% Q/Q", + "path": "micron-530bd7ed.pdf/Financial results/Performance by technology/DRAM FQ1-26", + "metadata": { + "length": 188, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "10.8", + "billion", + "representing", + "79%", + "total", + "revenue", + "FQ1", + "26", + "Revenue", + "increased", + "69%", + "20%", + "Bit", + "shipments", + "slightly", + "ASPs", + "approximately" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e20d166a-d3f0-5df0-bf3b-4c404e0f3541", + "type": "table", + "content": "
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    Cloud Memory (CMBU)
    Revenue$5,284$4,543$2,648
    Gross margin66 %59 %51 %
    Operating margin55 %48 %40 %
    Core Data (CDBU)
    Revenue$2,379$1,577$2,292
    Gross margin51 %41 %50 %
    Operating margin37 %25 %38 %
    Mobile and Client (MCBU)
    Revenue$4,255$3,760$2,608
    Gross margin54 %36 %27 %
    Operating margin47 %29 %15 %
    Automotive and Embedded (AEBU)
    Revenue$1,720$1,434$1,158
    Gross margin45 %31 %20 %
    Operating margin36 %20 %7 %
    ", + "path": "tables/table-0 Segment Financials.html", + "metadata": { + "length": 1213, + "summary": "table-1\nThe table shows revenue, gross margin, and operating margin for Cloud Memory, Core Data, Mobile and Client, and Automotive and Embedded segments across FQ1-26, FQ4-25, and FQ1-25.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-0 Segment Financials.html", + "keywords": [ + "Revenue", + "Gross Margin", + "Operating Margin" + ], + "tokens": [] + } + }, + { + "chunk_id": "05963bf0-88ec-5311-8331-26497cc804fb", + "type": "image", + "content": "\n• Revenue increased 22% Q/Q
    0\n[images/image-17-Abstract Digital Art.jpg]\n", + "path": "images/image-17-Abstract Digital Art.jpg", + "metadata": { + "length": 73, + "summary": "image-17\nDecember 17, 2025", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-17-Abstract Digital Art.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "0845eeef-cd20-545f-98e6-d8322e8732e4", + "type": "text", + "content": "- \\$2.7 billion, representing 20% of total revenue in FQ1-26\n• Revenue increased 22% Y/Y\n• Revenue increased 22% Q/Q\n- Bit shipments increased in mid-to-high single digit percentage range Q/Q\n- ASPs increased in the mid-teens percentage range Q/Q\nQuarterly business unit financial results\n\n[tables/table-0 Segment Financials.html]\n\n\n• Revenue increased 22% Q/Q
    0\n[images/image-17-Abstract Digital Art.jpg]\n\n
    \nnatural_image\nAbstract digital artwork with glowing dots and bokeh background, no readable text or symbols\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Performance by technology/NAND FQ1-26", + "metadata": { + "length": 574, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "2.7", + "billion", + "representing", + "20%", + "total", + "revenue", + "FQ1", + "26", + "Revenue", + "increased", + "22%", + "Bit", + "shipments", + "mid", + "high", + "single", + "digit", + "percentage", + "range", + "ASPs", + "teens", + "Quarterly", + "business", + "unit", + "financial", + "results", + "details", + "summary", + "natural", + "image", + "Abstract", + "digital", + "artwork", + "glowing", + "dots", + "bokeh", + "background", + "readable", + "text", + "symbols" + ], + "keywords": [], + "connect_to": [ + { + "target": "e20d166a-d3f0-5df0-bf3b-4c404e0f3541", + "relation": "embeds", + "ref": "[tables/table-0 Segment Financials.html]", + "position": { + "start": 290, + "end": 330 + } + }, + { + "target": "05963bf0-88ec-5311-8331-26497cc804fb", + "relation": "embeds", + "ref": "[images/image-17-Abstract Digital Art.jpg]", + "position": { + "start": 372, + "end": 414 + } + } + ] + } + }, + { + "chunk_id": "2e82e176-5d91-5a0f-944d-b408f4df9ffa", + "type": "image", + "content": "\n- CMBU revenue was up 16% sequentially, driven by an increase in bit shipments and higher prices
    0\n[images/image-18-Abstract Digital Art.jpg]\n", + "path": "images/image-18-Abstract Digital Art.jpg", + "metadata": { + "length": 142, + "summary": "image-18\nDecember 17, 2025", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-18-Abstract Digital Art.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "474ec016-4f48-56f6-9f6f-5cf117a6e0b8", + "type": "text", + "content": "- Cloud Memory Business Unit revenue was a record \\$5.3 billion and represented 39% of total company revenue.\n- CMBU revenue was up 16% sequentially, driven by an increase in bit shipments and higher prices\n- CMBU gross margins were 66%, higher by 620 basis points sequentially, supported by cost execution and higher pricing.\n- Core Data Center Business Unit revenue was a record \\$2.4 billion and represented 17% of total company revenue.\n- CDBU revenue was up 51% sequentially, driven by robust bit shipments and higher pricing.\n- CDBU gross margins were 51%, up 990 basis points sequentially, supported by higher pricing and cost execution.\n\n- CMBU revenue was up 16% sequentially, driven by an increase in bit shipments and higher prices
    0\n[images/image-18-Abstract Digital Art.jpg]\n\n
    \nnatural_image\nAbstract digital network visualization with glowing nodes and green-to-blue bokeh background (no text or symbols)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/Financial performance by business unit (1 of 2)", + "metadata": { + "length": 955, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Cloud", + "Memory", + "Business", + "Unit", + "revenue", + "record", + "5.3", + "billion", + "represented", + "39%", + "total", + "company", + "CMBU", + "16%", + "sequentially", + "driven", + "increase", + "bit", + "shipments", + "higher", + "prices", + "gross", + "margins", + "66%", + "620", + "basis", + "points", + "supported", + "cost", + "execution", + "pricing", + "Core", + "Data", + "Center", + "2.4", + "17%", + "CDBU", + "51%", + "robust", + "990", + "details", + "summary", + "natural", + "image", + "Abstract", + "digital", + "network", + "visualization", + "glowing", + "nodes", + "green", + "blue", + "bokeh", + "background", + "text", + "symbols" + ], + "keywords": [], + "connect_to": [ + { + "target": "2e82e176-5d91-5a0f-944d-b408f4df9ffa", + "relation": "embeds", + "ref": "[images/image-18-Abstract Digital Art.jpg]", + "position": { + "start": 754, + "end": 796 + } + } + ] + } + }, + { + "chunk_id": "87d6e692-358d-503c-b347-084513e0f6e2", + "type": "text", + "content": "- Mobile Client Business Unit revenue was a record \\$4.3 billion and represented 31% of total company revenue.\n- MCBU revenue was up 13% sequentially, driven by higher pricing, partly offset by lower bit shipments.\n- MCBU gross margins were 54%, up 17 percentage points sequentially, driven primarily by higher pricing.\n- Automotive and Embedded Business Unit revenue was a record \\$1.7 billion and represented 13% of total company revenue.\n- AEBU revenue was up 20% sequentially, driven by higher bit shipments and higher pricing.\n- AEBU gross margins were 45%, up 14 percentage points sequentially, driven primarily by higher pricing.", + "path": "micron-530bd7ed.pdf/Financial results/Financial performance by business unit (2 of 2)", + "metadata": { + "length": 636, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Mobile", + "Client", + "Business", + "Unit", + "revenue", + "record", + "4.3", + "billion", + "represented", + "31%", + "total", + "company", + "MCBU", + "13%", + "sequentially", + "driven", + "higher", + "pricing", + "partly", + "offset", + "lower", + "bit", + "shipments", + "gross", + "margins", + "54%", + "17", + "percentage", + "points", + "primarily", + "Automotive", + "Embedded", + "1.7", + "AEBU", + "20%", + "45%", + "14" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "5dbd386a-3eb0-51b2-b7af-4ae287f27273", + "type": "image", + "content": "\nOperating income: \\$6.42 billion
    0\n[images/image-19-Micron Presentation Slide.jpg]\n", + "path": "images/image-19-Micron Presentation Slide.jpg", + "metadata": { + "length": 78, + "summary": "image-19\n- Brand: Micron\n- Page Number: 25\n- Visual Content: Abstract digital artwork featuring flowing, translucent waves with glowing blue and purple light trails against a dark background. The image includes bokeh effects and particle-like sparkles, suggesting themes of data flow, technology, or connectivity.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-19-Micron Presentation Slide.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "6aa9d2d6-cddd-5f4b-8b06-1ab9f9c3a2d5", + "type": "text", + "content": "Non-GAAP operating results\nRevenue: \\$13.64 billion\nGross margin: 56.8%\nOperating expenses: \\$1.33 billion\nOperating income: \\$6.42 billion\nNet income: \\$5.48 billion\nDiluted earnings per share: \\$4.78\nCash from operations (GAAP): \\$8.41 billion\nSee non-GAAP reconciliations in Appendix\n\nOperating income: \\$6.42 billion
    0\n[images/image-19-Micron Presentation Slide.jpg]\n\n
    \nnatural_image\nAbstract digital artwork featuring glowing purple and blue water droplets with bokeh bokeh effects (no text or symbols)\n
    ", + "path": "micron-530bd7ed.pdf/Financial results/FQ1-26 Non-GAAP operating results", + "metadata": { + "length": 539, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Non", + "GAAP", + "operating", + "results", + "Revenue", + "13.64", + "billion", + "Gross", + "margin", + "56.8%", + "Operating", + "expenses", + "1.33", + "income", + "6.42", + "Net", + "5.48", + "Diluted", + "earnings", + "share", + "4.78", + "Cash", + "operations", + "8.41", + "See", + "reconciliations", + "Appendix", + "details", + "summary", + "natural", + "image", + "Abstract", + "digital", + "artwork", + "featuring", + "glowing", + "purple", + "blue", + "water", + "droplets", + "bokeh", + "effects", + "text", + "symbols" + ], + "keywords": [], + "connect_to": [ + { + "target": "5dbd386a-3eb0-51b2-b7af-4ae287f27273", + "relation": "embeds", + "ref": "[images/image-19-Micron Presentation Slide.jpg]", + "position": { + "start": 332, + "end": 379 + } + } + ] + } + }, + { + "chunk_id": "9ccb6a52-37b8-531d-8072-3946982e2313", + "type": "table", + "content": "
    Cash flow from operationsFQ1-26: $8.4 billion (62% of revenue)
    Net Capex1FQ1-26: $4.5 billion
    Adjusted free cash flow*FQ1-26: $3.9 billion
    BuybacksFQ1-26: $300 million
    DividendsDividend of $0.115 per share will be paid on January 14th
    Liquidity2$15.5 billion in liquidity at end of FQ1-26
    ", + "path": "tables/table-1 FQ1-26 Cash Flow.html", + "metadata": { + "length": 460, + "summary": "table-2\nCash flow from operations $8.4B, net capex $4.5B, adjusted free cash flow $3.9B, buybacks $300M, dividend $0.115/share, liquidity $15.5B.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-1 FQ1-26 Cash Flow.html", + "keywords": [ + "Cash flow", + "Capex", + "Free cash flow" + ], + "tokens": [] + } + }, + { + "chunk_id": "7ab9398b-5650-5065-ba4e-72f1327148cf", + "type": "text", + "content": "From FY-23 to FQ1-26\n- \\$1.0 billion toward repurchasing 13 million shares\n- \\$1.7 billion towards dividends paid\n- \\$2.7 billion returned to shareholders from share repurchases and dividends\n\n[tables/table-1 FQ1-26 Cash Flow.html]\n\n$^{1}$ Capex net of proceeds from government incentives and proceeds from sales of property, plant, and equipment.\n$^{2}$ Cash, short-term and long-term marketable investments, restricted cash, and undrawn revolver capacity.\n\\*Adjusted free cash flow is a non-GAAP measure defined as net cash provided by operating activities less investments in capital expenditures net of proceeds from government incentives and proceeds from sales of property, plant, and equipment.\nSee non-GAAP reconciliations in Appendix.", + "path": "micron-530bd7ed.pdf/Financial results/Cash flow and capital allocation From FY-23 to FQ1-26", + "metadata": { + "length": 774, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "From", + "FY", + "23", + "FQ1", + "26", + "1.0", + "billion", + "repurchasing", + "13", + "million", + "shares", + "1.7", + "dividends", + "paid", + "2.7", + "returned", + "shareholders", + "share", + "repurchases", + "Capex", + "net", + "proceeds", + "government", + "incentives", + "sales", + "property", + "plant", + "equipment", + "Cash", + "short", + "term", + "long", + "marketable", + "investments", + "restricted", + "cash", + "undrawn", + "revolver", + "capacity", + "Adjusted", + "free", + "flow", + "GAAP", + "measure", + "defined", + "provided", + "operating", + "activities", + "capital", + "expenditures", + "See", + "reconciliations", + "Appendix" + ], + "keywords": [], + "connect_to": [ + { + "target": "9ccb6a52-37b8-531d-8072-3946982e2313", + "relation": "embeds", + "ref": "[tables/table-1 FQ1-26 Cash Flow.html]", + "position": { + "start": 193, + "end": 231 + } + } + ] + } + }, + { + "chunk_id": "d08f238e-52ff-59ea-a3f5-f1102821c504", + "type": "text", + "content": "• Over 80% of our professional workforce actively uses GenAI, with total usage up tenfold since last year.\n- In manufacturing, integrating AI into yield and quality management has cut root cause identification time by half in cases.\n- Our coding teams are realizing productivity gains of 30% or more using agentic AI.\n- In R&D, GenAI is accelerating development by reducing cycle times in design verification, product validation, issue triage and root cause analysis.\n- Across business functions, GenAI is broadening automation opportunities, and we are deploying conversational analytics to accelerate and improve decision-making.\n- We expect Micron’s use of AI across the enterprise to further strengthen our competitiveness in the coming years.", + "path": "micron-530bd7ed.pdf/Financial results/Al use at Micron", + "metadata": { + "length": 747, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Over", + "80%", + "professional", + "workforce", + "actively", + "GenAI", + "total", + "usage", + "tenfold", + "year", + "In", + "manufacturing", + "integrating", + "AI", + "yield", + "quality", + "management", + "cut", + "root", + "identification", + "time", + "half", + "cases", + "Our", + "coding", + "teams", + "realizing", + "productivity", + "gains", + "30%", + "agentic", + "accelerating", + "development", + "reducing", + "cycle", + "times", + "design", + "verification", + "product", + "validation", + "issue", + "triage", + "analysis", + "Across", + "business", + "functions", + "broadening", + "automation", + "opportunities", + "deploying", + "conversational", + "analytics", + "accelerate", + "improve", + "decision", + "making", + "We", + "expect", + "Micron", + "enterprise", + "strengthen", + "competitiveness", + "coming", + "years" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "e345c4ed-c94d-5cc2-82be-a62ffee8fca3", + "type": "image", + "content": "\n• We expect a fiscal Q2 and fiscal year 2026 tax rate of around 15.5%.
    0\n[images/image-20-Silicon Wafer Close-up.jpg]\n", + "path": "images/image-20-Silicon Wafer Close-up.jpg", + "metadata": { + "length": 116, + "summary": "image-20\n- Subject: Circular silicon wafer with visible grid pattern of microchips\n- Lighting: Blue-toned illumination with bright reflection on the right edge\n- Date stamp: December 17, 2025 (bottom left)", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "images/image-20-Silicon Wafer Close-up.jpg", + "keywords": [], + "tokens": [] + } + }, + { + "chunk_id": "d0ee05ab-b3a5-5127-9831-839cce3f23b3", + "type": "table", + "content": "
    Revenue$18.70 billion ± $400 million
    Gross margin68.0% ± 1.0%
    Operating expenses$1.38 billion ± $20 million
    Diluted earnings per share*$8.42 ± $0.20
    ", + "path": "tables/table-2 Financial Data.html", + "metadata": { + "length": 268, + "summary": "table-3\nRevenue $18.70B ± $400M, gross margin 68.0% ± 1.0%, operating expenses $1.38B ± $20M, diluted EPS $8.42 ± $0.20.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-2 Financial Data.html", + "keywords": [ + "Revenue", + "Gross margin", + "EPS" + ], + "tokens": [] + } + }, + { + "chunk_id": "63263aeb-3d68-5759-b9e9-72b9c299740b", + "type": "text", + "content": "- We expect higher price, lower cost and favorable mix to all contribute to gross margin expansion in Q2.\n- Operating expenses for fiscal Q2 are projected to be approximately \\$1.38 billion.\n- As mentioned last quarter, Micron’s fiscal Q4 2026 opex will also reflect the effect of an additional workweek in this 53-week fiscal year.\n• We expect a fiscal Q2 and fiscal year 2026 tax rate of around 15.5%.\n- Micron is investing in a disciplined manner across our global manufacturing footprint to better meet demand.\n- To address tight supply/demand conditions extending beyond 2026, we now project our capital spending in fiscal 2026 to be approximately \\$20 billion, weighted to the second half of the fiscal year.\n- We expect free cash flow to strengthen in fiscal Q2, and we expect to generate significantly higher free cash flow year over year in fiscal 2026.\n- Any impacts that may occur due to potential new tariffs are not included in our guidance.\n\n• We expect a fiscal Q2 and fiscal year 2026 tax rate of around 15.5%.
    0\n[images/image-20-Silicon Wafer Close-up.jpg]\n\n
    \nnatural_image\nClose-up of a semiconductor wafer with visible green and blue grating pattern against dark background (no text or symbols)\n
    \nFQ2-26 guidance\nNon-GAAP\n\n[tables/table-2 Financial Data.html]\n\n\\*Based on \\~1.15 billion diluted shares.\nSee non-GAAP reconciliations in Appendix.", + "path": "micron-530bd7ed.pdf/Outlook", + "metadata": { + "length": 1413, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "We", + "expect", + "higher", + "price", + "lower", + "cost", + "favorable", + "mix", + "contribute", + "gross", + "margin", + "expansion", + "Q2", + "Operating", + "expenses", + "fiscal", + "projected", + "approximately", + "1.38", + "billion", + "As", + "mentioned", + "quarter", + "Micron", + "Q4", + "2026", + "opex", + "reflect", + "effect", + "additional", + "workweek", + "53", + "week", + "year", + "tax", + "rate", + "15.5%", + "investing", + "disciplined", + "manner", + "global", + "manufacturing", + "footprint", + "meet", + "demand", + "To", + "address", + "tight", + "supply", + "conditions", + "extending", + "project", + "capital", + "spending", + "20", + "weighted", + "half", + "free", + "cash", + "flow", + "strengthen", + "generate", + "significantly", + "Any", + "impacts", + "occur", + "due", + "potential", + "tariffs", + "included", + "guidance", + "details", + "summary", + "natural", + "image", + "Close", + "semiconductor", + "wafer", + "visible", + "green", + "blue", + "grating", + "pattern", + "dark", + "background", + "text", + "symbols", + "FQ2", + "26", + "Non", + "GAAP", + "Based", + "1.15", + "diluted", + "shares", + "See", + "reconciliations", + "Appendix" + ], + "keywords": [], + "connect_to": [ + { + "target": "e345c4ed-c94d-5cc2-82be-a62ffee8fca3", + "relation": "embeds", + "ref": "[images/image-20-Silicon Wafer Close-up.jpg]", + "position": { + "start": 1038, + "end": 1082 + } + }, + { + "target": "d0ee05ab-b3a5-5127-9831-839cce3f23b3", + "relation": "embeds", + "ref": "[tables/table-2 Financial Data.html]", + "position": { + "start": 1287, + "end": 1323 + } + } + ] + } + }, + { + "chunk_id": "2ded77f6-5d08-5670-945e-86ac88c60318", + "type": "text", + "content": "", + "path": "micron-530bd7ed.pdf/Appendix", + "metadata": { + "length": 0, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "92d5bae8-77c5-53f8-b283-fc3c88438dcd", + "type": "table", + "content": "
    Amounts in millions, except per shareFQ1-26% of revenueFQ4-25% of revenueFQ1-25% of revenue
    Revenue$13,643100%$11,315100%$8,709100%
    Gross margin7,75357%5,16946%3,44140%
    Operating income6,41947%3,95535%2,39427%
    Income tax (provision) benefit(977)(471)(333)
    Net income5,48240%3,46931%2,03723%
    Diluted earnings per share4.783.031.79
    Cash provided by operating activities (GAAP)8,4115,7303,244
    Cash, marketable investments, and restricted cash (GAAP)12,01611,9408,748
    ", + "path": "tables/table-3 Financial Results.html", + "metadata": { + "length": 1118, + "summary": "table-4\nThe table shows quarterly financial data for FQ1-26, FQ4-25, and FQ1-25, with revenue increasing from $8.7B to $13.6B, gross margin rising from 40% to 57%, and net income growing from $2.0B to $5.5B.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-3 Financial Results.html", + "keywords": [ + "Revenue", + "Gross Margin", + "Net Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "4852d9e9-df7a-53d2-85b6-7c206f496a40", + "type": "text", + "content": "Non-GAAP\n\n[tables/table-3 Financial Results.html]\n\nSee non-GAAP reconciliations.", + "path": "micron-530bd7ed.pdf/Appendix/Financial summary", + "metadata": { + "length": 120, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Non", + "GAAP", + "See", + "reconciliations" + ], + "keywords": [], + "connect_to": [ + { + "target": "92d5bae8-77c5-53f8-b283-fc3c88438dcd", + "relation": "embeds", + "ref": "[tables/table-3 Financial Results.html]", + "position": { + "start": 10, + "end": 49 + } + } + ] + } + }, + { + "chunk_id": "678c565b-39d9-5119-b8bd-819a1a93732f", + "type": "table", + "content": "
    % of RevenueFQ1-26
    DRAM79%
    NAND20%
    ", + "path": "tables/table-4 Revenue Breakdown.html", + "metadata": { + "length": 128, + "summary": "table-5\nDRAM accounts for 79% of revenue, while NAND contributes 20% in FQ1-26.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-4 Revenue Breakdown.html", + "keywords": [ + "DRAM", + "NAND", + "Revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "15fbec1a-cfc7-5fb3-a485-3e2fd925b6c5", + "type": "table", + "content": "
    % Sales Volume ChangeFQ1-26 Q/Q
    DRAMUp slightly
    NANDIncreased in mid-to-high single-digit percentage range
    ", + "path": "tables/table-5 Sales Volume Change.html", + "metadata": { + "length": 200, + "summary": "table-6\nDRAM sales volume up slightly quarter over quarter; NAND increased in mid-to-high single-digit percentage range.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-5 Sales Volume Change.html", + "keywords": [ + "DRAM", + "NAND", + "Sales Change" + ], + "tokens": [] + } + }, + { + "chunk_id": "57d0a6f0-ff87-5fb6-9318-d6b775367501", + "type": "table", + "content": "
    % ASP ChangeFQ1-26 Q/Q
    DRAMIncreased approximately 20%
    NANDIncreased in the mid-teens percentage range
    ", + "path": "tables/table-6 ASP Changes Q_Q.html", + "metadata": { + "length": 196, + "summary": "table-7\nDRAM ASP increased approximately 20% Q/Q, while NAND ASP increased in the mid-teens percentage range in FQ1-26.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-6 ASP Changes Q_Q.html", + "keywords": [ + "DRAM", + "NAND", + "ASP" + ], + "tokens": [] + } + }, + { + "chunk_id": "125be82c-5fdd-54de-887f-207351be5bf4", + "type": "table", + "content": "
    FQ1-26 non-GAAP (amounts in millions, except per share)FQ2-26 non-GAAP Guidance
    Revenue$13,643$18.70 billion ± $400 million
    Gross margin56.8%68.0% ± 1.0%
    Operating expenses$1,334$1.38 billion ± $20 million
    Diluted earnings per share$4.78$8.42 ± $0.20
    ", + "path": "tables/table-7 Non-GAAP Guidance.html", + "metadata": { + "length": 441, + "summary": "table-8\nThe table compares FQ1-26 non-GAAP actuals with FQ2-26 guidance: Revenue $13.6B vs $18.7B, Gross margin 56.8% vs 68.0%, OpEx $1.33B vs $1.38B, Diluted EPS $4.78 vs $8.42.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-7 Non-GAAP Guidance.html", + "keywords": [ + "Revenue", + "Gross Margin", + "EPS" + ], + "tokens": [] + } + }, + { + "chunk_id": "95e9856b-2520-58c5-a10b-a5d98f7c22b9", + "type": "table", + "content": "
    FQ1-26 non-GAAP (amounts in millions)FQ2-26 non-GAAP Estimates
    Diluted shares1,148~1.15 billion
    Income tax (provision) benefit($977)Around 15.5%
    Cash from operations (GAAP)$8,411
    Investments in capex, net (capital cash flow)$4,505~ $20 billion in fiscal 2026
    ", + "path": "tables/table-8 Non-GAAP Estimates.html", + "metadata": { + "length": 450, + "summary": "table-9\nTable shows FQ1-26 non-GAAP actuals and FQ2-26 estimates: diluted shares ~1.15B, tax rate ~15.5%, capex ~$20B for fiscal 2026.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-8 Non-GAAP Estimates.html", + "keywords": [ + "Non-GAAP", + "Estimates", + "Fiscal 2026" + ], + "tokens": [] + } + }, + { + "chunk_id": "ee710aa1-37c1-5f59-a23e-6bd1862ab64e", + "type": "text", + "content": "[tables/table-4 Revenue Breakdown.html]\n\n\n[tables/table-5 Sales Volume Change.html]\n\n\n[tables/table-6 ASP Changes Q_Q.html]\n\n\n[tables/table-7 Non-GAAP Guidance.html]\n\n\n[tables/table-8 Non-GAAP Estimates.html]", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP financial data and guidance", + "metadata": { + "length": 292, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [], + "keywords": [], + "connect_to": [ + { + "target": "678c565b-39d9-5119-b8bd-819a1a93732f", + "relation": "embeds", + "ref": "[tables/table-4 Revenue Breakdown.html]", + "position": { + "start": 0, + "end": 39 + } + }, + { + "target": "15fbec1a-cfc7-5fb3-a485-3e2fd925b6c5", + "relation": "embeds", + "ref": "[tables/table-5 Sales Volume Change.html]", + "position": { + "start": 42, + "end": 83 + } + }, + { + "target": "57d0a6f0-ff87-5fb6-9318-d6b775367501", + "relation": "embeds", + "ref": "[tables/table-6 ASP Changes Q_Q.html]", + "position": { + "start": 86, + "end": 123 + } + }, + { + "target": "125be82c-5fdd-54de-887f-207351be5bf4", + "relation": "embeds", + "ref": "[tables/table-7 Non-GAAP Guidance.html]", + "position": { + "start": 126, + "end": 165 + } + }, + { + "target": "95e9856b-2520-58c5-a10b-a5d98f7c22b9", + "relation": "embeds", + "ref": "[tables/table-8 Non-GAAP Estimates.html]", + "position": { + "start": 168, + "end": 208 + } + } + ] + } + }, + { + "chunk_id": "e1164a66-ea1a-5cdd-bb6c-840d028b0b9d", + "type": "table", + "content": "
    Amounts in millionsFQ1-26% of revenueFQ4-25% of revenueFQ1-25% of revenue
    DRAM$10,81279%$8,98479%$6,40073%
    NAND2,74320%2,25220%2,24126%
    Other (primarily NOR)881%791%681%
    Total$13,643100%$11,315100%$8,709100%
    ", + "path": "tables/table-9 Memory Revenue.html", + "metadata": { + "length": 578, + "summary": "table-10\nThe table shows memory revenue breakdown for FQ1-26, FQ4-25, and FQ1-25. DRAM revenue grew from $6,400M to $10,812M, NAND from $2,241M to $2,743M, with total revenue increasing from $8,709M to $13,643M.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-9 Memory Revenue.html", + "keywords": [ + "DRAM", + "NAND", + "Revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "b991c6e0-7195-5a78-bd34-5b7931d26c56", + "type": "table", + "content": "
    Amounts in millionsFQ1-26FQ4-25Q/Q % ChangeFQ1-25Y/Y % Change
    DRAM$10,812$8,98420%$6,40069%
    NAND2,7432,25222%2,24122%
    Other (primarily NOR)887911%6829%
    Total$13,643$11,31521%$8,70957%
    ", + "path": "tables/table-10 Memory Revenue.html", + "metadata": { + "length": 509, + "summary": "table-11\nMemory revenue for FQ1-26 was $13,643M, up 21% Q/Q and 57% Y/Y. DRAM grew 20% Q/Q and 69% Y/Y to $10,812M. NAND grew 22% Q/Q and 22% Y/Y to $2,743M.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-10 Memory Revenue.html", + "keywords": [ + "DRAM", + "NAND", + "Revenue" + ], + "tokens": [] + } + }, + { + "chunk_id": "11bf4246-ab06-5c87-ad3a-ee7b92602efd", + "type": "table", + "content": "
    Amounts in millionsFQ1-26FQ4-25Q/Q % ChangeFQ1-25Y/Y % Change
    Cloud Memory (CMBU)$5,284$4,54316%$2,648100%
    Core Data (CDBU)2,3791,57751%2,2924%
    Mobile and Client (MCBU)4,2553,76013%2,60863%
    Automotive and Embedded (AEBU)1,7201,43420%1,15849%
    ", + "path": "tables/table-11 Revenue by Segment.html", + "metadata": { + "length": 567, + "summary": "table-12\nThe table shows revenue for four segments: Cloud Memory ($5,284M), Core Data ($2,379M), Mobile and Client ($4,255M), and Automotive and Embedded ($1,720M) with Q/Q and Y/Y percentage changes.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-11 Revenue by Segment.html", + "keywords": [ + "Revenue", + "Segments", + "Growth" + ], + "tokens": [] + } + }, + { + "chunk_id": "f315bd05-b59c-5797-b9e7-7c2f61be3b82", + "type": "text", + "content": "[tables/table-9 Memory Revenue.html]\n\nPercentages of total revenue may not total 100% due to rounding.\nRevenue by technology\n\n[tables/table-10 Memory Revenue.html]\n\nRevenue by business unit\n\n[tables/table-11 Revenue by Segment.html]", + "path": "micron-530bd7ed.pdf/Appendix/Revenue by technology", + "metadata": { + "length": 372, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Percentages", + "total", + "revenue", + "100%", + "due", + "rounding", + "Revenue", + "technology", + "business", + "unit" + ], + "keywords": [], + "connect_to": [ + { + "target": "e1164a66-ea1a-5cdd-bb6c-840d028b0b9d", + "relation": "embeds", + "ref": "[tables/table-9 Memory Revenue.html]", + "position": { + "start": 0, + "end": 36 + } + }, + { + "target": "b991c6e0-7195-5a78-bd34-5b7931d26c56", + "relation": "embeds", + "ref": "[tables/table-10 Memory Revenue.html]", + "position": { + "start": 126, + "end": 163 + } + }, + { + "target": "11bf4246-ab06-5c87-ad3a-ee7b92602efd", + "relation": "embeds", + "ref": "[tables/table-11 Revenue by Segment.html]", + "position": { + "start": 191, + "end": 232 + } + } + ] + } + }, + { + "chunk_id": "e2c78da7-c375-5233-8f2d-26abf6731248", + "type": "text", + "content": "Consolidated results", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations", + "metadata": { + "length": 20, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Consolidated", + "results" + ], + "keywords": [], + "connect_to": [] + } + }, + { + "chunk_id": "6c5bdab9-c153-50ff-88ba-50888561318c", + "type": "table", + "content": "
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    GAAP gross margin$7,646$5,054$3,348
    Stock-based compensation10711590
    Other3
    Non-GAAP gross margin$7,753$5,169$3,441
    GAAP operating expenses$1,510$1,400$1,174
    Stock-based compensation(173)(147)(127)
    Restructure and asset impairments(38)
    Other(3)(1)
    Non-GAAP operating expenses$1,334$1,214$1,047
    GAAP operating income$6,136$3,654$2,174
    Stock-based compensation280262217
    Restructure and asset impairments38
    Other313
    Non-GAAP operating income$6,419$3,955$2,394
    ", + "path": "tables/table-12 Gross Margin & OpEx.html", + "metadata": { + "length": 1237, + "summary": "table-13\nFinancial data for FQ1-26, FQ4-25, FQ1-25 showing GAAP and Non-GAAP gross margins, operating expenses, and operating income with adjustments for stock-based compensation and other items.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-12 Gross Margin & OpEx.html", + "keywords": [ + "GAAP", + "Non-GAAP", + "Operating Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "e2c78da7-c375-5233-8f2d-26abf6731248", + "type": "text", + "content": "[tables/table-12 Gross Margin & OpEx.html]\n\nConsolidated results", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_2", + "metadata": { + "length": 85, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Consolidated", + "results" + ], + "keywords": [], + "connect_to": [ + { + "target": "6c5bdab9-c153-50ff-88ba-50888561318c", + "relation": "embeds", + "ref": "[tables/table-12 Gross Margin & OpEx.html]", + "position": { + "start": 0, + "end": 42 + } + } + ] + } + }, + { + "chunk_id": "e4ea5bf4-8def-5c62-b5e7-732ad9a9dcd1", + "type": "table", + "content": "
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    GAAP cost of goods sold$5,997$6,261$5,361
    Stock-based compensation(107)(115)(90)
    Other(3)
    Non-GAAP cost of goods sold$5,890$6,146$5,268
    GAAP research and development$1,171$1,047$888
    Stock-based compensation(106)(93)(77)
    Other(1)(1)
    Non-GAAP research and development$1,064$953$811
    GAAP selling, general, and administrative$337$314$288
    Stock-based compensation(67)(54)(50)
    Non-GAAP selling, general, and administrative$270$260$238
    ", + "path": "tables/table-13 Non-GAAP Financials.html", + "metadata": { + "length": 1103, + "summary": "table-14\nThe table compares GAAP and Non-GAAP costs for COGS, R&D, and SG&A across FQ1-26, FQ4-25, and FQ1-25, with stock-based compensation adjustments.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-13 Non-GAAP Financials.html", + "keywords": [ + "GAAP", + "Non-GAAP", + "Costs" + ], + "tokens": [] + } + }, + { + "chunk_id": "e2c78da7-c375-5233-8f2d-26abf6731248", + "type": "text", + "content": "[tables/table-13 Non-GAAP Financials.html]\n\nConsolidated results", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_3", + "metadata": { + "length": 85, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Consolidated", + "results" + ], + "keywords": [], + "connect_to": [ + { + "target": "e4ea5bf4-8def-5c62-b5e7-732ad9a9dcd1", + "relation": "embeds", + "ref": "[tables/table-13 Non-GAAP Financials.html]", + "position": { + "start": 0, + "end": 42 + } + } + ] + } + }, + { + "chunk_id": "2472ca80-4436-5b71-8226-42b56e920a9a", + "type": "table", + "content": "
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    GAAP net income$5,240$3,201$1,870
    Stock-based compensation280262217
    Restructure and asset impairments38
    Loss on debt prepayments1309
    Other(20)1
    Estimated tax effects of above and other tax adjustments(148)(42)(50)
    Non-GAAP net income$5,482$3,469$2,037
    GAAP income tax (provision) benefit($829)($429)($283)
    Estimated tax effects of non-GAAP adjustments and other tax adjustments(148)(42)(50)
    Non-GAAP income tax (provision) benefit($977)($471)($333)
    ", + "path": "tables/table-14 Non-GAAP Net Income.html", + "metadata": { + "length": 1020, + "summary": "table-15\nThe table shows GAAP net income and adjustments for stock-based compensation, restructuring, and debt losses, resulting in Non-GAAP net income of $5,482M for FQ1-26, $3,469M for FQ4-25, and $2,037M for FQ1-25.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-14 Non-GAAP Net Income.html", + "keywords": [ + "GAAP", + "Non-GAAP", + "Net Income" + ], + "tokens": [] + } + }, + { + "chunk_id": "e2c78da7-c375-5233-8f2d-26abf6731248", + "type": "text", + "content": "[tables/table-14 Non-GAAP Net Income.html]\n\nConsolidated results", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_4", + "metadata": { + "length": 85, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Consolidated", + "results" + ], + "keywords": [], + "connect_to": [ + { + "target": "2472ca80-4436-5b71-8226-42b56e920a9a", + "relation": "embeds", + "ref": "[tables/table-14 Non-GAAP Net Income.html]", + "position": { + "start": 0, + "end": 42 + } + } + ] + } + }, + { + "chunk_id": "5f7cb4a5-ecfc-5be9-93ee-7a5c3d130857", + "type": "table", + "content": "
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    GAAP net income$5,240$3,201$1,870
    Interest (income) expense, net(65)(22)11
    Income tax provision (benefit)829429283
    Depreciation expense and amortization of intangible assets2,2122,1492,030
    Non-GAAP adjustments
    Stock-based compensation280262217
    Restructure and asset impairments38
    Loss on debt prepayments1309
    Other(22)
    Adjusted EBITDA$8,604$6,066$4,411
    ", + "path": "tables/table-15 Adjusted EBITDA.html", + "metadata": { + "length": 895, + "summary": "table-16\nThe table shows GAAP net income and adjustments to calculate Adjusted EBITDA for FQ1-26, FQ4-25, and FQ1-25, with values of $8,604M, $6,066M, and $4,411M respectively.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-15 Adjusted EBITDA.html", + "keywords": [ + "EBITDA", + "GAAP", + "Non-GAAP" + ], + "tokens": [] + } + }, + { + "chunk_id": "e2c78da7-c375-5233-8f2d-26abf6731248", + "type": "text", + "content": "[tables/table-15 Adjusted EBITDA.html]\n\nConsolidated results", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_5", + "metadata": { + "length": 85, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Consolidated", + "results" + ], + "keywords": [], + "connect_to": [ + { + "target": "5f7cb4a5-ecfc-5be9-93ee-7a5c3d130857", + "relation": "embeds", + "ref": "[tables/table-15 Adjusted EBITDA.html]", + "position": { + "start": 0, + "end": 38 + } + } + ] + } + }, + { + "chunk_id": "f48634a3-aff2-5a2f-8f27-f5bf07c21db3", + "type": "table", + "content": "
    Amounts in millions, except per shareFQ1-26FQ4-25FQ1-25
    GAAP shares used in diluted EPS calculations$1,138$1,131$1,122
    Adjustment for stock-based compensation101416
    Non-GAAP shares used in diluted EPS calculations$1,148$1,145$1,138
    GAAP diluted earnings per share$4.60$2.83$1.67
    Effects of non-GAAP adjustments0.180.200.12
    Non-GAAP diluted earnings per share$4.78$3.03$1.79
    Net cash provided by operating activities$8,411$5,730$3,244
    Expenditures for property, plant, and equipment(5,389)(5,658)(3,206)
    Proceeds from sales of property, plant, and equipment6209
    Proceeds from government incentives87871165
    Investments in capital expenditures, net(4,505)(4,927)(3,132)
    Adjusted free cash flow$3,906$803$112
    ", + "path": "tables/table-16 Financial Highlights.html", + "metadata": { + "length": 1292, + "summary": "table-17\nThe table shows GAAP and Non-GAAP diluted EPS, shares, and adjusted free cash flow for FQ1-26, FQ4-25, and FQ1-25. Non-GAAP EPS rose to $4.78 from $1.79, and adjusted free cash flow increased to $3,906M from $112M.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-16 Financial Highlights.html", + "keywords": [ + "EPS", + "Free Cash Flow", + "GAAP" + ], + "tokens": [] + } + }, + { + "chunk_id": "e2c78da7-c375-5233-8f2d-26abf6731248", + "type": "text", + "content": "[tables/table-16 Financial Highlights.html]\n\nConsolidated results", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_6", + "metadata": { + "length": 89, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Consolidated", + "results" + ], + "keywords": [], + "connect_to": [ + { + "target": "f48634a3-aff2-5a2f-8f27-f5bf07c21db3", + "relation": "embeds", + "ref": "[tables/table-16 Financial Highlights.html]", + "position": { + "start": 0, + "end": 43 + } + } + ] + } + }, + { + "chunk_id": "593b7280-74b7-5f9f-8e93-d06ad748f362", + "type": "table", + "content": "
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    Cash and cash equivalents$9,731$9,642$6,693
    Short-term investments587665895
    Long-term marketable investments1,6971,6291,156
    Restricted cash144
    Current debt(569)(560)(533)
    Long-term debt(11,187)(14,017)(13,252)
    Net cash$260($2,637)($5,037)
    ", + "path": "tables/table-17 Cash Position.html", + "metadata": { + "length": 644, + "summary": "table-18\nThe table shows cash, investments, and debt for FQ1-26, FQ4-25, and FQ1-25. Net cash improved from negative to positive $260 million in FQ1-26.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-17 Cash Position.html", + "keywords": [ + "Cash", + "Debt", + "Investments" + ], + "tokens": [] + } + }, + { + "chunk_id": "3375d7a1-f78b-5bb0-8be2-e644a03cb090", + "type": "text", + "content": "[tables/table-17 Cash Position.html]", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_7", + "metadata": { + "length": 63, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [], + "keywords": [], + "connect_to": [ + { + "target": "593b7280-74b7-5f9f-8e93-d06ad748f362", + "relation": "embeds", + "ref": "[tables/table-17 Cash Position.html]", + "position": { + "start": 0, + "end": 36 + } + } + ] + } + }, + { + "chunk_id": "f377b130-b598-574e-857d-289bd0f99c88", + "type": "table", + "content": "
    GAAP OutlookAdjustmentsNon-GAAP Outlook
    Revenue$18.70 billion ± $400 million$18.70 billion ± $400 million
    Gross margin67.0% ± 1.0%1.0%A68.0% ± 1.0%
    Operating expenses$1.56 billion ± $20 million$180 millionB$1.38 billion ± $20 million
    Diluted earnings per share*$8.19 ± $0.20$0.23A, B, C$8.42 ± $0.20
    ", + "path": "tables/table-18 GAAP vs Non-GAAP.html", + "metadata": { + "length": 584, + "summary": "table-19\nThe table compares GAAP and Non-GAAP outlook for revenue ($18.70B), gross margin (67% vs 68%), operating expenses ($1.56B vs $1.38B), and diluted EPS ($8.19 vs $8.42) with adjustments.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-18 GAAP vs Non-GAAP.html", + "keywords": [ + "GAAP", + "Non-GAAP", + "Outlook" + ], + "tokens": [] + } + }, + { + "chunk_id": "9b894756-58b4-55ca-91a7-5a450f89763b", + "type": "table", + "content": "
    AStock-based compensation – cost of goods sold$125
    BStock-based compensation – research and development119
    BStock-based compensation – selling, general, and administrative61
    CTax effects of the above items and other tax adjustments(48)
    $257
    ", + "path": "tables/table-19 Stock Compensation.html", + "metadata": { + "length": 434, + "summary": "table-20\nThe table shows stock-based compensation costs: $125 for cost of goods sold, $119 for R&D, $61 for SG&A, and a tax adjustment of ($48), totaling $257.", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "file_path": "tables/table-19 Stock Compensation.html", + "keywords": [ + "Stock-based compensation", + "Cost of goods sold", + "Tax effects" + ], + "tokens": [] + } + }, + { + "chunk_id": "e677bc01-f247-5dcf-a2b6-86f9c248b387", + "type": "text", + "content": "Non-GAAP reconciliations\n\n[tables/table-18 GAAP vs Non-GAAP.html]\n\nNon-GAAP Adjustments (amounts in millions)\n\n[tables/table-19 Stock Compensation.html]\n\n\\*GAAP earnings per share based on approximately 1.14 billion diluted shares and non-GAAP earnings per share based on approximately 1.15 billion diluted shares.\nThe above guidance does not incorporate the impact of any potential business combinations, divestitures, additional restructuring activities, balance sheet valuation adjustments, strategic investments, financing transactions, and other significant transactions. The timing and impact of such items are dependent on future events that may be uncertain or outside of our control.", + "path": "micron-530bd7ed.pdf/Appendix/FQ2-26 guidance Non-GAAP reconciliations", + "metadata": { + "length": 727, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Non", + "GAAP", + "reconciliations", + "Adjustments", + "amounts", + "millions", + "earnings", + "share", + "based", + "approximately", + "1.14", + "billion", + "diluted", + "shares", + "1.15", + "The", + "guidance", + "incorporate", + "impact", + "potential", + "business", + "combinations", + "divestitures", + "additional", + "restructuring", + "activities", + "balance", + "sheet", + "valuation", + "adjustments", + "strategic", + "investments", + "financing", + "transactions", + "significant", + "timing", + "items", + "dependent", + "future", + "events", + "uncertain", + "control" + ], + "keywords": [], + "connect_to": [ + { + "target": "f377b130-b598-574e-857d-289bd0f99c88", + "relation": "embeds", + "ref": "[tables/table-18 GAAP vs Non-GAAP.html]", + "position": { + "start": 26, + "end": 65 + } + }, + { + "target": "9b894756-58b4-55ca-91a7-5a450f89763b", + "relation": "embeds", + "ref": "[tables/table-19 Stock Compensation.html]", + "position": { + "start": 111, + "end": 152 + } + } + ] + } + }, + { + "chunk_id": "d84931a0-d4b7-5279-9a48-46756054d2e0", + "type": "text", + "content": "Intelligence\nAccelerated™\n© 2025 Micron Technology, Inc. All rights reserved. Information, products, and/or specifications are subject to change without notice. All information is provided on an “AS IS” basis without warranties of any kind. Statements regarding products, including statements regarding product features, availability, functionality, or compatibility, are provided for informational purposes only and do not modify the warranty, if any, applicable to any product. Drawings may not be to scale. Micron, the Micron logo, the M logo, Intelligence Accelerated™, and other Micron trademarks are the property of Micron Technology, Inc. All other trademarks are the property of their respective owners.", + "path": "micron-530bd7ed.pdf/micron", + "metadata": { + "length": 711, + "summary": "", + "page_nums": [], + "document_top_summary": "This document includes: Financial results, Outlook, Appendix, micron", + "tokens": [ + "Intelligence", + "Accelerated", + "2025", + "Micron", + "Technology", + "Inc", + "All", + "rights", + "reserved", + "Information", + "products", + "specifications", + "subject", + "change", + "notice", + "information", + "provided", + "AS", + "IS", + "basis", + "warranties", + "kind", + "Statements", + "including", + "statements", + "product", + "features", + "availability", + "functionality", + "compatibility", + "informational", + "purposes", + "modify", + "warranty", + "applicable", + "Drawings", + "scale", + "logo", + "trademarks", + "property", + "respective", + "owners" + ], + "keywords": [], + "connect_to": [] + } + } + ] +} \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/doc_nav.json b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/doc_nav.json new file mode 100644 index 000000000..b816f3276 --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/doc_nav.json @@ -0,0 +1,518 @@ +{ + "version": "1.0", + "file_name": "Micron investor report 530bd7ed.pdf", + "stats": { + "total_chunks": 82, + "text_chunks": 42, + "image_chunks": 20, + "table_chunks": 20, + "max_depth": 3 + }, + "sections": [ + { + "title": "Financial results", + "path": "micron-530bd7ed.pdf/Financial results", + "level": 1, + "summary": "FQ1 2026 # Financial results
    0 [images/image-1-Abstract Grid Pattern.jpg]
    natural_image Close-up of a semiconductor wafer with green, purple, and blue photovoltaic bands (no text or symbols visible)
    ", + "chunk_count": 28, + "children": [ + { + "title": "Safe harbor statement", + "path": "micron-530bd7ed.pdf/Financial results/Safe harbor statement", + "level": 2, + "summary": "This passage contains standard legal disclaimers regarding forward-looking statements and non-GAAP financial measures. It cautions that projections about market demand, pricing, technology impact, manufacturing investments, and financial performance involve risks and uncertainties. It directs readers to SEC filings and a risk factor website for details. It also explains that non-GAAP measures exclude certain activities to help analyze operating results and trends, with reconciliations provided in the appendix.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Sanjay Mehrotra", + "path": "micron-530bd7ed.pdf/Financial results/Sanjay Mehrotra", + "level": 2, + "summary": "Chairman, President and Chief Executive Officer ## Safe harbor statement
    0 [images/image-2-Micron Executive Portrait.jpg]
    natural_image Portrait of a smiling man with gray hair and beard, wearing a blue blazer over a patterned shirt (no text or symbols visible)
    0 [images/image-3-Automated RAM Testing Machine.jpg]
    natural_image Industrial automation scene with robotic arms and green memory chips under purple lighting (no readable text or symbols)
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Overview (1 of 3)", + "path": "micron-530bd7ed.pdf/Financial results/Overview (1 of 3)", + "level": 2, + "summary": "- Micron had an outstanding start to fiscal 2026, delivering fiscal Q1 revenue, gross margin and EPS (earnings per share) well above the high end of our guidance. - This financial performance was driven by our strong execution across end markets and products in a tight supply environment. We achieved a number of records in fiscal Q1. - Total company revenue, DRAM and NAND revenue, as well as HBM and data center revenue and revenue in each of our business units, also reached new records. • We have completed agreements on price and volume for our entire calendar 2026 HBM supply, including Micron's industry-leading HBM4. - We forecast an HBM TAM CAGR of approximately 40% through calendar 2028, from approximately \\$35 billion in 2025 to around \\$100 billion in 2028. - This \\$100 billion HBM TAM milestone is now projected to arrive two years earlier than in our prior outlook. Remarkably, this 2028 HBM TAM projection is larger than the size of the entire DRAM market in calendar 2024. - This financial performance was driven by our strong execution across end markets and products in a tight supply environment. We achieved a number of records in fiscal Q1.
    0 [images/image-4-Automated Industrial Machinery.jpg]
    natural_image Interior view of a modern industrial facility with blue-tinted machinery and control panels (no visible text or symbols)
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Overview (2 of 3)", + "path": "micron-530bd7ed.pdf/Financial results/Overview (2 of 3)", + "level": 2, + "summary": "Micron highlights its customized HBM4E customer engagements and progress on its HBM roadmap, emphasizing memory's shift from a system component to a strategic asset for AI. The company's technology leadership and product portfolio position it as a key AI enabler, with anticipated record revenue, gross margin, EPS, and free cash flow for Q2 and full fiscal year 2026.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Overview (3 of 3)", + "path": "micron-530bd7ed.pdf/Financial results/Overview (3 of 3)", + "level": 2, + "summary": "- Sustained and strong industry demand, along with supply constraints, are contributing to tight market conditions, and we expect these conditions to persist beyond calendar 2026. • We are making progress with customers in our discussions for multiyear contracts with specific commitments. - Simultaneously, we are focused on maximizing our production output from our current footprint, ramping our industry-leading technology nodes and investing in new cleanroom space to add to our supply capability.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Technology and operations (1 of 2)", + "path": "micron-530bd7ed.pdf/Financial results/Technology and operations (1 of 2)", + "level": 2, + "summary": "- Micron’s technology leadership is foundational to our strong competitive position. - Micron has led the industry for four consecutive technology nodes in DRAM and three nodes in NAND, with progressively faster yield ramps in every node. - Our 1-gamma DRAM node is ramping well. 1-gamma will be the primary driver of our DRAM bit growth in calendar 2026, and will be the majority of our bit output in the second half of the calendar year. - Looking beyond 1-gamma, development is underway for 1-delta and 1-epsilon nodes, which will feature innovations that we expect to extend our differentiation and technology leadership.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Technology and operations (2 of 2)", + "path": "micron-530bd7ed.pdf/Financial results/Technology and operations (2 of 2)", + "level": 2, + "summary": "- In NAND, we are ramping our G9 node with robust yield ramps across both data center and client SSDs. Our QLC NAND mix, including G9 QLC, reached a record high during the quarter. - Technology transitions to G9 will be the primary driver of our NAND bit growth in calendar 2026, and we expect it to become our largest NAND node later in fiscal 2026. - Calendar 2025 is a record year for Micron in terms of both internal and customer quality measures, positioning us well to deliver for our customers as the memory industry's quality leader. - As our products are increasingly integrated into higher value applications, our leadership in quality is becoming a more important differentiator. ## Technology and operations (2 of 2)
    0 [images/image-6-Micron Semiconductor Wafer.jpg]
    text_image micron | 8
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Data center server demand", + "path": "micron-530bd7ed.pdf/Financial results/Data center server demand", + "level": 2, + "summary": "- As the world’s leading technology companies advance toward artificial general intelligence and transform the global economy, our customers are committing to an extraordinary, multiyear data center buildout. - This growth in AI data center capacity is driving a significant increase in demand for high-performance and high-capacity memory and storage. - Server unit demand has strengthened significantly, and we now expect calendar 2025 server unit growth in the high-teens percentage range, higher than our last earnings call outlook of $10\\%$ . • We expect server demand strength to continue in 2026. Server memory and storage content and performance requirements continue to increase generation to generation. - Micron has a differentiated portfolio of high-value data center solutions to address these requirements, including our HBM, high-capacity server memory solutions and data center SSDs. - As the world’s leading technology companies advance toward artificial general intelligence and transform the global economy, our customers are committing to an extraordinary, multiyear data center buildout.
    0 [images/image-7-Server Rack Infrastructure.jpg]
    natural_image Interior view of a server room with rows of server racks and illuminated green indicator lights, labeled 'micron' in the corner (no readable text on equipment)
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Data center HBM and LP DRAM", + "path": "micron-530bd7ed.pdf/Financial results/Data center HBM and LP DRAM", + "level": 2, + "summary": "- Micron’s HBM4, with industry leading speed over 11 Gbps, is on track to ramp with high yields in the second calendar quarter of 2026, consistent with our customers’ product ramp plans. - Our HBM4 uses advanced CMOS and advanced metallization process technologies on the base logic die and DRAM core dies, which are designed and manufactured in-house. - This, along with our unique HBM design, packaging and test capability, enables Micron’s industry-leading performance and low-power leadership. - Micron pioneered the adoption of LP DRAM in the data center. Micron's Low Power (LP) DRAM server modules consume one-third the power of DDR DRAM server modules. - Building on this leadership, we have sampled our 192GB LP SOCAMM2 product, which enables a $50\\%$ increase in capacity per module and a rack-scale LP DRAM density of over 50TB. - Micron’s HBM4, with industry leading speed over 11 Gbps, is on track to ramp with high yields in the second calendar quarter of 2026, consistent with our customers’ product ramp plans.
    0 [images/image-8-Micron Data Center.jpg]
    natural_image Interior view of a modern data center with server racks illuminated by purple and blue neon lights (no visible text or symbols)
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Data center NAND", + "path": "micron-530bd7ed.pdf/Financial results/Data center NAND", + "level": 2, + "summary": "- Our data center NAND portfolio revenue exceeded a billion dollars in fiscal Q1, and we are seeing strong momentum across our data center SSD portfolio, enabled by our leadership NAND technology. - In the performance SSD category, Micron has introduced the world's first PCIe Gen6 SSD, leveraging our G9 NAND. • We are seeing rapidly increasing qualification commitments for this product, including at hyperscalers. - In mainstream storage, our SSDs based on G9 NAND are already seeing robust demand in the first quarter of calendar 2026. - In capacity storage, our QLC-based 122 and 245TB G9 SSDs are entering qualification at multiple hyperscale customers.", + "chunk_count": 1, + "children": [] + }, + { + "title": "PC", + "path": "micron-530bd7ed.pdf/Financial results/PC", + "level": 2, + "summary": "- PC demand continues to be driven by Windows 10 end of life and AI PCs. • We forecast PC unit sales to grow high-single-digit percentage range in calendar 2025, above our prior expectations provided in our last earnings call of mid-single digits. - As we look ahead into 2026, we expect these demand drivers to continue, while memory supply constraints may affect some PC unit shipments. - Micron has completed multiple OEM qualifications of our 16Gb 1-gamma-based DDR5 and our G9-based PCIe Gen4 QLC SSDs. ## PC
    0 [images/image-9-Laptop Keyboard and Screen.jpg]
    text_image micron | 12
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Mobile", + "path": "micron-530bd7ed.pdf/Financial results/Mobile", + "level": 2, + "summary": "- Smartphone unit volumes in calendar 2025 are on track to grow in the low-single-digit percentage range. - AI is driving memory content growth. The shipment mix of flagship smartphones with 12GB of DRAM increased to $59\\%$ in calendar Q3, more than twice the level from a year ago. • Micron is accelerating innovation across our mobile DRAM portfolio. - In fiscal Q1, we began sampling our breakthrough 1-gamma 16Gb LPDDR6 product to leading OEM and ecosystem partners, marking a major milestone in next-generation memory technology. - LPDDR6 will power AI at the edge, delivering over 50% higher performance and improved power efficiency for flagship smartphones and AI PCs. - Micron also sampled our 1-gamma LP5x 24Gb product and began volume shipments of the previously announced 1-gamma LP5x 16Gb product to multiple OEMs. - AI is driving memory content growth. The shipment mix of flagship smartphones with 12GB of DRAM increased to $59\\%$ in calendar Q3, more than twice the level from a year ago.
    0 [images/image-10-Concert Photography.jpg]
    natural_image Hand holding smartphone displaying a live music show with stage lights against a red and pink background (no text or symbols on screen)
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Automotive and embedded", + "path": "micron-530bd7ed.pdf/Financial results/Automotive and embedded", + "level": 2, + "summary": "The passage discusses strong demand for memory and storage in automotive (L2+ and L3 adoption, ASIL-rated LPDDR5x and UFS4.1 NAND) and industrial markets (autonomous systems, factory automation, robotics). Micron highlights its differentiated product portfolio and market leadership, with billions in design wins and investments in long-term supply from its Virginia fab.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Market outlook (1 of 2)", + "path": "micron-530bd7ed.pdf/Financial results/Market outlook (1 of 2)", + "level": 2, + "summary": "• Over the last few months, our customers' AI data center build-out plans have driven a sharp increase in demand forecasts for memory and storage. • We believe that the aggregate industry supply will remain substantially short of the demand for the foreseeable future. - The dramatic increase in HBM demand is further challenging the supply environment due to the 3-to-1 trade ratio with DDR5, and this trade ratio only increases with future generations of HBM. - Additional cleanroom space is necessary to address this increased demand, and lead times for cleanroom build-out are lengthening across geographies. • Together, these demand and supply factors are driving tight industry conditions across DRAM and NAND, and we expect tightness to persist through and beyond calendar 2026. • Over the last few months, our customers' AI data center build-out plans have driven a sharp increase in demand forecasts for memory and storage.
    0 [images/image-13-Silicon Wafer in Manufacturing.jpg]
    natural_image Close-up of a semiconductor wafer with colorful grid patterns, captured in a lab setting (no text or symbols on the wafer itself)
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Market outlook (2 of 2)", + "path": "micron-530bd7ed.pdf/Financial results/Market outlook (2 of 2)", + "level": 2, + "summary": "- Calendar 2025 DRAM and NAND industry bit demand growth expectations are higher than in our last earnings call outlook. - We now expect calendar 2025 DRAM bit demand growth in the low $20\\%$ range versus high teens previously. - We expect 2025 NAND bit demand growth to be in the high-teens percentage range versus low to mid-teens previously. - We expect calendar 2026 industry DRAM and NAND bit shipment growth to be constrained by industry supply. - We expect both DRAM and NAND calendar 2026 industry bit shipments to increase around 20% from 2025 levels. - Micron is working hard to support our customers' demand during this time, and we expect to grow our DRAM and NAND bit shipments approximately 20% in calendar 2026. - Despite significant efforts, we are disappointed to be unable to meet demand from our customers, across all market segments. - We expect 2025 NAND bit demand growth to be in the high-teens percentage range versus low to mid-teens previously.
    0 [images/image-14-Semiconductor Cleanroom.jpg]
    natural_image Interior of a high-tech semiconductor fabrication facility with workers in full cleanroom suits walking along a central aisle (no visible text or symbols)
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Supply efforts (1 of 2)", + "path": "micron-530bd7ed.pdf/Financial results/Supply efforts (1 of 2)", + "level": 2, + "summary": "- Micron plans to increase our fiscal 2026 capex to approximately \\$20 billion, versus our prior estimate of \\$18 billion. This increase will primarily support our HBM supply capability, and also our 1-gamma supply, in calendar 2026. - We are pulling in equipment orders and accelerating installation timelines to maximize output capability. Micron is also investing across our global manufacturing footprint to add supply to support longer-term demand. We are seeing an enthusiastic customer response to our planned U.S. supply. - We are pulling in our first Idaho fab timeline, and we now expect first wafer output in mid-calendar 2027, earlier than our prior expectations of second-half calendar 2027. Earlier this year, we announced our plans for a second Idaho fab, which will begin construction in 2026 and be operational by the end of 2028. - We are making good progress on securing necessary permits for our New York site and appreciate the partnership with the state of New York and the Trump administration. - We plan to break ground on our first New York fab in early calendar 2026, which we expect will provide supply in 2030 and beyond. - Micron plans to increase our fiscal 2026 capex to approximately \\$20 billion, versus our prior estimate of \\$18 billion. This increase will primarily support our HBM supply capability, and also our 1-gamma supply, in calendar 2026.
    0 [images/image-15-Laboratory Safety Work.jpg]
    text_image December 17, 2025
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Supply efforts (2 of 2)", + "path": "micron-530bd7ed.pdf/Financial results/Supply efforts (2 of 2)", + "level": 2, + "summary": "- In Japan, with the support of METI (Japan’s Ministry of Economy, Trade and Industry), we are making technology and manufacturing investments. • We are enabling future DRAM technology transitions in coordination with our Boise R&D team. - We are also adding cleanroom space in our Hiroshima fab to support these advanced nodes, which will increase production scale and optimize fab economics. - In Singapore, our HBM advanced packaging facility is on track to contribute meaningfully to our HBM supply in calendar 2027. As HBM becomes a part of our Singapore manufacturing footprint, we expect opportunities for synergies between NAND and DRAM production. • We are pleased with the progress on our assembly and test facility in India, which has initiated pilot production and will ramp in 2026. - As we make progress on our strategic manufacturing initiatives, we will continue to be responsive to the market environment and disciplined with our capex plans", + "chunk_count": 1, + "children": [] + }, + { + "title": "Mark Murphy", + "path": "micron-530bd7ed.pdf/Financial results/Mark Murphy", + "level": 2, + "summary": "Chief Financial Officer - In Singapore, our HBM advanced packaging facility is on track to contribute meaningfully to our HBM supply in calendar 2027. As HBM becomes a part of our Singapore manufacturing footprint, we expect opportunities for synergies between NAND and DRAM production.
    0 [images/image-16-Corporate Executive Portrait.jpg]
    natural_image Portrait of a man in a gray suit against a black background, no visible text or symbols on the subject or background.
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "FQ1-26 revenue \\$13.6B", + "path": "micron-530bd7ed.pdf/Financial results/FQ1-26 revenue \\$13.6B", + "level": 2, + "summary": "\\$13.6B Revenue up 21% Q/Q and up 57% Y/Y", + "chunk_count": 1, + "children": [] + }, + { + "title": "Performance by technology", + "path": "micron-530bd7ed.pdf/Financial results/Performance by technology", + "level": 2, + "summary": "", + "chunk_count": 3, + "children": [ + { + "title": "DRAM FQ1-26", + "path": "micron-530bd7ed.pdf/Financial results/Performance by technology/DRAM FQ1-26", + "level": 3, + "summary": "- \\$10.8 billion, representing 79% of total revenue in FQ1-26 • Revenue increased 69% Y/Y • Revenue increased 20% Q/Q - Bit shipments up slightly Q/Q • ASPs increased approximately 20% Q/Q", + "chunk_count": 1, + "children": [] + }, + { + "title": "NAND FQ1-26", + "path": "micron-530bd7ed.pdf/Financial results/Performance by technology/NAND FQ1-26", + "level": 3, + "summary": "- \\$2.7 billion, representing 20% of total revenue in FQ1-26 • Revenue increased 22% Y/Y • Revenue increased 22% Q/Q - Bit shipments increased in mid-to-high single digit percentage range Q/Q - ASPs increased in the mid-teens percentage range Q/Q Quarterly business unit financial results [tables/table-0 Segment Financials.html] • Revenue increased 22% Q/Q
    0 [images/image-17-Abstract Digital Art.jpg]
    natural_image Abstract digital artwork with glowing dots and bokeh background, no readable text or symbols
    ", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Financial performance by business unit (1 of 2)", + "path": "micron-530bd7ed.pdf/Financial results/Financial performance by business unit (1 of 2)", + "level": 2, + "summary": "- Cloud Memory Business Unit revenue was a record \\$5.3 billion and represented 39% of total company revenue. - CMBU revenue was up 16% sequentially, driven by an increase in bit shipments and higher prices - CMBU gross margins were 66%, higher by 620 basis points sequentially, supported by cost execution and higher pricing. - Core Data Center Business Unit revenue was a record \\$2.4 billion and represented 17% of total company revenue. - CDBU revenue was up 51% sequentially, driven by robust bit shipments and higher pricing. - CDBU gross margins were 51%, up 990 basis points sequentially, supported by higher pricing and cost execution. - CMBU revenue was up 16% sequentially, driven by an increase in bit shipments and higher prices
    0 [images/image-18-Abstract Digital Art.jpg]
    natural_image Abstract digital network visualization with glowing nodes and green-to-blue bokeh background (no text or symbols)
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Financial performance by business unit (2 of 2)", + "path": "micron-530bd7ed.pdf/Financial results/Financial performance by business unit (2 of 2)", + "level": 2, + "summary": "- Mobile Client Business Unit revenue was a record \\$4.3 billion and represented 31% of total company revenue. - MCBU revenue was up 13% sequentially, driven by higher pricing, partly offset by lower bit shipments. - MCBU gross margins were 54%, up 17 percentage points sequentially, driven primarily by higher pricing. - Automotive and Embedded Business Unit revenue was a record \\$1.7 billion and represented 13% of total company revenue. - AEBU revenue was up 20% sequentially, driven by higher bit shipments and higher pricing. - AEBU gross margins were 45%, up 14 percentage points sequentially, driven primarily by higher pricing.", + "chunk_count": 1, + "children": [] + }, + { + "title": "FQ1-26 Non-GAAP operating results", + "path": "micron-530bd7ed.pdf/Financial results/FQ1-26 Non-GAAP operating results", + "level": 2, + "summary": "Non-GAAP operating results Revenue: \\$13.64 billion Gross margin: 56.8% Operating expenses: \\$1.33 billion Operating income: \\$6.42 billion Net income: \\$5.48 billion Diluted earnings per share: \\$4.78 Cash from operations (GAAP): \\$8.41 billion See non-GAAP reconciliations in Appendix Operating income: \\$6.42 billion
    0 [images/image-19-Micron Presentation Slide.jpg]
    natural_image Abstract digital artwork featuring glowing purple and blue water droplets with bokeh bokeh effects (no text or symbols)
    ", + "chunk_count": 1, + "children": [] + }, + { + "title": "Cash flow and capital allocation From FY-23 to FQ1-26", + "path": "micron-530bd7ed.pdf/Financial results/Cash flow and capital allocation From FY-23 to FQ1-26", + "level": 2, + "summary": "From FY-23 to FQ1-26 - \\$1.0 billion toward repurchasing 13 million shares - \\$1.7 billion towards dividends paid - \\$2.7 billion returned to shareholders from share repurchases and dividends [tables/table-1 FQ1-26 Cash Flow.html] $^{1}$ Capex net of proceeds from government incentives and proceeds from sales of property, plant, and equipment. $^{2}$ Cash, short-term and long-term marketable investments, restricted cash, and undrawn revolver capacity. \\*Adjusted free cash flow is a non-GAAP measure defined as net cash provided by operating activities less investments in capital expenditures net of proceeds from government incentives and proceeds from sales of property, plant, and equipment. See non-GAAP reconciliations in Appendix.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Al use at Micron", + "path": "micron-530bd7ed.pdf/Financial results/Al use at Micron", + "level": 2, + "summary": "• Over 80% of our professional workforce actively uses GenAI, with total usage up tenfold since last year. - In manufacturing, integrating AI into yield and quality management has cut root cause identification time by half in cases. - Our coding teams are realizing productivity gains of 30% or more using agentic AI. - In R&D, GenAI is accelerating development by reducing cycle times in design verification, product validation, issue triage and root cause analysis. - Across business functions, GenAI is broadening automation opportunities, and we are deploying conversational analytics to accelerate and improve decision-making. - We expect Micron’s use of AI across the enterprise to further strengthen our competitiveness in the coming years.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "Outlook", + "path": "micron-530bd7ed.pdf/Outlook", + "level": 1, + "summary": "- We expect higher price, lower cost and favorable mix to all contribute to gross margin expansion in Q2. - Operating expenses for fiscal Q2 are projected to be approximately \\$1.38 billion. - As mentioned last quarter, Micron’s fiscal Q4 2026 opex will also reflect the effect of an additional workweek in this 53-week fiscal year. • We expect a fiscal Q2 and fiscal year 2026 tax rate of around 15.5%. - Micron is investing in a disciplined manner across our global manufacturing footprint to better meet demand. - To address tight supply/demand conditions extending beyond 2026, we now project our capital spending in fiscal 2026 to be approximately \\$20 billion, weighted to the second half of the fiscal year. - We expect free cash flow to strengthen in fiscal Q2, and we expect to generate significantly higher free cash flow year over year in fiscal 2026. - Any impacts that may occur due to potential new tariffs are not included in our guidance. • We expect a fiscal Q2 and fiscal year 2026 tax rate of around 15.5%.
    0 [images/image-20-Silicon Wafer Close-up.jpg]
    natural_image Close-up of a semiconductor wafer with visible green and blue grating pattern against dark background (no text or symbols)
    FQ2-26 guidance Non-GAAP [tables/table-2 Financial Data.html] \\*Based on \\~1.15 billion diluted shares. See non-GAAP reconciliations in Appendix.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Appendix", + "path": "micron-530bd7ed.pdf/Appendix", + "level": 1, + "summary": "", + "chunk_count": 12, + "children": [ + { + "title": "Financial summary", + "path": "micron-530bd7ed.pdf/Appendix/Financial summary", + "level": 2, + "summary": "Non-GAAP [tables/table-3 Financial Results.html] See non-GAAP reconciliations.", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-GAAP financial data and guidance", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP financial data and guidance", + "level": 2, + "summary": "[tables/table-4 Revenue Breakdown.html] [tables/table-5 Sales Volume Change.html] [tables/table-6 ASP Changes Q_Q.html] [tables/table-7 Non-GAAP Guidance.html] [tables/table-8 Non-GAAP Estimates.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Revenue by technology", + "path": "micron-530bd7ed.pdf/Appendix/Revenue by technology", + "level": 2, + "summary": "[tables/table-9 Memory Revenue.html] Percentages of total revenue may not total 100% due to rounding. Revenue by technology [tables/table-10 Memory Revenue.html] Revenue by business unit [tables/table-11 Revenue by Segment.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-GAAP reconciliations", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations", + "level": 2, + "summary": "Consolidated results", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-GAAP reconciliations_2", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_2", + "level": 2, + "summary": "[tables/table-12 Gross Margin & OpEx.html] Consolidated results", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-GAAP reconciliations_3", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_3", + "level": 2, + "summary": "[tables/table-13 Non-GAAP Financials.html] Consolidated results", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-GAAP reconciliations_4", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_4", + "level": 2, + "summary": "[tables/table-14 Non-GAAP Net Income.html] Consolidated results", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-GAAP reconciliations_5", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_5", + "level": 2, + "summary": "[tables/table-15 Adjusted EBITDA.html] Consolidated results", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-GAAP reconciliations_6", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_6", + "level": 2, + "summary": "[tables/table-16 Financial Highlights.html] Consolidated results", + "chunk_count": 1, + "children": [] + }, + { + "title": "Non-GAAP reconciliations_7", + "path": "micron-530bd7ed.pdf/Appendix/Non-GAAP reconciliations_7", + "level": 2, + "summary": "[tables/table-17 Cash Position.html]", + "chunk_count": 1, + "children": [] + }, + { + "title": "FQ2-26 guidance Non-GAAP reconciliations", + "path": "micron-530bd7ed.pdf/Appendix/FQ2-26 guidance Non-GAAP reconciliations", + "level": 2, + "summary": "Non-GAAP reconciliations [tables/table-18 GAAP vs Non-GAAP.html] Non-GAAP Adjustments (amounts in millions) [tables/table-19 Stock Compensation.html] \\*GAAP earnings per share based on approximately 1.14 billion diluted shares and non-GAAP earnings per share based on approximately 1.15 billion diluted shares. The above guidance does not incorporate the impact of any potential business combinations, divestitures, additional restructuring activities, balance sheet valuation adjustments, strategic investments, financing transactions, and other significant transactions. The timing and impact of such items are dependent on future events that may be uncertain or outside of our control.", + "chunk_count": 1, + "children": [] + } + ] + }, + { + "title": "micron", + "path": "micron-530bd7ed.pdf/micron", + "level": 1, + "summary": "Intelligence Accelerated™ © 2025 Micron Technology, Inc. All rights reserved. Information, products, and/or specifications are subject to change without notice. All information is provided on an “AS IS” basis without warranties of any kind. Statements regarding products, including statements regarding product features, availability, functionality, or compatibility, are provided for informational purposes only and do not modify the warranty, if any, applicable to any product. Drawings may not be to scale. Micron, the Micron logo, the M logo, Intelligence Accelerated™, and other Micron trademarks are the property of Micron Technology, Inc. All other trademarks are the property of their respective owners.", + "chunk_count": 1, + "children": [] + } + ], + "resources": { + "images": [ + { + "path": "images/image-1-Abstract Grid Pattern.jpg", + "summary": "image-1 A close-up view of a circular object featuring a precise grid pattern. The image displays vibrant, iridescent colors including magenta, cyan, green, and blue, which shift across the surface. The grid consists of intersecting horizontal and vertical lines creating small rectangular cells. The lighting creates a gradient effect, with the left side appearing darker and the right side brighter." + }, + { + "path": "images/image-2-Micron Executive Portrait.jpg", + "summary": "image-2 - Subject: Professional headshot of a middle-aged man with graying hair and beard, smiling at the camera. - Attire: Wearing a textured blue blazer over a patterned light purple button-down shirt. - Background: Solid black studio background. - Visible Text/Logos: \"micron\" logo in the bottom right corner; page number \"3\"." + }, + { + "path": "images/image-3-Automated RAM Testing Machine.jpg", + "summary": "image-3 - Industrial automation equipment featuring a vertical array of mechanical grippers or probes. - Multiple green RAM memory modules (SODIMM type) are held in place by the machine's fixtures. - The machinery is illuminated with purple and red lighting, highlighting the precision components. - A conveyor system with metal rollers is visible at the bottom for transporting parts. - Date stamp \"December 17, 2025\" is visible at the bottom left." + }, + { + "path": "images/image-4-Automated Industrial Machinery.jpg", + "summary": "image-4 - Primary Subject: Close-up view of an automated industrial machine, likely a conveyor or sorting system. - Key Features: - Central rotating carousel mechanism with metallic arms and platforms. - Transparent safety shielding (likely polycarbonate) surrounding the machinery. - Vertical LED light strips visible in the background. - Warning labels and sensors attached to the central column. - Context: The scene is illuminated with strong blue lighting, suggesting a clean room or high-tech manufacturing environment. - Visible Text: \"December 17, 2025\" located at the bottom left corner." + }, + { + "path": "images/image-5-Cleanroom Manufacturing Floor.jpg", + "summary": "image-5 - Workers wearing full-body white protective suits (bunny suits), hoods, and face masks. - Environment appears to be a high-tech manufacturing facility or cleanroom with specialized workstations. - Equipment includes computer monitors, keyboards, and various tools on metal tables. - A yellow label with the letters \"DE\" is visible on the back of an office chair in the foreground. - The floor has a textured, grid-like pattern typical of industrial settings." + }, + { + "path": "images/image-6-Micron Semiconductor Wafer.jpg", + "summary": "image-6 - Subject: Close-up photograph of a silicon semiconductor wafer featuring a grid pattern of microchips. - Visual Details: The image displays iridescent colors (red, orange, green, blue) caused by light interference on the wafer's surface layers. - Text/Logo: \"micron\" logo visible in the bottom right corner. - Page Number: \"8\" located next to the logo." + }, + { + "path": "images/image-7-Server Rack Infrastructure.jpg", + "summary": "image-7 - Multiple server racks with perforated mesh doors - Green and orange status indicator lights visible on equipment - Raised floor tiles typical of data center environments - Cable management tray visible at the top left - \"micron\" logo and page number \"9\" at the bottom" + }, + { + "path": "images/image-8-Micron Data Center.jpg", + "summary": "image-8 - Visual: A futuristic server room or data center aisle featuring rows of glass-fronted server racks. The scene is illuminated by vibrant pink and blue neon lighting, with glowing light strips running along the floor and ceiling infrastructure. - Text visible at bottom right: \"micron\" logo and page number \"10\"." + }, + { + "path": "images/image-9-Laptop Keyboard and Screen.jpg", + "summary": "image-9 - Close-up view of a laptop keyboard with illuminated keys in purple and pink hues. - The screen displays a colorful abstract wallpaper with wavy lines in blue, white, yellow, and pink. - Visible text on the keyboard includes standard QWERTY layout characters and function keys like \"Ctrl\", \"Page Down\", \"Enter\". - Bottom right corner shows the logo \"micron\" and page number \"12\". - The image appears to be from a presentation or promotional material showcasing technology products." + }, + { + "path": "images/image-10-Concert Photography.jpg", + "summary": "image-10 - Primary subject: A pair of hands holding a smartphone horizontally to capture a live concert performance. - Scene details: The phone screen displays a stage illuminated by bright spotlights and laser beams in blue and red hues, with the silhouettes of a crowd visible below. - Notable features: The person holding the phone has dark nail polish. The background is blurred with pink lighting. - Visible text: \"micron\" logo and page number \"13\" at the bottom right corner." + }, + { + "path": "images/image-11-Futuristic Car Interior.jpg", + "summary": "image-11 - Steering wheel with integrated controls and central emblem - Dashboard featuring four circular digital instrument clusters with blue glowing displays - Large vertical touchscreen infotainment system displaying vehicle schematics and interface icons - Center console with a secondary tablet device, gear selector knob, and control buttons - Ambient lighting accents in blue and red tones - \"micron\" logo visible at the bottom right corner alongside page number \"14\"" + }, + { + "path": "images/image-12-Business Meeting Discussion.jpg", + "summary": "image-12 - Scene: A group of four professionals gathered around a conference table in a glass-walled meeting room. - Key Action: A woman in a purple top is standing and pointing at documents on the table, appearing to lead the discussion or explain data. - Participants: Three other individuals (two women, one man) are seated and listening attentively. - Background: A large screen displays mathematical graphs or charts with curves plotted on axes. - Visible Text: \"December 17, 2025\" located at the bottom left corner." + }, + { + "path": "images/image-13-Silicon Wafer in Manufacturing.jpg", + "summary": "image-13 - Primary Subject: A close-up photograph of a silicon wafer held within an automated manufacturing or inspection machine. - Visual Details: The circular wafer displays a distinct rainbow iridescence caused by thin-film interference, overlaid with a precise grid pattern representing individual microchips (dies). - Context: The background is blurred, emphasizing the high-tech nature of semiconductor fabrication. - Visible Text: \"December 17, 2025\" located at the bottom left corner." + }, + { + "path": "images/image-14-Semiconductor Cleanroom.jpg", + "summary": "image-14 - Environment: High-tech manufacturing facility (cleanroom) with rows of large, stainless steel industrial machinery. - Personnel: Worker wearing a full-body white protective suit (bunny suit), face mask, and hood walking down the aisle. - Lighting: Distinctive yellow lighting in the background, typical for photoresist processing areas in semiconductor fabrication. - Infrastructure: Raised perforated flooring for air circulation; overhead cable trays and utility lines. - Date Stamp: December 17, 2025" + }, + { + "path": "images/image-15-Laboratory Safety Work.jpg", + "summary": "image-15 - Subject: Scientist or technician wearing full PPE (face shield, safety glasses, protective gown, orange chemical-resistant gloves) working inside a fume hood or glove box. - Environment: Blue-lit laboratory setting with visible ventilation piping on the left wall. - Text/Signage: \"SCRUBBER EXHAUST\" label with directional arrow on vertical pipe; date stamp at bottom reads \"December 17, 2025\". - Equipment: Fume hood/glove box with sliding sash and internal lighting; overhead exhaust fixture visible above work area." + }, + { + "path": "images/image-16-Corporate Executive Portrait.jpg", + "summary": "image-16 Subject: Male executive with short brown hair, smiling at the camera. Attire: Grey suit jacket over a light blue patterned button-down shirt. Background: Solid black studio background. Visible Text/Logo: \"micron\" logo visible in the bottom right corner." + }, + { + "path": "images/image-17-Abstract Digital Art.jpg", + "summary": "image-17 December 17, 2025" + }, + { + "path": "images/image-18-Abstract Digital Art.jpg", + "summary": "image-18 December 17, 2025" + }, + { + "path": "images/image-19-Micron Presentation Slide.jpg", + "summary": "image-19 - Brand: Micron - Page Number: 25 - Visual Content: Abstract digital artwork featuring flowing, translucent waves with glowing blue and purple light trails against a dark background. The image includes bokeh effects and particle-like sparkles, suggesting themes of data flow, technology, or connectivity." + }, + { + "path": "images/image-20-Silicon Wafer Close-up.jpg", + "summary": "image-20 - Subject: Circular silicon wafer with visible grid pattern of microchips - Lighting: Blue-toned illumination with bright reflection on the right edge - Date stamp: December 17, 2025 (bottom left)" + } + ], + "tables": [ + { + "path": "tables/table-0 Segment Financials.html", + "summary": "table-1 The table shows revenue, gross margin, and operating margin for Cloud Memory, Core Data, Mobile and Client, and Automotive and Embedded segments across FQ1-26, FQ4-25, and FQ1-25." + }, + { + "path": "tables/table-1 FQ1-26 Cash Flow.html", + "summary": "table-2 Cash flow from operations $8.4B, net capex $4.5B, adjusted free cash flow $3.9B, buybacks $300M, dividend $0.115/share, liquidity $15.5B." + }, + { + "path": "tables/table-2 Financial Data.html", + "summary": "table-3 Revenue $18.70B ± $400M, gross margin 68.0% ± 1.0%, operating expenses $1.38B ± $20M, diluted EPS $8.42 ± $0.20." + }, + { + "path": "tables/table-3 Financial Results.html", + "summary": "table-4 The table shows quarterly financial data for FQ1-26, FQ4-25, and FQ1-25, with revenue increasing from $8.7B to $13.6B, gross margin rising from 40% to 57%, and net income growing from $2.0B to $5.5B." + }, + { + "path": "tables/table-4 Revenue Breakdown.html", + "summary": "table-5 DRAM accounts for 79% of revenue, while NAND contributes 20% in FQ1-26." + }, + { + "path": "tables/table-5 Sales Volume Change.html", + "summary": "table-6 DRAM sales volume up slightly quarter over quarter; NAND increased in mid-to-high single-digit percentage range." + }, + { + "path": "tables/table-6 ASP Changes Q_Q.html", + "summary": "table-7 DRAM ASP increased approximately 20% Q/Q, while NAND ASP increased in the mid-teens percentage range in FQ1-26." + }, + { + "path": "tables/table-7 Non-GAAP Guidance.html", + "summary": "table-8 The table compares FQ1-26 non-GAAP actuals with FQ2-26 guidance: Revenue $13.6B vs $18.7B, Gross margin 56.8% vs 68.0%, OpEx $1.33B vs $1.38B, Diluted EPS $4.78 vs $8.42." + }, + { + "path": "tables/table-8 Non-GAAP Estimates.html", + "summary": "table-9 Table shows FQ1-26 non-GAAP actuals and FQ2-26 estimates: diluted shares ~1.15B, tax rate ~15.5%, capex ~$20B for fiscal 2026." + }, + { + "path": "tables/table-9 Memory Revenue.html", + "summary": "table-10 The table shows memory revenue breakdown for FQ1-26, FQ4-25, and FQ1-25. DRAM revenue grew from $6,400M to $10,812M, NAND from $2,241M to $2,743M, with total revenue increasing from $8,709M to $13,643M." + }, + { + "path": "tables/table-10 Memory Revenue.html", + "summary": "table-11 Memory revenue for FQ1-26 was $13,643M, up 21% Q/Q and 57% Y/Y. DRAM grew 20% Q/Q and 69% Y/Y to $10,812M. NAND grew 22% Q/Q and 22% Y/Y to $2,743M." + }, + { + "path": "tables/table-11 Revenue by Segment.html", + "summary": "table-12 The table shows revenue for four segments: Cloud Memory ($5,284M), Core Data ($2,379M), Mobile and Client ($4,255M), and Automotive and Embedded ($1,720M) with Q/Q and Y/Y percentage changes." + }, + { + "path": "tables/table-12 Gross Margin & OpEx.html", + "summary": "table-13 Financial data for FQ1-26, FQ4-25, FQ1-25 showing GAAP and Non-GAAP gross margins, operating expenses, and operating income with adjustments for stock-based compensation and other items." + }, + { + "path": "tables/table-13 Non-GAAP Financials.html", + "summary": "table-14 The table compares GAAP and Non-GAAP costs for COGS, R&D, and SG&A across FQ1-26, FQ4-25, and FQ1-25, with stock-based compensation adjustments." + }, + { + "path": "tables/table-14 Non-GAAP Net Income.html", + "summary": "table-15 The table shows GAAP net income and adjustments for stock-based compensation, restructuring, and debt losses, resulting in Non-GAAP net income of $5,482M for FQ1-26, $3,469M for FQ4-25, and $2,037M for FQ1-25." + }, + { + "path": "tables/table-15 Adjusted EBITDA.html", + "summary": "table-16 The table shows GAAP net income and adjustments to calculate Adjusted EBITDA for FQ1-26, FQ4-25, and FQ1-25, with values of $8,604M, $6,066M, and $4,411M respectively." + }, + { + "path": "tables/table-16 Financial Highlights.html", + "summary": "table-17 The table shows GAAP and Non-GAAP diluted EPS, shares, and adjusted free cash flow for FQ1-26, FQ4-25, and FQ1-25. Non-GAAP EPS rose to $4.78 from $1.79, and adjusted free cash flow increased to $3,906M from $112M." + }, + { + "path": "tables/table-17 Cash Position.html", + "summary": "table-18 The table shows cash, investments, and debt for FQ1-26, FQ4-25, and FQ1-25. Net cash improved from negative to positive $260 million in FQ1-26." + }, + { + "path": "tables/table-18 GAAP vs Non-GAAP.html", + "summary": "table-19 The table compares GAAP and Non-GAAP outlook for revenue ($18.70B), gross margin (67% vs 68%), operating expenses ($1.56B vs $1.38B), and diluted EPS ($8.19 vs $8.42) with adjustments." + }, + { + "path": "tables/table-19 Stock Compensation.html", + "summary": "table-20 The table shows stock-based compensation costs: $125 for cost of goods sold, $119 for R&D, $61 for SG&A, and a tax adjustment of ($48), totaling $257." + } + ] + } +} diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/full.md b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/full.md new file mode 100644 index 000000000..784124b1b --- /dev/null +++ b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/full.md @@ -0,0 +1,508 @@ +## Financial results + +FQ1 2026 + +![](images/9149f0769e7169ba814c6e35977694c15e50c546b14fe678bbc5354730bcbe9e.jpg) + +
    +natural_image + +Close-up of a semiconductor wafer with green, purple, and blue photovoltaic bands (no text or symbols visible) +
    + +## Safe harbor statement + +During the course of this meeting, we may make projections or other forward-looking statements regarding market demand and supply, market and pricing trends and drivers, the impact of technologies such as AI, contractual terms, cost reductions, our manufacturing projects and related investments, expected product volume production, our market position, expected product announcements, capabilities of our future products and technologies, future events, such as the end of life of some products, and future financial and operating performance including financial projections of the company and the industry. We wish to caution you that such statements are predictions, and that actual events or results may differ materially. We refer you to the documents the company files from time to time with the Securities and Exchange Commission, including the company's Form 10-K, Forms 10-Q and other reports and filings. These documents contain and identify important factors that could cause the actual results for the company to differ materially from those contained in our projections or forward-looking statements. These certain factors can be found at investors.micron.com/risk-factor. Although we believe that the expectations reflected in the forward-looking statements are reasonable, we cannot guarantee future results, levels of activity, performance or achievements. We are under no duty to update any of the forward-looking statements to conform these statements except as required by applicable law. + +This presentation includes non-GAAP financial measures. Non-GAAP financial measures represent GAAP measures, excluding the impact of certain activities, which management excludes in analyzing our operating results and understanding trends in our earnings, adjusted free cash flow and business outlook. Further information regarding Micron's use of non-GAAP measures and reconciliations between GAAP and non-GAAP measures are included in the Appendix. + +## Sanjay Mehrotra + +Chairman, President and Chief Executive Officer + +![](images/da164305699c631fbb49baf9d21c6e0d322abbd9849fd5125ac085499077e61e.jpg) + +
    +natural_image + +Portrait of a smiling man with gray hair and beard, wearing a blue blazer over a patterned shirt (no text or symbols visible) +
    + +![](images/5dc07700d70cc6d76a9fc6af05c62d73cd547d03dee3ceb571b0f90cfe2ad49e.jpg) + +
    +natural_image + +Industrial automation scene with robotic arms and green memory chips under purple lighting (no readable text or symbols) +
    + +## Overview (1 of 3) + +- Micron had an outstanding start to fiscal 2026, delivering fiscal Q1 revenue, gross margin and EPS (earnings per share) well above the high end of our guidance. +- This financial performance was driven by our strong execution across end markets and products in a tight supply environment. We achieved a number of records in fiscal Q1. +- Total company revenue, DRAM and NAND revenue, as well as HBM and data center revenue and revenue in each of our business units, also reached new records. +• We have completed agreements on price and volume for our entire calendar 2026 HBM supply, including Micron's industry-leading HBM4. +- We forecast an HBM TAM CAGR of approximately 40% through calendar 2028, from approximately \$35 billion in 2025 to around \$100 billion in 2028. +- This \$100 billion HBM TAM milestone is now projected to arrive two years earlier than in our prior outlook. Remarkably, this 2028 HBM TAM projection is larger than the size of the entire DRAM market in calendar 2024. + +![](images/8cad2787caf6fdfe3dec57debe0abb2b14b0aff7ebce3ea225b1d5c09038930a.jpg) + +
    +natural_image + +Interior view of a modern industrial facility with blue-tinted machinery and control panels (no visible text or symbols) +
    + +## Overview (2 of 3) + +- We are excited about our customized HBM4E customer engagements, which offer further differentiation opportunities to us, and we continue to make excellent progress on our HBM roadmap. +- Memory is now essential to AI's cognitive functions, fundamentally altering its role from a system component to a strategic asset that dictates product performance from data center to the edge. +- This structural shift means that system capabilities heavily rely on advanced memory for real-time contextual processing, which is vital for achieving autonomous and intelligent behaviors in AI data centers as well as in applications ranging from self-driving cars to advanced medical diagnostics. +- With our technology leadership, differentiated product portfolio, strong operational execution and solid balance sheet, Micron is in the best competitive position in its history and is one of the semiconductor industry's biggest enablers of AI. +• We anticipate substantial new records in revenue, gross margin, EPS and free cash flow for both the second quarter and the full fiscal year 2026, and we expect our business performance to continue to strengthen through the year. + +![](images/e5290f0446b3472e804799d52b94e550d2933b0f0baff46d44e4df1939101b23.jpg) + +
    +natural_image + +Interior of a cleanroom manufacturing facility with technicians in full protective suits working at workstations (no visible text or symbols) +
    + +## Overview (3 of 3) + +- Sustained and strong industry demand, along with supply constraints, are contributing to tight market conditions, and we expect these conditions to persist beyond calendar 2026. +• We are making progress with customers in our discussions for multiyear contracts with specific commitments. +- Simultaneously, we are focused on maximizing our production output from our current footprint, ramping our industry-leading technology nodes and investing in new cleanroom space to add to our supply capability. + +## Technology and operations (1 of 2) + +- Micron’s technology leadership is foundational to our strong competitive position. +- Micron has led the industry for four consecutive technology nodes in DRAM and three nodes in NAND, with progressively faster yield ramps in every node. +- Our 1-gamma DRAM node is ramping well. 1-gamma will be the primary driver of our DRAM bit growth in calendar 2026, and will be the majority of our bit output in the second half of the calendar year. +- Looking beyond 1-gamma, development is underway for 1-delta and 1-epsilon nodes, which will feature innovations that we expect to extend our differentiation and technology leadership. + +## Technology and operations (2 of 2) + +- In NAND, we are ramping our G9 node with robust yield ramps across both data center and client SSDs. Our QLC NAND mix, including G9 QLC, reached a record high during the quarter. +- Technology transitions to G9 will be the primary driver of our NAND bit growth in calendar 2026, and we expect it to become our largest NAND node later in fiscal 2026. +- Calendar 2025 is a record year for Micron in terms of both internal and customer quality measures, positioning us well to deliver for our customers as the memory industry's quality leader. +- As our products are increasingly integrated into higher value applications, our leadership in quality is becoming a more important differentiator. + +![](images/7b8b2b7f13683937d368150d2fffbd323578d2c6f888174f2d956b92d3ce4a92.jpg) + +
    +text_image + +micron | 8 +
    + +## Data center server demand + +- As the world’s leading technology companies advance toward artificial general intelligence and transform the global economy, our customers are committing to an extraordinary, multiyear data center buildout. +- This growth in AI data center capacity is driving a significant increase in demand for high-performance and high-capacity memory and storage. +- Server unit demand has strengthened significantly, and we now expect calendar 2025 server unit growth in the high-teens percentage range, higher than our last earnings call outlook of $10\%$ . +• We expect server demand strength to continue in 2026. Server memory and storage content and performance requirements continue to increase generation to generation. +- Micron has a differentiated portfolio of high-value data center solutions to address these requirements, including our HBM, high-capacity server memory solutions and data center SSDs. + +![](images/18d6e362707276762027503d802d811052cd28b6522d9a651071cf6a71ae3875.jpg) + +
    +natural_image + +Interior view of a server room with rows of server racks and illuminated green indicator lights, labeled 'micron' in the corner (no readable text on equipment) +
    + +## Data center HBM and LP DRAM + +- Micron’s HBM4, with industry leading speed over 11 Gbps, is on track to ramp with high yields in the second calendar quarter of 2026, consistent with our customers’ product ramp plans. +- Our HBM4 uses advanced CMOS and advanced metallization process technologies on the base logic die and DRAM core dies, which are designed and manufactured in-house. +- This, along with our unique HBM design, packaging and test capability, enables Micron’s industry-leading performance and low-power leadership. +- Micron pioneered the adoption of LP DRAM in the data center. Micron's Low Power (LP) DRAM server modules consume one-third the power of DDR DRAM server modules. +- Building on this leadership, we have sampled our 192GB LP SOCAMM2 product, which enables a $50\%$ increase in capacity per module and a rack-scale LP DRAM density of over 50TB. + +![](images/29a02e83efb4249d46213901c0fdde61f36909469e1d9ecc6def688df1b599ef.jpg) + +
    +natural_image + +Interior view of a modern data center with server racks illuminated by purple and blue neon lights (no visible text or symbols) +
    + +## Data center NAND + +- Our data center NAND portfolio revenue exceeded a billion dollars in fiscal Q1, and we are seeing strong momentum across our data center SSD portfolio, enabled by our leadership NAND technology. +- In the performance SSD category, Micron has introduced the world's first PCIe Gen6 SSD, leveraging our G9 NAND. +• We are seeing rapidly increasing qualification commitments for this product, including at hyperscalers. +- In mainstream storage, our SSDs based on G9 NAND are already seeing robust demand in the first quarter of calendar 2026. +- In capacity storage, our QLC-based 122 and 245TB G9 SSDs are entering qualification at multiple hyperscale customers. + +## PC + +- PC demand continues to be driven by Windows 10 end of life and AI PCs. +• We forecast PC unit sales to grow high-single-digit percentage range in calendar 2025, above our prior expectations provided in our last earnings call of mid-single digits. +- As we look ahead into 2026, we expect these demand drivers to continue, while memory supply constraints may affect some PC unit shipments. +- Micron has completed multiple OEM qualifications of our 16Gb 1-gamma-based DDR5 and our G9-based PCIe Gen4 QLC SSDs. + +![](images/8a8b1d656c030897717c91a4b91c2a5de9706934c1cc5c2bb3101dacb6fd3d20.jpg) + +
    +text_image + +micron | 12 +
    + +## Mobile + +- Smartphone unit volumes in calendar 2025 are on track to grow in the low-single-digit percentage range. +- AI is driving memory content growth. The shipment mix of flagship smartphones with 12GB of DRAM increased to $59\%$ in calendar Q3, more than twice the level from a year ago. +• Micron is accelerating innovation across our mobile DRAM portfolio. +- In fiscal Q1, we began sampling our breakthrough 1-gamma 16Gb LPDDR6 product to leading OEM and ecosystem partners, marking a major milestone in next-generation memory technology. +- LPDDR6 will power AI at the edge, delivering over 50% higher performance and improved power efficiency for flagship smartphones and AI PCs. +- Micron also sampled our 1-gamma LP5x 24Gb product and began volume shipments of the previously announced 1-gamma LP5x 16Gb product to multiple OEMs. + +![](images/76865e98093d3cc2b9f2c874cf5299e7f2a3057f1a977fa2a1edaf67e41ba490.jpg) + +
    +natural_image + +Hand holding smartphone displaying a live music show with stage lights against a red and pink background (no text or symbols on screen) +
    + +## Automotive and embedded + +- In automotive, L2+ and L3 adoption is driving robust demand today, and our customers' roadmaps indicate a significantly higher memory content in fully automated vehicles. +- Micron is uniquely positioned for growth with our differentiated product portfolio and automotive market share leadership. +- Our ASIL-rated LPDDR5x and UFS4.1 NAND products, optimized for automotive and advanced robotics that include bandwidth-enhancing features, are seeing strong demand and have already secured billions of dollars in design wins. +- In industrial, demand continues to strengthen, driven by the growing adoption of autonomous systems across various applications. +- Long-term demand trend trajectory remains robust for memory and storage in industrial applications such as in factory automation, aerospace and defense, humanoid robotics, edge networking and video surveillance. +- Across both auto and industrial markets, LPDDR4x and DDR4 are also experiencing strong demand, and we are making investments to provide long-term supply from our Manassas, Virginia fab. + +![](images/58776b148ff840f505ca6617afb6ee6c96a8d0845588253a7c049924429fe46f.jpg) + +
    +natural_image + +Interior view of a modern car dashboard with digital dashboard and smart lock display (no readable text or symbols) +
    + +![](images/282889fa272df0c1f861110f156f442d7e0ce8261ae7462ed77f4d6f252184a9.jpg) + +
    +natural_image + +Three business professionals in a meeting setting with a large projection screen showing mathematical graphs (no readable text or symbols) +
    + +## Market outlook (1 of 2) + +• Over the last few months, our customers' AI data center build-out plans have driven a sharp increase in demand forecasts for memory and storage. +• We believe that the aggregate industry supply will remain substantially short of the demand for the foreseeable future. +- The dramatic increase in HBM demand is further challenging the supply environment due to the 3-to-1 trade ratio with DDR5, and this trade ratio only increases with future generations of HBM. +- Additional cleanroom space is necessary to address this increased demand, and lead times for cleanroom build-out are lengthening across geographies. +• Together, these demand and supply factors are driving tight industry conditions across DRAM and NAND, and we expect tightness to persist through and beyond calendar 2026. + +![](images/e9deedd416499605a104d596731dc47625c0ab441368681d3fe84d8da560cbb9.jpg) + +
    +natural_image + +Close-up of a semiconductor wafer with colorful grid patterns, captured in a lab setting (no text or symbols on the wafer itself) +
    + +## Market outlook (2 of 2) + +- Calendar 2025 DRAM and NAND industry bit demand growth expectations are higher than in our last earnings call outlook. +- We now expect calendar 2025 DRAM bit demand growth in the low $20\%$ range versus high teens previously. +- We expect 2025 NAND bit demand growth to be in the high-teens percentage range versus low to mid-teens previously. +- We expect calendar 2026 industry DRAM and NAND bit shipment growth to be constrained by industry supply. +- We expect both DRAM and NAND calendar 2026 industry bit shipments to increase around 20% from 2025 levels. +- Micron is working hard to support our customers' demand during this time, and we expect to grow our DRAM and NAND bit shipments approximately 20% in calendar 2026. +- Despite significant efforts, we are disappointed to be unable to meet demand from our customers, across all market segments. + +![](images/09546d03a83253b90d99e5ae02b5f5fe14965c280a22729ad67bb30a5e5321bd.jpg) + +
    +natural_image + +Interior of a high-tech semiconductor fabrication facility with workers in full cleanroom suits walking along a central aisle (no visible text or symbols) +
    + +## Supply efforts (1 of 2) + +- Micron plans to increase our fiscal 2026 capex to approximately \$20 billion, versus our prior estimate of \$18 billion. This increase will primarily support our HBM supply capability, and also our 1-gamma supply, in calendar 2026. +- We are pulling in equipment orders and accelerating installation timelines to maximize output capability. Micron is also investing across our global manufacturing footprint to add supply to support longer-term demand. We are seeing an enthusiastic customer response to our planned U.S. supply. +- We are pulling in our first Idaho fab timeline, and we now expect first wafer output in mid-calendar 2027, earlier than our prior expectations of second-half calendar 2027. Earlier this year, we announced our plans for a second Idaho fab, which will begin construction in 2026 and be operational by the end of 2028. +- We are making good progress on securing necessary permits for our New York site and appreciate the partnership with the state of New York and the Trump administration. +- We plan to break ground on our first New York fab in early calendar 2026, which we expect will provide supply in 2030 and beyond. + +![](images/b27956e067cec547493a491130022fb6391200a0574560ad41081c3c80017751.jpg) + +
    +text_image + +December 17, 2025 +
    + +## Supply efforts (2 of 2) + +- In Japan, with the support of METI (Japan’s Ministry of Economy, Trade and Industry), we are making technology and manufacturing investments. +• We are enabling future DRAM technology transitions in coordination with our Boise R&D team. +- We are also adding cleanroom space in our Hiroshima fab to support these advanced nodes, which will increase production scale and optimize fab economics. +- In Singapore, our HBM advanced packaging facility is on track to contribute meaningfully to our HBM supply in calendar 2027. As HBM becomes a part of our Singapore manufacturing footprint, we expect opportunities for synergies between NAND and DRAM production. +• We are pleased with the progress on our assembly and test facility in India, which has initiated pilot production and will ramp in 2026. +- As we make progress on our strategic manufacturing initiatives, we will continue to be responsive to the market environment and disciplined with our capex plans + +## Mark Murphy + +Chief Financial Officer + +![](images/a8a560d16f279da483d986445f71ccbe0b809a852d605817cf42e1efb1b5b1e6.jpg) + +
    +natural_image + +Portrait of a man in a gray suit against a black background, no visible text or symbols on the subject or background. +
    + +## FQ1-26 revenue + +## \$13.6B + +Revenue up 21% Q/Q + +and up 57% Y/Y + +## Performance by technology + +## DRAM FQ1-26 + +- \$10.8 billion, representing 79% of total revenue in FQ1-26 +• Revenue increased 69% Y/Y +• Revenue increased 20% Q/Q +- Bit shipments up slightly Q/Q +• ASPs increased approximately 20% Q/Q + +## NAND FQ1-26 + +- \$2.7 billion, representing 20% of total revenue in FQ1-26 +• Revenue increased 22% Y/Y +• Revenue increased 22% Q/Q +- Bit shipments increased in mid-to-high single digit percentage range Q/Q +- ASPs increased in the mid-teens percentage range Q/Q + +Quarterly business unit financial results + +
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    Cloud Memory (CMBU)
    Revenue$5,284$4,543$2,648
    Gross margin66 %59 %51 %
    Operating margin55 %48 %40 %
    Core Data (CDBU)
    Revenue$2,379$1,577$2,292
    Gross margin51 %41 %50 %
    Operating margin37 %25 %38 %
    Mobile and Client (MCBU)
    Revenue$4,255$3,760$2,608
    Gross margin54 %36 %27 %
    Operating margin47 %29 %15 %
    Automotive and Embedded (AEBU)
    Revenue$1,720$1,434$1,158
    Gross margin45 %31 %20 %
    Operating margin36 %20 %7 %
    + +![](images/5f6d2a4de9f2c14460c891feeffab121e44b28cc2c732ae8893afcd0b311b767.jpg) + +
    +natural_image + +Abstract digital artwork with glowing dots and bokeh background, no readable text or symbols +
    + +# Financial performance by business unit (1 of 2) + +- Cloud Memory Business Unit revenue was a record \$5.3 billion and represented 39% of total company revenue. +- CMBU revenue was up 16% sequentially, driven by an increase in bit shipments and higher prices +- CMBU gross margins were 66%, higher by 620 basis points sequentially, supported by cost execution and higher pricing. +- Core Data Center Business Unit revenue was a record \$2.4 billion and represented 17% of total company revenue. +- CDBU revenue was up 51% sequentially, driven by robust bit shipments and higher pricing. +- CDBU gross margins were 51%, up 990 basis points sequentially, supported by higher pricing and cost execution. + +![](images/db8ea56c06877c03e5a90cdfba90df6c55818f146e02344ff7f333b92d154e96.jpg) + +
    +natural_image + +Abstract digital network visualization with glowing nodes and green-to-blue bokeh background (no text or symbols) +
    + +# Financial performance by business unit (2 of 2) + +- Mobile Client Business Unit revenue was a record \$4.3 billion and represented 31% of total company revenue. +- MCBU revenue was up 13% sequentially, driven by higher pricing, partly offset by lower bit shipments. +- MCBU gross margins were 54%, up 17 percentage points sequentially, driven primarily by higher pricing. +- Automotive and Embedded Business Unit revenue was a record \$1.7 billion and represented 13% of total company revenue. +- AEBU revenue was up 20% sequentially, driven by higher bit shipments and higher pricing. +- AEBU gross margins were 45%, up 14 percentage points sequentially, driven primarily by higher pricing. + +## FQ1-26 + +## Non-GAAP operating results + +Revenue: \$13.64 billion + +Gross margin: 56.8% + +Operating expenses: \$1.33 billion + +Operating income: \$6.42 billion + +Net income: \$5.48 billion + +Diluted earnings per share: \$4.78 + +Cash from operations (GAAP): \$8.41 billion + +See non-GAAP reconciliations in Appendix + +![](images/f0887e354b2240d8c53d17166c3af7ea7433702016b891243ca661308d2168bb.jpg) + +
    +natural_image + +Abstract digital artwork featuring glowing purple and blue water droplets with bokeh bokeh effects (no text or symbols) +
    + +# Cash flow and capital allocation + +## From FY-23 to FQ1-26 + +- \$1.0 billion toward repurchasing 13 million shares +- \$1.7 billion towards dividends paid +- \$2.7 billion returned to shareholders from share repurchases and dividends + +
    Cash flow from operationsFQ1-26: $8.4 billion (62% of revenue)
    Net Capex1FQ1-26: $4.5 billion
    Adjusted free cash flow*FQ1-26: $3.9 billion
    BuybacksFQ1-26: $300 million
    DividendsDividend of $0.115 per share will be paid on January 14th
    Liquidity2$15.5 billion in liquidity at end of FQ1-26
    + +$^{1}$ Capex net of proceeds from government incentives and proceeds from sales of property, plant, and equipment. +$^{2}$ Cash, short-term and long-term marketable investments, restricted cash, and undrawn revolver capacity. +\*Adjusted free cash flow is a non-GAAP measure defined as net cash provided by operating activities less investments in capital expenditures net of proceeds from government incentives and proceeds from sales of property, plant, and equipment. +See non-GAAP reconciliations in Appendix. + +## Al use at Micron + +• Over 80% of our professional workforce actively uses GenAI, with total usage up tenfold since last year. +- In manufacturing, integrating AI into yield and quality management has cut root cause identification time by half in cases. +- Our coding teams are realizing productivity gains of 30% or more using agentic AI. + +- In R&D, GenAI is accelerating development by reducing cycle times in design verification, product validation, issue triage and root cause analysis. +- Across business functions, GenAI is broadening automation opportunities, and we are deploying conversational analytics to accelerate and improve decision-making. +- We expect Micron’s use of AI across the enterprise to further strengthen our competitiveness in the coming years. + +## Outlook + +- We expect higher price, lower cost and favorable mix to all contribute to gross margin expansion in Q2. +- Operating expenses for fiscal Q2 are projected to be approximately \$1.38 billion. +- As mentioned last quarter, Micron’s fiscal Q4 2026 opex will also reflect the effect of an additional workweek in this 53-week fiscal year. +• We expect a fiscal Q2 and fiscal year 2026 tax rate of around 15.5%. + +- Micron is investing in a disciplined manner across our global manufacturing footprint to better meet demand. +- To address tight supply/demand conditions extending beyond 2026, we now project our capital spending in fiscal 2026 to be approximately \$20 billion, weighted to the second half of the fiscal year. +- We expect free cash flow to strengthen in fiscal Q2, and we expect to generate significantly higher free cash flow year over year in fiscal 2026. +- Any impacts that may occur due to potential new tariffs are not included in our guidance. + +![](images/45c7455abd007b8fb2ce52e6956089d9662f19da777acb798b1120529697479f.jpg) + +
    +natural_image + +Close-up of a semiconductor wafer with visible green and blue grating pattern against dark background (no text or symbols) +
    + +## FQ2-26 guidance + +## Non-GAAP + +
    Revenue$18.70 billion ± $400 million
    Gross margin68.0% ± 1.0%
    Operating expenses$1.38 billion ± $20 million
    Diluted earnings per share*$8.42 ± $0.20
    + +\*Based on \~1.15 billion diluted shares. + +See non-GAAP reconciliations in Appendix. + +## Appendix + +## Financial summary + +Non-GAAP + +
    Amounts in millions, except per shareFQ1-26% of revenueFQ4-25% of revenueFQ1-25% of revenue
    Revenue$13,643100%$11,315100%$8,709100%
    Gross margin7,75357%5,16946%3,44140%
    Operating income6,41947%3,95535%2,39427%
    Income tax (provision) benefit(977)(471)(333)
    Net income5,48240%3,46931%2,03723%
    Diluted earnings per share4.783.031.79
    Cash provided by operating activities (GAAP)8,4115,7303,244
    Cash, marketable investments, and restricted cash (GAAP)12,01611,9408,748
    + +See non-GAAP reconciliations. + +## Non-GAAP financial data and guidance + +
    % of RevenueFQ1-26
    DRAM79%
    NAND20%
    + +
    % Sales Volume ChangeFQ1-26 Q/Q
    DRAMUp slightly
    NANDIncreased in mid-to-high single-digit percentage range
    + +
    % ASP ChangeFQ1-26 Q/Q
    DRAMIncreased approximately 20%
    NANDIncreased in the mid-teens percentage range
    + +
    FQ1-26 non-GAAP (amounts in millions, except per share)FQ2-26 non-GAAP Guidance
    Revenue$13,643$18.70 billion ± $400 million
    Gross margin56.8%68.0% ± 1.0%
    Operating expenses$1,334$1.38 billion ± $20 million
    Diluted earnings per share$4.78$8.42 ± $0.20
    + +
    FQ1-26 non-GAAP (amounts in millions)FQ2-26 non-GAAP Estimates
    Diluted shares1,148~1.15 billion
    Income tax (provision) benefit($977)Around 15.5%
    Cash from operations (GAAP)$8,411
    Investments in capex, net (capital cash flow)$4,505~ $20 billion in fiscal 2026
    + +## Revenue by technology + +
    Amounts in millionsFQ1-26% of revenueFQ4-25% of revenueFQ1-25% of revenue
    DRAM$10,81279%$8,98479%$6,40073%
    NAND2,74320%2,25220%2,24126%
    Other (primarily NOR)881%791%681%
    Total$13,643100%$11,315100%$8,709100%
    + +Percentages of total revenue may not total 100% due to rounding. + +Revenue by technology + +
    Amounts in millionsFQ1-26FQ4-25Q/Q % ChangeFQ1-25Y/Y % Change
    DRAM$10,812$8,98420%$6,40069%
    NAND2,7432,25222%2,24122%
    Other (primarily NOR)887911%6829%
    Total$13,643$11,31521%$8,70957%
    + +Revenue by business unit + +
    Amounts in millionsFQ1-26FQ4-25Q/Q % ChangeFQ1-25Y/Y % Change
    Cloud Memory (CMBU)$5,284$4,54316%$2,648100%
    Core Data (CDBU)2,3791,57751%2,2924%
    Mobile and Client (MCBU)4,2553,76013%2,60863%
    Automotive and Embedded (AEBU)1,7201,43420%1,15849%
    + +## Non-GAAP reconciliations + +## Consolidated results + +## Non-GAAP reconciliations + +
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    GAAP gross margin$7,646$5,054$3,348
    Stock-based compensation10711590
    Other3
    Non-GAAP gross margin$7,753$5,169$3,441
    GAAP operating expenses$1,510$1,400$1,174
    Stock-based compensation(173)(147)(127)
    Restructure and asset impairments(38)
    Other(3)(1)
    Non-GAAP operating expenses$1,334$1,214$1,047
    GAAP operating income$6,136$3,654$2,174
    Stock-based compensation280262217
    Restructure and asset impairments38
    Other313
    Non-GAAP operating income$6,419$3,955$2,394
    + +## Consolidated results + +## Non-GAAP reconciliations + +
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    GAAP cost of goods sold$5,997$6,261$5,361
    Stock-based compensation(107)(115)(90)
    Other(3)
    Non-GAAP cost of goods sold$5,890$6,146$5,268
    GAAP research and development$1,171$1,047$888
    Stock-based compensation(106)(93)(77)
    Other(1)(1)
    Non-GAAP research and development$1,064$953$811
    GAAP selling, general, and administrative$337$314$288
    Stock-based compensation(67)(54)(50)
    Non-GAAP selling, general, and administrative$270$260$238
    + +## Consolidated results + +## Non-GAAP reconciliations + +
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    GAAP net income$5,240$3,201$1,870
    Stock-based compensation280262217
    Restructure and asset impairments38
    Loss on debt prepayments1309
    Other(20)1
    Estimated tax effects of above and other tax adjustments(148)(42)(50)
    Non-GAAP net income$5,482$3,469$2,037
    GAAP income tax (provision) benefit($829)($429)($283)
    Estimated tax effects of non-GAAP adjustments and other tax adjustments(148)(42)(50)
    Non-GAAP income tax (provision) benefit($977)($471)($333)
    + +## Consolidated results + +## Non-GAAP reconciliations + +
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    GAAP net income$5,240$3,201$1,870
    Interest (income) expense, net(65)(22)11
    Income tax provision (benefit)829429283
    Depreciation expense and amortization of intangible assets2,2122,1492,030
    Non-GAAP adjustments
    Stock-based compensation280262217
    Restructure and asset impairments38
    Loss on debt prepayments1309
    Other(22)
    Adjusted EBITDA$8,604$6,066$4,411
    + +## Consolidated results + +## Non-GAAP reconciliations + +
    Amounts in millions, except per shareFQ1-26FQ4-25FQ1-25
    GAAP shares used in diluted EPS calculations$1,138$1,131$1,122
    Adjustment for stock-based compensation101416
    Non-GAAP shares used in diluted EPS calculations$1,148$1,145$1,138
    GAAP diluted earnings per share$4.60$2.83$1.67
    Effects of non-GAAP adjustments0.180.200.12
    Non-GAAP diluted earnings per share$4.78$3.03$1.79
    Net cash provided by operating activities$8,411$5,730$3,244
    Expenditures for property, plant, and equipment(5,389)(5,658)(3,206)
    Proceeds from sales of property, plant, and equipment6209
    Proceeds from government incentives87871165
    Investments in capital expenditures, net(4,505)(4,927)(3,132)
    Adjusted free cash flow$3,906$803$112
    + +## Consolidated results + +## Non-GAAP reconciliations + +
    Amounts in millionsFQ1-26FQ4-25FQ1-25
    Cash and cash equivalents$9,731$9,642$6,693
    Short-term investments587665895
    Long-term marketable investments1,6971,6291,156
    Restricted cash144
    Current debt(569)(560)(533)
    Long-term debt(11,187)(14,017)(13,252)
    Net cash$260($2,637)($5,037)
    + +## FQ2-26 guidance + +## Non-GAAP reconciliations + +
    GAAP OutlookAdjustmentsNon-GAAP Outlook
    Revenue$18.70 billion ± $400 million$18.70 billion ± $400 million
    Gross margin67.0% ± 1.0%1.0%A68.0% ± 1.0%
    Operating expenses$1.56 billion ± $20 million$180 millionB$1.38 billion ± $20 million
    Diluted earnings per share*$8.19 ± $0.20$0.23A, B, C$8.42 ± $0.20
    + +Non-GAAP Adjustments (amounts in millions) + +
    AStock-based compensation – cost of goods sold$125
    BStock-based compensation – research and development119
    BStock-based compensation – selling, general, and administrative61
    CTax effects of the above items and other tax adjustments(48)
    $257
    + +\*GAAP earnings per share based on approximately 1.14 billion diluted shares and non-GAAP earnings per share based on approximately 1.15 billion diluted shares. +The above guidance does not incorporate the impact of any potential business combinations, divestitures, additional restructuring activities, balance sheet valuation adjustments, strategic investments, financing transactions, and other significant transactions. The timing and impact of such items are dependent on future events that may be uncertain or outside of our control. + +## micron + +Intelligence + +Accelerated™ + +© 2025 Micron Technology, Inc. All rights reserved. Information, products, and/or specifications are subject to change without notice. All information is provided on an “AS IS” basis without warranties of any kind. Statements regarding products, including statements regarding product features, availability, functionality, or compatibility, are provided for informational purposes only and do not modify the warranty, if any, applicable to any product. Drawings may not be to scale. Micron, the Micron logo, the M logo, Intelligence Accelerated™, and other Micron trademarks are the property of Micron Technology, Inc. All other trademarks are the property of their respective owners. \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-1-Abstract Grid Pattern.jpg b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-1-Abstract Grid Pattern.jpg new file mode 100644 index 0000000000000000000000000000000000000000..a52f26018e8ea0fa2f13b34e584805bdea9d7afb GIT binary patch literal 201734 zcmc$`2UL@3*ER~$ks^>NU8EC=v>?@y7J5me1Vrf&iZqd;zMvowLN7uDMoLJ5fK;h7 zqK0Y^LV^m&AibjkVwrNHGtN8j_kI6=&VSZA>x2bsk=)$Ra|74j*WTB)^T*3S7T8YM zL+l`I?Cfl8?5q#AKW5o1*?76RdAPZFd3bpE_;~pR0D=O?j|)hMiU|W`Bu~rANJ>k~ zDQc<6$!jP`ORJt!)6mw@*VjL-Vq|8hYpSKEulscqc0N8nf#U+=f`a0@XQa>Q{vSX7 zc*G{c%QeFl#lbGmc1(nwLxla0r);NL_sPZn?ZNifgZ&r>Cl@ylFCYJL)`E@`Y{%F+ zIF500aB*>RvR2<dPY^4bV+3sOD9$Xc+*)Cf_RR`NiDg<3D*y zTtZSwSw&S1qz^VYXJ}+>X=M$uv9+^@Bam+H9-b)wfWV;OkWh4N+|~GN35iL!)6(x` zWM*Lsi;7E1@0OM0YHI87^#me`+}hUO(fP2eyXWzfr_{lr;gM1L6oWba;^nKEH;YTl zE30ek8=LRm?;jj~IQsbM^Vf5+vvK@+EY{aQ5B6WqCBiz_F-}enPM)vlVn23`^~oW^ z$#q7DThtuN;~Om|ubavXuqdc*8Q@dUgS`d%#Z2;_RMdZ?^zQ4?zMbs<*}-o8zdG69 z2mANAX4nKc*jax(4iUB=*xrNG)WR%|VN~yP)fASdoW$6p*i>sUY-*vbXYh43tY=j3 zA1f`yu%Xn{LfAv<@P97kSFK6mN7<{ghkjjHgHcO?(V-WPTPta@X#$Gd*`U?{wssiw zLi&I1fr|L|>yQ50u_phxH2}``ZQ-{Ieckr!h5mV0tY`k)U13z~k4@S+*nYqBZ)+|C z+KXL}(ZAgcz?Iz^;QDn9>(1mjt3Uw{%-!N^xvJ`KImQeECg4x~FAjAw@&EL9)V znN7iKzkBhLdCXz#RIs((LDNr&l7d-PYjtZ+5mVnDh984{0?Cxwun@MV(XkKpK8Pfj z5B|b~*Kg27e@d+Hj|=XMq&v|J75 z6bfzD9zu;3EN3o>lMCk`46ESaPgSo&TmZH2BD#j5SO?zms?AE?7)Do}}Y^oT{ZRal!kp1Qz;Gf5`@&7hpy7&U{Ka zBvy(39LKD{`NQHIl^|4!#v<~e5%{L}<>~P0Ed|+V4@Ts1%iM>?FGDND2JpXB z9XjpI8+>vOt@5w_lpbT;`V&F7S#?9fGOG3l>=fJ7(?|3VclA9oOP|n7R~xd_2KK!m zAwP#_J%O#ML%9j}HEuK@)k}7?>{D7s8ra~jZ`Bs20v<+Rz*AKTh?mdu0Dg7UGeA3b z3Bc6jsj|-q$Bzs2fL66Kf8agS9QkqC{ywN^2~%Y+%QG zT*mQJxYgTZ-cL8gFS4OuoU5t&m~-`IQuOocDV)Z}%p8 zv|@T9|I?Io5{w0+Ud++S(t8#3)si7!7khud;tm}g{=tJODuk9261-TgiKDfOzgVrE zg~8k+_-b_9?>RYGtBd)eRcl)G|3)E7?SF~F{{+WhC{+CuMPEV80#|6tpD2V)oC&F` zNzGRG*jKn+20*R5qpr`j9Fonk6(P41Jx)(I*xT2&6U4r0wiXjjJBK?s`zU=n*)8 zEp%7s$R?YE% zW^)`B_bNc8g_7~9hnaW=NcFkfKDQs{3fEq4D-pUdOuRNCF|gR_$-4*b(|PbIJ+UMT zmfh#MhInjm{FLT(`&WxP(S|s%VkZdVx->)T?<+Xqe(<3qbT>B`lt0%cHD9dL_)eWP zh(>lFQfhN{5jyd6y?s*?h3Fc4>BEMJ=3_miML=%ZA|jht+lVw>&0UY0osojn_`jp7 z)jbD@Au7i+hk5IE9A|>eX@*owg_O|1eD)M{+xZ)tAp1cvfA1owE`xwqHf%Cd zJ~AXomAnE;$hmET=D9=hv7Yvoc2(z{FKbN z*WiS4yVp4=p zi<4k3sMfKnnp$n7;^ghyUdK8U2}RJv07(mvq^Vxs&MeY7gimld{nSw{ZjkIiQbT7g z4*{j@_FUVFJV!4t_Xr2m&Y4b`<6lFgLw?YQ$6A|YUInDkr>t7-St+&R0SjAyq49eH z{eP0z{wHX%;Pe+Xd0Dtp6>n$3@;fwnzb4VYaLh{9-+ugW9K&Gr?}_wlrnF`=#{4^D zf4{pwW5xG9zNPnnU+}M3&`^p)uFMNWpUC+5OX?qNFDSiNgl3;|cpH`c++%p@T@%^R zn0XUOD1TwYfqMA3eRn)YHL&u7wUkMcDh)mR@#Z!5CKQfhQk-EOIV%eq3k=GJR1kKFB}eW~Mro z4j$tQII~)2zT=`?hzpl=i8*=Umqc@NQRq8|D(x^Ez`ZavdfM-bfgotG$@%=vmbN5A~Fu`%jGQ)o8@Rh(r(#eG?Da>pa-T!*Y!nF5i?}CF7 z;M(XN`RvKLg5*T!_4my3D^`qG>8O-nw;6jN>@|b{HubtDsk9E^642mU|{mz(zgwPyx!m&iwzrMx+iKfp0-xvvWv&Z7S zK!T95+aAu9}JS)PS&%S4pubfoPg(&!7Ankw2jMWc!Xm{J}AtQ|}pKw}}#Q6zLYvf4?w^P0wrkrewj}7rGIDu=)Nl z*y&3zoQww8^AI<~ek#4>h8nZ%F}fNS>*TSz6K9guU7UIMm!eN@wEBBt!FS$2?D}10 zwRwA%At@~1OY+^IPrnE{2qDAz)5SQf3R60KF0&1rPRAWZoQ7Y_KG9RvuVLSbUn>?m z^jza^^MZ3ORJdxD3_6Y9<4Y^rCkW2AC(inBZ61Np9ppE;X!( zr+xalR=v3UpCy!~5so@28t-%aTV8<*_3A$~%p!u`eqc)EoUz3D&jS}2Z8Ie+&S9Vp zoLr2$aa)Pu9HvA@-$nU8mwR$&>=b6G6gH%|SP*=5J)v1=RqKP{w6FSR7+*KZ%~f3~ zG6O=Q2mV04cw*4P#!9E9bTzXK{8Dti z5D%LEK9rA#!s5jrYH^EemZh3QVBZs10oUwQ7_p8y5~}^m9O^>lge_Zq2rVdpwjMJg z(I8O?q__gznnF;YWe2ot4r51A6Pl?uxuPQffW%pv!F3g1kRENMktkp5bq7e$Y&A6U zmaJ7O67C5bBR#8pt_-d4xrKkZ2hSRimK+axB`mj?ZzQ7zY4f_(thyVfk#a=XYs8nA z)_YYPf*Zq>Bc$!m?&|N%cQ^$*Y}=%>Z|zBH?X=jUQ-ALUe} zB7I(*j&H|&fg8A+Zg>y6>Y@xRhxPAeQXp8_nwqe<$eQjIYYO(0gst6{izsX)#0C@# z1q2$(w_Tq+-tOV4i$+GGGY!E(995`xY!j>@Vr){?l$jrmti3vfst~!zI|Bw}xm5`R zPv|c&LrY@p^_*TqZfwjsLcmf6ss-;{PZ44b)LV#|G{XwIAO2M##I4L7Q<3VVfN@rA zcULt7>74*y#`pq-cxg2ak8JF*xB4XetgS|hh8yvAamPy-%*-MAVhc01S=$YJozMbh zV^)5DwEZ>9bN+XPe&yz0;r<`I`@fas--!HofPaVcpB1u_ruNse{%iHY$%+fEI^Rof zi$BY0R&D>CcCc#uZ?WOuW%~t90Mzk+-zWejXexL8(Chp|Ujh)CR=)#`_f{LdQp&%5 zr~Vc_tn{g^?QXL0UNq2`ryI3I!uX&pUfu12Hcn!Q#>T^EGK(}m%@%3dbP%4 z4W=v{Z<}2C;k6G}TxV(YzzJSV?F7GOMAr!LhCiEMKZ;coxZP^mw5XIe?ogfGmy%8Z z8_*=bAz4Le}uo6Svjf1AH+AlN8kz$1JGhc&`T07D&6~{7H@yH)Yq0 zsn%?d`R%aD6ar-UY#E__XWeX6=L(Z2T8-l<-R7kn;Y#1 z&Gw~%e35ZegGc8|Hr_OVOjaW6DML|>#xH95--bcrZN&IaS;Ft28q{|;+|SrE^=XJA zr5wWs-V}bc0DlR3Qnc~|fZrPycfU$i8hq?NLMs}kw}6xgTAp~-h7uWn!XpzVpYTvN zk#aI(2$!BO;(aJz-#9K=ur@(09W*i)kR)jRC?C^@S$CAHI3nUig~G-gQm`Fa0~%-j z?uzMeQllnzXv1$$ER!;duR>7cTe;}KUKw$iL!gFDY5@{+ zx$@Dp{Z+MHKk|5aO`E!qWR-LhW@OIhHCSJF*h?m?rb=g+aCy3M9+Fdew{RsfIu}8125CRSCJaf%Ok_B&q(yk8 z)Pj8zs~2`QT(U&bY;z;ks{;~|;p{vF3s*VT&GN4))CfE zb9G{)2ejxod+gIdXZ&~EjOs zJxr?6M_XySy-?J509e6oA--D73LH zkI|u_U(GCW9yb1FJa28SIWh$9Ju!11_nH4Hf;m32svTAm>5&r!at0|yW8&YSA{~|2 zNN41$3p%HKE|@SqgO@0LGR3v^wRqM)P+IFrvrwfUqltXalLC8r=WyDpHMg~Dqs-JA zQ*y1(mloO8dMS6Py7&Y75pDsWEd|O(hN_z~Q$o|ZqNfL7#10$Ha7E?Z^A4f%3ls~R z^Cd6rj-5kt6KVC3Pmczgl{*njjl#-JqOLF3x_zvj1n3$3Mh&_fG1VpzYp(}W<<1wK z!wU7^0r38nu>+f^R`>7#zs{OGej&6}@(m1VMp^w;txvSiyMK8X`eG&Htpz~JCW*}! zl|BLSu%{$>SOO8}D0#;rzkjp%ZgvsvQoHRg!4RKZN~mBTAiF5Lpc<6M-zU~g_T%dxAJ;#8pq*E7o!JfGkjg`qU63kNFWN-S zEhL2Lfede0dm`kU?ztvP!xB}bkz_3e*N`+(F0OaZfuKYnXIeZN7HyDLaSd3J#0*v9 z01G}F;Ewgw`=xW6gX-M^k|*W3PS6V@hS=UgGj?X$=x=2^jK?YSEgv=^4gf-fX5}F!0WCJy8)eK-PEk&gL+wY+h!RMpC63S z5vU3r^UHqbMjA<^oqw;q=1U9ex_r|Y^B5j&ghpJ`%+60Fadz){zA_^rhxe>>j*|-S z=^uTWy6u)wwpeG^zZK|q*5BH-x2F`LG zit@*A)-|H7lV-q?Le|)+iAXS;Q3Ld{D#fj4)oQrc2JjBLHeV%d-U*j4Cjt7a3Hn{d_{wztf3Lr;yN;NoWgTzXR;!zBm1GrFC4MvG$q#1* z0{dNjKTh|T-l`Rjo6S1{y!(io%@Nh-#Q<$CaH~&MmyRbfCU2@V-x&@5A*m_*AzbVE z1&oH-!2YE#wYz!mD3{bO%-jd%K0@5Wy0^Fp+gBv|<=wm^S)%&>v4I*b81Q2v`_Udk zna;kd@KJI`9(M}&xQRQ*2Uh^+vnN>IEgVFI>j%H5U?8$5>*AfgIuWDU`)+ zT>b(16r2-IQ}HU^#>1k$a!R*2mzffg1Bx7;l6USu+OebsX&zu=yui1S^rxv6Ka`w~vmCenT|MYK@6CjCpVbW2Bx#=6(sX7J?mc`3wa!vM;X; zY5l$c|J^Hq#+Z@UCC?Iwuc8tfuB3q49*p3}Vr}8wC`Vairh4X}UInsSIu@bg_jDR8 zUyu`5I5EI5U0haG3e#ydK5Qyj8c#ynMfex(shqFr8)bBNnxfiguj%N_r#vIbF+at< zKmX=eP+h)lMhSTMjcS{^K4`A%e24B_5{`uZP)Fsj^(rRohiW{^?U0N1JcBs^gS>Td z^T5gWur^mSTPd@x^CtO8%-JjVa2q*Rw34vWH;#u5a>uXw z;Ms;I;?26#!Xco35WSx+mbkfNJD3(8Dnx3#$;n~ z%&`781M~_1+NEq#RLoHs=goh|E}C9B@Zj3OAewu5!p@W~$?GRl#TW#h(SHK(zNbLd zr`SP2s5AMx0x=!Ot*a`Kt$I#so}+x(egzh4UEnA}htU}xVq<@9^)pM+g#c#COKDiy zgM?aNnFd?x(^1EmpF2#!5H6s*N_mlbz|;tNpmTC9m~!6VW6Xk|)sIfu{Vpnfje~y) zr2n%R^p~>qO~m=^t_O-yP@gVsLx=*V1t6Z)M4X)pq9it|9*qMe8@km{nD3 z^SkmHawu1{v7sOp(M9qNJVS8jU#jl3>;IJY6w~V;6l{{YSo5GgjKn@PYW%`dI#e_QibZrE z7`BKvU5ZyYc}6rH@U{<@66-u}*yvsUwCNY{V8mig9H%s6`{kvi2K@oo&l62!%WIQ0 z_+eK95AfM$eFWk1@CVAh@}mI$`(s`EaCZT+{L0w#7FZ}TZ2QsuRXWzWg`uYZg zxqcOBNDC<5ast`dZ{o`ssT5Q1rIyK|7|J}}Gj_f^y=1By9HhSH)3FJCwxp>TZ+Uud z{r2)zyl`$u0q}Fn0Js({>BlPpT!%k}BQw4aXJdf7*Reh@@vqq++pB$e3M(LKVwJgj60bavwhg;qokCNzF2&+zx6hYkS=`qfsXp9Bwn*WIB57 zmvZ$kDpiu$;+rjxyTOwx_1`%~Uc4nzkXv#Cz)m$kLG?r(BE~cQy36G;@q0 z`Gcum0Cv9=nCg_;(w;oTz>iD~6~kiDekil%Y3 z=jFNTI4s^2EWzCt^4L!ISQS&sig!7PB2rZhT&$+*sW|EBda2GuX`?H8u4mf$SVy!5 zcWC1SK{`RYERG%W>ik>kxhue*-quFBHOT0jT(9q5xwDB?;>NzL(Rd=G?tEe-6;~4O zC@0jRYnjWW7{e^G0}38eJYhOii-sK+YFz#Fnv;<(OG-Q6A>pS7)qg3ClYzJDwld1= z*shZ4&o^A1;%ZaS5AH^9IaIC2G&Xs@uCl`qlizF%toJD3mS5G;^lnJhDNiiX^f9iC zlQek=2>`u|q)H4htl@UGk*_&9a4*&EaRB47PAOs_Z`_C?2s>)aTNHs3#myH+ZU z28ik`_o{*x>m+vE965jz=_z&0PB2zO+Xw|eCp{$|A&nuJmaU_uXEaBLJCE3eI)rh; zNzZi+lIU4Z{5@wza8l_~>2NXl(+)DZtb0s@QNcJL6uR(bs;fErQ?33p`fcYozKr6a ziiWPm{HJ^cxJq{dVBxknA=Rzmo0)1KQcByG#YT#{aE z+%y5v?5I163BRK|)V+PazWrY<_>{=Nc^iWLmTxC!UPl^AmRo1Thur>GdeRj&=aZ|9 z!|J;bb@%w_W)W$T+febXCfs8cY2tHAkawtnk^TFyKyY>E;8pF`+(CB8bHL?XL7!6Q zk|F@iVKdXW0WX#*e;%qi?B)(1Xnbg!+ymq!{@hzEYjI`t`|LFyUZvu?55i5KXLm5)9k zrJBWhzUvKE!FzeK1J^s#T~5^PY2i!4uM$KC(E^LLxTvgYLj1L7eO}OUwHS)nC`A;;0WV=d}Tzz36R%RNK(gCB^=ig^*)xJm`DzH;buI z3`|@7Cq4fs)A!~Qt|T>^4NgwiR{4sXE{Cnbicgn%6XBnUjF6 zpr$gF=w<#$7V*)?!Q9P9%hGj|7+1v25`kJFzp9tw0Y^q4dbVEh)M9ZAi{w>&R@eJA z1GD~X{~=n+Td3lNzypj;rMm#u#LV#VeGwYkTGqrZzx|SZR6E-N<&Cwur!>6ogSAy& zDlGgF&G|AYf%wY`0G&-ZiFw}wooO2IacQ4;x#=5H$=gY;(lH*sqMfw`vTeY0O z-}q7xnsr-9(67*Z7;jgZ{6dPlZ{~QuLpdT#ZT7b)y$-* zXiR+@2aM;4f|$}rifOX`{tY~`?T690KyO@rs^c%t@Sao?9fP$L^wX#m9_gv-Yz~$F zix_FmQ2uK{VlAG&TLtD1c{(nMe1QR}SB4Aqu<$^RD?`BJwP(?j>d=u}I+)S8-7Ea@ zjIn|bYvA&0F>H^(lz6f^Fx{+SS20HY4#l8To8mSjf6qi`JVf-tp8fi0l=V3vm8R4^ zUU#JIg((ky4zvM8D)fg+ewf1>gl5RLFD$mXxdG*$dkwY*esMT;^kuS#T=$64JFwwS ziz#s^TEjp`9ubjQ;7G+8U36jhwXQ=)c9-L}AnS$|rL2d5oj%x-eu=PoOjfFs=~aT5 zt7&R1?3G)qLPZV%A2=qvi(by@szQraxVgajTC*Y6dfd$=#F;Z29z6lEK-CeTP6oYJ zr(mYK9^-Uu7>z8NzpsCJ3Oggyvc7KON}g- z!&e&l591KaT=d-~_@^=9F9!LyZt*`Dkp7E5iRC@|hX(SOZ1K$p^_R%`m3w}V)!%(k zUp0_#d^F``)639nlDB`~i65GWtp6}b37rX7Y<#F8Mb}RFHLMncK=PB1rKa$p546h+ z8AoI@6Pb_Ul2sitcYWh_KrdWe{V%>hD(LxLwxmu7jT;uWgf(VN_kVg7*TF+}iaL`jzv1sxD6#8b|$p@b)l6h|K#$K(b$EImUmL zniTggjvPL|nb74xx-VPVg|se~J=oLB?U=wN>s^I{Yc+NOL0>I^pSF;(M!J#h^!C&Z zca8Ym_9iDf!(8^mBMC4ikRW)!O1R&N5c-jQ&aJtWhb`KJR-Y*TdGqEbDElf8n*jW$ zQJn%ei)j07%plb9{rLvIEx++o7Ce?+dyifcsHZ}c4?gtvB_iYHcWGjU}^Vo!_k+P_>N>) zOfFsed2JOKp*1UH5?7W?@< z4vvgNmrU;YT}etP0+-0ptL_jyw$GaAbHbc0N2pqks2>OKtiixVno6N;_?5C{C z1d1nm`WNhj)%G{HbFFp7+c14)KA`ed=F@#MZr%$!wv$T~&rJ}^^+*Zs1?n%4=B1!> zo)CG!ev3~1Tk@MRuUye}g^>OF8f?Flk9X#*L{nsPqH8_^w^7bFj#kMAuelS(6PaqA zyD=*+j(#|I?uRQ6a?OvT&}lPqS%rO=eY}N(KX)B=v3=g!d9vrmLg>9P6`F{*xq6ck zsc=sQeXKn_%$6BK*K4Vj87-hXDlN9WHD6$4sgss)>v!a)!83C&Z8Y5@CKkjzryae2 zWj>6=Alz?=@1x;QpW+nPolriYGgw9nW^3}yeyk`5RYAmQ4rCI3~V8urMDHqUJt?I-NcvM5RN5a{plDR1@}3UqWy1fMgS*eE$aaadlz}@VS}-$y3v2L~C-IY^Q8TT)rD!A3R@+Ta9!-&GO?3YN2f(WbkyH z0d(W-=sljssU^Q3ZKBJO4%qr28> zR@3xe9?EI{(A5sc$E^x?Z2!VC%zS2?+4!>AbPB{$aD1X)ewCXaeGAa4U;Wzuuz9lr z^mmi$|B){3;PekG`CsbL_l^FmGF1P!4JEy%M5Hb%6JMUo#-TV}m!v<*_CnSC_dUXN zPDD)M6@Ozxl(Wt8Ew`l{1!gE<09)<|FB-KZKl0B>j&OAJm3VasBMl}lzkR!?W+e@_9){XagY}mv$ z+gS>Mph_{5Xk8UJZXyF=iLG#NvDCxEZK~HTW%$-h1L4<#|9k(TancWcTK*6xT zl9pJjDW)NYyAEw{kAW+SQLOlkbF z8Esx!-_08!`cs=j5(Ya}lpA)hylwPD;XMWx!UbVLJBE!0FTJ2gQzF%GQWI@i*5wZk z>JI|3(47|YZNLXgdHaaMatsJt05By1myu7H^@ld5-PJv6{BBB?m~&aNAB;4FGJu+O)HUoX*UN=W%bS~Uzo6aY{+7;EWzsTnITTe@*p5h zvGK>NnGLSL*w!7>^Qjb=)}`BS+%am-T!(&=+4q{cm=97axLHzOGt*FlyJ^PWR4L)X z<->zM1Ud)C`CbCwzWyMvWat9uf+6w54%oP2I(*Tkd@c%rWdq9R8-Hm0vehDPXiB$h zYr@1_uXBH4Uyv^t`%p}R?=1e9d<(wASogH1HCXtgK=;CXtSXeI0d~rgLj+?w?%)bv>8QD%OdBC?FwcWKTsLkpi zL=2V-yi$s~VIANd(yHtjtT6H7o;#0Wcw&k6{%svwsx}tW0d8zL_pA)n|2a*eH^ocKH>-%Y5mb;!uS?CuZ($h|LYUmFqH z;f`f@`*0Z>eaXS%Ha7CPK~wl#H?+Plu7URkx>zHM%X8l#7ag=y{lZdY+(Vbmhzx($ zSHXb4CO!ve+#q60;I4If!NO1_lG*E*iYZ}c6*@-b`WWZX>oZy7U!RLcsfk^;D zP(sE!4qMrRTmuVS&SYo-kK#ShSZA>NMjiv7t=2e-eY80S`c1N&YGzj z1fJiZSmkk~y1aL~Xlfk!!H7^@s4y^SL}Kc=|1?f`sDGm(uBjZ_Ixz~`CSIUI)__5_ z60sP($u=Cfzxi5bA`jGirZnO<1zNaJgr&Qv#V_IA&BAUWWd=rCp=WkO-)wehIq~0b z8uQ&4AeY6$)S2YW>(r`)ZX$4C&-pd3H?QoTeo)P8XYN$*=z(gn*tk06d_qI?eJIM@ zd;9{?m>rh6G4^P~1+j#UMY_;C%!v zuD!qGFK;sh^sd15y(JMPj~TlI?PI#@0y5dL>m?8h@R<`A@Gf%kR$`wn_glz2a{ZLVx*3%`C6|&=3iWp+7)jfK8#78x#Xt6Ntqw{CN&iT=-{8XQd;9ib zg90~}|46|Cm#2=Br5S6dg%=b=*IUbtza}u;0%j0ng+FV{mN1Iz0dPjFH?nmUY~ZFx zx|7Y6hCfE+yK}~a#UMA-+i)X_JTcP&pA84ps0oMfarx6c((88Nu@OLsJCI`Vqy9yA zg=uoT&r`|wYZi5<`>=|**MRQB`=^OuBTQ`onD3{w^$)UlMOD}{DaTDV>Iz&1nWZh= za_klFE>@-1d{S+D_OsDx&7?BP67Sz$R9AXmYH@RXH5p#2t(<@J?E&{gj$EcWYiQ6- zOr3T^@O6h_sTNWx{-V!Qddsdc()YYEE@w}yqRPfTZM-KTbDmb{qy@lETdyowt5u=Z zcP?l}#V~WHPDPA|)mMmYx!x+R+g{1FqbcjL#I!rWj_vqC)o7GO1w0k7t5)B0%1`t@ zRvH;f@z3`yMlQ0~14xyIPAJ>axLPH$_8Q&%4>tGD)8U;l)WT6)*%D?kaXH=NG9{7O z+H=)lV!y%He`2Xz7!QH8OWV$uIoz&kITLHVk7 zSMkm5IQ&$52JZ5EHIKW6VR($pBT&=JA@&Vw zq@+U78q~JNzo$E!W;^sdnfFQDR{i{!(-E@ZFu$G+#og^Q~$9vONN;H4` zrd$S(K8VW%hQ_fNK zB@sTS`}8jA;?@&dpD?nSyqRo3xA*_jbn?VOMP-J7BU_T?iZzl86sV0`QkMap(Lma4 ztgAT(&i?S}_{t|@z2CN-DML~6J;=|iG=spWA2M{k?tS!*+sUXz9&nbkr$mxu)YGGL zIYBv-+F;O_M`-*OgT=AfH^Txw^%C3+PMW*S3xjcY*$%*}_vSm`21t=Ou+}D7iIAm- zd?r{F72=Wo5WpDASBK5~g5E{bD~!r%sAvo$T{FJf8EqbJf|>BR!(}fgzTs1%ct1}I zhJA(W>>`xeN}zgN`LTt1NZ7M=yg6X51dbd2smEhtpBDaH^*Al|?gU?*qU65%K>H`B zihf+HPNZ_^kBSUKqlTCnnh5k)JUnA(HXJ&pa5f3nUPi<&clvkh&J|4K zz4DWp@~Rxb_OOm__aMwOJ`!iSsoUjv84#i@l@=j6Qd}R(b4bVlS}Tih!PR#c+^qu2 z(IK7P=!en?M7caHb9hj_sYLluj|7cV$~7*t!Gv4(JKBP$I)@Y@9kVhztRRBUAIWrW z4oa|ErFQ0s6!31mWd)K{w}z=-QUMQ{7=UsPRP7>-F->n<4=7^OWnrap@E}kv#22Qf zgA(L_K$QJOM!gh)GJrdht(BeZCm+;DthTjT*Qba%3x(D}owxabn-^kJk&hG?DOP3D z(~Q{y-FVH%DO%`9()>+nsqpIOsDn_r&Q-osw_&*;9W4AI|9rfy{O*XJEvgJvlh`sC z#!cHcoAbwzf5ka5g29^UcK|eqTfKf+<0~ztpgc2c$i!_rMR!Wa(>Sys9usl=EE;=D z-4bvIEZ-Jqa*_;W8Qz=x`@TECzRjh4O&=^d^3Rd7l&`Zb-(2b}f${&`tnp_m`o4wQ zx2e?c1Fe5ulup+-p#}Lp(0YmZ{e94vhTm0YNGd7;j%(u9qs}7$%Fb$U<>gxjg%Csj zj`gHdcQoSS9D0v(SKr2nks&wwU4H(9t*Q1Fea!Nq&Mda&v?*clhmXSO;XX!P`2*?n z58(bc8ApT$!>E@dexR9(8%)M#Ggzg{Q8fMwf=^X}5-YL^8-zH&abIB6{;U6e(Uwn@ zlrCOVHiPNA^;oCvT&X!GsY|&{^o{K&53q;b{*F&E2=cGv#!H6%*?p3z#?~Z8g9O6 zoaBCvxH<0@^~(Hyg&&IBh$LXS8@`MZo zr|*{3{W=u|%yMlRtHfX&g(6CnM#~x0Yf`CqfS1oFzmRKft+*q{?7AvZ^s&u{H6eS} zuC_DRct(%tF54kjH2vTW6PhA8&@d90ht&pe!k)Tj2=|0G8|eZy+TMLqR{CxGW^a37 zf18c#!?6KMXx^CTUB_Dniu2C-|t5VJF6ne5KZA zy0a*&nQs`Fg~9O}H`~!%*~sh6lT?FoC+$^&g8m@P+1MRB6+6~ORwEeB|S zEfYo{`VFj37K?|?O1f@d_ubrfpX~aC>fS>;J?rNKzIKr@+>MY%+RC74L*9%r)&TK} z!cV1Ix=t6f5ispvCHDxq*mh&zvbgZ|Y4aEQ3!e-dGn+*oSR#kZ``DN$A(o(&SP%Gkqvtaa*8~mi^QA1tv zo*7!<1TA*DjvT3ana-3>emeRxP`+mNV!$*>WghC;wvj#lIE==UfuD|wW@}@n?3tg! zm-C3j_ z$DqU}R0+TZg6ZkiaL_*1R^P?B1Qs~*?U80)(H1KVD`#}mXS`0v7*B~alBrTu>yh{n zasDjQ$_4><=b6#f6Htt++Dp=$cU>U+rdm#b|x>VtZ5M%<^~jlm~WuUsGz6K z41~jRT8n+hL%lZ{D$0$SUywzSKcvAL0|eZW0#aTDpobEYmk~drTy>lxJEoV0vDf?} zsj?w5PTE!I#wPYAer9-_`N^t<%S#k@-DRdh@oM@#;Pai`D?t~99r}<^RsEe$6r>fE zblD&@7Wrh6HDjDTcTGtJ<1DD%h36)gngIM#T{!(n;nvtqmK|o)?85ii=YLLGsipii z;l!GM{>M8Vl(8CKe~ndq^)vtbJq=iWs=vz@RtxLzcSHDJ*E*$!^PwjV3#x3+AQ1sY z>s(3AX+KFR^(g(`S6#m)LvXFy+q}_pBiUPSbvs=pFt+6Fi`yy;=9Mc?%>9`*kh>e} zW`D4Ow)UDd8zWB6Fh#}v#d-Vq=b6P%vBW3hi+xA1FP5XT%$A*tu|AJaVHoepZ(795 zziho$yEL!A8|6Ine2<3G`FtDN?$>8rRJ@W{MdjCDA1YKd8rZUzEUxCpnFIM=z1FMw zEvs5iFTl-RpL-3rLBKRPz1AU(8rR27pmol@i#UI$*(Qp2+lJU=SLLGVnGk+kjbZvJb0r+$}13>*?8p|h7!*}J&+FK=+f;ok#$7irHHxH^h>ND9Mt*f7>zc;!dqi!7_z6lCiVJr}^Isdw$ z6LTuJeaOT~VI(&ev8!c~+g^L})$5ZQ`@UW(_*hiV?t7>(gMrH>R1D@!w~PhsUo2j@ z`Z(v%Um_-%aYeG{O|giZA@y7c1q8>I5?|A^-CByGS#P+Y1ex-@CW+W6ydCqx(H#@& z6FmT2n;v_k>1^EPwH*z!diQqJX1ZDdv%T7u3*DD%Y^XdDE7huZoTE5TPCO_zq}q3Vns{m1#+G&~tZ8z)?v9QI zLGgh5{$$+cQ_ssT$8$d{u`}f0p#OAR>+z^MF8qCT*~WG4fOoGAAAt#+=Km(cIA{nm^7J&1juF2w6QpS(hUNhp_=U z-Ay!BA**ex4AUc(c$;Zl+gg9%&MC`v)sC_FOM=dvlEgOtVpga3`bZ_{X(KyP_PiZn z@%Mw1-1TnS4S><-Cn9UeV}5y6Y!od)BBC2-`LmW9eJ$ zv?vW&a;3fFt9`Q#tExlX-OH)CVsEuQRI*3AyROzAO?emne^K^cQBCjr+AkJF5Ku~J zQlytiCrVdDsPKZrAJ;Anhkm%lWn`)|B_YGitt&2V_6Mhe$5$(?59l>=KMswZ1MKG8(RDc(NXs z$l2yI-0&num-a3Vrf!3i*TN+BmH{Bfa=3Z~ODFEfoiF5Xb+`Fp9Sx=P!4V$S@87c3 zVFa}c&_EytA8c1**;rivVf}@0HGhf4DY#zN5jO9VXO+O=Dm@6oz_Xfwj_>IcTwvy zU18;2L+5oY{GGh9D{<2!f6uFvAy(cf{oT3*V$uks=k?0=^uuhNu`;PjetWqh&HA&v z$n0p2AGNx)e5Zz=#I3&*^RQXrlL2gbhFW_nqPC4uYv{XdIAe|Yb4V167JCk^U3(%V zV>Vv|K);qqkxF71Y~RlNUWd8fskkY3KRdN-y10Z z75)4d)rWfc{&!5sZ&<)T38wPjn2`VJ01N)-E&Oeb|3@nH|3#`J>|$KE>i*zEM>8pX z<_CmdHxIua*1q}|lf&1nA2#pHg&j^C?k*;VfB$aT9unTu_B5vPZyF)8BJjet$@+Hc z-!xh~KzKm@PukHMWbPZ5kJ8yOE1YZ}T4!>eO*_Z`%=i)Z+8d4g<9VIrB?3=uW$L9Y zBEqJ@SJ3JC%g$$-Tzkb@T)yzw|{5j0>+LDK&FIk26kW0B7tuheyXSrwnUyFF5?LrbsVrWPrmr{wV zYS4T5^6>C6F1>XFb5fw;R8O)3>-wfImsfgX0>UENN-CFr=r%ssaOSm~HY;B@U3O2h zW4+*=K@&vdvh zv%N&#TD}%5{#>sUsWqKJDePzVt#Y4ubIi@v!1Ht+%3=C!=A{X8V$vhUQ-Z*vLNf(e z%i)FeTrw791Y%x3^4Q=K26;^A7!_e#3P8df>w0>d%uKH1k7s<4LDL0 zM7Y!1!ePULV>J8bQZyaV5sw$#-lUU&UA$bqrWDrG3BRZ>{iCks=lXnE`iE}SnR-Vy zO>|5|z+HZX3xzoLedPM(65Bi*2^H=>r{%AaV#C5KXIr&_c9yiINv%hyZ8+?<729k0 z@!tBnA)J5N;kI_;Mqv4Rs7U^Hr+)p0r6}LYWZE%LCv$CUZxn!{rx=}8G1QwQ*TO#rx^o2zMa~D zK~Qdiw{opl8CAFex)BQ8T53Me(KRP!QcQjp#dNJ*5GHP3Y=4H3U!2cL)wyP}z3H96 zd3^X8F?PZ9u57*|9Rx0Hx=3U8AX$FjdipJ7Ys-Dy`5Ugh1Z)-Me2FFi*9-D};sTyv z(SIEH>ymwX_WOim2=s29Q8|;~3N)L>*BTQuY}0Fn=_lw=%Vxg**^=;IKWzGSb&*Lk zvW07qae!uN$Q-jzNOjg_%(ovfN%BsF*fjV#h+B>Sf#?nXT)dqI)h~BzUAOK% z$%tU8B++W@t=V1KLFSSTGgJed?acblKuMm$ShkJRii8H+a|ggatfUjRhYYmSpM<{I znxeKYYpl~2qn4|0uaOcb5&CVs;2vLS78ugam=rnN__aw9r`%_k@+c^Eg;}aM=({dW zRjN*%JgPpe#>lcF7CYRUd;ArJul@$kx%FJ$yqa-I^&FO+kvucaKb>#sQjWQf2ig{I z8WzAx3SD0+n8GJ9_it>q9_aeYSTL;ik{2FGPMM)c?F`SzPDS}JZ5PaR;Yr@3|r4vls4OWTZvE4S&Ebi7)dRs zIju_5Dx|Q3S1uwx*jhI5JVOmc*;i$>gc{NrCX|YNhsFxCcLx!SDx7>K{9S}MLFR|g z8D_HC@ctNb+F3qkg#cAtNRqM26pL|j(%-@xc#e<2qy00$s>i9wah$;;GvnSt4{QH4 zFnRz!=W^MP_WqZ_o=DsD9vi21n^p`g57qKzRF&a(XyWni&`rs5?PL$3GN8VWNo7^q zON7odo(C=EF+=z*EQ57s!)8k3ju(dAk3&pg5~wV4zy z_K!ZmcbR&fD_1Zs&^tZxEZO8xfsrcVX z2kpxLM_l-K_U+&0`+qF2fBZ3I>Br`1?+zW`&ITrwK%tZBcUjyo2du-#lt}m4iw=7X zlb+v(YjpE=ci#L>lT8T2eZK8wF5zrdk`#oOczdy}ay9=noZn527pK!-_%}@&)d#g> z=#qX?YE^9VB2av$aI11D1Xp}nQ{EagR!t+q4@_?9Ob-@$olQc<_jbRmS6_USEX)Ea zJm)iNn{pUx8A{PIHE7s*j_&*R!apBc`E)D(2hkU-tqh*h|IP#2H1=z#e-yd6z7MG` zodb;YR7XWh{Q^}mo%qq1-~Z!1=mU7bysAO1I3c?6j&{^?K9hUdN};Y)_s9TXu8fp0 zg+V?Tqm|DV#@`FU^sn?=gusol)g&IUDY>RJF8`qUp;3G zscGXWJm?BqK$9OnnObeSeEBuv*q7T#q=qfMoJC@Z*Nc>x4T)bjH>i0qaT0Q}@YwLz z$749$H12U@qCSL`X9B6q$I<%n9aQJFY)Xu$nlx&N<7nGN*rU%Nse=Vo73`Cy`-mSs zANm=QAF)<&IuUstG=K;zDH6#qq-ZwaHwr2+8{Q(Gwpk)`%(5Is1diKc3|4K0Qw|Iq z$a2hszGO+^mVga;{MRpKyY>nMwI#*fVp>`cOzj@ALwT_3>mfE%{J?u!nk5s`Z$Gv&DJl)MFB>SgHyZ2i)>O3SIk( z6#yRZAf~9^_sJ@p?R`Nlfj&^yEh7q=g-A7I^6(6t2n9zONe*yr74ov#@MxybJ>Am( zvyBeKhDi)m*|$h^t|SXTypIKfZncu(zLM|0mUDUkGbL) z&(D<`KDeEnI?43beZB;^k&+i>r??`6lsR>BRf|(v2aosDPL zKv4{0a;3ldm7x3-JirJ{;tj;HT55l;11vp~^Ie=TRPdQBs2`)IFzNkmqv45lfhvIF zD@5XbZ(uzawK$lwp0P+3LM?9}z<0YZ#PHTthf6ZeMBdm2y|?P)mA~!V{3Pr23vuTz z$Ul!IXF53{qss$d0^-s-r;uJaCudgb@mlOhP9+%5^ZQrg5xRAW&QC$?mFxjk(Jz=> z?U#jKDo!=L^2)WqaswX@zbRnJOLS^wvUucY((MBbJA4Ytsob~VS7+0)I<8W@eFwfS z6Wlsz{8Wgo)!6d=9*~$t>*H6_P(Gb`vewa-qgkg0&kT^`rmDa|a~nFN6&I*mOdOc{ z;lQo6$50~4-lvoZ%`!|Jz>rI!LcHct$)20?MtuxF2mJK6T-AB%v#UE1sQ@8em&Y{w zHV?`&>iG_EHBSSkYN6a>ZI6Kd?2ZdZ{FH*y-NJ`X)_NLMwyZYuZ(h;G1wnGxoqfIS zH-f~MsK37wKz6y?m*=pe?^cUJoIwr?{fLeaH~c(np5e>3>SRyl-a|oT^JjU5 zeX4bp)d@1*&8$VU{x7yMuf~RkG&~pjfnxC`GQ4clDbu6}W;I~j^r%;qo01iESN<}Dnw!uXG>b>VRRbe)dssdvlPo-j zt7n|pDocpHwRIwKh$R$T*8o6?V>l|Z&*tjY70WqCpS?_0b;Q?fv4>SJXf;Bx3#eqh zcZxO^=w=m7LGrDoS=_Sy)h-%}T01%V*s6Y+TOh;U?wJv!Zf|8*xe9Y#B{jcLVi~C6 z&QFrd*4U3B8!VnsUa4Hc80(4@bpq3;_`X8)w=#JrjTI&JYBOU+y^A0(DlU*chjWvg zC5K$DFM+HNmK6urNKF_|+tX_IxK~O^2GXOpVJOViYt1IWM;`{a(;x$gx)UX{w}$4Q zc|S4CGc`9~RfSR#S*9?rAk4s}7+JUK6it7xGzF((XH@-9sidBRF{J8ry%gq2tmn%! zJgc3zvRIz)9fK~syU13D)VnNf@vb@6#n$?Y>I~|%kPDUOCX`9-i+l6E>NfQ838(1f z^Z57{lG>^`WMDSqoq`A=^^>}uak}jyLNapqJzn>TLpO?6T8;aK@j$uGqYarWSD0>l zAcp&WZKist$^A|9K2wq8nH7217vHH;DMfNnr9_ob=%GLLsy<6Qas02((SMVy{u34P z`)K}Ga)b)GqXxpMiSOURl>dR*{;!S4|Gc+U!}0gcwcvP+rx#hdu>6AI#No>Z=N^O{zh5zN70rKC*d1MXrep)=Ak;kIS&w}=8OL(2nBw5Q^lf)%j^mS@MpD3J zc}*y9OATUm#Aqml~Lu+O`PBEL^C6RywRPqsZ}~#<-CK3Hd`|ZFe3pr8Eobh4 ztW6vA{R$tVo}GXs%NtX~0tG~y$PUROdK*tEs%(Rn4C55quYVYH0%4RF+QVOLw1{Jh z%#xXrZ6T~$#}Z_BTTd2)K^g)0O7EtO{R#JrtUvNt!Lh7iO)}1)%nUx7r^-SvU<9cTE?JgfMOIq8%qNrL~j0g0q~r^ntp-LG1Tr&Dd5s|9(HD^TVmF85vFwLC&MQ*sym- z5!MC|$KE{tSnHRerom|v{!L5x%sESww9s9}D&o-sOYhV}kxr9a{eOOGO}@q?k%;0vhNO0(R=)&~kOkAnhb z5|aC_stcR@1_L#`xVV#EBrPyTnx+PvG)V(CvK64M(7;;xtUBccuZ{TCiJIAB7jm$| z!hqE3I^DkYT~%IhgXjZI^agCil`rK^IS*C?JPJ_CGxl8cZ0O1_*Q>G;QJ|q7G+H-h}kAS1!zQx_JoYrMvsn>*KLx*k2 z*`=KeiKxcxN%TJ1^`jJ)JltE6ABNekTqC(P`l2C?#z3;2vpM_5oLV4`JCA0CpA?5f zhGruKG6lx)eYh$gQ~1`GlOvZ1etH8UqI@M4+`=>V29RzG&*>%N@re>Uz~)OYJ6fq#+)rVCI;mbNUk*&@lS!`5Tw{{y(&3*zgz}!b z2Ce1T&`jr_Ua3Dif?dAFj{s7Vl$Q$FajorwAg22d%B~DmU+G-@-Y{EbyQnn9YJmS% zlUtLRxe?wdH;g$3t+&(Ra9Jz-r_mxp*hep=9PI*Cp9%*Uf1P=COM~nzI{E7qj{3k~ z-sT9F_vdlMEe7KsDc|dZ&?~4AGp7Lws3@}z`^=wf8M~JL^wWYaTpJdq-Zd1&vWBNQ zo3QUxMmgToxzPA+DM#KLqhxg5)9YE0dNdYwZ-tF4O4O5ql&+Ox1OGtfy*1tw0Ii8C z#F>i=B$&94+RKXLEi5zfxsG*pu%!|t?Ax_Zy4c{)f{6n=vk_x->3c?=Ac0Y{r3_L|t^xy>tS_M&s_i@Wx zGb8e2Oo7@rniMjxMK-{kjgFy3H^&g?vWSz~kJ~ewl?TPQJ*Nm#7FXpVbZ=`6MA?1G z?TM*QRSizqw7tyql{pPK3TVFD*YKJu^}aDX_ocRAov25bij)4o3$Xte?)Klj`2XK& zQ7L_=tTmTdu6YCA7z=Q^!3T*8!bw|c&Pm)GegLShfq2oy&Ysn?o}3eC3UpT3az=j? z%>Phe?XTgymFt%mzJ@};Gho-5ch8DG*Is%T){Of>_h<3Xk~_R|#CWPJ{kYB|uig>( zWUKC!PaA=D=-k&OREb=*xCJ4ky9}SP7J|&`oxV0At7#d7F0ZH`PjE9Ptn|U*ybB1&ys# zDihtt**NWnK6hoY9-v>`f>s_1i)UDuJ>J6EEmuo~Zr2j$E`L$~=Ehpod0Y|-Xc+7vQv~9@Nn0w&bvS}@8z;|s?iD)-2 z_F{duezh~%J_QDI+_!HOrOU?QeEcaB-$)N~_m0 z9~V`%$4bgEi34O4xLb7LO8#@IJo%X`vqtAnJ|L*FGFbuzhOuRE63MIl>tX9ABPiFu&p1zWjU+cndo=YHa8zXcaoFxw6Yn9oFd?_ z&Rr@xvJQ*su9Lccv5of#H@4fKdAnK4cF}8yS?hr8A6AjD(RlN6bYTIasg=O>NQZkQ zESWMe0vz2es5hv&%aXJa6#2n5M5_MhdKjO2*v4J)vqbH%3`^DkU$l?jY>S`4Zl3|) zXl?W0Mug`indfUt6*dWHs9yi>FuGts>ELwOWLb0vT zncn*eH7$jz;@w;EQUNHm5C7$xd|NEP3vW4PW6 zh@!qUpx;|b%g8C3M+;Zt&seJsL$~txY#=e>;=|hhK6N$XfWG~(AvgJli@~5ft@tcy zUgNKG@?ut-t)LaBPqs&B0ln-M3-kQw&}+OIWI2Q6lO*H5&^=?t1u*RPK)_B2$Xi1v zk{gtrXnsg2wg+FcaV|#vyjkT$z;+iVVKyH|TTCvg1`~6fIn6USnpH{p7B97IKfn|C z5GR&{TkVeViA}(H2Y(wMrLsi-K69HFVLX{eYI6_J6J={dd=^CChdkl7CKr=whH%mo zi-xS4e_)aEF!md`!9B7ayJYKO$k*YIaG`oB9E-Swh8hJ6PDX2x>Kj;T03kL-VDwhI z>u{r90S_yI-^GnqdDmGt|-orA}Mewn9hkIJvFH|7_7`Nl>2?4pkf7pn^wE z@5<;a^P@9F8_N9@9W;{jx25lBbchJk22;Zd)B07nt|9T)=a)+g(pFxHE1itX$zWxTqSam{=|4cxB_WxTbYzw79L`lbXK5)(A;DWIP zQs+a;M~Zvo9Lc}UTkKP1t|fa8*~oG}!^>^5LY|(T5tT|1n+FW|F!%7zOpc7vS3cM# z**7#O+|gfa^V##qpcL~<(C_`mq913PM~s=G}WL0mWOK??TDv9mZp8g3LN*rOJMI@UT?7O3ZzzOvFNK zccF#IZFt^LmEM!0G*yIB>=TRFm$XKWCKYSXK!Y2Vhbe9FZfb8G;xXcp8-{X(&q`be zgF*FopTJm)kUUGHQ^*_kC#admOv|jWwf~~WUi=NN_^p!D1X2-0g8!d{ z>k?|F>R)pdxBd%S{}1=#f2xJQtJ+lO8Rtcf<+da*FN4}v+)-t?z6ciw=xkw$A(FPrMznJs<#M$9lx!b|w*CMkNO7ef*$lBfX8+~8-W2^$iF^##iP6#n6&1l!evrBgQ z^?4-B!c4So{X~3FZTz^x))EE-Yn~L>9obCx%~X7>w-qZ`8^bU;sc8O9GmO*)_Ga9) zWwC{Q4C6_YaY;-fA}KxFtn;qaAC5XpK5%3IBRRjNuXdh|LKk1)}3*5?*@YZRI#A#&8Ukk zuO(*<^1)?~brt8?rocW7Z9ub+|5V5MujK<5w*s#1Kj9;6QVHReTRpCUncwyZ3~c(x zifT;;AZE!A9j8BbD{h_Cuo@cgrIcaeH*9fMo&av%ndtLro68o6q|kLC^|brX_&t3w zj)0PkHOPKe^Q2yweMj9XX_< zB35UufYjquexKi{(6oKPJur@xYLzLY=%LzGZo#-*=|1AwcGHyAbeTdae&6>w!c15r zN7QX+T;**byXlft#hv?BpfV73yyv*h`~n{=p@8w8hRv%cl~zBQ-NL)@Ns!fx%h5{k zz9sqJ?QtsnLK_t%i38!;46@ccT;DDtVg1%JSgRRUV`m$i#wl8GhTWTtazegV{Ll64 z!<|N~RnV#rzRtk@Ot=k(1J*q+{pxId8PO82QZ9ZGYfc-%$SLw9B?o72?o}V|Tzd?2 zLqG5yV=`Mx*6MR#UQ&=;dZ5Dc+mmX{dg%UCh{TXQ6B9xV zok(nB#Se?IE!6)_!}MTs_PZ`&wra?QFdbz(-Tfx+p*-KHgAB&Jv8+CDmkj-VxHi#} z(vy=s38%m(?0LQrH2IeCBEu`rQQb}Dh#h-8>atui zUD%ml$Hj~~8oCw7;72IO6C zvJ;}MGKN8Nh^X2UW1^azL~nskMPObGJyeIu=G&$lnee?=(B~%@+*o&d0%5o`V@_9bcqKNonWK^*iof@S zr}|4`rwX7V(x9mUeYJiaTFPDOs!qMp^B&rPl{zn}>m0Ix6iV4r$&;!99Bg=!C_-~R zyEDMd=C&a_jXW5>@G9V)ipJi~0p^_dPa2#a#wK4GHD+oY!?K+`Q!XTBF<_Yaf;KGU z*(>cx@*@5(p$u!zY{v^-RhXE;M1EvWTmJA?2ctI-ZG+o!^iV-`BTJ4l^c$j}@#S-< zI#%xwr%>hjzSC6xlfP*csE=q<_D}Kzn9%oxY1z=sr_|Am(+YZx-R~BI0V7(}Dr!qu z2m%o-*jtC8eZZ4105C`!eq%zVb1bZOrv^Rdwk7e)#rrlM_!7)th*MJHhOMv*6!2d+KB29~u9jz}f$>xBeZ#{*Qv_ zZ@2uvr!fAbQuJ>P;`fx?zZP()90N_RvO73 zV0M=}jhOJ1$PU2uIO&+pch(>k&MohfTYtnv{51V`NBzRbb6I9Mu^1;bgdgryjlb}V z=R(TXZ6F@Q_B9;`18_cfq!T@{5bWemV=cQE-_ zaj*6>_9``k#eF5!n=Z=S2@_JOdXhc?_3nS22EL%nbj81?UneFlbJL2l9{~%A->Bs$ zrE{Ue{!{@q0>?5stjBhT1+J-9c?If!8`C7=u9Wh=aHRfqz7xi4f%6sO=bFQ5-_3un zXtlO&>OWz=!XF2cE!ej;Zfa0+@%HK_9h~NDt`t-(cZOM=gE+QrUb?xqe^8sS*Jg*Rh^g1$^HbPJlOaU3EITcWdDq~5h)4-;tXg>LI>f@g`} z9y2-%WhpUq1Y(03m_iSY8zPtgV3itYxZ(SGj&qF@efMgfx}`Smuk>;y_47G3DNYsj z^R0FXuouFAM6x3OYop$Ufb5$i%IcVM-()@NWpU$cL*uN%A;A$@dMIGUtuUQ*asS>qP{gLwahz!g~Ea?JUu zTCVr3LUxo&*E=Xps-i2UntZaM8p*>8;8-^^qC&W3;Oden>%GQdg^eH7^xDlH)vMi0 zi<07Bb%JN74P(FxG<^HZP=rOrkw+*rB%$@QMsiXJjZ=jQ9efb=ni`W<1?%C7w5TK*# zl|oH+%6|rY-g~?U3FE~jPG1EPG++B(=;?o6f@^&IWvfNh#P@63__)Nbh`CS($ffT9 ztfWT-0*i5C4-K%Fw+iZsyXOJf&DVHj{4KAxzJbn&n#)9AJ(2YU0 z)nFMpl`}jXR)`4@r1=)VF^iW`0i1lbf_^?jq~ZKOUtcaN%#MHhUx^Y$*smb<(W*MI9s=a&ngYk zP34%mgD#TOX3JQw!zVSawK>}hqAEZo_n-u(wP-?IfD&o<#R`xV${c}Zd~hUagZ zQKy{UA$q?n^kPI(M${nTCz~!e+i=JAHeg{PhyQD76m|Mc+4gxt>8k)XLrw3uvmZ^z z%B^(i*o);%!saMuC6!$|0w4$0z2SA5x8+$vW!|*zOSbWu8*7dEGq7@twXGA8mJSw~ zN6r^p`5@Mw`FMd$3wTrYRBO`pp+aM)hAb8GvoDG_gFOVaquy4^cyg9@NQGhQvs;g> zzSo&BBeRsVQgnLi+M4g5RaCDRSA%j;1u`Zoph;eHH-)G>(JxNQWE4?jHj(>~veRiE zT+T3UhucdMtC5teS5FiiCoFzN>@Rr23ibUlxRp&a?PJbSH#3s;mJJOO&B_p&u3Vsc z_gJR(@mo6FXT+C4*V?9cS6rXJp;&xm$tS@92NlD^C=RZe4sEX0_$vAPSC}P*ntxmm zYpH}H=jx*^!0VQT;MPkL;HfG#Mhr0mLCG$I z2FBG!H*+v`{)&r+)(QJiL(I2C6EP@uBSMBF+GAieg~W{&CMhKYU*Trs=;fEotjEJ@ zP>Lw*oyb0g&!bXc8Kjatj>8=2F{^+U@ko%;5_PYpNR8OB;Bzx^+0kgBPIK!Ed(S*we(@^Z^LF)MmvF{Z+fA#iWjWt0(1D}RkGidNmGm#^^%4WF z8z}{T{dpo%bR%<}$W$(894OcKW}lcE&q=AK(Bam3t0%AFCy~*^O_x zBJ(&QXkPY}O=%1}WdllKIE(bD_ zzet$xcl8|rkW^HL2m8SWw(Xu5(jqm0L07NP8RC#MwlqMK+cL^jg=lrr*?+JX+CG7H ze7kElTC8Jzz0aaG)zbBw^}w|zGxRrngYi2ao7@-5bePhP3|p(}2Vuf96{e4pAK$Db zYl+t1)gqhbPud$56>dMs)1jBP4ub~iLIw%`;6}tEi|usjyID|s4!gPe9LvE!io@^p z-oJ;fe&gcUAveG}?2NyA!#95G5dTmuem_Y5wSD>DKRoT}@w+JYzgQg1I_`lF%|{0= z+#eF_++UsRp-f0;Kq_D8{@~pRvD8&qS{`t9!=Ene5(VVh^k{%Lj)%*#( zbWn)o8z6)!hw(0d*5FlghWbDBnjZ@29b9Oz=7g*J%gQHyhwY^8KwT}W`gFKB-mB1YiNd%bt6XmruglFf&sDt*1bA z@(tfYQ>^9G$@{>YX^h$`>iL#cJLeDK(4vR;ZIg||l#PHHTjBinvgrcMYr9SKSa|h~ zI&lHT@YxQq%^ z9q-wdRwj3w&*+z}B45b!sMf!?_n*nhE*;wzS!zG z|K=*Wm(0@wkah}PcM0Sw59FP54J_(kmy$y=96iLSqsgJ5`r%ZH@Hp-PNg7tJyG%_j zim>)B;wrmT`c;VZ=9vK=W6EzVk9LHJ>xoWne=>gcSxWO)__b@w&4;LoI}w|4XEn%- zWVR|i-3^LQ6Vl;{No17h3(Mh^DV*~7bj&zpdKNG+;D)20rfpAZk^;uIH)!+pwrbXKLzk!G5Pg}wi9HmE^%&nSUwWx`P_M6u9K z_h1psTJKsW=j~vi?8u(w>&C2p%zn0{Sg^Obe-xbK{KUjp;#~wXtB3id&<&WLrCDZQ zF%2nz^XNHuq7GFo06u>unTT#QAAc>b8eziw6O^P!0t}uj!>unEkGI?HipM7z=O?}J7j_Ek}RZ8xKR)LpZs?OPpb3YLWb8=HhSY!1$>DX_t#%fgsH64Y^JshON z5uAhkVVTlfdKJ3|fr6r1V4Hon9l4^%K~=L&22XiPQV4cd7Rl(2P-jB2irKdAgDy#X zj?w_ho{LOg0G2On^+(%~w*!DO_h4L?0|So0xSt#CL0#Wgw48xkx@t$-gq8gYkuRk_ zgGaykT{Ph4Cv7=zgR34My#*&m>~cS$7s0s6!!YW+kEx`qmBu67p(Gk7<;*XokZIL8 z&BLS?B+0Or!CpNHYrU0uKC|ni7gm8XGd9nAs4V6w`gQM)diiI%_RC37F1?zT;?&j>46XC z47cDAzC>a=b(ca_*38I=vq*}}$v^)S5hQXvPqbak%`aFCv*5bYg0x2r6vxKZ&-7#s zKUC?n2%t{ewZZIF`Yh8g`@SsQx3{wuYu3Y>vMp|-AcnAxCf*#kfxy#cus$)2RBCL8)`d##bg%8-z7KK; z5yw<)Z5*3g0|#(6g?b3=@GEr|W65!NqJb$hlWyYeiMa%uN-`{ch}n>`h?(1P!~x(5u**3bi7S}5I4v=phGeYX?yd`oxU6FwQPIA7IxyEDnZ%}aT3V3yO8E6fYmeRaoGy1m_YH46U3$prnAtwjQLRE4bv%|W08o<%!*GJ(~ z>K!ntbD_3@TJys`lTVc3Z@R|eh#{-v>8&d{64fIk6L?)+Pg$Ah$N96gy<2+Xpmk?? z1P_1mO{NHI*G-xCWCn_ki%^+$OsG7u>{Nr!WO%TJCp!>kec{OTI^4hX?nR%O6^cWI zvIcK&OOAy(*xvex`5zY7zX2$he?MmA|Esqb`MZ7K^lzKy|DFT=EwTJ(^Wa~r;s1mv zYhYh}SJbQFUkIGMplW8-bG7Rm_XG^bzk0369^__NQ-uDeNqFYd)a^oXC%Ql6EZ*et7lL0Hr zS8)c6SClfec+87OcvkmF+;rhS(LhouYL4Uk_AcpY6*|CkCRj+_ak|F-bZOpSRA5M! z({$sweiEe}m3hI|t|I`H?8whNSNdW$;hvkd6rVYw8wGTAi^D=3SNt>FZH&g!b!o3 z0exO0tGhqQb;LtoySO-YG0YREwSZCb;5cHnA`(+j78-&fecbDN&E&zd3qyRkkaKoq zcQJBi+Kg5$#V~nm_i<8!a*b2Iilvk~CBaK&REEgPHRL>vStTv`d*4_8f&={w_tyOo z#yEE^(C>`dI;SwqQd68A2h{Ud*A&=4ZvU#gxHzO<{Q)5$7hI+xdFQ)joJ#8(w+JP< z%vEOn;`iGG!)zvj3-R~?)Fl=ZkSHHG##Wk?Q1`To%&f5RhD7ip3cui8p}JqerM_eS z;{FB^iQ~5^* z1#f+3K1JDX5?dMHjX3Ya)9i&`K)Y&m0glu+)?o!I#yB@H^>#2+B zYO&L&XdC()3(sTZ(GvAK*~`^Bfge6?hdZI6pVn8Eybz!0v+YCp>1<(# zb$*huYunqJdMKV@CneoI`Z)_;HS;BLU6$3&Zz^iGve8)s*pjuxdt>k>^^>*q;ksm3 zjb@Ym!dAl4hfQN^-l~3oa1PnawEPhZ)p8MkQ3JLQ*cp?xV8AxIg&IMm)(5~!fdxqx z!`n)^LgkW1K!YsI4UoP_W&0%RLYcL-z6P~{=v~20HUsz#FCB_m=nR2K++;OjsWgVr1UY?PMe}199^(1+ zo*-d1@t#n0_vnMM9JPj=6;iaZ7oyKvCc4n?SIUD7vRp5wh0mFH|8PET83 zOn$d;yN94;b;>YRji{^ND3~v_5R~Pp2m4Tb8xpuK?Fymm5xnNg?6vr)#SKOfX zqMdqsBkt@o1s#f88KVa@$yBQmFtErS+6#4YxzW# zOER@2)HztkVi{Nr`1u`G;A;;$EXoepCkG5`0{zk?W!{;oPMQau<*8fsjlmvTLaIHi z$7p^cbJ}OrzsWJP2jDSpYUfz%1-mZ|fHGxl1rVPR>p%C!BA7WZ4c#2^_b`&5Ewcs+ zsr?CW;CrjFR~(<1wKZKbiS00V;i(iG4-3g61+vL+xQrIfJjvaYU}Xqe37_?AjIw~H zPYUEy#A5#TQa)2pAGZbKM!>%HX+AQQIfTtM^ z!*jOog|)jnohj0SUx%TUcDdi0nIJ3c8jLmHv6hZ%0H264L#2QOTALf00nUUcYxM~h zDTI8f0HWDdS5F!-gIeZYXoeYv~B5O1=`+=Y=Yl>sD5O@f0^X@B%o z?ob#t$wJH37v$(8Ig*=0=~HSo8+;3f6j#b=;U=!|@+0TXz;IrL8UFyXtHa{eOjNtC zBudEGZ|=-kIL$stk>$gBF(tZm^w6TcMe&}!RH2kJS&s0m)1WPV%e|c%!hIN0wnZ(B z!W!N+PE2e-W+7m7Syy@4#E0C+#^!Y6VHN3;T}kGZlgw8ubl73ymH(s6{Hsox*?5&& z*|_pw+vWcVHvjGZ{4XO@|D)VAW@(j*aJ?bD2BYtwXZeM+Aw+Y( zcFh4Mc&1Y^QgAb*eB~U{fH69Li&OnMRK)w~(e%gC3MKyK6AwoS9X+u4Ve2VCDfl`+ zgFL|MW0+f}8i~QNb=(gwn>f|G+AnLrav8f1OogjUbMNpX$d^%rN$Ul zr~V_t+M_Y`hO)-rmWv%z3yv8T_b3(eGl-2oH$FZFoVq9j?|XoNy16}fK_S_XnnC(1 zR&_S4Z21?^AC|IX=jL4B^cpbp;B)*p&Q8~b7GlNdT0Z06oE5)CUoRO5NlsOw1I{qB3lG*-&TJR6?b4ML95v zfrG!jm`_%)K7zA~Q#m~^(=JTt%(}Dbhw)w+0GJ8a7tcB>AMcoV#TKd zdd*H1;)~X=9>_g!NZ6o`l*&7W@7_~+rifBYY->xFqOAM5(grwOyFGRH44>vcfPF)) z&x-n4c2RMJF&-x`ym*7>h>wIE4RZGiLv|VShp}rG{-qk}TY&k9$yWwo&-)S!c*^sc zwJ337MNNza^_hLjS}-IIOCi8i!| z3NYL0N9K)W=c#aeDtq+`qsKSuZS6W+B=wz8iI%}<2w4g}FPYXtvmQG7*jsuH4+&G* zvoQo%spj>d2OHytIIfse@*OsPn?ZqAL=u1Pt40CX(tX3T_PR9!Xvat1VY(!ynJkLQi9 zE!|DrQ^Q!oZ`VexGrlD}#mSq;;AILe^!ClQ9A{@I%Jny6``pdSH8vlCH)q~krqoDU z)zv8`xfIkVPG)?{5*kT#x&6?Ojx0^nVh8{$`@(z{Yyp-Y;4NXH$<7ge6JTVoNBaG+ zz@E>F%|(74`sn(5Kkg(sS=T-72~Y?xo#>rAE8h7)9#&fgT!7=#kvaxo{j>Wo=u$N1 zsxOCX*&8e9N+gy_;{q@n^%chEYGwhl3kOZ>vT;&BlKnG`DG>J|srEwOn|r+{0l(jw zhUXx+=U;N6!VYa*rE1SzmvTo}+AFLHLS{Zw4s2W1(b_1AgGWc8;dP;%pk_Ck;_f5`Dh)o*eOlFHMN4MHYME2nOZpj(2BwB4kC6mEdw^LEy_g z7N#GSzLl&vfQ@q<8?cx0OnqqnZ~6gc`Q#?PLko%dDyyuNIK0s_Pzi6!2G?uvss`X7 zNK)a#p+(i-(^>wZ!fz#>aTli>55T4nWxXH`L@^~=ju=a0!ItjxJug|=GsN43yH2F} zFOUTu)$7=$-qy5wUogSItWIC9S@@ejK&jvAttQH)|r0>R^de<12 z;@Wg`muM>C-}|(MF$+1`<>6tXUm&)579B{_=ovoHaKM3`U=RikNN^$)h) zjYpNM8wcp=E{kmIknryo#ffvH;~IECTjZ!~8x%%R_Am2W9Bc7a4&}03^Nx z(t|_e7T5s9{=#4lFWu3Moi9V#`?die#n;luU=kCK^?T1in`pBIZPV}4X0DLE_$^dj z?_fBfR!{M@_HA$CV3gH~*c4`TrW^Xt@?qS~2JUw~bSO z!p8Ee7u&1h`$~AbOsnQsMaO)q6EL}>eJ)hV&aR(g0=jt5JfrsnaN-fI1@Zcl%=2Se z_^%)ThWSsiUmP>XC(XJeH2%o;ly!jy2V24^X%Ou&d^oD-BFi>04sWpn#hqf5?Hg;>n=~wsGF#ZEERkFg(b#IM=C>)g11?k< zbudCMbkq2_ZrkjQCudVGVFZ5Wv1u%$eZQ7XPrD_xhPtZU(DWe>4a-7qI#-KVs1!j9ve5fHDtqb2O-7|@0k z&ht373kelL{=cflCAee?`wcXchi!IL>^}UgRNr~@NkciB5XX66 zw?bEWU4m(W#5A}~Z4;N24BH9muXmd%RV!kBe_(2c8E8NHc2kzxnqCO!fdhf?SiQsR zmZheZ=c($gQyvKk9}KF7MBOBulB_EYz4f;7E~{2@}CUSY5Sw+r$%AD41xLj|l)}+(!p_ znJG4AB2z(TO-&IA>41Py|2}19ilRflbOYw*b1n6^z$N>E zgBw$^E2ktM`BI2?%de8$UXAbZHS)(H$m4?DmTF05P43KTaIrWizj>2IMhZ*0TJ_EO zd*{fPi^h~|q8qqIs)F-h`H^iKQgi4^v=U?c;7aAvcJZi}X>NB2ZH09sT2yAKfdHVu zTA*y-=jiGfW@f(HYk9N!4o4+pcE|uGeZgc~()cN|NUu~J zyCmD3qc~qhe8r(Fg=7ML{^9}i8#lcbsT7tWA|0i`{`Em7OP8p}@~WF0SBORfm{crPmvBfk^*59)I&K8kE= z1y`%RYsDUfgwpT>dyewD0li!AorG58uyOnGy`P@_ zvbNl`|Gbey%N9JN*vm7fi(~2sd}fSfD8Zn}6n({H*UUC*7Zbe6APd*<4A%2)=buXa z%FD`Wj~Yv?0Zhw&!P2VnpoM!#{||cz`Q3>KVHY&LgF5~o~N9vkZi61m-488A}Vx}Hcq~8zhSjL zkxOeQ+B&vv(;{VYTY(@cbrd}s1QX$zR zN<(AT&er^5$|m<*t6$}+gnD%FNtS>tka(qO7EXg%L(XSFfeO8&`3N)3+p>ZesFQ)ulVk zq8{t)GzQu#gsP!!8+>K1vyJM<|=z{?{b7#CIA9+h7~B! zq;+-S8T1#F^x{6823klY0h^g94M9cPhaOf8DH{#tc(szGjTx&k6ULzf9K!z_dwH9| z`}fIA!|j$=5r#hAHTsZ4q+-u{wOHpXHa+ua4rV&$p#e^n095_?1+$XY~TjI6iF0Nw+*jVBrlC(}Ta-XRVupb{ZytaHuh=FDvQcRBct( z0Lpy{t$;WO!#jI5Aq1D$GQU!6z&)E(Hs%D-D8d1(TUV-fut<4>w;|Q* zrY?c;!e+>@7@Vg2!Y@{B-sxwFS}YA6rEP-A(WYM7wERtgPIghn4#SD`F(lg@2wb7T$<$5cV z!fB(es_Y;L-0iERfdKfnA+3a#U4^nAnpR-17e-x5vwSIusv2EHaxOy!6N!@XuHURa zy`Q3%SY`^M)blyKw=f<6PW#|pyI5VoKMSR2UF2}*UjjXxi)*xF7n%F>e*wB=J^w@C z`mYt*|B<2p-)`1Gl*fGE!8aZXU!N5hagKgBwfZ=t`LFoj!CU!1`TbvJ)~XN@s;}3@ zUw;q&ayluO>o6-$|I_MEuP-E)pImQHzfSZ43;)irk3q8B)oJ{`#gq`H~5QgQZX zX5ILDQBw{?4h_qTr=-ep6xj%xx?0WVq)SKqaC8+{8qyA=kQa7yuW%k8Fi>mQt?|2t z?U~O&aq9W~_Cv*0*R9z}LMh)y`Z$MTGuLT-;JA+#O-1@cOyISO?^^*;m21~1y^n3# zp&c02V`%;Oux(*!*@X46WeRio{)CoGGw2rQ#8rmh-A(&idXJc)kEf2bPZc znp$pjs6W+RA&#MWN;hw#cCfO;NW}uZ7MGH$qa1I~e=WR>Vtu}5wI^m|bq(c|A zA1$V5SQ!E~5F@C&%QN40)*HNb8L~&@Wqbx62~wXgMAnGp(-JFx{^_}?(J^?*pBdsl zK$Ct8hpv0B=14UPMGQ7Ai?DYRLpOBp(OfXQIgL{v-=~W`O;o2vN_~s27J26!iQ3Er zL^RkJuU~&&CwO7kv4bC0=$&Hdio1m=kGIRuQI#mp?*9Jr!9CW>Kiaa9(3Rk1W0-&D zfYw~`lEYA-Y>2jrzW9|2qWn09=Se;ANigRpd+kM%#BwHuw<-urQ>c|I+YKwXO-M|rZQar&6qz91>djBfwq;iRUOj>!m4wc3};Zs+gPC6+s zP%2^`ab9_Z$ui(@Y%ovrh%s!*A!&O1Pq43`dP3!-l?LDcLBrWFdMnOtOD(v9;s%)l z@RZp#c)?VD-#CYUNX}1Edvnh|jhZeu+!ON(=LZtqdjf2_^!uo=w9pz4RkbtN#-|oO z7H?!{GZHid#p;c4_XUPvdfG09opTRRJD7Cqldl`DuUPY0YU^QcA(Q4>y^n#9^!G-v zjv0;VhoS;6cFa&8QpRrc+5f~wUXGrS-$lgGX{3PSf(0%2zDk3YANlA5TYbq5(kH;lK>lyN0c<*8S&q<%EQ4su;WgVPGDggst2 z_5Z|XwdGO&(1IPUTe9-nGrVFQoNSaGUaG!r!83qLDwkcHGE}q%iwz~d#7TKFX38g2 zTIQu}R512w3vN^@du_RIV<(qjWtYJ3%F3F)RnnQ71pag=Y?`B2?;29yb|jT-9c||f zeT$KTfOnUzY*9-uUd7i=@gt+k5QriE3vJ(1J3L=A#U#yMVr8cZCUW-d26D*rdQGpl zT~;q0UX{)kO{!6+Z!vL2da3X{ba_1&dH3mkwZ${C>%P17qFu1Ppm^I%H^~-(iIg1T z4qsctM(-c@Ul2Yi`N9oO{)(q{$$zT*JSJ#dZ)!h1l#y? z@^M$^)r>!XrjH>B5@Gcg^rAcB~6gqvrEiRUFkdIv=yvI`}iNv zImUW^`K2s}qCuKox6<8nz$8va0>Cj*ek0<*SGgS+%gjuz#qL&|uuZn(*=^pdD=j2D z#64I3PBV;e8%gr5=Sn24vE4raJ803!-Gt|1x&>_=-#GBX_$ zO142BzKmVXDJT(+Eh7EphsG`BN@lfHv5qQXA-FvSncoMidv`~M^o;C({}W8yvHd6F z`MCIMhmR`lB&)gFoB4&2| zw9(>GT1JUW>TyX$JU(Tgj^USg`J~10cF}eNoKd5ax96;xI<%aqE@p=au1A<972|G8 zto*(Zp&YEO@AXm*73`fjovrG2D3+BY^FurlCsn+&7y$BqibA6;Kpx-GH)E}Y<6d%o zAGz&8A?)~`Z)xW6@vMlllP~sf`vl0?( zbxk?I`ajowRuy_VbRA4l!p9J2K3ItUG*l8bbM`3v{O3bIX@XLMXC7jGtgq zV$XIdk3aw6dDYwRhj6)G#8;BfC=RyqI?bXtyRdGCsDKW(b*31c~R=XM;n+NjzB%%OaNlliK?4HRAsigVPdJ|H*wnqgg)wpYRl2 z{eSzy|6OmtL!nrWfs9+IU>R>8?x5O6WsJ;$*;~QM?A`^~-iM>mIFVi~dEN->%qYBJ zzWrzF;7RWsZg)vKZusKc8odyJz)9^l=n zJkKfD7L{yVn?8C<{DuRu)3tNHJ|6&SUohAc zg1g~w&l6eqC@D-FBT$^b$=;>qrN}k$Wd){H2c07Ss};jJ`_09jvs-t$FN6Auv$eeC zgZj(kurd4J;y9kbw)g5IB0#}$7g=tv{hTc@a`xdhpKBokyL89=Es10D>{^y@9z680 zzwK;1_s@!n=X)W>C?$t-d`0ZAv{dsSsIk$xvcUycN@Fqv#+3O$Jgm48n!jXvZowxZ zzXvQA($^h4A@~1_uE!Lw6u5_`-!#TbLGFMCdYw`Da8R-n+WA^5&Bt*tkQnCFBjlOE z=%|)yuKeYp2e^NQx|-sw{w^&0g^*WiV*KC;wXl}R9)|OnJdt&MqqM2o-7@aA%Jc^3 z96hb-Q0lR_b48h5(LnW;#xI(QIHfbIEOXy7zsK&z^X#c>Se-arN(h|U)jaFsH^`AT=R{oew7A0AnyNSpU*!4uMi4Ikf%-|^tS2yK9AWTbFcJ1`U;QAPsmV9t@lx*5C zFZZ{ksTV!?8{x$%+?@KI6S~0+k|%M7Z)sC9Ypf8d->;4L9MCR_5DI@?*}TcvVd%W9 zK3`nWV@?oCVL$AdyPvuKbO$75@GvQD%ktM;xhlhVQ(*E~k_xK{JG26@IXA+S+v;aM zY++p<{!yO;l~5v%3%Thi0KG)|EZqP0pnintE@setO5dK?49q%*W@>4pZdkhPJL|77 z!J6;0v#jo#*F(q%>|%C&##DBS0G=@~44~PQGW(CLkcz>U)l8Gv>xY5Zkm)6CPh#1x zgyx!EB}S564(%n9fyyZ5qT?G2)wy!?dBM%DAXW^p7J=PU3`^b^z4#?>7Qz6@GyhKfZ*hURaja$EC!Z+7zy8WwS=cDy6 z`-0K@voS4WHopYT>%*NK4USE-8!zdKdX^GxzIX5)!h21zH6P+Emascz2>H$$?$uDn zde>GcACzg=p?+sNXLQRZ-UOv#6zuD?PLYFG*vLFeRm9A3k`cngld>^%6R70UGjhr59gCd! zzkp1gS6b3pgVj1@F8{5E<{eqs1JIZxZM55CMz6QNL9wTj;D<6 z{eS-&&10c1;a{wE#9DCz@}pamIZ}}m%W|Jk9D%a5d1Hy;JjZZ?!9-7t zIZXEV9XM-P7~QpRJn*Eigoad4Rno$XKgi99q(%>kHJyEBH|sR$cy%}U zJg<#ewQ#FFHY`Fk$4n^BUm$2vd09aC91`1xwC!5Y^LhBR13^68pYiT zCmN)ov;-rpjEH@)0cKx7HP8|b&|HeJ+T9xm_hjz;IhgJFo{2Ovu(hfKVT7|EJ($z0 zZDhAkVZw!X?2qPJsAee1nWB^cv+4l@fdi#_E2v+Fza_3+}t)v<@Q2DJ)&y6+j-q; z$qwLL%SlIn1awhjP4BbUVx>l{4~u2cq&oTAh*x#d*%hB20+{HVlLM{ZEB1a0lRzvH zBQ_Xq#%V9wUj>ktxxeNK{g(dt5r)2=zu!uW^xc}}0t;v4%B-4V=ODZl`P%5Ntl&27 z_>mg>L|-L~yzOK@0J~J;JyrTZGmr_$!sS55mu7*MyRW^~T=nuF5Cau&_Y9b(u6lG@ zLW@3K#+on2dr?38WAJs21XLl1Wc`N-jL&z)q? z!3kMoAZl8_C(fSc=3|%pWI#q_0+e7T4f24|J9-KR`0TQeeIWJ~MuK^*zj-W}Q0_Gm z2wd}Dq-J>y2iA&}t@3`t@=K=YKTlCCX^oDKWmpHmRIiBM<2Bz#aONKoshz~j&(Hi# z4k8c40!Z2%lC)rZr~0qxe^2me_5auO|1nShPeJJa)nYY;u7-}@@XH0q&tKkMPG#RR zk?*~|(}ZGu{x7-_o}DWZ=QOMgY~``=_SKTp;3%ko%dM@c;{89zdzuk{Ky#(7hWb`r zjBRhTEIT%Woh$UqYd@@lZ*h!>t2_-bLvAVr5#FdT_#c4g{8AwHvqnY0B^u$ozB(Fj z;lnp;aUl*nWm^o9Z?T}|Ypy=8ZmBlvJgV(PUZ;s|9}8gx)7T$pZy?@V#Vr&Ed)N9~ z$3X7j^e*H-&c!*jb7 z$vop8a4T$9BidM5kK7XS|ZnGt%xkh)2r8%NE*3`=jOXg zpZY>>4ghO(PL7lo+Dkqou=8;P;$O*$J^fr#Tvn{wXbj}JM8o7yBwv;fVZN1-SNs+X zuk(tvhyCe{TI-+e&jQ^_+hblJ5ZNjwBF15gaG7*~;rXhFZQN7|UvB;^$p|WLKRT)L zKy7pgcxqQlN81OyMp*d;ReHMFG3YlYqrwHx;Ckm-F(;KIPQh1{4%i_y&U;-iaIToY zOOp1HZqAOli9G!ZT35kW@%8Tq^G$s5x;Tnw1v%~-V1;ct07Vj#n4~E5Xc3?8g29)~ zYTfxHE?J1`(ak{Y^PXZ1%WWYJCQ*bv-a&4R#HXgdX5s(r_8R-l6oi4b8Sn;YV@Plu zU(BwnLP`v8hRQ1 zulbmNEHo}w`LD;*u{Z3&Q)yrs4o(@X$ccv*h4E>blPqheWB&3b?ji5@T6W=Ksl7@c z^*{JB>Fzq36H%Evp_hsV9ivf17c0q)u$TX61wZgd%G3~c4x?lxJME8vh1LP@byM^o z^w$jr3ku6#?6t=3&b(9hs1^jhZgG7xjr^V%Hps=Aid?Lp+u!IuTv$rB=Q29907Xw= z_Db_jML4I-_z(rNizoGMP(D`XQ=X%pfoGuc*E0(869?tuAE+Ho@gJOwU&U9$Q#@c# zuD-xQ)m~DdNEZv5>G^8>>c)i^Z}=(^t$iKu|J@gns!o_md7^O$$C-y}uEv*I-<3Ml zYiOAms+sV&X>t^5dGW{kkf{l|JH$2_W1y(c2NuLTSMf)gJVOND;9H_T=2i}xNtzgHq>>P_>^F_3}N69w0z<&=ZAJ* zHEM0VrFE~mVuQ5xQx%&$T#_WSgHGM43#fx$+yp+jGQs-Y=M?mI-5{1j;=J0X=B@u|=2-P3zvnnl+ay7MN9DSnR|?y%`e2Tk@oP-^{%rjLQtHkzLJBCW~*T06y?M@>PCUgA4+l5F0dIDIGZcgJ*9@v=zPF^u<_^Eav;Sj$apre=JdH z0L?#g_i*jfVi|J-SubcE-`9HFuxtu1Bu@ksBw5Ra0BKf>hkc338Fe+a84porBz4*@ zwk-l4c>CnxiWxHfq4o|Qc7M&HxlrMOt-+Q*!nG~_Vt|lhnmaz(xQhOC-v&Xe(X+Qy>mM~H;;tU7 zv(rZSp_E=)rPz$vq$z&4rxiHg+hp-txk+4`SVpH~DCCu>vL|cXEh%d_R6GXe=Q-iJ z6JMsrB!o_(RwKlv`iJ(dgNq*OmrH7E&IxI4gvbq$9+_T6zPgoIxobAK*}z!rWxKW* zfCguNWNaInsk~2L<(fKYRbB}ii8ZP4(*%QB1RX`_)lwAvZ8M4Ze+g6 ztWc;oNV-mjryM{e@m63px=P)Y3Bo6NV*lP;Q#_{(l9W#Iab+OGL?(2`Bb^4|I{@89 zRtnif8L3yFwdNz$Do;Ni1v^l;G8|f4#kp)#l^qu449YSG0Q)-Erkp9>*+`|0rlm)) z!=A2>`FRnN2>Xi3>7cwlQRa7H>}Dhf?+fE92uAvoi8qzkC{^Y;CmXFRi)*1K-O0D^ z6s?WW$r*gKcO+F-Wn?Eej|q!^Wi}ZU5P_}be21+!1)zxw{n&<6E!e4aLga|6;b4fw zbcof%LT)Pl-$aM-tQ6=q9zFTmTUIor*wZ8&+ILJqw<4QAV9Oo^W4b9{n0^XJwdq4I zT3fpEo>kV)>q)DVVR=U0!97=tH`S(oS+F2_7tjhY!Njt&gBNM&44QuwYEL6u4ok|i zSe3+lE$68!WXnpV;Vv1=mD2AFWcHcK{OH$D8hd}};G0k$E+d`&*gNv|tOJL^alv?+ z^QGX%#&R!4+Wwm5EMSlS@$W{hc6aIgU+NsRWUpVZza7&j%rf+#-w~*^EGv=j9;d$*r&Ih-5 zx|ZD}_ksLq|3bmp*g_0dCN1SuZ2bLcwT3!A%5?sIOJg0wQlTLKfL&#gg33~()wE*v zKm3uLhD}1i$}w7gv74B0Fw|`>Ai?~{(h3svyO89?$e8t1PuftazpBJ(#7j)1?Y_^z zMQg)4?Qnpjk^1Lb3LGR?xtp%#yu}-r?I}F!{UB2Z?YVHYFADOvTv{#l9p|VZ??954%Tji1 za}9-uYeC=GA7_zR16b9Wb{*Z^R*Z-)%xQ1W$=HFuuZGde`x6$ljiC>)7#gKzDLgvU zG3i6T@V@Jm?E5de?f?Z_{pEd!*v3+HP)ni{lu9ZT_D}zixL~GO6R^_MA0*ikTx!rCcWq&gje_GxoOsSv3 zX6?z_Idf@9n{L-e;$uUDip@fOr`V@5j!-$jZd`r}9bk_Z;akLjOWG^SxBJaRLTa?G zEt~Cqp#1%Mq0Ntlm6eC z;dwN;or=`5SUHJCG8cMGhE^-Lw@0*K^&?-wDo_7lYaxCoS2b`2=jMPJiUZ8eH6sc| zfa9Nq6KV~%b+udp>%(QV?(gTlVe?cubGwKej$XTp4ArVDC8ghL%`iNvYMu1{RLMMY zkz(LU9~y~1SL^l3N_Cx_)4D5dHe9fgejW!x1>>1Y)x(KX9;G(lyBzL(k}|8Lo^}$v zLh@70U)^=Qo%zu31JUoZi2ywrJ&DVQX&CI67MSuIoGx*6=nwK1tCQ(Rde0TF+~n#! zom>bjUxHu!e51Jm|14tZ*~t=pP;WR$Qj~Ey&8UKeh|FUoXYvz=AZV)K)KqzD10wFNatrpO9C% zvjyAbNb%mD76QV4R{6EydD-H~9G2Mmg6C+A-%x7lw&>3o>n}40q(NDQFPdQIGTK+z zSZYED2l0NCzbs>h>T9JOI1R}*M_3mm^vni$U)PBkq&q{VRS0;?nrNi?ksCh%ilU= zk#1DF+eK=~H`!eUHu!alKGs zGX`2EK5LoUQk$rRG58t`e13#F(X@d}+)KKYpwIIJIWirzMIup|bD2vwA6eKjFc+RM zB%b*wScMW&gsUYt#YUVIFf4tu!O!dU^&s9>4%T)wACUC|6LR?g9Msw@@$T{I{UotN zH9c>8V#He{|22g<__dHB?aJq^0qPL5{xTz+b3dvZ#$jx2MHmAx>ylL6FaFf!leNvX z9THhW?u}D_b-&4&eSq_y9a)t($9D)dsApblQ0;Tj`ibOv>`-3kG5zpv7VT>Mfv?9B z?BF^FVj&= zdz! z??cHC8--GEb?``n%6K@zwKZJ>90=wA*aptw4AQv&g_6Ns<=Mdf+bRIm5J^Lk_w^M{ zzmdsR=d9wJd`xmD@{BR5qrqI=A~g#+)TQ%9nQ?*mcWmV-{+P zxwx(Mk9voE1ClwJF1qBo(0245f<=gmG2GWI2$}E8x=>29OUbCXkRr}=1T7Y|2ELF= zZE<6|MwsQ#{XmSNR3R7YDU}rv)$gRy)c4i8M~`A77abO*=(gCr?xd>BIR!rYl(POW zI_B^GRTND^ny>F|ZDL0fZBJAI=j5VeqR-cip9|AsIr)SuGadFaPo9@M7rUZ+OWt2c z<|{vvVClV)${Cg}zduZ*n_X`@|F&d*Ep!YZ&ChwP^^*pRn~xMHS3B-_ghJ#{Y^z#= zmY3dXJj#oh%LI~r_ZCzs2)&usbPoH(CK^(~1ugxgLk2?uscpFYSM zNf&P`jDF%3ATreS2_kvR3TSkK{labt5bMdW$;BwEa)n8EEKwT8?W5N(By}Af>4t{xl&%YjS{s0)INZV@vn}23;;eQ`Q|KCUunb-gAGkyBM53~P$ zkk!=#xKIE&_aFR0evA4;xak(cRz@DvmYNpk!W+u&Rsx;!h~;`g^!*wp683c%QH4&obOc4y-}EBl`R}iRF$h zba())4jg#HQ?9+ZdJCoyKKpIFBGxy4xz8?T)9->EEgusYJ-1Z)?Z+w*iXj$1Jerl& zEDiWjpjG+sJ2dcX-CYAcWxBKCmQd;=`(E;N4xUR;B7enhz@; zUZubL7CV>XiNEiXZnuT*7jG3mD5UMKsbqP3p1hdi=U*L8jx{3r8*~Z& zI;r|a;$K)c-I_J408H3?@7DX_n<+j=ZPnC9^?%hhyp#E%_?E_rb3EQk=%|`c{8h(M zsR));4ei>a!{9$FL_+m5hW6Siz7Z8Mqxh(}*__Btm`GIxQO2b;345ptq3_@#!sk9T=clAIKyu~x*3TyTpN?e?I4~P_y&0EDagj4r2E%QJ1$k0S@n(x@mIL7{YMBmmEClEdS^AsR(FL^HU z1m!F{W2n+;8*n8PS$kSF}xl2Ps($Fk!afBb;*@(M!46wP?vV-AtJkxm>T zWsjk+hUX=fSLv{s%9gCAo2%Y!@OAJ!4=FpiGpreHdnlqV&e%RbRVY-Jss_VJdlC6I z{dUKsmOrOr8eJsLDw6KhPWk$<8l}RoM6v5BJ~hL^TsD6Lr4SaoW|Lbs+42%nvj1={ z<*L5Gi(whx$PdV6YZ#5e#RX$5K8hQO>RqHQw*x3r0MSwc&P!lI)bdioz9Vk=5yw9= zdZUA^$mLd%ScsePoh0`)BytxY7sg z@J!?#c^8+eW(GpVsSjM;+2ybS`UH6_^ytCOY2|*urzfP!an-qwmJw4Din5L`f$I6M zWWCFD)tSO#Ov#KC<}>i-Uq7bF+4$hgUqs7ZY@0)sjr3D7aZ*90!>5{~&mt_4qP-g} z#3xtYijNupOh-6c_qi-?aY-4=w|$s1 zl|l&e+Y^JuN^ERqW?3v{XTP(vqrKjE8}b1x2MVz#vOcBR25uIofXu)*_kJk-E^q%i zAJxg7V}nWiusRBz1_<2Uw+q!UJ}TUou)6jngy8R=@` z8zl@vt#PB0CSWc5n1Y&$Unze!wCSb?w@snzgMX5EP;cbFfGALV6O?|h?BLL`qr@|P zyOF@-dj8@Nj5r#2am}b6n*D^r``C)AcO^mD|Il$mI1wjs>cRSv7;<(O0e+_AM^|NH zZEfHh*v6vqi?Z{xQcocAqdBHqIt{&H_`r z$BCmBpKl=(O7*1+5n4#>O^&oE)SmTaDa9{yK}xEdC5nX5uVS`x)Y}2CQ*rkT4B@6r zY^VXnEfW}@V88{umbOL_y|D4R)Z3g}%Y&ztk3?uAoO(~olv~JBe|%|b{w+?&DEpR2 z3^pr*4p@lARXxQ!p%?4|w!Lr79_zm7$`8?E|SLqI22ss`7*R&#)}nfgJJQi__2XVEeDPU3YC4 z=W^D&Bw-oFuV?87TJQ-jIn&y@>GILWiud}eA_E9=Ky z+%ORK(%y(SpTQ$>aSnX%ZKIssrg(I5!zy1F{CQ2$WrIC&SN{kNafi#2BN;lFes30LdV8nnSIR=dGP`aB9kdp3h1^N8; zx&GJnz45=xEgsv>$NTkuozLfSs2v~L*;@K+el{AZbxGJ{gQz2M>o||3W}d9v(x`k- z4bVGOtuhqR(=$KK@=b($-Qf#lCcl>uMOb9N=3AKcxvPo^B&Y_>0f9kM(#B{@dQz@K zGe3|A-SJ|j{;(9tgPq;8bhwOoezUaI%sKR-0Hsjcb?Wi1dK7`KY`TRSh&1GB3CftsR%p3a4QLw6zr zVWTWP5z&O@n!K;WobT-)j}$$e^?wLKjfj|1n7(7wNWwn(E2p?<9n%-nlC9E z)r`;OJsQgL-brC~P4K^LCI54`f%h3`$gycj^7@Q*ARa>`VQ@2RDL&wCG%@prMwVa6 zeUy~FNz%fcV#hXl_*Eg#Dzo{cLczO0q_+K*E)Ptj9bnz7&aE?ur}A=pM)~W34Wa>V zq*?|=$K7INJnt=pzqa3hX9vRM&Gh?v6n)!2a1tDteUsk7B@X(vU8ic{TKR1Vbx3=N zPf8fz*S=Ox#X9S8nAF@bi`aKDUYIw578KTT4!ryFT(4hbpLm4NxpZ3aPGf5|{6r*F`)16WdArp%*>u!E%(!9uFEd0nnOn;}5+3Pcmvx zO#()spd>85)gUjEc0-@)+>RbChWv=+qkqmgNAH@=yv zeD$Vs*Mv4cv@o7s8!TyAR!EV*gpfq9-%e%CO;>KX*i`f8mrs3Tfv|e41w`$-);Q!? zHX8}x482Q#@}{ej(h$#2-{00`q7d7C=h(iTlaP3?=c7cjn##QbExz0aX64t1`m{gC zRz$=ONad>q!ve5o$bztBMjTJY?LDJg3%yHlZvfo7txa4QQMsi%lr*7K^_xj>El`*A z+mXsd4#nqgaDhoJ;Rig$JGSlLn)GNaqQZm|ypP4mHSd#dq`9C*q1eK;`1@y@P(c0?MVF$Q3Q1pY>lpFWl)p ziU^^DWSaeAexonmd)Znt{Vp76571H6Q52y4Lv+|_fw)XxpN1QoG_wo|`h$%kb+++Y zz=D-QplQ#7t;r>^<{gz82lGDQ%S;04#91*(pryHEOnzXZ<*-MU&^J3s3gL z&4fYMH(uRP0l`e7!^0MCVIFm-rpDbY9`K0BE9t9$Gi?PjsZxmFL5{Lyg+$fdySn~e zB($~cdCKlFIB4|!n0M1ag9ZBWF8%y-dIqhj{QH(@_&bMzJqcP(hIr)^n+c^%U3#)? zjDYnjl6ycu1M?V^v#`GnF?;ajcLbfLR)Oy#>5-(_P<{P?A}jp9chL>$vI2Uf9?BhU zvzfh8ZWXd9mh&j`DV4?RvvLLUH!w3KzliXT+SnZ0oDLs8(uX|I`6NlyF@3&O$~W_# zh|oL$i(h~;e9v&~bd$1*I(G3i$<}WW_h2RbjKsCMqqJ@Oo0qaT|gi0JgQ&+qqfTmTia=n6=68% zhrIYj%zR1}FpW|XM9D^bPUK`@3gd7~x*u~&2Zx_cY-?`SiC%`OIIKj! z$>dFsZ6$TgKN`9di>o%P%QIsP!Gw3c^rv^(47ZEksAr!3g|QlR;pM9X5ByaM(EG*- zi!U%hOo=C#7L|QZE0g$XZCg$m^Od1$+ftJK ziB`I{MRK94MNhFx?SYk3K#KuP5D%&CFOle}1Z0xrwDkrYms4Jan`|SvW+Ed{0b5z- zO1_PqBFB^$kP8!-Uqsh%ka&~`i-752)`+jQzW+TtnVNRkV#`31H@nZMWgH__ZXU%} z#v-yR^N&IB3cg@q9v?`eH*Xl(CUhgEB?;?jDeSW-whXi)x zgT>Oja3c>t{MccI@zL$UetxVbDbyVDhdV$_Hr9yZo+*`QQ=Lxb)zlZ-uH&O#mR264 z7@Pc9tFLIL$X#kyyeUsJNRTE12*I4E{i0dA#E6xxzHe&7YqQY$#)4VK`ZBB!59PU& zXl{T!CgM5QeNK(MW#ZimyiXh232p(S(0ep|kGyJHjJi}@%`QIo<1e#O;m@d*_6P2< z2H{u>Bo&V;Qa;uUedBzmV!XFhKa=BL3;kB+XlOs-`^szcWwZrL(pCB+zK!&s}rruXHOzGY(G|(TTI-5>8b8?) z_z{F!Y|H^ksqUJ2Rl=>R#Wy4HwvF$SA%iCsAg%_&P}5APX14pVgYhr5y?gwg-2q@nL)6Pj-|^^o7w=#m!vB5icpbf%O~ z`M`6zM4mW#^QU-GOfU`w%IL+B>f8VBc~maqH5-8Z!oyjql#-@fPTgA|uv$8=0hfnS zW#U5dyHy}9_fx#8nSN3Yq3OWUGhzL&;(l14YH6lKJ%VB>31CSVVL+!^dejv)Ho0%z zn63+APg@Fda&&2paa3YQkJ+zmT&%Mw&8zkTc4-YV7&DxUi=}$*xgQ#Sr#ouI<0AaU zlJPn`GhvtYqnp1@6Jfudshy<@M%fJK0YN3A`nV*m&-s=@GXdBDd}0dTtTseQ4^4Bt zWpx_NYm94iT&3H7w{q;nTjj3>i2}OwRCo*M}kBraU_3ZS=Y{3E3`Am^--R~scHu6exOc0LiVGqgzoMGPYi>y-fV5*t>RIzAW1Xy3j3GLTg)%CG9i#A3pF`t}T_x@lgKgLr#xV%grsj6hMm` zp!-Qv)=~p1kse)Zv>%|6Z69!7Oc@;v)V;*Ym@xg#)v6XNxVUsqd`gJzK2($e`xU>; z=X?D`)VnOw((+Oq6<*gfO~PIHN=Yy3*yO5sK$ARR0TL~q$xgxToSRgh+uo}+JvTE)o?$1{cvz&mH6qs#7eSLe#1=4kq;AG;$m%f*q8Qi*2$MmNv>OYN> zwkK2uPIxjD2ZC)_8rjv}czG)qrwn1LXS1mu%o9jpKa0{6IH8YjV=&R_)9dCC#xwLd zT-BHj+SL;yo^Gc&5mw4#t3L3{D*&o*fdc$p3uQaioiRL=aVjJ{ikf*>sniBIZ0FI@ z{v1vbTJ#sA`(;fzfUtU>0~)Ysu3E!Smuf+jHAJdB#zY z$$$Al8`7V0qqsle6&4f!8Jqvj!Qw~fsCRe`7S+F1rvFMW{_C;$Pfqp$hJm)r{MXC0 z#&AKJzkh$wj%XW&6!lBpF(bv6THJ?or2&$|tnonnH7+9MVA`X>5pOTy9OcjLsEoTd#XE+k|l% z6oy#|K4EhUcp7Rbx=0odk+loT-a|UYRvb&8r|ZynCwmWjBR?;%3_TlP|#v&+c+d$Ky_<1T@AMTPIu+4@H2ae)gD2M#53 z^f?<=3JWRV8afmx<+7dLNL7152wFF&p}1;h@49ics9g8Qk!i7ETYImK^hb&H_ch}C z)+~zl8Gmy=IfaxFM}jU3QT z57aYhW(s%AIBdkIPtP-C(tN3Qsn_o>*$+IxYdv*VftfZz-C7A+EUC7h;Wxm+ys{~#*N($?)V&*kK=~0O?GOu4vj!HhNOH&19 z)$Y!aP9Y%KbxG>)mDSA2#Oqu{({h6h>{-c6I`E@64PDWHm(BQ)wW!6< z>zaW0qk6!qj#&}GCTFct-0wX1aOT#MCZvVfHtY8k%L}`@C)A$dgx9soyXy6g@@1i% zW1i+8Emlkh1!aSWq)EL}f`#h6nH*zswi}(s3t~}v3g3}}f;)ov97@!0`=*OBYfOebQ zAYZ`wVT4PWhx`}N@Y^e02s_L}x+J9REqf`s;CCqBb|y_cE6|KQObRH{8c^(!z`VeE z9bgciv`n0(d84!rqs`8$oM?ToI{s~9KZV$vC`w$Wu(BYmrd+72gtlhA-|1P+9JMo_ zRek+y*f;!c?Q7}b)2na>k@1S9C*tC}TGOBg;W_Bw|-Khch!5ejY42P3Aq&>}pj~ z+>!Jypdb>H=JAot-@?w=(tP}LcAiA_1vZUvy~E&-@=Ppy@SegQKPx()7b~fqBBUlH zwG$iLr6i+gCJYD{F>{Nf&Ytw8(qBuHlNvxTGN)h|T{@xc6xFSfvgvLi`MdKkEOvm7 z)5bN;1dj=dS3Qy z8q!?X7N-mGah$0?=4tHivc|KCFuWbHW;EM+!xkXNB5cEi^8xlC=HWR#Bmf$+*4%Q3 z`nPz$r3BIRIJ23p^|OhG_YTR`&NAqzF?>_h_-m8%QUUXUmr7_P3zD%Ws8d#J(yHP! zbC`T?yjr|IcD1tpeVQFZ5YTo9ycGMjSM8LX!WC=de!mpGZ8#B)E>?+-Y=^9`<$K(+ zO&7GnS0fYBHy0bZpT=|4yPie&365l>Rszyl-Ywx{+rwQQD3L;l!}WtfD}j~GsK_44 zs0lwjK~_hYjdmk|!M(^X?JtiY-UMcOQDiRTmm_3|g47yfP3BLCo|( zuCpF|FMq-6%hJq5W}Ky|!y2R5f>^r|%^fUWxYsEz){6!mK)T{us zCB4--sCB?`Bd*eKdg24-3;ty7DxN;J<^%;V-f}|-2D8;#;;5k}D9uu58#TD3`Jd{T zY)vHny6RoDlpl2uaEy`aogBfG%?O+_1kcnjmnk&;-tVF&Q5`GbTQme}lt~!Of~&o8 zVBJCOntL5hhA-3hp9{|Ngyn72sNK`4(GQD@#Bxzzt>p5kIO8s&k#IVdC{in7$eM0A znnv28PzbASkFbIdgRAJhi$q}4Q32cwaX6ho!<3ZuK98Y0x=!O#Ov{}DCf7B8C$blE znEd`siH5ek`jffh|M>|@Z+G9k+0GGGysur9M?i+#8qdu+1h2bY#3VkdBgwRzEpe0x zF;>Bk%57*6@zuVi1+2jbu))ZA%ztU0>x{hGHw$1NWu?7J7u|E9bP8stoTt9DQ86`e zH0>1;sVK8RIZ#(h{X-x@VW5}LWuVhLAUJ8o8LDM)E-|)vwijx;W}IA!Yy<=|=)65V zHgoS){`hoIig&d#cf90C*)ji*mxmYiAIR}Y;=2-}v{%Cpj`~j%lb42QY8s!z@u^zA z1(_3=nH@v1xXUWcUk9;{{$$d%6)u2b^B`pcQK}%eHYP$nG7UaVQFLdgvNZ54qsZ`P z9D!6lw%*S=)Vp%I(mgyJagFo5UdlOI@4lhLb`6XLt|sy&zpf<4%l|D1pPQ?Rqav%_ zy(KuRF&HP6zu+7Bh+}4u9=5pR?9lKDUU2)-q<0d!*lIb$6{-BH*^fy}?eGf~v&WVa zHdSawl@*vG4Sa6c8aK6&Z%6bR6S&)eVYW?PCMc&pj*oh<-jy&{-=9glNVy&_9N9f^ z_;WHa`9@(y^}UTiUN#q0%$nCj1L{Cbk9eORg;%QUrNXGTigH&5AV-c^(Fslk!~g{( zL`}PBsw0c-QsPnV3-@Ncr*A^_(h$m3zT;?}m2i_EwG+;-BX0L4dS>_jr^%{EWgdL; ziaQ=WXA&VIqV!O8cSmEPL_Z@|?XqCFfsar*= z9M^Be_?HsMWgP|AUH%mTquG%HQg~iV%tJzq`Hz;kk%fGpj;P~b0Kc$vy3{xk8ADKR z;v1yx`Oe3|hV6aci<8zjD@#d^v@mKv|7_$~nCm-Mp@EgUU##euiN_ezhlS5e zU7Zto{lk&;0XS{!{@EmZTTN9JlGG^be5yM18vFSQPajQG_9#SdV2wY?`mL1pdX+vT z2Mj+1@u=_kfp8>mOfn}smG->vYHO%Hl7WvX1rG#UL36QGm)+h9PoH^c5wkf|g=+UZ zCkCxk|Iuadb8iO(Oi{YED>)-zp&Nx|`jz8WDP}Xj>cwH)vQH<`D| z!L^f0cRNDwWqLxA9h?|O_Bd#!9KA(&!VL3>0M?)$nT`2R z92}#DNMB)$xsfG4ryn)x>I&Vd{oIb1bABue4oCcRSmG-nAqFc!bU^Zh@B^8MuXS2P@@+?>nL>xCb_$CC3;BxiRSq~$q#()81qN03e=-R zi|{>BTK2kpceXpHH}W;S!o)!wcT@66e)%D5BXHcxP^oz{B2Af9?C`kP8x3;JH()9k z`sOpYs>kko;b?pi^t98ISeth#4w`1@2f}~c5t$bKJ z**UjyviF=e{!*Vx${V8@h?Q}(eC2pk7da$A)2E zs6sKcZ=TzuKU4mK@^0eN+uK|tkJcva@k|7?B)CbCObDSotEJAQqy4_^x2A^Jn@Ls> zXC8?@ZC?Vq_p+EJIxft(@IgRms2tnh9b3=#Npz)lPurTOS{&qbda>|b{+czDu?k&N z;qnew*lKj%T5Ng_EXlW`?rd?C^Ay!7-jXXq8JErXV90cZgkm}ihAx{PJ}TF(~q zC4hFT=8*cA{N14r!@uOlONsJ+ARAY;PdN=|Wl`Qca$CJ5Hipt##+AoPu~YXpIAtR- zbJk`mcb^$iAxq3bCI)g3R^(bDbLc5p483^4>p+f z`QSNKOxl)a5YO?KmtrseKs3|o^VB1%hEXcxbGg2ZdpSsLDbAb4^MB=+%fej>qm zzIMxf?9h~z6Q>@tG;QUsb~4jFl`QKmw~L*@wm9-{*fVNGBlB7n7=}E#tltI0H!-4V zU3FxElPisgduat0S2&#^xac5z&en3JK38w5PXPaubEYi4rU^dR(c)sQI!khY? zO09x)iymodPHUwRu1yzN$x=o{F_j^;RV2+r7}Xs3gS4HC@A<*pY(?4Gn-b@Q=5BxQ z3d-DPGtUoqUD4&lZtVmd=*6Mt3WAVjNL#aUcnyp!09}mGpP+_Yb44*tfA_hq|@02;+AomS7;;=Tdiu+nQBmQtC>JGr*3*{Dfj z4Oh5=Jd|}@^+GX>kJA<~Kei-v+>#Pq?yq$^SJZ-O4fR$Oonbi4AS71CDM?w73;^Q@^DTHH#nEB)_u7{OQtE#IhFsT*+fi^x_^rYH#F#Q zkFgD!?pNNmXucNvHR1Sn-?#fOsuIcH90hUspu`05`&J?)AbUuzD+$eR*=J3$WlKub z;2RV^5svZbMLIUyw(Kc0}M)0`<4Z!Z<|R;FGA$`fx{vt~k_1t6P^y%Ln+ zog0cdU+8c9|! z-we_=S{1%2Hp>^_*YqEKA8__pNHs#U#*kT?%ENCb5V3@o7tRl80FP3%gwOqq2D2K9 zDVo0XMa`QjSERCW>mVwHg3^h74$U1_eo<^FhyP-&?1r!omn)#28Vg(IXn91$NST28 zoebX+k%^VdBnb*w`h#3e`hBCU?7H6*>O@uC%MdyGFR|L zKZ>w`TZ2bQqtrYmL55UHRsVPQ3rp_P$8Akj0eW5GMPG$aQ*{Jrz+~Yda<3QsjZzJ z@JhWfDcyeJ*8-+yJw)FT4&H;}3D;W>Jryko2+PXX?yBr))>qkZ!c4cg@^U;K3BKq{ zoW%hM=03FwXg1k|u&h@Q^be~|l&bdMW7z$2{4!EzT*$^ksUadaZv~0jB2j(>5J}PVg#uDWxpJwS0;jwhXAVu_-%dN z=7gEn!EZnoUkyxzhSWbXkE>O43(o+td5EXZ*6AXw@TEO$lns7cU(rAOszBjuZaO(j zGG3ur=#2JNvhRJ&y`mOPngz%(%s znzu#k_BurF)c2w^{^6C+o3GDSl$N~kaJ&LC*nI)gDB^%W(y-H=wQjiQK#Ut z5+(Lpr>)pHcE?zLWcU+LVEXsIGPhBV^ zp+{LQf5r23d!oBPE3LBkjcbJ(RiN%ofddH8q#D>~#O zpu#V>0VVGYW(CdsgHdA;3H{VLeveLp-od~>L_dSL=+nOH9Zpn~fQ~}-DKTzMlU{U^Ud-+ZSDz=P~Jzd?&5x8MTkm z6~_HZUl#bjSg;sT)ys!0%veqlheybYiv2Or$8WSt4}6`q_*x-WXXP(LZUYTSU8G>E z=!M=*Gye3|-CeiJMV5JG*=YLDnkgcJKDJl4c~0Jsl$}-QMh%SqHI@7=ov^iN4czsok;I0o;YQU@{|L7 zz*9W~r%_1Sa>AXH-|v}2j@L0A)Uz~dsC4?KSl3ik-fIAKL+^?0)Z%RdjsacEHebgZ zNk~D~UNo$rAM7~>?2Z2rSmU_&aB*tM(v=g9rZ&%Q;~tNds3udm^=79Ol5}RSX**|& zw*mL3N*Zk5Cs$m6L>Q!8z^+E8_vq^$sHI6iF1~9!LL&!wk*-D5rDHXLN-M;$%zc8w zbOb45QBm^yacPX$@0OJW+FpyXHbr0VzMeau(>0#a^7l!Y%L5y(Qlt0w%{ROh>7tKh zYWHPGk38&;__{l6_j?yCE`fz%oP0)Wt8Qo8<-R^DONWp7Bb*tXL2IGBzvdaUjc2F4 z(}YV5eq%wBrc*98(aJ7M$j!wTl`eCRZ2d;J71%?GOPsJuPu_Rk%S(*Rf9mB6#a;V-;9=?v2MpUuLvPC}{rJZBL6!q@)oyrDtKc zxS1l<4S<{R>*zQn7C6cH_1j>ewRXi;M1^rd$#vUeH)8-vNqMw2%8Yof%mU$#*U|Yc z8NS?CuGc#5QudsbdNc493P5uPs*0@k;agak&VDNA9}s6ag1?$dgWz<}M_OBjfoUT|rYQY@%WsrFHI$hbriQdIKZDt)QhsxaprN^?-8pp4o!lSn>vW^p)oN^v>%sP|x!!B_yNvx{ z!9J+^-ubU4b?8L%rj8OlW!C!Ma%uLGP!FSk=u7q%3VORlQnw%Wi@uDiy9gAX8MwBF zl{@l(*zzaI)6Rv2o?Cvy-fXQ)G zg)q?%{bAlqILTpwdH`1Lf;OF_?*=kiI$2J4ZGLu3lgVvlPuoN!s|&={&S}vZAV=jm z#he2e>d-D8cj8`axoH@)6Rj;xXQ<#MdQpE#9p-i}eXj0Y;;G7EQa$9V8J#rj`6I=< zZ|WM&b^H$vQ_bltl8A!=(mx`X6F&3oU1CZC`|Nsp4MtnN)u@2L`9=wV=-9F=Z{epE za9?jNgd0!1+D?+A>W_H+k*K%%=!Wv{__7ARE%&}uojoGcJFaNuL#a=<1hGg1Twl&o zlc3va-acvW$33@UUq;d+w1G);(WhRXd?-bGzuMkr=0?pM-OW%buDYju*ewziUM*4H zh?F#>$*Ry?+n!H-X9NOe10@OFSgoKWm%~QzfLOex z#O)B6;-MfTM}eZwxa@2w3X21kt#izi-ia)E$a_&V%W7!4zoB5D zORAm}K##MH4HJd$YOcEDqa%K(uhKgq{k_J1DPM-Ngyd|{y7J z5uNBBADt*~FUejt4UF1U{C>_dcxWbEsn5MP;;*e`Ra=MEIgWVSQ|^DP_R8IqNE5O~ zfnRlI_TQNNd(^6_aD^)P4KuFag({Rz4v-;|M4$9PQQYbiAx3aiPv{8>h!ns(O_sBs zuvuO#D+pYFnpRRvMroxnKQ}e3;o)yDn_veV4h+$hX6(0FY$wm> zeJKqmwokkK}b?1$1n6OXSUyhvQ`_!DsNrglwm7c^^NYU?0YQJYJe zJ4~8qJ6gCWeh$FTQJe!tQ+={*SYamSdBtA3gY`N3KJpwa zPIaE3Z#vx8QU!_TASVPUnHNCWq6%2KUr&bL+DG2`3iwf=%KmrAt(aneO9CqDK&&Om z6Wjf7U0jCzsBe-FuO9&o<-K591}rFJl(sxZ$!QZ+S*_$Jk~HeIztNbaTHTb`_Z59# zMAg3958=&Z!u;Fk(=`Gr5Z`Q$O-)_IHZmrmN!X~Y(6V2pUf%=dlRUm<0e(d;@C%yB z$HYhwS|eYt-z&5j8yZ#&Biz@K60>Qb~rV3sNUS&51)E{zdiJH69{(&?M!s^g(Z@;c{oXeC-9)H39hls={i48 zIWs{VfmSyCv^is9)Mm;Wzhm`!{LsvV*NpSNfmi}T-u*w=OU(+7y(J=;T35p+pv~{P zM-?T1A+krZM+Uy7t;t;XcMf=&lIa}7M0-_JlW1gIv9z2C%@WR-NlH}hFWN|7_wltf z%*@2H@6ujHf7h?<9Q5_*`JR(aGM5lR*zg3>^K z1|>2<#}zZ62dP&_sZrmyChsW^SMGPeYdQOW2mAj&x$QqAKKTDK;P_Vo?oYM6Pu_UC zGqgOiPBC~C*=qE)?QSmSkd*CWIvU+W?N4~=OsIUGKWkjI-HTA;n>oW?nfMoMSFqUv zzxo^`laZ~6w;ygBiUvpzC&3fNSy`{FO@s9(9NtpAchNW4@*ltb5$j_R%z1R%`Xp-A zvYNd41?IOs)|l~<%@ktl+7kcV%lT&KBl%JBp%qEnP0K$7)$Qf>KaM!k0}9_fRX_wY)*;Miw8r9;fDCr&rqk^lU@30pp5PY5 z)71pAOB>`qV*tto8s!Exi^l8cCkCYT)CNeVDkS)cQ~g6A{<-H4!>Q3&bC17Owt^mA zJ|S5HI>0%AN=PQrW*+?>UU_J+?`++gZW${I_FJvBDe;e_eWJ=qKy>ku;L361%fq+Y4T*Li)8m}VFg6-SjCb4L2*OwP0awX~s0J)QqXXMwZz8Fl z;3RQ)9Ks;CRw@zL5Y|8;k{^ZzAN32SwIe1-qYUUkRYd5 zU7zUhW^2M9hpVpfgf-8#2IiS;&$fswhkHq5W2PmGt{5LDXISAKwLVqIhk{ciJ%tamw7#5h zJBQCJyYC{3rq)dk+;@Nt*bgs*Y&Ex{%O@(?>g+wji2nS60@S6!UXAO}!A*@F3z4lJ zz~_sCuvt~2{dT~&3$;yqxorJIv+2(*Bz9?z63NzcJ)ZQ=vQtb)=mL%5pLt<8w5pMi z-ty=d#4H|&wemZIQTOSeY45UKH-~4g=C7F;96-P;;S6P1r0(- zM&-BD-6cDQXmfuy`XxxeNH9FBA@r2{Pt?cs>?ThQd;inisaRS%c+*9R&-Ap<5D8^f ze7SMI$_1Zn@uKxJ=l4pM3S^Y_f$p>)z}3EoD+0VX`s`aG&$)eQ2JE zin)&jJk|>m;x%h-fhhxc{{#8OkN#U-SNC7Itm#uxIJK9 z4^11pl&Rguuo=91pYZ8*$=z{2OFAkprH-QQsMD9!SK(ViZLsgSRhA0WgF*W}KW{W7 z_qSDPl#KAP145;~KDgD*X@q@#)AI!9V@uKKD=V;x2+L!$lMVC770!oz0kD_-Rqu^Y zn$5M3H^+UmSj&8>UE?xB2*?__!g2}f1>0;-)ZCYgV98D74VvY5pp8Zw*|s0>*rXu~ z^8~CI6YFyHsalUOjL?^-3b|2B|of)y=@oaG1?%0ZnFK98uz+ei*&HdS9$H}p^>}-qM|Aws$oROw=p{3%LmEa_{CByGhn06#d|I= zM9?Q`EMOWypV3F;U)I@SPMZX2GP2gKC|`8#X{G0vDA=>z89@&13zJftH#EQ%@tylQz(`zV_L53-}kKm+$;cUJN=pE;gbUS%63 z>W&-Je-cOtpgpC#cKQKAD%y@3Z(~4U&fOOzTEXYJ&$vG?=7f2*Nbz)x+kqqm-e<}B zBi;H0#?VFPyr(hK{((7iV!Ua3hR==Z$Fft|orX$2XZi@Y#HFua8^lzAV{6Bv z;s^h8B34ApGoC1QQDn>_0Wr}tgm6*!6}r6hE9sV263AU;)3LJAE~G|p z{sx%|H7Rco`UJZk;Ecey%Ig|b*j905ZZ@3Zk) zIeMMqVr6_9@fLW@Ym67vn?9HRylr-5V6Lp|81u13VI`i5JsFQkI%QH@X2n?{bvNW` znl>ulpNqCOI51=+>ECbM{iA2tt!cg;QJMUA`@ommEAEw>=e*vK0Qd(plY*(UXjS(m zQ+&X9v~kb<=&PtbRNR+aRvI0IZ|yKJi$na*Krl{5o&Mkr~4F$?N3k(${&#SqFgK?Cc(!kff-HOLk{2b#&0~}I5 zh45a-I=v}`RUUk1T)kA(wsbCUkfk_IL;OYm2&wP)2l)Ip|Fe_jN1w&lyfIvzNOQ#j z-*Gzg2S;g9!dU*i^&dEHCQJZW=YC*X%=rDsB(K9}2-@BrL~Y=qQedw^Te>#qchC=3 zL9CS;yu#w&Ctn=DjbfSHp9h9)egmvdH^ofXld~qip~nV%ujJ43ngDmj90?h9SRrd; zjsJLbZH95bSwyb4T}hpz^P0TG4Oov$kJ-gk4^59Vl+>~=MRTUu)CuXr>^sqNA~G)c zW`Z1NE0}E?|4~OLOJr7e$g9|vQM<`8Vvp`nm#Go|%#j6V205uQyo+O_xh$Kpnw(!o zhbCTW`19o@e;6ROowwimEs%*vjNMR@QQtc>5G^6!H9pZPwMH2viTl_5&e>^X#OdU* z7QL8us0DBFw9I-5#y8!$D06J!ywGg0N|{h$9KVLx0CO70qfTi?D@z+)*kU8CSsDuD z@!v$vZjXuJXEg+!4-B7Imca?O4ZXhGO4u$#W%Tnszjm@-!lNBW);H<%R-y8imJw|$^aG1Bc8G0@Lg_a(Egtaz;r-)4qR1;#u zpy_0Q~e#@|%i)7IeFScN1Emwv8p~vPuY!rLX9(jZ-cwa() zb2Du%po-DtNfTGGa>#?ye{u}#$)IZNSi+9~Qeru3qfNIVU#xWnkUXy`90_*e(dT?G zJm?~0WDA+`(C}(kSRx;38YT7fQ&Yljx5@IrrYez)Km#A`a@~^kmsbj%0yD;JRq?;X z`lz?fLdug{ZoOuhPo||lhFM)@o5!nE__~=JP%APxk=g-Se}#!HqZiHvKiI5vjk_SY z3_*kO$#w;Pu?yJgR4sR*6kl;R++X9{Rqa;)O-r*lk3a)|ok@BfV*}I3Rk*pYM+v`% z;$T$WARZM7CH|@dgHolV2TmsteLzu4CFakOhguX68x86_$g{>*ZVt=%inH@dCM^QS z&;D(*RxOn*YX{LE<@aFio;tBgQvN@j)GYbm!mL&Ke2;{Ii5)>k^q8_Sw<=MVYjBeJCaaogZ$rW zUb{fGheN(T^{7nl^X>S7!e>L@sBL%&eZ*w5Ji{Ne+|QXObk~a3L5SQHe@xm=YLq`U zRAgnRF{@%zc!e^U;518W_^L7=3}GtQGh2ioi_j!+Y)Mt?z=Jx~|U;cgcs<;8PjjGSKgbLtv@rX1maCaLL^iM^$lIB?>=UmU=er7|{+ zRGjwFy+@a@nxOOj{Xmm6=L#7bC6S-#zx}sgtgLUvLYggZMn8H&@JEu6H^bdaO*U1M zii808Kz2_xEoUSpn_?&~*|ycSx3`J0D4O&SFaLb-ky|iadisl(ixJ5bMXB@0C{|c@ z0rxU`)Q|?7r}>tme$B5XW$Ki}?eSkHkyQ=*zD}mJGugk6XF+e_xfByRX7{Jq)!~|<(Knn*_RZEtD>-tO%5*BfH=(G&B zvM-8npfGTQ*eH7S9J6emFJgOkoTyU$(+{PLgK%cD`rJN#fM=O^`p|x{fm6aRlc@g! zdm@J9zQ!$^{rY5yJ1o03S~Ib~J$6-=mZ{y+=|0^WU(gZ=8PAXq?d);^Yp~vPjSxG( z*l1=j@S1QI<#|$BJ|NIm;)}o_VNXQK0Izp<4-KoFXx~Pz%%ncr(Od|t&{{UAO>ftw zs@(74A3FI(;Riccf|116ixE#WY}6l+5AW`p|3LdOPAv(&77@(>bnhTZe^kwHWkg(q zkSt#ciUs_o&?=Xjn>2!=$De+weNt8#q61*0Gb`#3t=fx+K7ThSul1imak;vy2^Q|N zTYhEBPS=`BrMHvP&&C&x1eMy_P09634oKEGd`tL$c{ed#|?B|LCt zezz&4^T}+Mx#mFYc28!;*c$o)z|*P2oIV5kog!p>E@u!-{?_p#&{sodSXrLe=@Yrd zEl5vua$nd*nl;v^C%Px_dYXLo^o;QyCD5d*)(R92jNCoiti}-J+$-?T|=9oi8Z`m$FmWuZuxMGv+uY*eB>Wuid zcRaCeBNCIoVp9q+ck~yfvXd?A%4#y^1QlV=7{BS?eO`FCJ_p|&wo8czy0l*eUO73t(ldq znf3gh`@XK>-TFWOW<*jEO8}(F_zQO-Nkb$zInpB67`eLR>}QtP9*yLScKt8F#V+ z0Jgjb?i^RFCBkf zoAl*^wv5pY5B_+6JCcLG(e5<8jEN@LnK1&}ny~I_-6C{%C~Z}#!xJ)H*6=kO1U!&! zX+(B6kJFtpAHqNZ9LY|vNiFCT=6MvXtg)b!EyP@4r_@*`L^oz+VI>RNazzIZxG&}q zX9(0wQPfg5WPI(9fcK_w-0wvu9L9?yc6hc6;7Qs)$EX6P1y|QAEm`)qWw;oWg7o>s zL&XY)*0qX^N|G}c^G=k%seYWfWKfb=<+(yfBik9b(r@i{(V%s_v_xq4P!8E(xmr1I z4N*7&T}}DHsc-tw6CGb<-)tX%9O);5lwjH4xhVJ%h+V5GF7wRuEOT~8?_JVAZS!?1 z)!?BUZ)>Hqj=98hR(f3070Kr5twspyeI@w z2vgR5OuudewFt)*_ZXpd3 z3X^)umuj$B(~(2%Bd2Jit2B#uSK9gYV!UMVAbhsg<+t{u|TwL@EnM2WUQ7WD0=r)ZO21<&9_eA`fR| zvjhw8>AX`7;^gW*E&I%(E}*{JnHn8lmi(^H9J&ECmmi)cuK9Vf5kqwkh!qtj`08BZ zekcZl91G}&L$Fs0dWet5m~FY#j`X?sk}DwjqfUbY>?~ zY|JR_)R3?8_XEm*Reyl<6EKOR$LQ5^+)lN~<}D*GcH_i!M3sQRkrUnCay@*-#FPUC zrwU55aeczsuw!~3smdQRIpOb%KwC*?HEOq&quwL%6a7jw_N7x-d=iGQkpzm$ga@a$ zHgoSfO$@lAVk;D@`%8QTp57Tu_p9ahgLqBK)I4cs1ULbVd*moFr0E?WflzR==LMBY zDfS>MlCW;H?A%sOqc){7yn?PB`uwbznv}4XOCBPAtc~&{lmh&0;k@FY*b0)+#lw) z>NXvi{;7h;?BF87nCjQVDOg#pn?6ox6Cb&XV3d9G6blSmd#%6*^9vi5;EE%fLjkzV zHHgk^*Cy+7s)@gg;M}~~PM%F}kT76Z-p(T3D8Hh^uz_&OMLG)#8bZP$cn{^O$6~)+ zEKH^oG*8d=KlOdl8TE3RZ4E?y0Pg6{OL0wMu#%6C_K6ZRvJa@~UUGSx71DIASJfC$%Otfsb8wKr@hWk^+p{XhyV4O z>_`&Lt_5qt$Y!C0Hi~RB!tHkBfKQSR@N3gJc6lg?=n$Sq?JqQZP0n3Id~UeG`@mPB zh~SjGsPIiOxBph)UoTV_G_zFaVuG=jd~&3tc6ippmK7(vEEnPeFVs_13);!_Cc`<= zXt(rduI#Haq#J6=xchA~xm=sBTtu$QFeW9>HqXO8AMufgFfdi7 z`e|MV>~0JnKy-4d>*0by^8ZON(ZN5u^z_#Mmuv(7MwcJ-zhoO*X)_M_n&6lmr6MU^ zUDNz9EB&Nyq2I-dqkD5QBX0KgdOZX`qpyNkQ6ReOGtSK(Wru&_)_8;pIQLu9wM<5O zPV}2vZqT5Ftuoqn4z$)Eb>pbtnVVxOT{7`fNA4}{#@ut3hHC#aTWpf8RO4&$qp6aI zFf}9Kn;$e&6sY|8l!)Tku9+mdLnFQ>C?WN${?suLZe5SlV{r*EGlnr!JWn*c_w3h0 z&mEY%xls`BOzlYjC7i7@fNAp#lb4xxHP#Awh!Hkt%-y=2l-7X-1<%~W=1pnfx1*!S z`4m1_LBT`^mA5}yBYZug$=IT*jJJ$(SkmDYfOGNsH?q>yx3tDFoVq3^>M##2YXuHC zHlwQ&Tkyjxx7BS-Y0^3&h#J+gHYrlmVZ=V>-JjMbfe!$@Hr#Rq)>#EU!ZD{{E5B(VfNQ*#}z ze2aZ!&+b;gzV1mmJ1xoEV9hinzRNVwP8cxg&m zW5%shAwMqA6Va%8%hsGB*(JIke45#Xbx&g1;TLZuZ+-VF6;O|K4)NOzKGHH$ic*%X ztvLbRD*Dnk8LTB7O322JoA4z!Qm?&_g^t`5U7L#xkCf3tzn_tn6f^W>ffNaorjLu# zM)#OytX3Rl8$Nub1A}u{t9_c`mJ^K&A#&$Y%$;5e~W9u)vzFljG z=Vcy3Lar#IuKOOhxsP@hV(}Sm2q${gXIuCQth`t{5uBD=<+^OP6-#XzQ$}U?P;*6w{U{n)eO&IkuFv{F9!-r`Xh@?B^BdRPSLix-}M$qzC3c$Qt9xznLK@LPIN z9T5TGnc)@ZUVe+}-YMsd=`&JABG-_b`dIBYA;9Si*Oyj3!0IpEjIa;Y#;uUgNfuoS zjtED;Mf5)sD_=1tz^hvf(i0BX2ppiW<}qT$x1Ro#CM<%}-~o>er~c@2XGq}ZbeAA0 z+DpJPkpe@8^grzt4R8r$V&ibJ(O#_IGfsEh`~aW9f54X-zIgcL$_lzHAerPX%Siog z&!XH>&a9+R0>J6%1rDSsF%@ZT6#w%EwZdAOoE5P_9gxWdL zP7Lof_z0x3>&wr!o&Ru~P;YKaJo8MEL3Nt~7)_9E9-QcVSnf|aKOie|uh)r>#ogJQ z@CslLL6vb*&f88DHwy9?wO3md6AGz6Y!Z4SNc?W=KBI~ZzgJCtWwmrxd5Z_p>g5wOt#C6SgtovY;d zsD`Tk`EFDW^QAnYMuB9a-_j=L?dDTsiCgJLdo3@o0LBHwTSM8JZE7r430clZvoXS^ zno3}UPrV-DbnyNinA?I2+_oQ@G`*d3d3_>peW_#AvX((}9fZ3Z|D~tQ9-mN!oLGvp z5)(DUc9+>r@k_ib7ERr(Q{ZHg)|84!QlWK4H%3lp{Ut+g71RoON-QJ&I}E}-JunAw zsB$FZPZYbMX>{-= zNO|-bokzKC!Kd(=x(CtX_c*sQ_US3_*D^|i@>2q2Skf3WR8DdTLdUw+ z8$b{Zl3)>&phA1NhkR%O=p=)};yEzu&xjzYGp`Iv+)-0F?mZNRI13O0KFi;7L%PYb z{DhM@7|ym5ehc(MVXh@>^I=!oOZKjelAmn)-jPi;SNaJPX|##^WPKd_veIg^m_-tE zkWIH-S6$uyY)BE4;>m*$bu)!!tzwh1ZK-VKk{H;gErfcJvD%$_Z|m20j~n%`s&c!g z$%o1~Pg16GWSdr%0i&H7MeX%w(8xf{!upEiBVrOdc zro(!~iCPhfPtCyt-%`1Y5yK2`Ml_qVA*Do!xxH-iNQVeIm@8m@Ch9++W*u?zwCKOr z$DZZv|4F4~W+$1b7^}46c=NkfGI$U1czN}n#he5OcrlR7vJ{SGo+Fi7jXU_Z&#Dx%3(!C#*CwGG zDIX+{@hUZBhLe|>6xac7QNR(ETGPzoBG0gGc#3U87L^h)qquNkm8UT0XK7)govg#i zGQkK{rHxG^`-Z+0#?QKzZoJbCf{PoVL?S)H_+Ouzxs@6?Tu!4pl!Z2qqj~f_FW$QJ z<%+mY8UiwJWqphLu>v9Y>uw`?TXR(J!ft#h8Y)%B6)E;ZM+16qu#jjL24UuqAv^aX zmYMbPFK=e7jmmdLV^$5;Wel<>t}b}QR%6pAOt(DLL_%}tnthB{to|nvT0@k)<<&-G z*rP2^nlnQkya{{r-Oz0h3+zRVcO=^oeH&+4b|kC@)o}eoYJGo3S;i$A0xeQ816g<5 ztjcU#dHBOr7)K+Vrs=ji9F7s9WsSl(q+}Z<=Ux10AX33#oCN!^fqx_=L!=^hlo6XG zSPp5Kp1CsR5dtT<6i=bvm;E+07fN{+byA7vU}}ZDN0bQ1%Vn4Qc@RP81fJoUb!?@m(*HRwyzi$x9Poc-%dlKYmpx6`9pK5aY6iodH# z=LGal!mD7YH{)YV6)O!_i&Eq^eqQJY8{rGby5uVX>+seRa){ulba@vg$dCXIumFlr z3CmBJq!(GsuQh?8Yz=_v=BZ*H<_X1RrJXR_(dP@~!Lk3&Koa@^ z?8W#`^3|8{Q=#~QNkZxEQO^OkB;xXlOt~=Q0E3d1I%ZB8%8`gOys;mgvv3{Q_7-z7 z$tYY#no&oE_1$DXSo5E~;hBYms0<*y)=8{rK}LV`Fh6Qb@B-fO_!SNJCB28Iy>$QI zkqHhn`rqQm|K`KsBNJ_XL*msQS;VJ_W0wJ21$6#l@0kS2ShxEE9)#nSFSTkgBwU~wG2##GE?&#TWhfqZ9F_$c zStWj$6AuZ32c(&Ge-I#&4`ZZ zFN8x2FF!=O2w=3jJN&H(e__;BZ`la$7vRkEKGeLZ>RXNb9PI0rc3O93Zrt=Af^M5Z zU_X~3=FEQxwKw~c?Tlfz9ovClbiI8Q-=?bFznOTJ8b&YV{$#p`VP6J*VdQpH9k5+UohsZX-4l$f%tBhp<>euexPo+U_oROuv2o z4oTp%4_#okv#g`s#+M$&@QeDId_c5s+`7I*{k^g9V2PgfuCjyG@V)RH)NLC{Ra_NG zc%AY|O`)%MfPI9e^ypE&bXC2-99H~#T>yG|>ovFDO+d>E6MOyWNyyy~l^vJD>yazH z9Cm6KCcJ(&P_@Q&d_7JW8MIwlJ6c3QMU66>ehL7%^s9#*XbALWdIhR2LH*RW;&AAi z+Ce;4W-)v_xlqk{Arf5!q)f$@4v0&Jb&zkMDfJh*k<*zz1z`bcAKrh`hp0j-wAJAPbSQK`j?k%gEc@70czuu|tJSS# zEtv7tFdgd0+Yyt}xGW&FM!4g6BiF`@VOvYXSAe#k^Hpw3V5P;qDXNvX?Unt6$U#+aYD-3v=#*nK56o-Q_aPWAXU8Y z`!8j3(Yz)+2yp@Vsfx2_EG#==>*%AWppSGz>4)kr8uTyna~H3ngXLxDbdo!5j*Nfn zpV3KCtc!Lna8Ad%N{Bz2Jy`W|8KTcjoNn}aj_NG8X=OgF4qn@RV{*a>H6|h=Y1mjP zA;w@M2$lWYs+Z_JG%$No+s-DdiN~{fTS+0|{dN_66ugo)Dtd+mOPxwU`i|KpyPdq2 zeq=&tvLz#_7ukh+-z*p>Yod%PP4*O39Y!E@kDDm&lD71jK}rwt8F#HH22HkMkLOCzfKtete~g;7uOk_%c~jjhi?c35GTrLzqBmY}KXL1cxz zQLj-%oVK7_zdk2O)3+^vlRpH}OBUW(&3eO33>C4sTa}xDOf1_bp>R zScGG6dB+I~&jxZrIX!2`GiLK!HjV=jDBV+HTvl2bPrttP#HFJXc_HNP=@MFj00umv z$&VMH*Z&Z{7)Vsba{9}E=nw}`j|x3n-ZFmK-?(8Uw_~~m zt~Ki9PCjSK-kynCRSgmi*NXs`Bw8X!@n`kVKZl?ezBwVb^%a^wzAa5xEW-ykHI%*_ z-lutocao!XoN6&mc#Z;%eSXEo*Y}t~&rzk@qCUzJ6kuXhq-ew<OJp~9oZ6%6>4l)5lOm9h-LyFjaoNdxf!NX&GMizzI%%>ACGk5< z!+2C$U-UW;X)UIVz40;+90)!@ZXlH4OsEZ7QrP%J&yf0LW8c8N=cbfvjx8vxiAg}L z?mUVupTVoI|Lj#ToWQLN5STEhl4u^a_r-Z?O#)yNAl~4^8W(~X8`2sq( z`MFyL?`Xqt@p_ZoO#G^EFb@d8F2!{3mmq58tiR!up{bR-tM8Xz51Iz&Ax!(2UuU#_ zUS;2mz6Hf+wOO0`iMPPJzOftlX9D|8u^JM8JuNikKEIw) zPM^i=>5FsxR#MY&d=dOm;OgvO4i3Ar##B|y71M=vQhXY(@tccE2L#tQ7KI z*zPU7&gpZ=SlWJZIC-c<`=`R^W;G3ygk7q^MG2M1ov*-<{n#A;v2L z^wVt1nmwGzB;NadrNFxJg4OQlz^@qfPb3Ob^cA+a@=`*{vqcQoG{^w#INGQmwq^R| z?FSiPQbedgA1FVg+M=3cf{}k2^U}Hj#fJ5dtKGaG;?}p8WW#kDa>XD}L*%9rid^{y zVPcfCD$t?X-Pa(4aewz|qRY*#-8DT?a7?8dJ zh_}4Zk1;a}+cCz@Nv}06I8q+32HisS9DFeGBIVxzBR@87DMpe;hAT6)TLvD9uKS9; zf!=C-yES-vUs=lr;?>C@_#2e5W-V#UlBKVIDM z61YQd^JQ75)6yi2?565#_J);}sSGVIz%`fRzKOXIEHy?Pf6z>osH7`GW}=bCieIQn z=BDp=<)rcqa6tcc)T^fcGJaq3Bc}l#K>HSw_y&sZZ@Qu~W^dbB zM^%OvFI5AjMgydT$rb_mv&`=jdjq_O%T{?S_q7X>O@aWqLeUsOwb{^V_!|)FkXdC4 zuOO8hTem?qGT>Mk>xBv<60+Wp#`4r4+;2zO8-Shotb(NUa4(i(?{QaU-p~YE!4+37 zLq_9!r#~A<^PH3OHM5-AF?4D4bYur4SU&bf<-JZGc~efnExRUt9B|%X>wPe2Q?xo_ zX5GEc(?j;Ua`7BV)ttTu9^=Q6;Vi&3^3TRAFFETkr?(no8xhE;96QQr92*c&HJY_QT z;%jZ6Up;S6mILF&{uAk}@w5ATK+dgCJr;)Dqd&v*zq;yuS|WeJ@=;jNxBEsdv@4I^ z#z8ni@p0~L-NWKP1X6}U#XtVAm#2URl>zre!DH=KEZ?C^BbNo+ocHtg=q>fvl*iOT zXGZ;6%s-e<@rO~@Ul)jl(T+HkNNsh27zQ|9*P{V}KNb~pC@=QAu8yAqmn|LZ26?9D z-87G8OC8|)nevRqYolEsx5=v9#_p;UP}i$%jLq~qDLZ<0`3svYTU-&)AD)&vaZn@` zmh1X`#mbJox?jnmdeO07ILIGi9{Tf*RrU?3CWlaNQ{brum{NV;h2-{SNep58Di^WPmT_*B`>%3H^hm(3*8<)vjXTvYj$Y zSx5awfzf6gRiM)nWa|{^KZO2!wiQu)GGX3-AA$Q_+jfD-hw(O=u7gghWa&naR>w*m zT2|pbtztGnTsgP0N8ddQ%3b`C|5?Sx{#9@b^!-sq`8YWgm?r3d(y$Wq!)3Ba=7_j8 zwUXJYGi%t2{i|vJLB~&KV(LD+8!8PzTK^~sm!=_i2}R*IUl$xxWnWOk0oUsM?S)Tp z(MLu8uR$P7h9L7CG4)e>Q~mxS!M;G*(y*@xh>eQJwqidn=-)L3&k4-u!onyW?!g*biyR_SU2{<$-{$RE^TbwBo_Vjsf)*eIZ| z+FZ8H>q;Etad)EC?-ryVd(2O0a+UV7EPsEbc#veovInG~b29&RA4!06Sh%@x=%G|k zdNcehCJ^WHKZKJq)t|F3lXx>gHNV0m+?e3!82U38S+Fe7(SgddN$i!-IDGo0t4Lh; zYQuQRxO`^%u7ca^IhJ%76V*_{sB!)v^t}{W0?@%RP)yXpp3l2Q?RaJh{`daukq!7w z-JO9A<5djQJ*{~0+IsaQ$Anr^xb^US2=LgYxf(sd7BE!HDa43M{ilRu2~(l*4NiIk zw6n2;4m<<`uUVRJ+olk0_zEQr2=7?+vZ2E-f{YO`KfF?#;R)*U&wQ;!aZ~>xK=ofc zt9*)vJLN_Y56)>xN838Ewwa4hiR&-X*OxSioB(sEy=65^$0jFG-N0a;F@LTH1zeey zkZ@}Zb_w9$P(MnmE&gu{0J7rUm;+MJRs~_%Su>R~{pBGW`n_mlV;Ct2Zg|lP56I4-iH!)1fl-neCqMC_FjY%Q7u8t}iY9(h#iK#&>H+>RGN1HMN# zQzB;z9rr%T@Zt(%b|Gxy5v?tjYCuVlyT@cZl4t4K^Z0JT#nQ)KUt^Gr%GC=kVHsszjSx|hxa*$$7PO$7J2`^jb~fr!dJKiRHwaY@;D zJ%(>_wfD+HT!3FGlZXBy-#{dWiJEwC%H$Y8V#ZI{s$|NmYi5E2UwNq&JUGqb+1A#c zEDIUGzn6Zs6O2G~CVS&Nw6LK@@VGC`N}3+%fvP|(JW6$>31_#GnpIu9VtrYDS5)1i z=zKN={Tod@2Q$6a8Bn8O@}R=r48d8S(6Tx@oDpWf|ZvP>}!whhAz}?W;M{LVLvW5G6(P@fL#`?}2E8 zpo6;#Htum|$AAP4YVY+w+eeE-%mDCr@#X_N%a#VcH2Th>_iX_efG>mi9uo$MOy!}p zhYNYlfVMXHu%c6kzc9=mD=&F{kDbo&?&q;gygNg$dRSF8hok=$4kYfH_i?;%X^ z$dDS&^)Q%KDlKuXPq3keXPSj|YZDPV=3&;Wbu*AljswJfTqbu4_nCb3x*)bG7lZ#0 zezh~w)bJ_LQ7mA}IBg4o6E^>%DOLU@LlLJ;ahs9wRTGR0a`>8TDY#SUry z{=X?MOE*lygPL0Oji(q6$4G&)HA1HQ@jv|{R~W#^km?d40s;U zNH!v1=9Z8D97zt_hAEe_8t}`-#>-YZ2M6IiYPtT>W3TK(w%=v56}UW?r?>U3b*38uBuF2JA_y0dwqC@BzWBnmcxJ;xvimIUED#+whPJsH;**SG zG@0i%ocg5$d8!NoQPAeI#^5N|2zD}v7q>TW5;qFII58+x&E!az;UOahI(8%RFWd#u z33V=fh|kF)l<@>;Fspc@dvf+*8#d%e4&~hdzC+gU0~wgD%LJoDxtV$@4|)u7TVUsA zqo!mcOsDF8!dpiTz7KzXp>voCIDhp2m}h<0`F&P{=L1%Dhl2Zze%ZfKx4XDh^VSq5 z2M11KHnX1)Ld*`{D)P&F(5IIvD?9S5)Zg8=<-(uaJ+C-)bfkaoav&PY&vDe@xXO%7 ze1Vwp46WMT$(f4wC%WAhvhU1m2IhE~lv%`c2lm$Ncb141nO?~QbHH@Ttee|)O+o$) z;`X^lMV#a7>I#xeUdQtB9Ah@RU;f` zuJLBt4m%MaI}g&2ivD)Lb>87m@LE`By4$Ni6P}fTQf8O-a_|J}`Qz)9?a_N2J^J1H`m5MNz;8j9O`%IcHKNWLMYp zB{w9N%?0j&2_#mePh-XJ=xdX@PYh_o{J6QQ0&%57VqS)b89I{+M*_Vg3D4G?SsDNO z(QY-nGx0bo(;A}_xtH$lrLw8)mkV1Fx2Cv5Nmj+?Qt%ffzx`M`wwnwfdsw1!FfpSY z_x40CvVRf1tOvS(vaozi@4QC5x9dBjO+2}XF2hw;3RuLCbF|i3o3&FUUmL^+^G-I+A>DvTkIZp?NI%y!Y0toFAtxEE!@J-u z@d+Vmftsm6Rbn8t1$=V;E@1)-$ZMJ%jd@P2_|8~)mDJSmwP-7W`+T%&-uh;6skgsoq@$%jnR%HSHa|Be>CIS68EPU*(|d$41UVsI^*A znUy2@%C-^ZO#WGTvEC7%tDfi|`{yBI1ver7OI_=`i4PcrTZ3%5ZMnJ&mzs^7(ZbA# z>BEoT=1vSr%LQ*2`X*cn)T~R2_NTc`MvZw-Ljn+joKET z0ZFrw3R_xQPs<{6HS@dD(aYg-D8ax_Y;-|(Y8qwi*Bn@+i7hDaT&a03$B`SDg=*%u^ zwxXz`8H8&vc^dT5_it!BfF#ikryij!b^YC?z^t$vs1%UDe+phR{Az=LdqV=o zH7;TX#)-}JFQ`LW-Mxq*Hgy|QRQsw|ydzB7983B?oeu-iPK-Ol+`k(HgcTdtdLTrd z=a9gIxqde;9yV)5yRoA`;4Fh>wwQWhS?QYWxTW9gJmPqxRD$j})ugmhpaETCv z-~vl{!b6-7Gpn1`E*xLtt${Y8!P<|*c5ZkF`K$`+?buSecA^`+rqAUXppx^u{yu=R zV1-9&xiUOwMgng)ry1VGn)?h_QvV;>AH1T4_&@B25C02PhC`Kcq`u2hF=03KA9qe@$}SBHa4MG$V3zuFIdge?5J6clw~okjeu#vdKk%zxTr9eV@eo+v@vls`T7!;j07 zi|WUqC-sZJ?T$Drdz&gRI9YL1Q8vQ0Jf;??Mwaeegs%eykMScvczBtxU!{o|>t{PD zr#M9!5C6=YOtOx0n;jW?dPp=gGkzdnw?pDZD!76wA71M89AYx`OFMwiCV9?2yiH7Z3F__UVh@b*yo% z*L?nreE<_s7}@&s5uu+;^lN%T-QO14N3>neDFVZz`y%K$b?nkr9Z`8;+*Z_S7MR=9 zig0F7eYZ2m(tocJg10k9H2Zj-c;KRWP%6}*zSbI8Fy&ZB|GYH_LteV6k`EquqA5p- zHkeH?7rwguCjYe4z!eCO+1EQU2SbKf1=7}kQLYM>qa}ki3iu9w6{fbf1maFHr>x8x ztuM=)WOwalb-e=NmOVgO(A{qH8;_#&3uDFn3%OU(!DM3wj?xCDE9uD-j8VeJjO3OI zv?{sx()!jg!N3u@zu2wFa5pzRU`A!j)jE|^>TUAiQL;Ktu`lJlM)Fle_vh-GrQI;n z+dkz4>(RD3hX}iTP+)od^&^aBeABi_IOJEc%JcFfL%t|{Ta}3c|{fD5hHYWM!aadEy)+VHyz=Y-u z;amf6DqoHF|1)X{lb)_AzpQRx?^Ocevirm@rO95c2g=s#sJw+y*QFRt3|jmq`_e@0Z3!Um~qJthdT3FP;MJ^4?vr#k>{ zlo%cw{!`S93w0*~sc!*E@)W(=IU{3}%SyGa^j2aP+!LzG(^xgS$z9;0JY3fGBf=Gk zt+g9&`{ft8U~1KlhCneX*G}$P12tJpYkY+sE<2CVVD?>VT)6(hO3v|@RD%*&B2QuJ z{s%tLf);nI57Y1vnBQ$M5KRlDoeSr(P5h{loC-tYiyH$67`x$@_#Vb)+qxNE(Zq&S zaFZjiwEac&CJe-0dz@pGvcdkCb(w2hzNv&Jv&|$Tmv{9oaRYqZdLB`Kj%a;S(i10a zNs0ab%_m&Cp{K$m4)|uqO(3Q24kO`MxVcr2kB)f02h6N`SBb);QvKgcP-1+Rdf$Gd*!&nlbtf>3 zzwl=Ht^=!N#~`_;PFGIN+SFD`a`NrI8^(QdaOA~+e}$)4wWZ%!uyv}WKsy;UjC_v_ z5Z$^=m5Rf9cS~g9)#qhz4E%nB^qkEYUa^W=iIcP_}nW+`bkwW-1kd+*N z#ydokU%*C0TmX!-oo;Pxtqo*2Ufwo2nJxcan_V78d;(H0LAH!>ptBzR=*7iLsvF;r zC#Ya>HMm~>TaW?O^j`~7!8D#+ZCb3kbBL>*S0BmuOq~UsS!$lCAHpH(x{9%aue;9u znMLtib{hi1c{<)=k{q&lR*dOHi;5)7FXL*^q9i7fkKrNptkDgA4Ucj6Q&8{RSOGhs zI`0|LJ<}=2uqt>=K-Ko12HjIg^b+1vz?eZrlIa0x*!~r_k|ddxbz9A+WN$?n=T@8D zWxv=Zl?~lV`223TG^l6K*$xRvRFmMN7-059=o0@FtlmDT7%nuwCFEnVUEI%S-$g)W z#d8NRu@WBWzQYsc?Rl?$b1 zYia6?xuu?|N6@P9T@wg>Er?^VWDpL?`Q+|N0${u{9BB69nQ*b8GejkuZ|lP__qWT( zZ|R;GCh?h{0hTaw0e@_Q6jXIgynE6F|yxb^}N{E$Fi=4GQDoo2Amr7_%ZH>(MGw zu$J|e+ic77GAeaG-6Jh8Qih0Oer5IQyk)@>6%WyDO>4R~zx#wV&601h?1jA}(kIc& z7RU4g(+-#}&N)c~nJNFyN@-kb=O>n|zTj&WSACWmA9LQGmtzlC{{1?bLqX0bd78DZ zE%o#)NbS`ma?L$N3~BeRA*&*A;|?Nl<~=NQE0fM=ndRjs>V_M`wsZrHCdJhk`Ni%? zlHUSOh!}If#bc1gLdeom;F|-D_o|!TOZi}S+4zS#Bi9tR%{e&6``8lX&NdI5amaH0 zW?Xzh-l8X_>}dbA3)0YQsT|&+*f}rA-YVjYVzrF>)lSiBj_%_e?2A#Bv%nJjV%REC zuBYdBUEO2y0`MBx_|bTAn;xm{pErmnWqV!ff&NJaU0=FZb_l-D1k?HIxJmyj*HDYA zExES(gu2O%^0oJg32RmZmwLB6hh9ThhV<{ci)Xt8WVB zH817u_=k9xJIQT@QyXt_=Dk@<*JdDE^mp@)X8ATOu+nv9aK9l%)5TrRoV3?S9GE{z z^EO*|N@PxVf9p2`2-iz6)bBrpZ5Esf=-&~6iXNFqErEBsQX4SIdv(fxTU~JB8_s_S zwuS=ts-$dF0lih?Cd~$aAZmZ1&J^-ln}ju=)~d6H$VnDfmA)qKSzn!)@csqd^0h(H zHe98ZOBU?Do`~iUrCsz@pR4LVL}^!3>&L0+X|60l=MehQmIp%}8!hElXBUn^?AnGD zVg)So8m{U49caFGdZBw_h}v-IUTmbb;@pw85-QV98D~)(TvJU{Ag&-eLv#&mq@At& z<7mb#SeC>rq{@{myA9@Lnmp7{)smkwhjU`tK;8h4q-k|ac3*X#@AS#zRs!l4yydZ^ zTk;lU#xXEOq$mqflsqM<(%zW>p-lwbIgzEyl8)Mij~4)=Z3e4I%9PFM)yFgw9Yo#6 zg}tFY%Ryh)tVUw?;N$>osEy>?O(RF*!T;R#i|3y0!H=D-F$BA$6bbkKC-pFZp2HV~ z#Qm*qv<~48O%_M!O(p>?_%{Vc&D(i{-}6o&!PqvjND2)Fj3ZPz;8xG(dCJVr7vELN zwItF{$*thi7{SkWxKPcLN$ocudR^KW zV-pr|0PEORB54DvnYc9SuHGc2;{@%0!0Lf=Hiub^!Ue+4qVm4#mun*g-|^#sM`6^r zJhm1`C*REKQdS5CiQm02a_#huKC*FP6)88u_G5zJAZ%!+gXxA*yJyWUv<0HRw(nde zWtP)2duB|_*dhh|Gct+NQ+YoQGU?I0E%V|tXy_KEnbVm<^*5fd**Pr|cM0|G-Y;M~ zXmHmeAq9)DW<`x9&RV15NJ@3MPR0~WpO-|0ynVH*_r}AQzbm&-&SYi8R6Gz(*Z!lD zk+HEfK~ZX`z-wwlieTA6+b#4(FFLuaBO*=f&n@SU;!_RpYfg#O#dB}ha$AG!&-jfz}r>9HbxgX4sbOz!A4w)fz_z}q8kb3zX;6CEZB zgliJkZ$b51zYN%W8c>vitckOIGJNqcEx>EMJ$7mwBFmGY8EHZuq^ISys1s zRGw}s<(=D>^g8M$2xn4Cmz>KxiTVRq!?_!Q?Z}G_IvOC4l@PYKbFqiJ^45T0A&7vd zCfeETX9848(o=JnU$^BYs4~k$%~?dajwaeGYZHNY{yd|Dnk(-?Lk&Qw?2;G$+4mBS z29_4%AU=tVHi&^u0OG8c>};M~G>o*f83E^eC&%1H6?ZGRiuhh;Um&(+`H}J@_sB5X zxo>Fv(jpA36sq&}OFuAtr6dt%eMYj2&EDK&3_OE;gr#$tqTIPPVSGEQc>)Fi5~vGp z(drSJE?O@FSuTtD<#0-MR5!amQWI5e2NpTcmmWKco2kaNHgaXm?oa2|4A&+tjyej% zNz*6ujjhmua!Z5WpKhHTzv1BIXP>tbdgH5AS?|6pFyysz%gbm5JwD(h?_=BxdMQLv z<=9hYNHMeCpu>azlw}+6!WNn#U1fY$d(jQA0jWA6pA3jct!T^vv68%QUm3?uevhfV zzIf5FBWWV6*>(Q_hb#F#&PshYKya{$ly!7JFdp?qjGK%Q%jReP zS;#S4Gg@5PVmnYD)URRmJbqNeB!BAXfo0v*m3FZF4nIoglqDU}=2XKmeidcxqY)EnD!UQ>sm4d(4>|lQn^`rpZSLNQG|Vp-FF6)Av5pmTAn}g+iFrtrfHJxvyAg6d5H7_p!TlVLNwLcIE-VTx`grn z0FGtU+}P>EW}KW7rv=ZgdC1u@F3wSh-OqY)xsZ+0p;6m!2lS;sWx7yThxZu&0P9yY zrrR!-29>U!y+$*;AbFxk1B{OK226yka*(UYN9k5R%_m6TZTx5@Q>X#0OoS4O)i!c~6jqW{7>@MtJrWOKUOm;FBtPgUF=79oxZeK zaqCsJvlU=2KbzLnqA@3uG^n1{oMPC zz}371tZB~kExKL7jQN)5Kd1iyUbzQT7&6QE{{V;A^R1m1#(M6K6{5X%jDO2Hn30dI zZyL1fX(O&UiWSwSjFWEbD|vx~gZCM5N8wB%h%EfCt~swkPa61l#0j&+np#HZrNXG^ z=tn=DB>n)i)-8;_9o3?le(<-MvT^7QCx?UhS&d8uIqOZHbYl}7=LkDDN<566!AJx6 zfvt^CLe%uP@^7pqg#pI(8<74rm&>q7_yg0N*EHzPFJsTChpkUb6&y=GRBe$*;A>;U z-W$8uu9C*x4(U3#WghhXH%PeEZc^IT-JV8ZGNYw?HIqNGypr)ds@vFl34E}H6Ij&mbrzyNm#6=hu?fI#eh>pdihs8CuzQP|MZ zKPsSY?pBqIWlrq4@heO6WlHot4Qg0;hgZ_v7%m!BAHwIF#Ex*P5gV^#QVG#U+;V-h zTSBbis$%!!vEN#F?@ZKygIlu;k%Hi76kK>$Le-c{q+Tk>+^W|O5jN)=*#0=CjrN2+ zyss1W1xX*ux}lA}l@W2T+}YIpFRf{Ni^6A4&b6Om1KixKJ0Hv^8%W8`Z)rX=*5PR4 zl~&~s10yEAKf}5es^eb2wn$x<%v5jS9;UT)sLDL(P88>RXg9*%9Jlb+oNwzw1pgiALfv?OuCY>Tk_$%;1Ea z_VlE14-B<+;RI(K4^Aop8!^m7Rj^O%S?Y0yVUg`t z3n5&o81yvKF_ko3`B;3zBhsg`YnY}62^~Q^E_&8%1*Mr#gO11gSEzUk!VRcOx0fl4 zXwS=^;x(-)Nk4SPrQ@PHe+6lm)4(q^C*2qDDgI)wc$38z*3u0=Db-nq80V!k#eOEX z(JalR45Cb6Dx7Vv5Y%p7_V!7hMrJ2D&2~zox|1<;a$d?ib!TO%_(e4)GumD5XXJOy za`(Cpr>8z9w_={F%|m6X>Kc28x42>KPg?3U{{SCv#9EEGVcm$x{7+ibo`qfXH%dIb zsN$KoKGql>gmXh62j$|uCQC03c!Dj_xP)5KMIVY9dr0q z^AH^4B#wrmc!(MMwH7WjC^`oE)J*}%BCm*Xf=A^=t?b-B zVN8-G9XO?8)bF6dHagRmP}K{@{f0$26$#GsKnR>_@YZ9e>p;u37G82)u3aKoD3o+@ijh7~BOt25Aro}H>M44|Hz{{TA7zwlG9?=2fX z)yhFUw(Ytp8BejT+Yc9CT976?*SROHbSkvxq>-E{N}iE=8CQB`<(nTiBad3Av>v^E z4SFOVC)0Hpf?OcU;N*0qy71nYs7K~nxXB$CikV@fEn>@cId?qe&>qzTMa^`#{usNp zC89gix2m1h^$|nQaaXXd(9=vq(-n-OY^G_&p(nLmk|Ul*X%;=cquA8)8mW}R5%^N1W!r(ttAa4?!sncET{nijF)ppB#|ox$ zM>yb6som;K>!_`FG`t7k`L%1sc^45W1)CkK+%6@t@cxut%Eb&R`ubFQcAshCy&q;( z5;ghU`qzkjL-DlQ<@4KHITmr801w_iwX~&orp3lS&0YLY@G1Dy##4BT{gg3VWH>~w zRbL~neX6!|$?M*t^O!LMfmyn}hZuG0Uh@Fsr9EXM+*Kr$cjlkw4#tv^GNy~mHV+3L z)ZqN}Ak`#n`_!Ib2e08;MrLjXt-hmmZgP?qQR`av-XaZwI{r1wtlprHaZ-r^?hSN9 z5jgD5YPeWW;zxOHdu?=|D~1QLsP6QuTcMJHPpGaWN>#FqRF6YiHoC>#zf-+^>!vt} z)cDcEEEMU!YD}|r;pt;JfwvDz%bwC`N%JFMeXF9i@e~p;S%$+~A*}Bsz$n@7YMlym zSBo0N)~EbLdEr1uAEid|@#$Ty*M^{!7LB%__G<#~M7p#we(R&gqCb%Rvyi?bUK#U)bL+~bHmVJADsJjtG=)1LkALHRm4^{B&40k*7Z897`&mr_!Nb zUh3HFr5|GsdHhW0{v>$SwTq*t*|`>Ro=E;SuQg)X;MKFa$t34I^HE5Eob)tQT^7VC zw)vjrMsu1~eri(|dFxR!W2Y6*sx36f`QEi6NDWlv9+dLjdQ(!cI3`@*$F(ZIrBhi5 zH7}YmS}Bq2%Zn!+Y2r^qR8_KgH8^|@MOu~2smIt`mq93$GvnGOZ}$t#cns2IKZtrX-`x{ zYT_w17OrLZL&Gv^+JdP@3a2bc;0pIUO9*T&RyfowXt)6P^)*vL(Ie69_AU#aUA;|n z-aGLeu-m?yr$|1~BMqs4Pd2&F3= zjTDqu5$=miNksw{iYTB0mXeAFEfkcr#4Qw*v;a{>6bw(G8Rd*RoHSH#{TF-tN<6})ClzwtLkAJ_1%=UMo7eQUZ!i+Y2fGA>E>$4b#qbt81o zLkorWkaFwUo_#KxD=z55O%Ui$r`Nfx;}6VICz2KHlJxxgS4DNDUE0c32Ee@TA1NQ7 z2BLd{#9K`Yvk`_NHk0)SIjm<&r9X(8*eOF*p_BHOi2#-yDskpV-lP|cfgDmJhQ@R9 z{{Z#s%$Bgb`F0W%nkn$@?JTYQ5nz+e8eD6W zE@FuP0LvljNBQYoUBQ zOCVW5(XOrj(4F&V^q+Na^f7KN^-fS$Vik>*|LDel=mPqLMtEGW5fqK>E}_ zBT3VvSK%nMCt-3@X#E_557oc=^-V2+e3JI$oQwiN{3}7B4JxuRAKf44tvAe(-x8oI z4hSD9KUx*h=VCJ1TSBAxgZGlgBvtD0w>bPitxavI+!+Evb-ESl6xbYndJ{p7E;x<6 zwI6}c(wi-@$yp-YM^;mY1MAcDr*!u`*#->@C%NR**C+b;FR1f!0_|xU=OlwM;_m|vl zE}=cNi42jEI{d(lqkg^qmDA~7Kh(7OMYr14>L4&+x?l2ZjgMKlSbV3BE6`V<9X-!{ z{xtZ;pK_`#WMrAS^5prm_;nwhTx(7n9n`Z4Z7+Fw?s|dNd?&A!T}I9H+mnIjMm}8r z90B#Jx8DzCnisgzZf*4_ImXuohqiyhyzs5vgeFjY#T;k-Vhr}<7(bm|(|kd5VxsCR ziET>sc#h=Hzom4?5}(G7+IWgJblb`5g7ZP1!ggzjrG`<_WK)m8kDK+XjBaey$v@7Z z5v1bOOZsP()8(Z_MKiM7?@dp}Cq4pbwITrT=f_;TKQiV^1 z&ZY*Gz2)w|EsrrNYjS?kD!f?eQ1$#r1M#ij2>4QO6W)eq*<%sJlCazI_Zju~uCmAB zoy_{QQfd&d*<*dh)wrB>>FNFzAB;Rj3=qwuXiFRx(T%e_O|-G}UuvjOP}g(B#9^ly zX|JJI#=b9xD?j*0v^f?FD7a~QdBkrwM#^!u6pO*q`0(^oPz;foHUSwjC+q-jZ#Ma+*>Xb41h)$3dGViX98tu;s7ok zM?ycRtp@d6h}aNumdNQ%?N$o3FrP9I!1R%M28$cR7byHgmIRSY zLn5nsXP?TW0(b!%M1{K^NTxJtf#gfNQ`r9ig*rK8fM!J9?mxzZS2|S|qd<%twquZY zrxPS;k``q>aw=mzyvPXvlTeFk8coo#Vmc_G!dPqT5Vu4vxX59Xs-q04s|@Ui&7Lom zbZAI5P~0;}O~hX+@I?uUT$tr3J1IZXhK^a%ly@${b0m&gWLH7_FMxjlYJV~*4jcDa z_5T3t@%dJJ9P>0MjtfV}=R|ER=g7B|NWFGH8KTI+)4r&@(Rt5^m>7?6I{pTxf=i2O z!Ly!dB~QHZF;kECYLA{EA|d8mx#9P6Du1;wi(o~Od%AK33UZFXrBV**ooMZx=}%Ru-2Q8iSm%jT>tHdshV%YjLtGVZEH~0P}-hx1j0v zJ_YdD)b%AFY?JOU@i!a~rFYS)iO5@MISSL1$gg> zyhEv9>W?++OK)&ng^(232Vqlc{xsHZwU6y6?ut#EB9HZ8XQgT{kF4*YTYnSivC4#V zJYy&Dr_UOi5g1X#K6z;RBab_mA2#`0mfg?mOmL`}0Q74B03Jcx^E0?D-^_T zYiUSqeR^WK-A}_B-iXFq+Xi?^&z%P=T=Tqg;jz zf4e6Jtfa9_#kX=h54=ALP`+s0APg6rbU6A}4l<66!SliU$fB`E8#DQklhEUXQb`+t zByF=t>b+{%Uo$KY*vIgZkyV}#G0T9_FZY*`QkhYm*-y%bcgSsobPdG@DWf^YK<+BZ zNgYTilSaKt^{7{C$RmbLx&9Y5Nt$y$g=UaD9m|p2flo0J3dt0}(O{TpYz&;9-Q@Km4s9G&O4&ux{{{XdIDK>LdO4@F1 z_y@ve)peWY^59RI6py{{^rrkk@qDwy*0xQHZ0#<;#Yd04Zm>c9kEDPjPnFpZPipdy zwLG$)Hbz0$Yo0S*KGHDvPI%LD+han{Qq-&;&b7OA^ip$OHII+C#pP?35zdE=;eW}l zKg|rF$hcA7g3CY5tTqbH7>bbC-ZPX=^k|+H@hTRux)!sZi@W+7#nikNtm)r0DlVWM z84oqiv}1ad+>AS^#cFB3E7tV6g2QlTjDG~YoL5S(N~grNFPbvk$&|WJpU6ZX?4v?DB@C3a_x=)tEUcTMM=dkGh4wP51U=Q z9%3?Eq06!ULcLSOJ|MC1mWa0Y`$L7osQz?nnwFd3KMsgvz+k^GJh86<*L+8HtzSLF z&Q4DY*b3TnQP|^Al;6AaAF1mXI<@P~EQQ&RtyJ?FP6;E|HALLTNEj;Py-dB)27CHc>neekA3>VhHDw){n~O)HLE{}GR017N(N;sj$i_YM(yH9} zPfFEElT5k_duNQ-5414cd0)<^hUR-hB)6H+j{CAJqBxtYNg1^M$*ZaOUsTfv&k>S7 zp@%=MVQ1$d!CzxtKA+=z_*5m*EYOZIjy|=rm%bSB0&SW#f^*z)%~!LhQ&hy9m&9-( z0q3Ch?NKx#{RkD=-uMnzMw3grS(C2Nb6nltn|-Gew>He6a99onWjYc?sd<@?525@G zS7apMgIB``!j5xE<#JU|J;AIwUsFiUm3);ejPp?}O70`TB-O`+;{YG1tqmJTlJe~l zP_wbX08(oD1gj{kvTqJ}W_?RkirphElhd_*18D_^gER(b*qc`_eN9#H=7VS9T_vwB zwu!do*}sH(SCjbn#`0^|Z3U~5YJ`9R>sw#l3BdeuA?KQfwj;s*rcmCJBl z3x-lSsUswe=BOmB*~rKhJXZsD0UwQ0k({ZQM7Zj4?NX$h#~@;}%QGDIsS->M4O*4V zoihIA#|H=6tlMf=mZ$eqzCANpD=8-xGR=Yk>srF2=w~Wv?sfJ)Fp&UDqzpUcaa%Cl z=-Rq`;WB+WuRIb1#XT~lZDJd?(_L}IwWQ8jVWYfsJ6rDt!twdmjy{TOo4wHQ?73*m zC#a>h*YyiF07_gA#NgLYVdGm#V-ep)*N#U|#;Vk+dXkk`KIq{J;E=$E^dgX^cvOERhHPh`D_U%UY8AT(YtY07a@-}HSy9scM#Ih;seFbwkw=Pt5 zLOj2|sn=v<{6_KP-_H7NgqXHdhGG0hyu6@n?(TC|?ibCGm=bb&)H?{`l%vrWr1xbf zkLEf3DlaF#YTuT36HR3xRy?*fQ?W_1r==niPZeZFU9c&S7S`#CO&Y1Kh8YSG$*Dwv zx|)H}$?2Lpdm2fxxTaeVj9{O}oU#&mAC*zw!12vWB`bn*4O*EFnSEaaf)BMlWf|*= zq$Lh1>cn&)k9xH;Cj&!*GI5_;k~!pynC4)8Gg+c;dy1J|h|V%St4P#o=**oy=R;}Y zONfA3M&%xrw`s3g-Qi@wZ+hXDZot*6cuej#gPQD)DsbH6t%Ro4+Ogc+SX$q0SmRe_ z=tutmTD=>=eie^IvJ%Rl=@Jh8RC`u8!`}*+?c=|^QhIr6-nFgcuN&L=a@bnixHpmv zan8Z(S>h{uW}{5&z+sgbdrF%2Gd?1C%1a0vPOuH;zY@rw!alyW;*-tixj7{F-PG4> zZK+x60Qs^BC1Oh!Kj)e*{5fZIyW;>c?mE`%4QN^~Lz1RrdD^Nt7L0+fpz-7!%Z`Sh$g4Wm15 zRA@lxY9@<4aw_Yi9u5eo%44DY4MTydnNY?+Yot4oSRVB*ge9y37}~Vms>JY`T1qRx zKG<4XN+g? z-1l(U$S5j%rQG%S{72#4aYAaAg5F++Rlx)CsIR;QX?<+c=~t;VZZ{Neb6!%y?#(va zx0#--qjo$p|0eo)E> zGC+;%5cWS`xlW<;&+&ADmQ}Gt%askiz zG)FSC;mZW%76T{gRGRZF6JdY@^KwW&^>HTN6+sR?R~W@DPU;x@Zp+3fC5$h~Pi8@b z_=>FYx`7?L2OUDN>G*c2a*V)2+hlITxB2#`N#-hrU>&;d2RWfzsnZ&?YJ5U7N3h0I z=g%xpf3KxA(4=5S{$W4BS1NxBk*++#rGVXxXBag!c1l5TO2~RJG;Oh+TpVVv6ENDW zWS3S+2Row~A53699Hjx zG|%`*^>YQ@R?Y#sQX=G#e(zrXwS(c^8tcSY@!Q)(A<4|LDIq}j1B&eR&2H1fT6=hZ zKwwx8^t+WKa~|M-pQU%uqwM|ZjpAxkT0fD|>z@!K(&QRmg*;I#A0aG0Q;*T5n9 z5_or7k7l{1HMHz_xwZiF+~Mh!>{?9&QG)u{C5#SmN&0uLJ5})g?T`93 z#kAJ@M^Sv5#6Fm=(@*$|ez1#eLK}Tj+dswz?0q`^ zbkyn8-0yzF5V}9XoH2@fblRUXp!r!&^~pUs!5^(fD?unJEMv`Mfg*A<>Q7PmSEMW+ z4)GkD?JgzKFS1Bi7zF!v6_2TS4^`7yu3@*1`XX{Bf%!j(=B9XL^@})a<=tt&yyM=@ zqlbDr+-HJafAkaA^s3B_X~=0Hoxg#GN53_pC8NP3I|f^Uz>Yvsv)>Fl{wAUZM9bVF zk)9Ss{{Rn^{&meNbBomTYGLYA_>r8^6wBN~4(?T>KhmDA_%9ryKsqCE!~81h#RInR zULwTsFf;CdhaXC;ZCZEpCX6&napptd=O zk+qOWodG}(9=NEPB9WKKa;|tFo}XO*07`Z=Sc|`XXekp%^2FPa^#+j310amYwX^qy zE3D8msQJk5yzZrW(q=HNob@UJpj*o-jxfkX&m!| zj*O4O+9sXhy)^iXTLaD`FhawY_ch0BemB0=^`Gr1?iM*zks2`jnd1VX@n4H%@eTZv z7coQ*A%ye;`By$h&zj~Wm0O`s2>Mk{qFO}dj2qTv<2Y9`+%C(9B(dl+eaBjPPb?V$ zQ!9=EI4AJ#D?56|yWpE?9U_>2PL$MOu3mWFGt@L^f`1Mxk=?tM?BB$TOKJm&GRY=* zDZ>NoI)1g$>ApJg7LcHi4x3h^*i)vQ}hxqh=oFh2v{tKWDA^{xEv4^f9wwgV{}95^4X za?c9R!AM^yak%mi=gl_KYpWZb+GW(X!#_61B|k2|rEakio{~pG#6qT%L+_%ywUk4s z+DmFc#v~aB{sm1NTOL!WAKnd-`Sl-_br#|RV)Xvs{=*wXNZ^1 z(X}ShuGnP6_mV-ev;4n0!c=7xbIz3D^dN9;@f;E| zIqI%BsG2p6vphyO>K;Bptp!4fmfFPjBm8)#g+&Fo7>OAJoPHqx07}bMx*<)i8YyMf zHzTae-Fg~{Vu{N%p`vcX995sPh9-IAAa-bu59Lw^m&+xLbH+b`{`M%jr!weJOL^i( zk~TXJ2WjS;3~eLekETUTL~h_B%^5$zjtA#mPltRldEy;r(kS9*g+JFafC7(t(vs$t znB`K+%+~N1f+p84A-SFqY@&advtXY6Fr?o<#MZtO z(o;gwq(!B%EXuT{S1lxlxZVJJgZxQtEQ@vneMfv(}_p zWMpO9vqU-*)~E-!c-%=Go@RblUX;}?yd#`<#d7X$=~-x8^AU+y-Jw6+Jq=79(SB(7 z{`_XD{h);rykUo^>p*l-{*Nmwp2YD}Vq7n|buLaqMB#zzrl(mIM%}n#d-kd;wFq07 zw-e8qA6k&fvzTUCa&fjn$o%NBl&N20lRKSoXbp2OV(`DUbc`RzDAsp^gT~)u2t#5+C;yofclcAAN99EyY znz}ER4QO%4_nk0QcK-nDR7No|!tsu^-M@r(^#WDQGZHU3wxN4uyLVTq-e)$;3=U41B z`EPCnF~VeX&?16s6;5wPOOubp8V-X!{nEi4XMBeP9jn`{tt@^Ycx<%JtvfHwPf~pa zR`C9fW#H{A`$GF~+L?Y;Cz0)5UE`k-UhA5f7an9thxcd?09LV15i(9vdhX2M6!?Mn?Yf(@f3Mv2|otz*CQ2^g0K|(bKxsLhYTgm5^h% z2BeB$tANE%aBD>)CkC}O?6rMIO?5C`ecp=P*G*~TxaUy@uWBStnd|wQ;r!)al#UN^ z$TXv9?~-~7+BM||V>eHmba!_;FNZa(wT;MvL;O9nShxNU)of6_YAzxj3Y^vg<|Dmb zmQLGvt?eJg+Rdbk8cF9aKIy22BXn?MBmpK4DOY8O5vg6B@sE@O@;zE*A6-hBlUrn$Ky zHD&L~dK~wUykgq0ds{h*+o8t??^l;Jc_oNDjcn>VCa;7)~>u+asknt< zIT?EMO-+#Lmt$?bhN*tgN7ku%@v!6-YI<%2HbV;1GBTAerS+F1C#^`NF*QqkWqYZq zEtW~72nJN1wQ6k{N)}gUm64q%9z(Jd(`m0>@K1*>Z-U)w5^jb*RdM&LnD9r0B-F2# zN6*>Qsjm0NJ}S5HzMi^ulPcUY?ErrW_4?M)t0uW4)aHaEDSJt+5g&}aV`*Z;NVSv+ z7-CX~8waP>y!z+E7Mhtd+S~?q=LGXyef<9bYPfkCLm_6);(F0*e|KoZ?DDW9vE){? zrA`Z$NGht8Uov(yFT5|QSTW2+m)nZRbaazrp;q@5*XjO0oe)KN2$C_!O#XGM)A)Z~ zOq@c^)Q~y)3a3sp_)Ohd(_Zqq=UWtkP{{}+*K2*?mLR>1!Pl<`Ab(osuJp^x7ArIl z$MFH2)>5F>tk#`q%VuVIWqHL%AS}a*x42^+{V4L+uT%7_TuzQ=o!Lib2Ne^jJP}vk zA&lan5n%IC$Y~mdBMvzfk`h-1e;U0YUAQBfh@m5|=TP8isub6Xlg=ZpO3|qu2sH~r z-B*!Qs@S`yxh!&n&VM?cW*GE7)maJ%&gD6#!mLgKHBpuHCB<%NMH@=PIKl2K)Vu-U z;i=q5aXKHfr#U^%b6yJYtorTzXi$jqGT+@D>)fwzbngZDcXMq;MZzz3YT8M`d7R3Z zw3jrtL|!cMHIIaJY%b9;U9m(f=e=82msY!y>fy|ixyuvV+Ly&1A-eGvt0m>cg@Gq( zb^@|^$6Oy;(hWgbbu_5`rrIKPeXW%wR+YB7dvA@h7E##NC!5${`_#)IBa@Go2?^5e4m=2im#Yy(0G0my2@xin(jzYs+E-lIvU6-XgNRP_lrn=)$Y_GuuXI&xxSD z<;;#FZ$B_9ed!u7M?BYkHj}4c$CTO+Uze?6UieygiEC%xy>VGj4<)I!9Be1Phc08m z;L}+^9E^L4y*0GZF4Kozu3KJCxJZr{$LF0Tk9yItsXA9mnNSJZHBe zvv4$1DVX%Vdsw53PL-r&(EeV?-N@#*wb)bKLheaJ4wyZYVEpEp|buY8E~j z(TvkF#|%rgK;Y-lSI(X*@l5_D)ZTfC^3>uq2Za@X;t!0jJWZ*zaWNKA97i5{`DE(w5$T-KZP*9O;p}g`qYak?^EqALxFKzfz(rkp!-!%FlUN-5>83T zs*K2REr%fF9DOR(8imEV%1D0`S;c@n98-gLJo}2)H5%B?RN~c|IQl4}yyNHsmXeA9 zS}7@L0cfP9phD3}OF#t_Q9#19bo*l}GXsKn9CobrQtT9UCjz=ypje?;<0BwrKD9I2 z$9@M1?d0D?RcmC8Knl^Y{v|zWg`|(=!p)z4MO_OdYH$k^>qHik-AuOdL%!xt00Y*h zlqES>(zagd>#+@u+|alBG_o0Ak1X?=?=-C<4-DFWZtCn7`G@XdCy-BkRQ5g=w(xbj zYFeWt*Clr(UWT~O5$iX)uApuukaNPPU_C3?rAD3_Jo4P+sHaXmt5$027iU$r{?L2H zlJZFiShrxH&#hF3>h8oTpK~cZaCrPj70<@c0rO-+t<*6801CZts>>c4Mp+lo^{xz7 zDpgu*q0@%Px=WV!e}UWSK0LnDf)RGxGm{kDgZcCMS6>H=bX{gK7Oi5jyASnp}3F{@Vbb7tdf;8P)KRZagib0ML zkl6eyiProJs9G=v2320Hzc4@HSo(IqcWrZ++DM8Yyr-w(UFMJRBIGZZdXZkjqvkll zAB}aaMLGOR=Jr)rxYU>2&pnpb?%rLIf)PjE!9Stqs?RJ}WWex&>QHq2I@h!;-V^a6 z<9FF+Jp+8X`c^Kj@YYoY?3rbcs2$Bru$1SmHgQzKLv5dCd1L9|CfjL)uYLGGhqY%- z0foW^k~uwBm9DQ*@a6S};L)-+aq|wJLs`epvH9!Z9{H{@lJ71xtSxZP7WN6g&+tQ9( ztqmiKx5c4%&Mc**MnxlnKaPfI=TLrN!-I@t6vU9sxK|P=Jo@6A?2~1aUsUebCQVkb^bIIUkohPSKPcycGvyf`r< z0wu)kqVc{kSx@JY z`BqVd<&Z~j_dwbL6O-?>^Z8d-r2IqHt;vd4k5aae?)K_z;d_2^nfo0RC zi&3-#F_h|~@UBAg&U;M9EG-P5clWE)=-gm_6@>89j@q7eTop=tYK-aP5tZ1oL`lJP zg=eByOgoOPfU7~{cD-=zm0CS z`@oDEJ1DZTMgalGsWm^2d|owM#I(17+v!m#NRb9IKfG&@g(a6Vad4g0Ps|tJ(zUBq zFLeu^Qg?e06CAD?n_+L4Jw+i&ZRI$WGOrmNgHLGSncH{nKp6|#qgbZMbor1B;11rn zu4L|S&lkER^HySB;t{kA0zaKe5OM&upLe1je?w6Tkb)U-d;8Q(74lGkM{&WT#QQt! z$(W-!-3f(90!Ycmee+g;k&q%ZS1~UkK-$Of`_)aRxPuY85QJoI@A%Qa0u@jgWyUuX z{c3DWgu0^7+VlIEq>>!<5&(Jh2en)`7+gGj<8PRpsQg#2<5U7m8f?PcAp4|@5$-9X zRE@!gXM#@3{EYo*v6P^hrdl%*Se+f%bk0~0T#i3FjqRZbOsWiwepk*m{TO_;Srps; z_IO#d_hgB&pRWfsJU5O3l*+O>>bwW~4#a*GMYXXUt}gL3?0iM9=o>^DjIR@(VxN%1 z(|4!P*7Nwg!@A5%4~aBSvdYWF#3(=?LDc>^u0}bdW@no)zAytE57)2hQW(qUnBrx#HdvXh8Pz|4p}vINcw=0O<#z5i^EhI^Ud#B`H>qj5mX|O(uAeo$ zdBVD(XwT!%=~QjulwA3ia<#ju$T4JoJ9GVOrLgh#xdQoj@x`iJ0f%Q?NLSO)epRZe zPNt1m)jCx)dNT?)j_J3?$`wvFsmc1|?^9TPt%~j3qk*+~9{h9p*H@ncc$GLWqh1lu zI6!{A>a%z=QIAtw7%yPGhDG^hQ-+V~2Y=^M@QLddWjrfd!Q7>xXjfXr^mg`pmLX zWh;{li3f5!RQnz^F{I(M$%;{qeLd)~WA? zOnGe4H+r1;yA@hBnMXzg=EVy}X5hAE<1idDln1%!e;$;lYm7wdPj&~V^sM!jbBk++ zjGhWf$o%S1UfD+YnmG?iBkB2frex@)v1C9ab@IzcG3g?br~LgZb4&4lxuQ&|=KZcA z)+vW4{A&g&*;f&=Ovm_C{Kx7l#jr_<10nkDC;C>D>dG$2mnlPYvXfZw&xxUE@3d(x zubkl?6(H_EPAjnRcZF@dH)wTjK)97g*wv3G(~k9<;ExO1S=lzTtD-CjH=%z}YoGC_ zjjuJE&@5(XEiKRPgnaTln(cKx-AVR}qO&0J--$IVy;X(W7Z%pa*o9pPQax*w59TSE zoXE$F=XO8Bx@#XF-o*n&tZA^{TR7TX1_=7+Khmu(nc$xhfON~*H0$O#SkD7L?w)^I zhLo0!wL&(z#+cPdnCrOqDtdlZ6GOiZE+hw`BN+T^qttv2;w?C|!W7gXQ}^Ke*!3c? z49LK^RdtF+!zqmYD*VuSjt7)a`%(4$MMvey zx64;q$M;CXRM^roxlOHEMq?qzwO3+uw2#iNKbEbv3tPXTU~OZl5;T$!gI13tc$YPWK2 zWo?eW{{Z3kp==DgWa)PZ=Kzd~OmvXjfZf#7ld%cI|mcRZ9V z3g^a6%(mm6@fm6t}~6rNFm2i zMh!Hs@s}C&9qTS$#)?m=MPfy4Nx2Oyk(D_Zk!shJNc-P|3XQG)Y>?6Wj+HWrqME+t z6AVwzb~mA>v@C-^czadmMk6IefO>VP0s`bDn}6AEDtS#*u2Ls)`MCPj&lrw1G0U9w z6=ry3R9`U$-|q^IZB|yqEuU&uv0Tc+Ln}LlE4X*Abk|&v8c;;aWQ9h&KUIer+Ame7ycpCp6YivCn}?&k4naCg~lE{9?eqZ zlCf$H`WG0yc-#p0?N3?r!u-`#NX(!tp}ma()o@u#x2W$&v=gUt_;635rC`Gh2~+*x zaa6qH+YSvqB_pd5>}yEO&B3E?!(P;^2-z`_NAR((lH101w`0oGt$`iR+#2Bn;ANEk zMMwyf1ador&2Fp77*7ugYSoVNeG@{~Kx;|TIS*`#$iMJbr(+p(9&!#r9-mt0qm|)1 z;*q0l^%cFRc-K_10a=8xcL%AebAKSyKr8(zkarjbnx}!x_-x9IHv`XDEOlWc1wr{*YscEc4 z@DiiA9V;-t8yLs2tYrpi8L4Xra*HDt$5Brarrx+UWsn{RG{%U4YIzZ_H*%p*m;wO= z<07~8PY+3be-tqxa_j~%RkXb{+}xHZ4v>A1KmhBVzP`kz5w2jgEET=<(*k=e+$LPT+c#w&Z`FOAz@y?ZN3 zF=r7YXg_%R*Pb+chUZc-O{cru=~7nZTM>^l%V`5-d(#%m+&{I<}Fjs9BK%9uKWUci{_5q%vAaGt>Zot#Hatc>L=1 zzO#L3SId}?KZMrGyk(@3&n!IR{mB#Gcz02<1-!EAJqP*CWH>72Wm2u^c&^IN#@UO> zyo`NF&;J0bv_x7Sw-G-oH>l)M?Bg$mgmD!h?NUE8fJY1Rm8(`7beFel97;%uE-Zyu$1gC`zZw*W9f&%X1hsd&QkH>V|4+L^!L?+nA? z-7HHwD44EYk81N@iC#Va)wq2(PILBH&evdgb6p3-e;q`Q`YnvO#}m4L=r!Y$?UGW% zxyY_~Q@_0TJ0&Q^dpPV=*}Lx?R4*P5F^{END8rtf#-S)l$;hnb5R8hZA~^=4VUE>x zlZuOBV}J+aSvVT0mD4QP%{`r>Y#h_x5~tAAZ5y9@i!@Sck}SKs)T+UH8mLAOr6F=U z9Me+=3+!Q1(otBRptMp_&;dmhPy*3OOF#=nB@_U(w3JXH?wc0-o#bR7&*NJk>c`4{ zV_CMsOKWv?&R?8=DxN82G7wOiABDw7B&RJ8M+sfiT&UEwS+1@U-X$+PoG2CQdRCQV z;Jr9%H@jri9e!2*?Q`A@(-T0qePdHm<`n(fw^3L=Bk>%%wc9hhZ2k}{*r!G~dVJG6 z<%#yTQe(VDtljH3Z7ax54+5~TS1m9HJt^qU6LTEW2?HtKG4-!65ntNM>tn7os7+|~ z7*$zIZ@~0wruNM5y@nUID@kqAW&^c7?cEz0`9Q9Fvx>f6p4yo#Btga*2V>TjE6KM7 zNg6fuJu1_IG1ylFxuh^{1ruv8zgl-Z$zdl+`0VtFmn0+A^i<+Xi;oHDn#y_lRoVi20bV?@JW6JcaNu-P%}-;i-dW0JW|9^^bfXdctG1RmqdMVQ zl=M!|@H$On;q~p5&o7i+eOJHMxtVnPn^_LvMn8ubAC-36N5q@xBUQTDbr(V(?*aX5 zq>ohauZfpzsnxVP{&f&G8Fg6-I@qtez@n>qmtcIEgKy66*!-8+`6IawQHrr%Re?BS_Gae4|S zWs|b$Wkva0l_&Nx_%ARepN+5LHAZ)Nf)p9)Un+8K(b4h{^Rg} zF;3RlQJkUhGh0vBbqx`OntkNf?oM`RD1RRR0F`x?{{Rppy;Xz6+Eh0v!~J}C$oxP0 z^}?))gC)dXTd=@k$K{T{jXWsBOOQ@@QN}+GmAyPnIY>V6W{beUdR?F+=*38HKdsU7PB3?I&uC`TZt$VcJ8uTk(4YZs7a%}fy~MV*d%t;boNxMLwR4x>58il;`aSel zuz{3P;f_D0U$fKQIbH+XE7E!#UzZfKNb#0PNy+n)6aw^nf^h+>bQnm+!A z6{CHo-dV~by|tDU80w{zem!cYOs|9(k2v+?{{TY8d9`UyF7iheN;PS_Nfh0txhH-u z5f4&uarNy+P<*k-VL8sn*-8ubRaFffwR;#SrPr1g^A>i&)!!1r}FA4w*(;DEC5N@%#8m4PL%Fc zWv+#HGen{{EW{3t$Q0z+j>A1j=A?u=fKpYIw=AcF`I>B=Q76np6P`yHr*lte-i%@0 z5hWa)bOXQPN|!@$Gsdm$jAD@6$+OJ_Fg-cNH_u@Sj%da@g2{vV&=!ja$4|%Y|eS`nPJbhC~G*VLW)~;2eL28xZWbEv&{+5z}-_{Biu> zkm8%M+As{O#D%tkO zS|XgDgg7K#w9^|>y^7sq4I?-iVY}(iQ(mNxr~Et6qg;uvq~;|n&VRh$)9F!Y`gNax zG(YW~R4kg5$%}XyvOzsM^!Kki*0uAiT+c1!avQl@iFc5oe|H0n3h$=XIN6aJlaIp9 z8($ac^O@rDHjr*i$9vqV-oSRiJv;N7xBmbLriH4<_Kyr}VhKq3p_RB%=ucnBS0}Mi zPByfTMnE{=dk>bOmfAS{xd-m<3E#RhRIjHT56ZKSA#EgdQ>_%aS~WEt3trGeE#98Y zjnZfi-lG`qL^A(Dm zb_cCxEIgLRGNU-`SMpRTnnidARd~4s{W!%y-gMu+l@{e!t1DpsHE|fs`_~>`Pak*? zJ#o+HOCVWtOhI;z9ROT^-#^x{acgrvXC2A|X#ht#kH42EN=aDCcHBXYQLvGW9v+MJ;w{AWyJBTvFJze)tIDpgn2hV_RRFEPb-*GDPlkr zVJJ%5Rb6rYquX4lmDEg1daa#!@`vy|J=YMQa2n`Qm>^*eeJc>E7s{uK<6EyRLJ6oPr&+#k$! zs*NO)ozgg4cUhNEjG$AhNaT#iS~VtY(!El6S($PT#AUqAzk~P84p9$ zetoKWC2MKQU)#%XFHo|pDf|ypPfW-Q5T!A@)PA4Vvv6y2edW19qjvq|@T`mf06il- znm{=H%=tg_&HGi-fr&UMc=?cpKc^K8u$kP4RFMbq02GhMy+w;txldqr%wi{fxEP4f`6CgM3YF-JC%fP(vZqYC#kNx z!X6j7*Dult(GMW8Vt5~wQPcFBtxDNog&F0QfQ>*I>(@2yDEfcHp9w4MU$;oVbdXy( z{#CUkly1x?H7yev;vW!cz5viog`b!9_gt0FBkNuhf2&EVT|C!ijE-_!H8#1ZUh6uP zw{x+UM< z4D5E;NC&2R)rjsSMZ*B&{i@FQEyl&gGA}!0oPT*v8k{ay7+FYf?l}InOi2kj4A8$~ zDpfKnZkZNzKgFJD+=)RnT=Pm0U3NY@e~n0iKqv1O^r}KgROUIE8?s6=K9mMKgUu5l z{`_2z#~fAjqc=}Y3y7s3dOIWbE5WVpJH|TJl)!n*brAjudj56KEW6Z{Ws!Zb59DYP zq`Wp&hi2nH(ylXWVnnO?@~SerkD5^9g;Iam;p5)PkQ z+4za#d%bRDi(@99Kh(HvpGWL<)3q54&hG@3C1y1V{oSvN0I&I z>q=x8D--U?`>)omNf|n8p>2R<;AX8^-d))+YkApyiO8(vaS{2$M#s45YHv3SoChAO z+O$<4MrSBJj=s;wcM4V=MjtjVnZ|MWR-?t?ZxJqh!35{4b^Um+Jlv$@AHdS>0B$)` z+ZD2@+~#@<_XzpU|gUGIVtS+sm@@^n#+@9jD>017)r?6Xj-PgL7 z&VL@Y=^h@_?>t*^9NLTt6j5P|1;B4=wCS~X(089QgVyB!0Pv)1Zn}-)0TK*=!>RT9 zR&S5IYMP;&OVh4)TgURduO4H1eJh~xR*Kgok47kxU8HRtY@i6#Exz@ zX;C2kY8e9WK3ib=8qL7bR4B17`6C9LtWaf6D$_@fBu~Ut4G=%-+djslmc^-DPqZG4 zx%a7Kkwyx(52Zn|qi|h#?@MseI}a&#(azHwN$6X7iw;ITki2>4oETJ{7`#AFxr6{4t~<~4oQWObee&@VN2j!WrXts}u~{{R=Y zbiOn3m8XZ}O(xbw7cRS7Z~U@f@vA-~*DSmX;pBz_ziMHKkPMvqk?CFws@?sz`dfo5 zO41$7&UX@iwQ5eIQhP3jkb3YUwx3~*4azQ=o2LAxUI)$3=jeO)C zJ%0+sg64Z>4Ft@X?Vdj>=yd-88edxCIn&IHb>}~&bU9O=>cXloaAbCSm_FSc0P zKliDd0rnN>G3y!*v^L$uvAN(JaaHcT4GqefG~1Oq=V)yHl@g-&h}I=7CaiFe5Zl10 zwqT56we<}PQqx?C;9{WTa@Ke%1_;6QBbvp*tM_(VN>0YEPnt8{q zhdk6-wjP?PQAO?4=aSvzWsE{hr-fWsQ=s^69Xy-;Qvk0jjoQ6N*TB{~eaG6weq*mA zj+Mjs#>Y{T;IgR;yC4Bx))NgvFWuDkRvr~4dS@}>{Zi`ZXu*ymFseANLNhF9bbvDU zu8UBT?srzZKzy-*#b>lpN||g7R|aJ{xpLb>inbDT<$kQHw-K<(C4G%M%=3D5YxMw)$1HET`shNV2_=>e0i;OABH7!`` zr;UBdRh73caY9GN)ceQ1L|H)F{8c1OV*{L;4#kLs*Sv=tc8p<{c6-zkMxdM=H>F%R zVN~zr_7w*YCgKL;(wZ`s8WlA~L24B4b>!4+g#kIp^fh>^2juWORBWDOA2Dz{U{k+y z&aaoXswu}jis6egW1#I?QR-#K?~pQkR1E;_$oHzP12f@rquPba&Niyk*w#~Tc*Dwn zFWA+Q_Ly*q^Txl#y#AuFw!C~VZsMkoQJNwnjq*=j)Glq&tt@n-)uwg2UyOAfI&ikv za$tW89E$W`5BSkMLpZR!)0S4_e5gK9J?q9*+Wn)qf*j7rk}J?Ou_uJIBW$D_>%RT@ zuVVuWlUI{H7~>sC-b-HRr7fi+dR4YD%zt>BReO&046wAP5Rl2})yuQg;Ay5r zWiwzhw>G(@dYf0#NY)GB%zmn~csp)xgF+QC$y*d?9(R%jfMe+K%LP71Kiw>bn@m zn|DmJ;g1d7c%t+v%G$@v%6kub>h;Zj$~YIpdR5eRws0v~t{ndWtUiLNc!$I`x+at~ z9T|f0i;0)1K9%Qe(@)nD0C~q+>zw0<(Q0Qlo8^wh#*LzZ{=k|Yud`OKZn|QW2L~@M#Yl90K3`K+V{{R|ftL2b+^E;8jAB}a>#@k7& zo$$>Rmn}%0MyKGtUrvb09Af4p(c$a(MJxLBtt~&r8s3bE67S8A z{&`F|u980%Xd1W>>za4jrD6WCbo$or6f13^PKF-8!1Xw$b8fg|43F7{Kc63!L?c+w zm`fv`LgSCf^{(dQ!8Ui%o1G6(jq}kDJXbe)qH0=fpXqkbw0Fi=^8Wzq*F8)Oma#ai zU}{s<84n05(#n|wykuuS;-Zc@W%lL~%jLxvnIw!5Ho|%I&rH{=XdV-?@Sd96^m`d6BWSoG5)S>V;@shBlFeEpxncg+zN7BM8nb@(DR+7!9NL+z^cN$J#fu4VEM{w+Huh_9q_Oh;WaIsxPpx#ee-JL3&8_v_GA%|7 zVdq8U{yhibisl~aWhk;n!8$aGau3p^0Wtf%tTUX0)BXk1^sS>_Q7$#-4bGTpJ`?c> z^K^|%`#R}|%@#_pJYXw)%zl-hsrXx7(uUTz7fMEV!aj9T{pB4$5m@LP<8y*rc)?-qo@G)uDN( z>RL;M)a)gLJnsv+58>DKs&Ivww{9CiF2D`g`mZ0(xur%>+`*|u>Q#vxrsC1uAV2Ed z;j{eB1VE=dp4w1zwT4um^XpeadFRcQNktu0Xa4}zQ0$WAff-hXe}-N_{Aw%;cifE; zrkNIXNkbpuU`hJdOW?l@Nv`Yf2F%;Qz_JFxAp6uh2ZyE9FXMvN-J_An8=UR{9^=xz z55!unpMdl)?9C`EUr0ZHApRe%ZALcKmp;TkJn?)QHj;D=8Ob4b&pdy^2W*P*c_uTG zSg{>LPxGj*Q)xt=-J&GqvwsyTwL2FC)~|BAqa;zZb1NAJ zN+WVUqMR<-lq;gT@EG(zN*>QGi+Kc}PQO7>Fld_&q!H>XBTsa)qBc3Vl*D_fIM4Xg zr)dw#HMo58T?QBrpyTwaOu%#_ntW1E8W%+@2PB?F76lewwg+QKbNdu)dZhPnaPdY=4o+AB|I6fM54-v@b~J z=m*xPHwz@nCykMRzFY5pIKUr;T9aKx?<1-I0EF%D;|bzDE<2cj3e!S>hyMV1xj6Px zC|?hFb6h5E2Uhzm%tl~i!DG-WADH9Un&!sbs|Hq4l3R8;^ar8*s=TnElZ7iF8-ZVu zvVHl_=~}9{+G_6_-yCUN-x#uUM ztA7#v1$W*U(Cxf8rpv49=^LzSUNBBrw_FdW{{UW-q13{qIVEV8bju4r1L*Cm>S_(s zFYaSeoxFNimH3BW)HVH5TeoktBk&qw0lQW7!qqN(EqagHmU zDy$w^qXZsHN6t^N?fDu~Nk+(uaOK>m3~-}>lgSL7Fzb;~%lnn&pD|Ki3J=G%Q}~PGJwokcg3i|FJtiT`tY>Tb z6Oq%>ta#tW>uqdw9Tk9vHQOYra?E{&ae8ly^?g4Oy2pP4;B9g9fIT}3(X}Ucpww-) zrXt*V<04Yn94_vkf#>q15$`0B8ysMDABS4)qrUJriBzrpGHJJGtM3bw?)bs_VyWMJ zKGY(KX3}q=)GdxmD-EoDmy^1t*EHlwXmUm}#O+||-56w#@TN2XR;iUNKB(7|LL7?%aYskN*H#vv3Q^B7~^&Df1&)6Z`S?AC(CVZ5a^G zcYkr}+eVpC2Hf|~2j||KHM?#3qw|#djji}pwCq})^e9LaVI_%`yO20E!E2SwP9OSJ z;Sq>pIY>YA&Qbk))Ki2*iQ;ISjvYB4m1g7I(LTiqHz5I(gY>6c#*@vlq!ICuONQusuagvm+ngrVps5eUeoKt{Hn` zm`*zWv?jsX>|94>2g{H;{{VIA6scfcS|gsT(y4vt0Z%8jSe2)@+Z@Y)dm3s&W*>GZfMQq@nd4E2r@8ns0BsyLj5_bLTgb+hIYnQq`~pd*`%6jZN+r)O8wCApYfkZ& zsbK0RWw?!aAgIngtDW8EFcGkqj!#CX2@`jhxT(%L9M-({>Nig7RClYccw=84R-tOP z(>Ht-ADvXV@G8iSHj#XispKONe=6g)Ao*L9p1prcv2A;Ir^p)8=1ACobmV?jN@?z5 z={5W}Zz1}InW$-}ZRNC)sQx@;XX{i|!>`NF*P85f{{S7%6h*4qM{^cF`iznIRll+L zPsE=!w4Y?TWA{P9C)X4xr?Ul7$4STLjtrvbjt{8mOfZr6lsNUSr&91Ht7hBnieAUt zXAB3@x%pMC+{Zj`8-s$5FZRRY$ zKcy^kNH`!LtuK)i;Zd;1``lKJpQTGJ+CalD_{9-U_CYzMu4(9cRN9^7P}@LDwm#_1 zd%d()J^}F4TwJ;t_UGN3)9F+^C!pB)OG0f{_ymztaom4O@(&w$*6YPOi%Gbpv}B0m z&ls($a_EtTIBU-L>Eu%JABZk|L99!6aXS*t@+mwh^c9_l5$A44a5|>}h*Mm62`_sbysy6z=-$mEHW{l zTDNDbY8rdy-sOj?lf`7Lq_$Wx)XL`>Y%=>+v71`l%2e;U*;{zsoCMS@)yG4Tn!MJ& z9@pc+mOrM7pqh;c4QpA9h7g5>0D|#OmY5J0gwHr_yx$XJUXir^*%~~H4x;U;R z>5NkvFx&&uyQ_~2==z*OZ942LH$YEH!oTpgtEQ4#BO-Pnaz#w3vq)kf^u3QSqLz}p zv+4qhD4;@7OGN-BEhQ9yS}3A`3ieNjHcO;wqVnyGZH?-qsn4x>e}{DOscV+dIQ`IV z@_)L#pUN;$rWULf=oY!@v{7H^tBfda=#cfX&3X0qM zomQVOy407R=bYW>m$qzF#&A8WHdSowWFo!wB+#|kvaK0r?a!jm@b(ViWjR6-)(a zEj34HeAxum)>S`wdewKaR$|BSt``=*BZgG#Q+7;P&f*;7 z6ymDiF+9~~ib#oUkOxYPPa%Jpe6_Kubz@HYVB0Kz?0Q!rccWa|5}sQo4{==Ex z5C?jOWzAwOs`7tu3RQOmwsO9-k~7QlfEWt9qv-cq#nQnFox6ZzT2iL!OO+ioT$|A7 zJT2kp)}|830$ZjIb^0#0I z^rTpFJiII%n%-$4Bal%407~h! z-ydobN95`j>3eV8k0%w%Eaq0)S13A>%}k{qf3qz$2Fs;>$;|p>b93| zQ~Qp0DX#e9=~Y&!tadC`9})ajXWXqlwAR17ng)%Bb{N6-H5q>-?N6DT?`|2c$`2e^ zUZ>ghn*?3l85KdN$>Dt;Qv!V^@l}sIcVqOTSSnR@rp)SKVU|vV`t}2*Huy#n{l`62& zqPoE27?gr|+TC&9y(7UM4u)7pv#c?O!rSKsyW`TJ@Na}({#Dd9LaAdA^BJ449l!e3 zKaBiRb!%s`)0rUCZAKLb9%tcO(xJ<%Go0LQ+{y8uj9*)v!)<|K3EGf@kH)y5k+xn! zRr>={`2ew*gD;_9@TZw1WF^%khO(IBSA13Xm zCq3vIv)q(R3$$W9#>1&SDrsby$CrheXC(Lh>XVT2IoK4r&KIDm5Hw?Q9%vM!ZHp?% zjr)ljCOuX9esuFOhdbiKo{~g=rB?I67?A;`gajUBa)n|1EzKfreao<1G;!`OCEPyd zrQo{K)8a4+iS?@+aBeEkZnGly99! z*Qe`Lw7n+dQq&^0NFUAfHp+~@GGn%T)3gWHp}vkCF7DuCq-2LdpH(~>^*Lv>@J^Qc zmDBl}q~j{CdEL`_{{RZ#sNA|@D|t&Vc;*W~v5;e6aDDwLShY0D-c(Gi+lop3#uUL6@kGT*2*7Z?4QOwd z?9Jv!6!9NGH~xKUq+k}zQyBLB*yS0?{#347Y^P9kXL|dU`%FvMBQ$)wCu%6#anI|E3?-Q2iXyA{i?^NLn5Z?shSoK!qA8dE2 zCV0Ti_TF5d--zcwhe1yECBZ!l!e~$h8A8X9N?6uG~cvow&0|27TbgBM}MVX@!yQC?ru{`w2hxj zfaTUA0FdJuByt6L_G6T+s>H5KF6R0EnIAXjU9xU5y9C5|XEz91R1xtmMA&mfEz zK8l&fKN_ch;O%!!k!OrWe*%2|WKV`3fS+HcYX$A&l?j~zPzlNFpZAA9om4ilj4O+(a^e@4vl)=qV3_^qopMCh*?2qCDrynB+IGQ}Z9hQgq_= zh&W$kl=9X|%OgBQaq=!R>;`Z@I*s7-8x0ce1D1$jkErfzPWwpJ^ia!pr^)4!xul2p zh(Yd0Kz~}Ik|B@x4y33&9>>22ztXdnIdvhaBC`ou{HRe!+u=_N3s-R5KC(( zMF8$o>(6S@lo?kJfyqBH#z{V>KR4x36e^Cf$ghF7WQ)7ML7qis;59ylCTT*bc+zM7<83FA;1~=SabOvl}Gm|#U+X2anAV8dH3|E86;%d3|A@Cj*XwM zS~;$7?3Y69tL7?#yTJK*`Nn^z;ZxXL-|0twoqIL0ErO`zq4quh07}k+;bH}2c8R*| zlxH9Ao}l~Hhq&^H5=P68cSduJ`nvLe8cIzH(pnv4UOk@T1h?0;_->)j1H_=H_wvtE z_=cb$yNjn2X{2_C-H)W3qExP(pN{a!iYX-FzHiQ=Pz5HpGxkIqQrlZ+=YS4oM#Q%nDNdWdApmiJ!<{aG@(tz z3gj+4*Z>7bzCW#A(5;O3U0I<~xbIL;=TPi|RTN{)%KAo?J>}$4Mu4;Q_pff!EOhUO zz8?P2)eLKN+1LTeVtZC^!`}`;qv>;M+KgoeNd?IHhf$A8;k5P7~ zlNcIoz^tj*KU&69v$4?{dKc%II}$~M8OoiFjXz!0w5)xW?TQcJmy!9`M!q)FH8|t9*Yv@T@+b@| zh!+Dp8Cowx&>zHbe;RP(fsTF0{{X7AQ|x6bD|Wu)vXjC(Hl~JM15**PsAUPnU?Y%;!lr0m4q`K(NFriPuom6~z}!CZ z^{j6OcrfY{$0UejQZmlGe81ybemwC*Es4?eTpL)1*`!0)&}Z|gRNpp-JCms{XYlm( zF#aWY*y_`p3rPsEP7dDYywXT_p=5Az*!8PugddU0kM77a5A>$UB4s963OD!Dr}$LK z9g$@>o?&m~x`4ia%U_AoU$-%W$B8tU>f@qvph0A&hP9(x#2rV9Kj1`i`|~ zOovGp%!HmsGv1yCS8ohQtyMC_q?lwe{{XWmB9t>^Jcw{d44o@Ze#TClTNh!TNDla= zXxQ<=u7gkUuBBiEa)p*YfO~&h<`zX|-ML@bb*1wWj(J~Z;;uCsF`XqX;&F6R(Od`8 z2}w&uAqhnkPytCxMFJ9*l8QiYXcC)9ZY}`%+<9t!2UGn&DjgbEE>3c4)vdb2V-#oa zV<+lq3@1ERl`$N#RO8V47-he@jrS@Qbj}VjLf*K-=BH#m7*uTno|VmeoRh;;r2W#= zcZfg)jQiBSVz~Jm2C}4yl>D?t(nZgDTWk=;RHpsoqP4!dv|wVLw+`aEy(i)w^^sWT zK4Ol!uRC2psQJ3pnC>H#Wx1{C;-^n@zMf$jt4mYdwEqAaX!@#zoGLNvnx6N=+Fqp* zcnWjfdChog$WgM6N8?($cZqd71_mjaRCdo5=}^bT^|8|lQ*ZZjv-}RTSMU=l-4YTz zV!12L0`}HJlqE%WnpebNKwIPoJ;ij%YZOfCcN6XYSi7DJgV3!$5gCPH#>e_^yD&3>l|RS zw*sU#O)Q&YLd<eU3>BtvRl<<8#0{uO_Z2)hG9vBH?E2JRP{#+8Ol+XR z&S{*E5{Y-uiM~=cx2-)@qWOW{=~NL65pW+qgk=&Q&Kp0cds}e`c^{%Q|>Xh_nqenNZOXJw}LkEX-^~@41 zQa5bkuys!Z&Y5Tjn;*4tk)3qWcvy=qtK<`BI z`c*yXGaC6d3X7V9Jo3j(xWAQShBLFfG4I zSdQK6g1W!Fg6-sGELdZa(zlHThKDpyKE4&<>07Cje{ntCRzEgs5TP;%!+MID znS&N@ywryXKiVR^3i6bu+eeJsS~NgT%;1J5nyQ^o%GB`|Ic~nxjJtVZky)kOr7pr2 zWE;0;nqQ2#r9}?eCjffW-<%g;m)e6xQp8e5rwpcy%fTmsM$PT=@I5KumFhAmDpQNx zP3Y&4R+smZ9jZeprZX&qXdj&*MnRa)C@0zJa>BHHDlv_^8XiVb<$wwP^3)442vn{~ z?@s|oDA;cF5Yc8SiY^uMPh(OEl?GUR#;Fp^_t!b@De;W_+q!fOO2%#l#fd`iBr!aJ zN}hGE@1vFCa&k7BujXw#oE1GqNgctHF5of>xK43?;GrszcZtzyo;1|0krL?^;v#YL z^{&@R(d~RuG`%FXnhAmSwAd6)Q& zd$g+0+56{gvGR;9Jq-Z+ad#t8d*nf#R6AEQ2eZs>>n{%~B&mRI0N={Rtm*)kx9SBO)*xY3n-$ zD$AVprLZ(wl1Gh5+8+RX@%5{gQOPQ*LdDA;xZOeiwNBbC?Hq7M&m@@n2YU7o1o%=d z63FYC*hu1xZi+H7gImIkp08sHnnbmsX*PZe&>vdWLj9%x0M*CeC$)LkiZxmFtJ8Ze z%FNOdaHk;$WyL?nek{512CU{ecH06(ap#OzGa2%MXN}TB+;se_dDT!{t8+NPzj&9) z@-WEqM=|J7?KQop_{&?-$CWgxdk!(>PxpuQHG-4vQtgQw{h#YjMvg@cSO(5Yo|S}a z&MeVQ$8)kxTf&<7WYld%joIq|0Cea13a@qGn}v*A>36W*!P#BG{Frt8E0=i0h)XZZ zI(HOVU0>MC=GfiI2wt4zS5zy;YSS8&UgWoSmo`@(UZHI)9q{XreuuB?Op6W+hXOVA z8B_R=>s=k6j$~+xYZ`i8M0hT|XY1Cj9V^2gAW{GhLGixYb=kGKZgzIhrr+zD9STxc zgA}5=9CIp7lEHu<^3pOtpGr1v!u*cC0B)o49czB;!g{@vGps`9H6MB-^8xQzGOa#(!AH@@Ot0yP`Cwnmb{{Y6Dx61id zjB*~MsH}NxSF|eOmvFabnHcmV=KQIJV*9S9Cmy|O$B+V&@q+&V-;KI`MI#3YIZ1%% zjyWHND_Fi{%C8He?p6-I)dOx$U3&KDYS2VtTXV!g;ZHfK=84y9%Bw7R>XbB6m5MSf zThPK=r#txP+NZW@UO+i2dMRA-O18;xz()A=R<4u5z8jas8k|wagUjFZ5<26s{3_{F z=3yylYIrlj*PbQ3^JE6z#fek5g4pgeU1yGbOKG8L1H$?)3k2IVig~LZGS}_A52MGV z=#rvdImE}QL5%(t;#W5lTu$>d6C)h4&o$QhV6XQOmC1a_q;h=OASbC~kZO56!270* zNQ119oN@IWRTsICVmT9!af6C_vN6OmnA@##N^P>7OKEh{GWk-gWB43nzvODqoSE7| ziOJ`88UCN8X6|Jp&xAPs`fzDP&7F`;2t4_3{&}h9A=B(>95R0FsRKN#59D}1g+gt_ zZl)J_?mIMOqkdg~Dz2^MsoxtzCvrI-m;V4>r;c_=`B5n&j2@p+TEEQqp3!Ivrr0FA(ZELOhRkJ@)STdxCZFd?|pF5|zp?Y!+L=-R zWC!K)=Eon8O15MkU=GASIpb>s^zZspjz{ju3`qJM`-;WVV^V0U78P&bM9~~=SBs%P z)|gC@fwEZVjBqo?57Mo&?NbyE$syzB93R8AEwzf=N)ez|JOa@-%09mJoTA6=5?5kD zVqDrix`^6mC<_or{{XJNr%Hmu;qQi8^5!ybSq}_BLiI4F13AD;^yh(cag~m z=cl!2M;Vo*!>W{B-!Z}bJ?i0xc2ZgaBIAdTkSdshIdEftn4S>U6V;s&5<8h=&RcOk zh^dvW-0mR6}OsY8`t zsOz7`rdEyih>q+p2>ks|Y<{(*WQ5yWok0Hpge%0dCf1Wky*+UIGJCT-e52?9s<*xg z*6dfzvN7s5rB2KZj!&}v=Iza3&l*V&G2|`>&GUY|WAZgUmZhp`E^aj2`Jhs9hdY#y zO#cA&s;>H!rk>?SlkD=mS5e7tGJw8Z=v4k=82oBM@}}~`8@8)A-TWKwn&{#2oEK4u zd`Y84GB**;8s!joJ9?A&)Puts4z8^?hP8HrOcE{M4o`fM#=-g1v|ht-i`5*$TL~oG ziz1m00AJ(x^dsqBli<$`yvz2B=0&rF?s+!t-TwgAtd9lwUiR}?M0nwKwv%kEuen?f zbC1Urt>fPnMQ#_vnpVi+jAmCx<&UO4KN?Pz+08mrlh_??mXWgEs9qQW!Gi|q+#CiE8><7Pkl2~OB z878$M7~o2yii;GTiefUh-g-MwBMmB1@0rCw5){r@q&FvYMwlnR<)ex z`-h+Z098_%E(O@6I~xv@hB*azZ@^T@@srGgGCGMk+xhePR%v7#gwPaW@`s4<57WIq zHkN;uH%pZq1#yNSK~$z@>EBZGBDwj%+x_0WdgJeYwJn{+<)$UFy_OKCAO_x`d#`WC zsF_qs5Eg`v3Q96F_2Q=UT17@^NsOEsSLOrQbgfhBPb=%O*V%aPSt1tt&YvCJOaN&j z8E?lq1MmW_+W2=$@gA`hx*epDTXeW83aB{ghRG-C+PU2mN;;kDDitd2j1!g~fFAYg zT8-_OfV3r85eB0={E_bap!IH_TDZvOT)fRKivB6p8%5Dy!`cU(J>sDA+5x%Md-UjW zUOaCbaYmJq$;&IY)(6n%kJ7bb@g}ursLOi=%xu#!XO3n0TlKA|{8gcAD6n{sN%Ehm znO8Xf0Cay3)aHkwwY;=B22jUq32>DyT;ClWQ*FS?Za z^r@y#v(2(8#DD;DwPJgHjuv=KSsBl*LM@KX(Xz7Vr6!B!1)C!h#vFRos^k7xN>}}s zH33r}-&|+bqhP0@p-G}_TP{f+WG9c}O0nu(I zblQKsz3HiwD4gXKQBZo}NlQf_5|WmR07@xoDF`ZCX#C5R7~?rL6<*WOjq+oUyHh7+ za_Obcu4;)})}~47kD3Xo6pvcQ-A|#XjZa&b@{iV|cvSQ?Tu5`lr-?`u?slw2Ijuz4 zRG#$4QQTC@;GSvaQH2P`rW&11jPq}kkxU_@>BVRi$4Z5lCj?Wu=T^(w)fMBOK-~1J zk=}#yw*XW|AbJX}@eoZTieB25hM<*Y7zY(%8=37`G*Xa9p{ylmkC19*y(0kgR~gP4 z#hcN`${HQ+mE-L~#Hvhu)E$Yg+QY{hZlf4U$VSJ0mE#fIBMkAMdXsiO>i}!Ejy=ct zhoKB@Ij?(j+NaX=-AxNdP?P=I=I^`(=S!GWG4&P0*lV};GonE={F3RY_Vd<|JMdk4Mv4isnq@nSDz?2yGujJohiZgB3UwUI#k?k!0S;PeS;#ScV*(G#di{{)6}#jW%)s=l2s!F z3eQKu$*DZSQ)77IdKTnVJv#eUVvHL-XjUhj)7DG_nhLxmMZqMWmZK`ndgN805>B*@ z2p@FuO*5XY4zVkO6jEC~@-iQHu9zC4{B#Jx^MlXN85E)YG@|tDD`ofXJN&D{xJtYX~-! zSM;wnU5v!~nsSp5%#L?<+HqX8^F3OUx(Zps_-d=2rSF7P=1XP-p4H7=Xm?hw#zbRN z*m#>zv&Q!_C_N2ymcA>py-=4Z5T3ae*K95&Z)SBmcwAc3MsTkiFDwm5gC{I71$r#n zHk+xh?pbnu4PjsSR(MqTLn=KDa#zbFwS&3GUnQ#*%}w1L$v}AEuf0O*S&FXh%dVu%VI9NkrYs! z4(ekeLGreKl`~~WJerISbKZ*;8MHNiQS%OISS~U;)3c0{Gfffi%V!kI=Dy=scx>dH zcBib0?YYL&Kn|==YG?t;1YlE1w`5k5(Gr$s8;fpH*w(*?v>T05@qc#93pZa{o5TJc zeL_@r2g?VhI@e>ZYWCh5)5W!zQFP0WKZd=0CJ{zF^$K*|IP1G6*Zf0m;e99jCrR7q zGU8cR_;)p5mN3Yjzj(B6$RIXt)|S{ zeZVKMCbYy;(yFaASiI4%?A)}`GY?QQ zDv1mCh{3L&$Ht5k%e{y)lgT4BpwVbr;^r{1%6pDKN>akntF1Ofu!>s2oHUBsLcZ+$k+yT#)0!Z$oymjiQh6a%90p@SEfSxb zB>w;jpsr=ms?co%V`R5DsbPWIK$yd^01s*sHfaIaK|L$eJPY8b)Gq%3wC=!5ct9k# zdfJUXXPfw z*!pSm?c(`Zj`*)DkUk5ZMRnDvqO4^f-bR{2T^)ywudPPg*jDBqprAw7v>@{`q85*tZ+Kv3PUG6fu z$#Qd=&yw7q^-a6xeQJQ3olCLsV;WetYofKo#v6pnoD3{ z6WWjgj9?117C^nFQ{TNr;xcdvR`fNR(9@daVrPh98}O$r@i{Frek0bJ&`6_pGAX`d z`6U`JoBPykG?OkvEUOpp@^kZJ>q1;JxKT5%2i$(2g;EV1t%YKT)ErfnmDl8W)^zpge?GN$A;h>Q z-6zh42X-BQI$W%ybVW|yGB_vk?^Fb+4%lRP^VMIaN*Km-BzQmHkD8?;Hx9*@wuy-2 z3=N9ESFwC@#bx;aFK@9i+f!b5ZO1s>gnV*Q{1e5Ofzka+yXN#`dFM%F*TM<0>m zw2ay(Rk_$diTpdSFPERGpZ#*=St3((m-kg@;kJXfvIx!*U-^ zoh#7%3E@~8-F1yfnC;;Om`J}U1Ewi_DWz)uCuo1Qq;qKj5+Ps|5;@Cy{xx^R{vy-0 zIi&E-uAnWg;!vg3^CKSFJY)Q8R|vOgf`Xd6=yP5&@pL{Rx|R*nC4?DUbdQXm?;peY z*9g)4ys(rW0CCfv#=3n&!1}G6OEs%SeGW5}KQR1z{u!=zWO$|7a}z}4_+&z-^B>lx ztQ>vSu&-uTRWK6j?{3`y&lviOjnK$*J-Bqo4nI~R@mN#p_1N6*C&zcDS8cSxhS}S$L|vi ze~5bhYRXS>tmFJMIp#0Ym+T6n0E#zzs=Lr zk4mV>8A_GNAOKG|{0}4Wt*>KWX>@e=9tF6P0X~JP#eZvzEOUM1{^;lR#dDXC-dPg{ zrD-fBK40DsF-Ot6eiYbV;=*0_H*?#P+?bqzK=lKY`BzVCCqY{4pi;8*r}yD#zO#6ZNTZxSF%j=cI~B4q1iE zH}S|Y=N^Zehz5spiJAs(7)WxbxEUR4b0VarC02@HFk_K+H|fXcQUcAmZBhvwI0GKz z_-Y+7-SsFm2uW*;UoD#%k`u;0ohdUe)@bGv3}CKs+ed1=BUL{p__A}f90&S?^3)eL zg`~`}uH5k?O^p2)x8+$lG|E*I+@#9{jtr5Q4miNYF(RHh3){K*vNr>RjB{EsJh8X# zohRc3XHZBUpErM{R4EaVE?&`zISdI?{Y^uGahGD41U4Und$7U!)P#JX_x0&lBw5e; z$kt^e8B7z8py^PEVYn^i$Po8d=%9Z(l@hOUg63qv+)9CgyPOYiN?$tOO`tE9-2ojw zzglFFOACVB_xXRqiKX-A-eOqz0}01L=|$RMp?&7Ll{U1Z=~!%EG-QvbrC-r>#TP$n zcJd=mHnV)h`c_t-0BexkfV|{nXWNeEy^p}254W{|>zb$!@|JOMfxsQOZuP8S-mJoR zX~A_gv`s$O!#Xs&_Nf}e%DILvglF5{yth^HW!|T%U0yA~cAqFwxMID4_8-!j6uK#AN1I2%Fy zy3@e*ck4H7lc^PIT+UqP5 zNpCJo28iJC{?7n@2D&Xj#=6~sKWHm{jJcJ|6qT3Vy8ac<33mBGJBpUy5ywDOXZ00U?iTFG zt_+aM7WNq+lXd`<9x>?D^>XXK%*wDkoSYUOnHfKru8toZJnQ9nlSxb1opRE`s>O5I z4vF|;m3%+py=b)B9;mibFeRgD8)^)6JHBQ=?&MNvISyV_k19~KGAPMpI3J}$Bn6HP zYW(tgH{xq!Q1IrprD91e{?QDBhqoM+9n52&%vMx##^+)YytgR-04W&y{It0O&1^u$ z*Y|NuasL23rQ<(N)ft5&KRi!v;PA+I0&74KBJX8aA#;)j;5`6tKPo(c1{PIAw|2+L zRnKfUbL~*#H8VN<#RRU)gy>41b5OKUw(eRc{_ZNV9pILPnMQQnkdu0=|=YA<$p-6bs*C$=RNQAh%cD4;@7 zOG!usu8!WqXe{CdNZh-KBi5++b4xnch^#dHWoDEI$M**<`U76+dEq!_<}f~(uIwHX z71MKPtCmowK50+T@n(rZKlr&|rDep^==AaBQoX{VdH9G0y)U;)T9!C_OQ3IMpKJ_7#V-$^K>C`Joig?9E zR_4C0rVosI1D#(~2RMW|ImFXWFjAZ7t;D7*la>K9$;8cxnwY)g+V# zml-6A^srbdSNEl{z_7AXGxR+WrY^1r?+-wGRqanySB2sda=dk@HEm%aw=u={y{n6o z=4X|dlh0c8YGNUVkF$al>$+01H4@@zrrjKk>&0{$UyXGO7Xn8zApZ4nV&sm!DrLG^ zFb#O?7g931XGJW|Ykf~#u<_N6_7ZRbJRH`m*=bs&WT1@Xe+jP=R$t*b`c>;)X73##iqQ3iB%As?+$9 z;n%>`qq-|@+A~b^3mkT-oi}|co>(2LpGM9uX15eJJJY<n^@bu;Ex_7&n{D@v@c&lLeQ!iWIU0N_POv4qsW#cKMV3#SmdJ|l*oQ?%K zS#i);T^x04TRjfy8J;4RjoluF7NezViL~qo(A8VN4cpA6pkOoTYlz8SP~iJkji$GK zW(Hic`V(HAJbfyc9T}* zw9~Z&+=WR#l;?w^RbC95xn>@!=A|Qo#_m{9R2+82br)V4I}1c}k4nOx!r~YCkbz!( zTn$KUk0Q1Py(hMYf0kU6&{Tt_2>EJ5#zqY_u*-p5PA)FY91#JnYR*inwwM9r@y*)kdu?Z^c6`|1wGAPHy3|)8zU-}dF6IQp;+<3 zqzvkKH1ePidT2tm=5Fpy(;NsrM>Op183#1NarcEsw=LFyu`+eb)WG0w&T-nSZ(x3I zf}9y&+KVD35@ndP^&=HzN*k!+sdLw@Cj6bNkSSZaE(w=T*_&vmqRA7FCD{n>Dyp$n z$69~d3XP+!X&RD_u_Zah`>fow@iq0Ns;+#*9dldO`u?4$D2{m3arLewNTqq6ZBnyXO+Pt=V$u4+vk(`oy)y|rw4H4zFDay}cn>><|&+{og zPg6}H8BZlh>}qUfutosxDmfP;1EqO0Zcgo;Nu{D7ZRyKoE)KO3<%5Ugvl=1%ni|;i} zMXjyfqPL9ULY@tJN^+cR_UaU86yAb(+T&c+^%uF88NSYb;8BpH>0DD54Ts(4bmq5o z-x+E)Ay(l;@$|zUm9-A3;mv9z9o#|*o1=6&`qfS}>c^Hk1e4t2Pso{UF+ZI%%UtdU zpGN6jecysiiRWq8D>DB8bk{v~rrlYwTZkBAj+|E&Of;jm#43wpInZFU0>10gqr2EdlMzxvlEt-0A4vb;ks9C z;U5fLYj?=I2GHA!j(~cf;W$Ozx}*ILqZid?90`&vySXcfX)wJa;HLDnhSBK_b$E{>BWPQ;VjmLF3t<68in!SLHCuF;YPu-KQ zD`hxO;zV-IY|JsVWd=k0b^T}{U*w}#12;S+}cZzy__8X0KNYJ)~fg3 z2a{l%OuUNfA~yhiJ*brw*Sb%&< z>eSosBzHMR9#ArqylGvrY(lmq+7 zwx91I;irLekxV(7bhT1u2knK(nAC*;uPhb&`%(8H$d!MCyU&HSQ+v)Z{+d7&E)duxu z-~n4vf{Ny3P4i2ZCBF`6Hog(kr11u`F?obw1$FVmhl`7-ml=Pfypitsz5M>JpfV`;*|QAMYRNYawRb*$uc4?{Gy(CIU^3x-*UrF`un!rcK&A z5BN;9jY=^eh;)cmn}wNF>>m7l$Md8f57YHkGkAAZ1&Gg(>>PjgBkNq6;eUsy{vJOu zNVtk!(7Qt?V;RBxzxwrj_tBH+%ld|mt7(fP-C4=yFU#jc^C$JJ* zCJQD|u5f>csQfz9L|Z`-NMs$BQM*5g>0L$NfiKi=(=MXEh;&xrzY6EBAN zuAjnv{p-dhX{VLh9Vb>fC-|#PY+P<<$l;qjD9; z_+>DhX9v=tYiSjMQMeCIzm;Jt-seJU+L9U4RX%h;5-_YfFTF~oxs?=gvgc!nHFGVBo4}L2(E+s_X zMnM=Qhfi$OizF`^I$8<7>{z}a%8T3vHfa@ zX@q-QWyD0_WO6;lBF>0Oif4vXz*p(@&03ib3G8X)6(I;bjQqnOeMcXdsAP&KK4XFl zH_HD2b%)T7KT52Q{{SqK!Hf(nknxl1eQJ28W^AmA{YUp)=cubvL|Mesve$K;CoaGbA%)uugqRPcK%&I8tP?V1^9`ueibSfXE(BPmo&FL-jbGz zD!$qO0PBA`^{q3MbhNx)h}rb6w&~Sg^lntW?};82 zai>oS!BdK+(E@P82j0D^`@=9?0?~jErE>a@gi3;Zyq{1jhPGKwrnKFTI@l^xS{`um zc2iM+C&Ys0zGaPiMqh^%A8g?Okt$w67{bcW^l$bQ~+z^ zN=HYb=}>8w;%R^~5uSTh4Rc%sG)UZ2HLYjL42pR_rE#-dJkm20oY%LC#KR8`NF9!> z61xfId8H+dL{w9ggU>Z8s+#$!)Rd)ma#W>Gqq#~*^U|a8*?p@>zzEGjBz)8+dDFt% zR8$hWoK=YK2%ETcGdE(wtJ>raYjiTB3j9@sUR^Jb(I6|01&4Pwft zim6TSW2>?8W&M!>+_)a5x~)IPw)Z$$xF^=Uxh@Gjo+@z4c?SpDy&5^4NNF{6dboV< zQPO&zxgzPBl4NYj^sGz&01704n0Ap|O}?XbXxWNWk3wsqu71gAB zG;!EmO0=}~GcPl5rX#IW^_f%PU9WJRC+zUYopiMnb^S z$PoczYq!vJb+Cuekw=-1zLgc8jUJoxx6I!4KB1~Cc7-B8GYag(VWWqSJohFv9vU3d z+|AeZjjYf>+4ZhumS)^eY0?&wLd!Z-A0uGbp-(fMbe_kfOE;R*qPjZy^lMv$CNiTn zlm7q+?55-*<8as{2=w)`mg{9MD_j z+m2`}OAil$@}6kq46yuaFf&dpSFbb^WJ)C3Gv1txOKv%+0|q?tQloL#fk7=6R#D%Y zP*QC40CuW{i*eGmbWJW@PT>(j+n#G`G$m42iJOW_?9}kCjV`084S|Gz&b`YI=U9#K@qY^iG~0uCX$!O)X+_ zZSwom1YjMqtk<(zx9=o~DXLd|+odrr^kV8L62!({^Kp;)@oU3&IOvaf*$Rv+63$j9UQn zO#p#Kic%w7FHuZpApEBlGhpyVFvG@0A*mr!U90mhed|AAZcz`iaqQD^*!s-d>7%^vM;Zf5J&S3p=sL@f-;;jkPR)LS>{{S4N;x3t|GbOSNFas6mO>*-M@?nQTimfEA zEQQPGJ!syi?*cgOSXHlQ$sLU$PnIb%&?>`oene6Nes<@5V)P9&)h*AC*6b$WvVu@iUH$n7#4gIDAKbW1tuBBd;MQI}m z!fRMEliSG#RnUXX|I113ZD6iD6?sV9eGb z#oDY+NTQFJah}xHi;cu{+ZC&Mqh8uM^CD(Z+zQVDgB*+>LtOL2$!?CA%}e2BU5?nv z3_a-zqMkCfY_DFJsGcy#gO&HLeqC&8rdQa`azJ72O=wiNA1NNSU?$f2a=*%wR%ZLl zyB)xxqSUo2n554nkz1Mui9V@+9lR=7vBwn_m1Q;5-dt!|n}9Kz_YVhnZrj1yGwNE7 z!7kmzp2O0&p$jf%Lz|oBMY}k(?}vUD^R7tyLoaUTyeGuICc4-4X>RU`Y@90RfIC*t zk9=DB4@2}O_X7{XV$i>Q;}J3)Ydk&kuu1m1}7oA)8v`hW<(5syeg!v z8@^F-LOhAkRGq`dh+-yMkB>w<_y+6+K{_>P#+;lZp%|bqL&#gHLKQg{R?A1y_ zaV{jp#m?3lwENVCWhVq@+*L-Sj>c{`R(hRXr;X%jmJLePW`v$d=k%+urub`K zVAfNuw4{B~G0)PtfI|!>DGF^MsUf{rit2_U-x9fp)#9Dw?z5r4@Rp@!%<*F4G5kn* ztX76t3oNm+H@LyA9Y4gnuAP9B86)nm&TD!<6X`ml2z5JER_GgVPsgok%TA`NLpM6~ z{v~zfaiy{Hk+^pTqu7dYpDuqY?CyLkV|N#urd_U2A$Ml5F0?H|&n+#vBfAc3IboXB z7Yt1~b+?g_(1FV?Nd6?JLvJk3s2x}n{6mpa0Lv!Q3Y=#Z+xREK5^Gn8?H}$}mSI>* zi`|>bt6DCmzk&P^HCf=exlNX~%osQr+3nD6D7-zR z{{Uj>tR)aKwX8Gb5b3eBPp}j4;kAU3dJik_Wy; zQw=PGCv2(Pv8h&NB!z9C?7055eCBQqY-rf(H~LAGW~X*{`JQRfjDHr?8<+_ls5hcKBuK^8Oynp>8X53?4!{9>N_ z&mk1x=iD4(vah@;tk@&Uk&BoISs`!cI-0>{{opB;Za!hb{Ka&-7Ok#%iwqLH;P3;2 z$Mdb!qdg>I<5Hfncl9fHe@uJ*N*ES5QtgKUaC7L@>sr2{W$^Pr5yQ4kR^@PmjDhR) z6>ChewD5L=KDDTn7cvd(7ywJ2!`{5}TCu$GeaH5q-$3&+{IKxA1KZN2Q;kVlJ1@lA zT$H`)o4>&1wOOONx@(B1k>#7_lgu2J_pG*wSdZQEI<$Y`T9?{Br>46Ln@KU#dV;|J z0EJN-K*F38`2J#|q$J*@MxuJBayfkII0Y^v0|XQEDD>u_RE;=Bcwf4=Al0^1avZ4R zxS*py%fy)+D{_B6qPd(6RYH`Z#yr@Asp!kiKwyo~lx!dFwOkQ|IaYX2an)(gLqb+d zb#a{eiPoXO(Mz#Jz0`K10`KG&-Lp9v`cyyzk&KS@aZ#CvlPDwmkHH_EHbq{N#K?cV zI#zBrE10erp~^7?^~F2w8@@@CaQ$f-1|%5WSRR-(LhB6MriaX2V{26EidHP!%jHVV zZ8Jo1fC-{T-bbGRJn+%+RB#!a864EfB$40-BvmQgO&8JGq7g+fkraQcm$%ZI*9sKJ zi6Q5O11e2aY2%RL(`oerr-`05%1W{f@Q;K1>ePsF?p=3tv^Lp7jiHq11L$esq=ZJS z3vtVJ1MAdPHkJUugpvlvG6^F!HMCOPT%=5?I-C&c!5>m9Pibuonn_)jlE~i31IN9j zI0W;@x$F8@tawMm8h?khiM5?nLeK;a^9s?Dx8gzbY>9)BFP+?j9Pqk>JeGF+-ZSy^mLs{{Ko~f-&`lYP0*s=wULWdE325>&L zRv!)N8l|zh(6vvq%=TUhmTuKL);N&uVjmM@q zr_R=fm%gVNMJ*+9pFk8*MIaK2T1o&)RRx$R>P{%DHW7K3OX#CCo3godsk)O;T}z7= z&;UIKDMP$HaFbvm^J6N4;{%BOvFIQhDlo;<_o~>ceJr)669u8S1)Mj^U7mxj6T(str!g`7)9t zgI)-XdZ{DcuG?z2R;(I18;bO(=1IF$M{IFa>Lc22w98wS&d>?;tPB4D3kL+M1|H(M z%`?X{!a~k>Ep@gYE3&*$3xy}vy?RySCGNYMCfd=QR-xf?gp^W-u%*&2gn>5h&j8o6 zk#BaOVOu@LEPfuf67Jjc9&0+7d_3LejB8P+DB4R?%56M4E!kdhcMhVp?k*D6+ar!R zuA^GhHWJ9a1#vpHr0<=X;B>8(H41;aMQARmaw= zykq|W)n7A;Rb2h#c{O2HEI}nbWvCi5w9>?JTy7+)(vIcPKNJ=XRTZiJWh)xT5W1E& zgVLJN_oghI)XN~nCW>-|kzJz#p)usu#$akal}WU7PKtXK_7$d=!Q!WMJ*p!fc{J~H zigYBM(5oW~luj{)2oA!3|<@@KQXzE&v z+eAvgIO$(YM!WWW)eoennz>WGO$}T7adD7`t#bFeJ<4vuC21_YPEq#U1J|goiVaTM z?EzL+_B9uyh*oTweP0(FUsHja&RAIGO;$P12(L(<%HH8gDh6v3^FUHg!J6}{<$YrA zdA0I4%gs+ykpms6jtS{nmwH^#6TuZk#OQuyHRaXAREFsBD&cC<_>l^xIW&agrX*Bl zK{bO1=IE zh1KP;b4t>qvA6SNA2vBAvo) zB=O5M54jxvb**Qwyf_nsk6ObLVw6a)!LL3T%9Pm@v2|(s9ke=?t>O-cKD3kRcGn{+ z05!&`*mAB#T({Nktqo}kgr4d)Tt@PPc>JsIj3&G=~-MH z3wzY{9MS~ZcPn7kRbov_A8EShDO6raxoc<&p z&M#(Dn$BBuhe9d_&JT&9bN==o5bCcGbP8a4FsW`?>CeZcI$nR!XH&#O! z+~6KXc777@%irn=ET~1eBXjl@Ps4s3ms7n#xA%GZy{pnL^=lso>2GLb-tNCC73kqG z&OY-wYhrID7?OCe#W46@&299?j8C^|UfA`o64dn@jZX3IE?l#*<27Sa)E`j0X$dDM zfNGneE1uM`HPZduKJ%uMaaxcZ1jue_1f=kJrDXJ^3P(|1alV9#RQnoI?bE2JSn!(^&Y%K!Pi{gM-wOTBWo- zT?W-ukoOhD4W~5Ixeic}M{!+r@sDYv8PSaP(CP0yEg=Q%WemNAV9jMcu_+{EVys!~ z*Vcb7{{S(44Q*QZiq7opxrk)?*7l)SZ5fyE>2InzQgOf)J+-geB~?&zPipFJbd5Ud zJ=V;s=bkI4@GpYcxWB%S&_7|j|L&+eez*8HJUu>|sFt$j zVD9h#Jt0)?;=p3+b@;-iBKnEdLmnkf$M#|aO`o_^0XCrv?4dn?`Ff1(2wJ73 z<;A{zvSj8Rss$=Kq$K=R8wh<;lM%{2Qr^ltr%)I7O!JgbzzoT;_>|1~q}Gkssm;;K zx^XPxW7%9+a`l{X)0q$HuIKky*BkeqKcjTWszBKO1oeJ%sZHc*dYZ{pwbF2;Indjq z$sd{Uh$VKE|8Qm<^DVKF&lr*ZGvi}_TvvQ(w@~|hQ=FvxAjLQ?iT5%DVNih$wG(Ew!b7> zG7~uorD35o)qK+NAkxiP=sI3)RyVan_gFjW1d$~%qE?`A@ELiUHIm2`rca)S8}(~H z_^p8MU%?x`$!s#IO>6&B1c|d2J+2XAa~jf(`9{!nPHNqvJFGnB8!CH(;Lmpem^jQAh2Jxd1Et1ly^z%Tla20 zT=lZZ_U&@oCM*zU@#g~-YS_9N+M9R8HC?X?;&82iRC z_X*rx3QIpVW^0+=P)5uA2RlIoDakJ={thqXpiL_OWvQ3ClRlcfjN85DdUoqO!Cto^ zYA0O7w7Pdtsi}9gD^@x)(f9Zv{ls8wrFrD}yS13Na{6bE+#gfW{V}c2$=h*BikmFuDxBbjPj7-=STbba z)1v#wKa&u;_QHtxwaNiWIjcHm5!41Ol*@1U1iFx!r)JJ&|7+uKF7rGq`H?|TIsrns z1GK=M6ao?#qns$&$JRE32VR)l3&q;%lxPPI^G@?C+Ak`8$=z3-`S4zmT-+i~70X2b zh}K!o>C}lvVn^EV2M##<=}eWRtFS$L-n&qv?8fEF`oQ}Xc{9nj%aH!Rx@C|OwWTIj z{ge`RMKn?OAJ~1C6D};W!Ks<$=D|mQUBT&$C;2ZV6zEFRU4qpUpC42lgo}7mH8J~o zwV(RM`GJ2Ev(MtCtmA_By_i4$3m9GeYtWe!`tslAkV z1)jcDp??y{k-4?$^7EvRqla3Ze)@F>DMRer$yH(YZkL^ z%6Et{yAjY6hv!2fAO9);#Ew@4>;_surz{G0$z~Vyj=3IN6iDYWkLA)2W%l>*kPUC3 z+EvxQLfT9W5^V8a`==usN{KIuKQXgqbawuhEF>r5!hZKL?9Z3!=o?R1Ub{WA2D54h z_9K7*tcUXcYR1=!+vPgHNH|$_Zfi57nPdn)(gQUXYZOuJr{J9E7F)0(IM63xgs)Yv zd!1zHY&XlzQwbU~HBhepBh69uO>UCi2&+4m@;PTN4VP{0RPq5lGrGQewETQ>Ou1V9 zixk_lSoT;Pl4jkT=GwsMhpUysUbXL=AX%9e_lo}N#c$-F?&9mTTSIf}m<%%^ocG>z z_q^dPaHr$?V!Vv#COw8L7NF)Zh#(I--iJZWUHpcW*OqUlS}VUXG`Q3;aPJ+en|lS& zAgEq=*}6Bj34ETJlYmw=lIEeWe<99|3pB%+_!uobI>Tz^b;6$4k9DT&u~_hEFT&zx6YUf5qsx>@lqkp~$yFd{l7XhJuDGqY1`AVOOW@I4`F% zNI9KKxk7l^wXU^jdp;91>Mvi-C7^|)pIHhTU5(#>7DeHRK!2iQz6;CvZ6 zF3VE&AUx;*VPnB0{cR13%GZeZKvwK{ExnfHdHe>VwmQRj^5xip$hXL9kb&Wpa9xzK z;FqK^Gi3VtmJ@WIH>l=i9E-wzqSP2B%+fGWcCNX0zT0rXzq{J%0H!2Tdti=KstMZ6 zUPS?;eyCJS>dT#E@p59OeEZfeObs0rM^tFP&b%NMC86|7GuqrG#g#t==VfF5lBm53 zMo;V%U*3;%qcThVJ;&7%RL;~&(_XP{dB1gktsX;a$SyC+l{)^^*8KyY!>udt?pUqO zvv;&8v9pKy+Hl57aie5kFGAt0!RxvEbN@?Ofj)TJ@^1Ql5r)4YIXf<=ea4VGNEG6k zufq$Z?T6+T&h9@Fh4fiwQGTEjTL*^Kc#T`w+vvUOAlHE>)VB9j155n-!y8Xo%PvXh zvMQDlW_m$x%H&IySb3O>&Il%@hP@vckkXf;YM#kg8lSL(MzQ^kni%bOFH$L!vuSK?4t!Vk$?2kA*e(mII3=^j=TFqIR-#JMDuK(+d) z)cmPRr+1og?IjX@U;7x+!Vg|e$S`y3XUJg*upp69b|P%v|G+`WUKW-*WVG+BS^vHi_G&2QY-EbCxG=9X*7vWwR8A zcx-)_H~sQzg?L~M3zhNZ1UZFrn*zOUHZ(n&a?CmXsq8!6iz;d)#KxF{bCFMDU_hM z6k5JGW&DTqzcH^p)apS14E zK=>%#{r}l8Mi)cd9V*kAp{f>X6pyb;ogVd@Md|4~?_dUo8ifR7$w)V6U;uDv=E&qyMOoY|Nke#$|Xk-Tj5#6Sm=4YQg2a{9>K4(a&vM4*U66b2*FMj7l z3~F}ZPSf(>RKBe8tSXO69n#mR#DsbFS_SpqrSr(uinih6n|hG=5~p7*D-ypm1RYM(JKMgJrNgVB2acN{MO0!B#PvpAQ)an8zkBELU*iTGVJIa5kn4ZJ8@)@YLH z?>l<8+n?cv_Z-qr$xC+Ryp+<2z@tW;94FP^R_<{ffZ*z79R?%+{PvAwLZkz(!t)=+ z=|`K-36?VGh}tV~r<%6r@uB(4*^^?~e}X?hFMh@V_f)+k+Fci4|D$~R&@gCDULiyF zf^+L#p$UBUN{!$Q#r#^$PS?rbIj`Eq$m|erpwa36KRA|&4*33eJ1!zrB02}=Csw|@ z{~6(0HnWuZvw)Ag9ixQmc0@n-clSw*aYmp6#$QntzF>r2ur-f_Y?m4c(`dCs%m6Nh zo^)-E%ZqF9EaX!NCevvk>mhHGl74ct532|Z7)?Tm{l%LM-%q*#T%2U|1NkEMSGy-m zOu!>CVotH{$68mziL5{uGuDfn;Oz)~jeL&WL1CJrt6l8-YK@j{lK)e!3|`&crZG`m zvo#Y02iWN-LN*o%w);5?&+|SM45+cD+2JR>Rt%J8ux4<%YuO(WCRZhQ!R`K&Z@@4L zUY8VSd!L-7m-ac|_sQ-I$4$gpqXXfK-XU$uz|WuYr#tRXP$pyyd^e8iS;y zjo)H3MsLJKu>~LrRZ{-I!KnEw_M?|KPhYFrH-B&Zd8>CM zI-BIan#AotHS_%!D=pOOV(9i%6KeOO1;YxoQb!x|q zfh||n;kqmGz#Fvr^x+u24biAW&%-n_cw*rNqwZvjIJGhk_`^+u%^c3(4TVlZZj@q+ zQ{U)uWm_ca8}nLK>ciqVD_`E8%7e}3QG;&ijn94f_a8R6xllk{t++pe^LhILTlnIB zqz%K3O7q}*fpPs*O~(yN+D7zWFtau-F5uy(QUt?8lgD5q7oG|xPx>^&7=*Nxnu0s` zRB~dLi^7QvcNDk3Q0ZQr^zRH8Zj09=Yfo6ru+~gE6d=Nc!s;D_k;)-s=rdGxr%8xn z;*7~{xY+CLv+Jpv@kOxc;qYyK)C=J9)-SsOYeDP#6%Gx@ru)B)rsMeE19^A*1xZ%| z@>}dl+bSAi>seB%TF zEYpd((G%_!YSrghr`h;Z-G6nFHY9PO64xw{A@tg->aL9FH(ZzC16cnPdj2^2Cv45m#}Y=wNJ> z;lr8d*|%zmlbWXmJ;@YVWyn});o~ZIOW_VeYRwl5_T|&>2QfYzd29eP_7z10Fga`g*~k6&pX5I=bf#~=%Te;=bvfRcYK{CooBNe!V=BqaT>3CR zj6I#VR%F)Zi*ZIB{nsuZIwqbB6A3kF_^VPVV>moN=dBP>t5lp_R;WafRqsoIX0L%* zDJ?tytv$nxmE4TCqTgan8Gv{6B-AGw#QI~?QnxtKoj#Rj%qvb_3j7k^?0Q*MBjqD( z@C{uY!FRko1sBLn4c;`7&ynj z=;N^Yd#s*UwOqc2w-#Oxc-o6Of)&~F$%T<0vVug5+9ue|Eg?zgN`7?w6o zmayCtjI>rbzuXC~?S%o+p>|vm`q{Oq=ZTp4uGZwKMV0H??&2WD$sJCi)=vF(TbnDP z#(YbO-xXVZ(?J^^nbJGqTT4sm7*1Z&>tATk9c);Z+ZL$)OZ&dAsRuPWWJIUm&GPvE zo;fCy8SX`ip_Kh(gxXGowLej$h+fH1QJ97>Nl>9s64HS^6)~))oMn0mS${njqmNP| zqR2=jLp51=|IkjhwE2NHqvsrzQU5n|0L$@RYmqs7brZtaP3TeKtMAhrTv=+Zcs1f28>PJ6YL^Fk z9RLy6*k7)$7RiuPjU5Q{=?&Ia^-bs@9OmYiC)jLesMEp$0BUvcPs3s*TW*)n-w zPiCWV&QfYj1kWKtb+!LIR$)Aj*8?<9R zxvUJiV;Cx+Mj0NM^4MI~06AOBBMa-Yvm^QN2s?)dCzhs>=t}?M#WBO^ zD9x-d29l}y1A`Zp-xt2bzkL4JtFBNi|DS?EePY3EOm&0ZI~97!FRRGDU6jR{y++f= zbcq`=S2!$sY}ciNArg&q+%s_y8DU8n3HJH(yT8w_j!kw-K-$^LvIXT#gs|S$(xf64C4djcCbAS<5L#wTk zm5>7~Rx&bSJ^!8v{la9Br(X?=GAX~X)T+T_q?BY121Jn+op--*@AB z8Zl{_xzfgo4@29jm2g8`694S#>7w*PwCy#oUp}J*TW$0vNIdWp`~@rI*gpk0k%B6! z0#-xB3#tOKbYaJz*?i00JB?S{*m0^eh7TWk1Ez?90cSVORn z>4r^L`g*9O%_dcZl}l|9=U}Ivr-aV@^mU$TaYnPYlH|Mlb&R(KABH)l7V(|m6w}c2 zFY7N6qvd1dcT~4@>I+g>{Xzp7lYp{D2h&)q3}0ed$s0Z;J5VSS4ZBWw8q>|KjIixO zqck*j8(NtjhKI_bTpL4^>Ur)M6cOwmvN+7Is@*e-=wopvXxM=L=%`7dWkHts(T{U!nx_DBTi{PAr`}%d7dMin;u$AE98}tGy!)J+@?HXa3PrE{2{`Z#W z7cZCK>_BX>8Sa~i%ze@LBluhwSA1JziC+NfPH2twseA0&>G$04d8uV+^LV2`yt9@WP2DrsCiQB6GWm?`e?H@|iV zkBWR3g;Y7w9%GVvCHn1x2jS1O2b1;O){&(aPq?DgE$TqJ-84fuk$CD01`9XgIh6{3 z*9$R;OeF@{6X=`nYr6~l&q27SHw^l*6()Hdn*vI~!u3SuO67i3T}j^ykvFro=4Anr zhBI#$qG*xnjE`&DPlEAAkj+)}Q<)GE7_2qpjniq*s?2zVo)oLEzN_Od63?=9@zbJM zTXnX!AaTjI*?rw!yD!S+A~giLd-UA6Y(MR&qnR>=7QN=&_lC3jclQeO3#TKIU`-=M z#KtR7SypIdC%I_=hapF%&Q%}Mgb6MDABDMU9)JV+O;BThu6 zk3x%3!yT5*d~4U!*@gOXkmo(8$C;CmZp)> z2`)i$`o%C;j-T%g-4g?#Mr8+jLuXcaQK(o2uT;p)bbJ1Bl?HI<&Caa!4zKKlS+-YJ zevCm{&mTtlQR*2lqaDNcdMD9@mgAcil287A8*ACoH(xr6)e{YC&)l?fepM88d0+~$ zh+CAN{yU2sf(9syVVqXa@@Gv|xqq9!0g-u{hL_r8-TW2+;899j`bXA=aCqX&7ZqN5 zOMbWHkKJYU-@btI(=>+%*nL);#euXA_#%Q*=2o`N<*R^Q&0|4NGa3@x3{9*fqwMgvZ zPb;ahECtV4ZJc;tRWi}!h_l*n<)qqy%X6umV3S;CTX$@=mLUTdh%tN$&A zQT%m?UUb>+#B&i;N=s8h;)3;?GL{xwrqa~aWmsMf%dx~5fz;mUXj@F#2V2{FLT!MOWIg2pFZvTxIo1KMoV!{? zFzZmC>sZy(`=bwu_r>1nA(zM^(ecc^W*708rw^?ep>m|_k)QOW!ivhK91)9H$UXoY z;FWvgoPH!r?N3w_W?OD470N=xoTInjHw0Hi3+03pskggUd#6%)Vf7fDy+@eGA;$oD z!=^;1R|o^&a=>;mEhbf_Y}g}1m_SzddPkiz7K)Y(yLo53vdZuIi^QN;VgbXuUixxf zafONQ=3R)i6FZ-vG{!L$u>5!}F^Iy7AN$f(gu1)NZ!uEB4!1YT14PxfDwgnlc_%L1 zXc-B?ZaQPOom$fs#{#biw|y+AHFvz&kuEUgyd^aAtSISz~@vB9NFGa4J2K9AR z(5Vq=x#Z0y2N(qUq zKC`aWVdfmfxQ+1>k*eqCYRUo3#61LoxYm-|&idZS_38pKVtTA*sa06^u_oa^dKK<~ z8S-?R4cLnRcQihM!L{WR(3^>yjgL#W_;TxMsRxe7b&^=UVtZwO)9Qhe>1Wd3rVqI)vtWGAj@1; z!-t};|4!)zjx((5;>Of+w>%ZoU|E9avL@=uGhf!H@5`|S8aE3SG`&*(P?>hLUEbKF z#EMTDD7FYi;pxQM?4<{X9(_?c!H{cQ1)lRvffh6K`{w za9|Td$CS~ekNya3A3@k9f4lRGl^7@TYfN zjZDN0KAh{H!WnEyrM~gMR1^ffQ%xsQXm|iqs~@sW5Xs-1Hr&95x3nDxY})ZB-UHTw zR-@4#@XSOEm)Cf+j$NV@`q8`sJ5i8RA+8Eiv+HjvZ3eH7I5V@_z<9=>B`cpnuS;iM zT;h8*kgf43X;94U`z%LG^&Yh4%K#uLEic-iZ(YS&_>6GdK4ASDMTV`A5ej91TmWx3 zxxuhpd*HyTrZWcl>R}SBaMY~JyU0?kRgy}MiP4a~Z&G<-pe6i9t@zugL@1FK86n?-?Tti+yCw1`{FT>aF)AxmH#71mo6;!_|_oK9^7!`oe4*(n% zI+H$5R4Q(0yoDib!u~T@DssNna(f!QrYUpA@aykc=zq&4C{V>2KUO@!*Y^IqDx`nn z9DC%4WxI5gQL^VqCQkcc-EBrT=(-a#H2~l$YZZA$8fP7N=_-8sAxbYiBc?Og$hgSP z_a-icwf??XAtCw7+$Nd4gG?T5r0*+Q^g!LB=p2~d z#9c+p&{&-PS~)~~^~_32$|XD4CSl2p6Q5e5mfTX~@H24rlgwbnH~r^DBOvC^6T7L9 zA9?bdncsvI|In_vH#XZH=}6rd5Gq6gnfZPW+*Wy5M=0xgDGGeKds+Qvpi3yG7bDLW zY+K4>uC*I!J+7KG>R_*aE*lz{Zg^n#+0<;oY|{^xo%tf)GH>Z1IS-4JVDPS($q8RG zbi7&gY+nQP{?&S9$TCm!j$H`9L`zg`-VS_jVH#JO!amUl^_k9^%+g5we)J~k&6?#% zYIyawhO-&Tg3eq^v$*7aZ&N75VYyp;V%CoddsVk6CWD+zj3GWu zdM8wmQh*&oNAW)UK*5d+tIt=ZuKZbK^kXWOHC7@+{f_5jcMVVHYtCcDRHz+$(6Ccx z&9d@$oOKXk5rd1GDrlvc{T_=&AgyZ``(ksGyo{uM2f!N3Nh$O{BYdo%7>AKQic5Z7 zrT?LrV|uC-XB$N??4V(fZ+@d-Y(~Gwv$Y_4yUjqOvvTOAU|4_Sk=BRU@%a+9N3@Bv z=jg`lp)pW~L)Gx?A4ARh3dE90wkJ<=n7OS;NhrR;I+WNC(wy!g>f*ooEAQJ3+yX*0 zXRz)BB!>>2ENIunxdfAGxbKl?E$EZZvINKejrZ-Y(mKA6SZbPa%tw9h$oWF<&0)SyxwXpO7m20m zToQ9s%}$)$gMC@QOgoU#;i>*Xv5jqMPWV9ajC*^17S4$>qkuk$PlvKKfk^GXJC+J~ zq+z(*@Tc1?+*^%|AfiYt(5nIcUB%-0JWa{1j#}&azIG;2cNm>7Q|iP>mup0qg`S-5G_raJ6QNKc2!|Xddz$;{Ds-W4 zdhX3?M#eplhJD@LEYFh}hdsgZ2=-f1S__m|yXJca?rt9ujiI8GgBig7D5LgJo=J*Bc$%tOl60 zs&|dDEWDRh7vI&{s1{PE&+4k9FT~^0-5GYrn4wJoNO$?sBI((*T|T8vOM@%nP&csfrq3St%UfB4#M$EM zp!H*s3~Z2=uIx?}MfD%>?Z_!rXHgUTbgt}Mbturg88A)3yw-Xt$!~tCnE$4WqRr-v zlRw7~bC$|6eM4~!0t1zx>R66Q`#^0vRCZaG@d&>8VY>O5h2V63|BaUP1bH(C**(ji z?R@Uovd{j9@MKTnEt|iV6;kv77n<2_&tTHAz(uS*<$yQXRXi zwK!ukMJ!vc=!yl&!Rg+6e7RG#ce2x!wx?Hf!zj2%S|{gK9T)NFiu0F@agD&3fiwF=13 zOPgmaZV=M&^f#`@Fg~*%xcm47x}h*7lw28M0OFU6KKvm!E}@EZDt)O zrIAPver=C@cIPsQeppR`Q~ulQw`S1Ou$c@o1L=LO5qpz(%w9rMGaUS*x6jmg7!@tr}cu2-2) z#TBy4^EdIN%@l2WFDs4^!VO`T4Nms$xmWA9a6?MbPUOY@NqCrR+O@__JKGXeC`F*H zH7Y;ciRiLLTRgFCw0f>mb_aybFnhI;xjOq|ApPvTV=fIwZbAb&<2%13b-*}Ttx8{! zpR~7r71f5&{{=xZX6^!$FXtS>aOg9#mi|Ac&}aY0j>%cFoa=ebV#WPp*6J{FO*@s| zZE4c!I^(Y0Y$+`j<{sNrc!{I0aT`_}Shx9)?eu!pQp`BPIs8WC>}**~we3@p&ni&zH)Chde*aOxQI?J@zM?Z$_fexsFR)vPxf81rmc>Bf;ZRzqlXx z2Nd^rm20%S`(J?@z1FqKtdwpU&Biaje7ma~yw;NbFmgi(toois&F^Q;ZYF@WI zdO`G%_l!t)rl+fU$ce#? zox^B9gBXk}Rt)~F-?igcI5)< zrdi`5NmaS-U%C$<0^(fw#i8;zK^HW6D8AQfd1ES}6IhOcP%B&_vBKh1tIxtupzet= zF%}&?2WAEjW1+cu(ZoqTws9=IL~-it%7uWxKmOc=R(^luBD}O+S9w|xCem5wSA?pl z$fYc&qXD!>?;#fXN%=1EBhDC4{Q#<0HZxf~$rOYmVIzi7$+y#uq0dg!Yt%H7JsmT@ z18hBB>+_or6Cfp!i4UvEaW&uM`tMLvv&Vo}p)b5vaHa~5Mfs=467|6mP-^}YOfC1m zrFSJv+ILiuTsbkD?ZE#WZuF>UuChSac6!zjDXGlf(eFxj81RtSH<)x05!_vXp0kNb z9fe9WL%vLTOeT3XScP;&opZ$xnzDCJc8v-uZ>rR6DeJe3i;CI-kU9G=n#v8-mZCaNIxSM<88YQ&rj|1i%*`NT<;Y`) zo@GsX?jzrIfpxpapFZu`sp`Mq{#a<LyqN%$r{J>Qi1g#@p-@-SAOp|}AUICbvbz&DPG&)K=TW{giLj`@WfD19 zX{->dcfx&PD-r{ktuQIe50F2x68glPtoPK=GLs49&TaVwKaI6>ewRHwmu87kG%ar1 zliElPpdgEU-G!j!YN`hoD9h2dz8&q0Jrl>22vT7;PW>4vWcFsVod`AJp(Q3n*?|ck z;;isjqYQPd3Ey>!bwV)7nlwF%qL4*PA9E;(tIv50ADT}S*t8;_3qj=zCds5PMAm=^sQV?SRUZ9|DAD!|?t-I$;@#`ti3uo9&y z%usii7~FfZoj4;d(f+}tJkXTSeD2L(;GfXRPxr4ct0RLY)Hiv_cA}jBWPwb-soG&s z@r#u>EFEvmO6%8Z?AtEAr!i2gZokYqA4!CGiRyg^b_LhrD)WH|l46_f3L+&lh;ZAQ zRqS}CI95BlfzqC($YeiBRR8V|w5-J2CRvLsa>bTBYn6lzCsN&X8b`Vg(|p8ZtP=Vi z-WCqP()&o>KvH(Ppt`^1NE*qh1D~1^s=`7RLpiZTb+jk^ODHQr0UaW1_z2+(_>7j| z>&8$+p-*n)Rfy#p=RW~4ujr|3BKb(CVINFbUa67NmPYSh+iA|&g@Q}U<`t?%QvDa2 zv0R3`x0gw?(<0%lL2vuL8b^xVmGS0B3E|yTbtfkbiRPv^BV{V~2F)${R6yL&<*J%Q z7NAmNod49v^^UMSlGSU~*FHUJCgd!O30KvNJoxJc!ybU``GTQ%Iv-Sn9BWoH|OhGS1lsCs(u8p!Cv#xcnQk>LHhsP5(uGbdHzrYEa^Kuw2Hx~ zwK{EtAj#CW2yz!A?2I6C9Mp2D=yS#XPdEl{z7oedRQ(jCaFsQ&Wiw}~9{BWbM#;Ta z5ADk7P0o`y_i7@(WiTzE;jh+<-rDDv=UI zSoAYTc*3{+-8z zl@{3Win6*xk^j7-N!PTjqhjK;{Phaaxo}^TEU30g{nu~qkuAm2Wyi@0#^U#z+I;iZ zhHgm9Ixw&ESzht%-Aa9T?oW&`HO?uFlu8n2cCL!Zx_aScVs*8ZmArHce3OfQ&5QDE z-R?ks&4F6;pvAutHC(#!tG^}?R?qudIQnwgR<-cxpSYMa44INdm-%rQx!ieK%g=dc zdmv^8PkqQq{9jhk9q36gMK3x8cZ(|%?uNUDX`(2>11|39G_jyvlOTgT|0W03c^zt( z_Nou80T z8N`T$zg{kKG&Q5#{Y)FDIc?8%e9Lv6KD4gv$e&RUKx;Eh>N0NUle=M{ zktGgsoMt1^)4>yN;AER@iWH*czZy@dc;C*KcwK#LC26!ejdW`G+}JxZS?nCTYW_(f z5y0`bK}{%ZYti(ZW#n{k#5CE~_jyW9&~U}Sbkz``!9j@Ktu6cwA67THqgn=OvUo@{ zxUb2Uq{(J-H2$kkNWbqw7&c4$vE0poo_>Jf>}j8&1{1sm8LVXRAKy`25kH@`7;Y&n zWMM;$rL|gBrSyMSzYwCQ?Lo6l%Z+qUHF!4E=zB=i@i&4>}mv%$gt% z9tdladv(Gm0!Ht9KN_!d$&Bb4|9Cf7blg1U{pt^n!7JL-VuP{S-GQgjKcI3V#dg-` zQc8KKkFoMK{50hb5{Aqf_XlN5hS4?#{{v>fhohdTj)lnZ`j04FbdS>`aWb%C;>Z}5dzK&vTG42EeLf~*(K6WvEVO08tm zh#*C4&Pqv|g||M&2}oCe3@V^eSX38=DKIxIQ`Ru!{mQ8Uyg^;WS+@D+s(5uhHB)RWDUtGE+%*#iR<(ws<=O>6Q`6oN4_;6e2uS)y{70G4hLa zZ_c}f{ba2bbvkiSVrvj!eDF=C(2FTtiPaE1ELItwmlPiM#ahji9om*r$5!pBI2}!s zgFZJ;+#7%-UkJ2BKKC@R*|>z?$4Ru z_D*3CuBFC|_wXnFcGLY9<@E~Kq)NRN$pb>;5Oytj`UH+z5qY=Ji&>>E9AsCwPXFAK zwurv+a2xk5?2sRnIRA1|nySP}^TC6C&}c`haqvd?2Z4yW<{6=^Ffs%D3bz5Uv@4ut z<QW?8Ufkb9MbKl-V6aPg zRt2fFLC>x--H>8NYNrzv(~@B_;MI>4TPe?gmgbRA5n;&JI5wBNE!(ubf32W$iDu@PNeo@BP`j97Eqef?-ul2H<^!XY zmubTW-YxTBj7R<@pNT-_%iwV7$v5!Li{!wm>iOFVHqZVv?oJFe9gdqZCi~160JWu= zjQ%vb*p)BQQ{bD<>+~y@B1xY-EJU(v%nBKi5xxy@o8mxI5(NYkTp6+EjA(}!ns1$H;d#<+ACpu8wui|t({)J`>| zfEv>|B zas81u^bhzri0N)=y>KB11OqFFssMA@{bv!PFYPc5>3=#a^T$-Vvq3nalQhIx-vKV@ zOmS&9pjw6fMz+UwH{Kix4D$sZT2sbZMiZ`!_BKOWwpHzl|i(?0!zMzIw= zWtlM~?^KM9xKkB8Ib8TR!P=43Mh6lPB`Dv$A~?op8hn|Yz^-d5Hkyhr_fX6!@V>gG zjN51rQqM>4+GI)f3~LYv(EtNMH$1U?C#zqF@4T6@A7Px-8%Kour|}at5?iJCn2oAo z>|G*OYvf2vLqO0f$p{_ED$(-JyJ5L&1}7TB^a?jjT@t|P2sEL4T%}dI2hHkNsP>Y! zBDcaG8cj_T-YPo*dDLK_&C6=^)bgFc=;+x_?{=CUX;0pngY!zEt9%geOYeLM_BVw6 zflyg}e@=xXV++T3=3P&b-62aRq1&9h({Jjd*(46e2IP;uF}}y1kkZAcV6n~Pv7?(b z1(?J$!?skLcVKl0vIU7kdYFnvU2Dz=Iv$|$jOrTD46?7*(2~Z?m}dR#(~Y&CRjHS! z>&@l%eF^$IdhD|X47YH7X%LxJYlZZ>RLfeck}lAX)*9_XX_hGi_0zpWui^4unUOmn z#f``5ng&LNYBS#K%3u$BUkcfw%1nR-X;{CCKkoLxP&nKJYAS2b1rbBskwY3YQY_mi zl>nK+nJ8)Kv@>Lmoc&KZ-HxBG8yZe41d#+=QuL>LI1x-)x#jb{<#sXM3!vmP>jZvx zkNe(~gY3I_nb+Y~LwHoC__$ok+m6S-3Wk%#lxYjSGP*@}&TzcPPJQPQ)AVPUvs%lg zE!WyLgSgX5$^ zjcbI3gX6@Wh$=36v3^hoN3xpTj!kb^YHtJzql^e z&F6i;U$b=>>&lkAF<~*5?0Ui)-Nwg|&ft)usnu^bdp zUudFoqCXO_{gyhj@s-%g<_B_m^b8s(!+yarF0S)f;w?X|Z>UMYoX17J3mvb%1|$C zH!YFRpZyk_Vt4*xoaxLI(m9CIUmoZmCh3^!rq}bAKdqp>e5f(-9Y1kvW_B+2pL@?B_ zrcBmCYks)3;5J0EQ`b)M`~qm-dbKkJJ{mU8g~xdXwZI|cM!#3_P7@mJ3=;HmICC!m=b}{1<7~}pe$U-~5l=6xkl0Wt5=!%WD?gUI zk1LK1R(ce!ZJo?qt$%YPTar|siXq?Ama`t+hlN-=8NC^^Z=YaYY2htcwnH@p?Z6L0 zHO2X3?0oe9I^x#xm&-@rk*@uJZ+ralu1-vE$K087sqs&CKhCq*W8#DMUd$r- zkzc~~T{W`oJH$QjzbGF0V+j{qlw#H|B8@6T1}3bHXH9W>vyA-v#oF ziS@x4SdTP}Q_r2Tjv9Iy|5B<>ap@hW8ePVf+$k4_WRv{Tdv)eGZ&oHE^)NRYejqa7 zy>-zSSD*0{a9Z_`C>LaFHVU${N?}(T8_#JJ>IzGfFIbG%MmLs%HPB((g+RX1IxjgF zkxN#tn=+A`>r~yp*HjG3{j%pW+(|k?p8&m-eSy89O6_Oe-560JN$;*Nq8GJax8P2r zsPi@*8B^L;8==;0s!fqO_*JFo77?fWNeQy-0}kp-_eSv;Kkg+|bjS z`{a@BWeo(&pu4_pf-2=ze%x}kS=b{K2@vcFc_xJzZYgyE9q- z?!?Lnm>3BXC2}wuLy&+ZhHKY+>?)u#GQi(*jN#=0=6$UGzRR7Pt!wDa%Vwm zvHqM89_X_5^d-#usjKxr!@~X5IoUivxr^LsT_*v;a5++Hi6)U=Y*I9KR@_-(lWcUb z<5TNgV-V6F{O2DkG>FasELILNUz0DSYUW|~NIzveI0_0Wy!n8H%2nQ)*Zps2;|6riq``We$_mf=9YUh{`Mf;aJ+N@91XcPHTs~ zyrMC^8g0vl0!?8h=He!Vp4ZIVHgRg9a^tkNTaANvru%CB3|7~WPgZ-r&%W|U?zEX~ z(AD<;&1_6KIznteNSn*f_UMk~5kwe8SgbIlfMKCa_*135SZV*m5R$Yw!W2^C#(Q6N zwy`H_d*pr1Af7yGFFFXR`DAkkyC4-IxUKlHx+~592K5@=4L#JF&5xK+CK_gB_;N4} zbldV(>s8+-P%2JfC%%a2>06tSo3G-cwOG0kHvj7e z*45&Q#YdSG;f4-^A&eOi#Z{SnGn<1FE;NGLM+i5}5{}a{X?ecdf1PIqWKuF;|8lwm zWQl!loKgF=57Tank%;!h#O(b|U?}eMZi(&y3EoeN3`V+aJ))=|Z?Ci?H(GA2N9xa@ zQpIG>dbk6{VmL$nTf~Rg5{0Z<>w0@d%dBp5%hCZ+`d%I0o#ljM9Ot{+c6~JbNyD$w zQp1bMkA#~Ya4MBW{7(gQal6#fNRPY^%q7oMfL-J$XKV&2X6lLHeG4V zKV=k>;fkZ5$l^{s+THbOJKZp;oJQJYFAtSfZf;snNH_tG6|8e6w`Z95@PLzSL~$)$?OsV)3jwZ}Vk7te~7 zjG(_Qj|AGh{&4D^45r7bsGMUGsu%y@t5QEOM5|2bOSBJdg&AYu?0Eux3Oy*8_Oo<> zLOdQBixakJSs6ZAah#oBd5~o{g%*_t2b@|^dOf3Nz@M{2F@#-_PvxBrg_nkIQ#qNt z#vUs&L)z5^E(HIgk4TNVpr^)iygy6Wx@=a5VI)X4q{+dtO3ZQOrYy2g@1o zV2HJzhOoMEd-!e6R^BA_rU6J~5A^R1Ijfjo-ko11M-p5x;&(2c!4_04O!6^|Em zl0%{Myv~|{J;?0Usjc8d$Xn6GJO7}MA(U{txAOz{baW5HZ(2j$&!pbdULJ-c4?%S2 zTqksU zNB&0KP59IppIEGGM~|2x>G;*D#OR1ubIqgB$+BNlT;+;sBosA zskK{8eXL>I*d0jrfM!Lv#b`HHZ`=L*(RsnINL0kQNi8||3iFQGH{vKjYskB{vuplc zVMKufr+s-p@?O*m&W6|H{orYYU%`)}>C3r|+qU)R_1XQHP8u$)XcQ;)*&ddNl#G zslbbk`!lS~i=P!o-z6i&GB_yp=Kiy(!p7J+!l9bn*>k+N7VsVsl6?FXNz;|x(;bed zUIYf%o_oW>Mv7Ed9q%v-FPo1c#GPw8Y}D~}lKDlYx^-!58{%Clr2;zr61FxMvax$AUa_fhjTrwZ#<9-x3oBn6f{}e1Z zJTgeJFs}LWRjUg#*F0$HL;p1@A{#*1$(LyBpM1!&K?RP%)mu6@k567|ybN6f@ z_V&n*CZinNKS7)>2P8G%CK}xbO1I|=%NP?EJgi>Ra`z=32tz9>F?dq!SLw{ghdRD! zZ^P;*Zr1zQWgnV^cuaMciEuq}-}Wu~Gx-@$V&;ET+*!)unco7f6=@d~tu*to|0WQI z$2m@l0MKYzHXZflf_^4qeSaeI)geVahQ zS-y>Q%t#>px)v~Gh`_QKI15i`^&XnD<)eR0ylH#tDw})Rho6O3<|lIZW(f69JtT^t z4Rxs%wCkUC!<2z>+;NyHGzY?5JaR*KTjJ)uFSn2T?xm-(bd%T-m9a%5hd0dFqzIwK zcnqenQ(J5#-5T+&+`XY?E$UJp67M{&(wSo;RSS2!8u%jT_tonR^yo!+m%kQ31= zQ4+A&Ip%c=Qa>V$#Z*3_{KDc4&b%9wh{7Jn2!9w{s1}>aZ(9@NY%!!9V39)&Zvv-b zm`5k%Yfhl|@68_~#82;M%PHrba|WrTkt3><7(Fwmub5*;P})-261cqgy1EiZ54M6( zSD#1`!Ark{F_z%~`)^w{w^3bNCtxy3kgO#CfB0OYB!|U$q?W$${iuQmn0r&)W8gwnQUjLcM z9T;xL^_@2N+%>Mb_$XAP{j=Frc`V)JlZ|}_|EMRT9D%ZqEHUkU+hekY493OaZr;B^ zHCQS#6Fb7bUk-7W&`6Sb*0_AZZ}gcfZ_0^nB{DC6gahgUJ5>qi;Tzcj1(4Cpbrc{7 zA`nsz>g?S}RV2A`l`weA+p}7#P#xRgaDF!{Y5PS$OGLN|-;d@ga}gsvkeB261gWoa z%6WlbM59&@VAu4Kmi?mz`Q8_{nmSdA5%``_dWa0GO+Oz2$LfzLy*y zKbKBudnsmXB2ZJzQoE!J&pdGvA#@cGNBd;nt4W+L+-VDz6KL$z8KB{(-wKzisAVwl zoLZ#^*G7g{h@GQ=hVO)AcZ{xQP?}FWdC|TG6hN|QTsGjjM|mF*SPvtD^x=Ryrb@G6 z#AV9NsS9WFj+|xckn>+k6{4N5;IUc?`5x+B-55{4)6k_opdpxcWstfoQF3%zF|oPw z`*(SHt8rPtiKiJP?6qBa{uA2E5g?B(C=GGf&((h0!Vq~;Q86)^DD&SfM9PHs&b1V= z$-hbOr}XFVPYZYMeQ5* z%4!k&L2LV4_@ANPnccQVc#J7&FDIq^+c150rrpBy7y$Rxx1Ds?`t+3UP5f0d;Jc|R zVV|39kUbwW_Pm&;T3NdnpNBtKOMc ziE1s9(|dFiXYFm!47zdXqMb2bH>R9C?^E9`i`bz@Xga3!dSLZk@5fvdYY7bbfp}n1 z{@bG_>;kK{Sv^}O`Y773UgY%DujR#Ex_6Q1GP|j|wTtKwb^hV*dgI6c8Dw6wc_be+ zD%udYX0PplSMPiwIr-az#LMGKGKn{7{(Y2V{Eyw8MI^QNnl53X`)NWHyI;Cskh+q! zQ&d9A&bW0A-O;EQAFuXjJZ5EI52mp{_RPdVz!6#Y^P{^U;z5zUV~M~FHn~k9!`*6b zqzD^ve#6&+;(U@U?!F)x7f1_vp66}Z|3tq+)*&`77yFtc!pChw`%shxeF7%m%#ax3 zM-;rTgX9HuG^vkS``?C>hv)ht@i%i`L)QchuJq_vm;qpBY_&_1=$h zk**#-8<2-AVE2o(GkP6U4c!(K@ozadmhV2;-}6I()>~iw3NfRS-)yd{(BRv|gUPRhE9gp6!Nh;XQTbD>h;sE#a@d#{;%-v60!xtX54Oex zkyr-+InKqi!#YPMWZr8IK+-X$$W`5#;QUIpT{?=Tr3v*J>A48_$}+YiQQHy2bG>D0 z;lSc+k5#vza496=q^4nh#JIj`w&M1cWX+F%$1=mS*5Qfukz{R_QKa@{`D*1Zz&>V8OktM@o-!&JTJnG* z6xiu7ElxS~iGe$IZeHi_t_9EAcBF+2cY@#3pB{kz`EoU|g@lJv(Vi;{Y_(wyl;hfB zRA1Q*XKx{%g!J^^Lxnx{A#l6rXMaYDCm-XrkbXzzeR2#Sm9>8dfLNxz32#}hC44s$ zJPRHy@~&=vnT)sj3j)^L@f{^Kp92(JFP?N>-ggjmdttvuBQ;+rpMNB*`Uk~5JQ}`Y zDxHfJ@5i;0COj_{cWBP>PYAr2?7d&P_bwnN;nIUa$WMRFC4rU-1IL~E^mAJ8=y>+v zxg(w@ye}nTXB*RIR@HjMlAxJ_*Z4%x%dH)f zdOQ2LzK32N?GKYEpMkAosi$c-vbnGr+L(QyU)HpEf<4uglI7S2p$cnXPrNyRQQ8hN z5z$<(eEQMZti!M~Bzq~`WSkn~ZTl#X^x-8+ffFInW!yydmtGu74Ek;|Bx**8)Q>oVF09*}!VUA89mLVgKdv4M{~ zjTzYcdO0t>p}uJC5sr|>R?v;`LMPduiGy^-TSA3Spb31@zMO^?lys*@sm2YZp+=vz z_Fz<5G=4jsf2W^H+r+!xt5Fyo8&pn1bIt~NAc1n(l-6kQ!7p96rjT6b2vPo(66(y zqN7hVp$Uq;n6nr|)8k@f&TB8C$bxUn;?%1FYt!r-P}0#w$Wo9VXYw0uAtqU|bFZQL zpT-1a5?S2mhi;AQ;D5Yi-hK6L-=zC)^c;E_ikKm~FTS{KV}8m)JtI)1Lg8Ir1}d;n=@*@&s^WVM zn_b@>eCXOP$wB>E+$`Fmy$NYL4JdD2dpI?}1426nwAh6VMu7#T1$idVYVAb^^*3iPl-Q7IE* zjn}}w9&Y(}Qy0;LAroq(xl1{2FO(z1Q5micYFmv1XJa@rpmvdn_2gNu_CXRODhfA&q zl!LkImvUO&?La)he)17*t{ClSqvPT&!Qt;5G zu1?&wLp(I6&AGXGc&mK0*Ny48=)qCzElgcS+3u0L%{!@M$|NeXMF{w7)z^+&7Q!h+ zhG!~a$`Tu+ZDCO{Wg3<+Yb4>h!i+VHr(1yCFRURiF?sDVN0IpSAhx1&5!Jb5zZHXRb;6_Ux>VTdz^7@8GpB8>P-WsuO z*DIfF^Bk{yQ>ei2segEJ7q-g(AJX8=@rQ#MlS@~%;op8*Nye<)037R+-^k=`zHm35 z==o&m%6ZA^wG)9(uC36I4}PuV&(6_a=$G#W4RlDgTcnhEkB+{>vdwz z=ht-jt=mqlQUG(qjoNy$f^f6=BVq>WU4`fy&oV0j9O$Hwxr`C$!*sg;b&}XrdOYDD znCgUgk(eLzHFMPDu*M+S{EZ|Ht8~%&h~@<0Zc~~5^M~~S+9@7fCkEFBR7EpAm&LEt zgTv8zvg9>~u3Y+sGxb{7BA!oGvGIMfMDj@usMqUs9onKN+r#&LgzVaM`}p&IM5PVj z`TINP9B96J?k`efJUv=ql$FJ$CViC#L0Q1T`zy1Q&km?*`>%I%zF!aKaVBrC*MV7& zmkf<+xLA_H!@!SB$YCQ{6I1V#A5OX*&f)t_in@L>2KQle4vFfw;E|vvQ0+&1yqD+f ze0FA9hQkfsz^XN|G~oEl%rPD2NjK_wy@<^j{FKOzTskC!AdL$wr6@j;T(gr7_T1=< zILUN7n%7c;3pY~}`*MwHpSjme_lY}$_=erNJ*dyj=rgLIN!|v@00oYEs|Mz&Igp@$UhB0^>w7w&&UIq?~jqaqUf@{j+Z`6FQ))CixCrw)I5C`UOl4zK4)iHWl zls3C&+Xeo!P~jE6Lgqh1Vj(iZYzl=ecpDn=Bm_mp4Jq7037Y>(OV?H=+m(kZYa^DJP16*xllLQ6u=W2c88?82q)7nff@FD3tfa3hFYRJ@mtwzT1uzi-V6}e*8f&Wb zH*We(eaKWlG3B!W$UjugL)LHt@1N$`s%MhVUVL7l-S0##w-hd%83jVa&x z2awLg&IkA7F5HF)c=FG8HR)c}3Q^#u$g8+K;|0OfANbbvYt8PnZnIbOZV1p;S=$*b zQ?Hx2^8W#H5@LLNpab><@F?_>p-NAA2dSm}q)t;4;!$COzhXb`!`>;gBc(0g$KV}V zy851_8a&^)8MlXVLBUr0Pe>@VQ`?xhv*Bx_d6{|(0vtpsf5E!<#(vK8ZgTnoa$H$G zCzA=iod|di_>`zYubUiJo~OFyUVWC2)|N2k+UYF00O~H8@7W0*v83ut^%S$Mvm0N_ z3>idehj_Y;DPr0@|B?g{hP$Sgx4A_T5m5=yum`^|La-FqTVWCO&oz?S?zT2jp)SANKGg%gXLrW;Sh@bwwEgt5vh8;T0&(ACDVyY_zHG`- z3VoK;Jb6-N{B2}NvLL}QBsB1?eX|WCe4vkkO3^uP5hiPD;u%N>D`SGA`=hiH#!Xiu z_l*`LyslJ|2b+kIBEjMAqX9Xc-$0#B*K(Me$#Jq(3DS^v82v*Vdm%HZF%(9Mf_(P( zd}btbeCjDcOu$xvalE2YZ1HWU?@Om98ZGCZpWa(t<5^PSkljL#e;FI^8uTTa8SG_Q z94!(l`a#>sozorEaqrYUUsE^o8AH~0c45;)F5@qKLb?qj7?x;*AWf0*s>Pb5fGLX> zc}e0<0m0P#;XLI=2UGcV{Y3ED+HGdgSd=+-HQ{C~G`&11JJ>$@{)dUqQ8_Jeb}_qp z3%Ei&QWfI-%5+OV6$%#f7WXJOveC1a5UmBgx?UmwjZvZZTqWid-Z{hOrvCZ78w7uJ z>sJE>Nw++n$1j|``1B3~LbtaDi8e&*ZNV^njka}X5dN!A%;wT4GL0%(y?kOlu!(O~ z5o&fgY5PHmmqlwo|FR<(XIz;YNNW1;ob?&LN|>BEjY9ZS?}`WEceawG+atr(FJFIh z#yUm#oTOH|}_237;O@@bS zKhnJGlo3)C-t+(Xif{k#?!x_BF!!=oHGeig!-O6jIcdr2od6|b^F6S=b zxvz)n;!T!x$3K9DEeCMPp7*Ft#s9160sCK5f`&6%9!v@l@uS8?}1mX&}i}($PbF8t*L5#3YV?^y#<>_ohh{p&ZQa zt!vv#u4fCnnVIq~(5SrPQTd_0a>7&DkKQd(l{M6fuqFfJX3%?*+3y}Zo4_kX5Z}Zg z4UQfk5XPo>>s;Mw0%+nz*(GlCR}=Xc@#y_jemkjCKAQx*Vy#9$hAb01dY3SazR}y% z-=CL$ibpia3A6|0_BP?%yGEhlRki{LPj&Fzs+SzRLa_FJ1)GRp(x-OSpq07XFhW*pEL)4jY#EE<|L6TMsIOTqFsZ;1r8ky5 zn7T^nRb;=i@_fgoc{=9wmy7LDOzakA=)K(npSB2$^0V&7Ti-Z*;9H#nCN1#t@Mb)> zsWPi5>kP7<|1ndiTeYg8=m(sJa5!a(MYq>=O*x^$N?q)iL{v)E?1y@SU7jrst;mv{ zu6Ft@C~wfrdlhLFa{&Y0+U$TjFw6*0<|MJH593Wp_)GvW!ZaHm*i*=34qWvO9Y|Kz z2AJw|Lf>h{GZ(}2m^KlSK0A!}S>TQSFgP#UgW;}t{%4Jth|FH_;sC}gJ;Q6kL-~t% z!X%XTn7~rm2)nvBt>vs!+BWwE!)?!*KQpp^GwYHpIsf#8w^4ehX zd>j3OW7U&wz;BKLx>^CQQ{Bt5+YtaCUI=<0L6AaAvT2=;lw6$WD~2k#^)=wZo{xQ9 z@s5eYoY;I_F(C=yby%8$(kp=GASV-WGXdU6$&{fS1zwazuMMTZ$~?4}3NkM6wq%f9m4vC4r++TN{O=~WGL zFNk8(8~dhNLPtZ#N{orMwNAFeG3D>|PT~zSRCM@}A1e6i&fQ!{wdLeN!9Bg-8F5=C{UA- zmfPaytDJ9Qql}DMf@!_Gy@`ug_mJx%h;(uM15@{-rjTK`!Dqhikl zhp%RDkSn)!{q@EeQOToqKRWUXI+e6(7at=x8@JwHVUHkpibwDLH@D_Iz4X!Dj%91j z`wh0KoWiY@>y@WrUT16Xfn@6?y-1P9`V<<+i_E$x{P+ha?a18`_iqVw=|o&#Uem761VSq&zMp_T zFuzPcz3}%{|0BU%@wj}u(W&<~0880_$&y;5=3x0fR#UjegsQ+msO->HZw1*V8^(XD zMuz{wnsgbxnFdCc#?8Xk2DVDGsJ_{p;rQvMRmoB9-8P20Ns>~t$T=W{?fr~r zyg+F#rV8bXApFwg!PX7u#Q2&|NvI({SkON(oQBUPHUBHD0`Ycf{1_a!X+!!iJQkJ- zR;i}xL-|s6n#N*fiMS*ZLYYLE~))nFD<%9l~9r`HXW5sirR!ODTDd3=UJ1DZ;{ava&Y{7DJynv5hSwRRC-{a4@-)R z(hX0P&F({r=%!-Wt?mVZ7%x05bGFKf){E z;&mi~23W`aLUD~Ze<|WBy?m2_mDi@C1{Z?Ouql-lI_&?)C)u$hT&D`PH+lU@+|%nG zt{EUh9N$oNi`2Fhx133FPkr};!gZF+{jXeeW-5OXNLTOmE1!R(43)MhV8*-^CASTG z^UH+?CWU*;?sV$v-uOjoI>R$xN??PYh6$l?XC@lQTYVl-t8HYmqivnRok#voQjwQI z#+SPdQzf+Zyv_i<$bYO??G8k9%EC7){QNCmQ{w8`gx!z?o&`E}RjcZx4#J$&OCqz3 zWICk-kQH;ENWFmlFA8`s4yui=b$|nQ5T{nWihd#BwDQ=dQz3}lQ9A7Y-f)a|d*vU! zD-QCj3lic&f-D4K-{>t`zmZ1rJgJoD+3EB`maa@qW0H3U9!W#bwO zBb6r(O>@?di3fTp-ic&iG2UPVK2Z7UAY*Z!PaiJD$4usEHSjcB)00CIr#vbnYKk%G zgsI|yQNl^Dd)`A+XXk^v1qxrC2_G_!VyR^ftfc?Vdt8%=XlpJJO$HX)in0O(u*jaE z&rA%E>9e@5@4)axCHsOtqVnB&)=Q9vLx;lm$u;nc;!uWZI%;MF?Uh`0BoRFz21@KgAt0*VWx{=1) zPg2BeH`f>G*HZ2k4FK;>A51s>a?_Cv7odjckgLI`5m{b9ekxEo6g z+JxD`uAmwtJfB z_D=xqP3V8itqQ>VzavkVF1syGCVHK-&uTU$EZ(LttRsqki%q=$M~1qBH`Pk$cGeD= zUfB_-A+Ed`h51>3n^PHZ>=&!~RTqe+rCcQUogZsW#`~Ncz7J6bwk6 zWamrN#A-!!Jx6~LF%J-ZaF1zo-=*b>%#HkVOdWEW>0|GT5v`T+5y>QryUiB-pOW+V z%ZVp23d_)z+VJ+W-D6W6p5Arm8=Ef&cDmQ`Je;J*CS$@)>PtR&&;6~0Mxm2g;KG$6 zA=cIl)H}`_p#;U=ChwRae_y}@5xEu@5exUbZv!*?kX>qBjBnO_8;sg$1uvE*+5JBD z?TeSl>uA!+KF1C9A&F(*(e|9a@1!crsbEZZ*|B2J$>6iUk{b2OzHQ@nzDBzHz70I5 z1t&btvxT8kyw3kO=L%3hZZSnJUy~_NQ7MSTT7r5SbPWv3*%45X%mZY5Z&m}#EPKwL zlvmF;mw||8$}+F47MhzGQ-6_#2$=kL=(C_HK8H$dx7T8+AoO# zzbl*Np;7G6u=b4ga!=q+6}Mv;IFtFyGm86R_@Z1TQ|F!uU0g)Wf+~GO0*f$?%&h&R zV7||MNkw~F+B&vtwG!On`8l$vld9MTmj&3^PkSahAXvpx(g|0tPi$9$-2*CM7(0MR z{&^d6Aoz5+KSH>8M8G9E9b+wi-PXPm{NQ~b-UqO$mUQx7OY#j!6QBs$HJc?J1;;0~ zm-QU5=Xi$yP=CBwQXNyS=9%vm$olc7orUgc2NFnY=8@!J-uVBytKtUDxRIP#`enG= zVgs@q>c}P@+I`IUjGB)$-9lZo&o`;Fyn;XUE&Q6lRKdZYwy}IiPbhR8r9b!m5Qx9|g^MjVppz3n0!q3L6bRT6eiMzv>M#n`aE zT!eV3bq!y@uq~^-voO`1lD^88vP_Zx_;~U!u)D~x&?qC z3QE0}%(*hwD)v0y)FVhbjVnK=Y4FMHs}nEMlbaMQmf6138(t2aTj>M^prLi;z>ou? zQxJJiWJit`y{4KSB(`i^q|56RaEW<~4z!kMD2E^OAiHa}YMUw|-Mkrr;E2tBv~cru zCWT1Xl@bGk_dyR*P#rhxX#>s49%c;+-)d&91j|d9))Di?3@y+PFIcUuK@?tn~-c?H> zOQbiI0b_Js)}UG{wln}oB3qSO|0ZfewkmDkPW~m91@xI0i-9=R{F^X#$(R;B-E52< z-5k&gplln{-)G@sxYjv?$9I~F`l4T(G_pS!33q+pD|s99L%23E0|CE{`R9mT!AYMe zT3=M+p7EAMU+!>TmS@uz>mBL?I)hzQN(oCTagkiK!=In#5yNslouL7_MC|a5H*QTq zT~~0WbC|`l9HPNtn)lUN(HZ{v>aJAlT2&IGU$|e3?GU5xQCOgiMA*w9g6`EqEt9@Q z`x-XXWZ%DPun{my-Ky-=pQvRdxr0qQ>h;|$V8Vnng18>0VlRxN^Cx60FLw+01X48z z4@3Q-{`a3yZ-*i?K`%M-s*NR2E|k9 z?l5lQD!{i-Z2nE=8@H@*%8 z!gOY;>k1U7N|itgAJN5A59ZQ#YK2Q6b7!PJL5_<`sv8XL#nELqdrE5iROy}$sVVcG z*kmd144(FAu@V`s9r@1vSu|4D>eK{XNLlP8%xixd$coi8Rz@hM$~J5j@*^u_Sj;$1 zb76S4ztec^m!t}DxBR7rJ&$~=8BTx3Sp|#6FW=-)-M?1v>*!*muJC2r#?DNuFf^Uo?I|AMaTny*}tS~>Uj`pNr^udD2Bn_8>0 zjUyaFJOZ`Z_ZB^3m-P`*_wGk?u)F?r4y$TH(-mx@6Ob!)1)T1f0&Fk2nLzE!%XqNa zhKevL`~of;{NRk7|JjT30iKN>H(`c?PapV1OFbRIKeRn`QvQuP=!A_gnC!fa>TPh?h6201y%%z#|F9l?r+M?bm)sghJHjO|ReAT3*~wOgy-?E! z(evfyc_IBI+a#5G?#Zc+$C=ijxUYTJ2Bw&6k6AH0vJBZ@&)!|RKl_he#;Tlbu^Gu} z7~okBe`MTJURYkS^!b~1n3p&qv5RYgkz5Na0V}=y#h8zfY2<~E~4jMyZY#I%|a6_&Q z{#2d^SL4U*8B#I|(tce^bZ1J0j=M)MhgT()@)QBL-z4Is)<4;%@_^O#<3G*Cliix> zjeGEsX7bK$?A6!Q^j-74jqsU$=0WT~VdMBJDBP3ZjL`iGGXJ_E8g4+7Wyu+=1bqQ_ z3=WecWrSiI(0nN#P)9Yd9nNcOZ`m@r6ga9My~}Vfo{8V!A)sik0-zO6Y#@^i4_(4M z?}46~>-*$W)S~STh#%0PUUFlb*=v|Ek&`J?*tw3k^bj8wDi`a&BVq|tALvoL@EI*u zhU+U*qWGGbp8#Y|jc?SWj5lh`#BkR|(DN=Ofi|C{>@K?+U~#abfH&s^Iu~3zk2RPtHeUF*9KSrykx&F+xf$EFK}V`k07A4xOKj^gt+bL<7P5k z$|eXqieAN_b8@e2A(@T64dnsjjiPok>#Obw-TpJEO2g%X;1~yq%CS?9Dk*a6q92_; zqjSW%Vi6eJsysWwMNms>$WF4h2W)!^*%$frr%w*KObN znn8+bk*db^`7SSVZtF{#q*pY+JFsd>>OQQ(ZL$; z&NuV%Rq1i-FP=1KjgD(-q-_bB&Bw7OROKi>_AyBt9d!Xtf$)20lX@S-ui2Wy-FzON zfvL;zb#@lIYM&vQk|G2iEJ{6=OW}N7Q_7^mcj(BlgtkVRcRv9Bw@4#rkz2gIo#Myf zI`{3vOIuq=65?>;o_8bDS#9k{QgZ`0`w83;m2ENz(22*^=CK@v?rw3@UmkO$Ig{iX z_w$a_EQA!{3I@cNK(4ED)#t5*j(pM%BzrAOo)Y@rULm5#vqHmvy$KeR46Lgv@KAoF zsnHk4DJw3Mn;vIBNU!3iv8mQ#L@?sAn_p9Ca3jgTIIP~y zLq}_zN^q^^mUF2qH%-oikI4zkAkS2A5p(o3Uuu&m$ofj79_&oU$TUL1_c zCrn7!o_W;XulQmo=R{Gi<@DsI-MGhAa~T{y_NThKSv_?!?|3%#N#>sC{2%2Ev3kgY zmYMIzmJ?5iYCO(4AHUw6lk&+QtikprwYIBO+-QP5Q_5@_)6=F(KQW=$wB0+bEokLI zuLJY-4&@1r#al|9JW#?u?s8j>dgtTiDsT$n!xJ0jEit-o{?({c_&4Y1c`x)YO$Ac( zt+nsix9Yvl+NgfyGMqu!gZ#8?7;JIjPecAcz9il0yM7?8Wesvp``9IA#OFxVvcO0h z0q3LU786a_ieq`zb5WlVPsanloCR+sL6w&L1E2D>(XQ`?2E2z^x6I*7Ak*f^N=*u48*Np zEdnupowuH;=3_H3cg58*$Q#mY4HdLHs;w=liUXPbGt!S+lF_ z@W7;04fRpM*U&NeHBI-2zN>lizf`gD{Vg{ui`T5P!!n~#p{G7^ST#JS=n;hMBJ<(1 zL=rW7souk2WKw7M_tXZf;geu2RI7yaTH?(NH|x{l*-O*?pg_cIMeNs!53JlxMa-l{ z)8eO$1RBU_ARgRSJQi&F^^WBpA0hpw2m?GWTGfT=hs4?2+fu(|xUa8w-Vn~MGN)a% zGc2xGGSpvRB^C*S{A#)?fpj%hc5mo|gQTL?tBb5@OcG68gkd>rgxzXp6SSTuO&Ma_!TE)Rk8cuNO)Z0u={v9Tn<<{iMlZ-!>$r zWQNl*;r014^esG89ZSlZva6phpY@@y?j%1-)~d-(6vl(0 z>WVqYKBC2#aNXMfC+XYcnO^+=`2@Iml92AH&pGGw`Fwxx9>0J6^`5==x;$Uc%fJ7NG|7GU zv-zI|O(+kDrp=G6Gu3s#I4Vlp(8PUebtTAL=H#+H*Afz~f8*9gqFer%J`yA-dt+zo z|5;;RGI8>KWG}pR`wb+A!uB}khkp`GOT62;HSRf$w5v^>XOPGzv?XE!s#6!qN$%## zXZVlQV+lH)dgx=Wi+-7h+@Vc2wdlVZYcad&Z$z%&JYt?3)>d7^RCK|9qGeieUJFNU z3mPoxp%;_NOPkrfmiarI{NcK@{G*iA=j>NCBMq@8t{ zv%Rl|67nkPvja9h@?0h7SO@|Zz9aE^L0yXs<_{DM$WA7aQ{#_%h?P#1Me1df10Ld! zu=P-+sXN@%NTSY^z}K-SyyMDiqTd%o#F8C!ezGl5Jg2S7A!yf5`r2Lo_*BZ&-oFuvp)zr*(|s;8#<@h z)+sj_Uuny)xkzA6RF1Ukbbd=c-f4I1HJ_Rfs(0DuEV8LzmuK~q!=2LLeGtXgPW#KD zk@@G4Rg@v97+#S&ipaYJutZRHkzba#LQ(TEh(rZl@me&|Z5%@L)+^>pR!kjlj4wg`p(!sy*<|`zi62Su+%P(eiP~qe z?!EjKmnP4(cs#Smn6y~l?-{bQ;ECM*5E2``X#`+2&nt^G7W!xjdSN%!{-I3^>z(Xg z=*IgI_YZy~>A<=sDwc3lMf0usK8Qu@YpLjAAW1OINsOPb^MeW!pD-`KA6e2*nejUl$V4r4fN_HKN?gQ|zU);yELErjoSJS@ zm6rA}G_~#WKaYY|<~?Uadaq@Z!ckdRng-|VJJnFHmx%nv4{XL^X%(t`^pM!|XsUXI z>J?nORr^{eoo)93-xc+xb^4;qkX7YF*X)o{23~G~ ze4F&5>DemMv6N(%yHC`EKLObwCGbw|06AR%L5-aW%Br)&dTV6cjoW$RIOR+a zO`x2*2$5?JJfT+yH*eNiHg*LTtd?3ouP2nr{X8tzU-n5^F3|^#^JF>nb0V+&>tvn1 zX~^MnZ*0qz@1L|atgj#T=C|&$rf&b|m3qZa%)7@-l=tri9UZ0X*|q+{)0l@(G)pd9 zafZM5f&Sx-&hBraO9@QRTb50bRbJLcPDxZa<4Sr$7Gm%8SR= z4K&*ux$6kSp|qxvwq)Dyrhq>pgp|-CsdnJj;8wP;XC5v57D%gIJaTwo_55!&M~{F$ z&8-8V@yV9!DG^r!X9+B`o{N|YA93{!>}lxqZQq&GRu(0hGxxF0xyKF9m@9D)XB%#M z>%KhW`%*mgtZ*uF&j|B$?m^NOc!1?r`B9wAWg=;X6NVHt`fz%9yZ51fL%@92No8qNvW1+KQit%lrN&?~zTrpG`8*)xs`SK9M4RousB-`-dNy$=U(sRB7? z9!zG>w~cxNi;YQFJqeL2BG!S2X*%z~>urp?$W9p)k!4P4(k)2nF z3zheG0Mct)U#rc=dO$BP-6&%?>c!I|dpa(jzBlvWO6tfviX&xouoF^o>RoL0+nJ^* z(Tj{lqS zCfn?=i;0ei)JXXE75UqZG zyQ{;|aOiLI&&~T|Z1Q5_GaWnsHljPY0>=|i5=fb`OeZ_cB*D(6C^YSSzXl#+*5)m{ zEKgmJHuYpF+^jayQ=1jAjNdjxEnbyUuB0(^Ob>n`&-}i(g{J2%;lL;nE z`_n*NQn*xQa^t(iD&ZEK{D$5*!;C=z+ZQ9o)J0F=7Vj9LI9SG27sI9W7OJray6ZWo z;b<VVafv=7G|%$U#ki-3UrH>WPR*+rHk5+qN0gFh8pkF_xJz#{dMQvDC>O6RjY)brPk zlz+BbAh0Iw=82pINRG!F*JrT}sVl-4!SmOH$24z#V+eE~tDMWh zc|~%RgY$LMvsF!aX4D5eh?KO&0bJZP(1Segd2_RBj0*@f4wob(SCnre9-Wy9Q7ZTQ zLU~COnb_5a(~?hJh-w4qh-4MyShwyTa%bL=NssH&s<%+*eYL2Z-veUrJg)@Zr@&{SbpKHWVwqIu|cQ#Ag8SkSARp68}ij<>) znW|@W@kEN>(%_rW2AnwA6G1l8SKUKJ_a{=O9+B}PRH!Vp)a?Z1?-0;y0~A1kE;U=o z4@$;`T+(R&iO#!Y(5EFW>-b6VRF^XOyxsh1aEc}q>8h|Az@#53Df4`)yvN~(jeq)V zm}{35e&%e&y^Pacr1fv|eYT%*ZD*6Rwk-8+Uw&G~lJ$Rh@c!N_icQ|pAe}_=e%`5$ z?gwTyDK?!MxS<>E6j97S6ZXb{>)wCv|D}X) zLpXZ^lIP?NqrO3<-8ja1QZE~IwAX;m0przfo{fz4rl)=vB`RM${};y{^M}RbO!<+r zk@d7LBa&BqbVzlJ%=a@;UuVW@PCK%X(ud8vQeJZu@3v#M(8k2NC0SH$2XH}KC;_K-q-&St_J$DA|CoZ+4w9H`6t=VJ!k@==+@@C7R#;jKN9Mfh8IeIw* z3wsrXo-R6(e&{C^abFAybWTt=g4FT3h4iHe&?OmbN2 zN7{f+f$K(pkb8^YTFCK9TQWIkCMX7%R+%SV%G3HI1GzoAvXTs}(+AI)00)htptqT` z(lW}}`|2zEvk*fqOh5Z!_k%v`U3G4cX!)lXDF@U(AbpoOl!MDO65lw{)1b0;3c5yg zuAJ{T^QP=C8tC$@%bj{HD!w6n!Z)806!{=@O0My<2&`)XWr{`|(N8lM0$zHrJkQ_) z@Es_87(CQuzpqB81|0yvfV^)9H!M83Tl2PzVbV-wxukafGi6xSOW^z#l8y`)ifT}o zCN>veh3ROE`)56sb`4fbqQ0s$d(s$MpggdD^5wekvyR(;&@1h8KURiehHA{7hy;hE z{b639K(yOdGXr$5zuX|cY3q(wv`i0>N+B=no`+m+`!2NHF>{ZhTNJTa-zX9sL)(&d zE9a#liO@Z`t)oad3031H(~v>qTZNC3z|+chQ|Z)WRfOL+(fe0Uel|dz?8eq}or8}< z8QeumreFp%y$9{H6Z54NX61nvYme?*Fsr}l&zyK>`Zyq^I$KUTHNiV&tHfri*vBUv zvAnx&Xj**y-O8yYe)-z9>liIlF*$S2C>6Ys9hv7F7Dev``Bo1XZj>EUO;gT=#@+qV zo=1MDk6woQ@gIXXYiTty{Ii+(lpM^Vw7nh*%b`eX}HG8)yCh)N0&a(@;+McZyEz!R*3Vl(_z^tb6Dd<(`$=T99E} zf$BBG0@>;$fF@n?ZgBzp8&d?oz%hG3(=uG;=<7UMgB303neA+2<7euES|H(GCvt-k z{_@2m8T3Tw6L6W@l~$S)6INn7E^Id_SG}M)MzwG&8i_yI9^mm#R|^uM4t50`gQ-!7 zvBtkdU`!<$*mYz<}TVWT)d=beWvPV|Ax`U`@V7^Vg&hC#-hx>a5_}9j z(Y}9_t9SGS{r#~1d``!eVE2#Yj+P%C8_JsTHL}@zeTotTUAq^S=3#Hs8PcZ#zE4qo zrlbSg;Lk_I3E>6#Ckuw!d)S3>;$P^c!B}s=cM~xYJs>a}OJF~DR%va0e?(1^3mGfi z+XB_E%7& z+tyAfEoF^^gj?u=FTQ1eU9ak)mCw7>Syasd8z;{HspU0+U%i;2eU51CrRduD9Htw! z)P+q~Uk1i@5=bf{ANFmJk!Re>^?;l}x|jw|@QR`g&sXpOk>oq;8~XIZ=C!xbk3rur zc+FnlRr@w@c8D_?^Mm)2BM)xLmxr<6wt^<3BzOYkN+2iwZYa)9jxXb6F*sP!_BZN? z-RBJyoflB}j5HlhIX_I#G;3aCUNpgcsJLnBCp>C(CenKq+la}?tQ|GLk>z%2m~K$s z+dE$nY5Pfj{B+GhcDYc*afZvc+u{76a>NO9&{S5H3wfbf;aBNEZBZZ4s?cTXzuPp- zX)4CoJ&N{627=xEAjKc&jXBu7)tbuZlT}mcnfPBJtPDio1TdM^>+u$;Ulo%hoPAUH zmVa2zNzVRg1VCtlh@yb979BQb7wVy;%zHd<^6ZZeplwuikk$eZ8;S z-K#ML($@SdpM@Kk8gG*yWGT{Lb0X5xljMJ~jZUCjzlr=HuZn_{SB#3IiLEpXR0QUZ zr`fcF9&Vz%rufrD8%JQ^!y5IMUXYq@SqJ|`Zehq$zV;3G_L8c(4(%|+%iC(^&3CE; znUo7NCw6I&b@OHbG-u{2b?MqodfqDq=Yj4suRU-c?mZyde~}-AZgUrh{pG{7ciHxM z$;&7NB8|j6a?^F-ahJNuMZL{>t^kCil|w($xWjX-h@ENNWS3v z#2W2&QQRW61gjQD5nm9IeAleIb@=E;s^C}Vb!*TCrqK-5Xod)P+S2ittvqUwrZX>g zqpye;#tz+75gcUDhn<^nEeO%~Cz#>~vaOAEBOX}l*+Ens1Lv6DPp2jDjCPS2at4@% zf(Icf^YWIDq`=zaG)nrL^U%(dR%%Jvu@Rz23{&H2*Y6^oF6P1eJ#z1ZU3JmLJg=hA zI9Ja0^9cJ2&hr56F6zLQQ-qdJ6={^kO@TD-dxt$1A2ZBKxTt8jA?4lF%%pccxSW4*s0%sSrpz1#y6`)1-DHTjyw`aUqYJ#E(wG|O@Y z*h?m{%-YP?&wQmB5*Ca|STtc=?s>>p-h-H^ zo8BZNb`@i=Ya8`MQx8Bl+W%;HojRZ`)*$G)&sOPZ{u({-E_~<5=n-(nqSS${Iqio_ zPMO+&60Tv0KR1uP3pZJ8RQ3MQ5kYSX$MTO`lG+2ANUrgD1ERNPj-(n)-+@oZ>XK73 zaXOWu*3unusslRwY{C^>j}WL6Endt{Qgna~e3|%O1Gq&={~+_^q#knbb{E~1S%1cl zLH`BemgAnseNgLoUjOy3DBa#_7ox~wX6#buv;~B*nQ#nNRSUfGJj|Cvs?|0WMIS@k z&@uZ7y#}O?g?@@9XS;<=Mg9=gyz{Yxy8m9V z;1V)QdDt+*+ZU5fEv~WlfQh;i57923tBhIJ<#;7D4YZHZRJ!k5FP5*6MUJH{y3z^3 zRvzU+;TpxM^h@4TD)Y``V_Zhe6SC)e!UmT5kvy@zXj}*^K7Bl!RsG>fm~pV5w2m)> z7CF7f3zq)CW3|BUxIxDA)Qf^S$IvZPplwm3x$21MRK(>>8^NELxw82DEaIra zh%7p|JB^D`UBxw{FrxMWvyd6=A3p2UYnbcKo{7=;YL@DFefvqsF?8w}(z)+`Ge~EN zU!zCKQP+ILc0n%9IW~@uPMrthSZp~{8vQ=x;F(CT`>~BiBa)ZU*S;;fz3o6aQk`8H zx$zf~z`kHBw{e4-W+Cj^3w`Wf=f;f-f#R+#M(mdYRm#8{WM4UHfauHZbNNZo)um=# zEDFMUNq6^n#vXh}Gkmqj6!V9#JWclltfJGQFY2Iu5eeQQTuFV>NeA9p<$ zNCUXA4^DlPH7Gfu#L$S9Q?zV;QkJT{)j!adb4m3Dt5Vs8W;}L!yMwFT6C5<&Ld%0K zDM+AIz)xapRfU!b2-VZ!4uX>ScQzU$qN=xjx)Pvr*KkbIXalbJFdnCu%2jL8m0$1? zqaqDSD9%Q^BgWIl%NRb=6LA@r`c)g5Yx}Eba?@ftlh?7$T_9E6dxm}}(T-D1=~(e|Y5=bxdqhBHe*Od{SBXVjkf3y0AA*>c9<5->8-hD199 zq^&$*pC(w3d0W7GyoGSSFQMY>7&LEdw+r20KL8)KPi>|{4GxL>)^4eld(YK6L<9OU za9&qw+CyC~NIxsk8>+oC3tCE_y%(D7W&VjlIu*9kh(e|?Z@6l(ls9QPr{Xpx$b+MP z;H{W6m-Jr|t3|O59-K)CMR8%5%nBE^JB!}=srOVH?J0Y-Qd~Y`aMeN*A;99M-u5T0 z8l{=CJh)vxcMv+4fVWMsi&=;eB>+-v*`cYs_4^Y{di~JjYNtX(!7Z$!LvXGwD%781$yRXi-!6TdeEZ`URH`k1= zF`4ZN9htSvKOtwZh=B4;Qt{G_cHAggltGj^WjKs#1=uXox?%)DF~^m02|}-0^jeZ# z4>3~fCj`{K-HHnk?oRzAyb%>3DGDrdm8AV02y?{>2b3i~40gvjPgGL2$UI_1V4m+V z>!lZG%P4$AT=B;&FS;ZLA*%O0-gm394c0_4oHcy~|FcZ&1%nDNWUs}oUIh9OJEWF)K} z@4y@3^CCcyaVtpMLiujzxCCL!1S5q^a~Q7A^0NA zun4!#O+HHw#g7LPL#$NMwVWN~%&;~H$6|De4fzkrAjoQ9vex0DdLXm>`kY`DRE+l@ z9d`Ii7*h~0dUyCuZt@O99^HkdVKMkX^^V#8Ppql6_6*0ZWmPiAVvcZ7VHcpksza-g95+R=Ve!?(_OIuTRC zkbOwozxYE`OCQz0KTNo07mk?;-_`s!3DqRegay*GZlw9xznq#Gn7`F(XLF+|**Zb0 zrCsqS1WB8)NfYd7gK9X2$%$1C%$00>$2wCAoT1K|3g^9Njd&b)=Jd+kV=I7gYf5!s zC+4lHWp)hm=_|eB7)07pOg`Op%J?Qon#oM%@kQX)3qN1iLYnr~G96<&fFq^EMzlCH z2g|~(UxyrKac-I=+!A>>hH=i0{|OP}HxQuO0TrffR6X?4-&BbE__WoBf$D83?;_Ak z^(#dS(z3&r&zx!Y7EIVg+iwi;cNnY5tKZ|~C!F`}#s)@B!W;V{k#&7U>ERp+=xBr7 zM#iN#O&&Iwuj(@S8X={!K;3|cAJ>d<`YPl_LSJTT*Z_eqo(TFDGp@s$- z>4n~^2?w8dPK~D5Dwx!~G_x|Z*1jxrUOMNr;N$q(9|dmY zm4Od+$F(_$@58m57E1P9mN{#KJ5Ng*81f!x3Ez>>z^i8qlAR6a>*?XUY{6CjUm{ud zb(Mzc8G@jq!> z(d^B#yqgBXy*8bAlZ4(`+R;t#Q;3&o`LiLCYa?>U$Uq%FVJ~#5X+%F+j{B_cNonZw z^Nvt#hf(qosN)a5NnV{YPY@WKSky<_2RSEIIRqLnph$&)G(nBLBd#2YSN2T!;LH4M z655lIAP@euup!g`f>q7eDU}Ya@h`#B?Y|~&ik?D~G{85VvIDGCc|UQK0d2%UdR$bMQNZEjKpkB*J|kp$VP`<_ZTu2so?yFjeYCFY zwvS`_Bj)4hhNE@rf;HlG^_m9Yo1*8zhq)>+>E@~7*Yk`YE^&z%-y8G+6!dQ!agC1? z!Olu0`lXX9cS%YPcsdm+SJMq^)Mn0}* z7asQkfl+2lL4HA!yAc}yU-33aFQ6tyRXA2)ZnPWS*29@{{w{!8eWfoWbxYpQNZ`Z` zhS~6=`4Fne5x*7^rGFaA;UnjsHbsf(oQ(fd!yVfAou5MEUzT?Y#%95qWoj-!qr zoa$xvJx}zmDOVD#PppyKDnqI52}6k=ntFcZ!3>v49d#Hj^bN#+dwyGZ zUdy4U{4GAF_LTq%6vP;|R;=?B`V~-&!FSTzTRW07|G^$!)eGVz&~3NfahkUrY{Ke#gj zc=V36?duVQ=%PnfZELi%dwVJ7>{B%BBKSZc3-K!l6`UPvYQQp!FE;zbTv6eJIBjG% z;>`93KL-xl+D+c!ca;!U^4%o|P)m9Z&DshkrddB%E1(o^6;kN-&jzj%30N!8S`0;Z z)LurFBW_6w95*5$&LfNjatezSQW6vkhj4V9H}Un zAs|%qh>fYBEo(r1XtH*)w^@(H<3pP#!-Oii8souCQq7U3nS;llCZ%v339qjb;!Pn- zjPW7ibZx&$_l48Y0~I=if*lkI>no|{a22xd6T%r4Y3l4QtY>_gyO%i_hPx}Sx``>s z|Kf}&8l5RSGB<%0_@?p^fpo$WV-v0GfF_xMlT0Zj!a-og=W)SQYs;?jd}fnV&;4ww z2#KGm(!d@gFSG{?02489fniNn@iET3cG^-_4^IMABtoO@Nj`eszYOQdD&eruQFyyL z$CC3HNl6$T&*q4%kM&16vV%t6M^D%fY#h6NbjI{5wB_M@qH`w5f5JxbVMH5+E_yns1QNl-K}adJ=XD)gvVC=1*tCSJ-kjq>wCT)Ph^X#y zVhxIB1m-kDz-&+hORTo;1f#F(p;y|^T_L3k{Mw*NGSI>g zyB2-x1Vj*kb#H~t%LVlKz=Dvkq(IBphpBUJ@t?-njnw5BN+!&ca^HGr&CjNa*M)aT zEDY6ds|um@=HUo#iOCgPK2nmZd%!vOMS%*OO+Hm=-SL41EC6W-D>K8^)Hd5% z_R~siX+YrpSuyh%=(_&Hj=MGx zFxJd%z#nN!b`0fSCa;B;xw#Q;E|>ZsC^ttI$tBt^N7p6WAaT0A+FEEBydbi5E3YGK zGE&=pMIt|B{D*CwMVcogYYtk*d?6m6t?^=2{@}25>GUtphg9qIZ8!!4fivZU$ph!j z^?p8aIRF~DHQ`&m-9U+{*g<+l2`l7(yE(dQW)`i+RQ^`@IeM*-!)0 zN*;1OYI;F`Gu`VvE+F!7nkd(ck-&YpF|*z5>f>?2U-ciUkIMK|{BI6(dT0zmP|Ak6 zCZ!-s@sRJ5nmxLs#DFFTX>#z1HR=K^Y^IykmRh(n_y+58q08_eljQ_9KXu}`38&-x za!waxsz>JS&Gl<`cz2ioW!xL6*`?p#$QY`}>Hl4R88sn6ATVpByrrRdC?szb;c&FN zcnL@|vz5X2#RHe;OV3hvJi|QnFz>5Ct#Gzk=s{@Sk5}AI*#%8DGw(g-ZHm!nfO(V4 z*arV0+#jF_NNlib*BKx%iv#u$?BL*SO z=?n!FSiB(XXij$S^1;@#GAhJJzsa|z_lzj0pekflAnn1SV3L>z9z1>ZRi2sOq z%59iEhv1k*<9VF|@q0GmLy++D${-W}w($bs(9z`!{`Sv@fVXx46Q5Qd89$|($naaH zX8K9UsnF?j$+VB*dLN_jR+4NVPMu!fu99=!dny@vl&Nc=!yon%?FSocC$0ma=0?*C?6u&y{PG!@SukokW4 ze7B2P{g^>!)buhKE1-!oD(11C;R6|rp{lVq0+}YhTISPO4gGT<`Oj&epVg;K691vT zOGIuv*VvnSk2IiI|Ftw@Ipz9T*#NZO{EV+;>qZmumLWV$wgI?NGu^TYcAeEFPqIe; zE_-1RA-#J_VCo@QE@XQ2&g`-8_cjWi5HXxNh}*v3g-3V#(KRW_E}X{{_P)FFBP(DT$vf|jz*TX5V825z^(?m#gO5p`fQ&~HF;TNVV?hv2H7%S z=Z@ZZdK_9=f(ybe!3PB{i_}RPj$UOtzEpDU^j!~_7{95OrSCV=ai=da>fS#b=A5mi z^uL>p*LI-TWXI9B^{nftUE(G@q`(xYEkqWY!7@-L?BvHSPLgayiTIgNMmw@G?ozW=IOHy^^S_XHq2hcUvu8x6OU- ze?TOsa&u&1d(86_VsG6HM`Hi?=-*G9F6Uj|oXIme&5ul84Yd#< z%bly{?)`_imUA~wwU*WSFQ_lzC(8*ia`?b7KU~>ix71!vAC0a@tye{JK8734-d<~x?5a<+0vJ$}s%Ddjj>(k}|NcFJY=uCZMP3fM2AMpu z(+7SB8~dE&s`#EdF%|lP7tv`dqbJGzr&qWUsgpVLbC>!L%foYRFwqlGIcarC^r$#x z0zG+>%%Q(de~XcEtANVwKY_KNV655}5H+VZp5G#5AgY&3dDEG8+5$b(#<}R1BM+U7q9EL#pNH4}(c&nK2*qOx*iH z#V54s@=plW5THB!T=-kIJT0PG?=Ue}T3@lpBu7G})<4idYCojq+8qryOLuUZR$NE9H+Ie1nBu91sEqk(*>+$SZEYgX1!0H_da%4No7Mt9gijzB_XiuJAd*j$&41L;ew znRo|`Rr=H{G)=<(*_?YsN&yP>cbx_R?6 zq(C%pyaz}#r|=P`yG=#a$KBT>YyOR))~yICj3RbnGiFZoiMLsCt7^OVJIFyf&>;At zg5WefRc)fZqwZY@SL~7+3X6)=8OHBf5GiNM(^6-X+>J-%Q6*VSX`l~N5vPyrH7=+N zX-Ib=z{RQ!V6}{8arZS8Qk+l&@;?87aoeys$I8he@gQraOWNQPLlV?an??yJf0Obg zA=USx@rV>4xd7o`+GaJI?r6-3$J#ks{eKn%5o=}4N>Wg!1KE)#v6qMTeHty8S^a`= zXREhwFO2>6YOt?s4Wevpmt(0(p^$!n(8oc6q!>adWU@vubFW2dwtqIP!7&sn zY?mgHw_3N#YMbs=D3*$5c?p!yft)19z|8n-@J&Pyvfd&Madqk)owzpOx6Gr}303aP zAI838=%>{nZ_nK#yEmU)9o%0g6UeU^vWYDONn?KpK$oLX&`8*t~^@59t-Zj~j|86K=iVV{C1r=J%};>n^P#w-HlUYFtaF z{WOW-&anjxOO@O1W1zr!_k5FRoPQN^KAqWbrUt*~*!yZv+2QA8k@b1*HR#hlZLMc` z5_@xH5ldwOFDXlV?>KmLruN7GD6<8^6a?e&t8%z_JhS7EGx@Y1i zy?uVZ=%i_@T47l=1?$Wl)268reS9);S2;P95sI-ls#8(e1DaG z-6rBqqU#6De2XnRERmCrtj!*B+J}YEl649kK4R4&C<{FTd)H}U>RU{vaH{?|s2->> zr~SrwhlDTP+IJ~7fP5I^RYa+J-6Ohx!E<(!n+(G<+RLR=qs@P!1RG+Oht)d5_+tVxM8tAn&lyIE}~QxU5>3TBzi1$xP?$p<8;c)826M71QMu<1* z@;JIO0a~Sz5U;?-W(776{ShpTk?pHC&Ntk}fP2m0PHf{QUjwPKFA!qY=Pr+rN0bd` zDqda@4$!OPs5x0W7%S%cq8`?Am>&Es=kgJ>%hJt|w~QO^-NLz)x}(*uX&z|#D?`|o z>rexrbA>}OyGK!?`vFHwPfBuTWIc)~J^el^>|z`lz65knQ$kI3fVVmGXi`WrBo08B zC4(an-jCzJYmYA?E+pIBc|j zzdpRo3p(!^PeQD?gd#a|g=2`Q2pfe0%Gef2w@_|1+olc8GnlW5wC9J``*V|K#+B*~ zkbDiGTN`za+C+c%s#B&t61l!FT9_dDg!b!SxiQ!UxG*`&+ycLPrW0&a3sDBJ38}jmal4N)($WTHH8>@Iqq#}+@ zHYv~NM}3Yygr*XY@zD^| zA+paZckp=%qdZy{r0OPlp3+xz@FR)$rr_%l@zWPGv=1>)h7T~i z`jBVL@-yGj1od2w9bbmtTadRtZCuCC4n>APfi$J2=Iu**>n{A?6E^22NYf}@;wwWg zhC27kZuS#efI2&LCkB6UyPjNAQ6u~pG7yO)h3cgUEkv>FhFNLNyjv!&`#jo)A(01R zb`WkA$dzI+Qn$DOm9S)=d;u{us8v9Z9xJH%)n~m#cR2^N59$#`DYP4RS}?0(;^I^f z!6Qk&Dr6ep4PS5k8gjKtFUmxA(wI8>9@2{S0&kIBPwje&GeeS=_%Kd*=OIec%w$!o z1%h{ec}3dvyIEnI=C=+$VI|$zA&1-PZPSd<2K~>*5RQ5~xYZZnNaK#(;5b_G|92M| zB*JT}=Srh_(Fl)Eoe*D1=s^&|wn(O(UjsZ*`Hb@g{WS4zyGhw}g$s^(Yuxu;*jKIe zdi|sairq`C&SZLs@`U;+D5t(m^Se@;#l3{2(&~2~S5NCMG zT+7i4UA#yuP55ZdQBuLe05Lv&) z94O&_13Pr;bSSh8e|bQ8yRW$h%>eJ@f{(%%8yBq?DPk194f6Lo;KBQFfElt>c@?_H z9U3|p6k~gq3c@aAenpS4?_EspvAmpnu6~>t1!<*sf5d3Btt^lO7om|xI@*75I`los zMU`}1tCQV%#xtR@ABb@r!KGd_j9`GUoS*3Kj37ARGd_aI$ykwO9uIs|G~ZmoQVIJJ zo$iXmgyV5By1Bv91P+<=3^#hplyf0aMgZ5Lu*_%*8?P`J+E+R8G2|V1aJl`=t0~Qw z3m|;y1P6f{JR1)WRWn6@S_ac3xEnOs&Tu_yU+@6XPvph~`-#@Gsnqoc0rr^NzoHu3 zlA_A8$wJ{2>)%@`tOy*6loeD`S#0+#E4%1*vd z@Bxx88BF)fa<(5z)V=sIXT>G}_dg+ZORTA%}NLb6FUJnPmG(@9qwSlh85BJ%{v z0ERVF&UmSp_d_@#*K1VhzG18Gy$TM(rFoqP5wE-ORGCQ&&9Kp;@Lf<4z?eIALstBU z1v59}erJTt9YvqT(48=5FQ)XubnJn4bKa&Tv-j_(AuwLfy;J+Oma28J-q{5Gv>ZmJ zFAquhO7=3g}LH=>C8e4JAZ-LPECq5W+xlLRD?e z+4LAqB+)rVfL6+bp_SAn=(el#*CE<<g{ua4`5O|-5%Qahxb9pE8Q?rlo5W^_wOSA zS?Su`Xts%#qZcTHI({tE1xR!oMw(r>`5LWis^n2e$2TogKVmZ{UMwI+unkle-60n&xcowOiv1C~@LG1h@8$_(lm|5SN5pc!%VaE30 zLyNDW#CZtxp$lA8C0#S6j$*b_qnRo`Bsi@}q%G`=K<5^V=4L{jJ>huks@@(yEJRvt z=LjYP>3U9cK%ytpElx(Uof*GoB79BgKH6io*{J~|* z(OdS10@ly9_oWn`RcKIYaWB`YOyT6;oHIugj+3h?e&VE+e)=`z28-?l4vB98N4y~R z0oKoCsB81>Hb7ZP0N_$hjAbjh$=01vK`%(E^mExphPxOLU8px%po~5#Wf0*%W`Rdz zIVa#Ghb~<9K?tbDswe?zky5p4kkqOHwx`L+g*I+ySa7^3=r}#Pu)TC*NtD>E7itNr zT-&3E<*$gt53i6wY4@*C_p`wJVx`wU+e2)0Jx_il^~M+&nxdLzPSm^FWD^gzv+<`p zcwNc1^%_>NI*=%wrzFf83KA9pUP4tzHRQfU(ptScXnIYmwr2BFW-fLA{Mw~If8lBq zZP%e_M&T2>P*PnZqrB!5wm`NNhIJKpq)IKTY8Jnew)u(QcEUcSQC*9gxU>JKR^T?{ z0)uP*ub(L#=08jpZIS!-!o5MA*`YHWc+6?RYjk}Z#<6Hu7uZ}w6uFq{O2?S<{Db_c zF?6x+MaXytVgucT7zu-yAl<@7(_O9K1){%HD^L+j6NWXAo>p9F$d$FQg98dnNnyCg zLdjB7Wvl^h`(+>?IK;k)cDPP)T8=8)!E{c8j|XXg*{RJ7zTl40f&wC2`T?H!j7q$) zS*!0!xq>3k-r9Hh#v;RbiH(VNqI_cduH*c={$TQU+l{xG3bZxJh3z?$eqdaT7h1Ye zdiduZM&eDzL-?8#sexxKCU)c z$@gj5IYOf1IT{97FrU7=0}M$b?{E1pQz%L#;X7YZH(jc**!$w;}pR+#m42z;-I|H%^N3=kKzyT?QEQGCq!Gay{RV1j?SV;m!ie5#f zYSq@N(`jeset#e4{{E7e`XZ2=&*yob_x9K@fiFZm%YN-t-p~MySXI$ZS0Rp z+9|s2FMbbON3Cx(g28VekEfTTi`+n!QtDqP~~3Uuii; z^g2pyp?d}KxHQlPL${Se%BYA-I1tctXds9ifge>+yl-h6&yUH5;B(%9khZsJGqt**0xd&+gsO1! z--EM{H&Bz_3h*2{m|hPid;R%6_#x}7kh+sAEteDjiA>9SD=YV#_T~am|CqPhH5o7a zC@IvD)~j;~gI#Xb4pJ2)9e~u&Y-)I1Ji;z8Q|>OJ;xinRCHM*Gos_h-rhQwtHrJjN zcD>-*sW%5RW5@f~VTfaplVOu&i*gcW852!xWJHlQ-9tK!26rM8{!MNS=B2RSkz^M8 zT02A2cA#?p%kfhBKIqZZap2mKX%9X~k`G@{d)iT~mf#L0@l(of^Gbj0)tjK!7bVGT z+tF{raBb*nz!rF3boFYAe9i{lRH-yN2WqLf$fsZHKuFm)C( z5~t=%xLavs9_LY|7Yl#dC;<|&t?{e(tFY4JAA}QnVAM|GagO?Zm0=ERq=4Br0 zP4&`F*TaFv{lI9y=D&i62*SMMrFv-?tI40e4O%G^|9oXh@ zt&fkzyJjrH8iwTg(r5F*%b4=!3l5o)`)HO8LLVg9Irg}lvQ(|3JiRyM`aEVMeJ9u_ z>h7Wh0si*R%tD<@erR5PeI==30;0#iKK~(HKZB(Cnq3NhE*rYT>WU2h_{f{QTNzFO zWoc5oc7m0iZJ~i(vv@r{ny9v{bF(3G1V$fSu0NHVYubx}67wp4)_Q1dKLf1l+s4s^ z>!GgPWgXd-vhO?g7r8S(deZtSFd+>7U|E~t%}#T%)x73zBWR}pO!4hDdD>_8DF=Ve z>(nHq$2zYxHC1GOAw0E??K>D0M2{>VzI1r-lM}DxLqMs?o=ZE@;vjyI^l}J{o_>Js zBPCQszRa}7u7I6y6q>`pN{Qx>786IxHj@ga`9{EgU$ng7&dY? z_o%rD$?xqDUZ|DTq}ik-#l+|j-IeSM}jxlsxXIF`vo5F!lm*UJZR>| zzb4_tp84(^K}6PD;p>4r{~kMg{oQ_7BR_+wzmNs8(`KcO13;4!!9Vkx<{@Qb=tv(4 z{0^KYxR~sgR{^!7)-FMI`WN}s9#gu}v4l0bT*%SRiojcKglqQZ`3Sx969__t zJGDJ3T#>+=S_YH*^}ZhLL~$@fIKe*DfX4MYobDu!)hp--M*&h{q-b3KL|THLitn;m zh0N0NdxqD_e@QzW#n{LG-m&B#kk>u=X?`8dwIE47{Sd)^Fj18gm}nA2NIPE^338k! z4wr0bm0oDNFp>=>?@WS4x2a^^-+>50gX9V5X*6C%btE&)55rqwpr8Fky=|5{}X@r$^OIOgkI7^ zcU}o$OiKA-+gPG-aa@-?{3-u00#g)zQCse7#elU(Yh@K*Zfgp}Lj=@H7zV%e3 zYvU|$D^%01J-sZdolvJd70rwteNxI%+ekX$NW9Q{$S{z*fUGn%;Ho3QLd#@q{baLz z!;yK+BhCXvWojaPxXODasegH|<82`#+8=`dXKF{wcI) zVS$$utE-636r@S(e{kX4L+cB7&W6-9PaW^OJB62gP2XJ1ON*>O0y>{u8PsJKFs631 zuoDd;S}ImbXv~cScu=JW;(Ur|rmrm|L1W%o=(YJiR+A;+x<+fiUyzcg7=FrD1_hj_ z#2rW?Ari0);}5pZvmgwsGjZh3!SM65b)W8A>E4YuD|TBV@psQoU8Ig~VtA|6!>b`n z4&bG2LhM1yWA_H2!xT!yYmxRIK9N(XgH37T5$~T5x*v~d8riucMql1*eY>CVy{p?v zITJ47dN-=4W0l5~kb=j3R^C6|uA zimyz7)s|Xt91evzA92=t0mG7bxZ^x}YaLiH=G2h7-E@`#DVy(|ab63c-Cfo_GNZV6 zmd)KzX>V52Qyk8(bd08s^yLvXUDjVpQl`qS*xam-Hjmhn`{*s(HE9z);GI}|G=yOc zT`YuB(!W_B%ne3Bsc}~AF@U)M=>4xpzzp{=kW+PYelVbClS?bnbiqYz=)tI)j9KMJM~F9FsWm zzGyTZz22ks4Aj_T1v1yW!AH}={)Ia~dDwoXVQg3RHH$Z2+|XeSCkAY#eYXSu5~`Ap z7IDtlT>^_YnEkmzpI}%Rp=7@o&In7FL8k0kXE3hUg4ebotG7Y#KShn1YXXWC`|Xr2 zg}!NtepJxR?SHq6c)ouA2hXRsZ(#fn2WTg&<;qu~l81l9-|w5uD#Obzml4eooa5Vk z@1boPlkm9r7DiGURw|>{msekC71F7Y<3Ofx9pR1yESn!xnmf#Ym&gVUSiJm3P*~T% zGiIWs;pIr&Y4018q;mc(ZSC5 zzs!juDU6oQd8&(uZ$=iekiPba<@#?!&4IT9@5u&?~-yNK1(d7ZkPbix;63eu*9j72l9n@BZgWLe?U3%vRD0!}vI%d)_U z^z(ItubcMq1h|G&C%fqFl?tWgW{O?{XPV}azN)TzugW#~$`z;{4>bAH%+~3$g4mv- zgKznr+^vyZQeJ1l3KnSxSEIkHE}`K=iL2M;`>NCeN0>07(vn?iH#5c!;>OL8hN;$M zmus|`_=4R%vfe}9v#IWzkIqqM29qi=WzoUHbLUU)I&arPY9KlfUXHM{P8V@Ym-}~h z%opotQ149T*L%=}l)@jc?v5V#x5nhl0(311AM9J%raVbgF3sW>CMO$N24V!8l3%!x z9mj!%I0q?ASP9hi{*Ed?{B!K(zLFPD#yz+JYHcG|IlGQbx7UbildD*T&-Am62*vYc z|C@=v*Bj|KQ|oRcZB8;kxgP~!?MjjCy0JSViL6q`u0HhQd@wn39CrgB?V;%Pmb2({ z+TucIjO|aU{JGe_6(OOFQlUQo?v;|f;^#mVvu=m{FR#9j#}6aDbzmTC!y~X++BWJ_ zGiZ)EmY`wUH1sQ~)ruQ#Vb&w32Rt|d6FP@Gha6@!uuEeq+meyjDXKqYH%5TJFys7r z-iCl1vL0`hcBwV@0>Nri0S-f=w@49N6)hFyyWs+zwrk~aMMA3rDkA%zE1 zwfiErxsku~=`)V7Ryvjc{wl7o>M6j6&aAl_P2YPoZD+0l>>@o(PZ-yI$!34z0y_@T zDmVce?ee82f0gHiR%zs=Hs}XBLX#PvAyQp1_t$rVTihC3yqRHJ#IqZMCU2B$Xh+rb zE2%*>sBUlvJBJu}XJLcZ3-mRvC3h2;y!Vg&g~mwC#2RF85Bzro;%wU|dc*{an7cyW zXk78WAdom5EC2W4bwF|w&m7w$#*dR5kY(IkSYbjv!YHm_S~_odUQEl?VZ+0Eym8UZ^_Js{2jVke;FUiEPc7-5j-F6->b&-@C7kn;9`@anCLXUXq> zDdD@pj~Wj)w%wf#MN7bQUN9JQd-zwh8aQgh@AsAvYnMy7)2OY+X|(LpfppG)pxF!9 zRtj9-3c}=)@?zqs2OA_!;A4V;6jBUQFEAv_YS;y6@*KRenthgeRv~m8)Zdu!lngR~ zJ^$f!pc3F{^m7}B2jI^{`F^zfz9OFijQ~$_tY4|O+Ug%0gQ(+8D}X4nPsY`=kssp` zTqk9UV>Rf?Wd0fk?Iy66^58F8hmiKlRQ5eDx*ISjul?J*R6-Fi{l#O~JeSe{;do!P zrpH@7IB*yQ7>6*qjIBk2m<-pDq;E;lgg?b}p9izX&8QL;b#&v{wkM1bb`C^^e)-?A z&A^gRsH=22IOoLFIrR&Jrdq+p=$_4U8VnZ)nnJ?pq4L?tAdM526KM{lgjxk@3I*CY$EV4nex})^e%p&RH8DdiYlmHWnJ)U!ROYwpb~w zYN3A(j}7=r>;FK!nB0vcp{bi!kl)L#iOW4X$v@DNduI=S9!(dzrD>R#N=$j{8Sf7z z4gMBlkGm%NZi&=MtkeGCzv4!SJp)& zq1_Wb^tACeKtcg17;5g#J&030$Mk9oI9Mqasm?-r(5v{?P|cq${VXIO3q_ zC275(XE%WR1WA+XyNsjLcf7WcjbTyX8vU(*T@5_G(h{vE)iBLtUW05dePTeni;zJH zZ$4BVHds8faUJ1pW%7QQ3#X5VZDPQgh(x0;C-*>z4VYt`pd6a;T3=OxpFJU*dCg*@ zvAgp4=wVWZXXSh+*oEN2FFMdyfBQnRw58NB&Fypa_^G*5i4I16Jz9O~Vf}L7i}HUJ zF&RE4=vuZoCnR~kz9@r5Li}KEuKt$x2|g`)BMxph2Bimcsq?(E&wZz8HEP+5YUaSj zNDPVR_bZYsk$`0lV~H2vXT{*yf9SndX(Ak2;kTD_`MHzfiT~VjC(bQRVJ~v_Te& zWel6}VB1O5>rBiI8sT`%`V_lP$LNsIU|wqd592@=X!Kuht=dM|^fffF(ChOxaYlV^ z>!wjy|-|!8}2q{q2zlqAHw_r^T*72^1jPX8mj44ymn1b*A zK1lXoHNVJCV<*8%nxTCg@{3>;Ob4^Tr@PwR+8Da!fc4+_i6hwQj`9^leJj2r z<&?sp@298!_NM!f@AU^4+YkSD8Z$ho-B)zh53D))bJhv^4PVB2x_G+gC4Y_~^?L_^ z-R~2F;Y}cGt;4*c`UsCo{oAcsmgkQ&p}k8fnrX1On>~0Y$homu8}8C!VXt&gKOJ8h zxXs*pQ+{dmT(euFLLCcL@Cg2FRJ`Y6Wz(dz0Yj)vZZg*^5VS=yQtR}vA}os=b`Kp> z<~`YBRil!`icIL=GTD6JT6yb-bv&|c%|vfkXy20@WW;=tS%t| zP8{Nf;mu7wKV&L6x4S`&yE|14)JArz$X5Mq6500pE!`OcAtW<*hBN#(teKIr zu&gPzkT%2eu!rwo&EG!%q<8g~xM(b5X>n(*0$HdE=n2LZgLb23O|!Oalzr*z#rCfk zd+sl_6Gkvfv2wDt_h@VFB~u72X5Th;c`6YB(^1uc3ry<-r@x~6LALwtzOFjLgpJHV z+Dx!2bm}rE&_{QsZ^F2#wm-%3j`%^Gd?-lw=QUS9k+m^fVl)9gPD*-mW9;YfxK7%1 zDJ#O)OwdQGr_uDan1wC+M^AnfZ~2hKV5U5r$v%In*@N(0?*rS-hql|=t*Gr9liG3p=i zzlsY9p&Ln0KTX&CfMg-d-R6y9_o6kLACkz=hKGW-7O~}#bkBQvf}s<=&ViOTee<_5 znTupLj0T4rSMd~w^U7aV+E^w!salft{+IiG@4Fg7x}I~I7C_2d;yu#ayOE}sl_q;Alr48EAUi;()hUP{zG98Af3 z-2T^Y!t?LbwBM-)u1+6TuwqqztNhj_X7h8^+2Jjt0A;%Nq3=B)0RqMCjBAhH(|-Ao zNd5Ffz_&jg^cY;4r~e6ApuKv_-(C*>Wo{IgzP)&xvCCIzt6#%5$g2Qil@ds$;wbb7 z+BX_h?y8vbc9X52_b>t7{!8=ot1Kz zgHt7}qzF-UDSMRICU+E{@CM>R1-&Brr~OXdpniQ!ztT9`w8b@)5mBgF3QnBOy!BTB zcQ=t^-e5|Zd{kx;gXJ)$#I~(WZ8O1N+n9da#}J>2Gk9sV2V-X|+ycF6A|qN&$#fpR zGt1^<1z~cctPw@Y`puPBnSC&HnBYfH^IwsnN3e}%sH$f3P9lG$LCEU6T6aY87oZvM zbyh!QxJf2r#TrgYWbSd3bDo}2z`WDyL4xJLZ zh^7!iZy|?Kti0oq$`+QUB3g&&4>~fz|NOb}UePN*yNgWn&j$jNHz;TbQRPiUyYQsl zslw)(F?U=Bj4)E{#>FK=-3A1Vs>jlLI_Km)2_XeN$5P!UI|2s@iH_wbbuLwr=;J=? zkV6OZma_&c%@g?-l7C>Y?eXa+)IcWEb)hZm@|Tp{(G6_wxc>eL?Wp;we+ul=?;vy5>!G?x5%FlGrhJGf_hd3=~7bE1krCX@@Vi zNXVaX3+k{`5{JG*x^EASb0cB`_z^t2#uLylc}JyFKP18TRf@!=0H-zlmmt0BV>z#>_z+5ijsa*VxP z8kiu|_liX2%U81Q#Po8SCo@X?lH;$)6BvKe2(VYgPyImub^xq>;scPljj0Q5zGd?!5f*7RiC(XM9R5 z+$A&~ONtCHSgz!0rOGXo+TjWHL)f`DH;b~FC+J;Lj_-Ny)K?IrtCOc3c=M-A+h_5* zTp*^J<8&TaUD5~I_aobPDzFF!cgi1XXY5o8X3i9cpT={RwZ={I&9ob>eSFL89{J}J zuX!G3QPK)nh6tcyLSJe`!P92P1z?>1>y6*7AAD~d4c*f}s`?vouU7t6cJnSErDVTr z^S2J3xEr!X$aMtA$K3$%g;i)C@+G$UTA2~^8oRMd5X@%ajoKqMeO6Nc}U#6d3s zJ-!e2iDg7cDW6GjM#5?TYn9$=BPWe!ZdO$KPaT5Ph=>*px}E?~_e6ZN?;7Y4+3vz+ z_H_H2vIK@xYiogVM;ix}z8{|jJiY!dNq)KLVMpMe7kwI70fwGDKiaE5Qn9pGLE{>`iz#{KCIdmY+%P*+b!*1L)Pt53)Dcf` zOS5Bp7hGl%7qV%S=mDAqUuh?%f@wZP3j9yNhXwiq+c@Lyy%exfPD3gmCVqLP@4YU# zp)gNHIA&!Gmx;QHQkvp~!O|-#i&yUH7Q)8~JXbL(A?M0F?$+Jr3y^Ql;I5`=0Hbsi zaYOi1Nl&oaLEcHNM>-FbB4=>{MFLl)1y`R1{2%r_xG&(akd1QGc$ zs_R7}5ysS3?6^COVHGRx{eJ*Yr(X z_n(^Of?Zn^Lt+;cGKluXN@2CA6}$#^D6{@6$UGJaQ!{3o|Lz#DUQ%y0GQxBkkdlZr z%)q6%og(AcZ<-4Q3Z%L_jo>%f;}t`-xu(&8jU%-i!43;P=oPj{k0pJ_l^PRy+#GoH zi6{TQ#;MyEaF(q76ZxLt=tO}iEVSv~J2S??Gv_=3LyRA{*@|(5dh(LH+0ag57h{<% z>lP#$&c;NxjSca*!6USpryWBk$pC10 zMF9(jxBi6xOuC9~p2Dx^$|A&+`c7mTusil1`#(zaS!(gjLgTb{@LXRmZD)t7HAXK= z9Z3s%P5}dGtT7}%Cla6&*ysLwSt#_qPtq3a%i!)WGY=EAN)u-vt#xP(R2&`fO;Dof zkCaik8W8^zN`3PP#=fYa`+TuEc;h!IAbs%nlGhV8*lM4SJ6B#UE8wGN?60 ztZ$$yv)8_e#Dm=rv!U*H54%ZiicZ6+)atvlhiTIOGXZK(dvVg>ObWn>!ooDBH2TdV ztFVOaYCjI*#Z)y_Q$~OB@iMr0X?s$t;!yJCmi!&k{qTl6!fQs>Wrfq1LD;iRN$7$W zk6}rO@iLlOzNIF=f;MUO;*drafti50kS+iWU{zlrT3zCPaJkc4(tppH!0sV7=Fqy{ z@%#U<+^|p86j1Du@*dWVFWG=Ak-V)-m->zXcm$VN%p>*Mk4p76}&VFpWVaBTll_UI3sqfWjbK9UZ{FF6&V+vN}9|hroT`;*IGb z);hT&SJ)TyndRL<@*0E)fQBQET09(e?v=P(T%#?g=EHM_C*g182xy$}@$}kC7P#Uu z)3vqFb@}Q5P5)Z<|F1Q!*4UN9k5VS1BuA*{mCuS{YErP~p!@wW#a5jE!A)k8^`$pW zzdU%8@TVf^(f@vschxzoNp0Y6ih&voHTB@ShpN=ZorST}>&nMZVGvswVyFO~rNUY8_RztXw8X*B~Guhq34Pno1BDMs(`X9#~+InxR z2bz=4#U#5}U#;{Xw(A&`jGfDCdX#RmvBtdv7lU5Q9bmDT#B+0*lbcys<}4?H8??6_ zBXV+z@Pme546j80q=jZzpx1;WN#v)bWkE8QfXtPnz4wP<6Mtgh&{k%*`l(rvoPH@T zB1xudt_cZ!!Uc8q-azzT_|b~0ht7^4p@$anYjmr#XZL%<9g9(X-0L2whg7EOOKg#B z8e!~gK;~zI>XKm8xw{r6I5bBP>uU0R`dd6Feqsp$<};LLpF0h5AiM__rdIp!9@?rJ z35C%vD271UD<_t@n8Rh<0JBxq1!&?ELyN^RPwT*{%_P@Q`R}#E|FW>urm!)$tVTLy zXtsV1jHo#u)GA?JIwBo^_}cgNOMf-ZAZh~FajDhXE{|aqa_YCyzJdSI15G$T9lj&j zZ}hi2kDNRb-T=sLZos9ch^GH52Hx|C6g=WfvmIuUC%6ds_7!02)(S-0$-i$~TKk*k zWAP9e9B=iX<2{4Jr)570f>*x$vTTkOZaw2>_?HlRYV8@z-<-@_UT%TLRf7S*_GgUd zGj^>nfC0ftyihgkL+}dWpyx7pB&5#v7VXBoHkd;Wz95j64lj}en!4K8{hxJ)Y;%{q z1}AHd7jE&z%7*FgKDr8QG5nBDa_z-_g53x(tyz46H{MB3I;njFf=;+=!u~aj7rcxT zjt@ClSYTl1w6<1B+s>o!B_s0X+(d)`cGVHw0!Bb(aNUn_o57c0|FF`$R$r6_W;Cq9 z>uNM&YufFiu8Lvof=*w!T%Tx`+hyf}GA~f{C$6n6=sy)V6IPuQRc}2G1hpcwk)e z!*bUhWphbWm%;m-;%@92^HGyD->i*C)%T{`I_AS0RS7p5d7fof5xW`)HPSg z)ct%@Y+3xDqJ?&vwC5u{^PjVK?B6j!h`iaB>!hG*Tb;3khAv@we?5Z~mG7-vu202C z92sO7?}4H{X?R2&WHCzLC&YURsJC@qT>5t;Q$Q7voV%>HM{bO0Dzr$xdiwjokAB*} z-ZhTxOc;=-w?I7CPv|yfCqA2h+uPpPCI^f1_MHv zj``kgtn<)Y4m8$x&-8_f^payP5 zlIzZYH8+3M3tRw8-(N&h+xIy(;(tu=&8dmGP@0o804siEDu=Txs)|;iu7L zx%&#Oc$Qw2GW{g6SI#dJ*{N^aT|w{Onhig4Pfq{rn6AL-3Z&T4`nkhfE=X5fim6tE zKO&HYH?Ycm&tZ^)qa5xm`rjPeMvsIw^3CDIt*VaH z_DcTpI<+wcuM@sy?%wX z-YSVj@0)yo+k6vrB{J3HBAKe*>@BH3Wu7(BoLxbQ{XCQ`Bl^SKMfqof3gN3aq*^iArq5T=T_-7C%|8;Re08JAVCn!;&P62 z$$noedoBQ9+*vmo3s1Jf_Xkzdg_l%t=Ed&UX5-XgCui?sC)%vUnAwn-5)ZE(2=biW zKn6wb@}(m3AxKliLTmxp`=81k=2F<@Qb;IF9GQ`)$ZK5+hgxs$=OvoV7NH8GRSXd3_&4|JQ+O1O_4>0-Y znoUto;LCVvr>flWfgq?0m+=jM@Zq8Z>fD7&O{h#wkKah!pk3)HH6}OL^b{uqMI4Jy z$885L#{ceoy@$S5enKvn#jwcQl@>axIiKess2es8?LKIcCu?uRAo_bM#- z`TicVJnz@(hfh;atfA|BL-Vh{&(3%meA^hDvU@4duaY(WKp= z%PE@^(S(6cpy`#?Tk>7+oTOdUPpC~Bx3vnHWr$aj`~cy`AOOVl!>wk3*x?+JAvrOZ z7+<3{I*6hhmJ>jkZ&t&2P2I@m9b!aV?i#!{Mcm+GV~kEQLZ>@)0g9^xRZfmRoI!e3 zl-!t7T(IdD^yeqwD%(Wl^~4n)S*f4<{HbuNty?s~P7C^ucC|0;zD_#G3LgTyJiwb@ zv_v-gs2aCZxd9g$4hz}PgE2n{@b4j$H|zvo6UgF7^V5ywyGp~dV-!@pKy9@tU3VAQN<1QDXZBM3jLG$kZ)#jS6+c&0V(#r29lWL=Z}{?`IK z@DzH5P~RU{>s-t`l{UzeMbJksa0>SHPNPLhl!1gzKs5ac)T){a574%jiT=^MP=enr zB26W}<&rwT#k0lFnvN6W7$i6^;4}c_RyS6`-8DS)hY_yo;uMppB%W_(4FeE%rywqS@rGVa{>UA6lL7b%mqR|c`>X`4L^N| zlmNt19^Byc@$~+4s(M6}4z!o6j7x`Obbnnr^hX?zFbgF}PNQ*Zm4P?q38=iYH0f8Z z+reN%yJMDWOs@V@@8Yy87je_*03@lOWtMtDk_(Ef!8S_>K?l0l7RWwwx#5GJ%1qTv zt$l%(nqSj*4}?;%La1C=okz%tylI<6hQ_+8)TE1CQq8Ni+y!kLIY&U85yFXzQuLaY z_DyZ?kB5pC(h5P1uD~JXoqzi0LEGdJjH#U%=UuX>9Ci(X>?M-&h-C0xtp(hvb-Z;v zdrXk3O2~4A43d6j+9rk*F5fgftuV#ER`5y$_>b*$3{=xg6IrYI9`=>OO6$=vHrI1D z420IFTsQ9d)BITdY-pf$%=?Jt*I?1Cs~eaJC~>KES&q>zjv=zfC>JWjr_VhHvarF0 zcsC*O&*rXd6K~0~b||ACH!jp*C&dMeM%9}5xUHx@;Zm(?F(&A?gH0PRsbd!8$#;QIU z+#N<|1DS$)^$K8QipjpS95AJKIT*b)8oTrS;RitU`VY-I{IkkuK3Ev~4^@9d4D)%6 zF5g5u?V5b9R+hk}4WL!lJ!>TwXp=Hhdq$)`8m&?g0({-pICc8|*85!Bp1|nZ?o1(nqu9uCHA zwUFY=lvSfO$;Nf1ESl)av@JN4rvzz$+3PQVO}u z!GrFNX@F}$%TL33qRA(hk_DAK)4{2m!jdGE98{s*qBMxE#!sD42dD!S1f|M>fr1xA zJTjtfH_}RVAY8Y-d_5S|^}~*kR4I(NoYb)pj{68z5X#&dNSF)h|f0ZphhwW~d`S<>CFtO5NY3q)~6dDf03QNS9o z5aASBjW6X6P%uVz7N%es5ebloCdS#_` zE$f-3L8ef7i1pQ6iQ}1BTt=HP?!s*fK!}Mz+Cl7Z z-Oj>oSK2!)7Yt`7Ts>`ZKD?o04gE^ec0qD1bhK|I002X{YZwtdXYS;J)NN(+x7gO~ zRW(Or=C2j1+@s2W8zh%Tz!`$onk?`_XV;T|W^8`GZk=L(YVV7el;zvGFYoIq1cIOG zSW;4WVwc&PWnM9?1l;IV0~t_>jTa{*fgY|xLi|zJ5K=I;7G0|oAZ8y85Ne*t$^?_2`?eEgNLHq}e72FA09r#1Q21L6Ed(Jvvj8 z+8Qq4L9-_EP;F1--f|d@8}y8jJi*CM_4oX=oyu1hXZ3W0g|n6Rj)hIKS8Bh2YF4j_ zvvs*gg|%w1j65r7HDC&Go-(pZ=^_JqQy~X%-T7XTw;{{*-ZM~l|Dmm_rZj^ zG#|#U{7q=_@VG6|8TKQ#Y2J+_vZe9l>dUM59&-<+oT@pt z@rsCcV?XUiCTF?93|9$FQ3PdUeWM37NkL$0I_1yj{v5q$Sd(+udZ3UteskLjA7Uat7$bm z%EaXURie4i{V>%1VnVIEd+!aKSa01m!VWZkBHtKjBGXLZwa@3Bqu%eq3V=?0ssAi1 zNoKOM1LqWsOzBRbq4=wq%D%Sm^0_e**%*q5x7y>G!nZ8!GnK3WmZ^1;bcE=vHYEqb zWKsium!mGnW0^lQf*8|GU(4Jmg-Dr8w%z;rb5p`M&@anhEoQFy5#yf#7^!@s2DB4q z%My5R9Czydow0a$xnV9np$J3ZqU8Nbu1BLW%3F`Ry6tqC5>;8gV}nTdCSp91tbUPh zprUZxY5+hR*r-^)jPVQy%YU){ft7PJbFV44pg^I4*dIvbe({tQOgOf@K%4LO9}-9==88l^j^ zFaa}m4eb$?xzl1qArhI?LBt)&-h+yf^?yO*Xg8AN*4ra~y9K!PWd!$DW>1}VML`oy z0MHKs<6yW^9dH4Qje@F#ml(k|cTNeJD{3`8)FuL|0T7_@7H!&gzdZn|tLb(JAeTZl z^jI9QX}tj8b$D*wqN0>ct5KA~{>LrJxmG|`6&xZrtz1w!b!@>ePCv+z#Ux^gHc=S4^FfuuX@V3E1wL#6{(VQ;?0(FFqWu*qeXyG8FenGhVDgAG{F8t z)|roJ7lUyg$b$q`Q4OAlW0Anp?QD-!3Rccpf3MCdz+4IMx0&ns57yuKGC_#fJo$Hm zX#eBf?;cA}(yHFZslVFO`B`gCJ1%7L?<^=K7b?w{M(naM6ne+i%ZT2o4VY4ZELYFe zhRIA!RI5>m-Pc>n)zI5`!kN-8mtDx*dfxj96=txT?LheGgR|Mk6>L|=zOo@HoslvZXo1bJho%jL~z$tSo2$s(0L<*kw5H z!*_vH@V37O0dBw{zgj!G?4vd7QG*K4E{?sgv(f|R-)0xb@l4y~!J=ybAOgqpCCD|f zBPOuXv5<1*r1nc@9=u)(uU+Xx$J%4VG2whA|6Yyxo=k6}2HV9PHFgcc`5`j58YFS; zye#%Aid)SSHB4+TjOn2nHBjdEXq*E=qDGt6&$5y;mVl^|4DbrbxelB#YmFk_J5Ft6 zap|sZPpIr3?3i7(M`S1rmVAyeYpwy0#06lbuTg@wC)43B`9Yi(pFw2m>W6TgDY48_ z9Yi`<2Q>e>(!TM59b&S0mnD8fy;;C}M84j29-s!ca1n$2BDE=K7%;h$*pP@Zy?l`Y zb?||qlV+s>D6h(L4k#c;&A{C?kb$m7qPlL}e8!&bZGq7T*c7;Y6X9h8a^%7pAT<)_ zj#wYTs?K1hD!og1O$}Y`Gy8+SOXR0Eg#;zr4xH$1m8uJrH_|!djglrcz)nY=(ZB|c TS@sR^0;^jpkl~O2-~Im&$j1pE literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-10-Concert Photography.jpg b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-10-Concert Photography.jpg new file mode 100644 index 0000000000000000000000000000000000000000..caad191d2bdb156e9c5371abe8e5a8415527f937 GIT binary patch literal 70971 zcmbTdcU%%)`#+3(Qes;aL23jG!K|HuFST!CmP%1z1y z%SahRq&1{uG^GAqhZupsQ%-8}ALO5ZQqnTAa`Fm_O3F*X8@iW5q@`qJq-AB~&V7@IC!Xvd5Ev93LWzvp9=(G~qbDTpP1?6VIVB@AEBolN97aiLS$Rce zRW*~{$l>yun)xlKPIq;mIos2FuJ77)LI1$u&<)X;czoi{-FuTyr=C53G5zw@%z#`&Dm9H{6V`_#*Gvdqt;tE;R_kwU>!WTBE3g3W+X zkO-C>y96xo2$nPhiGZ+1XsR+Yo;dlmG+wEfr<**Dmq9R3FOjBkL|WeF#WXc*&vR7- zJyxq`hOazJoh-mdGQY_j8-2P|@IbA(2!mDu>I$;-G#~e;;rN zf+K{gAlN0T+9esDOnD>+UCkP?MQ3MtP^HxI8@vcC&MM|Zq<9QV%@cd0B8eCWk5i`^ z?aNztqs|v}6)<=5rdm%e7c=T+Zj`S!GjaB&oB8DU zY~+`_@@qd;G0IU~RJ%v3P+?6GYU~P2ITosS2135Vk*+I1tT-t^=#R*)PIptIQ4MI+ zr6dt)WI4NnAv;|7pN6}QOVRmH!?7FQ9Wp2Up1la{rMwNcT%Y8s?MlZ%uac}yv_0hT z>_(XJ$aZ5a$=WUjn5^m)R>UO|oB3sa;pX`Cw&lY0>20j3>f5Wp6T}o7OJ=cNlp1!; zc_8~T68mvk6{{J4vEnB^>WPC`>v7nr+6+7b!e-#*t>u{j1Ttv&da6)H zS=CHFRe1xJH=Wi%>z%Gdi>RopOd6M0>dirqsk$uim^ikMe!vbGtwkHU3|ICbU>~LD zH4)01$z$9{&XzC77n9O!`&;~Pl)b>0ldAJL#_jl40U`|flxkNBacYusgFeO9Y?E{H zlUt+jhr6avVnDb8+0@rQ90QtIDGN#xEhXZt?QGc$gpA3+<&V-Czb^&-fne#MM#GLi zyZue->L18f%RdlSck-*|1IIU&{(+oZ8R$w6m?1KXX@(&NGn={jxI$%G^y$aLW?vafEs-xqp2zrztg} z)%vFMM9~>;!45ccst}^b5@`Xaw1!r5gfi>L60qpZIV_gBet#qS!HBksNDGU8@X4FS zV^Ve*kBkvDNK6`3yIRKDozzy&xrb$!GoCs(W%83qSuFkR7BS<3D61(ATN5GWBVDE( zN?vhNx>inis#Z=m)D78}ij4Cp;<5pRy>Ql&lSvOIjpA&TkyMWSrowg475>GMeHRr5 zj(?Q-w@IIpZnKB}Kmci)hLszOV@T~Lqfxmc4lo-#LrI-dPRH;AGmx6r9$f-{5L0Ed znZ#%alt@{PcFJErAQPrpfJ%hk3OR-2<80NCQyKCuBnGvqpiHmx8W#!-xb)%-ZH;Kr z^%%0&)@I<(0~EPPX3`8*bE*`-UE0ZoQCFA^%v6= z``PBTgpK_f9u2x;Z`v^39KFm`Ruo+W%$Q`Tis!Kun8kn1*g`+YEGIbV8TT>wW0aZu z(RygS#Eh|My}Pko-gMp=>?mm({GeO#UOn)1MB4!8C^2pqDY}R$eT}aKXcZVU!_$Kx z6;5+-2Q&`M1$g}KdYTf8NkglyxW_B?Aehq%t^#5ZLZNybG|?j>B9GkZ#gKlc$hl97 z$n(>`et-%VrBJ#4G#`NtGS2P9a{aMp=~@M}QfiYjo}v`)<>}>?#b!vt2+=KF!cD;I zDBA$KAyNB3{dS8za_7b$$SdoxgWa#zy8M9@=^HJ4y!?Gh&~fVxYjZ9BK=y{UmD2;7 zE4Y+RtRZa<2yH}cUXC`Gk^^E~|HfQ0t2waE9bsEKQZ7fy^VcXq-B3d~^M;E2XHllG z)r=ihRdb?xWO_ZqF7iD2D1Hw;mX(d*0u8d_fJn3tFqbt{66%1fN%(4M?_z`lp)MQ; zOx0CJ@`I$1{HQFBnLJ}RRkvs0j0W(zCBWx0qnn4O%kyoS zswFDGi>N}m;KQ^l;*cg8H>Q;(@WT)aH;}j8ltLzD2m%B7*r^O9<89y_ZewYzTi2_( zLR)w!FkN)sSb{rM@>LWKY^U>&XZAJ%suW6v-vjO?(sFyV2AwHdH0NoFp&Z=a*ieON zruUdZmU&SDrvSwa0L z;79>C1Eh&AQvq=%!)>H2yP}3EFDvpRpqcXNRRp99S&2hgZ`WKkRwcF)!*PoJ4HnD< z3~;>t4Z35C=!bK_Q97V|S*>b91jdktWm$RixPdKK#4S{ZGB5l%irLmqY9-s2SjqUx zH;@HR@MwbObR|X?{7j9y(1upsh*rM#$%`2+G1#|8^zbNfyVLG8U_-g1fM_2qKL5vV z{}@ss+>>Koimv9ZQ4>14i~{d{qsJ~OBQ5P>ry{TTnO^})B#8@Mig+z?DoIpN6~W1# zZr6eQ0A@3@hFC*^7y-~`VI}WKXCPSK5?o8dhy>gJ73*GfxXsP=XZ?#eTj(M4$?JZw z7q({s;po1?i<@lluC2bQ}v|YeRTwtTbP_IyZYA7gwqWb14 zeAk?)3Oq98*K|9PH{tinszJMe!4#A!yR0Fr8CVGqlrSQ6E?S#Gm2z=rP+&Kr+mgja znI9D{DSrR*{naZ2Z|{K0WJZ4te&hita4GHT*tiHy0m9x`5#CJ_YY(w?81zy*V;<+ zt#5M2EBwVb=;z?o1e1{xDk5V>c(9BOJ zbD9aG$ox6^CxyM=M;9m;^Zi%+9Gn-?>|JAS78bty^yLU8eD3YNw|^jom^R<{uYs>ZK^5XV}># z-|l4ACC&eV+>9jFMSnQHAoM>;ez#%&*6AG=437A2{kB`x?t)j7`ZxL!)Q8fRhxMn| zz4Tt44+K?(@Y~q;yz`|8S608vpStLLBtPH%U}RIl+puMyXKsgV{={*NKmLvSxNcqZ zm&oZpo4n7T=eWmSwG)r)-z$8wu5N#IZ{_b4$CsNXjm{sxbzEE-`?+fCiyhrY2Ad;~ z{ee(?zxuTQfxN7I9cKOV@|W8$@fHVveFDF-k-%zT}0imqRB+VWQkC8Y6{6UJ*zZE`|`#Toj!$WdTWT_?@Y>lDn;zB5^WuWebg z@$n_`oB3DXXUV2?S6|AfhY3?%ODl4{k0#GdJK{eb+H*Z6`)s@@)cd~YQ>iC`&bJR+ zo~wshMb4ho zX*fJmN znCCH%4$~dG_4`hn4{?Rqi_H(_6V!%}qoY|5#Ji7BoYRME(9sPyQ${NmxWn2}_rQB9 z78)%k3c~M|PUnv+tnEs8?OSE#d#*Beq2=Xk&#$AymfHI|+Ls9ee^PH>)PduU+Y+ah z+mEL>;Q9o|ZBsZ7(PWgmRIU{6Z< zcH?Fx#rLGP-q*MrdCUVftC)EwZCB(NQQrS_oT^afo9yv?K|VE=wW@Kopq;)gBN^7Ihn!C0!w@D*dax)c3U{?&?@-kRBoNp3vY_j;3kJ)p2r{41bG=^>>FezcL z90+{YLWK9~abo>#Hc9Z_e3Wr2yINqsmoK-G2kKAqIdU%k+r|X)JNM7{$X9`cSn1@P zmp7F42`e*0>t&|}@3jGcyHNV%tWQ^U^NHOi-Di@YFE>^x^&N5EoaC;Pd%;)d=={XC zD-WEm$AqaiZ+$=y+7q{?BFWj5DV~cyZ{FSJWf%D(ef`R9mNm>R8IU_%_mcGUtE0x9 zK5p`P6I2{@8M9^DoV~+zEMfDiD3@NLRBFXk>NY7>z2m5AF!ZtJ>JBM85n({d9r|)#|66A5K z}7NLIfIt@CdRgHk8NGxv6vSZ=S{(RLx<<^s>W8HOh38{U>US0 z)4^M4Lp5L2%z$p%)}VX;Ca%m-NiqZggJ1`47a2EGR!A76J9Q zbO?0Q?Sn+855mqrkR|~TN`@CsICc=fo``#TVC(l^z5rQ6kts=Q&jKM8AJ#VJE)33Z zUF+By*R#;Pquq(A(Y-lu;-I&~M90Py|JndzgT_oyB<8qM>nL1(Rg2AvR@DBMWBd8y zF(+Q`@;MT6ibW>LU^6ye8B1$UYs$dZY)l2K0emc4E5ePI;4NQfNHxCp39ssi?z+Aj zxWM^9uK~zUQr)XPAg2A$V}$tH?K^-+1WQ_ zZhw7TXV>*lq+36yTBk3rUna&=5UdJG9{tb5?yr=VydVQoH6jiq1poea7r@OZ2sVA` zaX4L}okGgL3KZesZ8BJpHW3^zdz1R`bqwxO%;xKR zBjjV`SG^5vf2`vCKHyyEouic(3J;C0#EqBeKTl(ftA?&i?{K-fjq_QUlzOKSbJINE z4*D&6C;_kPu4+S9ivgLEi+>1LKDE;v9E}k2$)h620pv=w1nep!U~jON3;)IyZeoUx zSgOlcRH>uijL7gML{b6HZjz!D+00?5sr{AIAia<_U@pQ#@>veV*=m-YIm3I@EqMB{ z&6&r_6r=BleSb;gKPqxMHh;Yy_jH->_dgKdUpi*{hwZ1jyf)0n>--+9tie1hW%O!I z`1^`ZTwkc!)Ft#_cJw{8w9%OfdP>hu2*}}Uw2pR1_dP_-M;A|zs=S{U zE7ZpF{m1uWP)rrm8~_0#62PEzWA9WMJQ^gB**Q{8(T(zTOhqFm;7>_pJuJaRC0hB4 zFw0^j`qJLHfT449s?RyklQY%(E@k)bI&^A=ET#5% zLpk!M|CX6I$-JOsi7P8o@JK2swMxpZ2v{Ej(1xUfzL;b{f$2qKST4^_QeBq@NuhK_ zDf%xQcj0^dJw?wo0@LWe=T#Onr`tcF2Yi z>4~n$>yxFYT~}WD;&iwB)$BK7nAek;8^eu{F8b&UZ~GBczJ;=L!H#$?_oKD^R4tsP z>Z=-DYh!w^aj!eP5|FIasdHT|&N8$$9+c0K7W-3hf??pe{t9ulle8 zjp~tDo{SGvD;&huRb_?KFx(gd18hiIyNq&J_L@PNXHp%2G6B z7TL>Ab>5oy(ontY59DO}kypKz*Po?)F3fK~d$nT6Z?@7O2>W@5D#JHra>+y22PG}} zaz?9mZm_^#)S5t(0y9^>$9sH-S=KCFhZm)_YAei8Vjb~)8*lr^^o^&OPo0{gU)G8v z6|QX~C{DIDwAft?ac4cL^`g@QFd>^6zRcMwqVF+04V` zpTb(Wm?dd5CP!Ak>Dw&~`tX=vv0#)^{fmxz>zpnzF@S)iYy%7kdME$Fg9H+%@?9PCMmNIA!j#>ZO|vVcktb=kCSa60AJK&z<$6#H!W5g?{oc zUtDd<@B{Bniq%Nv1*@NoZW;NFoqZTH#yEmVXbr6}G(a=ENQyQQ+^r^WiS= zZ(js6FTNIhJ2KON8P7WW$@+-erd1D7HVY1kn9!33uP?mlnMKx26xvZ?1Eb!bbVQ3Q zCbFK|DCFrFc@e*r)o+)Ir=V6kOctWQ`{|gm*`L(8hT-!p8!sGg+6b?5R8-gZyvu)W z_&oQKIMl|=%kMpX3DZW$TnLkY=G8L-7JAJRH6s0d-d~SyFYbH(lgP}l_PX6J6;JnO zs8X<+=O^DmdEW-E) z@n!DlW#mL*Tri*z@LCtmlR)W8W;r&=v)Ti5pD)B>aMy5y>xUr z`U%~Zc4cJ4#_FlcZb|7~BOC;OUUV6J{mGxA>1s&~yT|24SNzqHlDZ#J!c~mL9?D`_ zH&fC+0TnBD6#=R?U^MsP*_AWvK5G^Jf#AHhy7YhTXgIv?$Y2gAM1(v2fz11Ne`+~n z*;4Yw+xFBc^wU!2NVZRTzSdO7RWoOkM}2orJ6b=Vp%l4>W6fyyhX5oG_`gWpeg86E zl=WIu$r7*dtUcmJ$vF%=1xR9XFyRiU^~5>Ag_rl z2kx3iaDl79Ex@MNgBGA7wG_<~Z~NpUl)VeMS^`StYp26t&t6@)3=2Ebl@KbpF ze80ncB|4aAdGn3O<}8bei_6@L^wVZS_jh;BJ6WWt8Yo}HcLV?@r%&m#DS=a^lxfin>qv+z#pxRNZ!Kr5XVwR)xCe1)2d?Pqo6|g#9k6KQ9G7>#9C#qGoh3}&) zo%}$%*L)SE<=Td-YcRSlW2fCfyABkWxxfOF#?GPv{#8_7Y#J<46aH=ZfTjTxbOo}G zMl-_Eep?EBpMCjxtm5D3a=tTq+~~;jjX)lstUdA_MfrB?JmROt@l6rI_hCtTs{{A9 zYFx(iO-33>!;|Sg<@`Ms^s8q}x|f%x1rHhihI55?O=pC(s}`^YZ0kJ?NE6d$#lkaj?)wggT!1y1oT{b$AcqMTiuF~*WAx}{&Z zX)rhjH+Vf9xY}t@l7%Nx9V7)<&Avqbh4Go6_Au;Tyy1Q}K6C9n zL9XlY(aN)rFrNMQRKxaMqL)2>Z8^51{!3)1z*>e2GnVnm(-*-H9zqq{8xb((PCVww za5#@yG)7qQfW~Pek);$lO>m$2)D1#&B$a2!W02huQV0MiO$L&dujIu*G;LEMGAN;b z(f~hpb_lLCE)jaoRsPt__(6Ss#I1UiudsaSR;l)NMhKuAX!G?~!Xd1vv5 zFf(>+Gf-1szgpbtw?|WKwIMI)(9@x7wc0OXFFWJ=&gHAQ=iez62kIVsWO24ac;@MY zd6)|*=C%<&GU^D)1Bu1k33y&BpYJaMEeGU3I#Vs#cpH!ykpGP&=3h)n;%FIY&oAOf zRUHV=ZJ;-iD_InR1<1TY#mSQ77Ub>!2yX|8`uaD33_vF-!Bf(;5Uz0IY4a*rV44HB z@itDLVX?R4_$fv)fZaQsmOP|u=~@!M#uP0REAI8RWu_FY!=bDVVKa^T~NmP(oFj(~cy;e}?dz%;ZUxnP>lH`XW9g9mQ<=hTdZ zMlh54>S&QAc}yh$e z$XROxT^~vs4Np*-WcWFA1=>q;a5hK}_SLwqo<=A!;vF>49w|P`HBWyLE*9q!k6$FI zl~tPZ#@ zh=m~Ol~7l7y4BiBpyfgRa51rybTvVDI>DXB_@8bVcrIx}|KKknJCeQpFd8rTdE&mi z&AFej>|@h(ly(XNL$ts2U!k!4(6awV;hDm|{fU<}Xni%4mEBZ&&eHL_M$*kPjLs$g zLk9bk@7352iMqCxawFY-DqmhQVZUkgX;VVHp5=PLcUGE-(_Je{PL{aB_#OJS0X@&# zdN{3dL&2wZ$PkPeCp$t$hXRSOoL)+Qzv$iJPn_MGqDFJ;=^$lK@ms|$KS{P_UReIq zmXo`YPn#X$tmHuv@Lb)wbZ1Y`6G!@oroTo!*Qnh#F+F=@4wZb#@ch=#4grKoYFt6{ zhhBGhOK1`%DQ@Wd_?k4q0Rwe7eK=9ZPF#b-(GQbQ@hH!eA{0QKcRzAtmb+bmw)fOz zrwuX>p}8_8G(Baa((c8{2V{>Xl404EaA-u?sGAbOl}1G~mV(j`RVtInh~pTiF2mU% zc!0qgby>22!Mf}~mkO$!!k}qB%zn!SZT##oWbq*n#pxQ~aN!CTr+?JT9a-=Mu$r~> zPC3A*Hc0kvv(-gvw7X5=tE4N^yl^=^RbV1RS^6o6umTxxyqy<=f-`r7X2$snRzOu< z@p)@g_RkQC4$iYr9;hRlhOy6fJ&@nZ-PS?eOo_W`hTEDs|G{Qldm>&wc9j!7Y74~K@hhq>c|@lqu-2*O9`;~J^$;4_!BqQ zDShe!Fr;;{yB!Tk{c(Q4UmWnehc7RQ2Z_PdsfgX>4NC&Ozu4}@U{)N)QL$!G?&C>% zmd=*-Zw!Q0oYvwYK@`qx3j^QhG1|(c{tRe--gfsf<78n-wNEvt#D8|7wj>~{B%>tp z=8Xl^Mbf4zZV)FNp%d_xFEi&q{*@1VWfyCGv`N0@D*q{8uc!1~%Q^P9a{;21>yK}6 zR0+plo+&@%Unr^#Yc4nNRXDe7OXbM8-tV`x$Gn@>n;MN?hMbP?e$!q&H4)g{KUvRi z5AfQ0sjvMRH17m-DH3%3dD~&sTqs-wG-e1Cthi6&yPf8O*!))ezK88OuuCv<==SlUik(cSgO6j=}Ly41Qrs2zyqzIM|m zrC}XaNK<*FG=f&F00Hfi@(xlC^^>!il#3PUa=|Nd*9E#wQeJ0LB~uP{fJW8AdpIVH z!*(7ierw92pUx-{eR_;s?PBUT*Udl5?C#i>?Aspsk}-BVwx#Da8S}30t|w*35uW~} zH5J*`ZdUUqhAAS4g`Qo@rhv9<3;h(cnt;tK4yORx^e?1<9;}{mEZuK^%PiduNWNMy zfEA>`C^AIqd1`^owgI44ewSn@vH^%|Su_C@IGQuaX{=n$={yjZ6kEi>!+#*fnghT4 z;y{y#AM+CR-SNNVU(gn#%x^cgDzZTvs(Y1oJ+h9ElQSOPCESwp+F2b#H*YR+wL-rU zI@P!L|FT@|IMv=iUbE@)(bfb!((cg-oLN-kcn@bL0h(-}Xc$mT;!amfToC97dG@ne5T$XWc4Jn{7UfeEJH6 z>vvz8cHby|9dM0&jnu1VkssrEn%onbSo8<7da(2z{MA8%QDyr>^*~|5`!9q8eqV`C zrpj9e)9?IBLjAT}`(WxxUAMIHl8N%UgK-W_pnlETY5Sxf%~rm9pNCk}_y zC%RqdG06a&M4y-M{bR_XNzHNzHxn+^U$1N3ylG>}wszt6mB!tP*}?CynA-O@ZhOg= zww%~E(V{6UBWjls_;<&CK*FMeDsJ;T?E!JR@wf^MchbgLI*y)dh{B<(b92Ha0{nzJ z4ApI7o^zECYdKd>cLQm58NDesr3^b8}%$EZ`Xye2R(t?tv7!kd7 z{~fo1dVKt!>hZ;v#KT%IXAT;VZlY7fBVppFVAlCzv$Z;<-XTEX!Q<9JxzStmXCLed z>FnN|KMQv~=t}zT8j}A#KsM!Mq2K5m5kJA*86}Pnq_|QF@JF>;ti%QQlZArl0dxA~ zI#?;LJN~1qPoY0?j@%fyv(Im(rMP8*u$fW4nNg~=5kz&%@qmzciCgu zC|JPm^0*CUja*}fk~CFDR98SV{Q?FIsRGAhANkat9(EpsvTq5Ct42fcI3u+Y0}2MS z+v5tl!^%`97!763O`R#L_)1f)8`V$;pS4qY1(d3!Y8i1fCt7w)3C7;1W4Sq>3+-YB zEWMJ*mCS@`#o6KIYydPXt^q_mUP$HV1ZtX<3ylq8cXDzxie*`uziP$Vb8$GD_u zY^jLcERf50MfUI&c~ixn1#Hv2fh>vm2zo4LDd-a5avaHL z{oVeylq3WyU)8n_9bMxk88ouCle0l^B{=MnY+Mj$=Z=g6J{syx1e0)r1k3*>$jP9 zWI==gQ8C`bFYjp?q;Gfecc}lX$o+sK*Sl>RB44k5aXqgdx!115sY8EP%&R!@rqB>0 zOj5n`vfXd*k6Y#)?tT;g2a;V_H++OaI&3bC>3Be%wKPsR+P?o1)$d1O+mj9X27EUb zqa=H(s*lr;Z1+&qB`fle@gg!dZp5K8NX6daZi9m$$VTL5WuK)W31+s+8-}>-GYTMK9nXBZz< zm>I=^MzZ3K34~R-vAOHzP1;>t9ics~`R?1TTtW~g(BI>VQ?(A8QI*cztV2pT1a!3a zfPX7v6n!g1RErH_RtFZx`6@6P>w|P;pjFd#^+?n)IL*$d5!#0CFZN^{MhFP`T6zpS zUx6;hRqGZ7=Fk-u@4*OLUc;^w@3bN?ie_ zSyQ*4?J_pm$VTHEjMANOBDx|tgJ+j+Hb*=c&h`F23zmQ=(QY( zr{^+y#?c?Jjk-;!jP*FqNJ9ciw#h2-Lr=S=F?p%AwH^yjbFeTK7|@=@y(hqo=bMH{ z4R5VXbhJV~y-0IwdaFAV*y4bE8fl%a0uLphNg;MkxBGifS7;LNdK1so4bCF5^6ltt zGef%_c1h8GlIFb2x=l=^S;f8lsNLgI@~5LeY_!kok6G8-?x`>suNls*e zEm>f)%7r{pg}mumG=y^@%ej#3BL$0JR}>l`r>elnySWQlfKF^;sN8MjI{6{5ab@k? zxYB8&GUccTp#aekaw;qLm;(9@=^4WhnbT?KBMnBy1+r(%C9)m?9Rism_|n6(%ojj^ z61{_L*Z#jSe8>RIsr|3IQVb7d^j&8_w}~RYOCfrCVLc>Ql5om^fD^zvo=N35`q1p3 z>X>^e4#D`lZBZPJiM$xm$=VZ#4CV8?tu~3;H%+!S!J^lW*R0Jh8R^!T2`p(D2=t2! z9!#+6z}KP9CAKPwPVKN7s%Q!;S5D-Y#rfSz)-nTd23G_eZKKumv{swHf7jFDSoA5Z zaVqgK>z1o+%+2RBVLYWDN3)m%k8)9=1(sf3%i0fTpMK};b_n&k;7H=@S26ANnA{h3 z9!{W06p?lpqWhYFEESU(Od8p3;`sr)<>>yLx*XjhR}|^Rk?-?gPw41UVJ)iV%uXMV zy0st2R-HU{=D>LG_Ow&d=MKm)GnSl{x@wqR2fA1bx|O63NV==Q_J(fBXvb4ujqTYw zh4jR^ffp)~`(JxS@2qVyGWA~@;%Z0uc=ZKAR)0jQ#Wt`I#2mGx(Sem-NXg>n9Khp#)C0w00 zkr|M`WaA)=*ZK2!7V%VQ8fD7r`r&eSt&774%3u30%b@I2jE_eZ=V`Z~Ge+{=>HA02 zXbpBG8AjJpWal%;2pCAG*k8Q<)(!fGG{x7vqFkhgxG!+UtxXapLi)83$ z?XpA;7sfxeI7--nvj&sjeqBT#A~t4=!t$Zkq5(hmYU3{JXM*j{h9}gHI#}V1VDac? zws$}o<>k)rbZDDLtH3+Gip|3c&F5Vat;UkMU{M3O{t{>(?GSYXl{iCOiLb*&-2>y@ z-S%Rt6BsIfS!~QGBFlNKCfRN$gqf@@PSBn8yHj!_`DI7S=b&Yc+ZSYm3b*nLzCT@K zSsYd%zqxTxXtLj&RP$@g=210ocFWN3Hu3jmM>ywx1SLORSjG6#2(y`7CF+d}TG7O5 z*D=vA*x!_C6)R8hG%K6qea`AjDfS$)7LMzo1B|t2+Z|0? z@BV0KMFj)ZjLVqk>bq2ShLowfwP=}j>g7}&fcMpuT{1ZjgD1Xy1HT8Wxv?*@tAhpo zT7-HT|9nA~Op-4?*}bmh>$R08m)mzH&wNw;IOJ!q_?96ncYr*uh*@1hBe?1;a>J;Z%7W`GMPeGXAP@zK7gx^Qy<$w7jdg zf$83;FX@=PRfMa9$*lQVhO=#wtPJJ zp!qfPd}idnh=)g)2hPV-T~;0*%Z#_{+jk-xF=e+cHcr$Y((Xg&^FwrXywYP_o@h}f zGk2CO#YLXk08g^sxx!`C-2pF0(J}zseTCfjiK3Ukgn8Hw&#pVe)Y8Ct$StRo2eupU zmZDvie9*t);%dOfA30XqCPad(b4<_8b6T*1mV)Ip{dt$1`^5Bk%rYa8@Uh5WIE|c7~qa z4U`+Q1?7gY-|2^ZLXvoHO+=cnbSjX!*@6M+6C}7U}E)l55 z`yt~%dA>@LO+?gX#R{M-qT7iCBcnf%=NtZ~wbp;B;1_G%f4*`hPz?GbO(&m2@z4%vu~{tH3q>iMs$ zwL6WieHs}ND`EDGjLD`Y>R2*+j@aCa=+=xrm7>X>1DDu5y2FDQg3}c{i{-#u^W}Wf zPw$+rF(nGT+2Gc1bF=gF0kBafvah<-u=@pk-qe+K+?x-5Z;1Wx`>kGlA=N@OiR+7Ifb>-NZ#!6qp`rUZ>J5nv&XX`z6AM%G5gLXi2pd` zo4T#CyOK-uu>jz0NaSP+u-Cc=G_h+n!!r`vn#} z-KMa9brCIRw=g4+!17bugnk2kxg)$j%*}al?YOZGw8M=S0C!2Z@75|=)=qLSDep$_ z^vLg@c)!K9$L~^1QLF>wxz+QCuP3sX6&010MHaA6+!^Wrd1^ZPFxTX!!x~M8$x~M% z_D^?5Q>U%8w_aheGp`~ovm2;w6gufbT`AnHFVz`L2FI1=JrQsGDLAJ^a#wnuKykOS z2y{HP;lfEZxivRmGAhIO4jLfcmM02gB6WJMWF~R%3q#bDr)twzJ;~ErkHg!&y781A z;Dg$VkYU94drozE+@NS=3fAoM?$n=eO5RHqoWY)D$nK32ctm0Xogca%TXlep4@j?q zrrT3HYntzE-EO#rvXt~Km@^7C-kcXvG!V#`^EQF4g16iYX@uRP=bo89w__&s-N~L> z_>Gp4nQJ#6IZW_;x*>f3JI1a~u0Ch4ss0dIcF7p`_^O=!?ssBtlXTDPmHWT9Szbgh zqpzvX)V_`Utq}+r6h!Q2K6~>Wr{4%06+FghR$XAthQ2N2*oFVvqEhs~>c;>3rN{r!O(l)`Wx?BY%64}a;zPtHX- zdRHp+2?DT|w&X_ZYGMJ>ttnzRs7JeltI9st0iP_Ugbk8~7mxy4i`Ojiab1G4iKYA;j(TU}sjek;W8l@os?&O2a=OAI;9^xh9MX3)jpG;+6r zJyW4`*czSgPIGb{<*%dG+yCw>}`6sW`^kW3G0&syY`MX@9#OMadC7;sD8BDNYVAu@m)(iUp!lLc>A8vJLNZ< zE_PpYAHgl!#Ls7^&Hq6BpB2TlPt5C{*mPVo|KZnmaeKq3Plx+=ujIe94XIq= z_I!eSsj{X=Gd$Bu3e&1xxT3V+QQ42z$t&E{ou=C-Ct*IPrd?Ox_dRuXOdFQH&T@x| zjDh9L%dilvVQ-Q(vr91sxn48pLpfzP^pzsj$GSMT`x?vk8!{0SrM&3 z!`zu!tGbKrRap99ybgcB8a2ww-BbN~bN@i@G5#*&ZI=ovCLTrQNe$qy_r#6(f1$LE z@%kg;LO3PybdK)+TVIOa+g7R+x)FZ6LB~`)x?{OAl~Q5$JaBZA{)_gTgUZc=0wq+mzT)78yhkKJ`=NFDfNiG1xI^}scCM%j;lVPZza zv6jN0F8B95`+kS@XzPaXozdkd?TTDu>Q9|tdt4D*gpR*y&wfz_aV5kZwh=#N!m=$? zurFbytzQ{a-$@!baCg-fGECbBN}1;pt;97>b5(ag_PBX}qEPgw9|xW-N{R~!;lJe| z?VV=TV)(Pae6{m>qpRQPJ=#!^=inpVN-MRVsNRKe^^;j>WpqEo?1-FsX4#rbN$EhN zK9La9#r6@+e7C7$AP42G8Lgn%MkGmLb0brmnmJ(Qeajn&M@hb86@3?oC~#%I@)f_bW;FjBg;JtWV6P;@|hsY*bmw^$MNU_)`R$Mc#~rXFAU z6%I2#c`5-}ZZQfD1!Us@z8hcmAi2+v6Y<;*=R$fTK)m+(vf?~rHR&^losK)PZb0ed zea{8ho$1~}IPdLR?fH1igAUyr6Q%-Yyhc8q{vr&BJoD7IQT=7${Tx;D@n738?;Kum zduPsEJ9L3C%Wl&6NMrhaLf?!Y2o{8C{0LCCI-$kbV6gckU&Xzlkvf=e@0S)Rt>i3l zk6bP;Dp7*nBssbsv`#kLMz!s2aev6GLKSrLu$Wf%j5V4~s`bYOKkANRKRA~7%d}sk z@9EK$H#=`vD{fK{WuEj8bGHjzsu+YhyWvG#K>4PIy1ly$&g<#qEx?2^jtIY^1A)J} zDOKAon4jLJyh@MV-)@Z4O~Aig*Al8wY0SU=XpgQlXK(WJzVz!8?{e9;#oK$g@aq>G zj@r9kSY!|4PwXrKGjhCQ55W?NW^qJ0ytF8dYMnX00lp0|u>v}z}=XgAXJ-Nx1firz$>@ROq zXMWqa$=el59oF*x{ANVBkt+$|k8BrDl%$r^v1m0@aIPVK<$^JGY#}33s8*3Znkg{^ z-j6H9py$Bx^VdvfsZNb=m6Pi7%VC(W<1+eY(+`@K`YD?Rx`F2wc`;K<27Y9$BoFk> zOZz=7DH3ib!Rk%xu+_=i?~v!E!EQz$b9KgNlx7ahgfo%LpF-R;u*X7YL=18Lvpgag zn3a=DxNFq_TeiczKGyT6?sh9ZH>)thXj0%I1+^|poQ=ncEu8Ir*bZ~3joe3jNeh00g#NDl5zkO`2DYe4`bF}A{XaAbWqnND*5>=Zl=|@7c`vptEW+?mp0JN= z@u+kw7n~GitGH)zin7JNC{v8_M>co)nCLl_q$hQQFqs^bRJ>A0QA6K|ZiXiPZYo-N z7tWN@F{yg5KKoE(QXNhp!xv-QcO%&DdkcUHt_0L(dHXLjyH`wH>HWRJ{)B1!={nOl z`#R;vikUC!r}-_r@n;DjX$Afm#A*cs54NgqvhRoUUi$ZSl$5+5H*jX9i>rt$@W%n? zQsKp-76SXa5|G3x_4xM)%LW>K@Mm;xZ~|9BAW}PPZw!w*jnhE?CsBWP7RYS^Df`p! zuxx;y@yGl}yw>w7?LSsJPfxG7Vjy1+KPH(#fhm7;n++b3I=eauTzs?X3hHgSCO_zA zh{A6fvVAhbCZ3jrhiw4Y-<;EN?%XoR9H0|U&2#?u!?qWwFkp{}rtT8`3rvUER_w$)ZdJ?!B zo{ZE_n2d;jsL9}omYcjqs6lNk-H_PIz zlUyT?J}L`xHPtWq2ESRT(hYM;?1wKJFO_T@yr&=66K~f}unaj~VmKAT@&{o8n=1w%;zGk#`qs#$;Ts z`Xe@`L4FsBf0}W>`a5T0A5IItV&pI}Bo&BtW&{=GZ8nIX#ZEpboc6!8rGad^cA2n{ z9DC!exeap=KIs8^sfi-_GPzTmy_s%-d$C;Z+>Q*%+GO_RWE}HIk-n~3L%S~)A@&Oy zpQ|Zjr_+-7yKw@D=xjA^ihk-eGkE|;&3vBb_-}kZ^AC-PK-!;}HDC+|J~IkBeDWVl zBS!%doSDyiL|PoBApWzqq7p-k%2F6K>ZNjqE8#$+gbEM zp(<_FUk`~5sWKylOJfohMFV1iS5k}1G1+V~*WCTucaYMU{hM5G?fP1c)#B^?Gqh6@ zKPle1xpTnNT+q(WJZ%Y?_T4Q&cH4T68mRNSlMr3Bg) zsk$n^$eZg}t-HX5CP#U;LMH9bNL%?O$;BzKfwu@Rs6h>v{zz;*nS~74{-FN7B>2$Q z`hB#6W^MbunmH4G?-!Ol!`?GEr=!qpthfEA*kBX)F}XsgYSO?G@jJ1_TA-H$Q+`AG zhjsJi;PTDgYmbWyzD{(GFx>vT9{2LWwfVN2u@^ch$Jg{&e_B_BL`*HH6ig8|_DyYe zo?M%+FldC&S&Y|h)C+*`H7h8{y&17fU3l2NyN&u8fkpJHWekM&!_^8;T>LhR+^)oG zbCcw|D{jxrW5fbdjO6=eF71WX+EZS#6C%1~ZosLsN^PSC_Xl)r=PO^$$?J3<)tYWs zt+w9&%Hs3 ztG7LJtKV~4-*;dcr~*9; zAEBup3mzX7%@#0Ji9luHS?OEOIA&}pE@0AH#jXG9wy%FaVY4ms=Gghyzt`qh5MDN78Yu(*YXTydDJ&CJZ&RnYf=~h@YvCCm{Ha|9 z@Hj<0C@&?_AG@}9^iEw|+NauJ4YSrWos|ZjTF)_av!3i!1a&YJMjw$BHH$LYBYn5+ zNBUp-`**8b&zB_@L&XyzGoK`7-9|mAUepe1(QJ(iZR2{ep{Z@PLgD1J%)N61E;lsK z=)bcp>8?P$0jZcnjccqto2#AKa=?;K%PECvutsn8YgsmLG4+l-@QAavYCPPwkSB(E3n z;e$gTTdC`P&_1fZJa3cR!P{ysu@Ej((oWvBc@h#VT?P1@3cC!H9!o>UF`XVheu;Oc{-zqf? ze#GJSAoZ_P8%jA!;X{voysvFW7UnGyezPLKSVIhX9ZWyG|DqK0e3bC0v|vc-(~u=j zE3avX>DG9!2*$$roEZ$dYY7^;aZkx8Nx2iOSZnvBQ|5h{*-{!Vm@#fqm)Jf0@CN4R z#rGFzudAMQYnM;5X`G15HZPEV12%6e4XNe(Xp}w;OTVVHfl8NU@X@_B`E;^p^69wS z`gK_qSqeI#y=hOZRpfG1|p>4nN+Fj^`{nLN<{XVFxc;bHHe=k4+UUB9Y1Bfho zS$8xK;bJ*~lf*(T-pKxJEusm|5|c?8?@~*UJq9hx5quk&{C ztXOB6?^oxVTLHyI)12$I_eE2UsuvV;cTbxb_1VwrBc5@H}3EguYS0UR~^2kTaJk zmWj?bsh{#Wsj$pm&x9f$o+I12JPnP2IBH9I9hZuk=d$oeV@s7-+gcx>b_2qmXnm2X zPglDMdpG^LWyydkQrGV??*7<)=QW7*x5BB6`>G(lJ&h{Tv*aucNlA>BDH4LWzST!_WmU2W3SJ`_9;xNw%71S;A6GKKt?g>H54jY3P9J@Fe(OmfXtR!u zn;H_7?!MKxmk(EI*FeU0H~~efOW;49PhM_2M(4(J75MZqXrZl=*KlSVzpP3hHnTa6 z*JiBZtTrocer&^Q9QJKt&W&wvxclXirN&Iocug=R@=`;V@>Aj5GA5{LuP^R=YExE& z_iUuNG^)TuA{`?q=cXN8Nlf?nq*}>=D2L!J$csqTN-{_0J%NYrqx7~d++I}wl-&4m z&m}(02U!C~ertj78}_Q-Yo8io`t*XyckRPRP436VR?M~!pKxLl7ybz9{2d~~*}yd= zd9!S*9e6?42eEx>fpWo>RmwY-X~!dQ0Z$_bvuieOLZ6Z7KRX-f0k4e_nXD*EZW>IQ zR~8trt^|-i@=zL^*6dz|jw42qu7x)ptGm2#TAIx3psw9+a_EKL$9L!7GwZ#vy@0}w zTUO3k1;M|suFW1_0R~%FSb%F}2m@;ZP@9zE`YI*dK=W@GFq5nD6(B`s1e6s|`@?`c zrFMz+zv;676Rj0toeW|E1(-WI*kASECHt$x0fsp1x}4iK!qMwT2kXYS%=1_^HQuVElX=$(_qOd z+}PUw;F?)%UEd;bV6lkp+38wJSARtJ6oW6Iv#wr)H3-pGmlDr;D3+nh=6NTZG??D$ zqLkg*J-vLbQGDitGPgu2jdvmC=T|#J@rkyVEt^fjoo;wg-(0V#dY@m27XPcPN)gCG z?|K8|%U^l#SDdf^p{Wf#*!Q|9v+XDTtNo_pSpUsH+Q1PHEtv2acg@HB^;U7}0J!!quhB*@g)QP|PF)Gj zP+ie){)mA!BIM@54Cv5?X7YXka!30zKQ;N0C5&)RML?xm$J5%hb@r7Dn;GmavVnfK zoTG%x5G;HqK?r5IX=CZD@e?UFsq|^s%&~Ynr(lYb&A;KiT$Rp!T4+7u z_xt=rf%Ymld$%0r&<6acHP6M$-deD#Y^6n4*;q20q3p1D&xyLaFvJVE;S8) z4`uNKF|4Vzx;*df1HwGlwW4X zN!@bboggxQ8hzX!YWV^B$jT??z=0v_gHe}_x(|D-iuwSg4J#L2axMRf@0M|A?AR0B z1J^@&VnU2SU69380AImpH42{xeb1S0=C_kq{|ccNq}rzUa=84=c^eC~;lOPTj zc^V6%H+z~>SN8rk^|aSk8~WV0JkO0jav8tuJ(Bs@kT9XTO02Y|lE^Qk9VS#Z6U)*v zM!dEsGJH9>SE6md(u87mH6WgwAvKj6u^*2RV82^iN*@o@1tQ9SjnGMUFs&>L2Jd3` zHs5^|r_@3(ae}7Ru>29^GC~5I z`Xu%0iZjX3r}r#@Ya>1_^xZuBcGE3UZajVWHauZg&?9K7{k#aoxLLJhC4iuu-U;^- zsoX&8(cohjH3@HHjzAJL+*}1`n?;Dy)`zajl?Own9LSJ zw8+5JmIoS_U@WJ3`@I72>l)PCuY;-C_zg&rB-~?CZ$IZPg$?|Yjn&rwTZYwpNLSp3 zTmHJBlnHjRcWSd?<_q$Bvb9^w7PY>d2fapU|1|l?Ff+ptHkHG83uvihgwX`b-cV}C z`<70Zmpejoaa?a_d+D$vpUqP0S0;@=mTK8nK31)8z#SK`eDvdiI?%S=i9_XgO!Cg9 z$shi-4omeMBC;8_WjSi(@Uk-nb46N4R$cc3E|&)MYWX5Sj!aq~b`EH&aHoC7H3YOd z8EK!#$Qv0M>-s{~xquPwMPM+EE4$km4#ZjVR!>)f35g(;vMOtKV2FkGe&OHooBzA* z>HQspi2qnY!G9Qks0R?-z}e+EwNXjI7*o`<=-|RtP=jVxfq)A8*m|gYPjF3MK?dqN z#C(;JqniZbIO=#CR*}#{O>kiKL@~IhSmvqm!KU+%xt1Tde*8o80sz}F zKVEY@S#x^XbGiIa#_!$k+3?L*I1{h-~V*!DBkOd-T{eh%AuiKmK;n1BV`iW2ep=&PjED;5TabIY^|j1^+}o4eVU~f?Cm# zjL=VZCb1VKw_9Ut7P2>NfV-2lh?@rDMr`X{S`)?KAr7AV>G$s4T~a$DluUo|PVK+j zJC2EV(OX%;WMmsWvQ{dY?9k*@jG4JQLAsoDuPOe)T-0|^k$a&+E-RPA zJIHOwv63-gNs%zp-WiW)Q_;#Pe^fp6#H)TN`4QO)b9RvCj|^`(CAOMBFR{aPa&dd5 z4=c*zi|IY5@xH&s+owFuOrB#Doc!=o7M`cZBdl;n#lAiNa>jt^k+al8ZA}>;l~Uyu zZ^sSst=yly34OGOL}ADKTiCE}>!&IN%a%CEWo4Yuwbbmf zt8Szht`qD0_aA`5kLAtV9fX%XDzD}JM8=4^g$Srm$tlcLRPI?3ptox-rektqB^Iac z2eNC1mh&tj?zro9&8Q4k5K~aqyvFm;CVi#YThjyu*avx-{yR9Y6W2o9TMD*Zq!|7C zX|;huFljr*B!>sxgi$5_wQQ(Re$`y30F`#w3I3A0HdjNpBaRv3BeDrChu7Y43qsI6NIDrkLkU*N29QMd~9osfBXX>-ybf0r2lDBf zJC4kCW)tYBxbVL_DsIeBg#KfNn*HjcaD5pt`HJE#5s(fWm-gTl?kQwdG$* z_BV)2_gOUus#Yrmq^ni%Ov4|!*wDk08HiXjj&&>ff=gnG^qRXLiqtTAmfMA9U7vwVU_OJ{VxQ_}lQ=VwTA!=xsPtKt2>h-&R$nC1 z24m7`MGjF~YFcRT(&Q0bwDKxK_s=IjqpjH4lp2mXu1@0~fKk#PqB}D{H0%6-a|~mD z=4bYYA&|!5OS^ELF&Ifi?gH0k9-f-@;;E5YO?tcF8GgDS2oDbJorvuGCa~WNo&qU9 zm6{=hnYO2%MK=$`qV&pU0#P7>`mqwAoLmq*9Q^#QZ5`2)IOZcwop3hwe0iaLAmLft zJNerPEAUtFZ|RpL=tcI&M5oO0o;Y@Z_@4xsQ*{qcubMiNAwP&? zpFYZ<#x&&6IFEN!S_`<()w_;pvnqy4C(H9Aszfcvk(rqW`wjm7S*{ZSpr9pFv{b2p z{)Tm#@fou3s2f+=Nj!Tn8O8;|Aqcel_N1bNOy356CeTl-gJcrKdXpf|gNi#+`Rw{u z{UOwjtx_Y*>MWdOH)S<%oH)rCWzi&O{m5b$>EM4@aqFRZ@da3|H!V&JcQwo+by~5K zdS&t)iBlCTe6NzjGN)<|Jt{abHc77k5QPH+8%*g6+fU*pSHOr>AO$PQ7<2n9JKy36hP zANB8+VWdlAx#*0HonTV9)u_ed>0^7*Q2rT8#HG&E1D)m!wg_$3@NxWE-s3y5*wzB{ zF7kl|r0A$Rtb^s4V1{DInrnH?$4ZXgnd}YJ=Lg`GYZ77!|pi3~)xhf3| zssWmk+`JfEIad=6P?9^%b+Z7W1pJJYQd4#}KuD`(#Qy&V6)S28$`7cSQ_nO0acNIo zS8kku)Mq0Mw1f`(?+8yo#Rz*@4S5RTgmnIFK3a3L&%GP{$*u7NaUa;UV^uv06y73M zq%PS87Cl>Om%NX3zInYmW4bDBScc~lNLiVb_6}7qNbia|B-yAj?e89GgnpceIV4Jx ztrDi2wjJ%K=QgH3`90@&kkT68cU}saf1I-i4Sv_J>&g=C@E|1+IdycBt?V3g;;xl5 z8in~~{35=-Bx{$j?)+Y(Jq%Q?cSyE3jaJITSJiI%^(%|f7Sk(PMiG?GkTuz)Qz(V; z*f=Ho*sncZtshb*ohHgem z``R`}*(BZXV&={54Is#yh!nkTz3b`d7nfJhI-7pi^4q)d$-YIWI4I3Y25eIdd7@@@ zgGOtd?V*f8g+jo}l$o3Fo3rLg$upK04Nj`R>Dx`rmo?aOs(v&<6py(0om2Vb-fo_= z;Uf7=6uf8}klDg6r*y=raVFP7ZTYkki#Zc|G@BM~VlCW3mW_3^ZieNL&Eeh}O>(y8C&0~Y407bN_m`9=hgCkZ*m+CNbb~!`smgL6 zjK%qGiy|uMYXC%#I~vHRI8zj^D_WqC`Z6hV&Qj{rxHQ@f_BwQZm&{mpJje-?{ioA1 zlB>Ds9Z#yqutFQIEX%n8=mRb#%=w`eI;n{tohRFhJW|2Nz6PO|N=qCaU$mEgO>qwa zZvwfUDs{a|)rAa^;elP2=8`Nav1<%mB{d~H4{$5V)PLZFtXS^%k*0~0I`Q#H@gw$3 znPi4QIQXZf-uqpf4O@H}uu4>y*T9BU%~B^Yw-=+xh`>!!vvovsr|kvjavU;%=uIv0 z#V%h(st6^RN0&4H-L=k1Lmgib)PLFgP3V@)0P(86Ug)gQhi6us!WPYHoO$~O)1#qM z-dq+s0FoYhonH-@nWSV}^75-&-&v70-~RbYI~_F2=9xAMp3dG}-Vn2VidXtP+-{1w z=3rJ`8kh+Q{HTYvw^m&lu|kK68QWbgp>vRs8OjcE1$Z3%S-$*g7TUOyA%I}_@1tHi z0R>EW0>A4Nkl$o}1<#g!Y72N84bbkg;S|IbI)GnuH>B=A>RO|8xW_`jw4V&d|1a29 zItX_yCZL6Fstg0pf1w*Idu7{HKuSqldWj$n6&e3v38j{SlpN5G15lxauFE>s<)5vb zhz*A|`G>RmrT5ejF`p5WIMbr)3vT%yYKXQ_OjQwhQ79M*Q3i+cU0R~|)SUm+1W@{79N--|uE@!++2_c4S)P`|bLP#J?_?{7-q(=VkMHge_WI}r*Cfi$*<3nn zt+6`UzH33`(c2)&x`Yk(RdI8UPcOmerD_LLuP1#3*J2AEh}`^)Ol@j^`CH~i@=~iR zcy9G_DU!1BSxokFMALEln0yGEVfDvmwWtg`^3tMT3}**)*T^+vDqsXH+ZJb1zFGa# zJ|xSRkrtmbl$QbX^UFySsSyR|LI?Mg1)HystrLF`nfzpm?Fzr8!~##ci(G)XhqqTG zPls@xe7&KvR?_LE;?zWrW39@aEPkPH$%NxEpOC3vEa#U|h^1hNv*{5E zwk)EW#tVJj+p5*mY1S21mpC~*Ui<7%v?G@ucbcxnoe%o>s8A)_mX98GgvtfMvce_L zyu=H>z70yZ{}rgEIp2xr626XxdE;EkH~hzq8kb5n6Vb6t5CRS?wJK*hCrrjtR8?Ae&EV-+fCWO7-{bURC6W(vIh8i zasiU$e>uHqQ?r|*ak8{dr;ltvs4}U05y+%67KA?mU=?it1FI0eqxbuC>3{Jn~`ZS5T|LeQUYl8uMNT9;ae^0xG5pc(>e-uaX@JMoZ%o9EV6 z8PFY>9%{xj50a}+boRomQyd%&!ybwE4Ja6p6KkIjpn4n~aXFnY%VV7O$D94CRRpK^ zHiGZCm=*1fVF#$4?d&Vk-SBHlf2X(Y561KGO!8zECS9gm9wyhZ1=-OWld(#?p@F5j zNys(qp-?=JplHnTn_AjMf5)7!RF+-L=Q&du)w4?_WA_|AW=q&Sd~F5o_zXM_V|=w8 zQpie^`-T{1Jek9WE?54+h)0$Y+USov?t|c$j7}fpZ-3; z%K4)f@(rF$&~Rns0crMKNt0I-ENjFn7%K5WZ+R{&ZDgv#8eLc|-rYa@t|uDNlJE8Q zJFHT!GPdnO_oSDw$)8+1UK^vB-~rtcUQ%P{o5;`bc{F+$)PJ7X%2xCCJa;Abqoi7T zReQ!rp6vyZ!g4WW(V=ubsG6hS=nSfFC&kfyaiRW>++4wYi9+}mw*DWQw#jY*6Z4es zyu!rxUiuMfgYm>h+04^j z{Yr#7qdys;`|CuHJv3~3!D({WF=X}EK>G~tR)j$W6f_i5Mzpq`6Vl8>9-BWP#EpJ4 z7~Q%$Aogo=YSL=o6m)2^a66pn{qhAAFl%+s0&=y21pC!U_p1$T;MaM8j1Pe8Q-e7o|B`T5!z&S9din8-uJDa#YMFvVPsq$=F`T1ko+>@&Y7G?ELGy*{P zxR1tAI6-|_eR3#56AEx;N1=wVu~jjfxG@Nu1eKlz%u4z~$-UzKq?2E0IAfQUo z%M$QJX5Ofs`@80*?p!E)SuOVXj)9bug|}Kiu4*y`@+-_>PC=6b!dI8Rh+hCQR+4GY3j9#69#l# z;{c`F!y_4G?$J~-u5?oaS)3R|4b8qt=%+Hg_+^#;YJl}4y6n@)wo$JBG*)(g*&}_J z?U59u-0{iR0qb9G8SCozCR@U+O9D4L=j+!C5(gYL8l8a-%PH>I|B@q}3dk@pKr{fD zCOteDfz5Tw|2%a?jnf0*&?q1W^dC&jSIIwq0VOo?|LKVe_|Sk8ia`N$vbi;j((YwZ zC3rT#-j^kN5r;}^Mwm9E)p8%1qspcNfdJ6`=XBMs@?$n3=Rm;0EHA(F)naZRy5jsYO(54q zVL>RM{V(DT3&%*{#^D-o_-;^mkSEGCJjv-)`~x!6g-ba(Y-Pg0h^l9rJBSK8g0!z8 zhDOmu{b`f3(%FE=>M|BX&-q>$(<(aE~#ye(PCf5QRpRs@8+&xq^e+<e z(pt5@I0_?Gu0#`$y(K-%@KlPI4=T&>G%jaPI&Tvv;K6OK^pn+sihV6SD(#Qa9iihI zy!p~V;KJ9*bCp|I=8fAm9~TuU1_={>Dgt21ZL^f2Tn2E;i1~Hi z%oVq@CEHz#u-uj&4;oDw)bgTT933A0v`4d{vAD8UG2-KMv>T*a6}H48Gcf$w;`R-G z%UacRtgA^p=*5JxX+K+1@SK>_|F&PF^NOS3R`t^83)cRvbLGh>Y|l+do@PYQ^p41Y@)AH2sjt1 zGGodXM9_WYl5&HDF92q(Y$$usI5JIPYy`jv>B53jD9UF(~1ncD0c`svnbH;{1>Ayu$Q z2Q`Z=Y73rQ+?*uYCvSi=kBI4om^Jm>ZjT&MCZ`ex3n&ADMf&ug8>OpVQlSFioZe8f z1M$9ix6m8CtahI|-fG*O>6cj^-!s>TotWL~@)R&VpS{L*z7Bur*$`H#tFq-6ZU3X~ zV+9g?1Cw70<+2v;k>xrS18|vNDf=(b{24%h_%t^U;KKu)aV0n}@BkF&>OOdW$^x~0 z^4}=XJ9^=dPs6!){})H5Cfr~Z0r2sph&OHl#1frblkup#c#?$uBgs_Km45IGMAa{= z;$i``O=`^mVv=NII?dRn?;Jq1V97whYv}$sZZbMdC;nEap>@LjRQA(;9BpwNxb?1$6Q8b9WvL(@; zB+dYRn<>+9qwZdbp=eQJD{T^+$?TejSm-T!Tr8B%I00i4nVj16+uv}ob_b@LP|EwSI z;cjUzo-d-bz6YBnqipCueGUfIa-Gwap-t;p{??1cWTD&|vbw=Tbe_ia0%DX+dKQb3 zqNHBOoC9es(wRi(kYTf;po04al7#2ifgf?6$obX8atzXG+9p$W*WW_XX~dBCnaWuv z*M|D7_jM$9Z2L$;Ba)v_sbUJ|Yc`@NCH#TVePH2J9w>9c8SZhz4a+Jk0}{A>Zv^u2 zg3`Cys{FBH}nIVOXQ*ZkldWin{tFL5bfA~D3@jIa@48aw-wQc0+yLC#U z0#tq30E6EUkiY@VY8pTlM`!xk_!h7kDkG%<=`r9nRfNhHKiWU|d)w~+f_95!KE6KP zT>ry@*oN_e^gjTPv6@*aK>nu$60r&tdn~PJuxFD>5=WX&yQyVSdr3y;k(|xeFrd-Q zcF#frHreP%?=~<*^1RoB=_igKZh!JXRoWn%?404l4@+FJ@0xdHjK~yd#gBWYHy>ZM z84JfAI7lMVQs4HQb8fhWn6?H#3mqB$7-g}T4lZN&P`;P=TNQuy7<9;_(O~f@9AS7c zG?;D`&0RH(Ht)1bSFwCkS%r6w6qNQL!kOrOasJx&ykSi27d==Ntzk58qoP^9gGj5D z%s4Trb1BxUnmehPk!Czv&7S0I|Il3QJ%m{Bw`lB$#^K+6e^a$~6JmHn2i27t^w|3A z=98viNi(ml^O&8V)^uN=K2h&JUi(!8eKg(oT~)Ncy)1dcprl;M0!stjcL9ns^%P|3oE)_=)*E+roJNUhx!yCGeD zuff`Sowg)aR|{(Ap2pAwokAbWy^%VnX!>{|`WmdXvuw3ye$hwWuIs+IOXfrLx7&{x zHC7wn75xC={MuYFrGD>N+wZBQ9}b}AnSNfX?k`l$`&jv{64fRowfFF`q78L_So8pY zm;G?e(%O5kyp7SHrs0?b)-_?f=FRh9%J4$&*28(VeeLkyy*T((Z7=MscMdSAz94v; zc^uhLv1qCIbJ1tjqh?U&?fSUu*{|qi*QCCY_4S{x}bsK<=s@ZG>aF$=A343BWjB>dq<$Ajdy(c6VU`M zctzfuI{6=^LL(lM?Vz2@A#|;%g9nK&K}A0AmqIP5vPBou8lOvVMpqGlzel9iHl^Km zngJyV26I|U4?6>JC_49K0K}}+t4+@Uz=cQTfpKJD+p=Xh6%lkwDz2dyzV=uAbgN_8 z31(5W$^KCx-{$m&p_*-U%*~V0?eYDJJNi`~`=%-B-Mp_IEx*3&D>Fd$AlF$iN80{B zW?0eTS)QdMsOMMXEuf%`gnxyxr>YzpE;z5EF+Gj}pv|$30Q_=GwV#S>TfY01(u6gwPqn$6uqqu z*>KmXRT7qu1_5aeiLmS>_T&uIc(N2IhjAIyzbLoH zZpi0=9|_XDdy?-#uRrV>?D-_*_%)PzziQrelzjQi@wMOk9#XS@N%_vF__@c`jUm$sITszxf;O5x;j}E_o+>Ixz&(s;{o~ z^oQ`x{13LyV~goC1g8D8$nQM$uVRXI8S3tj_-f7#b|eA*T5bz6PkA(u)pf?uD6n#u z0|5UsCL?Nqi(=c8ML~nN!s!8=a2sdn$T4it@+F z{ymnFdkppd(y9U!yX;{!DiFPuIPgHFXZeMS8tRsWB*T*Oq*w=)_t6zvcA;$N>c^$6 zuJSy)2s!?%71v-%BVyJKOun$e6Acr?_NJe_o^q?;4-ZRyZH~YF{Yo25Ou25)Ldj;m z{@HJF+%~~0+VqBc8bcM&8zEV}#>f!J^A|e-9iL_Fr)^V1r(?nI3V$!>WM{^`VZgsX z2&^hnAWVD0xAI3+CHv7C&#-6nG_WHSb@!RZ@wQKDPqlLRjP+@WN-xbAF)k2e4*|bE z2b+4gRpQ`D)w|8oOw|1KhF-XH`v7gx1CjFdfwR*OZeZg)7)hQ|5;gZ>t&PuAoqs+k z5X|Wgzvb|g`++_~M?3E7bu-w8hR3t~ZbyDNCGy+-)kiVi4TJ z6h&{NV?O3`p&dN6ZB8=nlG3b(=4l;y_NXBdt;-h%Ov~xKEXSx-zA>5J8ihwLcYi*;ui(rH|6e1=X86a!t^wqVNf z#O(!_lYvhxHRsAUoEG!2f@^PzW&?DE-|`ZUj*F%mJj|R6Vq!zD)14hhDS5>$VoO)` z{-H_qTqEpSuXvXr?<6d!dHg&u^ilQlP5Rz}!2d&oK%CjENp&H6rTWikoi%GUe-%Hi zQy0S69KQBdZ-u?FF=ZRAv7qd>q5O6_sj>>@YkJD~q zeT@n!9JjdhHq*xqxnd@Hc|#{zRgNmLtRlBP5cR4R8k2@6?n=#v7uVNWkY4$lb?r+6 z_uqz2s#Os|z?MXJjxtYC*!{7k*DlgwAE}pQpjPc~NR%7^HpDTxkm`yRf19~>!5o!Z z_)PvJHhowhR+4!yj+^5MuQaC#4d_4SdwTEad8&y{$hDuUqRrN}579F1Hz(#>DrVGF z_pL~Jjd+&KA_5E@Y~lugkAyT6!IZPoqsvD zS2@&DTwLq>+2rX5DjAg0T=)eIgdM!_8eeKAoK2_t%R|Fm0|~fchS3FWUxe6$f-5Vb z$QFI|t|o~PMf)#_0JFAiRFi0d6V;$PHvbxtTe<&(^`hgh1ejv{jl69suC21RjMF8D zT=QS9Jidk!!shz$Ps*BoVWie9#n19~31M{&i=`E<03$1)c|u-NF-v>$O=RxLEbLz+#qJ$wIN*86xCC57E&hv&je4uj zsS^fZ>{Xb`oq!H(JpkSY*Dl8v(d(7rb;vBsnNvCRre$S1!0y!y2U6yE$7x)|Hoo#XKpI*^^#m_HV;(raB_|Z7Ggj)u*^5V1OW!M*fe6P7H z)5uz1Aw@L^g?pLi*E1GPeqOW8``@(I!`x!iKj?V%kxFJnzmF{+_cKnZ?^pmd1 zPdjU&>!16z>MN?>iOJ#)_w`Y~F7oO*=-ZRt_P=9mVGN^+5;8kSs^*@#_&)#FDXBnx z^EOZkqM@Ns0?Pp{Bvz_~j;CZ_hbvwrk{zoXZMthhrqeyQiPgdLdMLe1;=#n}z9H=J z#GR?|zI2N(8DaQi#P}Q|mhi}z$@jJ7Jl1$SDcKQcYvPBk89Szw=~;YsU}+ss*A^38 zw3d0CFZXR^2-3{`@?{!_quhCDwz$Wo2S)=tri#Wlz7<}LxejvvLtDM4!G;H=iQY}W zc@wqsG!ZW5SOB6t7gv^XK(d)siQ|Hbc3u1YsaRpPN{>4X{IlaQj@54x-~4Xd#! zeFT~Ai5F2j;O3@1uOx(Mul z?HdrpLa_>$s8gKha?<$kpVi;7*YZ=eR6`gD`7|RU#&rHxH4~%M%AF-II!4mXVU~6= z|1UP5wXA8&ZA~MNm}@+Gi%tFBUFb)$HXY)NWmw84p{5tdD)rgoZD7A)*=)fkU9(|;b^?cGk2iA2 zw&Z00d!t3i!eS3nHS&pGZODz`)Ok}45>KVQAlhJAGU^RamIlHwqdc7$2y&kA6#-t3%4dXKKo);|jaDJPcmE~cnN{e-RgA_}?^}iIhppQ%2dH?2 z{*lQaC`L;n6`I_~1BGt;O&a(}hS(A1C3j7@l13w%j`(Nro-#s(Z%yHTkM+f2;Ok&0 z0f<-GJR&Wp%+E&pnpzlionqLr7dc?7+(m#YZqlkvQN1CEu=al z^t@(wQnPDSr|GGP0>tCy(xe_ce*dril?lIdxO&;p2BFBoH`u?RXM_9qlD4>WJa-4} z0;C>mL-t?gNaoa{ZhpUK`EzC#6?TGs{4$yfH2)RX_4xtdpsF$qkR=Bm+^5b!3qAJ| z+B6aur}nvr9&H0KON@`N_Fi0rY9*(s?8xm=>58{-gIlwZvlJ&oiJVQR-x>(Mq^DL- z!BckE<|hsCfdQu4D@e^ZW1(G&(b#uwk`1J7vS@A~P8P=blAwYf#oo!|oF^?Rx>6nb zB6&Z;eq}JVU)SF>%vGh!u@@e$TkO)_LDp8?JMu~I* z|4{WN;84Bc-}u-|meMGMvKyi>vhRj5))`yIQkF0y!emcKp|NCN#?IJ{vSmw>HCbl{ zWy>ymq7Bvm===Wu@9({?T-RKuGo#L&^E~(cxj)PH@)7ab=%n=SQ#K#M#&O7!ij>1C zQuWi@uhNAG^SUb;DMQNV9z1>T({~)nezo zR)*w;*5T#d&J+23%}zRSMv0cV6>Z%rh~zWGt?LQju+s)Xl;!4VpU-EUviyoStX?(N z*Zl>BGR}wbU_G1_^bCdH)4JyNjO*pM%LwZa^vk9?iKh$izNA^KeRbKt z*eqG$t{0zV0;g^1jPW)X$?zhis7^Y)#UF-^7!*BA+s#&JvkN;~!?Hfk*&z z#`*)2guTS_9DLd?Orn>oLa{N_5c$ZKH{>VI%)nkm`w5#8pNEjy3&OPb17Bc*r$y2c z%@uwj86XL7CfP4#rYmyp-cjTPngC~sx2{>q{j zkv6qcTk1)d*jT}x6D0(FoMukBJol#QRDfjA`RM^6D#uj8ph)a;7F4?t~F{vPF3 zsZ%Qe%_{XO;Ax0h68Z+0{AE)tmnqg_$D{Os0OApjx&uQb__{G9(x<9{94wb6Ix z)p!HKp(2dCbLumhX5wGtQ>48t?sF|c&QIya3K2!X>@sy`MuUdZ>;U?B& z`J)V1xxgm6yEEm@oc*!X=cQkhlXLtT!=rce)toL5yjFF7X`vqVy77zdT-f7=A0x+g zobjGxL*`2k(J(#tqIytK98xn9FW$pd03@9DfWXnOHg%eGFu?Y_UT16NzwTP4+})cK6eT&&W5ta*jUhXQQ;{~Xv8!_ zyJYZEr@w}4x-Rtx)LF&G$6TV4<=_Q1XKXF5n^yNU9o z1seN#*vvcksjJsPPvT;@MN)@K$P$U8LFF$9*DO}-M>_w)*N>k!1)n|i_2P<#Or_TM zedH_ROM@zu(rR*mRBV=lOD%M_;kCYKIa675mC7wZ;Yc-O^)GqT6$SS4c9URThjnlRz= z{>s}Xk~78|+w`Nr?mX>tjQ~=B|K3&c^1i`)cGg+=iki63k6my)@$uuUagOmcx0_d6 zexK<=$46Dsr-+NsA^?kz+Ubb*ENitDw3Tci{twlTPg4b#X3FRX%Nu?q7=KVPwi@4d zZEgrmJj>-QKPm@cx4mE0I?iX|>CfiIyC%@#10KN@pdb^9lB=7y0I@ZCebqnA)w%Dmpn0b%M=TlHFb%tBm&yXMgv9 zPOHNV+N%;z+OJ&J^AeFaE`A0s=YxIvbE?cM8V3Hm)KUJ zU}4)rl(I;q8U!F)3uku?TPwR(M1X|(SU+*>i&}B0YgQZHn0g*qq>Q9sG9OmeRe@}Q z_&Y%6sQ7=GqwS&Wdz*iBJ+2*{*8dmxm&f4Dmtd#y2L7lNi7Wu@AB{fassHRB@o7M< z)XB_%C?XUS(qIOY#>_<}NMqy>V2L?x6ilr$LZTjV8lE3@O;K?cZVch(`8{l8wGo7D ziEgRBQz2b;1fwReRV-#UPI7&%y37|}vuWk)Jdvkt4%pJWN$^T&83F!LsnK-!ByX$? znaD(&$U$Q=0v%sBggfu|fG7s5IRNR=CI-kf7;oXsINixvc<((kGt|~QzAa?J%Rtzv z;nvQ%JP(yWXkN1q!)`V)C7ny8J3U?2I*VLk_n+G>yRDK0aqoRGZ9Io53y=SVJ-R#m zaAWbcZFbso;%he3)t|ERBePDD%Cl_Pd=9mjO)Iu{Tq~r=vugk-u8>rT-7!!^HH0&J&A=i zJ$^}xx%Q?y=I?`q$?4Y`c#2o!7At0qp%Ligqx<-xo=h_le@6Wtd%k zt`U;u^oA*9EZd&F=+*$clR4hKHd3jn<7xkT+m&&?3*@Ok-)=YOE;5B%xJ;A)i6zoo zyY$zq3@%+xis{7$wb*A1bdzf-gdNSM3w4Hh@t3Tf3j(!-UOyem>5t4E>@E9}AGO=i zFQg7tWjGM6FuA9TPQ=;ch1)k!teuBMIp+H`0GePi9AAQLiGmlBu0tP*+J_ zd+BBr$iQ0hfOD-yasy@cVN7WYOIpa)=86hht<{g)>Pkz7So7@x)(ajt0^YT9YkgO9 zPa`+K?_%W|DKWoLTOEbM;slez-$dF>JuVf$bS*(r-B0TL`B=r>OFs#QC@Ip-wVyHa zxeGSet|jf1%tnxlo^@Ggw6@LE$H)DmX<=;Vryvq8$v+jc6vn-9YAX@f&P}g2zOR1& zXPIgfzfkB0p9rk4QpECK`G zFn79E0h@8(*}BYfXb24vx-^kBR}DIuC|)4~z$qFcUP@O^9Fh(!u)dILu8#ny7!uM` zgkZ~Q+gTyDA;KZM*6Sm$B>I-Md-D4auD#R@WeqWn@VnFU0?z1Lv3t|Wx;1MZwoid? zzsKoq#6mB~4>yXXrZ$Rh#NA1N$Po%ZWcx{PL_=Fjp9~nc#*edY_-_;{rZ5)!=>nNi zP-)~*dHH*!k&@Bn|EJjb!T_ueeTF`x+avIC05 z;c|gD7I?t1+f1=sdA)$P1lc1GgNrNAE9uKyTjg(kw+(sMw_u%o!!ga=Zl}>_b@tco z+}bY?igi-Daz|ak@!-nlHz2k}Ld)J--SV!QF+Wzve({lllKCcc7}^4z&!XnqKw z)!-QnOLu522_K70@VGJ0`mrz-e@LWVGQF)FVRv(&Uu#3L=FjyyZHoi>e(Z!E`Ag0N zQhynvGty6@U)^YORD-VU)55#rk=fv1zj~dYd;L7G?)lvD;L6x77vJ&EA*R9NyD%p> z(1!t17l3pD$X^(`RVnb*RI(Yos}Jo6qCTeHm9u#K`&8yRGx+;a>L9%GD+oItbC)Dh zID|^QTZPgym|v_I;5O**C!N30j>j3j0?HrzOY$Yx^~6+p23liq4WljQ($WrkQlHEd z(w^cPWR~RX?;~=o)VgC+2y-lB!h>xE*&ZUXLMtXRqZ6bzvNt}A*7Jw^m5Dum1@q6I zAKUq8dL^@;F^AYjXhZi3*Bc1Df4Q`2mM&ivvJm>P^UTcpMV~T%84jkUo81P($SCy& z$d%ZHey%;UdJgPRW7V8dt)7L-{h1G(tUa&Nzt-J2ZT?8_)JQE!z#~oN!{a6(Q?GZY zVTJ}R#8?AoRg~9d6_>Ymm1cM!l6CG;`HV)ZK6(>+k9p?j(^L4N?accD^pHdA%WeTu zX@j`F_2`RT~($UfUX&dTRA^hVWf49;l6P!b| zX!3|K0%WO^(uX#_*qwxhpP}LE)pCrR_%tOr$2f0r@LYmqGqiz@3lM&)T>$b{EcdG6 z7go1$)jQyHo2vJ-N2w`)%z4LgLza%`t*J4qNf?Ui7kBpM5tEa#i=li^Pur!Y`&azc zJa~?J!@OG3C9`)$<&TPFI%_VW#!D~W6?~Ppw4Gu*)@87xn^gkf^8QW3QZ>BgTS1(z zT|EJ3OV>W;e>e7_*fqqrWTxz3^04#4lf_%k?_HmGY-DYq?{$YhTbW^b?o8N*&YDoV z&z&Nhgi5Z3T|b<(UI-0^HUY-M!c3Em-V<*+L%z(++V}Q#pfN9|YB+Jb*4IIcwGG$K zqBn0CTtz+l`77hKhl{UL4Nk64T`kR1%w#DE^2c(%j1+oik37@yq%c!j+eGP@^}*LL z^d@old~gWcw(+i^v3y}2Umn)he`@>`bmi`Kd65j;iQ)>HnQ$PW@9b&g5@_0Ouo)1T z>rN~=Y1d%1e1=omh5*yT{@&xpj}6%o!Xfs-vU!Uw0@e-d{zf$4iVPI%o7hoPak@_!L z`x`?j)83~)x*HT@+ei<8?TzmK;;Y5bg6-sn1;|y{O&1=xueFuxah(3{T)W4-(X`*0 z{&owVeS4?=kF%}YziuvD`+Px#UF)vv$$yn*$+sBvEv-6~OQQq`F{w8AkCj;z!5~lq zO$E)6`BNLn*d)R8$YwQ|L6Y%Zr+D4*!?f9KK8dzQn*gL`$sJMBIk%ZA0k5D+RKMmb z3Gs8?qSR-|y8nD_zOPOQPhZQNTx1fuiC*5t?pJIM4&Ci3U206>-8XlC$erJ!~by7eOUo%t4ONQV-&q}*WF=Nx+h2F9=KuoLXLLt%9liL>lr8lFfw`k&{>L5;% zrS{mvnd>j;8!*mO1>Ja%$#4G2eMn#G$@1k*Pgm1Q!lu;yQ0x@ar%VOj zPV-Djp0eJ0E52qePStrpEai$rT1Vs4_+S%B-5{3>yDE=iPHROGLq=E9Cl(%byIeUC z;(A_s_x`02kNfG1WYHX!3+JBS8fYHja<{SuPKT{1s-9(%Pwo+u9Vmr@cl`{piSkqZPbeh3MHDr>( zPBP_{4HlADw&&da@-RAbxly-spGkN!Ja3%hq5QbXpGDX=v?`MiVjlUZ9{T9Tf&q+? zL=Xr;0s%VI^jPng)?gye?CmFc6Do@%&9>o&ju$nmX*%X9liV=z+`xn`wKTMcla2DiXsCI>Lz+qT-$mcK8NuvB^k-sKQotL&{H)(Hw zN!jdw&ckZLajsuJrTAhk(T8uO!1G(E`qsnM9Z~g&TB|O%q_`045t%vTi!*7_qZ7Fu zPrs>{(FdSt{(RG~P5wxz+;eMP1J92yO?*1*chj54Z0UC);bY{Y?H7@kIvc?#pYm27 z2yAP_U>4^6;uu!%eAF*h*4`hEZmH(>^V|A1l)nEvQ>lU%_&*9%V#n+J$D7MtB@e8VMjdr$5d34vG zd`~FjG!EXg7uh}zc6)T3o)}u>m}aKT%+XfsCvl{>&+k)?aB1@jWa32SY`4S)6HLa( zJ%m;_3_BHJW zlYA6(@J)B;&zadvyQx!a`G_djB$va9ylV}sNf+0&2+tO-_Of5nEqkOzZ}mEFQD>z| zB2YeSAycF7kH~cSCCg9vERFjsJ-q{K6rmiPNvW7PdwCi4zS2Y|z|^JDDoQ*kFhW4Y z3nESf0e8@ueT&E4j)eF+2DoX7_7+UmKIRiWuE!!>b&atssadJ=IR!906P>sqQ4@ts%ZZ%AH5u>jeVmc;icYk)Tp2{m|r zEvi_2n#FBp1m!^v??dc=qW_uom2ij5uIuRV8j;(k@RvxyHxv_R30Hjg5SN;Nb7*{tkzhslJIe zmPz(Kld$?ggB9Mo{_Cu+jZkCJY~A31*tG}zh>T0k@@?GZWA`0>5U@buwc4BJ0v4PR4fIZRf||Ng*xD@ z@EJ5xs}k|x-oZwbk#G;$XtvZ+f--6p%^OW`**SnG0v1Fu%P4Wc%xB4X*MiO%Tt)(S zi-67CK>#(}r532_hG;M3DVDEmmHAxs6#lwQPx#z<@ny)(eE@mvkgq743-W4Fr6WI; z+-Oo3QW4A@Z?he+I2hq%@Z)*xJvY5;M(1~EE&CZc&_vH;Fr!o>0)Y8FJ{U|(G4xEP zzdJB=8Y2X-IWF|CK`mUyV12_!`MSqA?k~m}zaN|b1;O4q&=%u=>?VYL6$TgqFpUGd zw0T2^2mUPA-l$rH`xpT%Hty+j8xlxzF{cFCb&tR31anff$UP?<$FaeL!rtPUmDBMA z!OUVnLhVvH>^|N}eevCUd1?7e8aE^lno{*!<@@?SR+YAIW;(mQH<=ywfVzeU7FexC zoiXLZJj%y|f~hfB85Cp@kcw1PTqr`P`oNjKm$5;z7Bk8hBRJe>Q}ZC&yeEs{*c{dSytQivXDKv?(^}@zK1KT7p3n6I;POf=!!9si)nN z6t}kN+)+J!KgN>ER^0B3h;}o})+e=ZUvi}5A>pY}xyI1b@kQwbOi}P@BdiaUVxhx@ z&O9I?Q%t%O-wjzN`c<*bjAOWPXki3)Md>Dfq*-Fd^5ciNJ)8Pw$IsYps&T5AgNY>n zaw1&5OhqWWY}{{;H#-Jm_>Bh_ z0rAjQOHc0f!-WKz5yWt{axUu6o!ZUz4bp0sbGhfmm;OI41)nxk*HF$oz0~s5 zFBjD$$=~7~C@>1~Ra+Vo%t5o^v69dIGx;9JrtfSUU)EcQ8Ux1IK4s59ckT3AiU~XZ znHaRCCqVT6hs4XlLA^8hNIwDRO$$kmUypqV3i;@0RZ6|ZQ};Id75Ezdwa8UDLZ4n0 za;OPJC=ryZD5$}<&&)7fHb<&$l)wENiY-j6ciA`j!F^+Ej1X^v=_(PRDd0hfH{S$! zrF!ly7!F9;4qdwrLBZ;7)iYeGVB(at3G<6yLb(%abvUoOf!C#$pOr6W7OY03vpXA=eP7MC5MUJn7_y*s1A=&| zDA*NvSYrvmUcpf2U;4c$oR5%xqesVtk#Hva?Cf1*qHC31lBCzBr4?W$KY}*K-q#y6 zGWQwMxPr&@$huTBra+^Tv(oy`8xSy1*F-zi3Fnnzq*z@ktUv7pPzD3G(!SOaQX#v2 z8vkoc=RBNb@xSr*m>!5)YUIeVs3J%C3nEc$X&lrl)V*_n`~Pj|9%?uy*Ms^V{sB@u?^(bpet zz|^U(2~?}TRB-_LQKm+gT;ZF_YZwI(AHgI+Djs_svI%#2lqVqru<+B>6A6ldd(a!+ zvB?s~+R(xIiUP`gp|DwvhWTU})zHCUM8Xa2O-Dyd0Y`~`*Y zjAK<*j5i zN`XTd4r%MGpGwtLdCQg`?cyfxZ(Wiv8ntf8^%0=0^IGNVB%LYJH}T-v>*W1LOmC8q z4Unt3B!(Z#`e<-vi4*I(i7$la##=O!`|SLwpYLCe;YoPhd4}8Z&r+~;F{XpF!LZ>n zM#^jLxsL<8(y))EZ>Z1BR)ei;cS5rcI)6qk^ceRcO*a9MZSR?GU3_|v!>XOPPlM|# zTFfoQ8EjQfo=il*gN&YZ(9f@W( z7R8(Y}5{KP9r7NdmY<}M+YYC z-T#5_Cp2{(_4uFbfS&PxZHvm* z9S;BqB}o+E#X*4a&hm2*Vj-bV@8Y}&I?Y(O@k(K?j|e)IUxHI!kdRVzH4CeSltL5> z;aL|kpI?&v9@a|ih^e}9_H3Rqw`#WH66hCf-8}6@a@{B0dF49%63x|%H&936?0(gU!H@PA?3W{>*- ze-ANvZ4Nwx=#D!GI+%>gcExj+FsiAKbuF7u{_~ZANJknbv z38^IZ|HkITv6>2f!C8p(Y$clk{@(}HAq=WkoZkeOF?f6?z-h&tliG#*&lZR_aUAsw_?t-Ydhwvj5wM6d4dPSqTiAC4(uLxq%kv0u4#H0fZlD3 zm!^xw;pOyJ=*pnvkK5MXm#5+=THA&MrEqTmr+1ATx3dxfu&bgWZX$hQ1*;yb3bckk zrY(ov+@XHq=pM6stA#>AWK}TLM&==@6Uilq7!SJZzaFURG-trat?zfONH zPsZd%o0f+`cla0r!-@h8-2lLH5C(#2Ur+A&Pss=Yp-=<_$bniwM&aGH*9d4M^7tt^ zgt3SS?FBp_#Cifo!r6;o1F59_N&j~z8`<#q^hYJ^zeB<&o7?0_rAL$sAJ@I`9G*me z2@n>BElp^y^A4;4Q$fHs!`)z`C@=VeYXlL_lLi+8XT7o+1p=u@u4;dLiO`xc=;M5| z@%de)35g1DWI7ZktXk1fZjXX?pE&EBIRwStK!`WFkI?Fw_fYNK)Urh}35uGC25kV~ zo*ITmf>QA)M;Qg-0CI}uD;Q{9B1^c3pge(vr;t6;7&(A0plPtJH)3uAVkbd?`34)H zGh+lcdAmI%c{P!vq6&E~$)dBzC?4)g)ID#{mZ$m1^oKFdYCBmHJ7J%m!tv>?`v%L- zYxdcDFWaA~iKtpvuD;TkZ~Un6W6$l%OOI)N^gv%OQ%CtR`MiNvnN!aYlpts1ta*c; za>mS9*Hqq006&CG2g{USinMmR=a{0PR9W*)HK$?-`CzqD*!ZSPva@`yYNQw=>3yDu zw4bXOXKAl*u)B5o0~zkc`W+W~UcajILV4rb?jItNQha@e^>SWfkR(E1OE!*&()uzH zaK&D2^s%V9XQG@cnle{k_JA;!N7k0JAbd=~5rT7N>g4jdDtCO)*IZ{Si4)J)^-3#< zaqBMcDyx052kBm>!QmfYW6lrai4=n>y?4LYSyM(;(r_@Fqf3;B{Ra&Fa(Y-C_HD}6 zkSP);$V;YnxM9_YsXNde7{QJj#R!hNQ0Cn{NQo#OfWEdVkOrGScFlMXwQ~k~E5%-n zYQ6FG^U$_(h(RP`9{tGYtg0LI)Ai^X2^Q3H#J0f;1;JJfsoQ_zsB}m3U3KiY=JS!h zanC=GOT{05H~*EW6yHh+u5V7zeG!`*p%lFKbb_foH?%m(l5eoI6%S}2f=YSDxK9~#6L$@b z=!+r^w62;1%1BUYGM<%|WsHCjr@iWX~TLQx3Du!6wznwv%=P&)W&#?uvNsaGh}eYor6YjKwE zYK;6*`<>{4_&t-h)Zb6vsFiCD*8`de*y8E28R><&*O9g{FEj!>av$8e6XfIQvt|qL z|FlJz<(Hq-1UbCxJJL(A3L-y&xAoXxzWHnKFK7gI{CR)rFDNhe*W|mupfA}TpZ1T$ zet$DoSDfrv`))d%A-N@K^~dr@{!Q-xPNNtBZwn1m{-yeJ+D_x^(Y4>bb@K2^L$sC7#c<8MY6c{T)-*(+d2mjCa77n2GkqcoH!08fyL2oUq-Tr8Z`~|%p64+CJ5w@~(>u8HI`9j=s>-GMI zRdaOyoYW}O%V^+A{63H<{`fX5UPH#=wpY!_apS$=#F^Q!5v$|% z{nwpa#9%`k*c1?_Yfd1`0wx0*VZWDtkypP; znb$VoHpODX{d=7L9JO5@e)cIie>mZuz}NmsnNUdS+29E)(Qg{LT-nq=_i;f?!tth8 z%#E&CdHp+>adMEglfZZlYbsqo@O~pqnf84ia6x(JwEEv~rEw=zrlkPqWDEZ|{=CW) zdv$9y#ZCT}%gHIC@6><`t`t4+{t||ndP!KlBlnUpe@Cw5lPpEMbEdCc+?SY5V3yNj z9cxXKc>J!p&l<@T91;m(uOsA zEQJU(okJJIrVU-uLrSc=(cTpo9=p$ui9pBlFs$e+tRhT!Bg$DG$mwDzN6(2B0h(0= z+zsXtF94P&n81uODiRndI!CPL5^h5q4PrvUQzx!Ea)9aU!L7dHXLSER5%UWxFvk3U z1=m_Nig!J!wGA?p=d>YR0 zoH=}xKTY<-7`&xQ_XTD;l95S>9%3r{|A#*De49kY<*|E1(DbkmI80(vba3g}`ugG)c@C3@Y%7TG*;JmSD zL@ePn5t_;ZG))LOz|H|+#WtW$d>sJm2J}D22!ule2t_fVo6C;gOw{0!rW@kCJe1$A zKeP#Zla7D3+BXM&@E0@(wmS{MUd}&y8`(icdbh}PvaRt)_Qwj&HcM~n7wQF~tsY5AV~CAUt2cPUj{X8| z>!wBgeews-H*N_Mo2D(Psd;rTYS$VJ1(aAm9hL4@bEU>(i3z@0AunBRzR zYBE8cl7UYFX6rI179Q7u!k5OqxKUtTGir@sM+qU5{|(=1<3@ooJPzR6M#N|mg`Nfb z2^28(`l=mKHn{4zpBPUOxBzdF%(r+1S`(nWT?YduR99HYtI7D=DtHrd;G1TQG|+sT z^b_3h?h+mpse0nSbDE7)P(TAAV6<;_kq)2kaj{^wx!4o#W6&c2L|pekiorCO%#iO$brq)Qqb5jeRtaObnh|L5)2rPn$r#v`l7c}1yd#Lyyg!ji z162kaA5dCkDw{u{x?gGxJ8#kT+oHv&zjCgS_ z!1mJ|NE9#tq~l0U2EYkQ=HH8dUN*p*cWAyuxc3EC@E25$5BuCAxq1}ijk{&|0Bb+T z(s23U&&kyB*T(9n=bSwpx7#TUQIG=ak8_^bY7bcX-J8)l6BF6)KXk;~YIki8?H^~r zGTojbsB!Ih&KWC$=LRY<1Z@7PBJTe=}#9Dw^rAg*2GMt5D-uY2#MY z!+9I8$nzWk@ydyP72qv%5CFi^190T^z%pif^$chv!Ok31l#uEg!H8)AM#H^Oim@UE z0*anU9K`@=h#3Y0aR-(1_&|GAQG9sZxAN4*k`!U;&DGea-GNF46i8<~c1K^lrPxoP zbKM})WM0L)WJW^Wy*Tr<0wa2ebO~fR{F6)*}Hfe{zf)qZYYjM9mZXL8d?CnI~&?~ zT0J#Oz2R4QKGe0!OT7X3gJXENuZL}M-ZW@FBsbeN5{wqBvQF(9yvjSUlfEg-OkY&+ zD~8x*-fW+eMN`6}yjr!;F5U&~E=q~WRm+E=#@vTv-YptgL)Ws*T;s^ixbxhYC;LXr zakd$-{AoBu!@|Jeor{ufH+w4Xco^76s)zuC1) zOzYk)tz&EZKb%dM4}NwW(oII$b&VjF6MD$;3%HEdrha+Cp7eF5A$vZmy$63F+YV$) zH>slFQ)s#L+3`#VAB4Z3^6Q>CV%A$#kf>HtZ($x$oD3xL& zhIfR%(ZM+V8Cr;H%tzG7I<)$N(3Rk`>eWzCRwc#||C z5d|=vHl67JkE;v}+V7(qVUgc%SR2-k95Y#55&1fKbP(ix^tR|-F(E)}z(nxYE^}an z=^k_S=^>jYJy*sQ_Xj1J2j6@q8l)OHlYAFyKTx&>lz6SR7``Nk=rHG4iceeFIs4TP zjuZt2LHZ9ZMJXS68EQU0zv3rA>YK+r6)9C}g8rTxR*;w-7$FMl2KpfNUVVcG+wYps6!&>HJ^vtw;GOd0NQX8o<|sf#}|^B%=^ z)q|_k`?B7-gBOB0&*;TY{+hH|lG4``PbauAKj-f1S8ntPsZ$x-Y0pYddxz*F80j1J z_b3SFOGqqdTbGxC)FgPxPgyVUxaT-!;A0flOjy*`@7_F=fy)rk0cTF*naQhocYhwU zVANgkAu9``XV1+h{KdQQ6)cZA3R#n0tJ@N9aTEy@sBv}52TmG1^?XO=UExcgNT{@@ z%PqwOyYax9f{Yfc11~J<)bgME)D)3k_`?kab2&XsvH^e|0l?&&hZwE$eUlN+D+pFq zp2za4JmGw*+U|%67QAJ@T60Z=@tuw9TPcCOJms5(MU&d`3hh$hH{wt;Obh~j0!L^8 zj+}sWk-LQkdc=>* zt9gE)ei*7=8X2p5QFiBUu;s&slBQav658^W4KBc%zPUmi1C+L)6yyzg zOinR?pL_}i#GiIK_#vj~QDSikR|ybQ4c2^*OzW|6L|!jiydsza{;EV20-uiW;VLP> z-xdL(7Q759S-iO|kxiR-O#~Z_Qusp%}^-DttKmZ{Xn8e3iXZ7)10ryGNML+so;r z6@iCc`I*1fuvs$+meeTTvwGJ(l+R>&et>%QcNH==6jGB{-n8Gko>#}lQ-SCn&c9La zwV{Bs-f4K6{At_Y^Lw_P?|_nN`Jcv&*2*8(F{F0^sGV{^eNz&XBM&dFVyxh-)I&4j z?<_y`Z23l;#8blLdp3lEW|hlR3r3pT^{ouQxxZ}&nAtHS#0yV}4>2Z3{A# zBF-&d7Kx-YVNB#Wk6@KkMOEQoZxVUzC<>@Ge8AZ)a-rbr#%Z30 zNr})pT6sF|UlJ{fKl6f%6nj%9oWa6*a7%YK2}@?v=e2s}^9229POZ4amWlJT9&cq= zR5*j0<}rq*jHGgL@Fd^!42lLdiem**Jp*~i@UcuX+~CI+nMBFRK19aZmNU$830Pmq zV=GiljkTk(V5wm7o$4vaK|M6V7LSs+LA=QNjr(CkG;{)b&X+Wv6F`?p7d*i_m-i*5 z*LtGWoDfgF-o=7%5Uc1*y|On1jq@jMlH?7fE;5npT;eJ|3T`{jW$UJuSGYvGGs$0& zfoq+Ru1ufBQ6THE0o`H(TB`@B=z+8?yZw{e6Ku!8n4wAuL2B+O<5iGG(qK0*(hxd5rnnAzY4daAIxizWWgr?SsY4~6QM;G&IJ`**KWZdMA8(DgU3Z^(q_$^~dmL+( z4ve)*8*>ZuL=#~1RV=(z$EvE(d#00H>7UG=zy{((-hS!Y+`Ng-dm22~1^0ANx0KPB`@)dMuD;syU{!(jlX;r}59{QhCiJGcpK)>z0ctsyr`% z{#1zIKs|>Vxb!m9s-y>s2StNh%F=oiz~Cp0d93>63{}*5;URJ88;sVEg41A1Ju=dS zv0SHK4Bx%H|s95x1-QkbNm@nd;~`$&iV4c`Fb2 zic&8-$tsQ|;K`DP>>ub`iGxJr(@U$jh6~k^NBx%V6}}O+oPun?)YsJ&HTJ@BeeSU~ z4gFcH4RQtolU6JS*I>B;=e()hZs|bc2=wtD;&lUPhsvxSScnN<0Boc&Pu}7o7`8DkuT?PSJp4Pn^z}&Qca7idrk`gz))k<#abz3M!E+{g* zx%x~C+8|pvhez-S3TS*D(aZf^C=-m<2#EQ%hsx0Izu< zdykb_8OXr{Je2uUOOrFRE?KZnlwo7N-05#9{%9tt1P1kJ(FgXpzis9m)6TcTjpu83 zdo(a-%b!&>kEo&KXQuK56QZHtx`U@F{|Vy24aA>~tr-AXZpLPv9X-SO-g z+*!rNs^%(V%ei_V$;`@Qi>p5?N-BL_i&pG>QV5QT!Q%nG5*HFF{3|cB=PC!CA0#tk zFC-pPVZ-iJ1`UB2KFgtd9rZ;>Cx6u22hz9s@9FVAFOD2iAT_!N~?-Mg!~(LhBmO zgwz@TuQ+5F{=4e)F5tcVRR~}v8UW11#{GBPH5LOUU$R!b-Fs}3rF|u6ADc^e0qAd! z`?N7aAtI;STt*ac2gszM@It$V}=a`5yHOiCr?97fcxB!f6}a&{b$`+?UHQo}T_ zmH-~-AmAqDDoYyTD$rq&7{iO7J*Lzfl`#i^hK&WlEvu10G^6;5_YdqwY@8n#p!A;? zhcGiJr!-Odu z>yE1y`reSY@5TCap2sJkTJSXz)v<}#C6eYQQk-+pb%5wv@!A6g-?tQrCEa=NtUDkE zjUwLf(Q_9^ocxc{c#D0Ei`(g7F9tS~{(?42Un~}|SS(vCv9*jW7!V+KfCRA3Opk{6 zhernt0JA8H5fh4n16bYws7Yu5zG5j5U>mM7N&}M;D4>I)P-I8%B0MVyO2JSmC>j`7 zQN$Bz3_zzF`mZG!0-FiofX%4@K*lqmZvZ^gU7!3H#MiM5Ke-{20r<&dt@ZHyQq%up z!2cjDx7@dG6|kdT6mR)~NLVlL2V+ieS}nWf-TwoF-L*}hNHRP#L_o%?z2xu4%3zd!I*WE=1I=bYgVAi$zCW0K z2Q(R0Lmdq^_Y@Z)8l@)`<&s(1p9rxYM$(v%NUp&93d+Bjc~QPzPaLLd+7sVB>Rs8 zLwi>Hoem%MdgK`#{ug)GL z=pH2)giapY#)~lF{*>huz@^VQOf+tEq7a&iX6B)-jfZ_MAN5S@SV5d4VZ?+wDt^P3 zQYyuLCL7}$qHSE)wZu94Js0=nGTMD<7Kez33Y!zRJVuA#T8#4EkEeJ6kWs6MYU~y8 zvu(w7#hCR3!T>3HY`X+*gaiedRwRes^s0 z+LQUux6-5X%lFRrV5Nt?7T!W*^~}xq+l}ida|hH+rU=yevGI@%2jYK}pWqKcBIFQ{Gip{>oER;LP3#V!C$lZ{D@inSJb}BDhc&p zH}GvP#GQ`JR_w{eP5JftVybz0`{J4<55)Px`y8REOJ(k#%F#d?Eh}>``?YuBF3ZMI zxDeE48QCYrirW`vTT#_M+_!?b=j!-1PG9ica5|$RUR1R}eWGS@)4>7v)wJ%`B@a%V z6(6kEtWPy~JyC<@fhrTIU0CNH9@DrpBp|NClW=bc##rW8-Dgd5nXOTUdG;J+lf(Mu z9`w3+u?DM1=8XSsCAsoK@p*X)Am2Qn@v9fq_qit+!c~C*`%}ZKBy8yFT zegiS+5i)BMq&G!~*;L?u@(ie33FYth2fBit@@x^6c`B6?qe3R+aqvu@(+X(bc zS=E-mH_OUI{y)PTsYF9YiGO4xF{ph~N;MTPs3d3Z%Mw{k2qN)HgXK1pb4Ah~t(W=cql2hoZENV&kD)nlMqDAWmRg zF{*8fe_Ig5mKF{$-9~mYoh*-xH=d0nO3dPEIP<5%80!=IrDV?T7L*f4vlglR zYDVtc-<#8#ob4F5=p)8&xHW3^xk~RoBZ6w%z7%tOAV+cPVuK6jWIt!?${^wKrERZC zK{v{LmK1#?RHy!|^JVgw3}-w=dE=E(tXav$rE|pvQP9E9^Yme@#o`J=J)t2CL=7+2 z%owg*{|68fkz_WX!>~pKG8K3stbr7{Kiyvg9>43!vHdrMO}oFzXYb&W|6Hp4h&$O9 ze(J3M*>WO@8^UZ80*5ilA6WieBP|-nJLo;%M$qWn8Ol9E;0?|K*)6dgtFo81JIo0H zsx54I8dSpuQinY94LF&59VmqH%>lpy4BW2YVpeh$x(1l@q{;wK-Ym5mH!5s!=AP8mgfE`Mg)bF6`ajkxsk)l&fv=ffU`mWp^O}+``zfp|(JNHmR7; z&*&kTe{S4n6cTaOlVr18=^t2%NL<~>wh(C)bI0@OdrE6wU=`7Qb%f?ILKNMuxxz6Z z+&kCOp7e>W?_iqaP%A znE5`1!&$#Kymt%^x(goiN=xaY4 z%nZMC9&^+A>HY$6gMuKk6U=~0SU0p_2{(iVeJSo#tc7L`*g{LS5%yWJ@uUliS`6IOOAt3ZB8pYfI1;X9DC+{JOntHz$s& zV_!%=?asEJ8~ZJT=kHQze5#L?;)fV3&m`#$c23deo1UZ?#@T`$} zJwV(x)`zV-gbY5t1{3bMEM9yd>}6;71LuXsChWkdpC+>My!s>iup(+{)`+VCXDH(Z zO|a38KEF^JS#6dcO1l%r5xpoHwOjvegEq+u3Q@qAu)8IQehD6Vey{5yYHZ!%ZWp4% zqEssLMDGCl{rNDziK?r>t#X#D!m%+y$(ersivW+JAvQxjGTRA}`awKLR(9aBk(2^2 z)sIJBY0OYG2OiH*G|(dd->gl=&Ai~4VJL*#U{ME8OHV2tw^sc_IN(^By@b1i?G}Z^ zgImPKtaLwZRpEdS3b_hU0x5Dm{q`~kRuU4c^!T714XW)Fn4RE~HCMXSHx4+~0u#M?+2f9V`8EEGZ)SkKN!G2wn0vB~mgnj&q9&*cLt7gaHz>C9u zOKm7-LYqJvkE-jq)W7p2;x*8Sl+}=y+9|}L9*2(Y+`IKa@rkC^5>7H&)$yMGs9;0R zgQg*mfmvPDtPyGF!mLqCWAiTl8(uUw!XP0CwWZ8|U}fs+0rcw!4`)u6f>yL=rYm?bud&?q<{)Q?}6q_fC z!lGyQ9R&)i4%n$sbGaP>J{+*U+@k^WdxdHj*3BgSExGULpq_MS^EyXrDKR@Nw#f+%c)XRDM-ovgZCSyrS_WnT?G ziHaCdZO!B@-?%ikibPs;p|COQW1|davnKnI2pmo4BE`f}bT{ZI4kQO?>Tw{c8FE0f zM1liSokHm%twzc^h$o&%>VmC&>QhVpNhTk+nOghb@0>6|?T0bH>!-)wXV^JcRv|Mu zC(Eg-!#pL6q;BHFaw4g$B7kn_I};1)dI-OpG__b=C#6!%n$aTkWSIjq+tli~MaEWNxkF{s6Q|X{ z7cegN&SmF3mODmL0O_5manWwzt3`K3TQLDXK3G+s?neV*)Wc!t*K&(1eS z>xY=f3D-xU$X5bm7%AgbMX+owIcn?Wan#HRUY4vwT^-AJ#(P3(_!`^AXx zmiG~(GyHCFLEA!8lg7LxQkc$I7=Vqz^Ve|bNK4}`2f_8nvn5<(RkLFGIws_Z#5H;3=yHXCcyds5+bo*dhag5i?iQ&WgzsfVaU~;BZ;!$vsR8+Ru$i zyBa`8hg)YhmQM;>7sJ?W!>av4j9uZOGy6gN@nc2h7gW18Ji#Q^W(*Ln4*8FY&yIX1 zi_Pc-`J7_B6(&XwM0=aQJ*+()h%~E>@e4cvOFPWDDfk1HGqM*Kry_-6)Ii)JehMm* z*AU~{jJaY?Fvi3s#LR;=!AE~0dd6M{*QJaeMr{-yax7a;f=TAiq(pH#oxj(HWNuvN zxq(wbp&fpn!5+1h^Dl$5JdfgM{YF=lQ!9IH~TlEYA0 zMJ+BbuM?`chsj~6_@I@aEO9wsY`39ko3&--#49>1Wsk$8xXq5wpH&`nj7Uv3>2ml% z6VASAb71m-u{gg+Zu8s#k~qiDWj zo%E1dQmjX6K&huKED{$O5$t(^IhA~7j@TlM{EmL2fO)eL+wkr&vFQaZ{s@sDU%geP zkJ;h(mMv~Uh0L>utv$_n_w>L@=Mn`M41i$DS3wM3b4W{pw4f?N!FoEgp$y@CRqAzS zO4irILjiO%8qwN4T!-z!5(TW|4ZpjnLDM`b$&Ei1cZG7C#kXzGp;U<-cIw zJ8EOo^?TvZG9^mK_=n1aQSXJKyT9sP%Iv!Bm~?db{IYa*z`B(yE@Z`<*JI+l%UX;* z8zz$mM~RHVw}ocR+hVoZFG)V6tZTAIvsb8Q4r_gG(W1V6^;_<#`<0tQX|H>q6%7|r zqzXLUtA5tzBa9eyb9C=p*6jL;5)MgMj{XMuXt^UNDxIXhru2L5zn*x&YS&i5w8qE3 zc82tHUSsHsU+RRU%jmiPbuDp_&--(7VYZ0bUT z0#5_K5}d?&Fh`uFTS{3?S%+FpQ3(_Y!*W7!jj9c5*NN*7KQEI79`ROtn}YKts;$L1 z)(qz?ML+5lH4eCM=LM>WTESaaTDjkr9z%=YD9ez0JkF(8qm0{eO*boQ%|;?NvXvN) z-uYo$2l^8@L3+SaLiq#^(hZO_kXC^gobI}W)sb+gfF5H64TsG*W^3G(OwE$HaELb) zPdTlNT|hzSY=aWiLbH4&Xp`x*G=|-}k-b(A=~2lMv3oTLpB&36=oI{AY`yV7q<6Ib zmQ|#RMv_-Ory?oVcxk<=A0&u{2I$z68b&~Vfuf{>KZqek*pbddjm>u!Xm(*c+-l7nM&XKa6AMB|Pf*a( zt8y6MizBJo$|Ly5vhtxFjh4sEVe_llBr3mQp#7Vgl)tH*HYZii_HscF9ZKk=OJf*9 zoezuJdv;}C2J%=JUb<9zK^Q}CCy9s7ava9^;!0?;Ws-A zb*=Hv-QV?I%M0Qy^uC2!ehf<|qtq8e$3UQFWwupVuqH2zYD3)|1b2;gsS(qtn!sqTOAQ|! zzCK*FAhd}#*eOA z(K05b7UYI0hLtyICF2MtW(*(&-|v}_zkeO1wQ1lg1yl75lJf*-@XVcq@@uE?4IE#0 zh6^^4IZN3&e+y?D!tar~s~6VLI*$uO!*c0yDdo!Hn*rX`We;hccR4rnnm$rCT&htl zHx+9u#HJI?rY#3H_mTYgRc_&j!aW}1ZVS~p+t62?l1(wot!+Fo*N2zOwo2CZ`%5&d zy+|*@1>RgmDPB;xz7H#}Zm-vDZ!tK!A>D3ER8Lfe{z#Bds1c#MzPB?ju)&w|+Rknz zFl(x;ehNW6nr}AjT^bSlDxCie|0KNB_S_Qkh1pZEh#LQGbyJaub(f<&kY?j{J5jco z(NOd9g4$@2jiRypyw1>pbpJ9&UZ%M1G-jiIx44v3cJO7d@Q$3kK+-BKf} z@ymixR)iZ}u{V<(-%$2vta_}q`bE~dUSi|~utZcl=WTAs4on=ive!&1y*Ri`mZnx_ zGxqer=68_`DxY`n=MAS_UxmF^=UZ|{C`}qk)_3#CxLk>n%*XhOgK?Tg1LuWC_qbfaflrka*q*Uzift+zXw5g8%%_d zJ6*6QprIIY55uKR!scQ|J_`|RNe^AsEBm15>c)oso! z?S|YUYGuP5om~yL!#cQRwX5;r%kh^Qs=@+~qwljjBpXv{_h?c197juf#;b@EY%TJT zBNkW6VKtt>o7uAqfFLB#vV`gnAQgfJ;BrAw(*T$BG;f3$SG(aKN>Xuy1rA&lSS{~F z^JQ?IysUhRC=riRDk==p@Zanh@*0ItO=5FBx!!;#ASw+`REE5xO1w_s6C_2|c+OF@ z9Og24&C`hrYl9pK)`rS@fK=WdVZeL%8=me3i4JHGQnjL<=#%+%RB$LDqXN?pG1i@^<0&%1?(;iwcw+8@ROh*q4(_Q z0ri}4^A`-~?@dJ(?7KRn+4O`~iQg)kMeLnA6^YS7axHxIB^ztp_9dU2H_>!ew!>7O zc&*y87*Au6Fn9n-!gW&v8}5d-Q*mhQ8n(8%#PA4lR;kkD1_%tJ3B#N0QU;b-b+nkj z@KCyL=zWjEWWCozVf3!`iA{W88T_SU@#D(>pl$ZoEuRbjQYqVqKrHA(Kp@s~mU-sO zRRn5cHmiI&kTM}^gV0^UK`m!Vp&bpI*Qeya;jp5MjbTw}Qyb zFPIGuMRP8zX&nkg3eGG`q^kOxBv4I8C0e3f1tvE=?`xN5(LbJw!W#z z>7_ll&6Q)aCb4*|o#8&_sp&L@nMqTQW6HJCh;K(<%>7em#?e>JGH-`fW0vS^aP-!^ zxEvvR2!hHRG7jg<0MtQdXj%(JP_%%91<7f?Y-_9z3XRqAg0`G&9fU#9FY=8!+(FND)41oVlVV{kPv&$&QqzM&2pUQrC|j4UPO zZHAgRaI%VN$+w!Rc`tM03G|ExW4ux)uTtWO#XEiotq|+U^cRg5mh_Spu|l)soY`>> zlsCZwIl&ZJx`13?cp*e?2>5QNe}d@V(4Lp|v0-Y;!_b_V=-k9%C%9nvj%9?+S<0NR zblVjJERSy-#qt|Zxx_kRqEmQm)-MZFquaI+4izdyj24NUENZ?{gAEM?URnh=6-fA| zq^NjY7I#icR+3vLa`=K=2@PV+*rE}Z#=25gwc<2mpfhk%_S;6;DYw)pn!YlKQB@|w zW6y1?@GHVTT@`j#Cv>!|JfpVg7cKUM1>{@#r>|xr_ryRUOHQ&{JR(jSJr!nroPKXY z_HYrWSN6;%_E(-nzu#`a#f_uPre}0cFLuiUP7#N1be+t=mzq=1qLM*Vgt3@@ zfj<5>dxHgN4Lp{vvg~-!#&T6hE=(TLfF4F>?m(2`7#r2^US8w*2WRB(5SuFJ?v5g`soF~xfj$dU}w-hg@Ff( zo!8Z9G#o$ZsEOI=c}YwOacL~WCKzimYTZ6bT(uZN$4@Pcf@ir7E%Q5`7{zTcSb;g_ zIA(tDm8ch|VTV3EXWqz{VDim^4-pUDhfB?U7Rwrc8m33B7SkU{jQqxAFE-G>i|5~_ z544GWEe?x+I%`8$jdT{tUOhJ&+G{G0F4*2ws)v8+EeycgUPBgmgmPN>0;#(mF4Y5i zf2@z7^Voq@eUDynk3~jsF7|^`ARVke>K4z%?>Z_$bM^btm`?guEDd`UuFyaJb9?JQ*j4-dArEk`?gMgHUH35AdrU4oeU4 zm;)^>x*&;$5bV%Z{G3D6{4jzOklzXoSI0@FB6sK_U7RV@EZu223qDoP`bOT)Wnt9Z zH-L=QQ~l;ZP0j9Pm#oCtuIrk}0kv~A10Ic)|HIU-cxks?|Ypx=$!ZuIgM@>q{gLdZ9aG}kQ#^p#@kd-OfQ^wJkH zCoWkog9{an6{6^#)&zl4o%$NBl3&HTszj;yjHq*3*$ZCyA9fV%Z@hki{Vg|_W@(ay zEf}2}=9n5yv;*F5Q87 z2+2n1sq@xF+_u65AGvNw62z}MBoE}5vetwD>kD}J4OgN&`Q;&w&X`&In@dCe&eKet z+`?}%8{v{r<7-a+caR`|{pm9YAj!}F0ZA+qo4R2vvrv6E%2C!dte-&zdDzlqtcUQ@Fjx_Naly9@GV&>0rU7HYqcZ8zy*6gfmoQliewP^7bDjBnH5i9Gn{RK*8HhUk zE=dNruPF~aD^NM%VCTk##6dS3V6K%rXE(!8?qK~-IXt|1SHqVJLUqnr8TjnVzYxtbN(;5lh**u9$2ut^13d+u%lVS}{ z`xHDmE<__Dii^ATfFtm~ws&aX`{3raFKwpF<+#)JYnlK0W7+7E&ws-ZhhdWfq>M4r zflPoyfF5KE@Jqw3AepjQ&@R*210BiMM$h4!u<%U`P0(l|oeYPLb2I_5PRq`kGu@|1 zl?@97y%f0=p@7z>D%7l76qX~1?(>9p2o=`DLC5Pz5~*mS5wcrRT0$S}iIe#CdLk}^ zMHoWq+c&JICuEI zX%wvWNX$ZxaN)=S_Vb`glwr8?QUk5YV}!XsXgCqvG^}hG3B`@!&F1n_>Dby-c>vzh zRfAfQHF}6sZ~Y?bgU`Fft^K1NR#Dj76|Te++fI&L356Le3D{B>g2JA4S#XD2Sr&)6 z5P!$4>k!L+WRUef&4%#u2L5Hnc;-zTQTs;KmL!U+Jj3wRcKW-WzUueJelfA3>fFeE z&4KM27HrY-+FcTj6 zr7)o&A*wo<=*VqG`H4zNC9H?TD#}ZZv6jnVenYz4xK^i|;sfj0t#Sw;k+M%T8DjiP zhO`$cXHP;c5+nAZdZX->1I`Wk1h8L{XCo}h|Gq%?n<;F8k2VH>UUrr&cp)u@TFak+ z(JT><_S)cAPDO0a8lUg_)PUhUfzUYSDHon5q8t1Ru(JHyWW+Q2{gtfj;Cf^d8Va0y-=41U$8;R6i=aF>RwFs+o zs}|g}Aydz`J@t<~KKe9=v2925Ikqxa#=exol+(+aNc6If#D-QE44#`q36!D>No)A% zhb%)o!XRZOK_$O-?`9FtzvF1DQX8!>ZdO|XI{Aq?_y699< zt1O`oQgy&?_cJOb;)t3d$0Vjg6U3*y`}y z1J}k?n?Pi^sBP|GCjm;$${qZdr12V*+3WRxYiubvhokB~&yQq4SAlw;$iS0uV9*8? z7rL13J>WAuWT)RM!j>pe3{$>S;2Wxb;G6LHYgm|mH_HLAVtgK4P2_rNphK9@vpz$! zAXJMgYoe}RhdF2VZFy9FP||8jWMNZj&B##4oF_V63+eHwZt=TcFk@ujMMD|bdU{g< z`||dd1F?O>!~1I)B{aUa7C8z=zHaA5V1W`3BoQQrb1JIB^<>*5UN5Latm^kG7^BYfIE1LRv=d*pw zi-1l&E#gypaBoXTKeJd&P~?dhJO` zJ-N;xHA26$wz6YjC_i4kTp9h1;^}4NER8P_u$(ei@A2P1`5k z^q&=F)r+ZH<4*r5CAJqMopJ!ZfmBC6O619A0JK&24DctJl}M%gD{~G7_<>i{2oY1_ z8sJ%Gg200O8W=(}z{iM)a@Gbq)-!VZ^fD5db^z=V2+*^7s7jK{VwPdmTh{Xt7)>%1 z+xH8MTkx#UZ)okA+|TOb!&+IE(CVoBpcO$myPlgsHLcCBBKfhZnjFi_2?vEPN=oQ) zuMj4`T8%n{fI!Im&{jSDx6ZKcNbIjH85+;F(5x?dk8f=ZZKF64*d8Nw2HR|~T2qW| ztAC=@9QENWu85h733BcgTPL;KK269Y;)ngi?AS$x9`&jeMu|>rY_h1L@r)vw-yLBC z%{ON@;+;2Iw?@^PWi+MLGnCuGiZRByEZ9ywG~B~oW8<9pmXbVRvSna_iOo&1!I|=+ zL3{SqXp+M*&6ByM>pL2;qgzU&O%7D`5O0i+dB*Q*2)x{Mc9osmR*lKCyK(Qeh&}69 zs+|pRUaF&b`M6Ed{YU7)#>1VMBPri)$c7JTs0HQVGdfEQ6qWL=rZx2|5{*OCr#-u9 zc*O$|QMX(*>UVzWpF^bX-}W(Y{Uei)DNjka{Nwnax$Q^L-Ns>v*hrL~A;OWz7?}YA zr*3IydWgGf$Tv&#;0``K-_~%DP!S+2T*0)A9k&Hvb!FLbus^7w)+c`Y^5CrBR7cAX6!@2VKgB({XhU63WWH4%;>g z6~a*Zf#u`iHBv1Nn&VMGP7Q6RNu-LJ&=_(tE(~k z3fAM-V*y(fUH$sqE%}2Fbw(B)UKm0zx9YIFg{jk?3VL{`H?whgB=c){zsHaZXvOPf z_`#6Z!`JGS*ql;sVN=*>9eY+p>Lq^7cWeITtxwke|2XW+@LLg#Oz> z_O+tToZD^BZ43u2q;7}IyCruwYHW^NTybG*7QOIUnsn>kU(}?$6pe%p+@YK#6+ z7=xl6?iA0?6wg}Es9>X~s2rUo6as(0i^$O+kZ7(`c-Yi&!5fk}3~ljpUGiTsJmcfj zSH}OOgWP=A!)H0IA|ZtlatkyOo%M3%OyI-QJ-w`u&;u@q@lZc;vE~VGf;*NZ0;1^+ znQ4oYz8;P((d~g|8##d!%Piu=V4%;_tonGNDm3THSqH+LQq9fEkcZw8yUmM?@^UywAjUYs$haJ4q@xF{Q_U{DC_6kvC8ijS<#A!(J3O zo;e;TiVn^?a4^Sg)zL*JSH7ga*)aC&oIO9)Gh_oj2bX^n8e7@5m#iu9|Bfqo{9-Dnqv|I*u{bA%{c41C8x94h-iRS0I+MXHCC*SLv z`=(gv`%B{+&2x)Fv{c<4H$?f*5}#RryRPx2=Oy&@h{T;+U=vT8zpKUQWZuJD`d_L8 zo_$?CGVJm2x>)8k)z#|sWl$5dJV`KmamkmCE2EN(Cg#qN4fVfPsXV*Tm0&UVTGz9s z+{x|xA1te68R31gm5IxEDj5Wo>x`FSkJc7*Z%55V-QQ~NVw>}q%A2yg{~|xpP18Pq z;_&3_yB&+3iW~w}Qt9TaZR$I|L_R)rHzsxN!R1%u<97+;w(dK7b%~<&lik9w!ImHW zO}C>WZxtQjI(99t8{W+^7(Ui!i&Dr;Q#4{dzT2D`{DzrGP8>K<*cm8o@WgM|)q%x_Kbt&)}ewch)~#wLJWk$CaWV#6_z$c3u*- z6uVlT6n4tjlO}2lQt2ghQDBx`pcZ-rW4+OOUUmd&LVoM>lJEOB|D|$AJj-tTMtsx> zgCCdd_q`k`zv#mCeze6tP9)9P5>wYNj&*wA)<}Tm35gtW-vUNHv&fh~(IGmdc~-mj zg*-9Jdw=F>=kr0=sZJPI$*SeEoAPF_#=qiSiBNa@-EEb*;+JF!^7x%&vq!J#-pK~W zj(A*E{B9AJZ_8SjDfVNR+jWamxlEE^3goMN5kuhU>BmIpSLw^VKDh#-QnTrLK{EfH zpY0lqJbJJGmx+x$UH{){SK*dR0GvEsjv#F6bO{6xP2LQ&vgR?7D0>JRhXe@G>FI}X zu24>EK<))>1NJbeWx%YfLd7yrDwdi6XAu&8nsoU)hN}(!U3yWmCgbI)6rqW=ox>kO z&1#6fE&eC1(f7c48+X{W4o0t|z&QRKmzS?kk^7NWE>fEE9ih)h*h4^`yfp{EIEZd% zZhF*G`*==+Hw8a{D~a2dZ+kdbX)F2kSkd|CPvdj1_Ddx45#dpm{KxL_qI~U>w`yMZ zr1!{njO!5AcIfgw!jI4C$nDXLoc-;+&E#Xx{fA%wocDNz{?v~-Z%_QC!rA)r)su5) z^$%=5x94|v(^%J~x~F)j?cOfvG1(dR?&r$tmL8U+@Z)7nX5HPumt`rbs3z5D>yMnho8H@$&|se|+${w9hH$%b53{UQ;Wb zA2^>GF*X{%Ki*_f_0Vm;&eO43ORi>6V99oWk^ZZF!_=QAtvwH=zj~r&ck05g#TvkWZXnNjvCpC`{dKa8CWefIQ^v7ukp9;^;~v{xb!YFt6oUjulO~{*UXY@_Qve#{-^F+2ICfgPa4|V(!7`WZAIrw!_q5_ z>oz(R?>U|qLMuPHqvl9=>0ROnjaL=dPW-&O;@%3>RlT^fl}7k0`)!{6rILNdk$F*N zShoA^<};nuBCp3w&aV6INA>+bLd^Rv8Zzy_9O?ONxH9#itVsI7eEa6yRl#j8TDQ5M z=N9~>V({kM%9fUc{GOB}x@={DS;{4I)9~$!Hvd>HzO`4Q|LHeMSybH5%>zdQD;j>- z)$E~;TrAo5tX$0K55wK}^&b0%n#lNa@X>eg;L+q-Cm%JkxlsiDrjdHhJkM|2l6qd- zF|LGmb)+BrrD7uPwcyB9^y%*Wf!FO$w<;EVVE^{b&)9vF#0IDJO1b#wIb}8-Fp;UU z$1<69EjVrVSDIiEjd_kH#XLxL+-`zTU=%yhF)Vu(*-&O}`gdStHaD#O%)UZv$-OE* z_421bH7E}7$c|2#LkV)=NqY%6I?5L>JGG^XK%3sG0#6gXHv+H%tR3CE}| z`UAAUt{)p7`PGE^b>j)avrTu{EoHvAq{E%80-L_?sgY%MD%*nQGREJ%-ydG3}^{mq*C=WyD+K?{YSrsjnxNmH01o{;sj1(*!Q!b;s(6F77nHyTV9sIn{XSlpn=-;y9P59=h6ncx#%qg=FaqlA9;V zZ_%;;1AM-we=hnjqRT|<-(YCkAUqvl86GSq;>bYff%x@r?r56h%0>gDWR259Q$zir z^I}Hr28k8cGy9WpyjckRg?^jq`g7il(u$)1Al6}M+O1$G+o{;OY!`tE+5yxQXqk-4 zK*ar0=uwSQvY2>;5*dd{n82cG95|gR{qoD(iG?m^$1eYZGV^H~_X%cV$y%_Hq@(pHv(mpn@48@KZA6~2de_I(7G|e74xV_N-CWwc zv8&$~z+79sm+q?*k{Wz>rVJYHs`E6yofkAaE$bB6r!~LY*<8)qSm$=mo3#`DxQk`sSSN#^FALT?XN z&rJMFg);mqLtE6@#j23h`Gn};KXhHX>G^772#LU99@LNF2JCu#(irO+xB1!gZM2-; zrC-M`Fi)9mx7$FXti1Nsp>vzcuKBaLIKLlehkp-Nu02S8i^o+r`SsiQx(be)LzPQu!&EO)*4 zVk3;7v^YtsobV0t>L_!1M%n&g*1MZ^4=G!Jjcgj2)ii(d&P_wwt6Qc*^ z{WdUX(NA8U4?nxAb3e7c^S&H$!|L2%%i^>(v7_(SKHp$cD2Q!=(&Yj*tOqtO<~Q1Wcy+mU0*fxv!aU@$xCvoGV5b=Mo(&l#d_<6Z>^yf zQSBQO&>P)HzO#S)8Zz|pxSxVlc5|h#t){lc_NU{u35zQjq-M6ySMH_TLkigj&7mdU zqcPR5+;2G=$6Whz{Mqh(hXb7Jg)9bf^b}8?_CES>$NeZ3yVjAi8^BajxJ@h)s zkVd%+F=la-ge+?6W(*iNwJypPlmP@AxNg&k_1PzE3J)xk0%@Cx}U!y*4 z70m_5|Mc9_f5#1o8AvxT#1)|HrMj>k&V#3=v$1>#1lV*S)-Ed*k@^RURguDgqPmca!*T*E(|-LgM(b}v z^FI)|m*Ca(=MEALP=VYMS>|0(8H+SARtWZQb4>p1yJh4=E2+-6X;S+#5n}5X7b7$k&dAUALsSH z8?bc$B;wXNTUy##&dRehtsAYH3+B7t215}iuk+-!igW3l;heWgW_o;SZ41X2vo2-a zUvXw>S=XQGjL+X~F6urmJ-w(#r}cK^JVCsR!|o^hEEN$MWXdlv%BCbEyh2+bgL_YR~bcQBubKJtJQ J82Hzn{|B;gmxurW literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-11-Futuristic Car Interior.jpg b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-11-Futuristic Car Interior.jpg new file mode 100644 index 0000000000000000000000000000000000000000..c46440c58200943a81afe305f2e4a3f823a10213 GIT binary patch literal 145941 zcmbTeXFyZy)-4=*QL2dohF(l)8b}n7j)Blar~;x=1cHT76al5TNEZ+yXn+I=9R)xp1ot`bdC&LfyBBvBJ7gKMlJSf=<{0zw_r&ja5K$|0OLGVZ z2L!?a{y={J4Kai8^YZfX^6>NV@d*g<3+{vp?cA|rr=<8U5va73tcVOdjzbI*Yhr|mA?%pFIDW$Bk4}nxg zV-6nD(bY3Mf;G3mSz0+dIlH*J9d-Bd^*iMs5J(IS3qOA$;$mb%Vp4KSY8pBBYF>W9 zwZbAwSviefQCY>PzID5~rS(qR-S)>%I+ZO)F>qa++}vE;eC+Gu;Jg5CTw>fj@+e+$69>MN zq+JSHar{uzoYJNa0Yz=X7npa*px|yL^px^f_N8rK+5f)_i~oPRvVULL|6JE3WG5E~ z`0%*IAV!c+mk<4Ysixx~BYhfg{u|O0;`;1Pdz5)+Y|uj16>W9H-OZX>QtNwuL*Tjh zEbv3(ETe;>2P~{6chBmo_IJ@=Xm>cylW#*`N}O#euL`uw#CeLjeXOO$>q6b1jcGYz zzgjy_U6)mJ&RL|;maw%@?*cN!WQ;CgBaOAS6+76|$0{NhZGAE?N2A3e`LW74m+NnG zE{K!e5u$m`4#qrC!T_g(H5Q@_ZN_spSY-=(J5e1umx@_?C;av$od|l_e5LV=3V(-W z!|v3u`lnAKr_x*mt}kWfQ6|%KbH2a)AuwJ+hczt(h&6`Vj#QBM9^_8Xam!iwn%J9f z@nB)@7nYS;ZjPHhD%a8Mq zhFtLf5JB|n%R4z?C4__)G7gH9J@Ek&c@9_zZ*BpZcl(t(SewvL2(*K-09Heu%EufT zX@Or_@C-(##y$8-L(gq=T_JI;j znH$m4wIwfO5cCm1i*z&sb zjDBHwCSw(mwp-{hHSCH;o%Qj%R4w)*;7+qpxqTkVx=U3ARh zXaVn-g?+MH$gwx08g}v71 z$M_Tqs?AWx^OkFTxP8N_?Pgj|anf9wT4dj5-RtOm?pFBV0MT#Al)A(EjyS4lAwU-1 zxf2|{W5vnm5j3z+&Dj`bLLoCS1azVshHyV&lYkkIIBpWu?NQVr#`Uf0vnRcM{i^xR zHz(9yE(b-tYfXbjSED=}GG4y(Z+nTSl?uJo*E<~~v4d0fH{>tRJ8Ac{cMhU+I>k+T z)Vg8>KOgdYB&{ms&K(qDPY<_4*2wNlVmi|?I@Uh#U$O$|0TZHq!9-5An9eWi?zca| zWrFxKgW_rq%a#q)dZY9grKQdZ{? zO!Vl z6M`k9tjQ?jfx@zaJTmV=vZn*tSQ*+@4&7KaT%Yx4X;F|B30?JaiJ@X->BA(ycwl7dn_Cj_ufJFMno!HDZs7xb~v+ z-0%{YPao#v(+bn+!j9@&AF`W&LxOh-dNB*lQWws$B4!G1n;Pe;cy2gy+uwk0Ie%Qe zEE|8stkKume$D*kAufcT!^gdP<(-Xp?n~P%q8b+}ztO&VE<1YN37?7(jKzmDu0C$8 z)`U8IlsKsC0&PaVI$2%#nF4Lo1 zVku;=Qd zW1n(YrnA}}iY|&#|> z9IfRF7L7D?oy;M`)BsDI3rHB0F^>ctP>R$pD1lXwEx>+nkqPYeTaAd6y56FF&^6{k zVDq}+dc_dN)a7ByYwQQJ71&jR$Fl*OTMf>-)roigLU%gv4>HGl+!(LWlQ5Q;cPCCu zG#Cy3GJy{wL|RQoLyrkPXq1f2DY!4uJo=*ltBz{(+-PS0`Hylpgicj)UW9ZZc}qo2 ztrNQI{52~@@18#;LA#?B=C2&Icd$`{iLyKP_nX_UF0G0wvL$*3rn~jq78k8er3D$M ziF=g00-teN39DxPeL_dYVSRY)oAgU@k>_zys%gY4u8Ltk3&@$}X}W7rTZ!S?ulcTT z;(X7I^p5o9JypK5^CzJkQc12pC(o5ETY@=Pz9L()`>qv6Og*)hmea4~pmgE@Jm&>W z@?JXQiQw*e(rS{HpX71xi4*21{3hj~H?AUWc&y|{BG+e!AOU2!nPJcM7i$joyGo@4 z%4QJ{m`0N#TSwGnHv}zyL(1q4*Hf&1TL0IYoLd^5v+!97T+(susOq)~u2H(fJ&T)b zUEZ`T50|eg=B-F3(4BJCgms0!s@&2M_7NQ zTWD$pN>-k_==Uk%Z(SXo>Vap2GHV7k!{K!tzGmviF#U0ks=%zXGfAw30K1>4p2t`o zd+0A0&k(eQKC9U$#}|*q$mN}BQb8{0R0x(>^r@ywwic0#$*Ea|16*u4gA#ET!8{E0 z&i)s210Ok%@dyrAGDSt*){IOMxfj6M^Dgw*!-QqIj&uH=O-2eg-mEaqOKHvq0+J2K z2lIVyB{>B{T}Bj&vu?0t9r_rrL$#*|5Y-`%)Fy`o2HmvfS;7Y5T`ia}pF zB|735diWtED1F=niS6$a)a?~n4ux#5K_Gj{HlYM>iUs^;ggVDs)KIpPtR_z}I z7Sf1`Wd&qoL8co_N$i!>o!xa8cXK|oGa%AoR<@^~(apuza!;j>HTd-e+E3XYTHSY5 z9%Cop%OKg`IQDF-&UzjHz51Bi^ks|th_SONB&Tihpga32#-(hl4}I;}U1|G8f{3W6 z$nod5n#>VDa+mB!J{a9Poh9zc@#}brPDDYB%df>K$%KJ3h*=ApdCHk>mM8r#fW+XXM7kBK-#8pVHN9m6e~7 z@0q?LEBQT$$7BmxWaXzBiywwk8`2S9&uwBh1M1~8pd}Y!n;9t|quzev`VCQ#&e}|z z{tX!}yl3Di3vY^1xL>A3MLe%&Oo2 z3KqZP$aIxLi^kakla@>=O7^TU^OmZDF5|Be|2l_4fJs2zg5v)a3-&hx2g(W*#7E$> zCpG70O6?~WRgnS-gch=^BdCUs06q{eVA0b_Hl{gDQ{oe0RY5aQ&pHX-aj94pHtRrM;jubH-hWo;_5u$7 z_i~Osi^S2H5>%e7&w@9liBR}x+?%5PA(DJne%6S22UdJ`A^0WU*`1Xv3Lgdm$E&4- zMfy#LroetWYiqKgSpX>5Rvft<10xevb=~fbSnWf1T}-}9`C0FKTOo!UJbe%pN}Tt$ zI7WRHwzaZ=!4#Dih#$ftWfP4~#(uLtEOh8{D&S}XC=N%Ekt`A4}ml zp0H?ZT}q?g!SoVj&1ah44hle+q zf2qiB{ZaLiiu={na5zOK`sn;U*M>mhI{rxuh^x7@%FFMbSOT&5L~9hKjDI9anwVMR z+rY(nSE6P<_hr?JfYT9+Xjex=tGoxJD<2{r+zHUYNQ&;{3{lg(^NTRG;wfBuv38dy z3`Tn~6dV##x_h@QAg}BmYDX`Xf53w5@2^y#OP0^WQlH(D*fkdVv zIW7f5q@MgPbvh_Se_N}`XOu%(NR)BsA8Lo3fi568P=qd>s;`dSs{xRcOk=FwbON6B zubPNz1Hb0qtHL{0bO^Lif3%*I4W0wY#5fZ;o;!@jE)iTXzGtdkdLyXKg{I7A@m0CJ zFz~!qdQM?bcI$u?DJ91)DhlPRihwrL~UBs+|wdDH$zT*?U1HvREA zxrZ5LdcU=a=+N^W)fJQu9?Y14O*hf+IxwEzeLCso%bmR`bo?V~sHtA##-ynP$#K#v z|FMx`Q^id)SLjL_jYoqc0b2pp>shFChOGiLB=H2J111K0v}$CM=v|g~UJ5H;DU+9XWv>YUBR}Ycwt2?ed%Jm8 zQFTjk&w{BRvpwKjSOt&ns{K*@X*Gj0^%a9Gn?slykV7!^B~^W= zDjllno!1x+*vXWq_-P7Q6=HT&vY8}X3-QKE{RjV&ZMJ(3AaA&Jg!5{3Dyvw@aJ<&e z?NuVj-`($GT&1`%>N8fDdg_ydcg3};s;?$yBrIDl3cPZTRTb5R`NJK>_GYT8J1rTy zUx5#>fn1DLmQN!?=1Ry2ZbR97M6j?<$fABar<2%lelqtnT{3=FGZ5?jp4~7Phwl__Ttb`~IYcB6|Afp-o+vBH44^qI8Og@{Cw`wlNMG9k8bkTt z8smA`R8xk1Dx!=$@C+Ov$bhm~K?Jz*K$`_Jd27vd3A)wIQ^pj%Glv`_UIqs>-MM>b z^vo{?>E3ex#V|s?!X5Oc<&@ughKd>Xe&~}=l8R?dJ_Mz<-;N68SIb%Sx*r(BOFuW$ zpPrK$#N{!7YD?<7kz6pH_)1uK=xf`T}HT$&?x@zJG;?^l@5gP40go@ zu%)5@aA6ng+MnLlsY$p+hK*it-;!!RSyatFy6egB1pag;H(*CaqsU|rVRpH;_f(g8T?t*E}%lSd)U|u2*T}j_YHPUG-NIS!m4q`6; zaB@A?l6G(}SdY|FAngNaJ9A|!Dwq}`j*enBxwQ%J(UZ|PqlW{K{DH_3Ac@Kf*rq{6 zVlYvaB9kYOJ56TDZO$G80N(!=&0oaVhlD$3N0+5Q<6xwI==2zQhsB-fOTg|C{~x-lM$LSrZi+f!HI5dtql-e|+C#uS2oxj(;Km6JXH4fYV1OBY&UfR1Q8@v6 z%TdL9sn|ohtb1d+f|QZl4+45R=5nrESk`_}T{sGb8K}jNpvPOY-;um>Du}&Wsk6Gg z*Q-CIk;HNj@S{@%LT;Qq7d&ua=G|DQR&7nrSj|GD+21`lX@i=BNXnjfl-Rzxt@rAi{_@j7$0GWlGa3*5=x3D#JJF2J}M%4|0j zm@tK9fZI4Ab9Hj88Xrn)36GZVL6dKkI&#bY*VSt@!o*pGLNUNBsK5%vfo!D!I36L9 zNboKJ#_lBBHxkoYF71X^=|Jms&>Y9_gHF-JIDt|un8AlRs6)*au)s)L?{hzeWQzv& zQ$XP4)&z1jpF&;LJ4aI=Ww>B0xe2jO+)0oU9dc~KaZP`!T4Kl=*hD%m3Cb=FNvRH6V|ip=eqG%h3bfI>`dZfsBpO8ChlQ?+QWncQ#$W8YZA>JcF+=5ZA1qsJw;_5CHC2Aey zBTnY!=1lp!1-eueyL8j#X9!+4SGbVNwJoCx0pCP0In|2}PxkJTd#)qXGLBTLkEZX` zIBK1*nEQ1v?}$#VnTzI<)WX#4**Plaeedg=)TzMsFrVEAgPNXMqrV#X=EhH3d`ovYaj>#T{@s&DmU|}U3?9z zmii8_qI5HCH1hVo*VQw4v^;3lyweF1?4Aq64%>#0m~mH*r5$L*CQ)_@*}?&0TEaYP znMfy+Stf$>E_~1`a*)hxi0Ze&nx^p5HBmM;$@cecDp60P95Rup(zbH8D`F-J=u;%m z*)!B-c!?Zf<6Aykg>vzeMN}^*C!+Jf8Gx!mm_g;>!vr|{EvOE#nb@mE5~hBis3*8B zCBVyauv5s*7B|(!0MQG&upDFZEs?@uRZG6-fq77NvWeP`AS+YoW{yvW{udaZ z$+pS_7vF*A=92lqY&LBIcp8Hgb2|kENvJz$%&_u1>|8o31`wRfdzAU-+Rv!DBAj-! z+GcfX@SZS*D+1+1%JFZV?z|J4sF(E|*lB0Qn7$t{A9>q+L7N%2p5S<=gX_HNdG5&z z`r(NMLnmg$Ja+G4;bGS)6XVhPea!qAsTrkA>!ZOT{OHp}3S~UB`gM-EUBrv2qr3G| zPl(%z)DcTx*yWeyq?ezO($LuKQyDFaU9M8fpFQ2vc{37OaoyKEPqA7YWWiS*5KIlw z1&QI<9d;aul*lL(+*xEGU>y$CBhU_lGN@X+`xL94|{T~9j7bg^qWEsV@vReL&C&a_9& zmHtoBrd0x0oefg`&sIYp~-j003?XMK`QOdCVYFV8^ZZwFn!tc zvA)UGPgnar6{@XQ`&BHp0FYiIvX;m$a4A8o;}~eP5I|q5lV=qVh-E3k2T@}P478PdtQh9riOEN>+ zS5t7W`tF@)npF~Q_li*37Z3OBFB9t>`RW)XaQ31Qv$^0_wuwYdt6t_S#o~?ez=nw6 z1DaQI7S!VI-O3#^UAbIKcu35Rl(slG#tbX z5`ap$w+$;05^lSpY?jCc+Nf-sZeU7buNv zC5H-fr*5ObJ3bLmSy}8Zn4g|tz1}DdF38<^b%|Xo<<@Oo*?kWaz%(07VdxCuQ9{b> z{ubCbz#9Ur23iRSPLE)qI)s5(0661OD4?{c+zdR)VPr59A$ovSrcg;t9q7`Ak2Y_# zcz~m<2EUE(eI8@tv+~*MwbBeBumJdJsAe`(#>I!%M6jxlDAsjnR4s$kXlSI&~n z;0+n(sN}MV8zBg~a|5=rF|qzn)dyrCj}BbQ3dw;E$;UMLRH08Bp7w71%PFVFIcfJU z3)=io+65H71NwNUOEfe$tBaQmxt0at=cmvxtWd2H9y-(5>r$%YM9|7mIQ={H# zvm90>URCbu9Put7dz-T9h-|uTO0e^8P>+CIK`Z0(CL_bCk}5h?popND#?y1u!gMNV zf^khBkk52VEughJ6_y;}GeklUf~1Ygy`=i=5|3-QLft+S8yibKsX7+Rt_KjYRiN`SjqZ;edIxy$Hozx3-ZxEWO9XF9=}UfDx(m5 zl*V1t6d$ffVbyDSC>cg&9X#96Vj2l1H!bb|nD z`M6QQ!}Wb*vOi#dyhPOa@Xh}MlPZ1!k0@jCN9a^0S7CLeyqiw4$%)FMxm1R@Yt%S) zRQHW)9Sp2EHFyBsB7uHKOrO{qA#honC=6mX06WpCoPe%4$;bfqI1mKd9AA(vUjB|e z*v!^uy$t6@teQai{h^j%<4m)Bj7mcp^MU|HEIX3b_B7~Lx^8`?<=FY`^pN{P3Q{^} z9C5iXdm*f<$yfJpXWFTahHIz_nR2Pi!83zigrAYZDXG^CB?LG<$lX8rTDkKgALVws zt1{2izi|g}?y+$*P>ked$7I6;eWrbrMt4@tZth}i*iTPfwUbHWf69Fsv=W9K;p zoMEU9m8Jp;Hkf{ak0d(!z~rF<**bIdgr$*lghx75na*TT*fNeOF!~HbE#vor$_5V( z^Z^9*K}z*}S>B+DMeWw$@o)CEMjzF-r+s`I0O18R{3pD$2$@e#Vxf+p&xAtW1wI-H zD=XYp<5WHBCKcJj-R$AYFzI)4VdO8=`y;L<*zuMNF?FQMRnmAg4oC%~#WggP0ifL3 zVES|8VFQW|&1;5ynV75h@XR~ekB5FC;cTUE08hJz@XF=>r;G=cXMIzJCT3i3b_C-y zgUW|i&Zmv+W?hvpF9BlpSYMitU7K+gtNmb;jE#X2Mv@k9bl|+ZpS~oDyldp;Bi53pF$@-#8dQY}mf=*ikn%W*C_(Mfm zbWT{cg6b(Yf~53D+kOhxf+H|B3SWZo_HK<ro%qb+CablyHL(N(D^4FRQhV|+^f0OKMsDXM+7=9Chgo(H#UsqAf zNB=)KaUG8NFB%F2H4Q;W=XJ7`HT#CBOmG+?pRkdlxm3R!e&TpnWB2HSGqIx1Q4L=t ztehSAZba}U13FsRmYdn!{3!6z0B!+^42=B`Qv+}jC~*K(a&RHH2CtF0$YSz+#6XF94vLD^)tCX07{V9dM zQ3)olY%|83UsaMXtEO(ZU+h^8c7_!g2!<%ZXcirs99O3A0GNc08_)_tY@Abn3dV-l zAatlj0{g6h0s;pU9DCeDA-fCo(GakZ*-A+%?3q)}z|M=Z8pakF-{tC$x>V8sNW4Le z7W+@E=rqW-UcU+>ASdxTE^rb)$5_6+&LcfowF=Hg4@Vke9qWoYb-Z2m0YVCBoUW_T zWO<;H6O8%w$i?TYEZj>1#NaKG3_odAZ9q-H#zG;5^e6$GIE#Kzx*9~@)MyOFB7s8I zhAD1-w?*$f16nM!o?#2JYI>1C;qLnu&Y*7OLraJOe7(lvz#@2q2^-bDyiT*!j|n5( zSBeR9yb+&+AG8XfJvH$7k@g#s8*}nE=My)on&K)WN~n?eAL;118|h^|!k5 z)dmBLj=H_pPxf+AY{kQ*x?*q1=gRS8Iu@i(7d*##sSP&2KI-cLPmvp}4j4Sv>6^$! z@iHUK>d!qPM4cC$$;S+tXxYVOZ^4on^CD+EtLB|Pex|it<Nw5Ipq?3$kN(r+V z_d0ebRwXsT#tvKB76*b#1q5z0WHMLpjHUDeB}H<5UuxKZj(wbe2r7=sAN`IwiJ8Xx!_M;0D$_BW(tHj zjyBn#1$lf-CqI_L$89SX);K(8TqP=DXLrOWhGLU^TG@fFyp}~e<=x69v8j+gns@sd zZ71Io0zGwisdq*-*1A4AtqUD&)xWW$=#270+gtEO?iN?h_! zoSz1skj-%5Z>fL~wcqUu9n1_F<_Tn8r5j^yHdR%PlQUb+@I7^oovWj+2PQ~vV@fNN z$2b5$iXR9U@?J9d7ziAE4z9cC&Ro2;$o0S+rGw}Ga$dM70u!KCeFO;>a3}!w9C4Mb ztMCE&FCbeb$M3)Ig5a##{BF~YW>2#Xpti#yE_l8QGHqm}jmC4452LO)Qw4-k<7OW> zj#&I~QOHw)(Yw_+!2NY|&}tK5-;TJP*N63aLk69GZdSk=C}|4eH#ZY^flf%5 z8ZylDXjf*G19|Yl&`eUA!?&?>s=-(@7rkkWG8G>fK^;N_(ki9QfPRD$wukMT1+Cm6 z!RU8^dDLN8T7I?f$9vtF##805iaj$y+%GACd48&bOrSYu-4U_<7Tr@$F61P56+&0w}z5(`p?d{SND`EoelDLA85Z- zZ_z@!v-_62b(e9!n%krnVX!0{R7opOSEGuBw#T@jOVYlc3>L+k_!x}4fDVVB9jddlwxpgde zZ_YDA_XhIqL$9lsp{>9y&rp{CX9^|OSeJ2)euNKm0PZN-T~&QiPcS*I6lF;c2Qy~d zXaiBi`ZOjZ7reisodNY(UQ1tlu0a5&(f!B6>2lYVP*ChEIWCtjgBVUBT9R@p*s zz1r>aETsEJzU6PonAANBn3erXa@ z#msVTzmJ(KPx;6+b>UjRQ^S1w#pxELZMq~MZr)62~=NX1wY7GyVSW3$^dsCi^L zVo9`Df)DeADCGl^CJCk@uJ5OIRkCl%1tf#ynUq>S9nyjV-*e6OQ2+esu>{WWzugc@~UcPk?`2i9TFw1 zdiB!Jd&F{(nIrjyc{D9)bZMe7{%#qq8;aK+e}InZDG3Cs>rZJ+8vpB;Z|2cx`GBS) z-wypfFa^E3Rc~+gRiw#l|NbDIRaE0ve#`zuR_AYsXgv8M*`Ae=G9qSqN2aCpxiVtW z9Axks7HEOsq+^k*43w1~e!#06RTc1AOg%vjj8emNbQ_H2wBDtYyGO#D^7bU2Y-imj zErdHZ3>H5?w#`)(?1?wcJb=}rF4`KT_JHDnCL6Q&F&VQvgza{5cR?;0k}Z|*PL&RP zG^|`UdEcteX=J2tWLP`Tnmj5;tg3S?cP_$sp9l=34kvN$CTp|%3u|^u{%7kE;6fZ{ zL2f}Tye`4zT^qszV^v%c=e8L3$x#P{N~c!sl7DFQaA11PPy}7(xxk-AHk=_y!O7z1 zbebi1%53dM5aaX=*FM<%h239>2WEWEt7VLaol1N0bojo)v~p8qH{LoA{@ryexh-I( z;JKH;yK&n}V$LhX&bJt?Vp}~VNKgdW1KBvRxg#7ZB*fH2_8~VSD!ARsp_OD{M6gRI z2g=UGk{l@rxj2t!f9`CD0mUBjLAB8pfMW?9cFpZE-7{A4MdW;-dOJzXBdQnRPHsS~ z0r0l5w|upM@o;b+a9knmCKW)M>hy@@fS9Y`j1ztMrmBbGyXcSuTqiKh1$q=5jj0cW zb+j4fCJbsuDb7TQ+5Q~HF_dI6VAkRgNY}v=HDwZg9J+KFLJ-&N=uvf}kn*daof@+W zVr5>dUcWT$>Opr@e2*@|okd6o;#m10hR=;-Lci!h@#sq$QCUj9{&Uqj$CpI8)+h$1LJC=JA|~`qje9_<@9gJ@ByO&wXRQFI>eXj{pB(DC1G_zUt2%id^0~b zl=TD7*lJb{^^m-4QGGb7A;oKNTkr030>2>@A&r~DSf69jD#Nd2Z@qnWB#MvY;iHj* zzadiSyZn3foS$8;%}wc^_c;F$s;?HtyZG`GF@;ud%CPV23> zm#gwy&Kab4`^6^X-8*Pp_iXkq_eWtGQkxII6RL0U(TwEK^IxB9HWua}NXWdK%c`|K zoA|-x6V1Jg>o`=JzL3m^Gi!N)MF5@c_Q>HJSU` z<)_}7M-Ns7_J+`Ml0OFJ-)#W~em&~*;&Ic6>yIxBH>MQ^HBDIYM&BXH8Kt;?-6R~- z>HDfY5=z~;g-|rQY4y@ddm{LK#ft9Pks;O2f$Anyo8J!%pL;^x;RQ6=iUiFAJM<`4 z#($O@8TlbgF)`z4@v@*+0Pn$KwZ?%hiM6X9f^#G8#&5q3y`ME&ec|YpRUR!os+(~}3}ch%hrXkDV_#DIVW%(M=!Xj|kg&bB86TB87kh6%J!!-6ihs_pv+?G? z{>BV21JChy9yoI4Lp5ZR$8ROEUAX_I?A|Ea?57Ho&ytnA-F%9R$9=dxFENdhT-%1w ze?7vEl>{2bysuS_DUo$QmKJv1(1O6fXfLfcf6Q#^W@0$JV8|yf>#N5u*>5_VYFVR( ze8v|Hb84Q-(k@s&{0)(j*mABViJm%}mopvKGO}nSe9xim?8UJ6ZAme+b&NNtC0_nt zK?RCIFu59GYPy&Qa$l1WrVaR9A92kNSg!3cATSC9k{cQaNT|8uYQ$`ulbkELaKgz? z9DL@ zPLDzf*_Uk?g1HqQtLzc)8}?8RN%$byaU8uJi2x6D7jy=ix5k9%FIVB_K>THDfrdO| zOE54f`^RehXW$<)S)}I_!6=90Id!U<9NSoB?ACA7^^usl1r$9b5bXo0>$9=bF7pQ1Ha!%Z?#MuMzFfu0>bR|w85 zhb{;~)O{$~!=j;t8+jb=Krpcj*TyK?C?ppCTD4rHaC8jrY%YtU^_iip%IAQ~X3`Jo zjb)J|q!ATq2Cmb#f4z!OkxsiySdrXJ?w0)3(@@mn!F81+o|Cn5YY>T<+i9YwBliFXHQFIo1>oj1$h z_odFj%nn&ZmEG@h+@9|D-oT{HAvf}RmqTv5hDDX9xy5GzndEx28cmYw0hzK}{4^S^ zHy?4EkyP6L8{!xodvnU0qsHshUGM94@7s^S=aBZxn66~Qwwi*;x|MqU>qSuEzi%0- zp0!*`FX1g=J&%p(?nKS<-i|i7HwV}!{+!t2wIXrig7?CRu}V|jJaJw(yy^iTN1AK2A^ zhhL{m20xn?pfU}Z;RcK+6wkO%Gr=F}5;f^!wrb@06%FKBj$BlVbj|g!08jA4F)+qi zR)SSfKrk)A_Ryhw@8T~dSvFKn7QA`mQRIT~xv1om*^P^&(%*X&SrXM z&8XdPv=>;jGiY`BT9oo8f6Lz}Yt2=ldaZqKe&S+XmXGJh_8Od>=S@b_5l;KB_nn@7$p4WJ2Ju1V1wQ*R~h{Z(svtZqjB zuoXF;;*f+R<*JeE_=H@#ZQ&N$w_m-bS0qq8$92avLcJ8fXeP06y?@h$7cgIZ+&ey{y&G1Urq%TO*4#mJ#}8;o-LV@U-rin^1Z&y8VB2u&&FdwnL7 z;P;B=y61xTk8d)l1efO5&Yvcf5?S;B59`o+Cb!f1Re|8R?2%4qQLTUG;>+UF%)t&m zJC7Xvmo$ZEIu%Wdyr1139rBzG5EJ*S2<21v$2*-maX z;=-!%e~mvIcWxW(^4rk}*gy{ERe(<(D=G-Kz(VHe-BA9dZuGNIz(7Tr0f;{v5OBP# zc=m7M$V`kLJuz}1i&C4zoXLz^+BULT-+(5UU1;0!#n& zX#p0!#=<9=PKtAw@Dvz6t8nPVA?SX05)`qX#HG%`gifsnRHPYZ-_^;m-*A2waoePI z9DVIh$;B>*@#;fWjC&;3kuML5{G!n2IHh}Teq2?A%$KsZWsSc`b%-xvQV*_$?cIe{ z*FtWrW0Gz*+X~m_Q~VPb^=!Lxs(J&%uG>4Gn+jjyd$?SG96uc2lq=s&^15GTf0Fe{ zr?%W@rDv-29R8A8e~Wd{eVI>nO9O|(gAH9}%6}bpjf#HyMrYpoczAIQl1pwT?5AVY zFg>s6b!?X{$C2U&zM;ECAA}UH@@BdwbX0N~L)3*$I_eWFM=!L+$J`&ZJwf1fj`O-- zU9PrE2U#QiCd$F7d9k78C@t~VYax&RFvaJj+4ItNyhRiT<1~|Msmy|F#ACcN^$7%o zrfIC@K8SDyy(aIbX5}8`{7ypZvopEDkk4KTQ!hA80#8$Jrkal`hMiAIRvdl!xV270 zL*GE+{s>GK-}Ql|s35;^X)*V!2zl;)eo7$@ohLq`>=kY>ah>uTl5Ses+B}(qn5-^` z2)SbU`XwG#FgcvLqJJR?*Y8wDIU~T-zC5&A+JE6P>?^v}RWUi`nsf1&CTbdfGprtHDg68?FPS@M(c@@$&d(VAo(0k!IU3cl^Mc9K}#CJh-_#LAO`ghPwyRGu=>>Hwpg zDJPycNcMUiume)bkQ+Hcj{Fv^W#8DZmUN}jGf$zl4kMOP8&T_V$?#|N#e=R|F@Nol z8cBF!D3^CiLuUQqrz;{~=+Bb+p6+RnKGalTbLy^;nN3unz~5`*QGrwY_xR5azUSI3 zD_^G!#iVQ@?%P?jK&n>YE$Q z4jDZ@X}Y$j6Sy^+a+}J>zn0%Hc#>augoHmZ*M4rz^w$xiIe7Qhl0iw;A@eOo4V zzaf_z)&^h5upIpLuqtZhzafdToiWQo=QcjYbY-oX+_Sv7&UZMk2FPa0AZKIMDbVw z!vgHGavBiHEoZlkinFxNUI~)ZMD$#JBdhwrdIpyn$%Pp%q=0`XsH%$eRFAS} z+8)}lesssf+_&{)@J(kvqf*7M4ZL4gx*Dpx_a%Q)7UHga;YQQv^}2lAeyioup?Za< zm4!4&dF$AVQa;Rg{}h%UUuRU*1v&2aAn&M?4;+4Vzjt?>U@@P*$+h-?p1Oh``1~vs ztK7W!GUYgFL6!Nu!N2b}NNf9_EaI$wnpzdcUUs`jH-8`G1Id>!>!HFKUz;v`|Wm6{p4B zgF}itK?4M9p=fX?NLqMtD_-0sXp!Jnv_Nq!?(Ptr^rpXi@Av(aJbBhkR#uXkbIv|{ z?|Da)9nEgGY^>0c!s92k);Q;qTHK!}G9FnkrdY9-U@3&JjWguT%HAu({@PCtx;r!D z`SyiFRqeP4W*22ho#q#@ke*wX$yaS=D@o#TB7Pa?htuj7%Roc*#mFepPbyfKfDW^a zdSH{heKit7YvipGHo z1cIE?vsQ;#iQh872$VnYsy-o+DC;HvaA1oIA^MTpBu343(L*OnBUNdC;@`krFKa1M zndvr|iNzdi61HlrGp`;fPNb(LInJEMK9MovX|+ryQew|Fp?h&R=4@LeB3!s2NzsB< z5u;AOl10N?SsJ#>Eov_&fUQ)y+W_=fpXqetGUX1hS6$2q(r+@Zns{1Vjh-vb zjskEu{>vj54OlR^Cv0<~xH1R^>IM)lBj|6mAih#{CrgRg8J}>2qU4n%K(0Dl*FD8V zXE*fdCk9Bt3o~8wKN$kF*aC`Q?!b9IA{Sc41UqU_Ltmbgf_4?0gqWLMe%ehm$u_|$ zRM)(x!}JPV+J7pe1HbYA2Ja~GOIX&pEUh}w@Dl2}oIR^Gd=qml(}1?DU7I+tPrmY) z{tFxbJIw7ww z6|*psE63YDqW=KTCs28JpE2Ys4BF7X_*n~&0JnsVXW z#oXB3PV_c)p&|aYekT9>oO^Po4xx+Oqy?r#JDBLfjkuW^q&Ic`i^U0tc%2ewozI{O zP+p$2$dV>nUFJ}?;9VcuZEY@Op^(aK`*F#Qny(ViF8986Q5e%-tjzXu?oycsAKQo= zl$+qiAwN?OYy(EPg%*j5`ju>0yo0Hb5aj7q+*{G^@1hX%9uMdo*tgSbn-i`O96 zu@`U5RnK{#A){#eKYQ=Zn;)|hP<~x&+GtS9BOnUVUQ}}#PG4EY8%zmvwk9VM}t0kDMH7Ist|poyj4BV!TWpcgKGmp~?CJ?2x2d@Zirl zD3TexN?5Ez>)w=7uD96;ck{`sR{M-t87CaCWd=;((+zVkSe~UaHUJScGt*}H#2H1p z*n$9LAJ=J6x!sb*>UcTlCv1HI5+^pZl};_AwSF>8+n{<6U{6ifZ}79_t_&s@wb~id zjf+QtkFMW9IAme*zs5E(4fZ+Xd%?+^eq?j7FC6c2$n7-#axkit>gLvM7JXsv^%twp zc1p4Jzt5N`H*0TJp7?1>|KqsxKKRLpe}%?^cC*G4TBJaDEJg=xZOKCSKN9S3O^i4i z(>GDT0WhtHe_+Mb6=rmNHoTx7CKgON@}KrEj6(n&ZbCF`F`JffHm=L7f1C)lZy3WP zNo6e8S8Qnn=~G)&x`MP4k3WzGwCI$ySjHxYr{QYJ;G(?mjFIjg3)$NplmY*gYBwQgnL z2^!=WXUUb@aK3t}h1BKd1er`BOZD%YyXX=#84_Ni7T8BBkJB{UQa4W zL1kKs%JbP#ej$$A5KX}ioJl>ofV5{r`e5}+ZkY&N-jm~rm)g| z!lmKg2j0w))9h_N<6?J{?u3cMO>UDU-!$b&Qi?S(4Bj6`ANaM{K09DomBqvk{u!;i zWt%bepVZ#tj(ey-e-VE5xd^`ElfjASjvkT0yB|y4P1*MFr!}J4`oNVa< zGW;u3=A4Xg&foq52 ztMLmFuGme6!o7+nQZdaEdhJF6V_dzNsMf}9KgQy#x(J!J6bfVBX0W&OkG%eZbp(OrANdz`ykeJ&|EHH;~f zM?lmLt3Fq#HNrzEJ8#fYMlQrnPDRTbv;$`oPe$+F7uUSyqtDIf>3`@_+EjYzy)T{= z`8L3cHvEg_+|DK)^)PXxK{ZAh$FiW{s)B6Ae@X$MiOghge7#cDV=V=95h;T?F?jaq zSdJ0q=eyUz(&RfzX(ZU&PG+LHI1Ds5I6{h=&hK%FWZ-`@MK4 zPXK`}`b-mA7k#d z%6zr8+-V@&*LopEgA7j=cSnG?m+}&GZ{oR(ZBqTN>GsB7z?Y;_J6o>(au*(dG_X&5 z2x0J-uOr-EDm2W3*=iZ`LYSX;pDf*Sq^bNWGF5m6+D+|fIXS8<+8+6l{N(zZ1MT-p zv$s_4MPX9=5c!s*wM%HuN7Mdq55^*NP>oO^OR0m{cW%_DqMA!iQGl{jL&n>^=Zb~LRYQ_4-7VO>P3 z1R{O|S~+8(E5pXf;k7%mqgi?0V_;&;ruYOSJ_cdx3aobscTv*~JR?q!i-Ilh1(b{rGn z8zSu6t!&~TiS|2L+*^Y6M}7a51NR?*7cba54&PeV^Tm`qs(f}b7qYaAsI%0*XP7b0 zzIN1~d8hZI+B4M2!UMNG6Ph@wkX-y5IJbB+gSh89^~?8&$XizZ!I%D%o{Ig+Jy7$( zx|EgiDEx&o+xlS}$zI$S2S@vC$#6W{^Xut+;ERJi?)&)n>aK8xAaL7E2L`uq{I zsUA+oA}pnUOgUcwHStv%weNo&R@CQ|rmzQ1ZNIMF0nHxL#ZcpIMnT*&yVW$SHZtFj z|9Sc)#~@euQ?wLyl*{+6<}AYfBE8ns!?8D;n*pz0peo*ny%ZC)hCz&BHdyw{W@X>4$9pKs2gJi<-8)&tu5%1!ERBX}V z63=w#lB$`I3d{#9Iy3Hw$|K_?>F(6+#q*^M`XiBIgt8u>NKMKB2gZ^|f}MjwDj#XF z9T*?>Wf#ce3bi`~eD7}T|BHpveuPJAm-QD*m)I;ELznvh`n>-Lz(^m}go56#(J6{J z=Ce(`3xSCmCfmI}c;3Q|C$8&H!0w#xatb#8Tnntzv7zH`Cf8aoEzbltok;KQI&l5j z!B}zzAeUb%Y6D~{q~kp5$E`w*w4#y zIA3m>-Dry{UqykMjsm&C!s+OO&tEJyqsjWZLmRjN{uEhEXU-7)*rJga;GF{-g$9NU zHh>|6>2&FQz0Mx$M!>!|Fi)2jFL>Ne&F8;r4$;&hzyT00%)Q#{9-CLWN^laxu$s#o zM4I@4wH6=le*w&8P9D$=N1u56a9tV3f9OaZ8^67$H=#rt8SN*(H&_A=X!ke}`6$1wkT z`g>MsF9G-)g0+u`updaC~9t9NWd{yL+R*SW#***yazq zeVB>|xcUKv4IdHti*!snQ zQiDqVi&d`qPf`ba$nOL3s+Ly(bO{ z9%_nPj`5#h8jO1F|E=G%zC{0He1{U6^*LXXh4$iA-CwLW zTU!1utMZI%nu(Z`JV`Q%kCQXIY+1WMFa?QbI0QwC1nU8<7<2DE$B8to&fB+X-$;=m7mcmh^zpo@EFO-9MW z+C8m5{0Xs%^DJL+3YCg}ipeU7YiG_$&EAK@=OzBWGSv zl@qFOj%ibb(Ar{a{s!=7%|66SS+N)L$D zR=#B9Sevu2_4A7c1gUKZVdGFx<9fMry(B8KMkfH4sns}C(0KJPgjSKl5*;M=yfG11 zYsIQusg+vM;^{vK1>eDM&#f^psmGOx@2)eIDJRKCAPTcAUTL5chnDVzdQCd7?-;s( z#TBhwUi(_*!#)$b%&)7h5=^cQ4JOVks=bNsPVU$5p^U*~0X$vB;%}}gt^f%XrQZ3w zP2p>M!>k~L37i)q-@M1A-WL|DCMCWeo)+T!;6sl;hz9&Ny=I40F(y501RYEN z0mlBkFV&xlAOeeEYKv5T#;*fvMDpu3k86MHhGO`CAqBk86k*zr%K}rA-t_R^Vw`ta zUSCq_GW{=7>U=sMryx-vjub}dIIDf?dH>BU+@B>oZ=#I{<^6;0=5A z*DWQKBKQwkpA9L6x6h*;i4iH?+BNqn3~}8sx049=J*Hdy{?aBVtR1V|onv z@-J41t#gcErI{`0Plf)`z|y;^JUK%{xi^Ngf_-G-kI<5BsNNw~4~B`O3e|B^W&os6 zyV1?Yg}t>ioMo5|zh_q|(}pF#ms5%2r6auge)r8IN{oZ^bh@GUuN_fbYgKu(ZHZ=F z8FhN%gl@q=bxihxVZN`iVss)HuH}n=f;+9t`~u=fqc++OL9UZ}uGvQGPXOBx+3&G& z&72hW)sGCz&Ab!WZKw|_n_jybi~|8?tm1wq+MlXle}G!zI^qw z0R4Ni*#Ptx%a@|5jk(bT)2g|eQfy;yq@G9L3XF%ic$Z+BHy>Cz?<7*WM-h0!lKP8(XE=K!s){KX!68Na(Hu@nIScL-JoHSxq zzGW>CGecq-eJ3z|YW#D38?#3bQPti>WY(W0z^hQen-k3Ywc!iwiiYl+W562U)P^Od z2H?`aak5MA8~@)cCd3u1a{XCu;{@Q*^uQ++G2F@&0~6)9Cn~$&VbUIhA?8cepPkO+ z#)aBF4|+kNb+j?#JwmeIAGkT~c(;`=dblNdf{|T082`ocK_RYu5|1y*FAa}xP|0U` z2UlHL)q^)$8I}isvEJP87U?#jZ|-*mn`4&tz8Edd{V=QUi}rqM){58J{w5Xbng=~_ z;AEnF8s}nu0aQLmxQih)@_5;7OPbk3s1*cQ&*Is5Q)|RV-dNNAfknvkj0p`8{SBPD#HqrLbaOj*@fsqXmxIR}BK8kN^7|EwXQTSFJ#7!RE4Q#SCcF z9?wzWt%vM5+BdINRlKK<)gK@g%h6P9hPqHv%_tP?aX*Rns-P2jX&-AyS}#M7!L21; zmTK+AfPYMay?QBAr=#tzDg`IpiHKf#Rdf#ASbT6*>)wD-6RI3*bF`rni94c^7K~*J zQv7lj2(a>q0h@08Qcibi>SJ}*H4rSfTIcH?hw)&pvR5491#71=n0j!KQ(Wa(M>gS< zk)7asTaC|TDX`TtjCG%r`3-_10@$ivPlNi#c!hYngyLdEq$#;#!pcnRr!q8(C2OhZ z84U(TpliD2E=4n3Xv^$n^>=$5>n*Ym!HTHZRB|{EG;RdExSy?z&Vs z>ve8GjLRvpMr%6ZQ2iB=clgkldM`QfgPL#qddzFM19avsZ~d+C!vZ;a1>*-EZ*q8= zg!Z9*^6X3Eo!Yn*{i__@M&aATzI9WsBLJeXG`eQMbYho`(2~C0yZ~o;q!B zm`J7>=$I)@B!~TNS}g-iJg%jw#cOk#?a(SVp{rC{IBy>z=H#Qx=wiW;qOK%F^@`Z- zs_+$vI7H+gHhmPBN7+X-Q1nZ?1C3%x&j?~IW=~;LeprzckBkgIaHl4uEUqEL-4lN2 zUZK7=4OPNzN*(LTV*y7FYzt6)`_xCDYR)yGWs5Np#K-7~F*XSQNT)BAF$zaKh;JI# z%u>^X$mKxG%o|F(M8nY;Q_H_td_ooMwpOHU66&`Ds$G$nCFkk9g*Mfe;+k$wz44HA zrG<(^&E&!w9ScM%;oHKb;z2Ek?9ZaE7XsM%UBi88e>@fY zYc>Y=PfI%yt(?p-NPIr44NLY5IvIwNh^gpj#xq<2`5skSNr1U$az*$xwcA0VR+0~; zf@kJ{E`WA2ho9o+R%+X?ibo8OrYq#0$5e8v)rJ1IxNuS4O#5;l65v=r;Q#^N>}Z>z zt-`8+?CG+(^<&`(H(%*}mokCxJ6y-0EV;EfwBrucBFZeh9E{lYX6n00$$ zEPfxvqVQB3cQO^X6*jr{KKzTb9z8jvIQR2x7sc0fosYdWf8ORzqe6mEYEnZ55d4=Yr~%u0LZt6i`b{H zI&jXztd?_Vqx5Qy%u&VdLFw$Xy$e6zJ6;+F>Oy$(0~7%hCe|Z_H{aq0XTA=7GGg_9 zBrXJprA%xJ7JvZ)s%r#0@Q*=k3! zE>wUn%JZZ-jaJH2IujNsC(o7-<|F;aZpI!`Tpym`FYIpcQ*RA#qU9eg*g%KCu z&01P`hqx075tbkMoWwX4s>7xZi|oTwsgn>zCS7oc;zGe+i|@!wVHtapxmUbh8w$LO zn@3Msk7I0OFK*m9mDxGDSS`JsH?zs@Xd!zQXu|7?HDQlGe`HB`6y`pNTs>)JN&p9F zZTVk@(9CEI(g+a@S06=w0)_zV?0`Fgt#3Sx`P9;-bBtZo*J~v(O95*#K zUDo^Gx(zd&gD|=!ADy^)aXZ~EY@DYwQ#8L})0p@sZ#QRbh0zQY+CF>!^wfxfFu@MI znKA4<VF8q4s}?s!GS|gjzq2 zj0v^76J?b^|E^64C%W5EA22U`FpUpWBJJ*}OopN`R(+dPItS5-XRYq|orEcHe z)CHf(UH53zHwH}LL5de(;*OUB3`u?5>blR}s$4@033=R_GkP;N(BSl3LaHon`Nd9C zu;IZPIl*bC2RZ{^K0XWbBUx2!yvC3;KXaSkZ-uhKWm=C_^e>i+R4v+%J*9Pm+m7-r zc)aJR>m3@fhwxg4q>V(&veRMX6w|+(4lhgE;b@}ajkc~-oAiBOU>a~z^rt{#efPQ& zG!QrTPhufq!W5UgT=6F+8|FQNnklp{U5C2q7w4=w?I{05S z(s)=+$XVh#7ttILsyY7U#qM-d)=MwJbK?fBU;WtLW};(c^|^FNn48(Q=IqdDEoFwo zi2_AJkav=nqz+^UFnDcmKi)gSZvT8P8zq^4H)FUaSaUL*&?FaLB4RYqg^)yTde&px zC@}+Slv9OwC%uwKpmb#w#%H$tQrXiA*@+m%K0YH8PZ8Tzte!ihK2&5xgSE#fMY-!R zn<2!3cU03)QX#IjX>vmOolIcx0UN$#ud)20x$3%zHnz7}Qy!~uZm!41;YW9aE}}~J`Mh=T=XhY zo^{?lWKzJpY$1`{6f%N;&w&pU>#ojGAjfVj*X!DJWq88A!$t+m(AhVYceVRosPnrL z&)!L%m4`A`!@bCb4P$8+sA&BN;}W9$kKXqLclTmu(UH4#sNvwTU*dL6lQWb zysetdg0$SZpGqlcA7Gc>*Z{{ZoNFo#$wQK4hNOoSrwBOJI5m&9z9FA5;14<9PWk|p z2(H>}_pbCr)=2(hDPwBg79`*7Qp8ncI4diZ{|76<6M@>2f(>2jVV1|4s6pAPXYf z&bu&ien)nO_gS;CDqXOkhh*ZEO(tP|Go4_|yz|6hXnqLrgR9;Poy-Vx)`>jOQf5Ko zX(S}7{SSazaB1pQ0anX$9J2Z*RVQNUQyYNw2g(~A0>!L3w`wt3LMNk@|mT@zo z$FdDysL1JQoo;#P>1sHS$PYhHYyz7_tk3W95B%_1&+X|DXA&R%o~%@o8(xnnB#ocE zYm0K`<6d-rK+*ozd$tb5<1C*{BiaxT3||wct{hxFelxci7^L@yZ@VP(d8ON_nVhF( zTp=0Xk@f)$%&uV0wjX;=zpb**j~5$>JLoSj*1LIPxkWJKqk$>^ z;PP1+#^3udl~txhYHN}%-2(B?6Ffd=XF#I2yq7(>MOW7U09a4%|Lp5zs0Q(Gi zj4g*6bCW1_5D!DoVYDY z&yhJW8W-|sv8{MM5$W;=Y-ci_*bfIIT!IR#J?0oJ=Emx;s5~T5!-zOigGZkNUMXnU za!r3_(Jrp2wJLzl*?PI^C_d-2N%j^z9z!Op7;*-CJ3m_jN%_641ZV>n zIK+^($fTxnsPW+Jx>P+vnb)s^UZBSDMxTlcbL7NYl&Sipn~&FFMVyk1MJz}v{4As) zc3368m-X4g{#jWh|E^5;BBV!#HIgYmvqFis`E4?}lCv{n7EEkDiU|7474D z{QNc=DnH#hyho?kc`LD@f0H_3Kvs-eJp-~8Ij@wn>qX7W#i>9G_MyBBjTbn@_j1hS zxCD>S1!A39C)h(WLQ8+#H<_eu@#ZHc{|2jym9opdc!U8(Bpl_=@Sz9$zW<& zOfTAG{<{O2dGw@!eQ@zy4ZksuE11&(SCF-$Mc4PW7Xp-+5oNS{$W`~v)tEDC>jQbk zt6_gErY1bG@`kNC_3dYh1{p2<+f-A?23@KjPnH_`QL_iR{bq(gNJ4Lyib{;|d=xi5 z&A(D9D*?ac&q+ryji}l~=D1t=0#TUyyHm=N;Pxk~LThjp9u1HsOyhMZ$3WghY*{@Y zmf0I=0w&Y&D0lKFUta2JT)dz50ByO8g-4x z_;C#<|BwR7k?wj8K?a5xO}SA~?wo@`7V$_5ueb&Y?jE|2k9;4X2;?Mvh7|sEv^Ht# z!sN0tLrJc>mq7EL-CNEu)-~nA|78`PoFPRaCE=)dwwNkcYIk+71gS_7FT?Tg?oY&O zi5=fB*f${(T@2Et6h%r>;tnttDdjD+C08PP$5cj_u5A`w&fu#fX#K568`S?eBU0Oj zSTT+yH!rs!H(q#oP6(}Z_<#MWWSGC28vB|fYnxA^a!q=+AuTxJ4#f-71 zep0~cr(8`USrQm?)`sID~2sWW-2TUJd_w2JgK)Z+6X=Y6JI)`Gg6#}(0>H3j|R zGr8wHp~BlpHHTkA-v+v7XVS0ly6Jp^Yr}qq)9~*t?)%6;KeJYE6K&EZ@AV4NO*%VL zVlW#%(2s+rGcyAO&d3><0me}-{R?-ZI+Jv9M6K9?ico$X>>$cC;zXC>sqG9psbtr$ zhthhK9`eCa`J5NsgRu@-UeVj*B_@WgK?kuPD+tIhD(1$gI632pskq_(C~S;LPjPR~ zAVb7h4|Fi+CzV3`v7OVeIOB!qlcXNPkDG_M)ne3;Zz9PFxRwe_cVRDKzP*;oL$0HT zz3is6Mq^}GB(Q}!g-(^whV#Q?3 zM|M4WcXQ740t3Y!n2>FyyE*=bg|R%wu%)qFv;rsnWW9hl+1}nZmAEe2m>6@4NMvRt zuO^hbDmAIwGA?sJ+<}#p5j!Dr+(i-V^Ekox#H!r-{&yFEv1@pTwBaY8GBLR~#{JLV zE79a<21J49Rgl*0)vHmjsVn;5`}OT9mFK9dELl+-$%*(xXe}iyfD8I>6vnG+yMoK1 z!4I>gR&TWuf^(Qt-GZ~twX8!(3zG_o)R?};Ih2{cep$x3qgak{^l_*ag0RB`S@|KR zggiLP5M_4y21slE9ANYI-HNAVK=D=iuf9ug=;sxX`;}i(1@6&tPNEHfl5?pWa+F)I z#h6ZJsN#&o+VO)_3AX+Y5Li{%k+2#690@}q!XGkexrIH*?cDncm~Ju#{?nn0!%ktb)* zwfA5@Xi$l3hXi+T^(=zAI-E{L(}T4~C&#WavSi{|^`~V*SPD415J*(Zl?fTt5W{Cs zo$!9vRI6~0Z?Bb*+jQTXVCeB-NtWVBqNr5 z34|5aH6<;H5HihgRZL5q?p_rBZZ@gd^)NG9P|4y1J-Hh?Hkj9L?S~vB_4YD?Y8SU1 z_z?T3;$v=zXP+#sI}R3Z>eeUrMAHU>?PJY8hY984u-340Vv{kxrX@Maq5?O6np2Kj zlX;bZ844dBZinrlD=q(h0!_V@lhHG&^u%Ag&!3hMJk${ZhP?XTL$u#YV98V=!GsKM;0H1bFQ8IH0tgZG(#a zli?IdM*AAr?XoN6v>B#PD^63DN#-Xu<9y21Db@_-guK#qCTcIC(V1oyMDe9wYE|#K z_&;s+WAEOF($jJ|nSFX#zXz!Fdi1HD+8w{{Y$W1dmup`0fgZ01f1o^(fSRt!)hM}} zyOd^}uJ13EEUlmMtE1wN%7$ULERK>RPW#_CR~!52%gsz}+O4EG>{P90ip%Z!q&Vvb>tW8@-RJNZ>oRQ& zp206Yo|iFnIEyimh)Abhcj(&Hww(*_XSLc?SO|4`fs$MbsEX9mvyv_Hc9C@-Mz{>M zsJ>T^U{!bCrMM)CVz~T$Fpig&%=}>9Uk?m^@nA#9dd$S(0&;;7obN(-_2+xrdH86J zR=~0U?Y$Tv9soCesm)*vU=ENK(F+H^1X5f?QT7U^cJ>j3y8PxTg_9Dl(Ms$4ZK95i zwxS9kwiGR-YIe^PFz2Qa`Xh7JXrK4x_c=QkfH-?v^TX@$*gma~v{Q!Ij*TV)d{UGP zL}L;&fDz(iOa<1fDNOD4Aw$92WOqjN^>Z3%)lPh z6=zfSr+uO?&33z#`IHq^fztz4vMEU^@6fk=SbH=I3lEb&-4a>l_8yPjflRJhn5WNV zEzRDX2|jbMnTu^7&*O606GXo0CJ!sF`}(#&=jVvSSe2HjE_uAokO!nU>+9B|9+w$s zduld@=skc)td<~47fEq)rjEg$^IWxaG}NG~|(?`GEdf~K%}R$Oz{@kxihK$@#W$Lb=FK}_@IE$$lNm=w^BSGapu zez6lg6(!7lr1nwHhT#jKd(80t26+sF9~M)|u65*%g@j38QDyIkbQyVJwT8p*2-&Yhr6Oc=8Jc=1;VZxK(ji%MEW3B`1K6p&dP$&fxIXkAj zr~Vu+#pci80;CY3#2IL!1uxlLLdwH}x5ymq3#rK*z%=a-&L-UJF@c3+GrUWL?T-#0k$M>EUN7K7jp*$kxb}q_XT)m;NSV#%` z+?CV-)QFtf?EYtQw0tGBJlp@xKFD$7@IY?ADw2$Pab=B)# zz10}ulFQR=lA@QsxC5%wEWS2hwcLmgK58o!11Qw^TRBDa1dPB~jkyInx$zdWJ=d7d z4_#vA1N?vPEE+V>x&O643vB*R`;({=oZRU>TYuN7(OaO`2f#wm%O}Z&d2zsQUu~aI zO&4$4NZ8rH?fMI>Cq(Ng;E4)9j!n!FO9E$g>fRNB-A$?e@51>*b15}^U}y=r=z6}{ zQ2L}!vP{vlkVr?~|0Y;_Q}p>K3?J<(DO+nOb&#;+OEUH^z;O33@WuK=4z(&6;7tcBQ%Wx+{(BZ5#k6>oY2>SrcB{m zZQPB|R5RJO5P`QVb%IK>w=$rjhf8 z4`UD8WsH%7bj3zh?_T;$FEu;83tm+TWN$T@(2P zoUAs(K(tW?0n5L}-iuolND%xeo7aCI^mK+32@5)!D!bBL*mbJad>%jij^*|hvifM) zHoRY6j6&3A1DwCTL6ZDwuGWEalbAmlAuluZp*M@(RfFF$e+*Ym04b2_U`@8RpsOb5 zIooitRkoFd`@%9>kR1H+If^(%!^2aAjXitnUW~D{HfaDN+3XYEVZr82CedT*DQzoz zDY$iXNrrR!17}wcFrVXGOsq~9`c(5kIDgFfWz7QZjTJc_3LtA~Sk%WX?q}OX)N*tw zx5(A8FmU!`Kvswsh0(D}j}H$o4y$>lN`9Y?r4tm4O=g~)d0#R|DIu*NzGB6GQ4u!2 z5nv_3p^VVdWGCe9Qt|g*nQN9Q?AR}WFXBryxw!*`>>_KL{c+IjR^O^^q#4g( zs$FkzcrX4?vyxNnyFuO#n!;Kn7AFaZncef?8Q>Bo$gP6Q-q5!u-Aa8wq~}JNcZTpiD?Ed9UwGvl>D|1qMt_neuy6XRlB1Ae$hsN!7mGj#(6J9wF)%ioi*H4Wn97$9{UXn1?E&;8YoK)ZEjcPZgBQe=~KlU6IU06Vz^(pFxOP#3F1 zZl9iBg)+(3lYmwF`QYcebbtNglLcwak7}~3rn%R4+q@I6Pd;-QiwZ;k<661sDS3+8 zEFBM;O&55QW_?cjwh|gVY4ost)Z6$gTS)tJ)p`;$QbkN+agHuvmaJUs+iumf`6ew1 zmPlPC(}UcF);LY%#;^7I*{|mEeT5>D7p27CKhJswUV)5sqkyWD02{cAK23gb2Q8!q zUT6#Z9GdJRd2#ct9vGg9!4LP;AbX-@u3Qf4_R1y6b^J1JRfBl)p6p*E{Q(mYukbk) zHN(R-`TW^2PBOf&tEK*j(voK>(mufUXU6*lX8 zBKaHxnyv5a{hAx?+JCdPnJh!!Nr4zgMR@yWird#)H^r7)Sj-g z8mCmg6XRi0&^R3V07bNF`{i~H#dVX7tg&Yj&uK;ABHtqaTd(Tzf^GD0d=rx9V=w*^ zsYhVDh5v!(KPo-VXQRC@DsiWrS$|gX!zT~^h|27RV}N_54)C! zXMd46g~wN=Tm8x`5-GLr?a#B;uT~nx9Ca3Fvw^Hqwl6wddt7(&eVP6qfav zBYS{a3Y*oNlN^>YJsx^(5y~#VzFfII*MykZ`)v!#$#BK3Rq@7MRhnnJl*jygE=_|Z zIql_t)ysUJ5 zbbwDXo&V!Uit~|B6O#JLLmKzU$FOfAIMYp`NQO*-SX!1Y_7U?TzP&xx@{O>YmpssD zQC<;K^DXjE$trDL3}hC-uU90~JCI5_bM@1Tq!%{pnr);Tx8*KPem|q-`p6#S+b*DaEtMAqN7Z@*(o^y^J?7X(Hd%r0pi|beLtT`pUhA^%u zI5t_ETr)a0qA|(`nD$;|NK@cS!~scOmewrsTp^L+eIdraOh$mPtP77Rr|WpIO>9a5Bl;=hcc&vPtYfi+%sKqjjH%fzeCaEB)-Wj( z;WHrm^#i;$5WXy3w;jIct=1?u+fB{dSr-zBI=(r9|JDxq5AeSW%@LY)rAyuqMY!`c zuV>n~F|&C=Gz2ri?f#H8R*abT;Eqf>sXipoy$hqCg{R>ef5 zFV8sL=zqet?{6y~E?D8Kc$m3=;Kx-ea|#VEZGGab&|DhaX$&cI@!!qASkKiPW3I1E z?+!E%=@)#f4_s^Y74eH*~$lWS`u+Uqqy(*{d zs!?~1@83n0dCgpv_^H`&j?Lhk4Xozsa}rtONEkapD9=8EO|w`IXwYgcq zF)G@Bm812lW4Ghb@~j$4VFYJYAHnF}5Eif0x7+?aHL4TcZ+V8by3)HKi>APEQ@ zr9(^Cq2H-8V~k-Qt;`b&S18X(zx5WutaG+Ai^dBcqscM<-7_^y8kEK6(wFVxk~n8d zTHB%IXI5r-{gD)OJ@_L9Y)oAF zUFqfXK1~zTiUrCHX?f=N>R3gTYDJ9}x40xUoPCEBO&RaYN$I zMOD!G!bd0sk*xhxptjuXyb{ z#$4>g>h8wQ==3T!}Biby|CTk+Em8z?Q!IXEOH0b3uC*d}o;NR)6Ya zQjhsWfmWph!S|3F8f588#K+siQsG~ZzWaSB|MWIrwPYlTuh^49KG?2|R3Py&llTlX zWm*HRdV*K)+>*WM*xO(OkDPVk9#fmANmU<2(~0^pt0?XJvExYGe0H62_9f>+_=N~p znQsqnO`@kp{Lfqa(ui0Fv8Kladr*hFF4EtHDt3A9b8Ky2-_#%pmOv?`MVkpk3)&_} zA!$)PBm0Eu23s&>eozft#;PK15Tr^jS+>ReXBeRz8aWEBv$crUCEtDucF)H-gtS)5 z-MtgOSj$Bw7A#ykH6(*mHj>Q;CpVvDyV%|?*wd%xAy#b9B42qwu+o4Q5f^JM5ql%~ z&{ST_0h3R#sYQ;IcbTUC$lo^`3e!y)kLkJ6OMp?)P+QSo+uaIDZ^?^uf(KY0JytUG z@$s3VMHrX&^uIutWB=;QM%_bmPG9Am8>!G>8xBP5F3j*UNb0uO`Cp`e-D!=oFjxYM zI#tzjeR!M6w(uy{OZTQdG+ItfJ%H5#D}mSagLCR+mP&%6+K+SMn8w%!ABrZOG(Eu# z*^J2RtyF9V7qJZ|Xa#$zSDsSHKq&cpHnNgugdabWnlIoEj@#3qw8rew9(J`EomZ7A z6Yu?)dU=wnLAD`@uM{RDom2M;R7`!$2m8z=;*{~gE_3z4$m!@g^CGr<-Y>n(VCEB? zxyHc;=VHqQb%Ej5h^gpr012I$@GKn(EI3y%F!)Bfhmd0!Y0 zZxY$|{t(+}Q{!MjpSt!&sbg&3>d>d*sK@*k--v-Bv;(z_ll?gOMv_r(W`sTIpX&YL z=#m+sr;%GNSNI^gMc$_DJLz8pVqZzNEGQ^vLSN~gJo6C@S;a)GH%|op2R}-pe&+@5 z*XnLfwY*pfBNs>`IJ8%!QgU+Y10HAMjM-35WO0sldpUM3F*M>{Hc8CO!=t_x-w(l_ z)&JKbt?{-x>Xit#rhal3X^V^)ZMB@$=Zt{9maiyr|HMS*ZUSOuPinJInFA#7ZzWe@ zn;@`f*s8>5zn%BDFMzjGJSS}xIQKBbts?sfPK>y^2*vNeSfSeYJ<3^cS`5iM_wxT@ zwc^t5Y7WvRn8kd+NExgkPw4u9ms9@?D-=*c`$hfNd|CTOZ17++u5&DJDCmDb$87%)QBt||BkgSDtboEHyVcI>A3@f6|J1R&>zGpT7LxfO2*py$6@juiVR8q0r<4FQ&v8P zvKv->Nhq*oJn~_IbtQb2{<_C(dH;eHr=kyx_A0IN)jO%}@SgS@#gs73IJEIvZl5LH zJm`r}plS7|+kq97cqO$sE?9K_&3*N<0<0!QB57B zG27YYYy`gaQ|26Z2LF#^x)f|n=S0P~WLcElM;FI9`Xd(na@?2UkM$bPHdgMO*avGt zp?sg^N(`T5zO0J$H)|d2gBgj|g*_!I{%n}&OKV&PEcdQMV& z{wMiYrTwKpytR!++0NQ*w7*J}bnR5d${*qtAh=9@;rZ4!V45w1OI1G7pWYO|YLENE zi!t36vrENP`&?h%6n(ZY?+?zEm~$fE6ueb0w!{75`O*EM2>vlj%zVhbMNt}rKi?F8 zXu>~?QDz*-y+%j1Q96_(_{Asooxk_SXyswfi}|B9L;FtO{(1^OwC%_Iv=nl%=SGHT z&1aof+x~hAY1OTO{{UDisIha$xusa1ExY>%g>@^8;H(S>)QnV;Yu1VX04)V^{{Rp* z`0o52eFP^gz;yJ@X$UsRbz<~9YE;>x#@(J`e<57c9r6?Jn(D4oJ|DC^@gsjK<<)XY zHRwrcV50S7mN%7_DA#i_z!lI%ZuZvJ@W=|f=OaBUBG4`*%LRA)rngyGEwFQz*~$EB zp&MMw5jJXC50FuC+?rcq^3*pclU2~^Q86z>4jnCMtxL&D&+isF3$_yVDdm8t!*7aR@yFqfPX6D ze0`{ibO}%%tUsM|(}U(4l6E|QQ~`A5Us(_7RN5_GiT6Hh-7j|z#ZvTOLN_hE3ebED?hWv1=6>fH~h5(k1YCPx*-Q9a@MY? z>e-8?O3}@=5gfSy1{`L)sk|?Gmc#9uip0#n>f@Q)h8*Rb7xj^0OGhVFL zA85CE=aJ-993ZZ`OQUCxm||TO7az1x(flo?C6|=o#HE z@gAK8wvBOevN&RS2N>t@&2kzt`Pz}a?)};yHrK9FoYP6EyyZOMs^m6 zK>12@`Bfh|pkab=V=6^b)EmmP^L)v2+z`i#PMxma#|=N;O+{~W%{+Dg03n3;w2zO@ zxgnN|VCq~~QR6n2;08?t{2@n)?iev3=sM(mB{>DPGUXjq5! ztsepSn$JR(;VzPNGZiaw2G(T9s69nt+`0b8Cp=3zKVB;)9;|bU)3GWX(K|00Us~&0 z?d9&BFPeu9|U6R(bf$R4B)mOGIGZYx;($;52YUyMyHnXZ$JlRR~&sPO*DM$>fXr9)ramOS--p+Sid zulKk-S8aYg$?Yxi$$Zz8c$VSEiEkZA3MSYOt_CX#lH1h!Jer%RpP|X#x0$L+ef@t5 ziT?oAk?BgL$#pZHrHw3uHg zP#GRZ4CiV9qwuFrPCAOUH6%lP&2e`e3V2)r@9R|VLB z_O6cHvs{H`Ay`*^^gbIKi<7O%1D@6OyIUtVv8$IU`{Y$0sqSl=@qVFgd#78QMcSLe zDtd#DdM~0fUfr`2T*}=3G|gi|bkG{%clpVNV0vKsR&`?e4^Ii~sZ@>kIZJs=Qx^3Y zuSxLFhG36R{@(E`h%69qju+s`_4fAfS^fs_ox8&RBG&#*Hs388zaC*9#s285j~Dp5 z$Eo+PD|6^Mxt)uCEb-;{h#1;h7PHcmg@!+rxb3^!+PM&L02l+) z(vY|1Nt!KVREOq_(`d~xkIfmT=8Vt-vqovTqcjM1Xx&ZC8K7hHM(Sp0%>X}j8>yM2 zGyx{+P1MZLO&}k-npQXJtDjs`b4F;e%Me~&T9Ni^yO=T4%axb;QB}#Jmn!TRXVrct zV7hF1$NJ&?jd&NtJ#<{&3rL$WTjU>i@BJ&-J}GLAr1(-f(BazUQGTT#{43@iR_&&d zi8&^`-34*D>ga#SA}= zn0!*--!oZd)2$nnX;&{N`>EE47_NcJ`j2lD+$hfSKN^9& zS8`4SpRIScehl!wl!31F`-`UQ=HJXBL+`Ya{Eci|+IT+a5H5$LKWsoz&*9LTow^M{ zKBtqP6J39N{&b)4l3i!-pU%B-*>PrT+kgj_3R5^R3zb5?|ZCCDpA_8-XLM1^i83w)i_Z zBkelGWFPKDpVXQiN@#^feU3=@isb(Q8T{!}TD_UVA!cUa1<5t(mLCcR>A!%nYm4}PQ5uN73n2-rez1q zCXEYi8r|(nx6MjKK^55338iDrFq(s*26?0=lQa`ZuqTRqXA;DH z9Dh2GH28jL;~(t`0QDd3Mrf24=%80Q;;W-Bn(9tLr-_IldJ4?ei%VlBI^fZ3Bc4AQ z%-s(&O@YAUdkTwHLlx(z0RELc1Wn=jR1BDA41R{D@dP-M+F{a2SMwE1Yn@4`$w#>B z$B7cf;s75jC;jZzrtuZ~sM{hZx+gWyqizYF+*QYn$F%PIh+7=EQa~r6HSAg?)R($FwZvn2McCvF zkVSP-YgOFv^N2e}X(k5gGvzU#QhrYp^pKAJ05Z07(EYLX-KMgI7z+p;Qn+Y(!kSw%jEOX4u5 zQe}L0W1r5q{4Bm+uVo=0DUg3U$hZY$x=eGtkJhxbdDsX_{{YK228v%1Sc}S&k8|in zRe9l!jsOLzZ=_GMI;s1=Kjl;`?@~(`?$xqnjAQbtE({`-~p zFtzUy*xh($Xi`(=*pX$5LoUu4Mon^>wd&n#me$}! z<;9kA4>$un)|DG+v&zfi?JK2Y&HP(0nXT#sbVdk200m=dahI9D*o~Of+O^N0B(f9C zva_GW)Y>92w7-`eYzPn3*OK*-_v_t=%hR_;vmMJhKc!|N`AGGv_vbogsq7uc>siQY z-`*T!?wxI=xU-v1c>Y&tfj^FGxX?9~uySKk@Xd|!XB$tbu5u^?OQJ~X0uYEn=s~YY zi$K$Du0dN!!!JS!Bl*{%O)F}Tna#0Kr)4F%d6BQI+Atmzka9@DB8w|TvA0+wW@TQP zslL{T;bcZt=m;E9BQieJ84b>BM6IGeN^+BplhA(BMtTogtZywNVj5wQ?MO>2NUll9 ztd`T|naD*14_dVg?!`(Fe-P%o->V77`lwX4euBJXQPf&(L_gjJStdV;$gfY-rF$bFjIl}{x`d$$$i%*@f<%ls(20re+}^YJoQvOeb{rOQpoj?Hrp{_V-BBwv)&w#l;H zN2tv!ra`V-9>ae5mu#Ew<4^XhQP9V3ryn$gvHDd#EkV^7k@rDArF8>DYn>`Puuj#1 z*pEyKQCpQ#O6cV+SN$63pZVwgY6!+<0=2Y@U$n!hMhDIIc3?i7)?8!e^f;uAFLM2& z32zlk0Hd6fiqX?rORMVKx0;@U$l_P`dD@#w+q1CMm+98T` zICzQVA8%3bUGutH9|0;3QueUd$l|VGb=Gep!3-rNeKVTtrm$UV7&Nt93yd_U9V1@3 zrIu|r=2F*?D4sBLBK703s$L=1Tf<%(l2Al?uBZyczr_qgrblDWex|wcbuA8Fr>}yd z@ikRPv`%y42Z>|3m2?d(h0|>CjkBY$?erhcxH8l+$v@Bhed&bPowkS6#tJ$ib4F<> zps~czI#tLmlWt^C%zb-Qrrr}Sze#iR%L5B>CL`xFQJ!~O+M!m&bwS51t6 z-ha-TY*)0bKMHmW6zI)dHUqTB{`sbTf%{CqivIxh=~ygQqcvdI&DqfYl+$N8AHaVa z0>n9@o^9tJ^3V0AnB*N}Z^DoWO(tj|hgje8rj`Kn$J9^*vq3c7>U)X_9<%_tqMMp1 z1K&P5zRv_&ANq)={)>{nX}E3fPHX8Ok4GK_v;P3Fl7HyAYv%H~IjM)jX*K)Hl?bTW zYT=5ZbTpbRO64{cA9|KSgG^DD=cjsxYhvY%zn^fbe(2?^pxur;4l0hF50!7dhs^x` z6>D{E6=RO9xtS*FH&l6}HB8FHo2V^O=7MM#{nQqy^G0Y358YDRK|SrvcGl4}vq;-j z6m>OOcRQ*gmUcK*9Z5cw)vmJ-fGrqXEOYpZWW!3_t)M->-u_gZNNO#;4w~0Y&^{m9 zCI0}3rjFxQVY)k2`TW0NI%nMRT)w&SV_emzXf%~;INyQfjnhBXyL|<6W;wM>!+EP* zPc)M@;S&7P%k=aW(%M;gqSoL3B|0tcofbOVS>{4N+H>=t)9F&3{S9MIG1gC0AN~>V z6YA0#ZXj_Z@*4K#QJenzY?^V>yg}iKl3V>h5R> z^TDXY^Pl_ZWPNj8u8!6>BHvKcjGx(aNVfWhn>%k1mf#Vdah~;Dscgk5$!?8r4{MNk zgGslr)GioF4a82Rn1+!Fz~po7U5j{uG`||d;r3$I^Ta&8&61Dlocem#m-v^%Q}}-A z{=i6%HxCOAcs&Q_ipNh7Uuqio)h^#R5xO#;O{7+Hq~wggu#xm^fn-LF)H)G@2*@J1 zYpr@u3hS$5aLoE#*z(I9n6v18E4Pk8EJ+~f$_V$(dEdm_Q1+`Z7-a8Uw4$7xWNTfu z6)2tL7tpo_@z7w7^>8*uatN;{@Xor|S_A>ynII-6&$mz06{>tkEw9{^jC<6&4)#Kh zDO&9GOwqd->0W&wkJ>T>(FPv1czE*0NV1`W`eu@&JM3xonsQs7PvV<%;opgOB0-2Q z7`l_)AFt(HhPMH;)TTqz_)nEyC~8 z^<=mJVN&npY5|{{W%He!tGMHJ=RY8YBMzr`X9F z_Izbe{0L_?v!zC3vV{^v4Q5K&V^{853TrXVXwGU#V;W{@N#>YM zLC`pwLTQB1O#>8~v0^vhK)$XjsWop)Bl}JQdIrq`XHw`x?6`>@NF$XnL z>C9}mb{T*h;Nbl_5m?PBveueZ9H9=r z_^!6zIKI(wX_4gJCz}?18~*^WS4&&O*A(R*T8p8>42I*ibiNzY4!eCFvj$+JC+S$N zYHhEY3Bem5YXj+Be}*SI4ydkBGG}mnXL26@0EIeg&cnqv={`;R8(OEsxJ34mY0TnT zIdI1+7t|WS^X1a-CbqP=o+XSB(nBeXzCCbpUa4ht^Xh3P1BZx&Dt=cO>}$)l;cIIZ z#IeP;S)79#5?(e3o_#B)l_#O)VCvPYc+}k(msgKGg{78xbKY4CAJ()4QNyKK+LbFM z%%5nN{l)vE^QszrM^U(j0JtpJCp`yq_zKX~?x7ONaT&re*&noPz{l3P>K?5<4tG|N zHTjKtDaq%(W5(O>0r%pjnp3h#Smm7YRc+7jhtrzmlhp2nY@0Qe`KiRAS_VC(2B{-c zrff$D^?cQH8FFiJ`jUBimDFTl7RdZ))1>m6u6Tkyw^p}9?#~LoeNRJFQFcxM=);O} z+&6+fW5Ift+HJnUBOJzZxb7>Yo>TT)nDh4`VfojQ=(@z3W$beX+VP)G)#~~r*4oyW zEy_v^j0llI$xnQLTJ)(^i;bD`nCujxTiV5F+2f4uCaKBjYtwua9EQc@I7HlV%y13` zd8Vk1r)u{x1I$i9W5*qhc9xp{k9Fb5tt7oRp(EQRw6PK&)P`P9(y@fnn`fzv!@BfT zIy)V1tKtZf%2>ilBBJvm%#nfg1mylz=c4GPrl))l-cKShNk21n9Q#!{wRoR?pQV|_ zyRhBn5XFG^>OPsKc#2V{S=;G}jnR$r;B}R|egd6Jx>&3(1{5Xk?sB(s2A2b%;8ymU zyL3VU=0`7&Va0ON$@|W^$Lm{S`5>}_58~Pf(1tS=Oh+nmXyzA^im#(soW zYW1y)c>K$&%MNoid5`Kj6_0|ary%sHeanOBc9L9~d^cq@V5>ugEDszWE7ZI{smH2l zuw4lQOSB0A?8B(99nvqNZ5r~@a&Qo`5Ho{Y9suzG)Aa*qGk)fN274aW-%hGdH`w{i z*8=NNm6ox0J%BpMoj|CJ(j>|O1azpCww#l}snOGCAxZVF=DQygsU)FFd5l8(3aR$! zB1{PsNO~ze)X4{)*>~c(eR4ZUd`)c}QXkyLWVm8@3t37^CvFQ5?WEDq{6f?g-%PaA zQ2g@9Ze{hyKn;22&`WzOXcdo{GZ|zaorP&?*XLZGIXLi~81_G0%O;a_K0R<8;5wZ5FvCW?J%bB)7Yb)q&%kS7-8=JeA^o^^zj_zAO z4O{FBhb;?k8Q_dp38S58%F;-)`klz`G5P~ohwmmC3Mt-$FlPcke+7CjN}5Lp{w#~*B4hAOIX#(!6SqESB>eqweFE{X0*5w$^GCB zc3KyUyiKm_5LjE>qf0AbxPT3#xybYt-3(ev(nro>GKs0V(v_D}zSHbrv3c6>%lj>u z^R8qh;E~Y@@9SP+;!S#Q5oz z#{-JivC|=ROPl+owzgyP4ogZq;B^b?O;A|RjwvQ$u}1PR>Jf8_mRn1kbXS_*NY%Iu z%0_FhwkvCGFfc}z;5sr|OM}9Ll1R%QexB8dsCiJ^6@gkfocU0hO9<8KBgZ6u6`vvDUYThoX(6g7Wum?6LW) zX&42<`m-JZ^c`uW4V?2vr>d#^DTWA-Sk|_6O)?8e%$D;z?3c=wWFU~(&jfV``c_p8 zLbAN3MX{TD5K?d4{^jriufyc1?W>LQ=|E=uqehnkve z-6p)PBC6Zubf{;x+7A`E6ayJJ1KyyzjSf^Yk@c!*#ZOa_xEo}}I?fX;%DuXpv43qN zP4cE!*wmH^_WPOPB%bw>-0tT0O-=v+bf)H_7ZH=lV1EjCn4OR3THNDkerU}%a*_9i z8;7j`J2Vqb%^4oF!oud0a}pInEKUF&Nu^aT!BZ^6-~c%!_7$$-d^-~=fcm>G8TyM* zdk^;g`+aFPQi|P?{h>TJ5nxsImKYt@*kkhF{Hn)?{2O(vk*I1e@LI9l3v(p7^dHu> zyffhO8BI@Fk1`*c8)-lEC;tEn?qZD0;E|uu4`3-M-&S#AajKuYPjWpcPqot{^0ez@ zgFkt|%eU3L*G;EdO5D8ikD-n_aUH0T$X(81=5Iqh7blka64IB#%rIQFk)w|vMef$@Tu)8&;%DJ{`B z>$he}t*$f|_>>xEir!dt^0942L8@zAn_|#7>m3boz#hXsJJeG}ejD2?Y z{40KB*~Fsaxn8C1FT^+Yn~^S7G(R!kQ~=(i(ETb6O6N|yCR>Z6Ez=%}lauPqaar=| zI%!hT>(NSoyX3e(txIR7Tcl#`!#$!NWRvArK8xJeimO^~N3e=~#n=`rdtLDzrgbWr z10g?&{{Wz_YgW7Sq!C6i4*^GI?fKRHUr=$GB-$B=%NQI5?erei&dV!ArglHPf4V=d zbJdyELY>{p*KLLogTi(C({1@IY96GM-3v{MCk^cZK7lMAiwW3_aN=iQx(8gP! z(@giNCBc*L9poJPnv9sY5eq)v!MgsHtv`mZNyHPyDD-C7e_kqE9}lgx+x01<2s!ei zZSH-z$RC|9Sv?IswL4zq{vWnWD4SHDI75h({{Rzu{Ri`|cTv4JcXM1j6Uo6R?;l#z z*7b(Fb+oroh`%bvm;pzi{si`!TkyK*kE_RB=vT}O&sic{uea+O> zbs05&-qzb`lJ4na^ojdjCs)m)Z?$|Nn>#mhl1o5wwLeYs-Da~mC&_TXFJi|5Lww2`~Lvh2;bpeEMl1t%^QRCq%1(^gIy5C)QYn_df65hwE5ZU zT93s0V4hx`YV7BY&OrSIU~9Uxwt7oPBWF<54*6JhMC`cjn#a*>mfka}-zM91<|RMI z)thyFCH3UD_a`8qlz$*7omukUPgZL~FQH0Sq_*6Sd49`lv;JF3IUa(t_np8K(xoi- zcOahU@)Yamg|10~SS~PkcRsbpH+_#@n&sR`=EWt%rGQ-V)YKWJ>r$Cc3v)u%{FtK0 zcievp>--^jjU@Jxx9+ZxisFlH!2amPUehjKF@!jg4&?r|q$Or^;wkgS%W_+oA5GLJ zvz&aYEXJ4czIh*EWdPI&&PPh%#gU0A>9T*0B}X45{nf zS7D|2vfe}fpL20xcZ?gj7jjAd`V4dKD)C!uvSnRwM4NY4o(RMKAgw(~=ah7kH~?9j z01{1l{{V(dVYOsF9+!ftLjoa%(CfLm}#CkmN?{o%fJ~O z>lenJBL3dim_kUhw#q`e;GX7#Daq8ABHO9YOEi4TvAe_>~O(%VCNpR z4BgZ1tu*r;#q=^yazJ!|IE+L)7h(@R2mn?2wHv84>xtmhCMgLF(A z-i_dxV@U42LwOyAvFFQn+an6}{{V{mq0&4pZFG?%&GyAui}Xg={{T1S59eM3b9Z^G zTp6xsXJT?$&QJOFrAb}~9x=&Q(Dit{LGbrR9#y@(R}4D8kYp$R1l6GrkNh1b16qV1 z?OFc-g?QrIS;hwYYpa9m^DImF1I<~B>>yxi^CABLrdQ~HLMi(s5uB3P_ULv000rI; zmR&Sww2>L5}4u~l2)hlA$i=@5_iOAq90$l5(p&O))= ztW3RCE^tTWYRo<$@h!dtch`S}zxTdLxGLh(cxERsL_){p%KPqCiTN+@ zliPAP{>xP+_!nd{W7Y0d_T2)Xz>dSc*-OMqQ1C-h2f47e8JA2j^gWx>W zFEZ%-&|AsQ&-UNc_s7<~3&ncWcXOL%!$k{bKn_*p?7e$+Jaxr#uQN?4cWumhCbo<33YP8NYhG_vu_0jBizRT}@Y`&TZ75rI_w)ei$Exb&_AhWfa$PpWR!;(y{c% zYV%EQ<*)91J*1PFCzxK`;PfM;U8sp!#d$C&RtS9H2(fo`$EHj%)i{ieV_rFl<- z?StyK7fv@rX>RipdS@6PsjkQ3mF~++4Lw^veC;a__Ec>ZmEpS~ZQ;$#aN3Q^s2}vn zIb-?Ot3)c2y^mS5lv?R_$j9bbmp-$HOT9yOs~5f@!j%oc9CKVQx2hT0iE8JC zJ3Ecfn?sGhY+V-5Lz-vXH+SZuzmqZ^dYbxA!+sg?{Jt2|T5S%(W;WGRMK9MGf?WfG~Jg}0kV{EQ{xUBn@!s5Eu!g*4B z^&2S6*)?5_K10n!Pc;WhRzed^%^9GY1{t85LrI!Lu(d}};ZxqO6UBQ+hWsa^>fQ#k zwbbm<%$bwSA1kRohNf>p#$2xF&q9{HcU<@@BzvXNFT|>PLvh1@-$Rej73P}9hxI)Q zqx$S33*ECc&Kw+{<)|h$X0B(Gu%nJnn@VxR~^ed%TLj%!;`MwT-K8ySQ0gbdD}m{adHyn&*7ddYy{URzvig1t00RLl)?A z0sb>iy@j;~d#EGcN8SKqBh*q~YA{W=aV1n?VvOJ@`uCVhi00|UjerY5p)UP!{^Huv?epqu7`jL`p^2H)l z=O=DcHVN7ZL#e?W@m+nj#2WsbwsW$r!gB8UPO}sFs^uI_Ro6CxE95; z46FyEjQ;>Cr6nDSP;&T|r&OAZRF-(jEZy>bX*&ATmx{%^oe5Z0 z12`S3#5PmRo7F&1?=irt>I=97sp7Vy5vm2;pn;CP>Y|eroryG7Z7)`aMv)j3gR!zO zYVNkM^sk({a zZR9Hrp#B`@x%*p_WiOc=A*II!`Y++$qPe`1;rHE6UAlpu2jSY9oL08f%53xY@c;U01C#qpUojxgkQS9)|n*5_~^8$h}X<+Tlhs|DY-kdrkynf%848O z8X|y$BafK+3a>#Wn3K+Vrz~v8jw&@~mK#z=0{qR!Dccq?mORvK#r~A`4at?b&vQc@L5t2>@DJjJmQ%auVm;p4Dq*7vl3MitG9_R6~ zcX&3>+fYOQ0FdObpLHwKYXFWTwpluR4_f*U<6A0tF4dgjMfwtd?VOeK73IO2S!aJt zDD)kwVJkq1)8?;vY*3Zto^qy&x z5hJ8vY7zARR z)0QR?vW@*pZQ)N7EM?)-u8upS1N@q)NS-9HRo%K`5LpUc((4~M3zU%w~w$K zmd8`yHChiA>k&9_womjDwm&10S-DbMHg-Zan!Ab%eQ(6#<}_RT{IdlCiEN)t8uoif zzSFdrV4gtw{oH@Nlt!Rir?~^L?dx3kgY}J1T+=4G)T9!nvwx$jsODb8kUI)Pp!l!P zmS}uVfXdsRKRJq%^uZNyj+aN7ip0UUPEAI|+leDDMKrGk>pN0bJkP{80xp zcxW&kb}Jv`Xg)5zwc_VMl|H-I`K@ItaxOGl_*vJ3#Wm&f(VIO6YnC28w7?IgT#sF{ zsr?QqM~w8l#FjrI@(_w$%!n`)b>ttz@vQkIj)>8l9pbG>$!lw(M)`aB7tY7_k zA#Wf1-^710`d2sMLVvVwb$g$fTA`4!92r~g{--~MLwR<@oz0fXB>UA7INjKm?@?ou zy0EUUnX}C#^OAF#&zel{pkyTJl3tZ#ZMR8;2ls% z2mJJf5A>_hd{)z#54C6;+uA@!L!N9ahwoPsXvjfjeacZQX@kyj}$Vp@arkU`{37@cwXYleQw;`#}p7DEM7z< z6!y+LX1mW7YZls*TET7hISdJzX0{(I6VL;W!mc#qx-hR+QGBT!<;?FZyplHNa#)|O zW8FSYUb%b&Sra}@X(K92TaY_*QRkXgZfS(pXH|tJib@Hfb`;W5Q9$fIH8Rf}W2>-H z`PaJXm&t2q1PXXW3;9=y=+Sw)yg%KH?ltHMe0alBak@F{J2a0)hf-GW`qMQ#4L@D6 zlFs@Rnim}Bx4vtNgHvV4&03cF!5_t0x@>W)W^gxu3+B`OQw&mWlS+|`iHC4Mg!|V` zs_WL;9g_Jmu2{k#2SRGqfnVw{&_&E5PN!`#;X}0Q;gbb2v%}GbuKZJ=xRlKb&(^6P>Q5(UW~^a z(SLUu*`l{E=7NVE0IOH(X?1jzPEuP}hDJsQAoZpoWN<;juA+2`;waRhEg;?Wtf_R_ z75-4fbgY~#^g0!E^pV$iSu8bMPb*Qff?&LFEAtP;R^9KzZ7%LT_M4)};G^yz&biMI zGikRzOOBu%de>hS?7D5zD#}a~4at+!)K;n~txr0oT8g8yJnvZWCatCxD<7X2=+W0S z&`TQ?XxKM#j90&=uN>*NMF-xoHBBPJQOg*RKV}1($7|R7(mZ}UIp3m>#bAEwO?0dZ1(|oVO5eCl>;(L% zqrN?BXHFK_;GN$T`CeMe2TwVjsNP>$j;(TbHnt!J1IPclEwy9l()eN!u$(NEoP zD183_O!TZv?GQtA<--LtJ<$8g2h`@Z z2ZhSTxvl|xbjK9SYsoC+Hg{QQvw%+}J;fypZZ}f@011%x6DM4L6$2w7!4=WmpbT+> zI}zMgWbpzzQ78@-8{DY_M4vHcR~Y+Hp-Hp*tH|<$|ZQv7h)C z_=;&+J7E178iI;f4pK%*JE^9CTBjGt3Q!<19M(9yBa@S07>)X|gcO5?D&+sQdT zl=XGaNXe+EqnKQ!k2nFs(ivEc4Wsd>Xs65;AGEh6m-D3w{@VU^8Kk9g*qdkjYx&ab z82JNPyS-f&)P=RR#`RvCheiAaXxOY@4}1(hQT?TL z&mBV%$KWf?+1=!MByG;@S(x`Gx^2rF9!hStS0}RAO!`rpZh57obc;`kH@E|*dXK$c zfcP_oUDkmhNHFL-^Ch^uxNJ!l~r?nlgQ9*hTTEIO?it-`W@>F^|r(!8By& zG*}HZzE{ug-+`XAE&aQHd~xYns^u6krnUhke(CA{6s$WVTkUq|{#s+|DMYsI(p3CT zDKJx^~+N-P|v&=I#rL-cAZ7|}(8l=t6{yT9QBvsLKYlwglv-O2rNUp3vwnNrIhfs<;W}Mq4Gy1!R3$0zUS7kiuZ0gzkIJT4JHTO91o>Z zcHJTNs|N}`oMx&I=Tjp{qD1N|*L*kO`vma)m)f+Xm`61cLxLi0IgW_YWKHI6I@H@v{?w!uH%F6jGxZ7?B`f5 zW0vg^AlwUu2M9CAT%Ne^T@jpRvN^F>cSTWtfd`Eg!xlVntFz(s(5El({$?#iLItcy!($YuJ_~v^jv;*Rv(CdDUH5uQb))9 z^o?8jis{<=Nv4|N!mG~aLJ0f5gYv~LR$spyP4$c#)l++>d9trf-AbBfGJH?!@5lTwTdtMjKliX+M?P< z68faRIz`)0i#9m|eLnIZc;^xU`Qo=N{wH{Ta5vxGL+vfTe^t$K@^~*ukR7JtGycnu z>MD%h0=NGFmN)aL{{W_ugZzy(D!!&t!sYVrk5Yrh9v{^RSoN(sMm-YZEx-B@uRHjS zuG=Nq(=2W6qW}**;7~l%)Bs0rm61P$4wVC6+-ndutVC{)VOMkg&@tcZSh`_L-Ck`% z>33UN=Hf*;`ML#P&jO;QStVwy&fFdXjAzeIS4H{Yl>LK7jf>84H2CK;KivoDxUJt8 zYKV~9_KPw=gwomAs;v$p3ZnmIV6rCe?q z=~*^)n$|uZ)bH&L)E4`q=oyRHu_2Pc^Qc4qkar;wfE8*0QB+nzeB*Sk)#q zn>#7S$Z0ehO!GlC2DS<*D4=#0lQ`)}O(8sS){`BLy#;m|W5tlAxCaad#y;TI z^V}V$n&$O$z1>go9pmd$ChTuQ+`w<`rE=D+r~nPL4q3e_f3|s&894)@k3m$KjHz9nG+KcT)~_a^pW?1P zF1hPgqq`#<)gvbhGpuP+b>voj_KTd3DwH?a>T2iP0Zz0wA#LyHu~|`AnB+pNwTm6Y zn-ztA>qtW%aaQNJQcpF`-rhZ((=u|7K_mU4OPfPV5ciGkHtpfJxB_B;frQUoS0>j4 z&nr16B%J1~E~+*+&Q*?j4_eIM!xPl=S;~4??1U`aI*UC;p3MuNm<%6!$JC_~Pn^|z zdy^b^Jt`?=GCu~1O6Z7E<(rYOZy}P-Hg2F|qLFh`MIV)Hhzop(iTJC9XDY1xYD73jrS zdpo$eds)-!gj6LQbw(dhXtNGvHxy!>^%PNHxZG0WmlR@)6p(;m*QER|@Y+vxF1dW6 zZLK+7-3x#9tBcTdS+&hx3m6oWBMrx}9qZLRUw>_(Xtw?u(=fjv^UC$dzq$TZttjiv zaN{aAy_|ZS-->nC*L8N76ZUxiNLSm|u?eK@MQGj3;ROiU1r$=MmS6=16I;iJ2C)40bf2O|Gk^M#B;bEhLN|_n0mZ@}zGF_W30zUe9CaNYC#VKDD5;p*RM;gHZTQ62?RiW)EMU2U0&3Cb{dM zhg#e!Su9fNF~POCN6-8P@M&_L6tQY28>b{!7fP~jWgj-w>fYav>DRSywcK;59&PMu z4kVJ@6{3&*lE207ef_I0`%k~oRpM<k$SR0JV#6O{h6Pi6*p+Z>i_vD(d2l z*KN&_`Fd2bzok^%b5kQ$sNr$GrSx-(u#Re!ZN*Nfie_?5xUbU`_LmEh)~g}Os+!KK zx>lO^9C?IdHT7=QS4&o6sY-8^bvZv9YX;`+Y+@(Nhda6I{m0?bptiDMr`zd^m3i@Z zcKGEGeZ~B_{A(t{F?*;q_bf~?W?5U;&=0BNwJ+~vwTeZNhuZGGdGXN!`=~uQ#d7>o z=9fe0VGn0bLL)Uh!+kt<&?peb1aYYJ{{ZXrT^6EOOwz5iSWoWbcZxUH&UV?64E0z zGz`svbayul-AH$LdC%|u@B6-Mz4yZ`Sm(p6H8ba&efEC#vv+&n;p^~tXPss$^o5qM zBIg?zdvO3uJnc?W6U(yY-{)CU1=tQ^+K0#iKDo&3;H*R@UvVTRt~y`q_-o z;kEL5khq%DWyt3J5@-l9i(OT%JyOWN)%qbo;{e$ur5{^W%td!#=wv|j$@7o?Nv*0@ zZ^7DSu9(@l8PIw)s!Is0(fvSMAnOUU_~!vT`JcV9WvJj^@+xQR7uE-V2(cg}kS&XrlbvXp(6M%fFP3E` z37E57qDaG76elCRlJ>RrEy^^Fca>ni+w=W1I|km#n~0%CI%|4eHCFF&+Zl>{-~+!t zFiQ>ym`obj(5KHg6A5QO=ubu-9wBmBEp{a-KXVF_$I(KsG^X8b>c4cr zg!({IFv+Zzr;7vUR#q2CC%m+L6e)9EVqXRQN8EnfnTdh(&{w4X`*K)ee6AYc0;$HRN()$qnl4kLa5cIa8I7 zaSr~1U32qr)9-Bj>+2C02YUI_6=hW$0%KRF9e#77-7|JcFE!H`1Gfg}@5Qv3A!2hM zsbIA({dpFl@|=*sLif>S2at7xT;tBehcKcx#Ct)xPfl^5LSYMO=5z6}1M^cOtb@o+ z2r$sGMgw?YwTjS{R9H+YZpCIFHbzvG&BfyN9sODL@d-G`&Hbxr#p-|%I(@&RQYQ0v z>J%lyPtF6R7`ORx8bmGtd!)&y$+sKvsY^rgoX<*cpB`i#lRn32`OJ>_+tcpTvweanl_3$jwuX(VmBFb;E7YF8veeKmzid}< z#mC=QexD_~?;uP-?4jz^AmfS>n)qspnK@y?>a%bkOUHxcJcBnkVjNqSGR zB7~>-M5WWQ{=D3t+hA<4u#hl)^sXgkNmqG-8oEN0llw6=FI@Bar2m#dctM+h>c^f! zxt4>kDOS=D{n7?b+78UH7;n7OlIJ^vh1S@n%zD78^x1(Q{Sm5!%!W>1M(O*>9@wY- zS`#PM?gexFrO^%p6CeVmqGn{OQRX5pRvv$?oZooV3Tuq|X>q1cnTYR64I zCYSo$`=x_f1V!f9GPt&yCVDijwy4*Pt|FX493JB-fF+>5?vWAp3O!C)P!T?fbT5Dl zn9Vd0)7wzxc09Fs^FWHqOID z8OvTZB(fyzd>Eo&t4y?N8e$fTZW{qRhi5WG3lD5i)$g7o_A3_>yIG0z$I|N@~ zIgoC!7ze#zm~CGygiSi?wjGjlahcKEPxR?B&ChrJ>UHmQcL>f@rzHgvYS=(Jo0BhZaT%q&fiB-AC=c$Uz}~i|^A; z%tvCAMXzM=^3FEO?chG`AacYAw-yO!N>E(R!A62^Qa!>#HmZ`p@*VJEEE2B#v@dBL z9!Z*j=2GM(a1}&VO#*``i`t9pB!_$y(yoAfL^a~Ch5@r25QoQYY~T#B)|Zehs;$js z%;(}0iHAcvA&jRgWJS87BZ-}|WyFe|6xjvO7~2MXttH=hPl!6&YRoIKH1jR~ZYI+f z$#q!y722MJ(yJ71EqK2+QA3EW%Bt{DR<4U9Y~r~+$1@H6O_gy?>)=$FPEFoSO@3{K z@{dypYZL;;pTz&Cb^`%}<>ahq5V)n%jF_BsuCJ_B^W@W`Os!@;>uXQ-zivp#oug9J z48GvTV%c?5(l!=aD=RFna%vv2dLv~GL@PIfRaY&DJ16&G6v5;640#hNQ8o@;Lo z27KiT9&Bz9wAxMn*?|b%o;~SEhqSmAr!yBN(f1G3GeA{*!Kze8R+iAF61fY}s^9HS z?(CA=g@~nC&n9N;5(y#xXBHdUUmgsw1-&`^$^|+ z)*^STcBj6%G?hrFwRP~#c|L6U76)r+Ql997iGBcax;|kqb64TlI;ezLUs<(133#|J zxQJTpa>+2*LPqa6`hZ7^r)+g6+GXcipVOqZRf@~GIV?Lo=}g}~Xtt|WHH&hiUo^eP zR=%=j34+ugOO){cWG6+{MAy(bFQIE1U9{5-CfUZ>*e-*fAnzw=u|@bNgI;qCj#s3h z`Re?P9-13jJ<}1Kgu8FP%jj}iG0iIoNuq!0KV61$i#o%q`D`IshjbN}QH(DIQU?q| zKoDIiDL3HGv${@NI}5r8abOg3L_GG$#jHk{Kgxfa`M!T1ZLjRiOOz*^1L&U0PdC32 zxo8bQLIOJ?zqOiS-R=%=PWq3emX-n8ic2Fv@ZzTrtMMjcQMC`QiD9mU-82om5%W3muG0TJ6ZO43Iw}lxz(-?B;iKZfV&C{_rdS3y~k3-tIr>s(OcKU`{$E< z4E3bTM1R_@!}IR=RGw5Pa}VgyCOJNsAfTjMnS4B(PjGY^IgC^)Vn#}Bn=Nrx6lrLr zG5{HUBQo6^{_e;n0A9dst z3lf1$RD=?g7F+lTgC_MX;cl0WM9C&ZN|?d+0xLf=Ld8Y<(r=bf0N&z``BUZs2?X0Q zC>RM*%Mu+$p9!4hb{}37zxLZ>qkjmEYjP53eOzQCBEVQ$4*ly++HUjewwP!tv@#XO zF|EIc990Adft9RuQduAS^Ndr~!Pv{GV=tvqskM)0c>fslh|E=^T+;~cEGHUZ4*UbU zVxR4w3{wX!&?l0${Z<=lP1DyZ_n{g4R^HUyJ#d1Q_YDG17A8(_y8b2qf>H|VB=@tL zvPX@_aPMnS-*4J=|HGh_Kk-cC>v5TgJT0E<@GA1M3=@!Jwf$gk95{EI-qr$KAAgHU zv&TcFMS3K%O3AYM;)*~AW+K#xYRXjsd5KCdYztu(xTaY@&#}KOy>33rMh-c9S0kiK zw2<@luogK{DeS6!v<4cblyymI+TT7aej|J+(5Q$#FxIy)#B&b%c3!k8W>qbeA=!JX zVD?fFZe5-064lT!_U<;To_>){6n7GTEkjt=tP8jvoS%?z0#IztcW>$#EIPS3tz}FYA%lb}QFN)3I$`(Ui^E z10d}I2QzFu&PAe+ouy3U$A+ETd6^yaTpwk>ga% z0^zE$Wmko~f}|6ypdDtr$L?Ul1~%AAxcTZhON^M+6mwV03+Oo2*~|mJ_Hq(|^9qgc zL@L88`rx^{!tM&Y555wh5yfL3kv>79)8eM~74qQMTObS9u7HwwtcY1=31n;C8j0dCMdU2IYTMsEvH}T#JaJoL0OwCdo4h0Z-X=Zfu{@~ z@bQWu^WU^>ges0AaE)Cgg-H`L$bG(Bk&M(^*E99*Oz>Dog6i>}q6j?-=@F3mF+RZX z3^X(R_Z^~Zi38sZoFv0QZxdha@nxf%^gB?GL3ztVRhofc##j9>El5H}^kWG4@jQ)|? zUX#abjdtW@)`bffd;-aNIQ>txWtCM;txR0b%N#glSm1R<5Xpa02^kOw%=$9^8NO*8 zIO;r)2X0^yY4Q3sq~lizGaxM2Ic```gf0$A_&(ZR9V?oAa`Y(77veH1DUto9^wTrh z{T4F8EY~S5F_KiL(52F`p`}`^DC@|?AP-`{{ne^1`qIyXha_p*O2IxzxbB&wjG3f zj#?$1=?q4(OE2hr?7kTg1j5H`6dCO7G<@bgn9^@e1lx}(L*|s`9Skl zJzR+<`jniX%8pxspYB)9cP)%37<-)TYSsrJ2s3(LpRh7B7+ud;1NUC0!~P&^RFa#G z?;hy_!2;_dVIv-oEt|bmdXN{SwZlrkduu8zVnT6ZNMDGM($(6_ZLy2{`hqw_CqqRk z8{e--vWBu3VBDa;$72Co;;}}yF~}>1`}cme zm6qlz=imGxO`dgW=k&F-L%PHp_x!u&vi3mL@*ZBNyJy?y0x$H$@B`Z}Er5;j3p;Cz zP2VT)=RLPod4~8#961(sIo1ImiKD7nz==O+ttfn;>O*ipAz9>z`XBy_Jzee{=wrg5 z9nRPAaro6}4Evv7uemLZ7C#xoZ?~Qr^`*!4X{*bf^b&6~3a*3W%It^e`fUZm~ksr)q}Geb^cxgg{>n zOXU3z!0iX{t|8{B(PsL87;7I(0d|w@cf+ZB-7Y>AP7pD3z=9JKp#EM(PKqId{)?~8 zmP6oV&QxohA}*i&3H|y#*Jj}DE-;6@!z3KW*;W{S0^tBXjoUi)qniN@hgA0S^v}1! z+xR9{F@phUSI|nD+d^H1EezR!fN7@c5$F60se$I2SBg=IariJU9 zKDR4&aF(OJX_{D+BPQSMn47DAWIb^i438M&Tb^`1Z;cvGd!LTbD|xB(yfdeT`}dK# zGvLQ3lJ8HG8Rs@X*%&*BzQ;Fv_~dA#I^0&R`iX6v@-?^2>r)y*P&|shAuc!)sL|T6 zU_oVXnSz67hfD8LTZ@zS&q$&n+d=pkuEA5 zcC6hZU?(X(I`BMg6%=womg3^^Ci}}JNvViPZ>GdS5%c(&o*TA%Gy*2~W`S>8p&nsu z@0(v=n=@VO{o=nqyO((>+k&dW^7q!Y8_C`LW9SS-wi#n7uD|3(;Ztv3zn3x@1r>M7 zNgte`t}vMbcXP1rJlgE`LqiK~A&^GE{NVub+-up`Jt_MWK^GpDhH{Xr8WLADB}xN^syxW2caGi zHmgQav0j*Mh|CIXCI^14CWm@sYV!vuZ}1;a%;V;ICdCS~qA>9XMg~uM@o*dv|&cG9(xF7kAMZeqK=bHBk`dh^~H~`(`r-#Fg(|x~^B=|GRc`7!(cu zd6v`E6h2p2^NFtMe0E2$P%vj3a6-sbK6-ou1S}V%k_GiOpE4PpE1VErL_Vr;d+xf= zxgUAc(;Z(wDZMr!EjZHUvKc#srRl`oQHIrO2jn3J?;<7}9!L!rzE-BadP*YuW%JpA z4P{nSTxjQ0qT%neN%C^Tld-r0!?jZ*$~C#)-;;UEjoAzM%0J`M?pNRn#L&tkGc*j6o(x zDw1R+5eu1fc?k}b6Hnfw5C5oH&B>+m$@{zctLO=+(L>{_&F|xA{LP4p(|X*Da{)~p ze&eqyS82MsP^d0D6)Ad+bM10lw&2Np_c|>U#SsPXyUWT6rwpqSm^DztGFDP=Qh~2w{X(4hn%o#C|6`m~- zw%G)oY$hyQ(}R-St2D1{$OX%|L1)aOhg}<6x8(O!j-RX_e?F%FsK!hv%lYGh%pKD? zh%lVlZyp|fa{|VjeQuXHRHVxH4V3?$c~g8!!zNSGC%t7ShVIW$)|q|K@dipLpoIF> zoSj4#+YMq@UCY%;+>U>5rrT~cnON&@`+%{7k@o)ulfjK9d1`t| zPzzca#-}cu6B8gOi%lyt*_3(oxcBXkJ^6nagzFEVwE9z1M5jjRpH?Ppj%Jiuw3D?( zUW}aGNoL9)^QDY^@}sQiN|P@gXHu)9RJ4`s!g40ss9ep`Y^eTxT9VL@{`@`i?imx} zGLmNZfi{&}rtJo`tE1mwecTZ2Zdvfhp_r@fEq|lZtfjlZ3E%@M{z1v)=jcgs=RFp9 ztx&e3?LcMC@v=v%8QM@E^dPDPW(aL-8h4|Whg^ERe)x8TyDUt` zHD^^#9x$MoAfvNs#0fs;U+0WmXzZkNg*92kobw3g(K_2WPbpN5nUK%cT$k2SE8Y!m z>5ZtL7{U;YCY@4)%l|Oq6SdYqKD7C%X@|XW6EWOa(L6bgM=1H-`$?*q)1I3zs?B{O zzr3WX9Ft?X#+9Na^-{PTIAqvJU(u64hF_`I7l>cb<_&7}Sg!@<3#kbyL&$+BEa{y{{+ED{%n7(r^FdgD zMYO>+mmwkW@d+#VK$N#ej_@`babPLY(v_eaRn0#zRue-9VTILZ4+5hjbj0r#*vfZp z$87$2G|7AIr#V!rupel1Bm)uvBeb=SI~{~id{hccR!vTdH1v-u8t{B_>r}SDJ-0aj z>?*yjPo^8M!!TZ9E+3WjcebOMfEtNIo(`@YE z?;?-8h({vl?OKUOR^(c*ElS-Icox=i+OhU`HC|jAjY@YbGV|$`QGC{rsrz+Tv*CiVkP|*x=Y(J1@13+HWgTn!hE$-@EA;$Bbt6JR)ya zvSQz47kxUkh{TIo{30Aj$pl_-o4eei+4`FCJo&mBMvT|V zlNou?{G~xp=ehOR>KA^onu81*p#;BGH9w5XLBl}+~cVlRD8bm1dSnzbKEq)W}VfN+dtMBa|FWq7{ zz8`vjet4*heBWev_nZFRT1iO_5O|@TajITVTl)5O$Js?rA#F>3EA2gJM`_W~>*)TP z{QI>pEowXLYxTHd&&5KXbU$f!_4d9+cM>PIF~*s;;PFBnwifXx&J%u=A|2b7+~pGT z!0#~!4H}LXv;W$Nun{ANHXUY-ZGRZv(YegAZR@tBEJWy-z;bPNOgoM3Y-;*n*&r#+ zA7evp(^f9XCaxTG#@x*H4CEO$U3Ij5SEuPe9_ovie`W^}866io>*T4&^s@bqj(N!{ ze3Pst9~#jmkg;!i$Twm1X{`5TZ7k~h4wTFU{B*nXqn~9Lcs~no)%7j8^#w6^<3^f` z8y-QZ;ro?w5DVVBH3AKM*gGIBC#z~o&*oy(=cR6a8G<=7Z_%gOf{ zPa)6N8{jJsA*4MmzDY$J~{Bhw${<&?bsaqTX|dnrcT{NB;}i;twqkm)JDGf_i~*}I#7 z0L_w-%~WdA#%)%4J9FBaoa+~aOWE26(i1;*zl?h$z!kP3;O##|w^rb;w2BxW*WEYk zevBVS)S3q0W-Hc3q!JDK+T%6^+*$eRw&G~Iu-!XL+djNUyO~et8Sg?>87vOIS|_@i z8!wT4aCmNj`4E&j%=2)zR~p$5*Vc_FT>TZLx+siNtDdQXUzg^&ae|_^jZX?8CE#|fJfkSRqdfd)9uD@-(&uMvEu{{rn%sP-x|^;+w>V zL_pQqYpx)=3)Tc*kVFF2zJM-BVkG-8Vsyfw{SRYW34Bk+?K5Y{EGQA`9f%Vkrx=w! z0%NBl+SGd>KjyE)UY0P{{)LO*sM0Ncc_z^@iJ6o7Ti>TZFLkt}H|Of1jazkcgF@g7 zNxzx1JfP*eM$d@{97uW2BMb{+3@R2_L@N=7U;*2&{AI7N3sWCi+He$bIj z8`i-Ba51Am^YO2rMOWZ+e*f0gqM?*pVNqV&81hWjB1(1SU1lgdMh;wNk$<`7cKqin z%Ix&icN~C`*(g_ccI5S4Z}@n?TI9hkqWdRns(FaSDGlD46dSe7;DL>9)_y9P*G&U##4P7`fp;Cd|ETja_{?np>^)FyLE88Bbh~D@W@KPj^sLx3&b9^I`!g|m+ zpj#eGP0hxXLp7{gM#f=tGy0dB5}K4Ufi@yh!Sx%DwYd zOsflzl$<7VQUA>{(v?*xW1(^k*uB`=p92?Ug1}mT(}DWs?gQCIcLX@it31z|WXI!l zalY2m!mb062eFuK3X&4_DL7g9CB;HoY-9Ysw^K*5(veem%Bptqtn*Rqu zW(LJMk{MoeneNQ%g;!Cm+2chpfGvJ6B#g;VYyhd`4+8&GB1nEZ0k4^v78bGBgG2KtL6RrUF}gBceB=lkaBIHkBD)NC3@q#^-d6x57N~ zW=#!*C}&t>u#KhvX`OpvktuP;>>Y=7Jz3ylWVWU)57A5BV>y07E>`@F?iW0AK*My! zh!+wiCbIQ(*eHEhf80y)z;vP#!#XZ}R5Exs!LP#W)xGs^#QNS&FS2FwIUSkZ%IiJwY}D%AMKPU$ zyjC=Fv2}W_R>buDEH`_vxVPcwcqp5=z!`7_g3@Q;-Z-+EZA`n#()95llsu3EeQ8hUups`F2 z^_y?T00S8+Us5x&6Gp>R46j&0yc_;?zJh7+K^=4v`0%B`Wp}o*_x>W=^w>#rie;LB8wh1Kb^3mcPAO$S_2H4toD#yH;gf~to2qu^TR$gWTbyUR zEsIvJF#iHZG3XAmx*ERC$Oi)Lq@G~9+4$pL27`xk9B6FJ0>tB{=+7sy9k#eg>^{E5 zXOBgXMsRZPx#D(q89zj4Z1d8sy;pzu8SS<}Cw>3$qcHWTzxsMgOj>O6D>{=VU-C?} z=A0gXz;Sd@GIbE&8T0zPl(T7iCL(Wf3zSk7ede)`4KPBo|Bn&oG)3P79Bl2wfAS*o zf6I$i4S9R6!;YDu1j^sgRP6j1zg1+typ?sIhn*LC*~d!ca@kQEN8XzyEc=3Xe5s{t zCrhLnUj4x3m5X?;^>Kx|-;S`=bxr^a>+z+;1|sm>o94XGxk=V~Y1x^odIr7OTcu1a z%vMjYrZX+vyRAv91NM~oNcSN%cYeZxWsaFBeP0JOl$R5mV9AOk-h?aAn4QRw7B4%= zNX~yycVwS$N*;G@tW&K))5zaYDF)0&)=!mWQ%C-N-|jBzBsb(W_FZi@-Wu_moKjYF zJQYY5r1A}muZBI@PKIx>%OA8k%NZaZ=u7cX-t2TC2*k}M^Qo473iqqn? zHp0+U@{-0~fDorF_Dt4lkQ6&mzZ`>e@e}O3%BOmq)p4}g#9p!4`3Pkhtac+wO2LDh zVFt)c)&(FbdytR6HvP32e=T4^7{Y*+lz`_tn zs%r857lMONlE3&RDZ-o^2hQ21Oc5{{%3kA~(K zqO@C5Tqz1hjj#344lA30(RKOgodGrt+t4P(J`Xas+E*@VUd_fui`&Rx|7*L;%xK=A zCED;u6vLl>9ZS_3cz{Na!ag$bTv zm@96 zI2wY&q|a%+&Th=6B$rVi`n;bFzUn&0iPG;7ty)^ah(Mau`|ZniWm~WG8I`qTeUfg| zwxo$v=>}aFsv^8x#l5Cy+OnXZoEAyPbI0D$p|7a~Nv?w_jH4s*MT&;WjjV5=A&2%T zD$74L#*E37jJ;S)w3tISdfhjuun^h?&eC~y-iqOqQhmpCH<8`pY!j;oP8YK{2gW?g z*C!c{F4L=vb+iHk!{xjI$$noZ#~Oxb{5JnH|Iw zJYT0qNq_W4%68B^#~S0Xv??}I!n|Hk8cE8KK{ z7yMm}{6KlOP5TV`nbmOYlLsKlJ<{#l!#xt!3;6&LSe2*|68MsqWduOqjD?L?O}!FvhFyQ zh@YWz55t@Tw@>i0n;8(CbaVT82nqvO1Lk+Tf9#F^hQOSbmz-k4S*gLDt8VI%KWy}Txz@NSF*d&`Z8?SPH44K!o`&vVF{h#J;*^|-f8C}%rR?&g#c_r&WU>4C#vzaksS{$ zh|w#++db4lom87|R~8%naY#j4_KdO->9UvXOl7E zv1g^Wv;LYk@$dM?>swel_PQ&{xCwfGf?r)-Y4*iW>2%{(J#8mz&9A}+s@kxY_gTsn z{VR0Cl918oFouPCH)YSWD90AzgMMjW`BZ=;WoAr1jq?$VPZmOkGf4`lz3jAs04%`> zTLBJ|?f3_YwgA9lkAolh71~;*Na@(0VS~uQ0s`+rFd*O;!BDjj#P^`y<^sw575-kA zgTsn8Ff%3;+ZYBM`SxcIa0egcm1DzE#fRUX1ZitQiD&QI6^a4Vdsc77^( z0)}aJWajr@QlzRFnW6g1U`ZUg7Ro`dsh@icN0xzRlMaISRqJ0r9bj&OsWyBt|Ca2F z0R`~XQP?S^3-Mb6;B2{70^drMY&VHtrd~?v_mMQ+ooqU(fy=EgS9Ph2xup>ciVAbR zC5J5aDZk{H%*H8^gOvG==b}txWPO3zIh*`fvR40~s*Sop>KxeiEp%$Jx}93H~b&@_8S=^gTh z{{76}QuBe(nSiIg8Jr?Y zXbSflT@iC*x`k2T^rKpfMLlNJO5KGSzXA0irFmI}* zs_}#9Ne*rPPs>4MfOS!>#++o%`G_5Iw^L* zU*^>+9uBFWQQt2xiNTEF`y=@-ia)9%atuoEWNvd)!BnB@1;$JcfyTxg$jDTCUvq;ml3%#%Llh zWq?ZAa^^&Fg_3T%`#+mj%#L}YtLS?o+`2DR;k#2b!vJF zcH>c&P+Im{yX-}-+64v{?t^AXrHi`{n8p$ci)F z))5Ua#9_ix(%kU+YI-2SvQQM}di%^uJ*|<|%;lUda$R*BP`5W~;d?g|VdR3vjy|!| zy9YR)0tt284dfmpsG!-Ve%u5@3r@@y6XaVTx-{M1CbX&@{|>@A_LFexlKU#Tqvr!v zZ|6+;m%o3G^$P?+SZbwX_9p&%Fq6<(^Z9tnK3NbbcuR_G;EHPf(7bk()#P)Hv@5~T zl*(l@UC>cxRiI^mp?I=_wpVgI2z(Ln4Vb74rmK_JC6R^c z8PyHkaMCG*Wpl$rdn8JvDVSY(&FADQEXj@$hJo0Qn|&nY5HG?sTG-`gn@pu20#f}4ESr|9;7ju<(aBVH9OdJ*IjBnBUijVh zRXgYs^ZfjtLz*X|$c?klAtuc*=WS?Zp@@NP#oU*bpA7!HU5B}qF2uSGeP$*vSauH4 zs7|ZkoRw2%H`yR;;50&IS&0UP6qkj*Sd^f=csStM0__TU@>!jrBQ~V6f5*rC@|K;W z@7NQO*hHfi+HsPI;(}Zg)*J5&3np{N?u1M)xCE|)uM_S!!RUsY6_2`Dq_bZeY8G!n zaeGK3#h;uiOdH7Ra+dkW(!7XNMT#L%=TBM8!~Ux<=Wh5G$hrnQ0)XTh zSk`zBH1ymGD9(R0lg9%Kx-=uWM)iMC!k?+VC2MvvQtz5A1=siWb~{s+(zwZ--3{M zxIaMbCNQ3gTOO5Y98LU?AUGPRo47OGhvy!QpLZKyO^(A(8{LJEWAA^zt~R+Uz9Aym z8oBa=Fqm=NZWz`aULj5!#xLXHMvDp_j_d!;eyp2pbQBV=X|BBFlkG1}Hh?W12UDlz z7ml26)Y90zEG_e12+twmT`e-ZPhAi;V?`;`kr#gQWZ(SaZNSC6xaidGUN)!s&FZVk z(ulffR)Oz^L9e}1+Y&20+(^-LNvX~IG^%9Eo=#UTnzP=IAod)bqA$Ms=Y;e-G5G`( zTa!Wd!;Zud9*dY-n3Btu=LO*%d-A6YBADOkT`^noHDc{9^HkR^_CVIvz0}6BY4q(; zxSI%#R4EIS0i+mXU0jU|C48JW$c--Pl0t6(n6+Uzt-2U`bD7>Q`$xM z&%!T$jlPDZedhvDw~J1ybxRTxkQsD8e1-RGcv``3cq~~<(IUN4WF0c@MZN1yTDl+e znHt6>to-!{5xVNk7m^(zoH#WkdGfp^$Vg>acUjkQ^A$W|Kj5BY<;Hn8=(7x@`Rz$T z1n2Nayel$InntMJ+nlAly4WAiV92fepF1h(MXC{lfTFQiG<&>eUz=AA{fK^5K6wX^ zg*#w5&Mqb7r*#~9yaPg$u8@^nxS3 zPE)))-WDpV%LQZhNP(dFMWfyXtB-5E{cD)$J~PvIlTMUYwDJtNOv|P;rH}V%T|thn z*wOJI8Fw2i#NV>-F(9#lS6@%0z=GhmI%1266`|HCy&9C)U0X% z!w=;wy}}LEkaS1CwimH|B(p5LI>(qfey9t3>`9>p?^aV^;p5|ziNRf5spZtHUzb4z zwR1YwR?M0mVLBN*Gkftjy}|@BW(7uXR2TKmFK%7agrXNI~*jt3ag^b2-7O`mltu7A1BIrfKN zMr??boI!LcQN^f3?(>)Jbsdl7=-E8_M2nTgPF!Y_au=rM3g($#h7e>oD3 zDtYoyX2zQEx~5?CmFJ_J8vFV0G{mKB3pI~Ix6TIfe5oM{zCY#WgJS(tq3IW-TAw&r z8#jEaOx5Nz?ZvnW3;+KpIX^YJRPm7v*ap)ox%eN2Fqu*>UF1eJ zE5%V)#>$|g?sRwX6b@|Kr(1e0VDslH4a(Tbc`DEu<4 z8|O*WbH&Y6Ivf4Ab}uv2sQM~RJ#OP*6-%u6jrnb9?QAow)W9_u>3~?NqkokmUKfZMD50 zG4Z<5Yia7i9yg=J>hRQPren)cxuu=l(6yIQxV>k|66ErnGmGnE5X`}b_&$=m1 zWcMUW{~b^8=a8)6RMNYl?(x($#2FFp9Y}k?D=nBo?2DCcZx^%>Fsu{kGqIzk$pgi4 zT~=Nid{nor8Y_=-3QF*L3rKNL(yO2C?Sfd)KeAbF`Z9QDk~@p%mR+erK~geh7>;<+20X>xp$~$H*mi=@&(Z+&>o$hV%3r#`yS1V&t|pfh^W%@ta%*N=|aGFBlWZ=uVDAc6p{ECTfjexMwZsnV2 z0|?52&z|{v1XGID>OaJ+P^D0jBE3HI^NN6kUTM+XQezPE?^ULJQ4&u*xn z*IsM}-6v0ec!|nn+yXWmKcJHK?7d8*I&iQKb9UrJv#UhhZ)ZEw-xU+k!Wu1dKo*@E@*@1?lPS*UkwF^<+y9<3jJk@gQ80bg<6^{ zI$}|$sX-^Xz>)?TC#4aikO{6@D(+;|zBqpJ`RDnxhsu|P_CIXjQt45><%TyF^&JG|A} zpG#a|?uQzOH`n>eLn6RGR=1VjwCsVM=?Y{<&?|g*ir{6&tn0eZo7+%zXGhGQ-676c z8E@?*mT?68YA+iv{XxL@tFg5-3OpIwAW0SWr!8*W^Mo*Y2FlityUCM%n$jFn{i?hj zsSrG+ulfi3p++2$3k=N*V02RBN?#;gA^`C-O-PA4DI-k|SQV3;kPpYKKhWT+a#}K$>>_PkKGJjU2hS9MA z-6qCq^7){E<4COioJjObBlq3wiG)%NUV(o4W^&0lrUOfoPmkp&4_u6$JfKu zOEFy9E|h&7dq$y`J&7^q%Y;egYAk5cmoLIL?{(OOi|lfs32R=ZhY$Q~F~UPG&9wDe ziq-)&cXA0jH9W2%1S&8Z1_IwYJx!D@QM{TF(|xd!!Q9t62Terg9GkjSArbE(e;v%{sf zDoe`W`^j!%c#Ig9zBLKd)KmV2UKFK*nArV>2;2O5VYGF%V5wcmgMG)zlN&1MZI&US z)j5puoHhzdN#^MLWzMV**#{lFtz6u)C?-2;%|+vo>H7_8-0kjl10{G4B!>&s5W*Zj zsh!(;hrw97(qI*V(?u7Sb)(+zLE=1l^$)|T^oEYBE|$}AqP{$)p#Z9&L(6HcJ3#Ne zapaB5?8hhdnY*)=@&@KFtP8vMx{Jqn-=WP>A~nG^Tj%+Uz1eu1W9F<;$L7I|)j2mhw-Be^QqxO7oE2l!4>Sz}%Q0jc8A%3dNG zE1-!f;eL^y^|^)wv#0v!4pZ;#lfMkd__Z)xU`#23*X0)Xn&RbfI(9QR8XJN?*Tfc| z3o@q{8ts$r7Vp-OCYFo4K^C)SYCTp-!NW|n2TaizJxXUpGD`yzVRM_eK9 zJ~!E`UtiMs@c6wgN4VE#fw1g$=Su;EIDIM9_46c&*Lxjn3lmwYN-Ji}mn&fC7PxH70+cF-lsike~js2lyJ6U(p_`wk{sLRGqyI}U zi7fs-TV#mplQej(3iJJe4k*Cf@+LH%{QA%F0|LB{Z_yqJ9&p6uuMh$KR}`Q>s_}T( zfDnL~Ckb+K63gM3`u3_=Zsizz=W~Ixh==e!e zMXS69-4NTaRa}dFJ*fX{C7n+;hQPW|6QRl%@wVy(HPR+QPR(RU>$YPpR-v+KifE~+ zJfuS{JFL>6>^EU(XuUc`N&M`3o#_6JR%Ge-VkYYL6AtP^FWfD_%iD?NX^J`6mDuR5 zvroua5k*78wO{(pQQ!(C+cGN0#d}g>pAF@?GcB;8;&%Euzfy17M$T5eR!w)Q>7mL# zpZ6C|zL?W`t=NP-1l;OMl+An!_CZGZXNme#U7xt ztD&-2bA#?J&XBPM@y^Bk^7gH*-EVkn4*c>4<>-oxsdBp?qQx4n5?8`A)ziNutet-J z_|RTmkbrV{tqA_JuXOTIQ1#(HqYs5M@*Lb-9sjDl$@CC;^|ern!#bt4IF!q@{aeEH zJjYauGJWvC56lfRG7M_i%ro1P`R6M3!d=z7li(wdEhiFAcQ5Vg;bq9uH?nrz^f~aB zghz&M&`0>PDqDqT(Sd#N!9bAog4{Ll-oo`-8xEyP^^H$~a;Z&SdLSAQW4HmT=Bj>T;b*2~Pw?FH1z(o4JVme^#Lvx^-uEV1id^@5TnqkKTo&vz zjgTne?~YONPE;Pt=T4z~Z|xvvi-=`=wy=nujj_U>%D8c+4)vZ05kbt_vUqqD#?7 z#U%n!3ZB%wlf=KkR2>&FVpCMhj7shtTgd}s#8F3v?PB`RX&Uea`E&V&3zTK`UR&iS z)QY|Mx5WjKXOHeD0+ofd970W^Oamt1<$9S5kF21Ow|$q*s*pE{0?LMiN&sgq0orf! zsE-Xb&H+D8+Takd)ah3f@mtff~LGcILP-PpVz z6;3%YU_;}Q9VfP7oNtj-m4E7;>3mm!&@YPT#9<<1{9C3z5qGNdQk983Z-R^h3|Mad z^B@NtCrQw8E3%~eq$2-EA0ueJT_7Lue`=QGuNrqiZ#5X^rJ&9`ibQ+tNg1=MFFW)k z!=<+y1b%pn0f5D9Rdy9y-J+nlMPKv$oCbwzEq`JESGpeodoR&_`2iI|O`R$2K_HD` zYps$lmFLTjo)1MvDG9fZHmCQ*SC%bun)F_^kCM6P2F3yX*GjjOs?Ulr%YM{- zu8CQypYa*V$qBmEgb&rTVOZ^mWAR~};C&U;?Fc?#H6|OvZZh6gT~Hnr6<8U7^+_8z z>5ZMxexGt_%CjGrcSns=Mb|O*cap>ad(5`=0h`KYfZE3W90`TTlA{)*g&U83xM8+A^e$d!?~$|q=D zCBVL*drOYSfmjxUwtK2;!C=nD|M`$O2M`fM6c@dGJ-$`{AhOt=8+cHtz^;}b#PHBE z_@2Xjtu0OIex1sD<|%#FTVCGO`c}(w{ape*hlf#Q!*hAM5V-E$8;h z(wJUT__0C*l9=fgHIlHd{!P|~92>*QR<)gkH*5F6&z_}L+gGIeMyA6?v=F(FM&4-s zeg%iUrXkzayJ#?R%EKOE4PE+M%ZEST>2#xQNhumGH(HYyD_|ijl%_JXF=!WBfR1$4 zrfMGlNSv2#ucbQf4D?`!^MnDl2T`DX%J@MFk4lq0WAu7R4i73oi%H9 zRFJ#COJQStl^k|El<&K0bgbPml0McAIyJ)06kttdneP+O3^3>sVFW^jg22`|FT<^I z0Bcv@FBN$@0yyv&Xy`zn0SwGMJ_-8w`}j)+v=8=N%OSNnpDtj@XwyqN0xtMC>!>1w zMnnMyP)HyNl)Y;O;E(|u@Z~M2Bo=I_Pfg|h-Hj-W5A+wG1m9snH`U(71a(PKiyhU6 zYX1Y0^FKK8{roy3`D}OE>^aAWB(yt3d%3G-3N7l5f~44QTE~_T>Hga+rGi`L z5u&1uXX=&i$}Wn@4qgIZ!^`&CQ@SC_85|64NCVhQ50EDN2Je9p;9k}>v4fRpkWOSr z_dkKx*wfM&-(vmZ*pV|U$rpIO+|KbaiU(s~jXq6Y|EUGp;An3VrPY`REjTAiD@j`@ z?B>HcQI9?A4$;(4*6MfZhg6YMlqW?rVy6aPh3v&8BOG{XBLc~J_G_j;;l_1Q7o4nh zw0vw3os7ZGkp6XuU`0WrZ)kFuo4mY8-~N>L(9cIeOI3Xrsjiv`ecu-#XyQjqnz*96e}r0{}O% zr(VaAZ2XsHA8u^tQSKIu+6B6|c_(|8%%8zNmK1j5b(NpWhpV}hJ5Qm`@tp+c9}q;T z|KW8?hl5J%t?(7FK_sRP~7Y0qnWZ!D}_455C z`vSfJke&*+w6$FGPUAh#k=!qs{Oc{Cw+Q?;l(4;2tda{WFK!90-nsS>@1L#%G20RD z*%mZR9~_06X69CYUM1dz;vVQ^SG7nm+Q*$LcXJL-BuIrDWAESf#>jZ{ZGCwdo(veM z^{4&7sT-^Bd2}@=a~^?OvsJ&m7Ba6GtTY6R*Kd85$J$;2OYFkp{#=hG=}+9hzw~O- zE!jLIML+DMiU-H5@9TP{CKS!A`D5{o>{^THF^vwef91%g0Qa4~Ezanr5Rg{&&Ze$z zi0${B?24{1XK$?0Bn?#EASy%cllbnhDD61j!jY#FI(9os1~^RU{@A~AZQfc`=+6y$ zIWeeZ|DMQD-3ei1Ew8Sx^B#0U!R+*kwQ1vtA;y=F!hLA)U0ttjA~^+t?rq6uYesA) z0&5pGgR}MD9RQIM0#F@GC>CqpILMkcKOjW7T`a*LRmobF>1Bj0LLkW>j0rqGe%k~p zp8U6&2lV*hoTM+{34*jcIH^mZv0w!jp)++@H=QzL_gR!68y@S&i-GLzR1z@0HZ1 zMWVgYh%K+LfFfyekzu*ksB$OrDT7cmuJERp@f!v5z6P(1RDltvh z)djy(i%uYS_ zsnXF88&y0hE?$S=O65!`8@d>F{7@&jM8%lMyDC5b@vAJ9%!9IK0S^|A62opztw>Jn znO|`T_0pC$cpY)SJb-kmJ<>&B>Y3U1w`moL+#C54O~BhJr6FqF&D~Lwc;gdjj$l7( zR#6F%@&%Fp!Yf2yvSe4`$lLH1DT~Ygcf|+=b;eSW>oO-MRmo6`3~hC7jFVE=J)#Lv zB@+Kakc55H^y!<8;td<22D$Ng9eB|4c~LT*J9<$M)LhO%QIR&hB%R#bSKF>pl@3Vq zVEeSIqKgR?CF%jHNP!CNpGb9AaOdcL8|$a;Pko6{3i9WAygd6> z6MsfvM}>4d_8VvH5z17zHt{Js+OnNp^ULw0DX;|esjEpfbf3yFQiXOEaTpzNf95q| zwy#<2rAR;|Ibu7(%q(llYHyU#qZyFTJff8LPS!t4@(*s+7*91&+IOg4?2#6k2|7!t zma-GaucsS^;*o9m5?PV>pe6PQ`-}Gqek&=a?0JE=rYHzirG$o5Psc3yVGR3G>D2qO zqa~@E2<4%Wuz`cSF6gG3zo~yb__C59Xe}|v^Q9?w^=Z(!AK2tNesGW~u0`3@%oxc- zpWdwNj9ynP7Jn7YH;~8Qp~bB(Lg|ewO_6xes=AA9T9fwktt|y#iUtO!N1Rs|!yYTH zI6mM;h2gRhMxUX z07}93h`Ih!!T0%wjy?N?j}Z7Dw0rTVrPoLFPCn|dR5joLulUyip5(0eWmcUtCk*%x zGL-zaF;-FQGU>4H8aRd~VU$h2WvJ;TF{v)+$GzlRq^R8Fk#M~z6s49eNRLb=?funn zAm^gAuO*8Zk+Kkcs%IT>4B<~y9knW)vk@W{fI_m6NnGez_+zqvCcl08`{tgt?s>cm z49jEKXMIFzxP&JU6%sxA-IOsWr(S=7a^k=Z(*C2I$~Hv;<(+_rB0?WOcM1QT7WPhD zJ@#$G-J2VN>QPAyg!zXh$_Tt`McH?}@S>_~i9f0{shl2f2!|&J-bKqUrN9UGTNa5D zYEhMC^ASlz6s&nq^OGhtcmOkE-cx`{JmOE@1WG-04UW!)o)RR3lMg&Rlg&?w!$G{^ z?vf{CTyYn)V(+36tJ&8kTU?0vInUMCg4qUmfvUh58l{5JW(C9 zrO54NIyhobi_hv^I^za$`*7T3d{I{Y6BiEr3OrDdJ7y62&Y8|C5jXnFPUtxLy% zvZXb`%Z$N0E6d}m$jN&;G_ly{i@}x#^@~Hy#b*WD(FChcy5qCIh0mZ;Hy0mVv;=;o zwLX~M@@3<3H6S6mn!rd{DZ%0?ZzC7voivWKAc(zbO!N|FuMZ7i<`wxQ z*)RB;K-GV=y=w$B;s?c_Mv%93&A0Tbdbu;ilbGWx2$j9i@^O|Iv>fi&IWwxEbU7sl z_whvTd719nv;achT$#Xhn=sbbCa37Qji~iX4chz9TFADAR z{h)EIR{SQj#U>R?42wu%F`TvB8O>t;C+{-c0qFqmSnwzw;C87*uOQRPN$SD8Mhx^7 zg{ul<yvQ#{SU6wiByViIjuBjkU?i(|oJs2PjIpUxClFQFwRG`e{*`B-w+YC{dCg z_kct(HIFqmth>|w!JG~K`O_ZSN=*{AU*FB2_VN|jOW(t{!QG+yTj`L2b7Qa znjQwewY-7@HJZ==Rip9j>i2(?YlFrlVHucWu_E%G09^kmd-07tOHwD;hbXPjOJ&dc zGvZockG1+)jm)kZmD>|I+m|n1g^Va87y%kJ3pQp4AZ61r!3isvK@VAHD>XVq{k+V8 zIfeVfp(eIWHpJXcQxv=cuAYoUXPq!#Ud!`{u}w0m_hT#<7D*~AxWhYQ+aZ_qkO zj(-v9aG-Vp@RlI0|8%4Quu;J>OOZrV)Vv?~UKT+>&ar_MSaY#v6E0a}ff=xta6(qw*+kj9whX^W zlAb7`whL7nDh4!P@5%?y@iG5E?Swf!jl?jf9d_)lVn6EROO7|e4CW1p@kUmx z;KDSYi3}EX=dam^!1Fe9Ct3-1T%|T@sN?IyQXL2-c^4-6h%^h`xw`0ZNn-+wYvFtb zYtDCRF|UEB^EQ-wo5OLEjM#ztdeH@LOWNyhzsr9>$Uq(3(DhmQf(kiLp3lLU$0llj0> z>TjJ7?QAFZ^Hg2ls6L-woe3hz;8i+Sh5aA6dQ8%){3zz+-7Lq3cz-D9B#M4IdFXV! zCSj8~VARYV%NT5*Y?wpPtB}i37sSR+w1Y&)G}X_^u#W^p+6=bp&y`s!Ae*ny4r`O# znj`oJUWUkT#Xa#s{Ml(66WmANcAl4^I~{TE;BAi_jhP_(rT6*lMFGROMO~Jt_Lwee z+7Q0xTuD~p*h>gXiX1`K0gJzNeBW8K(u^xjcSe8dUmL4*li)fuOGn7KF7v@P!iwIu zJnDPOuR5~U19zy_ZBSG!0+J#V82TAKzU`t{%aVJpVw$q=y8@Cf-Lm)knk8A8i5M{vkv57QUj5yi6asqN9D%3U%a-cIOTzX&)56RY+ z5F?&+mT#o?gGn z`07MToExHu*9veq;GiOi{uP|C(YS~-1c)fuV(v?Yo`}P=M_#FU-^IthVfVEVd_Ov& zG87Vljq<4|AG zo0cNVQpp!FmY%AR5O_`kV7!GJpvD2tMu12ChnJ!IqXIb}k8X=l0E9q~6_p7@MIOZT zaxEzf%cMYQRm(h9R!Q~CiOdWk+WXueOUbY;z@N=~LCmUE-8heHCaG%CMj|R{`ys%( z1^)T?U1(yCDiarRd~*^2ZI49VH0_o?m zFFnx}6Ix$R`EyT?Is~cT=ot#%GBr`rJZ|QQMKC5PGbRqQlokr88I5iwS+i?EWaXJy z$tARkY@`_ZcrMC&Kp%aHsUQb{{8I=@^ygD2&L1CD+lcEkPkLpHoX!D}QK3K#f6%LT z_AVhzLzo+)>|6&I_gbd=kNbe~f zN4c4EAd-2yW)e!#d}HdhrH7=-DISL*uL;YOV|c?0C^c{&3hy!;-f9qGf9hP8{C-j5 z95bb=rvf>BnW8OEZsJ+VF-R~vZuonsbH-|=sZuxCC*>6nFG2I+oJ~1@omtiKQINLn zi)HaMuDN&2>6Gut<_()@+xA1KXV>zZk&f=tH#aTGNY^?M$c^@nom**=43eC8^U^(&YtN_=Y&S?8>cfK^x{W@wW4g z#o3axfLblzSwXVsgLcsa+f)U)rba_Vzv^(?v%g*#C*`U=u7kro&b=dQK~vPR8Eh0r zr$*hLLu%~UDRt$ao`=$oW=cn0T&gb^jE>C;fdE~wA#i#-t+vAa0o`$VS+00q;4CL- z^MmGldv|B1A7`q}DoS%Kx7yL_x|!N(;wO>6Fl>fBxbKM0#}l04o8`LMJ3d1CsrOLQ z)F1tH*bDc+rX$iEHdg$==}EFAlsDc{FMPw=HR6X7)HQ!k)FdS+!XgtX8AHaZci77K zTz6HEp#cH0qcvh;1PUSxept{sfrYJ`Dl2h@qkS;Ls8 zTK-6tbItVA@B6B#2OI;6Kf}n5_upcBZ{)&6c38DJecHxJYJ~*Kb@U!)TAl{4@%0Dn zAv1zgQz}l!YRr<%l0Rqp`s^jDXSYPlLj1i(*)dFp4Uc)RMs0n~lH>bI2glN*`;9NU zwz4^8d4$c@QKq|dM3!wC|gq3Yct=H zq`8o+>$sQ|TSf~Eza-X=H{Hz<8&PRCWobg+{KUEB!enyJj>|0!x<&H$bBf>*zG~Uf zoUqc75U}q=x#X=H|J~Y0&WQZoPRb1TaqWpu95&5j^p<0$`_N_bbSHrnVK{I&u&h$%N z9DZE&c4Jk3)t4Tk%(?DvGkmG0%K;h|RfWg-vv2|XV-+W0XiA?Nu);mc`Hxro7Zd?t z1P!kFH$%V*B1PAEFX|tS=)fJJ6e98iB4o*k8YrnURA-;EFgMs$PS4ZUt6UHKQAEfU z%|A07iTfNmxs>hWx+V1MTHY43q{_loP0cYun(ui;hV?VS_gP+r6~EhrqwMn^8`4VE z|G6d`hytPv7Slw>c7B_emBIFeBqbpwt~+&i8oqdv62KBJn6}z=C-!2ZsR<#{J2*!^ zRZgMdnkrXuLJ~=nTCB3BDu)RV{y|xNStb5q+G0k~`YV$JVetE(?d26WA#0fbfIj8v z3308Uj?C@fFLwjd$=}m=#7sY1@Vi2a*_z&a1NXtf@QwY)v%i)nB?WQ}&Xes%R3M-+ zS5-K8K`OboS)oT`y#ISXFYpy_t?y1H+I~NHoe~qBU{SEd=lOr ze{w0y_TP*}SV!yZ{rVQZiK68sG<~tVxR~S1Ua8QYOiezixY|yS87N`@^5L9UAbRy3 zZ-=dSX1NE@RlfNf?drTg9~C2$?o=Sj?kNF*!9UokVU+Z`Nf{a@ozdV?QfnRAWus~b z{{t%6!#Db}Q^!2{2VUulTSwhyxVT$m^gDNTsTSqph*h1u_@{%t5RB}%G=gg^l5Z@~^&}8`9 z7-lE(x?<`%`N7e|MQw)tr~meRrrd^t_fPj$k< zar1nddZ!ROnG@x5JyH8I@=cA~2O;54gWUye{iWD23MNXrWO+81{19ZIvL36ph8Jd= zR<7}Ud5%}HsnvCnklw0RF2Z$iVFx^*wXs!5^qmYB(uogdIuP1k(E1WPP@GR9aUyBM z%NbN)dP0=rx}6=im!CaeF-ZNlZ@ysjebDW%&8A|a<$BXHwp%LA>6K4EC&xMaodPpw z64l;SxVE{A(|P1^Du~@_UIVgmT=fXEJSi?jh1MYYck(t-g)?MZBIhELI*ep+h(OIX z>#~H=XV|B7be0(BO*$kdANh;>S0WPW%#!xdQ>n*@dZ@;s67*3_E513Fv8+)M7AY08MzNv|-*^D)UX=1^P zBEnp#`|s+%wCqCJQ>58zZm}AuoDVi|mJj8%DI5$5*~is-CGMcxf>y6;=XUOT{*1mWkh~{uymQnVY>wUiT{{@00AKpIvqCHbTkUn7hH+S5! ziuJ&q4)7Qi@jM!ho&b;#5JChc0FB^8qf%joXXV0l#G~!3DLmKqoP(*#fc#SiQeiVo zo%~`N3)LdM=Ssd)&i=Cl7&r2o9p7vt{N||HeZV)P4+byrJu7;q8?q9NYG%8mh1tB7 zF@T<8jaWB8y^a4T63R^^f&@P*ZoYZ8=Lb&cTj1dz?8a+5hAgOT4LDEKZ`QUXrV3Pq zD~~Xt!~dnQKt>Os0EBQ+TUR*Ce^-z!y=okB6_Js8zdccaHQJ>`pe}3pTIAyQTleHJ z=CUlW?4^Oe1~u)UO{U!2dh~{#)u4)e*QK1Jcabzj8P=n^#$c)vFxUXc!hz}WWI%^& zDw$eam40kWS*^jgduTSWQbn{eJjv+$+L7uR*;^5ILae656#Q@{{Gp@>8JDznW%fEH zwUTTrFAtHQ5E%wlGEP`C3uWSB$S%Wx+D-*Tey72wLwgu+CR?vsla8xU4{N38P_BAk zMdwdQv)Vsh{aO2~RA;5KbK(Ze;EORwgS|kl^-paQt+lr!jusunEzyjzoUx42t%{;7 z)9LcWk(PPM$$?4OSy1grw{_>^iQp=n*%X1<=6;q+C1Af$(OP8}Nmol4%SB}YWg1wA zIr<6flZrg798`dOELo(A8qKB!RKh(l3!|>jjd5asXwgd& z$as8Uw4_1!$-QIWwHlM2Z(Gu^(DUOaf3i|{nhTD6sfE%{>~b}M33<_T8Ka_W1<{c zl-V<7lvoE4kDWg1btS)5-@!{nXeQT_^&9C)foV6-KXkFjDQ2P^&n$1qbb zqCjRw_Nw&HYkR7HKt{T%G}^$fG^*dVaz|Cr(O%|2Pt27fL(alm+~s?lt>WgCqs{Fc zg#?8?qm2i5>fY~Z2?dzXW_iaU6fc6J>fEYqro0y8C&LN`DTka_RT`((rz_~6lF??f zE8shcN{+QNxPISVs#RKSi^U1L4o6t%P#ro}iVGD7!&lYB(rcxY)vpY7qy)f3?Q(|#!#;6UQ*#-`u*O3NvSO7P+T(Yg~HZ-88~IUor6{sEEFgjNh< zo4bhJvjf#{qy&)Un$tyv&Ka2Vjiyqk1F7J=_((XhI;#5t- zxE1Fj14RGgftzeG7hA#{a+30^x^|S}y&xg9p7-m=yPETTMGDDCzPwQ!;XoIdN2y!@ zzKo|}tI}YNojuLb`o^EvXG(|gxp&g;eN2{fe;%4E^tc%tnJTxeuCWQCrz>%w2T_!% ze`N%JskyB+DB{>?7jyLKywP=PT~w)M*tEn_N617;bhW^!;`enz7Jh7x(&ZwQ7+W&< z9t?iMo%HiZrC+d>XePFfmm{O>IVfA*H%Ofi>(5M_cuX9~eRDbu#v{I(Gv+BodorC) zH!sJJjXm|cGAW8%q_hhk{&hOuDDT!9!7!47m!TW034|J_`lY^zQVp!y36Qy>xg5*M zT?gysxF&dKMErJ~Ze~|cU$|n0Xs!PYRk{x$>X})BZ?_<0ef}q=k9~OSJN^&oj85f5)*|M3_=l#@bABF*kJ4s+0eVMp)56I^`Xa|!|%k2gQT9O;R>Fu!@) zkvJ13>FoDNM<{4plfmeq0edYPOx5 z^Y;15ujm+8S{$@M#=sI|IWpF4R)P`8V4yH&1#Vu32s%9-U-X-k1%eC^`o6}hxRf1c zcMHoPXsbqWyu8H-dF7txAW_{2_q8@Nrtn)Xs7x-c6lHOLlkM_@za8`fvgAWPWDB{lzxGN(MzkE zPkt5Fx?k=6>}m$GP2+i-4`ll{UwZ~Fsj<|tdnbSN%eJxeaqV_Zm$G)d!XZpdcyYfm z=DgIxBAW;hGq7vud?JMakQ3=N_~qUGIzdxmME!#gTkLANe&4t8^Il2(W#lCjV^F_%fWqWr!MTOZR?N3+vmR3vSCHl=LBO7o2n48n&x;<#q#ID}8 zYZ}CK6V@|pX#@&r<1l-JF=K{K@fhT&OT zRj|%ioh+|UcHO6aHMp5Ya;u9H0U(ds$O+Du;$^Nkq(D%#`EC42`_QPq#HNs*!UK5O_`*9q0~hQYMdyCklA=`G8J zFc6xPa$-3bieeb-6^16~(?O2gBxhS#rT0e|+h8i17ZzbJ6;nj|S(Xr7WTm1kCXA#y zwO-+)#diZXrhW{K7BOt<*YHD@vyyA#9h=_f&^OjqS_2qZ9re2{cIgNQt*-*F%>y0| z)WYr$yVLZ2{CL&kUZ45yF-OsKMM9kI#t$hYNp^TA$|k%MvEu1jgSSzo>eJ^vS$4dL zIp)C;&4*G267VWY$(z!|ntF<~1gfkP{`&<@*IvBAqMF{1OT}Q~T&uFtgY9vR!{$2; zQWj!vU34k3YpUGV_I8^K>)5pW^;fbP9n&@zk=*rm+Q zPd9ROomPUXG7*(z{;tA+$<4a4wFKe!+jJDzd5tkXxMM3l4~XJY8zyhw{k>}-ulG}J z1%=TCLsDJ#o&PweaY%kL12dCOiWgshj%PoLZ@X>mcLof^`&{jgoIVaVheBH`z!LcP z-quQ>m1=7Dr-yR}sgz;BT;tVM;i1t)f}>m!&n8k+UNA-^KX1Cf^WlFTVjoog_cK6G z`%x!2#VqY%(Tt8g#PEQeeIKAii@+tCdou294?EdlF zrcg=dNmr1C@z{-y*7>(HQq!3xgk7h{SLDX@w{M|sbu|{E%o*0K6)`_5DgpsjvG2bJ z8DI+huRqu}hZmTpNtO%&CLl5H4bJ)!R6DfZ7UowCWyt532)lxaeYe}?d|YM-33HR& z$_=6iv|lLtlLqO#*UzgYQ(dfWa(e(TYT^zw#o2)KGHomdjg&f!ALX*-JaL5Cb-JN` zghC~bIg2Q0{8FHRRQL-qyF$XZaJNIjV_H}2*J8N_6G-lfrI+^T=;$~qKHk)Y@86)v zqEdNOV7MIUWhEkSr;m_R1ugw(>}T641GZEdU<2^?BPHN;lYi_f#f?L0k{Y-gDw>)g zzhyED;~Pr~d{EkK-v1K#(UHF0Vv^#7bJl#kQ?M{qemB+9jPXRIe~{vpl)b+1Yjwkg zg>iQ-c1p|xH}5tU6!gms$Cn7-jhtr}y_Pq7ytwh^)WbG)^}RFvUKNC$BfCb2MHKi&ACMHk0SUp-no4v1j=&9lj^;xv`fOSI;Cov~ho$f}Phz4?flpTk3dyBo zG^MWARzr5Z0P*WFZK8I5Z@=zHC>`FA!dr7k^t6IuW~tSW$`;=WCO|_mndLLAGz4rH z6Wn6F)p0i;pFxxT>-q;Mu#C2g*h*L4;`HT53pW^0VdEw%yEDHvNT4Z1y4BP~*NW*x zx=?sGPRa=)fCyLo9O$K`YicCTAU!l-WjykPt)N-TKu}VmFsx;L%m=MjOH7v4L7vpOC zJm;~)56qe8VD=M}16xpBd*5siAK8La>*VS*+Kxo1+1{%z;3oYXtjq7x{L!Yc4s=zj zyd=?7vnJskg+n#;IW{8=J-@+BA@H1FOS+hK;`e03wd|lDIYZM$2@`r<9N4|CF{X6z zoj@O4>-u{0yJ!>r0^acT5X$4HQ02eM_w1SPT36y^Bv9e{m1YDcB1Lv6xr;dsjPyPt zl#!LEu`6&j;atRrZ{|l|VkK&_tDe-5vnz^R3PgBnY#PP(l?jYFGrZmA-sZZzxYaCp z$*gF$R3^)@uj$d6I#h(%KBo2R3P zIHRT{Sw|5QSG3P6sp72>temvK+by~vW_N9nF-1leh^Nj=UI}HsI>e04Iufzyh05`jw2*<#(1RB^2r4zW6{`^iu`A(qqWLZ(Y%i$+UpMveDk(pLTck zMJfB9z<>x2gW-v;F`H~%tA5w%0(l^U!=qHoqKY-0Pg~?erG&eN+Bhsh3_AaN&|R#T zY*0d{V28BR3cqJl>CevBF?HCI0D{Ra2AsT#VR10PG`ShY(RaG=X8RcyJZcwl%_HLs(bML@JOq!rNYszIpEX&31DZEB9`ZD8=6OrsSsiiO zTM_a9Nsu<9Wsfc0^TWbN>+LS|L>mQRk$ghWkq)U0RtWxwmwyoMzdgZ!f9sej1g9{yZ8U52+N zK$8>sipl$g8w<;}m|S{}oo}=`O8@p@VSt>~GCnFOs>4bF+$Xm2Npb|Id>eC?f4X9k z+MuAO>U!W-NUf$y$ zRR-v=I!OCo`b}%+$JYmuZ5g|(fjw`6bI4DTmx6_C%6k`z)Su0akGkHaw|)wnKqFjQ zciMPmd|3dULRxUR`iqR+Xung`PdS8FX=51c%Qxx7*szcD7r%C=mDJ&AaZM3r25uxs z9BE(uXtGzz`;THFKEVoP``UNb);G4s3oR{F0UMask9><8EO%f|ZOvZWPkS?xH#R zWJ6nF7tEFI@u^P%Rf`M(&qe-dYmjgqfXWp$hFT2(9n%H3B z3FvSwYc|Ns>!T5ARre>hwBjZ);>>0+&Nd&xCV}clULTOgG(iY*j6lVS1o)Gl9}}+> z#-_8cf@0HMFG@;^okhNE=9^{*(;7HzVQhLlN$}_Yt5roJ_hVENZ&mHQc~f3iMLj)? zo#i~>fp$7=Eb)8ExeAcf-b4Xm8xwgUcw02wEMcJn{b~I|^jgxAW%$}Os zCKTRdmQQ{NUw;>u)<{*t=lJw?Lu=PEe}V1n=!tM%B5i@Ty&Sl)zVSTa(#+&L=IP*H zSHtFJ`T9{s9b(4BH>F<(h#aF5_iE#FyDOL*%x=z&n>F`@n(zI(TSmQlDZ`I7%GF!S zyES2R*|N>QxE0910AYWsCcL@B&T>{B=d7NxtS~Q^P})kDyRl04j>)r4^%9Y?tNn_% zOE_OcRjHp+V7u5r4CELD3O8&1GIW9umQ^S7!v+Eu)s~a$SeZJ$=~Ay|!)6pnaReOB zhPs>g&}*;h6$aLT8`n`rmW? z*KA6Ehzr-tu!Am3k{Eop=Psds$*O3}vt2@G0~GQc%py;H1v0mMPbLqB{!jh)G;^f600d1e}4ZRjw)pIfpYr7h}MoO-OJ=)fb5kY^gMJv?Xp zgvBFFlLU5ZNWh#Z^==d#k)KFvRDCm&7%Y(WNCgvl6#)5 z?9b;sPvAL3k4{=*aG%R3%l7}>y!-#*J&LA}#yufksuj&juK6utybW?&Q92Afo?4Z zAf7x=9M|+>S*q$U7TsLnGK!g>Uobz7Uzmhl)j$sHVhpV`s!k9Oz@mF zoC+W4$doIWz6rwLjBT(S@|b8eWUg-)^A@k80Vn5_4diU?R#D!^k87J)hrYckO2@O) zA;?}TersJ)PH$lCB)LcLx|C#=gKqu)j?}D)$|xr=ed(*guI{HdUQXG?EmS&(U&HJx zla#$#O~ftX%@av6u;x~`do|dYdP~P=IdX=(*%e(?^VK_BCW6HQ>VAm}qPl&ZMk|wg zd6D&i-j9nm{EI^GfC$MsR?TxusvHJ_gLb6Dvh8Vi5r;NS`k~RrHPlFIm#pTxM46?M zIIC#G9nzzTs6_(AYIskHHE#k+_IN@AdOjI>K=` zW!a?hPLrD@j<|_&h4q0&S}ia`ml@g^>7nYrF>An;WFHc|WdV#?CtUwU^XR}kS&ZC? zEK#?`xwyuN8r72ND0aN8Vy6~c*NE!1CLCLYpl=!PAKuJp7v5MJ)J(2re^d&F%knHM z5>{VG^DJI4&Zsg~F7-2ur|@WO)klxW^-NFNb2z&aflD5g+!L+sF3wTLP^X&@9TqQG ziC4bGY`!86opjtx`tTm_Y^|02=(Ak$08?w|`oV%%&S!Ec==-0Ufd0=jTs{n*p4FbhP<4lvHELVp`INz7X>9qxqg2)*cevam=_z@J{cPRqK}8BucVuz@^g{uXDP z<zeGgciu;oLNHzr0_| ziu*ro{RL2z{}(olgMb)>v>>5$2uL>w(#_H=-7MYG3KA}z!qUysv2-cjxrCH-cenrB z@9+Q4`^+=%4EMn7EDHnoe$F}9xk6<3FUograH`Kcm&+K*FTptDy=T#;$y?_oFzgml zit)UuBjb6lio;I#m-8oL0|jYs#`oVV(=&^X`zm^Vzj*D-k=2`{vwyZ>v?MEdSAVSy zBe)p#ohp;QHITjpH-2Y-H##}LqtRcwtCI5D&L%W>f6cQ2PzX<3eJsvTM%DWHBFkHy zS}B(S4rd;J58NN>|0$J$`-ACw@W16EaE$>V#utEvJocYt>7V%n8P!~(GN(-nd<7L5 z_4<|)0pzLbQCk~|no-R1&h6!rGFM`A;y7 z3V02Du5j+v9^vfa_VfGODL%J2rGEV0{Z{6zv~yD_VSbSpFu10YPNG8MLRzNXo#saE zS3rFpNtK)2(txjG*{eX@)*h~ z@XjvIdRo5DkrVso&qI0dR;r)WGeBEXN7R4UlVdq(eo+vgfzS{j&SA5D(HqnFgTMh6e zoaXTcUe`mz*7XVp72y21=UBQ>yUB(qY(yUirD=eFP`N$2(?@> zwA=U>gnX8dMkf zRZ2j$d70cS&bkaQ26QO{zYk~TGflvDVud2-VY<5(Avk4ufx7qx=|;)G#mRO5-d3>zi93_ z5a2ROM1Z^*gne8Uw}MB5 zhVo170f;s4XoBYYLNv zKyd^9JmXu9!&^~^JimubO5KTx!>5Pl-;e%;ek(gP(7UZ@$zzw=OJv@X)2=7ZCVUA1iC{C)+n{4berKU}4 zUO8%yIoCeQz-DZT#~#Mj+F)d!e31r&6nPAqbDZt)kuV2qVvTNn@3eCmZBu{k(DUo)IyN6#RXGA=VZfp#}Yj?p;4oyi-}%Skh02>=LSSWzxqm zT)*5=LKebK&r@yi`Jdc4o4Zs+{Y8l+A6$L$+0)`vTWLLUO#D3QI@bslorDBd}U5k1Ek1MYvD=9fi%W8G&EJPhM%Sr?@`MV5nUCBLG}x;cH^Jf527> z?tP?)yTI_Y~9>hikxEfPe@ArJn; zU-Y|X7(XVLLOKp*SW=0FSJ_A1RhP>xc$nTRmphS$I69x+<^;#nTeTKQ^;(IbcK;<} za|8N_ZYH7v3H4aTkjH?e1bJ{1FX2a2DPd)PxCu9oJ$FbO*LZfBQ;|ZddnGb|8S*{< z7$Y#N{}Bmj#@Cr_H-DKhcsyxKc*wT+?2BgF_mM#(!fo`vI;qXl_Iqk+fCJ+GE7i=7m_J`WRDHcD))}>% zoPKU}*wcGlILsD!>AVl{o?7&zz<`guB};8kEmVYCZN!U# zVNc5Fl$GG9zv0eRVRNo8?j(1&3v8ps%7W(+nPcLkM|4&P(mXa@NP1oBYn*)*rL zhf^stt_LoLipW&UN=O<-S}Jp~=y>#{$s2w`brTb*)TGc+MNbyAhoS zn+_kw$938%GNwD_K9R($cy;W|qpKLp_`X=?kK=I}=9FV@;(d63 z3?FXQG5Ld7zCB@X$%h{z+Wks4$Uxck)|4UUiB~R)#>==}&ezk^DR79lMkU~8gvvUy zeQ>zkXXM@oOy<6@e6u2!rfz0CPbc9nT8Az8RmI1~Glu5osz#ebyI+w{1eQb(T~P1x zyaQSEAh!jst0g|PxwNj+-uHZb^`k}6jEzrF?eW~MF!|qRs_SMjH{Th|?Zk-A%leyr z9c9d6#X>r}BB6m>BAEZwd%A^e9sM3#BMts<}G~Wyu4qtFO zz@WvM_W~k6$+V(aRqdJ@RnJ@>%Q?nfk91@D3FMQY=6{5q#y;v7-?9l(&hpUy;3MI4 z(btzAX&L+g^ifpe-Mq5Xzw$X1iT9giiB(_4BA1aP+ULC4r0Uu}N#%&%CeHF$asn@5 z#1r2I#zWa2+qjJHHeWTJo zmsBj1XmPY$$uVDFf0(;#@m02YGO>ysV~qh4bRNc>Y6`ybofxl4yQfW z+Y(K_w{5F|Ng)6yK~A! z*2O56ME$WbUk=w2Nk`hqZdjUyci&R@-UVD(jSU8Ow6Vo#Xbk;s@%2SSfTviOjH}*l zPqAB{@4Q<}ZR6+rF#C-N=_&@7r*aHS4OTAu3cPZM-s3*j{EN+zSrcj?9nb{miV*M< z+Ho6wDdlSULog$cdn1A1nKhZjPj|DZpruN7*2jJyo_1K=H~ThIw*cpl@r~F3EDhKg z{PQzWQv}|fS^uDG6js!*E5Po~8rLu|3UF{B{%1&p0rW}0JND_97n*UBVlHtR`76ejxeBx4p? zqXVCHpDDC+sf_kA`l_WbdeRAS%Nal~Tm&*_J3J_y00}1-gZyTFLizI$oxF3D>>QVyY01TVzr&;0^yTJVhP#U;fuWitT@u{yz#B(3k?&S8+*SUSI<%Fo09_7Wjhf z6F!tl2?aPrHA#iX&|-zA!ojYd{E7z=E(hBf+V5E`-tu}2JNyi39Nc<{{rO-k@4V@K z&QT$%Y0*qy+Y}{Gy(^r>n{&&abyX;3TF-d;OVHj>*9bMi-}jL;pV%osL8sx{f^7Z* z<}VegEyB9#dJg!9XY}c}=x#1SCnAvfW~8(Ki0B&f?c{gQn1wvG&cH(&7jcmf&qx{e z^p5x9Z8%$7PT!wwB_E1)kw;Hg(*j10LwptVxC6pV2PRp zizM&is_n~(B(6R)&8NC=>7>_YD~l|q*x6?6>eU<0ZBp9LX;esfUGxT@@x9_MCGGF} z5_?Wv^r{F*B@$XA>e=u#`<*Y(p&chg^}J{tVIb4R8K?*`sqfqEZFPztIvYCen@{hz zbOd?Nc&^^th`t(ktEJPk;Ob);9j(N1j_8$)oT&8NzdbUe0#7)9d=)Sy=S1gomL{8q{|!Bz8Vd1n-n}y<;}4Y zBAFXx=dBE1(|1v{CNxViFTIBZT6r9zsq#3laXj# z_5+UIc0E_#O$LUuf{AB6zvL9;#7z{Wo%_qM31k|kD}r0_l29D{oxgs}+b`$rltuDz z$Cj~XNJhU|V1Eg-yA(6`4B0heGk^agc;wlEVWLAl86iHn+t#-5w+v~VvonkE^}Z}R zo7^_bmudIA1CxOTWkVjjFJ@8aOKdv|VKGP2cAf?34^%KulqS`{Q=lDKz0*{FQdt}*C45B)_ z8DGPc+zQ8Xck1qy|3sw9L^$L&Nk!z=E4Oh6V&(ktLy2SR5BppYr4NX0gPkpTtmNTB z{aSO9h|+)<-J&(Z{>qc-vWRRz^Dd8X+t$4IK#^mjD(AzHL+BQ$J(ym~KKY?WwKY#= z>VuyQVJV0c4q4mTderT6-tH)AZRx+trS*n>l#zP-7lnop@y~Aug~{f>5DGw2B@RHM zDEa^M)BYbLl#DU&38=={*q@sKYWb2Jr{D{DqggO1&pd?8qU)9?vP-R*_7!@fI)jyO zj1bv+)^NKbzV~L6A49k4G@$Wi-7tAQPtjf4rr#$b@@Cr>qAYmEm5KMfP4`oFl%7=| zJ{Oxnr#YyHl1b}hx9+pIGbRrexx381gMiQCIa%V@$leD;c=V7437I0O2fMXefpqKP z=>sj*)u5~PzJBee+b#B9f;4++>Dp5|yuRjeHfj0zKRj%ig$f=g)q%~k`~3TY^($Id zS*y@2PeFYtDri>P&8w2hJ7ly+UL z%s{co*F7min_VrkG^V+U8m570!ZN}%PIA1Ni~LYZ7psC5*&SLt-)?4=qJpSrcITUo zT5&RPF>h>ok4(Y_BzKJH(d-YRpJeCl!C4^$AQKxL4CiL#4UrfW#4b1{0)Z2EF3d^Qo_QOz-F`gvWrbe`0w8Mh2)v9bPoQ7pCri${8Aq&X{3BL|e~O@D;BYfZVM*6GsL zf0cA%)0or+N}=c^hW1k@wnKR`Yhrp*6hwNFG({8lREa0pW8t7!WRT zi}WODL2V4RpZz&{Gi+l!uFoI#$uaefNiA(~y-ddXExs-0je8;A7*y5hRWh8qF2ncc zgB%kYgDFWbkS2BTQhoY4F>#0_g@q~nwPV9@+oKG5L07y)*_jHnxSF+$$B0cF~GdjC78a^C`t0x9X< z@PFd$|Hi3*Bh`OR8gOg@eVr8Azmp6N$Y+Y!YyNc^^q4sbWR~|El5S|X_!=85;nz}~ z-)G*k$)AX^h1oeF{ZACT6;F##qM@ItYITLXX^1{_L)*;qI6$njIcApy$=5sDLmH#Z zu@EVCb`f|1bNhz6q6TrEDu<7{Z#m=!3NyXT)3YETsr}HceP3>aWr!qB$uI@JvE*a7 z67VXdE?^mkKM_Qa0^%pKKN*G z&dI#JnF9YIEG7QHSjOZAa68^FL|YPp6okSWISwgHCcXencF(I@HCdpZk)bgDlS2JB zXxcbm z7|%1gI`&UkBQ-Lm<3DhfwBfCx-!||2i((b4nKw?0$+&!HTH*&g{Z*fr9k2*#wlA;z z+4wPFPCa^}LOH8337$Y(*$^$1OS7XIY=Xp#hU~s=Lo|vtX9(Yl4M22RO=IrIbFCV? zc_@b<6=z*r)JXR1*U-}yl5q=TF=o?#nmEO3CI+Fz5heRkGwZKt(q@bIkl$2^9NgxK zXBY!90WD%3;c|!k%23a$DBq#25f8D0Gi7kfWT58ESkY?5f(;)NhYnIU)0S9FS7c_6qVzDI$%xoSuIeC9_bt3JW#1k%-cUicGnjse?z-If3VG-y!Ts;BnsqNkCztPNS||D z+xzukTTyx4Y~qrkT$5J~Ux5*9GTjCR-S66C)`i%wG9AXt#A1=L%mq}pQGTxaUKcn0 zCiH;7i#_won`)q{)+ne(CTZe=$9YaDxc5;wFM15*_UB_1Cm$j)nvGlQGJkpW)ebna zCxr$mkV!F-d$$3>Ca=VQEk|`3L&JF*$H`)nMpo6ta{@d5FIWyZd-bqU==g=AB5x+G z3!SPM)1q?4a&EwxIG{AISvSFTId!f(iZP&qqfh zG%|i3Y!OW>xLZzFFNix@eJ+2wp!EKL@^Zhpc)n^Lwk$L;*~l3fn_UF?L2?lKoRH<} z?RSr!pjoTzijQc@bDufst&!)VO=N0Su@_W)T^^J{a&<58>#QB zzMf%Sk@kp&l*p`x+2`oIom}^EcYb$v_>9d-`i(Uyc}8MNF=R%syFxZ!LJ6v=?az* zX#(o8x{fGm+@y*Xk}WGqOAD`eW-&7AHWT~ZUH43FKH z0}B-nfhh@j^8&4wd$J+jZo~9+skVnnmpXNQfIA6QM|+vEM!Z?m9q zX|T-=24v?z^oTedf6mLqChoZ&_uWXbd^CKU@n!6oHIVl@Bh84)nBV>M3jo{x|IZ=D zEL`{_Ys58Z=CM8gFUrTb7Ap%Mm2T(Em*9Yz-GFUeLy@{LJFkn(7VY2SQKfeIzcnme z;8F2|C1A$i;x>tjZXxWD`i=3830BU`L?sT*Bes}sS;T;-Cf`8GqvliQmmtR1aNs%P z96q4e)~vE4r(OlgekHP<2TC9ARP|zECYO1qUORfCGPRU4wKP&RA37C3+-YS8&B2Td zv^kQDFN@|D%<>`0e#N{Wxy1E*xaWL9(dlSDYp~H&DYWFYXv?jOvYQP_Ql3O-|J=74 z%|s4n#J0N2cPIEF8V*EHFGlqig4rvg?6W+K3{mcPe;p%F?aQTnOidN!aSZpX;xpft9pPFtZ=A(QB&f_4+WwakozznOY8m zWsK;bNLC>cP5DEay$Ej{=2DNd{wo@WKK?E-Gx?#UNc5*`t(CTFW%QD3 z4HrY3njW(_b;QL_ZWp>594`eoy1G_$DDjGL-A*E@QI|(aiKDIa-JlJTiU)u^%WobY z0cYevs0NTGau&a4}tSyXv%CkrWI5IUh9Z^eWkv&A&6}V*lak(Cwe{C|%9! z9v;||7*Fv@^Bl=a|1(J>DxihG7{m}$Pw3G>57zZyx|?qhE0b_r{D$zv2|A`O{zIc! zJXU2rbB(Fekz#e}X^_9f)Evw@ENXJvY`C$4KL6#68i-Bwg~j}?`>g7`?JE17nhm}1 z5j_hzK48ZGxmv81^f6Jt$G#eoR63v~dn)p*^ai0|)3%CGdK{JC9v|<~tQP6P7Kt^N z5Zw{$mWr3qG)M4zxxM+)B?+?gCMT;mskqMUyaxM}^sN~!JL!tifa6rm%o2SP78h{j zfW(|OMkq(!(qi6~Kjy}C&4vP8`GVW}_oV-ar2-O-$?P;YcGTHFvBQq+-+fV6vj{+e zudN&~=gekh?y;P2p=U44G!K&zb<+*!cxD%Sx+65nF-iN+A6Muz5k3y4$Eu6gquQ<4 za=E)|UbmEmv>(rJy2c+dD|T%or~FAjrz$PCPbmk*I(~~!C4AdcamwN@L z`dU<#k(uuEo{CVIqfGMM4EOZ6*r3gx>fUS0_I#FWhF_GlpV(qOVj9;UV1}4eZwl8L zHIQw@+S3+PQ4A5Q&~p*>72;a^ACrfyM`q+-dItjrV{D;FH+vO1KUi>ZcjHCu2Qf6z z3a#yn^NkwD@sDrMHxr=oCjZ90|F28Xvs+lUv&|+!U6D+(!`I!YOTDYuXKhcjsBlOr z+!!FSLo$^^2~Sv}IG>$!-^A$Dh!;uzMOjk6R-cv>>xKdo1LNKV)bxeAoYa$lttrrS zV!X@En{+&X!}8M0GU{~olGl=VJvh|!IDP>sc-Tr=Z)0gLhrLT;lgCWtq9TP>f2~DC zS|G!g4S*7=YS(K^P}P^Oa^Y)$$#emx_R7N(;=sAf%lQNPRf)=AsJ(_NG#S53lIGk;g7V(Gc3bgu4^+g{#@6K2-~oi; zX%Q8)2yOOKeXrYy@J3S8F8RI&bav9UDgm)bp%iv+X^*EeCcj>vmr)QXJkBCPvo*aa z7@H}o^+`c-pO4f(J4%r~HaixQ3TNH0ZBd}me$Ci=O?iR*%}u6+XoAcy>Cn3L({0D; z^Cje=$o;%Y4gSgBpQn*;MtnOeQ5C%2Sl1XnqIE|LP8~8vnOmp)66aq#gz=NI7G4Ia z-V^*b36PZJ2pvC7zMde0r5R#W^eC%k!`W`-e zwz-%W2DQ?XKGwfM!G#$8SvE2IS#&DZ33Sx)Z-2}D(0T&e&nij0Zvqta_&)~oT^gZn zn(T=OT3v1@B(*F_d1|MAMR#H%ERvrxrkM4%YDw7B28bIE#A`3hx@-7%6yyFlLNG^6 z-#9`-WS#Y@zi{dPaAih=Rma=qay-K}Tlzx_aq`MT*1oq;MGEqT9=h)g#Sur%Qe}L) zH|!|XIv@Z-`!-WD@eLvMqEZNZ8Y<=Ri}H-lUze9z3jTjltdm_4e%%1rJRh@@L0*N| zgHOAUGS~+s5H^KWdBNjQKY#%*KNZ#s9{pmn)*-@OZI6(!@wTZ0Rxx)RT8lMv1CrAn zop{Jq4OP53%RXm-V9|ao7Ms|MqAPS*iiRV6QU&|VHW6>sbu;nT_3l|wxIUd|K?Xn!`uBP1Y&}>YidgAoL z&e4QY84rPCRo7h;!QzZ08GKZv3&wg9L=mGY*CJia zL1+}N>BC?4n{pzrPSWYKU3`j>=uge{iPD3o2C@FN;k=DL^dIAmFIS6NPGtej(j817 ztL`Hl=~GkWC!<`4&a~uLoUlY~XZl(q8f1&fz-?KeX>LQt#Dp(g8L{j>l&vgzyPRCU#D{cB}DMQoZPGn9M6(byHbFdDLx?eNiJ zTYOuJCq{wkRPBohAx`4N{IvsmZ!&50opkNHN#k!O)SAWXev0{eOp7$}wG$ODpK8pB zjIfxBP|BbU$KYlS+utyzI{^*2z5K}RtBHuHpHnWhGPMZ^NvECIE`HXFVVzgSI4Gq) z(fCk*z7aQB;jQ4spoE+nD;ijVZn(gU;ZbD69#_szrotY}#K zG{Wxu?zM#)7ORjIMj-JE|H7|Q`N$ujG!KD!+m|I37auq%l8?m%r|ORv%7R=4vmSS` zd(X;K$~H+g@fi1Yk{&Rxk?al!j8VGtHl&m~b7Fw^1n7Ij38#;>dZ^?iYmXrhtta5H z7pwCEO;E-tX!5GoN-i$D95Lc3(ayR*@NCXpO!4ZA$wA5rmJIR) z*NM?8#9}zUI=C^@WBOv|j=-*lCgQO+5$^j&xGlc;@b#)t*x6r{Xg0A#uae(@rRe*D z#r=*7n@8#ygegW+zAHZNGHzo4O}7{%vn%Y9Ao>@*&Xh0ist67Jnqn;Qn@#-LGMZ>S zedbaaLLvT|x|Koky1V!zk-HG1LqY&~K(pgutO4#l?IX&G={Mv-PJMd&)~3j}fS)}w zr@2X~zeDTFRIZ_qp-?kI?MlOYx(S|V!#$zJ+z|im1-ehy7mx0JkN_-SrNSrUbG=@# zA6Y#6|Dv=hV>3wG4o$os08L!-?=st_DugpBYHOW|zdc|k5-a<3q^o-&s1g5rfiWit zNHLVx`~R%poHB1$Gw7Uw>?FmU^x4wy-#VKzJ2J}4#?E#$i@0xs2f#VSAtvMSRV@l> zDWPoL_;I!G2^VX7D*y04?tfWUT>wcLg(}f{hb+Q_pgk4KA?m4aj`sXDpPFt*QhLXX zU@N$5*X$H=U6vdB16Nr7Cu8q|lw*U4lh7s6Kzy_Cdd6cxTof{pi8V6$Ug&*{p8QoE z{4zedqis3M{r%7TLef>l#UT6uAq2Lu^rCCgOiUR84;bRPE?BYNXp0m=Tz2C6`oiCH zvi;6VbcthcWvhUTmJ|)Tjo)~6F8t8PUX%Yk6BC1G|Mq{Zf3Xm$fzg1EkKUPkgl1 zDd2eTV5?noQYbd%uX>rDM^ug)40SD=LFEi$Sr)bqk7`-*$84Mz?QG1RE_XBfivr%P?ftS(!ZgY$J*r#u&wub-2f zmEhadyTqwU&jj0Ry%DJ89}ZfqI?d8O(&CHAApPd2Glh%4)Y9~&8ilnL`V3Da!VzFm zumYy2w#n|;DEyUivgg~z_}R(Qcj5T(k%hqF2-iLRsQkRvvW8~e2@NmaK2hVoE(cOf zq*WTs-kq9LSz6n7Ty6;(ufBT(H9CDiA272D1fJs>DzZ$s0lUq<8vDN}M&W-^vP8T- z{HxLJJ&c6Qzz0(b3gI#$OZ`ihsTouiam=)0gB8IlOeag6L7dp@@;2dwLJ_V`>y}E< zN6ehUdNx-k^VCOk*4~R;LjapF2Qdn1ruad@e6RHrCy|9ws>sEHH0~#!3egW2iA}^& zIgvQg&>zF%^?0zjI}_32rY1UU?ZQu;E*4v%1hP?2O%~dDi{5lFn$E#u9l8~84;UwU&oN>W-z`EEi<(FR!?pMk=00TGu5( zfd{DBrLK5-cvnZAUAptk;$buLknx3qS4q_Rlh@KCr;Ca+d@jUhlMLrX5}SJhFMkrw zH`maRQ(~xkGPJtZyH0+?*L!Cu(OV`BhDjhNAlPh7+zS(yF|cQ5Qar*sVO?@Wi3g0`vyTOap3s-1Li2gyPr}uq;1ui=t=lx=v z^`7f}^8441L{&B`?nW&fT3JJjI%QG3Lb`c05P$j#@f>|cpGTs_U%B9q0{XG)TyyhS zNm3=3S_}O%j%JB<$JzXx4)?Z}^PQKKxEIDS-^aIzP$96N*}`?%h@?EpN$#TLy%mjYuJEr z*i_nMVxseEc<5gg4QS9Ak2EXJtv>Pd*J4YMR@~4m^}Cz(m|JOHXzGZR|E`E)b0lcj z^{ziTXkJ)@15gatADYBI{N$OQw;q$opBTcR>U~#+p~Cj+u#|xRjfQXUtXltKP#0aE z+5q_4<5ua$eiP>zcDk{j}Rm%l`IK$8q<&Vl$)EOb-)^kpJS=5f}Vz%r=jt0rxF zh|}!t3rnB4mXqwxzGA;&ah}zjP2LbAwf3j*QTK=>Q`?tn$VBnyGdrNg$al*7^YJ>g z6yX~W%Ce>Nk+M~T{qYy&HbGSd{sPkE3rHx#G*p|4nU0IxJ?{)WotqOL)fYc> zLFj-9z)NZ{c{Xh7hopSJLRs4CmW;)wO$L_C&V2V4sr;{oz@(}ev7K^avW@Vq|IV~3B6P?L$u8d%S$iydzVtm!zn$#G92*0Q#T)V445HL7e-VbPuVO?W&^ zdPevlL!_0@R>zP^BzM;@NBL1;-BrPGVNRxVlmvS^+7`RhBe;M_sB^%eU-g`u(XW=Fb!O4Ry zJ#1k&QdU9ZkPtKu7%Bo;H2=IK|07uZ@ALn=G{AgL;Z2}I5a0!2W-4GIMlW7GQN|Qu zE<=eOTP;aerfXtu8dxjGoz5eOT8_>a%GT>j06oPXgX$Rr<@3o4R3I3V4i=pCvlyE061h$#kh>h1ult=ln4Jr?e)TbNr{g zLE~Iwo(w!etD#{!z0BgQ%|}!zi47lXJ+c9ll%yH53}Jhr4T(_j*N-Hb?vqVf7bJIvT)&=eF0(J zw_rY&5rcV4z3;{A89l`CB+ekHi^j<9zu>8P8jD4PN^)98M1tH*VzrCis=FoHU#Q-v zQnt06?6`j0bSPS85F}&k&}bSS?OiC^YT_7E2AFPEGb3pzhA4DLOb|;oM0n2#c&;uz`>rX|uGi6YLU?Ay$?>Qm&~5i3VWV+z7_Kw_POD z0KRQ#+@sYa&9LZ8miYYPEkF0>s*L~4kpM(6->H%S;yvE2 z)XKe`_mF@Bh^8#;!O&n->J-GgKtp;l%n=xE}Nh2xQDSm$%kuD)N1GCuB6hdfn( zgIGF%2G<7nQMJSYu5s8y7Gj1n>$Z}U0{NufyCIEhuz{8ajxFep-V~i;1d#}Gv#R1e z)Jgu8Oe=4)1mrd?{a}sPYqg5r!xcFr?o%*`CloXNe!5#_k>^{y(9VF!3ry}rE=lnn zwZym6T6~K3gm_Z#fN=nd>GW?o@_${ZzxwiVgrF9LhxFgkTw}b-7TJ45-;N*G!O}W7 z{;f`rjHVlR4Z3jkiiXQl2=brr(0bP;o+Pb&fYrg`;LH6E9`V=Y(Hr1g(6#3pKJhLi zb)$orcL4*{FOlC2A1+JM+keO(xK%{xO_dX<*44L) z-^&ZvJ@E5r7qU|eaByGx_MP>ceKs<$YH<)ODaYn*trMfXrs*oG+9M9f3XMK55=Yza za(lj zCZg|f$UDxf)jX`%PT~bm#sJ36Kk^mk)E`2CUanL37o`F~n8qkuV#qEsp_!-e|80i; z>)K^;3A3vDUsLq z4Ip%;vT7Ku7i{(R!@!Rvc2D%)~@gVHXN6ceK+SAy6@&jLgG-ffa9{j z{77V^eS+%F(Q!w!tx_@ZFCzti)T7sQ#m9VF4&2;xqnZ4e$O$LPQ(_vkKht1e*!}Y4 z)njcu&!DW!u-eWxArnlza=d=nMmZS9&2 z)wd6}iGdVsWcS$N0u#jDhZS)rJwvL2)t3FWC|@pKqm!&Xh5_5%{lhXaJu ze(W8AcFV(RLmgMGd^}f6tX-%LdY394?sc8%B^uPWSyG~e!hdwB_0#)d01(&Uj#s?h z$J>km<_s?dGi7-v>V|3hNM;jKMZ%`C89cg(hM1~D}Z-EHe(<-su()t zjE(T~9X5St;B;q8f(f^{0w_Li;dF=9-JjUlhZFtJx{_y0Ub?-tW}1PuCu6xNZ%m2C`)<$6BU?wRA7-Y6(T8keCu@2z zhZ4uaUT#3K_rA+cM|~5;CvEGr9By%W9{gB61Jkl{HV*~VVxaTL{;S=FRwIb=Ua^1_=(EecH&-qo{R*|m0(0&2*Uf)!qd~^oqEjE>Cu@lDyvFnN(D`4jYMrKndRn!9b7`;*-5$`<{`K9&?b{*`cUt0fk>(;<`n zX?2+L+P*=KRUb&R|EL?l0a7T?UCNGI(%Dd7;zjpQ+|szVge@r@%b)O>hCYRB97Y(D z&2{aTBd8naSg8Ft#!hRi3ibrQetVM@55IT=J|ilq=jGGEQdIVl11wAR;mO5uQ4uS7 z2#e7H$kT`@4ukbuiOL0(x98N=R^y{4pdDCGlYnU+$h59tVo~k}l8~fanExp_a!PVQ zL!$zX9TJgEQT*PZD<_bN3H$$S=s*0Q6|uk2@#ki$l+gF=m9OvvheJC{0zLgf^q}h; zzk=d19BdvI2m|4{l~10p)(K8}Et?+8cBoCr8>%BLtBdl1$<@M;i{!S^vYky7Y24J; zDPplSr$npZXm5i~qT&kHlwM(MCZW0B2DxP;v}+g_Z@eqHve$6hq9}ZXC27hZcaJzf zQ>iKB1XIdnqy7TONKl}~p@m9IlZSi?WGR_wqy3yJ|Muk(qDw4r*gOME1!fh`)@H^`TgnwP8n_=n2*|n^ayPe_&LCf}kXhl6ng=_p2-pCrfhM zG;Ml{>7dVLaWIpDQ=%k7x}dP^J>(;G!(OYhZKFp=>H4iE_yN?j5by3n7ZWhZ7!{z` z2Z&&TM-LH$VsyBqnuGrIO>q$~B51F8!HSxc9&o3stgXRCXrc3JE1%LQ;>=eGQQxpo zbz`qxHIYU`ttpLyv?3zS3}=}yhRY*f9}i<{}F|3S+= zT)%7E6w1=6JOW)6V!i0dyC_@XG+02hk7cR<;TK# z)U7J{EX-b-sR!8ZW8bnER+n5Z{`FSwT-;qqrX zF-Cwhva{R7ese^Z8k1CW@xC$VP4Z^=NSz7u+S@u+zDXJ$ikUFCJNnM0Z0Gw|meTjm z-6l()8wOA3ZHBfGKX7EWx#aqBKG;)`EqKA}j7tNy(mv6g?7{46=$af;En58QW3v2- zx+xU&S_9Yw)=Ad}>kKk)arJ5(m;GE-?R&-(Q?Co0^8TXy(Vx^GUHOQl$y>Ku{9Gbm zi~cf5D8PeobZ?^SP}s(uQOq16t6no8=Qs)d)mZBzH-}#D#UP9?~D=T+$`i9 z$>Buf8GmUnk+wDqTVLfPS)%B1l=!~0re$?~wNXYFt~2pp6y#^X*V`%JQo>Dk;Q}0#=RSTrn6d6c&EdQqcJQz3s{I+tlktOf@>##)GzbNMB{^)CB zVdZT*?4*NBl6T@;ZQT63Z22FLLV8HLdM7KIl3=&5hdWLeYwvI8*OiY`fqX z8_xa0kP?L{0)@5xkMLNVtqG=W{3ei*>|ZuFXPfE>+3o1gnN}XK>MSOSqkC2Ug?CKw zhF?$9!J1YZW}aLsTrcX0oQx9}NkhH1f|a`faZb{(2NE8;rHY~tc%RAsSKH>Op}lkx zStTc9>hEuo9hBhEJSy`hEV|1fMvM*uQm#AIaO{_;$ui#Y#Fd zHvvilWmQpv-Sz&*DDnTEeIoz&s6*RMV9n3X=rdYrhaT@q(MuUkg6nax^=`j~duMPZ z?o?tP;p_OQm`D138W=TLh}COIy2|*Ac|vGhbIYc~ZSaUp*%l8g07xgTi_-eF2mMnZ zI2Y}L=ZD^A6hN;%;IL{ex+|U4L-0CgcKi>ad>28UOx5yc5WIU6h`np;kg=FNEwFW# zCK`&Nv-EuObvAV0iRlQ4+zbDRO;Cf*64MU3u?Bc5WnYU3y?`>f1 z5q4jbx_hf}j$n5-VaF|=-1a=BN6B1+zJtSs>#o+KQR>JyL1A`X{qoD!-etRkkahmW zQ2z3j{}NNeiL!arlgs$asx#2ODOKGSU9FKlJy8xUbXWneaBQ*93lL~|EH4-cyKo!* zSC#!A6C!p(UFZKEIDD+H4Bo}94W8c=VTj2PmCW5nxOHm}SKfQp|Ct~bjY}gYud6XO zNYH-8YO{!r;m*Ib0^940ii;B>fq6`4wO|Gn{EuL9mF78)Gfuy0W)mP;H->Pc;fBYS z4(riVO3@^*RiC}VwU_}DF=K>QlZ1Y)9;OLIU?#E!5m4WJ|_jp%*n~QV~`J#qC{W3vp zO$W{u1%3PD$ZGz6Y=MKD9@Ts#W|9oj7aPHXC0jH^`?&|k@TjgmR`fBsjL`)deH%M1 zX@%vYgM`++Iihh2Q^*;*tpC*Bo=DO* z-Of+D{&Un$uytvl3ST~^zc8CKaJ#BLMC`lDxL>&BkLd9_WvPPKftm0*t6zY2M7SF*y{sjzssz$SychH=$LnmRVolP-p}Ilya{Pa-QQn zp(>hYJ`v>RE3KA-l8xnZm-)pi;u=3$)I&rI#ci!YM4n&!kdFJ*&}JyO9w^y$-*P!+ zeJ-n>cG27K+j7% z-ZhTbLQmt5Xg+cA_v34ET?#(%q*!W^yuQss%azMD*{!Tgu(!Zi^FN9{vk}+gSA^I*1n2<`2v#fB7Z@ zeSGug6FoV=3o~Z<`G>AJ*6309@{5|ZI`~}!9+UBim>!tH(i2<&`<+07=t@Bk>>O^R z!Lt^YK;8Fz-+JROa_A91$i#FTa*Txjco5{7!*=hCe-eyNy~(Wp8m#7@aeWQprra$L3+vFG{VwxxPb%^eERZT!t89gmYqLt z$WyFA9^BFHtWKg}P}(2g6tldR7f-U;Vj%9bk#lfesH;yYVWPuI+`ber_-s)iW4jf< z9~s1f*Gx_U`I$EhDaYbg=S!Ipx2kMC?Hh4UN~nHgRZa}Du+2SKVDu{ovFrUpwFwb@ zpEbwq$2451LvmsCwdlm z7S{(|>-$p6p?v4ek+t>402D@D8*t{LZh+9ho%OF{JScd*IcuqAP3>2Ylnb^nxR7v? zyn8(<|0YS&HP)CO$)lPmoLp=m%57!0HOec1!;5;>x^Por)nr!V7Tq~z z2yoYpD8ft!%mo?jHZ%iOLlLOn7i1Dz!wV$laxEq8;kvF9Nc%KC`Er6#i@OJ<-cMdP zDzyV?HI#-$N&AabwW*>SWN^9v1E+z1l1u98$2f-B)FZz83W^1#IUS2HEJLNjy_4y}MZCT=;w~(P%s201p ziqw|bOf(OUj4X}nbn8*w=(lDmVVOMUr1VX-;go#qbbgUxFEyJW4Gp2hKws^NBzf{o z23omg&Ult)Y+z7>Tzg6Ff8TM{+Dy@|9hqRHv*1!AZTj*s zz@V3SyN%reT7GU(mmq5^5F^FA@hy?+y4X5LpYt3uI8FaIZ+c<#;+P*}D(Re2_sSIlqp|GK?Pj=BKkw##D?zNCi4JO-Z zu$I0ObVRtMD&!EKWKuQ}5nWPI-Sp9l-_ACfWUm$F33P-n&IvG89Ol3Mlgs2FF;!d5 zN**lx4eqaTw05rGr1}vAo2C96#v=DQH!d7b8)gv1sO+F7_e0m&_5Zw_p7lW5)?`I! z+lZyh(DaV~Zla%br7~yt%4$nw@$nAk^d$DWZmN6HL~tt}DA&cA+BL>HpcSHt|7>F; z4fA%0z0ub#5YFfHHcD{3pZa|-NK>w@MP&AW2Ff2qFzq0Yz`e{}_gPusH&*A7Mx$g* z#|?B?b#b|s&f(#7>wENT3~F@V;)FU@td9F$y~-0G*;o^@d1uBtZxb>6_TP%GH`e!< zAkb7-^L6b`si5TkA37>t_o?6z@Hi84QFNr+;qZs4E&d9&ccVjw7p(D)>(6Bx2Mfv$j#p_i6U$oa0 z1OB*uX|tBl^a4uEN=dKR&=15*<)K$c;Fln0F-mzroISq4*}x2muy$gPIO$m>ExD~| zyeiHzC}wXC`;5`HjFMHmA5&5HL;RiKu?e_o_o0ZBWfEOJn8&4W^5m z2g;p0fq5~Sofc_i+Dcwd^n!aniV*MG>RAELAz7l|;E5lP8tm?{Uu0T5Lo2$!YIZof*%M(lgPnVUieQsV~+U6a}ZWEgH z@*b>k?#UFg_p6zbZJhuw8{dh^2XZoeNNjNW=5B3I4Bp z%32I0eUNgCT}HyF`qAX06eQTC0S^DoSq~?@#pXU=i^DS8uT%+-EEK16$?E#l&{zxZ zWv~ZLsA=G?{I5MZWSeK6MjB;z6Fxd;c>X+fjTC_jgV(z}EkvMs_eLe(8{JtA_g=?7 zHns@vwG)ly6%NrYNIdtD6 z``C`;3q2dvwe<_GuOgG+v-)Je4PA^*bv$`1&e*fFv+%~}#e0dajuPSIKFj^R!Yt^C zQoQmWlv0zK>7smCXOLiGy)o%CEPUEJpy+JH)l|8~Oh`o-rKhPObWHkF;kw~!j;_tC zAVbbHCtbY@q}xYe4IqZfm(D}lEGW5d{w|-Q|3k9x2 zxBf`OGCe#E$Z$DH&#B2J+9Sm@S2(!lDVzHfye1Q8`3`WV>p zQySrf+WoX;2We$~$P(3Rwb}gqK6r}p^hS@-xUOX??#N+#h##qPRI{dNc?J*<5@Vp; zygGCem%*Lhl%#82+=F^y161Rd94)OSr9;Jwf2T>nE~-ayxC0 zHPa}QraEaV#Eejf*C;0q2$$UH1oZRTTKSUXCacypXx`D^WZszk9#cAN%9{7*7t5&k zQe>Celk=RZJ3Mx+rhv|II79#pKT*7{&{3^!Y4*7upDuA|Uj6unfb`w#5g+_+1a$Cz@bb+DvIX+z6&6r^X@04Z z>KM(Q5eS+9%*ELz?6Pz5{m#v^u6)`LFIwP*HT@|ayU>lDa+b}hab%GTukYLz2WPa# z*NrN^ko#$)rk?{KcSU)2A2c`7CEHE6zQl1pJT(0?NnN!IvsZop5c{k;18n^hIJ`5J z@JGkX?=NlrU-9qXBo|qCeFUxb`P&&(^lF08{X*3H#Cp~&okzaW+pmK|zyFq}XZ~Cg zfATMUqHXTNUSf$@B5O4|K(BlvsO}}035(;`gT&4&1HwbMJMiG^2ytE&Q}sW62IR@& zC`%KT8@FR;N+GYSGoaB8PFGZlri&Q>6NsM0$d=F)C^h)s)Wgb^GM0cCh72Z!1;Bjp z%NKjifRAmZuK4c+3$$bxt~jk0#RJ8{5E~1yz0^OCe}zb8{2~jrNK8}~@r#BS&Sx5` zs-CZXdCtmbu}`{ZXR%}IkUpJyy((SMrD>eo0pJd{ zVy>;_7)?m!ZFcE&4$zu8(w|$bBM%Qa3pkU`2N6aQGcFsE^{AOvo1I#B#0?T(shQ6t z+(R8ryfU~jmB;db2|NFHqXhST9M-R+>0qy|*PSAchZBaAqR9p$5CbtNVUZ%+ zcwsKT0epYtzrt>b?!s+kDZ9Qn!xq>%Fo+VyvZNn&TdzyjfD8z8eoK~tA~PZ(V@&L*m4{ zKvM-7@X>GEmKc&H7iS5B2^-(GcP@9;<$v@ozefa^mFUuAY?`nCi`Q_?f3;}(2`xRE z8f`~|9GLiR05;zfH3}$s9bFoJ4_kj_jJrfUm>CqgeaS_#i=8^eBKMy22#4zQ9@lA)3kc8^xzaC6h*viG!;B^`WQW3$u zHot+wtiMFy{B2AjM;RezbT^r6olCLb3^``dU^Gv#TOnjWd(ND5RPaN;4IZD;)ROq) zxPEc&7jq1PimHf2Hjxez|T|_Yg)xxIX{p_T;BjcQ9Wy zOMf=WOeWJtSch3rrkl>3$FHdC#)|Y3WI-sLwO6ehLboa9-*5PJXJ!Rd3zqmLsI5~Z zFFV;jJZCZ_&i)2@GT1iKeasxXsd|nNc+wDdHuJRY>s2AY8$dji#L@K%?`t+tZNWBb z=2&1~8`K98EdB=$J1So8gCy-fCue(pBO+=i>74Z`TPNFO2-&)FYO%xxP&YrXAbhHi zF>hC_b3!Iz<*)xLe)U0Sq8(~81qw>2@=DVy=q1OjTSg;2fc((9gu`gE59Uci&iy7|rX+mKa~B~t0==KES}{~Q*M{5I}Wt06S}$?}HJGpm=R zJNW+GIv4NJ&Wx+ET7jG*xen6pY8P5OP_LVX3(lP~4+JX7q`N$Hy&LC1_Lp%HJ^^E# znF@c5oUzb%jzi;bxuy2g?|I(n%CV@8I}8b#_};I#O9T z%!8S|MI{=ne9E8*;yE_mQt;9|JNvD(bXBAUqF&f zBI1YD*Pv-oUg9^H&1Df6AOFh$EgwfhPOLc!b(~=`(C!fO`M)(`1qP`(AGy8I;Ii07 z?Z*?fdJmw7++n99+fB4CxPA`w&N5y9V{MF}-Y#@>&+nt~^W1(V$u2~Rvo1ogpq1F( zAXVV*lWU(nb)9M#cC7e~fVck;uTb-V%sfkEqd>YjLWDMxqL>fc_&z6uQ&Mbdo9A86 z>RXfjLT4y$0afR8D*q&u>W)@kZ`0;JuQP2Gf zk)2||@|$N}#n@59^WH>rKDKJsp1p;dO*Q^w$e2rtS|(Rn@8(i*_$nRsufS^gv4cM> zNEHOG@EBD=6rOHEwIwm5UiRn?cu47v9)XshaZTg<>e$Z6zP*0UJ7iUZLaqA3azcY# z<}%vngH4K87#C`S5feW<>|7y0N6Ixld~s&7qp_s>&$CjB;oisV*>x2Jrup0}7Ke)*f3&#q-q|r{VNNs(IOpNrF0PqAGQOvp;V)$u8x1zCiR=QypA}NDf?lr$l&NBn#!PurYsNXYw?=-(A_g6`GaJyF4Ax91PxIf9A33ctu@!R7ccGs+BzeQ8OIIxo3E z5>N`~JYnOQ`uo`|uUkhkZ7BH`ESf<3Ot{RglTnG?92Yw*Ml!dYXsIt>graD2M;TUk z65%o0_+?L`=ua2oqTgLfZ=-fp;zs;$B7e!zW&msvXl7P3$Jmxu8tO(IWCMSPhfwSS3rYn_?HL6xO41uvlTCmn@S{Uf7!$6-YB@M{tnYxJv#mtAUCWq`>ab6Rx;YyQ2hyh(|aU#4TKT_Vrr~{5D?`x?vBSp3R zbNZe^9p>wFR;3w^+dA*0Cl{+HbyKflUHKhV(UD3;6!tH{yKae~-F!xp=biW|uRGqd z<}?fEk+;-#kkgZ^xejcBQn|=6A`wCho3mH~N8eAq8h841YA3gA8rgI?oR6`7R}T;m zk02-r*|^$U)jAgIEXr~QSzbzHUxed_IfK(y?+{l$W^5Io|F9daE&WSwJRF*BH!PG` zr+i?X!6UW`loT2Jk}efD{I}%e-uMToTPD&`STfpvdJ9+FZcEbW?Kp^1bcpOzhXw`=!jyk6_)IVhrl znLK#|sa>Kx1bJeba~4Eg>5^7O)UhO`4c1b7pGq0f?o`$Dr*9^eO~=$t!n}?7)y%nY z=Ee#Q-dXh_4XhaVlvtC9O)S(|%ZmB3P9asuV!A5lBkdg4CSq)F%^E zTsR5?CzYjTrT&L4{$C+6u)hm&iSege(!}vlF*fHzGAfQ+WZ{dB@whgLw|NyddR@wf zxiRHs)}%6cMx3W%=rrM2#W&@~z$e73fW5|j9^sgZmYHqtAkN2L!hB?8LEt_4&?C#3 zhmx0GcO#fmRxI4pht8@_a2ys)A9S}U_pIChWq{%gu%7iWAv_&^#V@;3N)XhL#J19O z;G!UTom;uAlGGdNTOJOMfn-|DW#M)S;naPQC2OQDdH~#262uX#(_ju)OD~q3Q{4<4 zGAs)*=Kc#VcenWXgrWL2P)Q~kwVEqckaY2x9CkC}v61(Sl){|DZlN5CwEGn2n|}u# zzIxjy-gH*}OEkK6550O@hkHZt)2_l9ko@ia4d~pEB3^OvJ{g9PNbxtFqY?cpMm7Pb zvvZZH1B(O_BKg7Oc8L{XO4SnK@`NK}mWxCvSp;dK@lvLmw3#CHqG?|ft3j{HlTs*k z!Qh!o8vn;Jd%IY6G*;sH$4sk7c%0AeZAu(948s$2BLo1Z=prK^qNr$knvJ*;x}>%% zH4{a|NjKa1d~tk{s_{y#rp}Ob~iQrWro#i1hhv=N7{d^9EUi76KGa$zK^Ed zTg$0U)o1eIe7OwDvNJ_aDIsbP802s5Tp%f2t`f4bDHxL zp&nr$5mqL#zBD}S495#akUzlyE#E0wS&Su~oSyv=zBbjyrnkqBk-*mxOt11E0 zxA|DFH=>+ZdNq%IrKu{a#hFP+!EB(8RKZ@7wVwxJ0ZkfZRy@{^T|fFc@^FV7&o9z> z8F92?br9vlRo{fsM4|c_4q3`GGc;Ht%u6{1zFow3A4Hkfp~N#hRjze9W@;+p59vSF zg=Kt3kL*J58$bG9fw(cx`ac6xr;qUfraLXE#p(eR0b}SssLF~e_8f^Hnm?9{;>Fgu z+i*FBOAS2~8%s^WH;_b&BLVbkW;iomG1xW)p>y%8kkGVlhv8BV@E7TY6~%GG%F=&^@|+ zsG$0#$TMc-ySh~_ZEUTtXI@j4G`Bp-_*sYM1f1JM(E=xR`K>jY1NEb47e!3eNVLsW z>Vc7X+Jm1Zk~acA;#Dc|ijC?blrc0ExCtvRcm__EoSX5nTC3*UN^E`G7bf3~L(o|X z7%Y3dQ?>_0+U6#XSTK&W-lB+G*m}pwT{BYnDj_PLepaXzwaaRalzy8}FmgTtJL~W* zO!h2NH_z&vXv8g-3RqVW;{gn#6`ZsoJL(Oj+C&~7#x)sz1j7c%f}LutRB90+S_IYp zA38~UCeIvZQn$(+a6 zQ@g%D)%-C_Y887Wk>B%$W#T2`&5BNF#;7>un}eIcna%DqFk?LfXVqjZbg1>B%EZkR zmXF{pJzIS7g`h4+2KPXNLB|7p`0pZ?T7SnJ=~dhowEi7v3)Yh~e0^{C;~SnXM^Ey; z4~pYXLC4W@Zd}isH)?rNU9|-QdC~k0-$aho35+R7&Cf0Pc&wsh^DEqczH#y!&{(Rt zkhWpoa>lymj$oywDUG4rt<3r6Hop;RL&#Q7HGGRP5HE$NTgtgrWjnZSn-GO$UyFp+ zLzi2qfH;W_7DPEZC~s@VVipwx#B5^DXeLfOqa$dUOUhkP*C=|_Ge$j~?{QZN#0ao; z-?@*lOzMSvT?c-xh9ywRT**|wi;qrhqewsx6W4w`pz?kiGnRO(V4?8v_kN4(01 zA}$(|jaS**F^SmD3<^|S>@bnB+D_B6d2Ps%M z*ZC^rY=_d+QV$pEgyzPwth|DJDt)$8*AkOf-95xqv^UU%0;cspQ#RqH*l{lUg7ONG zzv+WZ`Egg0$P?`%<2S_dT6{6lDP?qOnSF)LVAj72G&$B`>34mMz>2qjgmZP12XVr@ z_k9R{^MejBxbYy9V&s<>htoy9U}R#rGzLMr&kwPtHlM1MA=Zm3Qw->tHY-<#Y0+SV z^^oLiebW==L=9Nys_*AAI1jp@H&{vgTKxFLXWarD9BM-4ph~bblTmLUR<#P32fP|wO=Zn9YP<&vFoA?; zvFRK|u+j3i){brp*Umm-4sE&*gE;!`?5Nd{)4u7n=5V%viU8MCeD;u zkj_;M1c#TOQJV^5!xV#3H0ZG)s7nf)Ed+*G}rOIM~vR1!md+!~UCHjP5dnC7+h5Y&HSUcWMCUlPc4hX$WWJP&LcwCeQbmk(4ralA*qXIceRPrt(uh@@BB6zRKqlkPp<*)< zH>7A)#GHcWT=jV2|@~b0?XXnQa%4Hj zQtC`~cnmw11iJvKFnxBpZ&jO)4)p)Gpx1@xP=<}FWO}daaF6;c!Z*)Z&+rQ|WtPDy zl)ub*jj=!a*27k$IKuzO;09~oh8f%<>RozFdV7DH_}`U`7z?d3k3KMVe#I%|mz+eY zDs%Qnuon6coJLDS3vPp#P5h|iP0bc!IoIfpZ%A%k^Z*ZhIx&jgNmIq1xjqqM0V^g= zxOk%IjRuwl%o;CWKeJi;lOhsogG(L8bT`CtE+$pQ0$~Av*3`!vmME+#xFE#iK zPE^uDGLB-Zs4!j{U?igAJ#e6H=f6c)yEFSDDFNAkT}^ANp+iQEDvg$hLHW_s%|z`A zJBpr$Emcuw@eGWS$no&!F-wPn)ff7I*aMxc@qplAZ{s9>B=|rN1~IwyL|{g~^E?{} zuXznD?hM~`LydxUndVRKu&2p&qgp22CRr- zl_j7eo4b}J)`1|bU#f0A%dw_R+4#>*`K8}H0V64LmuD|bsJWP^2d^*BmQ(-dWy&jr z%0t8}b7MFz{#R3>mG089mo5Y6sI_34&q^_p$I2CnFku+8ag3RM_3fCS($p>`K_>M} zMe=IYF@YjNBUP5@ycFAxOp9G@C%4GJTmJFCU|rL{4n+6$5tW>!rY0@*T19p3Av7(6>3+xC~9` z2btF&NQFvsfaG%udYPE&Mg;uUBM)xMP}c}%e;-Pnij)hI0Rwq+baZxz0~B^xP&WM? z@E#R>#_U&)7n149j~Y%1-1i03I`8N)6d(NPJ1mDZ2OutM6I_t4otH?dTbewq2^4Hz zB(LyWoi$pij^GN$LY6@8FRyt%L<@^Ix*2$gGq8a^0gy-uN6dex*HJl);c_jS%Wo*a=E1x_a zdyuY&2WGw0qy<2MRlo3E0N>smRfhZ+=thIkk6H4rajZt1NS@JguR5Qa3^iS&_}&37 z1)jUH-Tf?xD8+3UHz!Jru>t@Azc$)EqfXM>OlumGzkAU@;)7TEr|HE^-~%IpfWLZ* z-V3bRlGxkqeJ?#hMS*D5E>P)kM+ODY~?NFJ7u9;nX9|KNGvSa^fr3 zy`qHW*!@p}%)p>rQ8TXt3+d6)kU=|}DsSbP-rY&&daD+<|&o^R&sE%H5*rTaOM)?UsG0SMXVjG+syAUt9 zG3t0+%=0>;E@jp{S8-m@l(LlQtF@2#pr7@f7{&|Bt})0Z2sjWJ;3r7eH^Q%?|COxf zjC<}I)|Yj(W#}}Bq&py~)#mB{zL#iKPg=GTe0w)z=snW@l4YIyIvV5-=lV=o;9emS zO}V^6E~!`K<~cBM92QJA;ef?2-wdx#C4u`d}u`nFqpqD2uBGRV#!JsYt~* zU@*? z<7a-oT0LpPzI3dwk>92ZW&c)e8QHQ95%bag_QesCGHavzWTplg8sT>Gksja}DgB&@ zs7d3kMeJAh*L3zC=>Ao^xjEwkKpa(-Japd&oGzNRpEsymDo^=T0*Gy*nv7HA&@I(b z8xS?VpW3R}Pe1G0i2Qcad7{8t9TrZRG(aGrN}2yeh`*skHS^SGKJ%kA38=p)HZY-qv8VuO4OE@Z>B>`AW~;}({|lb&{vt`|j`F<^ z`wOgm>VE{$|9U?V7qb5E%=mvo5dwxZhfs6_x+6{B&D$%taYFB-L$`u>e>T#024uQb zz1Q#JY5S~2_&#xy!xZbV!Piy|Il}sAjIaMZlp-j)_191`!1(~vt$|DoBuq^DJ6yV1 zRdaz7d*`mpiIM^6{yEaIVAQ*txN=nTI$yTdtV0xLl*Ut(DG(|*vi7`}X6Bm6{7OZb zIo^*dHP4`l#X%C{%f{5FW^kB$%-F+VMqd~bF3ZrqxpcC2*1R9*oS zcA-Ct3F#NUC1sKPK*kHP+SGRI(v77Ysfv9}%s-^pAIi?fc?iFulBn@jr1)5n z)bFzPSM_R3r-;S@hsb}}#7P>+KpnT9{&cr!WuCi5|E>o79(W&@{Pf0vDDVsnuBi31 z7}Y>$3C07&_}>i_&1X!OjimavLJ#9=0P)%#j+7=J-ZC%{ugZ%rZkhDB)_%H-gzUl0 z!);&AryLuP>lMadL|em$r;mq02G#`WhWyRMaI5hv;brfnnLKgU{Jm@mY{~d+eBbW| zyGmq5n2&`p+QVnpR*){vSYw?{yVV9>#7f?&!n{becuBSGj)@jWq< zEjH{Nv%GrZG-5~CNxgP=Q%{&*aU2;DE~GL^RAva6>_lQEU!*)Me3_WFXjpfMd@rd! z`G`W@4a>ss&m1%X%qLm*O|PYTwtzSmfGN}QH^f`$*(vU-A-a#%q|7|Ggj13Uy)1H! znTX^2gm&Q$s9#F33unp@T^i0SUTeWUc393DOA7I->gcwd_a%&Lw6FVm34YSOh-0(| zKKK@gN{na+qIxlyFV!vlg~&U8=CX8O(VX9%3dsSv6S1gVkyuXREWdIA?v0LQJ2FmF z1{Ul0A2_n?;?%A5$n>neTkIJr(pFpoe_`HbL~+qhD3+fPJC<#jZ-&oM$zhHNs0pzM z3PZEN((dW>wgw6?p|RvXy(KQsX|k0ICIacUnzY0R+!mt7;nxhlOsJ-Rdz=%xZ5sPw zGkkB^%2DIVgIbLN9ANYr+MN@)IeAq1D!n^>$v->nP8bz2@6<(XGu|_%f>KH&=F913 zLsFfJbjNhz2+oZwE=WueLh8=V^8)_yiT+S$n&N|)VP zv}xG~Nu0=Zlj=4KUKc>zCi(}2oST#uwb(_e1KO+dGwvSe)$|_J4g{^MU^1Wp%Fs|d z_*oMrZf@xZJLxzs8)6KL?ma=pU$`s!F0Af97t&N87w-n_=L@-(Z4FId9-aMmf=p7=9u?l1$CEgbLd9+8@%EWxH?czZ@{!)|jwt>WQEQR9ebwDbyyMFf zr~(60A=sMLotya|IH`uFlpq7El;lL}@P5JSsDT?C(gKjXyRCJJt}G+udb1XBbz~eF zzFxYxBfzp2;DkDaf)KckxiDv{I{7l0ow@JIItGg@NW^Gwo?qA8hV)G@X1fM{Z>wWi z4JPxx8p=ITsH+#pSC7(spu>otE9UXI|DcYzY267^HGR^oK=&M@b0yuwP}`94B|-#>>kq zU|yOuM!}huW>!qG(yA(*KZZk%n#wLM^Wipk)E#X~5i?!|3Ar$L(HS6uATN(ioouL{ zT#h_|Q*KcRq1t=1mg*c=$093Ot~P@&cU8vtTJpI*Zm^$^;ZN3ht;tlSmg;FuALaz$ z?p>m#U1(m`TXP=tdq`xjI&vOT-SxL5v*5*#`}FDw7E$l(5Og+CFB>_77arr1?}!|( zD*j!?13Lf-L0pWbK9CFI{kMDu>fVW?s62oB#@QfC=PCJG`xk-gWD%%&l@SnQSDRjG zyrOf9kXhg~lFR48WLrTgjR`X)4Obbfy+o>AQx_C};kduwyh<@e!kFPiNU1n__E-Ki z)Jk?dd}j1WtlnI{sgu5PZ|Gp1EV8eaE-7fOnn-fDI*MJ1@1wp5y|p!`zG2Em zEYV#YMBF*_!<6XLX7#oC$&r>wi7L=ekqzT@v_%S%dEB~~*66?~rs^ksSP zs6N)-kx<(5<~D#L|6-gMu0-x+G@bTT=HN_<+S^th?aGyyAzHJ%pMVqMOxtq8v zutv>Hb~j(SMKeu251CAMzOjVmZ)>r|s3U>Zc{y(Qa`n-w4{7lrx$ z65LCQ!YKPsy`cv%qy?Qn4+QOQiONqFiwc%W6zbf0eSn^r+Y@515=FG~wMLv=nI6dCGt9Gk+PW8i)A=Hyh*O>KzoWexpXy{MjqPnacDysT1A7*&Fz(3zZR z=$N&Mmq;n9EbDcwgu=T6J&|lHe;>{KKAk^t(C=8qhJPXFYkwm*i3RNPV^4j2>mY9P zz~I#NZhV4%f;@bH+G>RM14GlS&`JRI4=XQ{Jv~2k(buEKMy`3rK(`n~9<=IOw09DE zi}tv+U2b3;K(rGK!#atM5y@v8zlnWr)m7-{nYA;g_aJ>$|F~jI;vZQlXP&B(AGp7M z6yC$(^cDVVQ4EIIEmTH{2nWq~c0QD9-wA6v*=c_cP8dm61PYV|eq|b+xwU=_kn<=wp;4u7@v@-fh&#TRR zXp3z)C35aly&apuEc|>be2ZJbxuu02m`8fI_*k|h8}Z(X@%Fhi4r730?c7sc#fioK zH>Wl44|JpWx6IJX3r%iqJs7PTKMD&36E3p4@Zgva*8D1Am8YDCh(3_P5JK#DqJl8s zb$oG)&+N42HgufZyi(45Jl0{lcXk@_Up`y(a`=eR)=&lHU)NN~YYB*bj z*aZ5~PqxXcOL0;5E&NMs6)s?WkTtPn=KO|cijg+_De^`2m}6Pn75BcY(lj`DZ}xjS zhr>{v@a$b_?60c+Sg3IcSJbHHJK(Nl**C`)s`GR|dfH6yo}=2QTVzaJP>cIV$~}QX zVV>$w;IGaT%K$Bctxomk1L*l3J|p-wIB6ye4+X$ZRYkGskc3i#Q5O@JlMx7aX&fEVy~;h z|Cb0;{LFP(s}3@GQ&Z}Ja;7^hyzMeKB_B}X}N(Q_!8bs-HccO^cXCz&Sv;3Mn?_~Qg3qQFOJP)=ZI?GN-yRjR4-kTTf5pyGG3aa8z!2%u5JvULj~}Ds7T%Dhwv3^-o5dfJ9f4W>FBxPRpV-~Fd$H;tnRYMk8n?_kBNy+7lp{S@-i2rGEzJkiducfhytvMcjVF}cJN=!os=@GCg;BU{!a3RI5S z26+!J9OZ-eh_BYEtu-xGX+v#|9srUI$1E5SrI@-$ZH&SN;la+vGyH#;dh4jB|2OU* zL`6YBkZ$Rg?v??gJ4Pss+UPDR5s(3+YlPJ3(W616Ll8!fQ0Wkm7EnIFeeZLA_qqSu zd++_Xb6&6OdS1`#`Ji;F9idf;TI?wQum=0?nBqf(NXjjdc^Y?r?h9fyFcJPj_qh6s zJ!g!~Y~neN`-_3gIV&7&51P$*If;neGkV|*^`^ST!K>|;gc8<^;(8l_mn9kRzI$8t zoF*@nS=q0S_K^W7p@Ung_%fUFw?wI|sc}j?H9#WP8ybSSS&mMaSX7p{3)M0k-^tpePXc2iz%ZjNPQKYR7-w3Px`!EvY&QrkT za2{7d1F^42hKKt$SxI^#(QKyR6j3_@UAVo2Ve6b(G0+|s)!-V5hOnOI@1kD^S3x#4 zU^CQGE^qvNSQYKKPaYq^cGfr4pLMN(nTDjiKUtdqdiJj3-5jb~k0)Opg^s$1Pf*R( zkl#OaBl_m-q$I+9Iw_3MnUk@eQRlIq(d6~g)y+AkG+tnX`n7VSjS}pAG*K;xkU?D+ zz}rbUzd8wT9vT#tUtMO8tn!$L%26xSltydP(orF)vr5sR_~<&HQ;6Hr4HW0ul!xta zHkvZ=je{zWOkHU^zwVY`q#4aJ3ky$iE|fN^+7vullYLrB>v$7)auw1UOnp%w<&ZL; z=|)-qU^G(NuN>yJuc;G==ZooJ@*EPtC?G$?kO!rC&)r7dOSle&)Gz%SYSA&XjD5TcD3X5eBP0DI8r12J?y_1}2Ad}3;qm++*jJ5qI;*VeviQ=}^Bw>s8L*;~ zVJEy`NBU)j%Xg#KJUg{>8l%|j=pvYtZs{Ehdb&v$C@M3ll?p7p^g{18I*#=9+nYXH zxX+it&Hhaj?BLYoRZQj#pZGx+kV)ce7HGzjxMb?3NW5&lb)?D`^XE-Vdb65Zm)WN; z+Z=8FI4bM-fG_@b)H=XjT@&+iN>YOy?7ydOv|g;Ky6TJ(vzM|I%0Ixr4KYYd^?x}= z>=7bnyZk3vCEN80@+|}8Zbqv|>^{U?q9h2x^uh!tx5U{-ft&Vm>Q5whE}g7~@lGFP zVFo$dXIn1yV!k>DPCt5hpQ%0Qi#-TvO>j%v>AwqnvaDkEOA1eq#8{Ws?sz2{BvyYa zm2j~)=w2Z&Wt2JmW5`rtjFx*(_Ijhc8PwY`@A2W9MqfM{N2}O+?DdFk;wq=L*QeVV z(_CkG^odYffkdZZ>ei%G*8*K}Ahy4#8iaI634HkW?s-Tu^eBA3JgBK(vOB4?J?;iQf9WQ7D|mknZy zELRihgp^-uv?0B-VV&WnyphQ~y<6BIY{a*HeeGpa#h?YlmB|V4Z*D5kfu%=yu$3=< zBwioEJ~obX+N+e~GEf#pXiDr;&;Bk=XyKdZYtRA$hlxD+;cCP+;%gAeZ7;XIg zgn7|$^d@8TOtB@+7PNwuSZ#`W@SOO^50?Hs$PN(NCaSrVk=B^gAH&q(*M2d;DgY;Hm=Yy1?M(M~6>jTk>r}s!m=lW>?U$iyzy30*Sw~8qas#e>^BJCK@5*h3FLRa7In3L&qwmKXX&M{2{MjGrFQN$yb?=HvdRzRXg*%#T z1;KnDB(Mp;~i>G2O_k@qD4zPZpYE{mjb~JhVpx{~h zfx3n`nis3<`$F53QII>UgwBC+lqM}q*D&_W(Z-74V4DWVO``DkOt)I^uvAs?Y+EBd zo5r4z_Tvcp$(DJ+leJXVj#jU{zD|R#Veb*8sq~PHj?CLqQ98`ou?u@^J3v|V?4?13 z?E6E2#-mIxS7ba#O4~&GlBrr#&z;(&uP^c7?zLgbkD_boorGTGFvx~ccPf)Df)wC5 z;Uucg-`W1TDbCbJQujU-&xdAvr=%TIk{4j|_?w}&uOmonQU1!xNNmW|<>Qe}%sK+x zE*vAU66hse+wn`2!=0VUWm}eVR#=9RvSXue!bL^i)GQ_&4X6cyRjQ7Tj|Ck_p4AqX zDgad8lTbp<68GJ@O+{CEyWfbu;LUP0AVVzmO@$IQEMOHo$O>3>C5`k?>>j1&HQj(y z!bYgZj4|p-1Eo}sGSB2pdAUouvm5Ju%MP&exTulxSjm;E8Y4^J#fR{?rG=@2%?g+B z5fo@B3>q}?>Z@XDaqW@VQ1$}FKvEam)OSJEABDBbKj8|{i!20nBb1r@?1Uq4LO1%^ zjZ*OD7I+^fW3C#e%Q=0K1Qz`ha><@vh0gypU8r$r(H3AX`)=Lzs5?b6NIXIrtFs0UHuulkvgf;$)V} z= z3?e_wVhYp!1Mlm&edPGI=0h!ik$+D@S`u;LB%Q_mSZy9agIcCY8pvcl@BBfE~@j+3J;n#jgde z(HGxvIKz@K8f})YqWu3Mk1dhUl+}pHqJVd$V?q950n> zaB2HUJC5`y>a*F3ieS>7Ue>JKP_t=9={+>8qf(4UzW#?eUF%+~JA_hdGO7PiIhxe) zu2-*pzrYDG*jkl65+&`c*W(vDH0E&Y^L#jo#0rZbD2lX1UCSyRivrkx)u5WT+NV1L zwccvgn^teDi`MxXK-}~K7LdKkt0+H%CBDt=8i-pd+NRcyZ!)I=q72?fJ*hR-UqexS zB%No~hYG#jSvC~R2&l?pl{5Lt`vqd1m--|U7cl!^w)Ir&hcKtSpfr&eWh8*|H!ofV zJOht3<9M#1n!ieo+E*(g>ghios=9CDr-COd67dcH;jxBA-s0Pce~kBJNf~b#q2*it z@yx$Yau~nr@B#P3;t553?u)WHfytzeCD1C64TxJ8O88xl=#-N&P2yihuTZ z%6l(}`*!|%{(W||I!rshOuj+CT_Fa1K&@ZFmaElzI`+EnsxDd&&ZMNDsE(foAY%`_ zHB~>+hN04oR(Bp9%_FZiY`D}XYO2O=VS7jZLG|45t)I`>L-GG|kbd8^WMW=^N zST_d@_g$B^cjZmEo!ogUsB@XJ`I9w&MbetytU=jLZcDJVdD+o4qMYH93gm7l?Xc9+Ch>!Z+|&$wAB;kT+1 zNR!FBy!ry08_?;{ zCFZg~RwG|5LZ%?YQUOl>P#@#ncpF7VsMA+XY)nH&@}&xp?qcm8;1d|oMlD6LvZe+` za1Ffr6OtTqrR{E*ARs__v|o7H7j$jV(tXM|L2leg#mRp_rawAA_Z}wm+yB5M-B<<} zmd?*}9FX}e;GXF-3e?g~ftBcP=Tj$jeQnv%+T9X}=Ck8#C#c?RWNNg=^e)HssE=sJ ziOyXVP#x8^e9%0aofuwb=XtU!dwRRjZYdd+$oE^ik9pOc?&s50pf_!5rVk_e{xl9P zx{ne+)YeXJ;cbV=Z0|ZT-n~f1R54p$KmR0b-1EAB^l%Mn}t2jmUMyNkm0?kx*6MG6PXI-WaGo z?*{`TN1yIFH?#vv7-r{F%BePHwq#5+4RQ({K=AD+h#T|$y2hox;@CvYEVqAHZi&rF zNdlS0{=-X=SGxTDKLLrB*)=!ae1i$$CH@t&lbXx(&IER8P46XM`!t(cD}`~M6PjzA z4p_~v1DLiq11aA>vb5yAd1$0t&Xq1>HOT-Gq}D(?<`hZ}2$2LP)7s(c-;2?mhex{$ ztx2|t{U%+E85nUdF~P=mJB;vMV7P=CL%hd^H)m~YVg&z zJ{}m;s;iKA;^<_OMdv)b-3kwUug)USxa;$&KV3Iw0UP8`jO@&BwHMK7iGus4=imD^ z4b=dv1+uOeW-LxS37)LMf6!;`P-iY!d?>U}q$ggsh$_S^?7MTc!x*ygKj?k2BCMdi zM1*ReaQ!46ULz*EN1M(0)iC^Swy_#rq{WAgj&+mf#f5xUyb(xjtQC4|(=TN-7`5~^ z8Uc!ixvC~zbk@?GG6`W9g1&t}|76GV=l4eKHz#f4*mT!T0V!OUa!vIJ`Tgpu(BtT5 zAs*c1SU0a{1Wl7=cAJ$e*t0pB>ZM)!GvjtzCcp*wV1#YlL<}6uCzcGYs4Oq$8~BG< zSn<7^d1W1Hy2gdeAD@zwgMeMY(zvta#X-dFQuP8N{{?iBnprJ%G%aV=XE&mK=#hqb zdFlEi)73%W#Y4-*5FM#!4!p4nTKQ)AjeVZ;buXWRJ;)g89TqyH58k`%psDeElCH?3_uQb5Rabbt}PsQEMusgQ3(5<>}YiqaZx zL^)RNHZ9+os<+u;x?Fo`YaIrA?YPUmVZ)uX>-cv=E0<*(e#1@9(kAc{*lgC~iiDn~ zh#}aKKgF4ecb}i#!^pJjxi`$th=hyOWgfo`83a4gdBex0XspR>(=A4Jyo*5>Cnzsm zD~tx!JAWd2*?$8$7>M~>1Fv{I;EbusYwrljdsSwO+^Q+p@DBE^sVnlKsEo;}TTOGT|6V!sb z2=TkXRgttX7udpne7W$S1hu&+S$HZz$HAUFjvmmPUB7LLd_$D=)!=|gscW}}>}!-- z2z9o?+LGn*L$6^XV(-tsfnVg^SwJ0RF@CvzfjQ7N{s8uKsfzK8%%uS8TM zz+Zvy>oFRWr7A1&wbyw+l6Syj0qt{j^){#PrZh^G4W1u05~9o5;z~f-iEQX}Caq_L zE8|GT#Bcp?jo$o87RFjFb9jg4TFTEQd*+bY%ijBO{IaT%DwbzvU{kj-`m>@52d4?= zR}GtNokkmctAwHjeVg0jy6w~$M%iO z9&Vrp>@6;83|@}0!GbXlJpk7%i2ve5)Y#dQ>&kA4yoPDq4T9gOX`EV?`fCQeJ7 zvX9Ob1iA(4t}<0nfq`eUMqu8)r;(6bn`&Ea{;&GkET)8XHQ8onku|Rwlt-$}<&&on z`o3uN@7{7-lx)u!T}m+EsJRNt49A(B*smZ|2`IUU(od^)6&NJ0}o8Lu8qs)MbKq?EqM zl_;aJ5My`flH+;OU>d~8kDZ^EwsJT9r3RBPu!n@wUGSj_vr-W4KYmohDYAw7H+Ir~ z20)9Ymsw!uO4fyyeiyM#e zO5X2k(qir+$6pY|-fhg@Wg*3f>PUqC&{4+J1ME&!w?zw^?$(8KnJ!8AXKY;-+kuG zLo#jLg*fCFeOE&^2Z3a_XtGt%rv#RwGqoM_`Zj=Rj z>+8~KZER}J`fIyBz?7XwnrJZpj6R4U>Uy-ga4ygDi=FWSnNpktZ|_r|m;7yifO`>{ z#$9>l_e-(-+QjBfo|kE(tGl>;q=Kci-qe|-Zn(Cb^kySwXO6=IWLBUuCci5`YZ6v0 zH6+Ru=UUw$$I!k}mtg6>{Uw7Uc%`@GV}DyEyG=54&`OlnPY%R9)UX$`(ObBVjyzjxRA6$zt`icIsq*_@1pP z@<8KdtV*?%Z%asUC_lN^xEEV8aWazO!FG6sOhmt7~Vhv~k`}WPx8bgsxhqu^$oe-3}6$FZR<7&$wVc1uL_nMCA zY`89j*E{_m^azYFwVJLXd0+o~e(9(mp6Li4!O>O9Ov=DhYN=6Jz_eki2js?$KOY2p(-59u_Yaa^c(9~YILN5FIvCD(ll?1YEwq|WOh^n4X8DBk zEO(zN*oTpsrpbjSoemFKkad2K2|(=PpfCbktuvdsfr;X0@5;!oG?Y-msN=jqXs=(9 zxWifBF5&du9FQp}s1+rKX>r%uHRcM)BJePeNfP!wc6FB60x_d~kv7L4j@1w8PHeN0 zhmzQm4N)UU7IED#EStJ?W+j`|OS6iy_r7T!JvUH@?YSqdiG6+@BO_eKEJBdwg2sH1Oou3A^r4!I!`srTsuS=st9>!>XKV<)V+$;u zX%J7t*wWJa$mm)qPP5RE;PSFOYdlr2J=L<$xIJV}wQ9d1%^fOhxvp+ASWrqu#@)m3 z<*mH?0GUmhHl2olmVO9)^_{sN*=K}zWV*CHc3mMDL*nSIq2)q{Jklhq0OcX98xhI` zn!dAC_=1P}Nu%D;fU$-vb;EqeP=mp;P~c5o`bCAOZ$~G%i0TPvPBR6tg0rYN;F9BY z3-v|0koHcjeGpTPxeDgH$sC1z_%^A7Csea)5+FG>`{g#cbv9`r8R6R=`U<)q$9QXE zIGn29ScGrzmJFJXb_+7T-dYCj%zjJd)UK%sHuRFOQZ&9a{!%3b_?DtLdKx^yINmLM z(zkzO(qc@t-Y>kHCi6vl|M)Sqywo$$*QLra$i4g->9Lna1(gdzXXfqlD(ndo&S_Z( za%OB6X8bf>+%e}SaBQ@`amtpN%~zlL2?qiJ?*BdDEd}CH3^svNdwi97E!ST+k-Z+l zt!G)X#d4)tQjv!=GvXV*Sw|%`w!1<&4V2jw%94MVy$_Pmew1m?#j7%|%W{FUv=ErD zrt#XhCjy_vvlS*LJ{gjS)w@<$F3TyPTMU{LC5MKgLu|3+Ks^OWlQ3K|K$=L@1^-Do z{qUO?pb78ok%zy;BER^4>PMSyXCwl*c8bAMiMs z(Xva}FkV;9ANT`Yz9wm_z_w`gGyl2f6t26{$#=!{-BbUH=`)3#R-tH1~~ZT!T$bUD-Q!r2WV| zP?jYlQ*pTHGtBDnso=t~t5d{{EBOEsq2IMwT}y$)v5+sAJ(P(?RDLl{`BNxuFt2^VWKVZLZ}Ita5x8whD)ORhF3P=- z*CG9pnQ;MJHq@w8-1x}{=yN{;B0dtQ?`Ay~X8k+TR92$F(U0QXyr9}+Sqn4Ah!Bg12s<~zCjK^obio;3O>rSV<&?}i&8P+niiR{e! zdJs1uT-S<&3~58v;c|t4n`86PGk6~e!&A;H41Kh!w zj+T&iWKrV7Jx%lx84gP=)`-)G;k2Y7e)~3@b4@)N7;JUV@%Ocw#KgOKCSG)CS?Z+~ z?NMW7msOU+lvp3qKP-mXp`O~TqiKbf_93Zst$*73F_1k{b6byryn)~pmo>f$B&nMG zdOG;D|3vz6idvMW@`E-do*(m|4WqU}xZ%a5l-&x+3NvCQc(S z$*bzKJ?FFyJ;l4PHt}u2g7~OlGT( zI%T?Xf~1~Jz&#Zf;OtTHl=iP#txy5+C{wYz_q9d6IEV9PLUU>x##R~C7L7r(&YaGf zYcu6cd5ImF8O@Q-2Z;IS#*t>x{DJ-j1<`Jxjqv15CZ01@iuyY8pzv_N!ILvhr5(`< zvNrT#tW#~g*vMEJcC?iLh%nvzmCKg(driZc-I7iI$GhSw!b3xs|M0MF&*h*GgjafX ziypIoXEf^M&P4lYPkg+%Gxp}w{K2f|+I??w2ife5aZ1#rid9af)8B%sb)&O3EZEvL z5G2+0~AxX z=nER$E{{7Q`2ap0CSmKq%~aXaDT|^L{UvPw!!wB1Zg#Cn|6m*4#iF9M32d{+3=jTp zxe=H4e_|@s=VAkrGbOl7vu~A9LbWWGrqC2FA@Iu! zuPjf<&}{BXfo=CF#I$9{W;;9U%PG2}Y865b~w5%2sp3COFNn-=lP7Kl77^>)soKrL2rA(lPYTq8`Q3mK(^PNN#IawR#YuJ7A$>+WoonxPvqF zx$FGX>)Kc3F24^JmN*1P#Tcef?Duum&9X)^wb{qFx9lx0xe#Q~W`AeSmZOIQV4S?7 zh~}Vo&FNkYr}Ggp9ADl)A9}^s`h?Lq_33sLON_3)nyE zTl#Le$wc0eDBX^|@v%BN#{2mxVAI*r(-hkdKZYZm|E#drnyydqL3 z@skYYut^$kduZCvHzM@%93lum7{6V=eIbxP|oR#xb3Jz15%+ye%qyi z2)Bi6q9KOPOUeo2CoPBi%|ilBTpI&l*%NQyn)X9J%-EG<8=N1!2Uf1DC27l5R~}gW zAo`Z@W|sTY82m_k7mqvVjt!CzAbwo zgtjN+=HC8C5*Ho3lMZ&mz+k@BO{9MLW=(Rc$}loJdH|kH}Zz z!pb~6T1>;(Uw6%Ota#F`6k0Wu)Env<^j6F(O5+rE0i44A8K=18BTl+;sdTKG5r^9m zIK`bn-b~HxPh7M%!bQBJ=xYkDp*D=xql`?U2ffAB43U9QuYFx6-T@>5qpWe9mNV3= zh!CE!T@DK>BUmUmHb7~T@KeSI8e9H5kW8mCbB*jkj+5lt^1$VKxGcVR?sTv`U5o#b zr=L$qUWiXOidqXbc(5EBO25&~EuzSlPWYa`nx36b$fX(83ky?t_IpQ@&|>xM^Xu>X z+RkB*Hntg_sRtC2K%m40lFAWX){=?$ZkQG7m#IyOq;%o2b=r;j3VR0ukJLpb0eqTl zPS#q+sVr*(VPe>7tMB5z#8$nu`5ZYvKi^)KkZ7d1SLH!H>wZw!s^YHAsOMgPLsLyR z=iOpvD+*0USQEQmKJ99oOf>7z$;Zu2$~sLys5ODqMOSVe9)w0hiq8SZGFdx7Zp-br zY2}!if?6f|1ijfJG=&u>4$`ekAThUcnykmS6}9@v43A*n(4I11{fw{6&*l&JKHx*y zVsU{hYRLwAw%;)1&oKNo)|l3Rlu7g8k(G+vxQUK6Y0jJb9?km@&G8DpupsV3)B2-2 z55J4H<~fdrw=21&y0~r_7ZmNCjvbVe^+z2I8ri7RTj=A*4+&Lx2NS5Zg1oy4V|iJP zDpr~2IpRH?T^ZXzX<2<{2yn}2n7)L(G~$&gc7r#YB}DRZ!7&H+{IwbHv$yS7@6jJW zZcRjoC>upZ;4ZhKk4bKBO>_QEPo*cP#_|iz0$OHHm_vb<*M;De&BN&Lr==!ZhM_NV z##frY;H}o22?>)GO**fTm|3QJv^0C<+^iw3otH>;Aw%9l%{q%_EY8i0Wk;gq$zG%y z+xjBo+fHPeZ^_mVctDwfl(PWy7D5@8B<`iTN_lea8i>(BKn;a0x}78o8yR4* zcC1o9;uKpzqDe92i%a|Q|C!th3Sm`W5Hfm&GCEZ+@1KF#A9uIsH&g#RBuK76+CIpa zH6rf$I3B}LFt1U=NMs^XoyMe9*v^&}>LT=0cvY=wIHcwoOgl74u-M{ltn{W~WMw<$ zql?>)wwnQ56IlBOQ9<65}z z&W2ITjBBZ|;za5Xk&rd7Opm0bgrF7!B`^KWHSHeR_-A8lb(H8EnQWt;iHIkJF}Fk` zy{FHv=`8L6SGCTmTlO{yETGt#6}_Z*PM&7rq=}-vrICHZ>`5tXEjk7Z`d3xtA6FDW zJE<~<06lm5^nx{0JFz|_LFN$fsK*o+oVI2z()8$;+&n}mn%AI%u*+1(sSdQiNTX)= z=h9bJ)HLPC^X@O)>g z&z@D52SzvM<{Ow*mP?Iou~3gazD-KBlLS;|uT?;(SYmquyYCe|6-UM&Hi$zN8e9J{ zTRS2K%H?gI2cu?^D;iKdbr)R^pL#s>hMT^i{U`e`xK_b;NV7u|Z!OV-{rDdt$~G^5 zqF{_CT6QGw)T1%b`EgFX zeKmjdGrzGqRTa-cQV2x+g${V=F@g1hURBtYm2-;Fm0?ORx>)Q7{g-)0(dW>V4v#6@ z9dP|)Q9zT)F5a2apqc7~(5IJ4a~a2E7ghDMWF4Zk%ugg#j(7NsJlDytI_G*M6rOO} z9}TS`=R3^gaO#5wTk~rf+Vw#)-@nz8S4zBPW{!KVGGjec9|4@j;k3jfz2>X|c)n_t zY!Bb-g@u}sucZi&>toWalEv_zfjdt0P8-rPhvdC=1hJ(Y>FQ7+x0JFOAVSQ@tbYaw zAZTwfaG~U&d7Y7>Le*Ab-Krr0k5A**nv_Jl6R+-|oZyT49nw@iFK$vZh0MK%RlZP- zE{(NO&44CT_|1}qAY8zT{lzTfwA=h_>Mh0*(e38TBamgKVbde&UOt_b_OlW3m_;Ti zYWr)-3XntOse@rk?yxJ{Q~w@7MuM+H?xV7F#$3>9uCEv7eOZ%cKz1 z`j~H3mo*FY3O!9n=vK|)2X(gYMI#k0Hxn6CEzLvbgsX0104q-M91thxyN$xJwk6iiy}Do8RTiMozD_q0QoF zyeUojBdRg&Qwr&~H@|3s`#TO!u&y#+z84{dYnz{M9=!e!?+yR|9D~8XCH9!*a8&@K z(&gFzs(???mf2RDOxS!`W3eG=dUbG@5zHf~f;1!2OZ*g5O8HZH1QxWfgOl7NQ*e{2 zxNx5HI21K;IxL8-DlbxQPB(xsMCNr4<{YnFqK^RO(DMiH+@xaOgcPmyU`Nuqn}|#z zA*||y@ub7rAs)fGxl|?ehHwW(N&T~irS4!Qv)6;S_i+^RbJ~xKsYkw~=g`1}XTdTG zuCww=_MPsCsy5}K@b6Hcti9nGw<#uOV0_I)X8V4iy3+caN zTK$30AV@b7=wwYVU1_8F6bEZf6%!3(bm3*mTOd1sbl;f0{-F`Vb+}0#x3MOT9hjph z*FE=Rb!mQ1YaTx`CtM9eBmaGAqo#>)iWSC>SD%jHlu*Fg0m^BpQPSo(#lZa~A7MjH zp6?&yo+0k>d02qgD!_@fG>|e)2pE(BSNIR_w$G$>|0zN@?XjEPbP5CWW*zU&<9lH( zT)xn*W@8}$eD$#}{VN-9KrBu^-oRX^bF{X`_M1@$;NnZRe#%VJv>8IOHbOHqVx=E& z2>S!r{aMg=Qrh0nj0Sd}lrdyQQr~=VPii3pDv;o7n+kY%)V-y!US`lx z>cc#izV6$oSep`?d}&!xC5Sht@`b~JLN@WD;(;6jKb?p!vNmtc$iEvtKG;{MzR&gP znDy4dLt-Q-c~Exl1umM2zQQ@mi#99*$&VP0{$yOf@bEjf!PaV3otbK-F^`Tp!#b!M z%V6*&B3xsY8;_+P>9s)rH^Ia1ZLi(lueES=y03hb*2>oU;5kglZy2_m2da;+6Dm{c z7Wkg>rl>9e2yzl+rv@)7?3+qS&ZLbrP*o7do$TWOO{$LQ^+x1LdIo)gI%Bc=L3e^1 z&Pu=Fgqt>k$hnLa<98oF0&2ElVj0!i z?r<21oI1wX3&uK7aK`v)p|KlS3tC&EWpy86{ViR~>yQ<=#sU8ohWk&zSMj6+$J3Wy2<5c!ne#=H;C-VWaM%sJrV~9%Ag_v6Xrs!2MkUtXF#fnD!V7x%9B)jaW=5VH!c}0eR{L&@)G~}eds^H#m>bKr+DLE) z?x;imk^mT95$`?t4{z4uOk)GbvNX}t!t7k@eNdfI}mUw6NRNulundp)H!GXfZGW9W(^+Sml-A{?E2&^ zS~9QmKX5?-TDf*tRapY2l_@)t(#^1$HXTz3kz)J;jRmw|$$C-!SA`l2e#Y}ukG0lR zE6G_f)hAD9_Cm^qweocSP1ad=cDTFfYl#8X{-u}MEJEi|rY&nq=CHuDb3;;wTBtVJ z0Gz+9WjdaKbz!?Mw)7#&k+*~*Yed1*k)84U#;lB9K8>-}wY+VXTq(AN9j!(H8#nEu zq1Uhaw&DUUqd?b_=6_V&tNCfFp~oj9#GXiD@eH>FR0b`k7)*=a=e$cc7y`bwfrX(Bh<=v)+D$E^{sb5aYLlty35vdnTUl} zIFq1-t3LS{L@0Yv&Z{XQ!J`t0ol|T~bOu9WM+cgPJpdmfp?E ze`{~tMN;;CfCo_;Ir94{4xaA*jr@TdPiOvbJl%ake{#<%x%QagKRkWy|8u*J6@$5O$mC8 zvds?9&0leq>7Sa)_`g`II(u0uhRiclTCd%|eEsjBFWul?h-+%Uyat4d-GLjRE7&n{Q}_yV-)n`jpEhV)OE0GMK2DQGp# z^|k3ddvlvz3MnrltDdS%jh6s+eq?N#b{|&l^K>78;S%u;1lI|1&eb=>al#*ke2-}v z!if6|;RVo1jQ-&t;9Rv`ts^u<9QyYd3jRDxiHcHHoBt%X9Urw3rYR>HZ5tZM>MSEs zQ}aESsbk8uF!gLMPmV`*Mox(U%ISz0TZuley7ym;Mw}i|{KC3QVxp7t2v6&+VY=fa z96lTD6a48$7y^Xt9387XhFn{vjG7+6h4p`&Q!7PDZ*2g;6NkYqJ~WKIaNw^phNFde zmCM;#v}UHVewkgFtG1u@;gj_$gy*2>%3`rE1kMD!;dLWOi@r(^_oA+-;`SKMq|TZ| zm=Jf(98Bg%NA$Z_jEyha_*q`ooFWh2ZN1B=&JY1@7%QCL0U{(N&{vHmj7Jk&8-J0?Em?^In;)Sub9%b4Wt zHm}mVU!O<){vIYTTw>sl(BE^V0tp2^Cp}xpl$!J{$DMW(A8;^HoA7IvgJN#tONGOt zT$Dfwq{7eC7U%&8{-yN%7_$scTjtDEm&bU2s;p%)9+bkCy>)=n4p7}kj}Yhd*+sXnC!y8 z!U9(|{z(z`yJ~4F>hzyC9*j>)gx1CwiH_aml^_-BoV}?`BXSWR4AKvDwc_0jFc;E0 z?7srQp23d$f6Oi2%dnQi%sYSprf&piwN@uoD+;f#u^sMR=?XA|mPXOq zpV);NE!CP!s<8uG{xP?)KSV!|{d0G?uj3v!7di#tO0tElj|-G?d2Q>gX+8nz7HO+M zF~?+3GDN(?bbleCfV@1bU-ai=h}~OMw2l)8#z)q~8Z#81gMVW1iW1{H1?f*GiL2)z z>T+ql_c|ZLDt$0@DdkQ#`j_S-_3{}SNH=yNM=WUeoa!}?VIM(5UE2694Z$;#V3qR? z`C=-M6@%XR$d8Q*kvny}^81pZ&pa6lpVa0ZQEY^%*{XzXH?UWq<@^%)=PIYt)0>I^ zSuH74a0TS0_M`FQL4x%cTY<2-35v+0UR4=Z@M&#m2|RJ|2Y$NcwJkq7 z&6@<-r8W42+Kni)uC3TSNm%3S(@Akf^)^wY$hg$t!Cug)gqB4!qPw+mrC5^&nzi=c z*HAVD=2Vr7qLCn0LClFbA?&V@U;saz`g-0~xU2QbE}DC}r^+XD%HdyO^mAomrlJd! zxBE9O?h zEwo6&qLCnc19SOic{ zEN~hw(Xa~5_I>DfQ4ONbZkPctJhu&gdUHs+Y2t*t)qwT!3ed9bXCo*Z%+i852`~D= z_aPI9E&*K1N=o5HzwYbQ8G6VI@xekgojoMCq~uxx9p{vYdE~UOPyWLfkYnQ}jNmx| zyo8U9fOsEKlCapm)5{+FJ>L~EyOI3Y`*4i32&dwFPCO50CTsDux?s5>)%qPfM}~5J zRJ>)bpMSt$$c!e2zBw`%!bV+Ig8u%87t+!sy6v{VV)P##v?xsfU-b^+?q79$J@7<5 z+GKe~vEqK-JWQ?dGIB9)p~SWM5HJJH?cgd3gEqhUbYdoF6FL_j%A6}+_oi9B=1u4v zSx6YP1mG=l1C|K+uG*RZx9?oN_;|#pbS-@yQ4WEV11J5>6v>^~eSA-Bm?~mXs0=zU z5kRiuyq^XEq-bp|wOyEbegP8xTvIvNBwV6*cFCe*l!mPlP4*&^sPrw7K;mkJU{WF$ zG8RLeYC&KBah~AO+eg0i$scXT@Z-NB!6QsN6_S)94>3M_NnkDPXk%#AWSEHB)(hBy zJoitH_Dm0tY$BO2HQ`LZeFj=XQ%85XIUK{AfCI(oL1VtO=r8r~hURg;B zF>oD5^58852ibZgNNK=|JOXeujd;=p=c|`4L^f35`-I?%i~sO&tjNVe9d968)xh`@ zg5OMsTGs}5Chp9$O!cSK?#Pi@OnVR5puq*ZPV6Fi@v4pX*3_q26QD3NKUul_-Lf{` zhfA;2^=;S_mW7TmbF0`zwv~zVFt+51P3z!x*I@lrelYga)n~cb!}aoiq)31uP)LHI zt3)SV3Zl zHl$gS0#jUMZ9aWvB^~%vc2n8Y3&XNUk}L1=o?X9O>>ZI|PH?xQzH7s!EGY-oCX8(c zD+NTEW3vtBY|nhp-{s;0%`O-w0mGDKZ@xoqW8=?iMkiPsud0fg3V%c~x~>0oZA1Uw zj)4WIO}oN_Rl9#?Dg()Q520v`T`gs3>dXe|R4~ zge{(FBZSxR%9zjq@s z>KalG%geL_iaUK`6x9j`aG?ZY!(w9Ll-Dm10O8tQJg369STS}mwmfayz&%Ep$|5Ue z3qy7MayV#;&jLI)Dg3f(laLiZUG=r6YCz1oyqf6R!F)k&L;cAUjthejW7N@V($^J@ za;S4pYs(L&W2G*$G-3B&*02??iHnNfQiwg${Hd*GUa!szu&%eD~#v2X+qulrJnY1e(4cGVq?NXn)|6RE?eImfxdI#%Z8PtPn zCuu52$$$N%qHVK;XiWjstV!v&v0~NP`4@M%CDOgC6YL_FJ^D-jzW{6@lirf}Gk)>T zKf6|7ySH2sXAk@Pb4ug0OH^Y^f2iFtD$)VjQk@OvrwIhCtiF;dIj+v%WRZpd4E6&R zV8+1-QIZC6>ra==UeR_iY;IQaE!#@C>Zi3`5ofh#+iFcdr16_#hoR%2#J;>;SSlf6jB(jUO)@#Vl!Q1vDc)EedQrDiz93s# zn`nXm09i2P{uQrpq!Jf+GB>U|&?!pBA#lzT7^IcEg;Rm}^sDyP6SxLI6WDY5*GX%m z$k+if_*T`Yht5G^Qz@iH+c}GEBrzETn%cF|x?wA+x6z}Fa!nmvV6b0IhTPdh5xOU-hnPZAF-U>Z2v$C#`IfMMjf*n{(={bMsbg*X6s` zYuppwqB0VC(By9#NfxBLEPB;H7Rdhq)Jie`0Fb4xoXx@YucE#` z#?p8a#4lg%IM4nTlig{8-cBxPYNzZQ}nI*HA^dSLV=N4mtH7D2Vnw>k$FhH28C?^8-N)0t0qee zMabKcTxI8qWNhKMtSg;*NeTIhqnQ;NO`e5y;z(i`LI={h`|lL9XASz-7WeZ?NEtb& z!>8U}5ru4e(&p@I>{NAZzjv?8AtP})_pIR8m#>5k&2&09d=t4$uqPf*fe z5!#DBqIF{x?wppB;VG^iURBL^Hd;N7krwH8?i}K(-D}ZXA&%Sy9<|J0YjR4&m(4wl znw?F~msIf(3Q{AreXEYX)gDq#;(r>?xV$fux~NGxCzDdy(@v5^vs`XHYO!qhrZeQy z6IrvVD+*0HMKx+rIG}BgMcnE=O?u9p0$WDDiYv=(V=~`DgS!g#OI$2!Jvgl(abjKU zjU+r&pz&4E)T(GXi9w~NgG-(%3~{8S%`G4bTCH<`1)>t|xraqwrn6$#DnHidSk5*>fYPox{ zr1~SNpIedmwTsJ7`Rgd>@E_w`{-Lhh+p%e7YnjJ#bC2-+E0febM{{wN54AtKwT&e2 zDZwuWG@49mxpDxf$V98QKHJHK0Erb> zw$W2bcY#=vUPox=DCOPl=}g7t@!JeeM{|xpI-)sxvd!+SkpcnS)sSuOwO8FEY>VEu z?EE<8gHFmDJ^2--eP?Z?*(Jj=&7REM52B8=SUHl|#IVt=qRYb=kgr)BVMpR>-Oi^C zvfwb8{{Y$H2lYRtVd_32Ylg=BsOY1BK7z33zjr)RZpM*>r1ds0HAkJZB!#^PVe}Q3 zE$=u6s64FwO*b_So5l$fJ;Sl6i)WnDQs6IB&!tJAVwDD%9#Nt@;h@DX6ks*Fu~&lFUZc~NIlUi2#Q#s4=R*G3!j5}^`?usnQ=8D@>CMzE7DsQp&3sz** zSb{C!R9{yd)d_B0xOq!wusl-GX~~j(nGpW+MPJ63?Kbdn5^p}V+YLVVRl7@DD@kXb zLxhebEY0)+y9M}B;|qIlv_64y$_8)>vwz|3^5!{&I?j5o3S{j$bpB-4+HOz+2yt~FZ=i%WP>+BihS=2ioqgMrR7US65WecXOsvpRZYhV5m+uJ#W zH+Hf^Ctb2GGxR;_^BGR@)f|nkfvBy(#rh;5=P181_BGR5>8CtopIYCv(Yg|~(^zbjF@2mzDy# z%Rlux@pybIaDzquf_-W4XWaQyhhAe~I)vz}c3b(=P#Nk27Y$#O~SP&CB%s^&C} z6lhC&N&D4R?n&=eOB!ivo?~?ciCR*7(+d+#%_eFua_lKI6H2i=o&{J-K)kbJovtoq zRT*xBH8NW>a@G#2 z587meNxsTfJIQB#AtwpBDe{)cW>f9ol=ufcRfujqeVsaJA0oK zoDdqZ?tDWdXXiD|2_=-^0&2`URo%fDM!}_i#j0?Q<%zY49R2fDn(|pm11B}VXW=Pd z22A}cptbPLt+)lk@lTbB)Pnd?!NX&Baz@l|t!&wNPU2z=xf!m=eS*<&0@Z2t=;H)N z0QynNjNMs1h^+i8Yiyv%7mnt%zv5kDFrYE6YUf>%&)=+>^*3^I$MdSagrO;Jb+>*Z z2nxl89@T?)tjQ*Rdcl(IiTl-A?g{TwEWLiNU9XP=>635ntTeSf8`50}tLxPBxEfRvOjVlTq2PrE6KZiUTsPdb- zIvut3^j2{hC47_7bICQ%{{X^CI3-MK6raRZud>_gHtRCT!fasV&X17Z*yqUj6{eRZS2IF&|OyM*;PE|3)p6+ zj>66;!%eBhkM>t5@~S$dwzAt4wk+EY?aTlst6=z53+wQHNRhmo=A z=~QhdepL5ylK05C(^II;wZu|6;?}zeu9e&nX;0Bk(#|7%0!rsAP;hC0U2Y! ztB_bcnE802CK8fMQoGG8$C6LdogJ)_9zm^`G_K@+m7^A&#sbxH_cC=>$i#zB=Ok72 z5nF%)<+}k>-AQX^q)Q+z{lx286Kayb%WR%fp6kmWsjAUzPrV||)bnj3Sp~X(bage9 z)5NFuNLc>>dCf^4pL1|`6(c>(SsHX!<1!qYompw6p`1k1o~IwBN{}LQ{{S(mrk*w7 zK)|XJPF=7mJqt~<=L z^@oc!>mLaCV*A9_R(A4T-7H}nZr?qK!WJWv0Ll6a^M3^VW!A6!RcEH_YilL7?4lKD zqu9|q942BV5Y)lP{C+ku{{URpoxR(OpWfv>lfhmdgIDn`pQMcKnor)h z@f1*5zd&o)J|KAS{t#rd(R52_5^p3w%(+No$p8R5oF2cfYs>x<-94|wJu&V!ta5pg zXWfy2{ct}T>3%Y3vyT=-s1#!b#k#7#I1HrZepo+^Rj%w~&uBR#lkqpi?-KZzLX$_F z+axkNv&$4^U4oqL>x^dyr)up!0_hDaYFf4763GczZOI&_K`ef_+CMt;c=W>9`D@#L z4ci-C6H!tzn}C1u+9Z-fr#@*TlhuAJc!uXxyN<)`@L2A6;$JTCk`Ka~dW;J^i&>&O}U*DY(IW>zJCh}W%nF5MHubJ-~X?_hu6Kyyk>C1fgX z%6gfy_`_AXyk@kHD6NBhfs-u5)c*kWtD;dKi1btE#H%vof4cso+PK*rES;l}Y;-sv z_BGacW?wHyp7w8&5=#Pj`BZX#oqsCI_bQy5c9EB(*+)L36mFoX%P{rDXzSi38ePxX zq9_?u79?k{L0q?jwa@Jz78rrx!tIRvisZf~>rZRrYx{7-1QGZ3jZHNNu#~L}qt-MH zb{`VkU0celzAOjZ+#KNjcs1kN?}(JcN?wv2Ha+Zg(0JGNsO?Q73 zPyC6W_A6Nb0POV_G)UnKQJQ9H=^MX6Myz@v38VnA`9^BmS>R>c6!P%J!i<`Hy~*e7 zYIs+{*S;ygcw(O3*4_m3Zf%e_@1@8KRuE;7mQq z1wLjyPbA~e8uK>1m4Q?wDe3|1UWf4kS^ge;JwS5$KW9R_n*P;AS&@>$%z}_7?B10+j<%Vn3Kpx?jh9$a@xExjg01J2@U($XZ z9Y@3ZdAruNxb}JLDks@3k6>Ac%--EU3ZKQE2=NkpBk=v9(@&qJndZD-IzQbeNzQj* z@#&gPx42mG5f`mEC=}L^Qw!7Bc4Yig7_-QzcM(>E>McM3ZtiFX+A9#i@+qdC8IOLn zNf}HN-jPX>no2v-L1-#iUu$F5qNN81G!rdL5=Wj^quP2^n3xp-s$gPXa5GNESFutU z?gNH|vXRh=D(Bk2x+tndy?P%+{?JnxJOyIIzNZ-f0LW6;%@2JecQy3a$J;i&@GXo= za7L#af8;5D#=d@o!2Ybim5)h03mQH`BF1=Wg>fR4$|0V zg5*`hHJ!5X7d6b!tUDjQS@P-*$L|_0N`q%xHoB@Xn5>(Pa!2QVbI*3f{pzmw2iB93 ztZ16k>TKV|RAszke)UjkkF7>zB-Ha8Mv_IRxhK6>X^AxvB+*fGRTPI8CYnj2ibVw! zQ9#9_iYNf3%_S57;ZDyK(M@LNoy2uIOo1a%E2*@XAXfpWUYOhhc&^IAbdfY;|aacYkvTaJhqJ8u^A6hKR zhs2LF=xR?h<;Qwit!IJCJZuQ4in5=XNcxJ;RC-rTy{GCAX*h7iDdB}*)%8Y^RyfWw zGHVK)%v)|xK~I_5#5!x4F~6N!v7B4n$00Zha%!lhq^?_6(CTiyR^~mH@$Ccnny_v! zZf@H$N=|!rrs?*E`t4v1w|9J3UuWU#8&wT45-{#h6@K)Nc{o$)smfbv_ZLJ6p=SIm zdeg#_xD9b4#G|%3t*LcuOE~sIi=Rd5Sa;g=(r0v=7tm8tOL9G>No-w~&qvc!1R3P_ zJXbfW%M{XwiZzUM+O-V&gE=7w^sEZHYDGHYaO2uVB0o++57(s&&yOOOc9TC5vu zQnTi!=Des((VAykT1=uu!I49K-1A#0EuFId=>s2H!%>q~(MeHN5@d^#-^(Woc8_YT zS(}ncsiL-dF@ai^x;x2&MOBR4qZ1-KcXPn43vD6fIVx+Xve5v+D_d}BsDtj)F?E^B z+h{v+m92O*fTQl#y~LK*F_)E7y48|Agdm#2xzw6N!(`IY7P+ONt#zq|1SS+#Eyd%=c{!@A@HDm9~+0bilfmUwY|MJ6mzlawu!e z{IJKLdbI?MrCGDm{88g=FIezBmW>2Aam{b#Zh28c9f`Eyh9~r|Jk#}iEjwM)?#yp0 z+uX-12(gkCfjG~8D!SO*U9dAq*#7To*0a$Oakf5vXwfSfx|er1ek6FV+s79gT$*e* zwyA2e2?ffkmfg1)+Q<8*w>}*Bs>i~*4c&$1z2vgbIg~U?q+qI!a>LTGEp$R~c2>k1 zI;iDwP};SF+Re+Dk`|Y91W8(WbDl(*#CL5a|lW0S7tblTr9rU9s@> zobb+)v5~_MECbuzis9~cxYHmJSl2!w>xQPH*!`pCx!22o3~a;@21%n&@w;euo^(NE zL&TAjfW!J9dh?5&ahpFiWlgFxe-%Dnw|Y?4WU?!VmtaHH|(+PsQe@=4~YOLk9se6dFNlhE`J19;lc!u}`z z($rGoSUi`HYzn2aagL-8-nEb8?-g43o5lCKt%^+@wBj&}I044gRXG5W)E@Q5y}>=| z7Lb}QuX9MlOC5iL{B?KWuMmw&2$td}Llp2cW6VN9;PyEK2EAHOjs5|?VLqebi@9!C z4b0Y%OaB0mGlFa9a!mx(Jxvm^)_g+o4v*qr6dglCv3n~c3MZ2Aq-Su>P7sk?o@p~l znneyKlQgRnzd=@Gv(GGXMVk|86#!$F1!3g67*K6@U z;g|jp;G?gGERX(&PS)I>X7)IRw~jto^F!Stj3SDyi6lA6sgl?ay-xjw>O|8?DWzB# z@@mErDUT=}D7SH^*cK&TO?93q@lB4g;cpCSmeE`!`b-izeETmGVPkNkx# zeAw3~)~*i7;;T=ZS2{U$mOU8N8Ffc*;y;~Yna{08)BDrr5~7Vay=FYMS9_9s)e|wM z7c`uWBS{w8vszE~ zh8$x6RlBWPNpqPnJu4<#lHDr1F>a=gWV9#Q-QAYty>|P*omFQna!EAWDtWGGqhO+g zNIK9lO(j2Wc&VedEl$mew2IWCwIb7tV+&T-MP?s%pD7NaGVS74#tm-RX@X)Zt7wv^ zDhM>_X1&wgWD>shX;{urn`U*EnnIu-Eo|HA>Z9(~E{);b}tuvBpSFXeWePPv1&= z)}&f#^sDyzUP$?9xN{>K-%e==ED5c8#8S*eB&$n zR^7gz3?n28hD}Bo1_1*Vm3QI=UW_U=Va#-GJP@uptoxm6IQRrt9e3gaNyAnp&alpY z^)(LTS7%>);tsfK=WjK~VsPKBbJOZ9#QfDpOQKI&t0ZX}cQxnL9lwk9suEtYJ!?8i z{ptpv-j^}5re1rZPg<_mKJ`?{b4)FoO`518XPBP#8FFcvpqiVOqhU)@W|K7KjhmqS zXuB&!CTIYjgEe6-d$mitC`TA3pD%JbyAUO-zg{Z9w~8T`BB6VyVT2ULj!6zqY3a~T z{aIk$Mw~{us9JetJ;&0eOF%hlllO7bvnN8bTe}>b)ae2s`KcP@bQJ?kSka;_U6xFU zohl}{C#671G?ZLw2zil-x`wAvBay7*R%OGyob(D4;=F*7HEt z+h<+XlS-~v=}kKmC34JyAlY7&@o<50Rf}h}GK+ks=1$NvDLrLUOFn)+|{jsF0Wdtv_oVOakF^i;L;4G4k7 zsWU|s0A`9PpaPbYy%YenQAGe0QAGeRYAGpb0HTU00HTU00HTU00HB--DXl;ZG^3>y zQUPe9iU3^H(acAFl?4=D!E2V*RCFrbA6i)Ccv~RmsgAWU=qaX8DldzT><@a;wlT=K zHH13WotZft*0E+YjgfL&c!xNw`J`0N-Tb}twe3vjy~Rx|I{nd1o5LwGoThVIw!N!OI#k>*)bmT<0!(uD#|P6j z6_odJ0z*q}j=h7~`c# zG;@xX(||M3Qs8>X}=~(_6=&&==nn0$c1}f>GcBz+71GQ;)r0+$T4G5k35W|E4QQWlCkQcy8z=>tEosUZU$D#MZ!gNmyY%*cdM$M=skhTt#VTCT*GAdVQ_VRo-KxYuU~y70A!(#8ZiGt=aq`@n8sSuYqM>qg)}k~wVW`}Z=VstkWr^x( zD5-Ac+QCIBq@*yQnr#Mv5Sl2WfrSQ&4ImPVD4+t0D4+t0D4+t0D4+>dlgq*DQRb40 ZcO?s&DJY;~=8V!BN(XUG6j4YI|JgKL%eepm literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-12-Business Meeting Discussion.jpg b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-12-Business Meeting Discussion.jpg new file mode 100644 index 0000000000000000000000000000000000000000..ef2ef0ab2e091bf4f936296118c34a39d0354ebc GIT binary patch literal 84296 zcmbTec|4Tw`!{?I#xhxwQOKGwqsTJKnnA-D+bp&ug_0R9)1tSe*4Ws|hKPwL)S#fy3 zu%9q&6C@-H6PAUoJciW3eZpa@KhVEFFd<OMMS{a zr@?VZL>9hI!_Y=lj^r=4DGX_pka$0K)609ePvC)rL5G5=w1~*4<0qnHViV6LC7-`=k)D;!$jQB& zcZJC==5R|&%gQTm+^lc7b-S_YPRpamt!+=*pFZRD@O%4Sy?)dGap=?V=aDa8N5`h7 zXJ+Sq&i`8YEm#)}68^g^@cegS|JS-?!McP*M1)1e1nYtcod6GESrNE~q3Ajrl9+#( z+$N(0aineLwYrBAn#SZG@&V!9i1nK>AGM|gOIuyp|GR~q{r|PH|1Ru**3}P53d4Z% zgk>RXsFj^d|M}wFq;qUG@(B{A$Rc;+gAF^7BrF_BrbARVqZ=<$iXfbzLE+LgLaMF} zHZLIFtQ2OCf*IPQ!*Gx`ny*@459M)@-%8g$~_d&r> z$a?Zx5|+ynVdth0g-CQCxcXX2@=zj{CESdxCkheK!7L#p6OCUNkH(>p@mQ3my!LxEhbSCetjznw1P8;wEupyV__YWrlmeIY zAeakY$Tcf~XK+6WPFN`xrHm!g3a+of5elIDlE5VmodgTP>7!U8#mEy}@V?;I!PSCW z1$R=c+$BVE!l^6aAi-zNIHdt}=qReyJfll+W;Y$OMWgvdu|q!|$F({|tU!K8R-kL# z*yb3wN81(@^{vM{?$2UsJ!=RbWI4xo27@o>O!_tDAp+eQPtj7cB&QB!vO1SdhfpkX zm#_2bx2&1E^Se9r{9Zo17K}1{h)M3v7<=M$0 z1M=Y-EL)NjmL-C$Zv<;b2Bgi7K}UobU7}QogsS!9@$Pyr6w(X6MZ zAou``?n_)<3s%A*7MyIWgr)|!1uNvS$V2fcs1?}=grH%F(?den8aWsu5LbsnDOnbX z9IlTNikBn_R1Rq&8`&8WbaWV&^MC+17l<&--t0f>b^U46*pk$i-jJ-@kEtr@OmN-T?B)o=`lq@+X|9aq z3&E@|w4?x$5L+M{RyP)IN}+z@LT2Xt-gWNm1OsF}0!W33KG}tWS@e`!^ppt@j1W-( z631#G>jMA|jh*OFD}*F>V!`TVM)6&v~3p1>kd1tqP?90-&D-wkiGOW9gri6KZo`Km1nm zs_@D^Wdeks_Fv(ff2y8hJ_;Ne0H$r|SC^4j_PHE~(-Pi=3Al_Tqkuti1eOGCr&9n} z@a6$mTuH4Ep@8Od(VYL9Wh=0M8!T5S9;Hw)`kI}9J-bR90xme4o8W{8ya6x;*8|{R zm6A|y3K=R!!stE%V+EfVr3MRQ;!yxGt%gYo0PnoD3>Y&C2{C|%#6_uLKuN3C59k2g zB2XF^*|=%|OkeU^2Dr;rGXM&Q>j~(f72pGgYDIyQM7#bfgDwTW-EbAlt$N5j2_UO3 z341gUy|!?0sS*}oNg!Qd1ym^rNkmfx-@QsE0y{{X6-d9HD5T^}DZ|7AT!XbGXCuc@ zuts)1)ex(a{)n^!$*VkBirulCoHOR}`M%Ws;ef^!sMdZ3>f$vM8plrsU4OlBPjC~l zZ2|7#v*k4O9U=xB%<9Io*a0y= zRnWn;?l0iDZ3KopISnnfmV`qhfji{Duz)8d-qWR!Pyj~-oyfI{C2IeR-(s#10uH{F z1h^SS0Nh*#u>ha})8^7`5kPq0{&-5O9500ML9v7quv`wzAr{#GU-Ab+hl6vm$N=D{ z&at8dbb?|LA<^JXKwwzFssOvdTpgOvJVz;rSSD%VJncP)PE2v|9J(HwjTEXABfsEj{gxrw2dw$KX&DOY?@z= zx`}c#T1vhHC=dk*5eXHHzAHyf0zDzgP`P^;$^AJmFYog>nI0hIW=C0Yjzx$V(W5Ue zp<6aAkFs{z=5nU@b@48!@#350O(RPeWJX;VEejd&_OkLbojP1qyWhkO;4JPUrjRjt zNPuxzl9Sw;7CrMXWC{buQ##u z=?ui}pJJtJ#OB#8BRF&2$)6_|yO#>z&w8I~=VB{AxzR#|V+P z$ILPdu+@g#B0JaR=R1>gl3t_~<7g|;!Jhj!n#-<7Xlm16DS0)hz1bgy1m|XBt0^Vv z1iSrWw5>Z21d+V_I#uAFxg}tUdOP&e7?dHgGjc z_afO3!;|~0%3HQurP6GuxnFhv_(WS_S-ww91LG!`mbcYfLaLTKoDU(}>QcfIuo23) z+LlMT(;pT~?pR;fk0y~zn4D?2`iNP&Hsz0Y!B1*jW5|tkXXn%1HE+L3EOooCM=y0Rknczbo?qPjVF1H^7B`{veYvIR9r9z){ddT{V#cOn z{$WdxZ0ZVPxhDcT;_m38w=&YsrV8hMOs#tKiz;=mpzqH74WhY+!_6-WrTa(B?rSZf z|Dk;0M(*?Pr0x6m;3Ua^MtPF)hBN(^cSo$p_Gd*cP0o_!KhWNPlD=K8%VXqRWUKC; zLk;Vt1K1Bbcm+Qls-utCo5`aRnOk)anbvGkI^-XjO6Ur^dyhORT1c_gpzCN{4b+pm!#3w&$K(>bK$Q%m-gz zcxmxGd$Hv5vuw4GqO@Y{R=eA=M1MjrwPCsQweHh7S zj~+UGE|&Cj8;yq~ZZJOc=)2o`O%>f`9HBB3?`6ER1sJ0EY_(DS5te??C5=(H^?|(F zPfQ}J(f8UOIM86Se=`MM$5blle?#-)iGsaGk(l$)!{xSGcR=Gu7m`kWj z=udutOR}BTt@_$%=`mvJM@y+5CC3+UifXm_ypHxOrK!B{bs*h8d`)!N*$`Vx80p+w zr+nfjhcJ2-@4z6XXr$zcv`@Tpd}fDh0dA#a`LRXH$`l zGSSFVdUm!AeJ;X{gtz;Fu3;DT=bh|=pCHoIP=~nQI@R{?#`;<`U;o3nyFmqQ>`X<$ zvf;B;Z!!){c?};mtB^Cmbjd4ju+sJJIxerMW3nkr%*0L`IYkmHjmbm8tO=q19+uyx z-0U&u_Swt(yU7!bXf3yq7&zCooL8Y4;h8Z-d{96m>dKp_n+S`_`?xmvcerQFxCkpI zFhh|K_#fTz-w?H?gT2gJYdc$E?cv_d)>eD(RG{a_+EVc{&vehs`3Xf%kDF(xQMrW4atHq(K^ zK)=yV?0Dcf325j$KdBsTjFs*%FGdaqul?3tNab{6Wjau1CEZLY+LSArC#6q?zI8LP z5%$5b1+HjuhXqFRIQ315FhL5kR5Et%P~-@YHg)i;AueI~Mvh5I6QM|r`F(gGgG%%o)8x%m( z^4agAVCez4fwWJ4+ZDnC=f_GoB2CN8n?I$yJ3e(5%R_@-MTZu zLM;cF4GI;RQ=Wk%3AsI7ycD)Nfh>&AdZu)vq`WPZ_RPdds9z~VDZx!%ot8i!DQ|c2 z3G_HH?jo$9iz{oz*T2zR^oaRqv{I$w-U<}i760^3h4`7+yHUHxULWpdqMZt-8!Kdxm3(y9tjpt1%=X|9*k5%!TK^0I;v1b<$OmPG63Rg60!X=!^5TQi zt}Zuv?MiKpV8;FK-t+Z?p0Pys(c-9GE~pYmed+^#Kp{}Hdqs5Fi2#o0P{KL{ng!sX z*ovY#2Qi2m5w0BoUh!&700jZK?CgU(AQTt*fXauIjPb!Nu|yj)3?nE*8aWu|ELy-6 z(GTvxFpUAcFcb#@2`j@DLk@r|T3}c`#vtOxQX`y2FCRxKYEu<@oG1!iLfQ6I*bqZC zcx__`zLNBXY5C#=N92hxcNZ_9n0k6ljKMsvXzVoU@}Y3fz3!TK&-IIiSNk)Hd_qiH zA^jibr28g&hlQJW9fIaq$1$HyA+7`o-GH~^^-aggUEe|^ZoV(sExJY0cczcC=iUkK z{_T&8`xw?HSLI)x>^;X&on;E!8@Ia3$UAOU7nw~-dRM4>&{Q6jl$;PyzKLsAz|B7c zs?AMRjl-2|4CczNPwb`YR;71nFl0SA*-WZL_OWu|UK!J%LT%Ohe%ZJ< zdA~!0JtCh}XcDCgmOMP-vIG4^#N}4s#>@0l#pfes9G_&(TU^-)Cq+2Tj+bzCS>}@` zBd$jn)Qa?uH$47j8Ta`8zTbm!vsyL%jfQH9Jx(7j^~0w74m@J^=7)yppJ@Pq%PHe> zOyH>{4W<;Ae<=(h9n7X<(W6*)LLb1}0YG6_WRU6y7ljue?zzOf{>^O?epKgOrD3L) z6*aymNhFJuPuW3K10KKh?ymL6_~%)*j__*)w>e|l}S z(Yp#py(M1r*E!aA1r1dY7PJeJt!l#FXC8PT;i1OexygY5H7D4I>KgFV9(V8GxcTkL z-s5+^R2v1vjFejtje6YfTp95owsDo|Pz;0ak@rda+ELx@foW3f6)gq^KQ-&Vcv4;g zZys$SqrHRQ&rODKcTb5H6G*l3OxC-t`b*MIq2`dcQ=@f6#gI`l?sxo%h zsd28t{_uQxv8D>G$|5_k*;JW~My`jV*ey+)8cgxJ#b$FQ7Y*NEf_K2QF#pif`^B7l z{^>I8y99swX3F*%oR~FuV35Bvu6c{}$b&oIq@1(myf2w{`gGmOTWiK{31a4Injl}; z2QS*f`pWjc*U`jyV~-j<4oa^t@_Bl);bnx&j!Un*{jXj!vl`u;@!hiTfEPBcVT-Hy zI4&ua={#Rn9LHqmMv2T2w~pn_Q`HGESHisQNLE+)Vk?>P|A!;2gHN%Sv?ZP80ZhbBD!LsJYx!brFef_?3Ngr zXIl#Ej=iV9v&)gUrwrsP`SSPoB(Uu6bUPBHgj6Uxgj(RHpJVf;!(86EXlF1*EKXqN z$`VF6Qdk+zJ38IrK}E^n@RZB0%hw{CH?|(S@_H5-Gx?R)gnN;y{$QSUyP3E~V0%kJQ7e7B;75tX+-Ehc9_Z|oev zgr7GIHDxtcus5bpDDAsB5Et70O!@I*tLlE0F`1jbMT3}bYGhoz;LYy_02E@?4nH^6 zPAKhs7#XX#(_Q&dAvRX&l6qnN3iR#m{Ry6w{^Vl)ZZU(32PaAw68jRk?|zd(qPq7t z`G&NA2@#!4%Ltx7v;y@KBKO>|{>|SlW~|%rKxORSWzD{t?Fq-@mfRB`op7j~E}b}U zZ2V7mHih$Rs76L=uk;GU-S$V*%kMr{bux71V3>8YNev^)SStdX#p+I>!zqQ@DpKP;Tl1X<}Z@@Y2tQPM?CKjn%jx zm^Ynyy+*%u_nRFrem9}~?NJD~)2CW*P-VPy1v;nE=|u~Smb~v3YxQ!lWKsF@+qT%A zq~E_^k$0-1&9=@?{g}3%wEp7IG;nIN?)jPq&1zF$v{gvezVlPHC;p7t44OV%-uCj+ zqsEAxPjr4h_JgtZm6^Os7qUGaXCD1S<$h?Kxi5b)>GSK)pZM;QBLz#G7qNq0-v5Z* zQc3?pmVx)Df8=($H8xK?v7(;NKh;}vIYuRfuuWo5DX#L;$FGdQBTo4iX*{zxH9yRI z#r{#Bt9f|OVoel@q_*a5_B>pRTog)k&|G4}+Gx>Hl7$UDMQiq`4XWjsSsEWMdSI=4 zhF26G6BMBJ+P|-u`2%jk*H1*G99GNO5^S&o)TTn5&4OY+k8o4^oSE6#Z6W-tkG7}8 zq^PxIFUy{^UKqZgL#}q#Gk#utFQYKI#v%GsU9E$vQoY*r(>UGeAK%M>#6uP+KW881 z9-g@tqp>WEKRUcTe#T=|6WgFR-4UmI;-^&YpP7BnvTL2qv||$Ab)+|b$gw?IBZD7U zmLL0M+pq!^hAaRK`S-6tohEfc!JpZGm#}&as69pgvsdt)zEfaG^rX z$*@0boW?Pa-n37P(SOWdT!D1cpEZN)7@hI>cK^-VxKPqkrr4huwVZ8LJu-)kpM1R4 zyj9|?+q3$Bv?+Vb%YQ2Q_BqfpY=M$|Rc2o7v(vc2H|s;1&kyYR7A9QOa--&2uY|>s z;o9Kkj`NFWGe$Gw!s2vy-jsg$Z8+L)_qgPTsZpKJZZ6wA16Lr`Nxv^kSH$KwOtYw6 zaiM42PY0+ZW;vwUl+aEEiOGk}zH)Mm`OKGsmlcHW zx{vd#^jIBIn0%vTv8_2!A!o7MhMkR32uq`FH_;d+)eU8=tt-0(#q!ezJS0?ZM{$ha zPmSA8xm&i)6~8#Y(0 z3Z!1+<)+i2{K`(z!o2r&%i&wfK^$q_G`DI(J#*9N>u*XC7gQO8G=kj>1-(FnXda98 z!pbfNOy3^I5{(BvH%Y<-H>CLC&S~aZ@fSC~i$u6)OyM$H+bh#9Ts#`t=d%fe*dx+= z?HaBSTX{=wIi(@EFofx`xiImweQ?)Dl_bmH>%S4aPUPT#=b66KPoAZ8vmb$al9HwH zFWf#c=J5&YNym9LU6hLUfMgD`n2@Lz=~3?JpWu2)_ugD_;og?^@}(THd7LRVL*}FW zu@jh}%a(V~>*}sRK`qwj$f@RP6Z9J((TaH97|y=^T4VVu%Nu-wC&;L%4U-EJvZzcL=fSgGv}*Sqvwh< z&l~B5Imml`7w${1m+>)ESjY?xxJ)n85o_L7C>Cp`5V5JdRY%_ar4?23Q(v0B{((JZ z1=ne^p(tZL#8xkD=`a_4nzh+UVvxM=wXnAxunyywNX4X`B;S)}CFg0>JaSA+8nH?J zc|JrN<{+O10y>;mK&=1l58uA1s+?z{_j*h`t{3 z8O$>?bl-#*KKf{3dYNyJQ>NP+aBM`z$MWIsAv-roYPd`uzaf;Dx!kd*>YZt}Mcmyb z9M#JTM^!<-8;VZqB(xVzD!pnN-7U8?_R}>wgDhvm&J0xG>(e~T_HosBNynHF_BDGJ zHGAZqZT||Ivpz5mi>$Ps>zX6eOejFJf^rL%>%DdrMS7W)tU%TBJdFi-aEeZ2Pqm-JyN*NNwk|jN7@=#oFpoEG$r5Ef6GKjJoh{jTnEB-0 z7SgFMlpF$HR(hI2icaPPg}KT}QvP{{{%{cKrx65{B#}%VdX3n36XA8?Tl>9_%o7D} ziw3i$r}JXTBNujO#_-v`{Idhnd^T*<0b~2(+;f6f`8nt~cUp|AADnc>_qjtpElxd_ za&FD)@!An3%V+f(s*D)HYlmmg{-G9km{t1O{zS%T@-BlbB)(9m?0Vc=t97zV!^gMX zZ1;4et8LL;Z#HWGvy*W9GeX2REmFCv!YP*Gc#LrOGu^Vz!Cac0 zyN8!@o4Z$*zHck(Wg??E_Zw@ixJA8t!JpeevXg=k`I~>3olD+|Ho&xVW42ed_MB2;Uc5&+l$q zM~-__+2HeG$^W+p^V!jF?+c$?@J(yTHW=>fXs^2XM7z1TP|KIB%J_tLK&Fact;Wvz zpO+V&Wg%+cS3Ku?l#Y8G9LFDsM7Yk5*{$t%a>T+3IcvwsT#RH_Px@q1By}DxD17vP zEnZ~T+Z%5{R+w?Qln%+W{L^qtW!(Q;dkk^Kqi^#x>N03D8IiC`G@ zL=89^Fa_vP3oZ=0b8NZ@2+f6O5GSL`IzPzco%6{&giqF4I%xkDGY--S2kwU_H5%5T zz92uUbujTK`Sox4cxciIkEYoRZbl=n+0BpNw#c9Aug(G+7-6gv>KeIhiX}Hv1728|3{FNi=6QUGW zq9mb7+dob=u?H;2{YtZN`?a{QdM9O?zrASc=Y1cZ#oA809_vrueQMm@6@4qvxD2nVFMObXqw1Qln7gJX_*)Y+neEvxY6C`mId3>78?Lxe zPk(LPowZ%Jo-vZkI`lGz{Ry9J)8Ua-Ww=lw5lcqVK@*1+Y@M8Sxj_q(H*z=8?UKA5 z0*a70i3NDXtw1CCsK&kwkLr{lYfStfK8F&k6m+ocxBDZm*M%ucB6%K-puLC-^|0%! zc9_Sh2c{a9x8?EI888DWsRq3oMiQ~^o^b*tX8GmF7yUV1+gE>4a(3^wWh>sT4Xmou z|Nc><^fm`Jl#i|-d(bVgJ3Ji8&SH=i*mMF47QAYQ;7u?9K*Fvepn*{uaFGlk5bS@< z6j*Hw&@Uvg9DzlF{y6Ak18Y;-@1XOp#9`iL=f1nnAV=ob$%OA3bw%<=-8j!}a7ZQ4 z7zY={X$i%z7hHfxW)ehzQ3iykS*^RRYJCr{fKFAW2yGuHDKYp* zu(C>^Q?60QEfE0?e-um#S>KIa)6enCQ2~Z1>Q|?j=j9>}bhmyV=tCKklNcl^kukz%EXV=#) z$+tdZqs*_mR!OY|oQWXeff@vsi9~>b0y+r~k?_GRI6JEV7!Z=d&S(L)2m#CF0HXwd zt#T|7J`8Z1V2%I`$B_S3s1a$phj=PEci&xnQF!%XyMfrqQVNA@&Z$SrGU0ma8|mn_ ztUNHN@|V3~956oz#uv~IJ|c#9dbGZicle|cIOxOpqDXR2;9~s@+2lGyFnWQ`T6&fR z3=`8g*k<#-WRhsd?@gPvYOxRp188IghLVpRFJrCD{a^gjYWtnP%B|@&yRNn5(sdC) zn_%>d>Y`<^7EH<@p?W1LE%pDKWl4G+*Xm}dKHGcBYP~a)s5ST83r4F5qA_<1&_yqxXY}%DMu-hkl&T--?1lUKjN%X{#r5ZYZFF(#-8#yl+9DSjr}-> zgtM_tRQCfsv30}q@uB&>8#43r3sIN>xh^84OpnETe?ku5(t93?%EEtCaCNID%c-a@ z#F$9stw7HEs?EF`pJHx3SX2MIFVisd>m9d0C-n~35}xqfejD|*%+@G$SQ#W+siZq^v63DCEp!o+J3R-!qJ6IDY;lqJ_Te;-17;^xkedeR#H+Vna=&T2b?QhI`Gy>rczN3qCM!1LAi{Ga2b>{dgzG0Bm~ z)nTT&-F3#r!jUIfqJ{POWF{6Z1;^*8O5x+X&OJ#?uB7V7yA``%`&DTh+?D#spf_Pj=d-+Lyfoq%h%#1o z>)zX`M4>N27tURhkaO&>f+^dTxL}VuymMu|TNw8|uygmNutMF+CF+43N;5gjGvgEM znn7KZHi9E5=W;85Bme0a6a4+$^UUNH-!XJ&Y@c?xL+#fm+bj2kba}47)WP7=ZRSM7@-~l2SO9Cnva5^CRRkw@=_71?n5l%va0lp+; zG998gg88U_2c<+4;G$U|Oaf!CB68t0tf4~L=fz%~ABfVrEn(yWXV=*o@^dV(`wX~f zqCo$GiDVKmeb50TE1_6+8>=;!o<&4)j?G0rete$mM6~NG!5E~q2oOeBalSl3DbHH! zaXG5(AVcxA=#4gK*1f-YRQ5FhZe~0u?yi$__Bp`wii!*6OQUsBn9@ng$#cgycYwPE z{DtDcY*;*^fZ{-_jnyRr#yA9yxi{o5611A?x9N0zv~xm6ggq7CgweFC%UxV zk2c65t~@c>NLEtUWTrk58>PsSOiM(-jMZSNWg7MEMqb_(VWQ!vYoytz<7cKF(m;eDG?(+59St1b~7(?XD#zS08TLkX$gQiVa-KP7~(KiX7aF+r*{0Lm3wK0_!@oQi$ z;i#DtC(2THY)0sBS>HYNlGxVpxA#B@H3spvmNH$hE3?Ff;e9W;HTI>~DH>5z& zZf`HYsdJ7trHB;DcmXpV^cVMI8X$891&1;fXg7X@E~@v}6QF^?fPSNZBG?&ga6%yX z2BEdO3O*P_V1g(N3s^%BNH|{DadnoEi#Zz)yh1PwF&#eWKI|VxjC<~L{mgf>b+?rm z%d;1G1$aT2Ww&}d7@Gxd7PaaL1hIUcJkMl}&sVNMYssqIcJ;(z(WAorLQFB@6E2DRN|VsW{C>WlIXuy%N6xIX8|)@E^Hh z`mq?Ei5OSQP__X(XW=e`MIPNbTHzDcR;sqx8~$m6S?s?8X_&4+@2fSM*d-E4_{VXfzm~4uEio!93F>1U{K20z_RP8d!0v(v5qS62F-6fVFxMDJIZp_QD&0m1P`7lXq~(YdOEIrVyHidg zEYiTIkn+m+k8?^K=Uv2|=`zipaIf=|9tj?LEf#Yku0+5js}KZn`2Q2Ln`0VHHX8c7JjDEnm(2F7+w8yQcLU~C&Ar`0pSd`OR!C7sF=W%`f0mG#Ffv{l|GTnA<;Wq|fwdbIne z+B7nKX{@VpvE=4A>tQs|Y2^H-lg94_Q}3!0vo&0EFTIaNzkaxJqnze(u9^{9nxp*w zs0d>lKJdTpssB1^3tkr>GeC+&1A)xhbU0G*gyH~AGQc!2@RzIFgxDD%%5s7j}9-*&lyYI6%e!{FT&T2PM~-TgbQ_E11$;X!h&>~tykOeZ1I%;%`{iT zJz-C?yH z(G%2(;v0Nzv!n9)Q-0ORv-n5C`HQct2uEd3(tmbkfe3x~e==~e#MMlE^>;`VSM50S z9TiNQUihp`Zk?wdQ1nUDoCzqhIsEi12ic$n-|p5(@+hcoxxHOjU5psNlvn{z) zt4!yb2{L||%uli7T~jmW*M!!$ztStz(t+S1%8B*Xy}FoGo8(6gxFZ*&6Lvl)6Rw zGr1AjpKH_Ula`GJ5ioaCh7{PIgB7Pjr4*dHB)QfH#MbDNp+c%)z8DNA<3t%zEU1Ut zb`>jCqIBAzOXP^hz!ww$J03CSMEI8UJ0a$eSD?o{>-kFCPZNDdF8?|AX$6X{DI9hS zI{6{l`qRUmIlVfB&!<4c79dQ%xAYydZexC?Dth*ztNe=&IPxPXPKYLfC=l!o0>jRK zDFK9XARrWY=zo1bgAB?ptHAmXS+CYrq(rlT_XiGDD*90CzV%iUC6OUi>#(D7(Kc-( zwKRTE$myLil*y{Y4zJ{=Q-^EA$~rA;FULseOx4h3|0(uKL2OV|=08<+{n%eKc^VN^{<@#4^d>mt#%?r5_ezCHMm{>@qZFH2(t!q-yvMH}j{rc2KFroZ! z2>74hu++Z%X7t8t)TlF*LXZEG?x~CRvg}@uUUx_ngGkeEf5OG1_9vA%%}edPsxyF5 zKd=_sC{&7mNmDp8=UtKeQ6*6U6bD@44y4Ry5v@B7lgi@tdrcLPZ@Zj4(PfK##-sT% zJ7sfnHx3UMrK95+Aa#YLv;+g$z%C!;{w>s1A6ve@q}y?5H4S;x_JW0QzaTx(o0_P= zX}9rfPC0542u%lv8@A#$FC}4l*^YH(7j=B-jC>{`8!U_QjjYgvLxUU+4Yz0zieJZw zV*dv_j`4t#AJ`GhyXm-byqJ=IDbpY9I|4LkSSF-hwt3Xk+vCur3(OvPn2wv)Q|&%ZC!y@X^OMMxcsKoWXlRPHCRL#>zSZ`w%qn#M9G_`zVu#4N!nhxbM&D?9%id4U?xzgq~Dbq$l{#0r zd$q&vXsdBg6-Op-87;QV3Agt7z2@tZZM?^{$1NR+UV$&BMd=IqP$tp3}a*FGjDtf_5}) z>HoH|-YxK=?XC*rc+=7@QBxJTUg-6;K+m9>^&jsEwt3+|veY(CFbA1!_-qLbZ*Pz0 zCvy#AGYn9@y0<{gnMMth+|k&F8c>P7#^cRsVAK8{J3g8-yRjK#f%3ttu|Zs2Ne2LwnzBg z4MNK*`IlR`2VG-cOQ!qi+UvwdkGpV@5QjotDzAs~h7z2>#xE%pJ82v%jp+|70sFsJ zx5{dMJ}aY#wNBszin37%%POtWAeXd+`3qXZO=_QgZZy6h7@Z2^Xix%HLop| zXA+pOHm=vv!v`ze{lfZdJj@C@7+b4$ z9f3_chd75y`B{b<(^FDZZHCyNKI4p!Vi);68_UPN6xRJ3b%|1Jqi(m5@7_~hQ~U4? zN_8WDN4wgyqZ^Nzbjr8PmaqrsklCU&r&YMWo;da1(&O?RHO9^jrnlTt`0V9-R_au( za;Nb5RQjE1@o~SagC8!G=&X;uHgFzxP$%Q`n3j0MHEBK7<__ha{RKJ~52QY-lsv2J z^Vm+58dFLl4DMi>l_Gq+V0jpbI!g*wvf(@|?Z!R0n9yNn)4C;!tAhT3l@mQHWlk5I zBj4$BTC|`wKv-OxIko+fC%Sgt#Y9`VtW?D1KG(9P^>@$E;#@!lIXJJ?J2kL{9*tz@ z-fVL8J}Oas#GI~g*jTdmgnND4xa*+)d(>NC8@xuTxAy%}NMF(KtZH;|l)=VZXX{Jo z`ohP%Zxz_EB=nLuTN4`o6Q@^w7G!IutaQajJr$-r4%F_xa?OGwWk#*;b#K9Bo<&K0jfr#$QD?W49I`X}0J z+0tut@8P&eFU^l`zkc|PgSJSn&XC!D0-E3LTH{GOPo~?r%BzEv>Tk66H}+n=7m!ec z)arkV-&+61Hv(rX1ac&o&qA4pXxN5 zLRM46G+knJMR*f?vxq92v4010;8gu!vuSzLF4izT=D~1OS$w&C%t&@%oq}!W2Tbky zp8k>)*tU-@Io|m{%NpBTCXg+6U3H`OWvg@;htvr-Y!DI>S1jMaQXk~CpF35O32WmS zSy4s5l$^4f3Tw#0ynSkY{E2Lx<+i%o{SJ~%B3{qCoQ4}S4j9d<=1lmBP8syhHl$H5 z1}wPw7R849KKuGCtnSsZh4MBW@~WL=0yE)bkX49QfXdUybn0NDho`pR95ozCraedk zC0l!BpDA(-ogf=Xw=qgXPq65T?VTo$8Wy>BLAlIJ4Pmb3N~^4VxG1uxJnM9iv-mZD z(8@>iW36UKwXTzJ@|uFI%Kb-r)Rp&mQ zrdwP{8_q{?mO^@i!r+)4KI_kyZ#^hTb_B)Nc$NOEuE?Igc*GHNCJGQ7sJ@6^1=%4; zl_#+*p|n}Bj}M~o@~9(uVEbiWtUIHPCGJ!C^nDexS`Q*NEtcG8n#$rx+Hgpb{ zfFYc76o6jJugn6M^pihI|3sJa?usvet+8l{n;O1P_@=qYF5EI?rtb89iH>mN0u0{a;zx83jxc|km7RW>dWdQI1$p{LnTJl`jdYD_#4ny8@YuI4M8`rI~@$X@{g}v%E>-Pam>0-qbY|A`X=wlR8>!di3N{Jg}`d8f?W1{uB72! zqZ1fG4(4cbZsRS-sAgjwb~=SXSy^d;X3|dna%rOKd+@jJ&oSX?J)X$vDgVKUDIi` z&HLS1%d^TqPEr~r_Y;Pd{8NVE)HT-?v^i1k#X zz>TgUmEQc6lo>*pWB%{-ne$xL6Ah>f0hgc27bfzL6?}In`0hFl20Msc7H!ANpT#eK zmWuaXe(r;j!WRaP+P=V#x(4^<$Qrsitsz)zL z2g2O0x+DcO>5T-!|kbuA^`2As{E&^B2 z5mExtE>y~OAgku6KpM!lQV{7vNfq}B#-4k`fITv6itdCc39mqM3M_`y;OM7QYKnl|e@TT*{q{u5oxyNd#o9~Xb6;n&flj0BOcw7^zb z9OT#$^S?dl7SOZ={gMCJ$=?V>kW>9{N$c4#2cjZcX#3Vs=wNUAM zdLF+(lfSX$gC#C@YyHV9PlW^Ip0-8T=1l4_ZW>~@jSQF< zKB*SB#kz;x&7eg;9C-dvtQni6nK3csw$0`2kE+PXZRy6z3nT1Dov^H}SmC23`06{=L&fSr)p$=0#G7%7G-D8K zr8biU`|YXf>k>f;K=Ohc0p}?X{`ZJ8XK%3~Dky3{1^-XP|K9NBJAtnKsX?BxDiy^Q zYL*q^pP%3dRudeaJB!Qspxt~3cGCQMP@0xj3BDHFlWL&o!sT#;PLzU18Wvqb55s^h z*>sC;tdt7%WLNMc77DhUv&TiL~EKCNlX^51EDnXw^yLrJNYNkPRo{*(Y+~gk(9@O95zd!TbSsfrYjXH zf}U*|#O0Kd73u-$3z+{uh!(VpK+#{&@&!ly|0R8p&BXG$e|7y7PX3h>^>d%pfUJi7 z!3Xm{Yp>2^xv9Ake#WKTPRX=8Rv-Uoc!$qaK-|la2C4{3@2G{@P^rSFW-J`vR35~+ zdO-#K;MRLx;|C+mefbf)Z;`dU7MjZKq>H(pDpnP|a&-I4E$inn_&V|T;(BlC1w;{z zrcM))Ey-XjYyEny<7zw92$$9IS#KnS6*kp-zA4-R$tEoHhIs7C4ovK{qw_rvW)cT$ z3=8KNB{a5s;j1Hs(<~@9wwELuWd`k^+N;ug$L)ZJ??KT}2b0{37xsi_XU4y0d^99z zNlsDIBC3_F~%y)F%x&fJ3krbItbqy~TE@9khCn7Q5lspy2j9nfU zi^j{RBZXTH&m&PF|CBI7b?iU}WJq_2^n|)-P_WYNoPh11N2R>3Bwph}>fHWD(Tc|$ zN3!`kmp%BLI@A5Z?IoM?DYy7HlG<6a`1O~MsBhSmhwrdBVo-~DaK;|hq3(w!bipz} z;<%kBn-+Y14byhe-P!SLC|cyeZ0P~*H|Pv<%-IawBq5;F?tiiMR$)=TVb}Kn64D?b z-Cfes-Q6*SNHgTnAYIDPEkid04BY}MT|+mBbP5WR=6Cyl-sgLF-VHd|n1ijkulqdL z`mJ@TMlVVaR}vVAum)Dpn%9K-+-=u=>M-@u5%YKu7SL z(k^dik5Lc?911QKJ|bdp3zTeV^=2PW5nGQ;iSDYj9_eYMe^9dO`YB0n>f=_l@TI|X zH8F@a>o2IKyI}&!d{TAG%&AJi`N-Gw!u^7GAJ5l|)#28kxg
    nx=k|K_-L1rOdF|##2wyN~ zgW9w;KM>sCSVJR9rPfS_U*S0E?b@8x>^RAzs?jxzj}CZ>C^bmj!q1K4kbN9_j+w7V zk>}L4Un25U5dc~x9C@lS)AJdxpBq@ezk7AKxR!An0)X?kr%K+`d_}3+>t0Jv=xwhI z_Rhrc4nbe{NUY@+Qj)p!D$g?T89*I^^r;2Joyb3hbG{tcIBOum%?AMVGmT4VI?!fjHP@(?-o{jo{g-lH{lapmArLhT%{+d7M{{Ysiw{T3M zOfx$C!#M!)`Bma+rvwm36t2f@N>Nzb7=~4e@AIGWsXw#rvHq+703JW%PS29~{u9!t z4W6Cp^D#J;{=sYi06d51N7yYt=YoD!VvNynvMc?D)BboTBU5zg~NLmIP3)>`czO#9w15NaZ(pwQBtIuxHz=zG~iQ7 zq-huX!}F-xRQ~$@bgoiGO=dTrMZStLmwR~?N4Zm9GRdZ2t-`|keb=1ISyRa%kUEet zUr`xj{`&rP=Ds=CV$i&EqG}Q=NfKL1uBp_5NUiYoEcr~a5WG3N9y0dR2%14}I>+6% zoDZ!Q*0&O!_V)8RUaYIguG#M)yU=wBAs-`|cB$$H?b!S`R)2=>Bl2c}kyMfian$5v z`qrOVc9XWu$Li~uEAC({dW+62*4pi-BR*c!yB|uJ=-T^#%r5)s<(ffbH`@HQYr}GS z0iEaR-=$m{4y6A8^6RQM`~knhwz^xH`!{=7&$QF8?&Nr(w0D*>hsy!Bs`)Z~tI3ig zB@MK$M_l6-)_77Bv(#>b8AAmM!cL&>99A{c$v21$(xR-da5pNh26O3H!nB=P#@Zxl zDk)Qo>e)2TUCQyyYiw53PKe?9A}LFHRb*yklgCp zOd>Gx89r_@KK=;Jdk2OfiZ3~0+E)XGZU+PUgIL0>rwGbRR%=q6VL3)SBbNBEms*TE zeWkVRZlRgl5HMG^N$=Mf_O1rf5os;RW0qv(b_b<*KM{3^7gmNRz4HGkXLz^Tz_*Vl&5=}k^{%;Sjz;x9gV1fh%cAMA5rD9vKbC7)5m4Riv$T&RsEinEk^F<6F0%xu0hl&&sNQI?=1uXu=QocQ?E~@~4Gv zq%KUWYriM^z!RU&xvdLZ@kQ83f3_Muz#NNIz{vVnN#Wah;PCX;97#9XJgn!2Y!Tny zu{5h~de2afBc&4RgXL?K8RP3+^l>qx7M(tJ{{REQ$NgiPE{f{zSl7>pBhjvw>spq0 zT?$DW!mE|XJngE#5^im@-ve63X#`Tsc^r1{KHf^GQ~cC^wbD;#tY5}sztZ83(DNe; zkVqd&!13jbcis^$!0)$UV*vEY3(~ZebuCnRY<1w(T~v=M(ron$>x8z6Aw;;@76JjE zLU{UUMjn3B!(y~E%Jz41)QiHXQ{wH zopW9$@qD_)+(bf)9DD>`K=tR=xm%p)rpIz|gf-ms-v+^Gnk~AYh z&EvV!;?>?;$@f~MjAM;cB%JY|YVE!r$>#V*X&emsvJgE61$ivd7PxloSfAFa6x+LO znN+k@kD)B1j9C$c!zzJ*D?~JEF_jps?IELW5(|jJl`x^t;3`YIt8G5cTicdYh9xe< z*E*){t2b2>*x&H%%G$Fba2cRxRP_OuC-oK8SCf1#y}#GINq?@~Ad1!Ccx}@;ECa72 z*kc~#n)RCNq^%nqm7V5%54pDYG@!WX8}hBs?Hi-gVUg&g{3u;U^RL=mA7fhC8QjN4 zkmL7meuAc0=5P67A^IUTsC6s8f72x&@G3vFZx8&w&(HldG+71A;cIZiBHC}lGyW8P zg7ls&u1G!nrmX{eaDG;~mnYTDM)wjQy)=jF=8jZ5lEFWSZD3e!p8hD8q5!}Ocia3& z)vmAe?;{GPWXZ{PKKZ5`XO`lN1=)M;`u2Icy!>WA<4{X=t4cP@bvq1$ho2+))qE#G zMi;;GrOGRt8cqGu>UQs_-Cav&t~}V*=z`1K1D{ItZ?*h6z%6~Po<8ZAar)PiAOojL zc_THVx>iDGr%kH-I*tBIJyPQx-sd0Iq8k4Iff4ea8^sh8_ZmL1_&w7zu zZN^9EMdr3*JxF-N!04fW;SIJU9enoB=UF}>@lS>?bn^z6;l$JKSd82x%)}2&R|W}T zu?S}L6(m-7vuACz&*`=@e~_b^Z+6FaPvM)kw(*s{(l;@?FhB4UUd`b zrzhs;2Wr=bX*;6V#^ox*a3JtU%4^!b9)*v?m{$#%GI{8BjEcgi!d!PZX4B!_=EoXu zAHqilmK3&Zseu*8RVpgmumE6;nu1Xq@>na5jh;nvni49i{nb(Q6c1Xgw}=T@8|EIJ z>OpXaqPhP7fYK=!p&-?zIIP&GBcxJ(pW|1S;2sQ8kH8=6QpySKW4L5z=V8bd3q>WP zZ-C}Lr9CRZok#q$m;U*u3BepgFXTVQi;PHRNk&xQ_Xnt{fa9f9c%7Ska56vlw+fD3 zMo2i49&>`;l+w7%)45AoPf%9`8f3P~A~?VVoaf%Dwd>0Zt7Bd-XL< zG)mt=u}-L)XCYZ|Iqz6rExTECV+;-RJb-7X7&)zr+aW9&87Gr=g|d zmn>O4?BcI1!3-Hk-XjL9`F>T=9Tk8dh_1%&!aJep+0JVZQ`6#0mR29Ta6b=v&Qy~$ zjYBz^MgesP>rC?_&)-_J_HVNBBq~sU5{j7wOa{y!!n1VERIX>*s+~qwQV3|%r_|Sx zYu611wR)`p$esT8UZ%c=(u8{$1zp3wE5yDf7{`k)!YJx8ao+w(-{Vc^;!Jlo8x4j*Qu50@?<917M=CRG_9wM3vqKf63D})|dFVLwRXYk|lu8ZN%j{ey>&^1ZF(qm$_(z*Wt zSO$JW?mK$d0SiHYqFqMFiK;@u;*&V@0O~&w4Rg-3$8aV^c&=hQh#VY))O*yZr+qdh z>9O@`L+kofaxTc(^PCLyt~bCwJh$;JvsqlY8k7r?Kj*6d0C;=)*Jp6cz6KcTE@D@U0d(v{`Q zB805cCN~WB;;Sv(kvLay5S$Z^l&>ZZGtE}G8_<#TKT7keSF+}b-wIdvS*$k@KYD*E zUz`4VAJVfdF3BECoU-)y6?DnXc2cVfkWxn^>Bz`z{ovUy-uth11pjU$T_GUXqig0CJ5vcRV}U(CU^^gsB(uq z5m>p#H5X=Tpxivsqkw;i{&Zn|Pv=#Q+#~*4r~UOzH_&7Ka(~}dZg3-(MY0r*{{Wpq za~iY_fyn2rL-zPH#H;z!TK*26U?cIVl%XWiQBqFiR?LlahX99OGwWIfBOk!H`c@-K z75Bz-wX!Qw@1QEBWCtVA*OyLKQct1iRE^qesUcMU>i$&2w0{czRSCU^``@9c+U_IW zzweVns=I=NEE2wgynEvd2G;c8IVo;QA67N%A8JGW@uhu8MC6 zO%SwsZ=n(0Tlwh04+9_(_-3);(H>n=Ycp{gh>Hf0j=1y{>vKZwGtA_vWX`(N#JXqv zEImI~n&2(6TcXJOPs-j_(mi^b@1NpFh1Z|AvB#uM&3IMKn_22D3z;O1PItzgK^>|7 zMX@`Y8{>~fQCQA|Dp}mpHD?yldsNj@ZAVa+&ENOeUTP3A^5mXH>X6Y2XJB zA$25Uy-lU*%XxD&cMmPRAZ1yA$>agG3);TV}|6<51X_|pmM&v)~ZzFCc2|J zMx%;vX3p!t#NK#))?iGL%5qOVd9On7s|U2WAas$JlaA)R!$;OFv|W}Ku}yZ@O}rHV zpF{ksu<-7+Yip&*XSzt4 zW83AB$1CrGde4doewpaZTsABPmD#0O0?O#{U|mohVzV}b!=*P7Sx zRn4W>#4S4Z=H?Vp42(W}Z6sg=@6B~e_Bl^v@e}Nh7?^PGI&)S0z&P}(UNg`RxpQj& z0NOJ`ks;b)INg)hxE~W}+P;Nkmo}QM{NAyC9?N^Fe`vkM?95cAHynYT-80g);Id};fpUi7vbuVF*cs)j zrB*R+E$WDj9HN>>N9JVpu1CX6!t-BcAAav>KIFA^-`V&a+)4Qg<~%!T8yo3E6qC+w zcZS?|Fv$a_O3mUU?Bvm}viIVd*VrcMBS12!E1vkDi1aIECy{bGeGP5dtkJ|7r#r~x zg#x*Mh^3oHw%;o@XUWJP-8IdFsGUU~w>8AY^0(5%{07kJ+8p;XFj(3nJD)%u$MO}) zM)x<@dbY1Hz`l6Z6}p7;7Z~;cVyw?RkoYWG#IqSLtw72tDhHb6=O2Y~cGr@PU2V*g zExE!nXFP@f036q;PB-S4nZqZ_qLJ=89#vvf!TUP(sIF%(;#fp(@$s{G$0ypC!`@}w zaTN|++Q-T+?$vtc#@;K~JnuR=P~}G7y_!n*y^cv}TDgWqvX#EmG5LBi7(buhOg>R}_!yVahsLaP0Opc6~l?lWlE%##YWEWO)5I&HT1Zz)vu)RM zuSX-ce*y>RSG-fN&uao`nrF?aU;L^E@F#Qa-}0;49*YKxrblZNe)sQ7DE>F@ex8-l z9Ab=A=B;xtt7_v8hjHNuA+@^uJ-lliF~PMJM*sr6S7BUVgdS6>${%4KLF9kDwc1ys zMxv<>Xr6s|I8My@U;YvA1gRtKI;(qm@Q>+IE{ot)>UAdi-$Xz5TJ!SIsl!O$_st-C ze!gR$!~X!UT|DwJS3CC8dt*0zK3zI@9{PrVlGmP=k9W(sf$Lh9 zJ|cy?}I{yH(_$w!FjO>5fzH|JAHKy=XBMqTK5Bqm=n&%|( zgpHTBwb{l1W@D518mDo3+Mb^+)X}O+lB)c<#~}LADzBp}mTOjZf7*TxLzU1hPu6LV z^`;$j!m0OsJ*EEu@Hb6lm%|{{V!4M+@A2m5=jk;Z4{Nu-dXWToX@eb|&^veW>P>>MNb^ z{3TidLAONH1CF3rjDI?wFB)hGjoJ>LNbW3_yz|awhs`h{UuWm?rD?HlB=Tdy>GSjX zQ|6LEsp$Uz@SNx;8+0on{{U`L)$KFJ8b$2y3|cj)lglI)wTvkIg?UtOxpR(dZ^Ke- z)w&b%9+gT>u~_cxZamw|TcVJps`98Y#;8Ziyl^Snj;Vf^3S+vN26_h~tJ=5=2p9q8 zm#_P)Ow=PeQ=Id~Tvm*h^*JvRwmv0l2i1xH0LLr6_;nPrcv|7Yvjn<3h#AIJYk~1M z%kd)nujBs!TS}v?qFc+5@;W!K{{UX0DR$Tb)Q|U^3Td^wbCQuL{{VQO=~>9QqhcmG6?u39u;G&B z93vRVqcA5mrmcCmq(7}-*{zGAvDTa$HAi*kdO`V`Lf7B)?mq)dn0%#ASvR(0 z%Myo`nNCS(C$XzPW|7o+mol>Bl>`C$Q*}9zxh*2Y-ImO7D3QlgoGSh`#@u+9{wf*l zRz=+d1eQ{`kCDL7VmfnOnj*%bPynN_uPxIxw9v0SZ>ThJ%2eS}L6v4Nfsy#ua+^_Y zEA9BP^LsX$Jqc~J@c?OJi+0?ca(@b%*GsgW3&Aw1qq4tB!|-g@-XZX4zPE$Rmvcod zDsrg9`=Im(gNm6i8%&e;j?t0EYUf|4-N~jYgv%_Ef3$gDrD=^u z;EmAClP^-GaKrGfMlEXI=1q?1ufPX9HC_@jkjmI$(1Xnq#L5zB>_fxGYU^{Ux7MP5 zCSNT2^%S~~m2nD58!KUgk`8MDE)rphPRIWKI6qpQHkXfS`EoiQG7WUXx{}eH@}T0# zZmxtkGO&|q+FSGQ5x#pcSJqErEa zKU&s#i%Cx2MsUGLna27aPibom&|V~@2|TjVG3YQ3c=}dtrIW{Z6#LY~k_@&< z=tmee)SH`43&fXmPam6Oy`eFc2Ww-p=cW%|dVSunqiU>R)9fTx5i=D?!Uk>1N7ooN zx3;{rPg5^tWxlAnqIf+e*Q_>2)p%>farlozYmK2UBLyZw*|F4n9Okr)mc|pi z)Lk_%Y|oq#R1tI{{W-Q{^NftouiZekJhW*EB&5a zmBz;H?O7<=Ng7>P@oi#5bKuPm;T)HQ%L@Q`1Gn-O%t>cF*Cj{aE9Mip50@DxnQeCV zw}MQPsKk}e2By8#jnvlnVM4O1uF!wG#eEjKB##Qp+POU&p{{V#3!bAOEg`_|B?3I_{{{R|X+GxQew$louoc*Epx#!ayiqro9 zgy&FY3_q}wJ&c6o^shF$n|736-(#|zlGQUrVGV%s0)tC`cKAWbksh z&po{>megYMd|7rCP_LgNs;^FR0R3x!#C|cpyR=wllTURL1|Mb(p8d$KPUlf??H0RY zrb8ioudQ_9;Wc?lTTbRRr5hxj&fDSUfi!+0x4f54xxBQ3Ho#Tqc^qe;uTHqIyOEdd zJ`{%Car{z3r1k5;^{U?sHC4UQ7Tm!yy2h+zW%&;PU>{oa6&XK;bV1!}a>=ChIHl4x zS9LAn?J-E|T2;yXmbwit!tNq>=z3~`Iolv7{{TU)fIX_!w+w4UiMh^d@mt%7mMc%Q zV>v)DKMZkL)5RUkGD9+kjn``_?ki`-lDkg|d2FF`hCLgnO6P|B{o~Di&SQ0nZazoQ zV57}JHe`4*%xSm!&4QB442iY9K?IM-n!9Sb_>F#hErfsi6;#qUjb~bAVly%WkM{*f zXwKd?)Xw9@tU|PvYsyXbU!gwA$tB$E+W_XcpAJfmdw&|VWQWTHOb~asQ{4O3?2O7Y zo^!=>UJi&$sp@h#1QzH1`Zdbx##nl{^8Cx3IVwWuZk#9p@GFt{girQGu>^5H^flH- zA{qVK9x=$SN8%3F(ruA&3r~TMp{dKOLDz~*ScQ6WM4eG6?!yCq^cflHheK080^%?ao zL1dB0+xGxJLBPI3@rtFQ+DSH@W@AnGwvC7#vZAcZY-5z};Etla+L-u8ud-6q(w-Gg z9M5w}wOgBrRwtQ?;|HF#hvQ4<()5dY(r+nz#x8jIhB{V_vZ0YlIXSLt;;b(=jittl z(lcB(5kV!6-)il_PuoUG^c7lu%1YgeJ_5Uz<3*XA?h7$h-Pu4GuAk07#k6`?A>gPd zwb5o0MDeejPs}mf-ntOX{{V4IeLGj1i>*!-Wz3I93r=3m7d-Y{WCjHORa*Z5SC3D# zYl%Q((liH`Y;Y(owm+Zb10wwrplIE2(! z*794}UT=+%HNvvwbr?1DV7&0`T!~=>dbzJM__cLsbzyF7=0{N=;afQTE1f7tSC3P_ z5t&rP(Ty1;X#R(i+~C4Jd)B;G@(muz;EG4{18*RbG7q(6Tn*8HD&@8GHuk9(*wRKF zT)7$ijcm?3b5M6`;OEk#m5i4jYK^0&3C&E(Oau<6ifDJM6v$1z*MbpcywV7rhLGQvS#yb+F^RyCd6c99@h z7tK@5fZ&=-`=NZfFrvP4TRhWkwP8NQaDS@?wzMbPsRBQ`YZed{0_1*Guc4AP9sv8N zigDZ-+gmVbt0gH@lpQ( zi;sQv{{Y%`)A%o1z0kZl_RAtc7mOTr1Z`fm%y_K-0L8|<^yB{kMT*kWZW`ag+mPYD z+Sotv0bG?Iy3UGvo~xwz`o~qcNUlJ&nEwD6A0a;CyE};Ou5Cj+N=RSb$h?7{L;Wk~ zl-32q2(BWB9XFzFWj?jkcz;vWq15c{mh78pT~{Hp2q&d+Ikb+oPWv8*cVh5m!QT-$ zR~(PVsH3EX!aA`-h5-H3S)LcyUh7SpNt#Sfvm-<^%SK#e4{^`uTNQ|jzNb2ik%qS0J1jt`UV|Ea|Kyg?il)t0q1}PO`9m}11qjo zfd`RK37&&LO66g>^4Jzo-1WwPI+I+znRIn+HJ%CDKp4Wjvf%N@Zj~Ia_9`mw$msAC z;IJOV&|W_7zk#lE!O~q_>607dBg(T8nMWXCfA#C6+MsX+Wi1_zTpV4QUg?8cN*5>P zJq1|C$kfPoj1EmlaM-kyJvJKRyg0Ya!m5H7wsjUN_bP zITmL>-Cr{Y{{TXvPSBLSKQtxF?P8^|+WbC&G>;H?n?jE(Zynl68u}{&Kb9-5*0i?N zl3N29kd@x2r%sjUo(|Lot>VuUU)jn;Y_aWZaB+}7m3rQlBdiwEO2L{%*iXI%b>Z4` zYF%yl8C1L@%@fLQbUXbz802sWbth>#@A?|Er0P?!l_c{vrroMX%rl(+99MVZTVicl zl&Z)i1Lourisv-jcT1U>qeuA{{a+o=tz(Otr#BevZ$k5mkz&tPIojD_=y)Jj^_A>x zaFKJClMD3~ku9aY((J=UFKqCC16=ork?gevl^o-{Yma@Buk)i_Nyl1SdZ~Ce|b+{E7KcW zw+O5lZ6iFKR~|AHXwAGt`BzI2tJLLQ%^XGjw6>a>$2Sg-qAn^93A z$J!i>k9zmrZ%~T=0K^h3T$tfd*dC;s@P84XH^e%FXRX7Iezl%cH!M`_?;B`si@c~o zwW-eii@%r;;Y;>DGrymuU35?Y9cdkQu03n%Igcj93u~Ci?+!gNOZG93z0cCEMUfsc zQuP9%$&`{aCV<5sB>q+H-T{Uvye|~04*6q`c^vfJit(H8^<%wzH-RJ{@QGWI{l)%f zxU(vbI9W*@7*xHoJ5!`zTv|TZ3acT*Il$>xZG16x_BfTLiE$Vh)Gv@b%QOL*jsF+5-&oU={=t zIO$nBlw*kYbtQUht^VyzTd684r8Vz$Jg>yiUFdqFTgwoXqJ?QOlH;aGJ$S`Y{@t~X z$;m`nRtECGObyJ-$K(xqFOPJaZAvE8ZEs^P6b6sf4)x`8XcredW`V?zB6*iA2wy38 z1^jx7?8Dbw{>?kvZn_F_Qm4<~e^U0DuS;j*YoVr(-A|Gg&kcjSr*3jT8tOa)twVpT z>#|$IE#P>J`#-wm0Dn>X*BJg5*EIN7Sentk!~m1cSrp+tu)|oudge9$RaaB93=Gk)BkuuR)|swf z{{Uv&VniRhUJ83v^Q&ISoXnh$?pKv8&XnQE;P3MP05i5YO0kzRV|&#dLI!x4rhUeH zQ^=T~`1oP;QAd?27&j7Q=s&`^r7vj~targUyO;_}NMmoq+*G7@sV*!ET<&*fkGHK2 z1VbAci2B!-=PFVC=R?p_i`a|~v>JDko~QGm!k+&Con!Ubj_9d!8A{5(d3ujgQ!Ig$ zFz9LiN_rp8lgW?!qLRiW?uVlrypEK4tUJ@iykFcDrrs{`ihtS?^B23=kIg57)v2pn zL#G=?je@_F9E|gx=Cjr)k%1Af2R$oaRh2KNNtG9Gk`;10`d6ug#4JS##oqcHwXUkg zt+g7rB!EPjA6j2GxpDc`7;PerKxhyj_eD&xLOB=*kX#MLT3;_!sh|+L`;K@}=!&nrK|8@H?Le>d@+QShRM{GC*z!6ATm?{e7zv z^GCa~SYAlb34ZPxyI+cL<@llojf$VME_R%EHO^gXZ+WE~OM-$(3ObV9kVX&mHT3kV z)SWoXzUPfn5L1$WBQpC+#e_N7@(H}{H(u8he=#k4Nr>vDwNx2>W z0NE6l7Sp$kwSW+ASYZd|_Nlx+{$2jASd4wJj(^=1HkUkq@R;kbxD2cSf8RCJO*tvu zbYGzplU%m%{$;76nip1;Mg~p+$*eyJ5~*~{871bEl5y!>CGEsdaG*Nm4hC^pejJw; z@T`v@KRE(II&t@XzO}_FQdBD>yBbbh(u00x=xGP=kNfJp&*B!fB$~uJbXW@pHyr-} z8-Mky(&V^E;gU_tLMd!#t_EwzbqU6&eQ_#Vvq>DHd%0PL2)r37APRh$D|WQlEOcW-enKXM(HKknoD*PLh?rBg?EFH+K>-R|<~{M+yVU-zMpBB@bpHT&Q>nByZduSF5d`W$Akq{$+{?Q(SzFxny#8!_!?Drt!N-oA$wQ~Bs z`Oz&|geo(AIQ|7a5PO=}wzd($92W6xhBXSKpa4?oT68*9n6{0)$yo22Q}?8E)SAA; zD&?2~^`~1E%9R=`Mfmp>I(PTx^<*DpLOya>`kIZcm>=D?N1>^_^1~|=k?MG-#LDsj zLCF3RI@cCk57kbJw|`VnIkrX^rsG(sHVD>IXtJfjyA`q{{UW;xk$@^3HRwt z2?hs69`qler$zRwfWAV*EQh2~rd<9M$t0dQ^CLvA2Qiex57M2Tggg#MKZR-N#QnRk z@SN66Y*})8)~1eecc1o63vr!|xy7_3^z&r@0MJ&Z4;hbRAtU~J$^QVLEh@Rd z@6Bxoxtm&?=Zf+F01F!M1IHiw0IT{&Kj9NV-2VWlH(&P!QSoQ|lv}J1PyGN@{V8+d zWEkMzpZynHl|B}2Pjo`k!s{x9QLv72S2UeTEw1eu7#4~^;ZID1nx_qm%x_&8R7T)2 zewZhLRU(1m)fHTS7IB`{#G@G_t+^&|!>Gvw>v3$@nkNjb4|?x>H{;u@yOp%m!Do?x zj#a=Qo`B=1uP;_FjV$0O_NE53$xxT*`r2TZ8Oq;D2dJ{X@FHw*s&e1G-pM^DiwvMlQxT{A8=DBF*G=A$~l z5+RCGP?csuk&#k9u_B1EJeCW%W@%UD$EAI44wUP-wX@XlvGT1>66NXJf5$=p0As#^ zL3=}Pr1SiLjVoH*5uZDQ-B46ADsBzg}v)UFreJiq2iF)kx`6jKR(m z&{=g_(xjZ9-@oCZCC!tYSvN^5VNX-VTbED0mLw}?V^mN?q=GodZ+gy=wOM0kdvP2< z9ovsMt1xRbCQE8@q)IStBww72cK&r(JT$6CXnBQNkenqO{H^%h&G8w$8U^qXp&mmL z%19s%yjPd_VQuuyOHsAggWKBbwuU(7Wjjvd2yEA`$$x8q6xPMmi4?962LxiUv^#rE zOT+QlUBZ_!Ld_G$GUR1XT=x2k{Pw@AJ?ObWm+!HCuN06~hcI5uGHSp6lQe>Xb z%bgf9yk{G}wR5FKUCwutDwOIkcb7vVeNG8=GX=w(vI2c7rjLDZb9XRCAmAtrF`i9q zn=4n5onlk+fym~neU<<&WMHUK6srNzw;Xh@C!Q{^ByXwcZ5!-t+6y;=NUn!1_gfj; zT=u78&#hXdVOlj=1F$2Xl!4c+TD^}@XA5qiAW&JGxjm{QrTPB=yt8D#uedj!>^S8^?@jG1(?$^hH#@(j1nQj0^Y^wq5pX*-aZlb&*VSC4p!8~zf_O;q50 zrThDu>&qNIwLV|vMi{qxQT1dShW5|3Hg^~Ve>$ly*S;Q1llXD1fjgp#GJg`-uNIZE zhn)Tu>)#C|&YN@*KyeIw=dUeXIh23dC+Kg5f4_;?%`AgSoCj=`AdL6vR}TXnD>C96 zi6LRRRT@Od1Fy_8SN{OBl>O->W9zur#7-2H5|p~XGv9Lewh~T9Jk(^WO3R!Lz>m_W zlTU^_Stavi44{xV9(d2KRBK458{K|kwCD1wr|daCCu{ZvHr|M|*~#O#rFpiys1FZ# zhf%i(&ue>awWCvwyND?jJ)YPbTMQ2g9M%fmq5CLV{_$sP{=X28OYDmrDXVC3wzsr zQDl#E+uF3KbH?JqdVLA5+S}pohvE$RjQEzXr2>H>ByhP5!N8 z1;M+JMIe)K!ZHR>zm9ijn)>LtxTcxm$vC!|p{ME*-CHfVCOK6A!S8`nhC8;n$G=*y z7m|`>T3fCU1S$Uj8q~MBN1avTg!=rZqerkNL!fH{Xr9OYWz;O&g7+CU&@A^j7`I|S z3gY2MiWqLRt9Tbvl4*9c4p;+|pTm#Fy)(mCcDm-2mkPdItMW|=;Kt*qub0N*E7PxP z8sE=j)`l%AN=_zVlJxYnZ_HEA_K)z{x8^ISw>ED(!5gbcpLlvz(+ZJ{R+KW_T;qAo zFGDF~l79)p%rzE59?8AA6`2&b0(l~PbC7exbgxO0*;r!&xhB(Pxz(=L05TYGPB_I* zI|J;Xv~q0e!{Z|~@1d3ihyGb({sl(a5AXr?Z~nDxO`~4IxpiEBy;Wqio-vZE`Pa|Y ze#TstDcSkl=!JO6X)BaD&M=w(0Du1gtxXIG05Ws_(f8=g0> z{qldtug7?*fm%F)Swm2a_GAA50W=HcV}PyC`zEfPdbHoTtVF3fElmr{wUtytKso1* zl~^=~_{|$m@-g0?srXmXQ;)PtQq<;emqZ}*JG)YV-`+IvwC%j}TPiT@M>{<0;s=GL zym>XDc8eTQA~G_Pe)co^*9EHhadi2vt>(Ukri20zAqDajjC1e$SFw2LZSY>CK^!9E z{`wWy!OGvMguVjt1X{{zlEo6- zJVsCifSsfN0IyNS;jH+%v^ABCiWnAP2h0$*KQUVxh5dqCJc+Gd&v-GmX(B4be|L6z zRvrGM6~?~zQM)2A13AMVlLViw6l&rsCwJ4gUiyULN)Frn&u{p3_KQ0+q|F`$vXURv zmgk?wyV(Bmt^?uQGuuYTET-I$TC1te5duAmX}t2ULKnEG4v_;Qb}d;W~pe$bYEz9Bh=tlzKy1}$BVo~Spw{M zSB`_^6_4PgVdDKovAKiGpVGSD4hbJw*Cbf*UgzdJ2+jsN_pheHMM|4*Tif(JyhNoL zY0zCqPKwnMM~zr6PDt&=bAAh#PiQ}RWWHjvg20|f9GdTSXme}>kOmJ?T!(@*t7p>; z!e!p_y13(kk^U9tW-Z0kj2pkYIMj@yoV?AgLTx#0E;iXI9#b5H*NX6?J;A_=-Bl$} zl99$hZ1d~XewFmK{C1J)Q-w%~5fC^YmGi!xs>3$7Z>BmTr<`ME$KC~$@%dL~78Aoc zbIYCD`gAgkqfvI(snp2&ZLXOewY9VXF62mxl)-HL-8bPhd>IHN<@+1iyYEIN*fc*fhZ#784W1nAI+0Ms0 zrfGOwD|nK`7|sKpcH^sF{3JGbB#p=a06{g~c#c(TTXNue zV=b2KG0k}So1CK5R-I1V6nSby=dW+z({!f2o=@FJoDA~E&{jr^X*I;sYWk9abvWA` zM1QD{vilR>s#z zhJHo1yCsdk%8}U;bN8HvLAd__D)l=34+j{=l8hCSK5G^HL*a{ZsUsExBzLP%FfpD7 zPfE@V0~XIT?>8&fx{gqZT3L1~mMy>la!<8ei^KZL$dCP#ctV`u!rcCpx+JL%wP!Sh z1WAyRdJGX?MxPdVSe!{cr__}H02&)f7a4NeT~7{s9bW3)Vj3;zJGsnf6;A&EM!nK> zi;;D2Bw^-8Lg-f`2iCpo{u3VwAti3Cf%e({HRqocwcRIJ(_sG3x`IXyGKj+>FEy@3wDEtda5=^gPsW+!+vqN@VOZKJC)xYiV_a_MJp0xKz1)`YtZNix zsq2nSI_mOE$h@~i9OrOeO(rHq0U|M4dO^kZf7z_qh7b?uT3SX$=O5iQOK@jh2`5d9 z{<_QC{eejvHak+qk*3H009|GO0MG?T8O&oQwU1IvPZfX3M!6mSU;O}9g@|SFc-nA( zv_JMiS^g^+_>VvQ2mb&-71?+@Ot;khEqxsPd3PJpy&Gt*YKrh?)U9=M3eMqe>~#Zi z5t1V;USh9SA2&*6*tTS_p3&rZjEt(%5120;AEC{9H-+@S4?}r88njFn;B3sCCVst* zRPnEcG@TsUMPqn8wqqn{IpB~wisYw4*Jct(48~pT=$=E4W*^5?nALA85fK;anI>i9tk45ia#}45pbhv z9R2F?tKp`amorwsUC%-eGt$N{o{--CnXUwu>n7dIy|aw|RoU5VXTvvg*-aq1wFfat z8+NZdyyEub1Y>(~k9z3jnlWvrq?S>cCfZ$**s)&R9s#X!bQMJvJyS+hYT3y59}mCU zd@&@c1adRZKU&tkMQaclK^$;Jab6j})FRL!iBY_wUm<`kjGq3r*2{Pv!GC!d zn7|#R{oiAX^lCuL|2 z7TAdRQH1f9s#iGsOZEEK zvv^@Z)HQs6&%yp?yjR8Tj~4301D~}&e|q~&uUu_U@W=E#+*j~oV_cEulao-kae_Vb zQZ`OK@lnWfP7Qr3JjOi<+tQu9;P$AOAY__cpI#|~Bq!x2y?fzUd0@On86w^@{{W9Q zc<^Ji*Rgyfc8RXv86Rgs_4$o?xxLm7rMn#%Tl-|u-pdI{;|w~=r}eAYs+Uvwu?E|; z2~_^Ia=$y9ZYus~&^G#$!varlByMq${VP6OQ4CH&Cm0oEGpeyn08~RozxoI7rhBb-dQqn9La8Fg_AJCfAhkD2g=Z?6lc8UF_ zH5(hb7(sJx(Ua+%kLz8TVrv%{dn=H6M}BSAfdxZkwnd}oq9 z`Tljv_-9}K)ztM#9Tq7k4=0wRfw-@K!o6W4v^qG`?sGg|QH}Qu<$3E~N8*TWJSDBN zXcjLd&YO&XGyI1ij|ETXUwec}VdAfMWO&sR=65#|GoICM3v$z==N&80$BTSMN6Vuh z`|#$R-Z1e!yhc{ib|P|Fplkv7*6J-YBV$q3+epyur;2|jX(gRfK)Z;`VSxM%SMXnn zbqhwg2II(%MHqJDeoo#$$G785p585U!5UgP#q_3W;w#YL7G=l3^sL{q9T!WsY4b6= zvwtie0u}&*J;A}OD>_{1KZ%_*l-zXK_RkI4wbjfQ=#3S;0D+eIhGMrC8c`9~mqay{$2@czGT;vFq7ZE!Mn$b|BZ=xa$`PESEv#986as{OoFA#I$#sq7kplBl4Hi6eGyar+V#EIcJ!JhW zrt{`^sl(X&{3rq4r-8|A;>J!NsA&a&@6J1zZoY9nG(`P0o3h`{#)GWMJZ*8XDUdHA) zKo+g?5~HXbe z2T_{0;J+GpcS|~}>QP8q=IF~A^0$1sBoA)2&xos2>iq7PGwuC-yQIW#m-)lOx z$_+@)@3e&8yLj(a<k+K~l!57;rk*pWAA>lT6UbsKbzc`YMnIt~X~Db~&|$wIphm zbIhlWkg(6UCyMU=(4x0%?>Qb8rM4fh6gt(wARb3=VOZyDUvi{?gSWkOE=;l)knVHT z;-0DpJAX?2`xS~)uQ{c=XSYhXB{ds#IbVmcvGG2qCel%!Pzm+iE2Z#k#kJoOMX^AR z^Qh``p0x*pWrp2v-cl5Ou3?ft+NAJZ#4+pMCzYG!d*77$^{=+ev2-Ir^T{TkR(W`} zDNWfOFNg_UJ>5oGQHZZC@I>dxxLH`Fa$F*-W4;d+>6*Rj%LIt3*z(x!#~mxiygNO@ z>o$6E;UnL6=z9FX1wWN>SDhSXX~qpT?%b*_P^Y(Z({*X!wbSmRd(yIq7=<7M_f2?a zouJ&^YX(R+iC{#ZYj8+znEXhuLh(JNmFA~=c`T6$gOG~Y$sSn19M>6j;$23^#y1zb zWt&fIHbTmg5*Q~K2OM%M(3T#C6>{McO{afei&LKqUDf92UkgD$hBcM7eGxCBh6j~= z)j1@#8HYZf{c6SV6{&9;+QwY;m$zIgs~39jimoS1h@iL=N_@uI0pYU1cJwBj@cnL| z#@2z=gQRj~5sVF|{A;00lACR#S)}BWOziwwFhYYr^4b3YrE;%39OI1lt?wDJ+=IB8 zv-t|-%(y(&N(rl?RGf4=p9V8^{#`onA5#99WLLtDTP+h!dX(MP>; zUKvDvdE~FK z`yD$FCB3QyiXnl=0;c;M%H#kWo}-G{K(WWNB>(~1tHw{7dUdahb>$}6>qfKI+Zl$! z<|BqcN#}!I1a_$nyk_GC(;vKM`=_uUok1g$BAvC#c#~JUy4JqQ6u6GvfppJPjO6_V zbYW@L#LJR8Xmi)8DJ9Hq%c94nCtgL+PcX?d+oah(r{{SqPKkrk& z;U2xu`m`VLAyk`JobY3Z{`saIQgin;a}(_v}<{do!U2XW)X90?+nM$M<>q)4#Fj0DaJX*j9Lz{g2=&!f8R^;+CLn zri@CA-9o;_s!!}hKl1x;!1b)CwM*z%YFs*u^cbZ&t@#|?OOxnoglZicKEG{)^Q8XD zhaBF=pK8cA_nod~&vUrce`&aE=4n6gqeu;}?AV5LcNycg15dWkH))E*^Q9x1TmJf> zJM;4%Xajfb7JzejxT&p@*3w<#=}*?V&zre8qbJsyb_*T+8s?b>l_V0#USlIN6$I_d zf&D6x)~teU5-|S&fY%(n@@V7IpEDJXhg!aeRq?!+wuHP7v<4#`5Cg~5SEBe6E8+X` zpZRP*kgpxnNgBLdz)jKz8$;-b+{S)g+%@aTpgM<^$Sn7+O!^Fh>{z3UWT~2 zt?%R2n%Yk_;YZ5vKfn%hE3>y|bdW|o#8aGuowVj6L@LS%BLo`q+}vA9o#|aB%K8qG zr0V){vAUI#_TFY_$rvCGazDNFt~Th)736mc?B%5{5cKPs?KLYHwGA-Bim=S5ddfJ) zGEH#J1EjX)ZiLO{?#E?2a!1rxG%7iJvR1d^bV@7YLv)zz=C~IZED0AIjD3oQB>uRs zUeojqJ6rI@oz#|4NWNa^<&150m4xRK8oQZd%Pg3_&Qbp0;Y z!%vZ(*Otm+_xx60B!rEeNSqR zO-ovf%Zy#1X&IGRpxU_3OJsE5;DFW{dv)5Ad#v zGf<|ZC(Y(@$u{l{;w^3N=d+4vl4#LiG9W9?eE>a(sqFRLen=Ks8ri^Mk>!z}U*h`p zs&W)S&TS(k^&&6{{VPV+D@iOSoo%qusUr--CmHAut!?d+)kQN3l&32=yBCt`W?3OP zc-m4}G39V_2>hy{@l5u|W7|Z^tI6DcQQJQ*dFS$|t?q}MQLwTSSp82uDx~U?q=3&b z5~`zsM?uhZuPZE@QmGmDJzRVhskrtwEj(R3v9bFL+sFahOXM7OG+$~l&Ej1PQHKi@ zyW(-^#|ITcDE8V-BF~t@wpF7U=Z*+J(vweJ)~gi2IgQCvj(9l)R(Yl&QJiL%yR-iQ zFw1DVv67bP@18Kgj>n3nZku8&=)_c(nv7R#jO~;Arkq#jkPlq`mE=^Cof&fLvDs%C zT0;urviw9p;-9?#0CaybTBLJSb#>pO-Gj&}{Y7wKl)keIuA`~&Gt9KLBjWqFV#t5; z4SK8aYs$3t+wpL&dihxY0KYZr&%zU3d_g78PA{<-ZJ((SCmjttdh8(ZdvRGB9sTw7 z#rN9rvN<-yJiWMHe?HZ&zyd49sUNG=aBHXLb-V1&W5XjCIR1f+1f9GFOlTV)+Q_+5h1~d3bz;pnQn;zJv zY+U{0&$T*sJL)X3IIpTll=meHF^)|ymH-`UFx|6P^rRBb;kh1JAW#KN1yf$V@SaS^ zaD?=?U;PNzkx0982XS7n@R5hwE`k35pH=?=(9L<-zx$O>L#GM+WO{7YqH7u1BtnWPdom9! z;egI*%_}Ty7lM6iH!b!uIR5}w{VFLwS9@F(UZ#{`5!@qXbHO!JPZo@mMZLuCkFrJM z2emPa?oLSOoF8h!H)(5e6tT~{%iLS}{{Twr!Z|9gQA+Q=#QQ73rp~7J%HDY3h!Wcr z2jBc&=kc#P_=TYV0A#~&s@YvVyhuETSWXz0IUj{}sXJcjc2{M{mRR>mwT?4fhs32T z;U$o}99$^DKE}O#HX{{F8yXUlwd~SbYI4eiT_0)WdF9ph!?k`WEtjyyi4%PNbI-XaN>RMKZs%sty(rz?)6>e-qq>b1o zc18yr@t@AQTe*BeEOz>J%vS(FK#n($1$tm-9@XA_HhY;oEb-gQQNw(^_r+&jYTCbu zwef2NqASSQ2_ZZU>InXyt#~gNQDW*wN-amL-5v0xohZutnO-IE{-GwVYj|ZvhVf!p ziGR5J2g*lk+3@D4srXCm7huOU7qTikDd3IUTj)m>8$oJ3S#K7%I@<(?BlnmDY(4(G zR!zm-t)GYOhKu3~b{0pvyV@%e#TLcHZDb=7RN*8M_M;*^(>^c}f2tt8PB7?Ygz zJkxEiN1HszLm7EsdsjQ*e-x*Pyf^ly0WOy@%NXq5quQ-SFOxW8a1R)zol=zW-qw`4 zTyL?-Lhf9zskNNYbL&+_+(-D=^QHS#Kl9OlI{e|nSJ>|4mtwRgSCNd7Y(B@1wJR%n z@mcqmuNBg?ZMjcTo+*b^4xVC_+6GC_7_OgZH)$@XQlmDLMdx2?g_P%|W<#soPILBB zk@jY8zlBKmawy0Od*hnZ*v>A@)-$cYt#KJEqByI-;h^9zp7WOhtyy1!v0{fhr;`}D+vBYtVzDds} zxbgo0#@j%ex{4q1B;@+_`ijodFXfKN z1#%b8Ac5|3dSaKtcaJ1mn`}>#=!1fJ_pe61IdavyT|YA^McHVzI;)qvHu7X6e5d7H z{jgjMV+O#Xo#~^+TB0^WBmKqq1;|i6j}LH0iHWo1NQ5wz9HN*l422T4fP6t znwsAtQdMNEw*66^NVwB&ozAGYFQXu~vsSt<)Sx7H2kDySTSM^-SCYY|YcSZ_qJl;N zB%GE#IQ}Z%v()GE_NS8DfEi^u{&=r3@m`UrP2!8`TT)ASWt1ugcImX9gXl$j)#_8l zyTWN(O;!H@13GjkMiy%K)ak#oK{E&OZNw`8Lhht|;De9TBC6lmmhE-~gV(hfN0y)G5YDb7b)^G^?Wh)tx8Ddd;!4I85100MwW z8SXmqUDS6|Br$*nKwOjh*UM-0*R+o--D-Ooa&xBaZOL&AK&fC3I#mI5t2*ZaPbH5P z2ifLz5^Q6W$*3-;isMsFE;Z+AXKz9~SI<+To$arGBeg;h(X(Dlh?$q>T>Wc;)TT(U zuaOY>_cFfb9J2s1$DqKjpj)zlT}Wr@2>dGo?oBq{8=Xez=H&U7_Oa&~X2>jmI`(o} zX;ht`Lkv`72&;EHt7%kePy$8-Yy7Doc%-^OO0WzG0QIULXSs^cXH^Opj8uAklvWol zapmF8LBRXO^Ij}sqbD@=zJ+@z@;fuR(eD1&eC&FJBOaWJ+^#Re*YRBVhIkTHG9|kf zLDYQ9U8Q~gNh-6#^HqNHZ16FVP^o3u_-JX*(**too&BEYfC>C6rrW|n3peXY<=zwb zCFkp2wa#lY+0$FPwIA7}4D*VsDZ}a}Baq2|_NWC0+OZXaB9hrG-?Ee6Y<}dr` zQvIJ!Ubj)>whdKH4)FCwA5&0VSxIQ&W>7QGjzs}uP(`JW`FW2YQ%>2WfC!xr(AF=O zgZO`1aE>B5Q=el-qOq=jvVi{pTlyNOde$uv7S4C)up^47V?p;x{An0*&+9@957STl zO`yonKhnKR!zLy0{m-2!MEgiQoM+|fUR$UC0LWh7Ki>Y8-lj)|n*jd+$Kd1s0aVrE z&@B$*NSe{D?$+K_ZUHzoj<Y^Rt;C3Kx?N$60Avg0v;E?h3Bj5i3uU%xiNZq+` zPI59iuLf1&M-w{{Twmk&zb}Kb>aj#b|_?yC;d|=n_EQxCB*|)-@HkNf@d}CPMB^XZ+^5 zzdbk^Kb=m6WNpG?4%}h752ZwvmW8HvFhMQ#lvkVA%~@5M!N)*B6z@9QA2DJ@L*cu$ zk4_f*gM#F7)1229^W4jEFlnS9IpaR{!-7+FDyermu}wJBYq8s|tzm4F!p`JLjlcqU zscx?(x6@MQ7|ZP;iu$kk*B5On!xX6HncR#q?T*#kX*cUFt)yjhxd<)UdymgG&zr^a zMcVdfMp)95UD0z*)Z@K{W`ahFW2$F{A#=|r3eQ+nu6~3D_9F|^3FyXM?>2K*0aOZglk4p zi_!NnsfBGd7ITnjm#gtrZyGg70z`K+07>$;;10RYc&|~0&E;@}sQQ}V?WCT4ZVgvZ zA?~A(S?JtqO8yQ7({S zjhxK}N$9+BS@J?xxIaTlXLIGpG8;JJJICE}+lD%v_ob3hM+u7M&uC%eifWv-R{V}^ zz14J6skw!r!qMF7xfl(0#B=no2l0b}<1Je~u3!1rz3BvfuT%rI+kYWmSEzVGZxQ&L zSdz>4o>j|vmV0zOWUaIhAOoL5D>!%?!}jsKvTB#;tM`bEbnn0i+w-M5 z@P?YdMA>ns<+0~-1E0g1^-qLn{VwCs+K>GXxUUmwx^|M*{{Y!nGh9J&8OuzYNFb5u zJ*&3(M=>`uLUDm>k^caF*CuGxur(i%rW)$9M|r60pV>FkJ=C$qW+A|59nLxY{{WR- z(X}C|*olN{xZ^0nmAKASj-s)=Wux3``c>V%utt^b!p4N)vj+{veSxR=2TeL>gsozD z+Bu(d$gS6Qar}U(P zBpna0AJ(^_QNs2#r!5|Y7WOg3fnyJjP6cy+5ber%U4B_VLtj0mLx%?^8 zTuC!;k;^}FYB!TAM()-6Cbv98*nS)gcwXG2ZEeK1JJxrLd^0p!vDK{aj5AzF2L~8) zfH@sATVDy8Tf<}&E<+6Cw_59bNu$TC=#$z+e2Je0Br1MV2tB@)@i?q&!{&1A^nQn_ zN>P;<#z^xo8EH1KX}VR;qLZxc8pQJ_Jb+K**0C(pMu$b6;2vb9RJzRF4*U+D+}E#N z_+~9{!?t>Eru*Iiq{$oLmIDOy>sg=huKYfc{gGmkM8AE+q_HF3u7+(zQdH_m-9DRI z;k?|EMh=CkuA2nSYU*N^;O{HkfCv8oU-7Qe#$kH#HuA{FJ*%74ye(&@`F3_wG%>3e zCQP|vK=1ff-7F@V14k{`aKtKwLOxc>N_4yq-V z-Xj!A&<+6{RZp>vFu(VU8^H%P}9)t~R4-CU(w; z@u%C(5j;Bnb)uC>tF&63M71+*be&$>A1zYRz$Ajww}1EzU5eKDx~!=Zw{oPAD^Vnm z`|J7BjFIln;r{@A3eI)sC#o=Y=H95S_6Npyf`8qv1LAg#sA?Kjidx!DGpy`Li3?zD z_OC}0pgl+8^uBaG*+={ew8CQQVe2ZXUdB#Sl4+hf;q7Nu(Vp_vyfJ$>n2Gzi!5eV9 zOL9BnvNX>WU1_(_w7MMY_7 z$+{cws|D>ZrOy!YCyVtD5GA^%jWi1kg@M}5l2ngRddi2x{v)~6ZFLE>+k}Xi!IR4% zU|~W0g?#}v?W?X9MLdqYQ;p@K+Z@dh=b>ua8RMk|3RG;pTihj1Qr1^6b&FD#?!sNn z$7-SMYsz&YJX+1t$jqc}B{z24#d^M>Dm#W3k+==GAR~(4^$i~8{{UQ`;w3I53=}R2 zKQ1crnc1~2r)^luR8h0x-7<8ri6-3^__qc|=bGqX@XOB@&+U8E%V&le$E6lAO5s{N zdA6#B!HkSn&Aq}~$G%<2y)w#cix(_PZb~hDf~?w#P-a^ARjs8g+Lg@cz=E>B0|+`&$(R%RE4EF+sqGRYx3 zrjt25@JJO8fMAEq)&m@`+Xw?bm21MfgckZBh2~A-M=!!QLAXfr3E301;UuqKu5t%l|H}Df?Gmmm>xc=6N zo&g|qJ9y1yYPXAUT19V=N3gEmq_97r2BRyfR+VKGw14I~LTZ~*T3F}wx%Xbiu}Z8- zz+C(KRUK>L-mS0ew|7$NA`5cip-=z=LHxd@yA1kcy+NXn_^2%pmH=6lXFm1c-D<`Q zhlOBv+Ht#h0=>M(t!y0s0NIk)c4bA*ylo|O&wt@sYpPWfTiY2QygxQrjFJ57qS3qq z7N4s&p)JptS~Wu+&rEZi^T@5Ctv=uN19$y&{{ZV%JyXOp!vaTd5c@=}<;0y!G5-J$ zTJu(6m{gLAX?@2-!nCzV8R88lO-oX}+dNRDQnQmTdCA}a0m-f+?#5V;%#Hx=RDoW} zqTg8RT8-YHZKpiig{TJVO~DnLrx`r{RjD?OB**8_;(zz%LtlM`wCF}pdma^OTBA$3 z;@%UsGkA{L5ZGr9pq}3$uV{uCp_Vv{Zhn|O@ldXm+Hw0*=vMaPK1`9i76aQAPHX6s zQyT$kB_+A-#{k#L@eL`<5R`Ve>U$Y;y(X@7Zp1D>biT)$g0L&>bfY!?9WEM@ukd#*4FSWWA2|XlT;wq^?4;$eM#_6am8|IPnJ3AYjVj^ z+9Lf2YWFE+*jVctYwkEDimLSHJy+p}nKr$9b0!IB2;J-REqbMSZ^H-w0ExxIj-O_4 z{RwN-tEZQhv}ZSaKO^E(t}=i{s(vK%N&UR|;?_UR)z4%a96);t&D?nTaX`=a zY1s!eG)}p7j*bK7#g5w2-a4P1%8AAsd!RE3e1!Ro>0JH^e_=6aSAKvfusFT%L z)*7dCFeBdvkphrrY};Yv?BDO{ZCo}2dO!tPFA!=Y<`*l03oYx zpL|!bczuPwl{Wa({c!t=IOE>Db4~vMkj1t?yeri7+xRTBJBe;fh4a80MoBwpoA*gZ z{1YG*-d%MPJKJM+B2w51oRRJ;T0Kc4@}PWz+%5-UR8$mTrLso`n$fJ2 z2M3tDf4yFQ6ELqy@~5IYtKlitTJCISDJ5188bWxC5@n1A-Jb`b zU~&CxuPwpA{447)Ro!|jqw08gno5-U5#+xSTqlSlsnq9mRStNNhMfk1GjqS?(WoIPNzIb+A=&1@NOZ#Vd!h6wY!Qx4-1J|q6;V+)1ADWAEj`KcOK!AKsDOmkbzH1!DHlxjLnl-?`-q`7^?v3hRTioqCBq4GK-8_ow-$H`!!qLoV?jjpl9jpci z8Sh*ivqx)jb0jIfF}t8S73elMQk^o{lwnySW#f&w8LtZuMwFbTH0{#IsfnDU7}@G@ zZLBS|%;MmzP;PH7Ky#dSKKZDv@0#XAaL^DJh*8f^bMN${R+bIZ!xU-zv%JTFl4^T0 zzDZ++&Iqh07{)hIJD}w!X3DpbTwH-6l0d-ZGPc}t&{l?l9G}~EBRNn?p5y>C*YKwj z3~r+wj+G{!(rNc`T{KSX9!Lp{oQ!dwtyUUb^&tJEPcSS9a>No3 z{{UE3Wxk8;X>AOGc5p}lk{2HJopIw2-A*CW0})xSWPKnvSY^{-@2yF^v?VGRBVKuXSt##m^xB0J^p0IEDDPg~ z-uak-&Y^}e)oa&usSsUXI}NWqyNKflu*czEa~`Yxm#=t;$7>AlCGi`c)XZ0d&<>U6 zWiWBYRg7cJWVG*ZQ?nZX0Cg7noZt5RpDz7UO}^4jc>e&ttG@VosmXa3pqBC9Uh^LL z5i^VyI6K=o{A(n5p6`DaNDlA%2C)AC*@bof6V?sO>NCh-hIu22Hh{438Tr~zTzmJg zqFmgv=cJOe+3)4^JgcX%CbtdzwlPHxoqHJbjG3IE1mpeka(|_Av+38Da<7?k@ty_} z?0QYpfaaJ^4Tq-izvtG*4jkar4 zG~xEQIR5}0*~_QhSzF2Gx9;7;%HfI2bk0vo>An#+{7k0f{#^e6_Il(zSs8#>LnXnT z)nsDDfg>-#;F{__6x>B?d38L+gmKvxZKEdv`wGJ|gr|d?df3+v*S1zYhD2ExYk;Ye#RE*%9CDESt<8J3M zKtTTM9=_Ds-~J_g&IV8uQ0nZ7EGqhqhW6JIggMr+AYmB?K zbF17m4#HC@z#gRZuAD6^N>P_W;V*StspwjJMLbsV8AkSxa9&u291?#T;=U_3d=m_a zSz?lPU>g}>&JVR;y73L?hHPS$w8frJDXrde`9aQcfzP3@Gx3j#@AbVQt!~{J5=1eC zP@w?Gzyq~?J_jqK2tu-ty`wy8)#R#54{c5;HUNz>@v992^75l08_-q>0Pfw)1}h`Ue7N4NbX4hW2x<3UW0X~jXqmT zbaEYhhUKXld$RukhbIY&Rc{&S(5x+(V|#z!c&Hd!1CN{j`#} z5+~3K5v_p#09Xg%Rt}D&edASb^ucQhmwUE56IrQ3PS+ty4cRivT-hhyWIym4f7>jH z!m5-0?Q?V8I$l8v$MU#h%txj=esrr(9 z^5u{$=Kx?-cb+TKQ$n}4T|(Y$U>Q-K2C3S8!3(Rimq0ZDlJV;QCiN@c#f#X7H>|Ez!eGCoL}C z{*}{7rnrsU?63d<=gvRjUj>M#N{x9zbap<9ok&ial(#wk8&S57!X`*&k^JABsN)c900=$yK@BUSVfoy62N?R+--g0Z z;Vn7_cu5cD6q&Yzg2Oxm&o%UTS`=pLPONVm{zsXOtti3@mgci+kiERoK-ie_ipQ2D z93RrF&3GlZXv}+CAm=%#7S7OkrV$$qh(U}4*Ov$KtvyQFVYHC#8RL=l72wg0HHoJf z^>;lv%0A9Zt&STt z^#1@E_?7Q3E$x-cDjB2;y9D&jcc;Xf2>$?>?w&vG<$uP$a`7$>qgB@{{JlCb$MP~-SlgPy0vlN9#ML@YH@&1+K=9B&OqSnxV ziRs`HRDbQ_=$_>!j4iV&yJ3mVa_i!22k~Tfw%U7(faV`HrOrKYI|}rTBJBBc#pSD# zayNFb6SoXLDD7>T7ftth9qY@=rkbp+_D-B+r76WVYczT{h=rA{oCT5E;e&kH_OBU~ zeB^=2V_xU4J*CC{v|`Z8VzqW{t&9_nO?aKiQ0;+$7~p56d-+pp9FJQispjJ~N>=L0 zVkgT0de+P{EwTn-z~?oIZ~m-g@tWFq7OvD2+?H`&_73!8irtH~K!?kX>RK=?f5_stn2kII?4lPY_PLnO%0 z^P_@ullW6NFZYHirGPv&084?_G;n%#sCV+FVTAcWzk| z2h?zDyM`@p&Jw0OYv`Dk+>PbA$t0Xc@*~%6+#Xc1=hzze7>;0%{n1}Po4;n0 zO7}f{0#i=x&4W{l=5rIrCA|H7sfO14J0HTY#><(w`M_VqR99B8Uc}MdMjb)O$-z_V z4{D%)v~;3MkGI>l1_{K-Ka+nt@Sy3kSjykH+V>e;3?^3RO z&p>JrR1Qezr8!^c>z=ixjLy@+QwVNm`H0KYyrLSl-8L9mH^4xcO0m%agz!m2zpYtgf;WRRIbQEOXNzO15-sJKLq6XjI7= z1s`Z69<`a`(k(QY%1ZIGPqTV;z*An17YyX-O)dKR9QcaMl4+G8z0;oLTK$?$^U61! z#1h1T{qJE~nzfatg$@cgl2|S`vXFl~R}ZDj2lma8-t0v2@_*hH-T3zTTSAM?OyXoe zd-%4u74(yH=3e*UeC(kKKY0}vvwm+ddS1CDcE z(cu#v*|9GkQ+{#&=q9}D#hxO#w((A_7`L+8mrgSFv66Z2NX>KQIDT4{DtXxM$JXYA zDfwLCb}`|kZ~N+-{0vB3FDj>eP9L)dPN+%6f8C!T!YBUpyTkWH#*&&=Bs47OWmcpDzgiM zr*S8oW419*-R><9J@UX(Kp3w@_+2zM_ZO1fLS}_5ShQ=8m~ZLLW?HctjIj)DFw2pS z4kK#R_+NIA>NgEEaZ0ahP^^i$PF!^z>ngE>i|2Ml!OcH&euu2-COLGgNv-Y0yizXZ zkB^rm`ci2{{{XXM>Biyz0FN5dX+eqzmMG%^0NWbvQIA1cnj3D@ROElGkbmRmzdOw1 z1y-jcy1M;OeuI}JdH1mvuEf~ffng|8Hf_3gC{uM5)=C7u19GY)qc_r zPI5<~#Z&OC5<{iUE65#YbZE#w!XiPBz~olKf@N)>b59oE)+V57L9b@6a6TZ`Wz)6k zY<3B4u0xrDI6Mxc-;-W_tjZeIf=941T#V~u2y%1`P43VjAj8)2lS0sWyoL32{ z&u-dTP^@S30S}x2yzz?rYz3u4GfQ;81LyHqRUAxZj2=xPP!4)}(>%uOnqAo#C$4L_ z<~z+8S-*oPHx4}HAj9(!&UyOrUahFb72k$z11Z?hp;O&>;E#IoT|Ln>7(}G1dBZpz zv0l?>a}DQ(bjdQw50n=z*8_k}eAa&%UX<5%dYOApRw-&Rm^?EKu0)<(p~=TwaDT07 z>GzX;me55MVGN!`Wq|pyk&{)K;*&*akMCeIvM>9je~oi`Y?l{$MEZ2MPaDAG#__Kz z*dsrsc+_yu5S%+R(Wy~IU6ejALo!|I@k-H>sQ3%m9CzllGZXJ%1em>;|kytpRHrZ zsNH{WUZvNO8(fG>uKe(Ej91CS4N>zcbpHT=dr+2&{Z8ve@rA31TI0>#)mU?T|m!xz2uuxt8-$ijcsL2-{wUS%A(+ z9-_IQ5k$#+vPFdwE>G6HJnJl{N}so+t!>X!F2_!C_H{Km=4S2~{uIJnzVFJcI=}Hc z(+V~|`r^K2Gu=`|ag)7z_4KZq01F?jbJ3{;A2D&7*Gu>lB**~Hd8{j^Esk7Da`;U^ znXLV7Lg|m`k)O_$b&E~BN)OVgYI>!$#goR5l6njp=c7t$qHP?ssZ{ofv~haX+gW*6 zC`&p<SN#Ut)FR$(7xmcrDIod(z)4f#jKbS35mvAB}G4;TyycG@LzqOlDnrEI* zm@78`V~#87TvEb0_1Dzvs^#q(U5<}c(POx~4&awu&}j++lq-59`{w+nsw>n$h&emS>DE z!x*I48DtM_|tNyy4{uav)u8%pq zx^XK&8N9VS@&4CJ@AQG@Xqs1++kbbAXRpk2UUzd1(Y%m|WC+cG+db>hU0~DVw6=(u zB8cH!^*=XWE9k2@a(gxAc+uxm=6wyuw-J0nu@Ng6($(^Dz$J;T&3LHN}JR~=h5Olye1OQ0S=UY0qq}R3~5ETKzZhG~vns{kP8%N16)btu|6PH7U&{#K# zY(`ED!7OCdd!(N!!wyF-LZN z#O1mQ(6qRNHqUYZwBNAfsWsty(rERnHyUX0&l$#v;_XaCa-Mew@DGm2~FeSC+(D_V@AQ%L8_pVFCRuO1k2A@o|w-KsO z7?M(cXc zU~HZ^?_6!eWse=}r!Av8arc&uFUZRqI7(KI>9NtU2s-d;{C5$3rU5A{9`U)o26?Q& zNXAV^G`T%`SF3v*2h-jUxzcC4Pwcy)FczmgzYtHQnmw2!M_5`T6rjy}8Az1Fj) z?{(O7jgN#apn2{RHQSTx(zbuIq(DF8(+Acg{uQUC_!Gp?>JdpMjdZ9FB&~HV;hldY z{w1>!j+s&YYWY}L;g-%!oXcp3{;Iwy9$<8y;SktpYvCQ_R4B4kNZA=56w1AL5Yzke#^`=pd zX#r@;q(lR98=q?1@frQp8R@uxD(4kXnxo#jKM{}IKy%!GDw%GJj?Bf4xH->CtXt+h z)ri5v06SGyz+*z(Phr80I5t#+QSS8XKqT;Kish{|pY3~ZUE;fG&i?=l$MDbxmJ$f~!tVpa-X=Z+Kwo zCjL+DXxd1OE~uk8-bMj9Bd1!z5wQRTVt$p?X!dYPcMMZ$k!&gV+>yBLJx+M%6{N4w zg{cpQw#TpSk0p5udyojhtX)P%{?WPlkNU>`b=&FI@mlzXO`3dg!5iTujFY%y*EPaw z1O6r#BmSDV^QnuxwJJ);eVQ%Bs_UVm8FM@V)AWWBSuCtw3~s>2^r2RYR{H0$f|1cM_E1G zv06tip&yu!_n01-{{Uy9;8g~>x1RnsfT@h_;0`OIzOs*0wGi7B4he7=jBs;`;VglQ z{!m8K9oEu4FxamT4~=!BC`sPOuR0N@B;s^;9v~KWVSJ+?JfJ71BoJzizlpB|awsKkK7cGx1LeVW%h@d9Hs!UX2=1rx?a4{}P)Gy}Kk?tmGLprcH%AdW02;-%B_>hBaQpT(f}PKqoxc zr1%J}?wR%{%SQJ_SbG#T=iVrWJ5Lm9kSN4ya_TeeYo7v|b5Fbd4Y4gX2jpj8`9dg! za&I{VjDwCT)-N1-q%Sh5g?Vtt86W<-tR;)4;m@Twx(*Hr#d-+la=xYAR%j;G1KWti zML5}k`GLTzIw;PgZa?X&)=+e20Z%xsZ4`lh2Lt_b{{T7!Xj&2dttb6*m;V4oHCy<6 zR9uoqcOX&q^95D31Zr|;jKwehi)yvtG4e>qpj@1K2Ni}ZS~9<5d_VDGd$f>k3QXu3 zS0Q@tK7z5d0@3MPWm2*vj$Lz{62$)iTI1|)?FH|gBx!E3!OI*RA6_Z(YPV3x&*Zu> zZ1BBn<#SvWHE7c7X3tjxQgxJ^)vWA}k6M;?x0FQ^zR|i-*9o4Z@vRGQDKv|Djlwy~ z20_j5Cba*a8h8VR3LJyZ~^{x z+UaoHU*D&m>;eW29y61UKgP1YA!*hS_)g{GFXn@eSG878fT>bZ#I>c}%(2+$Q;ewF zT@D9Y(vOIww!L{a$>f0f9PTZj>s;rFyeOJihDG_6jJ|M>Zo&ZT-|(+M^Hx0+Oo}j& zt8wfFR`KSSd9C<%-rG*MdyAQ6R#=g;Ps&L+uB;7uva9UV_fs{+Ql_c%+g6Vbt5d>} z+S@veGbtzYHMuXt{{R#~0h>mh0q9t${>Sxfu za5pcU`C>W_E1tf%xUrv7c%N*bB*_-sha0d@;mvJm@v`Vl(xXHc;z58pJu0)ba(IFu zLV#_UWRLd0`t`!%sZ@QpJEHlZdFP66u5T@@TI7XwF1}DgyLwhWl7Hf50m1v;zSYwF zLugv|-BTu22XP%Sf&T#4u4hLC>Trzr{+0ANCY=Q&tdAcSnyl6CdW<^fn;iNp+bg*f z7@edGv*!(teulAcFXlcVw74b}m?GegcI+g4!yFnlizuh@EQqcmD8ff(JZ&U@_30#v z(CWIpl84-olL<}%ZNt~wHRSs`ahqvwzcbNNe6}pYXB$0x39(2Q?{e(m1BBZjNBZR~nBtz~a_EKN1j z!qNQ77C^cD>UiWxU~QPfMsPFeD;va9TBXgXxv*QGGC>e*XKs7d=T(l@bVXInOgjGn zv&DEm&A35Jsmi17odmf^5m0pSpKz< zr=qNeJBJ8V9F{`-{uJ$C+NFsJC}$)LExA$o*V@LCQc#Y@9M+3$(D3cd_s#_KGaPZd zp7lSAZcUZ7yik}`o$^YLp8$8N{tw5Va&d(?=N;=m#h0IJ*BJw|EGiHa>(uo6)T(^g zyIVwybG9D+S+(Ird$BXLD>1+%w9J_Q0KRLn@ZG(TjhDf5+E#rWXD4&#pjyDSFUg29s44^w0f)7>%b*@juars)M zj7!J@GuPJ!v&-9csyTU>;v>mIA0wo?f+W=Ek&!_%ADAyV#TN+?(`IxhAgJW>NvGbd zpYV`M8Gm-b;C1A6{(q$<&CQ$)pKCG;baf5nfLn1mPyyqub7AD(A`i^(t#7Ny;;gi5 zST)OdQ&f&ip(n`@gyfN&j=tS%wv5d2CeYZ&2RN(e!Im*yo7B>;AeBQoxs%EHP7gdB z)mtbN%JXNGDy(~yla6~=c6lC3k-NT~jx4Hcl1&!uk~ppIXOd{sVui~N=ZsXkZm|WV z`hCCIzRbcr8IC_g_v$ht|LG*#(INS;kKNJ=Yl0E1e@9a0J*>f{{Tv15UBA`N?I@ ze>&v6V>8R*{ZZs#&-RAN6}fFS#pC!&-gQF~XJOyBBX@sJ)%6wm{f#N7Z5x013p_Yk z)S7?Rr(5w4dC+bWILNmXAkU}NSCzyuaJWC1uTJsh+y4Mz&u_Xw8dcn&<$C`BI`cuq z>s688>eqe$07On>rBu9;(D-s>hfqdqgoJbRF~%x?0LtUWQnW@gtdElT>7Lo|R{SU6 z%`;xpZtpJUm1K#s0BjAvO3=|fJ*8M)HNpccq@TMi-*i`YE`qI2QNP5jbJDHIALoS+b!Zr;X@!QFhR#cYsF;Bt^xX1n0m0B z6kz@x39n93mo%2*6NUBms*>^mwSfsdr4qU*&B~;R{#Jt)|jQ;?3wY)|D03NUq2>FJ8 zDyRPd2`$lV2-nxCs#m(4*SkZ26;5y|JGMKdtcc_2)lt6jt0v+-ybgYqSe%^Dkt|A8 z=xGmHa&$TCilR#zfBc0Y&rQ|X>08^vw-Lt8zDwxTZR^n29jD-GcK-nH3hq2S_9>vw zx2~%TUM^g-4TovL`9VLWDSRcva;>Sov6&*diK8B3w$>QV7*_Y%B$`5~oRgk8!99Ok zu_9<+*nVtUnYh#i!YolV-eQH%BxD2mj@4RiNmFbvrejl=3&u~c=Uy&8?nxum!6@>! z*|B9ck&Y8Pyf|FCenm55<=vAboHeuVcr@_s@1NBH7VA3g*tMwOf}0sr({nxvNZX}u(gTardP_Z=&a@hrp6 zwVhWA*5N?(!!V;Qr0GV<-AnK*87_Bsx;1=5;#SelpKog#OwDH#$-HE%W2fbc!#+H` zQQIj@G@H4>Ro-o*sy;lv&&h3UK8sy{ zfiLd7Rdsrob2KY&8nE&}$t02AK9%IF6BB^ABh=SL1?zZv*vX>n^9z~6tf?GZ20o=+Oi4>Bn5YGJM`qO+ru40qLAqHGZf2M(C&{d5n4b)FtmSo~a$jhWtWxZGOv3x3RXfif7u- z6k{L(pHFJ<^~*t{X=`&7@=V56jgnDN~9+UbVo#&w929KrXHgN~T{ zrzCW)W+r2(yUM;Y0Vlymg%e$)cio-UMtQSBO%OSO!=E9H|5;*klUW6CBV~$PpBeC$rdo;#F*UUfCxgU#d{*R^% zPw2`&kQOF-V9PJ~H;MXDXLgqW0O+Dk5IW|oyw$q#*Kc*|D9TAmB zq^^0%Jbmf|WYjtNxWKC|Z8}@WaPY_>lOZH3cCj6Bd-krI9FJqsej7g9c^!X-?Fs(? zz8bhM88-g_6zc8R-MRg1vG{*8My(cc=E^w7QWXCHjdK1VioyIrs5G~`T--S;<&%zU zDk&+;)Yi7DK88l3v8*YwL~O7iDFlzM4N0ZEQR_C+Tt(*Fp$#e$FvHOLk7^7WX#W70 z-^w#m!yb*67dMk*A1{{{a((k%8yLy9?i%yjzOr7+J)+83sK;CA{|jTSxEkdfL| z{&d*W!^3HNyS(7;+X(C1y;!xs(d=f0pq}Z-aMA~i4xXIRa%^9=ix;vnY>z5VKkb`V zhr_!`E^THBjO>07d=P61b=wrUk<-oCQPn|SMpS|Nn(Dq6k}c%}{&~yIMOgDwl6{S6 zwyGwC((e4tX5J(MG|41)I0m~Gux5|XRsGp+zMR*b_+5V0ucS{gP3pkOoNeO(^IoQ^ zllVSca^X*%k2j`p!@s^iTKRm+7pj*vtdDB}HwtZ~yD6oY$Q?;7os0P7e}!`PcP$@_ z1+1G!E*dlHPa?W&BJkRTa?nhR<`pNDZ2tiIv0iZ<)SvK~d4f#4UGVMdPgD8PfTHC_ zuDaXw9yYJ=wjWTe~Wk3&pkDat&ySC?Nw_BumdXmLXB zBQFS3xOC&1+>I_RFEr>QR+3S;4Xg+JvTk{*mbZ4+S20>hN{=lWT%4{?QSI8H@wU9a z7>dhQXoTvrG-fq@t@9xn&JV46F;rJ9sI6PqUgw`ubzR|8YiKvb$#NsKXAz<(lWCRL zoRQN6=DhR7){?G;dvfgrB+(-IAZ0^B7!(8sE z;%GG+IiXueNn{&Cw19;hfjtMcdeo^&MLidm{{V^SU0TOq@W$1r(ysR7hF%YT4R`l8 zOKG6auBw18_C0{kdFHsUg!ul0zo(!;$QaZF5|YOpXM#iEe(vDwu%~Gb=I2 zIRFafp}BQkE+R^7w3%_!Ig^q3R^7x*_r6j}BY7?5jSFxy@|@@MtmR5BNi$VaN-oEr zc!{S;(kqZe=+3RU=ojf*pQBpl2hkJ4sJ7YcT_NJl4}dpcN`bapUN! z^c^?U>B3ZW>Zzm8uC3>2$c+x{gZk5=StYrS+y&nwY#sUcuB%f4ZtYc2e8lJ6U{@og zTOBV@X1Ta$Eisir9FLgS%GSU}jk%?DbLlDK-AKDGhNQX%p%^U79u#EqI#x2fN}JOK zPeMjBU0m1iavnH`$jaX^86=+7%$R~Yzn*L#vyFLZVRx-;aNI#_I6h`vfRfbPbIWT4#mmIEuW`q^>Fa! z!&&Ni)g@~B9jBNiYj?wKec9YFe_H4KTd3-GT6%wGNLE$^5C%yZ0=F(@3mvSRNIqsT zdh|8Qc+_4?Woz~pZ#)8Uy>Xi6!BbAWF0}J6675P>IaDng#v}*KkpnS5txT#IY?czl z9PJ~oy;8ZFNvDPphIK)^EmP^uJIn(vM&Ouc_5T3%SJ6rjmzeStlXf(`J1lTZc;I0p zM?JXaxmjju?k1i7^p0DB>slsnv43gDzj2fpl#ZC)$^3<65~?sOS~Sv>r{2bODqbo* zu6oabbQ?RJCMhpev^NY@fzA)0>)Qm@tdfL=>>rdyuB+&HtbYpJI%qa>0zidW{-(9> zR70j*XXh;&fWz0huM)jpda;a_(mUY}SW3@$khRg4!qt#%KV?JDP6cz?)c$?mohyu% z5$7L8t$TENG}~B)d9YRo+?-V{Ld+ShAyj9WOrT@?xLWGK#y-}f^)kjw_b`5kN3D6H z`OF;3%yuSyy8Q)XYZ6OkpuG1ByUUsK`%!cdT+AU~HJ^AO@tlv+y&UEaRiK;g z`JQfBB^q-!Z}erqyg_Ae@+R4QrWh)qbs6q^R2Hz@&1-XRjq*lIFjLnUz^(rPGAoTn z`bh32Ku7_HTrNlDS(+0CzqkcUd5SED|3YS4`8#u3w2uiw6%4vJML?B#xmA3ylB zAMljwd;yDw;cH7qyNcfb07IU6Ryg70yC&cUdFH;e*eOy}-$nXZ@uyM7q@<3w#hxd%)$L`1)^9Nkfr$%@;F{wL zA2=Gfs04#B8+M9Z{K1AQiyz+P99EPt(8I%$1~scnw7H~r{tWRJk71+Br`6pRey6J9wdD}WDLFtJK3yi}di zWO)PvIjCd+V4ej^Mi-tvDhUop9V?{eJ%ivApY8kf>@67o0ORRi{a-%(0m<<-{{SEI z3mCdP+=C1CA-ljh@-2ahiG@4@?>u(Dx(E z1P{ufkPX~)sa{RZlgAXmx$0>kV>?uV!1bUgBdspQ>CG_>SHlzQOSosJH18}C+*0fU zkP*^k14lW{9q0hgdefJT{V6d{UjG0}Vk~L=>W~litD}lmhr(A36<_@^av5?y>8@{1 zIMpD}A2oEgzH#upzh_2yzTSlbjGTqyj+%`z`fAH0kQ<#f;*SKRh^(VMy5}P`x|aJ9 zZh_)rS06!G@ZR`>=TWyBh3&(aZJTG9kT^VJHPuP4{g%QzyL%X;oH<>)a-q5tgX`A3 zOu0r%k@U_8KT5-|H~*HDS(jCssJQ^3j2 z2c=fh9{PQ93GPvwX~yRqbgp+$^P#+%WOBqv!zY~J*0^eIJ2Naylhq@wh7qdXNiM-T zXcS1VF4iQT2q%p8tS=Z?UFi2#_BTlBByj_^$Rq-CGmiCt!kUBI&frLQ$eiH+00`~x zS-vaOroVWkyt)@RiSje|DS@}QZfivC5b2Su7xMlVY-&qD;()(bV1%$$&(*@8OAa?S2v(A8oW$b zZ~&YRr;3$V+D2MiLgNP+E1ls~lG;U*FPC83k=G0fpj^yj5{3Jp~~K4r*jU0yV4Ycn8m<)|R%r%IaH<9m6n{Jv#5 zR?pobeJdvBb&|pr<()V<`qi6B95Ni>hW`NBMRr0y?xr$+@-AD-E@3gusV3eCbqD%Z z2a4@seIf~wRS2BMNEjH$0~MtkJ-fRBlH7Ap>UaMDW6@pS)(eGMRg80TIpg!E6)t6O zm`$!nbAs|>P)_3Bz+hIx>$lfBDrt`)FfYm%0|0Tr{DnPyL8Qpolf$;)p7F|7y^F$L z4$^Hbt@TTdJ|(!31c?(db&n$>uTx&6VB~pje5lCyd*N0(RCia_?HA5uh%mu!Fb~qa zm&EqAka(L@j^ga3t8(F#CO|o_sUanq4Z3Fv0A^6ZPt%HRcd|dII)ZLY2H4{ob zXSn=Rh z0bXF;E7E=zF%z5ue|G->%vDAzPRy4lA9)dIIyKd=h;Hq#t==gkwvtRY;erpPbt`i% zzJk`_o!CjaLZK>o2Q}9xkk9spj5Wka$fWmEjE~S&1L8|nlgr;1whtBM<~V6aqHAW4 zO96<3r>^B)K3VVddA`>p8JwTq5rP05SCLyWlf-Za=XJt^?grMjBFJ5k(lGy8k{y3 zFsC~OdX3bJJpx4&Oi~}2lmH_D_O-&rqX4K-(6UI znFM)b87uOY_04h~Hn>@QJl7Do+ZFQsqq*n5;aE|E=7Nt#jqAxZEA~9^&$k4r((RDr zp%MfC0HIar?ChmWNj}LEd9rL^yRnS@O=sTS$$fs;QjeLG3~tX#)VFW+Z9o3LGyedO z3i<|kzK37nTtlO3%OA?3K%jN#TNTyn5Xq=ds6eu9Gh3L|`r^2M44D$zX%L-@J2ZeE zsu=T*`0HMQ;rIrxJ=hZ%nj3K-Wl%xS{{X7KcQK_>rADdm`5wLwq$3pPvB)=s?kx?p ziMI2}f#E7!X(gDQ-+?3YtMe=}U&C)Dy7{RaAeBkrvoPypQL z?>Pe*W8Br@DJtIb*qlX2OWk{(RiykEmS;(Iy+I;D$%fnJZ~N%+`DVQ*!~Pc0bOu{( zKEwgpL5c-?cjNVX8BNe^{a@h9|)iiVO{>@7b-Y;r8jnR8m@%$U)suys%@n% zGu&W|e=o|dTUk!>43t&wPc_j`Znq?-lW zVjOXfDGMxN4157o$bv}09SGoxxZ9~N`=cuFPLA?OPt(DY-8s}{M|hZExDCeqSS7;KylN2sjpdx*6E0I*$2Zu0*CqsE98 za&mV8inXb)+m&G#u(B%$B*=Sq>DH-#W?o)BgjN%Pz@&T0$c?htuG%o=q@g{v6-V#6 zBX;`cSGJ9&g`q1r0331%HM8K&lEe^s7#(x#QHF7fja2NfV6856;JTyC?@rxDQb0Rc z3?ATn)IMYR#ziFn&XM_^`dgm5w->S3Z@Cq1v=Q3yNZMzTBK-@I7lsRIzyU`ok|T9Ak*puN-Z8Ka4mjqzveeclZN1Yuv5)r< z`5G413FbaY1R|XD>OriJ9_>C4xwnjvkQkWr_lq14%LntRrMrR~30g^I^JY!4VNWD> z6_2eYt^SUaPbKVDQoL{y7jb6?rZPF_@UAQd8mHD&@pt^SV&`G z7o0NWe7~Pc_uKo6OMNM?{K%q-Q*c%19ka*1c|NhMYC2Z0b)?%|JXSDa65Qabj1I=B zz*(*2VsLZLJuB#OnR3QTE>f~x4-$qe*efNe>3VOCv};>9)_ddTn45khcFSju)ablz zpxN9+@!iVcv1i)@3~+JV)Yq34ZuH;@KDej-k4o`;&)xL@03p16OEqPwX3_@Kz95P1 zzIC|2G9Y1sL2QiXy+-mYsC+~&BDaY`a0nwH=bG|40Zd4tvIiS@!LLj3^l`;)VfLwH zV{^)cz;98U=(&E?3Pb z0CY9wZNr0}Xp6i>Q6%&={3PrSO<9TSzW_P$75V=Fr#Alp(9*twuK@S}ajxE{{W>}S z0LRk3{**zV2`#)ERbS2X2+zxdpVpVlvr+RT`S$>ST1U7>3$=b_80k>?6nCvUJJ@M*mfW5gU<-{ba z08zH_yPk7Sg$`Gs>s96w58cYBa;#VkgVUOmU$%Je(7KWGWAZgNoei{_rJ~LsW4eYJ z@1O9lpT|}ZEsRfV$xW@C^Uoaqm9*u{QYK4Sv~b#}%r-5aS^@mXsGuOhwA8KSh;5ok-+KYH@%+V2clKn` zZe+N-Gv7*8BMBJ@st!*l89gh}^q&FvW%Q_C+VbW-LgiIpGm)`mBRuiY^!!C)>fa2! zS11NZWR^7~jhJq~m1O9?ZC~}+-ZQ)$Q~4a9gG;Wyn+Q|1v0Nu0Z{Oa$A1~p|PuEsppDw~WY zc3PJ-49D!@UKj3pAo*I z$M;q-mLPQ{fde0}pEl*awsf9lONEA2R*dv1xR6vH2;@GZE7w?1yMaTINzK2`GpUq%OmUaKO>Z3LGWHnE+mT*`+fcq2GG^sX<$ z`o^91okAT}SR(Sut)MZ5E_v!e^7Q-X*g(Q+6n&vXa4}L zTzA7wF)lBkRctA>W?sFD9&53_klSgO{%Vs4l;K;cPn3`6SzZWj8%(gfx3;^MP)K9{p3nA>i4>R;8}+887LIDL6hI{tp5NQ+>6~7Cz3H5NjwbU za5&n)pU72xa&~QQ=?Eiiq-{Kr?Nj)+IW2TUZf%6_jS8|5c=;ITk4osYS0hO0O}=!_ z68$YB)MQ%*k*%;c84gdqY}-bL-oRu2TV#LVLb=a8gTykz3tKIS1Pn>7Ng6r8fG#>uoX0iwa|}%Dn#oFzz0Cgr5O|t>KJ7Ndw<{pd&`8;X>^tVUdG$#(eNNs(Y+@)IAQE%O z(~gy^;^^kF)L|k?BXT?o4#ekM&-YUizKmjXR9 zTbVGRO0f(c*%|(I(p&!kq)x5>04(}`7_MUSAs>crRx`?CPs9+%`qx3BP3CEKGiUGT zKZh0R(WJ2WWfXb!X6n{$o37GFg;1QY%2}`rb`>K{B9XCCls?jTjz9YOs`plrUmJ5M zaL~V)rA{zMtyGIijaDeiO^-kAEXGmy{r!Dw6GLV*A&pFm70*x(aYjITaxo*2LFa;V zQAyVw=>a>8$H!`t+4sUgJv&q|zZhz4L`8M9oCf1G0c{;!N|DV`<|&UkUX)xpiseV6 zR9DxL*~HG^lE-R*JJyhT*qINUf{OSgt9GtZ{{X|%T;5G^lZ6r-?I_>Qx=3z!soud< zoB{_ml{T|z^2=&kRFMf*D!nmYR&P$8j;6<@mei9^kjXwD+Y+iD-mX4&auog;tC}@} z*~N61<^v448(F~z>zcGK5yA5BCj{piQ{Jb4Dr-qrX7ksU0E}ZD2kBll8dRe+;PqZ- zqdCp4SsqWWM+$4Vi*67?B_Tp|0;96Jn$$t$%2##dSpaZxpYf`!514mnaVI(Tt5&eI zhS`86McQ%)9SN@7SMXiUqP0lrZl?QcG;5AQYtVoX`i1^xxebJM82XCO0j3rnhM3ReNcC+tXV$HjuO8LO z>pD)OCaL|PwvroQ%b9p1jAp$&Ei|jol^j^AoYZ2y4$>Hew{bn=NDy9mF+`m5!yt9w z9(nDGz)gDp09v`V)8|MQ7_Z_gB8(G$nDxDnI21WcAS|aBR`iFdr`bg=p;!b zW&O^;WL$U1sPwNAX%@Ec9o@XFNgHt(Jbra*^H|WYtYwPbCXQBLlrnHn=TRtgJxb@X z&)rzuURv8oi!{YS3{?wbr(9J(?78D`-c82`Ji=?-ei%gyS?U`y!!h6vKfPV_I&>1B zFhW$RBaw>Un$XU2itDN5a-fsKmcDhswnZr;%Z^uSWd1z|<5lAL$91NUFrp9>T_HAR?`2*X zf$c=BOm0cf%1#eGs>0hG6;dP0HpAm)UhF%p)Bm%}SHwD8~Bnilge z?ihyS8x&(7&bu2`VADaKOC}GfJ^ujDrBd+Kve|fdMYm@zEU-W2Y^t8^{1=du(+RGUmw-`2(2+vS|rFbukHGBJi73#3ZHq}d)0APC88hwtX8$C{a;FzGq zPEWELNUi7d5SU~iFvrSjZ|&Vr{(edP1t<32qqhd>zad7YXjs<2$;8Q|uQR`lj@S(A-=hE5#0Q4&Jah;!ydOyP7msEK%kS$z)Zj}E3FY&5raV#eBp(k*OWL`)-a7AC%NBSL$l0H{{p5T9lXIcX$pjVHY+Z_J@ zylbt+PY&YFSjoc>q=*K$eRXH$$f~frd2U$$01GM4KGnnP_Ojaen82ltp}0d98|BFh z$n~!CRE|jW-BmU!BvG>l5VwGfOR+wS)&f1euvzaverYx*d zBeY{6{JVdxaefiIlxjBU18ixeFYI=3KAFvOP@0w+T)hsOsm8Q#)acHW70mYY%Q_>O zH%K3hO(T-Rq>D*^&}h2Xbcz>0DmBh`bj(FR`yt88ufnx#}%fP5-g+4xF$zk-1-5JrFo};ZBpmMx3U)6(T|rW zgYurW>3$xMOG`@`BAQmajv(@aFraPiUOpof)mI3+S?YIEid55*z4j-VNnvrXGawD; zMEUQHqPbPIgNJ=BV^@l6ua`%YLO zj3|wWA9oyz#qfTuGWfSn5KDLuX7Xdo=peu8__n> zPIE)JlwgdE)FHzvkTLS~ry=6Fj-pp%8C|CX9Vuof3gGZ*8|Fboe>`s=b>vV4US>XW zaf4F3ZcxDIjw-kxyeQoqaYeae9uvt^#%TcS*$UkN| zFBx{5nR@>KdllPwT1nd85|U-wBc0@*;x(VEcw)}iPm+6fkVu0gb7WQpuCb_DMkBYM zY7UFE9DZWHc{pP!##ntU{STwN*PE$l#Blj~{m>;?mfiA<0iM;=_#Z^G)b#m`_iVdn z4dtoYauDSct&_3N#1EpVO>uLw>kcm=;;bSzFgWIn20$g z%I#)rTw1JCIw{CjARk(#ePi~k3GLN_G_kSp+<6d3aR7ky#+LwY}DNrxQR=Z z11fpJIpeoV_Hx*|u#EX{?|0i{&&E|yo2J`Z8h!}y&7J=MgW$Njzi7X5yJNt>`Em~? zyt~AHbb7X0Ms|3kB=pWu9+jxRF}bzUY^)$vNTEbVXXLii!S?EZI>zw@>{pk`JR6D_ z<6N-8%Ag1J74(#;#xDG}ea{)U%9l0AX_u2}v&DZDB)o0%mjofous=7iHPd)+OVKqg zE-`T(yq8aM!bVv(%BXR-9D3Ei30%jhXeUa0Re0kLWHN1F+qvLlAdWe#&jGY>-`+a` z`?pL2Gm*&rD@Qc1Vxt|da{A?o()HV`AnhlUCeY2+uGu+MQWyjLsxKBelf}A=sKS|! z26M$Nvtlnninx(I9gQ`S_UC#302*I4&+!rTs;egMl_@+{#${XSi>X#2fE?$F=#UoB z^uRj(o%3A1$saZV&OeH}$RucVT``mbO{|@3IqqC{Gdx(nUx#$EbBuMaDn|x#F~_ZT zelQ3A5&b8g3iPfXZ28?kD$Q(EmZwYLfE~KHF%9QXV864nOm&|{50Y8baPuCou1wy`pyuy5WYtpIgK7?t=%{ELE z?N3g46vywLl!SW!0Kg1?!bN@S_BQ_j(9*tvub;jEf;>Zff9cJ*{{VrdeFH)UdXZXVS0}8@)Z~L35|Hxt3cv{PHAUFewgz@J~*aiubJ>F<7EPtVhbY;MT2(2TfMq zHyK9N?f}O%pQ&lyMVnmxqBbUShB5cLXFp1wr9F!9UKNumEYT7N`l$-miMvc~ZA(?l z4xVoZ^{hC?93GXX)^km4w*d%=SVqtO(da))Z(>rrHuY~2$0nZ!lJ0-+Tgi-& zgK_+EUTLZ;EzQdSaArNd>M6H(am=qCmwRwa!w`@TPSzjeTV52knhkSp#BUi1y9_%5 zD*Zu0Lb`(-toi3=#@yF%z91+ODsP$V*Ww(l0kxS#8 zbK0lY%UZRhZz0^0><6zo98?qlTkV+47&EFemB<<6`qw6&CDue?R_Fx-0|+Vh8in=dv23#5 z%2DS{$-9Hd?Mf6{kmXf1q3*$!FR>svRZ@fLf&LYRs9MBtZ^CU2BqBqegn|ZtL0cM< zG`8<7d>G^*aoY#~0M}gfmk94GIFV;{F_E5wzrSkrCcV?lrz=k9JMha|*6#K8it|*w zlExU?Lla0!p1@<%Cb&NqIX*7cn4FV$CvST9Z41JhR*P}=s~BXBri=#NAz|y@ydTBn zH;c7m2L|p;0si%IO3P!iF6LHw{{ZV#9QMr$Gt!Zdtw_Kz1A|86>-bYYDhCFDB+tv& ziu9j_Lw2Og&JDYdZ05YtPaSL0{u8-6zQRtyZq4}ariAu8ZDoklAX0EvY;*KAm!it@ z>UR=#R01{pGEG_W4cJi}X92v|lLydo{c4|w!Hrat%#2Cqy$UPbvBOc?X4i;=Hivkr z!#E1K_hJ4u$@s|(cN!c(@Uw+KhA~|u__94RD+u**BZ-5dNW$+j2f)DYFh7-Jc&!rd z<3wokBWvIRZlE7reJkg1cv)ecXf^Kck9!e`jc;cqxzgVs_>|lQ9Ee*VTpHjsYpLP5 zvu(wH)16$SVR5(?&N}rq=o*?ddPUO-V|>FS22Ohu)AIc5#O<|Pu<*zBhFF$O#Z==Y zo=L}EmE>hq9O=}HJq&GV#&Br&DH1~zqHbRb6mNKPegRYmD(?zKh`th!#crOI{#XM#)tG=lm()5B4>v1e^?hweq=zmL^VHU4BQsgiYbni?*Ms zr>5J4)bAj*ZSnbvn9ttIjMls=cDYf)NC)z)oe5%1a#1K{Lxv!B{n+NZRMaiQODC1J z%Y3Q;$4qDIUT!9gYE*BXBy=fG3dvmZAB;co`=+q}0G@CB#dDtyE&kKvdmNATd*u9- z*F*7rgI#HO7#>@YeMUuQ{3efoi)E7;5!$)@%8%<`Tam`tT5I~;@G}0jcRN{arYg2}_j|T;10OQ|^$GV&yOkhhMx0gbt(T$KioYy*Y;-jn<&k z)#d%`j0G7v{#ABpq@PYgi;`O>j@84STj~`TkmeX)`LXn`HO>Ie zK^-g8^{Z)TUSNAE7#-{8ay4oV`JYhan$Ub}=18%Us#xQ59tYOCp9&(z zw{fvX+aMVnf(LQ!T%HDV*jG#8XY$=}xesuv{B&KXKdvjcA4S!p$$M-G+f!pqQ5+J+ zNXVK)jQ;?%YU!&DK9Mb>t*mS=l2;5!0~u`f?kbx%YWbYQ3h?#0T$8@K8_}ulBXpYe z98EYHi*-0PaM;w-j|dKnFSC@_KPxMfwDe;4^NM?EvJSNZNU>n^3rm#JY?(R*}nX zZqbM)mSz6{R>zP|I~;NKte+8S>1(7~+?b4aCPs#GTPq_jesTP(@ACXVu%j4DbLa7O zy{wyC5qNyT9<6I6@OkKRn>gf=f!FI){9S7novup-x@?Q(bxh}OVMc3=w0(1He811f+zCgx!aBp86WJP^$&v_2{laQx5<9%;a-ZIRdk zjBqNSiCMqlC%*vk=N>9sd#MpX4o4lUq9#+^*GNyH9Mro8IXq^vP0Yk(OxXI=e`uT& z`^Nn1PGW3azEF{qk@F7Xx(JxIfv1m4XBEf0%H(cxex|(#L$DW`Wq_NAL`7R3{VO@i zElY~Kv4`UVPlxn{?Y6jgIgnSP{6c|kw4F93A-0kO4)y193`vps)F*VARTY`N;NlO7 zZFAR}^hRNW5_Xgr^!B#wz>S|={Om#5zg;ela zMp9XL9M_gi07WLfvs|%;PXu1gBb8)apvT^?7MXrT;;lQYj!ML7jC#`?dUU6S`ef6h zx0d2`w~2R2BLq?EO$45|@Czfv7rvikZ~YA`-|NXf5qMWn@YVg#_JW{!ZOL7yBn``6 zw?aVAfhk8yZ@nlyVu6n19MF2xbAypfkw6Y3jMA^&Bi@^l(=@C14n1fYDIo9&6p-Ob zi8CfG9wXdW~V=^!#9i&!5t93aQr=$igFB#zW`cMR#%!llb3H~Mg zc{PNr3vpWd$tEN9T>iC<8vNOzY$JZoKrBEffZ6_4Pgj`A0X=t~{L}4&eSnPj&+00! zrVeq8eCLWwa{3h!xxuc6+0^N(t(1$KpK$NxPC=Hb9|C zoHCF&AA2>%Yj;j8zRi^=XD)c;9-V8e(Iww)8xRN0w*XfSt+4w++uBB{hU1eTqU_%; zc_{TAC^o;tNE_S)<^6MO1e}lyp7qWW}WG!5({O z=xdhOQ^+ecV1TDQ{cEF1mNb>e?*|+YQNTaSxovX#StFK7K-v_jP&!v@*z?w{9lfXS z?P5?mN1xKZaq&?90E)9#@BWo1KaG7G2{>W>)Ez$oUpaWAKk$}n+5Z5ar~Y-pBf17K zZKRHA80Q$IZgcBMMmy3dS~E>Fy#+8F4SGkyrOveDB!8lx{R+I@N#?yb;XHnJwK9YA z3pOk6cT+-p8(uArX!QXxQeXtfpF&$WKT%e+T20+~3OD zMTBxfvMVT6bz%yf3>ZgxSg0R8luB)xWa`YiN~*Mzdgvc(j%QF0iy`Y zeF%+foz!)iZ+;)kZ8!}ezgIFJymF_n7=36qj^H-B{43YB)QWEny#r6DA+;U$UoMb@Qx+^)YPX3 zJ784F80vre)s^uS19fX8ZWzG7r5$;L|hfm;sC`yIs2so(I=8cJchAZ1r5%L!fBMr|7X< z$Ow5)owRpm>H1>5deXEkc~ql$T-l5jI43^z^U#W4FUhx8bBPG)OeX*zTJz4}h~zqZ zW|$5KUPTCYT&e78Nt0`D9Ss0FR@h^hel;N>WH|e>42o(2C9z*lML@1hB@f<=@l+z7 zw;%jVB$MVlSxFh=8-;UPYz=X6&f^K@79ROLS6g~~{U&((G7xiI?w5Aal-Pln01=OB z-j3|bs?i-3KP{W04oKR*m|P1BnoVyc~in$Yim)fx^vi zA~#$VJu9s6p0^)}+G!$Iwu~K_I3$yte=5?rw`+UZBWcT+)TT4@434DN&Ee}!N>P&A zbLgST74DvJ@3XJ+uDijK0|n6AoZ-DVu3kVEXHZE{#3`<$!bAve$v%UQ!lN|1!#jC? zXGC{2EN~WM&-&;80MLyWx6#?d=13Tf;|elr6R7F7+lyVX=bw<$H2QV)DVbcUIR|bz zuL7kSocz@y>~t!W_DrI_AS`%*<36=&(^nS~h4P5cdd;@OL*x)h$;LYW0EJ}uhUnbt$s-JY z~ic| z7m{Ng26?LXQj=thlYtq>p!tPW*Gn4_#xZ#(@~=WGQ<_q|NiTKxbQM{ZLl(gQ0A&9FjVHsKNv?HUlMp*puH&2m)bm+BrLKoF z;xUgC>I1l$sEGW7=~MWnJiiv|N)AArijFbzE;G`$L`!x%BLsR<@qzrQK~^;45#^;` zr;OA-2h?EGydWL#?U9VwYj$854_sFrZ>lBV#5=2i$f(=!M^2q;d0wrOnb&yNRfg9@ zeKlng7LTf*yX-}I9N+5H)}8#&NG2Y6X-gB&b6JvS$x6E=u}o?XhZVKqn^P{c1i=}E zaw~DyuN9Q4Yp3v(mll!T-P)%4Vk+t}ot3f$)*0|0;YBScZ)!yc`OPgRMHB$E;rC59k||g2 z9MA*Po%qzphv&%GylQ~J^l zO#&VGQQnZ!j8Fn*oR~T1wKo*y;Gd-cSg`*9kc1yv-?CEv9k@Z&?$L67hOl(+`3c+q z0KNSycG;DDH8V)XSM7_Ey+_Ju5p;M`=~chdB0E7i!R#|o=@UZ$x!<(R+g35V83gic zKl@iuagVmnRE%KAc@-s!@4kaM+6gD{6aqBf%HLPhr1ED-uA`JGiJK+YZg*p^TF$u8 zWwf+u-b5-07y$E{qjPaA63S9Xj!oFg@K3En71hMk0W8uu>xN=zTLHBJglNh(kujTT z&vC{-txG+HuB&8|Tg|fMO2OgUNLB1|eREQ2+HKaU3>MaKJkkd;FkEn>zve}Am&rA} z&jgB5(n74c=tVB0=xR@*HKn6Y(}RKr-MG$a(cD0gSw^XsNg1UC{on^5rD8zLW&n~$ zxvq;t&@XkH*@dFV`=s2)M+!X!3=sRRCffO-xw5(8;s!+8PRHjf{J%P_1SxBGG&4){ zBLFZy_slcau;$*32O#4d9-P*Op9I&BZwxAAndDVaIZm^Ix`E$}kw{0Ms}QHt9cxBeO6fM|pY@HO$X0GZ z+(RGAugpjGJ-OYoXr0k)&ABAD-cUpw@4He^glUkLPUl=<<oJ}uRc{XNP50HGTC4@gv=$z~3DIp@;8dht#NiuF(q zf6}DqAMG08XKYNIDLnNgQy=f2N)KMN)12ot6d37CM^Qlc=|B!%_2|D0&dp9&`GH$( zuHL!bUUoRmdY8fZB8Oj5Q4S5IcOM?;LSlR+9wM#TrK@!?(Gv+%nwcuh7cZWsW%_mT1-0 zt`};NoE!`q(1v#wmdZlo%u$?n!K$7pg@ufAyj!GM(MVEoPaqt7b*)=?pX@M)h$@)( zBd#;RJRi=o_H9aQxw5FO2iVPWzuC89QI+y}811xIB?X3yb>zeo2ww`D`jdf=1wZR=Oc1xkD;y|%!Q>=S-i4|7|f%2_$ zlAQQ9QY+2 zLz(OAN-zi}kY@&ggMw?I=Q^uXj|flCSx|o}-uQnW{{R));d8NFzwzF+&q2?F<>NV> zKaj4+;jC9TnzVA5cEf86VYXv-c*c9z4n6YWru>=JhgYc+vCzl-Rem9Ee*8nH`xRpR zPN{bi;5oP4q0cE+YcELtNct>1 zZl9|>i^lg8HS{*KMX=kK0Q!R2&31nYx5MH6N>ogjomVH=S25y!D&qeD##Yu%CJo)e zNZ9kWS-3x~dcTGIHatD0J%xg_>%7S%vtuQQ1A*zqeu0%n^r0oSkCn`&YO;&mt*J^9 z-M2P#!8xw+?yoIkj65H^RRJ3r0=S)EJeKaKA&&!r#d=1Y6tGxD1kwotB9Ae@i;C~W z^f+j4%waUG`{kR>AH=7qt3qP&?_PuN4@%Lvg56l`jRt=60aaebl1=cD^L10uisL#+ z}c*8F_ z$25;`m1Sla$K_Nah2vreLC}u1k!25;xQK`&?&qQRtJf#Xm?Jo491nWKxYl(!ZbqM` z#}slX0p?}SKl=5(6?SD;V|-~c%@>^|)ShAIZg4*;&(`EkI?f?{Qey7g6>!Bk=jqy_ zg|DOn(_0O2hk+sutf$mt`Bqi-oOJus_Qf(RWO;3#7<$%OjJYT?e7Zc<(D^2bEbe~I z5QSAq0as}!j!(U5>N5G7Tzay`ts`11Fwa`U)vx2XiVIdpV|1)uL&q8Ip1rF|>S)e{ zwlXUyjueTQHN19MPtttr@sd!+Y-}55~2$ z`?ix#w*!I?Z>?c!`ku9}%>(KdvSuYJg<&Iu&1u+a`h~rAFkW zHFKVe_uaNTjeE;8+pf$=R1)6R#Q65@Zso8+7*dMEB=zZCj=K4|g99W)D~s_HMIFto z_jb)R0a-$;ZTYzGTsb^HeFpkm-yMHo(akD3(e8q&;zSt5O{`l>4z97fvap3o3_Ej9 zn^H1F&{-o}*_l>D9(R1;bu34>N?UziTRmO1DG*#rRH+jj=OZ7beR|ya=dsc0x(tTG z2bz0mH^u-GNHfcB85#cQC+S&!D@#c2E#iXYPbBia$O<@AWAgqd8LD?z-`Y~ibYn3W zQcA3hKp+90Pp1`9<|zJQw~;AOa(EcxlC{davC;T@!>o;Gsp?G}I&o7r?=eAOxgcPk z<#Am0y{X5o>oVU#7nKvR2%f+IKb307;+;h_==9bW(ptneD#>oFK3s-bzwmNs$tvoZ}h;3{dKkaL*dBt~b3c#F^(>Neh-W1lXE&MTWrp0w{ zYpPE>e)Paxf=|j1AD5+LXj+}lrK(t4+`34Rq=^_YsKD}@jDeHWy<_Mq2DM|z=JFsv z@2X4|%H<2MHMJ?c@>)&ja}(6pr1&dC({H>td3SkcmlMZwv8iS}DFcoTaC)i(u3mh& zB2q9bQ%>;?tE5{zI(_uE@wwdUpkSX~MN)c-OHZ+Tz+HxD^6rhnWN(f)5BgPvztgyvC6jT zDDrs2^UpPfb3Ucw9YWDBu8Ye#`5n0Bn@gk-l$ab75e z3CZXwGLjNOv~f@sKmMw)=rQ+z`q3O}dQBM5A2Br0nO=eKQb`a@;nlI}nv4#4rbv6@ zmlWZSyylSOJajaMCwg*`pL|uew?S$c0gyV?RTu_ z?iyUf6G@K2bNSbC;wa#r-6OZt)RqO2z-%8$9#k?gs%+iGXvt8eZX0p+3rx$lAEzGz( zmUYJ&!0FSP;$+irmRyT@Tx9&fXElHO8uwG$QU--pbG`Z+B_-8@S18`;Q$}|zt8hAEfLmZ@NFQGpZWp?vKGdUmY+PF25_=zWcR$QY9%z8HbWIqUQlF?xxkN5CFh z&3$z$#s$5U*z*YGh~NNw)t?S9iph-9M-JjL9k6?zpN(LRV^D2AC4ya!r(jr+DcT3H z{b|~brl&Qe4(U6&unXtu@=`UAwmlj1mg{#N-wSCpGv!e6TPPZ zf1j--0mcWGq}pN_RHT1$YCOj%Gueqcczy9^4> z(*^F6szW%w^I#GJ@VcBg;6@E>OX6GWt$M=URv6)UVQCRyQMs9NNjS}G3cRTa)8;z~ zwMS^#k_R9TGPxjn=DhR78oV;!&89nfmlr4@$WObtf;;u;UYBhH22`6+XOV{}HmaYX zuQs~Tpn&ZaMAqSCqdEUa#$MfPbU4zHZJ$R`x@ zD7EmZo=h210Kt9nn&-Sw8a+6mV0JRO!OtBkYuTPZ4m_;H0|vx9yW_Q6n&+3GlpBnF zOsMs%+xHQTHBuR3&Pm|$j;D&2+RObZS@ipOz>s6+z(c^@$MmRL3E}YK%+kJmggob| zZ|DVOT3t-{7QZ{T>9LHTn~`2uHu60iS=i~P({*>aw1(SLxALxI4xmMind#cOzZ_U< zI&PC4wcXrO?UWf7HO5$jgPe24dJI@g?J)+$jT7%L@TeSSxz8KvH@f}T*)+8KWW|xd zRsfI{v-wvtugy_KY;?+(EjId|MOJ4;M1_}=sm03ebAQ5yW2WI{NB~Gdx4xiRH_Ggxi1|Pm zAB}Xjx+TrhEz4^6id_o@R@>{ef!n2W+5;8v-O7%49DnFP<6VWGq;2lwvyvuAR1mY1 z@{$)AADwx5jY_eNsh*Y!oEzquwWGc4TK&qIBNkauR4>gV?HsZ7s*&m!Z!OwIWCjn* zAvoZkqwuS`L!;eJz~mc~k9-QxzlnvU;s$k<&H+=;Uf->J>XLPzGU$=)C_YQYoUQw+ z`0K;B4*6%gQpJz^U{|`_M2dxBBj+6Ayr}6`3#Vz<7ZzV=x`mjlw(ZM;7;rtwuF^TK zO{z|c;7FnHbH;1!@|SuEC$aLmqP1qnHL|tTak_8cARBYnn(iXFhT%kaQEdL;JOVwd zhth6Y{6hoCfv`%dp63># zHRLDDl;9qYD$)oY3}SV4?av~jo-3qHfrG|-o}#lTt0$Pi9l?eflZL~Vgc0T`I`oG_$$@_SPp?an*ZTZw$A%Xg{~o2tmtU!tOqEDW*tskD15O+H(d znGcyHreMt8yz%W>lQXpC$@{~RT?AJ1{en4K+8LdcefGm(bjK9oB&BvK#kHX|&D3%y z`aRSL*vl({k@%Xzx`b+4jge_Q$dK)I7{hZ>YRhgSd#AQoT`(FcPu(8AhMVDsGOVo3 zza;x`7#s|Hf1YV$XAfh`z0ELjnyby&msHTS`wNRpxi`2;a;qZ)^L7K%{{XFAo~?N> zvX)KcJh!e%`T#4h@m0i9e`J+lRNZ7B_zA8PR=7!Rt>OfKtd+_7x&D>&H4=?mBk1UA zlV)T=1c~O8<>%(v*A)!<#m1F)iwD|=Bo{zVGso#q+&`sc>ZkXy9OLCb&a}1NnRQ!H zx-~!HIM*-$xljA{Dhr<#>erAME+JCPFde5X1PCzIy949LpGSv&o0oDAX20eoE~d4M)-%UX{s&fkla~;jho12`Wox4 z;I;7;sJpTAaN~0$Z3G(cQENnTUh7jkLh!@M;fqP`Z2tg!A|dbr=j{M`{{VF2sCbJ- zw$t?g0JIX20eLz*zU*KTv<^M%wzF9DyLD47x-bMb=~Yp_P72e((g}8E1PM)%_6f+AYjAW08-Y$p~-7VWAP56Gn~%HBkNU9E7OX% zsawMZ)X#CK+9bK=GEV;h%N3Y)3w0lT2ptGDTQN&=XT~rNG3ii0LVrq$^(%gx#X|aw zBphVon2UFs&Y?Du7MRYzWQn&b>^7H=yeOEGXqAkccuSf7Vgd4=q z9qd=B3R;vQ8A&b9e+rkN(O<`lw^U^Zk1&2U%z^WspN(~XIg;Mu@U$#RS&Fbeyw@nh z5)tiHV0)A4j;pB#U~#nm98}%~bN>JdwYXD^?jNOA(~7R7yl_VJTGKywwPsJ$`sSD;*M0#10La64{{UW3$NjX9zJad+_yRxTcX&=S z_D2~10D+}@^&}(UKGbBEG6+mO5n~v1U7xChO){fK@K+PvV=s5PQ z@!yKjk$=#0>?(rr3I0?8V^0{=pdf!0cD@sWE1fFtIc@FcQ8;b45(HkIbCFzDoPXkD zZ~pakdUqZixFd|+FaCmPV0IAM>6Vgtk=p2>`hBl({{Yac_?p!H+ie|3{dfNW>(`xv zat~jvUxofroK|fL=TR}TxzwVvlETsV##l=XrGom5XBAcS4I%>a8x>_t7R#QV+|>;z zs%vWgkg@#DY|QaUz)Tp$yDk)ReX3;aQgyY@uaI`@&MS=%AU}BZuA19XF=!XIvRy77C0OHT&wjvvTDpu5?XQ2149{wHI^Ac2oEYdW??w&~J@T*4B zOKBL(35lF>hXgh~Dkmi7%^?@cQoE8t;p@dYXo*a5+&4AN+nuPRJ?o~rlno;K*vo?I zq(|yQ0)I;9o#Fl3#}>k%2PEfl?tQCTdl*#Qt9ZK(K5>6w~?Ioz^E?{ zNg3#BMn+vK2?!uFwmt9#Zj8ySuWn8J1U}*<93FeuTVt!K@l3HxX#LWRJ{~X5Ltyee0>zY8=s5@;Etc+C2D;&Brw~ zHa9bpVS-t4*ULiK{HxKIR`8$*4~A@+xX#<5&#iX81HIE#^*fC#O%^CpIQ+F@;J)Vh zv;KNkaKy>o%9}}b)BMKL?s!4dH6EtI>;C+=_*B-mSFlRV))zz-BZyo4E9hN+SG$Ww zfO(QdBHV4<+?PxoX9KSl$KCjX=G_Y0&2=L;M?enK^kOJgs?I8O<8;@14l$3~c07jS z8KGb%hDeF%9ZKi(uTl6yCv8sfL{&+&0Fv8y-OdkVT(6C-DZS+Tv*^AsQ&d z=Rdkr&*Ceu@nXjwlOqIGY;%Fuyz1D=3P(E|eE7c_(ebn5@UH_`$}}`LE?x z9w{R6f(uBOb6b)3f&ehtIjq?nFN}26BlxCa_<>bCW8y~e{-~F6+pNc6i)nJa5Z&vI zL*7X0sV{Msd{G_3{fU|3jRZuYc>|IO?SWhui0zp)Ip%AWmCGn9F_1Dnvq`CZKz&J~ zCr((zK!BohbDmF8SelNwxAt!lwJ7h5$kGFV4spO0*-D(_Zk}8ICkv9bkSA1E?0US;G}}u50Mir1elNGv;QX} z0|)Z1+eo=Hcv{NopqQkPs|7OSA3=)oO=8aK3y31UMOg^OI)E$MZzWqT9>T)jb$>1u zT#ujt{x$YEJ5r#OS@XD+?JKt=Uhv?KZCzZm0D+t5AniQ`b^<51K5SPTm%44-4RIb2 zjjnGU%eaYTiTtzM2a2j>AYr=HTl|5BA1F9s z>ODm)fGUtV9jE~wUs|Dd+n3s{6P3+p+&5YPs~U+fR1h}j6{o3cvszt45dQ#He)E3{ zDC|A!F4U27n^z>YL#fYiXw4S(k;dbKa#Wt8v!2HERoSh3X>G6Dc~IQOcXbDxk3;EM zS|^6@{62;lFSPxJUF1kfY~T+;=~DPo6*rSipnwwxKDn+NSVeyoxd)%Mu_N1ca#ngB zbfeJp?N>yTSG8EKCO}Z`M?~dLF_HP#IR5|-Em!CJT${kgKo#7H6k#$qV>tG#`L3JK zF?_hrGtO(ys|cmcj>>bM$nyPrNlhl=7^9V(I+95q^~vhsG|;c2lbZA&8HCNd;BV=k z_~yL#RXLqA{om_Y;p%g%opt^Khkj|pIqTknXy`?BoJi^x)LTMIl7DdOSn_uaR#o)&)ZD9v5HHHmNXE^vAU8Nh9Hs0OmSB&yjQJRTE_*o-<1k|#aNS!{uQOG z*-00KB9aSqWR_tq!~Xyp;9+?F?>%u`lx#DM1NE<3GE&tZT;k&F&f>$yAKDhe*HA_C zk|twv%m8KwKX<cx4|IE&N%zreJjO&LUxQ+g@&8~LbtdPnK7pq}8p{rmt1jFFBRYix-vs8Myq@0CWieUYFPg!fwyQV; zf%k_3s^}W^{{V?5xV8IR+WnhpmkYUN2ivGWooMP>ay*jx8k~@+Y^t;Y@I4M{x(}&O zLq}<+xzQwoX>MXM+(&;V3XJ+DYgbROz0k2?@2i)IR2Emkd9~j zI(39egxcK(@t?}QtKjX*C7hA0fJwO(lfQcNWzpnD8{=TZ{lEUTdI!Te9t~BCgM%XX z9`u_|z+vlJJ@op|*+yfJ?jF_iE}y7L zXDZLIzSEw2*U{e(?qj;qwE5zX%SS`<=cPm3SFylrI+3~6b(?lM5i=mqZcTBL0;-25 zyB&JY8y^{Jkz8CX3wZ2N+V{9ylXCC?!c49q3;lwz%67$c}Nr})!<;ad2d5BTv+=NR+T{A)K>n$gL9 zL_2vISSSoS)-aI=_YWeE}nVM%T$nauDwgk~;LK z8jzE;T43QNt3?ezjY{AB>gaUw{{R%_!T$iKrT+lXYo61;^Zr9k|~#{l%_rD2urJ`2yJ<+;9C6DYoK}wPg=1dv{>x;xDErsU-ILUg zqx@^6x427*gf|jwj{uXq1lJ*dA=_pT1Vjm~380dAq>eceWx&IhBWSE=6!$ianX%@i z8tL*I%!)J!3FPOgu2STG#Kzzg(7*Dn9Xc89?g*0MWn>_;jzP^=yNy>;lz=v}lloJF zu@cnjw8MpU8a-m&7YCQJT4TX>oq> zu(~kgIl$!OwPoEvC%1)Ik+IxlA8}gJRxzunlDjO0ml^jNq?+P+)kLy81Rp3p2d@-5 z;Dzgq4Ae{x)0*7Cmrjx5zlua3GMtPa^`I7M?dL4*6B(76hdsJg9W_VW;oJr{cdaof zg(e$Bi@dfw8s)DhrDS!`Nyky%VnbyL%@#{GRDyHOb%V+N$+wP1X9ZhnQ`4XEu3jx} z*_993U&3}98e{a{weUk6}=p_0aR}Fuu$z!eR(V=1< z?bwu*p&XIh>t3xS=StCbXO&eb)MrWkiC{*5;WnFt{{TF3{{XXBU*L@f$_vdzzPod& z#IVN8EHgYfIbpYk=bUx*uO%Avoln|Nes0>fbZ>}PdMiv~)8U;Z+mSS`NFb69bMH-3 ztrXg+eKu)-k(!dy==H5~4L%(rD~0=9&*yn=hWb|uKqsj+%L!9>Cf$h62op$uUDeKC zd{4X7p;%u?wvBfKl#r5fpYLbX)?4_tXTTN&0CSX>Df(v_r3y2YQ>!Pj&7{m8IFJPlRKDVQ9-McqKM2Ra#82jVXsCyrU<~p7Ypd}l zq^aS{)Q4nwh#VpT+;SPH25U&lT#dD3Hx=y?M|p9i!FQ_LY72I; z$!x^!Vylv&kPDv!TGwXrtTwji^5__1QCRRd= zg1=V(02XE*G3@RhMMz`3F}6ojvv>Z>UYWAUpD}KLVaX1JXY#Kx&@K!*LO$WSZy1kZvme&B z6H~cT2$ygyK2!Ct4V2&0^U4r~IQzz~+G+7N>xtPd)Zl*e z9Gb$>E@im5p4pj>$)qigpLo^spX;)Ogx!IjfPeby=dm!gMw7AYQf<@F;eI3AMFyp4 zVVFvd6CJ>Rv}k)$l&-G|>?8L;f!B=JJ88Te z}?`&P&bPw+QwjPa9%J*OgUnn^1 z9y3xdf~I%I3w_`vy`E6<=m?{7~R2B{3p_&fr(A6)K!bf{HX-7 zfwhQN!Nz_5wM7X@NIeZE1KqH#S#!BL=dEA3ZVzh6cFd#`QrO9%Qose^inD2@+Bqg$ zbZnk7GgaXS<{V^l%T6ku1FUT7M$ zo+p~=EF@@5<%1k!1+$v;JKKe|ja^u-4?9jXMXl_T!5o4@11Mk{85PG;X{$4~5NNlY zFmOrvagNnSSmrkcz&kVj;(tnfn&MkdA$yP{J%}AYDzAU8OKt<3q8G2op#K0$<*iQe z(VZ0BuWK0k#)SH$(XlR}1~>#7=dU~oEYk)G89=}!xIdzr>Hh%XDvse>ZPbcE+lCnb z0P9p|R+e4Lw-Ou-^4)8}oT|}&~_%pVS9c|Uy-3+j?Ke^nJ@{uR^;Psup< z4gUbvtI}%oeV5PkWWngW%}EvMk|18`l#!ekHJw_x$h~A#DJRi=#&zF>rwuDd8;>!$ z%W?dvmO31kx_+a4>Q^@M#BJ0CyD)We z6G=H~bYWhaO}i3!XGYGK1k%EiMJ2OFRfa*y{JU3^c%Q@f9v;@^x0W^#le!kc0hILr z01Eb92T^$RD@F5&*=j%@o|P7{29vC45{*$FXyG{yjr+tts!#H-TL*}#3htcU(`@0d zL!OF#tav!uIu2{5@coP$g@oQ@XzmV5h0L3K?H;uGMun&AdT`StzHc$Wklkz`5_(`~ zBNgnH+9sQ45Lhgrj1GQS#(z9lzia4^BW7UOJP=qUwvZ~RjAV$yxgED-zxAqjSF$$a zu;6z*S1IC+e^Ao=OL=QyaMtn1ghh-10LE(&_`6>$LDeSwK>q*=wW6~Qr(JnGv8u6# zPDVk&!K|5Xu1+wlK8sl^;>~m@AKFrJ>Ie8%j)~)4R^L*HTuzrUNWocTLC~LUR#II? zo!y3+XKoLflj)j`o=`q-Gdah4^$l*u>7(;>i|Arwj72JpAEE7Bo9NyxnC>vnIq0v) z^QePUS4B<^JDkCphaCR36t@gBhhg+4x@(Oa#nR+SYUFz=)d%q3iOKn_Tk!+@YV1js zX7?9Y0yUL^#{{vik3rToKMmeI#>ieqK*-NLcLt-oXpM{wj6IKd{qgSzqegr0Yk?Q#}xL^Z+q%Im|P zC$)LB(o*ITxtxq{6{B|qx*k;Wypv=T*QGjH7^vWnTAF)N0|!i%>&-y!(ha!9Pf$IJ z_H?=};~V>P$o|fEAB2&Q>t9M&%-^%LFCUMu$sB)W$TN@nYZZMnK#zcr}^I6!s^P zR+isY%M+yOj>P9X9crBFq?eL<7td`hcSbLUuL=?Unq0UH4303qzgm_} zH{rC=4abCZHB0*>5kLz^EV%ixfOz!h^Q+?06`k%OxzbYVMhs^YlH7Fd(zL|S7_pJ| zal97$Yk!|YZ5uo8VQX5w^j;ymg+qvh#T2Xx{+_H4dZ!$hk+B2qO_uY1YjtcN`|hCE zWVk*dzI3v^o+g)aOKy=e9-TLHSr^b2m3FZ`aqUBm)`Pj8al5W7X4hA;jR>Dlibp@kj(TJKn#1qV(lFzt zQ;KAgNX;amyDbWlt#z!KY&I*f?`u4(PWn!VZ+!tXzj zG@#qusU>|2cI$4e>>*VF{!!Zkt?E}sOAAlk0VA`Wz+;XDQ@F5}+IZxQx0K`@^{cF| z2#`X|zyLBc(0kO~GLmIzRgyATwm>9gR%F@vdZ}V^F~x04vdI+BEK-$B5D7lMl}_5; z;^yKtg(D+=HjHG|QzD_7>R9J)e+rDnBVl&!C!jTE>|EPiW#b^Bu!wf$BwW+31ZdQios_P8etDRdhJZ-Gu}w^S`04p2BEtzS5gF z{`B%su&+ND=k`xS2^h8S7%imOT0VD5Ml-?Uo&A?(s#+x0b1*BFVn#FgS3qvya=We2 z{uKP#%%CDHSnQwyT(WXe(fh|@NlRUcbgL+A;aH*VwXNzUV(sl)M>F;bbI@bABm8MagxTO@ z@fEz6QHa39am5~ZVz?|nt#!FGn8p?t3m9`7Z3nJ6rhT4F_VoM0y=k?5F@78iZ?Xe; zTOMwC>t1U%3osoC&0#{jt9z}EEG8w_=DOL~qZN}&AVZDExaPex;b7S%{yEzOU>{yh zaQbJy`QtYxX+D-nVqtrM$iB0)^;yfKTaB z#c6+N51J(4bY;fh#8+2!;y5hkk|ee=I=hB!fHS!N07^@56k1#mC$(tKJB`A*Cn;)@ zJE_76ElyfX+lypZ5j5MFC_s56)yVEv+!bTA?HLEHbq4!Io-zoG4tl05ls*iS?&eFg zAt9f9WU)8~vy8OVxkcYWy4~W)h{SF1LYYRCb#gp6Qd3gEnVrNv@xL@J<23l+%Vipq?Wo`%)oB?ohf z$jHF_YI{IU1NqwbFEWp7c4DiH)c=i?Z_=rxfCE0XGsX|IHvplcj36XW1 zbU%4**Ep;lE6S4O37wU4xWLX%?kl4Bg<&POxfjgux)4cj+@33%vW%IR*LrvV0BXLg z2M$~;SCQaio$(T9qiWYzK}^>wiVVn$5^^}nMPhsl4R=(8d=RdNK!c0b2dFct!Kab&D}fJzNV>l zYdxy~P{zUAw@>nG5*9LA#OITaX)v&pw;uJIBH)G?>sRGGRWiHIJ?a@m*8c!_`--`3 z8MU=yWO*{h^1yMP+*No`@w8>T3c+6vYg6dgvD|7`5!}f$yl#cK1aO^rUcD;{+Z{M^ z?Dy+==fXeMMsw|sX~535F@2tBWh$I6AcKwp=DA%b!JZ+sONV_)%#s|8G5715vHt)F zcC4wnZln^HC7om-n&5Cu^c3G=F0OU?)#JXlk;d)VTrh8J^{p*ZNg%O~VHs_)eCv*J z-}3&IS`8n>w)jzT<|yMXd9CY9T}9*q=GrDqVX|?N?dw@s^set(@Xy*LM26lo3N92e$Q6#>2a;Qeq#jg7 z4ZyEN!!ozGi)yBiUfuhUr(#rn;x?Sc# zqX6d>>c^Mb?bPFG`#t&`e}-+XuDoMu4c)wRLSu8a$8a(av9A9BRoA4joTPSE%@#N% zq|g3`Tt|kK#p4?@pZQ}i@)g&3k4?GMbZ45ynJ%F_lywp>abEN#7{>BEbq6U~9$Dg9 zm_!xRO`5owMb+BaCutq;LD$o8M7A%MS}G zZUckUnq*QuQmgF2Rt!1xH7R03f@&6!2`~viQ9vEj{6<$rR?_o=4tIJ~a(MA(kcG9D zFi$6bYm4CbG;Ybq6%c}gv`U=Z_dPlj2D4}zOVk71k&|yDl_dKb=5;Bx8%vWF%1;b{ z^ky~8BzJ8xwW7*hd&CY$G*k)9sOfaMG?MU&PUNnOZbrT|T28ofY#5 z1QHwu-g!Ud8ol9f58wE+XE!>mV<>P1Nf zhJ&vu^5uct^#|o)_*Q+djQl&MT1RUTxhWd$8BW!esrZw^SN{OK@-88tsFo42zcEwF z37s zD6DUkErIb{6p~HP10;uSmu$r!ReZ!{{Rul`?-qsy-6c?zF8Y>Pt;g zH1g9bw5=ZTj2(fcPJUp>mab}B!@{d}?A zbEo?M0I&Z5>r%hMe-WfDIcr9YDPlz3y|}Aq#TV-sU-(OGPC6*GIR609YeLJ#TCB6d z2mB=#mr=U}Hu1nzusv{bSt#arTE^co`-gGhzXNz~_rdyQ&E}D9EbjnmBw%t+b6R#j z4De>5<_NS%S7T%5Qa29FIp}H+58eoT2cuf~fg0i*7Tt^>``r5TkH)OUe?7jb_P;*T zJ1CSOiBHbzPbBl(u&ps!lB0@^3uyO!udtOnrp|X$(R?Lis84J(i*MaFFd}j@^%ar! zeF(7$pxE*1lGcxjqw_V*LH4tOHhx(E>_6Hc%Cl8le&Yc#Z9EHr4u9|EXz(*~a`v*+ z);O4?lY%CK-@}Z}_KhOWH2{)TWjOt7l=y*dp&c5^^G1dR24{_)BE%dhIOLk!x{f)l zCWbj4BI|`^xpBg1?PG8(tqEMReK!0G5D$m1o2^&0(iT4tHE35&r-pe_E}f0Jai0 z_oZW>%C;xdlIYr-rQWfE$cLcM{{X6_X9sO-A|aX8fMJEm>rQ2NIP~v|+3^g!wIK=s zs}g~Q-~xH9xUOfmaM8I_`PCs6!efFv(lYv1m-eKN!_Bs}mjn2s1N9iHw-Y4Ikpe~R znaAr*HU&I$fm>Quu?CwQsdAFIa;&>YwPP5;KP!G>x@|v67n*!B5FZ0?PL!u>YE*B1 z3g=e-)70W<5#ow8A;TWL`&D9DwVTkH1n&pwT?Bed!5K)S``mCk*Ev1RntrWuG+-A} zz=rSZQmpnV-8-|Qw7Yvnf(w%vjm}iRP$~DCs>0q%E=f4xb@Z-TuQd%)TX&HIENzff zV*`)yuC8nRBuZgY7jDt8WDIl7X6dAAsy0`K;^N{vsM#5DfzMvG4x4NSqhjviUw_KD z_Z{m}-%x1{+=&{f`A7$o*0N>2o9(t>gu+RY$2{ViRurXj4-g%il##1$%Bd&c1M{q= z!s@o_tF(=|$8%F^cP(*#(mH@}qmQLiiOcE`wh!J~)h!NlPem^Jpp8fXxo^ug)#+Ad zSgs2V%qL|BKA5gUN4-lOPwf&6s)pn1R-eTeaNW%W1q3oXNB|t<`_>gHxig}bMF^dT z*htUG-lLMsXa$sj58+&Y_ND&-mN#2WPpRFLQVYwgoxAkceZ9$i{{TAA>gn}1`pMeq z(?Z&Z+4mD7Azsb3y6+9eJ8O+`_Gu%J$~k8_Ipd7?>0Fkmr0REm zo^$^IGhGy5psbEs({p!aAse=$ z)~~-dqd4NJ1t9e`l{lq)7*M95pPMXL3|@c0AO8SBHQxLuh{*bn{d0&P*0?(ni=Iy+ zH~#?8O?nT)m|52S(4UneEsvqCQPB1_BhVnYRJfLA1;+BD)ST2u!%ETyhJTc);fFP; z1j;R7<0s1iVx(QFz@80rHg!j1Ab4$rC7N%W`?$z8Rj;)IVz-TChyXrcU;edkTwEfo z*nANiH(vOuU~+7&i(_$(a8J^cOLkW^o`*56rR-Wf_ZKg@FrniO$r<(hs-K3YxNRF! z`xHbU52*=$7P&2prGGH@6Z{Hv1i#pS|lcJW&PV&40K9da>~ z{VCZv*^$iD?vBq!)Z^D0dwVG}Ew(rLpC~=J#(fQO-w`cuQcFp$l@&-Rp&NI(t2$<# zb7gyTEu8HMd}M)~o~QMwe0KK@Wn~@I`{droZy5svnsBEW%II^eDMrlUN6^&POPxXk z{@VWlI)@l+_NnbL_Q*j00G@-V*18U3z3`=~*?}@7*OxHJ!ZB<61#+%prMP%ObchSyiyVXdsMr;cL^Fq^O2P#z&=jc)07P?DVjY zO?nu*6|K;+m6-`?CwJub&Ij_Va=pc~H(j;3Ic?b81KU()_!u7gc4CYuKs z3=i`)mwTlp{kUDl8zAn+a6PMX(M6Sn5rXf-9n5*;=CHxiah=Wal{uqhV&4rIiTGjA z(ru-QWQ!xv8ZDt)+gT!KWn7HnmK_S#6xt<_jFZS8;aOEp>~+v;D^i4-#AAG|X51WS ze_FS1a+6!EivuVlyHt~CHj<#sl5MLoMHDeCD1Jm370K+Rc6(^_DKwbr7wT!vFo|%g zdH1abNb{Wjl?wSBmIno7YtzF+Y`6xz#6D8wdmGTO};|1qc2EQAm~vg7GeY+^pq?pGg-Q@cvU6 z>JKH(^VxILxCi`dh1)v#YW^JOB1ZsvvlIPmW_Pg{uC0K38YN>zrU-(6@oM1lI z=O68HT@^Bd&#i4_)WO(MN&|aRvN#gZ*mOt2X@(=&;Yp*ELg0Xd=+;TY7^m ziMx)YXHU|!_2UGd9lL1p{;$%vRG+j?JKuf>W&7Lx|)#od@ur zU!ku)w~R?Ts2D79Ur=4!L33>I$+eN!45W1x;Jz~OtiBw+c_Z9y9kUC1dlOEK-K{Oh zUYc*3CLQ!9-r?fsD_QQXxDsx3>`NY|qGOX=oXoN#1r+b(_RS9vi3Z}qy)jD$ zaGf$Lc^t|2x%8w4mxi_beHKZ~@VuMUVVAM1mwqm{xYPyD+1O!r^u|SVL@sfPTrM~q z(&bIuu6J%zQ~nYd^TbCLdrQ@aoe-l8q`t7pA zKS4TGb3elbhDM9+0iAd!cOR8 z5nX42t|W_4k+l6jNe(`BGG~vdi0v%yc2sB?x!P&UO7C5;nxQnmb;o!*ctF8AEjLy zjnAxq_vwGakL^3G@?yvRaZm87u*zSaXsiDK+P7cy$MJzb!kjgGwdgJX0Pr0@!ixc( z6dsgq>(;c_Tf9;CEl+-Jll&@0*RHevs?YJYkNham2(;~3Sldf)C7r$0QaF9G?%{{RYK?c0y}_ps0XdMEf%ag&XDz;b*$r<{Mc zC;tEyR@cSxsYZ)Lu|`k2z`j)a{p!pb?D_uyk6R!71yAs$8sy+(>~+H`g{az$>N5W2A^>&<8<#_eIc>2?|I@|yG3)+Ie{EPo4i6sxPqqd#J>KUf71 zQ5{9+#M!TIWuAR4Jjo7Q%e$2npP)FX7vdJ&71G~~bibSYYX&Q%pL~*9T7pkSj{trY z@3h`-IV2mEQvJ+3yJC_f|qG=R+0zvL$Y@4Na}N|`<#mySiy{{YZy z40%UegUX3xT>aJ0rEkZ_hb2C@L;nCltmHE_c_XJ{KL-esH>ix z>N=Ejmlqyv&n%1!G0Dw9vm;;@9MwOy5Pn!=A6%LR)QoTzG4w%N*hTL(GBDz#X`wLv z^)2aH-)WcwouBu>q&L&4z$x_DKf;RvXU$yw;rLf~;bzW|7+`|txeF*RC$|&qToZs6 zG0jVLtLhqjX=wz_2wqq!j;5k(nOLa1JD!sZ{{1+>ErG>(_4-@OtUP{Gre!MRwsX(Y ztawvayuI-ivRz!fgtiq}@m25S+pS7=z7}FLjx)NZE0x%CcRDz9Un(pnagMZ;Sx(T# zc4c@O$FQwguA;TK4=kb~j19-BtXtm`#~cu(QM0+gZFe59G?kZG~iwyRrTuxP$ za!IWW%NyhS6NZfWx%qhG6-1UJBpi`bw5}r~am7=4Sw~ZktqBr~R$2Q)GbsbhfBgj4 zeek~Z^y^Jc_NhjRHwNo~t<#F&Y+RYnPC<$P0Q3V}x-Pdali>YbB0|r0Vt_aIG7t2o zGrX5twY9Xo3d+xAYL58zBd7WGtb5-d=}mcW2A&$=$m$+R6sSN?=4+X<(r0}aQMsJ9 z`C={4QhQcrkDyNTiAdUWk;gSWshV>$)%8s_$pc*vHKd(=E(dPZqidBV?z7K?s1zS58$Q(QX+vH^wG@2d zboZ!2Nf;ZKgT`uD088jW>%DRL)}{xhd^&|@@cgUDadJra;cL-YGTW&uoE%pL@X|($ zLvh)sV~@dWx74+$Y(!|$lFT;)w;*x{Kb>%5BX)W?X;~Apc`&ST#;T*BQHrB$e3x@H zaGk|-k%Px1XY&TOgrLTZ@!__Nj<~8;)=6)x-Cpkh09zQ1W5)^uG0)WIx#LfnXwHdE z-($)CDXSk4TcIcAwqA$a*CD3DMbtp&nPVLH=DTl-0cG(#P3dlfr@5|j&dK7qft+w> zkNMZ99UytNJ~q#PjaYZZY$MtQjyWV<>Xn+)#SConl~qx=u^a#@INrlpvl3eB18L*r zJbpD46`5F#QZ8NU78Zvi=FbPv)G%srT`?Bc4B&NR!jHzPHlL^2MrNDLyLIS`z(1Wa zJVgw#`9{g%h;gxiAIq(8bClK>i8B;M9DV0rMtucSy@>g{vOZx*Z-b(Uf~0*(6=n9z zW*l}jUf_|9BA#lwVpN{hcIAFkREYx$!y~`FK+QcRWC-v4EJxyNO?MmvRqSV0fM5Uz z3M%gO0B|Z~wKaysQpiBs1yWXE{EE@cM1ziM5V4o(-m+-kG8>+mUfHN*IsP6^XfThs zs93~0oC=1nXr87#<;!rUkprCm?BcW}k__N5&N`ZPzD#YIV<)i+f5x(KUCWg8Gj_Oq z?QhDNa;_9*7>}iDB#I>_Nn#_=1DcWU?NofzD5!MPj^##eJ&ro_f7%g(I3iGa2hh}7 zu1AIMu%{kT=N|RZYB$z};|ns9Bm|GZ)>g4BE#Qm&uB4T2JLI{=YX~IOnVja`?u@%t zE#Yl3k~TXwxxn^3)$8)Z!c#Ht3Z6%74rdZ(@36wytY^3LlL z-~o&gw0~OBC3GwMr=R#-De;|yu^WOuGJ4mrM%*_YtH*pXy!g_}Fn&gj$sV}HdtIz+ znmQZF*&j0a zhr_oX9@L)JU`ln3@xu)D6_4d^IIpApS>fv~Yfg2R6Cd$f@b z7iu*{L)A< z&P`Naq%=b$o1ZL4K9uDk_o~G(cr_xNjt@0Nk`@4te+qAybDDf;C7lqQ0&q=iYC2Rl ziyj<4mO&lR%rJ06}95816yeEi;ET}5_w-r#_+tiT7I<+)x2(F z2M%`RV!d+qZ9i4B*)u50@+*R@YkO{J>~>L=6;GCHQ;6^`rF~&-G;vPF7oxWyS5^N2 z30|lZIgitYtoutFO9hwg{Hf1AwF^e9zbW^w#iBVR?#r@x!&6Xm<>&p{tbA*!=btG5 z0KjV+@;MOVsJgK{9+j+YX5Eg4Um9v_leR9M*0bjE)~cg{2>$?p)?|>%Vcq4i1HL${ zYnCgVo}KBt6s~&D!k-l1Ykn=dwYgY=C5y;54ssqb{cG6k=l=kKtWSMw(z2E)vDE`Xr>B{q%E3IuQ!#AdX|bA4qB6*$MsI(e;D>0IyQyukpYA z_GwGDFeH$Fz|9=U2!@n|^t~TgKl=0yDnI&s{{X;${=H1HzId8p^pU^Do##pa06!Cd z-`4*C8hpqK=Sl!CLD0U7``38?0IzTUy(jimA+hG&f8a6y02;OSiMJeFPl3n!=Kla1 zX}P%_F0KdtwrX6+WpC`5hgoiM{{UY90R4Inmlyo@xc>mabN)4LZtlnYz1?U2yEpjL zh2`PkUfmD-YTx5enE_raD54k=%18TF=lp0Ev8ce1{R}Nue%X2sMg8CX{crK7eY*G` z`2PU8$NhCb<3}C<_q zbOmcGU$pBE>$t}N4ZH5~&}Tnd-?>Cj4gA0J@-vV3YND#9p>o6$&1(}9dy|ZlNeIt| zq{-=Y5B@t*EnUpqZqvC*K-@Aac~9<0_NY*DIP|Ma8Ij{oK4($-(q@S7V1;70jYrVJh-IIMXi zLc|Oy{Hj>AtJ#-)cJep}94&hty`7w`nRJNaZsoS0LKaf3 z+n16dl`MUD{OL(5;nDlpS6UIkB&b^bw7rQhItcz9BVfCL1dihrn*|e`v+%gHwSDl6um>!C>pxPnw6VKHjTp+!o%zCn^se8?BuKmR zO85HLGvXV3UE8}2h^Ti+*a|rv%?au$Da|zDskoje$d2Y(*n9U!4c#rBj3GpRQQg#d?-zUt^ovc4oxWpXLI-Ur>Psb zj~rBN<0lDIVu_&I(hOkYypK~vTl=LOffKnKkdNX$>(O;xb`2&tjGKJPTqJ6LQ(kqb zux(Q8vV>qRnHFpJp$P{@BWG$XODa7QH(~kA4G9Zz# z#8hmckWVPHHpRHRUh)r{{Zv@U3Y?XNVS_8lHjn8DUsPnbF}07 z*Da!v6qO(ywKX>&ZhR86$`HL5mFXM>qqFaZC;UX5&NB zESk-3gWR`B5E&i60n_xXJ7K9@owl~gV;(;I-#obXQNsQvw{-6aCAdCY#}c2sV@xxB zSn_`w+qKdy?M4Pk>B*>W#;IKAwEYb&OJ2@oxw=8QWw>q32mQTthTOHZUc_s z)DWUAlI5R>89Hk0?5;=6QG2P@msxo34Nqq39tY$iTq#~f0W=YT3i zLS#QCYD{B*y!uu#j;8KbN0#{WNrL|X#7OebG-Y5wLMBI3isVwt;>S~G)GkWcMCl}K z&6aE)G6%M6(lz_7F5_L3&ACEVP-S_+Q^>41yelrXug!k(d3MRmpNQ>Rc@eHKsgw# zscp$)$rVm}DGKo8@~D+3Ygpc$MPy(}WQOHV=E(<)`&NFX2bbX6(g`OWPu99t)29=24i^(ojk4cP<57V)==b*Da4e@Ww}vEH;e(;!o*bzLGfM zM>}iDyec<%?!=4(_IW?0djc*a%XP(fRsI>vOZbFYmjLahXJEr4994xQ0-(F)fE)_t zM@*iCere7yYs`Kj_)tkd_N}nMCldkN-n~X3y~w6s+9lPq#^@OsVU&)wP?L;_X*k)R zF+IeVH-tqdals;-WiAVKCzD+djyxSciKrWkXxXh?xt}B${LuZCOOh$FZ%e+xYHbjTKvRj2zZ<`egcj z%SAfisl`X1pM%XNhMCrAJ}#5P_A)}&YcjVv?khgy#nQ_wu#aeBP6ukQ14d#iCOUQP zR4#20zGM5o)S?X|ZrjFETcasVf+_z1@R{+W%;80{%#t$Z&zuvMW zamP*%sitiQs2_zc;evC=aZMy=p~WyAuy8@9vYzLW%_i_U!OvpTD z_z_C0&stzF1=|OV;*564Ii$`@eN8xirh$ei3@{EUPe2dhNq`0krU2RP+JnUk8HBds`k;}ij+%zob?^|FVL%9{uLi&FFd053!T070dbCsA+u z^M;n&AMlSzct7bw{{TUul2nXk#w$Y6OJCY49cEgo5W@=7(hxqeYrC@(&(fuh9h_?& z(=z#9Vv&YJ-nG&hkpcpdDBvO}z^HWKi4(X`Il%e~o9zd4WS+Pl)t$>9vQfvHe<4ak zuT~vB1vBjjJx@w$xMDjVeJBeP1|P(~l_T7QRtlr0ML=d|>vw`A<&uRttslg2*3ofsg>_ z@$FGLDymrE(@t@cerjC9u`#&ZN~~yr;GQynI^z6cBOCaK$s=sZX&X4_usjOta#h&w zuQ&1RoAtQYIV%VssjGv#CQCvYh1}=^J35cy&OrP+RaT6tJAodRd9=s`w(%qx#6j7( z$>aP@SJft$O`q(sf2(K~RXt8~@9kS(TN$0Fgzqf0?+;CLdoYa_W0w2}LEO_@$B-tq z90TRkj9F%6%Wgc6(~riys zcG^<~cLGdj%L9Ak=s&J2m}{)uNZ@>|L^=Nebk~{KjLz8q0G?8R=mxw0 z0EhB`PuyVyt27A^`w}-CpAwjF%{^dXmYivrV{0Q}G+cTUBX_$_kY22F;L#9P-h~Y(! zA<6^uIWs15D*d+U;=$Bi1?nqdTF1+LrK@@DaUY@$Cs)~ zLw5t7wPMaG1vSFYMz=GHyE>#C#u*^MRdl7S!@^!B20@~PSh^rbfp4_bMYJ~Ld(&6~>h zDxt#+C0mJsLzSlQRXD&j<06oVO(BHjM-R>>}Vny(t}kYm=P`#483gU6*@ zw1}77M-*zX>v*JMm9<;t!Zfe1BqaUq8@{QbbD*@;pPE?*fGf39? z&pDt<$2(5~o}xz>{RIF-VQdaRI%`7N>;+pE&OpihDnVz?O*^63pd%#WrHVH02_Cg+ zqk=wr(>&eos!kU@C~Q_K_aKgPX-DQMMghs^ttZ{8Ksfa2NPeQ4&z{tsvU>KW1F|^l znle4fJkoQH4IOC!6N+#==h}<{XvZ7@iePCs7tQZy6*|NZ_(x^P&-zgR0MKc!NWSC# ze4?o=7yc>7-w`n* zg|}n*kw|Fj(F|~f7{KQqw5t{r26Yu?-~dQb`1({k6pl$J`PLIgzu$8K#I>@=i~^080cQ${d{I)~_9k zr{oyx(yQAJ599dM(wrg5>N*OBX&sC&6eMx@e(GdEdSHGu>pkCMfyw!dV~@g_;(0v3 z4BcRj;Q;{i{6qX|eKE6PZ|mibe+s4WlKvvf!0ZDzZ+bkJ?fmNU4>6S#s4dfZM0+LN}8sCO>eN$Iyqq3Aov@YjdewFNAbZR(p_)u;@ zJTR?e1hq4BZ1P6ab)>WN8%!%SSViYMmmM)&_N}IBH&;n!rESe#H!zS3)?UXD58D6n?enFJWUWq2Ad; zgPqDTT$ZEYcGNYQudYG6-61ih#H7a>_QDVAPpH<%D3|a{A&*J3mr!-!x($&h)zAL` zTCFXO?87(L+f)91EWTC$0D(E={uSFp;e`<@$ufYw5flYqpy^VN4MV#kZ_<-!q>e{U zvYK665s2WnRAS2#V73(XZl=8T;=s=p?(yj`*jimk4)-y)KZST?gXStaW}HPK6^P(^ zRE54#kH(@!CQNk|dE-r-rZewNl2uW-obV}%3TQESX)zc^fNAXRJXZ~2Ti5O3&_dO zLyt<)RgrSiV15O-C;;twq#sJT~Nf^VP6rW*6!$biwFWp{!4ROB_>=yiK zk$KM(gChciIpK3%(xxRRU>sLD;*Th6x+s_`f+rb2?t?^k6t+CAd@xJk>MRNM9jc|p zj;*HZBHqy5TtrkeNfMk4o`m4{uH#L#X|&py{A z{Mh-A>w#KbnAovv;~RetcyixQw?)3xQCJm(?^JB@oPBFL*G$!XOL#7>{KsiIX=MZi z9)NcG3c2Bb1$n5mw0=#f_T~PS=#WDUHtPgX?TH3GDs1j9Z7fpKG^<@A*g~R_Tq-qz=P9YQYde8ZpU63+tY>ZA$tOJ9A3g0_R}^ zu+14%Sv#)IX&w1f;iAJ0lU=I$5SaQGd4(!ZbOV?prwp67AQiUy~RFOGctxbBh#9e zHQ0J_Q}&Br36VTX{I~$;CkL)5g(E;n1OhT@#P+UcEE9q=$4YX;y#P`G6sEKadx#3` z&pmNY1O?AL)P@Y;Zs|vm7>phO#bR_^jFHE^GE4#oIjcWs$J{v z7Tc4Wh_GysPfX^n8!~ct4B(H>m@L~3h8~s9;M+A*u~k@OB%0K)ln8Q92a2cmC)z+| z$u)8dR}kP4!6g25bl}rDac`Ol6lFV@ka#54ed0T&(0Mqmb88@6W4AojS*;}_?wXeg zb~lWYMh(@prT{WBr!}3c+q`}lyvZFGKb>^)+0N0h@|&e%c#at&@Ws!W8!kZ49`$sS zOv<*q7Q8-j;maHzJdwpaUj?*VfExs@JHxO!@ZF09I57iKYwkY10Nbpo?wB~C{*r>FxQ*c~f1 zu-WZZ9fbEGX$z5@3YmEv9Mc&;Pg+rt(x{S7PZVb~z;X|BN=e9GMrZ-p$2iZuCQfR} zg@jwXRb#*%s%}y}N8mF9<4f`1`#sP8M2^0Nub+MdS<4@cuP#hvMA~zxZrSrA{{Ra5 z3X%qVO3}*48%;?atFh!`@u<3RIrpVe&PnGzMJ`Y-t-7EY>*-Ud9k%@|KWXYInA+Le zYFsWf&0jGT;~;%1ywfqx#cUqevU#%YHmsHqm_NA&*yzC}3AApR~#wKN1cB=n^b zoPE>zRgf;4GW*U5%|>UOC_=}tUi2mdb`kt7^~F2OZ|=7~)V^c)Vt*QT_QrQN=Rk&t z-#7@|ds2vHJrMIqX4-!F>+4SrSaY0H0^Ev5p@J}e#}KQ!T#Ij@N?e1w$NYWq{{Vc@!4}$EX%Mg~t&gQ{_zp&kRk~z+e&7Oqi+}a2jZik{ zHPU!SINIvy$O*)2j4>VRB&`w0bGN$GkVt(q`PB(Ek;cURD>nY++Cr#@1M;kNzKUCq zDK2AaNAN3iS)^NbI=J;#+m;v}>wfY>v$Bo_K2|H(^j$~C+HRN( za07QR?3^h2pYW@M#Zm5h040oKJGAwU4K4KJL0DD#6vW)Uc`kH^5 zJ*mO+I*JCgBU?~T4L&O{G7oy0NXG{>@Z9ExkPJ&jSNe=s4e=*UyO`;7*skY+q-Iq= zF$4j?u8&mlmWiZYeV*?0EWl@KA$)#Sb4Atky?!^iv)?+Q+GLUv27Lh{sxV~ET3GT8 zHp5(2l4OQNln_2;Ad%=Z*P8FNuM1vHYZThR4psLgO1a0m{Oh*~0OUlU_sFS4qyf(# zT5bueGb@<_#z#@|GN|YPz!fT3GhlKDrYlfBd(!Urs?lf(T&o?5q1?<#>@m`%@_FZ< z!lvCxyP8F<3e5;4l1ZlddvbZrPVL;(CHbTsMtMi5>)VQFlEXOad*ZBb7zUIuV}aJB zqC0p4Ose1!ob;>msEtPp$8$?I z;}Vd+k2Pi{+&*5$rM1jg*(8$OI$-o7s=PrC)c^s~u6g(zsOjrf-cb1@1;8Pr!+5w~=vmTG%2#!W+TN0fQTBl(J?E`*!T-nGp5*^78i=;zQ@+AuIaPfe%w zu5ZQQw}$Q%9P$tJripu+o*N{1S5NkVYCjdZUkzR!pDjt@rU$~hcn3Izf1NYM#7CVpX&SRAMOzD$I28jZ^r|3@oMKzNcj($mHnj!&&dc^UPn;rV01?LlmRp0#FdQh3QcQ+IX8r8J)W z&`t$1M+WDL8MwebDTML$pq@COWXAc{GNq}hFPJm_O;utrMh$g774XEGQf2DgR2$p*J$F77^+v2mtEE+c7Y@?mej<#z%24r&%n7TV@?LveOlN0R6F zU>tQLl4`(9OEY%FDciJn6>>cs;O>(>`c-Q={?BVHiTBF|T>8~hd2tl`jEqitXCjuP zC%%DgrV)kRkPiZ+oLP;#Hvxg1`qWozCFY^FECaM;p4?T2)2*gJttlruSsM%;O*lGI~7vb{n4Dtl#DxJ5lF z8vZ24+jZ%V{M1cz=D(Pz6CHr02D8L`(enVf=Re_8Zc$!f0ne@~KekP`8_qG*iek*l zLC+(eXs}#%(h1JIkyfW>Q^6O0%C0P9u+21w;eA^Ymvmma-pXe}O7g;QN0 zgaj5IBoBEc5hB~lLrQQnwDkA%p(L~o#?RvgXcR|ezIM@;X5&9WT>a*wcc>iADck+f zFr)Al3ysnH{=6d3TbwlZk$%m(PBZ2yJelet^NR2s~B2&K<(Vp=CIN&>3S+WL5hIBrnY0# zV2EK96YEY%QYeC;;0$BdfpLM;y=3ODXrd!ot7H>Uoi+l*mFM23HwQk{!#+l zpGwHNj^VGSKnV)ORhV@RZ&z)$7T?+SNB&-}S28&N0A)LUO>~JIY6lr?a!-0qSkWAs zS<2l|kUs%dqqUb93D2k%(pr?WdZuE+|fFhPSWM>VJEwi7cd6}BT`gH4>_Iue! z*lq#G`nLW?zIz*a$OqoEjBaI7o13gp8^2n1+g2bW6ZEVQqZ|scZ}Lj4J9CWHauIW3 z?cf6}SbLFBm~FZVaZp}eJo`q_Fnb!LL$sN}#WaIMPsoX5m6lLQicBy30M~Kw)PbVh zdX|s>070)ffH0Z4I5A280H8H`C&LUbk$7AFdPk4@;)k%-=YFZjdX;0%R}Sn!Z2JMl zT#tBM<0sIX$(teT0V~L<7Tk5H{KC1(IrIjz^)C?F>3VbD&o<~#WGr7XY<3u?tPbX! z-c`I0G|nT523?roDEBqzm!372>OD@dEE6~GE{j^?>te_qwRWPGcF_I4Qt zJ9yi${_*Z>t?*BVPMsuLjLi8#ln!}3eJW)%nn;dgH^M&+uXVQ?mbvCdYTq!7oJlX& z(0;Y+&{;)gZkG1(EKs1rq6`2kiPfepg#hM|>az60lb+Sla~5fAY84Gqj{asi+P?nP zWL&r(bY_czVOEr-F=sgBbJ$VmrXQ8Zi5}qwIFTB0~n&>>Mmcl4LL5NfaYEgYGY@etDiwx_ib~vyt`$Tj*3n* zQ6|zZ*EW+`w&}a|sYT=yFUcV1wrkAw%^zIUZPZ)Zi%EB6m`^zR)_#%UPZdXUrhQuG zBnKn{82+^iq>SYhk5G^dlDGhzW}IUE-tRI0S-9JgWf&eGJo@4&l1NjQq zTTpiHbMlM<=~^vbq4r8oYjjYb?Mso^e5iT<0KR)uRwhMfmhO46z*Z^-PtXd~HnEih zCjeknFKH71GH^5Ai7f&fnZk7du*tyV>)N7@UQ2H{P)Yuu*0h^Xk8aS!bDC+fj3H3n z!ThTS7Imp3Jtc74nnG1_TL(Qq$*5N9UodWvfH^SfP@(jqtAMnrh6zKH{OE1iDTc_fGrCAzcsU$Jv5&6)XYR2H_I3AwB ze@f2NTN+*^VX8~!0nQ2U_5T0?Qw>q~kTH$eAD9#kI3uafNXPa602&uh0g2t5ob>n2 zQuc;&<$DuJe83*YsmpxLoafwPts7u-k&jBJEw3lA&lOG1!jCBn-QXSv^{)}sVq0t ze(5v(G~Fr|eKkW2G8{1X&VQ{l#BdbQuCu_~N)wW)Z!+9t?CyC=jO&k|`SO4L32V}H zi*M~MKSYhr1+uf}>0WA>I@F%lPt&)|E7*J|4m>$-Nco;TkEM21dYn}EH3p}Mf!YoZ z4K=S^0U-OVdR1{`;W5~nm7>kRlrUI+mCF|ObeYGcDbuAGAdHVwLurx*+=GHS0Q9Z3 zGthXymK0WNX(n9o(GW=d2^DJBREk+ZDKa1S;wx3H?QUb-nEF+iyw#L$LO%-CT;`h8 z(6}iqSO=VOn&Z4fqg(2Bl0_l|#~#Ply2&M8L9ChDKQ+E#?^7C$%Ed=R=e=J^mrJ{n z2Oao6qO!^yn$i<54uV871elo@3aOY2#YD_1XUCELjjIKskO>mYI+~Ty=v!F@h!x2#8YjfHhuS}_}8abnS3_8iuYHVXjJX{ zGkn@Dw;LhUt`w4mjuh0-2uapt05BLRdUIJbYp*geYdb8r@rcME`i^S2)h>dBX-QG} zqF5y)H5BNC@`&npQ$o0CUc2PzDV-R0;ryCJ^VYcoiZh!~?nSQtb*kJko+Q z%>Xbb9e*ks5>7zQI*OS}bg2}lYC{Q!`f3jU0AfK?Y4H=H!JZG>8NQz-GUNV2OaA~~ zY5ePZ!#WI_?TT8Yg%n(tcmC?_p@KQBQDoF6hA0%YaRN9UNvrlA8<>G4@sa+`dlsGH zeIHDSBsQW(=w41qtp>5QE040rxBc2%Xcs)8H1&yoR0AH>oqwhMp|gx*xUW>zY_wZj zMw;?y?GS$(z6a%AW2yMBM7FpwX~{ODDL7rAq7(J%e<4CadzsK^MdSfgxjlf%tCP#8 zXxRHz)KR+s0Lv)Ac@N+^ek4;3pNchv{{T#yEgs2-5hy6kKiwZO{A%@whP16YgF|X= z=N)Fc8-PB@IQ;WY%cyElx79U$TR+%(Sb_n;yt=^)@00S6!zQMW!#2+_xYn=d)oxGl zhbtKW0D&G);01KT(&FfYGn@_=J%^wj1wC|re(|Yp+YZd79wLNW+eIC`AM?ru+NbO5 z^*OCedrYeNga8Lhn%_@IuzHc}R<r)7pJF6Hj~>Tt}Un-%Vw);I)#Rh z4*Tnri?~tCH_>{U&ykmKSR1+HwLy1xWu-^vUR*~IJgS~qzJsZ(!Kmw6*v6L+*euUm ztAUqc{>c9T>mS0d>E0Z(vzP5MZ@Rl~T50}cpI~^cV+GXCQY*)Ot7^;t021qQVQP9@ zT&`Wc)hhLlg==dlxRZ6=-%~xo+~0nF|#Jp0yA-?g00v zZpLtU&lwe)G>I?FJAp~KSFJONs^nq~(1Wknq0CxiIZ@n~9< zas&2aaqbOj%V9f|X0eeQ9pZx{*)&#(<*+~S3Le@HJDtMYWA3t+^*yR&LC`uD>7J&U@~Sg& z9BK(>W5D+&yw}8^6P2bqcZZP67UHwr#<@83Px1V!ENje>)A*Ogb~-Axq+xF_m=IzO z@~8Xf`d68}_u7UbX8AW|VvTmMdYWTEsN9mI`+CxV>`5Vq;XqDh3RA z3P9~ryS759KjTPPsM;Hz{M3{{#fDWu0C&YM8CPyO?rKJ6MdY2wpTpLXOMT)_I}cM# zkO6{nPUGLuRJQ4yvv%oBk8_-4`jb^YX2h;x%t0vyx!;8ht6EsAh{*ruFj zo1SSE(3H!LDTT+aM&DX*T2gur8zKr?vl_U|*y&)%`TIu{*;rT2s zTIMAtCOqS@u3QX`Ykl=bO*F>HyVnPr&3`i5ZOD2a)vLE1jYloGkx#7xH+(;<$9kgP zVe_5_Ye!S?yzT_H&^}K)KUbk3U8bY@R%f|04_U(~dv;saq7Tyw-)OlLM=l+^Tf2CR)Cx!$* zYfGx4dgeOG%BapdbAeRvH5nx$Jncn4ba$-z+>5eC`)S?~BmO_NMfxOX`qZhTcw#}A zwH-w9-U!eB&0Kkx-lU4=Nb<_W5&G3@&{s!XbPo>oI-U3*_2`HH0B5Lwv3xd07h1d| z^ajX(@pY93vYt^Bh9?-~p{QlKNf5kWARVexmr>dzvqNJ9@<$cZ$eH2HVNd0s>rE?X z{{TM^{{Vas@vOF+a}r6-CiA${u_lv?aDR6zH>qF+J#EnZ?8nRQREo+3{c7q;Oa)Kr zO2A$bS!3mSN&f(73{`<>CBresdSR7E>sX$7?fTW2t{z3hIUs(NSP`WrpV=Nf_4%ex z=}wDGUQDvX(~L^Bo+S;I!Ry|hBm=1b0PEAR8jE9Wa6=}4>#+X-3a~V09bz8ey&vIF z&n=7TDI_tXT!{*}0QBrZsv>fuC5X?xN>>@HbE90X*tznb$ondwexo4djQ8TZPlr*Z zy_Lw$3rU}wgV0weKEHCdcDi&C2ia@A5m%OGU_X^M9UI2_nwx1H*g)rVEM-?C+=^%s z={nboG|d`6-dx0A&^Ac?2mb(CxSPKp>sq{q*Ted2IsX8jy9xYx;8o2t;j}iv9_{|w zFg$sTRTuiV>0M+R3<(H@+vFViaVU*HA=mNE8X?-pmN$g&H7l>Rc!N}Vq$egA8_Z$t zyY8>+T`rlU!>5h9Wr|xIW6WQkQ}uJx@WppA>9%nI76s$6sl!7E+NhxVD5Z8u$0mz% z=X0av^<&i4)zZ9>K0}ZzqG?&X1r-+qKQH2HSj>TL%Zx|Itv||Oa7RB{(mPo43sM!& zU=OLKVyyG-8&HBqwZKisdfy&c z><|9{Ub+z&Rl_jAIKuU(Dtlmh_oDrPX3@yOpamUzder4q0lV-5n#jCv;Pe1ckt~GC z&&s>1t?ksXSY1yDxBbEwe=fHTl?ewCoM?Ee6C5WTZmR~m1beRXc}tEB0S9x^b$ z;Z}*gk2chG+s#5^M3NRf4ZTm$54t{_``4=J5l=RgX!i-SQxcaA*EsK9O&+NQuZ(T< ziwFdE{{U%NrIopP20$Pm;U7x&_mh$W@;?f5R~mYa7V4{#f29Y@BWPkX)PYFK>@viG z=}C-(j4J2zsIg>oB!6R@=fdo4oO@T#`!v9K*CPk%ApVu__b2@lZU|6~^*mS0&Yv@2 zcdCgJXG*DQpOeH$KN&~#sntUfx5TNF!2Ror<&h6%$g7qp4gI!9%jT#c{c41cuxWan zc-yX=`%c1?S?YmZ*Tg{Wq;ewHx{V$ZxKj&i&Y3+qY-8C*Hkt!Z&t)9?b{rZKMz{LVRYJLZFCt)bY|tfahl~XE!JIP{HvEibs#~UfHxk4^r6TTv1`R%BGf!r zn`qigJu^H*8Nrj?≀%Uki9vNlK(Kkq&-o4o-g>=zKfj2D%GzdJJ%OL&1OW72HIT zTgD-_j3i@^lwzlr>dYGxPqBwiwODP{RQF@l)s=;dZe|(%O*IfE3k;+iP7Yoz5$694ra>T=h z$GOcbM#?xDKgxi7zxThF{{UW?LEE^3M{IYbis(1XZc8aW@#{?}?0#W^>rEw{i8%w* z^rp0j01SatN+jCHfDQ*wY8@C2st>vK{{ZV#Llql;i@DF~N~EaV+~@9(^V+gzt6{!b zQIHPf`hWH55)wXUQ^!H;Pf*NBB!D`T=|Ofa#(eVLwW4*X?_=caoKqFAVaFI=AHn84>5sksA=Dj7)~1{n9pK}CY&;RkhDBZ{jD za7!^eSDbtH6(Y2ah0iCCr?2HyjM(NJ(+H%I&hvwSJ-ZIS;YjRJS3i3jkN8yBr1TJ} zq)=&h5RS(ja6YslW^lZ9{U~bSfPyK9JX4F3GsZdmyMDB;88`qFjGwJ@&MfJKh!5VN z0mV(X+gzt#!>c;nFFt_N3vym zL_|`6o$31CXmg|Wxp{TnsZ{{Uq;V8S*BuLs)|R^vpLLQj^% zInU!xoF6c!IJoX+6vvi$IV669s_SQdk30GSTffoUko7X7ZY=pwJX>CpljBAIJxlcnLK78UR4MB>#UAG&(rZ0>s}sKUk=$e z-;_i%T@}^ra!__~A?cXnI;ETaQX}4Xi3i!7_doEb#zCdelanwb0zw%&6v^r;2kRKD41W5)D@? zy~NtEWD;sfhF=vP0ArE?%}WjzKzi3qR);K}qFj;X%CAo)*R3gfT-7{wHX!82`R+;S_jztm)~Z!rKX+#0R_00|RZ zW93i zB@(WQrNyR70hNi*1me59GO{YRb6x`%}LdC958V zp#U!*AM?!r05M$!bJ`0NdjA0R%>Mu}T@5fkGIHv`c-KZ z8QU1F;;d5`00%wkF~=P#dzxq&1a+zqbF>mUso;IeQ@=`wsN(8>pFPy&hM#|QMOJn0I6$MmSek52U>rUnT1p#h*={kjrK$+2JNYrFVTKjL3| zk-(Ra$X5pJR{GzLe!{=by%XVF1aLrX zpVF>g+NJHgK_f^I{q2pkn#NBE=~12|l>Y$0%|f?P7C*PN$K&m(gV-Z~g*|mGIE?+a zFb}9ApdJ|0fy%6^eSE|HY4CV&PKEcBc}VCaY=KSkjd_=0n%?XFJ?B{ms2Qn@0|%U+ zdUC49p%GP7bP6y9E;>{LAu%>E7!Rd4%50yPHDeh*lnxCD3gRaI?tQ%~UnRK8T9o0qw$?H~;3{9bMD>nHhlJTR2Vpg_rr{4z^bor9!X(zbH zK}Zm+f=B-VSO?3KpX*df!+2ZalzMiiy1u6|n(<%EiXh)CZR&kb>0W|kY>leFT8Exl zy7nD2O06r9qZ@q(N*K{9z)8!DpFn9o&g^p71F#gp0*)9l=m0!YggMKOBf<{!R1LC;IXDV|hS(Nt&HOI|ob7^#7ov>7m+gva|Lt2+s*SE_ip&kSagwL1* z4}4Ry8PQw|%c~W!(quO7NI=nn@~?lTZTN4&k=RDFUjXtiDua*@sr5B$z#0g$(O*)$ zjJ?It{n&1BgQ5IRPvu=lmoqWKFdg{dQc-vE6Du%z098U4u*FERd3*&d-81P^m5>Z7 zx7QtNR9S}v9QFsd)~J%ILd4{$AoL!!dMjqm;yZe91x*UN43@whO-?0dVUGQ%Eg}LA z+$cExN%f`9&OeL!fA#9aNLP&H_v7@W4sp4$j@)&mQCM}{dUnTpV>7#Oc|VZ<0P9td z<%j?&KT2$q0h}nsAZ1A;$l5Yge!le33Xc2FKSBOAJIX@js)BjPp!GB?nPN&P1Co6y zu@S`uWcprX8suq!<>AB-~5`G!lj8M_01=iFo1MkX*398sxyTPk(_n({{ZXLT$1elvQOYY`t>?O z6QS*jZ`sF9-kqyCBUp_;B1XU61&s(cB!~N|7xbx~9n4#ITn?V~4A&O0DTZ5^(b(=g zjw>f+xvW!qAVnmdbzGD0_y0#Dk_t+Xl9rAE8w`*d-3W*XNP~pZjno{Y5k|-84gpE2 z0g7};Nq0y$@83S(-+#OI$L_~t_kEx1I_Esk>vg}1oKFr?MENIe)?yes<4X!!>&Y1o z;u)U=T^(JGr&pTE?YIR$KPEp*dNbs3!R*2I+5hQ{{=xU0_olH$-&~qoqBit|+x!e_ zz`;^#R9B3Ab~Fz8^nq%xk#h8!z2oOUay?Yf%2{C6*#xRa1@Z1$GBNw|ZtDc>B}NY^ zNvjwC0pPIh?Cq9Hvy@-3z_g&0gf+_G74p{g;BxAYG50HN#?YuN&(S#c>N{fN1|NID zJ9Y_QMqhTh6rWEGRpO7+%nlh8{v!X$Rr(i4PF6?pdt!arjIVVeze*3a0_t+B*}*fC z-ua53s?8VHE^|03#=m<-_g?{0#PyWsJ=~;*IUG1OsBsPnWEj>k?rHFjP_wk$&*ElquQ@in>?i;#G$E~OHe%CIme@|w# zzTid&2y<|AU!ecM5-hBqD6RigNWABSF`F+lQ6Xb+R9AkhKrEg% z2d#is{C%46vT49W(@?lapM789P ztQzP}>617#Kqw~w=$9xTg~hdE27y^>EWMc_Y6sXctI+A#7DoE`Dl@d?I}BKdHNuDS z#7SRcqN;chYN$;xf?FRY4__J=D2z>cBqM##CmmbTN<(cPt`hdgqvuyp3wx+$wxH}Y z#VKuC8b3AlUM3Mj-(l#0k>&MD6l_f|$uCt8&O%6LkS?!=+eG8)ok&L#vAkKzEyIBL z=Tnu&)xuwB%muK`E)8k()E2<`fgJ|ARkysTjHcus+=1zDFzRK3E zV+FAeKd`Gcb0LP~MvWgyklojw*ZuE%_vNuzHr?!`pa7Dpa5JXwi{X{&+e>L`Zr( z#Fvh`%U~B&iaEKc_)=-hXYhf~iQx96cK%7AS<3~Tc(3D<&{NUJisHTHnHoVRGxt*u z#ay!n1JjuTRU;Y+g)G!8o|AAd4{1vy!Ow?>;uK_a3%~zlY$|hcfviN$)ESm>dzdFp zXqnp#qH{@;`al9i)ze|{uJD1P&jW=y?vI#*8YzadwAOaAb4N5{U%-s9}W`9 znHJg(Z3Qe?^UaS2=k$l(*5I(lHNmn9NL7D&9$_}dBbs!SJ=|v}sEAySlCs06Gr@y6 zcwpvPKrk%bCHb2g*9>G=yY6TuZLJ9pH#8;1>+A(G80SG;C|3VPoe2`Vhni-&B&f0E zR7vYce0TGHy2~?5JT{Vlu*UeYQkzx#AAE39(7B;i@sqa@oO)Y$W)^31nlHfMU!B|K zb8Z#$`D~s-@|Rgrf3W#GZ{ggFl-`by36VjWp|8Jl#h$&TdXwQI@p{uuO?5n1xg(_UIgP~iB@$_+7wXZfivkEC1fUv?virk`s1s^zcADbWu6msT2`r&plZjN8_M5C5qXhtl zCu)FwSfe%ckm~)QoL0_?V&UgR<)FxfloVKr6h!-sn$$Vbn!8;l>F({lBcc}1|I-82 z-jNou_miJp`!fG~+PI#G+Juh)??0f4HSBG@F2P$j<45=ex+mo6M@Nc0fL#1bE26iK zhS2hs(5?ey@f}aJU(gX*T;~141#`p3n_f~zYmOXkQ%myU#1@V<;DG-6wXq#nEAR*i zi?old4{P3cA5v% ze&t&kkR5B~crVF#*IwSoN_1sqE&Lw#1v$J3TYK z_3Et%p&kdmodb3zy*@|RMT(s2M8|2V&DJUULrT&g zbiuySS9d<;(r&7VXhxeBj;E%>)A|P%#sbxncN{a%c70AGDb^HSsazaocv>M8S}5-r zg9zKLOw!5QI-K*8Wtlw1>?uxd7 zIWEgwz?eA>pVZLil+&I@S;mb{7V0slR!O^}t`av{mmrtat$#-Jj2b}N>3e()H zTp4Sc1H;aJkIq~2z8+S#O0Ko2oL8Nc&>c&g)eCjq*(b6whG(a=eH)6VBMjV)=#Cs= zthCAMpF;F%r^OaUYKi_`LUR#CV9?5DkQ&}W`j}LJr|GoYLG0onu}~qo+z9M{c)sICv9;bGniYQXWqv%cUT;Cg77;_!;2Ka;5vDrqak7~ zuTK(?FjtuiAPxFtgE0B6ZP92&+~FBIz{K+}o;I;H3T#k1xbQq|F-}9zn$@{X>N?1R zs8LtIWciF~_iXsKAd~T%2cO)eho;_Bflm5}9EfuiRzR}#pU%r7?IX!Enr^k6raxQN zv)EXSiS>TKe>3-K0HcPTSwO;>`?A39EqJOkDZH7<@x4AWte!H2 zS!#WvJDz7n##(H-?GqLaASFe;nLKCh-^kx5?Nth4{)FKE8~`K>dy5uHGt7iXYK*UwGB|I=L|$ zV}8yrd0s&%l5t}5b0X|3y znDXu_sn1B3sp(f|M~j>n4fx({)aWgu&Efuw1BnT>uK=Evo5yn^BD|Jrrq;IX2T%TS z(bmjF%!|0rc0F}3cJ1CEo{pJA#;+d{XLQ>TqTxzGsphSX%7RtQ~`s9JJl6Lq%(24(Bpr_Swwy=`&hg-5FH8XD1VK4>sXKXg>@xDS^`?Y#J+ zsEPoD`r48|>p%y-C~CP55JVD*m5dGn@k@SzJwB+G415F@3d>uo|iZ`$o%@T2mQK~=bXByfJj0%CJ!&i^fp5r zmb{r9Nl7Pa;Xej5O7cI*hP&&!bZsWlfcHCIsk+AH6^ZY0513D=0%MhD`%Fs)KCXiQ z*066BF}<$qv4e$n30@gIx$h2CG6u)2$Np3?WB+d8;e^O#DmDMfBFxi;C11Tpv`qyG zt?Hda(BHO@tEkNC5;XDbHuBqhBWQV|RG*zs*mc!Mr)GG`Ld?BWMpIxV)yQ4_r@if< z#>z5^t5jI<4r(=dS51?HdgVFch?vQ@y-N}-HmZPCV-9OCR&0AUS`mLvS#r;ZF^IcT z-Nf8o{RcRZwEHkRe2n>SqGO67n6jcWH$;n!8!GLto;fw~ zvax%>7VuA*ulMNC2!lCHWaIAYi4KkgU|)rk;r!$GjpO8`@Z!@-tI7Nc(X7=o zNZt4+iQ{-5USWN@5JnWH##Tb-;+o*Kz_M+JlXv6Cq|OR>`e6=u1RQr9a@tov{voG8R0g zGA|AdGNal#cq1Bi`Xtbyj{ie|+?x)r=5vWyER2ozd+*y6k_LS88#MBV;Xi;P>_&0ArYpy~r!OpOKtSP6Qc!OO# z5O(HA`sQet{aELp3_o5&DxU(y5i@qp-!3o2Nj;#ZSF>TuV@FW zLVj5GFnc_JRx|l4843rlUmAlv(%{>Vwg@W~`SEZTm5JU~I+Dr$1rdsf<}{Om9LBs% zu-KAElo(?X*-##B^m`QzLuH~oeD^mtUQ{Qz3Ycw`^J0mwq~TlNVrv&|_!LUwTeNJg zf*bg|b}m@4|Eh7|G65fY0c)oCGsyFdrf)ZOhba1GehjRn9?wCFeY5JQYx_&l0ifV&J? zEhzyKjYD`ZmKri7-TZWjr2JR>KTyr@bh`LHV{kL+k0Llliq5*fAxMd+%5qMn%Og(N}8LO|hL5(b?enmmaUD2#!=Nfk7 z#gCEs$v8V2kqQJdAu^M#PEnHqisjYOZJjUqonM+9oz~v2Ot{N+H<&w^?;Z)dRJ^p< zJ$xTY-2Zy6t+-bfl4G87AhtScvPBR_f#s3?4Yi+Njblt&PsZ$0rwQI|vS1S%48X-A zH1Gzs)nKOPx}^l4^IDu0@Va;8HPA7(QrFLK8#_8LCu5Vbr4-$=(Jsy&sJ@*`Mog%#qEkN)6eVf(F(^xq)Bc*!BAPCo<3alrajKxq5pj-h ziV$8!EciG2gGBWwiF35|ktXPz4SRL;Lqj-r#$u+J zyb%$*%+08Ot4E@1(%_udyo%6o)ty0WerfKRVA>Taper>A6xGd6V{#~o8K~MhjdM3# z2eR&sS5u14G#aP-fNhL2zbPBHO66+JcIk&RhMXT6i{OR6`3^Eg`aai2*J`}jODrXIVUYQ>oY;=R8@M3-(Ea(># zw223YL5^Ro9dT+(+Klzi4b}U=y|)Qcu{1z{am9|c^qW%1s_AjLbF9MlnVbza?tifu zDG>OmE_k!rIn;of9m<$&DF3mRSL|}Aq*jmQA>LQgjBSB`TaI}l8`&I3wWIBCc`3qr z%GNTJ<2>U@bn_>0_~lfE_p^QpJi><1L75jA)O z&p*meA8}Nr6F4*OSGQ{iNpB(U-3<*pKEOO$&WqSK&l!_$^k)PwRc{P^YPs22s74^W zy@12>n{jInllNRK!0zlZt+a(gyosfX-rwoC8a9?|osBMoA1?B!GB*E&fi8CR79%}# zHQKP6TEcy*g@{i#sIkf58J4|>vL}|f$|~cZ52O5vO^|-4@+nb^x$(ApKT4-?z#%~m zeIDOOSRkq+Xz&{tV4f8#5y#`Q@&$JKntJGJ{vb)?dsF1nIO|3nC^ShAg`2pf(b8Le zJ(b+qf|qW_+^DQ_a(dgi*YEIb#T^S7mlOPaej>W}?Du-&bYSqZa_dplA%kj*?bhWL z<1NYQUQ(w{oik1bcF;Nxb@;%n7W-s5JZUS_^I@PMmnlPt$TPd0&~N;+Z_7G)I_9KE zi2-(-0lGXgr4kq=1eH?ObeH*g_}@&W02tjhZrgWuJ-8J)@bz0a^YL+hJ9yhH>o-3+ zh6%|SC}DxHy+Tq?=wQ?GSVw*jy_-eW4%rh|+0GP!{{Virzb3k^5jyyAhufVp(LNd4 z11JdtV*E&)tao%CdOE_}_k^^*whOmyq_+ciRrYh9)0spx-;Kr|MXpNg#n7TOi;9q}m#7lO z)t8e<-mc`(KBakTPq-rIX&jpQgJV?o8ND~|n>Fgz&=T6y>11YZTxYlg%Vo(IlH7@yA`-|d;zlhd&njH5k>WRcq_$njHM^|0^0(1Ng#ds?J3t?IpS-1L3Chu!?r=f$HFV# z*;zLvjr{zUs*_JM3E630=O+vfQMF$$yXA-Gda}*uE@;GtP9*3w)Fmn;{XBS^?6vFI zdkFqbL{<6$(`Xp;s5#dCp;D4r29>G@0`@VQ`JsUtU-| z%QEFae?D|mn{x?g4z$lF5o^^eD*M(R7{L3_(IzELwU?BSQjVmY;;xRv?nNPp5RKTM zc#RbKBem61F(MYx=oxJ7ZhrYyc2E7w=1 zU3J5x$#$g305h6b z3oPBIOLt>-_3UTZL;o@w==Dc3MHo~Ufy()Ld7d}I1U5gW{!8>$F7|q~R%txLie!4S zuR>qqQy;-}irx7t>8Obq$+OD7HHGJChRAMmC(fps=?|_be22| zh$nh5sgmP8G0?v+_aM5vcK;6$OyIYPB-x&}w2n5GOL5S*=2h4G$Hn5W_w7)66}8ln zIZ+=^ESBIxllT6~&?CBG6L2w&7TEXnd{4hTAbrpHDNp3e0jvkD&x;o(yR(HhK>^P= zF*+jLHRcXyeVoeDhMaoU&A^oFBiJ7eiKApUUkLBMxM# zkG*4C%+#ysW67h-#H%CFx=RoQk+HUJI*;Sm{cJ5Sk66p zyevUr)CUXpNC?H6-ImA}Ra7*w#|c?u5vB8;6NepII?JWTVuu2ykB^X2>ee8Pyv`h5 zxEtd@z)0FnxS`34Hi#?Rsj6no@JQsPNm=Ookg3(cF1xePI5U(`g-7H53oP&Ig*gKcL-Ka z>OWy95{!2Itm!5qpOO1nWl5%pWMb^Gj`wsb@n!r z4?6&0mK(gajJ1nY&mEg%Bc3Szu%*ueS`R;4Ee8+b7_c87Jp?-7^4c_8gg?%7KhhyM z->MdC!`gh8A(@RqKQN4v+W&x+U=TE*J}Iuf^_m3GG}A7CbGBN_HppsD#5aY!qrkGC z94GUv>+x&V|3aZZ02)4h%5g%; z-h^;XsPkL>Lb3%dpvkD*o2Dm7Z*I@PzaNPD>$%SxS-+~izesO3DSXJ2)`Wa)9eNnp zDSbk>)izaada09Mlpbdcrm&PD(VQiF>Dg-<72_&tE|*|SZXIg2!Mrs&N!3fEPy&(m zDmI({>ekGU$z^K|rtq=Ok=;?bWgjngr<)^+$iAL-O4#VRpP1^rnc6m*PB87R?G80? z15hWFb<;o>RPaL)Bq`?g1hap&kIP|Qv;U-at|s9Pi@O!dk1!lJ2Jjz9Ay35Gcu9hL z2shq&QM%4EJh1yjn4^!ONm90oDd3~Kx3~m>kZLilt7zzvwz0CFA}M)MT6YrTD2F0K zXbT1I<7*b0-%|zLI}{gFKUZA%IASW&yrBAc4+U_QIxj%;%>Sm)_OlCy0Uiw&idotw z0(p*8K9aoP`s>CZA2g zak$R|2mi7>W!+z-)~oRz8R{DFLK5K?e)jl11ZCvz2T>0`v%(9Zpo;CaE8XV>rF_Z% zrX{kkC7nI_E|csJ!))!)1yku*j=&?%cb6?Y^Nj$J*FX_UGe%wnc}K&mH$?&UE;>Z) zr6nHJRai;Fgbw}bufa)HZ+YojmYIt}Ub#w!2|s+(2wdt;rOefo6;fN<_|a98tps&5 zzSKR~4LU9!0Jl;j$j$KR9vb-@UVUjATp$7r`9K_Px*htEIF{N$=L}7vRLHR^v%6FM z1-F{sN7re=&!i3NlGtSQf6)P3qXjuGOKsnL8_NaQl&yL3KKm8jYj>GMaJv}sI)ihS z27mbp1_=8?U!=rNcnYwRgi*h{Sw&|)JOC(4``zcq^A*enfA6Q$VZ8t)Z;ZThQbjzG zd{XpE@vqba(`-(DVXy^`OuS4(6Y7|kb;6&yQkWdqsIK)?onu<8v2wWxL5qgsuhuB* z%xdwxjQ3t==SRA7EY@fU(x=OrapjUb&{r9h zsZy!b6OI4Njh$J=F^eZ=tI8X$EC19_&YL6^rdQ117yi9UorSl!_OguQ7of!c(QO5E z{jA`_SELV^vGH}b#1kU(6B+t9*)}$eXXURJ$}*ZnlCp*DDn(>;_yGeRBB0z^7Jxr;e+NujZs_iz^Pc9zfjEV4>vQVbTSn5%h4jzhYx$ZW5;3T(1tc_p^gH+NEyql-Vuw< zOgLZE3itn2BCOL>EoUSeoGcwC*@jRfuc}rW%(&v*VxG?;tjY&h4#3sJjqVGmj*om| z9${AT`)dCl4PN`o#qvKF>HlMVw^uAn+YnDy6)4ZRnHIobrZt#x%w)e6ZGurtM{ZjB z?4X)xe!&ftm#KDrDydcm0o?%sKs!q1!7l-F0%h8wQYvOw9kI&+--3JOf{ysE{+p>xa zJAYN#e|V?W@@dX*^4Sqp*E^(p((udP=ZTaP{1uBhmX6%Z>E=uo#fy&dUp!yc1`PcO z1&SLwC5CXNR?_V+Oofy7iI`Jrqeb`kngkd=$cHP`h(z@8_bJR{m_e*KTYH*eoLn+G zhF@bOEr+&=^5Irwxe>{0_jCEmOI6}7a#-afO!}_ey!hWch{AWKZbZE6qb|_M5MOxP zo9`W{vc6MWU(Bs?`{ORN{BZ%m@-+rqu=p5c2zjhxMBNwWEqeRE{>!B?2biz9E}Y(Lf|b8CtLnyV zxq@y`UL-sQOCu9Oa`Lc*`a_vnfRwFy+(XM{VTg78cW#f*PrH}nCfe24SCR*4s7YD8Q4g0vM*AfIc>edXMSY;y9Vv`HpUoEbA z-obK?A315iZnD%J$IIN1Wcmu2vF8!QY|D18Kq;BV2i%)m0B(z&xurF=lE_k5oU4WYI1&p|KaM)hq{3W3~iI9)nYAA;X$ z!P%Cx*15r0(DSwaxmhT#0kB$%K1yQBE+PJnRil3H>JTX4{+y^e^J z(KcGVY>#RLkYGW$yMiM|ESa}w@PQ>BNjNTR$bK$6x%0Rd*lL-3m+PjDy-FPaw`KJ; zDSqv@AJ8oHaP;GEi&-WK2$T(yG48w15U#X8>NST5RfL;pgn8gmFj_+Bw<`aL?tD%SJUB@d3JPxS=rmR1IE5vIWl_P-WjkR3x-pLi^>XhH;yV7W z=;?8McF`8D!$t5WJ@^(qWq+EuYx2J3Hhf|acF4gJThV#)3%DxVP|MrG_nBu3>HG45 z<^_e3jdIH6*^2aP{u4NMQ?p|>UZwwZH4-Jkmd83}5x42kBtxr*|&>P)sA#QQibRn|A)nw-6g>@Sphj_4r-|<;-_@N`E7Q|tBiiO!1 zj8bOxKUSMV`4+Lrw%qN+4*b|mp72b{9H;Fj^u^0@bO0p8lS^&JXM6aMYN{SGP+l;3 zPPFsrnAq4ck@JX-(4x8S;7Y9cx_6x{*IFt}WRL=7%MFh%YuQOVl#Wj+zvGW5)Vv|i zK*2^b5}A6T>vXTA?8#MSpTY(C3$#@deNBHAGIzt+eyfq-lvPKCsm9^@w=(x{H9WS@ zp@Z=WSqs+iiWe{7>y!{ObFS30Vdz>ovSIGYd#j~JhHuI^cMj$uiw}(qSc?x+8;_bU zx{}dQ%&j8z4!H^qef(=EHg`+y{prbRGH`5=#2hwkEccIhn;jHOOedqvCp97cIeS~K z@-04D0q9zZlN+K=aD$U@rGMhqwpw67nwu!k97QA1_x{I$Cb>qb<4~Q(FSo~T4l}|7 zHpa1ZE&Kj@^AzV@qJkq+j-vo>x{;PWgJ|c_g41_^FyeC>#&k0`a*sBfN>h#?v7OGT z(gUsDg;3%|%2^16SLBs5go26J{J7s;eYzyo;3M&$uDL~@BL$FJqO4`#DwNs(tH{Ba1g{Pb4Yj8D}aRa?PL1d1>U`F!K%GIZFKci6Mi3 zyrdhuqP7|nezW}>Q1!H{;GW7IG2q6}M z_&)%y>E52B|I?Qvh>d3Fy9&RXv5V`_Ra>BfV}TSHW&3DD(=cRQ({#t}nJ<7T$g%cW zODUy0__JR^!s}!ClMnpGOv;X*nq%~ZTZ}M^5bnBtQ{{Z0D8t7gxAnXE6@9B;ra(4` z>^mchc{Y~Jly{3NrrJ&HI4xR-zd-1LRy_5VX_eJBx2R%8D zQdf0U^73W&a@_=rfiYO_$uYAPzVr3dPx}M#>vyEq#NRM{^ZFAxgVuti;RD|I9t2_7UCI> z?JzOtcHLr#>#QCx5|+BwHMZQ${B}UAS1a(aCG{9vpDJV%`?4Xh{bxU;<=xSamaKKa zuPPxBd+o*f%|u^o?REYwrsV+3J6qc&4%bO=UFCF+>QfOnvArA@&4oavxQ>fI7-}e2 zl;3YMIpm5U_LF5kf}hEa>S59r@E9W5Ii7wGVWf$uCM>L!cHTroohq2`GNrUmBXb#LIC>aRqWrXLeO1BK|00;ER| zP0B%ZMz`qZx7D4E;tZYzsVbiF3^#&TlY_*D`fQvJFc!HjuBBFuesnYM>mGm^$J3vI zjux*Y4VHr`-eKD{B=V`2zi>Y2EKs~V(lY5TvDL6lEFt$izbs;;Al0tbT*7)qvM*A=x1F4DTKF0pG7vYNCZ}rnJ zMUZf1L)CljM*RBkwTlz1{uG|tvEBlkw#4LWa{t2=a}{+Zt2Qn5fuc0GfA1YtJ_v^v0h}fL3F%%Yt)-zJzwScV%B@`2Hs9#7 zbH6Fmc~wlba6=Mn*tAAW>*jvP_aFxN!hE8LmkN|3o0!#QT|BXj{(4({1eAxJ5HOqj z2h={)#_V99;F1)Rgl|H$ZsBqt{c`-Ea!!M+Z_}OuMe#0GatqT>(^AaYDAK=n))Y$~ zhD5T?97zOipO<{QEB_|ByegdCc|ea<3XVS@-Q5=no@LH44HyT|3;iyq{aM4*F56Ii z&n%7<8fMRKxpx#kK~$rRL_;6`oEPfRsEfX1AYEhMxc%}V&QT$}UcOHvvPFjtkb4t# zE~|q}CsH?=lTvfUwq8`3wV1!MVTGd>uR|0<(*nXjESffyP$`rEC}7UKLlwW(Y`1m} zwD^@+g&ZRlgptW7_q(mKqJPD@l#4<~92K-_8THoB`yGgVmi=$_`@|?5z1>xlnLQ10 zIJk@w8|Q^aXHuX!1Ie%!3kvbt)VK@@3%W2R&ILVwHx{dqvr*zgZ2EDt3j=;foNc1~ zMzuhNI){O!1~({IpXF2KK)(qE%llKEmNn`Nsoy@Ec|k;umhd`eM-hh;8#!kKqVS~g zywu1R8NX9qf&dH6ybyNQA9TX zcEM|nJRqR=c`oTo^AO)dM+VSwYhu(*{c4J2#%s%w)tGmda;h=lEVl z2T}vW5H&pT@s1*n^1+l9egLLfPFiYV$-_a4K6lpEpYM0F0yE6Du*VpuE zA5{243XjPrkGRTHTR+n(=QB@E&}KG`h){(}NfDrU0_;DPVu_VKyhVNX^~Gvg>>5&8 z<0o9{^eeG5(~?qYE@5^9cfD29eS&+ar()zf~d%>xqC-*_X#mTfou`hTq6RRMCmu)m zR3J0xa8a8J8%a(e`udY?D!v!7${HndtVRii4?i+CeGx7jUcUL_BczslUrp#%L$58p5ghokqcs zl^*~*loWpFU%t4X;-v*wUVpWeSmoU(QA^OW<~}`z=Gir>{@f>S(%B8P1uR)#PVWz; zCw~~WuA)t5rb~#Q(N6W?U)P4Hywg63PgIq1eKaC2@f#$T0Tt_g#TpB%F?XTGt0}Im zYm1ooQtje6yD=C@7k~E@H-rs22kJgx6lC!lhU72ZHc6<+&l-g2op93;e*;wio+aY& z%Q5xrw{i2%D!5XqR99OHVLj0jvT@ZiyMvS@ifkx+n__}8IqDy1ZpKrl*)U$iv9k7~ z{<>^TdF*rLcjg9vN@h=d`T$14>a#3c>$EF@4-J(J9R601o3l;+14w70KG5cp#6Bha zVXaW+Q2@~{Ev_4+exHChH2WJ8++Bn**IOs|AIFV}Nba!a@ASRM=~xP%D~f^TrL5hr z0NuPUN>4-gV1&k8l>h_oUurg%;}lt;w*{l_dJ5t9PPnr~o0-yQ zxu14Iwa}K$e>c>YGNDDM8pfm-1x+vwdY))6cP+y7@bOF#^(f(ysjENg5z{sYJuMj* zVJfPZTvNcO6mW32LqU`Pb(PI3S@47s5^~R@MAcD>^m-Xh0QXS6@pxfHB~C+Lzq4Gt zJtK*A+l-A z#B8@m52|MPHdoqiI7U@OD_q=%L$+i6sM<9^<{(zoxB;HR~C zAx*amRDiFRkPa~ zW$B4aK!)eUT>0v<$7nO&JzVqTB%mI1OVcSrC73FGNKO36mAE8et@*C=>OhKXE@(IB z_m{jUz8b!a34_Ibv~Cho)%K1k)NyX?hb4;>NQ{t7}iTMJb8JUFy~OKd+V#E!{+(Fx(ZcFb0!U3>bFT#&sU z9IHnKF4Nph*}ISmJ`=S_|JR$dFH;BoUvuw|l5eZn%aWZK#&XViDG%e~CGnkL(yoV| z7WTL^N$v`2xM-5flRh<>Dr$#-pG@NgC(Fpwk1j@eo?~Y;Fvl*kOaE4xY5` zc`nii&4=*=Xap>J9%$jk!2oMq%(KPTXx5m!Qm-+T`D$pM-x$JEm7*PgeNtqUzUr83@y#sAL_<6(fcTYe5K! z#6s*!iY*yJJdxb}1akIx*|D;l6|JgM)d?U+O#Q<&vw&WIF6DQgwQQ?=QI zVY$Iv8boZA7sg6R4r~-c)>KSA{aeTrZbWZu6a5G4)%=vt?(kS&%<`{Rhc#(mRx^ls z4OLM~7&+Y61274LCZ~W*BNEb(#9OFFf5)8-5g6qMz;WiSp_+>;_-9Rh z$dIJqtr!Q7a1l8B&Y)!o5vIj6_yL^CIu&`J6W}e|F#Ea4geslyv zRI0}6vu+r$@Wq$TlrsSc^s%PNr=;~nst$9&oy&^_US%KkSvV^fu*Q8zw#}Uwpj3;; zS@4@mwM}rVY`B3JNN~rs%WB~w&qAXCPn&PnDxbzHpsbp^3or7N$ymTP$6Ctxqxh0D zy7S8P$^ic1u^F7(zYg3Ue+Ve6e^;q_)gCeBi5T%lq zpI?x8Odfl0q=?$tuyf6bF<5{e8K=H!OOxD}#8+E%=8aL|5 zKdMIVp$1@+ex3C5@{5p1#y8Vg%j?*n2DxbV+Ib%aptUyI+g{d!FXA8Upy92CvDasp z1M?A&RTm$7s**vhzGd7g&6M%WmFnw9YX5&VIivJ{VKfZvDPtWJS0r?Fn=e-h32*_* zG?2+i4gPS?nQn;_DZYCmT!qQvbs0GzjpE<76$gEYHzA0q%@Px{E$&f9P1Yi5L zD?UuMJj80<3Wlexe0oj!AD|I$hF*V?ar8N|DE>85;!el2_?1uFQT4CZZuGGZXdj8> zT929HAd=5$v7Hwu_^c<&P>)=6`U`yEtR#p*&D%9;-tpPE65FA&G3ebvpW^eLKU@xa ze|76qP2zr+3dsa@q)hoZ?lC{9g}S{0;6j{|vx;i7&Az!ezvepEmZEOfE`gMRyI()@Va zsqVkjYV>cb?Ae-KCUSWi-E={Zhc8TR--jR>KY4^zD}J>?-xU89-v^R5Wsxt-EuD^> z&~#rtmCthNc>^7d^pF%;j#maT?Wy-#2qj^&qj1d%FQWI1AH&M(zW(CN4XPf zOqXB4?~ATrzuJF@EYke)8VIu9*3PY#{C=$NgHa6gFEM~6fw&7J-YP(Rs%7g?N5)gX zzIuQUWNeMC6P+5qeG8*arB%o$@2<q+8EsoA&MbZ@Tm6!vBwJ{qg zHwz~N)gOpuaz`CxJNwI z87)1R&^sD<2urw*yZW#UHL3iMf=1dcF>AIEN|XltEt(UQOSsh{c9;*>AHfE3Egme% z#d9<2XB$tIsLAE5KQPQGl}l)y&E2m8^OEYLT@9g^Ct|OZK_))~*o=R|Py-InKEvNS zLPaJ4%nlSAS~sIog~BbEHJA_MC-mdr4izGfukEevQ158iadw8qNT9v_gqmEscRRPi z^u()KF?3BVRFk?`nOu+^;AVU+Jjno$Tn((Z7P_Z}i?9lzL%PF!(3j}P8iX!HoGHqn zVoMw-QwinLf;)8EQ>&j$VBBX%m5`KbJPs{~PimVa2Kb8x_KjyDK5izHj^c?XL0K|a zvx!WhfLy|J2pRW|7RC?iITsppB;Uuw^xHx007@#`JPaUDUKJ5>0Q9ju1)}lYyQuf+ zC63{i$|24G1pS*bfBi8Z^!!LAz)mD7V9jG~=(c(5+JKLle_5sg6r@Fnh5o{^;6?s#&%v9i+jTO>l_a0# zgY%M*{{ZUI`|rMk?Z71b_|J6m>Y!s0&Qy(BN_x-A1P;0ovAM@r@X4CTjQpV{1Rg>d z@u!#!G~lC=TuEN8Mv-&Ur*&n{xgII}yU8n7qdLS&vh%hdJnOFcT%3qyp!q`V=euu9 zuSkQGS#HK_jaesh3F(CIPp2#SSv_*Si)OhF|K6V(37;n2ZP!#OSZ|?X3pglX^>M}2 z2vFZMDwDn=Xx#rJ>CD5S{=fJCvai`1Wf;5c$yh?jNcMdflI&yOlHDN6G{(Nh7}<%E zJru%_eMqup7qXMA@89e5{r%^^xm?%GIp=xKeLrr>GBTkQke!)sB@wdQjk%4ME36Uw zdid&CIJ+RDxn_Ln=`gmNA!9Aq^N)k3U8Q?*`%Gq&@?;l}u>jkO`CnU%Lv8r2$U8F4 zQ+auo4?5_N%5~N}SB?w6l8OyCFkhwn*=1BO)?JDX%fcHsJ~tExI%f8G_#DSJxCSRL zIueXR9qJbrJ1;!$biZ6xOImtslXksj4++xXZDBA=S2~=k(lXWqv*&Yz?x?1L+O+jr zPeA+6VuPfvII#HLx&}fO48f$1DNF7}XYLu6=4DAPl?C0$abAWF&Mo^Ig9W}KL6T51 z@ZYY#4%_AdBdd+Mg~*aZ-@K}*LaP1ZBRO>+D1wcN%-9MqiBy9mymgo@jZvFdI*#5GnEGeH@?1|H~Pa?Xx^~~f#S1dj{ zH9-m7>+@CGk+(X6jYY4T~7D&7?_(1C3_GUyCzAmq! z+n%dNCA(afo9gxM^%Ph^|d^RpfNHXhi#?O>D_zam%t?2J3k z`Y1sFQ^~O2?uES1QP0OJj7Umm2&pM04tW_>O5;=#>U0@}ENDFiQiC2?KlBACI%LkLY~)^lLU7YQP+uNC=TOR@&>!bYkv<6bubS2jFFvGTo94NYr3 zGQuIAEXkU&f!7Yt{S9@uMg%B>#H2@K^N5Zr)|A{CG7Hbj>>W#l77_8s_&t$-q-j2( z0EQlORD}J`3n9w)0D@AXzQSbW#OP>(&RrZx?2{i7=KQVDw}fIsQ$uzp;-eEeMT-3l zI?sp;L?{#rLD`%36(Qet`sFJWaZOwV|;!pg=AXrlq6}BSzjPUQ^r^~66N2$!u&#PB`_m0H#TpuT=pVWXaRYE-WV(5 z7ptXuHcGOkg`mt6DGuo9$5Rl9SPBW{3O~d(kb9u7-Soa%a9)_Cy?pndDQd)6w?3#C z-VEK+(O_(lMU%!gZ`yBS+@J}}wuZH0p8;Bb_zMcgV2)dIvi}`q6IhM=S4$j=_Pibyb zU+>PpEHR_wi61@HvEqJ1wlj)~$l(g#&(YoXeyMUBM6~F}Ur}B;a>3q~El6UyG{#&1HP`6^7***`$O1004%w{Zg>elEoAZp3{ zyr>O>W!rO-EWT8U7lyijy|gXZ&Hb=INq$l}{amsSvAI_Lp-%>tkbYguET}@hM`jt? z17}{iuAh9Z{^B{P97v;ZBO4A6ZT#N&=ZcKxd6&*Y4_6(X->47TQm)${HRwTl}Ur!UcO8 z6ap}(LZBzp`j5x!J?yXT5Z}-aL&LwIc4?A>)~0}uT4B9Gx=#-d{LuR(fC7^ifTe&b??Y@j_~MSk0N%( z1r&wkhLi~GdVwnOb@a$7Y=zd;^N}=$ecp@D99BaqZNP4mHl9=h4t_rnr*SHEK03iir(p9!`3 zzoPfW!e=j$Ui@l2{TpV<93Q^2ci|4!>0FER?M z`h`}>_Mg{2v%30RTDigFqVnYw+$nR+GWn#ZVT`gg^zM?Q4CYn@6qhy>jX8{h9(%Hlf8 z?jKQEMo_|s=zjbMR3jdQ<>vFsbn_KDZktvvqm7!E@|&ot$ry@KG(@KG^hP9@T)lwY zAQ8VFKqko@C^K0?K^qf`WG;-+fW49M6r;?7LJ43*7BtqB`wRdxc+ZKR6>x2E0fmNx z)p9bv0|E>>=HAUUp2P{0D%X4UC@m#}e+dl}At*wr`dwbpnp@7|j7H6Gb%UMKwf3td ze;c|-nIhIRIxk~ou$ZP)aJYF+5EzxwJytp?QpvD5wWu`k*`-%9h6dC zV+vnj#s!x~lycu0yz^pMN*&Oja7bpfP8!l*luIP!xuzMosQ0eXrS%$G1IE9Re$iS` zUFK$|pdn$NBLzHgw3N&x5}GzLnZ}hLAxxs(MKtkxdhsBOoS_#*q8t4Kf`XZ>&&U8U z7c6`-jY_6nQwIPbEccAjY6Km?954zsy0cunr)XTbjYuzhlQG& zxD`88ZVuim&DbRIBv~8()Q2+8-&A z=;u}8mDk@dA{aMeednR11`+#V)2TG16Pp0YZuQJ z8TmIwqOAwkuxl!c0i?eE{`Ftl+<2b~%iCVZF|KRV5pI;o@G0GZ4)1)4WBS;AlJ?0-s>!gG9v>$OyUGeR zK4utQr~T+*QSNT!=+fJzCmWCMo!p{ioM)>J{=sr|DaLQ7A``6AE&5Y<=R&zRy=bO7 z(AxIOQ@-QYwSY75g`EHz*IS9z8*+S#C$ED@PTp+?WB)4O>{4SdE@JEc|3$PaaSvxu z!8^jG+BXt0@HoTv?)BNl>E2l^uU;;4y!xYRLda4>9G>uGGI@HxvvT!e_PTD}roHi& zabt>8N8}2LO@Xt|RlCNd6UuKMYeoql{gZU@k(DDjbTkh{|FZwOTa9hI6t=PKRaPAP z79SA%_3n!+*%=`_s5yMJ_v7W!r9i(ctlGD-dH&Kc>Oa8zCS9J9kz?O_Z8hgjMEm6Kwf5p$ zwI?gytW*`UUlx5RZQb#oFjsWCCi>YmVm3Xk$V@~zqy-NM4^csj zaP_h8`A|2Lbgb178wd(Y%vdf0N^qA>(VkCEL(sYn$;pI`k{;sfTh`vpryG%^X^0?- z!TJCxu(qpz;WSF-FKzg@hX=qS4I2O<9pq7o5EfaCod7149MY2IO919v>W&jUB~INu zEij8i=W0X4z*4bNdzBd_S6SyPZQhmvSQn$@=F@OUz&n{=&yk|85#E$6+HJ4S@CYG4 z;05U;yz%9KE)^6e)hmm2U#J^`}cvaV?N-B#=rVIqJ znz6p{eafj1me5JBz0Kd_kK>oUsHyFW&-Qs{OmUeirA}Pxs2(YIfIzn^F_t~tZ-w0b}QE+_@Mz=Z7(OX)JW_p|4ug@G6fRe8eNO_u=|9SJk z&-_XC#}IpZSMIq>cbhdj&!3Mw{t7{WeLt?L{<*`;c~SKYt_P@t_TWmNx{-zlIr4q{ zD!1=lOAJ#KW%>#N8|}5z`iPU4FzGwWjxAz-NT74ShrwgweXhU zxxW&b-fh_Iihf{8Zf}3gR+=`Q#qY(N;Te74WHsZu%OGFHoB;4?(1$U>&EHuD`je7y z^mSI4%(Gdw8B-`T2$PyfO>ONOCoT>zv_`e(MqcboL$JPF%%cRt;Ip1R0ep3dzxCN3zNqzEKR- ziS=2`to)ww?Y9!d9$9>HZciWL16r{jh`uni+7%q!}4Yt2)^J#;8 zFTnQia_pvu!j83qZuUhL;Yon)PFqw6_xYUI$M1MPvvkHbNsq)I!MWl`bLL@b6Gzvm zu2jWF+~3u)?LLo@;>c~lC;dyOr^lF!XzpewJ<$m~{zMu2K1T47&umNxs^B8TqvgHK zKcwHhA`JnG}4H&Oz(f9%pvHj_UeQk3zJHks~m>3Vt?e~ zISD0#>Kx^G3T1i7__z?-_E0-1YZ3*aRr>*{uLtIjHsL)GfJ`@6<4LF#Ne>{rp`n3n zd5hxy(SA!XcnRw_iMn||J!mOcdr0C>QUHPuCtO^*{-KyGQOK^I6sf(7pg{`{f!`+v z5rY8jv|QsM;P$QCf^Gmof%EsHxHFE7x0DM78G@K)yd+3NoDjuGvDfcP^!h1*G+Z*lfRXhikMvSBpmj-88N-P_70%D2K5-c!#RRn+R& z52{2fkna8+q6}b*TV-h{QrvQY)pA}|L z%-#0BNYfu8uFuer^OgvO`3J8RX>;O2aw2!BX%8-^RV` zY`F>qdR=Us>0ddeAKdjz3T@!8Zm1A)qm}0rL4TB8bg<9&)m-K78dvRu!^eb3(jk?nQ9ojo!U-?okDbV1yibGmH zf7~oj;L0xTbIiPwAK)CrEI4vsy5q(WjTY39CIK?@6decA-ji@x7Lurl$4<(!l%KsTDU zdgPwk2&V|DHm?U=F%r|5N?z0VCm7bs3JgGT(|p^_{*?1ytbsbAe$X(|R3U5pqU)O^ z`0Yw-Rf>G(aKi2E>@v*=?RBS0psr5}O)TTFKSVHgmmVl79K;-`^q0+td`2A9wfYWI zxjVU4m-j7BWY{{wl8C9tas#9Z%m#lJK?|lZiLQ>1qftwek3dbx+0tu`piGC2dhZTJ zzXD<|GxqK9Bj1~)V9%-5e(;S5TOGddi2=Xk0WzqF2^x`U zbHoaR`n(d}t^Q?V?2J$KwQ-(Nz^{kx)G(A$&xGXQ5L{YxRnC*Vh>U(5g;S}vX0_gb zZqq1@W8Bhvm`ybfKB3r3`vv@Fhf<1NO)B!>R0(<<^j+)G(YeFdy-Q1bUL zro(O|?V)3x%R~%!Z?d1B8mC;9gmemepiYEyVCMC6gL|3cj2qSq>xJ)5r>%Szr_qky z&_;h1aLLt6-OC6wiH_CLX1H*~btMN@h|q{wunByBHZ|7I6p%8)uQK`WWQEMUYz9`5 z?a$`uD_B93xAu`CSb?(kq4Ny(ZR4@6&oqej5lz=_ypKChKM~=V_}jBy=NNOB{@~t^ zysd%re}tmI_qM{DSOy)|nvY17r#xi<>(Ef7Wb>thv$0O7J{#Uqm_ zHI zBlA9J=#bixt&?g^B?FMmMAT9*OX)=VQc|ki^MwyQ`|JG5^OdX80=t9lJyn5&_ogq{ zRw6T8=5)HCT|4{&(P+SVrH9%0W|%^+lqR>a@kfn=lCn&lvO$B;s6+uvG3kCqKF?Q+ zY|&}|*3yQ5(nQ@NCmxZt1NZNUuC07yP0BimG31WBJtIK8T`NJevfZn~juWu5hf#F& zbGlcKuqC|bk}A$G5)ZimC&}o<9XfB{4;V+7={BpAYcs4c^~oeW#Ls)S_{**IJ$D~Chk*QY#fXk)S%0I7q!A^Fa^>p>{Pk83e{(tJ zr6f*}ltK|OOz5b5zq06y+~VV$tAsZ)y^LN#?g) zHlCY%X3SdaQ25agvn&f?6wiGz6MZmY?*|UrCGE-aXeE*P*?1@$-a(6>a#}D+;4qP} zT2@rF$xuiG$8t0sl!g=@!x@;OTfR@tcB5HE{1ZjkzeEr)h8b zZ2ZZ$hzGK#O`?0jJ@*$U=Dspx;9Ge(k=jqPE5G<%-~%in7s5az^^faq&UJaPXk**T z`-MKAEa+Q!ALMZpN^~9i_7&fLn4Mj%({qkHb{-D@ZfJ~&x(Dz9^S0V+E6e0wN+w7@ zfuQ1|mloh!MRld1;ogd?;wQhOiooB?*rA7tjj4u%aKsH1gg}K{m*oLVCyKF}hv93k z+Q^~~Xwsc50nOn6M<`>7OsHgM$yZmZcEMk76o$x1gL{Ph70Q@<1GGZd1yR@W_cI{b z{ApNGTkyu=uduy=K%cnYC2XqXTk-S1XRADky<7H!|MWYCv-vOH`L=RYfFR>Vy{u2l z47BnSvYW-VEPM8Q-63fJ;@ZZ1Zc&3gh(17^V%?bXVmsL?tz~HO3kVbq#?i%8`u#nW1N2yt2(O_|7Z#(#x+q8~k71{lx>`^?|#7u)v z-Hpd3St%Km#pKbw$kjS#HhyexE;C`1KO8-^HS{2bvqg!-VwXjuz#~=5F-HBqA`73h zl#?7ky35D(q4N+IQJ|0Ni|i^+6!g`6Ts}XgZ1>+`Ns+2^C2I(|iSD?X)QbxCM~+j5 z%Dv>3+JO1{J(BxovUuY3^pmCP!s>8dY_-Y$wX%$GEturCrS_JC_2|!k)&WbeasH9r zzs}e#Yb;a4CD``W`w15EisT1hfADod)ji-RopYO+``dK*_h( zxU7Kr8{2jPECdNn3iQ*cdj z9)fb!cNi1X+JQq2Uty9kqW$uGQaJH19tb58U2!6ABsB6x?YDqYAQ*qNmVyRLLdd=td-Kz4G=Enq9>AR3xCEB7- z3b)SH1~?w(-5KgoNdz=*bIFr4L*@xmFbr6j(z~{0;KJR;$BLf{hO<*Xo@R;nIVV58 z|1_y%skAQIYe|8-&HKp~f@L=6227*m4W8gLiBEB{T8@5n5Q3BnzHqZz$=oRQ4Jz+q zJaCsU!U>n;WaL^Z-6>m4L-ecZ-~TQgY+7V>5}umxCPNxdc-6co3t ztfUR8!H6>Bdf`2Z!L+K-OJSys#}~i1<7u{dc^qDqoa-+Oo5R&dGoT~udeTFjZw3c- z6;2_`NyK+H2{-2S5;J?Xc^Fp_G$HrrK2*lLE>qz|;v7?*8{jdOQlS{`u0r(gP)M15 zyzm)kk;2(pT(g+-t&vRhPVjW|$YgX{*HT!Mj~E?vy1IG&py%GsLq2)vba;;faf)H{$In%v|NAo_d)F<}^0JhkOqu4lGw!=p$Jl zOE8Piuk@-0j1LY+#@EbxmkXnRmI!W1uFiYx}K%%H2T*E{MHI-;j>044?a~@ zjt*VME_qqOq zL9a)q!`;HuU{F5G8TeCB1gh{GVt*Neciw_srdWv*aK0^sEn@=T=R%2_YsM!{+^fXW z#)!$<#?clSN|M5?F}ps?)4>R`hwyeI8DrBg>5hh)tw=JaM!q$!=)2{0=h1stATbA2a5t+(GWSlfZfFTwM3ok3VLslcgw!)}GE)21E$)keG zj1f^xu0$+6Hz)vO{A(Nn%z^Y{9=7FzY86=6xb_x;IzlAQHy}4sin_FUS{e;h1grRl z+B08l2IRLbmuQ0aXs)u90Wdg6au+v2qG@2OE~ISxy^W!(3faMCTX6c@eSS9Yj27@F z0Wr$gK=(yERlDRd1b7iJqT?;!$-h^@=BM8=^J4=6Wwz0zdc!i1|GqUj_(EU*8_X}s z{n3J||F9EJUmL20J74aly68-8r#v6(7L~{ZV>L5#LwaBJpKG)lvs@~)?*@|}Q%d12 zmOWrDFDUiTxEpt%ohMcB55kZYcI+XtjUxnFF{qvn3);0Zn;r8snQ*wT#&x{(S&Z!b zvHTU>ZYH?qgA~q^bqRMHI%)Zy@$|bXWoQ(*1YF_d<}b0{H?O_BA;~E(`=k@JUa=gJ zaf;!;dtHr{X=Sf_={+u{adQrb{|BVJUFSYr`(NLA_ap5>nj+jXu-N{!-0~G=#uh7a za3ted6sz5=Pt)}Tv&51?kMU8z+^^a{HJ@bX9aYW*pBjj1Y7_4~tNOhy&Vwv<>@e;+ zPn$B82v~a#g&gW_v5UAm#+loudQ2+u?oB|j=XP=aAvU()Q-lSsE|(P?esL-Wj^ z241vPnlsWd8$D`2J1QpSXv|{dpkCyB{NZ^R19D60%N7UU{r)_VekA?iYAuMbnVu2CT5q?{mlo$sb0Kr zABN}VW9^*=@;M)q0Ucc3_Twf?cU{eWb&mg*JLL#zTCEvXeZ@uPYnVv0zgji+!D%_B zMz-MT9D;M`5}5ERVM?Z7DOU|tVGfnmt@=Fk9S^@eGg?=%PK}aH*QBHjZA8o9DPf11XvaorPl(Q z65jZo?`SXo_S2#}<@_`n*`TdE#Df*`gM~|{^M8;<{o0Ge44FOgop(OoF z<dTi6~=Xu8QT%aix*{&eV+Py@5h^bvfkgr^^(mn2b z4f901MlC{n?>E2guwbd zqIGuE%NoH20fDM9>IyaH&Q{+(uR%MYgvr+=_x%lJT+3u(snnybdqovJa@%m~?W_up9?`ebUu!!52>}Fl#Uv#^k&HGfl%)n3C~>9#GDXm5K8)ep%`d8< zbQhK|*_0YvDVJ9uH+B4sMq;!C`nJ+}BA?0ew*#61978l%Ich`l!%NZnNz4+P*^l^Eb?D z+V&VJWw?bNr9w+lO-8eZ_Y1T$$851PM1V7+%}lHp^-L0c`1^l=VDosDQV!L=!GTY9Ib zmeI-{Wc|#H@uQg4(*p=pL6|N2^gAJ5;nkth-Fb~LqfrgwyOR6Q77Yv`FU&LEf`pK} zH>MT3vuBPuH+v>Fe7tqn*?P%HE`J>AHqfc z3hfu;g!A!@yjOO9t1*oK%yVu#ranxrJ-sJEsgfak`|f55rY$mj~1%Te%I-y}cc#YM5MwFl#xog_%~7o&pgc{oP2#UX2gL%nT!f z6}1(fTEe#DK1onY8M2Zz37)zJ9F%uO=i2b6v1+p{%A88C zg}jT<=0C1@-_%f5Nhu!8p$Ynj3xij&-`G0*gZrV&2QoY?i)Gkv5JsNlluYQzLDeGq zYi2)hfZ4jGM>9k^&Uly@a`ydA7Tmh8Vis-EPT;f}_Rq$q|Zv%Wd4KhZbT zi~srL`#zlqhIl2%YFF_Nqz`YrL|eI9>_1wBKx%KUPTfoec~Z)Al8ZWPNygdt1JT%= z4Ktw>%6!VLK>Mjv1t3yQw&dn4z#s|NqY5Z6R)gJ^5UKGOCPhsBaibTJ))(?I9wHSW z4}{DuD;<}AkN2k(cIg3h1{t9x9ZD0zE61e8*AR5kS_50!Ur1wlM8ix;8-D)>Xvkmf zT|E;X9>+D<9q+r({s%B!J%jTjSB~(s({A0M3Hvd}>G#GFp1{f*YzMO|>mgNfeJcUm z2nJ;e{v#@7k&=-TeiqAIz~pNoJ}<&(vsd}77NcJMAR1&K1_n9l!2VI;;Pg!pt*-C5` z-7xq)RBf1{B(kur*>lq#fitfxn4gtAwVVnqa6fs^80>1e_dRBMs;8f9FelzlZPW0m z>Tgicbi})$$>d^54axFYfc7Ip*iwEsTZb&&7cHlKx%R*) zLaV%caGU}yKOe~5mXco1t(LTOCIS*WZ%KWgk(;NV|WTJNkzob$ zg5lzpjJKr1W;q-?D=@|iNZMS8)5C0}%}$LU;k^!Oc=?h`fj*7<^K%gqbrgcgcoh6| zuq=l&`{K(3by}jq5}UJs%H_s$`n2EvqN?Jl_tYRS;?58DG>Px@C7Y7`=nd_Q1c)SA z*DrB!Ok4J%Dpb~XK&rbk(&OdIQ_X?nwU;=nrE7d*dW02HX~K`{YjPQUc=QeMpo;RP z`T^eY?x~8{p=!2nsW_ITtUcVrtd~nK$U7?^xp%-P4Kwjt~mLm zlY+1_6B2gksiy1d?8Ctr%1sqfQ$AHmu^U?nkp~G1q#4EPV$uZO^vkIe*P~(4;HT8q z781Jj7wd{qT`2a~ti_%`MQH}SJr9j;YNI{{=9u5(?dABzxj{U&ME^uS^ef1QEPjCx zJ?xq`s_Xmpdt;1UiR8@XHEEaKQ=atAJ#9bHx2&MA z25JGl_~35Sn627KeWI`&LrL{s)_5ByY%xuJb$YoHt+^>ovgK?(dSQxHVb;r>%PFeI@_neGyF1Ug;-L!Rqace$Fr1k%WDCY(#*yHfd}>SI zg-*+cb6-0(lpXN?b2u}fyizUlSIN}M=^6h9f^sK4|D%Dv5P9@(k-0b?CKRFs763Z1 zs9?*}C`cgQ1RS+!JMvtj6N(mtBxTV`VpPFmtYbX$iry`~F z^+&3~*3wXt!KEoFUyaCHO7mQE-6%9~e@e%Nj*`{=qm}NF7C;C2VZoqv6@IkbG8X(e zT04C_*vopmBGVQI&le{>^Xj7>er=; z6u7(E^?}>%6KPs~e-FAf_wP!&_j>~RMvb4s;vB}bnW;GJN|4k%*K;K(y0(5Oi`HAm z)$f=KP!P;{VvYKJ`Ar6=Xr=Tm(N`R)dMsioKhV$V*fop{+4ypu5us@Oc&b0+hH?E% z3ap5&IP!G=ct~dUZQi8!6}%wSQK4h$bzCCBX|3-o-i0r$q>9iv!Hj)t zOP&RA*x*8LMdd#_BzX`p-EyInl5S_e9es}y3Hs)I!)UhTAdJ>>jnx)#yUt|-I z(tM7})-`lrc46nHdg4zI8uvXX;wx_V^K6T1y4xVv?Rys;$y6kIhu}9QJNREA%P__Ii`k*Zo!_ z9a*=5d4jyCJf3#ahdlGMt&fQD+;ZoYIcx(a@8|!D>SQ8m%dnAXF=^qpV#M)5FS6m* zP)_FD_#l)Y1Vv1uq-|fevWU=gG;dg|?V!J*<)6}w>02{hdi|e!Xs?V+p+x_fr zF5jT)7FC&H-N`E0>mkJ*;fE*Yh^(y{|F~_%UERoyp1NR!}nE?}W)ls4e>4!yskK05=c<8FldBLR#502iw~ zaXv+h^&tkp@pNLWhqaqe8dBR;mQ=7CsVU3b$nvQU4qxVzDk6=~aJJA5DKd#AU*<7w zWtlB$vhc|ui7BDc;La!l`C@@qj%KA zh;@~an1d)<``c}lO7$3PwSIwhZXN8y%q0y=lN5$o&SUy_;GC!ruFW(66}4hPk!RH6!29?dvL{ zI5Q|8ZK9t=N{113@%qsLo08>F9XB>Hi z!N;Ek7-(i9Kfd%7>!inv09kV1)>d2LlP{OlP zI;lKg6^P6rGx+^GM&52ai9aR!xQB;DyVT6aPw+f`{drb$(J#t^buA|=2VrNj6O7UF zUTK2q2wat3pPz9)xx?|oJtFy0mrfiaM4rcKoM}S!{_5~wBrb#gqh>JF#8czt`nSEPdkBSmiS)iMC#^F_ z)URdz=h0){w%?dT`R_!PieMWLMJ03L?)jEJ$7OGva|3gJa^3n5u&jqA5oCaWc*`iI zmhj8bZpqgLtM{heNgt6-U*x1H%Y>k;ps^hhkG9g~|U zE4>a_6Pt#7_20cru}^E>#k?JtZlg{~?tKY9EKDyi9T!8+&<$~oO?>kG<}gu)FL9b)q%&$U zA)5qH)O_zJ?QV=vTk2iuFSMkvq;krnc~zh2g8N`ghlAQ) z9sW-a#K(#muY8Mwr+<~`yNv5p%fz94KD}fEB`Vq;B=Hm$t}X<{H%b$i!t1%}_e+yW zHg@%(XKA&FOMfbT1rz|mw8#EPm$!Ej*{+Q$`ayE zPHDElFRja}_=!Tbrd8iE>73WUr??g94>dsHg^79rQynurQsAqMt}aFi*Hk{&A^7YnQ$72n4Cx{Z*Jl6|CJ@f9Jn zPgXb^f9@--)$K~YpNmjY@Y&F@Azv|j+C-10gZT#Z01B@M!)$r8^ zw0teeAfrL+I8kW=kIBtGJXKxn=*lPqLqM!i>J+)X3FWQmC6v&e`8TbADP6#pc5*0A z7%FiXt$&ALtap3VA^CWL1i@&E3($AaW$eWSjK`|MB!>W5ELC~w1wOxss-;)Nrbcxs zdnQSS+T%d=DsXaup5fBXUTgQbw9zkiK01+%u2Y_~d{#1SDUF|s+=T0N1HlCAZSIu{ zCEnq^jf>RvQ>5#_AQD5GtvJh`Lt@3lCE3K{_?$Nf^ij^sU6 znZIuw;QjLJWwJI@z`^~)5y_89#!{y)Hr`qF(jdw%MqGPh();aFkyMA_zOG}IiH!t@s8H&Qd9vA(Z`f%ehxB8Ap{@%}?7g0LD<_OmPH0)Qf zb+hEN2VcQ$yY;3==HAcYGNDU9l|f6>|7&i8mpRt<62<_K@+c zP2`$MUoXV!EoS9tI)_gCc7ICSn{dTI7YnFi*IW9{FRt;A9=8q$eXs`Hi+V(8$!wb* z-DLs_ue$}`J_T_l)h@R|!CGDGD_&QM$ct(=o=eg0kJW7)tKwn4#p`>J4sVlhB%kzz zcXog5%aY_wy-s+v#Ql}YS@HKxeb}Rx|NZzkC&>P}%KVhj;fpAhzt;I5V1fD%FcJR1 zfJVl?NwCu|i;EqAa;H==B?R5bIq_cA7^Q}!5WawWYxjTs3d1VwYErh+nyZMtWa9a( zdtiy8F`Q@ly(wOq0=_k;QA@Mh+11ep#Zj=|FD#r5NDJ;_A>=9#rt2m61nq?=<@L>)VM&ExO$ay|)}GnVlz_}#eG@aED9_v9N=slEG=P0iGg z%cr6p+=X}=YZ`$ZM+L2$h4c)&Ql)jfYrz29r=?c#F?a~*?X*UdKZ3`>QyV~z|^;wSXxhLWrPbPi< zZ%!+=>7^ppOcVIjbDW<$jtqfvolH`;Uft(@Rh_6Oe4-92c=4`N~^l~A= zlsPlNFw_ntPatT9DTVDw)@7Q>Nh88c>GBpBzyqGpvi6AM5U}E1#?W>ZNUxUK#^tt& zSi|gTttBTj%1}RG^h^r1D}uOAV%yNbgD_&QJYh1fV<{SKMT_`Jz6^;An2(9Hr3Id* zt#HRDkBCVqJ-BkeNz?-0u}z_D)t$m-ke5ygqTSP8_FylGlTQ)(?tgk{IbQz?U; zSH75Z<**gflX#j<$=5O!9i&5 zea{{4AQX2hrS!~UF}(`=)Ba_p(6K(4EWie^1pJsxC)HlgMLJM&sKklp?axf2ICqd# zR1mqjtzTFqLY^GwM#3oc-e6*_6C}ia_5-_WEWzDgfe|<=4oFb`J$ny}OdN;>tzoJ= z$kxxcMBED#C+0~MKX2U>$n{TUn019*#85LC~cr3h;^aFP+D5XgrHi}BCq3M(5ftfuKDJx&tCXHqxJHXX1_)CwH7?-Q4 z8DjR_NZ`rAco@TxLdVvrsX@>)Je3c$H>B9{AdekG3O^{QD1&W-Dh`vrbHfG?{%iOc zFZD%9u&Xae=*5Yw-3Y1UTBan93vgP86WVOxi{=Z?A}HeQ&Dlb;{qpf#|~(zuT?pFK8p zO01x;^kq}(jK86{pGJtcMr1J>A!_?F`4ma>!uPY0f zS&e>s_4zbD{$kTLRah&|P9|V1_{ywXc3x5VZ?WFja;&7Zao-rf=5qnlkm@l&AgROD zp@}zFyExwF{enYwk7y@fmDWSUpt)1hXsn~pwT?`U?7mCLO7y>=ilRnlMJ;0A%Hd&N zw0>jbfyPOH`sHqR%0HL4%{S$w-|4W+Qwgg3(){feCLw;Z;}3Z*TKoQ&Ct1O8QLsZQ z6JIK`7nrs#iC^E}&@w^(>i>TvoqIgf|NH-6GYpB+)TUuf&7tJ5*c^v2=VK})p%^*L z`4EOsnmMP|V#ORP>EIZO#Hbt^sho3`oJk6w-|PMT{qN@R^4jxyUXSZ>-5=lOY+2KE zt`y^i?ocj>ZtrYKpOn@GNWbI4-y2DP`}*K^rp+YtY+(WZw_%dfmXXAj_nl<1$1;++ zvSP4lM~mB6#XeyCzQ>Ipeeq}WA%yI42dB1pN-4;(v-$7UyPgbz+p?V}Du-V_C^Ym% zzw2ebte$$?d{-WO}PRh~Erfc7Tr0wS-)wOGx>e*`tKR!o-Gq;B23rR)C zUZY-i9vS02?2=*lb$)1HGZvHbvM)X@6}olkhqI4I#`VY?ZNUk@k5ob>8UuP=p zfVQN3H8V_)%=c?e@OC3R$&WaH-IiPKd!o6st6k8}hT#+f7ylZS2U}aPkZ6tkDBC!P znM5jz{^{kdnHpv>d_4XKlz0VG{%n+31YiHm>+@1Qc6tlGA1`mAug1YbT4E4EoY_e>w3E>1SNFT*5=Hc@?)b_h} zHU!PP$HT6!rhmcjYkKofaq@Tyw3E&qWBxK&vU0qHiF!+(5s3V@J*Yp9FCBJn`m|72 zo)8WE5}h}b6Z;r^ZyRlW-%DS03l30Sn%jT6BpU7Zp;58;F1^8FxpZxOcpTl9htG7{ zDkM_s{D>@VhlQ7UE7VYxtrjl?*oH9eXL*u1=9dHXn&ts1>yS2H`jiLl`kP#uii$i$ z-^QHKQj`ht?7~S~av|R*?tQIvrbxbO!&*1?0{k|*WK(NC#DWDajKx1iXW3p_*?=Di zp$Xend6!z}0>xgQ)mTrCxvg5XN3N(dAP6z$rG?^(Da&Te8LGs0!RBQmVz%OV)UtVV zrfL>ZfN-s#f1HTa@lNxjJ?jm!iMG{!9m?YSk&c?h-RD1|0Jk z%@5I@)n^Pi)~bWlBfL~CiOV#N;HbacjC|w>*R1(y3HeKmAuGKB*g z`)>7^U!BHQ%M&C`)kZALi$tZ%(E~3^ zDA>zw2VW)vyYrVIh(^<>EOi!m_ji{Z91-%zXz=skY$T1DR5N2*7|S~_j%hscBc6kIpk|!6Cm>KD0Ojw zK{K<5*;*-Gep8YanRTvazY33d&{ckXCo#s}dq!Vb*T-F83$>*dKM+}-3@ z#BcH*+#=30ytHi?{32FL^zS?KFA|ZRlWF``{^_ZD)hc+@!LtuXo--*!jJxH7`^4ml zIxE4Hq&8i=U31xA$Q@KaOBmI60NR6PoIU1+tyV$Nrl|7qP2JJqIBKMB1m%~{{)TUk z=0CQ3qGM2~J?UgXNh4;J^GIlgO^2Of+bGwPdtgdR;TpNrh<$ExXWFQ-&rpE z|6Dq~_b(y+3;el72*E5d_CXZndT=c z=*{k|x#P`q_PbE*XH+V7ftV6=%jlDoxOLhyk>yIbF+7KaWxn=YdemXQOJdj8U*ERH zXiKpLwn2l6pmIqzT*^^-y6z}2ImRgYczfzs^Qsxa@QGyjxK}qDPMj3W#<*m^`h@6J=B{~ zZT+ps4TR|@pRiO;G8yna$r(A8oAMVz+F;EUxvvq zKivreUZ{Igvy^S7?exF9=LNVaSiTH}SRDKlUPSqcaqaRGST7!vD-06IkhYcg%TG9X zHL|O9x2=Wemee43vMW5->=e-g6({*P@?@SQd;M7O&W|B3_ z*muro=E15xD8rZ(0I5E_c0vJ6tnZSmahV|b)Aot3rZ>ea*1`q-Me`9tMNA=n4YP+Y zYx!OtjD$Gm*y3tlgw48;q zkv(u+;MfIrgZq_#w`Im-`7YE}AN;`W#4=_dUO~lz`;b2hE;vt&lWr8=If72ucC3B% z;r4X%%fL8Vd7WqlaI$+LS+!M%7>~eE${61odG=j8<{^FE9Ta=_J*-LhpTI1NNf#Ih zawgL>*mqryUOrG@?fT0d^rQONP{>Lg$wtK|0jdomMsa!m=Q?j+^F0q~_GI>l*L?o| z{w&e%3*Bb(aoS>Rkn4k*)io)?>p}`qtPTEe^k-;J5e>RzQ&*$G(bjeqev(3t|OnVg-bn@gD(01!xvf(gDa?M`yvpBhKUZK9*Gm@!yLzuzmfz0=j z2z%TPlIotqf_83}e9)|lj(sAGRXeCEtoS}Zd&t@3YNTEtcG%5Pdf^;;U8kCyci3pF zCiKyyIOgD)gJx)v+2e)G&!Ew{4K2ZW&{_@5=8#4RxZZ8oWDI=+n252LDj27jHQ;+L znTz76h7obNBpuIc1yq`-qz{D9rGWBok_?s?G6k6w-le`qJ@D8ciK^C|>+3{e7;l<` z1-*XR@??q-zX1#OEFz>&GQu4;sC)eE>Oy&QfoM&ElV`3kw!mwBP)h`>)IQTpoNBP!E-5yCZGbXQJRL5*lK$ zWW%y?AvlD2(16!#A`{X9Qaupex<1k>=sm{={PGAX*1G8l)@&Lc#pG6wu=s_QLcpih z)j7^EWQ8GdOxol_mO&jLT3W5b+}N4tU=>3$oIMrXU^>0fsFAH#e{3}mw@9Dl*O5P( z;T%l;WE1GiFS~ROKSyrnpUtB8k2jq!$WeG5;2&`w=`H6v8E7(1PPM!qL=&ks&;~{$#ruvy|M~=4iD!>ZIC{bufS;6F4AW6qFOKx&1jKT7nD!6X}(ax&|0FZ5RPjBuvMXZL>o0YmeB4~#Mjq0Q_*x2Pv&~jVrPQcT_;8-|8O=o(WNUnr@A2h2;b)Og zyX52F*Ms$??kMkokI#IgT*zaKg{i__LBE&Jlv;jSSAf}WC9wZ+L5qGq*Ur=u#;SEF zSASey<X`f2q6l=Bsvz8M$*Ps>rXn)8b9{jb*ZMd z_(NH2x>bla}2Z_teHryP`qox0a!x`p;1+-csL;zcs(| z>+6157bDciw@*JO%fxH>kGy}F+_+CA9J7^1O_okK)>$`t$whj_6m@1j>ez{~g}UmT zxnMB!GmWIqR1Gx3f>q5RZ=E?d_Ha$ zE{lJZ*go{=Owt`pf}`2L(9!EX5`!kz17>%<7kV8ZEt$;*p1kaL`ODVqM(Ht5L3hdb z{VzdndE7wXT&LXx(vWL|!3C8HcowGY@ac&cEG=}O)Ui7yl#uTivubUaAC>U*CN%=z zvhz6Nj?&SK)8}PZ#Ek zrO0I)gyxXuTG5BM$!l*ygoYK(TN6O*_`_}C^hzn0x5s(YjHRz{Cr$4ty!+ucKG=5c zrE$iCR%O@^%k!2LL{WjNc1TmM*B(!vzg%K(u@tc`uyI;NbX*=db*Sg}u~)~TU;2T( zD`&!=OFlk~lY~2lYf!$ouC=Eve$49@d|X@eRWsl2z3o2t2=%Az^Lsp*BBJDAtADpL zO?P*j;~dil)*)rFj|{{Ll5c{4-%o|oIueg{CU&RF4CwjVbYcP5#}W33d}RHUXECt* zL8j;}OaUdl)<({Isq(0?^ROu4;n^TvK70a+|4IhtA-ly^aEd$Sjc8XuMD)M`PyJ}d z(qkgK+X|O7NR?2fZEoP|J2^V`OkukM0-S18P`|^c6v`S;BmdeDpmNfJ+EoWLd~MbA zasf|&Y+PRBA@ z;6z73%iEyOVvPax_$n#xDik-NqP0t|8xQH&WVwHrgEX^}8VHN#TEd&fS6~mPY$44k zh6&1(#ljM>3_-a93h_YJI(Vch%pzoyzB1jp2Y|^OBJ@E~Lfa{rwE5Xqk^16ak>j?! zUaM8~s@^)3J?+)uB`sTUrf{8=sm1Wk{w^1an*mhta)0|)R1DgnN1NvqU}f_I`Hyzm zb29g%d4fae`}a2&doJ`tG*ciako*yrG4Cz>sAJ0|*6p5fu(bDtJls!ty2csAiRdnpRHLRLPG)RGI zVxram;Kf+AX`=NwL6*u7@L2CRDLv7Zm8fW6Os{HYsWyY=9yLlbo~;AI{UTurCL%%c z!VWwqrl|yQeWn?k)W?YGxJ*}|(?rIjSo|u&js>KdyXR)WI=GY6#+%tJ%1Hxm^};)0 zqfcrO1^QaAPAT+fu$``g0%lwl)f?~vC19SZLxs_X4c!!4#3tKh`zV3~Zl0&%PNdH# z{ju}7;i7aX+e&ZbZSenqbgi#9VbuaKrZ*C!`_q{3tU<98AV%V_3FpnEz`jwCUt_aa zi8gyYD-xmWLy$B>N3aBXlQD1b<*IkaH|G>mcggte^A5lN&iY7}1AVk95i8#PP4l9k zG(}R`Fzx8x=FYbrJ=|9&@eTK6ck{o{Ynuwks8A5DCoi=-2San2BC4dLgo(R1nmNt_ zu4^I72ICv-NlAH#YC-Bo5gyE?w(A#rlsx6mg+%_9aMSL;#*XUtvHzsn2X78<>n8E; zDrCT~8R2}+hsrs0hb)q!7KWP8ve{fD+pIWQ!g#JN*qHROKtNaPa~6)t`KU2a!}qW{5muDPZ!@7b8)G7-IDR(yj0xXC(TJ+ zP(lx)GziJ}+T}gw77uHV&++iO>&SBb%||lZ8s^~Cytms}GKLG)#V{u3KB`#>VMpxH zt$Xk>Wxt{3wWw8N=`-#sob7>pH$G+lpAkGwUGD~4>XQ!kUNhvZi5qJS83rmltE=y} zuLYUbR`{*fgo6UG4CPbil&u%v%!uVNGZk#j;hN1e3_{rMFpBOTIj1E zUQ{9Haz?Z-!@;^nNEV!u^>)^HClTP1dVI^S62YKZ zDz7|_-#4*x`a-@3^6S>gtii?}u;I*~!MyzA)^Obhw3oYqGk}Ne=r2*4t^E5ZVwmwg zIq`F3w^!CK*;fuFcMGy%n1r-&Kvw~O!Wb(bR253^{V&=tALg}lMmS({Gzp85AI+aU}Az`kkLD%xhpK_Aj6K*Ys(|w zL+IEl1tY?UEa8lIxglME3=*+mMoZMT-$1$BC$;+k`;wcBJ0vLuxvN8I0x<9`|J)rl z0djl2p>Spbh)Wh2Y7N3Kn-@WIx)f>NOa@Q(dj=p1XL&i0wCm?>P)XoK(ge?wyE>Y6 z-v_x&{3Q>xNB0uuxd~q-8+!To188)tk5b(v>TkcjU`Vp9nQjP|ZEHx4{@crQHIvEY z7T%>;VSSD@MB4(bK^VESkq^BxbCgsza91FbSAXme zSno$4Su2uMAn2@`qN95OPJj(tb{4tGmScLe@j}4WsM$1ogL$UN#td>UYSC)yvJH5w z(8f{rYan*q23`*m4vt{xWc_<3JR%{3g_%Saul^~6!f3)!4}U&}Em?+WA{S3Kow6HX ze9Ts-2?4GvLFD4q4*0Ld@kxfEsiiGsFf&u(`y{-lu`~isRh=JW7zX3p$srjZ?omsz zd^Ox5Ds@QJ$ecua2}kFpu= zZUCi=v~%@-uZX?`aKmfU)!u9iUpjtgy6kd23AWA>ICQcQOo89$DJStVs0hY-94x%? z4+JWEX})F4wD3wbkquW!J=nDL+x(cL86gZKfwXN)0oZ~yRUti?Pfn57_~uNZz;16q zf99>T;T^#28^O|ykb%r(xkLRC@M#(lWcK$9P<1FVYThiV{WH_yMQ6_l4&a*IM%#0; z@8X~dX+)w?R!*TaVBhduU_2BzU<(tMO`l|@`K@Si52lx?bZ$IdMN8d%!np6>W1W}P z&Zkgtze^=uBMRy!2QvaUxdJ)oJs<%Zg&U0?bw)AvaLwYwNF#i;Y9P`-+ zberpYdA-(xTW61YSn&(r$8i@e->24#DTwN~Nlvo>#zB1IM+)Sag0q zB&J{kwS@x?5O8-tY|P{;AeX!FU_jF*i=Vc-FmGLw(p=0NFwX4dDX*rgtAF~M zK~+taUTnX*K(bcPv9jf48v}Tz5U@bsXdt3KQ%$P@1Y({vX71{dsNM#kc~RWxOfBv> zUE5v^bhM!f%La1NzDG+xkk5hX2S=NmS(E&^p@gYE2o`kiqY>A(*yw_dR6;kNMH^4_ z0ovs~RE&Au|2YI19afndVBn znu+acz=F{L;A>3kC=L_mDIl>?B2{%ebBM8A;ZP1w7X5RMd6XofWC`Dl!~xVFf*gas zfo#n&B--<@V-jNCN>`wH&<8+ZlFXmbOL1Q?y;O2AfCJL23Mg#~z$yMrz6H9xqMFt! zQ7}=pgUvx;875%{IVRRo#n1~7;E;i|vcmgSZVny>T`A#_QX!Tsq96f{1wfv19X z6==;Sz~8k#?^c*jAi!{=1l@ATQ4RiB^0QXyQx5h_m3b@qr@^pK@{lL*c3R;}0+VUs z0+O#xw1Pm*F*TuOa>1->_t(0;II;~_5WdLblOENv9}oRSHiZskmhT&&B1ql9L?(cs zZO+@A9T{FFeM~5f25!%eYI!i>3AMW9U=9O7Z!8i89-k}kQs@q%EE)y0G{Xmmo85zT z$-*I_T}Wt7OR30_EUN&5kUYTai7jnXF+=KO$`{+rs@=p2-?j2OU3Wg2-Zs96=2Ch5->|I`XsU3mf{o5eoRncg zj*Vv$QMg6U-9bP}Vf=uv)wkwu1(a9iHPEWTq$z9l$w@18lc(r)xUbhL!C>%A)ZUW> z2B$;P+RyJSvP1gA3&_;0c+PO|wIYGYMYs5{o?9{q>F9=e_^*ITNu|X4|E?P7V3VI^ zf(7ONViq2cCuxq}mM?U2Ws7Sj+VJB-ZXubhkTVxE7eiPpC-a)#G-$uq^SoxDH*}|o zzfPLh`=9USq&D!|fL@KC1KP-sA^Nadxc?@6LL~Y#UDeV^I}YbPGsRsLl9!}_CePIW zrqx9SbzRy}LP<6>l*QW#D}{Rx=Ig%ZN;O5YfUGRD&0^@xg2_^SF5>E9A^q*BJbP_7T+ZI0UcjL z4<;4q{V!GYqzO*9p=P92BW_5q-GRCI&ISENhU4l&TBj5PrW-4GKM zEp9@y?z|`gJc#Hv)yhOf@pup)@_&Fg4oIbw3Ee)|x%|8o5KCCC+6kT3>7#DnD#=Ou zx={>3U~#&HB#dg*IqbkF#q8i`TO{2!aKl6`Wpm zk1&&4hWStbztXS=XO($W-Mkz8o@u|>(gk3Rr>3svCbdaM;b=B6@YhUYau0E26PTMI zd1buvc&0FNmdSl3|%djdlG;L$J9aZ==u^zSY4+D1DX`t zC`m*}k4%(Q)fZZ3u$8a|l(s%^CFbAJiSC8g{0|6SH*cNpj$&Otr=O>t1X{M!@SWFk z1{rNN59$}J2Mu}aE3B)habi*fhl% zbtP8*12u@i^1Akr1R>v+c%zatqIw!?B zeXOlYGZRrt(<*3x@@@Wb_#s@TarIqyW*<+>O={44uRbpmw}`7H1Toa^ry%mn>#ft5 z(}1UEzilG6ihYKKmyo~9ZIl6g$GqvvgQn;Z;sGE^SW#0%P)L1D+#OmFL>Vp0(a!it zga^OBl1tT9pf!uBM0-5_(2x&HU}Q)i7ep5G+EK1?KwR5!p#ljvw{j8{QV?QB&N;yA zesJIhZ^IW;TPuRAl)=fK1X+Vuiqs+WxOCptkEL%#%(fB=eOq_PY$*V z73e!(r1F@u0mqjvNx+daNXXA>4?{;Obt@C64zc~Mz1+O4v>yp>8^Et=2(YM^I{<|% z&wVQrf!VNy|G=gus4CS4ExJ#{DRl{n6qo3#t7jBNbgixo4xP|f<*luM$~bpYeA?+? z%{kWT*{w@!>c&Ys7oYS!JAc=OMj34UFrG7Uw?T5a@1|~sCAsYl6S2|eolb_GWJb|2o@7kKu4VzLjr&!DS7xl$S`3gon>ygOG^5y*JZ{B#kCYnsFy!8xhdGrGRujSc2 z`_v*rL3$vEHD#Z@Uj{#uu&+P?(u1pGftB)W1#SxEP~TInvJwN*?BX1|#ve2Z6jAnd zoqj&$Sqx&Vc$D242tWJwS!#pc9X8vJy-?iY$ieDu_6Ot*=_NJx!#l>6O_~4pGQPS3xzai+6o- zG*isP2coggoP@vGtGbJAQWytONLr8_E1=+8+bPU9E_1S4W&gIX<@N7fvNR2CiEssg zD}ek-7KQe5b7+a=#yx(jaEesbJz#}U{0&Cr144)*sL+x-2*RQTEjyTE-OS>3`Y>CK z_8g_Fr`ehWkpXdDknTB%hu;W}nz*00*#_WeQCmeQ`(V>c@A7qmLDrM0Y-Zuzw6hp# zx7^flEnA)g^|6o_S?;P| zaS8S9dbxp;>Rrl4?prM0V+ja&0IbsYD7J!B=DLR$Pp6`6dBlOgX|7<@tUMl}W0%l? zU19NZdN|I8YBoF%b}mRikU^k@4-o~3J#}ngN|>@t^1Qe$Nj|%Ez_hoQ=WuJ*P%l?l zEsM)M^hEP`qGk|Cp+adoaC4C8-^GG+q*>-xDgP`K6nL74ttkeaXs5qd8Kj{u0He@j z{-6KV+$HUdZYLv*pgD~;)-#Rw;KAr^;wdCecEq5>AUdFOP=L#mx%xHy;{1Dr*m z3!2I>iR6cG;(YCi4!yBHJR;+p;>XnwD7A2$i6*TH_VT;j8LBL?by>>CtrW}F0q;h{ zMh6~+W{tJL$xV?DoV_8X}Q{W^P` z@lAL{RxLjyp3@un*B4v^^rubvWxo%2hO~%(61jJ&c_2BDreeKj94421>qGu9pDcOp z_@l2pe+`TO2Rz0%myA9-*{S%fRB7K_j>G2=TN-@S{@W+kzBMFU9z2gJfE_@m(HzssKpJwfA z`sHP+EL3<`^MhV>KsQo{c8bWUkV`R$sdV`^EFfoAEA_ODw9|Hhu_sXB`TF|Jo=f>z z=OsEdw!Zr{bljYs$_O54w8r)}wrHa&|8gu3Eg3r{DuYpUIWDo~$EEBhDH9b6W|_}r ziY~W^Czrhb$aZDbr+D08<~B>)Isf%ceL-7FuWR?5Hy_|>dXpX)7X~}g_lK+G7XIBi`l>Ces#Q-uqr~FeBddlF@pnTjb+bYy?*0{t>C3Q< zkxGqfXQ+QgNz8}}n@M)>4?Ab}0PmTVHY=eVov8bBen+n3SKQ(Fj;@ftJ)Qs05dE+g zpCVZMKA+$k|436G6u-|u_R6(kLs?y?)=OyMp#)estG1XnFlTC+hDw!$x&J#WSpu_atWHq^%4#*#_+DT z6!U;g6gDth(4Jl|iQi&lN_+r3bb6P}$%Ae?K(I7#T*Vq)9wPXTn~bhNt;A%u_~o$%mjX9GT1qQhKfD$qX};V2{>U}6B|hd z(U8IPCb$NOn8xpb6~R%sK;0xl+0@hAypDiuLla!3@7xv6+NYwot7z%v$<0rqhot$5 z>u$rTmaFq=Di)%5a^-)jnQgMMz!d1U`d31SC-muXr_?fI94#MN6DFHfj<@KF!h7zL z9KjBZs7UN=SKy?Mi-|gr_l zT#I`k4P|Tkrd*k2P!QL~JrX(VESS6hpr|=f1+U`}1d*G%%501%0r;n7GhYP&bkS(a z4-z!X>Z6(n^{+jk101c8yQ@&OHE&TxfYp_%DY$bgecesqij%sJ!}J>*zNOu2!48Ql zLZ+X^%JXQnTcK@2%)gzm*6w?e0CHT@Jb3L-ePmJ^*k1GIALTe#>UyA^x^dMnn+3jg zCe}^GIQicuSDq;J{9YHkSOo8dSN-{Znjla`SIx;c?T!~p^wIc3HNUgYhs{5?ZS&*Y zqDSB1zb3Vm*;(O-4Z8~1S2h143i@7m?)L6Tzbo6f3#aELzD)iys9)6`9RKCz^|ucK z+BHSGF06kWj;+6H@is@N{e%7k5-r_u#wYCNq|0Wq%hPKQ;!=vgJPYt=(@evJeRiF| z3S<7F{QF2}*dYTP+LKQbq~KB6@>-*M&=MnGZcgQ-%`!;q zw26mzQW{>QXVjXHcl6xG{QVj_X0@qtm5??uXtZuPS*BMsO4Tx8Lg@;EvxB*mHq2}wsDu^i z9JF4tmUmdArGfF4{FDxO2n}?7k=Sw(EdI$tXMV8h-hwB;XjvxiV^~L1TXFiRBMbQozVIu6}+{yg?V?{Wsyv@M3im~oCcQ$eQQ( zKQm*k5qr`v(ulnIR(w78rc8g5Ii?+qFgY>>Q6f zzIwPa-fw(S$tJtUCbRYA(nAbi7RNRrYA0$($)#mf{q?|UT_eM~!$V2T}fls)e5 zjT=)*X&+7lc$iDN(GE?jZNO?le?xvC4){b3Nn-Z3bYXc94GrW}0y9+MCAm;eg+do3 zNWH&LQ}UAlhmAe}O;jH(fJv&8)J#aFkTG2!M6Y1Ly^_~!BGC?X$JKsrdYd(o|LzST z)sOZoYVruD923>)J4ZEWlc5W!*M;I(C{zZUH?Lj6Y&nfr zHD77jde-N!y;%5)szqUD$YPu7O+rtFvTi*qiJ4?6y`Y|xVa_Wea4R7oC~D*B#bYDS zj`LeYEWTDvjH|92IxC2wIbF<|uhdED^Jc{T{+#4U?Ws*O{wie@=ts+%d2+q=r@~+^ z@H6Y}x9feJW%1W>;#x!Vo({@v;|>B^(_y1L;OQc^ksgu7wZ#kl>G@&(H-Tc(T-QAD zyjp{_>GQq#wdLduq)A0bbGS}{u5aGR4eoGp=E&XVpEtZDEQqStO;r*t$cO*BTr2!x znY3AcT08Zk6j<30gzio%8$5Z71rZ?<4^MT>cSG8~I4fgUX$sk8+D<>~Rcal!IXDwlG9HOB+ z^qs4`l^(b5Wdc_w)EXHlFpdg3hsLT^bCZ}8#j}-b#pBQALdy^`C+fQQpH6@>$b9F+ zI2DBZQwu~?`x|zVyP!Ddev}{qD?aRNw|yfq@@bY&((|W7@*TAa+#|zNQxVVflP2n# z(xcUfGLH3#&swWaZ19dJS!)f)iaOoWyioSL?4H0vwtn7)8f1-YwEfMs_E#=&|39ZS zey%aYYW4q$eGPPtxY4g%ehRhyf8LZ|em=Uw)AdHOcdgaajHGMX!F z3y4H-7XLC+}xbssQC2C~dL{h)#80T^1cXp0tXckXu#u_a8rEnN+f(Dg7zO#eTUlRBk{b*4ABzxwbC?or!T>99^0ti8`=+ZV{)x3Qj z`)ueB10TBCe>S`Ovu!}Fbz8^Jkc>Ea=uzSyPzMs!OsS}D3UU3y9tfN`vQH<%*lGW8 z*;$l3Wmna4=SpQ{>N}}}y&ESzUFk{-or6m0PQ%)2F6GF8`?ceAVvl(InN^##>!*JF z!R;I4^Zm+5XU3g5a`@JpQCqzIr9Y9N?&Qqs-$739E8OSvx^YYP-}}E;hTLKcOcC<$ zucD(V34hnQRDz^~(~4etNlnoYX{7IyXG1@Ft!7`2pw^xtdDj03&0mjwddY9llv?=< z>!)=8U$Npm81M2umnU|`tio?h;?I+_RW6eITC~_HPZWf{9{K&l|Ho`LVr7hNG1;Va zw*K`BVeNR+ogrJ(hC8D?XDPKOh|-F)y5xkeKeMfBf5ddL7gm;Dl|L&wJuzf)tg9v0 zqEYI+lqjau{65F+u$}<-q>fF=g4UCtDIH99RlX=hwbcS0CfP=k==<%+4VpFhiBq z9+hd=)qnWMo7b;+->eK+oT)n>ShbL#V3^*%sa(Vx8($F~C-*rlP$Y6pfs>*Pkf$mrbb(BQzP`Y$K*koo`0O1Cq(RMjvm zDVu{k{{6W6%7F2{?m|@l!J!Yf9sDZwlKdp>yxYS^^9OvK9VNnEb@v)&mXP<=Vs&yh zpXqU|n{4P3?0*-APv*?uk{G^w*Q6_N+sKpm-#ON}XByD)MMtw}Az~nck(V1`|9weC zkZ=}@)Q&#Xyx{8n?(*$Qgu8J3wNAHOlG})1&l@CpYcO^!Eq(ey+smFOW^yJz_jKOt zB7+H16I88ualh3_zgCHqN@c;?o4F1<){%DGS+4&BTKDt#YqzF9@kPd7Iv-d)V&2Ns zRD4SmiI;e?ddcSPz=^7hirw8a6Z;;RsE78x|0Je5auX#IU1+}i<&4kcBmoux^96DU zUEp*(=ltPOEP00<@#8!0`;WDg$?q=e3#v^U3~}%$|2qAvX}j7YV-mz~#uC;wStqWT zq*G$loaYDb-+tMGusPlHi2jt~dM^3trmo5CdpDi_ib1Z$f(E_)`BMlNu*mVAFZ#n&&%b3|LxfwCH;5L+pukwk{5Ky$DV$E$8G;9 z^a2BBA9P#Ut}MNXUFqKUR>XJkczKvVeEC>%SNdDV%OX^=&PsfH$rM$k{db&WUjkkMEsmi`UBAMW zp5*E#&1S~VdVA@6iLv|IMs@P(cMm+tX~xKXmH83tVEg>b>?-VLqxeL+jaweQ(>N(A zeEz#9FJ<$8z{2cjkAdopBY*tPUevbyg4=Hl-xqaieC?yAT`i^*fgPn1Dq!#Q4_3wR zb8aD5l7FA_M}9Be>|1U`1s*JF&ImU$5)0y8Nj%jJbdJ@aW!vQ|;^IxVfyD3{TfD zsqfS6TTQI=omSq|Yu@y*QSu6B7d_xWYU!Nv=F_z+J*^R=(!I)ZU*N}!%3Q05SCa}) z)Ea|&d=+Gn>6yjsqijdp?+m@5E5zR{$fbHxO-8>QHBl4LRdjeT99pt?>=sz(YIhW`&I0DP%LtBU81Q{I+0GP3c^)G)6PiBCXzQfDXV6FwPxF{>{q#E-6~U zr>4%}$EXO|_-4r5ylkUmdxWlSE2DCLl~W;gcTPb7Q)I6Q?mvLWoDj*wog$pDhOv0rzC> z2|SGN%{CQ3>y(Lgr`Bj5(k8$YcDaWu+*Ta8_$byT#et2mYqmUu zZBfkXQ@=y$ixOxQ;4 z80$MHOUxY65VQWcQ#Z@O%YMDmy2RxJUZ2qREFEN}Nj6TBBE5rzxZiK8_|~*v zY3_(RZm6M;taOj{pl@kWC+;-stL8~EYEfdFwc*sUvvA6$#^=#%v(QRRkd8^*7WYoa z`xD1%EL`pqx6?Wcxdr2)FH-KE9h%0^UkLtLn>n(LwYk@?Ip*q=OxJst5Fw&}yTaYK zOZeQ-;NH)}Ju(nk2Gra3tqb1%f(#k*>xi7g=-QXVNH0Up-M(73(`zM-n=g-jw#mOm znq4iMykpvTIV-^Pn$$%O|7{-|hY-1PuO3+j@?H>yalTNcd{|V{SdqoM-Z|w_{P1b8 z3QQ_DIf%Jc4#SjU1{vFfWJr4Wedf%Q*axQACYRUvIK@|9e>2!#Tgq)QWsWX-TApcZ z;+l` zcpv?hV@8kuJd~zJ>48)6@BmQvBXID=d&(QY&K~Y|LGkCR-#T`nT^0~Qfj*sV7_M%X zEu-T>JpdTAuxKJyM~Q%UlSpJUi;sYz*?ar6;KqSvGa7S(3I`z{LGt)TZl+WSQK$$y zn)7uaKSF1t2vg4DUo$amJx4T{1C!pI7h>^1 z3s%Qmkgyq=}zjo1El~X0vH~& z9+^ET8-dB9-PJ`SIUKsxk6oijz>bx}F64 zG{~iThg{~J)xetEthP~|%<9Y4x|+_`Oz=S;k$9HpJQ)T-tytoaTOvrxTO5_?R5g2f zUdZ`&$#O^VKY9}av(I$eF_I~xGwsg*0O8F=q}*OyLeF#}GLADD$TYB^pGA@xth}tL z1^{~fDyN5HiX@ge?IwYT1!O#a6toMBx70N+woH@S4=r**R?dEuN2>j)9JPhSL~*%? z6+|VZ8qA0q?Ih$JQ^!#Zo(pI;)id$$0JS=C1`q5#U&=`zBL;;T+5mE=* z;d9C9RkXs(Y82b~IOAwJt6UN>)}#Y-=b#iweqNZTJ`Y-ngKg_T8CP7n7^?lVo@&&? zEPZM}FP_+}Ugm`xj&0k}3i8)q_^eml{#EN&Csq{&ZvJ=|)aXDaTCmMF1{o3FOiN zfJS+w!8~Sw9yrE-I#Gg8r8Eq1$3E2K$E5%?a@3~*NjRiD9zQw?h8R4G08xYZ((*c< zDFT7hG^eIb05q`ZDF$)JUi9W AARD;%l1_X|@in~bN%_AWrt|~=b0qaZ#NBZpS?$no;(!q{;2M+eoR0lzi^Fa1MLdiUXaG`hNfcp80~sfp zP(jbFHAW6c-ZTK4a8BXYl#|9dG`oQVIn5-D5qAOYKmq7L;Lxf7$vm1|1Mbu~I2&jH zWm34nrj^J99+Y6`pT?XyIqg6V$<8s-fzzp=oN{xIT2adP=713W%w$t=K*${_-0_Y* zsYXu(&;o!+Z%ys?fI!m+cJAXst0ulet61 zMSUgBy|(*!{MiTalZtMRR9GoUoe=a_>M8n=RM6U(&mm!c_WJD=J#R9syZ=osLH2PONGh#Ffz0>wE}ysR@NKe zvrLb?oCYbu{AgWUMvR!`F^LBpr9rBj)zqUgO>N~7{;omK=TW-e?>N)#U4Oh!S`rJJ zwsS`%x|r}&wkk`QZRfTta)@L90BH26w9w|;62!5b{nb2Fx2Y3ZNF|Sac+Sz((&{3u zji#S%<|Uv^w@|Ith&0<+CY9o54It;{=~-GavX4=51(n2Z!|yh6ine~u7PQN{aVw6T zm8xSj?5&p3Sp=^6Im?{ltiSH?!FbO#kEcoH>T|{RmXkb$Zq>4WU~(xnVJC6D{Y50J z@>7F_z#{~EQaffcahgnKWT1%+1vnU|ILwENdI6t$!R}~^my7<2=N#g^gpzz|_}8%C zkMw89p|2hQ{{X^fKt21@k5RpkU$XxIk96ajwT>#4k=qLM)3sZd=0mrxIIPecJ|6!7 zkNFS%GC#~$OI&w^SrIMFtQ=9nrGAvy{^n|l90#T;tH&mn zy*P9f0GY_7BxgPPQ-@PbdH`IL&~Zze6Ti}te)RxI1vhp^4JrG^fj|Ml9dSxOl}5EA zebt}rE%-6aXnLnow{~X;zX>f{+LR9F@T|5(WlvDMdy(ZfSua0~}I0WxyTj>)Mcg>Huj6 z{{XK|JFqtLDns6y+zLQ7GJ9f|XxvC2T2a(}DZPy_4m|;<6$NRL{m;gn-*%7+5U&8x zqXVriB7M_90UL!hsm?o7*Y1z4H~8^D3OM)2N=5=s{VBwJ^$`2#>p%|jlygc3Mn_6_ zV?gL=0guK%S`=eFXBT}g#ZN#D+S|d;;ZDHbX#OEf>p;ZE&5qQhWZ>eW{w1gX038Ayw*a1LyHu|| zYEST{{{W2uD&P+EGJN2Ul;P5Y-hq*_;+Ly(I34M(cwenZ-*9~>5QGLi{uI)2PW};0 z9ch5^ou?fAYZg1bQ0gxXwz2)3ec2WH{Httz(_AH#^MAryB6+W~art)~RV|oY-oDf> zEoMbq7n(Ne%6e3nx2tJpw+#!pJyk|cRn=vYFXIU#C?lXatJ9J!;yF1Yj=(OPe;xg= zn2PGYjJP#sRz!`OLxck)nrr|F#{!=y4G|0h1ab{7pk~?KSkEhqX(Mb7c8aXAX@c>j zkIEMVuG7sXnG8`+aplB`ll&?Lb5~KK__r#G32tyY({V%4=yfe4Pt+YGxEPRq)&~Zm z{{Vz;(YX^`0(~T|f5$S(6!vB}WIybjRXDG1gmGf#VaNRRG}3*?%8vfV#cW_Pqsbu7 za_1(jn1iVVk&(q>_-<)sw|_E6d>`UsYk7Z*=}^UM7;#x&L#S?Rpw|7tUQJk@JFmA! z@U(tc@opT{uFa08udP^af}3JUripp|q#QvPl1DB&)}t#GBY=BTQ||sXD|H`AO^Vhl ztcRm>j`Xj}dsK`Eprn7hNr9Zh_ncFPGyXJx5v9P#O2wL@t^Nh8`eM9!2mDr^yw|bc zf5)-^0Cv1_c=4b73aUFXy^mM6Bq9yD&wA4fwE*i>?7!4USKFm!9e}ansM0-pY&#Zf ha!U65%UykQIshNq(i{x=C!hEd*G*k8*okg`|Jj=d@lgN( literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-15-Laboratory Safety Work.jpg b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-15-Laboratory Safety Work.jpg new file mode 100644 index 0000000000000000000000000000000000000000..ea70efd5bf1e1f9ded6f31d969dfb316507c2d29 GIT binary patch literal 132435 zcmbTdc~p{J{5|?0Dr%-#SZIi*=2+%{g#(K9<~-t5WM*2aG>JH3nq(sp&Z&fmA%$ zcd`dSAOHY~AHd%az#%|JS{foPB?Ey#WMyUK;s-~)rP*a7&HMEU% zH8l5V!Qpxb^!FH>AdyIQT?;hI)Y`}lY5MP%fMjK5I zq(;Ffbq z*>`ht^XT{4MV#W2(z5bLk8A6m)HgIfefjEjTRXR-^9`?mU~p)7?rLo4-9)C(Np3y1Q#um8p+8Q_fUb`>dDhE;gratXtj?by zY44i)>61Y=F+C;Ff-tZoJW;o%JT$g8S@ zZYq|pjE)x{xpl^-%OAXCeU`Pp>TQ(|nty@o&sSMA^~l4Mt$qg`9b<_KF3QmsyG7QjX|~Lpp7q5%DF-Zn|DH zDZMqG*-2V>goAr6eEXcow`TH65!mcgD#!f!Rc{&^Q~aaNSaqa_;KWcE;P)VTjY8KV zU3XDFy6iPP1e&Wd^(RW+J^!$d$uFLuq}fOjdl*1&C<0C1Zg0Fl+2(Wt67`ftd@1E10;{3wxS(u?`Ag&?4-q zhpk{}(At6!1FU%stZ?hVFOV@T@Fxf~h;(Wrtg~$|0R9A3g#>PfPm`rBILp(NKojl~ zOOoHnlBb2E8`jHS(e!|cNPSS(K6e5g(!jE7LF?J#ESK!y3mDlRTwoR}$O{9qP%CYL zD%4(2R^w%kQ>ST(rCZnq9^F*J3`R(+-QL1=2JBJxx(^?a{KmgLP3gK2@ML7?0Ob`L z16wzI`>yDF`pg%vUwPUUe`pFGb)LcQ1TL>uM|?w0iYh1M7z=X&uot(~axY%X9Q}Oi zt0KoKx-OR*Gv|9>t1FO1e@x8Qr402e)R00LBQ^CC4$M`y7v*MfK@?udH?y=eL>)@Augr zgt7K~cXScjKZvpjM#onhR4!WyF{0Ef@d%S=n^#^u{=^9RgL61+HBpT)ZhE!#MbPC9 z6*c9h5iY6Z@eRLgscJ{H!5MB&(=`Dh(C(?rW-g11#!e6J%M|72H*En4{Ul$(RJhH4 zIZ~D(LdR#%7^xhI(*-EL%7m97WRDxBDXM+5aYK*Q^Wy6*`41rH5Q3rmXRm9_c65|- z&Vjq*zZ#gojoy`7!U@%RqqFr%0v*e>&akN1a&^-UFru#PJalWuvmNp#^sj7Z-^7u} z2HDEyU0Q0VL&kGrHKuDXB14nSUP9YOhVakZN8-#YOS|u8n%)T;@F$x(bvq#c0_KFw zPrvXw2F67b3Q4tOUm8YvzuL=CkRt>PeL^jkCl=uD41`R>%;EAFlzNpB=fuI4T$4K^ znrU6hKI&!Vt)@;oc`3Bok<7g+%4`kd2L44(w>xI7U)3!I&L1r_yYRb7k<@;!Z#y|K zS9`5j?#BDLKA=ad!s_r}Y z?3wyXgmtUc^$Hd3k(heCFzU41Z?0|Kox^dNIi9GV&yGb1@vcFRm8V!gLiH#+nHpW| zafNT-^>?02p;Enn#&+3vTxy;3)_!~Y(4$*F9_(r9t9uS_Fa1cWTp+U;3k&w!(*QxA_a)(TV#Ds0OM0@fnp}W?L_{fh7{6 zp|7|>`tL1ZUAw0G^%O`yS&OJHm>5^FKc|zoaa54tE)_5mDEXvP)Cos~b^Zl3X|gOZ zMOMT{dl6O0psZ zUD5gODPsCaW$x;YHPf`Xa(TO-%53AV=Y%y;<@9LLQtLNjz!5QQp)Ovd)nj;N24;R2 z8iGj~mOe4>m!z$JmV8tnkJ1I#nXfD6F<#J`<>Iw!R4VVa-cCu!Rgf96vquk%?0|q( zR{j^c0?uhG;Ft4WTeK(Qlfj)6OStVXzBwwt-P|=Y@m@EATWs*<=21nRGWd^fL)^ zls3JRoe-I@QcZt75Uu@QbcBRwY}Kt#)dpm%;)d~*vheK+2?>K^BIUW$B410~kDi@7 zeFw@ew5AK`7DBt<=&gXx5EySBMi86NOyfv&xXj$H&UF&s3P#zHQN< zjSVhQX@ffp1T#LYe~Ns!IQ1^C_22!(FrM{`(*i>$4s1&=cv*XAXP)I)arA4c9g*p6 z{x@e)Xg7R7Mf@J)X0)OCuIyZyf%;xQ>4wyJU2;OD4{1=!Rj*-q9=&$L#T>+8KL@uA%@7RY@;eHOd%I!LL zO)|4p?QG=&a6s`nYRUlf%`p4OxP{$m15%{nWA!eb2a!QOA=~Xk(n(})V3<$Zy!2%y zzlCoi?*^LfH%GM{J@PX0Sp%CAF_Xf} z>s>Cw7mVdJHmnau& zkj+wn2^>cB5aQ8MFCg8pD=F-KSc+*8bzXLyU!wj342dzKWO;eE;~M3ijqs$e6qmE?X3=OdJ*-Qj+W;bDeM5i8Nx7 zjEmKsYCkM>5l-FZtUAIvd@!i}i zciqBurOennb36Lo|08ZECJtR#Xj>5cvFjQB^VH#ReDv|DD9)(}6-hxjWX3Jnsa4$_ zk1<7Cw3*~gYw?;-(GKc#b`ylI?UV5;%DW)i%)y-i$a-qcFWufdSffKDgr4VLU!qd{ z;7BIZq|NneO($1**a_}?6uR(0=9~Wc;80sVK(8ogZb8_Et7+LPNO@e%$Jy(@yFPl0 zwAIh{PxA?yJ!~1iY4Da z87Z~%ve`k*`)$e6R(*2fiZfI3QIZSe^uYNOA9Z9Gmxk`bS}tcFWc1T8DjWKJDK9MM zjvMGd-v03pDvN#ojL^R@W`>#VW@ieIUa9ch7+RKYR!u-eA9v}DmwS@yf3o#<`c`TG_6S2^MK*qd!x2SOTw0l`3$Pn6J!X+qZny{bp16t0PixR<>8 zIe?Kn!D&3F7g3x46;PRRVdl<_QW+#{h;9mIQ-77@+L^R9Rd3n=5L!|#Zrub0?P($o zx>*mD)Yt{lh32)JVUPg3S|{r_(TTyV-A0kzy9>si=S)SjPJXJ|#SP$z)QFBmTG$pg zYYu}9ZQy{J1At<{0!C%d5iP@658LuA*+!J_w9HDynb9B~B)~g=fIthgfK3I{^1p)q zol2Ej9F$$UUb=^XA*qW9=*;S;?TWuV|Qqv$eihNUm$kZT> zJ$~&oW<$@WvAU{6S*9qJ!D!?}s(gZh7rMOi=5I{H|3nAJXfyFtHZe$b!DKL?+mId8*O$9ah_P2yJSO$}?k2 zQnuekE2U?Gq#R==%O7u#FqWx2>)wYf*;!K>8)Z#Webi{6V4vW#E7FM%emLV7f?%Ec zlg;-_|JHlw+U4cbsV72??#NWQL+^W*WuQEz!0y)e$1jD; zll9Nlmlgg4pnWabTjYFRv{soxneF2$cwGnW`Q+%-G^u6H+7T*rrN>W2CiZ6|9 z>PnD>yWVu~EpD7{?YBE6jFmrUKq>-%k1BmkGk=Mx+OCI$o^P&??q5G2iQ&HfmLoQrHMt z?m5-8%qy7o@GYkzP$kN`C3ZMQ`|**Dxf5>+H)|9{W_f+09QxI}i?rY}5LI+P8E)d#JFoYX%eS*$ z;E`nqcz>V@jmXd{K%?i2dz;40VPM~|i2XzlGJZGr>P&ytK4@-hcNiZDG;&@0R<6O= zFkK{&H=VWA<17o!?UgkA};nAz=tw~S(=CX;t8vKHf5LD8M}$&gFfsnBK} z!kkM#9dH$-xXd{;Yrtb!Qa{Aamy!GBDxJ7ip@Vw91T>m8Z^4GCmBm3gfG*r?4O>tG z4f#L;Oxob2zBR8QGDyHNvY0rNQ^;S%N+gzoY*mx27TkW8#1!fHKf7!sjg}Sqq-C%K zbyhx^*D&#OZnrnEL+3w^%v`D6ru| z?f%qp(=6R5!mj(WyzlS1&(;g8e#9+Sp~|gPztZ%;oL{oCFUCCvRNZ@w=1M1DU>D|@RD_$Bp{Ph@^+JiPxgy~=DV%|oJj zbP%dB74H+i)TJq_Ehjs%FyniFciz1^%Mhykbb)-LNkL(CzSBTy4)Ie=iTlRwCw<9< zy^ieU7(4*3g zcexa_t=r1S|5QA?R5C)!@_o1T7DkeC`0weMmCZ-&&^K)Ag7rP~3$J^?Rm-+DaDQ|= zlvS8ag_!(a2+TiFC>XG3m+f-U{W9k>-EyZzRHn^lj^2A@ubeL8Y(Rx>8{ayb?1PIZEY}4P~)~0 zPU`N%+Q};=uWh`z(7kWIy1z1^@>pz*cznPBUjZE?oQvM^u7C!SsanJv1ofrn7!~Xcl1)5+| zED$1bt8>chPaI21D7v3oEryjA83;59?&T1811b|3$131JkXb-1k|k%d+Q;w&EzdgO zY@I#hW-u~(8gxZ`pzSLpa3jU#6GxInbAwMHL_q9}vm;8SSX82qLUl=Op?D1?Cm?DS zn!^Byn%=iCNS&Gt6e!8;meQDUN1L=Au=r+zxu&Mqy3DSJEOFK!W26afkOF6nDaq5W zTMsnSZYkEfucRT`scXHvTBi7`eFZpI`DM-zDF9M$?Fk_`xQn?9VKiA`6(ttTqgaSA zIEeyyIy8u9uVT?DbYYS_9b>r5>8hlp5Rf18rhPpMJb7+Le( zVU?G=HaPNP(8w~l#{K&oy%r1q6g64?q2%3cH<9cY-{X}^ax1iJx47?#4gU!}GfDs9 z{|x*#fI1WowvE*$`Edkmm6SKyRJFmWGn+oSiXtwK={&hO9we{SI$6?a<&7%c>oI3G zY>w{KNWhk?OB)ol*(VUY?F~&uNx$g*k+QZu$~kdFy;0&^pl@^>Gh4Unwq?NHJ$Za! z*dC&GI5gZiKITKJH_jO{)J)y19_n)Z{nX^QrGc2Q=J)VSldt%zsh!$e4F+GGH!ma}f0rhDrvT-UrwIXa$ zv0brluMY8phyzpp!9;~N1;FLA88!ODLhg-?``{nV^U}Ym1r+DGvV* zSZGl$#G!jd`XMUMJv|% z(WMr1`I~9?D{Ip~U#iy~`B}b{IrSWE=wFj{%vGR1hp_4=FxklJJFg3Hri6ioz>u-! zw5#%$odZisprJM|vf=P_$5R(%In;dCDKA~L6DRe@0u_5Kwa}!!yC7d_ePlBG=Av*L zRkiO&XzvTa+V6^dROQGKd4fMbG6C*?THC@P(!W zc(-qYe)LQf5!a^TeeUXIgNv{>C)CF0X?FYOusfN((h1(rYxAG>xq>M+*J%wL4Mh@0 ztp};Z6y;mA>toPjWQmd{4^eQyp?>u$IG>s2wdyG`zgZU=N*hoGPVguwR*2Y;lI^z8 z$x1?!yf6@T911O^SJY+<{mEV^T9eS5B2s|Z0p6*VHi;C>m7Ho=ky_aL3spehK}`WA zVyg)4#2khp&s&&rqKywA24u`;9Ol>h+8`x?XeqD+O6ZiadEaNcLZp?agb3az;_@8^ zzfMx+KKNc1~o3b^OqS)IgO|`WIPJ?Fta* zKsHB%&m3rex>DpKzeEi#|EBn8ta_!qj~h66ueAQ6Gj?Psqp6SazGC5WMZiLv7nleb zh%QI?d082J*PjvO)s-7(@075rkCI!jj@01XFS1 z^lgvj`^8T?Iz{C_n>^?>#0->h&N*iuR{j@IhROnwq_x5AsqdP6!KwJC?ul+rr=>5* z7b!k4ypC)x%COS)&e%(jD0TD+;n>a>o#5RwJbx(g<7&v)xS!KGW!RWK0!EU#)2FU9 zaIbIP3H;s8@;*be(E8>}%kh?(Hdm;=D9taGcz4W%??Ox0VTtpp>Da`xUyWz}H(nG+8!BawT$aPe*BKjW!y(w~InzY|LK;5ouyp z5>bL`*}$7qXCZaF2U@kjCCdEa{PdaLYt-MuNXNY*0mt`R*jR;-R6*Okf0vYS#lT41=do$n6IuFezJbjes?S}DXZ7cq7~e{ z4=&!T?UwW`aJ6UJu%wU_@7Hl3G7OZKBKI%(X&$NvqJLJE<@%^ZkbTv)EV#&n%H57S z8DFl?=T6Ac$Iji{cEyQrP6%Asnk08dTLhE;9KldLzI5bPlh*_V&_;-0Jq zx%!*CsAxg7|M*-DO2CR#+QZg{nWRH@YrLi22z^ZF;X_|;jptM&5w)NmXtG!5R4>Hherh2L6E;odEnxsS7TXK4_LM+m1@TsjI(02J z^r=A4&|PXQG$j)QEVE153~}>N1?a;m7*IzMM?jZ^vB5L;DB&9Kw_FgAP6x5XqG$;e zNYuNZ>?R?<#LJ;3A+Po-)(bF@2G06lKu+;91|^0MNjNj?4e|&kqhU6nc;RKq^E|kW4$cC+XH7elEzm%-{6Q>_spdZGY9H+v~Ic)e0kX z)n_<<6L*Q@@=M;=mP-A|j!Kwanx zALU-_Y>OXJHRAm9=}=$0uV*oDkHG}UrN2&>%b}5F1Qp0wqJu&@g?2;g*peiiDQOh-G(bk|Po0<_? z&xvcG=PL$gowL#2of5s~^JTt4U~@e~W4ozG0KrwV-a9?u%&QaR?nJ!tCZ##&zd*uH z8oF%$a0Uv(tn=~@CLzk^O^unxO&W@;kOtvjz)kCZ$uFJ4OUGlZHpKZ8(Wxjeob{s` z<5&m!R5}xc5DlG7A6e@!YH8mI8KuVs7Nra|_rDFavP#Vvh9XxjOcS|bA7XuPhxuO73Tda2cVo2&< z=4I9PfufH_Xv3o)Ks1tEOXK;)43b~;M|Q|LEn> zz$vAdHxb2>In{wa?6Ska%&T)lf8l}#2wShet~E6pfi?)kj1hl%Z`bM3sycejBGB;k z&c`Zm4^ACYdf*XQhu;D=!c0sJY@+gc#VQidG-_*F50v4pnJ8d zwb@y$)-VnKsVbB`o+c8CJzYzz*aKC))uoTVrXh6c!5;*D3!7~N3tcc*p=WZQ>+1p2 zIUMke6O#=Eny_9>5>M<@uP<}f-SzbPx zMI0r!=!jjK9W>Dv0+kmt`>WUem}asZTHhnEot)Bb;wnKNF8B)^hd~=)D=bN_z-3tI z;8CR}C9$o<+Z|G1=hmZDz{?_ZsvzaPzlU|UhfTme-APemVj{#$(XBj9Fv+P^`-Z4X zLPtFiiMd%fNlLFh@vC*L*uIeuCZ~6)k>tG_*Y&of@Tef}jKjKgmVC6(Q3?Wv6=|(g zOum`8@W{#+dVn*f?k_YduuoQx^{QS8%x%YONB8$811M$d%Q38rxV+1QJo`t=@gYJF zRPnyeI-@Ml6@a5R|yME_vz1#F_SYwl`Bfq~;&F*xeF0Sv3&~ z;}GS*k)+=#yvmQEp`XQsL&e;dN{RV>NI?BPL{H6%75%KNxKJXty79)K9@uipmG{@p zq8uUwKLqsd{CQ&q;;{Ld&#$QjT3op`k_{mTChcc4K83u1xE-Yb=&Z^l?&ILgXJO?} z`c^g!1nyUw1D@Z$X7#wIeNd2}rnk%~3S{sKII&}&Ro?-+=0~0^k-Kb&TuC3#GVSQ; zM#FCIVZ8vRu~REbtDjyA)!7MyO&8wrd17pDaX(&?Fz?`jTRc0^yIk7R^P(XrCMDfh z!?j+mY&TK+UU{5-+t}f_{v5N$ue(nfDyxzDVUn<}0~I0W@Tv@d#6ViBcf#N=t(W|I zQq_4Cjjd+%=j=3y#IuJ!@d9FwvVVpyXI+`!wsLuSLL4m`YZc4L96`${ppNK*>!6%l)_Q2~Unq9TeqbC>Rwq@y+IItjek!VXXcVvJe;1n|@8ZY>H;-cy+g5+>vp#30xh$oxPb zJQt$(&A++e6sl+MQLD$@_=Gaoj0opL^`D_`-gV%whK&v+UQB zil6gmOPt%cv*Lj)oS3P3coz|?kR{y0T|^9|R9&FLCSo4o*<0T^&3QY&CeUxca5w0k zczOeG^^-L(1)TcrtrvJDIiatfS*v~$10jiBmjo{wsaIm9NO%}3uhrS=R$>KGDL7y+F8CFUfNk!KKGJYyz9v23U>XW6;c8x}D%ejDUVg)q(GkDny zb-?VA-bsPA&l`m?AFbY4^B{$n8AivXP${&5+}2LEibV(OU3K(p3-i0q)LFrJ-9Ck2 z>!Fis5I#9@f5-xRpk;l}-LXF_Wn~Xq4Imu<`p&FSYlJ~p&qs6Ur{7j`w^zQFNFPwI z3QVurj*>1trhbpyb|k+A)^_?}$?pAUaRulj_nUf>BjRVhK9mFqbl5TS6Cm=8|9MS$ zVvjP{{$2(4-S-M}=lnj>?oF^BTr^xB&_65WeQ{E^)s2nl5KX7k`j{gAaw+38iI@3k ztL7yeg2&-EG@#9%J)}ZyOc*jfA!5glJduy%vv6>ojz=(#&X%qGHLh7Dn&X z_K66C!KcwEn7QZe%!l_}uwlIh$vd79z1vHuWFDQei9l7vrA3Q5=Gvv*QZhPmu=4K6 zFmY;cssJ`F?lg=P|I;NhCmMUGMt_0ti2v}x@O78%xcYtco)^D#b0$KIff;+t%mNa6 zo{-{|r|kG~fM(cd;jmrqMvMvNoD%Yr8EO6gRhmOusI=J(&dXk6A`$-?q8ogAw{n@6 zQk8m_*W!V3vw=T`t7EQ!}K4{0&tVlacNoCkh(MD?;L)#wp&hCsjSy92XhDq z@^*!c-!H1H^wsQ+fq0Y=^jF;I1?HBWg*H)SB`hbNHvjPo$9!`xF@UrC7$!+HG z(p86HYXuzSgO1D^rTVd=Xq&8=c>1^grr>v>>988c^-e$ zciYob9kKw`H}`s96Mw|-(&@?($lJpo27Cw891nzQh|K%kM4@~CSvj^{@>#ojb(QI1 znP*wkd18+>#7b-cOFVJxG^9>3-HcGID0V&WQ3?k!5R@415vT{ua&edhuMQzErw-7G z=X63c9rfnKV4?@6=m9-=?^dCsEV_)eHk}OUyIYAfAYxny+N&|s1tp|M^$@(NYz+Kdo7^d3*4V_=}1W{vHea_NxEDXWf zk{5~&`x*vvn6X=1a`|Avnj}l9EVCQf7DyRsA{SYNSeJv0^C>3XW^};)qmbCyrC!MC z5lgKE3?G@*Aka)!$esDcBd>beS^s2{R=_-X>a>ThO{qHwox!EQiGU;xC z_MX3GUm?OD9ov3!zbTkT4d3-4;-APn8MLeZr}Y*Y*{7od#mN}dkMPenuF;Z&bRL6^ zr*6G02TM;$^JtENpWc4$e&)Md4@r~R7kpPPta=b;upXu%RItyHDR>g9G^^3s3%fP` zHdA^B$M|fEIu2-kK#)FEX^;%e z9^4kAc`N2gJGZ8yL21z)C^D9Y(>y-@EMbjaYINWq^R_n&&(s@}Rt#JY(%GS2by#w@ z14Od~!YmOUPj}EBXUnnWx1hh3UYj`g(p@pcZ-lxw;QoHypwaKMEK-WrH{)nin>R^a zlN*rw8rQ8awgiawdRvmR8jINQUO)zZh7u^s6QpB6@MZP_NBA!dFmjrtVq@6ir4FV9 z{JZn+2J~8lb~6~(a+`#FdY3*5u8LyUncSiu|CA%vTV^qkMOG9mYBHx80KAytiV)7e zAu=S%7AJk#68#mnTnxmsusr!~TxhaQRxRU94r;~~h#=7eMfu@}CAKV$V3Q#vSJgn` zP_kls4#izOf(}4zH}D=UJb@ZfY@5=7z^uGEJNR!;Y0KVSuS7*^iqKlI;RB$hIZ&sFO$WR(;wNdFkyu)@7)+9~niR5fIb!XB@Nd5YkkZ{mI{D&K zt*XI1@tjywq>b>#pKX%2Q#mj`s#tucM8U*S+WA8;8f}oH9WC_!Xae)Z1NIi(;bw7$ zhUsn@X#L8Px9}y8v{11PR0*i`C`)&KHMf|EDPe@dO8c*uEVE$nGvr|kL~_cjb@q@F zZhlgWtIY7Q-FFekYua_v+V3Rr#AoAd+dCU1sX_f(M(edK_#(cmyod@W61tLK3R6{nn0-3H%p-))a8r#a+x#haoJ*C6*U&09uSr;b0+Jp?xA7j&y(dFnWKf946nL0 zKASGiezcz|p!VL>N{vzc;+45{p+iOV0L)S#^w=HL`KhaZ$aUW{blZ`7s(_&oY4CaH zSDo3i;N2~1w<=`r*%fPacDX8sobb59Dzgh%UH1*eUT7VY^K2HPy7lc^TZP9<^uKvB@(2=KsWCaU<5arVPkoBO4UPCjI8 zBX=7$r8IP3QgU#lUvvl0r)ov~ugDS+j3WeVy0hrDCfi*=l$8k8X9kr9aL! z-=w@r9G@u*xYCq}-`*)CAL$$$HF-EZwJ_s_K!3AWK~3>Or#pZlx15v7E7QsGLU&yz zL`}G3+8T^GB2n|5=r?=w%&5P-74{znvCU=b;%d%yHT1X1o0)qI_*eA1U}+JJwr}oJ z4!+P%kvI5?z=g)@@}3r|Dc~{&u3Frf%FO1u|WDntc=DsP&zKO{0+qPF5 zeUG{Mz&sm9oZ#&}DdUqz6(Qyi}q~B5HG?4FV{0s#k&%I55UrE(STrv(aI2 z^(p`$x-fVE(P4>YH)pROC+gCDsKS!^Ms&X19d1k#ak|BQPV9FvRpIN}=IgbJ?OV{C zJw%d@e(X&0)DSC=wJGB?t6qu8WO3Y6?ofx=*3{clQWBV`*DVfqb*yl{S`sia_F>j? zruuQqtV6`Meb!=(B8a(9V@WjXo4z8dg(GHggkOo_iTW)0{_8}$J;0njiX+Yfr9Uip z1H5v7r2Qtia_b;_Uau`VT*SG4!|hT4{VcX|Y_o0NrBbH9InY@D*8J`a8>ch;!Yn4_(`N#WPA;(jr3Ca(c^H8* zx^sf;yVc$wLa8&^cyFzgihCng8hmZ9b8^-wk4Jj7?A2@ zdSo7clH}L+Vy*C~p3TWln=5sNt3l#m*0kdI-)HfTP2fa7P=^=0*Ay@V7r(KoO8Z?ZMr}V#qc$+PvVqo z#o>^a3qK~KkqsTt#4uq;F@$|-$6p}*<@d||DgTQBVoq@2(qG^m`adPo|LvI=VneAD;-6H$6Nz+hM+Kq~U_y zt{zDyI=sM$2dLxH9z@3Jj^DsYJ55DDIMZEpGdE=Pe#lbnx!)OmqGtIL1$ z@M?8H!JBN5uEx@cxBUV6jWMxJ?6h=6!9C3VyXqsO!%83Oe?1$#>KY*(o<`lz?=(ND zfU@>a)T=-whFjg%Rze=K*hhPw4Sd`I$zHCo@;Ee}K5$-mtTd0xegZa@qnRtqqlw@#FgAG(BWLk7tfvZ$s69g# zTZOb*FU+Q1Y|N}37aBe58U%p`=h$SO`YD#~F0Xynyq7`he~ba~Iba5l)Hh6*R7>EA zrqdEFI06+(Z`ggx`F-wTTb_T^Q?p$gNKp}Sbdq-9wV4W6hVPt zvU5UA8zPSdN_CZ!dC5}{)dKs5a@2%Y(>dyQ^|my2bU}tk(-4Ve?-5ym!gVs)sb$vL zCQHUb%o6mV3gUb^yO3S|*=e|RI>(0LDUm`6R3zDTavzroIqX?A`M!g8C3DQL|1=(M5^#Lg*O=1Ip3)xaB+3z$x6y7m~R&8_P@w*YrG zhOLX@2s>HQig_JEckBK7tV^G8Nq~q|d_VB2RjunxQ$&bqX( z$K>lb=PrCZjA@B)3YD%zeT$X%*?ebWia*r!9&troee*DLu(07K@Pjl6YY#vSp0yeS z9UnPxmJ!E^xpvDwUd8-825e0^=e(bzSl$^)xz;Tic^M zxFyu49-JUrtC2?f?SGqEI`y5@K)O`UNB(-3NtGC{t1_LHOJ*bt9)}slA_kKp<*zO} z;3q688TmE&MQ~s3)uOe*Msf@ar7Y=uAVEO&Ez+|8XxzZX^&TF5Ce{g_lsk&gBc6*A z($}b)cf05J6tw#i?E?>~^y>s85>PzocFY**$<>m-uA+=Z z|9(oPJhNWehPkbIr?kQQgnww^(%$$Y2h$)vb5^@&-`2sR(1Rb+O4~@DzdVeJAMMhjQ%;g;c@2~1JQc$eo(y7-q@l@Yo09*G@>|sUPjhWmZX|xpEJp;r`cG3fy!Hy(&*tp0kFeE z!FY;;njWAd4utBVSO-a1W-TL%Gy4ZAZ(4p%T1nAjVMkP=f|b@c{t--PkHKZ z-G|c$l@Ktp_6~d(r#52=W6S>J&Z3n$N5vWOL#hetD4u4Ho2{yVSEm-_V!Fb?Lf#!Ov1+XvydNZ>Ph`jMkG0Qm z?Au|xZ>Z?nV~xs>uTg6FjS^D@L%eUmRHi8d>if3!k$CsFqj<8#=PctS{Ye&*l0Ohx zKlqyT5=eZn>Dk?q0@I({e&Ykl~a~xMp zXY0Cw1dtYyXcBlX)n1#DXzBam@W)P+O8L#aleD}l$1Qpw2jTI!LVD?uAf0>~CZ!*> z4DXSwJEna);<%#3+-R`l>!~Q2Ev&J@vu?|JmlrePdhSnr))gHrZWzA`fNWL6wku7+ zd8)=nJ&ldDRrv+gTee`}7UndWGFF-&$i*uFn*GkiW9-MdBqUZ?L|&|H7ryMoH{w5# z;s4t$maSsSB(5|q{RaOZn(2Fv^bhpf=wiZ`+pG5Hgk6Te>(oYm_#MNL=3=ka#o<3Q zvK^in-9f%;_ngMT&&}g&HgZV2jDUqauQzE5yo3tEYd5-hhAurGh zEc{JO@4l*Hr-TA7uM|4@_-1k1AGD{dUbP^X75jy0h`mioD|C$>gU zz2?NQ-EceJU_18m$_BUPAa;aVs$1k^*V?t@sddHF|5Tu)_)WW48O}}3xHr_79=DWE zPzTRb6xx?g%+I)_kx7l3vkUx{A^t&rPt}ASQ7S#Et&r^o3PxdAvw@*7+G;dHL)4Uf zci*fn_y<}P<+#Y+FVXrdB|BjK7*C#Jx>w3!Y%Jm_6)hmp?x*D?3rn#eZIE*;hlF&X zYMkI<5nCkb>Th7>ps4Z#W#FfuEa3fVJU3bB8z+8{%7dUwtk?q7!V57cV}1u47dwuP z^&F7um(c#X#v&=Nu#$z+F_uW55coo&&3`j=wBh`iZIS4AVwnIcyC=;e9*?&neXNzcw0I#w1k+i!wx8^!@_ia z!#ckK^R_p^ppm9Tjwwvu0Z(GD0=ezbrQn%Y#2!)P0>yAbln732As{R8BbXd zGC>IIpr(#>Ksmdg9R8eD1&TzX^|us`<8y<)EdU0He4m!)sEc&u4;={pW&YLb^An}Y z1Xk%}5xZ5Kug=^Bk+?W0F?nbzwxRXBYrr|rJ5Q`2{#AW@-=%B%m^VxlYntJ)XOblv zP){+pXm&auYtMR>gJiad4YJ>IayI*K(3nj+<+{*@I&(e;0raQ(WvE`5+ou`f(_4)C zu%0`JrIa^U%5z2MMY0i1t~Yu_*VQbn1_tc;^;l5>+kZQ^ z>iyd{FaA9qW^B>r5+uDS0sh61sBR*JE>5pjEY#exh0$f5E`5cSr@0%KN5K}#q-K}w z1gttG#TBkbg6&5&r4YhXi~m4$ubA5$`u~j#uHO%dWr=+|rjhg>I7&!4`JO9!H6#1j zFpazZ;>=aQLp?qHUT<1cO+_SN0>F?VK>bg@4J;nX5vK@=xNdeX`}$ssy49ieO!Iu* zjOJn28-^nRJpVxTE)x;_s%`#^(viz@uR^@>vr1o?g_zRUt`tJtp!ji{HT*x2%|cbQ z(?sNIl4A_XvS$FqXi+Wsrss0R&re`B=lik4Hq$lSXfi=HqOm!mP?TDf6^$i7FJ z;ln+b;9o!UxSx9Vpg{=4);6|;OTss6a{@P)f2ckZo3BxfMCn1DS)d~5XuaNt2Urn<_`mU=O=7D&C^e8h(`+DJ#KaL0$edf?9C{Euov1$M~{A2ph zs4Nh(VdW+k@fD4*^5{zLO^NadWe!kPPQc4$lpzqOAxEw;AXQl!EP+tr`VAyFA)kiD zirA~vWn(^>Hp+fTNb;M~yyz9{eU8KF-SUVbC&hrKU9`u;9SjR+Q`KUi!eV9%14v_3t=1OoHBW_X@Rxy`Iyvj=~eC`C! zc7vHR$Vw4!?Ae3&V{)6uCpP(C>i^sUkv!FMv^uR{!>GBrKl&XjdXt4 z5)yRYK0?6{0ZUzD{^GIrgy0R{@>wMVXJYBdBxR^u9ZSU?{3q3F&(&L3TVy;jx97pH z+jvGlR&O!Z`2+4(+D~OVql)5{1-qn$NMUDZ84d?f-!~`*uwcVB07fR!60n)}opsc03 z|1-1&QDW#bJ0j?J{mIrF#}uz`VB}lYvw6}ch7TnOEI3)3KfR8=0J2!&U8$P<7*E}} z30OnMcT~B;*nkTV$7zE-;HeTXK72ZWyg~PEg&pmoJEvd2H6?2^ZX|Wu_lU0FUIpRX zqCY3MB<** z3)yd+!TJ!pgzjkBmsgzH;NEC=AcT8trJD&`s6OQntRpXfO5>8thD4ukhiw|Xcy`Wc zvP)jv=H^yLg9cv~Sll9dlFMEa_u!}TXp6Z2o%5YHCh>dD{>dj=lmIp51&vF z!CY|_Epk-t_ewr05td?Dr_x|9C~!C=5;iL~%XxAvweJFhd`2NSTp0*q38O%R)5iQp zGRkzw43JfS^Sxdg!BLiJ?(}_2Q0|u~2TcKo4^~>Im?QTZ&UVcP*n;G9bM-4J44ou9 z11b)l#b5)HRv8uIDA!ZIu|=IB@@^O(A#;@mOYcx1r6HC8b+uhVqe?Q6{E-eQf5F*$ zSsl40o7(~aV3lIckmkhA0XCz9kZ+87AdOnpTh1rZJ6I37o*h6?anL^Gh#n`jtAKr) zD~xP-c??I|cPJA3iJXMrVqD0*LI)})SFoD_n|Z0fvc;R=0av{Bd?%&Tz=K#6-Y)E- z{<%9noD9R+!$&h?XHYd?O<>Ok)720c?S|xG`&rSJi)jCct^L3*2`~t(SarA0_DEUITWAg8eCxF#*B^QbX3XQdJ=z2N zj=U^4+=`j*&1RQx)u6dfoy&S1$3U$%MI)#_t_hx~q=7Q6^WBG^pRUQ@8L+#eHvcYJ z&+P&`Ld-uu_u_QP*MQ}LK*mkvvq+7*!%J>gjb1y#YdtQDaDyOGHPn4wJBOLm={7P> zWPuU<(N-l|i(AVBNNGzL7Fby3d8oEk#0DlWODulgugU!S>y!uSnc7aUeLO?lLM||` zscY9mI{y1N%Nu8IT)xgHP{=?wSMZSTa4AvkdZcT|yPSSU{n|G0zUCXEFL%~R;4SNk z)03m&5KRBoB~{z-kn2XB0i6IQE{VnNPl=L#1Wd=|(EUQDZZ@Q9#H8w##y1vIn^rzX zs^#P>$V|BD2kiZnQC;pdw8t=>)Wp2kejfzL8`dh|x6XdKo9-y=bJQ9~5Lr()(#wlx zi*K&}aoWKADXeI)6e|yUOx|VU=ll*93r?j*YGMT{f2QYu{v&_BQ)GZIsz750ooWHD zL=nnD$Z#JZ%l`PeBj0%2GV}HB@SD5U8Pt!p`m%Rve^u48cDKTBPg$E^mK>ApRD3h?XYPeo z5M1J?sv!RYt#vM--7Fa{S;-KI#bRTsysujKO8#|I-lGKNEIajR+_?+X7JHIcPWDMY z2bA@R`A$itM^v=|3x~*HKkl?4(oD_|OaEH7WapGC7;P}iqho-tP~~?(JL>Weq5F+L z&&{Pw)|=9cA2Fyd?qOG5rJ4>Gk|We?d`0s9irU3F&`#HfxWqTiCAZ4tmzKuzsQno| zKwQH={||c847VlUv|xAYd+^ayS%UB-T)q0F&eazc`)lO1UjEwNi49~u-W{kj=LW@N z&Y!Z~7PnyYW%_0vAP>}(_iZ`VtEU{OlGLdfNj0j_oniymE?gnOfD$sE9Zgr4pez|5DTP`pQKc=cXNWXlr^y5|;PQmL?r4|P zh(r~_m`pI>^TAvyOA7w4{l{Wt@LV*7(atSQRbH}$aV$cTW#Al3GrUuolL_3wk!w8L znGjdLHkPam;wMyA>|ONBpU`qv^eSIA4G4J@&rQ^(LC}`dwk9}46@YTa($yQeo4iAMGJK%b5Cb*>rdE=`wn#d@5?Jvo;nSC#`o5l4YQHp0cGIU#nC?c`sg2B~T_3qR zsB)b%bgHEY|S$9aQ7)mj|;c^iVIQhq63G9}6KNs!#0D7TQf*?^bQqmkfC$ zs-wI?${=>AZ2lg=TOy#y**6LSsb8}0-uIBD`sQf9URN_bcB_-Cp8Ne1$vVz0!RQlj zP9LqsM{Ut$3%3M9jQe$ZT+b!j1v!mhxKP=mb_K2p-(}8Lt5prz;HFNaa z%b13+Ge)6qT48PuLAF>yA(ifAV`(D+4eNo>AIzssQwYzp*hh6&f)sVZ)EikyI;-oD z9eP2wGsx%2pPv5VC%$V_9>!#4Hi^CY_QJ*$-cSYt>Z@}p;CsQWQfFqH!S9;mvszQ` z>1>fwOC8g%e+`8kn9sAi@$;+CuAfbp;JBy~u!62=w8U(SZQ}B%5eLTL#XDSRDS&Wxh zb~-VjznkNHkxQZM-DIo6pp7cy+A0U#IFNY5>mSI>M%Cf&gUs;T4|#s}{{uZbCq5fs zgo|`Pbk@EwEm%GW!mLsSxFWeecf-#VLJ~YPW^+bk!YWalq zU>K&~L0?c%AUQHZk0jG2(iyODw1-W)e}54;<&Bepns1+|LR-m)OL92>Yj1wwWuwr! zlX+jDsrDFq9yG$zz$I{A??3c->z8lGD6eJWz^zr*a;sd@@M=%?0;7t4@iQU=>L{3Z zw0ecma$WtQWw|; zphlfV_4M^?Ya)aB`DuaV4YzP;vmeI2A+96^(eGAv@>A~&M_oe$dW3-TN+$Z4$R+L#!J>O@dT)`XRI7s zM~ppZK*}yf!T*6qNhejp@vZkS$m8c&^9@4N%brl^3Arrf*XMp;v+VWUC1Ruco767C z=7PlXAahI)>pt~aOAJosC`f-*Y;p_oG(MCpP_~(ZCk`MVGWLnJa zH4W{#_?So)D1n4pxblQhEe8W%Rjig?h?iJbb@h@=i}P|-y%!xn>=if~>tLDt#p#Ze z_m=9o=?BnwGquMem=t1t6LXk(_-4lw8Ye?)m z^$hP}68o2us>b!?!v2>fAH$%bEzONXQysFC3w2)!5&`59BYJndj{koe5k|N-F-9WG z?GHo(`!qu>#{L?5kpJqGh{!~e0k$*gOOs}R_>XDX6%odLpIIFTdn3JIZORlRwo)CD znRs}HqNWS9FV`mE3Qrat#A7&}S>ad~MtXy{>+O1IZy#!z7aJ!P^ZLt~x{FU6*~nyU zRL$?4=_3L4NN=7iaW;knrS10KZa(qfqSPj2zdaXH(_Yu)cVIQM_wZ2q^<}PBN%3BS z&$?d3O^dVe%O)$O`HAsx19k`1{e>L^eW1z|hu-?lZ~7|c=d2&ec>CQxfh{$(aNsYDb73T*T;;wHS~e(K7aI72M)h{!7#{njt?xc)mThxTIp%T{H@z#B|ara0Gq|&!4ER_d`>Nc;|{j&Ie$JK z^I6h8#jA8=t9tgzfaX#2V;MfMU*^6~@Aid*wr8n{>N@3FCG)R8#ta?Di*|dZA!h}W zx#;fEBJVriWrerst($BC!?3uu_h$9Gyok9eE?k?S!W4Q_++ACrfM(cCGV9*=LAV&m zg<3<`sNu{ntT>sky>IdluF7vA)m+|1)OpyYq|M71iI~T=T#7k>L@Wm$eU>7&+p4IwUg6=(qcXJia4$R)3kq;psT z5}{&sT05t?G;_bY&H|lNhD?BvgX*=w`ggGl)lFZ_5BJPP+%C~HmqT}P9KgBy%nM4K z$<{C9Io7L<;aez?;aEW0 z;qnK+s2dCO`$G#QW(<8Tyem{rp(5Bje+!&V@w;* zEf8KimBPyTmTXkOG^{OotDFz9nBOUb{k;GfApxYg(ILPIk*N~yL#e3Rte{>le6yL6 zGA8E8RT@GeGfVSW+YcmD-LOP{R1mL>vKkxT)cb4QiP?Hjlx_RH8o>A#SI9!>lMUN&2!>Eq|8xfOB zLSFlk=_bUf%*3RhN0=&>ym5&R0*n2p?=(jJA2i=Aj5O)3&_3j2X#=ASs0fP*y-l;S zTC$tqgBCZacvJtweAjWltV?#yg)ik?zA#qH?}|}cZ@_N3%Ev@IQs#cz>5RV@Vt+jd zRb6r2$bL)^maK8^Sy;vIWethdREA zzdHn(n&p1EcjPm3Tw5Vf(0#{dp4>Ppab2zSNZx`DzgpFRYj$jDzJRay`$!^;(T@5S z@)7Yq_v~i{Hr4&cvBb8_{Pt)4n9)b~Tq3hDkHte^)YD!!MXtOxet1m1JxUizuU||) z8CdpBhdvl$0N&1d`d3^c#$E6*Z`A1$$U6~J_tGj{tEx`^Yz2b*N17~LI+O>moA`s- z0&hskaLKOiLIfM|cf%X%uEdR1(nO1etv?k%fW8w|l4m9oy7*-m*B;(E(1ENtEaSIX`((oVfpCJt?B}Qer+E7X0>oHIjUUu=O!Q;%`6L6m`xcOGx)Bm=hs%`Gk;%>YrAGd zBQHH@OWc|7g6KneLTdBnRGLp3^;!mXVCPF5>2=P$xdl9_Bj2XATlH1%hHLHVwI!c? z@$2!UzM5wSeNm?G$-}q0O0DSHY+!|~H@Tpz;6e{voZAWimKl-?y^L%S41+Zo-0FxJ zsFONaR0)A=uJh}U*;If)*yx6v)?LIg5~srZ_p-0PL%`&y2inV= zdDDE#7x9m`N?U2Bhb7=W;(`opO3oPR6>nf>`P$_Rf#C(zS1>6{w=PC~3^wtxi9{z` zytpD{FTnX7>BoOJBsv?FM7@f>NRK@F&=4BzyS5AXSNr~fpyub(jfQ3awv{a?Hz)=V za;2Tjju}L@VJ()+-Mr96V3_2R`%2mQLmWcOc1Q4^!KZx*xl+@x&aNff9*F@gDR=(T z2U%HV7u#LGjmbRR72Z?#meWc@2(7BnGPJ!C`EJ|9`)kQ#QD$2g9fK8N7&S^UVSC~3=f z0O03#5-Hwe42?8-U&-S-WsKl-yJu66yl345olr8|s-^}kS0@Yew{4yX0yW3$RMQ^J z<@LPr=%K8uji#^g^ppUt9ct?DLC@Uv>cLF15K1Nd{ChkSG_lcA6`G9tC3ULrO-@_~j10;4 z%5zfulCxqm&CYm$qD%O+VgaxC0kmeqD7_6P(O@B&tFvEO8wmHW;Xp`RXQm^7T-(r$ zb~krN-vJqW%h&Ry zkC_`-w9^d)h4Wl!4*5jqM$IP06_hR7O1S=ff;$6|Mioo{)K$)>U95J;TaSnH_W+OM%Q z-mYQH+q8L9KEEo333Wx$Rc&8g11v55&h0}L`Ip9X8^=Oc+?cEU5QxC&^O193j5R@t zi^O*`C(c>^4b*Xeu#l2P@%nfpR<-92VD_sOekY^n-u_O$YTN{7fcp^6)v3d{S6=ll zN1w#3S#&6>l}O%azvv&YGqt|oM800b-lpLrQ6maFXTb(z883pWE^#S3Y9{5=P{q^< zOv>dP*%CB8Y`o#~K|i+j(1uo;q?+3C4F<9zLFwWkXm0eaF>h&V6RoIZK=^E;+9&$x z{2jts9m5dsFtfhMNPBz!(Ca{Q*=hXQPk#Wf$*(4IT#~6DPh1!olXNMndejM-8}{6F zxlg~y-fA>y7ZPjexppSMM_l|(qN)YVQaYsI^DJotDtk=7sxg|lVypp1(J(3smW1{@ zmoOS8WRPn8aR{S} z!?a$poVkGta`82OzD(t_QguJQqbff3^wc^A!#^E5ipvhb3&fcWw;L$eHVIu!NL*5K zk!_PVDhcQPT~~4$dU9>N4f=6gkkj>XyK0bWx>KYip)>lnJlVn4Lyda7VI0kRrSB$~ zCiw&2NJh#}HX$Pf{7UuRLe&z-RQZnvU$I zV~gXI(vo5QBH2}1#&&}M#G&^2!O(Q5lQe90wSHNu=VPh|VW;k6LV$ji+&cQo^OCSr zDRI(j#iCvXx;hp1zvg=aL>{NOD0aH@7%4>Dwb z$}T9^4Ha$FcbwV$zEQY}`KR1_rUi-Gj zjT&n5-^=L6;HotareR&RPod{>lk#n@7AWnoN(5)yiZFn?OX_vm;uy#Eyy3P?OZnY7 zUuqCAz3Ui?sRJ{zfRTV6geep1A`)ld9Aw~3x@lsa3GR#2&U?)!DN97?Z&h$EF>IZq zDnY_6>Sj0-ve0b-C=f#j@j4CDkit~}Ga-TCsF?F7^9 z5^={f1?=Lu2-2rZ`iYTCL#0Z8o@}d=W{0OmeLQ1zYYg;igIVVg%U5-(Q0sP-d-U(x zJL~nW#?Y5dlp=DjgW(o?lwpw`uW~M-MB|-(T^p^5hN*o`mJ@jM4VC`Texg8W5mlY^ z-oMZjJND4y1pL>)kE^rdj}^57gWCDcCTo8!B#&mbgbn8$H+aV+_8~IOh!z;SKp5N! zmeaKCaM?bo#Z%WvD#p&4c!quw>q_Z!Ege;xGsL(yU;hWXr0@?E)1V(2zjEb+_lP}Q z%?)aQ>n(J${-Q%3@JxYMUHaZ=q3|#7z{t>kuabkqH!^~xm%`6-Enn7!re4ZHTE#S0 z@+H8;KQs(#9sl+7+{CNXplDQ|FDYD)%jJr_=PyL7xHKj3fDf-$n%2P!>=m@Fc z)4dtcmX06VsS--aTeUF9HZ$?gFvVZLjK~RYkzWw1x2D{#57?{pem+$-oGRgpo;x4^ zSXW44AW|8Bi)NG`A9o*B^vyy)Pe=0Wfv^3S+HvRfdqFEAtg0*JA7+8g)tC%2nJ+#{ z=e_qzBlQ?(E(mTo^44B|I0cY_hqPnIhCZ)m%*v3RQ*zOh#PS95>uj#d?5?EWRvc&L zMPmvNIkp-EJ6&==%nTpP|5V3KQB#;2#>ei^d049=(ji?-Jh^{73dn8a@g>Sh+mPs$ zEB)laZhlL})=UL1jh+EJtW18`6-$RljRjw1Eyb%LEc(zK#4Ysd)5P+>(H-%^4|yUq zu{oZH-YvrzE{t}Or1g`Z$L;za5E4;Li#}udwW@Mc(BM2$cSieU;os<=DhV|47VWTe zEbXIskt>4d1q87ePZ8Tk#xgA(*B={egbuvFknpUA)P$W>j6BRiDr-QU-bt8Ohh{jB zd^r*I{m;(LZQ%_uXHLi3>i1tjA*;Msz#jNrp{b(8ac{i0X98o3J5NkAnih|GzJ%o$ zOUFqzcI;WwH?^+y=2+g6!V4EmgMp?@KEhceu|}>uplPsP<)|<~iy&0tU@!vXto$KD z&ovN2Ue!wtB~>!6ok}j{|HL9@^T#3X)25$r%`K$1^(w3l>vRvyxe(>HfNSfkARgF$ zP2ZPy0gc!?1{~bMan#ilS)U~nC+pvWu7h$NP_@ExfkD#1SN1%#Cq z3{-p2P-#+WOKCDlpLM5bV7fbIjkHyOPcV)&-db`6{=M4EJ5MYTi*F9~`UBuEjs&qV}y1aB&TNJq2mlx<81hMk`aKQ^?*b?6Dy{)%1epakR zSg~Y`8sdMM_INEak=0uJ`rzzUb3Yz8QD8wJK(>aynLnC2!s9yQZj`+G3eVsCK@X85$c5$~AGVtm&^vx<9HO--&nvM{>V*sgmsvN3HJ|8s z+G@#&9BcmM)P+j1(WOXkb#aT8dAX@Ik8^}MUmCo@R()*%@38uyD}k71WeF+kChG-A zie(Ptc17RH8{h+|uo%VSEwL>L$hsW}Zok0Qk$)+p8#GfaHMRea$RLk+KeGEHCT&Cp*6_;7^wV>$ZaojU*u^W^ zaKk(o?4#b`8{=P_A9QQNsv1V#eik0uWsTO?o~eq^KznX$*q!Zfvmf?phkt`a-Ya-& zcec8BZg-3l;=U*691r~AJ$0v=V-z?zi-v-SL3BX!Zb*oq+gQ?P?1G?%V z9c%0VSFI}SfZq?mUPk4?>h{9|4b)8nCkmX?$;&{^I7+=NP9I)2T9?q zUt_hh>1(vL92ge?-S14^XR{itgl2?oM=xao2uA482mK?vO98Ubzu|BgFCo{Hb2KaYwFy4 z5>0MPNBZ##5+SbNTJ37n$`!3=4?;W_JQhYL-iMztK$D(vm)wQVt-0;?y^~)2p)973 z_6d|~x5cLe%#KUYG6IL>_*zl;Q~ zN)Nyq&r@20X!g2R@{;gc5w2b(9;yc(PY&X*zZl^efa^sZMJj(+-ie5-)eYTiN`>qA zZ02%pb3D!_LxqHG;(TW%XKtro8IJ>PJ$fjObtYJS9jm4!Sr5+p8powuHkc!BbFJ9U zk#bxE1<~g3dmSNyKt+F_E=Vrxu1JsJS=9-LrOw8H5jBv^fRj1-dyM1+mPLJ^UGkx= zdbYELl_X+G-VJ}xGfAiS!R@x3(Yu*76>4T#Q9UQf9anumQoRh^{lsCXU%%^$S?k40 z?()zRv=3E6gpxga+)Jxv&FwGP6~0VYFem8U`c|uDH;jLH?L)Q}J!#v=GU(Z z!ku*^Rn)3J%21tDo` zr87;q((Qv#?np~yV>l=R(~asMR8NlEYJ9VJEA)?T@4Ija`gu|57lyRTm_Gi)$x4;7 z1Ovah4KrR<4Yh(EFcn&Ke-`|eMe&~EhE;W)d>N@XhA^usmufYBI!H9=@+qPIf#@hwKe; z&;6!L_H*O>R5*mZyA66n!L!U{uI#u9Mf}_b30}^-7f)T^xIbzd9!)VReE1{ux?T4utphMkkm@*RLefMXh(BZ4U_^h0)dU;1S zKk8IV z39g~*>pFsuruC}{VMR&p2gSM8X!4rf9`CE3p4|oNG^+6_vK~AViV+M#ymWQ;%SodL zpGhm6ov?sc+m{JPILvyP@ws0)dwosA!s_bFhan3s1x#@x%!ZEg3w)FbX;R|w*7eBu zjK`Wzm$S@V=pqI)%>t_HVglh^b}Ay-GL+=>SP3PJj9NcJYzQH5n7Ft zF9uFU^*fC@nuooUeaCts!o3Aw_QWn-KQUGFlBeM%kt3OmeZkB_Y|3IoY!}AtF?UlC~?{2r= zE_|G+_1Ey^!MlCB_$B`{Z5#AgS9hc2*J>|>bp(eAq3Yo0p(6o0sFlO&9|1RT;f%ji^ z=N76v(@H5YMQe!HO_%2%f>eoFO>dUdhKK?5E)t=w~!#umuvc2|Un8 z;g+iW2U=S%L%v$VQ5EgRv#_{gDVK@%F^F?Q01mJ?9aa#^e^iJDG->q^VH>XA1rpa` zNx*#)vtsF^0wvgiV9^!iC<-{MUizRq`B!2HB6xOaNC<%AI#=lG-f7m~N%{*Z$@|KK zW>%4wVu6m_Qf9-uLs(~-G8;+jLk?ohk&aA@gqcTAbNM->0EozB-nzfv#j zPY2>xyrlAipCDgm?OfM$w^%93&~q1`?)fP;-(Mv@Q6#S|`qUs5{6@vG8V@}82$%G} z!?U&QUH2{Q^nR81T+@oxf4WCc?tJ|^S^Ao5lI%!%d(m56S~WCEsBsnPyxfW^U$2ndoE9^(uu5MiEoNRXuDQ@wr4iGt*Th>^u0>SFp#;i!%4JDKk+-S zD};;aO}t>*t@&lM78>D|Myi#VOBvKbeJ<&PRhbLEJegx@>X0;w{bC9LUyNzDUyiD2 zpEumgz3B|Ytxp+$`p~kV^+%Wj37Pz}M_#X5LKYrH?%!#IA{b?;`ujH$LB`_n0(05;7Nr~HQiGjtQhmKdIp3V8r zY);pAy)iqc^ATRsM7RSNqa@`IhQ97(i~|oRM~q`UrFK)o{}$FkkF_h_IJDQ(v-%WR zqutKZSwR{ac9&y4?!?)NG(3y}5Bh9VYl_`IM?!s}UQyWWFN9LWREne5*L%eVfvg7> z+@mhZ`|jsLBUc{5I_dgy!(2Tahs_3RQ+8`7O(<_XqLQcDQc2rUh3mGhEa&L|^=aDg z{$~H*EL!U&a&EBXMv0HxyR<6{(^gK(nhh6`~lvI(oxCiUfB;%-PhwU$HHL0Q2X-z zwi%L$no-Zes~?CN^W^D4+6$!Vhs)1POPf=VcVB!!o)sGoW}o#AVm`iUh75U>IG_>o z6n2&2Vz;NAj#ippb}E#{YS1ICmQx^iU?saNhrz3 z5))CCvbS7%z~jZuF^g)t$dfqJax%}(@Y|`)!Df$$8k-y~d98y)I2~br?o8;kqxGSv{qM{cif0 zF*Unh>(q0G>#hwh_`&#du)MhG+1wy&mZNdoQ2HMN-@#Bk&o_gh?Wgh zf2Xj;|$q)z*2U$k4>A5tLi-cFg~JF2fDxiZeM%$H2FJJe1&BGVBA>za}&eG z=je@3-;X|UQSZI{55&k~HPd{FxPD3D%k|y%G@SR}sMI_U)6adeOmg!uliaS7yvwH~ zRvUVYfb7lzkYUu#+$k*=!dWZ&5fN7T0csS&q@MA^(WCsc2K z(+xG_e)gK9UR0Wl&SE_`K-Xu+b)Rl2IQ<}1w$WyH+>w|k{ZeZI)l5C}BmVh1q-H%5 zbt(QlCOu%YvXGK7yM52S4rz1To{VMLZ0mVhAD=zJ2MRi!NA0$U>}|LQppFZnP3%7tsLtEygz+Nk%xUFcR|68DWT z0uYbn8_Ae{>-%&T<`2A&zMLd$^DNZL?U01g2S88J^g1W}$OiGxSnSk;k4Iqje@FGY zl=F!B%eb$Cb0dT>Z|A1_GhHv-vIG8P`cJ^E90{XsGrfMr?y=9&3g%A_)#*Pq93W?5 zH@#vyav0XZIU@N~=y5o_e#p=`vG%aldn4p`N%>*iuyAecaiu^l-c5>;4O&8iq9CMJ zHE$;Dy*mV^p2t{s0(Ir;U<*j7wfxg|BD@?V<&9^`vhb;z${z0$m(cAgXE{b0$8@=jnl)_}m4O@P#AZ=|^LIke$%}l;1Av*?U)EWW{Oe z_hTjZXtdYS%+eQHx$!r?3dC*JuA=@wp58PZ>i7NsAL}4XWz?v#6cI}HX6#FL#!i+b zlrjD1N6St84z7+E8P!B|4}to>i_&-Zuy@6H{^F^4&?^E$8d`FJGO zmB`>yn1{f&>gq7P2J!IXh*1ls6ss4S`D4!du5bQ<>Wg+6+ShfC3+E+d#?i60Jf+2# zPTJ~o$IIPZuKspe&39i*iWk!eP96!A<959H#y!LPZ&maRf5d)Sr@WbLO7)iZ8&$4L zeENKz4*{{GIC(r$<5|zhU-QV18T^HkOU8A@NP@-F5(2!#_3_A(-^c;CnE}V8_y>cN zbK5cFy8d&+F?lQdak8xcKm9(4hWBJ`eY#y>em3LDvdl>Ky=DofMCW zS?(;awK~30BUtHrN5|X8F8FEu*aJQG5qyBV=wQTfbAUSz>GmEwN7I1+P^f-g+c58I z90l3dvz-#*fA$q$E!xRQlNr>2ohH4TE{;!~BDSnY=5Rr_ z7;mz>qaBLG;jEVF=zbBiUo>5GHAgXp0N$cg&u5RntL7(NlbZ6mB9|7z5c;AFBxOB$2bOV0hf=W5acO;jP*gY4A zvs?cgBTqajXY?zU+Q z^~boCm@A&6o{bUCUCP0bB7be#2ND>R)({AV3vyw0@Vu^};D$F&n1mAZIC=R5lq$vQ zKWa}Bi-Z=$H0>3Hil`#ONO={R&q^;gTPGY^;0co5Ns*i!_6v8runt_V1sV){C3eKA zJj6?C?ed)n$Q^RYD!$YslWGyYW%BFFa~Z>?8JzI54f-J!TXTDXym^D(Xxf8HE*S^K zjX6)rZZD(myDhr>?pk~lY>rzgnIZ^Ol|)Rcse< zEAtJ-^tOC_O^W(bC2QWHpmJET{^M2R;&ykCg1Z0jm-~#;03Ikid_g)cRNVTuSHY^~ zjbTokFK283bN{{p4rZbg+LaE#^B}S6SnWu_76QbIeO7G^e0$H(DTY;I-+TpT> zVh%)tk>Gu8EH`Fse9JWYG^o1B4Y4Op82y|(h&@?|_a1Z!T35Rvi*pIwE6mGDZU<^N zx7>c-urI(9=8C0HI{JP@S=XZ@!(!-#BuNHelJ_?BXu&{^&tU0?vF{*CnJO<%O?9tO zv8}}zIx>ut8{WmT4+zNN)z~YgMw*R%N*%5RCr85}zwMZi0eyM-Ei@B&8+1E?3qfcf zxN;$uCbv=)-Xk z*SFN8GA^h>y*YHO?XQQ5n7q(=U9vaeWQ=ucDioG-Y+;!;F=p0l#f&#$uc4i)M9cFfhtO)wplWo->|L$WopJR^pG79vWQ z(#sQi3^<4p(?73%;P7PJFUl!sBslEa#h58S#MMv8cy&BVD&66g5Lf(?;oeN;i?};I zBJkwSyIUu3!`1bOaA=S0I4Vw3)=BiieH15G9N`8(25)RIucgf?-HeNt$ev{W90Fg? zM-8HL9gKA})UG}0gFiy+3ko!+H<}TJ*+mbLi34*`n5p$VcXRiwpV@quU5#&3j|O= zpT1~ROzA(o8$YPz8=L|_R@e1C49+XgL|9ejM@X2+rbmom9V30?5GG$YQ(Z0qbuTHk zzpAIU+8Og))S3D`HDuWOzlkFR|M)bhL8%>>LjJE}CVR89uU`;M-%*s608#)MKZlqj zI*3+Fr;1YWs{*0x6I~ks!6y0xU#i zL@N_g!azafV$0&$-~uELQ`AWt2|L*1qluG}p`vdg`m8#Vn4T|L$6!A)IlVBOaK+<5n}9NqbZZ0T2i zGxrl?AIAVg_9Q1?Ce7B1+0O+nDQZ|P1*a*P7ssb2gNBgw=!a>RI8?OzJE^?UR?^^aW+CxA% zCPRu@EaZ5v0Aioa3f7^$3*6;A6ju9%-o&udyIMbGLtj&xxg%-BoxJjs_FCZUOv8BH zAj#WDes)1%4#WLbgO{}9erB9&M72=Se3GfPEM0tQT_^kI=;l zX&sfE-R7@(9`PJ5SZ}^uHOZ&YI*6b`l;zAQ*4^zM)v*%S-7uxS$^$TY2#!juD62yo z&A;YyrN#k$=Vs*PAU0i^(P}7Mx|pi#22Mm)9t3_tv9PqF_(4DH=NU1x-T^(yRkakBAd&u zP!_^;N$k2Gy!<2IeODTu*5T53OP{ta+^J5y_SY?^$GA&hfMjhRu#SaQt6DqL&>rkp zefX}lU~~oS+C9o=-gy1fLX+6=8Clk~+6Qvyc}uw`)GiLQ;$6Wx;tb!fsK>tNQr@e4 zgrJ@ggfBD;w=&z;OSU}tyh^4`1nUr>fiK0?B4$5IKOHO*Yg(ukbg{L|q~cETUfN6~ zmxky!CTgc>>P=WPZI~&J^{&=vC6?0>rGIobc1ha!XAhJnfdxN`uyy@LX7h6S;%7j1 zyAOkJ(VQq69O`>Xnxd7M9#-Z1=Ak~drYF4*66rpNP)^KuaSN@5XB~T0+8)mRySX#? zcHbA$jnJy_ko{-xZ*eA%-tAY5_(%trG{({0emws10gEWgn2)<{G}5JCQV;xsuO2R5 zz&%L-*@xb9;m@2Po$dm!#kz%Ydms0I{I6tjenl_cWi+cdODNWU4E=(&voI2Aa+hJ? z>2Hdqk}Q2=Nq*KN&gFACYmo{lSh+yKBIOP!#qLP{0v`2l_>P9!Wx-6{Ne9s&yGy?+ z^)A9_5`GcWoq?dZX$elcHXBrCU zD%D9Og6nw&U5?!7W4{~Q$Y}MIMblsxkIk9_!2D)1P^H`aaaLg;w8ALzA+kYh3t#NL zlz~U61jvU(4YplE>#kDmt(_hyD?`QFDRaq7ElRw49qiGiN8qT_fut9+^uhXPZaFv? zu_pZUf~||P*H+}=8#nbVzcT-^t`Z^jE6t{TYx#OMc^dmc%z&>NQ?7KYtm9<1-aJ@) zq3!$cmlgLVpYq6_xjpbD5u9`w64+Xu=P(BqE}Q?Q?5Gulnn*fuF73QyjCmBV0c^jJ zM4wh&DQJUtxXRBe)UrJlib1Ii$_7K2S+xt`bKJMV_5Tq~x_Nt#jSqVywSpv-BUj@p zL9m?Uh!FzkX;2A>YgZHLqQdK)w|9_7&z&u6CcUs>fOP%oRXFxQX;t?pelW1WHBe%U z_F0VNs)owC6(`t+$z%ekEL|A7#o#a*jgzBz~ivl>02c<=JZ!8WX8 zM0Kx%vZ|`ihyYY=mIq&$ez?4l`_niR{ghX=cN`*Z+kkl;lr;YcQsDc<+}He~BU&dY zPAY>0DH#7Wj8NZX;d27Q;uR9bL&mA?4ZOjvfT=d2xrjy4K<9~#AF1BoELb4W^c*51 znFbfAc3C`YRp>2q!6uDYt%gYNDLZVLTS!~yj6KU3zS+A*RiU8kYOu92;GKFQTUZGQ zi&-<(vPWe3R>35JibbihKncq%Iv$&U%whVJv!C{iMzb*qKUS;Ms&`N|6{q0^;AW%vFy-hjSxLyYdgzYrnO= z8gr0zUJ)dcvgzC__zz?roD#` zi*27{`&0EoIfiZHS>nuyQ>qGR7r(ODPX4_9*eK~AsQKmO+_R9LBgjLJMh`jG`rG`x z@=49(1zy2J8URTflavKW_C$sf2K3TaSfs*mdDqNiRhuGx5a+Pu(3E|RL*k2Lgs zZ-A35ouAH1&scbTNqao42pGm;y$h!mlbz~$5Ay6OQ(Plr3mo!QQL6jJ1B&XnZnV7AJ`e(lj-A0XLjFuc;LBuBy z$9jy5S>5oR7Z6`lu2xAyu#s4g7ij&rhKeql9qzu;g8>u;%r4$*IwkqT;X+IGhWS$9 z!#%~kN(Vxa2K@YxKAUSUMUA~KNF*&_=18y-De9C$EtRg)12In+cyU+DpEfnx@Iw@g zgKsJa86=V(WjQC8|GaZ(zx?2M`0{0++q1<4-|xhcoTTqr^V-J_JPj<#2No0?v zv~wyT8<#hx#<12!(*_RRK=&4|IB#vK47!9C0^044bptZ@K2))e<+3V7FSBiP9v{@M z=6U{jhISzz3fXMzzCtg&SL`!z0|cY#7TOyyx>YHCVaAU6kNm=PfPT}Ufr@pSr5N{E60Kc73CP%@ z)^;?5FHK53Bwy+RL63){E`A{m8qi2B=-P2-v#1We%*Zfu4(I{K%cB{wH)(55M~hnc1E(|Cig>fJZ8L*`O3Ts zFYwdM3_v+%v0-j}x_^Hyfu@dI-%qRWn#tVS1*@S1ECywXR2muOPT#9^4xmZucDNh! zV9qXBTH=xsFOfqS7ifOxgB~!WyWV9N@~r8g61yY9)AfJ_@kh>7AqrZjnmZSw#%a8xBTDs}a3YbhEy-*2% z_<|u`GaT*+!>>5n`h}R#>ghKBA1QA_p~;X%t9O z&XMvsERF3A zjF>4&%ZL(TzTAFKXQf54V?Ox}=Qj3=%_YlD!#XXqh*0zo^N(*)9AEl}cIZV0`a^(T zZbSk&yb_Aq%kAt~>y1NNG@<3OxRz(7gJNC#(g)T5tDFL~Z%M%^d0yhHmOKAI6+Byh z<#XE;_@3*v?}~bjRuk3lg9RJvMP;dlsHXh60`cED8GA734Hfurmq4Q&7Mg1+gMgWv zlf15cK-kiwr%d5mj2&DCU6dCWpM}y8c`BjM9t457q~!+y@oXbzrz~D+wzAQsf zz?OE_HJjsM?*Y&aK9{nDxV`rVIv>4E$KZcI!Ipu0=v2cTONE2VJnmen|uVOT_B{r_Fc#>anpScOFofUdO zbB$fgv^hg^a{@pME%F+!KkF6ZY;m@UEdw&6JMd&6eUy3YtGe};p-7UO!Y-z^EGoQHPm+rH|IAMbqnDJCa zf}sCzb-gVDQe(;msDEVOM*D&NSeO}DX`oV#LF6+ zfYzsQ6ugzn&=56tF+|6PbuXfn0BHx2;R|Ur!??GaR`wuJvBMPB8xjp+@KrcA((dsm z7~$<5Lig7-5QU1#Ehy#JqCNJ|RZ|Yt9*lj}Y$L-}`J@38TAm~>mH_{rCfsj{mxU(U zOAo$Dc)_JCyHuBO=)R(FU+*l44+>8^b6|FDewdL2MTy8;JL;N^1zXpQm9FXS>XNwJ zhBPmx7xE?OPOR&|5&5C4Yz?iz^BV}B7Zx#FWKBI?aRmJT@fusCo(es6&q8TK?`TMN zzg#kSlXIpN$tKcHq%Il>yY|W^j8s7Ymea1&P6@_SVbh&o3=g~0(x?r$AF%1 zTjCNgf(WxvF9dn_szBw`(X(Q6_iS=fmY)DYHJ`8Z2&a@KtlFjPZx+Wl-`c&Wg&#H7 zY*`+Qrf&%BFbaGkgKFyl2d_nt_Im$Qj?q3pun7|@%KDNMSIF_$z|{2hSkuNuCywDR z_;PJYMbw-4OIf=ncQsl)g^myvoDPixj(;Q0<#}>YR_6;J}&d_ybe&Mdzfzy1-wq#ZF}n* z)kS(@23((pHL<9O0omPil@ASeO3{^07IRNc{wBO`xS|U&*6!r}S%<5y$)o!z_wGaa z{p(R8%x9VX4+jbdyY~5(D`!CLkb=u4w4OjD{KJ3KKIw0tXENS@%?jJ;#vE$pvbbhOiiS*rPZ8_cWHGl@rNX8}%1okyLG zg>#80D?V|m^@%tuAY~#Tw;iVgwlSohb4t?ERO(MA!6XKIY4@Dyvwv~OCS)ks*0~Oblh8vQNv=5 znt|Ys7cK=F<)rAZEsRxd0vULfQNi#}JhPRmh?I5GxXiTGhdz{+fOiXCUfe zyc*?AkY19rDHzkxg?8?YW7C*mkCt3D!^>y`^WQGT&%?^oD)HKkW|=O(8P_fBo0}T! zK17M&-qny_W-UfoU8zm@Ctp32*kre5jDUC|ytiU;Pnc;(V)?oa8mVlhlKag_I8Llo z{@7$;O<{d`?WNu^gZmo+&~2^|CVhBF*OxmvSYfx&chV*9)c-)v>Tz493b9uxLl`Zd zNyA(-jyDT|X!$=t?qoJzys-1v5@d2Cc+5BV9(7B)ZfZfo_}F*)*rE-?E@r<9(v`=r ze^_gHiVt{CSX}yH{oI(UJK&&RL_eT4POu1LVa|mfr6j}}_^^C8IJV^M_IZzGvs(1_ zxCiIS_Sp`aP5H>@<&sDf?rMmEU9%6R(_-<$l6J#U;FHxZ@YBvNBxJ1&f50X&=4KyP6*Ya!a02@A6`3FC_6U`f`IjqxlQ-8GJ3 zzf|yFVI}7tKl`tp)dbqviGQFfr=u8=X0|`Dv0%QXe;^jFb&)i-cnMI?ul@$ph(vVX zd41%3w8Vu_op=VSWC!UXSuEXJNOn-LMbR}SNlMM}M99tLSG8H(7&J1LRTCVc*KP^B zl>71-!5BFUlAgTpaLvtV%{O8f6eLtg7K8y_q)-r_K>A{pmd&x^E3B+`B$!Ps7dkqcv{%DOvrp5pb=16brTN_M0u8M?`c8!kGzZf_Yh z^Q>IHJo}Jnq=&Ve^A;?wRQz-A$F14VqOxo{gVpU{qq}gyYOG`NGyUD@ew>?M=cb+n z*Psd?3-98D;+J2)=Wg2ERsg}w52Lf>FW*kCp=HY~sg@J^2MTyBsRxWFGl30!%6KU$ zltNv_e9<&3>)1>?l(~KG*uJgr_scpzN#Gt~mAhqj-bj+xcoLnjM z#~1Ff!V67CriVE$kX~YjS;1WE51t9}CFHm$F&hmEAt=RUWhM)>X-$cok(T`Q;?V)*vCIxCTnOn+PaqLmj=}g)wMtx^W+`_b0jT zP+~^2TN*RE?KGk`_fk*tCURp-Z#k>^`aR*zrrb?H&TGX7WaSi<^`86;2XUtW#w&~K z{uSd;5Am3zPhV4?b8iS_nE=@zh1J4EyTd}hUyv5|jtUk!C?}Ls58aAUl?$#*KOUk9 zl|ME*g%3zkU^L1bVM)rYCZWiH!#r~7+(Kr=e~wQG=+uEed9N4J-p-t5pB;~i>v{Xy zPXT4vt1Y$ckVq$$6aF?7paGqGqehR1ywIaD$~O>lr9>B) zali{~BwTS!c1o}FrVQT%grTZ3!kjN9Y|Ul0xYu9KG^DOmDk#P9bVTqoqWrDf6Y1}n zpL`Z2x=Q&#W0g=tMV*CShG1jNA9d%SYrE|c1wojJ&k+NBbYD$(YM`9DR!;g6x>>Al zv~FJPyAKM3Tt{U`^+Q)IW?OGRgS#}_%%`%}nN3OcA(m6|xDW3>?G(4l#cU$iC?e** zPlyC~u<7(J-e!!!jIggXoP;&zh78@r?&jX-TyazV#S(5{Ma+oH39hen^*7mgXq1$k zQSUs(kU`|9(I%=2ZS)E>^UWi#KV~;C$VwrVze1R3H94|PL7r1! zxtP%&I@9pGv7GI+i};VkC62v}p!-bbPLZ+r6n*-CpertcBVhhHlPI^wC5PH>2OtUh{`jOr|Ai*E(Is=)KXpyngICCJ z6c7ZY7RdnP^XvGauZM9%8_a*XpwgfHZ^mz;{kN`w$Nl9$$S10I_NRr%hD4w@%U1!` zf^(Mv=>59U-I%qY3k4y)qTIROX5&6l5)_!(-FrfX4>S~4N;e_(Z}xAtSV|B2dLv?0 z_cvcqo%0x2b&P3Wl`dNUu%6|!YWb;6znv_-HZ%hpqhDDJD!DY<43F-!v5f;Fc`s1| zz9>)7z#ZMgXsu+b5f0&!Zk1M3@{vhb0oQt5HlMYOeoX&F^f8TR@MQ+a+5Y;1Pj6Y!`c-@KtVPwfa|xIx9uQ_;de?lmt0IyXBcc{@rk& zq8R`64drWA|9@wz|2_Bq=I6a{xuZt5QP<}XuBV#cY+=4RLMh=XvF|GiDm@Z*Wk#|v z_`V^v*IZ9y9Xfb?>s9E*8}bgS-+xt_$$NPq_jl*^;?XXugRh|;8AAio<^>`cn=ceb z$)E*-vKoxF74^tOB79WLfOif(AT*QZ#CuSTtpbF#d&G7xIH{y6fD6TEFFCy<#gJfI*Bll86n@sp@-(RYWyKVOzi&ZGvftcO7@qd6>IOuP z6rRni;OctYw>lI;3r8Qah}vsHD}c)%OPaxghP)g_@7DSv>$=rBH^Ytea><$(>F+xR z93x!EhjQy8(?c#w*Us;&K)%&7GTJ>{XgU4AQS7brZv!~6+JJF#;^KaJH0MkQhp~GA z7MOUp_2u5<>c7l0TRdJx&az7~EhOP{ty(4lyn6rjT9@BZsfcxTW{|Cm5|tI77ajf zR~U4=z#BVxz_FWwyY=a7!tcOzVMgPZ>u3*)nB&PTPx4>wif7RNM9o%h4&rA224%e9 zj&|R2%Ty9&Y|~GbNw`D?qGC@af8>y;e{0XduImI%5{EuaDKqgh8z)}0a+(~G_2wv~^AoT-fbuP5!7+rGRM zm&lyuJFk#(cU4F9j_!1wOML^vOvc5sG1h zI}`Z+tbd?F4BBeEMWdJ}_K6z#*?t?XrGiT_LH@9ahX9~2t*ZSFAtz%gMAzzoL-2T0 zO>TQy{`bgpsUSTKnr|;E)h+_0BLyw4ZS%}1Pfsny6qv}#B%6ZU!d>+G+k!LuAX(EH z|FZ3xU1i1WwO*LTm>pq2hcakCaaSXXt;L1N6D#G3beXBmny^sIhyim5Yk0BxACHJB zbIO}#`Z(E@f8`##{Nr+4)W-z(>*(Gx1%>te3vVsO=w-%(uov8(@7g6|%yVcgzee=g z&Xo68%Jm-n**mh2ll}yn)?%OtLJ5&b6@I7L?11i({gWA&8+|V+IYf?ST{PiJmGVwn zPQ-Y*%jEfYtOwb@+|TVK8MJ6$`eW(E!iogBq%^hLM=^Q?ST4H6cA)++^IV>dtE?^i zABflggO@K;&HYpV965`abIE#l)CAA=Zq{K%tHYUB!s=9=)AF7SBWV0tt8iDpu9qft ztFZBsGbOQchKG6!LBla|r_=cYY2G%d$VO@^CKP&mJ!UgskmhF7womJNfsWXn|MHUw znEmB^E?50d54VV1lR!2Pg>^}53eqDGZ5}x+w@~IbsSY}h6pL8jN9h<9f&y80^$9!$(G_P2yF_H``#vuw7 z40v?F_2_Fj64_W()76gL3WwqCZyAF*iz(9Aa@t2ax~1qUfthyK0y1O?SOOkOjnThw zuDTO4!Tvuo_^+sQ&aV~=ikM-;S5HO~e*GyQ2!BQS1w2fZJ<;lKFcun2SGwJCvyIH6 z`NPEU_scO%&bWuVz_Nus#-Z3MULx^23^N)Smi?#Ykbm?|`|51j6BZ%|uW#?FGH*5! zr)1?xvezQX+b41P1DGeEOy7Jm%DyokQFoh?v-g~?M>1h`92fH1ZlUH`NwF9JZsn>q zz-5@6S1?Es*X1&W-cZUoIAYe*|fUTxq$}_5R3l{^5?g z9(jO1bxCEi=o_Iv%;;)A_88zr8R)3L#gy4ymSysPuN-mbXcb>eD?&e4xOgayvrZc) zyf(Y_vCv`LbV<8ILHi*cxhl|1NadOQT;LdGPy-pXZCb2gS}mAgO<3{2;PvViqsh8O z0C>$6o4#wJSntR05O!p4IFo7MNEZ#@A|o#egyJ zn|NocE+9fH)$lb@LvA=MK|cX#k~5WV_idR7I!*}-$t3i;n!L><`mAP|X^@wSb-I@m zny&~BXJEfO%%YH zvGSFR>yPht;$_qE2pC{#w;kgKHlTvZxs^9OFj@8hk?0018Rr5g(SDChlV8Xx$O2>x zVMEI}5byU1aNYn-L_!=vj0)>iukg?Tu`!&^Y5?4qrU54C07wYB7;IVqC$8$7oDc0E zk9-3#kfBOb$X1$)ydRYGr0seCX%m z#-7IhD%Q-WKsKCG{KCpT@kv10#|eA?4z9omV;QTqs_ciO(Nd2#{f!IfMYJIV_pkA; zXBCN8%m$T9+P=0%!OK0Z(=!pA1f9#qjcZ8xkN04>y4kw0#znl*BPRy$5>cEPYTaDU zEHEj>g!I^IsplVv+BRB){qOgEiTh8d$xZYboM*d{3TyVwZ`Qba) zhCYYWZF4wLnwbfB$<9HW>{pkxhtTmT#GXq+Wk1RE?PaN&=K`0Upwq$wq>Et*VOtSq?!;5QcjUiO!ra|=N%?!9 zk7(6$nt@E0hZ4Dc&Ue*puO-A$JuEKSbb>!9lsw#`{Zi9OEvh;G)(>#j0WbmKg!E>{ zd9V>&X5;n-F`MfT+iu!Bs`LZ}i6H&nH|alSRlAya-Fe!-w<;SVN+td6uVcjXM)YZU zk;zYKlDXfr1ap2mUFP#-WcTx*uoJ^~iHr7=e*(>KrBAuX-W{R&S?@Bak@yXON; zHsX_hjn{i@FE&7C7x@14=d}sGKy(8)a0|xysux<_es(XzKZddMC52f|&_q~C1WgU* z=D$;67e7C&2Lfx{I8D=x9U$6ZtUb^#$imqEhga%9G)N)3mt(Qkqw(R|@3QYBhz?e(`OgXn#Mgsd7q;2@FT`D@;OP{`}i z0VjHjTye*qU@xlhNgK{Ng`$Wsv-L=mq&{htFoz%4fJLk1=X~->z1=o%H>q2L#7zr~ z`!xc%#w&$<4BN&P16vcod>{#a<2EKYk5jv6(+E4sLaQ&~SY6AE~?e^?&vK{X|zP?q&Ms1XDx z(T8@;ml(lY1qd$x^2}PGgcdF`rmb68^D<|hOK07eN$vF*$u1MuX+MB=3j_eMpb(m= zI?3xvqd<8o+3F*lv^epNhthAdL85LzDM~te+t_pO;y9G{=T-lK3bB#=bGVd8H|-|fC%i0fo4pUJ!S{d0{p^1s{m~9pXF;;x+0-4-I{5SBm-DC< zg4hX*cNMKb3Vmy@ISS4wDiy?qFm00_*f?w%M>a%AWQh_UJSXx3FY5nQ)>G!VsXdb?V7;D4a9?7A%AIhxlnVR65t4}RthlE28p zrO91qC*i=U;9}+_CVO7ju9B67-Yb5^7p_bu=+p#=G(`gmR0*|aFwIX0!xJ0kq$%uT zToEu~3N1?EWltCYmOQT4YE`Bl6wacsCXKZ0v!T=$^l5Z&zy#=YP$^jcWQZmT)XR1I zMeA_9&nmLL9a!d(*=|?Lmh%AqULGq_OmfiW$Fx7}IL3l1&!?ToN;xkls&ESG$#`s7 zItZ-Si_-Y_$n0uJl6Z9Dg?Q&+f!~)-=k`EGs936GGA>{ttBEp|23}Ke`P%7)6q;tN zB<$F5z=1u^D0<5R)&kIk3>9C!s|y+uayPEuZKA)XD%-&6ZU%+?=kyN+w=!wg-mN7* zl{uB$1FxJ;+djkm&X-8@g#@h#_37OQZhO-^HFE7K2MuZqsd7lRGuq7SFPBqNsJc@*AG zgq8G_+N2mFNcGx*udClfxFzxSLg;q6H+$8$`j2xK%!h_XhUeeiqrLIR(@R9dxvIG4 z=EINA!635@`pLY|sO673G%o~UcH6d3teoxO)EMQ_OJ|CMg3 zdq>Pb-L^3}e^qCZ|9}0o{I$!F>`q=_TVCsGL-3-h5zA!zLQ%Y@1`;*fc9biM-9G=I z?4|9Ym?9BQzophrl;%cDMLV}E$tb1>Z`dLDCg^JtDlRSsl`)V8Mx?|^idlwK;F5Kf zrI@6{az-xk>&;m;nh-???>ZzCOHJ*_D!S)g0Vc$dHthC z&NYy zOTH~_LbN#A*^9Cp9FaWH^3(A;PyQ`eCewj0)c-%k}Y zoaIymG4_GC(@b#Wr^6(G2uqM+cocN+R7eBSo=3X}3W>dELgL=Dt1{Rv1m|e5@Lt>} z<83HmDt&ouDtHGbW#9|YWSI^$=61dqY>#y)LTj+OlZV4KL1D@d<@O44v>@L{u-q

    C~oe>=1b_Gmm6Qj8#oB7@sTvljPxtae!!9%NPXgE?2m%H4ci=cvnKS7Ul;DaG7% z#rzZnmr~2f*nsa_PnCS^k*{*FzXEmez-lj@PBZ=n6gt|b6Z|9vd`0x5`RFOWzy_=zJjnc5~r=KLP-xVnS z4F#ck=pIzozlKuA>*~&gbtfNPHZ_EN#EgM|wtfJn(T2Hda?&L=n`WY*y}r`9Nc+b! zDyd37sM$mg`nx-G!h1ZkrO#_ozG1yRfjKFvH|&2*_p2zl%3Q#OuvNPM`ob;t6$hAn z#i3An^lTUlC}N1))#Rk!jOL?3y|nU503^Vz@jc&)GgU=s?BtoS*pJ_+^_cgiuNL0j zM%;{NLO^>I7^Mn?o91@)D5b?wls0wH?T$qs0xUuY70<9vlKRsKJ+F)6rHM!sQ<9Cp zF<&#DD%Hm`UW}8O`86_?WS4MJ6{-0+IC(BoYfv&LNSFjbQfRsNWjh+5CLeTbifco% zwQhVm6XLvo(KAH&`-v!0(*q76J8OfsH}D8z|I^n$1xLnHOmUL&LF*>J^>OGhBG;Ny zjW`0d{Uhe(3c^xP6_o>|dieF!9JhfsZt@73ZzINsuuXx6QsBh#GU>}>jKut6+8lSC zvHQvxGiyULEnafH6=!u@huJ|LpX6eV=~=eLC1)vPzk{mI^bkx;$o%nW#(<1LON2bT zP5Q4n-JxF>PMwdFA3DK_!|x_f7Jdy@ek&A*E|-_woPR?T17vK+CCrRLbL3W?f@&LM zyUSVt*Th-NojbfO7xWM06_-O4+(_>&aN(cITK=>lLD!(QjrC-)Ull(;>z|u<^SVsB zDNNa71?1sA4mMm;tqbR6pI7q$pH#Pgv1a0R{)<3*)u|SzjjffWgE2p=U2?u`b|N0n z8XaP_+mvLkS?ng4Js2a&&Q5A|{<`z?^GQ+dOQqTqai$vn{Q(!yY|a@nsu=UF!pDD4tk^Rv+Tu^gCo4|F0yjVO~@%>2=Vn3gqL zZ?os*Q8T?8b+z;5`X%`yuWl!GUkhQaws)J2eDpa{aoRvJnOy~Z3E`%kF(C*slsPhV#phbUPu;? z>5KOA=2v1GKgTBwc(Y@>f`aQ`pjiI=?cRow$D2{-6&4(MSle^ysehoa;?^UQ8L(XM z@s|bRZmPiHvFpilnIXd~h^+s(@=i-bv4RkU`Q+FFI5k+gT*!MRy9SRjAWT-4Z|T(> zSMi9Lr2M@ix=PyIjQ|8`^S{Hwr0(7k(`Tg}KxDn5tB!%bn9r(jle;F`+J%O++IKPsr)5Ai z91`Ig3UcKAlKx!rPnWWm3+X;Zwx9c47Nz;1+#$%_laK-^s)K~kVd0E>L|2#VD!h)Y zS^E|I-d)b}IN9%EgDP8tx0(E4G`UOKiBy6~mXfBXjkLH|Oz##W8Uz*Vz3f?$AOORz z715j!?n;G=+wz|C;UD|4+jAPRCY;Q-giw%$*{5!5L?iK9e?42Sy&TtnBa-g*@#O=y z$G<|Xev`o;w(+~+TU}vq?7a*yTiyYLTx7VV_A`;MMk!-R<2ZD#M8uM@=s(aUT7cf` zDV#iwn3mk*2x#KRhQt~r38*skD9BmV5A7JKx8J|C-{ZO}2qzDh-ap=>WRIaa-JRse zjb`v)9Xa6?C<&v%m*uG>?CBv6uxr|p5>%Cp%%^#&a{x0L>lDMPj5sZYllaeWU-Y7P zKELSPM+?yS*k}$QvuEHiKcBOC%;4&AC z6KW{CU;2uv1UZZvSvQE~@f21a{rh~9pECTxy)H^MLz#(umSk=WdK?b|S1R5UMp#(eChz!-Gpv=<1ya)JBf0R!Syp_vqU9k4CBT%-s8#L9v(B zrT#1$mv}NzPU>&|wHEgvB6IdXKO_2cx1N}f8c~}WAWF^nO^VqE_w^x@Q*YjDz@=m0 z3L;}KEuC}!;E?TJWlG#`{cnyZWc?}!)_IQ3BsjjAfgF5vzWPFU!Ko-fA-mH zuk~Hg+UrTV{$h<$)DsFI+GFEB$;^(YcH|MHwQk|QPwC|XW8m>+Ey>$sW4g#QBC^?Vv{TEw}r3_Z79B*T|CuTk04*PW&|$^8r*>)qfY; zyO7B_&&RXht;B_Ga-xWddHNv?ukxf8&Dj|OlT2Fj$C{x{=auAe-Wi1u%s;iQ)yzJm zG^hLW!;OZ zj^~$ZB;j$1PWFC!8wchX;p)vT87_X~6xFBr` zx(c#7cKmu?n0;8fQNt@?OpQUYFlG$EMj6LdaGNyUPNVOK>!$EakM_Y$<`jUsC~;WM z1!^Siq?psEwX_lV58Sft_wLbgGM7bc z2O7}@TXFeFQ9r2(0Eyap5z}L3d{}vn0YVw+aA48$eeN#aLUqRJG+%QdILCjdh^wNS z+c;+@abn@}z`}A(`~fuAj0M6Wzo1e!O&*H_AlNs6K9+K~e_5!tj41Wgocz!vLp<^d-TOVv$tz!jQ^@!Y|3& zW}AxNutAI&+0d7_K4u0Vz~QUSi@Pq4E$!Kj>71m^Q1Q?4M-pKB`7n=f^cIDwmp5^K zsR!1XtH_GDeJ4*tWxo3(^9WDzfqg#D{N>&(G|pw22UOP?0O$q-O*V`!2WvLBhi=B} zn#oNQ(0qf7-+!bTFZz+eF2rJB&vI7$G}a0KzBu1ao3JN?j>3qjX=B)UNmO~umweus znJpKuGR<(VZ2uWz9*b1Wsr4t;cYcVC8W8CZeIBzZ%cN89oMT2I;v7ROKUTdG+@Xk| z{@~JAa27*@-I?$5uC*^XEugMgPw2zT8HVUP5v<2K&{>tnM%7L_1%I$?V2j>f&Yq9h z2d|@sMj!$RkRPbjsM;4!A6Avc;<_<4>s70A36oLYO|@mh8^*iYK1DID*%w_v9(eme zFQ+uz*G8ces(#25wp&xqVxJ`rH0&^Eu$yZKJ%i4XtJitq$9gID;<&MBCfxF@kt3JS zC)(aVEz0s&+Sb?;jWmsY5003;ZoMzEpwKJ<;hmr;9ul>GIloFY^a9GLO8&Hra(kU@xi2^GjEkZ;iBNg|@C^?}XV{^SXYm<{Zgezx6Fg59HEwLWpHfC1V z`7%*;*}F9eG%Y5a+8Qm=J?h%ei7|*=j9PzS!%2@Bll7!_`v9dVg2;xVOKcOe^kKP| zVtLPgFN4vkargW8TG3K>F>{Tt5`H8CiNb$eR$Y#y5Xc)6yQ;3|#!G^sW`|F(2{DD! zs;T1qF64vC5!2m+dsZ#K!}~x)YTBj@O;*U?at;$D7wsO=!W5yVHab!SC~<-ftyF~W z+790!W`a)tWVQ(jB@&G9Ni#rxs}r*Jk!!vMkt!`{U3uLQ?VHv441$0;L2Vc>!x1Lf zgNFBGDtrbIF&Eq*kihi6%4Ogf?d2mVrhOg`DZE?ulO~BXu%C zO!htlpkHgQo#5b5-S`nuqU@izy3-ouTuc7G<4k<*&bl|Hvl4rI{Y+9O8PwWlK+E8A1hI_Qe z6Fkl_pNb<0YEP$d0OEb)Q;q#|Ti$dI6)V=wuhjnh!^1|c!DL5*Ag5-$(8W&ro}{P-Kpo^e4S=Q26fv|L;7%yU`WzH+P-e!4W3v4y6E z6eS`4?4%bAV#>9Ka=vDC6J3mNn-FdxFewZ_zaXdS#R@_8t$^E^_sRMIX+5@_c4|Eu z&hi3UaPn}SRT>Bg->H>OyL?{$>td%^Ny}+^wTOln0A$=L0Z`CXCi^z;87-Xc*SySZ z1$+1G61!^qqH%!Bb8k9>o(oHx5a^!z5nNCejtdO)7wf)Tt&JiDJ73Cre1ZCY-U6uJTY)=Ld&5XbMgTL%j!UiW_>qbyu0 zVHm2h+f^MQEF)iJ`2t1LZ{g)>?oorsmU65&V>-R8$Wf6VxP`<3A_}+YnqJ{5#WV$> z<7`Ch&)$GJDk1y!8xS4=d^E?_tH-AkuL=l|mf5LRxsk>p&KTnU+1AlfV6j?b!9@vq zUO@$Nt`r|~)84NI4olc9%t}cq&Y*c42`A5wI(yi zK-Kz-UzfFhC&Ci;$r$$UzUhlI43;>fa!ktyyEV3c$Y6Q=&!D{sl7_PWm+Gc&*k<)N zmVscl-08+M*K|Z|O(Xkz?C^G4oS9gmd*%oajvHvSx~z`XTP(;R+gh{otM}H!Gc2sa z%MqT4*B4*32oGNfBxQ{$#UU+GD?%(FK%A9f=9ugTeQd* z%(%b~E5^XK+%U5ew-&sqRWW>H;fJF?CANVtC&~iV{eAV{DFt`049Z>CMj8CeP0~_x z{SjKwAH=k0@H>I+M(9kU#8O0uQdPZ0#Fg>7R(&?ZVwP7Zb!)UBMBr{NIN^~1bWz#e zFzobJH`LjWZ{`)5hw!)dcfY@sN;y`$Bk-ru_Rd61=|AN$?s~cvN9o+DLhod`#wo>s z5To?uPbE@!%3-jd)TT<2dy?{R1Eqc@-U|Fq;Xb2MUGEM6l|WBchb~XAQky3D<*DyVqTh~dJZOPR8&-f)FqvQ z$2d%AoZX{q-1`n%`Q9>&>Z`{9fxud*X&SX&cBHdBy5!Y}%HXuTU=L`j=};ffP7i7^ z*-&Wr`JYIVW78p+f{iYeM!TEn0r6L)TzA>DDy=-KVI8>gWRu@{)H1G+*NtZ0>U+3nw*` zADn+K{V*5#KkLQGg%lIf!e^Aw^hNfMXp64L?Vp5-O^De>Ulgd_17TB@vdWKS7^1x0x>puW!Cq#05lkV}U z@#BJ5{uc+C#N)$n0HzXOIZDUH9=2Q-+S>ie&K}6Ixl>nycl`B}zrmPoXeQ()D>Iq$ zS9fZvZOF{~`A+r^cb{(c^Mew-#wfjH)kP5~D+Fkhg#y?^Ya=eiWIS> z&dy@7F9cYPaN+;~xLLd3hLt@t-FX)t>2)d#DDg}>Jz*v;>%30;pF3tfW4*mIljGYE zD{7za%?&fnFc4~n!Dqjab?@cuI_RCV9Py1x(GJ($PT>!b@O+;iaY0~Svll%)H+Y;? zsV2V^N?)pZ`9VeC*@J;tgFM~f<9ipk@*^-G2>(p{RPzRc_~`p z0M4sW% z41*_44tluX&DtufI+u&tMxkiWe$({ujqfD97VnH z!N61Fl(T$H3!#vT{7cOFG8f~l9AU#01N`6YHN8#-+8G8j>>cMKx)~t{nm@``E_(_; zJJ({}#*IdF$OR7g>KDeIWkCk_3;{oaMqft9L%T;1ASwq9-Vip1MvK8IZbJflXv};K zr>~kdlg~_5tk7DC=vKs(uT+vLg+#)b6ttt0_zK@B8nQJJ7{3Lb-&0rELxc=fXIJR@ zs7f(;6ff5U9-Biq8q|?YQ{{rasN~Wx<2WkGZU7Nub0FRr1iL~N8~(&TUwiItopl<$ z{9EH7j=wfITlTI8m|B}gYz067N{kHPkgzvynGa>(?g*nhckd*X`#SUfpPnH}=q%T- z`mIN?P-CMsbvAgjlEGOEzP#>imvZKjVe0!lnpZXJmuT6_veowF{SYzVb?~Xku|;3m zH)nt{5&41k_}#+c>3NByz!?4&B{&nu7proYl@m%f#3kES4y0yBJkH`A9r~_!0UoLR z=s6;4$23z1N1?+-8JKHyPLhunV38DpJl2kB+Rk?q9E-SPmPDj!~8r|!8J z<etGMKa~5DA{k5<91lI!i;5>3^OV4kA*x3e@qmVg&kUNU!1o#d~#F* zp2keIZzmYruPwf+jH}}KgL^pny~7=B|FpaO8$Je`qMQ~Qmwv1yM~{O(uVXay3vw-? zDX(HxYTuWCtwkU)X^K^$V{hm!$NB>7sJwK957CP0;*$eRr z^~K+_Xh=$Lo1xcIXq???U>3ef=){uc-TFR*$wCF=+8xDu(hlI$vf-7yPpqTGIT!V- z88YE$9-+t^&CdWw@O(HUF^3(hqhnf6P$j&zq`LBg9w|Z%Nlwgh+6E~KfTyO|Su&Mp zNW7e(=1yr}9&3Kqsx+xvlR>j5NZ`yvVJji+p|R0L-G zsV~-Y_ZuVo>yx|F@xiS=>9T5S@5DyQcz}Sd#+7h&bl=;IsIvF>nGUCzR4&#(yuaZ2 z`UOjlu`hH&?B2k*>BO@Yt={xMy3Yk>1VPbwI@Z(8cWw6Me1f37&Yt6I(@eW2vzQ@G zlJurA#Eu176SiJ1*gjy?ZLj5%t$A8tFC&Q*-C#?6Q1B7ji|yPuG~eT|XgOtf(pu29 zV(lxpyNrQM#uMIh((_Rtg?jTv*}k#*Ub-XKQPYmjrnhJ8Llj&BZ5wPcGwMOQf7U?2i6z$Gi z`qV!6VP|Ex#lDDKP9A7|F2LGldS~s{O9Ke7Y|+R_XI{tPevo?ZC2MipaW`{f|BmQM zp8iBL1Z*rFP>LwhFDuUT%Gt{yW)4_Od_kO@wu^rplNrXHEOL_H@HZ3C1)DFfFrkvL z>Gz344z(&Ks^Qc+5QYzRw&&N5z9`_%E}pb$-q_nJ_o#!K`(2DAXGOp-PgPQ%MYPWI z0-?-vVBiz~{?mCMtx@AEdBVxL3xg83XHR_?ot_Wg9C=XXtWpZwXr}ZP@oU*X($>kI zaQ?hqIa)Z)!btzOcR3dgxj%ywhjqfr&(4UP-KN~-Lq7MVr)jX~BElvI8Ji~OGQyn1 zZs{6K)do6k`hht{>Qa+?1DJHDEFbI_jU4|sAQNUVvY9KyOql)$h*J%>azK1ZJ{23n zL}dc<3n$+^%0%d?pU`2OCFLg0{(M8j43detul$sdzO9|>$7gs?*3uEx@xXW1k?%eW z{*?|N9Kp!39%i1 z{o_nHBOxPkkoY?Q){WVd4fz|wpCEdjAn|sZN0WF7lz4AAM+=SZ4_sZLl1I)e}h2ejB1a@=Tx^otsK z3e;WTaop9Y3-T(Sk~G}RaAE6HpH#8(81JnT1xT@y zl(yH~5!5h@Wntrk@LFroqn`*SOo>3{gQK37;~!jYuj`}>Cw_CmIF^%whl*$QJK>-b zQGja-h6Qc<8vBbuYGy^fM)voY9IU;7GP&l;r>m8J)cE7Bd~+GEYbYBslJkDu5f1fV zKHshM6&6v7wIUQTMJ%gEI^@^z`+2WBQGvhj%e~>Fd=oK!W8%46e8(T>zBkt#-g7W& z(v?n~7QuUCNWPupM0qmf5KrQ{Qu7}GwH$#PtP=x5rl$2C`_~0DV}hJ6D6r|@VJn=j zz;8ZL^EqqjM|3^yOkHAmFC~fK3`njLOLA^7;ahyvw&K6{AIq>fI1@J@{dw_?VQ^ zwJS$d(m(SV-X!nB=`QxroisUBiPW}03_!095P)vR%=OTL&;du+8+DHN>#(_!?jH6j zsmFiY@9H^o3D|qdl)%B7PCUl@QbO~Jc#^Movfhl7}S*l@yTxz@L8NlR>UlLWz+mHa}{pe-rmdNP#@fzI%uyc4(#d&!VmyZG)TTZsG zxwCU}uh_29!bW*B9kLOMK#Hk``*{76Ho0;nTX*7W-?KQYJ0>7bJ+Ju=9;KaPXeIPi zWJ#TXI5b^yHo5?opHCtscUqiYFRN;J7n?#|+N#KC!0T0|vW@)`gDFIx436Q=-tT$a zxRR9>#&ybQP+~#PcmY`WOY3Tsz}`RKueDJO*RXgz?d*;S`Xhz4KiAhG81kM1^mw3=jXh^gvi z$#TBgjNwZybI3^gv!%(@K6Wj7~ z6tOgPOtM79UVTwVs134Hn3UoqLqTl*{!Lstk*r6dS87b6FS|d+>6~Tg(hZc*nRR3t z6AUg^m$}B-E#}@xwx(xdLAgiUs-EwU7VPn^sY5u;M2fXYlzb#A!I`dGLAl@<&R|AzG&<#)vkt zLL4h~(;M@|czoaPbypwkm{E6yFRF({%qT=NMQVxUzEz9rA|5Hjf#nn{B?7VV;E_v~ zTU9SvV--AtP`KyI@;Xec=UZg2ir3d*vK;*MYXm0$xe{}pq4lh4X~AHW(oKt>VEf3B zNg4n0v!XFJ$U@6gy~&!}q1MNu0XqW^4o&J_ye^3O8tpR39^wO$PZsmpztphCgyEq{ zlTNR*fn+44IE0g0Rgi6xgkRH59i63TrYYXmT^Tj`zprkM&MzM)XItA)GSH=5Q$&}z zgO#Y4^m_f}dBr`XEMZmR&zT6eUYU$(qC%mw3Y0d&cKzG0cZ+x;$wr1p{C!#UAmFl5 zt(8xipPIs!>sqIdr;iW#wfu7^(#Xz0cQRInFrS`!Y_=>#NV#z$s5Wix4~xNDz%9wl zl9mG@*1cUpRb{@B^{fZX&7Ee!uSUBFJZ(nUUMiC<&$W7SUJFs7{`~0~HuE1gwlqBl zYsX-;qSVLbXM(E76zGE0H9gDulgx(vCpPM@iK_}$-vyDZd}27WeO6+TP?B!Y^{_pF z+Af8TSB07B(ql=T>hkUmv^=LwHw?|g>#QBGvH9deZdJ%CnO?y7SlSSk2U#h873gTu z4mr?1v6LO?9 zZe(D@mx-+(ekUKFb7O7mu}RiReMlSjl^my94zB-Tsje}8Obiv=Y6M`)&s}cyZxB8l zVY{>ca$+j9gYV)ythC=S@F2kdMmv*CuioaBW&3sK#@MjDh1^h*I#^5ocxBlUK$c#~uTpraGxxA&R1{P8bA&!(Mf;)sF~_ z_?5C)pN3I*&V*C;-r*WUkuoq_3*N_Vvy;7?3sT^K?)&!7uHS0I4X<>#kNhXS@Arcj zHoD8nPL2)x{KjOf#-xY_t^8|+BQI>*BC=2*5ChcMUWqgP;aIad1bZBt4WhUlXXNln zbZ-IKd>V0}7Rx6-#*l!0S~PQJW|DZmr-PqTt16UJ!_FjbY*OMP6Q^th)qd($vfZp4 zLVaL{6Idhgwcv-rr7Pns3AGB3NAN>AGMSZj_u^xbVru&k7>x|(sbo^WEl{~n2jr|&v zLJBM|`N9TfED|k@){n|#|4|!?k#G(Uv)3tya0l%}b77iYL13c5Qq`r;-@sdNRq2_c zqgaT%=9o$!eb#Oip;DfPOq3s0dIy^P)rQ!+z1pgNf3y3@lT6*~2H7%kn|Q}#3o zEom3(DOC$i^W=0(G=c{)(v`O;Lhe1bt|78&Wf;9UK!yo@i&>M6*TK#KRTHii_PmLh zrt#aSNR-(I8&LA%suI3+{=80K1aWVCa>VR9%?B@S1QiosSohO;tRt zhWjl|wHsJh<3h_3I7;y|AtU8kpaR3tp?T}#vhT-~=U3yahZc+R6&%y#v`f^%z|L2A zOA8NtiM0avg3>$=S!bLG_oIa$W4|Lr`jx2ExR|u3TU~TPqN`Qb0xd^l#c7kI`0vV? zblpysNdV)8<-=QOoJh*@u`;X%cOu_}3Kf@4GWMDtidJrBJ2 z!{rdb-P8?pi{xgBH86?L*HF$OlkF$3r2y3`Q3c=JR}}!+&81jW<~k~0GkJ(Y9{6xv z;>4Z9`d-P19Yys7K#JGQB}gsQhYxV#>U15!GBiKGthU6KBamx~i<`@>nX!kdE+-j( zQ+eP#R}PGQdBzHBPPwXWxYD8vNqV_al;ZQeCO3>p+4^TD;XtFE{Bn$6k*xW7?WbRs zX*NUR*5I1atmP;_NpEqs_lOH~g4m*3Z>0&5gb-ij`U00b>j$-8@yp~`Al+9LGuyz*I9%=8ZE8x^5t&X5~y z8X#Oip7_xm9h1fz@`+@Dt1c3BpzlWy%=ent43hlwiY@(blx7!UJTL4palC~qN`^^u zs`aSDPu9cP9Xsj;rp+73nQ8>X!#R|tE;fC_IEThGzW|u@HBeOPul7ja0{qmtN_iie zoMVuTpDQO+&cMJUs3mp9cb(`+NOke(*J|6AcVQok*rsb<|M_#MZ|;0i*+Rx!J{&FU zsmVMg`Q$1TYCRPsFrswMqKINoe5d~&K@F_fGh<6qjfQEk#vyAg)g;8r_;Axj4hO7c z=4MP2QmMThRCVkU`?X>Ll)U2$8U#88t~N&)=c{CRE5+~(>sLp7t*8LsVWc1_W{ zoKJj;QuW)D*)Lpo_KmV+L~FXi2pu;?UV-5ywRq?}DCW@OHvX%?^3syGXS}3W6zNop zsDQ}dz>OW8vHNz}sURXDe=O7H{ApU!b4ki9sQ$usqduDXK4MWolG@R%QA`Z2al*Tn;H4Wl&vxk*4U^ zLR{zMoC-q|_HC9m1BGG+y|WTYWT($7$@BBV&YNOccZJQ*Z^6E}-0KnUxZI*cj_^*; zf`4jqLf-6%%EZy4@MS8Ax1ko~4?Z5CokhF2w4C?x4 zCix>%jb&Yq%;)*-8V&ntfMgDb0ha8w!^um%t&HMu;x7&U6aaf{}&7rSj8oqP0N0V06i8%aZ zK~7VJPMiO%|1cD5_+hW#)$&WOmofX-qcBbTq?nv6Tu1a&Rj{Tq?5hfU8H{R(vxS3@ z)e%s~YbniOChVeOcB3YyXYpCZ-n<3>e6w&hZ{t&zw5d`j)S0$ge@_ z1*u15IF`%4BuH!`MVf9y6k7=#SX;V!I8PYDcwRZJu}hv>2oK%NWdeb6)VDNL%^1%+ z_~^|$mTgAn3Od;Sj<3NpdGM-MdBlIqmWPo@BQ!+wPwcw=K%;Oevv!O8bK||KAJ1X` z0jU0B1+(0mwf~7ienpi^7GEi)vgZeWdm%rMfLfhKEOvjV&7CuR&Q^$~a+1s-j1{CL zj6CUp6i=!*wkA4g8rv_3@#uTd@dz^uY3Mp=voSUkeck!^AAp5c{{LM;U*8bO$)o?b zjUM+#5uxf%5cuNXfXt}168?7p+`R)t)5ISWwv;GMFvUFzSk@WQC_@qZvK+aC=86rj zz!1>f4s;eJuwo&H3V{My#K}+#^(nT5U+eZ{B*MX^TgiKgMhNIMV-sAG-cL5T+M<%K z&u36RS=2$n2@wXH%V31UWufyI#aoyLXr{bZBizgDh|s&sO7>)1*~5Py-+t#&=XgED4}8 zXyRL8@%2N7wo%PU{pt{8=uzpIW*j^Ch%wiMK<}79y6s5C1~x{YuV_x!@>QT$`3d%l z#^a%k>+bwlv6^X@8aZ0)Y6ibGMz+k`AoS6Lm`<2Gq-wUYsOCdmKb<`yx zI!B>1y8{e(DSFaxB?#pc5kIYfaMBW)0U&4D)~gGM6~agrU~)JfuZj;W4q5&crCy)Q zsEU7etf$J@zEkFy>{C;(h*Z>1=(&4`bBpQ!dMC(#W02xLS&AMy?!lSDPefj|AIE7X zQf}*Q=#Qv`eP)nz8521(ps%i`T(3X!R4F=RL&qcaEPrQ<3rap2-?k7nNq8=xsXofY z)>gUx4I)|7!LE_}erP~Cz@sqA&Av_A_^nK|%Bu#<3m4tJ?szu5hcy*QGrou1_T)P>IO8ZSA_%V1-|xFOAZr=6r`q?RW$Blge~ z$xDylT}h}?y>B(?PnC3DsDA|t|$<9Awq54lGi&^kdcfB=I!8&QGz@wpEPfP!8F6Sg(O&%YcHkX#B zS2#RufSi-p`XVC#b$NLZPYs@e0{MYMuMrp*X?jr977G@ZCehmD2A)#;LqXe$M zh1uz4IXvWA?}X4)v|Dm7H?;;IvcA=wW9Aum48gd$KF?4L^H38JtZlF4xBbL4tPKu` zzDUosExzC?adWd2@u0{qHT(uVnETG_SL6KBX5=T`RF0~|rze^6Ft*M>{6a-8?4tLk zG{2(D6NQkPo{;ouw_RCGNLwz@l7OM~GI@F~cS^<8=gQ2GAFA5Ukqj{Upz<2HY^~1r zlJ=EPv{_2<$bj&{ynv^7#`MH}0n^V2odElP>3w5EZg6W;UQxCki2=svU-~3eSBa4J z@3ssuNp+ZYUF2GHF+WTT@dIY(q?*LtqYC;xi@|U2r2Gie58l(5@h{mI$yxYqy)|_L zZ$jiI8jOSTHH$I-FO;sp#q>JSE=xzsR~Z3&IntAlFm2_m@1(%oYospfdi+#uoTPc; zn(0xELAq=WbD!mue^`u6S|hj*zD2ZeVtt-#W&00c^6W0Q{)ggYx#^7E{#OjoIEwfF z-`%tNp6eP*b)t~%5@s_nOPFT`i|ev6D)I1Mi8Py99^G|Ot7b5Mxji9Ue9yPoqjARv z&Mc}KH=Q1O9IwatRvl~4IV;SgkRRXO#ny(-r^>(2<8$~pXjh|IOtz-^PM-UAreflO zSN~5!U&zOc%7)t4Y;v(R61Q`=Gr~Z070GAJIF-@sk2)2!VY(a_gBMVc8itw9e|RSK zFv$>f*6*mg&w+$69DJ8%gIi@`(W~zzs>zZNfc4wPAxKN;UYiq>rkDB^6K;w>cWA`o zLYw6K1JWMM5^k@AF&i(mLrekl@6AxsPW6MfXF@&4#xGE>-YbC0oGhO$%aOnkKQm2F zJ`LjJrrs-cRGbT^K2I!BlQG>-^CPwFn}q-C#X(QCQaY_MR2VO_)wBDFdg060xGM(| z-rt+*j~#ucVt0<~cbpy8URX^MhLpcxh#XtKtSr35z_xgsx61w=kJFSxCyK{Ke)5Kg zy>4ziBbhx4v^&}FQnbZB(t|Ve2d^F5-Z@|bnIF!na<5bF=ll8IVX$(r}Xtmb6C2?h1z#G|tIIMH-#+slBn) zd&woE2BLs+b)@Dm^~q=cP|VSrS`x!*{Xdhj+Wi}*$9AS2?%KZ+|Hd#-1KM@%zOUP= zAXM^jj#=$8-RCnVU~-Kz^Gy#gB^*@~Q6aPpi-~DjloRHbZ8do)p3J){;k-(!coVdJx$RAsohT@sk2yl=zL1 zOZ}D{l>6Dt^1)YHp0~0iGnodm4Kov1k?9A?zu!!mxCxH%k4wH(NX&#GShkIIA{NA3JZGGD`;ZaS6DIuR5PnMZfX9GLC-;udm(d7)QKm+ z?n{N@TqETt!^HcKsmQ5LrXGwX?C=rEr)~>wrLa%^4=`5;8PGsU?i1Tg9c|VBNc3k4 zh*vm1v@k9kt2W^|9lQc!uk=xy?})H|S~dr!MUYa(RVs>7+`uZ-9uNGW*QURpTG^Q^ z!^mPNOwQIK4ptwvxzri&WS>509-H z_s%SqMGb4gV1)Z-w|4gtfHBv&!b-XD*FIWw4{nuN4qB55@3Pz#vw;W{Qb->PbT9hq z*lSACTMJ%L%@c}bgwzy{NTnvV`>3a7*%#ICG*?S#@OqDHyOuZ(5cn2g-0xPm;Q}Bm zUu1f!`IxwP=~!j28x}s}n&ocI#>0ZiSlF3$=ZEz|ki6!>`8usK9?fI(kXS{BH|q=; zrG8=Xtw|w1I42b4R>3Bnzf|~D$O=65W%+kl)7wGq#xN6NEn_(TkAqJ@AYtd|tAp<= zEpTv(t)cE)K|kajA{$_cDNB;MC6#1Tm(DZtHo0e_orFNn#78 z*d#+1(#+qkM+T_g8akVwHLKAgZ_aC$w4$Cbax0~b0VElg>h^4lG=&-)q+N8f9w?e!%cUzZ)*97&@SK54{wy(T$v|5mve)1T-@Wg+qus& z>?a}(3+TR0Cshe28Fv(q*wEtC#OJ@{ZNTgCirSaHq;uJbvLwU!6+5SQg@oi=@6Svs z$Ew?lC7lXNhDJrfj2TB=dYa4qAK)S?oU$swx=x`dF>vA+Xf?DNWFJ>Qn$f*iwxVcH z_wb@piPCTic<+z%axyyZ%Hl+y6%A!`b~uSa0*|Yd*}A#5CsPJE7y})U{H!JNoZ` znVmzrlA=|z1O=p_$E|i{Lfc&dq3ZS$gCMWq|8AttevzC;^j6}kbm5Q9$bB}t?KQhD zo3g|K!$H!Uzw&udOT=XDa`;-m?bwrbh^J5JSU%>fPt9~h*nVK_kNQH-q!U})8hToG zB2wt3cF%TTh`hJcj0<}5cww%qcxzlU@Eg&$$5xYEzei>StI(ySsD&&pg31^!C|Wc0PZjh=DNHf4DrSL#&bCMa0a`W7WdI|S=! zmLq>p5`4}Mv&pJQhJ9iKcGZSWvMj?UTydOg2(_HjldTOkrhzAt1#XvJI1=~jj@!jT-|#iMPPGA6|8`He{=q@kS=DQGsCIiF&U zCR-t!^Y1i+%>@o|Pt zVr#CWsc`==Dr3lrnm_DQ^;e!=1|s!&1Z67iD^Q-1ZELmMvXFlUGc zbEWEJIM1Gw6k(dtz;XRg=397vlR%0u={OVh{{!$bdc3kMC?){SH3i!fJF&ZEJ896& zVCHB34tJTm`CZumODm`P^X{6&@_(b$LC?A+3Ovm&fTX!iw5i5UQMla6L8YigmdSel zMIaMQ!1W`Mt|bIl$uyV6gZ#RIm#7-9}+=j5(a~E zO;Z}6SI=RJOD>&b$x$yN6fGEad-1DBaq$ zs$A@08sf#gq1NmZ9DAEX8fB@R1%M^$>q^}j znZIQNijD4i@YolmL3KzZpreG#TL@z55lnk8Uv#{J6p4Gs9aJ$1KJX=0mdDzu7k+-i zhBdHpJtJ);ucwKH-WmFes(0N^u|JK>q*qXsrk9(^?kU6L8)PT-c#iXJG^QD;n<))Ks(W7#~qgQat~209UzKI-MnWyFVqz)I6FbKi!0)l#6>*D)6PK*Hi4)VDY$1fd?U`~wuSae zW$7F0HpbS#8+;qE1*_w&nIXYHuTyOO0pAhL5l$*Yd#917g<|S`s-GYEPkm|5o|A2| ztSNAuWbb}WZ-#R+)$r=A7fVh^YRfO-{bxDTlSiw}rm7inXL0}^KcPpMX>+<0>^1w- z!PW3osvgvF7~znT%<#b@&C;T#P@rq-erWFZ%#G|AV=%94G;cw8;6SKrPS(zOGVZ|@TQ56BwtO~=TiczkLBKr?)*xgB{ThPNrbz~V;zB`$D z_Yz`Z)Lpf@2r{553#sFxihfhO8G7wy{55mb6E{lmB>{c1<=+~?)JpiWnMCnAPv$g? z5v%6fhs;k^;TXY(KG@)TR=D5$lvco^k_I3&Gxs{7`MOD#XIW5!0=xbNQ`nxlCGsOn z!>iV&BrSO*J;JV!zEW8w0p%;%wKjPjzzM>G?(F~$F{$q_cj^*jP2W@X_g#I`;Xkr9 zh4917HkZ7gae-(hE5)Pnmit^~xGQ{Crn!F$s)?Ns^ErnM>v;4x5(ia$1!r=x!`o^H z2JqUuPE-B*VdHiG0m=!{KV*IAEM9q`87&`WqqOJ43Ky$=&fYwrNCbuA)^5$HIm-n8 z6jO~R%qUZ^qF5AYD!SqLh>!)=a?GCsp$U1qqw4Vo!Q%p+7u5 z%nz7up3PqlT<`O;`Zk3Gi|9xE2))OAwrE_?ezvT&P+id+or*IPg zCHWZh_=*1USYFi|1KE_-3vxP>n99xxxjQ^BCA>eVxKw+wnIhzf@uOrl_j%BFwH)2$ z-4mHXH*>LEa-(Xg`aE18xy-oSyVD`)oJ(JM?DkPB)zE=l`3e5UhaNL<>%gqN;{)Z~)M$}o&5lGg*XvGUU3G{kG2`^Q zws;)*Wi8y6CEN_ME`I}1N!HDr9M;R2-88u#!|*{u-%EettgtH=*9-(E57c^?KaXtI z#fT{du_{4RyH2bTptoM|i0In4y1}EcTRd9x7tCf#ML?R0suL1vJQZ#RU3pDiv?!ye z@O$|;Bp!VE*O#wLWYx#na$M%`!&=mwMu|u{J7`XFX~Ef8JDvXlbTESyqj3A1#n?eN z?_OWM6Fbgv7)^}&&PVv=>j?1`G3@H*?E1h47}0Be@gG2hk&oK=<@-cb>ai(3@wi82 z9Coo9p>I^CTTD93&BmXs*Syk#FA?dckvZ~(bp2=)B zeUhoL=m+uh_?1YEOoU0%2&pI+4s{tg#61@O?;&905t!3D5Nu`7jbgHV|9)R8bhC8J zNg(%vivF9S88hZn_SY7fWaM?kY)=;$=5wzx##8h7?IhXKpc_Nt`d@5~k_0*cn~EQD zYRe5ZJQ_qYgcNtLOn#Gkg~4dLu09!n9%}Y&t&<}?v>jS8wa^8zCh!mpIq+-dm|x#5 z-FQm$4KqpxW@!eF#Zy3X(Nv@F)`T$2Dg!NqhAGoh;t6htZXa~vV$|egf&%_l&fv_k z%*`A1g^yvzW8vk{kAH0U_qEy=Eod9ULQP;pTZtmEP<3-65&+V`+Q)2W$-y|NXMmSe z23_!IyovGU^I(0)@V#QbpPh@YFX<|b*YcED1r-#d5%*JVPmhMtqMre-%`cf4zE4l- z3hJR0kY1QlucXfb{_|+@ui$tmpEd?wnSo6B00=8Hoe-9Kd_7lEb#qpZLyQsb#fBDE zxe3zWUa)?D;=cWihh6edQZ%vlN5_$%xel)7=x9Bz`f_^?_GLVfZZ6%d`^U=}Cf_9A z;87Ey2CW8bY!X3LPw`Ph9&5{3?W}#ZtF($Etjl@`V-eL)eVP2{W(<=3$}s0v5U!He zpANbYaqsiJ@l$PrS-!_vSvp2!_G#sF3-XrXJhcc+8NoT_L&Z!_ADI2EWq1xZ3zEbu zRitt%Z|Iyxm^QQbiYt8_Z$fDDl?A4+ z^s?cN32v3&DqwO$ak*0^LV$3u|K-X4#>0_=qe0BL+)9<^7CJWJ!*R03DqZeW$9&(&N+S*?8n!=yQkgY$8 zeG;wy!{1lGuP6304L0*ATKMjb%_4mjSE0+!>QeL;A}q&zvRIOtF)Y!gnqsWSfTn&I zH-L^h>1=3wAUoxsH9WIKV}P55dSR65W!kZ=MD z@(&I+v(-GVF5g`V6D6c26b~$nswid^-nH@nF!kQiY&UM)e~1ch zqgq-tL`BV3tyHL4tvy3gwJKVxwJEVjQH|Ig5kZYmtBRs%X%X65F^gJ>5o&MB{e154 z^PKbh{VC3IPU7UcuFrM7-|v?(s34<*zXKdQx>3;h?(FOV5Tu=I@o5w9_EJznF_3Qf z2S9^mNv!nnbugEL8lk`{*FeiaW$?UUhu49U#>PuJ=@btGI$-{3R@2O!#;nYs-umTJ zokrg;Yuh`I#vG~IL;LZ)3qnKvNK)?S>9r)Ms$SyaFDUtiZ1u16ee0fa!}=;6KR}w- z@vF^l9gm$J4BO#nTznNPq~9N=co~NVvz)Ffq&E#sZ*0G8j0{{%qQ9NhE%H;p80dKj z>i)XLTKP#XORTyh>`|@#oQ-OUNOq5^v~V~#xznXdkLuJC!{W5E00tFtSx7i; zW=BtF0G|1bluC^vj(A64F$fCPh>|;rd0m+ONVje6l{cGSF{8@ z!^0DMjE%{H37idZyn}LYB)4Wal)}r`DJ_TNUsNl7#doZ5 zjr?lwJ-UvQbJ#D6-s2(f-w75gO*t^Uc7zV@2^|42r$XFEPk7_@@4R%3Lqvz?-bD0k zOd_H&U%t>r!VR9EJFDYPu0U90Es|}yMO_zo^9VT1Fq4L?SY=riWF(%kq3azg4}u7Z zf+Lt-jC@F!xmrC7ykS$2Mf*E$V1jZ?-7TO7M+T}gB)1?tjl9m-E5*MO)Zp-@kCh_9c-xl6W z-Oz_=Sv=n6SrtxVHgydsiphxabVi&p*hT!zbb<3$ql3O8u4TlI#2xXxSkgPR@uh2z zGqx4biNU1&?Mpa@MTGM?^zOK>Ex$Wo{hlaTB$~KuJ8TswtG|CA%SJCwG-XPP0;kw6 zK76)X+0w@^ZPq#9;4|W+9V2)DRG@_e+OH&9X^nvHNk@`q(7onEfgbG?@lazuxL}cq zia&~6Jw*TXY78Ntp5CuF2Jg4NS29jk^}O=lFOLz5;|SR+3@a<>`sOD$$oo53{;#7= zUG3{@0;fz?{`STXL_{^@<3)4r?iN3@d*~MtQ$BT1H)@kt%0Kpm!z^hwKb!rHsix<7 zXU!%T(@ZQ^b936M2q~#eCO5!~yJi>z+bD$x2!BYoo@%4eBRa168?y`s{6>F*) zJg-&HHY_w&ogJ^O4`us!kk2RRZ{JZ+xVQNO$F_*;Y(Bo<68zUx`P%oh1>zcu707Ga z+EU#j^~x&>#DqGddZUc;=+BExmICRAN{((R|L_zjpVHT)_+Y+B;d%$pcM6`)cIHi? zY%<+%$Kjc-(fg*pmV!gvX;{wADtA-fOlU0DwBz-?ZqX!kM z`HIf`5d2v>;5ST|-cz+o6U#*l(P8V|lev)|D*;bruBm`tY^$sLnM9sge!O9LUP1O} zPq&)ys+G)|)yO#&!g+GN?Y7k2EYpA$bMv8Db&+K;&+PCUN|(`R3e^F{kNJVF@jp+l z&}e5;(Z0y9IMSiDLdleIg=domsB2QmN6qv-?0dPV`sKTI0#9OjKPKOje(H1cUr?^? z9y;MwwPo78s!_>_W|%$Dv8#kPr_)A&Z;-2bmj73}v>vT?|Fn7a0l1FxQK z=|x^Gw=WoooXw_kigbHAqP%6E^Yl7o?| znzrh&2R;zzgi{e=qH^!_ifXo?&P{qcG;j0qh7{Jbfgv}dlSx&!s|C(?-$x3NFaFi` zfU!P|9F8_VTFCOR4MG;=er@PUofi%0lgctN1GKh@XKBWohDfu)7ybzj+;tSNm~Hh@Qy3|kI{O%i zyJ4{26#+M@W!Je*pWoAJTO4mH4N!PV8wZE|RG9jNfRk42MxwORaFJ4U5Vf?(dJX^$ zeq+bQ;8+V`63gx(JBU&JhG)~;bjt>8mY3M%(rNYr5<>)uD!VSEiNYEqvT!IWJ5WA0#fcFxRED9qMH-b-k%7;(lf`;MK53~)cZ z3iK$;Po{q(HU2M>ktgfz1M4|e?$$HR(#^grtVJ_?en5U~#aZefb=hJ& zlDneiR$zU+J}WdpXtzsmK}Y7hJQ?v!|DKJzox8HCzcErUfvPpLldmaW4YQqB6fosi z4Vrre@wyq5pC~AI=&L`LD(kEFpi_gWLWuzZ2w=3}F{M&&Gzp}=-^kA+xd`%I(cx5q z^+xI1aea>xyN7IpY!r3QPK@V48dWW#$Nd`r_S}rVtQ})YH;=j)jY?qhlhhem(LB2Y z<`rgW?CxvP_|eD5oFa_NWXFVc1{TvUC1SgM*||7&N(q#Y8-c4YH)uJtKUzx=txn=doG{+Y+bBY?2`)B`5_Hn zNs1}gCFwcOG1R`0EcNBE3O!;~vQ!jg<8(H5sq39FwMMgk5WgK#P@y|}S9wL5+-g1ERfZ0`G8 zapX%kHkX47Yo0&Xko?spyi<0uy4jjx7{}$|sGP?ALDqyzF-hiEG~;2&AqEJ&Z?YY# zsBG2>{@{G$+}RJ;0U8u0%=q(-32bx}W;8iC{=1V?|kIuWnVKR){Y zk4F4*$m{K`wDU{SJXc>1GJa#aAxGCQ^q|;uF>AMT8fd_m)zs?-f78vED>zPSqIudO zCb&zV)jaJJF8Zm7@qg}rWnOyWkB<#8Vf^EHonOHNegpnGIB$&;1DG@ zK`d^P2$&(-6>`kWdfEm~DzbjP(fo|E8h2quNEuA2gC__cQM4`$tFkW?rJJ)+`O&R> zEB6eHn`$oRbDaDUcvxTvcarF?RA?Xn^O7jIEgLZiK+rI6Ln8u*JZvAf{3f24KSuxHmdgNhlZGC9MS zgbGpN?uib%whl@t&P-B~>_rhE8WW=EimPqEPB&1KJOi2;yd^kTClg$pvfab+HDkqO z_K8mI@`pIYL)rVKyh)ANKU)MDSL4=YF_D zK)ddg=lH;C(m9b8o?+qmtAN_YdNJIzCKRglg87L}yCV)KgJd{iba2`7B8hH(c2c82-XU-C;5Gq7;C*eA~{KFjaw9qJ6P2Mt3|39 zrHxS*g?7`>(XcFCxt}sDi@Xx_)?(w7D>Za&PEw|4E9v=@R!9P@w`E#Kd2sH(Pc}9< zj(g6~zi;}#7jG*MIs-3SAyjx?^Z;TO52zu2`oR{ybtrOV?!&|OXkC%426(D^ZL2&p zmjOR3KRnmAw*8((nkfX!_5H1;k1}bCtKw#&U$stI+UHq(o{4FZK6|u@0*ZMg3(&(n z0UQ*4fP=yvspRr6Xb?aJ^9~x$%9S?Ge5mFV{aZjn@Ql%8WBgHH?uZg~D{hT}ktd$Z zZ|1XD4tq`csraYPB*;XXajP=)2us0YnT5H4gr7vKA4vZ3nc zTO5DuZmger?c4#m=$_OiK+Jbg^6=v&|53$GgbONs)5I-{Tw(Q4%GU&EOv)HZC4)BJ z!^Ky|>uZyZV-oMPf8NFNyO=3GO8-4RLU(+(m_gUiy71KStXq4Ukgh1_*Qz`6nJK+@ zogkQn)vD7Ue$3m|@mhVh^gK6j^rcIwY99QxhBajGJ)Yo!TLtmL>|wUP<(p*zvTjZ#YY-5Xp069c5qP^;1tcj5O6$_hc)r9y=qj zd7a(9u42fGWwv3V<=xj+?+?}8s;3tC;8lfWZ4q{rT-L~&j zuXD>l-oJ~jVZFu$(;cIJ^!%_fOfiMx!*cz!+Qjq!c&2@NmY-GdU^{d;S!I>_#y;X` zwN+IWRJ5Rqo@>1|pec3I1Q=%+2FlV;(R2m*sqX zn7*q38Rm+KJ!PCg#Ig};{srQxYw68I^zT&vezqj!*H~5Bb^fc7an^4RUYHHn-cR(t zDnmQlJ#6|j|CG4MF zb>V7d-fc~$SjB;;$dE~wovNIk>T^#?6mtD5v*gSp;_s@MCtZ6Lo)1o+a~;Qg2wdNV zD9>0`)ZciYcGBi3k`n&MS4i?q1*hEU``t?6*$DGK-ihG6$hShi$KOU83R0ixi!-)N za&6$G0A+yZ*WMo$Oy1g*rd9>~O!Kv^Wok+@mq^oULys%=phqH1v_xHNMQd3?yAe{s z2W1zSy8%_%9BIy;0|FnPe<$pk+V1A3dv~g_e*v_cqjj_nK|ISLN z10$(0-FS6f^GaoKY+UF&^Eftt;3XX^KUr^SgADaAH zPMTe7H-`mnz2PRAv^6+%c#3BxV7-s`*_`{{#5cZPo9&*5N^UQ44w|MYf1*1xXcceO z#+=5r6(es^ygH>n0np z)j~Z4r*K#Fc>$958l&QrxmY%P{;8V@jBXv!a&N{*3Cd;AUh5<3W*3#7{6*U&MY>J+ ztgnUhI(lH$#gJw-Rn)|}TmN>FFFn*^zaTS%jCSxOd0cP?*-buF~}Ci=)F%m^?R zl_|sDo-?9X&1^r|h994tbO5o5QA1ZB)dXJZy2h_wH$G-Usif%4NZ-6Sm~;Vuxr@fN zc#nkEsZ0?Zp<(_j!?6nF(-Q>g*lk zWwa7yd}ELmSKcX*3aTgDe2mw+eVP(LLc4x8-zqid%Tk*^%XwHGINkWqf$*#Ic3RnD z3`4nSIS`I={z3g`CeB@tC8p1bxx4-{gB|)Itj7lD2URV+sgL4r=AR5qvh0+Oab4eZ zc=@Et#Ywhtr_%8{^xZe%^MfNT-W^u|Jbfg?9O1&z6Y}tnm?RrfgN+`$pmW}Ka#>Zf z^eQU*-oKz>BQ^ql$pDelmV(7TQ6FBsV8QE~AoZ;U0vAfE)j=V?eKdD?Eb~7l>+E9zL1$gt-4l!AcEFclo>3W2lVk{2~ zNRCes?3>qwx5(lFZ}3v1)fEo1!Q;5cE5?(cxwlWhr%^I6Oz2WTzg6BRVd zH2riedUAlMnfrb7^wUn&OQ2S~xURkA_b+G`KCwF&(L231QdAky@n3h?oeU=&tyV~D zpC)idHb^gUuLdBic+NfUWz}d_wH97_tFZkb5dyMyobxe$4hZK-yq(Ca@_L=Jp1)fG z=+9ic*S(EWt#XMOD6F&T&&ir}U$e|?AkC;Av}u28s=83q@%Lf>fqq6NU*RH>x?Pl# z#hiLFYxj68cn3iP0{?aY8V`cSU2S&*+dQ_a630$Qv@-TF0u&<6@__R^uHmp{+2->e zPo)ArIdR!^7Ny;Ns4skEt@YC$TIt_&`M7-%&NG7noRLht^xPtb{@tnszEd*vY2|nG ztvcUGj<(b9qJV~^N!f(=a1}V#{I5~esXm8Ag{*l1JGE;G%8Tz;tr6FArejBH)}E_) zbF}U*d@B8erDlHYMDDKKehp+X+&bH%2g{6{jty4ryL;ZS4c4d6Ko0{&w# zK%w*BlZUnO^|-I3_Fy}bD4LR=kvog@%5r|pQIOa6*Q4zHLs369>E(NiBGrylLF8iJ zTj?Qi@R)@@q{`$gA&xUCXpcAJ8iO4agCY1DiO)N6X&+f+^D93f>=ZD?w*G}6vL)Hg zqZp8Tl@{#^kRi6~Qc-LxdiKrer|&DzoBJbn{+Rd;_uTt+@mzru?fgiZjy1{D!)P{f zrM^x^AI#in$k<9{B!sNsZ-rtg=j5;q2mP|nrj!&dVr?VndQ}A+eHC*|VF;az2~BG) zWD)`?x^OZqPDMv8>w5zl%&0pu$i*Zt9wo7;i&*Zd44909iSlZ9;fx=roF;3hp50(W zF9$!LFu}q{ZbkE)7g_BS@Ko|HgzRfN#RadBJ?rDcK!QK`Kw>h-VrL?Fq8n!{7`xgo zCot{K6Wl9>t0-IsRC8a`cx}P2fW>shveq&{4W>Wb*7=9IQ-ShnDQU|65WYax za5UHo4TmaNLYHzg^#^*h{Zk@igoGmmj=MNf`3R0WK+AhzQFiUl;d}WoY#Ej_Vw`#PX`O0 zO#~!0RWC{RRl-&UCZo=88|i+rr7Ny<1Jb_>Errd}I#}ZC4oT@eT~AR@kr*AD0K_HT zg^BM8`Am0H8QQ?O#^NY2uNsHby_kHjfOc!-xB)kIl&~^>FaF*|H$GLE$ByBH>Z;Y} z$K`fFcMFBp>WJ~t{A~xbUKd>|=sr)&jq;xywVrL!FP9l;9Q%6%!)I}It1#l_#7dj|YSPmd?T~9QuNvYNvqjlG&+PV{@U#v1Bu8~;f3$@c zp1jmf*JUE^T9aQoSj9dRrjcc)f5J8@vFwdUATt+Fco_2yNICj~su#ay6H%12z5|JkXK*stf6J3iA z{$j?NprU=nH$|;v9u^aSyb(KUE|v`0$>Zkl2BO}p9BQd;#q||u@-KbS$WihNAz>}E zBOV_Ocs~1cyV|v6u2^pv!!F>%Q)ixZTlt2nk6P)kRzYJ0)o2rqr}(j+H2t4t*qw?m z5XP{3#oK+^D>i;yQ!-p0lN)L94*Uc9@xV^iYoPo;hfn|b7ZjJ`H5y5sOz{}30cwPV z=`mwX;PkfUKq(L>;FEE_p0=O{O{_X+CwTPcI}aFtNW%(7tYII1HmctCXZ`f;``L-7rj`RQ1!ww za?28&31xqgqV$|kdzHQBIg_Kj8^59l?}MW}KW3x@l%cY}k5?1E`U&7aE_P87Z~XHA zaQfSRD#&t$jMRw1Hm{2wS>I7hBP7;BS$+F>hc5H2*nN)IC4}aB(EI%~)H5w7MGPlk zy)hwq^#EWoEd+|SdWeJNh(^VUtChBfS^%pZt6)J1@b|F$oi)sBO&y`$@p>w^At%ue zSLTL0tGR{4H+y{@e$A#FGTr|d)F3UnBk&8qm|uR)@WhpRzhY?QLeaM^Po+k^;`kgw z8r}oP%qgkAL$`_soOF3k=LskYq)=ufnW-P7tR@w>bm6U7`E@qdLgskDbPQWl69K}m zjQV>6CI7Gem|nE;+VUXfTm#eN6|C=FC@gzHqu672?M~+ibd@}7Dca@m2Al`RAK?;m zhe~vNG7d46jVW?J%C9w5-oL8{dHu)N=!xy}dwO1aRKA4_!9M zvZ$qeYPLR#)92%yn z_fUjl2l7ua-K2{Zz{l=OI3)1@^&i0%qDdpE+yv@>poed{bj^0aB&`TV69@yHe?ol? zaW?_Ld!tlvk3_G&#+xGbJb!({W)jGvURh0&G&CdNvqEH5j1x-0gUq-fig96wWLvEi z$FBqyBqRyMVm_gKA}I$HPvS|qvF;Bc;IZDZsc*%A)AR==#Up@OX{~uQ0YGb@Wx_B#=G0NGj^Bkm2~f_ZxI4ZuZePh=K8JXo_D29(Ea2Sb))D( zFEiVqu7ipN8j8Ey;JspdW7O+K>{QKHUI%9Vy;Af0phW_PBB1nmhVseePO$g4=)|E9Kgou|hlQhsqd!(JDc?$g(l>fPEF0cm^Nn=c+csuvI9GV=`Wyf8g3#QkRIVL|(M%uYG&WG?-Kl45p#DCwKT zS^*-&(i=ez<}$l@fqAn2=b0d3T=<;_S00)7*++ZNF^N79vnQ<>RIfjqQ$92`vc&I% zK1lNUqUgn;Vs!Xi>V(YjMZnE%8!+XMv1WZQy;^{$t;&nKJcb4VpUc~ym7KEQwyJJs zsa9x3(Ol~SEylik-p@&iTl<^c#;2-eupH*0y3YDthgDZ+amC>&$>{Q@4z!$ zNG~X8%<1ia_-~}2tQ&aYzf-p7xH${?kEq5GkXr-Wk1TS#7fFr-N{=Jrzmo3yyi4W& zUVFkv&jiPY4xjLlN@{y+vgGEJMUOA{r?*0a98qM^?D9H28_9j1}wwI*shHRP;e1jXCiP-AGf=v(&tGebw&c zi}#a)jh&t)b~hMSO^RpFX|;8eeAMV5KK{8(Fwo)G@4b}KEtafQ*8yVYNp>jZL%qGi zHX#e@Ip%`1pxoDyDtozXgr$+i;WNQi*e^8rMeuskUJ5Jv_}nz);Eb#KGnwK+Uwr#e za0cSzbcYQlcesp)PyJwBN+)EuQ|a?=dD;TNqvTZ zd{JK}sgy$Q$Xkz>&w)q314UgwFS1MPzXOu)VUqOQl&}9?vC(FSVvEHCQ#IK@)vR;( z5X~lPe3H9I$$58c2S9Uv&UqYiVsQJ}_b49?+l&Jcd2qad?^;7{{w}G~bOUe{NwTKO zTc?In&cBqNfbdi#w`hR%b4bFx$WFO?AJTdbifM<=4bM_61!a6%6%WD@fp~(P-eeK1FG;DeYcQ~MZ?pjzEyTI#dyd^mBN*G z-aSk{2T12$)5dv9+jibfSwsm!X0@O$8Tjae;z5Kx?wmHTMOxammmP1rP=LN15Xi2n zCf)ZoO{*JON!A*%(Yz*lIW2`>P0Hm7q#xfoBKRcH{KG?j@zE7?{o)>X8Jn=iWvRbG zWRsBxpKHL~sjGpIwZ#fgkNel)ewN&XRRpIM5OUx80(JZkGqN+n+t+?LqoZW$i2$}R zjTdm9wA!5^2e?!9;evdRu5 zJ9#r{v&B8KCd^maJdx%rB1G0W2^%e=zhYc5iP!Y=?Qu1K{8e;+dWw@i&^-rIc@=n38VNyiD;`r-2(k-S3m%EraJK3vj7>7J??IfnvXY`aX{{p^B=>!ve^el4X3mSM)Qq`j}s{VPLi@RQ^7U z)^yQLXy7|dKK`C-!h(%Oy^@Iza+*c3MV=&MIZm<$1-nf)V4#j?D6bf^aH&BQ0l7OT zhw#ze%Q#5sX=Y0`-VLdq`L^^NnbB0(-$1RD1jkZvneFo0#~r(5+p&!*A<02~)x&Q| z9T_(Vg|ez={Zr1nPonpKbcFS~|JBC`0h%JW{U_e1YS|~_E$IK$G8!wdPwyJqE==a` zdCVvLjdcRnFoD~Untqz9p2i&NkpE=fr;7yz0aKc)&Q<-@pAZNCtYi~m<3&Q$$|nAs z`~9pBUVIi1UU6e3+gV362#+$%X7yIV(0x?5g9`#h-it1vysgV*2E)mJfzD-MpEa1HOcMeq}Qjl?&9ug>JoA5RnAHvtB{MulvY z{~)iZ_5%NjXUY7Xj#eNV%|x#Co0;;SP;HTk8~-wlGHd6T_>Jd1DY2rrV+#Wz5XU8x6^drlOl%GpbHPF>hebDi$@pZVbUpmLGCc*|{p@ zFiH2gJT#L9!KWJ2|Et_OF_4MLltUVjnfFQJuNZFxe6dB>_7Q&HgYI>p4`OE8id162dJ~>fxB*Eo-w$#j z_a@TGiZksTV?uRjdv$aXN~R{By)8ZMPMYkxwoa-6rT%FtCUg11*CEPls%DI;(KP`rUq-LHjy_qj^_Gtx z#*BpzzDq*$7eDJF!xcfZF<(+D#FTG-vIvclb=RF*IWS{X`Y~N`-6)WBj)lyajSWx) zHU2IA4_v0hy5w$Ry_p9}ew*D33{Gj?=D8Rzex#GB%r%jW zzD#rBlnT8c|1p(2{bo|MDc^}P2f=s9>~(J(PeWXD7}QTqipO}hRsv49e0*KKcTA0o z$%>Mgd6?+Og?M=RG8X+~S?W^uV>x(*NhR0?3f(Uxnvo&V%We?|59jkjB2Cicc2PAXnXX8HcpnSm!cL0a62i= zG=4v!t{NCMyhLzu{U~>9D+bwp^<@V@gk@9gjgi$Y<wmC~L$p$SsK|ZCK_LtU7$Ze6UoY0SI}UHe%;KMnNuKC zS2|nTZi_WeRrgV#-y4q-2?Bxu-DRw1S;W^FI-csHFE#|fm4hX&->6iQ^{&*^7^&X! z@9_@)cB=iSH5ad_o>P4SXt#R8FYr%gI;L$M)Ok=GoOdA=;9?9Vl=4hffUex?TFTSN z^QJ=rYQ``Ur4#P1uzoaPbZg692ZM^>sna_ns9%7bOBJduNNX*=thJZtJ6I&#E2*ir zoXUnl*GZ^S*d!Az+KR{+lTZ|TY)}&!DE4ZjP^h+X|3AB5hXl<+m+=+qs3aFv+DIp`zFD$0*d;E@% ztodIFwg!$9DbC^VhH;#4AjetTf~KDN--ELk`X@+7y}NC<{$U4qh6Yrww>O@)XSn}o z0{musX563PbQ-uiqXFdR9lRj&-4&q34xLY0Pmt1e#LR@L>Ww_$6WsD_v2%)Exi@zz zzZ7=$S|_|SxRhC8kQd7I_(&k6(+^m)Dm*w`OxlJ$7e7&f6=B@?`4J0|z?@Cp1I{`@ zmAUP=PcMXQwkdb|c1rfs^a+8gsIv^~cuQs9CM;&u{HG%b_b*6W+X;6d$gs#0<<-gh zSJH7PN`t*0QXHlzJpNgu#eRDivG^~D19YJZyZXBW!ya2@yIBppq%vFGz5H!IJv^j5 zBJ-ZJR6UhHjQJO32SazBVwP&#wK@}Ccd{<(IemH?4= zd5!zZn!;4(j0J|<0J389QYOLyS?J8RO`z=OZE#{ZYZ+C`I%(k_8x#Za5>|yi!K2Bvss8|_+O*oh23fAb& zdM+5UQ(RSt{HQX_PE=@JtkV#u6r>=bgAIAC7uPu@GAcv@SMa(s8G3HR{l0sPt&9In z+B&iSZ)g6Wy}=k3w^+6*1PdF>lIJpb=VyXps}J4&kgo8ImSH*l?2@0Zo%K+F>RQob zr$yz(k*waxEXmF_pPSpaaI`Jsha&xPVeOp3>vr`o#F^ck`1md*@(!QlZ94Ic9VMM# z{*WGQQ5(VgBQYqt+-U63Ne)HLn#+diAm?&o2@4I$ce|>F0iU1F=Q7{)Qq15 z1GdN{WU)nunYBzho=OPoKooMZ1Qoh%z_6aAFlojJT$g$N2sMYhbg`rYd%nC2BBU7nSXoDX6B_{%!rYS; z-WtjM9`HrI=6Me%3<=nw92ecLoKL^=q}*3iQ#IGl;_!BD{cA_>jp&R~ zeDii^xevslw#^^ANHX)dFV$5%0#Vb1Qz|OTbTlJ$jV6EYhdHz_ItoZ z%Al7a<(SkGVCp&FDIOJ)nl^Y`N7h!h*^k{yq6Y}N=$?>cmWFx*c;`R zujc#1;eFpO(*2(PCG<}wd8mJyjkITh>!sdoqpPoC$j-n-goDIdM zRQR3B#APRnGC9rF7B#5hlu$+ndNj^mp`2PfomN=O(Qpq+)f=h0`|buzaT$Vy=XN%| z-I4c1En?*ZUBs)>3D$rLINbB?W+K=!F_!#mRD86;KRjH{Q7bO;lWq_C;(8&{{{XcE zvo+Bqo9lR%(=NhjC|A70kPCw`i}s49>cXq)JT|v8z3%SF4)92c7A}|s?v!_L+3t{7 z?~uJ%Z|iX44gsWHIcMZRe8`Ng4o?J6lu&I$cQ;U- zcjzX?DQJ^oL%8o7Yl6VY#=6jz!T8zJmdG5J?hFxXTGHUg48Q?OF-0!+IxadQ@R8?q zzq|sfj?iH_5R*|&-uLJw`WI{G9s<#lo*J7=lfdt622KAO3jYi8`;0FA6QJzK9TULn zOJ0|b{V2T<);<$hdE6eN4tt$)NVj9U*S&;XXeozm{OWd`yYA*yRn<8_V+15LU)uh0 z2%PM)k*O-!D^$%R42jec!*OG`rKVGk%`QIZUwx>o;Kr zc6P1dWrMOeI76xw57Jp0W0@8O{jE*sfL(Q;dLOdWw1V|V471+4A<TS5N&0VuonmJ#RiRqer*#gOr{j68>Etd(cgk6eo$cMrm3 zF{_3!0>3Ds$JY_Aq20S|m5zW{=UbxrdT*N9$o2V}@O3$u3XeRl)rCtH$Wg&a#G=ym z&vUvfFI7(w(;k%~gI=PovwBsky}_!ZoL839npxpcpXp~e=a(@l27V=5WY8cW{k!v` zHgqo4i4Z7QV*=stGtfj*TNp1g0VyvKD4Q1-9H+MAz;QV_UDm{M+dNfQ8S%g;WZAu( z6G%DGT*C-6+j8@gudwFyIe#c6e+u2Tf{$BeSNpCuR! zHao=bkH1ncfK|vq&^?!vnJz5JN=`D$>DNE~4L$0&r1CFBo!vLnJVIf21nCUQ8ncEC zuA&R6&pHQs-BrRRgGI3Hen>_3>5aAOlWb|}GZIvY$fu3H`l1N)u4OKRZNGO#k{i9*a~i8LMG{n@eRmL7jZW=y+Jr;OK~4^ExA3aQe1 z@|}$Dye7IP?$)kY$4vOiafx5mE%LNcYwD(T`H->rmv;KK73ME*ThUw#V3c~V+Oc)# zYtAJZ5{GYl6?PCmmsK8r#9qPrlyiE%t+^<(Oy7!OxQ6<1`xKfs+uTr_AzZQqMCeoD zu{I)WotNtH`qsUjNLBmDG4I-?Eey0=vIGEXmfBBck&}N5nG_yrBvhoPQ3l zQY0yJ+=(*Gn>cbqmG$*15?u9hXUc?5WUpxxbq@mO4Ki>_D&3uILBb7t)Pih;^*v6zhUHF0l@ z$?wv9PQg~5f|5c(1-!EAT~I@-DC)#mshX!?NitvVgoUnISmZ$58Gwr-({!OD#ORdO zRn>r#qkyB4`?RhcY13O@SB`@J1r^f=R~!*<0D|8CE7iwkk>sk#CJXxX7F$B@ObNpA zU>AA1YFq>`1;n`ar579)tNRziZn13av`2Dgk2S6F|Cl*=%jbXV18P4OjYWqBm#K=82q4esh|uJ0ca z4lZFpIs^Ha9du;p0kUy~o_L|}dNlEU# z;%KyAWxjDfvORv{WIQRCWn|IP&)9j{h31Q&KyM3l4juX1$dUb(Sz1ZwK)eF3>Y4j= z%Xp(4dCX93rH3d?f|M3F^C(_X|6-&q-}d@#bd-DfG@wn_>HPlgt}N?FcSedp z7P>_vrRJ%JDRxQQWlcYq;@jJ! zaBH&Z{b49No{Y$Bx%nu&|CqwZl}X%LZN=SDtn??=<}kD5LFDtzG#B!uLctarnr+>A zI0?qH2S~9(3!=&~+pSo4kvGXCktneTb z@6T8;u=cfpNx}nNe6a5s6VQHeb?@5SsHsVmD_A(NND$zjgLXSWQSzpWK7J5$Ei%d^ zGznx!*!^2ddI=XVQuUu!jAQf*W=!RP4_QRhv4f`K>a{aSz|sH;MgnjCVrpmdVVQf9s3=7|Kon;^KCzjWzAjb>Y?`tokY#9ia;hMOTXR*$|^=#nNBDMA>LqoK8(OX2O59hR#L=bGPIuVnNDl9*&P znjD_Y9#8Z{q8R2S*$7z<>b5JjKenHi*p-A2blxagjW#iC5F@r#7LUDs1)8nKM@9P< z*y+}0%x5IkmCJ@QLxdSMa_$ahhneKO@$I9ekA-t`nPbaYn01Y!?_NE>Kd{z)_{_tj zI+~d;0J%_8j%?HnQKq&8lGaj3ih+4^uy5eiV~>A9kM24HxUSfvxetmfyu-E94QOYV z#E^l@%Ph+P@cYqW-1w`pBBbF9Etq&?FV|wykGe+t%YVFQQE>Q=oXGJdPXh!rLU3Qc z!bIL!?W6K}ARF&4|6-c$DZIu|&Ac$qd206d0QVr#_i#j`KLW8c^oYbO8dhP0H10@J zAZ9jGT&Q)58zB|gKN_X9{xf2CB*daP{bLPRSM-NYBy})@w>XFX&;pAWULkJrobSU} z;D#pmig@+TfN_}xxS`l8RaESlz}Z+qx(=6_oL+M-21ypjzvp2m)~-!j`1hIAHumwb zeci-6)DCnYornsSak`hk<#o6%8C`T3?2VkCy%{>epH0eXWCz1rdv^GYb{@p<%e7EfpYJJIUbG%J8ydZ)J$BmuIZy;I_mF zhUUck^rC?@)>&I(|Aje?+mSCc_;b1bNi9*=I^3p#Tz$Q@*;}p1)+_6R$YX#0=FYt) zat(z^DY5oVV_e*vAlq4afP7PxbW_uPZ|J&}-@ef8lYg%Wt+d<}LA#yFhf`M~WtUI& z7926@0>^kal!h87hGo#)lq;55sE?LAeZ+@#j|z;^)C87YmTH5^x}Cgakous4O9{8E z&w>5Or%|hL)9g#h3&QDY>%%kXNwpx;{~hV|rFxjNN@rHTmZ%%`$9l)r=O)j2CEZh# zbC2sDCKWlX-;$Zx!LzTwG(Tcjh_;{Ns|?@MMRYkw4Hxtqt0SI%Y2?4_1Y7e#C2-c< zFCs^dX&0!`9L{E&uE6-^%C!N=LnZcQl235eN>-oC|2*EbBt*Vr@2ZAphGq3y=4SKS zOa(!n8~hJB2jmf|B!1zL8N2ANsk3gj&>I(_V$a% zeOi+_oGd|vQ>X$24viXqldkoJ;S_3f!F6)Tqayed>QbSxA@y^+ymN=ZOI2nk=!NV5 z2Lx1KfVo_%$7qRnE;|y+`jeB?cGpb>XxNq!W97e@OMxk(>+BuwS~heuf!DqZecZ4u z;TutTA3<}3j|ic<+7yRta)H$IFs%XI$+QeC&!TK}o{dBa;2)z#>#nJg*5ZmQ% zpFAHuZE|ibF33B;J%`#%4Vi|T9VcPlR79loqU9~;b}Jt-0T)2yYzyLMXK#MOGF(r-x`q6u!Z&j zZY63ZM37*YhBC{;W~9WDA}y`)*Z3=!m+eRR6$7wiwm$MYtqK@%mH;K2ES#l>E~c~5 z9w2lV1=xBPF^bH;Q@nK^>*lP7H{zF`4TbIwq+CDbMcqOC zq&4IUa}kfZ6G?i}rQ*!LO^g0-(2&==9itsm9~%yLslx^wiz* zwY2HackKw3N?v=PyEO)-ZbM(C2=`>YA=^~^LolakR8*9cFm)>$D5O=5Gekt0nTWGt z4N_|{VmSW0H7#f>b1ApHf}%g4hQ>xcLT1!toeq0%zLfOdI@spt7orTRl9`Cc#{q$v z!>2jHCOZ*=lRJ;9Et8~^>gYH+HvGkU`0A#*gU2%i+YDS2%t{uGseqK#MvU2a9&LNxdqU`?viqaQQnp~0 zr-kX!Um15c|C6fHFtQuJ>ceQ4Q-C!1-ITeQtwt`BTPSo((|e1W=;xQso96gC$%E7# zVzSlp5UYL|x|F-xz&DKh1gNwVC4BPHvrYX1eSF)T)451xdWT1g_>i?P=jJ2v8Q0c* zjPhzb+%?O_{fp3M!&u|{H?N~O>#7+R%eHY;o{%v@2%OC0<5@V3&O)`x{$HPb+LVoC5d`KY@VL)}!%M&`GA zJFXw6HqgO!t9sxqXYXeYLC{MD$)Kfg5Wr#@y>KgHosgCb>8*C>z88%zU`A;6Kh`3D zgwX6QECysdsnnoAl3843E}cT$XN8hcIHAqwH_pL?VO5?42>pOF(q}e_KCM*;0}Qj_MUudS5o6&(*N?~~>6cMXlnRepOk&~@Rc8AHd*XgM-EWkVRIny&%q=uz`P%O= zC=bs+-FHnM3M%t)JvM(S3sigOcjo+@B2W{p#cvuHZw+l30Xt8B;(WbY4 zO&y24|HhKZDtKSu#>3F^^-GPo%Lj)ai2Q^kTAB0$G?^ezG#{hJe3-~4?EQKQY*W{- zUB0%H-C&TpGXJ&&<~W~0>7VO${j1$Bq|q^t)}ca7;|qIzNI%jJn5zq|7G>Jcd4A25 zIR@Hg^RJBd&e$C^pB#4LU% zApL$d$+gJQk=x+sH(>o5nKTq?0S-k*-PDyR0rpK4=VY3Xx(gXMVySYQ$Le@3kG@x) zr{UZ*H*oC^JCVIw<)3fZVorFliSZL*uBgx@v&rfi+=orG^HSNsN_*&zJ-T+a1IYZo zhwlSmdFdek%Lx3D^Ngv~?fSo&%fl&(f9!9m;%%S8Ua5pG>AcSvw07|UH?b8-SSaJC ztp6_>z4(7a{r@+~T|?BPEiuwup{hoih{lE+IE^XIL7Yh{4HYOp{(PJu5sn*BMk`fX zGg)RhEDXneBj)MtY!gec;z&cp}o$v4A<;#-3pyZ7H6<)tT0qA1c>ch84Jl z7XC{F*6}mN#}jUdW3h%J662{|y5AemN3;Q?B_)QXXE75GLNToCbint!09uJdOoKQr z2AnCEn7ZUK0<9;x!{-P0kQ0`OLb*JFZXU1e8s!$fVQz;$T;F|`gau_QgF+U=WsJxW zb6`U-LQO}AZBvnc@(B@?tqVDIRtdq@3)#r;%Q=x1#ONu67Y~>g^Rh#4e1NhI)&Vqc zeo5}Ket$3R}*H;q)>Ccz}CJz3j@l$(-s|w4&GQa}9RXvAy2|SoX2d|8aso6h3QU!jFAYOBT@s zCr?xDq-BA;U0u}=h;fnmv@jTZ8lB zrwNV8@8Nq8%GDlLYs0oqg$F!8z6SCOUS^-WNqIUmNDv{Y1NRnmf6ZaBZT-{q`nK<03}69 zLf~S!GwSCl*;xD`24kee)i3LnuM*>8j5d9_$&`M4vrC?E3*Ihv1`;@_O0S;U`oT<- zO#ibr5LObbM{$ozwC5a*joq*-*uH;es*L`yAe*D-_C@>i-StKv^The)?MDY=JM6bt z$A+Jzr7Rts7rARsBrevun-teo%CzeB~k2TBugf+QRxhF87h~5veQK z-Gl5G((b!A!=hghq-0-g6{7g>+QweL5aTH=3#pN6*g5gwFN}{erYy;Rabt&|!RRIM zepPRu#kZonphLv}&n=fj^{ey%dc923lHpBDa^aZu)Km<|(V6E1zY2D~`lg2D2`?%+ z(d@(@r~1>2W&s88FS(;z3nD_QB?B@w?uGnnmM%01G{efvK!7^>IEOoa>}%YOJldplzBNA_Z;wfHgll z;St*r{ndHt_*|X|GupjwflhyCcGKBEIeU7vF}cP}NiX?6gMm5Wui^CXVa(N6c2M>f z z(}4oC#ALm-=2j7@4mpIOhO+wf(EA$k0#_>qR0^B6*JqNP@!l7kdQq;T#CYb+8Pl<; zg)xMjmD;#DDeRNVK8{Jz0Ep@%Qc9V}6My;je=>aS8(>C5XCS;hAl~M%-#E=ah2gu% zRIZ+qu>*K7*x`FJhSH7j;K_kTAepA%dcPNgU?%DQBar&?L)@dw1nEAb!)M)c74Ly` z@uqok=___FuFq1Qib>lbpJ#2y=b1YJXn6=pHsGZQ%Py3nF??8&OO^%!1Rs%Z<*b$$ z`eb+)t9Fi80y|wW3yeNw^UxmHQ*;*NEkmfLJ1)SJ`zYQ_Vr5&+^rC}vGgRZG+0GIG zydCmKg@AbiJydS3Tjy@dZ2;V_1as8?h@q7{8IJ=SPfm1(Esf zHX9d(f?Oi=tV9&ri7cOZDNHFJgV7jd9!?#vyCLub(xpWYSgI~I-Okou3G9+2zvRa9 zp|KHNl1Oyz))J{_UWR8M>$l;Y1#pyeezUiM&;~6zk^5{LmNIZNSRR)uZKysvfRGNi zcOCN#1F24UE(tyI`amLco|qsyGMr6KkHux;w+`mGR85mBMp(T4vi62l#>>8``;nS( zbh<9Pn3|3*7-@ez;Pjx)X4ZloMDEM?O+q2vjx*)v=YF4z8rG?Xo!q+57G9kH05&uv zt=!#K+VrNtmL8mDE~2Hu%8RVa=iO^(mPM%OWOYf&7vj&Ujxv++y`%9E?D$=q z%EtTT5wf_k6t!&=liDGNO|i3{-Q_L;-d$zRS_JBF$AoU5x_4X~$67rvJ$&B4(&U!; z^DSw~rnBeLsIS3Mr;*(gCmqbL->I_Cw_omDehvcrRqPh_F}!sgW<2|H=E2HE!Vqto zSX+F`dpO!}avRr_GIZE~-;r>f?gbKnaJ}P~^0XM%iTU(`>_YNq|JxIhAp8QMayHP@ zSkztS#aEe9dP<3xTqT+3r_sfqv+qCKw9e&P;)z|i`hJsBmtk?}U_VnNC2H6x9G)!> z>A-5lr!ae|2>%35G0&204554hrt*FjPF>2N8)-|VZ++Q1g9qCCT99spX0~CdNK!O& z?QB||jSz5H5;*xmu$8&dpks7A@BaZ z>@^3K0aDcED`qnH2KxGNf_p$O;~A&PuExV(-*n>-c*OxVQMt(M+-tU_u!m5*%j~x# zC3XN=gymNCYLn+j?oVyvx%d56iI1@BN5X0tX4}eS{r(lP=Pbi(4Q~0!aS6{`**?SQ z+`1v>OwHdxtDmbz-y>RH4h&3QMxRzFOy2so6>@8uV^w~5#nSD5QDf#_`#9WXKvrTf zmrhG92Oo3PxA?ljp)T|A8&e{o#^L)uC}x$t^+L5&A+SCSVyIav2`Mn`*ff^nj6(2a zZZ|Q^njh_*B|Izr%92sCb0Uok@jl>t|2}Xo0LxN9UtOX#m3=26v7lP^LRWe^^b^cES7enA&8Y$O-g29)of0jm=3a#a zzlHT#XABf2KkzXC=3rx8BK#-G-l+esZ0!B$f&HlL<3ZXM$c}oe=yLoLsE)6+9f&bP zTVRgzok+RQF#L4OcxhhiTyc=+Pj8*A7W&o%9rK9dbh+@j13GxB;1 z7Un-zeWp-9*>6EDZEK4^YhazIn`h8h6{WEyh>ofN9Zvguc=p9_LWzp%EPmv1vAW`P z3gePkGHAZgw1J_;vgEVvkwjua9Cc<4yhe6l-~AeY4u|DsOB z=F1O1)dnJ1Yf8)Z61aX|ZS;&X^ja{SR`3{KV2 )#y+6ynWq{=dP2yWCXc9Gp0EF zYp4~VF*{0~ow{@NWBAm8SLV6OmDZsyU3wSm&D&-3`kDi3!WF+lg@1@ELJPC#fBzjy zJSSzJ#{UO1aB?x(CuPy#|I4TPfdavP=|zDOp@$vigrzE~!I051hW^x!@&NH7NNwUD zC57}*QU8&J4TSu?rbEw8#YjG`R#z5EGvxDz+w8;?L+=-HCHRijmaOL9ufCO)axvay zi>YjI42|V+Np-6tzCk;Fm@`Gf5F-1TP7K{L1mrAOQqb~(!q}H;if*KV5$M0G5h=hV z&e<3nP+PIsZD_yjpBPmDubh%(!VrCJl#S zvHL|3IcF@ncMukVIVt(f&X!LpO-`fIzL={gfO42Va8XnWhcOQ| zt?Nxv=RbyD`vNja;O;li56CDpgSbZk1&y!)5$*86er2SW4vY#O#sXm&>`xLC_8A}p z|5;Dg;k7H6KP~|t0Knkk2)wxi0)Ib^!$vmHD7Q|eegR`Q_KEipWxSL+8-5=MKPmJq zyA&diZZDJ*Vp}*^V!tf65B&||Nq^e7BpAQ1OTM!hQTSH*$95xmzdrBor0RUV?j`g}{POP;IhOxX?Na|f|DURF z>opYhcVAT9~wpmzP#&Ax2 z7Tua_+o=E4Oc6@a{kzTSI%)m;qhIW@ttD=2r7>1&Yx(87NH6vK#riBiACYlsY#Q4t zd?)8gk5MO-FvRw(xpo?*Oys1-^|@YCapWA%PHOQvgLWCw5}TGQgW+`=P0@v^BS6%# zK~6zL38GCd(dE<$!TluuxuwOxTM))X7mI zU6jGhUnRcl1FB`vsB6gwwrAH?Mi*J}|2*>k=FJAa1C_0&-3IZajve;u6TEy*RB4gn zi09QIWOfg<-Cr)enO-)nP(ke(bLj5F&s*f;BHbkirGa_oFx*8o$YrcbZA3JeP1-!7 ztDz%-G1w_4`N8?~KE5$N7Q)(2MV*b3LD!@%dE}Kkj{+^h<>>u|8n=_wqiaf?v#w?LD6Bh=;S*QcjmL6zlNTOWM|M*bH65I@CAPAvf^ zQ?B5fPDN~s6_jgoqPLWVyBIB7QfA)j`@(;{uwfcqjeH(IA7{}gC*WPW+yQ0^F#m?$D>&L$v9_JA) zjZ0y_PGxwHYBo8FE3{UUc0hIZRe6ZIL!Kd1F3-h&-MgvxL$hnQFFgYqLGe7uXm)kw zGtMWkK4)1%RK`n0NkyW(?ow}arF0YS$OTm-2lnX+t}f*+#`=R3I65@=Vf|Y9hsL)1 z_dg2pv}iSPS75(0WMeKybL=1Jje-?~qrq z8_(X7XHW5!Bq!^C9oY9QW~aYX=a4>A3cGjVo{{mn=cku^op=gDG`+Cdv&B`^h+V~; z0d9{GD>r?sK}9RdPA)CMYe)u^93&6$x)*+_FJ zIQu*Qkz|R4x@2tM?JtVVK6-3mj?`BrTd-31Y*D`J|Al=M{~u6pZ9{eG|4Dx93koqo zf2$X{K_c;cGp?Y{aUMs)GM-Z9^BXx=Z-=w*L^)`cU^lD9uHiJK(>3*2>c|hx47cPv(>)ZE=PqWls5spWggd>}iXGU*YuT&s`(!4aL-j!YsMtGRWZ zLGpi#@o)}+hdSJlWl*I{*l?O&m}VZOD6*%AOt#>kEo75y#A{J1lok;K%vod33FVRW z!6c0dUj-}jP?QYVaUwQo0UW$2DPVyD{v}`b@#gyZOJfm!y|#? zBj#{VP;!@a8(&TXoJ2px@p3%32m3N}!-(8z-q7uBHRzgV5m4hYw5*kk9bcyzg>nW5 zFav(mh8>r+MBzPMf(G;?GnYtxFKY|o{qQ|mj@S}e2A9AG2()nc!mk4%@C^#`ll)8k zWH|F}m}LH8l;&{V-2o_<(+yv^4aXaXMG>k4Hu}+58iRsZ90WxVX=lzyXXV#R3=za8 zfp&EO7iS~)4N-Yb$k!>KsbJlueLlh7F03xJ3i|sLlinQ zB#?|!b85U{?QfJmcdN*Y*F@FVbh#YGb#1vrLGBlSzq!kmze}EYYU6w5GPz9)^j0jA zbmc2TR)_t`WQ^I)dQK_6ok^LmC!>GB*oJyA@l~=3X%-WX}+_fw|j}BJV zlO?C5D4jWPi^;W$>?Yqy#!52g;1cJa%_+>VJbo)V6cgOncDxmx#7%u}+*|=&#>`kV z*8jQey%QwAnsa>9**h<`_NY!_N_sF})nYdt+Vk$pSWBwrgxO2){gonq3eWg`m$Y2nG!NDN%p%WJN~f>)-KuKOE{^!Q1Q=( zse1^MpL+`FiU4P6S|jaZ2fWMMGaZiAhb(BbUr=)H?xt-cos^!e`9AgO0G5U}xA5~} zY^uv;Uy?dsbry?O(Yn^A2mY~RDGRTM;HG)Qos4r_){E!B3B1d%ky;V}^ySTWu{(LC zToo42WGiQoh-W!CgNc_)9nT{MJFtDldNob#zdo<2`m%aI$Z#PxcA4IOaE3C%+pU`v z);ZJ$<|ToCIr$I+>~+I#W~>n(KO8NbZoCwgC+lVJ!BMkyNhlN7jR_A^E;uIS=tpZD zrWt2o#}%EF)`lXr!oyMpG&fpky6tuTdh-T{mXu|lRCxKUZ<>F+p&VG5>hrPX@eEks zQL}!EaSGo1wU10s-#KqDdP%r)w=E}O-5<$7IUC0JchoJpV!-m$I;K_d}#$GX0pY@9ETx-e{k@%TFHz_>msN3e2uy zrkFq^w32A7W^mbizb?Ph61;0P!IvTDKqFpW;<>2?oB_%IsIpR|Inet6l90cIvt7-q zr1hd22;$MI@gVb3Uon|`))apbCD&MbM;fJ5nUGif`o3>_FyCQ!=mwAnen{FKq^GII zGX^i4_^^;;a-C33hPIdOHju$OI{2?9`Y@UIFu8pVv*buM*|Ca$gl*KO5qEsUp^xrB&!hK4H=pOOVtZj z)@7DgSp8gOpAXP;(*jp%Z^tDpkywxqa$JvkMhY@Yf{!nSy~Ui!D1%nBmEfyIUf!|5ys00(gSTy;*`*g+Yy!oO9q}?gGB;*ji=r7nvf-J(Y9Y zUR&&e+-VaYEvJ2$s3D+XUlXF5k@hKv+8Us^{>!2L?U{vlu;)i5=FCv4LVnC5wyF7< z-x9YSeH!gh?zOB9HPl94U-ZVs^W24cdHcj~Q3Eas3(b$%)8>ze#Se>HP7SwoGgk(+ zY*+1;B6jU9g8e3Or)U||7ml9pc22zS<5=G8ZDwQ15oU^a#BL%|Tv*d86Y3h6)-{_G z_*LO50q)^1UMnJEHecS!swy6>-LZ9=gk~=Lu3LpdlC&zuN%Ic&)ZA}=@IED}2loTZ zC@Z_{2e>p2xAOJJFY+jQsvQXnLjBY#xqp@552%ic?GhckHKExx`D4<|$#|b5v$0G= z;`2~SYlDH0hSPOS?nWB+o9jAp2N3iqXU%F$?z=66X8Y#vq{nW_a9yVu8=tN7PtK_k zI-RSVRibwBHA=%;SpV)Mp7|pcUU@@o=BPC9%E3i!8BC-s0wD&?rH4%F^AD2R-x$kv zTaC|_Sz@4DBGZn3-(?pu!88VSS<39jKUIZ#;65gwBHR~JL8N~7 zwTdRr@TGQutUb=vIj#Y|M`7RCcg*r;db=U))Jc1?)(8bTUZYCCZ!nHir+u|V7oR=W zl;Ba>15bjMXFwP~n)z4=JKLH2oIdzLg@%}qWlV5Ei|T4=5vUruktco2ce!63;e?0N zW@EYe8B=OZ&e9yDF6MrHb+0|Es!F$0)|5|geqHqb1A_#Hso&51b=sY^kjD1A@A~}= z+jYHL(v$m*#Ib63uoaGvOD%efrt;PA4pH{|PP+a&sX6O>vI2ruR^Jgc1VLHqu$>gR zrY$?3U#%!FeS%=LNnu{=U4o*tht>CV(8PAJw}2u#jAryeg_b^@d@Np_h7D=b(yGn; zjU~EhUr6k?*15FQ&0g^phXz_XRSga&hMrEc%ZH|SHty5hYrp?KC_R|%wo>toy`D?> z6~yv!;!*HTQS#hzpdhCz@^^;*FL7q@cy5T!XY`zU!jD|ANv%goq)`h8+&zvT2neXi z4ng!iVjDHF{)jOuNdA=mLT0s9df5xF_Hu+63iE2>&ML2$x!O`cHTtqt=k!uCU~i$@ zdfG&j(9$Fxued|uSZz3EI<)3~=~J$Hz@8r4NSaAM>;ev1`BQV#Be;^LFx;}E#RRLB za6rV1Up)kfr@ikOx8RH78^$DeLvhT@Cn2)D3enz3lXppt&U$rzUU(N2nS%LpnN%o! zx|XI(eBQ0^xfz|^VteO`m^-v=sCs~!9Jpd+f>|wr!biJ$KknnX&G~4XGtXZ?=-c>w678xevLGMi|r&Mh>0)JfM=&krZ zPw9o;f-GzAU$T>u3T$JOo;bka1B|~)U2A7IgoiPqr7{v}Du-C8SSu!@G@Z<=d%tnV z!B-HF+f?=<;1L5D)%cC&fOr}zBD`Hpf;X8L(H>^I7`V}BHwtp+NNAH2$r0On0btDj zkP)?^B6oVZgzfecFCyDD7q+>p0;XL~UeA8Dyv#i#(kpj!evtcdb36PmjO1`n{_||l ziP+5=#dAQ4GM$Udh_dfIgX6m!l^v!p8Mr<#ZeQn8{AI7%1pts^KJNdj@lP!0*Pa{-{m^L|~tZe9>Y5GeO(QQlfV zm|*i*t{>{to=?sXoMZkWEiG6KLkAL3oD@FoMLUg|-(j4e3J=>)paw=u}jm`5StfVON~YgCIHLSyX(sG2om)2Ix2{IbSAx1<=bRn%nHohBFCD z<}*6OW;XTk4NDr7cC8??w&KV9$^B67OQ1H@83=5M7N-wlzU2D5`7AsTmQQtld?kFv zIvIa!3T3&!tr&9$2<8Z|#GqWG!WD~p*D?C!ISimwbpsTLU$Sx8eh5GX#K}Kd4l_^W z9+{^>Iz(g)9TJMCArXUP`SDTcbSDb=G+MBEXa3XY zJ8v33XdvqH3=Nt5bXTeDCR5i_;sZ*AW~9IXD_6Og*W7}kjbFd%1E1WqqUQ!;ubNx6 zu(@@{_WIsBx^I^r2q~vFjfa&tx8T zU;Icwdq%md+S8Y-@uM&MF(zq{>$P}47M9`y|LH)3-ZH~^PFxOa@pHS9nDl%7pVo|{ z_`SJZy6PHD9Bgl(H6=GAC`gra>2|tBsg`>0;AWou!Lk5D+J&N>Po(_XIFwv|4C-*(XZ_SdzG0z^mHc4JZI5f}?R{v;4S_#p>Ztk;@Y6}t z5Dz1w&MO@(RDdiT&#Z|33pK|G&umYXq1urJDDFb>M(_1mrJSUlyT?@Mgu44>0*5gENteR)5dUbqXom@M-=vY{plnO@wW zNy$Y5Qm=bPz6wcH!+mNhVc|=CH!?}`d%X8wma%JdS(d+N)$)zp=`VlavT3OdC7TV^ zDQWA3wY%{zM604M3au)mHLTiHlE~;$6qT(g)}BNmN(_*bL1;j2803%FC6Za~1m^uK z%Ci7bsM2u6NcPY>FOc=R{)9%=I~F{s8aY$bp~}{AT>fdsx=Jr(w2?0m-+O3!FVtY> zZFYQ}B#=d$`)ZB0^2zf2b8gqL+e@2<{N4Cm>T0FPJ z&3Q?fqjKFjX)Q002ZB_+Y|>sDQOkpbnn#cCzsu>0*nC;hnS%BbC85JLHSOTEw_c8H z&Zz~B?#n}@XB1b&V*U1|U&T~XuPl>Ub)k6A6Oy|(;KauZAU{9$eF@s+ z!e5w*pTrcgCL#XP_{z(F1NqF4!>;n~F0qwPX{TRDes};pjSX;Ks)0=LE!Fv&WOt?i zhr z2j~IY>OkC>At?8!)V$&wjPU@#`EV=Xx}_ z7!P6|(x{y~H|+)+8e}Cz#+e1GgFo{A+2#@-^4kI{C!#Mg84*K(T$WX89YD17YD&U= zQsa?^WQD6RD6D8l^%2z?M5`hN){zeqBJ)Oac))h`dT&OC(Kx_%Nkk`!6lVza#=B@qv*$H-sjh{=1>=%FwmXG$Lg4Y`>Wpf)_3)WM)*6*&~-KJzwzL05+Qm(&n3hB(M0}j5|axs!%u%)zz-;31%e&E9=)Jq z%H8fbH=eM9filDZT$){cf1Vj6+txuS&*I8Yy!)0j>{6iYr8}ePhrDi%z{WD0*Nhp) z6iEF?#qVbFzlDbdKL`^k3=(d1ySw3{=>cdH|89I+VPs!_>cOih$nI1f@PR688j$5) z4|am}1>I!gv6~n~n-qg~vB5&32V9kZ+`E7yQX2%C?#H@XHkM3G2ZkuFS`douy{JKB zRf%(YjYMEJo>0X`#z>UAay5Su^6ngU<~CL>XJxeR_54iuJnliReG`wbKKMj#gEHxp zcaVG&xq%oB^4Clg!2&n*1ekdrNTJ;zpQd0q-|0G>rK4KL*WyZwCGR_dx1nti++b&#heB4eJccvXh4| zU06S0ukJ5uhRsS@7p$gM2_w1la)Q0EG_DypRq#2QIHZku=%TQmSy|1ze@lL}4C~|U61JXMd=hibWM;G>wQgwA(}DhkqA=ipLAK@^ z{;wReK*F(kRV!DqE@lo!TVE~cD%|Ffe*=H~%eiTpN!Kfg*XC$bL!1GJe9+lUO-zB9 z>&M$Hu~+s`LjfjrjUB``-0QoAT5KKN3`(YUYL-m7q2;Zgd)UDo1hWPJtP3&Xus`I{ zmE`Dw*wnq#Gv}*cx@hU!yJ`9<779MDyk{3U;%XZ;DFMVU8FLI+-eHHIo2qktm2mWs zqtiSwy*XTs>{BVQwQqo_MRx}XpQEbUqlPYMO>M^Fex)rVPs?|UsMVDQwVXd)*Mzd; zzukByc3Cct(#Hrbev1o%tND{<4}Mf_RP8Yz)uI0N*pAhyXB*}0G_uW`zM0+dFuhaf zw|$;DR-{LqUE_qWR(HAQE8XH#dhuXXD|anQQ6Z`Fh6Z>Eoi^%5s*Cp!&iwi{(Nd9Y z&R}+9sGpi5Lh3g^ntpmYN!T9B*4|rGRuPw0&8#VUYc`)clI*C0?)2N~y0!;3;0?<- z7kv=leKu&feA-M?H%F0Yi=LvBdn36Yj#>^hkLP~e!gbwFdfVIncMEV#WP7;>ItTs- zSnFAhof>X6G2zLhc%)PeFQ`qFxoP&>gyY726N(K)Dh+us-Ln4T+ zUvOSNjml8XSzO|%2)3gjVi0lVOdGgPqM#y_QAXW#05xxuYz7d^T>vq)OTy!gjW8fC zbB zFM`Hmm^r#lMiGP)8kD3}r1ydxVF6}K0E2jxr4139PYz6uAf$owkktJmONcJOWA^sT zbul?Lr*?sJ!=x4tO0eA=0aMY;B|~suH&QMPJIueC!ST#m%RqM+Ew>>%b9<*hv*q=N z5u~Z=LZ)6efj&H-D#lUMx)xAS0wx(^C4Nom*^DM~uXk z264bgg3Znynt3* zJMMBJS2wXW4A+kE?J(R8NCE^1OT{K8S*g!bvO?Wf#+ zk9!*S2_sSI@^#NSH#H_&#`NjMf(<`g`WJev*HN{5_@L`(enuqyHP|#P!9J)d`)D7b zZYUJ0d}qOionMknl;TMd<=sSb)StKPPf{B3GpCg!jcBtcj3}QZG90JyQo_$Z-k$bu zpVbGHKS|k4YiHC|0+{CQT>_nsk51P>$(|_(sq9_6+kCfs(+*x zC%icun3qb8<0Y?lWv7B_yiZ#cZ?&G2f|;CR-4Mk<))*j2KmKNE|7_Li#SvnwIw ziTrb3oAJc8w*8AoPp7utXTYZbJ3%8+6hg;kX`O!?*RWB$z*RGAu}q7t0q~ zrSsx`vu#k^h*07iXV}+ZT7W^uy{Z-)6XTZTnLliZEPdmU5Ao|G-P(1mWJXe+Xq5gV zkpw^4ST-!N;M~*{`M%)Pl{s0eJ>yC9v;Q>&*WV1M1^Y}**4e!5!;||}e!SN+soC#g zfydk$milR&bL-Y!1t$Yc_Go?WgBi$5)#!m z7rK+fQ~ZqNIgYP_RsIrUcu2z1oVf%+yVM`PuL`N}Dc^d;=+4i!tIQtA3r6Tbgg3}8 zMZQ(`59nx>iEV0At$c~Yi%1-mvvPpN4{qtY>bHbrs+>c9ACv1B4ebo*F8!oVQ4SBS z5DBBxVVAd3Bav8GwUle1y*Pe@=s`%TcGEjYXXs{xaWU#S%xMaXu=*f}dE5<>Qhv$D z3PO!ZHOGDJw}&1tF;6FVYz~e4Ar=}j9#lmRU#j5ZYKvLD+b;ejQS+C#7bS9>jk(BhT8LQeJ}8&La?SWY?~s71S%A1{jjHPjq>;gpQIc zJ0tFuEYam@DUvz}4DHEup_{X*PHRDuv}8YNpXaDKN8Ria<v(hWxrM-g1fL-cCY9n-|$NBe-aP~;E<^&D43nxD0rhlfVe~Yanliaq@tKE ztWdHXgp~wD$3Ws~EQt*+miG1}BX^$L5CDKT54ab)4p|?IhLVOfGcC6svkiUDx*ByC z*EiC713+ryqCYCc8YLqAAp<^&KmXFxnKkIk5YI<=0jgGY+TU{1a>MH!Ef&@gZod^z za4+Jz6?yHjIF%x`KHC=WASubWk=}tvRSr7?Jr(?k`!sIfOiy-2G}DG}ketT>~6 z0sHX3Y-PU0siN);iKr-ZUx|j6-<;cWdV`G1ek_U>mhKN~QUF_`*w**yPBBTXYDa#X!<|=eD8VxC#3xt`#UpOnE&?8_iyFI zMa_R{V130eE2(fffK=a4yOP4m*(lY!V?*@!nUgY#6{g3k=(?-;e&p9Fl?fi^ zSYI||UP6^!!V0WYC^V(1mJx!qBq+r~w#t*6+?e=LRRx%ULrKF~P7|K_M)I(e;q`&) z;5{ANIg%`@sfd4Qa-^vpu+g;HrCL_~p(3i`+T1od;G4)yP2ckArlC{y%1q3T$jy=& zQHCFR#f*=cZ>vj+8;$H%Uc0}h6fbYd9@xJf_Nf=|9cErvQse%5z*~&MbW|ZI-~rgK z&sLpi=&SIyWXY{y;`z&ay_5)eDb6FOeB1IW;lQJ z_{9Kt2fhObT7@n#P|tBWIL{|(e663C@eVdM@bnfOy)D!>itn?VtUV55AtoSWl=0}y z3am6JzkE1B)gdrpC3@xUHu;{oMytxh=YHvEcs>4eweBcis4ijWa966fOOMMGdsG$R zvb}CKz(Z57#J1YF;b4jW*T>SQ>0u8t<6N3!BF&_FdxXTtA z+(X=cfQ3zUlTQA02Y}ktJ1Y>&A1}aSRAdk-C6-~$!g=5ExPE35wrMsv$?CnITVxVb z0||{|{}nM=tEdk(I3ukJ(|OXJY5Ak2DIWG@+=Z1%)uev{o^)Umalx!H9YR^{)#8eu zoCW~B8=}J@PulpvhiK8C{ldYLpO6zFXH$*^&eIgQiZ_YZ`<3ocFF98ivP6liB`an< zEeBRq*a(<}*3Ja|#gTxtBsdpy{8Zkl%k@W6am16nOAH)0uY6P zR{k5)!1^FLQL8z)V?4=ap6sU7(Zf>Xni?^+`9d2Z{g81b80=e>LV!nex!&mTVx(^y z`1b5I)Dj$Q+j9PUEdoj2^<%%wnFyvSVC>vP4JmeAjEhRaTD6w8CX>A1SRI$S`qBQ0 ztfUOxWbFb7#G*${e>x7DCM2q$TZLSoMT|!r>?xg=VtjgvVQSG^PJ-nmVa$UFVxt3N9r|^y>P2^&b)dxIr*|oeoX&xXt)T4+iC@hk88$Zwht2faoGZ! zViAQeIA&~rTQSH_Tbac1JIPoox3!YyhW?7#1h|u35Z}F%znw$OZlQ)1%v6N0Yu2t$D937iTCN~r66X}@^9 z)TU4OBa2w`T`GV{P8LELAAbE!VGizSFLin8=A`X#$kTJQiUB=kvr=SL{Q^cykd*o`hoh^Ww3a5K&7-epK(Z z#%Ls6^4^i;*pJ$YGSG?Qg_1cF3_R!On?7`P#;1FBLnLIO>&*)fzz>X4JDp{+URX?i z`Vd0%qf<4~PtNeugN2Rm0Z<4pz;bm*p<<$K;9Ci~PWexKJwk$wRQJ3X1%3hF8f)<1 z%p(-9?~KFEre_U|5Ti0v1e(}E-JCgL3Au~#?Kuc@kI8=#CZ zBEzkt^M)G%UNJrUZlOF>7GuCdme4LU>Y>pwr*LVzWL z>Uy7cGV(TbEZzxg#cGo~uEy8DKiHMTUAt`N>+yBb3Fn#}bT1akjP74dOzn&KUc%wn z{P-DPR1H|6!G=tAszoOkPt3Orpf?@~H5U;sjd6PFn?RPPCf~;;V1-7@lTd~?YZ zrloi14Qr3nenaD5q}L&(p)hG%#A=t*YEtnR<@50G5>8Z${A?YHCzL(emXLeG<$lM9 zeW=#~;a38hd363!PhL>8vZYpT9xDNuJr#Cet}=S z@OtR>29c-QVv&F7Z?!#vtj*XW;dEY=?UlJhzk)grxh0p?n0`c0U&xEn@Jf~HA@*&L z)5SaV8wolqD(Zb77wVWUK$X$oXKB@`JpXI}v1G4({v=j4U>*HX9?pUx)ScRxem%gA zyKcXY`PWZ7raz1*6PKYzfq(J725%eBIVO@^Jw+;VQPR(Lg)4v*^$ zq1~pV{w%iVyzZ2T_cPwtM(1XL9H*Oc^+C$cTRs*1pXB*=B`oyIKsbl|5y1qRn)!P2 z@$dO=BL>bQ$b2nP?(0M=b&1IcuBtOX$tLRVTGJjfF*ioM+wB8D#U5|P-Z?ye>Ebyo zkoR-q&+;kp15W@PUN~5F|F-zn|6OZ7xN`)4QP~mY@vj8anlp0>YKz^j*pg^tkLg*W z@Jm5?VqzG{T$uDuFbyB=6IQQDuuP!N;z1`s2x#U?AHoeD*Igh1gco=GX=(mSbod=| zEyiN03wtdUU|_kd9Soc5zI%|DI<;BQU4hZ>!X+c}yQMhn-b@7PSX`DRPbTwnobEyv zMO)3Ap8bG>AZEBh(nkX3SrPx_Rhx|%w0G8^&v8^cep%u|b|Z_HHVQePWT4+Io{mXP zwE@s;l(7Fmui#h{&LrR}h6XXxU!ZXrpBwnOode;2OB{SBpbxhN5RSlnTPn`dCK}qpis5ARN^4V(-xRmLQBX2 z^!iY3Z|vuZaV_C_efH)6p8ZRFd<^9Q^Zv-QR@kubbEf(dUb%v5xnZXwn01a6&|@<( zeciNc)Y~Q4=dH+Fgq@_Q5Kf(w!PhV=i=uJHRpEp#vb3i z)Wu#n1TfIr$CK5|D^$!&1glT5muq?GnYWLb9t~u(+Q3N<{d{i=E75#U(>}r zyFB@13~5`+|4ttVuwZ8alYXH7F;w`BXT+}$-|ZZ?o1mTA{3&OX6%zCK)jXMTmct^4 za>T=#i2P3m;rx`nCF?1q^Ra%7Am1-#*|mHKU%K#W6Wr4?NM0l0Q{DF`wPzcIuU3C~ z95_?;r{amC*Z5sCsM7PqujNw~m$XV=%q|8zuMsP3wyLeQei{(g3HV{RFgNO3CITZ7 z7q}np+jX`8th59l$9p5J?uDP0>X>&q`64sm4JF8d(6^ST_xy zN&3_RCc0qyMEAtAU;}Di6vGQQ;{MbEIQA7+Aqh#gA0_H@f5bQbP(#&MbcN@oK$}|Q zWQu>1-Q{JRO!ls0T`v5?-_g{fq;%}aSgnrd!Fa;yA#&2J!K>SQyy)SGYGBPc2sl_G zP6BCq*Z;VS` zlJP$nl%?fk`(y6;bNnAO&02fZrQ~_cIyI1pE{yB5el&hAHjg%2{|i42nZNOZwEzx8 z(m$)sMf=KD&4x_6wX%xUhtV-r7?%a=aquvs`KJ0udPI7ICV*}gUihhK;Y4=s?O2+2 zBpj~Q#pHh3%YPVNs(TXoB5-;v<5`CMo}<@ey>YX3s~jmH>nVGxkT0b@+5r=k^G}nL z`c850M1&Qw)Wam_I0Pn>32*Q|iDz&&ZcSWr3j8)!-V=7mI!%W->YvHlIBzbQn@PCW zt5W3EY*S5?ue_3E!3p_)HqUiSzCDw8v$&MuEcFLUcW)xy(2^BWzbDkYj>azOZws2$ zvW!}L3ePsZ{n_;3M-OlkdI6w<+v}c59CpYEl8BJ~4`hf|tWRhS9^FeyKF*wGU>kV` zgG{^4ZTYa1Z?8XLjx1#8{s)Se3g}W&HjL>#vIlu?^cHYXNw4upQOt^Y&LF#|z`DV_ zeqJte$y}L)vu5`YON%3!pk34WM5IG=jt5^^c$xC`$UZ+_+lCSI46hF+a6J%1%k31R^U^RaZY$RSG2rVc6xf&pCeGa>0jOSwV5}(s`&_XG@e36! z^Gfu|allDfeeNQTzl<5yAR{I&=~Eg{^x8Jy{d|%xl9(|Wnvbri%yssz4Ti?f#ts(a9rV3LHZ=LeJR@l}w?*Cy7% zTEdIs@^IN|uzE}ZeEQhaE?nf;zc_A!DF^rUF_AMln0~D!^M$O|=HwvMtjgLUa($>&PDLa-^N<`fQ{5TJ#r-&S3c2pmUp1BMz{=>y-hzcj zlB76z+7_i3ogQ3>HFeG6cd=&peo6cF zr9^6Tf^<}O9HA{1jX^)aL_BNcGtoo$hi|GjzbhS?K0k+3E9!rC$Gyg@q)Xl(CjGNU z1)_dDIG+1(CRlzGR3jwy_eu4=&$iB@iPO2L2b8hu`fG;q_!R6Eqj(~RbcJSV`Ow<; zFqIjy89IQQtCZQ+s@gF7MhVbJo6%as_UkB2WB0@NpFwDKj~OpAk=my_ea>*OBReCu zo)po|OIkQOC1opor!O%Cfw6X-o3Yj}&`G{X$IjY`=an;x17()oxt8=k?|JT0)o%aM zo&xQX)`0?Sw*QwY?C@mWYTy}t<>9f$WxM*Kt9PS$pFU$!H_%~y$!z7oQR z2^eV0Wo{&Ks6^mg_XolG?33GS!|I;R_0I!ZKCLlx0%uzcg^>Neyk6$@;~@<-*W4|qftMZ?@8E?Uw%IquqgAb*XEyM#0&^x?S=!@vdsOo(x;@o#5!<z&(kD$7Fs;oEt*b~vwb&jN?1(tVQ2s( zTL}F5IuD!c6`vX%m$xO0j(Q=E@4yR>=$<#I#hrfbyK{hc_?;R~MOE7fh`mLJlNlC( zFXDe0v&k&~-nQP_2<862aP*@_dA+C6ybeVK-{CTQ&+b9F96M?B+L|fyRXV7_kpjp1=a(*fnMo#9sd^LV_h2@2`pJ zyzevT83=a_y`ms%SHwf5jSf3kg4~<3v{HvF6pLzMAi|cPsKDjQ(-L%!xv__;3ioQQ z-3YC3z#Z*#JuA+xf<29?zv=f+QDtZRWIsNB_0m$=XhLApXh0Byb3iDkcDewOI60(< zkchJ@Gd#{#bPs69HpU~=+F@!xuc7cx%5Dp(9HC_ZT`7lAwdSPePU|Jy8-oZbk2s0nj|a5E<|pzL#o6 z5cm;P{DVH9cx2pUJe)a}9L*$zxSd&I#H9Sri?cyW$0Cb77WPYE4~2gATBkdWG_3jUh|=Cw?KJo~gu#_IV{DRu$gj{dh4=@eIV$HR7gzzQ_k| z+mL$Oj)&8y+MI8!SU$>-zj4R*Y?jZo06?16O4`6jvCF!ghw<_8lQu36qz+ z0ujvc&*^EBb{lK&+ztCvOfJ{Zm`yW%+RjKZXfo~9O<)w?P$fUnb{ivrREG;Z{jJlU z(dQB~Sn$w_Xz^zf*dA@i`H{es+UMCW_i8Txgu7f-;lJ(c{ZhC(Akx#Wx4S^{l)WHah*R&bbRk zuoqCSrkv@x_Cr>1U>feLst9NQrkL}%*g~KryPlttI_pFjPO2m4kn&3I$q{foR!4%df z$)b`XspF;LzpnGO=*VeRr;MD66w|b6g9B}+WLc32@k87|6Fx7{VgDwjWdzebmbNMd zDQdavm{MHsV_6nA*tHg{Y+jB!`JimwyFBPP^hY(DggeeD`+~ah;Cf=XdZX4I!LwBh z1wI&IU~q8ZlvH5g%l!TuI48~VMR-9OW`!00Qx;_}lL*pc|5Ve^8QgPlQPteoAlKN)PVB*w2aNL&KD z*KRL0cnEfCIyz~lHKmg(WWGec>AS}9L9gCV_xMwg*v?%Tgu_W83mJJFRu^KXus8qxmz02HHdcrfqAErhxys4FG9WYtUPh2 z()oA)RrwU2^*H`5Y4hA`;g)Lz+XF(7v>n7HDzB6b zodPJ}E_L~Fi1MFDHJZZ)t>rc*db4+)9)y+OKF(aVb=z3F`CVGP`Jdv6kL>}!npVty z9b3Iq=8;pH9Bgq@P+2|+df-+dm@CzJ$)~8`qB38fP~mVfn|FQ=1%xV!A=}ge;LrxP zB*Cj55)P$CMC(DBOO#`c_jmb$ytCe%qa$hUKah=asP!I?Zp4|Y} ztp;ipOJmy`8i!NwC3lEMKTO*^u6wj2Zs}``6q1F47C>wypL{V@lqDjEri9f$c2?9x zw0N$%6RZ`nWLOOMqJz%@E9II&{mK-Jx5i)^mH9HI|V`I)>&0D;HU~Ku;_S-l7&Wbx0KjOUd+Hx6yKT!NCyE{SMyN$jk!*Ai~LjK zvF|B#WUu^snwuw^?pvPHHAV0YTX>V#wJ}+F-C2^jLBx&u4+)1^-iZAa-6I&zwvdA0 za15Zvd%D1-_|l`ACcnhyjQU>*wQk}~jwCxv%Z0@pt!%63_YVm4EKSy|?MWoHSb&8* z(TraoyXUnXmNHoV$HaC7t~T9Z^^b$O3NL)O9Mijk%QkmMkrHEXMRTEWj%npEEUhjh zKkE~-WNDzlyO)d~z3|ol+L`07nKD)PV{Vx{w00RBDhA8^fj=m}Xacs4!VV6TW@+th zhZE8PD|kifNM_40b-Ob}=y3Q7A4u@abkw-Y6)hU})}AqaKo7QhiiZbGJM=l$rl+xl zy25%NX&c(6qwGr1t`~wWr0ShxEv$E5LUtRFGSnlM^DT%wji&JBTB{GxdDSv?IEUM6`z6~?N&r-m2z zcX&F-W>To>#8I-NW9qPYQ)QKf5H0Pi-7|U%8gp+z%Bo-RH^TOmPYWh`*yK6NKhf_Q z*rMOkjjrKQuDElzBYWfz{f0}9h|9saIvOV$Qo0-#mSNDZgh5xC3dILg0>)-E_;A9L z^{KlrWywpf7U-z*Rlk=mi!;e!!V6DQI8EMOE$~W}Zu%>Zml6x~{TyyWDE!vAQEY+Y zn{I8p-1_)(GZ4qrhrZJuM6pqV?rq%e^Fx`>x8D3!pT7B_ky%RF3{Ll%+$wcK3gBKyubr=7({jo@2GM7JFpe0!HiXV3}Ma5f^D*x=c%Rmw; zN;dLnK4i2nd@K7nK%Hu`2#gmUcy+7FRJZQr09e%5hh6O0+`$O!;|^R*ap)%e@Y+~M z8E`sh=Cg}Z z>?>;;I<>peO&K;~yut4?YSTi>$*9XuP1ZhexBhOGJg_#wEovKm0qRWFyDK;UqFjTt z>|0_)^JOW3pv#nP8td43-Q4qKjx9I3#$@L;h0n)WneTtE+%5{`zfL~(P2uXIg6xv5 ztbyUQCZC$HJg+YP1^^FodGzh|^UFKkhm(pGNv$3_0|`BC%kn7G*;Mifx3wvP9%`N}Dd=PJ4`|tu1xOAy$Frf2T=qeVA2lCil4|iTzb)lWEHQmD zh=V~cUNE0Qx33F3;?TPrnX^R$LTGSOTAcZ`91}XIzx4gd9V#j4}V( z%9ZqPq7#HT{Qfl`Jh)*#P2*J8UuWQ;oBzQ=#GuD`3`nhk^jZS$fB6W~`yLXx2*SZOPJ?oo_g1!<&n*kd%++buB&r%{* z%n9bi=w%R&2?Whl1~I@tRBU8h>QRLpyMq)v?QH{}ukKeg1+D>A&NICD&T|WW>hddh zZfZK~cWIEMwg;lm8{HMBU^eM1RC?}e^4I+8KAt!rMZ=isWiLX+Aj9ucx^k^c)qD*-#@wQtdVmR31RAu3Ii^vUEzIZXp%hNyuv5 zjiojK*^TW{V|J;?zho65SDf50iSk>@^ILLWLL!lBA%BYQT+vAFR$IdjrH(^?j?<4H zuP&+&&l3`YX)~yEgP2YSyr0aD6ddQEt)@bRi*;4<<2;kg;~PPn$ zBK@KbC<8OMP(bqk9ek{G{d5N(5<2oI^G;JYvNF5O(0)5bXW1c7$Gz|BoF_{xT@wjs zhlnS7wOzWwHvXM1#Pu7+CfF?}I+*$A-n0m1tUe=3k*1_2+cJudaScxMgmUV?ZZ}m{#*XA?!p6}d>-cacL!-{T3(Prhx$I6qYSy|}^kLc6eY1XTO*X!Lag zOp; zWQie;)3;Y{0}*f`uMidZA-iS+hBF3T3v*8~zv6y3SQpyTy~}$d1hE!1HWxu9@{OK|8IRi= zsfii5jZ4}?f&Mpvnttouf?U7Fnem0KTdb~siaFnH2f)7dxWnSM7ivnXZ8X2BorDIV zI77zX$#r6dF_1t#__&QvE4W^Vx$X`2g$p6GO+WbY z>atVWQRBw5Y?;xB3T*q{5hIGFv!rC=!*7@B_@cCZybdiWAfLiF@|q&=+bhb5G1ZYD zX|*$#ujS^@Od$X#83j7Ji^Ws9C8yU2XPAqcaVQ^N_6uJhRp_k^nRx=la@#lKv%phAEQ+@KDR_RP=;a{=ZwXFR@CQwvguFi|CebAPRX#j@qqG!pS z+^Fy+^B@#okbDz&I9GCaTw@{N7;2tJSg-c3RrbP5Ne=C+wg(L zd~K!fWuNd?ZQvd~nICO*vg+~}?SsM-9SbGi?j+Z`b4v|+=588$HQ_lMtqaL~5!!>_ z3`C>lK>s3&E{ksZDEB~GwANxdLFeXcG((CW)((EqKo&jd;si+om+^8Z+H{O1Kw?>P zJrWyx|3iVY%bWJ_QjBeN4oiZRYlCrU=h!pqC-q z3*>@cCgu%!4~T=0hBLtH5`Et1rn<1LJ*C2#oy%?;4T!SXk!QagM#RHt5U{kP6sAAe z4-HJ(ZZyHGXKS+br9J<0?*x@iYrng>Tds8Xp|8c>fUA$`bodu<7w@!dS)}wuY=?`ajk%Hw0+z@#m3ba6RGFe25a;K0`k-O2pt$^jCiH;Ay z*D>l!ZmC0JML{ge~Wi`*!^r#EvIz!l{l}mv(bDTBFFf!w)xZz z`!RhfvD(<%uLRrAdXR@(dZL^P&C0yRo20(P|Lc#2pj&^pgA5OxFz}}r%}rI{f|4am z&Gu|rrzFZ~$6%HBA(?VUU(IfBYnM(q$t+}*hiLk6@D~JhA60+gpTB_)lDk6gLC||h zo`I3vJ2~qm3l`2VX%dLK|3LWR?R20)5|GPW8@+$J=hO(6q!2CG#<%S6HailgTX#<6W#d0z`73eTl@|Wkmd{~CBTr8! z<04O`^>vrkhTr5{d^9A4D*UOU{n_|ZLd@pI>Mfm|9v8xV?Ju}{bD45HSUfSJ&N7oC z5ll;u!q^*t&fj+N6z+t2`muKP;<1`+z;`Oc`xTFK-5IFE3iIjJi&K(Ke9nF^Gc4`g zO>{=5*-3M~NUdWpC;J$vrM<>OZ5h_`_*xrd9iW zP7X1QJ?0sp5H)0x5QW7Y2f{%xQ57H1<46KG4{jIu&!hf_{Q!H(rgmNyS~68$Cx}#Y z;i0h+Ji2@jUHdZ|4(#N!=Z&v%VPZRowwLA_tflu8bebtegIe2Ei*%^BCWPZ#`(L9 zUU<;A($22G&h^b_udg|7=d1ZPDP#@^6`x3M3${ea(1sO1mX{^(>DIQI^3nh*E@w*D z&*rU|+Oe==lPm!T=6!451D?~=**814zr8a;&g_&Gw#>@@POIbU5WvLr_T3X>+pj^_ zs)N(wMUv11Zu$Q{g{%L%Gu|A&SG3N9kcdfEW>5es(xT7Y+M;e@GD{RuQA;}0^|N@! zV=Mz)^(SYJ=K$m6j$%?5GDzF`I9m)r24!m#^#QS3F8kUnhpYVpeqo&LP7d2x39w}# zKu$7uHL|5O?iAhOUDBUpkdr>fq^5AH;kh(N?DWsOBJQJKiOL7@n>)y(@?D&*i+`S% z_B?5)&`urA%#4VBvr~5VP@E#Yt8! zrpAvz*)cRfQr%??Em3%!qGRUy3Gi1GnrN)ImeRsa!D1g@S5ZGJ0|jZ^r~dyyUU^>E z^Lrd{AcbiSHWBx*^&5lViaAn^mX)VU!u&s5In7nL{B-BQ~1E7`B&V{~-A$ z>UMG0k?`AF^e^|S?~J533qSABxLbRI7R)Wii-Y-xOy@zSf0o6}NA-!xd~aMR)1Ac< zYA3k)#jlS5pui6W(S(V8R@1#Nr+Y+K@~W1RdIl~#701Qu25-!ffw{=&1ZH}rwhiAn zGw$Fu#b~;vk=P{D{MJ%mw{F2f)7|@zos+?Ai2y#fLKY@Skbg<}aggM@pmr|t&IY47 zG3ba%45{!iM2?9RNNAZJ9Xbm-=K#aXt|HC}?xG9M0DSqp>IK8s@m@ zNskbgdW z=rFbc@py^W^*w8l5|=7j|4$Jo;B~h^0zlfww789eIlgzo5WU;JD2_smKKIRes63Kc zD9?Y;Y%{FJ<2VznWT6mi%FISzKc&gvl{ElCi-nL%n2l{aM4l=VRauG_!XL@;~9`u#!QN(^_Usn+a75E598_~-qV3ZJp5Jryw< zljL5C&>ng_Ul@;A0?ZnhFjTB5B!xSPWtbZKzz%ewvCBum`7~qL;CZX)R{hJ!DsBwiI_^yiH#~o0- z$q0V0Hs@1y$z4g*$HJkpf_i@~P^5iiLZ$&U`Tf`aA?m@Pv-^}7R@}h$P246ExngaS zeZ`_tblC)E-MX`4dROFWC;gLYb|}g3-i*leP>Da=vjDxEiw7UiL?|ht--Lo8u?FglgR--O2@>}+*b*fYkVfIb=Mr*VM&2lCom@{ zuG><9mf!jm+piV!e+l2+@7X)yy&ref&E5wFsZLh9ZrY#j*TgxT>2>9U6j%Dj?%K?{ z-+tG%)gwzW3__~4oYl)M1o))QEMV)Hx{6x?-B>-;STbD>0+Y_&$HC0w{M~T1AVG;u-lD4PhTe&$@fbq&v3IsueD3v zEGpErF99g65((1z9+XYplC_E5hqX@{Aor)y9aM9+ zn>0%6Uf%HhWzDXfIVzuT-ea5RsZuKnw(};pJ?U3KODjt^1TVO|*_-1++m%;tW-)vxpP&DN*f%ZjT^o7N1}m%l`88Wck-V*L^G`WyI$|gZYpI;% zC^wSdoJN^Dy=3TeXQP>(bYhz2`|X_ngPgtF_O8}lGhtj6LyLVnExI-0@FK78h0n4f zz)072dBCRT7UO15pGR*=#y4LPJEUzu7 z-YpJAr%!%r_hE{zm@gk_eG@2%ugP~y!Y$?|q?Dau23TG&DzH7}L-oW5Xxl4$h56co z!O7vP*SYPVfWhD25BW!dr{CP)-Ri!LUi9a;tQ(l3_CV+%SA>{S+2({-d|Fw7FWz0| zpsZXtS`_SEN?6;es2GO1{&RYH6I;j?gFNo!v_`j+QmDh=xO;i}#sKlu`0t9_S6j#U z9~Fe6d$^{u&F5>0yxuc2IQL{x%1P5QDMp#>B}AFSF@GCInk~OMI_K=yQ7NorD3i6NSL2 ztEVrhni0j`*|MlJ4B*aSLjN(ls<~$Bgl)(GG%uBb=fNZUEo2^LuMk)NZ;el}G?7VJEl^b@({t;oCV!;ZP9X z7LRPpm-dH;=J0vL^0ca{!?9E`A2x91ZC}hHGKY&87sp5gX+DJJ(dB=7vr)Hb7RaD> zKsh@un?yg+x|0JAR0;NRLe7Z!%!$S!IJf(wg*mlk+=5{d*DZx^wL zCI5@G;c^V(=Z+j+;4ptssauN=8>}w#+_KFGY}g9GQ7 z-sH|C!9pc#m=b5~Ms~}_hI?S_;Yao_O>1+Fu2@O6ToBqaHj$OHs=PitGp*}E6gECX zHvhEOFHdqX1xrm2kKL-UvLPgHt{n$itGv$IKB2&dxZD{*)yP1DmN0qBoYY^f$j006@#l&Y2-Kpz4UP6chdll z40HRJa{y$h;A5}9ZBn;oYKzLBz32-4%2D*W=Qg@P7U{ynM|F7iA85=2IjKBWM*D|# zoj>oL*WRHQeqFSsNSP>ZbB&CLK$O5SbG_vz@Ag4n!_x!b2_m$4r^m$S9onGOkG$U_Fv~*H$qxeXc)@1{9qoLLSydAnpZvd|i_zLH&-ydyj%6hTMa|{iFgv z=Vc*o>{z2`)4-!~^=dEdX3{IlT-GIr+#yXHN{v?IC66p_^U_ZyhN`Z98>ia`Clr>i z#CZl^8P^YojJ>sx?Qc%Ipm1YXy{vri`ika*uN9dJr}%+&U(4eTEZb{FofS-!rG_qs zB^_&3cYAo<>nV)fV~DiuYPoYHaCK&1wPF6oWW&e5ePW+(_I$f#4V*vjF1_vfS>|}@ z(wCn}+8g6)34h-$JPFA2?lM+Fd>h$fwA`(p0gB_LCmb^kHiRF;`x3PGqcIS^yawOt znpO$g<3GhERTuj|Ul+kkz<1S0zcRfKoa76lM_7VmXtr9M(<>)zA?JdG!syV%=$XG| zJ7s05gkGM!GUC*c;HM>_qe+QzXTYTW-jB?vQSUc#+0BReX5-2)IvQ;`kQr}`lU^s= zj;3@L0AlS0{F(nX2)%*9alc$22tskwa?PeB9{Re*8RLDh)?NFzaT)4n(=X8ff809* z+5Nwi(J6o*|FMwO;oAt`-pvOHx+q{$A)X{Sp%0H+=G8HPAejtG5pt)1V+{#NHTe-m zg43X-byK!JIo(*iNP{#ptF{H@?H@N>*}{&8o~+G}>G8iu$&;;AD~daU@~Ec6v^;Re<+@!MUcsx~<_Iug#p$+8^W-bUKeWt>v`yBu2C0q!JBmRn z5SGbr(hu`M(;3fMyfku^%I)0)=n;x8yxYzr<(2H@Ku)K1+v(fItXkL%fV0tI!N=eX z*i36&Q+ST-u9eu6S8)xZyb4q$DClR~+(Y}>Lk&4e%JIwHz;n!PdjJQ*Ne72SMG7b8 z-IY!+kZUKs28_h#jJ!k>LApBq(eXE(r9w6rdZcoTFIKqQel!>-RE_<%=&<9|eL^-3 z{<`g*ZyU+(Q`G;IEdA(d;o%EU4kGYS@f>*;8kD2B^~)G^O{DXm30JxF^_%lifB%%W z4L!O0n0R1MmR31g{mGv6#Yg!yvk+v9bv*a?9iM$jDf@}XQY;?jLE$&|l`TDcc5-Ub z_c+tD=h5*+xBdg2>)Gj(Pqs*V7OZ1&b<9p)S`Kcl;?%B61Pm}s*PcIfZ`+U8CtJSk z?J~hS2K)yaTH|QBLB!}wej#U?5>vZz9hT`j9kX>_knNJU2d2Rc^zV6$7=Z!VXc_r6(VSE%UgyPxc{ zM$bxniPO!Ug0#ITNiZgROR;~rde%fg&PymC^4@1(w2N2Dtpe~FbAY@6*eJy+%ALq; zu|4^C_{}(7ZY~t)fovB_h_!B+F)|?aKDbapQx35aJ}a~xaUmWolncto5-ij5#_%}q zT+o_+0hGE8&4Vq_kGYUA*OJf%1Xq^YAh(X3_$Rz-t6H8EV9ary9_luigVVwb&*AEQbA3K1;>6tcdaJ?A0wyrfaq0t_~Lr>KMUi%mXCVeSFR;JP8%B;+!mX_R0Mx3r2XBPMK zng+&j5ciHW47d6G($(t=p(mRui$5D#dLcxXfq0dX3w{L$TPfoq9MsV;cGPe04mL6uIXIT}U z>VrO%ixt_hlOM~HEbA&y2QYIgBJAQ8{j2M2Ydq^6tDPlo*f@!$7eHf}?cDi!uuf#Y z{70%V;|ysEu73K}+v$3Hk^ytF`knRdBY~-YtAjV+>vA)dKs)6>prH5SZ&^tf#x_RF zV3U)o3roj(2V8N4L?>zZk)Yb@i~H9{qkhJ)*n!Bm(DD^G9Xn0*c<=Yfaz?%lpgP?+ z^Fx`3GsKa@llR?!AcdE|PGBn-`zPMQR&U+P`B>Z+-0a3_0^FvQ4#b1TU;bb0`&;H@ zw^qq0LGA-&B3XkaXL^_Ys2+1%z1L^j@8xdATRWlprO74{c2w;#BnzmJ)#peGt%AOz<)9DLuU7l`JSL6&jODVBfZ7fvS*E?r;p`JyNLBMv1Jyoem-o(1JQ?e8 z-?ygxSWht=W0M~*;OuXx`r0RmR|niGr+)PTj@U#Tc6#m?)g3;9aXb# z`X$T&vf?8Lf&*&_&t5TYlKn82{%mOs57poxE>H^I^dTQIakI(dMPqm_mJpP!+w#z7 zVAlYCr1m9X0S^zC@C?nPhAt#Wb!Bxa>3G9NwQg!PZI|S86q-CTf0iL7N#%1&C4laL zG(=743e-*9Uz>i!GHy5m&Zf`btAi}M`>fl@G>$g13yb1$#|vBeGk91;JA#*5ya@p2 zG7-0;152md3FHBDoW%Mt^ogWf5I@rrEVR@o`w1#9%Bu|mEQC_hC$BJ`@b=69b;78{ z*n7!}qe0gHfi9aHqJSj;a~1r-EbTgQEeyQlYg;aZe+rSIW)lgwvqgPR_73ul5QLSa zlTtJZAYuoU8i-K)o!JI=NGwxxb-yIiSd3?Kkxj$)Pu@3?8Y?oRb~)-F*Ew(q``!qa z5Jy)Fmw`ltpqbonOyeH+>=p%VMR2q2lxf+wla6nf{)CwZDU(LJ%-~kUXGuRN%Z<|Q zUU0T77uzN!;|}bXWX=vQqzbF$HW52c$caSD-%TC3UeXGWRYX4Yk`B=+{l!}gm1ava z=N9M>d*jI>I+TT|^u*Orvwe%Fs`B{~ht&(tW^nj2VztM*#nN~MEMN7nE7#N73C{wU zHjQvRvbXHP&8WVilFh&1CHrA^N9!wPJb0hiFC_zbuuGi|P0}Wgx`|5Un9h{UsCI@KSww4jdySCp0f^0CzIJ^_O>dB@Tj|vnBF1ma=Ivzh}UfBmd+O zi8I!qbqmm15lvosZ)gu|!d^^9;qGO?t(B8Da&Q=XyybSIlv)UKHlyArM>O?2DJ|RF zrl-?Z8m&{@kF!a&f2;+A^0h1;3?3KzSZr7PvVZw99lcc25c zpXScF+XKyl@=r0dQiQTV?k{XzZ2wlibPhdeTAW!qzn03Y5;=C`wc~y>`PRd@nX7D` zO+W{yAxizeU&OtX4gx%gIXHfc!yzLjM=0G|R;vO}2 zxX#umNkeoZt?D9vEgNkOt84qhc8VdglsBjFs?Kk&Zp(f@>7hNWx#=sq?;dHt1bAH0 zA4}Rjh3S=zsqGE;oy~#GG7jO-8;TJR#oo5HP57BREub7O4C_W^Ji7s=`AYR$8VE{f ze(z50L#aS#oa*bf6PkR#%#nZ;3j{mazEKhUg*+JTmp@z;w)rsq^h9yi!;~P<`XFUF zfL|2r@;=i_{BqyQE#pdq0riDZjb{%HuUsChDEY+w`l~XLv=gekJZAd(JwVtMOIUJ= zPx|*{*>VXSENkQ<9a8t!z?ceg*p+E-TF^YCC|6=%S0SV>mBE8Qy(o*(?kfd9RJwh3D^x ze%8G5Gi>?c`U~LnFQ4(`(vQP-U@Q9D6URcV!(I~M%3t@44>D;yyLqsF%ytD+$)k2Q z6)DZ*B}trPY2$3;lI7T@IqpmkLm>^`j;i`9ZYVp@Chx1}mE1}iB(D1%wN zFyGpQ?v4Vr>Q5oiGmr0D21Sy`Ic&LHHTVnV+hTUuGYx4dIZ)4QH(nB62a2xSUk9zG+Fi{HWHDFt&uuCt#)ZOG z)*S;HTpk`hB^>e)YCl1~Xo>;bH5{=j5wb0I?z)Wj2mXLS>*-Y5 zRs)#O*XU&C7{LXt*R8YRLT);2wGFVp&d@a^L9|CQ?DA870{C9?_Z;#Yv0D)`&A(T^ zBg;O7luuK=wu&1EzWu}6Hc5Df6#tze{E-Mx*ANCqw%?kG6rV-;Z++gL9c?*>gWiiK z0D6+*Z5Y$7Ve~T^6=5s@abXdxP)4q`Kx{j60N5$c$n8`+ zhc(w^fj9CC4n|K2bzaWNBNB%g<(x<9@MlR>?lgYa+Q%3hM2RBDU9XV)dp8~YL?+sm zk4+2EcyVbe2~U-fygijXW<00mw=1vzrd2gKmVtwqYg@O^Q{AR-*FM{Bx+iZDi4+aK z60u~L^h)ew@e`>cD;xzG93g$E$ol%L+jf!$yc_rzIbv zUqQ?|P%qwa%S*jLA@E-4z1KH=^q0n}qs$+8J4LYO)FY$^Z2j~)Opc^A{`c~$Wy*7; ztWE6zN7VhhC%$J|Y9`CCsh(>Uk6Cz)BRb+b>HJfoJ7_MH#e?vMOb2u&qZ z9pV{&Oa(LxM)^ozJ(Q#Tmyhx-s$9MRRu4-o<>=*@@3a5F{1<^zgxh4@BUd9Lu`1dnP9PO)pZ8*6-!>2RjBkGYRKZ-<^FbtVz!AJ=Y4Tav-_B}G``5?4aJG2+szyI&a0jWLz6orSp(YEzj}CPOu8pXs`3m^A&kaRrw0=>gp|i zWO}GXUatG`KzBFQA8YYlWx2hu?lD2ngy)9za(6<6*L9&tkMmofK_;@h@$H{?zTD*#n5GWsia zidwyu{MEX(v_EiQr|e9>R;7N(-lp?AbD@&C5P|$4u|d`@G#>EpjSAMRY9Cu+^NbZ4 z2DIK-JO=|PatyO{KARguH&NJJ*lRAMiC)4mxND4dwKg!e-~_mXG?UHM$+WE^M9Fg* z7?-p#%Kvcm736=?oX!YkNmw?SeUK8bGcJMqJA64lBHU^&**)51x{Os#=;>~8FLB*Y z+(2U+|8&G!?*{TEDlT4WEXV&(r7B4n6$X?{2uf_Ypa%yx9ltHGH88*T&am*7zQXwQ z_WkGq$KCBME7HRd*N0S#{c!cV&-Hg~lKha~ZGcZK?=bL=$}p>k`$7XmiikEXW?d;3 zZWmm2;fya@D_*bN7eJSL;YZn1ChT(jCH58zZi~%u$!bq*jx~N%+q?XhM|w3HcU7jB zX{mv~{QVM+)3)h7(}ma-Jpj&`HnLdBGGwS#Z#90Y|3%>G?5bUy`YW-7kaD44o_b9i zDngs_pAjP6v0oKc#RA9jrLGw`)~y&_^J6iS{meix4a#{?apmZ1Mt#tjz%HLhglV9P zCxD)u{$vWLQ~8x9N;iU!@1~ML+IxLL`>`7KS4lkNCC$B&pAFxXdaKo{mTrZ>l8(!p z&a2$0l11mzashz^>(0-P8{J{ zFmn*9t2Y(!2>B~ZAt2tpzT9_RNy8mB1BhEl8MPs8kT1Qdc-_$SDrVOx$2w34<^=I1 z)KS7p(sW=|;18yK;&8q!qCZ|u=Rc51)bS!PG^H76(E#&{(*RH_TJ-W%@H_BmUGD{x zj2SKkP6F{SzB#k2jXzKbq=N3hHSX?son#{_pWM@M3vhO|AayLGa7l3PO>L)S0B{%s z-RjpC3p4{uHPn`H{#Hstr-{lzU}SC$YP@NA1R6~NKH#EIVc<;M3L@&rT?N?g zl%-xgdHnX4q@UoszOV8S?3(vAL1xJq_!7X={uf1*=S^7R zmqV`WI|AW4gb|LU zMkJ)(0TZV&@y~$&juDMgTo*-e@yD)GqF+Cex1E@IMgrLmD5^U1orf-#Is<$R52{p- z&HnqHcWNSBUP9Lk>y}&d#DWC6WF3BaO+>oMybE#S%w+f~>wyU?zVUlgyVv=lb=^7c zF}z1ghL6Jcpb||1#-i<#rbmq2VuOmdDQz<~T8^LZzs>dsF$n^?I71qi5LU?wifNtT za+u2;^7!~8CRS%--*$Rff$MpkvQFNY(jN7-^p@Pga~Qaxw*QkNeq z`=-_R2DcnfhyedxE$IR1xUXP%_Z9JzA|onj=Gf+0*&<;omEChXJFzG_jneEuxJHg1 zvDOc#P&xa65;jK(5L8G_Xs_~>DBFj24-#9b-SyS{0IacQpv<5=^1r4^lzYd*Y$JBQLo|J@@WyovhPR_iD1d&JYRC7dCVPB&_m9^<5p5D%2JI||qVS7*2B)moU1UsFcpVG|hEv{!q z$j)DXl|gfGs>P@L2fFDSZVyW3ttts^{k!^kWFq24(ldL)8PK@92&t0MGrx!`J37^Zh`R^G_{5<#TVEO-naAWcx8DjHuI|&RhD`%Ar&r7ILqeG`hWM z>&UJ(`c(8M+Efud{ifUP-gMpblMy{yi);UxPATz;^adzi`zO3l_ko2{qs^0GA)E(a zQ?kR<5Z4oz)26&d>GEmeCfs9V8@pZ}UcmJ>+@?N^xcbbXTQWpZyBP4fokap+EK72) z#&}rrG+N|ky3BxRf66N=-jWbY!9ewqG?u{eBx9wNm@uM1Kg1g({|k__c!yydhCnE; z7zEa4A0R?%Ab9=$NbB)Q+)dZy$QTR@G;jou&QXcb6L)Hn?gnt_VByyN{M==wu4(#n z2EsHw-_x&K;O{bMpm#+EG~lkwSrI*%QT-bR$`X`F2`b1c9UjKIHjGJGx> zo8;*9EFTjkk`Ce>Ft$$M2sBXF_gY3rrgP}D`;R=We0^tZ?q|wN2ezIm1~R}uuUnjs zSb?8jqZhNLE`q^JheqY!U-)vb-c>9dI3FthQeWN9X;Aq~rr62D1=4e8|0fiP|aCvr7&PlQkMVls>PXR6K+%M zaJTmN<;gX=wf46ZNprP)!DHaCI<({NV5VnyAmjl)-*NQ%&sZ6(WYMpm?1}esAN)(Y z-b%y!q*?@kPHnPda@49`IGEt?ctykPg+c0qcvAjqB6{B+ii zbY5mzp@)i##Xs2=F_UqacNge{dHzveDe8JV=Lw}G;jw#syoa_DH%oLFlLNbViguR9 zVpx~*x+^Up_0m*al$zdvv9p<}`~?9)(Igm1Mv)7?n8Vs_hALn^(*+`~8xdsm%{2hk z9$8~fHTa)+yH_N*3J?zjj#=0!9x3#Ld{J}(AO$j}gHY=3eEH$Sx=>cSU?CphQjkb4 zZCMgaeEF21b(~B%Td}x-lNBWbWdJYN<#Qd57Tz@l;YPl`vEsb|L}XxD#6F7`!WHbc zi=F}l%AZ&e?~u3ovp)O*o;BN(q^zIOr!tw(%ci*GL#v8Rn=i6)r?)H~KE!7y9Ge`%j9sdwpn^SY`8% z>=&$KAcVSk*MS1y1-ki)apI{xMe+0l)f8_GB@P|nc#CJu0OxD+kuG;UG=G|>@PS8@@u3o=q?7nkT z;Xe>thT2ro{KpKF3u|g)23BloxY91I+hn2^POpm7gwo>Wzo-4WQfV-Q6t{nM z7XNwZ9y3be51WSJr9T zf#MPX9`6@OCR%W7OCvaR1oD|^zjk2D13z>SknwsYb^nxsdM$_)(9_Gvvc>Aiy`a0- zs+Gtz8+(a5iQ#OTnU93E=K;!D)(8d;>oOw&uQI!eJyYbuFYMrGncJPYt7(l-l~dtB z%yt6b%rTn|osC`9=0C3pt?<)xt_#*`%$dE2AqOc#e|cC+q&158TkC-cTzT^I{@R14 z?2g~-c1T44MTfbaEAowOe~H$Id$CSIk8I32YW^-i)mu@wP;l~yAwezFJx+sO%htdp zcS&3m8hC9ahz4{so)aixQsLVZD0%0ZKOnq9evzvn^ipgdN9PEf{Brw)PHy7AA-l+~ zag49EAQ(6l6F=cp3>zU$#Ef;0uNvK&G0`|RC(AOFoIOrn-PTBw$B@Y*FUm2+2E2RD zXajZUV*<`-E|bF&iwV_?Aqg@-IdIS8Un*X>Tw}gd5wam7p|-vxT$aFP))va%uP~M+ z`BHPFXDUFg=KHc=yp-hV#oR54x5kZ#q=(yz zvqmVG^htLG>d-4d$5uEmVFm4zx{e!s&9#PVK+PwiuJ=NNqf3XC=i`fHL5cDiIcDj7 zJ`C2R8)_?|$7bMYIUEn)Tf~_g7mI7t`I^<@&D%v^-0*he1DQdLe~jMmxn>5R86#+^ z;pX@WED40>)jWQaNs2$L$Z{B;((AD##q^YqLA5kH(cCd51qwYoAo-%j#B9tMotI+16Hqu*9KyZEa?un*t!N8AcDp@@OLbOSn z7|l0I06g~|V)JvgDTb*7{~iWhjqH1cZ2kbWnx1nXulHt8@y*iUnS3A4?_ToS(&U!h z4#?oX!*OiF!vL#a8r$(pRlcMz$>ryGTTIW#Ty)yF*XIB)9C_XIOxnOBG8weZZ+-m+t#!HdV2+) zDMAK2XB8cqn7vfCd%@0ac{k+{-70@eS`*8U4Vm~XVvZ_Wefo`+T`13S(41){0Cf42 zC7pj|h&FgjQxCjV*7=^w&jt7u!h&45g)s}j9TZCl^JTg$O)Z4jsKU}r6sz=AYW8^vfd7V!nfYa4yW>rU;JFq*=W zxtl>opKD;he=cSbP-i+XfSJl`D`Ku5F^E_r^Nu3SPwm+H46`g#EkX8kEWhJ7&xg?-Afok z)8ppHVf%Gp)EJ1nHY?xKqNH%?gWh?{`n6E-WO9W<@4EIm(vL=|txm}j zwbS&@fOlu8TV>ysu9!0>^T%(G({$g)#cGUHC47m|d~@+eUs@+fW9Fy6D3zHNe{?6R`Eo=TJ<6_w6ZCNMz$q6uy*#_v{Q&iasUB{>V^M)}HEvJOi zmll_`*!JXo^us1T8emwZQX^y(#tG?}-NMY$2mH&D-`qUME1BqkEI>&+8pRY<6V($u zD`5caQt7s~KJtS}7kb;!ysx*ITB-253mqY&(>5F)ZrzejcA^UQ7C{%&B&Rd*OoCEa zZH;(KO45Seb3x>pqGkvVWuMB;C&t~M(Ezm^8Ttc_1k5eb-C(?vJN^mU58P~dEkwy? z<9T+g<<7BNW#DkLy4QeNgz32?MOnp#r{Wj{bI$IXf^}q6A_SY~XgR=dX*Zaq{y7%0 znA40Z!bjDLCOV40PI_b+Z)Ohra!zfO)p;a8UI*uoxA0Gick{4U8Jg;_>_#RL-86x0 z<7urK@@m0yeV*1awN=!3rt3b+=3u^Ft!8wIlUKIWCu7R6o9O6%(p;eJgdTCn_5%j% zqo_(hS%sI~_8-%x^jbmHju{DQz>h*dO2N!dplc&y<-xfkiQCa5fO4gi;6E%ZE1u(l zJsN&_7wR6rlw+1@7@|_pqd}1UkjwyyuUV@Qyfz`S_+C!#sS01D*m$1(%@VnVyt;RH zz;9s$pYI2+8Ej|cNUl6azN2F(5y4ceRHuX+#J~pkTOL0rrtZNVY^WuMRn^8@YknY# zLGysq?p?c(Gpsya&0K0xdlivmJrWxMit21#R!zn3?rH<}p`pzs8?pYfNTGg;>4 zlX%~dMTv+D*T>#_Q5F;PxHuno6%4wLU9SKYA;aySEAbQHp1(EJiwbx>4Duw?0~;E7kFXw&eofxCdGuS5q|~ZG8jqa4(gf!x014x2^kGJJA|w z`?m0(&lP-*_li@Y*BTVW6U|#jX--QC8scFy>#hz{Js;X-&pfVjKlvcOfz(W|g1`e; zQ0=5ZQw`5C0)%w)Hj5S~Axg+EIW0P=5>`zFxu!Z;fRBN0DAbeeDm`iP=J6mGlupxT zF$=vysLY88J>ctHVL_cJdNJyo_U5fxymv2HqI?$RK6D$f*)o3>oHd{83u_fkE!Eba z@_Xvxxg*hZ#fn5gF*wcd2^`zLzQg!&d^uEPlvmE&7zOqt=W}{u`*e#+=Xq@gKQ1O_ zq*y!Ln>jM$u6O*}@a<@59vZl*V^|2Tw(jJ}uUc5~UQ8se2%Qr7aoMQLogJ|ZXM;qF z_eB;zxrm0!-e55=xoZ%xJAkatMqGNtoyuya7J*{uz8QS&QuMcDp`nk4)hw#Bi9wF~ zNi$b-uT0*v?#S40sA|FstP8&3LW6NWE>>GoFM|F9vGmQEs=KtLJbm=>&dg57z%m0U zWjxPY?gISI+5+}+hOGL?nJe6a75fLYE0USfd1J+m!*eOBU48~c#rVkkl0fcbyUS1W z&D4$O&ycfya^I+(pK_1yO?NMLd)^m%o5-!(32b!8mDu=aQcp@~V`?o)k>kh6?W<5+ zwzI0RZEQ7Ur*N*R@0awopHACd>VukgM}TE)--RvIrycz`@9$-`%D?svmT}C@sWLk@ z^c%^6hjvo0zFby`dw1v8ohkO!dpdR|7heeM+TuX_8cF?l_on_2A_mHAhCNC#257qa zANE6~_Hh)+XL79YI1}7|!e{J;Ny&l6n9n+BUf;?-jIwTMh@X|K^c{LopUa}r@nqHZ ze3gkOJQEcs1w~tDE_|fM)9?NVf`m%79}JLIBCoW>Z%f*xD%W4TAc@qjy5Z75sjQbW z71_W0UB2ktcR_tA=ytSo371j#&A!f!2RJXpMwBBf$Tez@W@Zko!_7=#MFIXMIT2o9 z#0E-XHQJ6b)tn_Euyqa-0dPtJCo|c|V50$V^{pti+iBC8>;!2DytrpSp92USMAhW` z?-dIpOV{xpxB;k@=Y(aFF|#9s&gL;21gZ8AfaELaGKA9Nt+ob7g1Z0ga9$ zUMp9cinS5fmob1%Pg!HAgM(M3_sXB%JSZNW9@U+{-*4*fGAozSp-V!*UX)tdC$L!= z5X1VMR(KIoVrhFsVT8th$;?>;b0_PJi-r(V<-?)kJ2x~u7UQILiu%_$o|eE`9OrZ5 zdz(I^oJuw9`O)#LYegb%CfN;ddUTTf?Sd*rBJ?e4vEymj+myP7=3ITQsbl*xugOOt z7KP>RTh&+UP6#jHSF9*sl(`5NpRiKSZ-(?pUy71-_ZjQN8f-yLy=8xYW4)?8uV(KM zYn?9!#|nUxyt6~hWU4w60J1=qJEiy*S@`R$bs+{l;*V2Eu=(s~s?M6?0!bKcBu*iH z)HN3g@6U$zGfQ23(bp(7i5@Ff+z0o~2QSr+hd~0Z)OtB1WbLYshFdaM zL^BQKDl$isJty8!z<<`Mr%?>I2Ph{8pOHvG z|CuOZYV`zU2fZ`MH4`3gU-nFv&`K9Qk81c-|1$UT-us*E(`L10QcueOJG#*tB(sO% znQJ8f>a_Uq{J19cFEYj+@FNi1z2A$NwS^>oXRM_jJ}%Nx>@9t!yS4QZ`t_&hxgLY} z*s-j+|3DWf@2xt@dI7*pN3PDgKPiiE#NJB&0sq-8ionp3_BXdPVmHQbZ`nVu`?L0n z$7afW9+qL6l-aTbnFQ3T3(vH0s~rUXkl{3=j5|?sr~NEeDTa;L?3r--Xj{>9C{5Ar z#nNXtiH8lhfr1>{s&{)bUr32$(E90t`DwG;idsgj#^t$Ot|(5U>)XRw&O7Uw>y44K z49g06{i$Hjdxro4k@ZiUT#za6EsERF+5>smImef%{Tzjd8uPnA+PA`)d+-@@?$3E2 z$LW+pqw9$|?AKe~AZ+X#UHwS8pB1^8sQ99Hhn{VxQsax`YhOHxj;Moxu-(?66FCW~ zw+vdZbN&PEt@LM6xmk7Q?s*!=aMyLt(u5(Vn?z8;VZHy5%XDxVg`NlX5lW>^ zU*8Hz50cqQnm=M2YlkZG2TVm?Y5K}YvvZXDK>c;`B>2n6ntHz;>OZ#%-lRvEt8xRa zgbS2vp%bU=s;mfN2mAHr=bfZ%9E0Vz(|Yb0Wu)BnaMuyxqVz`H(p^JeDTZwInj)#i zNA5&WkTc6ROj9kwUVobMXAO=IX8@^o^o;&up@FLS9qMU0L@%Xh zRX&i{W3I7g8EWMm*ZddgG|=|?>Ja!_=@>n<5>i$44 z_x=0b{El(#!LDbz#5|70sFRc7^VYCM-S9JpxszY3w?4;6vWK0E7=M(Tf8r@vp~=6^ z_RQdO+I1vmF_ZYp0iogMI!3T`sDvAD-V;`~f)2xBf{p{=xhe4@b?cRz}?( zdR(6vyx}w3ypty;%}W#Y4QEnAv9Egj??TGRVi7)Z zdCu7#LJ0?|Lv0^S6WavZZh(cEA5E{Ef4pUFGTgb))NE?{yYhLthi!X+Jpw%4=-Q=f z4`#v32yOW|etkLf<-_oe%O7Lt(FzHmA1r>V!QeE?XYTkLr-`#?&F_gQ_4yAet|&Z` zT(ay7`pR4VuBvdWBc$Q&ZhgxesMV_noC}Soyx7)#7_#dC@gIo8>-}Ly$44P2Pf@RA z$b(ZpXzkzVN&q9FHm~;f>%TYd&nb0pPOjH^M>@#J-qY4_J;T}A8#);f3He7>*o(#XWV?wK~ZV&1uwC1+w8kX*RweSTJ|cmV&C3? zIEF?COF!JL;{5DLy1~=V^O>M{Veg%?_D(syx65J1mM;YT=t2wP>&pov|F9rF>0N=3 z9U{eaOFZqb?xb&qd%>Y@+mpJR4??Sh;q$PBspvIDA1Bl&=Zsmyx+u=j_V$lTW=&te zq#kMfyek@XHq>Qzm7rud#Pa$9Q}5ip3szAu&|B*`;qVR3smOZuIB(GllXlzd`?EW; zU9SfoV@XT)fAy5Qx4yPMjg#!wxGvi*QEO+xEg(6}SQVOQSa%qQ{8&$1Mc?~Ud9FMB zNto8#A+;?BbkbJ2$mJ$V@YO$DhsNzzu_vnLmV6Bxt?qIiPh8Cf08gUsvRv1N*uPrk zhvRore)JDvTXs%o<9MnAexP$gDhGH%GxWc16XKYxcnVM=P7-d!*X7A~Y7k;G_SSPq zwYE(YtnT15m)T}@y+r-P#GRwGhL8ZYe_A~=Qk0--wovTtRDf7HT>l*s;CrsiMV^_~ zufCf?ufP>PqA5S&J%reA^N;p*y^aXYP<(dKwUAs5dRDtlK3HtZyLkBc*^SEU&0o60 zufBR4j=Hd*hg$5|qc+qy##B45M-JQ@7dJ)3 z)zmiJb-2^G%eePsm*!k@Va}M_#5=4pH}}-w@*oufX^~NiHV_TUi5{S&U5o>!yp*LH zT8|{&`HR01f%P<Y5HzLP|3qD`$74TKYRtDaXUXtpON~%J<1?3@Y6?>%>l4CK{!T2b+gp6* z(YEcns@dkR!S+puDbQZ`iRja${zh5V2(wB5VO_nf1{{F*Fg+V~hJCM){oK4S&2j8) z?|F!@+8CGDH0UN9t`af(krGE9OC9uL2H&+V-b z-t=Tl<^@1}qmt*fK89PlgK?bFtz{qx$1C;n!z$}H z(FPzgE5gM!gvi#0R^&-d5;$nU79ugInj1f}Omr>AWGu6B1BYRf6Oh#p- zr5(|yJG#Ep8gNzI7i@%sc&{%+KhS;0z|wTG&S&t+@FgU_N@xT>`Y31^kLGu4+aGAY zBY&-I7*ArCcYVitJch5)6Yg|hpifo%!^W*dBqUxMhOh7#-L5$3r#z;V0@vD`iAT+ImsK#X=0}h zc!R7lBq=_?K>5gOOWzbUe@@Bp%!Ldnz3dWPxGW`T>i4mgVRt?Z)K|=C#25Q=X4eWM z$SS4zvWP4NUmqD%fIy~((-ja>3Hz|=GteL+Gzf_-Wwa9nL*mn+a#IO5mZ=j15^u(Z z)d+M@LM_dh9NX+#y$&)h{F9R@UNOBjvWN(ZVCqzd6?nKW0Ch)~j%4Hg}MnG1(fwH)1_wdb$iT*?U{ zwV5Or2afYsUIwb9*Uy`uzGH~S!1;m1d{SF+fYIu1giIEh%%10XAbwG@qwMNODUx%z zmB(=HpMv5pYWzqC1Jk@n#4?b>i5)%~B6qQPCu4Icc*L?79LnYj^`DN32>mO)PM|{{ zWZDL4KQ{(3FeBrp_S}3 zZGlaj>r}_{xOhVY}61~c# zyT~Q;{S9rO;^b-&J1B8ZK#gx7VrCpF_X26r!oe)xr zjT6Xo9LpYQt19(Lw)bW6md|w7?lYH7h>_Lw$M_JT*B!jb8f@wAlq~EoLE>!|#6Ryi zXat-kt%3t6R=4Hou*p`bcBaeo2w-Dm+s-wL?bGzl2@6lGOoVaA+ z{)z9@?0)43cU@KSu}WOI8Uqk{gKLM)h@)9iMiz1oNgJYz(nW?nxzVSy|7Y*cYeV?_ z_3U(iN$XxP@W%A3I9Ud*Bpc%Zjj(H^!SXMiFzX` z{^(@a^5}*Bg_I4Mn~fJ=En==TBtMon$}Qfu_|E&w9Gj|!HQ7b<^1Ew3Q@_ic?th#0 zU=y?ip%hcR|)Ss6>3@jM37_|%S56$&X=^8x@$p~{1bT`PoeJz^s%Y*Fe^ZTc%4r?hvUiWzr8XG4Y*-|P)Z#T|wbTPo zEJo+pvIfj0?01UIr)jU+Jy(x5mXBb*0)jPd#-HJWZyJXRdx4=S`}*0CmpEXHID zB^$#=GIcDM+$kmrD-w{zGu`;gST;qCa2hB<+g?rP$XylM{4GbA*NCR1;Lhez(%q{} zbwW<=`%VGmeW%(0wIUJJC6C{TF#J2(-~yJ%=Azucqss5McILov;nl7J5p<~zrqiLQYL>-6 zQH=HS?P@_c%UGq|K3E>H6%Q~{ToQ)VOUOWweyAZTVy}V~LwJ(|QoWBF9GVO{&!F{ z_0oD~egd|Zu-sv(`YQ~KnJ**>u3L%_G#|u)(yb%XU6HL*qe_dXq0bu%lX|1|a>N(eP)Ce3iD`9?PP8ybJ> z&g0FzK37SSi7>3qVSP0Yw2Zx7ULbfU=L#uylpU*GZOxuyWT}xv7NCTqtXB_ z%V*(f*4fm4l{14q91a()6kFBiy%Zq3-Yx2Ak(#!{5j|aWZ9-Jx?@EWqrX_(jpEb_! zr!1;O*IclIvM2dbHVY&o&49_-{nHG3+Izf5x^dIc{F@ADFLC}TfRVaO`j><4?P1Jg zWjN$JI+Wt$K)F5#!AmfeyK^sB>PA-&Etm?w-QXL}s@~F2!u*8(SO7P|6Sl@8; zhpWJ*X0MxjgZ22mm~XZZ6ezqS}(fLhv-6`>G6CFCZIlkV3OTQ}n)-wy1P z?uFu=N1pzbO~wQlH5gk{aP6zNbt7$Ba;(@a$M8VcK&SM6L8FF6xGk%*Ot=MvC3Wxh_Wt z=?E*#?YZ8aB`+}7V0p&^OzyV4BlJYo>3D3BWtL2gv3u>_sb0tWbNQ#18i_h*5r;vj}w;<=JrRqX3W%ho$d3Q?j=3Hmk`em^&D&5Wm8B;;HLQR=$lF9-xU3( zII{0bx>#2GuiesPW7YYRCC&6L_&hk(-)yIyEgHm4u3)D$#Pvz2BxdB#JS`D1fJ=?@ zKME?}&&wtKx>}W#ugCoo4;1gz?_{@HPj}zv)a8CGFO;yeGNv^ z+8RBZelr)iY669pf**N+Xf+X@%GwPbk#fti)?W*F$*=6`$XgTPeOOhhp2IKsPz`G+ lEiNR*FDxSTZ(o8zKtMoDL`+LULMy}yVio!y9)EfO6!)2oXmSxgSfQ~OsQdW`yRel<7y%*8BU(E8$4@voxwu6j zqGIAu3B_kh$||aA>U#PHhDOFFFu1jit)0DtBf`_m+sD_>KRn`XWK?uaY)WccdPZhe zc203gX<2ziWmR=kb4zPmdq?NHPksFZgG0k3qcgK}^9zeh%PX5(+dI2^``-=@&(1F{ zudZ)?-roHiE(`$He}(n;{l5bHzu}_ziwhGQ8w(rn-*91I`usI43Tzx!L0n3C9Xv~S zDmI}|d}@W{qQ+hVu(0kApq0mGLK=3F&Btf|2JOEf`}YA0`+tP&{{Z%X;#vU^V`2P# zcvutwS->S(!pCz}udly+{s2CoEe->({s0EQ-m*)aK6L#7eEWH{LkQq3j{RTu-&ljY zKY;Dm_bH4%zrXweT)y!-#|aERYyYiA`gi4A{{Ze}ByKT72sw-Ym;E=^K<4Am)jt6C zDDyvn_TUF*#y6p_UfsL=0i-C_{Q<0g{GEvWAXkVP%lR*4T<&rNO44k07StY2MYC_H zY8`czeHUu(DTDK4hg?MG zTdJm;U`C3v(%rr^50_!AD3bm)q8D>#!^6?|`+nM%cy{7U_ukjO6~6aBe-AV$;hom# zsv~})3@1KhxiCD?jE)!#@&9nf4)S_Ro-y3|Hqhjv`CCr$Gj@#)= zq2#47o`bUXIf`_WrZGb6-&+c$L8t}iYeWC=#&_o?h~8{K>Ot~CnN>h>K2@`f=Kw#? z+LIqIqiy~8TrA%_>?Sp1o%YoHD!_6$l8BlGBSq;bl_UW{AzQUpY~#Oc!9$Ho{*a*% zxP$5Nn%-xm>K#)$**%Zb#A@5fuE4Dp?&(zi9E4d3*BU;+&?zu~UKaLReE$!fcFwb<=Zn&(ofb=1~ zLr7IIQ5^QOo}=pCYfB}|h1c=lo}B3q07?4K>O2fZ2ZLHRx1ffM*VkQ1#IHN7f?3Tf zR>0g7Ovhj1;FyJNUGLOJdv@EI{I+t!z$=idWz(Or_I-3^4AzH@L+w{bMj7XO2T@5~ zz2E3UW%^Rh`c!}EubGv;o_>M>(Azl!UE7e9aB+%8o~y`g$ExGMInH~p&c|lZonx1w zr6H)Ss&godA!n)O%~x^IL_j^MS)8e-~zY-4P`_G&pz{St}{-8Do=N+luxr zbrtz5(*3A8HIq&n2(R_7AM5@RHoV{!7`N2`>_QojqfbnG_}};0VU*vd0J42@;oNv zd1y}Z$!9f5aS)rn@xpG~JJk_mWjt?J#2>(7McLd8C#6bvp!$Ug5pN2}Wz1{45KW=t z4jsyI|3*S1?`^Lp#haVIsRx2HawUjh0#%D_rFKJ(bXULGge5n2H&48xf6hbJV)MeA zolh&eB|C{9{wfVhvL?#ScB7Y){ABJ*cBYlRwb6uh7;)MqvS`ny6Rj0+->=tEGu~(E zg?twEO}sqVZcX+prA2T6q|y0e3as4J{Vrb)o@^N0bh0eg(AXcXa~5oJUjF@m%eu^_%ZGG)EGUDT2eH zOTQg8#q>X)R#@UTSFD;3%uWzf%$^4T~^#q#8d^*2&X@=iQgd9slNCW?s`{`8;PH5L2zjCaD^3ESAs-V6r@6_oj) z1Bx9VBeM=j28E`Z)1iL=*8ciOZ5tv&Pxgo|(4dr#@BZ5>Eqm4yPM&UqPISMWaZ==y zdRz=;B1owhwCamZotz*8?|P{#aeo)4NWb-jA)}}TVl3Q#2$QojpN}mMIqtchoL-uq zNp~IdLuuFQq?OFx#P8&fZ=Z-req(irc%quNnGMA{LZ)XZ^Gw@(XkC!&cY0k3e{f{X zeTb$7jyTeLo|&n$>vj5Z0NhOnRd_ct2F~ zyulUgB4F#Hn!BXk23uQjqFLokx7faj9NQ$=8sCc zc@!u^W^PuZ<0(6f%6Io1%Z`{30tU{z8P0Z4Tn9fYk`Q zi@|9z#g#c0tgFcUE#nh@L`N*ky_?;S610%qnC3XorTIRp5)8|KW(58G_$TQ%x{Lf8 zLA^XJLl&vtH22t^=WSm)%f^`Y>iEXXr3w}Lel~DKtwzR-w#6m{`2d9>MP&# ztE?_(R*=LmsIRF#c@~!kx_o13KvS$wejZGD zr;;qbwbiws&(G#Yj#Dl3w0yho9r>ys$XbPn1NhN`GHy|0mG+k z5_#b)Z>+Z1>F!Med1$VAh`w#FXGQN&>jrxVDp@Z%J)+UP&KgzWzE7#7C6*pAf1g&W z@}dFX($zXnGK6Dx1!)-Tp|O8OLiyb3<67YEe*N8)TfodJ#^Xl43f4S}V7RwOOB zQi|ebQs%J@t8WRjEVdiLRJY;BS?79}-TaevrSwh{@{?NX@m->PFC}|Nn7>I=Rvo3a zC8}QMs^)ItM$obMueV?+j^zaW6AkMhe8ayQzrr;tq$FAzLqu^P_t_Q)a4L-_|KJ!n zexfRVH`7O!C-f;>9F#xL+mLeHa$gdrq&Yze6n#7LEBvamPh^I2jZ9lz`LW(EcC7cH zs@zdmPUMUDBXCI1cSbquh^JAGk;;3DmHaQKYqoo3=|I zpO{vOPl&GI`l!wk-}wjRwB72$6_X1Fs!jQx49{oJCH%x{pl(#$+o=wb4w&5zcjw(F z%4sqH%I$<6>y6tN8vNo^5QpbL>RFTaOD_YDuJX#8y3tx5g>ftuM$HoayFWdZv73 z246w19mtnlB+I)6M|D$FLUmNK%p97UYm8Wukqn>(gUM`O_Bt`nZ^5<*hM*FG)|fClJMd_;k+pdMl0+vLA#r|4+`=sajzI9)^V1+_8_y=xit~8 zeiPz??Nfme!;%^F)99f}GQyKSdiDJMFYeK}3xy#ZhUx{?_7pKm4`pAObV!%b`t!XX zby@ohajA_>*GwiP#L0l66G@Bs2j2l9pPZ$td@4{>*qVq3Szk9rLCY%bmxdqrrGh(E ze*lC<`x7oars&K#ae-#*AnF=XHD3+792}oE3#@^g#uJj~rR$HIw`@&Z%-&S74$Fjg zbEjWU*hy&Oaq8nGf+${eh8a?>bjW=@AAguV{_arx)GAP0v#r20i;IDj`IN0d1DQTB zK^JSNAm4i^L4RyHyW2(Vd_s%}rq*&3|E^L|hPWGum@W`JUKd7*foxmC=O*1}_8`<0 zu{?_Z|8olld$j{}8#s@rELYHjr40l2N*v2+w<`N!oTEIG zZa#9nriQ6BeVuS$Eqijo!rzAl%+xD#)=jm4{0?Ji0hMB7FuYw{E6<*9Nr;EN)dCZC zp}a5g15w?B@9mupSs8Xba-)CWG~C_IrzN^b^DD0&J?P333P8G+ogz zoU|AWdhtL^TQ$8Em>V023EQ35^gf(QR=rVlL%=GhiXGCW>slTs8Z*k$#4^WTS9E+m zEDCHt)0pm3W~-GKFn={`0&pfyoNQ%nH{jR>&8=(gva0+6z>4=!89cC6oI{B}02=P^ z0xYh}jE=sWulW>ED1$7^oiKr*w8-?>B{m_!5DBYVMe+=mCNei|e~*=&P!twwzAQ%H z!uwQ%p^+dogdv#wSu`3`zV`?@@9cFs@O(FX7<^Hl%x^=-mf9uS(nnOZzRi>CsyDS@ zk#P>Cl7bKK)UHcDh4x^LGiqPfd~MVu%&q*?f)~a7r7qKOy6l2RHDq<#i}o7AA6HT~ ztA;btr7GAO`YJo{;`nZ11T_=hjeN;40!FECB#CcqB%4Df`e7fE-ZngbhGESdz!c$A zt^19S__yK2rk2$3h)dsrUa!R|mWnI-X=dJZYU|md@OoVA6g>1}JIx}0Xgi`83#i-X z)sg*pHHw9a=$a`#Udg!)x>NHhmb3hy=??zIBLo=NH)tUdMcw>bzvl4n|`#^v8OT%PI9Z2yy(@(e1w!N zfHe{;rP2HWTp4;Kms+V2`&!^1^<39dkYy5X|C|iQFxZUGmK&O0XixpsBgrygwfvbD zxEUb`rGu1_G;UyoKga0+<#Qi5ks1>xADOgyDzj%?-_ov-I4Zg{gDl7#k<=R&+aGXV z%8?4-19R=3SsB08XGDL^X^BaF3n1;hs{R~lff>G-YE%p7?zF+n=aI}=$`MZ+`n@<) zgFiUc;la()nUwc2-55KC;7IAyr(!n&zav8(t>Z6l+~RXg=|ZU}Qil>KRcP`ARPMSI ztK7GO-v?vbz<8k1K;AW-o4a1*Qa=3XNL8gCVyk5YUJ*Xhb(M6FRRWzYpyUP8>AF9PzR}9)yT_MrI?@S-cliYjh2B1;(nAs?%!r0s#*0P2#XBrl@Nh+|^ zQ_^|TEiytCk~d0K@9Ei_DmUdFG{plXNNF}{{_Ei|PW_XS~m&1gW?)oHl)LM?k!$3<~cq0})L zd{#|0x|kgC>dO@EdXbhv_kgE@5tLhba56Wz!hA4#JoJkj5$T8Js~S3&$?WP|z3M4f z?J*g*Z)9-&-7Kv4V4iKyXz{6zE16dD_sC}$Z4&957bDECV9Vg_7XFIDUi6Se*ADhm zqHU7od%oY&N27IC=eG72No24~f-6(>^MZ8L4mGhUh*^EF^%rgPHQM$uDRxY}i|f^U z5$GWg>WUgpZ5YY=Pvimqwex>c$9}+T?D(ysmyZ2%^?;2WKwgtk>fAYx?y&RcQX6Jr zjS`vGtp1coP)KLF%AP}I1jer^;R1O*j-pbX3U<@*3|W3h*xkrlCZ(FSgo*jSx#(=R0B8$P#((Bh<<^Q(u3On zsA1^*EvE2o6XvCq!SJN=8b>CA$u19jkcYx)bgI%Asi<5?9sxO=S7JD%Q+pC?hgRV?7qKrh!D$H5)DU0^gw0hbJJj9mYA`I-SqP<7FFdY165^(>@ z8;^S$Yg+#-o<0HBag7BJCkkMjG#y3D_f@mXc}H294!L*l zZ}eQ80>&=5f%9qiBP`q5iE(&&aAH4WkUIZUS-axy&i0e$6}j_c>@<+I7o?0Z360ur zsIm5(UCSz;19(lSmlUlIh(GiWuw8SsBTN|a^Jp<{s^?6wEe8eH3@mUZtRJ|?vGa}H zUreyQR9eS1-pR+62;^Qf)lXyhx<8D!4kjBHUWC}L(IDb?n0#xN-xVQv*vkAuo0uS>#C4+uoU!mkbHFPN%jT9UX1C>o63^TgZJbtDg9l_T^Q~EajsFx*NgXQVxe6 z=q2^u3vH-K;t=pjwCh%ng-_k<0dIu>JeNXp$=CzhyiUdh`9MlD7|@w~y;WyTlZlbB z!Eaq)4;r;*`MJA#qR|0J@6tmFw~~aIEPiKNd|?-Y!u?gMuR*&?Fh|aeGb@dM`Z7Vx z)6dJGRElkwiDM*L&9G-%I>9hwnZxLLn&j}{=B#w$C3C6lj2>~fPP)-KwBBV;r6iL$*pl{ujAg4P2(8fzXh$T z|D>AQzi|M`H!fm3FNt+O+T*3?H9b6H$i2a82=FL_yge1Fp+NB!R-Nf z)~bS-7#p1UF^zZJ3f;KxdA&mlF$&a2RWyYay>}*M>5b3bXL7o6wE{y5tvzXL5&Uru|xdrX@Z})vDuAc%hk9 z^E~x#A&uqjrzE^HyYA;=F@>o(09;VKyOGPt>UkCW)mC$1ui~sUTH%Wi{{lR}bisN4 z(YC?{dB?rucYM{}7w(h72Jn?ei>K1a0E0@}K&(#YN_xA+u2a@2f;2scXwko%56?4sX0}@0U-NQr3itLB||3 z9Y424uTxafq12(M;O1G#*XQubn(i%AdzdO6waxTvtpH@nIXb5*Cc;P#GgpvhNc_8| zUVG-~9MCN5d{g-6;wVxQ5^JQ%fqTJ&H@A)h3Rj_Nioc*4zYI5LEt((`} z&zQ3Ih-FnDKP`nuMs!9kym3PM@4l>N`DdwJ{;`t%FQP`HCZ6x;V18@xvehY>xwBM+ zyzW5DpY*D=;u)qyN}#6oO8t0$oeHqCLIAnoXLC3Xpr(#)#T97gTY za`NHBCDOR1X@AeQiZ?A`CYa*Z{i@1_?VeVN1&McL$o8fnx1vnfj&7-Br7Yt6r%T6? z+<~0WGWziPGIcyR@#-G}D@0TWp@yUiKL{YH5)>lDkIv9%9z7PJ$aw{-vN7P9Rf5&%DP&Y6N7U@KJ)rg%mHC5}hs*?4iA0Ulp11^>LaI~$=x7re*{vKy# zBBBKfQ)#2Su_)3O<8lHs zu+`MlvQ^Qkg;*xX&&99(AW<#xD7c7K(?3h9xOcpxy)ci^P#XdAy9sqy0}XZkgS=S2 z|`1v7s8lA1)yRUkh>yYD9Duh&WQ#Bk8JycvN^3k_xV)KkSn!Kqg%h$V!KZxrH3yPMy5#5(@om%sDmG{UvyeryVA#;u ztJpf*7U@bg8d4Jj@aZmVmRp`DlNZ`up+C#0R02`izxn6AX?8@%7o1mkXJbMvBy3x( zSOfMz?R1R@ z9#g`D9++gTzg?Ps+rzO?4mumuRH5&n+pGmrvi%BDUyh!o&{jLlS>|DH6fJUjPU_9xq{Qgo8PZg_k6^9UA{&#`{YZoaZ_7599YYqgfEX8^Ct ztXClk5&Xi)%khhatl_?N_WgxIHFk&5^uNWTo0{c!u^d)@Q|bx1bj+W)He;Oitj)(} z#jCh@$X(g(K=hGLiF_DD$!qjk!wNt~C;sDWiX`;{MI({{zWCL46VQe$;N+FN@x{=! zWP^?x(vG;}dqCo7i}^K_lk5C5Pz zmcyae`JZc8wbUi4ax_-MX5P|{^X~{?3IT*+tRdn@dY_&>mkhAbrOmxYU33+q&+hAL zQ5zUP7o0gYvYE~1zcpn(XJ8Vdoz{Jl`&&jGwf`F8>}vWfLG|eVurzV+YcpKhaoQ@C zA|DeeOBks`hq1@11=y;Bb>VU^Xby)ezAzp}&~iL!+0Y41^4+ghR8Hn&$8+4rNmOs8Oj=dNz7|>9 z)fHzUy{wnA7b>JWgJFjPp;`A267zF&Ny7kVyF^$N z<=p3{6igw83T6Q`g<1n<-x(u!{3F7#BzYjW>PH{@sd8?55nYYtSo1NI=ds~}Mx{7z zV}_L*Lt<6cX4f_`0b_I#p^nC^&V}nFx0~E{t-hLk9Fz^kjKvy#U9HhuZoEct=&-#@ zD_#9DJ7aQP+uxT1ggDa3h)SvR1)%iOljXlbD@}JLyB-2`slo^gY6+sa{1oyL>dZ}7 zKod|sZ?Jm0HGnCsZcj@hQ7Ci~pl{QM73KhQX7Zzc%`1taypbOLi(N;NDWP?<~Lr-7~3=;Av^8V zF9r0MP`02Xml|4ntqBN=BNs{2G#_@R=?`(?(xT}i`T{MS}IO`MzQQO2sQN>3nm z(QoP=MarrVF0d|dec2)7w^!?7Dz5h%zk9P9WCHf;;mA)<4Uk+jUpY;^NnjHG=44-z zud#O}!ei-2m{c8e0Z&r);E9lSN5vj{eU(|CJctWA(P>`#Wkdxl$q#r-t)mh0PZCP} zYnDK`M#l6U{&G5ij#6i4m>T=TGTm+VTQ!i~@JRL47~-l_$CgR#E(C91ufMl`ay#*N zt8N=&k5igp;$^s_nW}pVhHM^|Qm#-!5XKc6o|*lN!k}gwzxQnNlcrW7t&kQ0^lq+c zddOm`_g20Kov}oS(BEc=!};zckNar7Dfu(z3u6Jgjo;~(a-b`SiIz$?SGKM) z65dNkZ>^^k+B_afHVMwo)I{}+NEu06`o6N~Rl{ocpa1$nxG|*8r)WCGW*~_Z-0tWx z?tcxDCwi(I)GFwKid?Ps{r>Au?@CGozSL<34~IZfkk`VK<>+3gZlr<@-vH;ds~M z_AWPCDZ-|z-@bJ|YVJ)&yck#{$)^LJG&9ky4y+yn>=9u1K}~OVw>BNhG*r|izbQ+8 z)(~XG>`P5%QyPdj>mKc-lYPt)Svijeme*hz_i^0cV{{U2m zUa=VcgF5q{D(3?Hdpus16`ivl#q$mWI*xcFwTc5xY7pKk5#Ne!!IH5jx!K9*!f)y4*?UH`d@ z+)uG~sEOiH0`B**<<2LpR&tQ@eqYY>N zwokgND+`7resv#KcJm>IM6*S*5I^rs|L;9^JBK(v{+gdt9e4?(RVrpW zs#Wwppz4st?EKxn;95$LkG_%L+^6Rj9GM=Eh8dMRv?uRiG*B!an1HpMy6C9BI4gfy z+W6bQhvvU9Ha0iOjdvlG%&{%;wXmJs&{q6QY`HJqB(N_Loo6rUw`zR!6l*@uzeh(B-51C(VI4_Bn#PnO3K(6ec&q9a32DGLN<7 zR=Kd=RqkJu_m-H+KC_TkU yRY!P2@jmNlQdS*W3`&7p=5IX1B?9eBzMHRfUB<^& z{dm&K{jIrTVz=Y9oWqnQ0yE0DwC6OIKWmOXA&$gsh&vKMrm>TAbPN$>=N=Y5l;)Wb zeU~sXRPOXz?L&$T#$(~(8b^i0y}7;F#?|L+4KMt!CJ-;5#0RFy%4zY~61~l&l}g;z zmwHZ-s=9&5Z;dsyotXy(hDWpG-`*nE2v-XnU#b@7ox8Ahs$V5C_wNMETS z%!*$uYeBDar5*S~W2YF$46BWxAYRd7T;7LtEfPg4I@>9o98(hw%f+^pz?74~p)$%P zaAz~%6RCuLx1-3&U$qqr-GQ;$fht_t+iMF=>+$u$f98DTWjfU}M2-(`WKSW;ss%WM z->zCnW;Rdh_CBHG)FjhmN|3V?Tn4c*~lBS+@ z@jOWG%_fNZz@uwQWRy>dek%o^sn6gt8})i}3H?>;`fy|G2h_IOYw?sN{TbCS5Yw`g zvPlFS$_^ib4bT5ou16Xwworxf==FXOT6pu=-&AXvSBNq$b_S@wZ$T9854f?1nu;Wygj(ZbkN`| zLq|{K9sO)-=M>#m=VjTQ1q(s_^OiH)D~&jJ@%%Gdo1w8yiel z2lS678gSVCx89E_@!FWhC#qjb;|d)y;11%c@1u(r74=%10Cnp5dG5MqLus;4#V@Oz zllfJq0#NWJnIF~({>Mjf}ks2Wo6vN}%G zfev$^0EIZ@e$~*A&L%v*68V)|@wQDlBK5|HlsoNnGe@?MAoQ>#gvX#Nx5<>M+2PEB zj(oe#$y7U6BR!2bd7n`@+)Qb0T&RI10q>266Q!D+Qz=sLor(qErWPpUS|QTN%}D0o z*xsCUUnWzyFzV2dz8#AE^qzWTOPbQ9xon)3?1dQTR7dTx*w&L;hgLgG-=Pi1#|o-5 zv>mEfePs%`TVl)z{kqFaYk7KD+;VC|5+@;Kg*R|nEkJ(4byeem*auzi#AzLj0ot^Xny!n&+&5FeZ(`Mb5$ z4#Cwk_qh&x1LVRhNgCA+Tg3bWfUD}Kw(K*_o2rNO_rxP5D2yP^j8F7J2RJ%@j|_g> zeAZ4V@S$0v*PxRjhE~K6mYbnpf1TdzEdbFpie&JU0F-7CeX? zqDFc^Cdy_n(s3S1`Bd>9H_6HMemZ%j5HKIJY0T#u6xca-^IAt(RTxjWz z?=-)JlF_WDo$tSdyZ-^a`DMiQvU~d}(ILk(;Z;8$LAI1bln}CTyW5XqW8(@Prw!+B zaQBZ|In3P{|{l;e@Lgze^)%94rK|L(M6KlW4$BCo@Lb^c@ds7o*d-H z^Kw*CG?_twj_V%H70FL+*Xr_nHOrE>1GP(!#jJkhb|;0f{>HX1m^liiO4QcV0q9zsi4b6AeZB7OC&3OT>SM_ZR*F zkUA~Ck;Er265p?Y?7^HRAIm+GS}gzRt>O_27@93Z%+DPb*W^mZ+4RGab7MIftZjlQ z*Xoo`ex+wE%7RAe<1Shlg1d;HMuOE2Ns0tH-qPZL`;s4ZE2{m`M%^~%fqBi0{eD=h z&`RCLh4d+*j1{NCQwLXL(s@=!dXmfyevR{UB{r`mNEOn4FkuN+iCUmy_jVc0I4BFc57b& z2u`qkc`y5}%SQEz2%7dV=*8L~lm6bu&eLta@YvSF`wxIzO5UgJ55Q(PW3tq;LNMsZ z8B$v-f9{xB^HH&Z&8@zN`}#Zk-;C>ng#!p%JUY&JO;`ENxmnNCCt|uYqV&p)n)dsF zCTX0_;}8;==W3*1v+>+lT3en{9IjXxJfc*4{V;<_Eb|uj)wETZ^s%e+H?;rJo=)M# z-676mM_a=o-c8z}ljbE!0)7?sBI+b*$Ah5iap+5{o`411RM&;`=pCI-@u_yl!w z==QA#AHfwp{J#@4Q}jd$ws@l4C50)II?i7q*#xu!$_>TDTe9bE3zgjc|BtE!mIHmP6t zmn1HmF7&Z}6LbQhllj{*K0$wGIc#fzC{S-l06(q7IeC2UL*;dw9_i@XUUHeQ9Y4~U z-5mH>0LqONz^}(WJ4q!?-inJp4xzaSIs<$sb3CzZO%cOzFE}#u>6d47q8Sm6bO%he zL&O~d_BtaTHEC>D7RR1Q$^Cosw|~c+e@({f#FGv?hZp{C*YT13bs2!PmHDmpL`@?_ zUaeuBKHnj-@UqT5 zn&IbyL%Jl^OC`A>NB2?-scKyTqR{Gq?3ZVpGUPZ{&N&~QF&nVD?b?L|@!Fh}7t;+z z_8ar0C#a8p&j0*DjWTp>!{4hP3^EvjB+tYNZlLIALJ=*OL$Jvx$?gy7Xh(tNLHTci zJfag8acWc};hzI2B?#ZFII6G46?CD(4Udnr8#5*UCO~BoW^|9-rVQzTpdl5$NfX}e z;qw!`Z9~fS2VX;MGC99G)Peg64wYm6((u$}6>iqZAeKMSnI4*S3xf`)>VPm(cRCA@{>Nt1>1s@f^tmkX&DI%kf)L8&GbUj!N4mR{}!|f6Hcbv1RrZ zwH5hqN_rjP7Hd83gZ6v4bsj!#@&wo({y2~X>B@v?cB-qoRGl}Y#`%C>^WkE=+8ol= zvw~?#4X?g>9=wd(9Qee=vPf30Krktt;7=M^G6LXYx=L#o4ZAU!Mc;=cj-tm^onzCr zsIgwC_jzf=7q?RQy{H9~u2eDh9dvNS^OOSG2#JcDv6OfcgBCKO*7$<)g0W-|J# zO~~$7?F@RV)Lg!!7)dIKGEfqG5#Gs;H^=X!BudJadsMcZj)egJKG_U6vT8xDK+cnJ zUWQ!s*1eQgcozIY?hnA;5Jg|b?&>z`d!$&Y-K^ZK97iw+;E)VFY%+bCwuTJQKYOo{ zZQglPOslij)NJ7v$j#8HX+js{S}SS}CzXf-7Ux@4{*{7VLVUg-JWS71DTKI{9I5Xt zRc<3Y?+Aw%A4%vR;<;XR9JCkKOsEkpO?GRob8mN&*iSxE4vbCg0*tkGU94vldpxOi zC&xy+u3lC*x~g)#2r}R7y_s;fKe*0Q=HM*t<*#qO@JqvVtS&$_Tqx&C{Iq2T-9jRcon-nC??QYcRiW z+i=rz%FXtHitoLXoK3@QD0{U|AKk28*=*;syRw6X#i(S&2R6b83r(L&zt!(jU==T^ zBC)ql8bfB)N;(70yaseUAAVp-HyUmXTM@w-t9KjIeE2Ky7%yPt`FT80Qq$U35%-0L z&R^&$cGCA-9ZgfbueQ%?#@tM9&0?3@8InAQYHh&9_)9I$#6KQkCBhcl-KWXjSDIbm z&sMbX!L1JRl;sLl0Mr8S(Q`#Z@*06)t$FT#GTO=YQlCortvKy0#k7Cebt#mbp~ zHeEIy;}I>PES07=v2Y5PSUT41tjKf?)sH9#%*Lk`FDj7I)eqyni1;9{@k65#S_@K! zGMIutnC-WlSwcUilLX{~efTBb2d+$?d^n0p(9i`u`FYYkXBPelE!Gm}#Veo;IGymJ z_o!AV`Pk}_1L?Z8_XnMs|8;|v;QCq{baHc3sQX1rFn2BMZG)1-bHZw3=138D@P^+9iKP8bjHa4aFrL(TI8lmrw0?muD0I3sL5^D zqoQ;Z&A%%3Vg2rR&Oi3NaII;o9s63E=eb(y_e?hn4>3XLRW$*IA3qkUMp3Bjs~@%a zd~JuMDr?fdb=h6dc1uVd^5$DtwO#S)fKo4F#Atp~_LpN6z78 zF@K9&jh|wu$sCMP+yQ1Nx^IxWI|NLxH05#zHvu_g1&fzUfzhC!W!TOC7 zN8^F!vlumS(VQ19AfzvO&WNtck&M*_%C{6d*CJ$I<>AhJ6LX~9(bA0fSCW0`^sL5q z)p1|YiE2tBW5i%5=bH<7?e;g-Y)-kg4y*o)N3#np&av-8r}wIr-PaCVqDRxFxP$Y& zy?Jh4RSxnXXC*jy;ulSvII!?Sbx0;0%eU-5YRvP5NEj4)M8u+nt#iCYp;afkHlWxj zzeJ8gpfJU|^kk5S#3FPlFFI(!{ifRZ#7)xrrW#=)j#CoE9ksH;K_HaIa#Q9Vw-b>O zr43pJ=d^&xp_mPPr@%YrK1c;gk9hBclfC*rL|u~xMF}0`>hir}n%iWtym6GHo`)ca zTJ+%;o|Q|O?+F`5SCK|C=~Sehv^6Utaa?T}P}%`Q_=Q~|TF6?nna+@ILXneDH2vO5 zxJX#$T(Kd{Z^H7chE-ekIRW^*)@fnqVn}`es_L1)^N{Oz{x!oSSgP)|(7iC``a0bE zTVKkTbDqKvvzi}@_7q4bBMe?+9}KPxPq6f+S;~Lssm>7O;>6RsGI;DQCFR@-o+-AV zwyKRH^Zq6q@!7Uq6cgP#?wxVdp2D0VRWMr4y_b@nx4icUK>J>yaYd!%)6knJ&-d4i z=mE)vsTeOmNVoUlwSwEYLZY06p|rne(HPbK84$X+q@x5Z3i(g15(9qRT+K`1<@nhN z3^i=IeIK{ z@KRz4)I%Ju|&a2X5RFv7VMK&pCdGW@N4BJ1(rOU!;iim3Y%hOp*x`eGHT)d2c? zx_AnO=cK|W#)rDoi%x|RVH=@?N;Y2)i9mKF=MRnv1Jf^t#9s&TP^#^8rZW6~;ZPZC zz$>wJZd%U5Sap9`rNhkezft#=L2b6*zi)6U#R?P&QlK~#C%6@Nw?J_T8eEDO3&A~5 zq&O6JD^^^Kdx79qoI+{4`Ry}%_W#VB7yF!f&e_j-k;(ny&YH31uO3Z<2<0Ss>1Y!L}t=w_Fn-_u&scmnPv) zlfx~v9x~AZDh%(}Se*6|U3O)ko%kLMP{ovlvSK(Fsn6A%SKbxRsW7T~96rg0mD z*-NH$X9WnqC1ke<%ouMxhU~|8Igaq3%6$vr=xfhbGpt5&DZKg&G!`uoVD#q@s?d4< zc#yoMIO4*dpaw27ni4Qpb5{Uig=^2+xc5T5>1}L#7_Q3a!axGOPvl8#>bJ>*Ay?^6 z`l&%2b-K@P%VPaqN%8T7Q;xh$n!_g@(=JHhq%c&T;qvXp8RRZ}H#>C7xS@QxaOJUw z0(*yOBRQ7U2$+@-ZGD9*YLQVUFmDX`Wb?e0mAYY}6A7N(SU^t1(%tlIe=LSf4NXj4 zu-;TcXAVf_Po^wdy5yzwLL>&Oiz1>phw`gwRp&4<4L#-p7G<6N2`HE{c=(EP%T~CG z9%g}I1fYs?3FHr1RRYu~hS%zS5+G{jYXOXYAn+_ihqYDnKXlJmZ#njW$@0khL$-y80ByhBk zSb}j!%kzY4kE#z_n>%qW0SvMqRTuFywlwNxl*a;Iq5lOmm^K{>y@&;6g?J<@oWst) z!w>~s7b?c=XRvQF_`VaK7Dzu}YkNG&kF_lV{!}fqnC*Y6y!xLw|GpCIe+-k#puJ=m z;G|LXdy_G#?eA@K_W->`&Jya($%iGUynd;NPfw2@o#BO>oDs zhi8FT`ld3Z!?mljO%l&(hxqrT@2BgP2-X>4C0Uf84@^v z^Wkoj5_-S=^}!_MQaPe3Ld>OJyQVeXPg0#%)5U&#`itVcLWBe4!l74G2v+)gZ1QRlW45taMw( z%5@+fzaJxGh!@p$yW$4uUT)UX7q);mSO$3=-`aV|uOoz?@t^?{{2RCm7(>d)0q-aN zkQ_QKNyvmF6e5xY^Ql%Vbt&zK8NFt&vwjpgamy@=%d_uMYhTnfiDs8FHrA6(%`JAm zSDu$_cu&pvs`fI0!{d9``&+-gEA@KMQNB-^;-%QL;kQ=!%U^tLo(1DeWwi; zukP=FKLRx*0aK&ccb~^zwGE3a_OtW@O-iK{ zJ(5+Bq;Q*$Jq3ro12W$pyLg{-EAn#t|C(E^=gFBTmip#mR(Px_UUxdF;os4M2DBp| zDrQBaQsZtK|6bn!L`lD)E+b>iv0gy*7{wBW9vjewoC1S9!QjZZf*fl^vp1D0mQP>L zO}x^XQ!vEPY8Lf#ahV*44!(wr$V{tyRt~$w6OGDt>5RY|V3uto@r2zc9 zz<@(qV0F2Ws3}``2V<{o=ksKy^3h&0kc9(N;i(Z;exNvmuU^=Geprk?J6qaSNqd+7S^0)l zAsvSXpZ@&~@ab?~g(ve5>CRA};`}?cXP^EjzKke9S-5>g9mV*_Im*(HbZEJGwRNhz zx_)k32H%k_N{u?sFl4O_L{Z6=-;5<-OwdlyWsIl5XztSUpwA4JH||c8)Cuz=?ZV4n zbEfqX2#u@X+^3kZm*(c^7NR?cFve{$#!oKraMo%K?;jfp>3>lj_v<}|brpl{$esQI zlpD&Jc3#T_?A0%4k1!=jELMrtb)kHp3kL^$zjx6K2M|E6j}PXjZR9nm*`w#Kf6C1Q z3Mmdv2_4G^wBHnDlht4b*)bd&Sj|-p(}EbfG_&OilA}mvOqsSB5WBw{=PLvKH@3pQo-TsKv}{|LjtiD4mQ(9_-g|A= zH-U(6QE`~#f0DWnX@%L`50c|w?`R$hlH}wN6w9WC2ja`VtwXS)y%`lEBoXUe9{ zMkVI%v1mKir1f8+@NW(L9k0UbS9b%JetpgiHsl;HMAd=71z zbEwDRK+;{Zi^R%8-spJ02>TLNQ^xFU`}QYVZs4&D^BU%JbMGOG85YL8SrYlBduTio z<{8D>4V=SOk?@#pQnFJj20!*Qx=>e^*b%Sp8ZgutZ#pyaw0~S0@M`z$MUpd8jzWgi zguE=Pc5Gs8%v>3$q9^RsTrP5m;Ckg0txFKY{#TRJIHe!y$bg}5iLLVK2#mX_5_yLX z998NIM@PGtMkgZSymMpuJrz7CYk`0&6(cm19ILs7&oAVAQ1LgiNZPXMKTu3DM{f13 zWlmMu6X0=-k&)_*9S&)kqI;Oo03Ry{w~me76kHn>8HW$}6!-|No==)lqtG8z<`tVd zi>Twu;YJ|S))9vb8{Z2RB{C27ld(yK*jXM~J$OX4KssCd=@Lq~YlN6=0_PRd6fxF} z1ZIw%$&3s?LUV(fsN~3I02H?*Im?jL03hBM%(_wxxZ7atO!04p}Q>IAv z^j%X^j#E_8tpWaC#k}S3?e-W#l?O0Y( z_&i#JYXEhKyl6HZ9F}4d#-q~##NC*!iDs~u@y}TVond1Ypohb9M>o+LTNp}bN+1WV z%HtNYN@b2JvJcBK9(jSuf*U&LueVmT>>>?3 zQvRGB71%2I+dqA-ozUt@z_vahNEJfGIQ+i15g-ElNMWUq-NP;BqA7h-vO?+9m_oYs zox>o6K)`$E$bp1y^OW31K$l9u}YwFOzH;r0kCZx9{N)J!Ijm8SvUW2(!f zReLnZ;xa`SeDCD$$530rDD$hvm;~$fyL7%(<{C4qccP)jIIiItk9kJNA>OTMm`) zbzu!Y#Gi`Ehd*O@8k9nL)xXejApcD)$5h?H@QEU@w)0OlpZ;r{e`O=;e=Pm~{@!?P zz5c|oGD<5i{K5!5Vza69{KF6=h{WPQHxvv!YE9&Fhw9X{Jn>x?-FLK!3*$AETPW}p z#e}MC2X|<)^{dz;`ia#c%Bqd6ua%*n>b(#GPaa)HAUk5N{k|mcgE|pdoh_ETj z6P)%1Vq#U0uBK4xLSVEa~^F!y3N6#Yuw9b>lM2nGkGra$*4ulVI3)! z`7wow?WN2htc{{3IXk*Az-dd8ai!z0+6V%MDqaxDu+OP>=leqR1Y<4(&S+#T?a>@k z+yj>8Yy$0n{@Vc2S!-kB$f}{AKd@V1sWEmqtxIYi_J&gfD@b@9d5Yw7p(u>gCjO694Xe@3vcH(g)1hN-bpj(B?+ZSXoEnJtCqG-p&ERYPh~(o z<=&sD0Kq_ps`sazCfCST1&4%~?^fUX1GS|QU;0UPVB)2{P3+OOvDk#b^ z4Lkt?wbKV=x8=`xhnkd&h!C-ip-Cg9tSgQ=Cmbj#8uCt?G;Aoxh(sP8DR3veS3QtT zn5z5Bp{>0A2$38Z`t;ba7<`w;=PqW>|PBU1&-fB`QaW5xXNz5oKV2=v=0Q>n1^`x+Qv| zEvOp3BZX&q@36v*vd1}sLG{0xe zG;)@(l5^zoAVFIW@Gm~}J1>$J#o!M*;g#*xLAC(k9ZBC#zWgS>PjJAY3JM$`Bs4Ao zhFNkAPy`+iKO16(WsmXHjQT#o97&c!byr(=ZpvHOt0%r9Yl@k8iO96n)qhMDqo9UB z;hy}?9KM_WxTq)Za&{7e((vr#l59=88j>+ssh(Uu0RPz?Zgr0jt&Wam zRCv50oaOVcHlLjxYgh|^s5Cj951rTN*sQv~eNMR#%Y>6k6w zq_m$FMzwuLLI$G`vU?$xaOoRiOpNDxOE+m&fO|L3Bsb`;(!hX6?j=WmyqhDUt<7O| zEWrcsLXZ5&idCSFB5`u4gVwC6xY*cFZ9iAa9EEyvmC zyvW*lH&BZ1!sfD!j% z-=B-B5A9x-|In@2$eY;K4!{>JERaCvX{pO5O+7u^u_L(7mh}BG0qu(q72B?U zmp6ReNd%EXn5AgkUoyTQ$%SxQc}@g45QRGLgox1G)Y!ESk|WWYJd%db-X>vS)nGPO zl_paNMY1X1iD}n?2X}|CD>B#c-TY~brKyPS1V6ShRZ`6-VS0zc@p2=}Id)Y@-NViJTX|SB81vR*ZHbPc$qX#pxGh zuN0(&AvGQWBwBfy8RG0*0Zq}okyPzZpU8SWo_jVYCRIyUkX!w1 zXpPmt#hRy1Ts*YIE({cy&7808bUqBL1BR5f zR@<8Qi6K9TSBkIq*VzEnjg!WGpfYn8g~ zQ1?v0@~n>`ewifo?Xlr}#}}W}2uYTwitVCMv2Gcy@Q<|sy;q{vAB34p)FfC`zWBI6 zA0;BI7kJ#A;GY-KUFi0k?2VW24cEBUR7jrDc!VK;co?S>;#gAvA70F)Si!wG51Ts- zBE-C^V%^kFmnm*YYUK$@S1%LPZ@7%0vo%(RN*{@2h~@DO5Z>$g=I0Kmi6?8MPn2f^ zYbP|?{LiZg_Km8SlCgECf}AZKHkD(J8;_1#ZC<~LcP&AP6+SWZ8}lxOT{B#dA7*J- zXAHxG?0S|@%A2~;e|o9>C*Hq&x#*;l(V-*&~p`Zo+ z;Zf%#9{Js0YczBw(PHjtiR9lFaj!SdhM#~=AWY^|Y^>|UXfQw{ZXQ_sFMvQCLcSCl zDNL(e`+;Z)hRMnpI;1RTm9w3Zr_Ls5S-KL!#30Uh%|EF1v52-W4fqEO;q4(~Sx3IP zL&JK^p1-N<)G9!w=q|je(`xulrROEJMB`1Tn8mPool%f4wTH@WDsp!(Hh!;rg=lj7 zD{YW{K*Wz;aXupz&unOGs5)^aKE4CkA0e})C!G`{?pM%+%FTd|n8NEi@=f{}$sN>K zO_{=9KTOi`A~&oglFSLE=lZ77dqK*If%#ZJq7G0yxaq6SDrjs#AGs`JhYr0FC?Hm# z{Gg{RS%nQGH2Gmw!lSRvA!ul$jw!lVT%Sw}3`7l6x~&=$^-`=a^sq}&t7`(@k?4s~wU(CssW)?TehA~;F8XgS0(uE-nuDV~*L3l6?>msxzymYRxyDBe z%dFs=OfyzuWKSq^quS*mSyRDMC+24RbKq^y zCN7f9py^}PjyCq8;Uwnu{&aJzYM5^+`Q&e8JQ;WB`p({0aprbS=&#QyewOd-{By;&ypgH~B z&4g*BBnw-?Yb<><{%p$VU%p=~-{A)G>CalC7ARTF0%3-*_8s%9MQYZ~GPP2G=kRtx zg9-ou-((zS9hqq;*lbF*QF83dr8=Zv(H?t23wL%cQQ}PRXo)%93#Z^j!!Mt z!t7DdsN!&m!8dI~<3}7wI6to;U<>@T8L!xL3?>+?0yt_ef2w(qf3myX0Jl2`P{&)m zGla71t+PV@0z84ZLqC(xxHwRA2#rDnjk*5FKhQ`dmDd zxpXlwXo>!;bHQ>Q{BE#SxsdSF|7^JaFTp|n7w-Ren#=z4SHQNO_F*yM5|8EAxIb=& z`mEXM9f+OqXrrn16i9?khoTwO*i&JpV*i%s8Y^Y(*E}nMfW-)Nl{DR6BI&y>&kbU_ zA#UbwY49xUp!0RkTHaZbnhTbLtwJ(+S?{pZ&26>SSA>;FUFfo0z)jB3{+ zmtAI4js22-36U1$QpkL)*nwRM z_ty>~Pv*NRC29;^x-AFv3_cM?Ei8zXKo+n%61o=Q3105NsOg1>&iBvz!#7fFia27v`kVX1pT=;OwS9^qkI@+W)%_b2k@&cOXa1xP(_bCKd$~7s$1S8=*FKQ4 zv-!|%%mplb743|LB6G98Qbq9+^7;Ln(x3BwIf^I__@e0V(n~^Iw?cH*JU{IkX0no# zlS_Vu*njGoJYm^KtD@7TvuLH365h+}fP^Pw-8JtOtX0z6kW(wsAQ7*<#DK*9ECRSZ zdAnfq8v$GP*UN1?_bB2DOd+ndOC*!K+=^zc`-o26x7o7p9`ISR9a-{t0c6{u;;7+| z*BZaX?VTK+(Re$9BqJ<;=4t=JOcGo`$#i~(Sm1S%SOH;^Y^ydtTsTfo*3*ga z=Rf3fFhtbv#!cy;Gv&4iM^rrtOFmm~tlfz%zPTWl6)bQ9vBf4OnMK(+Qpcr2%X{Qd zJC6rz_K%1f-28OHSGo{TG1k$`w&A9#uHCAEswp1T55QZA!i`RkuOg&FQ8z-3IW`j4 z?kN2=mh8*wS|Q}@r*pc-3o#Dcrc@^>q_8b~=e?kcK!SxB1uYsPIn%l8u`2op6VYLW zaN;HQkvJsnA14F;3w55P1*Hdf?J2>9eFFh)o49}}KuW0bN>H*5YCvfHt6|!WDn))n z{yPG(N{-3{2`%+>@1%Icq$PiyS^~dsuZRXwYs^J)KmADG&OPluFS}R>g$q2J8=}wL zoW@YXs&e_MkmX`9ho{f+sriv%V5b#?tU?vJ8N-94(LN2bzJtlO>$J8otWrA)2WkIm zxB^yb;Mix%0Cx7u4b_qFl}Z#U9Co+geODL;{brvcbapSLi+>&Whq2yniF3c{DSj+GRNf>Mh#$m@J9qFhQDaL}zKF6u{P}E^hoEPXurBH000^~9 zxEvKz6VO}gr-~w`-KI~O0Eg9=5-H?Om*6xXYp{Z(Qo<@A$tv;@F_CR`3M_iK zt74h=4sCP8qSz;mab$~^HV=rPU8sq90Z1gFIpH%B^kj^qO=~{m5jHC&<$j$d&2vJ1 zm0kgDU>R41o^TWF>u<#g?po38GWL{P)d_30h_?MX0M1g(`i!?dJkC|N#b63&@qKHh zee%J{NO(p_?q?G^;6()ptzV%&c%q$G@_alF^Egr(gmi3Ryw)5xIUadrb9A%a8vStG zhd!Wfrmoj$~KGty~6 zS9U{*D=VV(%{<~8DNu$9k|kpTbfxP?EDKB3^OWJrM$z~b{?>5YqW4AlyI*XEZ#cCv zMg6A4QN#m}o-ctI%V?HSDCxy}&U~ku$XdS{7`5}P`q|5`d5i70eeRC(_e6BV zO6Mp%NL%}AcdQqJI^Lf|3JiUGg-aTEJolbxYwAP|m3+<8hid0^qG zo9y;P9XE7XV2_5A#%7N3T=^w>0OFe3)+_#7;a@;{+-1#8jlcze)_hpHxmC1f7x{^w zg5;~7&G-G`z2*6-Xzr_Nj`teowc}(|Wl?7? zf`6jsg4r&dbyVAD-t0DX#uTRW#VcqXv9Iffe?R|x8Z6y-K^h%qNqqS7jt`j_NKp1i zFI$d=TJGIEfA$2;#i^@>{f=1{`I@*wSp_NTKUEq1Pn>^Ibu^jfbynCQd}yVa^F5y* z_jq2%yP!stDE(9Z&oA%vK$VG;txFI#2sBE z@S)3kI`!m)8rQFKsA+^g*WY*L;)$vhMRelm30(1r9#a7 zlRh3WfGsx*tuYjBBL(1{k9jF4P%mm*OwK-3EJSg<62hj)-6dS5@eBwC!>vy-8`j;Q z+xwbF%2Pr#T6-mS2@f3sXbZTRF1C7UWm6F6E_$w+rJ{b9nX| znWJxd2J9N0YI1a#6k9Wyso93Ld3q_tA&oZbphw7~HBkt8fOlD&gZpuwkj;;Q6;b zKImu<^T)#HGWCz8KaIJ2{s=s0P;7ua8(wv=g$G1Vn7)u>qVhqAS_p2J%N!G4eTtuZ zqe`eW;n43p6_+fHX7%wiK7_*oI-ER3ZePfObs_8`iR;jiF5PnKcca#1LoM;}<9)7* z;C%VBVQZFCm%{2xY{@jvs5T`hOL%ga0&_*(3fsV!b8h&w!YrMQLKExTq`v?^MqZTZ z;G6P8Mo2#)f7I{ddz9OJ2THP?PvP|}SL^Q>e<9Mcc6L|jx?%-q@AdtYH$P@$x|@K8 zwsb!Awt}SDgM;LTc8|4r*Ob-O_iTVKrh`a|tTBMMua9B zl~|YO=#V6zUlJQJhI%*m0FbVtj(J_}!2$?*@6}P4P3wiYfnob)`rjLVNqN=c?AgJh z+MV%z-;8#%YDqk4Zi*~d>vD=ms~vsW`>z}~=tLrh9{pEZ$bWtQ3-}GW)s;o}T~kDZ zjhvr-Xs1bfULiDI$fa;Wvsd#D9PRvZ3~9eZG>5oh?He76GcFaYwmjmEjngMjyQ~}; zIdcrZzzcm;@*+<3cBbmaJvs3a-*GvP#5Ij`GyYI1*cQy9)l>pHJvGIAjT4Gqi5m1_ z-qWqwn!3)Ke@1y78>#Dc9w2xogZ6 z6vTb0o%PC0f37~_g1UqAwB162U7F6D!ub)ew?{={{RS5v31lA(T;e5vdkH^d>6Mf{ z)EedM72u@RoMGbrzM_H_+}KH5aH#8eptPp;!uXvm>3p|La`lffarU%%2^#y1-J;so z^gi{`QBz6vyeg)!q2zC-wx#D+;97lOOO++*U$+yjf7ln={PNQ`?*hL6{Gij5}R`#!|;@J_|>2Ci6j!r%cyDL-7ths9j7$)?#J#b?6 z6via+vl=$r7hO?rs%)6IsRcKX-XFqEgR?P#^Z>YpoU^r{!^7Ov8Pu=bMk{ZL$t_Jg zU*tk?s#k%EyrQ@zGlF~_%xx2I4R@@%{k@+hE4lB`Ks5{e!08~Z1%xbrC^Q+eBu$A9 z@XCMd9`Wan6daoCua^!Mke4^%?}=zNW6ilK&+Kl=+3tCEx0KeTNv!CATAkLr~KxE&|nhijAGo@uP zvpWxX-Ydz~JMFJ0>P<|>nZDX{@39b#qtZ;}0Y@@ZltdI9n7!J9v?Ub?F5y}%<@8$; z`o1L493ZeAv3nnB*3CI?DfiN#ZxbOe<54 zX@dN7C9u(^egZ4phbAq_E!&!IU%c+nB*X>o1@=O>Z8+nc&xUXAI}B++Qpb{~Yw+G~ zrPfjZmIH&s&}a+M5+?H-<2rkr)D^&#ott6!Gx%;ElMaF`32`C6dVVRUISsyZVDDyN z8+B9Ttd|uUWxQO=g?F=-Bj%C2mTJhKQet_KSOlYO8%i{L+mNl<@QtIPcd zy~@;NxmAAGu0WD7$-(qd-cf~YT8(>q-`@^kZ!&^Lg)1y?ZS_r<##I%SD^DabHhyqS zdqQ9K^Fj>mKPEMwRMHj;PpQQks6qi)>Aa$q zE*3u*{8vA2k~6dp58Dp{s>nI1z0aU2->~R(1&!azEB2Ab$t2COSrfRcDSr~Wk5!#m zYcLNYuOi@NlqtsFdQ@g(048uPsJ@Y@=m@d&FANe|j8 zFX%1CZ4{`;-6e6OO{_5`xoN#9pnufJP)_;k6Hh)#jL>cfpl7J&7x5P1{GMINVz=5g z?4gNw$3hym@HT^SH2(tXG5BLr>Yp8_Nzyk=yaF+kgZ~0_O#cExA73%$5!}YzW1iV; zd0$a3xQ4x_V%!xlRr03Wd*%-M3-DbmB)gQ*?qN9|@F!t4(f2ciJx-KB}M5oBF z@2V)i)m?`13XdE)ieLI*Am=o-7}b~PFnSJQjd97Cm{_WmC2^@JwcJodCTi&q0;^*k zb52+Dpm#(3<;)ZCb3fEdJ{rJOHzWzny+as4R&230E&U>{{*z=O_u+M!*(ApdSA z?PLEduO9^ z35~H=tdDX5WOGv#wCZf{8kJ)s-j{DiKDo)GX;9_yMSp!W8z}d+lAjN!sM24sl{_}5 zQj{q2lq_~P&glQByp}Or4Ew_ojDm|imz*iWXM;8Pl&&v*8LRZOa8&^7IsHEd+5PWG z6vBErTe^A0072MX!?df*q_@*M9h2K!lIGaw&^MV1dL$=UuYkO&J@W9c5Bgm-k*=i_ zUYA3^^p<2D9bCHG2y*VeSq)o1vlM z$BMt|5}&5sQQE&%bZM_9ByIinGF%jbST90E5r8nm z^GvL(&mB2K^xQ}Ij?O(6BYX9(Nb=#EqOV^5YP|^VlZmj-&ATd4-q^_M$KUK3({}XV zQqXF2Oj%xUC*Y~$`;tr>wzo9_cAoclDPDXU1an^j*7}#O%VethF~Amn!jd1=XJ2ZO zed}oZk)RAvk=Kta4jf{9^DH;e(N1>WrfE)}o5~z?AJ5HBakrd{FHP+MOQ)^~lN#Hj z@xMXq5-9m(S09+WsNd*o1~?y~eHgR|KaFqNrThzs3a)!|t8leE@-g8ckYJ|W72EL* z6ng8EHp;Ty;TXp|W?Tp5>|i-~8&1_VAZHA*--t7l_l=233o?@zGEbPS7z5CqNOB3G znCdre9r8OyFu#GA+_`(9EiJrWZ(c95W86$|ncr6eb@0u`>KMp#IMy|+xjg~rhy`-R z6gyx@JSB$B>7n-5&f=aD^?ow-n+s2l^J`+ z(Vk@HyO^gju>J?(BXj`yWYBS#bXk#{?dx!<{hA2`A;0zN_*n-HqudcTgIkrOK};f6 z=ZElFx)-`k4c3Xgs}uy2qm`{cD+#!Ja43i`-UJ~Lqw9wPO&vtpSr$hfi;lMMY6 zVEfnz`#j4Bi=+RnR@LgV${uIAjS<Lk+I z(iR`I)s^fEe?-)9oY!HEJ1^}FmlV~_xWx4XfZV@MWcAoH04^4xzc+>R^PSSX;dkOF zf0~Rz>5j#b8^T;%RG;cVmC6qkqsk%;5|_3t(j3wfor_vq8fXGz)diId*h4%r>%WJq zeh~bW5EcAomQQ99Ux6}O`WNtSqg~h(Y0=2V1IPVQD6`Tq_SMp_F&Z8;E~V-@C<4{J zq58xC(yri^#lJ6iW$Y8&0ilV;X~L;8@A$*1u^^|4Td#2OEPb43+}17z1i>y>ML_PJ z0i(qGX+LSdcg9t2(wBPJgMR4q#FLU=XWETcd;Gq0U8Nj?H`WV-aC-mHY%WZ}o+ueH zaH22rq}j(tM9uOZaBqq0QCgvNO9n>ss|l?nLel-G>69j^<{@hd+ECp+@ug}4q#|}A zsa3qk)Hs{r?tKU&!Kb}XuTyIz(Pubx=RAtcGh{DYTxC@;@RSZ`Yxi<%j%O-_0pUqg3#nhE>5>vuXQrQJ zFXj*Mv!zes|A2%>(^2m;FU?4JT(%s_rBIJ;6tmKp_ziO{xreC=ijyA9fu8a0Z%IyTa@JB&!qYe)QvXH8uFk&zJrssH*y+i8$x=gr zF%N@PCFT4JSxQTOtpw?fJQiSF!UrX)YLoM-rN%L;*HrXUflucV2UitqQc}aQl#U#P z1I9?~CoU;KAe&$-`c8^wLvEDVuQ^86_k>Hg6}8ode!o#@e71 zUCNj&$6o`Q5Gy~Gt)&C-4#73CV=a=^V7_ONl~0ToBtGSI&6kC}Sj@>#^%bKQJga6_ zLo0l4X%Oh9EE*RDwZu{ZXf-{Df%3W?!GT5~&Zi}ay>p4rW&oln0bt4HW2RC|Gasr- zsw-3}_QS(ZVHPJ{hk$9&lw<@8*7WS-7eGE*PBI2JO0wu2tPS9QrDoOW&i7N-=1$18 z^6*#+KdWcUkG|o|l8g82{5kh}-JxolG&oq9lF+_9Hox$-|E7T+*25l(6!{PEZDua` z==)dS>a;}qd_X$$^8q@hI?zUmc9t+Yeu#G|;UIDncj*bF!~~C)pQHQjpJ3dCMK&6w zj`BTes)?H}N!ylauuENIu}d-eM1$_l#5AEufG)R-(L2u%E=Y#^=_BL)35-m%xjbdS zQlo9HO|l?O8#8cG`HNtDzxmCH>tJ#W(|9oF_Ry`ynZAwzKG_! zJyp>3I5v53o=bdj>zGxL=(Y4Um~FI(K?@RB{?stElMN_cGd97w1 zpPzE#kNOUGNAJyW0Hjv>+jHm5DH2JQLa->-+Q@IqYZ?QpMBXu7n9?$j1rUcB83?%P%oR46 zLp8@l6rn?#tL2dR-=N--QxOIfAReETgXLZ`C_l ztBlf6n@jzhKW+0TO>Nk2{>JcUN-8jMP-$jq5#*h%!-)9{z`f|Eg$OPn-~@&J9xs@L z_?o9TawtzbagZeooavJX35_+lO23v)9KV%DX_?ujnGhd2)Qc`Ndb+klPdBnzW%4ar zh;ookVyQOT_TNv zarN?uMN*GTS^<48ZF0SJnoH|Vg<6|4bu?AX=|YvNu}agsPM#5<^}Jse^#-?c8CCQ( zKgvE0SK8v&7)lH;MHISZk$F*vZ90yMgAKSF0a9h#q`+%Cwgf<3Q9jkAi`OBuiI?l< z5bF2Uj;%LP!M{dusP{xclNGjV_#sHL-M8Gzf(~Snv?#eHH3}=L#M_L|`k_I4bmx(@ z$p2YzB#jPj5jD|PSmGAVd*D?vE47ImNl1jKcU~2JSaQ6(@Lsxtu&}V?*)&?@+K2ti zQr;Rm+$a#6|ATh&v)lde2ummI{(V@Q_uB!(=T(a^TN2bFygZ$aG?}Ej-Vy4K8I%-E zc^o+0#t;Ef>0K*4MC3(jFlls390HLm6?J;Al}4!R6s_xO-PBRo4;+%~u=0m*$~xyM zgc-1AifoNu{{HMqpGlPN+SMab+&&W%BRp&l5ppqbMJ+ZMUdi9_(quuEW?=^I%w8oW z>8(bMVC5noW>A%S(P-L=&D)%<*;g_qDTZGwT~k!VhSPW?5>zQlvH!@OL&&xX3{*vy z9`*EyPO0Y*iI> z_bYF3r*KK)iTi>Xu&aYE?Q-spnRI%^SATL{Z{PWZK7T{^T8n*J|+D(C*I-52MMNiLmCMx_w3LCm~UC8Y1 zagXZ_K>O90kBVl88p800(7kkNngDAT?A25H*tvsjwl^0Hq2fouVGx7U%n`c3bXK%7 zT^>btlMDgShW4e16qR=p{mXt+&vUbidfJDF z@%@eh|C1v$}$N>0j7n*74=Mrrt=y)j1XbH*@b8OeWmD;S259ad@zE92Rz>Z97 zkCmjTiQaNWHlCSG1DZDR15~mh#uijWzRRQ(=_|1S+qnGDVa8wDC=m*b&=dunF{?-W z#0qc5U%CpylLQNi#O_EAzf-AqrMv3j2y)aM(H-QWD*93-~ zjwErNvuPW0U;OSq)a2XNLSdj7WJz0~!0hvWF%wF;g2pfM;p*a2p`U2b@xweBJyupNt4!NiQ#>$8IW=x~ zj+j3k?SDDe_&Bb{?V!$J(31A<%X8DZ)^uj)LwcU00NX^Pk9a{;-7PiOSV4574Sg8$9yXLaLM zG35K9mZ;^kuzvU79$LqgVhK(zIUx`47Jk6~m9qxhA*$GL+=y!qOn3w?#LC?*US8Yw)zgn0PCMiPRp zoA3u%U%CU%icttnj%DZJEY#bs3Ks@B_~tl6821_FMB!qpQ zHOU~7hpdy{PFSJhOHo;;{>p7i^VlY+AEfa%WLOWeX3VHQ^0hDa_Pc?)rzb5PQ)!V+ zPms1Y*Xjb22=R9EbPS`wEzx6cVGP1lJ}tuHzmN0U<|iInmN~g<$lcvbKtpr6O_PwT zqx$y^L)1kIyYw{Q&)JUjO@L}`;);((HF7?lW)7HGXk}vh-r-eh7Ui;j&rUR@Nn-;_ z`NxfxmxBE!)kMh0p`nlLA>v1_WmBI$yUaujLCXyD72(-X!D zdk(gFE6=f<$I6x0xzVpc2)O~+8wRTky*7oOHl2>ArEc_i8Rc-juV&?Ms~7)zqOe~iH(duWES&Sb9) zV#YeCF=XFE!q|%}*`q-ejeU(Fvddc5q{yx;*%Hz)mJqUs>iL=XIp6O&pU>x<&u_2S zzc6Rc<95H@uGj13dV_BBEjMsD4Ctj%VY4$U*RPpW!5ff)SLI`(i0q-8<9YDdWmgf` zRygfWHC;mQ!i>2~2hnI0MY4T`?%;+K_Z^ddAC^K#C?Q+eF5gGXYZHo(lq`Ny4&dhS z=71#i>ym^qly?V+hbe<3N6#rR(SdAe4(CRz?Hft>$D9CCRjr*}Yvt%Ny2M9}$nAqN zqidBZF`oqWwxzU*TC=*9eO?CX`;suBnUX#;b?q&k)<}y3^1wXr0EKM!slbjK5IH@a zMAT~73$<=x!ROy5ev7DlffdgG3Ws0P-(_!a6N)ZoPVZ7111Ei}*{kob3*(_UN+CAK zW6~nlViqKjT%=fi z;ZpT@fBT&ljkv;huO|X6oGNX4RfXuIw3YuJU<&oy`fN+hQ_zTo^muCwoAF|!InM2K z9dl$E^8j^5?q5^4m&)~j>6nD0S^Vwh#Vb3 z@XzvtB@_1)ei-X-T&gX^PcO{UN6+7nLnZf|J`YQ`z_m?0%g!P_)`p1O@4YPv3!3$A zs;a!kqI;Jm*XkN(78dR!C=F?~jd^R$lo@&(PADoj#PuevM%^yH_9voCv zMLPGSdCstg&iic{G7R2myVucGF_XrDBvBQsl(~(9#)g#H**Q8;_8#^ClQX2Sl1uO8 z-QD}Ws&_-wL#VJ41Amo*Kc5|Wv%y8Hi5+nv>&7VRy`r@m>GB~X8<*2DdPh}1BNGao zb*Ui!OIE*^56kj?QodA_&@|HTkY}k>S6w3mlm+NN;F~1h^pMZ^Cfb z+B*IRT`F1F3#;48U2^}Z)Txz<08eL{Y(q$d%Celn5jpvQ=e7U~TL*;V1#?CQ(9T6} zw{ti<(DY_1LZ3Qw83Sxf+wToXc$+o0r;THWLbhef-!yU;!6bM`zI|=HMzQj~88x|n z8a>GU>izAo2|ao;ps=~{_H=sthMqE=-@Nq=rz&Kp_rsh+UN3E>QYexN{by-MpN1dc zIm{WHuVq(R%2(nz7nAh%p{=78%kuN=t2{a9l5l8>WnxKkM*8nbc3Xu}*k2F2izZsI zKUGTUcR1fZU(wg|*Zx)9BxECqml#i@yA&IzSI(XT-YsXnvLN`zJnYvuh(6#?7SdGF zO8cg(k@1L?Fi!5!Fimh4;&~Vj{j*N-8(N#swRz*;rdg`bSpb2|M&$$pHLKViJEbqI zyH#0(qr&4iE^9^+6(*Zb>l(TpvmVJboI%N zSb7q7K>xaR@k6EWcTg0^Wb;kzM*+4j&kT4{2ro?2Bci5HokMZXZExZ(sVkrreq6%U z@LigfsE|uPlCUwN`CIGq%~Au;L6uT!!}xbtCR`}V9T#nBY1 z2b^Rv>i^=S3a$w0-O#tJ7GTFLY32bIO6I~W z9o1|X$FWNxQw^R(J7bRlpW{m>4*D{ZW1GZg*xDkbfF?Yu1gJI*XX|wgPl1CqO zx96Wfvk9{btlaN69R79KL2jBEX=n+bt3Q`Bwl|lgDgC3m=I8eI^n%WcF2|qO5(1-j zyvc$^F@~mlsffiqHA?u@6Y4el>NM1-zOv79zX7Gw7jpX z75nRA-3VE0arY#TY^fjR(*?)C2&XRU_EBr;bF*|$%fy7|8FV<*Ej~;w=fVx7sJN*t zb-Us0@%&QNrl-%KfX2A~fq5)rPKyB|p!p&srmff6VZQNI%6Nr|Uo4C+UCPUCBMu!m z)o-M0Zc{YM)OwQWF>hD>c~w{1Kg`dvdhyBk4+)IHp!9{;?@Db4jyTR0^27CnK$oJw zp3PfZ?f%W3;k%tfh13KrzyhCMGeurxN=f;Iepq^cOy*l`=D>-sWyDK6PD37P#-o%S zc>U?bT5!ix7tg(@;ag0XozJz1Wr`&+B`jB(7j$^Eern@2n~_d~(*E^r9P!o<3BO+0 zMYop)nm`~6?NGh%Qj0{jaF*Sj_}tbvQ#)aWh?=rpK`& z$h(wGX#w?Ny!IOg*yf1zRhBdVHo-)+?{*tNsFCBR{Opndet9arYGVyZZ4Kx5?c?FQ zvhY~5JIx|2>ADE#c`~266iD<tmflj9rF13cG(N?BOZ_#c}uztQKz^n@VHQI zf*>p@E5|_}J~0Shrr{6is0NOpefP8ZaT%U~$%4sgS*W99xVdGiVi?E!iCtS zRL)X62g_cSQSg}4alnjRTUpq%EG|Pqf|N^Hi&5r}}>L)y%g~a^@wfDKqG8-h3%d zVeI_3&RAxp?JImZEdDq^z~&b&wK{FO{%U^LKglPv#;vql)Bd+RUAfH(pGJMg=Z>;a zsmTxZ4-YRIL$5kre$HScBlUAWd8nv|BY%d|SBpMKEUjbdtULGeOTM|6OrLlEHH$WW z$^V_)8Q>YGch`q+iT9+L8yxl5%eEeVi>HLEYjbSe?Ejtg_}@uvG&!*h`?vr5v0PWz z6vSy{FX@KXiUkg8?-)KNrS_ZM2@_)k)(^|#ygqEeIhhGGFo^_(aVY5-VbWSf@O6qY zZ>cAOuMZ1eXuOs&qZrKH6$`Sb2p-Z}nQcFLN|h{pRQdCg|i!Xb}nD z&Mw%x(N`i4GNDM)r#Oz$Kk?5IJSl-(+FQ+!X8g%95|z?64A66TG>qpbzVA-%i!LW| z`fYf2!dag`U~#-U6df!zg34t!LDRFNj7CD$Y5o-ZIL$;rF_k35@qp!yw=}!ouqvRW zJDx$h;vfa|3Xp$_&|-}b(j-lkXm(ADvw$ru2PkPtOJY(eWBlzejxUO#*Iet_jqQ(0 zTu)kgfO1v6B0aVOUr6cI#{hRvxLM_9Ow61jkQa|2txcMInoA02jgVBGOwJ>lT~G~$ z4FhuIX!Y0H-B5WF z_n2%aic@F1X0!xd$?+pkq-q&uNJ#`SdPc!y@dJpN;1Y>3uEX-J`H0G>htdGSg%m?F z=`KQ27fPh*4kb1X0y5$JfQP6NyOAgLVhD_v3no@Z;_8PLG9yITV;0J1Y#zKloAiXj zF*K(QU`|R?UCn{D3}J8vD46-qy3vIG8wxE%0hZvcm1X#2-UIB@?HE^7J9qM=;=U?0 zU#Fcm+dR~cfo+ihb9)59-dPQ{b6Y{uW_Poi14VNUf%k1Lm5>vN*wl7@#mh*ciLU{rs9~Fk6(0+H7!-F@JDpy zhB}KWCzd1s8v3g|+`c?3)yj^(I&;fb?_z&eeZy?)@v!;ZUjuIvl;vLGJ0{0I>lI&W zJlba_N$N}zI@Ak29Hrl*g#Jn&zKgBU+4l*0@_(4Lo<=mV`DeVWJYL;US4PUFoPI*j zi_sVitFmUqRx>+t2+BAJY8(M`ga`0Qp1GHyFPxB7-QQy2r5MDs00bdxJ_H_S`6M@p z1boufv4Tgtn2YrNtmv;XruT9N^ey5>c|C?Rw^s7)1o9wJ*42|YDz$i*lUdD_1p9fq z8+?NAA;AeZTF${3Z=I;QnEKfqyu8O^7m?9a)03yGZKT7J`~5>z#h#v6?5K`~2H)n5 z`xEQRtoDW9gd$YG6kmM;xo`ACwFu5%>uigD`keLhQl2%q6!(vhBxM=$XRK{5>D}SH zp{Gp6#R)X!eg(bxc31p45Ydm!q3?waZTR(bA+#q|1X#`-4rn>Gt#-D2=3+H^y&qF& z^|m=zN)G@}lC5QG?OtmSH(dwI7DW@z1!18rCk&(xg?sriM#zc&rRRks?&BB8#My@$viYzbR9B zU!lU;kTT;M!mW8wFu!E?b5p&`#juBA&GDi^l11`g>wquiX*)ejddzO}F8gCab&k7= z|166KzRTT9yE?jX?-%cKi_(@*sgSii-^b_X*K!2u1#BxkyyS0ox3cmK3KLt|UVMTdk8#yQF=s zI~P+CM#L92@@ zSjp1Oj97sIL6w}`sDj8$J`5UpT!}zC6%=H2hGO8)bBVrkK9c=e{C8~J&H~3x;Nd5((_C2RbGu;AFkn-_3$OX5v8ErWrV!^n?xmC_{yTa;clmCP# zSl>KKgG2iV$|om}MbkefTq?Me-?06U|Hu8O=68@Nu~<~{vyQU~t_IIS8#}aaia~1X8uT*asPt@nWBy@@RN{m~9G%5Ubqh=4V zDTr=jtt2k-?9*-0A01W2`#sZ|Er;Y@UlR7cp;?jmi zy2k5ZmJ1H)0~iT=wcjHfZ8+E`@0T^dO+$3Ux7KQRMJ0CbB@d+1%-#JM z53TLxXj9cu;YNB+~d@_v-G6M+WF!M^ip`%!VT_WBYk z$`Zai6NNnoe+E%+J*m_`GSk|KV8dBdQCLZVJY%=ruuR_S|O@(#u3~S;5-w3H*ooG3xLY9WAwJcTRQq>UL-R-rBQeOckJdPDJY9tx#D#$#zoS^mC{**-!$L!cWOGH}XU?S;1F%{k7FpKj)uS<(a zM9&-+e*}?V%kgw-&sVJNQOtamS8Pyg>WNaqRnJLN>}oGTWz`l zx|qS$+>ViXm!-!>y5RjnNNOX6By8&;4In)1Q*fR^B#JW9eO)i98V}tzq7l0gmtJSGh<)&`wJ%}=_Nu=YH8-7N@A8T(3t87tJvgtw;-D7K*ssx7^GUuW zCaxc8shgS+Nb3z47Hy`MpbsO)eY1YF3 z8>-=de!-XSO91BA`Rh^n?G5zQ?Obh; z(BJiC=(6jmSn*oODiL0jL~CgY@d*L^3?S-q8krK1Hu_xP*0S8d7Rme<>cga$b(+m# zpA;2FPu@*%M7IcU^z%fu(Oe^eCm z*+?#of1x48-cH87YO>HZ4hj&p6n`0Q4Y=c_*_PCLSW7qzsUNd!$0wmH}pPSjB+d75btAw%rk4Rr&wu^wz0@BmkQ*yKW^-=5t_K%M_{}ors7A8h)Nd3%fpsuZ zt9}x7`o%S5*RWK(p}SdAmNjNJCjdNB|NmEw5qw;AC&F?Z< zL)b%j=YC&0m)LuhLtCnT_d4HEdElr2+S?!2(7A4FE7q5tDjU$q-O!Wwm{q+}^xc}B ztJf8YY*RNj#=rHo7)VE8?tPGnLdBrk@|3E3!uFD?O<`byA|z77>T2bY^}JlM$luu) zckOR&$wn?G0v%UyDJHP=T|_-oNy*n_L4aN2G-W>h zIG39g0+)}z07Puu=c_dj8Y?_LEcEExb$#oMctBbF5gIFw{flRwEt4uTwOrM$FxMqLySWtR- znttvXyxNOhp`fXa(ji*dk#!hd|Kr)xQ}(q&oAsyOPOPY(0@kF!Udy-g%+dY(gm-te ze0ZM(Ca0j@Gz6ZDy*au)n6l;xDK)()Q=`47yEhTqcik-O^j8#rtv=ajF1M9yxf8$F zUpcBjG12%VhtWDECDIX1KQ@Cjdo@05vug>IXt@w|ukd)f7mJLjB$cTQ?R^X@oQeqEd$D36#*MH> zkojLG^h?yjVpDLd2Ld(0y2wss}eyD;~ zPO(Vu_hN=p0=A6SoLr@Yj!wr#X|A_iM@7`^$GQ4)pl&irb)2h}#^hti3a(cxRn;w+ zuTQ7g@bb&u3S}fUi1@N$!a>J=U&iC4yHnHH6*jTl#cVmQv`$(Kfk8*g`|D_vBHN>= zG&VgL((wxQI`L&RnGZo@w&>r!*s#wnJxPwG=cx69amC4L+{Q}Qb6x-U5)3~o2_iK4 zmR#uontj89DCRlu7T9r~#EE%E@SGweg81=EuHQ|Ck5>d=<9QV1WErvIM4+@4Nm$2s zZhsq8S4fa!#5(nNNI&xU0tWsA!cv$UeONo6A6Hrs5+4i78KTQ(;-Q;AJlR1%aG{+% z!2BY1wtQyNrNQ1CI%;6i6Ub7Ny5S}N4B`*GD;Wa^)AqIe2BU?Zy&LfT1fAn204K-C z1pgF{PkN9YFqR4HQ2Y+(xe?kImcSb~A2aX+gxS6gKv_FEYZnfi#J+siMgXWllfJ`SN(=RT|fG3y5SYY6(vJ~s5L*Di3OrM&M*>KqRVr#dEY(n@Jk z%baEWCC!!8m0JL?FxsiN7DC9ZtfZhy9B|gXTVX6k;=Ha3gOH!EVL2R&w@Le(y`bip zn3#!z$k_12!!wWMh_(>M5$R+^b5X-1tBq}6L5^IN)9~2&Lc=jqx?&hXYSc4TueoHeKGCecW zjve@Ss0^hK9+CD5z4rzJt%#rXRXFSg=1mPSxT zJG81R_CGteJd0KKdk)Zew?t2lLhD`hI3weF|+^>TF%nf7La7w7VcPAV9?lY-5e5{%^?Is~fo@Y>eb;RbNG=4D+ zhDZ~$Y*nKweszw+fBrnEGbiY|E6ar)$3zgRpnwE=CC_4e>0;@M4+U?{Q@Ug&FFqmG zv_6=r@L96V0eWHvAPMzPS?_SSw|Oc=)86mJ6{7+=k|MQQ&|3bJ;v3g2xl>fqLWOOA z)}@55%H7~(KK-H)3U?lcQD@fF&ZirSU>w;#*3>_eORZ<#*>Om9Dlg4{q~*w653s$B zy!!FSn|rgD4c3LUh1y0QerV6TK!uXI1^B6 ze4}B8QSxnemGF_9s1H%&FYSv8Bs`WuZ_;#4&7b=68{kX(P1jA4@+v4~;Fg|Eaq8ry zE}Vg|(?8zcI|UcoP--uQmJ#S3hswYC?}B8FZCjL{vo`yry;*GKQDqp-ai4I&G_+Li z%ywK@-2CQQb!I-2n73*&pU_kh;*YA7+q?1SE4AsqEn_BcG&&AYZq_?2<#jWk`Fa2M zSL!!n6)sfKp+C=eVB_Nko5sikU#=i3W7XybPu^#AB0Pb;)@ETs1gdhTlW+04f_F`W zvz43`-zKJ;{#ZXqe^n?i0~vMTZzy)NR5iL`25ATrVVa*`BPHbR02}fU!zua)V&uq7 zN$GMHY!zoAaL)U#(AU+>n^b45p`|zxWh02M2XqEc{k2L2LsL`YVB4U+4NsF-j{XrU z4^P*FIpvM!4;;~@^qg}YstHccQu55gft6*qUOqPKIq22bl1Rcy0YbLS#C8gXo|+3r>BNe#H>o9g zFJ2@{<1XaVx;u(jJ`&1n04c{Dr{q!`9CiWDWhNbG+Qk&sT+|_PFVqzL>aH7=V}kU8f85I8#R&w+=f=_ zLn-$QnJ1MLR{bWB2PK0*bdO8Wj$t3L{7*5VI4FcBua3EXVAdG$e--NpA49xOf||R>sLn;^63~@;8F_cbg3wlw|MrG!j^GH^U9h=kr(32 zc5=_SqvgI_^*O;3Z_$^d62_N3&%bH_|Ekk&r@Q*Cg zO(sAfNulee40L7o?qr%5{(7uVCeTvH8$_ozk@YXrNW%z(>9r-js+IS8G%U$Z`8r5X z;7y~!iBgN#<5nAMMnyomb(J`?1GzLhUq|TP0T34p-XFOZtXmJ-~20dKIazCC&y-v}TR!(c%{A~{NMF&i0ds>eC z2jr>Kj@vUBjMj-Yrg|2fWl9R@tWCM+I4uq!O z=9Tn2&9WL z?PY7iBIA`RISJM5DEsTr%tx4y4r*V2c$E|QMTI_8ZZs4=!1(FR zeS0d_Bjbix{LSTl)}G`K#R;z5f2Y?+6=zEf~ z3B9s1H|ll{gK&33L`P56^APya!BM&W0&OFJNQ;boF|A;Q6ANWuYER*EWrNlF+qZl};^tA8K`^ zMi)(G^;W&ROmh|a#pPeT%o(HmW?XX}c9Z92g<7=KM%AX+lL4(i4voJrfa()^*uf@8 zFk35HVp%TStvt!(Gl&L;!oV#8BzC&xiGE}Jd0x1NWT}yTQk@qt@Z2b2Y2VLk5JRnT zh#tRs(_$`zCDKY_#~hAs?VPt(;n4k zn2MzabX}T>8E+H5aO)!vgPv%duX@FXVrb+mt@rdw)buW%fvFqOTKyHgwt5RoyT>Ur zJ64oMzot0Vm|V_h@_)(^0=TflN(ikvdhUnkkY>CuemQD78;C|lW0voi@v*)lmg3Ys zPgMP@@ax0Z1ml!fUiV&x1>!48hzD~;SiP`1Ds+ynL`UK}T{Ne2LEB6$Uh9G^_?Ssy?y*U z*~^};O4%=g9?eR}EL1qwJ9w#etyGylDyiU)nIm9_J~aWqnujmbVsmZI^3d?rB332} z?Sn4v8@c1tR>(-X3!^S%{FL?WwD1-I>gdfM1pAm{9}aHSNC6ok572%b59Us+f5{Eniqmww<}?Akyu>J7!|?itO^B$=ufp zLVxOBYKVviM=P98S>4M?3uZhj#E$>!y1jB#0S1c_Xp9cOZzEfU{8}V}a4CR#rbDNn zq8nhJ9ls>>;KI4Y+V&8k%qin&Fas3Q;q*I|&rn|JT#NvPF#bBLTUX*bj}_s6ZBR3N zXW#!gg0g$k3Qq2Obr-bH%-bMjj7FcD!^Ef2gXJ$$n1{tk?Ee9Y_+mj_o-ra@k9(b0 zeS+#FVuu|M-7=@+@DHt`yDu1dm%o8e7dC8Z?ACkpNz}Gy;T7%9h@G5p(+bMm^~>`a zs(KQIiVb0bGyrDK^qre=a(KvI6<79v>GKssq%u||%p!#Hpwn;CET)lDZ*4Y(Tf006 z-zT-fBq7wgXdO`dVD!1ct8Zi$^de$iIrJ|_)sMY6-S{gl`hwU3A=^N93q?` zp6=**()rDGeRHBRVLUZE_!FSsR>+|9_^gMJWpgccxnfo2TH0(inRe>JRAlM5a9=UpZ@k}Q@zn>c+zpw_N~L+OM3@L zRs6i2tiLdQ6RCPW9E39Su329dv#pa*T0d9fUK^al-XtjQBot}MbX~wmo{2`~%?Mn1 zb65sHl+`;dSGd1yLI)@Br_O{&cq!7NsMP&RBMyf{0h&4vDM!EPL=7x+b(TrWqQ;*S z+4Y1+a>J-lp++JB?n01NHHSWiT}1!@=H!jVrL+#9kjaIWW4ze4Prh5QDiq?=#ASa7 zPxfNG$_%MBw0lJKo1BvCGQK@t-PNy1oNl>X$gtR9c>j%R&W`trJRO)s`EtqU6CSy{ zZkJ_X8pE*66=U@%yiq#sWqAsP0U9}cP?E5q#wnmUFHVt zAoVmT@|a#42H}PaoSjk4%45|}eLW~$O=v5>HLv&D2fHRgt4)&%IFIP7ewR}Q-WCxB zSy`EXua&w)@h^{RIs4kH6rS2-A;P4-n_Q{^OD%Fu6VnMlEShIpjqS;iEf!=_k8BeA zkosg=k5hH++@`5vz%#!lY(M0xim~_@I|y9CqfR7#INMJ#3bRPR&VTJGGEc}LR;YA=ChJCF}USA}^t zxb?x%5xV3x2NK`H5C+LW>=${iZ#ZKmqvg4rQV5Sq9o4T4WGsWt&<$w(z`+wf$9a#P z6sP_9Cs;*p=@Q$|clr9>9b9x=J>SHCEz28;V$DKQQoeOm?v3;&`GzbNTyz8bOpIa< zDxMZ)619>I4y&HF-iPmDupIPPOf9lJdAqErU#1)LB>UN*Pb5T6r#zGt3!6vYx>$;U zk;$-c1;(Yacf|F8^rR+**GRL_SK@|3Sdj5)vSoI;q_@ zR|y6@LNsbK_b%5!4_HB?SA<;P^(P!p4-XjQA9-a~0TEQcgO%}>6M!F-u{PAdWHf`C zXLf38IxO$%2`RJ9*q`zX6DWWB>zxd&kspXQoX|)MDKmx7G?osRGiRK=dn;VH2G5KT ziXkNcIM$L33Bg_*qY5P=d}5B?k<2yk>-jipWMUgr$l!J|vSjhxv=@_{^<3U5uSjqM zX^@;kd3b+(+ui@Ms;+ZLAq2394j={K?GIOINCBK&+oial`~^&g-z*y@linu+B7p&` zG8ujoaN3s6L^%HU;6EUk#j{l-T3>*i0v%xBGUoo#rx^n4FI23MSWn8+v!Zn&EBHC5k=X z#4ANxw02>Za;l6pw4VD+mf9$@qW)AM(@{}-ul>Ft6r*BXuH5#+zx840VoXVVF69p# z@UUJ#!4Gk(kPk|4a(z4x9Z-Xh%TzqzbLMz3zIxRe*xbR=wS3aJJ_z5}AMNyR6QPz~ zC9}0-2mVOuT!u!G4@SKJ*4o>EcjuP&5CE+GDz5*q7@_WYF78%)xEm3+$~f=a6Mj_A zXHF8`{Gxb|o?+H_aXbyDE;n-wH=(;nh;i+btLyS-wRTVZVC9S3Lp>W>#0l$B=4QcJvTz@SMf2_#lL2I8{O|9qH zwRn;PL#vq!+6iQs(J}k_GeY@?ClSgj-ji6#nb@egbz6f3z3-DraaBVlnCmIxyGgpX z#!kh=UfxSRZZh!*%?~{mmhoEuKhEXFp^NlESVigg}Zs_svr zLg1E2Gy>E066IJ>Vh+|8Grk)>HS_LnCGBlH?5-mrQE(Wf1RQo1vU+MI-Q-Mu<11B~ zr?zX=B>DDb$55m-tm?Ls5eKG_ImYkf*t1J1M%aoI82w@S75eCY^cI$nbJ<-2TAl=) zJcI|BO47TNkIQSdiL~dKt_a1cUcF+)g0t;(T;9cho+GmHu1HiCZIF@Rsi&(c4OB@j zx|GDWlRV_<_)j%*i!sd>Qnk^UR>EO#+SjSowXR2!JgL{E_DY(!B>q0y!9H3R@(WPCwr^K%^7jmWMZG!QM z)_OsB?j5Ns>6xPeSgv~jH7(-0_|3_JuRF#wT9oC#Dr57f;bVM=uQ3zINyAnl+PU*p z_M#lHWAIQ)Uh87VIvXdCn?0En*~Gpb9vP@%lk_%qywI0X7fQ3!N7s>}tVCdHxsTy= zSh^Bprv$9$Tz+|bm|N4&dAY6up$`q0Xv-Tp;*QZAnr%RH%Q9Z*^P)_1cK7<~D`L49 zwdOwGicURi-fpK@t1&&^q0hq~D&w-dd5g@q?T&dYkwflE8dgbehWS*$K^Sv zPG1ZL7omr>6pc#gG#-!Vv-*3xv2-3wbZ|Icvo;ks{qu^XD?(1VZz}8eQugu+jpI{1 zT{YkM&JT7siexdpiF)KM?|rD)qfRwdm978^aI_fQbpO2h(9?mhUxG0Qr&jJ+tGJ5- zKB2sRODWgEkFhXBGhm>7NFdPWDume(ID1 zv1??g6IcFl_|(1h;;uyfI|h??Z}o0mnj(MXiLGZzQGc26I|O+{uk?qI6*6_OcB@>J z(E{+46Pa+mz<)QKY5s!nldATs=$aX&ez!=A`*g3bGS?k23muJ2j^169zgA8ko{;su zcQAM*cH*8+i<<%H_w+sgMBl`g6?)W988t+_ePUO@e)x2?WY_fKs_Tgn0dOvI(-Gen z$809F{?O6kmIzrEWpacJ8Qb0n4}yM(j!>7jbYb*VK53&k(gK!EsR3D5LI-m}(Db=h z2OQ4Rxn+*;29dc_aoqu3B%L>6r|!v14YAPUysLx@H?4^qDEO9=y7y0T!=otG1kTm3 zuO~h@Xu^oTD*x7{Bu?)F3vT{DKt*7TB{iGl_!%xjdj8cwYRhnPjgr*)5N|5|q5As1 z50~DmnvcRBi9nYEwLXsdxyQ^UB#&x5u0!r-Zs_@%@`&^?geBFl3s$E=9n2?|1@8=~ z=_kxKPYuEb=p8vqmFtI07@(}ib^2q;g>&i8jhd`v9T_|64|?q_bP|UGHVY;{Bjbgi zky8RD)ICEj9mB%h^3rOaFA9#@`upq9#>b8UK&n4UQb*y&+Pc{>($@hZE%!8WGe&Oy zz2r_VBSX{xZ3sTfF9fF-bZF|Q(^vTWLPV=XK8@V;kEj6Mj2h^h*eq!L*6zy$DR|Z^ z>O9WLxg4nZ*DH&2vn^^a7w}wcjusHqg8#c8t<~~+D?2z{QR)}hcs0H-W35X&D`oVKCJGl?=g1;WQ z*Id(r?#&@;C&nhI(7XErhhR0~J|Z%82!RKIbhQ>@fLr&UyjYDPIt;ElI)`Y%K z=RjjOTHyg$Ht;#JvU&?Y&SF5oxJ{fautLb`P+xzhVFP+B{Bw+s)8N4sVBc)#E*=XRwNk^riBIke#cXl2 zJmAe38o&MIoa5HH!&2F~h)U(NPV zoH{>s^Q*yKCI}Ca+SGdF@k33I%a37HSZ>B|%l@zP@&8e5MDSI4Wd}*RpN<^u)=qHVYa>G zaNV(r2*06v=u!$FVZ=X_S5lqX!O%L{e^09yUy@i{X5>M4cO`+6+?T=1r>P|MzAq9h zMNm@~P6Q-|iI#f9QuLsH*tQ3{(^0?R&p0nDspX^X{=LS|OUO3|%|ax>3z z2pXYjdl@|SD|7^ZI+pktW7WCdS(G9j)uh*Z6+0hBGu2@^y8T0e4?9Aw1Cg&tg;a25#gNIpJ)%gN=yR|lN z**gU}MjK&PTc0sEzhhUw^RhX6IPs?4PJq71*MJaL0c-Tpj-m9;fJXN4iDI)G>_(x( zGH~Diy2>if4<&^6wd$U}Mgim_oKfvC z1`*QXIahBrcroW}N=M{RsjA3Z`VPJvwm+ z3vCbPJv=UCS$GiXw)Ke?^KBi+B_iWOjVKysG_H3^3MwP=m(S?pZ;LIcltk8lfIJ*d zGA*RLA+)=HyrD8>%zf)pb9S)8Jj$GlPK}i#?6tQ~S3E?T(YVIH0PPxa{=T=^_W#A& zo5n-g|8Lyq7-o=&sF@7LQr2W|2E!l>+4n6M#vYoXtkDcnX^btBCQI2P6ycH@k*zG* zvt5(HgruyOQrCU{?uYm5bwB$*oSBygX3q2cem~#iINrxWbY7>HJMY|g+;jO`K3VVb zmE59-EFxaqXARD+7qiOG@a+Q{z97xV-xojtEm?C2cEb|5R9fJhz5&wGHHse!;G(SiO^QED{=b6rfm?)&p{0*DNx}?fgtPV# zL@eh~a2c)u>~!uJtyKAug!C?C9P)172*1PjVZy!<4fl?ZL(-_Y4 zhNT6L&k*@93frEo{r>^cj|Ge{{DK;hinj~YkM4UGTRA0N=9*T4;4DfACJIVIpRN6z zZ*wHqCn;g^u->x=qdekxeW}IHs>n!kW?1@-g0HvgC6g%^YSX!PRP=%hO&$B@|Myt9 zBN!`pOpo5Q!QAQs;gFqEsE4;d_hXiRxTxH#QDoO!Ig7uQpJ1@1W7&n{FL_*-;Zg%- zM<~hD^4sMsJ*5T&hrS~=o1EmJbRMm-i>-lpIGEf>sPxDvJKD3*@lIKZ)AEe3LiyQ{ z4WR_a5sw;*T{pX^s!la-lusQJ4BjdZmZwe?DMGwW9f}jyoy+a2qMyY+4S?BVgy_%!%6aE7N+w9&EDpJz!P?TmcsuxHU9tmx=xv=`#SyAWO%qoiQ5zo zC^=#|{gv#HAZ%p^w~W@C5Qyyskz)up_t;6Y1g08F-l;iOS!kzIL?&SOHZ9AFNJq~= zJE03@;L+0ls(m%$ONhAy!O-K_ap^E>meOvUQrDZwYBa!sOFn*P49w=s7dFY7X>L8w z&SKIpGgM;!WF5P3aTLH~mj@zkJ7IP?z z0vV-eMfRPQq0H~D%|6h2^y_xn+< zhKHHoi_ASV^Hs3=ehhvprQ^?D$PtAem;z!u873M(B2MkmEo$N_3-1w^Y@)f!B6<{n zXgyH@!NZBThcxnVnA3t!uhyqiI=i!(TMEMO*Rr?UWty())s4i3gKo@t`)V0oiO0`( z9{d?j68(-~ow~UzZtl;AK1SYHw~0mw;NfrTbaVhfSF%~bB*&==v6Nl-kvH@ufP=sJ z*WuWa??)I(qGf;v;YuV(b|QS~T2p}33dTtOlc$O1GJ?aKah+1<*9(c%k)xkVU{^2$ zG{sG$o?!<}|AvW3(hnAyI`*TY1r0N_gqdbxOU$@hykB`WmBpF93SlZOs)81 z1HbC_2u*tE=?mM*i@35K`OLSqFTcNtTQI+0oF%l57iB+s)p@gXf74-5EA8AXS9!Yb z*oNgnaU8gns}|gCvDeRPdPrAmMU-)$80|xB#Sp(Owh5uUEM#W!!&zGjN(tMSy5$Zc zVHhz9* zQ10d>K?yH?6$%Ra`?Cc~{ROn8vcolRiT`QtPGxCU!%!3*sPqw+91~v_<4r1R`dyAz z$x8E!4}6;S#}1X7yB+W78fz@769qE%gI3fO^D$kVM$qnBTSV)>b5_S+}cG)3eWU8i3cQtO_5M>ft{tvn~s0yM5@z#_G|0Bzf)lErWH$3E^kn80_Q;H zF2l8`yFI|qsGnGL`Nfa3bLT=I6G#J}P&?0zWzPHS@M#q-%Y3y0dXW+ZV;>(rgHLHq z*bxS9L8~Q9Eh*=Jwy;Pg`!yGGRF|)(q4sBKW1k>;4N>zJRh{T>Dp6zUa#!y8;)i|W z!166K9h%^Dxu2^5rWdJ25p?@ev=*l&+}NJLDF?7l)Cu$Jfe+vO&P_2y>$ZkSC-1z##Xy3P)eiTyn(y@kfa$(poEETkrv>>wkz5$^_8p_D zbG0zmxYiZAU8dx@EE}@Mw?Yvv&WffL-Gb#c4CfumG4ybW4)YJ1p}ijc1O!q}N%Ubm zaTt%t*0cK-?B`>a4PO4m=Ln%J#cFEDLY^UP3Bxc=bSb?U#|s>mOW+^y}p;KlVkR^*wUs- ztelAFkIq%>6XkUh8WLYSsweKp#>i(|Un=*s;=Oh!WIhjVg zTD?rGr9YyO3lBsBujfSVlfAC>$*f)Kf|J?h(KPlAq+?69_CV-BmtA)M(B?yJMq0WS zn7@Aef4v|>J_-yxS!h}0HcJ_8*f;=>HLvmcXfiTGiIRLnhUFLAp$OzpCriQj9}XO- zDF0&h<~#kLv_D!OTaah&LFgy72|T>WJIuav`;_rSg&Cu7@JR+n;>Ib&QRf*_VyMwn zc434q=_n4n#=BV<%;!{vO6h+}Y7l(-YNsZoDJa75REv<8aQmt?-}6Jk*^Um}Whpm& z;a$mgicB-hOUH)K6UKcWez3kIr&41ALfKGueZ$^@^i6Y`+JFT6Vs2qj6))J z$~yf!u`+ok416b-3V z@7CT9i`K1HdC|E>#ITgr2<+Mi-|Xj@pRSrDfs=ao8=t-PvD}Iz7v&AE*+qJkmp(dC zj47IsA`f9#noLN#L6dFPuqB>g&#EO?>*K#ib=iegU(Xl@5AnG`+sXp?=zLz+a0+a> z1_5)gQLTqrS216BH>_YN%$i_akJ5Rc_sk(BtEWMs+ZzNXC;O)zw{#j_Y9QC|Kj5U+ z9uVEjwvbDaItR-->N}+$f7bQN!{T-a2I$M#w(@Px#Kqh3`04D!$7G&BCCEF~|Ga@4 zGz(fF7jw7>d&P@g&j0df@wpi6G8nDBmT#aUYVgTIsNY=9$*TpuVJjV%J1@)yKh~06 zl1FMU>HL#u7L~p3eV28lV9Fr}9tf7DxSPdKKt74bY-6D$!{|CKO>RS3NnNNt-xHCD z{Vz&g??qm{;5&^4!iJ?%xfFT@GSK3Dk~e?PG@pI-7`zVrfBtEDpK$ZIAo5MQ;w9yzy{^V$cESNKCwry?6^;;svOIe;>p>4Xjb|U0-6hC}_$4y3)o{lsQ zbIeMTXZU!-&K(|^1W8NBpO`P6RnUETec^FCVL`EOL1Pq+8)y9y>yJ%ceE818NxEQz zxH(e@Z=CZiSU}uWR`kZhax8K`4B2|$`{JT7sW4;go@Oo)k_kE2b@*xBp^lvPWYv>4TxhSm{O zI!SG)Z2ZC+dvc4vH|ZoQ-Nl7VW*|#}nCv#@VI(I_Y@Lqk=PLMkg_?BZ9!GSd`S-$s zgp(kT-JP;*nLEcme88_IjFx3LHTJLczs+s|b$jj+HFShmv`l6CC$W1-c%duiiP2qw zs?%^XQfpQG{hhAi_NX7(V>6|0#UI$n%3+A`_tMvos;>cFwSw>Y`&eM0`SNj3x*Byx zU*uJCl!vdssn2;1D%)%5jeeEq=E>QyW^GUvPPeiUSq^xwqmkfh$j?zVO8`3_>Z3PF zvUBgEa(8PQQZh=M@gO1U8DaJG%ag;Ekk zTFt$-Gu{?{Xa~_I9g6S3*-gkHKCMC!Gu!KRN(Jqo`1v*vJ^b%jn#4~OyBxamt2brD zQU4|CH+yWEwMao>I1!&$)4kK9m;8fZ*~T310CN(_RIq*PLwlNj|2h0~58FSUol z#bOC|{2Hf5ZC@{oN`~=Py;Q~oaDF@vp^?8LRbo+BOO7$GA7p-LcTFiw>S_{FcZ&ub zOb|4I7^Dp2lgnoc=U-3JG^Hv5(E;+erU<*Pl?c7XjMm(zU4eu8>mrN_1#v%Ca z2muJ>E|}XDLgx+txLU$H^i@qdmXnTmH_pOW5{A4aD8eIjCAIF{c5wZZF^zWVN5`Z1 z?K@(wbL93ULZ#8Fu~3}iCwVOY!lx3R(|4GGu*Th%=;n0ULt94S1P^6YAQkqd-j6edLfefmi?Apt_&{FLQOFsDl)8k}*mh zwh7HTLZv0$D^XkGzk=QcgS!U<85I<<7e$)JnT!CEg5HZETefk7-W@qR?sUR1ZtX*u z5eo>dQDX(Y4jm`P4Tp^x2YmUfO<&5Rv|i$S&%a;DQ{|WNPDF0vuVzoTOEU=)uaX5r z8f77#63?SNPJDHX%688tf&|j|1tU6kP)^gx3~}`^q?b%&Iu`L5(__~imAB`j@cezt zH{MbZ;cMU)0d;MYAdxwbTBOL~T>s~!+U*4=ui z_HXzSG)H}b7VZnCPFD))i0j>%c*W&50J6hX%k7)eB-~pr9fpc`WE@W>mBsT%d>*TO z$paX7P9Js?e`gP#=NGN{j{AG;d-?Mcnj?RekBwV&$ps$~A+5~h(!aB3%7f%(5n@6g zY$k?;nx(S|ZAgR|q#rerof>FtV~(tHCt#R>9e3!o{?1O_5kI({T4(_i2^9;s7S}!yQ)A8@Pi;K-oHZAN0(c^iQ{{1J@ON7ZKZKhg!HeK@T4m|$ykcxe@rV5z53DY z3oS>4rU4T&*qKeZ1^GMNun}qW7}z;~y^>P?(g? zKOSFYTMY>AL>CD&v&;0q3+Dg*`N(-mngaC&`D-E zhxpm6p&E8%>JZ~AhNmZ@fB!H`Q7&6wi7dV!nxLhA!8q;GU{2IRj)LsD95=}7g{P>j zJF-ztMK?ifO@N4}eG!4jt&JzejNW}((Umr9ok4olDBlua%XlFI_{BT-IVUfj>p;ujTDC$3br95E(JNAw z6;G$}{`9*a!HW)BAFZ;^kgJB)--yo=?(82ft?C`zj^taV$Ce!opFPA0)d3dLJg*)5 zzJ}Dy?Xmw2fhtRs32zqVDI)A1n%ro?3#T;3sG*C8?!A7ilfv9B`yWQEzI4@oX%C#k z?4T{x*g#-Ubi%j7C z5kf3SrC3e3${lBn5b<%JUpMiU#;8J0-}wfqIM$vC>Ax9^PyB&+ z;f+bK5iCzmK`KKFMRyXq;&H*gDYvLnR8f9;+RWSohzlW=b|alJ|g!UzO;hTZKLtGpye0CLXzJ-cR46CbPOf^e*`{0sgFGwjKJK(i{XB zqb?1Q6F33YAETKZc`vn!7s|QNKlCxXsG{xBVz`5Y?SLRwyD{2LS)`7lTi83s7W&)f zfqdOXXyiC=Ka(T+$?g7u6LFz<9{=l=!##(nrX(%q9V72btVH#)4HS}DwQhT~r38vo z$iDM4)5AwgS!}s+gV0xoYJrZHv68#b|3W*N68Z&BO59nYa1p(uIrWbAS_M74%o3ss zBKKcAoqQ-1Q^xJCH}tW86K{UDHP=RUjz>I1G#|PG^0YOpr(=;nCtWI`5*lnO1xg;g z{NZNLVGd4N`+bu`3HpiyU^tpPEvTFB!mSRU%xV0es{DVu)M{US_xmGe?Xppl*LU3s zPcg1fQ7)YlA*GqlX9uya$aO*)Bsun{6V;q(WpW?+4)dcYu^K*_MPw9~aWNn{2_xci zkRC+>KrK?oBX&!;y*0q(5gGCv0-gYq8Inzp0C0O|35NSnY=(0|rwcTOGxIdLt0+(D zCGXj6r<>l6GT+M@;MHXvAMzcSqFP3?rI!C<2x6W>_p|;!kFI`Fg7Oc~XDv2K_h@wU z4~e@(+Z6a64K{NPW)4~+uOVM#S-Twtx%(YP$7OZYjDAkgs-T&FI@AS;TlMmYl|p`Y zwbmlb*K*rpi&GizqcBs+o!LW6jx>k1K9NnhUXrqINs1&urG_Xck-c7vg`%5l_?_MI zepXIfuTL=l`6ZSeR|6A{UC9&rFdw4d;ECxk71I$h0>x#P2vi#*@rfIZz~k+71ESt3 z0_vST33kxFGY0OIO7s7P@%OYoP_Tx13Q#a90chk~MNLhXJWw-W1mQxXnTR2abu0e^ ze2)V3xZ|zJ?|Shrfe*1S+h#x{N_%EBUVCWX%(s{XuQP2vB0+PUuXHRj8^o#zzI#ubR)HE}Kw%`b#K-^CCUu2ddRKGBj!c2uxPOb!C*kY2A z1?%gD^Jh%mZz}la7``oCLS3!wjyZ-O)`HX@0bqQAPd^^*&ABM}_Bj#9 z%Mp8eJVJIZNpvLhukUxu_wELNJt8V@BSiM3Q#bk(h?3&e?1QP}tRu;jxt_4_kFTAC z;=e}dcS)jJycHWxx6pj&jGwW{`8}Z)vS$?LS$t=(04LnJ#p3)+Xhg4}<@L_@=!2H% z!PBkJY_juOV(4>(gVpvrl^?3hS5%VC4bQLIGv)YaF9sRv;6l!9>kl0{o>`dM`i;q<8vC4phIloT@@+g9kH2`&FrOMLyiB@c@-eIR zZnf08_-FoEY6rI^GV3l`&ptSuP~Z4WShc$J+Gl92=s|U>TeW_U7;>KRW78(t)tT7J zX{|;zI6T+)|Av{tRv!Ih{_R!AY}(hKc6(9|9Ur!I09k{gdwo_F@=CRy8K0_E)n1(E z$B`{o-kpAUZswGJgLrYwIpb^@kYvI%?z6FDW!z|{Xd$Te; zQn6y|4xh|C->ck*=Qv$^=cUm~%lH6Ws3^F^y_9CBIaAw6Tsh&d&I;#t=6%Zj4Hfyn ztGm7Jh6<_mcWAH^Dl46U3M_4N@G;LhTTjR16oWO-tSwtVz2oO<_H}||pgwMN+F6a4 zDO*3<{%b4iQ@?z++%2?k!mp|Ov9C<`m`~et8It1*%&5LrD3IAzcj+z!jxo%0ynQ8? z`$y658kQYKxhH-UZzI%)S&OgJBRf*lhRe0x{sUkJ_uvBVX+{VX*$FxuvRb6)Xl`|N z?@4A}d{L#ArZDY$V(>AUe5`Yygegw*g>q4%xn*dTQKw#}99{8iOjZ|^i*j6k*<`9! zE7idpwd{A+E zqTZ<+5DRp*WJ-EygS;FNs#vzQs6vXG?U2rH*Se33JXzmXE$WN+iNq<1lN(_pm1oN4 zl_CyA1;!x%0m?cg+t0RwXT)#&W{f|;rGdxn-KvcZ zDmPB4@k&RHQY3P-oRe0+g@}_$1oo7^W6Uvw>ujyfCHH+-8<;&!!xPrJpTy8RVtP;}iZ;XB{pUH2`v z;%P4O{c(DvDSxDf+bP&wc;6r(PO%@F+kb~vGP#o2yLvH~hVqvL|86PyWYHo;s?B(b z@yp2pv(?!NZ>ZG4e?cTh+4V62KDLA+@9=opkqEFY-RoM-OQlXHpW<>o2c^MNJXm6J zokx=kzphXnR7>U>kA+oneBQmNVayOvrNgQp%+Z4s7Kgb0Due!Bd8a1p3`crJjTgqs z`#^3S)3kIq8d{c03l4M;&$%IB#~oqWf;H$KN4<`&H?&M>1vdB%ADxeCgp$D>2JIoa z{Y~BZr31ViV^@pgpjqU)P?(NkMTP)3lL?{lvy_laBr`a+;=7POKevW`-a#g!_Ic$E zA@KIAbHA_jJ_jrX9&H7F?UnV=%{HPheRxg3@yNurId}$Vd8untV#2I3fJf9G|9^+{ zw1g2v^W*T^p2FK8&41Ko!-*|_s-Wbu;HhqUB;lANf3ZS98SKbwlK#z*S z^QVgx@=KB*eS6r6y+@pxXB zKV*xnAyMDdx!aW-2iJU_RQ+oliIA8`V11`*;=VpuRY@}O@uLcIbE4Py{Dkxq6(9*$ zd+Ss`HmT2Ne=$`PPv(7poVsdR(7Cx=Nv!^BJl%O{6kvgoWnT3nK}-oJcy?rzI#n-9 zL!gxl0q^KyyO84~xQd32xu_0HG^MvgRf4>=bP9-03NLeNZvoWQlr%KykX|v>5lk

    {p~#Pdtxvm&wX0nv!q7`Z*>?WQqHO#sz^ap(V52gPV#e%=11>G45C z%jWjY-ioss$?|qiZWnnR1S6{(tVU?1lKMQQ0a9A5T*aovm>oQD(`;cu=I*B3P0xXf zI$8mDSzKMkf;%z^)9eMHnoAf$J^lhf2`gJ3dEcA@&3geF>)}ZL&ny(Q! zlb~eTi)(}FpF7V7N+yjOWMqi;S6k4BizJytDY5Zk)oant-VO%QkG{VvIL-5yR$8a0ATu@gxNAyi*Cr>;-^R4p6L}vc+eYB~$3+n*$ zu&MT*0TJtdhebW?k&7-Ktr~2=XT{KduFdu;^UZ}R**Dul z!E2(x5YFp-ySCbj^s?ObfPBpf*Wbjnj=AHaO*>KxurLnvLwoWd2yupTi;bzDqK`v; zqu^AJ!9Xp1a>8*}4i+%=zk(9v_*d97aUorqNrZ$+zIau*tIIOBc@`V=wyMoI#tonx z!pxc9#orVzb@pz^GuuBvigL*y+}F~oYDhkB4%y1qBHh+uhlhB0Juu_n3t7RT{=y8x zG(2nJ0pPX)c>CU2dPxwTqSMd?@Ky+5>-*VqISaNo>{^xmm3NN1EOaiERseNI}pyloNZK?OQ&`_0zQ#JdO zd5M2mIq`8J<(qFB?$O(CB+oQSXbN{YHEc5x3K<)N-qzuoww8=I=xk8zN6C-K0?)h~ z!$oFTZQ>oB<(CV1O%6(?RPlVQH#`>%H%6P?-fJgAQa1m2y*fm9H*_>+Lf!Cy0Bo^e z--?G_L}@s<+Bqg`+p&|xz+9TJl(H{A#I(m9mW6R(6naJ3KVw0ii(mTfMA8)3EXd5_ zp*~AOjG#Ztu2!c=biztOZIO+Yx76W=%dl8{6aHP{!lR(UJAK8KINuFJ z+5c@9s?z>BvMF4X%#h)NMsUZykA`{~!t7I=&$MxgHo+3lFF!8Uz?_iQwUH8`4|>*1 z2`OuVlp#zc6gXo^0>jUZ)#EtbWl9Q|Ep(%KJZY1BUU{;ke1n_!MOH^IxF=yQ9XM^1 z8%>UnES!k&fSEmgcU6v^ys@NQ!0WE2dn1cFokPw<-Vd++dI`_qNYQb&0Uy^-?B9uL zU*4FTptMi+)txwhH+>L;Bb8YPxy_^DV_oHf9<1}8*zP|AvW|RdzjoL2k+ki1yP=8M zE#OgEq^albOOy&GtiqwFi80G5-L^QXUQPALkux63A&~f6p26763jLq% zKR|nNo}$`7u0MZBBsf^*0OvEc-*&KD#_Ary8JfTi+0b@EK8y z^Y9%xR%^St#g@!HY?Vb!h`#9tK%Toy7PQY zK7Wh?H0=T7pE5E+HT_zx7{;T^oh6-NH5)JTs!9=}gJV(61EC}==yv&n;pT+SE1 zx5Y5+CORiYCa%7D#Z3eb0T1hHS7bRq(8-35*Szc71^rb+h7)-_rg*>+*7+{psBhE` z5l!jd8fj{B2977leAEk_1xx!M4!h{arXWNquM#gdCBQnf*8X`}s^a5wygHlhZ z7nf2019&@-%F{$^8|z7D3YqPUKq91jc3H?Zi8h8J(MQm2f9Wq!m|ZbSi&qzUVg{m)Co_&yVTl0cDW|QY4IoN45Ub&AU}M_O}{Zv zXq0R!ZsxAICfW;ZUasxWNP4b*JXtH(`bW#d7|kAJ9?*ImyazgqYN5|iwR2`2m%+?W zssTi49*%LsP~a)28e?Wt)iP|uoX}#7-z!P;RLF=ULFOK@*0pwgGs$zO^;KIHs@ysm zw&8F2*xT1%GO+7qeDMVc=_rquh@ojXCn~}Kky-9VH+$3dxP76krdN8(k#bVQ5_ht31yo7*8 zNZZ9uS%=jPw{!A{FIk~n>z5*RoqH;NkQMl}kiaaRSq8mtp^t$FQx!y`teg^Fa3ak+ zz*wcxy0Cu>(i!ut>2^}ODt{lYju_;VKqB37;t zS1u@Dh=F#KW9!o00E=%Ne9AG`!KZy{d4%&^n`Bx;{;+Lz#KzA}+XgI)Wk9-4DD^j74^HLr$*LCF#pRoQuRwESpmI!+ zoubpv^{dMflo=!IvUFnJ;1-20<%rv9!f91r@ojNS^p!dZVe>{W5Pd?41E%H|lAgvQ z1p;w9S`&q7w^6~WKg!ff4mndwpg6tN@miX&iSCFIEL7y@#o5y)(1HDFkpbPE^Bv5M>VHj#e8uj*NL=JRC?^PH+}8E!g% zpN7H(-8`!T1^IyNm458C50x!Lj_{duDWsMuRSg5vC=XTt0n7^Y|5*zBq-YP*Bl~#M zo^Ivt3&<~w1;6w4_T_+bhScOy6Q8`f1SzQ~sWE3~5!(g2B32Syg4t7Fc<1o&cL6oM z1X^Hs?1y^Mn~`SLLOY%0A-2}8LJ!9R<_6?Qu@TkZ4kb6@Mfn5^77bD3P@d1yqgS-H z{{entn|YYXyH)s3MVEfiWdjGfm_qyZjPZ+|^do%@KWnT3#jP}Z+&_}x%+wm%LNrm4onh1S({}DvrT6 zDdTH*03X4AibZ&KWm;+O*wGMvROHZUx!{*2yuREo8qSi%KpUAvkA+z9`m$8&bMpj+ z;3=#=S*-6aTAwBdVqMh{U2MA$v8QV>H{~?sLcwOEN8T0t4=|F0pwt9H%UqKqY!rQC zo~o)HZ@Ccd7JV|m zeeW~K9F~;9&*@u5AbPqlM^2QmiA1p6%l-GF5tn(f&#FfV%!QHiO;BDL=b0}IdLufs zMHt5`Sg8M%cI=b?@^BxWzD&u7`H!|*hw}jH{qTkBNMfL0?%I{p7vL(bXSKgn+#Ms_ zfqv$#K%B}*^F&tFNfFAs>n}~=Y#mAP5BsloQzqa!Tr0~AhU$RM{>#71se4xL%&O|T zPj0+TNN_5%=r$>uEK)qU1Pk#Ao1ZKadoQ)?=s_j~2nPr4QU!m?9dYDrTv1FjJ~XGi z`_{gkSagy0T@tDzG2`H;(^9Vdw3o&uLdTDXL=uUq1`n3Bf6OaFP3KwK5gLO^Vw>1y z%n$GwM&6M%4n2~&_2WBLlUyh{;9G0kZXwus{Pe~y<5KBi5lAv+ojf$)gpc>!N9NN5 z!liQqkl-xI5@E1jR<9kw+~kYDC=nms&ah#d9%H`M&qdtd3j|P;K`KM-4;8qGnI#*c z0ekAH3CG~S)!rQVI@5p1zmuZzST@4OLBR%<&NVp-q&x9Jch}xfSi~Ssf0ZLpX~o3o zqexK~r-_Upl}-Cf_H=u~GFQHK)Gjwb!t|ybbPVdk-&D_G3(o})@Bd`p)S2rsx3-O! zAqMOv?)_D-Hmi~HrfvNB;F2}6EhqdED6F;IhKK$QB6qJQ^KlhgPx(cgksjTRT20(s zx>2zIu{z2#mdZTw@ZZ(iS)kh+6McS-7xShf$)8FgStxnt(qi6jyB$sPXKM*=dROFD zZB_15tF+%lz5?Ngxs+eQscQ<+P*R0L0oGMPX#EO4!&hEj=FP3~QJDI#O9|(%&2r6` z2>>hi#;6v`dw%DMhYs7(M|t2nCr*gmGT>TBbO}8xXvqVuxjBk1T74u?@>N1n6xwM% z&8xpGiz*s{J1|Wk9~5Y>EIYfAmCQxbK8oK+TVOgwU?Qj}K9qGkT2b@-|| zyV2~E-|1%uB-3VT(tKw)l62qR)Ud0Cf8yBCQ*09%Zp;kZr&0eK@3*8NJ2iYnq?RNR zDQ{}9-M%olWoe1#*2cSw(0_2;tKFA%*)9)|O9-<2#FbZm{vTka`fxUGbH1tNeMZ`{ zGfZG3Jtq2D3=5l~R(p|;v|Vr5HYq0K&AhH(P9)J_DEZ9O`woa^ig^dFRVtrB@BUfeVlJLeDA_-20T7wpxDuo&? zh5ifL7q6OnFFOYUZ&ons=!E6c#6-o_*LWmG-LOKhUpBkYmV>*6^*mQKCq^#V)`HzD zMW(G7(wtP7R8F;MVqy8ml{C?wV0DzALT{m-a3YgQJxK`I*K56D+fp;}X`g@ID&*_* z!r~HZ+eK%;^8@E-08yC-w>&l9I>k^%KDunJ$c#{gfQ`Gb#4t`eHthWqspA zWF+>^BB%pO9=bE1HR&?#-;La#?P%gr-4g$u$7dI#w?tGhlq=e%4p9Mx z@HbQJcZY<5u{6EZ@t1GqOp}ayyi+yL97>39!RL;io`RkYsKhp@1d2F(nxzJn$TB# z+*nJ_jgvebZDbM0#dAVG9UtVT{9V##kjlbH=;9ebA6)j>nsQ^G3o+~<3)N*R`C<_WRd)4G z>P=LCJ*^=I;!dRdN+`yxSp+8?BUe`A$nqUor>!{q*$}_zux&2UCN*}~=H?L6KG5D% z4E~5wJS-7?c&en>w`2UCwtsqGCX6fPPIR24Y7mGRRY=K`54TUiQYR;6B{N`Z&re9& zc9{@puaLbncg}#dQN55ht_f-PuGKfncYXlPb<9& zF-#+J4hzel(x7&X(miO(v4K0BH&6PJKyP%m?w6Xj!o*f z;!M(abJZXRp$R2Ig4&Tf;|64gP~(g!&yZ{9suUq#uR|oHK#vW|`OZuFxjN^xcIKLZ zA0U)1#rnjUYA?#x7W8 z*UzF-La`2a`sxcb&$Ldjmek`I*4(W^BPf6QbaxQ z=!c3a8$^+I9Ca$(Czb6Sw&H%k<=hN1U)_^frBt+C9`c+J^p;_3sio1cPwuoh7)BD{ zhE)h(3hGN8X?&?*kv<4I4?%(yybfkVg_r_4;SPOR1EXFYjen%2HRm8ecHLd?133s_ zu^4De-R`8!;L4R#Q-PS}3c$&qnqjqdva2Om%qd5BzG`DFDC^9w&%VRd{uC; zR3h)z^0i|r=oz9`?E{Mc2S`A9vaH+U<~9KGAP!cKjS#3>oNd`Fh12|Sh>I9xZY~Na zu{dH#G_qJvhOZs=h_Om*CHTiF11~qJZt#T}0>skq{iV)1;auq#d0o!yxXuH zK}LM+9{fk~+*u8_-?lJl7QcVJ`0lWxT`bRG;vNwt)5aS!i(94s8DTx!*3aUKjt#WH z!zii6hLc=)5uE@561Qdc9o5XSfJVwT6A=5xx1FIha;X_9{Ts2?vN*p-l9opOZ3P6A z$qp~8-3s*yg!~pgT|OF^7g|K=`$xX_jv8H&r%WmDs=~V9IC3cN5;P&u-yL})bAcJA~D#r3vVwWEZ?1MQNoKNM#0ZBxx1j>l30^U1GD3j$ z680TL%&x{h;@UMEzmvh|boTG8jCN2G|9w5RPUII|lNw__;v*py zO(7v#X?sM8<(dlKJXeJa*^ijqrxXrhGSQbe_+BM-dhxFE0ZWK;o;z*I6;f7W|DkiRQ?!(QBy`_e~-=M#Ef374oRV82xSYJOt)W{Y!ykI1Ig zX?kNavYm6YgJB^%{g&FZ=vry1Zn58uWAck5cuUuoEPOSwmhJI}Q`K`@oABXV>t9Uf zVk%jzSgA4ed{<>oCR5PYsx!Qk^Rlz_6)x&EhU}?kv%$PhD6m&Pn6LeNOSPesl6<#M zKtt@c5CGvRM9mfD{`8B$qYi)iMD>=v3vdv}G6p`NEVk_io-YbCp0w7`~wx$P4_VgvbGaIxTK%&`sH+191eLOCT7K3Wb4 z@h*h`!2%ndQC#^d{#r_Qc!GNFW4WWh#P{xbk^d=DTPCCkqUML64@}F~%G2IL+X&d< z4k86FU*?_ACi8MG8e_O?R(E+kzS$8zmCyGUBC;SW(NWi>MfX)Mj|*z&0PgwMII_Z^a`$-UZ$x0tUtB>vrMT!l;ij0grj z^8O0jqD{HPX#7B|-hGv*YII-LJDKmt`{e{MSO@c6#Da~Tq3)X*YQ6nRalv%`TRj$n zR36aaTE#iq<)3D0)`^Hw=uxzO`_IcFn;qg>dY@-#LsWt~s&NJz42!>9{74)x^m{6I zXnGzr908FqKEePEZT|lLi${pja`b7FO>no2V+gPIjhq9n|7ci+BdkvX15Yo3ADcFV zQOGQ}LClEH~E1k@J zh^O3jDkouQW!FPpvXHBNA-+15k-bK#!vv7m zD1_GHkiB2nhBkD!1(ps^v z#K>2p!imq{IWCya!p=btd$-#N9YqYHNX@G^lc`&xX=BbZ+5=h49D>jlNs}+B_Puka zMc$a2N>WtRld6Qa4_6?Gem!j{u_^fRKpRAc!=3h8OnzRQOlNwVcpem~J4&+n_`7HA z%7Fx6UcP{s+!AvBlF7C-7Ftz&@~{wM{TDxrKb2S4CL972mReS;)kI`&3#U_jUqvo- z!u0uhkhHCzLy0xqHqCamv+A}9aOr=cg_8orLXek>5=uVSUn`~&p*M%-%Zr_Y&Gp+R zDp)XW8rMy0{X&$sU}#r49;i?d89LhB2Zo5Pac;EWlZ0`@M!BM70=Ps|^;M z!w-n%zx|g`!jxFADw)%$Dm6z<_IFx8H8qi68G=uSK3U|T9Agrn6vR}yQ~9q+>THIi zw?;`liZOe-!uW7{lfxmAQiv3NirLHCES7Y0Hjw~s0cNdtvD39))ZSLAO&W|z!={@s z)iUH-S}z4U?zC;tMEMr8@Iytt#q%6@MLY-{EGtgG-VgbRXmdylFKlz3&D*J}#dbzx z*gc#X^9Ro6D0#^y2AXl#^+()QO3amKJ4&~cF_ug^w__^}bMsw0K(9SBG1@?2WCVfe z{uET*{ljFml1Jf4VI?E0!MkwaUZ^^1spkG|ZfI1pbaXddyQs6hEqPZ|kl|_{{4;;a ztreAo#RDjMPDC1!5SlFF1>Mt0K2z$<&A9E7M=ar$JV-YwOP#)fuk7M=yw*T|zAw)l z!{d}{Y12lB8+#S1dQxlRxBLKi;&->5(k1nSr6|XiQcxPiu3}&2%_Ln?CxVs#fZ`>u zHy3?OJVm>LG0#KA;M;$&?5t4wiF0~_1J(;7Lkn#3ZDzhggQ30c4S%iSQ-v7ab@JNyxHR|F0W&BVz84zriDfe_iANa(eekA^)0jIbzMn2A4;O>(GD)Z zf_u+Qez!%ChBNLTk)dVHC1iKk5$2UIS6|*9Sed(*QyMg~I3->jstB?`pZ^2U{BWFb z(2~J2O36(s?-lm8Gd-ddG$L_S0j3;kNbN(uq{U=4_?zK?deP350oF0IWXaq}%w6}Q zcMAkwW5BO0B2H$VCiZ1SoT9tA_eDyd2b{-_AeL=<(WEIMfmJ;|#N-~^CAnel#l&KE z;l~i^5Te{~B~K`|ND|`*HhdguT`N-z@%!K<{1b}|a;B3*_0+d{BzwQrx(r~#nu3+! zFQMGrTnxM^E)X)nET5@U2$-GR6>Ic^%DcL40oC00;A{99kG7TiA=9+Sau#~3sT|_=|&yct}s6xr^{kP1y}D8y@(_%Qsmk%drtDirgGY_ z<1qK&{{?J~ZT42&cktu%KfogzY>Xy1VIpBN@J~K?OlILfzy$&2dqR3o(Eqs6{+~A8 z|Bv^3zdBBv1a9cClI>SkNm~AH(U6UzWaf8ln7A8pP;~2QrE%Jjva}=71U7n!fy$|T ziM22=j>TI}2Ax+c4xHh!miW9rr*I!w3IH;Vs& zOzWqm(b5zjCXjEK`=Lrx!4!>%lg!flsV+vD&Vpz-U>?=NjY5ZOY|}40N(eq7I?GD} zR;g*2OG`{OQSzIYVII1J>)Z&zG}~HhlGH|sf8D%ACUCd2OkVp6X8p`5{UG$RWZCX` z?UL%hm>?+f+45%Kmv#$Xe1d4xuN*{oci}fCwW8N)kCcA=V} zAT?XXK{{zTc;!JFcX97BIJ8k zMBH6hN(ceeZzOV8I|9$MazUcnPD3hX}@ z&1n*Ahv^FK$p--k6!>_v2jK;y*1g&^U0^&6@5kjbv!r2j>4S-UcaesEXt_HJgH%(~ z{6+f&>jE!!NUZ!do$IPm{9Zq_>1MmtqkLWEnxLf4a~6kO%G%o8JyzstVv!t#!5P8f z`c0EW_G5b@*LO}(X#V%7y1*`LSh4{nn-Oy}g;e9@hC}3#4*a|S0s3e6@}XSD1yO`5 zQ9d=!3a>tcLUGjVwbAl*`(P*t?sr9m#RE zi@p~6c~|O=>L~1jXw3#pbdP;C2=C4dN@Xw_H_NlLMZwm{Z{`4FH7gm9kl}c^zl-y9 zn~hf;*^CGwcs6u;iCMi3s;AO$AMI-&ty9Jm;%kqiH`uEoZM+Xg7W7574~fkxqmK!$ zmO9bj=~+AnOq2ct1XJDoUHIZ<@4}w&3Zdmfx1)3&7F555zU@OL5@f#FVdlp;iDrLa zt96V{$1ga#8{#y(s2)Ao|5NI=SzotxR%wHFT6_38RCep2fVM_gX!lmPKt?~sR3&46 zO}<>!rL@L{7LL=VGaw9&i&(?&AKFaM;*b2l=iFz67*_f;?L0IYh`G+OoW5j`F#`Re z61+9ps>V)?c_S{CK4-}xP+p^-8E$Zz<{U~h+l8q`x<&Da?B8;#jq((W>nQ6sLeu^? zv9SslCAs=>LR+j)Jszw!DBcKdv@_G(evRdP^S#jgj2m>{Q1DVfO{HV#;IU`l^+)8} z3dD=_v!*j#(4#MQzM9LOQ6sfCyyS_vdOx``KoOV_wwb_NwEZ6d3eEnNCUmMZQ1!yoE-OX zy6s#$xkozCr|0f$V7@{XoXXnGN?&{4HzpbnGM$s|%y3A{-JM6lVDr79l0y2q##~u8 zQ633HZ;ur`Rs;dLp+u>zzW(;wX;W{_oUyU!h}1V0@@}+mabl{X`t_q z|BgOe3ZEq@S+hOr>A?}D_)|f&FoN##K^I822of`istx)U9pRw-OWe^G zaf5z44t=Ofku4u%TW(`l9+p%n^H1Ujh7QP{?Tf-`wIf?~j`*}2FMFOd0hj8iJN4UP zT54B?k$dq>QM5b3WQR&rM-k8{b(iP6c5A);?ccCl{R3-AN`$vlRi^08d=vc^BGcj9 zk>>QS-?vlaq(tV$-CEtORj|J=5g^%09vS_K>sq0KXiW-^Qx0oh1~0$3RmmQ`Io&m` zIbBk547LHOEK40Da~G|U27s8g6U$97`Rn4ns6X5KJ}+yAX;S(pF4RgP$`W3H?#cB-RIvQBMUUDwjvct3y&I<89v*pDjk+)oK$T@UjzNMEx&^u^7+m5Eb;DC)%sGOF?3aK;gbp~ln2l`g+uv}5Rf&FL+1iuUcH3kyW=Ew8J|9gQ3Pv>KMInRJ_V@a=^U z1POsb^CE>rMt@-Y?$OeoK6gY5qI-h8WJ5RNI)f#@&!^?8)dVtJv=1|U=7@t#MS zaXaMa8Y9cPLRyZET0BPnv@~b@)ZPHLHA@#_%NP*wQ>cM6*6P*X~_9Zw+FG%pIl#{3~;7FYr zY5aqP&p+((0{bd_P$~;aZiLX+(3SP%PZ>De8IJCyZUg@TNP>P;vR#C1;6h+`=ZdWn zu30SwT8uL7%rvsx@V|*Fo9evo6ZA7|Jxv?ny%AxA*E0|Mua;i2veqJcxN!YM<<;i1 z<+Q4Fvk_s0B4ch~lpqO--bZLSUeE50c$&2&)+^R=Mn?1vhRmG)UC0?VY)X+O{!F|+ z)y^O$Y`N57`ssoe<&#C`G`b)K?1p17W7ce}#M}vJ!%scFA&(SGa-M)Xx)rs=Rx%03 zGn)6@&wTl}M6^h32_ccCu%7+p0l8MO^&V|; zi0e@b#}3hVOeE0%O&IevwS++*20+}*4Su^G6`8Iv{BO@HR31Yard}B z|7^<0wW4#%w{8w1N*nB>6mUEiW6~mICf9VMoTUKi9aEG4iPqCz7M)8aheNMXlekgVt>qbqlnLmt$t(bg;JIap78V(2M zn$n3!vO*Rl+b4+gN=wk_>R6HK=nyh`Pjj9fsrmdvdbCu)qP%Z8E-oT%6z9UYkh|Wm zE?%)g+X&0I^iRVkh|r@a;`LXeV%EVO4iRQI9HM~g-8-L};=cPqh_ZgGf4M}{Ia)6R*ZJ9JAAjv5kMrmNuP32~2YhP>L=S5i zZ@w}a0M`uBu|Vv<0f!%nFMWZZ3g7%43w}iT#^8#9wgos53p`D#v}Wa0I92f#kOoQ& z`jicB7&anmJuo@aUiLm&AeS-KLtxAaIdi>~_-ua`S~L}`{bwzceFIv40uGumpNEnp zIIc^T826HcBMA7nVSIKVa-dw>F507L>{lX_Ks!qYgo2SYLU*Anp-S5yU)^g%w`;ls zi#V9ND(O%j#Day(BBDl#E45d={QQ1rFX<;Qx6AqhZ8wYKId@~NY_~uNF}5}9zDo&jDU z3!X`pDcv;#(o94U$! zvQwdLYIpHr?yi2?y~e|BBvn-*%{FeQXUw06z7jtczB17x)r~sd5iCaq^LB8@5my z*&)nrdiqKcx0s@^m`!x0ZpO1C2XNsURt&O4IK+A3KnPtf!P&BzY;97zs`w28OskTH zJfO!X8u-xqtP&la$mikGs-+OJn=I5seolPz?lB`Aq6;~BhUF6U@<}_C z^oLKM%r5fy9z^IeXq;^-$$BMuzcVHe#%q*<)lkxuNfRiEpUAzP}`YEE(zuYqnmfSio;^b5|Gz zT=@`>4yZ$0$zKC0D$O^lCxG7{wBtBd&mJwK4AX)1qs&c?*Z4d7$3< z2ri#`!cQNQav_+C^$L@ooPQgthAWwJ(E%Qu^uXh!YUcTN`3s;oS9|>%eB|Jsy>&jG z@{1>oMtzj_Q~y@3n{M{4zcC-BR(p!Z;_WG|J|>wD5MHte@ARSK+Osl$a6_IX-m9^{ zHjZeud%7@v(+k!JzORKmJ}d*Hc)=Sg>u!jMs8%9^*x-nl#irPlb-N~o?9XRks{ie~ zVCjfhGo`aa>+4$$cC6`1l6qa(;KEpUE1U|g3EnxrEbI$}tpuzGFrIUyR=DbX=%?&) zJiLPpJeK+vfeq91Ck@uu5++3Y1(4XF>B4X;FSFDcoJu5YVWyMxh*P|_=*pa%LgRmg zJ*V{3u7C9Avy>RwS;G3%~n%#+&0<<>~6Tw6acq z{(}RWeYS#3xKrc41irfxyJRJT=k%)=&c>E-B%&DFJ%1Z2>mGAN3Xf@+UR5(4S@gx` z^|3CS^K9CFiM1xW6lBq6PUyTBqGtwqvlm?Y=}QSOPi(H`^y07G(vj}l&ph1xRXvIz zW0s;vtCq9}*WmeUopvT~OnAP3_Rj2qm2|f^vlE=uGrDEqe*l+SxsaBQ6cM~gZP9(9 z{x_zS>87cKTY`9ZIR(mt<-8_t#O&V$;^t=Jaz4&Q*`GyHmkIN7!uIVZu=fB6iHb^@ z6he5kRa_%~GN)Czzz=+rmn5B|nB73Pv5o$NMGz0Z8;|g3$??ifB5)?dxpHYk>!t!V z#`vautGi;ptKCAJzN;4hl^0of$_9`(E})1Dt-0mAVL-J{fw1vV=7=6Ji*n*QIF@s5>p`qa;dlDc&QxZT45bRUc-W7e@hEj> z@^L>fcToP~K5u;N0?uJ2kyO8dDM~x7p3|Q?RHl^P11dK#Cku;4@6-e{1_QuiK5hzh zC@tW&11@KHb>N_zcE~O@iQY~VuW=JPH~=qJ2(b~`UsT4}7P356uX-j?k zXR~o#iQk7@b*0uzj(_#YsaIdyN2oYA)s7l}HSlu0T*@tT znDuoOC!q>Bx`}$Gy7#j5EoG|14xk+miD&o4pJTObf}m&v$m-0|={^WSVVVaT;`x?BALw zj?QC1BEaK7HSTD`$XvwRs8NMmUdexe#QUm-@jAWC*(>*diIhe`X)LTC_T>_T_R<-H zPVQf9vh5$?8QBuk$YLSMa17jLkX&~WEVsXlb%l+Q>5?JvUa!M`Xx+H|Y-`(WH~hH# z&fGrFzV|hqKiI^H=N=hHbw{p8azNDEM7k-KKpRF|Z5$2yjR6dZ zQ%qK|H|aB#vv~gfc?wD0U@WanN@%-y2uTd3nfY9l%dYW8n*aT^%3v?lq^bI|hUBUs zZG3#N|8_cYzKD*(L4rJHHxZDbzI{F&M3-&6{gpjd01#p!7D5>*ME+IB#>%O@(#R6> zW@l&^=11_`w1sEpu#vQFSHJ}JoS&?7`*rhQXUYc|cTS0V0$w;BZO$h8d8l98#UE}i zb%jCS-yWn~LKb1%TYVVXrczg}bZt2vJiICw;W$4AQ1g#Bh|7d1>yP(+Kw zfccI7*i6pdeKG+-@ziKsrEnsbZ4on;;0Z}Az!bK*Q^nz6dv;x>nIRn~rw=i-&py(c zb@g`k@_8_F?1CHlT+^PAa|{A56Np3A>|G+#da6w)?%x$}q;GTb2d16%}SzlD~SOg}LB$EzOR$tWEOS4xdGFd}s z#=Q^X{YW^R2G=pV!N<%+gc`a>$aTn_6*AQ$Av+mf4}n_D~b^Gc7#Y@<7bE=QL1Fg&3D>8a{t11M%7$s4|`m#gAg0oX{UB{XJa z;UzS1<{+GY#7v1fddy0HqUBZn$qkiGD{Z`zJn-fh$lLj54GP>%7xhcDuZ2&ShNF%W zUNo3r?A`{_OEU!53O1#ZyMFk5*$XyIxmKb6@|nZARm_vc0Yi;E)$C6o(`_{iShbIw zv7L%6_qCXhZ-PbXh-o)O;oXz&Dx#p*%&0`fy@GjeX^UbCS>(HEsTKnj znaAttWM)KtKlJ#szKw{y;Mgy=Nh`D~0RIOZu4rWPyX=fk2Xn5;OoYlxBTiFR@G}9? ze*pb7qJb%4^MIM0aK)jsP@dM1exi!rUS<7A;4_W@=(wGo_<{S+_*3NX!H?&{$YKPc< za#Z>Mi?+84YWt76euKM9k>J6lSaElEw*m=J+@ZKriUbHw(*VV_xI^*a?oP2%3T=^6 ztpAf|<~=j-%(*)=&qeN&pM10T+H0?m)Y9N&{san%A?0oAz)x$IY-Uu6Q_l{uKxMxG z{;v<(|EFU;|NoGX=h0Lt4gPT-i7@_skYyEymwT3oBUV zFQ-Wy^It<0P1fj54$_eHilp9r(i_&y%)LtP*Hc9OJkULiVm@oHjqmV{zLG@krfOzi zpT@M;`Pwn9Ola5|-%)^?tlscYcYxT!*!A|fSdkD+P+aUF;X%Ck$$d4THbQCakQ)b?2q;VsT`dU3;7)M&D3lXUn* z^>}3bLyCsnZ-;@y6O1aFiLwBs1FHz<6m}kvr-2_Foa#gYY;lo%o$nAuQJ5Xk5=sAK zUXA_UIOuEjdaP8gYy^&0#ZT|HyhkN4{nkaQ>Z+*mb$iITBZ|g(L*IapBL{BlDf?q1E^Uf9?(Z5*$Zyi7L~bFX-%i>$J6;&&k9KcMrYXfl3*+)_ z0H0l>zFNCxL+Q5)E{v2(vh`;;m4x5&iTjOpS%1`2skB_aI8HvJ83no#W(g9cfA|ZQ z74-dU5HAD^p+=EL+cO)K>X%{BnH<&xJMgi|KUlS7ocY?QDKiL1Us5QQd5> zd;9O_`t*tF`%1)u@?aa^UsQ;hjCXb_c1YX&q71vetz_M^rni18;Gv#d%`hc&eF@-) zX#i4zH?%bhj2^)hX|`EbIRJ~+0&R^cmnF-9?$#ixy zb^nNtOD!78Jtpj+V4qFrQGL@7q0iQqX4o6RrhMr?II1+kmwCVN?OWvL@CAv0)$Yrp zy7(&#mu0jw1$-|@)hxp-Gnhn!q6P@cCrq^cL+9wU;_zYl`pwc{#c(N+t;Z4H(db=y zHz;x2_fAhxsA#NGqy93`d8u^COghXdbM@baxG}*@}o7WNRl`uUey6aVD8ok$qul>l+K9o%jirPNq8OTt056 z)3UO(9=T3A?@Vp31|}MXUa8@{;>dwG_Vj#d{X*AzHL?J@9y8-YZQUNhr0nrqi?+xt zwD}g&l%6>}>lQH>s(1$hztkm1nVH)oK&=luJ(^xLnv{9MO)mw^bJN1aB0^2`Z@!V! z@;USraO6?w*!;Uv8Mudd97wIaHw$sf#aY_a04MeTDh-&6mJjH>50RXcYhCqcPSq~2 zwRtT`Ua_rK+w*s;+8*aWKurdDNB%kgkriBO@Ky>YZ|PDVryx!8x00Jh{;4wY4`Umd|QYKIA1d(d}Z>g85RT5Y$#2j$I# zPeU{NCaIFCY4#C^SuyEj(aRoVQwGb{`mn(Y6Z5PKb3RirQqthdFx+>PkK47%?YxS8 zJI{F|FBI%S*1lV#|0~sHHRtl>eshS|L4?)?yYI15R50n*LesMU;wK0D%R-Y3@U)@m z?=5GEVirG5@2}??kVcfdf|39T1Ti1VFMt%l7jqR{Iv3NHm~y4ZYid)LdH9U zV=1|~ZOuotqi|%K97j`TJ-g*40oN7(irNZDfMBe$_n1zuIrBH zj4%V;FAB}-S2KRG z7cMZGc!04mSTC9TcQ$^=hAk?4V{iC(3q>z@?+5$~X&8axO&QTFx5iM4c*)CHTq=Ik zU(9#OuUI~4M6H?N|BjWYx^z;cQ!6Vgd;UW+Foz@Sc0>K=GojaGVJH~^WLu%2;}l+QIe{y1BkxkN}Lvj*8Z7L zaPUqbjINj+qNUn7AJCa)VqlJdv~PINz%Io~VpWBj8WV$Ufw9lNN+ z0d^cL7PxyXYVAqI+gaxSj#lBMW|U5kS&ochrOCC za|nGWc~scgbJ}t3g{NYd{Nl5nJbpP#tfV%VGw0*(mjmM|(bSqXrT`!h3xJqJbgA(o zKz^ztWimf}A&UCsaZhoxsf}P2;VMSjbh(@MXG(D^Z8k}I zeM!*Dc-ry9v2r!gg02K@0+6W0J;=4in7GrdnIa`V#FTbc*w@NOtweYvX8x&DKKk}T zlPEKZ59;50K|ni&#nvE?pZiS8_lD;91FaXy8r24_U!{McZqveMnZ+i?g>*TgP>gO@ zp$MO%F|J8FDbS@I@BFp`**1as3F@b&e|TH=$i`Q$sHrK#gHXE8y08i*e_zTu;p{gi z#-i0?kN6KzS^tWTCjD|JhXPvCQD~h-&L?#FIFn~r9CWU$p%a}OgxA5hC4@ZOO{ks6 zkw)?#C}AeE-D`M}7{lE3c_cBYjmjEL4Oi0ZRjE2HY1y_!geHu9e79b`x^bBtVKL7T z>h$9Lq90ltOre_OS&AB-E-F!+5DhBn+a0zUcO}Wxk_PZ}fm~uH_7696_okuPsx)fy z0h&dXqM1n1J3Wi%0V>RQGS@LbweFPL-Y5QSxpxo;2xIy&N*7)6@V<4Gr*j&z8|-Brk2-2U1pxy)-B_u zrAhC!%qUm7TiKEC>6|@v73*+Flm#sd61ACyDaER!mAg!Zmx828fm z)EIi{tsfq`^aKmv=ul?{5Y^r_WsWn{xWAorboR})iB?xe@#kJY3uEK;J|9KSrH7Ty25M?wN`!?8AdzP)=|dHC&HOlVz zV=#N>i-iOc_4ZVH2-wf5PPUP=o zVrEo69+{YaQk8L}MUTa0X5J9#i z@sfs&S0qm5umR{a`duk`j`M$~TUjn6C3PI$8HCY^C$U#c@xiuQ$R2z55jd~x01W*F z)lYK*)G3(g@>Yq|S4WrvzRRIYLESIZUJ$>2<1m@(TuP`Crtx4jJu;~$G$W~gN!!O` zcg#;ZLf#9@QRC%>P0nYy?J_S`}RsFDUc9yH|nYBrm~ai z9K1VHJw%KpgvVw6*Sx5@Jj3bt+|S&PB8Y zKL0sIZF-(+a4K3fH-F+Q^s%D}*bCHcZz6B<7EFbS#F%Dk`VgbT&0?IKK|Bo$TYVC| zeVQM5$D5m^V+i1=Vl2zmM14Z$h_Q_&S5wmjRC9A!u2y1CJR9jd?^qA}?5IX2I3>wH z50!dcqJn&k#yB=yHyVlg6tW>g_vdFJcpPtslotVD(z_88edpzp&C@XjeOGy?d0e9* zCg^;_maKwDDZLZ3Gk{UANf#gIBjGIj&W+!<4{w#r?3J)E!ar8yNnWajYTALeBWYn) zCqev3EIi48Po@i2%v5gJe!g9j*0xDOeY}U##I~uk`L5>2S4^_!HR<|O+X~X#R z@>+;L`hvBm$N1>`zNR8c_vb}gV}5sOtgQs;i2ay5#jibIaLx(@X+ftmw5q_KAcH7l zgsgk|Vm+j%&mfal>^j}*?j_L~g|#)cCcozr%g8%;mAnYXJk!#-g{<5_|Y#w8xr4iev>!09B^p7lTn3m2LASrn1WaFP9GH0A>r0UUdPS{=Ab6+bnr1avg1TM*)0+L|2F;M%k%0{zf z!~F%httd>Suhqd1{RS0oFL|bfp4$-~IPlf=CqKVu@;h;i9}9BU3B0izNFQk<{*k;F z!n<3qRXOsS1~k|N)hr0nKxt18i*uB}Z8x@tHeEJQH||;k;eb>F9^IaqT5ZxVyxig| zoLwAvvsw%Fe|x z?PzOJogFG;f=8k42hQOCApc<*JE`=iJE9&{Yp$<}H3FOHyIzVDOfY8__BlDdnz>k+ z7WZ+}0KfgTMQd#xgaevDNHCB{*7)|nbNR%(y#+G(E?8GFwgJ5x{xv$MxR$)$xx{zT z+S$6+7a_OnvI8U&f~oJ~kfUuaW--LK{aZ>9K-|-t^lPd#ZVq!CsW7LlGpKun86&># z^3#d0TxiXO#AuR7o21--qf-}41jan5p98vl*p+0Xzx z1O#$e7>Rg$(+QXY42N}2P6FZng7z`NI40%J=Si!91jWJfDJ58 z(YkYQZz<8JW)9im*#DMycD9{Bu`KQE(v28oPs%qZAA-Fde13hK5UuA^QzNy09tDyE zi+n^=4fT6NjCC3yL}g1KGNxNmHxUr84`c^UUwX;vK;6E}$g7OZ3VTVe& z()7tB`;g%&?t2rb4#uvs9%jcEv20%)n+C82<*W+1W`@e{Kx!$=jRTSl@>(^GV|mhr zN?@iijH01B*?}Bk3g^{(N~q8r1^d0msG6NAN^X`P<~U?P+|Ztll!iVDGxACS_E>sXW&);Tulw&z;W9bwGp#)sM0b;CCj=y#*ZLS3JTu1;>O!1$ZK3DQNggwb| zU*Kt_2qD{y8Gtm7n}os=aDGo(ynRH#NpZ4`gvQP zt$aGxES-Q6ERQ12z0?2Krgy^askYuCMpNrUm5BP<4`-Jzcz8-}+95*SK2)=Q5ol-0 zjn!A`Al%Ex_WYKj>Kc*vs!MpJHC}+UUW#2|u!a;6$ns+qr%n&k(DbB<0!{?(qBAdbzWzv|+) z=C!+35mCTr?NxK5k5<%uNxZK&Eck&yt zcm{C@V_DepOb}3;^Z=YW**x<^08^75(L4o zC8bY%ItpE(DFR;nW5v?8DmHH_&J;Df(%ORgE-*=Ms=C}q8nF=8X>9vMX|;qJAn6Hl zC-<>D4u99wa{OI7zQ3OG`Jcv_uKJaaX49PxA?{z6q(PdWrrF*mG$6F|wI`mx_MR0v zYj-X@HP-C9MjcyFb-B~3x|4{eK(*^T&#t&VY-xTMW4wW)J1=AH$0(hLXKhtZjUDP3e4^xWktQV!TU z*lijQy+b3W5djrSa^R=4O!Ci5pWR$RT7L5QA9{#)e~UJZy&~*nwqGJkk%B&p7$eET zT-iPBgQ{4L+NEMf>9#ej%dd{O^fZ2+D^B}fFC=nvxMU=Hbs|OO4oN5Wjp^1|#=QpU zz(mbt6Qi2m*W)F#FG`u*1SnH-ZH_Po8MCc-NLn?(OyHMAA5ee3Riu$Wij<9Jt`Y`m zyf&)xc=Sj_HNTVTjX-ie3IhtpPx7(#cr>HOkvhHx1~KZ{)-tDsZa03 zm@@reNs33+^d(1mu$}R_g=TTAXcg3GoaRE^84h2*!($*{5< zF>yPRR}9W81;L%D!Ep5d+uZ)|Hn!zu<%TdW^l2W1C86Hoyy-VmK99E?Rl{Vi(vPMC zWgNhL?)~6~nqVvVE#oH_7I>`4)siw8=)VY;3u4dThx`+r7psagG|jL`&FsA08Ea2$sN_n~-|N>-K#GsER3*(x1<6+CxPOM?l~Ow2eH zDEUh7SBV4$&76&ivr<#f7rv1MRQ-_y;xgLQv6bx11d*9`<@CVq=P& z)&FuBrg{a1lm6*E#Y0;O-l;fmGtFR%)1%C7_zvJdE;&hD*z z!0STI7~GQXti@;o3TRZhWg;pIC(}@zELP#B!P3K%s(S4$lU8ELC*hE?Nm`B63G7SAra-PwEb6 zD8qsb^+N$Ua!VW|D`l#@Ox7e$gZxa!bl zEH^|h&W~2ol)t&~u++(tz#u^A?F(gH=)9SOPHt{;F7kQLBG0X?!f*+^*E>cUI$jP27LqJksFRacaz1|dOccl+nOlvA(O$C-FOGk>wc?;{plo zO)4&*`9X#M7A-HeyX$MX)+wXa^8pi#Ia&7nt!DBbJu9AG$DS1`?kTZiBgl_a=Qq{l ziwDe)63d$1UD?x4_QBm`PfR=KYsnz!!uzJ&37t0xf86<`63^00w7Xef4Mw~Q%@2AR z6Jgn&KL%(9AE%?@_ahb*`ErBuLCqB<_tbw9i~mQ8Azs1lALO1 zXC9Hy24^Y|PX4xL!oO1MBJ4PSb=Z-A`!P@Ec_qEV7arNV^%k7?qN{zGvpwuNR(?-llvbaFORhs-!r2-^n?q`uY%5&6QYD(lE+#kSWk$!Z`?o<+NdKk107oT*{1 zKCSvRupo_rv~_`~pgwKLEsBRdO0V&Iax(ze2~`2H*!*A$B!0Vd1$~PlLPh$$e>WG7@t^7W(9xy{N+@71(V+<@x1>JNFk)oiaz7b7>W>0DIe z9WS~wUuvw+bQnU)*i7TFjSPsjO(1F5Ng9OcUyj{a0oz7%1KR~7{52K`s#Ey}*)V@( zG!Xd1ox2Mq=uaSo^V3@z1awY{fP>k$bN3g{#ad>_R1Tr~#N`H0_hTWkUG7NUQd>8E@wH>$%fRAd-{b z_f9P$g1i0CXEwk13tDHQG3ml4o}uGZt$nZOOAUK{iA`5;(4&mnDy?6nkya)`y{bf@ z&b&XV($|w!tK>hx`+f!7)50!Z*@R}beDW`QMnYR%HT@pZ4Suvp{_VprZBnb$Jupp| zY_hcJQ`npH;s@L{ec<%>)9HWI;jk|G$+EWg+R zI7f6%!!w~5Cp6|-h~|9j>A~2FK#7|l=R1E?48VkbtorAetzbA)w7;9 zi!aM`FLz+HgVO0=^j4BIgC?GN3;Rr4b-2j=B&*TkV9J`}eb0efG?GaaKOw!` zpWWY8Yh?A!Csl!(Vfg^W-`dpK6A2865T@qzz__wFQC1#zs#ISSFb%sChGMdouflHsnPR&4Gf!uJoc6J}l8ddY8Kf89_SlXl zEf~`;e&W95Wf`XpZ|a&*kQf+X>|Z9xq8b=aOYWFV&x+YYEdMQJd68?8DHc?9oVO*3 zd_`$zrQw7cvW1lb)TAkHYQik|>o@$wlSQFq-4Mgxv;l6Ugyea0En!#&Q%l!9RPU15 z&7gUXaq+v&1x6@XKFtEJz>IqXA9uj0PgWYdm^NH(nQ3Yun)*pGmg%wd%W6d|y(`0& zI2Daj%CPUbf!;MiJeRGxE9l}rX*<-VtFlyM+%s-p1bb|Vr0d~fzOd6;sV-CNs>{iV z1^YHNrbuTcfrR@#LyE*>$r3cGK`tyEo4_6L+>(EB`!(&6#mq| z>doRXne=J8kl24%zUgL2F?(c3sA&WxU}}~9Ds93!*MV~!8?vI)G=a#A!O%9WPko&* zVKfeKk3WmE;_**RH$+z=z-W$@!b(phRwQ z=B@ZC2R%IO4=jba>4l2BaC`7|&94cL36gAF?*i0F+UO=^IW$ZMrbb_bGRt1do!A!F zL`TLgJAQ_+PJOC4Ayz1giOB#QX6=5`Xh$umz?9&%E%v9&VW=fnOmZpakaq4_aUUdS zF_MZ-$;A~V1v6CDJ@@z;tqc+;`x6TEo0-nic+LLQU#*UfL|%<0fH{K>V)0P+vq=uI z%NbI1&U-CF2;$ZVMA)Ne4$qynmRk|dSmbyV^?M1U%4poK+xZVLR6EMJXdctl1nqn< z{gw+#^k*G|n%sxO@Q^YqCNba+T0f9?ekzzT|F$YSP+4D1OXtt;;&?7|8XkI<^ft3_ z+2o8#D&n0^w3fC13HsK3(~02+0Z9BawR?q4Dpn=kb31o`Zp=-Yn=sOm7#5roxu)1l z?uUC*V|ax#SqTg)@-DZH9t2WgGixdG?DWT0#~jC-40j!&=)!)>tWy+IGNd0|^mjaI zUhDk_SgqBlP*?Jy8rbScUHZvrmZptOL+{qBd2(S;DvKlQE_U1G?w(~cKvXkZDXq7P zc~`}3U;(FSb)j-RCnEIam{;6fKYV4k(JqN2A>-rbR5sR@wADo&5lP*z7v_*1NWzL?cd9K+sivJkrKWYIL`lCrx1P}Tn znl~AfHN~EtntJc`z*>@r&p zNXdVM8+uSoV>wh^B%P;Wgka81&8xKwCp_PFLP%d_U zXmVTQD*L=A*Cu*}&DBMVl&Ba5m^#n;xXQg2{9+SkZq;oj`4Qg0dMfclSDJ|%<8vf; zZ%n#ZMD^r=+TN}#cao|L>YTI+cYiOINb~Oxch!NO#sm;ePgTNUPqs<{i6;mx&Eo{b z86UHC5RTI}D;}0B&D+QI5c|fQohxf*G*!{7`Og_!@r!fLgQC7Fwl<9^jB(G`;M!hc z3j%IuG-g={;-!D1j0oF;K~KdP7ixoDqP}V@!sAiF2!O0>>LXf`_@s{p(HhG2q7^0) z2;Gsys74x_m0Y1h^FgjVT@&T*jaFW&qCP4?LffP2X#HoGt_zY`1mkJ+zUZY&0+%u2 z2{Hk-st1;G7Ry3xuDf`cV3;h=;&<&I=7YjK$IuWNf77!3__tr;;W0P0BJF3E>j!Z@ z<|TEfgs?8QjF74i?W2Qkw3`H>Q1|j*UfJH>t7s4*_S8q1+>Ys zzVU88p_H$Wy&|IXTc5f@gsufs?Y%I4V>Kt{8c$HCpGUw_+HAwnpl4k!*G&OCHdRjV z+*jTqEY9c+55zFyHvGe#89x?05@H8-Hbpz)MAO4o6S?coGNlcoiE zV%$lbVnAP1|4dd3qZcLOyAz)Br5XzTQfPv!rHH!wjs8IQSPX1{_E8Ps<_^{jG%V?S%(Qb~cm<5yD8DXb;oDVB-zq zy@GU3ApU2zGkA zgp1u=@jjPyakJkq5!&&7T1vQ&oxr#!-Fe77mL;gA2WGnonK^yA>2$B1A^&b8|0ui{ z&6t)|CD*d`ecWu3+j@b~1UqUI_`mkTJk!__)(H`yr&t7U^B8D4Op65{6r70~mPe0sYr| zQ!8dBo@>Vl`SpnHk}>Wr046J-%8Q}ptU|uh`*S7XS9QMl@Loo`Agjr7ucp-4UjpvLiKgCHbVO)ppwtGT7cAFlHF50GW6 za%J^4PAM?i74rpm>_m>!+;<*Y_$=$!(^2xxvvyyrP0ea;7Wdro^(vAzrkub$IvvAr z;_c=6prmQ7{{Y155>Kkm={Q^6XTKCff=T{_y(28kV&;^P&*h1tInVsIiFwYO861GG zj88N(KBn+9W}-4eXY~Qt9z)N774R6gd&hh?3HWw zD@QPo`R4`_t_}$P?*4K9Q}XO2{cxqthppoIF=FNJ%npCX%+rTZvwXV1W}4Bykqpae z{2WZw%E1!OPcY7jj43@GS<2F}GW!iLpP%;!XGN%7H0LM#K)IX6&CIsKw~*F?ot)zP zqy%+NZn&4niXubK@T=_AE50=z0fFpEg_KyoxhLCD{7mAaK%eg9*)d+`%=heaHq7z= z0dn7jfI{h%8LjSST9d5b!z`9AUMxM04`XK3p8s5HQ|joBB)Z<3q1)8RMEVMQ!IWoB znD>46I?4&KrVV78+_c$ewn`Oc7sGK)Czm$sF3t!;E{P#QQKu;Y&VN{l@528>Fl(4PX-4#m31`sW9 zNHYF#^DNA#{uViPA`|1}Sl-(TBE)1n0Y)5VYSc#lbDna43o*yW@}tes1U0frcRw#6 z>siQ@|3FL>gJnj&Xhe9iua7{8&hOaU{{e8%W`r*JVC;T*d5kuv4xp2-dX;SRM5VrG z>!uar!Xjnj)J^xDZ!p0urAPs(@NMF^hM5yJRyo%jR(5Xu)nojd6h|an86$ywQY{Wr zH>NTb$xXZ`0Gv!#R#lE|FGJ#mK-eeUgK&GY-U(oGy)u2kUkP-*-0A|5m*)Iyug zhj>&K(ecneh5e5?qM3-t@`oK93YycoqpW9DrExV8^l+eg8dN&`d}zvBz>n9!nMQD* zG$g1&R*B|^2Moq}Fy+e0>}Y8gA-u#nejwXR&al+wA3&XgJ$e1=`ePsI80svVIU*ym z@x~-R&7ho;q6=eSMNHoWJ3R}pCsnf$tl`#Vsu#gi^yEG$PkTdoS1fBS&*AZakdSm= znw-#4LWl-4S|fxE4leW*wlq2bC4d3;LLkE5YySaOj5fAJ-$^(Ewugfmi`H(d|IJ_a z2;ikz>ivk;#u(o^D}3*z0#YWwkl+ma6?9lh8AyORK`-}Bxy2uLDnDzMejJ);?%1rFp zL*)ya{SmF*Y3so-^b|U5`2cizd8=sAC3T{*D4xJAgyvm&gaXvM{evT^5d6@4?Tg(_ zmej_`?4lPq6CR9Y%CUj%ZVFRv!wRsLZzR5UZ9R5$`jkD@E=2J1v>d5rA5NRYl=08)eE`PX8r4|V2Px}gMNbdguEC#h({l%!qfg$TnzVH!4g%5k z311m6W@vn&+T+v3v!0Z%$;Bb|tB7W?rz0#(rt5Rt%y$Av(TXGM-Kq+1hO+gt23!Ll zsgvTEpv`v{PDiBHK1Yuvi!*YZ_lT_+2)+JFxt~SaD2lmP4s-Jiym@U@X98C^Be)8DAPW@A9 z6C-*?{cRo#6#JHO1l3z@MgRQ>&hjx4EXJ)+Y)wQK-&H}J#EpfgjxW#@xu=NVv9*zn z*s)DRyv8@5h31aJOd1DySovnhuO+91&Icr({J2_P7)B=$JrQMiGUF@xS8+iK9)ED^>EEF5_M`h)ehLpi zW&Zul)t;HV6R%HvI2+0|%ocNRav|EVFea9(*!ym5`{v@pP^_@?SOq@82%OKn^{q+tAPAwld?B%s1`_Cw-yOy=5km7~U5 z94on;Va_3E$Kiv?IbCZqo6e8>tKQiD)73P)R59*R^7kMUhnNIfnz@d(-MVAe1mz}L zk9eDp0|8#BL@Dl;k#hoDif&1Z3mHZn$1qb#PLxIf1K&lx1jQkSx9U~z)!O*giYNL^ zc&I_7p=f(sd+RGHwe4vmk#GHk37s@;JCQcVw8I<9gq|an{$qXMhpcr}b{1ImaeiNl z_=O85JaB?adt7Y0pv*jV3(3<-gSPz#okEyQIU7rXt8N@Y=vDI}lh!N&-n zN-o6oxo%OX_0h)Sgh|En)Ttu%kF&Ln^Cu*L4}NaVy6(VFuS{?)UU+;wpEKP0qE%4A z9q8pGAEE*k3pk%!toeaI(u(v*)mwQz%(KbHqR7V!6Caz5}_!5?TQs0OKtEcg+;zno#lB7$@lY zNAMGmH644x(e+&w-&UkYySnVag((|*ZjgzY&2J=I=F)DxigU;IZ`O>z<2ce^hDsBi zlf`)3l;Eu%lAO;)R*Z)#!p-@Hxy<9Gs5&*W4YyFeg{&R-f82NJ&> zHEL(pD#?ftNp6vRTc|N6(-N!j?Tc1)3C!LfrltTi(bgpLQ1?>_rolvn8jb7;niz!~ zCTn;8wNxfAJN#W9-5mJ~0||ukL@tE-eN*>I%A-kz5tFc$>)j`^M2j^8eEcL&ia)qd z&Y~UtHqw}0Ehf7QOV(B!>?0#|%2gO9IQywO$f`W+zx#m}6peRN{W;U%Q-u~P*r?Q+ z`X9@m%+k8li^Q!Q#)?`lY`c$c`uC7{pub@%IdQ1*$5wJ`od!}wTryQKGWC?9nq4`1 zC%srb1QTBs(82VPu%r|-sJz1-hOU6rvQ;jjCjJq&+!8&R+K1F~7uaY)pp|E%Mbqyl z6no9D(%22ZqUB`L3Zc)oU*y8F2riq%qE(9EQZgsf=mH>ixPT4PMu0JEQO4Gn99g9X zLX*DCpA39aN<{_SYeo{@oS|76f#s=zYM!Ty{|{;Df0wL;g~3rjMjQmdxcORxmSsiV zln*htC-{T;ymW45Y?UbvdBdyzrUcgo;6_MOM3-!}0F1ak9A$-7WL ze9Z<55#7&f^-mq<(w5A1vMJ;DW94YKEH`)X7>e7AzMPlwm2b}lcWB7gT6Js zWrLRq)gd#$)yWSc6#rNn&E0v$ey^vmI$o*VsmhiIh#6$#l*{&_CwgzSX*-SP0U^Ar z#Dw|PQ2vSpr}`&rTvf4W96{qMbag?gnMfK(ktr0C=d{q3MZD@UK-Fdv$dwyr&<#`j zK%K|lXxf&m(yJ)q!byhgAwvE-3a_KJ zwJYHQCE1Nn3!m$ZW|%Gx1S39D;tzhzG*dx6M`ef3ndz-@(wFJ5N#X$l-xDUi6_BtS zx!>xaUv-%+6W8&CBl`mCI>BDg)mbRxt8|)@Ec73kC*Tcbw3@z7)-~$2>dob{`~LNx zYmTkgG00wU|5Nj05YtED6zJVPEoj-%+#=7{$q}>84}b%-1KuR{0%dX zmxZ1J2!);%2NEltl<+%vb50OVw^&K!bVxd1+Ys`CYaef@K)$0G_ht`<92-wnQ(glu z6OsDAe&6-nP+DMK)#b|8R%ul&i2D9u4sq)g7G(RYiNU{DTy>hKnd7|c%q`05JOxyz zd!ARn(_|yqdlf<*xY45D(QcYO*i~v^#Ihz=d>Z|hL6=Y1P4OVa3Kxu3@Mj_m7@>4J ze{tO7k_@O%vChl-$G!#{H#$);&9D9k5E_)*!3lk5+0nf6@>A^BEA_6l{14Km0`aw3 zaYbLx%?hQkDm(&VMq^>4@Czmvp&5M`k|j4{J{~vfz_KDMg4Ex3T#+lH>*OyXPA;V)HV=Jko0}WSw zG(n<->AG$4x*@&}pu*85o_t7G(hoJYS<5+Eq@FA2!$`MZ-&t2+Kf*ps#XOWXTF~O9 zUSC8_SPyPg^Vg*`R6Mr9A-)4s1bbo+$+)$c{k2?98aBhj5 zcRM!!Oie4BqSG|CCTKcPh|fRGt#J>my?+`6ZGLVFwa=bt-$!k4Un zz_vdsUpSPs>GP8%@?UIu(?|8n#g>u1200Ly0y~W43m1itVKQ9P%AW`s^(oOPJw2}3 z|2B70OzMykL(f|Uerh1)XG@%y_z(72C)?Zo(^BtQEk$5r2#aVb1kX*nKOgxN(MpqM z67>eraTDkWars2z7W(Q&gj-yQhMd27{Gxz8q7(P@ubOr7;L}-*$_FJc7IHgvncweS zSk+uvX~xa$rB5%Rp-VfS6+5JjJy*K`El+tKlJ06~WeJbOZ7qj&zNwY_Ive%(@I`T= zwAq!z|A)4>e2eOR-*5+pZV-^J8KjXChHeO}+2G3WA2^Myi%=$4hov2yT z%-|xGQdRVBM@DixIf=)^(-9B$HtZ6|?)847`5>GChz)9|jp4-xkU;j{BF*I?5Um5>Qvb{P>H01-r^D0uedEq}LKm#Nc6skM?OG4(+=hkVtvX;3$T5lM~U zNKFzH?Ve$)j%aS=^ivf98Z(@T_`A*A=UC0lLwAcUCLqS;B$O2Cq(&YT^y8aAFYKWE z4dh(&j+DbbVFja-#_IM$)Bw$ld$lpy;K*h+Oo9vsX1=Z^54s zg=X32UtLJ|Wu=RyuMzyDPHs@N03s=OE0uCEyzT4z4+>1;#OZ2i@X$ih@>Mf_dqrJ+ zVOq$E!c}j>D@}?Rg7~73=hCK*`RN7XcE(8eOeTWxtsLG|Lbds~W2EB?YZA$J0dBRtEvcv_E*pU1Aa!>AMVFawkKHC} zXAW`wo>l8n2GdaUew{BIEXiS&80Zmdx?OpX^pW-YQLy>b1Dt{ve@I=24+VVo6-6!R zutG<_^3c_jB=RNhNcKD_Ix3R~B4f~n30NXDpa0pMSm+-u4|;cD42nmQa4F#)O1%~H zo$k{h(R;y`A?NzE#zj&W?|FZ1P)}*~!OxB}l11dXO;mvPqtT`j1VCEY@biIYW+K8a zX}z6Z7DvQ2mG<*e#qwJ|*czXgh?LkzNwEbIxw%yEutDBsp9juJ)oT|1K`-ym&u9Z> zZYAGkLDBl2JmUhr2()$E+lnpOX2^uDoBXRq$yi{4BnMOeZqir}MV3tz$ems(G+*ZI zP`wB4!>fke?-P60ziI5w0KuFn?a-z!uL%f{0lUQtR>HU1pdE^AKG6%qlBl}b?D?_H z&-QzTA^!mmbK*AbO-LF2W1-v$Yy^R}dMnNv2y|>Jl3Ox zHsraLs`16=lr3%a`(I@cJ^x?S&g4tbJEx?M^V_7Ns*!1Lqd`H_FcwdbQm^&svx49R zaWoCPc6-ziu4)rkM+umo+&N^+M<0Vy zTCE;PZ`5>}Gk7Dkr9lt+9lHJ;)Vf~3d~n24FD?*5JU|^9EY1n3T{;^}INCv##h9<1 zd6(T7`QX2A+XjD8D_#4j0u|Bp+u&`nZK7{~2#>iWU`YxG5==~%q^qe{nBR(3IQ^5K z7ESoB&Z&&1FD-qlMRY(1cGh3{!d`=ZUe$1GTczbAFbd!O#?XVl0OQideX|_0-J^O{ zR#UX{i}F{!tYsnlxy<%Q;LrL{ybjjZxXa=U_WuCF-U~Joix6ajU5Y3LKBj9ILl#0c zW%kE}e(M2!PegcO9#L_Q&y2h7`qZVutW97v;4}3$>?Y%?RcDxvSv~H@XU&SAmD>(| z++_C?${QGzD^n^LDZ$ko#BPp@Iens?1LI@5_nin&bzLJ;XyOtie9G zC}9p)2e02=LB)Ky_Da&11G4oFDmviFpR$)PguYpki6L>(MD>26C-NPe=FEO2W1@vm z?q3%KOLLYGXpHcOS*V3vX_ow4UZ}mt#=nY8m-%ZTin209ak)GykJH(vkMc~v(>c@p ziEn1DW@cQzKG{ejmND*7*mp53Ptmx=1zjU3^eB)=)A0WyH&CKv>(Y20QeLcLVwLHT z@pWVn9r{i^fF{$Hvhz-5jspb7nA*nPw3%_ZuAY_eevu}7jlp!KBKi4ijp1eC8`Wk? z*^Ce`@uc9VW1;IZlYMh?`=gF%#>2i>^<$L!yo@>++IS+2UJ1K<@`C+vF=Yo9ckU~K z^LK9tYy3Tno;Kxqv8a42*2i-dEz$O{S*mMcHZstAuK*oU@F}Ex5n=H%nIiUaoxz!0 zo*6)ZYWk~oun0Jn%X!(lP^=}E8MUsmqWu=^XX~KIeS42$bzWe&G}WpLC#SAq6}P8V zw`$~xn5u=vdKd)wmUS2kpwWEb_>@t>Pt~q?WUaEEzdla(@}{{nxjGBg4XW%y%EL zzt^dS99hWPaA#4E{*pWfGfK^gPI2xR!|5l!Bq@9}xzI7x8op zbxKy;UomCIS4>OQ1Pa_PT?_1ADHd$_8acO2Q=9kzpA4{|c7+B5z^;0a89! z{0CU(lzyV+@p}T0>nQ&JdbFGpCDjNjMc;VQWe34AU>XK7oJEjPwws!m3f~DF?4tTO zM|cTqI`|VVmd6I6?5NA#!$Ql*RD>h~;33HodA|3iWqs6z8N${z%y}sO616N>r&XfM8~&u(e+RgmFl{!R4p-ca-)j~c+LDdrPCxe5lU`@aa{O9>v%MK6Y5<-72lPPBLHiC3DW zm(tTmzr?sf^9OZkcGWZ!D!hynZVRiaMqCp?SdrRd1zB%5G*spc6GyU>%2T`8Eb6+b z&S6rr1ClCq><@X1S;*cT$nR7Dmx%t(L z@ceLGpGj8JzANSuABN>OMNUUlUZ-?7F-zIf5wAik5gB0De&`Z(ca$L=#VJ)j@m(_O z6{Y<<5!%Ay5)#N;Wl5DHwC7pOJcRN*`4;wTj$H0p!PoX|2ckE-Ew3Bu&TdW&eu4^~ zk9XgAJBKO7tVryKCerCY%$ZwEx^#4J^fhEQl{YJc>*piJjH;y4E?*VDdDW^RB`FQn z0Y}*21i`)zNIvd{ml`eHoE|N$`rgJl%?|gn=8Tw+bP;BN{@x$m8{|JkY_6^@JN+Ig zn`V0L@f$gU7iQ8j+j)DY!rjI)phQ`EYrri*RYEe5-QX z`6sr3^~#DYPY^`VblSNGL*bwK#>=;kzwSn;G;pV>y4SxOS6WrxL{+}C=&RRBO>@UL zRDbHGC=VfSYochGT@K^_-1e8H+#b{On&s=lm9w+=9=UY@wfuJuhYVD`bZBunZY%-DI!u?=LqdrZ(WKGk0NaW5$iUwYj|i0ILcX%}?pss!tgd zq3duUUDF}$_1YWk!t#+j)2yy822bGwMN9m4?1!X4pF=wRXT_-R$+&sp70;K+WeleK zBkPBZq2HwOWTb=^q!wLzhUB!V_j$|BGVwe*r9XHs*o664Uv0iqb`Y97I^*7MegB?{ zAg0Ijap`#I@}T(rd;b#0orUvH4`19(@D5DKIqsp)6n9m5$_La54?MM`St*RKIq0Hf z&_<+tC2otbrU}2@wxw!rkpnccN(MpO1WHB!{@qD<>Le4iL}jX5uZC@jom*CRv2kxB zlcYacdK;#2q}akVpe-_L8xfss^#Ut{Z1mWWGadLFtj|F0cK>p_z#uM8Zod<ut4>h`=>8NcY>}#(#_uqCOHg=<598@J;_^x^$BuN=eg=Pj= z%FQmS^TyMyTS#Qis2EhD}4!tuFk{+=cQ zA7$#je5*eVq*aOuIs$RbXa>2`{D`hU2hHHZ)27-$ix}_-Hr=ySx@*T)+)Gvlcy%2w!zcV zlD`q^@NSKVTH1NG!sEPWyu4321prY~l-ht1mvhF~&JjH0ukMo0LD70+m+(+L-dj#& zqc^fe7zC(tjHU{RJXz5D573IiAG2rz579W~tB@Kjr3~CL!G}k|$s3yfPfmTYnmaab@e-q6+IF$nY*Oi3roG(DX!@a8 z^9AAutu#szi-So`Hdq=xDi49)<`Wl(>p@*CO$}U1Bde%Hyh8Pa5}N2<+^-LHA_Ved zG0K}5+lf;xXDnA`x-wJJF-h2cF5u>;{1P3S5n3K*H>f1`eB4w#*Q4s<$80U-$3PF?W18PIN`p(kHq8JNc*1cJ zq}$I7dkuYZnU210q?w}HrF7wcJ+;bI?})@w=;DLLK1lX)_UgrwqOyxB_aib& zl7dU+DF^JCqh#4S%UA``d4d#UZD>+wMqA5&80nnO`m`3J|k6&3$?jc>-AdcT#e=gA6RmN2Y-OZ zuYsymF;gZp4&VZJ*9f>irw&Js+Zxhd0r%kqIuGA!PK3he_V6H0V zR^l-gN@F{#%!LWf^M8M*6KNOheUTF<8D)@T=Zb)=j7oL>t2XT5k50p2e=`061WAMl zg}XudvF0$Afvm}04F3T>%NvwAR-`(5HJ)QOUAz9{Tr2KJGGEWzP&$q>1^#)!8*l4M zp-mJgE#%Ndd&?Ib(yFpdiOj@2vrVC2OGdDYVsEh><&?OC?e_@kkb4J!2(7+%FqK?s z5E*jYmSKb6V1kSACPvT00ErfF)AbzkC#_+o@*WmB&e2)whLHB7+oL)`vB4rBZw>CB z)gAJlrL{~uT-v``N1U$>?7-2JM1c!*^PqOSwv3&8c`o|#DOm(HL%Nf>dJd*VO(s6b zqk5xLBAAn{`5LtqWvRR`jJzUUurA5ZuG_Py#E2=)&`SpuVU>3mPAOa^mt=p_rMNQQ z!&tmwC62>%)c=NI`+Sym^;Z&r8ZCcTUzwHaeECFR{QVL#m{{Sxlx&&AZ1m+*|LRrf z%kO^|yK0f7GPxk2_*g;U^?XAe*9OrrWXxKAf6zofxs!poUD`B*ws31iK?i;=g8q&R zHniZaJ$e~?XZuqqK;+nDO zl(V=ZojWd`2#dc1GVT7{XWp(2-uVyUQ^c5!OLsRV=35RYO&G#Oa1)(+!o25`r#MYJ zBQbgQQqKUu!GC!~{;1G0&pA4LqDMbWzmmPZ^%36?o@o}KFlB~^Q|ud zNua)CcpV6z3N7}P#n$VY%cWG)h#Bfzd7`6iug@1Ym(>LxDVQ#2OxsLT^A~#lEYP=@ zUeiP9N}RrDh*htURXFEp16U2X1Ol><0i{K*)64yX1dgM44QY z{zZfG7FDeAWE5Az&_<{C^>`M?q9g$6_<3h&Tv6MmgE22pBV= zCGZ;secz;4VMr(Y`vn$bH-lx66ud+QcYg0gv(bw*M^*jW1TItjlND z{=rI_llS$c*;btzE9AFUmaJCkXv-q#PiOK7uBKYvX&mBNqi|z4RWq^Y_KJjw!P2;T zB4@r*v6Tl4=ClOUy{EHg@r-|#JjVAR=;!MPduw(6a3hyezN&nOPrDAHZf6VAUkr-q zKfFdYB42{}Pi7VreV}6-JqvxwYCckm>){KQdS|v*S?vO1G|4Nb^W)nB43fi&7`ZwN zd51;SD_;z21xr`3btROb`^GLn@8V2?)o!Gk)3C|M+u-idi zyhMiiN9uB3KG`BUw9!(8P#0~_%+UL|m5#Gb5Mo#ouh27Ge^~UV0 zsH+UhNM3+>gpyalPj!)|X-NZwIZFu$>tv|}o3Wgo9oVowNePH>oN`lc#pSw;YL zj<0Z|#d%v1=Ug>1rKEMFB#TDwdjhpd%OiN2U0xiHhMs+Kl_(%`uCg=+U=dT-ybz~u z|2<#?u&(35wbnG~ZPnI%dBC=Ct&yKB2XDHn06)wD^v18r_e4(?v}-MpS?v8sIWF?n z7{X7b(!V1&Gq(9?DKwS05OpdpIC<*J;{KqedrvEXUe}i&SxR>{V%=l6w#Ou z>i{aLIOp=+ru2W_Il8Akv5!BlS99)3;{Uw)VqZnv7F&;k? zgb@Q9v;tgOcHi+@$aBvnH4Tqi!Eo6w?Qd^S3^Yn-Sp0o`aenjrj@vjM|MSjZBG$Vp zvKuw^QTAE1(YGuW@Qn#RoCYw4M79~_tqBU6J8M#yYo3IajP|b*X;E|gNec$ETLB{& zm68n|F;alE^f5)B%tHeKL4BAIK@Fya>oRvaeuE zQ``dld!DDtnw-lyGCV<%o#7Nx3vTgW+rrtFUDl=URHD-R&>|I!6>;B+$g!H$Pn@?B3g2YAg~ zu7&x4z-hKJg_h$gcm80=ee#ZFYeHmm6^5I&fJsQ7GrL60S$%`mLcR5)4o-(2!HCE5 zYRxEUYpx<m1AuKTze*HTwB9M< z0J`yNqNrOJjD8FCso>P)P}1-hBY4}5*P4q@7kASgLb2|i^y}`t#q>NSJYJRl>CHf-gwu#9um7YX!~poBxWJazKu=zs-9yqa1S2TWYYg8u>R zO||@Pzn~9;1hqVdP@sH7YKmw-RRh*S{hHY#Z=1?=M*VQIKIWisW+oPq$pD}dZ zlqD$wRjz9=fsT5~1%7uI_w%|wEhZS6HcB>ERRaLV=dSX+aq1k6Ng7*T54B1AU zI+!?NwDg;|?7Lnb#?iDU0d$Y?J~aaOGT?Ds16_RA$C~gNQf;&-Lus)}s*lu=DdJzw zERUm%oW+MeM8G}BNNHJl5$F2%+(TD8(ZLMScV8#pUd^#s`de-ye72YSB?AK%9c>&s zS+z@2?|cs9Z(r4n)X%l5_fRO|NFmyAn^Hza?a}D!A=uP}aSjEf!b}E1w!#A5f6j)D z zziWnaW|xwf)k22?uVjw@cg$U?q|{0p)3L+j!{WaWmucgX#V45bq$R8A2YsiK*5shM zbPG!yS8t}^KI(9%m%W%tgjeI1=iUV~x3VeKnzZeo_VwT<5U?i{hX>CQLIk}^KzCMoWz=PXjbw8*3+YCDRw z(+YU)YCPA|Eb!4RrQ);_st>U_4jrj0?Y?mKSek=Ub4J(_y&vqXx)+BhtW z_RYofuP;`YWxk5}riPL*dFSJYD6_CUE9tM#ZPrE^kf43*PB(2{@qhrm4EEv__T4nD z6yqk#A*yri0JScgluc!WS8@Xaom2>yz@NIRaKc#nb+7dJc(9ERGQNt9aV$IJGS@c2 zM<}Wp3Qqky*PlR_t@la*@60p#uih8X_pB4KRz?oc!NBN+T9Gk~U0%evL-2#H3Mzs> zDaDLss!`J`JWSiqwh;bjSnCC}ip~vP%J;%N!wS}GfYaLo%Bym4v8?wy`K#wVxZd4- z@A;ISZzp^GTP_u$u{GE7RpZ!f+!{;Gaxpud0j$T4r4E>+LIJu6^p`vadd5Cn>*xnv zqI74X<%@&23RvXHdoZO;Pn{XJ`{1WxRbhmO1&IXX@$9nAG4Y{6J;=V>a;|E*3x0|& ziCa~YP-anC$i*P=?~qzDxLen3Iha>jtn^NS`UOsjD61@rv}N$cvGm@f)+eu>F$r7p zah?fgBd9ub;ImBdB{>y!`BvH)2h~+f#&GZOnATBU@}+nnQ_iL;jhUE}#)uDj8=dPl zjoJI<<(MYgkSz%+NXLCBkd(;%g{)pDE-(HAT-?62GrS(Oc`EBe+TxCl4c<;%%0F+V zkCy*1PuelQMH%9mU&|-J|5w-CEY@Z{F%lIZn*#u6avJE~zYxQ+7+D^D`)z1zvu=M5 zs$Geud5Xo;Gn3=FV7LII@&VN^PYp$AKln0~mm6dVQxkw8j`xQ1sr_+2ZsW z&K?IJFVCFX^RKI}E56`LNxWO>w{3K1DB{Lj$ucuhP`!U;v3;Np4mI-^ht9KmLpMVo z9iyT%6oDJ6O714H8;?VR7q$;x-qyy*Q;m1+TPT+rlh4;!`I504@uGs(r@?`UwEkt@3R+evNh~Qr_?FDhCQL1MKZ&3)(#MQ5b;~7BVzv*dv(nIs}|Nc~QBHa!Sbmt^28{r!j zwBHfmoK=?ZXXKg9i;&4Fo)@e6$KQAtqG{!zqV#<^nHP2b?TU;i)SOZGp-yVkK5 z5+er7=N4POfbh+#A&CTuh9s);IZ`Yw6nR6-v#%6ipnVGXNxJtTK;P?o{)YlHa>Q?V z=fIzbxs*;8GYT6{MAv{i1wFv!wlYQ3LC`$!>8XNoK`)cD-e^L9WHXduXsJe-*HELh zGx5L>Kf?dwZ1SbN+$iWNZ6ro%36Y(Tnrahd!dDj;GU2Crm*L0CORcV@X=*fu{Ka_q z=t^5MiuReWRI-re&jQZ2Ut*nC`DVts;|HOL^Tk9kDqtQBH#QiBUm24)|0wg}rTPTK zh@KKOSXB(P1vByqV*v&j_w;zf+VHC;GDf`DP^JPLm=8JvPq6>l;3G8o-Tpm#9u@+t zYRx7fVWMlBM9k2uCxurS+IvJuY{`ov3!f9q>zX3K^&my@Dd>G?e`SRXy)EPL^5VEs3hT8_h!re! zS~aU?#LFl4?KV8qP2b9I+aMcj+<3+KDqF%vRJviQ##|&DG(sM9nCDX-&b5}lZkEs= zS_seaaWU*>VS^k-9xbIfj_SPf;-3nnaptg8ekW_hYd|@{?Ep@~xueK2rIDm2Y{F&o zjwceOAP>8r7{UPRuAOntc3|98;?ib0RLIN!l@m`{C+F((dsPzeW=YNvXkFMS0Zx*k zzl1wwoZBQpyT-27&bVL?;L#l<9pl~CTkqr_EzfW zRZc~iCh%!uiptu$%c{I^b(sMwkKNx?j7Oj{ACXU1H|5_qOZhw_HKvG7y@}Au+wmee zm9VQyZ_0#hgaDt_jCj%7w$J{yF#Qv@zU}BL&YYH1ja>?Sh-1HgqYRltZ@OnoAVtf@G>tdyI!i-5(wU{8zwK*bT)X zT7jRwf%>fk`%<=mH*GgBbCE`0lu8&~hjhWh!#?_p~Jf|xLfrKqGA?!}~1Q*-0G zG35!W7+CO!;>QJ9EA<#35PLzdFDaw8DAb>s`eL&b2J+M%)Hq_!Nw=xWkjpZH(@m@eZXjU(1()27QQgBkUk$l^7p48sGJgZ$ok*> zf=?{*@r}}F3>8%ji*6Wxv5L(R7!h)@;5Z<PSU(X_&# z{P4ZOr&($G(tVsFN8rcfKZ>{>uZrdzVld#@j5wHnP;0%N~=w0Xz0LMm8sBP zZv!LPM!^(1V~Zs zBrK@-b5Et5XW{g}26;u+`>pat9y@hM93Oex^E4RFjR*)GC>s<~4+%a*UKpMYDg7k8 z6@RbaVV5_~zS^@!&FK&6mnf)MZRX}a-Ct+XmuAZJt!$zRpnW^ zW@c{wp6AV4Z5V4=4~In*R6J~Y^eiW#y!o>Nq0g7{va48$FWmRF1C2P-Xj=0GF9;X@42Hz)6sO-778T+>-+h3((bf}+3 z%d*@Xjv^F)`5or%UDgn@p86UAPj%Be4oDHX;%>)8Qu?>#VU=@7VS;#2 z;O6&3hTa6_eo(~Ob@hESX;KBFr&7QlNI53c>c?EyM^H@b0jAkO=%%<=Q7Gx&g-?H( z@U7hJ8LVKO2JItD(mClAAMxF^q}azAl5fQlpEHc1?54XK+YYpGMfB*0)GiM6M@P4d zfmW*;x z`Jp%wR+X_6k=WS>I+WzV9Q$tH6h&BKVeSyR>R_1GqLyETYxYwD^WuKtE|YY5dD@3( zYWYW0S@*!b41s>+NkYjlNlY!bpS3k2D|L07cC6zc)^S(s*1vhEdHxc8a}Y`zq-3An znpKYvudVaskPPiE=tL=&Ce1-U3T1ZQ(;rP=&iVa`R4FUr0awNf$U#O1FnhA-0rr%m8<=lU`FYU}2B~ zHhxCT{({UKR&;~-h3aO~_$y0a&G2K8cJdw5Hjp3Ho1d$f--eJ>Nc&bImhxwXPv0Cs z&*_wA!<0qAFEI{*lE*)UXHJvz>2)&MwPXv#3?MsOU51jA5$25p7+%up8gG7%(_~r~ zkpPlXAba%KrYc2_e%*KxkUcwIWIDZ!WVDmSk1p!;MQuKbis_@sxG&z*+4V02rCHOz zS^W)G6!Wtqt-%!KNqOquEkK@%nN`O1db|>-@O245E4-dG@Z6V=(zzz-5koG1fhf#X zdQbuHE=Q3D*}N=dt)B9XW7|8aOpgbtUmSfohe%ZD2>UaD9PuCNk0+Y?XR@+O3B!9F z`*o!XO`}-bTN1w6CBgKIBpw3{902FRybr1G+ zaWzMNV$&hxN>~>s8lrrGQLMy_hZOc>o5~OZS)eZnR`}%XwDF7WtCd5b-yC=qoBI=I z*GBSseB9w5jvv=idR-`**pr6iaX`E_{I?lA2Jtt|yWo%97+jul6-*^Zv?O^#lX4Ig zGncb(K2b2P`5`gzE4UHhkZS7to@}*K87<_OW)#f-!MdnDdERPD^=W$J`p`N7ka~P zb@-|+O`dZS|N0BI)bWZz>AbAkcKXvE0!0KX3eDtzT7A;VI0&M*9($Ulc{iM$-g1QFzBoCP>vRgn;G;5+Qci$K2mZg# z99_OGb{zQ;K93hpj5=qAadMm9nNP=p40(KK5n_w@$<6wZ)Ja^?S6u(M=*3M{ z@A85+bK0-%5B##s26`DAJ&9%fB2{F{UF~*z7`}IZ{&iM_yHN*?0*70x zo7usP0Y+FrPpYAq7_>XLQRxiS{p6YnRcUfIi&Wp`*`JEWWQw1KbUX;^YMR$o$W)w>R9Tq1Vnp}V-V`wCztSCqN90R(a84nrKlZ6h?nv$vXI{83MOKH6G>zyV9 zj+!zuv>;K5#X%zhQLchy)WuJq=6a-gFpGBy+=*sQ`4o zM}I13RFiO_@+7V+3QwUAixelRf*g*wDv@13!8VNuRbhFy0?5|;nl#eFJSZRFhM~q@D!h6 zlUQd9wHRm{aO5UK9eJqa3k8|K0DOZvQ%GDjDO5`c5nriKgPr{qw!|oylmw`VDwZmh z$hr2R;P?zTqlITSA(hpkxW02u*GM_b@e<4K3_22J=PDdleG;$IO3^1SrA^2xb%9J3 zh(Jjo(6^cp2PWjYll^Rsc7(!ryNp}JmU?5hm}RwB(L<^K4IBE6yWsVR`7rtHcd`D{ zqPtaQL7B94d!j&=a@9FSieODNz;%@BLa`>DkTEl5Xy0mi!HXml1RyPV!l|+% znqw&%YP#19&@SvH-r>ro-iTySIXN=oVd5cZ&a~7H%9d0PJv2CvsQU?Y2TH-qE#b5D z#yk3ZpoQnw;x^g<^4-BHT=V!tjNS4cJQ%gs7y|XPmgf$OXYeP#FSen!l)ZXbXoyO? zlLxI*ZKDe!`UL!bc%v=HUtZ|0#W$ta4|Kw^hhIk77EZ`tB@szVR-4}GymQpg9Pe;k z_Sln7FiJ$r0%(z6trbzzHK-;~Wng?fQt6A}^28k?QJyR1KacdM^AWO8-WPb-O_N%V zOCCV#s+nQlbkpKz-P8XS8s53{nJyE>q0MZnyhkz1)xOFca{E;@+cP1^Q@$DwNX%!Q z$$hSHmJJzH@@W7?pQxK#MLUftXDFAwk#)c;sRPQ9d8;uo-(NucYqM@Dplc)U&H$WKctZr z&;c~Xd@DKqmUoPrPU0%h*_tF9ukFwtd~2#}W{Cb)Pbla%zWsbtLwQUu*BaMl74g2h z&?e_A(9L9-Vp}hEeTAN5E*1kxL(7rlWYQ$kh=3q{N?46ayaMJsNMD&1zUA1vXj~ij zjuWxu9F4+x>CHb1s;n5RqA7BP%XQsHv&WHAZSm?SL8m>{guXfQmN08BPMaH~ex-By z?u#AbIr58B{ztCCWdZJO@rrVR8tJi;ld2l1)stEJ&6~!8UDe&)AhiDbm-GnLR+|X6 z0qmJ{>G?i+3~?1+JWNXwhRDsF0rhGmlJZy?LVuhQx9e_3>0V0o9{_@ZW*C4O@UMHr334m@g4KS+w%c$f8?vP%oV;PAz ziF1g;2tzwQqzf>@E?Sn{f51rM9U2_~!~XU+IL~~DiHad}HuLSsQDvt63QWR<&yJ4M zIdps%6f5e^S8Q=s;__J1R_kntCa&?p;#=ltD(D!3!%M_*D;#L@G+pU_9$Hg=Nrm1rH6QKKseh7`e zEn#^2!jNx1!XQL;;HV$tdYX-T%3}onl@5R335}|&9lC-AN(AySBQ*WqV%L!=g1`BT zHEyNH36r!~Yl;xx7D@shTkG6x{7pV$!XVb15&FzFX`iUm{sX*tAPQBbnrRsisS+P9 z6}?&;i2}!CdB|cG!azt*gl6mA`@dy=mh%f#r6j|kN}cG(N!aK#{?<|nY}JmL;Yr2V zr_nDfRnSA^;pzP4y1t$=amJ>71)mdW)fj-+c%TF>9FoMAououRI9X&1U8%Ds`b*5Y z|Dqj5!SYL<4}(yU-49vyEa*4aa1Ypkxj&1eO^eIG$-Vr*-UdNu{Pd+)6?{K*?bbC( zZ0>ZW9Oz{Z2aOPYjVXaqFwYrAM=< zHZgZ1cmU`C!uE5^I&5e&eAGy|Ew_fPoxLz&PHu3^wWKBX=yg3plrfiTvDQ~uRjeT@ zgbb75=+hF*FvEfU#;rk4wL~$q+YU-rmtA7^zHWIA2$K94|1Z(Xy1nM!rPpt$-0ZtZ zYrh<65byIX&Jw!a z`KT>&C-Tz4XTT^)(a|!>3`XnEsv92(GKuU-6xn?(^`1>r!?2{wO(Kfu-wAhTG1UHb<%Rw=Lw# zUY$MbTx3Xcz~6IK;nUinUM@E7u~RP{@9Pc^$YX0*Iw;CYDOaDHI%?w64{a4H!0XWT zVC;ZIyGX3)z|oET=M`pqoj&w>iK!lJOKEs%9Q3tBxdqRP+q1H7%$_v>NFv67gU1P- zxE^DCzf{r44pyKU4F19Bz>_LRc`i;e zQx=P|+LlEG?o*oYRmqbqn&zL9=id}JRCaZ|qAI#(z1kn=V9%yf4}IYwdX^KtYoRKx z!g_T!n;b)n#I#iZX!9>v1trq0cl)sN>UqJb{-v8`%x`i&ZE26ssegAf$J%2}SK35F zy}2}eI}@fe7=y2)juJ}dmujb8gSs;WE9}|PQkZ%qMsq=Fz;hx&9Rx;mbM2~bMmLs8 zL?vkibqw|Slhmgixz4qdaWx)+-2)S3Cr7rLz?r6Am{Dc_C~)1E<(q%s_gvk+i~|** zU=ufE_}RTtMUdJhdD0r9q)A_ZJ;gSvX$oTaKhsd}w^hI?GQ)Tk}#sbL+yl_LHqI(T)id>(w_W3~)sgMZ=uO|4eCrmkXu9%Q;FuCnhTy_lmMxptG zZ)w(NaS=|UvRF3H>3Ydy==a%y?3bfK!Dy0~ih~=YHVL|5uYPu9WTB?uu)_QpUA-~k zkI#Ce3{im}-{56tF*~U-M1j4dL2-@kCrzbJAh}@TO~|oiw}D%vRiR@FByLrSgb%^y z{W|E;K1JN#$5Pi`^0)ZIymjJ>QqGYIH{Z|Och7Ah$0lbKq%;D?FWSsf#II3v5KxrK z;hr`Emzfws;ZFHm9^6uCZHCSYIsAE~@q*7*z(4+1=agsE*}vmvT;6hjbdpd1N*dooyG|Zl4JXV_voI}nl%SibUZ3O z$h(->|A{eh8Xp+%@W$+vdkxE_#C>0?xla^Il~P&?hRZT|!^A&Oe3us*eO@|Xm9a*X z#cr7V8F?IFxnMZk0+_(RFLg}}p#{v3k7A&^+vfFjAhs--(JiQ0KAm;H3#2!YTBO-`;aQC*nJSYDdvn!CZrb})7% z4uL4))SMS_$M-(A`u^ULO(#?1NNgiUehl=IUsCB}UwxMoUm4pXT@U9Uo;Y6$A5V=_ zu}PWOS$1t-VK3F*N3birZZ?uVu;LQ{Om9xRwVKsXTbvJZs1TSXj*T~-!-JKncQXQ` zR&5f-ITNbqUZe=Zuw`{hLu_4Nfvs1jrg9TeF<9s&qD4Tcl8D;mK^U7c0dE^F^JOOH z*%nY~HIH%~9i5;+>)dfu^w1#(@s__K243LiWaN6!kg$#8&j|jJhlfS3^n4l?k#ocZ z=$JeY5KJgZM)~l9zM0aCpcdif^0tKG`*U@fm)#4N<}8!c16l#VS-7Db1?ksJ>&WEZ zprR1Q%Cn_1addGv$7lC@TM@dzkk+?5O$331^0b_ZJk190oTSzg8CBV2!Ct62m_VW< z6>!FQcvA*hpZxSeXYl+4HW(9>8&><{>qcfG!yyK-`mEdTI(3N>_&%9;(?ZGn&qJh+ zGA<7pt^t|h2y4^y*Wv?NH@h$U?)-vj5=8$V7XpG}=szX9C8T~LAZ`!oJS6XGYrOx! zZEj7GLY?g!1;@fcEETn7Nx*gm#+1jxuTiMB>Jj7pC|M7Ye-->mTAq;)`XoxRpo+}4 zU|)yJhX+01ZxO#tRWLO0q4XW?a!JBTf{{8s%@R@e7;Ye1RQb5zdEx1|47AHP>h@_) zFtXs=X6preFCoKQ6)3AYV|us*T|r@jFplfw|3lkb{Wbl7fB%Eg(j_r+qXsA?-8s4& zm6Q+$qq|!eFhW4OyBk!HZUqFS8)+paKli=|zoYN(kd6t2K zRMgT#3w`@~cgUR*7kTNJ4HKg1@kB|^wB~Gr{tn4GGmk2q?{`iIUs3jE$r4%zJ(u$y%8 zMb*V}bCeEck8L>9KUpav=N4t>hL31=^x1QpO%1F+;yHI)MJc+4>1$pW2BA%yBM&kJ zb=?Hq)f<(!-kj*?`k%qMI@6+&j93Kh*D9;a{zLx!hZP&k%nEsig*{yF6~vRD(CGf; zkou5X4$TOd0lgrGmSk-4WUfgHW{KO>Gnv&(%2Phql@Q|XgMRhB1ATvpsQ3@iy1Qet zo15y;e=`1(@e$W~jsnvY0L{HmBe2o|#UrAe1^rG2d|#2EaJPfNRYxcZ>l9KQ79mm#ynh zSGC9I0)7g={++o&7S(&`sI4q|EdIyOS?d^n+m_dA{%p@)(uPC2Rcg-p2Ny0I4h>^x znWE&dT=!S8t!7l5eyYHA#L)R-Vdg`nO_#$bEULHpm<8?c z2<7YKej_H0CPZR?t(Mulhf<6JTcJ37P4>(Q6$ja&6nhQULC})~^%v(8{gz|4S5i)_ zr$nZdUX#mUU7Yr>Zqs9JSP)bYR9W$UOZNZxx+bHHNwx1>(uPZ6{T*u zfucV!99d>psaIqBPJQfoPAh+U-ybjgF9AnC>_9VlxA{nKkQRkcf8SE?J>0|^^J2q| ztA?K+IG;-_CweHZb86-?V1N#zBOunx9&>~`QyNZy1K2)M{+qdzl3sJE19-ocHT^gt z0_m9impzEfn&&Iu{(OoSWQ4NU6eKhlI;8n3c2(0RJRm!u5-qh300MV`@ea;~`s<}K z2!3ZN!E7tAs~QavFrI%6xuS=Rp^RoO%WK)-*L2drDvF0H|4#=y0QZF9KL8tv_J4o| zgOEF6v%h!$+nROxKSA^V>sL|Dvdw$`C$d2!Dl2uG`Bzi^KULiA8rf-)$A7e6`Ogag zF%oisCx$=6m!Tl|l)pj4*of~A#iW;hlfC5CcZfvk$V0Q-uTCFEi5XZ&&UU+#Yli8y z=T+60okhx+k65{#J9h?u3wYj&0X8UfA<8P>+{Eb=GTcHR3XqS#Wv?IzNbKuRNoFQb zR^v9iy=BQ!oot4_8#m?41#5B;+n>B9dEdBWWB0)zgOeWf0_EG4Q)Vx4!K5))pY@_t zOe-vGb)Y*F_-(uO)#s>2&bWUxmIi-Y5ZUG*0Suv0x-CHgv*WR~L!xY5#9B_(Ba*Gw z1!5!9YM;p)p(lyS^~n|h;$t_zt*Xakl(M#3-nLa@uB?tWfG$?i;FBGgR2e76stIHfwwWR@-bv!D4FeRQPOmax)=t>lR;?ih|A zynH@9MNQ1}gB}VGKxdZO)4LNob#nf)+CLlqNcYo-J!y6_3Xd^8 zP`Dbpj87>zO)ohSB^-_25X+yq2%r`dTHO7){%)sTft40A&Y=S{I6!$MwC}=f$#~m8 zG5xAxM_GyWB{x1E)k|dB6oB(qv9$>WsA82$hBE&7ewX4^_R?mVQ`W(btl(EINmTCyWLDCcAE)Rz1^{N&g) z_j=h}P|bD)V<1=TWUcmt-{zO5aPErZ8iV0I^8s&-6`TF#(b4T?A;Ma|*@y1e?5966 zRCC$X)h&{^GQD_yP0zNf*)(^5ZjN2``)X;h`kq#wzy$*5@y|E&9}jU6`sUj4lx z5f(pmR4C^n+Bq*K)iA^SZKERlj1kA2x$EKWI zPE8tOfn=>X(l;GcLOK?YWmAyu#}UPg^9hdhhDDN9gVgmfo!ceV4pmr^cUpWRr^hbG zlv@D^wRpBNOU+H|@?{dVfEcZWH32(lOK4obCIE8Yg?745{VZ~tC%IYNwHIApS4AEFocy+?RtXBRi)F`)8v&9fZNdHH z0S#^3G@aevNy3|8tBmxdiLrq8$`je=?D5w|9hgrqvj)6(_^ra`?nj3=b;nYs5}B^U z*pGH&ERSebObG8bcBjqaY-TkaDXCEw7*BE35SRMY3^4}r^VLZ%Qus^i%yaiP1PX+! z+4Dzxq!^r6o{jaYsI|kJ--!bJI*gE>N)GlNHTEU0cibSL#>)E(^3LQ3`2JNhHkQ&@ zIQZ?>e*phlp(K#TcSp{^CYNmi#p2y1xfj}s`(blipM=V2cUH?Vi1G;ap4kc5`@quN zp4YR7vW8ExIGpHRMk5t?H#Sr`rf+h?Hf%j7^?qKll>VkzRSh9T0odZGgbOu@s;Yv| z-d5fjR@BzADG+waC7=zCnsixsvH}X!(;SZ}$ksg>vOu+u3#xW5U9KclowZ%tMA;-tA(&vMPq)w8Z|k-ylG+ET-XRa}kvg??u4w?A`RGr|Cvsty-}Zv_bzfx?k+Dk zK=-%AN-Jg*BEzn2``t0gL@?!Y<>=ruF>O{#G@4kml8jsC^s}~O zWnxBPAUCN@TSK>kDnqZXg18rVO>W8~8%5JIohE!v@3SU0U8^dtv@hDWW^ytO*bVZ% zsZ^|VeOB=9qU3EF=l;5oiIQV6*Tt=`T;g`9ipQD0eB&_PxEltcV29-fG7G2hgQq*# znK$9&<;awjI9+|^T6eGRqBAoU29+D)L`_r40%7dUJa?ti)xj#vBHO!Ps$^xw^Un_2 zv%TkR81T%vCkGnt{vPhYxS~XtLy8AP!lvHm;m-vgFKMT{7df14snYKzZp!krZ{W_p zBFQv+XW$2ul|I-j@$?Ie#hJ#CKe2d3%-zqhRjXXA6!Te&v~9r{zNUM^t*j=p;%?yh zC|_{ukufvPc_C7OK|~rdG%qick5kkgW*zo}30%jY&=w-ys>@2Y(WOiS(4Ih#!OH&G z&U>Y*xDJGpi_Z z3<8PsYq^k{!cEjfq%3YK8dvBux8CQxu=j;g6w@hL->Zdx@?hK|a+1{CTGgY@w`_T2 zhm2_tJ@QuhwZla)UwK2{LTuek5b6k?AK&6>)m2`O4Pp=m?4xk(^Nl*iUNb|#}2)CPSrTHnfz#~T$bCF|5WC6e3) z|LL2Xh$pVUL|GHB=?ianx`lvzhF$%03Hm$-Fj6LeQ`4Jb#WFE@cD(_d4eHXCN@C(qlR?HjI22zF)%W+;UC+cQSg;Xr0@rYM8P z)Nv2UCyOd7f}>Det6{T-Pk!F$@I^Nyx7SB1;JdG%`GgVr;2gC6w(>-^Ivp{qLZ{dO zU+>xZ`%@M74>QuK1rdcZ{MCwq6$v8te^k;aRNTVVDG6WByb?+xMN+4z8Y@er?F6oh z?$5@1lYbsnY^)3|A zM~~t^VT;dFqoIppG-uQ3+4hF#q7PIR^PV9J9Y*(Os#lx5uoWp8AP^L$#n?~1@z7_F zjxEV8L5kYy3<vrkAv`CH-h9y2L#{oaZXXUp{9Xo(>+)WUkKADi8N-L%c{lIc zn#gI9cB1+nHrc;sK_!~3Icm5TMz6Z4)5EvVe{);^Brx}TPj{Qe;U*b1{I-nS%-B3@ z>YIw&y`3#Sx>9Tei8%SKY0^gjGm|EkKr!^?+xq(Xtk9$mNj3hd&5p}qv4<#)c6L#M zAsu10#NPmh#^7DD*OE9O# zX6;;H(y(s+fqV~IyYdkThK^;P-&^Xf;tbk4YNubwoK8_f&!zLOoc@}a)FjL>3GS^B z^93Y1)zZV0Bz6e#R>p^KZL~W+aooUrD>3Bg+x=4?ew*Pp!dbmgd>D=bly#!h6&1Tk zlhd9!$!$8&1_)cy_I`G4^P-KPBV)X?sbBhD^*3iubR&+E^=DF*%O`ND3c%VuRR_)b;;EvM|uMhpGVQr^n1g1xte1Fx1YLT zrAWNX!r#%_bYq|zIYQT*ujOS#c0KQCM;!i`kc)VvoEbw9JV;DV`5lN=$D5Ft1aMz8;eE+)|96>VrFPzy79D-ZvWH3<&F^Tf zmOd@#QF`N|*O--{b)GQ*p0cdu#*XKuq@MQ=62y5+8962BB_J#z!eI6^0XY2G(Uiqc zhj~}oWMAU*SDtylXM10s@A4pYMb{=Ox~Xjqs1=d6Q}i5pPW1NouVymXR20IpgEkG9 zHu5tGBXqU|nMZqvRKZeJb|o+!n_@xmr{MK32Y2w2|YlU!iNV(+n!LJa}f$ zl{_vy8Ox%&ahq7)VXH{V#K_c24nQJ9m5n)S<*O8^M7QcsO^y>3Fs{!=nMyTYeI9<1 z!DSvgRTu`Ayq{ZeXZu!s$!wY6kI+@r5U94z+aXXbdE8(|TL(2y=!vQu5~tI~k+&ej zpn)Bma@Jb&Vs&Q2DQ!{qf)v@@G-5&3dhcD{VT}^y8vnesMnT0-F)U$drJPGQy(j}0 zSzJN^gTs_F9vFi4G_0Nn&1i8^1}Ndl7YSA6*75}9zTwm5(M)=(DO=keX^k|HirLC1af~q za$Fe4MMSkcOwM)e-0g^k$y==H12;oaVi?ayyIY|jauHgLFS0t$z?}_p<)H5CIx{LqI2BlL0!IIS zv4^b0q93FQa$By_(9KiPG4xMG_UzcqR@ONR|C&`3RxvDIGO6uYwnDf;+Kezhkfzij zS_p%-aL4Daq{P-R-^iK=P}!5P7~S?VyIQI+Z-h;21y!*L*5G{y$_`b zOZq6izi23b=N$hhb3f}1)2kh8w~8!@hR;62(r+Gvxad(~jEi|k23x`+6xcCNX@|N0 zbVI$boNT>+Z_zO6<}TSkCkGPKIXqbar%&gIhccAgI=$mAOK^VEXA!0}nxxD1i%7Pi z31zsoVlR&CRrBxp+u!a=TQ28P(0UKA0%?{64`2nhZC>^tjjWs6bCzAut%7|#v0G); z{eR7J)YgXj*WHzQ`UwrP6XRju8M*<}ZO2!ySb^fS)<2XvWQQj-J7H*Q*&lpxxvZF2 zCEL>iWAG1$dQ#;?jjDoXrfdadKOAtj7Di?L9zX9Thvu4{QBxATEQ7lU(={dG1)8x; z-!J+WtQMECQH#Dg@Q4&`WWa+B8DfI35Yd%1o#&^u&2)tyf`-?fcSic#wWy=$QsHE4 z=997HFhx}Uy0qMjM?NWRfSik$tK^x4s(IjgxZlY4#d zONs7To9&3T1fDz6uK)8YuN}xRbNHzHRXN@_Uelwe1$paS;=5r}7Vy)YTj71_q;NW? zN{4N>#Ag_E2NYU8cX~7IkJFvd`nmmmmKS%T6i-w&>!<4EA>$Q1Ffjb-v63faY+0+L zyQQn6$-I^R<%_ZIaQdQ(oroiSS}(=MHa_$J01lH-tn5>j3sybQF5=|zP<3hSNatx% z_=9A|*ztlOkW#1W;(O5U^jCj?V@pr-EKB^mFg8>xKQLBMGfi zRZy~zrpkQY>711HNg}2m4$U4d8~5|8QSlJgea(2NC#>v!mFS;(G6-Gdil$fkn%Q&y zDH`YH+A3GyQGQBrOK6U2Wu}XIAH7t+^_CHVb(Bwjs>4RZm2}H=%L>F}B^N4>{=Q+h zSWa``vQ69GVH_yPzEQ94xH1?hP;zC7Ayc}@Pgf3pH|-`<3YFUxvOUDG4u84M6YsI- zEb6G`OUnT~kR|T#d+~E7qO>Z`U=R=2k9*+l@ztnUCFOciL;TdL@Bo{E9+_?O#;(T? zm0-WCiXV)tIx9|oqK&S;IA5fh!ly_<$d%omX#c*mEr<|#N#;`KMKbEAJOnF1Kz78E=q2lJrRyf`EI z<*gH7Sq5oQSfh`7`(cx$eO{op|cR(nv$e+U529_+j{i*S^vz6)BuhjU&aC4 zWSA~xoTGB*;td1aT5(QDf!v^J3j%Ckv4VcCX%T8-{w$h11W=h3YY`fTGf5RA^`_C{X~D5_B>)_ zyx9$V_>FO`CQob}YjhhC3rXm3LfF$QlPS4)(K6RnVXQEBe96nZ{{Z6t_ysTmSTRCC z1bP(CCzG$%nAEI1(*(|OVNTed`zgqrh?3V2-tvRXsJ#{2XUq4wm(ZW>xYrGXosBs8hIsFv?+zK~>+4C*NR2h!erO1MY}*$DE~S8p$0%jMv>Yhu zymZct#SBA59PRWlo0O5AApY0*reONjHg(VoW>@sNV&f=JO1o7N)ig>A!O^hUfo!j% zN6dQtagPr^;jEk9lzPD~+h1os(WM!v8W{lGsN7^^_6;857GqmbqOUhjQxVoBWdH7e zW)|2hXOT%3M&Z&P%%_~@;Xh5^554ZtA*J5A1U_7r@qWV_-(2!eK26}lzxBT=FT!34 z-`kydkbK7h0Q839)?useiN*^#l45Np$?!=f)vB=5`9YNJ31ruIw}Fg#MKwbN{g5&} zx>qCLg7X#TOR2tBVg)Z7Ne8!;AdUve zzk@2^c#}WjKl@>rFq7ZL5>q*noRU#c=`?&PTn?G^k^{K&7~C=&KuqYaI1x3qZvo%& z_pFAG@zUX*Z?XBh5u5so0U6#)oeVFj@4e*eFydhP_?h0>@IzNbfl>@UVPjN(^hN?P zV$qO3gZ*;CG{F*M&%W|3Obg?N+9252v2SW!-s9+Csk93ziozQ)3j8BD@}}UaC9h}^ zZf2VG*2*yr+f>>nZ{=F*z|MMsyQgh;7+k=7&DfHHnP(|VukTAVBJ=34)P;1UbF{Hb z4%!`F(_hj5MXuGz7ZKbeAR$T4SaCpR#6aeHGC=U!w_b)=P?k~Y<^X$8!l z>^^=s%oCL*6hX`KE*~7ZqM}zc_oC$})zm^wQqS*swb)&SDMAAaFc$AsuW+UAX5j>Q z=^w({aUE&JI8y)3zTvA*ci53Klic__%ar?kt~RD?$SN&aA8~6NY|eQ39nWigpif!2 zF^b>Yl)jtz+Vw6;4B!bO z8EQhoT22i+>ERYKSe8t+>XgxIucQvBO_if$nYTJmQ{T;Z z3AtFEpRgm?XOm09ru;U7)1Bz8H7xf%V;si4rVFZ!toGVq|CUPH|IMe->-(H8dtg#! z?nIwlUeu873j(rQj102kvi|ob%V=m)%Y2VbCK1w;(lHO$$w?^^q5W1JTrP}WXBd*S7 zsvmRI2~pf*l{8C>OoB{m(fmB~haxbWH#vxde-uU#)tM<(T0882{S^5_A65yL(nn3( zMA8NriSg-^E_c ze~a0%Xm+8Ge^eq`C3&X=?+$Za<)(r#mIM$W=@S=h$95Tpex|^-A2$#5N}St86Bow7 z#ETn-47Egygw@ev`UyD(Qe?1nI|3hU1$H-@8-_pFa^iUQuOqRN?ekJQ3h$5Py-W35tWLjm2l~^|^}z9e{Z^J7DY&p<<`>(z`1e!UdQ6qsg~M$a zEHhOjw@2y8{B!7I2T~Mb zO39NT!g6N_D0^;qyb9sK2)6=>`wa8o^7Oqwb9q5#?w?z0;iIXlx+ z{8amai#Zzd&_7!%S`E|tr*@8`x*ETO86uv&KhDSVAr$r-N^Lpjz7jDj6?AY!lR_Af z<^^U_OS>VSK#I^3g^P~%R=%*7u{V9~wqnV+RTw45;+;AzTK!QU@|{OBd1S}Za}ZL` zpcX=Aw9 z19SHY&?CEVaGc2RB%04cJ;yCK(;zKKF7Yl-%w+I@`Rj=3X|jagmqkgE!Nkdw-8eB} z%Rnkn4Q&Tyap|jJf5twm7F{LT|F`LLPy`S3JADa9(@%$35)|ytFzs7^&c2V+3>GXs zKVZL1w9?wGSQNd12L;<|+-~SMo`;Ih|Lu&73D2_{5iOUaK`EpnZ(q?B`tcEI(5R3B z0;lf`QGH{Cd^>k{*$i$#Ol^6@k*2fqC-2?N;J1(do{ChzMwkYIG$ky|S#XTx1ptyL zw0om_z_F)(l0#m?K!6LJKrYH5bR8nC+X4s$;BHR7Dlo}l^U3{CObO&bMzosow*;_x z_!09RGdc&s1LfN5!5p$)xnp{B!wE*C*^m?A&_VOEjwF zV{VJTNI3wZ(nq9x;(4t@|0t9?ydvJ^L<(m9DotSeSUDm!ZS`-V!d$wPp~WiinPqRK zPH#E^W>s!+PP*jv6t>52ZzU5?#E9PqEt6F{;|q2f#HuVuUI9i@W$$k%*@d%R&hEAt z&n>A&byP|?4k^Y3MV_|vPu1dR50KuWzR-C^ybLGd%>}VXuLL5m*V3o3boT9pgX@c8 z5eyf!a~4&=hi${O%0cTyYddjQ7=8#ltI1FdYK+QBYqwy^4Dq|=ptAOw!@8)L10&gz zq4`41&P+j5xNFeLLmZt2b}+VRoX=n^Ni9GZsgH_Y-m;Ps;dlP9)Bcd4A}pQNd?aF&OBg zqGJ+B4*)v`ceKT;SB89FJ`t?$I>E|gRQT1o8C1njlsiCu{neo-YZu{ZCE+vGLu&i< zyZ1~F@@KWh>5dauFB21)Ac8G(1iHIoHmv0@=*cao)5i{!H6Ldw0}#VF2tIRSwAYJV zN1FLIxG1mf>3rL=n}C*c;3ENm?OFrw^26*%z6jpk$r2lB7(NaadK+Jc8+=<$+E&mm znDr9e!^(dhfCl|>Ev-cq052-+zLGjNtsYV1KeNaJ3;*eFEYnSZ!&ml?WXH++t|!d` z!Yb>EBs*D=E8KV~U7I7@42UDu>qIjN)6$Npy3?P(^yV?^Qn<99h2_*2yds>xRxej5 z#oujwE`SST4LX;;`n{b%qW!H#>D@t)jSJD%@=|~EMDqu3ZjjAB_hNo9CnkrlnS1n^gZD*8syjeaP(| z7a6Hu*=7>i)?6)%dT+}6iyQ#ox8A97mJh9cyoOxXvPhSFBN09Ycd@^Zz7XDd7+|JB z2c9R_85~nq4!=oQF7V^{;iV7cAEvs|=%`H2o`J6JzGJ_>Zl)zTI8PDyG^$?co_U|+ z!y(D5-nr)!a_mFI`st%#WNP<6`P83X_A{kPC+DhU0l(8|RfC$y^>~mJi`yADWkp)! z_}Nd)&uG`j^>9)AiC;b(2F#NFn<_96cWewlxQ+%otnEWMrDCz{y0@a9W>U6|_%}BN zgnwb)PPd;XomNc0`FCe%7f;A-@it@hxj<;4VKIhxNLO*wO~qts$=BaA>sk}%L3M6B zKvuw~!2SnjPvB>Oxk43FlPij$S=6+AjqfY}H`W`eW{aOG^dtJB9D=?InjI}}xQHN{M2E)|19RFQ5)gyDO|P1HD*I@y2LL6&g*;7V zvQ)Tkia4o?{>i`WPviQe?$2O;_;biCnNq16(wSNDFAG7%EMMSw?8RZ;VKW2V}#mXzdkCNu|kr_NXCmo`&R zstAdhvS z-&7WKn-9L^tr`)JiZ0wXz zUm2<%S8Cq*nmc>|0kVe6q&Q18*$@WltI0=7j0)P!oCe#x(+lgha1v z`X?pJUpSkT-986$VBIlU#lu_%;^V2>ZFr-1FphpJ%aA(BIcV_o6~8+)G%o@oxt{cL zQRXO{z*GBE8OBTV3TOTdVc2#Y8}R^@HSvPWaMJj#!AG1|iNMmsq>+~p>_(aL?bynHm!O_#_V^CISNdR>*RqOn@pkQ zwEHq@wUR_VeAWkvh5eTNrw;KdtC$@;jXPCMm;jlX5=bT;QS5{{XUEB?%uK2=O>#f9 zgBdpPVq?tRz(42uigG_LGWlC*v)EX;co0`60mKjUaz)zN63svxHZ$maDNgy(!xEb# zv4q$yIPF?tC0>lmmStoGV~Rc$ahIyy-iZ?H&#iAEkGx2*uisQ_RPqz$$@HTJ2{SYgn8gtw6<+tmt6@V@vHEbr;o6IG2hZ>Lt)&nqzJ*M(w>XOV&fbP|U8S zudcq?P|eZASN@%SjF`%sqHAwc)~i`SbMFGtq7 zh5(FT_4Un@#Cn>a+7$ZiMvvz52J}Xor>ExM8BDD_6l~`&2wlD0`>3krW=@oIw|}(M zx;r;@P&gYU$mtt>tyS2LFB|_iNB3XIiRB#zx zkg#pGk@S2uExZI3=7O^&O=f%84q-@deF}fu{wD2@ILr((;8nlwe|!vk;2{dI*rkUg+DbOqPh=IG!}xREL- z=$B)GdC1H045=FFC-;G?-pbm26I+2Z!skRe=ov}z`aQj}`8YleOYBj^Eby{$o}Xh0 zE8ZG$UMJ?wAo|pDVkLlg+6XD0RDuTIdBU}^^tV#%b>3nMJ3AZvUKEwQntrA4m^d$z zoUvp5>b~Iq=FV_grxJvmktVeR63dXxL{(fu!{}lbPgpD>Qj&%bmR|{>qN1p<3d6$- z#l&tZH&NJ0*gd@=yX96xZo%mq%Y)*0?ZRL^bf$V0y&aVc)#BpVHE%Hu2NKxv%*(L z?N>Is(~}E6=+AHVzz~v80WOO2vR20#QkkjO6wMVL-4%wptpbj2V$YT;*NR#~9O;40 z?#~PE3`VFElz9jciRv>#q1jkj4*wLYS!N(G!ze0mrg0ty?G|-*M2SHf74to(7m| z!~Wlwv`LuCb?lgxPaO=%l&hHEU9i&F_Nt;@X8RTi=%qcGG-*%Z zOG#gFM~aF)Kzr&=r(!Vfcf5wLDn9nRr{2`4d8^u%?-J&?`CV1T8>UMubaQabAqLec z13q@Ff4f*KFysPrd*P;@=(Ar!p#$gjSq>KA!uiQ5OT}&2>tI?yvg(j@I{HIuIwx=$ z!D76vAWIO4VAO`Iw6eEiH$!-4M=uhmlyOwdcw=}w()Ihc!x6I=6?JkKca~nD%SqFw zI7o+KN<;3b=;f0EubrhoYnG9!T0`G->DzYxVW%H$)BI0%&VLQhz@A;H;3x^l8Sz&Q zlC}VG>!@NI{}N-`xU_wadYpVOc9^XmRp&#%tOE7o0O)S7G(+Ip8XaxQtZg1EdXTT|JcZZAC3ZXcJy0!CT-aJ&}JL5JJVextIFulQwVw zY0DdV@!qN-!Cgug8}CX_^=N#ZJ@m#377A798&u_EG$R`sHe=IaAM%}i)!g#Jw`l&^ik^3!lw&07 z%Umrg>#~TyTSiU8K~|tV+v&0#pu+lTafT*qD`~UzBpY{ikQrlkDgDctPz=2z{#Gcx z5~V@H|D2=$cYyrrn5o6*Ax%o=sZaMm!0%u0hkySMeC_}Cg)t(~Yg7BiV3j!mNon}r z3;Tv5wC88hI{Sw7yEC&s2YZo<|sY^0rj9D$0OSIU`7!gWs<8q1JenO zsa{$V+DdZ3h^l3kxX+P6`nSV@(8iz{XNk<5cGptolug47=C53b68`bo3Uq}mHwjkS zH!6rEeM1fk73bb$CmiIy2YRUL$UYD$#HBfhiL?DN%K5sYjh8DTo+gFvRQ`~zz| zrx*wx+;7<#zBpY*iclaUXb(mYEG~S+9c#5!{#@tV5q%gJE@U>=ZA<|SC`9j^3nI=qz^VjYUqi7Gto#xG zJg0H`>v8E)hfmfPuJ8KQdYN*FL*ch7!Gfm8o{P@z;{=k)4vk?L)PF;YL#ObZ_ocwJ zcX>`K*?X%yc0abcYz?iroC`Bi!TWY{Q`&EwWLUlvEtN1@xOwbV2hNh+lx~Mj11}2( zA?8?JRe}(EKmKM4!JyUt@9<5BLtRew5ljW_^b#Bnll~f5e9boD7@w-{qL+-C;YYpn z@AMs8`iKVTBq}AqrW&-3r1Ttu77sdW4b7bf{sOco7RSQLQ6gkhs?wh+^;*naUlKh| zIPS9LyfHC4|MVjBef*6CU-p6FT zM;ACAQzxPU=k)i#+qZ!dsaS~LTzIu!_EK%B)Sv7E+rT^~nn~ipk$lS|y(jBbKI3yX zUtzps<_9yR=Tj!Ckl}HFeK5=hiEY%{F>`iBd*zF=b$C*Jv6Jd#mwEC^r zbvag}J74zi$02jMiUD@3g1G~3xJpt}POks52kjlNJ@1`oDpb!PsS}CP=v&G*uv>p( z*_Bbjav}pj)!~bgHT55$n}!Ahk>`fU`5s{fXlO@~FSzsg2l_e;&7=s=-!}iqyd#kO zz)7mYd(!WxeO;Zr@0yiEbwOf*W@Kq35|3<&3qVKmol>Ylk)j*y+LIjMdMa_L$)Rs| z*5yn6kEK4wgrCB^Bio-my0jh@)W#$ zH4iaqMkq@-O;VmF0!ixC+M3o?cd~>=+Fv)CJ}clN#{`f(?r+=?`|GEzr)$5M)M?4L zr~meJDvk4YF6U1%W+*qUP`AP!?#D^^SN(64>=K`H?)orxDEJFOk=aXH~g1>8dy*^`dFEzrcP*8{tY`75iNOcoS&6+ za3V`~rn)mS^XJ8lDwpqvzz?}Kzn)%*zR8*4*>mrl0Sml=*+|<0o{q^b)B+fZDG@WJ znNGy*Jm^@bMych}vEuuTQ3fAXtrS=Fm0O>-63%Cs-Hkf;(WX-}^N5*b(DVnd#gLoTA2Z6apt{eKi^y>eHrV|5$vGLl#DoFY-U+<&EI zJtI&FQMoZDO6>7}qO#r+q+W0-pg2FKSrN<&a*TT>^dCUrBxP9XV9w)2LEGw}lgAj_ zDgj8RoEX88LGsq(5h~`xXPh<~~vbej~-)W7N$$&6gjspGzkb<^85eT^0ulUjG zxRJC<5L~l(vC(O8gD%2yc7SRraU7d?9NcN5ec)14)I+m#qA;aoVrb>Cjc!TuHwc&l zFz_b^TqbsKq%PYidGem9`2S(%*x86)_H z&-P326KNkzlEAT_0K z!G>1J!+BRz7ORr~=rgzHn=5sd{L=BnWLQ}PNAX9G%)EsWq;!EN!*$Ta5_VMsiWZST zqn!hg?l6F+Pt013F+Gvy^#yFHrnZQ&NrhIh&~o79JUvdYTFW016)3+oR?QyDhk&f zcq^}&g-UC<-X#sy@CYnRx**7-IB4*2IojJVD>yniOUb&(`J25x`bI0WH#G{$>#skgIu~7V#fa~pcnNkXEh*6=kcWT2$yHw|^SIwjQ4NKP6 zV&|bOpKmL6<5g0g+XkETEqA*;Xrr%xH9|^+^({7gsgh!dyj-Y4W9_KKC*D2?{4PWs zsr*(-zPI#kGXDtUzcaPhW+o?_usPDgUtv>-(NcE{I6USnR~t_qkW)Xx>OT? zj%6O%r?h;&LueCU@197wS?pz%Ta3&<@i(AZ|7=*ohI1~r!Q>w}|BSe^gTxIrA{^Ea zO-PXoI>&{LB{_dS(PkiZUt4-h~p4rkOH8ksdQnaBJI4$lMod~^``9@Cu6w^&pJOOFzfLNFm==XRfei9wq36mFw+Nps z4#edtm(ag&6Xufyn9oA&7@`_Xi z+MSGj;4Z)}lop)Va?7xVv@>QYcH0cyNSy^=6v7oou$l{uZ-+A?yFzkwKH_ZIB^NT2 zb^xQN-GhQEvZ_CPi+U*l1WS{W8ART47+fogVT}|yA1>*5-yS*OOnrzO=;84 zRX0G_48yIE65ghPswuo7-NuY*N>sxqHss1PmWF~XeHI*b9GRrst;*Gbfeg`*wC55X zxUCY?Dbx@a!%&u-{LDw~awoknnmNcmg3-__QT-?1lGX223*I7JvT7%z47gvvmYQ@& zOT-n%(L=@D(Er4;84BZ>Ww9)GkWb~yGlgB5E&F(#_kA~(VF#+eSdQE$s_0~0R*m@9 z1L)nbb-I&WzPkxz-D=+?n&v5I>kSrz0GO41(wSE`rE{zq7*tjKHVJjOI)3P+NX2_) z74ph_B4)5e91#0IKtJV2dD?nL#TBsPw8qB$Y{927#GWYa{^y^PbM+aS&iVu6b2 zjR==14D|!Nrd|~VBpC2HF|^bFn96L|80xoGC;y{0HJXNbpZVM=*3K@lnxQ=D=`(pd zv|-!(7f(|hZ@Ue|YqX3fW>l7nSp4eI$s}!@{>6+j5V3rtY5XC2$xT*tKR#o1mQI!Z zgNSO=4D&-<2$P#b3*G_i=0(_g0s;GlrRxjrE_yd9r&oF?)G#isgyUnumx$LlBYlgh z3u2F4KOio&Zv<9QGSLMo1S6`O<|#+UE@dXb=kyVY<%=PAhel0vu8R6v-FVZYTQ$@C z=ZP&5085}Dj9?)8)XDtg1q(53Kf$NyTEo>SHaW9&%%1Ub)U;>M2h>B%sY zYdO*|>`NV5ez?|8`ka%U?k#+*0w5d~J3!a5pTAJTBmSVeV$ne-^oCdtb@%Ee^-S@On=WW zh4-eHv|2I3aWt6_sU2i+Vu1Z(A_4=sY_5;nmS%3-_00JL9h;6It_6$OSxg^ee5bi8Ecn;KKin#z|H~WcL@Wx2D{Q>Nx zd}OFIW6;0*@`NcOrRO*|pDFbWvWcxaJFHt&9h;?$s&22MOp6R8Tljlj2H}`P>G;Ei zXX*A;{lPjudSj_wQqbKmx-a2fRFj;UzQ#ZwChYgCA6dWX(lmG+(`}h}K6L^eF+qOg zp9Jlt)CPt{`(AYPCgs2B9A|9yg`ZBT!UlOG{r*%>scm!JHS-<7OTYahxP2)Gr_R8)q#bfT z5sq9RjxwkW;hfhT^Wa>8g#LJO%G}zOL0pTR+#^`Tx-5@(t?l=%_Y2lgMo>*^Pu&9w zYQw#e%)bNF3QXNZ9mZ(9S9Q0SSllBvfSp$ z`W%)q3!_NbmTA-x$vJgI=NweueXjP&=a~5J>ZAd1A1f$q-FD-vxe|w} z5#>wYc9`wER+PbNM%!pq%*ajN--5YCM8Yo<`J;-oEQa!>(-o4j!W|=>bVX8oSrtH~lOar;ODqFP;DQj214Ai}Qd*vvdVP%-7J>JOALcsaeK z%v^h1i+D4ezqDQLn3r2EQG8`V7G0IEoMlTbf0FTk ziWv!foMKwS@~h-D@Jey#nc~d%JZqc~zVkns-CH)MhJOzM0z$z>$_?Lr#NNpjQ=Go@ zKaqUGPKSd2MOGWo#h{#en*%$-#Faf$S(x_FBR6U?4Z}X`MH}Uoyxeieorz92amKO2 zV-zaer5tO^x;uM1Cc&F*D!i{hIyXOa57qDA4AR4$8Rl}fPc-s-U!&$?fQ>(G+E_I8 z7UF~KH4jXXbPN#Zd7{y}41hV>VoP%G?Q${oDkis3tknDw1lE%(-chsi7z!=>Z0zX6 zrvA+iT}_|DSG3M)=N@vY;zIr0R#JTiBIbMc!~bN)EZCzDcVn8{gfqt7VC|eicP&z) z<*IQbUY=DI0zgXja*AQ)uyvU;=K;r75$YMc4^1~VOISDg5u;!4SdV>v=HIR8H;5rU zn((aA1A#u!xPNy&-cFPj<158qHqvmUX|y;;jkP(pbJWoiOzI{T}x9UzT{Or&{~(54(E438On}_cvMS>G0>yyQL%g%^G#mpc_2@?1R@6{ zbaDS1?o{Fdv48ymgli;|cBqd~W=$830sK+8D(g&V^rubp)0v&gUyDd&O4;R4j+xil z{CakAfX5S9_WH(FTW|*ZSMS?P8KIP>Ot@Y{s&O(z=0gRG&;l0akHO7I3Xm_PM4vq2 zX;g#VdcaM`{%{yW1bvAa&_=Rhcy^kU_3-t{^Y`j-w&Y3GEzv`xSCc~MCe$-`H5`_7 zvCJrbBC7Qj9;z&_Sm5+5JJj}=YiG}SIY+-hp4g7q4wF?mk?Zwu(;3Vyt$4@lk<2;5 zh1a8AnVxRE_$4ntLGl`|^Y}JY`9`6_23nB5SPb_cV0VxO9GaV4`D3YafHJ0-HiH(Z zx_A6OXjD{lSQzd|HQ}MH?Dhy24o&E^^_pvTY$1%F&L4OLtXD-WG)AqBu?t1u!PS2sk`IBMJZJpgHi#k zmX{dAyfi}h8;3$4X};{g)7+;I?2Fn;WJSlMp$`86SpSse(jq*FoYkMp;xl&a8z4!2 zGM4R9{Q3PqAg%w+tPRugNilB%LPMi~Z8O!nK4>L=bqxnFz1Zg_2IYzr&_vt#;h^xO zuC+oJY<}DRrQS~BMcNe*I8xr+vr;q_Ep+(0)v&|4w!Sbc5C1CZHS>XN5;Mx=>=PJN@M-SOz-3JFlF@@M)8tir z$Bpji7?v;aH#2M8bXfQlOvEiNPXaf@Y#o6OQ~!dD{hXE4zPA{!p02e&QP}@CeWBPz zJM$`58MT(uNg!@2w`Bi;yO787!^o{+v0HgSz+d)d4VkaMj6d%27@ysn!J(9P1Hnp$q0i5^bgDi&ig>_dr{nU`_Ge?Ro%Yt)Fg6 z_g#>0%UtD(O&8@}@!$&-o!f8O$8QCf4$A>w;3hY5CZko!I>CDzU$^yTel||=N95~K zW}qpj$Sx}0%GUBHFC{7PQkZbvzE(N1LaLy%*K&Dqp)WgODKBOZC}@mJ>|!@_Qe*mw z$t{-CbCm6k^FRqiBX^6ZX)9uDN&NUqF(lErf+7%1BHYM_8j7TUTP^?}l$C42jk>0% zzpi}ezA&;s%lh-P$c;w0D!rQch?uW(1=iudIfHMglXU^XI6Uq1-7)k*)AODYJ*6Cg#QDSFn^Jq^sXRRM3(LAX2nS~@x(vXQkx$U{z+1=knPEy*4&p899u2py=$geHB?-`=_ zV+x7eEw@?Y98%n0AA8*36=uIBJ%NcAL9l4}&W_f9eSrb2Mj+T#h> z88B9NQfMB`cA)FU1Ak)A`)iH{a8?u+JX2*^8-t;5#-Ico2#X6C~RnVJP$L86kY1)!I^23=*QJ*~awPDaFDd)41Ax=2#1 zyqUbrsDz+2OR-qCiM&1P#hYP#pWi{j^XW>@yPGoK#KbMi=(3fPo>ge-RO^vk)v4@ja zoI=!5&}j_TFx+xC^`fM{q@oDxlhUhFS-j`}*N^?bG-1RY5?y?)A6K((QXtqxHXBatfwDY@ZR1!S zOa){XKWa@54Henhm2AONvhAmq=b5q7$>rt$7&X4NwVYdN1+2?{vuMy$ggo!cBMten z&`a~%8|ecoFm&k-?BgLHx=}i%asQkg=)T5BqBwhT7aWkP>pS_E(dcZ==;jmNQqVaR zeML`GgP#vh2G_8B=u`x(iAFR)1YpyiArp3fAdVGt&4bxD1ZHm>_^1{2R`%HJwn@^` z3VU>fmsX#=)5}kqjkmj@`G^9b45!BErb1glZg`VFABS<@ZawX=)B91xD(!QY4A$S* zNmhicc1lX>v<`mM{^r74v3h@BMFF7D#8)<${ADS=Guv_zshvtaLL7S|b>EO2c&e1h zc-qXn@utf~1n}7i2js-ajUP6jVrcMCf(|rP(U1%)R%KYhw@&sC^Aqxdekg^bhtftn z4N$GUzfmf6q)R5F6emTO2Vo2fYSK%qW730Rz6?8=%Q#e#M&zVk&@&A2AHg~FESK*v z-6%mC{INzdebKVUQC|2`TvVsE#`O1rT%QoB$?A!F1M+dovhcBmqszS6E6i^A%RKF` z?rO`p?3huNhwVIcK|0oxU@(i!TT7716GF{mJR3g(*wh3&`tPA`Q%lR8TSobEot7C4 zID91W$q@{j41H7wqpylV4njk>Vw8dLS8XfrGor)2l0%&dR~8}+SGu2hjm|pj*JL@J zfeciQAW9j|j9`LOCQqq5JzO_Wxeb270=wKGdJBVYRl4ORCW>yzw%Cfq%ur`+sl)6t ze}Rqc>ZF`O;3zqwP-kxYL$z92&-WqK+eotXycQ~X}w|lR_IUdGX+#%Xu}x`R1yqTm3asWtPc1_1u|&} z8|wV^|oN{=AD7ri2-DrJAnZd z_{Tmh>pD5?)5E&d{5#&Cv(}l9idWD5QnUM4fbI{!_I5o~ih<3oXG04M1V^R2*+RIY zv|vmo?0wm@T7VfwM^v7>u@4g=R1$Xne7 z{Yd~dVV=O8E@keGP=i>ST|FrLyHc1R`~G^|R(@$S#Ye1{cG%A2O|2@a87QqbW-=9y|oGSp`D{IWiK6cv%H^(7zozGd!uvmMpOyD z(?X{l?vaR<@v**$cSt^(=}gW1k5()ERKH;NqE0c*x^m7NQ{dNA zCK_V7cR_ZdqT{c}7l~7~TD^m!j&TWPApJI zKo*L?k(HY-?@>Nq7KY=_LnVP@2POd1=;N=#_?V))AS*CTou4CJp;n@wZWw|Zmj0`u zZlaPHA5O%eeUlQoiME?}+VQ2o(K%|}>S_%7TYy!`rV5MpqGihqSEF9ZD}O$yt~L%} zCOS6W<%`=j%ZIM4zw{hx^E$pUa#-4i+2WZYZFk+wm%@YOqL#~1V)>{E^PcwLt|GHUE3DTwTk2cR z+GegaJ30j0hn6!v8m7NZqCbut*KVt)9wW!Gq=PR{b-MytRPmLG%Xvit!>6)jy< z(h}KD-uVJ__NRCJ&P}pEygcY{%DWv64)l>}J@zKAywch&KPCuv5xB_B<^nZ7o5PJ*MPd^9b9?ahWhlI&RS^ zk6Kd;JvT|gnG$Hp6PgxXa^7Ca-!&%@fyC0d-!;;VF zSdg;+Dp=ox6^xWa^)dI-29|uDzWM@@@u8|?oH2sl1KE8?5hRDw&dwDqsRMOEoz}@* zqlQV47^w5>0C=t&R#pxTMk!BO6pU>Y&(kJ}Fq74cSd+?!9cU9603evypW4?~-h~X6 zTCcJAXL9XcTR&2t+R?UODZl^D`^$#yWTfwwV&IFBK2QTagay{qQF~jyi1qTN10Jr+ ztne!Nb~?ZpscmFS^(GT5pbs)B&!B50UMybYl{1iaQ2_8?lh#}*^9l_-v= zS>R2G%gdw%UNcqtu3Q@lNs$WUVX6Niy3L(DYj9ZRp7U_E+EEq8fCY!EM0l4o$2?MD zDW?>+uZ{D!u;@wO=l3yL1s9H12-znqV=2*Yen8HoUnnIm7&z@YHl>b!&kyq7ZDP*o z9|08fd6}4Rg2wx}acxPa@8(~U-0wjqtR+LpY-%f8$}MjEDna&U*^z_5hdB{_J}Y<8=eOprG4GW})=e%9=e28BKPoeMRYFJH z<8x{hZ|~bLLhZ0Lr9W~dx8A(81AC$gRkvqLYM=UbV|cA95G`zS?!h?YO>MplriTFRvI-b82&*uIR zPI`OYSN@5(E7yY7v*mZwT;4)wYUJ!wS9Yo)ee?&ia?R(&;;nPmogyS3yZmkoqzw;1 zYChi2J6U|=^xJ`~Qrj}q=ZT;6=vUcR;39VPzjZxIMGTvpOSg?evC~JaR|qQh#4*vK z1qJol!1r(qX5HSsm+m{V35%R(>T8yQf%?4oKfGMYGk!sSH9&CCReE?jD5q;_QS;c9NrAo;PmK?`AzOl&%Iq;dS@>^>F+#t9Q zA!jb!^XdVkB1dw)k~iW4{kHBf)dVH79iu;C1Enqr!7#Rf+LF5>*3aO*{{ZvHuev@W zd-1ZN)lUa5_f72DC7*mPz)Hd8W(}}iSXuwl(Kb$6XG^7atpL%Z2zupQ^tE=0ab))i z@opma`Pi^Q-Gs{9q-Qnzo)9hbV; zT{JSi_9`W<3sCF`*CXn?H9WQD@nYgH>JBuy?ysgD3?^Ixx<*P}fRFWZ?1H}#u~@&U zQb)O1RxR*I2IErl!(?KN+z!vZLnEuGpp$v^Ov#YSFA9yB(yb4($@91kH%~i-CIh$1 zf@zpX1Qz^umNhfRyq1drYCu)OK5=h~)#99fDV&80KfD{2Y(%L^2@veC^sQ(~Js+mpPKjPGzDHK)bSF8tAX zR6YiMSV@>iK{7WC5YAHG+ahjsV@3c|9k=)b&H_bUmfGmPRhMl|> zygb!WkjZ5I{u7Z#0zl6Gt!QYKUHfent(dB-^fT2Ff_U4UJ!mz3N?8%mIHWWDC5^dw-p#4j5E zy5qY*nJW!|`2N{2W>PGyVfa zFzmnh$bgV;sR?rtf( zjQ%2)>P2+DT9HAhb(i{c)3);Uv{Z=wda9;>G1S?#GHT!N?3wC4 zCWG%*y5D0th-mgr@aRWzT^reLhR8Y_dqlS$T-SfI717Z?Vt7rwWAq=uPySB@m4N7& zmj8QbJlRF1*x<=8uRSKOqY^P>qIRbl$GHdnDf+z6IbHNibqHnzQd?NP)#F$db|L*4 z>UIx?8#p*2tTa>L1Vj_tf_ai*kFOPOJM54r<# z_LMPC`g*Tu0z_!}a`u@sBO?EK`!a`?*qH+0Uf3*=giG}Gx6PQUb`W;;v>WbxcB z6?yLH$kc&>2@My;9LqPJ2o@~9lCXLb#kb1y)}IvZlXtbhuct_!bydMmrCIh}CMsT0 zeJLj#$%G1&3A!`#yYBobAiku#wtpuuaa1v^(UfYea0)&DH;e7x25aGFkbOzWf1D=Z zRHHu^H+75gt%3JAKeJvnoBL?skKJW>)0kK)VITJ`Bku40cB@E2gpUhG!FEu(iPb~U z5v@20GWQ>v>4N^yuzXS6Er@5UtEbb{^y|`n8k%PXIWBkLwO$%Nc2}()(-T|l@f>~q zlGCXcW~06u1+S6&yH{Mf=p4RWy)LHj{DV)!sgC){i7O;Dri00*LEPflVOFOrQUA%} ziW+g^ha~FArQe@q{HMgV6rHS^^AQnT79C&WzH^-D3;Mtne-rG!aw;wz+qri49Tgj* zMDi~P{t)i<;g2eNhObBP))P=_b}>-=s)?Kdli2O*fRZbLANjA$ecr9|^g4;1zkrY6 z21ao@utiRVdzZMYV&0(Z#Esv##11$H`x>>4D_z(0MXo`NQQM0~sV1U}tBvflu+SOv zqs-MuJuvJEU&lJv+qJ-l$C@lo7s0ATIkvf_3*ekx5#msU}oSo_LxQ##|XyV26H zfX4?=Y+o-+v2Xh2$e$`0pRarP0+(5)|AibXC;Ob0~H)5|+6(NA9_j1`)(Hmr50vIP+YG>6wE)Ujd3Ox~_asMQ=+J5z~o4|uvkZ>6&yOq^a6aK*;$w;3*Xg}I{; z`&cA>H$yuf^nkZ}1XPNtg}Gs??l}Iqq@y?;L>jwnbPNUDxK&6IK52Fn*7YRUL0)G~ZC#_n*v zxZeli*hY=00D#zwR#JQCTjemIx;?vz5H(d0+-DY;9GYP+u;IL}iEqEyR|b%6O4&0I zb1=qB=<=3~J>8!qFt=qS8JAEmaOuSMwvh4NP6jYn4Jel?f$a?={jnBsy^bfth@(Xa zRwc9>}yXzhZ6k^%)m$r~}$-wpB%?xrzCEtcBk^ zt3UJOx5&&dJxG#x=H(A?j=wwx6aBzBEIp}{kK8FZd3k#WF7w@HtNxh+^Xh$&>&x#7 zRWt>!4pbFzVc}}EeAqUF@)^MyH ziJ=w0O$%JONE(#*T_)xni!jRsUF+Hrnm=O<;lR{$uD?uSymjA=SGiU7gxiC{k~EKRs#)09pG9eV!7}KAiSq3u_vIe~U z6qhdSl@;d30g`9GrizaB-?)o*Z97<;B+q|NY_0iuSZr40C@(h;dYGOEo8o@AWhMw?5cw-jty~nN-&|93XU%sUDk3@=?#(c%`L=>p zS}iD&Q;6i;2kT@HMVYfqF2tcceYhADctc>C}9ccJjM1iN_*-_m^f3DUFJ6AM6KMVGQik+ zS>5XRto}DIs2TT?+w(`A2LHrlFok*SFH);mrl)6f=k-*f`+hM;E{ywH=~DzArQV}I zC6i7Jj}v~*Zks(YWCgp)RW@GB9{{D+Y=TE0l};-S}JK>U^d*K~Bu=Nr%kZ-}Y(fK!?`BIX|%(3M!N zR>C5OGZ~zyOo_j?ijJ7|ufSY9!EAXQ`rl|kbtjSq{{bd*=tWh{P%jy=Y~PKsEN*=Y zPBLR_2PY85&LE4IgIINU3_IWilvx5jkdMAd_a0tI9j>Hw8<1_A)N}JTLi74PsLLc| ztiX;h)B54ZJ_Zv);@vnRJ!-#x%(Zp{unYnO87G3()p$YU?_p^%0vr3UvcE3B9&;Sy zUZJC7E&S@@>P=X?;d4n9o&4-i+WkO=njV^RJfS0!S;T!~xw>s0?L zg|16}X35%bW%})|H6qfEKKH!fxu^E-bCmht9(4a1mG=92Hyra_D{fUT>p_ zX&;Rv@eY2GrkA#ENRM%mFwL5}A@!0HrfnKAmAndNn}0y3w)I8BSVP8%5bppeYY0^rcVEvJ)Fj}NsQ#2tQuq2WT+H~&OU6|$zM7s zx||HGP{hzFmkFnU%Vhx{$Gl6Vr_a-ttG}PqV3rqTO1(n!H5wnDEW;xrgQC{UT1Qo9 z%CK-h$P^AF#w(97xm%z?w#&uQ_59)X7Li^Pl8DIbq*!ewckgV_Uh>?CmJF{hxU9j8 zuTuW|&l)7X#K{c=xyvg#U0#~t6>j%tX{Er~9e=4J9N&gqv45C5%kKekvDNh3c@A`= z_O{14eiksBO%;Cmo%SD<{U>Ig4`eWgJTR*Axuv@B^tlFyjR8fi_{iim5<4kYc33#7 zzoP$uLTr~lrE>e?3%}iNEY4HL;a|hf2 z3@z~{A)CaFh?tcE-LtwXN0}wwOixb;7>9KfRz@h{#=A5u%X7w_APzjWrNMfW)YAZc zm^JVEBKh3zIkxY%z^lX;yqww=)Oq2a(PrmOEWG zb%Vr$i+!9gH#kF2Tv|S3fro`}8GOQ@JLuv2OzNIp)B@d5LBDl*d8|83cM_g1ni!4j zX#OVm^m?6o9IzG{G0RmDasZ7%C@w zsX+D=x^TVD28xG{xX!-1pc$WBb zX$+4&ZnbcNN@3Duu8Av0Nj0Uu|#v=E_S9NtBib zycH(Z(e+UYH+|j#7Z*T0-l8@5J+mY6xRLJLDc@8>z!OrU)sjom4T-Ti{+7~GBKQUy zU5XDU-+Kp23h>jkECOF(_lqL4=wT&n#^RtM?pHv9Prv&x#8jF6tJCTcHHFo=)y4&j=zQO?_*33)Gc8OCme2 z;o;WD5_~-xjX8DskgV2G{$_bhdZY1eKlT`tGp!VkD!3Y3p#5eAmONXr+pT9?dDxGNsw)ns&S#1Oa_qJLTO#iK73(YqxP)hkZ-q) zs>cMUS3WTZmbN!)zjL1hy*gz6W+G<;fU2EX1ni&cD|f20v+t%&j1`~k@#D{A8%lcJB*FC$-4iCc|>1xEG{ zS4y?`9fZg_*8h$9?xoX>~bfvgX! z1Kq~bYaTR2Dom&(u?4^lb;t>XMD)N+2>9xs5RlA0eGE2gooN?%1MM0s56s57 zj+Kkd9LApEVMAYR<=B!~T8886TC!Z$V?iT-#VKZBwG+rJh=U<=4jvHiIwQ-Dr%KrD z22}*(h%n7s*3WHLAXUb)*{K^$&P+s$di+F#`E}JDafiDZMedAdO;iB#tggAvXDtIb zINa!2^fU&i%wrdE*`@97%9l@dLX89M{0Khk12oy1c5~ zD!|jzRzw)Ea*Or>D~c_#&?wNTpATmBEH}7ZgB0lQlCOwP)B% zn-(ykQ+<{zo7~hRqdzd+K)a%RnWg91UEW~Xv^mRssh2Ym8P(!@dL1&!n-;!83(CTqFy zxz6$>oI24Pk8UHPJz`ewl5DQfI^Ps7StmlF!;3Hobl?H{ir|yMg#xcHO2<>pvUXbN zp~^};cN4#59}eoyCU%CWc7~Ts9+~!TZ)XL|@W{8DydKsUZqzViNJbrdyKV;2qA*hB zQs%?fClGSfv_-YO1NLS#3uW_;oNUwt}FVQEm)02<#Sj1<8z2LM5i({nsF z?q=&6+e&b1$+0}+9JeAHs%d=2NO+$o0VdpAO#;$)wtH@;Zngtx61r)ZO(_M(KPvG{ z`&U{Q7VlxYS8lE)qp=eF_-5&U2&+nWUjIUqrDV-LO%6>r%kX7ye)xG8pTU6xboahD zOi8OpxRh}H`-B;|2*21$#kAmfrJ}J#^;JK9PA9E|0z@}3boSF%X2R|k?OEfo>uYkx zY^sDEI z*uPq32EFr+nb71Lnns|;*5icjJ&XvY(V@6xTzOEw_B zN+`VmEO?wRJt9kF*Z^okHT5xmEn}5*AW=6Pvm`BQF+hVRvC4ywLlIhJGUd{6z zsgeZ0e8yD=IZV!Y8br=r%g*m@;knYs_p_)IPcxP+@55E~$fqng(Xtn^&85_L+?3#M zlrK|DEB{GQ2AT7U)yjZ;BG*~XPx%)uuKUNkOG=dP$w3$T-+84_RlP)h&#^;b+*8)j zTqTMe9u*P!iJ0f@x1nM`S3G<~`3DJNc!TzO_YE5{Cu#1E7dU0_J6w|0dPW3nSZ)0D zWA)qc+R^_3W5`Jv=X{9r93Mm$GTX;4q+tc5q`OOyOoIFV!?|R(Jy-JnPH(prmrV&;MLcM1HPDmM8 z@$Hs*3l))LAptXO`f+3QU8%zpKyFkbv)Wr8Xwjstj$Mvdv#;T)UhR0I2soWj*_*g% z`|o9C&9w#}r`?F#s;3BKH{2<5U5L}k4CJq;POvZ+!aMzeAy^egcE*x8c@(j5R>2XM z!wLW3ueV4{cdm(gbRGYmC`eon=8}DE7Kmv0+n6M)dEZcGe14sg(iWOQ@)8t2@FVW0 zIDhxqhItL9^twD01^1bnHa=(aTGCuha+Hp~EPtU&UZ7DrKO_<#%Zb?Wm%= zxhGxvFxIs=W45XGCiBrCt{?NhTiwT6j66vVr!^$B#Z*(Noeq#xo=|pyn?FYRDd3GH z6@koE)eq9hu0Vpn)vfwDj>hEA#&93j9}6x`v_r~F0LN={n9N0UZjg`)3W!ZF(#1qJ zN>LtE!q6;Ov(&#^J4#v00rMP&gU1P+qg`#lF!mS z2}L}G0%-VuH<|}V1B|>P<_ratMD1+Zp5|>8mya8n?=S`wGY4QbL?!4U`amYXAzz%e zp_0&cOP^OSC!SLLk{>q=uK$UO`WG=ZOT9GA%7$im)ego`(=CU;&@D-^d>d|PjqOq8 z#$Rhl8mtw0LT|x^>yv8$h!&C?GJFcSysz*uB5~l{vVctTd`(&^Zmjd>Im3$KmG&_( zU=ng$wB0RJ#iyv1*;IoKJ1213`;@^4NJG-9n(K*Ehr^u5#v0l_n^ zSY?9ok4+fP#J-XrULb+T7puXS^)lMow6BQj8H3;rTtj-8 zcWBT5tJnhkj|}_&>r*AtXJEKBc^^)WC*Dq5WN=d4{`Qrn!RM$ap_Ul_Td4@GUP(C# zs*;8vW4C@ZbC|D=avJ^FEs?2g%Xd|%LR7zgeR3!t@ppoBla?~6I96(QlwE!ME!|m2 z)sTn;wsylWtUhrudlQLvAc%z;no_GU)O4)piMXoNUqKP#dk21DLw*XxVe&rmXR=Lz z&AM*_1~5oW&P3mVERdCLLhxuKl6^fv5yfUD-zHv_lv(EAHU0HP)|G86W5oWWxCyQ^ z+H^kL0X(vAP)`*vMe40yEk)|t`UU=~9RU))qOS{TocoQa3%ZWH$Y^aAAyWk}*%~EY zt0bTn(UA~WUuMj*(REH@O>;*<(U~HPZrtGG=kUEu_v4kH4Ptn+zFl~nqpHEeSFsy> zqBtCNSO9>X+$uNH(0uECuwjdDhS4lCSF;-O+PjwneWnWuUC%S&axo|ODp!VguD<1s zi%EBouq<)cj(af8v%5lg9E~|v{aM$|VTsH8 zI+ai3iH1-r5|crZ|3)+{jdf`3_D&1Fr>$jSbN#M6cU()7utcEa??Hd3AB{|w@Ss;v zb$O-c{o%JgDWNTA2M>|=hLWhX4M>Nl&GrFN6eZ*6s=!NI!w<`KshNo|X@pf9qrIJ=N z^_`c!-I8tU+dlIO-*vN)2PdG#a|BCc`SJxWrA?j!Yh7mkl^aIx)a`zDRWJ0H#ST}~6Q0s3xWq5~m1to` zwq98w*5;}C#mX?yAcz}uw%$i2TV3OvN%QfbrkZqcl5!YeEQ8A`JoqPo-aVgZd2pve zjKq5$*}09IHTrq_sW6O#|`)o zZS5tTYRdV~J9k4i1-trPWAhwNKZU6sj2$+(F>5-o!(;&cZa9m(Bc`>ZrLC$`W4BVV zN@SyP5$6@L!8E(mK4+`3yuQFh=v!VZo zwzCL`vVX%h4bmOb%+MhvGIR-ybT=XloiivYEg&@v-QC>{0unRE{J=uUt^xZ zV0368X_dLmk0UypTzvpZxXOpwvd{%f;qx8RcDfH;oT%7|7~wAwzKp+pG!H0G28e%@ z@4Mi)5YYYJT!x|HYL+(6s!pVzUu1|y8v_d(;|gwc{(O~0w+PNhvP+~C*_6oR{I_@X zb{z?12qr53yc(lKJz>P(I}r@B_{nt$bhgyak+z7Zkb8_pCbzU0_aR}x?V8pGkPGw7 z8}3-CRsX8}bT*h|_*yO4mQ4itnGnAO%XZz0!hFGQt?^A6R?132DS5)gSg*`FuIzVd zXWdO|AY(yBV{2uqmk%E?18x|7LBK+KpbH~55N&CY#_Z#uoKd#Mm4VCr0 z6YM@>yk7k&vQ=is@`?6;`Pv@>)>k!q3MklfBxv+h=(c(}ZeqH@LjMvK;C}!4uBwoZ zk}t7aqZ^GdrrGz}M;)U4LY0N|9gM}!!L%Rg+jr}ZOlPe5rQTgrgyQkx9pXmU;^P%#k0!YtSoHLZ zBT-`kjwE?lUb4x|e#Y;fzKsWiHGeZt3gsQKjgf@q;DLsJJoE_wFdb-J-xeVL@9V!~ zN@2HH5F-C?frpmmyd<0T@JtI*@6%WuTj zImPYy?ggAM4by5DIdBG}6sZ&WE>@UyH2m#X569T5aPrU0{24yDphs6^nV-emAJJaM z{5x^UQJ4&8QV@zoL}T@KO0gkQ{y0FZgkn9SQ#ko=(H@~N!_8|7o`(j>-qaU#D5=5y zORR()pVEhSVJ)!r2o!1+Mneqwt_F0y77Mg&UW<|%^-7V$-MB^8J)CHS3PPa4W5k-c z?+0cZT{nkTM@&kDH!gy*ykwmG!uUE?1*6f>JfwRcZ}F90n!FGXJDSV+Y-lBgF zCFuCXE zw-(3bGY&G8Y%7QT^yzh$PZgtpai&8(Hr9S(BrZH#UvP|^g#g5}W}w;!*en)M(!i(s zcr0TMc{EJ$BfsZ}*;kz#9dNTNcBs-je+MP=1)x{(jnz5 zX|OlxdEBd*W%4HpOGkPvQKn)EZZvi>Ze^IiD`~1V1~6dyS58}m<^n$;8(04}cw!HQ z8*WdG_5Wyobu(hw1umAR)GL{4%7KRLI#cBQUv^o0qQ6&WnAe$oYp}=>ise0aW?<1@0rBU(M?J9G zQFiTem$o)OMT{cNp#j{D7F`ulwsCnNazQli8`=v4JHkObQ7dALc=`X(j*G*h<_N>Y z_v+-AQF1b=4cyt?T8t=bB_KJ-+@AkJH|)*}@pS}ZF0dOa^X=wu7~KK^u?LP03LK|B zLE$$!Mf*Yj(__Nh!C|D3vcvc^(1$}ty@cu22CdqCXDjO#;XT7@~af&kw+vN5q#M=8M#IFyqg{5|33P%-pU(G959FS zA-krA59*wu52(^^m9^iM@{x^%lHwOZGPEwAxSHe)uHTR1&1?_2{>)1!dDqr9-kK{< zj<;?TF+wQ${8|_LRqy`j`#?F=TYDj@Sk_iZNjQD!qtZFaN_3E4s%}=+n z{=&|vF%Fpa!JOj9`O%rqgsx;7$5DD+Z3Za$@hA*UtrniO%*rG zixyZ!mmh|0=J*r5CA9tO+DFA!)MhdBjR1#s7+cNq(&%2*9^1_fv|SHX^T4;2m(;2e zQ?|3o)q;XK)aCpPUC(uY2{KezHY2jZNh6GoKhV@LkHEA~2Y4rX%e6ZbLPF9Nw2*L0 zYW&04-NguLyqh78*Y)UEDJRh^;s_xI{EsU5Za zMoET*^T_EKcy3GEaLl4$SrM8VW}J8lpHN zs6jINikAc0^HER?@<7xbvZng7;FG;clm;+V)w2BW?s^_&AfnE?r?11Qn* z+)hgy-_=T}OBUA3SP6prCjH{WK4yG*)%eIYom@r1IN7pR!9t9>#VKB|H2T=jHSew) zA!kbF21$dAO@$O15VVk}cYPW?-KaE>82807OSgU{sUpKk7Y=KW^Uwq%Idf5vN%ovh zq({TCoooVs0RgqTYxsKit=BGR8|p!>-c8)z>B&)KX$l^UR)w!e(+F*d$1#{91j4_Wg!bY+}(pJg=1JEjpYSJ^en?oT~Tg1Y}*`P&8auTPe* zO?vgZx2)m6N@edrX0QxfYH8+q-xT}S?Rx1|n4zr17xoz;sk%6rmsccNA@d6c5*C%p zKOJ)AohH%OSR~7=ecxZmO8eozb>*1PIk0)kHl7%^7(Wns(>BS;sD;TiPuQAh$;&%g z+23OQbWOf{pQG@GneE>IX2rAS(?2_aXXQ4FBKs5oT5pAT@l5%k!t8?@mWa-=kyXi z59ytaI&|_DU%Q!n|CiOmZO8cBpq7b5VD;xH=V|f3F>_){LZ*OUw1ci6jkP>I-T5A? zM1WrV{bb7W4|l}vEY`WQ&5MjOpz2)5b4K+ED<+Hjjj&)G%3WNJ@)*glH-^zfh8qlH zKOMRSj3!3MDo20zV$`;D_c}GlxqZ-({t@Ed@g+@|YAuaXl~3YP8}r)?6c|h-rdz|? zs?M-Y`JpNP+u&^pT;pvjbpVgnrqU95cqFY$z5e^ONGZteo}SpTYs*1Lo7c6)IY~~+ zde8#PEyGPQb!J{7sioNq!}(zX^`PB7gvqxqEz6aBK3Ug)`Rv1~Ly-9CJQvIhqCvJ; z*uU6ZEk^NDUn7x5{$bRQJ+t$-qfByz!o6_W%|Kk{lGg=YY`bIg{1hL#@2qWQyW|r9 zc?zCOz1y1Y5qs<_VcN~=Uq~+y@!WU-=#6I={NImYrE4Hr!wj8W=liDGKO+RRTN$(g zn~}lT3Q21$q55;#)dE&67U8A}E5MLE#OHwnkZhcpPi;sisf8ID*~<-EL~xFbugJ_a zEpDcETmJ}6>ooIn7sxX` zhG0K^C>e*YcrlcfJ{`Imh0*kaN!0R-%U=0op!7)%Q^5H~w}0&Z&R2tyO~@g_ zkyuUK)-oNXQj=Fj)DBljyaE-@pQ~lY3TtU8p>g?~ZYs1-JUy9NS%}FN9cXrLux`_` zYm(WjaNtux@^JUR0j$f_!k!Li$ofxi_fDZu5ySm%D7UA;^1@XyQ%^i|oM&d-?nXP` z->VmxFAeu$3iQqH7F=jE?w8^rnyp%%sr#Q*#jc)1h`5LbV;12o=iqZD_X z0Cc$vcuES_Ee~vQG=?T9pO1HyBOi`5@SKX?&ETLnU6V%Q{`~iM&iwne zD;N?`1ZB7aO_8*SGhjjVtygdvn>H(}68Fhfm`vDFI=`dy050oelMW>-1|js_2O9uZ zZU%C;&@3PlcO1VeDI^j7DBD}Q!gXd=IW00l^MKtKGxqvsk|zYv5C`zch7hAR!tZBu zf?M2b?mQ0nS66V8a#tSb)hN^(1^N2{&P>V@_TRYSl>H1&U=`d6he+AyPDO7SMRoYx zUI9$}Uw|f;7q?K8rhSMlqEaO|3qto#QzoKt(|}r(i%ex=O!WaZ1=A|gU{E$zqamu_ zW*kPtmG1f|o1%orAu3!YaJs(~l6P6sMN`3Qpo%AiKnJq5g0Vi{mjjs8qgXgzPZN@` z#Q(0MA+HXhR&r<8TuhXaldhCpjLYUrpafxu-ttHbDarvhBR_o-dT{dch2$*XGEe1r znCdlmdt+<~5VMjK6PORb(?Fgi{>}H9>OSP78JbkpB!7bt&_yH*KRYv#pT_bIfso}hUght{9 zm{hR&mrN~6-+PP|nwZFdhrNo6`gn}aV^lsN`p|-7D7gZ@yGJS7kdEZgA zrUmZ!yrFyEJ%>eoJ$?ZAro3imKYQ4CJ8B)g6@OVGX}X=_4S`o_XY_=rS~YP@Fqk=^tY2m*3CrEB&r=~Z&XbcQa`2c$2P;xCVk1Y~_(DV=)7PH&I>7rd04S7{OYhcYRwPp8U8hM>P+m{cliSPL73 zH}|v3F%(Mq>PHwpPM5IdB%}`y?+LAUA4&z!FVGo=zp3?D!2~uO{n2;wD8QLWxW=4c z=rLiB47ZI|VN=2%9457uR-HDjWVC#rvf$xIHuv*)RfCk^L!s(3ba=}LS$|W0CY2J~ zo&lS_3Y%w&hzJh1b}fC&g(uS;Ut>EN5x34Y{>EmcMwu{%4MEK=>qD$z_wLkWx$Wnn z_=)kt>d0)?d>w#<6U&CmT@yIi=h+cFWblazEQP8EHzNpSJfNZI($~gwe4{vsx@HEVjqr6&o%sEsGB3m;V(>%j9YekxX;CoHmDQ_#BXr#nx7$@j)N- z2&)d`g_S=cZ0AWbv=)o6KkopAI^u{aiE-QA1D#UBt9S$syztns&0YK5$`)%^+Gw*Vz1t!ufc|;ZZLVa)8xz)=ZuX4#Q(HxKQ?tG@Us$;nx z;NhSe{cEbCJ;q&l&Rfld3dP``vKu#?CHh>}$wl@o^OKE*2@rHX;$l+tlUPrKv+t?s zty^uNBg3!HyC6XJf<_6@q6o!7noaIBhaMYeEVIOYWg*u`A+P`<;Bjo=E@OpT zPSyu09L#74NVqV2w|6!v@Dl9smPN)9OaF+;)i}Oe;)%Ph$Yo7MPFxfuW7?RzT;7t! zbFUi8g5ToH07B%bXut zRT#BDoI7UT^Ii6yck%n4kt0#-pv8ou*FOq z&)B&XuwKY)W+X|HYVn{hj{Ff=lf_MRRfCZ0>@7PBzNH%5xCS9Cgo@;bXxb)B zrxwL}pp~;Fn;{2C2R(CB>^2lt0MsnyW0_|$oeiVx_Ry~vFA!Gl+|8tZ+AGxK5yPmC z&rLtYNQ8n%)o_4=d6zN)17_HnLclODaz9wjDp&QULivcX*m}YkvXf4#<*@~+)l|Ky zf>tUU*)WcdPV+bw(I;^L8`(_qEyf-s8`j|1+KF=eY60gtyb506lr%U@jbaM=E zKz%@wwi5LTC}WYowzv)mTb9Q5ul@2$W|S;m>uS~n>+R|aRVxd#BqU%rmg|@FNk&&w zV`d`fHVT_-631F4&@y_x6F(aNyTO9tdX$p(@6V~o3$9L zEV5sHL&aqbZ8JVE;0U6vF>9K1rXGt7^DrFGvA5+wU-C|TKT13~=aS`93tJ~k1Rd4@ z^o2n(6AYzrO2<>YH8-$5BDT`L&Adeh1Z+XPx>JYNBrDiCxD0zIcBuZj2T9G_F`8_7 zxD~*+CR?|MR{ld%4r2+|$_n+$zTe3KzU+4QS)8O<#c-+rYM#?D6iOKE_IDYOv>6f# zhBd6LF$#;vmb;}R?eJH($i;$MJwUH5KG7r&9lm4$oV~H*I$Qlgg+}^$X%~1EMD8^E zHjpDkkuOze(3qq-?k`I*O_#yJ@fsfFnwCkX_ouxAM)cv^5LI$5BxNFostM;-zvH# zvhxY;;LFO|Ec7gZ8-CaAAsW1AQbPPz%!ehV*x4{D*PeoCNrdV@G((c#SCaleuxS?f z4W}OW9A>%|v|f#0c`T8W<($E@&yP-wS{z+nMI$VfUz1W9kjkgP#?TG)epE(C-B%=9 zK;xRm#tEinR`xcF<+9r4dW8M#>qN(++zi#|jZas7ZONjTcJF7_Y(mjxhmOH}KR7E) zmNZ`0d?6tqQka$97-Xjh#07r3CYCq@?G(VJ+31PhIh$pbpB@A>nWxdm2r%&3Fr1H0 zInBO5@94~F-rYCNH)?m*w(9_`D*Iuhz)75$07qMQjwH;MUETqw)_p)2pMHll7|_PZ zMZ4J60-3F$t0hpmHe1CI;$jxGKh40cE0GlA_pID3t@0pvmBxM$wBXt0Vh?iH^$HKl{37w7RD~YU` zc3ND_yHT-+(wU=7FGiI9t?=GKEsRnJevQ$T3juHM*GSm9uRHG*Q0v#w~Q@~8Gs%K zD}E6KJK)gsY>kSjcvW012-1p<)mT#Dn76NUzy7#qgON@bD9{0)8<42Jd;Aj5KL7Wn ziuknw;TvDq;<~p&*5@14c$8MIS}URb>%_jpj+{_Y=~@Qip({k}oA2`UO8``w z^Bvfd%b?&Yf%vS0C{$X*kj`o=iACpoHfoeDD}gA6vhmH>I3*gxVVbhTF@z!KI?GLD z-2K}~y<=|IgbDVX*07YddwE7*(Z2+ra^Fg}R{QW76EbD%TGF2FZhnzs9hrWg9wgA* zx)AY5`3cGT%N!4s2@pgEWpkJEX5;h6fn^D&l5tE8nkV``QBDB z!H_PMN!UQx6(PJw!ob|R_J9c{;u;qrnA6jnBD~G)V{1Sd=40&{e>&OBgcz?%*FF$@&;KRK?5Q`1 z1ru@@*0wU8?faFx-8dV7GAM72X-dO;hgziKfWehlu$kAp71!q#X zurlDYAW3%8x=nfc3eM4z2S0*L9;>_*3rr<{Ws)9?-Z#B8w-^)0rp4I+az;_a`1Yo77doLoYd^P?+4YER!P@B*rtv zA(3>Oncn!#rT5`x)f8o|RYF^p6%=Xdd8GYrLJfn|H!?Oh+ux7eaq^ozhaN?^x$7_5 z|IjWiWO)SGaOLiakK_7Atvt~G7ZlL1Wa4fl>>3M2a|^38HPP9PV@Vaerb!5kFPILT zuc9O0YW@mczLN$&KI$s?6Nq$*x0q*Bgi7p!mHzd=h_grBlnL~8}pj^zQOFw zLwzSc-%gjX59T@nf`PWXfpW~+ThW=I?J1V~hU=d$SVE+7v9^i;pFT9&H#!C$`6&et#>oz+Z(;KJpV#8))S# zGZ;ukh)%7NlYNGZr;5uto8Fa^L4XxbY}O!JXPq*P4CdevyMD1}a~}VGF@u=gP$h{6 zmtYDHSDZ2qFmXj{i#@!lB_|!kqg^=@+l5a15HKC|y$Lp7WgMb;wzt)=b-6ZM5+NPt zyQ&{c=!r7#sE{M#SS5`RBz0}5(O9(#&r=LDw>0t04_!ldZP(@U1ZMUU&I<D)p0>F2gdRqCbVmCchS%}u)l=z;Q^%Lu; zK@qWZML7;7&0a2!h~9cBW#v=^pKRrE!EV%HsYZ)iP2i2Iao*E1C`GVFij>IX< z_HcqxPkhjPQ&gTt`3C6s8js33tK3sOV#03X*Z+ zvf+P2>`gYpnPez4W!J;izFY*%6&f;KAiZaH)0Hf-qqTf7t%Y1neGT`oC|0(}KXl(+ zEy%!2R?gsTIMXzW*;k1Wd+=f`b9`*;(e)ytO{^6#M#F&noayqOk&Bac$L5E{lCGRe zzg6KmQ!U|8zhMc9_IOB3HQcS!ME@n5Lo=`BE$8N_Y$heNksIJFo~R%DTkFtt$?C0> z(H?8Vy&ryoM}ew7CK`q+XwqJX3P&DF76lV(>6%@BGqzi>6>ZjNx+!KeT53YAi>&pj zE-IdBTEv+s6eY82jve5X#_^joZXkOa;vMnK{)-OgAe^BbuVWB=;xUn3kUEZDkrg<{ zh}Y(=bA@ZSeOb2{i4w#@?Hw3FrKSo?{?9x*|zi+l0mLvn@I zm{U|}aaKkQFac<-TH8;Kp9s?2ZA@gQ3LW}~Jygb#H$xQc%?Qk%!YfrmTC0#>(oRjn zJ-z9M!o>O_p@JTM*5RpO6JVkNy8M9aDD!(gg8;eDGtK^t`;r{{6tuJq7(w`n>p&yL zeQO-Kjb-l<)IkX$FvL;NQ!=wwHwP8kM8#Jp#Ex^%jRk#U1`v!y4=?h(4 z7k$g)ft6&}yJjmY&ZTUK{rN&x8#lR9AKvh6{&jTSUV4$`my~tivm?#UT;(=HPONeT zp(C&(+4;ypT;>*0mfCiT{wJpSwG?Yz2%)wg{Dy1W`~@)uF40s#Pvy{LTN)OaLWHId znv1FL7Gp{_6qy2>g2CjU`Sjc3i6(Zm^3Y-9zno|}#M;!l+r~URB1Zq!pxuznU~cC~T^vo-4L!M+iu?aE5+cP%nRa5fG-hkjtpuSzGMO64 zXO&uo1{>2d+KOMY~==D4sor>(uUF{!2K_~h#6~xtG~5c7eM|(YoK#c*HnUrA*GP z1lQwrrr*IAJ|kDb`x?fbUK%q(6aEpxo^YOS8<(!g@%8AJ%^{M5kiLsmP1P;`IkB|Z z@}ncA@m8&@Jik^$m#Xqsw~hq>PL`&je_QK zD7QF(Q@YYWN=BYoWdkd>85sqL#ho%M^{HJEXaK#M`4aOS^)fHHVT7rLBAz=KH(@FB zvvKNyP&>*XE0cUjraJddo~H)`e$#AoEYlyFK_W+ae{!a79_1lMtfG1IHPD0R0Jj{Y zmcWsGeYDN#Bd)tSlT+&8=0te00T{iXxR)EVtVLTZRSrf%OJ#kvH;(ZH^Y=8^WwPNp zr0>CT=1H7iB&b;T5&nLEMm9?Ey;8sPp>f*;sj>VEm-8oWomJ@6wu|cXQcG-%l~SS` zJA{+)RCTxg{cQK)cTn#6uI1J=@Aitmb;dVet~L$hWpCo| zmbC?$|10&J#$;jg?V}tID^$ZNh4@FYb#xe{DkW0@SEfdu^xdSPvpL+e)Ym$h+r?m+ zi3ZE|bgHjR7@SqMa>yR+*tWTvqQ<%Fism1RUbY|ZY@RbX)IRv45U)NbU}#yDN|nOu zEN-f;b4=Ka)tI^Q={nxNn~Dz-5N|(6ND4<&a3^3wY>4E3@5V()oN0kU>`^JT)hmI0 zT!n9vzp=UKWBc)}ALsZ|4Vs0s%lqcbV_r-Gi`GjihBf%rn^F(0U4#6&9~@}}PGqBm zZc{T~*WPI4tfj?g$ax3{5Q*6oThOUIAQv`keUG%K; zb<+fF4|f@7#OE*{>vA)=e%BFTV$Qu#=qMl%=)1`Fc^<%|ANXJiJ=4{4@!sh6%PjCx zbzQOI9Ho&J2lC^6! z;<2+cuf07Py)88!vb~#cYnj7JIvSYdvE^wHLXIa2s_rG7=^))*89GbmC7Nqb@zp+$ zUiF9N6lG`M>;xaHb-fq6C;uFeO0PM+NRIJI|D@B!fKZW1v3lw`-}#F9lsPuEZxV5m zfgekF7@rjqU7YEs;##^^vlohG=ic;`{g9UKx)wYIox*W)Z<-SLMAzG5buH!!#(Lk? zgx`no$LUA2*$l=MXy@n1J4K%DkExJa8LkTBPN)Q`1o>s;UJ67Pzaes_475s=qV7zB zlkTzvNa_nMCu@iSd`!hROm@iLWWW>yw%1XCV%pL2z1AM;Hdm%9+q(#aGRL}-M=gG` zVl2)b!{ytd0?{{brc!lea>>og)*rt3dX2q6(*9jWa;_Gp>Dh(uc$PMCr?~EbKVX+i z%D;M54>~&2OYU^esUOGe`#^4H*8D}ZF5LABo_~LM1g2XA!k0&(6c|IVYRNwg(M_qG7w=d!kk z*K3r%vwmN+ec{14pSH5aW@RZRiT5N04g@^Fn7=CwZ6jW@8xX*OZo!W)*{lvkfLaIj z4_OQzPD3!!uOmTAR$6z7u2hVu&ue91=JbP%vHPgHrhscbSA?GIV@4*5HL5-eezA=OCQw+$MP zTj50UPBFybD*I@BnnHiHdv+|t<~b2MWWUW4HHC~b9&-{Z=WReY;vFYxR~QSQr{7>prFeQRd5#oQ9F+FGWKa=<5iqG2X5EgHgN^b!@ty%b9&qb%&i>Vx z_u2MIzeuz-_!>kp)X}J*&5?IUrbS{v)rccT0%5Xi;w%!)puAm;^)50@vX?^aOQiul zPVCni?+;VLf{Eo0nP+Q8O0>m%#aFISf#krvoENP^2)tEW4CVe2ChNT?>2_%+-x^x2 zmlfLSeK61F+cWLw>q#9USwY@xo6OqyeT{F%DIt<3-oI}w2U^Bq!-in($_FbPdQG9+ z(0mL-`V^GDR%{DTHXgdRm@73(X8$Q6ozN9?AMhVqjlwDHFS{2SV~4r-luX`Ya7$!0 zU%~~xRAGs)y8kt*U?)Q^*bTX1ey)4(*CbWz8mj!=dc5a8rz<(g)OjRgJyK_XB>gSs zfy~cvnL^%)Q+fY2MaL=mXW^*M6;EYEVmPMoprJ7X~7^Mt1S-O-eIaSOed5*ufiCez0co+OCrVC&woBedNK)}I+Cj{j>#WTr!!y!f_bPq zPovRQN-r+vrQVq19j&0!)(jb3u~G`B zWnA~OiLNbXF-e$D5$`_^cuxKm@&-%TecI*ru7-joIq^qzp>FrtN?5kvK`KTy-g3J6 zZ!J*1Dm5Je%NhU6_jWrgXmhAsm08k9g?H=)#9oO$7Aru%1ifOimM)@S z6BK{fpHV4wuFoCBJNaSjb~RP!#@fo2IEtjpK>x!sKZ*{1QiftBk?m}&>mPpBl@bm3 zYwecADs(7cTSn9fR{1n;4NsNGaP@fjmVJDmKG=L#U^w6hZ;K|eFYVcxo@2v99or+f z)U0L+v$NH1f<}W!qxdfqRk~{3=1I$XQGC%az8_XXL*=BUkH`CFO_gZ~H#{_<@mjM* zq{9ecjm(cH0~OAHD4c*o^(7ZsqDTwuPk(L6LiR<0)XdPH3THaLvYc!0ID&^wSzEG+ zPR1%`K2WB){J_zAlay@1a~{5@S3*7+*bwv)&8x0#Hl;Rnpg<$llegCUgWNQTX@QU3 zaA_~*9vXH)y`9EcEBUBTgZE?NXBSt2yf_H-IM2!$<7R@&76|-@=Fb@PWMGEDPX(^o zFz3Njw$@bv%s*zrxoL;t7Alk3uiSNIQ+e#dEhq2NXJTBZ&XJ4R{CcO8cnIHPVH-?_ zCx<5W)E{eEjrK+xWZGg6E;26u{yT}z@q!Oz>VL&$oVkI}xwSUwV~#i_WrJkAVEoCS zGGU9UM7d6>j)Krc9N(Mi63LA?7h&rdJ~gI&LfkPw|4bi*(tGqa8?*+Ny6qF)vyfW_ zgXarYl1qPl#l5!@le~gsR@{~-zjlS0j&yXn0p#Nva70+!2KTkFpl#fQoAK3P8`j68 zlP8|00<%#1S6z8^^;OV>m+$@@(xSffG{BE%VBTMPcAwRvAV5M4U(T}jio?WLO@(Kx z=>2P>75u|45bl?ayl$U=k7->~k;}1174Az-XNv@iw#v`f9A31VSJBC@zL)M1GNi?1 z7@8v(q0bV3ZHi6MV@wghb5jLNBFcy>veg^KZ74T#*4^LJYqi3N*cA3en)~UeO|a%& zcw>bfJ0`KPxS|`OVUXY?ZY!=;M-N5+9-KceL>5_-k)}W2!$NX~{2zmEkM5%x0qNV| ztymjM#2GADTz45OcXzg|)5ncT6j=T~rCy}(Na@qp&0T(zRE6ziL9D$`(=9mmj%%#m zRKv6$-wWe1+$hUOIB&{}*WJYO%u&#$E~}P3Fg^@owWc-Rw|k^5Y6p5J=67G^HN*Xp zexowal2YTcTMYpr$y~`h?iaPe&si_`W>@pz`hPN2J zfW~zk#+&MSgNb=26D=gA-l-1z>pQ5C3d4M=%wDR2fLvlYES@8;u?wrFD#3Iz8O4ij z3|{qZBc$aPLr{cP$!m&$Z~G$v>0ene!Ki6Ksbd33-wN#zo+7Y5u#_BZ1HHd5PIhHE ztYO)Ippr&Tl#jL39%-k>ek*t146cK!>yI&Ye<~CE$;EV#UahQ&!rUY4qbp^NN?fMgRJ>fx?iLaaqIR+z zAWi5MFP8O-y;dC{QvEQ~pe&xB>nMRQ$68z>?Z_<@PN}0xV)aziyH{}ZAKD#0DsHJf zZYgW|e>Jp&QN`{5r;E~HnQH827zek-AWBVDP{h<#Lv$i#b~a&V+P82T{#fYH0Yz9gvGChhs}P6j33;2) zGM0eAX0lf7IC)}CuwFLFW#wLQyA_xY1?D=st+7aWVGiFtXd5R+e*n`?O1^`O_)+Uw9#E} z0+i6#d8$;&mQ4V&|WE@45g=(==hJe4nF}4A_=V&S%-jCaa9)nh-Mrdp^x4JP2C6Rf{t_J@ZAV}<%>%9%v?E~)2UqnuAl!3W%~Ij{ra z_I^MW^ZOdxj@i=L>aEP7BDEu>)Vqce#l4)8l^zMG;Yi0^v+CxInQkwau2$BB#)fzO z@wx}eVsWbPR25a@8ZMc#t^a^qP18OuM1>Z!KHrPgaE`c2dM54u$9L-E58PJ>vieFw zdntWh;^al@R;zWqb35|2LH4J;@DTdE)}Lv&Wm#J;99_@ISQV)Qt~2absaZ6hE2eIS z=hvWi)m!SezxcR~A;tc6jj_Y%+W|k7sz+P3A+>_2aKy`wUP^`PVnIP-Im;mVCE>AlAbf9mn%4en=9=cn4 z5u*KZW8b|5Oa|esvfbwgI>w#~v?}cCN3}{Pqqw^Vc@sP)*D2vH=n})H`zj4e4S%TT z-8H>g|2{0ZBvqfOX9`DrT68J*`{rL8#h2f*n3$~0T-36$baKrS-!R?kay9#&@tqMR z@tU`HM39K?t4zG z34KgOj16&)!tPDQ(%6FfbE9!(o^jX~pHXo(Y&HViL+zc_SW8+EJIy!YmV`6X&< zxPqNMbGxg*GA8FVwC;2>DbDAG$?T`x)9} zg5G&Jt5Cij{gUTlso$mMcV5~hAJq&zG(zvAY1zXP@ip)sm>bhxPAEHx>AjC znT}Ca=1D@ICQld!#P}E^GS6*A_r_{>gepZy8SPkAG&$(AqvNqsVnVvqJD8jU_aoVg^%! zr%O1vo2XYn+VPt_tzF#L^X1{du-i%X#38CELqrsTO|8#u0jm&T75Qx1#zKxn@r85e zPr3@*6?UBxv*>(Twvrxo8UW3Fga3l5JbVDPs3;`kV51htJ*=T|oAUnRkLRRNyySi? zkpG~Qd~N(P>`47DZ}q0-#<^z>vIgvqKXMv9Vti|}gi^}9WAl$mO}MK)SC)-|_o~n0 zI#aV|@;U(Rh&txsvhhDcv{%nqgS|emCDNk)sX7Cx2t}LX~MP{uj_WB2z9_K}^ zA^Pj}+_3{ZHB-$8_J+6|5TT)OL7Ydcslg*7A3JnfWrZ~5aS#3{1J?UucQPRhSs29+Y$@r%)E0Mk5sZPB-IJ3|_LF=|?E96$Tm zKxyd;ok5;i7pt<;^GFoGOLuT=cIrPgdw$lv0$O^{`5>OJE@~BKsUu??!o|^kZsD~9 ziMlUsDHK#6+78Qi;f`O2!bU#)hi2CI`3C8tj~|&FP!h>N60}%t#`f=o#k}!fIs?S! zwxSEiaBZAp@d!bp2>%Z)+!*UE#3XrpSmwhIzvu`_UpiAuZ%8?P?=)=*(YK#a>@Thn zL_b%<^QEf2793w7n%tw$+HZ5p?EUc)|p{K`!S!k9!V~b^wtNyMiF(i zp1WKQtNfW~&);~Y5~SWB8Bg3Gc{tMBc*ov0jJDb_1}|$-e04!5xi6>fKL_-@vwcg8 zwl{xmLgn17jP%D)WE@W&dDc&&7Aid#`oSFNsYXuoNj}f?^H)nHBYLk;Ri+ebmmUH3 zcB3pp`93LcDZORdL#)`uGB*`Hoxwr%5=|<&uw!Grug@|3lkXg|*Se zZw7alLJ1n&iWL$x1xj#tXp38MmlpSi06~hoTX89r;=$dm6fNGOrO3D0z1^GLi~q&k z&XdVJ=gfJ3ugxm`A|+rCCAKy?I#{X{KZ4 zTrC>bq6I*7FYs<5DUA3Z_3v*Oqh7OxqS76PcR;|~a%W1?wG*+2C zG6A|M^vUo1(q-$y3B5!gI)$zwJs6! zHq|bI1;rB>0AdwJCxTly`5NIItYt!?0tN*>XN!Qh-Cc9G3_BK3c8OvAEsPT$bF}tr z->PabRHk#@wN2YU^mC?gHMzUP@~ZFM0{t3ncf_am`)V<>ta^OB~bj-79+A0p9;x znK6f6nx&n}J@6f@b7H><+G*y-F?y^?KyKuR9)wB-yLx$s4TkV-!!7#wUKdqFTJopk ztCXDo82Jvu?E3^-6RzCl%ex}&Y8|Tee?~kt8?5c5CnquISQGptgaH$jqIosn80b%r zNTeJDRw&%9m$g-zlVIxmDdVYq$&8;OLcNBW^0sSA)TGb%qK)(SUmU@H$Ja!}=_>-> zLN_#8`xZIURX|Png*w1J=v(}o2Xx&c$4}$jF>>(XY_1eu3w)cOb>v-VzPa9;d7Rg& z;!64w4^47QrI<}DX*O!1^gZsL0A_@jR;Y@=$ViF^Q26Gye6s&!CJ8n1B8DTtZ783> zAZ>&{Sl3+>!i)~`3UDZh0BYfBx>=QS72XgJs_%_^grv@xFp=69oCR(le7*zfFJbutdv z{i=|mOP)y>bWcne=UN6N`22+T2V8mA!s`svaaL?ttQIEAw}*)Rm~3%=D9Ew*JVoP@ zx@Rohgp{HO#dOWDQ_Y8Kw5%tP10V$tYRvea>BOcd2$}7nxt7CoBim2UL$bS?l$GV) zF6d%4yR1V=!VJ%n*jy936~gs%>}iDF{@VLPE`RFc9q5^2~ebE6GPK}|83L43QuN` zwpSvzA4Io%TPaB8(s-GJRxJ?PHpV10^#ZRe_ipQ6zVHsM5q@Y7;TYi09s8Lgt|m)|uFsJ@oN#vK*sR z#{+RS92xobcv=(1--B{#K1!5DDxxWpWkXA-VvH61;@)Odf!)CxwCc!bLg zV*i$T%?GTQ(ox<>V2#_1mzUJtPa;7BXDw4zb<5Y(x>(C{4(E5=fgz9K#s2|j8kZ86 z>Vs!~Txg@$**y;0SygQNg}DGTp8NmnBGPsjf`M}4+id%X0aDAB0fV}2Z+b4!to>tc z^(fZ4LGKRNH}6@YiZR;bm=TsFg5EPH%nQ}Ko!}5+y^-bFQ;W53ZCIccvh@T?&s9-- z?!{U-Iq^+I(@kWC(T)6Kqo0;epqjRa3Ej zbp$;ly^p*TBWvWVlI|iN;gSnH=AXB`*8gQyb~=#Ht7WPwm(%S4m{pIuUb=62NzFTzw0FYvxI;? zyq8jKX^uNZ_xel;b=lLwdB-?61Gqv%nVbeo0-a>DT=Fiq6~yA4O>N#uLC8RUbO(6f zRHl>JTRkM#GktXTz5r52W8!=tig1P`T~1M|-Zx2f*I1mDB6o8oo|1z-C`^oqr{zNw zjVgh+vCCHaimMInZ7ofCrsTe3++D`bU)Wg0nd5})CpUZPO4r%7%Vo&TK`P) zj4fsiFwe_%!0aId&) zeKp*PNDsN+)nnau|-WT&8?H^Vic%|NdgOoOr8=Ij-)x^Y46fm%L4j{|}D(pLtg(@^dVIGzawWSd* zFGmoQ*vV~8Hn^UF)(rjyBxcLI!p61mKdZ(A!-BvRU}NlGPy@*AvJC$kd~V-ob`k}x zZ#Ef*V*Fh3bN9`Anm>rA?@}Wf_)PV3UFBjWYO}Q*V)>9KEhVhZKjIDXYy_AFH-A2HjbF|wA!d*BCTweXYl5s8r^o)Xr)_2p7 zLXciPCNJ>zT&Jb&L6i8CKY0mV_(suxfJF6*yWJB>s^b)cS=944%`k(69miFbO#QR2 zT_K9R60(CV%)t{U^$|J}fokaFS>x7*R$C-q6+GhBV>7SG>CB~T%lgt*?tP}}HqOX^ zg*6}toqk$1XSz;R(Z1lKYqVN*g0E&fNtgoX%k2sFK5x`aDTx2KeCNAnfzVh71IJj_ zdPb21HwE%7L(Tkqsm5Q`+F{rtmXX4z^a0>8Z|VF%uZ{IVFZZ{T)^1MULHP4y?t=*V zyu7!hLXw1EXlf_N#NOWrSZz)~HrMLRkx-y#gggDA!s}{*k~6_tjUDw5BkATu{;FY- zGNKn$gILp^;}^9F&0p2|Z*6TIx+k&4`}c9asqHMyW(qvUwse0>(ITN@2Bq2twGIV5 z9qk(Y_?uZfe@yLgb4=^Pa@ix=y)~qCAngM|nJRuZuy0h+&VL!RCaAz;&@0!e1IL=S zuTw6NfRM#LPe_aYBmDd9`iF5 zB&nIK_UR@H7x(%P#d42<0;j~Cc7g?Dnt<(0uSQ3Gg^LPiuucy;f2u}#NSNA@Mb=oV z5my8T!pv&jGW?!g`%`-sAVdeyV#zw#^=9GL{W6;R7S5!^q#%~s+o*)a=hO;SF%e(Iv9qx7HB+QpYE6Y^12p8B%)cR`rQ!(%^^dLO6=sM z=1_Qp>!&;=9>RXsgikc9Tarn^&c4T2OnD0eCG1IS+4!Q-C{Q=v&MQ(^jy^dEu&rG0 zuTS|o59Y(3$&>aw#`}9h57Wq279%@KNdUFYi?MLLa?R#*8FjwrRb)Ry_$nXJnPmx* zQaNS`o5p~o7Vt~Pt z>o%YVT*jM3pZTr~aw0cW>cPA`j6T>Oqc`vtYg+@WQHF5vFm8=`1Y>!+KASz(@zS4) zk^~F=o{cD+2vX6*L2(o2lbO!k!Z3if+(C<~?|M*+rM++z5gCS#=2l>ir(=rbC;#P+ zwBg<)+3Di zuREtkF8~Xvr4^{M&VPTQKq_fhw0Y<^A4hMUWA8zaGW~EK_gTu%$4*A8MG>5oI>Oq% zQ_F`Glt|y}c7+9|ZqyjGv@6j^&$W3#rb@*~E+w$SrktAae^9vexE$&LjD>p4+s2g- zH43?WwFoO)-E=*Wf%kLRofg997zQ0!)w7DvUnW^6&pkMSyDe0t4}Sk!kSIVdq=4=m z*#Ht?X4y078P5zkxezg_N6Ms3Oh;*^wwZlR$?jF}aA`WojT}4NBR};8sGe;_z&=j=?PslRS4^qlbc9EZkVmsDzLqFf3ql@ zHY~e*A~&TbO5fX$ZhE)T@I{q@B+kZiG#Oe6kc}8x+uciq755#U05(dJC z2D$iqbmr;NJ}{dayvd2EUUaJT31TEq7}dgFV-|$RENAiUkab9bN%<7f67QBV^T3(b z#Wi}bpwe7y=`|%Y!7w@i4WP&Gr}!m9BCPj_--<4vA-Fz#bn+kiRs347+dN$E-@=@Y zrLW>#y%HhPb!r6EHcovb+kA|isnL#Iu(qK0y7Z}7p_UHQ&h4_v@svz{s0v`W##W^j z0jd8)%>C6hBdb{1@wQ3i$Yuy6`#eWOFXBebo;K+oL{=6ei7p87QC`(nBkye(_1TH% z@Rp$9w@tN9J(03Mo{*@Y<*HYqBd=I18VjFV_hFvQB$Du5!sCxLysi|3Lv`?gu3)1B1O+7U?n7 z&)J=FC=xScN_8_wa|a3pCRK9~8T&bN8tg@m9M0$rSGh@F5S(kyn9!9Fm;j_aw?^!D_F=utpYt0arAtJI51H z-5w?3t$V(@A4P}wi*x9zn5QRgN!Y$5Pi@(uwPz2{=c3=PF;h$k1IdnmeAuE|V^&o) zns#x}G0wDg7-p~BLLaJH%uEb7=PQ)@4}6aztDDq#Wd4K(6p!1r*$t`Yc};#iqDdR= zoQ_*i)Tw&GVw=H4@^ZeVzWV)X>zL)B}`GX@^jBVoz{OB~eTVbq1&(RJ78t`#9s2M5& z`{m9?Tfx{b-UX^n^zCA`leU+N>_y-Y38zHBb9~C*fC7Xw@y6Zt4_dftyh$>Fs)N6K z5oXDqZQs;_p8b;#fN)oL*12;x5u!=z?n6LGg8VE^>d=Oy{nvO6Q*oa6Y|3_eXbJo! zlMy1#xb;I%F=BJ!WyLE_-}By|ADwuth0V!@L@$ZwRirT2krP=8NyGk)1>Dqa)d`vwX#;HgstEqlGo2B zSV(Sz_6u8gO-(7jQ<6J5ia+KmA4$%+0!$uZwUobM>ddMMb%r-odqTUUugZ8Bq9y0! zhFPq^6+2SqRI5n?(vYdZbR{y%um@>9Tb7YD^sHB&6u-O?Lw}!5DjxpvPpFF6-;2=) zZ<^=3LMR<&@|t!}o>}oddSE;|OFJ7ykhDa+iB|9%!zK3Ta?bcp-R zL5BOP05?ljZxLC5M%UVS|2LNY@_&=nP#kKV|M!cVo<&zbx$1E{kSTSVHc3pipKb@Y zdE4?aOPQT!l1zQk6>WOtvXacow6$bTFHfBOdf4ggrSXO`K+`Bdn$U_?O5Yo=PI7)W zwDnKkE9v;Cq&m{`Y>4ziBL%mXab(|8py`Y#>=G+4pmpUJh2I0T?_V~M@u$@8Tj2Bn|$5CngHgH$W{3HbMPB&r|k7tOHjqy);0OA(>Yd{@K;%nHhd0ohTEhkks%% ztk8!c%P^$U7T#>R+Wkw7m#5BF(c6y1hPGjH_PXxI|1JXcvtG|PDcIfo3toQ9!^y|k z{)leLr9_fSSmQ1EyfJA>Z)#n!WcOU3h!a<&Hd7)6)7`1yL2fD~loXB(#cx(q)(+kA zo9`AZ9T_qrz%U-s~WxPYX9e_?8lvgL9&4*44C5?x8Pt#jp@!q)qcW|_(-)0nxl;1#zU4cQkQi!TMgxcBqxH32WLC$QRo zl`U9*H4g3U`shV1^_Bk-;KJ`KfGr?*8D+GClo`)O+0_ho~v%!82ml;?93+{0b701461esor; zKS(Z3%7HI+a5F}rwbHk~(Pa2Thb{l;Epe~~x%J)%9;A}OdoM*joLk?#FYg0|7|KRC zN7iQN3;Q3TP@WOd%6zN6E-VWhw zliL8z)t5-oeBajOZy``@!=2rB6DYmJ%M^=$g@Jb)Y~|@RuU6h)*?XDV-rRbJ%8{=2 zztdHDdSWQAq-on%#}ruQ_q94jJylw>C?AOP zpXVX|AAXcdzco>)z~9`&Ngu`Ho1}i)6Or>Rmd+8Tmzfafs?YkN*%m(pewyP60;|Y< z9r8T!n%V$rW2dZ{3S5sdtP~&_gn^+JpA>r-VY48HT%vv_>&Yjr z^NcI=Gd#RJ-lnOd^+D0c)Dbs7UkZH)Zh#8*_r?cQJkmsLs`)5IHLA6*;q7D{)Kf`Z zNf^Zkzt;x1NG!K&m<)Gx+|>n7dU+&=J@?yjyg4+?g2=7}i240~hLDEVl)xOr%Wq=} zf-|Nd6_%egqc7$N4TID?x5QSIS*Hxa_kTTM?KNc7Z;cQ02-=aG``coM^IvLu;esBPBDX zM>5+Xklm|FtaGhvoBXbEFlug&W`i+N+%3z9d5_~HY`oJioxu$8QI214M}TnrC^YQC zCjT-k9>OU?>Mmm3PCpd#sDEfOG+YwEt}TSZ!jcQd^tg3OVB1G1yIRAg zlu9!v2VFV{@l@-Cm_g~UM(F5d559>A-AcR;z6#EK43RXViE|NX>ppl>6=Yuu`{@0N zfj1PqLxi{+(lE)jM1bf&q{%DfVPOp6eK$RDbH$!b`qY9)D7v;;Rx@vH^M?CUM5x=E z8J(k6oE?30+^bRGO<>4>|aS6?^#>VBM8d7JmB z)LOd)vpnTTt?aD$vg%Oe3v8-z$9lNp1xYnqau_1uc+ihKr;RhGq-hBBDeR(}De|Ab zMP`Y6!B&MKHyC&!daTaPmGKjdN4yq$Ah+#g=D$InN-8g|1;LXJ$uFhtq&eYA!j?9Z zCEKtApX*Lq3&RM96*F1T^>VmT3U-QA>HQCIt#6j25UQN$%TDO7j1E8H z^__S5(@w5SU^T>7WqNr_*A zsh+XylAMC}Cxc7lBFk79HMQiwVf+VRYCf&No~{wD5Z>KQlk>({`hIt)MzO+UO|O@D zZk=fEH^JXX+pgrGR?%0TvJhjfJw79sNv1c2upHkZoBtuJN=-g!w7c_j4Rc z%~6q+ta?pBOlI?mPC;Gqf05>1?NLt@M>1rk0?8(7HAq_}74yK`sanBO6Jx77IvnAv zsc}y)gxBe#lX0o?=>}A9i00K@+?Cicg`$_Ra24$xk@fy{)lb56!Cs00y=iovvCJXi zV1{jMc+s;lFy%95_xvxH6%WIUJlinRLBt#aAbU-Ody#?M0%M1-Fb4fwsQv!yRJt^R zd#&ML*j1r;Z(}v@$mLIbu7uFAd^gq1ZHF!7Yk=hM9#VX8#iGXVhog2z@Q~b8p!1Lj z9UpnNpWwpQ(4QqJ2HSrR_UH*hoB8c(V9?BS3>K^J+z{zMXWTw=4k%WtoUmy_Eo;RV{JzIM zSECj_a{9e@U%&Ito$P8OZF2jQN(}6{f^Zl={#wB_viU6);K_4eoDv_WCS~jS(bEBo z*MIBdHcvwxHAS8#&o^fVC*w(w->MpEhgjTw#at2rub~G<3aX;n|-bE!ML$V<%kI`8pUH4M~g$A#}z8gi#H|2%esE7yhz-v zh_y$oZ&#Y96YR?DoPUh{26J`%3>DiglnTm6L*uF(4(Q=UF)y64pZ;Iom6&P+Kb*mz zwefa3g!FJD!iD!CA`LJNZK!A8}qK*Ng|-iEL=S4nJake)stA0 zn4GOuwt@tPQS#kvYp9MD`@3$kR$*O!8%fLgh_GZ`$)dH$ginkG4X62+@d+8RKmB%c z2vw*zZn!@1HdZBm21Nh*c%(*~d^R7=?@KfMHay$y*(!$*J7cCVWC+b=)ulsxyZ-|S zk$g)HPAgi{lUhz+l@s3$$2HzL09UN?C*`x+S|cBE!TVW=FD#Gija0+nA+i=_WLhhl z9bUJ?;}6(R=-W_||Ik5R(?P!~j@T@|uJ(-GnfI64@btwBb!zEyzB$>c`mk*9__6X< z6AGutXI0=P%DMkH-3&!FR{BuXF}-RDrCTNg&JkOmGHEN>VL9^qvj+z1t|`r;nXpln z-Q5t$tkrQ9|6{f!O^?omzN#fIJC>A3g2H%#y=;#IEUd+hv+wrU7<+kyRF^zY#hXYq zZ#@qN9?wP>tkpt_8YjQld4$LO6SL;dX7q{n4TgZ)Gx#W(wkP`4a>h1w%Vlg4Q_L|x zS{n)`$#khs#!o~{DT!$&OJk#0b*PAUy{S#d@w2uBevRFQD_ruzMIOm&Rgd=G7zirs zT~Nkwp^=Ay%6#GvdVmpF=lD?9>^wjr_^K-KhUqHhwx;(R?r=TIZ&cWtp5Yamr;RT@ zne9yPIs5ogSWA!oYnD!o^A_uU|9X%l^yag3hgu!lW7n^PiHKB*; zouBF|Oj5_uQKVzI3k5Ql8W{+U&q$2Jz>JSoyCQTnyJ*cxC0?0K6{vc?!vrK}w_;A31ZDbnn8BjX47s*928{t^^!^ssM8}p;&zzOh}>bj z5Nd7wn|ni3Y-?^O>xl-8ufwa6#dz)K=ch&t7LCGtA9_T4w$Nj`1(IK*(>y6wH34$w z%(YtlHOiC~Kcgi`H`7RGs08W6h0p~BOHz{)Khw-{vHs0z4x2TI*XIW?nc;5qa3%h( zI7_Yh*^9wYh*irYu zY>Yr$i)X77dVL7bjO@z#&MCzG9k^6ZJ`r7qpCaFz~V(39!RMiMI<*L@_z%@Atz(2PuglZfSD zpx8pq>OH+?lf#6g#TqMhMSBNX**RyyV$m#MyU4UB|0haS&PUW+$8S;yY!?1 zCl^R%b4Fi0cgg<%n#L$ESnmm(a|-fa{|}&ZI@N`cKY5! zeJVcQv#tG_b%{*K6r`8kuddzMbIg-Xy^&=E45q)TG={NXsKJ$Tq31f2RmcD*SzGCY zZx{nTgW_rEFhH#xa*w;&*q0`3(jB`o9>ciH8h0WS|s&HV;k zKJ8g#TaQ@_hSU*!fzF%?D!U5`?BI#t?k+3X{VKQtYamGH4U%jr&q3;w*D=7j^q;^t+G~V z@9H^mjY!iaxWX)NZE7TU=i>X4wF;!0G{QW|dIh#Gsl-nd~38X@~! zUN-)iC0nJ88oThTN?ztCcN8JI_i(fo;!hq3nYBAk1K%NLpUEzO%nqx3oR5sO7`EQ5 zhZm2XQE2_KFK0?c+qe%$JW3=>*Vg+AOkV955LaK~r6hia$b}Z*n`xjW&hw|73q453 z*L(9}q-;>pmd8oh?&&@A6)y8zdqEh7Vdam~R~Cy?->WS?lEqwD#^>HK6LZhVeWlybDaJi{kqhiTGl# ziOLf-?V7@%0L>FEqNy=>Sm&X}dkx9ejnjj9!QB{G_GEyN=)Npz4*}#r`rm%c*WZub z$^5N1m>S}rZmUXkS^2a$|ZGHx~g9(qr4F|Nb0lLnx${>Ja4B!WXP=UMNmsh(w|O4SWp_1n{f>FG*D zwInmT2nff$_YD1))h^#EyZ;EPIFKe4;x^0m8x2>%{^TK@Etv0l+5#7c4hazu2S{%b zez{TG>-z52w(_YQ&JS`VatgJ_gY&FHC)8+iWQi`0_euLr9L)5 z&5b%2DH$?pM?GgGKEEd-AM7P2PTYu5NHo{&y0tpA3i9>;89TwzH&mMYTNDVZpHz}< zF*ZGVXKRamTJJP>{gwy8XBEL$l!N}@t}NFuu<3M=ooX-Rw=`+3;Tiq`|GjLvv=~N# z5}RcJ+sqY(o9L7!FV4HP0(b0zkRC4Ch3A_k=~G-c(pG%gU)ZGlT6KkmH%m(L1lBzI z%x{Z7|JIBqnS}`Z%bxT5R*N4zsMidy9&iHA2OYO<=J==YT;skKC9<%!;=k2m_nP_^ zsJ@i7>)T|`0nc|7$*A2Sqgw}b4-Gr$6S>E9j?$4+Wy z8H;H})@y9J=WyqUR925y%>4r{KnblNG}an1QQu-A;pxIUYVcWBDt-{Uy_0hoaNLcao5z; zm11QwPDgL*2>dD4g($*CAS@-^??U+8p!x5-%C|;L_`)B~Lz9Ia!=Wwc2CDG3E=xyM z0&>9DNy=|ZO!tW`MMA}P-?XIRF)HjcA7qiE=vW|IDfX9$RG+moaAa5VJVl>IH zqfVTID>paW+FBMwrS%*a9>p*U=LYMbUXp)y)_mq2)P2b^0-R5|(_Me2===hDP%)() zLZzNGG{#oIA=;jYE~XJP?QI`Eu+UT?zB6K{1YBb+mVH=@POy{8<~_lCaeY%!NYn)! zFZOdSez>hR!eR1$#m(?c1OP!@pUPuWXAA%{iFUDuA0Xu@_HLW_B8vRXSZQ?L4~1ru z60)RsNZdYu8*(N69q8pm+j?`%JM#O0RAHPdHIT@A*irz-QX0Ma9_?IRB;dV{TBMqIQl~kv^!!DqUNBZrh-V+A)OZ6#{7b`3=R>OvK*G$Tri z%=4I((`+`zQ?*G~V7j>jU|WZwmC=jE>oa2YlqPXtx}9Uynn~4`QR0NrmHRGlu>}6w zmTB&di)UQ>g;Jr;#^YhEt0k@gW&Cxrw;^dDs07-)2(gg$RPM{MA(!$NgV#W{I9?eX zm8Ua$e$-IC;1BeGD`BKCq!F-h&9Aka*Z9^u5;akcJcLZq@Ynge7!JWmUYL^(%+)SF z(cUslNs`acTqm?*N9yql7Z@1W^hHUBp^a@zFM*9;6~j6G7ezcn zK7p|wo@fX!fdK~K$(iXLbPj#zD@s=Ns3dOil)2YpdBNr`m|rJt*}7Zy@QT?Kxy3h# z{+?MZ^9CipJ5Y~Tz^V?&*2g3O$^K>D(b*Wj7;a4J2^V^WqTyK~t3Zkr@8F~|ZJ2Fl zdj55R<8G}S)JnqE&njv{nLoqx2j=|<6^bQrVkVL&oy?{k-&aI+p+5GiNnj!Jw)8mP z{it%mX+ZD}2nT#(CGH|E|FT$25|hHK)@bP|iRgIbHJxt%23rf2?m6;j@W{i_ch|}F zdjWzxrLkj|+<$V{Pe$o!r`!42-49&qan|UXBIOMsq|*>=(m*{5O|6b0C*|J_4MU!i z+ZnLWl%51~!o!EBDhagq5yP!`i5V7kUIrzWIV>1Zdemj5ZBk278==-%n6~#?xF;U7 zDd%+~HEmW(OMk2$KTu+GN-SKIXBe=H5h_V&Gi+_3GYNHSaOmY4S@nez-^$J}{)v z%iG2c^kkt%&YSr*{?(m65YPg|+6P&9oo#MBRt=aLKC_pXHTM~LsWYE`0%J%|N^TDI zYW{Mpdl#F2hDk0-PU?kz<>hAlN;s^IXQ(gE^Zbp_`{h&mJO8O)POXYECZ5>wYS2$C z{QB2Bi0^4S=7UHSvA;AHDwx+|YgS?vg@)O2L_*{4b~ zU9*AvD_x)EV9XQ7Vo`exi_rh0SX<4ayG)2D19Ne0R=rS+bJ~lEGgK78V?dl1m&qJC zT(BZ#LXn>2`O~xJnZy-elDaGqc|W446w`MRMX%QjfIEL`1u7*tR7TB^$F$?v!tD%U z2XwOest9d;VG-Vuq?(c8^PWdu3Ic|Heh1}YJg&ljzcsXYapM?E^sT_!!o#OeAGyt( zh#johr6FtBX?Q=y7XhY_I!y8YH3 zK&JKdO}vxJLV;_-eGB;?KyVwTnaj)Zi9!nJRl1yMCw6kb3)o*Vg*j!NVr@3tjjc)) zhX^q2?M_7ePp^J-w!yvQBKn%k3xO+}d#@|gKRkVh_}-~*KQ_XYZ<=Z3t@(zWh7Mh@ zUBP}^QMf8Emw`c+1VHVm)XLugXEgK7gX#&6@Vd36Zr39uX_%Oi?FqWX8`J0($B)jG; z|A)_@u)LMqLyayA`%mEs>(=qwfN&*--*~zqazn+ZG#@FjldX>IoI{I&A*!Iyw{Enu zAI=U+MaM}mluq!EfTY2xe@n12364C&JU!5};<-T5jZxLzp(Ff$2`KLa{>8v--MS*? zj;rXgK}j5ICsO*Md#BJDW_NgA3+V{f|9CeTeRLA=Jvbt%miDF z-T#-6QA5Q>*K!!YdzZ!aREspM8#mHY^`Sp!A8|vTXPIuiWVCtM-OBY!sC(G)^`J~9RH zh>zDb)zzlH!fKn%_CXr_#s^{F&wfjqo!6zYg;?zRO_{2tY9I6N9Gq&#VS!nr2}X7p zdDUt^Q{fR8P#}e8)4t7qaU==L#2S5;grmSssus1Eo#^B(LbIM7{QKCTKX>ZLQKATF zSW=p?RzI>+E>qRGyJd5|lhsKajIMa~EJpAaSx+(@#AgcV=p0Gz!A7B5iHliK3S7J+ zo+Uz^8{CS2Wf%~~DyBV{CB^sWbbFVfkkY8u;gYLEO_qR!DF7NvbM&8DE`1|48lyYj zL~7GC94li8Bc=Qb0SyREJX2S)%+`D&8kY^2g7QE>CN-WRe{C1$2DB{w4zGn0-3X4E zrr)gY&!XK+{Cd){I3igS3InwOf%B1!c=c-UB(JQQU-Gs=DR7X_TOJmi z4ISlmfAbr$wGJY5ja7T)35ar?ms07PCHVo=@xYhmrdLFyP&8Z1_!jq>a5zPwc0m(8 z{$5O-2&8UIsiD~?{MkG^oAUEPVh;-;Y|(ZuIsoZoK%Jgi{DMYGoUOg#g@BQ}4kx<) zm?&r>KQz7o4~-GK5fN)xRbj7DRVq361$gDT<5kpD2Th2B(V<@yr}}6#XUJXy zIKfLss5swXw$yxiQ^yj`XU?YqeCzD{P}OXii86QP%3Cg5vWk`6D6LBoU31j=R=u84 z&}M7B-;PBo3}&&oswI;NX(48t6E$6H()!Sfo|CTiDAMjWf zi)}^Z!*W1}VoC1o=AHD35UbuRSf;ucf7VXtUUC*ps)g>NAHd-a z{m_M@yH;%Z5E)8-&8nIb;Ua? zmafp7t9Z%R4|=;gDd-~CsAG2W%NByvTmPbWKb{3hU@*-ZnqY6u(2(9c4ygV-xWh|0@ zsQylO+_`sK-L{lI`~JC0OP7b57kI->^mSsBl4y9b^aa6wC>u!A?bJ(PV7R9Ebhf4# z12xgef4xf)q4m1eLUrpRc^Jfe^Wou2&^LC<-clbif|U7I?|;;}Bix$>NC7 z*$D>Q{t6AD^$fVF)^cMw(ot^n6W(}PsC-&awz9**^BK~!ljBpi2{RN!$f9a7BwjgMg4EP!(ury16uNV_*pd{Mqy6M1Fyw%bgPa-tz2CoE4!#EGv~ZoGMc)x`nxl5c&znaj!dVhX}luFCdN;)z2-2;IPTpH80n z^@W^R{_hv6)nb7dvuCJ$_pcvBZHC!#_LBZY;Z>l-@Rx)%{~f;(SS7HJaVgD~_gh1Z zk1hKgY*q@C#JN~7=(iw=AggL^V^&)YR#oM0lhENo69*SWr_jCPkJZmfDPr$MPFoe= zxf?J#@NLy6($9~Xrtj!Zywe&xR(r>^^k z4(n&i-R9g|E>f{~!7V>kFGWXZDzF((P9$%Kq?RfkRQ$b9aSt>lJhgm+{F{_(gsl7W zKg$i?#%N}I5uQqME#v4%k_IMEzAsCCgHSy;qo5~_39x#&fMHEre^U!1b~vM^6#b4S zL(wEl6ha6;Ep#n3pjWJstXbXgP45y0^p}!k{4eEQy8Acc=j&SPh|Gh62gfCPgjf`T za}E*@l&>{f=Fk;7?oNaIOh%=G6{#XPknN{7VxSl5FC(Ywi+jZxl`hhlZnV0r19!Q7 znNmF|fwSK2tL)x|=(1RZUuM4#*~CWyzuYE%SP}*`jZA7-#^$Caw{X8%e~p818uuXg zNKRRv*G|l*dpPT>WtP8_+N4U;6KnA^&P-e8qeK&cOFY>*uOQ;atYB|~=tH^g>u=4t@HHX=GUU!G+ zB3)Ogq;(S($i@eXvSkyF=9S8_BHvqf(F@CAT;BQeccEEKQL^Ys>VIUWIxV0?$^uda@tqO5pGv>fJX(L28GTG{E;)Zo3Pn?-mH1;xvH3i^ zyVp-?Vn&>^j?A85Vt2sS!Qbxa^!2HZzO|x*eRs`Q?@a3=8f`sT9O~cPT(+Zbo6hg% z@^)iG?LIyfS7Tz_%0A7fu#Jh{IcW=Tw0|L%vi?&hT4`lyV&I2+7p?g*xJ44~9S>!- zQL)k5{#6iU|L{jMPX9!(u&M8IIwY}F);+<6Y5ke2P{^mvY<~6DZRJs9dBJ}Gxu+_d z^|X>`$Fxlz5i*jMNlE6A-xVB4>h}0IgF;<@Ms5aQcO+*T%@~{j%2ASuhnM zJ9VTP^VAaf@by2^{#Qt&@kF3KJ3OrCx>{VD9X>krl0g|?F46}(t8Oh}a{=nR!~O+C zJ-m`u66-l|g_ZJmrrtb@2-Yt}f%Ju2Jjn#=gN3Ut8lAbY-O0dHd+PrIbicGp7!3Vt z>Y%=hLlRF$x@l{(>?WGliYD~D7^5%;z&KN8$SP2Y<9_%GX;KJH494%zxj2;!!Svby zMryqUxqwd3?<_W?TGPSQ=9O6wX@nKBA6{7Mxd<9uHvn~le<#K84Ku!GFHqwfU+F_Y zK@qxZll9{Rm|9@^9U4;=y!=}|7K{5V4HL|TNG-Ne9Zi~O68Xuut|Cx6;7z493FrmA z&t76%`Jr|LlT8>5v&f9L*`~he3}SteFp6=;+aVDB!5tOf+>jm#OoQymmQ|(WzSQ-I zn3HibZRBw>*-UV#?!~I4(TA|OV((hUU6mtg@e-hu=t&>D%C{M)^nB+paJ4x*jwL;v zG|fSa1t>zny=Yu0gif?44o%^w{(29j)!ecR{=r`vEo>O z6`;i$`61;Gjmz4eD_a-_j3em-t@Q<=qh}%mRSDld4l2TXM6U=#C>^VOc~+jMuM_63 z>@4f>-l6GEY%e4}-gnj6qk(44#(G^0`SISyXO3uxL$>%}c3|D0w9mTe>2a2#udu5{ zLi!QRT)$03evB|&6OS{?h{{@xJZu>D29+|O<6o_2aquEXCDC`X z&cX5(vg(n*tNo(l>Gm@-m|ZjnKjGFxdBbfbF(njWs|_=h(F`K2d`6hbzPsNnXbh9& zu~5RBTB_CeF=8`Y)@l|u9;kVvOqMngSDxF9>|fq?R?SVQhvbwu-?-$$)IJ(D<`p}+Hd z2xyW8_~wBpW;8FrRvu)&N7H_QHt8_th3$(xvr8AaNNEF|*|U`m`C z;u@>`?Bl@*C=C6~C0Pb`8~59E4Az3S~lo#jB}z3Y*n*u{yDf z(%j&UC3IUKZ{O36!i6#jeyV=M6tkHT_Z7Apu{Xlr7Wlk_XCR*V? z-&twi-vL${b)V9Qw*oIUtUgrqUSx`I-Q{}=Usu?8TE>ftxaiH3T_!8_Z{W-@bnSWU zwF0DqZ)B<*m~TIQRUsgjpg~jOCoPQqsshDIAL2+~Vdl5guJT8u-xmfkYn9T@`%N2i z|0_Y#yr^g1XxluH!3!~s&471X16k;n(M+=T7<;vEVVntQ*EZlFD-Zj!)WX&+XRon+ zYH73F&$6X-RM!;&Z4@R(fpw;zr?0>X&EfSZI%&J~Z32?B*^&UYbUDqd<u~DEW4Le!D&9&R;y5D6NEBWslIh4V^7aUeB&wg46AiLZ<1W{?Z$?G_br(GqhG%;_&9b>FU zzs$v(%DUwA1x%ztBFkljc2B=ML^hXqaKztc9X86qjx`(kZALCOepO4*f=*EKLw~7= z5!~FJ-~d~D&18g&-*X^EqXVis+J)&6pZ%yhK-g&7XPH>!+ z*;%4bTVoQ(VzZwPr13x~Z_SlJ1$uS8@-y_PF(_Ok_1AR&Ij}42vsOx2GhM~I_Z)m9 z8P8A9xZ>?=$9nJ0l*aG;-QHW8sVJM3G+s#*;Yh$_r1=_OOFJw)on+lZooq_;s&b5^ z7*fI)gZf5haU^SSN0X=vF2HP{6b=`1@Z%IJZ{ZB_ehr8&1c4T_AH07xPU40=9K z?go`EsTsN(0byW<#-Y0fNdcw1q$Gxt0i>lxkl&gA^|?70=VIUP=b3o+cdxZR>wQti ztc;|=X4<~|_H=ZFpBVy%xxDgsot${gFSWV%Q2dJidHI6rG^X3YbJ7U1;4=1Q#O^wU zsIxdOIV7o)r)yUlpbND2IWldapK5pIFU}xlxClE;cq@?brX*vnW4np5_urPGf((g^ ztH!fc2#cKe*?wd`CJ{94WO$#j*VY0J{f#TtR_xCzVY?z5-?`h_O>!b?r9fKneT_^| zL)$;sPTQW{hxv{A&oBS@{*>2Ut{gNKavfXsf51DI*e&1xN|n?*&unSCaX~RCD&zKFpN@Mx^(k_P_7Cn zZ{4wnP|PSGDUJ-~Zj{*WQUA&jj!74geSir}+2xycffPiY=!h7{Y~JCrBFH8tR5VZ@ zWbMJH>n(X&#iC^uCTEc8XI03#dD}h}IcXN1B$STagO|_CwRkEi*1lTNdOP|$S6MmI z-r(P{$NrK2zec_1Du}W2;d5(^kMHEqE)Z)LANG;wQSxIB&CT~TnYK3{GxgQd-KaGR z^ZRmeF81MOnJk5aVo%G_&9Cv>+UB}E48E;S%Ua)5WrcX1WT~H&h-Il;zB}WM)JJ+4 z&&a)EwruqBDJPM=Iq?1dBu3NLL2Bg`pU&XNE+ZLJ%ac)b^LzwGRh{T92&g2LZ7%Ij;GLQpzl4c6_b!eW4b^mqi(rVMv zQVGDLh6d2MUm}3i&^?;xu}{@+#!CCTbv$Jagg@#YSDrsC%gFq~rdjZ7Yw7CB){JPS zlkR$_M9Wnb#AEiSX_h=#6t3Lmu5n#S)!D%|p2nYGdLq72K-dZ;AjPw>dc=n0&!h)+=d-JbftT2zP zn~EPq2yuA45R%w<%8&4H(*dz8V5KCoK9VM;$xlCV)kn)5M7s`Ysf7Y6^P1?s-m8?6;4~Bxd+^TZ2G67{+b@tWZ$YL22CWOM`M$$i9CLezMhw95 zpNm){6NgZrDJ9Y3(t23gjz+?Y)$|K>(&GtM=0N0muLzVgR0*w!7!vR5?3kM8q%VzR z007;igP7N=H}w~(HYeLI1QiQj#|Axdn4!new8iPaN#ZEtH@cTCf)a~k-@LY>J8i&s z=W$e6Nxr;*6uQ-U`TEZpqAlH9;(v{}Vs{cdjPl&~@T)N``g7N-YVfe+ZP+rv=^9^h zi2*U(D4YKP_&c;$B08-hmGhpjm^D<>)@eOXp3kFIdOdbSkdQok>5-ql-`f)l7QIuR z#|Qv-`A`U%Z~YykXu_y)k4dx50Fl<8se<=!zUJ+j<{1eLpJnVwKPq82^ZvZ_%;GQh z78-Fojyn!oeA})2gL0U-X^;Yy$<_paq<&60SxuBd+>3-n9f#ajymuZ?BhVLA-xnXN zR0De_GJ9&^@`cH~Q=%ExrA`{{`j|C*nf?e#D%haGWVBs4u)zGoBqsMl&){vl##@k> zJL7C9=Cr*~L^3!_y|;dQhKviIm<{ssHWWsCj1Zly*zM$}!R!>rj0BRgWFhdx3u)*9O zHCr`r_lXy1s{3rjQ2cGMgBY`O$K`YNe1qogAhN3|COcJhwNro!i}P7XUSyHKQ^YH8 zI_Z%A0Fu9))FFv0R2j6#W#=)|I+7~}d{}8DHCSvqdeRC!-=kScu~>D1%#U9Zw@ebP zpDd3Div~Z}R#FuhI0?io#Z@bLXi&bK+ZU&`9p*yy)A8YN)KY29-*=)TEUQ8XlSIep#KFlvXIgX@ z&Ze0!pc1UjVx;#yD#FLQ=+Da45~1(gm+9F%hXZSBiIa=7UQl0%kN(YTWeE2TfA?|^ z*L~o3?Lqrojzq?(nLW}zpj~gz^{ncKy*;6Y8{~y8!cz>>0Cqe^#2G7R{tWBNxdwb< z6w=yn?rRdFwSOPp$<%V)jqyckH&bLftja=`uYShUaJCW4z8j`9_b?oxm2GA54^_oV z7JR(4Wfpiu8pkg#3$|wsNf;YlL>UAH*$}dCNc5dO1{+^vp{b2kUnLh%`Tc!VEH@R< zr4M09PKJH6)Yt%aCQGn;;hr2FyL#dpsv@!jfB(5lwS73uiEZyh>MG`pr7_m6_oTLp zYJg4onn(uFPREt2MFE%0X#8g^ETFS+Y90eTHqxoG%g@z+Ln7i-u4xC=41rmJ;?v>h zBg|>l--}^WUv&vDR>nMo7gQ;3O5X0tn&H$Cy^ zBlC_^Si2b9vaPnY*svNw^9w5j~F5{zr<_BG@8#BkD>E-YHvV&1Zt9f_cW6X2a zr_8oG-P960OSSP6Z8*--A%($g=jH}o%I(_3rF(Z@3M7po#aZ$nzZA5rKJ!^``LdDPEl6emKI;X7ReQhH= z-jlim5O((VK#%0U^C|ZwA5<>WYy?!mHQAII%Q^~$3jGRREL&w;2EF~_Pwe1;V}!yr z&%^;}U0+dCeGfyr-w%e&m)%j79`kK6<*uEa7}senq6%y(o;I~}U5*}x9STFEN=e$i z`%cO3uFFj=5(POp+aoI(FgwW!LbWBN;u1LDnrgh7IBBrKQOQM;8C#_ucVF`lsjPpK zu9qTH_f2V(HIxJ^!qnd6SG5O9|M2%$N>kiPss$*2AK7!ZV+!{yd!&~4Ae=7UbLwTv zbn;yRIn!f}O6gy21@Ttr1SGv$u5v9+2?>rd=f9ygHB^ zFhVE0(2wSPFW}Bc`-JZA*5445-27sgupz+q;?tXr1$JL4b1h{GYuf=_E?t6yVJ>xp z?%~xGy0>{cThYMK;3W67Adjb^{dCXL3&Jo1%D5dn4VU#{O>}1;J3Og2kq=!>P#{kc zu0L3zw=v_R!@VoI{yobG$uasepNeJmbHxys{|-Bt`RyQ~c`1qKZ=5!FXBWmHY`ZbR zn)rVcEOL+zJ3UB}S{FeRsBjXrM@fOmJxPK&EKf9*aFpUH3JRs}YdueYLB>xC05J#r z%X<}MPLq(dyO^w&e;f`SIEzR(9A;1-Z&vUq$?L3^Gz~X5VyI)}%dc;sVJ4ACE)neigHc31cTW#g>^G<=kYAu*_bD)MFmm4VG)-gx=?+eGLWy-B23R9_ z&)X$_mN>m|lEJ^K)=zF*1f@L#MrlCfz=qH|jVYx?s&-}%`te?z{(V}Emb6;z>8Je| zGu9cznz<(~b8sy;4b)rAnLR1@l^?0xQyM?6nt@WnIBzq&3cIgr(1E3r_ntdlw9>>2 zG83@Dcagn+jW)6M)pT1T?X;QlQ%ItI)NAtiMBkoJ@y&=2l%z} zZsh3yqJ+(2Vb+`fAFmbwdN;CmVuIoqR||)zr;JZ;tvQ3kczJ;SS$)DAfSB#R?v8r!*n0Wxhs*+1k3CdkHT)VnAqZaUZV1$ne72>{Q zq*ZyNyw63LhEFkC&-OJRoN+WuCW$ca->lPi`AB#O!#%|Mn9Cx;lu+X^#LT49g8NMF z-Nk^Sg!$OT(=O&l8JTy$&iN{=$C7EzQHH|BiZgI_Eg{5#8T7g3^k-mSuYb@ol4Tw&OxRM zY4Y;5-0Fyzmq9)xpkCOFeqU?Q;?u?AQTXeWt1NYu$+L$&Ct!gT9wGdJ8%+Aq8?p@7 zP1T<_6ut1pct{7h1^JyxKiL|P1|w5ej^>vNy{d)VvS#pd$Ub_3`%JM~m7U!o=>0GI zg_dE+zcu6V@{9@c5J%J~o|(Py`^pBfaHfnW6Nw?-p7>gn8?Q@H5-{?LpGwK9dVW}+ zVc&T5h}`<6sC|8(+MZQP$wvh}F}6qVvOKm^dPB);sT@LcM)E%L{4p zM>k|GZ{(MLdp(}3S7%c`bLU-fXc0qHs%88I z$MEOML&|mQMEh~eWl6_!R`pS-PejhrNqWude z1=Ec0^s;Qyk0>Sar91IcBypX0r1zzbNZYv8%4z3~=YN=GS$<I*j1?Nlt}e*iQ?F zC@H^enX7yrO+-&)z98)EF(1)%81cT~3ata)Jd}iX=zl+5@O*Cn2>|Vvy(7V3T`Kcm z=#Vb`HS%GMLza8!YgRa?`|qJ-4{$)mMULCH;n}Mn7!`5CLL-OibpMd2{-h6on;jkI zO1WfSIuCw#Wbf!0o9nH&%EofKHZlWG9W)vwCvo~o(~<9|qlj}8YhmX2*FxYNYo$X`EWeA_bQ3 ztJT>q<&+IV<4DW^3)H5p-wqyiUj1R(aGm^o==hT3Jk!OPl~Q%s&O9qH`}Q`o6ANvvBJH` zJD9qJ_}<5!@6TyMJ1wQhKraQ?sFLj&Yl#dxaPQuw0jsCn59fKv!}9S_X2fb0K10w8 z0sp;EgU0P&QmO{5qLT}cOhHBkLP;=@gCQ410Va8NX~@#2SWgcL^Sp6Rd}o#WQgHgL zN!I8`iY~6xQQ8@qgEShP2}6pq?cD6BR3%f#I6KZn69vE)1M+_7sIb4wre_|ZXZ6A3 z>9I3$Y%+QW*MZ_L;DNJfn{8E}g|7Dv@_w%z_Lgt)_qe5^@v|*b$-UYkjsYQJyeIyR z>;agP!HR$gI``0^k0dSz!hq6-gGyh25QCID*ctVDR9=k!N_l~dij0ZusAe%q$t_c* z(+|F!pYP#D(UvQYZfn2yRMQnd!H9aaphLKBedN&MY=2#W)TLaa(oxzoj;WXv_;f^p z%SF2~v94&TEmBOcvfgx!$k6oTbh2VI$eCEIfC|PR<|dR*`E`@Oz`WhdR@urV`?l&? zg+=gRQW>|YLBd;c;%HI&55(tZT97iqbqukYXpjmAq2%)?#L&vRM2jcIb8=>SPBNH6 zJ0t+MEsx`#SRiGrKy(b6JjDTS_rwPq%u+67)X=dxm}0+)+^0R8!buBIukRht+`9bx zv#bqI%aJal&Y8=q@++5k_kqbx6^n!Sx70gf{!SNFKHfXO4F%k4X8xl;#!oHqJ*H95 zupB_NnA!cCjhqKJ0m0@D(4-2jO}I8w#u>k;Xukq8_b3CU8fMb|ChAlXB(-S%)q%dP z=pm80a92;wDWe5ufwbR~*Fru<$GsGe$em?^nehPcrlh9u&k@rd)3OF5MGt>x+n*`$ zJsn%YK9S?3PeA{ZA>o())ZxEc9v{f4D{0)k(T%9{ai*QCMqjo#wUZIO+HQ0; z69Fnl{IT`zBykfB&!k|HJ$8KSC-yM4+C<)PL-SX>9aw)L##xFnyZW-8f<95Ebh*X8 zs4k!c4qa7rmN;Q|K_sTT#^u-$r;I$6CjOM_qJmohNdt8f)c;ie>G-RW5dTYET+8R8iOUk@cnR!K_)+Dd_ru0^{l}pJ~jWQ zItbcSQAFEBT{^a7y9!Q6wd3tD^B)*$fpq|;SPGsslB5~?<`41Yo#FuGtsHuGXW=Oe zoN&9DUscvJ)A2umJVAOpq?c;=rC`k<4TeFxWmF*WygAastbv56Bd1D9IjMf|J#Mg3 zQk?S(H78Sa)ZL|)>#H+$&iooxq1mfy)+aj>AVP5y+W+m?^D2*d`fK=>li3KatVS>m zC)x{5*PytnC>dpNXgcP&ZTu_<2VJJ&`ydaf6d7-fne7Klb?E}_k~i;oCaSo(m$4J`9O6Ev$XpGu_kWbXx~+C(@)=Dd#V4LxgY}nCq{Wg~ zxA+3VA>Zr!+$HtxTM3l|q4=kz1v;{?XTf}Gw7ol(?2SL_Wh&O5vqW=JnC*?>NL&@4 zs+%N?BhsV0))F4SUb%XsD#g=hOq_IjPZ?~`_w!k*L1btC80cN>QQ=!nt2Am**ss^A z7zmeo?h9|)EL?IEvqna;c^MVmcE-g7K6`kBDS%*A^4@SnoPLL$2jwkisR91?2)2!< zS9&^btPtsBGu9dGEGNQS5YzgdAR}=T%h)DnBOljGsDv0Y<__g-XQgy-S)V;5IPMteelySB^&nCr)tIYmg96q4dVC!TTNhBy1Z z>`IuI9b7aZL~UW>AO46m#J{uP$0u2csFB8q|N%Lug5rh!F!o5dkw5 zt!D0iER%^H;C$tu#o8SW`iZ&Zl}6-<0M=)o0UoQ7Q^1j9PJt`(ywZdd=F4Z=#*b@ig?dCuBS$8rpY zsRA{9=ZF$xM!VxAaFXo4rHhY_;=;O}mp)2=bd(2-&ImVjDxKgy!B};^pltH0Rw%_h z-4?j!I#NM=m~<*v$?!xNL8lg#mMheHe&Bq?5jeROkdN6K0D zbXuAn7=rb2$N+?7Ze|jDd3u^z`tusP>`5M4v<50vkvC+kQjeNU80?h|CU1G=-nw)B zf0f9;03}Hr|4&z?TFO>IG$I3{gXJVT@nj@)G3UjmAcBByuU0WfD)GehXSO|()Sq19 z>^0THbkH3oiptqI!CRo0VZrKCGDg`=qFJ8xGHoY7Aukq`;R;|29DCX3WhBz%$(W|l z5w{VNIn6o2t~yNc$k-jG_DA30U%&Fl3j$tP>+{HMg$)f_@gw6a8g#(a7ZLm0RJvgT zX3fgyh`9TOTi%YyiE-j^YJj`^KtZ>uk=tI=vM)FNXh4?9S$n--R(<7=SHS-848`h8 z4A4JZx7ZA)=*+slU{eO#m1vuc^vOG}dzn5O(Yy?f)qb&RixPr4xIfBi4ph%(!sia= zR+Vb;Q@#*=)Uun`YfXTO7{r>K`kxmXS^2jwL4GaGtCjT|9{O8pDbaD2@sofU&`1Wt z+RA(rPqqtpv~L_%3oggIYM}|5!vS2RGXvo-9>*mbdb?ZVq{;UtotMTzAy8r)Qs8({ z@0oUsi$V~~LL=Or$^3XBs2LyczI{EVZ0Gbom@seUTEj@k;wq5lFm0dje|o(NkwLa=2XYV=HO%QqTBotph7O+2Njp zE7^s4hBaIF+u`k26iBcUHIvH)42Pd;Jk)%&bOZAkZCIB6%P$F=zuX^1$Bfie(H#pJ z@s@2niG7LX1qXa{{jyv>;7ddk<}#gX|_9wgw?s-TbrWM}0C#OJPx^Ri6iL-kY^fr+04!H`brwv%B*L?4QQrUW9Sg zy|eY$MMCH?lB~GxUrp*91Xe~TWA_{Iduzx1H$A?-7(!c__fK%NF*)wUulAnb>Ix09 zW@KKuv`pMBOt%SX?0#;koXVZwlSg{X+G)rx@~$to1ZU-XFg8zw2o0Mas@&BtetPOD znK=W|4SKaRSHHo_B-YVk7d~@iuxk~$8Q6H6s~4f!8St{0da9=Z3G+P@r(Y{eBUiZJ zk?RM{=w{$$b12DVF)|&udffaj2Js}yLxJ- zz|+TwzBi4$j7n>1dsS!IyL8@G8RE6CA!%$FdI0s1e|KRYc2KXujr8UdFz5o2;x& zr=N%sE8%@fJwfq*kAa(R>MF{Q$KHL5JZwp<_TlKOwE zccyOB%W(ki_d5k;;V&B0AkbI5#Ubqc_}aHZ(l(C0o4cPQ2+!MrOI0&fhuozf`bufO zx{eP4SLAlqpXHBtjLPnG6giOQ0hb#~5}jU4>E(zjmvzq-;Gtm1=a+H`!9@$*@{T+J z!gBxJrc7YP4D$m_q}wDH@qZwrQiFd)3othPtTRr=DY*WwV!dB8Vl{X_>ygH%C!5K0 zq1vwf!p1p=g#RGFoy?NU}0vdtz{$7PDh z*6f)3V)64;&oN*duuQ}3~xx~EfQobchMHp%RjT};V(SRX)rHmBaPN^2f>l-ZmCiJOkG=vK7O>}*jn;F<5ro31PS_3$xzC;vUy|$ddFur+X+ADof zMs%x&jh(&vqV79+0TZk#)D;FVPVMkCK(qeTeCv2^&ugsjzqQ=&|Ys@e%m88PFAQw+XHWKpFDXk*jpwppWtEA!&J#yG43 zsoBhxt7oXjNah}4X|es%d^>>|xj)a|1JXdN$Hf4|cPheUGyE3FC*pXj$?tTJ?HiYh z(s-bJi^No0&OLg8T8wDV@Qosq0JhxbqF3%rLWf+A=hP%0o!+P??ZM96MO zt4&1Z`EI2oY#srDd0DGi!pU?=9)>}T;jNU5%dI7yZ*0i%e}J#6k7J#n4Q5dzGOjRx zy^B+3KJs(}wN^G$Fno65c({$8n*&U|13M+7?em> z?wD^4qEb50JJnSgPyzVJTg}{^F%@0ohlQ+xkQ5bEFbpqD$Q0Mk^#e_RiKoP)aTW}B zQYbySY7IG0faf$S&T>k(BjU%t>?7 zEP8wjyr_3~hRDu-a5>+?1C`wV2Y71A62D%L4^u^rL+K!A7Kibyd#CiD;{aE+Z(XD` z)p2b910>j_zHw1>sixb{UM=%U)kJnk8OJ)TT*lo6o#eT%*>`Xi4bd}Bz|^U(2=-rI zCr1mwALR5R&dbtH4UjqGkYrK3Fs_wXJeA{&kTg7>AyM=Y+Zwf0<}*35U@J~+UBON^ zGMs}Y#P?+{<6NGX>bakap=F}MTl1!Vh9i?tI`7ohF@a@u^y}#@G%++dPgPN#olq3b zf!^7ui=J1a#dvJWFouBBh`$R7by1nmJdHiab}q{^GZD0d$+RiUGBeuY8E7(rBx>4> zWB~^FOz^x`VrJ~=3rBk6Hp0+5(6UJ`SfE!l=N#TrG|lx4Pv+fd}KA&a<^Ellq3aur+J8E(dg z_x;Bz*@c_jitbme$)|VfE5qT1OX+S~S#hogS-%7Z@T4Kh^NaSEVlAEMrqSUC!*$g& z9{ne5?-@u-;?nNwngs?hRH&rs)$A~Qrwep$zCthE4w5<8UD3uLMm-#K_UNPSsvUsj zz;%}D#4NVjl5q3bd%pf#Q2HOB-Y8776>fz%<&SvY{_c5})Tq($sua?U1yj0i!@dG8KZ%v>5Z-2$- zFrwFr3W>TBRES7Ytm4NTp}D+&DH`ggyH8|a0D<4AzQ>SC^8a(yPl;t}ndMJ7@i~~w z`wtL`Hhi@KM9!DfdQ}zKC&=nT`Y?TPQn6S1F1)c9Gn%Yy=35XoL#tV*h(H~1CvTx8 zX=uO9!6n~p2buz|rf>}c3E-fZT;t2^FN{dEI9w&l%7rs|{{LH`o>Z1R8@s9?Rj1~^ zg97JjIJ8^h{IOt*!U6(=v9@LfZY>zR*HXi{H~d8tNqZWJpTf%qp9wWPQ8<#S7YAGz z96H6GS3Qh}-@7kD@j8|qwGY?RfBoe%jSP^kUThW-sGLa0g%5n#%jMt~ET-?kjSOCB z{>;@dyVA=N^CPKII0~i+@142d+^0exYpe5Z-mb7CeMNAX*H5J{4c8sSaUYiVN(#}b zUiCg&{!`l}Dk&m4)7UKFYfSLr^=ZK?wE5(mt#scW_qY}0$|RAx zq`KmbO0)HsKAsn^k#&>f2?Cut0Wx@td!Ijusp&b}iltZQ`q8UbLTl!a$E{ScjL5~u zn^2`Vvp;lPZ3z#**e8yPeoCj#&cL!xiFcCC9q^ga+`C!}d#NM4G(Xv0K<7oY`gm^8 zz~+v}BXNhf*mnqhR|L~xKpkcyI9OQT8+@FGHhwGQV@;0fRAX;*OH#@4@Ywl&*8VOV zkF#QG==TRE$bC6rTF?4?$fZR+uRi{$HnxCPRLM0GSF_YL4Y{o=>t#8qP_E%vW;s^j zd9PL7+V+eR(-4ji@O%%SXXH9tZCPmzxkIOfj@k_u2i;Gq6D*G=7D$r%s5j#m&x4xM z#R39oes5-35CNPz?o?3>XnUaSmG#$L50O1vd$!hb9j-f6SoI$)Z%F{N#1++_%H~w< zrcB?KH9vuS-p`&9yzL&YbX~unrMeZI-m!RSzeJ{Mxamt5Ix-&^F%CfJepRxkT4H55 zcFnWE{GPiD;%6yc3XyATQ_IijiDzD-*5GDO$}~o&G{u2eV&Mh4N1npIIYT^aox{a4E)ie-sC|Y~+YD!z%_q5Cqk& zTb^nK@eck_aWuxl741z%+QSI$dTO-Y>Tcaj#7|`nDplC}UWoUuPD>mmGZb3Jt9zb4 zSIs+J63^T96zF(ZDy65gmr<2{b627z1h#CpcB_4=X=qWd4rBe`H9^NuMJdIUFQ(uv^3@40QzSdC%`BmhD zZ5=v}XO$sFUuCjr`lWMn{sYvq0q!8En=0PpYg+qnA5Ki?4%>+bt)0CS<#vlU9~gr} zjN7uB`04}O{wB|F=rBzrdQ(|(`*it-yZiQ5c^uabrv_1ft(99Y%NUcnnRI;r`^mJG z#Iy^C(Yx>)|E#5U{SPjLn62%OjLL$+0IwZ-fji^4R-rIIUt(;r)bTk1*3~-i8pZwH z@%_B`pTip-uXF_68a*7{^{vi6eE59Z{sg#twU=d%nMTECpPU6A$xQMx&=BV22}=NA z172^7ij-t0LhYK`v-Xw;Jh!i3&!3B0m9}}`;S^t05QvE8T^yWU0gt}e8^Qs_Q~ zzulr`fNzwfYd3}QooQg+AU)jC?>^L4?GU{`{-|-6D8T}|qEC^P`at5Z;kgk6Xm5@y zvsM@Q?33)MAOq1RW~Q~?T>c|dU-U&SBy22j_Gu#6Y1pExP4Y+DU=ogW+0*1IPkA}? zkCIT8kT9loVJW(@bl+&LMnU<3NxD%WV2d*NMTZnc!WzLNc!nF+@)h*|!6BB-5m zv=kGzij;&H2-R8(QZA!ZgU_`jE_Rn`abWoSd11|wZ*s8H&(F1pXtmnCNVa_zW${{1WXQ67lS|T7aTUS!P*P?hZC|;R zLkfZAbFJ6#9k)R?N(8!^pqX$Ut(auuzt3kil2nW7;Fx{7L&#j|XNvk>K`ts{B0Pzm z-_;H{HeBW$_VIw|pp!pl?jQ%~*m{bY#xr+k8{3}58C-wn9bIB9?!6~(@LzY1jYM2- zB#^Fv=xoY-GPgp)IJ{m>zuM=U_2!qkUj%D{FD!U`=5!co2RTkz5o5=~DqyT%um_Go_xCR0E zAmM0g9yvW14ABZ=T1<}AeGuR8tU|ZDl&P?s*hc%zr9|2ZN_X7~C_bQ_nL4ZW<)(z6XGz=3{+Zd;GNn_$?A<#|5-`l5|%Mx^;UzNUv+1L$>8v&7OaDf zJb%1XEkAo6fj+@c={{7_`^faX28Uca2sfPI3A8Bc=+9o|9XMz3o2&YY(nHbHj@!Q> z#!w(ke6|y(K=c}ez0uQW^m?kT?a-Jrh0N#!9z6^RSWFIK7qdh|2_F8bwW_Dn?o$GN zT09M4j}-_BXVh*M&v+-WA!O^jH;ctv*JL!-f-@Zk_avGKf;MS%s{ZP;QT!L?ittPq z%vi0z$i9;8hm(gEsii6O%|yJd#|HbFaA^s7GV}W~bKX>xHYtYgqy;$B-pp*TR+=P< zGa$w~y%n%vVG3o@-2Sbp`*)ixS<_?~Oq+wS8hBVmeU1&Da|#0d584tscNP`#y}x3U zS9Pb4A)I}q*1i7|uKFDdAebyTu{qqP2CBA>Nxbr=u)gM36!w&ECgKZnzpbP;j6Jcv zwD&wMyJCJXJ$DyU9KcTUZ+kWgA?5VWwnNk+Oz(d|S9o!|7SnPsPq4#v$&N#H!h&iJ zt&CuBn2x}6<-zy2w(-vx&J8BxN_WdpJwo4oj;<@&mo+HQ5UvY8^QZ8U{nawk5<5k; z&Oo}cnOS24&0(R-N{}2sIwWLvQmL;0xHKt>Xy5DP@Z+GbYyE zapbL~3;X1NkVAp4&|S!UvEhFJ0oG`r)XODAm;Lt_-F23)r(!yAFlq=RHF^OaDr|y2 z=Kiqv<#^P*JaO@Zn7lu4dr%y;{OI(K(M^(<4z^WX-ZmN`^5iLDX>y@uBawK1@CB)Y zj4PYn<*rs}sf+j_N0Q)FI})HSewMuO=7MjP1RjfMqcE!)aeJXcOhlFihaH$l3T%o< zOUce401`(?^ql6IzUOGeeY1l!@s|B+1!d!M0;5k%L8iqrxiK;=aHn2e?37W82~*w0 zuJ$Ck-LU?``dRMF=}AP^GkQ~@j3Y=YQC*HFNd1>I|A5VRd2OlR>OVjoh%NAt1UKb@ z3I8-LmLoYN#BT;-Cc&b4arR`ZsbpExJ-g5PC zVvEnS<-t6m(~*t(ZvCT*b&l~<_stVAsr6>z1dur1=;Dyzg@z$JqmJt>{!!+inAP}q zik{###VHBJ{jBpTRb z@}-y}JIN|%_4U#7WKRPPKT0Xk+9xivLWLxjv|%OUZS~>p&Ir&F!iWkk{ZTG#kCIRn zPj3B09*&F1H1Fr{cK{eOV_Gd5v@gJLerBn_BLfi4BM-+6Zor6Jdia_Oma4`MW-+`K zZ#^z5A@iB->QwyQd)Y;)&jb%0(J;)gd;j5QO@9T@;|3TINI4t1NzhBMkaw(RfC2UPZ9pkkB{~2K ziV-Kovfg6a{1==R;sZ7a@=<*~!_fhNC&6~f;{@91^rU~6_K-nPu`|0U(9^@3-tCOP z^EiOx&J;{{Wah1{P60cr!9)>r!@}~H^LvFK*`@yjwEhP;tUIgL;3j`=7?$*Y+|Pl##7(2xwF}!@7$p_ z`QH7*juZ$?YGQ~xX+t*Y8LTF_Ebm>huo@V-NF_`-UsX=diL{v$r)Df@fdoBzd)=c< zcNE9Kv7B!da4KY6M@($8V7umzHnAMyG6YAajDTkG;r{?Xsr}CeZY{mu`-ZvsN1gb2 zNy{hf2jA&6#W>hF6!7aYWojPdvi@+v#Rf<)Yra^_DQgAMpOg(y^7G5fxMq_IV#72> z+3Rsj1>2>Q3*9Ut1C`&izGpT#(l{3Be8#u3)dXVx=N9%;P!O-r-H$2b-cL{*fUUOu zhY)7w#G)=ZE^uAhYFOy&Z;7wxs_-eSk=R*?8LiWaWCY8g+ zXLi2Qpw%LbGnq=2%twF1iU1zzBM&NniPtZcWPaY2XNL9c%y~X9v`h+{M1&NKK3C$u zBrE4})0EsuQ`>P!{;lr)%ZUs?Wyr>Evi`mA>NL(GwK=i6-ID0@S zWu=ss@j$x79)$|y=J-60lK=k-CwD3B;xjypPUusufZA| zPObTJO`sQ-iObo)pPqa?K%Bi%Gk*K4jgXw@280Z08i=Z(S*YrU3vR;YhD!*mnal`iabXyU8C_Y2D3nTkubs77?- zH+mI6gEX7L7WZJAPgoi=lY>vw54qhoc8o$K_TOtjq+#FtJkZ3L!#qn~ZPF(Vg9BbB zv2>}A8`Mnm2*`8{D{in#Sj_MR;Lqs&to^?HW~{Ooap-;eF4}W9WkO~Elcc0$4gnMZ zgJ)Z&!rChiSMg#z3qRf6=2~}A_ZrI$TzU%_)P0#u^pnoB(fazN9)=z@=k7FkXsE953BsNe{cCc<{p?82Uwy`MNtySYPt?S(~3Y|Ha*pO_4g<>-kFsYDxFZ34#2VVdakf*9Ypeo`Y@>F#tXG3ms$RpJ5Jh#y@t3F!W0TXTFd z0pTkZLriA#88Q}u#jTSIl=#INt8cZ+hiIz2luzq-QR~A_bQbhd_KK1_?2Sb=iFj}V z*tq%x#qMgS;Hgn}P_p=;6j{C8y^1>*9U8yD=#J+0k0Nfg7aq`LEDSA>fbbv+5;=|Q2FhUc- zYq))NRJ->bl1S`rQT~;>nSgYM;pDNErnuJH+?8fdHjB@*fxIKGh{+frKuZK8IMQw? zkDweYQ>a+^@MopMMc+cfi<5*w*-BKpH$FHx$d_Z?a0p8wd4Gp_H zcmsbHoAtQ#vd3l5Uq5uro<7aS))*2=`i^F2JH$f$ho#ExS^N7HNo*k) zn#oEK$-G=25>alcNV7!yaKF!5I75NVB=lCL{~BT}aOiKZ_-u2@z#EO&)da@)sXvK@Pu@8 zHoNrY%yr$bGW9$~waI6MGM7{Sd?O!atn7w`{HF?q+NGbj`m}8jyP^@@Kv0i$%Y-IfBx^8sgycqsAAdD>@c|h_q@d(Ju+g#uJ>*HM2f|6- zp{CW3+d;7L9u>GcWp=j`b&`h-5~PI%HMG%vr2E+$m-5 zy*F}}-rnbRQp4Zlc~InJZlUlx-72t7$9U9CD|?kYj{q}kA4IpKhQr3%@)#0w@z8NfB2ngV6SZ!S zsid;g;dv?E-|tep9c=ae9^9;;xN8=vQ_1JZxz~}5CbVQC@^XI0ehlcm&;a+gHDkV? zEA+!ahRc&77WMGi!T(d+SwywfhV44I6faOH-r!bTio3gODNx*kyL%y6p}4!dQ=DK0 zin|7aQ{4GZ{_z?7lXC`pzO&ZOyWZ!y<(kBV_XkI{9oh-fT$N>#pTnS(%kU|TVsTKW zou0{QP8i=ukn>D;!kRMYt;M+X{e4oZ_b8>I$KG#S+Hj<9lI2>2cQj0fjDD?phP21s@j$$E zbjD1!L^nEME%aykX(bckEgq4UNkzxiV9s1@ZJTVq7wX5xO7Stc1a+b{yFJFq)q1wg z#9AcM`?-b>8j%4UIqSCkr}9plPY_Yrkh{%{!fouHa^A3Sb`tik*=?zYv)J1*5gtn9 zLah5*mR}rAoyY{1j30z=0)~%q?Z^;HnKZ5CNYJH-Z!g{g;pJPb7C%1>jCTA7&=oWb z@Wb`BPoYuZ7vjI9CjV$R(FuL~S9#M%k?>6cOBYp~PP0P>umgg>+S9xSlg$=pH2^5k zPxb2i)7UY*O3+FJyi6JFrzvi^!fFrw$IqwligsJ16nh#&VL8M$4$H~5NYK)p93#<| z;@j^-84(cOcxyrW5PtXQ9@y2i{%?!KQ#6UL>=mZ$XY2gp!M`{%yz*63uLd01Uug$+BmSN zMzEg1w5eJRCapKt8iuT?5{{VY{7sS}TdWJNW>|g!wL|O%M*M+eU#P6yTtW#x-lrrf zJaJf@R`)&_?Hy||FgU-K8*-^-oGtvVU)LClliR!)Y6oQ<|ESUcuQ77E2@`d4@%4uY z>_!Nh4C|G0^}=eD_B$ml8Hwbf9xqiQ<10j}{lGI;xf&I6>j@kZs3(y0EH*lvc=W>s zc^dHXqxy17z@PT8t60U+vPz&8JvljU)6qA2&$5q9&m>c z!jvpXneF;A6qmG6hXsz;V*Xt4)ka^#P@Lp4*TzMv$!xDoF>?h;@^qc(NIeZj-Xe}t z)^E;kRAHlWfUBhudYLOqf2|=bcycJ=Kh!`Pq`-*Bl-J_ zBh?NTrg6Su>|k8GSOw2BFh^g(`k!O976notx+G9XG*sZ|#L)BDyEoG#+iYaD0?36_4KqrPoF-TvmSvpNP+}6 zB;3yJW%3vhrLgxU{1TpDj3HOu_23khW4q8H!#Czz9L4U}%H-1{PLLdcptos|N0gX( zLnL$tlF+UCV6pWkk2}T4B*cwomV{?Fw-JjC`0M?rUy%BsP zzyoq}c&qS*c5&{ct}R^85dRO5MuRHo>cg<~r1K*Nairq8DI^qQ#O=sL@ZsAYM*`0( z_Q;=NC|4Jyz@r3%ScI#WhOjh*D>YM;{L6O1p(`%+Oy(iB(6Of}=0Ey^JwN+|`i(&> z-8DC!6#=K%u}60pm$?qAz9lB43%v-UFG}HErxdC5qXzYiTHK&m)_^Z$%%V9Pz|=u|24ZZjvjFfdZ<0Z?O9`X9I@RZzxbSFa6q<9*D-3Kq}E^#+>4t{W-DmGq+{9J*iMa9$!9OjB44+xnFI-GCJtQg6lYH6c3s5UjKmrhHhqQkSy(PadCnRhawH_DEsa)s^ z4|p`Tu!dzSJP&#{=;zCMOi$tnRMF)Mx#OR$>K$O(krL@ps_cOr0tspB?#wawoT>K+ zmpYeo%gAgCWzfF>*4%84-ba8=7w7!b~zKIiD3j%;U{vJCkLIi*GkgT zuRsMqAF2tIhjudi5&<87)p9=RRBX_OeCx-F0)XL66`Rns(P?08+O7^`b;m98?H|p* z80V7u5Jt}I?aN`hP5*s{IPiSQ1c#6;=QK|nqb}j5pa_n=Xhh9S+F|WBC$&Xgk6xl< z_N$$Zp1&iwC^J&hU`)hs)?Qk3BYe;nK>1}(&~i72noi=Tw15n~!tZA*F$ZS7u-{C2 z|MZRuvISvP+|LyC1yQyAPNup((~47si*LP{p~?XLl%Y8;Wm6a%5&_2`Q}#d0 z1&v9;4PB>1RB!QV-GMz}cW_)oauj@)U1k=0C1dk`1rTb;{O?a;a;k5v4Td=h_1twD z#4bUQ37XqU zDv)DFFR^CKgx1;eH|pDQs*n-&_xF!T95cr=7Doo)7GacK?Wf(aAC~BFVkLeb`<$nv z-T})eYO}x&ZvQ*!Ho-f@@!lq`U%fC?IpK93n%Ae7_afBO{QGZV?74!eO-3O`B4VVD zYLBYk)!Kz=SedEq^A5++@kb#bsMj`W5h3^!>2{iNXuyfIEq?7P>qnWB$2Lv<3BEa=Dl_`RxLk5|DV;GJu=Nddq4xlyRZShG{&T470>=Ly; z-w8J3ILcWQd|I$KaIZ)kM?d{F(fJu!(zpp-V!iKb7eaaz+oNqnDg6JG3#F1I%E*e3-KrQXl`WARs$-KJjc+QwO7niYTb@Z}3 z3*?aV#Uk$7;L2c2q^V|jlG0v|(MQmVUbB=t@5168xJ$egrR%*P>*Cg`a<1m%+&XGX z@)U^d79!IJQT~`MTIx(`G^*pipYTTak5p7H9e>{sx`2NA*D1WoGzR$>URf%65DO2+ zFUYfS-bk&Cvkz*Q$mFjH_5u$Fi%{eQSQLniB9a4fTb*a<-R@`RJ>P`04E;x$;>V&e zK_NenVbi(q9W5Cg>dm02`C>h~O&3uekxpM->5tMdwfW}57X4fU<1sJySjA7}LaFWB zx!u0Hi$9o*j4bVc`k|ZS`GkLCNX#WwfZp^^`=W@%lwIoFMwB$y+s0~(*nVNcMG>yLiF+)?aE8WIf*aGMh9AnUeWB*G(5x za`Mkqrb_f3rVZ2l02MQPrwD(oMaTl~h z@9%@2wsy2r6%IOBrp_Tcj4Q^;bYrj*c|NF_rWCjsq?T}C9;7(MQlOHCkD(5IWs#j^ zCYVCP87$*fm(`>tcl%?49Ub`XfQ3;A6t@?9Rz^nNWFWj2J;i+;3KvfSLa7*Z z>R4DfYPC%9XcIYc?dqyVbv;Ijjgs-#Nd1pdz!qD-LXrs=ti$SBjXo?~coD>6@jzA~ zR|o=<=qpq9g{Cpg2}9UoHLI7bJ}I()`r9dJd3Kq&IE7ZRiJC6UlMsJ;F<-{z5p!a6 z1ef$o*IWmqN9#T=Pl$M9n(7y6`GUM`;)tq?stSv7&5sYvV`j`XJ%FSRT_!<27)oLn5Ecn@dp{9&bqFFT8CmImB)AVLlP@=r1<;Z&+Y3X$M!Pn z#1{3xU52t#a`S4%zBm-%F~$|>B7oKY zKD0>|I-9EAE6WR23X)=cFth&vM$;RVy=r#mAuUBZEF^So94F^YoK?{bu9yhaoZ(2up*)EPVZ zv?WTvz6^|PL9{llmr88$n2^<}ah$)IP1la2T&Po%IH+hEk2*?W-ECVAE;xSbv-=-? zf3a zj|*M)*6>kkoR;@tEFtqrS8>l;R{p~z6llnqKTjhUR!JtwuQ-$b&`z3Yi4Bf0>WuYG z1?33db+K>DqJ1hW@S3L%pWK&1=;r~iuq9oi^uhZ%^-06SnE?FA&)TZP#exiAW_NkN zMtL0JJ~enRJJFbMlID=8rmGiD>PMIDwpyD2JTLn1lg#14ZV3a?>2d-B>j)L>8B&Po zBmAv5=2X6_`oN8CPRCwdf-75nH$ z^sL1EvaIP{W<2~Zdbm4-d{0ETDh<=6n4sgaAvEZtS8Hg??MI!B5b7i_r|Ho@GKm&% z4_tiXPtEo9tr1yypZC|Hz67K`H`6uV-YKtSVIe8Azat-5NkvJHYfqZVI?@RQfMapi z*JxMO_O{sXN~&)KjV{23H;+qx8lHQS%mmZ=PY0G6!{lU}P9WB-FOIkDS$?S(R;3G<_Qr+Yz1 zmc82E*N*qcy6`NX{4s6Jd)s&gXvLJO$hBb}rMr*~N@NEzXAFZIGOY9pRpWO%7WW}q ziFe_U4oP%h-}BHO+OpnE9^6p@m75b4A3Xb!3xcHG8dvfHznP5V9DMEjG@{~6=Xn}M zLUDC_;5*zhQgX@ntQwJI)v-ty$@M;|Q?rAi*YIXK*sNr#Gg5pTf`pXl!nKz--tOi` znRS`IuEwGN-k3`O^2#U3aEMXb8^iMMR2Ej}-SoBP!DJ_1OmU`q*FFXbT?YbS=O10} zXH}rm3}`U@=5}c@^i&%YP`cGpJ5j>XRBc3|eLI39NJzI}e`q&<{)3G2N8a zNDNPu5YtF{(2Fjv?8?+&jgYlq7sHV&XBomX~{;{ z{r$(&xrj*BRJTj8_tuaY4aqzMb?z2iF~nxl@%oq&&uh{Bpt5t=t+H$X$4^`T919!P zNU2liNy~XVLh=gIeQ~@@(Q8_&ZT=5{TT&@u&lEoK$=Q~KobmxqgubunwSJZ4H*p%> zG4|1fZ++USM)$7kSS~$tJW9N4{{yIXwlQ9=%PUz^+Bwk~Mp5yWM2U|0aKZ#Vh=`Fft=^zhY&^ zc~!L&ecwmlqqj|*>PB$fnKJZuwXv8ZWz-^B71tm1xWtsdJ{)xU=uBg2Q!)>;{I(J3 zPc>fR*O+XCeO$P581&xXk3X1tjz`0hov z%b5s!3&S5JqHn4&Ft3+r0A-T3pwV}aQ107=3mpaCIrN{Nh{(}-^uI6u9a4#UMcVic z)BsR36!f%^CgeCN{Q^0-I2aSFEA9JxN@j`5({s*iay>HBW`2hmH;XJc2k0F1gClW{M^ta}aRthZ=vP8JuhyZNRz{Pw z1OXx}WdV}}4VGtDAqXLWtKU5h8gVX)ltWKJPn9xy#3LfzT`1UHC0tVb^1;_rxV)}( zMCaKGZJ~VS$hd};7p(*tdx910Wba!+I+nuLiyc{_usKvX|#a-{CVyRY2eh2#vPcvIfU;pG1aoU`Mja?<3R?6|7Y#^RXbsF|ZUerF_) zd`CcEc>b?%3=w&`fU^a*?hC*<_uZ{(`JM6_%eH!cv--zX@e7wv8y5lLOYa68o@twA zl=iN9_6+z#E0m6kZg_&2U>(ov8xB_=5~Rk!+X<((d~PjAmZNpHsIgR>Vu4s%(++vcd=Q?*q3j}l$*XYNFNkAP0n|nxJHTd za7Zu)#j9KzAbm3UH zfKi~$mS-s;mFORxiEQ+AHy)r%6{H<@j9P2li&Nv+{d*?Qylnr#^;${d*&XQmsB06b zDRuvCnz9+a#=}}T?boHL-QDG=r2@NkJipw>TC_AZi)wU=rab$N` z&!#d$(srClk?*Dz0kI@EC-%W5RzB6m2KClQJp1Ar;69qEpc;K6@0u4GY6XU01|1PY z4RCUobj<|dVM-j0DuS*DO6ByOSmT24$c>xX+VA~y3q!m5vD-1WKu_jFs)M60-y)Uo zw7*Vh(HY%u)ko7W_xZ$7VozVK!Jx9$jrA;|lz{#DM&_irJu9?7DG*+R6#4dl(DxH; zr-mlA&@h6!JUD`W&{_3Mn3n$fq1$u-QBC985yz*VK}Re5@XFbBC%gh=`QBhu2z97)mYTT3w59H}|%%3lv=`j+MmSGK(g7qtz$WhQ$ z1A^;-m#Xz1dHBZ4L(Ca`;Yj;BaP{@}E8$7J#rAW@+t?jdvtTS8{u#AJ+~qx3P3$!D zA8cy&xZMiyK65j+u>%tA*+|>|J!f3%iWwXgcMaddmu@KCsek0Y#{Ee|eY5@dqoc}D zvK%sk0l!uT#fSp(O;v@c_BZrvu5Pf&*hz}2&Z7i{ZBktB0zYd}!a)1|{GqQ61GPmh z^AHkF(e*GN9eFahR^O&^$Uh7UA>DezjIks0HOq>T{&nlH8zQ@9CO2SM#Ev$3u+V=1 z#dvd%c!O5r@6eQiYb|mACK`9mUUvs!C|GQyi$E}To#2%JrDyETgXb3-0#{`;XV{p- zEn*IOH0NfCg_)AGK>sP6<)Lmxz2B#7mMm8CaoLm3q*fcq)W5Eu3#q>+{E`kG2wx36 z;iqeY>?_Anr$e({(ASp0K7{)*GJ^z1&6c66+~9e4qSG%-;p*-97^EYJ;0~(A=dYj zl}Lj7(r5vDPNZ%l%zd5j@J0PTP_fUJVd1*H5s|U4U${&L6FQh#%m!1y&TYc0Fyu#Y zNw9Arh6+luu~-UCG(Y*pu@S{fFfzHr=4<7as7ge=Ha}R3np00~?_K4q-qWlTnVZK7 z0#oMJpJzj*9^rN;cwY=a!uYM!WNJhmB-8G_;#Bon@H(}RhRJ2>Njq=(-__ud2w&><= z+$=(k9%ED$#YR!j8tyuP%-V9WGnJO!U zE%YgsZ7U0zoKgLIqAi$I>1n!-FHmwkrqQe|gd$Qg_iUsy#l^{`*jNCpK+N+8mM?%k zIFZQs273Tjf1*BhS&L6*uR@4h^Q`$a3Sfq&KK5BDR82k zNc;-~q%VBZFYehfC(Al2c`(8%$m68v&^{J8(0SC#l#C zZFS{GIg$28XCtUZA<(iz9mz4nTMx|2>RJ@5(kD{pz&Ubuw#T^NBsR%}LZ}hlGIRch zf09Mpf0WEC(Mw>?=%4WQtifO~s2KuSDdHB_Kc%z#+MX^;~z>`8%2{Gtn zQ0#B+mCXlpN0vcDqZNot{5u5!?W>uDEITG3gNU0cC#YjNM@lwaUnNNJhB09SMBM|T^;*XTtPOUe?2NT zD*+d0l7cxQI9nGi{D*%g{82Y-_v;tW=riPsev7%^_H(~!{pJ65+wQEf@vF6TgYEqy zGBS_wwGc>st0-OV$J5Ni8(C1vBSm_j7CxsH3Rhm|?E-7~er5)Q@Dnb%&HN_v-c$Ui zGg7=7$ItbyImY6d#}3ro>h(QJjMvro75XtM`j$-=QBlo}Ha3o9o12)FvTyHS^#b;X zpYllc)b6}g8MNFqn_WLs61(@7HuIQ&9DLwTZaAYPdl<8z%pgwKwg1o)HGy9}T1E5y zC>G3C2D!>_TMD3^4_Tn|y=^gDqj zO-8gkYK3ja0pfL7NhYiz;!@d6kiE^wHVoVDPsymD^O@lRM&{jrYl3B2!ZkqlKzC7N985~7OT{M>L^+iP86TfZTL#aD>f zhFQ55)-XK|Nl4eljkulY#d?UH)~i?l9ul+sQpJSKAv*Nl;8*>bwhSy0*MZz^G5xVM z3k_ZY`s(Z3g0v=`;}{{XEg#tt^7c;gU+&dkgJIgM1*+ld?GC+#Hy;84wk ze~+)5vNb9zcfg_;)48pwTwjWrwzJV10||* zp6EKE4smjYjP~E$Rp{9`t`zn(l49rp1x8W3KBGR2L*(aLKI1$@PTBY1Y9dWup%V=4 z71er+^wQkavL+`dXQ+#%a&6;(-ZAKn9Im>i%(30=MSY+gdHRY5sr+jHL4!o|_*I8s zL(QbXH}5)kVL#eSAq~cmY^Sfy=Qa;!7UhF|2UpzrSL2_Xx`ZJi6OkKfrcHR zJ_e`tH|rx4mdXA@E+4h@5{#3zQqffMGB-{cX6`=9=Epai!3a_`XVij0LWa`K&=fBV z@UUM99O`@jM<*q6SU-J#5U1FxO7Nh=x?_%zUwHmUn8seVR>-hbEPleI?Uxxwh(HHx_{=DTzf83C16$q`irq`$dI2Jl%q6xRP-2JLr zYB9R&Vi|(E0U0#DDY+}#NVm0(3=6c;JZTOt5ksnyNP@t92+10}kHBtLh#0)NY4Q<+ z^e3@|)vm3<57aiTHWG3t=dX&6*^5h3@@+rVVi>yUwow#_D^f-Uq5E!sh(DHI(Y4m= zeIrbc01Kx9o!)!7C=tl{W40$$D6;K##>ZDdejj#45E$d}4TI_PD7q@i<~Dqqp~1 zz^auJ-h2H>j6|j;?#lU~J?SLfdLWdprR+&yEb#O3mr8K8{ILWzSRfx2zt@QPKY&FL z-z>lF=S`$5hkaenY;t$f%Dw z1Q0;Cq1m9<184A>^M)FKsfS~%CP#qGuf2F4Rde@ZiZtG>omvE!^hj}!>Z8!_UpAdC z*R}YTThQdR*g-L-+iBtZ;%CP>wwFtrIgTi^80@xbmmxfZ5m5&!IQTtthdxCev+l#4 zaJeCR_gq$mF=F*#9ovJpj7Nr9C)fH~kZl^B@w$KOC4N{l3;>HoAYeKEqy(Z>rjeg$ zYC^$wT^{f=OXiy^wU1i(rDkMMI5*FwR{oQ03y?El3|iLKl;gI;8&8&Gdr59>0C-(E zI3@j>b7Am!&w3VwJjdJV?aJ0qqbHEW0s48o#V5uLaNzgS`=B-LR;+|lOk>&Brr*jl zDS)Xe>VE}Nj=j(%zTPlCx@s-!NM4;`U#EHFrWA#r-?7OXyyYzzPM`mD|57D1kT@Wg zhn|JMQNi7Vy8n({jMgRYvV1&jm8hC|x?ajb*MR>3gMR=^hYIgctoq-p#j2etMBCqC zRhsob%Ff-pcF)HxecI)BW)c=8=xiX^|L%&r2@~^QAZ24`ceh~^U2!Ck?Au&V263m+ zar_5>gbbTp?Z0ib&adv#N*~#qe;O)4XG!Fo6BK^ZdRH_R$Yg9vSJF~jnQtl_&mBT9 zZW`O|>-M;m*+8QI(?M{G^HFyx-U*g6bpLlkamp`A{|9wU+*z_7_0vslxlMv- z(}9IEm18UVV^+?}ZvFIUEn|88H^Fi}u^cy@JZ|1H9BmQelB-d&rnsj$b&J?vl)1^_a2Zw~S363Q=O} z7qRmzH7jy?-J|4U0=vddbSDRzs~K91uSLR4ekA+^i{F(rq6$=xYLP{Mo52b9EQKU> z#L`UZJl>oK2$pr9kCs!AlueNVg9Bb|mUSSbHK7<*!g2%m?ZAmCi&%YT5+aO2x(z0yx;9Q&7p9EY@Lg+%!e+1UYe;M^@kn*g|L-+~3k z#q2)1j=}TWO`cd_UyjwaYYH(>&DZ?%xfe^SjFVb#TSW;Oe<=wyjMOx&4W`vTnw86TuY@{P|GEp?DU7X8YXpAK+<$Zr;X9y~G>PDl1=VY~+x}HjM5;8}ZzHhu44| zH+;6C@R&7^t8u18^^0z@kSf)PEt!ibumvm7Jc8xmI(5{QOKhN8^1#SrLhSqzkc&H` zCFx1Sg&n6W%g7lC4E2LOX*cl%eM5lKHjo*rV#^V7E7w*|x6!6rEc8?gWJFz%F!inK z^i${g8X?JqA={%8Yuh5iY>w61Y9z;{7}ZI^ToQdxA9V%aQJs&&9@L!c?roEcZQ@1| zi-ORduQNY`ipi9C{Ip@I$YYCn9>_^ovHfzJz@^@Yd54O6A$kPiVDCFBML!|I^-^&> zT-GvmHM{qXI0au6a3}kc+mYc+UM4yRnfklk$z}I2jNFKiRE72k@~oBVO4~aFQtqM=?=*Re6X{-p$yI55C3f*y<#Cq z(~ibaUt<5mN(x9q!4+Nn8IE7|{FJpw? zBu1%xQ+m?rD!AlAtHQ~G1*p7Z98&WikFP$vS8wW~8=vS@FFY$bUr z3fw%jr(8W?aY=~S5i@TmqNxL>MeHo$ z%qj6A-;^1Cfism(*Ev^?F90E}M63wVm36M~>=%Ml$a#K$#WNk-=pG4t3U;K;@*3^? z_eFcAeRULV(r+GTa5A7MY>k(VHr6O1I61BdKo(Q^$d;>$twyQOF-`K=b4+wBj#i^I zUs6O_g!f@eO-h>TM4LR2mvtQ2Pk~NEP)|Ux#XlPvB2+N@%^19aJ`B&%j&mkT+!2oD zxpz8+a znF6Ae0T2SH=i>+pdz4`SB8WJ1eZjvM2*2NBmxLfu_+n|T@^e*#_Q%FGMv6W>bx3Oc!J_F9kCd%tZp&q_9KC=e(l82 z-o6;qTk{gAJ5j=MJ*S=LB&V`(AfL(D$v8H9SL&Rv)!EJ+kX3L zhBvO-m%sZ%U3ye1XJ{`;Suz^)dOvvudZ+p2QkR@5mgnlKU%`QLqmtj5Ylo=wX2hFt zrmK{3ymmXchx?T|Et;~V!^%2Bonks?EF@wL+=5(;Y*O>Fq8~`!Kp~b)FWF<&pzc`e&)vA{^2X6di9Q%AR9JJs~p23M2OMywl{uWm1g^=+TH8d~-HT%z5sD7~yA(z}?xn_XNr z<-(q*d}R@KU4nz71RN&(+?409Lfu!}5I!ihpu9powE#I- zXKU4hEhID0dlbl1E|m4}>(&2^j)+fhGt>3N8ew-n{-`kJ26Zui{A=swxnOUixe}Tz-jOiW}wB4kKyJp?1m0l-E4+S&@7jW z59~x$ACivF>@x=JolS8ZwPn{xVSJXo91Rq|t=QH@DAq{kgs;d8z0(vz(Ci#(t$SOz zHXXqhK%o&PI$JyX^Ki%NS<%h1HVp*~pfBer42hT2^LjZf@~E$(CUJt9T7cR*WW=Ut zzd(0RvlDe9>>;*;L?_HG=@)XGMT->sMf2zy=mg@M%BMss@HJIc{6Bz*O>7fk&R^{c z?X5(GDbQ+3uH@bLav%Kfjpur~hg>g!wp#4a?U?8Z&bTc%l`-ltCUz|XilB%dbAK$% z^$Sx#+e8B9ncmXw2jjO12y2wFkOf{fd}OrYXAe})tjo)yoo~qIMBc`qEdb)!Gc**g z%RSGRC1Z=Ka}~RPDsnKDK%ckqQKNH38iX(*t)p6}yQ}&Zq(;+3yuOp+o=^1EEf&p*bHi(8I;aszma(V|k4t;jvTrF{KlV#Y^2)BO)xk}1$u!_c z(e&N4h+P(VbH5qVtb{n>)%88zMp znRE7>z{$B0jI&m$gsB!IVlYItYo2IGeVvq;-)bb;|RRorz;tnW>Y5sOfCT}|)o z&b8)pUkIv!<(_}>9u~sIH1E(+vOsS=&9c>j@%7UEgdqk`E`s`OLVS z^xYyvX|86fjC&A9#Y*e*Y@fD~+l>(4A#$xNT%2Ou)omj2XI$DM@kWU-P?k)LlY|-p ztYW60^4hfnS~d(td%;o6Qj6vL%bTwuZ(rr&`X#QRSm;A8i_E{hk=2?>;$PfR*OQiG z?y3?f{MOE-W5Np+jvCI@pOSrYmDFp9Gb_;@^M>RFG<~nvVjcIp+2~_yu)a%c1Y^b_ z9O(F&iS5uWZBg;Bus>eR{va~H(LHqb5Gwb`u=GL|bh-nyZd3Gq7JkCgMxFZGtYidQzn?Xn69N)$9am?K90i@jA z81;;KFtw$d>d2qL{yEh%E+jPEY3z^LAvB)Kz)W?|E<Y_m{43e$c9pEn`hi+ zy}h|9MWX3h=H;)gXBqJ+5baDRD{Uq37h1wdihk~rLPwsqRcgyOd)nXPE*gYPrmEaI z-MbC9uBs6`rfQsg0dF9c%g##p59%z(hTo%L$ZjfZ8eD(5htDx`PsixFbJnvD3VtV2 zb0)QAV>WL4OrO1>Hr9@@E6`@&*JgtM&Rl=HP|_^2?3h+er7SH zC{XfeCkjk2t^WE2NaVmp%tWL5_tE#puKnC{#c2)mU9R_=#}_VXexzhnoVgMC-Gb(~ zs7LL|9|r+{@87r0b`~|tV&mFC`&>F0lKnonEyZS%eu^bsC>rG7VD|kOyPZ^zC=Q2m zhG<(R!OKd)_D1@llT1hyHCI(Mk9l$MdXEP(orU_a&`GfpQ9d>@L~HFsrkIE;VJL?yM_`_YctmcXc|94;lVdtT1Ry1!~bYbhNiV(etKDLK*f>b3U2fm*Fw zLD{Oe2)~V8nTZvMKl9TDZ%FRd`T8TbR2#(zrjWlftG2Q|;10k`5=M1M^--+ExCqw4 zkZT^O%|6@j_Ypw5sBNkfbuoU&T$iG{+2Uwik-#4>m5$|`1LwG|URB^Bq_r7M)-FOw z)R?eaQeKHDHTE79Y5KOjcOROk3_r&_>)Tl6T>r^#Z+sp{7zpNnc7L^FQa#WD(n#_% zN+t9?&uJ%rU=UPiwn8!oy#OCITr5?t+0TB&_vQNC!4s$RV1s!X5(4h*9kr`JT)bbI zervpyQ|4T2wBa>SNM?UbI~5|fY~Lc@Vz6XwSN9n^dVF56ri`k{KlglX7N<}jc^1NdkK5@PaMMEO`>gIsPAS7|JS!-0; zy~Tn3TfNcNVN11`5-U!%|;-JfR9#P4FrjtjT_m@&RsFD#>92NOpXyTX<23sG(z<1Pl@uEm;3j2&gq-x&1%%!<=d1Z zC^Fm}?ZSCeIq&&IFL&AxQV^=}lfCP~Az=fjnj3qdmlXX5)y3JuRlu{PbV=S;*%ZcFhMj z6ch3&s(UzK#tt2r<4Vo@&i~q`duEQ$$M4$>zriV#-X~2AcI^>47Vs+W&k{I`)fCke zIbKA&^OqxkjLIZvZ>5Rl7UXmiv{+hAl}Ok}<3dFfO)p&wf{gwL@GZ{UPv1)@(TTAr zP70h+81p~h03)10S`_KOIh{{*luq#ZB^f}I*YkF&;0@woQ)CI_*&d!`B<=VY+L%ag z;KJ166}Z2CJCYik)Sp0g`Ct4v?PtRIJY z9#c$izXM5cDQ`}kumsLg10JeA$l*2NkjM{;T*lgKhy2{fCXWN7Lj+4nEmJLDuj{`< z9ZzoWs@kB9(vu42-6`|)Dp%v%pipthx6BweNnkO~U9J<%JZ$1IQs87*3i@Rw26ej_ zG!@wV3OUsy&m}>aR;06lRYJGxfTHuKEU6nAW@v8t-$FT8!+=pSY1XV^)J!{pKX={8 zS70P?u%#-ZMI-FC;yYUmds5mufk9jpZOYbSA<9{f>tvUlkRKe0x)q4Oe8x-Bhc-F- zMx#>at&x_-ydf$pPsKS(pH9oF#5Z@fjpbr{zg``5K9{e<>{|4ye|DjN!9qrN9^*wS z9S=Ikt-%ve03fruxF?gZ(GL6w5v2NU?${yeOL(-F?u=7cFPB7Ztd_ zT(^1|loIE=U^q*6g5TF1DPu(q1KkrQ6())!J5w=D`$m4wRN$VT*sK;uK%x#C_d^yli=$gKz5`Y*n(}Nb4#GJ3R2#qY=#{&iS6eo zF`>(|4AUSX5mI$2tR_M%f^8)PAQp1}w0idwpJu{`Bqan&q1m%_Tg6nq={PQkwM)XPzNeS>lBEp)WdVWAh{&ku>+xp484;&S_ytJjp>J-vND z+`QF)Z)o`bgOP`iMjyZU>E-yV*KdCQkM#y+~QX>ea^>goh_Ep%Ua zwXLubrm1UKQ#FlzSz7Ta#x`!5+C;y?>SJv>w(jg7O}Lx7bf??V?%BUUw}#!>e|KY9 z|ED{9cVmCtS3gdlfCu9drr{Rje&jGI>zc1KIb<<`$YQwal4}Yyh^$~W0a^UfCc!g^ z;dlYLrW_6knEaAJESJ_~?qxXJtHEzGKH4&Kh%7oB{#P6GJX{d>x0?|CC{6m5u3=9M z$MI?e^uIln6-3v8v8gH9%5Z#8AOq(NKT+#USKZJ1W7?%Ql*w?W8p__((u5jW0R?2Af^(0IY@^74;u4(}-`be$j=xRBiiF9L(WDngiL)vcU=?@HKJITT82=pd2h zj|~X>8b&GB=s-V0{`0?W5ioCLJWqm|h>Yihhs~g2o`Jd4J1_j;hCVjk|Nb7iy6Vzo zsRZ6Q!|h{E=4?-R&%^g2Yi4{r1=xEU3tSR@##Xtw)$}4{|DZsHl`dHdv&!V)C={L; z$A(!23!!*>cTsfD*d|rjJ9etEg5ak@wvvsrL0UCc?}CWBE92P>BD@SO2%qq+-H*Mu72iKXAFpEB{~jbbD}f zxa~kRGw(;D3W5IKgAC<3HdpuxT-hF`9&B?SyMLh^vn2|_$q(NRjl!@eG~wEZ%A{7K zz6-+x`o;p8d`cdN)mN0+m#cnJ>eA1n{dlv%cnOJH>tQ=DX`V^ku393ylS*Z3IFrR8 zxn!wBfk>SuVn!&7va&E=qW~(fS#&*yy)%x8y&z7GKxNoT1OLq)&!_NYC{oneuRn}t z%xTo%BI;830-?5AQ7OCtS5mLZ#t_O%7ke||LVt6GcP{qv_YwdX4I&$`ROPsn?w$+7 zVgK!~K0eET@g9HvMWBcYv0WqJ0#au1^$RRmqC|cwW0y0niqEHHu=9mNai{+3p2T9B zQ1`+}k$|b`jE2?)AcXLK5=#*L;iQt9Oqg&Y%sjhu;+y|C>FCOFAEXqNJ>EO*e|#Mn z#K29sDfU7*0qi7I6hHlMW(&NqDbx-MT&(57w|wvs0cQ@LL1=Pz zHNyNRR%u4j=P=22$sZxW`y`+S@LZMroQXoYHd%rIY?2;OFcVJ36E>lMO+^ARI^hf= zw^;ONm#>AeMxp)3CTHUzR432!r#*s7mB;_tx3QR>nEemaD`0B?-0=t3CjlzN!07+q zpk%@moQW~$x@l7pPEVvkr5VPK;wM}M=iNOjJT2Cs<>qzKvjEt);yS>lO->nNR*>&# zh&Ep$r3ov1Rfi*&{L_eIC#DmdXv8Z1NO>r&V<{aHD{nH#V|V+QT+=vl(huGCW7kAb z`p#M3!_?BEcN6|`j=)VC9PAyASDo+=%J+xyC!K&SQC-8bUMl9oh_%6(niJ>H|LfT% zN$cW_iE@<_WMS^aA>)DDWEcrJOw0p--x#2gSm%7;EASH>kg>(I?!IDLVg_+s@*h0s zJ+DH(IRi~Unr;NItdE@@@j%to&tedxcm3(I9}nZh_W|!gF^0*Aoe9Xu<1m8(QJnPn z4|z-W!~qz?`fK2zU_=2N(0 zCDgb8{;!6H#Wv%R#@92v0~CJol_ok5K&;80Y=Qbn=Qn8^y5 zE?_bL2@vRz-^10oc_=8cU$T|l3|)*4|1sei@7aFRpO8iV6n>a>RA7OqF!+GVZa_e? zfK(v^B8Nujh*Nni8(5iuqvG6h7DG*!3X=&G4Uuz+Sgk5-eQq}vYVyfHa+RRLj-h%JPUv%Rw` z+%7V_J8H3Q()5N>}DMXsb*5A z!4i`SM}=?d=g^W^$M?s}d>7;FDtq#9@cy>0?2dHIPU(fkA9GaQ&i; z>8d-y3M)J+&EQObQ@#HUp=g1$baafg*B{Dka&zoPavirA2x3hck==Z$7R|ajgnOfk z3prp1O@tkSupmJ&xE2ZH{HlYdV{te!8CLqN36mjcR^U*0g1dg;jp)o)m}OP>A^~pp z9-~9F_7C>=7t#bm2aI&-0Y$*rB!aV#tYMAKoIQp_$aF;om=6?Xze70WC1@Ij861_D zz+|Jk46X=~C_`gVjj}Wo0XR=027&Ay+5A%`de1pO`0i|=AdRd@pYC;e2)&q2>JN^1 z5y;3u^Io(MC*+pqv6~HQ4hopjBw=|SH%~SHV31NlG--eps<4L{Rux$!vE~ra1*iu~ z5%RHzGEe~&jlM2H_>7$B9V`3G0GRx6(zS^hls^cECiW@tGvHJhDVc<9j8H%%q6rHM z@K$yrz_CJ!`Y-O%`RxuFkPQk?-2kqcI`Qb~$1$2f@(p9g1M(l?Fa0XFwgpkU5|{R;cV zApOPbV0-oO z8Kw-?mn5lz*oe;c1%(=rkOH5U*R-Wjv2#1=Z%}0w4!K!JKt@4B;Ys87D^wZ8*#S}t z2~ypM*7-2OVm1C`S!nf_(*WpU1$M=Wx73tQrb^Xh10LZqH^6|Tn1=W*GW`q^Sm^ze z!Ot)(Lg`UBfu5%S`>g-v@ZPhY$FacBb`EuQ#ie;D!%+i6B{J|?0YyI;mD>&!%1sGi z4g`?|Vg(7BU@-wT0D~dAPM}~5NxT&-5adtMxTLTq{>J9+)3{uSO*Wx*fCWZqmFVoJ zMeLC%rdt96ZXs%12Ae)%38LX`Uc|9Lff@?Rn>CpO8{uj-C0LB!gu~HjYD#iZ5F)fh zW~EXnJ;Fvcc0>a;B7QFxZ!8!qK-eD~@?8l8KFmRu`~W>P6u>roAtt~KavTI}xHY(W z!3AuU18_@C;2ww}XgsKJ8LB`k+Lr-U??f6SG-2gM2GWnzk&&@YwfOMg2F3e6fhg1@ zfBQgIIX@A_xSQ=nc9-ClAzs1~Y_TU86SvlZYajrl16;KeA2V@GY7iJ@t2x8FK;Vmz z>Ir)q{Su%OH=|W3d`!WWOB4&vfp_aLKy`2ak>EQhag@}u?Q4zRqds5t=CzC6e71|cZvl2p~6Yy zG7OnibFuXriq(eZ*iT_(X2|k_GzujKu?Z&1UKq=y;Bt{>!XZQLDc~4b?^Rg|L|j8T zyL&=40^Ufj*}sb@WU+yma^m0xQpYIzyQH}4q3+2nE6HdOmyWi|;41cBNR3M+^;DqRPnMi9&s z$pF&{a2w!N0s0&vq#PE>AOdPpz!V+ycV}RMkrE_`)`P4CNeV(VRN+9@+bdu~jIWvSzD9u^1L=G_JgWIIZ0fvO5Lb>0R0c@+h)IzE&=cgG?ES6$+?^VUyUe#aNQT z0!~e-)Q#=6*5;F+0|j#-i$!w~K?u?VqNb|gzzoy3|Cf614{Z1g zkHAfa*~j~VTB!@gsdr)50`O8W7sFW%EfAB$eq2B#EV|5a8_7?DmT_PQnTbv6UlJ;T zvI2Y@sg-$MNS;Um>Oi3)GhR|S0D9?%_HawH1Jor7tXhVe5lqcRY!1A@7^WQL0RbSi zF$#K3u%$CmI~d6{qyd1*9-t^#8Z1Y!8zTRW1@Vqq1M*@bMZ3Y3;z}{RzfCMC-T>kL zY?&Z9626`evFqR%B1d-ql33nA4MgzEEjDqWP@hI1+iTYq`U1HrR6%m-?i4xX-Xbu* zT;3{?sG%Z8gIrTu4y7U{I2LH~fM@`tKrCi*g6*vceJaZFA#G8?B=dca^H=fnSD&Gr z_+DuQqkt5uHiB45Fd7sd3MEyJM8q8cU5GLGNteN&6;OC>Uq!yecyDMA3yg*67tF3| znZy-QDx+fJV=7<q_BKwzNe zh`tB2;(;0hrPu(7oQ8+63PExRasuWSs56-m^OwY+skuN#3n!C+_zTow{((&b(5g&} z;8#HsyDU3d&;;PXny{8HXU>8`ZeHX8-fiZ^)6*%f4aY*zD^sGF3g~RSm zHeM1NsZ+@9rxDlK2~-9O(`yUz(6A3o2zlU~uqW{hh$d+L}{qxH_}?f+@BzG`qV2tyXdhSkclU1Gyqym6c9dv8XS2KOb%K!cg7^N zavhXA!BZS4KLg{3tQgV)VLM%~@TtT`ay#%8F=ngn?SN?M?ZA^IyRR>eK5$az6c7zB zmz*r#5FA~+Vfrkp4rgO`M2X}!Q;|A#gTzMW6yn4lViy|r2S=Xe>-Mw?E2h;Pg2R{W zdFCdFWpiZ;1h6XZ0pV9(Q(AD$p)#YTfxWZkk-ml1m_X(4h3XQiMB%FIhfolJF;E4M z@2VYu%y$bs@WPq68A(5RDm&^Sws#)EDGo&+;{Ax(i?<@kqWSRLa` zS}F2Y)v+q30!dE@fsTqipg~R$c0rJdfy|Zyf71seGbgeE+-#|U!!?HBFM~kGDnCWa zS>ElZVNzfO#{Fu)#EdV>+=Fd|$|FUOZH?w!Y`~YRs{(#14 zQFPOKnNWI-QxPMjNmo!pC}7!&$R=9T3QDk3_zqYnrNM6mWFR_5TVrKmH7%D6;r_ND3m;e;3N{@D$%AVlL=f-Ee;|LsOth zsyzpR%zlHKS{5H9rio!Si-_9={SLs9wjjp<#T#2o!YT=jav~KCSCiS{f;vHlkooyk z@E5cW0nv?xmbouNA_27n5-k^78aWEV6J?L2WNuXj|o*CP##nl zWDT+50LJbV;KZsT99fYXjMI%}XDXc42QZclD@dH!6b4m?jsh81BaFZ)z@f;*OY=9F z^T%6jE}S-H&g%gV9pkkpjlI4>&~0y~@;n%uVe!YHs^F{2VZjt`pW$HXxq*I8k3BRa@5_3&cu#^@d=uTPD!o38sS`3j3 zlof@9YNI0|6N||YqwHO5O~%;dezIi1x~32gV6*q%{1{URCIuF6Z{Sp~!t9_7#;O@a zQt4Xh;4OyZ0L>6;;D&kZaS8O*r2yc~171y6U^x?kD+S64u^;>dA#-4`1|l^`JcDKK zrwkG{R(1Gj0*c9SGQ%E(_~r1*m>37i(HD^;1#nzt#2FU-awi{Pfex3o$qT2-*&udV zUXnnG6B|CmOv6(*oN`lKdRf^OM2GGbBmx)-R?E5Nz|BZ9Y)!-2*-9$2s?!bb3SQE1 zp`3D|!e&UQ>)Ti9-EF`3p6ZP)4H?DemYn3MH7Tv#s)w14uBFm=3pmy##+ z%&J)0&JrGxI=Yn{*n7Mpq^Nq$k*AhRxvv`sLK57v?fs+W`le$n`4Z)c%^e9WqOOP^ z8+AFqW=*4%5VfG>QC*3waD_+0(o27^A|$avC;$jg;PrpnPf{sVWZDx{$UR}3|KitJ z0||&Xr3Ia!4HNO*Y;GAb_guskn5Xd_*o7Wc$D1#{?reZ-0^c@E0i`tXsrZY7ySa*TWsIjoMPL;Qy27-5vtAJ_~f+-`I z4M`G8#aK8K65ybAL6r@9Ul5C{DJ>>mB8nqQ_*XbtmeGe~{GYuAa)o0cvHn^GDfwALgKk|#(HDo0X#FxzFM=3X# zI|*f^$9??^&s4D*gugW^e_7OrOie5T4Iy)P62vqB3{X?TIY25PJJ7;}1HA`W7T6yR2Yf)60x?e@B3^4x zJjdtZ<)CdN56cexk_lk1T0V#kciOLT!RjJtJ|l|53C9%n)-?81_G^h)gfpOtU{wTG zARJ)HG6;qY)iCiqh2g;KaLRbh3X#^O7l03fk%#!GUokx;lb{?)Q_%(Wq&1brKSGK(fcZ-%3?uc3|IBicjYNAVOtMyn| z_>x7mIsL%geT&^?2stMrBt$5f-(zG2XfaFG9IIlkWQ>)vp!S@gR1>+6WG7EhBU#B8WJ$Clig6uC`bbe17$C+Y z1jepm33&pW4jgopcB(16Q<+nUu}dBl+<$U+_BIY2vQyPDX?<&1%gm18|Ik zVJ21!dvM)7eS0h99RDNQuL3!;<>#9=G8$vGXlFMiX%veN+9^28HQV=`Yq;%Mr{k_X z*q2poeXE_he`u)biL7Ldea%0$t!?P=RgHD>E$W53*8=*SZH-ob;c3DSpsxIKN=sDf zKozbrHl+C2?m1DbY8I7zvtH+Z&u6^anw~Sx!Xq~yifk4{G6I^eZuog&lxf`S#KHQx*6iSe%p7~9WXM@6G7~D))qv%rSHKJPyF`!%%4i39VhupQEn>eehUF;q4uqG)amPy)@ti}aY>wof zA7dXG+Glc|!bw;rI_Mt~U`r5rF(b&om!DcK%pJg=Z_>Q%SueQHKE86gaqYvGA`~SURp>P+rt_b*MhX&{MJ5Xvx*X?!N8J?^9e1YO6{uj1>_@Zcgdz5~@lL z(d;a^ryc1}>LN~8v8nsg@X-2SN^p%$RkNyzA`njzPqLcp+zF zj7|p_`{Yb~(DxxxAWy=)={>~wqfEd|3oE*l+=EaLk$5iQG8h203Dsmzwu|-biXdfW zxr2g7%?9vrH7b*m(8;+_nC^u!LeAraj%#DhC*S}oaTmJs;Yg^oP;!u2J!d#ubgHp` z2Qs2UJZOPcBIkp{h}u}40EGn-e!5`DY~>~mSAsuyxKVa!yA%nbP(+B;N8o&P)Vc7S zu1T4~L9zh0ny(=4vNwh2C5Us=gS!F)G@yB7qL6~j2m|0#Kw9>pEMwB`aqRBhOk8lp zD22a=6)Gcy&V1!(rzI*3^U$XK$UB#FEmS6rK74JUCv%M9X6dH<_8fC&!;y#Qf^+40 z&k{#(o@`t^OWXSaoZM7D7_flUY<3~i+ieI3b6VIR?F(}ELR82(+o zjftXbl$fpLt20bEF>c<`2OM@pPMKBTz-v7*J;uw^xUWf%t?xZuy)N|A3L^ zl6kkzh`ji74p}-%O3RO)Ss8!B8HnfqBnPzc--(# zyE$67yu5)yAxRny0O&Cmf@Cq@6bKltWWZikcOCi+8AKXq1PC-8$|G2#f>={MvjHv! z(LSM`VPX!7X9Q4_2_a*mrkPwL!)+cHo~t(*o9QVuGqA^*)|AmIBzS+mJ7@)STkK6y zm54pNQGNjvdiwHM%Z8EOIy)UL28_}5FcwJafD4y{DkDb)IU*i#UCa7*q7d}}xbR3w z{mDU+gtsngZecMyl+r{J_x5es%L5-hd2`xXx#Fw{BgWWuE@$lUIaG_Zi%HB z7KqpjltabmaNtN{!4eP|pe)5WJqih7_t4?n)u5S6nr zfBW=hoC34Vm4fuD^j__9vqDbwozUnvVm-(4-|k0!;%ex9D5-Q|RT%NfB5v&3{({Rs zj*PbvKN@H_&~0Z5+xH1tjRot#d*qzNH3hwa@iR-BU-a_1!_u{O>!zo^nf}WSlfu;X z{@aGmG{jwP96RuZLF)BKU#1q7TFfkoJ?gY^85M5m(K1g;nkHW=9?Vc zztJANW5V1cpj zzMKstw^pj6M^FYrmqH$!)j`xp5|u^`@Tgr-=hu{LQ~1#Mf&;w{h^({&L?jLHUb%KS zS>dp)mZnWf4fH3I*h3(BfpQR$1KEZR3%czS#Fiz3`v4)3zX+I!8jA_67M4IBhc#KX z9<2AFgxC|H*y<5xe3nk(Jq_T32#FGt@G&r11_inx0FA73-~{MSaEXZQ6B?jEY}JF@ zL<}e7c$ZUpaIfC>WK_KZwAnc-o*g~f!8ALPTbKK$iMaQHv(v9V)sJgh7J6{%m&jt5 zw-H~=_6S}qU1$_$(-3lOlikXfJ1JLB|N4TpG~4DVxjjia|M6OfKw?s@=SZWo-`B~G zQJx-$-R&BP((a2%OHPB5TW zr6ZVGJQOh$kO{#?Km}FL!XR-1+zJG83~x}6$|V3R1Wu?`jKjpb!jKXIBK}7mi{9(I zd?jB+zPM;AD*&-5H@7RuS3>m}BXQyS(EEPP|)T0kvm+c=K)m9tSOXsfp@1G zx2WtCv7$}P?Se6>WBF2O0qKu#fQn^kWw3;u0umgKObjPamBRyjM0U6D!~W6?0>c8% zEX#z4QmX^3L zgB2M1lsQOD5I`cD=+iN*3xo`xlmkDk+7W+%!TqE`64WZSQ8J921+ECVNP-- z@ggR9qyRq}h^~*OO9MLZtZ`$i9v;z7*DX>R;cR!C!r#D!VOA52^LW~gZW8ksDJvP@KJbs?` z%JX@+lWua-k)rRcPlSHYSZ)yd&&Ry=IwcKl&5oxkwR{6xkKO;5iQ%f;n(w&rovA^s z-*_Ka2IVAF>6VW2LbWL- z^#EwL9ZU8$u1N^fOjUe(hjWBbD0uW>U$)>zl|{CK+jV@a=dWi&B!khBg&RD6xU+K2 zOsDBP*g8XXr6b1z-MlT zQT{mGhN;wt;!f9EEtglQ+I!SZC2wdF1W4I|cQf}@fvjJ0otu;pwPP6Hx-gGhKa4Q2xf*0 z!a5b!Z$L8A{z91DM8Sj*i~%g88D2c$&zFDA#xuM{?xTtY0vwCopjh@jf3VXnd}mVR zqE|SZsyb0?(XCHc{vc)Ue;yS%*mL;zG2T^MFPouLP0XV{Nd-lfc1=Fjhfd7Z@fp1v z_dGJYIE>AEo%!Uu)oZ2XM4u#$g*{jFW_q*F#;@7Cyfdmi&hGxWB-hpP!mBZk|2%i$ zQ6?@nK~NIP@bmi3krC&#G;zf~wWI!nFBEH!m{qK~O#alLaVn2{BtZCLaf7%uUX;7v zm~y2uic)*E$yE2%Sd^W4`5^ZguTTF<)wKMQLmslhl$AGt%mv_|76&p0`X9GbQh3+e zZB9tDO^41>HV+LxBfT~`kPi5>kfYNZH|KoN;FkwoDHEkS(66A$&J|R014RZ1P_0*CvM|sa^aVK%J>w8V_oayU9->bZJ_{oF6e_;ln!X58sneEBZXa=j^I0sPZa@f>T1ysrLy zxju)pk#Ye&9p}Q|V_+?%m7_M%?6E)*^w0u1lh))x(%@`xrfSSh@`h#C)jhx?$W!kz zaLP_=&st()UZ-_xyU+|WQs6=a5ntjJ5@ zEMwb+89f_YZX2u573ak#U9i~NG?aSfU|47p!^3b#_eH`=TemG6`M-E*KH*lG3)U1n zMPwI-f2P+nRNnnEXGnS_?b8zuT^+QL)*q9ay~fx_y(8yu>f1FuGR6##FwQU8q|ITw zpKsu`*c>W|9n>vZKQFI&Cee$AX^G_FVggOd; zOy1U%U}Led@q35IJq=zag+AXHq@6l>T79V5Wy9`;(Z79tWlv;s+FEMLhJt{`H|*Ue zUEdkHuXq!EJJY5$g|zPIn)J=lkA?0?BQb`vp4Ku%1(0kIGiJ~Gx|M^Zs2 z)D~EhhZS=GQ*d_`H3Wa1s5&Nkwo!is#z^sdV~QS6Zx@ zFU)m84{PhQ{6kzeVZtXU5y`~f#_wQnmC#PhVC}n#cSV{!y_jY!3I*E7m{&&2 zb)hWgtR@q25F}j!uCXYV<U$V~!vlA=RBl!A;6uAt#prK(H+Df1? zQ75Q_1W3{8-Nm5dYiXDCt~{ZNaxA8Qx0My%tffo4!aY;f<{KGIAIQ;}pcrbhMw{!jzbC*Bq%D zlKQr0P?|5TiO(;Q>aQ~!{%n!_$&#JBPPIH6=#xy#X*c04sug{!r@y1K-Do`KwqQ-+ zTU~B|cjWmj@4hr|6;dfK(#aL_Dcp+ zBki-dvP6$^g4`+H-_)6=d{!Jeb7rkc&)r*tMH&e=HvAxX92huF_v@qAAG$Q{$XwAV z4K>(eN@oIpM~u^ER=Y+0*~hVtpETEE`odPzAs`Jz@lC z!?*&fh%%UDno8>jmWm{~C0H=TgA#^^t}B30m8Vbm_BSD>VEwICG2IO4FG>_pMO0Xx zqoBuRB(*e+-D`=72SgyfE6tdCtmNh=T_HD!2YIbAJ018af0M=~QX+1Rz&=lnLUjM;R20 z2tMxK6<}JypbIA!RPiJW^h8t#SUTdS=29slVtM814Ub#IQdXKx|A+7r5-6Q%Bt$MEAtr2mL zc|ZL1Rzvv9@S}GxOW8?XMzh(A%}SCqoc<{g*}MGHd7ZJtlf7m2id9(yCrRVuvfGyq z``_#MBsltN%!SbVd&IbTL(+&FoIMs1nKwA;$UMI3RBy&eQ-X;8 zb!y*Xcb6aBW}nisHQBwD7nQG(mN;;IUS=^Xr1~?zHDjZrtAy+9Yh9ZdE!(Pgc$~I8 zENp%@Y;x-pzJpP+o0HLw^~RT}tFBUNwR@A(3fu4YTNG&|npnMDwd?i0%%SGFjjxN& zFE7cv;PKq{hl||El8OT$D-9K!JUCbbu5X20A1z^F>;pzU*70}UAiLqan) zMUozw$8it2YT4jRjMb3PkQN0FKXnUQXUL^wbW$#4W3oYFDLg@@9>HsEXN7bSk3!W& z3+KTK7){7-)f0p=oK1X{?PYvuaZ-|{kkvycWr6tf}<6+G!+?}4IAam%{P|f;assaKpcuA(DCW|-K zf?K{!N&(RJIkdELhX7)~0OpvQ(_lR=;p zQ>%ER!RlU;D=3WgbSlOiTz&H;WI;q7mfdm z4x;_Sh$U&XAx#R75Ngq#5>y%PMo*&m;?{S!(T=LO-AGF2ruWdh6|N2z;|)a7-jq?c zt|VwS+kjbXIptos!%oNRN$xi|y-pSLLX1jnZRFzpkvn>MGe{mOG0T@m<*j)Z!@3I^ zYTMb7Dm%rBl6^~Vc4towi8iXYcoDoV^N?_)n6DP{B-Sm%Ye=+qSntNUh?ja*|8OkC zwJ(g`xb`Nm+=eD{t3$=a_xjEZmGk377SLBHC)m=)3RP`cc=TeI=#vTv2Sw3t14 zy*VxX@(m*g)+N5Yce21q*4z>>Y@vIN{b0WAXt?~u)SVC1-}Y8)Sh^>`^vH=zT2?8n z%Y&A6=a$Jg3_M~f+r4(yWz{_}xssHoK`QyADSbY3a_t+UF$l%~hThMuV` z&-3Z+8rh_{!MxGhqCdr??{V5p)gyrqS(K%lZmH@Nn>~)hy3Z zkVZ0}hYpqNfToL<^=gT^@MmP)>3TtdBG6WJBb0gJR#Hh#xh8p(F_(jb6@{Ae96e@A z040oeX@1jOCVvS2%ElL|eFp+UtP#VL+zIsO$EM9_f)a+VbBMEAZbl7=hk6^NJgO*m z8dH;J3!Pe1*U|nRdL7+MPO7;`^PAK~EY-O>R1Wt2%bd!Nw+05~y(=7$TvzWN{ z^kh%SUy0-f*C|OG2e!)82+K6fOr`xeY1~v7{&}2txW)pS(*D@4bauqg5?kdy!(nm< zjE*KCj?2t+Y8OW4h4Lxsm{Hu)z@$P(yvp7*8dt^ z`P%LN>DrM@W${yj_^@9@=$)$5)jjKp)7A&Xbmv4zJlfV!c&sP#?qKiUo`R$;UY3p; zHnJuC6>~}q9XH#(sxOG^-IKtNas2X9XP~3Q5K%3#X@`&JkIJ-1@{=dk>QtG(*)6Ls={U~Rb#4BstO}dlbraf?}EsX7U=@$HQJ=C)6;Fp_nK8@C? z2GPx_d;kLSJv}@3k1Beij6n+JWT}GHHEL*i1r^+=>4Nat!mc)nx|#R#kQ)pA>RRR( z@PCne`u4ZT=PacRGyt3crt%n-07+=n4fsCjd=MJ=6XM6 zHtQz?&w1X;)+d3j5n;3nB5Tun2c?(fpP_dG=pOcSsc>v?KhIP%^X5NjDFNm_{5$-j z`Mg^7&C-!FDL$q89RGY`UgG}NQM%qKqPTk#17AP&U*o&jiFw^qciUv#8uMS4uB};c zQ9#)>bsG&1H@x_^OF)NypgkMO--Azny;Y2x78If&&~X38HIbz31>O1DLlyH)1);*n z!QiCsOatqY1I&Be*se3R`pY=0dY(DyrYl)@vpcN!irkT?{#aWwQbyAbX1`>(=*ZOA zgCc5&ry=F0ivu-)8hV)`%O&Jkbo)y0FfLiw9^x<^{ayH;Vq;kC+k`qlXK`48nLz@c z)+P#HTo}6g>t3VtF6rzzLGdZ##WtC>?<;WzJnELG#F-MG_FPEu-^tT6ydo>&d)^>c zcBM!bY;B?qc8UBPH&UqqJ4Xk)3Ml#;>=)yd8*HG3TiBiQoJ}KENJo*R;Lfm6;yPUT zfTOVHshL4cJQy;IlV*S;*CmP!yvy|zWWq7o%h51G6(9FXt6NY)CrDjqde%`oXzI0f zoJ}Y1%GaQk3y@=pHXuf4I+bF|utFS7qB%2`P`n#71(W=o6>u87HU zO^r4R{{5byQT$3%=I`?&%E)d7+q}A=*tV)gEn`4Xfp20Rm~&*~W5-W%n>|*UTTENt zWM%AXcQ2{Py*Qw>e@4ZH*G*&Vw$#uav5xX3@#^5UJo{3=OvN?vr7qSh#R_fKDO z@~!SSgaj?=*y6L`n%%Uvqvp$-d3J8(Yj3_v?AiQJsc@Sp}&wQaGW0DeRrlG8K9f6C6+|d?$&i zntkc8jE)}WQ2osbs!D8tPt1k|v7+sqjZU(5jf*t48Q)VfeU^w7d?z$Q#Lr$LcpG8v z)v%1YE;XduVVNkLKd*CJ!|5g8nz2~M`NHbod995>A(X0dHEpTPc~7&i)Q}OFmoNP1 z9QLv0zO81g5GMiNCUc7PGQavUXXmmPg1o23wPYM`d#X=)3Une<(^ zNBslp^p(H!Z!R|@8He2xsp+$SlIwhPVVZ!>a~=yU6yd(WORVy9iyaDz(nV?)W++^K z(nS2=h+4!aPW$ULI^fx|T{+3NYMD%5#eBOcy_E4?-CGm0JMeL{TV@Xj1_i{6 zg1h|TVLvxmzN5Gr)1D2OZG;Dl@w}mo0OFK%`s~%hIKft>C{U~}E3C>r7-FlG3UdcR zoF9^KS-vxco5;8uEu>m_QjP!(%!~!L%8aS{-5EA{1Z9~%Wl#8B`Ms7iY{O-=2e{2N z+>dAdc?0hG!Zk|54JK~eYnuKr?XJ&u@!A&aqTO|BE6*)Cd#F5eZ~o^)RZA?1; z(8uw?*DF#A1$_>wX>&O%f61S}dAuoV?+o)DQJ;o6YxY;zy*lha<@%W-w-=)Ag_(zE z#@`_R8`pws9SfA?9+;C+jg50RR>Zd6Oj`z%% zJ|=&$vOB|5!73rVh|+C3dwh0iyVKSr-72Tz-f?N+$_EceSLgVA+mYW}xO0tZ!0pxF zm>XaFetTQ!ytUasX9wG+=KNB6+Hc;A)m=lDmP{|bj_sjeaW`z7VqtdMk`c0lH${)7 z_?2S1nccOTb=Mvwk_=$Ss>fC!cCqdOB&ImhnMgE*?CWB>t z`O8Ar)s+t(qtF>kUlewWHeK_cNx7dzIqMoygAZFRbQ1)dylI#@RBFZ6(PGB07j2}( zCfVKUP>=jPU-#>(VKrttO(WdEw@F4=G%a@<dcNv9ncHsl5MMGW$o#z&U_O;t@>e<>=x%YKd5+6OgmB*m{5hTvxwgfDO1=WCtTe$4Vx>~4#A}jo zGrc?M4&_?3ui0yq`CluiYTQwl6(ogH6-zrh^tGOmerc7x87hXV+tl$G@te43j#Dq~ z=x(#TOm5x%^iHQTFaH+foYz9Ck9j>SgvKwjtIy;M0dr%HtQGEv7UST$vKv{Vmrk=!@;M zs~j1zzul^z9;G!h7Lj#yQ3U(u6Q)J}K;%^`t2Mo>Rzt6bgopKtwJA4GW*qw^%*6Ft z+Tj=Z*Fs|G*EWSFaV)PE`u-wn+&!#zWG_K~nxJJW-E(2jF47c>&+uQEmdtK*n zr>pj9w-3YJ#NvUyzsRzdEm@)&kwuBjX7zlNy{oF`Q0ufdnTe(-ykV$e zo|A)5fmuBJn{gfOF}9XIMYBH7wJw>@B(MsuZ5zovLJhF3uDx#CWZnuZVp{dTpBH?l zc_1sDo!2x&&(Lv4gkfsOLQmQdA>8TqCJ)oP1ZCNG|9VP$5N4A(uu75MU2-vUM~;Kn z(7v2~G-XA7w%eOJAeBRhA04216l4cz#9Z;A=k_Oy=t+$BX{Ki3{%T zO%w3S;&nTZh}i#3Vr9zf(KR)pyVF|N1rN6uQMTrf zw`!J0DraW5X$tc6BQx{K#(oL3!K`AdZUuE7t$_$?-FDi5Zkpb{f^?A6xtWD@z3i}X zN1r>#!lJ8n1_~UOnHZ}XEOsdPwU=t0QNSO}4$KH~=Dr}%@XR_ z9Q(PN`@hMRRJ-z_%35QFqr&r1elHXjolbj2tQ$;6`%h2#TsZ+)|@_@T<1Sm5BTCX1QuTA-aFYs!}81-9@Z(}SZKMiQLY#0{F zcX^Sr`=tYm1vRF1Q_{J1gX10x<9DC)%iLj>emYv=5>JcupLMBKO1hz0y?j`dyndzu z52Rgb8bOfGbiYzhEX+@hu-2+q@B-uginKjGolhPdIUHxP!NZ2OXPehs<%r(&H(864 zK2OsACDiduNEM0Oc!Bt)p(nVMYUQ)Z@%pBp2!yEaO()%9ENb)hBaR*Tw7xgBASv10 zBUa-J--e#!`0shV1I4u(kG-1A4LrVzHL;J};xV&=`@`W6^!#7qyz`&*uGVwkxFhWo+L_&)EK&*q2e;zSp2iK3%=? z!QrCdJgqT10>vbm! zT<|_lx66KIn7>*YQFh|s8TVpI-Lyc#%kS?YAW}o^D6k-kJSYlT;6uMKyrN6b%QU-KQA z7V3Y4W3q%z+qRJ1L%S&yZ|pVQyW3VuTi_g8+%VQAYkV`kK69m3A$yHkN#U%@$Phil z<)Y75-*oG$oU4$#L4Qn1# zqBnk#)GUqJQf>Fzc*91zHqFGu_I%To!5+qnDGNKhS7z~_kUo9H|88!*G^{+8tm?iu&JW$QSlv%{nFA4o(=HYQ+)Hj zx-(Mzh3CS^vPYf^49_{zA1B#tqLe=r%#GhPaGYL$D=aB)>&)>NRzBP6HY*ZyE%N1t zNwI{01WRChpI=Pt)R{SM`CSjIfQN6(loDtIYUHSmcgRzvDauSy+D_gTnRV2%h>p3T z!W6;&y*`dPtv`+H_&l}>OW#U7+J0?!)ehyfko#N3i|ZxR^{4sPtuX_*m`yy~uFTrq zf=g&3bZ)ab$Wq?#cDPXJ5A8Z*q3YZW)1c!1O5dl{%Us5^hp#|=#D z?Ah0D7qLAeY+y>#RH{Y8Xo$g*r%(H8M~dSL9e$!nk8hV4jD=nkEhin=Nw&^WPmr2_ znI4jH;2i&&RubPzsSay&18W97Jh$ABOuBwRdM>E&E9{`>`CimLp0l#LDLQw3$);~!IbO}LH8Je2GN8=)c3V`c zF;{uKu<|9_zuoiuq{H!V&5oIsEHsYnmB(TKkkW`v z)9yAoJ^S^%;P4^Wv1sCN<%*!rtP5@lxR;#%)m3jcv>qx*JIsj|eKuSD$C@)+PI>cY zu1xg4AbDEHt~MX*Jge4pc!SgcDpTllxK!fcAm=B!Wp!JTpEYeD|DQg~kGnof37=Wg z_j7mPJnwARSSRDyilc(@UuNI`jQM$&Is7T3MgyR2dGL2RFert#C$s|p|M2wHQBD2- z|DXtngp|@CA&qoM_Xbj9bV$eO?v@&(OX(h?yF;W)Y9P{`(k-CB`TqRQ_n)2JvvYQL z@9VYuj7L0C?OPQ21jT~*$8rBh+e5uU!H%fv@qctvyw}m2lHBxJu&q-*ZSvEWKwI-z zLw`7C6S#3I%kVdZyObfrM*e9dRsFdRaA5gEtV~}l6 z6`(kd^r(NU$BoEX~_T5l~@1vL!;#s&u05oCVPp?0?i%aPW0)8a_hjI$jW z2e>!TU~pY#j>zA~L^K`8v7j@9{@lj6VUEnEf)klSb@;%B~9k<&^!=K*ot-p>kd zxx}&l6c6`orqk9`QYTci3p{Yqc)5=R*;e-OU%xgrj>xLpj35bB_q~GuTAj|#ljQL# z_bT~(F&z;M7uiW3r*o6NAo{a+yU8X(PWr|KOPFM62dj&9EQa|l!Xf`%03bVLO;a!@ zZzh9CglZaV5CbfJ#p#1)WsJR%Lk-9?4d+Rp&Yrg!=Eg3u7nkSe;{-5K53m~HjN@@0 z@Bop>d0f^AE^oNi^L(@krH@|%`kBYFTj1nq=%}*zc{tGlrL!5UYLjGfE4!LIE%5?i5%MZZw;)Xa~y!s0}*V>F8rTN$(Sysab^xRV^6u zONZ#r*{DafnRGC-drEK#QIP{W1nNSXlRN>t4;4D|fBi5GHR`&Y=||&oK_Th<9qR1p zYiCNVk&LIdz4Qt5#jdQuCdl>>l26vfPgJ;eTt$%Sb1g#~!O;Y33U zSFKT`Dg37Bsg_op?HIRkX>Tw@YA2=~vzgwC@dAK-RczX^HEmB_Ch$&*9U#na0u@>_ zfvQQprkf_+3xSzZ+9QMe=m z?MCC}guV}+;0%$fd*=_0Qt(9}UTLd!ur(3B`%L2BORJCh zqX%ysecHG|pydy7n5^p_1u;9I^8)moH_#LsLIh?IgVt3MX!V1T5o#}Ggv+0kAK+<$ zwaKwau6PG;0!>d0-tm$1?s*zxk1J&|{+_j|@&kMdB(=l^ITJ|gJ8I8=UmmQwypKGs zR1p+PbLsRh<2J}%+e$n!0DQyleTU1=s=7ASYR9Jkas+F-YS<4W@$uwat2WYqYKc(& zuCDfuD!$TsnA{Fo%uJ1kI996Foxm)YC$kj-&Xp~W(>I@h36Hd=y|Yh?2r5;w+@E|^ z)V#0n*VCimctu{qh8Db&+G9G%KuVNkkd|(S^n&h&#p^e~gnO;(3*}e`^>bcWSq+(? zl{zS!1Daxm6WT4}*yd&zL-fA+yfJDSs-zUee>x%ulb4?x-!e zG2M=Sm_-9d@gr8ytAecyc2v+#G$l_Og5*JRS7gil7?a^KASEd!S0i=b*<>%w>6EZ^ z_s;F3#eXaq0g?u~lT+zLsF26W6q!JO|1hp0mO5;P`Ra@Z@=V8NmXG9_>K>C`k{tNy zi!E#PVhyuXjWbDh@%f*UUp&2_VJaMCf z&+c9O_Bx53KU5WuPG%6Q7`d4cD}=Jivwz@(l!P)0bz9zYQODFzd5f5bS+|$C487 z+t3_CkCEzjq5*WQa3dGKDM2)!LPYNg0V~(Hin#3Oh@>@Zd^RQ=WaaY%Ryb2lM$A$TNJNo30D{$IIZOqG@v$D`eMtX+ zPnoEe!MmiL;z^GSMwYF&2ypYLT$OIsyNqP_E|e!$Y8ULNDdr8pcsHI2N4yhl&~E8oAk^@$>*0`ZsUAo^W4vjYnRe(?Uk=BIdfQ?&5rK0oWy7A zbH7z6dVX3TU>(pfb4gkSy=~#_A(_RQwat&yNEQCAWy`sxD#hpv{i2|(Q@XKnssB{| zalynIHeCVhph#$|*@3HuLmJGAhxWeZcml_@Q8mh2a$l=lz5_0;$<+MghKbJ}aE`u! zW5_o#)fUHwY0x~Sooii!FEU3zu^(`q2g@(T0g`V~P#Jp3wE1!K)X_XICQ6m$&bOq- zh9BvWFwHyM&@d`e+m^gsX5*t-NxAKq`Y!R&x40^nXtwc#f})y>TCA7Hnr2M33y4+0 z`2CEA4%6_wU>+5i%h|gCkXtH3im7Lip_#6X&vU_uk4jOrEuA1H9;VVTnv`Ta({>N0 z`6|{}bYt)z(&x+MYiFS1VJd;wF-zL0&!DHhB(LyRO1P4I!46m=`V=M^^lWZ}Bv;nP zcSRk3k z%G0=zKIZ0ip1nzc?Ro^a@61LjzI1+~K{0>(^R7+7un&@N)^>_`>R@q-oeF(nL;cKo zZ{!Napedw|5#?yl4?$5A=1^d5Lu7|f={0FF$05=JQ~~vcX->sx6t1K{twl7k>a;Le26Y|;ngQLFNdzri2LiKIc^=CAw|XU)+au$>w* z7q@Ba%DbLy(xp$}(OOW;q-H&NpSop{bES*hOtN)F#DO-}L1BW6Z-cFr{<@OO zO7ErhGrSL>a);wu*i+|tfOHTh8p8BQb)VQuf2BexwMW2`^K}GfVaSF{7OD7!Hgx;B z3}4(Vk(ZoD&E=O3G^>i@lhMe1-P14rGc8+I)m5mcutX!b?H+j3K{;x2bShjg0u!&- zLfdx{wv8vI>_&-86>n~+Bs;l1AHn(w9m^JM;72GUC zHl(XZjR`9XUkX4(Zz8uXhOSG^6=XaGA>IY=ST0^ehqW}P0p$i}M@45J9<*&q5o4T11I$mnzee7tL%<>!3iUC7-A%8R>PU^mHrR%I}mBRwW&OE;PAlMCe*nN}uM_UW(M+qvuq&StDF z?YXSHYZ<1bK^l#Ac0bKbV=ENc(%8G4w-e;|cu~fVJbL_EC<6*0%03k(D*NXW^1sIS zKS!>A{igqp{;xhlN<|s4RHoZ@hIYpjwe;tYu>am{o(NX&Thuo+rb_6*rZyNn0@% zTAY4csZx|DD(gG`Mf);5?i5wrLEFI_VEc`FG?jt>NA!kS9ji8VR7nK;kZ@^dPO(8v zMYt-NS)wjtB#R>3nxHQ%?#q*ikdHgjk<)PohcCNb&Q=_a>PlTskl=1J=(d-^h)|GA zoFrSpS%X6a2#>y2VI&F9rp$I!z(x296}vhTT6GjaqJ%2AU<-Kb`psd8E5}CgC<9d* z|7v~7k+`m2cchFpkNmePwMw}e9x)dH86V!Gc$szxxfeqqDQ`KeE%nPwgCtah!1(f% z!1!{fAOBmOh=P*Y%wI#^Yx8p}tD*ClN%f^T~ zTmn~GKN}pkV;+ATIFt#$3L{SKqAwe&;NB=q#0)f+=a^J}L7GSoxFV$i>_H6t`1#xN zME(W(pIE2)|btS9DFy&mnML2bU%TTZ&%JIcUN|unmeK2}YNj0@5 zaYKVmon;*8MKB(a(VqCm5c<7`8Ui?`VK~qII?EPwM6`cRv7^2!D35-JaiR0-fHhDr znPwxCDZ!-;6IWbx+Sy6BJ$0-psg}k(!3EVAAai9c5rVCUV6EgN^Iq+g>^}fF=6bcV z4MOGhlsh|wCpx8c0YEWL|GK3(TZEG6Q{b_oWvEMHa6lP3C$-2^1|(RL7o{nP+uuA% z)EX!E*T)&Rmr-U#5XlcvRFLOoxe7KaT&u-SJ@hW1GL%8}e~CkcKu{Y*F#1O`$r3M> z2@D(~a?ENZK3FynQQh<1@|#URR&b8-si}@i5Wg}?TIzkjI6N0`%#H5OHnt6 zpc`|z1ry*Vx=NFB@&4cPXUbfy2&gN_K@=4xx%unGJwB&nwn)nT8@-g2nviu z*Ny--tY(or+rqJ^neK}>`oH_*Y*zRN`8oPk$u#E<>v8<(-(x-{VS1=|_b}wBHGuUt zmD6?U>?p%s{RpE{k@0)%B((BX9kWgmkkOoMPz_?C8SK0SR#4-KILI8)F@)+kEaMNe z^V%!2{DLqyFr)}PkDlBjR#pDOuy)R}qqZ>#6gaKnWDH_|M{?zE3jM+i)ri=fpRs6E z6p>X+FGF&*?a4CQ!Jtuma9NNTXu#J3Z2HGh-eRSA&~aYdI#Yi%KF75m0%RY(GBsmJKO6-@2g=HqgDv5$T zJ0Idy5mHr7cOtvIUP@**1VsB(NM$#dZ z3=#qjk$yQ6WZ*`J5n~y0vVa5Ex^7?dwzhkug2XFlIwY4zWVuHi z%W;qlj>dDMu*Hm-Lys?9@&vCpzh9~1b!oqpUV-pa{_Pf!b1!C6(!8fd^@Xbl=_~y8 zI5v&OHrkQwZDQcJ<2{lJ5uvz_*@hJKxLo}ZT5AmOXzhfCfx;HR`7R!dl*LoyJ4NC7 z^<(e2xY*c3FMWd!pGE1&ZYsNGg_6!T-E6>~urOwc?$2ab+CKF+qa@)q@TZ-&QBULw zZJ1ljUJla9mqGss2(ufttCM za>i{bv#9omyMJQhN<%?Q$XAht06fsVMvynpCA>oVx7$@ESd-TwBlP8A3f&_sBPWha z*j1U!T|#_fltjvz*BjwPFZ`=)Ktt!8aMxY}0OtG^h8ziYjbxagG55CE^Sr01Ycj1F z4Hhqt1(#+Q#Zu`gaJDZohBQ<8%~fBAhSQjtQ}^rGjJe=-bNBNKcPt$n6|Nx8oBd)m z9cK&MqiPrys@@uTV(#l=dvbk8r{-Cym>=aJaN>W5t*%)W|&um|*t%%#nf$*QPROtbS zmvbuU3o1}1tRVG&sM9BH&tkN+_6>3$~gjM4s`W@*~fn*yw_1|&Wik)mi7p`xQ&YFjYJoEw?h>_s~(&0 z+m{oPkG?sI9jbPh7Ney;Wkma8v(%J@1OurJS6^QI9AUbGo%WuVVTHv-q#7lLy?KL* zUP@Jd`%FP&A`Yi}c$5sJCeqZ>McS4k7|`i^aI*BDbhN~v2<yGI9>xP5$MxmfufJ!;I~#)dL=xiHWCsEpjeSbYG|uya(8!*i zmWGiIW5g|K?!tVH?*5`BRG?~p=dZuy2+@BA4dgAkWe-X;GIG9T1hh{0`llTXayKdd z2)X@>7HiG=U4N(iVEq#=iUMaO~<(gtK}vJ|P)WrbY>0 z0io}od?3f7uU-%@s^T)Rqke7~3eYXj*W80}_p3^myFl>E7*geG|Dt)^2jD4j9ywlY z0k?<$qKRz@PfSj$g8~mp7j6^5YKe`f@~q)h#FTFPDmg0#UW<(O@Re-5hAJT#eNaH$ zGlB6y?Z^V155Wi&(PGHRfK4e^zlFIuGPcC}N{nD6FGoH#CceWDJ(yJqhiyMJ*_R*S z^nwNGLPsrLCT&MO@)RRa(%lhHgQu?4435=FXC{J0(t2_R;ZoaXOj`gL*hhiyQH!7- zsqUdcl?S9P1b4|-{@@;Z6Lc)Hc`9HC)$flx@|XAcqUSt%J86Q?8+)s1-H@^L=L{Rw zW~kLnaJCh4U<_S0>$mx-Hin(g#M7h|MW1E)F8J>2Wu5}nI>h(BaR200+5-*TAGBsjm4Zw5F1eeS5-F*R9AJDCQ5;H7eL^_|*LL7qBnZL8pN>sBt zVEv0`-PlUonrx1`pWOrD7*}EVj!AO#-~`hFf=UL12+G7d)?@=^^xk|@iu!kS8_5B0 zL%3?qQczyvImsQ~3#%dU%I^x)(pa6W(T4FHyW*wF6}mzXL!XW1*^V-Ja!+vLls!yv z9Vx}b-!z=xeLkYe-j8sIB)lw34zfwvyn45=(sJ4NM2NV>X3>pafwVtwOP4NPSPF3r;2!U z)@DCD6~%edS8B*|CvF6{ZQDWE9M8E(&3DV;t- zid@-rt1M%VYiWMMk4U=sxpZj5VWR^sLvWv>ILJ_2ghrFa?)|rnImo6vFXpv4#OLEr zFX|r)ld+_G=6D>?euK%ZEm0g=zMB&_8s#Et0q?sA*Zf zvU31T>p4a+|-hoET4Uq-D8uYHXXykK7qQ#7}Oc4A)_mk#HhkF)Oe`O3Oo}e z#}lveEhx`>>GQ|!8z^Fy4Klz+AN^HSXK6#)VTaVgYC}LCqncwJGCG`}QFnnlgsE}-2 z&%2UKVNaKxW>(ce#8k8?jiw-)m1+5jhEpveEUaIIDUG2=GFE8J#8)dNgBjh>tUf-O zO~)XAYV~Pf%q_A+)GrPUL%xGlk6BPuO^y;EH1=xYZ!0U#*#gfLuo2Dj_N}hekBX3T7v7N~m{oSGlNI}gctnJvzcW@R z;5vL@TR_jm=e0AnIE&tH@EUQvtRBMlguc~}JBgJzlyyoq&oqL~y)*Xq3y-SJn=C6q z@{)p_TGPNJxo1Si^4wqdFIeeQk1RfI5!vS2dOXtw3%2bIB=47UE+uXW(HH3zCWoEx z(Pi$XHT#AZc;`u@p>W;p; zgn|5%oUsv#iIU4=NRVFQn8i-nL`lI+Rha%05s>ZEVOp$2V4F>giHn-SCk9*|4S3!a zsWmE&VWt-9kQ)c4tk&O6q>$erUhz3)(Poqz=Pm5jFK)KU20-LY>00vSLSigxu7@s5x9qf`o!2B{(SIfpMC;Gd zBiIC5jN@9yKbXhCEkGi2uP4xkV6C5aGRe6L>OL+3|wP&s-FYlkY?a z7{ZE&=p?Oey<3)OoO&*=>9|mzqmldttu?{$2(9J>+E(|5Dyj8~RCPhaNrYHIN_^b@ zj@z(ABZOvl0w}Ixov;2@Rr8{pXM!I>vW|eO*!><8*_Z{@4QUEglHroxTsXJj@-?V$Eql!A3)9L7zGlp$TVxC~p`QjWrNZPr05Dj}`N`AVjC}VzB z$RF8tXKy&gk49*fDPgT)#GOrtP8RI4f)Ffx_F1>sr07(nMC~br&LM&9v7Ay#f9&!N z1TwYM(43^~(!}9Ft>2Jxrl+C*nEGTulA`-5AHKZY^q!N-D8d8V|!^Sy5nW_88QF-l=Wir?{^Y89<5o`hqf)5~CGtP~(FBlYOALQz%X+%FFGa3n@zV z@o(tZe~`}qA7jWxk@Pl!8;r4g>ykCw{BoS-T_fc^Vn*WfL;Ygje8|hcXhd$m=W==m zSGrOenAc9g~az?Le)=>3_O2P6s=hYJl>=V~mhVow*VV$;RzYk9`j6@?w? z$Zu%|H)~uxUXk(twC9ZgP$|5dl{9ctwxl62*=0nDZ-UrTbr7gR$6!L-HHHwu7HB$H9p{iL!V<- zz0KQ}Q7zCNylIOc$l;`G6>0M4`Gd9ghh$cO)8vnl{P^e@ecMvnhIEVX3`|3*6}EPR z1;0gt7DsV285OiI>5C0?q-d-O@xMN9-Uh`AUx-s+In1!vCHBz9N|1eu+)30>1g%bb zl}p;k39wW7HG6n%Q4_?5z5t6!U*UUINP1KaS%3688R!wwnoy z|I_#ZG;;W;CwNY~?x=Ih-)jCY{2}zn5jyPo)il`U_e`ILDsOD#U$pkWXmcfM)|{<( z$fIA&(xva+ixyR674uIVtW>-zyG}Cx)%bS!AbO`Ejrih;V(AuG zEDVV}H6F*pc7>PK3<;PBr2fE{m|!m5R=d?yhjqPnQg1g%-E3hG?sXmzfk&<9imJ}+ z1I0g0R;>xatu+(1T1kXh+JR4i2VWSz?kC5ILnIy{HdPi;FjsLrzySGxGvOlv?W5DS z((uEZvL(QYMXU@FHP&gfz?)aiIJ(5dzdbMuU*&Og&8wZDXT%oC5A-v{0Yya0c9OIc zD!PWnJWW;iSIx}DaFc2Bz^Go zs@>B7j8W&3gc>l!kg&uoqZjWR>aSOb!JoTS53zrOCLly=>n4i^pWdeo2?`Sp?)Cmg zu4tw1viyl-kJM@qgHiT=qa|e;VU*>QHEy-PW=a;nO)8Cpql6xff`~`~>(-}-3dP3v z&-*?A)qfsp))}a_$~I1z>y>wEB1@*HDzRP+ZKFFfOx(*>3h$pyq&21p-7{BeWqjB5 z4Grj*tJwRA5vDR)G|4o_;a{Wj)N~SY1DE$;G$c;kgN*4{KtowmEmhas{&>7gHuZMH zjUXpDm1MXmDV)Pq`HRNO27Iqd@(dgDHWDw`4YV|`=5WDlDayS?QAM6wCoDkJ`MAa3 za3f1Ve$3`wZ3#2Us?LWOPjkQEweW@dG(iJRU2>Id1uDog|5ykEZC!e+({@|-RRTDN z-Vx$=<<0t37KPV*oiT08z#OyJnRkbeWj0J2DECL6m;_2=_#*dx)rfGRbs3*{W1BTl zFozAODc8x>&6<1o59Zoz>0dO4RxX{}*{CK1)o%8qAX%-VW)FYdEMJXlJ?$Wt$2Fhb z5FOZ*-T?QCMd#x)CK{(0DW~*xWJRqqhTo{e^*9fg%usP8`7SAAv#qe%ffrJJpV zWwY@^c6NU0+{&3le#E|F`diA$S=$P$;KymjN|}8Ql~9}wco-;yhV1yGLNThn4H>f) zXe#0{1U~<|zXusypwE0FzW~v6vGy3PP8s}n!Aw8JhK!7Ye)1zNx~SE&+xB=W(xmix z@xbm$SYH_l_f;j_3F5y1u8N66b%*k5Pc)cm6+mql)YK1UfkB`BkSmsrgLvl@p~BPF z18YUUr$^xZD8Zo5<0_Nd77=4T$N6J&0v$u z$i-!`a&(^9?$g>LWduC~gYXqXrOyj3;sfzY{f(Duzuhbm=cOhtAX7!q#?0y&;KiZJ zJw-?h6||bw{SMHNuex%MksP@15hMSS2S%`~MYBFE8vg#pV3cXZcDam!L3qzMq0ZMk z(;V8{%Ov^dKTo&7b^1?uBXWM*5w`JVRDGDFa#Npy-Xxp^g~#dStmdQa5WAQApn(nZ z)mkm7-d6>tFM`b(-e#nxZXAb{?@gK86TchznyfBfSk05HmBAcan>$jhJX}B~!Z0~? z#PBX!S496-A(_#idt3!0QDmVP6CQE=Jc7{(K9|ds#mmAY2)rypz$=U|DWjq5=O| z_5CTDilC9OVlpvRywXgD=W%r1C6j~LfR{qQ!9wyxTeAtZSNvT0;;3)kCvyKR6DYFn zFPb8b6`FFhFg3zvbDXlkpuV1C&@l%z&a&8EkYRC%o5+L3qzEw^X|6#dBaYWEb3cr7 z-Djm)X{@+{A49BZW3*)<=>+U#Zo$43-Pd<-?i}CH`Bn>P&;53gP`EJXfLGeTD5snc zv%nMw&+#)9-c;Q49w?8>U~h%)OXwKp%rgA^^cM}ZO;9Qh0X6`Nq$ZpACJ|-dUfdIQ zyLb=YggY)mJ2o#A^nv#ktqxv65JBbWJl^Ry8vn78{zZz20>-)try$jBD(^No;o(&Z3#!znHf-n=M3tmsp<3X<$_)} z+IBHfebEQ(8T`dK={eye5RDr}Iji_be-VubCAP0QFCxCxJ@P=?+M2R`2-3(H;_{Gc z1qMo)Kd{6?@!SRGxb{Yljs+ZyB{M_>Z#GJdv=0bUe~f>0Y*aaFq|}UxcBjrLir6;a zIVOsohenTkOuEFI`+omD0>(5kcboB0_OPVx^9c{LZV?FGp&aA*K<2BRmIw!8lz@Jf zNk)p>#Z`1JztFd_v-!=B;Nb-rVOdo!X1lmYg?<{j+-x*8v>O-YykUxSq1HTjN?ogaXWU?DwwJaZwp8$MRV0IhJHU6 z!u}*I?uDN^4JTLB1jzR+4MswpYS2$Qf>9D>={@k z@_ujv-X1rI?j!BSrm0yDrW!RNxM55ahqV6dyBsJ zn%7R=qM^A!8v#r0^?m}J+%E*AMDh$xNYLOyC8IL>_5mwZ>4R?kt6nw$( zAT<*v1p_(S{enAk|1?S{f8IjJFeP!nD(m-ZB4z})3=pU&dd{`IQaKri5w+frxw`!F z1Mf^z)^NpOPeb6O?UdS9TSs<_!Npg=9{R0y z>+yDRdn{D_CsW!k2+Sp_D%e4@Oq$TA`%kxEh{ruYm{ zf=PF3vttLu|1rl<6ux=iyJs6(bm5gio4~3?x(HQy`#7~+IZW4`Lse=ryi8`H6@SUz zg>}sIfSi3Dop8DF1q^4|XY$x#Cb`moWnG)GzJcr=>fsIcE`Oz zeRLsr_syisou;OWYnMeFg4Qqxep+f}V{TG|xp|w}eXDhUs(v0lc^zwH>m!{8L3p^n;tR&Uf9OtmH^AJ%_x!ndsseh*X&2K zLp7k0#QvtpAf%2tBJiyt?FqrfAAY~OMd^cPx`+*3T6b68TvXPhp@X;F`Qx<)h8ybK z^kfhH;0#8EK8F^t+)JNv98i4Qm5i;p@Fan0E^2Y6XshRn3s4kynA1=<7w^$0(AubC25B|?R==h zEY9L@zK)A37}!u$^LD)EGmzNZvT!p0aauka;NWq-+!OH^EvMEwh)_LKAXQnGKGzIV zs4zvvn_vqUE_Y`t9W^e#nTA1)ziVY<31i_);y22t7MfMyB+gGW#>J8AhQ5F44EdFj zlFXd$UQYMtOr!fE|6YV^;*_?d8Cgkko=>SplMM+tZI!S=kN2%7v2(LGYve{qenXYy z1RLHD z8^uk?%_j5P$VgeMY)5eLYhnEql^E z<%i`yn+})k!hi182RBzD);Sw!J32;cjfSzGdTw9x@4C08q|r-=v$>f&d0gnL{)bKp zfb4u6GVM#jk1P;dFPR3X9U5u#^X=dhjT+R_gsH~q#NNCaJYd^ck!M+S4gSp=e+~h~ z99jvv%Q^*!*W6h!NwzJ8Y4`WXt}pTj%I0pwCu(Fq%T_i}A}Gs|{C%Yaun1m9#&7G4 z3p-XtH&~n?Te;E>A^=8K@v{U1auEuHbWnFqW#Y)#pMXn&JOP$nqvjfp+@sQ7)CMz5 zI9U^a;fT&J=`4jd3D?LkARY>j;*KQYb=c9 zU>1lAP&}w=%dZ;zOF~)XX+ohhP0C-ix=ioBB#D7vx&7J{2b5%eo`YlQx5bXR!G9W! zs*+3w0L>yNIVne~>;W?8fnl$HMUtbZ&`6)?kaBSSMT=S-@}rRlhYs(?X5(ymGAFAK z#5npz=#ocEq6_*lcL%3z!&khbTqMmY<%f5TgF5|aDu#wlb1MR%r*6Tf;QEV zBD{F@b=tk*U$jx(mWO_ttwz;}F`cAk!{Co6TEH3KI8@uGZqab7_igXvjG6in0Kk+D)@!`vgAu$v8qPE`Db5bu&aJ%nK_KWuvD2%%VV=f^vHc z(Jp)6+0TvSlh2r4e+y7eruS9!rYWy`$;nu2Z&}3acaqFaml9RY94aL(7z&*vk4ltJ zM>o`MdNe=}_gDuAybnHYW&V&#>M;x|Y@sjv%*DAVY$9%RzT4=O>6)lbKby%gE_FCv zd@fYBZFbWJ`4j{4{-dLgvMCnSiN1JQxwoXLE&8G-!N2*kl&gaAlcTC;6T4a7!E zoGQ@_4Wk~**uQAr?xu%Yjn4~PKLO&%lA)%}P2CJ5nYfBFkd6%Z>jTy0m5ny$_gxcd z@2eR|$yhJM`vY`D%Gz{S+b2OQ3mv zqQmgsV_nOPIPQUJItDHP(Y)jR$Zxbwo5AE=;V5TIToCqyY`A2!V61#cW>*}Ot#csZ z;Op1BV{V9nLHnZs-&???V`^r}`n#3>-vwEB=`DX`H`D%%2L+^klwlit-&rxbJys(4 z+eH%|_3H%J&g;39#Gk32^6Da6uLFzbFRqFs=>2xs!-q; zZO!dBt2g5Yb-bVb$;H16qNh)kH9v0F?~l}|?{uST{qPsfw|a4JQdam&O)0km!J#In z`8fI8FzfkRyY;=iRz+{VE@*RyzU83L#Xv&7qV`v_#{-^niGl6=w9jd$O3~a0ty>wD z%%7a{J~y@>pJeRy&Zc%=R(!~Zw>xA&m(`=wWH0fM=xidACQl;p;si|f8L*&@tEX~mbzE4eC~{qI}9xBNmeX{4p)Nj_}^ z;|!f~CUj4}qhnwyun%z2UkoWx%HOmU?*}stGtpaAzlx7}x5jNeN>0YsrzP9({ztg( zl>~ct6mH;U?$}Oi3Du7&3lt6x-KCXpl5&nv@MSKim4b>i zUE+{yeK`vYXaCR zYEZ!LyG<263cmA&Nrg}1K%mp-`?<1=EGJH}ey`Uh#hkIoFwMd{ z;LY#^#Qi*4kqW1}wl)iF1WIp5d?Z4)k`7s(56=@6D{^KAY>D=3ooLFtp5+jbyze-e z!6N50laXcZAt}Abmp8g7XrUGExhgHHcPw>fqZaD$c*cL37UsX5j2=ABu()N3f~mmy zJyg>b6`u$v{|Q;*(^j$PWZ+m~N)_N!7m!#CE61{JS$KVj=4e$|S4J8`r@eg&)<<7W zCzKx(KNq^>H%LB+KuMmFDpa~$>%EF$jxvh)!ql8b903wP;P%YHGm25Cvroa)PpmvT z#xxUEE0&N)At5rmcEExnFAV2ZZul>#MuL<8RnxSb&&NJ614Dj4msLXs+wkEl@imL1 z52U=Ry`j9U+aQh9!+qr>-nlnpbDO5^X#{OcUYgTzUV|lA-JR3X^yenrKHRVU?7&)vwF@7GLb zf%(rv;UcBZiuw?@$o5GU1^x8csFc^u&YP20Wv+^hDvc>&CrWWs*Cow0Sg# zKQ#c*izE9)iH|ZqzwBIUinvAdX1(d#jb*FzLeKl-P^rqtfwfr?lcZO@2oyD2jEbTe zZzqc$WXj$gkK0)d1;ophyBbZz%fVbONI*K+r@)Z1;G^x+z0K4@lFH zl%!+u)WvbaRVNPXFIv3Yg?auFcu8>cyNVDQ!X$SjpkLN^CmEtvt8HtGAluAZyVsic z049ndWX5T4J~mHIep^|yR2UR5BX_Wf$wKwH?Q-7t@2PBDT2|8!{IX)j>ww={Y^!Z* z9rNGEICYZ`T2Y1F&YIx+lzT3Gx?W;%G~Vl_&*f)2W0!n)j#<=%EPL?=gSzD;i=rhX z>=!E?5i8Y}wdCe4a>XI`WL4Jb&2dNd_Ci6favI~xCm`)j!cM$3V>Pu-LhI}iC%K_Q zTxQ24W!q+kb;1EJ_|dkA@E63}aqE*gU3GrNY|d|z<0iTutd%NhgbLA}t3#UPF@R5P z+0Nw$4f#GWN+c$cb#aJ1U3ZBhww!KdvSXhueX=Zi?0(W6UFN z56s zp!f4=`Y?8vd>{Se>=#Sijd{L>9-*&7-$r4YO^GUfZ_d`t+&AlR3-!vJuNQZquC8@` zgB4EfhQkf9R+Xa2hAjB5RuyW!EQdej|6$w;^w6439jQxDRQE<4p{`f$xaf^vmA9mZ zY>MR>yYI7%p9u=%l#U8?aci?kq;9w#5|DQe(eyC)`HM!&?ZZoZXZv#PQTq#5(ze*B zkGA*k2YBMGdP?D`XuhcUAf=yjoj$poy)P^6?%k}&(aM}m9A-cnFEynp2H_9>+g%I$ z8X}jF4r@~06yt+2TYd%D&{{dS=@OOJ<*ZQ(({$1PKhLTM2=*8s@P7JbPB04^IN3It zEfN;3e#z$ZP#Tk=6ZiGkN>Rkh$i*M=o%4THi%3v@S^d14y&L_@&d(_36E{ySF1G1B zj3Q@My`^cLmVKEr@lL~g)Wc-fUY5GArpw%{TlJZt@>YsNYqYL!T*F^9QML9B8A+0WBus;--^Z|BS`Bc@M&}QYM|Y-{<|^37T4`rsQv$)Y?u8cWiMsk| z_Og40nvH|s;lhGYdjq)(b6_4b{%o4VM)|3WlN<>2NzHFf`*|bJzkq38o) zVvarR{o_7gY0_=&TuBtnyJTvM>$dRPz+Phg2I&^>^rz8~6mZoMc>a++&@3R;v>k_p zFKs9)YjfspUYB1z^T}-bBzlOJa;-yO($7O3Gvscxpa>CMrHW4-j#cQ%TGF^w!kBC| zz3Wu0*OY-Atn>cvb`ye#(KV*=G5`dTqS^*v&3(;p?MRN=pg;+EswCsffbE;`nW?@0 zWK}&h!E-FE+C%U*6hLWL$ohbI+Tns zf>8i#<~s-av9QT=Nt}XVMn$_Pc+WW!xfjNYyv7Yxf~0XJNv>aeC=p@LTF_1l%b zZd z5?lx=!^`H`G4m#)v1eIMR58$K1QvTf4QPX|$us4xy}qp8yPSzSqqd}&dBX3BPG$J@88qJZKFE~ZorxWGt@EBjL9-c! zH%{_G^6~7R@&bJ_e6jBL6m)J|+M|VVpQRVx(lI`!X8Q_0s7dQ#fiy1Nhq(+A&~7TC ztV_E1XU}YSZs(X=FN43*ZLkR)CXC3GwL}V2Uh!vG2N4>6j`}UIF)ydju&J_9>6Xzy zqQDzQtiWrZW{zCHnDY@%vj?XPiC*d0tG|5+E zd|`olXY{(5xbGsvP%+aEP61SkYkTdMULa`0dm60_ms*?N`i$m+0tFRAcGzo*N!q|4 zFJq054qTt3Q-=-5Aqu(_DJ_p`l-(=2L4N+ALrT|Q1~wa8EMx|EoOVUdg6jI z@XcpFaDK-?K?Nbp-lrMfgL28!N!*_0zRG=?gjjvsy`4SNX)K@WVNOc$`MuOFTf}^C zMwR!Y_b@*Gu0$D&&?FSD+R~y7@_dW*FxRK@GJX6eH(GU=%$9TOM@}><^D?5co1cP_ z`U`r;!2Ats)~I)83e;X_xQmXKrfRv0IhV+#-pc z1IwX|zhge)bJbc*q7|7voC60o1wj&+K>qGM1vM{e$Q03V z;*2CX2Xm`4qhx7?ycECw-Kzi$ksRTL6FSex2PMmM%=lwCH$vm3;@nqd^NjDRlRJN9 ziaiTWlwtxiITwDEvo65+k0v^AT3=KQ=u}=ODVgb_@rZ@v`!KMo!cKmUIi zv;c#Gp9BJgh5~9a`oepPz(NW8KV&ZBgq7oZAbKU%2Hky1o2x-3d6HeSi5zIjZ8Vm( zRH|p=?qnZgU~Q(;0td=tY{F)u-pk4|v!PhHg^=~21SChXZX$*Dk@f=I>)FKFO3u5O z3luEnO@sQZqQMpikPo|2!|^XcNZ!*VdKT0o&8aGE6A_aq#|rWU)!(T%%6f#)9_O20 zLuk2y%yKA*PR3lrw&10)THnU_C%zsQ7$WdKX*QqDurkv}c0u2dH3GnbUT;O!O;g6{ zsiTPTScMl+_Yis#W&37n4l(hl8@VZ6CLxY?tJ9}xkHWOuhN+gGvw-|-%#`F}ZTZU5 zIWW>se58?XR#eQZ8_ZMUX6O;-*y;zwi6H-dQ-6in$$W9VR=zP?_`Bq=p>lz=1LHf1 zZe7TfS!*{n^p?)7w)4ZYm@W3==WZlrAD_l!DuiAH5IgmI6rmnFCu;tnSjwMev)W z=CT^K*skS4GT3M02-b8?1rLw^WZ}}!m_=8r6+{<=;egk#gl5&>jbFR`z{Tha)=G@t zdO-lW@zl?lSpCgFO9aUo(KBuuMK~{V`7_9Qft1uCjp#u`=!&c+Go!4nNVxFKnun$K zoss%W#*S74OLE9tX`DfdPd_0$3Xb zF;0Aq(3CNuOYEJ~Dpa~|)2olaK2>}(e!7ug&GIR*nLF90gJ5~M=xz|37=ae zhQS3g(;aCrMINEgk=rN^sK#S5^q9T%zGtTX*DsB%Q>r~gE%^OpvN!0R$(q?-00JhT zz{bh)naCFN*IV8*d~*75?_UB6M9Q4j?H|c6{j$WguLK4No!|Dv^#pD9;1lQ4(;5Kz zy1g)c%?!i0KxR$$kw5w<=WY&4V%8~czP&mrOgnGpWAd=blL~usN(%K1VtktN7?~Hy zo?AHMK3({lQM9{V^qsuXn@K1Yhtc37m~cK>{9`P2uH{P4VW3uv)yz(cLb*~Nh(XXY zz(V{IXVgtp7Q3OPPYie?wqG)~<~$Kd>KTf*msx{PSPZiPTPZ}w%F^%k^w-iAWus*& z(yG4(n@K-dov`$O%}+(Jr##P#q+vg=c;PQYtPIgiC;D4Juuj0H(U;N|pIG?WoG7&z*81YP9^py4 zsV|r*6*aXyCcA*LExV;da@S-S78OAJU;9p_Mld%Zt90#;*oIgzXS z`IM#)UK(u~g(ooCYs2=`T;<`>Tsa@z}-&+%713Pi?;$>%7}DQ|Iega_u?e`GX2!9NJ+3mEuE6NsrA1ECh>}gGbyrbtsWCYBTI$T6{p&1F6igc zfg6Arf<+a!cDI~Cp!FS$-dlOGEA)}9X;3c=YF?0XbVo{NiIP4SQvNMO`+bPX zhbZM6RO(2ozQbKfzuli}A+&jdwOkR|F)pX$7HfD&Q0>#i+QpyE!Q2CPB}1T89VNgA z{ZSdi6gjVw`!LVTR<-HLbkqfl#x6E81_1~^4@zi;8XWd29?%GwiYx2SB?{9YZVoc_Xg((OW13J*A!+SUGk$M%>_ zrhm+8&%tY<7R)U8%%ktQ-mLvgpuDP^sw<9lHbp#VjRNjUy|%cG`7U|!+}K*CkAf_w z=}|=h^F(hl8vm)@`j7TwS2W~C-6Raw-ppj{1?y;iIyJG~A32&eV}B0(gQB*eapq*7 ztdoH~>_SLE?O(i(`DC*6^`?4$&!5oJ?zQ$?nT`;65=Meu%1fxn_d>>fq-t?i*Zn-r z_4X*AKf}XfSX7jFLC$rRM%MCZDrvLdE`8;G=i$J~kiNyq2#^uS83W{I1&HgnhU`(=^U56gePmJt(IQ zb1$hd#;r{OqcktY;Gqw0)4^X;{cMkQ9|+7wjg=o({#hJ%n@yPYo+6xS%t0Y4s!3iv0DKmAPrQ(Zaho)-R2`2>x-ATu_Uu;z4^;UqrkaO43wA1Iat> z&GO|jXibLO$A9r^!Jbab^FtHpO;Ke|Lue`2o|#ne_f@*|<%4ZGrujI2mDbl9FOQeX z-{0`0=kHS~Hs)r`C*kyRFwOYA&ke@^$rp+*m_t}7WPK`XncO~jr98$17AHv=?SBa< zEk+NoVm5SE5Oieb=0o$tXxsE8l0l5Rcd3<`6nA*UDPe4_aGoD(S^pmTs*dxts%36^ zI<3Msb?m{lxP8=)VW18qra$5MfW=_Qy+(33IKx3>|IYF)Q8VszUZDRG>M}IP^R|O6 z{j+cy`sL)&Vu@<&G9;;$=rlibN)7 z$WbG&CEeFQshEHI@cGHQM|CHJ_26vZvX?)+=RHjEc*`i|x7sxGP1%gIta*t_B(QDi0O0aRc`+v# zxoLi`w6s)i4G~ug`CuXe=WBDV%y%>r549d_DvmpBH(F$8=kvzfy?gg=ukX+pHPx-vsuRWO7`&su--9hvh*AvNoSJMIo#Hce|yY=HQ zL7~)@`mV&fzvTN>gg&AzyJvs7+~%QVnssi()y)ISqKlRtKjSNK zCv$H`&$O+cSjCqUgFBsZ!yzN!AT75k-?TS1#8dDAAD+wZ+q)Lv*~bY z70t{3<#&H)s_UFuA3dEcg{S3)%MPDyv^U4;e^_tKRe8$0mbSdw3g$4e`10jNxZi>- z5VHWyXf(EDR-agO%n)jYATJF{Aq zQkUwbf3o)+k4ITWHN+6Bk>=0vT2c|WpdMYFU$2kkx5`>Irh?EO({j1-(_{|-tWS!) z!A-xS=Qir^mW9{CL3tg}5OJ#TVI&7-7z)kV)#b9<%fFKxg(e&lO$fatWFKNne{*tr zGC06wqZ}cikP za@7l}US_Q+iM?C58U{B*D1w~aTCBmYjP>p?Sy+nM1CNHHr-`7J)g7~9qtx%Q@6GMp zjKg5i4{nFH5jsdNxW*}#fsiy|#pxSIPQxQpesgV3otJO{4NZvS)9JU%u?b&zV$|It zKWxj%m#KseRozujeqCvfyaVzttpt}mejivoma%-)g(8kgtpIcR-DAtCsp5(Ydk>o8 zRBYlNF<2O*2m>(p-Uz!Uij*i?qyOFrGho0E5od*q4y_aeTvk&OY~6Kp8-D!FovP7M~%o^c93i8SsBd<3*VhV{#btd8q=+G-cQOZ2aLWZ6J2%@L8nB6n=~ zk;`WdW^p)82&gDqvv^s;7-Mz#Y3#^jE5*y(zzza~z$LGQ#(x68van|huQ5!!d<4~$ z=OsoP6ROEv)KO>(B@(U4mdrL+n|PO(MmC`#d7{&nC7fNH8bbl}g|g$0{4c#yT6#Uy zb!f+u3(`N6rbuhjbJEnhQGxu+inyx*BNtw)A+vt0)8W4iofMt17fR{cCHA{c+NCk? z5a5uROWnu2m)(gOb0(jT(7M`Iz~=LNoS`ujx54azzO#uik*y5aYDC1$#R@;COAFmH zUr%#f%0D~Ql3>^<%K54lJXle&a!h994Xb?9sMg?|L$1wXC&N)dF23q965B!AS2!Fg zIqaZGj4a#Y+wlz2IH`Zs#>t+jkKKWhEa!>VMxKB*NB@JOg{LWYS5ND>)xaXo>|?qY;Gdoi{iguzs2xy7qw!gru>9N;^vpGM()|`5bK5F=BX;$YhXLm^eM8s z%533g{ZX@E&Koa^KQds533vGQ2r37wzheBnMtgU6l-fjzOvA6X8RoP|#YxL0{TNV2 z-cVRPnHZBAD^Y6B!;a#zjdT5`s_w6~{9f}SE;m4Zo7>xO%VELYtd-rw>kW?B7>>H1 zxs~u>L3L(TwPN0?`ePc+2L0$(5u2racekf+zm!-X3{W7%+dRG`tDEcm*=qSkm}C~~ zWgK|aP}-#T1B4qJ&6WKF~M7gZiud?@Lr#y<{vap(c$bFKOP8kk)rS44*F2RX4`zZpqh_(tf!YtNde zZhmFi_2pjz?QkN%85NS2s_wS)=`k<}%Hd^Bm)X*k{8hGdAh1|~_R_iU*|B7l9Owgv zXt7H53Q*xVfQ%?8JPFdyTHOEoYq9$IGQLYTAVlRrbF894{Gq0@Sv&JtW6kW!&wrN8 zGI!7X)Wl=~-A%`|*~*B&ra<~BO^|H2r9kaRi$8&i3<6urWjg_cwL5UTh}Ac8bxJ1{ zN6%*w0BXcit6}fIB&$S};_yNFnsFDKc#|~DTzY>os$!eM+{jOQ4S&xJ ze)lnXC(~6~m9C>un=-%6c5~>*_md_!o5u0~IG#Ul$%*gWb1>B`lHvSQDc;ZJWiy?2 zh3Cy;rX5nOH=`2j>8tX3z|uSiT|yzLmeyUR)w8-^uvNX}3lN ziCvm|Q~%;MYdtz$`5o?2K$(@b)!YX^<)ax&&6V;?7~R*AV`az8(U75Oy4m6}%aEa3@iz)KgEyk%Iy~%*2ZM=c!Rr zBl0{<^o33GNunMwUo~!<+;_db4tQxKTa00B`P|pJ56KmAEqFC}rZ`m>kkRe%OgZX( zsPb(WjP6LI4?=2N;}7PpZZwz9rol0s$=nENgZt877`IG}D1RcN4+&iXaa4@z9DV+x zJ7p>SG1_LP<`t_chl%Hi&ezb|9YmPZT=Z3n$w*LAzQQeY(n$WC)dwJdfDD(EXbXra zg`-Zu8o%{ATbTQ9U!K2RX-=u7t=vdjQcagtl(diWaW2emGW2HH{A2g~sM4VN&$pu{ zv{rfQ_oT0RNbdfR-OK64o(t5Z3>HC$8-ogys63-2jOhHu%$QTT96}WbXALWDs=~_Fnn$*Gc6h@-ldp zL2^Fw<)y8#4iz44t)tHOb8#v;ptu>543TL1I1EvCRNb7|s7KztmsNOC7Snp?gvEQ` zzk09(ZBIccF3w7_N`#3s`Pj^EofUmN<|560O~r=M^$is+4`BYGO@sVmLlfkG%JMHk zp4T{RqGv0?(rWWFvSw6LRS`NT@Hg}AQM{I77%3We608oH-4=4>RiPdErQh;vu4edG zr>@o0Il2;e;U^n0^UkTa!5#vn!5sS|8nN3PAA)iphJ<@K`^m*BcUcUjJ3DCy75;;4 zxFaV#z!Z+o+tmD}<63Qn{F-3z&1EiUw-#c_-;y)FpY_Gm;v;fV`@a75y+6(nrafi2 z@}bz|b*7@55RmnQ9a6?Pbbas(Z-42)in($PnKVh2TgpmlCETL}EN@8`m$$yC-+FjV ztJaD6bbtz1-&v!F#FBC&b$AZW{nCGRcaX|@z*6+@} z5U?AV>%j2k_=ASyttT1ljQ-hvCxS9KXZhj1ha9r(fl+IOf2YlLQQJfJBp6EeQ|hfl zzsDasgfjEg`wPn7%--{Hzx!hEG;+E2sx^M$W`@G+M5<6*s6ae2 z%28IhOTeJLO^CDNO8NPMlEe8KZ31T7*jT9fY}85S=Fh^*tvyRYj>Sn2BEi?+ zXu-MK9itW5k=aScS)?OrM)!5~nefT4}075M6HUK9xVG|DeQ)nz;nWRe5 zT*-~8-14X~Cw}+OOxZ3-_c4pXg49ng&I($vn#|2q!F~?i@)--${xi?`_{n$U#tj~4 z1}$P~Y$Mf^uQWLidG1KRcYntKx^v^aE*r`G;|Hru+}A&eh1mnPJ1=>xYUZ<_e9t*M zy74&paUyfb)(2vA1%OyH1FsAMp?K(jK*X$4(f@Q}|M7+C|Hm6KR44qe0|@`WR8z(k z+OO*3F9v6#*N1kX%ra}R24}T6_$i2CxD=Ly-a8(d;u2?d<{fJV1NEiu+gMw6Hi*|} zoOUqUu40Ml%S#(^s-#z6vwB zhxze|#uSF8#r6uo^uM_%!e-n!<<2yUTW_AG@}-yL)6;TE>qHQBb=qm>30}o>QwXO!2_6^A>{k}cC#vuj(lXxa?5=rwbt^D`tQ5aKNC7Vl+2*bATMO!lc1j6gyWYB&6qd0f1d_1q~_0LG+pIWQJ7Rvbg2s$VdMReAF61NCpAW(D&@`0+aY@z7 zYo=SNcjX1!a4m^XEmuuFNRwp|EeX^sm)&(MEfWnOzlKc{35CaY+5A`?N2r%85B}Y9Yp?tFf&e41+5ADYW3Y zDGPjLiiVHGh2xY_q}U5Z3hJNdYG@-f>4f%sa4;W|%eYer#vKGp%FKCN!oLr4s(~=G zBERmi^I$5#K9Xxqw(OlLkAEbo8ACUi-o?-m!bu77S%#UpjGPEh#3HFdeDiGPdY#5a zF<4K>FX??*hCmFb8--=!DO(HyEd~oevb*|o%-1i`=fAaOOt--w`NM!h>uf?#Jczew9neK#hGZ>#tLN0RQ+COU~R%<^?yy@UsW8{C4Y>yb1lbL;XciJ^zEG; zXvM9Ielzo0*nN{sf^`!rn~qG?0otOSVcq(KPt(Nn4Bg~~zq>4nmVN~gFur$Hs2FqP zXvzhF{X1(pXo`ZgP0=cM^#+B)=2FaMA16zbhL@tHw4P2KE;YPmLYoYx%;d*U9J{j^ zno}@*dpaFmyQb3o$^L|+x~kiQ5z8Gj~63X!wGDN_@6*6&(|MN^m*lWh&s! zM5tD_ROhF%r<`rVp%aj~abcFt9mls2y- z+@A5h4QscV`*@f-uFnCrNanT^@@^qW**h%^5ymItiU%p%~=Sp{c+ zO)3adcT$5KFbwTVNZ^>XT5W4k5p+E+5t+fwTP@7-j`A2^;fAV2w^7fY7)G*)6QLZM z8b4ah57BlBnEAZyR}B@4|7|D)8a(BAk|AyT#2_p8Ujjx}We{Z|gWvo}LwU1`RJH=K zydRZ~L3L_354?d9hCTC?ybXR{oZp1GRl~$OeIUnpE*zTWgtUh`NnO0xIvg1y^gP`L z$p`DssxE49y$oq~-kp(-i{^Rs`J9Rdit!td#2y8TK@j-r5c|@uiPOh`f5c@i_+;7f z&+3Iyz#DmfFV{T#sG2{$Cxd1f;dDWf6r-s?##X=>70=CyP78G8XvCR)OMOHY^w}WZ zvKC@&W&P@kE9kZ8=}?*Zb}LO+IobhkFAC_2dmF7 zV9jjgD5-Y;5{YNZF_iw*7360r2imMpi2v$QAfdX*=%{fg?$)+JvY!hj4Hx%6*XSpfzGrEB`Id$Gk74~^iO z-qgF*SCI`Bmt0X|KD<{t6b8vkl)C8&6$}N{R&5HE^%U5(=~T>DX(nx%ws!cAr~FG$ z#jX4?cUoS%#&v^=tjYCcaLWV%?W*h}K-t*Up5XYgBpenm5x+hN%P{MmD8cTigH$lcP zxc3xit4yG;MNG97OCn;mIXD|ql;t*eea||SbQhb9AKr=de)~_PfE0z7K=bNvFdW2v-b=EB$deXg;!zRgIg=9SD7T2G3&yE;mS@X`oDz?!*TU0ycm%2UuW9oHg-wRfx6kn>e4 zpPDuwx00DH(F%V{1<=k?g7v2D{PnZU*`y|tiQ6llP6`Rn$d+X1Je+SKOxEwRYnstC zx*cQi8USZ-dPPnwSMhDfs+O1eW^vu&)>ynU9QY z-L`w5jK@Ry00Gbcsb2v@1OR#Z{};0VjkDhhdI8}p+Fi&#Lc<0>+JxNKviq2Jqy?)J zs9MeeTScqDn8!_wb1KV^&m?mJFAoYB=8bk&1!1d;q9A$#KBW3s^AM!EfT|`(7Ukr) zMwiaSu1uk1Pbp)XC`LwdH_NgphM$==U2cT2(5ZW)q?doFBu9%GCj5{Ir^dxHgXY;q zSq*I z^$p^|ASr1}O#8vz2k}AN|2KjK1#6BJP^jV`!gA(ihQTXO(BkFO)Wa@R0ziW7A{7U& zv>Q9kj`28yn=XbWcAZf|-!vHI9a)k8 zXv!D6?rd|++qV2G0$f6j(IDiZz`{QO zjmFRqeKg`?0TDN^f>7M)$t&fF%dE|Q;paT~#D4c|3a!*Luw8_wfnrO5S>_X#upAN3 zwJKn6%^*`T(!uJjuU+4b@i}M7`I&%D-T)HBC*#-PxiSM1nzMs2T~+NRLFQBh+D6!D z@Ry-BW2YTluQL$-3IW^EGm_Ja?wRY$F7{<6pSquCp3}MYcmn`WmlQ z0f3;(NaNefhFvrlD<1fN!{{!Aa2fYEa`{uCqKVX!wXi8Jr(jMO8+;-hNVu0U6RWop zzgf+xo*_dHCsXT5TA1ZTRDA zdw$Kx!gMK`Ei5*DS(AYlq2mba7jcveLPd=;Ts&4Fy~aH|+n4PplK|KbK6&P0ItvBK zK@$R8NspG%#vGsM=T^|gnp?5&b-c;}gF!)W<>U4I^_f_PADj%8jizY~#9;D}SD)|2 zcJTBNEaX)66n7%*<_rSm(~Dx#))PVBJK*%3NhnUS&sI&duF_s*mHY!r%O9X2hctM> z+{9Dr`3#F#)+b37l64CKjzsP2)m_Kx#pfE5?hGebn~y09ghSZtq4krp=7zeFX+D`9 zBAt{#ZE8%P1i1`@GZo2ZRK}((W_u)tTN@@Sd4pNMpCGJ_c zhVXBYtY0-NjVu4EKb~JvhjDy13!E2>OJ11ZPOTNd7bE=j0~3v(Ad72q&gm5!%Vo-{ z@f!Z)Kmd>%J#5SnA}*#sTaJ^`m!EU8Gmh=07Dj2lfAE7Zpx-EEED?q_yr06k0Q2Q# z;bltD5HYDgu-o=ugBh*A{iD^d&$4Aecmx{ucYN_yLR%Gx;Wh9|!E3#jzxkJdD@?I+ zLC(c%MVn|^8E2%GAuC>$B3mW#Ps2b6)T|jZ+GY9TcW_bRPJg0IT?y-3gs3C69W@1H za%e3R>)Hnm-f#Y9hN`|9rzm=&r*u}DyscLt80K+~Jx2P=3k1Tce@3oTAe;j@xt4SN zXsz(ok~tmR8q_nwV~xT&qsQ!kFs)RFjGPDxGI8gLU*!T?65EbQ8=xgI9Mvc}5B{+s z4MO7~o9|gPo{0(5!y7(6AtWVm#|gH04oNVuMZkkhZDqS9fA5trX5;!EtxBL?NZ z`)bOw4)+R4J-casN@CYlM8dtl)vGCWYII>^^X0ohv%8kmGMwK^fzeK)#FiQNEO(%Ikjgr!?8?%c) zd*jGvHW_DvKd5V^tBCjiDh9_S_Uu2jArRZcS4t&s&uWkhIO;sslTIjS`#74PoBwG? zYVFTQKY@*(W7S=+FA<9CrDs&@4a@l}DFU5c3cpyND`l_XG;t5~FJFN6cC3Q!HQe1a z+*$2f+pPxSoh@jUw2!ub>Fq*Kt6mA|yl5X;G$Wb)oB(Q=n|rywVvS2ISk?03Kg^na zIoDh+s4KH|pn1TdM{#D~Sga}X4e7yU)g~F7xIIhq^V#juSp~VVnUBqTxyL)kyS_-H zw{0&)S(}-sP+jdWmqh{sGOB&6c6=U5etP9cWp4O#EAt}O=Lj>nWLAkr8dpW<_KUIk ztQY!96e(pf<98?8%wwt_eTLn zg36M&pq4yGmWaw;R?#k9nI&xJm4cJSzXXq1su;9~_Dk;H_iNj%QMihHR47(^KM5zd z%~uwYVjIC*gr;V71a<(A4cel%u$UepfAZA(8;&mKrg$ChJ!vI_s_Hw8r2rFQ^*HR1noYX1kUZuozd z1i%XbC_`4X$&-6{&Vu3~J*AE7Son>wiIrk4PS;gRQETgF*Q0ad$Fri;g}B+Ig(LYjK%Q6FCO>$JVyNUKr5Lps@r$n*?+ zbYaaXj>xsj3Mk!eVeDM+NGU~(8w)&hfa&Dnx(BZ4Vj}@$yH^wTDF+aVSza358S}y( zP%QtM#m*A?(vtOPLJ&^PT<3)N68n=9Pb&&`S*+VSw@nEMK*2$6zfXFiRbJo$^IqdZ z9W$dSN##$w^3qSu5X)iDUD?8Yka%?z8bN!k|CDv=!dS(-x!u%`7^fVj_C+o>eqU_} zQ@IeR7)3)=!UsQuR@uR#i-#lTI5lhrQCW{-lkVmW(pY_&9`<-pSU8}r%GJGrF}`Pz z^9otaLtdWa= zF*>d@X@m0+zh*rsb4QKK*dqWHwqIhocxQ@<*iF~_M&w| zS%j%XOG9j*4+AI5j>klz9fhH!9V`EH0AlNQ5Bl1slGBudlHx^b*b6?AvH(pSSvH}e zeFl3JDx&z__Xmd0Qh7cg_aVA2vMMSNhrvLvI*yz};x*sOnK?3@?dN!R$zSx6GchUY zFWre6cd#L0W*l+`DNSz}fsh4Yx|??4TKPm;n^Jqa4a;O3TRPfw2x$;MoXf%DpZ%Jq z?rKxOsTxh<^qIfpWJTEMXNXUeW#brLD1$j%Q9)m*^g!bAd2(>Cwk}PdGv@DvLlt8# zW%-S^VbCffnhRT@hybA>+crBHu)Q#Cr8Mw8qx+08lD{dMpH10Wh?RylKRXNeDl0af zrKCbNvwZQ4{e<$Bo=8Q0sbi_7nX`NbMB#n7(x^4trx6P)YEfh)x~wxQ`1+l^ovyY zrqBybjuggwFnlp58fV=XbT7mENeSU7LvCOfuOAvpYHSfzq#O0zTH@TQNeU z08c{3r-L7mN9E2?>naB@*1T+B)gd(s;NZ;3kxg4>%p_UC!GQ6n*A|zc@t1hT7d`xg zsM52=e$A_)BnBKUQr&CZ^3hO!y?uJ+~#5^X<2An3JKW zg=(MMQJ%ueNd!)59A(AtctBsowlW`TiF)Q#vr|LCReU2e4|7-Xn|QS zc|nVxQ)`{I3HOSMqq(d~#fow>=1L0iuUhw$_bNCP1@XX+*m?@>G6>Vhbt!g4s7duO za*5NlS?!K&?-t2no_7vnTPX|WwC+PvdO@r4`P0l9fGXK9b^?agH-mPtk37bs##*SC z)`NmU43gggRY>L1^SXk}@@8Gy>$lWBG;Dx9``lcHif@b{%6(eL0azqq=NU{uQajM zTh_j^_s}~qbw5eM4F=ZJ1i7?H&sw&uZ4Gxb_cibyces~0u_t2BH^B8@g4cmQ|E=cOJqCQTAi*B-0yj7`rV0mlw1DS;Hfq+qB5zAuNQ14;>7|0g$yw}I|!l(M6 ze#m=0F>P~cfA;Mf6{Hq--1*8TR8q5H{gxTGHmd8W;F|ouuXAc1Cr0eJ`c|Np)E{E$aJ45inUK6XMDCd?1uCsHl!KM;S8)di^f=CAp((YWs|bP*L1j zg|lf&3sgDCQpGnLN~#gjNj4g+Md(dLg<3KX@@GGO^hfbQb(O2E4Xl-!JpOfcf~u=3 z@>yKq023i=bI9yXsfAm!LwNm?rG=Q0<8MND6-xx#>Fo7LOgB|!@)8?=AvSCNJy}6pT ziqmR)=gq)}pADx|SA;A!17SF5yJ z-7AW4TnT0}ne1I`^3)iwtNuIRx$Te3Q`Tz)ruhAgrEY#UR74-Ytq@0rZLuOEh*r72 zMB8o^Y#HA^>8?a(h&WaXPSm-hxm?}~G+^kDyj5mj%S`(+sB2rEE7B>>7kC`>Lx5ZZ z7kX++JqBWY|En~XGE&|}08DQ}nzSYWB~u(r06d-n z0e}QZN68N=wA7z*5XRzjSu>}El0l9SxqvSnLkMlP!TNC$OO|8qYn1}<`gEe?yImAY z$#~ng%9B4TV2V4@l00;RKoOZJH>GuP0NuyIbQzi^1pw(wQyd}*{CF@wol@%O1mr4# zhL-5V(4J0?8ly3DtWg?v<2|y}q1-j<(|SZQ>&f*7_N_4*Dwwr{C{>Gk02;|*p64HWi;ait2lYdR@PpkHKl2e7s`xkxN+|Fubh=EC8DWjUL0B9WS(4l7zpB86##i>kPfS}+OQf)HoJjsP{a69A@Snknn4{cysC0-Yw>XRE`xGbhDOQ~SZrBTtXDbO| zuQ`{eg7o}LJee82j!nBvgYGtRYo&zEB)>&T>Fe=tN0bP;lk+D$bVCu0b)n!HSjltV zbX;Us>=7kW8a-1TNsQyvU}{#<>-oKp5X!k5rthz{Wq4Fh)PO9f$5cI>FMu zqrs?DuI~!FEvFK8WBrIK_nu&$3g(|eT;HgA>c8@PXI6YFGdrR29WQmmxz)Td=E+8oU4_15K2R4EByK{6ci%Tri5c~R)oe96~+UdRZ0 z;yIIMk?{o|lyQroef(cBLh+5Q4GUdWV4ToZPj--Dowk#7Q~=M>-o9+u}-bANpkc1u)+d8CgnauFt5q8JkIX z$USHayIUgZGaRh98(~ypQib!+KSuH^{>GTyF|ImNKJWjg^ z^^c;d-A@fq79l1o61RBXCnHlnzjRj=kf>QuFeQq5Lmnom@5_txN2TUkKZ$eXGQ8*_A3YuNa??xoqR z%W?>%Q}1^&AEwFEl9V$Etj$SXcX%^hRC=l|Bhm?J1N&O>N8Mwy)yv28qhsHeQgxX| z&X^>wx#gz|ccJ8Tz1cf_epgG{w+nJ4r+nr=WuBsNRZu9&Fx=9i-Hqw3`L$zQ)ShT1 zp_WM9aUMKN{i$Aer}AmjC$W~<$An8AWae>5fVl3!&tLr-qV8wy&0Lx@Ht9 zxJJQR7mQWh1MXpum)X?dGb9$a43)$9#ksOdfLq@>9MdB@o6Wd5|C~FE^Uu> zy2$y}nZJk1-Nxz}a|=6W()?h1&HXP5r={BlRaF+%)G2}%eKpX6}Ra#nQt^LmjTY7o3#2agAl>Z zU1R41+Tt860)>exjK}z>vgo~X$98M&`oN6>ri(D}UGcB}%tq-Jq#3^b3%*_Ma+uQ`TI^t$REXT} zn>43d*hYoA$e_S5s;lVYo_v`R{K3~rle-BnnLbIIqsRanb zRK57HkD=^+C}GUEd0HQX5>FvQjq!}-SFpHBTLS(CuzxSkGwETa2YmM zV#iDnyXv04VQmpEQk8*0N^?Z+plJJG{D&^%Kui*PgP2hIkvM_9)QE7EOpri*IU@YW zu9U8(2Apvus$!PCpym@yOmQXkw%0mAXFaLc{-G|eH7?C)y|JV|zlCSZURDs!GgY$S zE5tpc_8G^^zWr;PS0d4(j0~gR|<53tP(=wrX2DueYT_D5@ z(=XlDl9IBaOWSYCb2C;GVdpi*3O%OMai3(?_l}UIW{m0WDWo18S){hJq{aaAOtnMg z_}f6MqsnokCKYr86jm7tQ!I%Bx7-zn>!!=F?Uu{X_I@q@$lI;FpREk3xNgG15_=FI z1XF!?v}%ha)9#8ra|F?Tj#6se){D#6A47wX49(t{ZJXMO?au?@6IM;f#YCNpv8RTi}&91_+ zm0-S${BghlmiNlf0Zm=j9j@rf#5&<`)1f~LbpV7UWuONHxQM(-*UIkq3+#~UL-i2FxyoW zvo}SLQONBJ!N$R`PWAts;jehvez}t_3_Y7bsyy6qECT=4s);qzz0h!Srch*hG+=jv z#bSDs4%0>-)Ni`XfU%yyJJb_f{|fcyR9MM+^I)a&=e9DO{(v&)@%5LqJzaXYbSzK& z{A~K@M^p3HuQCn}AJNZx8+88F?vvB?BY%Br$(49tNJek9X+1Ul-2UZdNao9lZnU+- z5rup2-&@dMSyJBZ)(dpuJJYvXo=FY;6afb9NVwZIU-q#(H)wDoAt#4fx_%?4()my& z=CGr2slQxzWI>Qj=LT@)sfgTkYc~o1{Rvnki{nIXkB zaOb}@yyML8CEj!|JF0tbwOXV2>8r}$B=_SSkR>Y516L!(?Can9DCE=RI_osWr@v>m z&PVvAIqhhq6$|RiJ1sNz!Db$l&ohlcXCfJwgLcN1eNwA~?!ayuGCJvjX?_ z4-a|oT9?z@aPJv*!z<64y|3F-~QiR|7YL;&M%B6R=)M$_^6YHxCdUlG9MRWaYt(Jt2pHqhkr*`tnk{vxkC$>15h?L56E~{4 zX7xUh_%E?dn$t`wbWM?}IexR*v+Nr~{CR%w(nUnMfBh87b{ArB%1Sx3qHOLeGg2kC z_knl{W#O{ndCwHdifuqs^Mp?^y3M9cPc$d_hOr7l%UGkZ0KkoF!+W|?_BEMd7LF*cW2k--{2SCnS7wQ!~`IY=)j zyba9!3H}MW*bc4$usQ7dV%1>{1hVnk`oZ^&OiEcM&8Oy4xP1S6IE(N>Q=s_TB(3n+ zvBORdt{COkwW7jR23kk^tTevqK^T>mxJJ_DqB6rWRIZItFGp;&cBm#cr##xO>`PgE0q> zx)u9-%;cu;(kHoRH#&bMBKPwwQv-`1pMM^CGw+XIyBz{?O6(MnfdGN}<+SiZX~_({ z_(!~_AP#puucoh~rNl8*KaH=7c~SXyYujK*;gP?BOCJP1A8IQO|Jy-pSFYd;XghH8 z<8Ig|hgb{d;mQKTu1>Vz4~LV9qp+0WyO_pcAT zQ_H2*FaSWpUHGX%XjF!5SEtFd>* zAw7or@sFeS6}zoF(rz;?>N1j0GhHN^e^$ra=W)CKz*k;)sKWnO{{;-AmKxOap?To; zv&%Ef9)U?#p?6(B{C?m0!3D!1WOUv-n;>!Bos+xcyi(Ov(O@m3hCt= ztPN_6ZNWCFxyxbHNJRUN_y^3guMP2^w7kdblXf5Zh;pn(j(6640+*$U6ieUQbD-Ss zs2)~9WI@#+!wm{l$cfgn&}^(#paFuaFt(0zgXZeWF6m;Is_HzawlYf{yf6Y__wK?& zR98tVUa@uMv3HilQHw}8g0>kO;8RzHEX{y4v~5-{1+OANMU&~y{Qxr*C08tEZ3iY#*X6; z@!>+dU5biDDwmykirkuZXMn0t8m0#nNxFq8eHN9zw8YvsIJ;%eksPbq9X-{GMR5Y> z=q;|*B{g>^0TRH*X74J}^+tfvUUV>hw<2)iAj0l1LUeBql&wnWxtNbSN3ZK zt&}u1UTMs~Uh6T|IeYnU0DJJ$QoF1g0-Q6s9o?tW`3^s)%-Ie{E1%X0CR1(N`Xyb0 zUX|dJq#uVI>W-=J(qEj(H3C|L49`l-stl+jq4DE;9OZByvR=Z&OxU6Fv!5>8|5_^D z&*elov%wH*A@TmzM;xD2`S90hXl4T8?ax(r2jMC}>mkVU3sM7cRJ?-jaS(bU0K znGaLGK%z(HwDyaftpSH_%J7GxIIH-sEGD zPVNT?!Xo+wL?x}=`5f~hhkD==IM<*S5D*f*LqD`Dx#3rR6<*SlAenV?x@b*Hd;M8? z@GAlO^wAwuIgumg@2uHPGx%~5$$HTTYaOmxt4xUareILa@831}Tk7#{nuRf@O*M)v z*3tRHpw-2?&9yAaW#|1ja`n~>kBv_iRv+>$JM0nkXd<)!-EWlx;g3MK7tP}zPEkvK$$j+T^ysPA=Qm#3 z+O7^eEj({|cd)_S+PIZAysNyby!8Lf2N;m;u_W-y!hq|3VhI8Mo(ksH=EflC@44p| zkQ_~AR8FSq?UoB*<~kx4*Z2p?2B;s%a7}R^H_?RM3}srb#DdKzXh(}ay_;i0g|n0r zDv(*zOxe2?{=O6u#0GA(1d!$ZQ0^^Au`;VIiz6~(P08`ST0UabtN>8Q$3&!Jy~~(I z0)HlE|BlB%Jz_N#XNj_coy~K05;y{sBOe2p33QBFh%n}!IN^%G9AZEbLc1Wx0V?wE zc#?MXd$cB_mD<(JF}|@V&cQu|gO#C1QynvM%E%OX*Desr$kwXxW(6}`Knb#XP& zI8&4Q9q+J5WWP!J3{Z!2#7w_s%-}vX-yta}OP;v+Y29<=IP}ji9u(P%pP&18cD4_@ zi;J{`zE8H;@_$-6PM|A;uZ)y6>r5UA^!9wgLEEB;#nVnkdIj1`+$*xxKhp!Q5dNBK zj&ZcHALGawPNN*k9_i-iZDc{L}Ye2O%>vGzqk zNBnqqzVgr~ezW{3E|6Jr23g*DFvMj8v#2GsZ`{0=lA`8_UcF+46D@!7gzJ<0zoCUE?hPlNYMo|IJ==No z<|%w7T%M%)k@%Kf{_+$X@^O;JJ0+C1tUnX@-#hb`)-b^+vfT1GL8gDcm^3l!ays=> zK=FLQdAH!%uZ4&KbJZ6){oU~!PKs_tu8l715uZMLM+_?cfcYHbh2QHw{!n~lAt1SI zSpUa}K8F1cfYgg}n|Big|L1>s-INVC3Bfl$B;QnXV*oBu(p>w?TsIxPT>>v4! ziSHd(Y^qIOvb^63$$r;YF!|oL74wGtUv4y&t+d{DMO&S6786mThiiPi3;)pEl z$C?IhjScw2&_z0MW-fD!U8*)Xg0gu>z=Raw899O=6^I8S^|ruCsawZXV5CnkK^(kW z4}MK~Xzj9NkX35VY(v*atVw-;7SzgmFeWO412L>Lexbwp0Ae77N1nZxE1h3i;x%TO zt-&im6t~q)a}?rt5=b^-?)WJ)Ir(gCjk=tHGcQdaNI-i~pXVW`%@PekBCqZT7ZX8j zR$t30Mz<#~R$k5_#@_&Oj&;6R)-kF8IZ$^g)xsmaae(@b{uxiRo~KR(Mv7&=c)XNP zXW70)xlue2PSKs$Zb%c@INn^7QOinDpXk`WlyYlD+5Bo#otD%R>89~t?d0B}u(KhL z(IwZ=V=JT~mr;jF0#ewQp?pp&OiDPqsR5Vs)l;48gyAv1EL=;N(Ui35+V1I?D;5&gjU ziQc=_2{N~sZ-<2ggoZEjYid&cL>gE!d~4;PYlJR2?b zu)+`TNP4vU{BqP~#xa*A%ld+&!h|<$2`@Z05&2=z+wUX0KnO@-%A6qD%jpNcz1g;D{HijIudT5t`c?U=;l7PxJB>vysqKoTWMu1i>v;<}_t# z2LnH2n9N|^P{hGQo2;dab~v`(Vom^IlV_FQv^e$EjYz^5k%wC`P`UK2b|}fJTlln-;kxQB}&&Z5Ns8h^C^Y(mZ#2=G64itT3p98$^QeLJDL4LJ}3XcxmgwxjOHBUw1pIo-xe12VtsrhD<@^rp6!81>&Izy7+(t+c`) zT!j$IfE2#S=_)DE+pCF4%v3&&SA%QwWZPX>JdPD4>#L+wAg&wRz(5QTkfLoc=j%)s zNU}0?YppJtjO(v)iifxvKMLVrXfTPMw5P4C3=feiqgQkr%*_72d&kr)c4<_-i&eJC z`4Ud|?*e+f_tJi+;c`!{2V5mnW7T3TiiyxP{@&{_-Cm2vpX7kM*sj8}=cIQvPdVts zQa#5+(IYng&TT=w)R1K(!^$MYUJjGXbx28M~0|6r+LV7U@LVn^<3TswA&HF+e= z)H_Fet*M|Wwkz>b!$@ZXayF~xFU{-SM|YRql_<;Dx-w5`?)uW9Qd$q2DC8WU_t=Qh z<62o5-v&9`IAQAD{@-}P1m0egc7#hS9jye5w%sW_?b~!#SGcSCek^Zf1`-?uH>GpR zE}4~dH$^^}p#bEfytw)K0ea@sWj`Iao@Mc!jnM<}OBWn2x6aSwL!VCBH~u#BI(o5b z%en2g)rb0jYnOI%|CzTv=B(p^N;}cAGuOu``(3!z^A@9Hjvc2fKA*tPDdl}~eV%uE zCYaNXCr559xmxXe=wX#E_uoD}23a6&xeZlp3x>WAifso7*~c!;&@9qyA6C1DBU_Ks zZdzFgvJ*4Z;m}v|Be1cSCv?(MGmS!zht_WLw3spSL9~K=kM$=lEsB-G*eFH0iDYnB1`R5bvc?wr;7XkYeNL~ps4o~JxLTj43 z&}CF*5tOtlQ^TG_e)aBap$RRJq#P33q|VQy;6%l7WN>s@pql^05{2MPVY7|FXa&up zlBptpe;`yh4g`op>%^s>&6JR@iikPDY4dA1FlGpCRVDmh4$Het=%me4iR!w$tmcIL zV2?G0?WZWXwQ#|cob?L^F0V@pHUf1>fV^||I?fcDP%SVnYUVf>fv~VPOOkhx5oF<} z?b3{t!;eredE(%Qx^jcyntu}$n~~S}mvB61Uh7hQGhi27W(}Z)$SEqfPTsZXqyJSh zA%|#}$|U5>RzOd)%FQC*lkkwtnRh}e8ib8&++0yhw0uocyEPr3(?@_ zO2vjOlv5>AZVCjrSwK=U^~pHP#+K_!ECS9z`!mI>2KR`m#AIqV!tZ{)^s$ZGggo7U zwM^vf;BJe2otJXp0z=7!YpES>+ZPaF<7Re!K)^od+^JGG#?IJQ*`#6o%T9eIsJe`g(6AY!%{psa8X4RyL-W3 zH(Ac=Y-qb(E_w2nwrtnbW=jLgWT_o{nR#1xkv_{Rid$>bay)9$`H6nqCjv?H`qFwY zgGO6i3}Jghp#K#Aw&xT|Lls8N54u*rNx+qsC%c6v2vS$>mA0jfTHK)^QpsK}5h~&g z+J_f58L%8H+lH&cxA@H|(Osx-X5#|)#n{$4ziu!7(=6Bv={d2D ztI5EF8qy@~!+#)#&)(ch)*x>hcqfOK z)<4wuA&131^&L;>q29U}GUfq`wb}F?P56!|e#Z`swj1XEivtSA9QO7WVF1!Eg7cOt ze&*2YA8C^E%=h6pJLJ}P#x@_!7CNP8aI^h-Kb10*sPHz+q1mE%BG_$PVcmndOcV|P zrw1Gn%4}9skXKW)5;#Y|t+n2Xf|4`|C>@r(DH6=zo=|b`TVDlddO%NUPH1y<@1m3h z7g6$JULYBD*`QuQ#U8-eNBC#;vaJG`&h3LS+U)f}cYsPRuYlYmWw(q8p^^LjnFOp_ zpdvGt0@gmUDt1f+MW&xFF2CwE#EAw;rEHeK0h<8 z!5Vv>@`M;NYDMaG&{`BL8cox~E1*Or(Sb}z&?uLhqVC0>7yXc4y7M;jB)v%K zR8{dK;Pvv=Fp%bG3#N~4gy)73t-{RCR?I?kttuEgOLZ6Fz9PJ>-+)QG&kp$nw$LtL zP3Mhg0c@=(&KU}hGQ0~mv8B%Q(ISfpMR-yM|BT7Cxm0mczm!BoM8rguVkG_{CZnoKh{7VZMERG$hAISZ9~ zpA?ToY>4YM98CUYyxZpvj1O(9$6MO+AIXQra?S?dr~sRMonxV#C_6%1_H4T^_{Y@Q ze`3^BKsh>MH`ZD0??(x$%LW z`rjeqb;q(Uy_Lw7;|`k$OWPYOKW)Hliq+icJDKZ=aSwO22Pk_cxrX768`?$rS{`RL zC+@t_Azcr8@RuSlVXph?=>roP9m-Z?^JBxa9qB#`o3Q77;otN5F7JLqmu7+VRU z+0yZ9_$5l3`2jLUewzW&2K$RNc+*gdi7kas$6BF)-mIm1J>=7_)ySaS*KW{6UIM@_ z>|G#i@wEKK0wWa!T=e6hcb7zXIL-Fsk;?r13K5*<{TriUL-9#`A6saTT$DzwJSGXv zZBkJXou+cYQ=Wv8eDsBPMqbh8C_(>DV1_~4-Ace4X@(wWo`w}n`%*I{nzbB}cJno^ ztXRvJ!a9ORGQN6S`qpGgvbmZRXouUbX@K%GIf&MQ%M(GS{XGkgBXbG9XIP4r+&GxI zg*|aPOdGKX^3Ha?Lw)eIYM5p$;(2x}whF{0`;g!R7dq0LmakUvwQ590;Ls}$7sI?ue^ng&t{HX= zn(s;5OGM;&oM@1q$3Y1$hZrp`Ei zcUJe)B1GZG;*9o}=94gvjBLFn|Fw#v^?`|lG|GhpQRu^Epj%}zbYj-V{IOIMsy;vy z-&K-(-x96;r8weg-e_tBMyu)t*0+4H{uFHIg2Ka0mChGCfd>O=!-mWnH3I0VLCB`F4vm}&k7KfjK4T%dW!4N zIj$qF6@6_2aC2Aajq8}D&9;V)Ls3VwPV`9jrn^1j70$h|*wLmpYZnKb@^oxzZ-f5@ zYvt|y>F}k~$60xmf2zg3+MVc~I(I!1mzB{G7uK&*JV}#)g)v&}7Ww?|pIXip)ow%) z&U{-tOn!YXSb`ohS-uYP?vFCFn`?h0~?4NeefU>03v%K?m{HIxywz5fM7 z4$UFkhxYpb0U6HzuZUzjG@zhC3e6&O9Zm;HlLMVYWN}N2#M7RtrqH>nn$SuB>q#Br zvx%oz`~6Wy$miHnAykH$K8%H=;^#dhN&NE1PcK#^q7mm`c6NB9V8$>VZ7%Yy4HeNh zimY4b!(yaj%0tSvG5L1^C>?&UP-q+}o@RQRS^=0UCdvZ9Qjm+(UiLa>*=)lLIt6l; zXhMK5T$A7*5G<>OGhwrv8K-sdYFl6|5O|A~H z5zPH^xwZVTvY=ccnrE>7I1DD)CqT*~R)McRfhetqlvM}r*r7k=g1QN(hkqG%QVgki>=|nlu#Wgs~t&Y?DxT1^|V#mr_ zBjT|Lmvt)TgMDZ-Jp|^$D-t+3?TRGX6pZtOLSPkT%$p@@2WYi6j@CA`iE>3Zra))e zqxI>XFGU_!F5|n#-&`|$Ytp6V(dsObZNJz3Pz)N7ngy+_wabo+%PtUvINg;~8xg-A z4zH^9xRo>86}_^?XB%C~>{EfR@tDceOq8z5ptXxSetGItYZTSGOZv{FHCJ>3PxE~l zaQTY0|JyUL;S2n(x?e?fAPl zqVSbfRmjJ+r|-hL`)`WB#IcrGx4b=^~dD=!<94__TWa8UfA&z)Y0)lQl#Id1CG86 zf1Y4Vlah0fcW3YluYY7H4%90cwp#_1yQVn{Xv^mWlceyBwlMKK2Ed;y!oCY?wOh^V zVh|$N?h;TI-z8z+b2( z&A~&*HK(@~WgK=h5)#V{ZFH4fOvoCzL><$}WccCd0ne3Og3uqJX!GQ}>QsrbjjqJl znhwXJDJpwLl}u$@12M6bK7ZwBGluFkG8I}&EmndqQrWTP+*R^QZ8LIE`X_avP60}s zQQj6%7(<_SeMN+yHZ=ld_{vMn(97ye`OpD=Rx~Jz26}5=pJsEQp|50j6;1p5j|_tQ zsR&F_mbfcdE3jFoC9Zc6=dSDh>wDZwnqRY-2r7~;rlQ8T&Ef5KGPf%jp)onl+Dd)p zj7hy7^ff7cYOg`P_&ounG;KSVec}3&-&N!fHp~?fgXCJ+9FRGH3qezPWlA(+P@*z* z5Z?P5>q-EM9GrDeTMT(7Q8Vksu{9cOa8EPS9xaz>F_*l?Pzl{xw4Fw`IP*-$5Hkyj z8T2(h2<*F~sijk3;0hE$N|xpau-)%72oXR$gOS!Ms>Xi z@;xhtWdCby7RAugB2_tIwhBdMU zIrSGW31+>(wyS7FRBNl4ES$g)tF~kJ{Q?L%LwY-a0gYnXr4)@ivePoS$cprL6CXhw+~z}(Fx6* zaW-By3wbOtu&PS(FYCKt&m7Q(#Lu4e$(e*DnlIRe?6~G!#n?mZ-?b?U&4MUIT}pDe zS&G^d9kq3`2O`S~WW-CF@Apt{o~`rgymW1+Az1yIUi5Cnk^AamUn{mK5q_g;OynT{ z0W!C&YOb}>WFLwWZ+#m-^uFW5((}YOIv+o35&yt_n(twJV7-s*={_?OhzNhv(W*a9 zGn)|iTlLpnzIY-o@bbCV-u3`9vz5;s*Zp!*-;QFw{_38ASom|}x6B^b2FpJvQC6*8 zS7^Y9ByN#@bUL}|*Y@Bq>`cb;{aqukL(eZfR$|`RUlY|`c%Gh>(N2-9CA^>3o)f;S z)=Q0w>3F8|e6{D~gl;Nm{fK>+jNXupTN zk^n5q6^y}#{rgBCn5Mt;dopm=^!Hj_HUZ@KV!*5nW@ZMY3+%x6fO7-}>(bahY?2@7 z8-PL<1$m5u_DoNwxyy z`#_|`Q=-=oP2x6I>K$^MB8Qn<)QHvzpd-pmBu8)OI9qgphEdsz1b?A|1t>yL)V0S8 zEf7$Z84KF}ScWdOunA*_R_2Hf`fx<%HVF1(C9)j~zIG=juvZx@j9)3YC413pjl?;= z@DT)gcn^?-v!TV$WZ4DyQcu+0*nvU7A6j}QZ=g{+9G1_qRR1vt3m0w~nmyN1vtLvzR@UZhuoo8M~uF@bCKQ!o7jr^KK4 z0aubTiP0rp`M50VxlStJ#SbmjVpxf@Rx_oKN0YklHsvdWBOG##v|YU`L1v~GEt81m zLcOz+Bd0cbYh_2gOxFVyb1o|AorlZ(%jn%|`QQg#9F!R6_=0nE9_<6_D6H$0r$vF@ z#j;(u9gL7g%|ARR)C1#fduokMm~`w@X5=ik%Vyd46-at1rC1tOb3a!dK zBTsQ)MleUs>?@g-N0~`7tOLlJ-5iYN&2o)yOmRg-&z}8om$snMii(gX9olrRYOsG{ z)4dZPz=AnmY6l08@*shvy&$^sBpaxL2NHZt>Er8j9VR7PMBW~x@JdPGVZ)s~giSz2K$3e)#Ew*+2GFDo#1CN6j1`-)Wej>%L`kYEMA7 z0ww?Mj5c1uV-?K0W|dD~qZA&@dv%-$xg(?fR$w(>k67z{n(|TE_w7L2?>l_oF0~y( z>p#4ev4$RXS6NPC|6wdRPkhlLindR3=yBvd_KAu4RKp1RS(3}*<;!OhAI%L5@tR|aSy(X zHL1i~CIy%`m4e4@MV)N{YH1?(t5Wm$9e=e1?8b`i`b-Zy!*+0*YMVl}DT?(d0FI2rsW(+1yY7L-JDY7s0Z%X)V#lOLZt?)j^4<~Xtxw~DC- zcbHa)M-!(}Cv%|$FRrY5R~Ohh$GL>Zp_gh!{<9UZ`}^ozq=OqWkJ@L#O7AoOUXNQT z=rlj8hAP`{qX@ATpl(OS(3{j6GIUi@DY%%=QwEBRDn99>mnm$ zmvBzamTQTZ=I38ri7b&WQZW)bP@I1l+02Y~<|sDYn|fsrY82pW45G}lYUDi|gtEwn zO6k<9WYLMS#a-=tb-8Tl-Cou%XRd|6FM=fz9`sqqE0%*`er&z3A33%HqmIbx-S!$s zc;dmFD+Q?8I?FAbr}dv4QazCand08ppsLC3 zumDwbo-)qbK?fa1oc0wllR<>>ZjC!JKPHQYaa>6S<&=Yt;4D*MnG;w`d#H$rs*acZTW=q#4$^dABF&eyla^{}9%Cud07aKm#Grqsa zLALy!%pBK7%YPq7uc|Z>Z&K8_b~8Tx+r1Ce&$}~jYF%Z7o)U?ZoiKH!=wea6^@B4>v!KD;)jUPokKOQsiM;Nl*tEKx#7B%bAY+xWGj6f{&?K= zo&LUNgNwoTqHdf^o$Rga`|dWLsH(SlC@Om6y`eJdFiiV=>`C8Z=%?!$xkiR33Zxy$ zddyE$6p5Fb-ARh5j{ZhXz5s${Ts-U%`tJ@sAh-$mZG?Hshj%OPK@lF&l= z8<|8&s?dgdPytMXPuL3%ESV+b~vs5;G5 zGnd-~V8v5*ZbU6Sn=PBplM^~YQ$4qJa6wey?ots^oX0KgFyhJh&bjJ};oi&08jvUp z^YO8V#$H8x3V$r1_kmgxgeyT5QzC-aLKB)o${&aB0pPHSmHA!;YmgmtBWfZjLL+UI z&M5i3F;sm~&4yZIG~Jj?RVaiDoxnET3jis=ov3}81g!iBRhg#(Vra*a30DE?Pnfa= z0>0%7Shus{)~PZ}ReViQeI}-(AlBQF1hnsOP;;L~pVrh(MvCpxV3pkN&rIgZ37tvi zmp55y9Q|qF+DU<`^a{}0E0_IMjwg!N*sF%GVySwkvV6u<;eR$ zRgFwY1{T2kX|wc!(~&^5k;&Cfu9gte=mFO<(kEeZd|!EIuqSp?l~y?q;h>0SyQ0mY zVb_{n6f4pKN~QkxxP8ySN_b~zVpZvG&y$kWirIRFBiir9ph}RD#1t3anC+QHmM>A!4LizP_sWA8&dOEYfHwOK!|#~$J$Rs_J)R>`D=}Pq_36S ziwrMx(w`ZVW(VDKh9%iWOqf;p3Gp}MfE-i9%}pgfA2V#-9FrOpxe$|{Ll8L^2fELm zs@5@M&Kf)kH7idBGTIG7zx^e{zn=7ozt-2*-yke>kSNgS#*WQQ`5fZ5l1G00tTo=U zQ?KxR8+k%sMwnUSsxR&n7PcLW^Pb%w2dsBpKG(2vHq?V{J{Dxv+F5X)5Vi2nScKt^ z+TSqM38K!sK(8JQZ=-;o%sL?|@As%_Q)fmub-sJsxkE(m@15f@SJMK-GC>&~?GAeV zwX^(*q6Xr|xuVp;jw6JtsikTsUhUPMbP!>`u`Rkrq^nmfO?Tkm)s_*(ged=)m;IUG zNn&nyp+~^vpw9Sj8w$3f{><+(`R}RH+!~s{oiBI%&$d*C#ck98xmd+g{V?kXAQf3{ zC|wY5A`2;MNKiXc7Bk=Slu!Xsw-#O~e8Uh9}fEb)<3hyH3pvamMlr^j= zT|2*2%uIAbW+AxMF6{Q>#BWeXhCWe34==hv;mQF)lwRf!C?gKn2*l2^wqr;xG4?9T zsP`cQtU_H_aRta;u$|rrE(8m8k_@E@@KymM3`7g$mtUzFb4|LP^(AOFYll6`J`KdO zg)tNnG{^|NK#Cr46$A1Xh`}XLrZdDJl%7gx5XNG^PWge!HWk0531G|c&EVUE_3^oX0v*cT?4

    ntji(<3ek-QH87E zT=H-vH*w)Hxkq!j+0xnEi!<{8ja{T5XQvi!Iu^yk9%j$w+HAICcLo9PlnCRZ*gCyv zaW$;3yXA6n3p~98VxJ~2S9EVg;y<8k_KmVf>|Jh=cbVO;8BBMbedRRZu4@%elG5UmlVv8DRGz9N(&6)pD zsTZq%b>(JWXU9>M7iHGRdtW?a(rrgu5q{eSG8dPXBpUw%geKcFWlV1sl?8!R`{Sl0 z{lsWX;bs7|5hY08mL^N#V^<5Q#9Y$b+<$kfdvnq(M0Ge&S*`9ubp_~mIO1o|b+S!JUox1VQ=Um3v z-M1~0L3#PZ3lZaw1}+!|hX1}BSeY|FshK6zpnZ&iRCD6scx;CG??{-y!h#AVnDzV^ zd8E9AsneSnZl9N*$5uV~WHfZk%Ecw{ESIEmHQ~o`vy5|O5DHM&y@v&HVgy!hSjH6i ztGsW%Tdk^X9vHT9kiF*fNxWVvB2ifV;_4Tad_Z(?X`&NE;WgkuO!wD~P`-g&$ zZ@l)L6L91<$#-EdJj^=gJEZr+dd*D{vzd(-i3{uA2+N*=70aB< zYdxQ&5qdfdn!6*zK)%Y&U!g()UXT~37J^Xd6o43RIL4raTglV>5qtjcg}a{6tq#eTcuO6fJ1sO*25x$qxs^@H@s9Sr*!k~|li@>pC+%>p&? zMO}!JzzZ3#50!=f)^+&+F=j8bx5aLW+<%?jA&#u8A~<1(KjV+TdfyVUVra?zsh|IY~Z@%F=9A>s}eYV-W2j)ipJ>xVZ%8+~mE?i6) zj&HrMLNg*M`5xu>^Teu7P+V-X*tu6n_f>UX!x9>U2}Pf5xCV`Z4R* zsRvu;CjDQPR1^`0b%55HgrX3y>nQ?9q5K3SYp>+lH|)=NJ3w=UnjXb7EGwE)`5)l@ zi2h(G0fMhZ=(Bz}DD)tneI0Yqa3?@IQgILw;Z4)G-Bkk`HaGa2q31AB!&NmuIP!q= z&C~SjuYQa>qt#xS6$BSA<63L~m1EXDiW95tN5ZkS;?d$})+Td-wSw4{+8*}06H!hu zFHj~y;V?Aou4jW$FJh+}>QgOCBPppyYLE-F6lXNlXDDYJXli&V`-@Z>z`H@Ua{G=W z_gGY(fpDZ!)lM$k!Mb|O&sizeBiBL%fga4pAU0{!Djm22Fa(=kg`*rJQD6&`gV6>a zrh%6*?;yD1%+z~~*6msToMmh!D+s!Zzm;rcr%pc{f11O$X7uY*l~Yy}lU-RyctTM# z)VEWA_b?i_T28%TvTBm-DZe$SSv}tZy0^7bT3>3~*|dFC-*lwgBE?X*O@s%zu=8fq z>{zM1GhhMin|vF=*gEC*#JkZ-nzsX_8y36nQfLC7z@^TaUu+TX7S>7G z(O&$R#Ro+gk(tbRU?{YKsLCe#}!jy^5e*;pd==DqPw)0T=^-F_`rpvFC&&UbqcL~1U zu;ILmhSEuUGdTYOc1lg&U zrW*H{G0FzUM>J~KqkGafTTO6I(u999gq?HJLI$QI^~mS){p|XI8fW7% zTf569=~FY#%~x`RheO%fF(}FA_C@i&pbXjWlC&X6t(+C(bK7C{HKd>RjC$4${=BG^ z=Gz933uKAeQyJ#s~r&<*}*~*zb?YW}bDHw|qOjf;#xm(|mmP zt15}27wCsz>~HA8K3e(uotd&vsj^W~V`3;Z9-G-0gBO>Ri?0KUK+iPadmWPQ*z9{= z!NpH|B|=%OdfBNuGbfE_Zhi$Yv!RoC=0EW_sVl8OeU2rDxiu>O89?9K7d3O7ZmYa?mZz2n)I&LZk&zo zX_MTo@(TZQ1A^}dnBp3O*j_DE7dHgnn3 z_0fVYgAqJ(zcjKx8Dc=OgZ6t*8>VV_|A&+Vx=VOIJ2{aH%0`D`Zc;Gbb~p1e1vKVRo5f18_T4b;KIYQ z`T8?q6q03FE*}<1&3R@685UheV&%6g%(HZVq{kor2fS5-d-$Dk@nte4ceNs^TQ3uD zYrC5f*9`70g+GBjSELHoo}W=i{Z71G_!4J&b^pez7GJNL>$!jb8Ycvg-~QpUD9gxo zB1Z7InOzs)KX&pKriZ=ZK+P>OXFEIJ_Dxn_b%7o;6Z&k$_tC;mow`8nR7yy2yg=gx zwiJ!#Z1u;El@+BMqMWGCu4z}2lw>|7&5;?StePPC+(MWhAgS3K?`L&1AXsg7SEt_Zf2m>wBv|iE~HRo7rwnQ=##<(s>7|r*8Rb=_z)&uQWPn(2ZtJ!)O zmnBESFg?ML0gP;6w9a>Ts)T~`L2Ov&P+sR|-=8L4zXY8S)CsBI@v#oKu!bfklJDt> zRa`gUZ?)9<{L?kgB45(K(<(yf-x148fGx{PC~7$zxAo{lG0(Hcnp;0dNMaW zl6Uu6#VA?JC!VfodZ$q07_r{V@yH8*PSLnDv&Z5|v>|_~@Zsm@ghRzjGJIv$)+ZLj zEcfY#v4lKeCONuM8`eIqIf_H1Bxev$hFn?#b&81dC`O_pq!$DBS_SyTt}KJ^VHDrHILMvU?@1$xR0U3~;ih>aT#(%F+I(yuxXf8gITwd5xZ z4z%DxyiCbj=_%2p=J6C;>0HRK-(OvF&5pmfhxZwse!_0z zzd$E1cRK43#xs0Z`%LgpB6X(Y#82)*Mc-imBg;#MSKhwjFh9n5D^75!0~n~pm$~qw z$|eH+ln52PIv1*L5}+>;tgGH>j8Ouyhw!HvYO}^SPuKO!!7JBb%%{yQa|)XkbuRe3p(T2+nNHn+_t;xal)9ofDIh; zH2SG#n53r$2;``_Y#e`oT-nlK9SFWP7$S9Dy*3tSt4%PM&bnTUgO^)5IGL%`(W26= z9=knpDd3{8c0^ZCUbh-Gwo79L)j8fGWk=d1_*E2{^J;V-i`_5%Cg)n*m$qQxW$=-j zDmD;%TZSX}j?m|ABsOXhBV*PTeS`n=E8A)V&jc=YW(Du!H3xxNT6;}L>wB=nM?XD1 zWG2ocyd!}>Pn{v;oibYH&uaARR)upMUpzVbCSjadnBfQq$uRNOAF%VoZ+2=R(K<1z z?&ycGHG;7VkJr3Vxfi^jYPBcNv)vKy1(nLchd|sdNWH7;nG614KuJ{ED97Xk~Ik1hv92LO>c8Rth@{ zj(z$GoumpXkE9u}`2A#PRFE1hp`g+k*@~&CrAO*;reFnL5u&Pbz8f~q;=J73|A)uIw-HVu>xt-cVPb6D##&T$jUDlaxtDYX;a0j zDfx3oEN+_Ti|$XCOJi~OvrD1WLVU<{Zz>x~U#VB}_i3nk83w~bzd!ZEYo=9jCu{rR z=f@fzE&GR}JBA3yd+usY__$WXXDAY_ zIlONVt_+6Y@722TqclUwrul9)@t4)2K$Y1XHPezLSM(;;xE6-zE-X}a12q0; zlHg0T)gx3rYM^U??ZjL~v7DD7If?>tod(Gi#Sbyi{+jcE*b)i1Zfv@mWtJFP%jC5c zrF2e=Cl>gcK@-1DN?e_}=d=1es{`8)l4;%hz`_S5{QsFXN%Wf7oA5u?W|-bOpK}XR z9$(OijOd+SErEMmri8ZH6e}c(KO1s(9jw9PV{>QzY4!kE$(r^`k_D)rPU-xgN%6b8 zPcP7XPCg~?)(k$H{$lDD2m+#Hi#_U;RU>&1s}NUL2}b~HwA`*1OYM~dq|a&Bp5d6r z$KI;yA%I;u)nY~!GS1#{z^{+VZ@y;!ED6MAp!>Z9uu2$qYm^?Jud|0KdR?c!ef zu;E3f1X``xu^@>MO23VXpm}rDN#*Ce9fMb+a_NUZr}UE4F$QTBo{&J*(50ZIo=}L*zi!Rjl-W05^eEKS2OjmWGwG}P0^_2( z6xdu+@O8c?$`)er#h6`OXL5_F_jiI5rB%WPXUZj68be-9*Ua4XD>#y>te72 z{Al^W9Ml(^G1sd(E$y{@ySG_iO?DHyTf(;mLcx2hO2lMa3?@$4U?P66ey~9FIu^ez z5UGIgo3l+M`G0H_-t}FgIljnEq5W3iGJUVATi>h*4`CGs1J88mJW5LR4)Qxd`&Ns= z+Zr$P_R6usR=d=G`1dL~{!gsm3Js{CCon$Q#>&<67Xn2R*Dd$v-SwvQO+39ekv*Y0 zKGEupKH^pT)zi@ACz5TovwU4W>V5iD*SyR&iCfKX{7lK4SvW;nNoGM|EXL(%Z^>?j zdZ$zuK5e5*fFWQ9J|-ZWM*U&L6l;UFBLtuKsl!I2k5Ps&GmN}6kP#oHeKC`h2t)lj4aL8bT#D;VV^}{b+R*iUN$}8 z*1aiWi2wA?1KewG1gc4+xy+42Wa$^~ag-L4^JgxGxE#s2^?|QP$kO?dRq@qiTTAXe zl|~}4W~S>BFRRQI!z2Igr%Pk*J|yMY2=5h-N>tVJ@vq)Lb8aMWfAStLHeHzAzE-*7SW`?ofX;XS6u)PI+`& zHWg$brA>1Af4&#L0!p;5!SkM1xSeR%+Up-?u>489k0aklJa(6j+3l5c(%IvX`62ri z-koUEzTwslo0(+A1?!$I4#neP{u+MG#I{St$5u6LHR z7DVS_j&BF1Y078~0nE*mK{vepb5V$$G!~uRyY51~u3-DOwH$OR2XC0Bgii`HlpS~E zwR?D%Jcf1%+hQtN)QG64Gl8sFCfY%P6S7i}bA1+c;>$WJTxw4vd+uc$gt%<Lw%IqsQgT)BGZ9AXC!P~Yjqi8Kf_*KBamOIj9) zAy#%=a1+C+7m@;%Y``+qKu^*@$|hb|-KG=lpwGCCBi%TakQIw|I+`jK^J5VB#9hXt z7tfh|adQ7Yy_ZJzIL}?9jyPOPz4Ko4C!^3x_(e#c{Vmyy)iN?IMilsW3z5S${wq2UM|NiO6aXY# z=nVQ(j_nlq(NJV!qP2!g?z6A_+40#$4BFV~**?S}Xzt}R6BlpWt9fzwvFOP{-Ms&K zsezi;0zo(CO9I1Zp3Ybb&|n`@ceOuEIU87tanq84L*;zg9KG}b6OnX<>i@gOr0ajJ zyRLhFq*cWIbU9r$_%<_s!I28$ry_rU%hNZy3z;dgcPU{fPa|sZo|&@AO?xxC(CnFq zQu8>23Pxn4g=F4^Yy$KuI^Ji(9H5&SNfV}xRx_E4 z4_fKGeLqJxeR^d{J0;wY9?}|Cz5bL5e*F(nS_tw&?aCMSj*k=3Ao%5W-LxoDo2ahN zEt=G~YWZ=8rpnCwv}ejCU z6qgnsfaCcBcoe;LCb>_@`1VED-`unWGl7`@cm=b>QE9d#{z|u1=SpSvj@G+lRr$A5+(&$bhxjD(qUF1m zs<9RDm$Ba8#OHMsK0W-vR|L15bUPV+KC1YYDfx}iF6_jpQG3aO^|e*aI|kq9i#yC7 ztH!BHhY1iViJ=Ug;S~mtY`b$yTld*il~0F9UDJYxzV}&X8J)?%*FQCt3E4U?z+V(K z+E#x{6P^CA0m~zhqdgX?VRRzjkB+N428t4_$rT*d573;xFE_zCaZ#TwDL8KVSDC&}D{r z6cl&771|pD0)q!K){Hp#dp0&A_J09*&&<&;tALUDYrgxPoUN49A#unVz z=AU?Z=mgQ1AI%_o2v+pEeM_E-)oKT?{*r^j)#K7~8Vupd$A9PZ){SZjh+dW3sq7RW zo-U}N=BPVAyXi1n)@V^M`%-sA&UY2$A<@G&I7N&3)FseAsu*x!e#m{wo7M7tB*}4G5clS7@uPwBF`IZV zu@|jsbL!^Ktm_>^ArFm)CK2+*9#Q`fYi}A2_2d5kzs3xM>`TpLFd7-A%~(gap)kbQ zm!uN1m1QV94N~%+>{}X4Sq4QEMN*@(mL*%VH8q%I%a&}tf3MH~od2!w`7Jj-H|DtG z95)89xvuN^d^}_sbUC^p5j{P_uNa215fQs50Q@6*`{a0I9;aI^>^ zp+RV+oK;F`#muvT#~;5F%-vEqoTLffxVCYS00DbdRpa3nOEI3tA5BzTNlk{-_zMz ze z40l%%VwAI~MMd@hoaiQ9&{qZ|yb`1`7Om|Ye~~y0O(5g-pKp~@!BFZK*&mQy>5gOb zmb%8ftiOB(fim){+&R+mj?25sJZ?N9(((Gwh_n~X=l+tRUxN&&4gtZ8yQSh2VCSIj z8E@0vNP$I?4!UTxLT(Y?kni51t-J2RKUX_i}t*%A{khE#W{L@v?`5HoDg^P zLMNE(EE1)HkaiN<*JM?mUrGK93~6&^9Ih+Xe5u+xML`MXYWSK0m)~w|R)&O}gba>7 zG$kJ3>PHVTgB)k`-%TSACtqD>9+)7&KE)>xk&FX|lTz=M5baut79~f9D9hxnm}hy% z*5_=OJeu}VBYFTWa*IHDMQ5v2byJhhxE0yJeozB)awW~8Ku`9=Sjp-2i*f#!Ph}Os z>lm=H=G(Hdb02St6yxVp>`D?u<%TwqmdB?=zzO{0%cD_=ZH~BBggAZuQ8>~rdOM|x zwa{lo64ZTt{ctSxca8UwA#TZwd;8E!?;K*Q+xMnLlK71i`c^&)56h%%?bPQ%^6{Hh z()F)9;h_I~<#VSr_#vkK>z3ll<-O*AfggY7rv-uk*akadNW`2o+8fzDNIYlKfkktr8qu|V_H)M1&8)d(L#-&e z-I3(X`jSMtn3cU3nl9?!tRl#d*sZ!>hw^MtQaf(cx_IU@tw39L=q^>2LV`0|+ZO~2 zBVeB4nvNmFK8r#U2lgb&u~QrJ;-KC&dN+#_Dko)G1MapBdO7Qw47F${H|*8-P6={2 zag(xBsfj6B&v}WZobjo0dwQ@+as%=${d-PCkgLC^uZ#&n?GYux?}3QhZ9TyVLMNx- zu87MzlW(i39Y>Ai#Iy7xAKYVkKR5lhUJ@Fg4pybHh<1II z@abO!{b!Z}PzfEvCi8^i%u^H7Quko|2J?Ze1y{C}acV8WoXH2s73xgr4qd8iKJrJ> z^c($Zmth$c#d*fttB=DPM*#9~%fl6FBcujXOaBE}p(a%B^>fyF!GL4H@vu9;v!-4w zJ#>ceyw3S}-gM+Mno6t~^*IMiafa0D|MZ~ApyP7#E(PD^Zv``N=tTvIEIV%-~E9s1lIZHlL6+=R&Ts}WEvFL+!pqV8Yc~}2q12xXsDUn7C z5=wQ`x((`o)IJ7>3SXX&r&lHf`|k`TsF}qz6y^p}RYU@vZmGspUHj{?WnOOFYT(Me;`q*RNiUOd8~hN6WJcS62d zY;`DpwtnoUJxZ^&X{MyK{!&z~y+F3jzvuD|tQ7p@rt=Re5|!suA2yxa2{7$lI@fXn z&{op=G5d0pG<{3Qxbyz4u-d=ZWz2$vo6VJWeGs-5T@U;zQ9;RiCKb)^lrj~RgU@9i zVSFc)NW8BmOxJPwYWO8&@{VNkFR2_Py!w;XSzxtwc!jk+XiExsCVn|1P_+9&*Uytq zIb{PMG=9n^vI8$hUCpf%NJ+jQxZLdQVa_{|62K9^Ds$f2@RZe^LoTkT1J6KHv%6}% zlpU?ff9vZk-xA|PN!s0+TfcwaKR2!WGR^gr4DW|Zy$ZP;d%NbW?sh34?$W$7L+~TH zO;JQZv$%rq%Hd2B4@sa?cpx8h`}4-3=UGQ^8a@oDCO1fFCt=*Vi0W)DJ~!M5(GO0r%jJx>9^! z`o!JCOPXKy)u%|x;kRpwgV6CE_f{19~)H4=0$!uuK%O)_M_)IXo_9X zA%n0zfB8wie}S{~D3$W&LsTQ|Pi2^XWBX@N`BL75Y~Q6_D?MyvSJjrhOjSNkm3*qV zxmAIB-&OXkRR)TbFkr=$3vI0;lEHc~G8EyVt_!EFtQyRy!t8pvT{*ZC;T zkRuR_j>8UNowCs$_gDA&56HDs537ju!+JaRGOT#S-k+dbvvCFo7?_o^E>2PhsCYY( zYj9CzRNmf7&aQ5p$LFgOZMTiuHOEzl)^j^8 zaT}&Rt68)-4Pa0&%eeK+d840^mmC$_RS|-<@Bg{pf&BZcDodqZTPRiqk9^^yEneBW z?UYkTOf*dL%y{{0v(2BC=svw?Vs0|YqMm~MacCSXjn9l8vaPU9Fqc*Wvf^oJy%oj} zI&}<_P9SyTmboOv5o#6X)QxH+__4qePZmr#IeWhg#rYX!f9GgN(nbGEU42$lLvO!} z=gj{ph`Gw(==3$H2b!va6-t*>$yR;cerTec`ruxP^<)A+TaHkP$FBEW4R(4HXr2Fes1(}as{cxC?KgAg&zp+sc3JC9mKRdr|kfLRT0m?JwG zqMu~zVR}ZgXeeKL;u1oXT;4Qkd7~@~OxZn%n@?zZe9TG9JnyKBXQi~nsBa&*4Cafy zvGU@txd`~64FrF%!O1P2*Khue{|hMIKL0Q9;>Mo5Y)m}yKbheF+iU;h8zaSK0m|$U z%F>l1XU-4c+bLrWPqjm>6f#Y3%Hf^A+lhHm95}6*YF~v8>+VW{r(RS15R%-wHZ%tn zJvj7>viet}y2&1OBdx_HB{SWqQ~0a2>G|r;M{8bT$f}SE!*?l`mxD`#h2aE{V$9w@ zpF&m>RV*-;j$4Y9)(c2b!yRUO5-M$Kj0oTJCHba04$(=hRzo|L1AdBO+k-dFP^!== zs=_5YowixwhN3tjq`L^%c^gtG6~4+4tPzQFG{28dd4b00=;LVb*LsxHQp0*hE308x z-`QtnJM7^f=b}>LbF&V|{nlBpLGZ2KJWyhxRM)EiZN$^sgwN3(Y{vr_@axHBDKDlT2UonLPK zmk4_lm73$W$3{L|PKo-(gn=q8GdNcMuFA2DhG5Ys4-Ob2h#u%jn9!cH`m2<&F5%|jZO3S;Y`#x}QjJDv)YTrTb|u2^noNNZonCOguKlZ_g_U6h z)vUf0tVk1!Ts8-vdM@#qgxP{_J{i5CjBmA#c2U!I?x^3G7Moi&Brq275KX(jEqH&4 z3;0asJC370HGuw)^LN4U1dahn1gl8lvYNBz$q3O}!W2`p>@RozW-9l9dsGp;OH_V3RM)ddgQ* z_=6z}{x_qdDzsQnYs@T%G89~7IZ1M4R#Ah6X2zmTy7dE152bZ<7q&JsB z>yJot;e^Mce9iAr-!48X&Ksz2C{{h!J?d_}`=!c2*gN*Pp8~1CxFaZ^^0h^?wupO% zqL+X1KH=}dtIi=e3E<6m&aT(M!ADRYdN{IzmEmCz3*LWXEl|JG@6v|Kq_s+V;g{bX z79Q3F$9GxF^G-{@?^yGLTe!kQ0(zsg0G@Y=jp4g04-l)QneolL&-Bu95yg8rqnE^S zK6aG&yuI#;`;GMxeN&504&5gdeTpRExmwg$)xxwsp}!tW$Ea}^qK2~6lRIO_L$`+> z9l8^LTt`oGD(|B&C;vm(lf11s8?7`8AVDcBqBG?naZoxENt6xo7LXUg?mD=Rf{T^T63w(d}uoErfcUxN*ZRKU^46Nzcr z1%Q+DouS>)-?OyF6SrCmxcP7Zzt~}F+lcCU+HVu61blN5BOQTtT zQjiq7CwRQk65(4}*T^VY)mG5ItKZLi;eOqgof@DuOBK>mZS{LRih3wMmy8Iy|a-#>(P(LEkRUAcR(`(T_BZXNF z`R=KO{^~CTdes++q7qflL;FSw+yXmWmwU|X1K8a4s>N{~dGb1cXtr)JyonKTq+e*W zR1ks+EQLfKRws5zn`1<2;mi=o;WVzLqgx;pF;C`yz-QF8AOL1|CJdZFqZxQvbm>GG z46B}YgayjI8j^TE!}8EqCko%wt2f!#JJnOr*NA<=_hd21lIK2mIOvy*IIxm}DxI$JX$hVR9xqAt5<=Ts1%*vec87NSvi9t5~<1R*^pXgni zvvf@zJtY)Z56b(dy4K5ijK(ge`;m0bbU-f-CIAkC=w3&b2ioQ!b$)lulQ4m?+Z1b~ zE+G;)lbE=z(LBrj13G5Vbh)=l4vqjlybQ*xr|Jwjil z%47`C!-J4A);nezZ&}S@w~ii2bM;(=?rVxiuqvgXoB1)!r-O&_QR@4b!I2i+DToZj z`X_p=(7)FQ!Kd6yQXV?f5{qwGq~la^0`e7)&s*Y8dxZ7`l1$?+y^6hpvA1NIy9H>J z)qY9?RYIr!Yl2>3*`ugB=)5Zfbmx5-Y`purp%Tf0>a=O74SCL@M0;zDnUu~didP0p z-nDtZ7ByThdia6bCr1_%*i*T0hQZWO=E=f?IQ8v;t4B|!T^*Ki&VlH3V8*Hpbx?n= zQ#K`Of@NCYGEz>RhQ&+`^@|cZzMT@|CoHz!WvJ8dtdYKER(uxhQ#T5wp{l6e8|B`IB}}sW(ia}3 zbIk5HCmKX^XI}0U7?b;%QV(vHhQxk?KE#k{FVX1!8(9PMJy_4bhoj+Xs`Q_1Y(e5UfL@i@TTIWdN<6D#8 z^L*vb@y$C@uMHF>If!LKE0 zh#8#f6_j(kCVCHd`bBnC&fVK6q4kG9XVs6$t5YNUg35HT3FaR8eTNq9c$V0EzC&k1 z5>H|8Wg+J`aRcKSVbR@vjA9lMbGq^lhf)#zx7qhuehM|+iSvK&u-o6+F*D}APoW@0%J$%Ccl%>2bR#WPcado|)Ff&K&(i_{ z8*LG&k9?VmjltY*#f3!n`(4vn&iEwiX3izU5u6ZsQxMY%7Gs6&76+y=TLqUcHK!UL1pFolGH-C6V&@1tm8M~JlPTUV95mfWiHdw90JcF( z_r0NQFq?TwW(-}ghzZgpJIVm%slu?Li4?zy_rE);^;DrSl>O;=$ayAo4cvA{E$&fB z2WE~QfbJ~}{*%BATzoSr)p{Fc-tm9|gX(OZOB~cx^n&i5U6*=4#n*2XR82VXA(6Io z5Re%Rm%Clym6-l$G&MW(@_{9vaSf`*(~2>oe#*2B?7-0;%(RyJ#BALLeCuQ|S?vmW zAxStPSohlFGj9sKY&R-Bj-AuOzyv?t*+SYDN0s^J+xHo|zLD~xPeYAmKfrJ zAyZSZw9H6uMe3SSjPMl*Z@pRS0f3SK*^p{-)JrhPF3Uri(Kj_#rxQ$h3K-(r`(+&S zw;huo=yrl@4u7$;Nrn;Ly&`5()Y|)=Zo1bOM3j$~+29PM?99x@G6!(?RhMMlvkcgld!H4Id%n!@XCtRd|n)N$H1{0uZQGnl1e46XW*UqXW z1kJPHlV?$*1@l_zJM2gPj;LHr;qQlupQW5k{%egHzSsBTA^O6;N&?G&9FOqwOp@PP zZ&s~>$3&qffW2yU{vYgfv#Z0(L>4IhZjzG<&2T+r-ksHD6MS+uA3#nZVpT%_e&UYN z(nWQ4{oE? zqDG~g!4ipV)Y_l#4<}w{PeZ`!CGkUD;rKZ$Vte=_J@{+ENQ~v^w@2KDbH zcuvMUlml6;p7S1|jS0d|*;Ml%zEA_(_&(P%EV&^n?v=YorKO*$iUq#NU-|ykhNJnj zkM>4-j7%Q_L`^NNc*pNLw)W8Y|19>oD4~LkGf_Sx51kOxarM7g*I@SIsyiLi-;0>C zRmXaoX@l?Z8&$+JO=TswxZ`eLy*20`z_E zlThhmvA>sgjS<|ohb#cp9-hqK#9AJRAEkK6FD1m)Z%mw|s#|`!1>;ofA_*{e)Ir`$ z;j|bQ5SB5>k2apMwa3C&Y95_?VN^arm02pT4!Izq zYR=SuwXz?%G}@=3p#G>aeu!!u&>=ihnlky`z4=fgYqNsen5HZ(Sb-3x58MeTyQ4QA zvk<_N9Goh8h;B@jz$@vt~@=3pl)ok1%ub8dR+L@!oOv;O_&VD6E8?GDVj z_(Lkd#<V`197o&u20d=cS27q^2~U^G9xrCCyO-}Zw5{dHn?f`zJLs4 zaGyA`t)&`v5o@Y)fdgbweLvJGMF~Wp%|bwR{i=S|tIfwR6mm7OhFX3~uiuMc?i|1> zivNvd2!_)hYAanu+oxaosHr1yWuW`7N72T(|I5l$6@JI@7q8VX#4M~?m_y;|IG0a) z=!#Bs4M?&+>K^z`NZzT!u>|cE^KPX2^m#s+S$N$Xu|vIPiiB97%^@!cC6rW_r-#gYxV1^9rA5Z;K&O;BA1^J)E8~6QS zX=R$9avUT=tk775UUWPWKFOwoQ(0aJX~9O!%)jBtmcJ=OvT8L+LVo%O-5ng{>4A^8?JymUM#fvmj;El+X41LbEnt| zTBe3m6pQDf#an(nJApkswX}Cgp{VeRF67P*>35}TPwUucyik;i#y?kb$A3Rw89IiR zR6McWM%a*=O5UfYb;E-4A`PRaEB<+wC+(2<=BN8V*IBfjF; z<*g_Vz8cLiAYm@XCQ$3n^xd_&X>7ftl$7yuOR{*=p`hV+IXv-2QMXEUEE>+0Pi6>C z>!Omq{srPu_M9upGeU=mI`T2z53VHqLB(B%7uN4@W$;~4O9<^7s54(XVQ^tM40I`{ z8oU~bWp4Mp8!3yg1klmLe^^VxPPfZR(5V`wh1jJVfo@fkAI8KLnp-@11lA4ObLX@&eaHNjtUt5& z1(yxAU7Jzui9uCSo-nqz}cpd*jlN&5c$7D>>e{K|f6^sscu!jaSn zokiyo^zqo8gJpcZjEp(wj03(v*0R$b`L6E;^EZ*k`gIm!(851`6=fL+-gXh)*?j5J zH2=U0h6gLYJ^TDbGwS>}X}W0FyXwtHodtar{<1@ble(dnxDc&vhiUfauHWMb&XyhW z{8L@`0X`04wv^IteX$y!P8A?91UOkofiZ)I5cRKH2YW1e1f~x*Anoa>e%_g#mMyTZ zxHuAZfvh7J~8Y~kEbtk=1GnE@Ocwfw$@Z^?3|Hj zyTI>q;gTZ>e11kcKTQA&t;9i$z6u8>C!!Cj+8WQuBu6saSzznEnnzK_VpX6FfuirJ zFe)t9QrF*DZmwOd5?#=TxoVc@YSk@aSQRs?LL`~U{``a9#Lpb>nn&BYBwcClo6 zPf^4$f09e#`ZvV7(A-}Y^&ir3K_}|wT(xVbon#WFzJU4Fe}1ah-?I8C0$HRDryc$F zthl?bew_M)sps@ok|u#)x^h}cn_v@`V8(YZVM6{_o5d)6DX|Zs@wTL7m6r!Olt?Sc z*GB3$+nX^^HZ8$GM=1DaeP!t5-jiM2)Ad8A7`E+{jP**7Z+`>#_>`fQzpi)IuShpt ztx2IGzwEX0`pvsXy4d2?-j-tBf`Ao7?{v{Vfl&ao0G)cd_M zLs6Q>C{WC2T7_R~Jf*7#xC!RYhJ{YnTfySF`uH}OY?B!4 zJ}uF1yV5zIX6^q(bLfpPHcmAP#Kcv8{rWLr0*UWh+_r)1Kk2=HPT^VVx@&0 z0tjl@UL@n`I(UeLmSF(CZ)*^n9;MC?(5_i(lTh1*O9xkVQ!)={@>`xcE*WMjsnJQ$ z{Y`(N>8VHsy~;&OdiQf%7cf%qfQxy3>6<$bOD7vTbvXS4d3)?moi<1aUdS-wp6Jkb z`k@!-S=p~n8e9(%^Q$*-epy8!YQ7NTb4iGJ^?4fT@H@YP<8%}CbMLj(MtXp}3I!Ig zHF#cmyc1#FZx>sc8cEn5I-G_9@jcM`wG<$KLVFYRN|k$Gp6M}$iba;!kPPt*fxrDk zv`#xtUDtEjXMU`}e2i@UU8VS%kQ_J}`<;E3E1#pG%Qw?1^=VB@RO*2USr|Bx-ywgE z4yy3zynQW#AXIMNEcewfUUcJtx$M(em8PpX2s+M15E1;Bc zch52xo3>k14LqIzW&M$O!o`#aPuB%W9hf<%EJmxlnQOHOg%+O{RZkla*3tCMUVh=tbX6>)Z=kCE&jtF>N>}bQ$U>oZIRs8a9lUMe zsNp%mI-Lu;j(*QMv^^F4J-bv~cHtG%tDY@3*$(*B`X442E;K%#_q>f6sh1kyxjbjb zcj&=|!kYJ8nsn7W*iD^TaM2FNxst?7;SuiE~zV2p{G~Nv3HCzw9$h z_=%}qT?e}pvers;fhj&jN+=f#Nsa6Qe0uM}6Stgr7fjAE=g#Y8>_7F{(6#OLb4Na@xewY*Z+5XzX%-o9$3SK|sL8H`>Sg^IYQoC+ z^G>+%b#fZck`Dxm=vzktg|_z$ykV)siw43kqnCwUGZH@Cz5VNn)Sd2M0?5Bdohx?< zFJIg@Z+zzM-i5xAy=vB?U7MPAmiCW8tEKEsu4Y#R0XN#EnqZEQd|UMDbHHN+bFilN zahSq%u|GQosgb}p#$6UqBm3vRy}DF>HzXB$<&j-%emFD1Yi*2T?rcM7lBJy;qza#i(gpgmfQMk>$8f;A=@xz|;gL>P4 zfnnnc^v2;9m95p+HgLImU_Ur?i?HbGm>?#g7$>LNkIP|PJyW{M*z>Wggi4SzPkL-^xy=~+)fXxXkD~uAqo=( z65y^E5<%;udNU-6@avU(Kc11pnzZr(rClo-^tCxF^uK^KELr-^D?E9tjz%@ezPbB~ zmphi23j|03?#qAWE57G%b?&?=gq)k@HC4CrfGAdRq##1oSwklVDl92Q->W;n^mXFT zu09{^QPS|5YV=E%t-2!L?1*(Az{@p+yo=w%iG8V-VDs3c9V`B++M9@zJ}WP%c0;hP zQCh|&rC#w&?hpRfxkX!g-b^c4CZkTWe)=EPx^c6grw+0CyCRvR2ws(01kbcG;~Ajc znnM&mF3>rf<);^KK70A;!!ny*q9TO2PwEUkt9Z`YPQy+I+zJe-$Pte9HJrVSiwb9W zK0{#3U=_H~ZF_UwKW>zH;}e1OwKl*~gyha-hLU0IMN4^CI5cq}kIKZ#kL0I>3sE}d z@7kj?BcVV1Cp*D-MT3d84WyLiexy-o&1Ukrhb*ON_|h$z;u*{AEOFk(^)@B=3B&DK z)W)*y&?a>slT~p<$@YJyWdG;m4ET?6_5bve4!4%qb;J!0+T1+RjOyphypFI>-S9Ym z?m^q>(!gF(O`K1diOrX=V_#sRqf=j+3v%ooiZ_Rb%!4cI2jFz$A6Xe zYjJs6>-H}o;_Ebq?}C0UQ%x`>v9JUA;$5pW)jga5u<+{lY}w_lVx8Te68s(5C*{6~ z5q|Z>aj8$4C)xdfP}@b(cE?qT1Y{{R4_>W6Jhv>`O=r$r_2P_W^Ipmri|4Jn)gc|jlsvEm2qc=Kc zticQdvq9B!*Y91?t-`l&Y1|6t@o>2tICnS!7+7munTi+Hy`x)cYJ>!DEm-i}osg;R zK~>_t1=rh@Vl)z9+^LdYU^egW7g?}m_ViwB=lA%G&&B@9VzHqk^v4!lI1=oI62}M8 z(fwji-1X<+J(*IRd9v*)9s0S(kIL@I$ZT4+wrhR822HIW>N=K$P5Z7Pl=mY7!8nzZ zPf|Ib?{$4PU!d=!l;_ak)05w+we5x{CTE}34Ck-2QwBs^Ibc+d$7rriNxh&cP00Ve z7W(UF^;KInIfTPdIe+oa_tfvE#A64@{w??vp3qO z%$Mq`b5(>_PvwF+XU@W*Qf~<}R`5=r(q}}hG5qs1ajeWKfX%@U4eiUd43ou?z5>@l z0RQENaN%}?Qs=YmGoP+}M{vB~lC>lU3#RS4hB|X&4I1}nYKN)=O(`-7A-eX^FW0Z* z=RW&<^xS#QlDFW*2!7l?gV-(B;L7}MnXherb--1BF!PO`;uG2A8f@$S)!012o3*{x zPc%LqU7m%C=kj0l?fqoUW38ZRwT<%MVi&}e zg6@$|u5@s0>oUMt;CBTvc+}#~u+vYKn33-E=(J_KeP_=fxEdXz1~0>l^?u;0Z&e%! zpDMp_@>pNK8)W)dYed!4al-=Z+|cGFhw$9bIG3#lRLd=c^;HWbq4%?m3c`W;sX}G2)|2Wzcw3Y|=qlB!(8oq(V^Z3|?<^XuX|e zVv(Vs9dYMPH8@R60c=%#N~jjVlh2RrjiOjs)S`;JsVJB2L`!cpJ^oB1v`c55p|*eK zjUZqLfm4;SO&wDA>;#0jZOGQ}H3r`jdbI>TR+3IbE$&vHrRohwdO-)M3gJjB;1uxT zE{%*EN^3=r#O_G{^V9zv%v;;yjTgTl#R1`6f7SX!pT$ire=IrF36_F^gcfb^U(7}g zRh0yV#O3(jPjM@_H0`xs$)_5eLXIxD;`M|``Joaod7Q*82+AeFW%YlaZ$#=6H#Hct z?>&o#Zkg6?dox`kWldu%#Zc4fhJtiCGo}1}&-|&@aT^MOU?sl1X1F&Jyw_gt01f3V z+O)QzrQI#Ir=m~6O#+-@=vwH;)#V36bD26?nZ|#O3tOI^3pCYNNnSuu8+ZNA-cNi1 zN6e4pOVGVsQVdoD*j>8aWI zsKp=u0_OftoPtcT?dAp>u1_!jcaCa7cY z96vV+Mx1HJi`d~VXoD&`mJ^OGsQLEvy&Sy3HS971tFvfhjwv$b1BnKJ2FhvJTUvdw!jI5KG3KWN~42xQ~dp}(icn#^KF)tF-B;IhAOVBv&}9q z-uW;_(~i6$(lo4X}x3$Ng#WB&F;yY)e7e_y|C6r%hQbYPWc=Q0k+8B zVIKNsWnxUp6DigWY~}KtB^F-E4D2kY`xN+kB9Af<@=74x7acei{;S5ML$=BA<0JX>4{%;9^fJPhPlD ziaV+Kww*!Fp?`r#or|`&ZPe+-u2G%*XbO=hlTSM?`xj|bf3B*Vpyyo_ufwfmKx7k9 z6uZ2usr;M*6J<5?t{?C+!pqyJ? ztH*ij-WJRsn-yw;t!#f5-xM8qc?|c%N^BzT(rGWujkpCdfK#Sq;Lc}t64C^>dc96_ z(rWF()M~mbe|rhSer|M^iUs3sa@At~5Kx|F272_WgvnM72#%o0_I|9;)%pW&QBw*{ zFMUCovp&|})bVtX7`;F7&{Twh9wdkMDD=mXBdr5%hqG(%OebeXl1mEy`L_7IC28^T z?IW}bEE6^O>Oj!z5pSV~ni;h>4n90P{^4k}Evzir_JiGW>bdK3F%^z$t3?g4=O1|z zoG&4TRJsCBQKQ`v8q}DCS15CDV4~73;P(;@f z<`Q>*($drLQWNN$crkK>7@fYSqKX1W?dvJqL>L1Fh_VB^CZ z(13DzT~Kc^ue;4bQhw7)NsxOYGF+nf92YaXUUOmbQ~6scEvsvzy7@hy>E5h**Je%R zuK_KA`H=_N{|K}XfP1lB&mv}{G!I?GG-9Ljw%c?Of#)WEe)U)B%k3emifuhtze*9S ze@9x$v9oJTg*rlZQHJV6N*&zR=px@sg;?7I2?(|hY}JUtU!M2Dc!4Z@`MGy?Q@gHU z(a(!oK}YnW55|yM`eYj!uvB##JMQ`k`|dp*CROxbpelB6#mwTU-$<$u#3wwF_{h8rygddTpiYH`4|J;o#!WLK`3W2il?m zE(EOr74*7BZ}s>}nV{&e1V8;aTjIV{vyb4Q$QgcS*$&4v$nq09q=5SZoZvRX2_2Yx zdw~Ef%Eqjxu%1dm9?FA$E=a4K+X8)gs=|OFGwO=hW(n-NE96~>Fn|SC;bun-{m~uB0ohyrs=Q4fsUvvh6wBi5E zBt$9C=8HOdG$+G<&6c+lyH{*ab06%QSVU3H+tmvI#Bnku+8%s}}O(vJZiMRB;2kbBr;NEUyDP}@ErP^5& z@;vC3dDZp**5Yog_ToB)K5{JZZ050kPpS&79?%c9Ka^l{y0mCm!+jbh=xZ|iJS)*j zJX?-k@I=H}=tG7Ryb|SL^q%iT zZAz;;RWVj>5i34TwStWm|)r1$^H zLO#^NK39g^4(>Gv+rT$DjNPhM`?)p{K4W1i7{Vj$kfj-N(SmuH*!fNVm1ZQ|qD*v3 zGZK6|(|J_hVG4YuR_ONoMEJ@~bx&QdxMigi3NG#L!*zJ>Cw9oYUI^PLQFLFk0v0HA zyIYu%^ek4ND+E)`gQ`v}S})BExsjueVMe*FfkTM^oAT%TO}(|W?)$wuy?&i$3QvUr z5)T(d+1r>-1a>@;mJ#Eh9?Ygo?)gm!1V1q9G)%&M3+=mTWI37fIsS9O6;-h03to3F z8@n5XlAWx+-uOwGdX&m+l*_>#ba%a*bLT@mXait~XF`jLa(D_i+D6AEM<1dYvaDakBcfBOba2V4rBnaz*_StIyAAp8u57gV z4Eaz34AxP5{^Jr8i5(9X-Ine|JZ_dvKJ=0_EmR84SU7T;G(7rB`=hm`UBcT_y$ALR z1Q$D}1$`Gx4Mo=9auUB=UW1qOx~A!dgf1`nON^Ttoym=f5+BHy4NnO4Ogp!l@6&6X zzu9(eO2UQn;^p()32BL1^ycgc!X$E4~Gs|0$8WNKDzrIh*%&N~Z5Rh0|99IxNVn zEFGEgH}{@b7?eMQW3d2tfZ1|+ErK&BEdil(2-71$AsK{;SE#jKoJ}Ja=HM%{%trvz z^x=?JjA2GO{@blc1iVGX;}OjV+X5G4q>7St2T;0c1NpZrV0;rs>lJ}z;>PdyIG_CO zS{>*K%#o*F88u2j!cy}!fLe}DYuB<9$Nk_#6InjUrY5cu#sSH`lE?Xf2%}}pF6`k+ zjHJFO0Yzx9FORVZo^IQJoiF3m{vUpb=XfGN+0;tKP&<%rf4&ipWIPl3!hl^({a!%+ zmPs*E^hr>8OrfRL10_Y({ZaNIT1BCM>}V2EH2X+s3qTMd?D=9+UxCMLmmAx2A>(m# zeIrzn5;}#roz_hi;Wz9}>Ltim^FUZw%xdYe@k7~5Ti$fY-S#5)SBW2b zrjFz!9G|5>=OEsz8)ct;9G@ANy?cqu+V#83Q>nPJY~9`;qGBx+7x+56G|4s<8L#*YjH0`P zPncS;zWz^B6n)65eQKTm{m`@Dxr*Ucj?s3k`&6@$iV^FOjQ^Ky8v1{G0#bhX|DB1X zO4Z?!AP?rnY#)4z zas6!;OWNB}F%0vB?@=QzG2E4C?{0=1Q|Xo=be3M`%4Nr}_7((*%3XnG95QBhTD;i8 zu{Xr>ov2&#{Q^P7_>SHA;S?Ls_5SA|H ze?UhLJe$!o3CFT*ItA8CS(PWEnAoUhTv3jNZK4gnHkRY`xy(LsSD3!u*x$j7s|4MMTeg9jiH8f8YV?z(y*enzdBk$m7t zf}hz1W{U4=Dz~;&KNQ?QZJINcTagUji3ma_22v1PI)*{EoW|@ewdPX_c%-LEAphGz(cNQ1UR?0GOJ+`DWX7KoOOZiZgXH z;KxqgKlCjb(*OxQ^b=3J;o8W#E2TD4OLR&nSUH>w+qZu`yfO;%VQ+sKZB(C5(8PcX?0nBBtz8lY) zzA*k+?i%$`l@BVs{ATa$xyrNgx}=WDN`p^0Ruhn6--Wt@z9a{hOZ~7fQ@;E!#@;=i z>Hq)#e{RE^PfZzf2w_q=Ol^*%V#KfwC6!PPm6us5t29K~Yvx>9Gb_iDbRfLs)l>}S z@N!B;Et4flP8H$xd%nJ>&-?p%e}2Eu?~h&jw?8}|kNe|xzum64@XXC5k;%m3t!BuJ z(a*OXhJ((GHN&DN0uxH^I^2&MWk8UQk?-m!2kP4Yxb`9EM_2sFh0q}tV&}%Oq;BPS zJR)7?M{;TTMqK04jdy;#B=FiF)n@w#%OZ-=6JJl}t$iMNG_s(4`sC32a|h$YY%;>fTXK?$rgWwe;5y+uSgcQ+pY@a>;V@=GP@CU`Rq~3%XP6=td^8J z*L{qWQ}aHH_>E|F$qU#J4ZwsyO0jw>=0_XV$6C-fx?qR>4PD>bdh5dbWwEx!!N(Ij zy8-i9Fu;?;^TqVX5QZWZVU!@CAPLqk}5Hk$!!tQY6gv3(VgPn`WQgBV&|B;>&jZ!5N z_%ai0(5h_v-k=7PE#Epl?W$J^d7jKBtVX9bPRtQrq?hZr%;6OeqJC8V9)V)FOwF=3 z6nimJq9C?oSuVjnL&Q@*MZtYZg(AD_RmG!{clnQGn}~n8P6yvV^#*QN6va0;D~sL( z=&7RjoP1OWW4P#0PbxE7&{f*hRE{cW!%xyfXC%04n@xM+9M9<2F##FWTD0)j0$!DL zF%t`Yu762!(=~A{sDbBz8>3ekq2k}bM%9b9K4yjhU_8K~b~?puGDA^+`pKWLl{+cE z6s$GE!>uG#dOp*SA?&9zY-jPEIXyTea;chXCG3c)Ms9{k^@on{!WwM9Xp7IWN~+r6 z{1ZGo6aDyfoxp9W?TNj?5sYji@lpXsMpN2wq=uBe)3c+R1nYIPsQ%G`U|gQ}a2$+} zr*d){QDlSs&dFq?PKBv)YS+@8&wWD*?~17G^oP@CO;P{8ZuxijK=ywdo3>1k4Qx<{ z@V4su9(rYvc8dBLeJd9VRBIIUc(w%vzD;GNlAjB?sl_MB_>j!}xpkl5LyrMeSXK4+ zj)}>dy>)ZE6K=AyXLq?2Runy+F-Bxiu9<|Q+oLQh9_`3>6 z8NloDFD=~T;W2k#o*!ozWe?$~oQBQFHu6nDWw`O>)ycbp@yfPm0c~aoM!wFaGIx72 zLt0~HiqcU2_(k9cK(dvE`4z;m$;yV%j@E_aaY=iwBcY$|Ei zzSRKNlP|w(+gTw?TFYp|813d>RI%V$Kg5+9Zij1QCXO9uxz4t2Qc}Dkv&ubQwqVoP zu7a@<(^9=clw%K5@v5Ucv#uM1_P=>`s!x{-B;$LV`PLK#$MawWazS6ml&vA5n` z%PT#|P%X(~FS=pU?3+_G<6FGQN$~3}wG}gTT6?Tn%P4bYcJjq&opVK~U1klpEahqOh8M+bEV z|F|6S?S##qH=`6MsOUi1d&@a!qLa#ryH-kS`HOdTeVTVRz|0g+Jg6_UpP2Xc>fhlS zb%1hlVneHeSG}c78&|M^W@p_xcWF=xgBG_dCDNU*nnPQB5W$hBAH&YdZCG&$IG4 zhloHe9zXlYi;AvKAsax|MIBSFu%YnXMQ2brKH-)p*rJQtQNLI3xgb_x$qTpnPfXj_Qk5F)e}~|Qg2FlimM<&Z#4WjI-p=&#pN}X={dP5MjsNUikxis zA>Ytm=`RZN)o7gGz{u@Q<6AD7dy>Mi{5|}fANT!Jh;a@p0`lWF z6$j_RS>JS$ReuL&I(z70+_)nQIdq2B!CbX6&$LblUtcddHA9m~)JlNoeiUmc-7Liz zxEQ0bKd-^~lPnN)GKC=gAuu{gWw*ra9M#R6|6ax_PHLHg(2cB9vw|Bo#Km=8tl}nG zh`^g-AJO=_{uNN(s10 zENPXcgUl-*K3aR-BE(8sTVupfk4VBa`8{63C}qTO#8>KmOrFtMZb+$clF#iY74_jnHs zBf1rHd7-rg=Ex<}ANAYp1vZEJ9!B3Gz+{pz^Rz+Aj?`(qMkn|CB>y+BqI3^(o115= zNq=Zi(h}+L617tlQ?VmBIa~DO^BbN8$ADL}d7QBYw3eyWSo+x!AAJ2WTRLu6*luk8q|; z=Ue3LQa85g?JMTd@H)~aSsbp)&BSBbUMIVw`Mk9M*6#T@AD09&kk!sWEF|Aco#s%& z^U};P8TbhXpeOkAxaS1OU-nb&xgmj zU@69Gl{G&;Di!qCFE-WTnuRWuGbV}U^`O($QbFo_IF5}IhMwdrU{Nt^H=>t5a zW?zfXfAy%Tmb`5_W1|#poK|+VHS>r07|S5IqRLDlO-}dZM{eeI33QkOR+dMzp!Qa61IbL#{93$G1 zIGM--;46WIv8JoA+L(sKOS*p!9d1dU3i#kgq_%A~ieg$xv*`I~OQMU;?_7;c(4){_ z5ikKMjtNJVOn&t&<3%tHT3GafMp&+IOlJ$%cV}A|<&1M3(Ds}CArW^W$oOeibFOc2 zis2?1s7HUf$^tlIn?p}Ze&P*+sjZ|-uKsLdpO0fioa!Zrx2z*bu<~l^`I1LCI4Nv5 zKit#`_M!q`xW!(-8A{D+w!4~vJN0YY)A;0@j^Z3|OX`$oGmn5NO~J)FnM0b{9wDSi z+(ds-W2;Ewp$9Qw@Z}cjCbQM>(Gy4MSzXzzm30cxvKD==CyquaQSowD1&O+}xAT}n zp)_@=tOEvRUwOw+Rz8F2@4^{o(UaV9F4?8w+|UhRUG;YUgdCrT?MEX0NB-J`li|p>zq9tesul>ZjR6`|6*cNO4N9zde!_IXhX7cU^OB9MP^>oX zy{Fv!Y{)O2s~^RyqHh9ST{>;XPb=iCgxe=Wx-W)~{u=W1T%U3w2S$jRRsxRW!mQ5m z<&#dH9>1|Ql#=$dhvwg$*(gOYx(0V}T_x*pem~`kdDOZN;gxa5c{0AtB$TyzV&rlc2IXWi5M5xA+0HIFM%XGIc zzK+wfj1(BI#bnzI8nURICe0a-QB`;u>1u<V50F)LumBq8UP^ z1V&TE>F3W|Ri`L8mltZic30JPTx<(0%xRrCJry`10|R>*$^x#ZP=EHx?hUKqS$jo< zL`c8g$`}*D8tUENHI~}oKc9mkrJSe>swH;%k`AY|?LjanG{meBqK&!P+WW)_&4=}o zB;Zrkp6yw_)1F4!5ptfa9e$55kK&?~*R@~<*K^2Eh+{FEdV1gpFeoI{uF74qizV`w zNUp3wWyZyo2UnPa)qvx+;MTY(TCA8sB?-Ql!`hsvv)sToVuw~!Q&kASmQe7;adyfE;e?|05Q(GMeuqYZa1jPi39T*eRDOappb5=0RN~niE z7vS))a|yg44C3^oTNyGbaoAFAbaJEqAY{JS?_w0hgWQhx)Q!_#r(Bk-(g=Q5;=khV z8#Uk$vTw9)-;JWRc_hzrR|e9DKiPN)_zHQKx22;7`pOkd)&tbB-bwjCU*QZ5=s=El zXjb8l;i7C`ZFv7}z+ikuF&~g4o{hwOg*8BJ)Zom#xcv){BkSGeMO*&7o!!}-)(U3G z%B~wvBEd^Ud$pBqPY)nXaF?7fFA9wOAfR9*EZQHZO7DBS8x>hwW2Xua_RqC4hi+nN zCLI~Be96srKjFDXw765vi)&mNWIb*C9iSeV->&l8XnV6OOe@bZlu{>~sk8cKG-Kq0 z_N!CBnH!%4@%-IcDu)iICYiB}n`GtehurX|&*$k~0lheY4OYsFe!cARqGOU9H{G~4 zyF1Xcsgghw_Lh-S9MumBX8q-CyAH8wLAQm#DJWx0fb?VQD%wM)p#|WH zsa|Z?&k7(}07s_}tdp!fFv1Ca0_HZ}_qQ^rT!3+=@K;|6KqxCIX3*rM2A)H2-vvOF6&{-F}c0H^x%0=)bvSls^O8eB}dPZg{4S#CUcWZ3A?= zq-|d^o;{uofjdgRm5YmOhZ!SSa+~fI)B>*#XeNI3K$gD2NfEJzT<$s|t<`vH)DI@a ziYK!Ly`Z5mE!~;}7l2O%cx<$iy1T6MMQ{u5o#HW2Z>65Qc3vm3=S^nZ5o1U457;OO z)BCMMw7|dr_GwlZi=IpvfwL6x+Yd&I&y;Wc8;D4vl!C91frAB(egM%v+54;>H0wD@ z6Ef9p%nD-|hbe$O&y#dYO_nP&gQcAH5wEe6fV)$UYmQHZ*6&|^yl~(|s_j}#gw4c$ z(a)iLV~6|w{ESmaVv2l;3UZfWgJ2=+GI^o-mLnPB+?qzXHL}32$6L6&_QIS#2V^FE z$inpw=Y{cV@@9y8kYTFMb@B>+v{YY9sJLb`btH7)?Dl@bS`>1_&P20{8rQ8nD?0N= zsh^Fk+iW(r%@8@NB`}78gd9$W-g8BsownnT8T{|mG|+qMJcNn8<@+*5I}hf#C@6qj z2%l99b!-IAbSzdL_B$uI2XAy0lfYm7EVK=X@5<8z8t|qBLhlDUCh67l@za5m=IHol{qHrQGQ; z(nf{dg+iu$5Cl~P{d$v=g!Xfbm1$l#6xfMXZpShTuSsw-i`9$|?&26$4E>Ag-OX`w zTA9Hb5!`Gm`!k~8M2K6;P8OYFkod5;zXB=NEDdLoAy=?HM5n6K)~haUOVw%TI<@}> z5^iOq;oI?G4$=0IdtBhHqgh9u)_MYp3HNYS?8|_*GD$DvnHGtdSdu6XUnp~o^U#Zy zmsv8(R#o*i*zu8wgmII!bVIXR_A%i8%YLv1GZMj0_Y`hf@N=0M=i}}X=HMqo{jOb$y7gOTO7@OoWmMa+$lYwTK8=frLF=H?@)7}X0{P+2%!COLRm5C%RL8I5o*9q zqhXU76{fc_+q$D?HrCt{Cf;uQDSF?IyFP#e4J{1+YE2ADL#^kxhc_p8L)zu=9sCH< zIX7C);*=Mw6)rRAEsDYdIyul8N$ov8e5)sy#mTnbLP>I@cs-WH_0~J(vo?ZNOz_;1 zPqfcZ^RWI+fRvBmo;G5{}DRJ301b`#;LXJ`DwypQA#)JslH)V<^Q_Puy~=v z{h?QFKdx_2TdFym`Dn3Cb-BuV&uN|_VnF0ua4hMK{_PyllMvfDF?9LO6Tk2 zT>|4z8=2eQ1O|;v=m(9yQnN|1k(Vq8>U`V|(i}A&nwurODea6rBJk0C`Kc0bmcy&> z{T%NOt!#LHUhO53O8IR^@%f^eyiQ43_~<*&?VD&CYYCfW;hpUBYO|$ro zhQDXk)*pO>P$lR;E&2T{-0*mr%C*nYjLPIjB<5p?O5Ij{&)M01#os|EsTGW#Ray2n zl}A`1q5Uqd*P*xzcok!_k07snhcQl6hJ}Euz90MzIVtuDe795DI52LoIG%sM^b_?R z15gW#7M<1mY+X$DmEomg7;3QWD52TcJTDqc156m449BkcK~F7_@i1c+DObeh$YVPu z)}kN|pdHcADT9CIt%we+qrg8AS&^e^U1;9O#i-f7IRcdhmIk;f;NDnIyS9ERzb+$= zNd6K9g9)6M`ErmK7wW!5M5v-S31;$^DvDb`QuVq)fnkU=l{68aS6o@X%CE(W>H!QJ zr-ejLa|-^V6}|5ENK?w*3s|wOXg(*kM3~v1TGox9CmBW-u3{ABByM(j@FjrktTcdO zAgC)7OLscb_vgH_OG;Q9_+F(>=Xf9c)YWG20LMo_2Tm`| zi$C-0mWQA`MFdZ{=x%}6^4;pKfKUrL&kRu!Mr0)wXe7}ZT+*EOOC1PcF*F(oPcA0= zh+YOHOZrH8tzo!D%*$KHRk-JAga~3#loPOI2eaW`^2*)Ka35Hz^msR+><8=1jeDGA z=a?wMcJI|mc`Xz9=8i>CN$of~`Iv&S_6(8a+3BA;N)FuFVR%V0WcpM)&c?|HP)%U- z*AXmtT28|VDci~>sJOjwD(wT?Da~5lo=5G}3U=+qwH-L7F!b=fU9h+9X|qMXJlpz< zh1nFzrl2>5y}JMy&wlwU@7Fa$PRG|1F%l!&x6}J81S`Yw$fY0u0Y9JG^KY5Nf35}+ zWG})Gvd?obufNnc9ZSnfn^x*UFjtw(T)~ zau(`vzaa~#@4Ko!d7_*mVd?;OIxuXssv-&3=Bn?Jv<}M6gm(Lt08K9wYVpmiU~bXK zw$$V)zwKY{DJt}qc^{ehm0E@Y@P<1Tv!=Z*74^CqzTRglQg(i`IydDt_EYJrj}`hw zW+Qk#4#p#KOUN2{NDj;3Y4r=FQ^Q3IcXH^RbEDRuq6@()_F{VBb2py+YK&7M?5;I0 zAEME|ZoN#>B+~ujp0pLyk4l})p?gpf$pgdLm3ByMJIEWiVS!fj4UtYX?;QEVK7>-> zoa3PF>5f~;f&O&?IE&FMY#&NIwlW;C_AJ2Toz}4@L!2r!B*vDM8Zhoi`5XnsM+?`` zDO_Nw942yFAWHA>~=h!Dx-A8_w%*;Fpf0UH}A@}^c76ISAP}Kd3-{z1a2hRH) zX1TInJ12^sgD&i&SUpu+&RJ#G$)YSWcSSK4t_b%H_J$v&a*KO-O7}pG>_WXZ$i5Oh z-Ej@)l8J$~zx2S3bDQ1|E>;d(HzV6T>mq24e$Wrsb6N!yTtLSv)jv&UeiIwlhAo$G z6}>96+O%!zCmX%&n7lF=rPjTmQS3TglsaIB0KSO;)Oi|l)IBc78Q7Ms)?aYY1YC&{ zo^1v3kYUMyX+iu!U8;yKABJ0}2v3|6VnLs$rlef0{s3BJ=8|dEelQ1*+~xoE>rf7L zqng*$amLv(&Wq+pE9(KzaZBg|nUx zD!f#4WouT#xZw3ZOPxChBQ(+-h(Wk&tFG-eC3&e$=L3ni%&r^RYbrL-`|^$i{79z7h&H zGJNd$(?nHILlRVLbpER&xCr>G-5m@29AlnCKpE5=VxhjvP&uKvpyLqf6j5kWZ)V61 zckv=UUtl}u3;i5s5M#~flR4JO1BeK=End-8@mw&@FjP5Vl2y>N7vzRS12;2Vh>K$1 z!*wrL03U25t_`X(SnhohZDTv@)7cxMzv^fGs}($}uZQT@jW!&527kj^;Q%4IHaam)3c`qjH2+EFhEI9-9AuF@!EHi;RRcJuHyp-QdKxYY zLh=?@C?FF;@p|kYA5JYWUVK6HLP-=HVO-|9_c_3oYT2+~R~Gozdoyi2fvlK_+`FI= zIQlb1^E?}!!4N##2;wz`A#33DV<^e{NfIk(8`Sn^yO898^W;{O4cxwsI!f`v@KH|q z#vTQOZyH{4a;`Y%dIe(>=>k70b0jzJQkm6uH4{J4H5f=RHgq0Y>62|7oIwqCvd);j z;b-Y}`)ZI%Rb3Rp4y+JH9AQGX;;45jZhG3A+z@N~vwz}YWF&D=nEinuabB~q`KkTp z@j^FNAT-%-VcMg*U)|{sKN%WJeDvI6RW9nR@l?~88f|C?A-Tf4sVZ=iGYt4lNQ z!B>`XZ8lo@UrQ{4|Cep(|JaIn-L&`u+2DIV18e56NMme8imJOOVo+~tjqPG_iw&W6 z;soTVr&(TCuwcnW8H*+|dxhz{5#w@Q&$n+V-$Whgb^3feJZ7jc%gop@;PyTZsfotQ zVDT>EpUagRe}oRD;DHMxpTJldb<1^GFwo@L>L=%-(O>k0(VSsi9esh~h#wJ|mz0P@doc*9>W(s628ghuyvvab< z0@CxYrRun{g2P#gOT)YH9_IEArTKNz(+s{=i(kU8))7}aYIX2ps z86v3%u;G|_cuXiL1o~L%vGcF!-)7}n$;wCu!0+T|r5QV=AVVnRM<@btWwjcD98n%w z+0-6`-=_o6Qxu%8#LydzVMaHyi4cQKi=)mzFv0{m$qP-esWC#;mTf;6<4&iNk5aQ$ zqxstLC11tl6oEs=cyR-Jg7><7R~Hs_x(#ZwIU&Vi;PE5FT-Bs{9j(#$60Myd*!BtK z&`XAudY2Ui&awnV=dFCR1rY%>qtz?Js1%>AInJ%#F>gP?3#MFTp{ zbN)J2al5$dc7zPbh%_}WY`y}=R>42o^~Y=+jR*~L!5rJab^>R_PQ?#0cAwheGFDB` zz3%4O7N6bK_wvE!R=GbsYxj;7o?IQ72rxA@*=2Mvg2;Ltc-gZa2pPYRSzWFHeJa0M zrej~ITfY&=GbeqjG{8O}kD982SgvY&CtRS({O%LHsa7HYfz3(%FbH9g47{;z70<)~ci<;gr5 z9=QIKabmz$Wp`1Rx=jhX4di5;S#3RF*y-rs;bgnS-!Ph#U)I<@+GnNzqF3VXiwfg^6Xy$ ziME8}rxuRmOYT>=*;B}mA_ZZeECR)cGVftwa{|{MtQHi;uHS{42$1K-r@8MXl!jiY zFmBnC$piGzLe0mKwK1XlxU_<@F|fv4lDoOnhb8a z55g@^#5+tD>4cu{#k|a$C5gV8BacE*njZZXSG{87Y!2TTY*nL}YIxbhCUQRRVo!Zbnh@zi|1(bG!HS_^v^ifE6j}ZeTK>ncdL?vCqQM=n4CGz2x+B6s$`xRy$&1~F zZ5Axv(ry$=&7WinS^3dsP2yE&{S6$#m=dn2svS8QP&l##TdZm&+6LYx%Hx)%0l}*1OLdPfKCpByNi80oE;)yIH#8wM%L_)kawP%5~v zqP+1FFPF>IVTOg5)n5Pn)uCI?3fMbS^g-*q%`d~MJFF(lGd{}N#MyvA=Fev&Ev|ZA z=w#FAi|CXh?l?n_;}`uT&D?oV80WaeFQ$9{@@M(M*tAv*>@CMD7=VbGxi*SoX4G_l zfvN%mVrb^L!B=KwzvIYBJLv z=L#A%nz`T-%Tjyw8T)Y`030dgiU+RC_FBSYGSaaQPk@*J;vt-93W9W`Axn0 zx?0Ek-Hn$=kscS0-_+K?^XcITqkCrT z`r8_oQ9C2_?N3)&hAx5timHmED+Axd0q%hMx5A?${qmgGXG5uA*%13isZ8+)zAS0i zYEuOOqZM~?njcP)#Jtr$SiGt=C**5uPjuXQ$wIl9-MIP*@*J>k+x2nQRoZ}f>gx0f zvJ6mN9m&+CNzfAl&bh0PVMG~ye!qF|1@a#*AqF9RP-S&$n<*f|NP4t)M%(5Lg6qo0 z(#|Os^Dusv2?xWgsx}ZrxYIdAghtyg<66f)=!~wrOMz@sNaQ>Ms)C*CXCUjOdeKnI z}yt5Zd;n0eK&Md%w2L|eRa{-bveSzejnaE>VQgqf3?)bF9b4ee*2%t==} z87aUiVmMh8{e?BngV{C3$2^ZL3v%S}P#U7(%l@&B8q?J##2X+4P;SV%rWlE{)%@k{VSvLKOZHEIRdUeexWz6{{_+<@XLXbpxd0IF%II z{#;ha>!q3(DSJ*jC_XD*qXow6I5+?2MdiQGz83EP7l@)R{~rb8Uv`-%rar&Ji;_+l zJ^Hfn{q%~T?EL{bt}2?Z-!df(cN{`Ed|aALk)t#5_7MXasL;jos}VVKpsLld5jXK` zh#-XLT0Bj0VbLf?Z=T@bOM;vyZ^RUZ=3kQja<@@ds_2MU=z*ZFqEg6|moBykRDkl% zZn^h*mR(@@>y83k=1lo!QS8&5Jhj;Tr|Ri`G?AB^oT@757WEnZMgw6dmFHU9W8h8> z)Tuz&&UcJ8^c`BOHbVLKa@#Jdx?Lh*F30nG>BpDSp-cxe!#!4LD%8^__zkg=S`63|TKQ|S8y5$WfYH`mx|p1; zJQ#D%r7K?(kDa10>Pih-io4R2E3ZTwR6F$kC?4X~#Wc}62;z9(K=TxboSqU$X2{62*;5?X`I-IWo?xn`X4^|6} zYcaq1O`LRG80NIO!p~p|!Vc{3lWE_x?FRuDL$>j*hYQ6!iewZTt6{0TTgH?MJxDK>!g54>m;k zh39eP4c7la`awD?uOdHzPEt@mU#C>jdT>7~dta3iN&qks@rvDOkV&wQOkK3o7S%X_ zVSIo*@Da+PHI$)Sfm9~mir!~v%uBz?6Y3<$BZX}!wyNt2$TeyFym$NguN9}kk-r~m z556G*1Y;~(yzmxLg;+AkF4LaMTWo`>t9$slD7uGpBhFL+wgmdOk~uz*!F zi=Kp#AF11;E}BCgqrwr5&Vzpq4m@Hh^VUE?<8f;0z?&T$ZWN>sJmN=?9c51R z;vUZPNCu(Q602>R%yyu?S{Xp_qD(5CWQ`iS;9fu;_l&5;R_1Sj9o_mT`4wU-D;yz@ zx6J|Vhd`t8KaQ0f`~nkIxZWqo<#0QtNEC;gQzy~d08$H%!~p5OEpGwJ4*f(j^G3(m zx*d!bnD2UXmMC@4oAZU zwaw>Aa*svbAs5B`P-dMh<6LVjh&M9aksD9-gJQMUV*J+R0SI2rWLLOPm@A6PBww4( z5xM#|%1~duV;JX4tDzSy3=^>x_j^&KP)*kfU;RVx?p-;Id)7{b|rv zW>Fln9$-;5^I4=9U1igPt4O(JyXxW<>hr)7ijX=*<&e*g9qkQPD)0TnwQRd~aXtsY zmvC$KZ&bcc#A+MT2g~N~B#PM<#yj9(QaV;V z5P%03IN8iw>0hnhcY#mRUQBVZ7etxRgT7@$`8DQI>-B&k7S{%sSHVF># zm1P2!4^|sbO7AU}mzE4}U>C^D=6X$>9xsu-uKkJLCp(z4Peb&opG$~!CR1{*I41Yf z^RdR2pcdeDp(QKU0%zQ=X7;9t%h;ts82w=gnjgF$ONVbCliE2%8c3OG>$4{lN@e0@ zjfX2o?JG~B0&X1Sl|PKtwL~mv(Q)he3f!^+p@q#0<6b+CKZQbmUxH8bN zS$YEWl_fMP7;JevqNuc8_aonCro+s|%TxP; z@Ee-(T08NCNXzT2iT*+D>g`DZod&@rar6%Y>l*{dZh|b2`%rl?MP^aTUMc|IC#3cb ze^A7~WpVOl?mbUXQdT+@{ z9lqTY%$q47&yv!QWbvgnUVMPYt4Am+hrg*XWRF$#T{((}1JdE5W_ZO%&-6F8tv@P4 za~Avn?YzN(Bk8?MWk>UR!OlYfrM2P*(`J1`Z>kOPqc0@??8ALJ-zQV$kVkH}p3em; zv%j}_r^fX_5Rfn{IijZorI-LYCJgrp48?cArbzKFSF=8N!*{s_*iO4TJu>)K!Bdx- zIWJltdMx0}T@V^vXASNxj&A?U7i@gGh%0YCmLy*N1ce@Ya5n2{b(6@wT1LMSHtqKi z+6Y7xd&#)HHUas^JMmrdiCU)djh3|4s-0o9T+bO*X(AfgaO#A9lIlmjj@<^s?NRZWx>TgD06zsOe^nW0=3a@7ff4M<1* zm~EvK4EuqS11bbp=(n*UV&`I_0ou2uC`OzulX1g3e06k-*K)J zq1%>rYJxl06HTP$24O$@SgEYe`1iKxUTyc)uM;#q%qzZCcIp|Js|L5*hVT@H-h1U> zn(Z2BH%m5D3+cDo)_D?~&a_@}^Slyx`*gr;E+eVVS}Xzggb{ZDnOz?!rt!AD!SxQx zr`saE-Ou-DZx}J1I?ySOvkm^8v6dBEPkQQjFx~7Lu9^V9@l*;n4sOcns#UaQH&wp= zDRZyex|~9?fe-U-uV9t2s<_Vdh`+9%T_bJ&|?>kn%1dQUJ}+)#tL=z|sL zSzXeRGaU8osQAzf7QoX(b*M>n)S9_E1NF`KBT+_*TBnFbPUzve_N;7!x~QH1Gt_kevXdnLAI^51bdr4*@OTmWpB=d-}dMC5nG4YPWv+H=wd zUR_OQ#k6Og#?*8-4A~c<;eV__wchggoYtu+e+||IujmtWqgk%!s)f;u_jb17v6qU1 z3|+wCTE==V#r=f!pL5Pjw?9zyO0>9ZwzK}5ZySsYu=Bl@l~0&uyTKmp>nGn^CkHW1 z7)0x-1A)Ge+=GWf^Z)YDExB>m7pF_ZK%v#a@RoQmf`QfM%TDO~B8Cm;`g1%@q{&l0 zX1QS5-YLGgAKEzUb5d^BOFLYSu_g0YScO}{TyIMdxwvdUOLhvsu|!pv;zIyUSLe&D z)j&+9g||9U7X&S;9v7q5-3mGY%}dH8iBgfssr6R=srE1QALT~Q+Nj8^n9PPy3FbRe zMwt`a-unE)%4Bl({VnPCzc|ScQ0EA&jM3{JuxQsQdH#e``ufUpf@ zdm`p?K;CwHjr|SAYn#{WABp8Ur#a7V-gv(zA%jO|?SFe@^8c$N6F^7*Pt6s2@Dj*T zt{ME*h`h7U`evGQ!6O@M?}$n{Ia$XDSh7D0LjSmw`gg*$zcQ5~IRJyH-hbpnRP3_1RaZ2YK-LEfbcQgV*$?QTq^P@{7eO?}W}z7I8Pjf7l)RL&rP9hnd_w*zK>M&WCPa6oD1azN|BtPUkhi zf>S`JY$b28(qsDfUq>{87Jqz2_?vuHb4X)ZSsVDOlHG<-KF-j4Q{d}lE<%KJfH)z& z5H&aJwnz4alf1LMtg_JZaSQSW=u~;)slw6wd zpSLA{`H#Qqy)QKEcAIEW>4uuhrihbHuiiWSq#`&5xRhZK9Eh5C{m~7eN8v}FvaZdh zda~}iVt^~Gcr@duaJpL-9)HOqres&F`ABQV%oImvWd8St*@zXAi+jlJHK zuuqmZ)ziMdJWr|J%8hef}idcE~jth%E+m_dvx@%+=xa5>hyMBdgX(YM>Ne3 zA2gdL2Dw%6mNkgImn;r`FTWZLSHqMZ>6TV;^& zawjw;+ernAl2fbMKEPdA9&k0o%n3hVvHS?)|AHzT9c#R{TV`b{pE5x7Seg<4LgGu1ZIc4B=tICC@u>nTXX2HPDKT zfu*w16?TT7l42rHM)(;z{6S85QsV+B#!?4)QNS~;iS^T`fOwCOgo~M*TlEt8vQOE| ztLPF)Jc1qVVYuLGcH9=<&JuIOIXU|w9T*|uNXS47mM@FN3CrL-x+1X%U)Jki=eL=b zH;_f&E&QTlZgLttE06DQzyMjf6=?#}&#cLU}Xj7@KyKGZl00wu6OB(7>Eho|aP zm9Q69P1#<7QUp>0m(`Z4c66=V#a^gfT8+6eY1_6Zo%uKLJo-~&Wv@(sDl;e77*TBW z=+x`Oc3)1Y618<2exo%Y9o2$MpLeHY^%5&XSo0RJ)1_o({-QyVL$WnUNt^AQ^|qmb zM^_&A`0wmzziYk^K0dA%d^DEEaTA_K+}*m76arF&H!54v4#fNo?1FI~$`tRy^y&J| z3ztM*4?A7XnhM@NqB0*S^x#K_+_sD==6143I52+5Nw~b6pJLuMV}|_RW@T)n_lvqh zN@OXiMib?TYJCH~zbdfP9K1(fvtCkx65pVwVp|V1C1%s892M+-rT-BF^hSJU^8p#mjmjQIi^%Cyc>$jcov5&%bqS_*nro|C0|5cm^s zKO?^q$47DKR`(Cxww91*)c4FwZXn?hA*9DiZ&l2|Zx=r;K~8+53cD>D;%JMQ=ip0| z=VDDM{T7l<(j!5g4$sUd5H`SfmBAi?{LWsRp_{}LM zkfhLte@OwynBBULj0=AjP`Cqywwk_&7oS%2moeI>yp5UprqQTS5)!qW$#z|!2>Nes zBWUahE>PL)OYXL^tX5v=FCcs_v6BaNBk_nRIa%Uzp>sG1W?XrM&32LC31Y)ycKcc%_wWp^E2hmerSBisW z2U`v?6=WxlT6iX`0k=h4diK zw(mSi7T{I%U&p!0!>94`_^G!In5t6MR7B&`f4-m05+!`cYj;nh+u6z&p9SJku($W@ z5QP)US?%d^C%?drjRgGet}J@rAO5qH(AFj`!xQfW>)mF0n>fO$<)D2uBOC@TVEaXfMq3`qc`Qv-JeBZy%_wxN*et+T*YYkP}SjaU84Hg8xBoH5?r$ccP$s;GQ zU`$9URLm^7j+|YVKSs+R$hNif5!s*YJ5LXeBBdS8;A&@@fd`dauc>OhK4;sM0;EO^ zew6WiObKxG56-II4|m-1ymw3mGzNqnZTux3$Igr{6T!B@^@tSe!MT$Rm=F(r2fs< zeD$p-`e!b`kPO^DmGzQbyFIl837;>Q&{;yRRLPvg!Vdab|Gqx8>MD9PqZ12w`MvVG z4n}M%HcqGim2s_CDhx87@tfyKO~m?<1{7<;tiT#6Nq>i38t!XvN%U)D?ye$R0k|Ks zy(IE>XCELtkI}z4MbNhcrEQPDqQXIfCQfFs?8^BQHG2)gH-cp-{zl5%=0=d-tk1h! z+6Cj)kiAy|ziSlxz~t}-hvUi)*G>{FefvDAxpqQ4g{ICq4O%`8oc4rhgD~G_I0CZ6 zB0@|N5?R7*Hi&6J!I#QU*~VC*Aus`*x@<0W*&tq9=lrx=+}3uyV%f_o71|z|#Vw1I zU+RO+K9 zaK(av?6-EL_g?0~bd-kr0S6M}7h>eaev`rCRG>_x1ob0DGD?LB{`*L8u;f;=2QbeJ z^;Wcp0l^Gq++#;cAxoQ{U~Q87>b-d4S+443eLnxZhLmUooO-EG^LKv5>l}itPK1t!ZNi#ua{hVW?Tlha5LwYiNpta+ z4Orz#;{A^-t3AFLg*mf9c=x8-9C@KN0%T02yr zJ>gce|1kCUKnIEMYNqh&O+BP`)qUCxyx3xnC$(|dmN~s}+Gt_gclGwnNKddy02|{8 zT4A!f-rJ3WQ!sMMDbkGBY&SAC;4n%?EuP0tIJ+8JNSlS#OQl zi~jzS^~Tg<>1R4p%6>OlTLj#^Nx>iJJLsJ`8YBF!-DkX-B!ii#*QI)DXk~yZ$=zNgrsK%Q@WsmTANvD$E#J>HZ&AGJn>kUJrLsSdbD7R{ zxj7?Y*?bmM-F#l1H;`_jZM?rWtx2(xl>KI}(HJsqYFaOwn5db15v#*wzSq~O`Efbg zt%QXfN={|Sn#(-ruzln<_`iSMi^1YHZ9{DhN-)UM5xj!F+gI%g0_tKi;(hB_PNrKc zvK??}(AXHIO34<}!zsi3ca$ITn|eqZ#FCV^_ptg_Gq$DUu++K-W>OD=>M;w>)1ld+ zsVz!QZ5JHys)u28i54Lms3x=~LdmceE}F5T;$G~`YY>3j$F%hCFQne3l9b~4cHvgk zKG=x_0lE-1rj)0$G@6~>LtJ&$1VApNpg#{YL3L+30qDGGy0v|Do-<#dit6h^r!M`FvBgh7G-z4hV`R zP5_p=$_)47aR$Ng=Pm*_Vgm)F9K2A0cnwRJ7Hm`|9$!JOD^Xha(4r^sh~3MH zRVH(*l_MSJ4x#aXfaJ_$u*<3wQm$qLFE`r0WJ?WnLW~a&07jWv!3O>0Ni4QqG%B=4 zF?Aqy^ufO8-YR0xB2wrD z3&wDV8>-2XdQb^5tCzW#ym9@e%ozNy>gzIJqMU*qOKUWK{X6Jh{C^I*uK(?z`(H!v ztxTZNA8BIfJNm>~7LVAno&(n(5an>E)(p6hzg|}Wt>I_b zxaKcoj#84t&@B)keAillv>aF!a*7%#lyF9lky-OJsL$k9{YSGYo?h&+uq z_e#cUL%%8im-JOl{C8eLN+v9njpF_9pz;+NF#oNX^ z!ahI3Il+{iiG6Ax%VXGg2&gFrkJ{oNSp8T08iy7N5x0Jx!HPrTqL(Wcb1%F2R2hAQ zvo#Eyzcg1J3Vw6uLC;z*C!DqM#F2b~ZL`a1=pL~f-`O;fZrZ#Lz^H|dI#6TBnH>0+ z!tJ|3{cg>3vW=1KCCUGH2Tv`Z!gDhw$VUCeD|Y*M3;GoD=bRjRcEnFvU0T>RjA|HP z43O`0wKZCdn|9IAYqexUmFvyZ+R^1WCv*8G5vldjBC^CM^uwd6=<{wQ)xtW)i8`Zj4o@2*G3Ns>7T;R}JYV*x)e zZzx@))?*5sRqgyzR6GjAa1|0jpjlQU$Dl^#5h#0$$u|@|uM^b8>bM8Ob*SU{bKKt_ z!NCs3$iN~Noa4vFOM3V+i4*-yiD8*gNZYM31*2{pk#=9dstX7IM$fqzIR>!hOXPE@R7|i-mY;Ya-d|7h3~t^(}yO>Bzohag`K1>=y%lz6k!r5NlXijDYWA&gbss{jQTnfC4y*NZGw6eOr$FFfUjFJKd7Pkq zNvkGQ-i>lBCYKd{2>bS8yQ4Ve%pYLCSlEecn80<91Q5iw-D*!=Ovzm7PW47&n?iCO zU*dP$B6>KrQF=}CbYWV$<#iUk3>Y_B>(5wg3K_~dM)S@$#R48sr;WL=qol3a?)z36 z$@yChuEOnNvEEHpI(^(u(eG> zh-ku_OKl+<;2_7ic)tsHX>5}rD_sOjCmA5R)ajFd6w%6@0bA=&7w*1UvM&t}AOg)L z-9A*5h=AzTwicD-x*+f|9M`RZw-bExv`5Sf)Vx?COvX- z){pvxS}yzNkPlfa`Ps$g@*bm3_#CGoQpU-s4VL<`9S9jd(36N2KTA7AH0Kihv?xU47FwnAE*bvqOVCi=^;lgG3Qg(`|%1Vln*NjT;8QJ$H~ z4t^7uRc(7*!D@;ixv-rn?UrZo5p#j`LW>a#D6s@GwwHLFq3fLgr_G=o`F}$bxxcdk zP7f@mEms>cVioWDN2AsUbK^k=!94-19cKq_MCnkzTHRtD6NH0WkIg$^`ND-#K2sDU zz$JjeS}%OZXcW%I;i_e^&y8(Ow>rUu?FZF4Ji6~XSA$5u-OwgeAX%#sxKJS<(}oeg zsCM0cvmF2f3l54c*iMUOH3A>DHt^owk4TxtOSjRmrF%W4sGh9s;Xw8w>sKKZnjUm_ zhXjyv(-e0oX9DCxZ|x<;mZiHVvLvhk>DKGhnQl}f*~OPk6cNt_Eyv#Q_!CvuONC9fP`k!a1~KV}_&H z14N|Z|#!~n2mf|CkpH)dG7v92%0Ct zQ)R8{*9N*;T9CS5=*oLKg;PhY874ckl}A z(e-GBIk=)hxkfN6t?-!nyf>6=C7HNVEq5JMFm-AA)MFP?NF<@ivBqmVpSeSQR^?MW%H&!}sR0V%K?B3gE@BvlUVa zpwM1f8i<5j)V0G=x>`&I&E#KnX%vE)$XaC7pefES;jWEvSg{Xa3&Ag!jcuoZ0l}8@ z=3*ORRJgjv#*b||1;UEb0I&+Dh*Nuy+z|!Vm`MSVLUi=R>NxZ8bO0fmdC3(fDXe9W zN%8xu{25Ay4no@VIRhWGt9C>Ig(UYKT*-?(yJtgVF_s zh2b~^pOx0t$5ibmm-d&jlXDS}1a zqmazda+JoHhZ9EWNT$ENLUI57lO2wAof&||`|x4uPi5pLrHzJJD->?RRZu>f^N2Wr zn99Led#|TkXMzx>%l0mB^8U$Yc5LNzs zQFn=b^V;QSfvWnZ8y`pN=`0;AQiFN9qIAa?YvE@H>D#HGE zy6ivFW;-%LS@>P;=-%n6FAwkis`Nf$C{qw;7a}vAh{} zjkN!`ikgi*mBe!sT7W$0yS1&|?vNs`PYP;+3Bv0c@L7ah0()-@KUSCtxb+|mHlsZ> zkw0*g=wM%&L`A*doQLM`H5J5sspmWN@$Rx39v4$FHz}a2EirbY-eQ#{Rt2nfP8{dMdxe3C?I`@SNb_XRv>Zk)W|S;XJNA$S^N;J+>jErDV^SW+4VHAGd!zyV zJ-x`vMgJJ<;+JK^Y%672cxk#UvDhJic<3kieC<}B`Y`2lOw$%@f#^~9$$*E!lD`|< z8Vz!M7dtI)de%w$`%=G?#_wOj^uX1Y&jl##iHaf24rXvy2rIkuv?I64nef~gMv~l| z;Q`297m-DWHVh6Af^DT6qwED=a>&~ESR$JC{NkRX)>dE(G4lDizib5})wXM5)g5|P zXAUA5Vs48DS-U>hPp0)wxtuCVZCe|F`mq*_X2RNV*UBF?PC8lcNlOLSb(b#pq|Kqf zv|%>Z-8bKN7!f!z6s{i+xFWTH1xuP)QnG{0Ru`+acj*RL%YL%90J~?sKT?FaKG3O5 z&?*G{WwBsjOecs>v~N-YOzdmQTe>leWGgTYaR5sU8i-N4x;KE8Fl)Nk-qU`M_CSgRLEE}+G33*_b94=u?{D7SGC;)Pm={Ij{ zj1_;!qZS^Bh3Bn|KO7E3cDqI{8!KtvPSq##^7hLqUGoI%?)6>#!Yad3uDNQxREws) zk2Lbq{yqNCmZgX3c85kG4q&wMXqP-P#K*R`n~^by7RoxYKL_h8?;O8IbNNQt$r>l? z{x#8mUqb^7MaW?%mc~oH_H=R4KX-S15MAX$$uX;Yf%SnS@)}scy-GmF;c?sf}xxYTf1;?k6L2REZv$~)zgQAxr#U{gC)K< z)6QES7?rk9R~7I6YqA9AQVrWv!{0QcUr_)mN;@tI6(s^ctKS5UK~~XI*|BgcFhjEou~dsIL>-QhDg=3gK|{F{{?8 zcC>Mv&Qgi4+7eGgt*khsSFAAI7WRNd@@S(`vT&7}EmKFCc7DtP!e3Lm3rigE>f9=T z7Uz$pt$JQ_w)&QVDHHiNCpIh`WIO)UJZX8==V6mYvBNEMQC(`W18{e|!w|eS{lFqS z)rxCCt9bJzdS0sY;Sh$BzlOgApZFmY-?o{n-|ng2#0?K~1=d{D^2M)_W#4ktI_RA^ zx)S!N>rStpCw4O z>mh!+Qyj%csUGvTdXPd6q2zr7)qy^pb!nZ#c--~WZz-sVEnX>n^(Y#g_Asb~@+oK0`En>jY^)t=c zyU=2dPlvFHM&^rREZ70$K&n@8eQ7^875g3?Aq`0sa0iu=L z>%l)Kni^xobsRV{ql*NaQ=iX#O-Jv*{(wJ$s`-Yndi@$ob-(xYON@+k$#)T&C!A$L=i2E}x2*8fMltr?8biak5G`S+ zFTC|gJ!>_bTqJ6W>C#ZUEG>m3n^3?zQBbewbVqp3=f5|Q>=r3Txix$`WR?ykpQ(ZCm3$9}_>c7nEuijt#rGnmnaYIdp2 zXZl&pYfpx?@4UEI@1v=LGAA4e2w()Mjp%h46@Ne4Xu??hsXsyjVSyni3mzG4H4&G! z!K|PghOmC1dk@d2897vSO*u%ILzXxzGvXV1kewjnKrlovRrusxv6C3tLjuv1F%r|D z&Fgsvos{Je%?+T3_y)N`Rl$svsbhB7t5kRw*bKSZlTUUUn-fb>x}aOW84gwup+5}C zg5-`4_m68~)Y3qa0qPY=dYO|nF>MX+j~>(=5dI2MJ?(Rz)1o-aqMp$ zdCxPQvzd+RjRy(aPMEcH>nnSz6)5HKmexJ)TvhfQEx&bb_}gM?INn|Z!X=$aO}PRM z50a8aZj>wSas_%x*%&XrTylPUaO5`G*X%cfyj0L{E@uXnq$THdh*6QaGt_GI?!V^#ChsuTj{xwu!62^5CWTF135tLl_>d=w2l}r7`)blWltNr;`jHS{jTp#Za z+ZY@uuf?iYn|u$}0m(1`qsG$BK8FTca11$^2uVC5uXoB_Z) z*xOza-dn2rY%;QZ#JYFnd7V^-wY^B}v?r%$mh@~@)z&Cyu-w@Sbj0gSM!Pb1Vy8U* znR5xhFNBsI(F`2SjxDPlFK|)R!1PlFGHs^lmqF(Q!d(EK5QB@ruU*s+8lvhvuK^_& zs`VtrLZN5dv3OnhLYd~0q5*?eSO$BUH$mYZG=xKHjNBlLGM6VE{i&orSaytyx({i2 zDpRcQl{_^w22xLP5>E)EHFDxX$i)ptlqin4`3^7R3wckw^sXMRib=@c~(Eld!H%i;ascv9X}1PWsvq za@Xw98YxCpSc7;XSic4s@c9r>)`)J(4ucl8FbhjimKU|iY7Rnj7%CqGZ;lq^iAbr8 z=4%XN1Yd5D-=!Lw^ghwbD$G4`If0WcVV;aC+u?=%dz*KkEHhc{gVgKzD;?s4FppxFK`OCxVhW#r zc74%yIFKWBG7nLmu<8UIi>|N?h!9Z-uRxIf`9!dxXBwqm4VZ;;VW7X|yS?T9(BCaf zB`SPynig{pNFM9+-FqMeLTxuGoWN;#+D4qR4hf4z4nL zdj{+UI{h&SuS8CKr2}0mpcbQ`p|jXWdVdA>Q0$&nDr6F_7-08Xmu)2DZa=j;$~f^z zER2=%I5P@R+nGfITpLlC)@6$?v1wM4I)gVR>in0gtR=w;cqz=bcU1&(F^Fk%h^_XN{msa=s|1V## zRsW060COK!-zPqm`RocmNV4Kw18Y}}!AE{bwL8jqEKbCOYmlgZQ>$^PQN@$e@H@3a zf~r&ggA`wu8WRNqLlV1mwU;K`#`Guf*N-vF7~&(D*H4R#B^MYRUhDw*-61NHcaBV( z%QPG83TSs?9M}euFRk`$Xdm3YK#(jsDCd88oN0J+JQ5^DnMMXS#Nx_xvYfvxwfiF% zo90U6XK8wJqxPzhQjE3^)val7)Zu*?L-TB-c0l8!i%%XfHq(xt?$t%Hq2_aqET9|- zX3wvcD^S&*@g6h=&{C9cZ29Q7n5$jZgJKav7T1y9+u*219=yAN5i9S`bwNno(^w;U zVU04IOeGT(ZGL-41nnc}ztwLjg;6+ZZkbUmxNzt;dgydV;nlRvC`_4T)D*tUq+8Mk z9etDzBww!dGLJ1M32@;yy%DZjPQw)Uw8s1pitD{ySD|c?H4eoywfz1Q%zWWh0sdg69iuW+!$;Fz{_Ry9F5*6OzAVWn2g26XW*=29M>3RYd1 zP5o8X-%F!pq$TdKD^yQ+HA~&WarDS7%ObOE2Q?!?8Zl{@PB^172H`#>4e)AI#`uAQ zIejXN{baJtNsu~V1PIY-$yC{9mvsaHG9ym0Dlvt&xWJcVv>vWA#3fHuw-LPd4iZq9>cwI5ln+5 zmw?YV$VYGkM&91qt>p?CC@uQbH#Ga_| zy)m+m1B(Sy>mJArMttcvJGfY`@hdf#r(vMWHOUo55MWC&SQNT}&lsg`hRT)B0dVjY zo?mhK=G^m*8H$`pYX>APA_?j~zSAn_(R!#KL(ugSNko%W$UFJK~?*?%HFb>*&o%MKglksc0{ELl7^y)ZS zX@jcZ!t->Z=A?nOrwg%fV#^yf7@%m1_y(%iAfeP&pelN5FLYY>U@;cR*7J9}TkR5= zpy(GkSoXx+`|OU+UUWJDkmyld757?E-X$*pkdZNQ*W*U&Lzc&xH+qwlL~v1JVYU~S zFr4(xmR7LRjw}t}Apf%Lvl)7=Wm->N1{oJ&UBmhmU0oF=r-xq~OND8@!lMoKr+Iyv zukXut7x$EQ4Eg_V9L-kQ93Q9Tre83O!P%Ip6m_qAo;WpYh4nWWF27zxCcWh1SEk0H*$eAY9LHz96z(%brLAmeIS-E0x zwf7L+C%d&!5+XTV-r)Jp-FV`LR-KBXF|JEV? z#Y!sgsEmikV5-7pW5SCXUc;J@$f{6$FSZiuMvXI zr*p^K39sTnpR7I2;+J5F6%9x(`H>+^CD0reL}NR9UGTDYE8qZ1-1Wh@fxaXwonl-h7}C77MsS1x$!V z{y@=GujT8T-)gx>>~n-tA%m1FcmIjR_EviqCHo2(%T+P8!+<$Ssmdh#>J%RJLA^S} zj6v^fNV)ac*E=HMpX>#LeXo?FQb>4$sNU3Iwke_!4E^@Wq8ab-R|hgKha2@5IEPWQ z6`U|}Q;vw~k1Tj9(M7u5u1OoimTqO)TxpH=m~U9H1?kC7PaavV^Sk3PvKE8D$R7zs4vHbm3@l8-M>qITHfTzdQO~2VMP^M>Xe=ZmiVo0(t z$veNRR&hl6<{wcGF@a%CVtwzbqq@=-nTo5*ulD?S98 zOT=}cq`1-ug(ztw={0hzPBNNhIB7pl<|mcp*p7=x3C#O6jTTMKe|ttWn!m89Tj>5s zSw;*mwW_<}r8;CT&bxMU9)paQ2|A^vSEsMTv(VIh>&cx%iQ{$S9Wo{-9>*P&l9ra} zyy$xF-jO_NcC_z!M$0L~$kBgnV#lb$0mbN#rfMu8MwZvt0yP@u1w3jHCoXf0bjg)h zhu@1kn|z)_SGM75ov@wCw=rzjw$%^d&dgDIO&4@h#_e+Y%L@J=aoVM0;Oj!EStj9^ z#p*pn30S7TjQ1udBrd3-A#;~%}_vx1d`s#V{v7* z>%55M$g*siVUqh!!NiX+4(_7*YZ}fuIbZxpDg@op;+MZr5*hO~-H6OpE{8Nb6H1dZ zkbuW=u*!JaQ6mPq!a9=+>9N#w?d0mh(26MYJR0w%^-?QZG@~h(6(ob(W8tx=7A}}G z`BLG;%d@9A`xWe}03>OV`9Ad22Vr}li5x)c%g1uQ(s6;NeA?_!v)2G zM;i_q!+k={0eI39M+V<&SpZ<<;MV$`!%K+5wHhwEyoy<;V2%ZwbREEkF=p6bffclJqT!WM}gQMf6D5bN6~B7@gO zRc>$ab|n@)Np`kt4jxheLbG49(U(>~@jsYv3t#?sDEC{rnSlPKclb#?Sn$UK`Pk|H z`<444(5jPim6&O7UhITt1_p$LX8Am-B=Cv!>t0v2?@hQ!U0Rv;+>OnT1qFLROcax` zuM_)2|8vygGZZ<+H!4D01%MS1X5PtD;rA)2L$HoeL~KU=!3A z29_QiJ$i>!*mX?IOC5VIBC z)`b}YGrsc!jp(0xegrrI7*1-L>H3AZ@~s!wtS8WY0rn=^MyF+w#7tFtNUryZ6lV;o zBZ0*T{0=$?FeYjsyRGUgy2T~AZtf5nIZl?#V?-aN*@DeVYl%}w3tkFze)nnOKN=E# zv8C*eICgTKlj~%*gH%wnNdCaWmnTRKMl2^o8_YzNDDH}kSm!u{ff%Vn^~ zX&^<~e0$c*o3CQ|?mPAlgB`3NTRZGaEn`9bOpW=%nb&()6*TcJeg>PLWMc?Zj@aHW zh%18JP{_~Y0#v)vH6U(1&nct@yyM!vc!7rK2QX5@jh*eg(NANHjo|{vk&H=8CA_6h z5G`@JY=<`UIj?U7k6q7n`eaea5)C#>>{OIbj*<;08usq~JHD&K87B(ps}$c|E<3NC zYd7ko8Qs+bSuVQ`y9eUk2_5!apK~fw13r!+xJi9=n8+yQ#sLFtWnN(P_&@u!TbaA( z6&s4}-$g#q(7j$Gul(aJ-ZI@|a04}ZN<6J{!gNPD7itUKv$&gKo1%0JZfZ1oH!?Z! z)cLC~{{YByJ8r}U{_Xo_n^aZf@itkvXVffqz+|%OV3H~6>3YXq;#qB|M~^iu>zB)s zRZr0yq}_Lj)!~Exod3JZcldQJw}BtDqTBz^SEVE@&s^$O?x|rbqG*@Fqa!&_u7WOt zBfd1+N=z3EP$gD+o|?CrV^&GM*YND89}1}&0+ ze`k!HPeF{ay}L^|v(4!l@1ggXmtRia`53LDcYHHp`qZ`!%zUgN>hxa?dp?Yv4gLc> zx9Xk1(*o>!K`-P9FjS~Nji?zmC&F*`O`nZU<@99+UId4ZLIn|(i>3NM2%K0Tr^5!F zZ7W&)nk<*Dt+P1oRVwkJ`GiIY&huC0Li*2VAOqq325wJNd&3{#UgRQY1KQwztdBmE ztR|(tgYgG=TMrVXA03m;nIN1!VU*VH+Q=<@F_Zc3_DsFmXyB`bwU850ps$|#uv6s{uldFMEHF)?VA7YeTyghux3id;ih!YXsu^s~+==qt ze-bPF!f`n@BC3_vG#mJ()Y|}gf&0x#@@2(KnvK2=Q=AQ>3@mU|qU<;IpjvO?3+;xr z`&kd12tUEO1o1X?v!M(slX8_YumbDAmZgKlpGJe|L5i$MpzaewlgnE|52n2KhpbP{ zwT4zSATj~E#03gaJ5sjH%R5>%la1q)+Fd#@TEg=Ewa;U??o;utzcb2*4UOtS;oo@w zNn|Cv@>1&2Eyd%@_di(+(Q#$L3|VkPU-hN%-HPHq1Vl`nElc~mG+@?oPdfl)aLzB4 zooA0p0?E>PXATd*>nY7ls{7QT?@@ihBcBV*t{g_$?-^Sth;@m1SAI z-yQ}-AGgksp0qu7@KSosf zXl(m#uk11YIQ3g}pDQd?DspMsDMD=|3{`WO9yz*Grw{s+PdbxrBL5}37D#Fda!}`%~DXLn83^lCE?9VQqm7Ee>7%3N#bla{Y2~WDLcllm|PNpNH7VJT%UJ;2qct(W{ zU1$_MF{qy_=MH9Pd}h{_C>yQ1g2k@n!|U#ka})R^cc@tOt&Ol(nlJ^^taeV2IAcEqR@;;#vVNzAiR)GUB)QJDJZ+T$!j~h`D1^e-@M6bD8K2l z^3MmSZ?dAFr_}TYejY%+QOV6R0`aTRx0V{=)S)N$@nB^%N#nuYZ+<1X*;m#3b`0Og zJHF+Xt9kqZeBBH%-#L+**EsFpLfJ`B#>5)d!^y zc-KE+UOGi<+3oJBo9Zddwwv73ZE;vHMScJAnjO(M8~HKsQ+_42%pa-&V4Kn@OGd7P zX_F^b3+Ag%_B}gEED4n#;)>=-?0v*-E=>xe^Z+RCLpXgBy=nVZu=C}%PY2Fx9$kHP z{mSp!VXYZ&-vx+h3e+Y+{naDOZ-J{&#^17#epN?E8&K0{>uQts9 ziGlIY6|O(Cnp6C_ck5ivd2?m>sc&~uw=^CK{ebzpNH>Z%NK6+|QfRn&)5X1M{#a(v zv&NzQ7s|X^dfxXSAV0-KFV)kMK>xO+;3Ykdc0**(o^fPSd#8=7=^?YHgbn+Hd>=V& zzkw{*{eRDH^*!`mUA26-w0(;wSfbq>fwY9HX$~f9PLaHB-9rbv7@lnZUe>p*MkHq~ z*=y>H;ry=pHd&dlb-x=S@=UzaJVp zkMACddUrXKvXVYHRjRLYJA2)JnWbzh z6?_YS76gW@c17WC2>h-_R&b{J3`%ofd>TLd2e`$EJhI#pQ0z!LmUU|_Dgj1cIejDZ z^tW;PYl&@f+Z4=X^|{N6-%?Dm<{q>b%lPRM%n~|{vSK&L)aH}o%BC^q{E)>e=pm83 zboF#A1BQ@C@}g35?c0y+p+*mt+Ih^)DJ1!%OrCu92l%w-3b6%j_4{WvAtve6AK=W0 zaI4E#6I&|cyc2$~AmQr^zW;f}`=5pqg?)p#H!y7Hcs>}e{m;gO|9QC9y$@oeTF#Qb z5s?^J?9f1l4IK2A33Oqfv}m34^pJkcyK?P?=RtL1?Kkdv|Xkg+7p>}lxfm^$`lH%`_iyr)rAJEE_p z{qC%%SXqAVdx@zNJ4~AQX-+vDcGV1Jxl_6UB%+X`mS4yc(av;?qcKBxC8 z#bYX<3=gRK9#y@zwD#j%gv_%(5rdOuVlGckz>r|QQMQIIn_#876D=NTTp9>1b6b!e z!MAG_U_=bTx`}6VJSoqJSky9SY+2>fK!Sejh7n#w2DG4;YS{$Tt%T>CqzbBSo~5;* zXY9ZL;dhEf8Pm`XP`aKxH|23)4}&l1zgxDysA;^n*+iK}?BpzvLe2vi;P|4$C+=%v z=9#EA7AKJ*z--k4#ilEdKv(6wlfXQVQ`7(lpP~)BX!`A!RHE&(t+;CAMYIHPzUsb1 zI_7NWdxj>|bV0L=G2)hz&^Bz3c&VP915r%6vs9N-&gmy&+8(mGS7vSi#$1@O$JirH=+qmRywk82Z&7~WM{KJB<*Vf zBznXdD)?bScYlkKv^0nu=YDCIdN8a2R-u22>) zdA(eM`MZXh9_LV~wXN+t=ivmQB67K_+cFJwgr#W2v0X`Nj$F%q`wbcE>wU1APhD6J z5Wl%#a8TXUCrV|(2@+t`yVnK-K(^}0`!Tu;?6t&>Z>t^<>++717Q8FFN$)z$`VUET zviv1!REzwV9+48WR`_i`j_^zNeH$JA z0i>eW1H5#ePg>h+BJOZP_aUi|T_?^%cY)wYT@C#?3n@SmjfMfJniV`RNi-67AP z#PaX&Xg&78ugc`h+vtOD^vu5{uw*ih-aQv(q7&-khrhq}q`_2PF4tRq7%XE5z7d)~ zwl&}Eg5E5DMi$ePH#zO+Ti>}uWZ9fvt+U(q?PK$+&s`L?4wIeibf|U_uJE&Szh5P* z&$V|?P5GSjvqOj1il*xfSz&*V1k7xVudFA%!p% zdRii!w!F_b_+h5GRotjk{H#~A!K0p*pDzDY0uZ}C$jQ^&HsaW;$|raS4r2C?(?%gm z+k(6`GJ4GlqkV(!beucvF_Rur)SSJmZrcjtQRayjbV57XL-f41&kHg2*x zABX6XgTC4Wjb?5fV51@rpJALgBAU(VL_vOJWeoG6Ex8&=t zY!Xv1FZv-53CRSuhFHsJ(UqRNuY9)dRIGmeea_|4t)ad*DmS-_judndV3dD#v60m~ zf{{HP*sd(?Ko7s^w3NbwDbZXg*RUA-AYsD@~l)+z*8qp^h3qt z&BNiTmN>iA+4u6a+7SAoni>hel?hLuJBYN4V*SSh?Z;lk($`$4+Q?S~V+ zaVkdFUgey>Jb75mTO)+lSUL}Xr7kwfRDb@cwy!L8GBndh|8aVJ{evSx$%@|{Y)P=y zfsVue?vEfX9Cf4K{<{pRu`Mlzvi3~c5Is5Ivs!^QlF+K-m3S9qL__b#xyKlrovCuL z_vqU4gzz>FpRv2~NOt4K#p2()qFSP}qn95{ef#orwH}x9P5Hl`$mwx7ayLQl@Yr^y zO^|QO!wYv0RvQ?)<(=4xGI>sz`kztQ{(ErczaYz_HVDY$$ghpoBSsV69;(-cLzO3| z7mn$I)rVo!B0ImZZL<-BOx(#)~v27T8B8j%9?f!3KQqU$}~dunx!XUxu!bw|3? z`}6b=4F64oUBqG8myEZ>PThG<^zkTHEY8VZ`8<-JdBwscJN&QPn4ai~Odv1U9;e@A zs;5iky%Pi`{jb8#Eh_13kK=zp#Y=e`T#1onG8)>HC>CBo)D*PD3);moPpO$=V=7r* zQ!<;$@ott8DJqD}WR68vqn4V8I-@XXI%Dd1LSc-`yr5msng4k>Yn`|AwjcKEUTg2) z{(Qe5LCY~G?f1)IqNZ?&g<#UcJ^Z&lYdMaC3U(01Zo71uk5Snn0RYX|g?9M8Wl7gH z;T6A9g590Ai%<$_3~}%PrWu3ou<-`%Z4^>;CP4vxaPN3~dvLN+gx)x2o2aOFy-gy=Q05X~R?$|C zM2AfOARm9nUEc})usp7=lH*SI+pQ)Wl|@E|%8TP`^j!{7cIXmdWW$EKQfy;Uil+<` zsF!dyorgpK^8*B-#dmAXJ(vpu2pFr^IaK`Ao3IE5^a-S?(TE4)fa>n{)66Zg)zZK$m!w2DthcZ9EHnQMsw| z>eQ+VvF&}>VE4KpW!Kpos^z9LJC_q&x}%zN>KI4h$(edhvq!2@P}U3{@6|mwRSUi5 z48CS4EPXgVE5jrfJzT1rmX?y#ET<;z9!#D#Xz*ZhE$wxb?Bg`5T2tTl+UpJ;=n@~_ z1@_vCcb3Csa+owMbs>=SZ->ro8LaOXd`^mYY0rvXYQmllBmb4jzhG}~^ic-6F=ALC zLCGtF=S zJC-qM@7O*@2i(BKG~eyzOAv{Bqaq+0$E?ngMD_QLUBrUw66_V*N*jj)`GDSqc`wJe1_5Q!PZnCyMxP=M#=_4)p+qEv9&Z2kgbs%m8%NTD&vkhA8%=J_e? zEDjy$6ARg?s2jSuhmzMvufEQif7*1XkZ91ujd9R&6?o<0f@Ka!F@+pWr*;$__=;?* ztDeBm54-`Hmv5fOIkITT*s6{dF7+h%H0Bscr+NKE%F{%jQUFPMJI<;+V>{n1-YWZ4 z5YW@Jo3Ji8ePLr3B)5uBs%mpz-<>AaYR=6Wm3zA$iHga0Xj|92T`kGVwm^b#qYF$+ z;O*)x4mZcKj-{E)aqBI&o*YfCF8(uPgMq>4X8(V4b9nLpdf)!L#!>)cD#G{dthNPjiiMPJRa0b|XUEjr~EJu$%# zPy0sVme7cIy2UEbax8f=wbLwU&&Aqc??Y9U%wZ=)^~|M99$(CG$QblZ_>$N^)7I62 zHD9X>*o?f%Z=u3}skC??mEVN@s@1t@OOj>z{_DfC->3Mp=5n``nQN{>q1Ejm@FN+E0KX(w<9|oPHi@C zB$(OB`rc_5LL5NaODN}~K{&D8;PSTC=<6Cq0oA?8mu%cKlAAEHJyFTYdr`)BN^gP+ z(TB&lPQ;An39f(*&^ty~`BUUo;({}C{HFc6NZ`a&tXJWd+(_2?LL4k9#a#w<Hn2bF&0Nk$5aQ7X@E&K>8h*E`$RAWD+nh10-;+XmSavZTBQ2agZ^2mA zHa(bFEaM-(>d%+^Tttl*-s8Rh$sAaivXY+DT)n<+;MvQWe?%qj}REN za#zl`NM2sY2Rb>T95q2{*gBq}E>h0FPCDmnKWGd7I)8ob_Mdh$HJYaVy~e<+D?lxI zfvPb4GmvX}m+BGH?#`OLr96ppM6HP_j?VgPZD`*kuPQc2XloS(cfod7)$I}p)4)@F;c$}XF^gqS78gB|5Z3qXp`8!%}-Mak4y&~Bb9}E4b85yMaLZlG7Y7d@eg>6wQoFU@IElr9J-VZtIX`C=yww5RyT@tJ_ zAaEU3;u=3q0CpqCElC-)SFC)6rA|!k6I9HV&#R5~pcgKmIxP*npo}UUF=1Q*gV0)r z_c}ZOi(kKdxihS4xvpbCvk*kw>S{2V2_np#b!6k6=a~5NbS+l6UI@_m)c)Y7*zhf@ znE09?GZ1RT!jTfv3tF!BJ!AB|YaQ0(XEqRMIdc|Vir U7(Zw7<>R-~ILhz^Sg)6&+~($d!4w0=C+3ehsu*Ei7DGc+(TFfuYUUIzrMTf26hm8sbV;3n{v z&6~hBHnxs#TW#&PIoQ}Z`#`t3d%)rFEnEEpeLVx*yx^Yyeu=h`kns5POHVr+ zJJ0{)?Z>c|iJ{)CZoICxy_SxNwyufxkK0;XH0PFnl+kFpVYjswZ=rx)XpPF-wc;*V4nr_EMT=6It2IJ1mUNDcl1ih zE;KSWx3IJVJ3+QOLt$`)w~w#ij*w7fSU4&IO(4dS;^Gs?scGqy{Ztw~=WuS`k)y}* zi;B;bl$MnMGd4bKa`lQnvF9{Q$1ID_OYGkebhLtrqgt>>nZgBSjHP)hc+vAHAsxzopgP5X~!|Njgt z{J*m7e+>J7?V8nEr>m`5JY5s5U0Mss9aYBU)eT8{u1>C=XSfXL=@hmx6lvj1hC*oq z9md|3(|ATl*&!i=Dr5}yl7R#noX`<}2~tXF>AM*Pl2BYti}0?&hQcDUtZI6PQi4F( zN{P&c(xRxneK%)#GESFS4?q<}uo*%oZ*D62E(PC#rNz438BDTjDi84|JBpyDYb#QQNZCGC)A ztan?Jqm%1a9Y&n%+Ax|kPrh+TV+Hs&*#immCIgKV@y3Th08ZOO{vDuplrF-RBQvUm zdf@lrQ`*L6JggWPVlm->x7-wrbZ1m4vDmvkX2z+m7`2r+V;d=57CIvJoo`LXKjhdt za6+Qtw_?WoOTg&?N`QU5f>tW^Rbtn8J9$+2aw;lx4#v|0lo72Qnt<}KSeb*4f)=YC zpr|l}iLgf`$NRl|9#CbvSTNIXzd%{wUR+3+|XtoUHbE%o7KO(ekkgq|7yanZM=`4X2iHm z1rnWHExC=s(@34u7I1D2@R_sQO$Ir}moL*!))E?P3H2mvwS){E<;YAK)=(-6)9DsA!Oq1Kz+F?hcr{V05+KctQWW^;_Y1?}*+O3Nv7 zh$V-~IgSW^C)1xUUsoZIOXVEDjIe^gKRjYjz*-4vVjf)!h0l`UlPd3w(wOx$J!f%7 zU(-6JXl)B^m^k-n_d(#_eHrZyGwxO#=%3JF)d2bQX@U#_9VQM?xxU7^x_s9V%5)hX zOs0t`A}=gopz^FLR$?`)8-Yp|a0plkpLu{71tVbSXDekCQG@E9ziI>O9x53=aZG^F z$AsWGS#X7_i!xe+>yZamgkk-wMmfns&6!iM55q+A0L>mR9cqa6W-DSXJxZ$CiWrJ! zJ^h5>^fev;ANW$Rvt2m~3l%6pw;h#6LU6L!sap=`>{ZPM>!R}U3NWieZU^im_GWlj zpJ#aB?r{<1Ob1Jz-G#^g1#QrROX)OpXGwxne8$j>;AGTT+?4 z63DH>Gy2A;N8f1Nc{`mqIcu{4VM{O070yfjt@ z*7bq;02yHerz+5XTLK{JN$f3!m9uMTH@SVkEM!p|L(}Hx#G1TaU3(qT!gb{&j=g3E zRnonpAPv|^)|KEGtQ}lIx2eGE(1v)B8jZKx4J!*opQH2zi&L1R@Mm<7D1H7VuH)&? zQSvZ=8!d6h;ik}-Ni5`FiUs=jI|DV*!jlMaFY!IMR5H}GV;fb7CsD9y7~ zJ80WBxUol+%u<|gH`oq*i$lM3^Z}-vXTUObv%;V%e5y6WBZ@E6MXh`|{o)^&v%meR z|LZ?Y2II&u>xUNT`iH-Ze)^#mhD0XE&HEiVL9qfpxuJ$e(YzlVdkYVaV`R=YOMTMe zL6+o001toC8>2~0tBq$5%+=}1NY0FvP4s7Jh$xb~;R|r*3 zw(FrGz3UNi8vlR;5`pHnG)20FRddKNQMG<|LnhRSKTJJgEv7#+2a^bV!HT@#2U_U z`fZ8UG$gu9IHDS$mq!vJiRo|F5(j=ZSy1L+f5W=sCNoOSA~?x(AAB~Pw1tzTHlZh} z*N06*^y?9ZKj-KDqZ+Efxzmp#wCMSq#L+2LGx*>ADnumf8qQtFC@1sQ5dEx?8w85Q zD#arFrH%%Sa0M;Lz*&ibf5!Fum)}nI6?hSvRokd`1@I9Kc;tB5AyeX;bAoB0zTzz)x0}H11R8=z9tLj5e<-x6)C(Z z$56Zp-Xwxy^wQp*No=Bdps*$^tPF*;1D0Dc5;ZHQQ=_dx%wKpD^zeTWlSmw_eBluW zeaP1lvQs_6W=}zH<{(STMilQ5&gi@e5(2eyt{`=@_rGB#=ZDtovDf0A23-NPW18&q z>W9`eYdUfDlmp|}(Ue-jJ8_v{{<`(%OKC_Nk9eDaR-)?0KBDykT&<}+KGu1+B;bob*KqUE73%fOJ)wKF@EC5{J5+j<|8YCQVhWRy>|i*@>}(% zXf)@wXYBaxK;su45sXvLlF;6rfl%KS;XE@Y!bHHrv*d7w5w%RjdrSZcOcY2`IVQTlCwXh3Xsco7ngaH9_Dxoe!Bqkq1HCG!G)#v;}!wlNv0##M}D zP;&*&2xNlrH!d#)t?TuqL2u(`c_OBUAakt?Z zRSYdbg!wEDv=b_5+*Yiy)B~t+wqvDmNPJBo%7ZM#3s?j&&5ApEu!@Cak&KmSVsFyK zVzt1dX0{=wQk5tdgwgswK?N-K`C=|3n7XyW8Q91z$XUEDJCbTMCd{oqL~#%t)p*LU9E7borvTls*a1E5P4fo_o;qY z$0GvCQZMhbLE`6pp8BL}5}y%!A;%q`py&abb%x+mlvp>K z7pDmf2^(o35-YI0X&TD}4}Ax*AM6i8Av1;e}fioWdHd*Nunu5@Q~y(#;lNbT$=8cAu<*9$4= zgTZ%Hags9#nA(7nr;mdsX(cr$?dX0+Uqr#sVnR93HQdfrRPryoJs1Z)+h_%BqGNCj zzlcSvfhv9^Bc`v@=C(D>IJSaY$c{fD!w(8rBs9TT(cKlq0HBaQHn`+OrV@)r!lnME zkF0=CSM6+wX*ofFG8&H{cUL<`rVz0Hz5-Kd*?>HS&UDZnH*=?ijiJM4S@H(i!gYD`z+i&W}Slp2cfsGSzy&v)OHZlsCDNRWK zB8tbq1hbM#5mpEnM!6&ry*Q|#d(tN9UYydlh$1kPlSeCQ>%IhzhR=NtT8==!Kmjy$ zJ64{$k#`Szlm{PTyaZ~%)pj4A>2Tfu%vqS~FE!3?(-O`Jph;6T5uD7o3bQxp#R|JW zd(fL(w^28c!@>h<8#4k?QsS-8GyJ%YMW>FsqT`AW^O6tz&w)!yyE-35shsrNee%)# zj>ezxC+8IJWxF?@zfoA@^jiey+>z_nBG`ha*hN{xi>L5af(A|9x>)6BNafbRjqV0b zJQRoMzx|yP@k+2<78inM|SsHO{A|u zVa@F(^QanvFeFSq9$Bf0Cs9A;V*#i`nuG`x2xOn_2Kl4Z`DTx(kwr^<170YzBa_yr zbU3+7_ds!$Z{qzJZ=uD+$zoy#Ktndl{$&8ESy8u#+WJP}JXHp@r>3D;Y1G;e=j3_j z>dAFEnY1`5z|1PCnyYFM2+_w7HU7{uaD{@xQ+ze$&_*>?i8cUU5EL-Y(LEgY0)bQN zOM4`ra5h$CR{^8&Mr$N4D94cC1P!=tsvT zP)swp3S4X!O+Z$>@KWsD@XOev+#gzdxBt*;ci=Yte)xx$EaAxiyl`zY?xX&bM`bXi z4H5lF{D8IhQ!rb$7Vpf%;bAj%UKv^%Ipkg6ReEQH3CP-26A#n*`(V8)Oh2uzJ)Ji- z5!IihdZvE07iJK`kaO(I4d5g6Tm0Q&fOLn@5|bJY1(4Z+YJ3<&Mu|$4EP9{;8F8Zc zFk4W~uW-@RXTTyuP_j@mpI&~2Fvu~>4cyzhRTJbFG7?RVPyNMcO{`pb+01amH`2;Q zU+3GsGDUaQ5X{0^`d_NHj)2Gy`0^-NSNgqbxA$KK9~L`~^<7;KolN(yJEAO7$E>`} z?;9IBga`mjE=F%4HoT9QyDgm#bX!VYgaFXkU~5j^k})QaDp+Jr`B(DVHMrCPqmA0( z;~8F2`i&f~APdO49DKVv@iV@IkXn5HE<l#xhG7C#c8 zS+>F=O(rOk*$Zff_pQapn~8#!wlBz}T( z3}FneYH>00nkw!C_WNi`R*K-IQ1n8||1l3~8 zY&5lLvl7O-<%p=TZ4|33A6gHJ{zLySZnyX=N!PU*yblx2yrv*_UQ=2f>rCiO@#Y|R*$}X4a&B4UQPD?6_bq#F zx)?gFF%j>9g`0Jo$-?H8mt^jokhCb6r`n4Mkm=)t%>{Mb2CP2jCm-Nw6XyW9OzmJb z#*V~yj*2PGHVj&WU*fc>^43f~I)`9%0BL8Tt5h$wR=0G~Ypk;N@_&=z6JRl`v@>_L z-#lL$q#w$pcUf%H*KaYD+#C|?W(Cn|^xJba;sO=+;wwIrC-*_1;LF5*W zPZq9@FJI%6ymU;=Rv+3ejveBSn5r%W;5d|{?Fa+1;|7MV3Muw!Rv?(=U#D-hJ_-9? zLT*ewX_sG`3%{;`pW~4+$y-(BBG--9h)!#(u&xLiRFr&etxVac@ml?*L;3r7QD6-% zDG}-~7P8N34+%@;aR9ta41D~iP#32!ELmi6{zv3y5OB7@map&j847Z&XMo6w+{}}IHF}(w(I8qxWOYv}Blz~5W4uvB z1-u1HZpgkG}8zf=QWvHbo{@l^wOZR z>+DzbYgjJ%%#K*zyS)nuuXm)J5$YjeB82^&J)>cGZm`A7g#m$yvN-pM=i#zojyJd| z-Hse*(QE61^HlftTe~5(-%hyouXRm4EcxUbViVgK&nTGb{5N(oSxTK6sYWp#yiYHf z>888^eMHd~8xuTd3T+6MSy4O$04zmW60sypqT3C285@O#E6Zf=Q#O1+2ch9@;qA7aFd8sWta zIcfp?6CZ=sS0)kWBYus0|9F zJUC9LpI5wDjYs(gkm78_aGZS;{t9ap<{*QFP0}>QsYXl>2^mU~#if$w(qiKhb1%A^fD)^}?jR6PIcpdafu^?Vf7K`%F)k&v z?31)U!eF`Xi|c{uvYTm30wMKNWm<2M# zlrV#b4bX!4lVrE@b5e9YsAY4DBmCQ0%(U5`s=}Z0Cryc5x*M5NR>w~`;6eE+RwQ!V zXhw9Zk@I#LDei^Tx1X|*7dFYh4(|^o*W`wKkQ|?cEuDi?D)+sFgvimbD?W&1+5l~h zrnVN)+NJ1zA4GAuHeMS`t}=GZXmuw4E z^dsju74@@yzAq04lN-n_x2A|`DYa#I%WpDLRcuKBal7KI)xGo-7=(O&P&9kStS;Nd zQ-!&vt^-+-`^~CzT-dFy7!n5=2fg=X>`gfEEk32OVf6%YVs&q`Q7fj`5^S{C3aDJU z`WkdD=8w%eFCsNi|1Y%|zIDNR?+>kqe*Z1*n;8GtJqmdSb;yZxpx+N+Tr$hq+njcC z3sh`&xC5m=53Hp*&R=Gp8QV*m=aVD{w(1A~WJpMP>dqBY*rqw$jh4cfj*zPP#v2tY z@6AolgpT9Q1vNM*$a16Sd>j4zef^G}&L#vi=9Wgj#Rh6v)lrtal_*IYlq~Ud@tpM% zlp-SLZ8;BFIYj+~J24&@8!rt>zd? z<#{;sWtIY6sxYMnsCNsFL%Qom1aWI`yHVvnaLU=!?^LZyFEqeH`?m_`e>vq5Q0ZcI z6DLWPxXs@YUmj2J!fpkY=9T}oxXSa|LN3bUxgsYPGuUI%`9C+}%v)Qae%f~VSoP*- z{?k|5dZ!_@AW;#a2oLo{dtm_RXn4iOzZ6BdR}Ae|ZldD7Mn|I#PP^x`Aezu!0B@Z? zKiYW8@7pqIW_LJwbi)UNS%QgZ{G zxyeumDkovw_CC2Ec;2s#`$na@qKB$AwNbrExOAnZT}2!w((qD8xu*CA(Il`K7z2gL z34tJY0k6`heCIl+_xB?m{XmDM?ow(cZTRj{^cRs83e+{uk`t@#?t{3#!l=R`A3E?eB1Qlm$%@5YkipW9~ye$Ap4&_$+5wU^P3KzGDDDs z6%uYgrXne8zbbyZLNXD!zx^Y@Vdi61Hph9@bNoZX-Tx+PN$6nxwFw_}wAa)6$*&Up zR(Ai5`rKI48 z!EL8g&6xsnnUoe`&`&y=HCsufOh>SZ8!Y9^!f zB_8zQoPv6whcybZ9WQot8VPIw!2hihPvaM>tE)^YOr=*CZO#DyYPKnoaEjP7!R_r4zS%~)&2-Ij)bo4O`|rbzD&p2C3%K)Nn0>uA zs~uGl_sNs^ppKYd`P}^+VFDRi-xS4gz?RC@D&Sl0lBjh5GCN= zq0{P2`8sVOgza$i(gL|Z(0Q?$#w{BYfHj&YXJ%_a*=$=ffraayA1J>YBZleDN+LE< zuY8PMz>yt~tujZW@`-M>CzYH6MIY)jte1tx*6zQ{dlYrY8pak)s0TFFO0BDDoC{VK zMQSP_?6dz@kQWQw`N>>+Xa$}n^EhKSxAJSjmQMjsHL(%#Z)_yues}yumZc%P`yTw; zb;0eWxcpIia7*#1rty#>Hlib&^sj*K3%-<1hrhv>dnAanlPRXLA$48UxAFs+man|)W0R>%oa7504dN% z5|eYP+k0vkOlg3j73t5m#y?kkktN~lgyPySDKtO_X)f=@pcvcB`8AEQnDguYWDGSv zZuggtB1YGjVd9k#g2#z=28}s$Y0sZT*PjyxA1Zb)*7~jb%|d;VXkv1v z;FBJikah?x}TztP7NpJDLg+S%c;C^9J_;mRFvRMZSoXr zWF?KL)|b29mcb)t_;zfoH9^;vN$Z1Lco{A~w~N0%^wC?`<)CRswlXNBNS znhS@Ei~4+Wd0RB&r#`g{d)Xmc{#DR7M>;Zl@=eK$r_#iU-0JO3mHDck#0}&-UZ0~P z3k|urA@_^otJmC@v2uLLf{7}YW$VCrRabff!YdDQxE{F2zD>UcTq`GNn!>puR|=mU zk%SIa*Uh(4_Tj4&P1NErD znaI(!e(Fey<1qwuH~0J&ble;A?Pnd??W@AkM&2kk;Y?u1;X^2$<1r+mO9750r!vdF z7?;0T6Wfdjz2NBYzmn3HHs3W-nVu?LbFx(_e;O5Ka%p3Zh$!}x0&GEjT5eghIiR2uT80T0U< zw0VJjyGzW}G-wm;HXd&|>kN;9J2|-m*N?>L)1F7X!*}Moa>AO@>;Rh+I{c|J7mg%p zf)9Wf!lr>Um`ZkHBVHNY5B$^+2ptqKIo)iwrh~k@WI9R))S7kXR_R%B83tKMS8qm2 z#!29`CS`>)4hf78wHg#fw$Qls*JI2aQZ5Ghc23cjA7+` z_3Kv#gPo6-6&;gjUmv26_=18Q*StdP(7g88$i*VG)|Rcd{bnUBZ3FvRtPxyTRBv4Q z;5ksvZbA-JH|*h*o}z`2ruT1KKw&*9on1fRE|GfngNPopjX?nlQoXO!#@1<$C#);& znRnG^P20zfGVlJKfF^2M;fa2R#>9;Nbcs38%qit6gbqo}>u>~@*VQr~z5ZoQSOD$l zFH4*M(*b}*7Zz>yyteCqYY+>mAj$c`Urv9LXmc4$L(1fw=(xpp85T^bVI4Y~)^j}} z+*ee)8~8D1Aurg~TsJMZ{TI+i_z;groACWr3Lg*h6V3MFJ;eHMpIiEFB(I7#0Gs6G zBYuy24(^AA0TL?zs1)>m#cA3Ei&awRrjL7zRI2%RCl53yFk1R|MA7}$H_R_&{@ua( z`9e*C*aP8t9Z$0ETu59Woqu%aYD^&NB4gV4>$alEE3F%%HtbH+%%Z`cj8t{ETI<6{ zXtx;@s=d7InBc1roCy`AI`_r~WQAlyMc9B9>9c#_o~g3!h-}b&$9e#{>_KSs&PaUl zH6jL<5(RURCr!4z*uBW^Lh4+YZ2+b3h4t*Ezr79<+o(`8YmoG7XXShB+o?CZj?wc7 zFd#qvhnDe$eZb?|iSIT|l@#yscv{s(KiJnqvv#hX-9;~dZqT3f7}q;xDrl^Cy6jjG zn^s(@)<}jC8v{jt^yEoAFHZoHLS2`dSs?gjzjOoB4r2-|yHC;A8Ds=(5F1YgiLFe4 zZO}iD2_>pa@h*j3dI82U0L#wV{b{jPxJX7KUe)&Rs0^Nv!sX_;l1A zMDXB{P#4?I_n2$(BC*CKSuON21YncODMf(^&+i9}1%U4ZU;2uA9vD|xN7%D&inU)! z(~vu77Syj70r%+J-sgo$(xsYV0^3KxXY?9)UJpd0?VW`;^*kLYm;0N25O^(ex6AD^ zIBJ_3RSmpR@V*^!#BWZI&#h8JD9>CsQCrv3-={m5ILrKjsqo20YvW=u;y}7x<#}d$ zb|V4%X}lvHnz}nai^xrPUH|pj+@65A!w-LG)mv=&X1W9RY9#jY{P7FlKmGDI!1>~q z&A*<=Oa7(gXulm#t}8u|b5Xo4vPMph)#;FtfFJRX9-*C2K{cZq2#XyVJI-VEnbTPh z^JoBFJ`jSbE5<-;U5qL22jZJ+ZQ(h)*L)G1yN*g7ZP!zCy5CG?!#cO!;bR8g`m0+F z=!ZK07C*4D54AI9`<^ulB<$b+E`dV$B>UnrA{WLdMxEX0A=HhIrxP#B2ufq9DP@6N?>FAxHx!`AEUROzCAhn%1Dla z=iYo+T(Bs>^A2~M$pH_BPIEqXiyeGBq#HA9b$1gNuhM3Q7E;{tl_D>hyYe*uQBvZA zfJgZ1xqZO*s;j8BOANP=v!SNoF8UEsPYn)oFv`dJGW-l|L_o)F>^kP9XfI9Q;K&Iv zdBaeexI)jIbkV;b49B!v0W%dNe8N#Izy<_; zc;7`c*U|g2S!s;#DK{juNcHJ~F{-|U*fR4P5JVY7mbVXV58j4Xjwqg~9;O}D%=+LM z+dKmZRyhO~_A#=&dMKW6NA4yy^)?n2#EW^d1Sl4YwVLp@y`ym?G$C*PJoAycvkXP5 z0mkd6R55CRCdf0{0Wn(tj)Np8W;zQ~n1t>wEj$AhA^ev1MbK(_0P4YA6g-3*l)M|Lw=rjDRGdtZ}{t^ z^F@m-j>}1#fNyq0*0vaHq-l)YIrBis;Z;|eLIB$9Juft1jGtfmvWY{c(1sP>A!l6S z;6&x1Om4-~)bg@GsAPzar~S@sR_DaaLmF#F2SLs@`1S0ePMg(0!{9MzV6FW+P7}K= zFA&Hgz~+l%^iX=5c}IaQxyp@-lSh@o*hP<$=1?l*^G5~ip(Fe;1^g`)F< zp<;q$UtDZJa5w_v!ZIlXXO9{H7oQO~xa|PH6P95~X7Us0{+p@cF zO4WOs%|?KajNpZyfFzS5bn38RDG~boSVEGS&IQJ)S?`%6Lys*VQh}=kv?)JUGsYc6 zDk%fkB2ng-_G3&?BhY|W*mE?K?|?M@wO;1Iwd`Oqkl+K zD$zZtyBarCTS@;2ASuIVI>ya|=A{mDDRBD!xB*D}^vvz<)%lLT*hFDHfK=N(0y_vd9Df%_ znyVqi2m(PxJSxlHp2qQ|qHX4{xtC0niTaS6{}v9)=*%&O!f{W-J|Z8xr@QFzEv9Px5g1q4 ztbiZ>+as^UzF(mx9)j(mj)dvO>w<3X=z2Ai{QFdl_cgPPSW3#SS5JOjapDSpKl($f zyzkf5VNGKw5_xQ}^jOQzR1o+}(zynW9MRZ{INTgj$IV107%*DT6_H*XBDJ*-R7AeU z0CqG0+~`&_nqJ**8g2bZQ6xsL*A1)8EwYAc~ld2 zOk6f5nos2PIy{>xuYS=M^AYJc3T`4dTo$cbiFYq@35@fG2v_h;26V}w?H2NL( z5;7$4T;UFNSk!^gbCAC4%b*j#(!FB~*-d`JRNz=J{#&m1bmV|QChq;U_TBDMi12ZA z=;9gEC*k`T4;VdP4r#sMv^ZIU+;>N#XFb9^C1dUVYYvOvyY5b$@8_N?ka;fJbx+MR zuXp$zFUy+kFzn4I@oyia7uYZMuyK#kp;?d5q3b8^(EC_f=+rNriP0%dQ-V}rOHiQX z)2$jsr|dwBY6!bo5YxYtJv!jfE&XswTSYV-@lbTIC#bI|mwf|b9n1lT3Bm41JBPTp zQbD-zrm_mg&b|4>o;~jaVQHn1i;qVP=ubTYzC{Q7x&m`Sr(AM3NkW;ad@{o$$%BD< z#eVqk8SqO{Wo@?a9X=}ciy&p;QH7IeD?g^?0d1_Sg8Pae%h7&Am}ty8&(t=82rJj; z$KKMn=Es+;4+jN^F8f!>GW;2hoewgz$RMlrs>a%#4glX9>OOi0J&N^`;@PDq0=$ad>VajIp?X z*Ot;6tHQhcg0~h@5$u_tofa)d;xV{lar)=fBxDQ zcAQTZ)pc=A?3L!RnG%b-BioT*krBAvM{E(kps}wzE4^Qa{m?4jwcxt_??b!3n@&eQ zH~FDuM*C&C;D^>!9pm}Mj~4c;a+Xf7cwxh8H)GzHOSFCO0{p` z5{5AuVHilyy#QLj?cJGN)fL0`ovZQDD2n(J;>*YP&GeRE;-Y?c5tnu4^U7JOC#o+5Q z`mZv^IkMLdU@H6peP1SZH|h>tSrE3A><6FnH|GY#GCvaMGr+y^2PJzL3j{>0E2o+w z`Lm<9dOv`22L2p#k2ZgG--@qe|F&S*(z)tOFEYRbz08z>^5W@rcS}UsFj{3kxD$Ig z^m*MvqtIpF$C&s0C)hUV!#+(%S+cxLJThP4t(<9Vr;>Ww2lE8}v4@+%q$+UJCYXe3 z#QT7C<69nOA|-#M2p7p0cty>qFI z^pU^%15z1jFYHUwF1x#J%zP8WeJ*)7@|T*po}O0u-E*5g^DzJMeYE`oQF3j@b@RKe zM~iW3D^f$Qc0SQ$J$?T7{vFo#7HFY!$9u0>N2}IG`{LGAe)^nHt{8CI;+meAt;Pk+ zk;)34zLkRb#_N40SsR)V?S}VU1co0M$7Azj%LtqbWVD6;z;Vmox3o{HtGvAm@+R8BfsbPT(Aw@wN-5$T zW$x!EFn0KL;zQO15MG@2(k87r5!)1hvnng$9-q~3xtC`4=*_ZtuWcSZQ8giX|bK_h}#*@v2)9KT4TKjGtOW1IIV`U zikk#|tu7dQ__h-LovNG@Y-WSCB ze+Do7ZA87e^Ge@+1CJvYKkXzzajYT|O7!SJn^)~^Mh$!(YKRDn@-}6A@MJ*ack4A> z-y;{u=}zYBDZ6qZFR&rJ`4QXnFPr{V1`|Vr&h&QJ)3bKT6>S4c5Uxsn}X6KyEP>E ziIX3B&jp@X^t5@a)~e`mWAE*otKPbYg)=AOBAIEb`R7>5{PeW(^vTFQ;!e7*uRKi2 zNR-ws-5E-E`t%r)>!xdboAx?VE}bcjC{MBP8yeUHpHk!f`N=aEUTq+k>K_F6oQ%{H z^Lt{fN~N0O_PX&pC%lU|7XnJAtRcB1X-_HZz6LHMv5f(tq>|!`oYDXl%{@P9w zb^T{68|<9oNdeY#1DTt7N4op#=_67glMk=(W^2#19$R+3c9#Y#C?ew@h}pcahlK4_PXj6dLIFVdjsdedEhREeV(2 zM9p7vlv@gPs81M2)_VL64U&c4$9OX0HZ|t^+yv!sT_AbNHCIY?x#z0|q1?BJO$LW~ zwIwb){BJHtRljy$jy;^leIJw^MrYacIs4S{*CT1?-LK!px!;2=q&gs~rG>D!7M9OA z_fV_utMA(3T+qICj1FLHWq-O&sx)Q#%}CZhl4P9G_coAz=C-8u*;IK4&XG8D0@_*c zM{k)=xjipQTkK;5oIpnHPoe&SIAyRvq7T4s_9M<)59oW50!H-AkubBjn9&Z*Fd^;2rlk&f4b~2JFw^%kA^=M)${~y% z<35|G2WPtk&#DE8@Mz}bv$&u3(@lV!mr??I#r>$;k-YtP_}>{-ZyqifKQ;_>PP2TI zn0c)!u(Vz|P#W-Mai`j;cOFM3j2Lt7-LG+_3nwe1CCB~d9+lr9gIfz@rkUVl^j8N6 zb8VfWr)o(*U7_b?_LC^aUcqxqm8H9+`vfi*P&e%daud|ejJsh;!!xfxcMML1ev~Qt zr0Zv_d?hB`pdC@#9hiAB7M^QYv{_D-1){{hRy_@mJzv>3 z;jMDDz1jIZ&+-rr{EWqXV9Zncw}*YNz7IiCFvHD+SIz5JYqRnt!)8evi*`H*0jYYH zJP#(3^mQ%lG$7HWPN%uH!tX)$53N$O%yzpUT7vLz4A@D1&-$+DzdrAp#O=*y82VOw+K}1yXCMitHL#M-|)y_Gq_n2r-@@vihSyx*P<#aLH-D@$%-jqOX$qCEIh?6K{NiE+4veJ(63IT6sXM zT03=Ifp~a_yeLzyU->ZLlM~*HjC9YKt^Pe<>2Td@IbNRCvU1<|yZX47MMdpBf%V%B z*GYQM0+-p5Tmy2?rbnyokIn->KDo!wEZL!GYFkeHzW2C_`0OFdcdva*dqMTQnwt>`<(E6MugD^oP{jxgAC8A2wGRXx$ju% zOvfe7wRCYgtoLNiM2$21yE9BjL=B$`N2Y&k#t=*{E`v4+8=*DlbAoa&bYbj z%NBi(&fl}uSUBj<{;rgDbmzpYp;d6+!xtA1e?8|E)LF8wGpOsId393Mfoj#e z6JjHstIlzY(=Fb+bXql6y00a*<5ko9ChUEX=2fbYH!qQj239y)Et8CN0Jga*v4 z;XE!5m~mPZeKPg>Uo^dURMP+d{;!pmrj;h;&Ri*|IlyhDxp2!;apYETq>{U3j*M_` zG$){-xOZAs;?@*V)6B|&scE_D)qH(jr?cH=(1VgeT~?px&j-vB=R!)wTk@|PDQNH90*?kY$10|Y0;UoZLz3?9#({0 zj8Aa-(sw80soLV68JG8o&U+qDBeC`V-f`1};4#{YtAYDz_&X(3=;6Eh=Am+SPY%@o zEbsB&INsuz7w1crYOS`eXZ*6yn}5=tU+&g^wewzKCGC@wbmZI1f3z+gia0$=wY3|Y zh2;57WaRF0X4EP`*E?@bpr#(x8GeFhY8fKpXk5tTLU{K5}2V z4`%}C;|~+@F6BXA_>%9d|KbDM%CQ;(iM zwcRU>1WZ3MMBJ=yD6eob?di4vBhpz5^JVsZx^?sP5$U`-tYVnc%ZV}*emmvxL^GFR z+8L&QKHl?9*8y1oH(P9Wj0(2P=6E8M)@N&YUr@9DeA35^J!M}&huuTdB$i0;%*J;{ zx&LfL&v2yv@0BZ;0>;g^9SgqKkaE&nr!{KJwf3RUsyIjk_X%wq@lrwK?otW7nx-da zzUFBBbfRO#qXdq8ZCssb)5>{u)bXetZ_=yWMheHN^$8nE>mqqt3UUYnM@WwLh0pzU z#XSN~+MVR&rYvuQarSSw&EI2d3rC8i10|{nQnu5qmmx6q;bIXbGg)|2a3+wFG#TY6JDqy_Ia zDB#mMP-@?iuF)Vz$?^pJ#`e{jL^tQ$sRL&qR=p|gvIxn6l1p5ER_$_@Fn5)1V%Yqq zKDo(Re-Yo7f1&#!@+kK#91|@!q zdeUR9i_M>wBCyts6J=*wm--voJqpO?%Cf>>1eKgX7;hhPtFSaK07%GQdoJ zok|r*oULeSLQLk7D96|7*O&s+*Y{2Qwerbp*(H87JZpQ#i8G@ZSYfZ~9H*PUc09)} zQnT4;b-4~D%*A@DKOT>b%8T;s7Bdt-paVbpKC+9O?U>x#g)VO2R>##q|K)PC-xqzOEv#Kp@BBXRpT$!)EG;?R$>O90>>7RPc zaA`8oNi;KIq11=^(mO|T&LH?R;xoRkaNAK6-??1d`JzJGG!Nf!MrJm0yxh|cn1OF0l9E&<;0QN8D#HR_NhMY*XG^TG7Jj|mD=U&bg7joiQQ*y=cBg%IX2;ZyV z-EqwlU@&IPUdkDcT!EviOt3856?r2p0v7L`cTpwih8>gC6~;AGl|P6C-e%tpYqQScWfA97%Mco)_mD$R4v ziO9;oh^Ki=P<2mfveHGZ#TS~}f6U=)v5Cx%;4?Sv>3$I3xENdCNO6S+$;U$?9ni^+ z42P-ZBWJ>0ckgp8_uvo31L&~Zw4qejdSLH(cC$3dT4c_RS|2kLt+%ElcYn6^gUAP` ziz=d-5A>ZVL1!{lTRiT?R9p&(k}Hn z#p-Q2EcpLjLM=8y-WvF}nKqXk(ijtgypO-B=R6}%sq)&#(N4yR2N|43tBp;o7$y>_ za_O-(S=nLYA$I?qd@VA#2T%_ zkd>t~ZmS;ARoVJ+wvw>QU+7I2TQsk+=VaxbE{|(7A`lUT6GkBFn*~- z)Jyrs@I+gy0i-7-quh|MhR&t}d56!f{>!y8Y8L*Uw?X>exp(H{nJSY1a;;i^YPcjh zU0X5#;GqIY+8S~eQuBCc{z0Q_((WXpv%@PLuh zs@JxV109?W9L`v%^6evi7{-~w^hneqL3N69M00GGw6d8l-UilU+VrwSxg4Gh0}SHx z^p`6EvVT6$8~_c7QMG$)dH@79^USJ|Cma!1`d3WNach`*|G&IxP__lx?qxZfKA=n6 zJcVU>E_bmrOtPgo1_Tb{!^el*tOz)XZ|praH+eq&;_d-YB`DG zEx;d$lndA%*{Tt;Ki;9M#2Y*<#?fNh)L)*)0*0O#32ajy07v`HOks3T33D=)U&H?7 zH8}A$sXDxnd~~7O*`=F56aAblN#6Sbct8w;qi&~oRF=rn9JTVMoE0pa_EG09W1|aD z)SL;Y`9wQ@Y#(8|9}yvFlcY=)*6Iw0$wy*En_uCX;GKcZH_ob~rFm4I1etL7yz8yy zj*#sXE=J_5Q>i}(D9C3e91*wvi!n1x_E-ZX$1hm$<)5IQY1Iei^|!4YEZP!lcNX&{ zY?QdMgY$^=yf|H69_zJur^3z6AVce}M@HFsav>4jv)No~=X!aThYV5L)uG~YS+0)M zFCsMDz+eHC&q)q&@-lXwRk63`&Eu>4O8nc#)BYd7-9j!V@M0ZG;1&QGIY7m_(a7Ea zkbu9A$|20S$PH6{&>A7>8m>_S$ME)a&*~u@8a4faIvxPI6(@?NsT>N5U-NSF?#S=7 zJq_am=GVVmSo3wjc_=Ct_r7)+xA*=oLWlwP&DXGtV466$EZe7I!>(`%*US^er8DaV z4Zo({J)Yh^AtSzAUxv#eeh&)^x7=-lT$XhWN~5Re-5&0WsgEraEFqQ}?3HMYM~>(( zZogR3yXD^$+C`fl5sI5}dhQkgDT=RopiXzunl4pScIVmN6$lG=fh~N@6o%eyu3bN)!B*T*FDULBeH5ALlU(Izqn@+Z1C64IU_?dvrs+MbQI^lMf6Ixm> zcw(HTeVp%axEv&9Y~j0E_MwEhM{7!hx%Je83fz4?`XA7%J@=1K?11f$0=_+M6J74L zx0z-tIUN-_uqb}que))!;Ks>;vpc9O)Q=y9n;hbtY7CEx){Xu$yCsv(V&zPJtWG7y zl`6!BPFf<^)U0$+_7vEb+S9uy& z=7=d`koZRMi=pZS>u?Xo5M{ZJFNUT;m0YgB9=i9zbGFfD0%jfFJc2FO8xPVhR%BX2 ztX$pGCNLu^d9cgg-Cblg+sg$VV4qSfMCIT$M@rb!Te|oE1rp%TPH-9e{f9dF{y)@- zhsWjr85q}hRIdBt%O|~>@S7fztKKc_HtrVup_g>X#8|Kh#9Y1o-2&NF>Rqk3fu{z; z#rc{Phz3Sx2uHq~h;3yb(3mdGk$@S|-efa{^3Y1^lqv1=>X`J*Sra2r9D5q03ox-f z6$8j2-Duac;k@no3)L|Hc5>>Q&!YH4OKe5D=QoPzomzMLADUv~LQ5X#98Jt$Zo{Hr z+ldJDO0UWvC;)!0R$^E;7leS$6?xAz0zYjKCzA1P7_$d$<1q2kXTsfWg|)NQ*Bz6o zXK-#z%4=Ajsa;Cl0Ja%1MO=TP9VddV97JQtBI1KtPwq|$bpXza7ZRqUW(Ig+MW?ik zB@u6x2^y|1&+Ba>9i5aG|GaYEGjZug^$v(N*;pPY$92Z(U$|W8-C!W6r?S65ItQtd zJKARCaKY*`!f{o9hppvYg1F}o45|>~r9gv?fPPTZOwzXS33%T&^$Npi7xNvMp*jL2JJL+G;j%lTQ|ay(ES57wx6RsGiJXI135B19`k; zosj_80sAL2z;#X?)Au=?ie&r>8IyYd=#dh+Z#7NZCWMft421_-%$ngqhlBdufz#Lq zqfWmM$QSN!)))?M9p)!eN>%q4>P4eUbHA3{sxQ6%DwJ`Ni@e9FA z`r84WQ_5C2W_6O_OB%>g2Zx?g=LPu-CG$3`x|Y)e->x0$uJjuBzcKyw(y^a&_sH9U zX0a^ckX-Q{a8S9ABWLNacrO_Bh3Vf?4sJKyrthV5UnF;sWBNi{{?slFo=|QEPoTIe^hbd{4>XI+pc*{=s zGf)6ptZ%E@Wq8D(L*S6i zoKyAITf}kf{$p1{iGN!~r%Ys>yY6co6_p!hq^TMwd*u+uf*kUjbk>W!yXp@S=&*?+ zPUt#cisz>&+KLw}B-%m8{S1pPLAVQ|*rj4jU`qcZq348wmsOZ+c}7CkV{-@3~qfKm9|fq zW!<>tq$f}wgJkO6?I@A(hY2$ih*Tfw}V8b~!jl<>7$MdzU*Pcz@mI6_$eT`xyF zwRPc^Dap%^-2tNEZnE{xyROeJv2OpEN2X{wt3Qh)GHu|%2 zIsa3=UH>N|H;0vKDtIx~yN;)ve$j2b^qu7fLVC=yr5IVcJj7ie<;_q*-s(Nx z3;jmkAim!Gy!mrA*oehVB;%C0rrBL1CE{awqdDU@?b6Z(P@22Vl2@X~U66gW$;jGT zX0MfF^A(Stw|Q?C9gcGn-AaG}4R8=F_ZY%53>X=Vw{@F6+hh{+)fVYT7aPDK#e&34 z52)5AR5Zq_(ad}%L%Ax;Ww!Pzy6v)6&A^jvJ)otjTVwr$7Z$moGWb6K()2117g?@~ z$4}MzP2~fOmP_e^mrXdsAX3wb+g7*NCdm3-wAR`MBN@uDqDQ7tWOcXeRi1+L8(nF6Riu^uO8WulecPCw}J|N;v;VLS*wS2kJ^iAy?c@(kQVr-WFUOWi|$pmF%eHlKA8TS_I3*wQILk`@>5uBch@lvW>y_3b^29VK{}apB z?EV*r(KzO*HVaOgF9$i%U;fZQ7^;|g6+CSYR2ywpQR#s_f*N7D0>@8X%O|%LQt5OZ zxBSVB!oQK@DV(btX;+F_cLGT3B=ah*u3i1AYY;fZ6D~# zB_#Pp#6NyWG#SI{vOx(BXNe|3(ovC0@^x5?$U$&JgocavbxY=1`dd)B)v{V%Nq|vXs1TmKdc7bNZSV zEcrFo*9U?*kAEa#t^oC3ZXvmyMSnlW1IhhIe%5BYT8e+oSPCd6cPs!9oPbaMDL$T5 zkw4dAJs#Jz?nD{AvteeU0ADJR zmu_pD8uSi}B@27kirtsO9`8O%VBR#Qk2c5{ZSu0If68<0_dD6mq3doy7va*wv(l|8 zj`#=An3}+9)GAI#3R@+fZ7~dSDPC&vhedNvfu0B8)*9K(k2(Gj17;UluhZa3uoL$j z-LMyT=Dmm!_6@6Z%VBC0#;C=Hh?%5g(`Mdq-4D{YwUDDb=C73;Rs;sz5GK_r;E3Moko>Gt<9#As5#9AQv`e*5|MapLtEUur7hyKB{~z8+>D_Z*u<6<58b1E2jQ^u1PSU5|3ISU;y z29Z40;BU?5wvrd4g(wQj6cF8BVh6GzRucG$^zbbd$|10;PTmfkBU`DOW9fR7ejdy= zCA|>+&IGFY>-SNuG+I>T7s>!TUl| z%3@-1c-x{*u!1hX@wj0)!vAHCZh4lCyo3D19g$>eZ{U&4&#Hz3)7RK!V?>rpL<2zf zYoI7YP7+Ch>T3(youEnk3tF{=&(c;~;|BXyI4m1F2bo#+KjPu_uL|E|mJaT*{xA6J zREEFTqhIs;xN`OY8Bw$1M(h=zdk(;j=QPd`gp6&FnXMGVnKvUpbBpELOno5{&P@pa zA;n6ofh=FUbqJEgE|WWpESyy>mW?b-KU8|^wM0G23Gj~9o=@I7bqHL_Ne@zKlZjBq zf!$nM)pTG#mJT!4v#dhxfRaUXe4j9ezwdlRtd%WMdtbf^MLfwJxbjVaL_Epb8c0KSfTfo@{;U|{<$L=@}Dr}SQC#JUBVO3w|M1nHs70tNzZQflooVJI#glaL+ z*BP)hT4X2)F{}#0!Fb0JoaFNdq#IKyZ-{}+>9v~BK>B6!+nEGD`iVHrEPXw0x(+G! zb@qYB!=|BiBbI_9Fc>$+3=CrvJ(Wq@P@>j#&-dJ{fF*lyHWZTtz&4-943dGG;WxBWLu?m+VzWZ|OkI;CP6HS*B{B&u8}1=9Oj0n;BV_rH z`|Wmz#iqkNe#Z8GHa|0ebx&w^=eEc^ye}&mPR9NcKo4n4@h=&P0V`9!G>L$n`ocnJIby2l`oSu_|X) zvg_m{p(lN&pK-7MJzKoUaSm(S|B(OyF7+k7U3onHFPFpI|MK;2x*3E|L76vva$qYZySU7HF&s4RW$B z`rW&O(~bI7JMgg+pboCmF(^nJ|P*QszzdLnAa=Sa(_S1PH9@X_B{Gu*Moch?5y z_5npMEn^F02If8{TRbUxDq6Bkx{=@UFINlVk{ogLY1`hg48p|!Ff3i^KEy%}F1E`t zETX$4RbAe2&W$bmAgANZP@8}tH z(kQ5V&-$lDqmQ&!B@?vuc3l#wm9n3L;P$x!{w<<$%t2+sI)t?xHvQzuM7Lg!KMonX zgHUESIXPU!@->Lt0ad9}hPE?F-34Qr(-#R%HPBJyFLf>SAnrlLGjdJHUyt18g*6=h zxiD95cFkLKXUXCOTOe^XOnDjW+QOyo1hl-hpTfsREK>;Ip!CYWSn`eTxQ2AcSEY=9Y&l$&{~aX(ftT+$AKxAS zbm2fShjXuEn1y*?(V}F3w=o=}az+gU#mVM18q(7s4q- zzDEBXup|E1q(6b_XvXjy=VHRXQ3{>irih_|;|BOw31kiuB7mrNP7+<{%p_v|Bhx#u z*1#j&BwpZc!Z054dQVoXlY) zEAu=k?ZxEsGIT#Faw19LhRAS*t7CLTsv|2LA_QK^)};VsTO!O$C)`sw6jZ@O3h{D< zc98k8>5sYn^W&zN(^e1ar>0ysosX(ImBHKsA|s#|JyOUz0v1c)`8fesd{b5Yy{Ec* zb@_!&ip14Yoa1*AZXz_e_zt9)^Ek<24?D@6jGv?@LKCbL%=(^qs3as*#BjfsP6 z|AZ3>`<9v~$Wgja4T>`xWe|1O>fWfC+y}a-rRY!V=eb_&uu~6G&KO5sDvRx-^@qc7 z6Hb0_McygZH&2{7ZvR{16+LKhG@%I_sEb(nO~!n7OgL{@86; zdV+M_J~QkLDaZPcGql7i-o{VQfNo`!mah8Q>d1>XqAJ1@5m2CYg1shZY&vTF6zmoy z&GavNO4MM8GHg>^2&)8EH2`_xINkbrdK(6Bas}9B$f{p1mlwBvAXV~Jc~p%2C<41| z#vzVZx+)hAQDt(g_Tj^#N6CorL=%+acOBVo_vdksLY;1+U%9y5B4Y~E+iw3kv8vFO zsizBZ2b6q(og4RPNiyJvx)(IdSqB(VZ&R)_J1Sc_azB7k!*2(&gI|fr6~3>Ib;|T^ zA>l-lV8svOne0!B_0|RtfLWYsli{q{Duh}LIX*vY>uSr#OM@NrSNBi;-)gx0`Ip3{ zjLVPyOg9yeim6Fp7-|R2bd8T3TZKcfw64sF3@mfza8T9~2jGGt zFPf=&?d9Z;n-3KAOk4XuyjY_l%Cg9@Q_SJ?<0z#ln`5q(=UBE034&Q$C$FdjuY0cs z08U{f1_kx7aBxbU(bH;dLc(h>D@W^zvS~mZTEiM2t;qpBv^s`0=E9hc;oysu=``&J zrVQTv%*k`Y*@qAcx_=6HdGh5W` zNQs3AH$0*)iDaZaK#b#j<${)R<>N=3N*sJRCC6C%tO)GvH85W~sT`)k;10saP|H)h z#+|NrCZNBCr*%mom5aO&1~^N@vk^@zy^726)AOGL6?W zC7FqoB8upb-A!fpf`n3(pBCP#BX&*M)us$uT%qxQop~G{=68vhdgMU#pGu7;BxuCP zSg-a(4DBC=B^^)l3MKWYx4BPXO%@Thriqgoy_U7aYe0lZwxF5Nv>0~iC_wVQH?<`z;{z{vCHO?1q8XR=)mW3(5jwI?`^25U8 zqb`mYoy6*2Ss?dV%k5-|9WPYWn}Z7v+OPWW-dQXhxb^)%)&L0p@!H^{oqJou9GDTf zvU3+ZV@ELPIV~-V&n2r<1fv0-$0cEBi6=;$JyPK!2a84x#>xfCa++|U`I&vKA=S&d zoIV!}7OaMiQ6;47k}Q7!nyO8-N)W69pg|)mch&l)H?2GiE)hAQJ+t=sNWsV#4_tBZ zPhAIm8X=a>eS8h@C^}>=D6WBaK- zgg?}7Al8Eg7_|CKzkNxC#wYSCgRcjZHQV(rP_EZg!6#J)E{baVTcA)|a!|M21;Ii; zM<6Rdd>*<_k|lvG9i}zcCJA_qB!YwQ;@>q^GqMaH7p=}pv<1s)cc&>5ZFTV&5x3j~ zJMs2l5-gF}KM3*W8?(`JRKPWyhUYzz%^U26ES@?vzMR|O=wKWF;zeDTyQfTb9@x*i&>9I=D;A|m>D3>ZmmyA27c^*dTW{@&pH2+k$#+a1#i9+r*a33vyMPx z%Ee7E*$PsDWHi?iCV+aa9Ipb(1d!p0c3vigHW1Hn1y8Y5kjI>5cI2WlwN+4OTZnuS^cw;Ziujnfyh@2+`&G5j{cDUuoiH~a&fuJILGRb@hxSx-{-F+afmMQEB||jLM`B?>ME%IE zwV*c#c#IVejN25&h#*#I6L2Q;_Z5ZOl0 zKWu;K-1TD~EphJpmuraA`|`byZ^2&b{YAZjB2HFgsL9cTL(-a2i=50*QNc+VQS0-r z=u(8n;{;}8Yn4UD5T7+MWUPxD$+;>7+We$B^M_^B6u#JtgAyi5l60$S)8F{x6a$-} zM=r~@AAHcQ%e6s0$qA8duB90XNVJZV-SFeaOfBOo3O+!Atr!;TqZT-g7ovoSQ zMn~%S`uYIFkEz2)SudEISoE99@(;JF-CX1@j_^+<^YR9k2uM1;p6DrBqCw;)3@ykF z`3!`A&t}XwOD8G8c*P#PSEzwg_NEoj3ip;r{i&*O*60OmA}4tNxu}}lntf+Ru?24$9@IUc2YJtE0Yy)R_{O&8!)n$zqNB`*x!QMuj-r3R=pNKxa%Bp( zoO5K@N_tbvy*1YWW9fke4jQe{tI0<8>CK82T~`xXPNUumIJi9BM7WjXuEpwzQj%7@ zl1c}Pa)@v!G;6G%L&|xA7rvNw4DX{yJyBB#Kjp8!hcV-V->07O&m=}Icr6$5b;2hJ z2*YXC2bj%Cu1Q;y3kwBzko|g8IM4l>QC}wY)lpj`WBO_6*IUZ1M?J{Fk0q_~yA3Bd zYeTR3cs%i|>)*{N6h8#4@%XH3goRV1f)~%oLD+vai@0j$#5C69WzE~~mN zj`T@R)ye*Nm`quRePRdPetO1W2G?1drztn`FBcpwgkjuY?kO7CS!cmswfl(Nw8EB1 z2Zg#D(hlm1DwPIEk;n@1`%#vjIS9?_B=-BeCNhkCr2^{f9iG_~Am1gK(tymmQDQO+B zBCXWd<7pH$3;6@RnDEG}53>A|F}KjL>KQX0rz>h56~+dfZ*I~08a_*Z^QL#PTuB}o zI2R90Km2s}EK`K0bvbUQAF*nYm+&gy#XYai5IA-}uX^bpzpB(mxJOw~b)~s_Wzjuv zI-ltFhKK!}c)(P=xN3we{Ds=bAQ%2d20YuYviteD`8rJvskHFAh%Us#}k?^5n)dbfa*mBuKKSz-ai{QN@nhS#OvH3Vu-T}Y;O;X%)m?V*}FLP28^Z2*{#f8!7f^w3u;dCm7orv z756kqioYaHy)((}Fg-W48qR5lYPrmn<9;sfV^o&|Bkb7*CShc6{%5yVfuha{vNnW<K+h|d6y+JZn0J}dzhcEka4?^aXzZ?l~`YWvfWhI&}xHE+1_h# z@@evoz3;cAFlH_4AA@v+=r%7?idxz2SDV*EZ@2%Z)y>jRFNuQdQkF+_zmb60ffv?m zXZtEZMZ>NrB#2`~jsDzx_PhO>LyuV;zT>3mK2?)Pk~M1kcHFWSHib2Tn7o4MiBSOL z5zNH(*2(h=loPfFBjAXfh-V97=;F~UxyP>L$~*J{eMKf8LKK-i=ad7XCqs%{mwN4A z6>P)zIqTMUyb`ZTJ7)K0kRI+s)q8b)rjG^fd{$jUD(6nL$%bz+d|uUn)S~aU zN&d_k8wmThh|zkvcuQetGg3K+0|~A9}Aorb5Rc(`YwIA z?M3Z`g%mCW&DxH|oqh=31vKBTg3Os=)lccw(S?u|u;ZeDX%pJH9yc=!WrHUGzoBe1 zK6-NxJ$y;7iTAiX|G~*jZU4^K!j_fu`y2Lx+LUB5+zU#q!pqN`3Ci?LSKk(V@#BtU z26}GtR#mi;caPK5L}m4{*TN!3baTS55Jk+7H{qxSa*eR`w4A)Ah%d! zZQlboSlSvUGW!s}=-MOLHUKPn@bq6UA&d;D(M@REO>zX8YE;Xta7R2^_W8Or{K<)- zUG@*mb9T;f)GfpiZaIimUbI_DGda8n9$-H(=Y}aQ7uoSpE_=}cQN>DceY^Mjl?!CPtISr+_^gys$)UVkEiw@bmns78)80~&L7mwZbtWo_kdwC^#ZJ&g>rO^dgF_NMoy;))oksyrqqM4wp^~e_c-kueM9wYWTBmpj z@khW~y@B(6c~SztW^AHej6*`5z+Pc}X6JmRyqgHb7Y)_Vy(HbVJC0tS!+!!Byt=8m zztpY1Gq%=0@YQJ4#_wEj{4HACO2d4i=dC&;EqBKgqfv*nh1t$#(-*EC^f*V`D?Ts6 zbMi8->>f>wdQ&G1x(+KI@ACgn1FUVarfQ?5dEq1E z^Fk?^r@izaukUeAIg5$maf845m%G(F(=k}T&rd8ng2b>-JE_@Ig|koEixayZuE2X6 z?RerHF;zL6&8+pc8%o;oNO;)bnnzwmO~r{Y5adYBc*HXBz@*ybwZmRVDK^|&46F2> zJ?@_F#=pO~8VBBZ6(sD^C+8>E7TK+_|HYGVpxA}TjHEvi-yeGR!2Nw>R~=CM=aZhS zG@X%ncLMYnKc1J18c^g#XhGd+uM%r@;TDyHB+!F6qGXho9j^8M z9UAl&?u@{=d(Oh3@F}hXz{6S_VY!8;n3+(t7_v(A)sof~rYAvm^&4%%!-e-l)H*Rm zqDlUZm($~bYOBMGmqJ|j^KxSk)U0NMf@yf2<&GYmss(FoKw184t~r$~UauUFCm(H( zExlD?W;d=JmD&<8a}7-U1n0Ck`58PGz1_hlzPsUe<+(4SfG^`^Z~OYS2L|SQKWn~l z1~Wa8o=h%vO4o|}(qM=#ysk{SFSYol)f6+NGU$FlpY#D6iQI6IUHiw-DMW8G_D#1S zND4WP(Lk#sVgJ<2ey!5u94ZIK+I>v9n^y+5tBX(vQ9{>qPDB`uj1)oi-?n@$FC)ud zR71wk(PFG$Y42@(9aq7QU)HN*LOo1p>o(}8Re}OM#4LZ}1M(8_hhNUcpl3mH@>U`> z!;IKO*T4JtDKn^tEID<3k&@+mK+S(|MEgU}8CRS^`Fv`dT8OdxixB+g&-HUUoDdyg z^<)A6^U=A4%@MPfO)76>4l7>VxbEDm=5wcnSs12QPm}_ziXr6x01fEU zmM<2Q?gG?^XKCra^~zMx%U_er4dbDROlRU)sXu%My@kG2mT(Z07VipExN z8;4SkLsp|doTPFcHlTJPdyEj7p)Vd55r2L6bf^u{TxH6m#I}*FQ;Vg*{wPK_UbSjz zx&1@?2d63>pV3zi>rNU=(URa?dtYhXbLIg1O6H^TnB2Hq?~R+Qa`GLS|4wwmX_ zFZn2I;lWyltSNkIA~8!sB+F^C_@)S_Q)>|sd@!${MdVR{^6x+-n)6d{#hM>O*$)+`}gB!al+%-m>YD>7N3Zl;n=_VAJv?$w3#7FXH;C$ zUO7xR-J(><&u_(ns$A(o*6OB4px5A#_>qhtNg<~%zpX0e+M2`NV5FTp?kX>Zs-wG{ z`oF)*Jx&j|>2CrAI6RN*I5uON#N86}QwrZRBd)@px) zI+|}e&k^A{2I}gByj4`0)eV(w9^V)7Y8jDxWd+P`*CN({1z=zcmNp0w%D6&!E93~9 zJ?>VjnKArp@Jilj&|SIg?@EMg_0i)#csok!$K@vz;$3Ob!Caecs39@?is;!b6AI>6 z7>J^7*k!lFUMk(l)2q*)HI$u1>ZsHn?a;4ORomHVMp&IYTQ}CM zcZAeE>+*tsCPRhdGz`%|`OeW_H_pHY0Z|g#cL(xE-o_tw*DmvwXvZ#Q1H?~R=qZ-I zfO7ps9t)iuehqlvo=-pjuHM3>c~7!YYLm)hITJ3cSfA^z9hBCx>0ymMz8`eL%6RKK zC&KbM^1W&XXmgj{xt@a)#<~XfY>31W>-A^T3r!ZOa=kvb`_^!bs!HTfkNeHCB^<)PE^wBJ&M>kr!{@sO)FAH^5+>u4|nW5pospI1}a(;pY0Km%mQ ztl|u5GwP>P+UM%?KA)-*aP*0#2O|N`3uSy-cR9E1FFaKy3b%I{)*r{!pb-IgzDPi zHq0!H``DZT>6h5c-crh1drq|K^M!L-jRHi=VR(8HKdXILR;^IA?;_MefddGZnuu`7zt-e#Wy1*9~ksoWBuobN30 zVas-1IO7re1KF|-?_KPOAOR!B!ax;eA15l*NqC8p6)^>>qN2!)y=npSLyG!e?b^qBpQF}$E zbZJ4}`P$+g!8noG>Y~RrYfTF=_pGMSukWl-{^i2CNQ0)qfj*Z5BAyN!;AT60StPYH z2GaWe*#Q{4<^G&6?wBAP}t;cW9q%%*=+bfZf%XyqEytby@J+=P1W98f}r+D>`hyj7A>)A#E88| zBm}W3CH5#Hw$`rI+TGXvn3S&@~gore49#`=rlLC^LYw&B~h(c$DuHrI+@wjXHunZgq( zo$&XoK^;S4C5GqR%7?mL`VdYysYkX7sL?zwvoMehM$VL$nHGv&=%@WQd1<-U@@Ohe z5o^<5cGXPNI zH@o5J2=DD1XBmk3mW76|0#rgydYaKzUqN>ww>*6wygIVGu-WhYu*>!}QoXb0WzOcu zM&1jD%chID?h+Ogz;Ky>$hvUnKLXF)sCy$noS7g1bJE6obLgHIiyrXJ=c^0rfEMHx z!JMe765{o5o3O(##Mtm7PZ&;g*YLuShFko&EtUO=q}DD|A!6Jdx5$wT0cI2v-x#pC z*rrz*kCMsXMW#S@9a$T&ktc-gPMLRUR7vbHfSa$7UvvQesi(d!k(nu%TnW=2sUXrNKj z;~F{RvS)o#7BGaG?aHVCk4^>eTv!_9x=u#bjy75%K_+%Qtm55hVDoMgkMmThIY1nk z{YHy(5AV{qLT#cnyw?$ZNH^SbTpz#JnW}5@jo?5IH)p@HJy*XTE0-2|U}ti77_K=z zP&GiXX)}{?2DZ!{FHW%CL>in|7{8~h*B^HjJo$~+y)ZjpRK#v#`kj#J^%|7gTTQ9D z%ST&Hk5gq6-Isn;N?JKmG;`DscQuxs;|0c7m{#Tby_Mw3Onu%dQQOzh67<*x&=e_F z+HIO+ugtmPz$H~S(L8aQU&)afVVM553=fr-n|@k@M)*9Tz_lV0C2RePkPFseEkh1a zqmy6?S!&{iV@4JYQDxNj1*jYe|Au+VV+X`NJ-KGBG^7L@K_cbxOQ1V4F}gfXTI*dP zOc+^ECwGwNUmH!bZ1Jo&9c-d!K9yuZ`hc@m5;)?ywJqTqt^J&m;T5 z)Ylc=kiB)~llf7pzc#<4Dg>3dS8d#?j5Oh8uVWGu3Tk(AGaIJBR+xI+r3wn4ZxsaZ zR#&25L#bN|*}rV!=zvxo3}{xyDzXM4I*S`c(_<4-UiYU!Wb0v1d|XS3^Kr%vHj(4X z?4hNHmH?^WmRl03c$qrwlqCR60ptF}hBTU|?{JMQ>B3kAN}=bHCV zSYNL&U@`8G0L1fIe9?g^#`s6csyth$?CbP#d{4LU>LXkN3pS|xI+lZY^&s@bM{l*Ma(dLXZg|XE{%wwIG=84SzcTaJU6tW= z)@vinzHTYF_8Dz^srWKYhg$tthBG_h%B{58Ctuc^2NP?a>03)2Rf*pDzFFnt$NHQ}z%E99cINue=0^T|e}T z`M9AY{&F`^Cp@}wsNg77&2Kx&D%#hR`!}S7zfDHmoSMRo;Ku4u5fQF$kl(zIYPof+ zNP{Q?I>*xP%*hAiXMzvv%VsPvi2!1 z@Zsc>rKR22)e`Js!~T|Lx;k}A(4=?r;kcW%!u#yQK_s+Uq~lND9KkH`acs7<2nF36 z%Hgw#S$0!Y;SIJ@6Leh!trVAr~OU_P`1`poVwhe1AP*3TqVTSPu&o7a@ zRMzp?SmiBrMeBYW_1$Z|+%xb)N)v5DI3y?HR=Sk1a5@8EPOF_NMpMBbc@oXqNEwb` z@89Z?SIT%8sEYXFcclY&KAY5VqTBCvbINUt=B6vRXi@XO7`i@7tqq1}k?_y()K`Sy zwqLnLyc=s?q(&!wG{b5CYq6JhmJsa^Jsi!l+wVKxXyKR02a+7B;cJIa$wmFgDgT|< zgL1!H=a|=(Q)rR9a^GkT?>2=;p?XJ@rppOxV0D?~atCCWxv}LOeTjgsqO!fE$CUq2!~R`(aq-PcGN)GY zN-662N{mNL1_n;c1Go5OSLV%|$o;K;7k6MdxF?_{E+hW3W)3xRqLU!3p)019VY+I7 z2p5u0Tdqy4v$ggJGAIj?S6}|qnY|cRv8FrA0FNW1l)%I1Mu9Y-R6~Egir8UsiEn~P zO&)jw=@AQ`Ijps7;%8(7aAG(+)69Nc4&@d8NXCZG-MZuDv?+|v(%{HY{VG(b?Bc7# zN_v{w0$=jYq6YwZT@{<9*ZEXFT4!eO(qK5%x#$}LPXM$^a;GUL1k*O=c3x{Gsp^jv zV!ntD43tevRI#2J+M~Az2p;~-M}d-Sdy|n4Y|UtDJA6>(`AM6FNBln&E4lt5b_$Eo zW?7vnam7_Fxhwv6V*Ifokdx9iDp1|*_Z8#)-Mw8P<95wyQN(M&)&@L9hw< zn}OBj_^xJ)zg#kzhh(`8P+p4wOK9tuWmwOi!mLCIaMqNS4t*W4+m!wZDiJXqfC~gu z1jBTUiEbEZlB~AP2TM2Y&|zb`4y)Z-_gjDp1P+zeX`NyjPz6f|#D&Y_36ciHSwpc$ z(lWNB8XLCSd&JESjGL?AUBy}>q~aRP(*@D6j}o+ItJYIxl^d9_AQj$Nqc^r$w#<2x z8V^kb&Dq)gOt41uH)VizrouU(QoAAmOS#I0q?%bIrU`=D1i_y)uYC>$9Y)+QZQuDn z>4B?k|3);T&pP9NW`NcmHl?Z!pPpZ{s~fmIB3LctI+{%AzyAI2z<*7i%D=_`?8Nl_ zhoZ^hzbPrY_xHAORu(O<6<=xIXwczZV~Do6KKSZret~-T!R4JS_O=pW)neF8t1G*T z3cIgUX-{TrM|C8mM_=8!^H+XuVqB8Vu?KNBks9uqU_l=Ks6DVtuXm7Esq%k$(5Iho zoobj44(of=jq4!ZZWBm1+}Uw`$a|_(%W8R^2^C@aS>PIHX9!v9&|Kl{5-$Fgg+n&Q zdZK^4;WT284?lPR&URq!vf$BXwKLWVI6zHy6=4M>PIA-^l)SAv`6oI%L0_EcJ|5%m zC<%9%FDdTzOg)3&aD=?+F$`^uOj4o6G+iGCZu76boDU&zz`Pf48p*cF9|QApt+G?A zX7<5XWS@ZBK=amzMH$bv7`#p8FPX=ee^*C4=6931thVz8WxBZT1&f7F1cilWlx*=r zD*yS=idT+AexsYV3YR8$BEN%*Y`7WxpK%{jj~tKqYi@_bGg05BvRevUxKJ@Q{(c@g zT)P>GGkAA){g~b(r4|@Wbe-b%IU_8~pbffiFDj^L8P(9ysBX;AQH9u@a}o5_O53WI z)LBPaBe%;AyGGv14U`$qYV&7OO<(-6*<-&|6Sdb zwDOc?BS9A;?@SbW?R>aIkhSiGPf9mT?Oe&J6m=XJIHr#A!@aCT*)j25w5$~DSYIXG z(IW`!e()KoEiQ=v0F|r2Z^NzFFy`Q%u_%XYnOtJ|$tSMC-?kSl>*qkk^?7gJ;tev( z1iL;HB+u{^bIi?K5-Z{_XF$h3pNk&*?B3OiQCHsy=oEgcBU_*{!i+15;0i zlwk*5l}(Q{;OZLz!sc-O={cdXF>YeC{%`&@0q&)-a@%CPjE+6@Dm)|giHl378Vn7W zuGBG${&Jn1jj(`uYK=)K{6o$acZN@4g?ZeZozu!&<Gkcz>!(JIPF0(7(&IhS5(RZ(2KVP)2Sm1eSkAZ>T_S@qZ+Do{xR<&3%HdHTX z9Gs>tei-<)vCg+tA0zNImL>G6k?%(ZeNS{3i#dC4Aw*5=xxx7ox2b-eLdf`C?NF9f z+GC)_XSPwm{130e(<5g7=F1~OsL|bR&gQCQ3q>v3RaC+@FQk3_k5s~E;rkb^J*&@7 z*M8}Yy|Ec78?h(osO4!fS6=A}TN6D@oV@jgD;}eASz8s}TU&X^P1>}_6Ll014AA~6 z$#8PC?a&i55QPBF9&i{>anSl;UD zT$P!0SD`uUEp@kxjC zg)}!XS`uHky$2I_`tsiUMObZKjw;HAU9paIm-MG5Lk!Tth5@qEL{|ywJh#*n&>cj< z*WG(9`W(Zr8oEEo1p2)9?I+meSngr1y7;AiGWLs+u53mNL1Ta_&j-E6%pT(h^uN^~ zfb6{!G2wI+^$*DN7#@@!>@tZgrObh%SC}GB69voH5dP*W6CyGr3^rWoKZ$S8=o|rX z4v($$s(f>_Z|&0{!AbTDJxvd+hXq8-}&d^r^js5MZEuams!|a|42D;n+Z)SfBn-* zgAReKa}pxVia%BZazlMaOGiUct9h51W+6@>ts9bQ$aLVJ9ftfG&h{lMT&--D<<1fc z%E94o6c+ie#gFnF4jE$2Aw8wk6HMV4Y#n_k|F=U8=4cusS0R(zU+*}H!WopHJjsKI z*}tetW&+in^_aN^-K^ELS_^finHy9#=>Tayas`VQ*X2wQK2*HM)8tfd5S{XRsb@ zNATrtNwtacJ0opv!HtH*b`YlOD7~i6sUFUT@*!6^gCr}bPN4R`U${uy30|_*KAn1SCe{|X94bi~kz^ZuL%`KV zPztcNJ2Bs%QPo=0a%DKPd7+?-rzsSptMvQURP^YNR1cxF&XzD&*7VzPg8s0+6Cm_NxGf8Hcui+%6Ab4&uH01maM;Xy!?_J zmD)cX4IX(T>6-ahJC+AN-+0+AUG1D8n1+M_bB2TKw}HiUDFde=#aG?V*h7jbb7Fc^gpe1@{hLUzl%c=h`f=4d@YU*5rKXu4l{{fsc8_vvWjTF zuG3Zi^AFLX5zKF5q8bey#c#l4ljnNSgB(O~=1B%_D#B!I!uv%3Q(@Y|(Au}MOF2g& zGpoEaLFv_vT?#S+ecnQfE+rPP^^8PyH3odA!cpfEJ=-*J;Fx`=IATR07)k|XANgaL zAJgsoxvn#@3x;0SeX@=cxHBf)(#S>D`XRnMdvGnym1WOFeQ8un&9kFHk96gIJH}>VpG{wh_Cv1*`Y3|AwiV$n!JVj9G0qspK_QIaF+fMw*&!7 zEZ)<%G`Hv#KS>xqm`x9`&$Egbiu~$!+-kaS6SDawHRnfx1XcKdC>SqY7z%}hF77*E zdVXU9ISYeIIA51di`r+!2|?;|?I$)E0im{jOuwqphN3|FSQ89=D2e6+v)ncC$yG%}9H9UTXA-R$nqf?uIPjhUr;nxEvX}25omnCtJjo)k8u8=7LiDc4C5ljqN zUOR0TGBYafqg@#Uu6cz_)3NdXi4|c@LuZWMA*vfOeJB}k4}Ga}lAFx*&FK|QoJ%J0 zUYsJP!tn5T2krQD>0C~;7Z#KPG1y{c7M*)!<9Q0a#Ctb?q^1+f3Wg`o%cG)fzFkpPw4E{Z`p)ynZM;QmC{}>b_JwJb5cU{;ufRN57P0vUTCd07WT}K* zs$uz=&W!Lc$L&Q_d>=Woo?%BTNal)cpgGQg(EJiB!$0L=UqqR3CtyGR3n7A^RAII1 z_bR#wTGPpW>2|ywKCkQuB6V!&zSW#4K4d;0D_n!a`oD7UtXnh5K%hH&eb}-{Z&gao ziKacIoK*OA0ah0-#{jH6UAD;%opL9=^<=NI5Du5`aupO17cXwC5^I-W?Sk75YW5No zEZgW|pA7dF^L`!tR(Pu#?Qve?F=9wrj8|_cd$h)AUuyenk$gy>*|>JD2qG_RE~dox z63op&QV@5caq&8k|9TsWKhd@D3u~)Iv(SW9&D{9CUiwg+Ov6j`G$vqW{jDn9A`kCP zRnM}D$c1Ly(Lc2LjA@PX+y6fOM*iywlsElL_3UrU7S)gGf41!Y7d>{kis9|OQe9(K zo`+_c^6nMhTi~fkqNe>quPDY13r)=Fr!0Z3;ZLVdCi!g=7hrK+sMg=)4IdqC?<1CFlkqx$>$+N^6== zK%8831K!hUM@|mkVeTQG+XMs2pWa1Wmj|}45n2f_-`u;O?2q3F2-q~rFw-h1C!DTz7RPd>vSNO&}7_e@s($WxOfK5Z8qw9I#~k)%xeqd>^{@4mnnJz z|Hn;-8SiEJA;x&k9X54M$Uo&N!W?-(M^w1+NB9Q2r@H48j!z^B{&xI*-Tucru3ava z$Vm}?m2Y?Kp-bOV&o_hNsV=)MV#jdkZ~u@R2tAQ@P^v^hVd1`p%d*YK81Ay0GW{AS}AV# zpAW7;uB*Nzre}vW8YFKXGp{fw6)P`NeV)=)QPcSOmSVAe2_F7_dB9SG6R^nRK{bKf zaXmCDo^cy|VX(P3S2dm@6%^+%ASm5vcD7eWd4vyy(x5JAP4!3o@qn!g+(JU_q3&@p z$u$k=DAlhiOCl$V2u7V}c7ty9U}zClbv-G9mY?UwURS58r$_T@cYE~3Bc*|=-$tKP z7V6v9#^$V~oYv&p8HI!Glb?4AaB)7~c0)_TnOC z&mDS!L%vK~LFQh>>L02=)YyTKxs?df)V+e1+LIk3fc=oMw$B+$^R<{-OaGCVRX($0 zbh38~t*%}QgWg6r3ZO=2uhui_aR>I0Ut1pta{j<$!r=K4GLZHbmyANT94iqg)H$7$b}QL@b^t zJ(eB=3~7b%r+6V+o7?oF+laY9xoffG=W_qb;ENfN18yrF51Gt_mVH<Y@2sWrn^D|0^hUO@>T`&tZBRMt~1*39K$Er#w&W@gsijui(PBf1itG{!T^v2 zo4Z#9hvtcq=$O><%29RWGe>MDZX_{gTu-1< zx``;@#l}7iUu<1n;#mSXKLv~&ZSb69CK@6I3KzwEFdIHQgdQZ#%>*{QXOvB6^|$*? zXDRR0TH~3zdNQ5w#DL}xQ4Qb6UNg{#_*ZfabPcU}?`}&~_z?@;NXgDDh_ew3!MCZVPRDcjUFQ4eF*G zX9;omshis85@#he@RWp)?=-IAaWg&FagJ=Tqh>>)jNSy9-=C4IH;}0#lqt_Wx6aV@ z2&D2(l}7llT5$O5blQ#+yltI8-FKPA8%}#3y%v`yB`R6D8;X|`j|I*IlX%wmtK1Gb z3f3}>?%S!%VB`wFV-e&rQhD9SA3n%*U?VIk-9B2s$?lq+9_OJG75X=$h`f0NEjX{TALetpP)ei}bpvg7H4V%sYyG8bT z!JP{Wuura`JTrpE!6wfSW%{}$B`VdSL4zz~{4Qyku zskv#B_%oNpTu|M08$)d=5ezy}fVK+BmKP}+==(3ZXLyg|E;q9FDu!J`um#9Jf6i_nEfqw6WdW~6Suh+X_y(26kruG>v6ID`)wiQ&epekMNayQ zTte_4ElzeHO9e`_oc<9kSjvKK!LT+gA)fI0^!#2gt^jL`VC1gQ zOdsa@SgiL=17)B97=7V%bpBoz7conBR-Pl@08RZXJ?ZtR>>OTwjxKhcCVHeMs)1UW z)Pzj&;LQ1o1-y9x(k}c?pd?EpCSDXBoz?-}1{3*dByAUifq`EO6KcvvIoF_%-0C-| z*Grwl>%$kt;C$(6KpWHPJ->C9fYxw)a&TWDP@w+m*P>4CA>CQ~JYXto?8wTJ^ig0r z2lL2j+V4vwdeQzk*ZdfiTr8%ZosH5~L}#|<%Y7iO%@z_BB5JR-KlWd6@d}e6ta<(v z8HyNgX3?JFWps3k#c;u!8`s#5r$QmZt6ZlaX8$Bl1B>6M=>Ftdz}5L$<{#PR=JI&m zx~Q>0(|XIZuZY~O;aYs^x22M}y&|XvMAG1mF|VvlTM#&qn5OY;FHxvF*l`R{Y-%DR z7%DeAP*=ECG#)PiaEE`1{^vS@^g*#vtL@Zq%&&u21VH$qd#D)jcqkaYa(R@OirRc@ zcw9M`wCh?U^_=HPV`RWF6oI{sRU=ja)KZ^6L9DPd{kVlVJNomo>TDjQSPxs2A6Uq4 zb>nsB!Ee5!pm(`>V;f^{H(m>vVCRX`9bfYpgertKs7435`JHt zYG+m;<_&W&iVt)^u3cf29 zQ1$HjaJT_}*GYqk=(|fKi6RwVOe}&9NLJ<_&v4B>jUSId6Ks_&z;-Tcj6U6RpJy&3cCCQ@RUP3q~XC;+$CV$B5Y%`**VZ&$PpxIGFos9WsFr9Zxsgtau$< zd-ZhmH>fCFV)sFiA&y9tFlQvFMkWh)udwu+zz|8TCS$aYjn$M2_ROcxWIH7YH8WaI(&RTfOA(7%P>Ygs>4pY^q7M2UQj(wsWUF zK7XrIg%ywC%g(LyyK_m-3K>p(ZuEo3^c{Ymud#h<5O7oLs@z=&w@H|lp#|~$tp1|a zIDlh?0Qk6!*m(+^@;OqNcUsFAoR*zy9*}@1Gl|TQG5z`-^WT%`C$`Y~_F7W`OWpy| zr5L?B@B4OQt}EixEXbUe0@$un`R!*D?-Z)S?8B>*7)4M!iQ@Wl zKXyaEm?(z?3J2{es~MgM+wTi0MTbm(eQcs_7HuOQ@;F{eKIZ5{RzQ9~tC`q>;l=;QYl@XS?5pri=)w1SmuO>X-J4}*JhT{W~@3tuw zG~I$))wAWB)Qb(OQ0vTH=^EC2?K#WOS+B^cUBnjgaBcuB$tycf0ieEtO&$ysm{f3+ za~f_fuK~(Dhu*q3X5ZyETQL)T=+=~gnU*Gt=+Up&N4bH*UuS$Q<9UOBAv<&h!4{yF zy2(b{YJBqCNV1)f+O0m=ZsAP>SsOqwIYwkAISe<1TcIjGUNgQ5-+IL1^!Qti;(xOw zmM<8xpQK>>pIrQpKbs@+Bf&i4O${Ga@)HX1RP`&l?VS|dHDE6CpTX29e)_`uW0z#o zq#WE&{T?Dt@$$$?p%rQHyx<`uaafBnG?M#kDhS6{A+A^3U}rQm!cK#K-k8=pYFcz1 z+wBm8-%?JzF%6Ffd9VjJ*L7_9T;wfU_M~3m>UFw9z+NEhD=c34yOM|L<0UPVD&b6& z(jL}nK2JbeEr?(|b>4Qo*HVtmXcn6@oxVL|e`v#;n#OJ!!P}{_^UIpLO5n8*rN<}* zm=O%CaMzU8e|m{g^nFWJZ{^1(=!)O*V{1!sAd*_56zDFGI<);VY@((NzuO7_=_(t3 zdSYj$Bw<@@0+CuC_Nhw32w845eMgpb5_CH~SH$8A&%nq17rHQh>?rcsfT!?SCK7Fa z;6@Wcjj23*&?Yt(9xE+U6|T#K8{hfr^2z=u4H<2LUWq492r%03p&$y$AG^I?3ZVmo z#i2zudk$QrXfcskn2Z0U>=d1!w|(WtXyWL)iF96Z>rVl9xqeyQh9sa0142HdgOZeh zDUC2+fSb0b#G2W+X$}Z$uwgb?+{EZYiK`#3k8HoU9o^vqZ*RmFGLgIWn*+mQQ$R{i zV>CR_Z!w# z+D+XY5c<q?UaE#Eb#jVHYRAGh)@6DQ|T=|-ukYd>LkR&QeVvNJ2 zcvl|-JmJfX!Sh+=Zec2oYqbA8Oi0f6?#HqahBD+4v<}Fw*q!lAloDCrw&(=?j5112 z{d{4a%tbP_M~N2kKa-*X*&lw^XQ1~Onv7&eJ6Sb-fm^R?tMFxzKNSl&zd9EUVO%k* z#m89+iO=Q*+_UYVvzRvp@Rve8Wkj-!h*kr+n+#Xy8qsy$f$GY5!Z1Y>bz2I+)b;w- z_D9jtSy^@ptW&gKfnHDJO=RBewWdI4QPx9!mg~j&!dlLx{^G+~nT90*@nFTlPsu(NU3e;wqn>)sZ8z z@vY1GxjENwh~($I*mMz6ul(KHxd%>QsqW;Dmvt_ix1B~Fg;gelWcsDkO2I-4p9(wF zx)5K#Svaw>9RjBxjby}(EungZOg+M*^3xu!@97$SVLp`2L8xZF6xaP2#x3#ljdW8q zq$)Cc{82OTgc7yX(S-0GZ@wvw5Z|85l}uO8p|-4<%>Ud)IG$k`u|jMHy!|!->G)AG zC|xFzJaFA8L<;@HuScc7*JWKnu8(F5iK>I)KW+jey8nIw`+jePF0?-9#7Lw|s(&*F z?xQ(ZQ>Au{?l8qU*8ey^gmS0sxVx6k`=MpUiM#BMVT$kXEn#{uagFPDgB%Vto1HJQ zZ>>8OtmNX|v+U?Oa;C*$gI05cO2KwTdT&5%l1maJ5ABI(JQW@w>^_~h{CRR#BAH0DQf??q;;*vl zWkX3K9H+@h@Mj^Lp{y~krNh9)=C0?^gfK;NCGR03&OUGL8uO*Yesa3GL~Sdmt0;Hb zOOb-ER2H>B{QB>rv>N(}XEa1eyYugSbAi8dMnnb1n9ns(<{vuv@Y8`A00TX2gkAX9 z+sDZ^YY)&H#6r&y$psN&8D<^DFrKh_w`DoX)$8S2(Pi&Wm0A80vSXOHQLFc&&@$<*`%AKiFFs*ci zhNtX=V_1l61U%Z?x-Dh6N~OD8Qimd(zr|D^9bx(H8wzb}1N+4}Z^EZ_PZ}kQ@>8_# z1eC$%7^OQ8zo2m12!k?S_86 z;Dzm!-O`u%;x8wdKZbo8j%!;jkFGp&1ryZO;R4h2 z+IsNgEQ#Sfr8=cs7=AaF>tUb=E?Ld$`X6M$Nbb1VE#v?3ePpj=9-cjW_Ro^kf01)z z&@K5lBlsWo1Cp1UQq=q3pp!=&JKHx~byok&s5HR{vKO!s-m5L z#BZ@JU<>02$5L)s*W)GS&x`Lqjr|_R^tzGPd6Rsz92SNO#z56Pl6^i8^-NJ!CQsQf zf?+LUQ}*vJjG5%p@lS=jXGav*k%=QPKX3H{@sZBAo7VY;g-kZj*{1W!)pRm^g}-Pr z>QOmo8f>>mi=DsJV#ctA^MnnZOym}E_4;Xvmiru5V+PR%MKd+oyECyfUvt zONcLEa=zu)iFUToC1+194eiqa*r~|lg%4YJ*$7@6s|l7mKH~Da;t{eTh)gB^K_g|>c6Tc8S6=SlE*=TGpk1VfZh zvpe_4?QkEL-!bD$Ly@V05Y&+*#!cIz9S|)iP!A6HkaBX_&)WO-)GrsER#VZ?)chnRmyu@!WGAG#Xp56hQ{{@Cy{Yc9lVGI@m##iwABrC5){eU)F?F}bb84ZJm^ z+-#aci{;^lX{Y;d+k6A*<_C2w2H?}B_U{HZ=GDY(R=bL45R^WJgjMtCH>yXs%psba zu3%9ay*>gd?S_R=g(_N=Td>{Wp*gw^N!u&H-6ob3O=GF#l@}_bSQNdSj5}hlweI!! zjbnI6E0TkWT-S00%AzVfbx0^cFqJD?PeDe;E7Z|fcP)0TnT4~UqS^t`;6AN|C`TlHUD3td%{f$H$nGN79@|_pQ0Gu?|E*<>yFpg#kqDo6*lkn z6>Om3m9^1zC2U@`nT+XvX(a!Z{o|Ct551b~#19>6XhLgf=CDWD=LqcHr=fQZys{^} z5!k!B+@n`wkoe85l*!IK=C@6M-F498hWmp0*m(^_jx^5I#4Zix&oy%K$1BlVxc;rh*&? zc6i=f>w8kv(N1BWoKKL9K z)>g%B`t=z?VUbmrR>8|(_Q>h^%fRosjIt>$9K}bbovkAUoTA?W7BOVF=JzuE1l$(k z^GqYzsIy2x?@rTUbO8v~0iP`nUd~ly z9)=Uyy|nLe0OJh1RA9udxaiq-u(O+Eot=bYzb6|0V{{C8_g`)L&k zifD#~nkqXuYlZoe&a>6(J3s~g zce}+7&E&eLG`uSt)(VsD=|q=aiZ5i}^r669%?`Zfdx~+<8J(`OVGYjPN7*QX%AUKs zgbV`}R7qBGaZ-=+S!Y`|DuwGRm@I$hW9|zLUAg0^4(}KAt*~23>0l~oc3tLs4QJ*^mJ@ssxc2K;NELgC zAhd%V6kMPcoc7^yiY7|!kPhW(i5gJU@~otG<*2S1?8B#-P3SNB72N{cb0t8p+O&sI zNbugqla_ZA)4=5J6xub0g@sSBJG;ftAtJzU8L0rtmo&opG`Xfs*0?ee?)n}1-%EEr zo3n`)=C5-ILa|v^7<4qmr230VU_a%NqfUFAVC2P4;Y(ne?NW`~KOTm_4X>ODe?1H9 zb@{zsM(F8st1jnl{*LDz#{Rm%@;73}<5(Wmo>|dy|1buR*g6~Vyx8BElz)^?>k6_?C{GvPJac6z|N@cD{maQ(b zXeJoz!EW{O&GpXcaY*3;Lo0SQ%?JOnO+6+!7&#*7fx?WyBGDW#IQv1^9CTWCV_mN@ z*ZA@9Z$)wsQ6xFQ@DobNMis0>Qo?kxTfYU04Xv_qd}<4HqXA9dj{}!>ib|Wx;HE^^ zIh&1Xl-g5Xnt19}S8O2A zVo8)dM5f5QU?BKfvd!h~woApz$;b3#&VF;?c3q z1qydpF*r=IX-2-fzIq@H;vHC5L5gT=h#Ji>tdzn;2SRN{one9#jt?Z z`y72C4y*Q)`+?NmU!H0T&A-X@Dm}2zmh``qG1v4$ch+?Inf(XE^8)o7iq-g?tjYQ* z;=*uda8c=YRl^S0bt z`ja!M*A4kLTRmBJbE!Qez>U7NK0Gur@9_*^S6yPBRf+SCVlr$H4`+&K=B_r0w0ZI5336INy7sneJx4Wfz-0|~`Tc%OnJXjx;i#t1$4#@UZwNB!pzGHshRX zl(2SleRTyN%9k_f2Ya7njbNZS`~%ra_if?sogYDRD*-DNye~Qd<~3CXGIch(diiQK@4LNxRihT9Cz@{TIQgh$?x(*!l zVl+j~K*7`HoSthwKdSoGRLD6V_vz)WLQ)R0hVys&BT}vCk^L|F&#muMmp*&8mgmY> zhe3BgF7=(hz(HvN(RUXGR3#}VQy%L9F2RUv`{DE!SnevR=lzH}7prCdY&7Z%+Q7s+ zcy`F{e02MA^?i#W!}G;}0IVjdLy_zu>+u%DtUEIzsc|)sjNq1XjFg|z%H){b@i@pFJvR>6}S#{13(P|J*CHMEXqgyHNC?G!v+iZ<)&$i=F1L8xY$XuD1tfKHLZdV2GzFs|Zl&GFjai#>t+%2~MK zAMM6T4LdI=0^tMr#0Ael6^`rooNxo~?jbh!9oHT`>#P?IgS`^`-~Y~kd5kb({FWz^ zW@4jiQx}bDSorp|*ohCOxegN`Rl-uPFpNq`Da6^8mb)1SH{ayla-@|1V*cD_z+v2Y zgf*V-we4-nt#+?eb9V-y)SiZn_%6T=*qt9)StHFZH*c3C6|_;1Rov={xTV2T{m$Iy z=EvoEpLc$1#o5DqxX|+=N6~^e8YrJcfQQs^>V8w$tO#S zROjMPWvPggpkzr=tXek-->S`|0nn0>+zdW4$>4S#EAmpcq%qYk(A3p|rI^b`y^ME~ z?%wgQL>E$vP?v-m$hL)PN~UglE5%opym6o}i}1Y_1Zekyzo+4{O7S zREa*rUTkUV!@g_#_{8^muYHDBcMojbV;k*f$F&7vOXLS{ppA|*--FDw z;#FQbB70HCfS&u_`*kX^NO>vF3}`gC7+MR8hJt7mF(%>wt_AxM)kDYIZ44PZH-Qf= zP?0xoBm!0eoYD^1?%TKbS#))!Q#Y!4iZMY!xk_W-98ax3%ld&qWtNB%+J~zFiSSOlY*OB4qLvFd; z(~P<9(He-uYv6Dw^n!J^8?dwjFSt*Io+3D}B4?N15W1j6mFM-tQP^8U8F!ld!>%#l z1(ZuGPwNs!o%C1s$j)W+%4>3T$kj1CQ?E!WXY7%9^rBuxWc0>aVup3*^F?P>(EXHR zp!$VLuHoNHyO1nz8QtiMucth-zlGb}3dROLkpvq!~(+Zogy8x9b1AOJ` z8QTrlFJn0afF1j?-`X>ezNf*q>^UjH^3YLit1AJD{XCL4S!^ldDKOOr6 zd(^Btvg=Ty0R%m@N*CvXsjTO>?Tqlc5JN#r&>t)Relmt@975RGj~sToLQn4*`_gpO zAIEbbIr&#ouDsLsz{TMiEjdac3fJ`FXxG-(-xguPxcje%Za9)%d0G%F?t!#^JaT>a zd+m3tg{4^oA-Ac2KPjerp_YD;d~2SQimPf2YSs(__>rN4Om<|G5d53N(qf%Y-u*M= zX+@;+e<MH#Wk8pES$8XT^gMTPciOV9T;9&cT8BoBTLuJMFkX!ZN z6&qyFOUw*^6%3awM-;PnO1TJDo2Pj8J?l1a<_hJqKJs^eovNqCKKmnp5aJ^U%xZ3o z%q2*}a>}S>;BTf?=A>uHi!+JMEu4co6M4RGC5>xNW}-9rJU@px%qC|yq!oJ*lL4^t zr`+Q?2l}Ke(J{>@bn&lKZP>;cj+5)q!qGNys}8mgy9-0HX3y}cg08VPKu54qSicD3 z&H2x-1`_Z(VoF_Jy`TP+0GDM%O-3Q=X4Xdtvsu}~b4lHuxiH0VX2T5ha)BWRecK0R z_?iRiO)b}gS|574tTZHg^RFi829%Q2SI$?=A4$UvY-8CLnlwrvFfAj6J#ZwT2XcRhx0EG+`m*gCSs-q+MU}tjl^nT-@4qqvq+$ zs>+YIcr8PwSPhDYS$Src#y(zt{H|dvowd;oJYk)cXr_BZ(AiRfzEl?MXy;XMpxgd` zhG-} z1T3{a=D?MjrM5of=KH(vbM8Mm{=@OS2e0dTJ|B;WzwM$`19dO!O@_XxSgG-!9CoTe zf6!KeJ@xDLY3PR|b|Fabfp8i4ijpL*2F>rkCmGPem|Hzvk&wPp*NU+C$61fZ_iaQUo?QNkbI&Y3*L z|Cf%+SU1lR9Yo-O@cN##f=NU*&QKd@a|$>`i=!%m-u@2-QI<=AcI>|M+V5Er zmye^(&Nfx;jb~k|FDNTKU&nO;TJ-ctFQ<}SY857CAH&Y`(Z((V>zY8gW;OGvMCYqV zgjv6QF?7d+UIY->8w4ShxVBu>zlt(P%FnvH zMvLHtgP3ee(2`V#xm2n{?s~Qs^)gJ+J-!CNQcag)dgi&eK{)fvdQ85Bc(kj3hkOj- z15s7QPyTXBJ65YOtF!%P5!4MUu^C`7Vmr~B9U0$PXp=>QcKTf5Y*E&HPIzJ~vprrU zZQ2JgP0c5D4eI$tBY|i0xvPrYyh+Zf@Q9|UCsm=LoBlyTXuYLSIX!Y*iDWkPc3BTa zk!;{Y>5vx(9hsu7eV5QH9#S!cqPGhc8tH{^t;s8YNt1@9?lxLig%;L$T356FX%sl8 zBGjCXn{VzxyawaFGKz_F-wK;->tiJzVOX@X5`gKRZ>onOC|L4~+YIew9G8gK$I3S# zFN*|oS4-s|JfB{K3i_n+fSoGaHOUSefzeuz^<=ACnZnB4ZOLxJ+hL;#l!Se$n{fCs=k3~@+NVq0;RgTpgt_c5S zT1?Y1yuZx(ex|LM5pXS2oQ0Hbo%o@WwgGn7FhRWxLe9@rUv|@@&Ycb%pxMkHN3|M# z^qBr`SgF5zZv7N@%_T}j1b7k6vUPf-*K~_rgE7j{YRQ*5tOJ4@LYg>=<)5ncJ?r3a(H}5`m3pvJ!O&~xi zONp)kRn=c}H~8Nl=|$la{>qREc;Rzq!G^*UCYj~xO4S-aqxaGCSUi5{-i*USMa-gy z#EVlYd$DhWvbe|$=0C2V>BrAQ=DAdAbBxKCFC56@&ZL$AUs-RDX?A#ajBT}bfM9~}lswNb75@d_HAs@O;q zdSXQRf#3F`qeo~4d`TVf5Bc3hKud#|)VXCLI1ud!vz$@yCv}pD?_!qu@}BBZ^e?Fm z-7Z#;j!fF-ZnWVFU}@)c>JnzNXI!{}Z?*dbC=|2vwwp1xPyMyC{U4X_6O4DmEIHKs zBBNbj=hN$%)akr|TGq5aqf4g5ySI30_=hmSC|u25AZ?N76;)Z%<7Ctgpquqy`Dci) zJ^(eHUCr$pMpv(%`N3V2f6DB;GT<b!uP+JegB9$~zl~%HoOASFFC>LtE4kwdTP~`J+_v_eW|(@m>Q*Ft zuG#zgQx3w2(FE>(vhwx5v63XP`8(D8lyqqA^h+6U6l>*_>dt!+&;PhGI(lNu+f<(U zdKt}N;)3}u1Fphg6~5_oU8+6jccReme|RzI;~O^v89tsBuC6ulm^+_2M=33lP*=$3 z{R6clJb#s#ymGqwnxD|7lFW*yIntF9O{pQ!o_OCA7 znfk{}{{JF=P7kJTf4lY9e_a3HvS`G)OQkJ@EcfmoZB|EnlAo7>(+R^+IZ$v%$ZB+J zkvlZbfa>}-=(luQ+lbUWXurEEk#R$53JlzA?9l)CL21Z3v-_}dd8KJlC>hs1{N-I~ zm7ZSBd>~SQ8?MaeYcB%jMOM|F%|)9RHYE}R>JdL^?h1EJ+bmnX_3Q%}P|sq0d8W)p z{Z)Txb+4q6nLJ9l@k1m~0qyrvo+_J1H=vdmf8ky!w&IEfm_(Q+JB`Ia?wZ=jO;vu7 z|DsR%;7Es#X(}}K29lL~b05tKUYB2Kus*ZK81j@y$jnN-I>vBj%^3K2a1?0^wwTon z5b~6W3GW!F5%`#9>{u@xe%7U~=S?0iti&8UqlG+Edbd$n3YWM3(P=}WzQuwsqyZQK z)*6K7MQ{k+2&5{1^1SF5l_iUCDkp}v!NpIncRyVtTc;@EyZ6Am!!*H0p+;N8UQwYc zBO=1sA!OGQ12Z|1dtfs@Aokr*2w32#*}Us<@i!{)6d967LL$#^|HCtna`7%{Ln=#&B_sdTj-j^oxI z@(4={Evhoz0IS~8cHzB?E509#rVh#S{_d_%&-c%;|T!sBE~^#;c`i>166o=-dO5_ z%0xRP+)q#K_q_kPHWABn2Xm|*m6@nGE8KirxE8zCBn$VA(`dKl9XVBhSE92sMS(*@ z&S@hi%q#MB;VUlI`>xXEh|^6sHx>KuHrtAy7;RFRJ-VXy2>gBly*fU*&!S;sFQzk5 z?v=btYs$NxN4p2m2k4ydu6)|`=(tN%)yh!YNDM+&wCt`sWH1U%&v~Z9F<`x^?)gW> zfU$Ou^=}E$hTwH=!n4BX8S+IA>K@_1)-aoKUpoQska<(*)qz_0R`6sCsk%u-O37Ylg;yj6O~21@3siMC|aZukM{s?)%?- zd9*$A;i^dwy8j1>YW(;qykJ6@2I zXR&s6&gaW|9&FN(%@xI2v#6G;TQj?${*a;IWO1+e(&WdH2ydMrWbBJ;xNSxw(LoZq^Gu|m_pg}f2Kmcf-bLo850|D~(qzql(3 z{q|gMsXyVV-*J86FA{FpAbp~q*=H6@=88}V9--cPd-KY3Q^w^*G;)2uRU$ik ztncyD1AJ(e3Vx?%=a`K!KgTsn*gd_p{&pPSDF5u_f#PG@p>jsCj^=+{XIFZP$8`jg ztX7i@7XBD_3p6^+njAKj-KiCF7&lOseTFp}-5g}btp^6)9bG%vaXp-Wwrj)o7vBI^ z&Q_e$Fc~0b+ud*|Ga3CM8Ggm8Z+h^Acv8`7;y;CVUp%h8{^GnCxzk&=wgxDK8EIgA{UWEarPSy!Y?$)Bcr)a&>WH+X~)j%`gF^WcX8|L zh4SgW5c@~z-S39^U|xllGvCE4g!68LQb8vQ5OR<5Rb2AcqPkjyQxfMod0g|SSL;jCRjQk$_ym?NC`-Wn_t7=k+6prQ_ zRWMxTyMxGIT*%Z*6&ga1a7mNulkDqNv^72yOajlSVOa;6JoWJI46gXb5Yat90W!Sw zc2$SyHlUlx%+M$GXD+eaF`sbXn7rkmQkB6rxhvM+0*R((0Wi^v`sB{zp;8Q@=g}RS`>qClzK#~METfzJjKo7kbuo~diF$#!(1wdSTO-iQLP6TA%P`dQO{`Wi))xE%>4OSZXOTU)? z(PTQi+GsrO#DWsEN7u;G)X^Dn^m=s+@hp*CkHDixQqZMqpcnypJGZMqIyx_o)2 zabRnAZLw%l)01A6b!UIIGdlI^FeEYd-=gCuisE#ir7cI4Vxrj%Lm1AABPIzbhiG zsrFf18x=u?Bq|+XC%QbYCm+1ff9cJCTo>5WUmlFfCS=^)Jfz+J!@X%5@a^`UA1&^# z7shyej<-XVEwyF@+v$#?JFTI(!Y*j8c(=1>8^Pfovu?iXK}XQ+{O&m-Ip2YcwB~3! zw{Gbs21c`-MaTaw`k*_Q9Y{fL8sE(;5?h;BNgzxGs*K1M1sbI;6v~CC*H2QWyuOaS z?WP_wx?Wr{I0uRPqXO~5`dDzb6?GH2|SU1ltSmBM1TM3amvY883%k7M_D)44#f@FoS1=Dc|5E$Y^Y)ymr_hYRqTG@Lliw^+aTj4&exyFQlGGQ;3w~(P*>1>&l+vsO^p)EIO zF~tqZQmLu7m%#FkJVMT-0ZKTk^4{=9^WQuzA%DByyZd0PIp%*+<>y-#c3&?ZzRK9! z)qLH;IRTF5J=?K}UQwH>iD}|wH)oQ*syujqy-?e8Q%|327doNDSMr3Wb5f{ti z9XB(G{Y68M@*`;kG%M{+u#5+7-$8bThXsh^KV-?QeaV1}zu>ooavCu0Z@zNMQp>@j z1-eC-M4jsq1ESf3(3nOm!|9pyNG`hl_bdc-)Ix8~V4PX*E3!8U%Hw)ndoL;p_1)75zsBI%@zU^Y97)4A?czN; zpJ1dA{rwFN!1@#(?52$I1!z>uTzbs}ZS>-jf4zyS-V}h;g(d6l(7_zxgT|JeuyaS7 zBX&m`rTzj})1B%ilz3S%1c?A6H)Kgum|P2lmTtfKgd?)Il$f2@C1GwloGZ}bnPSvT zneB0Hw&IMY?K}J6qcvMYZWoqgsKQ!NTHcG?RAVzCM=rwjlnLmOBBx78;A90QYSk`cy zw)_)mpjs(8zX~m^g>u7>c4B?G>%mCN0X$UE7#M~bg5}ubIvZp&ww3uD&_|S)N7ker z(;l-tq?*8oY0F;#jgzd{!aqO3T_+8HQCFhrc>_gZ^Ck^{X(1dOt$E8pQK)8LU^xiozJsY~Z~@(QHq6Z2t%wt$>{vUnTBqbZH-mw~WA19GJ4Smp z4;nx^Crig+GjbXEe3qN&Pi6WYFLEF~3$IsRB-~d?0A9-0JETZbZ?3%YYaM>KZv}MN z+>Qw-CjRDm^YU6N3)DTl<}PHc1WD+6yVer!2JRe^zyEVD^l2v}uroISY82cs{+_tM zkMCKU_%hPNfyupjrrJ}WU%Bsp+2nfx%*_j2KM{TP#7ru85-H_UbT_vJAIoBm1PDim zR6QZw8p`|?)7c&aS--o_wEZ}Zd$#d9pp{84u#%DLpM7O7p=d=^EK2w@>+2DRHTjd0{<19Q_tlmb`R0K5h6OZBp=2py#N2%E5 z1=*;U0l2(|3+_UYozD`uGq%6zd`U}x->CuIMy);TMRs14A+B_;%)NBF%kJI!%*^5M z&eYB(Ju{>^gGy^(b-77!G!%~4TGmT~dbhR$R~Q2vznF9lUx zCDG)Bd%`kko^6rnYgq-e_^0++=aI~m(dH{^iK3gG_dh0nI+}RZmam+BCwcb8RUgc3 z4Q_PT2=T2ZAiD8)#oS0o<9hY4G_?syN87;OwEM2tdb}H~z%xC<7-+8MEO|gf2A)3{ z4%OU7lq9>)<)ylAL|$y&lLx4F6X0siY|3Cfg4n_?P^h<_FJ$R*uE2n9^+e_c2ZOo7 zilav8CH2QYm(Xw%Y=AMm$k`<$EF??fVWB3JSG>W;k1Fdmx_Z-t$K&`CHd5R3&+JH~ ze^@Hg`&%)E9C&(tFZBdp$VjfS^j(xfFwL})JR_oZl1Cx--%Aj?f<1EI8%_^9g zQSaIHw(C*u5`Vq{+Zl)QR26kp%LS84W&?MDlgVkd*>?ZT5HeQQ{!U4rbDt-?zd>~P zikdqQ#ZsR8#iBefFXpSw`nX{1LcjpiGSs@QsurAY1k2#JF!DLqadP&=PbeoqJJVl3 z!N(WckT&4#oe-L|Zq2EBkHD-fevzMu*!0_fR=1I??{0Lxo(6h_hW3Z*?CqqSprKtp zjv$+~TD4-OtVgh9K(icJezv$^9@oaFIomJ4_2(GIjyJgB1wBNMT_$5qZgnbcTvl~fQ~YY@OacQhUg82ZGG)_gFVyMo}H0> zCMGsgAH|`buN2Re5|<35T^ z8y#5b;wzJ%H#FtY&G(HoF`|aY0j|M7FURK<%rPTz|N4*HcG8E~#=b|165B=z{ z%;Wf4_wv0$E-6=>*4Q1HVR%V5fo}BUgn4Uk@`rM8d1n8ZYoPdagiOqibB)p6@F&t{Mlex=@V}C&#EzP(Ij6rkLXk;8dOhZ>S*kQZHNFm3WvOap&-;M*u4oXtK{i zw}lg{9*RCY%Lhw2XAycU<4#cLH{^9DQ@#MS=Z*(WI$CyHA}7tkJJ?50k(w(oDVzP&qufcPG~=ocqmP{=jWT|#b(a@sJv zAw9fuy9d8ztiU49{_%FDB8tPAe4N_Z9_1WlPyGw7>ir!wmBB36d0R^P0ldhze5)Z? zy;7-ae6C1UQAxknRT6UMasg&tOWlt2s8(p)onPQ;*D9IiX~yxCD|z)G+96Ug(avvyvG0KVhjH>Wm`9hztuc>!@T zPwN-@3@uyUE4BWfb75*#Dk)lAu>S{DR`*Vvm&68mqFjx1vu`b|Ebhz8%T&=($b~{|n zga^A3L%9fe|3PV`zt0HvbY(?;4iGl{sR*$>ciKTE+{KUlvHBZ5T!|1h?64t79Owe4 zkA$*cguDDte{}9l6RT+-*5nR!58+hx;NDhy^9w%nD&W2U}FFFauPNkk;Ck&*3zcM0(+~ zHywltLZ-?kX=U5x@35wiE2XR0_XA6AZu~RZ$$*S+MO(1Jq~sJ7K^+l>!KQm(B&#jU zQFZLHaymGrqtH>4Q4Vem{Sx+QxH&Td^r&zXHH;Ab96htl$YVs?<2E()z(dIG^Wt0` zVP` zWbR80-Df#(a(U0zEfv=$Fv3mVowDskh255NR*?tS=uRxVh#*U*03Cufu9bS7RGP~3 zMWvRH6*a99QOaum8XvO`!Awa&1lj#Zv82PHH1MAn`E3TxbL+UE;pgW|8x2-EYdeLK zhXO+KQ^igZsYgCm$Zw%K_5M2m?|5}*9e#EnoVz2n;x4VM^aQJCO2nSG;5xBB{Z+dx z{<%U!zM1ZaBd%ouKK2Vks>6bZmI+thY|lF)cATDbWQtl+S6RY1y2Q^i3&9r%cYRcD z)ZH-Na>er@^r&wF{9$~zBGaJnheL8(-mx+!+@H;qC5o|W@Lg0@rR(mM*j$L;hb z6vbqTv0*~)+99x)cs{|8mZ&#x^K$jb$Ib2e{!*?s4L>8k{fa($EzN>6_Ao7y2J5j> z-vOwomCQeqCgD;)v(7yZ2E-v|>f2SNb*%ecayrH*(>ymob6x_CGYq>wW7n^t@a#gIk^NY1ZC@wA|t@x)WFMou?g1S?M;fwdCUA5i&pK6r);0janp1(;q_o@9H8D7)rg_nx#xKe96Kr zV9s-*1@O{%g>v>by6X;?N(KCtd?G^Ks>4p8$6lY>Ql+BfG60_Sw0IPHZYvVv8LNi@ z723KizlB2K-V>kmlya*Mvo=_Dwa@h^%4@Jl6ywl41eq!X1IuJ3x)6&L7TPmE4}@K_v#l@f_Hvf}Ip zhpaE+%sWliSIV=bn6-LDPbX%Fa1fzLXCCmu-BaLP=;SV2vzvyUa(*oi-uz6a=15s6 z%C&o>@lfHtJ(AY+0QH)0|7=sE@9R{BcEaH~dNXlAbX1CiV05$&B*beeLZ01Pu{K}i z9e2Z4sChxDw#VVo5D!%}2E=pCw#dRfy&j3TCg4zvITvnNYHs4JUp9UQbHnE3%fmZ# za|@*%$5MnU4=!HH1x2?7(yU7fdPk%)hs5{4r?re8MI&|M4CCoRnmcxqyK(^MyMFmE z47wXm!ajcM$i5R9Pkw-seIeSeUlDuJPXnFGnMRTAKQzwDkit-B0gJ&WnVSKBeQJ#~ zAs=VYrYW2=kl@E-F`oP}!B{a?iDsM3EoFMhdoQkK5%<>1Kq1ZSRza7*pmVxbcTOhH z>yySjV`<(~u-9BThTYe}d=-MWJ-bj9^)%&j2U^;g`viIe-#caItTywJRn8c8nKErD zwYO9uwLUs|N^1V>=drHcu_n{@E4_e`jJta4{h78EFziYZ%F*`wezbYJxRa++2;BjK zDEoS&v9|Ipd$k3z0K21jio0{A-_rSewC7S5kAzQmpp*nFM*rxBfR4XX$Yhlrlm7vl zee|xQ(pF^1HUDrqWw=e978f8fn{%lIMj-g^0`gMhP9&VR&)af$y-2IxgL$!@h2Dv> z(-hywO*4#coXp*~IO%FmR(c7&dY6`9+!?;+kKH@;g3iu6!V7&16A=NwoztN@-=QP* z+G(piRD+7XuAVL5uir^6Q#-By^g1QLAHZf?p>rGiUTe4uigza=&d;qit9YZM6%wZh z^i*1lQyWz0TP(6lWMY~k=3?H1{(y48T&uWm*%!`2Y|lv6c2CXr=O_xB(;BWmlI_+E zr)KTgP=$u$cPw)}u5`*i8vD`RB+5`T?a^y`cu=|Ev6)ZySJ%XTB_AUc^E`aZBYS!G ztzcC+)u2Wx62}Kw0ax@+M0|M@a`-Am?YWp&)_+|8pQs7;;Iu$I>-)b0Q~$@9>-*w( zaKQCxyeey)J?z@pHzKNMakR4brPOrqb~4V{g0%RDbiy~vA% zujI{R=Zb`)Qq`3!fp6=uv``%i)Gw$6R4Czk_Qcq}YgoA;i{z54GmmG|m2E640~?g| z$zfc8Y*ds^4*H-xr^Be^Mz`N3r4o6c`CIXpkWf$hJZ`A6OOq-ZmQaf8TZh@$qLj6B zm1?nOT&J&t!v8PxKd%co}r~u}b`la;29yO$mHd?#NC=XdK!E4j&;x|Lk7WZdx z(E_kgg^dl)S()xO!IT3Y&CiKLRnPisBBP>%~(<2$a^c~vkM>Y&9Ocqc}A0@4H z13#iWxC;Zrvn3!BSo*qiQ|?{4RE0|k+xRM3!!i^Vw%mz>Wg zwb;8S#xYQ3qA}U9_BKR?#(6?1>iwX#|W2EJV8}_sA`!&JX{|T{0?+_5>=z)7||fx)QEBN zu*!2aUg@w@eMAE)B3rG%BG@pSLcAErOy6q9?iRqQ3#{96x964M@Ml)N)_LS>Y*a%k zg7;k_3u_s1?Nxm?=C~C{s7QNOMc8;|!p-NHm)hTcb9ACBYcEVB z<@f+oOTQwX;HUuz$z*rLV`TrcU|!L+0z=VfyHUR$M}{KjA=e*~admhRc9v zc23tdGlMeMa-`+`=4_s*hecA^$Y;0t-*+ccI+nC)XO>hdm;{@^tEU_YZ4J&#Sa5^$ zg4LO%OCRu~V$z|mHU~W|T^f^T&Q3EVC<$4(m8L6Uk6rKkOJFB7HLRBoZarONPBP`M zMg9=r)L7B(@szFGIF^Pp!b-9z@8nYkF6{2CLnPM5h>KTFle`f4tA3Px;3wnRTUGcXeJ{kx&$w{h*p4fAX?;j9fgR3`W~Xf?*?9r&PM=GD2lQQ zz>SF)rma74{*bLpq%5AiCPorEvkOglz>wD&|5-173asgf&dP!{*N!b$kO23+?)QgY z(F1+mVBzvrs_aY;=PvHMtcA_F|L-#GiZ;}FJqw*hgiUKYaOc8a`JA(5|X4+fX{Uy&D5uDG9fQRTVatHRcdRvNMYG zNhf0sMs^*ASTj$@6yV>!D)(X#A2fh`00(&m*(}Crhj1i%)*ldQK9;5I+=@Z;x;V}@ zIxDEw0R3^P^&aJ7jf5ToCB`B{InmkpBtvve=m2^K2sYcYYi2#&Gy~x2<_@SKYXUUP zhD&Ph&2*gt!FiYSOw+w1nudX`-XQVMHPaBjkRkhJzZ!H(T5Y)l-jG)xrtsJk*CXo9 zv`!yo31Dx_(kPy=tot$K7NcQQ7PbxDBa08&lKmxtyVhn3TB$;molOBEWSAF>Eld=- zDPTkiNKjOPm`tjNqSYR|l*`7aV-Tyh)bqYbp{92b@pqUsJv@&%DPyJ17h4;5E%_^Y zrCYZ3$1X{DLmC1N=*FFVhHO&iuKD9|j}bMg+-dt%a4;vKv9@Br8z*O<2kRlcXvmwc9PMyTh5%$Xh{B{J~nlj&RdG(Sb~|cj;4jIJSFsiZ|@2?BIQldGI$A9P+o|&2UW9_oGwjt zQ1%eRGnqL!5lqn}*_gp6syXNy9p&+D+(C<*>%HfW zX2Olb6Wi7gzF=JTuKu#&Yp{D8OLysenJs68T}@Mrjt5_gE@x%rDC~Mayr-}wx#Q)U z0ls=Q5UP?>qW<=%37IA-S8(a6qba`32>$R|Sh)8}IS5g2xWtQD73 zLtqdU917J(bO6Z0#zJY!$*HitFzC@mtkRG|dTCw#jU;7J&P^K%zzm>1M_boY7T*N; zaItHlHl9267GvadDIf6R)gm4&390<#LaN|-3oVTxe#d1EU8V&U z^1!u8{twfOy&tvOuQWLkl|54{mdJ=0q#4P0t}F|`rvWiXr!_xQJ-uLY^;#rq?%DrYwh__ z^;3x`(4LBrjT5ejZkB3O*v@a_u!!T9C>Q+z;!$Y-{0vJ^O{d1|b5rG%wJe@tEs+%` zu%S7?44}aQymWF2jqB^hn9G3R|9urZxK~;E%ZK+m}ZakuX zH+`KFdM5A)tYtTagecY-)WAplTT{Jb9o!hYiHw4)C&)OP{%xnVoDN+5*~%-($|s(- zvz<1L1BnMPe>Ggoov@%>E)@@@I{A+_ZYh$wrC)ofcHw8<|NW-x@S~b+)k}?!_5NUy~Gj$RERQJ1ANe zdKBjQaLsy?Mjkj;SbQromc4Sgd%Bz>}OpBY(OG5HqC7(9Juo=Y{@uP!R*B$NwiNp zp1(tyH>zhPzv@kyv+VQ{Nz^5j;ZoZcY@ucDId4|`KH(c_1dzt{4!i%i8rM|^mCQ@9 ze1pq-UJ%{8vA-=+Muh*a)1$uMjp|ocmEpCLCP$q0?R<%(dZV;o^v6mIJn4{$X>2p) zP|zXQn$Nh$IfwJLWla$0*}=R`9&E*xzBwo3{N|Cwqojx2m)Lzxa5?%NYSKk_MPIhE zZE?NKduf#20X#F$9u@i*^rmh74V!=hEJDo-4#;I`Zl3A-w{SClA(C(>h1fc0$uNT1%1BYBOR!mBbQ9 z6bb^(?<(08cA_e|M=5!%2jXU_->g&aM*E z8RyIE?WBUiTILXllpEQJQD&cix|{5GQ@Qy%)q~5gJUgp`<`Ot${t?I#Myztz6kRmC z`U_px149HF!NnmK^y0d={~n6x=n+rdXteM=u3eor9U>hFbSPzl{ls3PLm+sFgv+9F z13rdQ!Ndowlg=krU1Zm&q#B14NM~?{>Ac{cM)D1#DTlRhIp`+2m5SM;#(eAZqKgt4 z|8eOZ(r?HsH0Ge=891Akb_mW%AKwrRK;^!RRFb8Ht&oj>k~6>avsA}jWo6S0AvX%K z8BF02k0l>}T<5iuu@ud#Fo#R=`!0Az9`|(skAulFQ{?XH9VgJQea8xYalHEEem{&l zb7MzVuABLGR9gAF21c#y>Av%mF=8vmx+&FRUET|Zi`N4S1cUX&36F{-C~~V*II;+s z>)DxghPo^5(k#!v;V;Ev=dK5cHmaSgP@R3=cNgPBTxoFYl1QIp+sNDi_7{>=(ydg499a9FK?^nrQOH=}qaw%&$%IO)y*|mCcLn?OBB5KmX z;mZn19p&|0NW!)}klUa`1O>_!I?=~%iqm^R9|=Y{Fg+cgnaOwtu%RQ2il{QJ{q<)`v9P zkz80>6E9i~8#)Pp|JZcLsl#EV8eQySImc{96}l+;TV|eBR~`<-QgPbw2`go}n`mvi z^yUai$IuN9MF(WEp7RG$Rs9GX76$i+9;#p+RRD2!IzSB_m(*k0WwJHU<7!}6K7zA{ z6)e5B8L)!0L0@Ff0 zMm%OM!O~Agu897}bsbAC zf;bHoa!2f0*}*4bZy+(>l)+dhYlphwM1Z%l)a_+b$wt6_NfRALz;|dl? z1xRW%ri5W{YlA)Ce#52avfnh}ay}U3#jS4a zjJVDF76FTANM>7B9e?`5xoGTMimK2M3@K(Xsns*HmBpXz#oNjqTH0(V7^Bmc%^o_H+&k`5vO31WKN%uk+No8NW6+6YY#GGhpBa^G}M^FAm)>maQ-z zLYgnp-Z?c7kk#OeBw`3EtNs+u76?X zYPAVOBf|C|eXlpNn9;9^@BCUwykP&X^ImUUtr|jbo0@5=2ClZa7lqp@L%VlA@qNyq zW_0{GoArFM&)`?m$C<-IO#52VlvyvU$m#Dl-Mju?>TW!oXV;wGElo-S%|6QpOuND> zw&x&sH!dQFj#Qf&!edpWek_qaIMz6pM@{WB)jqf#W!bc2d!Tdx@>b9PzfRYteA)BdM+Gw8^BLb^)RV| z2a@X90x?2A`aj@YE0`h@6DwPRql!dy;Tor`Ww=0s|z1%!~tNBI>J z(_IB(G@SjLS`aG*2#T~YdIgw^iPq8`7KE046?89l5Dc191XQe{AQvEtH_$EmpdVp# zI_w@rIhv-VW7d7fQ3fOPqCU~S$Gj^7_)TdD+9ikWBecJ0d0b#7<^maV^L6T8NwLoK z$Dt}xS$UJdw*E^J=2Ox8@K4Z@RA=F6TVBY!^%hwv+^drb(!j_5Y^ojR8*>n<@WomL zD`C()-v(u$ z6{evB@z%d#$19C1{)H>`ToR8lj+Tw)niDCFqnfy`-c%>+69kZP10T)griQl5`0`wd z%mKvkY_xCS$=K@dBMP%9&9+qpfmam%)`>Eqn?A-Bg}MaGuJ^{UUr683J>kjh`85=a zbUUebM(q!u3^UR#J{`s zah`nB`_w6%rLT>mpi>(QQ+5crisETzH;zU-2Gh-lpMxjeLo7qyqqgzP(q&Zx#B|!s znEv~Qb$0GZGV*rN>n-OoKHktT3ziRNzf8(t6D95uw4H+i1qk&YYRW5>Rlw!~(U-nM z&_v!Ry)J@d^{NnlZCJM;Kg7r!;+As--Cq1_7dc!svFEn+I*3BGa_ z&Oih&T-%6S(9OZZawpZ6iX96<-K(9Q+Xh4^FOoert=UE{_94LlA@sQYr*nfpqU~Nu zZ_A*Vm9&UPw+CP5;_*YEnQ=Q3#->3XD2p0TsC~m$RZ|`ULr&$nJRZJf`}hxoYA>fK zDn!03H=HQp^!N~v$Qm8D02!Vm>-yj;NrXp6e;_4;AMIr@}zsBWwz9kna)PZ`?g+c`lP=hc?hW|JBdwDumfUE)F3*d4Oto=GeNA)f{}oZE63i527j*IJv1I#lnt? zAv%qISrk~l2S=%7#Zg#cal~xar=O%qw)T3^qNiX#8b8k!M@fjiP$x+?NOHY31c9~G zrcIfBHD7_;jXhQ0cIEGc5Q<~^rH3I0kns(@Hvj*)p49`E*|$c1U9l>$WT>&@1sd3m zU(l$VN2V50)bq8C`y>&GXC#uCOK>WpOD`x>XY56XPijt4d61<=_+oNrzl5^ecNJaC zgUQp41+TdesrDFn!xd%fy0n1ok%qe5aHvrh82#~FN4X5U8?ENhZJ33h0{LYjbagH^ zJPN=z<)GtM_MvL9bD-ze9N# zF3Bi4WF?|KE5fca<9N=NS=m194eU9y0c;+U0n1@CaFfnphqW3~9Q^)w2SF&-yaCq| zda7Khu`vpACxrE9jnf*X{PK(QA^x$ci?*NRW9m`gcDOj{FglC>`11fFO)=irY^iNu zh|bx1_+XG+6_CREoIK9%lrVBRe+fAWHo$c&P%sAl8C#yLLt-k%LLF<8H`uF^+~&?^36DYvIGni}5vyFtXXjjQ4)k7=#4*RU)LYDC4|n za&rheJa_^r&HjI=de5jP)2MBfA}A;*k|0E|fJPEPDG5>?S|9`hq|hRufI_-z4x`R z0&^G>6L61%5iiZ~f<_9*@f1)?uJ;pK?gGlSy*Zc(^E5mXdA3)JvdP&>Qpt=_*&mGG zR>q_aX2I!M!GC7KO=(kRBsQ_{%YHTS>lRI31wI>i*GE{4o z^>T;wFx?p)ct1m{)bdFirdF~XRt=o9yf3@Zgv(T}RzW!)lm~*O^1>tZo-mZ;Tu3b~ zJ8iN0kOA=`2YjT%%2D4l7+sRu1pd!EQgVxn@!86q+y{g)`fa*G$W9a2T5RKG8e zeVKWq%;%(q;7J?>?wH3quG<()X!+ne(3*CYkmeaCvQ)&&Z0Y$)=J%LN5o7bAu z@Vv~WFs;N|@3C@VYV+nckM3g1f~i?7>f9;v+;nGJ>$6(Y&vz+vxN&0POwm1pe#%0R z73;xd`@R6RpWdcHr9Zv9n8xu%^S$WXl)HWNib?31-D}a=O!45k=I5_*oI=+G#n6#v zUSjA%Ge68`J}dVLS6ftEp*Q<$$Ic7B30J;<`2BA!9mpSBe9r*Uf4ma@BeQ{G-_XAO z?H?Jv#o>QsHoafdLWwJ^3TDZF`I_2(H>l7;b{D{ItO}+V_)qyc53taIg@-prscZd1 zBc7BYg_-$RfsF7_QQOwK&9x?8qC=aRtm^>@{AYc~dS_o!m^NQN@{T@%eGA-Mc2D`5* zPvp=VK{A`mZQQmC@vYytt|6ZQXN%~d?_jtiZnQt-aK&l6=D9t{Ddr~2hQu}5F~AEOgnjbhhF+OLk5%jy<il5)UnX21)ejtvqC6EHm3#bA}2Wh9BenF@S#l7i-V? zkij9Jn9if<;YS_G$C+O`n0mt^*y(*pjEz`9gzf0wqi{~(f5k>%`5N0)#CRJk<`jt? zj;#w+YFQU0F1^lw5G$@BP0;e-YvnXuV&8n%leuQbdNN?p!CU2kD;C9_h1nqmqz}&b z6Lp`E&mxy2#4TCRfo`o9Y@P-*sYwkW+mJv!b-yj$d>rY+_xsG+a-xVXNB6aDmM1RJ zw)U-Y99;P>6sB>U!^vrakw1apV9}zCyz|x!l1a`?A$d?`Gcr6?KyzJZKfqnQv(qw| z&@F|ZNGas;!yN<-2DWy$QMbOcw8iRmw@I-cTn;>Yl@W`9Oi(K*b%Q`oqW+)c>KdoP zAzG&N^g6e{p`#N{vlF2`=yr$z#Z|^Ol7N?=W_TvUC*zb)N-eD3()MNf!Ag%pwf} znI@Ea9&tZ}YL~|!P4hrW$W?wew^a7Tg;lwrs+yhp%JN(fp;7*!;+C;$`chBtvTR$! z?y4ErZGfU}VIVM9Oz?g4#u)k2=Sh(P*J-WJ@tDEkZNYSOQJTdpEP1J2OhV z-}6lCkput8Ts;J~jbdQv{r$a1#=aiA$V_s*KHmPXkK4vu^15r z#x69w%}Vhw``lun<0_eMzaNV9L5x2WZQQ13y5VVN-ZO^@fr2W|Ary-o?K8X!wbJt} zh_*rq{M3~PBs%98VV1oEg)oUr;Kwf8$-Ldo>&r@Nez$4k6X6M8tb*iP<-0G zq7Hp-2Xrw0vZAEk4iWZ;%`v2Ux^#)Am@@DB)**xZ=x^-I%Lysk*c2je%89yse3x?U zY=-~MWo2M(@7e)o{n|6$Sz6R%(>}Q&EXLu`j1T{nv`Z%OHtFt|NwI!h9S&hR}wV=^ts*XoQ&b|0I4OPk;9euoq)zbwd<6didqg~=i z!P9~kHT;7hDO&m+IrlVCcgcB~{D7vs(0oWQr6kRhl!3z-*Ajw*o>i0sv6qmStAP{N zkII8XMt>W!yX$Vcla|$udpw1(L_O?zw#M1|25*p&sJWAiSz{NC5cn{17G#v?6MjN|k< zik0alBs>BoaR(PX(Rf57M9XBeEe$lI7eZtWFVZh~pSD0-&_i^9;HGdUjTK6}RStPg zatQsZO?=0Hfp%J~KEa$#eX2a^Gt#y-d_ow2EkjkD)i-$KruCif9^F=y_}UjiF!jDw z#h)UaBqBP5X0a`*b@c9bq(57Fu8R7JE=M~`UcSIB{LbABXDMckQ;iBD3$r4iA@~e_j8T$spa3i@WOik4%v0ADIRt=%?@h9>m$x ze++-Wa-=nrbf-5!5(muO=Dsn+iNlzxcr#SXLS61!|30mzO$}D$RFaVzl>ewH*bA+D zAa{Lg5hYmw)Xk{1wuort&6k0QG&CRfRWz$`zlv9;t4n`hu|;4xg~ zvT)TYsd?Xu^n)wVH>idRoIz+xFNL{K9DaKL>1eG|@`3yeJp9bo=Ww>iT8WPyn*qqF z-R}%r*x`k@-Xp!gNnfyo>}~3;>%R}Of>5yV$y7vS3C2DU&~AsVU)mnfU)l_9P~^On zdFb!oBoBuFge}+#aH2MW9&jw-rYoO<00tmi%JJq$ZAJ$ zKu7zlQS+5I@DAa0C2UVoARfnS3g-G}7ACd5V10LisxbV#S9@3Uf$#LsyTe;3+#W<# z(mJ6hXr4b83-WC;`~;fgZA-;AmWurs6PGnTrvqzSMnrKG@sX1G2ig4ZJ=Q0`v)(4# zzCM2#*7Bz1FVN5$OK)KQwQP3d@ZF-@X4wg(mKn3=G^@v662yqS$StiTIC zEM84>vzH^_X!<);d)~*;O=|~lkQt9>p$4yU&l5*+hkYQ?zm${0M zT&t61v(tnm-NUjaVxKT<|331q6Djj%^=A<9GFlw zrTqBy!TZyqM637}cEP=bqYck}i+CLY1tfn(ZUd{I3qAh0@%h+Sy-iPTos zy@m|lGcz%s)RY*T!f1F4<101=xUS*w4##>w$NL&aUbtj>VScyM13{LcUufLf_Pywk zh5;HZYBv(i9Mig=GfrD#p?b&b&7svMHzh!qli^&D4@p`3^SR%PAXu+#&Nc+@J5mzL*2p*HDrOja+WLLnsCVJR@LVgch! zo_bP>#pqs#^h|c|P^I4q1=>|rD7hcVnRELzA>vrimxxy6K!%tk!w&27SHWvz6BugYYMge|YuRJ>oeF9rNm*k22QYu4|fa z{6Es=b?A(HD__}P5cc&M@B2rSx?Y6;3b^v`p+M%`Qo3`~ba3y+|3g6^zN7Oh;HxoC z+M0wx=Rm4?-LYg*_EARyX_7~TX{2A&XNeVMo&)vzTT6xJTik}CDfhF2FsCgSUylaY zIeQn~3FW$d@mOp-n5!z~ngslsyO#25Go+x}NYMg%#kQaFqS{m?JkYTu!Vmj&EVzuK zS!cn89>~i<@5Mc=ccd&c;!{)eWua4K6F$qlfkjm=9)U`CGe#4d&paSDu zM?c+yJ*>YWAd^p$KH=GnoT6SX=i%<;kwXOwe)O}GJx^h`dcan1?$*MWIQK7sy9hgP z3e%|Q)YvIppJhZxYO(#`0j}=g!2cJ zQf^Dp!XmTmwT`B&XrCixsL!X{PLTXcZ_+>2QNFjsqgQ$J=yP?8(W{;F6yK}oRkF!X z5|gX$`2NT}=HJO9Xpcj_P!k^U}@ zfxjG5U+Ij6Hq`w;&Dk zjBI??wF~nBT%x{iT~=&QW~u#Z)p-6QEg+B2JhRq@*|*yLHI9Ti1k!b#*|k8d(5EOGdGdRkx><-CGZ+ewV}=qVouY(J zcQClm^320tH6mG&3o~FYcCn)@j?@^&&ldU>xs%*0^Ia6HFtRTuM3hsDW&4mDj$6rW z-K&`N-XSvt`_{x1(z8d+>j~fty>61D3qNWEZ}mHzW!03GMBaeoHA)P8#913eNQi( z`kCFcd=fT&BSse5gAP|bcu>dP|JP?bI<22xfUs3aSn~V6WH7mST3=8(|A2*hLS2bR z^|Q12QxBjo1Tc6|&(vV!QH|>|;1!6$f-n76zya}XwH)f8HJ8@*#&iWD3|0e}3oFCF z<_rBE>Qvz#O}PYV?m%QU7aYxfVY$XNYfJu^f2>S1cQT;*0nik87$V`+f_ygQM^JFt zl4Ty8JBB$FQYd-HkjfCt{JTBn(GQ`gE^Pd1tp%76qqq;S0UQgCz%Q1-R5Ma;Vpik9 z6}fINO(Sg10@cqSB`~83w~OGke%@PtO3g@Dd{x8BLff^ghIT5qKegI?V#b8};h5o3 z+GJMznf3|a-G$)+bRxcl(*0H4Rpdm~@1|6JpY$u~44brU3pD@pyu9BFkC*H572jo> zjz7%vJMhxu^+J(xJ~5F8C|x}ox6WpGr#Hrx0>pW(Xet1AZ%W?t%*NN~{GKJ1gikx{ z+ari}n#>Jjapy`{{Lq7HAPIB|a)G>`U#|Yt=EvAWg;&AS=g+}!sS|oFnN`XB=x6); z8X+%{hQ=N${sr?fNa|mJVZa`4sT+PXK=H9+RAj=d#%pN>v6#Ht+*;|TPekie(nD>w zMPYA+EFz(v9<8HvnQN24mwxb6?8)C>%t3{f^mMP{v`tJQ`i zX)`9JjqvhLSAf$-^JBrwd4~h2QTvo&x~57fpS!-b>@-3zcvmpI z)jU;zo=NK98bB*+{+@mLqPK9${Wy9zEc&GqH`XuA68r%Ol`V*oPP_;*Yno72f1r7Z zyL^!@f##P^im3Mb!>!(zz!0TA-QLWCwL;bs*A{u>OfC8Q>VUdseE|)s1>Mgx5yHod zoxkaUupIaK-x=U*K0Hupe^=4WP|BMvSOer|7$>FR$H@=8F>IpFMb z3s~wFAClM=X z_h9t%?$0IJALO)&8f^kb71`|thd=8| z!!tY-2j@N&?tBiIxjz){?rZLA9Vv9aDDXrvvU?5!k0nEtvBymGhc?(7O|wLJ`}zn;{GmWrqzK zaVDi06_yW-emy@oP;JMdM3%sk_(H4|iO?6}f1>?{mGomAf~(^G$*!@MAUdbF#=OaX3OI}xNf~b^w~58XG>1(f-CM3!!FY$V zR}2-!kTo5$8S;6E+x@7N9$D4vEPChuXQ?|pEzbB5Gx(Jo6M0-^j9m%TY?Bqp^4I5G z7b|n#e&rb5rG$Y1O{`xmSfxZ-6W8tZ0j<@F6bWsbXg(>!WyN!$rxa=DU^ql-KZgt9 z5M?t-IB6dT1^Q2}cqv$F4|O?-*w4JHTIByx+OWQ~=Kc<$+cUFmS9BJ9(uPV|)-A61>-Gdhe<^Xfq z&vm;n?qzCtp;?MIS3bb=oqJgqxIvqqHR!(#geZotq*FG8#tq2YgqSP=H69N)ZOCv(+DcC$Z7dUW~nQ* zf&XYk(%D!ie@hyuTqi*K6o5K46LQ^5GNy^2N&u25#9klxQzrsY0tBzj8f4j08d0p( zawF3VADAGudpup5zXmOI#%pm{*|!Q{3KrYMnzNw6=gwloSzm;XeiKBuW_H%iF$-r! z4t$A0vkw*tjUe8BHDD*#eL5zj+*ZMx9vwR?^{|<`bHC8n!`J0>RD9RI5yAD0NBwTc z7aS~Cd|FddgBNampS|b!;{CUcd5*|Afv+=eHSaS-w3>H)m#PU>&VL<6(%P#2aCwuX zC>RT)ZHJm+B{xkY9PZzdQG!a@tm4B_tK6sIw@;3plFVEI9{9;XcqT>ZjSuMb4ioJ2 z1KkGV7Ms)f_^w;q>C*$(M-KRuw#4PDWj&kM3T?CS8r??K_IdG%scj`$Pkk!MYu@LG zrFo*nv5vAlKwqFqgi%Z!HAgWP04`r;Gu_@(9`&XNQZXJ<04t|9M(NgK*G)Y2n2sJ# z?~lc2GXO`TEN~Wxvc;H!kV+p!IT0L1Ts?(weOUzy2@Nj;r^1<3lnhIfWtHCRTfz`qn#3@fU=9~0u^WRIh+(+H@>0nsf^v)qHCzeUB{ibP^tGjvUKt`>cJqO* zTN^AUy&mx}!aprx@Oi(HOfI&PmM! zme25G)Fz-HUZBwqKl~!WTAQfK=w-E!m%7e%d%&wwq;4S%aBkey+32)F@;$HJshH$z zHOXI+$@Q7{BEm0q@j>lraUSbk+-uXQd`0M42Ufk#b-s!d8Y~|; z2`LX)6Cf(_`IKGvnptVyhHdye>5hSnhKcsORyY1`dFG^H<@2Nx+OJv$3@K2LKWL>} zv;?~Z6NrP+6Zf{KXL>&kQ?WFe^-kUxJ8dC%ZG!*stieGa-{nVsDN&KJu%arct6YbU z@-c(it1GHG!0Kns3;n8_Q#Ly0lY5TEp>F9%?93G@mh3UHM?XrkmD%8nZ0Zr!f{>l% zC|K0~hO^YQa!5#8{=X!C{Y|#&Xe*ZRK!gxyMLe46C1PKuHW9E7D%84w_EE*wUWe7p z(Z>WlO357S1yWCqglPtl&5uGLKSpKIoI}tUr6A+4SR>h z$PlLTN#zfjRu}9aFqbhhAzeK)745x3|L%t)_WJq4{ahY%c*8K;xVd{w6V^jN`Xa*5 z)`LDvbblVL9s#>b+10o$;@IHT_~M$@HuI7pgkx9@(^6zzwNxb9-}r&cG74r_X1D^1 zMvy8l%K_$25?IgPjwO>*iAKj7m?PAq5SR)iABI5z$vozkQ_dy8?W-h1I4mAc zm38HyylC5^AOyQDh}9~DbT~1I1&h*RuV7UyPbfBs^lj$h@ci6mGV6o4UP>LM&CEOx z!7C}nA0YL$rZoLm-G98+`megr469$!`SX9Pd-UJx{$1($T?@YEWgBm&lEF}P10;Y% z(_1KX@hQj*y=S)lEi>%9o$0U{_1d1W^%;K+wEbh;eC-nwa44<<=<=RHXz;QD!KEhy{ zOrn2Zfemje+UU20mW>MOa}WP+FmH?V(PF-gQL<9Q!gv zSutPfE8U*3UsIp}Pax?4xF8d}$hBohf%FKm+&$U_r8T*u-0rjB$#vTHDdbNpr!v`* zwaZra+iYG&o(5DSQfw9OhI(97N#8&X-8j?#hZLD;`IctcrnIqEP`@3DNh4)Y5QxaL z-i^pJ1S?4@M}{=uT@~O?C?En`cqrO>alOIm6p){D^fKIll{b1A6SI)_ZUuR0tEmGm zQ(TTmi_X_qPSE(ZABZM0-pQxHSwy^*RL^OxzMNtoN>={l3%7pM)GN6&V}ZVRhh3~h zOrS48La&{(QSM>J7g)j8qvE+XCOo(@y*PhHJ)OACIB$>MXT}y)9&? zD5sy_va}u22=tStyn8I13oo}F?Tz}=KXOJPWZ+GRLZSoR&GFWlSA~{;p0hr?$8wAf zAmsM+{FMt!LK{9ZRN0T>l}HOal{hGZ$C&w1h=RzfR)Odom8($)B+1oXYt!AWD^#q< zpfUiwib+5C*V}%lyA#6-X607Fo#tBVHlmsvM&>Al(Z><2q583clckYNxoGBQLl zaaPvV&DBmp4nX9ld>|_7vLP%cEi3Q3IGP9Y2;OQpwOzXV{Fu#6P>OWw%3tP4#Fg&# ze`J6!KYoq;=Udf|`%|se+-7|4sBs%V^cFDFSrBTm*84;g>g^e-pb$~CB-A^kgN!U7 zxwa?XCERwa@Ta$qww;Jw;0NrGZQFO&*HsVsBpdC=WQI+{GNAK1#d6oj^ilo14KDB0 zVJe56X7Wm02Vk^G!!35NQOZOsQm3;%mDcsE>__;CwsV!DDliq<~CGI(tU+w)hG7Nf%>8RCWznQCWv1PXZQ$Ut!Ch;V2jf9qOLIuo#d{PfN zrh*lJEPwz|5UCO@0)J2j+eT7s9$u~8Q_&uw<<|+`B4Jf~9~CZ(pP`@6#-a^(pgw|# zZ0ZG3THxX0(7HBYsZ-w09#AnGU_-o4D0D4b-Io6!5 zvBS+lEt-|cyRpTE-hp=>?<9u1vj#OUFDOsN^tZcsb({1*fxaNB3}*x^T#Z!j6rOwB zJh=GLY?~=#x61*A!-&AdPKa;EwPVh)-5sfMSkR8ZF3qn|&VV9*AEVS7w@?vS@)R}| z+b_v!1wH7D8GANR>75}rPAcMYXSxI47Bva9&mOtKI@i3*?D;-edIb9+vNOa3PlF{| zwyZ91U(5W~t_37RQl(gd#*Gd2e% z3v@sKS$zHfMo{kUe)r$kZT}OKe0KgDP;|TAzvBUgy6mIV&A8yx>{?2hPg`Gk`GY&O zku}WUDxr87)D3Uc9oKc;YuP$2NC7$x`=He}gH;PXtI6L6{_;zG%Eb3P29Lyxr9H>L z)=hkwbw&n|{D|s#I)_qXH95Sn+dzV|%5P|@BcHdHPU|?H8q%aYptF~-locBA7=7TZ z=ES}QG(ymLeQWL)v;OrR%Q->4`y;=$o8R@O4sTj$1|x9GgCgHAgiB`cG*~Pzp1YmW zg1*&!rAm3a#+Pt~*>(6cyre|MDk;Q7i=o?NbmGzFsSN*|@^Zqv!^x=6?TaIJACMtB z;jyuRJE<7wG4I=7bqje&n>^kaJi^LbZ1qrqSbP_lQS^$a2}Wk9uxdGG6atlAE+YiB z|2y6urnzi}*8l+^Kvvm$@4||299MYl;A)@9P(H?Bs7o^|LNxB?xNcfyv~s|AY z%;z`WU)F{uevAKh!Rfc8*D`A*;>=?VDB9QLY3LTwe#lElkFU_opEUbM2)nK%^PR`u zB%+t9iW|}>PLC&6SIamK7^u5k7&nOYEzkW{+%eqydiIGc(_P})YIts$NaKh#c%AIA zWmlDx^yV$_Yfqz9;{IwZifGx^diRKFM9CNjMN|u_xXb8BQ`@D^aPR+(D_AmuRQ5qla*RfV60o+7#6Y z$sCg(`VGbK-V!j!#JSJPk2IoYX;fJnha^yuF)gbD9Q0-XHCK=YE82b5U_0b9e}O|e zOoP}E#mW@t%AI8lF-hH}<2nR-Z?9+ z*eR{%1`E@8uvpYB>h4fj83SlfGi%&taUn{SPM4TJ4+!@BCawc0}$HNbr6hl@)ob2^!xZOiqQ=3f@ z&3s;46mZ(e`t)u@o>LSA#O9Q50SG&8Gj27LNZ5=FhZhu|G4x+xVWHJ*cHuHC5d-|NZB2CXtwjO!5`{OfJ=)q-Pst5>a0 z`oyn)$&*ul!ESv!L;rXqnIP(IT*k$y1YMug>8wCYX@5EJC!xJi22~N_(`!;YpjT9R zd{=9ciOp~Q$&|>Odxw8iJj{it%J842Vb}c(BivskC%}sj-w-o*Vxd(pyjF+IydBOT3(mETbbG9iB5ZdPHc zeUcc$(PPq}VXEeOxdk!lY10l)|t%HmYc3Oko8b_OC z*Q8wTlq2uPjDo--tx?q+4m^FgZTOc_B4k{9PpmN;y{(E8dYFOB*ZW+{;q zo2NR3?>)&waH&U1GWn4E)s>BU%rH>yDS&|g3&3#wLE zPpcAo++yo7spD9C4~|g3#Yq8#3hPomY8MvDe}867@{=~dF3{8_dmss&b2;rVCQ&Jv z-l?h7wN{OGl^1b4#{Gfc6EpxDmaz@t~n$`4nfMH_e8y>#HZ+M;Le;yt3zU#}~s!oW# zvy5PPgIAhK^&0o$a~ifa8%I>V?w3vi z=?tB_7R=uc6A?qrYmJ(28RSr-jK*@@{{X7w5E-__dBW%s`#-g#P4EgrhR?UVmK-xc|u z1-HP#aAs!zC0=dJ%LCqbP8+F(xfZmQ(PUQH_9W1h^rOYwr*9#BEae0}`^;Fp?iDN* zQmU>BT2BNINv$tup}Gu$+-c6Ddm7@lSGy)Iu`8bK--Lh z?P_y8tY4wvZUGT$gKBfHrF&Z%q=fq=(wePUDz6-`c0iFS^3H^gBM%B3w*6d)OG`xhDUO(X>@0_KFnT|#VOUCUkruhi&d zmZ-k_$Fbihe*9a>{}ZBc>F$+}zk!8*_&qM^Fa|TS7#oQ=zD@5-mGe+&I)wfBM z!+ZU=GGmI$?eCC2icA;)s9`=@#1kSQ$y}(Ji^%K@iis-7Iq1$I>w%*;pZdu30K6QM zx>kpsGc||qCf_k581o8v*~&3I@fv@mSag_H-Izc;wFJ^u4g=Ne0I@Dd zcZH@l)A}Z+veT_PcgjWkxg9(v#St4BL3ZZQgQ2e4A48NW_RJU}U2GPmgrz;eo7Xsi ze#lyE&LD6jULbxfJ*n;mPz-^Bn!|g?yZO282v0V-ep$JmC0ai*}TxrXxIP`A+ zkbYyEW}D$jkfXq^NG-*?4|9ZeV|F>J_nc`5*pet@m%7js5h`+^wHeAw zLs_3)A_i|&FOw;{9mWAMyjlS`^>cah9wNnDXI@A>jWg?B?ECkX-=1Fjw}S}ycV^W8 z|9>@5{zFawQ;bxD)xfJjuP@coSPwmd@GdiSGQ>{{30yM6HUcHehIY&3aAp-~On_I} zVfs{xL5@JRga#Vb8uv{=X3t4j4Q$*=_cD4vr9j$OgOIf$<2_KLm^?rfXH|cz6jj_yP%gP7!@2 zaa#uUoe!s6u6pOY3#+~95hTQq`?_ts%Eq-|5q*>f+dfEgl6gaXkC|-l!0x3ZFYy#tu;@ed>e>cCZn+kF+Iat<;w0T~^^0;DYY(>7^EZ%9)<%{@2 z>3oj5dnlH|q@)M5Ta`oPnK2ESP7geU`s)~1HrAb_Nm^Igk2|mpk;8Sj;~BTO*i_`c z0CNplrTUE!LU80{BfSd7Z&GYEcp6gfg0Wnz1n|pVo=^>|{ zOP||=@wI`8P1kI`iy3=ZD%Pm`$lSqQC=v+Pl|8VH_;LYsp#2NPL(cx|Q_ z(GfiMGPF44}B%$#YePp%Vc zC2@8vS{Xv1;9Q|E2z$Rs(dUTtf>LR`x&STF93dt4C*S;QW+M3}- zY1j!R3^!#$UVm^MgigKvY^&rhoT6SL`KC$HvF?OB?M-8E%>-2$RMf*e*3&fR&XXCEfFZ7S9k94|4Ubd=b& zPhnU26L&xabI@UUqb-J21dc%m&V6UMbUjwja5q%gE`Qeylo3_%o3RV)K!Dz>AIo2O#vr_qM@x0MwTiPD;Q zB;@p1lH=sQppF^|*9(SIN zjWQ4Ijng==?)}AQNNO2Ma0v?lt)=h>E(qPN82+Pu~j1Lb#jQ_Uk|KDn8>D!T0MmItezrAYe`!_phjvx{>Tf~wtR(OyygT1iG zM1F19&FaRrazuYKsnqvrP_GC1z=COB>nUP1B6I$-ao?qiJJU*q9|AX;W{R1@;ES2LeBojJHZ`7D{G5rDJLm#BN!VpY=$xdqX8Y;mqK>G0jrfpT5uLJ@)1@mlGy*o z@{R}+bx0lpE;$Thc%>dZrYv>(ufV(I7HdPBJ^OavQjM6dxX}*=IldTR$$Hm(zIs&4 z;pKD5Y0de)v31q!gkBfOI#N!>z&2^dJX|{N+QNc2Bx-V#WuO#ShCGum>^hBxCYsaa zww3$AvTj%Xz<6G~)K;O`b@$HMh}lM$iU81Wt|)ve&5M){B#l5q1`fyabxnJ+k_1~g zLi|)LCXCXCI-eB;7KlYlzU9jiP08T}epbMMJDS{f7sl=e-+=0>UR^W>r;1Rq)^H3f zoZmuEeCc!48@0!SR){?g){0*qZgmZW%;*JG;IY~Q;(IdxepKL8l$a`mB7*xmmCY$J zJc*#di?%W1#tiA}d1fOq!}a)arYm#4PF;M}1^QK!P1XbOEg|-w+wXeBJ}o-$tEm5% zdoymWRPa*PO_%~(8ucSJsgmB0e~n}>3l1D#;j z&!3ttCH`4&d@A=3qOE}I-ix;^;_D`!WFDbclkPL(xuTE<;l#qC=a21f|Eu2w-wjZ| zS)8@`7qaH+Umt(}TSdQpV6W`Q>f1jg8!pvv;N{cDDq{Dc50$<-5ju6j#j@vq|@&gq8~kPA%ktrWC-v_6*uhPh>7tN4Qw>1zHiMSn^_SXCPlvDzIT9|^y2 zF6s7=hjZ?m3~K0KR})4r`n??ylDjg_s~R-*r0S{~6b%IhZjX@7zR4+hkb6&t{7O@< zwy(*z24QKoTmbRTP6M_y5n=hz*i3SCFWQ8^Sx|mn@!S2rHr)2Zj>28NJDlJO3`O@(g7TD>&_C8Qn2PI;^EwKYiUzTKvSrJ0bDgQQ!Exn_G@)#iwrxx5OY?WWan@mdX&vT0)% zDy$6HWCZT{a|o{B4o3k(3`_0kkvP+oSTtD_gMdv04^tYT&*Ad0Y|ylrBpc^ekLYYm?< z`$q&Aa1eS>So!x!pfe{KA&cDoe}sw8G{bZ<((k)1Fm z29b#y;TKwRpjR41b5%ftl#739l6%_VB5gL z**Tpaou@?^?in`?SOvj$ts-0X@<)TgiN9VOLV{T@DY!&~x;h7f8{jD1r(7sl z;xVbp-P;-WY+Lu~U#gg!GUaukgFnJvjb8Avc|CGSzMv$;ng4h5*m`BC*K3Qb-g}|M z?KSpw2&JYS=fxrEu!~NOij@p6{ID`bpSK_BSA^V|42o(5*2>3Wv&O?MEgAk^lWD#? z>8~<3EW>r&((HMRJ@cC2ws@^)%{e7c*PEs&D+mghr-AFTHIZ)a9u}Mv$IREhXTAKQ zC|AE}6KW6{X4cS?dM8{c6kVi-!%jvzbnDG18 zq3nuD0{y%njcNrLpC@Ga*9rbD)MvB;6KrIGlt^x+o3AlV_*3ouye%(hhG~g$K6OPd z9zQBBG!u69`swR0VQ#;k;#n20uC_z9coh)1)4JY4iDI=VsD969cVKSMMeI@@20}9X zuWs!i;4&UUaJ3s_fsDyWDh@$YF(}X%K?pGrl7L-@H+2B?n`OY|qb$aCcU0sWTd^{y zlW^IvJ)Rl$WfsyKAA3i~StWVH40S3FbmP2YhmX0}c#*K-r#*7P)QW_+r8fr|6!n1C zZ+kuqU*BK;(wJND*=bC&HH1xb`PzKr`o42bn&t=}VEK9;Nc&36Z?gC}Mycc@do2G4tc`k)MUWCq6#~a}8hw*Y_nLVD6m&$`X(KC(OF% z^7t6(=sz;amvYwxt(HwhwG;FUFg@KGphSXKP%Fi+Gs{6)f&R%{?O1PkK!$H_vjh~G z6H}^=()@{sH5cB9!@9Wc>H!IAOPd=!3dmfV`GSVIdoifo=F+%$4I@(aLQj1n$6}#SE2lB%z-=MSoX9?j$4t;FQoRmY@4qlxMDPPS7;zB z09H_XKC?h^C)9tKL)1pAM@YC;^#XecvOkwI%6m zsmOQ=6Y(OgGGk&zO)RD8St&b#k9=l460 z^T!^Keg53`e(e2vJ+IZfTf`-cQYEB;Owa+>U`^SJ30GQ(ZbLQ~kXvR0XPrCk?KxQ5jK#Ae#Xiy)4*rKfn6OIzio2h9AszZj~G77D@_8Ts0Fq)`fcpoJ}D> zI55GtnSy5ra}R9E#-|!)S?{NL;&AoDe^zLoh|r=vS_A0I$p$EO_|5Lhayi_v5=x^S z+Jg2-{!(G9ayaT7Ss(i{O*iMYjPvQnzpL3*yF&uDGi6i#7Ewin^&G|2O@53t4^RKv zP{Zq+Ncl0Es=;8UvXy_jsy>})*W7s|mOd}uUpTQ+5IpSHSm%V@v;3EMBz3T8i=ezx zIjG;59m9(ZqFysq3j+wN6}q0rR%Q6Z>ca6R;{1E6oA$E@ z@lwc~BcUD=X)Ss1PXTm>$zo#$8QWr@i+J4H`sgD8bxnonWL~6O6^Z}-=Y4UW z?7?6g4S7naC&XINLtg?C4&Lr{8{_yXxIy$keU$=7ek0!hHgNk@x#vd)t1p^^m;~_B`$^=n33#V z$&`n|lUj46+2GU4KXD_tRjY+!KXr&WsT|Tj9i$|A;feR;?IwlRvUKK=;u$K%lepSc zf=Rt=o?W-|2yP#&_0g1hJetw%O-Xe^ zOe~_3e01R1OWf1va~!WzxN_n5G&1aH*>#9``0_=BEe-YqW++G&>$YHWk3qeXgXX@Vio&J zf1+v18E3-BUfuB+NiSUqdl=|)-OfVNYGKX{Lb}4ZGJY<-fir8HJzt%?Absot5*zP+ zp85dB3sa5!obyDAa3Yi>MexBl45?cKo}d6a0d?awR@0ao3(Y5~h1E~1&~sJ&`M~!H zfh-a>5X>mXc=-7NPgnw9-IFGy|-6?rjTrSj|V9QLTH=G&8 zy+Z(;ax#C?nuf1-eo%*9!)57;rgKIHsQ|1MVp$wwElLLV+ksv>?WAV;r0jYKBF2G!#pU$)J(I z@Bu6RGePe{sZ&B7Nc(9TjM+2gh-R;>vuBuGsB;|G#iH^eLFnYOC(Z1tC-5{CEqfxNZ6>WYg_MT zL!_Z`H1GOXV>_1|RSA2hvkP|Z7>XgVvN+~;d#Kr?^nQoYT{u{Al`&7N(T9FZ?zw5v zU~}}yWG?X1OHtwoEILY2Wq}q9jSEI40BY4@3D=)C$0zi|k&;Jf{kGSRpad28w2l3M zn#!uxitu)Pis*NlNLF$5L+^eQ;hJ?V$piV5|6`K1Po$4fC)<*$LnT~RayCXVD zA?K&yu?GN`THE4?S_0*7*%7aKFo-w zGy0(n3MS>Mw-!-j-x~GYDdY~`Eib9F2Hx`?+;VsH#zjuGdLr>>dBJw_>ruN^==vB^ zO0Y{nu5zKw1L2$0q&NIqU#WLAix3t40mAmgy9K6%Qd(d3m%A?dy>yBl7_u}EK8r_m zW5DjDWq(JO??Xt%Di@^`v`Nd2iaV`hsUr1VFe4uLqm4VKe0ZL(1%E0jPWIb>VjCXK zrqM03@0ivkLgR?Hb?DygLruP+cKGJ2guvF)yaTr?EAal?pa4}|QPYNI@BVKjBr}~0 zG_MX@X=I>%VltBB6SYmlSlVb4Qc^ls9=O3ZI6K7*xX^^?j(cXOH?JG7zy14Htm?mn zMjBjfmwy58+JBxsiN6ip0&(t5xX6~wtRH-a;8Bjl+G_(RUTMF+`;342QRug)&b|`` z98zS}1JCfPQ73w$ekIi&F}-2q9y(p1b!U~BztE7K%GWY3P>)9{Q2GOo99rW!Hl{cZlf4#45SYWJ ziFTu4^A zI{>Am3}clerP+SBDCMPNp*Vg1oACpb$HVz<^{G7fTW_l}H>xil$|nuK>bJM{jKUgD7C!W;*PZygR4^gJ>gD(GcXeA~=qEaMeON zdk!9OU?Z6RDnHo$U=ls4x{Q&W@czv0m+etf#Z}V=AhC`OLAV2lAtMY z6fvqYg(d+{gsX1OzIp)z)G4Xr#l{>4onc0umFpzmQ&dq_@Kg6B8lWh5iv7X5hqGN! z!eaU;ZN>fBI2_HvYviASzy+Wcd&nSyL>LFFPTC#8EQikW{L$I?20PlBsJj1m1nqwu z2l(>0e=Cz*C2yom$jaTidbuJpapEpd3HcFb!1FH{G;dD8FFB7&5TEJb@UY5(OrUTW zLWFMDPG~@MJ0ZH@3T68XVXYNg(A(J+e)&e=rS!54@#?wvo7=k=Wj>}gU8(Urg%T~= z-+#T|RW-Ea;hreaLIht1;XoOE5?IsckqQNCqbsNc?F%0FE*rV)C|gDUn2jPGq?u}7 zo(*y%J1O&aCofoDEV)E5>-$Q8Cq00&n+_np|O^f)Xe+$>#B{Y`PH;u2;XIDfA{D=Er zX#-C8Bj1H~RlYZ~TQ*y|(K{`X+!ach7&wMSY)<<;Lr)$u>RKOLvOR)x`QS4}eWy;M zyV92tJcG_qO1&LB;^wF6^}HiJxy&5LDB-jKd&1g zHlMVIKM3s^j^HW`F42rC4jXVdODm)$aVZT4m6XerLHV<|?wJ2eP6UZXano6ahx!Y$$_f#*sT(c0K7~O!p$qtrX&)Za*G>HlnO^?0wXD5wlS6{kaSAiuABIZ}&yf zV+ywK@E?_5$8?tEP*a|l&wO?2RFG}M>h_qdu)dcP4Spr4>06)&_gEfY`&nJEo*1XL z*4uC)NV0#Le5$v}{gq;0e(G3S?_I1RIoVc^lJkN5*G<#yvp-gsf1T#KuFWWG+R2a) zilK*`As0_6-W&NC?wV}yO?>Qf63{aB*aCtgbGnUXkCK>g`R zF|easD1rPt;{tl>WVB|#QO$$9^wFHwnk8C7z^*$-dUql{d*REmSyzM%(`9I3@;EBt3?A&HQ+W-|7a=AFD4>-!zDdl6W`41{Z@6Zjby$($jOk~SA zYgpI6!qI>b9<8NRM#tM(?Bqoq>R$aOaWw#7J1XChk#D08cSi=EKCy0Ia=FUg&Fo_% zSK{@vhMSJ7_(6loM$X!`nLA-K4dsE-F2BeYMm)lsb@r*<*F$Cr4sEwJZ!0x!JZXZ(-a9#U!qV;4 znFi0`Cqo!^&Eof?$>DoHa*p>H8fstM(t58WYryA>UhvZI$eO7K9}jka|DC6E>(j{s zWtoLH4O0e}C-+*W{0<+2e!sBzt!VvZ>Pgq$LoeF-lYV*3#r4H|jgeT1$5I1nJH1y) zSC26l6%^M+E2=1y-W@H?R_H)q_+6&)T&>~y_i?Pf_3#Fdj3rO?|66wY^bTT?L7Ku|Evvq1MzlSTaDp;ov%k0rn|colv~^H z+vkrBnA`t)ZfvxD%82#er2XsUhilv2>KlJ8>r9MU@34uFe$pw=HihZrf!?#h`@Zby zzQFtHNF4fg;l?h0ba0^yv%CB&Yy}w^UZ8EYcH(Kk1-~EPRH{M1m}}nk1pMpNwZsG+ z&O_Z$@asFs6xTn8K-wK$ec%-$Ydu>-OFq*_F6@bj?IG(hx6{68+EZp@m`#sj`}fG* z*s`Nf@%hm*D&6n?to>~W_4!i8pbmt`;m7`c9IJLOT4gZuy_6jx#Hln)s7!GI*ol*Jh0<<%dfWHm0U^?^8KT*_;)ErDTQwvUcK>N zuRA`DG~P(Xc>Kso^39WtLyJd9y%Afypu0RZyQj$KzO!sIUw(3n;%CW=uX2`dM+x}r z9;2xR*Uk?{4)_F6CRZw$L+X@{lTO-XZ)uvh#bDLTjF0r7L`jk zMISgFnjxR^_!>rjZE2j0J&}%53`%_;Jp6-bcm0ztzu-qr=i16`^^Uriiw*OUpM;(r zkGIvjp*1noBky%{p zc5=vf2;X;sEO|8R-rB%BPbr1h=09zB z+pJuP{}mu1o3CT>wAb!&)q;~3xq<0>j`OvfbaU`<>!bEbakBfaNPjcC_l)+T{nM0* zeg5WR^+o412N!)hT2Hr+=FaauoAuq*A?)K(>{>bbc;Z!$*&ncTo-?nXMm-9wI25)o z;?!U7knc=Z;{|IQ%}47avTXT2pDo1QJ$K4qSy36Iy~W{gWHdQyw5*4}{H;7w9d;s# z5Cw_-Nu8mo=a?UefUm1YtawF|jKy#H3a#HXmU_(5-S>X;hPaM)_%B`ev7lPjc$G)@ z9+nLF7uoFE>+)N&Kv0mXN%(8+U0fj2=tfi6+kL$2%Lk4u+iY%IkGVa;q%s~{_#3h^ z6G~rOO=!6xAAmc?Q%aHlyFI>pzsjk%`5Q*^nwNW~)kne)$@%M9t3ZTN`W8(=T0TU@ zrZXXyk0oDx&9blW7(F{|dVKB>gwigDIm+F9`}y$sk4o~_OMV4L&QGdm%{qxMUZ|v+ z_&un2kaaQFATq}F?%>{2A?DfYtd-wr`MJ9tbrIFG6Lo@u>g0~?&X4Txi2kzvh?UiN zERF6J6_Wo}Yu{Ha;k0vU;6lK~^IAG*J(Ar$*fz`8xZwaL-GdnTb;|yiZqKADd0XAW z_RM+BuEfj-PQk-I6S_UO_q`1G{zZH-?p{g7ly>=k@mR%E9H6KdbH*&%dKFya3xFZ* zZLl3}qQ}!807oy7`zm0aRaSnlvpY2JHs{57=F~hubAH(Mid@DOCzg;+id1zoNU_20 zdQPLD^r5yXja$S|89(qFCgZ3fVbw%%{Kjf7qKam1Wu_{rU5Cf5;_Y=MF}A#O>OtF1 z^r@F2=CCy@fe9Hrr9Z3HsIRylkC=01{N5sQUXB!WP{M>*331OSdV78{QSPN<@L2UG zEzhx7ud*`45y7$>1V}{d1M%Gxf$vkECb7mYA zrK7<;pkWQ<;Q*7bhm-@vP~7pc4a!p4uO`F6mnVu(?N{4FF>Sm;WMqr5yq9T@+(FsU zCe4Zw^FtM{KZYSTmX1k_h=x>nail65-;8?2g3MW3IO2~_6YVb}M#h}t*XYU7bErr* zN?A~A76szbpl>LzMpW(TI|X;>RuL7!FE(yO+nOKykAKPZ?Q*GneSE1VvEf!?)?Mx>LAAc0a*(rq`<(N9j!vKTX0gc_ zd1<;Keg;6YGaaW*3^oJD1BxGxUv3?W8Vab3I#Y6(<75KE`%YvRZ%7zY6~8f`vIA>f zLV-&RpvCF>EWaFfGteoOW>%QNcWN3Rph={R<4G(dNy2l}RDs@~?|t&wh>`|EhJt{` z^E42Xm|WsZ^$)_K4$!c2R|9p@0+Yle;&~3KJbRn^N_S{8-?&1+>I;G-ZaO|-Nw@XSYbSt=(F-vfhEDh$-b!@p~8M^W%)!T)>t$zjZO1@l+ zIp`pj)(uXGE!o;KwFejwge@%?F&2Hg?tc#+No$ ztW9L;HbfB{tERpK6Y}eWjZ*2w%Jt1bM=G}`Q!2U|hs=Nd- z3dU}yXF(OA1R7iqj;gi_j9_xbj4Bl1HGVI;;13@PlhTB$nvROHRp>R%22AOy^DR-< z^s0dgPS0niHinBu#L|oYICDJ9KZ}pb1Ow#TjcNt-G!QWZdc>kbMIFKoo8%~rC>6U1 zUDizDDv=`r7=KJ`-j~9pT4T#c>@spHhoTB;DIEJ-E2QA!9C0b6e+8mj-u~8@Cg2ka z`)u?6>=`HITt=Dtc#JqIzwV;D6k*^Z%C#59s=ibD*5Scs*^=#~69>nW{+m^$AENxG zRds`%8JNNIi{-hcS!NI_f{m(WMjoNC#qew=60wQQ@>aD&9YwTNseuJVKuMwKmma3o zsfb-M7a`?Z(gq!xX-$UI2Bc~#oz^x#n6Cs;Gf5#p^+0o4P+8EH5EBe=jB>HG)@+<0 z#pO(E;n?tJY^j2OgGcDm|y#g&z){0!RZ>6f|@L3nhbu8kO zK2|xF-VKq5y(m|Xc@d<+oWh6*!THf+B1SmBdT%!67>=K-v}9+G+Ip_~_t4BI^MCuZ zuK(XB$#|z*d2L;SbXsoumpY%%o-Poj3Weq|OIZ;mCMYo&c5!6U!f{G8SSZ@W*%=s~F$)SQHqRc)^kMyiP@DdX%<>-Frpep7v}?N5 z1Q*a3B6fus>p~9{tCP2|fq0%zs5&}b^D81W4VQPesQ=@@K+%d zf!C<7!bc8-K517KP2Jc!_F=`(v_*e9o$oZka=6=Q7fV#jLTVLJtHWyw3~5r1lvYNJ zy9`al2vk=>cj$jaqs-HXx2Xc8m7Gt}XB;tR6rrh@{b^rZ9JaGNtM#99iELafHN?NY z7IF-E$|N5YB*o=}jHM40Hlo#qj)80eLOXlZmSBBsAL6qea8hT2&*ZoJ{|C@?8Gk^B<*g!i|5u{4cmee<1(> literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-18-Abstract Digital Art.jpg b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-18-Abstract Digital Art.jpg new file mode 100644 index 0000000000000000000000000000000000000000..952eb1d69e70c4d5ee6d15eccccdf9f2e919fafd GIT binary patch literal 87349 zcmbTdd0Z3M7d9T2fP_Vn2?Cfa92#YHqxK+>~sHu>kRkT)7u`ambT5Wyr1hwt=kKgC_zP6*8$;{k)&UwyrmV5K* z+Nax?Dbz?xBt}mUgV6*3VLsi!gkvTd85tX4Cm9`XLGwf!~ zoMC5g?=ahQj)T)&XM1~sKRnlqh#<(UIf20e-a(!|h&TEYJrffXGgC7gb8{PSM|(%_ z|LZ@W&R{GjVTTQp4D_5Z`j&bImU^GgVP=7OV)b-?FrWYE=^Gegjf^Lm;7q{{?Nc!N zdIkpih6Y%yp&_{Y2kjaXVQlPek;l}$&EwzYR0>+Cw-bM~C7_k3Uf1&wy-%GF=5T_3(P za`)c-(FcD#eDvbw_^a1%-oAVPCn}d7#$ZA$@b`pZ|09?&?9D^=#yh%L`OAon-(21k3+_lI*Ks z|Hw6rF*DEu#xtyBj8a{u08l?9f-h1RDlH!q68}*%L3&lgSZL(ZqNZ z+Ltj5rF$R|hDm}jMP;l(-Ahf-VLu-jK0%sS_<1Ds3k)BoyLf$>t?pkS3kHayJ79AU zNFS8}ox5@K- z1-MX;h`wk6!sxEZ@c=_*W%P%z45D=zKpPR6Q`(01?&A>mRRw_&e(2 zvkyj;PndeHbk~1RE9b)>=j;Om&D?;F29De>8YM_8binwm^1M9Mj2PAgY>0&r3pMc< z%uGxHKm?#*;tl{VN{9+*S!Y^+fqKY{3TNpcLDnxDBpKicG8A$DR>cG$?9kVOJ5YS; z4x_vjPm=-ZFG_Wb_Ftnw2k zQbn5cjPLCBZEQ8@rU{&lbzUarN16Hrb? z@%q1dn1IfNu_oMu#riD37bB@%AQmt=YeFnU)FwLfMe#n7{0X%h=(J}mnRpaMikDRQ z59A4-adL*BHu|@zbXxuD%M%bGNO(RJRBd2VdfDKUP94ua?mG5W9kY)9=hSr#`(5k; zFx{Z%QX#BUCrSXHwE~QX(gql+LI~#qotsU}{>wbjIZz3De1cqMfcuORpeo=etO-r_ z2q$F3Xjo!u2n6KUy#aM&jEG6nIl2Kl;|Wp$H&-Ie5F_4%yBnd?m~b}$LyeAiCV+(U zh7Pv6hyM+R&!5d;^oGw60&nMphs-?qVc{oC--jPQVN&`Ry;>Q^{GW5y3euvj;};0h z@&|=FButnMwa8bim`Dr0;)JFXbz(rB$_xNA0Z~RLtY-wJgL_TCpa$Ihh4w@`??Wy1 zZ+%u9>4fspsULk1-0)vROngPc7**^>Eg)0qj}~i~{|qsKr2l{zn86j$&Wu7Cs?vW` z4sc+D;BVgmZIr+H@OO8`kAGZitpD$+>=)sWCJ%s}oPR-q@V~J`A4LkRd$ClA+RE$8 z9fm0W0ou8Fg)aX{ZN}DN^YcSS9-p~IhAJ2EK2L?}PlxEw&nqgbv5s;X^4ZJ(M;qbM zWS))*WdJh3b!ob%(2*xb|F|reeWhL`-(1H`X8-kH5Y9pgii?OoS5dx;QBtg3ZKJ&Yg&PmM%!PLL%;fr?u?M$ivmTf}5-4$g$p97{ow;0Wu0I(bj zQNMQaK*KBjE6jB4H~|iVbOI8nSh_4qN8Voy0q5GGc@)YypGkfKN$7|Z2W*b<G#UU7*gd z@lW=@+yOGqi6{e4m>$s5XH)^kLgjP$sv0;~7a9P2==@&?2{LaYv`nB6jUp37L)GsF zs{QJ5Ks$N_x$Fy|Q8=+}(0Q^9u6)9*xcq+B_dh?GHs|jWp>WZP-*soUx-;5;paR1@ z36~j-#<|XVsAq)5J{(Cf-9hq1RUqQ-fjT9g{t){xMW{c*j203@k~0aa1we&2GYJir z=vD9pMTh|uMnc_(88V3v_f`P1<`DGET>>pvID(r1NKhku9#El+<7mnU{HRQ$ zu|rLfV*AA|s6BM}(K)d>%5D?i{W}r)LfQH#kU+XL5sjwUp=AL&lQ>Md+76pz34-n1eYHRlBE7VR9{>RB(z{|GKnF=8q{fUB1QfjxrO1V0#eYb1p^gLu<`^4 zo4ddh5KYmAbpRrME($-C>b~$0TNfn%39MzBL9&kDq(Xur8Xa&nCY3OyS`N}`OtoBp zpAe%kFHw2I*Q9yoKsIK`#NTG0KSQWRX@D`F<{W$SanZ=h)1NShmi_dnpBv$?pFUyk zZT@=hwpbf|C+N{`A(K3Qz?C4E=J`QjDEv8{dzz3uAndt%qusmz!4c}hBn*0M?(|3$ zv^r`s0T+oogfw=E3;LbxS;zp3H5eMs)0IsC1Lpv^Kp+IVFqERi24HA}sv3(Xyug`s zQE)QRsv^b*bdPIini&-h4nbIvxOKd%_gz-{o~T0Ozr| zn?=QX!Xc6{t_ZgZU~g(J46YDOghZV%Dv*a(gL&mzL0n)wgRDq+h?`fUQ^HK;bAx}^ zRC{h?K4FAUzvN2F74fsSj@Pn;93pw#6KdlYGaO{VG&~fN0DFa~ke`o9ynpH<7Z00> z7&cKW0ai|31sC38_~{!(3qpr#T{FtsjxPf-2m zNIp?|{wqz6jzsB7g|a>xktV8g8;K_b%$P^>be$j-a}@!4!pbA%rV~_)pp?Os=|W;m zw_K%{RV_zZC%5w^D=$W6BhzV$q0fMU#&TI)6etz@gkicdfP6A!f?EK5l9`}9jm=&3 zGWcs+wOJ3ujsH>P1uP^jxCq1t$%qa+9KsY4Kr-fWG$lF&D5L^#Q8dY6s*cA1{?d4W zd;z;C^qHWAp^JfL@B+@z8ZKofXw1Pl%H&SPY&{shrc;(K zRtRYfeLI+3kH%f#RAy9^9}U6)BvJQv0+AX0IPts#kXXMUq<5rDBc`HJi4t-~1DMmu z#j$~ldKj5BUuaasEfa2mCrQ$Xk?=bp=OqSd5kQx%kQjR-60ndyh$s$ z?`Ak9MyAqR1!w^{Z$Ybv4eHFW`2wg~vfdUQTQ93k1Q^p81W7ayKlKx?gF&q1%Noil z;3^C(Lbc8vbynFZ{eaHE2Lkk$C_}XXcb6Cq?#&_pz!6=NFRQFg$~f`5a5COcZl?YzXgi;T5N*oTtShyAoKBH<)YAE;?>iKp9! zw980c{>n93SXqj3Ty0*GCo*&1FN_C^i9^7vO`O~JuuCQFd1)F zHqx3hkhp3z@1`W7IPS7#!W#v*`h5{zc7w$&X**lJ3!nigFK-+T)MNEf3s8_F9vEG3 zb#PmuAep^x0O@<9u~bJ3!`dwFMXV9u=S zdg!mV3}8Xs1@{owDBvi7%97A4z+)OxZvuQRGAg#4YkzJ-J>E6A@kU6t<%kHkWtcb? zl8w%TTb!QDvK5BG?-1}N)M5Z*3yPg8(TN#!%_OKOfZ4EpFJeq%Zcz15CzRkSGp6$n z2c0EG4sU_iX)U^0?@+d722v-KXM(obZB1Q&)biqj41H^XvNl0%?^2}M(ZTSg<-}m6Jg5NG^~)yZs^-M~u0lUXbB@wzDTdro?G7q-Q&a!+c`h zDMFz9RvwX_E5`dk*4j5{kbJTDH)^1%#=~b8;_w;7?W&1w*Y;(9`8q_w9dvoI zb?gCOdYm3QA)ucg+W$l2q2!$zuEReqN95Mv);IMRl^r9w!6a}ZQaou6fC|CKs*b^q zhBGqvn6v2Ih5cZ2oH7j~aQb0o#z#fR%P%^2u=I?8(6L4jmwYffi*DICH1j zcGOz$>kxSWo&XO6H8E4VtcP(!CkWR``S+Ia*L{d9%nC8qC>{O0Ts+aYgU2J|SQ&Vj zgW4!^3WFs#jXoxviz9LChBS3ombD^EsVaNQN*O$s;KyK96;!uOs}qg2`(4o29D2Wf z^KrlUeWI0;K%v~+b-0Z!PQPGK-P%zccOO0hUlT+m0gW7~Ty8zf>x@r~`&ew8YEg#u zoa$}gw|lw)t(+hnN-nm5!?bTSdGV>AX@_v=Kok&SQ3gObz89!?5p)8TJ#QS>QEydsFntuz?_Qn?DUwAA{f59N zfO9^0;ZtojuO)1Ie{ z$Gm`x2&hjNq=@zhD`o!PmxTLtxNSvV%vfw`$HeRX5)YH7w%RBa zbD1d*9IjKRDlE0<)|m7iJR)*g{Ic<7<22%ulvgr0w_mHqXuKE0dD&A?(yo$tnBaBO zK0()WYnh<3BYT!xTX5$xJLK0X9Xo*p!+?zIGptR^$7JrTG1(1j_v#?_asBVzmB40gy=)A=&@@-bkA8_`0EGvEUni@OsufMyd+ zkgl(OL4sNgo31#d2u=k|bxZb)j`*0v6mHvCztJ~nf12jB8R&uES^o+1&18_aI7{o_ zx&QQzcJK?Wx!X4T{M!8qGwFfZM#-G^#n%DodMbyqf9I_?EL~>Y&f<2JyIJddg|f?nvdBeR4%7YLrw%ME$%tZ|JGAs zIu^|nrK>e@@Iir@|tEtZoUuAGoDx+m+sbfaaf$NuZBu)WJ4LvSXd5UL&@?w zdu4+mz-)L&$>}Q#FZi(!9SA)(T+$9c0&@Cxd4LD~$S%+7Y)inMH0Oqul9=Tno(V4R z2`SBe;J#8lGk%FB``kA}$-(=h`|JCg|E3rqzts&Te={I+R3=zHWIvhHPIhuw0VU4} z`?+c?W>T=*(xNI>Syx|&*&0Q)DruSG31aNyWLZGxsW{|^wnSR;rH1?3p18gIko~)A zGrP$wP{y&zG0T<^)zkIHScS-b!nD9R)fnaehxiT_EYG6Jy_K#F~Cn|CdYb;B&r3-HaYoe$n$&?{ifqnHsXKug)WxY#ygHf9U8oH zoyM#A&{I-JSt~_Iq-{R8Nf&`Kh&Q(bS3bd)ajUgx^s}UdQg)zqxIriAi4Z}T%0}Tx z0C74*HUOGjARYtXLOVEPz{&8JN$=bijT}A+8XB8EVgBg*qha$jn-_(@|FvNeN|pDU zU(D<R8C2)yccVZKUyrDtsVoy`!nh^_Jkr zdBPoMjD_S=+h=nMGR+rv&k>hgReIZ6CJ(JFkjG%14{VBe%4BZXmf3C^_JiAtm>iau zK|5#JQDXfAx&OuUp&X+rIdK`5267I!i@aX~7-c+aPyDJb3oR8IQe}}Lad?;Y?*hLM z@w?^;4U^Z%v#Ks=srnn07uX@M71Qh*jRR!Mm047)V7|ru9~4!Ib%L|+*9^E@kRCpO3+a4Jz3Xs2@oV7Q4Nraeq@ICPb>qsT3rNfGiU@ub2-p#PJL zzmUb}i{*B0_Z!LH>3VHZl>AV^`NK)HUHid|5gKAeWDexDfPSqJC`?8~9`kqSC5NCrkTWcozE_J1od0(F z@bjO~fBegK2K1~TZKmgmPnd(oy3n57zQuX?%#VgW92@OB{rXJ$vd!AErj4yO)$h~r zwe_cXVWFi8QfQ??Ipf$ZB`M(;)s)s0036G3`oci>4T@ehv9&YXK3;Aq-c9Nh#sI_p$c z>c)`nNvEe@uBMet5g#=f=Hpid$giuTVeemp;Zk_;XThchIlaU)9Mr!lwX0z{Pj5TO z?#jgx-;C5Q(chQ1z_Ko&F=^S5Y_$Fv-7#=dv)XT&A^W^hJvW*icDgWX7e0YB+O{nD z;3;O==9^a=i^7GgOZhjv;YZcJv?)(g7?Y8YS&=uh8S%ZAWb2=VlT-$d&SF|zb-&`! zc`A>VIUH{zV8$4T71Y@V2)?1$Cw4hz2nz40&0)XME`b~zdZ#1u4GkG`j756_zIP4Y zGLye-rGfCkEDtqnN8;=`Vjuctq%A3|ln?94rDve@N`&DBEpif4r(zgz#dCP5e~GZI zUEf(#2|b-m^%EbmA-RH*7K==RvJ=uCxy=q2?~9xt*SGtk-fACOppjy`eRgJ!& ztKp&so;u>pJdSLd?Q2Ef88cQ;y`(WZeP7X5*Id3_dE((U>CzsNW&GRmn90wmXK8zp z^!neQg=9q$k}Hw)noj2NQby-fOye#NYe*w#aWr>@`FsheGGoyG9)m9vlN_YQ8rEFu zWT;IWa^4-xD22qvJDrA+aMJTg(mZ?|rS8?~)YpyoP1`sR$w_+go6kNbe~ZlawRIUhd?R^U7L+i zbrVbkJ%W!+;rbc90BP&sb%R-8bt6Qs3cz`~p3|?j7U4EP$-5mm8#tJTJaW-5pD=eO zofkg+`LPN#MFK#0bq3w8rpJGu{Pq(joa!z8ic{TU@yUlI{=xj!C70Uj^~=vgs!HLOCvO{4 zv!|4kqwqeSQASc{eURCzqqMD5esN$!N|)74x0aoin$|-j&3z%hyhJBPR&3vSyzS;! zGTSE8N?cZOlfyM+nmqMprSQ9KW(}-};U@*|AQ`6DPgVy7AWs(Dq_MYgzJUYOdaEM> zeCOS2Bzb;hrrzA-H$QH<(louE3pegK#NXIhEYl_pZR%=fPPL94kz7XJ`r1B@W6tj& zc|;b;l$#3AsXZUjn53eumys^T+j^vh7>)aGA#5|UKV8iU3Ap%>d9?EC-dknjwr#18 z-uZf8AgUAOYU}BwGZ#n!1T{P1t+MKuDC9{rsqe1g7135)vwgMl+{ZYCv@T__8qT;xcdsSZZ!S|y0tRb0^|-w$bh9JxeRLaj%r<`Uk#rb z%I$=DZ5k?RNUzsx#S}p}9f4|Q!jl?CDj_`RCS)kdBYTCONH4b%G<=A(^)}PQOWU5X zM)J%-#||tFVA>=sAptbJ>)`xaxa@N1yC~4Yj{o(qjf?&+5;30_f5NEZ*MgE{-Y3j{ zIbNMO68z-YyIFo|8A!EL!}ZWz!kk(5ip0^fQ3YH&q>=7~6r@UV`Ia%)Tl_m4_De;% zOkrp=*)w=U!@Z#t&1mkZJ>@*7I)125KH*1+zX|rq_Dcubn6oq9%*}yh&COm}0O|cN9M7lX^ zawbS(_w|ru5nukYJ6e+s;6fZ&-RyjE? ze;j`;J7E{+z$Z+O659TP0y^Z7KQJ*sN+v-yUx?8k6mnL9b{Xh2bCSX8>kZJ|1A)io z@ClMlgT_w(-ZGYDtU_#5y_G#lwFuE(y2VEdGI^ZbXT`43vNSGn=)Sb~3 zK+jgG*^zu7N~quL->5U|rMZV){|w{T)FXxIFap?GWbWz#fm5h(jE>1-KgjH_p;dMK1>kFoxEYYjCQBB>b zsA@<79cD{4IiWHGo4mc0UntlZVoIlwgc1Y3b0o;-=FB`e8_a@u+)H@H(p1`U0fC@QKYc@>0)(;a8C4yz)U zAQvEUDq=-<%`YpeY=s|ELZzYNFn-r)nKm3cq^Km*T?J(pzs0b{DYnWHYO2RJY3-H5 z@T=u<2JC~>qD$4k)j%2UEa8>`zHL|1L9su#_F*2QlI7bt(%XDu^U#~kxZGMXuWeg; zvDPP&H)GrAo>h(Gkv!9>6BkUUj*9Gdt_`psaNU7j;5(Bas&Jp0F0y2%?^&zQt3)*X z6H0%EiG>;wXFduW!e3Z`+|;gTN3-AvHifR9-slmr))mA|f5w8zvsvl&YFnAXACP)z zf1}!pBK$dI11MvM{W+#^-{0bBbNrn`e(aMb>5?cPCbGSSsWz>XFZWu~on*mqQJFv24{v4RwhnLm4FM@|vS*c1(hf&Qt6j|vK}c?K zOJ*lASj(pKaYHh8R@}DLMx0DHZkM8}3HF`}6 z_g4BWmkiyGZA~y2uCIBMn{YD^roaJNN#K(WGMh z`Ws|mk=9VA_Ata@!oaJ*5)gwREmHy0O7l#?01_-%oU>_T0VL>aM)!(+LM+s_iqYZ$ z2el)hL9%4sG-RvBitZ6|KE13Rse`)^^U1%5=A@h3NYZw)GG$;%6l=5x5ixuQPn*qp z*W&pJ^BDB6&}E&?HXwuPefIasS-Spx_LXlo5b7210UZ_rt(#W+HAEx z*De(q+0fXpM7Gb$NmaSebn_m%IAuhf6^e+S&W5-`n%1VX-hKdnxZx-R;U0Lw?30)c1a#~Zx)8}$!rDAKhVk8MeUaw$m1TK!H^M9* z_riw4Hw}fz^EQho>AwpYU1Kb4g9X9T;zF7yHt7bfj2%*wXj`8_w%(I25ROK;@a2Nu zOYC(8aSOSJsU`Kk*~`7;SU6d6ky+K8nZMs%Xl~^`rl?X1$H@r=5q&s-!TbCZJWzbJ zY1#uheEodU_Co|ykS0hN@%{1MZF@=70+{YBHrqRtuG>R|p^1|RNEn7euKfEGWOL`4 z1UvU*oD4W}o6 zZBdW>HdNTCcU`WII$knlwuNuXj%G(Q48|!mln}-v`CB#VxwFJHZFVSvY3o86fcRbp z46e}|>QapE&YgzT$kY%9IJ=0|X~fSK)0jd}{XsEU^8u$|bu~nK+Z$b?xJPiHO2!H9wjXnL>_tk9r;UhX_+3)4Ti1so zCcqKNgW`0a+FL|f5clF$(~YJLd!4qbT=x6tzyvk3I$cPlgiMBh{m>>AKhV#UC1%1k zBUS21p6`;$Mtgj1T=%TScm&~gt&%J{W#DoiV*BsnnC9Kev+$s&iBogv*A=-NY#Xr9 z8UFfs%m~~m2X3KCZ8;ei5hv5y<&T8)NV3$6w^SKy()!Yh#QX=X-UZ~t<_PFqNMwC| zit<=Q8yc~p-d_6x&Dv(u^MvI^^h9=nxaM2LDs_@{U%yPHZv}xZMx%MTzX;&@L#1bM z#rnVT>>Lr$)~r$yb_O@P*En@@D!<2jB;6?&dub|4J;li4=(?oIv7A#9%?^N2!a$mk^< zDXAi`&STQ$AAIfx$WsGn$KRl0se|5Huy&mZhk?x+Hq2W>+JJstTT$^inT~OMNQo9_ z@^9)FYphgAyJKm9Vzf+58;H*Zb_Q{G-vzRXOl>5lJ5>Zb-#+-UH1Vf5lXZ=p-}6^2 z^!capcp&E|%g47rVYYJS!YxCZq24WouIC?j>%X?5(9;7O`x~-Wm(Hcs`_AP3rdm*S zAwW#O^-6Qaij19R!CHDv=ILet;$(^UQ1tX|t)90Ax?e3{kvDE4o{a2!$TPi3({?MkGA4Zz^XfifJZBx8`wKrS7UbmB2f`A>XT(R3 zlr>5c;X{r4vJ>u63wPJFEMqTMdAA&ib6LgnWjQ%$vshI!cXGSaRCp3->q3$PTqh9B z?qj`Li8r@FD`vQbwe2OCmL@c0LV55yhW;|P?O(8Jqs&BXJVx zzM+Yg$8s7c?O2Va+7a1qT48ZEsd!@zqj%Cs{@0L!g_WSg^;=gKiFr!_rDz z9FZDH*k~1WBJFBT-FH#cZR)O>LxvCfr|)T5^P2pz-#EBw<9^mjy~O3yBZx$Z$L2vMFwQplmIiyc~MoR!B(e zn#nhjJ40GBceJ9+F6RLMn$c~LA$In{G8|^>3%WY^s?=eClS<~_;jd$ID;<$4cUO2T z3-%e@Vggog9{jUj7}Xv7-u~;Fw0m8{xtw>4?wH@?sr`hNcUL-(Gwb+cv}bYEEr@2F zTqeCsPAhH=iSk`uL<+f!r<9(f@lPC%Nu@dEmu1#(fddud2(a9h zgSA=%oHtwuTuYwzkgg^sabPs*EX2FnbOy&E;@UE?E z>@+2osLsduTM`u)G*3|lZFc;ud`R+r#m(WD9TBt)qPDDd)%qq(yh8Jf|Cp?MiMA&2n&jNLP`) zhVYU}1Xb)I%cv?95&weO+3Z4O&ew({q%zAB7Q=-BO)$PsQ6^a3rV*JH7!_x;A(@Gq zSTe&c1T4PeR@q3%#t5r%RBkyC4VIErltX|BZ|*g#UnsJq(AL@8?OE$xt0&pL4b-32 zh?OvEash#KhmyT+p`0`k<|s##dMHo+JgyUp(Mu+Cu5fI>j1B8s+%9;gWFNf@-Udm-}lg;WfjLOla13|E7y9pV`qb^g>Ry$ z4C#l)Sw%KY-O5Kp*d0gdu6sdIyTIZ%a+20W<`L4rxjXXXmDNMHZwd`WC#J6up0YgC zGL-3dZuOfcFQv5_Gc?mYad9E}+h9WA`;eDe^o2v|;`5L&d~spKHTh~cr!xZ?kfXks zH=A3(LS-bJyK zkrTlh;*9`l?`D<%R!faiZ82HD-Bw_dHnP*MCv!CXfL`Nx!u{fVf5xxwL`qxx)M&?= z5hG=amnuN8Ca68NA=J4kU1Dt>3~i z7E!)E=f+HzQ}E43huv~bZnJ|fbT0np`TQSUNZzCf+klqu$BuS1UsNKs>5bi~kG^H~ z?%&WrXfM4zl|Dbxb2Z)Jq1tu5cH7a2B-w~8P*F}kVvSGO9DnS#-F|odp`Pgs$QLnYchVOw6Mi%aV0@XA*bI`O)q zf=D^3$2eJ6lwT1%>-OQxCG);4e!I=fr=RHE^EvM0FM>W)?X2fQMJ8|DC+_2goc_z4 zowMd#NKQO6Do-$Ke0DjfeD+%jecFZwy>dq6*)C-P+nf8Of3*6Ak=!$(OVe1aTH9MrTP_U^P%PIWpjr@0!5_XcUyCs!dwZ?_<{E0oh>&`&?RMr!_1$ukq)@9 zDQv5~U21sf@k5$(Y}29QlW`Ueh*O2#Wspm4P0`qEJsg2EP)Udu3A57Gf)J<&J7vTC z+*_}@!1fG6X|A2U=9#3IgecVp`mZ#vEDLCexzVHYSSP6}Fh@K6TAM=7X3V4RGlp!v z<6Y&6PSS7M3Hmjn7P6iAoc1`&o)PBeGnHQ~pcn_mEo1qIo2ZC8kiSl9spFyKl5Kcv zw{?)Z4Zk`luG|ysJfh(AKMwf4@{IIy3pqJZAgp544GmtnC4t^FGzLF>bh=k=S>_{b zwxuSxSN8&^u)p#W8`fX0smk~DCEnW^*wj1M?rO!tlkw@rE5r%&@3O;oi3l%AONd%q z!eMDmTrc5e(fs(5(?c0h(&qaaqOmH!VT|YVRVD*x3ju={*y^!Bo^vVLwbajYNp7Z>d&4I6#hIOwK z*CVe5>lQ9;w!U(?m3)NT)I21-F6TyP^@To8h=0 zPRG~M-b~m0Q=ZRru2=@D$+Oi?fs0pWAS>DNS00hF=q8GQb3~OtAMdb=WD60OZlyd< zOjxu>n71lHp6HMQ8VIFs*~M&IvT{)L8qpSj9Zu=Sc+MTYn5vDrP2+hTV!NCgGp~)3 z=WZ*Ct!T?fTJ4^O&|Ud-SGdG6oPCT9mNV1_yxF$NS7Jw+S+hh}nEgoiv{&!e_gA>D zjU2wLow9Lbt5xVl=7*kE@?^Z7w!2~VN?PmrQ?p-e+x|lB8|xdGEMHL5Hh4L?cS>da zE~@4Qdal5q$y40UqvMv8^0Px> z^8iGHT?DrC4tjgTCBpJ1A97o78C=L8%>$Ltu-La-kyk0aa`edi&HW%%vYIa>6cP4y z>9@4Zb<zue|ouR^jDZRl5D*emzcVGU#2=-R)jExeHS{N2Vd*SJ)rB8dBcu zfVPk9P55T&{J6PW*UGs=2fQou$V-g&juND*mxu2{2I>gS1G(p;1Pj&{*aWYaR@Zh@ zoO*%Uvp>DNp}3uR_qB`_zFyP(f>{Mvfw45VPvFNDpjC(6VLjLCNWrfn*o<3FO(4v_ ztzAV=-tbq`qIttR?XvN2?EKE1uHv`IhA6VPB z$&)QVusb#}<_VZnc2UWw*JQkr&Eu{A$tdl{$cLl5-mlVm#-i>QyR4V~`0Gxc z+Z^EhY&n*)^pBgjyZe=*EXF}jyG>u~#@M~rF*XhPqy^tY)zIpiV_A30>A;uC=Y@7t zpXevEKmP5e3>q*GCw20Y7Z*}1?!9jL_(6L$%-r4Eo=)^!puH-%y!a@qrQqtm8c&z| zdwmIe2CnA(d}_g^^R~*vLB|l*k2(A7e-3PsZ?0?4w9N;*dsndhkVIdHw@v-?>s~w7 zTD)pWQ_ik7UHsNNPet+2j0Ig?dTDFzsM6_<^|d!;7O&1aI5pMqgD&Tsm|GKk zoX;s=W|Vt$hLHJ;ZU)CX3;9!{9A{v|S>ElDp+-$-YPu9;Tt;iqvP9n%t5eI;j|a?N zoMvc8x4&XVTXzc%Q7xFEe1vtMT78f9R#}aWX&btGkg8n2u5oL6&HR4Ryqne|NBjF5 z=!>e~Hx^`vo@A3ZA~!XefwJFdvO>io&eB)C0$y8@m&&`A?&5u}mp@-44ySn{8f|v8 z7kKz3SWY@FlyH{7`|yp4VWf*hKBpw)0Y_{xR9N9+;U`m5Rm>)X@&l)stSS_~xbaKVfW>?c;NIFqF;_aC zb*rpzat>{?l-M}_3w-I~V{)|zS-UKDYGZPx+MP07eOjZib}r_XQcrCdXt^fk27Q{od z9#i&TKYUM#3JKR#ZnTVV<4H+pO%@(hy zdGsCo12Oczvf^Q5l(BeoZ-4!QOUkL?6yN4^w}za=?WrfMn(g-a{3OWtUHPgRlh(>2 zR?;>yNQrAlS`6yk{Gq<te$x2j595(^Yd47HVX{HK_=iUaM&Q$Yqn&p1 zmygdw&ceR2m*wNxA*X6mE`>ghsyW^?mGU^cQ<(bX;xbLTWSlq7-{$DYmlsAXm(3HW zF=qG4jil;8y8dB8reJwid|rc*BAQe(BJygwj@$-6X=1wes}1I~n@sOpC8-c-9%>JL zP&8fILpUF%2ySei%ITkVi_J*80WHoKC5h7aKxxDIZJ2gygCnYW{OXt{t%j^ufG2eG(6b(EFztFi|A2C36;9Xs3dJilQM9{gCssT9fDtTBq&oo&AU z#^OZptACc~#WdQKZD`54a(o|(zNwz+U$Porpn639$c-=l1<9db_^xe4{kA?x;gEi_D}TlrCE1PKW`Mm zpa1r5@aAdHOZDy>obvgl(L{~kf46ZrRz*H4ltH_jnZt*>=LjcNkT)e;^w#@WKN6(3 zC$AEmEVp@x3u;-vZ@BKLUcAWzB=fE0n7KK&@|I$8)~%z4vbdd#TAc&gT}?+;oLaCd z`0?E2v{PWM{f`qqZ&seBM=rr9UY>JmW+v~1MI>jc7H~lQC}XzDEV$;d{h88srwfvo z-`S=+?{9d2zEnW_%}PLg5D@>swLO3CBz%-_)t$Q!mSQ#~h6V{tvljPXzecSq+SDF4 zb}fGQ&b0}=H?^#YHs7@Ih3s3k9sKFbt~4ikhaw+D!#!=q#R9($_w9qR)Q!iYjD?v| z)ST0kUbV6V_wF9uTwZqDMc&0P9A4I3B}1+!ExWRHmqU7u{YGD& z8P4#Ao(a~*|eOvXuS1-Ju|iE$a1fZ4Y(B!+B_Lg_11SwkLo}zCg3th zvGY{VuPbE3Eq;%#-MiwFU3cmm5?5VRX?{fiY*z4Ny5rX3l}-2iY@qVKP2xdhqmZ@x z$m~w7V~vICz$V5Z?MH8qwT%7>Bf*kt(HY7NY_QtuySP=+4LL&%v4&>xH)`<7W3L-W zYK=#$%Ij-yW(utys;2phA0NwiSEeT>C;07OIc=0l64u8bOHaSY9ujA2wzRztCqGOT zhrwi(1B5L~6NB*I4Z;)nvC;NcCKWMM4{ z4bcU{R3;@CzGg?OzUeQ>e3>(}2dqE5Ppw;`(bc&(}1Dg?o9OD|h1lS5|Z zqyU4k^1Rqy0j2AYMMhxNaqH=?tBv!zyLV0d?GxtLf{BdL4WtcEQhwaJv4}KJuyZ}~ z{3dSr{GKD89M!3#4-=PQQsS4m-oBz3mDe84xNvrZ;&@TmWX?CD6+a8eRckKQgeExE zt0}9_T)5U~(vWs!URX$Pt)0X?W^w^lh3wR?osOS+YQYcIj?#>w%*H&|JxEd=i#~Ls zB#a~SwVnkBWXDeKQzw|mn5!4Ek~7!L$MzMySiQ>RKyUoz;9b6>j7x)^DLIMX%f4H6 z^UR@3*K&GVW{zgMbPYUxpA)}s=hZfjQBZw)*%F^0o+Q9E$JbB2>UeCpX=$0JY%*26 z>BA-$_mkGIT7GWZc;tKI0?wKHt)^2)>T+koRz`9-ubiXEqAoQ`p5h=FowKiw)&W z>9J>T)r*aB>!O`H;?rZVosE~ifP>aHRh%z~f|)=0IPZY()93q+HoWYX_SCjqkke-@ zh(GAAsb98G)Z*eD(3@0$QgAJYXK<>P_m+ihJge$Ucuh(l}{Ud5iRNMjJxtS7g0SxRLavX&FAi_vm~} zN28;ejCoE+d9NMaukDFFyVUiE?~ZTV9vN^z|H9HexUACey4ou*<>NnvAXeBdJRZ z7)O>roXoMb;Mblq9jNv46+B{iR9g)tn{6D}B?N_f= zq|VyP*zlO%H;0gKq|#m!^%5K3|3lPUxHZAPf7pbGv~(j4qa;R*Q0d%Y#Au}?Mo2f0 z(lAE%2+0iw1L;%Ok*{G9i5XhIdm2`@>>0AZ)N z^Tt`P8berQ&Dmy8aX^cVnbBjH{#4iQrnU}KsW(wdn9BB1x-j&P%St3|6haT>Q_9V-+Y z;gF?p2*!_BZF$I@f(1?8gGv5g7mZyS5!ef0VTD%xbrDwLt&N;sWF=uhu1MJV1Ne+> zJeW>2pN&mPqcRYVOw?q2gdOsfG&SCZU`BY?Reitc8Ib5VeqLL-WQfqg=>htozC4(% zs(axJ&&@CjMgD0pJ0v3G<5}X}FjfDz}YSN4z)zvRaqqFBazIXH3U*Vbz^sw2G~- zz>p%S!_)xf=*lL0GgO!vBHx{y4MQX|W%8od3NU38ql!{}2VZP_>|7q43GYd z)_RcriE0hkf)9B(`1h4jZU?owtW(m)Zl;;-M&y8PwYg8hlstsaMsM-s(dpUCkEs~M zd)crC(?O_A3C?!*LbYfgjZi-`^+Y{Jo_tMdT%LMafNe^aAw=#l-g;dC zG1Jvf{Xr7mxj7AQNmE{EK{L1?u5u_mS)4suLyeH=h1TC~kE>en<2e0F#pHF(bHoMS z>!vw&aS|FO5y9&Q{&iE-zm*)VKNs*)a=LO#)p51+&H>QHJTd_LL06ZLAWFmPP@Ylm zTf4q6+UlEV-_$YQe7WEazU$-s4Vm47EaU003Tx+LHjQzJ;DXwfX(K;MGpn=AwHi9i zkI0}EOY-tZ5<~5rm6%1xY_Mn6wJ{qO_A9j(GDgVV*rn?G(S5pP+(*O00qVjr&oan( z{kAIowRPOz;YDOAQ>&(QQJx$Urj`zuCi&a5qgxfgGjErSozbNV*C(yWMNUS;9J~zE zxX~4;W(9?8V*&0PdvG9%WBF)pxuSKz*MhzNw5g; zq5GN#kb}Sq(Zfhg!(3CC0}(&UDBtk|69h_g2&p>|hD^}><0E(ppFo?jXf+UFYh2AK zx7L_6Z~{n>`mvKjVcuo@3EgZ_QMynRg*7}BZQgB@j1UN*K^4XfEX~9b3|Ga+8gU3ti7Ne%Jkggr z@s%fpD?O}=7)Vi>U@aOwbWL80OflEu%HHvmTlr4Cz(NbV)||D{DU9E-7&UEqTRS&) zY3AZTEmW&J1vU|TDz}vTyrkY>j*+sIbIZjECysQt{#3!Ir3 zJu(|t7Zc5F)rEbRiZJwfKkfW>#ZgVdg^a=^h*nzV1#IY)dzMmmHYiFH?5^_}-;CI{ z4Dw?D(`(3z=l8VHPng$^!XY9SVz+aJFV(OjY&i0E2U(__x?Zi|`VC0Ig0WL9snMvs zs}@_00@uRInk*wT(5%<37EJ~$rvTElE}`Q>CN~&i)I-tMUomag$N+?!YFS=QGuu0i z1MyfD($Dw-_`Y&R!5~lWy*#YHr#8PtxY-sp(y5>iSDN_bW>D2kHAWdWnU`29)?=IH zhbZD4=_rJWVxu~1hr6Ko{4up*6^4VNg?b$wK&jp|d^tZBpZ}?1;G98jANd&WR2)=? z*NQ-R1@Cnj&Fdg4+N(U~29Ti%}mx;cYQ4QNS9OS`k3vbPX-2#1`mXo-hVa9Wh}lblQOCyR|Y zCghtn&v}(bM4NavCPJ|S^f-~kK8>%Y@~N6?`P^7xyTt?KZuK)q$5K%RWbT9!AkJdY zXXv_00jJ&w3~y1(Un@|@J#A`9W5NrUep3RYQ<&;LE_Y=>)1P@z)2diTREN5PC@ENR zK_)+BCtOHB5jkfg3cg?c!H#d*0bOHO#WOP4~B zjiP=A^3Ia<^VS9kyQ6~}4sb>Lhb9bfWNT{0^y`@VVJF`1f^}^&<7ywxa--kgxIZET zjQ@5BP%5Jm&KgF!v|dVColHM+^SLfptDF7)G%=s$U2Bad7Kz%2q&_Wd)d^K9*Ee*+ zj^lq^%QB0pQ;59rbaz)|A2Xsny{4nm`AYH9sKaZK>)Q~nyiRrZ?Cfdc6`Gk_nd3Xy zb`W++FV04nCi6Tj0y9qbNnFY|6yq~&HWc8#h}x(~0_=y?vamiHG=TPrHbJyF!^)6S zVw%I+0uJWQ*VAzUVQcloNAF?HWz(Q@SE5~>caLLv)RZ=tRwnPt;sSNVWD1{ zgJ*4hsog$yth<_6?JkyXKuOzrrh&b1?InhDC$#RXBkD~|-Ph#VEk9LLA>MimZ@KF= z_#t{1qWvJ0fQk52Q&yR+--KaOX7m4r6bKQ0<1cxVrJ`L=ulEQUeGaG{+uI$UExSw2 zgV1*ODPhq|%K&5uaJ#FD>5HEwy^I3l0TZJh#N$Z(fjfayG}WLd$Z+-kPxQJ?pwUcY z%?nVEN=2tc-~hw`LC1&?^HHIlVJe%CFhztgY7%I+RyUNCI3u8z82cG(0hiL$+MbVn z7+Mo+7Hh^g)ZCH@t0<5@Xz4J|d=FrUP@uwYG2$Fm&A%4vL`Q*aovfM**x#_w?5!$m zcqVNjhV=85eM0`zLL&ACde>1+Thwzjp97Vj&8TJ~vu2^)8Iy}M>Rkaz&ANS64ee)` z>RhpFtLjW#iTT7jxlMultjxxsq@mN-Jk={3xQyv$8kWht&D9HKq(#Ki)TTyb!M@97 zX(MAY#TF>Q#bIDSg>_u8*bWOC?$TM)hJAzhAx`8yALg@@vQnA1QA5&FjLGcB2`OnRF8lGt+%v1~2Bk)Sx0oLs| z%DXyT47+a{B>0D7 zM?wcwmu{cnJa83Equ{q9zO}6L+p}hL|vKb6eS!&od2%eIn{@2jMHhTC~DKQ32Ld4F(Ml^+G6_K+KtQwsf%Gg`KnMz2l{Nx{H2nW0g&o zZ^b)?pFajagpqQe2p<(Fkw_+{kwSxiNOBlDIf`l#r$aFA^8iF(JPf|6;BpW>8sh- z!XOcSNEp81;Wi)e-?Lw7XgPc7rL$+SND>`ixO-cavPJ*P@#+c1lp&!@tvztdAtzWQ zZ+yVa9#zEZwGYB@id!B&P`}hjDwVG)(q-=F280d=O;xPv^{3-2I=}2e^pcytV#ySTJ=zP2kVzF0?vGEjc+o?w~Jqii{E? z*G*5*AQa9@$-;-wQuvUQbiXsslIt2*!V1oYc3UlQxn-Lt7wxC6uKbp`7g~CVelqkg z;YPN99T&(MC3d`z{O8M%$y=}3KcbB)43{Ri&6r5pf4uHem^8yFKs2&J|iJmzHC5oOimeQZ_v?4 zBJk%qon>i&WR(ODIvauvSsmkpHYG$GNKwAQ%q$)g3qxmCmftZx{FNqP6#@j_!%+XR zieZsouMAXHYW`^_(kyQ4yB4ULR@;;@1{in94G+CAD2}xj2i(V`*`em-0Q2^ctab zuhG_x9N(FBf9dI|i{SKchJ|Bu=QG2(0_B^WeuYj2P`&0T1zV+CYO8WoHE7jNsZ=(T zhiGcPTM4&Rq3w>Xqclu^IAZ=rn^L)`yV)~uXc-TPbaBPj5;x0O<6lrDtJE^=>wdMc z8vdwnwWhlPcbFiz8UE#NF(wZTM|kgDnYtWL{@a@4L#k?V6{EZ?$5PhC#@{G~enX9- zabL^k!?|ceD$=%x?=$=H`}PH$sjK1{bP~yHG?2s=ex!y5x1aEj8IOD+KFA} z>^qv(jqk%><1u-%&(uyuOuvQ(zD(|vKlqQx!$}$}0C{$5eU$o7ok*pd&0juO?l7$E zX~XJWlih7`z*2$!-kksEn3%;P8r+2bAlGd7B<4q|lVgzxQ_*@E)@c3Y%Ta1rkaZ6T zJ5xLTI_aRrzW5QUn0duDrD9h@+?K=ewG_x1+PBa^<)(yHFZn@Qt>YF*n5Ze18{|YV z{;ALvk;G?WqGGeBp8fT506nLnI)DW0bjhoZ6eqo7&(x$osFIUwkeh;L>6fc;>+uuj z$Jy+K&Pv=zoz?epSMD&Of6Ypxj$?zT_P&WVg<-Pq1sDdHE3i70J#;9xvXaMR`(vJnDtFxLB;(70Caa+$4l%gP$3$+AP#PX#s@ zON7XV2xgmiB7;DogHK@x-V+V8+@p3A7)gPPaiZChXu}0;uqxC z6NaE)$jO;Lsf;h*W&vJa*KeG~2Ba6s;8Aa}}bmJ-3AY zGFCuj#eVpaO5k3+U$Qx1){1YFh2KGPP65 z>ee7mM;EmznX05W0;gMP4ma?1R-eaPVH)}U&xvt=o{Z^Z*y%qR9?=BUy<`*%3#k2P z`v$$fVSMkSD0)SIh7O2KWft-defX^O@NDu5NX zy#~*@_dM|D?DOZLGyfUd+b4t%Su451x85K>TAThTSK)~d z`2308yr9UDwC&XoV?#sk^#Ykw)VI1+pWCy_iOiz0r!w5b@Ew1WvvUo4DefyxpBPHZZl~K$Vwzbz^GpQg^O43{%vlLj?>J@I@<=fgpgycU zPVWQ%oqKf&gIpOF&rMq{?s%iChW${5LHEZFy8a`Qdu5O-r3RLAR!lHDdfiI(*v7ac zm)rB9apnLfQKYs~$L)J)r2`y7PGc1H@@6;4ZO@xP8PPFW`IVFWM%6X_8mW6<@$Xjs zt%R7Y{;76a(rlupdShYY7Wz-k+nxIJAu~f%o_kHDiCx5e+})-1lK|T5iakU|VCtK@ zYVqDL;(*6$@D0`pI$@NnDSF4o(-9F}=6GEjkMH@W12_k|y@A&k0>)*dH9({&X90`$ zTxHre`k?xKCU2Hi^LEZ`y?Hm}LKc&=N&CQ*t0$-Ii&mmO4}RWuI4O^g$E@?@h zh=NGIkB)~{O&u!j=#4a;ynRl6M2r5A0CQ8Vv#O@w)9|{@Ug2}Az)SP;d&pR5hD(u$ zeDyU)<)vJ>VV!HsOEte}3t(M*nkk(@<&%}2^&ZsR{8^=^Gr+Hh8}P5$Q`gG-J(=)Z z4+$)jzql=A!SNTd4%=GWR6)tbdWe9$(~kDDnkqN8o%qU5NRFMkr3SFmxrY6r+kt$e zzw7&C$25tigQh%?gR!azbAZ9p7XIADm(8ub4l-aWH9i>Vb!-!~gtR$0P}R`|)o#}AKaX4g_g2`1H_&;B&xJ|>_s|QZWSpKzn zcWiei*LwTjlY09C{d~jD&iE!-fR7HMRKeMjQ6^4b_G?-k7Jq zv+3E6(2hu!5+sEOWUqB4@sgNI1+c3wwOT4>+KiLsi|S5V;@sQ2QsFWtH@~m4F1}#LyYr@n5gUzvoWc>#8=I=mHQLuTn;T1|>^kX9 z%Jv~9e^>jyIV4ijnTS`uplDD2JQ#U%>Tm5orn|{J>Tc6Ek{vAU=_qxf@ovf+{vXjp zc@JE5n`l#gT?gB__1O3vYQG{c@lg%@DLezK)r1ZuXnDJXj&*D0i4#EiB%QQ5|9 z7mN$;M@<<9cm^GaePUTAEoZA$BZo*@`7}z&OJuL#7z{ zd$GkbH#lxv!&u`ylI%o;Ho5o$<$&HkSG4|MdW|p{$VV9(LF)^f)!#UwzGAfycvIPD z>`DBKyIu!#$kKeqXT#9b^ z?CXYCNNpC2gS>l*A}q!S|CEZqi)P|ZBrl%TjNhHG>Uv+QSduSOze=6hd1@Ic`KJIs zesdGJE8gC8>GS-H#MySPNzb!U$n_r(ZcZrA<74}zr~s%uQlmr#zpwVCUnrQoJq`GUi2>&5nw zYabGHg4$H9383y_J4dq~KEmB5^_c*F;}ENluUMYaJ@bz*t0#2elItK>Iu6Oc=ZENC4LeYn%fc^kzDUEn4BQH5lmv>+%_tE`i0E7bvg~g0nEWGR~vg(0g{{};TB@IpPGtF~gCqL@cZd!P+^flAb%9tOb{yq(uF;_k}y4q>LM0x(f_^~nh>*uY@ z_bdQbF~avNLAA|#HDe$g6GG$uK<%y<;te9w?Eq^eXiFJdOC9OVHD5j`X+*?@@-F_C z-BtSaTn$iIhp+Kj)G*GfT9P{QJgAV+4G(w%R{$1C(pg(sbT$y%xuIr2=)FczttntY z&haJabQrWkZ!PXWxP>vy(4F+NA*{Y_@~qXF>5cGVJ@1dJt))Sj$(Sknf@Jia)=eGd zZZO>H%Q-$Jb-Ysb<5tcf{clgJA(sp?E{jt&f%ANsVYa)1?Kw~J=BQ@KPNCg%lP9mmVK{P}YT>r&l`45vy*e3HEC z;334W?!Y2*&9W|Ia*lT|P3)amWz>Ve0mv_X$LLG{ea>O4&isk8ED-jwz|@m3PsOsn zDa|(K&-Z~{#wk^r4+Ri)8~QrR>7K_nbOPL=SYZX7yY2<7p>Z#*oW5$@RF<-t7xcx1 z!EwQCQ^j_FebG}!{log?7H%>#lM3~oi){V(!`kwJtBoPGVz#>XT)g|ohs~|C!kbV-vJ>|N0>0ecrFVQAz3(cRR>tR!k zFq3*E)aiSfnQjBuYqGm9O1rcVhM&mVZms^Q*K8VL-9I*)ikjkbX6c((lUGUoDyOi^ zP8NzeQN~SI1{@mvj($T9$GIu5Z;T4?`eHI7mf|NG$|Z*ve3Rygyi?ZOIOB3&`S+!j z%NOssk4uq$2Ri^lBH)kMXq!KQ_u{xf5$A>0~(@kyu??!)Js$Xia!s0KTBvs69 zRrU(RQi~7U2s2oy?1}$InZ@P|Hwdfg)Gix*Ti1~%P4DHke^sh|m;DQ!k|@a3mBPYN zIp~WVu=P9i-Kjm1(+L8(fa~MO$F?m@Q@N_Sm>pJrn$NF3dZM)J$LH6xf2`lsS#|L# zL^&R}+3fvB0zou!TjL!>Az9Ix{7hW-9FFh!65J#o`&_7wEjxSIWi!S~cv$E8gXh-@ ztMvQXt@;wAlOF|m_0mX5!WZ%deyNze64ric7q6BEN7@b|NsOymjU}2df{5OJ<_567 z3SYi&dG*A{x@)J^IMdUYdhWtBRxI?8+fzZ_VHL!EcYXCJgy4$VHNHmO7@sduy6v5n0xoVsV)6An57pqb3yd}N;GOco@q3HTKhJ%Ay zoMb^qN7|9Hvdj_Z+mWK1vEQ@W*mG6-QZe+lAc^iq{S&W1W*LQ;RRf=^pO1`8ucVeX zu4kepUW|5(x~u;w5@ve6Y(-FwliL?Ha%|jxz@a9kO&pOfD<+=RI23I1gHAbU(6u^d z&6pY6@PM#X>VJFDegtVG)d|{j3&cKu@|lDw~a&%k4f|6|v|==P#U& zfAfI(rIRGS*qKsB!#+8Hohh-}>v%~~cC1ZG%O zuh*B&%3NzD_~0d>C9xb`hMMnlY=YTj7e!y#)Hj&#)HgtAw7jPFP@IBFoQVc;Ts)qd zM5%42R)}`E8x>sZ?m1nj3(Ph68hHKIaRlDr`q_Wo4a6`2xqIE=N?z2~BLXJT4Fb$*VO8%l&M3LGr=&)h3NAtsJ zz?LToVA$Uy&M7z6@4z3f)g^A?y!NKLwk{R{-@oo14th_~m)ur$74F)pNWS+U_#Kw;@&-|Dw=b%0?B4S1Cw7yL?BrDTT_(m4+MD%O* zNH;YgH~m*dQ#?zx|9-Ls-`P1<4%hZzZpv)o6&^+egXhs8a?f3m6zmY1AOlO7j!gN3 z*7b!v!<8W`H9zyXqN~!J_f{Es=V0@@rc!Y#Zc2c73g^9W4(MURr~#B+mIyGvekaQI z9Vj?@arTKdq^Ye_`$StG4L^O!C@!Ur&9EN#M@~F}PHFHzao3Q^M$WhJjTAxX%JMD2 ze!=r({UV6GD`?h^yip)q-N-PE%lUvOmwM?w7Yi4Cro~x5riX{Ipm&_xFz5k^U_sg6ONY`$Q;8i1^T%$w@LrX2$4y$KI6 zND97CIt0~v1OhQ2WqcRL^B%Zb=c-@hlI`WsE7W)K5@*f)k58R+?57?QGX)e|X0Cej z(|~XqHTU)oaYm{d=@O2&^#;;f&P)+NW5JOZc0=RLtzVew-43{1%x%o%pxptid*VHPa+;hX+KXlZBEzCTO0^0Y;I72`2aZey#rlD zayt`Jua@bcg>se59q0W^?~tgM^{~`a`F!PL@77M#T4ASNZio!u zgJWBqH`@I)xgmb&>`euR=pXs1jh;;d`xgSB-8PHNc(&C|*4ERS`p2dh&YTRZqVIEU zE#D(hCT|&M!CsiF0A7I0pgnD$EVh&SB+e8@6EA}Uoa;dnG znddNhvNyH^`wF}1MQW+)uc*JhUm0YGpqwoN>v>bz>9;?haNqBy&<|kGREg)4y$`|| zGaal7@%5Gu%eaUdl$}}%L=;Mt%9}c!7*`v?@_^YbUEB>RVIlBkD7L`-b4ia`c`23E zK5|GXn2qMx=$H0hu#WU3uSf*^Qi5P%F%Cf8lS&*6TUL2^VZX(z4!p+GR4605^f+2 zNEw`Me~sSly0Su1R?MEZ8IFB!56Dn)>QpS~t^RZ3S(JQ1K0V58PED_EQ+zg0m3Si~ z!LH9{Q%K};*R!a$PGgjv&*#aMTFps6MX#a9&P>65YIMFDYyL9b#~9ARCA1L-0LFPc zBEk`-7$%#M`XSMQ@GjU>i~v=&e=u> z*D772a`d88Jl0z3i8G>8Xl#&>4w8in39kCyG9!+}g2=0-okN&0&TwJ&y11}JQ@GfE z=-^|U0A3Nn#ba+uB%+*;_q?vwIOh;~7ne;l-)8KF-SC*n5qVW@NlkvgP|7L*|L*xk zbtj;~`YaHnQl%|2X-`YW@G|K!ov`qRX zCW4zja_FlbAuS^wJAfr|h)ds?~}OqpLiA}mI9zzju>sf*UL2_s^7BfT|`3+8u32iRpH$Qn~*6B6Z- z^zH^n>Td8=ca@!70O+4;p<|t#b6x$dh<(^SDf%3dF;W(7oYM$L=E7!p=dB^(uz`3# ze0eC-_gX(XFXTsLU#dPA^`JaDL;SH?1eE$2h_1Ub93OzC1X zJxo1%sVEStP`rx*1@SF;|Eg9;7i>!nT~ArzTN{<`>U~u##`j+|?QwmD%u9V-YgeJ% zq6m`7wMmm)nar9Lu3+VIZEJ%Py|Y^8Oggt3**{qDCl(0;*GxJUYLXIoV~d@<*qZx^GPN1LGJ!YqOYBwMWh-v$*m%n^Tk1n!CB2xv7?CO zT&YBLjQux0D%dn?*`yUKi;&gFMZphv4Uk;d*mOVNZfFAY$OXDq(4`LI2S(0iDZt`& zMfmGZI9yn0x~j;Mx6LFf3WZlmA4z{7*2mgJmFRbcsm_nxGi#OtV0(^y+}oGD(QQMM zxIg0brCX;~CU|0>N43{jy!|Lym9MtAbbh`Fk|WX7R#dgIX`{Ex#EO3Qwz7aP5f0uw zv0t{?uxDz`zvT_CZIz$@J02u?TIOP$889yY0-~(2py;nRxjL-^i(25{`7-KM4^3k> zHZxuA^=OQBOz{fh)|}|D|JD3Vtk?*;DtcD5TE`Td4X z4^vwk`ANLauwAX`DsXWXr&l*D>@w)?BYdYEH05dkV)#a{9v7`bYa4oLj^kscT(1uO z*zgQmyy)-Nem@9CRZM=Yf0UACmYc6(cbWN94MN{S@;k8jf#7fK-gR%)G*BqP z*oXB!0==tT5eiNb9l+zgQkjEhGP(`QVDT^&7Ea@^hckZrL!DzpyYe(xU5OM9+q=|t z;S@&Z&3?m@smyOz$7aG6UYOS{LkRVLLDyD zkAh#ktpNwTX4Jz{yef>r>qmHaM}$+h+huS?!Pj72`KmcxUb=!Try}_h<4MaU70p5p z_IT27QDWyvmVnK;uD4!pd;uGahD9$sJfN(1C9 zM$7zF_6InV%p*t}D`yV}>FdzlHZXT zDySBHGpy)9Q`*X|7c}aOvV4>+xTElA&&92#X0LPW&U{UcOgSQN-H*5z?(yJ4+;xv{xZO&Pv`{y zh?_woLQ{aXJpzcyF46fWi;vkBf@fNSIXH$0uxSK9@tJRw9%f_1IHV8_a^9+8V6C>M zyd?JCpe*R{S)}&&LfpSb|I2)A zf^=ykyukLkk+y9zO<3!L_Dcr_sgto#c6~uu5TNy15hJ)GUnIaxdw{&SkEY1XhCUS` za%~FdKo=_8u|ASB)tCC3)@D^nyMm65UCJsM%w&;^H8$(?TdGvg%&U!TuJl#4u*)yI zlpxMtDoh|>i&^DD+3zyh!JL#y#7HPn@91#yvrjc4w*Z4d7NdpaDN7f?y1bz2Lopjp z#3qH&(&S$a7w?_?1d!+8hV=^ZHuOzM;)E z#%|Q6WSGH%s;E zWVE0ug5#o}YbioEHnlD4(d4~sJ!k6G%B;^6{yA|*ApVjTU!0NyI}vg*rB5D4&!3@m z3^o(mVO2{D)@%Pp>mk6aDDTm4s$LV>+wM24ncN-DkWAr)&-ZyI&!&M%%azrz@8UAa zl&r23(C4NTKxC;@F+`43S%qXDiREM*Xj)9pX2qozh`=%wF2p>lIn$?x9V?8=yoE~y zTgur%=mWR!5o7aYw&iGMzC?KqKuM~;#B?|{ZEUE^^%w+htz?+{9L-Mn#{VlC9$ zc=f>4vVx%$&KsV#lIo$9fTF_4nqS4=%iRM$MCnWG^R+I1+w*Firj2%*>| zbZycJDE*4NauMQGD-`(gFA4r0`TpNE{_?)yk&5>T2Prik8P++4IZyqpjU(KD);bLH zZ!W-m3M94;|H#7hau}l*iv2Wn-2@4gy{XKEl=3g=q|$7hn~3;HAZx8>07X~YHW@eZ zDO91{Q!QpKY-rM${n$F>v7=C?467~jjesg5xpk1ymbH^onEmhvpo^9*aB=-wP44qe zy64SO#7DIGzTZ|iU6Kt={8oWa$SC#rZNP7fp-|5(uW$?0tj+pqM1Ol1y23&r`NP2}fPy_)!wpC6T8nR12=4fGE@FWV_c zo&2!6Blf%hXSiuxUPC+?w{K0~{y_<}2WwWu7(`q}rbhhX2g~nValZ(qCOS3t9ti|1 zJv{4FQr%mO(zFco_O=k*1UgIX`HQvXwFJ0)WO5_taKP2Tu#_B+y4@x)k3-PXeN6kq)EWy)Q*1F}^N`bh34EW2oMA-IU z=?fS^q=z-;@QLZl$~M^=vtLN(p@Zu@#a>Kmp(2kKB)AT z+MQmPDA)zb%eUnaZ>D4=$wHnQEP8<#ZzZiHGVBHV79yCw$)RQzC50nag|s%(hNGYV zo_C0w0R?>1eA^pA3HCIzc0o`%X+foCqu_#z^`mURelp{Btu89Aa*_r~M!E1O53;r< zZ`P4Q$)?(7)(sGf%+3q=3Eckyr4LsGj5W1>IvF7HPjwKzPEX-X0f>}o9jD=mzOTFS zte=^y!_>h4KcX@mx8U*7%+0lk|L-1%w5Qthq*t4kuEvv_miJq|8`ic)+*bf}rxpz^ z;C*K$_S)+{wII%+_$sRG#ah8p5%()K1i#&G^gOfE<_<$bAu-faUi?AedPlsSen)5h z%IC1EwaQB_%92+#03B#I%g#%Wl@aN#oyuoiY2r4&20N6ax{Vb#9cWzg9t77^i#V0b z3!c})-JP|_L*&hyrpe&yWpKsdgnj-eHn;M{PSRbQi^lv#TYDPN+_MWqnyRewY&4+g zjg!h^63x^sfP{z^1iKH|n@KYlj3KD9XL z^04(ghS-vHM~-3Dh0$jGBRWb|>6b2x`Gt{voy-(gc<^^^_gT5fg`HC>Upc5t?dcLa z+5FqJow2!Po*CbsAJ~+*PLan7ioh~8^Q9BRiGTxqZG&?6Y(KdbsDgI8ll!vXl-^r82Uq8^#$ZIm)C&mQlb(lv*rh>CEFk3@O| z2hr#lB7MPb4^$AL*bt;+T|_aPe?76PMPdOnWgnn<1XKM?-3&IJ_S{>)m>ywnR|=Q+ z-h{jG${R9vYt?!Kdv)U9=gd=Qj3=8kF7?dZgLBxoB0Aiw-d@7qrUTEFZ^nHjNs^5p zNs5kd?1mN@=2lJ_kaIr6X4$Ih@>esc)mICi+t815jZe9^DIWMP2Re)Ie_1#j(NVVO zsgoB84AK;JGo+iZD0f7x$pnL!hNnOyI6IO_v*pA{<~(QS)7|NO^4 zif7M|>Y-=(@;`iVo3_Fh$+iaryx;3_m+aUy))ZZ#fOeHp(u)|l0*TqT1#5}PX^&*# zC6#bXtq7)ytr|aS+$ht3M9SK1=SF#um{N9YkE6u)19Y-j6i%}Sq~`gdJ5SsjI6k`B z5YfNI5aoXup|Fyj^JThzK4o^FY{eFxT{G?3i4U?8 zZej25ZzZf#Wqc>^u3zeQngJPM&WpDFPDL$ z#x^M9)#{#gox#uN=*QXk-_{WCjp&9Sc>@E)(oXjENDWE=`3idPveOF|kCv3?)KyVk zejY_8o7$Rs`St_F&Feh72)C?XNklt6U=e-W8$AnkSEWXy_Aj&k` zyvfC;Y9!O&B11(4nEDe#n72aHKfZAM5S0GM>e$-kNAQltgb1^Z9LVW-R9_9^$LXqG zCo|r{eraQ=Og7}5{uouYWG8_GwecgU|cXiE%CCHR$ml+t+ zB$XpYd9I;U0~Y}{E#jsgT=)g^Jj&DMt=+}hM$zAUZ%ZU+6E4-`Ej0D9hipKQ@z{E~ z8w*$7(|^TF4>%FyuD2=WH_Khj(xO3^$ol>w*4!!*Mm;{TjDGWI*etZsOh%i#*OX=j zyCm0_mxa%>$W72xG!TFaSm{)vv5N19(w%|J;;%@lb!2U7Ui_=D->Q4XiQ#G86el3G zhiTRYgtSuHVVj#l-VG{m9w1Aav0KsR{#DuPq*}vn%CNllHTPD!+oJ-D=YGX*8EPl> zAH0jSuk>A0LcqO`!Yjk`J3@Mdoh5qwX#<9BI~Rjez|7uhrt25i)#tXl8t6aE3)Pi; zT?-5XycW~?KCw)mtYnuiCgV+zS<7;)!hDh|EK^a~e-I4^8|>gc+Z2u*liz>)K1`5a zoD^9|rI=|m9@h$nu#4Yj6#rvB1(I+{GmBEuQTQFLYX-LU)c0+_wazJ)y2yR*Ep@Q& zsn0{7Oq{5xk^5az<5P~6ck1wfo*j`gtI|%xi+fpIF>2z-|s$G z8NClTe{i$2+8OX96tshK*y;ef#Mp96&J;3I=qZs)h-K>oKWq0pVWq2RbFI01hc^z{ z*@x}A7r2EfsE6m-_5v}1gE@=%UhRmAwMigy!p^e zIY!52UY6UIrjsDY6FO?s#vjJt%{W>s)R1@lu5B`Xw;utE=kN_xL#n@Cj=cP&CQ8Uwon1 zGTvq;REjSxfNyP((wOa1VCe1iY4iBaJ)Q{2n z!nv25be6WeM@EI@25P^nU~E?-6GY?Q=9nueR%`_bp7sc&j$jN+{p-J53IgFMpD?l()zzxEFth*#6^C%$ao z(=ovI)hAb#?B5-Bi<@ynD7?TeO~epQ;_;u0YA^J<<}nW;vOlKebmuPQBlgj z?IY=c^HOWG;g)pHpNjp8@8Tyn$zDbp`-ZgA;wr}PB7RlH$QpL_*X%PGgWUF-rr;hn z#Wvb|0YJz00K{SNtFhzOV|??H2K*=4bi#m??hFz!LZbf&j?6o^eJ}h~;B?u^JDP|< ziFqm-^w`>Ku)BaKFoH1umf65+v(+t6D9+K(C;A{chbIn^IOATxqy#z|G!E%6kY=A~yo{p40NDZu)PeWQqTeb)58FxqduH8Sgd@|LQ zzJ$IJE6&ND|9=3CKy$xCS5VikwFrbVx5%Rcv!TFX&(02OPfycfnj-Gqvs7+XR%S@? zfkv!U?5`MgJ4-(q+gj*m%tklk6~tUjXSjFixwA#RubBS;HA^&W2;_Y!w%HV@)1@zS zc67cEyR)~MmgEecgwf*r+bJbcW=4OD707wdpOoNKo=lFriO8oJI}q@T+WS`ZE$NUg z=Aj&NAC-;`dR^CrG_}&$tY9!q?yo9ITHZkDa7Apti!}*te1jp|j8lt^wKK)x>r|SQ z9g&xGf*IWtqLM1^o1{JRmnREZZy{n>oReF+Ua*$X`D}54Q>D%7`$bJC?oiY$O^mx3 z@l-=Q9jXZ@(zG>u{{Xj42pj=cd^>Lx*XU#eiaBoVfz)uP%`MEv)2_A$c$hD(M62bt z4^dvLsKame76d38Ys@bI`%CN@8RDvRxfzxd7xjzidVj!)clhGJ^*FX6ZDk^V* z7A7ASUEDF{wD({9i68h;Ty@ju)b+5MKW0zE~s%PC%hFbJoM8uN}*; zq)K--IINlC*u-N5S6^*^=C*eJ>(;5?>Cs9Wcz$EtR*ou5L!Oo*P?e@~>ZcuP20^Qj zXD&%Aw0c!%g~7;EQS91Bqt6$p_~w!Z4;x^L(M@`H!cO^7Stl!iN9RuYSYs8Ni87@L z>diZcg7^L7(AIRY&1`a{kSYMr6J!&`T!#5rZI+rBBc|@P8y)c~0}g6<>_yr$=aW?0 zBRjLkIHk3^lGaiE-ubIuVKrs$?lxFuwtcQY8c#4gaBE^)yIa&$A9=X0^Go<)Yj>n9 z%_9k1WD(Mb1-dhqB|%ziQ_JQUkL5g6NL2<;TD`7mlW97P)^Z%|JXTXF@6w^@&Y1Ge zT{1^~?V4zjgJ(3U6O=3z;(|b26~UxbZL!+;SH^9isoVI~?N7xk;;8_*V5*;Xyypb) z1xT_qL}2YTDOl&JhElzmdsBr!w0x^`k?m5(XRwt+4W##~);CdG%jU`pusl`CZeLlu zLObK*)vPSzf(yqv_pWZ^Rhn%_&z?f=+>&Z7Pfa#28CTvoQ7mABySK0Dh#v`K=aj8_rk3)@{o#4;_Z zVx(|ap*2rH@rCY%F|~k~8RZyvr_U!H4p?F7#;-Kh(dhmyw6wJHN+O42T*c#C#~3oG z+C_6(zl-m+TfZ{slvvLQ1xI_SHm9a}@&4>Z)DxW6l;-&?_c`&HRZ_Z~bcxW5ODDGl zRa7@aQ^lp(?7&4Q`PVJs$GyDs00rX51dat(pH$Q|%|b*D*v|mgv8d9y&-FREx^bD? zOKD?kY)h)B$29v}OD48iFuNP2aS5u=s6{j1Ox{Wi;eC5l)>_2(Hgek}WJ0P>m_LTJ zRc=Ges`9UUTk3VtNvD)?5mEslij6L8?9xk~2GhsZxSbHqE#!eWN}v)#o`$pT?=D_; zj#n|pSoP_ena$fnqB&ajuiia7PPqj zYc~G?@QnEiKY9~uE;&>X#>Zy=3|~msik(2Q=*`$ z#jS30)|zF_)x6Uqg$lVDJhd&9qe_BH!6JFUg%O_hcf)pHYQHi{`=E?&a5{nXtn15o z?{)c7!9fM^s@!MZr*pqa)RZN$3A>Y4)X?3qUF-&N#w%;YchSXb6^*x&=%53Rr25rO zR?_7gK1^XE19rpKvF`1-S<*v?Yy>?=<3iAJqZv3$WV#(Lfn#kZgLgYhs~|slheKF4 z7IR(d$f#A$?1B$&^?U6T&!L+;D{`^N^Gru`RPei ze@YEvtr*{_pLaC(qB0>>jRqHB{Ast**@)u%I}-bH`Rr;iaBPU zR#%jvTyWh#3KIp?SuU<2WLa|C213Yq6*_AQ%mkK^Fe98(?+zGs!DCbMu_w8m?f9d##IbKBIJ=OiIk4bF@_f7+Z~#v0NXU*01S0 zoSJ;ED3Grh+v`4r)K9wDYl{T$=<-*EYj{ur>ucJPbH`$5ZDErv$S}Kx4wI=yxxaeR@r;_q=q69hH z>r^i!`&GeYK#05rrs7pW7zZAeACWJ0VV>r)ZtU7IPn)H^MoHHB^tai8ue{(sT8Nm&NEW$JKYIUrsCj_OYy$DWi5nDV#;KhLrITX^IS|)p-ND@lRE1^2}drG_NS2fj4}ms*Ac8TGGyc&R`gyd zmfpbHuOk=-(z(}jEQUrsPHHtL4y$uZ+CM_YzMnnBO(KJ|R1ztKZ9E$4G_6M3D>g>J zz^-ooMV{^12g*+sOuBU^DcPbp;+k`^Fj}0}%>l%N2D+~a$84y^KsO&+sp69g&OEGs z6)VhziEg*ENW@FoVvM9m0;>|mw;oS=zouCWxxQ|dcT=&B-U3u~;;Wd|)Ql&~)X0Y5 z#MvEaxVbZ9H55@avjtJtrD;i|%8>5c+MU^RaE#UWGZO`>=$bJe+Nw~YBNQYK{D5dr zpyg(-a9|5mu|%bEY7-Gt&lsRcRA5rmsB6%4vE!vgB4bZ1Saqo*UBD9}#I!Q%x z#U9J>O6oa0Lva$Hm{__00LSB^vAzK@J~_Sw@%@`8{*8)?;KWa!cRTR3)jD=ZC2?$; zhMnfe%!93BBeFKuJdP^Po2j*tAdIVas4o{2dB9_LII2wZX~o6zOJR~Jcc(#e*3xil zNTi4_C+0OgS~bPxw3kreqIIg~v83e`tv!vaJD4ujQyF9HRCTQyNI!iSE8i8936=-o zkjq*hn&wC7SsD}?D!nwG`@l&C64<3J2K&*fV8nnPW=n!+;Cp0$lMGTX6^ zKton-*y@Ey&0DFZZK%s0*!N*8q_?QPa;PW>O!FRJPXXa5roitkt@eg~aCi0*u=xZdg9kE?3dX!i2tP^8-`^~R|6QX$H4Xps_Ky1%=;DF3~QS9J0{#!(akH(ZJr4A6?h9) zvS%b24SP3Z*Yi?2%!vMQ-#w|VVaUJg1OYL#kL zZvF(0aw+DBZQJtqsqW2_+m9s~X6|uXTGoto!5S^X%v+KNYQ=kLkZv(B9=NQT+Zigp z{7FuaewM=P$YU4xs*JnA1(vnnT)$fpYG(b{_iAe^O)lcXS#Iw3%rmtqJww`5g7?mR zN=|aMD&0r`*b1=jDYC|)&z9K)(r=Ll(6~Oen^rn47pbMD$t~5$GfFU={Klody@uxO zZHO0Nlo44C4Dm4Y35*`~BFPoJh~h%b4{Amdq_514v*A5nc)rxI6B!w-_JSx`R(8hU zy^ht?TlnSIP4ZF_c8q5=%i2pF-lgYVgXM4B`#TCzZsruHUEkVliqm;bIcRgUJ!)Ik zbyO#D1bn-=tqobUyPKq4Ha0sD|>Bv(;6FWpUZM%v=qX>DhAgeg0ejEZ%X_O`M)l^#w$LZ=jS zT*jsyNYT5LolKiz$gF-qdey(NTxt5860(V8m*xt`1M#c~uHFlev<4`^<#%&erMs5& zEyczDWCs8qswt^0iJdt|ZHC|7&o%s>X9S!SUe%?kO=fgEacbKn{xr`ft4k)A_Cm0` zwaFvqJ7Xr_{9@E6rLtWrfif zh6k=|J4PN<7gMdnM=WWTR;JlcH2043z-}0f9y(R2*$L22q@;CG>H2;Ci9Os_%^D_7 zcXh>c{!Pp}OUWFJu(NU#y<3ufJXkfPV&QIc_W}86V|;YD;E=I|o<=YfR*q3Igj_nA z_PgP>4GeO_aLt|I;4r3198y`9o>rFQcgqOp@inWcPp5>mi!w;DW6YO!56}wEyG=Uo z_$(fDQ5E?~oStZJLHm}o?mj85EpDyU-fc!;=f29jt7j-2Q;FIQF^Ydk?rESyo{qvo^-!7PB_Q0uTnl4({AkGhFdg9BZRk^ zj1qBNUV){p<^G>+ZdP!9ZNvUqSFK*?3q633?M}N;u6;!u)?_$Jr^C?j3;89nntKaJ z{lt=E3VA&7S+_cK*j??sR!^643)h~Ru7dh2du?{+2^;~OA@94f{#Bo>*xlPmH{FTa zVetFN>wq-Y~RI92FRt!axJab*;7Gf8;P z2#ju;lI_Mi%CRkOX4(iJE)R5?PZ2I*v3zo z-Fj9$i42js`$GAOP6G^NW}6a;rGMgLI_8~ir?tvT$to#`ZVFFYJcgo{rly@|A<`zE zdt_^MXGU-QzaM|CPngtK8tBPP0CQu@m0OorT*TZqT-^!imT6?FEv z#pFgpepCao^)%cWF{^5BQi)FMP`K3YP@Dtzd8ut;wTA4i()pwPlU95~r5!SO#6xM2 zZp~({pK`Otj)XM4ue~Rsag-$mBb)~ul=8^0v!PxM zT+?it>vPE5FF7@&C-z9G1$9#P@+Q(Op}N0`ZcfaWAT^EQvSh}5nfk!79tQD=Pb=`kzls*%cmUawPScRyIH&VQM z)V8X!-5`EJJl9RC+1baYgBpOqQf%IJl{qEc=B=(i&o1%Si@T<{kg+*66fl|Q5vNjd zTQ|NPlGfBnI^!mzn$E>eaE!Iw%Y{=JG|Pz>qSR{4I2@X>{hxAE%w+j#+_w33BDaUl z?}6z~xC(;;m>k4N98_%O%>aQ5vBHu)YDQhOu)3pjQcCK2Pz}4?47a?R-#%LrNj%cd z*^rWH@Sx8$>6Lm4Voh^i!>sO_)=j|iQ)(KMT*tI@?^IO@>rOF&O2jxvo13XFn<~X} zuHB7w65c}g%y$pFTxn7d6)1U1fJwoti%5#p6GCXGL2b9auE27PmbO|W5Ca9t)ysgeA}3Ow9QLWYw36P zuA9Rbkc)k+z#5sP<)Et0tt>y8pQw0TLjiQp0=Z^Vj#T!qPSr0U7SHp8#d*x-Ug4N` zJW_G6^7{2EL*F4%%lOpD@XMIjAPzCj|D zHy{*4L;8P6q?R` z^FQ7Pt#c8Qa;NU~Cls*W$!j;619CgiWh^>#y* z?Ins62v#QrsoJ)OCZFZ7b#WlzDdbYQqME&@xmFmOaMQD@f!eg~ua2c9y_Lj)AqNGI zTEJ37n@Pa*&q~moSw{j09K792dQotx1=O>t+O4I`k=r{=i~K^XTIw<^v8>KoAVFY# zD#fkk&7GWp!TF)a4!%(|yKZ6`KJ}Ah zHO`fM6furOZ~Nw}Szq00jL!t7HzNcJ*U)XPwMp(9P??m0xW0H4UgUE77_{|9p1I?V zX7f4Kk<2j)-cts^)qCM^M z4|=g@c5NMjA1SKFIZ;FBhUtS&P(b;ae;T5t9WM4SZHG|0Rf`b=)r}+CJ*DErTY_`B zJt_xxE3g*hb03&sbf|B2d)bv?o@pdS=&X36*D-XtrJ-QyPdZCKof^ic zjon(Px?udIk`F^#I&83;W*17Oe}|rH*1R*|-3wB(ic*5$x_sC_bk(@UM?+Z3SF`AN z_LZrPHfEKr)>a^aznX9QMz4H!Al~Qhfq*Nr_>bVYygf8h+9qL>rh6LVjkMZ&M>`cI z#{o|wvP~7pz9jWzx}KKy4uW~oMkAF6aayG$cFKy;Z)4oB#Vof^JC-8^aCzj_{X*=e zgjTI#6ge5*2h3?ZFpo5C8E#TryOx9O(1WxLkOx|`VSPVHoUl2~D-B8#Z7s{1TUxc$erx$-ci%m$(Hr8e?Vgw{;DsEh z*s~K}XpzHfIvIhMKjKVdHDNE`?GqbIMFE!qSZ8RWP>#aYYK@IP=I2qoyt|Uv-AKwy z5O^F`WEyi`>I)B=5#7G?4XU+LE6qY8@y9Dh8}9+dOqWYxBF?u>B*UB!TC1&*pCprS zxp!Q(xw4sT<(?^m4m`0{bm-z(=Mxy$<=~a(ujzWkvdcBpu*Wk2$(Ne5;mfElwRzs? zL=nbuj)H`*v6Sl0Fn8!?=??(9jtOE`SzHe=@<%cL_oc+~!&lQnxZ#B)y5X58v{L7BCY7&n#$Xj)ZleKE3#_7URY9~+O zYedz2LbFQyQJj7L_3VBj(ro0=Bv|1>V_%m7ye{7E=vi91Q4*tl<>3Bxy{mZ3QZd16 zYjDjT-hEHjlTPMtR*xmFV{1?LJz-|KwNRrNj~q2WN_cLsroEELc_3%y=#?I~dv$mA zf3!z~x15UEwYaxm4ZJYJnOtG9+}4U#(U?xHsK)D8q0e1NZ|+GK1?`o&EPn6Ox=18j zO&w;0M>H|F-+r};py_RKt0B zy7SK#!X%M~)?O>4y3(z*?I8>yUgv_2Mlfn$3}~%?XtG|hD1$MrH^s*0G!Z&t-;Ia3 zqh)gm;ytZ+(1f%`6cJr&QKgbvp>&Y5N!0t)n$CwFkcqxh)5-G;^a8fLA7yb0O)r!{ zVfT-E%h%?YQPj*gA0nXueJMEmw`D9#sfqS+m%6tgv(*0pv)&txJ~swF9gj-iy!eZ( zy}m#T7TgjCTDjrNWp-UAS)ny2+w<+4l zFSQ7t3n7vBK6dvcoD?CxbaptI8WD4lC%>vHiSm2RNWDL6l$*MY) zf@ylWhRAumU@7Mma@f&gsm`Ul9K<%QBohBjR2S1{`;-1@QK)mOpD2 z@_8M10a}_h;_IF!wKqTr%Z|P4=-BKn4xTMUN#yLtF~>^guUSG?M@0zAalFrwo(Fe1 zIr{dcg5p>MDz{z@be=S_gHQ2I)xEQPo4O9w(C8i$vug#K>37K>$gY}c%T#k>u{7}U zypJ)H=3;ZrU$BxWp=_U+inZcRDKza`NK=8-3ZrkPPdF@!cGR^b&6QY6nwH6wW;vAO z)}@Bn$r)xf70gcz#av*DdQ8DKgHoBgyN%*QxWy;Raf4P`?rCG%TOzA4B=gp;!VzQV z8KezdhRT1k-K2za4Rl(LzZ~{vHR@Lt%gh0j<)@?stt(xb)TJ42*% zO>UNKPBDxM;K?yKBZ4YdyPj0q%t)l-O>953l9SmIBqr|cqK*r3TW{h204m>pdBtJa zT#I{!bL(5zI_kB)0pyHRS`k~?xg`Sw?C#TfYBF1FdG4BCamQMXHOM zB#gPHc~qf}YCzbhVzp!rFe%Vo%Wm02PJ9}V%gM;3_a2f~kz#F1d4?IPy2t|pdeZq= z9Eyd329s=Yj3c5tYb%v#0M9iY{Ojcr-ms#btCnlbUotR>3(rG|{!TmezUOK&PxXzyA{{SLK zMQeOFxQ<^O-H7^qfzSR#j*8~3@gC+|IZqCyEzb#(Ie;;kfNuGwfh>QxEUUY{Zd==4 zc#iganG`#P%eO!)9#|t15YG5o*2W5a&y;*qG%|=*Z{3nN5Pw?JzqYiwo*`izlQ3VM zhkC_TRRz@@ly&2+Z0S0U-`iN)MF?h1xIGk9*onugIr)=pg-F_Z8oc+B{{UvCoGXpo zyOUZlXtx@qv&xH>h?6Vs;l7oYi*0WsZ;hL>jQ6DUElp2OqFlsmm_{U5EWF~Xe4bwl zOLf7mJuO=2R8?3}%kj0(0;}paFJq`QSXO1^?CnizQj${T>Q_NE%Qn&%XFYLR7FVLi zQFAj1T}Bo%b6I7XK41vXT0m9EBr){@hKrO_H6;UB)DdlM(ifAegTbgSttHbgS~*oi zuTk2e^ApTZvw4BN_~xxyJ*mE5?GYhN{P{UGHYKMWR;#dix4N|IrAzJL0l*bCxQfeA zjW69~Me3omij2f=H5Y?P^DxKd0P|Je!sTzi%^)i-aM`Avh_ae%V`o)uHtO*uvWbiU z4>3TjiKiw;`#Q97Hvs|dPy0mLbd7HefcoTjt2Z!Qi#u!g2(2O7fj;s4DYi@^pp9ZmQ$3;2Cc771?)97-f4x6n&8=S-5fxs1R8*3lzHVed zn8R-yJ7b<|8{0S2?o1QMwZQ0V-jMo*mB-pGLm?pN=~eHqCA__9XKn6y$?ZucxRal> zv~@^~B)&-@NRPV#OcEqV_iRJ=JuAA8SkuphOeQFnU@?L@HG|=;BHL8d8_!@PPDrC( zhB3shS;}#>iW}ZbH6CJy&vVU9Yjbeu*5-Ct@yO~b$BDG~?6or#0|-FdLC~*Y(~aA3 zti01niB6QM#!s%qnr^!$n3AMO8JmFE=BV9#*8+L6#m?pit6SP!Oyoj^$n>bLeAJU< znFV+t(QPhhsmGb}y;;<1I=!XFixsuKzm`|2rg%bYCe;tySZ)3t)qt#~%bvAf*teU* zZK}DO!v5{SGrjSh^^Edc$8J=zkDG5w?fe&`>967OaW2wW9Fy9-O6j)93byQZtCsra zp>1qoig^}4Ow>zOX0VPF;{~ca+p!nhWQyJKv5pv5G2$2|x0*M3ONRCBSr&S|)HCni zSd-3bZBJF6@?`TDg`-`JY2j*9t$vL3F9nOwu!AZ6*F~)TcImG*_^litEHJ>SbT1Rz zMWrG;1Z3@5minYuo+e3{fV;Rg)1|pjDl}@U)hbFhw#TDe!jV9V0Kfvg&&866^~Wf1 zK?fD+^Ib(9v5>eZ26J9heJA#wvowR1EyZbC+^HiBnlIZ*tkKtaUCg$|R>>Wd)Vj^B zytfv>U^_4*53OO_-N~mye4;=Dy;6qu^HIM^1AW6uNhsN!KUbWp%A$5hv3PRIH`tCo z=;`ZSTiU|W71{vDO7iyxSIE-j~^N;+RNr$L}M|Gjw=d#hrQN+bTR?LgUx%di!`WibjcV-Hv!aFm+4v( z!Fp#SJn>aS-!Btv9G|pzeGB?#hAkrdOwtCFu1Lp9=j4f7Q=aWdBc*p2k+HL7POP}D za@$Lo$WV?CCbykE4<`*>u4!OSEAe5i)cty zz^-pvgiUf1HTlV|Nm6m^sje3i{;=h;W^^7K)85a_2zQm>*BmtanXlxwfk`2{*Q?p- zuWw`n@^5SFX)& zw{{GrP6+Q@Vt8o%waHLYn&z>t8#9eaw^K_TwcYfSCzc5d=}|`nL;}a0R;P$`cd~VK zA1SO~v&pk?CZ99vdbFnrLz+7>__`K~j}Z#!atN-fABtMyX@=Dt7Av$ra4XMP$W9GO z6O=tTtfdIfS7ypCF}zu$aSh$nvR_XrXJfdNT^^a^acN{Al{YdB16;(~ffn9gdXANq zl3PBex^AKHob$sf>eWqMPf~41Pw_EDwBfVtbwYs{E;6=FS3>KJ^qkcFte zwZN8m#GHzIysI3@^>YBpu`>fUKwwi6Ny^^9L@S{9d zl0DqewRYQ7mJTzKOvbhy$yz4vq~q+Zlhjr0KXJnvXHiu~3xx5sX_j#!zX8aXeGP zIONwmX`Na{2qGSo;+P#N;jzJ|76zw5U9u?Tc$_kv8q|+koF>x8rDhNhN;Xi4ojA{Q zddI;Vx4H4n;d7t(mcRatj*5feK*x>mf9x9n0Qxd2E0(v!cj2O=PRQdP-%-dOVAUYsq*6`?cQqlS_NFSOyi;j?DRZW;Ewd z+EjLt!%wH*T4~84ZLm-CZ3mN&{{UT9RwmI`&5c5w;0|l3y3lT8l0P#t#~J&*v02)U zq}SJ|k=lrRDglhrx_cglEJP#?mn{A2Y^`x()^Q}J;1^+?qk^?(MVC)& z?CNa(RzHWWYRjm9!a*WQcW4Z8t{I~ktbgp-=eUsDkddhVDt3Jf_KH!o^pYXJv@Ehs zYZHBwf6C4(t*N>4>mAM5tkvSL`#X9mS#H#l$3rWdfbBeCfusJ`)-lml;#7pJM%^?^di~&+>Zl&ux zjXuKSIrmK~AkWG+jMOi1+McE)$wKX&*~tQ-TX+m0w)J9pW73H34Wq~*J$T$jD-hzI zt8(V2X%rVBAh;9pmM8gDdyA?3$lbRA82JbHew73gt4yX`mCw!ZT6+EMrQDDuq-v^* z*i&NJsIaS)j=z<%+oSFa7EUuJ*@ zRv9GJkx|A=k;dfOF^5*xk#z|$j(H}vd{bj#4a_iUs->TT7P&iUm`S;bP&yitE6HKM zS&@snoP$lKWjY_d=Y!ELjcbnz>NZve@>K|+XBE$LzbEU>cG|a$mh($nSA?vrxMEKf z%`3z9S6W}%C2yR9bDor1=5+Bhrxz;G)$Cy3YLm?zF_*%snVA1(`43Lw)yatbgz|lBE^9}GhCM1l3+)UuN$p^b2*$6q zi`3{Q(ruSc%42ri=CE}8DK2jIDClXecRtOYVaKId(zOq<+!97}){;Y1=Lkz($+bN( z?S|}NRX8G&WMz#DU{;Q+tF6-cYAZuZFs+}lm6=;B!KZz0CqfPmCs})`qD8beO>Fot zLmHf)Xpm%XuRR6X#^YX(;4N}1Fkp57_NknlnaNt3qeaCn8R^mZYTY3$BLRr*UR&bR zcctnss^wR;dWrEg3vO7j9CfcB@s_xnmY~~CKsc=9DIz8Hc}hOTU%5ulTAn+|BYFmYTD6s zMVtv``G=)(x+7m|p**ijhT}1>ToIEH}R^$k#hm={r$cnvS5x zcKTKnIzvsUnH+JN)@#VaDh3;g#dUfL{NHXV|>#sDC<*7VJEQ$7sqplQQtbI_NXf%QVTi+1uJ2JAbLuUqY6f{v-aRXzj%fUW$mv{S zou@TvC7p&8u@$W;6s*;LOGIbk8i=eF3w~|bND_ZhlG??vIEhCK(916-*)0ety?OY*tbI)$^bq1Ti z6~Cw=7>NU#$cD}4ZUto|6&`ZOEBWLWYb~tc7?@I(k_$A+D25OD8qZ_wot~9FKaBAfF znfaun%Wk$h{S!v8xVI8YitKqgtnUzpik_)KkD zNvJ)~Exyrp4JHPST=_C}#a7(*`b3bMQmp6wV-D5oWg(acsXImoOM)iZdAP zicp6$JJxdJZg}3uibrny^YTLnYBN5E2})aNwTrHNd?R7VbJ!h%g7;)`O-NFi?Od-9vrbcF^s^X zK%|!Mn&PZHI^$OV07QfV>aI|J2mgrAvjV+y{X_RmUERy-o z6and0r;bZobwY3&gIyz|+^T1uP$Fkyqo6&jVmsSgZ5i#}Fw?hue_F_jDIYEmBBY8q z%bzsoG;)(TsXH!)jiX$ypDV`@5S|MW%|j)O_x}LtPU<^U4{;(yV2(!ev&z=&I;EAa ziRCJWmAz>+VJS`;!*ibUQp)LBoH$NG)OS~HbuRZ9JqA51qSEx6d%J0F?oR-5Sc__Y z(IklxXwkALOwyfbN#Ck2rmK5rqZ>E*VwC8;T$_W7ja0+M#zUeZ+gX-{7>npP6AE^|^Puo+-#` zC@?t{cG6*}%&^JARog=cdVYpWvs_7*9M$i)%>}=dlCV!q)%{ycg4Q7DkaUazu~Q33sWl0c9!VImsghCRlkiPIRmM#XYTai)r(T`x0((y=}VT( z%6MfNx{-RCSK4HDv2R1jtn{$Bn=&kIw0Ff!*Y9y6$0yy*c9!i3w2UYLK<2cJZevoc z=qpQhdGzYgj1WPoZ0Fg?txZPH?RTzFe3hR)lUt4|ZDw@Fq$aG!+}$&{T=7 z>i{6trqohdR8qa!oT=1|@6evlNG!Xbr9*#cwxe>4*FkMGW&wlRvh`RW=r(yC+cHo+h3y&^L)nd%&6(Xn<8uBYx(re&SrP7>S>M_Po z1oW(D&Nwx5Oq|-?$0#`3K_;7e5js|;?0qTW3n-w|kgAO0x`F`3dH(=|d`S$p=$pag zn(W_FfnzVYU<%}v8h2*7ekzn`T2EASJ~y^akxjrOsr9cqp6tUZE5&r)E%7WDQiv8n zaBGg9Sl{N|U07(?@NtrjY(tvZwPQS!UMsmc2Q}+jCZ!y<@u?$l7_S$%yI9=d*F&T0 z#7GgCsqAZ7iAzRj8r3lJRUouIh|G$+ivTMlP}X5kQ8LE4*)^AId^Egctz&;{#U;Ru zHco3gczCCFZHmvT*N-g-*>z^im${Lcj!jyJNqOVTf!fr_fw zj#p@tmgOH=kG8AS^LR zXErx2uSTPcV{~R|z90L1`*z_Mt5-nY3$SG2K5TWZjUL$`wNRr3^{p!Bsaf-Ad-;A< zNnzJIyJwO3kHLir5yV^50=R3B4oz(Bx)vXWeI(GMoU()Us+T%VmC`cF5|VudIAU6Q zol?uJtruhFrqj2#mJpWdz^;!<_+vJsYv(!IPas#IT=-tn%TnABVU9(1I&?`LvVaEy zw5wHG=-|V9S=MxAqDRdhB=CjSgFe?owDdL2nAi+rzLN3In;rF|$p`H*kT!dOx-0ij~*w<$)?IPU@tr?qyM|HHyk}Rp7T2zzak{dS zaMd07EN9l0q$SnhkYH8aIhV`EIjkZ$=BI+?p*$Mqe(~Qre(jOa%Z;4YD_VK?1SsN_ zwG!ZiR`jVAoaE%1Qd${BPLx1stFzw~nR#Yy&mAkS!K|H9Fk?J(P?2s8w8IHKiV!*BqkIi(z5oLjpl5}n(L=B~#&vVcuxQida?G{R6)Rz{Tb0E7cfvy42P;;0rU zsjGIg1Ytog9I(*!&w{~M##exU&^`YE^kj5XJ_)dI8D0_p06_Qu0MU`rT-BGnc3|)9 zwKDB5_9izPc)wuzCyM9dk+*Qi6>Q21;B~DzPm`7&D?@3urd7adzo%-2xh-6Lo3WZ? zT8xmLqcycDD2d?XDCl}bcb;Q?>XvCTgB`1l)Ad~7vBhi+Q8A3?H0L|B%c+Bu9)+o0 zkd_9YXLjuSSk_(6psY9|ppNxdoZ_(OX_|WnC|K(jWxU*XJ=JR@GC|uWb#xM2D!Lz0py-*X4g}TU`KvyT2oF|k(GK?EYwWxL&COq z7m0EgXy8;p5@Xv1(Rnt;(lf1#|kylkFzmmzenGzP6g)DIzhj1xUqt4zb}&Ce>$VKwCA0 zWPKJkqrtU0j_C1_9vJbA3W1!t=DkN$@XfroT=pa<)Nvuoasss z{nt>|EY{BBY)7wQS>hQJ=42pNjkLdLx$`IahrMXtye(=~K+ZU;Lsex?(0iQZjk|$U zTgqW=%hr(D$Oi9PiHT5Gy=a-!c+1`BU0p@yx=%UDKzr8gm#`QJQaK!Axg|n+)X2=F zbHJ)Zy=PF9Z5K6)BA_5BO6Z_;NDwKBbQLg^&_f6i5Tz%fNU!=x5tI;+06_>vN&*B3 z(n3=dkPblz0TdCD5}KkkmB;d)yze>Z`~GC+n#^SG%-nnST6?c+ExSR8=H^!B$%A() z%4U7*;O*o_UZIUnUqgj(kBRvOM>@X)hw(+!f;31@XBB+{gVFYszD)OVMIFsRT7q` zB2{5_RS^*9_vh`!S@Sq{so!l)L8VI;z`*ht0fqH$n9pe~#Ki)T?<{Wo0W#b_!d4sK z6z5;*b}}dhx%TJN@>7}G2V;)m-IV;wAo%=jqs%RO1~C=)y!S>KTLIG!8_Fx@zHc9~-RXkPuiL+hgjZo73O z1;w3!lW%2tz&m^tjC_{4nk3f{Md3b%YOci7N%a7(0K9_ISDjE(PxL9{vQxU=fr8%a zcDa$-*`T=Jw`v#O=j~5uc{Ep~q}VSayljof*UfYD$?am(Mx{QTAFj_RD}<~dit~-* zs!zvM1%9R;TsFBU|LT(aS**3JbUS1WoMg|N1k#lNON^a4wG(syTXr3^wR6*k8S?vs zt+m!*V1{|nx-#wkmhz(WZ96F6W!AX|l|t3bmr_K*My~G-56*&K8|kl*lK{$heY#_2 z{{gp)xCHweH|+3%tIC9K(-x{5zMtI4jy@vC>I zfSPyA`1P}tOTeA>ky(+1a2mmWK`XcCnfBCgTe$~8x%+hnPjFS5dFPUbuuto(K1U8+ zO-541#yP07gRnRx;;Lj1(I7r+zM->!2DbmB_`}UR0So#Dz^s zjHMMJENQX52$t?TjMGh+R=NXx*UNPq{(T2FyzzH5@?&ei;d`It-AjEa@P@rKVl6PX6{zSV!!JhiZ==QlSdk9S$I__ zXdkSfb{%T?IsEg2!(V3!(-(*ZDO}|x(o$)VOGf#1MlZZDC#s*OSFYMHeMXQm;wtfQ^TuFhz^u`H~_lCx1%oxCkI$<~$R z#5lQ@66j={F~F=hxf`zQym`rG?0r_CLEtSATSUil0>H`FNe|Jz$Hbi20}KR6e5bpn z1&>`f@Tg(7X7jOW;8zge4vH!6w1GhcZ5XwazqC&z0o8w{O^ z{0{y(A6tfv?*+DhTkM*(Ikjew+DD%{t#eZ_@0XKAO`x@~n?W#4s&Xdeu6sz^5niT0 z%tLACrQq1}(=D=rSKRlmnjuPVWOgtew5iA{rN11}Zh6|CRR6kw+dLyr>}+@$f!X}} zMR2FpfYGA0%IF8RJ=Z?4v`VZrWic5AcPank_T);Ye<_{=s^8Bw<6Y;Y!U`;M@z9C6Ks!PQvgF{6ZqsW_>gsTvt22 zmb9csSrGD6=4Rn)KX!G#jszTLeoZ}i{`^YH=cWepW88#WdOwF=M_>#VCe;Jb`T$Mp zc+^bwNPQuG?wGjtL0$BZz|i-WUzp9ZDv!KU4FJJC4KeXZ+WUxnM8&7bTP!BT0l{TY z@q1L*$%v5@ONzR6R<+Pa{+g2m2Y$!=y)b4gy!U38KNYUqM9~{j$U67=X4$@d7E=9` zBCgw~Vywo-bQk%x_0+-JAB_uIsKHBh1mHnjrxdGfj^82RnU@o4GyU2dmHV)9^&7=m z#L7E0N0vN#6}6^TC@K7_MoXt3Q6$I2Rml%&elLPb^V&7d&vTugeaV<#b*f!H@{RaA z6FvBa=DI?$?263HZNa5kM+mPvY>pbV2CJE(q>xWC5)A571l-RJoEwal3e3ndPx_%& zSLsYlL>j>}ItOkEswl2BK?bP#K5H60&ZV=K=B|>P;2AdsA*ima<-D|pj%RJpKVM4< zUyHE*6po0#j$(6s@2te%7biL1%ko#OSuT$y$;X^_DbfQ!HvjgtYhwHLlU>-cyT0*r zF5e3%k5aNgL$saMHhptQ%b9fLV~B%>IA~Rz_zU`k$M7ay#`vZ zcej*S$5O&EbC){#*%Zi?btK=eCT zS*%cqwPE2Gsjtc;$A_qnKbBt{8&4n!gs#F{+|S~DxGcKWpC>IB=3pDJk6h1UHh#t9 z{#oNAWxkco6YFaHUX%qqcSPPdvCM7*H21PDLbxtP4kl z6DmzEufyHP75`wK)r`iMV#^c&=Vh@CrWAqCx7^vohjos&`8);eCS4LCz-uuBhmfCr zi1E9>W|0_X*TE7vQ0_Q*Nw7ebL~$=n!vhq0w<9Xry1fj--~#x-e5ifzVNP{CcIw*~CYtIo_~5$VUUq2xoqM`~ENo#o6> zcaj`WCoyQ$*=r54?MlVBy%z=V=Wt)|8h+b*J>nL!eA>O( zDVXNtHkqxb|ID<}S5tYx2S5xKeb|ya6UAng4ngjJ`(f4bpx)tu-tvnl1D?marZptW z@CE5xppD*%gju`JK$(GzG_GtAsrbX^d%QrSyb{Zu$Hbk;qM|#m&Yy`dn-W=A}^4$=;b52X|5z`sO0d z+j7I70dRGVbI*T1?3;Sy|K;qe(BVoKF^S%R9^MG7v}oFooI%~mX+o-*TtpU|aS6K) z@XLZs7flhTCx@S%SyRd}Ra_j`D)E_?SNe7 zQA65FP<@!|kLXl=ltw&aeVitb-Vts>A@@Ui0B_WK2X3}>s$FE$_-Wvk^6u}U=F7m= zU5+`S)$5_z^6Y@(8EN6Xicwx=>FJ8XsXw)DLNgFIsW^@XyM&+V;D5!ERi&TP3t~Y&#l#5s;=EjW-Tu9m3$g5KNXzaG@cC!~hz@pfYeBNN2?Er@{SdGoSIWg!iNuiav= zYmE(61?n^5`rkS>mug#Qmg9wUu^}N^jJWDYsSVY&Fow8XES3I7)Ur*zG(<|A%jJ~N zb1)~XZDb4Ecu*CoeKDF?qOx0QQ4d9B%#$-wbf<6TN`?bQiKB z!SO9p%laJ2fUaDYVBGT#4jvSKx%<@~ z%V>FisC7jx*nMr3&a#cmZ}F4qUa67x?~Dwe{5MF*cn*heH&;K`YKChOd~I#;5CXgN zp^Zlo4hWiUyjy=jv9-NfLN=8 z-|>vArr;*LO@`~{BGhECX(rFEP@a{4v>Lf&TO{K?oPnXfclhQ0nc)TW4F&eAPIY9v zAheuTSN+eG3+!K-sF>%fIR$uH^A^nbMQ3xIi|>|O(Tbwah+xxT6Kb;#T8?ZYWPS$8&jf_3gYF%2G=jM!=)cBDba@sL}I#nM~&D*#? zUc2Dmc$5D$;W0KSJTJ7b(}SED@bE#y)Vq;<38&e*1yj^<$>uMInYG-`&)OyCaQYJf z^H6`^Hh~=MgI$q-k<&7ak=mAlSGa+e<`Rn*ZL~6t{;UeyGKgRru+oxJNQD7@n~vOB ztCXxtY1(N3tWWmclsa`%c#T?MMsQT>C1{Z%Ev;`mD@N$; zb;Dvb!$X0!RJ)mOFqS$2%wqiZFnoU6 zV9e>sx?Dzp()7iFYhV1R;$3#DE!h`vk5dLLUd6;azwtbD0n;C`c`Kx6dWyPmr-26i z@XgL>rbT(xGtaeaq)^S<6%5lKF5^xb>^(fFcvZe^4RpYU5gE=WEQ3X$ejy7Gy z3OK3$I|(|o__~}XbIp~+xeeW-V?{7ij%3XgesufUU(>4s@%>rif2VMJB1L!smks;4 zSgw=i7Nyc`EH4v#+3+VITqfd;(Mv5)&vtDgJa#-DUlY->i||r!zA|~x8Fl(`H;H1r zOeo&>@D8+#y=;FI`o82q!h&mgCpvmQxe5dW{mzc?xJ3WS$H|?>{PQ@eD<%Ksna=t7 zMLs>+dt2{7_^~w*zNHI)AAq4PXLDg~RXW)g^A0s@HebDWWg2=svKNUZHQU#)Jn|EZ z?9MRFK{x8)M$OV;{fanrVe47p3drd@d`e8|PTr4(00jNqS`$a^g89%3L9GWi@Bn}+ z6?&@seJ%kPAh;MnQA8E@q{}U(E?GJae$s_AKGCcGqQLJk=CqOoYdGtj?^6B>?+5`{DU(^mRZlRJ2^#X_E)AU73Gqoj_qiv`(`uRF z^hH-gWHo*^Z`$ZCn|yCZa6%sa&DKIn#K^hut@Mn=(yqH+vEz&zX?E{-p~$nY(ChWX zA^d0Qbl+q8^37;=O?MIdoFKSy?6~okjG4XswOrB%D%#M4u(w6x}$+^Y|VX7`efmb6E0Nc z&Z4ezauL*=_7zSpT$(8hrlo(Omi%$T{?1@R$KeNIRv)d$(Ekh2h)H3fcKK zCA{){rtBlC;NK)J`V{$E(V5| z`a}r4w`!fauhlf$P9M(ely8)!-J7B|yHTI=qzINmG%Gr`qw2Oc@)M-kFL*sB= zZz8u0Nc*ZF%8g5&rXw-XysTjlnA&`TCOo+7qI0^lP7CJpQYa7>B`q=HnM-1AdoIg| z7K|KJwHo$n=p5EtgNi+r_emS`uB&*SgJ-bJ*s+LFsx9-M%?WH57VK|1^vPqof5Xn? zc=x$AdG~OwT#q#t|57G>Al+)rter1Cx-|>ED%rx*Yj?0ovs4r!x(+%VG?k2%Jx1a^lE%p;hQxL$iy#U{>#`miDz)PN z`2xj9HQ;VGR5fYj`bY|Q1GC(T3>BD;HGaCO3G9?7J+mD+^|JXq;kvJudc&3)OO7gclg2Lhn z0M$z6*fuS8*@%<7fB$fVl`?T}7fE0?3B$(6TpSn4nZ0_9E6Y>MTY5D{U9$oc<>Uo!cfSszfd%dQ-On3~Dy|=-*>Q$4ag0o@z8TO%lJG2UY1QRySLOOwNn(__@mAdFMbC+vg*1KTO3aYQ zoF2r~9#wh?>hUSlEYV1Y@JcFOoSWG#);B#Q*x|_=I4b{KV%Qi&ouZTRz!vY~+Y#2K z_`Icfa;}1kqeoTa)UwlCTq=DSAE&)^FR*lm-k%LHAEqhDA_6i8?8RB~bS;hBPN3Ui zq@EXv>eaYKF{_Ki9M&9=KpMDY*CjnUb)dD#We~`k|a~UJB)$wp}xuLw=2FRRBgXxf5wdLIbxtJQKt4Z zc=SF6z(jAhKK*gO54TF&E&c8+CawL-DT@__Q8q2x^*~Z2f*GjfXq8gB;P^kj0rJn| z;8*1#K(&3_;>faeEr$-8`+M`#mH^U5&vLFs3oI;0)(23C#o)y74%oG`BUq#lF zadkYZCDIWg4%nH_H+-b=HlkZx9aVc!*S`h5hkflXZ)>M`g+Z7AJ?OMw)Sd0R2m9Rt ztj4YT!>yDb)W_@Jl@_gvXpjOY4MP-@%Lg1cD{imSG%eo^s0RozAcYav872Bn$sa&& zJ46vA4}R%)RYP8mSl+Wzx~|1c)7jxi*}sH1$BaXyO+9X|B=NW zWdgtAIOx%&pSOx@u50zDf&d}Vt5$Al$7ttSUqWNRas?p+c3-lfe!bc6AYo+j`)mR*%{@_Oj0=4 zO1PUsY9^el9NBg*K4%+5d^k?7Z@c7P``mY$-MVkiRqG(8%>}A%Zx=Y}tsPi(-rkJV z`lo1ey=rRYQL&3|7}h&488kIT6$J@&fA4gv>9F?KrhtQ)%jF>r z%b)|AB1@-fM|k{E09tfz{PXeN}e!=jW16e#QA9{=*q++^8w5g%yPl=hdwAxRNO}$=Qa9+a^4nnrweXM z0@+Lqn*bp`Webng>WY@SjMGJ2QuTb8RsFE5)2<}fqc~1T>NW6k!5x8OexlVjtfx}- ztdyo`t+lWd%F_!x+mhqe>9AZnNT3FfH=r4AITJUw$j;GlgGp?$XU`cU=SS7=XY0I+ zRRtNb4PtjtEg@T+C1Qh1c^&e{2|e}$JQj}ZoT^W`bG(EdKzB*v5LNOmn$DXrE8(Y_wZZmG&s493?m;< zjh@gzfi|k@z_+O<2YC!}&RKWM`3M%UcO5Noa+2FQIdl1vu?I)l3(WJDXt1aTnxPB! zPs(Db>TTLcPd^mtx!KP=sAA(Mxm)abD;au{nDiWK^-c}9cS#q?x7F1w*WRbqzs$V+ zqw(fK5%$_R)=@|3@T8CREU~wurx+L7aq+OWa6qbi%Uw*~pNvYWx{B=-tsOO42z<(0 zDxET$qUMld@L<+gL5iV8=gsYo=rc{2p7l1I^lzMUmyy=}6{}qO1NLFN)NL=GHKs4K zp_6~XomAy{2K-v3J0Rn6{4FC=Yj!s=^MroToZqKR`(b(IAQv9#-}|WOC+5lvv=m!( z_;SsD-GpetLSIFOYgogcxOW^QS2>oa|W|XsZy5BSX8{MF8F__>6RyhsC z|9Gx#iVpxU_5Y?HALiodP2w8z-aN#7Mm*0bPIA+Ee>?)W@H8jz3X z)>q8`hTpjxR33Fsh-kh%ZOr38wFz%%^NXcKZOrI(5S~cX255c5TCRB>UIOUdLcm_% zSFYi>pt2jR!*Lw|VuTu@76q;pO}pI)5y-5jxMu7uF<~h(7#S;Kk{%g|we6@B8Kl&- zG1_1Uw@#ur>pg%My7(aM+Gu*hwLb61r)kBFfHybib(I{y6NA1RZl5{h`p*3L4V|y{ z-IGC=KcU&FaVvT>Q9~DSw+m#n)wsZGbKtH@($DqYYSRuK(`(^FxKAcW+Z)%?Jzah) z;F(;mk^nFVwd)5eyrAqT;}7JrC*4U zWrI!TYpbA)K>5n!bb8=FGRMz*^u1h}rM=g8;WOTrjS-9VLE%US1F3hL`{%CJ483Tq z8$2XktYlJOt}Bw19)(X8=g9%@I2Ar`*J_tEouy4q8E z6|vu|&}))ZAN5^x%-&^)#Bwus@cn1G$}rEog{fjz9^qm{F>6~>fkY)aTHnTKPwV;g zQZlWKoc%%tU9)a;!p4WubE(H3%LwY3_RPk6CVs$v{GR+^+iqipsPA!EQ+_DE-&N8= zPuh8N=PvtwDcuNm&M3(a_<2?=S(R4#45F$SijBDJ<1lKzfD6yW!NV`TeSd0=8@F-7 z(L`LpWAVw=clMAj=|?Y|pOgKl-{Em0EkC2y=tb(>thiSC%;@+X)|1rgD)Z=QKQ5cC%JY*P!AS0m4} zD({KEne&L^OjHD^;0jip1%G=IJi>Ln)YeQPnvAo2umUa|P$Sn~Iya<`D*B?ojo4ru zUsK5mUqF1v3M6A6W9h*fvmygZJ4KZ{iE5`5D2gHc+}7D>J&v%^sg9f!diB_|hPZIWj0?xD|@ym+iLLssklj?Ak zte~M@O}kUOgK~o!xUeesIz+)T6#RJ_YEYnM=9Eh^<%bkZ(}@!xPeaRf8fRd0EvMNG|ZX zo(uMsY#qS|K^*vzk2I^p&%g-(?3}Ed4BzoxaKoSD39vfmv3nCVcm_o=ZY1BaKl?qv zSibn6Hmh23b~S%R1y>2?IzM1$aSxbY_#*4Tm&Fj~qkspo>XB#*%Xp_YlBp_SO4z9s zME@rC=zim~-n_q?|ISHeVL!B)g}EUff`eR)Dr>J6H_g^ZodhfuieHE?y=OW*ugx5- zeaU_q^3dT$tK;{$Qln@mhlTB~lEEA8+M55)U27z~h&VMrV~~o3?>6x!zvW%Vc@kYDXlivP?ra@$ z3k3iGDumqu+HbOemCt^i(gI)7!{q^+^7ng?ZH z`sZYWG{@!f@UxqR6a2pvvY!Tmrju}f~((%vA@zlL7TbT^Tgqj9;M8Wng zZ8%huOT6wc$|%;W6x1fpffYMpZ_cY49_Wd}$;n#R0hGGy@*R5O^$W`#DzfHNo2DMS zHmVFcFfmtM#~fogiQ#z|0Js?RqedGOVb|A$Ie9{Qw*J}+Jb%C0J~t}5C0GUu{_KO& zg*F8);HcrxO#ps79oqj62|*T(HO45%Qpr;4n9?S5zHf=bc-tj^*Q<(gb=$_ zjA)5JS&j=eO+b7z^?asOAcM7CcT~`nlc7SfQYHv?VtH8s3+^k@SyhY^;4+v*D@cZ* zeM<9+moxAlQ>5$nYO0R2@l5|*aAGv3hZVVm=Cu?wsN^jb6g#^msEj#P@ySaNOoC$! zroz(DA8gB}LYpqe-#iq&BqpFR^Ow8rZWt$Xxz#ymAW zcH7B9c?G}kJnp09Jd`YJl$buz-nU6h{0++BW!UR1mJuKwH1zt!eD7weYcZ-kPBEFS=I`OAGU6d z(QcBSC3M`t=X|Kk!mX_sV`E;#c!)wi+KX0XNp@JM@lvim4G&|se6Xxjuhs|8+RP~) z#u&;~GgB8p zf5=Xb15dtd>w9L5nGpJoqXu4C;&%R~lMI43jew`m9=CvanhP7FjJR{%V(jxTO4^=K zh(IoWaSJ2eZMUqO?L3I#OR1m4+CbC+Lh^{K!_jhu!fQ|P_dSBiCrF2GE$1@$T_u*8_yr3d!1p-jjiX}!)xLk1m6rE>dPzQN+wxHpzzGvQ7NRzGQz|_ks%E0;wZiH^!V}^^mJC%A{;htD$ zMKW%3t{o5L#^p7=338GPs#05(6uPo=D5l5~6WGgj5&{Z5JQC#2Vj~Ja4p;-~M&l9R zXBUGKf6ZB7ZVMj+-9%_klt+c}lQdVD9{6SAL4`=M3$n2`D>*IQasxE^?*o(`(!fQ} zingeO=QxAidRwP<(M2ojO#Z6N(?_opbjv@Q34h_dN>IRXA#0L@1elW`A>Lv^b3qL> zAU(p;R>#PZ#05t)ltfOdouZJzr(uo*N(M=#)!)wiwW#~}i+6;xNcFCGvUKr3T)5@` zQ|{PT!gCg%^w0Lo(v8A}*fl%rG$Y+^Y^VKS;uvrKXds6X%H90QL0-V^+3osvw(T;{7rJOozkR6 zY0Xtu5FdoH6Tmy=#**bCYxzbESdPZyZcFc*&_^TsG{fAOy`?kl@_zBox^{6?1I$bl zekvG?5U*=7D<(MGkHM#ADUrYwGc~Y!a;d8@^udAZC>;jfc44ci036fM25QaS_9q!P zTu7xCMbkVA;{_{e81pG&eWC@n{Y1gGo0T-NGQ7X>SWY+SimwDs^a&mrHFi1&ST+xk zB4pM;t=5lgA>K7dnC5%{Rp}T=gFB8~xgB&?ckN5uLu_5D)jVcTxjyHDS^AK;1iJTbJrsHp(fdU2P1!e3ZO6^EAI!l9nM%q)h|O zHG!>oMZGU5D>d^$L~X~!y2x(codWLyPy(X)Sd41Dwr1cMRS20Nf>j$_cC=2ha+7?- zWBIUnT5Zgf*S7T)EU?_s&Z4(!bS(>b)Fm7Yo6z63eQUK52p_W#Oj`|e7v88#Vz#D= zU9wqE&orK}Y`EH5uuj^%6B&$y;nvxYhUA=_3_^JGYWu4D%sG&>T4X?aCJz9DuMVEv zH^J2dmzdc=mSv}9vYM9Da7Rs@zJ=3`AqzDaUFHZ?VO3SkeEa0Y*xpASJpZ$tAqdwp z*<7NkAOW1gAF8@-`-Wo7i>fbr?-XE$Ka~Voyk5qUS$Oc|y)@yYL?R7FpPf#a{T6Lx z6s2$HG5le#@m;MXxld30{CtV>God~c!02{-J4pgDo!6@0_%@D3jg$UfeKBW)$gWk_ z_`q{BnHya6Sf8f%MVL>%6V-X}&vcGam4Le?A1pnI?}Uq8F!4DuYAxHr9bi@+yUwVb{H}@poBAIql5L`bRh|#w{t8dBbcr;Y=ij%g;dS)JMa!;;D>~^?} ztMbwK@oI39Wt|dK&SoyP>9Yq7YPkWmSj+{5G9<_qiDE`fc9HdUk_<`5mclsZ6=7vX z(z)*HXz$1{2Ra)~UypC^N%{{Lfd2JQ%B}x!nM0wMdmbOj+`0ex{{!|UPS~r%Pa$rj z=_H6aRkXEp1#Brvv8XYePVr8b)KDydc=`6$ulS?*YF}}yG$w+9{~sPdStaOTA;`~zHO&NJfSyc zw`R0QlCtO*xgNuOy7B2tgVz}WFI*Xs+Jh$*y8XvOtn|oH%({h9PzsaGokcW+X0blx zDMm(kP=LM%watTkcJ!KEyt72*#mOm}zR2i4TtmwQ#YOf13U^+v$|`ff`sx9$=oH>w zm#Eb#E_R;H2GqjCSvPPl7X_TmaWAF~9HdJ@HyGq1#V4mbAhydxu$FvdDt|7^dGV8I zWs}3*DzaOXSq*Q+2Ku>qo6XY+eH!SLP#gq*hT`7dvG>-$wfYJr0@(UhN-E0=OK^dp z=KXSM?%MLSXp5WnS1~mHxsB6`E?VmD9IhkCdYkDCl z1^8)xq|Fo>YBivTcLKVZN**g?@}E6dUqD^X=B@S5(>mb^BlHE`-)?P<>RTMFSBra3Ca;C{oeBzc9U8>0q&U}qFr9gkMkb6o z(m{A6L}qy00^)-P!1iv|K+kxYxuI__;H?~RB>%+&MfDFn0Y1|iR#-Er+b`*Y37|@P z?pJ=~1?>dkDX2)Hr6H3E_sq^CX9`Ihshx#`VC> zC!X$HwozB+R<~jTn6qE?(JZeVPYwB<)rjRlrD0be0_M10l{W+1_Sk_O6gwG{QgeK=nu)B zd=Rg1@w4Fb5worE&@l$ziBF>=`;TeBmi*U}k7HFx{}|4u7itIbKLfh~41^a6gBBv= z(=9rF8OFfpYMUsbsspj@jPiR6WqG;F-9{Bt6`<$$9iP z3mk18W0rGOO(G0F#7y%jB`WS?%`9EQs4rR;v!7MN%-Ek1#FIvFsHmN--IC<>=!06Q zep`KpfMgbn#=l(Y7Pgq8Hh}Vnt*L$Lg&?W~Z{>+H;JIl73bp4tCB&;lFYekaPKQAm zj^Xo5RlxTi=ji#>ybWd#JpxFx5Q$7nd1UBsb_LoxvQEqNRlgqfl=Bk*j7IQy9IZk} zDY0hLA(3kU3*(Ne)gHhToK)Y=m)HfLge3#T9%lg~;kK5;9C=&EzU02^#2!P+vd#~2 zP}Pjo5WQsIk$uyMkLG#PIZ=AwO+!tAFSc4TCo7pKKH~8{ef4=nXAr8y&fBqq}J5Xb5sdJm-xExix zx?_y)?i0z2QcmUuAz7j@k#oSR(qezQdGwMKPK#z7WPWLy;2cuVqRKaLMa-s~`_wk! z6{i>0)*t(@PR#r|e2u*4qrj@#No9z5j@Nlq z^~e=TkI(7SE*C7)WhAPjL1*9*PIWUL%f>$-fNP#PJe@w}?Vb z6=^R2t#D8M(UKZNd&OC2`XyKK&z)a?dvpHl5l{Ykj3bL*`@gdI!>bBiPy52Trjyu} zyu}97M#>qMJQlzJ3ecZCh$V=%sA<-`;iZj;0?*69!5Io7OPFZyth}L0Sh@d75JJjB+@P=i5TqY-wc- zs!nRA;Jq?cr!=f_4lhI$Xh?7#LZv<-wGN0PRkfHY;_}sRF*gG4z*?OMPt>`xco4By z4rx@MzJvPB0p&<$&8U}VUwxbz&Yu_R3Kujp&w05jC_Z4i#kEWu76KL*GGNjgqd!N; zk@I2pMsRTAVeOT7cxNi{n8;LH$qW8tI4=+6y@8if@~wA!anC@}0tZ|Okd-Le zU;)k|%DsY-RDcVJ1avP7^H#QsG?KSM(}`oaWKIBkJn$sCj*5uU@7E$$C$)Y}M`Rt) zRKyvm8Yr#FK4jGQajBr?61&2WG9E&Z#n{6P;5y!AQ&^X5zm94P;RQk1Bk@&^88Z-< zbrTkjag!uPWG3073wqb-x6Q)VWPuy3#^kO|R&&$3Gv&$-nrA-3EuPZ%Xf&-R1af6V z&|jx}JTYp70;e6A=b`mpVu;Cw7qYN@^DqnkKnm07syJWN_D#k38>)%HBIRj!9FgOg z*Eizo`Zn~%YpjM)2Z9#R)X6X0kO8b{=7|Twj_{1Ms-Hm!ERA0st7E59>|}d;qF=-j zYB@kDgC|)f_Ni5sT2_&`{8TP-XmN6O`J?K%v_dPpm1bF2lE~7u4opx?KuYx73U2O% zrg(#=gqx5F&GmAkBWX$}jV_izN(ytdzlbnm@g7AV){5ztC$VeU4&?VB&#UZOFI}gS zF*>U4fh@hqC7`UEWtzX5MO1kwWPrq(m-zrLa#sP5Y-Et$tvm^c8^N^Bi|}gPc*o-q ztdf)h+P)Cb0tJ(F9Iy}nM`_<~c{r${6>Xes^`_A$IRAm?e-+57!g>i#gOo`w7QB*( zK7XSkS*<+PFm&L=DX20==Q@*>4_05Mmp?ec2Pg_0!_ArO!GqRmD$U1#E;;E+PY#j1 z!1x^5YJG-T*jZojp&s~0FDgd1DIn|{Ao}O`pZ8hX_J5=R0NtKOzQ=bS|1tZ2uc5di z77l9+dp_|P$4~QOUFVcZ7MaR6iL4?!Rc?MQ`wbvdGbr$wWApBSWk14qh74B#+4xbZ zWG4yU1$ZPxhyM{CuJJ>fmf|9pn?>h<4D`pvY;psJTzcr`S3i@D82%kq#wHq)YJLkH zhNy*FU0c3r3?!jJ@kA7P3rG(C#VX&QoT-WkcYSg~GF!+1A$7FM!u>t8 zA}IFGL5jkJKcEuXb>p4eNL;mslhU+E4INBPtNB;b=5qwwtS}r{PdGahb%qA2g4#ZM zFdtudP6Jpe4Z(P*pcrBMjy`QgwjS5}rw8pt`aQ9ITxzaV^l^C>n3p0rUo7oluk{oR zg(4J4v&FoejO#ZjK`&7Dy2vyq=!AR`CwTVfQ{D!JRb-4Aaw!~B;hQeKH2F}ubmV!i zX@lTtM&n+Lvs(rF-^jYH_qNbT-AlIfCv6voF`-x;Wm{WyMOmU-n&8#IlBpKx@X2>} zw{1<%`{3G6Ui7Bw=td4FUYvfPTHP!(BRvk#GBIo>GdeSaH$;zPxP=#+!txv1UsJ@6 z*WzB_cO$|mzERP;>RPJYWo1qN1N{HtQkAF%=0teXb5IvFR-Su{HG*ZUL7Z_y=@NAbeDi#kY>qai2;zxY zFmMR8vqk&!RG+~TSSiD0s$Wc|!pbD01udzhn>ZkJ{1Gms#y11K>4Hc=L}(xr@%0XG zPj1?3aXV=ai4>573Z2W@iTtv_5f5DI5LkY?(_7)iZPucz0I5#EE7fxl!C!d_J<5bY z#}E@QnDU8l$z!%>tz=#F>N~R^(l(<%4>;tX9VBxA9s9PTWQmeOw}Exg=^5{nV2{Nu z=u@Mt!}IzP40CPFC$xZ~F%-KwXU;3GS5JFeU0+}vXgNmQ2!KoPwis?3t`Ws9q0$XL zi$m7)+Zh*jk&Lzm{#51Xn{zkqMPKqPZfpj0@mG`94g+6le&Kx@MqRTL^s!bWr2E)W zY{B5ZPjL{x7?0$@tA0myRtr4!n_24(Kmu~|yV@+*QmOXVZ3QdP1Q)Slu#aJ(|PmlYs8N4&3DIgczi93TyfhP3PPO?(f?km z4|7tC@v4nLkGfUlBspaDVsu~_)&F*xV`YvEkwUT|FPb6Z@ZSvt2Lh5{s%2bEIMYE3 zYOLsIgm9}=PWG@pxidUyYyW*Ji%#fZe!bdSAMlw_t~6_bqAEUr&NA%Ym_RpGBwN(q z<>Krf{x5L*!{JD)2`@Q@N4`)0y#fSD7L|bio_}m1vF#L>C`8Y1j+m`iFOH(YjWE|+ z-;hGK@K-ReNfklq$0$`b)!q-x_)!k316rqS{>j_73CVkSj<+g+lz<4J@m8#hEFDUa z8OOXwyDgh9plYN<4)4dGLch#euFk@~aiJ1=$|Yc`(^lpNth&qH0g7~kYu%0ysFhZV zz2bGS!$Rp4BT2fH?;bW+zuM--vs}w1VMm1}L~oL`v-I~S_>BRxwuY`9z|FiQn|U2^ zUNwH-=@c$~@<#@D42W?-21xagg<=eLWVW9754B;W`Y<6DvL`9NB0zw#pnj4AF7EKOZ^iqa1;tWSp3-am*qj zWA}%tetpsYp{P?m)RL1ml7uO-u=;+I4GCqcs?qxbN6?UnmrP`GGf>%XIwm1j(#)e3 zX=f#D7Ti~?p=P~^I4u%B6fL4B$v0A28UZId`wWOJ4 zxteN423MY=73{Ssht1cEox^#+9k|ip4l-DzAddt6nKJ#9iYFpx!hi|)50lHN^7o#R z(1k%!$P;l<;Nsb+2S<*mfOXWY3c)daV)cYmVzmgkJX9N#-Z~I;g5Agoq}HR=rC&D% z34LmRMOUGA^Nj8|xQ@vZwcRCKq_0Qd3b=qPJU&zwPzsg0B;UFRyyC#Q9ja{KiXQQk zFC=vvxqd+udpG8xbRRS*qZwE7%UDq&d95coD+DezK@o>zR#1`CYJ!*KIBmnm5y$O_ z<}?l2XTPpJ{rT+jtIrbu%Tv7;@b&1z|2c6k@j|aX*zbv>R));D*}|&3L*jDM!vt-* zf#hIP+e6%GNhMV5@k-tj(rGtk9y!iVl2X(0K$-MGRhcw|bJ0*uuhoH{cck;SiQ9td zC#FxalbIK~ifIn)Vw#P#HE;$J=FNJi{0jll=LSRnRV3H+7o$DTg4%W7G}vHs@2Kt@y|Lpu3LGx;eR3hTO+05U7eyA zHV-Gg6mS!rlO;kV%%J(iot^J(;# zYSWiudeAyk)1^PhuIvcN*bY~v-M>_aoxYGO5v~tqoYlk+t6?-E<27k6UwasxzHCLGE|BE4qu~nJ zqlm6DdclZCRZE#}N_WAaago$Fv@S99e4=rYitAEl^*W!YtfTso33a(xn%)?PnaRQu zJUv<aaOvl&yyDuHM1z1pOA5F;xOp_z zJW^h8Nw5f83Ge+N1>~|YIz}jA=9zR(HQmM8PYG&@8cew!8GArG;)>v1s*GVriU+Ue zLJ`29@{-#VOnn;mZWFj0jLQh(rk7jBN*^7%E7*BJF5~F7$)}4Cp-vh{(C!cPU!r5j zM?0UE+nva|I!M?~c*1et3=bn9An5mltsatm`^~TaQ$@S~|K10Ks=E4O*RgiA zE-MSch&(Xx%I7R2z^@{YR=tUqfKuyIO^c|61_tnChC9--%@&*`Vo5nKI2hc;*upks zdCReNxM&+zNR`lSRD&q+NBgtg=9q}X>Q&cOkUSE<8Liu9W$W6CsN7VaSv!|zcur~o zb2_;WQL(vF*ca1LUAQzS&`*OGuAB-z79v!*@lo^!0;{z(ECBCS*%tK_XjjR4CGh|U zrC#X@g~ktQdvr}V3qx{3ujbDO!#VFZJ7gEVZ0_J_O2UTIf-2@WX!>^}%Q&wc-|T`k zLgQGYg*nae!6s}|-97!Q_KIW7h90*Xj^7l^-U8NDoH*yu%4&#|_`LdV)c-1Drk6kL zPQT&ij+V`4MYJRunTrgeK^7y4_=xPu##m(494Ef+1Kq7j*n)=kC)LdckA0$8WqrzvG@P3V(ySL8d{8<+mR>m*#@Jw7V+f_e z^y8;59d~z-s*O)@OI$dpj6+9w9d(X1zP|P(HTV@Cmo=>TmjF&>+W1kbQxw7e3*9K? z?dBm~`;!<^Dymw`!p2roq{TN5ppSVt5uYDYym$)O(5>$FqmQ?g`8>UL37CuIu<^X- z@>aX^%)M#2*tQG0A0sOp=On#6|%X4u_34g{4>1kvSl691RS1z%o-Wap`0W z$9v#|;LMJC{GL{uT)2HAiJ#LTLdVn67@k+EGyU9F^U`GYaM`h_!E&l22Q?PmM%LNg zqLm_l`Nbg?i)!_ZZ!y=(Gbta=$gHHWISt-jbYzTUIyfmZQxsh=nyjtn?9STIiuDXk z_BD0PuKTF+RaHk_j6`OC>|Q^G`cnh{KtAsL>-6}asulKs-@*q7({{+8TUhjBJEBP+ zZ>-J!%Ozb2pJu>QO*8c1vgK>YN%MHeH$z3ha*>rCR&ipZI#;u>P#6r&w}$J~J*#VH zAgo;Cxe;kj-)d$2(|qJpTFhxlXow9DHdn^122Bs%^5;?N<4K2zJTv24RF_cR>gu1d zKSt{epJ+E^oH$y)dYc0!>eWNPTW2r5vNAD#DTtSSADK3|?Pte&mA(Bo@d_>Qmtz43 z@6YhAC79{Zq_|gMcSc#Gn%Th>h-fr~#)!im4k}9~m<6r!NhXxk|JB#J0eN>fC46(T zr(WqT$b&@sK=?~tA@|!jOYlN7uzq9M@g+8s)St78ZEPX-8Is^`khq7BzabW^m4qp_ z4!vaM)V9eFCByg9Hni90sxQ^;v!Z(0u8(-KYqA;_~Us7{5OI0?x+3L#1~zk0^f$pxeMY;%w-Ne>1f-Cl2UpNr8gR6wWJ zQB9SCw!lZa?~!P`mLsz9^+ z&c0!)6*1_|V9*F4dYbPnvr%vSFpw=Yskt;yM-{iP^=r1k!L^+9Wp-HK3(OtK)F{!@ z>jrCxsK+dumFG9lRd|X~tg;=&@Q_lFvl|wn%c{mSi4kmCj-<7K2{zFTAY00yGQl1? z2uj?wSfmlG+f?-0}!3 z=#OW9Z0Aa}R#(>l*abkqZ~rY$@xTAiVu5G)X1%I?Um0tKfif-s^95FjGx!xw|h~|%2 z@Sa-uGsqY2f1GI{3Bo0ubG z72c>p3vMaKi%yj;a&Zl+zDRdQ1|8J^hz^^5reFvoBLfm{2vc!oCBg91DJ!=9$^u!9 zF*Pt(B(EqpP#C^h&PbhIt9?K8oXobzciZU`nGSrDot<_8wK_3m(WV6Iz5wOYh|+y5 zZ|k}~lXXu)_zlw)A`8tYEp~s2S*xiyv=0;5;dJGkQuzzOsJ^A0(;U6e)Zi;`@(vq% z#iJyjee-+Q^!5K1qsFIG7|ahh=l_A!9r{z8viQQhYd#eS6eIHo=?yP83x?SGV1H7z&W$>W(1+qph>%Y{Pw8R1ONJv{ z3c4Fbr?u@{8VW|WlWqNTc%sGJmO2mjwbZV+PfDU9wZeLmn+uNv^7Cd+po?etz$uzP zQtaoh&Ku3pP;WOuSlmv&O}YN8Rri#mI4Y>oJkqwmcS^8WgEdez_8wK)xse`Hcp2aE za^g+1rVLB=#wfu6NXk3%nJu?(QNL*beDjEN>YTkL#)o=mvoWKXi%96a{mvknS$Q}Ewr^j)cljUhcove(XW zL>`Bc-92%U$2DKcFVcE*H7e-OwQnS;t&j9XUwtyxQR}p1-(QOS%Z1;yMW6maf|m6d zmv8V)I3KR9vO9455u`2`g_yjirjFc)Iq7`b8Xvd2i6xwxF|2;fdTe&K!`rA{_wrf^ z?$&5rA4+ZYTVl2NBq!$$p44I53zy1iA}6BoxV!>-qHWc%Q3$&hA!jJ9BwT>ZRj3xM zt!5)}CUl6ECyNSBmk}umvy8(nUE@Ki8d1wE2Rg|C7jN?sv1QfJe*h@Z z>d#J(|2vhDcu{uXXTZ_4-`-Ok#xA{SQhaI>n* zTqQJsfHy$|SjB7svFM7ZK%7AsK%Hpk66ETd3lEgF*X`!*UQne`hK-yjha*_t+GtF^ zVPO$U2=NCMh3N{ovUx>HRfH6wBFX@;vQDfHx^+gR9&^N2s3`84dnqHGb%1BD%FIp! z`7pRFs{64l6 z3=SW&sly3-qU^dV>^4%V6Ys(0IKf%vn)c9K{i%4v+t$R#+ z8#>b7MW0~_Os^!9ZZQw&W};23RD=3i4fn=wK5&tpzN{79-CAfj|8O>5N~}37h#0DR zkL)U>LlUX$Q2rjP_R`FXLwe-vD}{rDRm1edHl(%tmJmpz_VhoyD*{2ytfaDut%FSF z8$4c(*W`dc}Rk ztJ{N#zVQ(Ip(>-kB2P1^Ii9E4TsKWUD%Kah!8YO{WzJ3y`{cYH&+|@QDB2|83=qY$ z2XeRVdh>6eJ@U5fuG#n-K;es98EO8iX_v}}%94#=59XzL;p8jrNSbfuj8KG8E5}tG zyg<}!9ye%Vk1pPIe~gbbSDBvVgq@4R9Kvrhj%9ZHAi2tn-KtbnqDYxN*Akv-hK-^1#-~O8 zcz3s#=+#^7{aTN}&e}HUzP&VOa9F`a881JhTUBe4$ZrSN7B2JP*GVy~V!P4cY!lU? zR^#9)qVf{E(65@6T=8!A`n^}tm6LPa%DrO0i0jlhiGoS>fR#ZO!49s#zS}om^(&N|*D$h|l(+S(C325!odS;qD-MFfk(r zr<>+AnHk?dhU+^M8A0eaF(chaWNQqCn+m1qOu_cUVGrxGF{Wyq{q6oNBr+&;J2v$! zdPjU+KC0W@%LiOCm?;A?v?tV6Y-ttGu{2LicRE|QgpO;m@L7XZy^!PQb-;Oex`Zwl zW8obx+L@_dMP>N2){(K{+MrTf2|Y7Vwe8Jac~csUC0k&m0E#DP-VR8DA;(}G{u9^~ zml+w@t6LSwO`fcyTZ&bG566tvP{$>jLq(Xo#p|B=l3DU|^%0!xW-1pcfO_QOM)3;s z0cX)S65RELAiz=p(K`I;cPB_;!Kk3~Q(6Cc|DVmyVwd`S$ga$!Yb=JhtFh}tS`OPM zo7Kz{xt^8g2JjAbc^B46vw^XW06GwE!k{M6*g=w!t7>>JG+WEV+>LS&hvL#}P@C$AHE!y~J zn^c+OYk!9+83FnFmg3%9BXa8ejT?8{lWd#fDruK0S_Hhi_13j#7w%``Skbh)Cb5rT zuO|y@J>F`|5m{`i^d>zBQpP=# zohhR@f|ald9&AhLb}r&VipZIQ(3mK8N8y{uTdjMha|k@BZWM}+BYXfBV-~NsR0_<>gr{wx>cb`w&^-1Z^r44;qUmrd&ZnNsnz@Djw z+lTT>d&{UZs7_zFyh|!)00pk3eRf9e%@E6|s(eME!y2ux8o5RfH$tt=BgrDTcjcR| znh-f^G-0H?I|fEl1!|pnvp@Lqt)^g$53+Kf!NBCwf3A2ms;YD>S?U}M2|-P9OVh=$ zUuU(9R#Db7L8BEec~EMi!iM`29M2o-W9TefdSzP>A;U0zID#3Nw_{XzJKAo{_=<6O z9HGA~o#*=KM;stTqROt1MbjAMJYvqT3jZSyw%l~s$Is-Ec!Jf`cr%k2g2^a4$~f#0 zgA}LkdQHG$Zikch$a4bVo)sUn!w^Yx_3}J8zsQPG?!ez3%mU{fI#8}!uT#Df*%{|` zV@$Rb@d%|^mX_s4B0SqkWz-IUwx0A;6(k#O8IaT&S<+L+MH{<&~ zV#oVw9>RMfP+)``%Co55Eb7`ENf;3T3Vy2@)5T!ha^(6g1mkDue#V{}bmTZ&HwA%= z0i!HPwrCFxeT>*g0XCk2jh3TSM?9-Jj#r`Msb^XH|2TWO;o?BXN#ij~WDb*sV0dXx zpNE@lK|Rf^8lFw6f&CI5fc3=u3921n_l3#bW7p#SR$+l3-33y@5G>R{ox@i>TYp04 z(AT-8(-P|y>;m?8%9h_I*MYt4`G@51SorDgZLnDBXD68-u5Ti@rqPEghV@||yrnDU zJPQ;P42-Hu?10g2a1I0W)8@b~%u_Zy?cw0hzeBY5&}+9nLsQxrprODm3ut!-K$W#p zY5gljz$!*k*wP13%g)pfwROBvNA_bS*GRaBu5<-(17|1ws{Doc2+)4r%q{4KLPjn- zHaZT@u=1($jWIaNrwLX_Y?~1$TziB@-?R=0#Oa7+mh4annGN*=>H$04s0f72AB9)m zTXmFAO6E8DLg}k0*P7np8;U*mv3JX%!i|F4^?9mT&fM{l60GpSidw=Q5sRef?B7J2M=0|6&4oAm@U#87zfdDZLY5wPX5JZr5_Y?Y8{M! z)b(J~t6tNU>LeZh1q7^OYXHk}iA`Y>jgcJM{Vi?g0$mARWCEo|Di)6)XI-hHg1-UPKA zhpGz?7;$t9hK-+ldST>(S2igm@Z2y`aicqd!`l1he|4Up(*LjH+6dY|M!lZfmnDyu zeuMk_q9gRkRwW+I$S#&0A;YskSPgVRdRc9##(I2)7~4XuhpSq}n!H6FLLYKPo@h2| z?ili=F4d^jM_71M>YC%2vs0R+Ew+YD{WnH!^I9AeJ z^#JSzO}5LmcZ!s#;k{?USixs$1QBLiD}%Ggrx7i~(Ylq!?P%jRS)mW|j)W$eqF7Rx z+fnbASD`wp#N6E*2={I+6B0D);AQci_^rj2`er5)sKo^< z*IhU5cikigiaijlgH{=32NE>+>zATXeU$TmiW-jqgNIb;cAJ_6h}#{4Vk&++nW|pZ z-Y-hrcaq03emG%&Ybj{R8?J6_xR3T4BX$jajqRpWxc5cQGxTdi4AO<9>h2Z>uBmPN z5xA{sL<%Ww3yMcsYh;8niS8E93{E|U(J=4DBk zr}X3As%!BrEy%G?7!!PZS46#*IWM-+NMz3wYe4lPbE-KXzt-HZ?tFj&*=qSJ*rb2R zJs>!Yqt~ORq9X%BOuc#^^B`hU_IJ{F zW(?uL>ZeK`p91vGNnb!HRdc1w?jx`0a&|!QGt%@hYg1T^hqtthf3G8~VM_6grxY81 zE-Wg}&oY8zSUz?^JX5Vh;VU>B&ubD72^Yx>k4C70&%D?zfIoL$9V6@SIEV*_t=1VP z9#4lbh{q3tAjd%L5z$=5Xq4hTI#n$UFXQtT z(>M0QO$}96+vgu4KuRa^iy({ z>MHIy%7*~!xq?QjbofT6G!nS+kUh_8(vR10j)?3@^@E{Lxk}exi%8>IcK8X*1!M~_ zxCi&xgWt!>;u?Qp#`JR4_*gu0I*0g9Sq81TstTO5Y8){v)0|F)8=&Z0QN6uDR>AoN zF_#Y!Vlb<46%IBwB1dE_jhWqo3~{7oGF5y_12F9my~)6udL6joN@{w~NQxL0OG_$> zxwBl2ZyB&2-n9i=b}g31rLk7$czu$Q>t0E23K*`I|8e>*8N(=VYt$Xd2Pl|M0v#Lc)N?H zwB+C`Rzf^ab~CoSMCnH0&95)JcyUo*l(i@2lbX~z+9+XRF`s1i3XS6^Ny}BQ`hTr8 zJUhYC6|C0#T&EthH?bv}7a#a>K9P8Vhhzok^x!B9-8NMM6Q!5cClvPHvb(u!8CnGT z2t*j)_ZBfT$45XP!ON94-l5=xTY9&Z-pO<;kP}b|wMPYO%eJ1xP@H4>`2O3l*D4gL zY7#TOY0kQL#JkE^eH1FrX5-PAvD|0N1TpHEFniW?qjU`wTtOIa$O>Y)_-`5}hF|v9 zAF3E6HJAZXE;U2bA7fFyUA-=dl&MMFYJ?lIypAV~6QE=ifcp&O&S8L{ee6S9mro+Xjb;2K?pUavl_pPtW&B`Bnbbd> z7QdJ|;%!5xRGGb&aI&xdWhtRihCXFl|3Jc+i1J_GdjF zM188G`Q6PqGdX6T)a3G&N@-mX=&gM_A2faNzDDVAzJqJo>+jw<$tTnOs`FqGg6nUC z=Zf86q{wbv75Yc^(#}+}iOa(9-PpCa@gE z#Y-8v9qbcP2hY{Lj^!rUxzo)AbC?VFc{Au&V;j#MRlHVNeZ$9Ed)>pEYdVh}(brhr zE8pDYQv6h_%*{IXqJCcEce%>Zq0Cs5&YvONa}Jo^U}^7@(s$Nlj;*^FB>f;r@nO-I z(flJDW#_LPJCsJ}XbV=m!`=3ojJ-NWr1uC6knbF;UcWM`O=VHwx^&sc^QUFy?Wy3h zr~Usf3$OR?zuk8X{80FFb@5Eohsr!4Hf1Xb-`jGP+H?~H%H`v2>*oV90<)rxPmQ|U zU)_j`D5NtKp=MN}a5c!B@!BiSHYl<{rvt`(y#%kwlpJx2F=?Kz(dE+8E2%Y<-0Cz> zTnp8meApD?t3JmCLw;E;6YPEutc#ZiWqnxi=R8v7a2X+ef+*3GmW++#$$fb9I*%}^B^NFDPp-bUMwK|ONC4Uq81cs4W2ANnmP5(psL|M z54a#)!3|v&b-grT`Py5j-Qw{%fmuee4mDS2fSg;dwV?ZKxj4@?`<%)%=4Nu!?PnKm zKP!zbk*WBAU%l!b$z>#M&}rS4n=5LeoV77ye3Ro6{>Y4n?7Q7S zY5DseG4E+)M*I8X**Np;*ji5F^A8sOFUmeXuG~$MyyP~=vFToT zFl@ik$|pbGyrv@X{_ujh7_H?DEEygk@~)I@yTmGuew!|5gZ^+wBv zfW_Gv=^LAS$XOi{EU#CA^@3f_XS@xat--$3(wVNiA84@46kqVDW$*(pDv%?cg6v|9 zeTES-v|$UY4sw2io0k|5ffn|!_LYBT*GJp`+5mxCr*oE?FW$BH*hF)E?GNIAQ>3fA z^Beb6LYgQbYczGZY`|2sV%HWFAXfGk6-^^H46p$(W;;hDt${HEt+Ffok0TVV!D8^Y zpv1b)-PKXT&-M}=wOkv;1fhN(EFh4zt7;0wyWM3&`AfC+>#556yHS}Lhr5_`w*wX) z)b1kS?^UN5V}b|AgB_7mgktJfY3}qP%3rio6Gl(zfjZaoy5*dEP%n$5FFo~EGVxcG z@h@En9vZ6~Z0yHaHxxi!UyJeKx^_sYAw^9X2by9r-8)cIH2dK+Okc!>!ZVmdnO$^I ztAk14(SEDlzJ$Ch4eB40N7fL6az?y}H74uCM!?T7qU|3tr$y7!5^T^qyFpPZ^JCOc zGPSf}C?=?KS;rRczljTFcsqDrVaFCgtr9GccGjgAkEjMBWBz)^bstEsZX)Q72A`My z*yf=@=L_%(wQ*ZE^rAQ-r&|U0oiwj*5pN?m>i6*&mmOKnA8okB;vJ|CHMWL@oY1qm#d3o1xG^lZ2E*#XB~o;&eGrJOavE!{m^Pqq)MT#nz@C zO;?K4c>M%*q|z)3NW9)6Iq1K2CzUPF99ERSpSYMLi1d6At51(6&IW8%kQeG6QcRo+ zIEW7V>~R0iZ!@p!;a4x*Nr1%d%gk#v3b&QNMZSEi@L`C)kBd@Llh6B}duWo!5w)C{ zibv=1%)0*92e_G*#)FL4Hd^zF+wfB8vhb$Id%HK;j;7XMjJm3~b3cznop{tT&+Dem zmzYB|@m8|DjpdmEqW5b>n8dz(KHU z;x9Pa{KVsqYNq=7(KWd`uRQgW)mDc5n19Wh;%$fYsBwFlOy&vjxeB8=(p6yoLS zMosCcLST_nmODrA4Jo6+!6Kmf78}4Zf0`5`QuD=;DFUEPzFeeUh1mA^mRdmF2|qqL`w7S4Vi$>z}O>J$7~6hC0_4^?}ggVb`43=+jK7UMYSn?0_YvU@9sQX zzTy2X-gIkv0&m!#C_|)weMgE7hN6m|u{RtaixuNn-<+R(=!f|SvYqa0h4%BTTzW4V zv6?ki_`(}nC}g#0$;W2Lczcp&4zPrDkLs6JL%Cqb4{$wp=@PHQ{uLYJ_E@?RW%Gu1 zgTZ14N6_)sn1B%0b^WBbq{p^=yQ(DI=AXC{ZQjJ`AV`1mrs{`UJ4T1etAUj4;wi_|p%mReI-;^N!!nuYERJIjlOnTz4*_D1qve6Vj@nbht-#uIAt!`;ooxp~OPMAmYl~7B({f z7_xj}!?UHRx{jgym%_=FhpSNk(`ph^-kki{`a$gLiXI0 z(?H&~({B(f1G|jDz^?9OwFJth?SMD}qi$TYVgjmZ+*(%gW|(@v9YQe{NxKo7$m{Io zcVwKE@_|i6v%O&WVMVg5P@Mxd_y$tte*f1HYw?471?6T9L5h_bx zhg&IJpuTpZ9ERZuTR{dCs#-`#Lhy5iba^8i-uh4e#FjxX3^Nq@f^XqQ+kFl@bB^FC zmK~uWnh&y|28Ps}L_lfyt@PGFIAACVgy>^|m&C?b)h%M&D29b)bsZor^3!ze=Mw_*aMK(Gw8D1NvvUP2d60DT8-P(f7*2C+z(Q>hpfK2P zH!Qt>ypD5zO^)PbJt{K3P(KjTwJ|nDE0HZj7_IA2 zGAO2%k@@w+p^A}?qU^)Qvo7cWW!7+}IRs@LD@oL+n|-aPK;Ve!i`U`h4AGWEoQ02q zZv9!*J4;r?i3VmJTBX|o+4;(p#nHZuL52)Ckr0ScJ>5%5mnQ^PK@v)TSSL5M(Q`9e z#tcf9ETk_wIs#WXk(|>=LE}4BN^^Yd^G zuXm&ycfIv1mn!If`i1!868%D|w}H19cE}v`YA`&pFCbe`?})9;YKBKZZz$gvGd@1>$n}3!E|SgTbSBwoTVRngIkFI z<>?wKG9$aHP5HsMy0SV$7FD$%qF@MP*|5S&528@RcotSs5oyGg0+aKqsV)Rc)mERv&<1xicXBnhiOTk8b+yKDpp;tw7!y zI$v@d5j=B~^y%09+NGd<53b(UTCip&j~+nwkaId5UU_hE7dU-f71p@~8$@-34Pczg zIPxj_@)ZcsUX08t0AhC*+6G}}7!<&p-i0PY1KxHA`901VWz^Gd1kh6y;MmC!+F+E4 z^?+q#Lj~H%5Wu$UZkm+E7M9eY9?a#88k%g#!)ysn*z^Ixn%djufnS=`rW zXy0t%0i-F`@Ydn-7VivmdU1vuh7!+p!=MBnz^EmM$x)@Jxw_?ag;-Nz1}zv5XK&Qe zWd(3`i-4Bc7*G>lztz|p#xk{RA&02duQj$NZ>5zBd>@5Qy!$6bN@3)i2GC4_QNO`bMS9ZXmHz~tH_dm z19VrgN--F;3T42a<6v(FWm87(617J<0s3*2wRf%>*sF&3u8R#n`Gaw;CQ5z4secqV zwR!ccTRJ1l_kYFCeb+c?31|K)Xq;Ah^S(cBxVRttU2`WDnygO_?7H3;u&V8C^62FA z@Q!Rn8RL(}`gYJ+;H<@=JOgFfitw;1N$SAgNmi z(4pSeDSrw+dG#9fSkmdPF>+JtPj|t9Y1icZnOz3TDQ=U|N2RU4DF+vCTq9+V4;yA- zLqag#>T^|MPxZM}PfD;B61sOtnCr4v652d+nBCr9@V?G&9sc;bxV@Y6f2yxbGecb` zbsT%y-uCHWsN3MT&NV&00+)#3rnr2zI5F9+XbQPhe=Naw!{6k;E$1j;ZY|OFm6EOC z@LzYH!P?&Ae(N4As-zE(1t}QZed+HXvj)l_5Mia z4F~r4!1RXPteFHhc&7*Kfg2#sD9n~$YbOXTD;LO@|Wt@;yicnR<{{` z^YYYhV&h+W?s}MHjil{e&$L=wHfE>VE=VQ3Qu2S910(ifCsuIAp=Kvx?IYm+Drum3vg+QeR zuw6%0Nw!pNlxStBma?TLYq0G{*X8w7V;5@uXuFG#q(9Ej;5qSMogx>YBL++taZr0i zI@$)kswDfe+rqgR_9TsAsuIN!yIiY4Gz1ukjnM09NvK6tefgS#Ffg;LgO6jjA%TXL z&5N)>%SVPKU?`?))g@)lp+$#x!9ojy&mEQ}a+0{kV8xn%lm?_k9_;?vI9y18qU4Ye zVHmhlH+0qLW=5(yf|pv|()-x%?1BQSEpHWf`b?ua%L)9nyZpf zQ`9pRMh1EEB`jo|leH>v#&se`c4`1yDSzhn_xGFoUO%WGIknNd_EzfjP3tcK%_Hnv zZ9yT&`^M1vi&_IV12o^z=N)qiulH%UoILvc(cM=DhxYnfv;{gOst8X7!HYKDJ*lcL zdmdn`_V%XC7r*b@lsG$_cc86POPU^gy?5xqpz?Xo!7FRLXXNqh4fhfs%4xhoHCz~Q z8%6C>OFA_lKOpzj*XMU%Dhc=L8`NOrK%uB&GwtPxz3+dUG;R1wdgAMCcsTs$_18AL zv97Prh|E1^4{`_}HWfer2corb@qNMksjVZZ?rWsTdT;Cz^H@+yPB5p*vV6d9T%nM zt;woQGfy;rB;2?%@;2CV*H0lS)bVq#bZf&;_2(yzKQt{5Am)!w5FJ~ejQfzMqip2d zO}%1 z)WoC;*y|3SrgDDTf5>?Iw&G}kX!h#k*&nUeK7-F5chh>(f4be?a3t!4#Wd8)=hv_6 zH!q0V+H>DNb#s(W{0E}6T>gIXwbMV4bPbVRR0rQB?0hE7IcVcUA2q|+JDMtaqmg&V zCKT>c&Dy^G^x`gjPf49_ZX02LvV36CG4TG^t_$nx%rgFleBJqRr1L!a_xb8Qk%&)E z(KlOznIES}HV5w(tnS_U%i;Cm8z$37V6PB@+qw3(;nX;}>)r?Io_y08eRLs5InU|) zY3%d~w1Dzk_r1eFdV<(H{(AAX?TLY{`A-)eNcmB-_H^^_x=_pAjT($mznSX?$p{v%IqFK=}5D5KVopt>0a1N(=^G{`kier zzCUb2Sn!Vj1nu5Tx_NNOw<&;tCvI z4=fbwy8UKMdk4C;To;ZsguU z3)ueo*2Xz=_QRI!t?{L;JBPHD?tb-) z?fEEw^_0Qu7TQj21SwO+b9m#MD??3B``?SdGQN|VG9Gjde&BgBK(W5rOsYN+ODagt4pox(;ZSN3~s-OKG#J`>m0vPG9`%Viz)j5w2P9J}2 z<#YA@m(=(l8Q1gr--g}l`0Lt;)&9EA>bq)h#=pc3FWp|{Qk`_Ld!I%#>$z3kzP=Nc z@#6ez>`fb+zCAoK^4ELzPK~zA-(bm_)o!VP(7R_nUP~7BM)JqFU(i$$VzXMktFupr+)1srv0$;_Fp@Z)*f`fMPpp0Ti))+ z;5C`2w_Z9tzYSgZZ8tW{c*&?yIJtN!`d-$fjtB=I1?($(@9C2&ejmCsWG&4HQ3Ti; z>)~z;evP$)zNXaPBiZpccYg6myZd|7hl`o*Ucl*L- z8s)JYq~Fcl9P&;a^Y?mt66}sWjgdY74A~(`dT)WNT1>EjyDk zjBo}+E<*pz4_Kk(L1Uj0sN?=Tq?|+^`DtQ|qfM1I%HDc5{=fewkxid7>P78)T<~)?5AQ2wn(OVj+Xs|6%tDA z;4-k<-Uf!pINZGrW-Aj?AuH~@6riI19b%Bdfl2w2#Hhx#(h|CmUu?`FXO|b;5&i=) zyLhLHbzRSTe2{q^8RteBR0)N1nLMK!H%(EjaM-xhkDAj4g|S_*NDq-U`#@)X;y@ZP zm8W|G;jL`*R7>kd*}k>|_5t1vGmwBnz~=VV4Dd88?v2EG0Kg^Zjn}pF2`<-iM;(0t zWM6-Ft`d>b)|W=;tdk+8j2e2m_EQV{K8_rn=8Jd9l3e{V9>elW>h2QO%nm}M@*45` zCGST8+BfPd41(fiNX2|VNfxDA zOW{Bi3-benNggtTmMEJDY!(r3y|O1B2oIdl5j|~x`YEF)7<$L1$)Q%x6F20zZB`=w zFdPZHJGK%tK*7LLx#wW)tm@m|;yACjkuY~u=%~eul{cuAgKr84} z*FCWE@CLEc!4X@pfcfCRaom767i)uzI4k=+J&WFYSGOtzKjg^(k&pBD`^))(Ww9u7 z&t$ke`b;f6LLC5=P&Hz0eEg1&NW0NhQLT2P_n78&l$i_GX8{b>i%R#)y1bqs+77E4 zQLT|k%2Tvx>rhH9ZPUvAksjmW_#gJu% zXBuEabO1oo=#sr+V}N1=aa51l+=N%)U*bKl10>0&3%BVF04I?RDo;=syJ^t~TxGoG z2=)VBtfHsO!gyc|N3eH0=Kqq;v_G=)=l<7j_5F%y{~!Q)bsb1s0sPu%{0~I^)$?02 z_g+=gn-z^vin6ArfJ=k`(ra)<8w8VBO(6y_A@fh0j647hO1u7a#;goU0lXOKi27r^ z{(18sL~#XE{4==c-v;&-I$`BQ|D_k@sd1hdX;rq9*pi(uK;g0niOhRE6XO1RVCnm5 zgC|7@lJf8OXn{7Bn>-9;2dYPtAVOaw?5Dp$3O~Qbp^MN?J^{Mr80{^d^wHAB-giWd z(QE8IEV6vfv3K}2Ew6(XyzNz9dljqM`J-=*Inb=HFQF^J;ISy`INav^rrDygDr)R8 zG~Ak}TFo$9h^A!`M1FlzK!p_9u?!03!D6q5E!W194S;Y|ili~18{}6$X+zMdcBI~s zgkHd90#gaprM2|co1nYEtAF&yn}$eVGV)Szl$-{Y0vj}clIpETFKW}7$9WdmaFs>a zUo@*{X&$W(!eLw94uZD}`jz{YT;}4^b8qdA`CYPsn=EhaT3+`DMc&fs`%N?IAIN`t TGfeU~2>h0s^OgqwdHnwYQTDF7 literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-19-Micron Presentation Slide.jpg b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-19-Micron Presentation Slide.jpg new file mode 100644 index 0000000000000000000000000000000000000000..283af9ed66144a6d9a1884d861c4bf9025ef50bc GIT binary patch literal 121273 zcmbrl30xD`7C#;)3$ZY87ptMWxkBUv1Ub&H6ic0&4r-`>*Y1B*|oE?mf%*e9!mZ zJiq?@HpPh(6B|RZu%J*Z;D5^Vo0KSuovrO8TN}GclP1~Q+ff}bM~BIi9j3cXoq~C| z&z|MsK69q$oWQxB-t&BB&ZIBE=LH2bnatU97cE;DvNUi$GX&kl!rtEAVX}jpqoZ4h z*G#XF|JVOL@1Qu_*$mrQSXp>eES)W^oGqSrQD(z9Z7j(DDC7TGSXx=z*iN#ur%r}1 zv^h~MEv&38t*vZqtgYeO@56bDwX=QZqef{RraRxIqY{9}si=(1hF)P@y9Dz_IUX`?3vLQ7s{VkbXv3bkZZSQX1 zv2%A`e!-r?BBG*FSyjEi=736FuhHrn8uf;gr;Kf<&$OTYsPnTf(}jzdE_e423=UoU z;`)u@uSf3Oy*K*JxAz}B`thg7PoDn!=U;wB^Rl2=nWqK+nkV*`c{#(pEUm4rtS6y) zSy;Xchn2InjaRU(O9XFHO6FAWkQ_TKvixZ4d3&Gv{2yr2tUjvi9Ol=)kI>Y}nf*VT zSnmI)nZ20U%Xtk`9IPzBc~;Jp<&@56>gSZZb=N<<&RG6zM!*i*&VP+fdrtAS{p^qP zyyui}Z@hh}Y5T*dTXZ#X)K1RA5-s~!80FA?nL3+FSJ&8JyzrL?Khc7P@hNHor|{rg z|Lc$v^g7E|4hp}ogvYg1^7#W!=o%wGJzJh_T|@YQqw1$1ioaz$;j`Uur>Ep^H~!36b^;1!Hg5n4M(GYaVU-C z9ML@g^9aB??A2xcrOFN%4zhz}y_o+n@ukbNUyUV{}lPa^0miCB^h%af3fF!j64yp&UqPs zWGo`vX#R7eFD74*?OH+5$>IKY51jbmPaY^h9x&BoJmFv5J>hoJ^%P~IY$B-t$J5>F zAthqyCNK_!i$-e-pm=KQvGoHKt>(N|=LdG54-S@APh=J0lv`bi8UH%-|A;DxsNg7y zx`aoBqg#`KH6CSTLO3APap>x@aZ5p5)?>(fCbv2pi2D3ah;AkPi^-55X1DnlsX}-3;4TduOA6XXerLKng;<7qx@&(;wn@z>> z!EE?QXUU6WQGBr%DNzF!fN#U#CHQso%>D2|N=-dxgvnQ8>Y7|af;?;1Bhi68^`jqL_vnbV5Gb_whV4$#zrYYS9koykN|aB^O(UG(pT&Q!yli_coO_C zH~fcd{LGLY%c2Be%0w^O3rs5OrNA4MoqxJ!=NJEh{$G43JMjxWZ1z*?zUP#fE9+RR zFG>%8O$5iW67x)MT6OkR2w7siAKesyWpjC8EsJ0hGn6{yh3gtJEmB?v6`w+0<}PdC z5z-lPL^3bj9^H;Y-hxtN_TKe+tRo{3&ty>$9-+G6Ne$-EbuThIT&*O-(+NVp&I$vA z!hK`#x8K>|ef4W{8 zeZu(k78~CH>=-j5wCbE9xEKRag>DL@*IB>JA~JX(e`3&R$dAn^pfyo^U!Iq!EUzT7;sM=al0@=fQ&4BY z0XLwVo1rZ(mXJicz@vimF=K{EE!cUso8B_drcVJl`P~={eniwLJbfK z48j}*H%ZALYfOG%4Il;-gk%(;OvWw*`9MXC0&|!}5E5QLoCh)h#gW(w)f0HsIwm|zX4p2>qrnCwLJt^k&j4JQ+W)C310fsR1~IE>RaFtYRTjqVDhwq+9OCk|L#qB$yi!L$|ChrbsKQO9TLQPy@sB z)cY|hc|k%4d@;8=+X2v&yto0OHjr+ToRCy~gouLpL1T1< z$MQrR0)PH4?$33$18~e5!u%|Z5po8~7z9~vP7IEGM&1BD96V)iYs?{Neg_tk2SfFN zZ@Ep-bTE&;}`*sfqap0kVQ!z&XG8J6^(}-Nf!|qzhGw2zFDI2c;){osv+2*Y=Ym( zK8zg3OMIC?Lkmb=a3EmnvdvG6o>MAvQl3-Rw?C&m-q?0`>vu;cGVCJBC4BUII`;&0 z61XCdWXNQS1YiQE#6KY%4sn6X@kB_#izu>cqsaer`n-5=q$sJ$iIbRJ2ze4MN%RMn zAaGI|dC+WANs29@yaRj>kN^xt=^qbh#K9mo0iOa;zCvS9o^S&M2>`S+c@>~4W=WDz zvN8rkwUGd#A`f?fk?SHLfNSjI2(DlcM+$>QRR+TC9b6vo`$Ao?h-2e2rvmd?63He2 z%*@x7?kJ;_{d}@m^g3w!)c@du#X0AT44r4vzz9SZt5}xU?OwjH#ZA4lPAm{ zgW3od+b$dI!^#N5}bh(n%tBMUex)Lb)ZH;s|SC9$Hh- zj2Qen6oPEHQ?NaJ{DQ1>k!5Y%#rzQh631NTP^klWd*QEfUx6d6VmWeildm)j5IO&m zK7UOojHiMU$AprLA{)>HDv1fOgse<86Pyx=7$5?10pBDZlr>dZ8{fD!N>aST? zTWS#*5hzfg40qs;5FZftsCCI9>p$uKi6VuH2H820VgeM_iEu=%P(YAN4pfpPY??!sOe$p095V#@1_0iY^$t}K zgk*3NEg5239V#2a;Uw(oJmD;GJXu}jMDNsz0Is6?U^qk~jK8igyt5j51jv(c7%Dsn z`j85szA&2?D$%Pl0P?2!m=D2~PH=ePlTjb-K+;Kbl%l)Zyy7|Jd^6_wBH@t7M)hbp z$r8-9|Cvc993NPjixkY{)=2^?gfj$-4c#S*=OAxS!trN}G<&I%q+LmLXD9+(c!29Yrx z4ip^WX|h>AbxV{@a9}iN=w;Q_WM7q&??)yN3eWkRMG%o4q*0s@oN^;kH#z`_Pf*Zv zyD%fr1oHk;bLN1rm=t9VJZe0Xkl*ZNb-jwDE^okal51Ozi)T=WY|$jmcsCv^$XR9) z38p82pkbPmgU9mtmQqJu(%)-#h1n$(Ih056x{e%tHb5gFxGlsXFrEolA3_SK&mSBc zP(bgH3TVZuzW`u`dNy_D8F+GIgxS}q>$d^$2b)*FXyr6@ls*>bfV2E?`KweCp)~R$ zI@Y5ogagDX@CcIDke}D1J3&Zs!;$($z5*lE*^Ir2tceICrEPN_0@ub0e9hqC3uB== zdQyS+O8dg=F*`a4A}~RKGHEt0GdNAK2?;pD$OC;1F&hAs2c%NshZn*qqNp5m`va%= z{1Ipyptpl3NJ%E(34~5#5rLuxJYXlQH~YmQDqL&wg(*RIz>DS3DfL+U3)3PKgLp;8 z9`yI>60lndV)5p53?fSvut*}@7fJsp6D>z4WLG1J$n{JLPsO2I zY-agF-@=Dne`#>!V$k8SM7k4vIf#lw9-*2<7F95e*~2Q_k+PVG=zr19PfT|L;Yg-1 zcPFXnN0IY|YZ2pyd_;DEJc3w6a)p3I0wgAQp-;?aLj(;0AGw?WhC;^A20w8qj$pC^ zU--w;g%HHDr6BE@qfUvY2GrVtLuVqKL(zg$q~z1k6U|MYr}+dAG^lEztK&e7krbo4 ztT#wq#z&ci>NZmL$;dSm8Z_~aCm;UbrOc^4olta8D|vJrEGJZ5X|@&1nvC~9rTeRS zi(ZqfJfxvfSilQA2}F}k000mS3VIZDZ#1$O2h=k_Ne^5;$-IIxF)gGF_xHm^=4y;g z0qhhfHEFuKl9?^R=R?++*@^(9aHNsonG@p$EukM2@m$h^-2PHIN)`%^Z;rhN0d8jB zK-kHY0nf5E8jDzfwO$x9nLXi{(pZD4(r%<*aQodLo&b}ikZ{PE<^v)CmYmQ)dp$K; zD0AjWl;%M5_;QAHCJ*$h@h;C26;!&W2UU%Uc>J%fIsQTKbIOm)VS&wDk9^;nl1~GY zO$S^+7skclg^+M0Fi~Kzkcog}FiEmzzKD&97~z|VfD?o&vJ)!^2rU2zRCXA&vE$X! z_~Nw}K)yv>EW)-!_jiq#OIYmD0_a)33{4FgudvV&Zi^_jC#faDKO z=`X90U~)$oo>^5*L9v2}gEgQ!n2v5TmkUdNe;;XG8yF4b3w&ceV$KvaxZH88lNYlv zB`_EM=J3gJu^cml<@fUk#>b2hP(=NM6!G$M_JcwiA^x2Vx6CJqYzR z=A7a1dIGTC8boA8NX6Ama{!O2&u4;46i9DL&TwlG$ddy7kp64arBiLi4n)aJ`2ARv zI7lHUBpeVH02Fly^DMy=>=Ysxo&!B7f+|GUr^5i}L-Cjl)8#;LAoA7+01j)jFOY&n z4X~0W_!||rI#4u}#tA+;K?rMRwf+{h@8=T_CmNwj8gT1BIWKPYXJ518Hi58Egan`t z1ndY=DdR6l-Z8Q0g$g?y6EaXBU8A!^)tUnMqTGV)sl1;!gA~-k>Y0P)$ajh`2^q3b!%zd2&iW4|2-GRTO>nIr2ER>NrKn4B@ zV}SZ&12NkIaW6fT*QJH3-MRq$l*}KgDHz$0+N34aJDFsoloYrCt7!o4*OHuqNe>ZG z{NChua9D8h))>2(N)3;gC%VuV2P(_?%wXu{y$kvhOnzobBYf^oFqUoKi)NqznGLeu zya>}pj_*a-lS<} z5QQ#h^5SZ01MavLCdB#~O0Mh&fjmdK=6I4!Ab0L?*$+`g7 zA%cl2arjG;h(s4Jlz_&GB_Jekchxi9fnw=Km1N zZZ9nR9hA@o%n}~Vld=G<)seG@#SpW*VF}s}wDwjIS-Aauu>MaXfVs0kUbdJY7Xvv8 z0d8JbA+;O3?e)xdQp5!YhgxL~Dq9rNVdJd8cIW3~)a4_eB5B=B6gRGF{-F2BkIR#u zQ^t#7DC6xfBLcdR7aJp`;6}tc&D$d=F!l#@i*_|ApMUPQOl|=9WN{)>I>0v|*p_ak zLeYxQ2*B6|x*zbjz)6bG1x1Io#Myl1ciK~Hq4#2)gIkOa~h>bgjM#LX;%Krv5R z1;hd&B8p8UHJD&x(2TyQl@a5qO!l$lv=igUDJil6Kp#n6Eo2C7`dEPs=*6Uji*G){ zXiZVOMS{t)y4mHrw6-an#t2k2XU#0g_tPdlQwaABH?W@>?o728dL^~rU+FBWTB+C` zabJ>E^HDx;cH8Uw#iyrm*5;P-XLc7XSIA3`W3!a`_>X(J8fl}7(TnqpTLXPB993On zMPKfbE#10yKy|t6tND0&ymv!dWB0X;85g2c1(|&3SfW5ybCpL3A{yx#w$&9{WghZE z*@g00v<7gCFFgdnZ-Rm66tU%iDPXB!NHDXw|3aDRc%_%ire-WQ_bq064z@tX0UaAf z93)`}!0!lnPSDC)VgZ$S_V4>nFQ2HyD+%ZX?LIW^+5ieA)MKPj-W3k~?DWl}U~qw3 z6X+-4Jb#A_OvASTB!?SjXMiJ}fkVIvD9GrO7d}IZJcJH%Ff=_VL`h8y=z@p*dOc>! zTnK^RuAAmkUBVwxMYbyit$?b5Vbw@!r{0k@NXv+MOVBeBOcKFtYfS^_U9if4zTf~@ zNl>)wp_AT^)$!K1D*=Eh4JiP?nF%JhhV?Uu-~`KDdW$2xK@jbzRuXnlY|ruFUaZ^^ z^alzn2})++3|6bPX8vO7j%U8I8Xe$kPCpy125m>E1*XlXnY0)+S>zSw)&STUZio)D zthJ#C21VBk7|p8*pZnas^H+`R)4pw2`^0yB>Cz3}4fbORLVcvTd+v;?OPzw;UBVps zO~ymkrCk0)z2yk3KCJS(5s!<5^-vvkL0gBsp404cDO+jOuBKYRfuc*I&l{l@S|@A4nt zxDMzOLM~Or_(PeR{mblmA3mH;YOOl;k&ZnJ{uK;?^nlT|08>5)X4^t9{wtRc3lps@ zs~HinXCXX<*oJ}v7UpJvkQ^R{5@r}J2*Qqra7XnaSn~$qUl2JJJ$2dNI*NA^Y~E@uA9WRb0WNSWX8_;8%{=0?h`;3PwuM z(#ijVX!Dpd5-H$SM88tNo&-Q@l0=_g z$nU3aGO!CQ592lqMH+UUu<+bK_-Mmk#YoBB$MYf&Eo^FSpke+UO@-%HMLU1SU$wr8 zZ^3arY$M-rsFk8MEu?P4E3=mJh)S$RGtCEz-_P|}?%g$V>LA2H_<(0x;|;}Stg|gF za9%xQC-rEEv*aHK3b+3BQmaX15R@#s84Z?$OyKXGB z+{J(RRR1VuxIk7IwkuIEth>o(Md3~vPc`_MU_{08FFAgBjCXRENR8*q+G$i3S5zP= z+K&y;3io^w3?Sw3>E6-r_%sh!lWI`-ZN(;t1g@7Zo2QvRwMJ0^e6PX9Z1 zA12k_j2Z5YPO7H`4u|!xkAiw%g&mYT?89pG!EAAv1Hu`=ws-JgK0wU$S#;BK0KDl$ z@Ji^5ZKWj;1aKBY4OKoAeefwnL2E2wfr>wo$f)qU18@ZlL^Q|mLQaCEGGJkVBsy#X zb{o&+xHfyXn`70Lb^&1SA%Emh?%h#<|g%<}*sg{XcA3>Q9##IM&8Vls&KJcB?~{SnFpBHAz&46tM6h}q(y%YXK|v<+B{POH5^$Xih_6WEmm6U zy@8_LMz5N^*U)x^Drlp1Jn`4FgPOhDf)Dn7?D9xlmYEpI=%#)f7xnqU!A)SDoH(N1 z6DC>5AIU+mZvp58qx``rEpzZzvEpE$4tap7Tvl*(Zr|$t&BCgk*LJ^d+r&QS-p&51 zGDO2!&uQ}>u>8U!s$Un=lyH-85#w*QWyR?d!KKj>VMw1dmM7H8?euBh0meLGjmEn! zl_l41*RvsYoT?hvD*CWaqeW~1UegrVX5nVw#ygkCuJo>?$;yr9hQv+hH4d0B1ADyu zSOS5DB5DMd909c&#1WQevLZ&>PDfejdH@ z$9afr*f4x$8}a1LeO7NQJPBd0`0(4yZ(n(o)Q<{zNI-r6-ZsFj#KNb2Bm{H=8`(D4 z#g}TDoW_96j%jX&C+)_hx|^{lbC^&y_`htqt*n_7ae8mthPT>gt@W>xemWd14%@}Y zw3}MrgjDF>C7=$pAB$o6G5E$?iFp@z5a>EDgx4Ui1J_zeZLA=Jo4QHT=k1owhd{5j zkwouB0m&mYh#^Db9Li27{}x+#KtMw2b}r|7YrV^ec!xY9gTrDH#SUrX#oQ9ID2|X* zdhyI2I(D1UZ$V-kY?-~qA80hWBELf{7Tb;A(gAXGr3p*2y95aXT&ybh?)G{BoM3+M z_d$Kc4#*jZX!RkKSa;}<4$#x3*&ZRdkQUttB9j7x`oZBQ|JUwr0ngVF2Vc9JR7ZS( z+i6CYlxUdGj5YceL_XpO$-E^criJ`IVbR9Kg@8hxdg^U_5IhHw4F6_Bm}{b@VPA*C zxGQ}3@l0#7vlu6dGgBP=+25xKq`yZV^KA5LvhzQ>ans7oky=zkZP(k-A+CvSi0j{m ziBYZs7!YfX*(F3IBg2?4PfDF5pbp-X=gWiD9|x=|-yXW98U5?Bg5?v&YJ57RcU9j^ z7CGM+q;9XL`3;XV1z$F}%{?F_1lI^P04OrerAP7J8WWQ!CLCOtOUQAWcq4#T ztJgu<&6OtKINb4O>}kMEeqxVtqdu%ZUvr`GG{5k{dga*Rp#_XC@5MZi1fPb&XNF{g z=f7ji*EEUqE@AF$#bN1L&N`7-@vzbQjlmtvvwJS3*Sv2C%llQ_nk~$r=rnOfWqsvb zTvD;-H_^MhJq!c$GwBaCKD3nj^E#}_t6Z9Ed}gE>BHdgMyDF)hXd0eFxbe2JPLSOc z-D=sNkB}Jh0Ln}mHEgJE`suxB;M*+7fzOb&IFHq5EYGm!rIgb&{(K zTP{I$*@0!$D{XV)&Jf#%p_{eM*WO$SS=CRw09P)-G7K%r@kWb;n*jK<3s8C<#k^p| zPG1@i${)j~mL6k&n-?k?G2%*^7j{c{eS@7D+pRLqItN*_V(=u}3>Xe3t;$Xx4jX`{ z&$EIrJ~L)O?i`f)uY_W6b#~TpWSbvur%lmCLu5tbA1mC+DgucPRRmSP7+m9DXtR(? zX|NEN&x0z>mjX%jZ4Ga*V8h48IBCvG%Y(#rL-groyF`tj{$S`{)a3mB17q52^2 zN<*3Z>8f*ES6E-u)MX}EmdkI8uK%OMW@!K*et333AQ%El$Wi%Cc?T+i(@&Q*!`=h1 zAqIW>C7p9@9c$Yz{K0N0q^ zlQ&B@{Nexh>%UE!`$tK|xh~@Ib4qE{AO35WkAHHJ))V)$Z1c}Gz(}@10rArCrH8@8 ztJx&zB{Eja?Z5~0bqbc`A6#RBVK4`OP!*z$N+;&T%~FU6Ei^yEFqt|MGVQ@8w?AI0uC=LnUmr6xN%P2PVw#)yv3rJ=!MfmqQ*prL;O?j z+#ZVcZm##S@4D)PNgKADzt0;v9pjJ=5>6|})fn9ZSjlK!8lWh*iiGsafzQ~s8?g3< zG3`wgKhg(DJd@*28# z{LF%bE4>dA2W>8t#|(oFT|LZMDlLBPPG&6Bl3*4pg-bwWggzF%_*e#64n3R^2?K~#2fx1Hn%2#2MXf`Puu5nwwcPt>S@7X>Fx{ry}=IqGyI0D32q^Q63mY=lf%sha+V;j%!n(JS9K3i zhYBPql07cRS9#Z`HQ6Y_;B}?{$-zLz05vy4kWw{Mm@54K+yJ%Lk44z&!}?C^?TWF3 ziVXE$V^Z`QnJEt7^r1DP?|m_S2^}nAZChaXCeA!-FZtJ*d~3bsF5$@v@5u@F0_!{4 zg}zat?7~2N+u+Btn`3-_k!yE++ZzM1XNJpuVQ=ft(1Q$Ew}^93u4SBuIlOC2m~d^T zm@}BRqp3JVR!Gfi4jkT2JI1Xs;J1;*znW8s@9NKqD}6iK3rScHeTF_=Sn~5fKL6wV zW!UXC=R~k+=bj7qJ})`;`Ln12*x4VuPRiQMfAiyr1^+s^bThz0BA$DT3#wSMFr*iV z4##F0D~d_jtW%^(6emI#2c-5$^M(38+)4PLtup;lAwx(g?@5#aQE8eOm&2}_n;h4w z?$%kuVZ!!$?6|A<&3J#oNLlleeB$ca#siESlD*ZcRHh)+(7)pFP~GLk#lMM*4qTqj zH(hHA8EH%1=dF+P_?vCkKWq~5^YWXUrc?8OT?3%%3}#-GgQ374n$5M5!Oq{M7v%NdIHqMd0y>gWV>1}MZK%E*be$LvJbkG z$!0;%qwy~vsigVgnneF{xLFy&6_%m~9|AUM)p*nL8EU6c;!Q3YK@xzg8+s3xg-#vh zBb0t9$v}4st^uAXEo=6-a0Z6U){RtU7h_TlD4(DCpo9kJ72`-5fS2&XPZrSJjsH z*p^sy3B7$D+*oN-=++s!D*Zi=?w_G_ zRp~+o+pDrZrj`&b;`)@t5q{b#$?-eH>5-iz6D4fKq zg`Blj4t>#kplH2s%xDbob~GMTystb+pN(5UDz!ja&A#V}23w_bjtFoH8Qj>mo=A}x z)8>S;=-7|Os&m{q+0gw^H#PUwK!p?-deIt$&ga?(p$Y;JR1%<`0S2c&}%`{pl-{(Os};^LU3VudJmUvq|i z(l@nRLT&`fEr%NRsrFqtf4qH3?dHOcHt*)ZuM{lK{UTR);ES~`wXVj9l%w)(15f+& zn<62z*0%X|udQNl^Y)J!-k&xD&o3l^cXSEZb=$oBxoP_Df$6fzv;*>d#b*IAVvQXY zyyG8xI)@qKx*zehx^f_vd?0a_Em!$t4Hi9tcp+J5>Jr&kK@`J-+3{vIATf3p^hRTv z;rDU-lk)_b0vfZwpejGh8$MMs4=PqJ{uBSN3}bP8qacE zeU{1bc;C3d&OR2+5i83>B79<>RH_U>P3FcWdv8Wapyku9KRX>tlai3r0&Fg=R)jFp zX)~w=1UElFva=zjC;Cf;B!G5H^e#+TJ^&aPa9b zo;p@iaArg;V>j2^J#JL}b9g__(U)#8F^~Jlk9}g8w0n})Z?A9w&!}ro3MwD20om20 zL##GzI9`bzm<217Vg1e#?@lMBYe-bG1_r^RC6qwQd@N6JUzf8N2$`9HBB2CCKnIhx z;1Qd(#4a6a8qh~=K!5lndavYg#6>_zp97L&i6FaSy%V*Zp{l^YiU=gS*CRe^gI&2CF58sq52KAa0H zqMf{V2_GCDY^(}tPVYO%b6TmO?h^i%86ml>@@`$JkaJqZWzDg&V}jwD%VGjBdAhI} z@-AeMYqYPb5pV{t;(3Wto+-R?K{>CbDJIL)pY5W%nP9&r`n=kArJ`(Z8^`6aOB=UG z?OD5}K5gY0V6`BTpWemekEsutHuQfmAGXYVDSu8`XHOPB=FZ0_N?@1M%d}jl-uFZK zW!?vm%0uU4qq_d|RtyYRs=VzV_O^3}Ft=v$fGNEv@rk}{&?krQFOHuPZQP?xiW2r2 zx-ybtJ)8r5S4W68EEO3?Tgv+|u3E?A)r~apl1|IY48|Xf*@BClgHpGA z65G6H_DesO{zIHMnd4dur1{IKQ1CzzcWa7BA;}w!ttlXRot?p>5Sn*BXoA_T&hWSc z`N8>?vOJ=MsM^Eqr2DEvAacDcv8gG*c0p~hlR*hW#4ygCMYdv)f!fKcZC@#(OAk0i z?+Z;|7_*aqaW>~#UNU!lM)qyhE6yC|w;}GI@;SqJ^R#QJK01-zGdS-_{$I zmbJTIwV}yZ9@ZP`CzO3+aNRvu`zAnctolw^+U37pw6Zum$mxIH*;-E`dgARnRx@Tx=POZ6q6xD=BI*G4c4U}W-Y*u1#8mx z8gHK)Xk}%WH2Zaf1X5K{$l;yT0RpwAuA$q*W(|HTdtcfNn>Zq@zs2dH1{Z*`6&JZ)S1nSEycBOhwYu}0}!W# zX0zD!fU_jBEo?YanN@l(t8J?98oDAZYWE8NX~C>7R7Ls8+aDz^7Vvyeo}v~Khw|tD zEmYwbUox3*t)0xb5SI+QPXoLDJd)k^u6L&tT1>&|q2zfhP7{?G*k z!E9Qkc~8FlQIh1t>&r-yva4^!&+NOZe_y|#8&Rqv;;WoHYjJh0f0$wqbt_X~a1RLX zc-l~k=hpf!@<}?G7Ag%hWIqykoj;Z~RqfDQbW-s+TAUovBa0lm!k;Ug{yrnk@aQmO z)7yvYQ#p&_7dGp?YZnVc@-++Vz|y^T7wYF?BZ{z4Sq${c88%G|dD*@o7J;~yOy1Y1~q`R)#UcS)2e ztM@Z*sF!Y*GC4K9X2?gF!G={1Q0kdiH*rMpS#{>C|B`f|mmmrXGn^y!bv zS6+Eqf8}Y@2Ww}&*%lSbZMydx@7@&7nN!1+X;DmWW6GubiJ$!4!#^s`t5kC>BxIH) z!?uLDQQ@RG*7BsIVKZw^dPpX(YUH?aMgP&%j-w$D53It54^^vuXzEmB7;MEe#-X18 z3X5Ebt+Yrf+CJt93pRLX7+W#- zIm;kQgQ#UwBji&o4-v8k{^ZxN_7RoilBL4dBspy=RXG^Xo)`HUZMuGzdc*FB00%`* z59dPN=7<1i_rGDfgPgn9C7X2o-JGveeyL{;)7z6AeQ}M(%Mqw^o!AAWD3--~*nlIk z2305kBiI>mx~}CSvkoo7l*c;WjDv<)b1?-L4&NwAKdLMvfRPc5#I};Xe6$po!|@L* zj?7t!db=A=i7`epZFi+|?jisur*5(;MFuL2)FZxlM2W`0f`~ zB?C2s8{gL1jn-Ui>{+TRQ9anb=(_r_BkQodp3O@18zxPoRPbmuvb| zk#eQ=z-IUbmB$5D3N~7YD^kx$Ll)JiCHRc)tr7=idENlB4=ppRaO0i~CUGz$h?p1a z401r8Ad@}Aoy8!6d+6Uq%&TIgB%Y8c1;OPZhZ&+!OI20;w4jQ)k{Z$0(8Y)8XY+h@U%b83nBwt3g)%z)i- zfWlE>(KB%1=JK$?{aGH+gP;v7KGEIx=JU=aZ)r+{wWuUT%$BlL(IU@|<^oPG`;*4j zW9ib%vuwmE#23{TME#sOBK0rFiPh@ES4szOi5shZF)ZJ10pL)ggTRK*mz@*6nE3L2s!jJy;@+DM&1fY%LqC(3URt{=!5Ef@RR9y-hj zlHaJ?Hp?+kv9wQB?LWm&y&<6|SZ`k*+^Et5DXA+6zi@Q3AVgKPf4J(BEU#afkLNQ2 zXQ~tNZM@yFE%&A|(@Wfgq3^(TN@9uNY?wPP$K^z4A8aY~pQKnfs7%jpF&5;?&*}rb zR&s3oWEd@_^Dv;fqj!aCMM-bG-x3?TZnR`Dq{4Sy8@*BZ6MuKoX~`Vz%*||^6&p!x zlN+hUw9MNBU#Lo`<&&6MyURnU2lxlwA2+cN4=HS28bf^=e%Z=sUY_xWU~b6Y_j6N6 zn`wXh!2xp&_D zy;S_+(98unlb54bfE!reQ8gcf^y}i8`YV8{e_3xcpeJXNk z|5UzPPmq?Gy;pTmwK;%UBAOHD9?w(>xS?{VF28!e+M(&8MqWwH0M`%6VQZaZTr3qK zJL*&^ORqf&Dtp-L7$Z@-e>puQQGVXwF?iI_TXcp$s&*hPhWo#MRYo#FI^v@Bz1TPpDPY`%Xv zEa6ZGH_~SZ+c#bURnVD#F|(%5f!`Nt@jwx-a9-KtP+gRx#|Ehi7V~(tI!PdQdL*nm z+hjMG+?g5Zr56sRE)_mdlwVm`NtMW$=?97cPfyP?IQq9}n6Z?mj=$dINmGz+}6o3|8u4xImYqm?kH z@vPis_rKy)C82>20++a?mkjRGcl2+Jf31hvr%IeJ$a0^4hFP${+!?TmEmASy*9lfWpz%BOnR4ym0klh)xra|A3|d8qXUB1SmN56EY-Y!MG5h5q z6&|Vuu+3?Kdyld_WsT}Ul8B`^;3np3V8%I(Q4Ry=^&UkLfYYAxj{cT|zUvfS8&tW$ zwW9WFV+psxyX)=#X754g(keMiRJxj-)w~3Q-=~z^Du7N;AKQ4n7d96MGA2n5J@BK+ zokl@4q%}D(LKbOy2QRH&p^g5oXSJtPz1|=9(bW>BMGcJ2eA;N$D%Cd?OT5nuXmreA zeAad_jGI1;|Ew;4ZS>5A&wrVD>^Wt;0!khq*}LZt|0S?$*MQdXZiZ^7{btN+Yc#4= zjtfO$%sr5&eJg2>%Cg&w^uqKSpkkuNjcv{%#ukpVqAPRsR909?gwKeMy}jY0$C}cm zL??ei3A=tzY1QJ0F|kiEakA`-rMe7E%wA?l+dbSoFL{f;dfv8UGxs_tMvTSXox(Zi z^HdZ3oA4*>E#|DM$>$||Jt7|s=`ECxq(@Vmai3XFB<%0lA6I(s4v6`TmL7fHFt)}t z{iJKC^FqDbpu^BPUS7>LMbSZG`Y(+KzS~}Zn9IK2^k&|n z%kTCVH2%c#XRFuvImUc1-_3lIJk(g!;}sLXj8`*Q-Wl7(`^+SX2{Dvis%!F}g+FN5 zQyMm?M1QO6fwhFd;YhkqO<^-+5U7xrXp(89n{6kOr?YIRn8D!jm8$t8C6n*6{Gfh1hhG1Q71X>6BGBr5gB7=j&isCf%WmlBY-=5tD9agV zLPOfbwi>bKhB1zFYw*EPyhOK3{Y~`hjmy~Xh9+Uox1&t%k@uHujdbG|jFzrH8>HBz z$aCL$Sdm;*>4!aOyr$zN-hC_cwxDxV#Ymk_E77iJ3rm|sRv~iY0Y1Q%G--Iz!6dXOr zZ|3P69*o-4*N+#CJu@7uu6^8`vhFIsVY8}Ys_OWvN!q{l$aai0#wu1QzRyqn$Lij} z9)`TVX|FwPGHY$H>aC^v5C^o5zkYz-EIM$oIEoItoCWO<11t}!)($4mTh7dT6#pe# zQqaxLRHP^(p%5M1z}}&Ve=$XlC=8{xV5<66fF&`H0W0z2d_xw&!{}eTAem@vT(G+`(c@c<}vi z^`~C1Is53$A1hvf#_%Vz@9em^cOIb1Op;fxmuY|SmdwIC>unD+YP?0E-b9fi?buBw z&o81UiPgAa=1^mSVor~=XoIB9UC1NaoQ}=ux#J?MB+g>*34>p!ZSx&D+u*a+`O!tw z>S&!hQ|u%8M!DC8#@9~1$iXsKTAplw$fQQ%ur|fH9Rq>c;H}Q=kysjwM82GJcPd4x zv~$*dM;Jr6?R-f{sUq*Q=B?q~Y|S^c)i+hGb0zw@k_ROTt3$#RRd($s80pNuvX(-oHI(? zi;{e-Ehu+v%B056w)Af{u-Mtu#IsG7 zvlW~2N{8O)J+UE+VjOE*UKPQig>PZ+Rrnf|NJfJ zvF?+W#~Q-}$y={4+dql8a`^M=q?hI57uvF|MNS0Dao zrRl9JuPs@nO{fp9Vy0ZST_SpyyRErM`qWSv^-tOj``R}AK%O&qsi;UmTYYNH+|7rY z?BW+}W!@FUHf(D3djQy?kSq3kSIm;A;)B}!6f->bjFt?{k5SBEeD;|jXH*S%?Af}w zjdMa2p?4Z!vyAU3u4V{%XPcbcguL1ww|%sC<_o>wu}Lm@v__K^zn^F54&}u(*$pK( z%(-Hak2KW_A8cN!{HAt~RO!1g5O(ZnHgi%_Y!9$uZFTLt7pLxXzX87jFmKyCa3c2SZ=O?p z+V(%&T}kePNcS%|_;2SYMJ{^a(w>8a;yAOIs@P(t$O=Lm9a3X=yRGlN<8zCXwF9Q4_;N;Wr~td9^Bp2EGYXXbjS)< zru|Darzz`+;rOECmu7DE4{E$Fyt+u|yoNRN$J8Y4WLN}Uv`Y0A{Q_5^TB2o8TKgrm@oDqNN!Vb2lgoc zT<@mg!eD4x)3}=Zi7lhYPY%7c?rojRq6Tgr zfv+Lt$MdevWcB#TaeKqrrRwiT_rD+1pPoB&>b!ZMK9MfkF>}aapzl&>K;^^oHcWl| zqUX(|vMa0V4-(znJ<&t4HtZ9t?*}BUwn?AWqyAR0?R$pd^8J-lIA347T%-#gIrync zqI!+zHndui9_mn)IkhN(b9_h&$Iw*2Ldnsc2Mlv*>S?$Yv^xo%YVR7@BoJKP09)+J z9Fcjubgi9 zc43oqb6nXmN#T}ew@KcO(Tl~0+V3Yn@jYsE{=jDB#DznQwvT@BkC}5ndH&9W+Pb^@ zR42;^>LC#C6kj`==B0@#7v&dd*@ms9XVbg=3-w=#R{wZ^o7nLIKkZGwn%w9!-40-YrB@t^6%G%9M)O) zZ5C=0{L{C$>=eIu`TX%;vw|vuVVA*C#_elYuMk%c?Fb#RpE*#(df=z81<<}3(sFAc zR?8d)UK16oa`a=&@W}6(4jWzc?GKl_Z(_pl%i4o)4=4v<^ASIl$!SoXavn9DkhD}C z|8;EK;$%(AF5uI=F%T3DH)mr;`Qi(o`us^sSYXBp$R7< zt}Jw&#NqMwV7FuNnf>C~30SYZv%F-F1>a(f3mcng^Ex+_2_N7DBb7Gqo2-_{G|!m> zLcv8zLF!#h(5OGV^y1;)w*Ue93U<3+CfB~^+kW!;9sP3(@$#eKo*jdKSbxw!J4z4r zR^`OAn}vLqHZ*;{u-wqS0Q5j2@WPI!;KK@KMvr7~)ASjpY0{-b*Y|Ng-)p!V-+xxK zv0Uri)Oy%|vEThw8x*4l46JW|73F^?@_C2Rt?J~?i&x=wz44sM)c7D3UtXffj#Pv( zM-+aF_8e>v(SSFFETAO_);f8Wu@(P^sW*X#dT+!3OHq`{GPbDfW-M7|gsALe7|U!J z!^v*OQr4nP3uP=}jAblYW;dfOQ&i4T*)y`tXp>WtLgff`s?&M?AJ6mp|9-z$r3o3! z%>Dh`_kCUOtELtv$>k&CTjZ2ss7$_CBKfk;|7acGvRAuQFI=pvI&0sQn5Y(R06(a@ zPTVxhbPjQ^98ze;!Z0ShF+znU<#qEX3~Z7P$}~&gSV?BGhjWGVko;KHp$sDb4p^8oN9tMi&>N}T`o5|s^6|&X^bN!7$uGA_6z2+_0Ey4Ec z!})nu>&?(Z2!@bK-_OVg?l(db3}4=^hN_KfoI7Lvt!mjh33j*ps3YEJ*Oq*oFL}Nx zIZ(y^ShFryo;Sz;6zXr8tWU|~gIyIR@0}j}A#^Sktyh^+0+hQOE&$qos63FL)q?$@ zX!G!~Kt>8jIc63;@hHb6PluD z8Ie8>dY}UvBDNXJaYsY~O>*Vxz*M{O^~)hx4soQ`4)Le22e-G@-jNWs@)jX-NQt*i zUkY8D^Y(x*1fSY5bvsF|t;Li0=}Hl~$szzR4lct-N~Z1)*lU6z$|P*BiK1$N|Ff|G zcMB+>ECH*vDoIv`8GzoxpyL=|QpyEiHJ9$d5s@|byiIWf^8@a<12|~kkBM}M;!=jS zE~>GQr4o5f4BYc*KLkt^=ue3jV@mDcA#b%%(zCcsYOz&AdtF*SWu6ttk6VYeHoiAo zP~U=Up5QkOHgETZ+NO-w=9LP}oi&;jV0-Oh9=&`}JG6Gut*PasUb;Zoi9N=-hI;!3l9_-VXxCi!e*$AjEdHN{HF4 zKU7wN_YC#@M)frkGMnyaTJrYuJp~ zAu@GxSzac(EB7O(@xh8-`uJ0ng-g1{gA_q7JA>bbNO#r{7Y~V#Jz~&{$X5bM8TJNX zE(zqpvO*Q>xZ^~+Is<5C;@w7=R}b*Yu&TJ`shgEUy;)Q`*ysYfVB-MR!`y z!)SX4i4g}(Djwb~5gT}rbL-r#nq7^qFu_g&owI=2%SbfgMYdWZV2l>F04fOfQU+Co zZH{bWToRzz9IU@Ns+&wAv92IY5TR(K7EgKn2w6x?`Vns%A5`U~G$;x#7g}*E14%ww zFd)uQ42k9jo4O6gwE+8@6)~LODR2V=*5XMd^I@+W7!D+R-EPRqg{@*?{`q3?h5Aex zM3ce(!{m8T(Aed~%|O%G0EJs0D0@wJ81Pe?XxAipdR8zx^uZ=_oi%J)wX6=8<}=2u z6rA_0agM{TtzcyPg1-FX+2T6jRzSwKi(+X>!D&yOa7C8uYfA^q`Th2g3|(dkoA zNb-uecIE&i$?oq3%pAs5|9TNUT>mW#S+YVadm?J(EGNEJv8CzV~hS5c;}v8EZrJi8_!6>)pwP+2ULJ_0t- z7bRBK%A08M+>eS1SPj%i4ZE|B%VYRhDQIhmLyawP^TUab3Z!IUl)op$mt;1NI!I3u znX=X*6$@EnG*{y!WkS<802iY@b)`ud09LVX$k7TOU-?_}>NmdfQ<7flApaM|2x22K zAS`IKSwpX+ZX5^%`Fl#x2;g}$QE@d9LxJ==;#R%n!)gB78iM9-5b8@g4&x(yg~4#^ zWIu?q$PQt;vx-==mvKcagEHIVR$>9V_kD~0@oc5Be%Dnb*~XjkT%!X{4fmN8>+r_} zfl=J5WWuxL0PFeMp?5GPA3cFJF52B+;32!0{PP@iSY*CQPcmF~?LIg(^dp%z(ffj$ zH#_k%2xnQ-!gtgFonU|wkpBeizur1_p4cladGI~ZDNEpe{fU7z_7I@QP-VhR0pX;W zVX&@MP>{ee(TQ}un|e3y)Y0{X;Q@Om_k-bxke^gzXBKgDJL!MI{ZJmv(u-M`Yn>W3EVkSpO;F0AaM)zt0F2&xUo zQERv>Nz`<*CzdhTbzU)J(m}7_XYSmT`{Z%0?o=cB`PR?#8jerT9jLmIwoW@7i^!Pu zsS{O0=exImOn!pt$ZhW8-Jtp35*zCu-su$g()XpqBrUkpA_}P;fT$y9RQ2?zvNH;y@gwy>5UdiSuC zcXsW&E4YNOknE|(*}xwhsV{HQ`=*df+_ zuf$O+a{N;3^CRf}F^Idj4X)qADB8XKOht8(@-$?oC$GrexzIl9fYlnLiZJg?^mW3& z0_kTUQBovn)W&m8zH23UVgz{W0WqqTD%0z=-w8HsXmXuFC(y1b@cecFc^Jk1m`_CU z8b^z~;Ax^~3sIp~6}mlZaEp%BmS)PyhNMz?HtGBVgxoNuM&U`Nbm-5^!R?2-(_N}+ zs(fRXaA40roPx$vz*IXEGL7@#ga^>)JCNnqc-uEj|E_M&3aClIUfIby-l_>?ADzG- zItHu>HBEBB*N5vkyqgGSrh9nmLQfNcuMo2(0JPYB+Vyr?q)A{;3DltJruwP^HN7x^ z<}xu~N^LV!jiJJdyIg_F>dgc-uKZdsMelcMoY0bkh|8t^BtM{pmc1^kven8L2RZxT zdBxz+;t1-ny(i=VakOtw4reRR1$!yrxIXAJfn1&S64m2=o8du8k#&;<1s>2VR!FW_ zFf+{t_BPRLt0z`8(m17gxS+kVopXM9ywHyP*NC-+(brRQ%p_*N(W$ZWt8nz2+Qs!mQ{4skIm z$*#Z@WeP%g+1nVUi_13=g~_SKnJtTP&-M0JbvaHUM(Sze;Ld7$!BDk*%L}ljGaY_D z&l^l7L^sk;A`-~btyaoeFqYn3fL++4wZ_I0m-{{79OOcA$sE{TVY0& z33Jyy;t==OI9vSHM#%>l0%ENoZYwn|h-vZYl4*7WB#&J%Z7Q-34s1gr)yRv4S>jb| zfs?X;nO4FtNgths&R`TIqVo|w0FWpCC5uRPTB<-x=6r0VOA)Yc%GpJPCyvAwCb zbX*fm^?fO}nnL}6h7`K?ahzhH&T0NAUFzGe`@s4h!Xv8BX{tUsG+8hb?QR{A@_2?* zv~~)?=&z53Wj&3C4ShcOw1=m&?PcH;_5Jy=KWE^9o{qlzKdwcdJp-mOi~pBBa93e{ z|Md%BoPz)V(@seiQ*^y3BHm<7Deo-^OUEBaTmz1?T(<6`pl^0`4Ne6Ab9I(W-f=jKVI~ z6V(ZF3eCv%0RIVrB#mA6>V03WYs^R@6Dxg9HQgSxzTut11^1ay#EtToF}d7}EV=Rr z5yyc##B5@|wlAWGQWD0`X$8>lX8p8`K{>?PNI?l0n3Y`G(^N5ForZhT9W&Yy!i7#GI5N!Hrr|Z z8>TJ?Dfp2b*x2@j@`vKBx72QJjZZ-iqx06hNC(r+3xx8XEW#yFa$6NBO)(fjRQ2ZT z_wer~2foGED`CTSer=(qYr?vVQ1LRmWq94ZFzFOJ@hP!~}S(D^)II}wObJLIW_@)PcdMZfiCnV%+I1` z&!kZ=1z4)=G0YvnY^aLM#(-%VK??vWWPlHLmJNW7TGny>lA2;Wbg<9SUl=C}oc zr`t7W9{$`?xqsY3K{59}wM+R6Of{@V%FSLTt>(jh-&K|iHv6+>eKj_dTFh>}q6XL= zmRKWU%}yr;Hh^R%JCW z8+=s4*caT+_fsOGMr)~7%e186Ikzx$aaHF3?N}5YwN9x+ z4m*Qw)k#?cooe(hZv-&9;bA|UrPgJjnx~PWd|vpH|y4~6a}(Km)>uo`3^)o z{1P93ScbrPJ%Ajjn5lqHYUPz5!l-fY54#$xcA?60GPv54Rqg&=+_;B6ky_Un1cWA~ zKfscv!|R>AL^>e9E-sfOHJ9f`5eh%p7)~D;4Np z4j!Bp^#6lv4`OV8-mE0m0DIfQF5s#}9}X`y%^Ef<=?f1|d3ZN}>jDWsKLL+-#Z2wG z3c0z}Rfhl4cnfth=NM$(-gY#zE$;2SfG>#?g}ak{y;_GwFd@Pn)p(Z-#5FRL)D=Jm z(hkH0RF`0o{jjpuRANyVzilAoQEFDIwZXM0slEOZv29?8pJI&ofXUKXd)H6{hP)a= zbzd}=ZrW%v^%QE==Z5yvdo9`Bm(b{{AH!DTXcb6JN}o+RN_N_YO_k)S>lI7A zc2P3-7SThmi{ctxqjtfzLBj7!cEToXqBFo;s0P?RGYncD`NEgo)>vL5ZP+1fpykA*(cFgRd#LeW_*h^D+Rdxl>z2D8VhI`%0evJTOl2nGFl>4$i3@@HLE`* zpm%C4VRn=hiH00veGKA-Jvs~hA0ma3)i<)j?|y5!5Rt?zjO}`o#b2H0?!B26SEKTc z$2VNn`&Q}2*%pb9Jn2Vl)!f!iY&;SXJ*2bH!e%>{Ox0nj9>U7Y9s0cmzTEspi6Az0YzxgSF8&|(~;w5Ku0X&Pc5F;FYJ=j~I}Wy)PT z&D^4BTgFJrY@F4GIW^kAQU+KREXqXI1e?q1=`q zfhs@5jr6$k-%=8VNezrRA8)T(Xjd3onB5oSww`Rb5QJ><(oV)UQ=bH%hFF%jhe!N% zBFkTjj{e5mN;h+_2QmWgO^oTp4mD@iEB|e*gFXa9Gbk&^EsPOXsceI@AZsof(I5el zw+0ZQgFMY2(%9nEP$ivfhI+hVN@kR{)P6r=Tux1y8uQjY{$#87+c#h?{Qn$nY(Qc8 z>q1}pMqBkiK;pY}soxt2l7Chkt>5i)M~)9_(M=b6u-@_u=)<=NPxw7<>-ib?NlWgF0DRjy zvOF(&TX1QG90To!Hi|!_?9^cU#v|4LGxN`2hiF(%&B`&#&EG4|MJ<%`)m$f*`7amT z=+zfdr_4@pEjzNqSQ z43Kl(B-Y9%swR)nK|N^-qr;wt!oziE0k!oh#=|EbPu?ac5Cv9HC-yRW8Z7x3{K9o& z19Y}q!LtD=__bKmIWIuIqLc~yGWvA#;s|BZ*zg6+;LLjP%pK$~(t~nMq~GoGgwPoE zY(7h9)sf6S*|MK+xUa!Ig6~R*{QtQ56=w1nt*lINhJbc#|B_7 zj#?3oqD_>paIWQRZzjAJk^NB8JCi+I^w*g4#ohFgxR1ftQ5}r)hY4STb978j%M{To zL|tKneLpRUd2eJyQ3a{Em`yQ^+M%?O@9Hwm-NCh2QBC@@LTfEo?QDt~MH8?acvY?5 zlIR5i*f=(7>XtI1xSEUBYa_vNUcSnB%e7ByRXaQP)$QZS?-ts5`#&Ic{a&~ptbqHx1hwu-S$4OX!q*6!r?MeA{y6lW$9!$_;a zBl8Ty2TPtNzZ$wBqwy~CtEAq*{fCI-mzH!EOKl5A%b}-v|2&}f<}R^_Atiy4Zl6gd zxv+x2fqYei-|l0DHK|Bp{ypCwqeGedvWs%%=2RC)IRrLn$8Ui>W?Wn{DtHOx6pJws zr6zhwGSm;1M1DTVO^$Tk&fX`|Wxr*V6mT|-V4I;mlVlqopIRc@>m;mh(f{vdq(|vv z?0{i3^Y`@_sMCT~69JmpTp|uy&~5gzFtw$6Ok>xYy%6CBzwDc#$;-i=Wr@OmSyYAg z)Pc@#octSE6+Vj6nix)Sv&`X}SsfLmb?!-}wl+y4aVX?QR+-;&aEBbX{JL$s`#|@> z7fuU7id)PrALGzqI%{*%Del%|vAN3^sZrNAF7f?Li&Zo;B$MCKo{SHBe-jbrT`;xc z)2!lUOOc}?c6imTiP!JD^#0^>4~-Ml)Zf1U=eq6xnc3jOnl}zD)&J+vV%qZJf`X$_ zFMG1kFo=XY$y61^Sy*HO>1jQ_zUu&d9ihRYGS&BORF_zU-eux}&evpTxGemyT;Pc~ zxs3{avkRJ`)k09$7t7UQXeU%_f?7P#Mrbd9DDMy1?9H^Wg|9i|`7qZKc zM4CtfHp2KrN<6Cvg(j3}nTB4Gt?&Li5$sMkPKV+i>HnHol@3(VAu~KvMwM3kN!vn~ ze@=)=IbDLmmyA0quh#|9a_KG&-En{GGZs6#dh{8yiO~?Gp zpFJ)%jZqslrA@Z{9w1I5w99Q_e0lFX(%*;xeX5A4Uos_Ng(6aOqaIJNAj*s~u!ImC zAGqv|-bQulQFx{KOk)Lacp~d;=_UuMCi9zc;o-**92w}JI*V6sE@RNf#tfDj{li3I z$%l!HsyCl9ST-vd)Gta+TOcW5V$BWHjX(8X=n!nrZ*w}K8g69zhYm&VLY8M~eXI~; zKQ~~rI@K7eBIVw;AAWZzuf!8@VdBHX!BzvtPfm`U0!dY zdSwhqlxaJtz+NEFreW}p;JL*N7pbyV#5h{4DR>FBB0p!x6ioR@cXJd9v`!;-c5Ki% z>nG+L+O9^_$ilTw@;Vh{?Xd2Sm;`whQ82279(@My;T%sph^lpWx!l$4--HB+nHcTM zO{A~F!`>=J@^b83lfl=_Oq*`E;dG$BcRzSjL+FVKGMoG23_AlR1h7qZs#Y>xEc_6y zLH?L5t|l9n$x`4j2JhWd*p2Y-O!?rRY@#{eW?`|-k(wx-_nL2Gb(;Ahq>zS!yqL}* z{&gerbt8(=`FLEo?mJ$A7BgmSp3?HkwcBNcJT30K>;x{cA5(qIX{dQ$gM00M zTN!!a>9#M~{Za8IJ5B~~>sgF7m~-r^cC|tNrn8_S+4Z**Wv>mT_us^Q8Hc4F<&8F} z*yHmDHM_nDeq6UYC3Tgt9tD|$P!TfL)PsB1@4Pz+P z7R$5C{X(~Y@NN)?IfV)pf zH=1P!FTuV1lqFvZCR7-wUTj|uZj4;=tVx}^VWO3vUbJolV@sA1^Pvq08(-Ip%ivnWR5$-#w}$`Cn%XQY_5*XD|GNbXzKQ>PO@2jOzU`k! z04WW!)yos?7edP#zbn1IO#}1++T#wvom9lVdDvhNhQyp;iV;Rj7}Ee}$QK)qy_d%z zMw1>elIO?<4dXR**q--J3s@neRH>j>I^w+gEM;Ec+2;ma@R$YC6haVoafSMsuO=*)cyLbl&AQ08*wh^luBJ%OIowuTj5bLB zUi}VJS`6g{1t20#tp^vdgQe$D_5 zZH@*fZ63+uDDcqa@K`)s^?e~Tv~)`mT!bD-e#c|+t$}H>rt*6p_`eOWJU$Z8WlZ$F zne3VMUEmBm)b6g=1Iz*~S~dGvZIZ_@$tP)O7}%@jKE|4D@sq1pWE5-V4$m8^t_1LS z+l=S~U9hjsRt4+)VN;LM=NwRG8pQg@r3Uj3MQD$9pf`KrtV|%nsDkxYlelE|QRcw+ zc*g1K4|%)*XGo&pp~+$J`@4IrR)xLO&BLZ$JMhMpCQII_cK-1<{*c&S z7N9W0DW=wH4}{yJdN^lj+bJRPeaflS z!je(y0+S05ww}1f>R#Ac3E6?L>_!iyObkP3DNXa3Og4QW)peA$+I$qei0k|uWF`j)mpk$zJ?&|uy47W)s@{G7w%-oajM?B(z>^k5Iy_)51GNrhibp>7LT^5R(9WA0JoeC2{0 z#O-W`m+IBe-k^Eop za|lSz8!hYL)9?Ly$*F7rs7RYQn=|)mbRngj^F5o10a+{U1Hs_u1&;~L|QDo z9`{AIHvp5v=JIfOll)3N#9C+tI4;d)_IY4;B|l0rgjiEaInWMbE=I&TkrfCj5=Gch zM{(Hc8ORHV_(ac{jyICdpV9#X7Fe{tyPDM>gluoIqf0)iVBQLS?iR1NI2^vc&g*-d zB@w_pZUp5`U5MKs{s(Bg*|k#*X=>nC5pj>Dz0h_Cc2k(^2?L6@-`4C#8TtsaLwu zW**H)b$G>#x;rMoyhU|F@_84Mmaa>=4*s0dudGNnTHLs4VOpxzlVa>EQ3lZK||6|&Ep|f5G9Uyng zK)_&dUohqLDi3Rz74VPAZWSfuCCzwkWqas$BG*?wNTS0Bbf%p>J)a~#&D3`%2sE$$K0@T z4U$mUo;|An#B?Mi?zm+q)bozYxO>KGD^-eyC;k)(BO`mh$A`lI;k@FfcsjC2EzTX* zoNaCSIC8Fe+#>CZ`sb(i+~;o{1Dc+xjw133`M9ng&Tp=*h0%^IbgXGF`(r|W3$&DU zd8SqTRQ1tGv_+Z^pEO|Hyuk9EVk&VV1`!=NA13e-M6)7M}w zN6e4Za@~vRYI`)2>fIMeCI<&?!T))Zo8VhheBBta!wE6Ot>Vi^=3{pNpZ=Vrb+@_K z-D9+#}`zBIE+mm`20jpNTK*_K8|IyX=R>Z>7UJSt;Jmt>Pdms!%8F)F?yhU83W}Kn8ZLhvUnf)JGakK%bn?{l_K)2gF4qYtH|*n z{NE{V@L{)~Tz0^*C;ormwV7@H?Z=AWuU)9sJH)jU`H@GUqW^5VY4D#-V>9m1vQ(n>$_>0>h6OUs%(*!T`m93A#gC;aO zyGL5cHZNGEYQFUnJDvl??W;OGho=aiNI3iOGkYH7KXj%lpbx=S&6-b>N9&J)K<+Wx*!%T}hMi-OYy6!QFyLLGQ1}Mln zQ->$_Y=zwFHqEv8;&>l7%?f{5JTam5ye8*Nd2h`9x4muczVYKpe@I1{ck*^Gi($LJ zU2VGJc}RD@*GqI%RSxEW6W}*4qjn^_+vrC)d#yqC@kzzc zrksg4+Y(czo?L}U$(Vwhh>l<~X({a|-fs3zL3jHMG_ zBU;tp+vN-{_i<6~o%&W|5lJf3P`@;wU0cx{j(tdTuI;BGujh)9k#-9KZIJh><}&j_ z=A)wcjEVE~L02mm)0nAc4&o(>ce3Q!OVTdb4^ly(sqDO!ILv#-;VGtSjG5?_+~?of zIVbhF|5k{pTQU}USElYrwVZEFVrfghscQp8R!!DZaNk`J_F2)fxY+6bYEt52Y;wL zwDBs$e{0nb)DLZ~U%lsc7u}S1Xp8nzYrd1!Zb+aNJaxaSVXeju5u*AVl0@cOhT(-i z@UL94t4E%q6L?1HH8g%a)NQA|fh0FoE&i*Yxs#&HIL&uA@ai1zoQ$i_km(0U2zFf= zsab(oxc121JKDez;fQcW?Nx=!=5OxgY{5U0Q2$-0I7fv0a_RhI929zVIU-yX?GV}xpS&LM)JyQEQZa(cW$!W; zuC0+N)8IL@gx%tvm+l40IN`fkk;m7It0hD| z2k?dpM1fOx)hKC!+OraLD(g)DKpf)Fy(c|Uyd!mOTj;Db@rSoHp;hk-zuevquv!Z@ z*GO|*f&8cFoNKUT%-d({(1^E3SQJ$9k{C`E-%f{;T+tjQka>mB?U>jAfU@X-oB_~7 zdaR0Mz}B`?Ac4}g0oTsvM6k*$$2uCw06yf8iP(%WsaMFNmJ^QyIMh>kHnvR74q~n! zv>g5+r*d=1kMyf9?u1Ffjy=7BSaJfy6C%c(+qDS)*fJzce0Z@hJ|x=;KZ|O+A7~(Q zsCGcx=RP2QNHMLMOxm^u@y8C(h)sB2zm|{bLTnRmVd)C1$KIKS{%2D&W%Y#e#=~hr z{lcO+b0BZqav^%#&qrgA5=O&+NX%d{j*PtKX9=&`)Br`9(@$ol;;L%s!$tRTM}i~! zNKr;i>Tb#e?Bn|zcd;Q6XE36hI;NoqzOv%%7D5%a9WJ?q>(3$-7A}|fs8FLNc5-b| z#l63hvF3`<(Sb}df73Q`;D0vV-sHRQ{=B(+MR$-^QzVZ^a7e0p8bX$O26h3@3e=q>ai7cQl!hhWl^zl2tYfvic*qenPkFj0jgzQnpk9oXc zA2N!ZCT2Sj6xX)Yney-I%XPh+wUjUjTIE;wU(ndPLV7p537xK$r~T9>Q}Wd8o85|9 ztBZbaUzlrOOzsP53UqUrsN|6c*?Q_Ucbd9oWva1cWrpUU-C^oiM4>6Pn_dnu8Dd3V z>R7+k+8Y8nzc*}~YKYr*ZNhz}ox4oqTt2*|d#fG+TFy%$J_k`B97~Gm#31Kw)bWY2 zA-v?>{FMf=k}r-raqsHozNover&LIS7b*$T+~pae@X!5OUDA~9MSN%YjF6CFi&vb# z_l-Y6nq!O{Qlu7)^iHH-{j|e|M609Fx+0MS@R@jZ+eeI9kW&bgO^#zWUZSa0z%Ny( z%J!&u)4V*;vPO>fs_SLctMEkypIe4vr-1YZQHmYCa``)QP z*XRHEvG=LWO5nutKR@e#IQtLp)impX7++$;7&{H{C{uTnq87?DPjfZ2rQfboTgF$B zMJ_zKN}P0qAG={_vmIKTKl2v?Y|ZiiT^-m`JPmJIHF3!@@>Gx?dgVbg3K z2tZN0QUI7);DnQ7t!1FP;zK;mbTz~RRin@;&L%2*oZ;;4QmWl5E(Wl!H-=pHO1AcQ;Q7| zhk-lkO4X3@v*W1WOc4i&CGN3jH0|m8BzAKOgWcm8yghNj61p>!qb1yBcU3L3_UpA3 z<>C8;!~GwS$K$7x%n1I@#P{h3cbm>NPVocE-}A1HWFg;JF~|^MCi8L9z>Z+?UZ`#k z)V1{^w%6m#o=A8E3kFq~LL<7ym>Ue)MCMEX#~3YHKp8JH_cpfzX-2iTYt&ffOt^wS zfUKV+<`NT2asbODVfKc$&-+c_n#u0xzh{wyf~vHYVevY4ty_g@gmPW0zHs}6$lnXq z@2QQB>0P8sbDcV#HS{!ZK51I+^BIY^tABi_+vBrZdrQ*Sxh-BEKel8L8B-Pal4Ra@ zXWTw`7jZ;7?~ul7k!1vru7L998kag8PxH<@kI5*+hsRb@ckT z^J&w}xkq6qW^JkpVMlup{{rezw^>yM!$WDQkY@xF_1 zotBQqGS-!&G5HDn5}SRnxeS3PsyjZmCOL%g;`9Kj?0QVD){piZY6eB_?uJ*lQ9t(A zS$(72pHkR7%X3>w(~rDdCuun?SZM6aP`};aJlplS^A_6eZRF$r@QFYV*OO!sJU^Ga z5>#hKubjlb{9UU-j~){5UzNrgGOdM_J6rU%$p)r5Q+yY4CPb0sXOkq4T z6Vq&qD6j8+&w8qR;#{_U^Mw=Nv!t(d90#9w-F>|pbm-}(%VJNbzXyo@_kY)Ll%SAx zX(gC{{ZqxMcl>IkOZN(USA(rvBT1j8SCbLteHq$6CrbvIdNgg^vsKj>S<2tpjEPcc z*%eIW*9ommj>k!|6>^v2_PY4b@x+pg#J`j-lGZyOAcd!6wF7-LuZVpLaLRAxs@{XY zMr@~qA7Ud|Gq#LJWoP?E7A#1Gu54!uQ81&mr$nsLXGh&GsN%i=ywFRq1ogdMz8e%y zY}rk94qZK8tvIHrqTazBVUNu%Z)8f`a#`<80um6`OxO4qy1O_S-MT zrM&Rpyslj6%9}F@Gdxt(eE+QU8)QpSR-pS}+Q@rf$~}%~gOR2>VklEZ!YZs1f^bpT zxZ&QzfP_AK2^vJu2R85xmXhkV6(j*9F++ox8(~*aqL2Tu+i50vuL9h&4i^wh;a4z` z$Q+N$u4mJm4c#q;=-gpY53zt>u4a(0x*34gRX=*Wn4@x=`d~F$ogcgbm;t~w@VgPw ziiv98Zq0r&mB24E?V8Qo#0H}EkKT%atp`4d(y<=E%#Pzk8Y=;^vJn7eL}+QtA{(2w z!PntI;>P1J=ZShnKSe6RCK&{T^+-Mf2OOJm2%~Z=NDU$*1vcbZUS$PQYO z?|fd~?$(Z)>bt+#bNuOnBIhNd+MS&3^X+B6lxOD?i+(o}6T{+nO?(g96;IT>tT4w@ zyS`5ZbuTa5um@72wF3{w8WaPeB*0lV&cS%1v3-TDzvX=6sCJt7NZRPtP2>8hPOj1m(9==6o0we%+<+$ia-{pU{k_n6nn=(6GOM z(?R>`-tBwT^;q^RXZsUYdj6Zi$(zKgv>@8w&b*VhMHXgr1hu0NDR@zeLm>WFP2RZY zJFlw4#;prEPI-p4?wQ8Jb&&TUMh?wSq22M%Z^VW=3Y^FudMKt7j*Vr zQ{Zo}Fl-lF$C^>LFqZRU3kj$n3Zq^_&9*&O3plU$!r08*(E!Nu&j<}(bMY&sA`1J3 z6;&msV`;dq|6|~@EsUB!SEjngK0^OZeXII;bnvY}WNY}x;8Jiqsoa<$h>8rbMeD^D zo(ONdr!v)`kI*HeI|Rw2%{0I;ssz4F>dw!e0LE_t?c~J;>h~Q2=<}o(1iAj5yirQ; zsh!=6AomUkl4)@k`49hkOA8pvK=xBOSw7g*$Tqd$9TEdvZZ-9424*jqodZv7P;&wJ z7#o>b50cA3)}t5I-Do3}U=sOWRRCY#J7l+vF*D1OUTfSG^@T#hBBgpISBfY{ocrQF ziQUayHC5qT2# zHv4JtiehvE3uy6-5SMXLK+4tsA`MqnS8*# z$cbS!P8y0K`xapd_$QRV>z=KcgRE!He_PQi7e`&eaT?h_(`60>;9{)G;b4D_vy zbNLL=%zG?uaKS!vPhmc@Qi{L0!%M*0&m74F0yl+1w4>Gzw~KKvkHNd@J9I%f!uiwe zd9ZX4!l{{#6I#bCi$GG4SSRj(!F-<@-wf;3QKBf<4kGlVQ2E}rT$Wt3X#|TPj5IBt z4Z>o~U2<7iBHK0Z0E{idPZq6+je?^=YWT%)qZ%6b8hjk>#oo*=~{mY@?sVs%y zNO=bJY+%dUnd@05C}A#=5^Iw37i~bu)NEl#OEheY)qewNI4MrG*M?&lkq}#`!gw;b zt@T=#Lin{T04#n*MLL5u3-}ABBjp`&YFhX>-%R3vu_b=EdJv)H5ntbmNX|Io{ ze?kd4=O{Kp+6gs2ja<(*O?m9AbW^WUn67$i3{!3ToV=v9`L-xRv?;AwDdZx-MOHhI zjuAfpJdUB7IBq(KdVKHUmXL~F5Om59)OaDQM2~-q-OyCeq(>NZ5eZuM=h> zFY>N3s>++9g&lUJYs4!CTFFE-J6|3x%=aI~y+vdlez=V|gtzBq+&(6tt5kcE9nh=P~AZ(mXpHU`EwqWl>+MB`twpkkIqK?igoRya=yoF-6&YN}bF)jktF=1Z?T48pDmM~o6?BLQ$6<>uz!WM;GEXX;g; zJyw!j&JrBQTqEs{NvVQA7j_Dufa#N>3h9 z1W2TXt}oJCD4~jo(z}X+1=05&-~XL&zI$f^a}v(LFpy_I&)#dTU+8Uhw4dJNE&x~o zY;PKw{R|W<2mq$_as~=TkXGzU%0R-qTR!xaaS&oq={kDi!)Iko(%i@=aoejFbJHdh zmIHvm)JU$AJaY>Vh9x}vs%OxFg!g4@7A|)?_bUB!xLODtM$f)#zY|Ys?#=q=aJle^ zKeLlh<%Q?@W62!GqjtTxv=SWtbXFjK0cvG=x z%C^B2n8cUC0R{jN$ox0i)S2iD65O4PD$m)~Dxe_UG6eA}BC5g+n6-YDA&`uEkMH6# z7X4SBfU359mOf-WPw>D8VFrCK{w*X%1X+WtIl!h6;B}^bP~a*72<3yNId+kMIe7p% z%k0MQTwsy`K#>@Y0jv!!D*yY_Q=Rxv1KcnIG*zPrh^p2$<+{*SNnQbhqZ%$xHSv>j zxM|*Yf^JS64A}K4(4K7&Tuz}Nq73zM z<${AL08|7)xZU^>2Nu18I~D$x$zm;F9P>-vE@cGjiqsRZ@coj`-Pa?z&`q34(g-FQ zc2ypbVuZ)*HKz-R-Z(tmpaTFGGhRK<{q4u;_(-``N6u zlQG}QbF9Bnp~le_pwimRZ$GitsEkNqG*yKe>gYcwby8eltXs9Hl75k06zy7^aTqvE=a>M1`Z7}3mO82<(d zOUyFRMOc0IX0=j>i|A|Ba2W(B1Nk zYB$dL94AG8j{Hml!@K&l!WKd@0^me)asNHJ2G1jj4;3K40w}&n5?G;BxMzP(DO3PD zSa-!U=$4PiN?Bb$z7$5*VYI96cCr!%TN`(JO)BGa)|FC1iwNxjCoDBEzL|8%HyKZZ z(+w?aV#*|;U}&$MQx+)@#LHE45(&imOY8XQ$pB$&P7DH^m5b1_)T`+N#j7FVy%J9M z)CdHE<@I^N+5OaSKiVe$ba4KcT^BeB{`3&0^`3X0&-G*JSgz1>fKUi zY=~ubdkQT5-tq<}QBf2KZ!So(utQ+Tnh@g(>t_W_-#|7OR`^ z+Vqu|R(%i!LRPxCMMff5s~}Cyj5zRBwW4N(!$C^IWW9h}`mGIK@4<|2K*4&A=;C1p zIq%cyBd~hqJItx6a(t3@1TyPBIw0Mgm5yzWX=g@W4{W*nX}r$+L@=oC@+ouZ*^s74 zy#E}m7iHyMYRthae0%!JAi)RE_8cUaI9l0{I-5IsKA`z4w>V9zwCRi3vS>Kl9U3`3v<~tH3>n*zi9!vuxG!;a;MVeKh>;AnyOo|Rlluj z?G4kM|M)b@&aJefu^znJEn`%0?Y7oe72~HrSBPV2GgdwqXM+`8SV@L$CU<|2d(I$4 zI(ydr%-(7Iz;WLEe97{|c*QAxs|UnD=O23TrlZ!m^ zl|`uAAWah*QKGt$T=Pv8T_hE;eeYPL&OMXbFvt}uOssZJ3#s}<9)8rNe&!Fm&eGhY zR(oU&Dpa_zNj`+|d(`UBr53Z#XtTJYo89*AX|FOurMAr8a_;SXS7ly$`lubL94CB0 z9UI{K$L)0TJ*UjD-=m<88?DR5>D>|?bLV&^f5-g_nwoHuJM*>T^Fp_}KE9?6 zH1sbMOPH-q;m4`#Zt4|V7FC@LmF@j^f_KET)q)bH?>D;s8t|MV_KAO}*uuC@eDGRh zLFL(0(M6jLa%lhj5-;_1QnxBWEY001nF(CHA|i&Ta0hlPmj4lX>cK(Mce;8&EswYm z_=kU*FK~g)Z z_NWNNKoD;%;o)ewMN39G0oc1j2LZH00V*TanxgD#1AG;j916dh#{r;DB8pVf3Yh;% z_`adr88xU>Kun9x127K~m#3a8{cp{m`jP?Y)PO}CAdzEb45Phyhb>6BmY-zn$_CR> ztvt)$Wtu8Aw%6OFN*Y9%VH2l1lr zMGqUR;hi}-B>+*VtcVKF`jTZln)g?t`V)5lPqOjl#lx*f z{G%G4?trIj^*^deHaE5E`gafChs*pUns}ZtgRofQi=13Jn+7Cz5ltUqoEnq!`)`VT z!h}C>E1hfc0O@y@r~phJ>{wk;lR)B&EORAYSrO?g1e}^lbRNE|fU@<8;_kHgspo(_ z>%3%Xj?1F!6}k@mJb$OP4OA~3W~EqMcRLlAtuxuZPJTjW{xu!Io$IBkp=rVK$&BL@ z!mxf0$ONwvTFeuq7@oyX%hOOCpHS7mW~y#q7s15%*=+tfR2#JIyOqf0hn8}y6pfy# z5&1PUn3Wm|Bv@Kx{@xD6x5qx)>{SGZ2&X%`Kq>I}J%j_zkqLG_qePe_dAb5a`trM4 zJ#$DOFd#JjLalJwWyRg7%(2)*DoI)AebqhX1mvHmgX-NhN^eyV{?2SZlxXT%$5&Qb zRwPq&bKzohh&wTU7T0#LH$F@8aTnPp_@VZAD={r(=J1(wRBa(*Y!cM6Qb=*!|BRX9 zn#NSb){ar#7w*dyI+&?t5=s+qqY)b!Qt?8(7~S^5wmZ+yddLWmDjW1(lGgK{e;bSJ z#WW?zN}0+l%>uccb8!B$C|}_JApV?%w{VnQAAZ*}kG)u|L%3hb9ku)52FwE3*YP+U!-B~L$hD_|^- ztccSBYxYjocFL=qZ1a|V?}bMMEJMDOKrxKIl0dRt@0k{^V=LjPo>g}TGK8>tZ*SN% zgt{EkEq2##{`~54nv5fTwj;k~56743>A3Q<7pvs^Bo&|c zK_TK80Z-BRj45-|Zd{FL>^U=#k7wcnv&Fjxd&NwehDNZ+b6rchgoCm#y+#O8EBm(^ z5Q7%_7jLEA7Sfy{#S0ZkbMDYUwS2HP9^aJ@(|`%?`wK4 zyuFgjtPyd*fMQyFrY5M3`K9JPzwsHLgruR6dCz72&SxBs#^0>R9BrOeGKIq2G_`NO zw1MYMyffo`lC_!NxS4lQS!k%g>4z3u7F@0=+kCg9g-=y;HejvZKJnV-2<(UrGQ%@H)E_&6z<#onCQmoWW{hD%>299`?C&0{F&xb!YjI& zC%V?DgZWc+2dn;xmrfI%VXaC9Ti>&)G%IMvQM2HEH=CrCeyjO~q-joFJ>zHfxhDMx zh~-G&q!7&jg4dG=9M4|;dkDxLdI+0~cD;tJN6!<{^Q2Ne;kv)LTogc|?}dQAF`WKG zy`v`VdamF|p>*rNd}n9FfjXQ7?-&d?2QbkCBu%_qEE8YO>JTd%+7fj~+ZI=uRIBv? zuW-O6hn1Hm_d<4TAy@n>$P_>S8=*DE!;t}W1K1~4;$T~w1TdZ|YDG!C8#)UvKd(CI zA7Iial@(U+j2d}}Q$P%s4T?9^)surSOke2gtEq8qw}|JUaVsUrqJU{9O3E0}HWdir z5un{66;dyJVmX(Pl`voXKjJd_f86wxUqNLD(Ag_LCQQG$9F%1C4~k6vQCHIN{%f)Q z-*TMUFk5jP?iKPkpRw*=rp-5=LwwbEvju&>hueamJ|Vg3*Hp_7s=eYqLKGPubTPz$ z(|Q8XaP(*uY0$?h=zFxB_MnSD}i5*TlR72S67fmHvq%YQ#6&z;r zzdCdk19iCB)5RWzW!`~li;-8-D2jc#5xkffM+?w)*G(=W&@EO-MV|wc`14h(C4W|W zRHyBg*H3n{>nqk6yGTR1LFyKI0oa})!@n$SVD@f5P#cM!_Bfe$qU1w~BuM^3#pH zt#KqPwu|TL{W9ft1MIqQyVyi)yV#8SG~|%!pI++*yaX$ePCTgF!LX~P?xJ8m*N2-= z=XwEcrTWXLe&XEZ8-Eq*Gg<3RE7>QAimsC`!OQj}eMt2_r&<+}zV492qEt)b_ zU&3}gN%4fgHyb=nqwL;ZAqIz@gCtp*4 ze8UjeWdu6wkf-_twktbz7<~!ik^5jkofx~0T&$CYwKxx)0~FhZmt_qtsZYmMU*;R- z+hu#?TVqs%bdld<{glDSNh;zEq-pU-%dkl>n-8kfBEodt+XwbX?&mbyv2OfCb$r^o z{&rsx&S)FqMSczLm%Zb$L)yPOcA7l6EFt<~R>|dixnib~9#<{bMBDsv*&&S6GeB^? z?%ZRdYgL=nc3`%S%Z`HJIx*#U@Wd6$}oos^8$$Rn}zktn) z{5-vWX*-}VB!$?{{Zg&)d8*A&3}5u?*@_yQtHeF&dW-h{XeTHKR7 zGm`G=ao%q!@eOTs8V_ijbL-;3g=wZ4Okce4#=Q|usip!|%`xF7tAy_MZi`UY(L`aP z3;>!e`%lQjOyKU3Rt~|N?87aL)URI>tXw?146e^Y&4VkGRl(*g;ehiim_1RJ)jpJi+2M0n#8|N zAPhVd8+IKgbA@r=c{yIVrc7(9wgTSH|A2wi1iH?Y)o=?lR+{HieKiYI59CX}Z2m;7 zw2A=qx%hc+gb>YrSV&V!`3ip~E(7SPhF|bAOPaUgVE40WU@s^GerbW{Zu-mCxp?nN7me_-RRLUoA_C z%?ZOMvj%hect*j2H1K|8@kqN?$1t`1QFWt#rh;2ONSOFJV_VLm zJT)F0bC1AiXqz{Hr7{ZY_cn2wxAtUUOlQ_8Ufg*FAe*|jZ6^;$_yRm z`K6y7feu*=_$mmswRQTTLNTPMY@tVr>(cWai!(uljjL`ibX@w+W`2~5Q+h~^t=H?}YUh1D<&ZLYcDs@dod){3kjexaC%xYejcmPKMJd`6 zJy}V=F@{ar-ZS@~9>s&_d(N;DcSza|kE`+y^c%k$=89Fz?pPD}*LJeor`NNu%B|PC zX0A$vb2=p4*z$2a)c>w_W$dz`t~=~mK=lHm@Ht33hshVu9YEMjK*6l>p2ecsvOkBx zm7hN1_~^PlGjVxjuUPlWc^!mXIj6R)aF!O(-X_|8h4Y9%2G@?_D8yP3`gsnJzIRD+ zIVW1}?vzZgxcW7~_WZicnSO()v!)wM7e7j%iF;ftF<;LPt<)yb)#Ed=gCxVAC>B?? zDoI?x>Bbc;;B?TfGTbo*Q%l!j7Mik0vWWP{c86q3ohiA~mDQ8o%f!@-qJqt7!jZ?= zWiX>@wk#p?n6N%$7M5bDMYA%HF_D%S-`k_jEhOZDXPc%uSVVt&(z8r1>S5!KJd%)! zy5v5CQ=jFdI|P!gnr2BRKB}NrcS^V9&*!X*9dU2%Q+@eo=uI!U3VWmNFurnGm5eNP zP+Zt{vQ1LRnq4(BD*21&=pHPzq;Jq)Q!03@2ku+HYL$;>QQ1xoI9tG|{E3%RC?6w9 zGJ`A^k071UP691OUz@7T$^&o>tJw+r_phZbN~l23o3~Dsq>F{Xs*N#hLz*VrA2z2g zi3UPJB?D!9e4uo#JbfL5AE{{t~ zB@pbw(A;p(L5ur+C3TjLba#M7_VW(0>|ZAJsw3i)bysyoqY9kcTJw$Nw&w!O4^NV8 zCTcQt%}tQ(T5EI4Rmf0ffMTFaKx2FkodPi1lNv|=T--(?D%h)Yfy~w! zA^>+_1MQ>xWh_tUNg0lI>n8*Ru2T(^UpiH%0%wXI%ZV!W=YI5lY<=3V8R-CE)CW=4 z^bDDj%9%5Y*mlh(3>noemfO7C!CYNe?4eyvYp}c*mSF<@nG(OOZF-r#M7-jeN88zK z=E>-3IT;lry`k0+)e1IfPFleWV%G4pdm|NWav0@b6R#S)-Uxqs-jBuQgiR{|X(%?T zU{Sk8`^SU@M$PNjuqe0AL3$9;v+}J&uzUknq(p9fvqmP82pjzHO z@}*u#{d2Fi%Ca{ClU31ndxxE;re7MEn{$jvrT#P3EXOpf2lj5j`>tV=TiW@SixN6s zL@knw6>d-~^d}$mqqM0T*WIYL`81JQMgdA*!Cs8|ROuYWYi?`sGILYoZD_&?y8sGA z;V~6Drc;n^h;hhFeB^fV5@FB%>NUFmQE5WcnF_w4HS?-fzgZDy@QZ&~DP|7>B>1O8 ziqb5ZWWvl5)eSbLQkK}}lq~*P+~&FVy|bDV;W%IPx&Vd1QQsobhO_2t+zUGy4F24zZDi1gdfe>(<*QKu-uj@ zeG_6sw+fgcUCa4!t`MH33H|ucQY3l@+uG|h@;jlR&vWX!EoJHLf;+obIeDz)ok8B- z@mAG%gr2nGFWx<*T($rFAfI~eO>pH-QjHWwbVN+wxR?(x`}0jbx>!jJ6xe!qY4?q! z#?-{LP@^RI@;-}tDTI+)7Kv)%g=q(!=~da5BT`+QHpgn~*=tq^0d{@wdrjm;0v(MS zEpUT@*Bp^F>r1)3P{(I`#-YxB9z1lD(_7KRv3xMpzo}Oqr(kHdfmr>hr55}hR9=?!mI*FW-Xowj@KcHnM zb#t-wn$xD7T9}@I!b3y|OBr_eHIIVG({>YFkw9p@_7y5Gd%6aam?<$ngG|Woa9P;g zhqGx%?^tJJHR`@5cC~{&^=s5D5B3$FtvS_)fif9&8v$=F=*V)sbhE=jKctkrgsvDr zb{tBW5t~WaPjlu2ZKC?#2W-fx>OIhEJT%C&$6Y8(>9DkzG3oGrpri%iS0q=3q~U>C&ea&5A5ENe0p zyDAueJ6&P}FnO9Hg}2gUmt37$@P_)OY4bb1w7M-hL!N{!J-jlVin*Iv+xgjZDo_ScY$!i?b-3hINs}r_oW_PJ zj2pn6e@&l(8LlGf@}JO&A}W5w<$J|Fm2~)w^n-%O1OQ6b#QHs!tLB%$9jn|`q5p58 zF!cod>(_q@Nk7`<6Z(^w)j@nY_dxP%N=75{Ru@4>j|X7SLV$CEWXT6EB=C$>Mk-ml zTatm)>%U7XA;@yGW8p}>t}e#7WWVCa8>G^|rw4Y)llj1xE+_aI6Igg4JQUkC{^?cx zVASWi>it2B+*!C@E$+i~P+0J%u={GNgGS4xsc~wP#{1?2#~5J=%`>E{L_%J__;ZWg z*D|;BAWT60NO#mzdv3~QCk^DEdH|VGU0qRslb?GvZ+B+^Su=)}dKwtW4WMf%LMAgF zs!g+ph&LtLvLuQXE|zmzI@JkmV^U_SYIgzOw5IF#ntez`O?mZOmc$c)w&vGN*f5IC z;~PoT10WFUr7t$jU^5JHYH<^KOA-Pe^5N6Bek!(CE1S>vteb^7rW2pmmC<$XgNA1s zVjdy_mF_Qv&tBz??zJ&))&`u?j18qkmuB_NAm682c*4#d4En3p#e1uL%;3a|``dlB zNt{U2GQ{(Dj^(1vHm`62X?SYr4(i%^Td9#AUt6H=R0f5^2Kid^~7Z3#3*gLDhb0%TwU5fSrTzPDLyNj;iPpq2EOc`4~2sV=BDmV zU;Jv9c{prhX!q4pXtd>~DsXtIXQ|isq-KHyTpXplxH3zeyGe~7&obL)Vd3Hz<|mpP znSXe~uQ;fo$=J7oF8iQ54g?|97J8!BY)#n_d)5a~SKcD1TS-@g&heA>6(A)+tAuLl z#XD$Q2Fn?vKB93k_p9_0nTRuk0u$v~0~1M`N$XU-`^27e5q_3?pS-;z zLx<%AVx3VN7)g=RhR2@u++wO9J>jsW592mrHYFX zY;^r!reH##gn%tnF5&+DYvD5nL&7hL%l^!CI7TT%?O`u^?2mMRZy9@F`MV;T;%d$? z61ngvvmp;PRUz45+&fsg7>Ikni6zq`d_&}wHyoaYFHwXU?`o46PAxK)6Hd-`hM>TF z)j)-ekB#yYEdhCT$J+@BW&xHC%J0UB&nT;!i9warC(U+Py7UsP?tcxW|n=hiF(OXzaaevch*@1yU0W#@#b$RpS z1z-B5qbW+cwusA_vGK`o?)-JnG|{TU+a5zJSBYC?YM`&C3{VZx^SzxxA8yYIDI^bYy81@!pHS>Fu)Wx|rL|5_mi?-;3- z_Pn=C9!if%AR%N3KYL~IkLncwaop7dexae_5C?^?)b|1Cy64fH;u&DGuf1F!jdfm= z=n9I=dV?M*U|*OZPP@V%ECx6c%%kDwQ7$c600Dfv7M}t`xu^iwgdd+UK+VO7yDZz< zXl{F~eTV64!e5xK`zL*wUYdZ7V$k|o>qlpj7J{4dbF$!<-7&thKFeo(; z=a^~U4f=kY8(kRoW#GDK!f}l!e$skV6h^l?%Xgq!vafQV&gK!eMlA((CDxp`<=r8E zRkufEpXbT}LbAJiq|ev@Muydy&LOqZcds}6GD~gnK^Zjo;vcC<_S)*`Q$bWdePmd{ zp;~QaYfZu9?NJZ$>|APcAf}*8HHyU({7Ls}x)6-vuXjzbq$w7s`*oi$i2Vpg z?y{R3J-k36yp2rAX@zZziu zT}`LY=q&M_sq`)>F@9nFoI9h`<9D@_?QLKXWohiizZ{peiAl(cb2ivBfnd+G2~=?> z&o}rfCk*K;n8+_BDJ)D0RYn#xWwm{pkkFiols2>t7HhOLQ@xPY|kSh_wX!W9VIqn^W00ankc% zrIA(zsupd1Nwac30ui9We_=WYK2x{wFH;x7`(GwagAG03w-KAA+bKe`Zx|29@BH6K zE?C&u9t%MGTuc5PE%IM1Sh`bPa#sTPYdtRWzH22}X(=^8y}VirblJ|I1zx)W6wacP z1c(9lw3sANC7Ftl1c=93CQ?aOe!nB66A9=efv9~X!6AbshIBG3kA=B5I+|F=pLqe& zc^7R`wrk7LD|=MKR^b9{rA%^ws6TNhT&v<;V%ea73(^#D{FJ+0tvGRu>jRX^eG$n2 z0=(atuFn5SFaK7}zYax=Z*37e5_kYoP)WMR1)#DsBj{uQpqxmLOHt^#ZYjo_dV%Pv1mfN+2qJad8t%K3UW64d^~&ZEG0P1=W!~_6&6+he#_9quh!k~;%WwKC(DCa zzS@Jz3uE~$r>Pl5%#lnKl|9rAEu6~b zLiWCw2g;Wj%QV;O5@uLC?l1O);oFUfDt|Azk7JO=1K-mrLa=*dHkvTGjmhfHZ9pF=T-JNYZKgW7T{lP zt^Fir4t|FTuJTnM+Y)gnP|uQyCT{oeJIdb?V4{agH2Ia_UvEiXE#O}!wtAD*?-2TmN- z8Ey>h`g{|( zA;}g*;g^5BVDyIbojpXt1^T$br6qleqbpyu!#}`cE$4vA3T4=ZzHDUoFB2cgel6wF zuRg6e$h`G}eA7_%^oZ*3y1C+)FGOo#0^~I55?Qt~zZb)%q$3~|+>VkZxA%Up z{2Z?zP&4e_ue>jK;QlJ4+RKzJi}7v!X1jWFUrj&deLrJGWIW2ICE|IfSeOg?Y^1zx z(n3&KnJbj*eNWQk>s;?D7L!!9Xb(Af_2o8fQvYmGHCmu&R0LZ{V}j*{H+2d^)KHsY z0?i45-QCswyZW9e88e+Jo^#)JH3j4xknfYu?Cj!R&k6do25h0GM{%z22X@1{f86VC z9!|Y?+H7AW?_Vah;r1V%&a&9iZysZ%kp^F=+*f`*HT<%X`Y^#0JK5VtDrah0&=ewO zv&rPf)XovZ)w8It__sfkJ#mgL$W7lz*o_y8-ezV_vR>sT_usZW%zQil;8xB>S#N+a z*jV0nEf_L2MNnFaA%t{;o@L)KUb&T(6!H&V)_s5PI zhOAD$dU(9x#p#1})h79h@1?3j6sLM4&Xnze?mAw@-~A10j+Xg|oWv`|-H-o?Ro*|Y z>asWN&=Pp`mX9QLs^YP;=1XYk=?Nyq?SRg$QF%>0`K%tA}Uss z7gurJ0k6qNkPNs@fOiWH?NEluuUT7sn7x8fS6GNyfrV(Y6cJdprcjEs@4_LM`gqyt zK=U?R>=x@htyQz~twk05x=t4La;z(^G_{rC7GJa=rDlsv<1^ICq5e#{^cb4i_nfQq z^9kyFNMEzYIa3{hErl_vV#_M)+UuI|e7-1ky)e*dvS=t|pyrMD(}AF?v&CjyL=7CO zXDPqDQDZ%vCn05+Q2y=s_xyp;!9nW7O;~0?33<&5Gzm_1o2rZ7<43>!g}EP-jNKMz zh~4&8srl(8it;aiZj^tpFBi@mcN3pj`i)}HF9^cjiSpSUltkJL`B}3{qB{AX!3q9*IIt+}ahBl_V8u z6C{D^jAA(D6HWt8D|5+SCSFo6@L;~}l)-KenJWS!=}>^NLFe4x&*qQ>CV)nWjs(VI zQMlKCnW~VaZa=!@U%y0TZ6siK`Qlom6%^JVD@ZYDtHJ!AOTyB+`T^rvkH8!rN z7P_yl2)7HU1txNF+T6om)?$Sf`h|VV+<=>wD(8uaId||>R)*FfY%#j}>n%S!46dT1 zt(%E8v!ZLYQei4a&E!FR;JZPZC$KOW-qPIRX8?#&)$*?n(<+sEB54#oh2{=ts zEI(3D;xomKiTIbnyNh!uNF6ZNf2 zVx!!>Jj0=kbWZ<%O#n-MyTw*~o1*yeVd>T13r2w6;5XfA(bjIz|BJojmFjz%TA6uw z8r`VyOHc=gvC!C?q9SER;}Lkt6?gB-R*Bl`{rTIcx7hl3Ox?Sdm=ZO^i95#GI++zR zHtcT=E{^o5Kf2GLVBbE^J*mScdT_E@w6-tgj@(OZjd*e3f^@t2BX)wfMfiDJax3c- z31r(^X{696<(rw_Z$g{~^u>t&=xHm8|F`+IUSgs2$$sFswRv?7ajX9lehZ|LIU~|59`Nf+z zqZhay(HKP`8-UQtov}wbPwt4u}0F_>PQK>yzRUr+UCjKLwUU&z=0a2s;W`l|+!y@uUnb!i&AQhE zRg?=?NVwges^&VU$zN&+-H5V|V3`7Dm*iW}iAkBYW}2;3Pr;X^q)n04s6K_G8Jnjr zmWc_zRSlu?iY-f%GHM3%Bv*|wQ z7c}#pYJ}ADw$J6Ij(i2nLr5X}?I$3X#J6wZ?Y4>Un5q+skshHrBW>Jm2IaOtqKw!A zuIO+4dIvRrDWN}O75HsJPE+94KovG(!D?pbrGY8JSShd5r!k}Sk=j`5ow~2v{xLt% zby4JF75!mcv$XndcB)-c=@Q+{X`yltyZA8th%{*j#~d%nQhCnh%+ zw6{-Jz}AENK`1p^K;Oh0)}emEgJqQ2s~Zt|Vf25Q;sWh$*e=$XK3Xf`2d};OBk0iE zTH|Ij`HS+p3IIv(k&;Ph{oPTGo?Am-Ab?$Sbw;`EL9EZ+T&3@lem|!Q9Tl+g2?DK$ zRYsagS)ZyT!8n*m*F>5M{kHZeowfyA0uX+mOp;WjB2py*ZqG&F!w>=F9-gjbMn|J{ z75@fiP2Zh5#Fy>5O?W$yY@kFMoPr6EIJs?>7~!|VD_?58~eJ(r2847DTdd{&-xrtuD*s17_6tHL^@bJD8C-cO_O>JpI6Bk ziO3Qz*!;3X5w;HK^OQFZHhEnB67P9tl@-cCb9V5MGcHWA{9MfkHsD$+ONRZ`4ej`J zT!*^!1=uKfe3x_o?(zpjed4NNA~w|vF+{x>^xFBJlH64!_kmFdaFa40o1<#Hd3TKJ zGZQU_eo0neL2%VrsXBj)0&sF)`v1I6{h!uTY8r?$7y7@t>;)ax$CBT&UK8Yujf@gZ zkOVB$nhj_O6gi<7%Rod^%;~HI)N<4u+#Kh3i|fK`V|_?hWNij}uw*ZDty5M4aEK4S zh>8M`M<8W^nBKp;j@P&%!&{9XAf4zslg79xIIvCd<)W&e2RZ{EkWB!cN>4_es~EQc z<|EiQer{~3GO+{$Ta!-I9-y7@f!vc5m?^uL>%!o2O86O)5{Fpr8 zyWKx~=I_=6G9-^%P`z~><|N724$fjM(1j}nJpcK;5%Xt~R>mg;->822%|I7dMjh3k z)Njd4tjiO^FEoEn9Eneh;UI<0JYRZiYjNjYl4iAqbtlgWremqZDY7O~$hHFUMx(w)A zR)xKC&*G#=ld^8vOw5FkoI7X_I<_C7@Hmr>pXg|vW1|vh`fTEC0R#F0%yrOiR$5|U zFyBM>3r%z^Q(s78#eaPS+V51D*yI}iv<&4TVYA^Q8{?3X_8FC}+QLI~TwBZ@S%vs#4D{%|w zibTU^O`4wZ@|w%ReP3l>Xn(}Ef~mH=_ukcza#^g?9BN-xQqs~vI3+!DTI%`KF_H`V znwR^VA^;xEfJ{AaM2;`m_p+y1pRoH#TDC4mfugjg2kz|5^0*}S$VPZV*4$K`iCc-1 z{Q*?Q@E7LH1ZKNF z8x*h7C|fU)2j5rp&H|$rfS6kGY97^6^`ynngR(Uz%w(O`zf2D!pq`$gRF}8l)eDsI z6Fca+tvHv<3lCC~3K#KpehLF6^YaS)%q3C75fPrssP+A#@6Oo}hnP33*JOSnbMr-@Ug=^Aaxv5o?GsQHFdM zE6{&H4&C{S+7WD|j*hmrEJK00e~t#9Gv74BSMJ@zzb-%r-6TtRRK5t(&){7`a@l8w z|MpaW#H;aw6sVtps+Z50{%D)FUtqFtp6%{cCtfU*#I)Y5p;|OdvELmXuPxEgu41XcxLS7?EE+{f?)PX{&Bx@^94D)F*HuSLl3 z>f*L05tSNpi4>7jZ3WYySV7Zw*2?rNRjg)%=@&5i0`wyw8V%S&R|%SsOZl^)Y_y^- ze(all|Jg2~gN0vzjMb&&LNI+|x(T=&s-EWjhCqG8@pi=;&d<8?3gP(H- zX#xp26(%om8mTI(S4 z?tm(G&!$}dqP(f$$;x`~-H`7)w=}-j;LwShPB_(8is_i{^o;POB$nWBv(FTr&&d(? z=gK`s^*!Cj>Kb8%fYQIKXcojDc;(Xs-8UULINH4xu52Lc?TZI7B`gES$}IQh;qxy zHaA@cvC`x|81P4-+z4gA*v%6fJ;u^aEk{+V6mV3nP5M3aTbfl%Sfq-l>lEJ8IVEd) zd?~GJ5MQB+{AuMjv-#TsOqR{%{Bl}~%G=beL3pE>RqM{9lzHl<)9Hqf8Ze)q*ZdBs z^Y)NzF?{tK_)9C;G(Sq?=h8k`qY6YrQgyxuWA(YF^Yh^*;>5?6N;(Z+$CkRcW1t`U z{aE<_;p#1e+UmcsU0O;h6m4-WEAX*Wxb4 ziFD*6#F{v2a3IhlBB*1eEk!hTX7rhj;e zo~;#GSF``03>o2LTX#V2=>X4W5SpS<$T$7NXaa^x6Keg+Y z(H(0;r>xBFGV1wI|06BarLotTOauyJ=n0YuV zo3x5mB%fP?M^S}QWSzGsLlr(RNq z)7!Rnh@Ou{yCjP)3^A%a(qb2L=lKd7sLHJWFep~W5SA9t3W>8*5_+%M*(zhtdgxfc zlUNbp+|Q>SiUpf~G{_aut#T=Dahl>|Eey%9)+6oUBp|pDBPqZd#MIHK^G9J1=Q>P! zm=GydT!6H~1ZNpE0aIKjAof18n$i`7B~{7{X=`*-tu7`f=Vnj7D<`_JbSqBS(}crM z-}gkn*VR#MC4h07wrfu{f+-u_az9)1I z+>H88Xx^r*MzPZ_LkK#K)W+D6>3mwk+55Zz5T|}I68(NE{g3i==0I!5;*X^@t@m4RhGVg( zDwdZABg&yle2cn%t?ZJ*t8Ha8v2e0m!*{mFY zdHzBsPpz!sUx~WCM1vHUA3#xHr!ggQ%c*$fI;7wY_h>w#*-@vrWCKKwab1B63MN}r z06$Z4!bU)_@Ho2ODFQN34ORj1emYe1OHFum$$2*85Nee?JM+dL=xi{VIA|sn?p4m; z8$2<~fv?L^y1bOFX>VYWuAyxdZ3U8m`N_bqioCGfWuu>Or)eq7CcfpcE7k z%OHUSsqUA)WWEIx4x#-M90p_@0gt=e%nemj)L_ z9GuGU=CvSGJz08-?h9079Sy8<3)|O)+!^JQ@3SZcyFok6?{YcYSjAu!^|~`n6f`g> zs-I-FA%reHpT=Wmre=<((hC2@2;7Seb|RQZ-OoyP;3ciG{l{UnPMKsUzFG|goXqyD z(LF>qN=_|nH+O`Gfs`I)D|3uGTb)ZQ&)LH)q%J(9KZ7r+xlMORJHtQ}Cz&T6->AZR z^glB^-9kotdmqBnBzTwz?uu&1AZw&ESK%y4BN5-I@czTNuXyXqT=D$Xb+vQwr{?yM z=h`!IrIi|%ogk{jhNe7J^1X);1o=b2k23s2n~>%3ml{7R=PAHvoj?Dc;&@mij-UQg zlIFhJf+n2;lx=JGmE2NCf*UasbnaxsbhqzuV~Wgt;V?6*TtbpjxK{~=BHtShG%Q?Q ziv*`oW9H8+5#m6DAIzp-My6H{p9#!t7+HDkz@oy^tBbFRubYuoY^th-G*FF@bPxi= z-_Ts=N9rhzkS8*1RDkF8t^2l&>(s}=D7+3)_$C51Dj}AzVsr@m+if=2(Nw;VJ)U5U zCqU#QRDgd}dgzc^{Dz-D+F|PJmL%D4kn%5^Ui*GA6%lh&H-dtM!A_*Q2~cIA``pyi z5x-HnB6O5f$E)0HGquJEovEj5DEJYp84U>2MUdB?O!WiH%6WjWXf?PeNn1bCW?Ia5 z^Sb4pqb;aNFf7xu44;y&LFaV=!d=5IIrcmJM1wK3fQRFXvjgt5pyyp{+STBu@e4_+ z6swl52AwVXa4bIsoBCB)y(BF!%y?9k+z}|9zfw^eT8s z4jc=^jJ08wHQ&($fmC$=z#@$#A4nUBi=JwT;9{GRps&~2(Z~A#^$+BxsN`izEYh6B zo%k!!kY02{;TO=dW(D14SVlj_x6qw|Upn8XzfYq-Rz~s(O@Cqq$|rmPjpdY3=EzVA z<=JJypLe|NYW)ITRc>+Xg>FLBa0jkj{}{KAI)46h`yITG!~BukdybBvK1j*(Wpl${ zrm$Jidu&pd67XJ0xchCfZ{vAA)Gam)@Yi`T%`Td-3)SJ!Rm&()7_I*$paNisf{EMU zbt5>hm?&1Axp+968&v+7cz-ldkLNRPt0Dk2oGTSjkL`y`%+sF+g&tTu z=it-orKoD7Nu|P4rLm4vdwzus;2$4LQ4$e)n;Gpq$^mruQtogt_}rl_c8($ql8;Hc_h3K!^@!H^BfEMEpuY3;3#Ho&YZx6M`^R+d zY)v=m6{y0A8a;wyKQI;FJmY3#Zy6Wd-(;c6PW(^;6p8p?8MzEI*6kIT%8e&~r(YJp zGiIu^8lk7U$0abQFS7X05$h*rVoJJWdb>xd22iAm6gvTW((0Cv7SwJVpuawhuB zAVqXvGV`}I1A@jx_!wB2JF=Zjr1-H4gvgyOjE`v~>$|ENsB+RCIYl@H{?HQ4ij=_h zGg8@hq8xR!f1*dVxk}W-)<~+$293yZ)z@sqdYMabI~=@KCiW7oEy7w?dAs=f3g9}K z2PYl_`JbL&m48V9=V};_0lk6jt&gY9Eh2thdAE%eB$MWZRD_hI4v}f zo?i2|S#%Yj;q{~_Prn{qh1gfdxybvq9oc*hP@cf@)Q$8y) z>`qpw-+rlqSQ)jb;=_0Lh5*H%E;ROuU^GrKXox*j_G!Nc?F zjXv$SieacG4w3_yG;@Xs3zf98gR~ZzYC>0RqSP0hW-L&tsd9_GeDDlk(dQi$EDlWi zJSL7_X{M|yNLbk(ZAEljhyUPk8=Z!3s$+I)>6bB|ichuUobBUsSDF!v&X8_Mbi4Ae z$~t4bnqQ@vTBq0L+gfgAEyt$0Ps)e0ia9y0(=yzHC8wV_lYt>YtlJt->Z9nA#1F!Relv?frhXiB&L-Z1>QoxPnotLRCxT~Nr{f=b;O7Eiq?_1AG% z!#9nx%xOCnd8?l3L0jM`v*+Y)Z;h2_k(D zOoS6Cy3UOsIR)lgCHcOe>?Y|eO9sTwtB3Gm*B}88x3%GpWKICOBiO+7saMI%W`^X3 zum2pz%Nu9TcaMeH_?O`^luG79xw9ncon_l!cF-=rDM=z!zi@`$j!HH;2COFYy5U_A zRZz4?B~ocq5Loq#Zgym4t*&X}R~&eGk;B<*q8lZ*e^ABCUCo#V?TZ$D;VBf#735~H zdxkA4Ix6g>LUFM|4DOOB@72O$D;LxlP}-h~k$rY#+%O+|HYh}qlT%X(ITlJAEShj+ zd(f%0|K6Wp(a-mz4}#QtAs-VlzEuq@CiNP;3df6`qx@H=Uc3P}1Zc0T4X&r=3)bA^M7%_6iQWks`H0l(P zymdM&y1ATblnqL>mEYK&Rz!{CtR0)smyeuB`TjBm(0TKnjv%^&Lt#WTt4{m z?Rm4{Pwc4fhpX4Go;b(g$)tk%s(1-Bs?-JO&bme0iH>}f-0VDH zx3I;GP>HhYe7@V1d&+o-ftN;^4TIkd?cHh2cAV??6f6hiVr6@jz0v3py%FMGaqaaL zbR8JQ9T{#B-4tWB?$4x zFUR{ilkuw6{Dpk%j9 zsM>kvlD`bLTIo&p6}+6ak*2g4scf{C`|M@IM%yVhX%=a&mRbjCmKW)`PLg}hq?>nS z0qsNFRs~M!;87-mG|Nw~tb<#SiQh&(!``WT9Ik=&U8WuP~Klz0_rqjP6U=Z%D=fVtde? zkeMv01}s5>ZZGuMWm$>+;zbo&lYlPF+*IHVE$L`t+)KFLW_Yc(LVRL0^;Fw8!o!2N z1>D~Ok$U$t=K}*!IBW>Xq7A;!@qn}(uoRs6eU4@z4)vm-yGEiN2A&bXk1%Pfy8|Jd zNmpjZ?4{6d&y=5HU4jtD41zE^?`iyK?X|e&9&9i^Iz;ezpC{ukPZlrtkk~= z7*((m6+tqE{p{(I8q2)>bhm3ynX!-vsVzwiuuqw%D6L@PP#s*2>7^(EQWYxmz5vao zc`+oo8&r&wTD+H~sNBP1U(KA%imr;Y_;VX+9!3f>znj-W-MiD>=yv2sc$wYL4(-@50? z_6fv;RluVja&Jc!D3;|#e^MS&6{V}=3A_bLwo-bEsp~=Am=kH*l7R3{9@)xU6lMY1 z9+T)J6qQ_3BwobcZRMN;F+eY=R0vlKmYn4}FpY323vsD3SCR>_M6N%}p3=1lJ8DMR zx~r@tx`V~OX53^-4**t;fXEB>{y&-AytFN3`b?M=0;UZ&isK`obGZH1R` zq`+H&J8fR(`+oqH?qp41SyK4f1%+I?uZ?TmwC{!9xj#-Iv+)dq#aY5YH}QaO4Z(#V%K`yiIq*X74rI zp*+t2Fr@urFRo9mBsP>~bw`B<%7FLWSj>Bp_VCH+jt_#u?m^a+Y!U>?N}cnrSiT1` zG`o43DNuukPaJN#&!=F{_FYjy@5;yu|ECNRyt_700*GL&+=<7@U<^v9e1sQP4Z#>? zNQ3!LF}CBtiw^!yng*dEyd>kdh4k-T5O#{|Sw()9Xxeyv2*2;(fSDwqIPQUUyqj#> zl%?GVYdac~Ln=5HkmwV7qb@@1e6DsrfMC`M1M{Bh^F4);fB7n~@jlAkqnJG4fxWqEX`h1rc;umE+rg@5Gli_W(ThWjq zF-iT5VVnsQwu}}v94K)PJbtvAUdESj|}-u4`5utbI(9}r-lCYI9sucQ6#2_qfR){D8mpKi3ru--kW z!Iul3*Z%jqw>1j~UgwIW)p@Q%$X4S(6m9LN63?rgFu_kaAX?N9UY@kACoqeI`x;m$_qqOz0baN-H8qHP zpq(=0l83fUCCK1Xgt)Mgk*%v*%x}#rUuzL|X{x!VsAKX9&d2jx(^OsVj4s%J1 zI==Y+Sk;Yec@IcVJn}NidIelj~Irv=+BwPeASrt6a7x(O? zfr--^_>57hs2l0Fb{H&e5K~K|`?LsH(yqyxk9AqT8WIGPX18Jh^|QbRBms4 z7`m0XUqpjd2yK11tC@y^k_>LaUv4V<7*ZrF(kCR0DY&3b)Cg~{%;*nx&pJn~*)`KL zB~bmTpv-!STm6Z|ljx>ztu6Ecgi2K)gEgFw&_3<@Xd!>=wHVT&v#1@&u8;=ixLn%% zQ%MA7f)TKq(r&C=Xit4};j-~z7D7(hYux_l@g}4}c|`VZ&Yqk2E;)GLp@41_x2CKX zNq`U5qp0+84*VtQASc>MtEj!N_(r9$e4u>>$xB&&A!WyE7l{V#m!ypfW)j%T;#bTm zqu~G=hD`xg90Iu>W8<1kuTAh#)QV8CfM36=!j{&ZNC!b5M3-ty*WVYvl9l8re|0HN*i`d78a5lQr9Gop;iLCS4|aK|pX8-nrjA2WH8gA!wUl?`c~w?g z>ANf@87)pE)S95T9pz+Nx{XP)C2S1}0Z-)n!I|!kP6I1KG6rnwA zNHZ80N|1t+ci4=V2jyk97KA&7WeM-JVM@$Wx+o*f)yvUp65X{_NryTSe`mDbBw#G~ zWbH~{`X)*7bFH%N?4~n?W(?@bbA5J4N9C(#Y9Z7_4ZA54PZ+_6Z;fw;^n$LzsD`tL@usAG<>!hk2hv zZWGfYH$kN&me9x;WLY)SLby+B101O)@5BQh zKl6%h*a9O(*=H8!-Xcu=%SSG_(@D}R-mv4jEo>Goo2K4pPk~axuu3YeV?^~SkWK=c z4ug86esi9z`BkGFI_@bj;$;Cd49GFqqf$if8}2ndp%1R8Hl4XiGTrN%NDRK{e}K@< zL)9hXR>2fC!BzNp-m+$fGGFtK-Wr+wfY23s!7Idxj8PAJ$JBe18&^bo2t#-`!eS{g zI-!!3v$mabZx6qdyqVOkZuC*JNHZu39N5pcbPO2i=#zI%vi%_(YCR6UD)&xHPzdM3vgHIjyK$M=w=qS`Hn3Yp#}~Hm8H6$> z1fH0NZ_+}A^mQHcAeRSDE<&mowdgu+v-MAd&F_RoV7n#ZngIXWde=Ja*u%tPrwMMB zq%rmQmMa#?E>d--`e(KyF}TLLYN_GfTuF+=5?{M{nN-Tn;`au4sZ`@sC*AD!_6(66 z5doI5X-dz2pE%7h(cxjg2mDv*b_UbK2$qfMQM?hyU8gj7ie!g70d#DyG8CL-(3;zb zUr3+f4%@w^CvC_|6@cyV@@}_Sj57!=suMcd*wc`ddk9{~k*MvYGWEJOH&3td@$BXK z)k*}_8`g!fw$g0T60d6UgbH~@WL(ar8%g>Uqn7+;z0N{xI&f+z69Ej6}_s0D0vJ7a@~NW&$TmdWoD+&!$$%+_dM$D;gxy zC$-yt=8z;KqBE4T%pU?oc>;6PjgQLM-l=3oOYQU{?*_9p#KR$ZZ*TjFwD0g65f4+& z%T$vLsn5dl^ZVlZLar_nif$~AiM8v=;h>QUmMfh$_{sUaKk=WPcd%Z@mW74CDv|1d6RTY0DIPfzzcvPN4Ha?oJAf8TIb zOAlT+7v@I{zRKt0QieFrJ1&@ZH_B|+J1UCaz;WtRAJ$SDED#lM%UEgd8hR71Q-bnKpi@JM42eQ-=-6`YhUP85Z&hVa;TmEn{mf;tUmeefVXFO`gc zOu657KrF@Ma|V04SxEDbxkYf*{K^EY`r?a9G;{K{-*TS#juRVfe{WV7;v^(QJ`#3S^%6AsZ5_Kx98om?@@ zwR~(0t4sClS*2{tT0b(9UlPc9+iE=mE5D%R*5Owt}}4gj^>)AKv(o+J2gE zq*eMRyt^@`H59nc{fS&4f{KvuMw+XrgVQaKw_VlZj;$fP>p@A(Y!3#X;**%x3tM5Q zpu3t1D4&{04o?G_b#VIFs(U%rR)$jymMhcR%TiMQv#YQ5r+R6dLUp{KjUu!yIVN8`5-cGq#I5`CI|X6{$ux_;i*sXC zY5Vy37a`JY3e1S!13-cg$i)<(1`0(66#<>nITGL3nQ{_e! zvBqt@>cvx|=yVmL1tXHxMp}5Xg}x64++LYpG@)@<7;-HVl8l))G*$FqdZYUKrQWyJ zZ}+nk4rEQKQz@aO?=h@F<5JfVa&)uRANZn6i;N7W&T2$G7Ql9B*2^pD4!>r5n3@Xjz!@1`=1 zTk$Mg&($@py;+R<zJP*OS68i`gd57h^1;V1e|hDX&wcQ-D|l){`;o%@HM^GNVz>B^gj$gR1PbS z)3s1tYodx-h;>@=?p5Wg!JxaRq@NItxz{LzCs)ShGZ6L*i*n;8S)mD&E}p>5sma;r z>)HB}P~@feULIsFMP-oA)6`z=xTjs31FeT7HvV|w%O?;*=~N^_IwY%G(uA&~)UMgj!)?)^U4V&NXDE zZDyyW!Sz6J00wD=@{=aqsjAj#rWX%cog0g6aENNq+;8}9o|C=%w-1Xa1&OvyuDZrs zr1%I_x&A__|CHXr)$oH4-V|6Z0rQeiKIppq#dm%<=`|^*TWSO}iac)!4x>ciS;nOE zayNL8l5O-0mBCjJg(1$z?SEpwG+oGwVmFG8rvWWrr#5KVY$>OzL{~Lka;Or~z4wvu z80A4$4VQOCd6NKD@|b}YL_@*(LOJDNL_-C$cNx^mWK14f@=#nF##B%d%09ftT|$eq zm8h}|<)b)tsF--v=0+LpG6xxZr!Ug%)nFqW4q6GS%ZwU(Q}ZNyINdsQ<24*UXad$m z%Qr0bAaRycjWaDW@)bN{HqEnJEpgHuTLsm)^*=kNoWGRmZ6^3t@JySmMQt`DV|jPp z3U%w?kDtk$vuPI7fgI#Z+9#b(qe%XtJy}q{=5P~%n+T}580^H!#k@V5)V#Ga<9STI zdGh@n9}<{V&nf&<&?IAX-j6Fjaq^AKF(ksiz^Mvu(GmCQ&j>kW8cdW(r*d7zraNgB zG_<1d={}*e>d+@b!ya`o*{dn`aTnxR4VLSfxh1p5j1HzRFioMBIl2tEdir}}^ufxH zsI)ku45REzp)$=(ZI_m|u0jR|T3=rpLfmVrZXVPtecZ9n?-y4eUq;&D1eKwgBXM&Q zD!?S-(`JsgR%|Yq=yuPBI^<97P?GlcbI=~Sd~E0J&fMzJY5mYWktVeGyX#l}F1)gW zx@JfeN&HUos5_g=wrpsUg+Zj0&Ncs1*O|BL8!0h-rv~~XEW?vlLkHFRs>6ZN#NeA> z*|5*0htM=009?=E&hOB+y4YFi5NT2mYurgY((wNh9Ap|QL_zTnj25@H`ic(rS`RW= zZO{EXsqL(!+~VBU4-i#MJb)g{w%1b1YtW?-1Wrr?^ZL;P4bsuH zf5GSt#>u+*R5S@_Yv4Z&*zW&BjvjXV%-$q!Su=uO8|0AL%#fEspac7#_|g&2a~ke- znlgo6_u75^+W$-kJJ70EScRn$}2IoGA2<=t$8bl6df0w-`4TVTe z(h5320rVdaayjb~S@ct6@8>e+8?H_IcBLN{F>-ow;o+#)>N8pf_dheUXuhM^Bj;J5^TN0skcOOl`P2TTM zeU1`kqqPtFZZ9?u0f9tU?$QgBJ>i%n4R_ifET(|5_HLR3#;bj>1ivY)GRx3yk(k&2 zu)rm3btL6|R4Xf!LTNQ_NeL$=*Ws}eEx!h-75NM|z0|v{`mOed?^E&m-h!?= zCU%95p1Yqa^ie4~S1@;LC+q|BJ#I)-ZCSHuj26;ESPZMN9HTZmb{mvv#7PhgOOZ=9 zxs@2YYglR&S=d27a?zZ%xbSrxLFbj4gD?zcEUG0E6!?3dof&%nV5x1;$whlDqz4cHktlzvXXWFo4cs$n?UH04%3fCiuG{5BzoL)yL`C-md(R1B5d2k=$G4DL^Jgz4eOAG#V1*EVP`Yq3cu z9Rhs-pTr!f{^UYtFYGnMkl+?etSrd%aDjg%v1Ql@EowU|`86v-wf_oXDCRK;#S z?$&bfRD3>Pxp;9wc6W8C)Gb%Tz4d40hOV8*LWFY+L;rz^VEIhaj?r6F3V)MR=)7>^ z;&7d3H(elA;JxKbc*-^^X(T0M@Ki%-glS*O1$?zZ&8F&_FF?80hf+Y}1 z3x=54PUtYy|0c{~Lm$>w|HHry=$)yTYuZ&JVz{$B_?xY-V2>j#XCm(u7j$%_8HSxJ zQyZN$zm($wnI1P#Tybg8sc>0?;NNM>{9|&6++?b#lCY^G3Gw zDE2D#Pe=Fd5`pTOQL)NZI;PXh?dWnZ?L)hcl(C3TMR(i_d4!Q!seSsFqKBX0u2lV8 zr9?UP4@j{?2dl0s1p3dQs-ok-zgE?HKUD9fcMS622)6&UwlAsBVb`iEde$< zY(!S*=woZrJkjXU>hH!MC`;cXZ~**U(qr306_I0hQ4r%*m@r6pJr*e2_e{bqbHDw= z#`sCa_4f~=#y=C%nj~Svv@MfI+A1K z7m zVBvlBK3+>>1{>c+(ON5QLOR21NkG<-@vLMXN1iyaPC299;l-|K@b?axGwo*lW>-u} z4GnRLdKdGg*hXFa{&MP>sHtfqrqV0md;OK3KxW` zlZ7QaPsYh+iYB$);8vsUAfAeoYy z0R(&{QCF;J$nfYv4T8-{-AULz7^XMlY(b&As$5%cRI9<|ulpfjBulCBM@OklZ|ZvX z^|;?;a38rgIlugif<6k7qx*Tjmiyf_x1(%&1U@6>#5WfugCt_0rfUf-lb5Kcq-dTm z4R0=;`RtO9L54}G<7d+w8Ya6DWIm1DH7%Gl07CY^arpM*re zBH%~gABny$q(+miVDax1Y5jvl^brHaN>QeudW@8jiJU0KxWJcvtD~zb0rlVqC_7r7 zjWzf}5c-B%YL<3HEH%of_+rigLj<%M|2v`5M9jO+%SsFHw|q;(pX-j<>}K-bb(|Rj z=^Ut7U;Lb~x*HxZ|Cy*=-+vfsze(k8!Kjv;;FYobTzxls1b1APFPFVb=^$q6ueCj! zfP)9oUg1UE{UN+nCvp`H(yGEXm@h1t(uCt`-$>Ok`&{re`OP)E5`=cGR)mIouU+$gX%(+Rof|JL^#`_E_mUHh%sbgno8 zHYy8z>|tBgS;1{9K3Hn-s-Uf1MG-h z208!Dn^I66RLd_Rk3W*eKJf4$LmGG#Rv|{JW6^P!;ewf!O z^jX%Pz8bFJ&{|?dK~Jqj&EubSN_EvXB9(rB`YEtv7-e^Z#Yw1hq=i@Bxn%TKQjJ)6 zQAvdhO})JShUwrn|CP*yi_)^AIcwxK&Rx3O6Y2I=Hq#@1%QbjH* zo%%I7YEQ^uNlm+@VAp$%YyMH-D|16feJV`L^$guQS&3sd}Xx{Rp zWu~EQ^ZV)C)bjNnRsxyEX(|SldC~%hPc<=9C9J0WEoLd z9Cfc8$r4z~oREFF)b@|J=e-X#!paChmj4ys3S**AXVq(^b9=cmhM}vfA?gbM{qDZ?55_RGoM2C5mF67D z{=<+0+#j;4mpi?ub(*>9%(XV9Q}5wO$k~nbdj!_f8HeZg)n99H&i%t(>HO}eQrK7& z<*=Hutl5z>xg&n-(IxUkH>%HF@?8n%-Np9P!M=mK{-2!L!VIUMQw;R04_zpf_I$W} z!VFzuTmF`$j-;60mLx&6Y$z><8peAE+>(=YG);aRYe3npK zOAj?wTJWAMw+c&5=@0pAIU5phHugLbX`#rZvgTn3%3>% z7Hh-Fa2rH6v@(P6Q(uubzIhH9Mv}T4b_p}Jwz!s%hWbmCNIDc`^3~P4Bq%{jg%%hz z`T5%Wc?XfymMBH*pdjJ zSbdAO&dKn7HXd(mx73F^bW|GhZGX^VxM{xXi(_NrH_F92hIjV+0&(~DEWXjHarA3) zlINzlW%yYl1>vRZ@YGOrY*>fiQiyI1pm8Gr85?ot=yX?wL1A3464|LhT^y_pOq0&5 zP)4D7pO^ibs077f_Al3#a~)m7vXf6^ZBCmKo>e%YtUONf?J{i`RA0&{PqQWXpOuyW ztOY`r_UJ&GFPZ9G-a?gXD?RZPgFpv}kkOaQp_2M!NjR>Vn@oup zys9YAfp0Unk)kAjkT$Sz;D{Cx6Gw}-)fO& zmn%Kw;6`Uo?2XwA^n%-JTk4XATOt$sZ$m-I!p-?sTOa`yUx&RtpSqrn?MUd$Z2b4Q zuH_mys>hMf5ToyPLWRtmokCaDGi3xOfLrs@K^31|%yy@>@tR_kLcl9oe*55!oAq9 zBM|bHR*Mt1WSGwEM#eVPbo_;>d~J(} z>^Fk&jDz23l6aMI?G@>~{S$F*J*E{PVFmWvH$TvNQAtw6dd;UqqaIt}rBvY5Ug;Bw z``4SI_MA}>F2)FR8MJema2VzE5@EKhM{pO!)CgTFiQ9++IPyD;?V>U*)1Z*7W{J#6V9DEwl9cO&pQS!c_kZH8 zI0*|XDXS3aiKkjypAy-Zv@obNy%w(Q%?`TrowtyJwPj>$@`s$V&r7)64TlPrsWd7` ze{qTF^eA|isrO+&FR-x{(aszl%q%V454F{!Cc5>?05 zG!v6%epYG#6n^@bo0v8o?AeYxp6h*z7iiuo6!|+$z`^?Eu_)odNKr{lky6uHU3=QN zA&)bV#7{2@|F!OH%4kk<(hEv1FhkKnOU>XZl{GPj0=wow=ZRKhqwh90J+GYojhBrr zg>i}Bw0G>7WaMp}7UU4jjnD9!#~#m1U2y5=(W~0Wzdnlt%6Q&wfSU3Ln;6JdHh@5tCw{wDLW}SgSM8bYHt+ ziTv58$4zt<%!!-|Z3y>9-GE)Bs)Rs=Ojhn>O$au|1^f#8vbLX_9rU~1#24Opx@`Ef~$W2VW>xLHYWROdh?5{!2e0(s-M`LX}5g%040x) zezrhpZhT0ZT($N+=2X{nSyFdc%nLp8N$VkK?S^Emrt)X-J+N+AqUu@siZ4)Ame#}+7 zR5Q0`IwiJrQ-r+)NE7?o0YBDznzXZ*bhEU=bH6|Za!FHuJ*A|ioJz@-2L{Yy(Cs1@ z*NhEU;gjl&V)Q9N2ihNpMsjt{i2~O9SzM5b8+RRI+y1g6rwpmhKZ-@RM zFY+05X`I>Skt91q4RYH2C)^hC0xRR&?59OukVoR>B?Xqt7wMvk{ml0d+OCsDVaSRT-;j7KKFQkY)b&~Q;V;B>AA<8t1Cenve{?k zb9^&rIKYU>;S^{f87U65#$w_$i|$=c@c-#F>W($dlXUh__L^>%rhbTC5aN4R{j_R5ZgcTr<&_gBFU+oYY= z{#%@z#|jv0D=&#pp~;uNplzZO7L$5Zhs=rb2Bd|xsBD!lllsvY*%t8p)WMgnDfYLOrgl#y|EuCbbAJb_Q8$+tYMOk_ z)ZvZf#rF`kNE$FtQ9xnGRl*zZfE$Sp3#omYqL^sH)p50--apKAPVmgfhsk_M!2l77 z**B@nhaF}8F8)Akjk@+*9b=5TVdfw|0N*=3sHP?qEyyW~>ue2~1axz~!Qr0_HE)Si z8*Q09b;68Fd5{P`tqgwgk3+GSQY|FEvJOjB^$;Ke<*By7H9`tWy?Uh(bd-0wVu}%MRt>R2Pn3$dch%y4nJz1Dm132YfR))qkduzBgLwx*HhzWW_^>VMxuTXCR+K%R z!8$j<*}Xn*&qxWrJhPX=s=FGSv%jKC`XW?Pg>o~RW~<3B=pUdW{MUWVX@-ku$tqa0 zS(F3korr|Bdv5!^o@OO^kJUkgjosS2nJb3`HwTV&OUH0mrhN$aXTu+J*OfzQNHvy(rG*e`d#tq;<}o) zVaAzj(K;smt%+prPp8s+*hYxX6?UWh{{`eg8^22n7I|{RRiW^njc<2pb_-fcH+MtU zvo$RW$5J_E5L{sV>-}qxkk4Zvm=dS6uS(aj@hFi(U9dRmB}A~3hP2(noaGmGVd}md zxzouqr|$bm20^Swf=DC|tWb`Or?2B)sUD=VrteA@Y5IT9Wj-OH9uC=>}{Ma34gFk(F{OYaNm?M>wnB-mx zKT-6m5leZc&SbY{h>^;-M*90!&N6E7)b{Aq(9{wDb6Pf6L0EjksX=?DYRIW^w@Wu14e~Pkp8?6Ja zxYaiwy;V*2Pg>@U`9H-?`+q-rrGlvgs29BTryH?cxYS7RLDWt<)=d_xKt4&^Z9~LULebBp%^LpzcmQk5uWW9bR%f@A5__d}VJOvFNZS!s zl;WFfn8qT7s}FA~lbG0k6&}VP!fU(O%=6UaeakkNa@(*m$XDoVrM2;JC` zAP{Nco8F`9-E3_fTZ-jw=R>J@r&iS0ZPo0KJu`~JcFE00xaNbON~mG%?_y<$=h299 zNk$vemuWq!)hRY*?IYY#ik>0JC)S||eGN>C!1O#;pI4lfidCT*Ugxy@Ila+st-?Q+ zw+E8XUWq2HrbixHcya#lua}_{smd20?vf2!#-T95lHt7uO>;c9YWnVJQobHL8 ziJHcXs78_5Y-^Q2Ga%x;#lD}dX)cYZO@OxZJ0AOZtB`m*Tbz$CPkw8nmJQO4n=q-3 zptOrQ{%LXVkHWhq9` zoMbi3&nLHP?_|HSwo+xdhym(rChx|&EvU!ZkmLW{779GOG$+-$ zt$3|1;Nc-f9)MR7dGR*U0T$NsG4(jC8^4J9ocR!kQGSELBIT zII0&q-kWQYCDdS#LtZg?;%!pwA2RAdJ&k0^bv(ShjDDuHSbps2hFel!5$o6fKGUs% z5g#y*O6D*ABhM%z6$|NH0m(d*NwkjjqK!E8J0X@q8<&12zq%)M?StDjpXbjv-gY_n zs_L1mQH)}rJ+?dP;Nc#G&_G!7Yf39T4uc-GKwZZI{KRIWms@ONK9!8;C$Zm7F{Z3u zxVOeYBB;DksrlRrp!d^Ca#++Q*^ehbTF>6<4x%&+kSDz%yhR@>ilBksMU51*vBH7e z)YrB*ntjo?yoI5edZTp|nkSvN?(Rz0Zr$)IKQ)-+H4-NoX2n0wNcd5c>q}0>e~p$B zGmXx6ue~E&qbTl#KW^wYa)%B{4DuK0n`qx9K>1h;U0=oYIw8yPM zr}%#2_CT=2O8)?OifW?V&ZR9(K(5#r;B?JOA!2fRdRL&?_zbLGWyVjXW$OA(rHCrZ z%H4)*NYzqyF_K8+ju2z5Mowzfk*_6sd(tEupOIZrh0%nhRoYJ->S*8$rblhDuU}JF z-%4eN?J|!p-FV!8i24etCRi@vv@ZdF#=fH|$5A4i8%pXQqigEFA|*zSEW1!Fm-xu-3Rx#ynd zG>U0Vh@2|$F;(%OsN%ZIuMNW#F{Ba+?n@VCU+#APl+9AYJGkr&cIgAIQfQN?Bf0#m zB_vc)%^>-J$5V=xqiu>?ct+`w$I~^?-e~H&aMZk!Zx-SnHaz38{CTQch2NbVR)LN5_rd6Db6lY+~lcEv$^pa=!>8WTW5wT zSOA0!p0($aX^Hlwh+IYqKYs@uD^aDjKk~%1fMX@NsvvEUFi=1_AfC0I8d9ehsS1lo zmczrw=Iq5hyCdxCzj*oziL}_X`>oSI3V!1ZpVGRCd`5#uTX<|ENr1?1!~X!UT(!m7 zl2&4&i9Puz^QooHHr#}%wp)EqPda7Yuh>YMd=UMY80q;}Cn!L;k0c)Fv96a#(uT8s z*6%l#Fi!0IR%WANZD%a7%{g~R!tDNZ;UuME)GT47(n7_ek(Bz5w8<`JmmXw}`2O+E zYQ)!j7<@~;+P@SBMdcgCmyLIw!nRV8b;j6!6> zE=U1KKaEnRn)izOxv3$%)tVD6yw1^NhC~aK@5KqDrTlV>%|Kg}KXOdB<%;MuDKzGE zmh$#^rS7Uu(e)Xv{Ym^cJ*LQ{h|AhQoR6(@&USpM89Gu@wT~#g(k*S}*$vUS9;$?Y zF;Ja2afN<8#ddcV8f@N0A*@}!)`KF9Q&RtvJzG{Sa!@fEaFEEBFT?MBQS3$uip7%cqsYg|pnpozeI7LHJbyqB_)-lrANlA{&>EKEBE6Z1nf`y?z8vL;sr8~r z-czL0FRX)2^*j3u&d? z&AH^b89#;vOmv6`{Jy8ZTpDT9Ek_3SF`us=rDN=2dy?tXPjV}JO&UEoE!E`m$eaX^ zf%=Ndyt1<>zhW=E9I=-btlE8$IR4O!j^rQks3f-2(>T6Uo=E1cK{udsL!m%A#{$C8 z?%T*<*eZRhwauM_s?8fpWg{{pa!K~B2Td)1<@evEN1Y#WCi??EoYQfzH49k4dkm6S zARd5@N7kCQhB6p%!@et{-=g*Y(I1F3FZMf(E~yAU;Qs)HHq5=88{CRd6zWDOV~EEP zjRH2rA2|1{Ta8Bcc5x)~6=F$U*cd&rT_V`o$^NSL(f%lY zP5I`b!<|`>@Qm+geM?F3o!qOZ$XZE6uCkPoPfVYewRw)cs;;Xzm+bIF+ymtT2TItt z)^#f-^459egm7b#$pm_RD+1!)NvD=?C@Q-XwR#FzdM*!B3Yd2jd%@a&hi_eB)t+^m zaR34Wn)JE+38Gy1ZLOlSefzh79B;?5>0ThVvKdvFxLl4w#w*se4;JY5Hj!!aeX+nC z+`IGn=QWNpl~?aYnJRCwXT+WvXVR@;((inQwTlf5sz%|-z&Q7>B)D>AY>YAztU3{1 z=cD+J-%-`C?4DbBt&u^sc?aG&921XR*A?*}NQHGfJG6orgU9>dBmg&Y=~08LB`!%1 zmiiul6~wanP;gIA&bKc#8_gln?(Lm3&fq}@@~8M?Ot92^MQdi#5Z+*Ho&zX6*LAOW zZFK`QHnHE^l$1P^&naY8<2VGK!n*3>`6^h(r3B7K$HbBOe`k?P!Z;zvO4B+_cagQl zgn`qki#;oz*7W4kZyq#Onk7)*Xi@SShQnQJIT7J~jB&v~D_2w`q_0VyZEPBQ8?av8 z>Cy$1Hq78;aa|6dtdhA#FY1ldeJZw}f2a6?KFvPWl#WxUKGhBHhcE0L&1y`s5AO8> zqH>JA+KSBJl?hYT9R<(W^=JZIMS_0p%hgSDI+l?wiQ6#qpg(qH?e(Rz)Ll*)iZ?>0 zwZ7Fam5CR8P1duFCrRBs4%{{^$3}6NT4St!Yz^i|+^EhfZ6{y6(wncD3E`lZ<_-Q8 zrFm@D>~1vq89iZchO(MVOZM{a7wnfOqkjplj!jpa<{Bm$s4zCy7~5|G|YR~XLI4*D_hLkJ;1lRJezq=0s2=f zf2Ce&vzV@+c-!1?XjK%wOxl-HrO3^w+nYgV^e7~}YhFzfUJx%fUZMsfC?J7>6)KhnDY z02SQ9{iz+KlH29r4dqSTZejf^h1I+rrQU93iR|S)89rntxoJ~`pstS|Q`W~tZShM$ zx&TV@z{mT`$LCj9#9kcKKjre)T=pay@x4#Mnx&Gn#;J1}^S(C81O48$&AFN)LvBJk zj1h{{*d^4$bR70}K8pJXhwl0NC7gi&0I(_vJSE||y6Lv(=z_j_TdRcSCy;#_tjDiv z(0$E9b^0|cMRhQEI;&|O;Qj&d)SWDqf8B5XwM5?y{5IVnkbN4whA$cTfzUF^c#P>Zw{4((IJS?OB1xh{__ZZJ{S0WAGt66^;T2imxo7KT>VW}s7NSKm-7~|)fw&4_-jIhe3mx<08*l?rkke4!dpbYMHS$U;*D-!y{Jrn zhM{d|RdbfUoBsN$_D%hwzNb;2LvFQ%f6FxjqJ2)vF_xY+_a?r2H`kB%uBAWlDvZr5 zb(PQKN3ctF3on;b-{tWxjR^Vg;2-cHD?Wc6XfOkdz5f92jQ;>y@a@ccvHa;UJ!+L2 zea#kK-1W=Cfpj$YQU6MPI)0 z9p&CcWJaEg*0Cgak&81DeNAai&v_ZtbpFXZPKZtpY9_d6&N!&{=8gqvsm|DhmOafO zUgXeU(wUM^CYd{r$Q07lpL^*}365K(7Pup@@D7zF)!g^-M;y{x4Z;F>cJ2J>o-@)l z8w;4P^)?3P@E2vz-S+)x@HDn^UeBh_YvzlDEU&+5Qa_z<_?Jhy);ue3bVR1%B=W7` z{{UzFfmv(Msg|mDJk8rJNxH0GY1cY!(YCsbtjX0u%~b%CfDLP8Yb4&Gx{#ha;-pJ= zVaFckr4b~3a_#6kQXxmV0>;nVijiRhrA$=gEr>m@EwgUe8NFzfU5kX=o~KQs=PtM+-N-ky_ ze7M(aL+*p$+}C1>2{)+cyi~T>33gHP1JHZcd}2$7!u;9q>+48&OyeA!@GEx4#cbj8 zWXARx$*W4n)Zu-?7-Q3|gE#uaS(g(uu1Ptm?=DQNdx2G=jig|>#*0SYxmPTgoX*F|6e(n}T1Y~2stucmb zVkrWT%AUt6+=t3Iqsb<jF4mPITytb2m8)?q z69`rtb&FzcY=0AHsib{Abi9fO*bIssVTJ%Ar=1$8nc^@HHM6AZ&!#&^3=zoE96Gq% zO7|4odplcu>0{IhF$F3bHXVIw99qQEt+wf-VSvmwkbfG^PBPJ=`7Uch(_Jmzprd4} zG8LZy4hI9;t4#_I-&<-*W|aYupth2-=8Rjwas}FBV zQy9)b!1Slb6A!ybvqgI%v5c3<;CfWcl!fEkj2oPk$sdhhwX=JvLPiGv0DdZSXl={K zA_B*#%@(lziHxxwq=CUXz~Y}AWR1WY+q7GF46#alP6Dxo2IQ#bQU*E5q$Y^lBXa0#p0eWU=HB{CLK z!lCE?0IgJ)8C}c(&nFeRb3M6*0JRBaetm@i|;-N&_8HaR>4 zU6N_>FZ_Iw^b{7cfZ`iaU;EF!GN9XcoIpeMNNA?zoeX8M`V9wFZMc zz&jv}f49XP(dxzQ(%8(DK=MdPkZ42TEe!#}PhFvM-_UTE)8uJbDx&u#x`&)x^#8TC}=N&>3}D$Ne;HQpu>? zzaC)&&{QI3&PyEqXylV6$h0)*v6(L8Kkl5-_Dc|Pz1#4`Ro_eI4_a!Q?k;9tMYM~R zgCum^GAh$*(wM|}MM2wbV;OzvrhaqR%8ZP<`Rhl|g@mx+bai+pgc8u-*m1R6Zq(2UBV9)n{fAFqDG-i%=i&%-B+;>`R zoEV*vUW53{X$+1OEW^tIx1Ap7j3!HdXtwF!~y$RhF?Y?IX9A)I@)M z4a`mc)OjDBLM>6wQgm-^+}As4;vHJg3mxO^AHdv$S_$IWByaS+KHg#6jtTv0y|Y`E ztJ9X2H5wbJz-+L`-6Ye+)!~_1W@%MMHnVi7Ew%j`UN>FYIqjXx{HpYFMRUcq{iNCL zDcDcuX~wi$*wIs`BCEPVGTd9l@l2;G=y?XWqWFof{ghkWtT$0(46-H(A6la)nF`|K z(ba#vhj{%evHLW~2;hIbHlNn6ai>x0XDlj~#=ZXljIA}jANw-%Pd4bhpoQ`??lDzU z!umd?qoUaAP>oO>%#Xs6{{YsiaogI%IC;S-=sMR)YpKT;mn2rtY|${mMjKpyIjrSP z(_Z1tR}agwIO)7Stm#sulSE@een}UVC)&Fo4EW?Msb{OU0QpvLPio~nLv*%n6|JY3 z0)XU`o=G~`l|}OZ00SNvS3~#d{_x|ehl^Vt!F}PqJ6DWMupnj`62x$R zwVu8jzSGkR;^HCGaKRPD*m#3g@V&{2a$*M$1i#SNqG+Bq({+SY-{j0kG01WGXRUC} zR-&|bN1am**%=b)(M#4d8~5V5t!Km1V{=8leMeDVxMG6f4agf6oqMG2C7{|zbv1Nx z&Mwk3k2N-UH4gjsE~H@*j|&WB&lwuHRD7Ae%Wu z6a6@@R`bIXteJ}s&rx01rCDAs%GqM<2Rx6VtQ(7IU~;mdL+RYrtnk*B z=_%_I9Mfw?TR7z#t=tc0s#3)9oE1M>E^|(l;zgQV(@r~4%_dFIx#o=0Qf7jX?9oY@ zC>_s6_{PkhF|=f27+5Dm{{T%Su1d$mnx&2vW-gu8&fsgU_{qnNEeRY81kbH* zxg1wp2Z^RTR_PW)+4>*OwNq5IyoX_E=Tp`@f%6Zc=Au+nE-rE0^DqdbHQn9#S53S| zdsz*n=b}F={{VegF{tPohMDtRKw2}vamcP|QgJm-=3a1V+o=fJaZ8NU>|Y@5{?ws{ zYCKZnnq#+-FpyM&P&?I3`%n#4%1L%FiIk7LDc)-T0D7oW;DJ=GG>iUn`qKH!>r@2; ztyP&;FB#YCN1iI4DZo=}$aXCm{i!_jP*Q!)On3j;+z(oRkQ>y^Nwf_z$c|T zII`m>ZhDpH1L<2#x(&Ua$dcyj;Ep73kyQNwsFHWkv|?nQgO6HJPJKP8<~-z|UNOx* zA!a$oed)Nn6p$uQ80M^{yp4lUZ9QPWOQiua7@^gH>1G21Xv{7gds06NW`Jc$__8L~vm8qrZ7ut$F!pDeEfMoQ|Y4EK~)tfP6Ge;njDH#}aE5<*qbW7q* zYIU~~UPyeO^^m%X;MVOu;+>$F)*{o2z{yE4g9wa6JVx z%)rEAo#lUg{uC0c!iu#6HrUOC{LFczEC)eD5_{5PHL9jdViTG{k}xUaK59UK3bg|_ zA|+2+kVX%+O&zqdebxDWYW2O0X@{HUBcRSd{dKHqx3Qa=E8191birawsp`joD^Njc zr+lFOqH*8&R+r6|XIPK`3BX0;^c`!cv+zsHq)B^r<|a{;hz?u)Dr!m;m%L}r+c=3@ zNtl)-`_+qmId-(yO|lg4-M*FC>-uhyr@>`89$m+i{{S9wjoxAPC$({R_pL0OoR5@n zi}zG|VycZ#YOafonHAm7nimKoh<(_-ea&S^-eY$>3VK|(k5O9 zMfI)iV_3%HwB0*m`SUIq@&~0Q{n2;Wr{x}%ZvN*VA|I_~Z7rXzQjPUAl#*6N1Z3m2 zX=#usK5Ov3iKrsBF4a7;?Liu!k$P6P@+KSHl0zlj{#-z_b_2aunkJ2iILJ67(AQC* zY3XGeY8OLzZ{Qtq>sk8LaLs)rme90OpW!|I>ln@R+S(e;Heir_DW~42k`g-_gsH1& z#l%r2IOiW)VSFwC??^>7ii;q#4aSsp$v7jSq{TaKNaHk{MWo3RX=6D?DBZ}ZBS(`V zL2eHu;-!mhaJ&-0rHf@`cR3%0O9H@cCG#XZfxwEh^g5%W9xSG`niaSJd7y11#ps@eH{2kBMixRv8nX3EC9R0F^ns7b&e1>>eDKC6*&_(xKhEha#f$02s)nt2Rv` z7Bp4EAts)&;Pe#FF>_DzIi+FqB50$_4WxoQty^0%z_fscJs9VjsTIeT!IuLg2Q_%= zvrBBY?YC<5YKMF1CqBcLF&G7Kq*eBnCo%r}m0!EGb*zs(k`8iecxKwe0CV1(h4&@2 z&Dfz*c*R{ybptr67IS0@zvhf}tR-~xCBDQr5sY-Ki-qR3CzFn~nR59xQPXmHiVj9< z64>p{InO4cCz{e^V%+LJ zSfRIVZ3Jc*fO!vA?O8?u-S8^f!bdwvHDWCy^5lN*Cc*ruQ&iQY&WcpyO?%H%FUo;B zoc$?Gt&B#1d)G$?g&Y9aGNKOkF^|LgX#M1#Y=gCC>#O^!=AT%Eeid_;TT7dh!aEb~ zS~lJqo0pa^uBOpebW&s0RR%Sl2qe<{W3TEbKe6u5*4)NyzofVJj&(iwoNEj>nOsd58A&+kn)6`s@7 zW^zvz4nSy|ijA%htx+WeQGwC)nfG0^sdQhoXTm2Q&5-UnZJqzOwyB4 zGn$OQ6&pPUY;!e2?ZqR#LeEm!4UJ}~Sksu)ENeE*n-R@KO)DCRnv0pRCYntlG}3BY zBvNUnf@!4G+>vQ1GfA2*A!#VhCTRe)wA|5}tAOf13T|l406*!#pbUM|RQif;X>&!x zQtiI2s9Jy_yO0k<^M5+kxA8+V?Y^{*Z}!w-zrwlTd8OQG^K~?hX~$!(gHzJ(*afw@Ldj9}C)X-edY}>8ok~Z}R6{Ko4xzikF^pVs} zWh*Wf_YCK2^&i%=gjZHF31iAPWyJ8<$-+NRE@j~T z4Qm>=RxNuPnj%;{N^T?abw~xk?I%YSVQiqfPdiSy4oC17+Qa<&RY|O%kQSCR6CRGd zt!KWUaSM5t?YIx(jPv!aBP(oWQwe3Jr(dP`%U02*5nOJT5&RPHN6`1LLD2kbrt0gk z$>h!b$vFLbSBoaMa3k_Bq8Blb;&Jm2#-~-bxWm}o5dEQ-gZLW8tq4!l^DE&xqv$6{ zZe#g^usscA-{>mib-R>4g1pa3@wTa=I>B+eNXJxuNUu%M{8y&yaVdgeCH^SSQCxGQ zR*$^4JpTY>;{ELoYWCJ>o6EhnSyTJKR!zr;EUu&ySlt_0x{UPAdYsnKUBpq;bKbcv zTf@^@?vmVZjPgLO>0*@Co3k~Hr5#qs1*%$J+y3WMvhx5xE!0Clo+}3C$i_U#U(Ag6 zUe)MvYLi-n8?IBKjfO$|>M!hFH&tUcoibWo9=Y=gs^X&_|9w3;fL)PE5B1~S(FhPGH?gi zJXdM)cHv_2jhC3dVy%j>L(?VX8LmDZS~-JT-b4%%$Qj%ZnLro-dsN}q?xJ~9i*G`V z>t_SBKW$mupPl1P!F_ShHJvz&IT1_dAUrlPnzM8EKebI12_uD*5fvN}Rp+*z8;6DB zWsm~cVSq=dJq>Evab3uSti3rEDqIfYsuuc+P)Vt7#`kGxY}@LSLJB0L$NWggis|%y zdS-|0O0dlgy<&66UZi+UrxL>@u+=uDY70TcETK3B>S9D#SK4JOO?X{V~0o<@2wY6)lImgeRW74yg zJENu)>C0n2#>+yFOA%keJKV@lW{I#<(~MRG_6)KUZeD2n9k@S@cZcjZ*fOAoQ~Ybt z{c5G=k8+WpvM4G2=(*3~T>OnQS4|~r967YP9QhCVQTBO2&yv~tR)p`gpnnz(faN}DiaZ2yn7Q} zTKLyhkNseY{{Y@WO5Qkk7`W4H<{pJH{EclH)#{0rC|gEyVm$MXJ*nPVJrw?Rt97ke zTu5DXs9`6r!g>77RyX#i{#~*3HPsnuse^)B6#TL}if~V_=T=4Kx#@5D)5YXUa<-BF zIsX6}IgqT%qBHnZt8XI=qB;5u)Qx>6endq23S(SM-DEU#1&J2cQ^^}Y-=)MTaPp}> zi&f6|`^2Bdfg5%dH*{J>hn^)LBxHNFA(7Z{c&4Ne*j1>Yg+^zcP=9oCX*nZB5;O-L z57w^Q>Gv0h_YkxDsL1^368g^6?uOINe$6S!AC*Tot9f$RG8}eP8$hh3DgBdG6t^z5 zgcj#FcX6p7!5Q3XuX$$%KF@EtzrQVI?2wX4!9KKY3UYG(!lT)@yD!??WL}d{o_~V$FQdFnT8YOUqMJBW6wGD^{rz?^)i)a_7Sjt>gV2>w2{Vt8kfr>u_w?{ zFmGDfQi?gHIY?MuIAlk3547Z|=xbW<#do(B?{5-CZwp|siaoi5>c4wW3YPbUfjrZ@trM&>S~dl4Ac6~u>;{3EqjJEo9D8W(0iDz8yp z7K7nSy>vX~#1_bYXMU%O)40(sv>Ssw68VvBA3OC29O9v3A$~~iD3kbi!lnyP?Hj08 zS-}MqzR*3*Sl0Y(O{-bQ6iss*49X5sKA!c<-)fSWm@Kayx#|OQ{{V$^w-<&|LX0n< zu79fgN%Pb;Q7md2n>@0}WM`UA42}1Xx6-jBa_^20`y!+&UGfHVTKaaC(pkRfM6)K7- z#PS+WNIcWl*p4w>FZO2Gz$5~C6W*xZ!YtbkMo6oS;`BxeqHN4?BVCm~pJQ!zAO+dx z&-$P?kIt&f+nAc)@LE_;t4kH@A`QPGW&mcR5hn;aBdVP3xvi(@vfX5%SlvEcXLdgt ziuXagEpXBW{mA>wcp|0Ob-R5RRWRw2B)LzR1C<83dyQuD=(nFUNDuJmis7QF&2y)r zv#OrNa>o&olc^njDx|3n(y9*XQzlXzCyqS$<75b)RG=|Gz#YtlsyGc=f}z0 z<`x}IQpoSgr4LfkYO7lo)^?ADZKKkY%^AsJNvh$GAI#I%j#M1&mILOUx}sc!40ssi z=e;Bs+7NTn|2wUae! z7WsxvXDeeFH9T*?Imhcnwhm*WwVQ4^tDb&Gtzp_UlCW5jocbuuZrfgAvQHqGxlYZhe_Ko}q&a?2n$_y7+!Ri35H1PeD3!AA2)SBmS zd~1Dh3c`yLpU$u+zq`03&2aCZQ(G!ks%s}>RZgsyuE%vYr{VoKAxcQ&romYko-~w# zwvnu3u{~>(E`PjfjN-J87Af9N<;H?v7N(8nuX}s`x_2YfJXI3eJk%ypOlfPTF^qOC zK`n~~g*SCpX@xM&RnOdIqb57lW>7oTMwsWVM$#Idbc0gin0nMke@dua(+q~8%q&|o znuyJsu3XY?q~rwVN@-(J=8V)Z99+_GrV~xgArkJOHAD?OQvvy==89`mC4qcYW$CQd z6Fnm~$FtOI^upB}H7l96JiS8AO=hMuRJLT;GgM5}F{qf zqH)KW3BdfToyJcz@m8kMr8?86vpYEAhf!V7TtjvK$4C6Cjn0@6fFYri(T6q8##%Pr z8?ql&IIWu>5?DuL#i04Pf4Q5Yv_~7kgTx|h55(fQd`G$85fp{ z%v9%eErwsiwOq5j(X|%x#II#3=pU4;Gx%=$HS_G{4m$aeFl_!6tZFW%lA%Vb?rmrv z5-zP+Lw-%fV}eKDTJ(J{#Tr(#7+5Y-%p?6<@P7*NM!DKNjayGyvHt)qQ{|9*#7{5;2P(xLY%eO;H!Y;)+em$o*uTo1;T`r`^)@|aQc>&cc!@# zAqm;mlTPtwsiH~b-H9DKY@JBQ(D$yg%f$Ltu7M}p6QB8KAmG+a#tV5Q@nxf`l0L$tKSz+7^LHe1&JB7e#%bqI6!u`>ZlJ?0u@_t{kOF^!uB4`Oq@6VOdBcK9w1> zxd_sEO|d@l9jjF}1iiBsg4u0CIh4Af1d=h=(wlP+{7ZuAw-T~O+vbhFLQXNqT=nZv z(=I8pEiYMpG3EaNMuv5oGaD>}fDeDIX3KDj2nqlk40NT22@G;b+r(+h5I*nLp?@e0 z9FTjCGx*f=A@W(5B8G1vCLuyMLy^s0wv_<@n#^e|9o5xjfw6)<@T+iJtm;e`P!d0a zW3`Xab5ES!oVi)i+DQv2%l-nl>>taKWP)G_?0E*bRJ7D0iDhe9RXT?Y^40q-LR3F{ zBRzUlNhuzTD#hr|=I2MU)!mxr1X&yYT1Gjnd*1*_yEap#o`Jqo`cY-9Fd%Qiu8z-I z;BElaYaNFwah9m(G%Yh!(C;NPK`dy=w3PvY6^Y_odx^CMmcb=R%Hl;h0oy%5?b^MQ zPt+~vk82UnV@Z4AjXP07LZZ3oz*H&O-JJ4Vx-;iyhUrNXqtGF8K4l<#*RAT_4%BSJ zHMD7QANwQb6^ynQ7Me`6T-YVU8A9!tV5sMzuBBO9LLDUai#-1T?B^X1<4Pbk)JdaF za14>aS3N?V!G9Ws^edFWA~JmoVAi*b++NL?D-I7isRUyk>s;u!fMmo-J#cC1qF*^7 zZ6t0u3(XEm2Pu!`k3&w_Kf7BaLA_jgk%&kAG&rY?F4iN9jY1V2#7(&VRJoH27@*^z zr3}Y`#cVt}EMM~8dA_mj&+AdO<&;DBHVVhqN8S2jjd~Q4GQ@H@zz$EfOtv$>8weib zHE`;(fs~FI{{Y-TwXLV#32lo`(zVI|0970@F4w z$DIEF-H25KCL1xtsz2K1x_wtgg3B!0g|R=~LzDSdBgF!@?v>d3y5hQ^`PPMdr2`wv z*(9G~z^98?*XCyD(R$UB8=ky!5@5Lqr>L!)Z4*wi zDEB`y_dSI}dEx|8QrU9ewTE+maWa!}CkOa*PnLClCu2y)OzdtvLk*F5^z~OqO`Ccu z`_~((>mM%0lg^YAfIjM0WzM9|e7K2K990ZtoP$c98Wi6#(+W?hlT9tehDSL)DXSAX z+6OgeD|V3JEoemqcFEIoR>?b>>_ul_k$&<|)~24|xWaIwfr?2aXE-O1N@f`FcBuEQ zqO2z%hB8{ZTPyhBfmY&0+C3`)9w_+3ImbcLt4jOa?MSG!#YR=QDsV?}S5Mi|0EtYFI)nT*42(}j zr=#CvRTftT)DhJ6E!@+v>C_SRs}}o~WNnB*JeTMwlzEb~yv%-Yz1Rpxe24$g1Uu<0ll--js5KV(&cU zqJjxbH9D&)Inj=zpRlxVI80GT4+>DkUSG#1luLPIVHL8__vDgUh zP!WoqV0ZhWzJOJ@G>fT^5$ztPy#%8L%5|oavNJ_UvAd8JNjdFZ1R4Zj10gPdy1Wn0 zwIH(cSyl*=Hva%+es9jRl~}}yR9#FPUk_f)1d|lEkJdx}6}1MHq}l+^WxSIg!Zx5k zp{|lg<}|NsWXq^917s8iH~y`)=PGC(B91>USv0N{B^llhvp z4v(r|w&z#Fe-pU_@~)Oiv^^C!%PbIMxg#}!fAKR+fRwXYE*N`!znydYNzdSknyokd zIQ{08r-b5<4RbD7c2UJ>c^Y1rnUKVwcVk{>fAIrWy(B@l*?kxMYX*Dko5$UCJdS_c zHKcIWXYn%QLahEN9;tofZ7$R>w_`I7fd>`MUU=J4x+Fpt1wNdbr~(pH?*iU^;%aQLcRrM$_bVt=B5q9Ttk&7aCk1qN8d3>XDZ{ z>7}Xj2)5Jv)J-mX)pEv@aif?--eIPf%|pAIUDU3`Jh7(dG}6@F*rlk4nrWt&rk0=u zCltbHEt+9GPy&-cH5qCHnh9)_8eG&{={M0~Tyq*s(>LaqG+0+8T98yqY5S^@3ywW1 zCN!n^s2%vE$!tezg_?(cDi(T%VL0Jqc2d|lVjPcGSt?3VQQAhY;iRiG=-`% z)eVSNG{R_RsLNC~Lll}qX{D*9sv7~s(q@`klXWJ%mj{Ar6)0aSyi;JwHnfi<_GTwg(O4YkaOx#$?<@~n01it2nfY349EsjEIn z+~}cxW2H-myN_-Cl1zkP$q>Vj2sE0$&Wa3`2PTfcD6cjD5QE-sWd1h zJDCcw95Vgk^gXMJ8HX&jJDW$FCbg*9C)saMT|zpf>y{tzF14Wqx>fqG*zQfF{{ZvG zb1-r97WrSpp_lIikg=kM$m{p7( z#?`&$&WU&$K`cy01EJietzP(yzM8PkkzO*%pD~6G4RO}Go%ND=D&|PL#V5?C@HJs4 zvATShHg0&@et7fxVzy9>o!;j^vE|s>hewZHg|2L^J7zNd>x1>77m z62I+p`5KTmJyxbXI)SyO-@o{A0Ka;(JIL zM$fQ@&wQ70{VUDvMX0%0rH)zUlz|#Xc+PwN8LvY4%(?M(rhn(bKlEgk=E&wLM2OoN ze(4nyYb6AXqlUZ`i}L7~Eq5$w1gf5KiDhAh1QW)3RiXB~CRim%jCsL`_CMYRp|{tr zB86>al$%i?m4M{#=lo4WtLhCn+^!i|e2U+B2Vuo+8i^9$hT=kxh^R^KW#{TxwlESdJzRL={Q!yNIv zVM|Fy<@Wl{$lv9Gu(`&{20YBTU6aFUf#sg%Cv=0Ble+m|s52Ipmjuobxm1k9Y@g zhIN>aUFloB^)RW~^>||I+vZe^NC}GfRzm zT2sQzU8m`ctv_#h^W(j27HP&ICgr)pSwr%_tA8ei!#=rTky|KPjz3d`9hrNJ@|~Az zq*BSpS68SR;LlF3@B#fU=o`~LEc`6X@1hg8Cm8RNlUmAi&twV$>SPZ;ViHZ9B?bAX zyJO^tI*!*)u)#2$q*wSN3F6m+&qS3jV2DH1k2?W@)uB92Tjl~ST*;ngldDOd7bIG9 zrTTe6TxbHSfpd<&{}@>jR`#zl{E$WMNi>QNvNgX0LMr4PerqPlpT(VmxUwX+~y|xDG*H&wRj#$P=D~mq!hqkPU`_G3swZn%9 zwPLRp9%+;h&iozQwhinw#k`fyI2)f??PSC}V*Xdfl|I;Xame`2iO+ls9;)EwMnHZi zdbsWi*>YvY(7J9SWS>8&?8?@^YIfz6fc?2#t}x`wma00$hv2TQ-QD8`91j1yx(Rtu zv=*Nk^l+rweqo>ywW4~Sq~q`MygKd5Yk`*%b@||U(M!(80R@SbIeYOghjV-PfaLUG zEo=MoP2xl6vzoLC%kr4Px;K!942BSkLZL4Jz@zO!$O)NUf#1ILZN8=M&odKfYDTjksi zv)tWnv`K5}y`**zlt-0}oCQ`MQ$y|FWZ6SmWRh(?g1Ms4w^W}#j*Y%ujK=POwLGa! z88@>R*PUzR_nyM7c)j*iK4-|!6@Z2LH}i|D~sgtmgc4&V8g~;p{_y zL$JTI)O5f??qtrh?GEis?vduUl2u6DoWGSNeOjl#r7$_xYI(MWOqk@xI@fk-(eP9^ zpniYMh}=Z}{JQ+wBn;CieSgc}IF~Q+Wj_5(KSQnlpbizJ?(y()x+Fu)?ND|LysoRR zYnkc*-*my7_8*ef14U6BW$rN+_ik^1|5$h`^sNgYSIgXT%^2Qc=yA_@OA^4?FW(pA zYbaYfVA3Ote}l-wAAR#H;bHyfZdBW^AW%TJ5wK$AQXM?((fDcn>*GUqr!Lm~ zpLLMTCeA=83aeQX&606(7N>p1C)nS-NM+|| z46tT$6fAOXNY%oF{9fHohBsF2Uj9{k%S+&l;=B3};%CfBnu=|XN9K+!f*%)wFhBCZ zXL$yIiu)w5AzJ4Q{%hL*&|7ytTi-q0kDyD~p}stKUfxiimsRq8pZ?>5~FKa{eYn{`n$Bcn+Vsr!158E^z2; zij38SLj+?ZOK8eNYPcg^CR1mi2n#NSgKpr>xiVJbQL}D)z*Sw|gM_glOUf-}E3)=m zl?%juSyNl_ZHu@5*-eiL%3Ydfu-^pljuc3XD-|!vxNB$vTEd-e{?iU=&&`Xn|#9t`h{kzR$PvM4Z4-&zT=ItW#3;0EM6SjhLpcB zcIt2&9SGGJuI(wf2CPE{&J>SQk#3+hEMn{`_EIKk7po}|#cd17C)5R3CM-oNnYBZU zL+jU|YfxW*cB*_NjN)n*nmD9K95r#sR6@6&Q{sDpWZpCosPg(Z_~pG@UO2R=jZ4$g-Qn1{uVanVv>*NZ6as%1;41bvkHq=~)3Wn;`^T-oA)Ya-db zY{_X;2M>#vm13pO0gjTR)n6W%ad1~eLF5D^mU1Qas!blo!5O2`Pb_lT#1}Ze|2E?P zXsZ6@GS_#G&F~v$^QCom&fUFb$%#iyQa(3>YiSuqAdNb!^`v?_?)vK;6biAYjb2FZ zDlr6K_l4J?#HFcl$uVTsWBOidi0ivvqQ~FSFBLfuiJ3U{a^TjiH}+dV0)P6mzlgSB z1ByQ;`4N+`Pm)rWKgW*spoNN$_WZW>`jy8o(O~E71sxGKpr;x@rQYRZt)nY`(!O@3 zH4|I;bMPJDv5sK0H!A$@RCB7Z;ky~L^TwPt?&{VBQT9mA1~-nC7s|^XDSb}IgxJ{T zc^j&5O&%XhLN?JQop<<>Wbd5-{*$AJ@qSx#{JiW(jnd|1TSo&RXswf8gh zLX^+N)Yi}cvHo3&h~_A!R521$5-S9xTBt;WE7aJJbgL|*6?O;0LH>pQ$fG|$&HnL- zhS*ivyrKERCmTdIo9W9idYfZOVOx-R5;r)A;*J)WS0gQaIB9F_6=Tw$5#ftMhV%Yh zD<`1kHI)3-F8>;8-W_w@c06Hr*wWN9vX{ECIm7$K>FSUJF+!Q|mH)2kU%P}?e6Q$L zRI2oT3uKaxWj%X5JncP(Yww}*`2oJEzhP&~9}ICiMfW#Q)$Ya$ zZ@vk?lwW9l7Qv!b^+`P}w}(7*L;d-Hz$?O-{9k2uJh^$-aj9P_-*;VqzRe#QSBd_k z#XR8GYnd`6RuURNB*@G(0~ZWpXel z^17YMTR?aI)pR5s)5g0vZDM}Zz0)p7%jG?`k#^o9l2;y~i7x~#$Pp2q7teIDX?#)e zA)Xz7wP#gYqbIkwdTZ4atuj#;iWC`c(SdOWr|Hgb_jqiKxF)o0v|183hWHLYj;js8 zD98&wRPOaU%{~1+UP|h(R&4p4kuPhP`@zuG%?=Ba#9FWJvEghrJd>*4-G6>Lz!Voh2FQn;z*VSvMY6 zm}2h*B6uko?~+wesf34_h7#gQqkiEn=p`w4jV_OBy;&;X`je_4a?_DCt!kW>y=#`< zXLy>)IE`5TJZp`UxjA6BVI}lnvb^)&pZ6D3{Bfw)XGhT`CGG(K4_Z@X5y`aZl_fm* zB?ntymkr;P zW6pMH+Mc~Nt>&B?yF+esFRBPhxj*!Rpg8W*N}ZcyQez{n3^+@L$WsQb7XKYs$0mHB zqF?JwwJrf`)tML}gVmUNLk;eYL~eV>oeJ|D1HZWP@ly)^@(h>SOTS|A(L_5ZiwG{8 zoTb2xgvCioqi1zs7wCsgXx_C=Qjw6>tDF%UzYyFr8KWIswA?2%bjv_3&!yzYlD8Z~ zNw@Tls$D+pNhC=ozs%S*%RA!L({=61{N7lyZTV?LeBSN_Dv9rA+y0)`(>=!LJ7sMd zY)319U{~|{5%T+%n5Db_(NI1ybWvFS6@S_a^x}Z|bJ(vp+XlX?;JKGf&v25Xej(T! z4I9i}jvX!1eQ5Mnc3#$uDGPM*XyTHMaG9>>sm)=w+zT!&yucK@AWZ`t;uY1_?%Ri5^XmQ zi;tCde=|--^;*Fmw_E&~j{ctVOx+~9o$cnVHm2j-e>8EuS&dC1CvGbL=q9$zS|)Q(5-(LiE;?kGBg_nx!wdWh)hrcYGVE9y$FU;0ye| zkHJOPYA5RqQERHA)BXhkvPI;xM!Actv7q5CK~C=U&KGe5t~n_O^(w!Jx8#unbdcxB zB$z#9%&dR*q!Jv~#1;{Fr%mup4G$Kxu@-`H+S)POrV^bjCo8UcCP2aN=9&=gpC9l? zSh0K;huJ$nl4Q}El>9AOF41|z&a&@1q1nSGIRUqlbUQOOA;TTA%SOCq^B8j!vw|HZ zKBUruGI$pR&eJm~E7yTVLKG1JuS`fY`(%(|ka5Rqg313WdXKBCeI(pB?fFFzmvZ z9lu0MB#IM@cP}p~l)yEZic=BXMD+|AI$5+>I%dMWiR9Pt8mn0xpV$p-Q4DZhah_RP z!;C>>Ok~Wz3db$w{y#OatnnqvCkiJ%!?1xKh5ZZxVq!imQXe0n6yd&17}Woy1-Q3d zKz2inF6>f8o>gtxa6HrNklX~CqY7?1d+Xmg`APr!uTng)f!5M7)R^F~hmzhb zWzXLt`t=wOb^_>{_p$>=6&a)K=*I2>XQ6r@e1JP-Iu&?(jRvm|@S_0+Y7L+P99Y_I#{X~3Yw$Cu#t@z374f$uiLhOD@ZK!yR%WS+<2KPU5DEQb`LObxi zQP=zUPYvyT9%GNV_yX#A*V3bB+DPFWUvhsFeyw%=-ApjD z&dc5=E85Dpi}+(A;*R+>pXUvBMd62H_O$TLufsXNeDi0ouowKhQmlh`dQ|eVuq6_B zhdX{jQgNDBkq4h#e{nd4y02l9TairleiNCUe8bwlR^sWwX)O7$?kTY|_seq~|4Q3_n5tbPK_hY>DOE8`JmgrP z!bJI2=$y{pO!1uRd@*xT4LHkJWfgK5j*x_U$X{GOv$FWJm&5&mx*t8PeE&nbDe;GN z+kZ44^b90*c1q{gwTP+=DRg1!B8Pk%q{gKnsWMAk43 zbEk>8(KXGimFE+^&>dcqUmG^cF*U_EY89?vMBKj6L?od?{jA=s@>(3F31F>kC>1@y z59uG6e%8}@T5g@0hgi+}X7$%9mtyYj`gf{ty;atOYmbLn@E1zLB5-G4o^ zO{o~U07;&k|NQ>%k}SjLd%q`3%ZYIG9DbDdb4;J^iH(4LzQ%y+`8Mx_K^vQNBSgb5 zDisktv1)1-FTlr;B(})fdya-4z^ zk2>eS8~D0u_;_x~`FuyN8W6p%^D&D(8j9#I(vo(c*KTw^&pAgs30V1Oe>Oc)Bgm9? zKl*M!ldmn2=%kIDY9d*>XL&oe&C<(BMLmlDWAN{?v*3O0{u*&&tyIN~H^b@$a8*Js zNBu$|_EA8I?OD{2rOs9o7=u{3taSRTzUh|~CSj>AIYfy}IM?p49XK%1^BoNM6Q9t7 zr6zA^nifsARDNR{x{|Pi6h;Q~f zw<(#AsCpkA_a|l3wBI|)dN0W}`wmodn5B3)4a9UU#BhmKd9yEH7B6%e;6e-@m?aZA%urTkn}SG$b|@z#reveR2INq^y1 zoA$5CXmrD~sM(JNzOUrZ0rPLv?w(h_)#9SQGPN1|>22{DdaALXq_ZpJr*io%R=TJd z7x4aPJk|H9iOgMKk8Q4nIJqaxK3vtBWzQg1UnAQawm&*(_$4&t7YTL@g?Gu1lRcynj!l3-tKRUUJfIELs%S7|>;8Oj5iFFNc&u_R^?4w=Msd_L9FKd%Y<=(i ztzaCm$c5$Oc#5)xKwgE{lh2ObGS6;_{zsF24JMo7`u-PDTQVnapiF93b|8lut}()P z*Op%M#uq}8+E~PDVnk|12&eqi2c1Miytg|7au+mh5{DyTY=^#3mdJTHG6_PVzh$d&rt%>q`(!`t6o<-I-~b^*&8g z&o@HJclKC3(q=W~kIFA~3P+X;Em+ zp6O>gH3FfngRlD~q9q;CO6acU>xs1w%`FnA*t?B0w%o#(I8G~bow~0dvxB9S2K_kR z;5qR;l1iat4R#@D6guGrk67tj5o)cJKXzwK*M)T+3{eD@;oj;-!H%&^rcCj&FaZ1` zo+a}@2W!Y#V=!A?Yn^zYGkFD>Xz?w$(g~gV9za$`Svwsdu;^@qr7G|_ykkVkK-d@I z3=(eRupcv4T^Y2<&XgC+iqM5eP1)5s4u---Ex#M~ zBtKr0<{Fc1g;;)v>Ca8!EgWzZkXUfl#+29(({(%h)O8Jnl{eZTT(Y`_ITw)~Wm6@F zHb5t~xUTparijO*fS)L&DVB$G|NDBbTMy9QclyYN*s(jny4ye!$`Fsq{-6s1vO}n9 zOgXU(U`#&Lhu_dp&-k$nJ#~HHrc3eaE;foF8r5xa%Z%J9agq;&XPxNlt^4qdPu;hJ3i&r10 zfvA49>R#40w=1YD(s;Tj+ps98449QV6>gwCBL?wa|A+{w<5Y~{Afvc$1txycy5k^r zWMFA;av+L`^lNMCe9LVgu_q)~9+ z#cKn9aNztU-^6@d%Va!CxxRVfuA5AqQ2L%k|M}7!Y((Rg*05>2v;}kvlgfXK+Nz|N zG?WHZFd-RB7=8Jy7$IAJoi)n=FQ|vW=z4XE@UJ~KwjZ zo0#v2O?jdRS%o^Z1BMtaUv5bK0 zrs4Q^#X<#F@o%3+eZ7Co`Md6Y@00&%QY`KrAe~Nu$A5CDrW7koVi-Bv{q3_bW?19~ z>w&SkcHlJVPiq9uG9nH$BAE6Bdhwznvqrn68szDu=M& zCy-aLww6m+R*SyeeN(p2Uc9*D9RAIP0CP4{NY9aVh1G+(41`q%CJraw{gCR2E)!dG?dDQu8TiW4S&9G^4^i~hJwWZ)b1{v^!9CdBo zM!8EUn13ZeJ1VU*;I}{4-mGyPVVr{}SllNmk~>;VuhIabT z$xLBnCb(gZi!l8VRab8jRyhwHg$%5{qKr{o6J9Q&{7uF*%;ow~jASBxVXUJVGLf=k z!JeD)i6_iflcq+4j{F+r?vXu74w zZAQazk6qzVnYIQP8c7%;u<8U97+ z(QybgqZVao9XHqH_#e$C({a7u&nQ-^Im{{wsU*u)biBL3lB4uSamLcR8!D@e4-&=! zrdCn&)ps>nRrlcUKSuUsts)l%y`JnlloAZzn33Mq+dTiSP0jQfDSk(^1@m<20ldq) ztNHjwM}PnQk4AVtY9v3mK zJl*vI(Zv2jDmM4xpr>V0XCq|hpM8In>6kvVI;hJHqXCbW>jQT*K4&F>x%Y0Ca6dLl!l4esp*0O4m@+u)auGiPHw)nbg)$Ieg{tVH0 zW}0PmOS(>u0aVNXh2JByt(beQ9yiT^T*>sFfE7ZRzjS^&uIZ{SSjW_SF?fn3NPpGz zH4g-sHxgol|A(f@>P2{eWdS>_m$8INhX zLwWNnMYB4pOZz=~C*WG7?oplWZj-uIX_T9iI5?G^+GBZ54zj}eN2Z%oyH?&+Fp(Dk zs;?XMCddf+7`QWqzZ|WWOZ|e?amhHqt0D#t@O|3jipc#3wP1ope!G)VwuCR ztmZa#J(AGVMC6yP)heX(au(2mvq^1wMAK}qa|1wsnl)>v4@ZF4ji5232L0};SN*{c z!g$l3_@9HqR{3dg3(dz);7XaI0-5}7>qDA-h+^dy>T6ah?`#;ESoYc5JP507Y?hUR z5xy8Ni)5t81WGD~#j9F(q&*$h*)&|OXn`qb_J-{nNU=D(dzu+J3H?P}ey)ka|NirN ziEpqIidIT8z*+uv$LD`QS=$X3mcyaV#+%T8mGE~*-bdE?rAdqepGw`?NfpwurFf2s z+oZSdRvs=#l}Jh`qIFJEVW8U>Y$}lzKFWZ8(pp{gc;C1Kq6;4D?+?50ODb`{MQSl3 z>>I8|k;=bPp~=F0{k<)BuUMAfXGO==R`K5nGxmnjkWzi&`F<_8?r^Yh${rwuaKp8XAKfP`7KQfh5=+)X zN8Gzq$o7nodmS-%^u*8_zszyD#U`h;id{)U$$$D(AAEdS3B`X1uedR)-VpN`%&cjOzJb=$m=a2pDH(D=!+0-nn<4?fkjS8`jz6fg~(@}_*e zSF!u%t|k?j4$tT$c%t)kcV>uCSzv>YmCLta1!{v*Wdm*O?nmb?!uPC>XLLVW^E&GB zhfSM|(kovez6`^`HO7@cAvba>z~!{xB#T2&Sm`JJzUCP)I#^kHthr6r5Z9Xd?Ak~) z))f^D9H@D)Sh9w7rS%Xat;;7;NOlKylZ4Xs;hcBysXBR{^_CFVdab8|MaT0V%8$A< zzT|A>*mQ-5eW}mZ^wyW;j8x9obNPo2ope~8xR=>-a86Aa408I9=JDWvG)JB-t2Xrr zstS|3G_L;N(wMo_mAz>?IFAHJl__Y zG0%k3wpu@nQSMYp>(e%s+~EX4v8gnpn7o0EbWLX+x)mx_g0 zo}u&3@cEH8^gBgs%WIP%9Ds4GEcjw@$R7aE`U zZiDUqkWQ*g?3>IN{!(0bET8zPlTCV#y5H8J@*51k(Yc(JPh(t8al5hpw!2a7xW7n& zsg$+9Bq|C}ttLnbGR&KVd!SNNtn!BxvOZlg`?jNGVgl*9j45*ynOs+{6^BNT3~_itT2*hI<$ktw3aie9amf%toYfBK_&89x>t+E z)(oS5L{@q6@JnEH?&+O@| z5N>uR5|yvR5=+iBc;G)Wu* zd=xE76193WMa_m$vb@Z%C-MS4fsWhEa(o>(O&HF?3B?nH7_;5(=J|;uYkD!Cvs6Jw zI#K51A^m+S)Kk1VrQr^mO-#O>z@t)~rkg=3yskRHQ4Zi&sdY@=vhdd7qpzu<@>3oY zNW4?P0BKcoeP8gTWBr?UyoSYD8mzQ5cLUv3T5Av+Un?S(RTpx+X$x1W050My+|O{= zEC_RlI4SjOjQmq=j)hd9%UQMSM4~0A)Y(gC%WD=-Q!C;ir7+)5_sGUaY6wF)E;-`y zkz@8h#X3()Tj0`ZcD=xL*Y@4<=Yf@ULAXphEC9*@(7GLx&ZfyAq0;aZM00f)vlVsTtJ$e+FEW@nr(FyglC50S5zi zV2r+aL=4IG6-x0w3v~$|B}LtzybCNjghm+`z$7#wS8XFLb$x9PwCJqad<_eILfO?V zL_120K>~ZIG8Y*`U3f1h3b;u52rkf+-HHv6-C}}?&cna;T&>El0v9tI?x45`pLOIJ zRjVCB29W*9p0x0@dDWT_)1DVh>|?1OVy+{+4I+6HAbtI_q-T?ssvg5j?1r5fBnH{} zw6x#2Gii!9-4wuPRyp>ljYL(~johq1P4c*EWJ_yf3IV?v|7Omq)eCB&Px0Wq{)@+! z+_r9&+Dp?16{7NhraL!(*iJtP<^G8Z@>^c{HE^1bbJr^)3S1|1YQ>R?Wy^WfZGNE> zF?D7`_KeXgTe@kre1WaKr@nW!Dli7!hzsc#Qgr*c@Z)Q_MKvCJDz#}E{p2miu=QdaaD!ITA(=&P z-B#ZX*I;X0@*bn7<>6sdj}$SF6pB^O64VLb2KNyUn}5jR#z8^@xs)DMrN*%Glmu(; zTR~t~v?cStaXWd932(y+8<1keQ^iqAfeTN?>X=*k+Gf&ZHPODZdM3aRRzg(}FA?$C z)*##DF}57QC!JjcuOq$0d-vOiHmX8o8HK3XGWWq1Ui9d-ZCK10R|ir7RJgot$+r3E zl|1wYD>CWF{BQwmipMlQjLY;E?!iGanhHrY#L zv@lg>>y%_S5yQKzm7>?u8&FBWyb_G}R^(iGFjhRC3XI(OISI##=o{1ck<_vrBYcRtlv&{A`NCPRq;@2LnWTuw%N;33_7GYtZ`N zUm?Mwq73D0N7C*Rkva$=Z%GmeslK^m*d|nxhW44WFx9EnGo6+I%TdieEwYkLunZM^ zO}`sP7inM6ws@r7og=Zx)5bp=V*E7cOBt_HG_ z(0kqJysC8I>jIdfuf??IDIw@UHw~>@uvSf7FssB=Iru&dH`XJ|6&9yo7YA0_!3{L+ z<3G4m{cE=QSOrb&_CQu#s^xw1BhdC03!OWW!6<%KE35kY?WJae22)G}<44#VG^tNw z!JbPE0EG{KzylrY8ZAs-n&;xP;_vYzvNbFdUm{d}G1JnAzJ`8zd+(0$jujbHMjcOG zRYIJ5LS#pQhYiy7~z%INdq?Y^H zwzV8sLGI$zXs%#|Hi*ODHKY7@?n6BgA;Qnut_pb5k zay9JrBV+$_at4tZ2$E zG1f*{9pm5uwz)i8ctQu9QZkzsayitaQdHf{pY5Z=hhnE4T^|4W%pFBU@dHmPlM9w{ ze2E@6FI`DJ#W|{xXGway&}Pqfyr$Ca({ViBs6=~=E zGMqQBz|xal`YbcXWXg}y)sFL}4qXxp%uUfsAZbXlw1l$7L#iO9!Hz1xsgfpnSGL>b zzmVOk@5~9{;vW%k>%NzVXoKxljB?3bLv4$~Sxs3jurQa;=kc@Ty&HEq#-(4?=&72< zv{b;eF#4A2;61CS$chl3M64y-_jc|v`jz*g8vz-XK)?1zEo5d>pLshls$` zKOd5E_c8AlZw^+G9;(BfhvOw%uU6XYSxbyP9EtA-6UpvXQvxA`6$ietL>; z&U#K9E;$n@b<$NrE1>P_{D-X=B@{7IO0F`1ymH`6Bou-xxmNDU+s2n$bF2bV=m3yG zi8UzccksTR<028@fhc0EV+@dwY0mJq-o7(tll(L^NKz4KQUEtTK)mkYcUMS+DXm%Z z?SHJ|Vul-va~N5T0sW`aol{?I&9v-Vrz&w<603hnlGWsD=397(9-rPZ_P2y*)iWha zJ)CeYs%)>)yepJjuM_H(Us^j|8h*4E9y)~**&c(8R!H4R7iiJ9T%n|oBpzCb<jyNvhkq3I2a2&=SE@v{LC0ZVVMLlQfmVp_yKoG%ya%c*iar5<1>+66S3^*kth$7c{l-HKmjrzrz4P)G?(LQI1c@1mBFJn94Nz%vHW zl`fJa@vtZpESUK&@T0Nb@WT?S2{qoQZ1v6=jz0f;J(5N`uxx_LsbEBU<-s>?zL6h1<4Azdsh1)eR90`w@PMts~|0M_gC7mQDWRI>- z=j>hys>8PG(^lBrp z8tB$u?pJKI0*9sf9`ZH`R4sQStLdLeYjpCn(-`O=@Ns77aeG@0c6h?7?@%qETu4=% z6VV=v607#N!Ud#-PAO=iKLu_C!PY#WO5XT)d;*el9%=9p8;)ys;^OT5KR}m-FO%9J zJ;6j>(INvQeno9>N515;3{a^|5vx(vPYz8fWqJ?Zp)?Cm)GBxzY`~FWGHIy5hU%zz zTlt3Q`guDa9L*#oXY6~^5HTasI=xY3d-hN{8zBSC}79S$`lw)wy z^wf$sGowl72J?dh*XYzEu?nU%Qf6B7ixsH*Az~$#)NI3m1NPmjb)J+>cQt_L(8 z>O_(VLjtTuL*+d122N^={S8sR(?UV!$!m%y1z+xD(<=6X%EJ90UiFru=QV5z$+fge z4rmS$g?vDoOzMfrmBpKoXik_EDC3HAuLyY2myHsUBX_W>k1`l&CO)_e{xlrw0;f)7Lab1~%oy{R+){>4mc6rnX z%|c%nkb|I@pgUdZ5zZQ=x{L8Y874fb4+_l!?d{&B2wHT0-G3%FW zwVb||!*yS{0f0-5mg@^?buTs#`2{Aedk&M)*o@<~D$NN8h-LbEJ+LD=G|Srz^r7ag zPIF7uo0b8C|20Jr@D{=yiH0Y@gI9)tV3C#ShUCwN(I7 z7;Xg6gU1@S%yCO04LmQ|)xeQim~ir{vl`*ndgTbGAP1~gnGk#H#kR`%VA>=E8Ckmm zKQ*icuj*@tRL8Mtv2tj{y~17d^v?%tyt>-l8iGLryc zY~hclyc2yFdvVFb==-P6v1MA0jAhCqbV!ycY1WXavTeGXkuu&D#u`1_Ju#-yV4E`f z1@h!)zL`EeEJ&oM)Q90$wMEhLO2di`O+a;E=ZoDR#FiSeTj!nIFCbx>#Msvw@~oWl z#wr^f(uQ@9;0Mg(*l)L!)xiu8;~o_EO{bOz5pZgO-iF~cC z`<*sM8L$6kPRQKKFiVzhyULS7nVM+Iv>Usf^nyCIM^ZhTH`F(Okc337@(P*)IrxhF z%jNoR2BB1dh9P{n6N39A-isgAY~FK}>2V|vs!6Cot)xw0jS2*H_kQT3b`WmB#PB;>LUudlJ+%3$~r$aFApIT}4V%AUQz7-_j zYtYD{#?@M9q?`GhgFdQ$xP;?WzOEpAJECRauoC;aUn|f&yaQjY=P7P#h}L($!uHiO zkRXeV#(wh@xw$L2&0-ohN*Nk((8}^KeVYM$XKejuVWEN# zAvCf<5h+u;p01`8O88hY&Zx$#7F~!daBQcryXdH z@-MR!^pht0!?~@GZeIY&Ow?eh>VjBE+T?6zBHs;z=~OaIIIC08*z9a{rnT9wb7~Dq$zEolD;BliC|b_ z8HQ=FyCes(w!ALI;X<^FSzaKREadAqd(xSTa1>-*4aQ-eNVDV?xc~q2%0zP`NM%ip*gj?LVxf1mn>YhPl3#k^fo_ zq)sRqe9y}A_0a~tM?d~|isNERCsg|;OQ!R7+P4TfP+u z;7G#HUVE%n*7g){+5Duata9VFOXeSPVoj2$w-e{y_rc;V56Aw#5w=z}irUc=u`4Xc zq~C$_-qIYMOI0{i>qC~D{>hQ8YzSIsfI|m7Shq~HWCXqkvKSx_k0LN->?iTd-Xxj9 z((I=7lnkpNXA3rowHOOhL%}eb6WRW$^fq7oP~U1MqY0KVvJ-J)D%nLOx-K~`4aele zfx(sFk2M&D^c!RpRbQri`))bV6?-X>+Eu2ZZXK?vbO0EPyBP*Ezcl&3-9q1e>=Gl= z3qZ4FWQ3%VKuNgMlqF8yf0dn{p1Y?_9q#DS{_n0Dn>aO5Ohw!F{qU z(Kq=xR6UWUC7wmnA~X4@E`Z#plAqweZ{!`kZNuc7e1u=t+)xW|tZ}Dq1nB0Bzf<~| zm?zK4s_Sodj7aPY<#Q+Ui?DXAjLoe$Vb{hzvoWg`u_kZUG0ViNoAEv%0n`VwD>#FypBuYE$@eSGnUHXWn?gnZWqg})wCFHN?|6f<%9?$guzCVgeB-)(M$88R8hs_~} zoaQiwnd3->*hUIT^sc0k&U-#46-$U0v?b}4w`hSr%I{b!$i9P6OBD}*!9oa#7e&`UC8LO|2 zEKQv>+{+$qZ%Pp(I-(^T#ISAj0>_twjm-xM-%^keW9&)T8Gg|D60Y4$0?YD3LQU|| zq;${p3KyS1+3E6;h%QI>$Q>3fQfpA-UAFA$u6#=$3No=SD17>H$`j&to1|B5tNJaun?wQ1l{(o*I$u582j9yWho^2uRj) zT0;kVjw|^h#0T!so8{~VUFFhE9_#vQ8GJh!l*lM|lk_i71{MNbZ^;8e{ZZ`?o@nGS zLw4T0^b7xd8EKO5bc9z{9!rsnwfypS)q_Bfte4`4o6U5@Y`7m2f8%p)e{RI$itTP- zl_Y%ja)?yFo|bjteN+Tp`kNM;u|f&_zXKZb^*&f|0uj5_Q%-2=bQs48{W^(h5a1QWP|2$ZrSLN`9!Gtc!6*QPvF_{ zQ1|!2L#25OVsM(z`?>c>ar8L=+Hc1KbZB`=?UUP~G3Rg4@+iG;1=}WTKPnZ55#B9o z}R(}ER~hCQ_R&U;|X`CehDI6p@eVnV)?Nn8wb} zGOoas7UjyZbsKWq%P-9TV-<<*`|!vrQq)?^1NuxII%=HdYFwTF?O%;OuFU<3aY{+F z9_g(O#$Mgl{i4X+7pGc|1{PyGTlPDpX!*5Hj&*y0HO+hWo2E2Sjr%-CST`D*y5uqj z7dEh4=8dkmB_eNrEKZX9@*;NbblIh);Gq5^qo*S+(fp`LqHkVH-BB=w_nd;y8jrtQvjOG zr`n8!#pH4o~^M?RM$3v(rCk3ART+zT*7 z1a~`XG|#`iWtZqHc5W9CTy!u5Fq)f`4!lB;x_QIG1xO5fvee%@JWc5KctDjy zAs_%35(IiS_chq;1e>8q)iK#XF8~EvWSRhj2N^mr+_?Ewqts1#Y z0z(O~wVYomZ~scZ3c|Z0z5+MlydMC)gVsFId_Ana8s&Bds!wt5!#N-a?emBjMd;WB zMRU5oV!9$ThLA0EygqwfNA(LQ7MKzc9NN*6hdcc5R8w8VVf!PdOEo57=R;zB)l1UR zI2toSs}gNFI9r^`Hm2IU&`9^$DyZ;cF+HQ}NBmdFZ|CtU`p+@$At#&KCDxct40VL# zR|t6laa$xGMOO4lDxi{s+PRNH9Aqs;e+clOx-d^tPE`~HNr+k{{^YIueoMdok5HN6 z)RYkIgsftZZyJ*`IgwHo)shVfPPuvENXG*HKKz4P#!u@j5`oSsqOpVKbz~N~EdO20 z)Ea9F>|#MAx>(<@VCx;3Oz@<$A1dO%U8P%3?orRGrM*8AB_%YqK?`*CUs8J{KzqV( zJCe-6@xo@l?u}yCJ}4!ORo%>_Un`w#fByWteo#?J+>nUtqsBqkJGk(!2}`xtNh3}r zm-@7q>5fy!T%P0i$UM}zIZ{EEZP%lP$GFesFZh=|mR>%$&Of4alfZS#vqdc*oOv$2 zk7yS+UP?1|!)0$Kc7`^nE3abLbD^arvku6&bL?v_-!QoQlo~b z#pr#}_!>Bmn46KMVoBIC@{vajQ7%s!J@0*zF>kZKAUv0UDhrP4nJ#d~K}X0_Zh)fP zv?_k6Ohc3~c{=A*qP*Wva7x$`E0NC4^hN(eT)Wr&bIpk6T{?kNH`&5C{kLw>Vp?Fc z2~kUw-2H0-!hg-zQhRVV>13?M*ugQivT=@Gv+hxzY)yA{cx^FL5W|+HHOTNYv{70q zo}+fhTK_=c5}5=%=eT(-zJsSjubKpFL|ww2uoFA;oQJne7BEUg>a*i3!mJR*Do<@S zIpt%%fxDA@jqa~GHZ^2cBvz*&6Iu>=7Vkght6$)_T9=+)74MX$$TqKuAx(I@HRkXF zT#zX<-W>iSi9;WCwxY<&qe^&&E$+Po%w817nXw(`p|EwDZCi|`+tYO^0}*B@%nsPz ze(D&(`~|+6NqH}=_3}f5aX_6JGtEQa6SF5R-BGEY5l_23A9gITK4wo@(4O+O+PH)+ zadd|9zu)Vsl)U4|*e8zPa(D%Yji|rd`&6320|vj6f>rZDmlLdXEw4+8*{KZEX7sBj z)I&O5^E$+0Zw{YXnQpiQJ2~~JLcu##%%DDMxNj^&5?gu}=D`%S6LX>0hCGV%nQ@~Y zFhxc4R6Dc?`Em`bzIhP-5gD{m*5KitC7H!Ft3>hLoF!>=W<|*SWu7Kz(DWDvm_6gM zQm9xlTUGG+LPf#NQvlpl01+|-(5vbPo1LJj4^Nhc(y4I3jJVJMf^{Nd5x^P`gtBB{ zPCZN$wt#v8cMk!6D|n!9kzGNE!-@&S@Cg9_Lsl^;Ik>=EnkPbvA(%Gc>@7pNku}1( zPApdNZt;H~@WS~3Y}XHR5+>->4N6rxd2wy05+$$j#XY$b7_=k#4P<6a&HI3-0xba> zz4)a{mX(*JqI=3Kyr@uBo7yQ<{Hlj~Wq<;I7y2%u>i1CfZ=Da~KZX9I4A9~w^)CS} zyZT!C|BGe$ug4LjP_aC<&%SgSPEBGhzFfy&y`em{)eQPsOUOo^>XN6n0vPo?VUSWi-F0Xj>8i0?b3wg`?G|Uq1pa7W4QlgLV#2%Y*pRz`0GPTkh+Rwi(~0t#L2{KfE^0B12najgxoB-Pmu5 z{#luZD;>Q(Va2SwA_`fKWTnVN9eN_)#Pdo!sryDgrW1CF6G^Q|p3rKH`fwZG5KdZ1 zy61W3pirzSt#n}S(fkJ={pgnonx$y}8#tGl>6qryZ%!@wU}q@sEN9KqEaxp2_Mt{V zKNx4^-@yPQGbJqNj?MHDe)VCNZSEfY8B+4K2!UnQLZ3*^Brr>2j+O?L_mW6egcp*a5N|I$@90-)HUym(?U( z*aB6_zDkYDdR?Zh7IO5Il^_OY85d&bA3ju>1MXnk-u1ocE0ZO z*`46aJR~(X|9qh_>BJAfP-y9r6Hf6hY95n1{B-Nu(b2N0;n7^@224=+;G9SCRJM&1 zws3c(l6+=DZPOHD3vD5D)Lh$zz9N{C{I#aY^A!J)%35v5F|+=8fDd_-8pAMg(snKV#D zVp^=YM^2X|d3{$O$q7qcOFOJtp~#890Xbg#(qFL#DTLmd`NDC4t|BC41Jf}OtbM43 zuoAvH0lnZ8k-NoJw*vG~Tw;^P3Hf{HgPJt^9FeuwIL9!p%-If2^$rSUKL1eDeq8H? znufi-Z$2z^90<))$GCU<6O|%>O{u*l*L9i zp@tutJ*#_!H=3@aWh8U8vQJRoHIgJ1hNi3DQ(QVRsrgqs(LsCMw~ip3b2F(~RXj^V zlRG!4W{9t+NGks9Mgk2FZTHzjc8FBx`fV3}c#BEY9d;h@9w-~(^AsIDsod@=ZQ(S$ z?KHJivWIH#NZw8~b|%9G;v7j_P3fJNXb3nDqE8aeFG`k1dEsdgBAlKqLX0N!{}Qfn zpsj^~tIwn;SrJ@cm@U$%@O&!d7^t7I;wV8)Y@s)t;z_}USYNRu#5qwM&)-ckmQWY%3LZSYQ-N@E>%3dIR=sL~Ucc1No{+#E8{V!0; zZ*^Xd*4Y3$)$I_7ouU!#bzvnJ=zWBt&p7yLxQ}|9la3}rZNM7g3ij4P ze-L`3Ua({35iYf&p|psk)*E$W;X0~QzIiNf_+vMwg6JIcK-x3-j8FU0Wt-{p0ih!@ z{)^T2R|3RvHD=Y{+hJR7Ak|gtF>{&3>t+lM6{wiSw}xs3{t;XGOTv@~jT{0gdY8RRE;yAXILI2ST|Ws{j= zs*3ZVK!@ecQ2hcD4`sAH-80Ecgm_TpHJIiiJw{X?bSuH3({RS|mI~Bi3;F||Njb)V z?`r4p1a#ECF?l#(cd#_3M2>SZ(Aw12Rsw5Rv|58J?Ic>5Vl(EFWfYHA>wxVimWxps zLuRb~t|`c#AnKrSxy+C8yi()KP+npCwl4pGt`9=t>RoPE91`j;6KE&n5)Ui2A=}Yu zH$%s(1G}@^=dD6bd zdDomYy3;GXljzy>W6(orC<(Si^wNuKNwHW>fMNh*jaR9YwUT`m%;N z^IdUxn%#F+T1}J`BbmT8^nf8>4`oS034DSLd$*4o-^IrW@!K8aDmf$4+$4w`u~5an ze-e3SueaFI1)YQ3er)R7uhj#^)pmV4-8{dOhJi*-Pp}Ou=RN0HvF;_=L$pk=gtoTZ z=cUNp5&b!f`xd2@e`gilt}s-1T6bpXytVaz^J>XLZocVgQ&Qz4 z<1GYcAM%DdW5y>cQm$?^up)_1Kd#nvUt#oi-&*zHaj>che^q{@^wUotvztnk*jBK)XCt^dt7LOZ(E#z@N!gZy1<2oFm!VRA%P0t-5TS9G2F>j@WQ(ZO=p8@`#(opN6@ts zgwjBFlL`_C1GPnT2nUMpAnJkHsngSk2L8R?^<>`dapob5zTq6Ux+ zsE5Y1P?0T!Thvp94oM?&iO?P$lVpKu5LxhETUs{lo?C4SUKHGX^Qi}PwUW2Hjs*( za@jpk|2X-4oc0v=!3XdMOUhoS^L0^+r}$cqR`Vr{ns^cKeyL5VGVei~ggXL*BBa`f z=UrU;K+uMXtg!`Zq9&2tLP`s?og=X7z9)o9<31acR`5b|!!x?(SyA8Wcvh9s0iV4! zuV~k!-)Yra%yM++%9dU4zRdc)2BGX93+i*elF(%)cOJE7g<#7$G}aSzd=9e<$5=-t z@1c=$4RTD!`aLEYm3xjM7sEnY)T4z0vpzUHp7hd>p{CwrigTUReP~a-ZB=G#8)un$ z9h^nJ(rIteHhHJKkus$#Xx(>FY>nWt{cpe4E_vP>_evdn$vWkH+=T($&K?j+>a^xpdJE^$-Pz})iSwVr0S&;_nc zwbwa!t_DiPvvrY>%*qR@G8s()R`jWKE<96&e9!wFq8pCFt54#bGu)q3lU{q0Dt0O> zB5il~V`0gaObDTal_B8_f%&Z9Ts2BixfO{9C+Tw*tuNC0k81QH5h15_f?xGaG9I=E zUcHNbe|xX49`-hKpWOUE8;Q*M6=kt38TP9(td`xk2U!HUiU}JanAnsPuVi-ue;aNA){Q%*5G` zXqYnLv+3{QPgZ;W7nd`jGk%t2`2Y253k2MiMJ(S(r3=u!T5FJ=X0=VXh_HbXA@%@p z5}wwmd-w)Vy?TY3qE0P6EoD&!e^!tDjLS0WVaBzO^awBL9uAucZcRtmE7<&lWs+=Q z6&eDOvL;D|b#$k_saRg|>#@#~J{=RXd(wDrM;0-~m@lr~<@gBpp`zm z>EdMTmgBGq3OIQ-C<-##7l&!lWZMdT5Zr;@R3HWfabrgkK#s;P03v93{XH!W<8J+e^s{(Ion%i{PaIx7KP3_4X`1z-*;#MoU7N3S=O+~Ur208pXTYxid$1#qM+ z3!ZluE&+8-1FE0k?{(gFyns))<`;fKFuV<)Vcmu>dakf!IZ$UeY&lmIzd0}24tQkY zO&;0na8NNCh`~YS^I7PjDV~-CWZ?+{->XoApn?+k4xFw)k=_e!EvZ85g2v}ZEW)M+ z1cnMi%8-zyNCD}dTSy8ShZ<3=lV>yMcYwWA&4G=7AV#zUFRS!$T9sg7)eVXfi3hz1 z05&1jkO8m0P9ee6s0=dHxedv;@lN)=_@1gM%pl-Ja(!5ll07DdzM)AqU7X`lz8O~y z^wXw`m+_JT0cD3M*3Ht z^yLSB|4%Z^PYI(<47k3txtxebE=Wk5oSRI(EOT zR2h5v!vv3)WhbUQ=CsMV`V<07`%pxetrVv?3$G{Xb_wOz1{8Lj; z8VeGPCdjFgaV{$_o2GXVt#sVmJ9WNgl0RJ^yCvQ`_RHhVlS{3&ejS}d4ZcvX!PTW) zxT4FXyO2YDTX$=$1_8_Oc2R-2SVD9-H`YVnT0FwXiKj1L2n$P04L$nv-Nk*(2#uoY z6MpH!`YMGCi zWNA~w2qlsL!Me0`%UX4$weadc=@p;Xa7we=w+RPrvt2u^A!6>^`#vs^Y(NPqEJd6$ zv5+`8Of5aE{}>T#pvu)k9@Wp$PZ!=6IbHb&B31R~59C+x*)K^?MT8RWAeC1s{{kUM zjPfs~ww{nm2enoRA_fDm4jB?~7tAjcvye&^Wct)}vvWuIt$Av26hpwJOvG-=+k{k? z*2HxTo6UR1cc5*L)BNm?00RIvT2~HLJjOK{W3Qjj!##KRsRTRN#(VFCWm^9frzG|H zf0_A)dS8Ghs|((GgVT$0ExYEWk)Dn5P;l$t-cfs_=m6<3Y>E^te~x>%DLgM1RhLholPXCwwsNvZI z&+d;<-v>#!%~$#CzKw?4Wz@vp+Kq9K zwh3;_Z$47VGW$RXJC6K{VKEC+NeKvgD0&m0M_ z5!z&H&g}wl40?|kG)g94go87X1wi`JBalC0RY7~nz%o3y8%T-ZM$z{^gLRG~~GDo~inn+Nub9R!p z9)h_dc^IP+m=jtpn=C=`6e2^9JF0+S3aJ(#%EPJjav`-GO0GziL%JT8MdxX4_^1^$ zg`Pn)tOMIzzO|x@7IT zWW=}YXV-OQtz{_<_c(1+gFBWjYx>R_M2dX+woAeLod!2JEzJ5i%!53*&| zHEuhIF0VuxD0Z5YHB(9o>8KikVXQ-tU1;zm~1rJ|Utb-@D{&gxc);GJW?4 z$``E)i&ThL2ZuWat51V}Pi3r`MLl=Vx6z=Q4HWasIdNq<|IW6Zu_T$SJgkl!a%5{P z9$6Oo1CiYVKhrL1{N>AaF|h*<|DG^gu`c)nIl%q{p_kWd%5p1Fg_kq0Pi)N21^ ziqEdi-Qr!KRremT_qcFQbG%*y=@BxL_bc-5uWP%afx+-MmET8$LS8u?TFqK!9c;oh zG@Y;#{NAsY@%q_>h9U6IS8b3mA zS9`5@PiJ{x?u;z+sxS$%zzUywi@hMc%Uo6=du`(1Go#V|WI4Jz`qs7-$1N8^Y=eF! zZq(j?Hp?nhdo%b=BnG#?{PM~^`)}Q>g4K^K3sz9 zYW!?z6|?Td4Xkn13UfuTVg7zQC z5YHa#qu3udyPTlB4Yhs4@2!~Afw8=3>jynY8jO?@WpaBH`ec5^5Q9zDpVm&E%>Lj$ z6ncX?IpTYfhdfFSaF@wGCT-q7w?z2;#5C{7(D~kmwVZOJOT*v;TzH(NAXZEwfMWThQbZJ$mKYv~LmHTk3d?0D| zTJvcZt7{nj{1jPpw=68rpqXXY++4oz;u+Zn)3;AI zvTrDV-L>g4?X#{BvntS>G~u!_p6YJhUs}`Qr}x;tj|M%Xv|7Yw_8N*BE5rAcsY3?# zX~E95`X#QOmbKPi;1(wLNHVgk9@=hnHz-rzr#0aU%P*Y5!0{f_g&~{=`oB zmne;Xw!Z8B{Sj*>=VZ1&-PQGHuHttn@M8JW*cgzkjn_ayht6?(Vot}u7fn&8f_bFQ=< zGqU-xq1E*VV%Js5{ivTcz5QTEwsT2kv0cqdqvA!_Y{HMd_tja(t$Ej&;heR-mG*w} z0oN_c`6W*bn! z{7ceSI+Pf5RvOE8o=|CCt?NAv6kVpsMmQ;w?6*U9w$sYOKijN-P$;pv!_I)RDE-fn z;dSda{S!%0N{xPTSQN0muQRCe_Z*<~CeCI(V>PjgYNGp&=0^L4`q*siN8*$M)m39N9X+pNUv?Cymg4t~+qpxBM&XdI0zEi!GN9$Uy+wQxLo{%wyMaE|?VbFMsAvxeXpn$8Z-3M`u5GS%J6G-`1y<^2 znqc+Mn*%wmcen63*408YJxd`Hze-?DtvBL2(!Zs~OwLwGRZojyf5YOb3kv~xL5EUW z@{P0C(K({nl+f^&oF+$Z<@;Ab!X>EA!W>9|`QWoQns< z2=;fHMmDVfebla`HWbjm%g9XgR;GfIz)_MT~u9H0m;DTmlrI_P&yt|82Y~cB8V|J`QR~)D^uc z!r`88iD}D`IRw1I(k<;1-ee^`^=lRxxz|5)_SS-<$h8`+rUd*gc7T5Hxkt>L)BEH$ z6xnJnwI`z&T$2sYZ#5SBDmo}W9~{3b&n0flxE^(Q2=%lnTV~?0Uy9jHNk{Pq?q}uR zxSkqK7^g^Fx}|BWcI?Z_n7cy?x&#Ku739kX9t@=Bw{1w&R5WQJ(X@JYct#mn43FA9VdxPwcCpg`%gQ z4h)ui?;-YOhIm@?`MJYl7Cehc-o*+EjFew`n(D0A&pWKO!YsrbH$4Ofj3!|=pl zHUDzP)v=a0UEzHLFTde#r34(WWj%o87+Q(5VsZv<*Lsv8V57ufAw2Jx?}dV)FYgKG zQ%kM1#5^#=DkWde#i6Ha@Q2GF>f1&YKcw@qOWKzFf7p<&NTr{IDRFcr`C-fPO84a- z&8eUD-v-M&=gPs$nI5D>R+4(wI zkeflsjubfW?#sG~q8(;Y2NSD%*HM|R$dKyXD>_CSUNK(ZlgE=fhy{;1GkQK9M^Dl@ zz+BAho&B7}*_>;itp0lzi*z0TzwD=q&J%}&r!BJ&^?Dz{5{vuVB5m0sV$RiI(Q2e?I+Q`Go%{GLdY$BE8CBXZGXvTP$yAH-PaUG&Yk zy?G2868uty0;KYPo!LBpZiIReFpniwsO_6d2R9Ha0r>-poI}UBdQbucw4wGK0(uFZ zG{CA5AUtg%oFy%a2Pz4nV?3y5lBWPrY%wbYGr=hvJ*d}z#1-IuY=pr-cnWNaE+FAW z`30zC0!C90rNRVl3m~MZ7>(mX##mcRn8H;&3pB-EnHIn4g_R)FIvH=;rIn08;56;+ z-4F;8*ST-lohw7XFYoF~#o~6PqG<6ZxhYgQ;~pZ;#hJ9VKMlMHw4-9o$dlj6u^guD zoK?b^#;XE3P)MvYun-54KQ)HG>IQw|OwNHW9IO^;C5Cb>2-l1&?spYpoN2ahR=gE>DBW#^*_c5L1%!6S0YoG7U~8=7z4t&{-joEP3UWi3cq4(8~ReCN8?)k_KySEZ5QPjTg%t@(`M z=>3&wFPDiY8hoU&mk_Il5n*Y6+CkwOwnP{qIC`(%s-!GZQs@x1-41Xrlb3%xQ-*!CF)}l`E)NAn0(Rg zk4ea)A3D3@0?V(hesB68xCQwtxcVnh1V8xxKhVuK=5hSlPAA%U)i-LS8P|b--(jX7 zYSzwpg~tcA6qsBJOT&_*MsdQGEv}6~n^EX3`K%T1Ow!l{7I1-k!!%nAL{FBUpA3~F z+Pb<4{#7Mz!U&M=i&aT+BvJO7i60 zG-F!^;0CC0aK$4acp5?sq7RfF=Y*hmoBwfh-h30=&F=i)1{}@X3zLYD_U9D}JJE(^46+@mq2nJzZOiRm%M((3goJHVtX7XnHZu7mi zxPcK29JCb{69)hb!l0coB&J&71!mUFsi%aBX(AYP6BqFqm?)6;yV!_`>8UVTRQqnAqJ+|G#??P_;i#85cCW zbG8STA%$O9v?!7tnK4q{IvQXgI_!c)Y$sq5V*gOKn#l!?V)Q&kaGI*jiXz!fA*jRD zG1N*gKZHSSbq%8$QI5kTzp}^MoFC@H*-(Nc14cKF@+2n+xG_*xRH{O7!3lC&+0|P4 zh3MQB1})DkmF-(^K)Mu;*v*M2)TSS{^A6JQqYtBjjH4HN_UG5O#JMM>)|Puk QB`9{jq-ImkP5qhpKge1u+yDRo literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-2-Micron Executive Portrait.jpg b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-2-Micron Executive Portrait.jpg new file mode 100644 index 0000000000000000000000000000000000000000..ea1991cef0073aa15d62e688c6bbd2e623cd160c GIT binary patch literal 249370 zcmeFZWmH_z)-6~BDV(6eArx-Gz0lz98UhsVg-h@Nfgk~bdvGZ%Xz)NN+}#2c?!h6* z<9n}1_vpUwNB8^2yZ7rc?y0frPo1i9=HBP*wdY)O)zkdb8sMdpoT3~62?+o|dVT?( zmH;vUbTl*|8Y(&v2*kiZ$HXSW!N$VECLvg?Pj`dHHz$<42G%Ffg#Out{-nNO@jUzUKM=d^~jl@X=8hQIU|5=m0P9 zk&y9`o_YZ^&)12H^xqG_|8XF_Kt@4D1EOPKVm+VG@DlI>2^skX3Nk7x3d-}@0neWU zQ1DR+UUN&K5o($P>7YbBAqn~DAnDp}Vy%f&dR_~+Pz+3xSEOX*42(?7EUbL|0)j%q zA~J7f<>VCweqkTOBO*UVMJFbGO-@NoOV9XL zP*_x4Qd(AC_r1QMv8lP`$IqVLUw!cYfx*eC>6zKN`Gv*x4aDZw_Rj9!{-3k+i_5F) zo7=npz=Z@r{vWWO-~R`&{|7GoXIw8(P>@l8|A7nXh4*tHK)hO~#1?K7m?ZRk>kNPX1KNK__Wur8=>HP3{}b5%g=-Ojjg0g>dC2$x z3Bb-1!1m;p*)QlJwBQLaQ8yPqR}(n#`%L5F3Gf$H{s~ZERZWxkBAWT%@t+t2JhO*_ z&dvP2C%_eE{1ZTK$W+3#`>%Y^fXvhr;ER!6^)W!6m-zR({J%Zt{Ci#gXXKp!->l0q z8c4E4d4!X}1kpL7Z0XRgZqQEg3===omDSmIkdaDN0fQmB&>cCTOx8GFwJq73p?&3BydT$UFkndQOGMRXyGt~1FogpX3umXZRr^Eb$^yCn# zS^d$|Cj1>odfgV8Ja$PfEg~!jPf1@2@*;XE?B`v~lrpmXrpdQ0^E4tGM}+P5#VD_) zM5HHQOl^Hy-NPE8s2p^F3fGMbDmz|jIbkf&K{`_{$WHRB5NIpi#VsD9IJRe)v}j_i zG@*|5i!O^tCBb6BlpJDYG)HOw|^MdyM~+RQK)K1Z5pHl zJRTF zVPLd)y$DfZEOtNHkL7Aug~q~tesLEHSBI;1p&NZ$n59<99acNdLSPYnto)le1(%nw~4p;6kwe!#Gfj)SBh3=v$mPTq|_!K*%HOIck!m2 zw43WYr#L&NAi{Cd-X!s~%@^HkF_(FMRLyNa7=15@xi_yhA{4{|5YTNulems=LE#Ix)PJNw1 zxs~+j51E2*R8qNj*W<~$A;ArUCo?tMr=#N?IXYqb?k|jPJbu?;ziSw}-stw`veBq{ zQiU{W1oBc&#|*T=BT7fDj;o7eaQ*w<{G+E>nJ2~c9|`;q+$6EQs;IvT621CMv2CV6 zLdeVH8!lpm<+6aB5!8F3p-h)F_@!xQywY8o!`JXa8?t8aOh%LJ~Rc-WnXgu#_QQ(s%Fo}{4gFN94rp| z{b$5YWUL!QI=#NXD4|Hu#jr2_bmWWD+yxuzs4k>)=+%8TLvBX-8SQ%d{vY%e8N z`H-@r+Ji^gYQ|w6nFdWJd*{o`)Uj-U&lj3=fA|gA&ElApU-s?c&dQm-^G}G&imwvU zY5ex^CvB5V7_%ZbXf&jt+|Xl>-Zn4Phrkhet2|VJv`@}e6hRbA9ciyKLY_k7Kbrix zX+17C+%lJkQl9<5`JFt)qlL&$v3=qAN87DNpHBc6YiBngQ?8Mftq1pk7ca7lL-EhT z3IeaJwHL~`+27T}blI%vYuQ4+l$VF?jFnW=>I>A35GC?eoYU1?1>dGq5)0jvr;u$o zgy(%BS;*kWaz-^8n>H!km8v$%>3P#l5wOZbD-=ISdx#i*Ujxr#L6-9RoY z(eqRG4HgXxGT84c5gc*%o{u?~&O5TT-V1i_?1iq~Ty+HCma#QeHoUaGYh!wAV^xDv z#HZyg^_?;3VksGeVDu|p%}(S%K`Zw$DxWGHTPy2ceL}YQbrhR;4F!oJ(>P(6)8ka* z+XVUrx_SsHh&IQ6?c$yJ?%@I5}__+=LBqvvzjb~B49?m#Kh@B6m zB^}!kKiSjc@bv9gbh7ejMqso;pbN(Ncb-@L`7Y%~%kb^@2HS3G+n>?lU~gRaTg*(P z!L)C$oI&Y%kg=3OrG!JN{s)JKNilVoa^ufbavQH&J~>9{BD+J!!{*|+yTm`vD6g<7 z%;d)@J8VYV#!U-?hUonWcuEco$Ri_5QskBh>cf(K(zre#St!@ntVQ43 zzKE+)!?@o`>`TdQh_ZCc z=oJlq4_!@bmh{kL^${sk3B^0fu?^sC^~!-D8-()DlB_Rg*6l=7TAuew^8Kf3@dQQ( z{>0U@QGc#h9d(STx%ne{yei*Pi}d;#y_Cs~B`@3q;}l8x4N`LZZucpV0NZzPc8zXA zPM!*ivimqLyjo(srS6U%GU`cjVeKt0?Y%MM@_%Zbr-+5Vs+rG*dGz=S5L~own3s zKr{RVxD0pI3@9>--Wa_7sF{eVm`;=^j{+d|JO(z}Ib9wo7LZ(jgVI%f=nu4_yaWUX zacfjVS9oV&KTSV0aS$zRT1mmD8r)auPl4h0GRWM7UXNkMhKJjLK2l9iSVj(!D;#i?WUt zTM7lXxC*0<_MdHnsVBl2^IiHH3NCqIRGq(;`dYaHge@Jq=3Q{AfO3knn1OT(*1YV3xr`*$*81>dvdy$6 zuP)Yr?+U=N0Hm*ync6neKFdMWGvp~+@!#D$WYCVm66A62*i8&5oV_t8W5{#So(;>! z3-loX2BqWv!WU{JN9i1!%nE;FylOyAU|B`SCUkEFE5%o@lLz2y+#@EwPR<}6Y%(Cl zkIeuHn3L@_$r~3+o=W$ZTi+>L$^3$_Tz$*RP}>oPU>$$F_eLFg^WlE0}CP!>-X%s=1R2eYPS#f2mZNe+~ahf3U$ zReqIW=LNdW!~XM}M{WnL{WN0wn0Xa<1QOW>O@AjBx^@W!nDZ4-!3OlX8 ziZwnzDo7EOp44rg#PmPjwU7L`)nfeF$(Gl}6!Suq((Aqqb;!ZD*q)2cSe2rdj9X~_ zciP(H*Bh(m@7al>Q6Ovh@g_8x`ny^uLD>Lc4C+xTL=0gi0&^Suc=QCHX6wp z8X?Hvbm^%;9=}UCW*)D|zIXIAmeaIy@Y)r(?fKA)h%MaksQKkkjXvVcHi$ulYdc#IV)tM z>|T6(bzb0NrI`@QB@dPwqiZ`DFIDHMj~yrP@l!^pDLEhp_z$aA)V8_Z5`+5+e@TmmDSa z56ej8(AKDv>#0Y|(?FBv zldOE9ds%M006h8W4joNQDU*=1a7;;wcW~mNAuC6U4=x^R%cwq48tPQ?M%6?UOUv6E5PIdaHVF@*3kify zR7(l8nxn9GfTMi5yn2MAY7_7`r2E7;!@AR)ogETqXZH2+61L>e4A)b65d^xd^v7e< zoFUns;~>aW*!)b}*W5Qi6oj{@I^(f)M4_wj%~mr#J@_n{>?JjZ(Aq@NCD!-AV=B%& zz5qe(FzBrTWJhKxZ!`$A)~f0ztf%vU6LlFv`=Z;J^iT~qpvEEonGI*WN-<17=B*>s z?eu1Ig*FGw!kr^2_1FWMI~vSgJjt4)gv^D8@XY)J`KCj6e(dFdYG+n-`YU`|0(z_D zzXM$5Z#U}N6xU#)Iw>0t#y+UMX)d%cYx|)C+6TXB#-h($j$X52wd%bbBJHvrUB>&M z|8D&^MeJ!$X#zcVzCvK|)!5CH9XD^}TpMbDRfn;I7p+-ZycSr0E+0|d0IjBw+j^%| zvya}#w)-G0O0m089J69<=9F-oNbB~Zc3(bUn7X53dzIl-wKsaou8Q*tFoFRZ#`JSK z37MjvhM60j+q0a>K`hY0{#V6XFEwR}0RQ4Z`iD}+!9OVRF`1yOn+z`q&7^z(P?}x4R zsQ0vhVSx~MUa_jDPcEB6Gnk8u_o7&%uJQ|26G1nx@G=}}FjpnRVvyJv^K&QDYdLwu zY*}0I@==b-+J|La@wAdpPU#rF^vG)A@`!K^$Zw|GsY+^vomCBZBcsUN`C_#Wv6qlXQ*e4e)GJ)0!uNS14jb=q$Yl+YJw@-8RJ@y3$zw zBbkO->>-G?D)H0kZZ<9jgFz!YjkZy~6DMg+x(42@_43+Z>du{ZfF+zSx=@kG2AEht zGn?bANmdA@V_6IK#_QoPcnIYr%fL#rg7$q_%T0560_eW|K@?KUW`6CapSzFx3Ms)M zmf%~_aw|G~ne>k4eP3hc<)`Vp%S6=`~t z?n8_cUg{PScMsNmgncd=zaYM%a`@radq? z?ze{hV|uB&En`DAD3Pit=hf|od3|pA)t(BbmXWEEGQGNOoo^6tzVu$x92MaDMbT)s zJM|2K-BiF@Oz-nP7|FTfg1-z6RyasLU)YH48-5+BT{6CNr)4Tbd=W~e#>1%I z3a$@xfAa%@0~dqvi(+RCsU;5N=Xn2YnV$k#QBnhkg;6M>8N)M`)P4I3&S8DZAFAtL zBIZD35B9@!2o-F;)^&EJ`4q-jgPLQ3+?`PH{?TahTxO-S7IPJ`r0SrGb{{k>EK$Vk zm^|{QHNNb5oru-K!AMD31$rI$u^PB9$v;$|5gW5}cep4X5PK5_VWZ~k#Np{xIxkwq7M^I@hAl`e5S6eys8+X{{{bOX(BrEcMh3V2m9}N%*^R;2UN3Gl2Kf7 z&049Wv#hY^2{pf{Jm_yw+_V%)z^vUACvy{x%ejDU)A_(WI^#Xaq}hOdOcjUYtAQAfSiV82<27kAF&F0wjb;4g)72(rDcB8{4WqNp@px4y@o}X^(PF+z>TWA2nWJm&bi38_1-LS% z8;-7cFh_g$EVx@I?pXEZG824D+=V04L|Q;^5aW>OmQKT}f*h(PUAcE%sPo>mda$^{ z#E5iDl5>z*Qa^3nM@X?O6taE!>TtQT5+`a2fm!*qH#TTns#ir4QWhmPaR`x zBOU5&l3%oJ7=N>F3k$opZZZ?P5W~+ceHJfAk{cFN>k>#50u}KGQ2P-e`mh_C)4UY7 zmC_QPBOYV8RFfwnLYASIA!VGAtQ@3Kdj%WJ7#))L*RtXyx^#*lL>?2R2I>umA_e4Y zDHt8_{5_`H7}+RJvax{Rq>G*C1jdgWQgjj8VKhvC_S8h`Sd+-R)ZDOkB#OA0AHLSo zp;LfAgMw@bX4Ec)xOgc+Q7Nk@+bM`+(Bph8%W^h}7~B&tspA=jpJ zB-R~A#V1dH%c^?_>dVByz~`+%g*3>C`p+aeyv|l_5&hl;MA2Q!bSPS&Oc_?n+c6(i zl|6jVF<2|(+6k`Te0?al{>bkkhOl4md&eWE^CgH|CHNg9H5&`O@u&W+UdB@UK5DCQoBqYM9y(k6tOL6#omix^~Mg6Cu-?QNM0VU z`|=UEj4r?9J-*R6v1QBAb8Il-YOUGayW;Z%u>A9j!nrW!OJ*j80>JZbx-leY-(b6b z1v{89VpkGTG8t#KX_rPOBU1cu9Ny4N4T#DF&Rs zgqK!G6TY;y@XX;+MxTEy53m70X6Ud^j99DHobCXoNS1bn}; zjP{$TkVdor4*yn1ph+Yp2g|r0T_&r-Ikux>T2F_?*V0nQOIHj*rcaG1QaG>ZgAZDo zU-xpv1R#}fzhY_YpsR0-h0L%EC?M^WCp->N#*bDk7kv+%JyUKXW9<~W{}Df1fAG+njqkqGg-1BwvO>shm*3rWxCgN2I zwVDI>vH6ELS-bG!^@JzDY#EHfwhH7TL{&wa5EEGZIg+K;yiAt*t+&Ob*!{#{s6m!L zBH@-ib)>E-)YsgFRSUDiXc>uY>@UvPmdSdn!op`drCr{!%hIj(oNakLZmE!90CLXs zv1(ijjx+%YRqh-&x&epfshAH>-c?d>!d2F_}Gu7WWOTMHaPMlLkaBek~_1Lk*JIl zfn?fK_kJ3ZJ+@BABXh=pR zyImhxdQ~Pqs;6U>tj8oq0tUx=h|ESCiVcUjMiP{13H<#_oldueX#F8uz86tu`w;_J9`qX77GS#C5)!<&mjm*qkg zIjn5^_e*SlxyX?yL+uYJ-8!({gi%TnWbA%HtQ2IuD>t4W3iw`?gT&r)qK)xB*AXO4 z^hN%(srB#ceq9S&Y$t9Lb&dODvKUb0-ttp{TKCz7%IGEs=DnH5Euod&w9~1+if*ZRR?8 zzdk0jSh?JTZ;3-oiI>8iTcxhFTLsJa)b^001dM*XkEDupwSDXkNp;Gu3l$k*#Fd|F zKqbA}PdLZ;$n_pU5r2azI9&EvLZ2!Kt!iFUBK2AbzX9uPWG6+(xELmt%I8yFa>G$yKDClIE z4~JJI#DmC5Nh6hG$NNW#L)MT&0~-j86`emUmwkZ#fDJ(1iIh7Bb-*al_#pcjS#a;* z5NW!)9GqB?qK$0kT@8W@Nl$oybyTSg_y{=zn~M9#!&^wk<}>Ez>5SFZ+wov#VRs!7E~ib zj6OR#u>xIx$t&~`1-{##G!#9AP-^7APu7(m6uCT@YlI$YZ99;A3)fF#%3)DWDe&Nm zJOD5Gk$X5QD!D>`-Q|BxbNAFm3|}Zrc&qeW$HzdH=KV1T zW~_4t9NpePh+tcMQpRKn1;WVZW%-jNgdbC{IVQY3ulZ*d_iSsEb=VE=kNq0#7&C|o zJl2x#!G$&AymUK)o6a(qUu`S!cl^zqGSqGxkDIv3 zh7=hk9;%ZvDzx;5vNmZ(+LC58x52_AA&T|a7LRDTqtif3fM1vneTKrZQoo>lRzeQo z3RaOVzN$LrXeo{Bb$D{YuNkG7?re^m-fu~%SDPMFHb3?iNb>`WUL|lH43;d1k-wGc zD?-1#fXVxOj4{Vz^eb>x)or#kh3Lgx%Y;h7h+?uF^^AqKxd$kiZxyVVrm1DMNv7QE z8+mWY6M8$$twhx|pdqRWJ2;_!IYWJc^~Z`v%8EQ#HKgD@Okd5U=7f3NOP-$#45!at ztnHjHy@jb9Ije=*+r(9G>9DIoNX?*E$fZIE;&%n-pmW_8p`SCgWOVY~?`~w_(?Q|# z^uId*67#U_+7B2A05lMtB~6BUuJ-1kUfv%~^(C z7cm|p%5rBDPg3E>`T=B9Be;jeY_OiL>CSh|30^kpR0xVu!IEmx{tM&MTa^ULa6d>+ z$+R1eqiXSr6dCvYVK!-f!mU64pWiN$K`Yhlr;p#9=&|AjNGMs>9s@cW_DD?WJI(;n zvaKeRGPJ-E(;MwBXg_V>*TXct0KWZb`snGRNfxwKPgdwMr(cX$)f=PJ!$zgwwzhh| znaG^r)hmd6;ys!r?)UQ?KMlKi-t=Q@l-jCE{E$b1)Q)JSzH112ubdL z&J^6Vh(GWK^*82*Id*2JhGZx#EzmD#HQi%*j}7w+dtUB)RjKAOLO3Xua+dTv4%Cr3 z8>Rasi>z{UeqQmYGaWoHeKQNF0?pfkSY0AMDB=wG8@C#m#e3=GgNBZ>64ep1!Ml|o zz#g;F5sPGWa!f;p0PeqK$h33Z2ni&P3!DH5t@0L)up=H9v>UFsQysX`PNUS8vdlUp z!jtma(7-CP0I3^6d0BE3ftE4D>9i?H_KLQTVi=TNi0C8*{W((W&1D`#NtvwID$~~v z#gTmn?|mo2(4nff${1y1k(j^97BYVaxg}rU`}D6gtN-hb-bF-TZ@zvHMq(PrwS8gB z!H_~!udyDxA5IJIbuK8#u3Qx=qlcYdJ7KEg}YVVvmZ zy(thF`2?tzQY23)9$b~-pBawl6hHj%fTm7dbTwcTKM+c~9wz*=(I8ikDWgxpun47V zQ1XCw5wC+y5?uNzpcvr;q@XiZxEOO8Od~AH`}&dEa*{A6?Ci~kdy?)kAE2Bu#~_ua zr@(Z2t2iW1`nlRYg4?42PXs3=$H_2_LoUju zu0fon+($8w$raellR*WvebD{^33Ypmy*`HsHyPw{WQ-(TdYw>2I-e|6)kY*SM9n7ouN#=ZC&s}FarSpp`# zNU64Y%Z(Jj{tQ;@>S6GTM7T6@Ll?kCZ|Pe5qBC3RhkaDthDu|A@4dgiGIfN&x)z_% zL1Ivmm~B~BYuF|5UY1FhCz*|u4TWU$II9I@%UYP*-~G^@$O5 zZ#S2hTYSdxtpJ|deC2YLZK5ptk>tBF>fhJp+W7C=2OjL#S;In9g!V;pjWWXr>&lUW zG5C#?-`WsZc_W%(G;cVv+a^h0Rd|9+o5|iU{B5D##O3nSrXE5r15J54Bs9?zhRo*s z%qL=Op5eSQvlu7=MZC#^ls>Dv-#OVo6BxTQatTs96nU_7wdLHDyEwFb0*R-WYn!d5 z4=5L5xQF5fv=CJk!0R!va95fi8z$Nl0Nb=+T>JVIM;_l1qtf9Odbd;~t%`d4ElIG* ztmB>MC}u^3_Ow~Z{XvbtT#x5iyH+AjoMA*@gM9^WGKyQ@;2Fv7lx?4s4bAK~+oAz$ zPtbJ2)G6%qr{$I{1@NcJKh9%WYc?Sv#^sTsg&OSL(xoVtW-+G$cH+b2;M;D~&IkKR z+ZW~e+C7$t&IT!_yODdOIZ_1CbU(BDc*iHeU3q=&;|VVMDIf#r9!l_ zTo?XsWXxmGXh~=M(MuWECjjUA7RiOxpNv9*Plb$v93%sUYdPakZP)ZO>fAD$)$bG0h~U$$*eNx%*gedAT{c#vc>#am!EW?RLSYQBnA4885~0fz4cdJ1FQ zIs08654KKHtF}*oNotlC*e~-cbMoTX`6pz_8Ge`=nGVl3VYC;W^$}g%1~WZNNWzt| z(%o<99;Oq1(I)LswBY=v&Xt*)EpjGd-!rAnSiyc($-wMYK}=ajN%J8nCwWr7QfiYU z?8QXgad1p%*IL5kDXm4@odNYAFm#u7fWk^C;x=+Rj&-G@G;vwZ*dtTl{%A8)6mfA~ z@v=iaVoB$8Dl_2GO@)TuZ5WBdc%r8FFXH@9%DdwBA7^)v%nHfMNkEe1riJ#vCM2Q{ z`Y7w0)0$7uo%xsQ#Co~jQQQXuj8}T`>_xz^5tLkH!1yk%bCgjxp4gete%N0PJsiY))wB1AxoV%m=PNgW*d}bE!X+ofcE{-WJ~fzZ`pGiHsCmlJm2192GSJV=p6u z!YT$YEh#P-L2SXN(Tq$J?%E=7DVnHgh{w*IC|}I7YJos;lUG-Yg^jK$BC5Fc%%i-m zy344;G%-xwGZG-8ioz9kcvfVLg=Yo`Hr;`ues_gDkf@u~RoO=w8>Q?G>IWerM+A_M zp3?%5=?1vB`_ree8NjmtIp)l6fVyumBuv-b9%rT+GZ7tI?U*NZDy5tR8t+MDJ&@fk){o9pw?(b%hp>tMKiNg*)j0 zcu3bP%ykEOwZVnMu)`5&XflsGo*gIa4g#|+G~2NLsRyLfP*D^3@$7v*$%r<#ZWutp zXA}?+345@=Z``!}aRiy4lG3gDDiTFFY*G#F-V97`qjw)=bmx_ZrIIuIYOTa zOz9Vw_eh4AElS8a`vr2-+gK*~h{*MBF@L4-_hYowT>+x z&Caz_z@I#kJWsGE5|i)e zbbGrugQ(R@sSz3d)0Lc{jvqQMoJS$jF-64PG+eaTCwEhg17Zoy%=Ap6-+L%`gfB)5Z6&57` zcs9!#ScgjgCwid&Gd<-0>OQ?9Dmgz5)D@utsy=OJrL_pkmkxP}GuuulOzDh18eHY= zzQ1%Q9XGSza-%A`k_uC`>?}>z_YEnMwwy~2mWTBD*({Iq2}%B0k4B@nX?(R>iFHmH zL1AX3jxZA{C;ZJP!L}(%9q62Fv7q&?yzofuJJQ@@NXG1MnGGVdDGf+LEh; zHlOv}=>^24ZZ%?|3W0%2vr%5O$X^O5ud7%FzSrR^Bp6Cxbnmh%OgXqu=YaP>8pEyO zMv|4ouk!j>F&%Yc-vh3;3-Pmmzfp&wmrw!!zV*}DtrUr|$(kvYF?aU5@BISU)31>Z zeqY@ng&9KDkCKyYS?eMRByRH(-ahW9X3uhVG0nN?i$R`guhr8KeAh!IXKNx5;Uou z>e=eI03RXT4s~QDI^Oc*-dE(w-qd z#N(KttC*wfba^Mg@~j*<=zktPr6?%7MU756qUCeu+qIKHIt6m?144C8D7i zXl=I3>o`e6WTRPapv!Uru{Nrt?8602VwAIGdTgyRW z3C>ulj0F%T<$uBXv-gV>(RjMF=TdyC&_Rf`@-fr1$|0BO-BDi_8pa^pMHZqymTu+?0h`U_ zccHhl#z~YLw%H%w^qddwz#NQ;A~oVi0I!|mFPLZ^{C+jBKz!L9d5ELyE=JkDv2P}_ z8?Lmee@80Pa+T=d9(<&0{FXxzj&GJ}Cx0B}c;7w|9|W0(nw*ETr*gA?=fsd36FTNT zx{02vZLa$fRisEgEBz}&kRsvzG=4vD7y1t0nn0#NC-ii|gg#s*aKm}upJmK6cU|`w z78~B1+Vv{7pMW*~$BH2S63nh9?rLH)PC?NWcMb*l+4v3;k320^ap!Y1OrMmhUGWPQ zp*a|9w~bj89;o81U+QdF7*xYM(KxFr`P+3WbzO0SE!an~0v3xhMk$qGJlPHSq1RBH z=@h_boW5b$#n#XvXZ}0XHK0a*@>^ujVpTeb-Tu3A?XVUtMaM_#P7{lCG>>CTR%gXf zU2)_7t}{{OI4M%lvsb=p`qw!}(dJJ(dbBq}EXzY7F*{Xwft|(d_FPiX;zFa3I5r+5 zEeyI0G{`rRx}!l4(lrc|o@~BaQpn#i#PwnoC|0So?2dJ)P?+ z7@q#N{5axvFj#>iz7Bq>e(FJVg76Cl;RIJ;0Whs)#mXgJ|uEgsHk_*k*28p=oS%9v@>~#Z9 zFWwT#)*y$2g&sdl*JP%~k0#GqUYFROqWpr*+U3!2r(wYv(Bpwdw#4Lo+JFq4S_5b<3) zvlg$irTY|2m3b|vT+&xojA&z7Aob7t$-M_Qh~kt$92Rnn;ANH3d0p*7WQn)C-0Z9N zY!LnJ2TMfw$<3U+l>;T|YoE=q`w`g^V_!{_ScOX%I(#nP=)CaQuQoQSpxg952ctsN zr;Ci=lm{yUzOFYbQ-5tgT+D8(-Wa5m!siH$YqzCP91vp4FPkFm1_^C^{N!jb(?pH7 zyj_Bq(oKmcrJ|NceOu-QSpR;b_i3bd;AgOVqHhIlFf@)8{RelMpU1w#+u;lb?ec{(UrG;o1RT3&ss(mfs@@#BJwQif`s#CP}6>=LtLvY>m~ zZ+|ldvXTK9PKC<-37Tk!2*;3#P2HnJsC~+!*vJONXP&PF1DPql$a+N6E^Nx0-k%zaDuPP7WbW#~Xl)6~->2dZPmKCD1WS2fz2_0sQ8G>ldQ$z(??151{L z)C?)>ul9??1^TtCi?VHvk!JjOrSKY{sXQN%-WuPi+gFH$yL{b} zeV@XEvGhBSIMsDikQOluBU30Twd5Hth(dD}#h<&kFK#7o`?K>HFL?k8BX1>e$r|3LT4f^7he`|)%Wua{C^ML=lZ8Pdb~6T%Vh965f)_XKD_KSn!! z84^KjDLAe3*77;w($|tv08d_bU&E%jtbetk<`~kUDHsY0rVg0;KN?mndKEhz_iBs2 z7Q^_{^Tw&C-F?=Zz+mMU&%>kwXxfGp#?;a*mx9vFo=|H$a2vfDFqH%6FSll2ou>QY z+xyAeJ5Kk-x5t^5NJo7wWh6ah`z$9HEijI9!2zBai-Rk4+Z7qR?4nQN!J|p!gJ@RP zwwQD5=oZ^PY>xB}!8aj5A91ahykLbT&gG4Q(=-+T*rXJ3u0MC5UAx>LVvK*Q^rH5d z3s?ffpz3%D5Bzj6oQe0-&=uF&NVFJ7@ zf2p8iW+SwB--!ovfANs3m4Q{}R(a2Tl#})Z5UL6@3boF_Dp?#CTCPez+6O-YU{Jq$ zX|`msqrHIxwY`w>UrwGy@9`t$P!i42Bvs$tN4ar=kC3#_%s@3hhN8@IG;@Nm;+15Lr2YenCZJUI@}Cp30pDP?aeG8H2v@g$@0h z@=Ux|kP$MQNGg5`))^{&_?@n0W8${O`)$Z|{I$Fp_&|C{CAnDP3E*YWXz*$?ewUef zX2LPioI*LRhpStRrg{!$+k-4R;#$k9P-G$}yp;GN><+5;E#5M7(0UQS{*ISA53&^B zSJ2r^a_6zCsSWONL$uJYJpe`+byTckIr+KPhny^;-LnuCfxcB$gACG0Cnm>nwS15E zrI-M8?wF9}2fsO8yR1YXAxiyBQ+*p)L~Mb(>9Qq62;f1fgP;usi|D*uD;{r!4(z~G ziE_ELn!zJ(yp`#=q}&4#?YvvzesCE;Ly+)RXRrlMsfDIcjt+=7*C%$(tc@-s|FTT| zV80)>6?Z@Lc2wrHg}ONVT8?{^BXfyW)wt44ajz9rP6I&O! zd?&3{x~BWGd@NDTFNcNCh^@*Cpa$*cTX;!y0Cyw@B){u(6U8&~f{y>$wa3Ai9pC`~ zLHR@ju6dRLQ0;yhTscx*U&E~M%?E!CO_b=ousFOuU9yl(&kF67cx0MbO@d%Js2!QW z#8d42Ru?1C%vHUq-84`ER1#<(6J{sWMKhQoZOOcejX&3}y2UxmME{b@eKw;UV~6ar z3|dFkEath97E>lS6-j2gI+N-F%{tGCq}*=g^cDSqcYdeGaQxh-5|?*z&!&*BGLjmB zLf{_{QB5M0J8To~@sA+ee6HKresJzWM2B1XFyV6fZsJ}}At%~z34amhi9lmtp(WR(I$}T@GDz2v=Xd%Cv=(9s-ex;XhJUuu>_l4&PQFtGS8W;l99s z91H48S>3k7HH`6QUf#Khzdd+%VVw&%f7eXeT8DJs`D=6?)DIo;>ePGjq*ZO2w|r0A z97QAbLUa=A>zoejWAUZ%R9%stS@wQ9n#WzcvpKI4SoQvHxYNY&YHPJHV5-@Zc*1Cv zvk2qt-M4lXO61rzcR1uiOs9scXsQxcZ0mw(p4y>=P&-;)U&Fv6Y;p-skPhV(WH*K2 zGGs8ic=C6&TJRwT@xg+rd?%`NKKkJ5&YdMLo6r;BC)gZ{diR_gw{||v&LMrtNn0F( zk7p3Ph^ug%T$J~HiQ`Vn#{78t1fbMs@rRHbM#ldUR*45W#vV{U$XO{>dJ#e{I8# z{zVh_vMAd}QS`9}jS2xN({+e*e1aqup*%x&n=ymqK#m_2E`A>*vN(R#W3#O~@Fh0O zEof9i=D<)N_gYYs-9Me^Rv0dXOtW(ebHMH>AuoSfTcXzW{5X;gw@9Bd7Ef03Mg^)`RxhrjRL(MwS+h}49u z1mm4|6@foTeJ^611~6}~J^PDykBH-bXO#S%;NpdDxCd^ z#0jNI8wJ*Qn*kX7<^XJpiaT;gTd@WUTy+2>BZa{#;d@wa>ygRf?%P409;O35w`coF zU;|)VJ1JOJ?Q=3xCUr9H5G&wdYPLjWjnn5q6F;1^`&bpM*ww#DJ!3nPh@1irK2Ue9 z%)^e+QOU8C6>%2JRT2dJ^6B$={wERkzMdiNez3?C-1;`;n&0phbkh~g3LQ7^6e;Yt z)iMS-#H*>zB|y;PF2U)I7I!O9+}acZ!6gKD zf)$ryMT=YUBEbt3Ym1cUs;q~9KWM-hR61P`c$o< zFjgEF7rn|Dk^9l@RR2E!I--Y@|AF6T@v! z=HY=`b5ONne-9Y=<78=zF4sVYM~)XN+am`$Ko^>Jf;~KwW$L9H4z=9bXc(I}4&i#$ zXHN`TG)go5Wx%Q2271G9KpWd=g~-GjlJwfGzkOQAXK^}0&xTM@$AgR3!IndRa#98_ z0QaAvDOfWxM@=MUh$69sz?i4u!xLl{q`NkwLriDL73NApNDuV7ikd9uTdkN$B%$IMY9_7C=V7a?TNwsl%to;`ja&JRMnI{ZzR-#$>Igl{dMDI8CNyCJlu zt8PNEZhwYB{v7R8lkBgmz$46kPmKr9Y_C z)Op&Zh7oVX-|D}3F8vrx7MHYfR`(lsrL~eRp;%aJ&z6Vi^nqFkbL4GAuf->F274;v zlLDk2%1=#ohk;-8>yYL<6T*puWuFKb?BuiZty>7>A{44@Yyt@Wzwpw%b~@Uk-h?*iH(XYhdASO zkBE-NpvoLkT466}&Lx?nZ_cLf{qRq6y-H0df9*uaGG#BB5IjSasDPRAeO<~sXCj|J zqZSEN%9aRc$z02u$)6j{_@!p|xwXY4NWd)6*a3>e%`N+(#7Y9g@@q0sBXf7rB$cNh zd`JX2zL{t$oqU>`!{tJzeK8XxS$%8 zVyEK@7{j0bUVVt|lg#X(+Oc`FsIu7%n6VnJK@M4KY}WbgzX1Am|GMM_{@}1q8N=}n zQtxF!ae%m0d>_)|c5l!7EgSm#XuFiX04=$3mz9}t`*(h7jb^#an#iv^az!b#z#YQ? zhfiu|Pyf`ZaZ5;(v_~bY$ng=S08&OKyo@67gofCxNP2S$l!-U|x5XrDNro`5ihkQv zf60;#S^sXPs+q(hgs(0_v{&o=LFKi#l6s!VE{36gQ;pd(2&Icm%$aTCi%OPY0CM5r zO5+|ZS4Ys6(*yQE20erfcwXus+vN9!_1*Sy*n;48s~_oyjaYNp$*a|dx6p>M+!&Qf zQVKIgvZX;!nN$qdl!;K!4<}47?NtBhzRK&%8d&7wvEMepA!V}bpNqao@r%2|cv}+T z-m3~j)LysGdZoYzChOY4jU= z^lWF@vHJ28G8VspE>^vh7Qf2T0x2VxB{nIdyK*Wc-s=!<6Zx4;}*cn&C^b|Q*oR6I2a|*52T$J7OS=PGv8tW2@S(=JXQE6YZ?Ya+UzyN1z~5KkAG@i_}N+u!mAlg?A!Ja0@>xFChAfGcK~!sm*~=Gnh7?DAl_7&9>| z$^QU|AoooQQoCtSM4sRRyy8iGYwTt**qAy!^?Qp2g}bv9L{TZ<*iuoLJ)42Fx&ex2 z!Z7jRRI^1bW6$p(>hWF9WgK zT$*FDU4Z)8cZ?#b_c%W%={U8pjP*%~^IGmu>L34W+;=@JP(A|h7Smges5xQ6i*o1S zy3l#9uud;s1I~Y20{XuzXZ}sTFXkVLh6Gi2X&_V^=a*QT8KZb_`QUl{U>Rr76rt}k zo7e*dH}*n-6-m2@K`AetX1K72#Io{5p92FRp5n`c!|2`YCR{n)WrUV*xFgknNE`Zx z6+1UG3T}*U2saJFnPk|z;3{{Fg_I<}NfKX>j%EHJd#9}=-*wW&up)7B^4MWoMIB*Y z?~+VBipWcsMHV;fdW`5Qyua5w(fMNp6Bm!&Kk{8PF+b(Bj~j)$1TC=B8?3?yXQmpd zfYLn6kh^gYLFqdjzfAzg?dvLGgX7Fx)jo|R`qhw9^P)f;|#Wg z2)9jUjc?3Zd`_d!j%s8`Oyq2p)Jx%MB)5K$VYyK2Yo|6&+h*SBIxxe49%p%H#6rm; z$()z?f=sbtPgJ5=n&3YO!yS4UhGZS9S(C zxp9t_pETSD-S`4FY^e%x06lDn5xV}{Xy<9}GKR|Y%7&}m(t%RBC%ZQns5#d%ANM`h z5*g)X6^TbhfqG1NTZc|hyoFOa(WyPaKUW8uXoJyT&_hh)ULfc*Qxo0ek0Qx`nEBGa zfo&haH`9T!KYP4Y)F;utY`Oj1U~XEIcLUEkkB8mAIEZN>FLsjE2vdyhT1|@jchsbD zW(tC1kZHq@d$BByKLl-xH{qTlOB0-;+D{mb3Cww;wdSk-9~oJ>Y9sa>;FYBh0%P8-JlYCPa@|TNNcupI)``TyqcDGbOuQ-uCoULYa&fpC!3lC0V!Js&rl)on2o#>dwKukm|S69~5O4p?MgSPo0sw6P}e!gfcGvfRO2HpZhC4&${(jW5q!H zaNIwM@^_V)q*0P60YBzXI$OP7PZPp0yr8%*E95a0JibTLHD$!8@$V^Ea85S z@_3(!TV-kBqApfSEY)ZNzQ%u(%;58i^hf;JzNolg{*W+3xgM3wTg>J0K_mzs03zq7 zdnT`~#?5kN{oD{mz#7LC&=hejHIJ9u6f_u5FBe&`$4vPZE-x58S1) zN@sFf^=COxbKM5&n+{9pA%F04b->=&v*DOz1ogdIjNj?{1DaLe^wD=Z`%21t!j~}r z!*r#U8cNQO>Rzy0yBMa7ni3kWhFtxu2^bznE90A$c{9I4^l9mfUyB6z-WYJSsb;=J zbtJNXT4-euOSb3yyg3Un2y03e#+fW&=9}lzs6?Eb|IkRkmSfgYFY|j_x#$xgU-4sp zR1wo|{ceC=hgH4qrc;z_Py$>byO$k0mgoFB!{EZSD3P?Nf>26J{@45~4y_?h7THPl z!OnJuQ6T3P{}7PmbqAnuLxR{%q&{u?QWLXh~1)ayhH6j7e^+E z+zfb`?;qM)zJGZu{8y=ak^&hju&j;TZEoe}vUz61&--h;fd!ns-#7s}3c(91+0wfQ zs7o;?8r~G1@8C`0>H(kK-9>b%XSJ%2~7nJV^6bur5SLCN{`Em`BQhjU>^O#*vi zr2OcFfo#7Plzq~LXMjA^@dIrHOIASf(K1X`RyAaq*0Dw6dz9fp)t1q5bk=EeU1mIz5MzgCLmsjiU zJ)Ta2S?07P`+m@Fyp#%|Xd*Tnig~%UEVq`^{}G?-b-U$cK6KZ=mJ> z|7-l0L4~~2@x%GKZO{I>$&8fF&j|mhja~jn?aPdLMLD;@(|FBrlV&Ym&v9_`(}< zmEVbG8T~n>{Z#1!CVOWk^QJrE={uqNI1tEja+tvU>Eo4a8_kHhja!F1{_IbUrb}_g z$&yDL52K_3G&z$BiRr1LrrE8sl^`O zoiXHdBBV6;H5{s}iD7;99B0HXDNM}(cW)Pt5Di?aDi7(sF#tskzWqbdPCDK@sfWB- zfM<)OX)yOJ6g*4I0C=_5>Wmq5Z-U=b#fi8bYJfsv{aq*fAFL*XLQlf-2(QYE+Tf}m z-e!`Ecn97E!Mp8Zj!pN4{}ij5+D{O=Vnds%JgTqv@%WkCegGI)=678D7%2-v|50{Z z4RLwhugfd)eocXVaROTBV>*$o)Q6eWZeCCnUsx{oCh*ibFc)5D*Oi}KXyxI)C^8}WC{lVP{SWxFq(9`o%>CNgI0L#x3#M)Q%gxwfG(N}+Z>fZ(~g$%iOZyQGT zIltU7FI!r)=%Lj5rQC2ey*9U&_wJgq^)_s|#MmuD50oRz5LsB?yq|bzu zJ~5~3o)dK3O|jssIISE^MKkK|Nd0@?OfoLm=Qo0a)vVhQ$ug+9LmHu8&t^OS$>`m| z+oZvc?wFVW%C{VJB$JKto`eh~VT4aRt*5AiXxC67q6Wu$>d^?IEbPXxPT}^8%jPc4 zKoh-hr&gh`rICD&=R5_UB2KmM`&AXv*@nZGCc^s4mJ=66)AW4lMQtDyXS~B6jWmn+L*y&rWATb1v%f#5 zS>n08+L1HbR8g<>4E0$rB45;3lh`eGRV_)hxGJwV=9gW)R^}7Kbz(lW5g{kTe^d^h z$?d5CZ^V>yS&kBKKgxsjvYI(b6|te43WbsH_@KG1R42A=CZRBx_{L!9zO+M=d0ocu zvpyC6@83c5Djz-yZERd&kP}0V1i0tta#cpomn@f5nP_{C#%v&lbZF*6Aj7T1|4e634Y}XjYd0jHEyPAV;F-=r2Aqe!>%M`JtHJz+jt8NY%=!~%gvFhbv zaCwi6L>AXyh(xr~4DYKn*)I%!KK$lVM^@s~E4=_^|F_kD|6S!DEqSA!9J6t(TG#s} z=?+8j&1++mUUThirHu|C9?RF3{{Y~uX`Z2=4i}d#+KtZu$i9!|QbJ7XPit5qz(J-n z4_lZyo#O{?S{>+)k!}K`9B?l&lN4419rd{Iek81OvzJ6xF#4NDvHAPvy&a$A{#j^M z2G!5pOpgzY{aEEI=)nOy-=j#i7uDUZJ_kMJ$C;UnVbs4Sw=H=0gEU9GRI>l33fq$e zp|R&-^4IgB{?&>jqwgxa9^In+#HxIeLmh8_ytJ(iyYD_Qt;>9WZVGD&*@%)3UZ_ux z6){W0QyJq04ce|2UdrZb(WRcVey4^UD! zNuTTlzX%-o7M_*#zIf=C^FILay;#D_$Wu0(;41#s@rt3VDqy zKg+-NF2ilOEIw=&lLu$PCp5yYoWGzYZK|t%@NnH{h`_R=?JwYBcrhc~*Z$?dUa2w} zLm)l4@R3h2CaT3z|KokNeMyqJn29a<)0u{;YF8m;ikYV4L6w>iKKgMf&$rO}>@F5g z3#-4vFt*o#0mn*fMI2*XzF4h;J_kHos0;eDH~lfzkD(^{%DcMrY!>+SP#c^Sk5Gmb z{}w|GmUOW$tYcg$Z?XhEkTtfNLU8u+X9kC zF;*$r7ydl6KlF2}!IwyBEC)qaQ~^Dj!mt_Y0=6(Iz>-m0)da81Ik#U&1_^=^J%5*P z@7gd=`6C_*S7lh@`gO2=2TyUf9Rr~i%VL$@%ByMydB7PwiINHL5=&v0IZTzBT+~eb zt2D*hD{@KvuC-7riMU-(e1fuJ(@0`vTi^ z7^eesy(x_nWr_rQ&o%7{ap|;P;P<7!6DR>TPjlCXcM6YXnxC-QE9`g8Z ztnIag@xk?j(YV6C_4;E}G9g$@dH0fT#P%W216GwaI>1{*9ZyZIxOge6K1rlU1n#!g zLBj+)T3^p0RXAd8Rx(x z7y_y#Pb)D|p*?Z-ogLe(9u^`+)=he$=LJmg;zKzi*asepdQmibE?IEV(~}sakIZv6 zN2vDD5eUSWMpd@XF2`knbkf;oi1*2@Gw0|qtU213cQ2Tf~zZ@+3Ok3$@qQ|aUn zN9HKr*|c`*Y8X*Jr~u{v#psOx17b0#SJZYUs?{FnJt!iM+VyMlZ($!_)n$Mj*vi3Y z!@%vY3tfr58~->>Yrh?nF@b^h+=Z_ht>XE86W-*0e3!Yt{%61F&?SGio;J|W+CU? zngA)_2sgY{&d;0R3HPHa^BS&A&(^KZo%(}DHHI7I`oSsJCuJby?S7K z+&9Kj!DlpPa{5e;1+T+_SfjJjr5N?PM1w_a7oM=tuyRlf0A^Y#mClDX6uW28rD|d+rtsV<)q78o4qL(jF zUA?$ZaaH-<$r#5p4s><;8f;%$1RNbLmY%;Z& z?%;wF>V&r+-Ow_jEO)|>{M0qCr{)c!`ijp07Q>LuA&5opzBue&dzrQ_uBnMS*&!?l zT*W2u_JAcNU%Q1@xU zXUR*@F@V*NudlIqO8lzvKiTs^qMMxWl?COw(LD>#`j2~F$HWHKWWL&|0C)Xdd}WA- zwQ`v0AQIH{i1k}Rhy;DZDblv(pCZe@v1#@Yqwh=?w&4Fd>m1$_jM^>LFi;)w&iTE( zhUwJ$ux;772O01&`1AREqptA-3HM#2YdOo+D>KpB<*UGdM(LLr*eGsyKh1M}^B6O_ zO`kS-)cY27HACVYkoG z!Ry?OZc=$`cP}ba;#G?)nal6*B?FcUnp$=9W7%3ZEe$ua28@bpi=>-RRt9dg6TLz` zniX-rAhSe{CLpW{4u>q9XEN}qXV{}#EOqG*YWaigBafobXNhr5#D2NXhUTD#p7N?) zKpO+}#uzcol^VV?F&)vmhda}p%5KZ*q$g!@WOFvYJ*;`A!WGsoY%JU2c4CWdRBG#| zNN>yK7%0n37Rk41?+5{jm0M=fb0X%WqP4b->}~xO=@M>#?JG&YCM2~CJdQAZ;_tnz z`7YKbR~<3WxY|J;YJDgPpG=)syLz_Tyosimx~!s`aR=BVpoht|SuBS4%!k4Hwo;~V z%dDwKJM6LE!L!@{i20z7x6cinz&t(lXdA4O6cBtEdw{M}NqwoF5e zb9r#Ba&+$b_sHnB?aOq8nW<}J55jYHcI~ySOFn)veyzT_Yy@!lEYDiQuhD-tL_2`9 z{siK2%5sIQTxD&PZnY2Vd`yMfF26har@y%IF|Tg%FJ{28iA^X?+2b>2@(1Pf`?)OQ zA$KUBRYwuPz&}+gaTd5+G;5>X({7iTfrgZ8jn*joa;F^VkL7h>{Vr<)^e*g`U)faM zX}+z3KkcAvVx?YQR&A9AX7CX+_qR3^P}XqxJ6jCQG{NkB-Z?mem4y$3SZGwZG0Vi`{`oifQz{E}s-wD|NVd|erF9*=la zTs2V4`6>6n85LXasbXOWZULULbS1cwmeE1hu#eF{rlj_oP^-3_HqQ~F)C=4&Y)}WjrQgbb zf1JoLpdwI7L+UfBo)cxv`_HH`ciFauV2kht(S%Bg-?xj>1+w+lXa2gJ!(zH-2s1G1 z^RB1d_L1o~y>#AFGm)0ehbq*_@gt8tgrmKvVCbn&+0?OZYh%1lW5X49RHFLw=Q-<^ z4{Wm-I%R2UN*45PeWzrphHg)!&Zzt))TO3zcKXRl5)=E|1N2;0p&3YTM$8Qs|4qJG z_*!ls+HS0rzi6?aGqv(%Bi{N`qz`JtR%})UR2Qq15Q#h&8uU60HY|wUp#j@*iF|EL zQ9lNbHa}Z@TghayW&ZIWo^1w-_4}OjZ^?>OC^7$wEmMvqzCCkLD774l5xqvSuXOYWz#GOG*3X zy%hFAcK>B6B%R_p>07KXux6n@wjT=+x}&&nvbPvuv#H0%9%hZJG`^`!R)K`l@>|8| ziwFt-@vU)JwWFbXbsdy_ygOZ*l!1yFdWBnJ>?Wp=^yGnd1Y=;B$Z_4B3Un@hBs#7g zHzvV`YbLr}tns%i6tGG6VL7ON<0PQTtNK9C>oKEL=l zHJ@04uJ$_Axibl$0Rv$`(3Wk(O5^Kp{Ajtj;V+NgL z2Fd>M8PLn}`Llo&O&`Fw?Y4r}8#uu^+m`2v^jpAl=*G(XR|9isiYv_Y$g{ywxuoBK ztt|f2C3F#J#^GJ`U#CH?>k#$b%(xykQz>tJM>+rg5r{>27-Nwukv@fw8C^XCEB&?n zxV^Zq{s&GyI{hyf2J(qgp!uv>J2?O$vvWz?-~F2f3fmgKvHfE(J~8CLRC2ymOFR(v z9*mW~U06rA#UvJ<4cH!@6U@4AtxTM>9>53l2K|gQmO#&F^$>>R}1!Mi$|Y4;#c`9-vd`pb_KDB%Yk9AR;s|DU0J_5OIdP*ALiKqtP zdC2JiZTMgQla>gS9vq=SNMNN*S8HC<0<8%MQIR3?952(;MH@6RereskiqD`a4GJi<}@K|H)trnFEq)eI&m5=MgYN*R{^%ja)p zKqBcU_IgWaO$aw)J-^uM(BpkhheN4L+`Jt{hphUcpOHXyKijt_{-zGD8<8%~&No?e z{=D9*L+a@gEozPDMCCe6^?zVF{;M*mg|WT*VC7VkwT-Bq7D7bYnz;qg;$&F5_Oga? zhK4g!+MxhAeekJK-u4|O1>Q}NVY3UZpKkNcYCCbY0lahM!=O+tC8hvuNS&zn9$hJv_NTEd0-3cln zsc+yWr!{lbF;Qr1nTpd+Ie;bTb^)#;6bS|}pzafNJ! zJASyqMUcUARRI-y*YkNVK~(Tqe;n^1gQr;;00)UmYF`L2)E~FRVfZaZFCKnEW{7el zT&V8wf;sFqD!Z0(coN#cIy5+9lMPHWt0#- zlQ4WzuQkHhr=GBGfNKbkIC6tLZTjHXy=9wC=k0=chPz!V&6KewgZ^sb%mbvYLa@|r zg}RKT$!wM&lyAIOfa&39Pgt=K)J3+z{WGVZk$e4e$4h|EfUUDu0<~%Iuy)Ad#pkaI zdFxxB#1^=<5dyZvDMQ^49Gc+4tM9E7e95U-FVgnW^#7vM=fCUnQRn5M#X0y+CirUE zekb1}`{jikv={e|zk#m)yQ0V_gk-?$sF!uLvt@JVbm^TA$Sx^`R{>|x<(t)eB_9qM zIceMF&M^gSh0j{=h;{ z`g`T#>8uGWncAjcseU&)BZt-aoWL%e zl7@iLvrkjxX@ZI%>=YPqau~v6h9jT~z=aZu+3^GMC<}8KZDhUibyyhQ(ht#+Z!3%} z-uq6wo+n!_CdwNbMq5s7`-^`9c;Cfka!`mab#@&v314i~qo@u=*Lr8=_HYz_5# z9GFAON#6CMqwN#8`n_N3BCXX%j`WoaXdvTa6T3ssXd!w-QiZk$kz+ekkpw;G!sot< zaEa>MB=vBbU3Vd#I`GGZta;E`xM-gG_DP+^w%D=@u0uA+-W-<&-85)Pv0nEBTJ@HL zk0Yn_N&X~8k^txZiISk)YhEC^8X@lYE=OB8E&ew|{azzAk<+9@UL)HWA1`mic-El( z#v#53lVr0CFOw2u^N(p6OT|Y}m)Fa3GIpFarf1Qf1W)qZ)80tLyT!PIu0V4ZGlQKz z?vTlaVi5YqEUhcn>8VR;~yIR zYHbt2C=xv)gT6dxV}?lIU;d=WjsCQ?1an|V$O*sGpfO~axk}s$hx_3z*e#9K+6Or1*MO0m9Cadw*9iR^)eZl0JjwB>8+3dl8q+ z$15bajYrBFG!Hsg#C&^>^5j!_n!{Ydam}&aMMNweASO_bAsc^bqYRJ`D<`5#)?sPN zu&1Ma(_~CPC{ba<@982b+Hftry4AG8hnXNOv0p7vp(YRCE&g2EA*IGQ zwDeXiW6Nn+SqUm*;ZTQR7LjvsLeVtRvFx^W1-bp`LRBV%i3HH4?&oR0$B*5d*AQAW@GgL z>TsHUFAIsqHLjpG*B{FawJN*UfVzG#*_WXI0P+?p50(&kgn($Q;ct;=189(y!1I)Y zS=JgpOavaAYhJ2d-a=c^PfkBAGuMSguMO683=d8N)VWxCFxc+Q7_&W07QYct;pmv9 zEqv%atVfSOZ1(CSp~NJREm7nlRog7Gnom-yjWFsvLFxObG{QVHJ#zW|?gWmWp2!-a zUj#{XG=B?H3@Jd_lfd1jbs{+G-YJ9<~lj2SKj+ibxX`o&gg zWyf9S_qxpAedjvNIdPht?c;HKc8eE%W^zaDJv&rrhk#BQxSX&z`Z}+9X|o>1Q-?UO zx{c5eff#FO_n0f^Y}pK`c;8SRyO(}?E;LqFggdm*#6X`vGT>l2xS}R}GE_newoP3q z5-dSz`y|QcdbOmNzVyhnx~$nVT=^BME#e7H@O~nnx}{wWg&vvUO~a2GkB_s{B_gnS z`B&wFoQr#C(U}NA@f`%edn~gS4|o~o*~VQHWOQ6e_jqydtePosKFb({b}6XS-Di+| zu-6Q#t`qE3+X1&_W1K2^dvR;?D~OhpyM{>gFFmpIaYY~_P;ST9jU zoCI>@{ni#$CE1!+4tf*iNX+mE!$$|1;O((PB69cJo?3gHClSb!SVvRn9T-jgU?Je6 zjpy1>a~qOX2UHv5;pSmD;Q-bzA(dZxj+Z52M%R=C!w8HB2sYt@0 znYIyksRh@{$0Ao?%nQ=;iJG`MAJc|#=DI+c*cx?pIzDoI!z5nIehAdd3=Cl~#vZff zR!c2B?PVjqn%pz{-P3IDC2=(9LjLAyto9S>U}H^k$(*z|u6tSUhY%if&W+Has zp+89KJ?JhsyvbB3nT%>p6EBnUJ$1yW?FSgetD58noH!y?7^?Rt`Vm*&y!757lJN1| z#|moJII5?oroW4X_GY8ZY6?}a-Mvr?N(YZb^1XG;Qq7;4Mld6h5f9X?Q5TU9&rLSA zTcmB;&9mYeb@51DcUvo%i@nEu%%%}7LIXNVRH=GOz99DhS+gz@{?&K5-p{4!e~MyD ziGB^?8OM=aJ|IN1k*tml{!H={GeJ=cZD)F&g8m!!&gqiHu;96bHnJpd=f|8fw#rLV zmZ|r^07IktxlQ)B-d`tY7b<(y4}Hoc)>SWiYh{>Na*Zo$={E9h;q#5S9-Iyg92Igo z9H8_0P^z8yJvOvjg=_IgJs$sGRaB8(ZO}@hCt=dtbyrnq2p@#j8{JkeAzJ$`-q8pe z0pZIsgd{16Hv*z}RjC>)HXjyElNV&G6#jOkqK|YJO*6&lPvZ|(twYt2NvlMlb!N&8 zCH7?*VVY>1`!sbfcUw9E)a(~dmLS6emf4iM1BSTr${kgN|B}8h{l1qWr&lY+QM=#i zI>EkW{CVrC?n3F*B9b;q*l3F|3LeWd%N9r%EEYkN)nkS3_}b=0~` zVnDw&-s(K?FPTMpw1OF`>u+`@JP3U?c!(rgMNu36syV+AQFK3^1r8ip)sxd=S&9k@ zxn>r(1!I>e#@CC;IuP3>rd@Iqs`MiV2TCRN{a8EXwXyo*&Y^;xVZRi}0wGzJv~pzVDZ(Xo<|>J#J}t9# zpBw-k@TYrn#A&;W{iy@|_yMyr{{-(>Z)kTV52X6RRin7V+d=@oJt^i;hEZi9$|AAZCt;R>L z5rj#O=fbO8Ss!Rl1sJ9@*H~QsJB3dt|A5`!4mrMZm>5CUu|;}PF3ai@t;BLcsZIyqlRVBsUvoz z-&>)g*>TQo%Q^k5_J@u3=vE6XfzMDewmROWm3$Q(LtPg;LQElkpb7 zSf*5ii~fFPKFEFR)0?`*rt>HyJm}^4NTuo+p%~vw2$YYZ;f1X|Pj9NoDSkq4| z&>n~AiXNE5AK^>C{WxWu*urhGr1tfW(UN)E?P4|gJm)!sLmIK0b_7U)ah-Deu|-C9|2SC6q}muBkTU+~%CVRIb!o&te{S zP3w<3{(dQz^4NV|5-SokFB^Ar=pt$0)S}7gnBu{mEc+MsZiQY<6>N_eW4AVY@*uK5 z8uP@uv8-gH43X1d>R!2PU4WJ)tDo9!@w_TI7WxsC-P{FczOXodX-sj z^`c7`1+4sN%ZQ{xeG>#BV%ouVLPj}G9EKKIFOtg-Wil(hE&Ko*^HNpk;TJOvdHRDI zc5mvvJ0y$LhgSm6OXMdv*9#NmEw;Oq>~DwrQDmzZ2S$@rn@Tv=XWi^xe-`2aRHl>; z{%Z6X{G`fwMlX(H_m5bzR>BBF_f)G}(z!_Y>vD$_F!xNO+Y5W5#<9oyUkhBX_W2X% zzkPk}6*YqQ%_WR&u|EQRHQ=I|*y_+c#mC_%8;#xhkgXv8Q(OL5m3V!iUAjfV9?Rv& zdZMe>IAR>5K-KM(WEWeYOS+6b)1y7<=S>FPu9*{0INvQE8{r-8*!@*@F6kGD7WKQ# z$B^6yB!?;vN?`snS=xtG56IQn_|r=q8|+_%6(i&GO1NrB{2gjcoIeUE?vA@EdV_F~ zQ5L{V#{%+%iz_}Hv97X1l0}20rMn|$1F;6jL$GI2vz8iK*mP=@-xla)Pd{#!HdgYO z#E@zx-Z1(2JA(e<1Z_2I z_NlsS*fIL>>elv?Z)J9Qt>>~sY0)USU359#A?2?4ZVnsIB2;c=q2`rs^3(Wo3-9Wn zzp}adNY2143bMI3Spw-_2{v`gURYYZXi+l!UB2b*0Py_-|)5|92Ze|8F@*&>iQO>P4a(aO}INj!>jbAdc}R>!yiOq|>J|3Rr1j zP)I#4{Q^pi8_6fq|y69^__aQ*HQ-dDG`V?0Cc3>F2zYg`UZC4 zs2*FKyh_=bJ=WfVD*@%r2#}`K^7&?LP9Qp26EyeIr;t)LfAZp);m@TW!@bv+X4 za*3O({aA0=f5&fN&~EcEP{lc2Z?!as5wn}rO@yg0pxl*Ai$BsF>YBxmZcyy~F|zy2 z*BGBvemu!sMX(F8Ll$R!Z0S-Y$i5hfBpUY{q%udptJL|C3(flgZgVT-CY1h-k!=*^ zxz>?M`{+LnquM8mPN9;!1ipH zY4pVSZ4)jbo*^`W<7)25v7kDhuH+eVvcwtBI#zCM=?6dKFp3wK+>;0@JSnTqA2df9 z#jw&NapIDD;x}aIxcD997M2$0{0$M(59Vsi38&Jik$?TRD*ffdf!0I7tSZf@=r3uQ z!;ioRU!)8zR1C3sMa09xd#+?-q*i0AU+r)0p^%#Go8&xcqd>uSDz1^HnL zt#yY|@sp6csLifJr;E<|LBbMcxk-zM+6Z}~+itPpkUg1c(^Pu92nMG45)DEFF&wrW zEoTO^0$vCoxW}2{2+J;L;{oCpaVBWP_aDIe@iDB&zTI7o?PmUZ^Zed_G#MW{e$Sh>{)IYM(3q@bS1| zzj~-oj993C4AGEYPvtaqe$<@467ZxLK~EBu4yFf2{xCULyIDvt^RC``LJ#t-ZeJ!D z-l^iRD}?in{SRXj@)U;u-!`e^J3=oizX()yg=}Wg9B{x3IQv%Fv7q}EQ}m2v*V->WGr{DV=JkN zMreOVjZw6sNXneCm2o3temD;fcr7*in7%#e!R;v39K5!>RK+Z1Y#TkVnnT^#p{!O( zv+AvV5`~d*3RMsuuv^nqNPcS(Q07%EALz~GuVZ1;cH8!INT#(a!{C8X!tv?ZcJS9a z|KyI=-vt~w^`8otGT0ZXb!6rj`^gvQ?8ukIG&E+%cnDj2FvPZ^_4?|A=&pYr=$M+L z`h1p}l+cpxD5` z{Awh1_aTgM{N`R3YvB8xu&J==q)dwLU0Py>Y;EW&j+wVFc)6gWJT6{*HR{f$puL+# zkEznAqw$H&5<7cteOlYB1-JH{QGI%x^sM`j-quEWE|W+Kwr9|n zud2?ILbP1^ODp0f@XvZ>)+&`}IVH%NQX?gp2VhG%Mk04J;A{BZ7l6+?W_i$i-IiJA$)0sQLB^lsC)$6Vt&_xD-7d=lJV=CWz8o>Lv`$Q^zIm=J@j~`*o5? z;78_scQg^PCiL+MsNKQP?1?!I5 z7qC>$*cSw9K0qqzENb6rOKX(g0SKxQzW;_%tI$^>(GFTL+(Ey=SBCnSiO4Fj?mYCJ zwB-*5gja{UkOsuL#i$UMN0&SA-Sf}j-#SvpQ%KwRyzy<uMT)z-6P%Vpu;Rg?c%eXpySo$IrNtAx1&Tv) zcWLq7oICf-+`r&=&&>0@+WS>jCX>CtYkf3)f_(}}DCuYgSBTN$tk;z$3y+g`ss%lC)5}z8sYFt<5#6^5;bv?l4npkzn?QEli`ikuT01rMJNtFw{P6+wW-isVT^mC?YEF56N+)A(){dp7g>@tIF{9zTkup99fChYPCgKIerba(#?t0`iTIzz=*dQtKc8K82+|4?QBriRIG zJy8cw83V|IhDCIZarSoHgmo){7$g%ftdA79%UDQz@dv;27R|-TUjU zX`CdOFoH!Sgqbt08ew%>Bzyc`_2ynIT0qL<7rc{&61tjVNhUi5sN=N<19 zq2`?f#CGX(SXNa`I?5Q?+CZSfs}d6Dj}D%NBCFM#CPVVJX~V1Ww~;+w&)z@3Le+=C z@m;vCzN8JyY>%{4YcA*QX6`R=aUJcdJ-5z3q4UG*{Ya~Qy z%e=qb*A+9)(5?KPWxc`~3Gct`3xQ!=VLa^AV7}D3C6=~rY8Q{7^~g@d76-yMR}-nq znmUZ8mT?v}_Uqgj5*DkwH->a6%y3DxU%#T9)yr9fy}W<7!kbt9vRqIBcC2*k$B}{cihnXHT9J=tRzAM>m zaQfGdigI{}-To#=kA*r$E+G6Ap<_~%sXo4P=~$U&WtBpcQS~D4cO&U=3dKrgbM_ zUk!&YyP9mlp#o}{RgC>7iRb@b>;F9g{lB~Q3?g8(_XhKYipSqWSXO9uQGd4TJ3kT} z$wvu0O#jv|HD4DH94}13MDbbDlTxf#yKxpAu634#gs1OSw?4h`b3S57AE{A7%Epoc zNU&RsHh;L8o3s;PgM`Sp!3J3p9o%Q?o%nz@2?aG!O~pmNWb->3s$AdQRrork7Htf~ zIf}tzq!>oJn4hZYcBh{;?^>0J?H3gy3oQ!dN4N}S;%Ceebp{?eA*dXZe{FkPN3ZzY zm(lLd&mLbRTi(gN z5#uP1;wI{qw<+;OeWSH`!Xd>|LSrN2_&6a?ywhuVD`(}vRKkF*7JUiuV`;EM1!t}P zNaFGFnNBunAdS{OQDO1)N2nOx9$@o|ra6rIc%`&3bvD@7X@d8uK9(@x4sEj39xg=R zdF1y=g#e0co<1u4EKTHa`5slsF7Y_yEf@t_NKsh<27J?1VYwjJc~dF~Cu<<(n!;&~ zT@e2V;J*b2wNRCYz-2>m^%8yn-8A-ymE=EU+F4J$RwNkV?Th`bS|QG7V*gE-daSm2 z;Y*Dl;qv*7F5=Vqq$?Wdb~@c2o-OT2_3BuoX+GQKF!u?e&xC?TLZM@UJ!YV+x#l=; zpng3FEm%|Pbxm5l`N1K9-IJNR+t4(lA1ap~aue%c@9RjSaW>=?X{8AOcuL%2UzXbW zWi+eUbSj!7>{(99g@L)nTU&WD2ATCFUzhjHtrNY4RO<&}!QF@(JmuHmp5%J27bApa z?L7}w3{I}&9*Lq2OO2nYItcmDt|$jT%1Df*GkH@$ILnG&D7jcolCuw|PTNg_<8&Ij znsUHjW6n8pVmh&d1G59kpTnExOgy`w5x+@d5(RPB^0qHAO!W2*7Q-t^zUif10=CEk zZxEUvl}Q}%Atd~~n1lg*C<#UG%(f2R+tpoxpfCI;DzTz&nNxxx z6j7vD-0n>&B6>IF4+pEYl>XnB>#?0FVn;)HB)fPyDL@)pQT&t+f#ksC4`A!dryndeD?okDuzrtYItkI zwsrg7mM(0_cP&;^Y2By;ew3!CSQSfXWCgpx-j6QxciUc!d=@sbi2lXG`4W)y zN~#28cs`%=Lmfv~Mead$#Z7;a-faK%#R`|shcdfdDL9s|{#1qJC_yZgO((fEy|y0N;rnbvCf2b zwG>O0t-D7BVRAK28dEPmQOl45vQIG~soFl*G|kdjZ5)j`?LDW_9dOpKcArFRk%+LX zaf1lMSizvIaYK;M*jNeK*&FU zD#FImQ@|{ZWf7WsT4SL240jm#i`(vPN}&O-Qw!FA*fosye_;Oq0q)`deM;qGB11cr zd=~!UEIlKdGtn$3TJq0ISKa{1Mt5Wx|JETaUQvZQ#&@z`(Kz#fav_w#8HSQ5{U93Z z##E#SLA%eN0-JjyV@lj-`c%*A%Lsz`;x#qdr!H=Ufm@fxsn7@u;c`?yLfP+r;L{Cd1z2l>2ZwaE z_@6TIA{g6{F}{7Z)P6zi`$Z79xlSRc6J`zmSA+&k>{I2#_QlRY^!wTx)Hl!FKrqW? zU>J;e!7$f9DekG8Yt2**y}ClYRF^U%a&?7Af`3{vLjcvUZYWb${NBzO^r0RsOd&i; z-L^*L_jBb4Wv6}}Fq2>^9FwEb4>0wn_X~dNh6`QSEJ^?+Xb@#Cj$9d(&+E zsg1GI?KOjBGr)!>=Fjq{z?(>(|J5eBw@Cocsf#miNOddF+HBu{YO_((c zWVK!_Kf{k#dY59{97VVx-Ftj!!gFJvdHo?JJyf)Or`eMh+jMaimFQ%};EGTC!GyK6 z24-zAF}6jeZ@JV}$KE~u5Zn3CKnBh_ly9DS$$1_Y{1JcmeS_#`wn0J&?7)LB^G@70 zZz6uYo3ERw8*Iu}wpdy0zHiPPv%blg|KTPR>K09ACGzSmwLonF*h1u*cKRHyQ>fglaW8L`s+ zevaqMBxsYMKY-|?CP;`Jh&QCWm7j3M{YffDWK`B3vC7%kn<|3$FjW~(WVQyTx{qsY zXb|r0LyPNh8w#S`)%1tATDa|Sgc?ea?NU5oIWSyEJD(g}Nv}PE? zoNv6*w@r3`ozkEd@oMcBRBK@0gU0Kq*Nq7Th-JheKs-TF3x~KK7 zgw0!Kl7eQq$m~ZM8=VypG#MN1I}{+A=ATrNH5>XFKtZq2FQH|JtM$$JKJXer1(9U$ z<#0-^0_^WUGTSmV$WeNnLiGP|1J-}9_y4#F>%Uni)KxSN^T2Jtqc4wJDc4bfX1WJ- z=Uof!2!ZKhX-wfrOJng)=LM=fRP9oW?pL3F$l##_()batehY1wd3~tvu-#5dr(7gm z6^tH_dq;vxGgJ>Jv??PK6C)26gmPjNY*L8a?cU8kV2r(!ku(KJOCYPMI&E@d2BuxS zlEX!IDK;g837$NgA$v@2uc<3T9e1nT{sHD{MD!$Zz^L3dk{1h`vFz1e6La4_=JYAC zP;SD`uPWYwq^XCNZ9!KQyONKU<6XUukW~4LCxd?PMF3)RxVoLPDwD3eg~U=?X}tIE z)e~5CTnt)b zpR+5EeJG1kk9rz)Mg_v8lm>1rBh5=PL~s{OGk=X7%8^s?GwMz`Gs6V?gu;d-gsWek zE!B^-iVXU#z%$LOb;Y*I@qC}?5m%)LqA%=vV$c(^vkg8LDajTW`*NB%t>y@&pYyXG z7-mQBe-f*wlo^tyK76J0?R$M!xXU(oeNCTNNS6m@ z?oP*O(PI%sQ*%P{@Wj=o9dI@~!EQTwCnv+13lEo$m%_>-&{=w;q4X>u`7hx(-qrq8 zhl5(%)C;BraJ)s5j$HYmV5z0LueGp?p6>oY`b?hO3!hiYiEpmei~j*ye5#`!C~c|8 zjbtzw1@`1?rz$6iL;n0J-Y*!co^Y2v`<2XT4&QL38`{RIHnZcweg837hy*~5jcTc# zZ%(BJz1T+%$sESAO>0n4u69e(>5;$uO+aWVB$wt+|EZJ0*@Hki7Ct+?8`4t_93 zZL8#P*zW-WDM}vDDCox`>ruheu39F0slcMTwcsG&lnnZ_WkmqhmmB?#%-8;8(SFYH zlYEUBy$4FFCmvN4>)-(^x5GWD;Yx#-1@Qm?O9gyX`1xQvz=Akf%iUMi;`=$3wz8p-e9${2)XA>z#bzk5Ul~nzr44fM2(B5(3Hb zBXb`xq*9F^7kno-a33oA6v>H$)Z0VX9;@r^rxyB3S4C-p&VQOK#48`Cyjahob3-0E zB%WGb*Y_XUY-?8umwtXFUl zaZ6-XtR_;!Ge37{Vo(K~%)zZ>u&hVZPd;)mCuATReo(Mx!}lsz@c03VB8ByI+7q1= z7nRj0bxdt3*)=tT2=b7>oZgRiH^<5^+VBDP0+d+JL~0_d7!x&*5O46r=nV%{0oDh_ zA$fBG)KjcoiOP}s&+CY%r^)@+XFZd6j{_zs?!0%eD=n8PsPSINNbwvgV9qz&-e-wL z3SKZ?qKQK#_4FBIm|Pm!7*qny-Z0XU-qJ|D&Kpl6;ee$?)=hrHeGsmkEr2u|led`3 zF6atRmTK+l$46hB7%?KZUK+f6^+)CfuPg#hU~&}*r#ayP}w`n!o^N3EL^jips%yl;j!l2{510A3*y)Be%D@BPK-53 zV^LwS7JE)IXOfa-$vE#}3>;wqMpN8!SD&uKJ>x{`J8RE9Yf^W$no;JM!SV{7LjE2}b1qo!k1Ky#Ifk5&M7ZbG6-~|7d;b=rG^Y zU77&dT{CGRL`XLG|;xJH}31uIFB==wMNvLFXsOr*GX=6Gl0C2c!XVu)Fi@ zCz~95F{N}X^_RM4*#UomXjdzD`e|-z1n3;D{M*k~a#ipEPl^qRg0RRVc#;3t&r5yz z7!qjMcWdBba+o4-4vaF3_Ur}r=H$M%-HvX;*Kj;?AIWZ4tsV9s8`ni%1DYivhoStB zZnp;Na~>9lsg9|7d_lEojvsIC%jmY^R7a+pwm%Q4Do!=|OIwIILXr-bE@~n!_&M16 zqcY7DaW|=b|A-lsHaE2j;d}YU6iX^|MC_@cDzfH|G6D)_TZVyZgv)Q-g0TWz*)BqW zaAK_#CCur~$gU{YwJ;d_CIzjQU%-IaMbRnGE_So~cm7gl@jv<2$uZwTvxqWYl2u-a z&SuFc6sTJVpYDT;TURx4c4MF8W-9z;-~`=f5h#^crDk^S?~MsR^U7KImYbrm4$;mJ znP7)RnSZ9oOxGQZ6NX#q3`|oF{_$<(gT->%pmPmEhpQE79W+h$aMzZfEz16fl%WDMrvd1U*!v}j6Q zFRM(w5CxC)xc4d671TKr@uj1bf!e%5`0f+QoNfmNG* z0er^DY;BhBJSJ&VK&;1{oa4PgnxBIb3?ZJE+ZS7jLh7HzWDKt`B%V7xZ|14;-_xzD zkLC~)?&r;Jr<9zh$yd-J+?u7<{*F&Csq&~}Ld z#UNP@G!m$&SXO$LxvWBw(A5_gHfoERH*g{q13Az}IPZYcySqxJxm{d)az+w0P&ML1Ijrl-zg^rt|Qz+R|42V|8R5;54XmyvdMgCdO_aa`)5MTqvFWX z+EfQx3<@8E`Si(N(k&hg?8r>0q z=OL>vjR8woP@*80fnRny#jU>DsRXLR;?68d8uh5`w$-;0f|xIXqz{#Ls64|KjAu@r z6j1n7(;J>T1liAuC73eN>YDq_^@D??5uuqHzDB0y_a0Pv<*UvH3H9}>Js>Ko5p(VUdU#q{zO>#mk zSwaW{b&G@b{S*0b42xax_NmZ11nD*t8l|2J%r0Zgtl7+86LD~m=y*rT2eV48_Eg4| zk4#t@g~q=hrKvQP2+$H>SDvcPBe5XARi@^1&+BnlBO?xRoUmfSGqR;ObjIQkdoO2N zd}997ogo84PaJq|H3XmwEV?SI3MhjDxsS?O>gfIfNe<*>|Igk)qe?BYK_tT6VI?&hRZ_y43~hA|Y} z_U^e2)9!Y=7cPpt;kH^9A6za-wmK!$x_ERbo?cfB8>fpe8-3;vvl1xws?R!+qy3%W zZA&784U9KRxYLVHwbaDfzba9)0_~}53zOhLlEgg{G^|OsR-dmHU;xo69!{um~wri zwK7ti_VT)HA%r&bK340Z`u>(?f|gJF_4d<7(vaKlA86w;rOT9-_vQniA*6NCG3?fL z6IRL3mB{`4P9roDR(5^b%TwLb>PU2_TT;N~(!N>1;8VdLUWTE{$uaxoH^8K+$n8{> z&6Zd!`xPNq(fdbKCjBXXE|GiP{YT_l;chkz<`!MtdfwNIyWbf*&%&5F(3Q9Y3hHhV zcfbBqZ942MkIxqKRV}Az8Wb}773u7zCRd)Ot*9^8KZTmkmyF8XWCfc%(Q!c6&uRtrvP;ezA}R}7mAqqe+% zV{m}*xS>O7_lEE!sB*K^Rhjbdpplwqx7|alk#}MNqzB2EKWT;vacoK2`t1u%EDqSE z#0bl#b<>bxp!OPzEWTGiXh;Q%lg`HCnR!LTod-(wE`VS*MX>=dXNafEheXf8Be3=ygo*r=DX6Ncz^Pg&7=<)b zyPcYfnzzNICl9-4ztI*A62}{zk{2=%MA*gJQ5Ny$y8jNlx%x{!wZsO#OcX#%b5=Ts z81O5ec`K9`EZ{p)2(#o~+XzRwWOExzDHhkSy!AH~-`v^E^ocSP?P|OASvw10CvzU> zPJBH_yfhBt0LFEDKhwFSzZ&wT+dV5&BBo+Yg`SVTL@v0;^e@%8YN_Lyw4rpO-uaWJ@ux_I7h@q6X)hu;a~MY+@5_;}Pk1e|8LW{s*9)D3WeKHgNDYSaHC-dZfl?<`QsDQh^E0uDpmb9-=kH=ZZ%l15Y}Q z9ml1G;QeuP*a`TymzS^glhL>@B z@iG!;?@m8tN|Qb;LNEd$itI{dE3`jmo_jxsch6{wBTIA{A>tK>VaJJV(f7n;EwuF6 z#Pe5A&Ch?WWGxJnnj}bWMm!IeIM@F?W>2V1^eNWg;BUCcz_;%V`eP2GR@7qW%LZsv z#dt+5`FC>x`KC%G#whJJiWyA*KxVCGTQj9dRAQz>mHtlfEb2#D^q)D8Oi$!;EZMrI zFFA7KTS$A|7qcIRD(Ti+vZ|W|@L^r0>%X;_oh_+T2Wq}Wc3Uyg4RyVkt{*e&pZ}L- z>GOXpX8x1c{~_v}{}b1PW;239SsB`7Qe(oy2h*8 zv%;(j$iLj{!iyipY|D-f)^>aN2hdSB-E?|;`Aj<(;$!xg^2wJr#eTqi4OYOO@QmQ!fuNBo*<^Rcg3;8GQ1wq@eu+fsScYY)^OZvUw!(2nFPP zS!6zeET)ASPW}WK4(R@9KrB@--)~$7HD>23_cCz%RiBbSS|YDKB&43 zIe!@|{8va@1x)dPeS?Y_%WBz_gl7gfW$6=jz1Xxr36L5}Bmk9Ap%wMAYw{=LAO>rJ^68_p;t>eDB|AI;ikY1oV%L*RJM%P+5CtlVQwvk$XXsf>6^lc>x*RQ9R z8CMUOgZ}QShM3`Uv&8P_q*A_e#+ksT+`5*KV)smm4hU5%qAHdrmqu`|VPx-?G2ZI8 z&SEJ>wg+)3>bU28fBLqQ+xp7!N8kBSJ#J@Lud81RKpk;d;E95az@D#IELF7U}2JNj{g8s+j1IN z``qy=*{JMX9VBhedtRFHhM4BOJDJuF5#5p!=MXJz?5WD%`*kW#T+?~&6v{|o>*##o zd$_qEw`GE~*90MD980%J=JPX#jtH~b9aIzaYmxXBLA`0mji%N%7jE8RFw35#%dWD= zRt`5;eR~b9Z_<5t{mtbqxLSM)LcHZ-bL&r{JSehqoB{^vi&%3?U~5W|g2&WKPuyQF znwl;TJEB>)W^GBW5XzpT!r)`m}~JCmTJ!o&MjDxI2$DcYbgz>p+Stdv;MEz!o_EZ z*iX>ksq_K};`*A%%U&)G6sJaIh`EIv8y#eU=o3EeO^)FVtttHjn9_!SG^W^T zjZv+ONB7E=xT)}F9~s=Edyy)uFZ%!>(t2M8mvi4yNpEZmH8vL(_v}UelDR9CfrgIR z0Y}H6K1Go#CaOd0Ia9 zjvnIdGV?p8VB@G8^-p5;pCne0r5&_O&m);##_58}7U(eWxf7dpy7Ei$O(rcy?EO3h zO^Q9Gprw(To1f(?0)t~~2Yd6IK(wVvC~w?*`1_Zp48E%9=dzfNzROTKBDdctCP&NP zsOB2C=0o~hD3W>k3$`1z+k8^nJ*kPeL2tfnd7(u#h*ema03!fri1&0~RfhC6!h=V7 z?6)AfP8>!m@afChIClxC?_?dHw#`}@!R~{CvZ+B&$_1563S7-4GlYyF0IHR2ls{=)uQQRd zI+wp^5jRO>(Jj7g-aTx8Q1rmgoLG;UO^d+a858Z{J4%DLhA4#&G)*}Lexb%ayb`}R zC?8P|wI?w!{uPx%r+4X|Ize<%NkT7_(7=n+s|N53(?eW#)VO}7u;4ujnnacHl{NAz z_;zUYaTSnRN&cQK5A7sp^QdFPxp30 zgNA9m@&iz?4ETg!wV!plp?V6N?gSgtFKp0eGCnv}L_E(gSH5?pGc1BOvK-UNU09l; zTzuVK6NF4t58EZ>XLVM^U|xQGa8$?@ZCENZFpp?qb{sjp0H_(V_U4p1-pNTul zjsi!wlQ%=D;ErUg2{MjpFH+wxPdbPs$XVr#o|+chc)z3VG_EbmB;7&Kn6jm62MFdG zqEPex2hAS<)JJ{}oyTg;d1zP7jm5@3ryXgo7SL}ipSmy?HAXYBBLTb^ld8~`1m5X9bocfPJgVtNxw)vZMTf({>_XQX{1 z2<`BM9a7_foA$S)q-#qER2?oovBL=PXGQF)xgJZTL*&9OZn+X&UtwH2+8*l4-Omh- zBkwaty|vK*egy4O;UZ>Yvb^=FBsV&J7d#RZDE2mhNVRtep$|!45i;VFwzT__37V&+ zke#QL>_JUglSIu`AHwhLdP8J@9zq_`LQ6N}&*QNHN;^SU<=c|oSSMjBW&JwCw`Wck00nF=MdL)Sx?-ndfv+$H+r$OJm zx+mX|9kZNm1wgxddb_04V4(Y(R~Q5(+nk$i$ebFgbMbnIXTnpwNz044zMJ7IEUd=} z<`F5eSp`|%==}^3K_Q^A21b-VW(1joRavD!6^*TuM=E{v-g5w1!f?=<6q-y1ksL+1!@@>NiA(?&2!lPxC## zB24*&JZbyLkmB*<^Lo=otZ-cYp1ZrJltAH>S8jy8+3^&>IJj;*NqxfW=7x@8&T#Pe zxYnCl=n`^{lf;JJ)Gu*GQP@7tSFj3VPa?4 zJ~d^mYSJ{!&Y^*IB6mS3b7xii{nIok(f;aJ-nfkmKUX8IXu@AjvDhH`CFYzU53Hrg z&*Ox$a!JkW2NeDnA}BCzaYsXZyHpc}vI69w7ei~#y z3Hb-`B`5=71j}6O;y!^*ls{9+^02*16BQ$I?QZ-XD^q!wsRaGd(F8;6e2S&m&93>a z9u_irU6Es*V`GT#LLVis-t~ESOK2lV4{RkCh4SL4Xp~roo85id&knji)F^v@bQqsR zVztA#l29LVMfs>1dOCLl1YQq~cLr!)nOMy!S|r%3qB*~JZ)b-J_{QJdi2|lc@S$Ei6;>Zxv@w8cof6P~ zj)#LI$bt$3M zyvbk8u**p}DjSzsO{}LF9Y@l$vaNGtn89S`4{+uUuom}m*CjR0b)7Qe!j>P8M6;}5o_L*4^;G* zre%JORX%E(O8=+uh5uRD!vA&w5jOw4F&u~~{K%W(Em5#s6TBHQq_xM9o$r}XROWOo zW;$vV)5ti`E1TjhwE0UEK{^~7GJV2Z*I@|{(}Ht%=ibs<#4Bz7Yz~F^#+Xe*zeCD+;mO& z8WEg{$4W58udw*SDs$eQFBs2uq@-XdxpLC z{Rnb{F$%p4kP=nnaA{OMrI-XFrAw^AT%z6|pRlRam?rBq3%J1UC}OlfGz%6AdwK<@{6Ah&31K7NxEih;zez`NT-!;v(t5Z<&TGzcfkew4w%daQjVOPsyjD(1V7%*ZI zC&W3UcHq3K-h>{mM(wByDKLLiA zX^&tQPvUyqJG72VxQ}DuMpny^z|tIFx^}xnNPj|IYsG~7T_&=^Dsz*HPS7&O_;S!3&1kBwe6%=$h#FKxDQC9P0) zoK=2U81U;WK2~S4`qW}%SWGS||4cDx;^Ru1|kQkG}On+CrDIWXzYS1=} zjGp;kGtIwJ4Lucuu6{y#Qe(gy?r%R1vuo{mta@*w5N%(;{WH|BKKFJ;6a*szDL9sf)L@+??a3Q`(i=o6yd{KrIT)%Tx7po35mfZt|oP zu`BMOt^f(>?{yQs@NM_kvK;6dsw&Rz-yzOE(pa*w19Ng-!n5VI8-j^j zMd~qs%Jo5?iPCJart^l{7mst-eh+eme74kvM9XCt92iQeZbD*;Yz!#aT@~DBV{#@s zG!F(mbQCP@{y)6!_%C`Rf1_g()2CrmW|mGVB_=uXSpx2wb%c@sxmgOUn6hL04sgPl;D~$-At*Vvi?5arw+8-FC+*CV~|7akcwb$C@**4Imeze28B%Lzgh8 zyMW&l`Ozy|#%lF85DJb5#?qi1YhkMCoNLOuNyZh0m`xDr1a|9pa=b>_%et*XLch(K zCziKq8eVl@l|Rs!lt1IYbem+Ei<>Qt#!7v~c>v*Pj=KMRfjF!7)ePtOsGoQ|-l%xr zwYnQBy^DRTKdZ~9S&!68edV7tar6%mA*xS877I>Dc>%X4!n{w0KmZTYvTT z!M>LmiryRyRV@;VSQ7p-Ne!cQ%Ke|@T*SQ|eaS#99P#**d@GrtK6|@-5h~V?%u{w;R&Ek-j#|Y%Y2>oL4e%a z&g>1K_bDis zs2b>j7Tb!Fow4>z<;zVoaDIj(n93||6RXWgE6Dxo^yU|MhM;sgg3fp>i+rw-gsRmg z8F)uD$t>qDUV+)p<=B~IH;msFT$PI>GAWs2rQ-{#OVVk_MKdy9eBg1Yvg?wQ=OT*7$e~wVV%jkVn?yBihvQq9do+2x;$Xlb<+f#i3<5hba zVXI5rGyil_8C&kUy0|FUJOhZ zfLbzz+6~iu@WQmXI-Ha`np&o)VX^PeDW!q($WZKFGG~$fRO|4Rjp@R^I#Rh?%WRaG z@8`X=@^Tz1w$5s0$9ae`Ciod48$3eThkKfL{sabtGhVLB%`{~o02qfifhm?qiNofB zs8BPO_Zw?T{{RL0d_P@uXE43p%c+u)6wVR=v#?vMi*gX-gzt1xBb_IeGUX%14F4eG zM{mPzV2AU^7Wugj&OCQ$JpB!n7Nv`;)(}>SlL>U7H(qn`LhiUr-lWuEhPU zdYbIPK8aRxK8yrgCn0WF`w`)T2wa_DW1%zh6g@}aVY z#qBQVmV|*3c%+L!aY{Dh9B-vxjZiID59e|`~oK;i>c(|HdKUBez>0o_iB<+pkdMc zB&BrH`!TI-QbakXho~(iEMBkvzIY?M<&TN{n=13YxpVkrs+s%}36Jls@--s`V*z{< zGtzm}y7`Oz3)9K?rp0-!jmwaZ`y4ady%fW?@nEN=pS4wLTXu-#3uv~MDm zSk>}^rFZzk8(&T9lXQHR89Giry+mrST*w?Ao@!m3JoR%|p=f=x)$Wdr*fSyS&57Qh zk<^_w;Z1EbTgJ__L$=WFYCW4Oi`4Zgd1>q4O>WH{ z1P~a_48Y_cz{Y5hyJo+_m>p97*2rg7L}b*sT*M7TxEd0nblmOZ9v zD{Y#C2VMD)MB{96;zWRfDO}QpXdxi+EfM*6QvpJ>+2ea%WS++x7ZFcSVG4`_BnHU4M$Dk=e=+x# zL2ZTMzHcZLC|ZgXO3)xhi@QUE6?bcKin|vt+7L)^cX#&|4-Um8IDz66EmC^fXU?9v zv*&!e=bn4_hs?XOKCa2T=FRgwzyFW8m4zFIy@U7rIBb0_x3e|1&3p$e^xk?Hy=A7# zM%0u#u(;6;WeBYQGL|su(hmt1-moC$8#pCQtM!eY#G(=}FL#E8rci(&)9pegOAG|t zP1Z%U^#XM77=J8nC-^f@K(J(dXy?7U9Qg#58Gedc%V9(wsER!D+p9h^5G&pLoYvhV zKC~0S=65}O%9qx7KacNAy_BDnz^TnLLTJ1bsN=+`!Nh}tb!%uUL#tKD43M2OJG%;QA%s{L=v7=9s1DU%u z`s2IH3~p%c^zN;HPh~H6>vXra9^W?wgrw1@f1cXOJ^u3&?IGZhWoCAspIwPA#^1vV z{rIKC3uf31=)!qA!`+!AVX1vkx$?X2>zt(P>>!Y~rcvq%VQL)Nt;FmT9AIBdo1>Da za6>ANTI2k(hs?HPqmLNH#CU4G((RASbHoatd@x+#joJZY%ie}%zEm)~0;JDDxMOye zd&^5pkHcn7L+)aFXJ{>BG$;}E8x*fWV)57yw-thK!sMMD%GH(zm0OHUK+bJ6(^=sx z#@W-pML)j|60*3E>Ogt3mc;G|W;g1hi5MqeQlMcmB+*`+LWADqJK<+#^2I14QU5@3xpxV|%9$Smc3T4Ze zU=?=I24q)y{CR)8)8LD}6vTJJJr(%5^-O4Qhrt%suEYu&^*@0BMd7_;ZT6O;z_8>9 z`ZGd{xQkQm-DF>e7@@_yVJxhZQqi`D_nL~YR?OA>DfAR(DGcC`2xXvrytAc#6;hvX}kC|FYW4e%=#~tj0LrwLp?l zNCubKwG3FwMOa(nhfPjh+@_BR;`6~PjXtDqvtwfVTj%6a7pAoA3qh!(hxg*tcJj*+ z!1kQfce~V+Ds#(fWi|%VZ<%oMB$Faqo(&PPcCq&`6BibA%ADe;G z{QA3BLs+wkhgpT`^L%%8!d|R~zx0*^v)B)vi;>2+Z>QxZD25L7oM#eZTAN|GX4UIH zZbAH2H$srFaW#a{au+|XXB8iTalk-QJyebi}@UJC))I;X3RALf(qe6mi+Y58Wp1y>kbn zF-9!lKj#GFq!jz`TXzENjIso__0kV|g9xQbhX@ix68m7$l zz#`LYcUCf8TdCne?iE^esUex3N+IXE14G}29EWTTU$9>y^;c8#A{B*?Y9#Y-E|cf$ zcp3Y>O+8SpfG?|vDm94`>Rff(_P^s9vI17Gh13$y{5(Ou?VU>)Hi~UO(=k_>#ELY9RuYCp>*D;98kd}wO+b%Ce*z~s zZV;}HZ^f}jBPCfx>%|YA&$NuS8;g*34J1z0-OCX_Nh=vr_T#0*rdI{B3b8#ZRz)=$ z!#6dq!-S24$d20(Shrf|M4ZL*UTBROSf*OI#sOUyytvt3?-X#MnqGj>lh@^i#Y*07xnb!A|Na+se_aRyzE{va^0Cp zmvin~ui}nkcLPj{X?50*W@R*AmLapS74}dS*b_Xlnbb|_Y?fhtp8P!d-;vzc`E>NC zlBI_^SLO+k3fDhjM#E)d&CQdZY#m@XV}KIg@9fB@5UupKr~DyMP1C2)BR)59A=44i za}_$+{53_3<^9Pw3HqOf=PTNf#yW+i3zBDqszy`<0Tn4A()q*Pwf0_NIkTNNJ3kSksV7|eKj$LU`FUSg9(-<#{bhILT^Lq!`@3J7lLY8m zr)uFy)hOs&nCMSln8Gxl%Up{3T6e{v5O{C z65v9n+Q;O_`ehtHJBsH~b?*N5WEMf0-<8mUv2FtHty7Ne7MQl=Sm<>09NP#8eD0oM z?dxzrYV3h=$xLF^GhtyfHMJ9JRj-vf9yBWL3De%Z^{q?N1^w{{x-W>$XYM*NC5Z7Z zDpSUf$qefFY&OI#NM@Pa=}wR4g`SWp$VF+1)xX5#zA}z;RoqeP(Bv38+suJ9d!t`c zwNF{Kw)=;D=WEv4*~pG$&Uqr>7E}*QE9VDLt*uQly|2*UQ#!&X861rxR{sPxF^>9v zccAB*$8Q6<1nv)0HXJ|H8GLt~S*p!sU8((hD(Qb--h7uh`FW~0+t_uR@a^_e+yM0d!=OL-irNvHlt{mRnHw3?Hzsru_Uxx_6fpzC4pV^bO2l~YXmdBQh)_345R)5BCmQ%JQ&itIP>tUGK&q!l%p?Nsoh+_5~=)f zeucGJxvv;8>awMS*b`iVP1-cVa*pKAqOPAT2ZK`ibyf+6a=||8q1!-W$aW2%J7H4& z*z#C?V`Xp|TZ5OHjQmxl=wVw?Yr;6Y@{sDWs#ULD^nh@Hj; z4f+=0M*_(zdCr~Bt1@Xu0OFMzzDRotHoN%?;+srM2(-Ad`siVjTmFojG9^#^V5C&E(p9$izH?*| zeGg5|Vf)8)BSY=uP{i&5LU#J}jjo#T)6mds2vx-cC}`ol#AH9-mge(ZwU{7?uZ-Mh zI}>}GUBaz($35iO2rNL?5u2ha5Y1HMxw+!Ysk`V~C(ngbn8}STj7Jhz?5;j)YvuH< zMy1$JH^vleG$aZ)v-)^1bE_nIRBfg>#kjR;?ij9+@^TRS{9EXYH9rVJ_7%Cmc58xq zhT->!5g&m{%|T036D3s0M#vL2tV_=TUlR?h5U47Xc4SABQHlURk63f>At!Y)R&=>g zl#FuRncj>3QtG4!^+7isL^>X;zETrPH9nk&@AJxi|DeYftoTbHt^-1CXKOag|GKlW zqj83DhJ+^6f zzbfKP(1#+XPD-vP0>UVe?*h%J%0#E#gdq#;~J6eIWBA?n5jgpCY41V#+;LPFi! zT`nTKt`%~$u&Um2bAJoOt>X2 zkfv+7Fo^4|WbxwLwd8vGcNp%G>@A@*$T$N(wKpq>r}m~^957eL(3>uep<#0P@<)<0AfQ5C zr7nB6a|X%Us4(`e#m9*kO87_jO!sHnh)K*-87d-uT=RN1GIXN>JCSN?BPa1MTN((n zi)YQpEPZJ=&CE7&?DtT_duxU(pyw5q*t~AjCTW`i=VjCy7{fW*kH#eTVnNKSvV5w! zj;}pu)$|Eo_CdJJIPD;xW)Xt0RK zPHQQRJt>C8SC{eK%5EUxOuDTxyYnAe*~h+_&RdjGfbIcbGG;mx3r<20d+vDvAvr%; zg)e?YIeaMoTl7D`ys{&=<(HgJ?YNX;*HaU4QTacBRUcy$q<)N}m}rv0TkTzT3_ylN0IML;WyeYg?qDEHa4w`@5y`NqEU@=WhE7smU4wVnM|XN#XBMDNdAV*T2o(|iO#Z*?3eVjN#oI1UhOUCCNh|} zUW3?_eovl4KhKRIhTIifhfKCB?;`-!(_~SXxh1dVHim1A+Y5`uLl(*xY|ypVe14Ow z7aF=#%jw%+@;E{9ZH|wcZscLd*=Rz;r^{6piX35;^M}B=`$@b}8Y-EQ?ifF1J+W1IG(*A5f1K@h#h6f%%E3yZVjnvSr;`M!=ETuY=gq5B+Qg2Nd z04T}Niml%sX57Q=5#G*Z9(o+u$`jeW^ta=}RwO=?7o zc_{LWt31&sK^z|nlsuKH>v2qtOWm@e0`OJw2j!5k84SX{BBLyew$DopQ4~gix@~l< zI}7T^Ias-In)9y1#_LUkx=^eS0j}C>p|!VWvV`#JN^|3+Nv~lVDE3x)^Q~iEFW`j^ zhBrFyQ@7y@Owsn4QQuBlqRmJO!<7)tcDmI>A#CB%qjhvNqWN6*o-Su;mP>z6`K^eD2=veo=UwLyQa(Fg%L|}f9k!v)H0KeU6k2I?UCF$piS6d%+X4M=|(&Owjn#jie%Uil-rtmO8YTvGEa2Qmr z$x|i?!pN4UVeo138>J&tpF(HTu*^?FE;dicF(d@~m}G5)p3CXbDeUZ(@>-1iL~PA3 zmmD3v#xRWoGn}zFS&W-+-Oi$^^LT7&q=nVB_bJ)UP+@cCfk}7G9^c;M2BU&yQFSJ` zfXYYQ-V!oYQa|&+cd`0CysTx}3Wu9Z2V%;l!>S7Qzt`-1;}t8udi8YJ4Mbq}qj<(;x9P>7*y7zow*%XRCmJKv$exY;Isn01jK`AVXK1<@EQm3_b^;fpZ=LhK? z%>+HGC~ySDW{!dGvanMbsZL$4 ztT!UfW$}uQ6Q<%$2)hQXY`4o0`g>MS_*T z!O;u4`I z3T^yeVCH)|UJuosoDW|WO0xh;7%qBtL-?ZUXiqHGt6g>SV*5srco~B~A&`xb={#n3 zKa1QG)-m=tx(n?(=1>K@mQ>lBPB+Zq9d8l{ zz*N@KLu_NK(B1&UKbb+Lwd##99eYsdTfF-E;Pg0 z(A|@#S^cGv9}GOPxqdrO)NCoDjZJH_NW*Y!_if2A<=|00hz&iI+%(3@E$hG5JpSMO{@=Fkq_qw9 z!YN=^1(2XsY>q#L?9Y*n%+vn^Am}D>3gAJk*FzqM{`P7|f9c;Y_sU6K7#f#k5^7?q zDp1{dRIWe&6DUq8$3YBk8G|f`BqK*ixo^ZVMK_>-?M{BU!8(%#yykz0v(VGj7uvg| z?ywOccigk>^$j>wOrg$(StadIT$LL^gk3bL*N8SG_HX6}fj#j@*d^+Xu`{< zs`Or{I6~tKl z{7Aw<)j^X*+y!|3te5+z$9kkZ0DuKmfNzS7kZG%Uq+e~Vsp&Z);V-vB zJuA!KHP>}uI?PHxF^%986VY~X;)K3EyO zTGd{}+Z`#5p{OB6l(f^I-S*T*EpJ!%0>w>xt7r?JqJEw|&>x((jU7|}`G~@1!=%e) z%3sTwf^5{kb}q8PCKcz+$eVqgA{G8*8bV*g{#O|;prqAcX38+~iy1NVj_v2fu^r#s z&4A0nQw(Pf0~rqT>}qY-Xz^=bPbLCeOP@Ez7ACI&Q{?Af8JN=w#h=g(Dm!2SNlDFF zJQj}~JGWQd>25{GQLx*T}1Xh6WnjWV(E2 z>VG+M&k$+&n=HqzOz0eAbk`+k0P!JIh+hn5r}xu%phmH9+zDRVV&fS%=G@iFbzFOM zUG{25ea<4~-fvPlh-m|RRHS#B>;4#a6M|tgpP$~l-Ou6c9JJY*D{%hq^w^lidVlZ) zdlBp6RBBMs!n2D1Z|r(xWj|Lqi+G*>DAA2T-TnGI=fN`LCHc~(6KFdrs$vC9Eo|nM zQi6_p(p!tODz|cYdUZ{2PkZ9BuwU20h2N+PY)Y-1l$1W9P^-&GqtT*8JHxJ`@pPy) z1Ji4$>(Is&8|kS{OB*zRbN)iNp|x%~ugF{v9ku5S60Xp+(^UGd8o5QwnQy76GNw#@ zhS{r3-Nif3IDXlS6$bu|)0~HsuV57nQfxs9TP)O2T$V8;pA)+Zq>;r1`7%UtdNJ)< zcT>h>1CWID!3;JXw<~gQz`530DpOD~xB`IJCU;0&+>Y&=+EIjV{n>HV0TOAJ&eAzv zlaq+wg~icR#_33`XuwH(wpz8f%=AgnYVdC)I-1ORMTa-_Y@EIQ=I7nC%PJjB`)~$? zz)xyn9ZEGZIIk2wd}6&<VZ9C^_t4g<&?krGkiJMCj{%ug6cCKDUx#&O5 z!f6)7i)8yAuqpSysrnJ&cv@A(k1v0m!uUZ$LE&60;Lad*X?(fIuy=qD5EwO!vkc8bh7nHftyoI~TzWOwE<9f*jpV z;m11+L6YB2GvJRx`rMP_9T07Lrnu)l`asacnT#$1+_12xspU0LzRo~8Gr#qIPeAyV6@ zlfc9wl95Nv3f?<{CfB3X5z?J+z95K;&xu_F!#wX>hr(c|jjO_$u0=L}O9>1=>LNfHQ_ir|)<(y!j3hi-bzWA`M|Y8Dz3P&w}Q*p@e8!oCjjqi&QYtM4ZryhHcu({ zGyS@|O9V<5u5{$pQUBUtuIinq&vvL(GzE9)giAs7NmKjzE!pq1z13i^b}|nThG|%~ z_+if3c*|1Hj~^=(iCBv3SinnGzpxyo1Kq7Pu403&uktr&&b&6QL)(&SBgKSicCU1; z9hP;f%ZPpVowP4A0(9_|6K`S}w{XVnNX%VX3-XQujob5X>~4b!N2n=S2W;cmtD-W3=8gwP5- z9N+&o*Uk;MhRdkUs%z@Yx3zhhz5ipv>|YbjFrwliS&B z$v(Cm8l7w9 z?T^lK0QCl*{{p`|VD^gi=My!-!^UTBZJ99!t}`yh#^**4DHAitQR_8orvU>V4tM|)widfpSL$C{aRbSqI@{&9ylg4Gzn z&}S3$feTUO-r{992b)~QQ-nA=ZIkz0-szcb42!K$RGrJG&^k}FvCKBqp@N@$Cu=Pa zl%n$GodFg;lMJo{Dmp&AtekTM=SWLSxHK@Pv#bYTN$6w;Q5tbmT~lSXZ{0-l`&C_i&n>ltt2HF)E_3Nb-PX-Qk(X7XqaC=t?H}H(l~L`> zo$tV0DMj#mpqMUxfu#91iSj5TpP12mvpoGm=eJJYST2-*xnb!sEbkyM#g@M(P`94d z)4|YS$}})(EUKtUhnOQp4rstwYc+@N+PYF)e~W#YBDOR=;+%}xh4Oc%P26CPqmS+g zw)Jm0nwZwvULW% z7$XU2m*ha11$q`ytkcq$^!9=bvRE+JrCA768y zmQ_>7T=7$Xt+62K0p{Ary8U^>CGd-Rar4YclW9tNWMt)bYh&xmo63CJKNCk`H^ELm zSLMUS{{X$Rpkd$51dL{H9PDP8*4XfE;t5FTZ~(wA&~gWZ9_mdu4z#tj($k(}qvHYL zSZgNzG=pi>83LWLOck4@Ab+EH>2_1XlV3(jO-lOg$ znaY zIctew10v{vHl60je%0Jj8;a1b%bvmbkvbjHveTow{ub*{w~6A2-^|^@DyV!3xnh=z z+;bj12ObX(ql-RExn~C^vA5JC9C<#s|LwfXRQT?k=0u9Z;d0w6-IEMgV;rC58fyo4 zl&sqIGF<45prMws`g}L8u+~YbX1$w#ss-RT#|D&&v2t#MIZ^U!w<3O)F&MI5&Q%ua zyK{UdvKp-4JV+?0V$R&UM7A0Hw49D!D-a)G^|>IGMVT|}kSL9)z0bx(7oLZjCQJVS zmW6j?*;E&dSz5=qRNX5?Vwul(fFaqvA@=t9O%qiDX_eBz--Bz0LK-g|;DR>U@-A=D zax5Co#=N~RYz^BP`=8vR>iI9VxzQrEI^yicX8FHSJ);8aUr|YOIu*z|+rN9LpBzu- z$PD(`&bB+gnrntRdY1=%KKQhsyjkT?EMw%Kv&D>}B|Q+Y7``&P@6xEro{QO9V5t0$*^rVaC$}Q>oAro(q}tt%7i2>Eaav zTxZ_^KW?OR@P`eP!>Yq7T;n?-cXLJ+x=p?C7w$dOiTw$piSWW;Oh1b*{=UMXPNdPF z*WT!B3C+K3CGra>q?Z>Rc5n>(ztO)SZ*1#hc?5bD=YHZB`+4pJ2KGoohV&oSy=Cg%kafCS{6R8t5UW8~r3?L48ez59R?xhj6y3xTJ?B=^HTg zB$qbn+gKzsE}l#B6cc_*y{V8{z9auIGo&e~j*gEIuYZBz^blZQWMskN{chB=v9a9( zTt%zC&!qg)52}bRd$kQqg?P>D9u-T74SH<^>Vt|!x5Errt_7{(289tD#h8nT z4{qe>kihS=_cm<}^ww>}70=!Xu;<3MaWe!9sn?PW=T|3VljFvock7w!2$oj&y{R-Y zlCI4x^!~xO4M^4gf+)ej3wz>yjc5AoPcjC7xQ}ovnm~&-%D#1`f6+{x=Rz@nyu^e4 z8dRI__U=xoq`Vxa^z36H9T#zn__fBXxmfXDD8^A*`hUvA{&y>l|1GPFTAQ&&X&@sreJZbO}$8!-7=;&SM22Jp5XKA2p-{IB*eos^IiG7%cueYm*oQ<_JexLRs*A9 zQ7EaK*A0-RG6;MB-R3_4p|qnzq4A0UQ`|UFWj!}Wo$wAGA!@$L&K#c}M8ehM9Nx3rJy49>SH2t=M@7bw7bn11oe&VRu7mP3u~h@$@d^< z*rO1>PnDM>#X zhp#znU|?q5-k5QKJ+4+jUgUFa(|U;0rQHal_9DW8c$Dbxv^>^w%^$a4JvCB+Cl3{P z{<{885yt+^iQ@8IOj&esTFs`u1|(sB=7#-C_)H4pK0@L%l2p;|5k-nQX^|KNtFh7) zNRqHokLxl+`lrSsWFhfJ+^6e>HxC@_6_U+zAK$!+Skl@Xc zHwH||+9zVH1S)zzKUbaRSZK8PYkkpMGkRgvSNi-W0@ctmv%Xf`7O~!YUGe6Z`y@$1 z!mh$!W0%yGqo6mJEI&02Gu<*AJd)jic|C-q)v<6KR!;c2? z5(StkC8L37WodF5(9>-+QCG-M2SvEr~p+F$3fH}jC5QgcDN8po;U?RdKx%XJ;*b} zNrVJoG$(|;4XHOsMJ0y%?B&?id2+->fm#I?467RBoA+@eOwnH!Nqu~@n*5wQamwxV_jy;Ul18zy|1*hnL^c$etV}?lRjza z|Bd`dGy937q>nddLP$91^NuCnBR?bRh^KMhOMQ4sM&@^smr{2zraH5y9GnB=;mH-{hHEQ%@f5S+rr#naB%Jh^ zi(Y%5H9DJ_Z5b5Intye&i8CM4@V)(=Exo<3_iAgWx5~NNcCu>p*qr>sW`bgX??#Vq59~f1A)^%bzF#-_+%kMd?KeIoeY*FX;1F73>G16&hAGs4 zEBXDbMW|7RV&k`Q+dQ3YE_5Z7uNh?HOpn2HK!nX`iDO1TM{(wpuDzRBifLhWphm-j zll-Pm^7t~}LgCrqN*H<=hI5@124KZ6ti|PlR|Q-7n{tM|t!Ff!@{^&x`BS4zeW=vq zyVYw^rS$T7WhEgMTnUC!^G(3lPc|1}g6c{H*fB5fU`+s)q7wvaW%}!Udyn~>h+(;-)x6_}tR~VHRsh?3K-?oJb|W(0 z)Fvkn9-@e!PXJ_xyD#${d53JQlBhxxT7p%yOf^;&9J{ZuYKy)otw4~T!a^2`1}AT# zB1c#eF6ckL{AKy@;!U!xMtnje+Xd`EX`6sl;(u;eIf5=+S8aEaW@BuZM%<@w>aumG`hguuIUJtn>FQs07lvzr#Zx5>% zL(A#gH@HzFZK~>RTtPl$N!(O19h2eennFf>+wM;Ge(+B_uN2En^~Rt^;9sFGB9ESN zB`c*?NlQ4{nG3S^KB?*z;dp>sw}J)eUX9_%SHBz1hK@QCTu*xn^`4ey8Pcc7c>K^H z4ZQ}B`DywR_zC8kD`?R5!RKp;SyQ;26KQ9?c73+|=*;iR4tLR!lY4T>(XV}? zHnpe7`4v%$Diyh@J@o|R=&!z-1Mxd?8m&QwO-}R0o07h?>S?S|TbioorJLO0-sZ>D3XvO)|UJ7IuRbqoJ2kVdWvT z_O-FUIlZ&?1P5J(NbE_+BQ3J+#&lhulmgD)V`7C+H4u$w^B(Qp9Qr$XzJx>#V)4e9 zLM~xpP_W_<0P+OyXc}y%xq->m$4ImUcX{4AmSd(g?wbf6+}LL=*0|+#KRvq((oKuP_y|ix|0> zk;sYrROHt`KSh80nrDO}ik%R&fZ-*P6zA=1YGrBWV!ccWJYfe52@L~Xc&BU2q{U!RcfX%E-Y0 zA=UO|xFo$qeC-ss{g2c2lI=OosxdvqXtQKc4=vFtA~46Wm94Juv#2jVal15A5l(6I z{fF|R#11D{Soma%q*4iEU5tjU(+v^Qq&+zp$4&G+vMyd6-f{!D z$c%_V-c$^~|CJ9FHfh97kad0g9K%1rhjjH}XT0;TN!~PR1R}vM@=DD|u?57;1Irm{ zW87kgj#lIvG^3PszbadAR*eG@P$^jul5Cj48Tk(|igg>)#-%(uR48v9NIwSD zDj4XYfDppXr$d#k}K;gDI=ZtU5N+Xe(0ieiO z7u}d-}spvN&$QhDac(K=zLxPvM&5zuX5t%9rg|GSMxs;s&GntX zo5%n^L5WI;Db&iFGG>aN>M-k_om^%H*N?_G;(q?5*uw`STm)*Dr3|YbsMUC%17593 zJQ%lQUaeo)lVu`yLgPl%P4yGxA0S6ohUJ$uox)rafFs8}ZvN>PeDewn!}NWzEO2#M z)j>%3(N2xH{MJ=?nPG>)2O~4#DTKl#JR--oWJ%?3Sou<8-)^!2Xw=s-SXc~?+68Od zgzvMGVMBJ66><7hCmmkB(#PvH(*sCI?4A@KzMFo}UmZd^vS$A# zEIFRtN?}8WC=$?oI(Nd1k(W$C)(Ft%W8k{;EMT zbeE8%l~c@Zj3>(voYAM!vaO?i;beKl2LUg_t$p>%Z3HaD$MMSqO!fXjUUWZ>QW@3$ zFa~~E_QB#uk%4`z>dEz%=xCLQ3AO$xh$mbgzZY#iEXM77*>#7%_8W`O-#=0$p~dP* z5wtOjyMMsRx3z9<$R6#lZpK;D&drs=dDi_L9fL^G)z#H4MQ% z-RlK~8Ox3&5@bonkCusN7IA?=BCkdc#M>vY8PjH$kqz>DrcYLgYM;%hNk2fyo4<6> zosYIW_HITF0o@ej2}z%1ymoA;lYlIxKVap`7Ddf znLloC_hJ|2pDBU#q$FfU0{(238|n8|3==#Fp|G#UJC;9a51_kpqKWNglM~2Wd@y{? zQRP%)9;TP^zRpC6F9%uFJbsy5KS#uq6FsHGd_q2>YDJ^g-Wiy1T5Si0|E{+29 zfN^dEdWI@qcn2_!cfJuQE1E36EnT;MEtm3lBbQ^ zU9NQV=H9Ui5Pyp+T8432MFiUHZ|sE`{GOosK}gf7VLw; zn2in$8~Flz6`@ooCZQBelBNc*Dduv2*dO#%DVnmX>srtB6c^Q88Mz_$oo)aNX>%6? zZm#!rD^2MQSa5mkNxb^LEcV=8mu4@^N6W~KzNGy!K6@AQ!-aIy*9hdLy3vYZA+F{Y zP$4pZphX@+9_(l7eJrbNhR>x!KngoY1EE*($c>cqrY^~g{vFqEOZSq2kJFtGt`=OB zkFq)I2?b>Y%VQm|8a0P0gvz3Z3`79FlWWJi4%P4LhYU%-K)vxUR!E&xWR|7m;PQ=aH(7^&}qSF7S(|4}-$#>46qs zzU7RrmAtWqI`|GloG`MwY?o@khZ^-z^^(HP$eD9B_NAH|bpl#wiKZMUPDk>fK&R)4 zNjs*y#Vz&z$sRFWihI6+fS#x1xxgf^`l)X)-hKkqni~&TCH8`FvfirTe?wmEa(PS^2RvUt-_p*f+OaW(#%hp}xPkHOAR< znN$UVjvA+$-PhwUk(I_4&L9+`nMI{Ltxx_HxG#hS!{+P`U6)=b`D6>fwU6LsHumX4 zwLQonl{*$kFlTOc5k6Pk!f5r%iFHWAI2-M%^4QIteZSpI7(mE%Rax``H-bZ@KP`izDpQ+j(mEM?pnmh8ndDnu+QUmf0&qiu5LM zGLW;-q|D;ra}&Nc50Xc^>jgG3Yj!z-=8X&j0<=M+7UQj^lJ5=hW$k+aR6)wyhlx=H zVx}e@a6ovpm^@^H6xdkrqecrVV7a4yuVGV{N%{|<({b8Z|7>wWK^^O&qAR-IQezDX zB#V37N0$lLqeA>Bw$~iF+vRQIwEFy|-o!W>P3Uk^$w9*amr2(k+4IAhAX~AxG)5r5 z!fIz8HJ^7>LhzzG;Rf$R*=m%uqpMhuo#Tj`A~tvToo7X}xn4F|CtQ+4Rp-+l#+^m! zV-h2@z-1pLFG`+C>hn;$Ep#}vq3b|n+n(5-j+78jxcD`?IOTh@(`6&Ohb&#-150`E5YofMR2akZ%)ewl8?{XSrp?*uD zaMVd>H5_t_=?11S89sGl^$t z@qRy)7c56Pbw+=Mu>i877=(_0x!=x?(YCjK#qy=*Ct}IS1gMP%6Pjb z;CS*vf43d_g8Qq_xlZ?8LR{&9R?F<01Hy(|JzU}CIO=nmphiWCo(C44F+qH1kijq$ zSFxUG3^h`KN5ns^MD@7l$o&g@{DVJK)0Kl(#)#JR9F;=VkcPL&JQY`?5=E9(!fab zk}Z!>C`8QeRB8Vzi5HIn>#Z_=I@bQ)hhkn_ zr}=*i=KKFMIIqS-NnJu;9%=TF34QZis*%(+63_Ps-O7VQT!0rf=JTfqUAS=;&Fhzh z#{NCoto4!UQa`i?c$A0UZVa|V_(qwcb=F(GoZLZx8$@XL#eWg6yVmgVAAf?=Jt1RD z((EUT0Lrk4DLlWM&TikMJjE9MuWe3~ZWbQ&og6~T?ZX!CKRFU&(#T+ZxooL7ht4g&` z%MDW`cvyM)aTaNgZ;NqMrKA9=jt?l7Iw>_MUyBtk%g@u4P{a?{|KZ_UpWVC#b=A|> zL+HIp#r}@e*;EVrkuH7lz;42U2=+20Gm2n}L+ja>ZLs_iHVMFP?pZ zt)~eJDPV&%r67Bp0&6iD&S0!2hwaJqv`K#(7Fltxvqvp_FELJTpW7j$$SdCQzo$EM z=X~VDQYWnBCW7^HBggn(7Hijzb+4nd?0EHOt+^#Wg1QL$9#vI@Z|l0maew*Qh&d{6 zvMM$h!u0J%or;dh{4qAUt=isVn;C+ra^9jUy1U(T8wFb!+j)OrMOMc8iVQKKrhxwv zEHgH2O-yfG)w6q$JFKO5V86-VLGs7hNjR+pCOjGW1fh3=H@SQ1dyVe;EkdQeR);_S zYg}64SR0?$SyKu$Jbkhxf86XtcIL`$q0nhFi1pHa5EcWQ6gS9isB#c&oiTu>C%->x zQ;Jh__U#}Sz@S(nB(#p~x)F|^eWcG-stwCN-)ZLQ8s;tJtO*h`-RWMor+oV0R|QCH z&MhOx7q{d-C2hG1XNh0G?NeF1&s${a7vvkHt2K07?t(HjndsRGe&Tl^Q{hQ}adUiD zCZ17rQrIxG=zhKZu`ihTcJl^Szz~MuKYq>XGr{(5~n!CFeEVJ~prqLFEx zs;vMWFg4=b>&(mXpOv}q2uSyCBdaS~On)1aM6%q7iLaRyFQSGL+P^xqyY#t{?@i-C zcOmS}aQk+=dtMHl*kpBT_U!C#+2chUD6t%Y_ws~3w$?OEK^-ukv@XF>xQN!~=lG_n znZ2YbwU-77D)eFx6xDUH6VKlff;Y8+-c{433L4d~0{Y-Fez7&C4NGa1Qfp5-7-xF~ zBR#wkeGGqrh3GLc`#a-?8utRLwb_OvzwNT!>z2$C`44b&Ja4O;HnA6Q3w`3)`jwO# zMnc~fPPqy0h7AetAMZ$K{Pi|&^6i@D*GuCfHI>j9%)`U~994j|)Yf3nyt%wV3u((E zMM`qk9XGD&lOLgUpOg>u7Ck)KzIeO#SY$Lw8Y?9ur!kl`4ZN$Y+wEwM-#4bakk40N zraGciGf^(a?-Sg*`K#$%)YaIWc8`tw@j~5ap>A7)a+IQN^lznLfl0XchYx-9uf6ll zozZz}OPKE@^^zml-TctdZ^~e$cd>4iRC6+&gsQ)-ejYd`exRoc;WSeOnX#ujh?j^8 zIf*t6lf$44I6!)_*!Z%LfFtaRA3R$cPeO>2YRUX)JR`s<4bs>%vb7X=jmnK^G~%KD9JQ?o4^F-gv!wh3&ucGGuA(}Mr}cJf^63_fZH zuBXQyT$WmrC31I+ekX+yRqa}AgpsC6Z%lPLL170}DPTo`Tylt~bGCnsy~2k1h&(+9 z#Y}Cz{<%=?{!ME}4uxy`nMN-ohEkAuh0wgtvkP4SxXiZt4Yl5s7c3TkuEeDdi7e31=eZe3fxpBSm^CJdrM?Gvys#8?yFQ; zm)zb@Q;(G{G&Ztn1}Zzm$!xGhi&nIvDG@OW-JLoJZ+vLal{!N-A~<()og5Q9VpO&y z*^*;sR_+BG)l(3xpkQeEMqVwx+^HcX6}LuQNo{a8pd4qoJit8Pya~FkOZ3TEs=^dB zzqR$xix0JEXnzREtf2KDpn5$%$CGl3`L$H^n^nAnp)VabW!D5IZ`hi4Q%LKPc@kikmr0;*i4wEBBG0KGT9g@l@1CRp&-vRfF|o%=WlNa;Wgr z{IoZSIOhtQK$1)@1#TBQ?Z&UwNl?d={>nZ0Xvc5C>dFN}+9+#r_)IU&)_meBh(oH+ z_1AswL_1loD}LE%6eXrp#&Y-J?yiXGEX^fj@BSU^tl5TSgz`q&=*!MLj%Btc3sgQX z-Fbnk_ki)X+N+~vu|BHUX**$xACpt$30@+7cMxZEWwS zL7O%O#aeL(%)-!Z9O?~Le9ga>_rpu8QG-u;Xxff41W2b=U^2rsk?c)cnq;nD zsOE`H?sLC@g*&;m6#{N{LDPI-!{)&x+O-Jg-*DHHX~AkYJFp}M6GPX}9+6_RrFxQ! zJ0l$>Kk0>($}m=*_c)|e7`F$tdU-<&43oQbyAj{UXB3lTnf> zS9kSbn@IWVY*UV%r$U z6S-VD1@HTcgTm%J(aP;u-;O@*wx-6w$E}xNrsv}a;&biKA_koeHpoXeIDhQh`o&(l zs`v<=@@SVeM@k1b@hVaNw$Z`CM(SOLh3Q`!v|?Pprmt8>LM{UCx!{`L)>5eOF%+tPMa4SfGJ)hs<9=fR>=&!a$v&PmByn#u)bJR5+ z#G%suaeaT#;;vITSp{3o)k)3VP0?9wDPjAW$p#xmb9HOC6c5+o$K)IO)ZWQcDbMG@ z@J?NW^Z$&@@PF|PdHJ*L#p5Tx;Fr+i{{Vj)mlBrhgZ^w@t07Y7PpGp00gA1DQ5U?4 z`Skzrk4fTV2w}<_Ipk!6_Xc9?r`B3ZQoq{+f&H&#hvTNF{XKpp1da6_Zv>}?%$D37tY)5$#2&-j?UMCI_xE;PEc8| zYIV6q0w?24HjjUi6`qtT{(pEUf=s(`Jnnn%D^{fYquHI=7uQvks(#oWBy!Ewn)r5Q zf7w0e5v{}E%J5|@`*Ze6Y@-@NElgV>Dx^J}XO>Y{K-9?*-Zn?E641 zsadE7>Jg*c$87IROvHV|%Ho{^IMVSX!NmxMx{z zegrl?{vIb?VSh(7?+LHfYiyZsp_IZ^RO$&m05S2b5B_X(CiX@~oNC_m+H%I2b89E@ zSM1Kn8xvbSq&Ekskw_zsM8MHh}#_xb?znKT(qL3^XR|?^O_}~p1cr7F;c7#&oWe$rmzvi?) z2)H_A^JxV6iWZQzr9PE=&&3QoNx_wS=I%b#ChI}=N$QG{+PJ?br&7G}w#b4J8YNr! z+SS3N5>q3O$-|ND)WOBrVK5DdE?MM~RvNCtu-ZaF%%a)5!toI!JQ=RxHa=T>mTC!d z`K7Y&G>P=%+$g8iag8aQnNc-|U;cm+K8_!CB7>=6#OM#*9JJ2on0d*Ji$*snDtML9SrC zM@8xkI^+q@19C*iCeNkWfMF7v?ovALO`|C+oMo%sH=j?NRw5gakbphO zIT!C9Hf-2cNjg!^Vy#cq zW16JgUn8Ubc&EeiK#{00)!~rOI~L>(_Ui#4TX0A2PeF%LdJrOJN+*0CORU^ zy*0pdsnh=|Ux(L4%*W(=iTEtylhKC_0?+S2v_Kl&X&*N*bnrko?D{YVc zJlKJ+d4YFn?yX}6Dpu6HfZnia-lb%X)=-izA@?#qNIG5vQaH@5#{Eq05A5jueJj5K zYU(_pZ&E+y z+HV*L72kapp#q?z7vMf0aX8t`(R6jT{AoVP&P-LFFdN9RI6xB1rq0#vK%mMn>VEwP z*ro()=W4o<2I+mlsV6yzq^^}|)mioX{d<+|uYB!y{;S^$KS8u3rtWh&{&R2SJ>YgY z9e*x&RYhC3rqMI_%iMb~=PBF&Sr;k`lN3?WxZuIU*dQD@s^Mxu8XJ&RyM`ef+3EOBI2 zG(ER932ZkhV!cPGkK#pI-jc13eQuEdK!-CW$$%iud<^ik&MA8=RuO9E&rrLU5ao&5 z+?r&b%Ag{rCbwE(k(4yaFja{=5$rhkEbHmwWIKLP3?=YtvUCENCe7tFFD&yUUwQS| zI~C`EuZJA)2MRy! zDAOoG(>fpx*z39UcaK6@XQt?AWIrqvAK>Tj{lYAh>7?yeBa045#8;EJ9VrhgzW$nw z&)Y?SyPLC(uy8V?u8t+8Wy((OQ3ocW21 zcGHc9A;<^%2P#1>YcD`P;~U8<6R9Ojf56-eN{UzYE2o^`{=p61?lZ0VLEarF3%XCY z0yxPSIS=Db`BZL46tpaRe-b(t_;jS`rM(aiUZ8K#*IypN!5Y~o$+&=(87Kz5}fP9?8D z5sXX&HhJ#)OX)Mkf_h`4BU3&nus{|%#c2Ai1{%;}nYI=$`#V^ie2dMao(PZosIza) z9DOpso1N2Ho-d~m$+5Csjk>|deWeD$K9>W%I=}2JJ`%qavsRr{Nt!r_D4e+BJ&Olc z^#Gt)6b^Z@yTVZaLSwtaEsr}c;JiSQ{8DvB58Zip0Agf({-;x4R%F8j`M~xZMURvH zCy^rg`Rl(hD6>fRuQsDzQvF_$Wva5`z`wqH7XY^6dT684LE;F|S=Gn!H|u%~~}=1)MnMSc-r_8vtVWwUUsb zqYat$G4Ki5w9|varq-A7vhzI0>K^Nn;qxX+;Q^92Ao?be>;$!q|n0tt7isvHvQFEE3BL(qYMq-fG z0^q*tBbWJ@%%|O-C9s8#x40qWSbR<|r%JWQ)v($HT0Ggq=t{iBS; ze*p0uEs*}Y%oXd9rQqYcUjS-n!TU3v62*&#k#V4J%vM3tz}Yt|F)=ab)HeDgLZ5&= z?y`q3Z5DWe{qgvF+H`ZCXg?;C*s&72_Tupird-Zt&K`arGr;rlM>YhQJeZmkp=%PMl`OEk}EIt%^^)x3ptSm?Z0Tkd8dYDBY?Ti7`;J0={!uG@FP-P$!b!X76OBvMAJNc_&%vR#59nFt@HuiE9IAhN z7^*W}cFnW&sM$IRvaeFjgyOh)k{k<>QvM_OtMR_%aysDKgQ|016hUs4Lh^zN%-w(= zN}5yB=-7@6$^7gM0ga4^zI-O@$M!9w8%etfOa{gCOD5z~UajRD=$(4O*ye}Wi~^1mC83A)sa?*RAt#9oJtS&9agjElYUA-kc;W z;kz_`nOnNzAwT1OeG?I}seJn>B0?7tB*CX%yv(g#?=z#;;_d9HV*T3Uq|_-CORIz+ zu6bShLj@5^uByBPA!)H>6O++DpQNAXsaHlQ8^_&qE6zrvARr|%3Q0Nf9q$_grc=*eK z%5V#}e-tn(DkTl>z+<^&EXv)oGAaCTg=md@yjpI@JCwwN(OK8!Te(3`(c%4r@L6|# z_9W{TcTsMk`w~JAgKVT!g9Ty5r+zUr$-NwxcAMUT=#G26NH0G0p@F&w9l= zDu4Eq`SH`-n0tkn3E2HpQ9wrTAnMJK#OF`<(*Z9IDQR*Bd;G_ZeeX2P`LBa6kwaC3 z-3|Pp<<$4hHv^S0==mktrxW83)kSX5soJ}u6 zrj>1LzBZ2g(>E9%JKVPy;}T)?v&-+f1@a8Tb=gf*^MnpsTITmTw=ZZ@z<^r{_Yvv{ zRr)4<*569^0%QSyKD=C4$pfj&q1s*z#>qvT6vo0lKY9*5K zKKk4b`<>l#U@4ZJ+_$AgN;)p53GFfjecRChJoOj7U}*C(w6}su$CNQ6G6zJRt}9M4 zEM^F@MmwD9KLDz)wW)zL^ik3<1#hCRMn8-}r{#(jA{@fW} z9fDz$c+q};i0U>)0T$V2x6SE>111wv)ts;ly#x=a7BZI+`laEryYQ{@bT8||21uVp>TSiSG1$b<&D<24T9zQ-2YPgGrM^2zZU&VdI;VX z@5l~kP9Q<8FJ{D%BOdn&WaHgL($93Rb6QjHet}EkPe#B|4$ec|Lxd*1yx&joY5Ve4 zTsW>1I>br6vt8!U@DU4nDz5d=h)ZJ2@CJ`o*iL)b*wvb5Cl!OoDBj0&=g`cxY(y)S`!GG9zfq-Beu32^4#heShlpgs=rB z00?Id{y~8%deCD(Aw0Au>+evpiK{Z}ZK>!3BL%?lwp>B|mtx9_4rThIE>55dn7pWJ z)DAvfL@xkdC%5}5)56iR&N6UcwYWIrThU7#=gRMQE;xLmwx%*lR^@8g7_Q^jl~%^a zJZ4NqM=Wy!MnWuAk&ccW#B@5w zA%Bt&GJD5OKo&BVs-TdK{B`R(|RhgBhX*0`B4&Hv&CD~i7Y;aK;dKNBo_nEL*FQGHJ zpSHFuYYZ{`&99Uhq-qeBgU{v ztJRxR=9c(Y%hGUc{#DDbpA5O=NF z4Wrc)JDz3xw)+cmUoa+6gjjDY1ReBZ_aLAn9Q&($_1G`gTj4u0cW!C%G+VsyYYIlZ zq@8y#Rro*Rc~dA*eTPcbzbEdO;m)5XWDpHZo7(Pi4Le5^O5C&U;M$By#I zIM>uzI%hyw?%oU<6(wYF;RxrYzU_!-ft_Qe6$9C}fXt9QpeFv-(P;ihRgIM0(|E!R$X@vywuF%>OXp7{pSpse!CDWMSq+DN}<5M;7b#{Sw{y4U*iyKoAyjzK-E1OYdpBo*yy)Cd* z+d%)_{JZr^tZ#NEiCHCst=`5~4L}p~trZ5tRy}ng7`P{a$qIL|kQE5IbZ*9f#scN5 za~Vn(m>g;9MSICL6llkSRSdlafnka)Mg~KIm-DN<$5uCFpL$u5dT|s%A?_P_q4jgV zBK;PZ`z_`T^bxz!rq`h?Z791rNizw4+pqLK`J?UWa&yj3dyz8w&bVFTCPz{sZtf!^ z({ST&(Vm~4g>1YQ%Vl_Ai68YOz19=*kqb!oyWGw_I$XFV>+?t~_CKrIucnC437x1a z($x^~LRfpZtTHMQaAX7hGo>w+Um0_NN+k%2HMz9jWyJ+jy$Gs#c)NAb*0>i*t))_u9~}P0c7jk|7z>e ze-nhDe&2fAYaw>&H;;GI6-RcJ0;F z#*p;ZtdYovHAG&8_zc!dI|9RdG(brk(?DW4UwA=e^u#&y&o&s05Z@SPLOwc6`;vUrCP9ZVgc@k-*&Y`YHM6 zD!Dug)@Z85*z;+cSMde*Qrnq{Ci-Yz5@4`qPF``X&PKhITZYXRXoMJF);w)6sktMf zxBpEa1_>!!Rukywr-gu$jZOCT;_tT=TM+W_NP3)*5C8@8rqF(Bz*F_SEu7Il5e*ho zTEyt37tvoe5_=^`$}Y?HihQJfN50~cMV$1|>P!3PN*%`3EI}VdE9!A{vQgG*JN`q; z?2|iW3FIOfBQWDtpD4R$b=9KDuE?Pc2P8etzv|`jX~D)ONKi-Y$i=>kYWc-IDZ3Cl zYnyv<7XcKW=G7{Gx_@#l{<{LYPB@W1Jr(;??9W!>6%-(&ldFOO=0mZ8*(14NwywQ#IN# z*Fe*>LAghZ|E$K=WO>#E01c_Bh`5g^6{<#C-VXg{Lr9W08jwCjZtK1l!@16!jUFUd zpTcNsUMv>8%tSd(u1WGEx@3;wjN#uwk;Z^uW=8qdhpL_AZ3yBk4C>!R^<7*A+B;b)qAL7 zB8i~AYhDZuHe4vzSSU0ZW`FG*>@wZp1)Yf4Bl#4 z(rW7Vx>)l% z&+m>>jQLY@H8d)}g{3}#vIc-m+WR6hKaM+Si#1Z;>gu*EH0+*lF_UHH8qd<6QbW&M za>MkQQ+xmrAJh_b#1Sjc1EnnamZxvIfuNLxz-jJ|U67rv! zQ~2BNXHm$rw2jO6HQv1nqF?gTe;loKu49{LDfu3?dbJz3-jpZuL=7$Mn5CVk6+kRT ztbPa3(OS6%lD!7l+ zz_K9Qq!A<*S`9;Be28g$H|TL5x4b1birs0#EwPe-E4M~sE9YAHM&03q8^+>uq(JKx z??cX)nzq?^g4W)X0=mhd8x#>%;Hw6-mV6If7ms;%d$8EPE~}WX znp6DtC_iNUqcB0ZB^pPQW@N>&Ssq3R89M zLyk2x49NP64R|I;dQ0kl=w4p7tuv%d@-n+>&s2vBU%>)#-S-p_(xT!2eHtnjUKGPD z^FL#|{#XD1zs!5eFx#?~w!0<)d8j3oX&}U^E;eFwr=g7Qo5YQP+IzzthJ2(ng1IZY^4|-V z9olV5^bWn!Bmfw)`1&l)SphWHuxixWJR;nYP|NMGpdOWW$iy? zX33a>{z#U@n&x~|)y!M)TP66oNG_=fUOr~NGoMLTX-=xMFfD;NBmFXNNE4%$-niFQ z?=J8m+YzNq7MhZR3qesquXr=mCW`^>yUPS*#>HY1M~_R2zm9|L7gK{Gsbk-8;*}PB1B)Sw_okeC=}5&$xTgdsWNpUsiK9a)^@KE{zyDB|iKs zd?If9W0okjj#?wd+hOmXo2vT;G(>j|WcbQ7H>R{i(E9fwQT6Vh|DsrETGw}^KJHKG z8xS%(#73%N-lQEc$I;)tqmGIoOG1B$?6SEOR8oz{*J;)j*+BXO9yiR&c-m;+k5l zusri+`$5A$lb zAJV`_>14I7)xi9$S&gyYG9xEP%bUKl)v}JpC%(PX^`S{Ge-e;wB6`tXA<^byq#&|hc!Fex@R`lEx})7K3ZB@!_kWx zriF$~^7hwVES6Y3t0BCPY`Yr?*vdC`#3@%H!c>7`$EAgB_28YU3&Sy3C5xR`{0ECR zhNcnszG^bQ`v#}VTqfQTN00V><_j__DoEpx ziB2>*OxRn0txY=!0o|eSdIFeXpr{J^R%k2dPOL{Ep6UFQ(($%@j=k1II}V6B=qNvR zdM&^Nl`SN;_OY2Wg$S$Uml=P(cJ|b{QNdpqJVlyRL{W72qe1%&9!YZ5;*li8Wgo zTH*%FHg#X2O^W{kn0h=cu^_K#YsWHnsQu(WxFp3(YwfO%74(yGwa0)s0F3_t4*B*W zias>yuY_!>DiiKZtv4y3nQpiq1EuA0laKGKy*&%{*W^*)=7e;3kjIBNLV%lBUx(Fo z2h~8QibMYTE{BZ$1cLPBdK_7ycLN|WrPQ(Fa!Z`9a-h&0r)aiRlK@QAjY~36X8(2= z|MeIkvoi$3QKBp@(aVb{awmF>ZsF<$LG$Oc#VCnQD}=gD4v^_!LPS1*T`OiCa9LFyTsyIeE6&6^53I8~k); zE;Vs0G52ZBuV4%vwuENx8z6G1u-E;YT(*;8XQBsxG~FmD$*ZwFiq+bFM=Gm4^(W-Z z`yrM!Q-gDZm<`6%fc6|2q>Wx@qk~`VL$bnEA|;loj5)4@UMI7~nc)P9V-n~EQ<&4v zet(>o(1fC;2u5^pxaCf{?KTUIO16t#J11=EL=s5d&+=9M*h_THThFMDBw53NF5+dy zJ~seAhYN#Vvl!wTWhUR=NOX||iL;Bdomd&LIAmJ-$d zR8PO<{q6yi^;&8gq9PyEe(_ajlp@c}akR&~xj2~75tqM@j#R?CdhOvG2_ulp+U^s;g=G#>i1Yl7lQoJz!FX9AzS&g4Z+C;vZBm!6`IZ zK4>YE!hvQ?%uG){n>L!f#UB+UYJ>KsvTfkUMeUbm6th5K&no^`r-_re33aasl4tv& z>3`hJjsuXsEi2t4V`f!%Y#gxe?+)2dgB2pndyV#9l6Dxr-*LK7>0`luTlYiaEATHJ zVwb0+sB)w5$}?>jc4gS*V`kDYyHU;yuFlwtbSG;sjkmn$$g1SL;26^M|CsFuReKVA z*P)NV7RN}a6C08!ki<-Ez(Nwl?VG-Qx_ft6Oym^UFv8a^yv>0YhJkU{^ZMW3u|n%f3EyfBT3W+b z%hNpy`{Aanq%WWg<%@#!jA#&EEoA(tSTOw;}(= zs6ybJri|-*DLz^Ee%`S^HX&(Z>umOqzQIwE*+w+}3-#j;h4y1suIP$Ko*R)PC?PL+ z%(w1C1Jfi~3<<2}H2%1d6Fn=kYhUNjzRlDrH#()@Pr3mi&$}24pk@4nJiD?X^M$=r z&PVGFp<-^Nhz)MPe;yQbY`d92!0@w?Y2xs)lshr1te?<2+t5*5(y&-7hk zDiqlf=12ovC`peVN2@J(rS5yQ>f3lJ}Dcda6Xv+|oN=?NIW-~<0 zjcct`&rq&oJ4m$R|2%gIaFE;ELGoR zTD5!nd81Eo+HqD@($O}`Eas#BxlvzdVmzslpCyIXOrI6q*M^O;5FEm~3KE;jgImNH zz(8KKjpu!wREXvE1M|VME6m7z>>1;LwMLqTnnSb}ra+sgP-R)wSyLMBVv_r@hS2M8!~LaB1% zV~fxqn_xCb)~YraDpH!(@)#x*jXUr}MCWp39(c3GDed+RFaEDfhu|;*n!eL9z6?)x z=bPu60DR*NAypCP6g$oR;=jiz}E4AW+r<7jsHS#4OtUly3LjO;v40dVy_l!uEsB)4D2R}x2c>tRby$sG#; zNluc(hCo^+@P-Af7)8zeqMY8aP_7s!bKCz_75#sF{{MDGt?+3Gz(8!j2;Dhr9U?II zhkrLl&Hv%#RQ~fxZk*Qk;ZY68Vs5{dR>5uM?S9f^p9{l4cnrUnD4?j-^s;t0dS5|< zmomsP=`Yry+1x6s2M%gixH%8W^ggc1^-jxIxj9lwJQ;_oJmSD&%%wma+W@BWwTVdo zXhvgL1o;=N#WvOY)qHkAI_$(w@hilO7_Ko%)+HqgeiWo_KicG*UN=!Vucr@{6~ghy zH#26=p>5=JPibRsfi(gNwoNH1o~ zikFa17JBp6syq96`JJMiSif^sA19@h$N}LA z>1ob}lvR5fpWxr8BoLBOix`qW*4BA8JT$$IkvS!b~sNmv&qf z3=r2n*n`4*hwTS5&pfda@b!y;n~^d%RdIs|F%=<%xg0UObKGDqGCaC(-I8@4_J{NV za)*VE#m%N&4v4$Fi&nX>`^kNoTj5^I_2}n60(L{8)fR?0^)1mUqwW3iHCQs3o{l!& zFE@GUO8<5`UZ~ib!oyU#zVMpcQ7&%_c#<;aapkK3DsI<=#Q%^D*rsiDh{z|Bqoa;_ zA3;?ix3zj`YoA{KuzC~npy@e95`TyGVjgxsQZ^~${M4JhZhNb9JXy<)f{p(C>s)b@Usi7P3SR|I< zdZO@aOWl&dl|87$T_1Mj8{4egN8ENPcPJk4uwQ@rYqt&3-I6yeChlvdt$&(t2OH#m zE_hy<7JIBFM47iS{-8M@wBQA%Y#g`WChou8APMNgN!E!lmn&|Vby)`n%8+ts)&CqN z9BgPZtpw!4yeC!S+?F&Ar)^5b^GQA1IL#URNFLGr9)vFR$-cIwuR1NXOsJY$NHR1h z?u({=c12ziIyy1Lwp`NbH|+lCuoeX{^xK z$({d+rt6N)Xn2T(QB>Xu4ar+E3?O_d*}QWmlDmd5y|DpSJ&Ro&H+-)XU1ul0ugz`3 zR^f{eXFQ5*+4bq>7Jyr74w5$o3(ahb8l7|JXN8!-IP&+Ibp0raM17PX`5dy`v$wl!>$`%+3k!BZiR*+ zk^ceY(*9x6b5)uX2DGs;!gVpnKlutn&3~%qiSu>3#8g!3}7kG zOv!~*j^|PpeLFEbW6p{pBO0OWumTcKpZOUiPlIm zA|knWxSzRLeP}$(KJZ!8x5en)-Tt}omd94S*D@zI6r6I;lTp@bYiL-AXf1Uqw?~NL zH}vBub4IH0=D==32q9q}OZ87>6E?3?y|$@8zrEA6DR2-zwA_EQM#}H2Ndq~)b;X&x zGUz8(V4dZs+a6;Mi>E}?uh2GLF|LYsvMFph=xz-AD|Uh&(^0PeYD}e(4Na^H+x?4c z8RP;t`o^yDDSQ5RfJ-)6Gewjq613#(u3c{{9GI)fJ1=)pz7I7GG28O3?GlTyCxg zLa~XOLwKftb?b#`w6!&avit`iP~n(*mZy-sj}+`R=jq$e$K5%)Qpno>J*$*X8V#nU zH7=fy$3y`sVE_%kVXqktd%o0)x+wvYuw~wTLY19ApioP@_hah*oq6Zg28n)y0xP2r zl4pa3GRu`cF|Nj!?&zk?YNKjz6=iR`4AV8BQI_q3)U`me)Es2t#^$ z3_FiR2S_mA*|(FN7I+aS&?ocOtxaKjK)K@GF#qW~niti$>g4kP6?G)0w8rDHgCkTa zKoE770}zQ(K(EE4N5O_H7B^!C9ej__GAH&XN$m%BebR_hY%wi#TGd^GDJ!z7TcciB|0F#9ELZ+!@~} z(9WcfO_8WbKH3ty#{~l&RseIzy)cUF5acI>o$?TjjS%PF(Ig%QuMREYdNL@Lg{c zq695guS;^}7HXqPkRwVt+z-beP~A3pI00KXOLH{$|Fl;=wzcZ2ZXm_GF2e$cgY18; zN5nio>5}W&P&Q@BvpwGt1r)BG5Ar0Re^O3vmr)Ag0)i+G8m3(oJSX?&iF}#pW3)i= zf_FpW9zIUdxQWBaIS&X#&P*M6P@p2V{E8TM8i?P5V@FV}c<-#Dts-C}=9jQHo5@Ou z`Tdc<(@H`V;jGn|=5GAyv&2Gr22~B7ElQIOKh<`uKNotoHLX+HK2yLzR4_p%?WNu! z#AbvQ$J%a+c#lF`^8}%v{aptS$RlA_n!aj?t1-4|B0rvM&;Q2TTST?lhwr*Uic7KL z8rcchMlxg|97Ju2O6ro*UDQDPu0_)FD|Lu)mGd6k~Ko z{ULM5%YM;V(zbo2oUeS!+8Qete;74_ZZHdKJ-|nDQ@~EvpjK8K7e`3wwUK6S6;B

    A+XY+qG+Un&r^Q5 z^;}Qk9RMI~8{h6UsfCZUWO6YY|K5TY02q%SwzOTW{8PF&`-|xfPrNAWt&&OElyS`t z$!pSoMWxr~pyq0hF#23A5kVh7ASqnbSj`Z5G1Ep8t!Z|b1;6)JLCCzClT*hd&PVj0 zAVU`kMMUh+YT++<>T-yEn7Uwx$LM@6>wqFClQ}#c!QJ# z4ESL@g5Mh3xTwY&F23qbFh>=b(v4*7Y`Pyw8znoCR#v?he?y#=6}qB`8`&r6TvD@7>T&x5mA}Gk@g#Q zLB12=>LUFV&*h7m@I*geX)!5eDK4Fr$`obv6C>6sNAP-@ivpD?#Ay5UXLorXX5w`j zLd9>JK}@(UkQV-qroQ>L(P*E@kW{-Ez$gjeDru=O0sTV+fZcRmMWQD%C{G|Vf*=r# z!(bA%vAJ>VF>&R@kUhGf9>D&?PVyJj%M$>|j~^MXNK$`a0;@{WQS;Y7#Q*csa%R{h z9|@BC@1<1$P)v5=EE$iYi-A?;z^MidnAraoCjeprzGXnGI+|p<5Y#q%UmR^QR<&wM39LAeT!&rP6;;ijQJomOvrO7czgoVjP*9&JpuZ`oaHY(xw|A8D zZ`!mcM-b#({Df{GRlI0RK(*mz?dKky`~lz1t>%}RQa%nkDrSV$seabG`Y`^nPEgg@ z2h%E9;RxeQhj`7zJ%I*2GNKQbw9YvNfr%f{otdL`R__^PQBBLyn#d%mDs)JIK+sW9 z{0`3f;#~NeApUg~?j@eEIVDBr@cj;A(_BO>|B-V2kd%k>{D;fc#`-v2RT#5FKX>r3j(qv3zF&0H8Zz$X7Kh<$>fe*o1zODE{U!qC zq(sjcXH~j7r}_$m?V1K#KmKrTT0-Mt@Ntq96Xc>qyh`4LD!2s0L!>mlWsmolrBg1a zhJ(0EXkndIFVw44#-+N(5iSH2-yOmu8!IiF77BAT_-PB0xM7KI^*gG*DQcBgZRv;h zWYU;Fzglz%%U(<{@YX{;Mon+(O*P_1#*$(q{IS0`v>aF?2Nb&{K8WUcplb(;k8cC1 z#wlKdO#{L7E`I|}eqveyTfKG&fBwP#W#*aj1uvXK)>EAp)-D95G>IfgAyU$q2}yoN z+dr)89)WXaisMYvK~8_8sR0}ED?B-;;XH6nT&_`U7b|kj0?8r z*UtNu;BNri8{Qu~;Kb}P z*GKr7>;_6Osso;v9{B`H|I|SJpS1DM(^C3}1OR6{@E!-;u75OR8t|29D+8=PFn8f? zUG+(NZhv3&C$(SKX+(cP*3k9p;X>STm_@~1gH+iUy{Yd%s3Ck`UY?Bbb+2g@^GD88m- z9xdl@w}SWguXpNb8*ihh2k&h$|AGvLL+=75Cj#p`rGC@v?0p|L%1|84k$$zr4YhpT z>gK!N7%p@&ht5~emj2VryZmyuTgGkoqAu`Jx)OOWNf*n(qv9#}_t)78BNyBp5|k*g zfu6FgajNpSn&)CC(%sISD3p%KMtPvC!1gzu=L2TDe3H*8r?i#Ac+4yp$dS6$Ui>7} zn$%adp`J(8dfTRgDw{Jt(i_poQcrZ3Ry9cbz)f>@K?F-5G2c8G2h`@Cw7HXWL;Isx z*9Cf!Q}UEZJR0qO8n~a>@SELEUfxCG)qRuc%DE)2Q{oNcNVH)wH#_5B?TrVvAkd?Q z{t0il;XIW6cs9{sX3%I zDwE@T4nRz(`M49!Ph3wb`D5t*u|$#o#9A(3AONfWtUl4kor`3vV^a`)6mbSYNj$~@ zk+lJe9^ijr5MD2z=~X=kgS<}+CW5L{vdLa^kS*`#43_%>w;oDfQEar6e4^(j@YxKCYAC!@gNL-=q2w=~S-iAo^^f zJ&-+!P)GSItrfH_;6_x6IZ-gdXUMTm%)(sNDY+u8mDpFW-Fz}Q`NJsv4X67_mEbQm zGBWe>WNd@wsy$BYq2>$OwK~Zl<7hbnd}opIvhPaJ5pS-Qbez&SCHv*<53RBz@>2@V z?tjSnx()b^U$X4)wk~$}nLliFN-xC+Xc^8GP>2Xf;yGowndL_ieW3db!fma#;w7o4 zWH&vosOqsIzdak%wN!Ne-X?q7=j=!MnzA}HcqZ`4U{gu!238=Pc@6i(z z>S(A$SyvVHe0L||&;W6&d~p4pVKu%$cy?{}I7=0MHal1RTjD;`45SQTDxeEDgr78o zrT*}V=;bNx`5E86B6o7~=NppimwLjNS6^cy_6FqcR01x4&sjZ>k=p|n7Pqd%3*3b# z-Q(I>MkmolJn8V#MSb7^Tl~jl{v&n(n^i$4$7MTdoHvo0uzre+XlVYL@=}v0aaorA zSkiPq+O)sGn}X=wUr-o5Pu$`k-m_eljA?P+JsI4FTZV+a>pJ88$>irO8#c zi_IXSk?wP~(Qw-NyT|^V`_R82-kr(fg&!-?pU>b`HN%Zw4g6?dE}##73~~D77m#eW z#dQ&>qISfumkL=6A_H5g-uD1%tR{zqt``q6kU|`@+p#*6{(>;O)_!$2Sc*6yRe9`9 zGiGFk4FJeJ|rc!Z*>Nto8WgirT8>`zNAEOv==NFToa)#prZuX^q4GKRpbtwTZ)@IJ9L?p)Ir zlZ3b_`+pk(8Hq&>O&VaisVH3pIA|O|uq;7O??3rrIR9Kp8~5Z-3lbMWLnKUY+fT?f zrXEHX8T~K5g3C#U54QKOUa&-ho{2I333RQ2-^x4?ZU^E!a`M|N&ZcH8it>FjDVtqq-&Q7`2K7a5J*n6A6 zZ@@;6M4q+`%$1RMI&bfHPL)esANG-UpQ)&ts;Nj_orWT0HdDHt#YIiEI!Vs8FTZMM z$Kd<@ShNN8`C6M@n7o;qMs;05_MbS>B?Zes8pwl(`L}yxQXxtrnSSLiUg?-xjO_Xw z{Z>#I%5pS?0&nEv9)WCpJeKYwRupz32!9{GW=4C9QWEhH8cGt|-V-IGPLMJ~SoxYt zFEPnz=hdJmRhAYs_^b}=c+;`Q$v;p9A)f7F?V&&!RmI-Q{o0XXb0{KPv9^bZITIo% z3gUXDMpk)`Th_f=$(Ea3f@WzDm+bxPf{Gm9@wn15e+@*n*8o|3WTEr{(5tz}n$m(Z zglDQA1T zjXSl{t5SU*Fx=qfh#L`wz2*CbXGmIauTXEOlYn9mGfj>cCrI{aacP7Au1KhT%|QCg zsJZExM+IYV7cG_EkLK%}M!w}xwK})?+(rKVEN3Qc9NFmB=xrknq7Rsx;^*>mLc2#6VLxs+8r;&3^tE0-i)yyCzi-?Qiam>Qp(qu+@>zqkZoDOSXCnN>{bv8oKa^;O|ws(MBm(crijbQgY;KKD{oi6Kwa;q__HzhcvkaCJ-|;|U30&4s!33n`{r{m`%1 z5$&ytty4J+*Ui>TsQnV~SuNiJ zOoG{2P8qH!Q&DDcj{4OifZ|!2x{fC>y}=>-!HM7m(k=jd2>^8!Yl2s zjG9`g+&~ZR>`Hrcrmy82b9MRWxOQIZh<}ZvWs%t*n>N*s=Q+n>TYX6CM+qA}DjQGM z)Gv4Ye?cqiXIpL`!*SBK1bRK?_UZQkN}(!!rO!c*mg^;(WFTUOl-~_kzld!K zB;om@tXx$OHZ}^Dg03*tDOD9ld^Pu)9WT6N3krTLwa4nY#!jo(YgF^*L$@knSB@&U z-q>JmUbl72@gr_I8XhP6eu2_{FYu7obq44>{h1O~y19Cj2cDEJcJ0Lw%2>TBsF)4oj3Hqzq0q8rp()N)zH zvxx!>%N0UCUg^G%dLgbA>u@zD)Yf#DgpKL!V4FL&p z?39OJT;EH5;LRs^dzB`=J@D;RDHn{M6o{5(DgG-c?4+7?tIC9V zdAO=oPc!+;0v)Gh^rA<2e|AH>?4^yCY>0`5`{~B#RLwfRAK!=N{K7Abp$=kChHb~~ z2xg`I$@`$;?{s|UPc`{R7frJsJe{BO&@b#zc370$Hq1Zk;Ys=yy5*R`@Y1>vTV+(M`t0G%>MsZ;ZREBy$*e*8N#OS;%R@1weD^)ghKtMc?XYvd z%sYJZc#v^QQS=a%#3VUEqv?^GrgWoVscMsx|(;N-Y@JvAI+nSeECV-gYKzWhuPH0n@I_k_G)F2hF!!i2TH z{qx!P+JkUI2ysu|R|?BYdi7DwmXjQ^9V%%9@8+WMm?V>Y|6^BA7EX!k4?>u-(Uv(+ zk>)l_xIWxd_ON(J0&;o9)b;8wyTtl1`tIHvF@@#*EZhLC2%iV&~{zvgEg)v3$xcBj%SZp~F zoG~Bt-Av2oxTb#;LT1Ruz|UuzDMfjC-(rKPCt5n+lH>r>4ojvbF?gKC(WLpzyk#=I zo6*-4e^+FG*?+&$nSwbzmnYuU~3G3!lcu-59p8q#b^k)gG@C-3qlqtujg6 zW4RVC&#e89E9;!uz_)WWIEmD;D!3)KXzPV4qn-Xb8zv$C> z!k3`;sZ`RCs9Up;)1D3>1?2(M3Pkq0w6a2m07J#^-x&^Ri9ptg?6Z>T1EViv<-aW- zxm>8@**3)T7gSfXyC=7&m$n0mJ1oU2_F!$vVN>qeBpEWrt3%Vh?aIs*T za+sEU<1p#Q+%z*qr0Ax&q8&`AK5!>qwXr4B@?bdBLKO1;c2F9Pl)3p1*r|{FspiY- zl~@NThx=L(UJobm$sG9xE!kTc9j#XZs`gQ0oZHJ55&4f6H}NbK6($KG-gx*7d0|kK zG8J2#k+2?<<4jjf`5*Q$ov7o#phxz2dpv{F5v`xEzOyvGY9zWBKma{i)*G1I;RD;i z{{Kqzd)wC*E*BZ7+RajH&{%PswM-H>AU;EvXDg-y1z;C-fE`z7EO zDP=8D+4%{U%Nk#{?=1A3meBQ)g~ZvJDM!?=ftW~1xQ{Y4W$8KA8+?cv$jFc(>#oj@ znTT>&J1yxNamdBOx{9_HQVXBjqV??haUCu3L zCa`2UJ!HP0X%&Q4Xn6m8l7ek-c7EWNC*+NL4Z5jWUQHvb)=ak(l$B?k_hH~e@wn`? z4L;{+lx@!q6v z!(u#`@`87%Mqu1OhF?_2udDV|8k^k{wSp$zU4;4wFE=~Ckh!X;4FETdA2Pi4H5?_a z**ErGofo-YJZj{d^6{9)wt!yfP z0AP)xu0KpHB=xcU=N@!?pYkOR1FBWetHvI;9AKAEUHV7*)1>pIhJVu&BPqUh>rTDE z-3cZ3tsVx@u2Wfk)79cPdsW*1QHWxYX9xEyczd5|k}231>M2FPr{)17m?@9j2t8}! zAec3mHO70-om0=Wv8d?ELrFjQg0{Cs?9l!UYy|6DLyG%CT=R469Mjq|1Sh5FHL4P; zu#p-E1m)y-DreTqcV!fShpn*PN|;CTT^u_sS$BW>OMyW3uz)l@uqMcg z*~p_rP0aW0zt!)1;rY?9hVBz`%k*cvbF$E#B$3gQSir9~kp+ggx7AICk}%sxej;z@ znf=9YuWP2wPE1pEArvN}nC6h3cr|buCvB!dS>&d17uC2#d8ztBOW3!lqO4Z*AaX^j z-cDc{^Z3jI|L!(2&s#)>>YFIOaK5Qo^ptn)Fk?Aa6x$c)N=4#j+4Z%~H-$V3Bzgu9 z*fcIxsO04KbV*nJMpprgUtM~zBa98GKlaJU7S%Xb0_QH8ylqX%w#H^$8hcHi@6=3A z{2HqbS^UVUmK$m`cvS?^e2#RWrC#rFWTZagbC96VZua0Oz{z8A@o*f4O!YNyX(XjV zCN!bSnYBsHWNUe_+XW7MOr?Vh@ev(JSoL&R`pApLNdKm;F*34`sc`SO%jkvCs#oGR99mA+12o3*r&%Okbq=;SWk3ea@I^@H;ZA_jJ2?Y ziS=%@-EY>dkP5{hV6BOI#p8@`{*A|1Wr4rZLU%ynnw%oJT&%IS*<^@mTuytX>3Cfv zpyL=836e6UtldObgorTt<&6z)e=5=N&fACTEcvm=lXhG-(O#3EAaK*$bMj@2GE~jd zi{3U6x$#%O;MFVi=Ltq`M#nN3SrABNMwhI#+&3z|Xcarq#kTUb(Rk%{p#AX)C=jcM z$>{9>Qr0?p;}uDV82$G0c??UFM!3;}_~?n%&HI8Bxo|RuufCSI@B6Z)htXbZJQWv3 zq@T$>_5gn#y5k5@j~{QEkPJN;enl5FGob2!>9}afzi|7M#dSOR^1BVnfNxAz4fsav z`$nol@-QJP(!Ss}KNcHS4GAM_{(RgA%3-Lt_8e`Db$p_Dj%;7u<4-h-%R*j!6+a!` zp1>3B32}z~=D~FnmR?52jr9Xn+X5~|$4UuV%}LI*i_NNnxqI{3_9l!?b=)u?`Dsbq zz(?_KUgwSwFh#bcEOEco6}&<0a_em(D4XP)5gbM0h-fSpkW;08BuN1rVjpNQjBXDW zLeA$-5Uf|dEK(>{|40KdpEH}(xj&X#s_ZgfaTok;4s+w)tw^%*gvKoEQs+)ggYy=c(wLJYT{5cJ166>yh4k?f5+Mxl&O!*d_}t zFYr#sCpYM)-O^8ckGt*58GZ3*#FML+y4Xt`H|wMKg-^A^=~6D zr3#G4T>{PbobAw~GF(-l?@4a1=cT3CdVTB;&-Udd_qrRjtn*c@SFhgF2c`EaG)@>K zv`1#WA^dYQKJul-(?~a9j|1`BCBY&Uz3shy&to>b(=By1_S|z%BKA`Ad^%16xzgIh zGz+5OMEw0F4L0aRHwvt7Db8Hd)d}5zh0Z@o(u`O)s(5P;4(Q1&K;NXE$iQ-s=4xVX z8b4oe?S1`NmG?yYMD|DX=yIy8p^(~{$$4HLmO`BkP_VRkk(j7#dB8zQCaj75nPIx} zDt|s}K+izS@S1UR$mi(6uu;Ypg@?IqSg+Gr$*1OE=5fNsSYTW_Ln*yjg+7=*4`+V4 zo+!{llUZ=sP>0cB7JQrW-11{(qmV00_C_mp*-@t3 zD(5imXGGdtt`-Hkr~n@79`y18v!dP%rDu6hx6YDGrwey8D3$#2h;E}w30+$&O(X^l zQFmf7Vis3& zZBB)4PT5aZ<3DB7N9Tf<8~h8^rn6zC91iv+c?yOwE5)6iiM_Sk#s*hFn($<{Vz^upfg_ybe3ez77g0}%w}GM z{$5XaZAuxe0W$6Yd?$~|bnNO%q-kracT=2fL2ePTyj1H$3P`;b{}Fk>hCUZlb&Vh~ zd^|RS*2r}EEQX)yZLb2X*Z}parN$m5=0@|Xh*jm~@t~-Ihk%Er(ScO5?BUXyLqa6O zmnD?g4c{T|l<}*i)RF$UrCy?{Qxb+xk9_BHdK-uTV0JAiWR8=v2jvKU_yOAzmCDn6 zx_a7Abow}WkEar^eMK+y2g8=yd|yw0@I5KL9NlL%4ufiqn7<$oyT;PmwV2HZM%oxn zLQG|3()`!mPN##KX*KZtyg!DkADSG~nfdWH^(!$P=2pm_?r4RAAj-#ngVb~ErXipn6{u<}W%=mRNR8)7>=?N+%^UNC9fxaCyXRxOHKxsF zPw_H7z%zuiA3~Ol(+#ktQY&?%)zGr+f9Pv)V|^Y+2;qzb*A8XdVfZRL#?TfVX3{0P zG1u94KopoWV+ul+{Pd-8^T_PV4p`OcEF-yt`hFNQq8TEnDkfCk4r>eRv%u0QJmV{{ zBYolCM#T_v8d`XsJ~jFtu|HOKn1R?MZLv#9bCk6+E$ za6mzJWfOrNBb;kSmpU6Lr z-eG#%U>)hb0SGjnKEQkW3@~vb<7zuE_w>-oLAdiX_RKP;spH zJ6w{lZ{5fL*`e`~o-+PvCQaC*`DMa~h`ih`#&)+=atzbcl701cpR0y-KI&MzMuuz0 zY(DC%7W`C5k(@5F#>skezKR2u^lQ`9S5$pE8!L!*@udRoO`Qc|2S zp2jC8ilw)1Wtcv)T~ZLIJ$O7X`%*dRlKoR-wl8{#EsXFbEHFO_NGRkJR0 zG3LNpR831A;Le|3VTldPW)MF_U~ewj(Oy@~xy*M;3_sc8>z;JRn9o_lBfS2Zh#|nb zN0##=A;EE%MlGyAoMtx3Ou;JxbLkyNL2oK}hrg{Qb_}#p1o;JaZ~oA#mdC%!{9y>4e0uqB0#i4Uac;;p3|NGPCDS?0yrA~ZKbM^oRNsGBBtER&!t8Dhv7hR1)XbS~E>Kye8iy%cnf2el1j-1E zz;Ru~lOGx^M}W@T5)}~b*wP|uA*HL4`BJqn{Kwhr$G1U*vjBf4*;!#>%!1NiX>b2% zWuMaCiJOMq{_=R`(nhs~i9z$&2w&Ldm5iZ(Xp&up(s~)KA#3atLxQJd+lwb-y9n0O zRB*TF;3Fed_qK|2a@`xIJZcFObAgWp^Hr`m$3eAMV%s4Lo&)M1zeU6$uTyEPPAN)w zeb#y%!yfxdhnpN{N0l{*VoLxeylM}y&q!CPp^?U+(9qZL=d_CoHGI}_6eQh<@7cB7 zkO46jEzY^Qi=^ihkGJ7psr8$AJconf8Z9Qa!C3Fc*kK-$&oksp2z_{P->hfy)8tFV z98jp*FF^{oikLCC8J<9TaMDXO$fA}*Z9{E)apx; z^(1{&VRAVYV9^o;EJVY^PxbPWSYSzu@o$%$9G;l~opqBYt|3P9PS$|06n6;}CNY4| zF8cS(Si&O~u$LPcNQ77KX3!(XB%j@>1YU;C*@SJ&Jzx#~Ic}6silW!U=x#_dJtAPq zyw7YXq#2m6_ei*q3+Osax%qzDiCyH@ci+5A6J}%6!#s6LR%@|&7-j)@GMV~9odhwa@~_FO2E2}45rOeD;N zzkHCCjh4iku2XzCI4o%fb5}j~NVfeXBy`J9UC+u7-KH>@;_R8x@nEwV8a1O}{LH zynBuyd&(Pzcq>F#;w`zN$ku#J-c7~cm%l?%VBhqv-|=9TySOswmtbzXox(UH5i*Qz zM@G=Em=J7<7Qaf>UzPAqd3ShXuG81uUwVGt!0M>rzd9XYq>KFT$Duf&$Tby6*LLNR z`3o5f{yxoqwP2Sr_&@y?K!_DoA;U>VpY@^s{aFNcyUEkvUVEm9F?BF!1%l?kb~{Yb z9wS6D(3@n^>yh4k+ep>pmuc8zD)pV|h|X=k*AdZCSf)8=JFy3e`D#P4-rHntXMH?c zGom619~nCysLD_8^AJAH9Wb`Pb(a1<|L&8N5$upHXj5E9(8i^w{yC1xz*vRvG|_=+ zTmDjO_X~EbA-zqG>;?^v#jmEvAs@WA^-Fvtm-!k>SPf?MLi#(EFxXq@5^JzROuAMH zB~3$(szR13(+IC0=d6uk(=G+GvTk4MxT0OxtQL3+Yb3B}DgOnf#e7dcppPK2#`FzA z!#2oVnE-AJ(y%nR*A4kha@izp_+-2aQNaQKLdD#!ojhO1by$_$w3w$3^@8_!2p!Pq zZ7{J$&36aL$$^hUrCj+|y_3YxE%fX)y$ml3{Yl>cFJWRiska;$@vnBji|lrvK@P_ z8wT!A73DU_S7_fH|xY`<9Js#aW}-~lsYY(;|X6t)>D{R zRRV5nzk%`hVSU8w(Dm$dgbrW&^h6x1ql`jLG6b?*B@sF+YkFp3-D6gc?eB{!aQ`B* z=wY~9{!)ij@jsw)fu6g*k1HxSKvuOu&T#iUynXdk*a+|^peiKH?1?{xU&{a_ezTT9EP_HbgF3{NA;C9R*y z+CFK=5o0AV=-9_f>^uLl85JA}lopz-?XYTf(o&+zC(dd*Fm0RarZ)TQmp=!} zBy6}&D+Lrjk|7glWN02%j1saO;xY7$Q0$xjWF^!yh%&xv=CUHbF^D-t4YWXiNYk*O z-*m)+bhTDQo3S^hBMhq0GNs@WP-}nDQOq6Qo)gh434=c#xttN@_0G^K41J^-(u@BPvKK zO&xthl{QwMzTADLlt(o!i7)J%zOr0@N^m$#uu7{g8*%pnwwi@RlC? zQ{H$2-*Y+Yw(3ATtQ&tu=Zbeh^{3rNZCI4p-Mz)32vS~9)XvgerJP^PQRzmR>6XLC zG&OgbVQxBkx^VKNw;@Rr9ZmJbSmcJNMzOCGw=s@E~{vw;#5E!R(j3S%y>UrPNro@ICGGL_J z_z~EEJ=D)D87Y=Slkxff^E)%v=g)bEiV+BM;jOmn*GcA9iyi3VXk@~9G_Yj7F=C@c zoJnRCQSv}eX#qXlOgq^&kl^oFuJ(8xmD3~U_b>#>g>Wv#<#5HU1y@*_w2W4FhQhzdMMAPcSrIPU|V9=}?VK72^kEijB$i?JFN>BQ2P^_s*=lG$QtqK8NoJ(wNQ+#oLDIZV ze+v3;Ez^*VFIE!;!W9s-no0OI3GES)k|gDUt2U?4d^)8b;s<|v&Up^^$W6TQ*kD_z z5(l#WE!DjlV*Gz>uMzI;kLhBtyk(Z*j^c%moe7?ndAxLLKyD`}W!;%XJ(;EmWwC4- zU7tzrAhUSbZZj}_%NtoObL^hT>L<(K7^bOe6;TTun>tp`vB-n=Wv8Su8wSQKj?Ss} zwdKYBCpn!%p;pN#E;&&_J(2ow$QN`2J*Dr)nv2&zZ+P3(Kk4}Wc@m&OivX?ge-7z+ zfp<>0QNZse@x_&3df<@9t)*p>QiwxIr^vQBl1oV~x@K)Oq%@HIX`|>gDz6aA1-BnX zFE#TLh#HYvY_1u31%FLrm1*px>BtIOZhPAT8+qL2IRmKa1*$O|FzqKId+r5G*&(I0 zQMD8A>D;8KpR#3drv=)-`U{QK~n9tgnF)-LGm`|X3wJ2;nNz$op{2Kz7%*^nH=^0h4-_nZ0wn;4EScar8kOo+QA9V%IZVs{Hr}Js3&u zF(?|_*!~^d(Sl_F&+#+JRSbZbyo}Od-Z%01Ndz{N4DQz26gV+C+^eZp<9ul-u-Qgk z?UQL+l=bO@H@A~Mf&8{{i$&fm_m{E|Un<~E7+1gL6@w{4+)mlOE~Hm>Ru5o^mbtc&hHUO?XX7WajKlT=9tCO`jO<~11j0} z0y8BVF;m8|dphn92Go1+L%fq;(y5*%e5OKXw1{dj|4CU}%uZD{^%^s^rvt*b70W3p z0eoLRZ^r)VH4SO()4Y$3bgEDhDdkqtY*mgksUxYD11rl*i2C3RT!r6qs-`PtEAs28 zFno2GmQ~q1Q=YaM()7`L@zQ80$_!*{bn3b$qK2^Sg638UZu zN~HP4m_2^vs!7aB3^I4XTE$^ExhzxTz+mP9yIQI~dC z4ZyEcdHxMj76zb`ui=~2=3*% zUJcB}Wa^1`uNA8gBEHq*gfo0XiLrL32p9ME@zl)yB5;NX{fzr?fgCn2Fwb!FZT?oc zQdp+XY!TCl62Yq%i6|MmHY3D@mtk4+aGtH07Xr#d(k_Xz^dd#y9yTdde4zttH&E{iq zkK?V~T$mNj{XU4-Q9F#%yZeY6hM!0@_-Hw0Uc?lvsJ(m~y<~hO+RllP(XKOwuE7m$ z#!foLQO4DVB-YR*`{uC&E>D3)4?ud}s>#4QF{wB)&95}cf~wF~SnGUF-9KF;E{dJCb|fKcrAi-ZtVeT4uy8d-B;qs8Vt;nareR-{Srp7i|sc!5cR z8JAw^X0DKMj$$>^9CnN`Uc11)jFQ(lgXgE^CyI_+NMOSgcJFGrgbCv-pA)JItDu4x zHce?D48pEw>Q|R2(_rEG%4P{TTz0MfX=KvFT#NnU9&>YL8o~j?tENs;WoGTBPD0na zrSbddO_kfXe2;)zW}KlZ`BD3^`kp*7XNj7CqdxWEiXYe2K00`!B*(dpDc0H8NANNp zmCVUmd%FF%#bXgSry140%VkoVT_d}kh&kwZr3%Oie%`emj$X53XBn|IQ#8dW-ho2shhYKwl4b9+On zJ@(E@;1XFv9Z^HCfTxn5@e5AmLj=j&&21>{Z=cd9Tjn+DxCh;Yi}9gnis0gp8fX*% zw~q6Xow=bg4j5m0*^O zD7-L%aFA2|uJ+{F*I%6sLVwz)%!A;+tkQiLi6b5xDm27DmYpp-dW3bX7cx~$GUF>1;2g}TOX6@T(MtsMy-9`*a&d&DS`@l037t@eObLKjnqVXO9`PB0~#l#g0W zAkUM`9(J}a$3IKBZAuY?<+!;t-xB?H*jMM&Jtv+=?YpP=?QT^nX8d2!06@*bleH)q zo`deH@+f#89x=HW_@5atVJ1*=%p=7=QH0Eqzayq;$A3w}NNNUm@%j4IP%}w&N&Ke9 z=*|M8I3%BNKO?3yyVqU-mA8vL5BIyaz{L`df*O6PQXm3ZdoCF*yQJZX*?N_9hB>v#LSiJ<8IFl;G{1 zJLDmE+)rA7tr)4$Kx`y@y7(6ws3y&P8DHc5uEN zl6GB#0x1K_ZnkUWh@R(7I<#($MUu*Aaut_Xq_G+ESMyL@g)NrU3YT=hL)>fKC9%qo zm0Y$_yjk25)OR}swW@TTp{gf0v+MOpNsgpkFSkDf5HiDXSb96Xm#&~?>%;x~HG1m- z5)c0zNbo2xuJ&8PgomF$T8M1OEMlyx(57#jGGtm`V}a#(kxuN>bNi*C#Tann3t_64 zw`41rSx#49_v3TCCBgOysl`nk!hY22?@`tABQS$Q*}H6#Wp1UfK#+;|ai zXDA8ErToImsd-Oav=n0>_lcy-#5x}IF>vW=ftl+^TG%?k?V`&WD49La;&=4n!rS(N zoU|x?a2l;-Y)zs!E=KC)(Pjweecvh#kf~GLox}ypO|-rl)?(HulR2P!!NGXg@yk4_ zUEMPp(=y*L^Kyjfs$T5qGwq62ZE5$nqU6A&hgavTzaGt>|o&e$7S{gUqq9du#7sx#HsC#d$4%Z^7;gs~nXwEvB(BcqG|us`|b?Zzs*XXj5UgH38BT2MTWyA4TpC9&SzWAxbx_|Yo6hzS9L7v90O znMn={mQn9^V*VsN`?=HyKl~;2EZ2im{O3~pZTczCT9PG9It1nW{{crqxW2N`G`XsR z<$Xmj+E>(7ax*)k)@aRC`%?Of1=9M8IT&*y-e}ELH%sa%KGeV76ml@;MZD3PsrI%0 z@SxqN_k|wKhcYb_MrxV6LHt6S_Qd}9r|je|WL^fBHAn4<{qaZJ6Z_(qvygKl_Gryf z7m5AxN^fKS_@mj#IgxiXW~dw3{{X%y`*MGLQ}$90WM9o0svmBT`{IwcNB!|fvX@}y zMf}m4&o}Y+aYx(Y{`jNWOR#ezLo^dv$5DuX%Rx*!lq0NypRt!=bE8i*x%yUE)ZzaC zo`RI>Q2zkWK}WHdVCP0(jCD`4X);9`yGtP20Qaw+yh*G}s7|pvUaK zi^Uwl8HWw&UL~%0I`+a%(Vsb6l0fNQY7$hf5z}#Uxqrcu33S<`Vn24-{0(YJdlbaP zxeUDpb6yU1OPi~F;fjjfyRGFX)u;#UVn&>=Uru&i6WQeoGPzoPHPEV=(9-Bw$UsjD$+;hFz&{+ZU_(s z3Kd68RVeNvblnt+>*@_kW!(|1$a|8;%BzBD#z9%;-~)9F{ zcNa!OfmOE=-jLEduLnuLS(ty*x&Hvrmb;4Jd=5$0rvCu1cK-m-mb>ceqwt(n{xjlw zQ__<>(w45GhakzT_E4Gb6*1R}qK39KB`>HZJ5)%xofV89W;+jR*IrVsTUadi-@IHO z;XajJqIej_-@5sx0U-KVv5Cf+GHC^@E$A~!!S;C_{}tsbEmNWchbc2`6H z00{KQ?lAmCC;TH@D)1v8vOGl@F@1lsd$QO?HVYvioIRd%~?y( zSESG2ke2dqxh~sI075yhJdy^2cZvF=5l$6jQ{ghJ9aAIC4bqowG}SXS3)Zbd_%&Em zhRov{1?uVbHFp5(2pz+p`poIbF*R4^K$V$w)3geOP zD?=cIQOc)il@1wt{V6?(lv6OKw~`k-)Q0@8Jn(cJ^aT=n5z!rqis$H6&z|YQygNMnG@AH&J^aZHY$z502~TBjODeD8pGksk#6fF78&_P zW`Dxjw*LU0GyYY{`24i~AR$KxpjRj@jpYcJ+0r{d@RI?4Y4eG2mrvB~L9`y7=1vo6RzgilL97MS$4 zfBWcDPsN!(=h6QF@3CHM_BjV?aI}-skF3kG4`CD0PsMqU{JOvW_9;K%YTSR%ss8}J ztIs~kB`?_|qw6#B7qE%wU+}eVds?6S?vwr&t=Ihe{{Z9l=ig;8R*{r64(0EMY|f92GF-*v$KmO)XvG<`v(^_hpTiRqT#5wGq!7U};0ebxJa z7~jm;F(>`>YszhI>ziC;)bMpK_C6}YBuyO0URQ~|WIEUhCr{Z3$e}XU8xiMnr)`McyzOz2z>>=3d{{Z1+>e&0S zezf2CSz5>UV*P8D`wM&08y5Ga^{MwCu!m!)KN5A1NWWTW_=Bj;^3EdIdVr?6pRl*R zA7O8LKC?dKRARavK8rk8I{Z_@LYFvouFa$JuGw+LdC!Ls+G|$MbB*|~Um9v&rYMRRjA zuKH+Hj>A&58={q2M{3s7^@2W2pOuGNu@P;fBbtil2%Lh-H&n@!a%mjM%+U~3^P10| zS62&~nOUJ-tH7ukcQIxanAU=Gj!h_}#UX8?f{vfS$Nnd?5Bhd*{RwN(Yt6g^Fq6c$ z3J*(0Klu7rq}Np+gy6({XT=AlHz0C)($gNgfRP&~t#n=;J1)alHxS7Jx*Z@CaBCW@ z)|$ENdQpyBFsH66E87*66>bk>RCGCymNC~A&|EJ2aQNn;jP>PVRt`(=IAOGqIjv|+ zN>NF{?Ml~Sdj*B5j^}=J#cPhViwdXBpP16uRXFma3Co#@MK5n^h))YX#g;POfvZ9X znxUv_wwi2Zr;u;#i`3K_{;)KgK+A(O@SfG@w-d>EBTFlmc0Fm>))2m?hOy!~^)@kq z<-wj5e_F~1H4`Zm)qEUgmCaJIGYVil(}~K{Cur$l%)9X>9UkeF`$>iui2(WVN3dsieEGK%Q5_Sv-8MINS8I=EyUR-LPY)rAXZ z6{T+EFvWA#k=0Ed6hv(mdK1Scv8|;%X0;)Fis73&QZ!Q>Rk+uRqYGqJ=*?itEU!u! zWKe>ijW8-;rei^1Z&_s901BsW2R>uojBg`X`#DyzWY z_U4eLDZuedn)MmTh_=+Rr#YsMwIDbY8X?V0&N>XC`>_KquJnYdT3Y(yJ_u9IEP0 zRDv<G(|lHXk0>bCJS zLvwD7?x2+nI5l=1E^QmZ_ic89Ij>|uio_CBbDGBKq(ar4oEGOB+o9xf+LGc?jf+yx zHyc-_KrQ5WO!)oSIM1zU-I23uD#oUZ2o2^pQ_{No1ItLbCn|OV*;c!k%Pu*bK04i6rBxb-JAnjZ@@K`dT@+=DJ7`x zc{hlFn_8Ta(BiU%y8i%nx38dSFXnu8V^#jjeJj|G$n&MSmG*W1?JwEa`?aVoPo*!} zPo*&#UO?_B@)y#ze#m_&`yuqE23OhF`?P(Xdo5~LXnTq-I~gJY{n}q3dRDQtJ%u2&K9pPxk0596)H7KL zJ68AE52ZBPA4&vstEYt>DEkQXt>bEQ(v$3`(txpswjg$(Y2dzbTgzsBDnV(E)B%!~ z`xFdQgg(`+_Frm}Ejn3FBZ=QVfX-+cjpX}KVb5ylo5R|YvV*7Gg55J2tFd?=UXTL1 zWVk%#h#$_DfN|y=cBJxstGi!^J|S{3W`pZ0@%q*A@UKvgF(uv95d+m%EJYWJ>J~iA z#qCn2opUJMmeRWVgs$&d(RAMnTf{_ddwVJQ0~`VGip8DvOSHqzVT}I(1Y8QnRpPWY zjW+c#B(z&RHN?vs4nhuF@~-Q{I_KM=0T1sZ;nel|*CN_lpgairQUP+Co2n zvpt+()-OYQYH2)*gB4Zf1m>W5Lywj~e;SOj{_y($3Cn?ou6Zg5tW3J}sdL>0H&S7LajWRg{b5s-B~Y&U+d` z*`2P4K*u%Ks0P;+!)Y=Q!n-IWEoj*5jw_KnJELlf`;7Y@{!^`EYPJl@STD}KYiCWl zkIxG%g&ZC|{gdy36{3mIsipL*IoXw7Wm znTl-~pV{|8FzllPw|cHL`P1exoD*Fw9CWGnNUnBbq!%a(LkzYIRY1 z8y50bq%vdXy{g-rJx<~S{4R0y zs(8(NQI|5g$!jsAtvftYsOFHIP+Ai;8S78YG@1)SW}8b(b!r?t~3yqHC8 z3&_W^ir>&YIX<&CNGWcomi78qsA<|ork89Zw~1HOfyw&R%2or*0DwDHeL@oU&5omyR)IL_Lt&Ju z??4_Ut)t&~l4p}UA;O^kby{e`uQky4uF-6D`EW3@s9{{}&&~lA$34#Yd??SkK*Ixy zZ1>E%DX|c#X-+qK3YKP62XW)BdsbYxFij@to953r>536n>S$SoOJqUOMmkleP|1p; z0Xs4cUV!<7-ldd;w+@48t8l`^^N@Y2NFnEunz)D;r6$Hku(0pVOd}ZQpTeD=N=75r zqy%gz;8S9TTd4v@3unKjO>1Q>^r284z1z~dn@cEcVc45XcWPpCv2(%p31xcy42=oN z{8qav<-9t|b%{qL`3Il)64z5*bbb?tx5j)*gNl%kl+zQMl#i0SS#dP93q_vi3;`9_ z-{>*gMIFRUMtX|E@QhpUI#+MvBeKdR9<-#BrpFf4(ML=eD_xzR&j}grTxOiafPE{i zw3qiPz}K90%VEi2i8ki|l50s?TXZtp9I@)AsI;u+u31b&Z8YNS%$mA6pB`V!4xF>`1T>D+FRK$WRCp;QWi6hCpb8F;lLQpUy zTINR;-TX#`3wSO7=-973^S$fV!%a#E#&CSsEasa~dDI!JYl%gYQW5~>pn32*)Dlie z;PXo5GQ&yvR1V(N7nn$-$uzD_nBk7NrP>WNAC{aA1*nUG+NFXsB+;UbmS6=!TNPtU zo#WN5q(}_KBn{>C-GiRBGTTC~l0CP?!|WDb6Vp_4=AmKq13%8XO?n|Nm#RhwNYRCI zH~PBk`o^6xwelK0s1UYxMQDdw;#mX)~cPTj>u4k#_nAqefoWokrynoRmuqjD=RR3GT? z^s0x%g)fBV0L?DiekjFy=bpv|rQ1#dmlOzQD8)F8&;sfVt^P$ z08{A205=q3oCPi^fWRw9M$+3<)MK{|v>sUdRUB7m;7eilsU|*^E!$8#OA8q6t>S_- z3aT-HIIHY3X)*i3>?t#nYYU8AQsRZ_OF#$(D5JFkfDzC-;MG-Q1e+V~)yjtA5h@&Y z?LZSL$j@%nAH7v=XV326?=4#DXak?|CYJ^LV96SiMgjCS|eG-0q+g zz{ooGuPgB#v5!%3!e&m0$a1RP=KJeI(t;6b*Uv5iLIf;e_m>9BZ*mk>Cut1iiI+_ITf2E z+>Jn|7Q<9<%pFGK%~*!oN!$bn8L0qJ&cG^`C$OzMO%i zhpQNm;mEG0yD0o8Eq{#onjVIvJuy(kaJ4dwjMnX%MD08g0c(=j;8#uJqC_zPSl7lSVu7cKM7~0({6HWmcMQ#NI172%qdoRX!VAbno z{phUQVtQ7UvwY%`W>Pe>f$vpsts%d(Xyi~>at&A~H02l-YE3(tw5)JQq!c<<|;ue*w=L`%1XyH;bN5(Qi|NjOlGQTSF!2#Z9IfPo}&kx?+G zjs|EVqzdGEih+Sskye;gOgN;>rn-q7x$93SJu0yUO-QnF%@+j6;}q$litiQCbkcO2>iv zR)h~(iK9$0#YD$B!RC{8MIqvm2P5P7e9sQbNXb$Sc@QMmw0Ols_74UTBVNg`~O*BuQv73(K-dGB#CMk#ScF$hH^I*fBn2I7Dz*rJdL ziZM%yF-!!b6u6>*6BJ^UQse&s*FX-27n)HVV=Q}DbKte|XO=#A7_7ez*m+hn>Jlg? z1!Jwx4@KnPxWUN))>C~7M_!_vpPHTzBBDPln#@GYLv*BGl&7TtHBCCBQ}a}ZtpF5f zG_BH_C<3*_1h<->cy%7NBr=ywp1s9OQSI3I90Sx)1fRv}T=$8*KmDl>pKbuSQPBSY z6?8f8OP(`HtF20moL$l3*OqK%@+FTFDc~^gRHli%=4n?B{{S_5FNp4KZ?C+&7*p*~ zpPFOSKA!c=v5MnWST#FAY7feW5rxls+O(xgpLlXBYAfC|sj=sm z+Ks-Q5R&2cso)PlBXt9y0F&A8VY6#KfKYtqol5lrcIe-&Kb zp@!Xjn154T>pK&Yb58nG2%z7UErWF->C%8HPXdq)%{XJdM9+`AQ5S6TXt-Rk!#(O2 zT0ti?<0AAlgxX|IryNu~hQ(SSAG+B!3!x{cG>B@#j%oSgp~)OmmGRbsM=>;;bJm@_ z8dV#P1p#6nhX?sp-8)Ub)TI&H+q;weM};3#+w-PvNZ1m)kEKfYw+$A?m4tuu&3GS` z0b_Um5gS+;e%s<31b^a=-!nS=lhl7U^Xk-yjpYN%fI()#lsA6E78JyEXJS&G~fW^G}1}y#TdY@vIWo0(v*=* zEOAZbo|F#aJC4<2$4$G`rrT{B5#7%KRVQpP1CmZF)3keWZKT_$a)6(yqULj{ZHhWi zhV69yvPp@j>z;spD{hnuC@^ik#G;mrQV^61S^%H~BC`BLAx$%AJhOpX)raCkh#(%a zR89M(L3_rQhX4?#71r9sFa|4`(6i+wb=HfJE6aPHk3!s0b5=aO3RGMHflogCnnX5M z!tNYY>y9ameW~#R$0wy00%ZoGxQy*4tg8%Q9x0Lxf-yr9dYm4iAD4R|$6ji-Gg|ue zB`y{~I6PKQ9jnx)?-X&(T1-q(DQFdB0T`sGP)2A0v{K@XQvn4oD8&F$0L2)kpily1 z6(kU~#lk|rdDsC=Ju9N{ZGgJgM65Efjo*zAL6E1kQPbfD!kO<|lGR5?LS4Z)y9snVTHPL%2Zhd7p{H4MKigYQ!h zyFd%tn0nL4N^4L9b4mtjgixmjfF$X~M>A}J*qi_jKRNo;Gog_hJmYZUlLYMUc{~@h z*y^`7#xm~GmIJxTtCxC=x}#g^pJNL@m_BwNFd43l+Z$a#F^PQ8czj{2cE;Kpi-UI2 z4c!LdG7WPoPWzs;vC(NhT~Ni+o;XlO*&%ri#X#`M8N;J5zAJn9s3#{F&1~tWdEVAn zyE)>ko-s{ecAjf}pp0|nFXc@8G%PYUB?FERUHBb9K@+f_+E!!>T(O}m>PyC}Y| z*19xBP89(93R2wTG%5*hH(kAoHrh-u;iW!f`}D02J+bLY(t>!cQc=+uxj8dFD?E6d zU7X}vG=G~Z{*|#%#4{%D_{B|4D+HT+pD0{t?*nd5MLyjyz#g^G2Ug*RLG4VNOAMZ5 zK<1qq_g0baVzTw|K&r=WzCA@$fl9qiDKV)>?M;A&!LEASobZj6i}1B=Shq^UP7Q0> zzXO`VgGTIpfNM_7MPM`&KjYqE;uzdaw@wF{b{QyQ-l;urGvvzS(vgfrz>BTKcX`%Z=at$Y!Q014UIcGWK zQKwO8DMAlgk}XWlX!H;;E}LFlA;|9D=0e8W@{swKcr9lgSL|$toN&u)!4Q(<@Tk>X%xOjNTNI zL>Tk7jm$mIdTJ0BgiVVODxj(JgPN}i%XO<5_pPE!8p^HLtqP96GMv+~?@BY;hDoz$ zA!t;dwV<6dn#r(Zvj(+OC2NYRvpT803sE=C*0il;!wxf93JqVjIK^|#8^N6wn}duR z+_pa`=DF)89A>wzPfS-k+|imNk=CPgPh5q~MIk4>D-vP&tQPR?*#7`xyl4@M_Pu=! zx)tK}K3T{6?FsQNOz4qHEnM58gDiCmu@TyBjxL3gI*f ze|CTjn(gdzRA;Suk8{y`8xb%oQgCrm#JK~dT1bHBG+55%W{>8{=}Q>&=~GCf)|h|} zlmje97y~Ai$I4D9N_4+=n{9@An^CX!&(dtvUgll{A-Yw(2((kw-%b{n!){g(jJ7*+K78F!@NJ z0uz)5r4?S-I)!lXi`gNsm8&;WjR(E&#g^p#EAjw4;8XF znz1HZ%;2snQhjkuV;*@?nlrd?IG{jC%{L(Q>q)_Frj!gG2Nb|wPg+b4=QQI;-7`l} z2xgD%K1Si@6fd{iAb(o5Twa5$OQ?A6_T7LH!))Pr>+_{{l@OOQxvgqWoRWQyo2}z4 z^}TF>M~rdavK$z`;Okc`?>4U2HH{|}eRV|IGW)$`?JWcCb{)Tms%Gm3`z;c-^cIIR2T{q7BQ z;g#ld)p`{jM3FKu%8|`@AB;3+(lzPgRm4npr_k5Fo`;(AzlvILmcrisSLRc(zJ|RP zVNKbdOhM_2jgT6!PDMmUYc`IT5)qU&GRu)h+KG^Zk=ByXm2_AX?D18SQaEn4C}3)@ za@4Lqntl3Gh8?IlprNNUgk*N7;{uzJNKAySMq;Zbn@Nz-S7F=*SlU1ozr-q6>BS^X ztTU3hs)S+iD)MqF2@?`kG!obldQ-5+u4z?GNPbaCAm|AMt@nCU{Hcd?RxIWj!4;wv zAE2$_PW?=r?DjE&PCaQspWe2X1@xz_x)IW!v0_z|=x12XGR`*xgIm!^_#D+_wm|B5 z>ss+c7SZcmwdpi(PNHnCBy82FE*KCwq%e-lPK`F6mCE4F)X}|RF(4e*MQJhPn&oWH z06Vy?`%8oX4CL1=;?c5qIs}vy8=Lf~13Qq6 zeih;~#$&jV4+U6?_tOfVwaw}t1=8=GMt+smLabo*GMss~Jc!_OdV1Aq zZgKLU{A;<@J{#(23t3Mb>Dfst->((OUFdiEbcJs%WQdOCc>L?O6NI*BG}MuvyO`wi zYII(CA4*adI&?q1yox0sd4Q@VM?=!IYjn2KEzuVlP{->8PDIK5224$FoLPkQt#83xf_ zVWK%n*Bz_TblI~QCcMe+dMi|PF+M=eS`v(8R%B8G-mM4sskto^yKWhzPPHEXDavTL zi%4l*fgIFo*c90xjL~qhM(_fyj+MqZ#I5x;*<2mSO>z?d0Iz@dtE&w6F|V;0rJ{=K zIl_tsDW)MYMFN(96BJU>KnN*lqJR}wECKFo)zL&+911#lV!W;and2Yr3iWG|w@Hk8 zAgt+WfV9_-TO!q2F zY|;;UwkaTEAMTol1BYQ#wN_J!@k~X!D58K6(`m+&y#OT{r8I}1S^(gDXeGX53O>~X z81*&5TgP;>tQ$ymeiVE4uUFOdX>KHEi3Ez>MOM-Dbo)?Q;|lTf^Kwc0*F8!qa=GeZ za|cG7g|2iubVfZg-CTUZ25M**WRh{z)VKqs2tH;t(Q6)LpER2grcP7Rx9Os9$*0=F3#mtGdj0}UGpt7IoU3E~m!PE-@ zpWTtW@f=n9n5344)RU9DK6_FmOk$yuBD}9!d=dcIJu1|ZhFpr`v)vzR-I<;X3>@++ zrH0B4Urq950FXE}h2@Ubx1rp5XK6fiuAC&Botfumk^Rpy=y6)EgLS1&#ZZ{@!KiFy zA%Hz=)$SpR;x<_nGLh+Cd9P__OupWO5tGGjh{L$6#PTrsl|3YE>GE(Zrm~lSD~!`8 z$X;uyv3#(vGFmgDXx>xarddF!BIQLZ1Dbd0bG36q89~6R+KfC06%vE%R#}XY6+MM? zP^oDn8j!j&&mh-1<4ZX&ye)McOS#hxkEL|IGCe8>Q1Hs6WECSHg?k&LiD=K7&)y># zJ;f$!&ZQWSQn^4d1dW0C)m}iXS=yb@xZ;{uG~jY6PZR@0EP(YD84SO4@m4WSBlDr6 zY(|oU^{Ii#s2(tSb@ilK8Z&}8^`w~7YjVyh0HjCHIj3heLmGIdlTN09Gf0*+FU?Xn z0+Ts3q;XEfikzAyQ9|%(U*>T|h9DRe!vOn42} zp_bue1TQDjpo!W^l(FMAtXheai;=pD2nL2=gNknaCr<-zQ08nL=GgtauhzB5s`6+5t1toiN~F~xC8$sG#pu(y6$z^w~Qw&3pT zT;}F?&1l-YCppD(&0OfF`E)wVTcRO?%nmAfp6EGFV^#1Dlc)xCi|q3C-5&$;u3A5bx7u#66~ai9+pA=rjrmuqh7C$% z1Xi%GD3K^8af}&G2qV~xQi|$lvWiv1C^eZrih9Jy=-n%>7?{g_4p~rZV&jwQqW(39D#$=}<{HZ+e+J zRdqa7gwM4CC8j@meh++9ZNcg&V2W1<2u3TDoBebBYpS|A2c>ed{{XLlt#sixW>}4o z!AnILuG5expi-IuN+_b31pMP1u?+W!O^EZc~ zWBtKd)Yxds>5pSuXm>;pwPK!?v1Hz5vi&O*QznOotI5d4QVWKwah?THBtkzf;E~a) z&@~ji75dbQXaH`NPD8eJtB0jixZ#1I3yN|nX_1_*K7bI?8dFK$fD(#Y4FELss&Oi< z+cKuz-Ri}}zDzFtz{syAzwth+Z>h-?cVtE%B~BX|tg6*i_d76{IpSMT=z8emy$E?V zh2hO_`&P#px0d1LK>>ZN2M2PGuX^+v@~?&r&EgB3gSYIhgOB#oy;h11xLuDjgkfE%pEzgb z$Td6{f5P1=2p7zcBA}2IW6g7FkF2thrdVepwJnfG6jLed!LDf~IULtXqfWOruWa9V z42s_iE?TPR7GX)sgR#)Y`9WEpBh$t1n#Uh942rc2;Q%L*!K++J6Di~oiuNSAm67rh zZW7eTC%mQdznjP9#z{b3aU=f$G3XYy-l){>sgZU zrFrDzt$S2u%@lCWT1SC+rB*K!Ua0Bj6ImQrQ}GhQD?b!UVLV6!k?3ob98}A5syhzG zX*yAwaOROgo+&o-O2U9fFlmU7xu%i08j!E0HyETAE0k#YUVW+SFg<$Ihn_J^F=YCC z($t$YKRKe930@CMap_D=ildr&;8G__dHb})BK4u)A6i5l((MJP8rW(sd8yQu7!=Wg zP6KhkracCMl4^JmnO7YuNUe$OS`k~S08d)D!LupUNmzIU^``{IF6A8!G3P`XHNJ~; z3*3+l)b{duGH0=@w$xB`s@EEeAGl6yI+Io^D^sRxqUQpl+RoUj7BaNqSwkH9XRUR* zSBAA+RhP=PWJA_T1_gO?-&3`0V=P++YDoOaq!FNIW557cpxOLBw?zi}<(&5s0gC8s zd_AS<%bROu+&_(pCZ*e;(>YHMcxF^}p7a1XB0Te5%#Mo+z@TRp3z40XjIs0;ZqrXF ze92DcG=|-`Ap0b%2g@6_+`P&b2v#b3B}g;<^fjMo`s;uxjW=iGSn2*pI5 zi-K}Y+17Njb8&1iagu_uBAoQ@I@eWibhBXR|vMR=)lO~_Wd91N?GuEeDl_Qf*;L&h}J+oD2vWjE#M!(9W zTY=J@_JNQ%s*IY|G8|Lg8Ly$n$K`fk)|>r;z2;%~R>+l4r8pHeueYh(he{ocCeVMo zKmI#W{{Udz_H-lbT^M1CaA0YyM8xRC=6|s6^TQ|SQAwoU#&P799;$O)XaLfM-P5&S z)+N-BV-w5MLe%Z`8KgE-d1dkl8Lo!b8RJ2|85Db=uEkV_85Lv9Okj`Fu$R55vUXNE zn3(a7m3B6*R-}q}x^I;IE0=uvul261ChW@*vKlFA4J~$@p+yu> z0Ywy0AfpsfKmetr6rz9u)K{wML(K5(WBrQqzT&-4!;R+XF%M!YDr_`l^(1;#{W4Nf zm(bNmk8RA0=xJ;uFl9JCrzWh&RfZ2WVmhKloP3n3aw#VN09I&2kw6k2l~U`0 znzkwnftbwzExT@vRNd;Gqx`t5lmN8aS`7dpq@|-7pbRe)I}I(PRw`eNj-tGJ_TGDY zmRV4hKXr$HrF)j4VwcwK8yp_SxpSaNZ9dsyNtKRQcO&aqR-&pNmIoT}*Rmn_e`hLM z%QJ>mAOIS>MNU}fy+>zimX_i*9^{UOtsvmmu#YOS$B3%x)N^A%#W-{{5kVYP7?C;4 z0Z#{th#bk9m%RWX=~s6*FlqS#;wm zNhsVrJEX;9BTpK}n~?SX~(%<0GYUc6UyP0OGnWCiXR#<~S$4dQ>rqu2~;Dj>?|1 zrd9D}miUC~I(4iaI(}ERcbE3-b!8%T%H&s_C^pxx8RG`084Xm9TpXK=gXngaGiMyv zNoM7gR}-hr$mY763Hl7zm~BxT$Z!Q&M#!pAwsBU_jMj+ck{NiLM zt3=V=TSFjWjIx49(!O-ptWKA$U0N=7#^f$@(zlwnrNXY3u+p*TtuDX^6+j2^c8Bz zw6fx>gBd-AYH28p?TV;tBRSs1_>*FRQWH{y?fol4!^7H^s=;k7S;77zz#oluwxYE; z6lEr1Pb$W_%~XQc{{UCGTYEcsC3Za#b6%^Zd@4tEmsYtgj(n*90I9D;(zHD{NP%Lq zwvHe_x+{jRCi)|$6s@Z~x5HlvKt#CLq$fOx5cK+*^eZ0(cwz-mx=piV`|23h^TWTR zew4+Dk-Fr4YmLpbqfUWuqiL3)eV)SFJb!i|DrpAM#{@4fwT#c3B*7{wvBx)8%_ z92|kqsj19>ah__t*B>uRq}OY>FGKaB#B9SS)K+}H8q+SExrb{WcP|*J zUrAY*+zBKyb;fwA*4Od|F)_gh2Q*&$nk4?iBN|QBf%77q%?SPyIUkRDzcMA}nuxNSRB zcw8wtu7mB8%@4_%tJsM_mmXo}Z638`VLSWINtxLRj zs-j$gRPjfceQK^niDXr7pc;o@2faARalxpJ%hR3JF2l1P%5mx{2jFz#rWwXgDW%gm zIQ&H%w;slI(fJS3xu+nzW7fM_Ea07cjQSeOxbU-Jy|mfSJ6L~O=%Ma28q1j+?Gp{}0gK6QLn9;F7!i36`Siz+6qyYGx2YPOwg zBTt(sJx3L-8OY626CFSLqfCXef}J-JH`1$Hx5_DX5qC7nMo`jE+IdAu+2qD zD65?qS`x2OMkjw1k+>lk5<2tJre2ld`X7idbp2H%S(vLFI4zKC+2EW@3~r-_P(@8D zs-dno1uQhI-P0{oUAFN;oYQ6LTAV`@l1)6*P<*14&;uoZE*F39sWstRrR>^$htGkTN>gPg+rnj4aGC)uURooKKqMKzdV`|R*c86pQkm` z*j(~Ct}fTigyDs6*h+E^Ys#0Ay+~N@ZDz<7qZ`38j-Q1o;c}Q z*JuOe;}v}l^Eaf@(p$4mqVXpnxb2QJ!NCs%cf!p0uYF-ZCkY(LZWn z``D-FieMi#0b!@*YGi{yFsYnz%_KJR+&`5ZW!Ufl8Km?bnN`#<_pQGQ=~tSS@PgVF zVtTOXKDE*4KMQqR;*;MGvRn1}{_*@fSEp%OHI{*Duvtb2_no?vSh~s^3)wktR?&P7 zr&yv%d;XI*p+6us*2H1Ah({c28jgyifmC9dn;7X*%I9ZV##LoK%2e!F`)Y35I@IE0 zjMq0KFqbN7<&EKH^5i_$S1Xj+7i*jzwHMpfW!hS)Troz-JXR&u*2|D{MZz(;*hM0+ z$;k$$O}n_POASUhR1J`8aU_VZ9MeG4L*7gdDznHB)}HU%AK_f~rKrr(8+$08Sq}w( zBnle{#FJRo?R3Z*Nu(d{j;6V-A6V42tNrU|Bz9o%#A7wFrubS%xHnpit2>@Tp4HOB zWfW`xY$zQEK}y%u*O^B`OFZlY*10bd-p8hEP^)EDagcrMt}`qr1E8!=6zF5b)(aGD zk<6{R8yyWhXhdaZM`Lgt513Zu5{gFS`t|M@Y-igpH+oiuhN&FP z4845~7b(G`ppN7q=zf)UElMGPL8?()Ln2`9#Z;O|qkP8s9+hdcHm-D@WHYDmoc1+J zZ92+ahk`txO3RN}iEyNIQnk#CyU=k>O^cCrBDwI)Me2}g;@K5;J!3s^YgpEj+)~3XhbTc)h6Eo%DFt?GbEn)N-&@AV-9=H<>?|JLrMHMi z!_b=9Rz?z`C8^EzqZEU$T3Xn^D58o$7|#{x{t;|xC_U@XpiRwP&xED_0Ev4qqf|_G z(CVdK$(o^m0Gi)$>b0b&a*w4%P;w7L#YKT=&UVfGxr!c#r`ES^+anc+cM*c*fI1wC z)U=dHibr~wyA7_T*~jkN=~N(Y1!zb!5&Bk4O@<&}GK;CPOe;p_ReD1@-QafTV*F`uWcdsHxpU`WSAP`{N@vA2YWCK4GO zfE%~cw1~i^N)d*}xV&WXEpzk*;+ZM=fc2+}jm~8Pv^d9(1qOn6r0qayZAH#2I@x^9 zR(S{^KmmR0kMYNgu52$CO|Xz7fth3&&poRr!#*vZ((@N`%PcL+jF!z{>qncE^kJ6P zl^rggYJ;_W&A7l%=HVPewG+FnrV*RdhDdh)7p*J~Z-o_J5;p-P=h_=Q9h zyn^V6LGeAn69Mz=r7z-(8pJH3H|QuIEawYk?ZeN0v@gZje=UVmom%;oYJ--Xh@tmC z2HNT16dh7+;5|W#*az6GO1sWaD{zMnNZXmCz&eG@jWa|bp0mVuZc1!GxP2C_ObS`> z*qi$I2xp^c$M$|0!6DH~jkFHk_tw=>Jq(=)5RC|;=~`Bs0CVH+D&eprz##f~n?7@r z?hE?SDn}iH$JwM>k?8u75BZMZI1Exk_!11dGO*1Jgyx+4r?{djHWd?jVi)?G-&u4SSiiUK zs6xPcYi%R}D7!w>fmf3M@j$9Vr(UU552TwB5E`F4Vw(tV z%jbmb1KjoC)|$|Ui6&o_(F*Zocf%!fwq_J%ROwFOAD~X`QG+|RXt{Ye&-0yu8h*qY z{s#kox|KmBpFXN9c`H_@YkJljWaG8$gO=2#D3M;^y(HqZQe73vW0ROJOxOzK|lFTc9}XbrzSV=-WSk^;~$CMKL((ID|(tUiatF zwfmU!!lQ$b#!9I+>aR+vN4QIZ>R`Lj##Z zysE)Mew;1u^6DL@4^0k?Owwrs`b4i!N~OJgXH3}UC2e=TYn+U~v^J)2_cGkf9cHNC z9?8?U9>dIlYZh`1v3dK&8S*KRBe&tJ9#QTW#t(MWy!b5>X9gK+Q9_r?3kjt&IivyznTNOc;u1sKOTH>PFIIIyI4 zT?QCUx|aX@cPz0_RhSy7rs|E5enr@k!%Y_LSC<@x?9^8?N*dtK`FF*mpI%C71xyxI z2w8tl^H-BoTpw&->dU6MfuDynqkcH>q6WSqCaK}gJ z9?dWI+7=>J%0tV-)!)^Bozq7fo&eKBy_~f%?-6kI*%z|7@UWv39!q~)F+Fl_nyZA* z#*K+)-<;9-PhM|iRUK(1i;BCo`kEJoNl;5FlzpQkemxu~sz|T3*wpt)c2o}2mgauL zDI;+&VQ;0&vqv;#{kJ=2e{QUv>UCB5;(?FE#WJ*SPA9g;d(;!;Ajp4RRw_?q^tEZ$ zZ?|nfcclodgOS(8@$=Pkp4!e|yxY6}D-8e^BVArtgGnDvSKKaWgbH2mvRy=3_(*&r z3!>PALv>$&yb>oqaY|=uiV=dnz=EtXkb7%-vz;{`DXsAu>gcm+Gv{(LW;;Y>x|l16 zvoX}N2VT<%KXkiy&=OJ3NrEf^12Hs>s>E~UmmQ`}?t)}OTw(A#>r=^(i z^qALK-|xyQb3FBCG(C0FdrB<{H>IEoDO6PvZvz%X?8VhbD>`X}tM*t>`E!Ri=xu5a z1Fp+Jzx`t5)?#FQ#fDa@rvZzr6UBykS~dOwd6m&OAq_591)~xdYi?AQq|lpT4t~<7 zm6o%>;w7xb0{aof^4kl#^0GZTc$Vj~k6z`JyH+8Q^R)7{25oF5!SetT^Z1{8n&bDj zvC>kUPQil#ba2McuRd8lA0GJO+D~&Z7+5J%<@3th-BXc={wDx5@se&XL!$d(p&sG) z9<)#ba&a9ezF+_#Q&aVejWgfbeI~Ch{_Ox-e5KnG+I;6BM*g$?O<#Z zW>5K6ghkJ`B5mZR(O~;|FL{`y!8zwGAue7esx`VAZrC0Hcqh78TLAt;P3>~N!PK-V z9N4#!F2kV%n9VT%ka4^bpBqIQDh!&KSq7L#n8g@; z!PY68{W42HcUvjupIt_7Yl0TyuA68YCW6Pux!ZP&3{#3jwLCl>qQGBh&7L`M8x8(L zcvdc1;7T&vCTfR4Lk845NOaQaHTWE2z#LmHrDzyY@RmKK?W3lV6qZhR*Ci91f6)9v zYWFA%ip#^Axz}whOo^Q86g7_MsA2N~KLRd{ec-a%_*i-m`k%ujY}bB8pK8cIlvaJ@ z^w0DGX}tTm9@Bh2oU-?1pllw4VZwEW;2gtFEn@dKajoQn@Lou(U+Gz`1V71Zk;%F>5b@^ z#9yak!V9XU-|xs8C+6GsJBZ|#i?ZRkJ0Wx$OhTxFHfu(>sHge_uYUoK!z;2x>Ay<& zw)eRcLd{r4=g!!7gOO9~p`MDNw3kXck_^S+n&*G+17?ZO+>L1lT9SNzG%exs4`rfK^Z z_xPKc(C~NiCdm%)q=0$(yy^+VXk>4qMT`uzKL7x7$| z4{g0(&B@|EbIH~%GP%9Jlq=aCI2CIY~m^)R$ivbzd%owV`qG*^j>Zh`h}s}I7?8w_9D;;oaiPw-*cf& z+@*K(p{!w+w}$1<%%IA{tn09QOlm49xB7RC+Lz!j6TZ#X%Fo`C*V?`ISYb^ZXOD~0 zyIMN7kW%wx>pzjKGTN0GW7xf^>G&qzg~FID9+~zZI05_yJo-rd4R0@~fsqV;T?uv} zdsDMjdM2*6k_0=|21eVAtg&<9x?VTTombbG5@$5Na_YnDn#U^E`}rjI_Xtd3A;<)~ zKyQ+IB8(V4`?87Y+0(gjE#!(o-&8H!IDHKb{8B0g3ywW2$FJMVm`)AmA)D8q$#Wbrfcn|+db$9_+`oVvESTGd=Nqs0B+l6LsGjiB`n=ASeeR&$2T44nY)LQjo$f&iMXV9ZLzfCw2ghqwR=3i?G@7 z2)R)Xv;vMPxre(FQ8 zNx9DuG_ugzLzUc~loH_KsKRTaU7lmzaQc3q*Ctejev@ut7T2_1kD-iQQ|+(7{Epg~ z+f$SxxsUa9=xZ8>O>WY_3FAL>+BG|R25my~^M$QqVLO}WZcy%n#+DX{H-4ka#X|Ny zsGOXAyycj!n;$JUKB~#eH&og)^2o!lN^yy&AMm}e?Sn*occgY}XWn7M>)7CBdMqbA zoexP|m9`d3?`JRVRED{7RfLn)X0JA2hrrRH*Z<9-+JY*O&m1K}+5ksX#1wUENBxD% zVv%(ryMCfMPKfa;@{Ssyca2+{_-cU$?3U~G0k=CXA|4Wpbi*!Zx$fWIqWDr?ZF8T} za^ovCKStKXFV_{r_k>2!sC2XV)otZ7YeRib;c{0P6c6>fQ0}oD< zg@v%NXGrSi=`rX5-kIQOc;(}tbdCt)?{o_O0-LUnf8g1DuPFS}g822{I!7v}JLdTF-M`#MovREd1d*x34HQsS`m zs95@d&0pp(8|J4;=r%i2j=wK1e*+wpBcO?Fw>cM#oziyFQHAx{6Ac=h{(dsXX2AFI zJJ+R?gzu`bIMW30BDLYXWz4A;db6iO0E&Z^`m?6;lFsLY%ZkTCV8vNFNSj~*CPReI zqGI*K{cv%lYhS681MEd>8Lb4N(2{o?^;$@QE!_RsP?|S)>-=Vn-9t&NNGn~GE9dQg zao~9LN8hsV=3f9;vDeyuHA5QR3E(=|k>rVc4#UD#rPjX6c%akL6WxTzgGtputka!9 zfB_rcmmIi5L&DhMOi?NQ z3^LL88N&>ewpoJpk`?97tb%EcwRb&)Tx1p*@1W~?&5Um!^eWc2rBw$wc1l(_ zTV&+wdIWWtUo0Zx2|QsrubS8glhDhAnM`Ri#`Y6qQEX_dL}i{zPmPZpA{QwOj1HYR4MjCT-H(gERMFKi4=Fr^o`q=(gFOa$n9d6ZZXke!kQD9 zQ#C4jd+5$Uh4EXT1(Lp-Xl>42oI@I+@7ik^er)VH!b6>`g2YPYP@0$ft%qWs=Qb8e zAxdR@N}KPfX)V2|%FtTDVPS1Tq-5GmL}c0yqcE7NKr8ukRgGA3X<=Pv}OD zpXq`ZtNkUGsJkoY_YzYFxB`o76MnzygfW|vKZ?hgXu=6GMLU#NXjQ}&fKB^XpD=F< zMh?{m?7opy<7DHZw0uG}xlU0EYFw$kw3iy0aKTh>5|^lbE>xqlEK*xZ1m$pbEqx2Y z#aU_K`7tu#Wb#Ix2+~YZA3&MnILWU#FivqHv_$&av6wBor|PeTehW(G-p~q8Nc)gq z>hDuuYvu%gx)|*x9)-Da6>5xYDb#oTj(bnIt!Ba^$r86$r!{SjoVsAYxWjj%0)2RI zz3>g4LII4fhkH{~3gq@uoLGJg+`;jT$^5~%ZkWM1TCBzRJCvAGC2(B64PmW{fYmxM z48Y;>-4#bw#;3OiIKO?Zl()F~>Zf=nI3>Zj`*4^->(*)6%g@rw)$|OwO;SLP5??QH zYuNIDUCxhl)Iy&adK>-ufC@XGQ!p)Qv7#cRG8Qcq2q~F?D^p#>g2RXgizm>^(f()# z%rwqoPX*`%{zQ9QGF$-X0Q@IJE0~7-Z-7SlvWx~uHNO`a7t&$Pv-w5@jI_0YB zh(gRq^6JV~;d%gS!E^FPkUH7U*Qnedb<;d^*d7}zeVOodOWYuzpwW~QAwwG5B@v=- z<(A+5rT&rAr|dxH$3%iWx$*O-JEPeyo7>6Ea_b^tHpExt3sLSo!_5~}UE>|uo)oRPK$FeP=mACd-YSdX4c)FzUi=CYEi@zoD= zI~O)R_sT36rX~7fzb@WA-K4sqZs@P!KW2phR1IGC ze57AT4y!qCWBTt-4>BE(nLN*H;mtz|_rqognQBuYv8hkRBE`|r>-d^ZSMn+}s@?o* zhiBzogq7A@y81aWz-8mgJVG@J|8kXY1CH??DnsPhh2R^>=K?Q{h*gY;+fl$UB+25ZVlRow!c(5df!CYN83f%u9^IQVd_A(gG_8Qt=K%)EK@p03I^+XdFr`CT)6eLf)Sinatk#(U?&=1# zjW*SM@s1zo^y>T(gvn?+HA>G;)J+*u&j|^VcM=cbPV_}Z@$$8>tf*~xC^;=wgO>gf z(Dt+L9C;j{eOluQU1;i7yi@yhla=Q)Wf?xYrJUjE>!{jw=<22Bn2a~xsqNtI(23%EkP6=Dc#-5^wu7mAb0u=7Qj-p(a)h-CfcU(OP& zTs3{QcTA+VzofX+T07JCDPz){09Qyi!l~??eJfvxR~epZ6S)eiwJgYd-P78uK|y9KTm(l5KKJ!g8PUPjZdQ6GI8^VKFW_?q&y{s)B|}ku3W`8lL*; zU+wzcL+>&TyedeAcc`l}9S^#40XT$d9P{enlUue-Pr3rQ4}0#scE6YRG~Tl2Yb<9*myu3lG2 zbV_e$>=a&cC(TuChiRQqOC6<=J?r>e{Hlv_Mc~5%7YF+0>h4|f$>oANNR}e*@oQt? zR}14x6~{UXjq9(Cb06MR8whD*?q7l*bpvZG&UDGNF$YgA&Oo6tfoaSDn6LrPR%gx_rRW(Q3O_$NAuJAX%*3)N}pH;=D_EWlSeIf82U$2XzVWP)Wm z*%MQM5JU5mES3+ondoZ2p*AnL{L=9iB9c)Tn}x|V!5wQnx46QtAJ*w8lC zpN{Vs20Kuvw3IB{4!>m^&JeAD-7?NUN?E^lep)?MmC?^^M6ALqALVOB-Z>oVPJ1{t zp5w~yv?WkLL{{K~yoq08FNxmPWd5fO`aA+PBskffP3|<(-TlEjGqC#G1lVk78*MM> zI=ljeI{P~E5iuyPMwsc$^SPU}zRF#N=QW#7`%GbAgH2X_YyP_xQ7aca84yzZf!lAzJpOYAaL0smUz;Y-6jz1do zEZ7fRa=zo18P{x`Jff*YyVZt`n@HUN@14>W_S~2ne5` zI#w67DXgfAHkR$YD`(i{O83FT2|)UmE4sdQnK(4vJ1t}zV{rSL2E=Tm5nb7KnOV&t z+FfKBH~XT>x?sI2?Wu+Hs@j4lBv&&iO@P3s9PJi4eO)pby;vg~vX>bW;;FsbYwL#1 zztJEmSLqLRKADv1+-TB{N%?|3n#Zz4^DEnt_IWL4{LNV7c&A=7?I$Tn?>tG~+gf?D zdHpC42(^cxs4#OBXAiP84%8!TMPP|nQuD5?5U%&_uHq^sWl{sWrpiFH(Ui@;=$vC8 ztu*MKS8O%;wlW-37f;%pU*9@1o-2UXi#=3R(OeTtnasppU+*oQZsG(XDbGU~o!5SQ zpJ5(ji%e(ggQfnyOfJEw0tAt#$j+~#;8-vti^V~P=U0GJ4(b|kZ!d5$}WZch9_b6 zi#JtPbv8daOOBh9vAF6BJ#!q|;F&m&JBwpN@0P{OG;5ZRf!wK`I$v^P4V)bC6~$3S zwU{Vvxdv(qB}7O0EQPeuY5xD^Qz?K)Bt3B&zzM>6@Q$p{)~4 z&-kV^jY*gZeO^96#tgR*ZOTk*O;j&2^QIG%{;L8ESaGqY`O?^{v@>Ja7Lf*PqhL<9 z53bg*yWYG_ec6ck%iVeQ2fDy*U3(Q@wA@aE{Nv!~6HV9YKF$8R5=Q%Cf2FuPDQ8-H zthC39AuAC@PT5p!iC*@y8t;B^zVXf#gQg-!Z&*B9KE0>-qZ;o{Ex(qOlF-}CuSWQ# zk1C$kaN0{(8)pyLUw8m3sj9L8m9hV*`{N${zxl$4@Z~|15kl_bAqAfLc6h%?b$XhX z2jvg^^Le=W3&{q;fPf3$rGh9dntL@!3rnNBX0NQYoiDi3kx1bN_jzjX=cd52iS`;L z7b4z@Rgko;>nF7p;EvF9rS*Cy>5u6;XFm|kzJRz=_)2IFR&$#uxV!NQNT-<(DW~sP zeM9f4exs&BqO=istTN}4L)B@@IXLVRnHs7G@+0qZDjRl(w>UUc(&>+r0C31>XhhcT zN7GjA-R95$Bq|S^|ZO*r%5niT?;4+&c@;8Ncyt+ghYGebeD|L(RvG#f1h9{pk2sYQrlG|LCX;W0l zNKG8+ok*_?@lzuX{#J~1&hNL|q;q{uQ&OdhU#LLX@0%oVuSeUz62>3V^ObsOqh}3f zU!PEqNEy)8fjW+qJ2skWY6*K#-L;nk_L1`V&&gHUtIUu14=D!7qS~ycg9(8#&K=)V z^cGm^S}vJcCRuP+TJQG--rBRKfF>kcUQn~Krr(s%X%Dv%T~(4aJ8kJcF3c6) zJ7r!dZ)n{`s?fG;2K8ng8^nQTgMWOJPV>7iAfc;lXSV?9do}zCjbC_nsM(OJG(fC= zW%@m`bh1-$ad6y|DrEUeWgW{j%5htw#BG+L`p}4EoIaZFA&{%7Klaki@480FEBo?@ zS-r~W&PH#b-+&s#WKZuup%bM*6Q-dmxH8gg5dAK^n4?dx(1=xgfn41cjhcM_Aq#uE z-aI$BET+Y1!LDKUJ%Og2Rii;>Hj>qPydMsVf%fY{9~d%xFJ6J#?``zr zf;tz-E;LKI?P@gK!W^WI{7Nk4*c(-;*Yhe1obHI-zIsn8eSqJ%h&;S4)S+V>{0gb> zzBJQ3jq0N9+u&~rC1?8>s38(^tG${LuJ@kpUF@kPcSgd1h|W-ItLc(NtzV4uYn#@C z7GYe1Gq0HWeC=d%OO@Cb3->4C7ITU727}glNjqP%SyzztFPXDln}?-(?VjJwoc@)c zZ^&yBf_8~Hp7dH=wzKa6?2ay^Rx?k8KCbYv+vlWOsPT) zk2U8VzwXw2VU&~JNB$2vnB+$@Dlem@g9Kvp_e0C&2nzcc+|Z@JDM-%3`Tt$%Ty! z|DjRns37N7xPiDkY`T1+1zF-sk?BfV>LB^`hA~T9(8yKWyTmD?DutI50o913#TQQ* zFp7+bCnM6ZR(NUiB*p3KodltVt?er;92q+|PNpB|w#@Eu#8$g0yDo%WMehoztY+Q` zPc@#QK<2D@&}&vQ>~_x5=f(}{LN*9af=b;_#z9-+i+4hM3gcsNNOpYMi#)%Q;Q{XW z;dp~7=velCdZQhRth#*#G%kAg4AXhr*$(SM9(64COyUWk zaVGJ&7((EZG1Y*7C%*n4pp&~ytCc(RLd)Kb?Zn~ zRJn$&(50bC$nqwWA0+H$CTcG-=bNtjigh7#+dh%b9c;2J&P1qxURNen89x+wDt8NE zW#mLORV5{4Cv!6D#*?}&XtfSlvhc4kSyS@G zpk?S^dR30a3(N`M6mAm4Tcz54u0hU?_c@2=`wu)p4ffbMv}MHX3smflJ$4%Hy7ap; z6o4w+0P4;`3Mf&zR7G|)I9f3oAdmx%*ceMvNfoKT4i6l}LGBYi z#;n9{VJ}Cd^U_F_5*-A0mOs~B_Q*)=xAt4enWycXkSeo4QK+9-C-jIg;yq1q5+3#G zXJsB-XUl;|!uI$D+wbLH_$2kp9o&@JGlq>ldfO0K!%fsnA^XucxVGT7=tn3T>0B&R zd?VWW;|q@s(JqwpA``_F`8j;Naoih(|L(=0Z`?nC_8sFrsrVgq(j@lkbElwdT>k4I zy<&!hWWrkMG2Y@s=iz`8B|XsDl4YJ>*QXhFfgA`72TyEQ3m6kBG+NL zOoaZYOiqLy{m(kj;V5oa@U-6V>Up1vs6&I_kwQ>lem}Q`p+CgtXe62lfipLd>u*gT>T>_+C0Y6}vWPY*|(X(%&N-lr^ta}7+A z>;HcLU2kW`;KuWG`8$4$q=^q;4l5($xa0|1PSYo{#qhSv@i?BUNX9TCLA^YM{j*n0 zOU1u8dv`eVb~w&GC%HpDVE+`%4_LgDvxl9qqxv2WPatkx&*6~fRw2EbA$Y zsg+o3ejxp=Iak)V&xe)&aD!Egpc#x@#^MD(sY3rkE^7)sDNLG~D$Y?cQr2W*%q3Y# zmI_hbxS(2nmUG491i0gX6n>k5+M?G0H$y3+_ekstq=_=9Z)78Wzt@eRRr!xh<82zn z808crO_*kZeaiov_@Z(#?}guY@~_&Pzv6>))eM%y)_zSzG=5zS3GRtgdLo(AQnR0G zfLGdBRhfjmFwS=THA)wx)NGxtEh{j1VgfafF43EDY+}b9arTKhDVw{jhr+ zTps(5sr~_IPKw|2_r_S_o*JOHxCs4~0c=c>I(!KWPAtwVbAQXAxk=1}J9^~u4-oVZ z@NoV65xf5H*o}Q{rR@<&&#S?A~SwxKmBbIe(THpq?}Y6eoFS(!k2WG z^q0%~JZ|RL+f^!t@x12Di~^)adI&O=4!XFUV$2mEQ*Q5waW^o1fYm^78{Jj6#+naK zd^Yb;P8Yn@*+*fi94s0?wHGN@`oA`4+Ue0A#VR}?G!d*PSHwS*C8OJ|35h&{mr(Md zKyXwjDm0t%8E2uQSgvxQcU5KcbD`zZ?iS>V5lMEx%gua;5)%>r1H9o2-qa25edGNP zAoDzUQ%~y;OTR;s{qaA5iz2Yy0EY8e7d!~kg3H-zp^#}^OOP#;I0pHtPqersk?}WG zDUAh8oJxjKC^*vD5pG^~@B;qzGO;t>B%U z3U$y&$`3Qn$8(GRVUP_{T@H#<%PX^5k(#viv(N?SCOr{xS^QSIG0S?6eQSUMcR@13 zPd6BUKQF7v59+?W&QB`T3=A7<8$*e6q^cOwb%MlDKb_1r&7#(0o9MpGVEbPDU!2Fl zS9@BdW~d@a;%#G2)$126o+x&1gpHBR<$my4;ge9`wwW`^5>tq}p`}NsmOt;hcBkfq z0xPu6^2ZnctVd68t@gsVB@J8t-gf^`Mcutm`#KL_@NrSXvDtPsBajo+7KT$H+Z~B%l-4SxeSp@B$bXd{p|p3-3er%{09hG zqIwco5|HgQO}JImD)Y-W>9)pt;vd4}q9rn+SD6;dOl6u2L;L!Ez)a(&?YB5t$&m+o3sdFHVDJnqm;VZ8gC>)_++ zA6#;BHOMdjOQe}r;=|7dlQ8z7w$6Ukh{4eL#{!>|1|`3cR>M zteH(lN9J~mLtmE}n@xJ^p7WL%Hkccy(udRR-r!@;eIH`|+#l?fVjknzu%XJ;)R!lC z|d?iG(w#4p@qP)@+S;=RzsO<|Ta7A=(NhB;>%-37T5(gIca z@p_?<%nv<;`S}O1G1pgSgqXHF3^3`~RY4Im+c<7{Tl{P5wy4MMS$u$QjH{En2{8~Z zVReWqdPr~1LZ**D=(ijH(LIKoXnym2`MRs3-zU7T+N%M%hzXj_4H+@fls+-}qzdt8 zAupv;WC=CcPU44Q+Qt#ej!e9Juzm$S^yOxEH(EmVrD59$e<_pG+u-Axzv4L4-SECC z3^kiuDX{y7%(IiUFOr0XCNygacx$YB1T59~NhpUs!w?!6xgGFltB7io^$hbL`0yr5 zXv`BDL2-Djnou~SRcy32sPL}Drmaf#@dZMq|6YZ=Dy)~>JmN`MX89#!AK0y}M|F#y z?;wIP&SC9QYEFcosNk>D+(4Dg8kum_>(=ia9r*boei`x!Dw+2>Q5Ncpq`UD=XL&s5 zA8Gn&Z8Q>&xTl1P0c*>|>$wMCp^1W@BtmwIAh}Loo6m)=6x+UYUYE)cozm^f9BL#R z{bu6~rZ&g;2avt{2bf7(>Rd{G6gb~qc@yJWl0qe-o-4De@RxhB!f=w%&PNLBy}q^} zC*+i$sb>;r{tVl!2V*1N))}dBE%FZ5rGb5@8+f#H)@cVp&(ERRf8QBk!pt;uj`UQs zf%+MZdCrwB6n(bKAs(zd>ojt4C9aFxb1x##@{kztV(>*TF z+XoDzD<@I#$7*i)Pv>v8iT4Pne1{NZ*pZFvt+N(nGVl%UZQ+m4l}D* znBho!9)NN#+C$_Jm{|yw!i*$YAyPtsXTW+1sr-gDI(Jz19|}$*24PH^>p#a{S$K6@ zODvw-A8--&93Bv;j^qASA>+2OhhpLIwl1#*YKiJ)vHIJmftvDwBUA#XIubBpuHELR z7j>->VmR0Z`Mo~;*W?Awz91%*S~fj;Y)(Z)!RF=d#M9+|3iaC;Gnd1$v;2`%wt)zE90`<Vy7degQ-uwbUzX{n^zW z!Tj-CzYN6jL|&YGi&Lxf=+hp+fzI}b+|c;sm~mNQIreu&+wSRvVJ2nNr?g-8!1{@~j3(?2t+qa-IrAR?OKA}- zdd@f*wJEnuGX#ze_6e;}ewwfr$>n^)xQD;Lb?!^-&O&w$jlKBrN1Y)%n9FTmYQQqa zq<%(`(V_F2%v1d>sD)>CR2=qMrS*uLzPeWBr;ND6@*CxUfEqXw2{I+Ho~yqCcl2z?cll55JCC(`?O{E(*>e(s;-UF^PN%s5oD() zf2XRO_dfVDsukeF_ClONOTLu{i}D|_q`OLrgKr{B;Ko|+umvyE^#rgE&tA_FhquW( z4o|Z{_qN3<$y3{*!S-I>udE#SvYEWx0#vS&Lxzv#ymb8288ycCN$8(hDJ`7P*E~a~ zyWV%}#>h%pEnY=U;b$f#sn_Joi61F#HI3Htb>_FXjsY?;PY=~#Z|Z-5^w#o;d7CGb zphicU)TLe@eAKog5Ihv=^>my+SoX*Ef0u$C-^uJ13YlS-DD{Ml!Wrrdj#UxH!-g&n-?X7b}Z?Fx&zC zZ8SM9y`o=Bf(uJaS8u)ZQu$FHQ^uHO`NFO_TWj*|&xY{Dh}G{zs)_cIF~ZKPIoD0MdgyE@ff`7kl$D$~TP}m#0DbGbxV9~5wwyNBw>9ypzABrB>r2lfkSnW2y!`sW zqWL2NuorTvRDPwpJ%@5Y!F${#*1pYQ=B(zGgJ7L`4_GyHLAuyEIF@E;K^WgVA`wkk z)-Pz+!`=rmwfzUFgkiHbM;+gm}2?KNgU3L+Ak$U z6@!-PwLYDl)HIE-M1=LOXC@uCIaSoKe0xHLU>DijlI2$mIHx*L)YH{jlD*C4hKWZ- zDD@-vZR-8{m%w8W+Tme3SR*f(-tpF5Lb^$~NGxTG+j>*r;25mD6@z(9li%Bc;f8?Q zf>-jnSdL!=YWlpvJ-o=j;l9@!IR6{=gW~vJKg<14%YY$u#GCbwQ?bOjc0E#g+n{hR zDS(Mw(t{XS_k{rDK}$^!k9XJq5%C82S+gkP1S&EA1>Z`xbPWq68oro)`%RXHwtU@a z@Er8tP!2Wx2XG=QItqDnT8fhHzD4RgPh-_VRCg+)IR`a4P&2s5qI%{AbbWq}#HjQh zAz;i$sQ5KOAOv;n2KQN4`bQ=3%~(;~!9?{m_H#RDMg=CfH3_&7S4vmP95HJFGfBs~ zk8{d7igX30ucaWRL)AKv;WH3*LN#51g#$wq!}hVRdf8(3KHbbH<(p(Fh=3F{KhtqT zLt3`Yi|?ka@=dii5!5dfs!4%f3|**bi28O34Rh2H3q@txp=a;I ze9#r-m~_n@%kA>xIln80Gs7TRflkc(*C?lKR#P#9f0?Ctx0G@K$gQlkhSY-e#1#Sm z`fB7@G32cO_=xjkztn&`zuza{fDVQ_LvKVf$!ur2#5|!}c$bw(k)bV3p>)ZSCRdv_ zy@WZ_DHmn8+?oNobORqN!2+)Ehm`L#wi5KmP}Mb9bQ4vVqT5)Y&r?x#uPfc923?W( zw{pe21$&8kH*tfv@A2IX;j@toIP57Z_EVWvVgu+r3mUJecFNS?iLYaVI7Du!+Qdr# zfXJd`Mo{QN`C^z>(;REZ4>+_(FV6_fs;@U##)Xk&P*Yp1?P+>*i*m`rW?0!l^G$DA z3Eq;AnabNIrb{kPEc6a*QXq3)%x05h+%j<@zT-_N2Iv{l(jW4ZJPScm@PnuUrKJA= zO*#TS0Ggm}IP$=z>(z>5E9J*;aDwnfB(A8}1Qr|1-UKB+#dr^DUf7e{T~Karm%@@= zq-4E<8{kti{H5&DO95S%$7l236hM;iuoH(j?pI4?=Z1$;43sin6&4&OtZ8HWv)-^L z*GR?i$ldzmZab3T7FF}+yZt526B^4oPbC`QlIQvdxHX%3!jPRgoqB92c&$C4N}Al; zXfs{QnDBZrLYl5~MOs1Da592!G3ObdpnlQ$!uAv14C-8$Nv*?)px`pWG9ue@H;>%@ zt0OF15(?XAIv<$CR)nrNin=953GvncOtg5XO}|srI==&MG9-1%Y5vmDgEawJIdth1 z+=b_2LRYbKuqDHdK|KXTT={$W~bPGQ%v&64VirB78FHv!sHYMo04>er9y@V#0L!}Y` zoz~G{2f%gNvz|&rGE0yf_YaIv72n%RLkplE{-KIea2hwm^)`wxA z0G;hA18|UJ--;CfaxwNJbC;#=oj~r4FC2LTz!YV|0zHa`W!ji>{3^q4P#t$rgEn?x zzfpyfw3DW}Ttz7))KFPDL*HjM5`ksGo2snuGt;Pz^`19(Zy}`7r2y~ALg*w%_j`#@ zzX2zxu99cAh(&06bR1EGW2`yZ;=8|XUTE-&tnWd=1%c_cgDYw!J;B&Pit3RDyGRLYqfm1dM6Z4n- z8DEFDHB|*t$SAgJM1;G>_O471XLw&sew$faC^1wr@*06d14l=0kr=Sd45M1mMl687 z$zM~^CTmjNXOphIrS#ji!)KEcZAv9R?^?W5coe3*r3v;jFrGn)mr<^&R=VDwTfxte z25Q{X26A;o^Dbl}f+Svx$TUj4UmF_uIFYf}75&6?0Nqum5s47-cV#IonY=<2ZW>T>b3MxGT>MFVzEJbk%8M`Xk4jiJvFH6tF4CDa>At|36iLtV5QiD zUZn}O-{0lOiYjK>Kq)t|YB8_9>&o=oH=&m_Gx0O$(h3Ul0^JQ6v*C!cg6ho%(L0To z&B3R$b>lC}-joU*jvSgGmQg!m4JU`8wNgvIxUhc!FI^AFR(#r59#hn=QRWv>-|3t_ z$*>wCY@O?a839lUd|gxO^~v*&+rCeE1zSHmfZ)KnQy_w%?9jOWw?3)NLc(bKm=mCN zr*_C0aA*{NKqkrk47HiExkQ5dp8vNq;yO`mMLiFI6BS+$&h76RsINgEMgGG$B_3_( ze^&N6q`-imW#$KCj|^9?SNYI6YzdF^H+qObR5sPmqEZt~#$Ik&G_?T$*hJAhtkh8U z1Pn8s#?X^3n|+}Q2Y3s|(IQJ5WS@`z#%^$ZPY`%44}wpnPfa+(u{Da!vd-4|XPJXR zCO0u9Mll~+K6=Z`fJ}6s?aLJlp#ib5Vo)xNSZy>#4wUZ)0Ff)eH5k?6Kr#X_mz=WY9`G%g$kIN_QT%9;geaqos;NfqpevgB*?Mo3d#Lj;;4RVi5y z7BwoK-vrte zya6V%0hnvJ(ePpjcaHS`bc!3 z8zOu_I)n_*RGagA;WO9qnceV}67&+i3B8`$TRSO5Mmq~&GesUniY1t$@=#Nxn4asW zvB)fj@UPSxavc|bs=$>sZcyfxEM3(D8Sh&8BQmB?wm5<@&6^M#B0Y7Yx#bzI8tWx@ z4<3XU7cI%%x*sYp4^U!Y{1FTG2``I3HrZJV(1~$BRO~tU2WT+()0n?p{HJLK-|Yi< zpzgY8%hN`kdK;=YC-c1~$r7|uf|^H5-QMobT|AO|a9b9iF4m@(HHnT|0C}Xq|(Sv!GM_2if5VqB`FC z$?Q@29SEQb$}XwXU|r}_W=S)IIyvKm@xgV(>|yi0^Xv1!Fvh^RhzvwVnI%0Of;cVV zktN0BS0@D0WD>2aHENd$ojX1csZ}yoZmQF$zgDGc;j_*#9%t%SPId!}VwpUqNjw3Ls>jY9?@Kck|sN>YTaXbR%EW?zvrsnC=VCaL3Up`H>g?BX&!SsXs@n?oXE{O)Abk|HpE+Z;5CHLFRAexj*VP-l=jxdrDsRogIBbf%{mZ(NixujI5 zQjDQeA$5q73Q5v=*Q0aZ-|zqVzaJkw&)(0opS|~5d+oK>{;s{b#&C*+L~!;-sswrK zK+G?DJvYqMg=733iJg(|fewBd9Ni608z`i@F)Hc#n3IC?YR_og_NkRiSe38R=)Sv+ z4t zNJbV*;=}E54zPo=9*#vJhS|v-rd|=ggt_`+$2*V6r0VRlp3TVPD2s}fvbD^-ai_24 z@n_$~xFwp6(k*u~zLsHm7beQPoSjyDE0voq!y1(0FF1R74G(N4dnmZdv*Ps6sdAQk zn5^YSIqa$9l=N}o zqMwCKGzKhNBM_)k2Px7bt>ni=M;TbB4kslS*cK3fEQ^AH!;Y+xWpLCH7csPf-Jr2V zLkDMF2GS~Xj5!ZT>D2CQpq}6ihlC>8%5C4>vlZ9qAI&ydW&qJiI5q3}GAUwtV={)6 zC|>B?hFwfgzfNjlmXu+gzw;Zm(|LP2y5mH}Z0`o#d+hyuCnE`N`1pJ5;oM#U^D@7PZ6-m7P1^as4khW-8MQLS(DLAN|>7v}Y-u{O7uZTX^ZN zygQB=tcY{%GSvmP1-k|Sp~%Gi`E-n$&lQT#xo;ry723k6`DyGYOkUBr7fumGb+q3XUg&AG>z@A2v=m{G3pODy2H5(m<0|8o!pv1 zFI=zze}Va(Pmwy@u3l)pObe$^WC*eTT%wR8<*a**nb?lpJKM;+m*f2xsS^*CH#4}9v);@ySkCpv#+|!RLTIxN4mN5 zDpkrwCLNt62UFUmu5KfVna1S#aGPo8yaoBn1V7|4M$LR(7)rn|xHeq5h=R?u-v$Qd zLh{k@iSm)!rXP1EDoJ^Zq{Ru&B{FBX52lC%Q`)f$(P|c>K%s>W&LrM;Ev87APsV(( zcjQ`;m0FkMe^RfZ>aUxLnN!E#?JJ6`oivdUnfCCtG{F1@YDV;E9f5 zs*ZAn$n$-RDo+Xoc3*a?C@ot>&vH{^kqzx+bYLk85G^!X58 zZR~{2cUeQ~Br6F`Ec3}KD!h<~=c)kvc_KjD&@H$q7_qvRT5Cvm7GYHA&b1qRHnvM| zVZ3xxD}G5R|P_LHojGfyw`TNRrd9V{+8|-e1lS!~AtG@gPxwzHgsz=p7NcYl`UB%5q@kJyE zXcU|nslofTSvuxc)OJ0%} zoE&c1oTq1|hiqS2DqViK^H{sIAMrNvdnsrUSGi1wTzn!$dNr|R?LA^K1sCX$T2#5i zZ4m{x@wC7&rK?*_lkV&BwS0#MoZrj5p!V|yb&2a$(jR=e?HC879E5j#ilBI&&BL9T zV_9x3{;4A5u^Z=G>G4t=Uj|ot9N#bcEJE6A*IptbrSW#M7*}i_&W?Z~mtyRcwvad! zLU9{`ga}ePoe7}!v>{?Dj1;6Z9A?Amxke+=n~Kq>cZ6jCWJ(?$W*XpW zJO)V^e;f}AIwH(C#w1{4e#6VFF3-M!0K^7wtPQ)W)V~PSePj#1Zmt*liD7NJj5e%#fwN4O7TJ=^V3|^+TJi0Y5YTDtm^aH_ zJIT5+^qkj_I95)qFwz}oq=k5sMJBWws*IK>xV94T?TQLEJ?UUbsEif`+jA)dtbr)F z9HuE$S41B`nu*P{fdYFDhc>{CKoFW(%$32vac(29$XiGsB{86Rs0OC|fB%ORMc9I7 zg3)EXrhY1lKgm%R;$<1<%?g;ck zQG~&&HbmlA9trQXX`+u7B-%s&B$d3j<{V|n#UhqL_HFjy%XF!MO zCL_gSRE=fSoCpL&Pf6PrYK^;6f4Qbrs*?sa9h>Z>3fozzL6 z$3>4T2v^|B+urspS=~x}qN2rtx~2iyrOeyYVY9xIXKin$c2J} zNc=9Jjq!S0WL0xTT@8RcNqm^aiIZll1 zthwX`_6T2j%thJy*t?fyG6J~G=H!wVQU_Byd{IfDo6b9o(l@16yZpJnxz?WFELx8; z;D2m}>epWS|BHRMu~1F0M!8K|5>T!{t5F4}L0R2ve|>2n|Fw&feM)YGv}9YkBMk%+ ztUa1iq|c(?Sk6Lj?vFbHOw(0RdhSWzr+*MC-30jOe6&bFe>&bUyrXO9`X=aT7AL2XFS-GG>k&_#l zlSOTowT@4Ujbl1|H$)55NL@>}^rmg;ElAvBR7kfXPFC22vaL*ki~f8V%YZ_t@Z=Es z@F~|S9p)r_0p&5drXiKyG!jG)KA9?67}Hs!aksQPjr5~Sc?;Ccr3H#Ux7vj4>^Owx z{p@rZccv_i5l`RgE?@3_Yc)wgFV@As(X*uJ&tvYWb!>zB#=fsf7SMf2UCTH8=EZ}K zj)Pr>8yZk{>Cef1T(O<*W7?aY|IISF*v2o%e?!RkbrSri$p$%^145KvqPV9?ZXi(Qy61!tC zjlSM!h{qUb73SLol=$jTL)%Py%i(pD0UJXb;5C}6cp8c`8q0HK!!V|)&UD`DWAD@q zf6TXI<>mHfIa+gWtST8_o1RlR*q*MQm?dy19~4j5GF2ZP4KTBCR~!Yq~6FwA`6anhkCorZ$L7#+S9IHKYIU}=8&`j^M`B4GYfRPc|zPD17X#Q|L z?idWndk$8s)wR|~XGLebKN@oe>&+|RsnI!|@>8+HEp%ta>oB-H+dz<%*Qnoms0a_) z4|e$F+rYoYkm?1wd@_%RXimC24@6Va17`WEkJY#Y{LDuBrECZZ{)H;a?g z*f&*KvqinRPp2+civTjPby=rohkN51_CrRNH`>S}XoqZ_ODKS9ua~JV9CKJR zb+J$sgO;(L#G3v!Az=SC9KeZxpPmaZ8gb;f`A&B&ML zHq0(p+geGlFufHmq?F!?^N|gKv!&-n+J*3q+=4N?z^RL!;bZpS(EGgn-rOf>U(R&E z{U)+xyy3lZ4D@DrZyKaMG`4FSbA|-wjVWP1`ofufAO_smPYGu$&x=Sjlf>SMH_)u) zLUv?N?1&aoOThBke5J8#%oc9{kSQ%_lVLqFwVWtqO_lhXn2(_XH&DBp4Mhs>Y;KzW z*G)=#(63!f2iFNNZD=u$=81{`QC9df`he)-+9Ot$;Tk%?;=NZIJD|jMey_U!y-$4C zG4!?pTYoWR(+5W&H%g_YKu7%p@H8-Dw=J*)q)F9ZhFsh;==nLGaJU`5Z*=cOY-R7| zUucmaiZA^UTM7f{X~X{5(z$l~n*8cSvp%c@R)FI2wEjVuzo*IbL!jeZ!t*X~rLeHJ4L#8x03yen zDuE_>D;1(dsAk^>URuhU67tkit83xuYIdu}-iapGRK?(2O1pl;2MuU|LN$9+`p4ke z>fVBHP3$`}o(%IoU&8dq^Yk@JVdy|i+_Q6$h z6ZI++Z$vFsYN zZhkrAxr70q^oMRts|{+E8E0wg;c1%;QjjGShNP2y_cbgxfG?Q+VJj#h%v81 z7WC!biA!6w<5(bx_D-Jqc8QI)6zTc>_C=rHec>Em`}>PtO~5_S4N1Yt)odyh;r{_T z_UJ@X2y}x_ZA|xqn^Prd7nmOZudy-W*^B!!oxPM`rF5nfl6@5}1EVytMR%WY_Wa-- zdYwieYJ4{4qH%OGa);M&#z(f~(VrFv@^-(9tRx<_VAMg{J9bxby-w|;lBtH>+zFkD zJV@);nKXK0aiGxg_UT%`P?^n=%gC@F8TZcE^Ame;rxSaX z>7g&d6LoK@H+}DTcP69w&*UfwDg7MBuCKlZEYi!CV8pXEhgp^{V3+Eq>p)-_zMEa4 zl~!zv1No-C6R|gYIdchfB{hWrVL@LXK94=;c|>E-{H;i-p8AGhT((F-ddn&{8_QbsJPw zZ1cJHAhf75Y)740@yxc0mtfYZTi-5$OE4p?n`^%%*!C!Spx)2_$EBTiCRzPgdn{wd zijCHdudw~iP4R4T7npf`g*XPwvYjOaRT$SK(z0xi{e4|@kf9-i))d)2vkg8pSrJpF z%tdbKD=lT1xE^J^CO-=7?-=@B#@^_nQJD($q6n{chGYK6s0(F z;?&bA*&%>X&9mA1ciztrS+8w=(rUa%T)FdpVWuU{11sVw^p7FP6bJs>GZOm|)f^Jb-qCTlyxNy{X-IbdXM!_)a>6UVS6 zZMr|Fs+KrbE_@_O=Y=qjm*4!bCCkoPyJE$}>k039$tiCJ7&HY_ff`*`yjh9H`cR{%=X0I`zZ#=y&tB7m1BI1bt$k;(ZUB6TVHyn zTw#E}fy|D?9-Je14^CGF>Xd$zt9sl*kgk?a*Z0MO0*wwFXSm1gogH?+gMd%36jL<9_ zCzHT^(`G|Y`9f9=>P;D^ej4O3>Qn<$C2f)rLI2hWO<^D$!D+FfA7K96hvUC)kNtxb zt*nl$miew5H*ob9Z2lbcJG$tA<|opT$Q$6rjL@vCyZ@<3w@5ejIo?}Vy;!Nupusc( zvEcFOd2P;mPO@Wz1r3I6p0IKSv+kAs-eFD~Jf>#Ep~X1P#Q6Zv4R-s5xPpo3?A_oL z;VW2zY2;x9D=)Ze1F)8TjC<3J7Y`=K|6x|r=ixEzl=W%FlON^}s|LfmQ8d_e*3FUX zv*Zkpo`!Hb+xxFEyZuvq1?%rcu{%}6><4Mszu4xFBf;QWo)MpOeW*cu79IzOPO@cS z40vZMrn59H-ebFeF~%(*V+?im+K<(ycLWE@`HAQ(%z;?#=?Z~^k1-FqvJ~GKuoEm`f(mNnm1o4S zmeNCwxZkbCjoW~BesO=%<2DWKZMJDw?Dt|LTAg~NSaa>=%-;0B##{#G9`}UK5%kkh zQyQ&5)MpTS$7)T34rngejtsO)=O>a}*~oP?!__Fw7z9w0QTksVM;@B@AhnICd06zg z^q}5YsqCEjny#KeOKRPzJO<`#zl%=64L+E#PW#`KN4hUBx1wP|>Z4w|JXM*J*RVJS zPQehq3F;Z~-T@N`6?jc!?wNRS;5BE$=m7C7=>N~Na&VQ>HdY{`4KX*-A4vmO?SGEK zI-P>g@L@^Ld{GYX5zR_bS36)$Blez zM7lGQw$MqBJU_7P5$$r>DM3Gx<;Sx5!+l3b*y#<S#BrhXsuFrDcbq#o3ZKG!iA z?hy%6^d!nu;T{U@I>{-b{ZOz!lmWc}On2!Q16nZ0bbHW5`Z%%0p|tVbB|=?t#l+`H z>fRDhcvRU{*(1T7`^v2j4ATU6OnHC1^qcbA6$=2{g37BR@a$vB%y3?c7Jz8(&;}u) z9U%SYoyIoK$ehQHA;SUxun^b_?!6_^d@WdU4gtv|UVuXW8EuAXuWguT`C8KK6U}#k zacal@WgH*!v!_kir$2W-Yx>PwvAA82!U`8@5ms<(MxHYXxY@o%6V5 ziZe8hD=UMbd2*mi?>Zji=qJot)}5Q_#9&C7eR0yQ`fL6c4^M9`1U%ikhG?kkOb6_D zjIkd~m2_Go#Cwt~x7FF;?g)?2-qvw+rNjUeK|ba^F&z!%FOz$EfJMD_jV>G6KgpQoq(qn0dH@7@uAMyb5j zPArw{P30m?e4=|pc?tnjP_a>sh;Jca)zuct*R%-y$ZF5soz@;!Q&C&LU)nvW%B7`f zMf@3_l?#a?rG(caEa-uKN|86+Sv+fg*RsW%Y4Y0~$;XRHB8nkV37^HGJ4CXAbuo5K z247l`V2vynz^*4)>NX%Yp%tob4I882^zcX$7xqEHQ(=3)0v7!^ThfBA+1XngsGlPe zC7dmi4y(F%5{6Mr^*sX-QgI=S8(q@6d=X!|*w0l;HwllC(zm9bHROp`RJ>Y)$K(JQ zq4S(yxd@>xK~eD<-U!#Ld=Nk$#C#bJDPzXKj9(sU)?6b&`AT^xBQk zT>0b2z5GV02F3S0R}}fV)s#JVkDrI{q^UTq0L+q=ODkT{ub706EL*zLTCrA14Tm-NLt#`Lyw#_|w?)M4%8t&@9E3 zP+=((%Z;KV1CWAY_ET`?^2N)^Fl2a}-C7YMUIbVXg!sL93DX)`#HY0z0>QNiNU49u zIpRqtnu=@N#`u#!BSHLp1>W{z%6uiWb6h0Qj4uzPN05^NtCbygC`q{lsm?~+Kgfeo z(7${Exdy>v?H=P}%}vuws^dN`hWJ{e%|GO?=Df8u5oB%>h!7HmK*FpK!k%FH4%;zb z?Xj6EUXfsJcTc6&rln*>o;5A;k@#V^uHm7rC zjMBwC!UD{;3$~=U?$;JPv~npXHE(mY#wVdeLf&Doq{E&gqCuNTR9I6Lq=>l)qokN5 z#YN~=b7Kik>U6XtGJ%u;wIc94oUkb%IP?dIx7&>3pW!)~(Z%bXH^z%Oi?w<6ig{{| zwH8b9n9GaLj=3f5qB@_RWCnpgca}fPsy=VQ+7wb_M#7LfrQO;BT=l1Pb9TL$k&uwA z6IDcEJH{GM!7k(xnR0D#R43^p$PXC1G}tx0b}jd${22x|0RNeYPjS1AZ-%94`!g>1 zc6lI~HHbEUgg61u6PaPKP6`Cd#g8Bk0P-=MN5;d*L8#=CR%97p9KVh4FHg4;$J=dx z%LCvKY@T#hD0Q_}Rwql$WFV;6UynF1C+Uc&z7|{i?Lm+3o%(ef0JudfsxPb?s*Yd! z)qF3Ew~el|nrg8MbMsRO06hm&x;c)pge7xM8IODsXX5`Jb|jxn#hzmd;uGZ4)ZjoW z8devCU%~Mm)>6xWaFswd0v5Y=quLhvlic)WE7B#2Zc0)sNlAs=U@Y!~gJf9=yQC(G z>Oc@m;bcI(83;fg(`aI?9JH(c`^BojZY)R^t%EEFqbZ^o^vzvW6kmI;JHZdSaZTi z5I6E}@K~6WY)GjL*m*9>t?DX}1NFQ#(Is~45DE-Lq3O>+AJTq*L;0+=F~ z$Y=$CCCZao^5_&v0DKb}x2Ob7pbex@uz;z9p=pRREC4axFgQIZ2GJ$p`Lu!bdC;43 zt#h*-7ciUB6dZz23@qRp0(NLosXtJV3?Ep;$EMJxphS-Uc@f4{p%vL=4dVoWr_Bc! zyS0+!&sdYNr?bA;!RYbF(>h$>Fa;Z8vm3Au3VKm-#UyU%HFd8*dY8`RnWz1mPD_Gp#v$ezpe)7din%Mq9U_@xm&3 z=ZWUL2JI2-ME-aFj+gH@aH~PBm$HcLmkPQc+4oH`6HJCyA6w71xT`ON?mz4Mrtmw( zVW;sc#`oae=5BX9wyn0S_Biy2)kR=cRM9HkRKi>z*)X|B=@f?lDw#s6lq;SS0pJ@s zYlG8k9fA?-)6A5$v~^cLn;!o&_~;#aWA~*Ee%Cwwce%chf0veTZm?;X{A59u&7-|% zbh1A;e%dBYAH2!>bnwcnyT|b}Z`htDZa;jpuwZnI{GcZ4WySiUppqc=lPPm@mUZ*YUd-5 z&$cHg>;JCXU0`7QPGBJS^7Q1tUcjzjPZNNY|SKDpeJ| zXP<5=yI(q-bH)4BgDCeUvgH;h>Mx}YERDbL+4^2KBiA)7EQ~q$%5bIck@W4pfBqQc zm<5ZDA872Da6R!%%UN{eI(Nfp=!l8Z3KBrcYe%k5l*_ZdY^TOHO5(ZA{z(-a5U_ zCVGp5E*yGwbm^0;=05%T&e~2j&;GQqu2oX~na#Cd_{W*Lk&6~lThA2Bs$5I2Vuweq zY&oiZ`Ld2-*W{wwb*_mT1MG;A|_t(^IOEdB6 z$8K4CX1ZNgOM|GPVTkW7a!It>+;7uB3#d>?l*P&H1dQ;k!8SpRg<_q9F> z2R3$J-lI2K@bQ32!9&I9ve#xG8B%AUJ(C4K&3lp0qYx%^`0_x*cI|i8%$1JY-`H!8 zU@pntRp*%qFMVNnj{A5Wr(JbcNj-IcrL;V97ww5)ZDS`nta!BC0f12$^`4EAmkfYp z9WQxF19T2z4Tv%VeUAt;zXv;)^BYK(UWHh5x7Qlb<$ytI?M z)^XM3_gjyt*Mu*48~r)UL1Ux^m*Rxde#|%UbDM#QXIaxJqR&bsEUp|+spLlC?_!wG zOIiU1Dep~d$02LEhc^|0UE|LoKIhB(>JqT19DoUr4?w!(*c2)txmOc~JYT$6lGwTE zi<5Ntd6q6|@DD|a;xztlYhYQj+n%~I#Vk2kMXM6dbY4vjUDq%(tjbUuOS7W{<(1z#W%8#gpTvAE8%QWB);Z-98le{(0pv?_Nym zQW@EDsc=dVkJ1%*yp?rUtGvh~F@&+|R_WjP3pK1&p&8lEhUx$T74qF^31n3<>>DOV z=5+h)_i@T`?;O80%|@lYn;fdE-tql_Uc&pMgwARcM{U`LYhHqw`fig|pC--UT9jLG z#%b#$lQSC6ESE;f%+uB|ejJ?rbWdXcQ7^UprFR;1G^P5Y#V|NX1YEc?MX%@35Q z>-P>n+uWO`lbCX2?}-}v*dimfWjQOa9VEP(KjYqEzK8!>Kfk|Eez@-ir6R%nm!n0; z{nghEEW4gYU2j&|6P4NWsfW#^eNg_pkG{RR*rjg0r}SwXwnyB8VY>2ZldYq(#;uz} zc+b`fM>;9F52IcwE@|o={s(z>ardm%onG}XmdRUBtjd`dee=9|BXCQIb{(0m7&KnY z>3nu{+0yw@+_yFzq&;i(bRN1%ZChQl&$4&XRh>i8u>EOQq1=0&rC)rKBdWsfI&^lg zGgIFktAD7X`rwOc_S=wc{cv1UZJ}b+fx9<8_KX&$kZw9E)SD0Y--yk;>$)BrrvWCy z2T4z4@{u*61AF#4B)xhO6-pEBcC{pJnqNOVyd?UsS%OOEFwZqLB9+Yiwf=Kb&YzpU zroNk&@4mCxNDW53ZhaGWcq3h>>g`SK@9JNBT3+0ScFlEQ*S;C&oDU_tA8M+y?XPllye9f03`%;kuE0XvO&rAJZ>cF-GiY|-0eSt| z_SA|mcSde(JxPi+5xxJTXOJ*jb=O8O@nG>GGYht4uiO^{2mz0HQ($W=?sVXFTS-%1 z;tbRl``sAoyL90E_g%38rK}$}dtbkNxr*oAQRKXg{5Xtb6m2PanmBch9eP@Q;KwHg zucMFFPG>$)Ibs0QwGrd+S#~Yund41-};iPX*jCY!^8&Lb6 z79IYkZ~>DSIX3$G<0rk*yP+W?M##$R-yq&F9@HZ9*sC$h54sm{6@GfC0Ce35t%krPI6`JOo-o(vkJnJ0k zU3RrQV&2LZLBT?A4Uf>VEz0lW=t#GC^3Tshw=K51C*A9gC9_m*Ls#G3c6a}fUV@~5 zfo1KWR(ECCrID*FQ+s*mjiz$@y)oBr z_MP(l=J~BxLp}2Fl#J3^v%eQST>8lGr(VMCZzRmasNLk;yXSRl#Qv*d1(YQxKJ))h zd#4oo)8(1Jva1FPmtONP|XYO7tFLQ0?0AAvo`?P-J80*|ETSc2s zQ4f5&YjP}}Xe`_Rc2&RpdKqD1>4#q%yMsRpGKMHOdsOI5&7UHji)=#M;aUP(C4C!|KkUdjvyiq>QFY`ytGmqKalHsGuCw6>(nsUq3Pp>72+K@4AKAgWa z!p?ky&-l6NOHFSGM@$bW%{HByfE_i>Flwm}&O9rBWm~$mYMK6#{ef|(*vGEhy*-6z zD<%A7fkjf1ZMpEf*4IsYx0LM-IK9LFB40;t<(U+D@fY>vfG@fOop(3yye_1_vd7xS z1;=d7>e!s3u2b}L-ImtK%{G-~Q3r1K94(jq7RVE5*2*5doa=8|vnECHsE9Lo#Pm(- zzWcma8TRI>i!1{^T{!gaokQ z+xbt^@vr#fYMR9iI&|ioufJ8d2+YYH|+cocN{r5!%%ROZq@H8mc}s@T3xTf z96$fRIUe}K8T6O`=J>OGxx*dTN{C#n*cYAl|HF^dviW)>>D4HZoWG$K*{*)p`?Bx2aR)hUL?bjCmsatBa!b`W4%Z>Zyp!71@MK; zCGR03il1N?PEIrvE~cH61d*Dz+g90lQSM6Ck2>MP@6#(DTQEO*AB%A}swBvE=P89& zZMS<$4_9xsO)8dZ(o!-#KvVN0&o8QGhbS##A%}^|bSD%2?qUN$dZoMpL+V*vQ0O!P zc4hicJ9m2axan%m*3UH?#$orH@;O_p3v8_;3&G=;iePTg$l|dB|A_Ihhb9h@2#rfG!Byr7<3CH^*!7 zaOpCDb}HGG^hx^D zdj|m=9<213>$&)gOY&HFw;(*@A-m(?x@oQjV_e2ecqZW~-9f+QL0q22`*GO@;|4tq zzx}|Ty%xhluKxY~W@h}lQD=NzE9@|%5vz?Rw(e(-ef${y<$r67)4AUc|AS=2$AL!w zKU|cS(7Hiw{#*OHK@s7Kp^L1zQl>PCuaMtJmy(cwiUgeXm6C^D9;|?3T=u4NOvMK` z;s8f?2@ywvNHL}o>Knl(kd?+9ZP|2b64%gy!AIsZ?FQq!bMnv_n*jn}W+26_96~H@ zh>T(^L>6$Bz{G9^M@k;3)D$T@zI?Ecdjgw#2& zG}DQMS*Ue9)v4uz-*)WH3}*79c(;#3JXM2LGI=LGE^j2J^fi?}KJ!_9{L}Zo;Ih_SNuPS6kyT(eanNeE@slk z#s1e${J84A`ppL$qtcxIaOY(So<5(rW$5V!+x2?xiBqJ+B{4yJ*&V{ktqvVq59dpn zNJ%{~DGCl)$=P*~aNWcB650HRrxge9>weu|aB06qS?ausk>W3O)9k(3EqBdY)xbZr zpu>^;!P-C`b=nbeOWrRKB_;H5kwYEZ+nJoRgIu~F#0}6+XVFkUHxAKTjXK3B3_L7g zfG^7d2Tn0u5$V|J2}SxA@TX%K#U%OvoSdV!;pAgn2b`2T7n~e=Iult!j=Yr}gtds! zWd{X22A=|fD&OqJ)*}9b+JL>nfm5@{zgk#H{3QF&i~j?-sc=9sA4PG*p~1{dE5$~{q3}^r8+iyPN+U!R za4NDvODjt&MJp#XD>L&E$IQ$`O-=ju|9!vn{onUJ>#TLo*(~;AEjIAG_wOFA>$-1# zzyIA2tRlO4xB(Cd06@S8@cSd+0%)kK!_?I@U@(}briPX_e5Lk^71{=C^mO4yh9<^F zhDfBT#TJyQ*(P%&61@|%$r_8p;Y?7r1pGFKEjGAq|N0V$rlzL$3T?#7m56QYk?Xhp ze?ETK0jo9CdZ2ru5Hmn!H3YgE^7|!V0tKrlF~|0=%GV6`%rv zLRD0uYHF&g;MGas`+(|dwKeOp&gy!ULohQMd|PT(p@#XcyY>419p5c%!ei1jwfF3U0 zxOnL@D?5jK^;&M;bzYH3EGaH2Ei1owzq0DV!|Iwx&tJT3Xp}WIzmj)$z3qPY{zK0{ z1B$_+;gQj?FH=9Jl{2$*Kj#zXcevq2tZ*$16%C0h;&6e;9v0?bFVoO~QB)Dz&$K9&L=M|*q-NwU zf&-KZS2z>A77kQ$4bec0LBocvs48o17PKrV@ivwCs?+}H2J)Tzu{W_FkV7{=Ivk0ojKCis^FA&+GAv#+ zOs)G>y4g8z$X8;i8~lDpI>E|zv%2GGNf<5aRl^pmk3CuJxKaZRlsSCwHP=+0E;+%c zGEcl7lL93KQnNs6~C!(WW2~ehPsZ zi5T1!!0GB10ve!rspPOiE4fUK$vkC%Y%GA|!sVJ;EM z82*VJlU^|Yiejp0+_%*huAy6DM8K@v>f=lHiGtYCe4sr!4;?z6re`q15Dp9qXMVP1 zjWrS8QB5ysWv+#mj(b$HPh(!er&%sa1x}mVG~Qs)s{2DiNV&RmzTFho)C>|eq{|X9MU`tY&jVm0e9bJ%XQF4gqM&R)x{!CHo zg(y*#B^9VeBM4LsL>8=W2mX+FIaWhiktSN+xUSV{_&>b@{S)&WFuyXgl>YfQ;6`#i zv+v?m#wo}zy)4#y9q(Yzl#1=q=ood<_LF$~f!fX%cj4x#j|B|cLq#oBgH#f1xH@CG zR&DR=C`OmBdg9T_!HiL>$EZ-NtQ*?~FY@Qg{p+4gWjudym@Ki2oAhdRZ;S~@J0qU-A`(U$ zwi+B5D@A%3#snMwP41Hz!$7fu~xh_jF2>TDWEeGW8Z#u z+=IBHn;XLz;9(fLQo!D6EAD#p4DLEgvc;w~#V`zw#|@u63$P-cCp?Hcc=LG*TOmcw z_jmA9`0W5=>T{RBapi#eHVpP1vK{J*keKCs7eiP&SWkDV^%`~u4~(NeS+t=q**dwd6l9P`cU8c0T5epq{}u?+vWu4dRvvN#a_k2 z?qAn>4{M2NC)k;oG8X0TWwS74OHk(OR;S>uYYs_P>@rS%^tZ4#V{4%$+sS?!gk$8u zMsL5XwuBU|`xt|JoF_p$8>5xu?*!2~9GZ*uQ!z~0QNF5g5wevSjuoa7~MG|;e zq!fVrMHQ=1JSSWMfZ`0+fuyLST$6lLJvfEbod-z9lwyf+mwyL|4ESJN)mZ?x0Ujx+ zGz3C1x@1Q#ji_cIF$pG&G2_HCoQQ>QFAZn6VqHwitQ9y}*GA>|ME$Ay_ z9xYp=0rJ6vWS`nGTB{m>p74?3mjly*BwaG46Q>=}DaB=skoO8;WS`UYjF5J1SJfPO zX6MsTdKW{ku;3qM-uX<{4d5gF6~$^O6dtbO%dj72)12#v^I!;Dv9KKj_{WP1C1NaE zgo_hGaW%zq7GUGqDX}C;41{V6?RIiX7Fc{&dysr3n@#nk&~6Tm6s*2^gFypI&Lp7v zOMY<@3TAavmJE#G1**S+LHd*5z@5b3z-ZHG#R`*um7d>#>cnni8&*P2BX)PUgAP|h zxLNb&2HbG#Zy+d;XYBax*gnyKFnf!Y$lTVPU-QtE_nNoGz5e#K;Z@GJ{yw*&AmBjY z5IwsHdpP5EG`l0RuwJ@vio@6)`w}y`-<);4srqjitM#g{W7W;mA2~J+k*{oRi{F@U zCU3GnTO8R}oaMe<^VSD>l@=QA=U1Hbj)jR@#Z8b*dVux6>%uiw*!qU8$(8Zr&SC&2 zMZk6PS`Z2xkdH|kMaak@j5wkCoCLnQW$ObpApSK*jav0tx*00qK8Y`eMoAVrh%c%( zY#k#Z{Y-`a1F7z+xCmA#8U|t=6;5R`GT;tmj6_z)ocR_7e5KlWl3a+eA2=O;J6+C> z6xjIzz6_rt?i2|xR8=6%r$oPIBq+kgArBB!^wLQ1u>N?VstlLVh&fH;*^CocN+#G8 z*Y%e_3owj!5E~`R<$mZ@6wiWaAATp$fQk-9PY@g6{{HQzvS>#xrH82%+E>XR$i)FV zl}X;{F#$Jk!fO}Hu7K>M17^WEQ8%Y=sf;E#BZL~urX)z?3*3qTXTTU@+@)?(HyyQ+B;NF-gzxw zM3XT$ESBKN?YqeCyt6|oR2K-k@?C=1e3e%OYlyeIqyl!s#}-+li}w)%eeH_%9UqWj&~ zb9Dw;E3n^h_k&@p29j5h$NgX*evoFTl=?@cMZ~9HrM}VikGP!ThZRkGuf*hLG!IT4 zP+o0wd~$0y&rj3x{>cp})4-fpv>&tMfuCL*ACvs~qdf5GC60@L9(2TD_|{IEU$gwi z`%5P&+{p2ujNtpI;L@4D?WrsRH8B8!n{w}95uzG4qq?uJZ>Myp%h}Tn)>oQ5{9#{WxUE$N6U;s7Zs{b zu*kt7T46ikwbNwvs4)3kZV?K2TrV-s&KE+^a=Z%uk2yxCw z)-F4wHmB?BuiDx&ag6&oF})NC8S@(95AZ#V81mKWUM{EU-Rbo*g=eQNj-M2@N~ff1 z3yeT_cjR|;juV++4XrH$W4UEB>K_lb{WK{UkJI)C2pBv6xgw^XWUIj}2(H02juwp* zHIWA!n%o__ok{?&>3aVGdJcu9s+O0*i6D;dPqaFiIl;8;K278_WBaPWK|OU%ucUv zdr&T*Nau}_u2Q01SyF2t=()9Rc>;{(d&q_E7{GpOc``aN7iv~|t$94;2SbpL9`#zs zyXw6=AfQal%C;E3mQ!ULnA5N?k9pEmD}Qs|3{>i^ZKOG-yczC5U!c(UzNiTdE`_Eg z7iE4#TSQdUhI|REw(POOIIPG@W!>t(;IRAeEmpTK>4rq!9dm8sQi8ed&Z=xm z*DTyXgMd~_CP)?J2rau*=hmp%1{H0ZXB(Lun*K1P9kAnhCF7ogj3vxeGKy}119_4` ze*u3+JVDfQ^Tl06G;DQfCycI2S7j@O5yr|S#$vn9O6_p|A3{i%d~l?6!Eeu<(UqpF z6rXWSx*#pLbi|_FhPYxWsvcD_MDn+ha|g9`xIia6l}y4#sHbOxSQ?q$DMcU`4G;bm zJ16{3Wu6=1Ct(PWmoTSSs>TbU@Vo}hhy+F61Q~Z{^!xLd&~suHnw;g2pA&<5KOP=A zEkbtuDg|k+QRx>OqMA`D2*EdOt%BFO0+t}#1VgL~;3iT&OTe-LB#6ez>Sx=?_I#Z# z4f_mkvG1-$wX992y>I~eN>Y+f& zjiXj24&+F-*bZl3fbgu*X6f##13A0|DJs}y&9!Se6=2AR^SyWWF?C1Ws6O_TG#!wx zw`{g_vIFd%z%cJJLNxLQxBmL3S+bu@xM*sc8|Y_7te z*Z$+?&AZumY#oBA2JtM8$9|(|@%@OKvC?Z7wZ}YaH3#!l8xy-vym^5%^teqJccgku zQf#AwthOv%9m%{_3KHmjnzp6&w@I6QeX#d7m6Zi@tRCs{@_qFwM`Ql%9}N0UgOJtE=*ESjtS<;%PhH>45kAfV-dh?a_adBsfme@>OgIGQRlUTm!Ht z3*Tz@mP%B){b;sygMZ+6^oHy5V!2MVqm>6Ei^Ax%PUN(DZ^4w?q_zh<_FHe$fJPx^ zL_Aw4FU$ySahg~UN+3>#c*!`(FUcX_qZ*?|B!CK=?mV0eGiU%%KV~GzBy?N!E>K3{ z7Yuo(X7$dA_%wITB{X6(7_S^3BPuG)*opp#^a|9BFhR8=o1Q~*;ii*+yCs^V-js=M9;_})9xgvgJi&^K{d;>^VyI`-`@Z|Bic?7 zHQ^~c=~xL|Hiqhz=fKeG0Lu?j84MZiu7+M?dJ)p`_Lyky|svYF%S^T4! zS}}HduBH|s(&-v($70(M1KikpS$BTwd+pg40EDgQ6ha{YGfM1WucEa{xBj&TR<#xY zkn{$7wKvs1OdoglSZNM(C@3q5^P{&#htS4Zx(Jojzq6fN@x|YMoe^H? zt9xC~dhMI0rrtwwHjf86s&q`(Kg2p17mSyr+Wv zUj1$r?D-nIcdqy^5tj4>*_MV3a$~i*yz<-UEgyazGmHP?l+Re{!yn- zwlg>H-3gDu25vA%>608-F*TOu?!JusTMq_H%y;V>RN;0_bc_7yEw*jAy0f>BpwS@uZ(bMHm<6r&XFqaN-tfe2DJNgW&Z@AOyfnT zCmyDCj zERtH!7J93%T%*nKmv6SOrD)qNMVV&WxY_jO_T-&|%5z}^w5Z~*L_ZZntPBOOU2LZ^ zjY)92=O_l-0Bk+^MD8!Bk-^hJS`B{^Zy>S<8&n z41TSI@fZBoW78Wft$pDMxUVDWd_U+J!g<)Qf!v?8x;YU-k?tO5m&kJdAy5u%J2Q}} za|WkV2m6aj#R{nIBfMQi#BacosauTMhHayak)17b5Cp6@ptBNz2Z66}AQhsT9GFnV zj%_eq)viwcG^R1YguY{0C(09K@Mn2&K=<|c5=g5rE&}$k%Fc<5`@=Qj?bMCZvbEti63l#7$emYPoW-r#Ip0Ahb>jF7?U3%=!d8rpvk*!cv!?pWvoqCTQX`PLd}_f_a=Q3( zB?u!C1J^(y!x~QcC^agD5Xz!(f=bAk$$yU;(Z9&%9)8N3|KtFB&SxxRzlAxaF>D!E z21{QD9{-^FF8uOZ1M`?;W;5w3*x=)RRaehAx5NenU-ym3d*Xzhy_I;euIbCMgMRaS zGlJ`@!)-X}(s;g&m3s^Sp;ypOTU#TmeV;e8r-I(I18ny{Hobbxj*@n4ms>xpl{@&3 z5Kw-3-+3+D5-0WZ`fuWt8(hbB?@6;#HL`ln0k+@rvS=Q_9~V?Tj*`9BXEFzOHxxdP zl~Pu_y*yDzD0fa@-cB*;1PoL1NXYQC=1U`v+?}Bp5#>XzXUy{C{2Afldskgs8JuHQ zv&YvD-)fqeX_~n;elOmbs(Jm$y+(0p)vBTLAf=f7VcTp;`~a_VQX0k16Y3HB{qXQ2 zo`s8>4f)B8fe`Yu6M|v**_n{R%zRkzSz_tK@Sh?(H-d%SK`H8Ef?|m8U&Gwf49dsu z)OS(+m>;)VDbQGz)q3Q_Ikheo37 z8<1F3H#goWMFJS$>mB54oPt9Wr<`RGJiw*Rd|GUl&!>BQa|KjmN2JER16N7Nfh@+f z2Q$nhjOzOUgKv^W!``6LAj{aR0;Bj)I2Wrh>#qPi*?)baI)?NEYWAuT@*5n*jBr=P zLz7@^8}7tBht3M&tH69J+coH86C{un>UnlNYh>!dwR1RK|D?+fN^orL{uv?BO{2wN zfbVO7n1lDdZvZTH4UDLE0Zb>dj>qudOdT}@SShB*-0Vk67+wJ9TsSwTm)?Lr#HrND z`^7N?)t*^I*N&xHrxK?jiF^CE(4tX~^Ssk{vh(#5OElgZI zGj;41WzQNVDz2|&h3S!RMq~4y6DlB%m zs+luJxu4!$yFpK&_;jDw`p6un?~u(li7upVRex|U&W2xUyX1YxcYABd)1CDFkNEs7 z-tftS&$Y2%X)Rat@6VLx|17+rzrt3fu&EVmVE%TeZJ7nlZ|3&dS0jP?m);*02q_uS zr*B`c5gOkblqs# zqb|S`e2vO_AFuht_KPd_A`izYrK@xFgpSH|iIMkKd1)3Y#LrnY z-pvk+8<8`0N{mM5L=U1CKu-UJ=PYm^&Q;BjTBa;?0r9iSqTJNBe@F(Pi2XH%jQxy* zrjw)qMxmZl;Y=DM zfDWaPxsi;fom*iQK9x_0e4;k1I}^?$6a8#Yo=)@=g4#v<&Cb-Ha_u;1Ev6q?5DghQ zY3*34jc;zGy4iHX;CTvIl`W|ja!9jocATu!XD_Hs)C!4o$_-Xs(aS+MHQJBd6>4jb z)S^*lslXt^Ka!1E3Y*&v6Dc_QBl1*X7YS_vQr=SHdT6AWL#9G_NpvW+nw zu;0L2ciomO*5SLlhf+N`c2mw4a|XqWhPzjZ(_7Y$;GSf8q;{ujBy%t1;;Lta>K&dv z>8?yWc``R12XTi=O+eHOW?S80tTS@aUx0>S8%!lL;#pxhMJf43MMLHPLtpp&Z{quR zx~{^w5Bd}J=i%`-{dKRqjtZ$~CuOy}e()12j{G@5H?a^hk~VJp)^C=>`>S)N{?@VP z?xayXIYDjfU$(} zP!215|v3NyOv5}}$PJgfla+*YtH)2x8?M{{#h$NJdX-=px*&!BdVm8aQsA|ziJ z%PMKM%DLimxTLTCS&g8=J%STCKD)9Wn14L26sgiYn+92cQn+a+pTk(}WRB!a5|PR} z5u(+$Ma{In?VX&P)R&kjPZm_x>e%$97B-x1rLLghosz$`MltUBuEe}Mb?>R~*-low2+hB!4avt9a5-&arWoqmIU_7FfEkSYe4%C1 zoJ06)(E69AN244B7)6vNkAzG)uX5*NcaB4B2|pR$!P$nLjaMjXBqWBF=?n zoai>@jDdxAyDeUf6+>LN^!XZLoRQgtM}E*K;|sa9TYA&zEZ`JgREn!sn6$E!RhaA8 zJo~s3NYna+R*|p@uq2x4`cV{i8h$!t&q0*TlURAe4WTQ-?tFRYw1UN?&p7@ z^0T8w+1|9S9J{iY+7}%<9`i2oh&~&QBa;37qgJxmcN=gyygeous_(qMS?YKYrDd<% zWLKwGEBU2@k6aG?j=3q91bD46gij(GqCRc`6uctcu{}cbs1aW2w?DR`o5mq?B1VeUsK$#FI_#uuk!(?<4+Z9yxSx%Eub zih!S;^z?C(qq4eLSts1Zsa-15Af)6Y6cuC6W5o252%i0PH=EAkQ+_IEB!^N_Rj|}v zLu_Ue=4_ioCol>t6FUsEr-`lKE--a;fD+?;H}CaERBR_sM^n=n@rwk3vIPM~>?bE1sS&&OzrRTCuUgwgR{mMR)Q!Jif8G&?N#BfrTGGC~-j z3h|izSC{sn0%*6jU5=%*b-xP%1@fUKS2>(1fXHx9PR#s58?K70gMV|(6I(gMa1l@* zEX;I9!SjkJGdXWB0ELY*#4vLehao?lX5abj(%>+A3EPvpvJHWk)UD{@y#Eq~Z>6<$ zGoj<`#C95!-WX-v-wyfZln)qu*@L@i0p>LeU8#{ab^-3hvkkl7J&$9poru2-7>D4}Hxd2;A{ zKN3UhPOplMXKHIAg-_wsOcM91LzF z;b#g{)#(zeJlT?5ndg*`H{nkIltXNrntrX^N|AX{n#Von&cGWEeNNn%>Shg!9?a?P ztB+FMZ)#u*hMbjVxlo5BJO8ZKm5@W?n4~hzHlMEQMxR7Ee}suC+#7*5H(H5{t0$EG zZ1i=Q8Q{_0RdfaDr)Sdyu;|Ej_d%71({=vLh*TAa@=5fcdI!Ib5UyT^YlJMsD#iW0 zZNBhTwi85Xf|CS_P>OzDlo*Ui*8iRUnTNgB=xW+})`@IW532=?rr$5hb)I(BG#iY0 zG&1}Fhe$!!X;3bb!V12_FJ|$l$B2;ecPyU|uyHqfgDN{UBZ6q0@qN!SY2CB`r%w3) zo~C(eGc)-1>`6QR+sGP&@xb9nZCb}hZ=aDCzT4wqGPX~zxte!v--^S6s-RZa$Q^;J zLtTR2(`CVT>ZbkfY=3QjTNNhVX;kiAWwn3px(xk-jYs*-CwDif8g_D8H2T*>QzUEs zoPh!xBog1TXtAb*Ibpx*XqeGW2 z1~%?BZS7Gsc(406kXwwoHE+=3Je(92sbEpF>td3=99EXXrvp&i#a?2c3&jT?UwNzb z17Eh5@Tj}#W&CLn`0@)F{;&#TC6meM)C=#=lBs7Z%xxt`Nl^}9_wrM#z5lEO&F#$G zoY+LFvKH!~oFE}UI6a(IBUB>*U_11pP?1VR)wHOtZi53QX;hmQt}H|~mLX#zBP<_q@h82Jfic}ff&XHnj< zd8()fXF-3_;V<~Jkqhl-QH0VvS}@xsbaTJ5c<{n-ZaSy0a__3u*&Ypu=2;}ob7ll5 zx5*~`213&^2i7uSBc$Wt)Q+N?N&8j0BDGoeQyz9$?`MCZ%Ahna@S-yJK3u}w{)+t_aI z8p1^hfT}nZ96AEOh>WmxXnz5NIx4k+Pcla6`!OOOZ!`)mU_f9#v?a8n#6X0HIY|IR zm$g?ap>4RGw2a2|u}&P^sO=}^<6YrQ1Hfm~vRYIiO(;31*3O^9P0!-w34#6`*~zXv ziwK0IzLi!lLRct4K@BQJD96cmKh7GJ*^O{)h!8v;H$YPoFc9-m4BlJ<;A5~X!kIRO zyD!1USL2PQfuCJ=l+lH)UV@dF@ptCkMS|hJ_eX{g)^#86`MM(iI;#7A+##b6(Q6XT z!y@kDx=kZc`mLU~n_g%E<)_M8N3QGsv-h=u_ldId7lFqD$`dcOhgCznZ$1zmPhX0% ztciS-?Q_eUR8zY$d}{Km$Tj_2E6+O`@;oj2@v)n&W~lMK#Agz<4a;NtHyVQy|F~1R za07PWWc$=iasJ7WDZ}oN)1KyPp(R}%8*WWKK!kNUxSkwJ3T2H_Q%@bR2S3k^a3vyd2+E{+T!x`=Nk_}SINg*d zt$h0~_vS8AdsnkG_#x_aTlmHseJ@Id?!Zv=wQWV8vcJcEcDb41_aIkXz0mdAY=_pJ zQ-!w)N7{@q|1>#myn0G{{O+fYoVwfW@m7-5V7#a1(#*!Dsm14YjOjh$?|f8Vo0xp< zzT4QDI2y6Z$n`;Sv-y;GpxmYO;~O^R{i+0+|I5*n0H^W?H0!jSqzc*hp(yspo z&i$LgKk<6;Q+A)ZtXXgCFryD$6aD7z=JIb(1jk1A^_x9rrJmn+$nW0py{squ4xIn< zb-lTy?UZFOa=5a=0a%NwJX2XQMBnjNX%(;o^grDT}c+Et`$5AVA%u zMi_&hXz%!}lBwC9LALK8u4uI5g^E}{geV}A?`Q1*%D|b9!PJqGA0m{XQb%#|ljGB( z^lZEK680R9nbQC@D&U?*8^HK-NLKsIgeND)LkLCklW|NHT`7#s9?n7F8&s-S?S~@e z>FI6L!pO1#<_AZC5{R^*X9S1(oLQh$>wwcfzPog#me!+;aQ1Yo?U-l#Vb)3ecI&Zs z7IVVG+6pPAQ4P?6jExaBvrl8?9J|zJ5h5J63k_66uY$oIMyrhEfS}zWQi2()w-v66 z9X-=vs`w355Sw5na*i&j1d3M?sCMkg0zhi4uHi2nsB5sS z%e8GWHC3_O9n#aaI&Ia-?bm6yyivXOB;l>d%O%Fe9fh{BcKa;dWpBQy1?C1OS8ds9tnE8Bu{tDgG0QFr|gmp}oN?`tLWh-@h^QHU2S*BW#Kb9OCr{r$=X~Mi&yz$u7 zat4}paDHTc|I$jA?N6>j_1ylvSc3GA8QuPq?Np7xf9_=j{8N+p&w)QLkG|cfKb!;2 zPKw@l!0g}_^XU}5l=Hr}A(DI7IW*XCb(>wmhI-< zK1Jh~>U34us>xJ(M&|9wob!vKDo;2Ob;A;HicLB5_I0A?erfi`c~!nD_n*Csh80OC z5>8?Fm+r$ppK2sz&;nY@sH^T@I7uwYi^x!Sf15Y`ut)5=Xm#1h#oaAP_Y^&74tH-P z`)f+@Urq7G4<6*g2EoxN=8uxGF~43a-MOs~_VXgS#`5FgA5H23UB7_@&?#N3UXIqA z+24Qb3j)<|wKD2S_B9Xle&{!vyM^uG?#0?$HEi7tkE;#*Q-tOBsM3d)*3G{G!m7c* z>itLU=eGS2&(zz1u=3+M&Pe3dPvI+uzR2(Ku8ZV|5EzwU1C< zc$Yti#B%6vM8cRIoh&1iUqZBGp9LirPS>#hI-RRGwpbibKK|CDRdajP%vghewZqij z!1rUtRi;~8)+b}w(S)~6gK?!fry6jt{l3kKEeqdBR8CoUrVrmt=Ld9k`X5J9POP)M z)p)i%>uvRT`Sq4<@*u8$uw&6J+1P)0ii;y}0l=I_p7t=^Hg;#Vevejwc4(YAb*Y#t3mgYyM{h~km z)c1b=5+?Bnp{7LpuyU?Rj6j)A6ZbmT5yr%{CX?XPjXr9s$)Q{g9RQr+Sy!VJ6{>3U z0-D-tfpZ~u9hTu2(}KLopQbn=<89Adu;y~QkDHEtktI*rCVJ=QiY ziU)^$UqT(Uh_tSbYV=@gMsVxn?lteP2B~(vinjVn`#a~})=3}^+FjmKw%<9wrJD1+ zW<1{~G0klNCMhxD8TL0l8X0}xsIzB_etKlNz;%!6u4fa5;h3w783B~| z9XU$e>Jz6p&ObIdK8yVgoOew)wfSLFz|hwhqM)-#^!t4Kg5dj+aX5__7uQcFnbd=Y zo41##QkW~6nWC9ukM&o-KTOb$%qeJtg)bZFbHDZMu*w>d4m_3&&##iOcxzCIWmYD_tkJf&O+d~k4c zWg<^kpIfiEpXY-8l5MPe=hRgB%0mt@aDfd1Y9Yugl}h^typq~&j}m~Hn=LnS#QknJ zH{WBgHf^0uiN4mDHM&* z?kitmw4`*y%b0W3=U&2kL*LI5Mjsm{#Q$mY=4cKLfYediwYu+heTy9@S{}lAzuHgU zx%J)0?+ug-8y;K=@!95!ZA@G&h`!Z2O+IY#Eg~T1;f@2imeAd24j*w@=pjrO4;6v$(3}vGcwqR4@A}Hg=T_{UBncTK|@lSu_ zu*_2dKjRnVkDZ4D|D>864!e&Xji&B1Q%7W=CZAQ@nUzwV?u~rbfAV)k?2DDz!P}Cc zar!s5!wFu~j3?jjMpXY=0q5Cnbk_Jeh7g#K6 zMBpOfAvWXbr(Q7dg~cCVjND)O9N(FnsZ=R*$0g<#?cd+uVzw?mbYJJUWwXLZcTe-x zu0iJZ`MA1n8gL#Ge$Y~?{_riEE4#(}^C-_Y7@sp~JvqS{>}Kuh&DHBjiCjFjRhBE? z5}&`#xos(q#d`LA$JEIurxNpa;XB3ixrN4u57b88vf6*cW$$s?$!PWM=HD9&BN~zd z91mzm%m;3Y4v-7CbFYM7*uJY#9DU_c_ohCx#>HSMdzDV~g|j32M-48h@YZih#m{VF zN8FCG6|re-e;bpC;5_X^6s(!}s1coFGnFI1pC$H_S(V2e+p)Vh`sI&r{aWR3U+>;y zH|Exp`H*q0>vptxca25dORCtCE04W(MNQux<2#5m$16s;ta^OhZ5;B0W-^9z%1 zw&O~XLl~_EnzMvnC5*5TpjA!4*>70k0GHk=SzmZ|$<7~AKagv(!woHaMVb*lh8W#% z78^818T_L9qDb?xoC4s0!dop1SQ{Gmp0bYuOK*iUr>e07i$<8%bel}j?yfHgYI zDR-DH(m&p@s}{?1T~ETdCRfH1qM!wb#>upHJ>DLx`Y%oe(He8Zxu?ARfdqQrRoNj; zRXTr8WH7*;G5NpMIx=3@$f06xBW1c-xuNo(I zMQl;1RE9skFtp?IOC3A%6YGVAu01=v`kSMlN0H*bI2K)f5&rj@_fT*(7Y00?t|KlG zGRT)n+1t(}KNlHfEuVZ^dD|XUm=p?1NS6cacODH5Tp~qthI{r$J#~I$^BdT_HZ%J1 zYD7ZwfwdvDwVGrlgleBV$` zm*B1A?hmKMXT|gh|BB8qZouc=Q3&aV-fQyBwJ&rZjSMY7d+rb3dBjq@5^TG8ThVD; z1*bjvF#%+sZHd})Yqni_?>rCkVpID1XU}uq(cHFO_}jm7&g$-p#N_a0^G_`sr1zQ( z_n92_n7cgr)k?Yh4?E`cX7RE3Rzr-hNqUM);LK{A|D_}63j-I+N#m3LZ>y{}|Al4^ z&}{(qmj|nhDvwVdsk(1DcvGF1X_r1}u?Qbj&JY0BE@k+G#VKk}y zQR5@VrcU&ti*}L8KgJ~sTD;EvPl=lg)?WMi`HSVcR8(iR6Zx8V{;dZ#vg15GA@hCfwS=$OA+PzM8Lg#^chk41v5!Yjc6T3( z+!ykz25It_QdSEw%*xHpQvKsc$eTMaYvL!{b?JLL0~1v?PwYE#De}#<`cUSQt~15k zu7}ti&t8|XF#qh?SEmolaiQ{V&zirgFCpfye>@*#)Fx5qdG5Jeq5t59^p8dSG0n`A zr$6Hkrha*L#ihI_WZ%TYAMRDiTg*ZCl-jI!3qhVY&;M9%K5_LyK(;L1oKY}pA9TfX z;M*TGcwI1LaT1$gYTQ4kY}m|l-G7YkxIZEHWM1E33ZEC-dHF+7!0QW+t5eG6wu?42 z-YA2t$-Ud-5yCEXEHPS1vi#e@;x)Z;Uuhff;^{;0x7g+E&il$h)gcB!t)+Y6z_Ulo zt#K>OASBK%JlOi?MgPQuSgBekM0fCLPDb}csMr3kG!j|>p zZE$2l1FQln1q>AJe#HKaSKLfeC_!y!HXc7Ggh2&Q7&Lb*-!BGM1w>><0rjnP;#ivF zg&cwMXoL?hj?ub~&%5zhKWG-asYlNJgrbEJMz)T~-5djji-=s0k*|^|58o9aw$h>hz zR#CNu;68BG4QMm<^EI-ca<60px>Z||g3n}F#gMaO!7K3{yb-!J39WUY`jYdEmlBYoo(msAb4^WmSOZTf82aR_SL>}hjir}H#lcweSn zg!AU3dmq0&dFpzCwwkhh+#u`Tu(Z*Xf86Bpse{3+e++dr9AEBU^LXw?jW{^;YtVU{ zvmRA=$1012_3!4!v&Q?%(aV-g)OA~uL$4aY-rtHv83x0HXwP4k>S9WN!omnidS2my`=C4xMvLLkoucHYIl$B%;=!2W1@>f4)6+cjxFUosXQi zWBxBEk1aNYOTSW6wMm;aPPuq(l2?D~YR^yoa=`QN#LjzPwq>-F3%UYO501qxeOOcS zOBs4~#feKN4(18O@kJkouDo^|B903#>%Muf!5@b_2L~ABx`&c3iJgw* z-cRK^QF@AMEr%g)tU%4PGDcdK9V zZhe#QM2!-xyf)v_PU+67`--`lgDw$YZCS0KU+?I$;#iua;^FIph>ug{+uk2=p$~bl z*k~L>Qe_O^WXhk#c#H>YmV2B)q?<)OyL>Q9Y0M4X*2~%F^3S`4Z+o)iF6>BKqv4cP zn{=^iN6+Wh*MGeK`?a3DqTvv=Y{v2;T#5q|0_@hXK^U#G+w#_azXJ^8eC&~=OT=8-VGa`)KJK&PX22Pe1uW2QhI_Imbn zYNj;&@`;H3m){LLG+umGGg~$>Q`2>JXO?8>uG5!6WX7B?#qHd_uT^e|l{v5W-p$IJ z8|1Zbz+HN}Yvm#*W$cpDWFp0?%kaeYqe)v%h1lp$N=dd(t5YZ#$5*Jcx_H=%+61{nJ z;_-f>OE3Oqr%rsvw~863^8u9)vVX+=F92gfoW6<t?n(3%i*}m4wl)&OP1T7y zbmr22t!;3AU-%3E0F6Vbq?&qvXDl*B9|lPoaArQa{_miuF79ErfZ19fmp{Fn4rM;x zia%bb+Rl@xUOp`)5nR~;ms^3AjeDH+AB843^nD5zc8_-<3d)eiH%RT#7u3|3cTvoX zZEOs7v7CTOpO5_MOioc~5?o_1D&JIi}NSBqNZDxwm4@s3w{+V9&=E)t$kFF zn28%y4r*pRRd+Rvjo^sf;;Xr-SoQ5y+}AYO(*#CZnhiCHrslA3Q=uYY+|#pA1I;`) z6w?!lCVeQ)N19r93kqsBr~-HyV~U|_IpJ5_}1U644YSWt1t6(=ANc&OH;11pnT843J2?behW zgz4@R9=O0gc4tsZ{9FJkvf&c*Hz9d_N@Vz*yQQJv$OEYBpQkc>qfyt*SDf4FrxTNJL@0|+KUf`=2k)7Nc zVPfr$)nzVHbFy{F?8B`#;?Tk~8S@nLky7^4U{}4RxdS%N1xs&nBx4VUjQ;=&R0@Qh z$E6J0n+sVvv}~Y@#@;K=ZJ%rZ0M9l*%Bs3GyA^OldKy8PdE_@^R-(NV4=(aQF?&Dp z55}R$(MH7;{o`cxrFH}5UTW<2S%|fn^J5)A`@`xgq-YQqZlkqL$6{+csjN=*ZEJCj`6%VUI*Fvp{IKSNcK3pPpMRqHGH zqaIwFX1D$*l7coM!jAs6rq?z%xr%UYh&w>*g>IaB@=dJoN%(mkWhlkRr+t5({*-K(&L!HKwNm}N-%RXHRwH_9-*g-ocH z=pD!J58YGz=ozWAOufTLiz~JqM&~h;_3c@=Ru^{jY`BHNZ@S$>eGOLmrG9crJ!?)K zR(lu?sUqSMerV)5QS|iuDXTT8o$sMZ*g~FK9G~Jj{VF(NX`cgtIVaFogmxC%i4A7{ zT&JuuALMEa=%urBXCyBX9I$1NAUSm#UW*X~rZO zAo?ll4QE^1`GCnG#3}cf{{SKHMDv-MRy?e0_Yijf024*}gNk|@a@oPD;OUe6;D5)g zq$>~iPe%TAOj`Yki(W*^>V8<5V6pf1ra=lsg5qqb{skOwKjBuBS$R-He-hn5I0d+L zAMCX!zNBiEmd13`Nj$9XcV@1tCYv7i8MU2}fi*oA$VJ~g!x=I$?bMp+mt2!iw~o&A3*ZJFT=N!t zD~@yeR$lefA=)t4OfXn~%OU7L+Wu8e7-hPcM-rEIJSggEY&Mkl*xXGH+gByDJCgEs zHkSk}gXnrU^QxD?>4Hr@Fv}qTPa#qZA3>gW{{T9UN!H3IwzgnpZ}qb&{{Y8S_8-o# zY5Kjz^G9c>+b#SgU_6Ixk&fg)U^?@~D|8KQeuhn@thZB$qX4cl%*O#oa!+hko4dG# zUt_W-?CNsN!woBXpKn3hwdK>HyR&#S3xt#cOt7)rGx}rIN3~%?VR0?XLn1UgIgy-T zAnr{o8ZC7lWn|Z~vXUZ(F}hh>?xOmi@b6m9;x>zAoHp^zrbaL#xEqWA0JKkI&u*P7 zKKk_<6|%GhT1IkM{pnl#&HV>zj>5^WWH@mY5slMG{Sbds?cS!vCf26azLBQuOXpcn zC7?nCvCYUC{_Aun&>EwwigZNM&&IUh~E>9TE1D-jfZ_S*a z!k7(gQQnR~#yF+O`hFARd zc|EzL&N=)j7!r2$rQ{k+bMG_w7h}k&OEqL$KrZ=94+kG~LPu_Ba)leXN|AEk@=&#b-Ij z3i36lPE**>xW2cw7-AW{Gt#j&O--(dSh5!#GgPCuxVn(Dk`KKG21zy%gIqOhN=s`a zx)c<3-H}BJE(tyR){?X`0Ukm>W7?`n!WG&uTfIiBk-55kYY5MiV_8$?+?s1yASBDv zu^sBUb0cjXD%4l*KX&RQi_q19WgN)QXxxF589a)`$rhl@@AYGY%{~bwM&29hDLkeQ z!a=l-xu&~=@~FV9`HJN)FWIjYp*~E#y&9mJ(d@t0spp~WY8(iLTa((X+iEkf?%WAt zJyEJ|RhlX&X&VQTPG$uDa{#EQjF zwOkh#x`EkrrWrd*qmRT^bFp{FK~e$au1+&dELpm+n#M5a5+ihxITcA1$a(4Btyy1O z+qv@FbZz&Mxl%rragxtbJ8hg>PJiW_pOwCcwFphv%aPX`V;B^%Cc(XqG1K0pmda~M z2)T@*AFV+0G=~_->?jk|yB?ymk1F;rK0bkg!>`c)0F6lce#qKOH}l3aS06GDaa4d1 zE_$&X4@ztgD7X?t?6PM$2Ncuj$+u!PMv_ct2i~NXMGLV<)rWe~YwKIHvh{X9{QN{Y zC){_aWzwXDeC_Eiw`pTJA5Us_>=#{1aYdEh{{V1vnx5ZNxwVrpNEHtTGBTulgH@!E zu?oB>=qgy_NR!N$hW-^j=?vCbw9DmJ!r0xg$VMm1WBk^#?`$tErwel%eBX74D#O_K zs_gAE<+&vGt!rIUeJb2QO58@=v=R)W{{VsdQ)0QDxLd|R!#V5i?NP-GPast5*OOZd zrr7HL05+L1g5dIPXKuglw@=QgUL=+UtlBn>{{ZEoVS-QiQE>DinbtquN1P6=({cX* z3fj}PnPr3Q+O!ehSRm^$!^i&P_53*=)r)OzW@62O8uILW5Amg*aQKcKtV{P-`kzxz zN4VnJdK-5dTpFS&(5`&tRVVCn4YIGV9UJN^9WBHTTJAPYhZ_$CkEJEFAOUkD6hu56 z;IZ@-fBSK6iHlUX6UEVFLy03E$G_0#niQV88I#W?!|rkpI2i0b1x*@@NW`u~IT=JF zr@5^r(UBwjWsIL|iyMSRcDVOt?mm@?Beb#>a(DNq=1P)#0g^|U$++()InOz+iFIXx z5-nygwMmX0a3zR#`MaNESxOg9j5~mCSbU@Yd5=@?=sQ&)vP~uzbCy3+O&fZONl4q1 z!|iz)xzX+x_8zStNmJ^*{-ZTkXI(NTw}k%yr7*xzZF}&c_3_T)DW4(gSB` z;gs<-pnt4BgZ-c9T6a1;uxZjsqfEE9>_2@Zf9n)~bgxi7$6CU*w3_baVTpco>%dSt zl6wJC<6^heeT}@=HwrD=W3S9#I0y$)PhnjIz9xcX78lm-duS8}nVW2^NAYLV&>F|Q zyN(zQjj0V5b0pZonZNjv*nTx$(#~7SHb;yAOEVr3M^a5ok3?FLyB~)k)QFcyze%F% z@yIbOWBuRa{3|O`z6ivH$;iPZxALu() z&Yh_?wVQon)h;28H$C2VJW4y3@AM-likG2ROvJLZYq?lEi0$JWrIorU{-pE*rMNY7GxvwE@pI##vrm@lmnbQuzRIELnkaVc)>%hdiHVzI4s`7Wkpf80eM zGf3kB2lYOKy%r~}+3DW^70-@u*eUYYO?4+60%ec<4Ha+TZTO$Z7nZi_QpznhLW+8^ z^CHVJ?oU5XlvZzZST@nZKPIcbDt0_nEsFZ7A2kvpDkc>%;8bVQvS!GHu4*Ps{VKr} z!nK=;tYlMY&s)snQ`oYqbTs$?56)}6PSmv5~bXyp!K5~7n8IbJD*dr;yvA{G@B z)W)H1^_0zGK-C!)DHPHxo=Dj;JW^tvjM0kD#oRzDMdF$TJXJ*sOtBqWj;cr2ttEB9 z<(Y0XR#4S6HDxszK^(J;>$;QP`Nc^$Fdkw$Zz;+%cE z)gdrE4lsU{+zv7_X?hLBqv35 zVE!~?p5lTAIQ*y(eeSvTrN?o`DKI(6#V-W({OAEOo=2}*G2b*5KExUTAOVmKFb+mP ztp}$V&$TEg8RCLQaoT~6J#kJrqyU`II9y}ikcG$u9@Lz5p-(+COOcMCW`IKHo_lwo z;FJCop4{}KBcaU$08g$cImUBFMo8;QF+k8|dY?)NAY=~xC>=OF6G1;WxS*j3&N0%0 zdwzUUkb09r9Zz2UdQuRa4uX>>J$R+a1A<62Z3Gj8K!cKb#RD9J*FC9t#U~i{?r9AO zcsvS%=I$$XFC$~xtLhr0R>OV&Qe&qQJmA6N z>ycFDV(qmC7qF`>e(NsFZV+$>tzOw~V^aHJxg&wvvU3MXJ&dvix^s=@n-878%vLGy zS0d8pp5=tZAOmq@(9}1#Dt~bjjjP6eYFi_c#OC%3k>yx3jZ>0B$7mdOr>cmf+}xAegeHqm zr^ysXXo#9WRUJN~wNO{_es5Zp$$7X3Ci zQS?2kcGKf64PM%JpjLt?r(HI-Y*Ve0C=(5Cu;u#X=)br-Y&Ih>a4L?C&xkflk zjm^Yksi~sWXSR@inYO4H+IbE29qK8j$=MLvgSV|ZD3B+aC?Iwp-k}3D)imuv732$J zaC(2wDG&b3{HlhbWp`~j7btMwb=Q?)?0OolE3~cibnTASt!JstrQ3Nj?VdK|+sZPa zdix)x7r0u*1;wHR!Ib|14nFVy0A8SuH@KGxlEm~q>u38u%ULD#$-JwB*3x$5{{Xn% z{)d{Qa<8X`wq#_Rq=bwS>?pm!WRk7AG!UPd{rLy+dk<>Su+?RnAG7K(E$yld=0?NB zKiPhx=~*`LH0}m6pk%Q)3MxtGU7`zo`5^ZkvGp{0hmor`nzz=&`xap|Y0Fz4CHj&4 zUx(#cP}+c5?m>+Z^PUMlfKh1?4fcTT62@{|931)&def6zgHa+aQrNPN7I^;v)-Tul z-;F!nk*d)dH#aYEzz_nc!YJsXq=qDlFFH}U5%MAaBkFytN=*R417BLox0YaT8vzaf z00L9evE`MglW{mbJ%u}un))He=4ZI`l1EuRLBStN*^2hsPcQpg98P*o8sXGqg2SQy zb)OnXWfILggEMmy9&!HDe@dp$B+@f6E=lB`b60x_+D5(ahp!SdcWENu*|E=`56zGF zhgCmXuGrGS2*d59azXz9R1abP*Yc^YZ)VT|dudrOTyg_(8vBlf`yQ2PF7(ZMQ62rc zw7heLxVB8IxBE%{AC+25<|76)lHz4WLeA#^0m(IXc@s~7+rYu)VTqmj@1G@VkTEdDao=}mgXJf+BwY^49r%QJg>~CW$#W+7JBW=gy z>0KSpyQf}nw7o*FB88g%*#eh7>-k=;J zJu}b_vcYL&&@aYObyB1Boszebe$pF!v< zbk$_iBoXOTvXuV-Suh{#Po_V+T2NeRULs>1<)N~@+F0D%FYeAcC6~9 zlN8_pH#~=ci`n*K0P0Vm zqiqo_Us)c@@Iy?3-^N#WR(FwHTQ;8Q#}22=kE-_RMM3Z?>Oc5RZ!E2U8|C5ey$BOLlsjyrKm+=GEl$<94$h+^m3lNcE!W7?GC0D68DV+WPv z+JOnjC%3%^j1xz`KRN)9b zJa9i6E-(v>92`>PBdr^p%v1C!BgvNx{Z2 zDG42Ld83d$XaV^n2fwJMjAuFI(cE%z?q~y?dLFr;XjUCFNLQvuG`Kk*&Vzy4hA=%w z4i=Py=|*~k%`N~w^op>U{C3Y@byAExryWOy& zu(cgvxiH)PGAgb1r#1A3IPNzyrNnj&JO{eXe)P!|tBc^@2SX+s&H|bmJTYuT&Qm76G7$&B_z5okLFOn4Q58gF; zCyon5xYUA~Px`#|$Em32d$UCd&hUf(02I0zX&TtZC76EnkClBYylFM<&ft2c|tL zNjBRkLH@V7^sQ+wEwHI@$sN%g%Av6ybsQMbKaWEuB5dGNcp7iMASYr~uUdnsbBQ}ferN&UtWX4g9`A6?B z;oh=}Jdzb4?$1L{j^cKlJWk6dcp&2ys`s|ndt$u;Iq`#W80vk+Qn5BN+oK~HO%iT$0arq(>OYlH)U59)W!7yqT*0cu=yUR2Lr9g( z>g38+FuP(-5q$@#^))T6vLt5a=OzGf0P5$^e^Xaw&?U5s?W-bP*md(@Io<3>RX3WK7LK|<~xNT2&k=w2aEa|`c@(%%HSwwB=83$nrlI{`j>RQM*hO*HZnF}hK%yA zJqn%(t?4JxZ`Me6N2Y2H0=%(;U~}AhuRn!gyDf?|Q84>M3`pSTZ>od*>ZDN1b27yg zZ1OR}s*XvgduR#kse121)GVe%nhdM`$rOT(?ig`Rp5i-L4W*mz+&_OVKKcIW{{UzC z)$46?<3m$t6p3{Xe|;Q>d`;{J^%ZS&?Hj~@KA9)>eewSQmr0J~bN}YncLT;gwX;3ukE$cuv#y^p zJk1&(Dr4?CfmoMU5WxYov}}+0Z7BZ$mWSK@pKkqXlrTwgD2*2^-+&(E)U~aU?R`yI zyg8`BIwMA<8zp88410T>!;E5~ySk1TO_h;8KtH^Y@MdA$`+5)SRufs1OoUq5u$CYB zYFuPUeK2|ou?!kky4zyiVR%MLADTA&dWvl=mgaoYPV-tF$L##s@#XjZgCzd|(Xmlm z9|g48w4WQ^TNpofbmwDnjwC*0c)|4K(OmTvosOzfQjO8&TeC^Jo6?Ne&|L9sJ2c(g z(i2Q9DK<2TH~w@7sCODo)UH=4XK?kTBvb*Y-9^R9PmxDzh#HUzXiZo~867F6B5K0N zp4BLcd(|l1o_iV1Y{YR@-Ul6xZGbUJv}UGMQz|vuImc2Dr9^tx{Pyw5#yPC1?M!C5 zDp7Gd=~hxI;+zE=F-{9vBx=|JPM`{Kt4Nl>RH|u%tvPDZ47Cx-1GQR@uLGr1izE(u zRmi!=t!kMTEFjM#0<13}dejl}GgcG<&21KEBuF~+rvb)0(4(HX9@Qd@4!tWx%$ggF zjG6!#rN*GaY@GC zq3cV6eX&6Ua9gD=LB|whuQa~hvDY*JoB@t;pGpDe(9(g&G!f5mC>UgVd-SBoJc2XL z3z63}9`slRDl&SSQOfkk;Ypm4oF3Gj!S$pu-^0?7U}uU^!t|hZ&#f_ujAK1LC?^LU zNAjh{PaIO7m;_R|SYSOnQm%R8l&%MF@ulae`cZJO+;i`WKZxV4JC5h{rtJeCodK}n z#V!jSpdZ4OM)ElPYQCd*mQd}F4=L(JMCPEb%_5u833+=Jj60ZvCOEEUdztkquycxI zi;LN}K2A?c)s6`C_&2r;gm5d$t6o)KWOb;AadYN_gdcdGn5_$|sBhx?6{L$Bj>CY}#a^wnin1=8^|YPYLr}h1ab;vN67k1Y zt5(sbg>Ckk&TeEMDtc4x?BdX%H+Pt8f<8p_toR_i@dl!DP7f@@qK``8l~pA5_g7QU zrB?3ia^xz$A(4!#Qbidu54(zsSCra7mR7NphaWQ!;y-m6>V4{n`%bTV z2i|8soG7gusAku0yxV6o5^?39;1wn8(CCmG8$-I~hwr8S@35_7J-?ZnB^gG~N^9?L z{tw_sE&AyZ{f^UXk^Z&MIrbGbmYZp)#rD#z5`*(6_m8$}o|&Z!h}!z`ysJq0EPs#K zR2TR1UR*42sg4!mIM@93tG(Tq%&ENT5Z)N$)1}?dDsE9OTKKSL^9l8LyEujH4}eoBTho zz|Oa|lI~JVe7|{_x(d1duGaf)?Z9hs!B!dH?@?Y{swdekPwtfb$Ge(KTV#$uwOr&N z`>Xx)>GZ2wn93?XA)2!}eTbKbyZkFzGxceVU$C^3b=*J>3lQ`Rsp43e= zw^ty6^JfYd_3hf$8&813(mnLaROKsQD`Gv2Jnl}l}mi9*?-h+xWkkbdvAK@28B zS-$Axx20%HqR8Z2K`PH`dGl2L->B_WZYELtsCYsD0FP5fEp;r#s>^7>Eyyu+F}TQY zsH>@GW2g@{+x;5k{{YJg8|7pE$^QTv%7zfB=gY#5fd2pysHK&smpC1G#XFg_Np}Ur zcSb9SnIs%-BN(T#7O{mmG9Lc`-K|vB1U9i+>S}IKbUFFr9?ks?MK*@hNA|5Uad&N> zo3qFM^8Wx@b`yQfh~s%%jz>~YZ*x}R)F!ZzO^mG@4y1n<{{UK}nnjI#s7^8Wn12!f z019LfEQGwLY7VVP8LqPFy0UJ%jcjiF%^3aDpK<&r>BqHD)hw?p+BBX#Lc{k`Dcp=b zLGM-WCi4!^Fr$I(S{i<%bEexeI|*%M>u(_3t?!QCS_yh;P>wMKUTfjA_fYrus2QGV zN#wCVooqFwzOd~jgr#men^{f?{{V3GKTfqwr}l9xENr)ZHnQVj{{R{UhUILNAM*0U zlm7rb{{Z^+p=qm|r?|~Koh{bNSu}Y?-M?8%{{XE~=*#K!;<0V5%<>$afq7?ciYi?=+@;dZX){9c z{N7+V*gKro#JZ-Ps}lyGrru58E#xcr&#a0%{YNy3dKkBNA%~L^m5oo!o~(UMI@;7S zHN^Z|uLXxzKiwYndS4ICX_@DeIV?W#xrpvp_8#ZZ)_vX4xpi2^$kbsYAHqMW_of%V z#PRDlcGo^-q^2bUXpDW`zJwmNtq{_6G>+ot1k>(WxFDYS_y{nJ%W?AQ|cMjsWx?)FN6>V&1J~e``2v$j&6?BwkUM{gODv5Gq(j z=g6U@VVI&})BUC&$W`q(QkzJzVJivYm+o3fGOhm6^!COLI$KLW5J=NlTo`X$<*jZX zhCl2%{zjO27*8Y5DRyF6j!8Z0@B>T=TmC;7Wq;|A_5T2S`F(0vwtW-^x{Gb;`-sEt z5APp;PtLO2bgz@SiEpVUsc0sRZ8KlgEXUcnyR?z8*0YW3+4SxYxUIL+w5!u1==TX_ zsgd(qTuH$72>!obYXySo5#dNT%sye)$p^Rl-^=S+iXK@2MLUVYfz+C(2 zaxL7~()G*HvdZQms9LwmAe)?jx_XMb`hBI~mMt3QG?9CEjN8k zxM7mwV38>@F~C~Ab8{V(4QXt^y9`Y*{x9}NxAUTTKeOU%SjHxCiKNC4`=i^}*0bVQ zmPJJci8u;6lS^$7_w+q~;0B>}FN`m2-gj$pr@VqTIRN>QjE~IGQhW)D-q*($w&+9L ze`dQ%;PAh^i{&5J=|y2tS({V0E{B<#X*A=-M91E}_ISt`;8bLqMm?!iW}@baSWilL zLUB)6{i=12sYc{Vb|DcGDgOXwC)TZ`Ue#eE9+j+VT*6hXXRV`Sp{Y{cx?t9{Q4YAP z@LH+KBbwDwNXy$fXkhszM^Dm?p4HK{k{))|XVX;n#XeE$V(q6Y1I=i?kX-Tl)ILF8 zhLx->O0RAR$mvw)wcHOCsuZevR7^4mH8P58Qllppa}q_gkZFNyTJ{y(cGhuRR}7(M zdJvt_7AU9Dip;x+Jw-i5CMn=mAvR9}agb{4N_x}~b{tj6z~ZeFa!A*&wN9d=LG4qo z9f_@CLu1fzIjIzTaDAy14)q}5`&Nj}fzCZnJt`I);Nq+vp7jAdAMvY3avDMDj?~k~ z0-euO+|y4}k6N^3!~pjP6ur+qyV0JVsd1bW&~Z$WJb7L>p@5{ z&Twf@9N>yvah%}&D8_jHbifyp@7|Xjk_Qzfyb@0;#S$4LZU7bE_DN+#%tE>bc;I(!ow1bdJ4Rt zb^r?Co;CfGz8Dy;$ml~Fg;L1OxT|89Y=;y^@VFm6zPFq;Mj>a)>Zo#A@<|4WHJlQ7R z`E2$yq6rMUr6(uXy=y}pcF_5<{qNyg%5#-Rc4u`?Hd?a8Hn2Q<(`w2G3TlV;eck2e zCQ_;nN$rXtNv|&fQd9$yM{3!9Rsuf3G2o4+LD<(lSAC9GDwN;5tNe^zM)4wySy}Pe zviGTNEMwB`P3`A0bw#UVMNKfZ>z_3TDUNF`h1R!x5D%BIJrq|3MD|;CHH70h+7I1Z zzT~!-QtK0#XNbIKqknR*>QaAa>EC730A&9FmXfDht*BhXq}rPsFZHd&k`CVAT3rs! ztz$Zc>y)?awEqAX;a+uUrSIECvhHmPysMvO(RUf7>5*E&{iUg$u`ekbb!w+`J>Q5m zB}mkC?xXbUL%N)l#jx$T(mwAvhpC?CZ{@l@^#oNKWNmGdenO^ z$E!-NmX^h9sblcnxw5k>CZNMJOa0GbQ)zZ~`t`Hj>M)2SfQ|Emv_G-;to;{6n^%@g zyQvITas1Ox$FH?BT-JWv$z^1wD+u_F2l1a%?^KoBrx$s3eUCyFlJ}+X^hDk!mg-xG z;En?wig5r=Rfn#7njp82!`7nW2H74sH{{UL9GfF(aVRi0xDZYPbwVT;?`52b=ms-W~;h@E44boHC~pKt4{FIm~x>M@JUn3mM;z#F!#r`yu9byT}+!jr;d zIl>?u0oV#JZlu4Q5W8bvmN?gHh#chxpYidueXvz&RPqGj<-PwI$Jw`~9O&oo9*8SUJNd``_bL zZ*C*ELuF=8!T03;`Tqb4(HTiyUFj9Kn15)sn+U8NsxRI@^V8}*D@Ge#KU9rwq(!rs zwhHGF%6+{nD$dOv4aLdX0v=CxK7-z(yt{k3Mn=fjh|qLVOwQISE zaljtlv;`J1g5xuhl?V7e&%J5d>My5RELwv7r5gnyZ@n1)_Ii({MLv)Hpc?Z_xcf|J z=8`w!NdBX^_Y~2SO?sH|q?Z!>xpGhW&01Tguyy+!tg0AE*+4ef)88q3y?VBxDcnd!KsJwALcORI=8sTJAr)9m;Vue~ zo&t{K)Y@#fkl!rU60}nE}eM_Q$E zIFk{=2!(#?d-|Vxeryv@V$@pQ)uf~BWGnz}@)&MMAMFG8kG*NbVWw*+E+J8;PEI_T zANNwXqA~lwL!Yf`kh{eQ$VS zy@BF&`Bg~3RqeO7D7UvuSOwhRI=A|$k5~6TgZb9AhN~^Mm={(=ys_yF5_06{lC9N1 z{HZSVq}G}5G;5!<9DLS~^353Z1EUY8_f9uQ86WD$ zW?|}k55}lRJn3??LoA6XIS!Szi}QpT5^&jiA!S)4oR&d%cMeM7;E+zXCLyw zdJp!!Pj9VsV_&p?DjyQTy7(^_mg)r1`flhL_w}rOL&BGFKyR!jO-jI=iQwd~voBC+ zPq;PMz>Ya3kRmB|V}PJ!lUgq0N&z5~{WNm&A^!kX{>qWx)czE1{>cofkG7G9VgCRu zf$9hQU&^zTG|r<0mN)%5`BL%`iPzIV-~4|%cA=ojr&>!k zlQLgKaI}G&<^ls-_+J{CRYQgC$Q^E_VqMcogJ3BZGW=cS+n!^aI?fQ zAo}{%%bft(+Q~kS;_#e?bw4~q(;barQEmCZ4u;{*9tfRC>nQ%pei=BDP>GM*!zMN^jB9Ib9N1JbKYZ;pUcdV(;djP%D9 zP2$FIIIYGiI`-A2F^vr%f^r_@fl3ttb_NunLht zrjKfoX`)$T9y!3P(er^+V&gn!u0#0Nkw$Hnqv|-R)ky8mY6y;bs{@|Gw@8N}>(jkR zr=Im1fyF;K`c)!s406BYP~ZY7?0NqH>r`ZC>T1#-S9{7#(v?-Nt(Z#UNj=;Xn!B(-e(@G5u+BbJHT1 zpzr<^F$g%o;*?Ra{lyM%pjnwgB1nkI#(bBc8^-*G?2L0?^8z-mIo0692B&o`qy**!HlU_w_5818oN&C-9 zZD_DX_Pd^)RZmk_*GRRQe%o$$#yRr))>P8Zdw3%XR|g86fUGJ(qG2T@;}4&5-G-xk zJ@=Oyi6U+qGB^UZZX&=LIj`Y`QM~qG`H7iQe;6%10kKZ{jsiQNEv9y;v>Ki1Uc!AI#R|nw_q= zMU?TfWa1}YbM+Lm*+=196EcrJ>P{8$!iwXZ7dJECp>eL%?@@d2WzeG_gTr>4S8M7r zFJJCdnmfs@MI?HhYbCOBZX1SQ$GvB1x^2d{deTnhMV!eie=46_)T0rupQnwMiNP!X z04)#rR~+cO@<&-c-)~RQ>4c>TdpJAI>fc5EbVi?vt|p&NhTBP1Z6ZEDb^J~K(5C6y zZNG+WS6J0lHQS8Z+McBU0C?1TCOkW8+MV%y$heBzINW{37Ye>KxD!Jw#i(Nj7CBM} z_fKl(O3T@&`?b8g{{UTC>d?eSe)M%~Ll^!g9wgQ>xt8K%&!Ug$1#0RS@OXaFw2dl2 z(-Z!cbUUB!sr0G!IpOfFlW8^$y81!>m2?Lq{gpM3qiF+BX)f;{$wfBQJ@YA+DYdFRHTW>vET zZLm+7Hu{1))h%aN{?T-X-ajVBE+K_~ne`^OG|B$};TY0I#-o|0pABth-4p$@=~`*L34Jh`jpBHFgHF}Kg8btg0RvXolf_S(n}*`u|J6Xo=*|{=f6V5ov?yK$m>v~sPyUG3mIVowhSK>uA z=+l4EwCO%&Ig)YJ&!GPRX06*tX{}70;m_ze>-r)4tdqdxkLDMa+{P zpYD%RYCHWy?K>ig3|6u697DIc^)>VeWD3J)mk{tf7<8nkLOLh(}uS4Ug?(x;zs$U zwsVP7=sS<6N~?sU_PV}?853Sy zGjAIMe8cy$4v~MdJ^f8mw6m7_eT&8cW^?yc4y5~4893Wcrp>(eULaOoLNBvjAj!Om zaHN0JMgF3y>o&JKd~v`cA7pM)R6CYa>yiAsReRf%SjC;OKWARA(8zsH^c8zTxVW>1 zS!R`6OzA8#F6EH+0e`*k+N}{J_C{@`pjge++42%@gMX*mc_C2#?go(!_`mpRjod$PLL`4i~W{W(7 zY~*G%?f&uftMKbFY9LzcR_1j$3v&>}%l#O89%~v`mf}*<62*_+?rQc%nHgRl;fFu; z$bUxuR1qyp_gY)&wsKtQ?k^!HYfhv6Q1mO+Pr3Xn3JZBIrV+#sm`@6SomtZ^udS~H z+H}$+r;I84!?+`%A6ng(<4M%gOLdYRI{EM(R9rzZ?mbtY{S7-Aw&T904b8-M@U^AC zAM@5wKh_^X=(PpCt6aI>cA|{4o{B2n#*KSvb>!Z{Pyx6|Psz9Y!MYJpT+X39i+*jv z@72$#qQQ00w-&cGk{f$qvRIxP8Aj!E{p09s2+4BI0fLdVRv`7H9&7fFIp2U|0BRxq72GVj6sv6`$b#W);oUw{sU`lZ*ue0Y_j3 zTegX8V~+N1!c)ur>63wvz+u7Ugf_To{>037&U4ogm)+2qKpk1AuD08 zm}L<1As(mwuR+_bWu4PVdCL>))X>tfj;vkiekfqK=(<*-+zW`8_cEXMm-tV(s@jf= zdEuD1w0q=}AueO)Fu&}dQ~Ff`i=;>MfY3$;AH0{uij=Leo!+R*+RoqrPEP}_YfzNZqS~Z=y_XDnh)1CR0BiYH>)7d+ZRY8$wmOq8 ze#tB3DgOXvy?u{0%ih@C+FZ0(P)L#k&gDGS>?JiPVzvvgz$c#F>u%FfxbXaJu|N7l zfx0qssy)ZA@~W$iE+9x&yJ5nfe6Rk?`Br%*SI$mJ&nK-b*sEXA=#*&M^76w^y*-aH zNPqTPisMh4!^&iI-*Va5gW3sWsFReGI&_wsKBSIjuO7JvuQoiVf2Y#D}7Pw0-M+G`%wYlH$N6t=lf; zwVj4D?OevEXMJsS;`S*PLV0E!AI6JYQ7Vy>*9s70*4+sO)jK5N*iqS+mA+YplGq$jHYvgLMG}@zTDUaM1bK zOO=?!VyA=xYB^V9Vx@rBG*+ino#S%!YryGOpop>Nsl)AB)}WkpI@4i8fJQ3(P?L_es$if-kc@CD%t^>RVxF-P&Ob_>BgxJ>Qf#D)5T@nEDpLRh zr!^9SoYdldqb*sx^fgxMS9ph~K~<$dD%`Uv9feetm6w7AJw)th$fNw4&AU~>^{p8R z&m4-OGBf8RicAi6?GEG8vL`|qkSn6QfU2H3tZCyuneAMaPKYBsj%Wg;b;0AUARJaD zunWydsosPJrV2ap?^KCnkP(kswHG90Nm|4F_4ORCcL(%SKd8!C{juXYDX1IQPgep zyGOXWkRm9~Nv2+FHkMj*-pwdfRVR~QKlt0n7hWd5YkPKOv{S$bueqp&JMJpAd30Nj z@xP8Pyi0o4*38Lhrx+i_T$+V}G1ZTHv@9AsV{?<(yo3PNgt#1rk=*p)qPtmuJ+m z9-(vgwfj32Jrzev-V@s04=A*7Y6Wq>u{F;J7ExpVTQKFlDl2p{%w(00z;ew>?_(;I zV?9=uHDb`^OO|`%vIO~=$@|q7rKz>^t-LB$-EgtSKZyI)r12DT;?CwTGR*v|{oz&{ zOBz%!e`-F{{lSm9pIXjPzNo{VQId6+z0FHUv&|PZRr{^Wf%#L~>tD2Sk5nKNNXX(wh2*xFlb#vTTG^v#+UAS*2 zyY2mSG_?ql?@VYkH%}!A^J7!RvPlx#Oy8Z!&_c}e1~3Ck47KXxn)B~1-DZ;^Fs9E?+&l8pRcLU+}vJkx}X>> zGaMogh$~iGxco6__I53uLPf;Wj&=`0?^bW19u%_>!n^OL2XxrTQS562Pt)YoXJ?VN z+poZon@0ZeH2WKW57@4)svvur25jvcx&HHe3dWX8-E&kf zOG|Sx>FNFgw(izH7S3*MrexIc16xSkFdy7M(yLs$_;TIspgGheTu(Rr^B>(`>T6m{ zUT5*Y!~Oxfs=gkBE^cGc;A946w*=>+UH%l{gLH%lNQ&Em(OFK34>>~Ss^^pGndA;jyO>DRBZ5oTIXT(y?`uQ-;!+P7>eBU=QpVI z1J##*4^MyROx7)=(yd~*w~sb5jmF_l zMD6t*0QCIoz3t3aw&zc>ZSem9%W1*pQaca!N2lXfJUunc9&5CSG|U5XvaaM|*lq5| zwmGX^#^v@_{mL44pZ0}$<$acsf2@3S?5pZcMR|Mnxl}^FR7f!#L!yuBed~JaD}N9{ z5@^zSGC`8Me1~bzc2@o)`qpi|f?A=u)Z%2?#3W6_NZ#YH^*@DKn^w^gT0s@0NqKZI z`OGook1+b5MgDbcYmsU=x7T4aV}!bk0#x?&VeEecS$CHcT~1`5l1>3$fOZ4gv=Ii4 z63r+Nblk@oAL@r@K8@%<8flrhsTVGEnRGj6yM+C$ZrUXa^29&d9YcF|tO($H+4jN< zNauvlENb4Ld3|qsWw4dnM>u6aH<%AlM|$3W?Ax;}7P3L7>aL+=0G}z3IaP1rJqM?~ z1UsVw;p4JLwt$a67@9@t{^&lQ^;Xu}OUsWh7;I#*;3)1%^s8^BUq^l{?yclWeOm(q z*F7MMG2d=DZ^N3#v9Wu(2@a_gV|21NL|)%X5Y zw{LxItlN{~2nZ43f3#v4ka`9m$NcKHzo%YCiE$5?Hk1b2c`eRkMfdI-*F4j=Q7L_v ztjue>x44&p1Zpsdy%1IHGHabJ1^jc#3g-xl#2HBbgS|y#X&ORrF3T;toCOEwmA|?@ zy-ilUypr~P?gEfDs;I)C9RMALC813wtZke9D_{9_6c!i6U|mGNx?}#*_WF;?pt#Xx z(8!+lG?@?jjfZkc@1IX@^-2qP?co=9N(!I#QU3tXL+XFo{&h9BzkdXhWtMm(KfHof z+=uY(_)r@y4B2Ip=2muM%*QNiP6#5?Axo>22I;_ScpH$T`|te+e*AUVw$dzhEw@^7 z`FCYYrk^usb~~x=J&kjEf?bQrKG=>YCoi~jnrUfqDaFf|sRYu=aWhFP7IryiCz5NW z({;veL9_YOBf3l4+(?fnOTr(D^jlIC}j z-JRU-W<5;}6}i+&rD>ifcAG@F7T2xI*_$dKcjd08l6_e>T^;% z$ssJv#GI0PlUj=$dQ>7;!(KAwBVqoi9^`uZ3f-5(aoa^Ep`lo8qjeO@2HOMf;-YvD;agr_M{ zc&w)rSUVi-v5}k%RbmM3TC=BKF;pc%@6B^GTX!0sr<#~!8O>KxLFT1a$KmZlA(IBD zjAV0AZqGew!K-MqDI(gpL7JUp;~CG?)ir9FUyA69Fle>7=cP%L(;aHMl1F@0sN4#n zqD+oJ>)w#5CzFamMsxh>PIK0QFc@BQ){b%NXr~O2C;_?13?6OoN^% zLKBb1mxcGE)PFit$m4-P4LpwH*NQ>Q1K88gUtvfFdvvBN3OF?+HuBs|u|%cW&H%<~ zs-@T}sb*t6$#?C7IK)RmEQ!JZF`Oo48$7j?sRBy_IP zh>kLFFrtBg02~2G>E7V^hCdHERcQ6$!< zKQ;6VdF6snI5s=r^fhsW8peFT-A+G;vsO#HNWn)+cJH$Rf#m}KiBd3k5iH($p zsA5g|XE+tDZF>dmK31Z??#erTYObSUJODBc@E^MC-l4LRDjAuaY$O5G6}xS#K=2Dq zN-*abSwQ)cu-6vxV#_kq#`5D&gwrDOS zP|_c|hqYPLH4E{m28qiS`{4cDeJeO5^WE%Ve9YkIwOXFxE#o%U#|QpeN$W@8ku!pp zk5R2)}^1~AJQCg2A zejc`1V+k&&D*W$^`9hsFiSX64PY@Ru5}YK!Wl(>OVOl{prEEf_ z8Bye^9XVLwa(p@(?BJBBheM{Q(F9n1!>FN%X!SiD~yw>$|*!lxhblV7`fa-dHBGFC~ zFXElQx;?#3T=4TrVI10p#Kt{88(JI&`$7Kb2l@7??%|8Y8zs}`^6uh5qQ*i1Ab-y@ z-|5z+Cm(mGb9hwN#vT3q+Ktzo=Le!jyecncQR!L`V?afffd18W%xV7sJOTb9KfUSc zPL2fd`YiWIuc^4qF)+`Q{@V{*7x#-2hnnVQ6cR~v4w zRPp}epHEZY=}z$N#qwN8kVzP@3!9~!ZjWYp+&z6U&T1_FC9%4neaaRZb6&f?;T5DK zbke^e2lXeQ6qnb+ zPDNzdX=n>wYJs4&iMEtHG_F5@e^c7DQogJm?PNDivW*}bmGLQVtVC!VX_@~3x({GG zxAir7_3M3Myq$XAlf+jr51t{|5z&X&n$6U&CA^=^oFX9kA&oxlYB~Y!{QA}VSYy%b z(*FK5mRUY)OLiZWjp#iGf2Cc**U@ZRpG9vD+MPjwzm3TIIaEGc)7y-itbJGRspPnz0Mc23QuaM|~(S1y`;yJ_}pB`yfLxl_UQRX^PyuhOpLwaaU+ zC8P_bPapP$k^-+IIR~Hl=G}nw@7EO<{3CJSV&_G?S?xzMD~wAB@40)Aaaoo&Q(w+4 z?Y5Kvv%8!suXb-hGglh?+PX!f%GoSt(A&g& zj2T<|&F|4^};ebP!3W_?0C(gkNH~hyvHq03%{}9%A&# z^&gc|*6eO<-6E18nsSW5(YE!I{jI~PKc`Bx$hF;!yX(feWL1nuYY@ay1Dp?1dz0MN z-462WN-cb|I$Csi$jZ)tyR`6q&uXt{W+o4JaNbpnM-E3;J&}D${W{clw?^&N!U&;3 zwitooSN!+yPt1o4s~c0DGsOO1ks|50%DYnDM*YKkA^sA54_eB7CQU9!ZB`dTjfIHt zWIy~*y*AFuEr$C0a@!+dC^%_b`=k1v-&|INUM7-N`#!5}72S&;k|Du}f3V%ZT6!`R zY+&46OLH-K2s!LH3Oxs)tM<0A+1su4+J_DPrrhvkKi)s={{TLo^|YQDw36KFnsZ5c zXcV}$VYXb3%sZ&=MROOI6I|T0%t({C=m6{orAVwJ-Onx2oXA#8=Wsa3)6%-gHOSt= z5BNx;Ny{JH$N)0|{p0FBy=w;A)*Cxjy}dwpa|9=>hx^CVLwn zGI7|#tq4`pT{@i>qv7eV)?G(Ry8hCHoXM~sF+bj3wUuhH>B;t)&^+V)R4vE{`_1(9 z?^4)FHi)UHtgj5mm{C}Q!~4hmpXXY?+L}j+N0+Q1vA*cHu&>DH{k84&6%J1KE-9sE z?sC@`65USH%;n^BmUGnA*kD@-7Up#iBXbl-{{TX7=qpP5!_)YDeWv9S;v>ut5#1a6 zlj-ePGfgeT%;ng#a@nSh6HJ;7M)C{awOicEljl3t$Tifh=AEg-3+r%2XL#c&aT?=y zp>_VXfo*RMox0v#px#Dci2nc?{_*x7t!La%CDhF%tXbIPpGq%cWgcT!ST2D+vO^HMxm?GekvqBn01C4%k1m%f zI*ddj^i}>d_4KU!o2h4%yBvA}(M2y)MXtq&HGAt@&$L@i(Y|>D?&H%PFB zT}It(rp^TG!pMG=&e>T;=x%NqmIOTGtDiwo-d(lCoG4hw2#e80H0`*h4s8*!uIRdr zvx%8J+hQ>skqn?7*sSe7$V$yDt`Z3Oc0K)vy=wSxSk!d#^L*&`b{hxk1Li{VAZ(34mYH0|`=+1}Wb@00J?c{N;$Yawa2!p5k*tw#t{21`o zrJ#I3zWZzhBWbRyD)O!S$h-a&S3H_;Lu*?>mq#^;cyg?9T6a?X$2?Xo+`!~l*V{wo z`IXFjvGWno)jfkXYIh98j~pc z^U|xc3dnXP--=awt-mzodZi*<3#j?2)xRpDkLkdx2=%mNWhj=_{Y^}-@A=hLZbeM3 z$6DDl7olYvk%N#a;~>>_Kc!5sxlc7j+;|7ACsXqQoOY!nziK!GBloJk^tv5qZXwF=oYuR+6ERuS#m#^{#x*-ihCkkCs53KQgX4H zi9PO|+Ip8yo-@(wJ~-M&-uc_>T3TKFu`yYImfZd2ZhtDC`oS(>Y4pM#A@`#lMPlSq zjf`1ecgml3w5&BI)1fIgC5igR+~es~6@JV&F!{X?YK}PG>O$SS=8|n}I~w!pacT%| z>?AC|dE-BMYaV7X{_Q~+>sA)|?Br2u#Tj~#de*wFp0OIqAWM7S$z=Zk15~xxPfb~g z5-e&ZlY`VRtt=N8A8Hap#atYMJ5pZT-A8m{o;F7vh@quhbn?)G=c{^EsS@LVJDniD zvDIVyH0-w5#9+>&<$tfGWzD3|V{pR#3Phal>L2OrS+GK6CP_Gqbw0JTZQ>osEU%&a zO_$wOKYcya)~bBk$j8~ZX%Fkr&$zs6rc~RA#}T0IS5OTiF(cvcAMT<40qa+;^i|T~ z{?W2=bsK%oIf(Y$I~u}KJ?+bU#pQFCVtQ3O@6?scHD#n$=BAUW-lfw-3eOCj%#!k? zR>jqX8lIlZ;pqm|`K~Q<`@@do`@*@o=SX4uC8K=5^43o7f5N2F_rT)(%J4-A1zAoR zo6u77GnFNNb*Jm{B(=0u&DHoZTEuX%^fEW4M|pWI^}@?4fd2q?7+0c>+*a%AwjL(g z9;vp<@1$2Fly5-e`qfLVK7S0p`=k%Y{`z#~i+VWf2kBVa zcAkC~S)%%pR@nlH0EG%y0&zEH* z0UV$19>>_!I0;Ho_ZQ9f{{Vye9L}SD`h>uR5yqozP_IQ5a^3{dVvh3ZLA`Oh2v6|( zJAXk|Bh!~bLv^Y}{+FCXXyd2*Og#suddQMXol5ujo94KcxFeI3*j8>G4WiVvhLK;_ z{5iL4s>QF#9-ph*ytfW^#d&VOy;JDJ_=oFK>Y8lc8L{%M2b^v)vfri(fuFdfbocMt zs>3?p$;j^{i(LYK@^) zN<7PAmYt`|sLo=9szify(m&meM^--7YWDIMg>1C)(O5#+$>FAN?;q$p3f{hlPu1-n z&%;tIA!IErsT*40j)h4)k3~P$uAH`gq+gXC+r!U*>fZ*c(2L_gXe#yxxWti4lF zH_|Tngf?*ew)X1TRUW?N_dV*xm8@2FV^q~pNeMsbmc8dbr_l$0UW2_Bzr4*APKeQc z4SYKTYPY^!{i~7oHH;FI9-)}yg&v(xPg>{h=bHB7OQ(^gxRB+8VM_c!yGg z=4-FCdFeSGWxR`+%Jv0100&HTt<=-((m{2n+oiU*Bwl1z6U(0{oczm?_p#AWJ?Q3f z-*ZIM(8NfS!>Ar$>&3D4OO+CF5?PP=>r+V`#J9g@ zoh18ojkSqGys_*@T;{J#>7qo|Hm)VqWCd<6P~<{8q90#Y`hoAFWi8D!u0>}8X!ox7 z>=2~)QMlg-ILb!dz{fe@^c9;YhbEO$O&)l!Y*_s7oEZ530G5mVAPn=<-m@)jX1%$N z#>O)&l67qL?oY57Jo{$0Q&KuJ3rz}j{n^dc<~W#-;Tit`mSeE*)b^&dHj3uaB8}C| zDFbQxbM3ZP4LrnP&fdPtI}gvAtz~3&DH<@|MYKbFvp?NM^*;T8?_C@gx^IjcMYjV? zzKm~>?f(Erano`B=O^h@wX12Yq&ibUEv1AXJ3O2;=lkc^*k=RtrlfSLZdQynzx8iU z=6%emaK21IqSLL;RG+;e{_*u=kHW6#T9xDx%+h&hMo;z7&byJ1V59rz+uo_!S~jI4 zv$EUiw!<<=MnVrk>gTZU^ru}<9qOl)jc=vf@f0JgA5-ebxvIUl7qL_Q)h)We*U=ic z8f~|TZQ+|nm6yp<7fq7^hp9u+v){d8M`gB0pHf$b2#(U2!@GZVeLc@_%A~c0uAJY> z?QLgnak+Ro?5XO<{{XLBLtC}hV8*{~wh-H3xkH&ydLQqOna?#>?q%$`-?OFgt#A7J z9Od22mu@AU6(Ui#7?*&n=qdKLQKh-mFAw@KIF2#FnY{;~9nbZxd#?`|3Djk|OT8^d zQcHe(&cSi~C)YhIBIaps99_?ZTy2UGz*QXp9)g|BofT3|QrUGOZ9?%c7V7*z$b-4z)(5>I{{ADIOk7SV)uwNtxgHP*KkDeOU!e{%w%>091c5GYddJ| zEyeZX<$vXjC#gu|f1%FbD|;h{TncUzpT)y$WdEi|o}8>#K|9+hSrWVVD{-3~eOqtD;>KGi$v zQt5ItYH&Pd4i+XDE`P$a?q`? zw`-(nTHecx3|n2EMU2mZA^!knyN_z;?5tvn+wIqR@&UwgXPI04Jxx=d?n&-E-0WfZ zcJLH-#VZFb-O)a=Z>U3kB5DyxlLLdeV;@gyEiI#pAGO^K!am}6kpBRMboy_Ijn0^2 z@f<66X1GY^fAz!t)84D<-XDEJGk>Dot@Wymd1x@v4??}DUSn5hFLiMx<-u}(Zh&Ke zJq1r`XueUoxMHxOz{giUf|l0NceCAYjwJiTp=0PNmzS+_;B5;U@Q6AnrpM%sTW=G} ztM+T?Qtr(3jrySf0A$tV@aCVXFWU60mw1OS0}MF!70z2)nR6qsSXY*B=qg!t2?L?F zc7p=q{%u#$gM@VMxAIj~xX?V`gq7C5|FqUC&`%Hl3_nYY`iLZqPYV z&D?x4{e5XQj}6?#WF*quT-f!@Mo*VvnkJBO;eY@rEHOunNCrj zrm}AS_Z6!-#(Dg!7VUs0hJiCO;Et7>GXQ!TwR0vjn#!JTyw^OzyBl-Xr&ErVRx$vqDAm;# za!nJdIO41>bJSw1D9Ndl83&)rwlTh>;c#)B z4!{pQ;}ijb{{YsfcH)h{DCd#-&^rzQ=NbJg+dK=RuAidD#D4NmnT>eOnt8X7A^s-- z{&nDRgB?=QNgenYCpOe=G%Wm&m zQt5E&w#MqghUFhTS?0G^O333^le^rmgnywsAj#E7|pC{ zw;T|9)Hx83GEQHxscfWb-05&j5xmpBZtWL!tS=@=;Wxdy3x#7N*laYL>tv5L-WyldiIPX(h%PyS4WnvrC)~wvyE|)tAV>_3i zJZ7^?NqafY1s5g6-bie<3{iE801?C{c%6Nu?IRM+*?+s{_qzTHnEa zYOrd<1$d6&b^00>q*~sFG%+HBB)^BfMI*fP3<5FIv^7mK>e@vmcfjB3-8lqRUF&Hc z(S~#1(wKRh5$e#%5?N}Qk#T`AA2fgCRPS^ttqRT^X0(1*WaJV4HJtP>86+9>Gafw@*4=SW|_pOPvn_I|D?vxm| zH$AI;<@X+7IHOyR#PLUo) zzE*-54jeGSC(wJ;cXy3;>Pm!(K3Jn20Ii{~+G~-mj z+U`|uL-T7Qe9^c2NBk<(mcp~$l~OUKMkcvX_mT|75&i+{f1s<{HnnvmgG;-M`$p1M zJE3!epY1Du3H>V(1{M$D*9X30C zDr4cckLgXI%29G?{!UWb)e&w9PY^{@A1CgnxJam-_Uo`+wmZiv#!CP9^Zx*W`I#10HyV}LhTLvuId)^f9mzhG zcK-6-D`TcEa-3sTz?*M$?NG@+Mc2=m2IJn!(|Wpq|YpWyxyeLns%QG zKy@usBE=3PML22O`>J~nQ`)kXH1%eYllY~fp!y2w3fhjU@m*=$eVjmYE?ck%_&~=! z{{R}tzP_JOxogQAd+orC02VbJ03Cp>1o2J$S29e-Ijv<-hRv7vTL(qk+Z~Tu+=D~0 z2)e$VY|Vc0dDi+}#EMFv?}62nbpxsQ6_&AgmZqw2sfTS6XfUPy)M52WSADgs^W}rm zM+g4^9bd%utjQiro1*N+ucbL=$t9Sd=4oyvo_on8x-ItU817~So!;ZtwLj!qh$r6G z3nKUb8mfqek zn)dER+S4%ZNB2iz-#Ik-*UailT{U0V^vZYFa6BhU(~C!^M0ZU)3^l`#;%{!dSNhhD ziKx$fuVZ;{6#66yx>mS6z`m;ea(fQvwPjjawyksaTVSx*K)d9gL&7NP?0tO);3|9D zxUOx*mY=!}8r<$?Z@n}B0DZsOJ%?fV)4zY~{t4Gf^CeD}x83*u03?kY{UJ3SJ}GoL zyxk%IS>Q!JYBo78JCp0$v2Co*qRnw|#@9@>;zt9gqNwiY{{Yv;SkVQ{_kU`(xoZnM zEHsmjHV;Al<~@ghrEOhmT3?D#$99dUBvO+M!M;Z)yo3A4xfHJBxzWn1r%I09dpmxM z@j2V8rPK;ZBx2i6xA|a!Px;Fp=9_P6_Q3b|$I8>7J3vkgp2(j5r?p?a&~Bl+FRB4= zr$V_BL_-H&!4GnI{8m$9LwR`2zG zU68t+t<|;5wxKN`y3u>(TkY z(HzaxvRz1SCF6`FU~q%51JF~fZBbM0_j{HFKO&AAR`ed;rk@3}$X@R7813UfbZ}h% z0Cay-+L?25*N^7T0wRn@8t@c)3edt-l%>m0Wz^V%#YuFZWY_IQ)uH>0oJk1(0A%+b z#-@|PHr`3qH0zUprZ3D<1|%QsuqX(-XzksD;byV5Mt3rAZWRdZ>Tk1q^`y_-P7F2Is{RK zMd+ffX%Ba2DZYZ??bbiM>T#cMdWy!x&RSVp%eRjPM?c7W(%;`L!C2y7E(AD8$MgEo zJ9I|y)ij?FIEPbDvD_XO;Rni3`$M%+z3{!&grVegdqc>cGnFISvTbcz;mKpM7nXC_ zR*ZV=)ZhqN3R-;mz)x{0UIrFC>1EO2EaUKM?P%@%fTc`noPU}U-IqPHQ_H2r+$ z#s1HBc`(0Ez|}2Q3#~Ce#PGCB)tHaIezeCIa=rc1#}Ku*7=ceGyPs21=vtwan`N6V$Z1zj+s%2|p}}$gD+ImB0!~Cjbuh*6IEuyU<}vTbVxB6AzvTIoLgq zYNq*&s&m0!8otMC@FFN}{9|_>cU&Dg+0UT!BjSqI_&#)N&lg?l8f%o5>|JDqxGR|> zKhlcAqO&x^)%K9)wY^U2YFK5#>rg!+HVtv9 zS)Fb-M)jQdAQSr5^`*$$)Ylyh*fp(^W=+~t%k&dSc(eQ68` z9;ACvJN2al)bUEaeW@0Z_?*QoaxVo~H{@&R9}j9!>UvZ&M8UGW2$*uoZJkwzIvrHqsv|;189%HE>1>+^E{6S0~KOkTRP` z$}#CsE}=D~>TO)3uNhwTYgf0P!Mui22;d)D&yFDw-zoWzK~pU>Dl+PhmIbxfkOm6| zp^h?mb6oxAop%l6MRbger>W^xk))E~GLobmf)7gSVb?7+I0m1qlfR-|fynySEwpT< z%(dL+rkzndn7BO)deIVnm(Fs!~XnY2`=Do~>HCHm@zT4-N7y z#-98#KX0d{Gg$J=Nf_4g3>jzt09X9#IizXi3aj$}01gFj>e?&KXfO1}6Y2T;s(xtC zp%sj{vX6FiD-NENh@2c=rj?egbgO@|+z+%u=jMo?yY4*=E#8S8%ztmwW%I3%E}dRK zVk;ULTIxc>>csY|G3rv-PNvDq?d!+h`qidn)B8J>=bMi%F||i;T69q)QTfg?-9jGq z(HlE0ZZ)}vCAPmKn{sfnpH=kwS2uMX$J-hi(UMhdyGX_^^tpn{VCU) z7xr9Y{x=tPEPr((jtTT{LsjS`-`m{Jf)%rfT-*Tc-|CR}_deC5qwDwa-F=zk7uNSI z1KP{Vw``B@pK(}kEVow#oU<|M{VMmEx@<*ZVfL07cOSq#2l&+{Q(exY`$vu|TbVpL zAev0#<{iPNfOi8OSN)+}#f^jBO3E9`hB3Nno1%~N&{An1YZvP^ob7KUe>+W_1RkW< zS1tCJ;>a4zO1f^KTOr8V1=ISlA9cS@YSYtGJ6Pv0ucDqcw6z8sC{N10DT(zuoj8+Xz4c@gDt{+V_U z4eF|UApL7{;?BxgZ!J75&_$|9F?9mSSdovJAY;^LIqrH4j=Zg&pFQo<*hD_gvHaIl zN(zJnsc%ercQl`wT1!{59jDo~?Z(SlO|`YJ&pq35w;#p+0=MpU{WruBwT7UxL2QHw zbqwXp0f5qvmK&*8<_t9v+TD+rCZ^eR{PU8_nXw7!?))}orHI2_jiQb==Q`6 zq4}KlMfGFf(z2XYt_^d0^3$<2o*8{L&DN(RZKfy<9FV?6$Dl*}{X5o9rGG8--f00~ zMcWh#FtM+oAJ@HYr;0BSJH5OOq%iW6Pj$)_9fxpv<+|pyAn@&k@h-Ke!#wer%(3aR z@gX4cf7*no%cw-){AtE#1Y{ z>AHDb-MQVncRBW_UQa#s+fQ=;0O>E+upXRt`c|xQr-!Z5=HV5!c~^815s_{WM`8W% zLEEYAP~u#4)w{R-ea6VtZS6(4)|m#n=tA72K)tviZ8Cbl>OQs3-RSy-t;9+qYs9&4 z-Axusk)7xrtcuwE!F=FoOM$OU|COC)p1`pro zmhI4YtoeC$Ls~L~vL=IXC61iyCZRAZcXFpbP(5Qq{{W#d>(Fu6HqO#(`;@WMlYFju z=Nb3UWgYYNt)rn^Ye>_0Zd}Ws-q_6el0WxPMqk&~pu3hmG8uGjIm28(%X@Y>500q4 z!UNye?vAbq0O1M5;MLA(9{LZ6-jPY(PEb z>u_fy_>XX-_*dW8rc;XBul4;Sx{X$~TYp-An>6(;9RC1kK@OL35%Bms=D9<;q;NXO zI^=W$GaXdI-Uhz(68^#qOjRGp9U1E0O|=J!ND3P3q_L6%Cf2b!jA);mxL(YBMCnebLwf-H%~f7aCp8p{Pk| zaS79OtvqH%lm1z+ss8h~>rIwe^+^1*Q)l7ZFSw`gIrToSe@beijCrN~f5B$)n`_v| z=hFWGha({5(|&5N0Bwtn?6Gqrm$ zKECyn6{q&>?>o4)(d@tyIOvP$2X{32wJEM%JBwo0I%SE)ylU7d@$SdDs*Jf;H1vJD zd3u^8>e_`3p6{dceUd6(N^fQoPZ!y2ZN5lthXF^Yx%EBjD|iN?UiSEybm$H;ycR#) zUs8WgwbDa*XX2|#rPi&!%8(F0v?DkfCxs)qKHSv}Z%Vm^;*xO~(qdoqepq@)x zSWeub#_#qQ*ZNUy0o3GsncUmyw%>j6`6~Lc#X5PDQe~CX?D}*cEKq~SKdC*c+Kn?S5#thJq&1)NK~h^Pi#XpL(?Itd2?1 zl%@5296tM%MQ5k`)hA@Aa!p!PAXNdpbQkm9MUi#t@)@_#>SmsEs2IW!y$?SfWnQLtNgynwh5e?5F=%Sr@I>{J< z&=j?d@#@Fa(|V%`Lt7)Gywa>ZLujqx`=J}AS*#z1&$suhlR90@cP$OXO%oyc%ZzOl z>j)>F`30+(-9BR`a1eUtx|i`Kh4?ytt{!NB3NF-P{-h3~q{YFea#uGhv5R;!shx`M_~c8wm&%^y^|eJp!T2~ zbRoK$D@;LX{I4${>doz44vXUNv?A%+<)YtPw+klR$e-g}ct@GI8;R%#Q(BLmCIXpX zwLjkf0H$Np>qAM4;n?dhye^AnEzgJLOMu+6hRdMzt`b{$l4BgPvN$}c9OkFf^()&8 z!EtAB=^#1a{qLynU8nZWpRFWO>k!9nct;YT!2EqGby^tLt?Z-CYx*3@81$QP@&$Xg z91rdTSu!+}%FU9*o=tQb--WO3GjSjKMYD1wL!LeBI?~l`;`8Rn1MaErYT9g?Sc+9+ z&j)kUJ_1^pyk&VSa9d7%FSnT=G*R#><*q&;Uo-H3vn1oN9&A)rH8(n;dunl0#z$(; znQK~e?^)A<+P<24qvxTdjOSX96+DBQgo9jCTAh)44nvB$1f^=C3bh)6R|}C+#;vu< z#xh1LRtt=QSWu>VcdZyf$oBQFh_eG~IryoC#w$7?c&YM8&N-{OO2vk}$9ibgMsZbK z@r+eDTWBVWu+Y@|Y9HdoMJ33`E;y_?W+T?AO?bfI6G_yt)KfPnx*X+&WZd5W0JT|D zUOC9gs*gLd>sU^=V@T00%{gLoRhW}e?k1cwST}Q38whzlDK>M`q*OHUnd?f|A2o<} zQu%dYMo&RY<($*xC!DXdnmoF)Bc&1ZO+Z}aNR>J#YH&)C7mssVMh0@2sJI-|$|y+w zf~8gMpVGIAFnW$mai7Ac82^o8|Amko@l>?mPnm{lQy-O5ES$2W#{{Yvea?;d&v?0#tI7JyI zxqlGr4|g}&U~F5*#s_-P@h+$>P*~eH%ty{eV`$OBZZUM3jrh-}y>M5n%=r=5P8_Q2 zlE+SVfJ$(y>S{}?p$;O5<%jWBwxeS$qyy}~y@9$i^NPh02-wLl-Ruo>Npk30)Z4Sw z;k&h4JzYUx;G4Mm)?TfnTw6}9o%ZebBpiY(I$hJsHu9&YJ65%pv}{ya&P?`YWWWsxo8@}uPUG{|kRhG~h}{OfA_LyOA((Xw+rqjQ1Pexk6Xo#$_uU) zTUZG|3z5&YSyf8h)nLl@_Zo;iLj3)&>%h#F(-Id5&j2KAVlPStN z`h!&NZU(^wL-#|csrRjWZxAh{V^5kNwdup}M(xgjyr;3JUFeqa#r>mUD_vOiE03K$ zn0ETrCD@*OKYYoyfnY&%a5A65r}&TfRhtcR`quTNfQCsvSA~XA$M;CA`K5cAc+aLk zrCh>k(PR)*>8c@yG~~EB4m)f;^HlFI zUgW9(76cn(LD2{K{Od&elWvIUVewRV(TzV_wORDwaS2cLw!{5~Jx8(YRxY$J_(!5T z#7Fx|2f?*)m{cTW83E`=VcxKj`Z!VBZ>@x%n-~Fcc_U}?@y8fqUe%CNc)4b;r z#N%=topn@`|NF+VJ}453h%_oC-61eglok-AQzS=Cx*QXcE@`D}jM~W2At5!o8_6-c z2Mp%#`Th=na?Z1JI6Qm5@B6y0*VRY$2}q~H?SBdI7W~okkBmfOS{!@lwp7Ityaln0 zGe+f5HSL-eo0mu39&fIOmd(g#%}TDb&!LvJQ)HhUA8~HPvTZ!5b>EVXN$PI1a(DGT zyKBYknzdLJ-AJ-!GEBKq{f^1EnbdGr-P5PssHR&hQx*U51+!u1Zvzv7R=DG^TwEbW zRxL-@;WseD-}t4hcLGv$PcabUXaq};w@+Il8rlByVjdy@QjP;JzsxR*=m^cy`0A4S z_qf~1uVUwgy^Ex`EwVd!Z)f9{jz_3+_}gz=oa9xv#w>}Pav7~;aNhuNDgmXa*99iu z%1r?5$xEwowEgRq?qG(wo%g-V>x_4zmL&GiYCS3a)?hr%tRZ|5q-Has*WDN14p04! z;vyG5a&0sBwVYclIGfjS^s+WKR(QrRS%BEd{xmxbT%61Qk$m))PZ2j>x!m1L(Mz?B z***BEGPv;7qf44Z>Sp}PrTTVj?b7bNP`L?J5(yyUBmxHt8@aK?m4>(l&&rua>fl0X z-(tMfPD|v{=1WbG5$!g=i3D4HMY@UEcDOSbrP7$4i0D&3XzG(1LTo7UCQphjDD@F+ zt|07VHEW4q+TUdscc9$0=X(WH?OMDtiSBi@7ZKQc{K?}HBZta!{XyRQ2lz3ywt*tZ z`3WfjQQYG`N4G!pZl3Sc%HKWokw_p5!*yHh(6{#2Lt${()e&3_J?47f_S z*+;Wi3NgOw$1pfOJcTW4Hv~|bs(4UD}R)9zLZdER*o;e`HV+RNsmsH2Hd;xmhX;Ol=R*h}ULw1y>~j&FGB&vg=9 zZSoRSLFMkT`kyVXz8U%$cDeD3Txa_@lK0seKT8iJUU@TeVIFTa4ZWlMIQ{a^@W_tA zPL0G&yb;G8g}_N8sw37{JXi65iaR}!;;Cb=s;}4oj0aJO!l2ixndW8XA0=Snv3P$ysCtKpL&&0|490S~?njoi zz+1JKGATu{Z9@XbA=9>%EiL9vrOp6bRWIjGeDU{*jikcM~9y29Rf>`Ri{1G zlKez0QX_lI`)EkiC4!&&8SzJgz~(n#@B+~Hf-2SSRtRH?0m0YQlxuh^oC$1p z77|9QZgS3Ha`!X!VHTF%_1JF%{e2d{1zw*c-qUfbo%no-XxxX4Ff|=hyyDYF#1)ue zKVg&cT=n8)uQ`#=RC*%%Zlt)6C<7UVY0 zd2DJL4gL3FA@*Vg9o4$NMpVDtvcEn(efPYX$hB~Gq!KN_p8D2=)6tRFD(Um*eyHjx z665SP)EQWc_ILB5O1}7yUYQWkYTRbHHep9g`-=UA$e_=AnRqCVQH>OXZMxblLR4!J zfBf*N7bLs@2spx2*Iy)e(;bc-eb zM0?B8U2SsTgRqVs4e*JjR=UcIS#RyCwXPnyQmAappI5XA(As~#S;=~)p_8KTri<9g znpoYJLk1{6b9bi^Vq|>o1pqUQz7nukyr5^k8xNHlf2Na-w2tiW2>gjkuH-BIEJawn zDB`pp%_i{OR(D$k(07qLJSZETi0}2ZrO8u9Ssi8ThCP69biKy+!*0ZUggli`mryn; zt@t4x_U+9UZf&EX|K(a!fTLlRtQbT{Rrw|}SD2bJ#oR8QaO{0q$q-CG4U57=mk4sqMQ{Nw;tbqQWYu1Kp<-nER&4J^)IZrcbts~%|`pDLGrp&5aAWU*M>FN}!Y zxuB8STxyDmXgsj!^k&Ek-`f#*6_;0OXB2)BMn%_lx_zl_dg?rovuVj}pz3d3-xMoR zhf5v$IZ@A=EUu;_hdxR0V__G!F&K0Jq1l*H?ZtcW(#c_Nf;UmG>0C7Q-?u@EJyV? ze?gkb1E&X96QH20>U_N#XLS5#d=&{%`+7KAs7~ug7JgHSru?|u`0JlRVG|JrKxOKp zbR>Ob-N*ZISErQytih~Ojnwql=$NOXB z8|%z}8}R$h7h}9Ubr2ItIT0q9AinSMnyR2}1y$Mw3FU7~edypQQKOA+7Qfh3xzng} zX&K-K3|?bcab2IE#fzOp>p07C`l`e)oD0{U6hThl?W1nyG=-E%a*`n;M zi}=SwX4NgOL6#Z_M#qp6EIc^aSa`Cbj!Dh5He1Ki^b(sjF9n>9!nNKw^4STrj`d>SV4+ zO#g|)9|7qu(H=Z1&_!&CF(+o*Ik}Q_Zym>k9lm5(1=?W>O;^v;0{>{q{ISiw8)I=k z??=#gx$mAU3O31+L);mR*hgUQD+&>IloweRCp2d#jNakt4Z8!dvkg#Q)0zpf)3GL1 zdVl-#kP}#G*-gbY-l@l%y5}pk~kb#xKpY;tVTt!H4&(JNi3#S^zq6>|&qf`q1KXMss|U<~Tfu8OxvYLzJh%WsxR!pg z>*>~*+F#qJRSa0eDjgFD9G$fx^bfx`hJBoW0NsonYG~9~Safv?Ax*!W(z!%tA|sj2 zVFH^k6vy>^t*)o`lv|B~{VCJ+K<8%#=j`o~H71B$mRii@Fs2#n`~$%U?D+gro86Z& z3vxT~1KqCxf3}+UKP8~~ayvE8*Lx%Q+vB;LpCfU5P@G>9s%IK~$ZjiKu;taYtxmj6 ziU%Ch)zRC#!{bxt%o+mo4hQmcCtVv{_hbC;kwUhOXp<}EUJ1Gw^oL#d0N1)4;3CbB z2oq;?#MjrR0OwD#;ZIvhi@l^xMe%ulx9_6^zA~zHRg&7x3SZqBEkQ>4-t%5tdP9Ez562#QTK3BF4Y5 zZWX<77V`2o)5l9}Ey}UR+5O$*(*%|Ay;R1^Kt_XTsh=^UH+nU*B+YM#_br%pTREKx z$yvLG?>zzH6EmzpJXgBR(nkDb?*82z#Ra}0yp{I+<+XHz?#~gl=Q32OZ+oXUqTI0f z0_nx*;gq$gBX{Oc2GBxNgxP7g*KiM@;~>NT!MF^GNkx#A@#@ zBUhXbh#p5?o7Aqqo`g`W`Q^KVO5RGCDBk0(P=3|2*n(zWICgl%fhmh~CYgaa0h{pL zekdLcuv_DsLx4yvf6&Dp9{O=~8o zW?GQW!cp?LP)S}nU1Oq^mz1DS>RBJ`DJyZd3kU0jd`(|4 z%_Q%HvD5wa-h%#uBiAQ(gups<dl-!bfbvyMx7hJd*|K9w{KIRU=eOcM2S zCkzO@>hd-R*Kx-JsC+t+%uFl4gh9aTwmohwAHiiNsD2ZpSj9M#|Ca|JH{=B*be+uY z4}4;}7OBY1I~zPSjaO#cl%q0H|45crN2~OycKK&nlDozftL2>6*YdzWVu0WD>epup zpjn~=SRjp1iX0R3x7RStuxTaW8A=a>pn^Nkik=xAv_UO)@0M97&o+ zv2Kf{{*)6LyuB+X<6|Db`A*Va8U3mw%PXk>cJ~go#+mT{;+B(4oJEFd!R$;zL~i!8 zGo%V~=gS9(D;y`%>*!3Av&UihTeWqT>w=6~`N{6~W-_KW?O3Kr!(?Xc;^IP(jZP`` zue5Dinlt6lM&@z|d1#|#5h&Clhfl<@wxj^ey>6%=`NV}pdt!RB1?}wb>PZ#ik*18U zh)jtin*H_kmP?`$y1#P{d+lV{#mS^~Rim*5(Z(gXa((z~r4Uv)RI{UpsePPA_V zEsyuV_5WyiKS!d+>a7RfxvBa0%; zyEwxxkA;`tQz-uQuOf#1$?~`#&{S64!Ez-ts}|WnDqf&#_5F(|?{yEPHm(;0-wnGV zP$qQ|hV5g?vhW)fb^kI18S1YeGc+VH7?1n6EUFLH*E_E$q9d4jO%}fV`8@g*^XD92 zERFbnqeeO;RAgaJlH>J7%5+5g&&azbT>Xu}_GJYHp89a^NwM^E<;e1=BU6|-jARpV z*6-&3A~})rd=df^6x4}VzD^8LHSk4D_A+C)-n^lcwd0U5P8u zXZOyg9&xYRHKO$)zO*N7P{Ta9B4Xt3uqMy@j(;Q*?|JQFjT0qgCK5v*b>BiNK5Ii@ zV-E^Vzq!WP)0@A!oG^OCo?6}EEw#I?Hy?BDU^t@p!_QTl?gDqI7t$t0)cp;jc}blo(cDYPjbI^Sv)n|v2Epz$@Jy6aY!d3Eno z;aJVYE|O3Aw%9D}`wIHqcOd|Ile;5fljr$gkBzUZ%{S|%@I|^Y!l@-qETV&8qKh(I zSi(eqS99^+rQgK+EC0(ey!9$tCeT{a>F@+`Ym=wWa{CX+f+mgoInal#?}SXUF+s0_vgM_DGp|FG`A5nUoi#6ps_NoU$!7Q%!yTPLJdY_&e$^J zB`@@$u8igL`|GS82dp^L&MNx8_8_1XZ=&4bU9_qErIvQgM8d}>w7Uf|iq9WPZKsyd z{O2vAq88Z8)$B1TDY5uG_~Up_#2PtA_bX_$<+gqbt)#rQ2T#ym`Q}90d`VKk6yJv& zv~5(*g^KKy2BY!zk5X?IvzDUTb(e1dJKjAk9DLUon8y}alFT_$H?W_b9=JCayogE_ zP1FrXd2AO#QW@dapM5#=GCl9FDUMA|Hb{y4^Xmu~IG}tOKQ-P;k|X9;?w`&9N6Mb_GQ`9 z`!IY!va8QwT@lswou1CEqr3>01A%&rp)+_^J7Amrjo;IeeMhL)oz$+{+GEvttFWnf z$LwM8jfj!zYvX6W`I3Anz#mx5(v{H3z-Th742AayREJYxo_j_CCiHHr51-1MDFZfE z2Q8Vb9tx6?Ag|es6BMcm;wLbN%B&Q0z|t<&8zZlgZp#Ptzuw4D#fx4I=yE!4r>`{% zDx9IdK%ZBK@zm6)hT}SIfdQ}y5Nalwc z8p0y6J?px}m14(zMrBwGwspqz5oT<&+ZnI&?A|)xm)p45vSw9sfOA;;NN`_EIA;&4 zg*I#&jC)=++4auv_mEXg#L{5BV~P4$VRLi84*%eeNO{|U5_#>^OO9XD zMD{7u)IaINs1(#-myKC?;ZsZpvG;#HygbhH=%T?PKCcJYsiFGHjq|6ezg`s(_sjsv1(QX(LtE1!(NW-6Pdv=$ zj5d^2XVQ+ASQ*ySP0;5FG4oGhT0TDi+kJ2orSGld?3L6W5)0NQd-$C>BV3RyGBNovssdI z$axFet4La1EPt+A5ldk{giEr`kOZ3|o+X?_FQkO-YH6q2g^J*cEi3$B>X?uBwtcC0 zb>{ToHAb@TbU^&6YeV^u>w>LA+4zmM+CJT@{)LKtv&2+*2s}exx4}?cl%=D%t4Wx+ zx>~r~P(K6L5V+-3dS_K&hSNMuzCpg|iYF;4eQN6d!`sv%bm@1r=d;vuo!zd8Oyf?c z4;_73Cgz&nnc)59TNra5128n#Ie2vH#()<4HO_?4JvQxyw3k(;h`N_=+CCc+Jrb1u zQhpKm$E7O2!f_mw23VY-GQJ3ixfrVgsi%k6Q(6|oW6jmK)5iP8s+V0EO$P)pfV@B1 z@^Z-LRwK(&hEZgv86kfzt>jxIKTUhR(-oJ7=2$upEC)(1+Anko3jOHxsdMQiAi8 z3NY`?=K7U-kQwCD_(qCxa9+ck3g;QG@3*pVRmMyx?Y~cX-#|TYdE3g#Rq$?v^b6)- zFJqT_r%UI`^Oifw&wo#=J@FVJ2{^=V#DZenXVW21+?8Lngiur}=Xa%M7JhT={pzT0 zM^7fBvPu9%vT0AaYyuzB+hf<&TgCs{2(%A`T+yN&NBk zL}!D+FR~$b!<@ZbLv7>_x2W_;3D|lfONJjP%0_c=+s)+bA`=DT_kBl^=#)n{i1{oP zIj5(P-@v#`6GtzG5eM&KFPO&?h+71|$=_oZ?UzsXyOsX$!W8_$AG1KKm`s8ZPjfIMC)+%8PW zW0R5^vUqKp2bS!3Pwb{WQhXubyeP5CY!sV)TENk$<<9b}fJ#DNUN$d!%@bQ0(~9&g zT9M%m3kfc)#+UP%o2DG5ORu8jo9Xb|5$&qoR;dUvuW_tRU6_IglrOf5(#0h$NSL{I zCkT*)Uwll_jR#3;n3`N;y;h~R_BTm!#@>sMn{;bq)qM7lKN4G*n4@2k$u(hxmTBB(@Gk;n|^WLz-yJi)J2uTye`{*-)i4_a}}!n2Ds1 zXoeg)hJ!g@FYPP*Wf1v2;fe5_yz&C`#cTA+L0PbNazE?hz;J&@Nd<9Ngm_TN^}R_T z<}mpmiAoloW6GZDPA>f0LS*dybDwZj0bNTAlOXde(2oH0&|$!30Y*2Am?SvDf|Ncp zE%ClKVl;NiZuq{v97b!Vba#c@{fl+ik`5JK*&$MQ49+0trxko##6)yu#eVJ1Ox57N zLr&qMGU9h8%|T`Ln3|W_!qba_fWIt#+-J`Ef zbFR|lDUjDhA5@=XwB$hj@Xr3<+DW@t9GPP6K;=738mk%Yx-2z%-=1OHuu0~2n_EQ? zE|YkYV=^b_m&!j1{j!u|?I*MmftO!yO8J!u{B@?u8a%&6Mmwy1GBSvJa6)>Ks&Y!- z#4BQgI}#t*iDIzq)wi0CIdwbCIY`PgQCM=AMERy_Ho8y4G96wZ^A7qI_EtpY8a;^x zk9s44Z#3~{>6j91O2-&sF+i1BIE&rXvUSOZ=!BacQl40U^R#A6_6@u55&q`AGdzmb zU}Rrb>Ik%^#lGhQgzh&T91|JX1$_(UGm1^!qp&s}ev;u_mBYZE`+JeTztPqY@>z1s8PhFNd_gIdU0SaFpowhGii2<_$1~uf6}8q^+-r`J>i&Z*%V%rZfqoY`s33#wxdqJ`jc+BM;AhF4raoj zAe6`AUAnkv?smLmP*pYeXZ7w7-p|@N{?Q}@G$>oZRCp$y;Oof= z=bA++VXRkPT)Ju-;*Pe)#70>K1_i7GrcWLZr`m}-J$G$Nnwx7phQ^JX`B>V$A)z)LYXHiva+=n8!-x_ed@jKg%_^Jz+P(o8mMHBRDVYzN1$3k z4t!9{(yA239qxriRRtwPtaWO!=P1O_bn&Vs!c~uY9W2$5eD8d5&HtHPbgjSRNH)EK zYl(0ADMuw}DV9%8*hP1dXIqzzd!yd{tgaKDTo1KS`;(Mu6mAu-Wn4JWW<;->Ly!4| zY{+_hB{f`)fjypgl-Auk(6T;EdJ`AF36$&bCZ9jLJxSzlGVn57Z2PI4DR=c{bqe_y zW435JGJVA97n~XQGu4XWlu>rGQ^6NevXUcqYrxKJ$2@wg-|b6#)#JkfEvA;!d}}Ld z+cgow#CaA1{_&68T_l|gN!yV4@UK{IG$T1>ZUH$Jtg9+F{o#;tB?s%`AA21|WOt$X z{jAZS)>jT}Tw*90vK>k&)PdH;-j;+RtZvZ3NdI9S7frk0r>G;}E1c-iXueh3?qWXu zYT)~j{n4z1z=50WT%;)Od@Nk7g1J@N(#5x=o}bE&Eh-{Bzju2!b?9iE{Cr=vCpp-A zJ7_cq{Bk4p$X##obrxK}4f^ZlkN4J(4f^u}?6Fj~Ce-d)OZ&P_Lw>MdxX~=B`un-> zB772Ni}*56pZ5j`#WUvarEJR9 zSE?6rBf9@{UJf#iS$|9?mLRBY5!YgfW!60#u$(uw!wW|t-smXBP5;44&<_D4RjhdR zYb)gsBa9uteC2&xf6h}18`c~)?_4Ba1;sBW)x#aNJ~IsIz5%6urt-P|ql^OUfJb;| z;`kHOOT&k-I#aqDrxgjF9_QYY><%=ETgt;hZYMYxg& z22L=G3ZlL>p+yiI(CSbH(~a-iJdw6KDs4*oUuzV_Q=#46PB3djco&}zS#*<3@xqL0-3{ByAD>p}eU;4rk;FFJJ6yhdGRY5HUa_tw z6*Zx>{(c!Z?>_bva?gXcv0?_xAe^>LhT=+*|V~~zLr%0AKE1R zI}4meycr5u0(GfnQ>xZ(suCYw1*`0&0@;|>5&lz*{)E_ncDK?ZpSv9pV)ywio5F1# z#!!>zf7!?F@EDe{k7T~&E3j^4s*VWQuWfd@@5uTDk#Q5AS=ze&{MN&H|1!MEY^2At z7CrBZ{#j$j0x*5ebGIG88@u~j$rs7X6nSFL*^X$Wo?c=nGhpoH^_BLsRNUp-A(T+9 zX4+vzngHnV>N#L#qQ{?QkH{gk@B9f3dmetO7cynQ_F$ehNkD!(dj~eV4|H7F6uv1! zH=hVpVM!iN;kM+G%#HG!>M(BIG4ii?ODZP$|JH`8O#RANT#-rKP?x=AS)E0EO5R`- z^Lwi#{<6uYqlfA;=Xzvccb9LyDP2J5=e$|N(>)ExtxTHMvHj;A90S0VAto7D^U3p+ ze{SBZ*kCHoWR8)SPwY1FaOgk%(=(QFyujrAaP1+Jv4_<=iGL(cqb8vtN^Mrxr}FI% zs#l^8Naft-ts$T&E~*FlI*?hT2MRF zF7PMV_f=+;d?e_2GNC><=_F3+=t#Y{iXm*mz@fAGA(RwpUV{#4Ul8QfJ_|AJa9NOF zfRXy!-ILO5SUoS#jrUdrU|)^zda_qG1y=e-A0Is?p1TD5 zW8qI~wT+D73T6~@m)N!#T<-ebviy;5-RXZLCNyYQ z7xhKppxXOP^2Y*L1cvGTt0bND!wZ6){l26vW&@rS{M@B%P1_@>&gRC(y5g^Q-lXga zUr@S~4WVPZ^OjE%!rUK}2jnFu^aS^RbR|#Ht6Ov7?eZ z)hp0{GJ5;exJ!6UHmq9D{H2m*s=z0Y-SU$z8L#}B2oTEz#6?%_h+^`rUkZRHwC!%n zT=S~*P&8AS)*l3?VLk_po%IWi!8brz$0OZV*NfL%e+CmAk?6sSXdaE$}H;c8;6K`H#dtCaBhU1mvQz z^coXO_2Ox`I4+d6lOM1s&V~ZqjLQ>X(-iJ<1F8f`tVFHc+uhUfIGeikj#&8A>JOFa zq!eb}G*uJ^rBgL5?UKR0E#3c)o%w$Bnnb;3aiM(rrn~2V=Dg6?W0}cte6;#Lx17fv z34P>-!x%+Ef2iCmAWOD=y&@OXY&0V7zre6AhLH-%O`)i9l)T9Mk@?$;ZCw9T-b|dc zOM(AP%vf#wy&b>?qR8;LY8t%5IbqA_Vy8eAPh~i^L>M^tn-F`KTeIa+o620B>WMq} z@}qg|X6?a68a&*!PszW)JzjcswC)_aM_vL4kH{>(^Xlmn(kef>RbMcJ9*D`gEv*~R ztWtMlSzgGc&9ND3p*k_?EU>r5$RA+%-F?gJFpw#~0%t3Jka@C8C$>@mkiK zn7qZDmXW;^Uc()ax>O?+(MVAF9n2H(P8zB`nKYqOYEBJS1P4$Nrmo@?a~}le*yn4A zwI1TVWLArVW|YhPocz9DdQ{-#n7qIl-D%R4HkNbOBYJCTJ8}O=tibXu^*@6%J=W@p zTmM7dZDCWcZYufHLj>ZfUf@C^~By@&{^sqXIJ@z=D-cr(~P~-^|^D-VH6ZPt1N;4xx() zFTcyMnwZSC;nu?rLyv_z?(R`y3h=dw+`kChgJ+nY*Mc(A4WI9T*790 z8y}`Go5qAxth3H7wYr``Y3G1$ZNaoyoRKO;cewbiJ2jRS->*g~f1W%pE`#mO!8qo| z7yIUqyc6acVQZ1zM9Og!og>7@GQ*Vev-R*N_0(PWup%fx0C&xMz@;oB3QM@LdB z&TFl8v}YY#{x`Ob@!k(`@1fs%Y(Eey-%N=+Y8zh3*f)4qvern7^eMU{Tt5PG@sM^e zXQL(WmZA-d`aajS5rec#tvC2Z;C(GdCc0_t(FrPc$IBovKh{ao2N!5qVf-i0_~K}J z&-62?)LAK5084nCu7=?ZH~{gE=0v*2vR`jrWC_RgC2|c>c|mHoG>UPx$yJX;XicvV zdrBCf-{yc~^lTEQVg{TT(kQi`*_=kW_Ao9+hMI@MT}iS$e1~g+Me_FGYa~1*Dh2ce zCV-GGk`CDQv^!^(Y*QceKD>DjyaHp^zyFccRA_5|xnGq!*DbkWIHNL_p{WcLk>l{39@TKWA2(_vmwThytw<$3m9B{iyC=)Q>ab7OSDZZ|Jqiz73nMR zOeop@g(!NDmckcbZIn|xS&3V6J$5LeeB)t?Z5B*d**4skkY@N{%94GoUwr?^=&>p@ zxbg5};@M~SDY5mes6f3z0iDG?QrS`L^I~6sFVEnewm%tT>6RSy`Q^Y17nuaT&U77m zFv=i7qSCU>%UXbT<=MJvW+bL-&<8$#F$$7aZ&5e!>uS85deOe<#iH`?(K=X?; zSn-nO1x@0dp#l0%&>_3>Ugagj{A9Ji(g-A zk^o)ks&NBt`K6lV?6`&?zMxDF|zAzmFmFz*VE-)9n*bZb^sYt>O zg~omk*kXT3ckeigsOoeHa#8}vU?~J=PI@R+5`5*WT3amLJO(g(&!)l z-L*(1AvksnAkco`;Y@pwl*kxDuPfiN2fi_Ev3VA4$%)MAGF=s!tGLJ>I)2>y8p6M} zcE(fzBArCy45rR5a~F(NFamD#2g+Xv2Fm^vh_GEal#Zh%`(@^U6!mJ6)K>U!MJI$2(vkUtivVk$`EMIk}p7<-Y zIVQdZ+qWmNPg19?mQ7_@kX2c*k|?WjVg#E5x8Ym-iJSWy?F?h%B6yw$4Vv?dYuN^;vb(;R~Lbw9yVA< z0KzJZ!dvMVr+7|yH9BH+0fN%aSlY_r`lVGyXeAco*hZ`OA3opm0wz)0@t0t+-8ucg zF$K?|mu2gwF*%?=6RVM5b>4sOGhhUL-Hi~?E0r*rODh?Y?|<$*L*Df9tV=rJ@myqy z`XM>P{GsvrSHxZ!*=P9%&fUf(dgg5a{e2=u>-(D_G4iLCW!zm5bL;F0Yxu5FVU>dc zm?>@1z=@m8C>?|{Y_TDN4lKpHfR`2}J*h5Nb&r%A_>wRk#eYTsOYJZ-*%4>mMYp8k_izReir(4$; zG75Ded*qjTh=RtONX@L3G|x2Sx0S61cLiThr)XSHu>vV)`v5T~^_O!La*soI8-DP* zlm*jwyjxmeSwx`qT3eHeYrkruzLZpdRLaYZb*Nvtd&&q^M4}lX&we;ZlU!)aID@(` zw$-g1?*tdNw7x$SzV0L~?Vg%Z23YAW+|&cnR0e5pM)mpi*_*}$7=?}0d!=N15a%$Y zBFbLNpQd3W+T$`L6DyKHthsePWufKT#hc#J2r&OtGqps$KSYtp{O?D9BhFnuIL=cn zE_v(i^+5e5;60)VYPh^LRi?&|{QYfFs#+?3@2&ngv$sk6d{`)c+V7GIz$PLx_VyT* z-E6Irt*)H6Pf-OXP|2>S?+*jtbw-RZSDB862diL+b1xlKUpso~cq<>a*-o{5ilohje8?Zak|i>YVaLU8 zPIc)Eqo-f`>w(^v9n`)zIofiJH=Lb_pB$%$V4nrNju)NwA8(*-hUS>vtG-}f|GDck zw_$W+2eTx;hz`jpj4T*dL}jhmMcqR(yxSRCd?@*j#q2Gv~mzXVnBsRVy5*=f)GfX%Ij1<6shwmIkJC#>`IF@^kZxX-v88a?m5;vYfF`G z?vXzwB8T|d{M_(5LVbx$#=^Kyq4ysNC*%kkoHySaqHo}E)`xSlOcR;B|G`k(3R8l@_X@zjX?lNybJlo;R75d#`mi#7$vFHBWx0m<#_f5x-D9A)D z?cX)zuR0wI%D7+g?(Zwi%ZGN&d5Jh2L-C{lxF)m+g_23rIcuE_nVd3Dl~q~>gq`8? zO}Z)B&8}R0jgaX{X3G)GzZSNamlpk`!?pO%@##$0&DgY^zrZ;&-pg)5-rmMIIEi2t zJv|EC0c|auCN9!OsP`rz#!>g~3!9dH#1FYSp!0<{C06yuTcJfm$0l=<>~`m=bq*s* zzG%Tf2=*Td$&5J1z@f!I5>9ASZlttnUi*8AJkJKG?PkEnoa@C6A{QPt^sZ&B3FDaY z&e+5!#C~ve;ZaQhy2FLoLuxP%tcHpQ?}g9;KK%gfXl^z7h)wR@2J0`XuaW zi+|(Cw5GrwsT~mLkFOpFmv$E_&e+V0s5du|^+pNKbAauTnL}i4xg{U8)bq8xIOO}| zOhepy1SDXEhi~AnwsrdEcDKH0tWcdmB`BSCV-Kxz*sid+x3q$V{hqJ~IaQ;4G0)Jh zu|lzsVS)`0z_t2dxk7NK)} za-q)q=EOe19h+sd%d*;Lqn3l+f<}llfAnAIvzNkeg&UchOZfQ>I*+cC$F8Y5ykun~ zyVI2|iO?;nD`hCO>oHL7czjcs(ch(A8ILn=SnBW$>23(6`M z0pzXe&dEvIHNNOtk{d*yl28oyb#Q+4kGJxp0?H6`{{%Dsfr-r4vuL??%R95=nule# z4o^|u1`$vRDG!b3Gc#*aB0vt3YcieJ9j!K5s|=*p<*;9P;QO_$r`FcyUHj~QN^!L7 zn?54yJB6s*CQLK94^T@ zMUtN#fl_9r>N&n&?}GdlPhLjW=cg*A>R-eHst`fPg1}3&&>Wm5^|fZ*)%hC zv{aY}3tAx18+v;taOelsAhPct$>ml}7DBafP+{$Dt1p4ye#`s4;ctHaSGPaA&Mj>l zE96Bs_@yU(-m7L-i&XNPY9G+Syg%e`@4Pt=PlH%E;|5_ofZ9lvuo^kqowYL~e*yqj z3?pLlFMh52bEq%GbpJ>NBjPl~H(A{1z~9FPlpN@DAEb|Z!}Irz5^0t-b~TM(sbvW3 z-1I4&n#?!;s8;$L%N6lMP0RSZyQcOZ!$+mRK(1QU{;9~eeNAk$UZ)YY-C{;78!$HL z-Dum_q6!?%=rj~zHJ07;8)pRaEnq)^J3)fSg{iOnPYaLSBfvPtKbA(9#{&ujLl%DS z9xHjEYaT{!icfyyc4Eh4x8OqKBEx+OzT4`PX^T}(U{kYk7rE;BjRdLud;{(mPP$=B zLd7#Ut?;hp%wGlh0v~D{Li&ccrv{kfPS`@8WyK3*YYL^rho7Siiq4t)RqIOYx*NDAihuXDA z5E7%P6=El5()anje<#Ng$KlDH`#P`l{G5P&nIP)x;&0!=uyCyFlU30mF zQs7}bl)ZO+vouM;dS8eX5XU<;Bg8k7l)u-a8&P2K^+ea;IIXgN4%p}GDYK2z!lJN? zEjKP1yzM+aT`MW;>*{3UBi5B?MJgfJbX%ZS`&@UdZ4#yS z1cl_S?a?h@jwueNbhMKH0wlw&#UwR_VQO;Y zj7dKiz$mi_nC}CXoO%BlG)SN7{!B)uP*yxJ52W(*4xb$ZD8}s*FT)7oEP+>?%P)52 zHfD0CMtbYiWTK}Q-mdeNF1aa7;AAds3qtV1{?gq%)1JiBO zPqw-UXIJd(G`Fio{}WGH)2ccA`Eh*zLC$_5D&NNxj~D3TcpRo540_Mkt?eURVE)pC z{poTiE->_^y9YY9rg%1@>*s^to=2|j-5^Qfs|^I zIF3e~D*%cZQi7EfsHbRt(|Qg7me{P7FWK?zFimvrv`_$u-T`xTzv*R^|9)FleRlp5 z;x^>>`ukYZGvd7?HgEQIZIdZy&TrJF`blrENw4|>NuSnA(TcQ%wfRCQc6I0(jofGJ=%c!cxQ^UX&BVp|Xv(AlXF4wlMEPluhfRA;>e5!D z$hVD`qdO!tYkH|R|JK|3XU`n45t(lHVE_(Ur*pkgcy1fIyaKfr@P<6;ouYMrlvS;8 zgG>OqFSkNAI}r66L$h{bU8~Pc>~d1QEs4l7%09rKp;huWQc8VfQUWMS^_=r4OO1`= zeDyuS_t5~(%$o`|8K>)pU7j!V4`2l^j^A&LzARf9^K#L@EFm(w-XLS$n^X%vhlB{w z4gx(}Kf~3l58haj2k$uqCVZ^5pm(EE)F2Rmb)x@8MCGo0v*Dl*`9-wvTxGV+}@WgD z^TMdAam1!tv4oc$8(Y0(srQ0F7#YU z?v|&3pCbF67q+vf+Qj_fBQE?tq>P zVcW6iY(cu@v7>FgF$NlqPGPgLy?e(wso7& z$w58voAs9fHPa9aG<-~~3qWj~?OT38@`XAgtRKi4K^uCGo;UbDg|)~L@$Uk(3M~3y z^?$NN>gSqb*Y6(Y?u5cVf5FrsShie*H3Bo9T@|!mi&1-#k8rksfVN<%O(3dK&isX% z%20y;dT_=5@;g6lbZs;pF#_lfP$I(B9_^^_+w{Ny zjU~}^_rk+{TRxtttX!XmU9@(~FLhcvSb7Y|0W+M-u{NN9yJBe{~ zeY5=}$ihxu^e+&uG&!cB?xpJV^E0tz5-DhcqKpy`u~k-dc8Xa7Osa2ZxG&x%mA~I0 z-}Q+Wy&~K!McWi~lPzc%&;#a30!Ab1X}!IcO@$rHjNy*o&v&}l6@muo zKK$=BukK#(Np*MOY02$FYbV2Er9i487}6|Jw1(0&C+F=C*K9Al=999Co_1yc3H2+{ zDAhTb6EHO#NL^TjAS9b(HET4CadRCdr>>W*P@cT25|>9nwZ-+b_xkg3docdBnLiGB zm5Qmpf}wMeIw2H0p_-MZxE# zp2Co^_1l@zXg}QcN%5GV*d$^rR1!*z+4IqeN8pMQzG9o zz)kBC&H?5BDDIvd7iXysJ$`ep>4Dm5bZ7B!wMK%M>pLEL--E^~KvcYmOvt zrd{iG;koiH`>(&D;AC<<0&f$++OdjRLXXF983ri12SViAd8eHJ2G0lnM#m|UGGSiL z{CdQcXfJ^*iHn5>uYVMi*5^q=PE2oFPQIO7r@3f)^r6tXn=MJ&-5m&bkUm`x^{Gb? zj#npJnz$z}U9KZ%@|~S_4Gt*0&$#+OY)>}h4ZBUE&xe7-uAJF}l@&Bp54&+v$#*z* zqTyCZvS;8kIdTCBN1f^Ip4UlYW?b&*ocEQI!D&>|1oj2?Ra05XIvo0tQM?`(Jbm}br5AkDbCvo4;;1-Qh$XC2JS8LgEon>pXL}vaM2(8J@>*Lb&#&D5>WLCBt?Y(&~S__lt zp8E&FNS?!kOVBKjK!#!2z>;C~wO~+9Wro_p`~#CndwtUZUuffySV%&Co)Au9d*k4t zNviiy*o^4R($EGjkd{RvjCY~3*>-6NE|npJH#A5z%@(2Y@e-xYp&R0g`o-DIi#nSg zk8sNGW@EjEzb)qJr8W=OQ%5F%Q(h z8G4yyqM2<3(4W#(bT8&zo(6Q{WSfsQzI;_q3K){)^xNZ9!<1+@Z`rp0AQN?m7!@=W zI|dud_o%{!pis}7Y!|vYYgb+fijl6{IgFd0st%221x;6)??=kWD1o8WDe0t1hg_;&K)gPashQz22w4;1Q7UzB0A{DON zIHdCu<>@LRU#X*=Uju?Lutq2gVhIl9VQN#WG?O<)^HH-=?5GnILLCrmagm&n{ucvLLo!Yk==vhMJ<03tV@u3|Fdq=1TJty?)5ke z_=lrZH#7lO76(rQ%&--q`{k}YZERjfAy zg5zxUzPh?#3?_s-e)1kr#*yMatU2&Axn^D&Loc@d?Gnm>Klh8@^N;CLxh1g4K=_`&6Cjzjb2hB&iRQ8>`0RjK@E zULtB&F(e>BSFH=;|GWLM1>(1yQeEbgYXCi6*J!Fk5-u$KDN74AyI=PPUQ@(;Ja+dk zLYx3+t4A5vy?!>K>A-*M)~EG!zHRlJ4`vE2b#?hx8TgMT)34y9*Rv`4a;@(le3zx` zt9X1TPV=pFXCak%o1Fsx*T<`i(H~QgA`kH|Xt4bzuj((|)m8l`uVj(zAIIx-p1)O- zxi5P6?s4(#tsl{@y7AE=dV?JA9l{C}r@ANWY=4R^rn+MaoPTs1*N&q!vY(1)g=)N7 z|H0;f+)^_Urd@?JOdFdJ)a*qE6$|EPEG9STHM{?YhJFP{WNv{%D_$coxQzf00uj_crhR{*@Y zP9N}(0NJmoZWAs{;c-f%H_i3X=A+8W**}SBiu37`0Gi^F@?c>wRPmD?^mS%s=K$CF z#XIIvB-H2o7USJ|6QO2Iq%MI zx#g~10Tw6_k>E^PdB9z9JRmDAZ+lJKSP=|jM z@+lYiEpw_cg#D$rgXEW$`~DGc8MO#+<@3>rm{2`2u%C+OuA9!iLaC?Qyr`NrBC~m4 zt2>|7T=auvuZvQ-1V`!&07$aHv9_Q76)DSquCcuDkI&Y7ABT9AEe7L!o>OQuab+;Co2l3 zXQ){{s&={E?wenkk37oK=_1n@15`aPqRWo&!^=a|~GP2Q{==brr1K-7tf?T)W>-yYZrFs_I#9L4xXXYa-a$R_h_a`xnb0)X}XF z*`^CC91>5V0ZCXvL+T%GI^fqk_lYOv{sPLMh7Vb*d4)Gl)f$nE!Y5Np`(!xlNWKFG z=Zi79UEXs^+-T@tQd%et;zI2TwuRg^_JFf*qIyprdXMG%g&3z8wtu+uqV+MTsRIzz z=l974=iC{JdCc<;-c&SpkH^w*L!?Oshx={At@fNKeDW?{Is6}m!A)HMMgKV6OsP6* zany1>W%&JbD*Y|f#m5tD9^|>nbqHQqKju%sKKrgLv48Dd>;gUgJxW6V*w~#AENO2o z-M*;Yyk;8PwK;*sH#s!5gpOKcNyB~@L_DZvpU_ZobJ~My!I{=s=+U|sKcnJKsNwmB7;P&B0@Sec}b zwm{@{X6I}k8gDW2Pkl==tVRS(_Qr8#tb%ZV?>K^(bEj_V?LUeF6Oo&u+o^HdOFL+( z;)Nl1e$z8mI<6n;El|SPg+)rq&j|vEOmeiSu#pIsehuX zk8XClkNeTV5A)=%sbpDT$Uc`6%{KQ@H>odch_s{CM%$T*^O+ zD6*wR1~|RT7-Q-{=5P_6c87j?;4#C=&NqOQ8AYMCBfF=Mgrius33%l--SEMZG2I_+g%@6-oPzy0X1ORm&i$RATC4_rRQ zy7tz)r|Tr3cMgYm6QK{p-l*LINay|4d!T6vl~z%V>Fb&O7wW;ROBtf%gI!oM-#L=F zW-jx@-v~PUZPodvGcfJo*)10M#F2bF8qNqNW)^kzBVJY`q6+H(Nfvb^5nYKy6->J1 z&eX^8_C{QMd#FZPFt1sW)8yZ>T9sSw!f52^Wf*_9%`)4vGO4a;@8vZM8z7Y9O%43F zz;~HdTpkSlr!^@-E+FtW1DPLMNBu36*ZumhTgD5Z9UoKw{w59k{=GYu9Sk1^HZwz~ zY0S2nj9%fo9p=P-4gvQWFZKa7_tRT>Fn`5DFVq7;b&BZsk6G{2{_^VeXn#DO4IGiO ztiVNx;udgG5E4?evE5NP2U}#>UJotBc^7MDlCozrnNk5`I4v2;pZMyBpU1A;8`bd{ zArm#9UC;>RhO`hF3#-E~ZJDfWzd87IccIPCtNAlyP`p`=DtX)gxDtGL2-Q0%OzH?& z@;PSvMZ2Vlsz-!tAha$vyWjlDF!d&j5NBZ?UIfKy2eUq{Ql3O@7wS66{2Q{t36DQV zm;R$bsS?;iI!wCKeVq&Jk}$6}+(qRR%_YJkA5T4Dwu|6^lv%_th@4~H0_j}wdW)xl z0pj6$|5?832~1G2Ohmie<*|itgX6;`8pW2sh_l!>Cn*DCvgoswD|53j|8XyzV`yv@ z5M7SKYAk%DN$A;t6)XNM)s{!=w$0waaiQ=X%wV&WwM_5EH7|EagM%TY`kb{F>rmyjcYmf&uW4r zRkBtZM%B$MYk2N&`d+z2*#~PVbmcqukGNys!H-4{YFu>jTJKbvjYqI8o4Qr@jtLL>QjD-T111sUx>hU0F+=^`Nif zQ6ryAkAitfzbQA z!JD{70gQB+m=54S>D)bO>9I1`oh4;+4px!G72CUNqS~0IBsgqaek8|AZ4)GBwiKme ziM>vjmh|SBx}bGv4l;ac{#|-b$jSu9@~#Hobos6DhIeLvvolDc!ruGHn?DSDVx1Qk z6(sMLJ8}*;eRpnXbl$iaR#3{)H@!%$)Twfn4_=G4x-}~U=&r^#$!kjnQ%iEOR^R?n zJb?eCege<4bkjCOk3*CsAXI|x(c8FFiQvM#?2}EX;ph<^?TpGe-fR&g$v-T znlCKxHoD0ZoX=uw7Ej(d)pI=26-|<2waD4`<75FfkSF5_F;>cKIS;7WwaML>!ffj* zZBkxfs(4ut$LCKhtP~7WknDTmJ&fNSO&G>=v|c7c*$oI3eiw$^G+Cpq#lC_S4serC zXDnV{u-m%qfo{J>E2OlWL0Uo7L#elC@x^TQpFJy0udBQ}D+ z982rlW$`YqZqGDtm(|YzNT1amjCSMFch^=1)89E3Em7)|i9^*98sGaVYYC23j8yQ!1(q@6Phh=rc_B3AmInm)Xhy>M z^iZJP1tX4?rMtM2Q?9Zw$N=eXlF*Q`V&n|nsbs_)kxl-?fr<$X{jy?(?AF>v7nw+DC3+ANw^p?ZIU&Vkr?sImXY z{g_~kk!d|9X`-_U3)m_Zlj7aoTJH1Qz~Y$^W5OgIZMW&%VieJ18W4smw4#v+VEux+ z7;8OL`HevDf7a5Xk}{mEb}*<_zK0Wl`wH%=x7l-q7{@D3)8AyvWnSN@$HHzKWf>Zf zi125m`PYAgVO1sgcf&DOEZ}K&&z+DHSG!%hmu3>5zUQ+V{3R3ToXg>CCT=dkM;xkt zE%)G5EGfC?S}fh=+C@9qo+K51ydo(Ng>2NS-Sl3rGt|;-g_ms|K5m~< zY7|zguj-7_3Flc{=!hc~6^oAu@AabpQ8=aq$ePhieX>yuxjZnaCn+BF`lTe3ZJ2Fa ztyH&KS;LVo5R(;g>sJ#%bq%M@5w8mSg8nuR5~iEPQ!;s4aTrN5Rk)NCmB0%q#XD~F z-eCzT>SVl-4jcGL^NVO{GSxf zqAIw)z8=wouFbq?h54KddG*t8sftgwWiOD7w@_vHoO%c7ecxHE1u`$w4KhCz-vZBl#rYpE|mXx?kt8pH|(x$c@%ubUEDXnO6@NwxDr zz>X$x1V{-k0HSJbXKlB@&H!fyIhMH(HYw(dF~4UnZWg2Tm6_BzwC_k zbmF_2oU~1VezaHVNu%Ri`JOuJ<4qCmRX?wv@KYBU_30$8Z=UA+tc;x6HM)oW+bZ6w z1*O6cY&?>ovJ#nmI{`bh#h3yGzVJeSs<3Y(;;8gAOkVmEIX%6}ceTOU^_?RMDv!As z_~S&dBM@j9I1v@h|R zvL8R+yPqo81oymetsv)>OKJU`M!agWKuPnJ!sgX`gd*z-O;tM3XudO2y2Rf8TVzbB zp4c1bOR!V@D~GS|Q5Ps0n?Yqh8WBy+WFzlFR)?n+%-@+A^|(up7L9nB<9FhMh#BA5Il@I%nb3m7?BDG{I<~_wx7}G^NkpGx#np&wx^(DbGUuG_V7QK#v zB!|buZniR2^SMASzVH z)T>?SHr=d7sLr9gzFDoG0|nfBOw%knQC$fzN}P^o%^FH_A`de z?M)GkAufk~RVVOIKGZnaIU#zhtVG+sK{@&?@vR>rFt)$TB+}wkT_pOe(i{y3chtO{ zugvQHNu}}%Z@5J7!F?!yU@3K`n$@_-?+3MCX8-2|i0vr1^aiC}jhI7oaFH^Ppof|4y7a`XMBmN#j$w$Kr41thVbkHM<{xSf%?BJR|zwUy6 z;8e=1XLJs1@Yf_NRI%?-9@CuKUzg3>k1H0&P7U8NCfkWuwSZ-B1nB4EN=RQy`?d@D zzt*>Ga|y~d7At0z(Y@P0&VJZBvT3~hp&j#jjfXWhkU!JUs1EUGO_SmE*0Wr^Y5K|4|<9T)V3zX!GZ{nhU(7^u&EE zP)^fr=-01Dk9R04qs+%Aal;T3t}4%>{o2>Q{>s2EFPoss&3v1~ivmY4Y#k96kfgvA zUX_cViPsu#EOIntqZ|HS4JlFOX$-P{w~grcQi9JEwT)Kf{?4}3 zu?Si-`zwiN&O)EdK8VMXEZjgHlL*QJaVBf>K+Xf*Y6oHc)3_5uQqG3vg6_{iCX3y5 z0ei)v_wxrEno4VBesn}NxL?#8RL~OVyxf^+@6nsnWwrP103DEkdRb)bvtY72%pF^q z3dwj6A4=ifQ`_v%Non3PBUc32y288fVSBILDZUhwdNl^nQ@2$1awVYZRT2-s$LQQpTI9U|jHO8)Gb45@ruH^76GDjR*e5O4H7*Zfx3j~gSx&dj2%*u6TINe( z>J$4iomm}CN^i2~N*mFWj#APoj?c~vAMuOW_P$N$nE{_k@CkW7jTYLR`=ZMxrI5^6aLUM$J}Oj2X` zQn4i)8gr`E<@~-0dNBXmn+x#7(dTTvh#nsC{#1OicElb`X*_%GcVw;*20#{W^zee3 zEE}+cAj}7muga(!{?e?V<-K=Cadg-ppv47^?grTEydRy0%wOWx@$Y&>`r*3t^0#5u zYQPo44yw|;k)SuR1GVshjJ%twm!Vlt|9h2%<&VZEAH{y`9q|&*v@gBGBS(TA$HO8g3K!Xd`~wqCmzo_IW;UT+^hQ%PucD=V zec)VROPrse-SBET@tHVMR{NFP!SCvCeP=eXR_60WlH7C@UPt7)w8zN48uf0pH4X5W z1+3-~@meKIs#c7WUsfa76`r9wv`4mffZ4e{tB7SL!8GhyFYRqqqY7I#kdaFKDRqRjCADD?ib73PjxX~!bfX#`ID z?GyndUs$d_-ub_(bxW}Ie2!(2bY!?oQklB`*Jh!MS+muvV)L#ie(w-a zHEoIAiX6hp!*t0+reJerU1`*cwDj^iY?l|%-1VmkGJCsqhc^toROoMWY0SN-=wPr% zH)y$dLu422*Ss+EFBDbSqLo4cwWQ9Y^+w)jocCJ$Z}?Tkm`|&Ks8xc0h1M@DLiX7E zSgfpeH_wpA50rx=Zfv^+-cSP2xWB-m(^)yOZ|Ape3n9LtPtyIRh8|xd8*Tifcv)$v zJ7qVud$}C?f-&|Rdk7K7HRP- zI9b#|(;1;2o*4~hQTx*xGEKVAK6X@IHZ(^NfG4b-$g!s#MEeH{)RSi?FI}Rvq50L* zpRu&LWnU8~Y}jY&xcC22TyoOOw@ke(h$gQr$=m}97f53phU{W|*qt4{##(KT8G(zx zfP8@X42(J#0AVI~PX{Hh53}|a8wBrSfTz5m^Vy)fQr-!x06+2E|00b${ngs2#xapC zslvgUGUH=Q4CXAf4_c9NI(Cu@^FEs?d330eF_t$gt61|BtM%mF5m3V7M-TfJw_>Te zvTK`;TV4cPbBYR&Or_VbRRyHOvAIp2%~Xo-mEe=D>OX;0=Y1*fJJ>mh(6GJE>@KLp zq9^Cq^sZaGqt4-UkeURr|o08E+Fs19-vAxt?6JFSsC_?6FuQ1goO+w&bW1upyDJp- zNzN%f2|sDx0wZ8FohuKyX4<4ap{h)y)Rd4>Xq-SyOLOw>a)F>(%~=HAq(vUh{LoyN z*N2N8m!xw$$C3y%Nf9+S)(o>(g@o?u#NoXojzTVqS<2Y;6t3$vw(3JwzEx-m_I(#O{jZr+&aloTeH)(OMfNF>}? zP3hbK`=cf*xgV?;TvGVhFtRO98AfYQ77LZA5E+U2!I;oFDs^`b1aGR8Jd}$Duua+c+QJGEL%~-*Tgu%+c=dz9lfnnks}FtkrLb| z$8F(6skw^0hd-*%j011%nfIjlvBy;u)ieOzUB3>g2DyRl@;N$tVS%eyyn=VF(UMS0 zw?Zhb7DUbWe<-q%mT>@j5-R=mv&4}Y{}a;fAI*s52&Wv7)H)QGPFrT{$ri!Tn0Ql3 z9GU93UT(J7Ux=;zlIJc5>+d0FxBa8Y1d}?VpwJ#awleXQSZmCQf+oA)&{TkgP*!H$ z<_n#=%;Zs{Ynu+L&m$c%*-(@54c(HT{+1svI>vjGV%oIww{F?Aq^=lK=|SG~(uI^K z-=JukaaMgUGe?G*cxbu&!7KS+aVRa)@J}uAOhSJK2wWWTv)e$Xk+HH9P^1sqsKc zvNA3p@UC^Ad?R^qH@Et$4oh|}bzzWo9COr_|KOrnKe~gGzq%SzCYks0xc=@}X{T@T zVtAZT1NXkK0Twl`Ua-CoXj%UsAp2==aZ!CFMUxvcP2dG&C$_D z#%7G904Dj&zVJ9H)-eQ6Qea?-ql(b8^Z?lQO$)`4#ZuFkt}BoO2gloBUF+fRv=l%j z;JTBe_b205Ej*D7UbnTLk*_ReuQxNY7U8DOIE{;Y`9)2LTMP?l>E5zmAI!KEZ6$v5 zim83>N!cvdDmS|xrHxv4n|$Pq-_-^BHoBSnb;a>L@97M_-U3750 zli~+zRf*$X+%rITxiARi)y%ML#ysr(aJMJn%Si{2dp0&G^dl@}TqNxHFJc%tN_y=* zMNh%qJ+vualTP)#?qB6K@c4IaYX020=Fj^?M5u#1KXF=Q`cI{Yf1^#~GI5SAE~KFO zk`xn2hqS1l7ROsCRSy?8I!RM?OtyYp1s9mdfo_`4b)lzVo#XMRqq}$9xk7-g+vLlmzp@Qt+F<1?i=Ubb;R<4~zC)WyEM(<- zF{ZB-68-i{q~yNCuaDnpk8iD5R~KAM_o9;C)}%6vy!zL!%JG#nJn6DTG@bJG;`Ch# zyU46?hI_PpqTd@CSm^wcYILP|BL{wbi&0(RSMF7v&Jb7D7N1OaD7K~Y)6E>WzyG<_ z+J|UJ&GnvbxLo?fJi8-qc~X4)8)fr@RN@V4ZTq&CjuY4dZ%3hP#2yB540Sxx?9jmL zfEMgiZOuUjoK50c>2>@<6}n}C#b&4B7U19q)|@vgD*6>oh;Ie*nzNU>l3fZNYAL7I zA9l3q=RAdtNhl3Z)pkQ2@)x!hAAY5Vvj~50->A_)X^2zKL1L=qWr|=9V}fWg>OlT^ zm$oL?%#cgY>9`MOZei^wT`@l@>ld=%pCvNBd0^4g9N&Bl?ic(Fs0Ms3)>G*j{TTT@ zhU}8U>b7wM2B9aP1F94IhvhO)E(H?Nh~=^~4d|69b;KD|=)M~W=$203xfClK0&WvJ zF$iDoeg?~*o&9u`h_tu>6dvg(iFK}5g`jSDh%BQ6E+6gp1jxaZGCO~95K29utddGu z-nM0X!w+0zPbhZXr>0wB?2eKxUlMlz3zXcsfzmWM0}F?+g`dro1*zkw(=@mI)ehCD zOC6<5Dsyeqg1}GoZuDIhvDL*?in>r!*#}GPl-0{HLEih3*x8f3{WijcE~Q*tEA1O5 zL4u$E(@^K>=j0VYD`qFH&qzwi#unFf<=Ds1)@O_j#TKTXd#*RsB4YhK74lo=k;@~F zyEp$Ll0_P}4uerx%_U*kAN{#4jSN#5B@-2?h%Gy$V={{NxwhH(#Gc)LB=zxkQiE+5Ca1|-;Y~=pKVAQ1 z#gcu)qCl{@UH6}g-)y#vnIkFs3;E>33$IFZ@oLDunE-0`}1sROtQc3}|#>dWk{f|ILDQ5Cj z^NsLo{F|b3HSf|7)(xnk?%P7m>dOVLmv*cq1pzo}64eLZ#EfFiHH_3)58g-ls9n~< zh&7j%IgV?`NKt34`Gw}l8J?>eV!{2jA{T1MIQyw84B6w`>&HgzM)_20F@YqIn}0f6 zz795FI|s^~IE!Nh(lZHH?^-?na*Ige&3d#{p#_C7WlF@ij6Mft# z+7G{9HC|LvlH}?tG2P9QM!jqG=Ce%>yby!=YTxp)h$}%oEtnAZ{7}d0mD1|G=R6GC{}L-xZZzp%l4QW7Po12DLo3U8 zzcwmE*XHY&gmBFZcdwQEJi0e9nq-F{jR@a)WC|ehHDI1F2BuUIrz3;AOQnmowaYt6k!;p)vjmFu40%YK!5S{7NU`+aOmJQ;Tq!f~{A7G+YDLS)RJCk4+`c z-J6cvnws)!fD;G65A%&GI9T3)p?yVX_P( zsT}{j>v@&8SFcn4#a^s=E62q@3gLRA$#wKsgaGo^qJP93_e8;&UBeG&)3E*pies0c zZbECtX8bnM&S1Z1Orie9f9XKJz}Q|PrDa1NEl`5fvlwkRDSZ`H`_!-ivIY3ob|G5b zG2Z#oXuTCJ`=4^kBE}gPpS{`IgX=z_)8#T=Rc6-ChHIdp6?hVmTkHJ z!zPvcsb^_=QK4j&2Up^)Q%s(Z(Ub>Jf`oVeQ*#>Pvaxj`NSH;RBpY{WDp-#lRyw0z zn@bP5m-kGlTuZvsNFx`weD0UiU10+9ZkFG=s9`DEN+wE%yqh*W0d{k0pED)8Y5!5| z#%}C06xyvaAi=0MH5VdMAp-N8A_1#3uNd+_#S(@rEzK;= z+a{ZZ`*U6>P~c677D7{kduyLK6BeC$5Fajnwx{Cz{Wjymj*+T!o+JEx7J^LdAon91 zKQ^GHh@~DszFsaUgcn5XBF(BUYZ~R5KtNc|?rg?(`tM1jwm{91#;Bk5te-wmupPDC zpWu!-ifQ`wa`pwFT^#*FBiM;0)`Dw=K8~e8d_sYvFs8 zO2^#gRrl;Mp#@8hA1_qO+xlG}4l?X0v?s+CTaE>K*u*#gGPw5nvj>Ht%5tE|sv07x za$ciQQQ7qOb1oO6no9*utNjON{^#p)^H#H1wf|s{X7*-wFQv0hSXY^)g@*qpGt%xC zH5IntmIXWIaI;2PczBUmSN=&dJUkrBmNYW!kSj3e_2!f6r=ERa@#3Ku)l$p4l9A|_ zi`q%DzOe0n{yIDwfFlIA_=A{@aR)RaJziZ})`z`r9YMm~iiqDh*+g^c7hf1ur(TZV zGM57#%?^q3U3J?dQ6cGrlfN;an3b}e!$VBpd^cB-Zl0WXFVRS?K&pJMhhh;|B< zHtGkIy8s8pI5(@{|1z{WAz10<@JfGXOxu2J(h9`7?DV6R=rA$nPe8t%8lbAg9(T8_~N0uZs z2X9t=yp#Kv;5?|H?NIcaqkwc~-%&B9v0oKq^CRY-UZZSSN+U#rSLosS34=>dHvLk5 zL>nKMo>4Q8F4GWGYCsV$4XknF5wA$cHYr>kgBga1*yXpdD8jgH-s|DBQKx~o)S z_xsNtRdFG&^jw83EDbd1=~^`VMCx{smlApsQP33ivf<)-GW**f0f%2Ekyq6{a_E(1 zCbOx{p{s0G);VOPZ&4n6!6)crB<=xZj*bScq`cUWZ){vC7vJ*(>_-H{^w^Fe+@JP~!I9qJ_pF+#F$HzLuk#7M&pV z`nWQQ;d)TjK0#(%Rf04Spm5P4Jit1f?B^1pkDvKq-C@vaC?8i$Mua#NRK~ap!z+`X ztHM0lb$0@B?GY|SNir6hw0!G%M5f*4Ihd5)0r!c#=TiKOW8Rg?IUJSUSD`_$hzJBF3pPtvW4#?w6zhB9A>oFj8qPP<3w9${lWD4kWstWjwQWqT1v;iJ|D$e-z%+amWu3J7P_Pg2Id*qSIf1nr#m(gg1fs&=4;y zx~oto`-8z$0+n~SL{K=%V6QZADJ=&Aaad**9CM98FsVx80OBQn&un1cq$iUx1RS0*J$`YTgJ{phvNsdmhjUN1 z7nR8jAkF%gX%Vm-*gh|gWRstJ|L0G#golKtw(LJInOv(NdvD96$DVG&!>Znnw8sqSATH;V4Q9a4T;Nc@?bd!Mj9lfMU~W3MDs zNj(4fQ#Wx>|M4G#C-<)OD!lylXC*z?b4zbEGVf-{)`$oYYm8Oc`lQ46`D@_Q5!q<3 zV?r+F(Vi+@gTj_ywC=by;u=dVl23G04iz^(@dckw@s(Jdb5Jtp$ap{Xw~^PVBQJy4 zwV9tEwvC@pPao|o+$R)Hbp2EQ-o;Ean`Lxif@rg;{ChYtk#mM9k8Aa@ z3UZObPZ8|&5H1rrbBjf#y*Vt&5-4f}jC;sn5vE=qM81`R!HWxMgf$AD#(trgk5J>O zf#%I>kb9Uv3Dt?UR#6nfV&VyN6k?`iRt_LyHYmy+7#JfVAF)Py}}t2y3kp-PfFJA z?7JP4)6%hMeGdM18scwP%#F{4Kni}IF=F0dRI&LADu61S5EG9QwS{d@?n^AuYn6;K zO!sgINT&>4XM{y9=Aq>8K9d! zEM!y{LXch@2sXnJ(2C56z?%dOZe%=p%-W@AvgNP)B?Vh~Ab$hgn|OBL`1ORTYC%O4 z(_z0`U=U*Sd^ZcS3T3$fb=yurqckMbKHnHvo1$&rJ2PamE-Fg;n7#<~kj6XM2Z6J~ z^#bpB6Shnw?_(2Y!_`|__7_KVx$M{CVV+n1Y9fc?c7c8J%#?VP5~Fc@%`)MpS5(W9 z_M;yJACe&!;VuPItnudl#gxNjLsfN=kDB$sn=&@C??=5PW*A;G-t{-Bf?w$P$A0!d zwqBdr%;nZ*yITE>IV3JyMG2>NqIO{cZ`3Swa1bu>cx?-IZj!(4ErXB5VP;VMYw+4w z7NG!|F~<%;?-#d|L8`<`zjNJ>2>`K~>)UL$>O+TzUj6Us9Kmdz?m@}Fg*?z6vz_U- zXC@z6fqkz)(11{zed<)26TyBk7b*R&X-oyCwjBc4aa)42J(;*65x3baZYMS_Tq{yq zROlEeoT};_Dy-KPzuV_}-v*%gjo|uc5>?>P0nf&k$epbeaEt>H$TNa2>sSW}V`eAt z*;yCqbd-vAPIczJ^5fHIkGE70*|N&Y zCnaq6$yt;+ReY`WX(7E-*YQ-k{E6cy)JuQeD6v0Xm!0vf6a)e2kcj9Ym|rR79VLb_o@z>WzG1b5m<3p5RSJ zN@A|QV;1km-&NeI{ITImUO9(LpKNbwx3-38Ap8|%riL%gN8!N4@j+>hpg>Mav;H6n zcVnEVliG9Qy%i~2qnM2_#Muo?mByegS- zRfGOD;4NFj97(Ub$1d#41$k^ERpq2ric$nP%EA(WXT^^7#_?M}y8bPs4KZ6P{P$EO zW?S^>94s#3!}!fWx)&_SwC4SV=<~!`Z#XVJSd(#dk&~fZ z?L=Kii*2{Rl^vB^X2JP@*{NSO0i$JKG5Xr&q`)cg(0`-xJwHG{h$>;-<+_XMKhjn@ zQOhOpR>MC?Si`zyGvl|=jD9^@p1D;b>)BS;s6nMg;#R--*nJ0jWhf0fg@0kwZY^}> z(QE&3){Lp_jwCBwdJ5*2>91h9r%l0+-Q&Q#t!(QEdG2D1%|np)dKJ-u#W_F8n`<$m zu@%NINa+<{q^e7x?zi9h;F1}K6o9!ZXb^T}iV5a=+Iq|K^>=%jcfx?eSFKG&qy|pg z(_}OW_{Hfg@o`AYZ2&Dg6s+Ov2KbdC0=F|JkAKgZ5DDo5>ZTioE73~>1 zUCD9Bl$6H#BH~V9XW&`>o5`tPCyl)clVZV3y#fHmwFIy2n<(Ds)FBei(tWZ=YRK6< zSpNP50GQ`!!%z($RdX28g?!N&MK_y2jn7dwva-mdHWJtbiah8X{PpJg;KO)+7a0IK zMaqoh`Vrdbo9*RQlE`BZdA854x+JG z@u7e)p?^p)itsG5=)yqG+9;$#0A!kRh1Q~dCI#@HrEc>zPQWzel#9h%VkX4ytv*4r zepTMlKQ?gxyB;sY^=+zY2PM(bCCfi?vb5o$WOT-dYyTejj^Fi-*$gHggYZ>M;g`!3KM$PO zia1gOWs+obQjq>$+?R!}<&H4mz>qbarr&!nQMn0G|Dz^L%!<&i&)ePp=x%#*<(R4H z51e3>PegLLmk5QUdF-8a$h=vDnCiZd9d+v*%P@|T3jSz;YJ`2&$0v9XZ;SqxII)nB zTXAb-ej!61Ny`+Z+T$wr;9pDxagRA+-OfH z4`A9xRD#l&WGMQwmDoqU(1cwfDk(Gl`m6Nez4nmVulMG~1grYC-iI8PIGPVLrqd>0 zyR}_8BondkT79&gC;-zonv?@kpsYPIWK%>&&>gdTrZ8TaK|}LE*UAyqiPJgdvFAnx zrg=AgfE%H<)zX^wJjG~e8^(gB>)-6rhJlrHC9cws#n@7d_H?=PY%|47TlnelM%Hq! zM(Yyucy_slwqu9#CdGwn*81$nXn* zn!WS9L+}%yreJvbEb;m4CJ^*(&8$Vw=<1v=?YX=x-TW ztA_$WdH$@AO@GSe!Aj|pI93ql+R;2Q!A5K0BDs~zcUq& z%R$zG!)AMmYgHmq?^xu$uc?Pu?vHK7{XF0W)gDEIt1A*$*dNb0ebxaTT2+O?HmvzF ze$`kEu^k7z<*9uAZ>G)}%dB-F>ThHZ**s-yiCino^?J2_E$VMlj1%BXNp&3G%MN{6 zYPX3-zaN{B%pD7pjw^!%V|~pcG5?-S(iq(#o?oTIpIe;tO)7!nDx>{9LQm9{pxofi zEDI~}7xN?5BHLD61;_K0lT{UA>I_YM^AThLZ6rSCT(dYoFEL`c`XgW|u@{x8GF{S< zy-)u*>ydOSfFdb<8{dVuPdq;`dAb`tcr#+ft8_6*30*=g&!eSBF zGmtR5Y!WW3+I*opvWQy#o~+%GEOD07^I76*=Mmglv}nT@^8%R1J<6+@4ziHsoX`s zaw%}|g-aCPIUh-!l}&n9(mvjF@dC}zak!BgEDU7G}*O7!qob$H1g%Na`G^p8^5-nq%t4sqi52#XODm0 zK2FjS)*qBQ)QnHPdO+nMteHry@c>^+mxsM}pJ*t6t)!xU-CM2w@R9o&-$d4Z+=!27z^FYv&C*}r6rN6Y-3VSNghO&<4fnfjxt z?dy#!wNRIqheo6RjG|&DxBc%>n9o|VXJoJKyZBu+<%xD*O+AHwr+B)?0d2?C%9w2D zrTeWIv^A$l3v31CNi0@HF)P5dRQ?L=`?L?nSiODuo6X7*!T%cqc`8S1>|qd`g=&aJ zsJ6seA%BYQiae4C^3I9J(!abqES!-q=`XG2-M!;RH*Ld;ny=$(%E|I%>OgqW%(A{xwPJ`cPp$rB|RB8 zowX6zo>G(#x=)Wf8$bUU5MjRJTsmK`?W!?8+v)|JoM#t^194~K5g&|q1tYi#5^%=l zzc|ll1_LOrudQ5loQuG*76z*{UAR<+xk2>eA?IJ8tLH;I*klqDIH8Ek9CkgzI;8z! z_2>&6>Bp{$Af>Z+HzcRH31 zT=x;iB3TmwHe@GcsQt=zsnh%hA$$@^OdjAj+CWhH)t1bwzdJr!lMQAC=U^$?;9WuT z)3Btz=Sy!E=Fc0r`=X2AcRuotm_M&EF;ba$l&KEm_oZ&_ZbzvpqOS1@fH|^}9L>9OZ{e6k58X`Vm9H)o;?86Pmw*PIcypaTo+o-DPDV;sEpU ztxZPv(}0FS%tt)nA}SPI6_T$`OfBCb;B(F!Q!FEFw`meepnAb`M)(xB&r;a&N zL!e7BrP-Kf6Dv8BLRrhci z6jfbgD&63oOfz-;(;oI0;9iFuT^2pe%a&b1o9LhDPxs!(bF*h(Omy~+F~97W-~3yw zanclj{pV(G(pgR3f??>bzdiGCw%MO+nBgU6%j0X4QRSBudM?BKpE#NWs((P06i-o8GOEoFF#2zc?$@RQ{O>?_7a*B>?-&u{0{$7Bfl3>>s z;h+W%G$D2X=^=m-u+m4Yv=UYRy&LbX18!C{8;(;|%@uMoDq!(nkg`QFyi=e9#Q_hI z{68`df(?wTvHlo;D3HNQdoL{W3w6DD*mUM(R|j&xcA|r)aB%~Rk+Cq|saR#E9>@H7 zLQ|YK?Fj0Ld3Ob`r^*Y=yd=t4^uPXO$Th{j?!_$l zv0iRqQ6QrwkSTX$SaRc5|Db=h9Gt&tMku+2Fb+(zyk8@>DT-B-aEge!>e{<52z6J9 z{PI67SuCvEW_oh@YRdCn9n+?)-SK~>u*aU3YyLp2jwR1PCbU7`6!*X-D7x)rty!OU zO1PO>v0XL;sE5`>^~S>q#q?* zQXCP;ygymUu^o$}xz74Fnf@*WSlwd;KI^sr*`Ei2NVQ-XZam&FYkJ{Wtwu>{;Prqs ztgoh!yO)YQ6Aj$#chA%9G*@<~y$gV+H4mH+{JA3*oe5NLR`aZXQ`A`n`MoS>ByH^z z?wjqn-DuRT;9F2~_jI3EXZ)&fppA)CQ9#i+HUwS)#St#bAQv(h}UEa&>J7Qy8gSf4&8xKF#eLpRHOKj^| z!sE(Y1lLQUjBU+VEmGC2nvj`5_l=jR!O_N0r9=H0nDqy=n|q#7G1q(HBz@2!C?L@c z!z1H3zdXY%Wtwy|37=FAnrtc;=SyMU4^y*GZ+|jby1#1@#7+_6JF=kR(9|Ap;dH(v z_5rJp*SE!hAO1%cbrY9NDp(Cv9nYZw-yjYJZ8<-tOgI8`uBI9#-D^wrQlSl7a`&&^k}jV$ zo4`oLfnj}J-qof}jNbYpmK-@7SR62;NK@#{bk_qrd7{O(kz~=cYAQKf=QQK+XRP;c zFTODmI%gtFb>BimB77l{>`Yv5N{xoCX0Tvy`DWd8^)1*cB(k;T6PW5G3bC-`nwE@E zYjLvSMC;-gaX5xh{ppgU=q0yjdtV2?f4M9UqO&kXG||Q|(#mI7XEZWj6e#u|9=v$* zf%z|eTGecOpcUL-f=2uPU|V$wU^F5-wXT8uyuyIbn!M9KgOaX!8J(soFFWg z&!yHMHcq3m95}qYq}|6D2uqNqnq*#*!Jq zZYcVU^{}vhx8?%1jw(|8EI-if`Q@NiQZBw2{(NA;ZeME5^5_6shH|+hkt>t}q-m;b z0jmI4m)!yyp)Hl5gbK=6f-s3b%XboRlTz#pp*^3js0?HmqCpzIOVYJaRzB`;G}Bv~ zHB|kD^7P&hCh>*ObPAO3cndWAvS6=E6{ z2e=~xyz(gOD(^3#0=~`-+j&V5fB9O~!gSKyrxw?~@cte!O7+?y2IY{N=v|m79bIJ3 zq*v9;YrgI*1i0$ogIFKfu@mZn4m{Nz&L|AC+mfvR;IiVX84+v}T$ypo?S;hQs-8Lt zN%*PL${E(;nAZLy8>+lj9q26meQ5lI^N0Q;5?@n8L+l#!q5I=IFImG|j%EatiwCzt zYPOjz4PO$w<-?y+b^2P)$PA@vjsE2w9m2Y9}u%qcV7F8w;!5R(53juAl503Epl> zIw(5`(FINU3{FFP?FH^1CJ!81g9*k(IJ`v5oi_Db`rs*~O)%w?tRS+%uqGdTjiZ)G zrZC)~T1Sy6Y77MPb{$>8#m01k`8H&pFa5Ae=rgGP`u8Rb8yMC1@hA+$RWJxL7Q@^^ z$?A(7CHw>OJb<|028~YvnKa7hB))}?qFYK2+jgvOygz{S znY5RT&`+UcHqN+*z5ttGKIpkm-NK%}D~to}+p?hVtGJpoMn{}#@{VRKSSuy!ZV-a87;M+zgR8tPI#W4_6?eWRjWPV3PI0}2 zJTI-U+-vO3oeNNVBJvsn(e<1Kjrx8k5o=4%yp?;2rU}CTyg{DVL^u9w?D?Mh$%rDtspjsw>M+?bm)@)yswyn1XtGzv*%qh-PgogRqq681hS3 z6@!_bTT8^!v&CNfh&81Z7E@ds6l)dd*5ban)9c4mA~#?o<@FU?JaO>)mb@tFHr0@& zg4w4NV$c--`)$hMa%E;VF?A}abk)9ry~y(osE_Ihd z>$aW}#=0VIEWPwbtZFiIW%&@-|H5R&9vPYf7LR6r+`VWW{4V2UTOp&vVhR;R_;$CYC z@a%Sm7Ghay+-euzw7lR-_8Kq^9nlFQA_*!lJFonT_+HkP@PuvKQg%38Qj}Kb)&h54rAR7i*Y+!aGKBfqfT1-vn{Q( zmsivkcu-V3QnU2>kbcm^y~K?lxk{&gssQbBb>3}%Ko<~U$WO@Wk;rw-@kW`D7bf=h zOy$@kKaFoL1W-V$<$Vi_|FucORf#{LzQyDjz|BI>2#{CEz*<$Zg5>p82AVV?F-mI*nd;4 zT`0YYo2s`cJSL9%NOMu3u~n=Zq0J?kYKX1tdoE3Q$jb)IURzJ(tDWFvxJ$@J-G#`` zL%9To+&0>fk9$pvQU@UqLhs^tU5cRP^xzNGS{z(ldLldbdIdTq_T^^Nthc{;wRJHO zi+-|AQ%mYVS`VHgwR>;eR5nk&cmy{H4)T>V@h%;6Tpzi^@7wsaX-3t68mvPV?#Y!Z zuQ@#UA6ZRXC_~9~`o)xc)mS>LVBUk#JB)M^8r>|SkF9=kSSVFNtQws5;Mm&UA9xxr z^-05)8VH}mKLl+POe_kQAuCSb%mTVn_qDDpA`z$jb{s(;CTXG+I)lBl+`iDwXIC4& zYFvqq&CfFd%;+ZU`tFOzhA=+#VNGDKjUW{@rdIJi-uFd%rx6ti)@*^TMG+!RQ?j@OGxX4!~{0|5CJd}}ZlRk6%ov(YG z5$ zY^L{wwlntqFpH!v1<1w}gB3G>u+dpWLNdWz+ck%0m9CZ_O71XX_^hFk=_j_C6 zLF1imQL4=;fBzV|LYX_-;x|HP#T1wRx6krJ7-#-&pxqx%J{;~}xZlPQbYTjFLjK9S zOwU9(ulZ5uX4pYG!0*xPHf0v9HD@ixKN zR;HT7DT%?)Z+`uLKKHw{{q5J>LEY|HTBViR5`oQq8DKmclDeigh=LAK6^8=*u299l zWu|kw6-$9GWH~34o)r`ZgDo@Vcpf*#Z42m_+j?t4UVU7M4fiNGW0C7QM)G9IbeUVT z$wY=}ek1&OqFMH#xg>u0K%T;-w#-V&%?V`@m;dOV-S>F$k;LuEfPbIFTUM8pd>5<> z1HMXTpdJV?>=1Fbi+SaU*le@K%F?!=obl9B`O5>gFOf_>jV()&A)$n&T{qC1q+xBo z2QS{`0RQ|C5|$QL7JgP+F`)6EI1>fL@3D*5bAuiawf)bDVo}MYiFn$blVW7^w%yt4 zV4xeif3;>a#8(%oPiK{?Ul; zI|}$6c?hZC5Y;wjd_`aG`mU<^ZO)VYuE$rW!TD3AM+g_jtZ&^7trA;w?Pl;LRV-I` zh%BN7ByEXeXj*!?G<2eIrE~8R0dR+aagcnZtRZq{<7rpZC!n zn8$u$b9+k4p`C(Z4xmtwCsiglPdT{GKWQforjNoO4g8YtazV7eOwbr6F4RQ$jEUF$ z*!XeT%C%*tuZrW?o@(dwc!1IZ8YOAe&oc2(q({FIBS?IdplL?asqv+|VAZSY;QFi6 zpsXl;ghRWZKxBd%`l&pflckDr$=$fD+M&g+BX6))rm*>-#E)m@7DA7N3OHcN<-0bj z+O9^zD0Mv7fxe%@S*7(_{(ek1{DF;y*kvu{X^*^qeI3U|H`>U}Egb&d zOGX=3H%PASS-9pRQw-G)Z4F9dhD;zt6zOD z!h^%(>2^3d#~V70N_RQfnjy4&ZmA#tI9<@^8@Tcrb%oDx?i1OfXGwNl8y-j1hZF(L zMoGl`rY({LyH2uRf!k*kQw=%h7Bf&I3Kl+jK`%l`TFb0Ua?Lt&yD$j^@t^gaGt1VM z3Ku2LlWiuQo0zO|Qieomxvo&tdD*^L+=mKMlmueZVPi`Ke!51TvB<=u z@eGnC*0(1g^wmv%BM}XG(ey7V=*$WuFVx4|E>Z9G0NgxQ=~?7ti_ROon0*}V>$aNB z)7<pas; zu&?iF|Bnokw4@Tu2{N6}1slomJ1BAiW=JxvT>j>zc=G&eW?HHf?OT z6s=Yj3u;0Qk6h2)NhSGsoXdxJ3ae@xbe7&@qTA~H-J~?mwLxr6l-}`F0r3`;GxuG9 z@$QgfS^1l$nh!SY8a%vOxTVnKVHcUS%oI=0+%yi}@#GRxbxUfqFdir@KDKxU{YQj^ zw|`9)`T^VwM0rmKVfOH#`m(0Tk4<{QQ?ML!-*48!V)ZX!0yBFRLN()Y@(R!=438(((Hig!j$;qr+Fmy;F#o1~!!+R!ol@#BC*P#Hy8buQ1dyM?N( zrd(k<;|igx%s>C&=`%xr;uzelMPBlx8n1n(p>`?9J|1JbR4g|~kJ=>1H}hn~bOa}9 z<@d?(KZMmThz$dPk<2fiN_Ym9`0Gqi7(Ync`FRS(^OXXQvkK$6OtIfHSY$|70*zYQ1b79*QO(=`hzTG;D)1(eGLyFw1T37mO zCUZN#%GeOfrI1S4fovW6VAx_KWHf*w%N?&erdp}V9s=8}Y5i;-#xVTvJ|1$(d#zZ^ zUmbJfy@AG-d%nZB|j>a5coIr*Ex|Sx0;!p5)&oOB~DtaJimEz|q zSK-fdL9*k7WyS5^|B+$sd`$9Vtvcm-Ij^lbd`F(@Jn0x0Q$r&3T!7GCql!c#wltOD zDGkGu_^f!t3I&)(hTac2gM;Gfm`S`hP@iUa&10UY=VG*-l@P#OZi1LFdK?!4fLe>l znOVkUC;W@i3t41(SlgVuE9`0_wsi=8{qw_Mn^c%=LGk!aa@LjyQ426BEUuV~tK^Z2 z;4Q_6F;Vgr4jj{zwj412!p^)ia%?5-q0w@qt%|_%PnX;ZMo#cwwd5yHvh~GK!)eS# zJzCG%ovs`H+h(YAh$+A3#&shnwDKEPVW8m~y_a#y>j;~i`^c@ViOdasFS=%jmgh#Z zU%B6?)>taXI+_?@!cI}AAxAoefi@T z4QKt-eMWt+^opaz*8gRzrnSSD!9HsmiJ8u3y4?ve+%Y#erSw7|4VXTVS3W;GjIC%F&W8QF{@VqG>o!S( z9QwyS_@ow?A3iIjyMwJ)516IH)^rvJng1-VA2O2$y_VZ<@5eVt;7(E$&*IgWfT$7i z?Q@8&mM;=N)qfkaVHaqmks#C31D(l1G{shoezHveK-k0BKmMx7%RE3Yyol#PX1%L> zVdEjlI-<7rU~ud)4#xq@P5g48TVlRoHijE1>+e+oa&J3h3;uO`i6q_?T**vHnfS&O zOYO#{=_&jjMQw?7g(!9pZNe|KpzDwYG-|ZbtSy%fz`P<9wi67}sl>`^`+HCcrC+;$ zp0(;ZO7m%e_6_YNmaF`Vv&OZkxiw0(%V_?}2g#dRmy=vJo9O)Gs#h~mxoYEjL&Z$7u)!N%?8>mfv z>1JwN@o2nyE!pq_^n$%h1@O!QofNadHPx)4VS&8ODyYTv~TN6 zmcW;<8&kEA_5rn+dx78rs|*!3x9rn@`BrctE9x7V)vd2?k{WgI0UA&bA8=&p>C~4Z z|ICY+7#oMDMHnOtE>+>Qt&8Tm=cc2n`3Noe+^A2z?3Iq{?GcufIvl+5s?gp?vEw*3 z(pt;qDMjUTc72zThVTWGmW&WfF2}-oktzfJC@z}YAf_Al?Hgoxk5+&ix@aQ}*O%&F1|5FLn;j+@_Z??};((OFX;&YVS~ zu{9lP(-aXH>h7PVIKDqsGgO~UL1o{z4~7xs$MTgDw+FxO8wA5d;J(e;!G4^(QO*!~ zR$|eyoK2mh$Qak(F|j(0%MKdzv+b2?Su16y7AD`e$$->9nA;XsYPm3W#?6L?8nfgW zVhYVql1q_JH+xvdT#<`jYwqu%7SQyA|Hy*t<=gL0PMyW&->BEQIF#meX#&VIS9`R8 zt4L^r5(t{u$wl3zAS8B{vDGGAWf5{#3U4c3IQAkF$L=0BS% z4Z~{@7segCXDCUVWX?=&n+E|XW(#4pmLDfCZ%bs^@7iD#( zN@1`ZwOl%mf3;Z>pTMV^LD<3YMo0KGNL|?KZ2*)dm+{K7Uq{s$Ii<2v3z1hZAjB~e zh0Crl<_YE49T@H1W|E{=aJCUpBfB38TP(wvTmC48ms{+|m%S4iUjzbhsU|E|2Bw&2 zs4dsuV)SAHOotEOd2*3=yEdj8PhDEwLDX_HqQZ|E7T!^YOI;hZH6N%fgH-p-i5(f= zkoJ>fSs)G)?iN5z^+WyWZn1qF-ToXalWJ&foRZhl2oc#kcPGpmxI$V--}i`WtoR7k zIEzzjiM2rZjQT6`diuG7@{ejVIO)(Wo}t#e81CkuH)GF9ScR*~tbu8{$DvRQ^@K*$ z$f@dwsOKiL;=uw`&VLP3j9RR`h4Zjk%;I@IFI4R*xAZpjkY}Lln|dtP<#^#n7>fYQH-7NULa!g>LRa`thv? zP1CQ<_OujzKNL|)*5eDBqZ1^HLTOb3NnM?pvwX}q8bcQS2s2&CqaED9)O>`_I=+#1 zgLkX3bTL3fyf~S&Nzv@rI_LhZ@Iy^snE`TXhd*J+)VFPOFyzk17i+m7<(}&+zWr5D zdWi6lhu4nl#6eTGt_{i!)SG8+vgyTb%K}30hwR0;276GVBmABsx7bCK?*Uz@{wZ;c zwia>*H{fR`I<(6AN_``&Fzl*#I=3~}=l%HwO@(Ei*f<#l== zf~)`i(~L4M-MHI}2)hFEOqN|73v8wMjD?4=F$2X}+EkwGNB5jjw?OOQk9vCG{- zuM2MSO=}ED`C+B@{Md(KpTivsbJ?gK3Lo1g3;n#_3BSost&x;oeHo<4A!PO?!iv2_gsn2+lU>y1)IScXaNir_A~9PZUa z0l4~XCm%FmP+w;7t66C>D0^QtI`tLa>p&JK;GzaE?iXtsH^0iu4@AwJ`Vs|A{ zJB7=7@$|H>Wn=d|Jeoq;RVKM8AxAGw3ZxF?RgIXnSKIpI`-6p+VNi&V9MN<-Vf}mR zW@La7gO%>oW=EQx&13w^R4sfpZb3kKAFoW8_;I<R8kEGFW*MdjQ@h~f(>`K0*({45P&MUZ84%%>)N*Du*Qoy{ zRT+D&-(Z+s-o+<}Dj1~F@H?rsY~ye?*0WW&%cTiHm2+R$T`*GnD@82a8xC$1i}LC> zvIWv^CCbV=ua*8GUfvm371Uprzxm92Ipx({R793%?rdAxe0?fHtoP8!2Es3nz`{Af z)@ga(Z^-xHQ{Vn&i7^c8d`Uv<@~cbjz0JVHzFM>`D0$cR*!`m9l$U+(u%CtShDl!Z zU*oOg^ku~z4uFd%BEzc~uC5>o;9Kf~|0CNO+NtGZx>W9n2?uGV+4H3=Mu{*H7g z{{Q2*p&0R@tZzwSalwcI6ddVd$cA{DVat(jO>LZ1ZKf9y?``!s+(-SP)F%vX883?va`KcWa|$+gD1~OFPyH4p5!2c*l5z$M6VuPd(|g4!@Mqb z<&~oLqN`Obj80N3tYbYZcd+`^ia(nzswBHCBnSY$buM4!tyG}>;uystyeFEQfUA zMfbRN?5r2&g0iapy#z>uKb}dh-!1w~wnJ94ysPcOE@&)Ob+&OMl7eoTRw*y(a_@($ zZAmWv?b@kgN92G6=i$pM<>RunU|=-T-I2~QFE--EeYv6`@(tK7hy#N<0)RFsZE21$ zOB!B$(nG!hrxa?bugHGZ(>4Is{z1#GEzJX+sWq>XQSAV~{-di3q_wNMv*8VqW2{@hmH<6vi^RqaGYT zXB@o|6fQI(s9;8Rg)hKYIbR{(In=uyC#-auy9p%zz!ekIM{8{>JEk9H-JEP{b`mRN zmP54mnzvp_{zo=%0`rv-$JEA)t<<^x`t53GEF1E6j^ef64ITYuxT0Pc-?&(|eNS@6 zmbZuVJ!-MR+#&?~@D2s=i{0!lt^w2Jim)}?!3CbUAyPs#<(E2IHh)#d$5*P35ABHW z1y;kjT&rL4V;7z-xVqlI-=?|leArPbu4(s8-wxFiV-xAq@5%p2>1%q_QK>W?K&jCK zW-3uIK6zUuS1SgUWasdA%}j!WK?>?avJKaexJwtn3orOyJW#Onb@DKkl4?SyGF~U* zU>~bAp#Dh#7Q7q%1v**)NMR^Qe!R9!wIW9-Ek{5^4`^(5I*Iq^Gy4x#nxa}QRWf4!5@i*FQpcKxzq+s zIj2ZeRu1FF3^rR9`bLtO<=Ml~123DrO)&uhNHB)+24kc$o4+b zRD(G?emXgsffg4BiLB?AZvUX%p-cQwuw@a z1q^hrTKi01ZLMqFaJRxCrb8-o+q2{)=BqH@Su>bhbAv)#yO*Ioi0Yj)ilej$qKK!s z>La+!Hx*mpdtTdi$@f8~N&rQVR(G$X61wX0^S zIN29xuD{u|vojQV*K5mpT55Eykk6Xy!kYGYXqE&8)h}U+OPu>jv$a3VJvJ>E?}&|r zjQI4rEpEI^T^5JGk|oSoq?Zho2Yuts_;IfDADPv0wSafRlGFIYt7$hmD`(oMOSS54 zHRyg&czn|i$&JRueXlhvkWK>G`= zMkTUty5ap^3}z>->6~#)4=gV!S-Z@%yN1(Ve$V_pZ$}xxNkzY?0~m_zfM=HBrvQWZ6M^l`e|l4pkkM(bzb@rY28B9SK`> zLF+PIH@omVgc(3UFlANym zAatNDtszdS@&4jErbf}92~SauulBfo`6^#Kxrt;f8d##^`FHXADLQX?8V<@*!0Kgr zeVH!y^1rX^?>T+|PzQms_yVPAQ>9n1XHWOw!5Wuhf`1S+OziZ&>lH`;QlsGU`<(S2 z&Q(jRvZY1yCCMMHQtK4wA?2XThjB`JITvXry(>N6F2rTqJR%oA$a(X*n}EW5w9Y(e z15yxu1Ow+~o>;N;Zqoxu*ke3afwTt~6L`WHhnd~BZ}?ELax(EBS){=X%r@B1SbUy7 z=SkAGdFv>f@ZALklJ3#@?8C%+(?IZTX!m2_s)enZy(f5GpVK$$D8w~ws;M;h@e)QTPJC>yo2IQkf3$^k*1;jY`Kfyv-p6rb6wfD@81QQ&)dKUiSd3N54fdiFYaP#_ z6(XW9UE=Cgu#Apw>APymFsh#72=AetU+$jMxv^`TKh3UUe4&-aHKH*;ITc)ANXk;9|<$R`Q4=moB z`u2q)_Rd zQy+0iA^%80VPe>tEw9373@qi~BKcjGLKd|^z$^X6bkzWAj#RGJ=mb5@03~CLi=^3I zs@#l0NhJn!_AEMZ?jD}k#Uh{ znI}GnpHJCw%>TSGPB9hky_^}LYa{$>lGXc-xZdK!$izDgRD~2f)W)fbZ;d(k&W-%p zyv^#<=a52UI?FIhV*>%YR2{L?SbMt>@|pC&;zKR!+jmT2r|X@d3Gdo}ZPQ81(-z>W z(g#7H@?p|Vrx>9{b(tkG|GO$b|IvE5v{&(S%_*+qnSlpGZV*ab`CIMmt?qxRnW!*qb=*-Ncvv_R^u&x15>(J!dz~ z`(@%Nu-Ovl22oG+PXfhVf3b1x=GEjRiM0k8D9LO>+mx==H@jcsu$$X5P{h4O#Zd|= zijE%MgmqA#(01zETj6>#LAuQ2+!-iazQ2SNO^=(E``7>bMJ7a%U9o*BJrKJbI&9!T10>)f3wZ%Kv;1Qv2|!lH%Bcg39gcAHRiin zoWt}k2v7@MV9zS0OW4&Q3(tPs2#Q37mMV_9x(6tpO?Hv(%O<12kiz3m?VAX7+Gtzm zM28FGqfp;2&Ix=KUknrp08gnNo+#t96SnIJIL>q{12N{mH8d@cfY1v9nC;*Y08p2J zSa|JDU}-F$H2R4$0V(n)suPk)lUK9lm%ItN@r05oRG;Ff%j=7RK(L^WyvxsN=1QtL zaeVWO!jnk9URlo;7a6|1aSG)XhVk9ID1(vWLNRah=3O;&KpA#z=jW9#eGIAq|{w zFDofbri8?Td)<5@U*{CCh080IIKWg5`n{iI!Zd7--N?Ri&MmJ6d` z0D0kvJd<|?n)Zzzf^pO>@hUtzSsv$^2UfW>fn1y?Lq3Y%YDFg#ax^@|jm!kr728`V zZaHAr3kmtZyup=~0{sY=_U7Tze~?#2w}lajLtT*u>4`&A3A!vZ0a3agN4QGC@+yt? zI31;i=ZZeRXjmu6WS(CcjyT&LZI(}qXSj}Oo%JLa z2c-AiH_ZeD!17b7aKgTyzEYxhzu_@Wwl9DsgvD%x=pHOcWI6&{*=2=z3ME~%s z_`Yzk&BA#y5i8y1vpfF%>9G%no}Y zcQ-SLS+aiK&If_acA?7902~^E3hAE>b1#TxpOwMP$nCi;q&$Zoy<|5h;L6WJDF-XS z7A|?LNkSb=erBW1hl{jtbe;D2=DTRj)i}gKl{Y>1PA}oI+S0zVUtKlAGClilOLODN z5#sHr*Z`Ns1+@3I3JC>-Ulhrbb}|0S42b*4BfZW)EY8ihrkec-opMl4zW%>loZ|AC zVx)Jun_#t{@x2}=?wc`ca=n;C41@>6K;U?=&T9$%spiuYgLA6goI5XlKpfmmJmn;c ze+ViZUfs;pygj*;99M4UsZXQ??9gR3#ZhX}DDY72$I$gH-<>F8bVx_u*{m}mG545O zKSE>1%cJ^FhnvUewRehYoH- z75fm0^yfb^7A%M~NthmU_`Cf_i(vPy(*z`RSZX~T+8%{S1jxy5t0}uaNv?iRM+ zZ*PCOw6yP|@Zk7n|CfP3!J*CQ79M)ABo}FQ#}#6Qc!cno?K(QWgS|*PET$0RBvMS6 zNc(exbsHg&))cc_&vHW)+q;*$n*zMedJy7dZnNXD0K(so#f+lh0k;amHS1YON%1V+ zMc`Xw<2!ZqK}e;`)xnB-V@j{SucV-`uV+*(k#dKxtY@MiU$uR@lP?_8ztk$7!uh6q zdaImj)`XJHX^e}F_u8;ns?PMYGrNqpi;K&B0?ILXS|P_956?8t#w5XCoWPq%=gKM@ zG332>jA&R=ZI_ofoF>cxHzC8WJ9}TNGkVb1Dl};8jkd2y8P)h$Bzej}>01NwI49)H zrErB4Du3?ar)}#MX7=0Fzv-4NMVbZXH!(tXoU(~3wi6~+B>{=l3ghe){qz^nBPXK@ z4$bga`UZR%$5s)CP1(s_ERtZKL1FX9i?`)psJUM@2&io0hJ6JMMcox+p0y_^oP<({ zEcQNzY;*dsMRI&Gfc)nCin%%ZPy90OL?v|d`x#+YS}w)PT=aln^CG6QJ?LTCNnRFC zMzW8{*xE2}U(?r?Lhso?jNR;M0IbJ;AzJ*FFZ#SVB9^mDA=I~kakE40j<`L1{$YW4{#fwaxW=2AJ?3&aL zV%EFVm8M5=|09#;XnN$QC3z@YU{<1hgxDd}t{QzNjz6r?t%R3n_*%g!v3yz5GrxuR zJ%R1+I0k1qUEfldI|`Y?aU-j*cGL?v(&5Gt9w=osRn4~LbjgbQef~_9emH|)Ggq3Z zCj-K=a=Lfr0rbES&ATg7N9(|Ng+%t!e$(~vSVo;LHhA$B^cIVTdqwk1wJ4~|z1wNqS+E@wZKTopXzorhRICclw*=KbgD5mGF>(Sy;Qe{M@d^Ive1 zvNvzz{O;Y06L8i$jS0mcbRm;xxypLTo6~7uq*uS4aL&bI>t+-l z$Jm|BP^&Vf$*WJ`tBdr+nL3K(`i+~71i^VEp2poSJp}-TqG)H|dadTWSK{*Y=(3e& zh|lQuqdZ|$Q>VIPsM|+lG{pITK}iEAl)0DCf{^fLqEQHdsj@-0c8Kmc&&2;LVe%Qt zB40w^%gz_Y3k-(=9|NxXVnO%#0f&8KAc#goUJ1{UKHx&=FxT81`1C3O4|O_TWKBzv z;8ohQ5Zi@0!F)d0yo_~2mR{JeT#l90w>IafKRMk)8W+F%Kbqb$uF3xm`$ti+06{vH z6p-#16O|T_hEdWmO1gvzNXKZTyBR&Yq(i!-yBR%djPL*Y-S_=}xMzDUuFv^7&f|C= zGD1-WnY4%R*V+0dT=QcW2ARj}WxS{k71+GX2IKeTebKu2CZ|zon(&PH!@Ks86crkc zR8q4IVBR2Jbnk(}gry!YJXnj~kM7Ky4 zk$o^z2=-nP=(-+bDD748P2Xo|^#CX6*?Wic+-lKu1cKOEBhHS{FVSR^AW&HYDG!rXoIuIvB4 z^weoz&j_Nvwv4PS5LA$bMh3TfylmsPK8#h z&()oD>0Z}co+SB<$sNeYr zR9D(vw1`P!3Kj5G1Qj+MNxD*8L{<3XR)ce?vPH*TJ%%)EBOxWBiiSUjT(1NV(-)aY3mgezvE z)c$x-jRW&KkQi@99jO>&MT@!0eU}G_dyVTuP(hqaJ_EZt8x@mdT2t;{JxZnp<_xmi zL;S1-5Z$XUZGm~sd+eSxnJn}2!xdGx3d#4z$Zz&^mh@eRc+OZQrq9s_;vQL{g5qcq z{Cl`V?8`&(j5Vo!?+eoU?Y<4GSdI-6ItFQmXL`?p8I{pX__g zl?CrhJSbc&h8xtui3~Tk(q(XTeW)w#Z_3?fX6%#?4jp+g%Azq1`UK7JZ-Cc$X}wGq z*jD$E*d_10Db*J_^8U$t2Jb%q5}&LH(X~y#Fnw9KXTKs6ZJg)0)Y`b}hSv@$G&iLk zclJQNBrZL6hl_xl2}883r9o^g5e{nn_i3)4AEhmZpT4iABZVHQUGS2v6&; zQn7w}i>KPPL-{^0rOG!B`Kt7^zy~B(9QXal`uD}AEa@JS=824d?OXiCRYjN@bE{UQ zJaDngZ@N%w;JVdvGvHknOw^jbn&qmYz zMMQVORBHm^sQ@$_D!j(pSnB0Kv00l@eJ zFpx2~qSk~_wfOklV>>|Zj;hg`U^=-;)vL4?WNtCN^zIXHiS-Jmb-&y$*4ycn-{J9Y zCtfx+Z@g*yOC3ET-gqc(QQHD8S6WRKFwo6!&Qy1WRhlxBgMVBRICtD(G$U=K@4NkX zx*_ipuTY#wi})jNHLoB#!1|-oCpx$Q><213_9h^$uOBM#>g6! z8navmQwUpsb`h9vJBv!e&e3_ZEsq`F6D}u{Rx3m3cj*b!Q`$#j-ms)2~Z4(^l@83y*??cB&5?RkVWk^#KJ2D*rt$tWf{$cZT z`=Sa=U48hGFHOBrC6Q45_l87i8a+08N7y_XRGO}1*4(UT;jp{|UA1&}BSlFUAB>$; zOw-z@uFKsi+0aFV?gNdemkvyy4G!jo;~(Kj`VFXH$^|kCf==c({?( zQrIc=GDXvb0PM8(;?jHiQ{u&i$QKZq#!6Ts4UVA@=qbn%BF$7Qvft}y?>Ox$%SH_9 z9?Vt@vVBAoYvZ&tFovLdN@>;-ZK8|E@D}^$c;CiIZ;bWimx`lphSP{bp9{}N?yRe; za*pkk4>%T&E{=N3W{59`l5MT8y$z-=3?SkN!GATh2y)eT7Ywa@7`dt$1eYs4VIEo2 zuGD3?zW-20vk&&*r%Q0xWsb&MBsr}ZHVw!DL!e1@vj?6w5-}r)=+!e7bXh+2e&qV` zuC(*x(fMVzjp#4bFUttlrzg5LJ?#I=^T|i$K$tB^cv^#+?@InC1P#H@3_~=PDWzu zQYNXxMf(?QyI9`9LSEmMrcNThO_WU>VWog0qd2_bN-)XO9W*={8ih6;0;=qi^EoMJ zpSp_ElY5Big}~}-*@iFEc^2KAzG~4WnA;WT0hYfNSNi=dz>25j;!D)ta^h3m){yF! zECPV302*LoChFzIB?A1~LwU9bVl~^s$&(<5-D>-;aYb?HQ%Y~Oeq-48vG6Lxw0NH% zTAw;_(@fB{r!6q_DB{IXj+)?>H@W0AOLC6!{5poG>&1`3MSdQXCOX}qVvDHc@bemp z%fq@Ruh&Dx%e!qNljw$pkEaYAdIM)&PCE&p09d5i0gRwv+@bqGIK6;(k=am=F8I#o z_k>F_w>dZ4!t=8;`c2S5-FS-+K`QH|=B;3FJj|`klQ$|eq>L%eNZ{puT~d={Q(H+z zxVLKEk_e7tN^Zg4q~Y*c6yE!PYtnAZ%qIfAR#O++y0-Z@b;ToiMd)&`%Ufy=UTgC| zG|lcTO$;}!do=N-onL5|)YYOBSPMOs9(y!7)aEj-LNC<)Q&zY0(2 z?|{Kf4ILlx87XHy9O=XHO4((Jq9yyswPjO4FO0N)Cf{ZsOveA{5zQ8X94ka&Pq+#nx z58@ZRSf|$vPTl=&hxQC)W38sC_T9@(-FbgS&z=Fm z0V;IWph?_bEJ0dPnxblm2z8|RO@)O?chq;heBB9`>~|QY4hCk_q$7}f-Py9&uL1=! zDrtydtbgbId#0vT5pnT2 zy$L1KJO&XF<3ByN`%PLypIq|;&5UD9R~EIYDH3(%T*Yhp1%GYydx=QsJIW7=dAbU$ zCCpXJ-`QBqTM(_*5A8w|TC4kBc?RJbt37jOk;Sbrbe+L_w#twu0n_wJPaM z&u5wx6+dX0S=2$2Ync9R`m9V8kDB$1;GLC^zn8*&nCw{aKOk;d>(&A4Gp3yB|n0@n_XnN(bcQYj$(8D163N}zsNpYaqJP*wERF8xY zCUW0IU7Sfh+G%}@&Qc=ci$cDZ?Nb>n?6b9qaGr%LJW^+ug(?RZP9XTqroLpXB)lbh zknthQxi6pF_c1M7$OYUVb|wN?g#Q6sX~68QdZK-l;pkvW(K8Wt9zA+Q2GF9@E>6Gs^*AGTb2i2m5a z^DzDe0o<4#S{0FD914`2&ttZOF<*_$sCSgqj(%er#PrbQrZXrR`dF=3+nxaWj1ZFE z587|`i$bQdkuIJ&tMabr`zAaqGlMV+%sUonnTIbYculAUe0;E%*;`k(6}HY?lVJdl zVCu`s{Djgu=(sA*)^eK{InGpACg`=(Z?j{L+Cpp_amF3L9`nYsd$JHS=oM%&Gd@B) z@54q(?FV&St)65$^F*zE=RBfsJd8x`LYs8uX{r(XwH(jbJJ6ii@gpCoIbEr#oT4IV z0P$8<(r4T}kl8WPjs>%o3fPX`|eEaQ@|uSJ>EO8A)+PvSJfkm2|HOwV)$;ia6&2y zWWyAYQw%S!f78WFvw(XCKF%~++nddK+uIpREsJks{|A|*Ry{~~>$e2R%CK^kRq>SB zh`tSvwm$+z8H_=oszuWmZn@Vqm&Vv5^hlQj=il{-#iaJ=2hH@}3z7}TS{4n8Rx4j7 z3-G_oVKnE{Jo8}BGX@D^>x`1|E&cIg_*1-AEMutpUJXB2i7?H4<2zP~U$nu#g-FdR zKt4xqdq{LW|A4FL1Q6S~m3^>yu->L-^YFe(!C$c<$o^N4TiM+V-fR#PjR%`}U6jMr z-cu>jiEJb5kq*lOVVawcy*u>%ZX_(m?B+I@{8m7auZxo7RsXg3IyoF_hIZPTYFTCL4ma8`d7~F&~70}M6fk*&W^I)Us2pNTQqVL z=^D*Logw>QJ&8ZNF@1Yg)LMs8gKnQHaoGp4pb0nPQrG8IJN-II*!OrtA2m|;Q!%u#S@XvqMa%;>Y|f?J7O zd0CcHHTO(Ut?A3SO1r4>7}AZ3bdl;+*nc>x_@KkTRP!&N@B+IHExh)edsd{ATqJJq z)r(p%NAu0bM0r^~nXshoO}zX$Q*at4;6S&z2I^GPEUwjX-|5!WkYa zf%`_4rzdw);#UZ#>5g=d1v#bv+>eefMx-4X&-DY0@WeZZH20b5t*^gk zu3<|Ce3)iamU9MI=GH%*ae1|4EK?R|YOR*%G2MprNMdO08~%>b##JV>Rj;caQ&B~N zO!k&7tWW=18a7v1dCX{b+0p%>5+&E;f`xWsgKs06^%V7VfGarjJj_WK*f^4te1AhO z$<=}rsNMi!!wS0=m&Hg>+Sk{xF8t8`0d{_H6IluFhRuTu9HqM?{o{DdT24Xp_Rh^L z66kqR=GH)yl}i2am16LL;kb;m;_K3h7q^&Uy=%K4pCc{Ik-Z7ZQjNXgwWo|3J6Pc!$j`q#gumx^uHl#;7?UPzDtto< zfnnwC8N%0|kDmRc*{Nd9)kXvsQ@k5@YRTMi7(3w=D3npoQ@^x1Y21L*#4aLQj)qy> z;;y26KnHGYj3;Sax@1Za=}wwWJ=_)S@s^h5Bz5OY-)ZvH(nbbSju>*gd(_^Z{_tq} zC7cWOyIDv-F7{c4$-HiwV`l9PwXdbh8jLcPDfPA_LnHbux9IcE46r7gj_p%?k&Ub6{~O0gRy zeLyvjhB57?uoD)1_c$X zrE;i+?xx4h`g){%FYFq4=^7a;=n}T->xG5uxz0rSh}|vn<@s|o7ACWT?za^;oMNj- zsK2|4yuoOfTSL;5!d;s6z(hSHVkPhky@0;o-j-vt=|E+Mo*n`$H9HJa4y==RYFp|BLg5=TxyW+Pu@BuDKwYF)8PZNm4!-cQYLfA;< z-SMV=Kv2*QLN*({vZ}K7Wsk5O8S@$+>Ud^2FQ3$J17Q(^gRv830pXko+-@MZ!2~XPJ*^5R9u-~_Myun31w{>A?vI0 zHSlDbt4mUrR>X^+bu%AiZ74c>TM1|4}T>Ynl6uo9Y)66h% z0D}MShoDzVuI@_j%rPZ)L-+%z)%pjQs|fOQ^JF>QGsc??sjCkc05zvVmHbu({@%PY zLjyUTOY!*WkatDVWf{Go@FshWy#Bd?rFIy_lT5RRf9N0i4bdY^=X8dB#B$i)dyaks zDM5{{)ztxIx5p8emvu9mFdnWy-P|1R^RRnfuhMhz$C*1~Iy53@<7i)h@=oT*yFH6v zqYjzN*&wWJHUebS zY=aUa0TW5y8{y<3!LQ@6hp@DTYsAm|E*^22h3L@ITaKO=HmP&LB;d-zbAdqK7}L| zqqCKY-%ZSs>)|WAY^jr%!h#Ev7xwAsD`yg16(15D>9fk*-coHMoA-z!OR^(asiA-A z-lWPjJ{k9m_&zWf=rCkmaNYKI)b^o%(H1xq+^wyyT9gtLD>0~A9O#&!H%KcH^>AY8 zh-s&Ba~)aE^|lI0#JYUsJ$an%L-~c^H+Z=Z(kXsqlopq=;c{cpl#vH*s>rzCtI6^3 zKkME7h213Z^%wCz|<9PxZXCL9la{1#o4v4I&s#kyNceg%b{*KS#`a*rOI|T zv7?43tDI^&p=I-x)uEGG7Exi@QOXa>2M6Ea2=-Ya>d(tJl*AL9Cy*GX*5RjXtB&dd zqJ{}36n_k-`DM>$#S;rcX;7$;?;fPXMtiLvo-`Ndpvw-zHNSfxviTrY+gCyn5y}MY zzoOUVNtX=wZU$cHqn0x?Vjo+Ll{uI5cw6T4)HjsYbV&WVJJm4zLbK|t`7$x^ktI^N zRd$;P#P-P|I+@$n!e26e%$-WzZwy@^7o3r9{q*W?mr^o5@9Sf?4^EEP8ai*%rmOj9 zBQ&_O{GCLy;>PZu!}_-Xw$C>}z#NYmm;EE%7jXVt#27i<hh=me8m(>cgL2 zwpOw6iJz5gi)t<+DEYW3v6N>bJvcJ|ipGP$9r8Rm47WN%5TX0J_2GXXHNGpIZqK?x>>j<=54GD7=h~p~mywXmz19JKdzvZC8@7<O*c+@Lx1xP1Uo1Ib5|(kv2DEjC7lcJ z9xov6pXVmAgiEo2#kN(=8=^~SuPQqFw7_?oMzPWpIo>w%B>NFK4e2VkRoHIH_GR$T ztDs1#y43cA(`TWT@-EF1MQVlT)MP`$jL-rqD`IPpyRO^lY) zwybBXg(&+EsNLBX9Ctm+O-AN97My)bWK3!TFF%(2;|DUD^aXnL>+COpst7Y-uMnc% z%;q4{RUDD$xs`Oa`Xx@A7aRbtMYU>rCqU~%+}qBQhhtn5WmP z9%J;s!iWoehO-5oal0}J&8?m%*e2J>hi3ShO+KchlmO?y_c~Q>N+8nEN>K3Omn}-) zg9?+je?6ZwKiP1HiPNAKTB=(ng)UA1?A=RS?cPRwt@!84#4;oEm?p{?^jl5VHu3w5 zp`K;>-``k|>5u6zVd-XzEwNuF`sXYJ;{+lW4aO%8^b9>5RL?gbS|+~WmT!9R#dAkZ+hSsK53+FXB5SGoSuX!)|?zO3P zxyLonVoMe#3L8$Co1jT&Lyi?tk271`?<5)3Ku0bfbf?xj=cE=FGIeEmkn~wFKb7VX zFE0x8zd&NX;GZ$y@$!A?3c-VlWJXx(;p)YRn3r~v`W4W;Mk)_8Vv4+JZ3~yD2lL07 z8crgcuj90>3uYX&$u8fdtpa^6ehXvTmX3vb`;L;$c9z7~Nr!*hhM6*5ZZ~F{NI)re zy3;lj_vx|wj_ixoHTP6^;I$Dd3%kc=Qx5aunXi{+be@QwHQd7#{*vK!xpisHd|k4c zU0i@hraa_oOLGP$N^*R7?PwCl8}j6#n%H)iuAuqh@_)W?Kd9j|<8 zRUWro-A-^0@44y8$#P^K*H@IYPl?%Bf2Ak$D1KR=^eB~sY=85@f-b|qX^2)sDF>qG zF^yPH+5c$wOutriuNg2_0;%I*ieA#Z#ubOjvCZ3fjnuqMwhgL?dKGcdd6j3>8KEY^ zM_jpXYy++}W%^o1=D_MHW~WMWdThUq@IRbi&T5U*u7wz-N^wMj^5@}yal0Jw=?Vw? zJe$%QKjB6AI8YNm-egUIVM%ox8=t&`A*`FqlA?#3*!s`2z>Z0iv<(;KBL|%+Dq)9x zAujKpD@LBpWe5x3+#Li=y1X+Dz^I-6*1=6MJfm01=WGv0bKu*HVnNcFZo>cQj;raS@3B>+zuJ39U+GFW9!YK6%IADW%@O*?<;SgU(}DgIE0Ep z_b;oc4U{FB;aN^%a2Ud~K8(wLtwqeG4h4;-%^~KUZ2ipcY&|uT|4gdXZV`wL_nsXL zMgAyK580p%^rHXSV#K_h&Jw=ySbAG zT4;oth%uFG>|vcDAK{9Cf-tkfXd_|`;k_y=7~*;cYQ2x_xTs*>+jUp9N0URv^%%VkiYF?MMhOz z0o^rz-pyDnOgzn^qIkjj(rE7XTZa+4VhMf?Am(<;@E4QD<)I_>5B|ug`{VO+7Bn|% zd^UY6vT$rb45Wif*?BOYh8}y$-O7&sRm=0&VJq782COyKc=x^-k#z|=@RS~J-WQC=>6TuQFLR^03I&E>rC1= z|IL{*`(VTkep2QmE*~Ya_@!&m(62o#^rQS)%~a<)+iWm0pnY#sSh~(e!mt+c`rY70 z1lFgtBv@ziPSe8WpjNX{>a6|?27Xe`uQhc)$reTT>O1dpa=xyjqgX#>^xFaB`6esP z%^!75qNh4?)B1b!29(*JYu@Qx*UUi2$>0DGiPaRu|8#ij`3i^RrK3T5QYeD!>+Wq0 zWA69Zm^&=?AZ7LkD=6*WwavWNGQX_MSH_cYdYXgOcQOLG>w(Q=5 zrBe~nMe&rM6txGog@3!9Y^ZdZ8l&T4<;!xW1q%*~i?x$lPb#{P-=jGQNjhh)pruY; z0lg|kWkczYtiDW28h`_66q@hEX7F|o(9R+ra3hl(U_JArF{OMWf z8$rvsGFKxL1it|qN9#>sDuIK{5EQ>yz1)!npLQVOYSVpa5~cTL|)F@)^9%j>3y zL4V}prp+1`5aW;z?G3?4noyGOvp@$H*hw;^FbY1%zNQ~mXrKR~v~;4O04)@1D0jxI zVG)e5;oH>oWVkG(e3naKrQwfnW@x?YFYPkXm3FFO9!@K}XYe&SP8QF*U~|X1j5^f~ zE=qf=ox3zwy6LAwKJVt-V{QJ{D@|ZLnqPC^4H~64>-PTFszo(Rc&r%JaKW#!o*!mJ z!{Tn<_oPPF5Y1?>F?lo7HsAO>j0Us=Eqcf1?)*h+GM{BJYU=*iPKxpHLlEIAAVh@? zo|nUF{(mz|3~<~;?qEIskBR*MY33cCxDjUPk`dnP=|bF$1c+4F1p&*4>DP{G$j^%! zat$#Vwef{BvDFrSDiBMr7Unqn*1hUm6(VM5csppuZ%*YNjeO-t>^cm1e@c#1D$xebQ+zSFX;Yl)f*+?N!av)|D_9rQmDHtU5eAKUq5Un1Y}EG!74)ja)7P(qpYO1(gUpGJs>#ab+?iR5i)Wz8cif zr2O(1&6*Dh(PZOxeRjPO?OUDaJ>?&>{$hCQtwi2Cj`!Vz^n224>nD8@XRW>q%WbY{ z5R|EWW8X|kirZSSI+q~&ur8bSqk;pzA#y0%abJ+5x^F{$$GdVndDq>X3Y3U&O@WYU z_)Q--Y!o*9M~`>KeLQn!#%vxB$<5ShmA#xKx@xUj8sT~D0y7U@e_S*C03%fK0N<@7 zlX(y}Dvf)6E#Y1cJ=!!1K{03}ClS9N z%rl8*&+uR54UcXtvE?{RQRu;aA{EYy8XF#LhPcI9ZXG?>A`FKv}i5w%e{Dy2Uggr3Y}!&qUY)4P7mL_Amow~nen4-2b< z<%qBArnjfxB*4DY?DVrv9b?{$BXi#B1P!R*j8&QKx}aak^$15fMP`PdV-AThP2JRZ;W}A8MRb zChi%E6;p@%tQ0SDMka-1)z4C^$RW4t7855LXf8*uH^SCAYcnfVz~L=+>Emq+dU*SX zV&Abb8m9pWEsV;I@Zvq0N2m^s`zjFNerw8> z%B?|&%rdgWg~?Z6TY&Rtc#jFPn3LE5+|;h+G8k8UgkWeV?oC=}W1Yy)>>)ep%heww z8^NO&FwN??M;Y7DJq@D$x_>nr&JywjZPm9z8}0Qx-dZ?AaLQsDZHb`|F7un`k6d-d z1)>sKK{4yN?vFQkOuvS z!wd5hJJl~}EzS$Qq<#AX!S`>2@p`lvUEZ&)K5qYAvr`O$@W=GVL00r&#_KclV{lIL z>(hgMeD~}Y@Ts2S(C0eHU$%3X5ua9RPtKqOmQu&xZa+n%GeSU6>_|b=mzci%#>h3h z-fx~3Pl_kCTHJXf-SYd<(r}Vb?rUcPZuK5SGK#=vP361ytZ3oCp`zvYZ^&5N&FS(k zE3775(_gH+MKo!ZJ`LB7bb3lLKo58c!`54*@)obDglc$+CtufX5>TJsCur=rTT@Ia zP)XUFLu2Qyv=}=79s?UYL{cMo2T4bG>YJUlV*8-{y}5|DlNSc?Hvi+(U⪻5Qp?YekqRDMXKW|NU=(4qp-1Xfbo; zHDFyHI0KaB*V(j*H^EPn=f^Da$BI zS}|^PaV_U-)ThxViuz9g(TN;s%8Y>9yo`l|j-5+REJL_#?LJdwo!4%3)|?GojH+t4 z7_DhF)?j>t-y}Yvj&MAC3V70=uOnM^USI0L5=l3r_#uWbP6_f0RJLA@nkbuvwg)eI zX;r0PSirO|om|FtI@ zym|~A+Fe_%=eRw>Gc-mwUA%@_NEhSX=CyI221zCqmRVOpFWYS@9ph$c?~w~uDpxTp zRU3n@MT-BXuMH=#lzm1k0w|>Xytulp?YyV3@Kdq4mHAA9)VB?D3<+=6{7{gZYAK3P z9twi#{_Tok#z~F*fuS!`Ri_O9njlM}-p(BsqHOmJ_1qGo4#m=!!+aWiL=8olii@iX zak6UtFNBc87Li$q+QafMww|9<*`vHbkv!KTW2tEZ4lv&6IRVP^_}mIj1_YD&=ofh}@<7$oEgv?&%?Z074YK zZ&5^ZH=WA%stV|_>Prck0g{|4%2NfY^EH6e!Aq0GKi)O_fq+fokfS;|*M~mkCv0y& zeHSY3{%MB0Tu}EpGGMS(jJvp$p9>3k^|8QFG+FIE^RvQlgo~nuKj4JoBfo?y&(}dneMtlUx@HXvFKv6G@53HIn2b>sxr;TI0eRV&M?Vev0whi|8w0MiPXSZ; z-poZMuo_pzHPq=XF3B1mZu-tde?Bf-QB>|U^9=bS=F*kUh8}I|cYg%{*ixa4vgY@L zX`V46)?BYv`?~d(IaXWCz2s#})O?dfH;H?O})o?TOEBv+@Bjn+!(KmMV66XQHaYK{Z>C`k~sR zA2lx4zhRX30hL9LKN(Kq2~J5P z(A3UNM*_-RNt9dx_g>#zj`dp;vZsbBlMZK2($D(=4zka{A-m-?#p1$1Ve_0GL}=GP zjR(BKNCeN`J+R-tpD?CotF5eIl6@rs*S~E%g6i3Lpk0KMf)stk zrmM~}%s?NxF_w*YHa@m?5{{PG96!`C=fBi`Q)Mm>rdpr+Xi@q|=OkhJ`Rd>VrGiW= zp&t7}3gA)gkx_^+XLbP8UZWF5wQc>FgZ9k3N1{ims_CwjS9%b=AC=OSead0&Q?q(j zc0at0ardKd#C~z9&8V~d`SC<0vHNP{SewJ)kyjbap_R~~aDi7dQW?Ncgah}itGw4j zxnle7!H0EP-JC7F=~<>Zh3&;fzQ?T$OpUGo0EhOYq_IWN;jGu8k)@vVmaOfQMbn&j zhj{3MPtWx`1wqCiApzl5vQJFFzr5aUu%49H^&oz@8$U`JZV0<%6Ve+r`qqMDImC2z zRSvzMSl62?NB$WsmB3 zJ~b&w(lj0j52*rQ)JJtfLJ*ca5g+H)7$_$L{X2R}7AcwJHFy-KSI;v{IVjf}(pSep#rph)krCO$Zb^0aCk?u97t(3vev5-h1U?2bWaz;<=zGOBWd^)o%}PQRSB| z=x}_A3zvYes$y5vmvc_sL%3ZA$_2d-ex|6G_trbch#r=JT4B!*D(*n}2c5IsT1kA1 zRx{(|_rsq@CPX{v018Bcg3DBX$W{vqt2d=%tx=FbjcFY;bex@&P+2G8v7Xt$J_NG0 zv;^Xp8@HNvOorE~>M)Z1Sw8C~&KMTr$l5eu!329=YLR5fH%+ZE&1?tmx^=`f&hvnO zd7loTAP6R3o;isk+aDa?97*2Ty^dMgGt?*UWR2{#5w1pUw%m#Gly>+1n^=$?xaLAheES;#X>b$ZlN2J7 zY*6RIb|NBOYX{gB%l>@fSXk}*?PD{0gdB0?ygnH;awf$yLU*GS0 zToSMp=~)%0mdHgRy`*m;ZrOlCZihfR4VepNLu#Yo zP?sfFzH{B<6<(2N2ZqDzE00<7&6)3mD+gHLllUAvnCiPOrMCwNtgt;FEgO+^!dtw& zg^6A99d?TR*4^VOz1{V=4udBB0rrmwg?8K8lfZ;024c1=k>dNw@4$oKW0WWFrg;OBXk@rPoMvT1WeV?rfEm z+rxL1w2E{;wXo<#D#4<R7X8>P<=9BWO8xHGIl4)_!yF+DMTpbQRz!{Law?S14rA>}3>$ z$da&dkeCO**&)poctns;*(HeE`&6)uz4mjCa}t?fuoSvXd|15bPkFD6@znEdPy^#* zm>4Qhn(>)*hL3w8=>X2mmrBxo(Od1TC7maD5%;8i-i1ywyw~dS-o#Nx#_IirxsDOj z0|;G;$U=IBvhkSyvd+HTA5=R%^|JSA^#3Q4$-nvskhR+modM1su&4Wf{1fFvaEAwG zRfBtAJdTjiYO-^79$JIVSwQ);ttD$PZX7 zR14Gx@jDPwV>BK9IL+aUe66J>@QDzjpsrLlNIoC0Zap;f`*qJ*rGv7Kx$*ss-{#K; z?!?=2eWw*gS7_3ZXH2H_2~b^r-NPmshObUPY3PeD3!SKZUTBrlrrCVo#qZ^(vCIMz zEP(q^+cNgd%}J48czD0Jwj2?sTkE6Tn-HGl-6+V_m z%xtRb_Z8~3kSQaCWaljmKf$X1_hzu{2?_DHkz;3?n)l}qzZZ#DT_P|`WSmfvSplX8 zls}tlIe~Ax-1|SA2Cp5Y#fo8ItZ0^d0>{>KLqQ7fi1(PMq)aED1vOek`3# zg?r#fZL?=`6HUM!#VYsCu+`$8+dMY&awK7^1KV&fF|-;Iy3XH z+`r#CnOygcgUIB+MXEtlfc z9*H6-BysxVYU^$&MSLpK$4&pOui?%nX5tWk857FPGyhRoHA%S_$I94gfF z(B&(kN=0!Qjcc>uPSchV@GCocHe)%>Z152siHximB2hh^%qc_*PS1e~;fHHtX)G`L z&E_1qP&vDVHx?>at=NOadOyKhVRv!hJ$&+i4OnRoi|$X4GRzy>oqkyevfKJ*20Hd$DiLGRSP{F$$Fb90%TZw!U}#z|n1K^KEB@}3^; zmXpTfrnnIqgKcfw%?FmX*TDs4Uj88DQcLQa2G)lgSbvq%Z?5HF&B~_6Ddups;6pMY zaPwKm+J{5nlPiYK*U+L!oazIQ7W)So>)^j|v*`S5=wx}TAB5Coq#yXM$z&aU^dx)r zov^UJQ83h__fwr?dQS2*v$wc}-^<({xY`HkWn~ zvZu9S>zA~}fgJJ= zeC;9p%T#SCA+_~oQ&6hgsWnfH;8HBtV$1-bHKkg%p4OG5SY+S_1sDe~Zdl>KjHIro z?Ot?yh1&?t+&<5vaxLmk4IcdJ%84$YxKge?%?M?JwP!eE^4*kEmbe=RZnFpG5dhNvf zF&^Gs%m};p$zS3EJHCqV+-+w!#R47A5{zrpHahEXln?W0a7SRHf@7l+Ugk~LDxE!$ z>mrQn5qO#)Fck2$Rd8Qd-!5~=^{v4jgtxseH5hV9k6lprtULeqQG#~d+8G;NvHGCy z%kh#SUW%8F0AhLQ;GTZkrCHnuuj{f?#9k-hJK@r(T5nakeQ(-{hjthHKr2miDrvjV#6!gH<0TbGeuE3HYC7N&QfFE43J3uj%L z;V&Uei`GJLXID?w+ZpjFl`oi=Q(~Iw6+c}9iuP@1o6zURrs|27;EwjGNvGn7GHl6K zAwl}xN|Ncjq6s*kyX|{QNm^^IN+Sa$&mzM?e)~+Go$_(iftG~k+6yD1IiceG63(Me z0`HBgt1?a*qeGpYS#%;f)o@3*BR9hvFPRKr@6}Z=))#nqB#5lfzPa&j9nt@cda}GjZ<8gPH78%PpZ5aE{QuGP)?rQl|NA$JiUo**NGlzZ z(kUQa(lAO&7^9mZAR-{$T@oV(8#y|Z8X-9tu+iP!vG;G^&vD;>?w=hyj$PO5^^Eg; zoYo=DdAbLt@Irs#GBJS|xYmHgJ@buaI#R8p-W2+>k|X!>B%0@By9$1K(p2f@L&8P5 z>M!``k}leF=g-c3y*Lv_udY_gP|o^3TN_(<_Iii&QO^dZzzNTsUd-mlisT*Qyii+# z!d2o?TT=r;wb^^II?_uQ=Wr#%$)P3Tqxwtt{WYmL7}J|YoEpGPg%jly*XRs^P# z!bVu6dMu8Vek@He;-<#ls3sEsi|Di5mg%fZ*1hx*t3q(1>i9`^@Gim+Lwhir_H>GE zE!FwF{&DUJMdZj%!jAI1L6U>M+Xh1`{@%JSzYm18WoKuPudbQfS(k|o;KiJ$aU{OO z#Lf#hG-mo%T@LD9Vdv7gGZM(sgTvDrv(7LRQyY4S%m~aqRagZ`ndcV05`7N4qwKZz zIrsjJg+<)WhUc^+F?FifA7Wd#bu}SEUeoJvAz^*1qyU0t(MV zM~42E8Mx&>F3fm0JxUrfVga)UXw}k%T2@>MS&1Dg33}vD@9ga(-v~kI7Llxt(=OiM zJq*&c2^oY~;Q%`zF}`hrv9t71^v-PGt3I(qH|*}qA_a4#9;|wDu8>bTpP7pZaQvG! z{9r)W!>d86{Xyr^ddi=gl1vYJP71*a-(p58@FWKf5zIKHZAW-n-R@D-j>=y%YFX2) zf3zN|nhIa97JbTAvY24DYLrwh_!#+!X&Zgu7vjtu3&HJECJBdVK>NDjD8;ZL6YQ%Q zpq#K51WD!V3N}vTbNd5t_Nr`({_Rfb{lzZCC<#mM;;_4KH7rXJ;_niDWFjH;3m`$q z_ILB#w|y7yK??aF8@o)J?746KPTZB-|7N}`_~I$j=kp~gxh+%l)8N`{YlwY6y{HUC zH-!&0wkAr$*#L-u`sj$28hz0dC{P53!z59pfc~tv$qOX#=IVOpH@%ZonKRk=G6#A z5(gQA?-b8srBBJ?CT7e8YOuA*ztJfjQROhKH2wTN^Icr}`JQfhZG-NCc7+D>*H1D& z2c)WPkFZAFMb61_DO&FWhMp5YdzK+6rnAp*wsA@NP8N4~kbtRN*mB!8*r@=L<#9|u z%d%a|_qZL+4L^kYei&i(`!dvr_q}Xg>@7v0jXtItT-hngyw%(c(CTo3>8W$}Vqkr1 zHsef|&W3xAjX4vma-@g6Q%!hkT}si#6hEX@VX%HO?mEx8^}NT?p_y8>%Hdi-5M$dF zdK}07ApDXR45mMOKE)68b~l-!@JU*0N~b;gKOOH~p6-0Jbs16+z1E2EJe=#cR<}t5 za^d9}c0?icGEG=~6d$1NaFsfe6;UA0{j=nvdf@Vg=>qQ)$J=KltSzPX2Hx|h&Si5* zvhSUq{SV@R4`={dM-dE7ZN2 z-^%3{wp1=+s?X1KG1a?*Q*9|l^JOn%#s?O>SuSw7h(h2pQ=tQYTeNJ&ANEm|{|LAz z4t;L&#^HAZ8YP~(6JGLcX%C@S2Gur7kMS6(V39H*p0wh-YV&QCZ&}Jsxu%}y_{}9n zZNO5L2_z0MzM)+-5Ck?|+l%wGyln$f>P!q)7@R9x*TwxR*bgvnQ>j>Op(Ecuv+cN4 zcD9`0-at{o7!k{F#+pgi{x8ov!n2%#8gEYn89%CTidt_+#CWvO8B2L=nR=%|gl0a} z6^Jrk!I_!%NoBXSEDf;TNz>b}$G5c3?P0zVL~OGSaDTKw=(#wPbcI^W-?HI^0(>z% zO-9QJCbjUE8ZvJ9x%0c9E6Yq^SkoBCuj6JE-DYb@u1?_q z+|Ay&63Dq2(>c8?inO?hD`x=o|5cl$>4m3qU%@I&xynG?0wepJwAS>w6#<|kwQ`wX z-n_Pc?U-Yt`_JlcYDos}8VvbCj+n*c!k}6stcT!FSE4iPEse&k8-NX>niyCN$d1lt zxQM$^tU+WsdZ-j7C>f#P!$o*(8x0po{s+TvhIzIop<(mAglOTxUrab&3-D>r#QR$( zJJvW4CZyl}Wz1ExO59Qs!*HX9k5-meTaxNAJumz1T66W;2*heWpIFdNsI!A8*sM4- zzzK#aPAaE}6+NuK2(Ms1Bm~2QI9FJn!s=*O%F0utQ54a_6SPTk-n5Iwu^rlDLeHe+ zZ+Lur5OwmdZWnKYzR_PnLWd>>t*!DMM+1*WN3n?7dldld45#FDuenLWZW#|_Px*moxYpi55j^Zt`j@v!^?wA)c`e}8 zYlGqx_$)B(Jkv1DdXCvCgJl;7X)C`kFe(lNSY_BH@1-Dv4k;Q>WS6oCi#PvhK=Bi! zVX>KAnUOYkEC!%Ey-v0J}qw{;r@R&u(}D^2+RmVatM_dloZ z>}pu=q8PEhoXUzUfFg7XoH_OuMsW9b$9ojDYp(o){)mAT!v6xkdAmX_d<)K8m8srd z=P5A2w3{;|xg3339Hbd1NVz97;w(1{$A9PA9(wvv`m|I^%6q%vt!no+#SL%~B??O| zkB(#_sVwy41NX#}M?Puatz>Q_i{X4nsY-^ z4)ph#=%&Yf4PuDQbFd{RAQ3dBd_>KVH9*&S7XW&+ee_!ZFXoPY{e^VKNwzQ8pFpB% z-`~)9aqsJm-Y9Mo`X}|MC2+%UEvX(~g~O!PeT10vI8j+A=Jp$32aYfhHwjJd{$NQU!pUzcp+DYf#1y1htM(>x&0`T7d!GDV|Gl{vK zJ4d@q7wyS?uM9+s56DSO;JfIuqkoY3UL&n|{Ko_Q9L!i*36xpO4AM0czL)|nOZdwWhyF4lTURFPi1pa0W8ZFoYGl~Ll8 z6--?U8;z-IWrOpe!ewsUatw7>k3kvK9-cq#PFd*BAaANrpK~tNyXLK$*BJo42v7Wx~sX7X&juja{e0Cb=DW6v8|g=E<8 zfEedoRDTn0@nFZmP`2^M)l#!umh{mX{UmGtx74Urh3`u0V7<}{VRwhw9F4sX-4-*3 zWEf@a?>@p@o#yf(MqzN7AlZWwC$pAbd3iL z4H6LVX{|*JViR{}B31>T8^|JpZ;C^cd8yu%Uw}jF;VqXcM_1_gCl9VXPDtwNln?4M zCw^S(za*f^9QIRuOlRMi_Rrm+#S8AIrmIiG`1wlujB&t_9%16sCMxVUvSU08dltn{ z!JfVH-61hoOYqF`IZDUMzMY2##=oBpo^+bE(Pa_q0VI*Nl$du6E@bYWwuF5_Y-GCK z;HL*|Z*QTX=BKB_#v;RBRVshVS@5UwgB-E)U>x9N+kW`hV`S!wlg}-SnVv3*Tq#=f z)v!7qj^%_na>AZ>@FE-i>_ZUGVlK$Q?XOUp=#G&UlP@JvSq_iNa;jLhPL0>{Z=1J~ zSKG-YPdu2a1;@NOYSFOXo9ZE}ics+6H0!Ei;m=-?{k~3!bZO(6j{Y05vY8sr)(Soi z@Myviv#nXrU(pJQ?uS|wqqDihmgb&9a~kQ0vWCMV>t4#Bvt@iELhGrEdmYrb zVJcU>p03Hx-6JAtGv*eI`Z|E$Q{hOs;KgCm_)C^`HSew9}Cz6=iN-y5Yw|~Xu5>++! zUcK4fFClxD4_POF8CLE^oNqJ%%VRJa!JIgrN<#ufT&a!%{xaWety?v6te97-! zOGDGRY%7Wfdq;&IWBM0flO3E@aH$zr@-Rg(`8|mFHKz0)RE0w9SdesMO#hf~o!oD@ zF$u~2z;M&vhD(Q~TqsFITl89+&HaU7+ES7tUTUSTyWxv5tqYyUS8fE^v-?TFlu{4*UGq~9`D#PKQs23>zwN!X#eC6cZb(61bs4^PR9~2P*_6NU4kn!X`pNJY$ z!1UBD&a0u+7hpnj3Hf1@r`f1;XwJ}7+aB=72TFPUlk0I^?L#NoRr=?!yA^z<8a};b z!)HyQTnlf+x0-uU8%+N>j_9ejJ09rQ0jjKOUzlS&Gl)D@m&vFzLkQj_f4jR#%4F`13%P? z&&MjXbtc^9JK%xY`-N?8;^t{md*g#ov>A+z8;!!-oRpS}M_t|pkUn`p95!M+5-{Ie zlP!+)fToyq_?DM$bG>)lI)D4uj?j*whso#bnP4g!wZpV)V51muGO&S9#1NX;mZqya z3czO!iW^X+|3^@4oP;ZDsX{uepGhflfwXpE+>za+2dk2TL)!su<&idv!{j9L99r&5fG}(fHO}MwSCS*;^u`_X%p!0yjatJ~0hXC3u1O*>2|A+t zfk>%mZj@AG-^i0SMH$$No^XrsBg~|oc(!0AaxCWU;al6BTnc{KibXvBxhW8jpbg_F zHPpH|>71$`o*Q5%osvNHuD)z%05k)@(h~FGE@}tOK2^Ujq9}cllDBnWB(mppi`T3Wa@M7ILWo=N=kIqb z6PAWh7>%}T19dB2xwV3~^p$L3b~9h!8LTg@O*lVh6__J}cvS2?tIzj;2ee3*0~?&j zS%RDxZkfS-?@L>LH(2H%tC6P_P+sS3<^zVI^HaXUh;Mty67z$aW2vv*GrlzgmZFZs7CncKNhUVw<@gXyk{bJz=dk zW-%@n(elOJSlC;S@Bb0((VpYHTA)`A7qx%OI^%%`yRb(Lcg|w7Zli?yh>1l1kYWe# zUQFjBgI|S9Uy-Qa=g|0899Lt@I2$76J){26uv<(1pNmS(n1@2U96Wj*<@A~#7=WVO zfZN!lF6y)Sd@nC2?lNIA0;cA#OC2*8<@D?*XPfcYQ<{v=ghjKvrETNdM=HS^XKGd3 z?KGp`e>~RXo*Zz0XR#kLHw4_IzIAo@yh)}c)LG>Fp5%a;^ngM2RJVkR(`{4SeC+=m z8yzl}M1lsOo2(#`O|l0u1Sb%})li8$l{=8XK^^K=r~sY>`NW`;{W7=0zi(sIP|3|E zQz&MwH=(5WLZBhNuhqp_4F2@|$kW$+dkJ%**TVFBP{Wh`8#DH4|Er zveFC_8w-yn7{g=vZ$e~l&Nqy@H*>RP^m%~cY83tcHv%~lcqlyiyFvDF3lRYNvfdL=0i z=BBUA8$z@d)ui0~`wsOd*{2mrUz27Nve;lku}{{MdnAIn-k{gni zoT}nSFrASHCAS}9T^7VknHQ;B*70K!Mxude{hAw{)_~mE5zGkkWP4wloPn#+>YAKR|=mS<&#=JfNqju-+ul zgkWMzj8 zJznL0gw`17ay~SqAoMCH)yKlz|84(DE=c_DfvNdg3Z4i-7f`(2(gCpO_DFbhvWN;U zYB}Hw%=4jl%(^Pe3BP(ILc?E=(CXNYXXRJzsq)-^HNbEAc7X`8*(xC$o*|@Ik0{6ga<# z2#&sf`%ZBMFN^g27fU26lTrG>vB<_1?2nFqUmuZoWgm1jf@_nwFH#qv%1uW-u-lq? zk~a^fw-jAUyUW-2;(iI@noi1!0!2&JC7fnx$)9bMSR0cSYB0jb-F*o!W4y&R3qiJy z82WO4;qj20)m4nx%Gw#je2=L%uNjrXe!Qd?A1%3;$P@>s{7EW3) zQ+V0CKh7}67~8&HVZP~u2A85Xe)av64If-pB~`SrHe7EGQLIhUYKWJt;h+@Z1P4`x z-?W!>j5mZQ3~5^|PtT%+c8Gb14snp?4b|)fNvOw~a!{m<tg12h-Ig+3Zov0hkH@N5GWT-tnV%sIG-=!C6`46-jN?DL%=|XU7=*`qhHjgQp^7 zE$>~cOSe^ar5yGS!iAm@+izWvt!q|S-9+at5TWR{Tq^f=VI+B9yE|RK=H+L!wUK)> z#Msx5uBM=BX3^DCyYDGZK?y9q5yB6m0}q3)HbWZ5YrPnrnNB;DW{rc4doOu{ubu}M zog={m)it^2n~LPM@yc|`K5VERGpX1n>j%WY7hb2SC&fuf_Tp1Aw72%u(x7}`QT6Y~idMX1W;%s_m>DRgZg*6SSK|xF&>$y9p zFb%_xbw_5wN2K8&jf2+q2PF%AY~fGGhe6n%{O?>ho+2i(4qctE6fyo0&K|h<)iq_@ z!FbcQ90uu8PmJ*3+1}@79ou_f;3fQLi%iw#RHn1q$K9&jP8D-br8qmUx_ZqVEa9I z`g$#@&Pry*QWbqz((ytF1s5(^@xJiqC2yz1^4ox}g3|yA$1Eg`xz>K{qCYw%d&{bI z&maAUc|J237wunjAtFb-K>)^H~0y!&PG9Mwg zqsxjym6bTTypFloF)8i8^~mz)e*};GGk3m`t&l;wSSo+95L>z~v^;M*KH1wceL5?7 zOfw`KA|{alsIQ9=V{wD-h^n#U>B%`?|78g!FCoP4Q$ZgI?W}{h{}GTn zPg?ZkK2uEPR#VUVH(qqN`YU^n&i5RrPa=@)$FbRb1*V-YlhaDfs;>{4@`$TNm|0m$ zj{AA@yX%V1?V0fT#;%*GQlglJ>%DWHMmr;2AeQq(=S}Emd`$JxmcI@@@%5MX)f=!m ze!=f9Mtxv!vjO=UL4VH9B(G8DDkuQUB|bB0MsUtr=`y@frFbrpG3*TRZ?dFg zY&oh6&5x+%>oOO4r`tj}7vs8o=Na*(Y7BdEMBB0vysgy+4ofKq&hw+NTCE8Nca+e( zrG2i^Rqa;YiGh?LAdA<5XNUgA%(>Xl7Ui!YT^NB6<|-3sGu!(Cpz1JUnLii-@jUsy~ z@c>YyCl20ZU6Hr!e+pvnatfl!LbtOyJtR9TuN7@fTE~2-jE;aqO zq-1zL0`XE%TRJr5{)fTbp8lyzTEm}8laP47)#drTcwhdDo;SJ&l4N#;8g6BOsU&K!;9a^SGv1f= z{)ttIMOOex-0+Oxje37DtJsK(!w_zwxqd=tdt5G#Y#_55Y9_UY6?8D*bA9+)JScl< z^`@`^2J-J7=XXmDhuSx{l|-G!vEt?obXv@XH|KXuWg>4Sc)qW}YR+Q=b;P)KLEwe1 z>wnGGrWmDjr}Ff{L#kef3R}N+Il83vW4o= z$(9z;*5_2bUkAbUaPVFmso?ou2pbRODbNc;>B#t6>D z_$X`$O&SZoFyt;dmA^Dpa-4hUqlAwaILeVKd|3zsBp*eeLdE@UADSi~SxCjCWJys= zFeZoda4Dj`b4jIfXO7o1u%A{k5b-YL=u5xcrp|e?r9XOP@Cp{7FY|LTenXVPz*Ikt zibChJk~x>}i@rEX>WIW5Hgxn68dhFTR(;PL_;J06lbDYLM5U`x_IQP{cAxyaAMDX0 zb%~68wB8N*8G~4Yuq){GRl9DIiH}+|ns~>H0;MVZ!Ffae@YhD~RNmR*|8k`{q{)K% z4)8{g|BoYP{`JFUi7ms=6EPYwsJk#&uSWSEUh_4nVzb^BHp2YF;Hn8)L!E+6($9$y zttc9L`)eUHxOt+&fpy0eC2l1m*mOVZQV0Ehl7w#iV*ped-*D+KP6&*#`a&LMEx>sE z$Oq)`RARxOfpGODdEccb$PscA1c zPB!%R6+#eI8wqj;kAD#y_k@Ifc-~X3q5m^*z_MKvETGA)A-`ef1O4=_LXJLl@q^(;hhT=Mb37+?QHm)w= zqGKbi>H?<^AonlR7p`VrBET_e_=7@Q{DCW82_`@}G1j`@=_FjGlxi+6r#!uJoOf{&tJM^Y_HrtJLp73r zL80>)(w@Tf>3#Roi|46Ao3F=B9O|FtYk@nkuH8EG2i0dk8-mpDl&19ecWp)XDDvd* zFQ0SM_nhtkCnD`|V<+L|`#jq9&~t`8I~47D833ylu)1d37AD*E z7SoiJP%q_`YP^r*=?^%d{YLB%c5yq*$b>$vTuL#|{ zE)wJLua!+ujKTV0oqj4Tt6Aokli=el#kM&wYu@W)QnG4%C02bay9BJ;DTQ-q?E$?0 z&be^9KbE?8QSoNQKjOz?726ttO_Et}1WdjF3y&!m%N(~3r~CH9KjiapIu)1K>k#`< zY)ay`Y=kXt@suK@UM)j}!1&Ffl!go5J0e1@uo(Z!tz`DwsCsr7(QoO~O)E2!Y4vUT zb%m?Xr?NPqtciWd!OLlLl_eHRZGI8=dpE6%jMvoO4aFGW_FfH-PiARRLu>Inc?Z0g zhP-Yb?@#(>-bV*|3l)}7c``Ovc(h6e?hzSE=c%ri)0b|#cC^tA8z>9s`&_w=f+yb% zu0-y(etdr``^=%)y~2>->+*!#h_{7{!wPCG$;Zt!#yE@3%fm%28|c{JU}Rd!A1aCo z0FBqJR$nHRB)p5ob_?$$OlualbM-TR^&Vl)MW8Po&}u5WhCuUqQ4;OsrJOb*{Cz?k zI*`>=X}dJzRLL6P^(}obR+zu{SV{`Yt*h$;!{v%MkOe<=O>E%AN zR~ncTxw(ncHhqtyp5Q+7C^lzHtUewV$0O_P&{58)+aZnJt&fuHr#px zLcIVY(aU28GPOD~B5J+88~gro+B-(n@DlRRF|1u{p^9iX9QX!pDttUUt7PXOF!a}X zKq5)^jlW+oRrJO!&z`xxMZo!DULDn8uD;crteH|w0E=l)l>lm6K} zT{DBM3S(a*4A5N_rNDMQ>W9r+KBu1~jrzuuT6$d{TVP~MlcTpmwJ%lf^`Kx)#bJc5 zSI-1usE-<%>RVx&Bbj#68}#G&OPHtvha&q9W8zyf<*u{C4>&et=IG~(YXtuzV3Q55 zT)$h}iGFx`PLU@x+aN+(YA$Sp?8`~k(+RIO1lPN;_a=Kyc5Kgw*VyCU)f{=}z85kY zr|A?3mkik^g<0aSiX4V+ck`X&it0Zi#;@PR5hfyKx~1^BEOkHAhRrnh6@Rhuvx_p# zPC@JgnNi=Yi-nWQ6&mT862e6AMDBcerCnb#?9PwIvfKl1JUEa4gYp#+PbB3uncZje@)^kcxE1JkZPr)Yp?9A$X`h8CC=70`crc7U5SLgBMzv8{%#WHy?NDa|Vr zH~KAD^byU6*Dlbr$V%Dz8Te6bt<&J9-P+~5JUd0y52Nc9)Gj2L* zP+?~AH(a|_rLOxkE&lMEL&8BtW4GMJ_HI#8sVc?ViN{|!1BlLqa!Py)?vYM=mOfwR z|3UsOA6ptRK>Jn6YLK_tfLAZWUwY((eD%?)o){EoC};=Xl>FL+^3 zw;H>?!d`+!*VN_xOfokWT1U9u^pmXPUGU~95Z@s6QKwFPtSW)Don&4aYm*lFtv1Tj zD-o+@+tfnqOth+t&k&YXU3Y7p;ub2HizxM#@5Ire%$E-c?!~1a}M=3S8x2T!1k@A}9y zM0-9B<(?s8L5Vij=~34ajOsNK*npfrQbojKp0Odbxf#=pr>!rqdc-Rc_XCxtH1_D+WbrOcaVE z!Mdt$A((a_1_r;}9HIgigH;cJ_v#C=3T;--JJ_GaSLQ6F-iYD;Z>-Q*2gl{|!Vujv zI?xV#hv#C#+b;^Td=F4zz(nfJh%7GAx2L5iJ;s)~lZ0?$hhML}iABFme7yZbogzLu z8m265&KbS>-2a0FjAXLIV>I$l`>+;ZlIw{?4pArR6e|AA%Ad3Y~N0vLWCQ?bx~EtkxYoZyxApK#*wpDJDT zn+?giEHy5ps#PDd<)H2rz7Gt?yXVM^`^Jd(6glhk|AxlFGoJ%Ay*cl)hvneb@ipwO!v6Hrg?0L}@6b&5<&09qADso@r9Hr)#)Pu<%Q@ZVsTmYnAf*kfVNGwIVP+BDrz z{jXLZ_2`VyyW{c&m^yv|I+ZY%c+X!!o z7YGR02^CHw{pJLTK-)SM2#B%GXE~$VI#uquG1lwx>VQT!2#pX4g?uFuUqW6w_S%}` zk%Ay{*ns=Zgn5=s7w+ERkRU)KzdjvC1&)_FHg`%suTp9X*Pblq;&xudbtX4m>Hc}U@Zf^jwscPDpSmIg17t#3XXBT z4$JZZmZ{kKyecZ4#>Q&lTat3W8V%_)<`*;1Ntvf0)9ShJs(+wyM%GjL(OG_ZG}+eQ z=TD^`zQLnsX8IMaM)z>;#5a|LTQ6kLRnJs>gsCDPHc1y3Pkd>Kyf!iE@ysyUm_sAp z@2#LQx(5L@3cJU@^CM;VCo^tg9uB3Bba2H&#Wn?Z zK#w1;p`U;HDtgmYK@7N7E4w(~go?An3wx=B=EFU80#Ne2!Znq%TsWZt1&5K;zL{Hf zupymNK%5S%LtT9QQMHc5q>fsXv!B>9+`QS&MxHDVoniUj%`{Gmu+#TT09V4kc91IK z7ek$C{WN9iK61QrVLNf20&~A0nptg3K|*(|q`o2Sc&7D+h=Hl(@PLQR2?qpzo zcepaAlIgn;b()V1tw}A^Jt)jRXmftFdDi*r=^1Q7dnn@fw*0Jjnxa|LnAbBig}Nu$TkK?N-hnAViJL=KlU%^EvXb_y6_Q;6)}3o=ZEw*U!1d9bMl^5;blm65hMI?tC&P!0bU%YM$acUozjhIH z7#8e)GOLJPc0d5=gZ>z==i|f$_0W9-pUwe4i8vpjjnsyhCVR>vLW9CSgu|ezRxwzH zqjk+Bvtlz>^hQ|)XV;+oe5V!$IS*ij=RNHBd5TV5mlxVyrDC`&3LI#kq+T=EFwt>k z#UB3%G)f!wpAd&sS5M4e@9VyIknu?9w!RJ6dAXyVc=q!Iw?9N;ttdJhpy7(v$;~1C z8WpXN-rYVmQ}xU^dzSHBcrD25o=n%>`UQ;UYVVtpRSa#+>V_y4&C&FYA*qes(Sxjv z48c`NeA@-bw}^$D`PYXhU?l@&Lt^m|U6?)ko$k;wO=Y zfl5O97Mb3+3AfOnp4XOrs#|+{PPB$>#VW`7LTIDLw}#&goBd7#k1CxT0jc_ZMGdB< zWf^Xm)?Gz9%?&fo7{)?EJXta4^{%|zsD434Uoo?c48@Wb5vNC6?B%46!X=9+)sVds zJC|_+d@M4<6b8Nwhy`b5e%xqNdRwf(>VKl?&DuPc@I*2bXsJ4zq)k2c@=xh3v6Le5 zW0kT_yYaAYLvV@oW7v}y%cbf?PfH`xFXV~Fo3Q7rcs0MT?s!ww$wgcT#GyZ8^p-*F z2(eL*ABJ2Dh;vAj2K61`AB6vZ7)Fn;eG1i{WfrNM~SJF;{N#T!*Rw z)oU3}b-i@KtN@_VBdfGGCHL@bPOSQ)=i|XW5-zIV^D>s52o{4I99g=5IV{Jqwj0x(F!ODwkg6#3??L zOVv(`P|8*OfS*JmFVO-t<)G;UZnH$?X@P`^%>7rN-uoH6#hb-{D|Yb7k-9jT|Apss zlyS#`v>Sf@6^50V6dASzjWe2mD*)Hr1*t9r=Jtyo%6o0x6`B_}zrxHuSsKQ>(iDkG zr6(znVb(D?loJ_E)689$URf=b?1yM&>#@Do3BnuMG{VMTK@k-uKY$S(+P7=c@kkGb}O8GgfCV8#{j*8|eJ((MtOQ6pey0bEg(N8`_<@OvK1QsJ$004ee<9wC*^?{ zxrj$Wrv3Ieu}WkqP%9s_ZC8t}=ZG^J7Tvg>;$3dzS^1Q1vVw;Ta~KwdOzmiP1SRbr zaVVz+Z%;aOX$u(DXF%U&j6dEqa zStsuHmFf*WqVGajM3S<6uw+iE;^@gP+nY>bDgyA7TIE*I%(V+fP=H1&ZNn-jXyh2K zN)S)zxUKVH^3TSONg(AUTrjG#RujgPBeq5)dOYP1~3)yxo*e`D1R1pTL^ddrVCm z%E#1+v+KtgdAt=*nP~prtfKl47B5?AJcb_W&Q#|LdfHRk+ma^bf%i7H-e`{F(aH7y z_=8h)Z-)$<&^)O}%=IaSiL3(&I|uqulLXyoPY}XwYqhLLa^_fvN}y-D{u=*i8~_H{ zDt2dSG?l}Lrd2ASG6#`>&ynXs;h%qbA?GjXCkK5aC1N<`t(g`0_#SU6A#}9c_T?^e za@U^isYKqTzN+U((&wh+FPVAoXfz#@Y`w=G@B|*<6{TDp&bCzjL>&+(1SYs<|ZlNwidK$9zk<8?dOGZipNT>Y+=Gy-$B2j5;hu07#(dUq9% zUxGH417>QXvo164S*G5P3*C0pOqq)`aO)}QwN1*XoIPZIvwsuqPf7SK zKlGOHlL|E$|2-iz_ep`A=O&hd2)tAt6}m^Pv589b1~9e9^^Q4cUs3+1pqmc(j{w|O z;%lOF=Sc?|!9U*8Iw_QecCh`6Z$|pT+(Paf#3= z76 z9dFiSzf_}vCrh8Nr!-1Tqho#HFMC%h2z_MD9;S?$-Um$;vamD#>sIJhDlj&E^ehBZ zr#ye}eq*`B9z-=a1!Zg=@$ms#A1DOnOi|l3p9#@Xq6ZXxrghWlPG@R!SLN9EV*P8q z{3E?wHG_^{-IEDn`=;nr$T3)j2qM|`ys?yax%p*N7K;ox1D(k{0`}3KhjunGo|mEE z4z0dh1{Ci86Mgnde=BZ6`y}-*-cV>-9;1GJAUU-JrA~-(Ury-r;Xw4f1fCp@MrojFnRl{PWyt$U9tO#X)sp%H`)s9 zk`L6mq9;3eV}YSk9jfmc3wdFRhWBALdMx6Z?;n2sTP#BHfm?CT<8$vPCBcmKZ17p} zMcKV+7@YuiIGdzU679PI^P;C}n7^r>Z?;#;1PX9UC`@nF%}GAg{mxZUg*q;bl}<6# zL=2Bb?o@CS$YR>W1`}vG!~|%ZB%gn4y&B!6T`6H(`0YwOWNbk8z)$@Q^9C;7w5V^K^ zM$!J32*Eqkqt#&R&O~^_eQ!zegATD2I0ZKH7OV?+A)&puO8e_?YLDC@BF;V0q|Z!I z$xG_-DZJjkam$!8b+O#d1XUjJZ;c*F1ly`+zcE#7ezX^>cTK(IRmfb~=)K4b0mopY z>^wae2&2Uf$>6{!Cs$hG9pIzYAnmO)cp0yEUzG9ey5P-2TalN-(}o8Ox?4K%HV@kX z;-RQ?%JG(T+MtL!iu# z<$ehonj&%2!053$(bX`vk`wGU1W1x=c=kO{1kPnXKK+~E7F}j_f zv8Gu&u9YCRe*|<~MbxSZ1CuR5r#2uzT~Fe_^vOjPQ9l#&1Wa;mp&8iK7|k&(g@xAG zzqoZ}KJn}+pHh3bF6p%{#o4lon2*nlB76D3dSW~!j!xXs*r<~d!LMVjP1a&MZROC0 zmYSpX)VFO4QJ;Ug1_U4xemz6oDjy_a!~T;VD-@_zN^On91DHo)lx~WtfR;w;z*C6$onCbHFM6=qK<6bii6oo>SfkF zKcG@e**D1Qeo1%(5j`tUje`o(31ojcRqm|1)S?}t8`|A_ZettoB{cP^{r{2m-cd>T{olW4 zYMCQba*y1*9HcmN?}Z|Y=qI<}C`C;vOD!|W+**=baD$oR)*QKWA!4~QXYSlt*7f7M zuJ8AE-{;(a@*jtD-Y;I``FuPDe?-@YE1PWU{^o3&&IzDFaf68rhGjRi+utMyjP4!3 z+_7*<-~2XLyX$Y4VkFTF$l3ZU`E%ecj7>>9t^du`Dy-(kv7)D!#%}AqRW;&xp#VsQb2fL^jLq z808JK{_%asdpF5*S8+VVXC)WCqRi9h;5gB2?O#rS`PtE*-|G{0c+$_yM{Vyg1|vE+ zC^s?nMIK3~UKwXj*Fy8tZkwAr-8JH%C6(Zl4pJYz=b>PB10RWdcnF_3h{sMf7jH7^ zACyBF3IPO7?be(jc)wcq(v7Es-2J*GA{Ol$zvm#A<6R{ZM>R$qQV%hW((@zsIqhtI zVyWHt{pb7fYWbRz=1X-7@ZX1Ku($a~IY4!ph$j|Td~wyD2L)`t zVUE3kl0|D0%krw^?ckn5ns@c175gQkfKB@fsy961!QJvfB8;+AsfDokJ`JYBYXC0{ zToN27U$qV`-weAf(+ly+a4?rz1-c}T39$uT6W}-D0I2XA(bqau!yN*hg!mbpAQ2~} zeN7cFs&T_+ad$3NZ{_X~KunCj@FYwVi;2+wnRV#&dIn}=&(uBO%?EMRW8nCS7!!@$ z*hLG1=-vDC^uNInBFv!u_6^>D(WTM`&F=%>Ypwv!4Le?UM?!?kh5UEi%0$6nvfZr< zIQRM~fSl6fOfrxCP_tCEsfhT7O|r0RFB~Hni?!x2mU!WX&GQ8Hqw7;clA@8_`f223 z@TAO{L|ECS+L?Oe=6>!EzPg#A@t+!OG#`^VUv|ht2>TRAHg*o1VBe7Ix( z=EuXSlav%#VuntnnzJ#p;WQqaW~0a7u>r|JQ{Ks{EGI`-UblC`@g7%hGof7Gi3YcC zG?nE#Am~=lrPX(WlT|Mo72(Cfd4Wn;J)LE zXH!6PmbUPR&oyaXW&92OYg0#=HYs}|3TfqAgt6pkOflp98uh2s%w4`6MYR{&fCI6XI zG@B!p-ucaszkUR$bfE_%%l@49~Vhtp1VaWb@iXz|=bqtQmOb z)P~uEbWl_kS7+aCJ^d){Z|uE}9R#hG-pj3hxm8nN_(NIMH)6gU8Aw1nzz6 z00%|+_|e#1F<`QoL!hjGuMD{V0`^AY3~|O;XP6=7HMROPLQzRrKc27>@8S017HP^R z{y1$6OsG%1c)fPEkfn{>N}vAFVI@t>V04b9hjLRQD+HsXd!G-%zEwD_(m< z_Mxl?x&1IU$C50{7}Z{BZr>O|YdY}5iNV4v5z|sC^c!w$@t@C3Y~-_eMW?p3o^mA} zjYN2Cg=(p0w@WJ8f4}}f;ws9`BoA#yhh3~zGQ1mSMob@#E)<-`t>8jsL;W9Ibvs(e z)O)kHK~;mI#(nFHo7Zc$oGo(dr9%&!!BMzZJ)GDSbQ2C<<%o0jnc@YFxV>1R$CPl6 zszSd3GKUjAzoY1y#LZYLS&p5Ac7U8y0a`3ntB zF@`N7Hx_I2vpD&P70rrmXYu3PfQa$+A`PDMEs-#i!IrG`bcT8oWDzrrNsRtzH z2KlH$2}h1#`i*1!5I8O7WNwAtUf_|_^6ZMpk8;@8&yOMOjEB5`ZPoJ4=O}~w<<0xv4_M@v3RI1jy;^A z%++!rTZO`t0KMO@;W6RSaM_O{T|0q zKyf-HQSfm+%mAvh#X5^}V=sJ*J4Y6z%Tr6qiT%I`@3+XLS;2hyfPujct6Qu`d$5X} z13uZy9Aktq{-`q(Q=sMcVp>121R=~DEhjBsTqGdi^s}>SKi1`BITpG^jA~M=CyK##x zgrAb5IA*bj7sx%xYS2FX$PhjV-`{X9G-bK&6{82EP}gU*giD*7R}k#x=kn!2Wm16+lfJnL;FUD8jfIdVusE5Go<7j z)HD<6Xr5hOCe~-}&&WUik;`-(uw8tD{ZwvU`NB&ttzkMw2yf+UsIqI(Fl+LZQdxO_ zAd889#u-#mr)OxTb`a=*b>-`3OKxp@m>`j_Bd0OoF>zeKs20H`mP$R*nvL$#AC}9v`-OQ_RoG0_zZW?z zyQ+{6|JrTdsoPI}!+$hnzhG6-CH%aRBqTiTlCbm3#j(9_LsU?qW8-`D2f5KnrQxA# zkUs~Hnu9rnUR_6g$-)ycs;-(si^Aw+N@^cN9EXU-u2c8SuW{Zxfqz!UM1MWxUk^TN zs4s}Ki1@c2RMY9fwXD8wI{Oh#tB3ox7;Txc!(yL)*j@WKX)cEh*zq^_Lwfxzsqfz` z8CeULreg-i*=`);H_T@;WMmonic1$|Rvapn$(fT;#N1Z_0wq`2 zm$%$}3nHuABaetu!fmGnOReXL&)H4!5%v|s%_CqI(}(x;?nmyxQmxQy&8}MShS(n$ zH{31I(S+Vp@s$NOPZ%m0yT7PHJ4tW_({Vw+W$@AhaJb9{J|}{BtA1d{8d9&yTWEWG zHv8VYMhO=!=d5**_ig@CDfTABMLn7gIXq6a{-QxZq>1NI9@8dAQ#%dC9XF@K$EWyE zbZp-)dfN5#s8*gxS4M&0KWLhG>_7F%_NfT1TIC*3{jWsxh01%fLmL6r_OhG*#GCRGcv;b+@yf+(r9I90 zo+qVhq9=et&-Jop>&UC;XHtD_c8GYu587!0;kW)CS_=jLRQ&LWg zsdyxRdS53X8Mdu0I{-Q+=BoD@+dexE;(q*{e@l_WV&y1V%-#)BvPLHI;yn zuQ-M+5^lns)YZh+0TH@c7Q$2E^4X*)FCXGOO@4&)Z9$kifK)H+)XAPyo^jVT_((4l&yS!H{wVc|TVhye-{M2rr1O32 z8}OX_uzJM|srSV}W3Jb8V;z(~PCUCWx8hl4paBTs!7RT88db_#%W!kLTsJ|Jxh#x< z6|t;@m;ja4UsGo(Oz(r=t^y})q#S%rieypLb#Q2b&d74JTe)!NPW&4>6>VNb zmOLP1QAV(9YT)eQ!2OOg5(+)q!WDMAKmXa3wckJ48%ZHT80 z|3&9Qiy#PrO=G=bEAr8$Lmz@fdXg?D6deUV$v+-ecqMRw6w!CQU3R#0&{?oxxV0-g zA9y0%+K+--QhdNJ43ZSTz%)I9_LQPi!k?2C9OFrR=NTp0k*d@f?)5Mrpwdw8<%uVb zome9B2z@e!&Yun3kHT13CaLR1I;Gw7Wgi1eHg=ua+^L*a&Kd&{Yj*@6oK!Z3FtI)y zAKyhC#xeR1*#v-hLa;-bi__X8VkQVxeKr1%<1oV!p^El)Syq_Lj;X~Rjj)mqMB$+Y zo0^n%xNw$2*Q1Q zSjL(%y|7J2)^9Xp1^>vWQ66s{Zn?P^&Rx@y&!BJ&t+hzjUO|=oR2+L%WBbSOe7;W;`5TJ0Vb?91u8nxhLVP&z`mBH^p&eo8?bApv}YfF zRwZlHG~ldqiSW=p*Y|yiyD`EwjCaeLO~<+_WKxh?%@FHydsJhKOezFjs5goKnLsA? zwQGz(5=ibZsj(SiQ_tLGU$n8JDVwTbH8mo$1%gJ}qVH9XXOPd*i#Ksd@~qA(|MKLI zBP(%`iiIX;dHJaNh{|U!Do>i%kNGrLr*LGeZ2=$ ztfnrPjGF(E2S3)oP`9=hJ7Rgq+a@Nwy7if~jyy7atJh}~1WZaZQ5cT1)rYiBoxRqB z;T+dN)xygLH8!NrKcPiCqw>k{3-a%OykLE@2$^{P_XbdxYbvaRGg{dVm~&gzEXiJHVN4y1|tWQhChrUq@wq}IGxH(2rqKmPlE`0`!NrxZ(n1P8dl9$ zSP!;0yNi?t-YA#U7;!Wjs_S6%_~!!?X;%QVe$2Ef-x8<%R(S;!iWG|)(ayIv7};=J z*SfpDkdxoEhIL-&&RIU2a)v6yw*& zk8i(Nlo8975vyzVmT(NzsK6FI&hodyD=$)YCbnRw%yIJ9cUq?1#4o?$e<_5E=nX;l z>|ksj3Lj4+1Mhz(io4zy3&{;}+cOiK9avK8@bz+}UY3rot&sTbRPg1~SnQ-ufD?5= zEBwsSkJ=QabbPy=Bt{PNFBCI>zJ5e}t~|g{xC0TBU|k8@w6^;|6>kKm34SmL@Xj;~ zdaDkr*L>t%%h<0Ee~zynd(@A*+8zTmc$g!dT2~7yr$vxwnVj(Up^{0QpfaF|wdX7l zBUAuO<)pO*GDxU^(U%fXc^y_25*lcA2_R=yc!(*>FwdO?zYB$r=R0F>(a|U1oy_ld zWv2EtmEW=!y~v+qURz$`eldBYr~HbHcKP?I2^Z1llHuVxZ@jXIzU13-LYkP_hS%EH z3?G~l7HL19T`+uQeWq&Oi?5%S9!e2zx$pW5@U= z+M@FvZqJk(-1PAXKSts$j*ZctOdzNdG#rd-|1hJmsv~#Uk^aA^wW?_mT)CmhBMt2- zUi~p-E055tdtBG%yk^C4g2^0P);8UE;oLP6p8}Vfb?ReMs}Gvj3phrP&E|8L8bg*3 zezb<~h&GQZ&on3o|1hzIEfbVZW!5uXufmVIh%YO|g>pXYho`DiYfN=_OPd2+J3bcJ z1h_9m_Yiz!_)Scb2wBy`#U?0nb$#2D><>f4_zr_Y@ES4lTpJ`=M~lcv?9 zPn}PJDA+}HO})4ev?hk8lBR8wkG#ViI#?=%)!L>1J`(v45&l0-%~1zeQo5)q8n0$5 z$WE#gd(aOwaqcs(i9Gn^;<@FrTPTA5lBc2V6gc>F!)3tea@Q6#=%?S?qXjDOFh6JQ{{ac39NXAba? zOS(5XIa6KrgAVUe+!rm&uI2E618vZq`mtY)tZD(nI+J6WQncBp z1UUE1$Tax79@2@Qe?2a$GJP2!|4z@MLjHcVtP183wU<0);CTx_b@rv$!Sb{(Cn+g`n1Vo6^TB-W2AlQgHAldT;ep)tLg&ueIrTjJdrf2TqU zr+vQZ|1LQuy!-l5>d6%yJ)3)30p+#d%FVvs z`ZYD{)k$>s4F@1J^9OQC z$`78)0AIlU2`FsU=3fKytn>Av)i|GhF~Y_-ajzj0--@+@L9=`XqnYYLvd<0+VD=Qq zCCYn2mU+^5dG_%z!mN8-b0xNu{1Q{2<)oB_kbmwwkO}73axkwR;+q*NO* z_0MD^a_OlR30IOR>QLt4&$|0W@9kF?emnYovcKQ1>Y_53UBS|bdTRHcmJ8;+tEQ^P zf1d$=ktI7J^mh3^IR7Eo3zLnpO7SR^tuUC3*)x?ndl7+kE5&aZYjJ%txr$8-8eGe7 z6Q0u05QOx}R?v|v^sJS+r%irP=3wgp?o*WV=e`T(^}@^G%2@oZ?u)>( zb&--=4*jssA)kRrA>_H<$`o%?KL9PB0{fnr7%U!}FXtojN?&B22It7JMN+&Pmea^d zsDAE}bjFKbxgtX`BHRRlQ0t%>w`|^#yT8))G}HND>KDT*lbX#lgCnk5_vsT}TIOr6 zOy5963J4aAHU#ZzFaychbd@i83~sE#Pvi)|zHc#+rntsOC}rz`NQkl-PGn=GAZ5l?&Ekw^BJuk!eM>T*;bKKi?byvbfl3PV2tK2bFv!0!M?r z&d^>zi8obOq>)T+hml*>D*G6s_j<<68}c&#Mz&RxaDC5B;VG-35NnmY$4hxJO+I7o zCO%{CB)gF@m`8!g-xeqR(1{THwx=RPReV1eDM3If$RO56MGp5q{J01=P)Rlm`HIQr z;S)dA!`tVy{DnxozgqyG0(q|kSP`jK{mdHG8~+gHK8j8sS5~cS#dmQ zh^$;WoT>G>Jkde=9O?Sv?3ifS!7g9(qmlFQ&*m)Kxq;u0x=FQa#((hP4X=K;O-tkc zOpS!wU#(@?CNvMzuk2>*6SUm3GU(@T91&jnEJ78nUsYi34xvi`3rgQdL8dzz`{`(Uw34XZ4yM&j#Y`Q~2>=xFE{y$bRfL z3Y)PG?Al^$iSEdnK`NLm!*y|}`zUgY+y)nK&X*bW@bVdzl{%DDI)nT(_yKaK_f6bf z$N*r(Ga3C)R=ZrQaP;PUcbpk$%Z0qk!GP%t?!%~yv|>FxtR&-ni&d^gRnKjMV|5Xo zQ}@)8i}9vx3-R(#UmEogZX)$9P%v?2Eer?ja(}S@-W_)P1-zksMOkbwSTcn{8q;{; z+iM1Fb&Q`F(KMH#*FPzzuFOQT+3sxMZU8Y^w2*OGg7onuq zEGy=IPc0y=T)#7rJ(MkgAvT6dYf5V=ZowdAQ@4JhS8?>hIlVZY7FGKLnzT2dq7KLQ zzb4AZ`GLiHg}!r(z&;k$3)im+W1iB26;AE4IO7zGr!N1%kg@ueW zw?fQRG|D)8)z(3Um&M+E{PIWMbLKQOIm9{oC*GW2#Y~#7GO60~)o9s{3*U%}Wim*Q z;7M4%>rEytn8G?jqplpl!&jVwKF!9iID+BVIPUdTw!Qtpl4#Y`*N=b12igz+`u{Ip znRS(D2bvmMNbf}2guBhnp9$e2&$@G(Mgugz)nOY|!*^V!*+I;c$pQ1)J-0U!M7;nR z_j@#$>WRGn{sUbvuo&`ccFqrRFr+v#6U_O#s=N5#z3WqjU2Pn?1W3EGR?{oZOs-Hh z3N#k&+<}T`2%=m#%~W)BkTX_hT~$pZkPO`{m0_1eMT_@o)R##eJ{Kd-H&t)zX4#D- zlLRbt_l-O2CsA=oRXtB@2SmTY6U{`T3bPtf%`K{S&z>)Z07m@OKHRfc`%KF&f@93u zCYN&$tpf5p3r5S;KE~Q>|I|T~*RJr4^b4UMQf`98LBAkOP1_G06;ZKN{2+U8v!uk7 zxsNc3VeYsK;IgaJYnXn89c#5#VRlxgeJ-43zcS(>%P%3`2zN9(+R)b?2LbYS?HD$4 zr@g+V96xy#vW4A#q@0e4>I*1g+fQ@&=2yBQ5*~n4e3772l}BcR1Rz~*m+DkClvc06 zu5UL>&rx^YtvInIcEY{~MFIL<4U6;plEh=7S3@8kycY=gs7XHyN$&c%?Dq;VLt`~8 zT>usjTAX?fwD{HpB&ZqsD*pcFid`-Xnp7s1Kr$}PO-qb4Cjz(ZV1guOF6(m*{1`X? zcNKwTq(9G^pre;?miCnl^Ee=Z+2Tv1kh~nNKX_pxpIvo3F+%fQo?J^9K=X(Q;x9Hp z;Px$PG4~cS0sI8n#i{4o5YQL^$El(kJjT>KqW|MEVXLQW1HMrF`Qo8_JU7ZZcN9m{ zXEn8!vfZcIEjCH!bZleDrS%Ne4^(2^cdG~mgWm$6vsOsr3PM~58K*o@aeB2qvKmeQ zHaz*9Wyojl?^GE&)Qkb}R+dFMKLA@oSrFhf3ypZ%Yr=c8xrAh+fRp_oROh_85dYvP9ccEIyXdar=nt51E8dBf+}^%wuh9=x-vs9$J#v}=z^yF7XS{Vaxm5i2C ztx@d|6Eg!)EIEq1Tn91V-d?4DNn^u?n;vU-E3!|9zD!3)Mnf#$)YbxVw3N!`5reo~ zmi5cQR+Q^bA5HA~Ya>UuI6pMN&*6 zNA$oPm?hYBQpGu#aoRdn!HukL7D5_l{uQdaWkIWHL>$*X^<`}I+Q4Z!zt`N9 zJLJ3%f6$=kX+HR_Eqk>&VPuu{-cLt14bWL7XxLs!?cCA&KHZjM{$rg#n>@Mbza)0! z^4G$P4+hF<5n{*Y>tp-jYlVn2+J2lG?6rEceas$9JO!A2qFIiJ#Y)z`Yrt#EbNNx_ z{NVY`3tx^)bf1^xCHD+Fpsb~mvlQn%CwgyiKjvl-Y$lF%f8Fe-HgyKP|IsL;w?NFj zNn4V}=9l1D#d1%ormuM~3YV5|(Bz*aU2}D~OpS@ffrhL4b_?daq9)6F*dg}&4)cX3 z`Mj_Zvt|Io8uAkF0y|s@D~6NTax7hvA`nUbQYF5nsjLyzzh8F(k^fF zK#)PU9bc8;e;28g~p$~7!wF#J(oHr#^nAaKjf*ej8b;^z0g8d~T2${*{otnYED{Vz~9t^qB|A(2QDn?lGtZNea2KlKeOgZMkV90#b( zi~FhQI77q_nc2?{FQh-*vk=hOR5kPBK>F&Oom6Pi&rR~uNI?YiB#*t`TTlD-2xg|V z|Gc-9HQc5BIDege8=c=vNrK5T3=$w9}&mTKDCR#owqd?Pqgg$hTI1ux_@N zlgqyzxM7y(3h)F)u2nFVl`N8*su8Br=y&fA``wQ--{+GwN5k3z@g7up2=yI%_nW zXRRh)7FJz>ECn>6cyu8LndEX940sXzD(MTgoZ#hdi@gfxkAjy=wqC zymbv=vesH|vfUTBetadBe7pzA)_1H?`VgHod2N~Yohks{_lIEEIE%>CAR5~j{t)|Q z#)KxqmPKLY{U)&MSH*s{P&su>}h>X`aIXFJ`?Og0Xqu8n@{ z@#8Yl_0jp$i`tQceY%+Q`C&5UWfVU=s(g9Bx{#u#_V{o4t{B8;q^3;M9hYLQ z7H_B!J1LmVdY-=-fbO@ySOImp-cq7=ax28=%dtdLTRLAO_Z_+u0AHPvhc?%9_pN;6 zJJcj#9Qo?CRh4H)V|eS35^J8(90rPmmDPu!N^K*PkB2m2gTl^D-+-L)+Y4MeDTJuv z#HD)+ehls5Ey5L&I63b9h0@~CmuBT^*yfD~nhH>vM})F*sO*z(EYt(;Ma1X;V+jx^)wrcj9@G8;IY)Y6|GT|(0~&1jo-kA zHFsyIw)hBm9sXw~;wHSW6e8bw#C?HZr&>kkYn*fbL&t{)bF_2P+~Acsrpsr>+6St$ zc(K)%znNY%Qeg64l$LH#LdAuw8{p~`Uo~idyojZ3}Oqc~H zE!NvLevw&Ycg&cuh*88xc2#QXZ!mAS>>J^;3gxF=+>_&E-kzAqKmH=aJN_z9cpqW1B>cREEVH<9iwH32GcX%qzNEix6@r*TuMeL_bNqyr|(>9 z5n|dlx?Q^&kh@NDj&pjIiJ@lBNrqWvO{G}6&X3c#l7bM|zB&H-HJ$f))3~98(=Gos z>Y#iKu;qZoYD5<(l@)JHhPF+c2T5OO%DerYu5H>95qe2*8t9JSFu@$?E`dvDbJ%xUk&$ zt6o1i{|pn#*ZNyGc9q1Hi`l3FF*U!s^Jnu&b{jQ}DrmYmz$B-9HF!$vP+EHlZ`LoE zA0XXK&W>kFGNt&~Za~)DabbVt?fS}S657lkydB=F9FWSy%!&MM2VRX|cJJ?9R2Of; zXVF7+91kv)XMP*ibShuYu{jx(Oj263L>&>)4<9s!Cb8?EZ&e*Mq^~~p{tL0RZ?wcQ zLIW^E1Yw`ZT}{ZD8@~2aJ_o46qb+~xsqu}AVZG+S+=*kbF;q4d&Ss&oV{PeGIoal1 z*QrR745}KkXm^~p^;>!vxn!%k1o8}4&40pw(5b?vby3<~-q}+U;A>P-x?8V9y!j3p zlEF~o9U%_9^ojuQgC_U2)mgMP)n9Oy_6|H6156?EwOE&g46Y4nTvNPf%gNKMf}0@! zsYpNh8gTVd*ZJ1?O=79_jRVxdagAQ`vL(gQJSg;Szv%TYB8_JBpm>L!aBz2V+OTdl zV$<_2d|-ctViB3_>#1HbrxJks=bGU8CVS+{R6xo_T|$Mx%ZJb6Y#$%T625$mjMy4z zzq62>5GUV(fsgGkfa>1+bSzdp7*Wl+WEqhrnT*A-4vCqvhncNcrQH z9cDZ5NrHgeMUUHO-megzT=qE0gYeRt+th z>38*mik0nQED*sr8&aa)i@qf;pdE7&J?6 z1;av)x8hGq^Z^Ro$bfyZlAG2`r=x{}RxdLvJh2#KgKp;rE%Q|x%}v^A`eCx}h%-+T z^!cPG@irX8{f15#Rr`p|49vBa?8>>+QaYNnJ?!&>tvk}h;69t+;;)C}A?DLbU|z1Y zJ~p;w3QhE#8H3m&rzWu6bO)%)8?tXDOJH=g6uY#;I&m-IO;fhM!g2q$HIaX%?<0!bNOX(;A_eL z#i@*nk~i^mwByz}A|?VqiqJ|yyUQ@~j)=r7s{bf7NWlv|#K;sfa@)_O$8f;(SXBE^ z=CqQUuB%bLzl|8SzBa5}>vR!4?N)+RfanJshg2e0^CtCWMEBjo{DyWtbi#BV*nD6J z=5KkZbsjqsWs*k_y^_7w;%beEc)IqOG>@CWwCD)S|9J8O@_CP$5OoeC)41&%eMG4i zCcUU%)X<{nSKMFb2{g}=iaUorTyCrIl&Q@6KF89Z0QH2MGeRYEaT-9AhiwebmIP)I zbk!VEf+q6kEZmj0BSlBo3-Yf5uZ-f-u(c1+HlYFe9WBhd9mmtWp(R*Ut1{5Fx)mxd zRcKVk)Bt>$MZZ5yZk7T`HP-@j-O>n`D&(;fn-WcYw1$ej#ljx&(zb&wtbwEDma2(x zRR$%=G0Lmx%f_kb4v1LXx5!ok>~c0wx|_@3W}LfDw@ci0{T^ykH1lPS;~)UW7ay7p zWuQ2HRHwTM90uo?xq;>ul98F#d(OjQ%zOaRR$V^Q_ zfk(ZAQm55hi3c2FIg(=^wp{AwS&s=ftVSKK?LJ9+6*`mEM?b2Hoos6v5Npn}i0+CH zm-P1I`Shuoq*EtI@3g&VRK`rWYnDXLCq;Ctn_OH6B#!6Q-G4>l+KxsRbT@8a z15JBBts&grGj?L-Ue1k^E%UXSni#e96MvUY|GE#RJ3@nOR5rQ9h7E{6H=ME#@U%p< zHPbWGG`SD3Px8O|H{S5;FXPy86;Jjb443dCm-LV1-hHB3A2ZS%5tzd+bZNzbpS4zS z9a!uqu>?)O)w$nKuD!Q?BBE?M{1q_qY_^BX-q!zpUnl)QIlm_z$~wxyUQA z+$8kW+ZX(@E|>b`csqcn_$_4Zy^+?&Z z>)!j&3)ee!bG8v>wgH|zjHVEeiTI%H3~yAOgjQeASmLty&q)mBJr#K6#+v>kvI+Bz z%I>~Sk8#ZcYTdD=SxJM6L92jJ!xFelMc;!#+dBKW@j;-3|~p1>U+-LS5Jd2 zpA^r{^s*N6m}?9;bRCuU_-U)iStP}`__YVmF1XFY-ch`SVd-Xq@BeZm)gLa`H#rC> zsnB)my>cuP8Z6UH#7jUh;{3SdIBsf^I^; zIETAVW!&0Jt0K+6023P%RIk$c{vSkHvK~P!jgMi=dSiSt@nkKn*32s;V-81m5qm%;GiN z%K3>u{QjB>)c%XkhT~s!CE~@e7so|@yqTvxo2UP^7u}@Ui{7VnN7m^Wjiq4_NDtr! zNKj5JyOc{GH{v^Tjo+4FGIgVTaIG%8w^-HO;&Htb>(ph-Ez?6>g$zuFJ8Z|y!vamV zUx;BxAe1e?a?^py-n1?^hO_r(puPDWmQsw(5q&RR@20;PMsxf|w5&{tR1CZSmi}=G zNtAUdw=~B_sG`ujI9K-Rk1__=&TgX(aoNSC>i&l+!h%z`G)BydTg1L(s4Y(gLw^od zT_()4Ck4nFTkrtugS}Tq!uJ^m{)!>Hf2J=An^)gT;BMtuh9OPs{VbU=^R^?{jt&`f zy1CCp1=D^*`<7JML$&9n5gA+dwBuDZ^M19Ep$lZdX^i~NTY{YgHb4zlJRz1xssZtI!t~fD}($Xqm?J+;HM=hlBOC%0Q z?4w?`Tdfeh-E@Fb4$X;GKfHvG^Ej82AOpEF)q%k|HTU$g;v#zyUFCZMu0KSlHS0K9 ztl@Qyv_V@umurk9m4L!!SX(=NPW_;Bxqdc@$@`6JrBJVv z@DlOa#y!7IfLDpl)c~0dRKkZz-7lB&c@B1=@g85doYO3Je+}L_2gFdyVB*otP za=b)XrOmo_hB2Far&WI7qjy{IgI&{s0S~vLyBeZ9Isti<>ZAuuMx%X1D`Sy{m>O-2Le=kv%gpUO zP*AzF$|&#-NeU?=7L}P_qw~FzDgJ>Jb)I3hay>0k=rRzG|C#>VI?sjItnd%D3bI_s z;eBfAde?BHHuaL+?c5WwT~U$s;iaTq*ro`n!iIvVM@WI47b-b65BynxB^Bb$PV)>e?^2l6Sy@6A<>q zY{I3s*Kd!B49f5A8f#L+M8dc*y?<_P4 zDhoPNH&NNdNZ+hU(knTiBkaw8;P;xd##dbu0ZKU)?yyf1&%FfLpFyr$uh3)L?*%*3 zF`4gd!Q050XRo}Ai-wT1q`!5MzgInZ?`1H=Fh6qOLBd;1R@2N`0sH<4FGrTcAsqA12MYe9YgP2 z%#$A}NH1Spy~)CxKeBDV!L=J^`a4uyAxQhAH2>jWZrt(6jXkrAV+}I86k*kw)B}(j zU8uEsqr_jwvfTBIE%RBDD3@+t1%ggUxmReQDCy{*r!KTsxqDCV@V^^d+5hR4$IhGn z(>fpg+n*$JE*-y?G%0HeoI^Ro*V1ETM(D&Ht*N-+EWYa1+7Mn-HQyA1BXxg&7&7XG-CLdK(u zGLA?pA?oQI5-faWm!BFiS;%(3;W+Cx>-ZPlP>ZOCm~n-pBO+hU8!*9@Z&cal_?_MA;Y`T#)?>75V0XbLs505$jxWnkyAt73TrC8(kEZw=VAN8W?d(*9M1- zhvb(!Chj=mcnK~U7pY-elxZR7&pUYeOy^LfW=_0`QwJ?*Uy^0mo1m;HEEqs$~XRPNqQ@eMbc`6nBHTR{u!k*zrd80*7-nFgq(25wY0w(?>fX0sXU?anT4t-Ga;IIsB|Mk( z%e-K#zX87+(j>UQ1{Xd;xsF?gMsN~0&`y|{ODAnEDHRd9w zUpe0M87pR~!eneVT53Z+Z-9lef9)APY`P+ieLdeaSdHMDFghZHeqfxob~& zwdW7+2-dSjMIC7NScnX>UoX%zv#FuH&&vizE7pJ z+ss6-3n^xvysrn};MLi5&RD&I*#(Mhj~Fn$4sq%W!F&r|{UYl;Mt4lKa-!MKJa?0^ zyopdJN7vJAm)Oi1TxPtygJ>y_-g@DW^Tx@Td3|qmzob3 zf8WY$OGmw{?R%Ux#@Bl;*<`L4X*q8Z?O>OF@NHPDAxeiCZ;rIQwU$73b3niAZ$zh) zd0usR#2UQpC<>2educ!Ufm%bmR?R9o`Vv`Y#JQMfu+SM?=LqwDuU(=Tyj-V!vzvH{ zcYUU~aezkk;gy!ggMC8{?KOl#tn;SKWGv3=D1tB^Rl>=*mOO@P2;81{K%~hGrAM$qtg7NSla)dtMb`p z5p;Or63)va%A_NA531eouaXOtkW9cVokTii#kpa(bNB`4K3%r(bSa=}`3?8R@T%T!A zv7?`fL1;$cxfy6FRAyQfIV5}^)|F#^t>PT{(T`eL(GSoL+H+I(Q=33*Lq13Kml6Do zklWTz)7-+o^y{$6jNkPm!8o4_7sP`xaIde)S!nT5&EM1dFTVVLY@Kyf(~bZ3M+*`n zL+OwlC^=v>h;)x0q0%7;j2xnbg3d7-VbW~$K#*1xK~kxa0@9t*h{*kX?(g^ad(L^z zdH($TwX?JBy?0#K>tdk*0!etp&6_DEHY`^uk+Ep}Jxo?AHa$z}myh!k7lONkBx0o; zTYg>Ueq6DO95i+n+9p1bGJ2Fc%Z46as^`&>!el%XG)9r%$TNJ?=QZ zt`RZF*44kh6x~ztGRj3g0mY1s&r$TY@C0C-hot)N4=@K}l{IJ~%63_=tn0eQ_g0C0 zY1+K1;yE1_WtOKpg6azOj%bQbHCQB8?8xjIibw1>{phZsRH8vMmxBu~$lPPYI>e%; z&<%2vyv(@M)3}>kTwIZvi_ zR+5hI2UlekVD&Oq#~;i5xZGl#yRYSh7;ypdXg^+8+UZIcF_i&S7Is&8IOit+j3;Hu zL=Kd@>Mwb^Pwl^((p-HnqC3KRuhOl~qcv!(|w`Azgm z=dC!4FK}_IG6r^}ebn1{zSBzXyWcBl!5nka`o(c#9QFcZN_9?4qn1c*@UPv3*aCLS zKv6Ga+r$??xp4Y%CB_X4S5!Qj@Smf;NDCB{w`)b`7B;(9ZugJ26cRhrlJ%PyI>sB1 z`W3#vNBe2gA@}S)JyqLh^`1uH#tejc1TGltM73ofdrpnM4{-BB`f$&%B=v>p(JiE% z^pt+;lro*Y_r80xRoio8s=hL>`JkXf-UyjQrq(}I?SQzG>*~x85BnYeY|T19w+OD- zP!sbPrg|=HnL7uC9wY_HIr^;`=X^Jon)91}dRouO zQ5=^uBCAr%!dpwEw!4%@TdpZpCMH*Q}gj?MohY&xPkm{ zRQ9;7pwx1-&8?(sQl_kW8LtaXjYi=LBDoB*p$-9OBeo-yFC{>3z2yvDx@;BiKY`JO zGmQ=2c6X(ZIzr7zY3(v~ z8ja(M#UAmPvUhU-S&V_<$r%H~SKqLKv$|-i*$9hKsw{xK84H9o|Nk4gL+SrEa+lO>o`w)yz2QVxF*)dMRK8Rr`-l_adbA;v<*V z;C3Kl5|`YxSk-=U((iWfuh_P4wv8+I0DFb3*)Ow(fez8NqH_H0~Nxm@9~j4&dOXd(r${?Lc+m4k3j8gDbvt9vG0V&*oi`qb6a z4PpRzZIC;$e`~yk|$DNPIWtGo(g&L}Uco=V{ zqsj9eWnDJeZ!jg>lyVCbwooCoE!Yq))OdG%H}F@rm0HCm>*tcWDW8m0d$a2c)UHvv zxj;p`9v>gdww8<#3B+TS&EM<0NM!OIT&Z%l^qfH-+5a(b8YJKSd08tF-P-;6u*%Tl zJ;R^LW{ccN)dV-|n`3k`D??vWt8_(3&jr^6ct+PaO|Gn_vaPiSPHwagWvz)2jZa?D zmA|hsA7UXDbQw#CCEZvY_49Rr%GC1u<)sJR$Q7;|wH6I_3oCvVs};&*J6I*2^5mrC z3#W2<={E%sU8sw0)l^urhzsSVbun`&JMoJy1&|wcymeQhcKlXn&WsJkW!TKa`|C_v z4sP2XUoLOhyQGgvjzWuCbSER+oA^s-5TBgWVpKPj&`Rlj)2iaOzew!UH;8Pz_@Hx+ zu^oy&r@GoI>T%5H_|NW$CY6_PPD(iq1P+)zj!}1sqD4L7o^~pyFRhv|#Ptr4_TXi#Mpo5@`kFgTvdz8onE62@su(40OK(;7u~^1lpIEcy&({K3A8^#Z!vB++MqM_^%}^JQMJl1CjFW(uTi>lt>c+-{n*C zjPt2E-1Nr&9XUN8;t9~qNqH;EoSiq-O! z!Kw&&3mbz{R;mfNkF50_HTf^48{dp&oVqaAt=8uRQthO#i)~BR#X6=fZwFe&G~+Wn`Iw@;Ks~Y-XMgu& zt>0q#M}6T?GK+7ginq-)cshRcwSzg{wPIVS>xnrru_H>pptcX(6|QiUbI88@9m}bbqiu zS=g0SqEpOS?1bF3>GfR9LRp-T3%>SaM&F7^j0M-rKz0*prx(TE4YPcWKVG~xGw<^t zH-&$?99v;_jN7w$_#uqlHKDI1JBCTi(LBX_eYLJ|8^!unN70l9G#@+lzPhgU%SvL| zCX>!N@Rp5%y%&SCP~)gUXBf{@Zk>m2=s{h5-q)!@3J-UMd&GYyvde4gMC2^{uZyY6YKBUm zgEfhO_R>w$df&5uKwp!TItsje8%=7dZt@-sow{sUsP#;x+-Ff#XZA_mLJ+$O{Fgr_ zr1Qj1HJ5Q^l$E54@T7y-<$A4cLUev5)Vm}HY6o!y_WST=*2+k4#9Bh9)#c5$)imuH zo1JLLTck#3BA4DUTVHNMcjOf9zGP)(KVw!I&*+gPQx}B|$_?KA=v>`?g^miPZu>^^ zc0FtI1pqas#3K@fpj*6Q9{HHfWCtYi#)s=PszE zmQwqb^G}O&W%aHM=&LprA3-ncMn=!kpogbx>z@6Pt_XOLBKXP`0rp!s)X{|(06-M+ zzDuzMSSYvy3}Acu6D0_05tPn=vGN=^+XWnDQZ%xUXaCu4iZqy1#EVc2oew{Z-nujf zEeRX}^B;|qSD8PEcW_Oq{|K~8CD&3IK$5WeWJh^G!j)(_?|06`$0P*7SH0l%Sf z9a6z8XF*-8L-gwFLcS{CwCc$txZk_5>X{~lS|oSEg)1e|4T)`)J(=Urr9T9c$^E1$ z_ySkL3g@TdYKIIHTf9GX{T*E`w`T^A$gVkw%e$>M2)&96Q&0EkXXdVPu(}mGoQ{UA zroq$sMxv_hxotS8^eR#{5H(yb`{jCXUvoHDb+Zfb9CPW(AKExuOmNxUW5*N)^s>*7 z7INhwGGK?+8T?pXgE8qq@N<=4u(Ionnw<{U{#eHF51Da2XYR%zEu!~KY1@uUUUW%` zRdL~I3f6v=*Cv|V*qJa`rUlF0!FB~zB#)QD?vU!-9FUfH2YD)&TWCKq0eFvpfLXn! zs6iQ0_<$K@d7w6?a^RIQv>UlI6jZs|+&b!WRd@D5i@&FiZM@2%Wl3EI%ORwj=m(!? zjv=GYez)(Vsmm-3Pu0*zlbWs!REBHA zdTp?0>)+1sAK731!#nRceAMmY%rlxW5_g%@kFnDwzSH!ayZgU9Q zP^e3Cw@QpEgPSL!bs*=89m@_bs7R6{*X*e8oY~R71AMao6Ja1+|7N$3M(gkRk=z*j zoQ{iMN%kf#u$w>S#}%jQ8-W%xJX;s#39aw@Wh%Y(MY!m8f(>t|E4c~X)f4z_c}b_F z{i*U4k$H^`-1{yuMTdgl@h9M5|NGGu0+Be?|Z)x6J>SCg? z=g5=gvHVrg?f2UwRw(4KM|{i-8!NGUi?nlHiYy76H(t9q?tUn1n#g!>=jjQ3YOGvFiHR9!fWQif#;yn5uPG2>CqLa0`x7a!OJ@ZMHX)lUCd6J{G=nN1HAfY>8IH)j{=Vz&2zX$7L^9tZMbQCt4-};cc&;lO1?QmV zhpu@0f({_X!mpE-lf(@ud%2#n8vJCm!PeJ%(B2u5TOWG0Hj(~XzTwZLtBp;`#)o-^ zXuFIIAY~A}6U1h1=-jFYYpXimt;kOLkem|q`9v(CE35v~N@#V7tH)!xw-FDl!VtMZ zCvqzuG;(On7_2VzPN?M zzPY9m3!Tp|N|AP^_!ge{{c6N0^xY3-4m^B5Nquo@CB@ zZ1CZRbc~OyzEXbIGcm_Z=t>vap6S+aD|!qcO*O)cTiqbbRfh@jzQYyxVd^Hie1BXhDy*LV*?_?P zwnKgICiZI(u@<0+q&ML98;mVQ5+qJwWBk3pPM^!sU&NY8noQ2tPgB(Ph3cYfLyOl#v3Gsykv@FC z?@#cq$#sN>FR`jn*_KOB(HxdPU*YEYYAjN&T2@J{Hw?=79v|J-Heht`M6Iw!MKjoI zVUOXbR}8DqpNYR40{qv4xyq;?Xi5cCQVz~RDBj#%&L6aln0jO8*$Il;H zA(`niHlI-yJXusVGp{m^{a(xt@0WXdluxO`6IOigaa6)1tB368r!i47&y7o3o7gfO z!A|CD=(a-tJPm5@Mkxj5Q&P-gKd#2Wb~KMw12kwyz$2+UCaStCnrvXu_rynVp2chw z*QI3^h58^k4s{dol5tf5#}ojk3UDdl7wkZh>D&9?BJ=V4{e8wMb+QHae4MPlqX*0b ztz~GuqcN1$VFDeo)59yo9uxNTwO#wKafpKJh4_`>!`l3>^>w`=pfvMwY=Ulkxoy%ZFiTO6Y!p;9twWV)9n=JxQSf)M1xN+Z(;aARP&a81zp}k z=6yC$3|vWiZhkxu*BMeW;Ac zdaFKtTiE@qI5YkTGdIg>`jv-vJcxZs!Fr`A{izM+2K&+Z_@f`1lPh1Jx&&&lZyKAv zd%6FmI{>gReB$z+1yCG(VMLlaoE6_wp8CzCC!9mRzOS;1liS*JCRc0I7H)dUS{WGq zKD$RwxfBt6g147Tk8+4Us&;EW;|=_?91njK5BHI>s>di6<0vN9p<8*ev0bMKA_~? zAWXQR@xr@PAV@bErxiOhf~7kgfIN4}U)s`OfX>z}9TzY0v_~iGj3+HephKjjI~8)( z9_mr2c7xz7*tBgVpj}1kQL11?zra*TOR;-XsU@)dyV=>RIDg$%%2aJ~iql92AugVy zfB{c3vR;DXGQBfqjq;bLqRJr-L{m^)n;q$OcOUIf!M3-h!4vz-=Ti47;Pd~#0;&0K zj0CW&%U4%FAt{M}>jdDVI|aXZX!UWlt4R}9-?f2qgRP4v2t)ayBPOblk}bO)yb#|i zTD3yhGbYVk18){wh(;u_K>a;cMYlyk`ABV%6%EEyDBP6?qPt{7QO{PS0zWAxH_sid z%K=~*AW;Gy&o8PfZKzGbIReeu7@f_n1GS2B9T#os<22ZSY$21nHlRgW(*_9x$BZ=N$CANkRDa=x>B7Os3%JO)%1Z`g9M>+h zE)fwCaLbF<*1u(O!9AOmFtJF@<0=|cd)a2=U2L4b45(Noik|Lpd0+K3Uiij&mQTg) z1(8cRp5xIe@9`E4PjKbdEuOGVwZix5H(6}^^;v!zfx3L&ju!Lr_!t~K9F+UMd6Pz@ zWXXOcA#Omq^ZMn&p@TuA;DqU-T}OKdY_*vIu+?rtZn`{OBjsT|ah!xJ{{y05*ZP*| zO_5vwQ@cQ`MmI$XWo63LcI%aic*(2s3QMmaR7(04?^DK}a%^z4-@M{0p=^AGLga73 z(6~^12R|X9G0H(&A2E-s7$e4EB-R;~Lo=o!zmrg=%_D z`Fy#XukPJgen|}r-TenNWc_z1y7@g$Jp9QkZi9FVK1)PMEp(EZ9r< z{TG7gTueDcbjz{|$e=;2f8LD>UbdUIiJhPU4xz(Ft^! zp1}jq0bDH8(0Bp}8k^_+a0#0`dU#ZvJd!w`#2_%0ghk>w0VUD#x+kb-5Sm-(g25aW zL7($+9j#(j!wG_G$;L}6EL$3kTUdxPwvLgu%a~7Oa%nu8T3T~Hconmz@-z*39+FSP zG>9V)RrS38a5nd5L8p`|&HV1O1<4l0e?V`5J1aowFg;UKUnG7|KLVmlfK85g-~W^~ zUbXX8wAOK9i%|!GyA+QHdDU@flHMol8POpi%HxmlJau)>8TbOALY?G0uOUN<+ra)O<$-=Bl0 zXr(`@S;ppB5l+9O<*{U$u)IjZ5uLQoR5#TQEVrf2ehG(X5r>aFtkYJ}e}DXpqH z@SkqpaUd?QxX#aGwT+ev%u|(o)w{0E*(cW1vvg+`C4LPlJ-$jvzA6yYnrXx3Cvf;= z^edeAbM(RIQPiKY@MGauiEsEpi65qFCVv_y`amg%cAllhGUQkE`HjNNWR1IO8v>`H^ojjS}9nkflzr<0m6i5X*3ZY1(N7WqBqaFwX=WYpaZ@Oad`fs!sfT5q!4hUT_<(h6d^()SWB3a7UV6~^d%%mW)^ zDxX@B;`q?w>Nj*4>z`%D#8sAD=^fkg5>NyhJIO}`o8%dM1Z=hTURF#AI-`k~zA8-L$1>Zg5cK{C7=O4S!M zyauL>XRCvDEfMUC@^98~eDR^N~67J_Bc2QfaT71i3*RQ1Be^zdSRX3TrM=Kh;<8Yb2uVprSuGZ+8 zG$%T}+U3vYvPU+jHkK|oc#^Z9Dp>kN^yIkc>WhFpXg8v^zECKkFy@p+%F4`1M-{Y% ztC`Qab78tjy)QT8>UHuQNTbrF#8xMEAD&wWA0d6FKJtJ^G84{HF5UhCWdva7)}~qm z+es4!BSj63+UW70@?z&?h(+(eX&wf&;W2%rQ+~)1B>3ch!_d3WM=8XQRpHF}&us)0j5fS13LOp8PW3rt0 zoo})&{}or;c*n?q@WMw7zSpIXb637=acF0-=Ui00 z3YV5VmaBJ5p+>@goIRBxbqA{_2e9)(xHvvA5I>5V{|+9o(>2J|ztism>-}+^pKTF8 zV1qVZhzC6k8zfR@n>~v1XxiA&pUyL2auO9x^z&qUzC^-BGOd`bLzIs5sd*srT)c?1 z(>0|%s{O}#-VtZBU=+Xcft8QprLC@!Qi+4{39x))Y9x|iAa z|AQq|Uf{h*uyObY^eq*Ty?beghkahmsZ*m^dwOK9Yi`dT1@*u|r29eq7Z8XszlxmVo{1clZR0>ejubC-@{F&XJitFY+#5Wlgom+qRO zMjMIO0Cp?(_Yhek%0#r2yOd0{2n>v}(6r)|Bw9@Z9tw(k+YcjgXb1`2;os7uED`rr z1WzF(sLpnoxW*G&S3#Am02s~Mn!4qo9r3%HageeygNESi7#W5T=}mII^yp1xjsw{4 z4@OP78lS}*XsO~4UW+%nt``n0Zk|$9Z#n$@+a@uy`d)bAVh}Nyd+}}d)wq#>aj`va zEw4d*!`c@Zj<(~`D_aqgbrDy#TdQ^4~oZNMbw z^{UW+_Go7r09MeCM3)=AX4zA#d%RfCOPDdXk(L*l)7c|vHx0MsNj2+V4KV&`JP5r} zxb&{MV#>kUr}OS~V3=GU?lb<=^jTsnGb%ygQ*A@acZz@;&i{aND|#-pLpY=o{f-{= zp`70boiaR74(NB)4OM;dNA>+9_LT_U_Et9HZ7eEf!eZ1n95h}@gIl;0Sa)D*BHOwz5TJMB!;J7kLmi>BUx)C45Z?vrdT>KiQd z0T%+}bxyWBzPYO|&o%Ri;g(1l(aKD$g#$DO% z5+6h|pcD@Z%-sBAXU64={-%ff-%wO1Iji~8Vk#v0- z!$rgR+<`?1;$ikjflo?U)e6>l{lnhP`}NQ0bju9OY^QhUl&%*EN(x#VcVo7FcQEV3 zTNHJ9sKe?V&aA4v>Z3Ex?Y=8P$ejG65{Ji$9T_R##PofT5*w6HR6JB-?Pn=PEq%FH z%N8SE({5XKv!VA{N!p5SYR$Xibf=vPizkchLc(ME=C@ryxib^cU_<3Qwv3Or@3LzA zwPZeLHbVa~_pyJ?84;Q?i|ktNJ*oU#;^LhmRf{(%S#dV;PFp=^ zGs%Hf$lv6MRo;{pD--)-o;b@kz~C|GV=!dbl=?l{#34v+2_1PF??;??aB*rnS={+p@P7N#3$6 z$C|8ueC7H0u9hS#WD|*t4eg)Vvj9w-YXw!w>K~mH;o^XN+#pu5N1R8df z?;Ik+%E^(me~Gg6P=(MxyQ#EwMp! zvk2@knA#|aNpmW-Ab$qK)hbkAV8-C@KY!#2Aks*xgmXo2bmfesETfBw4K9w(&}rvwOOn{1*xdD(&fp##qA7&AO4rA3bGB*KJ^&5hPUTz_} z6!>eUAk!v3*4%PgtPP}IXB$CA@n^2zTMK2b;41>B(^Sw^eY)NQ=AMLGvZS{r0@SIj zZ?y}WfkTV|fbBN(hma9i^|Z<~Pm3uGlq{O!0f1pnX?atnIyAI+4LC|R0j_-9T`yVf z_Kw3n>q;)-a^&yw@UorzZtV)K4o-!7$NlnmZ_|)!9Z#Ja$E!Brub?Y#ii@Q{FuQ>t z3U^$HrK`5~YLCiBO=#l1-?Rjf-<-R7aez!dLs<)zFDI_q%Wp-IaUUX#g)0XC2WS<5611i{aq@;6wK(plJ%cbRKBJQ;1)HI}`hB>5(D4xD9{kt8oo}wnw z?=xK_B1ei4TeA{jZ>$w;$OkAgomH035|Fjuz*j$cvC{m}X^zOd5qtaOSHk-)TIH72 z<~xS>-7SAy4=-3fBR=y9VQQO-+$h&l*HW?ddxxuC^R?2bLcBPy+7y7|Y@VIz#X#cm<*Ue2A}nchd;j})Bpwm%n~?pe#s zX(4o&J^QT6@4uTiz2YVR>P3S^qB$bzheq~ox~6q$qlVVGm&+tj72Wr}q8l?ZvC-mj-bHuJ}_ zp(z^b{VF4c>b=u5z06WChCTc(m0y{6#CeioiRs})WgoZu*N{K!ae4&z)}HplpA-8-AW zDou7GcY|y3cDT@LP9009OaCKFF7%P3so`N^&=;Cf8_5V)vt)F8CG)WUMp<8ruN*Dg z7UGSWuUz(4zyE-|hmoKz+E-;nuhOpE+4&7L2746cPr!HHk*58(zRggOGSHBhR1#jw zz)1qeWf5S|e47d?P%>N}@!HLjCNN?`+edOrL?%RK9=-Bt$4vuY@MW3H#=N9Ule!hh zOS>}hbbojXq}S8w*&E88PxaCv&;(_p8cwwL3c+lDS>^^Ke@{r3ZrX&l59}%)uJ&lY z402$`{IMD>V_O#sBHRg4`cuuCwX>Vi`{v0N$5y?bjo+8)181-8TgGm5ScC-4w27{H zM-MRp0{Iy&>WXsqOWh>y0#i`lhAC+~*!Qt%tjR$GLSWC1El_rXwG-0(e(1S?vBO(< z3a9#Y+1Ynl?k}(HlT+)>kJJ%4ZO1D6F#*w@t~1>6d=qZ?Fqwg`{MqdIE(&!R&usee z>-+CnBQ#34rq66q90b#5majo*MIXudp#{|+GqW7kXc9?VBjrHy)7th)vCPc!lMPsmGC91DI)|muL|`kkr}n!FxFRva{L1$reC_4ys^yP3-Ew5^i_%hDP!TtQjrEP zC2XrGnW+dk``jJ-?0ocjZdm?<8Rw(}D#-F0b*hLe4Vd02U@9|UbmaFSbOhM#3s69_ zgG=Tu^p?3uWYS&Aq$Nr&Qp7Wa(^n2kXd+UBhWl_sheq@F5Dgb_Fw;>*>hffp=@^4V zmzQSo7I~WJmw#}mT&2}0RnGa}G@YRU87%L=y^j8`Cu^N=9G-o9{XdrGn}DS`zU|5W zEM{2WCg7^ZhBkX!;rtyio$Q8wRBl?G3xPaY&z$sz3m$+|XMsx_v*Ej2sgv4*i*V{3 zObC#H)2Iiy)sqaO4T=~Z&y+eXf*up4S* zm9$G4QXMo@wXm-n@A76vfjNo`B^T7?C$_!j4-b(d&M(s=9aU3=#1USE&mdG6{@^v>TNU5>q<%bUcW5Y z1gFxqO#GhVA`Zgzi~YR$YLqsWp73?#xV#NtWVv#HdB_ zrt*;(!OB#-Iji7IJ>qr3z2CJwQAezEKMquw2<2}2f%WPbhA0n)caRdTd#@@9u`BQJ|ZqudisMTwq&#+8_W2)#1Dt1>@ zTd&Xi<1bnbMX{~EmSYQ@g@(@W5yQW&>>~R>lcedn)r<6GnE)~1im9W5Hf zDlTT6>(H`qn+H0BxxZh9%FralX;K&XG`^?}Hz_XpG_+QqwO3a+!?x<<6#aK%Xk4cx z5;Q#PX?rvE)EoaGZZCNfJtj+y*#J%rb&pqCdYL8rqR#31GUfUw((pXt72coS-tpj3 zV^?;mFfwy$v|N10-dUbi-Q3&pJ~raM9A3Zj-jWdnj#zo}SZb`SdPP8Fb|G6rV;%bZ zx3>fJtF3<7#L3&{z&F%>eFdFALp3i+_!>{GBM%XBIMHXVb5jK(pS`CimZsHE<07|+ z8PAuX&m%ZwCy(%(CdzQo`e?*CDGv_O{3}Y9eY%Cyng(z04?sTBpL0-iw~Tnr8Qg_+ zQgF_JS=zYvbw{&@AfVi|CRJlLRb|r2X}*?B_+?rI7d+*efk_cXtOm7WH#oB>dkOrJ z0$4_A9u@JK=l)PJPtrZiXVN}{As4^>%T+E0hrxTI1Q0jaBJC>BRn~8J6f#{XbjUL%$@<-Nt3e*W=1N_S)ztm*~Ps< z=WW`NXvDSr#qzd~SVf|!1s~7T^bKu?!&QS<-CDR!70zn?{LNCk&%?#aZalfi(H!JT zR?)nD;#Mr;TgYjT+*y6NYqC+`VNwK1pNEFE{mkTW2Mg_8?qL2?Yr5e$-b=iu3mc%O zK%wJ^h4Q9v;g-#Yuiam~v(pYheWf&+37Bg1vn$9ny(yqBuL^0*aNeEkXIpj+u%ha_ z{-ir+)B3j#>`Rr16^2@b=Ac@CEPwJV>q*s3t61fi>}8eJH3Y%HuNKm&7Cj#vV#IFS zP$*sNQ?@8$^3IS}f-kj-vJ*OQuzoGfK}%abba?ES=A28dG(-corM~9n^#y6z@EY{p zoP?j&aMA^nuBNp8aZiZ#{X!#GUl`_nc>j|y*!|@v3Kz~-s#LC;6naZs&aF4K_Mc}c zJs26E63J{<%pZg&`%bt zZqVP581-Yo$NFbLe`uFVO)bo=qM6|Lk+HsnZr{>()Za&kk6K)@cHTB2^8J}*COCO9 zSBFuYB}lG?E+K>PFxj{pe5C?P3w7fP$c|Bp-GuNrg*j*jB2!#UlnMiNBZ}K2Kau9`MI|$<%f6Ko6%}fWhc2KRdj|n*ld+XLBmZsq@eqp+| zhq1xC`R>eyGcto~&gl-4oDLQj&rND;o(-g_t_7Cgx$WjHgqisZRvo4)ie?YcT$ec9T#$B1RPiM72zi*%je zsmk1aPJG*Tcqg(dUxB+bs0CoSFz-1W%~N2k_xvMR_t6fA`UXlzcs+F3Dq3Ah)`%ka zBRupp|8VFjlzD@c^i?JJR?-on+eDqt^l&`t8sO8e*4Q)L(d9lTk;%yF=&BRAnYhgS zMM0n?f`UM!`U^=ITMxdQjV+(DTR0W`ktdHJsR`&7k|)4qi3ZY9GM*}A7wZ&5mkau2!)YN*xzeOw3y5X%1rwERb?)W=J9*e?k-pUv#tG1S+9g6fKhLU%wsX{}xW=Ma!k zk7p#cr_Q1w^y(Z+<`cgNTN+d%D~4zpuP8#J91Zh>z4MROo5H!sL$7G;EodQTL zJV7+L=EqZI*m^(uGH#Pdv5j4$6FK6So18DK1Mu74G#zV zB(T7-yEDs-+{!ZN)>dVQ-NwT~Yv1mKD1QLLtxAIYGBuz857Ahz|xU!)^D^Z19nl_jSwx zmo+8NalbbzV8)h_!PlqadQR5L#CVq?8<6@8ctffR2(lChm+Acub}D|FtiAA5qo4Xv z50@#7RV;-E8{wM5Q*j9$M!i^YLBJI8P6cZZ_ok-%MZ|DV=XdzD&g`*QfvEos2c_M% z*8z4VGHW&GeZ8G6TM`HwPRY{!)a&3y-yaz-YZYv|=RGh>C#M3(ljy*A=-e9Dqq6n~ zfr@3!Hi{*E#=UnIr!n*aHlKQg%osJdb-mGZen=qEvie){`GhfSrv1@-f_LDLzKqlC zV*lFc?j^_IlJk4rSD%|Y{R8^zy^gTVkb@d0fo2NkC+<075K5&iAgd#`@&;_G z4w0Rbqvdt7U2i(rBo>|IFWoPvWKkjS{agV2J-8$l3yH^p{2PvNjz`Sl*K0$eF-9+4fWFRM{UtS$R?^H~=+{BAy$OVW@VKb3tr8t6sl z)E>pa-P=zYr-}+Da2=}*kNg7~(>Un|iHs+mADXgVy@bvjc0cykx)vC=qoYnX6|1ZK zG!2ybVtz<~A=>0ez68%GFY(_!avhxX;CM%aHV*P<8**A>huIhY?id(fQV;_)&)+(@kV(5*}D ze)X*y;_-vm0jg7tQwkMdII4Hw=LV|fh!ftYIn1!puShsfm}M-IHR-;Pd|^I>N$mTu zBpBGIz)K@Yk2$*Vxhu3t_y_bVedgJEvi-ossk%?S6Gb#HefK*`Ms0+`i-*mRw-u~$ z0jl{LdVKyH-R1|lVJn@dovr9UFz`6$&r!uGt|5n^s)WM}UZk_>93mM0NQaL1(J1d! zjeck(@8k`T@7J2RHJ{+6-1)@5ayhe0qD4*7Lf*BRuxI9s1^D`;=-W=8mS$TyZ6X!m zBgWe-kXw=ppaUCr^@MQkPlVfVPMi=x;(|cSE8YFI7|>o4gPWo1FH(NivS(q+;WtAp z2(Bq)`h?ur1EvFCGxUmOr(KJ$qX3u$i$D?s~GQ9i>BAIs>S#Coh9d zP?M~8B%DTrg(6MA9lq_+`z;V<&%0z>ER3UI{^p=KL#^0pB+xAV(@rO)pE9vIq*{@i z$<$<&LEYA4_I8!wiz$7~FL4Foj%1+%%aEBV%c~W{mz*Sv z`4gm$cy=ORw=Z@T(oWsO8PIPk@p_ScS7j~6RKb(%8(AA$_llv@*Fe059zxspIj@S{ zYzG0M_dLda5z?99TobszP8G_QV`B$SqL znRsI6=ju?3jBPAccX+1c^5vGO(RB0&!Ppc=nk58H2Wlr*(&%M1FaO83t{ToA8Dqhe zKEqF;EWPRojsuqCJH{N3yo55t-+nTF{lYKOyY7wjE8gC_b#V*^YO5t7Om>+qu^_giqcIDT1Z%j z>!dg_o#Uz;*V$`+zkA8_Hh%(x4PR&P=!*tdvSwBKE7Rf{f8=(qBz%KrWe6eSD(DDg zd#)<6*Uj;^F(uG#nIFy}O;xwwP>Ou$YkTWcE8IEAV0ZBHaZspAnxe%^yk)!*Z@q78 z8%nhxF{?Pk+~}~gnh|MNl3b;bQ!&@){F7PtepPhJ6`mRA$LXJ#j$Pkh@aXW#-Ta-h zs!QZ-OIL0*dzN@+Z8)r)f>vZN*WtqoMIx>YgQraet{a`BDr850XR!HOF-414rO!2* zR9j`qeX=bQ33m{ssZzIzHL_$fz+N?0U)-nG`zn3i;I@XMv22H2La9s4yNJr_%B)Uvc zsYRxU7Kj`wj3@^+K`{1v#ZCEWeXqJtjyjtLyDOQ>q);aL2P&_ln;jh%G9O;*_LKhQ z=lMdODR>~O`n&HRr<*7;b|LS|R91e*r5G$rKnvf`d?*&b^181@HqW^b&x(&X#F ze}B$Kk>`Gdf5-6Bu*ZB;TUk4|-9r=3?~w&S?~#G#G^leYJ%`)s8dKLXdPHay8Lg4? z9a8Kb$F)#5B__XnM)-&o;NAi;1W;3HefeW~T;0EGK()X+NKk)ceR{@e9;93jG5msO zHo=n+vVahY%-T6dquI5y*0nZ$w-MydqL$j2Gu*~`VElr&PJl?8E}mh;Gv(JEfz zkW{R=S(%ZmEP%5rOB2{Zg^Dp!@K^Ox5m-5OuO06g?ilF_R0Bq_7=W(;ga80jMwZYd zsENr=h~8A~KG0iPt-ND>IX2giQ~iVI1JhAb1$qlVpUBSz0i!-~eQa-DK#BN|S~ z#kS#uP3+|x)NWz*7%u9qV4ce>dV``8d*VZmZV3IC^F9J$R@FeR)l#o}pEi6Ls?|5Q zz+MwP?)T*0)S(k$6Td1g<#4n%pg`_&_?iGeZU`+Eo_pel74#LvFMq;ijUhMgP=jL4 zVjWxPzj$%a5w*)64%E$>81;65U8^>13PXIualaGgMsL6V!~_wJ?>{@Lsp+ISH0DK4 zGvsM#>sy76b#NuaIX^pC0K1+GTj`v#6+UOH6JF0oa9+!_UgJ)TYx>EFtadQQ|L=ak z9Mw|G{A)BJM77f-c)Imo&WG`;v>~9L+a>tE?64M@oNZG(RMu!*DRY!2H#K7Al%^vT za+T$_4!mwB1AZc5emUIBf2 zU*FU*r~3rwMBs|-lfURRn15dU!6Mz~ z>zV6Y)^Ih%dXyLDx4kPoz1P&3Sm+{@ab?%&1|ru!J!dziO~abkodNa(yvihRz57lh z=LywnNm3AFO?(N|yQ!?+%QAb?lgzmcrO=kW@>C_1C0hM&g9T!QxB^6QHp`!E)QHF6ih^xQ5^W_) ziDh164emU1?h;S3h80z>=6!Xz8>Bzs7}%dB5GG&?-h|9hTIQcSsZyPMPbo(@cgzJA)y?D4A7k=(=k^5~`vFLj#W6#x}g*pSW zds-JZeT@6d;Q{y4iAjpmho6-8PRU1Xx*N_FAPEC=L;w6E>IKy@9F`rMRSg+49ow4{ zC)2!5z8>2bF$wFAiG`1vZTZ!!pV|9v@ZUXBa%UfP`HT+kW5V6)yN&lV81JdE1N`UU zA7odIqNH;=N0blm*Q&*)qN=4O{{H0K)_bt&9n|c|1j0R}jxpSC<up zSLB!Z0y^ZDI%ClC!WnSn+VKU8rxf^dub@8sL!mCAZ?8#C{9_mU1c?BS+ZN4Sksb5R z>;O=r_yM69pAnNYV?drEBA9?9=C~#+9hS9YbuD;0=4hV1+~fWkJ+qss825k9TQ{(n zY}Rtg7=V5C0d!HuZ}8b2ZSu6Ej5o!%NFe?b`##5l3Jy`46d0Y76B`sRF1CV{VFehc zK{Mf)NUKmG4Af#E5h(z`M*?v}bdkYh4NQ826f5X;iov_`BFhCxqv%q|u3ivbxV z#!E~WN4#&7q%_54r&%dtK*>3OedK4{pFOKt`-KUu)^yapmaq#JGv^^=cC;>gtNVCU zp|q55p}H7UGtAl;B=WEZ5DK`xsc49D{a9k4pL1yeS~de(M#03)epFkOJ&k@^{GpN5 zu2jF*7C^;~dfr4btgQU5%jwx*v@8h`LLk`9dFRjpOcWdSGFJ>VgXY&@@A=cUi9%&@sTs2%i zPnbC)EYF!kNl{`4^q;L>+zq*RHILT#i4d+CiW$A-21*t%=6 z>#oa%?AW?GHt^T1HxTnP@t?)hA=el7wF2=S*19?MOnvKA1!tr5$_eq3W zH?8v3Nt+Dsg=M{hZYw8VOR3brhOTSv;t4~eA0HS5z)zme zy7Slo-&B7pzWn>@5OokF0pWabt2OHjvXJMk3KDs|>x6257Kt{P0GW1ZhBHYB?W4+}J6iULw3m&!HEea>vj#P@M?wj90Q6wjfmRitl?mb5zN2M+ zsd5sU#+p4_;$Kp)8)O@`QeDap%o4zJxbp@dGKs&NSHAxE+so3le}BjW3(k9=66<(# zJ03M>aejOlAe+%89{ug2bd~tAn^>y>bgvKGp`#8G+hH~$sVQy@Qlu|fh2+f~ux&qm zs5RIIzmz22SS^>8sl7JO;>QDf&F>J=V0cr^HV|$zPC;EtvF)oUbRYTG*ET_CcC)YG zOZgBq_Qf0euYEk;!P{rW7ox;>p;A$4D<3NY%D+iS*r7-xF|sK}t&o$eVS`F!T=dpO zd~b$%Yx2Mpm($ZL4C2>Wnc{5b@?R%)xBA5Vj4{BaBGxo+%e?%d(K(bnnd^Ui4Zr9v(>q5`{#}JAWhQpgwQs?s;WC} z`Os)D>=p{}ERtk4|DmuiK1pXE`CyJT4=OY(d6`C zI_Ix%L$L?v=5hE)&Tlohe^S*vofp!oiaxalb%l2VvWNin$Qi0wPS+9MF7y}P;u%^R z3XzB8NdfM7JtZWk&oBiw?31-@0+F!z2!3F?9HfDx{KiKY`z7Gb(f=A-(5LDdr^fz& zV*t<7zafSHWUqkAt1^{Cge4w92VzVG20*b(^kQOg6fKYUBx_Exl#jfzp~0oRXTzeq zCfdvSZT#fr9A>!NUC}R*Gc-yuApGrDn}E{$4A4%*@z+AD>C{3orNAjZ3FrdnjJz8; zn+Q(!&saJov>+ny5^q8E4&dn(X%kHJ$K?PcuE~L!*$_eVeKij}q}-&9ViQN=&|ZGc zuk?gBw>*u2_>OLwv=8zhayzpz8&oHMDebIXWPUiv)-pA^cy?f(%rfK!`nGWwG278` zOpCI=w&-@!aG;IzI%bo!J7phvYt+GQJi=CQcY|n9IgNHj$QgjLFzgMMcOpX??k~Pn z;D3x2bo*lkE{54%XD%=Ip>O!iho+8(nwAoif=q95Tqm&)gA&@&LhWMca8}AI?CiBJ z`Y=009vzbS1UKYm(qrd06lF)O4fMVDRlZcI;^hZxiFYD}H+DV0>4mwnoTkFu>{i!) zXEC(^*68vwva{)s$IDv^rcXS>sA1{ajbGVrKgNkE@>m|zW@8`t-n1MZU1pAN;yq^Kl)c+Y%Aq0vup|) zy(ZkS*@)VC2XzTwV57o0ufQqrPF}!|LS5YUnLZSE;P!<+qhhnUq&31y^db4?!hTMk zf?MQu*MJSaln+RP!@_`v=rGn4-3=CYR{7>VDWZwRpek`4C>Q3U_ND$F8>snv70k{ zKY3Wbc40e3j=cBvZq`T6)mcZ?k_;ex%fNV5{}7`k-4qvbWxJ{sFpYLK&4F*R4msv9 z*cVy(2L*XW305{HHY=VcjXcLCtE*x7YMTudw)j{^@@v20OX+^F{zE}(YhpefoqF3J z*e9BqBFS|5y?$HhTY3_o`1oyBXT9qUvN-yO^r_!ADevLU9y7T<^Po`JSPN%U!zt8 zV66$L#4}$`K!h=X$l~%giJ|W(*_5!b z0JjxY-zgAs7V?|@hXV88td{Y96lAyq(y{p8?83iUg@5^7TSh9Dz~#CUp)$MtEY(0v z?=}^^qxlqI2ov+JjH5EvlXfM}IE%r^T5O-A)b^88Hq=#Ji3%T!PW^D5b^%9nZ=S2Z z9u%84hek>uw&=`;;c6=S15QhJmqY%>;PqAQoN9rdqtDQzrhxaZ%su=5t&;v8hHJUh z?F+Jhmok6xrYzUrePXPhMEJkX3pvXjol|fDLWl%{<x1_0d`z;Rl7df+0nNF$3w@LU$`t;4wY&0x?=!u9s zI)_%--Cepucu7Ruq2_Ew$}YD*p6yYs`p~Y@gb7_(q<#3Be~CjqJwh|da{ed+6uDSo zn7HpO8h+bVvF_v;!`YJk>XUHps3OwMIxKihVRrRfDGxzPVMF+{rR!Wej%C)?@y9_eM*mQFbZh>c0{4q7{%U3=Hr=I9`$EUV9>NADSS7FbbVMor!R|QE)#^m< z`5N_9)NB!rMR=YcvtC@@=uy-&;C%f}xu$N4pP7+-o0lqP$)_e|LA zAD`rWzmtTZ8g%bV<>v+B4CI@?4&)3uUum>0n6u8mR2E-n&F_x-+@99eY(A&K30iP! z{CZtXS%~fN$hz~@gb>xYwcPgB&88CVfszV#yO6Pg;AKqmz{c9f)d)|KKeeWjy&2Cy z#%k`f?CHDHhuDwNDpPZzg8`hB-#QyTk@1V)l4&3MlwS2*bBX!G_4JqC#Ziji({{}( z(w_7I(i@3k*DD~d-M7+dq$Elmx_WUcRS7j6#P#nQH^1HMRHTip;R)-bJf{hDj4AWQ zINfe5zQQ0ed(F)@4fl1U3u9%yQgRK${{2M`%&ex_DAf0c2k5{x_P|KZr`vDZdoR%Z zHSdJI4 z+tgovuD<_K4Dr3XHkDWFRI~0yVgCz?<>kDm8FC|;Pi{Id7hBi24B0Y%X zy>)EnQIFA7U#mfPq@eVeWOw=!9Y@Ply3?N zw?Uh!SR2Kw@Vzfc_7ftTh#`;a;wJBW;w>n@&+l9gnpndQbr_N$J1kK0>)k=d&6T6+|Gkc+2PiXcx&&&H{%igI z*Xk{>q`$5a#EL6Gv&M;{D+_2?`Oud#okh{Rc%+5Q2&+3HkoPEY=)D+hQk0v}l5LaO zbVuH`YJcM5zBym)qFSs-v&=9&WuOqZwf`&i>BNfbpy|ttIf0}fUJg0gondMu^|~?d zE87v_JFf3GB=;w3(=8w_?BR#>j7SrlYW+yI&e4lZrn|b}{~_U>(q`?B>#hB_FV3~^z4aui-yxXw zWKWl_?w&Z`HCw{} zmF*CeYZfWm%aSuJwTB0NFG<7lyiCAx5t6La<<$%@3%AukIs(kqK))Er(7`gzTS8F{2bKF^wD8qm5klB`hv;7R5z%>9cQNfi~PgM zZ{7UHX~Wpns$WfEXpvr36h9;FuxUTcu4dX7S zsxZ$;Q4LqpoWsd>tR^%$fbrCN(2?i-Y~*=kTQAt)#9^6iFS*_P_t&HIj2%Nv;Ju?> zA2o%2RkJy_7F?MnCo=2f80;GCy7GkEu@W^iHsWzxqw~%gZaH^DD|<3+;PQY76*uKk z)A%3DmVjp{*W!-$T{x=hjnfsI(T3#Th%Nreg)&>=JH9>|x95(GB_`D6&Ws(J@4fio zEKga?xT5vts9F28Ja)E+-(2(HPHJxiScUhP{E$UKRpMg+eX7JnWHwVachGZB`=5&; zcJJ2yaL#iYURQUAghRmjLotFac!+Rxr4~=he)-poHA?K;7a9Tx0RBm3^lHEdZuB>A zSwYP+{PDTL$hz67Q{AG^t?pH9M1&vp{PNQ>&j)B*X@PI7yqNTa`JUfjy+XZ&Uvs~V zM13KpX~+gjVCfnCiUQW8{wjpp%?r_+=_}i-8+@& z@fCC~ek?K&^VP@Lt@eaN6Ya++Z-npRVS2Hi!5I!6@J*bmioAUmLAa-}of%`ha}ivo zG=F}VIoYqGuj2rmS~s>uA=uKzqH?VI{D^ei2vJb{%{eY53Y4rlmz_G?8ifmj8>Y2i z#isf0D4O#PPSd+cw3?;y>sx8FtGQiVs`qsa!3-(ys%fKr9sO`WwO&N~MQG;DsvvLj zR-KDWU;9jUOuBfs?#p8&{mMZU;k4=2lCQM7q~-=`?sdLhZSQxnZs;rWDHK4<@`@8( zJ~UeJEGP!c3eeSw)D1t`;pHbt^%&O--zfXutLmX+;uUBm2RpS(@A8F&V05d#l}haY zP0!{`8#|f%EFUJN??|d{C7i1^`D_C6jQQO6;%t?r$sQW=l=SlYt+V*zF4o;a5OS^+R$q;OY2iOlM7_X$LUQ?O z>}_#D_~CPfxWaN|zLG-K%?_5DTi)#T3(GIZuS`Cz*>lIYOgbAJv@3GP8Sa^?oM?H9 z@XzUG-Z1-;Z&^Q3va*c~*hv?M$(2uV(*v>UZ&W;D?Dg^HCO-`u>y8V`bq|@RUPwtLo9s*JriRmuJk?ncLw&YKG(rF7rOV} z^@I_ae;J-AY@igM@9rT+gs5Y|3NvHWVX~L7DI+uylg|4`+t4IVLe2N-u;dLf5{sUo zK}wa_6x1Z%xBuv`lNH``vW=KUs!vumUw0|ZKe%Hv0KNCF|3ax#GZ#jDXQs)7v4P3t zZK18)TR}fq$SUQRRu8CCE*^9(f6l^MnDemPOQfd029q(b7M@?Xz}>gS?tIR`vNI@c zCP`4)i}_(h8<=mxZS*eJZ+6||LO6=SOn{upMh=6e0rk5P5GBGfRFsm=(NWeIVM2$! zD+|V(W7UW`iIbMZD#Yf%#+?D(8a-dDe_|BE>8cm1$Q@ zS2%1@h?Xx8mv~K&ha3C%5a5TSGoVhO0y38?PSfb~flJbG^#RnA{cb-n&*M}s3KS5n z{zD1P@Dp29%a=Al)>_YCaw3OjC0hOhGuJkuZAAOY&MD6nj^>_%I;jrJKD9FWD!^z=@rn$&6P3OvH94ojgXcG8I!WEr)vhYyzcoqf#ZvmU|w9m%Cpd7JDqu+)-;O&VO;wcN-F*>$}Cf+YY{RM#`3 zGGAASlJ`}#`WAWy7BV*Q^m&yUzXw(XWYe`sTW*wz1+AWXq7r8o7Yb=bLZK@pds*Ys z+d`GYDNv(2`hu7T2K@~t3C@OOfU?w}C{iOK>Fmbjv$1x(>ll1_?gpfb^|(6=F5B3m z5izM`CO@CDpk6}eH5b<^pI}$yU3v+K7m9}0zQYlv>?9qz%CZnvmit;1oV6qQZn>*l z`hK+!px+qesEfKgXsc!AWEq%zr_WQ;O)A8-*AkrWxe9g0NroXFBVj#!1$HfU*uvBl z>sK9%PQI{dI(+++;}wVzzczfD&m((_uUTPy8ETi`GbnTS?`q0VtgmlQz>4vBX>yaL zl_OA*0Sdbrdp$Q0l~l;LI~1fTN>S>WiG4d9=Rk zk7tLq_uyIS1@^RE#t(TKq<`!(sY9Lsmknadd(Qh+TZL5`>B%j*^ero#^{*o3A*}9o zJYhz%Kb%GlfULKz!SSC6^b&Q8CdsO9@`w)(@^y-A z6;4AMLZ}OfpSPhk@cP{Uk+ibnFLg_}cx~=8)2YMAizbWTiAREO!rcrkXeivf>lf0L zep<^UiHJ-Jl^jT?xX(o?suv(Ru0>>#n$|A_`=SdBYgTY(8m>cOY{$xH3t0efeLxsW z>+KfS4z#0HITuv#UFOMn_#Sh`aXQb%U}(16OoXR=hd{nHyXq**fXP*752Nkv80?cI zgq+%sW$MF_I4SV@(Qz|n_8{s5%XPN2LFE4Z$IZ*%`h>xDrczts@2I6@T@>o98yet+ z7IjMAvqve=W|@ti$_b0pgaFQJJ9)W`-`{Nja^$>={fM0OkEf^hpEC}cZYTTfrGNH~ zNzt!EfJ!WY_d zVg4D&()(1Hpw=gg=y}9m!~R@(Sk6&YE;!u>9X#Izdiow2dS3tByLK!cD#sjb_aR!p zxeaW=?7t#<-2ubK;d<)f?y_D1I2Hi$z+$N@Oh$ zHUbP;Es8h%q}Pp&=-BLTxVz0Tsvu!D5-*lBpH7K+lRyjKxxLP2pA!XuZb}QYYmMl4|D7SbZPJYUv&`O=sH>C5r((mhv`O^PNd~(zH+q? z>e-8r=-WzGyf_9QG+!DWR@(oFf?j#oaJ85gx}kdupTl%S;j;zKY z^FyQ_1?Asg1?maZ%)&Vo%HGTn2G@@3t(dI#JJmCGv^#g_>-?(^#SR&|U&jp!+jf3} zGb;!xP-)&VhOB7Q&LHt zl>XT2=l@Vd(Asriw-Dd_lq!t&lr$(ej=?M@Hxbg2$;?8q?FbE~$hdf~42 z>lcbv(Axa|VA-pA&ulifj!#V`!gmOJ%#t z_&rHqFZj*xxo=UhMgjUwk+T}{iVk5f+VYoUWgQxA#dU2PaAYP-tqR^+hspl5W#)_M0W5S+q3lMr+@; zd>OI+O>i){vXHfX_kc>rau9eXDQh!^qS^U!x&NW)_nid4Tsi!0`8VBMNg>uYuD3A* z9Pk1j_$TitXT14y05u&3{d9&}r_K6q*kD1f2p{INujyvAc|!I)a9;4JAG)P+{4jFr zV^Ljw`-wKp(b5URZTsMM`7i2>^d91X+4=e} zI4_)+2r8+Hq>1)cbMpwc=yA&)w(Q6H|;}uo)vosOHsMyrNw|bV?C`Q{oY^1IHsT_RRk(|7Xn02FAPv zU~~AGwKBU+>1oZ2N2~d5RM@*#*C@?KNCql|6vt}Qek%l>5sfO#UC~Pk`1Cm)piU6C zO<^*NM;K|v=cga#cqJHAzTdht(rls(vG5Nb*RJpQZX4Zryo`w zx&{!H1AVwEp7*GrpAboSn~3K<7GlYHTD=SQ){;`O?&nM5$1uNfyg!-l%o&7$+N8Q% zmxaWgled}Cg2#3CXdlR(p{w{Dw9*k8<1VE$fj~a9Zqtb*8g#baW3N51YLl5anUgJw zsA4@ZHX7{wV2OyROr?1rY#0wG<|G53fXfLQgK_k^(&`=NVY_zK?fNuU_woi#h^mz4 zNnTl1jR;2PF_1Y#rW>?xQxc=+}D`MHRFJ5<%(@4!Hhn(Ha)H9T} zt##_KH+c)7_fl-($#$PSIvo>;c&DL$zU^YK{g8(I>Z3 zAn<0R)!bWUIWe||IUf6Bhb;4zj=zK4B2vp>2z)v%RYDo;L6Md}MMK`yfELv?;JGgh zl$Lan6Z`B%Z8Lj_Bo;Il7orhwJ?PqGmeyNo90vW|~=|=>;5%)`lyOy{pTa>3n9wLny z{$k_Hfjv-&D?+6%RwI|K_nD&>C2P-x1^0U#*uoqtXS0?oJa$IN_KQs6Y}CNt-O)oX zK|kAeMZK#brJ}YBxCnb!k1YsRNMyufonq#wY_GdF2#BQTa!JXKs6z>q7H&5;GkIbM nX@KDkc(OWx3l!ij0^sL)TNk}t|4`@}{TKGRjovT&XX*a{N3N+l literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-14-Semiconductor Cleanroom.jpg b/apps/api/app/data/demo_documents/financial-micron-report-530bd7ed/images/image-14-Semiconductor Cleanroom.jpg new file mode 100644 index 0000000000000000000000000000000000000000..1fa7ed09be30ad81fd28143dcddf7b30b7ab5c7b GIT binary patch literal 204111 zcmbTdbx>Pf_%(WvVuez)xVsfE9w^1#wG@Zo?hqWB;?_dZ;$GYfw78St(qhHExP+Vc z_uaX7=KgiRZ)bKUne5qfR?a^AdDeQ?ep+~1170f2E6D>WC;)(hd;w3(fGmK8fq{wf z3=0zz6B`=~2bUNR_xW>NazY|}VrmLn8fpqEDmo?}RyqbQMk*?{H|$)ze1d|4w5+0% zBK#6O0)qViO@e}rjg9*pmkbY&jGvy0p8tQlJ@o(tSkIQw%+OF608|1LGy;^TK7bZ^ zozGDIrvd+WLqSDDe};jHg^lwZIiTq!fQo{KhKi2%>=`;ba&$2AcL1H>86iEN3XVn0GtQq$5iK7Yx~`c_y}TvA$AUQyrB z*woz8+SdN_S6~0Y;Lz|03_deEH@~pBw7#*q^>=$`cW?jf{NKgp)%DHo-G8`H0JQ%F z3;F)Pfc<}P5g>7)qNAgsWB!K=1=SbX(Fo9=(eq&t%4lI)xDzq(hhY)FO)jYG#by-H zJ|(gAn8qPx5?p6K`w!axAp8FZEd2in+5ZFV|HZWg;G&@*Zyp)}00OQnl-Ww?ibRWx z@=;SVQbKV!(0p-Lx!Zyif6?hK=A*WClR^fo>>djq9SetQT>FbCMJ7CaakVY&s2&5h zo4xR47}Ja=Q1tKPeid9)$?w6kq=zoA%FqL(W~qrobR+epA}WLHX&XlVbj$RD`Z&$A zY)s>z;LT#ipI^X$PciY5F<=$<`po()DL_By^mBTz<+~zllu*zFrv~x7^uF$O$rrMD z{UdLbIN{OQiJPJe3-|+vv)J`f#;tp^{lu`&bobAqbVI7dA9>ZbX08pbYRW6t!Hb); zeLuRj)WVwWOJw;?-=_5SCJ)Gz&V)v1DG={#6a&1cVD`aeC24FGrH67bki%AdpaeCi zE7b{vw@=Xte*3DD1z>hzCacVai{ztana%hV^#k18tO3se9(KM6m8H}!fO?PJT_FHa z)z+St#@4IVZ0dL^Wpf*|t`8>D#1IeO3OWUQZ4-Q0HSsb6axmwkAKhL6fBKx! zd8vd)@cvdmbxdK#N&vj3#-`11eF;l$sX$Klxp~gdl5p>qHl=2lg~h>YH7Ffv!xJzm z{7~^m|1Ao-{_PpnNa;wFW^zK)XLoOI__nFt6s~0$adX!DQdkMzE?7lMX0nAC1p}Ut zmY?)Rh=<<|{7$@#%%o+qcMI*IOm8s%WL&MuN-I=AOW1JhEDMPLQWciAmJq3pIv~?4 z#KEG@m(_tz&zGE$k1umtrr+UfkxE@f7wj|lxPr049rj%Y6QsP;Ns4)Ah7+`D=>n3KmEW0`qFGZ@95WJhjeJN^mbV9k0_Jthg_q5N{DfLrOSE;309D>W=*Sv|+nH-%2rN4{oJ9a}uRmW@s z2Qk&9d>{+1ko(zBS;rJ;bvHwoeC+c_FtxiXGQUts6=Y{a8Kz^OLh-IE%!0E z(AlaS0fdyv4 z?1h6YnYpY$h+jEdKGw5F*&C(16vtQnFW71eWMgW5maU^mtb&=<;aeXUj+2{K>?f7}B@$}alEWMiZO@K)Kap9cP_~3^fHL@^ zZHr`<{NO+xGm@Y59ZE&1{@g&ej)oWU}}J0$o|=AHa!!ICTmL13R9 zrBDzBuP-qAFIB?tZ)Npi*=qxv$;R21ZPI1=6jotpMBV`(p+ViN@1>9`wK-Yrf+?eB_AOaWUqE5*(Qn8Lh*Rvdw75V&27Fm zR7Mfd$`Xh9=A%OS#;MCtz+iD09N73Qp&g(#Gle?kt^kXWey-pkpmZV~qX}lWM`355 z$;?Olos#A$2fTrWixl^>24HA)s;buexHmoqBp&E8x0uA*HbdGNl0u@Jd941?nTsVT z%;-DrxIwg3G3|Z1J&Omov&RYelo01t{y|JDE%BW6Z*YWuzow$MIg& z=1Ay|gFXo&C`#dRwQ`xHkZ z`2;9iv`(I$sr{Rj#$Y3faqbn_lPP-`n0x|^BN$VA^R3QuMT_a~bWOfo3)<9OC+YKJbLlD7+~}R%zZ(A_RnmS+wC)exX>ngt*ybg}ZsP;+@}B^? z#99u{$;J%>tyv9?+}Dq&U8IXX%L9_c5ROJP#V0^5aY-07(%Kzs74oUHzY`CpxWoGA z38;OayM8sSp;m`KlXqG}+fVrEq`oR?2%Rt6^+Q+cOrX2^m0);>=zB_~Xv$Wbdz;Rh zhp6}#;?3@^XU`p+Q&eI^S!)A!GcH}3;x`+kSLBPL5~&{jYMg04Gzyl5eJvjzG&Y@g za<+fvV%3t<`->w*GT!?M!0NnBs16orz*@s;yPmu;AcwGp4=7Pct|ehfo;$*RkTF1h za(i;TVJa7m ziWaLc_(h?J=qw*|JDGKAp_lC71^g2ba_v;TiSmLIF%0x1hc4I4*)wibafY+EHH+t3 zRe}Zfe3_ZFbqYbou%dqB`*mUpX54>yfGu z{g#emenDD3T#dXRkjiA>N=|o)&dFj!00N(z038j2$Z6~8E~rbj6%F1pOO%uUPLmB3*GHtLTVmtZ`SJy+FI961}b>4Q^n zG>Xgdh9w8U)0GY%u^Ci)yK`ET$XIQufz=xA=yP9F6yojruFbKUg*5fi$!a9^@$cFb zz~MoV?_ZLvUXt6yNa3!OH}JWv)(O+5Zg&9Mdxh-bQ%GWUykgZho{ z`&z`&2aVo!4g8<{SwcxtB`-A&OuIjIqP@s3xmJRT)DXB$RYVz}!Kdc3rvmZQXS|3b&atiu&}*o9GBK{BIBOg;Fe_69FXI1T%nPbcd9 zUuOMLZ%yIvh(KS1M()m+bqkt=6d5$ye6=?oG(tG)|xrYHh*N=uVI$M6+!=J@H71L^Fs(utwY9hDuHn~ z{db}ZUAwJ^F^Mmd{z#4_xWkGxUJMwsEsgvuPJb5&^ zDGUw5>p6poBcFCvR7=aF5=Qtd3OdKAUUX7ShWqQ_W{$gb)xVdc@<5dqB`dRq73)0# z@@a&is2EU=ICM39E--K?w8NbG`S+uYO%gWpHu9?~PuyuUwp@lnKUOs8a?Ey@Ok7Tn zW%%dJl^f6Ju6~OLUQYn5>4iELRlxk0eNWTJdc%S?`zcosiJ$rB5CP)==7Fj^(^!Vq z95Q_F7O#gp1j*N^)gdu^Lc>c=c$_R(az*OHJGRMxwR#N=f%$JyB73$wN8dH2SjVlP zQXBG_?qD+1+&lr=Gu0u#Nb@|s2|WJxy%_B%5l9>;XSt%#DEQM~XsYDv_uIp>NxJve z4P>U~_M2CO17Y7nICsk6ew24s)9~vkBuKKDD^Xn0JZ@-Ai77cXZ=yBt)q8;him!?_ zt(N)zt!4muKmJ&&uqD`|!EmL6YF!Z%4z0$0E0o;XA5?bVg#q?jKB)uO*H@ByHVNlG z0r7$>zfI+F^Q)}NJcsWr3rv049HJQ=>L&i8Ny(r|J`2PT%~zCQF5xmi%24>6H~0ip zn+a7cH;pbw*`O%d;=1>Wj0ZM{xo~@R@%*9PubRm#M~*S8Cg7e(_Gid8JeIgF`zGqf z;nqmVHZJ0+T&cb+jk2dk4c^iuO?tkR1pm&bCOCyB_)CM~i#|alK$Y9vqPOL4G!?c0wmkYfOKOqin1UuziC9Y=YXFD0;3FgpA0@hk044r~L#x znnEt8AjSZpYPF(iLUhF;<8hX|Z4cRY80ou#N;3tLLp;1bi>bS-VlGVbvYeaz7qo^j zTGh{tJ5iHEo42$9i3*cf{VkxuN6z}z#3Lj@j30V<=@<_)g0wKktE+mFfXd3vTf~!P zh;fWlVza(DXmjwh0T?-27sVK%)VeMVSUU*+?hm%e6nA!dE^C)byNVyH9}8Ed(zWrY z)oQb@cwxv|LUh$$%O9tf+kwSRaT^U;1ZFi?Sx*`$SQK+cKX_Q!GhUgQ=_ToZ+x9+3 z){k$te_6YPn2cVE@!BxbC2ZtmB}aq1z-r8NM1o#LlJ-a)Alxs12Np_90~Q!3&Wi%X z1?{q&xd9%q=hD|Kk$g;i2V>VR@)EQLc{LXMaR4OfRWdCdRWw*kUYu{Cpv;a(hm)ZD zhw^pV=7F|0fi&4kul~st+^hLss?ygvo2y4X;!hz&W1?!MziHt`OtNlHXUA90D*V;q zzim%|O)J8$&98w0Z%2U zCg1OH0YwZLObm68ONO} z+VcA)Y+>Fq@s|zC&Lr-shbedFQ(1IzJhy5_EJ^2iXiYZzSN7ui!q_JnyZ=6N`cx_V ziq*dC%%FYIsBikR{+T(1qI4q>Kb}&(zCG)M_z_N3fxK`^K0?xA9%@bWAo_ja#b|Z+ z7&5xg-h6C6ke+JJ%aFm++50k=O=!<^C9x{=#-m1rp?WuEnfvQU?(|G0a#!P;5vv(t znuBmh>>B3gwlqfG+XT_gYUwL6@z$x=$D3S-zr<-Vi}e!NnI>4;;=+D@#{|V&H68x5 zSnOdFyLO1LVKpzxxGFW@7ILYz%vGEd&&*GyR98TsIs1s$G02tlStdP=$1V7#iq-dZ zYtBO9cv{VXFPO8hD^}`T`IOS$A3W?z+Z??6x|s*r zcAOT(zXc2-^8{v&Rf^A^0CGkb$3~bp*lN0(gqzvv9Y7aH!O}>dbM?`7!9VjF!y?g% z;*gsszSD_-(NbNBQL-}iw5hiPmSS0;WV^WwxtoRY07aQfpB}3{UD3#z%l(&|1Df&t zg}d4EjAv}^CMOnYI_JNcX1>Tl39Y&I7%@3MSE?262z~;<$vvYDL61Hy zE@RK_Rs_4r#pa!khBkAXJ5vH3OmRQ9FvBX8y}Aa>_jvSn1+pxoHI{L1(&)>UkkI4o zJtVSItUyH(1oBv7zF5Qn73)RSzLnhwCUbkeX$flk2jP(2h!;7}vrl~-^##F-h!_$x z-aNd-k+_nyOg&@BV~fTAXz6+x5Q5|fidYD*ywjxjw3h+Qd!QrY58E?W>BCQe-eAWR zQ|f*BZI)>n2DO!e#eEmG0ixZs3}IB3b=~*++PMn3hX&ZXZZ+ABrP-liu7NwYK?9&? zh3P(cI3*e{yG_1gP2Tf0=sLUBh~75E!VzstiTe(FV9-v%BYR}`!-eZa#pri?ZN%sd z6R30kHU53_HzfH2eV2U#YG#DUr5e%^hLvCaw(6dMFbcJZ1hM<6SFdkAC^gs3GG{UJ zCH>URWU<@y$lY0LU)t`JsA!K!$`9GSpxJ0QHaGYNWshZN`xc6W?1WIZeAM!NCM+w^ z#k9j;ty$2F+!OFV^>zyFO`ft^UkfbmOeuM|2nbetHz+E2<%Fi0df#qIDhjv`bk@Ar z+(q!FtVQ#E3a7>laYqu{TFsBoc||sbww87$S_1I31R!D*yBa$4+%=5tElTU=)2$;( z0W(cIA*gdMQh!)Jv-0t0Xk01Q%8`7XYU~>%{>h>IR??mvg?K7ZcdSr|fEYkimul>B zSEp%2!5fo*hg-O#Zt!QRVVSB~fbvx4(9!gw!k56wvL`6<7ryL)8g{0!SM5EHT^fpTQNwK5&T4I(KUn8R<>Ceao}4}kg?aeublJ}?l~^r zL{GIHN|^emnyrYsj0~CeHu@I3;VY|OG^QQX+kXp_#u0;Ugu%k3F#}9}H=xD%l$+6yZ%*+ zmV|>`tRmiYfx7<=mZalnFDcEj2Jd<`x%+e!h|ffogZI$7f$nT-1(=tR7>ZLj<(G1{>WvzaM<`Y|(ny^WZbeAr)APgYL-Z z_&os-y>6+Fx34csAyysuxyJ~Nsx1hIsKlVHtMv8`a9hJbYk*q*+o4_M7Zd-VWs8||46Ug+ikcs^ z4b)uDF^>%-Vtc#;-ApfE-`6Fc7G_8e1RdKZ9;KR80@I3bpMXY`0Y|f~h+wM1&=C#V zuqtk2Iw6&eVO(Yx1?mHnF@D2Ejl{2Lw*-%zO2SM{!cB`0wK1pizApEgNknjTk|$u% z(Qe?_q*sj6%`UIY8|&R`&PXY)<)UZ4@M38>Y*TOw3Hp@N@bNAk@qvd8ERiY zGB_MRA&!#5UX@z;a)0fPV(Ih=h}Dt3SD}>49?8cWq|~TZEB9;O)QarCvS*CT9l?Is zPt|AG5y;rfdcPHxovpJNCOSwQ6)lE8l7u@fIl-wTzj*{a0X_}&8c#q)W25IkQ?w)> zccnu?!&z*^xYqI*?OYu(?)kz4XCu7Pc3!Dbm6TQrdmtF1IxAajRN_$=`a&QD{GOpr zJzxq_G=MUU4{KK;$WfM3SBbMa(3GqaKBQoH#l$)!wp|5&0?rBbCuQ?spH>LP)EHIj z+u+}F9TlyUcDQ!2CGO|aS5zEJ$IkMf6RT=T_pEtU1dz>4q+_*&uNLg!F;eCA=+2+* zyOyqEYx7E!3{gkpi$1VB4B4rzqu)!VI%7Tx4BzEKs$FRb{X!0yl@{ zpgY*KgvAcH6}@(vj~(0kxj-Axm1yQ@0>R(s`fO4X=<~C-%kN*|CyUm&1}oxB_l2Kx zMp1q43Da!V{hD>Z$V8u0#39*AFu`YCHhKBA!noO6EoVY4G89ch%yrpaEZB|1x!KLQ zQR4Bs5pAi|3PR}cN?9)3j;eJ9LIbJxfAcd1K3u-~Wg5ri5dB5b^&{s3gJyEYiyj=( zZRam{Cm%0zFih+(DGP|COnG2R)Z#dOpEqKnKPO9^Bq}CeRwc$7J2#hZutom;hl`WX zX;q_fS>r*Avv(BEy$z|P(rf)t_V_hey=G@)JDy6_xw@t}wLTpOe^ZwS!DTAziuVP(Z`uGg;?BWE%);0IC=)W)8mCfjlEkPf2kGB_NKi2 zIa6!Gyaw*E&dBevgM=;j-f9i>n;qRVnjcQbg)NHjRBz(dJ@^uou4`!rHpC4APYWO$ z6-jp{pnoJh?1Zz)N5qlV@N#_Rx;S(8oV}x^uY1y1ii5_O#>T2^Uio;g_)oy#L*C^O zQc-%(>}f``shNssLvobj68>iUZ&1ts;uMZDGEPC+*-#t_?{&ienvo5!;to+J+*dl( zUlzr4wM1&FPH%+`-$n_UGyV(1$g;f+F{OxrjE1r^h5!ifS8rRd!Q#nKRVlGEwe)z< zxZa1%9k#T>X`rgaYQ+!TwcAD(@vSDv0tW{5(U{jp)TH@D>O_{aC zHcZLWu_vo)RP(R2ufq~EHx6rj28qxvYYj0v3)iQffR7^`RGB18UAorVnWZupjnnFz zb43`*Sr*2h!&Y=%nH;Z+WG*dKu@nW-v_>Jec>(*#ohA%gwlt9)gOEyk0yi z8QE+&{gtT6i~IwekiY2%kY!X*4n>?-GOxJjztvO!x z($YcPKNF9ifWXsAfqhSY8?VZ)o>N7wM+Zy2Xd~VlDQnkagECTrjCTh5Bj1M@GQL}g zrGK>Nwg`7L_oP&!*N%V}E85PP(-D`H$B#g$K+)Hb2Va54v|g?hH9L=uKkMmt%%U|kXrX~^^ycv?TdX! zTbaYWdjI)sSOC)PvvUVSQk=c6qYc;VPeN3=7M?b>&Z#GpFKG^+y}cXA$_u#rFQ+s5 zr4CPeXx7BP_nrJ#p79&~KBJ;==_UI)&jz#W%RR@$i=oN;Hqgk|Cjgr?H)KU#BH9kM@gr%{9sk%eXEEM;NqlA&1ibjp9EC9>fS+4CpTNq+x7nDxBBv!X)GtV!a-?8y5( zZH>+>#B#rsA(`@wb}Y+o(dFZ9lgZ>SOXOuYY$^P4zv6-nq&pWp0l5w#jfWj5stdRw z=ZLc|Pbp+cYRUVP+xZBhqs9=K1;Ec4s2Meev=2>B^FxSHQoOpDsK^*xuRl5j818|u zx$+k!9*2=CG>X0!~}vkH$s7O0^>{AllmRCkWOk+NL_RG!Q-pZ9_>;%;U?THw(Adi?dG z<;8F%le-es0JVE@gZ3wTP!UxW2|JZGg<2SGM(cgyd5JSkcu7g}i8dPFGO>2D92$kF z$oA(hZf5#eF?h^s(Pp)XS3dq1Ao)0=tWdi+?! zM0;~qE?C%3S*9UMj@oQbx^ZCr)i3sM;GKRZ$4BR~4~&h^bnKJ14ynvTfLW79B`^OE zRFOkhq9m_BM!}81igg)K6A|Sj@=+%=+G1UA>FZZUt;k~QZ)YyvrWN66GKL24_slyT zxCy|5I8ok6VN>ic^O5&TbW)h7|EBKG^k2J{By3C=w(u>7x82X;57lbNT(Lh-IZFJ% zHB(-&Q=0`@qeBW?rw)}|F0UCCyI#48R&?Z2nfgN?9_m&9He6`3d-5%M32ecu*Hf)J zVyJanF{*0fN%d5%+;?&F9v!-=JT&Z&Yn$~&`D5K6FSfshj~{(f3b(O{U^<#yd&c>_ z&P2fKziB<>Ff6c6W92|T+2Nz-I zG{;neua>xr1m5?gY#wK95e9QABlcG>7B`i2hq;)wWDatE-9gV^HQ(#qL{a~kW-1`| z>AkL%kNIUVCF{&RyQQA7%OCrNwU&Ici8L6ux_rt{>z&6?8yDk8`2gh8kUsl9}Eo@{?m5!4X6Js?G>nQi$0sp5&qN z`*;wE0&{M>+c`exm<*LoN3-Wc<@lGza9&sp{aLbE!KM*Y`{~*nawd%^Q}0?y;kxoT z77@`Qvh8nr3`K=2zXg|L`1yU_wa9NM8Q|PsC1@B2Z*3l?PR7L-9Z4aYzBOkIEN*z| zTZ(cxzS2pjuV?=3WfPtZ7pgk)N}klnWE~5@{DqZt6LM}HD__5&K#^z``2>hv^a$L{ zrpmBlELyWYdT99)b|D@*>vZlmu*AdkyZ(V zXUvD=7Nqdlo)?wO6tKy^ifvT2kOm-5cBnjWXBydSVrq!`1@E@g;P_PVmviI75!utX znkb6kzyNJ`X^P)8$P&>tlBp`(LiZdqPtRzVJaGSQA5m%2~h8&5DB9ml(C^1i~M;X!6<#+|GBLiuc(e~Q`E`eHJ`bWHSv z=Um=~R7xQ`@e#5TiCPyVbA}<_MdMdhy~?^iyNr(d0#fXM_LX}+5f+ukHz>kLoAb(I z=EGw>Bl#aT;INZR^!08S;{-}fI2ldr8ih;HX~DjeLD3VvZ* z!dvAB#PCfKBqIvf-9H;^(uDKucqQK`+0N4&fHSoC@CwXV-VrvwGmMTRJDsF7*WL+|Cp-O+#e~h4CozjJ@BsCA=>!^pNV>cb^OZTm?J) zj%0{W%Jb2C#@PTZ1qXB$_Bo<5FXQLn) zzybZhQ!35lG5O^Ac)}MFQ+tp%Q4p8OJ#+$}EQj;dC{6m{ym}FrK)`(^-A5d9`C$Qy91Y!y@~Np?qO~DUP5f zhh(Ob^r(^|8#fjHuY5j#(FZ{n&)=z>iV}_d2le0Yp>is4VBrl(_B-s0VH6g8s0xVS zxlKyPS+3{B|1eU|(-SO-CN))qB9*S+Fym?&==cfaOXFPfyH^DVS*`7TTsr~Z%GP;3 z+zkyhm-;^aUCsQ^p1{kIyV;U@2m7n4th&!Sw|Q>(7ULnPDB+ovF=SY1}lkXlY`>F*+6@9&k|uMXU~V6qdI09gZ)WMJ{L11;2?R* zt45?K8C>63A>$d^b1LWA^h*DF;oy@)PDY=dR<1n9I+8R4m~B<7#)>dDLOm|(N%caK zfa!m(B8+HmJ6w&A zw?x68Tjx7 z*dd9-Jj>>(&|e6)=VMD*E@%e-5vM}^jOqu`yt|r3Z!s4*687@T&ViCrZt(7e*qtWI z0ragK+XJ}jUNc1oOh-W7gtW(z*Wo4wIRpz00Cre^qT94;YW4v$HSeIKN(l7P8eJ!$$E{iq25;fD$@JpK*zFC=uJp4*f$Vo_)U>zL||$h z)HL7GWhWxyME4R>>BYVn3!@BcjS7~gT`g+jNB5|Hh;%TiX$@zAUfUFxeHabCwl(==U@mMHR+@&i-qx$IX{_Z=L{CiVkhQBDk73xoAnU_5l5!8sqOA zHNj9F=ib z1msOrIzY{52<7g_${CUmkQ0#bt#~Y){9q9lIueSZFO&#*Zu$&%47Zi;jG3f0K0?rI ze2Qn$r|kBOcmjf$-5o`gT$Fw}(f3I1m!4tDqVtIKg9zEz*oW@sisq9|v6)mdzHdOdRfwiY3mSS2}}# zU^<%`n`QR|SVP2?PiK}>kw(^9ThY6A9BpQSd8bn@=lPb5?vV!X*4_@d3x-60dLiZ% zvFm}p<&0DFAXalmW~dhr#efD9!J)%jhothO&=#WFoyxVSCxSxvvQ0Ip&A)wsZQdYIRHk zg?zh`1z2Wdt#-(VSs$LM)Rkm3B^V4$Yp|4qyN39rQ7~se*2o$rjv!^aXx6)Vg^f?6 zGqWMX6xbyw1ic_WbAiMePtzZbrZO(|82Y`;@7;8-YsL8T=2mhC3)mI*_c=Wx-6~9q z!WFjiK1=9$^+QhvjDrgjJ;|+ro%|t|>{Qn%D~?~To;)H)*AtpQ22?2V%ouMLkT8)<{ir7U8 z==GinzZ;WHF}O3|wvf^MXxz1B!x`fL@grUSSh9XA*smnLXRo^d`H@a$TzqZ z<9m#ko<|276hiF@5t3rlJLQJO_{dVXfxW|az~xi~Y%Xnf$L<;v+g5P7{V zQXSRQ7U-{4kd&cm@K;jiqHJ?oNXg(jzj!*5A+U;c=SE7I{#%YPXlkx7=srD{tL&bplT=C z)5$)K__^uQdl+;P9zt;-{I8#L*a_E(mc~K??V@zpc|oKhJmy-bo_3vy;AXzzLC|mM zx098-mMQP4Q+r9Rzfe=3)?1cMH&)T%BjY=+7Aev*^bCADpSVrVsW>$4r=F<9U zJttSIu4bd8OR7ZWCZkX-Ibz_yjy!Q8Q&qmf*M_d&CF_or&MTIXh7$WTNf$M$k$yJV%?qx=$|JplnOutk}Bcs!gX^+LsgJ{pjdcBU7PWn(;UG)+CyCu7+o; z(Q5D7cAt=mk|pjiS;NX8INr{|tt)D2pWr;3>tMg@>uKegNV06DjfSdk#mkop%F2OI zJ9YM07q(&vm?q#?bkvg{kqt5CTT1Gaw7)7wn@gHPiAzY=x3T&4TGNALR)OIdX_lc1 z+r$-*gV*U*PYfI3%Z^0c%~qhqN~N_9&ZWXW-%imLg3>Wdl?9MdO(_AqDwUQ7z$Dq( z2bWXT1_HhE(VYn5+}|mTwQ#5#`GJhh7E0vOvLmM3@~sL}N7};8#_vlW0zwKetJPl7|NiQ1pvGDNPnQ&#M69Sxp4w^w6#RWW;cZW^<`kmgIQF}6 zhsju1?n`QMdFV3>q3ZHzk6+=wDSS5+!_jMUr~^a&=?%FHMPWHRef-McHc4cmeo?#g zW^6#7wGx-c@q^C%&g?O%N&eVN=XaPW@&F?I`>Oi++F9;19bdwQly^+Nm0&xYsDHNX zJ*-r4=O*%M%kya~i`JI0(CoYbzd0TQdd3#5Bb@ExbZh;N@CC(ZMJZyNDWeftktX6V zQClUW;p1k`#@C2I6vTd*X4lbJO^sbr{e|eBd?A>iCQqEoUjXt4BSl3#mMADmjO}=4 z(^K37J`_>~&a@M;wUivE_$0&qvtL@J<-3obna2dQxPGVY2rST0^vcOM99U=+A9dZaY0*v$WXGNf^(gxQoGem-V?T ziu8SJjDZH;H~WtQW)JyfA?I=Dj;_CimktBCILkV>4;uu3SlN7L4TYenM>gWfL#kW&BoEbTrk$CGTVQJPdhX~SfW;?KBe%Fq|Z_Lr=s31 zI~n>5clf{R<+d8yKJ{otlSiQ+mYF`!d$|YAIMED}OFDg46uHB){<&RjS0YbmN#fM2 zXIsWH^$;YuVFoPRm>WA=X}8;l@-DZGucWwflCJC1)4f%~(O~_7=lus~!(XuD{v(M@ zhT#So*~ANV$EmPinin_o)(&~AnjD;k?YD#0dK$2*fsP`*ONM&2Q~#1QIRPSiI&@Kq zv!!NU!Dygp2%`a6cILrf{r)8}C~;5Gr8be%5<;Xt#x**&>OkjTk+AeNRPH-%G*W_) zt3cW|bR!P>DWH>eJY_xLH+B;KuaRM@^rwpup6U|-XF}L^94XK1Q0Ofpj|9?tNK0JY z8RR~2HRo>MYa6l<5Ks^*PI}7T$iwUQ$9gUWf3D16Ds;GZg$(ZE-@x!N-iA1~QT{6; zW#K_7;?GC$-NG?r3R$-^bZ&5b9~h?~0Xnz%e>K$CEk$_>^q3~{zLa^D7zHiYFGto> zmp-Pig|rEQJ3-EW^Hzs`56B{fCLQr5jOTk8B4ghwHA-)^=X_ekxPOUx zbMuib8P{@=q002s2=uuho#F1>_t8#i#~TeV*0YM5czViDJ-iOa;@jh^d;+`|S`M-n z)zIXiS`=WPUyLA+x*#EEMrr1_B$C*1ud_nV&p{0|0?M8#_p0ux-}_;W7xx*$qZe^H z{f$^1`oHNg&BIH7L)yWSEyZ?|Jb&M(A0@19RfN|_{|xJw6`C2;f|4YMZ+hw+wUB&G z0{PAAlf14}HQccWbktTde)o5Pi@>a0htf_cx#`w+)QS~zbze5k<=^muqKvP~&Dqd> zO9wu+4d3_UM5rA<6AggB4WtQPd-?B!qy8kzC3KfR`-*NJwCOna{cLqCuc!A7{tec4 zlN52wgPMe5QR=4^$gbYUGY({0h3!0=@Wa^dyqaPrqfxb<3VTc!7o13 z1;%^zoxGOJ%ia{DXH&A@axvq3rz5$*-~v=@q%W%zy-$sxEo>k3>ds3xW&_tHKgTZbt&OsR#ML#;Cls|4Nfk!`@g z^g?Q8CC(dZzE|`wo|FV*lRvNllI*R;8%dY>IPh|9@__Q8yV*$dvv4sArB|FoWou&^ z-$RMdyJs}{J;%w5C(BEA1hGE+RR~bME@q`J1B&`zJE(_;5_PfoZx+7>yo(-oTc79s z%Sw<9r)(81#UMYpg_|i}Ww(d0qz@?WJfjN6)3EgUNDt~+{7|rSyIt{fl)FvSGl8E& zk=AKi$`6wKqf6oPVc&$jP))En^q4(vSL{1&yy7u>i0j@}Qy|N1<6<>wP{hh1T(+d4T~I+}QugLzq!qkNjnr3>j2L4;0y>lN5x zd8VXKCc35LH^|Z$q~R)llc+U@8@VF7kiTYqRh4-7b^Zx3aCZBnI9!vU|1159=5Ju$ zTt1707304Wq`LC{gzqs|hHk&uANkGp{h{5tDaR#i6L$-ds=W2ii)TWakWNk$1OcAi zz4*OMHR>`6epiuJ@2-3HEX(w9vRnQH?Y7dD4&BmXh6T ziV-z1T9k9aN0)4hNw|skFUDtRE^Z0|C!M&c{Q=|WiAjX+Cvt~J1w2U5IQa_7jRX8*$H|2CqB;QTCZg;VO4J@#bZP{zUDD6_<9jUbFOqgmv8j z#F(-*;f|~?Y*dX{83ff|)_1xXX23AW6~)Z50U$4r5=YwNeSg+&7}DvK$=cI1B^|Uj z%X!()iFs!YU4M)%Sm?A8G4-yx6q+_XzxPVNbnK{Yq>eVSB|FMa%^RpYRuSCK4_a=; zoNKP~n43q1tg24uCDE)#Sxj;i@A{KP>6lKbA2rk$S5?UbSPR4LP__BOhK6g^ZhT#t z%*b27>Sm5HAzymXNdDd>*)}gFHm&wzitLRCzK*Dtnq2SVU{GRM7}6-W9rt>-S#~M* zDJV0HT3$_+;8be&2>>Z4tZIw}cUK!7%~fazGFV2VLo}t*im?$Ke#Pb6Hin_Ua!Um( zQ3GevzNIf1j?WVq$jo3-eeyj)4A=7C2ovv9ln+X2Lw4o6Kp!@L-Fl$!tN z-Uh_iN~C9zs}NyZdzi}nP0~JWPbncl8+qsnOH~VMswEB0iaT<(5R3)+w_qdA)s=#~;l7(A<-ucFz&61N+3&ELCAzSZ-Oka*n~bWY1H*)Ra9T{n37`bJGD zW5dOrUfd%P6&U=RH_)CnFkNU^d_I3%rnom__Qs+#m&tMuOda$Sw_ z(t!_}IMGG*CzXch;Yw@)@HAs)0eNyvHsm1#6+vJyY`#p0sxC%`D_BAFvfScxkcoyx z0wi5}$cgzkjQ%gY+5^Fn4gj-WB7= zJ!RB{&uHEl&YsUa$wmoMZ6QFa&9lhoQAwvV&Y7bgs_rd9dJ7wgVv(#^%f5uWiQv3&jkc7oIe37f? zVaIftNso(T&WMXO4&o=^JWCw(t{P`-jA=TfAY7Q=e$$YKuyvGypBT7Hm3hAZU0N$@ zN{hxx4l@mbHNVR?;k0&1$fWfx8mQ6Ku}1Q**Q6w>+=`T?%u9P+m~AA$RF zHk|185zXZR&^w%Pe8rhqa#8W8gz4sRc-bpr zsP#a@HF%p>9OE@1h?g$#kaf%I5%Wi0pE5++I|d`(X*^mCKye}l;|Ilb;`?cn~WPazjBBq z%$o1%BmWIo?OWWG6$zm+qGxM{Ia_tIRXRNqg~rY&GHr@XwFQNruJu2mof02JO+O#G zQ#$c+L2Xw)5C68kUTy-Agyk>lqW7vVX*IKc>g7NBKiwABi!HBaJXu&1JCxG_)x?tT zqjQrgr@^~`HmN;XO5x={nR}llUY@YZQ|Ns$`WQ3RjV~8X4||?%po5@wuIx1>)mazH zYW|Ct%8ko2JlJh|Aa?cINi4J#+THw#3RM8%dd4VV5@9Atlk)Vw)bmOI?X~>WOG*)$ zr3&b$C!%f|J?1GpVMOJbI>L;ZCWS;{1}8;1;qQCcv6K<7p+0E1*&s&I#4MBVk^D-pLK##U_@b3=I_i?4G5@d0I=*Y*3rylTl{&a`y z9nV_aUBMFo(=LozO^u>{yA?t@82)n}rq-W=cdK%(qDwF%mlU9M42j4Ov;>uClwv~2 zrk#y6pJ+|zV|DvpHs%Ys((8DQ#BV)+e|?4vB@Uc_+Ec?pf zR+jH#b&xKuFu6zA^P@vhW%E28^`bw@7A{_7`7B9_w9&rmY01jX=Mo56l(4yVV3 zg6|r!eVN+Q0(plGyZ?FZbwl&kB*M_d*K76YQJJjo-p9vZ{8uei*RFNYW_|ejQNJOBg!ki;#RG`Ep}4n&aUCjj)kbRLmc!zJl9n@1vY)h}i8S zktb+(rc0FaN!_G4koR}X(Uh@q4nJDcNhQVf6s~4ATAdn)Nv8bRg>j}DOH}|G)%xbV zxw&JNAwdGu;m0|>Bdg7I-YQ3={z(+6Jo7rD6=7_MuHYi=Zd%e*=dYZ&>dZ_Q48Cze zUr5wChM9zsC@nL2uD`IgdaPcT6a8A3=*|V9PTf2{(}PjiPM~W1FYI(WnJ#6_$kmg6 zN9(ctQZ8R6&FD2toiybf9yDV9V$`b>UfY~s7?C*Ef$An)jZ#7*8PjQXHB^V;Ejaje zV}~M->MHL+pW<^fM%vs3So20b6u2&e)3+roM80eqvOg}cp3kB5ovaKZ3j}x9I7>0) zdDEqp@D$#W2Cd!0&t!+*rTL8DWAWYgeAh&yFrk(I=m$O=+ zbkvT1b#^A>wpBv9rn#Zj)D~slu>6&oLUh@g1Cq|uS`X`~?s^kpp0vY8X1X^79#vhO zU;gVjUmXvbleeK*FBW~oW@Z^6tCrR_Lt>xsc26FrYH?ITBL;}M?(CoJTINO{gnkIj za@qM}HtdYWucMkP5Af44l=;5O6QisP!kzv|1H1f&X!DfFektf}b}OGVI`y|K6l?-V zRrodhLy9jIAVvoTpN{@mXQod6{l`otaS|bK11pQde|Fn8K=ymWA~ijzmJ45Vi}v#E0wZP*LDa z(--{W)zAKige4bqs-!dn(6Mi*(X;$hgJOh4zp!g!%>%N@oxP4# z@Tg{;+5dC&CFAZ0TBu_0sipVVQTmRgM(}Za!r!@$%ybUiH zF^v;u+MPP?{gXxl)RD=Y9LxP@S+(%IS05PE`Gm-x1caKfl#S$(T)0&JKG~SN%F>*ALOPikXx5 zx5%}FDoBnd#2?+nDlcQ3*&Gn;^4UumH~itE8$VH8fjwF=Fc8< z-)Sct{Lq`u%22RelW~u)_JvkDr`O0d<~8o)yhjgl@A5#1dH5Zm#YZLYvM2%LUI244 znI-u02y{~(EI6vBOB0#;?Ehp*_-(WkG+ipd82tGRzMPZ3`qnj`LpGgsR&&%pVk;60 zKhnj3BBhwLFjNqilA^=YdqL37md!bMQ)ZU&Z+C5uwCi01wx6?%vmlm}f*?joddwCS z7h}zJr3vf{TNA{;!ci)7Ti5f(8E&QUDWdP4#G~MhRKH|j)hhI4=o{L8`>961(m)LN z3hN(IDQF!u-*hNB{&)IQ)~uJ#;RnP6ftTlBr)xrPu6Yj@qc<*19mtB9=!Z~&|0uc2 z+2w5%*r^aCkz`XM?-CQ9YG+6S)9UyM`Uw)TF#luJK&B(~KemD^<~b$az5i&Rti*yO zLU|+nh;SodipQ=lJ_(g)>3J$!>d3PdV%j_iyD4`0ioJOhM`Ke_raFgS7quLDUF`0h zxVkL8=|C&`@hNUGB+K+p!`N4PxN zZB845E@QA`ZXWr^?MM`QMypH{Rk!{4Fx6MgJg4L}{A3s(z`;AuV5Vsy4V$^KCgVBm zv#Wb$|FK`KEr02Rm;>=$?qLfL^KGVzzp|sW&UU=KqI8lH~;DLTG_rH?08W)9Hm}>m?Y^f6A z=Js8g)+)-si+p1|Wzcmejt3ixLwp9p%!{2us=tZhy0idYIR!OGAsZsT7Qc=9- zOcDNZCbI{(H`na#NRA*6J6Kr_erm*l$o4*285p5dGo&QfH0*%fFY`O+b1*+-^i-MUY5Hhkvq z8uLP?Jd0DnB4({AtB#vwFY=@^ww22e#r2dj&vg~kl`P~#9xJ`>A_|`>rW?Ba;061@ zuN%Nus5tRlGXCYiofDFNaMOn?>+5d>PFr?L;47dA4!53dPExp6Aa=ivl_*tyeWWO7 zw93t;EG7x`iqTuSWifNO(<(TWFf1*Aa-43fOq$R=3m}@Fiz5SSsN;y&nBRaDf$C`^ zk5o+HicF}uF1$dw6P$p22z}fbFP|~_1o?j+b=Qvj2ZuYHE5b7MdgT6$ZNDFxCwlP% zPlT#Uq0O?grUk=^OKY?EOQU zTONNv9Lq!Vv=vA-Q-ZHsZg2J6{~=-e^$g}IN=g1h!sou-QmD>&JW%HC{e;KQ*9wsQ z?(8~6TMoD*cU*#Jc2heKb8+Xqp-WQFGHf5f>*k>Oj4FfQ#~a2IJ5b!A1+_?!5h@dg zkXnUIk97jsBR53qe4?+Me^}()7*q@1Jy@yJ5i_YrbgXfcf`edq4@Ims7 zf1LUb?ge=CKy~_E8>|k_b;z|6}(1{S_YTJ&|vqS({fC#H;1lVQQ0{b zTFZP7l8E-;8=vHa?UbJvx&=ztuHd4=FR!74QhG=ScPjsoI90!EVj@`4J?ZDLq4)~V z(k_3U)~(nX-&D~Bekv#sKAejiKK(q{60P{(XUOe0qskVJcC+Spt87k{(8m%scR8OX zAR$g=kr8e3LTgkaoY5|M?G}|Ejj5;c7?P4ysmG9BXT9!JxQsq7G`FSHQLv+~UF)SK zhQ9fPtGJmxl%1G^+yGNjQO#Wtka*^;kCGpJ2wE^)DB@6AaJ z5%w*#v+ao=X!O+Ttw+#L&uH=L3P(+6Y}}7kC90ifd?{iVc^4(k z$IVDcOc2B>E`Z1TDp~-hq8f_6AMQWg_$%#ISOC-a3*PHzZ;d~4am@`n>hHX;3ht#3 z8O_6TQ=q>hlp_YFiv)cm4o+xZ)H7IiDuq*?t&^KkvhqS91qU=o(h?&5Bz8b@6cEmX zFS8y1xK^jlrsTzIM2=)+|3eZ8oR2BYBj|L5OVF_*y$(gw6_|3wx=5m)IXYIIfaIT&r7OaL}KUzO!Rk1sZ4am8!yB$4Q#w=WhANmjJ7fDZiH%t z%#=QDK}6w(TT8bp;-mWe)lkBq_4`siVKxQB?M+cG+*mb+Gm!mO`ua^A+Xb_cQ?kYF zF(c1#RbAoVu{^pzGCmN&)%?pynbuy;PW#R$;Dq)M?GUD6m~!AB>3$&QxkM6F?(Nsr zAAm4R=T`NzYf=4Iv|92NJuMIEwIlltShnXa6^PARxjYAD(T^+0Rp5U^OHBMYWh8#a zt%AEDo)Me9B+?hLE8w0i$M5-*W_3dQt06K1y2QQWsUPJF7B)r$LJoo0Q1bD7%@dgQ z%yi`QZz8fwFv(EN{+GC>{Bjy4^U}<>nh;rRnm3pFI9Cr^41P56dPU?Hm7s~_BU*kd zDZ5@oQ0swSEZuDQ9^-18$f?)Fk^bUhlUlX=9QnPVq+R;}h)d6o=FN`luMqA6g3$Vl zlkdfBobxqjU&{!a2QVUyER@cX>FRoBF1nWDI2;F1n3&RbP!r6C<%fi% zaD>vg;WnlN}?V~|^{Lswyw$T`V5+jq|kkvLCQOz#(J zfm)OBS81w7l$Ui)k+qC2shK(krUuukWJn-zs@jv>f4l*gX|>F;x)Gn8*yZ7UgmS17 zY@+O9y*SBvIY+GwIvzM-x)+(e5;|whv}jBGha`ow2K@es0(fY8bx;E_sHM7%hRkml zb|Aexx_Ugrru5XO^^ojtA|U+f|3(UGac>y0`CH56%BYYu8*{eJB63>UL`-rJ;t*73r z?O)Eh^CjA2NKK#W2bo%)btkj7H>CJ)H3;{PG8H{Jut3({o?cx;3D-q-H@5s1cedBW zejMfOsOjqN5B`X_4BozxA)FC=C^3*tF$C@Y(rRJ)2qexJ>xEg^!tQ>_N|Otojy-(& zXw4?cy7R1`EOB2@CWR=YLY~~6Cz?7I?TYpLB=?e%9}q+BgO$v$+?FiBecx5_5h>W^ zs5ymd+TAo!4e7M}2I8SzgB3|ouz^ybI`E}~_0b@vi?sbgh>Wo-( zbkBb?O4#Qu55&a#756n?eL42nSoKD0Vo>?j4>H7T5*th!aMRJoTa%Bk?KE-N>_8g& zi-XdF&D}2%&}LsBt@lhC@Z~&V)ipPpk;yq3ZVMysRS(T!bD~JcXCKFn9Rp}jxaNWU zED4J5`EVC-3mjxo-1Km4>kv?&Zt1p@^UgO$G6gS*_dW%Ql7lgi>ZU@j%{b|L8Cfhp zZoaeDz50ahdC8Yo{{yaLENB#4`4sViv^~|uFus%=nQ=HOQp$TMYt@}oW;&G-Ju^#+ zt1&-a=|ApVl0q9JPgp`1sM}=>uc2HlujRyVJ0A!~mf2jI#_DxH~%b2_+#hg5>AY zaP^bTh8Z8pC{icMi-n>PxMzw@Nm`AuV-sX2Gxn4T`q#DrA)P1$wP|ptWD)YB{8Xz< zQwY1P1c85%_s>pkH>E`Ax9RFrr zu(i;=_DJ4)^}`PmGqCQIGj%;}F+V*)KpbP+iLgj?lTlimDaV$muSOIq` z*fRYN{KQy6*r}4>4@G4-7cKpxbZLT?k zyReCCwFT>?{QpQovjqk2QSf=l@-iB#1?Co-z6*l;*Bb>!t4jBuwmuYfemjE=bu)IO zVL^K|mT|u^3y^8xf$C|AWzL0FO+Ih`8WAou-281 zIU!*p7jDbHih-5UrN&o5)Oxr>qTz_$pPL`E;e;k$r(Y*&o&GxEy~AN2B)%)<=UVDw zQKcK8^$NV+Pn_pfTpH-G)|jrq5c2gRNNpV$*~QhxVhPwr=I?UK7%qtCu^Qas#l|30 zcZ4TUi-^CL5+a*FdFi{k`PEii#U$ZyCoO&~ER2-9+Y5pirD zi;aDH>Y|pm$=p*>Ls4P&NpG;x2b9WO{Xp$M)=zTd=DBSgQFk`H$|3nd*EJGxxc-4c z`_18~L#lAz|5lEBE(OmV&hu|)Luep3BNQ{p*QhN9LKf@vAh%fD>vDNrdP!Wb+wz)| z&yZz@WeZKuESOL64W+j_Gi6YPm`~73G!8RN1gK(ZKhG(!a|Q z2^BhjW-H^~OMd?4aex#CI3+I?zV;*78%2<*^_%?%Xpq6v+1U!#t}}5la!wnIlOwN% zas!FT9JXt2UenY#qK!nd&(LsJxZ3SEc&g+|@e_eODCR0RqhWBrQJqPIGa6Hw!)~=8sh1`9Kdx=-z|qKO_?>MXbOq0=)mY z(F0?e@K@zM1pwFQIZ-E|HMC|uK1dnax#mN}v41P0F5oBBFT&gHKM8{+%K_9<{TPC4 z2BpUB2f$JLS%9CowO6F2RFYYKxf|Wbld7NZXAW5uuge*S3=s}|=lE(d`=^Vn0C%$y z4dmUTLY3F(GU)3!(IQ?*}bw-F$GEZpt_C1Qjtn&Xp=#HR{L6HiVMq zTkUAjp$}4Q*E91f0nX5%e@GsWq0_};mcwNQ*7Q8aRGXG{rnL!O!5q~Kklmp$2cer{ zj=xi_c`&tc1<0^q>cFF#Tcg9%VIbDnyyf%bf^za#?KhiGGYbm-IL+vfYRbs*dL|qF z2lM%INK%sAvt>U;n_rFTgpR-Cd;Oaf--oQPKCb`ODb=Z=2?dn!X)rmQB;Jxx&lcT+=bk34-FeT|A?|)o6O3Sqd$FSHm#G@jzupUK|K?cFnPo`*{ea%DDC>! zk&431Ez@rNEH{9fZ9L}n%7@cchK8Joop$CJLd<6Myk__Y0>@-Vvxc9T)GR`&u}Dp? zwKjxxvRjk=mr>?^3SaoC1re18&*YZ^!t_ZwG6-}35n0k=pQo)`S9!}{nU;dX6Rupv zS1F!uMo4Qv_`pFn^Us{yg?67RPU{-RF>TY1gRn^1R9(u>(HG;(Q6F= zeEKrNFSMrcP>6`*`55il?<|1-K$6#}r%&80MfF;ANv{SPu0$7Gp7`|j0kB;IfdUm= zY+s`~_=$|uGv`<}{M)U00Yenpd{u5;L#+l5hB*T7E~G@YG)NbSEg-UsLG{dn~cSi*sHq;mV+Ra$C_u@G;Gzpakk9&kFh z_krwdfjjcAUr@<|Dss=qM^*~`0)#Gfr}hEG8QQR7`rGtc53(um^6j~sHZJo81i&>+3Q1=SS9E4+V57hFC2k+0g@!4k+olu1bog#An+ z@>Jz>)sd6m0HS&4nMu&Pp2iGsw=)Z0k5a;Yp9XE z>mHv16HcGhy(+mF3aDkKY+(gxp)fN&EcwSc7b$!X*?Pa(S=bu0Zho^>!*1}%h?Ril zsZhS^+7I-H7=*vy&~6iQ(6}3rb+}F2vBcl`Tc_oKR+P#3ms14axBbk4_rDB##{bs( z!8_aE#Y>J|I002$%ehpoc`OGy8Ydq_-~ET;L-_>f69O zJC7t+M;@L7!vkn_4mdv(nwFb?CGQd7Bsueyp|c*>fl&3h>ZiE-nhCU`F2dMRYgjGu z+Dd=!3TZp^0CcE6B3E~*oJP|q!^#+DU&r+FA8m7vX1@ZQ)T|1x`g9zs+t=8dyu59; zF1duhiH%}7wdfleG4zr9o3^n{;TUlM?G=*IsDLA5MM(Iu228s#l%s_6os_Cq9!Xzk zNF*Osg!Ij0@{jdZ*>IAT5tc^Q*a;*b(>x`oeD3x%g6JNB2_6cTSCs`q7vnA(W@5IY zWgCp$(gZjnjM0Y)E(TwP9(eFFi-bVBQ+DJzGt-AYnxO_=s_(-BJ_=u@GrcV%=<>i> z8j)?nT9mgtD0s7f1)v0Xw~=Gmq)nKpX;f!2hRcm-xSH%t{! z{~Id_%!%#zC00ajb#7|>$ZO>#kMJ*vM8z_&S0MfCTcg}q6O5Og1>`&?)>;5~bc&Q@ zH;l|*O@ZvpX?G=yg_|>uRF_}d&CE4DKNMf`CKZiJQL=*)2&d=fBFv6Y*;(CF z6X^4o{Fy|W$Yw%Ab45|tuL}jHEAtHb@~NZ;yu+>w(qnr>hzyx_dPo8Q@`1hy9E8kR zWELUY&t5-^QPMAPo~^LO5-j9JY%)kUfC(7#el-ur={#s+ulzP5fsCiuQ| zEP7Km*&O|lq|lhlIw8NG_zF3*fK7xS`Lg(<-`1jE?Tz~}T2*0}kxQx{0m3P~humAe znpT$9w+D^u9};&cg{SZ#)t#iwc3QNg$n&9mFUtUO+ow^~S3CKT0HV}`3a;~9GAGSH zA+#)GpC^)&xz5s7f7*M(HD(fR*Fsy3TP4z6$ z0$m@2uLW+xja@v!Ylynt4dhE9trEBWm5!(Fw)B5UU*i$xaWcsEe9p{}a!XT&YJT|V zV7YhfXNEf4b}DW<+w^OAt}MA^ky=#~2ZlooKXZ*lLU5i2z~ikgz3HSGUl%Y}CHBSd z5|D$`7JZpa-?DSLd|s@*>aj2hB~E=gX*@|K=}&*7ldFG#B`HN+??yz|(8OfM(#jzX z4)NiAxGpWQQ0?3=l~67&Xd=wXx5izm{INxRqSNq>&&=WMQ=95q<0htjYNSgtQky4h z(83Q74->i9?4f32f`iF#i;_%{_e@-B=a3oK2Dl^#%K1-)AwY8FkR98QAVp5rqh@=Q zD{IBOcxoO&MM0lgs@D5qSfh8u4qO9Mh{c;DLL8F?t=7|)T$VV~lK+<|M?VL-%;rML z;sf5YR5cgUM-&mbA=UYea${UY4%_1U+%Ss93D9s07~QkWVN+04X6Na=K~-3)PHx5| zwcI_qt*x~F)0?THD^xTSoWe%{^9#1OyV0zp@n)pE-|`(;C>ORdr;X8Yyx4K*5MeeP zA7_>x&fL8m^{&cKPcW5FQxVw`d1hS9%Z~inr`Gt}VZ2RnZdoiV04*!4Xn{#79x&Bkw)tBONmCK1RSZ{@N12#FF#c+jrXz|3^vvh47f9sT#g zNBn2G#rfLv6BDU5Q^4z+p*u25>HgLU{pDIe;OGzxI?yf1!qa3!9Rme1z4e;*jGH{~ zxhy&iCY6Q^E=`2Ae^XW zPQ>L4BUa5!sIIKOGUbUBK}R@wk$-Fou+j+TjkC7zPb4yK-8O0=t^aAgU-XVFv{H(o zTey&fI{bMsUUb#vM7_>RXiPYzmG<+aD|qIz#L>Tc+0%3RIdcIB zgK#~VIWheiWrM!C!#rIgv#mgmik*YDxrETrbs)5AA^lM7eg5j;I!DE&CTgafre*e& zivM9M_?aLBu0NDLUyJX-CeG27SBShH zHN^1OX7;<#u>DBYGD>*IVJ`67E^Z=ZK9@wJDSB&(7Y~n57Dh`?#Nn~ws1izn%j6=m z*R*d8=s2Z1yfzC z9cJX>xQ~{`+_+XYpdB)aYF2k}!ZiI}l?6mW`xjqYyUX2G*md5;S(k^MhD_+Hz?Jn3 z^$bpV7I5W11hF^T`e}R-xz11P6z`MlPaPm6rV+M6HMwtu-AmNruJ*l1ICo$LXme8` zW}$-%3OldCTLiNh!2AH6)#5#$Nb!|>*+odEpj8W;hu^qQxN9?6OdIOV z^gd;jNDsvQ;=#E~iFw3alDMLMyO{3xBRAJZq+S1G`|8YuuK!3tXTDeN^y@C_WNmen zkTV6^$7v%7JU8XHFz}z$oh|<$IggmTV6p=ifayZYFOfX2h_?~qj|*N#A2632iUII7wVt-jxn0WdRLD)gtf|Ayc_r03v3a+vf(e!L1$%?EQ z`I;3-66pk#J|RVF-H%)LEVR!l&rD4^-%B$c{bGx~&7AZYw^+T!`;t@+r9tCG*4|&T za>eM|)WeZa_NraNZV5R;Qs+W@Q=@)>G2Nw)X5cyps$pagCFyCT&y1w%KTSDcT?(R6 z+2KDti!d8B&ep3XE6@?texVS_CT#W)&f|_3?_7Nat1hfO|8Z+lNAMPZ0A^9)I&NVllakVQ z;eHe2`)mED_uRk2qq*4oUDm7W_SRUyr^(tCc^;U9XpIcd*5xIPpD0cf`3gY6p?}r{ zKtJ?7YgCDkY@ydw-^5m~UN4MipnjuZnFX*h;1>NW)*f8VUnk~!G(eYT#a4C zyCVC!`NvXi@MQ9WsH7yjh@bmhMuf-Xgv;~nGHOPCM1c2V?!-?J+_T>~HGLKDGn!c( z`A!-$jkLP^FU3hUQvOIH*fre)Va=#8?x5#E&DVttua?*X$MP~c+bK76%vw~ z%EO{)SrtjHM-R4pYD4h*(Ax|&&kzUJcnp2VSE@EV$)-}45|nT`k!~Q71${3}IuZQK zY=l^n$wzZevpH~79@fgp?UVz`%0rD?NdCje{YNs`r_j8MjG|8R>$2nOrqL5jw=FqB z9g9OCl!lqT&LusTT;R?fI&~lfyrzMUNdVTJ)ZC8WUSp%O(l;jsU@puobLJu&$i$h+ z4tU+Lzh08{!zCj&qd2lmI%6>8NP^xg(=`*#G8FPSsX)brWKuGohwij^Th!o@&%auW zGK3q*CR$)o{8g4;eDbnmOA}3g`W?`N^Ywkkm)x0Dd-bdu)5l4lz@KSinG6A!2_m-v z*)C;NOI*&cHXx56L_5>eo}$1_A-%{&98KW>Azf?B7PiQQn6R*wbnd9H->ci&CUScD zjJw~c)i{Risaguf;s=nim%kZ~j}$o9rLS=NCI3^Y&;x7{z@#?Yf2+RW3TW?yyWsP(!Y@az~E(gmtQ^X*#X5nUr z5@J)8uRM49hJ$$cfr!U$%f807z}PB(sb2(Nm?P;?_bQPg$c*oV0LP zVdsFNaIy4>Mcf>3SQV0aOY%bPc|?{ERRkmTDh3*hlx+O`sAf~JU+UC!^F4z2aN2oB z4}h?WI<|BSr2^=VneXwHL?*X+UrX;?6_ihP8KJm$oR{;wyojPbqvnEeN_&iNzNDu} zkYEpPlW34J8>zig`c~!&VBujpQcgljEGaZ^g5J3J+u)v1iLHokK29qL2lSN<=y4f4 zX0?rJ&!)UkZx9Qyf$2)-H9vFz#3LNxahU*g*qDbbrV@?K646ARQ5xcrIGDl^7plS~ zgA`VQ_7V#;k!a8)&XK`bCddpR5vl=qLS5Xj^$0W1UIe`m#cmkRJ{0q`zrfhyww~8y zXS^9RDp;M&GAw4L%#x9z$}v0)Oi0rn-3NN5A|v8=$B&2`l)!@IJ`~Zz>1-RXugRe4 zHdEhK@|VLI%c&vqA^7yPNMWy@k3wdbB$JFZUhh*%^Icbfa5aioLvCJhG{e+#^L8y| zO9+t4o$*;XH!_A>9D-DAC&Q9f&6F*f!l;?B) z-9H8|$=moH&IvbyAYI(E+r1Fx53LZwxrLzR<#8SAco`Lsfz(Zx6}X{eI{v&FsCBj? z`0R=dk>6Ivs=EEENuzS0OxSg?=8gt*PI4Hd$74$BruD^R!8PRs?;GcMyoj0&u@?IQ?CZS=d`F>@{Y%0kr$}u|4%7R*h1vA{$R##4@PxHryn*EpUy5Sdv1Z3`vURI0;;TG~*9mg=e1^~(*M zdj8Ytyx|ms(J#`H)U75H@a!-Y(;BLgM3Yq>w=|H6+nCtY8_Jv>} z651`wsN$dbXf{J?m>BlD_PFo(dBVT}7d|o=H4a`XH|-M$udguQ?F2%|m+blanFD?8 zH~%5M+gUC#u*O&9dFXTYzRr)47f${(l9%%uYq($*^Cw(ZidPoEf-P#;cFn8B8rY^S z_l3gd!OL{~FFUE;$xA$sA-T=uytj7E8Og_%a9KN+&WUB0i0mfcX?8O9FR>+L$E^}n zkH|`@_E9rg)xYNEzsJF^Fe9e78HMEVna$8wqNXcR;$|WxT7M&6Tc0M`dDwZB!yMds!?0J@sgl!X7Z=dFMi z7IH(d?_C1A3eaUM9XN0Susi-o=G>m9m7f_ca#~rk%6&*ix9bM#Kcro|j)CVVd;kr{ ze5CJ%W;5(O9|4Tq&xELD&+mVX_+J+aG07RZ7-&|(*iEL-MmmLxwjys!_P0s(KS#-9 z@sT#^?iYmp;1^cf(J+^H-?0<_VY>g^I5a-zKR5vul_4nQSj*& zSdeWAeSYH+I`}vD0dFZ*v=IeHQLOLfdI;g3XD-vwWrHgFdi^uF)Hg_;ok}l4&}RX# zH$A#2$%yTsz0T_se#QCM7e_l{dwHFaUNc_$u0mMCUTZ=?vV|@spO&)dLDBET`}glw zA3I-Kcdg$eO+w80{sS{d+FaYmx*mewOiAtEL?33Lwcw?1<8l3lImE@FWr||3fC@Cd z`;Y&~ietg3O4pIjIMZ`NG)Ls?EA>Qvc+E{=bd!^r2(ZjkU9kqaOKCW%-Wfbk`N#ki z<NpbeZ->!q9zn^*d4dQNfn-se?Heh0B*n;00i|_Qgziq8NmZG$T58q;3@w;SQjeD+L*xxVZ5g zYk>(OfbUboF0x$TV%nUsL)u;kKW(b@g*U?SaPEpDU_3ZLT@no+4A@emO$X1ZPu2XtC zZFXQq1Du1BQ?O=67n~@vd2Svd3G~OR0(5Rk&}EljQu!aUivYRM?_I+#9VWAhz+4fG zbBB?uUCxr_#E;W>;Hdv9eQs(yvn^N#!gRvG<%t0*jrq&R7&XZbIC=p;fT#ZtsgjC#8|)CNCAOXEJuF{4>Xf|_ zFr4q}jtQBh46tQKJ z4UhZOJ;#Vn4eEzj(F0pPEw!%U>v5+s=Y($Ye8U0U9okCJ+L3{8Fkn2p=HAyD_>fVAX0IYkyY872z*1R5kudAKLMTa+Gg-t{uqa}dIx zm~ImE;caSq`e-KUp^kNc$>B8SI^pD!kCEqJv1}}&;Xk+vVC9^1Su;!VrsoM^2V`rN zX!C!8mN)WQ=t3x$*zL{1{8qRzESQ41U)49Ke(}RIW5tfc2U)LK1{K7pHLw<`U<#KY z{z8D@UazDB&srI$FS%6eJmMnwirX{bHaO!*-^DYb4|Z24H}P~vh<}KnOE|OEDnP7& zE{d!RqO3QDU5kvye7F-5F|?C-A5YMpnPHK>$LTHx13}0g<>Bevs%W{LiXerMJ(G82n53_1Lr zwiwS@&t~T9?ldG$dyi8T+pF!N1S6SF1MD57n)|L{O6MEGvM~uua%9np6`v z5}k|;dCW1HMU32c$y8mdL42SIykh{!kBYngYU9jj+a#F4pZZu$;HRu-ZzftL@R4xr z!H;tX3kf`zOWHX@374}-C>{+xZe_Qxh9nP#OGe(cSbJNeSO}5lbkHtLY_ST=nX`r6 zGC##`5#30%8TG4*z78bi!xV5b?Ay9WX^*jaM|63zql#9}0SG8h4oOp!@AOB(?0~vQ(b-(=%g_@VgQrwT~-`wdV>X@ccmS6uva((rv+-AM|7=tH1 z-FmmJ6|~tpwLhzZ9*x2#8IVY(|rx!3tY?|ukVb3aB&@*7B$-?YCCvJCbvpLXm zq*|#B{I&E8exmmh$ml9a8!Ua8xCx*!wau7i)`=|ZfrOUapB%AgV^uve#m%Ek!gF1I zsYLshbhqjtwRbel>F$k24_y`zAyyBuHk zUs$N2qze&>(mH&;t{^{G?knW`JCk*uwjt0ZwqM50w7}i+#XqE9?c@TyVHI@as&el& zZeOm29?N_D#=;<{h1aVQX(?>LKnL@jNaZc z`gfKx#oTH9etTs@6ybp7LFT2+4wj(Ez;M40p41eS&4~+#o+XPqF_-~mwwbBCbA_at z?+xTUFxB}KT84Kx*7-!2N%2?diaVl4KRPq5q$cDgJGNAWTC$OhUYP1-CP1=AMFblEzn#|a`u|EeG zsBXhYu#wlWr^Yf=7CHLVcf*wYX~%MBvC{l922T>+$DX$5ANSC&qmkP%(?z` z^H;*>$BS-;3I6~`Zl6(J>20Y?X&tl1gp#C@_|?h##!p0d8KX7K{{X^6aKnhe{3^ND zZ%Y%n{&cQQ&bq5z!41SEB~>E_8SPxdYWJs&=kuzsdvN(JC+cbQ36*a2iTQG`AkoycqwNYXT{s}sdsesD zn_puWPtxJKlgpCeMB}F6!Tf7V4Hr+5*UR52{v~$CKPu+$r@7Pa{>^oil{r@fh90#o zuD&6;NeqE-K#&uTJ20YCzQsoN)bzgzSi@zhUo3H|Zna;Pey9Hctz3-SR*^Vjuw~TX z;fUQ`CxtbuZAZkDCA`Za{?B6DPu(HM_}2)x@>uGvZ#f{OxNvzei zG%SWsCoX$+*%xKPQ+Of&2~GzK6YA_mLN|cfBP$9Vo>@DUGAO zI~-GSgVL!GIY$)iQZ8xuHCGD?tD1OPMHL_i6kKTt7MEzH$28zM=~7ZdF-S#7MkxrT zaoCGJsVT&BO*x|Dux6beT2vmS3T7%>Rn636{iJe-WJIJ0Cl+^*0+2?Y?IilzgVOUrnyy-_6k$*VL(717c8 z)YB$v9MTiTW$l@#vqPP#7c?_eY=0V;QIC*g9%-cUK|ZvUa|bBGYPInH071Dw^u_-G zm039=t6v;gT%Y=4{{Z?4rF}xjk$6AE+Fh4}@3hPNjgq@aBS3_b8FBpU8o`WTWMW6$ z1M)S=fr3aq>vKxDFK-xQ%x+hc*EOWphE()y&Ak|&@Dq$U{(t_e=)62qc{KHzL&n%f z2ca3oVCr$ScbAIFMpzzw4QlAO2#ajQ@Ur9i8YHwshs1hATIiOFttOvhp!bmf0Ewrp z`Y=%O*=a$4ycJD(hOwmT8f>3zx`Z2PB|cep$Y63uA4eCw0nmzsZ$t%&?b4VCnobdXDy znnUJ}MoB%Z3r3X|3q)?MCeiIybm>cY-sc}M2_XZoH5R;?P3Vr-!|ZM~Whdn;9B35% z><;JCK9$~ENo#j})kjcGL_qJ%FKKUlX|W4w+m;6!n|u6E|UzQ<3weO4*Z1xR8nSc8o1uT7Yjh091} zEDlLI&lSb~;xTesP^=ef?QCTFd)LokDk)C0?*9N&)x^SFw9M@;19{^OEjI1~P{${x z4u7S4O^kc|ka^s4e-mCu;P^!TGxJKX?jUx_JcEpJ`PZ}ApjTXf3jY8~`YLM2kx}Z) zS-VvLn4pjU(TeaJkw1^W{7{wL?2eHpAhR2Sm@Ue6v^kZ z?1(FKxQ@StG_?c6ymhNf;-4Q$6|}@fZqqs21EB%)P&DIPZ;!~@6A{yq9vs#nin0(Cj>{A(mqu`6$+NQ?N*5wD?Y zi9R+M`qE=PsNK7f(w&?P4!-q)xPeA6J9^Lo0Am@&0fAn6(iQ`rF-!-jsOR~Ve>!;e zrcIo1XcH^46E4>s0f0^_SR8Na2*(sKTaqoaU>)3jDpY`Q z3m?Lsglz*f;^Wq_Y_XyhQ^s@LlU1+40RZYvTimCfO;Wy)5SZYM)kkqlV}IdsAMlyn z5Bhg+<}2G)Kny9+d1U_pc~eeb*N4 zU0CVP?Am^yg%z8sliTn( z{OX%m9ke)il2wmm7aO_39dXj6x3avD#A1zzk&31}pYm%`{>OSftfZ~wvzYyf^gmI^ zKU%Y49mSMZ&|Ji0mJ60r3FI6ppb|tf={mNX8qIfUB15t(84S-~$iqT$QRPD5)(w z_I(RWOQQ(2#kJI$ZRkFHXXHQM9Y#I8<&b7JKg6BCD!eL%%uL%}J|&Fh zKdAZ|xdh2^3d*PN4u_~dl~P@aM40!Z6k{CGO5hmgo&_{@r-CWm2dJq4rUOm}i;lyw zOHJKKWKq?CskqT6T5c*Ma|r7)5334|wF^{Mr0 zupjdDhw05NM|4^}Y0~vb85?F!R z{{X_c+t^x2B#aN7au54Mn$v#}K=34fMbm7~50Ky1s6EJ)QI0E88uk@`Z9#sZansTfgohkarzvEdrvrCr<_NFl6 zshiu=(!(EF2l!J>&HDq3ILFpQ{uEh{n6r;cOk%1TbtBT=0rhkK6w=y}KXTlDXZq1* zE0)Ojs@LnCI_uZTfBgj0HtL7??Z^B8{{Zn-M^^sQk4L_c+(JS@A;63MlkZI}cMJI+ zHscIOUTVO&`!1ebedd>g^uhkL$3+C~H4M<16J~G$&O29AS{&OmOH!Ndai!1v^o29^ z9Dhonx6f@9YNw-yA5l@6WscooAQS@t{yC@}Or&Rzr6#Tyd!C2ldzQQKMvW_VBp!d= zHQD$IR~`$p02~7+KkzeNGcK)lJ%ko6;h~4_@zb?iv+)+GXK1#tTx^MRg#dS~v$ z>Nc%n?S<5d;`=oWND?9peg_p&X22luGoES;c~S{?6E9HShN583+?>|-BINdSm(^?| zLR1NnNzeO2yW0&bOw)DA?(LpSi-EfN~ zd~PMNQ6<=MJzv8LUtXR-cOyxS+0Jmk{d|5E-dV<%^GdSVUP2-HLpcCRp5xn(*0|pU zd9qqY&B|>7PvKon>$bg=8x z5I@{M<5e{)o1g5o!yK^ns{{WZCe+bZjO5%PP{JtN#{{XLp{{Wzh?YD7N%NxB2Qe#U_csF-P z@lpLMHVxILyFXdT{#8F(rF|nMu?fqu*uPW2r(}kkIiDM1PYJ!iGyKx3fh5f18@4}Ay6p_OAG{+! zpsz~RbttrrChJg`;yZYnL+_4*@jO?~Yp%h4b|t(s&vi76;uz2n(lc?%Joc?1OtdKI znjzFw-7D-7$e3 z0Q~KqKMJ9z=+@emo%wlLZ(v1`eB2TIRc+boMY|oBz#yZ?ceo=z*^+es0IsoD(Us!A z19Vlhp5swxxMi3y$;&k?bxo2^B8l zkHHxj>)xPdDh4{{r5G$58NeCMFhv{x0Qmfb=B$X^jNl5bctN{}=Zw~Z@z?p)Er6+U zYav|ZeQFDMWJyC7$gLR;>Cy_038@I_lAL zkKD=s0HF%<4~A#{D`y_CKl2srml-;pfF0tO`HHE0MsoO3C7kA1wmJRcaDDyi%O3Sg zSe^+)(T)e+9;;UYS4Ch()5tX}geW-}qo|+*(w?YE#UYWA){J7T2z{vvjNp|YjXMR% zI(DP2N2;p4Z@wS2Sy)C z^|Db$3$A{(&v>IrgHy6;?-hx_`p8OG0jX{gnP}2p|1mKuGoX6`y(GyDQmb%7GN1 z?os?Ke~Ev>u`PUKru~vHFv%3?Fcc<7{{X73H;=S*QMqn&jzEvU=U6n>#i<-)+YJv} zlTFkDI^zn>_gzm+`h9y>M|0u*FH+Jan@fYpzJYga&)&!Ik?&l$iuG$T&&R(KNFk*xJ!% zGDmLa5>H|dGx}GYE!$12J;Ex*=g+%hlZDG?AIh{ZH4BR!S5io3c;l97751O@k3U*c z)Di4@Uy1CYzl%hYW|CtVFmekY_D{Y!u8YrO(%xUle~o@t-d7klYQ4j(u`XQJxFyIoe3&DH9yJss_in zu7#CkjH8UJbrni$Ws8W{3VqEu%I?gWL#dq<^Gy=Smu|{O1V*Bt{*jaZkvBi?{{V$n z3-&9`ogtXzeHFbvhplK1MQoBXY_GO|`T3vz`yb&=DL4H0xIgyT)siWMdsDI%$b6?g zWt;y00KfRtO}=`wAFQr_txxGofu59E2QgdhvG+tjP*3wUJ+_~1Igvb=M4NJ-yBwNY zRCbO)=x{;Eq}GE^Qm}$C{{Sqa2hc`1{VD4R{_2mSim`OKm(5?fFU|Bml);Q~N+Wkn zkIN&WBk4xcKZ(COkkVBXahpBc{`#ZK9{&KHKa5iM6ag2OzMNEHK$s1qYYcSnRyYEp zlpVPPIOJ5xXjERL*0_%2ZZW@u_*L>NES#v3dho;RT6x@|he9)4(mIgRDWp@#;+kpz zh>B?k9`z~am`5~*GN8M+TdPU!-z8>^kmIfhHDu#7(ghc=Bj)$;mx)=~V$m;KoM*{3 zmw%<}I(6b%-dMaBE&*8FVUBuN)V$+0o3WFQwG=kjcQbiW?cP^#9XRV<&bMeEN4*TBJT4ppIqT2< zgaje3e;iJb+`)E>pt+HiJDI`(4(r;|wNhX>BxbTYLwmos_Nol6F&fzwr1+W7k z%1GVpFb!?^iJ@3+S;Kj(zbj{e2T_m4pwO-~49YdHHIxD6h}e#K&q{HX<925_sLIHx z{fA};&EgABph*w(s+YDF_S-(ssU^^C?kO8C-_E+@sCWfF?w@r309J4DtQ~hy(Jop6 zq**<(u6OPv?&*%C`_(6MM%ego?QzE>w~UdpmN-n12jf|?S;nr#9hpJkk%C2bz8WIp z8zof)$QT8Fhn!b2e*@cITtg!5bYd6-oK|$^<$Idcf=M)L>0TZZNoKM!4K~IAbi;J! ztXlYEP}1#h8gUecWB?))o(_4f9VS=vw9@RpU_@p51sML7yA+dZ8hx9&4)Sd-K*o5< zr5U!}%92Sd9VdiPL8eAz`2qvG+ZE6?y}kW}_fcJkhi4d!kGqn3)<=d*eU*4Ch0i0@ z)vbE!<{c{Q%bcL{+anwh7bdu?M(v}ri_sk1`ouT7eZohdlmVRAFM9ZZorDz!pHbSX z_*L2T+s~BkGAPOQ!8rXYO+>mx<=8)<3)9=)yv!~hG;6pnhi!V!r3EH>Ux4gnwea1t z%Na*)@(_CRKnL;_&|kOBX9sr6W9onU)fR=h^m{2JP#vw~Bo4YK>f;DX#qDx=vx_)eT|sqhEj52mb)jO%BcEbj^kA{4Ba$Mv6dvy8=?Q@aEBp_qjsj&047^BY-k1+ju zFZ2B>#D^KEG~?#H*9W7v7w9l{Q`hx176q*VhGq-b5G!^!;nD zF|tU;PVBI2-De(* z&j<(Z=2*rL{qk$lD4Mm3XT}INv0r|bH$~^YGj1C+@_t`$TIDt)Dhr%*nn0bn=A}IV z6r7WcPy@t<)@t@8D%LMdK+w~+(u zn$cfF-m@v){+s7`coueJN>qbitPg+CM9kD>fL|veOGJPrG z6C9|fMh*rAN?VU!Dj3Mw7p^L$?0XeR2a;;>$?sI}^R+$A5!_SQ*Z64P@m!uc@?ZK_ zw8$9hHak~8B)`(UN8!|q4~)0FZBY~CpTfS3@YGP;&lKQ?9$br_pyZ0z?fi_IyiuM+ zm;@^1bk1pp2@sDhSIcw2Z{a<~bT`^?hC)~F9*6o>M7ITH40t0s^f~KSWM+yG*xXFq zx*+xyyJz9pwkyjszDKQ7gK$4L3=R!#*j#b3x^|_u#U?d$DZtNt!9l-t2z}2apNMkD5C5S zFwy)tbiWkEzm`+9lctN z(niv%KU(tptp->n+N1Lo^qVLno@EeqIcn#HBmL^7(WY>ZIH3hhryRn z+|Q_5tK3`2%GO~AHQH$!z>fCn&OlfThgIXJKc!2H zTNIi++ATUcw8dzOte}-;W5GV(=e>0Q01veZ?7T~75RgW)zn3SW{{SYc>#{{-;#LWsM4slAntpKu+MJ${+Wgjc^_|`Y&ox7V^N*C&b zts&x!jM3J$G7}teN+{f*bswE4nuGX&{Ade~?M*z4QZ&&I<5B+r0u>WoLO0Ft3*7%CVVxo*oG$8i8K#!1H_x#}aT4(Pi~ zG=_LrIh6<1g+|vl_~zmv^>J9((&&D8rkaVbZ3q0lLH_`KiiqkqL#@<5_t>J$3kI9b zKy@2F)2aUebQ4WFji~5kKS-GW07@*v<(_JemEs#+E5-VSo%Dt%XMz%v5K4ymh6(MT zQ|V4Rlm|FT$o~MYP5u?i8^xRA9rPPmB)WWOYpyefZr_b*MLuh?DnE4d{{R%~*3xVG z)wCs~ftisVFjNpu576;k&5o@mk$fIF(c_jSNda7(b>q;~&k48GH4Ab+%wmX;cFxhBsqBPAaE^d>LWlc`lXZk`Vw? z#t!V}w!AZ_G#&xe;QKs;I!XJ%_jyczN?(RE%LJN_l>v~IO#Il#TFUl|NT%CqnuFm! zjGuny<&Ureusmbn-7mx1To)FiNhEo85h)nP3G6G9z$p2M>5ppi+ooogW@1YcNhiH} z$A;#*PXt=3{g_3ukIeXC^B-Qk*OJWNJe_$xYM^G@aaTPO#=_M!J8eLrGe>8YZQVX! zPr|HN;w?7K7XiGXw-#X=@5d#FkzOINEmc z)|8cjHqPTfALxk0g)D%Qcmk{V#v6#dEvQ`>(8M8pFmQT|154qsms%VPa~jI68irRH z2aNQrzlay}{{Uc(G6jw}W%){=$I1=>`d23Me8+r#6=Q?&w&9^(6=R<_AbK2B`jRu0 ziMEG#RUkGwHD5`Et+e@7fnS^)epP}IT+7O9DhpFvmk9a^>6D@{f-h#7|@f{?G@P;lDW*& zdKu#O*)#j)H5t^ci+`*K@HMcua1KH*u}wEg1fDVIw<_8b&_pZh|=a)xy;Qy+s4~0q{#;v#TK{r z@Uq#i<02^eLX%w$*%8{l0Nlf9cJeZDM?W$fg^|j9L6>)h?GdAu zKhhMjI2k>CN#?A#CqPe8^t`9sPaZG!;2pAxD_pJTLCp`U4 zOSpM!(2P-Q1<9V~PcQFKv&cXpkLOj)fP!jN#ya31S_&k0`J!S^UbTDe(5U6VAz71< zQ95HJ)m`TuYL>)F*_*>qh-=hR=oMG3zt1Joyj( zE9mcrQyH{NMsB5K{{Ya1eAnT^POEXp{IGxNUq*Z?4B8c`9TjE&0MLi{R*QB@*z{>) zSz{=}gIMxfn51+))PR2q*bX`q%_NAj4u1+0C8L#B*&xq;3HR++A(Lt1v#)a$tfXVH z_Nc8j4dA13B5-~E>D>}_mbYMyu74W14;9NtdI9TN7TTy%101bOYoOEaZBbCA%rYot zk#Y5>8KWP@znujVQBa$x7p$m18gJWeiT?mxpPd&0_aAV7yg!{wRPJve^1u?QVnzTy z^*IP}uz zHwkeb-zYifI3S8$XGTp<5sWPB_g^2O_N)3%xo0JW(K@q!@bn!x6${@-leJv`0KjWk zQ&TjN#NGH#TQu{UWMt3GP#>4yp4CeCN4B=p{?#0kNsm5qMJMB$wua?~XsfvbdWFGXTB8mtj zZyLzc1#}GAQ^+_|QtR4G64{MXz->p#bsxj)S=RR!W5phHvi|^~w5&YdaM=V`T@AcZ zXxfBP$>vA3-4A>f;;@{Os~vt_#Q??W^giOdvm)S-z~j)@ z&XzY18neditS~_brz8IWtzSgMGZG*6NAj&25)U~F1}p$RwL?^q9@qfD^IBc2D_gp^ zj?JT;_Kc}H2LmS^X$+%hc_o~|)a84RQC$-1HrG&RBp%?7L9R(aP!Mzj0=0CqHc7{E zS+}6$xpmz@>-h?-Zi1inQlIz@TR5gN_Z5ps=uMFQq@I@YpU8h&&bytUjot00+PvTZ zN&Rb79&l+TPcbk?YNFBysm@v@+^ojPL?U1uB)J81^zBzZ%c~Lh?T67xt4xlBGI1gv z0q^QR8oLy$G-~R2Kj#&0Z4t2F82y|~ z5u;xox`TcZttim$&s|1O`{aLxZf1zwnOQ_mX4vk*ipmP1L=lhshHs{t@dD)m;5#Z}Fr1 z15fW3f8a;_YfQBvJksZ3$c*8k>Er%g)ct8ji>Lnp&$NH{$gBCJ%`SF6RoJGwRg#>g zx~c3`VyWF*gA*YEit+;vxUG!UTb02KDtVhUNRB%zCW&2b(`x`n0-yFfV0eKs^rqOh zB505Nvuu6`kycGZlO2%(w}f&cU&@0NWB3I~O*GQxVPZXroa_`7Pwt9+HxyhfCKvmo z@~=A8E(9MKZ4-bcvDwGapXpwTbL(Cs@ltuMe02=aNYTo+skPMLj!kJtSsB(|;g#Zx z8R79}s<1}PKGB?k*@l0uU}_P>n##uPRuQlI{+m)GvOOoaJ8u}=n8w-nrrt-?W_J%{jBy{GZy1JI$WjZX+VvK-F zKGDk&TOJu`E?&k9h}EKzJf}n4X9xM$p~2yA4??&)J+?=x=B6{f(;T9#nd4^jTaab- zxBmdZgHqXEJU3}BqjeqhNuS+IDPdmC{ez&!K1~|XANu4e>tUru%IP*^{@AS-4{-7e z3n?#eBbFzR$TM*s;y@YcSbC+*7h0Xe+CXjKX8B{?oDZ#fcZj8e9}-%uP)82nyHOMX zPETR2^r*ni&Iz!_B^fl2R)B3&?704w zB8tKB9ge4{=)O*qbk@_El$jOwki6u7D&)W64e?}%Or9dUCmjUm`d52LB2RRBZVoyc zZG7FSU28gSq;0OQCK5sN;JG1(sr9bc;RMpntLkQ1WPjW#`Iw*L_oAgKMO>9r-sL@a zy=8ci&Z~1g(SK>#k>v#VE+1f4L(zYEUZJ+O8;=BUh&2R55^!XW? zq|B0#dCza>ll8Av_<^iPYieyt<7wOnxaaa2{OiZGEmiF`JIzIb+_us#kpaJ9m1)}{KhG!?WmTz9K2Fn+$be>J4$@uN)e;6c=vL>bER`Sgz7@obnET zTJ~!3Pk?(xt?$e?v2(XP;1nGz+UjAjq+*QY7Ww{!tW9%u% zop3uO>lY2&~v8g8u;hP!U{rkF$M0!mL~6Crp9}Vf^ajXR(f{nfjNREY8gBoSv0| zc9O@n(nkb!DsXG0y%~~E_l;#s7=M=?Y9#D5L20sh_pQvQ)ENoE{OaN;4gRtJ0Pmw% zuMC#5`LG|T{*`{#?JgG`9CZ~pwx!t8SqT0n{OUn5=o!BQRRQ;<4bNJhV6kPUVlln{ z0Pmwmn0n>EI<8I+;YNMv<{v3xXW4NP^^r&BYAmX;I2ou*ZJH8#d8ya^;Z7v(>qYbn z6Bp^gsOD_LAoWxHsR{R{w<^QhmBPj+g(mYY?*Su;hw`oc+}AJR4i^1=Bvbkt>3da4 z>I-9lgHgKC{8J=%(o7IYNetwIr#L?2nuoJy2;CEZcz09vt~GT(w)k%D*6SW^h*=yPu$(B&fS8zN}@2{1^J^^Bx!8B{~Ye-^5*J?4#h7N2n|EPGjYjuu5c z9jEgkcH~!aZ!*mm;m2Y<4QBY0MT=R}TPJ12AiPoz$J6OqbKYL@PAb|SZ6s-Ld2<9g z@{W9_EuK2BKTa#?>wKGNiR-a{m3;pIqxpKKo2bA*`#^`xjQ;?%hj;7G>t6o=Y=6-p zKC!Rzu8Wy9VpCM_;Yp*3BvoYdRFz?ndiAT7Q&1&fRvkl%Ohlec4={!}%?$S{a;?@;v`dX`5bf+{$Rt^?!RryPpSqh!b!9OP1| z$mXOgQ4zG%+QlX^X!jhBqN;EEH2c5hl!ONKR`vQH=UO9?Rpx{;wnkKrSK6J%OG*t@ zwQ@gkA?-nfW=z{Mae7W&c_s;5x2{{WwekI|_<-*5i_JwZOenk>UY)@aUZp}&A1 zyvOK`{uB%Ny9p1{Hva$$8e_UHW|=75huW)uXvNcH$Nt%xL#QDi>MjWH`?{2$Vy=e5 z-d)<~q=2vDIUlBK;XHJ#iLTY7-l4+`I;$=YMOjTnt57N{{TO`fDiYHuMzl}=MNZ4E&vSB#zr%YW43=9-wwUw zD&GOg!914s*H0fGC1kfB;pOHw{$b;StV8E$shUrY8RRz*6IK~@%}aQ^ItYU*0DaNjYIaVk{S1V z)VNEy^gDlu7iD}?r<-e--0V_WSQS&aj&WR_$z^{Il2 z^y405&y<`_BjOK+;nwW_*=ZU3WFt5TM)BNqtZ#)IZEi1_qJ8feQbTU&-`(myr?25& zmg=p+0FQd%d?#rguZ{dm97E(%y-%qowQ;;TTQhYv<~#4Uu0PdGf8Z65;*0Ix7}VFv zk#la`wm-(Sb}l~XtbY`!9v0NFIr1B&OlYLa_RX1p4sQ%T8+7u^e*L)&2Vv@Ks(nUR zQyN=<>@YuiIQOnoz%i)s%8|}ybI;%_q_a);GtJpYmcQ5e{*+3Z+$qiMP8rZ3h8_4I zZDX`xesyEdz5f88Xovl^Km1Ke0*{%8Zt3i+U@l}56gWZnlG7t^6lQtM;|~lUU_cCbe%xwi4`OK*dP6ty#vK} zt*+l%=&){GG451(;BmP@-H%=m>F-tbFAV9D=@Q4LjwOsr8J`MRFi>)N$0x6Ps98lJ zPu`KoS=hb)pQTyDlO*?V8inhY*elWeb>;Yf!V&2Zff-Vhe*$ANq`rEk-c zAGJg?X$3$JpX>>jtz{Y<% zXvb{xP9L2$zFw4z6bujXr)Qp&fN(KEBLk1E04eKECkQ=#DWs2IdQ**{_M`-4w$s|4 zjq(CA2%zAhq7Vdl>CFsecSmy}=s~E$oMUn3pL2le`c!+cIjWAtJ;x~{g2ZE`N{+)jkxmVs{?!ORxdxuRy6`&EG$(X;z$YLYuRCN9Kx<1-1i=;t z1!c^q%TjuiK&7#N!Wlj(LH@&kD)(&@BS4J_v-@b+P{KPb(_sWxe4KO*-J9XOc3C%ESfqVO8GS&T^6{RVOOhgND3=YPtjKDO_E+pqzvF=np~QU;=C&fS#=SNW<~zB(+Kg!Y3V)_w2AIn zoBvFyw;yv?C~LSTta7DPa##<>yt~I5YFg_edqOtcCg}q7-JXBO zzJ>nEvc^8m9zK=L_^-kCx}~bg1WJ=f#>Dgc4$G`TI@OX3{TXol|;l1g>%Q6?*n* zW7lr(P`2YUG3V6gw(oo*>2Odwsm?HKZ(i_JwzrKaASAdWNID`yR(r*g)r-tVbIn@x+T;k0yl2xK+R<6 zC#h;|qo-fZq}~0h6$Q4HD%}lzH)wNb0sjD|Yy5?L^JU?g^&5McgzBOM-M3~-SJ1Y+ zS-=Nq*ZEVDu#+aLm(nW77I_9$ADvnzvo%302yj%#=AhaKJG1NQP{v(}U}Mvtda{wA z;3%xs-lsbo>sM{=lpT%;Ju^@}jBglRJ-&j*+_KKn#;8YQydUH%Y7IVLEligTTLU-) zt#dv-COtC9=Yx8g`g2{hXZLILrnMc)ji=;)u7A_~D?WQNtW?V{)G+@53f`!0A@dO9 z?@-X=Elz*z;2ard{{X(tD_aHw_mby1{{UEO#HvB*%|*Kzu6E9Zu2E=AtP8VmOpF3X zKPrYIu323$>Tttze~n+95WCc5j-%978-&4Ii1{FVz<&^`gqKlCK7yE|!TAT~X@&@U z!iVZJQ!58RK^P;_lNGg*joDn=!JqXJANT;ELkQ^vL;l%{l+$u?O2{t8?NQMXeH2q= zcU8)sov}|N{N|91jAE3db)e%bJw;gwT!Yq=a}(6mNJ_aw$E8a|UC`v5SY)4CPPDki z81Ga?#`8#hDcGbVngZiE0E3P>s}@a|N_8l9{vE0Y;eqww)9u@2 zJ7~UWmd`XAT5o#hHLww$X{2D&u}wVDafoq>=XK8o=$<0HTdhjoN0m?@DJfBprF31X zq~jUyLYs@xhb{L!F+3}y%i#;_^|o)cTHUf1-f^%v9W(vYSemYdV{50hkw62j$U@mX zss8{Xw=8@|Wo4!Ki%GoR9_D#`q8B44J$*R>sp=%$=~9q)jKwJ`axuU)=t@rUIH0bj za}z<+?Jiz_Iuv5070B(JSEXCr{5`aV7UB&#CshH|rbuq}=l&kNa3!=MN)6Ke(VhtJ z_*48v;r{@!xAv5_?AF(EpDs@`l5vyAZuKtit5+SrbecT`_>1A(Tr66D9+uhul%MdK z_!2L>9j-@Bw!i(2arVCr^;?@*t`^GNM=A$W*By^e)L;BQz0-8*?`|!vJg8H1g_x6% z%CDED#FXuGetl0!*R{(#y*4B^@w%ga?`}vr>5qEzZAn@y>53~gs(ZP>}g&$gc;MUuWh8saHT`X z0ui}#0pRrSTKXiHR~i+q<XoG$%N&dBo@k-^M`gm3}MN+B=Jh8y*`qyQn z+XyY~VaNKgs&VWG_*LPkdk89by@iOY?Pty8SVuDIw?O^G`A72pm2}__YR}W%TkR4P z%g9b@+fNu@zQ35SLJ1^rRF&SNNd>Eo@ZtXeiu1(hw&?!=m35wMS;>(H{0LVe;ptF# z-^9#WKrbE+?0VI7Bca+KV_3c`BzRX+@B5Sd>pvMyhI7jibu;?YzMj-TO!@Bk*ybMFa}3pwD{}%=WCtF&5C7$!wf~ zk8V%lS+Y9Er4)|=X1HhIW90(_`I_y19%*sKrRo~Z=>pjmk#qj<%zs>dHE%>rb#@N` zw*LSYX!1!XnC~Ul^Y5@`v%FPhWodb?A^;?k>OuhHYVOA$Q~v;<)V>V8x6-^pq{}`f zh31@mt@n2*9e*0J zVVdO~7GPR3aq|j|cyN1lIjtR2P<>iUSdj=3Mj}<~&#e~J$$6f=@QcGxX&2`GE(=R_ zVUr%*Z3UG6HS4wG-wJgly76u0sL5z70%tq`83q9V06f>Q)@uDqZs)^HdE+0QGkE9M zH4mJ7Vwz+m=N#0Pa^7%%3QgDot9h(&I+{;0vB{uit=Rr~r*7Pfo`*F|NKbl{5>RCN z&;;zmW=#;fl;^pqSf4K)Xor-*^`yoZ2PH^74mqS4Dce1%COB2~%{5#c{VJBkvHiWN z=-6N}Q11Z`)|F-Fr7;(Q9ZfwkxDa?1Pwg1SK{X_IKfORldXnyJ-kd`gO=jKUVNU$j z-BrwX&@cm%Ybt3>ktoJ7%`_<&d@(nX;xv$gLZctjy}v@eitEGi&4Lul>&W!QcwdKr zeQ+xfpvL(=oL9McF4Y^t%84Sd6TS{Q@F;= zN4S8I{;@~rO-&Hu+1vPWqmiVElpZ$#eQTvLC-)cft{cO8m8sOkaZc?n! zyUuGuxg7&oMp;hmuKXJ5miF0{5i)<@LYr#hTNWv-11Sfgs_m&+slm8R{b@#_Xu~3O z{OMf`qMgbp{Nj=qs(y42N$oK14+E|e??xAtq z)q6|Wl?1ME5wcyU*e=49BWkhg2(DXKvAxpne7M^x-*AEVhxpfF;tN<-M!B_-Ty(OcraqD~gt)P8^C zRIYv_YPzBeW%44)3>qangAPVHt_5Oz3}^KJ01CaQTS{-Ec8!EUl3yOj+OXwia-Avh zF1sgPPUrA!QwFuG77W+-B>MS=pZ0%2Py04;h326Tl@cZG)$~gh&BU^Wl^Abwd*k}o zy+!`1euBInI6SD`z~l`709y5(6IXe({zmf~6a=2atu&A&cVu*QR%iBvzuj7J)Em03 zK6E}tKK2ZcF-O+2pNi3Y*L5H&jMW(!JVxB7vGl3UQ=3nfuADr@;p1wa4u1-?G}}@$ zv55C`{x!0wmgY{IZ$LQbG~9qXieK#8KtI%j^Z4rAGHJv-MkH(qi z#5%~M@u_~pX1y(<{{X&d`yHoiXn*g4OWPs)7u`I{fnZMdLr~X?5^ahW&+hgxB{{Vm!T7UpLQtcV&Djp-qzpEx>wfqdc2@n0V zKjBVk(W4V1a=2liQP|d|(da6@#7ag&jMh`D&!vs2)8y(#x|Oq_Wp*I{0I+|;l;1)| zaVP!zZ}6$1U}MGy%5hRwF2&iE4fI`-Q}m3}KHLxdvAFyrtJKq)S_PR*-ZyJ$$o~Lz ze~l;Fq$Fa_fBbj+YUM{GZ9+#-GeTPoOBkOiJdLSXvm zHJ^S=x7Q9x+|sc>T-8)B1>AQLk-jnr2c>$hDWlHj z-X`&sx_n8ez$dniL~9_n&rA${6wOyv@vX*{a<}j@S(N?h@)Tiywa;muHP*C!IV`Pi zBTIB7fMW#n*P3njj&;pK&Ra{1z+jYZWeB4q@~pWZHSCL0ibE{*$*mIF)-1DmrIQ|~ zBOk77GUtg{F&l$(pIq}$h6yZkvw-Mwq#l^4^%&8(sqYA?yw>bSX zUW=yNs?V}8RYq_#!0FIdGfFXWiL@J>lV?{3`5Y`Vr^r5K--10eTiRR*qAaM+2T!GA zMS0{Mh26IdD8h_;4mqvs9vO4lYA7PH4c^C>c-=+kiXgWE7)x}DcI0QD=U#(l7+}oZ zHnIHp{IcTfA~xc>kPLkoD<#X!cS2~Yu!oolxOxX%g^ z{{RV}ipc)}pM?Ja*{P{MR!fS|>%*&- z=HA-gX__Q?P67lN=kl(XTkxt2PYakCZl#jk+r-I@u6G_l{7q>YUD+9^w9YVU4`-&? zNUQt7G>1O_04+uNyc|JNvaxzF~_D`gCQ)l%n1BfHAIUsYOa&X z=k7a{jxp;^qBb<|H5*CRKeJmjBxiP8Ii}j{2F_U}NKvF^+N|Hf?gc{nd_YI#-^VLD z?%IozJ&knU1@MlQt?Lq|wWt^@;ET+VhX`AoU=v3&h(702@C}*#TdFf0lVnLh@HMZX ztAOw)hWsI->bDd4cKS)@k^+(|%exD>9ODDNE5D)Jpsaiu^5?G<3}>mPgOkXkzqj+K zC~?3&=*r`d=S?RR13<^*9@QXXBOL}aP!s7=Zp>rUW`G~RBN*#aAvhTr6tP;|GVe=y zSx-W!O|?o;QG2>h!?qPmIOx%2myTOEDsNYqB) zw_{$$d?Vm^RnE^asLtQEHc#Y5G9M0jCBnD%dhIKpp5{V9^gP#_{ilcNck*Xa!%MpR z&n>>ahUO^I1AM&Tilre@0NY-`G~WP0#J09+AnL6%l0P$CCy#9O>n&;}(&0$0B)3UI z#!d-2=bGMzE}eI1C#Q8|I<%Y>+eQwDfvfY=%YWrw(-z<%_RzR;=|&d_gz%pNFsFO~R&x+xK`RDM9C>bA}NM14g;b$4khb#2{BHcttz$cNK>qe~X4j~_lS19>X8yy%{+6)RJ z#U*WsHE%L$Lk0z=MMRd2olE3aR@)eJF7U(6&C*&vIrM*f^ zLl?qwdDb&KjGr@%{VSzM#~1>+FAPZ5)=WyDA>8tMuhzPi=LB<_Q+kd&5*bJ|o@vK4 z&T~c&YMYu^aoSEwjMTC|oYrX@Fn$6bqbk=m|wR5$;^lVAOx2;qr;~`2nRy?Cd*X7Z#1>lUzIx-fFwGT7mxnJX8MwK^4(A z7W6Smq-1%+8Mo13L;d3z{{RZj*KDtEG+VIeA1BKcDeu@0rn?=ckgY(|?s?L9e^Yf< zHm))O=K`l59lz?(5%pnR?%Kxj9#>7{uJKDiZ$q7*LAZ{}cVLPZHaRXh;<68kFZ?~L z>9;QgqQ>SRlcNj9|DlCz(WMPm; zAdbedace`NO5h32K7zF66&5b&V=hHE6=TceH5;fL z-3_4ir{MnpbpBM*G1Xi0qm_p;5|)!AAb0eo{0&bdDM<^(D5mEfsWWisz4U z{{W3OhCv3JWgefcA((o8w7s1B2eglJIIxd{^r`z*MwO4wu{T_fM{2MPte}Kk5ud;R z0M@8+)V_o1r!;Mt{{RXP_!s{GT9B-{=%1OT$|S?JF}h-$QeOK%T0t5riM<@1e+1@|Y zKX3L;c<+d=q?^JXAG<@#poQD7On+51+eTBATDi+vLNZz$RJuZqX>6leBnDEUK_rk3 zSl9HeH%*2U5R^#H!x_lH{=F(2`-u(QcPQm%Iep4;gIn4~v+8~p*0n3eF80MtE?0oe zesAeozH6ppoNUq0*`E!IW&3PelTV+U$t$#u-D%e{cx%XzTG;ABVz@?@RVoj-0=-Yc zJ{Z#VPYc`K-Pxp*J2++|(D7S0z7Wx&)2^hB#xThXupo3HO*dLMu;WVV_dK(~+E$mT zCNC|cB+d{Q>T)|)72T9}x}28#RP2{;13BisC*jqFweFjFHP+qH%nWSJf~Rk4^UZaY zJVSb{NX5$neLiDZLK0JxZ$@27#nXkBsPqj!3tM^Z(g@>^H%tjt{^}f$%-2g`%$kj| zU~P~v&wS-;%j~c8+gtfA4EK{;E<*1dl1C>WTJQWfG#5JcoUlGnMZh3~gU@m8S;m`$ zIa|0&qLkF4cj)FU#=xD!tG7?5rEEY#i5T<J%H{HS zjP${+DJvo>%bHg_^F=K;jxKgF5u^>3&U5nsE70R$u<}PI*1Z1!!OQ0B`lZG?1VDc> zYfr=;BfHZzIIXR;n~TH|hbzORw< z$!=I?F)WAX_3cymcgK?J`ubQ4R7jZ%d4Y~{26}#Vx}@4LbuFVuqj}MRlH$0J3EU$1 zyTwGFw+co9Ki;mK=eae;cv~mM9wh$&uYvs9tJ!0qdcZ2$=Y(NGuw`p5Oa(2DoQV* zd&AxbwAOUXSGso*k1_sJ$tAJ>0J7$^Fa8zj?;JwbXGY!gGjqjg_$n<-*xF+$0>dC+ z{{Y@BHva%xX-Iym89Z?U7JP#Lz^oGBJ?WA(z zFbD&K-l6bJ9(CQtegGY584~BBl*FR6!*q(E0XZurDdf- zEEB%gmOxQeOL5b!YYWO3dzn^NdL!1p9;zL7<-+{p*o^ybTJ(zGd<$;2+Qr0SNN=+S zIrk#F>Zt5Qm6`BPIG_{I3YuGKV$R@xwFd6kp*x8*_7ro)1L;lO&q@^YOk|iNsHrpZ z53NQ+4x^7+Mq&UM@79pdW$?$1BhfrHVPZ7N?%i$t)-e&l+-l(Xi*Vj{G%My;ZOhPY z9-TR_HjehjYg;xN#FkRWZXuBt7AL7aLF98&CDpSoPl@kvI%3_w_^R`&)XOoYuVn2vwnolvj6RWh02Ie?i z@yB{mt>5Q#UIo4lgNAumjym#9H-N-_b>dvK+TLGyuXys<=4~z`h@%VJjfOGutMotj z#Yd=XPic>{L#bYF&-$61ZR^+N6vfkZgTxI^r6=}Isz}{qsb{CptswrD z=zCgU%)azLXFkRa&Ez+JZROvE?sC#OxV#9e)w@uG(vJsKR2M0>XN$V3z*?3dR1z)HG%@X!k87 zK;^9@Kl%ynTE<~#W68V|rj0{fTSvmPE&~zwPpy3)4EGR5v2Wgp0_|==1B&4E*z{Yk z5XY#svsvn^m6jX?-blts@AVbf$$hrsMG6QEKsBWO>sVK^v9Ae^f4TBW-ZFocJ(P~1 z)-b#|7yx=2)V04`j7M{T-J~DZg%(4xE#hY3&MKQnn&=--Nr`bU4d^KsNRK}>u1BEa zmnlh{+uN~M`Faf2k`$E-$BsVo_XqqdPU^xK?4|P43>5R;xi~IlhVOd0;PlS~dQhXS zkefAbmnBK#s9M3S4}<{3V^SYG3lC?D2SDq zbyQ>d3a11mCW>iqBaDeugV2SmcDihnw%;`eagk6fgko2x@h_*itI$ofhEw!4JgIam znT@#8W`TELFwa#q#nY}`*a#Tl;Pw@7S=UCLpj}>I8Dor<2cb1nE5WdWnzpZR8^{sm6{$G8ifs86Qq=xD_pvFt8(dE0n?Sg+phNW#dnNTWQ44UP?T zuqaD-kbepR`c{Ue4s9f0IR|xT!??j7@4d8Vk7|~z#^rQLO*FaoG=`PYA{xZ;ApZb~ zZ>}5~*(S0)L4&5S&PZWO=ucyT)thdUsDsAqIWdo2rl!_3`R;rz;o0rw2v!)wZsUuc0|Ttk#`mjGi8 zoDx4A{&no14sN1-8EsvQ;O>aG_fKl`-5W@P=T5$gR=u0x}TF&D(ZzFljk!V{TL0O|&PhVwxeg540+79OALG{{R!)UD{;-0IuKwyzVs> z!rbY}{iQVckiRh=at~Uh(^p1mHgtg(7z#1;s4gwlmnxtUQd-U~ZPH}~2N?j=xTe+6 zlWCaE7zFSt5}TBQD@x`qr*Iq^qz4AMq^!<}%IK!>t0N%!Nf_Ka9>$@Kos_GRBvXbS z^>vr86;^Op6E7ura67O609vjhdWcl|(@GKEr5u1qAorlwgb-||l;Wr0QwbgMNG{4( zF_dJPM<{vs6=iVh8G0T?Ikqz3vf%aRs@cjRXSZftAN zBisGckHV@8yKw8aWhXrWs|hkiFV4yo@sfDWbJkX8O*Y(17^EW<({*qE06mlb`lg$^ zfBEPS&=XkL(&CzF=}c?5aBx`t5dQ#$M%Qum0$=;~-{DX)ElZ6u@7f3d0MJc%6gtJ7 z-+}FPcXGFo+=pC_HstsGYu$A_k^Pl;1ahC;Km4;l-8J(B4TnsPxM9}1u-CJ(&xy8| zH8o8<*+pt!YZl%KUvK{aU2ADtWzMGYrkf|1Z7vVlrIogvb={9#RwVk2nvJuEQMMTi zZvOyyfAy-yg{WPdsBA5fOS-ZsbAWTtCbWLjk#{op9Q=;B_&3EH*=@BovbQHI1XH+F z&!siI3F0>&z1K^Q;iYq$+J;XXc`6beA~gdGl0U5@f5L9Zo2?Z;1vM_TUiL59KE^Jg z;R}dohFPvH;sID;NvH!)K`->r7iufk)6RN@Z>P;O6N7}-WY+u;U(r3K@=Nu zqXCbv>MA0q8g`qz9QkEC!C4)vM&5EZ!W1_pf#;6Y-EKxr4(zE>^5Er3{{Z!UYd2lg z=CqF8Ebb6Y&&o`v%L~qW0nTd&T-K&KS&HKN)>e@J0IRoB7-OzZdR48AoABDZ+q+jkv1!8zA5}p>cBpl4n=RWLzrF1Q)v6r#llGR7R>(uAf7M-FA6jI(He?URxLrK&2I!Se+OCf_e-WdFvhn} z0=b_AD@~$JBV4qBuB9Q=?G4XRJ?o_M7yHlu0DT(9qfSo2N-vbTr$QyVQhw_E#=0De z)n{(+D5u8KE|b}*{rDQ;!_NR=YetX;^$-Dd~>xxH{6 zA=WLlI}H`lgred!x|0JLU^ip){#DobyH0|C58hl{&v_}fhDjq<1SoB%2c>w=i|wwY zxsB(Uniwph7U()El^c)eT@#NwUPGrR6z$mSyd|qzz|-E(AVoH6?s}Y=U;hBDTHYzu zt(MjZ(UwGz02v^BwF+Kc#UXKZGQKca6P#AbBPYzx%f%Y(lghSjbtLH@-7IoIJ%9S%)z4|5 z+-r7mfxNu&j1K<*FygKFdrgy3y_)XeJHc*(NlOEk=K)52{{WR-@V1qAf8#4-6b>SW z>)-z{vD8?5D3jl|4&L z^uG!GFw-@?0^d-%y`DK_Cn}i72a44H0EKVi;GErGvV)E`f-ruS$0gRC4wa{^?Y;EG z+nuJ~ZG-POQHqD{x-(-MR;bzQn}hj%+Tf{$#!91}EKs@sP@k?z;GO>JWu!~VWd5vcI8w1pSqN-fm-c5L9Q5!}h2^lhD9COF6Y8^jCeKhX3iWrou`MVdH~A3VET4Pkasv_2eeMNzM=ipwpp8DaTltJVOJ2OuwBaqj z^=TMj`AEt5;fWQ=c+Och4NAu1PGx%tnXoW2b6$h}kE&nVnC@*Y5dr2qi{@Z`_-s}d zrKZ_Gi3O$I)NzY_J|+Fp-l2iy0xL*iD9~^&>XyengGb@5UQHJ29eJ0N^^+H2S0(qr(XA7qB=JBZp+`+*(6 z6Pz>);_}3-bRxOEn6!75S{#BT6q6dd8 zeH3KUt?AuVXqjcw?tI*v{I$F6N>;a^> zSmz%-w90*y5my?#yMv9$#QybW8<4&0pSfu)q)>|{)+7XFBl0z3AGgYZmd;J2aCZvY zHFPs{&f`ti3^+$SleqNyS4#x(&lq^v6*cA2CZ>uS`aMwF$Z>i~BRraAQ_HQ6yBRCsJ7(KzR1uZUQ)Fp?^Fvq-%u-%`f zL!P5hx!U-DS45la@;D0pIp}Lw#5#-?nmy5JR5rpfp60mgeI^)OJG08;_?-U$N|(g< zA^bOTSgObh=YK}Y1M{LaZtOQrrE{m#wM$8Ki*?xSJg_l}ge@86R_Ms`3}uJ|A6kpTJ|om^Crvk1uyqh0EJ+y19kcY}ri8t;f+(YRBlU{uG7$Yc_b+8$QokZ{klBG0D|P z&!z|R6oXvxBlXrM!T$hnAJVl}T?n_4oBf^RIXc0rpQ&nseH+DLSj%sJkM@}V0F7zO zt!w0!eO~Aga6pjOP37jTFAu5Rulq$RJZZY5MkwXLiI`x

EQcoz-yD%$1-75By}S!` zqFo7*ckVFm*xfJ>_*OXDc!s0tt&XfD=*7uMO$$sT#4zd2J;T`DTgctf zH^bi(v}Fd3A&ipVUggG58SB=)JU6k8bnABcBa;(ub<43@$HaP#)!ms(>xTW}?oU0r zaWRX{GI_+-FYRO#4_?#`$7>b5WLc zLRLBN+njg6Jml7-mNGWKs$5KDDDn~kIaBh0Mm;_2hSYu>_~HCOFZAsKc#%q!Ei8HC zsUvlEx~lk&Jznk`eHQmJH~Qp+vauj{=~IKHDstzn+y4MExubMNs9DDFv%G5@?6?IS zj2v~YH(G0ly|rg|kxiT@%qKZG?_HJmhqboTB@*dZa$M{SxGwCUZ$Vi%o*UMd@)Ei= z>mpi0ssu*>gyem5UX4l8Rifs!>S;~WRifs!Ib$T~pfrq2fXnk|0=4b*$UH-IMv*WIxCw`4&7OuE9@{NLpbvIWj@WE^SQB8vA50~USD8}O11E2Q4 zL0DQZhP+E@qRn%!>333GEyPU}YQfkAR~g`(@N2q~yw^l$QN`NGpIFdg)S?i~WAjh| zq+oD+@!q*zFF}^qU%#|#;Q{+NRk%kEs0*$#bBxzns9X4mpt!cvB$5o~ zSx!bWKqK|*RSuo1`4>8MmX&`e`ZVnK1~wa<#J~m5x%}(ru<)9x9Uc0frAJSZ*mSyH zjp2Ii#Ct%4iy-F$<_Ogbm{dK2DheOeWv-NvzV5aOky_Ylj&3FwioHD>4sYw zl&sN~B?tgyR~u1k>$y5|8JhN{_BzI*9`@iP-ADmR$o?adjPb{%a@x7oqp^2__Uc7H zFp(y16#M6`bp9mpSBUStMSUz!R+(?CdMYcn%Y8<6`2=v`JPhk#(I)**19hpeX7sH7ve}Q?V4$$ zj(8PD{o|7P9dlWlqHF#e_>VTJZKK@FBuI(#TeB{Fr3a^{^rv{zY4r^@4L-+5zm@ka z^4&BN7GhWKoQ_B}&$+{LO;=8RN8z7@))<~iqPMt_LMbHv`viWV{uRyLX!oh2+Tz0J z%nu=yc}FS%0A{@+8@*3Y_-)~BELhzt%N@a;mKH99c>o?b9Oj|%9<4sNpjc=Y-W|EM zkPXIY?c4~8{mrM`XFm0lljbC|c3nOn_&rpSn+@BuAw)Cyhf{KaoD>@Su^*9QljdRH%_XqFm= zw{Vx2w!?e5R4j9xo<6?S*V$6I6>$)djJ5-cxZ4{^hW8xg(&iPRB;Lhf=wo z^4C>Tv$G^|S+j2EuI|+Cu11PR*MW=ZUljCP7~H(xa=`5I<7yn`N7p@Sw}dQhruc>~ z9%v>}dn!+BF9&ND85!Y*4>;z!9}ns6Kg5k1=xDl{O&!!^&gy)%OlRi#c{wMX)lCyi zi^K72THXGaeRZhZ+{tZgZ5_L1BPZvRryL$CUYir8&L_p1cAuqcFKuX~xtZLMa)^Zd zsy%QzpL|zrZW3)y`_J(Pq?4rk63#1@PzDAtGsviK@1oQ+?MF(u@UE)Xjfb+i7NkoH z?Hxz=NzZ!eWbkIRrfayoId?77Tp3Z6Mg)pL!7N)N=~+$5!8ElrrlZZS$1^FE+-|t> z4gI~o-QHqdILEOB{oz%$neMKeOLe+wC2I&sl2eu^oD=U|ZQg<6+rc=K!S@-E0h26e zIP6C5D+gKd{;h4OSy;)X+{Yw#5UVQ2SRv;laLD4hTA#Cg!^Jo>oJyxmxk%#FWPPxg zRSbHZ@tXCG9r57p6vo^Xm#6#Yyvlp865Q!l`V(o=izu5VjjL{84+B41^vx2=XnYr< zL-u)tEOKQ^9FLs!t~92g+>U8AZ*;-Zd`WY6;>}A>l!FXph{icmwbA**)YcuJhwg2> zRe654(un3=$0&IxjyU6`ZGD>O{(a^D0KN@tkDO&Vx=PB+?+VgVYn1Q2O{vYJcy{+u zw}EGt-ZJYO90op_uP1Mb_l(HVXz?_ROBDNLgAM8QuTt@i>>783wJXb*Owhv#c8q@( zJ6D3+c&o#Avo@h5LJ3&%mXTy8cpM%MbBgz9s5I1-&YEg1YIS;l#VtVUky*Bp1XA2a z;oMxjf+h+`;EZ;zU&4P7JZa-CW$k6vB8C`ZGD#$GjN_mf }kX|icPC9^ATJgbCq zuPk7Y3NhfGho04i@b1FX! zp!kA)D)MW2rCXTdIgf@?ah&IlG5ORD@iWDDdgSp*soTJ-Bja)r!jpl=bBcq+ULUr& z)MMA|u1vmV#7`zjVS?ir$4cg&;e1s|d`4}C2>OUNON%0_u!!l`}OfRX~3Em}K{N>2NB>PuCE#;=6 zU@XK!IgPY}Nh2Hl!Ulfm0glxw4^Vmf)<&ys6Gh=ilHpNTMhc8na(DITo}AYlE6Q}-nzf9o z(@|>7=`|mTSNe1*IMQNr2*@{5ar)F6*TubJ^T0PcuAddXwbWP1Y(j$It06u6))$EE z<83iSEF752V|P*hdR7*bsm-JOH@&~NL@t`lCv1utNHMlSr$%y{)XqrKv+@4`j4wPP zscP-0g?mY(ML?f;H(vh$O6aUUBY3F%BGk3{r-$vZ-bV}#PZaqBF>L&QI{i;^KNfbqsouj+N(A%(<~X$wk{?xFs8xQ&qoq zB0H8dHsR%lIX(Sql-lm06v)O`A3s9OI5@|8%Gb4cZ0x4KiWy^=FwDRMo^hNGn5&jg zZLjDlaOOswfWYG$fIL>Sf4g4(zt)CazU}U6_>aU|oLc?eq;9JmDuy%20F(!k4?LR7 zu<<6ZX=0ZWUWsLCg05GL9AslS^{n5BHxt-vTAcA9VzHsY;Eetm>0MJ!>D<*tSlaQ2j&-d&#ut`W&SOYf0t_BT0X~D)vD?QUCum+{cBl&v zm&+%E_~N7S?20^pb{Y1f$O|`cHjb5-b9o)qA7!CjqR>K<6~49?O-Bv*6B+ylV?U{aLlbj)R;)#-l{ zHK{xqu4|U>7=FiS8~J{RYE)haV`Edvp(6 zo{@ild3S$hcX8)=R#z$-Ofdu@=*OIURsAIw!q(R^>ef4DoeSG#RFG6-ui`6Uin-~n z@i$u2H4AMme#B+C4#ezyugY+KWpQoi%QoY&JhNT*{EY6nq;wO$Y9bVD~ zytkIt>SY0pIpZK6{B^0dFBf>?<6hJSrQ{LW$9F7BSuj9TfO_=pT-5im-bra03nXIc z2nF_%caAyi4Rkho<%Q3PHH{wU5nRK2Gb4;c8yT{bJIbvPq- zgvlE04iB-({CZbTo<7#wNYUC!A~IfBEb_2WGIDc@@=Ze;WRNzQ=I^{s$r}tR=Yh2S z4RzP@J(i20O065LI%4igMnfF6FLvomYmreTv^xt=8DAsD(aNnU1eJ0~$LCtQpNqUx zZ6=zUcA0H+70gOyxR)SA#o2iPgNzLS09x|P8_A}%U0GDa=7170&mH>LQF-FD)7Qh6 z@0FKPw({g)N6Nlq08V>nHI88~V&(O3dc;Lt>UT@1-CCo_FKk-yBiGs%0l%*Pu{e2B9!ZBBg*umpq>hCqMM^Utenoho*Ycs{)6SZwW!$#*! zSYVPnh~o}azU5_ej(yZu6ZX&`P8qJ;pD~H!9gZqL65TDwg4@hcS8$B(eXrzU(%$UOtxykQ?UWKJ;DPlD%n`@$hG^>Ob`?=hHZR`GhD@oz$ zdo@xLz5f6MSw;#er*gzTGbtNOc_C>emukuwqJlW(bLuMoh2k6czjEGKw`>IhP5%H> zl25gH^_7%fC-5XP=+>fHU~sn<&nL_V)6o03BCPm6`hT`*b|Fvhq}y<^_BbaXb>M(2 z(0FPJjHMN-GivJOnum}4K{fV1)f`(FA(40^Jx@6NDtLTzu0uF@Z&A0zenwr#AdW{h z=C{dttIu`h+=VFSEwO?WfJh&QQ&XA0vn-{tn(Wb7ePZQIX)#Cw2_~BDQem)0iVJK#FnQ;-Woj^5SZf-OnRKD# zo6aNy@^Cu))^??M_cl^N<%0PIZXb=j^}y|4J9yciU>UGiOO9Kz=b=MZUpBb0(*XS-zt-fc30REO!O@~;&!dB z_+rlO?O}#%mzh{YxF`r$zH$#fwG2KzyJ{-2d`IxXN&18JM1WH%QV zSSU^xEWA^!(mOeY*x0B}TapK-HPeEpPIYAJOXYr@%;eecejn6Y{?CP0TvKTF4-aaQYZ@)RnuN4;{o zr;hX+e;B5VG?D_?65K4sV~x4MHPdNd*_N7P&!;Lavx7g#cs~Bs^6!Xl+WT6+)FHMo z!z{-!GH^M_=lWJNn~GL7r3Bl#^xudz)X}^rb*tEzt`-RQA>FxKoOKn%X&yJ$HSlJ+ z(o!hG69J9f7jLHSp69J|dK~(l$H6Un+V0Kmfd z5`#Ow8JOU0`h5juh^aWkngnvCD;<#W{;Octc6adEL2+=dq9tZ`MhZ_Saqo_`#>3;k z7F{y=j}k1TI_5ID`g78`JCCx(@fHszdEf7sX$d49t(+6d6(x&Z$8Bk-MSUEDaE>K^ zz3Gg7Gv1<7sGy~NSdvYig=6C%5RFn4h}AiQF3Omj1#nu$$l$%n#atO{t=jppdpLgNZ1`}-m~!w#F1-vmgw*$kPEb6 zZd`>q&wLYH1^)oV$GY(M*`C%Hyp!a~KmrbWW4k4)4pyhAnQfC!#7nHM7)c8;U! zIvUobrOSPYtr)9voo$beuMCrjH7LVKfJMD?hxZ3RrEO~7Fw`wB4Yk#hl#1OX6P9Sq zk`)B5(0JgQ@+~7o)h@LocN|wH+8`xvaK%qux$FM`*RH3;6G?flh9*d}Z3hf7$&7K& zdd*HsEwc;R9jE*yGRTq4EXi`ah!Z$b>EH3iL*q{vYH)aaO1aZoG&fU88|($X?@VKy z=D6)LTd4J>hUP^gF|i#+RDv^u>x$_7OMeEP;LR@KZI#?hG?6l%03CaEIjN;NIW+qU zEw^J5?_KfDzM9fC&9HO#P?8ufGD#Wh?d@F8{46gQf6zh?`}TjvqPg)*ccXMu7Ju%J zyyHIq0OZyu_7(pC{QSlG)l!|~?;~3zeRF+hZ+~Sit=ua$ysNe2agf8%*8zE>_%BAd z0&N>dw%o(Z^EfO!9AnnK7D5#XJ%(%LzZ7e)tN4oY@sY~B#Zx2hDBuisuB+IcdLKj! zJ6&SfCDo)wHh$^jZP)~3ApSMww;v8X2nUrP_SB{+i~v2x5m83MvgMja zw>IwKACRR;>FL&~_>*6TeMW10dw>ycq$L9P&OfDPR;r9{mZMFrjcq^SSAwFsjyNAw z^HGtR66H5U6M$h z;!BLS-h+YGu9CIgf-c)0!C~VchpA>7$4-r7n;;@VCg#C8#s}j|;>|O_dgMA}^F?^D zL1YcMg^Xw-EJ*4PIn8k%7SV>Yqt7jZC7cFENiXJM8YEn0P!cl79FS|P)zZ#uJu2NU zA~Cg`YOf;mx1Ml1imH_yV{)Y?_bO@{SArq2w`;SjEMfr=32;tO9N_%C_O7;7@Dg}$ zN}hX2u49VWM`&)r-{pcq1pPs;D;F0R*K$}a0hSZwf@H@XbCc_l*Vekf5={d5XGhYd zw2Xml&SypifZl#k2kLWKdrK=5d!1FLx8XfVL1zw^YRW)V-UvaCgE*_cBJh8MrO-`< z?WC4ga?Z22*u!Pq0s+Vej%&&^^|gjOd2Od+n~SJCpK)EUjJ$hodEmJu(&czAZP-O4 zxW>Xl^v?$%{#5YuQ9u-B8{WPnpfxz3cx0Apg54G@}?uQ#+ zHlKBHlgxb2pMODvj+|G|cAA~Unv%SeVn=K_X)ponxE|w-{*~?;HTJH)1p8#??T_}9 zSe+O-Wgq=|+N7-p*5g-g7|-yl!BMkj+fg{@i@tIH0DRW(hW;D)Kl?vYva=VU-`j}h z4L(M|!G;M@jzFxl;u*#1it5#rIvB%k@0@4p$^5HtL-7Wlu-(mU5Z=LH^6V$&z`((- z1&FMr6(yrKaE+TVwSR`*4%6k2?S|~TsElmA$qJ(!;W!wpz8vr$f^@GC7PdEjdRj>P z{?XHhp^jmA&CB%V>37M&bTgrnIM36npW3Em=c5(PrmO6jK=zK+{%WLCc&ka8Qa6XTQ^!I!)k~HW zX(AMxTAql*!r+yKF62he_K}Dn`*BumU&F07@|OedsMOx<8yW-D>_Zr;deL~@{Vttb!hF_Ox!Nw5hrBPoQ z{5cO2Txv5ys&^`^Q)f9HyuOvjX<95=hl_QMLOa;J>ubqft(q`PF7CO`K&iZE;k3HY zSj$zR>e}Cfyg%YN;@4P6>2n&&vNZls+R6%?5!_>~ zL2dBcM_8@I0#ajb%IkTS_Zi#BtdA00%^$$Ug`Q@%xY6vw zKi=92$sA+!tP7@#RJOIgmU-6YVlNc25!99$Ij%0_#aC9+q%an`g+Y&e*jC-#@#uOA z!PT3;Alck_!%FZDp>Ls%hv1+1 zLbWKMd*-zBB9e0Q9mY@L=N&nxXx<+1%(k+NDCRB7hmvfvDH~1>-ZPU~c8~}@9(aMC zUos%^VD!ib@x@T^?xP$UdqjppRTfuHtagA&{PX$Jt6EV?MvzKNQ@GMR72!QAMAWQq z?OIz&WsRqf;zJ_t0PP^)dYaGw0EAb-UQwPe4{7ZYBr7YvO1E23(a#T>_0%N7ykgQSywQl zU0fGnpgZzLd)Hkx8Pc6q_E8Q^G`pKxN5kKTl27Jo)@UxHLdBw%W)yO-PN@>g8foJ(8QSJD7^HKS0|$(pRUZ-jI`A|)ozA0p_tDQZ%Guq*;bnG+p7v;sEO~VvyD<%qmXYwoyDfHoSwZ* z{Tt!GfHf13E5QS={{XJCuL!7VsO039hNhow;FY#mnjJx*fpP-5ko_R2 zQn`_B?KJc_D$&UsAi?B0{A#KGDz!&k>M$d&bmgJ-0ThJH|5fx@lJu@Nc0u{)PhSZcbzviplq{+ zZKI*%8>`5!^#+Poh>#cfgG>k4BDULEms9ZcHq&2umT+Fc31!?-aUnlnO5fVWCe&=q z(svx1Z^HRpv@_~09BucSNdW2q{ttTT{2`_MH-}Zak4e0}7gJeD9HLU&1Wo{K5}ij9kdpYZ&vIrOXjzuFBvhT-Q{W{{Rrd!IqqjyqROWh#+L>T%EA zJ$F;^R);m!#nxy2^ZBwCpJ>U)A9V4+s!8B458b>#Nh_j|Sf$?LIsX86n&ZABc-~Qa zbz^6CrYKv=mLhSu_ccSnTFuAX-u7F|Yg=n^8q6e+CiwI3*VeP-U4?dcdhdq(5iX?W zd&`fuqakSg)>Rl#2Sq)xRcyRD;H#MuNT-k^Wba#s$?MpqW_X4*zSpkq?PU)WsBpos z3hf6usyckp$7mP;TOTV9cGBdH_7#!W-dXqp!EI7|IWFanA|aL0ixnjBNXMzH%XIjA zcW~l;4kcMeBshJ?AO5PNd2||WK0^|?XcRJI_;)ZksC64?A-Q*ef}9<>1g=LyeznOu z?la{Covw73@ccc8OR}_UJBX~=r1DY#zyp!HJpC&WFMwJvAlcnKd%4826Iq(mk)~+- zT|wV20Xc6)myyBc_qERdbY3`e2% zu2yIGZ*6w*HlHf&Cp+Fia60kp=~2qkFNF1JEgO5ybr~DD#uVrD70c_$HCtr9wumP9 zRoIilo`?EYocm~0Z?W#a9o4isES+?LDTdl5J2oG?+a9Ni&ARbUi43yGCZ7>yegj3x zJ$d7;az71j=7e~W4g|1A9snMrAo^Ba%-g4Cmm!*F1d+!gubn9EsB3f6q1Ut@5$P9p zmysW`#~AxUIuMx8818G6z5f7(T@33CQR!xV*x4gUq#w$%JT-qLUK-c&FV7Mqbm({W zu5(d}2fs_Dfj-bKNm*TUl?r(yr+RLtqCFA_ei!N&B3MO}UosU2E!)8hgN*mZbN5~! z@C-HzBdgS9Z!04AZ7QQKtHY#3y`^W?@-8jW`kltHf=pgChRMk)@ zNcQgD-hx+58uB%)bU%l8x{8)fRqmseBvlf-0#EUD$5G8|Ua!MnwK}hd zp_1UrRzmO*lb%oKT(+AP?}y!W88s;Fr7p4g7ZHb9l(*d-2t4M68}++_RanRs&_OCU& zwNY_0vqAfP*n%d=TU2&wEuq68Vo6iD?H~;1yq-JfUnB`P%IY?PF}P^uRdQV-mT-| zNBswHzwf920OBjE*ZduzU3f$O5y|9=*ULm$=JLkiK?ft;;OFaHXZ$LU3jYAmazFd| zANbamPjZiVxD0LN3HOSSzg$<%a{MgudQCaGTQu{Suz7BPCcc94MdX$q8Na%T8RU*4 zLa$%KyskSbwKEm<#qHvwE>RL@UBmsaPs7r)t65WaWhl95jX#He1=chjX>}XpZ{-*n zKX)ebNyoQNE0WZF3F2d^Y7cJKjIzrk#!bp)a>t%}R@UuPPS$PFE-o6$)y!Fyj?a|m zB%e`Kb=!-r8rCKfoS0R~mUqi!4xIJPYet-L5xwnVC8M!7gghUwX`U*xyP0iPHJCKN z=_=#_oZt>~T;{#-{{Y4p9wgK+tt7IV(s|@DhPglhEJ)z@#dLaX(_Y6jr1vW{AzZL` z0CSuinuAO6=Cfmad^MZ9se(L-97f$h1Rf4?+ltau6%>{4aa|nx_z%VQ?!HIa%xp&Z z-tjS?QO_MKZ&>(i;ya%aYZ|Ok+I^DzkR7f7I~hPfpGq}98~Bd;)V7;`w(O0$SfpS& z0fKTz87Bg(YCbvfO{Sr!Tehh9ffcreUO#k69Y4H3TA0V3Xi|}S9|3C#r^lvQ={8eG z5ZnXmJNoCGaqH<_CZ*v`S#(W43ux`^=15XGpD0#6x%pI(O>r;de-x#Vb=D;eiaf?x z)D76_oOG=1Ys6ZFl1F(ip>;I!S;XvuW&xXTTmk^Bt5RH=j{Mpki1>3f@-@}Yrj2(r zz{+4MtI!N%BlM_zTkz|}I*d9*b_s0e)(47QNDvZTi2VqzKGIEdQkCA``r;tGgfll@ zxcc#30o1h%?K;9qZs!vbAUiG?WQN8uo=+eCs>U&l7ckTc8c)Ly6~}85J;j>LW?+$v zpcJSBjuhvi`c}@N@Q1`&zLyo8w-%P;?M7)woLDmK3cVC^FlvKp7gh|2_bwcas!cn5 z{-I84w!f+BXTwWzX{g-7Mdrr;07+l~C^!c+<22{7a)XMwR@>m7ypt0)p{2nTG6qrR zm&x=z40NvZOzR0ieQkf=duDv zTx9d@T^EJ*T}%EEo9U+2?&Y7&k1O_)+ZP81uU@$ovx=ItAxWdM9u(E)jzDf=fxs$3 zon2QasO~vED$bqZJ!Z9qag{}Oq< zlV8*A+U8=iwiyo68-jDf`+C*;7$%x9RnvAn<4gEU;%y~lwm14N;^BxBxH-rBp7qe% zd^^`=*X?f>^G=c|9u|>Ob|>V(JZ#3W@cCA2AZmCd?Hu*BRBR!~93 zSd&}2Sxj$zasW`p_wF3~ReN1t`a}#hOPp@NP}uKOu*vIkooOYp%csLX7wFzJyVl}5 zX@=ry3db$cUCK^CJbf!aQTSow?Q&V+eM3xmrIgL&t=B{A_j6jh_OEjWyQq1WcV$** z2^W}g`Fd4owadxlAlzNT<#MHEAbl$a)y)+tN*w(T!&mSowWw)2b(Q9=V-nliM+!$N zI8vyg`*X;yZ9Wm|WGlA4q_Qs`dGJrCO6sqCU3I0%kjl|q!Q{n+lZTHeGIxJE;;*%h zR!Iz2*OCaA1qqb=zS*jDRJSm?JEP$JUr*8O?IE?czLpV*8_g)CdHF_p@7B3?@E?h! zy^-y-`zZeac%kEImDxvX$nn;-c=`v0Zf~_Ki(<(*kVPzr(&2y~ciXwF4L;W6QY?{b zdhM@LXPWG(VT^zVer`Q0CqfC_t8?i7qZ(3`L{_5Drd#DwcuZyZs_* zr=ECbmE-c;bie>QW8Sb=T-O)D8q7M)_5Iz%7nbovVU(lgiRicbn_Xvh z%f&S0lj;;_)b;I(8jMm#uy$G=y(fkBTaNunM|*HSPFtnE93#~hP`RUBf%N8IW}l6teR z(7ZQsrdw)J%YSQeHPjME9?3D5Jq2^xKZJGL4Khm$xGdvs}zlowW81u06*Y&PP!y2x$1^k0hmr$I(+)X6x0*>8%u~8bzy1UZGkc&>qT=YAS z2y6E@a0^Lc+N>RwZb#h&fA#90i##E%YdZDh`ee58ER*?(D4kC6j+iHomBr6-D|n2~ zHC;|LGoeV*cPhgpk5B1}^%?wMBHRmAyN+p|Qy$1Rvy2{aaljex`qr8>DP21>a~d~; zb^SA2wYY15i^#eTC-nsDe`%lD|dbB#_lso>rB2qU2#&cFYA@K)G@pPus z$a_shCzxQ6XL9};0uV|9dQlgpKg^2Y$5Pz6OwDskmwj- z$!(;Ca2RKDkUMnAtWo1TsUwidtIUkDh?ShK3C}p-`c{nk?zHw7!Sw^htjwEa86+XW z&OK|;mSW|2wYiVg;uvLeH z5=u)2#t7wk=w==IKoSIYN9}r!~a|9n}HsP|M zOgr*N9s2dJInk^B+Nqqae;%yJH=3VlR~G==#e_kTQ}>b)(Lf`os5Mp}0_rNT#V(kH)P|{R=z2Pr?_; zZ)^l`{^lU2PcQ+To;mu~De$fPOCdpQ%_4vYp1k1Vvwz_()@*zMq1-_<(=_>0aWfVH zw;#Rhz^2=LTAdU}a)w!>`ErMDanDN0#xiY|pxV&8u6PT?mwI-dme#R`gU&JoGJrBy z9Q)wb6L=%VO5muoXuu2q03S#>{{Z#ZOMT*Njb_(I)LvQONra@tu0S|t033Cx=htn7 z0wNgXWAC0To~2mET(%OV*+F~<;wA8P{gPTFR}x&uEJN)8-~}L^-MZFxkKpZo-c^e3 zJ3q6;3WfVh9FJels%oAm*7REg{hNJrGu&Q;K++Z@6*yjdRI_WoJhq2+CYTjtDS!fF zz{%am*0zgMiqMU3vE2A$!&^$O6D{#h&tY{W2joa?b5VK8>5!_fi4GO)0)#=Z&~y0;{MRMb#~h9T%O*wUdx%< z+Z6=cI*l8^clLT#t8Ef27gulu7gIQ3TXufv&IM;r@bAPi1UFj7m^y%8aNUv5@YbG< zb*Pv|5FgtUGg=gup&?(x>os&Gh~;Sp@fO zZf#}V8%$a?I0Sxqt{+3T)OFuFM%FJSF-+h_w=6NvGI`BSSVFZsOJbuI<;rz_An*>q ztLwHmntiyE-aVV$?o|EO$>X;amwo{8FPV2RhA6<1q>aSuj!7f0O3l+?ztcQxqN8e; zFB9A%JVxP&C({|KvDw^QYg+Z*rGKcrO{q?h#4e;%R1z@G0sjCvs#IzFB_4>E2hC<} zCxCoSa}blthD`EtGLfG6HM6bw3thN&FzUK}jkL{{+{_%3I%E&6dH(>2ZDbl}+H}22 z{@!b5R(Sr`Cf8i^!6%;e>sqDjL3ux&cOX$A!UrexttjE?wA$)Sp=YUvej7d2w9qZ~ zn>Ez1h>VQv#9?v3#~f1J_#aq?=Gr4?73xD|u(iuiv=#%NI`*q}`oi60!E)fKDmlhG zRxgVDQ9hG@5=pAw!X$|>k+wnOl1Fc?ROv<$x(GowR!;%z7q@n|LdwU_frBbrU5%=} zPaXL+ud8?q#1~qYiFx~(;jvTa$34!%esb8z2ev;d=HT&V(Jd$T&E^mm=1vIj)9|jQ z`&qlV@E(tRmp417mGbUNu^i(yi>RWUntB&1cIssK$HLw$02x8%zVUqN8qcKK+Nb)M zmgtrzv17NM_09hP!kO{cu8aQwe^37a6J2G#sz2c$mIFNBZ7wEO41Vw3>-7ez{{U#} zAM^Jw{rx}VSJK3P@1bv~FqM{c$5 zI^@yY=vULr0?v^@@))0^;1h%SSI!gsV6wTKI`~GxLH-@8-MZwhbro- z_V!H;lYOKlY|5A*WCPRmuHRSi%IYFt^eIa-7A^tE_Z6#R`f9o?njMUfykY#L1bTE8 zsd3{-4)|iXxJLwlz$xkMYo-;g7M`YcA3=&*S*2xC?i58j^b7)f0Tn9-si)n>giQ? zTm?+G9Ch@kDzR~~aHl5y-A^*QztnVlmz80HOIe(hWjHC02_qdj{VHqtmdejcFv}e2 zG!HN*rsi@_+}oV3&g$`k5Y$3iqoaH&O%us~ z5fL%S^7arv5<$&xOX3_i$>6Ow8?%ooI(mKD?ff~Y==UBSmj3&~)|MgeJd+eLt8T!^ zCxPl~dsvGcjj6S{@G{%Tiw>%zaA0`{-nktTeMqjRw$e;*p|n0yMUv!+5cC|^w5{%p zT$t_lAwP5Mk?=h!{{Y*59nr0))Wwu>PXt(dkqRSVkUEd3r*Ls+n881YFJgvE%`W~s zg^y;|g~G7!pIXrH%=(ge??#4AGWIZ%m6l0u+h7Gr&hCP`_rL6W4Cx%12yH^|cn;(f^@VgCRVUDtpm)aUV4n%qk- zmZ>Mn=5fj%-&*dan@^fu%cfbzg#FP95Kpd1tGZ3yjlQWI38&kM*cVq=l}fKp!=AMs z_SZTcORF=>Mv<-w{O!u21__Xi{sN=L?X}WP71SGY4jMoQUr)xp3tRD>{jJnVZ>L(t zYa(tJ401uvHtpM0CisB5a@(2oyE%o!m0mz$f#?>gsVB>FF8OY7z9_qRJVAc!qnVqb z``q^QtlJCAt@0L)n%MKpbm#QxUi~(Ur)o19n#S2#@XeSJvVC$!YDj!DV;0}^dv+Pi zFWI9Tzac}8x|JG@LB$>!;``~Yb$=!)l?p`>0Y@kIYuDDYv^F-^`|hJ@;S5Vfzi8X- z*S&jgwRxc0UddHg+6t$3Jt9eXU4Z0 z6^4bO+~35LNQpDr#^?xau+Qb!-nsj&b54%SPct%3SPjs{4&8|3vu_YEWa(Yt8EC)>yAJ<`e0Y3_}x8qgLacDH01si zQSf~6&ET&U6&9P;*;Sg{a6nhaPjWr$n$rYk%5cmWpbi^x03O1<4)<2J@UMh*o7l7) zFfGYP)rc}YlG#8NRC?o{`21@eRu|g!uq330>EC4gLtiS!V=BjZ95>2x zYm&LQp5@Rh1glPd8Ft{WqqU|`pGr1-jT z5Jzz|HgoCr&`8mF@-h)41C9ah?_CKcpzSI9%=5i}#8BK>Tv=QAb5zt4<}J~*)~>4* zm?H&9;B-7!YvV5mOX9x@*h6n1g8B%PV>mcDQI$WZ7R@B_2i(9h860&Tt5qB5fwp)Zu8W~t=x)udtd~)M>=y@fb=&Gc`qkHXdr^&i zJ!Y>gMk&@oHy)0X>@a>b3h1V(u z)3}Z}r%|XY&tb}Ek!t#$n;g?LS42q(Ai!5?3_E&vJl13y*6$8m-rZtI#-`iUFy6!> zV;QeV(0oIs>w3N1Gk9lD+ZIb^Bs+=qJ9AxXTljigG~c4y+Nw%OSnZLC&FzZ7d#8405Rd09>%3W0UDwx-f!rZ$u8Jnx8jW==z&YB7b6CT*2nFn;JG!gyeVY zf!?}X8x2a@%JsD^G{xjuPN>-=U?Xrxsjp#OQcMDFt^of4u1Qc$tg0l&;z_54QJJD> zlMVnRf@_^(u+x%CXwTZko&^_{^u^U1&ShDGbj z00-AK7sT7^TkTg)zO`t|yc;$a8;%Y^$EQ7OwAQtM3O)S4Xwpa87EDPgwT;UW$MIEI zweJYWCz&>qW*}e_XAmRuuT|yM+F_C#MlBkB6hWmN)SDmQodrrYOld0FJ)KvTeL$;ke{4FNcJn z@rJjBe~_$OQEeNoj13z|ZBya)l9Nd?+pG<+pyPRW=O=-R;S)q_okmrR&u?oU)!pTm zXFc)xSI`gR-4@2e-UZX;v=YU>Bv{a@44-^fSoou)rPONiUzP3+yjbJ0=lNDV^zQxR zMNy?=h4Dm^TlhCek57tJA*I-)jq8HEpYz_RnI)G@OQ^2n5+%4|8w9DxarLiblf@n! z)h(>u z%`{pnuo5VRSDbOuuafTUR^`H+kfH?Qx}zBKf^VVoo}aJXeV+RHG~0U^qDayhyp5^@ zN-`a=3mU(pc&g=e&2o9#(n;1ZEaM6k6P%x~1Fd_Ph_y?7KS7%9?ID;(S1ULHPdLfo z*O|A(B|^b;%LXNhj7ExAANS2@*y7qU{0-t(*Zf-QExd?RPM>sZCg9Jvlb=edXQf)- z>V>r%wM1lHo1YnO!2TbddN+f7S#7IoQdvHeZ3$Re+E~y6MnEF~4_fIZ({yIRhSCU6 z1eLc#oOJ3tR8EwtJE40^GlKA)t7-l=@*tCaquNMzj1Me<(E9sVoA_>9O%KJMQ6=i;XKwwVLuf7~0|(9m9EHOJPYPBc^d)7jf~INR$sN!B#OaZ0`9^ ze>JL9oE4caT+Zmv@Y^E%QnFzCmTglg*~lxBbNbfd)$Q*!O=8ANh|=cYQMo{l!G2#g zh@{f~GHE~B7HFOfvs;5Is?G|AQcf~3wduYgve6;EBl`lt%PSckV*)cVJe=frG^G^P zhZkpa%rCDr+g&z8J={5PRH$G{&tdD$cDlLqHOs(GaVZshPqomz#S&;5Y{jHd3AaJF zKA2kA)I3R}$9*IpXOcN$kgKqP8*%(P%w;bB?zWtG4=C~e~z{6#LW;<)VPjw@Y0J+eZwE3g2J z=Zs_8lf_c0@I8kObT~f;_!iR5VZGG0yy}36WZG3+_3h3$>)yJL86rdBy)97XMDNq- zQV$m0VE*PX4?q!k_vu!(ePxqJ(`TLtt_8;S*zv9x9C5{EQ7OghY0B3_jCP1<(cZ86 zz{Wo)a52~a0I!c)rS|y0xZn59bl>ohF2JWxw#OuoB{@I$&)4&&{{U|%{R8d)0PoU& z@iomle9z(s26{(|Kjc-_zw6Fl{Q%d-`V^XQx`baxAp_;Yagok>`ewe<)hx!NrCr$< zC@vysivgT&1ZVZH4R3%JkvpK*FOJZ8TQ$M|0LQD-YUXI;J{ipf{{R!Mg``F+VFc?M z6OsnQ)cV(Bt{a8b!649ouXHaM=iKa?ou<< z`qsJdRB1Ctbql8e<01jK`PNg5(#0twnA7fWQLM>r%zUuNby1(hZ4`Hzh{-f^IZdPQ z;GPY4R$dK++@>!smuqAhApjg7tuo``wuK$bkaZR^F>+CX0Fy-}C4OL&zQ!h+Cb+PI zYpY>zBcM188x^_ZyStlj464A;0+7cdI|5ud!BV*Bcvuek)HT8>MwUeM#<`UuYf#2ZgrXL?`|#{CVlKT95W07 zyZqw42gcU^XNWFcN&;g{=dVhKPxybJX*z}0pqjjv$_DpoQB@FQ*EQ34gTb0+t9ts3 zx~=SN=)b$Sb;ba|AQA0Za=Bi{^s2@ZyF8xe#{L^(9K`5;_zm}0Ipg0PS9js$cs>~* zibg`SNOGX(uWTBTd;y}!>l2J;s@0+E+RBzK4b0-fA%TiYsKc z4P@Vag7cqT*LHB83qm{I{{ZZ@uHF;07eXJk%5p!}2!T=f=CzZksjfpU5ea;i4dCYpF!~aPEaM& z0D7#W2lcCP_-{gMc?!(&`Kh&oOfsV&{uLHu=019hL-A4(og&6n&Nu`7Yq#)Dk$(=i zYd!9tYi`KQK}F{qPjGMzdMVMgJ2GQQCy+N*NfiG8TDxhdX!h}ap>F)3p_qvzQM6!! zO*=AAEAc+RtZBtRvn)Z0Mp{Ww^X=_fUIg)Ox2ya^f*X~KYQXALMn=FsZ(g|qy#r8h z4oM2bKZqejb;u=?1bUB}j|^CNKT5q*r)w7NcR${c$c32p9ew%=#-Ft2_qHxvmeBHb z)b1K2O-goq$z7y$mAZX905Q_7-?hc^S*^G`M4Q;{>b<|nuAcA0{u*>GcjM~|qmnWs zi+bmtp{qke(tI_i*(^WsoZ3ji$qhG_80Y>4R^u!44bD8grJsnNBk>NBw${2mw8T0q zd65oTy*`!B=w3hZi);3_^1(FH$%SZT8Ak4KMle9ldbIu_&@7}ePk#%LK*?nzIOn}U z{v`0kD6^aUSb;cJ-5C6iYWaHHN_Nu6BjYKq4zYFQM(-yM$fG=9XCk@B)Fz+EN%q98 z!nYg2`e!xi`j3ZneNINyMxj3GKw#GMFCc-_*o=O4K|CYjvv4FcI=L7!vnR{-KDC6Y zHtvp!c&}#;TO5CgOpxi;)28N~;8;NZ5td+0VcJV3mjH%cMQ55<0zmlOy?_VOy&rTV2%k)9CkG|DJiQqbH;W!oqA-8 zL(;O_*Lx;udP*ZZ$YQ;_u6q9fO7xv);qITQMP{&S62TpyeXGQU z%N(4Nd8~!_e*rF6 zkKH42+r}%Nx}R1qHFqcHVIZ6w^y0l+PWXRgq3Kulms*^c&|gZf$m#|@QZPPmdaVzF zw7C2|K8+>y!@$kG8I?nz^apn}HE5~I<(;XQ6$31qmA z=17c(VDg&`2L%2cSJ9DpSHm(`-<^co{-^0(28ZA;hJS5>T`N@v{jwlacpbM8N8RtpeYpItRJ8hJM z{{W9vSzAJ{uBV>d>Q>s8qodthOC`jP%8px>`B;7Lai7FjTYD_}#)6jio>!A$B)@4^ zJge_GIODe<^yaqgehpa1ezT^fa)uUG8|~C*oZuY$RmnUl;TtQtF5>Y7sdCP9G zo7TR}n)J~BxMUVF}1DtoqrEdpkshd3e!a7!=WpjS_cyyNG zonAsoVV)1STEF7O+i4D0Btm%q0D&6qBhsQ*2omq+##qKMGvBR8t9U}%{uD_sT0MwH z0*-mjMkRYne6b;;$UYsfo2%;16_s)T4aw);yUT}hr%WTp{{S_BC!jk;S~M@z)qmx4%Nn}P0Hm(+FZwsO{eROWfDtu=a$TG z1fWKz)UGp+Z)){V3IiRT-scSBCEUQA6U|(03rgJ2_Dj>BfAi^9?cvjOizb_7u}3l3 z%PRnT`kK}hWSz|%m5v`u)h?#-<3Y6fN{mP+qh}=jYteN^E2PN4bAkNp4iVv~Z+_bv zAh;>B%=4Ex$v@uptCt$yk1e#Tb$XE7qUGgzkSdJ#Ij2^Xr$$g)v_$n<98S4+u19@u z_9&q?#D4Xd62qUZStaDOxkcQ}*3x8k=V9qrCGiJ=!? zlP zfzrK4hp6{0-RO=F#yW-l%&U4&TGqRw(7vWA(34)OC*zX)~>r{syy0q2^@ zxA8B6+H*1SC8Cf>8}8Ht`qvEURFiMHaGWBJPsN@VzSVD0pbKvjK%om`?+x5zu=lPb z{t*3HH5|!xx;Gqd951#zSG!-$;mN#~y4B%L@vM=Pa0cZ)2=xZD5&R)EjvG=$+J5s~ zCAi?FSKCJYWahe|iRe zbAy`h?tDAp-EztsgpymRZOa5U(-XQribIH_-A zo?5qUjXw@qeU{qgN!_|YK*l>&PY`%#TGg(DS$KwP3v!q+Mu38Q&m2y8i?2HnPhZkgV9x{sN&*8PycXRo(AdOkVfDg<% zSB_h15yEiMDd@N5arRkvM+qk%A@{GGygi}Is$4A6D!s6m#iJ(O1?+Vt}4%f3F7R91>>&f-5ojg6^$)xh#ZX?Hcn*e?} zqf^i{hLS({o5KP}2t|#ZNx6yA>%U&80N#d9+eZJ?lusq%r)epWZ&F`F?e4EALyQMbH*qhh7TIMslg&)NZTa z@QCQ&^fJZyFZk3dQbFw^dR5WIw1@qZTsE_)>NYkEZ2th(??5+07~w$&Y4;(iKm8A+5Z4k-oP9h@Y(e1sWm2(?GqCh%2i}y z3i|c0piP8^-b6$Pax(CJK^5hP!5%P*@R)6FL|rmx9XPBiDJdqk9rY<_9$m(uw!-E? zGRUPy^MWd#v*P_n!k#(2g5u&?tmT}sg~@HPp6l3ty(^=(@D9HYxov$EPymE^gi15l znxDj93SVE>qQ8dxg~Qdysax7*0ru%YQ7xTlGr{L&ADNMaqall&yKZ7boSHI zd1Q_`Es;r$a!*`U4QIkL!Jz8D+4^P2+9kJS2&zs&>;-Q4tHWBRseNW)v1woyi5rqY z3Py9*rjqv_hH>S-n;cb!zo&hnAXypMkc!hOUVXX2tDYj!bq^JIgG1Bof}@KAB#$Ba zR0$Yzc|B>-d^^-uVz+v1v&W18&IliuTC=O$c&keBH-@jY%_irvg&c<&K4W7MK002#P-&f;(7W(|kFLj=FNb4m zlEtKzuh8z#w`!M8@IQ^boA~c^Sfg2OwkF$^Y_|0{;Q9keQmyTjx*0XDh8}~Ze{0*V z#`aj#G?U#52^iYsdHZJ*6701^q&vf>2~j^*{stLN-MEos(KGf z>AWrB&k5XWOR2+V@Loctp^Y2l=y@a8HOeo9d{-=Jt)Sh;5dk??^xzJAR*r|^FA+nl zTYbMqz7az*xMa>a>7P+t^x;2gcJwW{^4RmQ3RsJcW6he-+`}2mAmeb&S@BkeH0^A( zX&f=LmPGRde7m^hcd4!XA>#{Hl&6RFAYAPvLV9Qa0IgAv1$eod5PUnS;Yi6alhpHG zk2kwx8*6jW{5hdpU-*Mn)u+sn#XDPD!u*W&W+UrYJZEpGT4=M$brKas$X%dgu*c(9 zG>ct-O1i!h=r`>H&hkc5JdjEB|w^pnK_X9k($G;#b?Y~ zST60+<#-xvOS>y^x5(Z5)nkFrJo8&ZQRUSSVtsE+ zI$wkI6mb?H-xKX`F+Owhlb*bauVZ^Im42G9n5^ojbVdF~EHcUuPJZzGM{4ivG)s%0 z3dg4EdU=n`iIQmgkGy;K_4cetz8FOu=x**VZ*9{)TqQ;aZbi$_J^90*eVMbAhT<1JjNu+!w@jb4Abz?HcZ1da+50@ZtM$$3IPEBva z;C*c^VM#RYI!l;9Q7&C~f4l+axvN5?q})cIbI$H`6t~cB7B_8|PZVxse74~r&-49p zSbiYztC^yn!)iigTXzM&iyR-wSEcxi;rEMf?5`dOVxD(c#4*K));+EpjGTTwYZp=Q zFODaQ+sM!zWmy(UOWa3>z~B#?+}4!UODmR~Y+)TDEi&>+t?-%B?p3hgc!-ZY_pW-^ zLDnU(j%ig^-r;v7j^A`+f_(*g6#fJ8alE;_n?;o^Q%<;yD3KkR*$4mtPJ34~ci`_8 zK#NkRewL4hSIn_pdb5bT-uQ@4U2*7**T{ zat2!`kN&x?$HRUOJ{^EVYil~IFlH}ty#A)R%Y7SOvDI!_^legDB$3HzK~hFeD!Hr{ zIh}LDHy$05Z!^o1IpzTz5z3yNS7NqWJKO4CXo+4~?x&98Mq`y#h>!v7004}4tgS~w z@lK_st3H)}c1yHC$&-&#DoMT{c)9#ts9ZImlSR9fO3^IH8U9c@9-YCYdEoWwU5|)<7++sn$oejc z_8Y=B{IkXtf#aMX=}=t!CGjQ2ww6^bR$E&bplr>89dn*>S*0f2i@I7Ita9s28Q|6~ z-GSS+=W&ydu4(#~pwisve`|ZW3qb1Iq%J`cHFoLnufzt<5A^$SG5-Ktfx`6QdebHN zW8y^BG-)FcNVbaNSxmf=S&nhvsI56u9GbR=p50Edc%|+V3&E$Dlq%t|yNsT-=)N4h zv9r)+k!Nui+M9mnBm=Z?D{y!}U)_SX?g8W(D<0e7w}?NmTE=LgLf36G4Z*U63}E)@ zO-9<>%`&%$wOA*zEn_RlP=O*vAq0JU{{Z@{iSbPLH{Jnk+fNx9gea=)ShvY1AQDY? zukhnrw}WMkt)zT{cd~*{)KzVB!Co*=4Mr{OqMuMp_@z{Z)BuW(4_-fu^`))K(_@Oz zJTGHorwMOiGG01@L-GcDRy6idS?Zo4g4ALr)5`@NvdP7Ib>@%a$I^sW{u!K1MhFu| zpnv1l%V^&Zyjd0YsdcPq(g?0COn6rea#?UNq@LfUMKvWWvsunA3Mk?ze+utB0j}9=x@=OXQ2EYDInRGVTyDMa^7q8|>`k@g8#(hNPPN*2 zJHbc85W@^9Gbv!sdUNkkWfXEADY~`M^}F=}&Be{kj{aUZ0lR_ye;T*pE4eOpJvLO0 zqKw^yRb!IF2PU(8XQXS=_{Q?;$3(t(jD&fM8DRYN9@VSii!E;N#2Qt`p`zW}!)q{A z0SbUfKQnuOjbqCBuVZ_Vc+$=bzloagl#T1AMA50oECD0=SD0uTbhchL(c!p@WPi#?M?zwT}`^`$jOuM*>OIYJsA^?IwIUm#UtTExg66zi!@crDr9K5${Y2h=- zLhZm|yT3t-?~6$pwRCLw!@@1Ed}KFw5v$olWgbxT<{z7n)|XxI29>FJe&RbFItw|1 zksIvk$jSB1b>1MolE=j_B5Bw1x01VNhSlUw{A3C)d`CPMX&kyG^zAbBg4K@SsK*Al zXUybl7oo%Wi%(k%?G_2KD7Mq><6({)XvK4+C9#*yjyDW;p!DbY*LmVyD^b)n+Y5_p z3#Nzc(E?0lZS@Dx*DK-;7sMAgY_^&;+*a$^VaX@C$n8ZaHs#E7q}-Y5KMtBJUkj?a z#_Na2>0EZ9sQ7{p9bEW#PraHwKJHU4^#g@tl0zPy#syx|bT1X&crGa5MzXz%Sq|BO zLVoTup2r5aCDD9Sccnn8wib862bm!G`R{}8O>;$Zo5CvEot%3X%@Q^&vEe}{CxKo= zH;k@4A>%DV)@yk#Em}{XbeUxtCoB2%_ODCOZS}Y``4d@NW|%k#ZL@}4b;0Os%KUHP z-xBMZ!_Jx%rd2*(_(&-tjO}i@2OW)RE4@%H4x{3L{@eIwWwCAX%RCaRdM0z4@@so* zX1KUmp^GuLz-8o|0ng)JrzVH4C&7f%4YaGN!~qbIxbYz%asJLLhLge`BZ2KDwb7=z zmP>FXX@7U+9o;%}Syob)x{i!HjD6-u;N3q%xxCda?_g2pTS%})Dmqu$Q35mExh(>G-d=Hy$G&@4q4-0>dbW{qXzyv{{`?tmz<|B_8uP!19v1PQui^=o z&%@VGY>eu%#~A(KEKYjWLAzO*Nh_K^1T{N-M_-N`m|yMC?~*C}!-9oKDo3xqZTSBH zSZywA;R+8ireaS$19tbUZ-c%J@eYUMIBzvg9`4d5jzuou1sPuTrQ<&j>sp72+Tsg4 z#zeYeq(n-U&U4ba@p0x$$&t%>tCjpOt*dHwR+C9A(%k&(xkk_d2kJd52U*i%e;#V5 z%)y%c<=My)jojdLu7gAHwcUoj9r|fku{w>IXwK2OfhVUOKU!wF;13mC_@`68g3ELh z+_T5~#pDyg>rsb}?JdJ;yA-ug2^)J$mbJclq_^DCMo&9O@eX+enxr)F+SZ*szDlB5 zL{>|WnL*jq_CHGL$AY|DCZf~40TtpAx9?Y#+q9{{?khuB@E)aq;kb0uX_t-&*CD|9 zm)^Q;*=t8~hOupGY?dpVOJQvzCy{XyDxOA8Fa|q%)%{mQI-i5B)+t17p8W8g;x9ZMv4=BD4Ph4t?v|Kj9s`{{Wzg{DpY59uM(o)GcQJ0EBx| zLXs;IAYtFFd*Aj}q5l9rlm7tU2DY8PINZ{onL>eO(O2&Pi=DUx6(p7@0uYP*wyvJ1X8^k(nw)g@=5h_34 z9Ou1nc)vuu{{V!iRK1lF>eZj*jgJ7F3=!$gEKe^a8u^tCTRSfd_@>KI@+Y$k9FnMF ztl+lNc<=P>R6I}P4Npt)1hx}hd3IAfDmg{MV07u4^3Ml&OIEb;Bv<}WSuJm4MjNrX zalB*bLHsMZ@y()YI?e0?V2-ooWM>1tdG#mC_bOT%ns&Q)6q{{zc=t=eArO=Pd8>Xc z*1Sh)r`u0wd>#{M9_BL=Ww|+F$0UK%rFqPr9Wkmkz2BPg^4pHw`vG2otlcQH)1ob% z)(Mb%wv%0aJV%XN{5`|7PkV6ye+OGUfRC_-QTq?Us8o5;;?+UeBnZ0@x zKJUbOoxEhe<%EwhNlc6y*N??AFM>t<*Ko#yc@?+E_kdA=0O{7a+x=4NM2_A@R*7S1 zTmg;BNm2Z%?W4ze;2SGR?ph-q^z5v>{J<_MC{&Yv@oo#7L%N3M`s>5L+COd6V+J80 zfz*Zo{m|U;o-tH>H{$I_TJZL+ZFLwl`C#(To%Vg+TN8!_dFzVe)5W7q)1tIQw{J2< zg7SA#NeTd7M;Wbe4M}Rg8u1Z^b0o}Sd6=A>m;RkU3hRun(i2)T_4cc*!UWH*E%8yl zNfOj3Vt-DxqIoe4&5 zs(*-~BRKZSsN=mc_!Gr*ZO9hzjC&bC#b~14*$!J+O=sg|vs}2h(;&KzH+76iSQZ5Q zr;lO!*ID7OiI-j@)6gVwTHT~<&lWS4+DK*k9CfdsbX(aWzhe_e8OIy_2XDr`SK-!( z*Y`L2?*b^&NMS+hROI`5Qqfo)+1U7dRn+`NV$Xl5Nb&-9Dng(P43bC`)30lKUx@Ur zH8l|SbCtwz8iw249y8olZkrTVI#s>2^E7f~Ox|8d$ZW20^b}uNlj6NP<}zh%!P-K3 zBa(lmZ08kO>F$f{*zNAMuq_0-+&1x{a3NJ=l209vy=PkZ`%8*Dh^%5(@`)^{NjV*O z>t0)9<6v|S`jEDKEO`O(6B6oG%jIJ_RbI^MGSD5QmQfr~VEL>CSc77u9-o0<)T`tDX>L_Em zv4(k(;}`|UKQZr~)t#&Bz9rJEq@PFeC6uzGe*PteP6j&qtG!*>?mBGXz+<0qJ&V;0U-;PIO6_K9v_d&i7?Lw#qZ zSl;Q*RtdrLM!-lI&pGMdxF7gQ^$BIf1_n+(a-;DT+TKF{0Pw5InFBC^^4RyUHHX9Z zlIeFINM@SZ71}lBj_+??MKq&0>g+8~Rnt5}szKq4y-M~8F6DzQ5Z&pADR`qkLzRu6UX=-bX5Nj0mZ1M;?4jDD5F_@38EZwgNx z%pYUZk~Nb}P=rd1SDnOXc1}4pP_l^9JO2O(`1?um8#ESi%LH)57zwtF4CB<)9w+f- zm4}F|^!B$)hPg%9`?x|}1D<+fyi34sET7uGS}{=!Y6u_g*Ijd?UQee?+fp*|lw;NL zkZ>`;C#HK)=M^h5Ztr8&^({*4?e=F*ynx!GN%n~TQpXGOk@$+|JXhm8Z63zneM=7u zvnL929~m7z#w&;Lq*wOZjl*3q*e#^k0m)yG_pWo|F24R z3CVAHr|HvMqAX7dCu-v@u!H)8So(#YU(U9-^X_&Si7LcL69OEJcjOGzRE(t6in8iy zXYh93HjmekSov-jtE57e!*vQqlG9{}N{{Z1~``780QCy^ZE2W zb`KHVU+Gsj7B;s8NXS+j1?A11%TG2^J4X^ZBpDd8V=8z6V~l#&9BWGH zDUI#1eou&+j-hgAxm|#gyHuX!`d1yK>Ut)V zAeH485~4QcY;sRbeJdlu*08sWV^)-=_~!x1`A~Un&2^_WWZ9cNAZxmv^o=uI#_Teu zDh5YxpN((mciO$pj2EynMGVe&C;_qoAoljJKGiKuB0F6!cDODvJg~3Iy!!O!y&uB% z_c~98ZLC>HSi$X{2`8VjZ4Wy1zO>#A6ALxir2w)i2gfTGrlF zGRcI5GKLX?23V8RjN{(C+rs*Bk5)5lt#HsNWRJ=@LiE-QA3hk+00m+D;aICd!u@%vd{L=cyxtkTxX+TevTZ%L zd0c-w*PmOkjy=x6XYU3W#djyfNDP?vU8QtD{w-YeB@ zZiJT7+j(TOJYWsXp-CAXah~ASJ8y`3rS*^YteSS8hjEjcHv{y~O0lB&UeM@TCC$LN zH_*mI0@&fn03X9RtW8qQ?G{@Yo;dW&xWp~E^eTFRR|y4;n0_SP+F#oXTRR4l;lj(c zWClF8FmihLtsPI|)|lGU+a-tETGWyE@ONjNW3b11^1E36&8pj~$>waj8Dg7r^Llou zuC$fWwIg|UB$(XXOwz=h<#WK#tqr905=izfFUR(Gvu)CBny6lm@V&hEW)$ReBL z&k)O}$_=~1_l>X!SLSeWlgC_Ry?B#qamTIO&uJWx+r7eViTH3h71?;!rp{K{AHJY)F>>(@^WDp}6 z&*xVM%@MWA>~;43D!J4xd%S1Y&5SAX!kb?g6jCo3fuwm#uK2& z;y)_!{Wb@>hSd`3VY!&EmjnrxZ%x?dyC02qUtsX3hAsB(B1y?y=LCf$XX{kfmJ=59 z;!N6oi@M7N+Z4{kQGk6-a{mDEx3%y2_q6{2zjV(7_*i(Z+AC=Yx3!9F&vvRyV2`*x z`d6#}0K%H^H~jowf8S^PDX63*nd@4W%2{dG7cv6L8WtdT;=J`WD+}pTX43&Ecy?3r z{SNBxd_+L-_4zBf2wUa#;=Xs)JWn>Kc5K!+g=1WVS0i`jSXFNJMaQ}8wl|t@_Pa#( zMN3=BnFdMOjz&rJt~pP`^jCN-ilYKKNprXJs{RDOmhSnjZcJJTsH2GonGAtr4zhv{eqJ-{U3~Xi zB(F5{-B`4nkC=ALk5A6NanS|E&x>sQthVcgV~&6WlbY|>U%Cz^VSMZ8Ux-Cab1itp>OqTWf)}-1lo2XoynbyVSk->ABK4QZozj0nO z;q3re8@t%9GR8beW*vD1kb8ILyAKq2+Aj=1s{*aX_xe1d9OP_NTBdz0H}riXcP<2|EsX=Wy-?c=nZbtm+!nSGuh7-@@&bji+Vv zbLgXjIO)b~xbZHenr*h3diJjDA(#lIQI&E|IOGf(IU^P3KB6BG_&!xox-8Di{?lV0 z_tkY?7`gC{xu&?1!6UV^oU#db5FDr^=O?y0`d6Chn%%9X%r}2$NgUgoc~ry|oUrT1 zTocx|^nVTA9TG?^?g~XM;b~PH0D^<%BfVugs&VGAb~>fC@D1e83XKv+552}Dz&!Nl z8LcTTJPoC3cS&z)6Gd^cZiTxX;ADbwdS^BB^lxixMr z^xJ(dL!sWo6uaHLvT-0_Mo%1cqE%dH6>S-uosm4cmw^7;4YYbp!)YW2l;@mh7;j4I z?5w;QbEIBJ4xY`W-9`ZCKp4Nibb*0&$bi zV_gS_Zf4S!;%iCKt);ar5dQ!h22q`-t~1uUDoMsS64dDzz}^KH2#u%d$s>%YV=|n2 z@@u8hEVMlyAF*1)1ZvsBK(ZW?4suVmc*lrk`&0S4gG)2Bee9kGx8Yue;Y~|Th7CJU zvY&IXC>Bz84hg}}rE)euH4Rxg!F z=Kvm_y=#7H5Owp%qT0vi4Le5=@5+&85A^F%$2N$Q$K~moU<}-pvgaR{2E5nD9wC5% z1OVIQ?JOiKu!GM%jw&AmYw_D_Ggu|x+ihLT#?K*kBOa!wntZo3709jwiTrLMEt#aNykU`=dVdGcxrSgDf zIM_V3Tb>E!L@~7LRcIW26a@+~69nbov@4q4aI+))v}CD{AW2NC=F1j0rt)oby;V{v*;M znad*C%HR^t&6dFEaNkG-OQ5@F)0Dk6G&gFJ2pcB{w*j3uX zJzK-lXnMYbaCGe&+TCMtjG`@y-!UJ;bK10yooc`@*zNM8<<9Ye`TZ*&z*fm^;mfN? z!ZD6ucaT9jB=b}18t}Ed5F|mFe(@(8Xlgw`xHQc=*(IK97&fd-97Vfv1abyX*0}9g zM(|v>#^!B1_mU`&4KoM89WlDQWs`lZlM$GVOiLb}I@iws01$OaEj)T&RWm?{83uNf zmBxL!tlRhbP%AADZ}8*zYT7G%OKT^LCRRP9kOzE{bKKK)&k=ZA!hT%dUfHEwU=rUi z=j)2Y@Ki3|7||}0K!qX8#5;^JKaYCxKNa|awXIF%ljKh`0S`M=6M;^XbG`ZpGU$8N zhm1TQ;)(6mEoYkTn0%3)xFGSydUIIwc;n&axV+LW1UIzAF;X6HC9Rl!^2QGi6k~B61}bFG(z)5s;q4~U+6TFg zGXi{o7m$Ft^v~A0&lh;_!8TgL!FV1^m4$ionUtv+TwvFff5LaKHPUI&>en*~8Bi`m zf=^*t{ykY^@vKrQ3h~?9w>^kZX+__3IY*TA@$oajP@(c+C|iatYUF-duD;6S!G0^! ztu>y}t3w0MS~51Ao-@-XyhF!&d{F3?b3r_8W{^87ob^-9blwhzZ}?Vra!0+IH<*af za!GOCmE&uPYV3E=>Ank1NiF`-rwf^Z`9#tNVc!`wTg7)C1n^abl#}UF-A63!V)9&_ zuYrX3QCzuWi5@&Gho|RL_7c%Nz4zQVqRvkgZKU@Nf zf@=9j)`vp>01~_+8aXEL<;Ds|@ZZ zI3N~1E6ug*J2?Cg9r|1so=u2QFu>1TewAOtjh}>mFe9g*ZT|qFQE18}YjeK;0EOe> zhF!8~toh*n;d+1HRo3`Z#eNv^ZLQRCX;FNU2_!=j;HfwR89o01m3StN29I&2qq{&S zgcmCy<;cM!KK$e9UXAd|eKj?&5o#8V8bKwjSyP1ycjGk*5Wa|Kp~`{vNpGkD#b; z^|NPlr|I#=1>0UCNgEHD6p%|}(}B%IXD)5DDp5%tlP`oPxnUdW@AGFLy=M36P^)+x zN(}cJXikS=kREHyZTxj%;oCddS!Vlm(uUs@gg()d0OJ@Z?+)PBHiP0F9{a-a>DP$M ztR&20nE^Y~Im3Qr51_4Vj8swY8jt)UKW~?H&4{})4>(Tk)QYJczk+_!4ASXCaljBW zWVdSbj}@kr{w9_ixNoMpxpheVh=E*s6P|I&tUn^fZktW97m#XkVUgYu<>wjeo`=?? z(#NRlpAmc=rt3F)ea?*67ZG4Hh|q~XQG?0!#bF5E$@X}%Q#gjYPip)N#?xI z#rE%v$Jo*BB5-FbG68|D81$LohkZ6veOWo7w(#5|x+ zOi1~wmYUR1%PoRhtg7IQXCF2zDcN)`5Wej73r~rj6|&K+-q%XKj@oyWq5v|bFdIDo z0JJHR{9Ew<0G+a3I^m!^0kCKJ;<&9dL6Lkrqo9=wEYp?@GrJ`T9@R9u51V%+>l9@X z^2lVZnY6XYcSd`K3|-U-1-VGH17o5T(h$2QZG!( z&RZXya+k+4f#nogVhyM($8 z5;BI|g?T3fj;ESU+n{nJ&eKBiH-qNBv9q$W)RswIRwj}r+6n2BPfoSgPo^cw3i^JT zB4;2;2{=Br=H3>G1@W5Ype&lALugdvat0Z|@7}xr02BDO%fnh*O>*D8V0^gHjl}x< zQ|3`uWb(G`iR1A0mXn8VBTL-PmSwjD`*Z7ASLeeJ-I!BHvoeCr#qBZ)2cAH!%_{Ed z;=&nj0{N0*x?=$J?+pdS=pa0AAp7>gHEZ6~O*D2^3E^9b7HPDd zHF83sTU<8psjg?lE#Zw*#Wyk^3tnC77Sfb?P@SZXGBCokwB1*1$s?>l2^VT+yPa0i z%=0a)2tk#d%J(O|K1nZ#O6L)Crg(N+k1FrMck&&;$Whpn>CXDx)qW`WYr~&s)2=jI z(8ARjBV<#xNWfE*=qqwNu^2%(Sl&Wa21Z*Q_^%uIt84bZh=_!RN4A6(CjfA1Dl9&)H-}N zZnk#i=Vk$4CxPCqYjGd!{{RQYB(Dx^WeimF{<-NXCa#9nosVho?}+>bp<4|%O4T%& zpx#~4426l~CnS0ST@U;$F9d(kZqNS!9{&Ie`G3f>w$pAd?WIJyMcTl&NF@3n=U6^` z-}Bf103N5z-ov$zx-}@CeL8rPZstM;{-2F`j-lb}%SkY!MrmK}`HV(A_)=@V@pxwN zey$Xd#?nkxZ%C36*{xta%n6G4d294o;L2si+HvSd>scChpL=(v&vk5PxKFrF=J^{r{*{d37|Tt{ifrD}z72hX!3t_no92}r zt_I=q2FXx8#aYz88reyCDvf+Vrb6uqhd6G&wTBj;sKGUaqAQV`gxpIp1Y}oB;tfVf zHQV53j71v_s?Ck9&pGY%s8EF0ytXOD#pqa%!L|ttTK@oRs4^#(-?}{W=~HWe4*WlV zdlb5hk3V82RR|?iV2%%Z!qauJH`(xzypJ-*z&SbRJoCZ+b;5X=wXIL$ABZ&TeO~I` z2<}XawnoC^908hi=+dIE9;C^(^*tGW6!=;RSfs#8O8_1+!Aa@HarxHIiF_$-cV%m3 zE~gEo5P(&DuDBTNezoG7ou;uG$#o{Ra~v?L$CbEsCj-#+tuGWvZg0F7YcP2)Bpf)9kJ7?(O0GITEY+c`W$AVoyJZO3(2J!|TmP z+R)kF!8N=}>|~UXPUD`y*OlMss0WlC;hCAX6=F8_8P9Hq@vmU9n&(aUVWccGG8MJM z$TCRBagOvxYJCfo98N+XhEHtfZ?{?*2^cRUV?N(4cRn8QhNEev$qn7a3oA}!4G}B| z#yRQDaCg2e)NZE`-rdJNnNS)wO`=BVa!1q}($kjy+u`ITZ)b*huP0!FQaB7gOy`g* zI-HV==2{KLM&;Llv`dXnN$!5fEz_Twq=kPB#woGv4~)29=^57 z!Q(Aq6v%H;_VCKY;GeE)=+!lA-64GY*%oKLjILbm3AMA+w;esJ7^tZ)ckv~|LkCg# zW#Tw>pFYa=$~4`+Yg>zkbYqqmwp)_>t9zPG-bMz?zhaD2ka)L@LBE3LHjXpU26 zZv|T1L}41l1;+|n5^#NKJTdV4&q>y8?w3|ux#Pf3xQ935?r82)7P9bIxm2+CdZPH?Uk>BOB{< zR$I0z;K(uv&s0M5t0%KlqBi*!XXD$e6`1IL;2gVWNqHE$f; z{{X^EC4?6uAc0#8Avp}7^!#ggPw_m&tQRt6r1?@&>+egIJwn=8i&yZ3cPV8Z#mq4X z?NSv@jKeCsPd=R2GathXOKVuIBi5B-jzcm76A}r>e@>O5<2^3IeN}Z0Ls7j)mfS{` zurbNnPc_WJ;VUbPGceY*WC#EuuDMa@YcFJ*dKVeV9o4^tlTFZ}jc0=4MaxUNWn!l* zj9}#Uth*lwC9I5PdkvCm3wWd}aDZ~3c9rf1kE2_SFHpEJv%5QarH()YsjS(2U#N&$ zWxtV7a05R-{c6%Sin0^j?qu-)0G;96t4%)V3q8r05O!@D1dQUm?tg^vNgF(Rm`3?Y zY?IN69BuU`wO7SA-`ER$H7P)8a0z_>014yTss8}NSFB3vVY<2~56pLuo7>y#QtDfv z)LT0r3V34w07=pAVr%udGqN;t0geWG@$Xh=(&7`ax;D7EBsTsE=r_rtpV*t^s&)udRwzmc|px$F09%=BGC?mOA0pH#V=Ev)QfF60BV zfO2uiT8mY+)^4Inr-I~JkT$)$?<48~tSVJbX|pMNtoJKDFHWQ)S^P-$ znvjNRF9oZ<3rhLV`%8}A-;HZZQKr4#rA1tKgnkfBrC-H(w>DQz1iOajY$^I+X0{*U ztd=%9jhkwgiDP{^nX-+&T>2dU0GyhN-%_-i<``{m8fGjTsU!eE>Uz{SH<4Xm%?eCb ze=y=VeYqJtjP<7~=_|8XQ*GH9w%-ISrj6RuRJW9F)9ZKjLF zSEhGkzSiS%a5{Y6^}XY*M)kZir~R8;zSH#u+WXPmEU3efO7$Zd=lau)MBgbBrzN?8 zC&CN4t>LzlS+sbdCmWVb^Y~Yw_(#BvrFeGU(BH)7Cy?bMJ5F0Y@N3BI{7K?_3$|!= z``u1PAG?X*0*&@}n)%w4&GY5zlbmHEbkP{6Mdn^26}b$s!cax@--e~-F1Hg zNh8T}ny|GlN|q3kM(mGYdc}v~Y_}pP*Tih%2*hjqgPovY0&~rIwd(keZxr2^*R?33 z`%7$vZk!Fd$nStFQhS|S`!4QA)%68Pp<>VWmf+(&5Ndgnb_BFNF7x4LgK>D;hM@~w zO|^GO(>PufE<%o-Dkkt1jozCkmtm@0+r;tUDxB>d4^^%KZhS+35PN?S>T$NnkyU4H zs@__h9((7ny;h!27T8%d*FGoI1GC_#+8Zbjwtrg2b*SFW>VHxW%LSQ&{{T{o)ufkt6jpFX+z@(IL*NSqRVTzv zw}1x6!T$iqs~1YtE^hR@mbsE?U2ekt^1uzeW1cEyxeX!t*cI$+Xr$4Mq|>p~T6i|q zT3D?$fXv8;=0m_%1o!FEvb;<1{WV+6%dbNUgS;n}PBY(w&1c@ut>Hfq`D6xL0&=I3 z^AF0rLfY=}t`^YGayd9DBV|YD$gJgL%8cEteGw1CNz%j1i&TPn6$k;4fO0|SA4;hY z!@2Z{koorO8IVX~GCKkk0uW9h-3hK5@{WYVBXeGs?Mr zZSAAa9$4Og7CZ`;q9_eo%nHszvI*BvTh{z>|VH4nG?0H9ac#&X!%su{4JnldgTSSTJ~!=JHZ4?x!J9SmF8;o)5ni znvS8XJ>Ai_)aM>k{{X8G(0f$Ytr~H=BF&mn={GZH?DlqYOO49SD+0Oi_iLT;Z^Lb0 zT-3&*WV1vNA~<(E6&&>OL^=4!`mD0EOOr0^-qSyRgsM=nqmw#|tNMm4$Ap@2E zwUsLBqiG=6&bIhVsI|&$yR%}z;di%B`{ufvuY|3sY5H1RUquAgP_%ehw~k!51d;1p zy`POWm62LqLU$R#Qz8EVKDD{6c&_f!&rX#!3zZW<%mT@RPU1PwPX7RgezyMrptN834gUbyRt3k3JV!LfL9Aau808Tp^!{~M{u0j= zf6#vK{{D~fsdG`#TBGQVZ9L5u@jty257NB08!Ly@9!XscMl}(jPEd}#``4>#S8HRZ zPju*AVqL{Yaa>HEF|(3iGwjzW2k@(?U*%f5iXfui#M&Eq1>webMA`Jn2DvW>nPkxJ z&8p0Wr7_B5ISeo`N23btH18MMJ;K1YlQTIde2k~JPM+0@J|fu5dLd0d^4zy*okE;< z=}_%GQA+UK)cY~iC(@R0v;iA@fw)i<4p{o0D~a(&psC{RO3B2YTg-P910y^fS7~YE zi*L0k`z6C~V4ysTLge)vS1;lZh}xZ}h_xL)Ei+HqZ#j)+D0h%Od939q(OlAEomJGu zi#EEAqZD=u zvEbz6q3vBY-MrU68Pklhw)t+Z9zhzC7dXH^)u(rEL=yz6c ziM}KFuftkSn>y+GRo%t(j21XzkZn+^-8XgQ{#8VaNXmW7b%EZdhOc*}jE6G1M zIM4O2pG>fy?RHk~vIv^`1vyW;FxmeA3d5J;pN-Q}vA6pbmTuK#97(b~_G6me@YluN zQ&N3SJyT1zA7pfmKFC>1ZY(qN6aCtl(o>R4a$LDqrdN%n)O78At+e=VCXhD59Bz-x z>Nx|3>G^f5o)#CvYx(YkhIg@(V+A32BZ}$2;X5WEe#?G7lP-FHyHbC`ek0TF+*!!Y z5(mx9fr0szalrSkhlZ)x&)dY}7UnY}A2`CS#58#O-&)qQon_NCRE!X#PME02B;)@8 ztxmc4jXa7PEiTD+oaK1k?b4m5d{Vr))?u?cMT|=(;9UvG!jt_fRuX%Q+M|-U)$oWj z8ysUKQP9`7dboECyt!vwNc!D3q2oHI+AX|QtZcYtEqf~ftDJnswKRQCQN7nROIwK~ zmMCu{WElIuF#Ri|(>!(JzYyFpi$>F7+D_(3k0U+FHFh5v>-H1MWw6r>k-pNVHshWM z>WQUr3Zriac!tTsM1|TtQjSl++csZj=!CIJ=cou;L?2ifg%{-S)a*a zlk$UspUSJ*_@>HxdwbC}(uY9eDP&R9bYV-C+|fM`Hu#BQ;%n`8!b>&zqqR`aw2zx4 z9zE+9P4Kq0EbRKU&E!4{#|PeQae#Bd_pe0pzr`uMIdP_?rHnJhdkZ;dW4TCeyb(}n ze--q%xS4gisI!&`rZE|D^&y*u;ZRv{1J~O*B7zA!*kCstz zGwaCPTEh)FE2Ak_igsocUKN_p!@|Va-8y;u&&tKKx2;i&!`IQcFK2x5AsImt=O^{A zQP6eib-Qb?w8AX>vJ^7DNzc?&@Z4J5YIizxTAk6kff3{x7*%eVILA1wDp7HjnXD>Z zT^>Q>7_Ric3~CbF>5{;ijbRTNILr3;uQR@q^4`MI-aCt_?k)hG*r&@GIpI%VuS)ut z#MZ`u}^P^{{RnAGPj6W_Rrbd^!Y0(Qghjv)J{^l!st-l{{U~>B$|c9mg91`J4RP& z&Uh8O9+#%eYl~-@;{XG>x?|8D_0;J;H}GDms@qy=`s9+`PTOLPz*lkllpoHzooYq2 zv3Vj69Tc!wuru}G)hbFV?T!h=@vF@;C^c&-wE1O*ZT>WA?c+0G zziXFF4YE0qd5tR$KIrUgcgFt!6?A_KYOAJP#i*IYVqDv80}azTq0)oZ4bAEEX(5sltjoCIayac)-{aMr zke1q3uPM(Pra({rJ6_49-8Q%3>nqGnYS%ENn{)G~KsY}2F`}X_bH^<#=S?mfW2fDD zAvTnF-H08>PTeb}&@`iMp#7fO%6OymoP{dc83We4D@)sbLsPo2hW_`=gYSHV0geyM zR|k>nTDpYd%T1bMvp<%E;vttmLF-nbZ9&TCowkBj)genMR3io$L!W+V*E~5suc8Z5 zWq_pd$Q#QfoGTCOU1pJg7Okhta+7Ls2$-u0lMT~6U~|{CYCux{!3=X;#sT}s3g_#Q zLygvio?maJYL;sx4Rd#JCj)y+0XWCu`c!)5rQ2w?&_ii(?c`)_w{Mv5UfAZhBlw45 zbpyN+YOZ6M;`_)Ll{h~r$^9#!@b0SmpM`(7CKqulM=N9`58Rg$?e{XQzRRhLGeLw>_t1kLj`Sam!m2ZFI8|w@7 zxP|3HVSE(?f_U`rn(I7O;Jb}$NLvZyxED6EGg-+bLu-TBu(gQfTxQqqKjTmF6(yHmu+}&7PPjPcHO3E0jI3wGf zcNooICbzj4JTFH#_8uU6OT9AuSzTMgJi% zeC#iC$87X%b5ih*j_qM7bw~4{c-c=b-Mdml;k`oM(8Y11-EU<5+eW2|H~BT{J~-F4 zJrdH`>6Tw>l4sgcTWbsq@%dLfW#dcB%{~jr?hS~M-0b#9~4=rU-5}=D9naD^h`>V3!H7@%z4eo=^GhUhQEtnx2ge&fZLuCP|O@ zPquSJma7Z}p8n+#K`iCeV0-cHL#KCe=kh!%&erPBMY(44)#kelUDIZMQ_y-Wdyfok&{OqW21{!xxGyZtlT4pbI^^apy)p;;nd#gh>v-1Gh0Yi;Sk4> z?Yp;H`U(qwAsaJGFe9vlAIk!$+C9Dh0GMIExXws1oC3dr6%wHtNjA-6U8Q7r)`_LQ zz2ggc?c`ZEY}$@TU@?xk>FG&zru;!oVtHZvafi%utDMKTJPcyJ$_pb4hBq_F z0O;fq{HfF0nOrcrnml$!03Xt+QMQSh)Trj}&kfdWd~us8bdfc>`77tbw!{5zM$Shd z_55oGP+uI{__I=$-%Y-n1eK&hAUGqBx^enf(SkIJ@7Wqv z+%_J(bJ+1&(xXl-xQ=bw%^n2?i>|z?r%Jq0pS)pjZoGYKq}F^rs9ss=5G94vh~Q-; zMtD2P=hD3@KM?40ItACICw6%_&rUP9rD|Q@;S30EAM1`Qvr}0Dz5NlWl)@EcYtY$^=sEQ@C<452w9p{{U*w z=#`srVBpDoPImZi`2PUYBVHMMGnvs>?Hdm)$&7tE*U&fC(B4@~Z*Vt4>xUV}MRLc& z+A|UpN0vzc0EO0bh90Blr&bb^x-Oj5Rhj0#7nHuc7nWSXhQWODaB8HfFmWVl<|e^~ z547jjy?VpKnoM&E7fzN5Bm1SME!QKCYOT+NbSq|9qq3G+2+qNCwEm{5oElbQoYQ(9 zKcH&usCbM-Z0;ZtG6GzkrL)h{t?MvDCy#YzxyXTT*vXE#&MVg~d>x|}^9_vfx-h_L ziOC)KJk~C=@YV~>M(+J??#o?DU89UA%b%@bilZsPMkX^{$By`uRkX zA&@$OyBoOfYohTLtJ?fP)U^pBF1~09f%5GjZYR0NpsoJ^_)(S@db(cEso%!XGO`9l z+CVwLBmDDS{-@#nEW1x7yU8rXh6^x4PjWJ9=}_dRms3i1j;D*?Kw9EPwX;cKZUl@% zqWgi={XOe5#BeisirOY9&HbAI0p!{b9=(rR_UL>&p~(>c0Ev5;!u|1Z8=&q%to>8K zT2{GyC%CY=*kM?gY>EH_q1+ohRr@xTi5MPv$DZ(7{k&Zz%>)sKD z>igkEoOU-qXStCkSmt4!r_E}X^WhzZ!o(wON+6_VL5PlkV4ex-TUP!Q)U@q7*H6=L z?`ZAO!5n#30zs3W53qj?$l?>%dv1zuX@_K(Qf=3{g*L_ zYK!MhGj7TVP)4DN><4aZt<_)}6zlz~;^S#FqjIda*1?>5{&l+k5T9P~#MU<&{FB3a zne&$3FhU$~3h{%C=9`EqSo25t%>;6JQQKRHvygys`O_OvhWAXn)@`k=;f85qi}zAz zYcAun9e~djrr)#HnkLva4MmqY-?lJ5y?(XQXuk&Ze-GJR!7TUMls62A%H8GdWzQo7 ze|nx8nx$1uyV!MS&j%EaHqT#zJ%Us+Df`*LAo~$h$9om+j-wvawA+GI{{YuTuDU-5 z_%-hq%`D~nM)Hjp+kRyu)NZY7Z-=@lhTv&$1-|l!XR)l^E*KZYFkGyy@+^&m{^M7wTzoLo&zOVl;ZWsd4IU5EG-=-mbm*MU zZ*P^zbqglL_kkJX>sdy;5njcfYiF%kyT&dqQF3?(Cbg~g%^3KC#?sNt%l4F3ecWeo z=~wk1gZd7TeA~Td`rS8eN0LaCIrhz8&^{69v+8!*rKYEP{{Ux^WegrBR$xKF+gi&P z7Plc`tjKh4BTbCu#9{k}#~JmaX>D(BnkZPRM}}#M++QuzCb#ZB7{aO}7kBf5OJ#Nd zJ-Mq8{5OX?7V_=@03?+|y1aP8N~4o!rsp0@Bg8yGbK*(-ORU3V4abx&B;qJ}3!G$e zn%D4!t?iDTbsnW9&zmU?3d~MaW0G^yy<=JUeWymZT8#Jie>3lP+=>u)>N-<2-wNou z^gH0bykj0dZR^+91L<1bnR~5T$CP--R$D(3n|tV5TPL=LHxWMBQ{)_j)AOtPPN!rn zF=cvgkw?h4N)kiQKzrAzYab0fEvn9LuJrk@Zidy6rOP2WP%=PZ2LM$_{uOvrdMEZB z>LEOzwzdb?=BZJ4=1hFb);Jwz+fLPdJ>l(NRr9>HNn;SlfOF;^d8nGp!56Z}Y|v^C zB!Hhi*e>2Wanx6=-h4l?n?<+L?5-rVuz(%0?Gh7{obGI&*06296WXo8&F$hU@Czzo ztEvu@run`uge}fzQq>QKydj|3=nPS3X;pmH2j<_&`LltV#?#YOGtT#V?w@5WguJq~ z+>#?H?qg%T#v^q@Cg$8h$f~mZGw}7!pCz zUEChWCl!w~vaq>#Xyi25@2|8iZpOz`OZm;jVmYJ*&>g3sBDtLx#uqm_^cq#ZoGtEA z_Q`NAN}rRj80az3*Rg274D?+iPLe2%=@Ph=l1V{eKp+5dSQ<}(JTIg8pZi7$F2>tQTZW85BrEd+yM}ISk+-q3yt)D(5yPk3 z!b8Hb#@m#P9I@zHy}oY=={laDEI0Pjd2|{OGa~ z;$8{x=C3b`G_6Yi07+YChTsyBTX|s@2a<7Kv*Re{o@gP4;h`QxjB)&U#}&}Wt5{1U zaizkSlkSgP)jeXxuV##_cOjz(5=6*2Jx4!EhfVUeeahz4o5S&l{6Atj{{T|CuOj%9 zr629sm?N}dS#TFBhm)UfFF$!yxqdP?D)WGSIK_JB#T_G4zgQz%_e~}^9!Io| zitiENoyWQK?OheG#2*M2!PKl^*f6KeU(+2cXIgs;jWkDj8p~;NyGwb20SC5nI{Q;; z^%IrP3-I@bd`+lXUSGqe+UTf~szGxUKujCW3fRHpjO6jsv3@e#{e$8xi6l#KR@NX( z=jJHP}R~a;ilY8 z6rU{86kx6Ot?gS}(7ZLKM|-GUL1||p5hS?CVnHBq4`59R13YTZ3ynt971U*lA$vUG zEtQ%`{{VW;lB@FrUcumhE^h^CWgB|50I@xCa!4cBC#7;4PsA^Wo*NTd>AI|9+F4Ay z0Xt(Y)GkJG-yZeVXg)UZw!3<4G+j>Kdx8w6IFyX~o|IS{xnE82Zj++kYKa(g7Nrpl z8=e(La--XtxuW=X(@5~c>H1=za*)q|mM8vBcOVOHPsG5J4X? z9<{4xQ|ys#7)^s5gT^@IP%Bu+pTYV){e*B|S=-Md#t_V60FnkWD~OB17TzB5jhyzj z1(W7{#F+qb$<)>@n%WO7Dxk;8`Ib>tjWN}6ut zO{0sN$5Q^v8kB1RN9W5j9IBJSJYe341* zMUQEa9pQWNoK_WKRs03hl{CHMPsGtIt7OFRvJ=Pqro4B>wpQ_6$ksE-1kt+6{##{3 zvFthmdYbl)TR}RE_R_8Fuq+BxBB&v~3D2c*_CE}4{{XZnd&|qorD;6u$+H>lfDJj) zl%;V(X)R6E@dD_c6VoJUT3Ka>fIUFwvV1G5UiiCLyf^VV)2$@=7m$wKK|BoO)1`Gc znj{ivaq03$Bx<;5UbrNlmouOQemOZ$yH`H~%0bO{e4R-GR z(*8C=h$a{;r;Lu(te-%Fjs4@*FU7&LacbC`jkUw-Q+SG9ddxv%sU@PwL$c4q0D7Es z{cESuJWr-;&`Bkg%DIGQRR=phzm-^rR*Lr7ZnX3lmhLi=!s=DOT6!}*iW+65)VhGZ zxcf>7!=kLRWH2YLF~>^hul1Sj^=paCBJPl)F;K+-IK~gIJ6E790-0a|BcZNt`$)FE z)MSv_$WHD`$eUZ=9YLv_8(I|>Z0T2m{{T;l;~2=n`i>4MCerQYn9hh2VlZ7sGfcB+ zZ0%SSXd;DrJb|zWC!wslyltbSVmZXCjP6s9!(WBnc=b`ikHi4;WyChk@ z(I(Q-w&hg>1JAyHO71=(=>t-_@cJOQd0G}kiYW*Ofz%J@S(cs!(8Tbd)O9Z^BfM^e z6$d#z271@4Ky9V9hVtG&v=onQOa>bqWDNZ(eVb{axopoovG`9N-q+PGW`zbe#SE>x z80cF+%DX9SVbd*bY;E`XFhmhIv1Y*Ol6uy>H&aPA$Fi0+{`(ap^fcScA#iu@ZOS+U zy*EzLvJ#!$*~#ho`+leiBnySd{qVJ;Qd((7D~aFi_bc{^1G#2g1Ask!t4sSTKl$k2 zkgUCwi|0n042tI^^Sn87WK=JtNOK?un^yG$`<*g7bK1H?QA#6aB2BwU!41%g>M<20 zSrJ(nGO);D+*J*11k!G1A#)OM1op>2(x}?o-f0@VR<^L2^qY;%o!r_tTn|(}osYF@ zYH#HK@Y_yyoM)Qyx74*)zn(bnu8;2+1riUAj@HBhf+fiVpnX%;o)aR~wt!T|fd&EvHoT_S? zl0F*xcjJ-foK&9?3C@OIe*z3?&pJ-kZ@eYf0E-dt0kX|N+P(Z;` z{dg7b153BmbQrAl8(_%DHzV~2`Wl*3N@`M9MbwSeju*k6JwT5Q^~9GD1BIVyB^x}c z?~Z-z(xRR_h=kC{u8g@>2ZLN?@58++R*Y%44QJ2JtMl$1&nJUg*AZN3x_z#fqglxy zAdfkZ%NWljpHrHwRO#EE8}>7%q_k&bFosb1p+Xk^scFUcIYjcBXJJ2jN__ zdHH?)%^#7?Y8INjb0GfGjPGI-5#WseHPA~c+3D>p@5E8%+vqtp6!0wfkW6I(V#z6= zyn9to657wGXcsVA{La99pyO}{Bxl~WVmx!g*E)^w#H+|;(kFuANldazvf-mr0bWVR zBLx0c>l!Apd3~>H)>jCED|a4IUZ8a6`TjM}X;*rdnF&aIHtl}=J=3?|KIhZ%uAjr2 zL+R1nTwKbs-d#TVS$Yp@gxg~a?-ObE(>=7SBalis5uOe+kIS07qh3X*X>i=Y<>7pk zVhAK;#uFNi69tldEq zeUTxs6R-d?J8s*b;y*1}(I>K-#+v8Yo)ROD6_`fDnK>kp^v5+<$DR?>{9$Pf7m!?; zV0^_S4ybwS$?enfr+6E|5O_}BLM|S891Q1@I-h@PusFYqe-Ll7KFq-l#gGIaj{KJ)>t@di7od>MTmq%A9uyMlEK zkXZgz&ip3Q=7!q({zW1=wj^GJ4o~DxE4tPrd9+*DAy8#_9Z`Kw2o;y$dt1#r!j@K= zf`qz^5bKb8liHV2oVUfh-ET#*wA6I3-C8MOc`PgYzz#O#{{VNbaDET?rS>guF*c(- z+I)?PVYqW4LH*!QJRWQ48@p&PZ6&w4jbwsUDCol=gVww%@8Ct+c%oT!jYcRm`7?%Z znUEff?Zw1U|TT9r89PP)={{VcEn(1w{ zTUd0vTNzi*TWFRgcl*QvjFVBv;r%TqSp+CqPCz&`CWOo#aO)bkfo7PaTHX`63}X(d z>+DabrE&fM@eEpOu9I(lCC$M3%FV}^d#ADVIrcT|Zi?ttfGS1^Jq>W0{{X|iI_ci` zNp-Wn&hP_#tbe5<)6&MDinW$c48?J0a3B{F1m4J70|b>IcjFc0l3zTR3?Uv|dznl~ zc9d2ium!P!k;Zyw+P!;2(r+cuXSLPi^5l`p!nWe4A9x?HHO>!*(p&0FJU6iz)f6m> znUFAINg$qb2?v@)ofm{}C%f?3S(gSQu@9IZD}jzhVfePqZncdTSzDWDGOyRnjDK3| z^vfu$H0HOqwh^Et_66ATP6tZ!ABZdB`>lS?Z7W7;Eo`3w$0kBbw=15RQBy68hUca0 z5*YkFsJIy`1Dt*axh*Qz)pQ$s%|bMiM~!j{?l~hk0|)f3olV}M;Yk{6V;$7?r+iq( z9F9jd%xQ;O({0k;eI8qdkM1xlmGtZB(;}MFRwGs@Qqxzlg-`^M_acspK+p26{ZCl8 z(_+J&(x1B+>P>Stv1(U(wWRXcOA4#<;d6BkNk1S4k860;N^_aJU zqVazZE~cwqQ~#tE%v7W$o(o_(C`G8Mb4f3imgn|A{pbe$~+ z_EPA`HpP@cgJUnQ@A7@C$vz<&q0ukAEM(2SrlSP*+m8_|#V%K#*&nTXGkBZD$WqYw zF72;W1>TdV*~#_Y{oiWlJbB<9c3%ORlHK(kdj9}Yj^gQUS0Xf6+$t~oE0NEsr1k^F zuKYaqsLXDlC1Oc-?!#fY>G;>8{0gy}%i==`K-z6(RbX-(1ZKFy@bAW~**+e>CmiJ~ z(Y_0KL&SO?j!ovPq+QzE3t1FNDq(ZS8L3K3b|jiT0bfUp#TwYxq_v9h&$&L`j5HV< zHe}-*9tV8pw>}?a_;N!SkXbdvhR=Eb0G2uVaqZV0)yDi;@D{1!jb_hK(_x0`R@}hR zyqgSXIS-!wdRL|Rdcm}<0>e*|MUq>15;kGbkU60(O=RyAhw(<0_U&HI1cukmXe4ho zcveg(WEcSbde*0e?1JcKHHtXc{h?2mp;Wf!1mJYdL*u(26-gztcxpRBRT*TF+$*k1 zo-#UowYTBBO;R5ZBIb3VI5_*KwRP4W6!A2(NSd9Mv|)*FHaUS-`ma%1nvQ|u zFYK9RH8}pBUwzU_=0N&5kO*m2mSm%c-Dygaa!X&p*U- z_4MgZp3Yeo+>4tKL}~j#1Z7YxQM)Mx8=a#^U$7p zSCZIx`u_k+)&Bst3r%r50*o#R2M3Zd{(4u?`i76DY7oU8^_9Hw#KGedpkSxlHRj{t z5>4YnsNQOyYSO&JFoI9NxSZ@Ky65~TOibytPZQqF;hj3~P=F=X?&A>)v=#mmM{J7A z)O>M!rrem`;?OYUq=aW5pHF)4-qzkoL{dW=Mg{`1f&i{={{X-~8Pv5Ae>IpFAw0dw zAp;*yl!kFWF!4ic{sFqWngxYzuMsycO1H{MKZS6s;9a^y>N8%(OGEaAa59V!BR+>8 zm3p_ty#nX{75eF#2AXCmD*omr$=idI?_O_X;V%+H;n@89JyJ|Y39>FiZtM4RjPVkbb7qH z&YeA+lSr&Q)xd7OE4hl*Ztt4rc;gDHevyxxBbc;8Xitu;MD4IS11QLTj8jnsqGFg*0<+Np^a{4e7;{7Ywd1ei0W zyPd#x+upgD5w0XyBLKUQtIvAuhL7TyqMA6gjV{^-jkhhVz_Wa(jBOe1Tm_|{i>={} zB=E)jaw!LJL5z;{oZLZ0Gg8~-zJ+^S|dJSapfhn2(XopZq;G{UX|k> zE!1@l3d;9ZDzhqINReB}6b3@Xk}-mL@6CG-l=4YszFTtyh1fD2di5Tnyw_Lwbr!F! zTxyqBx|DM7aU3Eo@vqKEBkRGY!5(30@e5PA)uDUe+}p~mq8ofi#tsV}tTSHyp%-TPNi`G1Jn$dr6IL zw8$da%x=W+oI2|>JKO*EFWT#k+^MKbpVgXfI0{s z7llj=jUF(i4n`?wiaDYwDyt#t4@$_=ye+0^(koc3@rCGycPc;NB;v1L6FUnu{xO*k zqW=K(=mI@HS!a$(O6-K;VN7*V{9ljq#d#I?iS?_S>1^9nlVP3X@*x9o^aYq>742I> z##8}-Fa>x9r19HdOCIEyJ<(%kIE-ZD=5F8TikUDrExbjm{hMT0d?qzjiH(#6b9M%fdigv&Tj6-w25mlTZ^b0FhRrHM6Lbf{x#BGYWA>R z$prF7C0u1=&JW5udsC+*o}{$;8ggsiB!nOA*{tJIbLBO-8TuOQY`jeb8eOVuasWgxG zyBhFoejkeAyn>=o<|Zi-;Nv(yg=T<}j;^O1X zj}6;yDVmRm^=sb`N2S4ZZGnudnB?2eeTGQKxUA_pNj_}?t)nYPRBKr_iFtE07K?8r zkl?sM(~>`!HG2B$*47X7whI$*Nm5tw^sZOJo(G%5i*t+r0G{A3=8O_}&#%_HYmGUk z)9x*e#EfP_rJVH0?TXRqu_)r(#u|OTr6=2TH@4Chwk7MR>(3>6fN|El9Xn96)U;*1 z)0{n{Au+I00Utx!ylYeN--t9#X7Xu`-Sw=$x;u$vWp0NUAY+d~kzSYKTU}Pe!}CL_ ztWm^^w%rhNs0KO5uV2QKwXP)H^(b5GcJ5399Y5`bh}w5;DmwFywY%X>MmY1yuCL+s<<-`abeB4b zzVc<+9HmrXZ3iDO99N8Z{WkdLRgqhhJ01@zK;peS;N*`lhA)~&W^c8TmO1_vuDLID z39S!2(>zJ6YF9T8eW={XR#r&Pr#>w%i{ zn64*JYnhP^qZ98FkTOFzq?b2p*Yd`0-5tLQ#1 zyV4=j^!7-$gqmb$W(7k7)DHDy;n&3btzX0zT4tePWF@;&1h&XkfyVAZ;}{*PugA>^ zQu9pKK-MD8U zismgm7vej@1X^sWr}(7#N9$b)N0-d@u=Mb*f|X6ran?LoY&5$G#+=tbZi!qVZU#Uk zAHtt^uG;B;Ws=WLGQG{eos=PVryWn!is!sT;caGpBF0wKH(J`*#}iqkc|K+UEEsc+ zI#eDd(r@(-3;2A;r`)x~N6L|71zA_7PCIZbEe{#m`W;@etic`KmAf>`_fdvbT;l}b zXP1hLudSotK<8-`#%q9?Pm~lmE?)~<#W*FbnjkU zqiK-o_g`)j%_Qr%EL(Gsz=QK1bIoN}5UHzMP*A(op8mcmw~I<^tEZ2960o8l^`~w? z9_Fw|j5LU2g68HMJ9}9nCSNn}h5+XPb?7+vt~bP!Nqg|!oh@z!!oI|bBP25w>Nw~J zdh>l!>g_DF<(M!ha+u@<9x;>ZGgIwk>}v?Z&gapd4e_3tu4oa=F5*dIgvl~(*dtt= zA4>$!mF3?Jw5j#y*7TS`QT8CPyNm;g0StZm)$L#5 zzPo2?&uwKbW9)u4 z)R1alW??FW7$YQXQV%sUS6UYdS)PsXzV7ntNWWX_k2Yv-ieqRSI2>_Y4~ez^02NtZ z%M^N}HLE$o$fWU~rFOpwE)snwQ7Lq{Se;3canz{b;=E(U8s*LAu)8b+EF=ykA3v$b z)}~zAa)W8M&sFfxiLWkodkZhOv*JLLMW&Ue+{$?t>Co5}84OKVQnd&i3C_mr{;Yn%y8kN0;QWZ2I-ZXtp4U^_v|c zQMx5Ym4E^`@K3R=U0?05rAl<=g|2qx5@2?>vBhKF=*Ib@l2??tB(X)t>YK!uTJ@E! zxzQ|A=$whS5tL9mfN_)ULmDLThNA|pphJF^$!{!~S&@1NQ-fOKJ_pe^LK=zWOXAFcN%93{mtj-t0`n{&F z;!E3CwvfbkNor(;LXeGvf)09-)~NhU)x1TgYF3G)>O#)dq1_=HOJovw1ap!6t0Uos z-K5?piWn|lSz)uA%tm~KCPe5l$L4BpiW;OgUNh4zu3=Y-Yh`6-&&uK01K<1Jl{RvT zyyBx$$0+*ej1u<4?3$#m?2PvkEGd!$w&T-)1}k>M$C~nJam@4msym4Y4S0Z^p%)-v z4&>J=k>0e!yY=8YfHH`6p3MrVnZFgXLBg8mU&5Ag0p^=GP?)hco^TjC1y~QMw(DjpjG;+$#r&?OT z7jbEb+`~BF4D;+wXldRjvbr+Bw(`Yo3^F{$t&a`G0Mwx1?^L)l?ik9J9Ymd zb14$K`i{%%&2(B##+`j{_R9et2`Km+s}8Dt!LAop{?hRLg8NBHrqU$Jc}osuw`16M z+IxCb-wZq-srY?wFYm5qh+duFVspEk=X)RT9{8qN$#|n()|bV)rJ2_ZD;(uxi^)dX zy+=6rt?!3=?y)wjeW%@At;Bz3Go+0Yp;s6f!0I{uD~<6)5KZE3ertr3nC|4;BCz>! zj&YIFuKX*xdAv!g6En)Ir^&IOgN%yol^FXjYm@Dpe6~6pZ;4ah$K^|5Z+c*VE#w$g zE;u9u{VPcPL2W7(GU#4RZsBtrfQ+67ab6qY&0O2ZBqrYBVNs9BR*tuAZ62_72yY`^s#Km?hhRTCTU)5_rQ1601Y}AVJX1!8saQMRXjbxE ze(~giOYR?dx2%b4#?%*$;YO&bWoRmEy)>Z3}xRq5pbKPtxXjkvag^a2j)^PGEC zEre(Iqs5n29w4>QEHpE!+}uFZvU$M(P;uN=TwW>F#mmP301xg2umH`n;xn`!LH!Lw z;(2vlBjSy|w{M}|>Nbm~7&C6U!xqo-rp4oZP90A2>R$(IA!kHI0xXfYInN&T$5FK1 zjS?W&z9KUwiFszXQOBr5ArL7Ww<`>SJNK@mMV=i)PloKp0eJvz(SS$cUUesjx9Q#& zy1iMf&7)3HBwRLG!2yBi6|JZ=v7388wEd3B=2+rOd?JScu{m7ubI3TW*Fx;AW7O>A zGb5~raOs0tdeoYJnE*tyic6U?V~ZHh2P}O6??#<_D%sCCMU&keE_#?y~C~qG}r|LIeWrX%_=>9}&xq^d?s2vY)=UOdydA|>F zfKGW;>&qx`g3UY2I9(0}Ry<1o(FAPu1j_eMP4q zW_8)+1ssaT_>q6AX%Jd1fbt`dkoP$IzTVZ-33A1@jM}?aI=f9nP)U|3t{yZJV4wgG zdgJ_O;$IQPtw(JRr!-JXs;vrbBXJ`jaa#WX4yV+0T_5clklf;85?o`<9CKVpk4>$X zy>SdG$#oe=4!i)w_4@Hjw9`>(6$sk*8FS-r6lwEGWfr$_5saK8?NT`IG4EdR`mWtR z(iv7cq>|kpPcgBQK=!YgCDiYwylAfH^UxdyX1i+ngtA-mvB>%Gx$~% zQR-4kn1b*fz;5!u$G=MHyg=SZhA+$f+iln#-T0-4l%)OPOGJ4;fxK6!YJMoUv$zRy z6zq4d!SZ{I4lC2VRd;h~X>M($Xv;ZMx46Z5ufqM1c&8gc^COW*9rOPH)~^2m$2V}< zX}0suNR4H|AZHmKg1alH7H3V}T;v~9zKZ(YVwkDJF(F2HAN_i_{k^RJ0MJ5D(ztoG zE2;d%R&wfGD+FwToQ_8w1}OgkgqCmlo_h<4b8k3chCe<#IFF5pQWqFf21 zU>`9C$LsH1{KGLY^#;CueKsuuYpp4Eh^Soywyr;lAYy-zX+S+8s=T4pp7tGE| z$T5(8GuE{JC$!c#R(I)eQGBLVMp9Lr5OL}_C!cC*HV2}>&e4%1i0|B>!2GMmul^wT z7RvVXO_t6*$t%Yy{e+Or8x5bEn%($A;v4?}5Jjm;eh?dGSkxgWu1_SANv|x>bWIaf z@!r2KuEo5&`2=z9+7Hf7O=XCxsMB+oLrPF}CCZMvFNog{YqU&{W~;D)3WCD%0eZhS+$gOqW)*((dj2o14JK*fqFN`06qV z=CQm{@pn$}uZd*v@zO6RxJ6Q}2HmG{$N&@XUSHsCYf{v{CTNSP-5akaMp%Q!7jES_ zJbfx(h*uHYe07l_zlC_#iEjnU z_-tG%vMe_Y19Z=tde%PK#uAje8%mU0ijPy)N5}0U5nb&sah`wCz~lb-B@ZQSSce(4c>M~p13t6u%gv_~B^4t7shlqS}Uz>Sx@_4pm2g(~5V*T{7Xm99hSnP0*KXHa4!) zS35 z-`>1Z(g>xL+lcb2fG|9DiD<)nkTOm)$MWl4oV`pgb+Pzw zMc3L}`E>gUF65K<{$wODVbn0sTHMieNOV|ax3+@T3wCYWAdIg*^;=QXrna2wI&cee zqvld@vXAb!^gXLfP1K>(t-RQ>7#*)0aLC=k9< z1ody5y=iG)3DNYOMQ^S2+X$yY^2roLFgV9t)m=})te2DPnmpz$bRE&k$K{oO=ahrG zpI)Pi*6{Yey4AC!(6y$cZg$;TKi-MGy;Sxitv6C_-LaM>w7R%u7Y5Q)Q;;4U;QIBb zZZ7AvxPlojt&yd0sJ>Yw`*o}LS8>{+mL;8z6$d4aP_)qp+uKmQe>UKskamoE3dT-a zD3ui-v}nx>G(^H96Lx1@ifyzr0DjV8W7}flv+i!SIPPRrh?v8Cr)GGlrh~*+NU~b$ z7S_2Ra!X?W0Io$@Cn>IcOrJ*A?{uWm?=ITafMrU}i_2ifBYVEK~G*TGVn`qBN(>ddsr^B|I$%G@!2|(-0;E~w-Rhy~5vmzH-h1IN9>V9@A zMtw2Oao!y97Mb90hx!~^n@o{NS~YY{!AKj12Dlcpi)IT;V#UXf>@D#9SJ|qr;~Vvysf+O#_pM~Kk+|`WfG>5 zrOt1)S)a-=TMz_Xa!ILgmt;}jB+f-lYo3yLO>0sy$V)$o9@_tJE}|F7n^~I?momqCRG1LGqE_jR58J>pu#|sX#3y z(M7~^4T})NZXTaETIfaf+?Re@G-!^@v6&G3*y=J5UJY{Bm&Bwh&n@d85Sj8yJjpT%JkP!DC{l*C}anI32nLe-YlgmTd_Ywwgo#zR z6&c6Yy%+(C0LPtYf#X;EGf8nKz``E;N#~zq`PKVtVu`w836$_z6B(#)u-CeH{{W`U zfAPkvti&lQ3m-v93i3yH<~NrtB0PMmP~?8KEGr<1M9i%wIVvzJ8K=2aCMz~}>Kk7)S=wW+=lJ_DclMdJ`vNd?`FE6PSWL)MlHHaP|Q!c?^blL4_Rq5JkVKK zz|5OVTfwo2^e57;T%Al?NNv{G&$R3eTN&+|=B=jqf_2yy?$bmXQU3s3(MaY%J+t1c znbmJ0L0=|zRm zuc&x)PrQ^*{hwmPPu z=Lg)@R2H^gF}Q1s<0Ozp#RcH>X@C#ba23!>jPP#~gL4t7U#Z@W#kn4V zVR*WRo|h`63yURENfzWUnTI1Mk7~uab6EuHMhP~0UA_LF;)Ht}y*(z0R7lA>2SP`&s853Q zk2#3Y^g$P4_7%-@J{r_+BeB1_v4C7&$7)(QbAap!#d06;vwUo~Uo+YU2L{~Q{+)AI zO}4cgw?o-9?+$5Nd#%JeUY`ZfQl?vbbwiJ+8Kt<;tsRk}wzdwZtD!&EouS=b-e^|0 z`i-%-yOJ&P!a3M|2l?+_BjPU_>oV*2*7}X*<%~+~A&xAd3!Vrgsr+imZpXYqr`ubd z(ZL$XMhOD5?(7}GNmk9%uE&QX@g&!qd>qz%Q>$us{{Y%k-CXH0rXiZ(bGtkManBXZ zd}Q$kqYsL%tZns6l!X>BVvWiGN`sC%`sd%;i;W)bZDi;pP`2u+_myGW`Ey=1s(eWB zZmFxyYCIV&-SL#f%+RZr+DPC7n%np#t?C-Jz3N|TQ{8z;g;G!0|=8)^Dxk9Z_ywTed)CLa*#TT3QsX#wBmJyGh%XybRT;OmpPe%4N(KD){W?&R`HErK!X^aNl02Q{ehNv>*2(i~}aT* zynUhCJSZ<D>T-V% z{4b{XUrV{S)8@FH<(GWZTt-ot;{apO3gLAxf=eE!eXIWf!bvnDd7f5OwcU-xoN#N{ z?OS|iLP0Kb!OjP7(xcT`lSjRDICX)_k~4#}(vnhEV~m_#(c;bUHFdw3_@d#~HpU*) z9Q$O~vE1pl@1QoFcD{AL+5DLo9J+G{3yQh zCyZv+W7Oe-Igu4np?4!_$2&o;Ag%alP z%+3n$;ij`SidxI#2}}$)QDq?*}W^fTWr@?JQUpCqOLTwsDta@xPd zIW+58Wb-5}6qxr&7jf(jb$<(dQzwe_8SiCD5!OJZUAJK4oO@KNFs7{)V0C3Z%_;2k z7Pu3bu27yzq%VvQY*oJzX!GmZOb}U3Z3$Nn8sgu~7{MG4FnOw)zO6m}teP&K_$2=R zMe3utt({Wq`zuNFRleg#AI;BJRsOY&DaFmGBIg+-xsNXY0EKp~ET{WbR$ZhqTPZQ= z$2HRUJHXAOYe!ichn|u`P)1@XdB>^t^sOKGNu)3r-n=Uw7tl(7_CZv9U9H|Hg*ACC z?j9LoA1lUovf@-7euUQi-3;t@H_;Uy9s!#^ZtzmTeNX)#BsjY%ewGRuciz;TW@ zXVdfTUpl{xd_fvBwz+;3b^ZCR4-NR!#5$I*4X@j7t|4WU9&CiHhhf0ysG;btG_LG@ z4>qA|Z3mky$f=Mr4tmyo#-FC#Yj&ESp0VD-r9qBBrd4m@VcchtUPooBT51|RziNwh zn_yg8x#lMR@}Af=nQNyDZxv|J$vfOxJm~g!OmOH!494EQYnFy5o=!4WF0D5WwLRg@ zt)o5MuF^3$2k#a30CuZ3*Rb6~=0-?d*Cd_c7^njzkO}Ep)~L5u z7WYAhUNvHU$Qk`>vQbe+N!;GK)$A=JXPV|P7pPXr{DnttuIZOSysNnWTx7A^{OjaD z5nfxv;_nc}JW@*}mr^859Dy0$dBtGM9Loy7n;|XRCZgq7`qRYP4X&gvJn3z`i@+7h zLn9G^ji7KkXElvy;cX_~IIS(E3V`iV=_v!>Wv>W=a)sbZozHmhXkk>0eldrbnxg}sfkM$xVT zXvn}Ggx4M7D~Vw6uAKJn%Ob@iDfa|oxoZy-%OY&Qje~Y*&U5%xrFC}1ud+SD^6t(H zW^0*aSjwIW>OF;Vmdm7B>wnsoT7+u>8(Q8-&&qn`eKY+l$v!p8<4f?nEQjWXQ;>T6 z!nm0W=S zYFmq+GTlmxaS13~jBn4hwWmmKXl z&N%Der8O>Bpjgzph)-!OOw6k2Rf?Pt28<`55-6Jsg#ZNzg(4F{GjPfwi#GDONz znbfX0_pUYkKci{#fZCJqBa*9}@IcNz25V|DdDnLAPMtYBNUQcRM z%R;cR&-Y_I_4ci4(o@}u&BtTZ^j!+V8@R7+Zq%HUw`S7adv(Qhch3shN+(xnnM$x# zEJ!?duO#r*z#7~eQPT@14l&NklT+~)u@oolB**|KdAI{O>N%w7s5>SlHr3hQcyqz} z7l+tf>6VQgj+oyg64*TBo|MlOXgWT=_VH?qbLK<|iJ%Sgmg6}9kIuNK@g4LQ@X0#d zBkw>^O}ja4eqHKs8hl0Z4}|s6V)hm-1iv&BZex+|Gr^?esBJ!~8cRp>;%Hh$<^~k^@1Ep`X1VYC+1D^HETKLk&=IL#`o5XVEJjZ~l9@)w5Tl#(U z8ilc%+EFZfitZkuZ9>qo;)#gcd1DwuaS>6t)yllOJERs_oSK~+tBT9Z=tm)CExR$ z4X9iBRQrJ$#}(uM0PwK%3l9=my}q@7cWE>%vnXQA8)RURdE*AVnEXYj>N*|fp{8E_ zqsWPmIpE`rk4k*VEn{+hQr_M}9D-fL`>`_*%nEhA^cR7|Tfttt<15$dE6gI*{7EIO za;C8pG=%lE`?%~Esrs6mPVp|k4ckSlVr3%%poFf{P65qAs|h|;xZGM-J3kQFE!Ks1 zWb(9sYKfr^v6tFNJm6K0F5Q2!;DsbD|}6CNB4^Yy0F z=q(RhzqztqLRjROi83+-Y=4zRWV2LPX&plx48Y+304njVR{s896Y5uM9mJo#iQ@>l zAOr{L)~E5mjqmSlG~C56p9+Z>n2LaL>)hmYq^rwdbk^s+>FkEWBN@t(kVwT>g>>t= zT@uwdhY~jHeYs^kHb~^=zJIXRd`OR@>Nc;aIvZn~c*|^OF5{7(r)t}W#2!6e5@>F1 zu2S{EG2FT^RP-2C`nt}+MXPEAL|x{N#1RMp^zcF{6qsXm=(zPbVH9y8Ep*7W!iE0u&o z#h9=sBOZddtA7Sewz5LHfajgY2?ht{Pw?)ws~->FLvea0zJ_%XNjsDBk>7l27A`!tig4qVXIT7m`EvVt}8WN_?Za2B_V9Pl#$p+fC5# zZ{Y_9}kN`aY z04m~S@mGq(21xZSMS$9O8QWyHYOt`9R$~%C>_{!2ewnTs%f)^s6Isc5s^7+v%z)w-hDMe`3QJONGnuEMp+{AJ+{TjC5h)^O@p_V(7N%qsA|&c}@8fzW2V zEnmad*P0Ag*Hg}d2-FvVWaI!dj%&^Ayk9Nc>mT-vs?8F}p{6-G?T>19r>Wjs>er4Y zoS>11-eNw2rK{-1wAk#tGvO->3+XQHbsOm&M$#pWApZc+N2PXeDnSc7lZ6>L89$YH zRQ@8kl%h4;Bc5=w6=BAC8LswCRjm9GYdT6=;z(3POMqFAPEAgc)el)cPM_@-8n^m8 z#O@Cxd`zQ`;E};KLx{zExlA zQTVYY)u7XLCzY)2O}6$d$K_&r9Dh1WcN>k4`qx_1qKP7nrXZYW$!rpPlZw^*R6q04 ze>(Y}Q`K%7^`f%Bd0-_IDImiR`0wdjzwnG}AM?&%`~6kSD<4Bjq2gq2j4*4%rT8tW zH;LBa?;w*-5tI%I60cr^1ZSG|+t!gSBxr~Z2Lz1PMAs_O7QDIGsHBpTY~#N}S|xo+ zy_tvLFAmS3{j<+WRLO)@c>|5BoE(nlH3y9Je;HlsqU%Dsj9V;We4XGDd;Q^$*0~KM z;%>ip;=5ZfI?UL~bW_Xnzj!0Tz%8D*1Dflm@!*{ob)Z~x^4yg_%A)FJ>m5VDQuxGb zcd&Sp-bJ@q)WYl+8?(zwK(5YN0`CG43P1O7`r|Q=t z;^WN{V{TFLn&9oeE@>Z^>Nz_} znOaVZ;=LzR5yfCgYz75*{Oe=Kx=y=ycVYG&I$8Ex1(3(cR?i};={_jB)pYs(&EiNd zu2H^eAc*Wi*cC6|o@=pR5!vb99<`5Fy%OC?7Jqk?XXNrZIO*$2E?r1%Jr4`EW12n`;{zBxiq}3h(xPyy z75u2fIgaCYKQo%Wb@2kq-^wxDJ1eYli-n=g;vz6DwSQW!xDM{IQFZwZ9BvwsdcZAKO-% zl&NAMYdF!!GVjjgjC2^Oj8axbPNST;rMY}~K?S=<_J-sz&OJnf{{U)!l~2Xq9J!mq z+N@SFz0JhnY>99Q%&N!q2ALm-E$)wxn z&z^4n9C)hwNb*10JsVqs|n6dct4jkTxXD{f-Ww1+)S>s z$WNLPa2wwht)%!h>@RIjWNT!dCi2^ztfx5@*?7amacFm%Ob;B4F77rljmRpz0&qD! zYoV~7*5VKJX@sh$oUtT*bI7RnaqLr*vF7@W?Q^Ey+!*A`#OoRaJT5<(tH0V4TPnmN zn{0#?`3^&OAJ)6QM&0GKb$|gN<$hv=@gkzqt?pyBKWg&BGj3H^XWBhC9<{GGb31lE zRPi;cUHFSvo=rPYisl=c1XEi=O3HU}yFCs&kIJ!cEv_cD0rcD7I1ouOoRiLYGov(jrM8xu{(ZR^`UWW@L}eR9fN)t4ZMDpOkP>q$m?!pxqdeR zCh)(BJX7ThoQ9!l_ue1URbeGcI4Z;3S3fUp;?B+sR<(Fl zh?^mU5<%o2YQwYmwS5)Jd6CB;{{W_PyW{XDn&^zQOG0e*k}f&ScWG->AA@9e?1d45x4Kb3tEt9(SiZ4Scr$?Zs*c_WpYKf0=*p8Otb zHg6L6nY6~#ti*Q~a%^~7FwEOY8Qe45CbN=Pvm9QRMIV5gJdHvz*sE;XQWeUhsSBcW8dsZd?_Gj6v~P5_qpsTVJ%B z71`7&y>18BIIQH-)s+~xvDwYx^!qxb676`3v^$hxagGi;@BqzFv+#UrBTaKOf&7yK zBu?8^PI)=%c{Qn_xW3Y@@0d#)Tt$+Gb08V?Kh~nWQEPi;8hO%3E0H4`k~RQ3f&>4Hfuy}Zp7aRd9uWWgsrI3ILzUQ2JO_|7Y6podJ30Duh4{PFbq z*T2LfhSp|+W|M;1X$$j;@|%wyUH<@U7-yJ7RzR(=5Dy-ir*~}y)X4Caw}>BKiEdG2=&%|xg#cyH0Cm3RiInP{Yk;Qtyg|+Lu{ao6-izFLaw-w}^ z4|>|x8apV|q?ZqIJfb+zO4tm0kU9ZNSE&4t9@9KG<6|YN-@$TXS3qMCD8hyvPvKpg z{3(X_#`k*muc$rj60h#!afe>O6Vn4604uc78E&m(m7ZZdMWdBj$s2*e;QpS~U&WgB zhJ&UV?_O=k{ZvqAd$*v+YDrsG#}wPLMXe7*w$N?mx3i5ShDl7jQE|!YYHdOZZrV5m zYS0i&gpt%^10SD0)zA17#ui>Fwzx=Acy6uA`LND;0Q4i1#|FA9LBREwgD=B>BsV|tz!QGSJJKCYprhDAgVT_M&O@*8#USZPgZSf!&aJu$rOfV zB)mfka7gb_5i6!{O$Ut7skzj?U`a7s~9;vdNH1 z^gflfVX0crtICMamS+e|q^l5l!0BAi#J?G9J{G&upwc0RM!WMji6bt4Tw?$ldM!nE z&RBdN)mvJUXSlRBlStA?t&({;z~isKwQl&ELc8$igKnU-^DiFm;X#ThzF5Fh&<;j% z$@Q*I>*6g!8I6?2jS~>So|)-ToM$gF z*`#?i5cryHzE+1TzCatHbp!LRt3vROv#Z(a8g{U+3^#CqLW!%Zu|(kPL+K2jd^u;EF_H?hy(86Du6ouYl6`JCwRYBH&OXEu>GFhr^|V* zyKis3Pw>~pe-!E($Jyqyw?t@VZQnV;Sj!FzdVy0JMaji(ig(k^E8? zAE~EhxRjaTUMcX6wyUjONpS_ib(Mm$WDql1OW@-IBEoJ`H3Q{XhdA{(ucU5mVU_&J zC6O>NcF0s#Od6JzH;GP_Z#?teM;DzSM=cy53t$27`By(_ z@oPjDvP>h8eE2IYxjhNvKY;v2VCg3JElHfVt!3g9rs;=D(5@a8i1Tirb+7VitmHK`(uysqEul!b95(|((!JA;`@y=OI*;gmAEUcT&>)F3`(iapnnweax`|4snW#D?6QQOK9}lQ*ShE z7&}J$+#LL;Ju&pJV!pkK#s->Q#448By+|Evgt_sqogS$SYV!*>C8UtbuFAxBTn#z+$2K06ON|2eQ)ACdVa1g9>Qml)E&!mP#pRSY8O`yYa6ln2T#)UjaJ&h zray0uQG~6#Xi>-WHNjo{KGk%1*xCz;;*0@nvCA^IsTjxfuS~S@1oPRuc_K2l#zhQr z!wZlPt#X$>Fuahkp6WPc^4P~1VuYUE{cAd&Rehr+Vupjc)yqOfINm>J;v=DI%*{6*KM@kOJP=VT4_oQ zk%SBnE8P3nVXJsI#244n%BcQU_c5y}<30F3mD#t901H9%zJA2^_Va01Y9wHyM{U4o z)YB)|uC6AM4dF~~Ps$3Df_wKBTWC(_pICe`yS17*r?t72(B(q>pdZgC^Qs#C$HWLV zO-9DYN|sGMrjV%?`SStwKgf#pI5m5yLE8JX*Lf}c&1P$#6tqn%Ri8$^TbSpI8<}#s z$OnQbY<4`ROFtf6*hgzFlJKgYWJ`8w7@n^%9t(d;iASwqkmdEE;yhGya=q7=!=1W^C?G|jy31$Ve*Pb}3m87gS zc$riDBhl_7j#KCTw+ZG*?9Prr=jH3q*1Hzd(%Qn(%_C-t7=kFrVa0G3o-@{+8sgUa z33ieIkx2*W39gF!#QM;-(-tESn{{j+Os+m+dmhy3#cqkID=kY>czVW0%h)FeKkE7# zpX@9D06bsvHHkK%;yZQonnXK$d4&F%sK4Pd*Z%;!7ybA@<60=Ev6HLIx!QQzd2M_h zsm}|Pks>UtI)h#ts`#f@Ti+?J;_{>%D}d}Xj=qB@y?e)t2A+H`sDEnDBo-b~GfN-d zRrMSn1{lg$*9qv&?*0RS}7md{>B2TJU;kN84n@g|)C)wLh(FEn{DI-Ra@(2*GT z?_CY-ULCQznh)&n^-YZVt<6*)DkfU9&a|7eb*u!N!_h*U)m6f6+QGPC?dwg{uP)PD z)Gck*Lz|ZYvC60ET5Y2{$%4-x zRy>|@RH{@`Ql`n2BGXexgZPVGkz`n)mOmmf%z!V<-TC_dHL>HZN-G;{^p8-untO}A zvRlUF1zn(=;=M-8N$@77tWJrhmWYlPEQ}s~1zOdapBN(`itq z=$I(s-1GU@Yir>z0NmaG0A|Of>Ex0as@yTj_VueiBJdW0bL1%0?`#s<1-FXU-svJh zpp*PcPJUxt)>4lvX0;ofuVc!5Cv~Fub5*?3-emJ;Dgp=3%rn6_#&Bz+@x{%QP-vGw zZ;lzHwkVLur-8Wq+3D?9qWCr8Lo{;zyHMU+Eg-&(k?;Plx;;zb--fPqyEMDEYukGm ze5AOZ?_t}!KQT}Tezj>QK70QFk)JC^jKSfpF5so%)+Z_s45|j!Ami8Cx~r`n^zR2< z-`h@Z(mDL=e4HJnKqHS%wbMcH_rUI+Og8a?-6Ks^uutb(7k}`Jcpk#e$5geMVYp@% z(R}wLkOuC6k&?8NCXJ4htt_P_`E*PwVe`#=5>9e(aBYo#Ov zmNf*vRObOtPPKkpj|fVJ{!a=qROrUl!2{GTD_21Ahk~v&{ZX`V(8Xw~_h4Hx5tD(` za&yH>W3j|r>$-Z~#}(z4p%hXT+RA~pan5n?#}(;XOcCi?E&b-2uW=!S$cW#@KqLys zZyac0P7-6^q5lBlYj?vwCek%{ZeG$>R)=q!ybu>XIUrTGp1{-UcwhF9i#$(jsNKyo zB=IpnXJAQf$Alfg>?^+bWfr4iZ!UF6u5IPFiNo7vysdyKQZvCldGxOG>2#LVqrOW^ z39e9lv$KH8e=63B{?g%<1o1SCKI*VF30S8#^ki4$S7MgCJVoRH)k<7BGV2Fb!0CguI zRDKuL@2)4f({&RQ&jc|{lSHLMql1it(1NC?@m8B@f2YRxEb$mya$H9@l<^WV@|+&I z#xc%n9|_uN_HkXuYb27~2T?oEDS1u@${QeKj)N4#q2gZ=Q&6$^fvR1LUDmQTNf_*) zXEjC*FkC?rK^?=~Mpt>2a57GDk6~W9;yb?u-`+}>y4{;htjR1dc8C=8AdDYMr)zcK zNTposb}_Ee^9$J6f;)7k2b=svnt1$Os3n~9^5`as7>HgQ;arn;P%t21_4KL`_;0|5gP)jRlloSC@$Nf5 z$3LccS5MURn>}+^GE5TNZg!B#ho71C+;#&yXCP)m&e0CyBwG#9pqns}PY zygT6g2_i(fh}~R2$lMUos{R$3;jf7+r|5R}G3ie$+efsymuSH6fGgEC%^$$CXg1bO zsY7XI(ln6-x>&;!21wiwwQ`m|2l#gk&mqawUz(V(771Ciy*(c|cGUqSdHEehLBnhg&2D~QCKq*R$?*hs(` z0D4tRp8;tS>rJSCV}_8)yq~lXta{|FM4;r6aF;F3C$f@n1e+^q8W$FxTx}V_7+x{? zSAk2YUha*=^1P}FIF>_&?O#mD86YAk)tC>JGl5(!{kMeduiRhh+H6xh62Yd=KBNJ~ zO=vk<%_cA^eTe;G;T|uRnU|%a{@0FMgmdD~N%dW2y@EMm39DnoZwfV$>6^P3!yERsk^_+s2~~oZ8fAV2&CD~h^g&r zxus(N0D#qE@f?u`Ewx(&j@T8;d``cTZygAtNun^sszWgW@N>9+2E8N0+FpZWc^%EI z{_Ce*n?BI5pVF)NkKyK@t?Lb{IlFs@05YWNiZDIY{{T7`t3wuX{{RZ-xo;LshDll` zw_?g%;~fZI(cw@nOet~js;g01lFjM!FhXfw{0arXj z;ma*;p=+B2xO4;u^GX#TQADELY;<92s#R3vu8%7CS9&A(i+yjq9(|Dcany#+YS+cT z9BTG@vS>Ocp0Qd3Jtd8fnQhq9MVYs3&Rn|A0erX;XG-hB7; z6&H{0e#`L!*A0ev5Da}-kro_|yInS-eE&T-N5d)L@MP zg$W*|e}MYbSBsj^!WN0={yv&npv-ROOkr?IIj>*%Wo*!VIMSkK1;j@`T${Lg{#6vAILY%DDW@J(Rnx4B z-f7E>G;YQJ002dEKNRI?yiehlRx01W!w%dE?@~LcwY z^KDQ7u1qXqR!=S74 z>Yfk#P_lc=lI$|Eo%vzUIP(6rHjm<8huS>X7Z&=o&v682&x6Z$+A;zD0iJRB*IS$# zX&G7dF(X?a3EasI&9bXc5gXX!dUqe4TGBi+k-5}Gu!UL)94}p=m9Q$-yW<~+ejyP_ zb*IUGVOAlewUu@+1orn8(CL0T@P@m89)n{&%Fl4E5=L#E00B-wBw%o-@~${hP^jv4 zWw)h{YSz**Y{{?|EDmaGOH}^dvwU(SPv>1eqo(N^O|-I!Wn_hX$#XCTM`MkrJu2mn zg{8{&_bV)-UL*^=5$Zcu;d9S^GF)G2TJ?>^rRCH~J-N!oGa~L&*8ugw6{GM{{`XUZ zRJp&rxtW8iM6*cF0LD%;R{TY(_(R3fh-Wvqa$2)|(MHN+2d^P(-n@I=fhEO744Wta0~wTKDZoI>3fCx9KVl|wGRS# zLT8+;MkkMZ4loZH{438NK(~f6=TWt6^w}U|`D2>;g5FOD>b@7XirPp%#uc8~C6LD5 zPI~i<=ChZ>{ui{CDHyx(j$BM_+~kZNbBgmx zWp6JT!;X#IA78F(*Wbk69MY^{h+al5+!c??4B+Fa$j_~D9&dt;=#-g!z5B3aeWRaE zzt*#yTeX76pz4TQNsr9QKWAXX{{XvQHK|y?IOL4;iuR2+Skk;n1j}VHl_U9I7q^mTAKCkpJU`kok@;4Z1J67c8%1w^7-RQq*ZuKW9x~FkS+)NFwOI$5S~Nuq zfw8$bIqU0Q*%yX%eLe{`#K^MYSeGRA2>avdGd_FFu%?#7ngv0iPt5<;wk zHknx&N|iYQMtCNzTzDVDgJ`$#<%FZY@0S%3jF2u<%QZiVO{m;j*=f;!>7|d)mlzB; z4i`1wTwGkpL~haTiJKc&mKo?xZCZFI!-^J1&@ADVTOcUijC47vE;LOy=X9CcLP^?X z3(kI(af|92bU9bm;DUc9aM3Q%M6LHp^v!eM7A)S!U$fMN2_!dG>ZIozU4FcDBE1?t z2TPhC^rtAq{lO?<{HtqI`wTi$8_{mo>Yxv_5wQAXo+-6-E9!jhqEG#ye{EwGiiu~L zQzev@$>*kP(mo^ICZBSF6;3VHIUNQ~c77g@N0-c-O|{dLZWW%?t8?f`6(5LvA*XnP zMw;u-GI@If#^FI7dy16{Aa7j{CDL^jywlB^2#|xEh3THAyF2L8ICO2Ngtt&L)~@M) z4fLx$%j8_d85oxTO7Unuh~dvDh@dn zmx%uWzbEEvvK|??dCYGVqAqbKnh(>AQ~v~%*|oYQP=qw$`DI-oL(^&8Jz5O}WZ zTZ$=l3z3XJnt85wN$|gjWbr1Ga@t!ZtQTu6lKJtFpabYhJ!`Af^qa{wE0zIQa~4d< zqd%=@hm4dx;;*Tn;mdVT6WNiBjl+-kO*_ZdH~NQ%CrvF?uA-Dms(XQ){y3}pCYyXb zLk-Kqf+Ws%$PV9>j&N&JQn$T=-brG$6EQp~aHI1zv~HcyPR#Rvhp}pL=@VY+aru90 zaL}Raunr0I&oy_&7rKPrC6`NkrV$iIg&D|D3Xb1g*GHo1+KtYO1XJ0hQo@RjwpDOW zJ*r<7_(EMmdreZ%fpr9H9Ei<;%02zEYAdNtI|VyL^SSJj(S~R9CkHFHj33UO@fK^F zO@7kSXJ<%k`N{V@SAU`SG6OKWzmX@95PX=QAbn5tt6nAWEIuRDKG`&d7$8+4 zut4|Xu!R|@KIXpGDrvLEVY_i`Utu3Ecxv=bHRYe+UBgbrouiQDj(?AeuP4H}B`qD6 zv2Of)%EOR93htrsyh-5a@b{UrHi*U@duPArTTz6f^$PQZlI1+}M1@~Mwm)WrV!@Vn zi{@jFFny}e#p#;f^TF~X6mK?Rk;Xnnt!+EPTD85h>ECVB$C5VS7H4J8TycU$XMAPw zuD{|tJqTzvjXbt;$j^sWz#%|9pJQDT)mZ075!dw>+ZemxfHD*2{7q8uyn^!Y#|tCp z7PjhG}jK+}p;e01Au%1JbnMbYuu5 zS71582D;8O#QbIP7hRK6lE&R8(xZX?=s1z_+@DXSb-xe3A-=yN>3WoB?V)m$BK)!e z>*#A^#$FB6JWFb=aVpx~C;GdNKO^cY?}dCZrg(2o31S4YIQ`|p<)r+5J*f`omHb8V zKC}M-2{q1`mvTjGZz{Sf@yH&dn!?ic-BmTEyBeLqmR3 z`u5r@b+nLzvdJ(|I2}h?ev|Nq{7DVP-mPjCST-P1`N8!#swJK|bEL(s{7%)aZYA>K z`&`cW;G8kwR*#Ro6Qt-`HLj(l+(-+!j#wpN#1aX}KZSMr&%plx5=-Js?Oxkh(=F0# zhn6z(IA94U2j!~wUj=+$tprK@MWs$78C4QK2fy8*W|ox|uauW_D#-BJ+3bzG{^<$* zE2q?N^i6L{l_QQPxGG8&OfnJNp4FA8d@b=B`Epw7dVE1Gp@sa(90E}lN&CK{on`R< z0L0?etv|H%fMghB6z;^2PrF`wr!{>JwJ6zYV`>-oS}vQV-`m>Q&Rz|xxys9p&6ED^ zaT@1{?6o`jX4EY10bQ%Q*5Lf4d*JcUwR+~W@U8B&nYz-JV?H3+G^T5Lgye&Iby{3KP7e3i%?fNrv2t|9yQi2^^XncQd>bJF+kG5t0Bqx zLU*hUJ6E^UEr;3d43}~P=1n2UQRt(w_pf5U_$jPj=$2OVYFcuecXlrFHXCqJ)Gw$t zk^cY+6H-xut!V+zL+3wF`{uN0I5Vh2(c=?#%O!OmU1S`TVQb{3-D3`yayEg}J-u{6pCLX1R?k!ME^hcDJjk+L*1}G?B%UN$=Xb=_ZN=23J%Efw0GY8@lq`ZQ@Xae(XTc_2F7$Nzn;AI z^!2Yoj!5QR@v^%5gPMZ&8yR9~?@|Yi#tO;?2>ffHlh5M#g|8RW-Ml#O1?|SjokIh( zjE60T1f&g>ui=DSOGwq;oD-Xf8J0-+dD&~&aR#=Z-@ z@g|^?OX3f+*{cn*#0`GyWC$p4s#X#-nxQJl9OneWFjim{Gwb_s{28 zzYx3>HJ=@(qL^*a1|%oSHu|5XH2f;mq!!RWh_vY*EzTFnP+CmoLBPu|u4~aG)Ab+& zTzM`8NF|Y5YL*LvKA#>aw+*6&{)`tP7>)#Ny*dn{KvW^>RPst>J07tGcPAcupn$6amr=4EnAGQ-A zV6fw8+!q6$eRKL(cVaZ{I?f9#Y2#R;0TE1@Kc9NX@lAx6lUfMjLKHsdeV{Ht$0YUo zdQ?U>DYelYADkg3-J!UVRtJ%!ZKN|UM=g)MNX{`?zA)3;_row*NjQ-b`WGxl8-)S{hi}K^(0d@w>>zLL)A9$AR>XTa9#WFlQBH3`InYStD)9#weoLZ5j+GyuA`*@zs z%yu@<`-a__+ozV9HLc+khA)abVMro~iKWR&OMeXgjbz$*6UG+M$rISJvqZ=T+GFZM zk6-ppZuk?x{wcHgfn}-R*fQC~@|h$vD`P6)bUkvywy}}B7N)npDe)9DGclUcypZ|Z zv>$G3OTs=Qxw`T7#*;e(6cCk-QL?!J9!IV-&0V+f^qO7$%&iKaG$~awE(4xWD?d>1=Z;~pyVD@knkc1}o+(o011K$? z0X+}3XV~~h$MfImk;|u-irQp(U9J$b0NKgGs!~p!tWBNE-Zh89iQ|j*)FSh)CY$Ei z!LV`AjPPp{%kYie*^TXdyOMYSmd`i`BacJJt#*DT@E)gg;{8ifO)Xj(=Pr`8sxy@X zBPWjaj~Bz8UMV8`9+(ySuuQG?=fH zQ45IFBK*z%9`!}+4wTj}b*DY-5~xH-)wh4V*y)Vdrg)pede)(K@<$S^I9wKW0Q{hT zO3k-xgjLR@T>gPKj(?{D-)>$$SDp+v)3GlJf?Gr>L-lS{L_ zgI#12OESuYObcfRp1$>~s`w|v`mB&iJ;(1LZP77sN}Bemrl6F!Io{?&_=5Xd@Sd+J zxKw1e-x~!DvAOj6SD&@Kaz$@0>>gj0 zZz+(YbPbXRYNX!+^tYCBd`iWflOmYcAI^cZC*fy}^%m5mv!3>8tmIi_5zIp}l}z9t z!m(`id)vFn);X`@nrD&x+k2M;?)6-O$7=NN2mC10JUOh3`1N15h~hDJ3;5T>lK8scOa9%vnGM<@^BhPDXOYk0 zUW|45WjAdVy6uUW5@g3hdS<87z7_al&tA6F?5?g%@i>$xo)|Vgd*h{Zx`%*u`L5?g zztX^I3kc-7bs!U-%U2x}A+frDie$A}Qgvw+NFlTB(z{QHx|?af0shiR5_1%g$Rd;O zwn*Uqb;3dL?_DY)`*x5R9zxu)C-YkLPZ{`L^Tgf^8ZEu_R*#6w$m&>pxIW&L`XMb$ zPlTHL_>R+6y?s8~$cdvt9lzZmFIsPp^=)RuRGu3fsbEjBtjNgS!{LTLm8bCE!SZ-N zO1QUuLM!QKo6MF(7+#nF52bHv_QUPg_n&8qeTvcTk-}p*AmsfGOtmQ^%zQ1Wc#8V^ zV{>zNG(KAFGbu&fPdLEmis`&l<0Ij}4B0iT%QfWx05;Y6n8%U}C_Q;1x+uQWHZyx+ zX&LCY_;&vQzMGHIq?^L}EY>jF-daaIa0V$P+_I8MBp%$8(xxa6C~t_`#g?k+rzOSW zt1p;t=iC@@P8CKM>+4&;@VF=c0HJNa@F{=7w=KRM_&&|pz|$ah$x(Xzlkc4VwR`>* z&xHQ~=p29WB2WG*s1BO`X1CMsCO`{E8#vEQfK7bSbMf=UzH0-f=?(7CvCsZJUrb)X z_M0i~_igiIWZmhIMr-C@7yKf%@XTBP0JZM#ZdsQQLt;+_uukAe;AGGTWAMA;wz=Z( z7ABK>YhewylJb7yBBvuB_30Yti8S3c#^$t=LGqt9h{5#82DpEQp9-Mx=B*cqEw!00 zRib#ra$6YBKT}Y6yTM*4zmY|Wk~WG$!Qybr;QEY;Xb(ovwf!?tU9dg0QWJoYp1+{Y zR`K?!HjUv~uV7>nBd6a`!>^&P7r~wm@iv`3y5B6#vSA}nGmZvx@{E&R*N<&sz3{ZK zTu0_a?zvyahJerbXRlp&gGRf!x@2OP8y8{);{fy(&wNR{)oe9iv_mbyv380R5S*ej zws1H(70~=Mv9{CnYp88v@{rBAj!sU|T=&HdQYO1yEZa8qp59?GY$(H0y_g)XpQ~!I z>nU&M&jB%)VwY~=hi>)h-Z1f2qoQeYCX-}t@1uq}7!@o6u_Gi7PfGH*8YAL@VJhH2 zN!r_oCp~M?{9`PZcea-{=_zKjDJ*g4H$ymB#y^M_Esp&uP3UP#-0J55Jb&Z3kln4d zqM=a1Rsojv7|wrM-Pe47;!821#_RSQo2z3j%w{q2;fT*UBY~R87ak#++AA;kM)gq` zZog#4I&-@j#V^K_aSpSmv{tfA=FBPDg|YKw)^Lk=)akC{rIEM)0136kdI$Enznkwe zmJr5EZN_*D)1_^ASI56-@b0Us>QkGDtX)Rv&?z8{^!#hbA@Oa%(i2a+(Qgg3TfDfU zxAN6N?pqlo^gXN5J{D+@YknHlVV>S=Sf;sIHjoZR-T?$-iq$O+Hnlh6_`Bg5WDhX8 zhC|3_x5@kn&2@S|igbN;#%s$9!0G_`?Q<@rR-4&bf+)oN)o&2eac z(p98WgM|Z+peDH|)UA!vD&0Yqlc$=w9Al+){vTHU&oaBKnH-#{$iU)&GyFw*H2LkL zw}uhC=0al}irQq2Y^9bIRAz|blw-FQ%J{7wT}e*z6*ll!t#pqeF&N3~lZwK&wvtslZ9jS4`rv&rR;+a;RBK7Y6eI*9 zF(Up)806QLcG>8@RTJrG<4LTc^HF5^6#T_RWL=vHTKX-s`SJir0F=+m)3z%g z#lAhT@J^L#*EaLdJh5gt;$K1EH7~=j65M#BMYz1UhB%5(u~iA_v>)qT$s9M)qp7*t zEoCGFgeM2uyyxKtr!K3fUrBK(nbJ%KZb;f*6`;$_6Px=ouyE5=O6l{n88GXdp zJ%ZbQHV@9XG@tncS^ogn6#hcE_#=>fWS&B@6#D=G?K^YFdZsid#LZh$(KNdT(xw*T zNm(So7%PsP*Oo=&&k#JOCBH^HK?EXTk4~SZdSAuHBTUm&5EZ9ja)E*QdxP4%Ud9bN zJwo1jB$Cv^`^-%j$bCQ@j8<-!Jf5Ydk~>R38F-KE8kv()U~R$x5Ubd=27VV*dcd*4CFd(mV{0lwbhiNfq1cGskNTw{vQC&_xg?NeU_2 z2VCS=koY^qx7WTTj7usxSS8^G-QOwe`BtX2py}G7XPOI?-yxP{jz`4If4$F7^V+t8 z=G2Fw*6G^LuyxCL*5zfhmBB8Ns0=Z|$RA4R?ltW`CS<+3gfbubX|xRcX1H$+!)K#i zAhom*#K$gIc6bNZHF|FuXp9QWWjy=3gK3a|F-1+#;&qySzh!+XhQ{hCHu6>=G1uFQ z@vjni!%)-pxop1CJdC8UFCidgoMYP++}il-N4J(ccr^Q{QJ5&%ZWTc%6~*|MLXTFt zc~FG7SxS~1e2L}VzxO;18Y;D#ER0SRVea&ds1#KI*Wyk;tlf2g(QwfGhFldp6hb8tPw0AgCLWFet-(h z@t=$C^xOR~64BZ{s0LO9jm!b<{&}ghB{8z<`qWyBOtV}Cf;NySayJs##~y%I_o)uU@#$Pmyl<-8eXLJ%>$S3?GKk;TALCor(>=e1^qXjwGZ2bT z-8^7`M6K)4{VPf>^NMz{D_>s#u5Ruz)DZP%^IFuR@az)(zIJc4xFwq;fTNCUnVq$5 zjBM^*BhN7@4nF|j<5KGvW)BA0w2vWXx|0OQw3D@XAat$kNjN*AVFaG$iPm+Mw~ki3 znNmOsuxx|+RsBE3n!c6cy<+|uh0L+ED5)~;P#BPS;=J}h7kO4cXEt(5c>snZ^y!M} zbnmrleh1bCwX}^hHM@CHsb>4f(zz)n;M|p&b&^u|UWCznSn&$(SFqERcEzv%0A99y zKjY63c#Hd1&SZiJW402;*6|qFHt;&<(AS*J;ja|SH}0%%0uFH`PNzSWc0LQywGR*4 z>spSh4a90~x!xND?i>JXT5WS9Hy3N%*)6C@%8xwxlO$!&Jk_Z@|fO+jC9t?j)7V z-O9vgmSTA3o|>J)oTbcwoMfvIYmrYAG+LR4>e>i~{1DS&h&?zX+)`a?*Opgi(e15F z$aa+%Jmhr;jfbb2*)*Q#_v|s?xDDj`sV2t7FiNvSQFTV{@vq4%nqAa22u#ftO+$2v+%y1z6ET+`HE0S zM?FVs^Bdn0UcRqu6`XUzjN4?7%>CXo*QP~OV{!8y<$bQDjg6n3JLXXS`N6Ch{9~uS zt2L;Uvn)#c0(2S1YL)%J+Pp2L-AchBK_GYP2hChggxYleHVtJUMONHm$P+Bi`g)PI;T0w8T@H` zSjAbI(xmD;B7X}{e?NzHJw^d`W|BzMM#ZKHh9S0tliQlO+Wx5=EG0K_IOs*OAJV$7 z4r!1>7NI4c>v-rJ*zHtr~* zL^3NK*vTw==kOKf_b^8kV(t|W89SX1W6=IYRyAsKO6^#vLzP((Y1g*eW!(1hYZ|Qg z0x1UhE(ysBduP|$r|@Tpq`bF=I6l!l-brHjB)N9aBRT6?c2QePsz$`B&-1X}d1Wec z6z2qf6_cauZQ+Jny_@Y-z?l?$(fT;{>5)rR^(h^selFGJ)~z)C;waW^NtVS8+R1r zm{>92v*kW!;wz<|4Px)ZS{39lw5B+LlqmUu^~d8^UY8_vGg7&+q2INvFBYp1l2In% zCdD`(=QXLU>rG)a5hc1y9ILsD9C4g={@_(M7gEPp&JInr=E8Lf6#M8it*;0Stb<7qut${ipYT@!?q0cwND4TapOJ$t6+$ z0EBig`SiW~t2a8`jZ%T$T~4;v&ocs521p%ikJLPSq{Kmv`a#Av#|N2{^{$7+m#L?C ze)956`EaJ@$Dr$8ZEdaScfJ$alR+iJ$G8R{5#OHTx|=dc>$E=}>3Y17TSD`+(wt^k z!XI9lt%RWR~PUlZ!sK<9Qn6)^r5V0Fk2PAgQWa?fk(=0EpEpK&%Llb=Q zwYKbjIIKHePfXJ_2rYEzfK`{YfQ*yRL0pf-IPT87s!0U{i-oku095{gYEW_VA;IxfAc#ROU` z5lZK24Q{GR`w{7ZTfQdo^g14_+Hkjh&~3{u4stqk(zp)@YsXRX2Ado+0oUBs2bE)r zSv*>U6NMxVmJ@!y8D!MhV&p{*}zf;r$-_P`R^?_E{bUQs|Qc$VokT zdbWxNAr$U)~9F^gQ=V_{pT|%FdJBLKmtvuq;0#&3cS7iS476 z56dD31dha4#~u*ce$~4@7pLQ2TW&$&sDsYXhPF^7M%uvb;|JEd>(fwb#!+pyK1&Ih zF_P*c2>vJ(fm7Sav00tdq}wx=KmZ=t6~|e44$D)qitAL>H9bk>P~~3es&mdTxUC%z zN43y&{YKW?O=~oFGLPciFaXPB%f~Yn)y~XK_pv7?{I4cR(#fIrlX~qyMHHz z?O42*+FiO4orMvNTkzuOnxqeXV7Bw3O31P-Aqd9@74z1splVI1%@i@s=lQrWh{%ZM9J;ntL02X+fSTgDH@8ATLA5dgxM8rll4+Q{{GNq+Mvf8MK_+$ERAw zZ~$=2ZVmze04B8Uyg#LVnKxNmzIo4-27e0TwcF|9@qNJF&9e%t{?0N7>smS+UAC`o zX4I}G1o0aTEn{_iSXX7A6?nr!(C;I#)B}n3i35vaDx#B){WFT_v}-7=t$%Q0@}>RY z^5PMLk=Fvei{loJc$(Z=q-th{<^gP=uw9Ry0{WB5=RNDF_(u+<28A%cNfB77kIiP> zMmgaB07~iQT(?F%&zi?m;_Y#KJ)uc{(aR^xC6lVhRJscSatJ*AqNSTb%Sw*%I=zXIw$5b=HG@7OfBl3Q_?j$2eyhCC2d zob)x_c#lh6Z^N37m{>2`t==Hb&=NQ`;a?0ipA&dr#urwaeVv@>JBnP`{_wi_Y9QU46m@Q*0%Mck=1E?MTmBV;r;%1ZLeNO5)E+w>Uc+~lqa4Hwc z=rC(n;!dMAo`d#u4dq;dl7t+UBn)T%1C<4#nQE;@3kV_S(KCCw@NGrc@C9kuz7OY+c`gYh*SprF;%WK8&{Js z!Dkxd?){Y;pI((7mFLf;PSH-nXJfWfH(=w?*CTtT`1W>?jd-XJuq-ypy?trD1#`WJ zOR=|MBvy9mAQ;?{Dxc6(ZY|{}bXO6=o`xWQon-0u7y5RCa5W81=I%oVOHH(GIOiAw zyoh-7U(|J&Y*0ec>{Jj+aHNI7Z2ktKQspfxo|S3i-7xFF*wVv0T!~D6X^jsd#QND->k4=DLnZ?t8D`rAQQp+!}6|bO-v1a z=9@~6DEG3o(Z&gocgP)kb4#D9HJo98boK2Q;tq+bq;cEN4aCghPD-K4JPda~N{_@k z#+jySl31>j8I%CVP>Mmw&wBEWGVVEiX=gMVdXR3TXyj!IR~Y_v-}udk{3Uk-9E4B( z2BNJ;7~K$|t(mW*>l$tS)Hc#-5wV$qhG^6U{HvnZZnVuN*%)i+pFGXfJPT4~HYVNnvJ3!l*;I zf-#R?YP_I}cOTzW{&mHCNu3i^(^ZMblIQ;bA{x!sl3E&Js6urVXR)Y0CTX2dhEtpl zBai$mwa|P^;olJGtvucuNa9O_8pKr`XJ|R+9dnPuya-)G<>0a7IImLpaVkmSE1#Px zw~gQZg{b!REOz2>bYTm}bFX3Hz{eV7N~Gi`*y1}!r&{0BwLJ?@jpVV@*3>(sXUdEM z%yWQ3{VU6^yhAJBq&DI`o4_H-BItva?m}zh+a9#-vRP zL(*+c)b@69h*6p5@>M`w9R8I?e-rBP-!-g07Sv}%$TtIOqY>$olTd1R+J*lBfVD~W z9X?A5mc~Hw$VV--B)@Ao z-R34G3+9{_Y%Y5A$*ilt2VCg0!)vQW6I{PJNl6Ec;AD(CT(m=Kg zCg|0}Zt0)GptRR+0=J!JvSjhd7$22z{wlw{v+;h50wh6J!mT;_L86;0r`6VeQP>z6Ret4B3UJL+}ltCyB*1FbKk9R z88(wMYT6#tW2#y%lNH2TJQ1tyX_ht^Opd3DQx=PNGfkUNcqe_JC3Dbc^sW!Waz%OI zy-G-9JZ%w`!1GBMA-kMZ)*42!E|qf?y`+U@Q`i7_7yz1V(lJe3^;-$;8bcMdJ3vh5 z7|l`D^@~k1_82a8i)&Y$fgDbNvHt+;dz$iipIEliFHx^r+Gxyhtc@T(Sl%+DY3p1i zw~wu@%#Sssw^uTMt)6+ew4YOs2em0Sw2q?XSJe6+#J(zr!*NE}_JZcfz|uUiI+rAR zCw%9va54N<@Qj3*wA;wzEJ#yh9tW@SR}ZH#@lKz2eG1Qrtz@>2+`i%lRQ#mlk8ISR z59YMerPQpighuX1-5hM2m;k3fxZ<;^Qp-S$F6?*O50AVRsq1!HWLi{`O?5CxU9BTk zMn2)IQs(=>zi65jI$R0~jzoKLgN_I!k81J%014Y#+x$TAfrS)G*C0xP$tp$&C+nV` z)!1v+T7}<<)es3GMw4`|s1uMz2pA_B^fhV8tr0CH@jD&Xg{1fn_VP8fwb5>6W0<_C z;|vJ)6&A6ti!Dmx5u(Q&Fv?Wi1y>KxB#zzdkob4tW1;xY%JuKyxPP3Gs8p#aM;^5< zyQu#FXZXiaMUG!KYlk2YyCB_zpGvpx{6t0bY<1S2EV4_utv1?EB-p|#gviJl9=&UO zQ;yV1<#hXBFb3Yj$CQogY7svF+ZyD_qtr^rXF>>Or{Awm8B0 zp8)+4l$QS4xG&&ncEo<5a&e;HWXz>pYbhC>)(<)CF;^T##9LE@V& zX5t9cc^4jL3{G>EJOlp#)~ouQZ6>53fArf)BV0;j zWB&jy=me9*NhEQ?5?I44Xdh&Zc8&}ho>GcaIi5gPji-yj6^gLFc zq2g=JD@C!fv5Gv+6O$tXO67>-6`Wlc1teF$&1QC~<0*dDz`z_H1VP6K>)x$@!dG$s z06^yd0KWeK`qkqwYC5DcBV_!IxDs<)zwn-4f6qIA`}Syz>AlfSGuicv1l2T~iy0Uy zPK608oPHha&L81NhT~R&p|}Nj4-7DVoJ|noewa_l+wAuEBPm_4a&U@D>sA_TA z>lSvuYPFW$bu81BAST@I5*Is9d$&Z!oqYvD`2Hzk@ji^_m>yWbAao(dMr+u7W#brh1+ns2%{Q67qecJ~!5Am; z6y-@uT$UHKcDc$}d?RzLL3DK|+XRUzR)nrtHh%Hz*jGdGCq=RGU6r+iYY^H!w58vp z5yx?!Ds}jaY`R=`(?SGls%P57?9W_!;=DuRRCbr%H@crk$@@UvJAK{x0QdD6tf6c2 zqORVjp!i?FdOwGCi+i{60$eq`p~e7JC%#T=pz%G9lXb2&4;eE$HRwQt52*IFEUU6fZ*wYtM6n;~2-;^YuJlk}lNDd;A8u7wtZr1+xaP=m#G zuRQyV`^o_z&U%IeJn_YLcNe-&w__|&>bBBC!K0AHIRcjdvaqI49PwiJ`u?Dy_q>Dti7^ z=6bY0XVRpZWp#1482W)-K8>;)#&O2fId8;Pt_1THWbkK)732+JDJsMg5xG9x^IMu8 zji>3wq_amtcp>r${LOg(0FIjK`^DE$HlgLO$Q7d}ZV2NzJP}!b7oFnqZjln+O)4;x z=XYQnfOCOd>8pj$dOCHMpQ%f!#}oO}0Inll**>+Vt_lX|2+j`%yrMr4NZK{-qGT#9 zWHEV0Uzx}tarxHFa$GFXmX*%z9Bw$~uiaY~?)7GdyQgS(cVg~sNu!m><`+^QkeQonu0P{c$ zzr?;(W8M5?;f*K7G1{h}Z!-&dk;!OINAs$>H-;L=TuY|b^?BkMDGY;ZbI;62U&gpS zUsAf$uH}1ZN!-kv*b>=L3C0dFQCiP;s!8NErtpl8ePXCG-?=fssnBG`@L&qn~H*Cq{n|8bH>^_6yMm5#Alv*R^&2i z@LICN6i^jknPNb>Kc`Q|u&G(W+Q!hNq_1<>HQBTcPfzl1bsJS(*+vs6V8DM5sjf<0 z7s4~$7;mncaJK^D7!CnZv@YHbNaHoEIU^f7^V6vOs@=w|E~9O} z8@V9+9BgGszE*xdRR^v+@mC6$Mm0UzSoYrx={i1G=5TkbIY2S&eL1d2Le!S>ElM_$P)(cdSs9yg+s_|B z4;727YIk}~gtK{!%2Qwt9D|9p0c2_3e`I#S+ynu0x zarjphz8CP+m#9DCBW=Wzk?mZNdvbZM6Tr83I+6(uv^%_+)U;?904e7HkL6Up-qwup z-P^=gIKngMBk9kt6iIYs>b?`#?0g^L`zzaPpY0pjq>a?7Y#1aS1_!NaYfyM%_fVSl z=UR+OAqrM0knO<64SBAQtm9j;x3y_dd9f0KgQqSBa&g0U#bD_WJ^rb!3#gRE7?q)R zz)>fe+t-R5Q&vccPD!50r&@Sg+fj;m^{CQIb`Oc9MW^c;XM-=bYnz=zUAIezBYGRe zUAO>a2d4tOb5OQuEN%5gvRj=-X=Q8cfJOlMa7Y{h*11m@!f$>iUqAr;?a~jxg`z24 z#@Y@umF|yyk45--aD|4sZU8$@>%>!A7g}YUrU`AeB$nNgL~Osj%7oepAoE{6{7|#D zj`qU#%@JXiGJq=W;ZFqqb$zunYo+N(5lQ<5Jg_}B)u_|^MLA9NJ(5ok#u79Apr?Q^ zA|&vBwTcXvhC1TBr~VQtWIH2-1|)pyhmYxAxu@D4 zU*VOi4MbYR*A8a5bt4PTc8~HYa??-U4r#8Z9Nz=dIQO@TokvU?BcJ^ay$iwr01sgB zzlSA)=T{dI%{iDzVhPni#tA-^&+1+t@J+6h3s0`yq>r7KW>Ue(9P!f?)N4@3W$@0? zB=dCyRzeyS$#6*^;M3)dw6Hl(t0Y|qK}O1M&loFg54B&L!!cgzO%q^*06?0O}|+S_f+Qpvmid7)~3!GW!uTmDWLi<{hDK!m98KAsmpW2Y` zwzgPF@`)@!!60Y8DzrWcy}C&tI*O{}=HG7057V0C@2{GBp(MJoPdX^tQYG85iVlN3 zW~t9~!0rCkA-NbA9fy9s#cr_hYW&F0+sSBpX0@VteoMQ5?QLSt-tIUNF(aaq2XVOP z6_zw#3Q7(&%RqpYAybq~=XvVV(JBZ#8T!8W(K4G6mJXa|WuXCb2Eu}|qZx~(0RuDquo|*c0 ztz+*ilH}u|-`jjNmgh>m(v!q?v%9=ofioy{)NeUhE*nnRWu^vz6Zdr5NG9NbmTiXVl#6`r+WuO-XDC5(b`@~Q^F{w^yH zeF_Vk`J}h;J)FWx3jqr611BJ49<@h7))xLfQr6~48sge17!Aa%(d;-p*FE8nA5EoL zM;)eO*xeILh86;AhxHtnR-mCbFz&CoA ziEcG`_3L<{k+2|0OOeid8lfM=OPlF!qirV9T1Q>%@Em6&oaVZZwJ4SOA9()&I*EJx zjV@h|TSoX>HI}lvknt)gM^jI~+U{uOGPp09nWw!S>M<}ybb!pz7ILD!ns zyS!4pMr!@-^LMJNUind^q_|v#7#LI2J$>s!$Hi9qmZN7KywN;T!0jOo(Z<{X`PU)iTOmJ* zE?n>mCP*1ScdctmnyT$`CnptiqMpk|)?<qaZx1wTRDg+Pe8K=J5B!v#DD|vP%SRQ5&IR_!<1`S>dBj zJjsy?j!xDm{{VzJ2_VS-0Qd^kH~cH~{{Wt`{{Y{ESh^R)>k*+^-RttKH*EQj5+}o9x{7Ve;_;UIyYk#y%(GZa58_%VAWIhVl9aLTE z@|d@SV{$1YHOBqFP`#2y5lCcGGI>?$+vdl$Orn`l zZESKEz8%(1wP+*oyuW0#-L~oHQc!F?068^-XD+v?SXsz)E7&Bq2%b~hxC^ulV~&;T zdUmsRmoh+O+NfB#amx_7w~a1tipus8f=59+f%;J!M?@*g&hk0kdq%w1G`m^j z)2;-u%^aZ`A_|bhBC+*<3w%SLQ=RQ}(y)+roW|KX?UT^gd28c4m?uz{F_9xGlGD3j zXSQp))Z|!n*(G%?v}6FqdF{xpT~23pxtwin4-2>OmyG4`?d_H3k#N(@xL!~Jz~h5l z_P^mT6pe03(?GnIBB-WBgtjn>4hZ1%ucEv)Gu&xW%Q6JGz!HUaW!;Y3xy4cOFNmx( zo4BC5wY!zsHvROeLoR`AhFC-T5?xB|XTQ=>$u8sUZs{a6MSw}vP zdgU!4U9PSWFdYvVuT_J_Iy|%6ve{}PN0Vp-oB~f$Gm75vCa5f7h84#7+W?YAI(mEJ zii?q5jY{X2Sa=#sn`_I~g8B_Y7RZ7`0yf-p!5xKRc;i9T;+sv4wCn4fTe;tF3pkC7 zo;n`Y>$;`L(xb4NIihxE0l}59&F#-lwa5HL_>ZsX8upiEq9v({^4LsF$Xf)GI)l=t z5>8EDU}^R^4F|*b+C))a$!~u%IVJ7>$sjSXJRY2!aroA~zLl-(Hc$o8F3K(IZCNc@ zLMcw99@*nQ)xWKH$HiJ+lMb(IsA@eO>W5T_i-}--#`s?5M3*wCkug`liIZA=0r7Scty8?JX3U? zJj7*bB}1GOp8o(!euLmo7Fp`rG`AWTd-clLpDi0D}Ebd{B|-Sf?DN1^zQOiu8CsV6&1 z=Vp3!^c}gbvfo_LHSy)!YH&=z{luGVepv>!;f*4RV~;yP<0mw^5m#r)2g3d+{p_bk z)P_#rg59YOkK(Bv6nJY>Qdnh-E_?G{^YK%~TAzjVt9fkUh{r1sWO9D+831G3HQH#K zpG2~i{`{~ckI6qVP)}M)oNgD;;%_`Hcd1D0ai^DoqxnwY2yy_=Ur*&+-ktFG#5#7M z{{Ux*qztS)&oN5@>5g(g$E|w()xErg%QLctTodLM0D9GH5A56Ew^mmkdZF35CM+^P z8mU1ssWguy{{V%1;ve`)bqko`3u;7Z4&%!eMH$HL*0S#Y7x>ch;cjJ`BRQY`r^T!m>7^Nr#j!tpkuS(G_#GXOp9Rpjk@s5_O zr(RqwzMT{jO%1yh+&2Tzf1k#&=F_$Nc%)c#O#CG^HBCA{P z0AAq_=kr!_qt7d|T5d`@9&e`U+S}>+Kbxdoqb1zi zgWItxHUKllJ5!TYv~scdcJ_ONjLjXhajAH~> z8h$D3uv~qbSnPu`?*Sxc!1wvMHI&>U^;-!=$$FYT2GMo>N5fZ;-D$Tm%x-sqK4Ty- zJdN0^QuqVJ;oL=JQAx)1x&uC^8LPS<#0_Us(t-0NuzO`g+u}cq3VfqibCrBFdY0 zG|CGT)RUUfgT;EqpN20cyqrTcGboMXh{^JnBOMM;6@uPAydl~)x|Nvow%o7tBNUvW z6|qa2THLvJ;JsD>cX1Ak6wPTOxJiM-mB$O;Ju8j9_;Idk8g!AxqRx}&JEX?n<(ln$ z72}P2SMjZ;pF4%qUKb@DS9x#=>Ivey-xg~+et`*gxqDfRi~GqpGL`DZMsP?U&b5Zu zi4C=*JmaIS0YtQh@veLf}AdW&#hOr@Q$iv^X~Nf{2UdEIC(ff z#7M7ovWD4XS?%4OB~VqjmINMi$*vdTXN=p!k85RWs&6Jj!OIX7bQ})3B8f+<1t+vj z4L`y=ixbp~B1da!MXOr5n^-qWQHrfpET)q6Vz0}f29lgHwQ0G54 zQ(lqqDI6;nvLXk=sFgYXa4}Mw1Y_# zD7McC@Q;Hf{=?O*4Ys8O_YpG3EYO(LgN`tJ@mW%Ma`F$0cb9j%W#d^wxW^dDZdN=uX! zUoqgGGqu#Us~BOm(Dg@>IX79$>0O4E;2YgXMAD_7NR*Ybg;+T#+koT{dRL(*F4tT{y)4D0$ zS-z*6f5HPT$pY(6V2ZA*7(urj0g^tSt#9i#EotCH(+-<^bdnZ@qPCCcL?HI`?ODD9 z_=Bc++gG@^7Vl>wlF*}0%*+QQo^kk8KNGxht@v-nT3(l>%^XR4Z;)CbI6wjaehW#Hf-V%n`+UAAo0~n-JG`9zH`X_T22Xe%8r7t zo8l(4x`{)0zIc;n;Sk&DoL5h*c)P?JPk>)ll6d~wFIFJlP{zVND`~D(sxoktY}w|C z;Xe>C$kDIA@6oS8@Xm>@LEs%u${RVMzmi2UqaQ4h-;;`|7sOevC5+6|G{wSzu$960 z=Nx<2N#M^HYCa(G1=Y0k!eNFoOi6;sc)=uoHJU*7IXJuv;|Z^=@2}$%i+#my0ee&O3R0sPQ3i;~D)jDv3PeEeBH#<-VJAPCjMwjsW`O z1M@YjrfAxnS8zzSw_*Tu$}j`yD^Byp_cPzzH`^je)!0U)?K?BZdGxEE67hrF_=ew0 zd&uRAWPAh}`NutefY&Ok2QE{k;p!;!Sn@3b$HZE9iKDnWHSvx~8aW-UoP`aM+;r() zfpYq;pn)XPuCloKe3k)wam9L`wAOZ(rt<8_04jrQg?T3fq4hOO!kVR)v1c{asga7v z(Ya`d3QRLk}^6(&k4yt)FS=abADp9|CGx)#OhdoWT=0F@SOp@-?;Ne;H~T zEcZ~_O$!xZf;Js8GtN1zTW=m~Q0may3!7E7c(>$@%pZ)dY5N5yYwA+wQRW?U@aFbg zl$A8qbTJ~n_FT3U56_CPFT-t5X$hpl*f|+=86Kb3wf-S|K|USTqtl_bGu*T*CQOA@ zLC>dJ=dQjY-DvaMhhMi_%NLqlIgOMPz!>20JNwr=!Y%H|B$rLqbW8gyL!?ULD_{wY z8PQ2*Jyc`3tvk8ze1~|n$%B;x3ux#0iqi3Citp_-h%W5q0y%9Xol%$gh~Q*;Hh!`JVy7CEbRb^>3Wle=aD}n*X zyXQ;hZErgBSGILEP4Uk%(_M{@RZ2a)aB6gT&Y7_UE%;kStuXL;r~4JgLK>Hr?X zy_rCc-v~|>cqHb$s^`T1BfOU6v)e~)Y~fRKo$gQ2@Hy{VQ?BVdp*V9bOgnFf{v^}2 zL4V{jTfu8AZUd_qU8Dhl$?N(G=QO)LLekYYXtzax$^QUFe2>z-XTrWJ)irHSTYIt) zM+-K?9yuKk(z%Nd8S3)uwsw9ZxE8QnM)F51FO~qo+H;ZD6&0#F?(C?+MOfy==Bf6Z zbxk8u&zhYq-vDHFBv)4}YHK{eY4>D#jf3R3uhOM>nPYknWp^+_xjQi$I|#wLVHQGo}Mj`a|Nin0cIq|2%|O*-$d{PAz_u0vbV^+xd~ zu2)LGVDq!2iEQ{MJa)mae7@Cm9}mp0eW^ulBMd6bIm!GpTy@vQD?85`r-h)hnSXrH z$#CtAx`IGZ4o*k|tzWZq*b}klap^kmks}K&N;wp7Ra<7pa(%h4N4~hby|}k;?CnH( zkVG0`3Cf+LC#O+Wb#IAD;TvmRLTf3fg5u$=%v`G|QOg2J1m^=4r2ZQ_kD>Mql*f0YFwQDKLO3jdxTBDS|@CS-!(`;5P66DDYu1Mp7wLm?6tET?W z@qhgb_n(^o0F6!Y2gH6g@Sc}rs#sk@LqLN2bUTQU_Wlf*DGOj4YYDcEN>VR5Rb6oyPN@CW~DW(8hp~;x}~zNWM=3E zagz9pL)uLlzrrW5LON##6i<(;fohZ|d*{#;i99l{vX4aO_b zbbk?eXHwLb1-{ds;mAm1iP@QtIT$0-tZUZZ8_^+-S?sikRgMpyfFHmClTzB<6H-3W zBzfh$%GWTB+03i_#d_wuG;!$;9o0+VE>BbGR5WcrMAU6o?kzt|VY}pOa2R&SINzP3@90p1|Xt*`X5%aw{5M4V+1)&Za=pEXN}`ARJ=3ABlSH zwwJA1MLmVh!!#SWMBzXij~t5H@XoiRz*-$H(loVpJEetLi#NF=-`2WKYU(Q+2fdJ5 zh~iSt_Xhz~R8*#?9#(4y6)VYI9x-L&{XMR1)B7IzNttkb#XRTfUW4OnnG;j9iAtzw zvW1Boj!s7%cpWPDhkQq;c$Q}V&o)y>5&7^4b@V$(^{IS8d!bD~*!4K%X(HqmQ<0I_ zip%QBLB=nB<hS$X28~aN%a5kC|nLx_? z#{_Pc=SSCU}3ET}X3@yESr{4v$7th60VP@J+rb!0s^92nZQ zw0M3RYnQ)gk4PdSSyyvP8H)AeIOB@%bngTBOZyd~v9T~&8NB8J*nH{;$;tH1Tq6j@ z#xGLmHzu66ISoh0nn}O75Us?-4Q1J6c_9O4cW{6F|mD&=ZZo?P`+a z!u}DHSXj-dbR>TF%yHYVOjgCh_*cLhS-N$Q+FnYF#$^S7ILD_MsfwNTVv~bbiRIdi z_kK?6iQH<*C=m{ew;0O$d)KD;Z%~s|v6|InM!#EPuAX2cb}^Di6_0u2Zwg(_9BZY+ zJaQ1I(%ZKL`tikeULDlDIW^g~(k;>%iWVVa*~L;xJ;gb5C3Iv%;caD?d&ppA0AMJaT{iEvM-u-4 zbjIZ8^sQ_`;w))r!|}5-wx?oP;0v=5-DaFZ!Tta#(2g}KBV+H{d>Ur&Z((hO$M?O)^^?? zV1Fa+Svp^Ud{w4ht6FMNTB%{ZqL7fo{oG*JqiXujpQBotudglb!3SrW_}qPmQ}|U4 z55?L}w#FDWD=98=aTHQ=N2mt|j#aq%ovdZ}k6qIA%kgih>Q?sdBSQXTZ8-I;yPp&2 zX>eK{M&In}pfjU_xji`){{V?vKZ4TSYaS=JnTu_NJS^LeGC;sRE1vK!gWye4Red{8 zxKnYRs$N2eXD2!IKhCe2D*{Z?@w|?FRBQCIAHS$2Ig`t)=)I#P$)~ zT_WE>12UtAa22>D9AnzK#_(5ynl+YP4%9?A4fdH29)h@~PV)DTh|MV2;(RNqX~#op zjFJRvz!6C5`wkTTb=iHo82AGtJh$OpKAZ6O!B@I-Lt$eiP%OejwnhhjF`Sd>Ouq1! zg_h+!*&J>z-@ZY`hQQ%UgL2F!_+q26sulPdNltVWeo*le`u%Y2X4dO#4`o*zr>E z&Sffz9A|*PdE*;refKT{k32?IW5SWa@0z3gLf?c0dRMMnY1S5w zmxeVP|T|R&NJ*UqjU7*CV&Lg8AfWH_vcUS(~q3JDPKYYQV1-smuH{ZDzy9 z+SZ?_*dmLR7e?heG5k1OSof4TuU0=j!E8w-h` z)Zuk$LKWQ%yOW$`alz?X+NX~6T`n)*T`W==j0h$LLon+~F={C==GN%Kwea4ft7|rQ zms%Z`mbTLgeArGMvD<^t*4K*koBd~2(Bid-tQNOI=2^>t03XASr!_UdjI3D*xF^is z7ZE97a&Rj*QuvvrT56XXUAj#z?YxC0Qxk0m*ow7Hy#?(o?o{}KnDIA+=4M2UUhQ(g z=NSi&#=F~*6Ix0q$COa-&rP`Iyz5i=g`mlG18UY%h+vSsY-3_U9S$*y?re2PwXF&r zM^6FbxsQBeP*))O(d^|G+v+D#NhR3%-{DL#ABxQmlp~hN85}zgnluN3KS5ic9Oqka zh+Y?(RtXx!8JPA0O7%?w*GvA>n)ghw(b8N?60?n{(tOwe{X0~sr0$Dz=2kd(d>+$y z(oHR`-?*| z_+x*0V0=lhTDV(?bm)oA7_c*Vb3} z`eC@ugIh^CiIi}`dV}~^tK4{JLAO7@@ZX2gT>QTv@tnp1lOw=xQT*Qkkw{`|c>F7~(e6AbZqj+m|h#qXQn@ zD~2?Zp8B4&^9qus$!#oj9}wc2&RG^xZ#5(c^gTnghig{a1yF&vgb!eM$ zEJzsRo|W`&v8-EY)NjJpQf@2M%`2LI z06M%r3x?ZKw0nD-aLE^)#`Wr?+}u)9jKnF_e!q^(Q&TKT5SVO*qOoY!l&*Q;_sFE1;h%-zkuPm*Slz`am0j)`a6!&^r5!dIL_JhneOnk{y zebDQjr22o2bdh`}@g}L_yX&dtTMK`-Ng`bC%dp%?WdQ#Gg?g@+;Vl-%#Z*?R5E9vC%mW>#1w(8n^K#{Pl+0P#S_0D*c!M+9; zi+g#eS-sW!w7aE|$C<`U3|6qK%W`)Bls@J z#y}z%$@d;w@ml&Hg?=4)hf+3{nq`H=k*PBIa!bdi3BWbkwwdAG9s^+ou!qY)7BE`_ zjyWE{S2X1qK`Aq7%i2xK+n#Tw$X@p_k=KByJ^9UQ_=88c)%;iCSG$oyS=&u;BRyS& zfC2Q+HM^zh9vsr{RnpEVN0w2FNh2HskU9FAXNj!5C#ijo=IJj?mXR@nNFCE?C!CSa zGASzXmdzQ)n~ZEx@pZ&Ht(4Z245X-4hZW;L@UJcZ0O$Mv039{z?w%dJL`mN8M91$N znKw7n1D^F~{t%A=dg_1p6({)Bdq{U1rzOz!4N@C>jW&3$z%Vk9`RR)J`^VlL(sVn^ z6w|I{`#s0aAeD*0+&Kfe=lNIATAiz2>9gA_ZD~-bK9%OTUk*afpKa7)lRL;+vT>1) zKN_Z%?8&)vS!`!$-wt$*JH?t-r)hg6uN12ylP2u&Mh(u_I z^siIW^jPAu)5ai>+ad*-GxKd2B-S^Jd=sZ?gi4d8r1BD6w;{55KU#{>QjMZ%hrAu- zsm^>T)IQIuU1|BkPJ%!^8Gri6^sYb023<$u*8+@OQ*Hx*-((o`$G;nQa( z)ilYaypri%5kLoVY!QJ~bik{w={j^8ckW?Gd{c z-p<}Op4!c~j%$B-iyRObJc<25DUZZED z+^2?P)2G-a1aHG+VhOeO|oq&m$Um3>t+BOh>#<%5ZGm&U`+6ZFP zE*<2O8NpC7oF4xG;a-98p)MuXCAgfo+$*)Q!NAQ?x$riyyP$Dnxk?-~{c+S0iu8XD zcxpWkXLzBCDBmo>4+NZZ`d2+TNlxc=dXRL|TeHbOXjIidwe*=ovT4eNV%q&>&Pm6x z+t0mwM}yY(KL_b6sD}GswTw%+c~*%gax zqiS|OG1e|F-r6fmxqQe!eUih}jB(PpbX_{&c!BT3ME3qqm$@G&c4+7aA>*rkS_|>^ z+u=35#f(v|Ku|qLN{_(sYL^<9mt%flf*r3a9z!mA0n;_x_@_X$)@8OgSFLVi$WQ<~ zvx;wqJUwNkT--}(eD@5>5rY6R&TEg@&)V~JJvED`^=?nK=y2Z{G$|vJ-(HgS9*Wl9<{M^; z;bt!`cHjar$JV*ehn@t#@c#gaB-E$Aw4NagL{<_>1dw}=YV172FXI-rH2(k*=vH%} zk^;VC?f~Ex+DP~5PQCFSkD}>fD>*!=_Quhecr057jCRLb(D<9+OV1VS5Z@q!$+oug zlgyWA&Ric)O0TB;JJRGfzGbbUxr#D}Wx+eLJN_p%&r&y%XGA6L*xCgq@r3s?#F9xh zjk*P5FbT;&=i0V@%WrNLVQ&uP2R|Smm0{{$3D;-Vpw)EkR@T{;(l{PDV)FwP&pm04 zqxiE>2HrmsX<^254}VI;UjX=uNS4S;rwfIUlD}v@ zz#g8}r1$Pg-*`Il?z|&=JThWU&9%=MIIje|@jkJs>kVd&=10Ag%gDnV^~mdr`UAoq z7`@VbDS4&q1~oGl1)-3fDH!WqkB9ye=$d8Y-CpYD$C-slIODDh9G|W+Q2dEz*M-db zcf$LKuBDs(957=58t?dj&W&2Uj&DW{#{cGE!wA0Rv z_ST^i&tXHiraS#>hW`MB14n7VT}2u-C6!2QgO9E;RS7tvmF=mwqUvx>qKFG=sD4J$ms}o`uCGwPzvlGgvJ#u(@Y5TZtq@Am?V^@vcwA`u+C5u3TK+ zO3VY18WZxyx!?g_?ct9PO=Eu*%8PRVR#M2>B;)DLbDFQheF|Ml;&1Jlt!0)&Ac!KB zP)>1=N++w(CuHq&hWK^j3(Y6RHZVgphT7^;6tL&zZ1J3SHJ|YZ#vf(ZCDQHf<6#BU z%_NW4e&*nM3}U@=!`}_GZ8KT6y^B+OsqK%M<&;pQeHXFBWR2=Q?%i)iPx(<=1Nw-kCw97ag#UD(MbJ+V^5m*vR(mej$0m_R;1qy^nFLddR?}@1-7Yg0Eg`mKRbs($ioUMKi#C5zS-km z7x8`1g*8iS>k}g@@pl`(M1wq!uf2PAyW%^2OGUQUH22*UkOp2k2PY%yE1J^24rvK? zvTC~7yS78JIO0zzx1eKPp0VKuXf3Sth+?<0f*pYY#twS>*ECZ|(BW-$3t-mQ*7IIN za_49}n5&=mf#S1%Dcen`+G#>Lyok(_i3UjBlDvOf?eyObUaY1$H5*bK;41QJA0Buo zTGlP>qS4~EaS?LOG=*?*G3*c&EB9hlEzhC58Yb^!4Vw zAK-qTwkzTsTy8G0E=O)nVo&hf#V>BKA=3jMBtL59i8(pr@UFW;@E?n`%?DnG%Cv^+ z>L`-fq~rKc&5{QkP^ByCT}aVUc6$D&1Fz^>wui4rr(F1wOS{PkFi0jXfJwk?=V`B3 z_*bZEjx`N0SiLuw5F*Ia7Tul9!x{7#t`Ph-@f4ryq_Am@NzKkqez>li!#)Y|U8aTN z2o~bb-s1WMg4*dOFMt`hX9FXH#Sm`p=;xKJh&)*?mt(G*JwI1zNm5=kWM#J}peNJ$ zVzuLr+5Qmtg6b`BYJz(ycgml3G8FuwbH#bP{0Y|}k>h=LOh-R>ZhZd$FM8=b9q@m| zIz7zg^_w@-Zy=Hjh~g}F0;dO&jDw0JO({78n`+0N=+i!yA&pphPZ*IHoDP4z>CI_c z-e2ALhQrIB&9>a2ux##a-$7obtb8!Hc-3D}f;MH@nIq~wII8+z!fEVn;(L8k8C4__ z50md>wrd#CN=|mv+Ets1k3)c#>Mbuu(%jgauQ?Qy+yLMMQ)lshFet5J8e7>`PQGpZB7Mi>pj?cO1k9mNX~yAm963r z0sjDN-Nk)zF7GU$WRZr!8sj6TeQP^J_Vx6v$Ne^5ze8SaYiah47V2Rh%V-MEhR$7^ zmB95mADw!?iZq*zV@!tD+%80_To%X6&TEr{!}^L_$!&IRe$zIlvouCt1S-R5;B~-W zDOHndDRg%TXy2JXfNkZppHz-W%1Ww$t8tFC!!E?T6ny+DBR-Ou*#h82}DAsr6VV{@IGwx zFisQ<0rak8#r`O35j~VrK?-f!^2pydNXP(JW20M7ZKnuk#(#u{&PFT7JVo%|UDS1Z z`7UqlicN99XP#69^v2W89Id2rNrc=J$~NP|+pf*~xh#H^}DM`4rfI@G!zfVSH8 zt>DycjFG8Lh`B5~jMm<-rAs~3w*}%@NC59Wv5sqt8BS7yvc9IUX-M(U26($x*EN}} zuS^#Z`9?t`QS9>0dLONITG(ZR(2I3-5jXER%KYbw=%Dc2ai1>8*+$ZHi1^))!kejh zQ&G5v+93_Rs~V8SpBT$$jEd!jDY}kI7FIWe;h~g?IJ`?PUYt%PSudM7d-d$uU7b#$HSd>{=Hhh zLtR0IB*Jsu2 zqL)#TqIuO_L~5Xn^yywvsOwt$YS)Plt*B&9-~n>@Uzp>J*Kwlgw$N(&G>dlLXt{{G zn~dka0M9(tm^?dar)j#C{INB~%G?s-9$7(ioO|HZO|s&7z}GxPnDcdgMH{dDe7Mi+ z*1Z>3YxdP4c&;R5dwX_pTRfP@tv2ezP1U?J587cCw~nU|Ai>Da27P)D#;;t-Zyu&0 znd6MDwWuJtU-@U=vyyJtEj6i?shhai815MW8EhKS@stzI;|rpf6QB@Bw`7C?@{ZZX zMXg=fqFe-)0(`Qg_kf(&O@F6HeXm~KNpS>A1Tk(BwlkE^6%vcpA;qPuI4E?h*^10A zryKE+y}IM;UW2D<*FG5dXFi~ieXdv~a#C3c;GA{#u3hePk_#XC}VGox;Ga)!O9YZ7>3Z9>Du_e*yaK45_* zId5#|t}6%Nw~6(=QukV0{Z8uh%UFU7iJ8J=wsD`Ob6za*q&IqXvRj2zkOp)@7b}jO zcj;Gt4_@2FHoHBow76x2Y*1L|EnJupl_uTJP{c+BbN(mPHBC!K)U=y>2^!gg;ohFgx;q@o8`$(GP!W(HKR{sDR-|)Pb+HZ{XmcE{083t?_O9H!3Y}UVq zEaisJW%Y-W(T#za511NH?7NHq03X;qUJnrE>Rs;Ep?JzoBN8&`SFZ@fbN0b`| zOMr5Ha0mF+{S(K(X}h%_mH&Dce6A_=yYUZ2Z zZDlp9E1QP3jvILc60#O21RQkstW##17yck$>M&~BZMBvB){)GPU8D6J(cv4b_SR#! zn^d^EUo5)`JJtUHh-ze&F z*k>P=Y-U687sZ!aJl2+)UBHg*SW74>8A=?04t?r}hkSE=bF68}tKGp0PVEVbBnlbG zJAF@D&G@OM>R;P(Y7p5-)=zGT^C-y3B=j9CknnA-=Ck7cCP-~2n#$hjNh7NKs0Tcr zopD*oyOxr%&3se8zK`OLu>@DQF~%oQ6(wK?$~$JE@m-VY8cmJWrlESWNO1X?uz1dT z*0;ncB9G!Xh%YBWwA`dh$FV3Xkn&Q=E$&E`H^PN;p1zfOwk0Up-*eqQ7GIct6X|z% zGCZsm$t+1`JI)1droVQQpEFM>8j-Ddmu6#4%Ch_N)bE4f% z3SKI_(ApT{byfo)4#ZPJXVE&_UC-f*vtcU1cP+8YvAY_J5yOZ3lc=oq@cc-KS?RU{n z(TYy(9E*>>r$423TBn2uguW8b8$+`0=IS{n2148hJ4ZdhJ*&+_rNXQ;$Usqm$CDrb z09vu1#BBtb(|9KO_TyQR9{%Q3YfFwAW+OgRW4%|gvDB^{7_M#u0x_9epVp=DLl5{$ z)d5fV^|&7(;lFfw6{BQlI=puA$l@nbtr*5|N9Rs2O-P#5x8Qg#q4C|Qdzqw&thoK* zWj`tRuS3>#c(j3RB{^xN+mA2|FLTXeXgYnpkBjXiw}izQh_GI`B@SyR$1>WV7sC^c z&XD7;Opa?+E?Bz~q0K8EjT#;Cva!L(LtLMNba%S9lTp>R?MCKznnJe5@s(p;#{&TK z#w&lqj+1y}OmOV@h%V+iJRE;Y@XK!*y|v}c2KUP;JB(xu{Kpkj0mKVC^w!Ljtl8^6 zFVNWN*BX|Gtm-h&7~UgW$qGn2jAw&cn$L)J6udA)cQPs}hG8MeJ4S2Hul#E(S{rkG z?98NKHgYSp*JASST2(^AMN_yQ2g);7M%UCNO9sEG!s^2I;g>xgQh)tbm*P!FSiRR+ zS?U)-Jcy(UvoHyj>s66uxtK>9F?0D;mpa~`;qMvAYpRmAGg-qCnMO=(6nw|qtt?Ws z6Gcn15RJ4+ZKM{{ZlWLo2n##O8Svu_fEDt#Nlc--h(nYn0Y) zZ`_0AL2NfWAAd^iJ|3CxJPQ<&04g}t?|6k*6m2kFVnqMTPdux`G{cBJ`X(e-mFO>O$DToDq%8v0nKzaLMt}zN!y>P ztvet1^IV+u)Cmp@&8?v~}{{Rzs zvfIOo5v^Y0>Q%+GG2?X76M%|3`;*WLv*Cy}v1D$w+1KrN2Y=aO07q^+AU#PP%}sBl zMK$z#jr1t@QD@0wGGYXCk8@g6yQC5{NL|8`sym+5H00#2(Au$9^Hs63lgpk#<{pR? zvmd}!T^Gbw8szsWVLCXrAtYR3m5zFj-7C(nyeZ;q-FD+rp62=Fl}ZF$AVtqXf(3Pc z3GkcveoLE)A@Zb{1#h1@1K4mYR+2=SpA&u{9|*THPkGy16p~z@mVo2ON{%+OGIJ z#`<@REv1su39Z^96_!=wKE!ZpeQ!d&y_gFbZDNW;nMCn#=h#(SKLp7H_R>qNT=`Iv zO|%HHME?L#1y+JBYZtd>S#1N0c*yx#2;RfgS3MS~s@X~k-pcTY3QVIQAw*CH>^!0I(_Uj%N%A#RL(+Kex5JCA!75L>M0V4DZ?l{aCj)q5agR>5hCVH6V$E*On|2VXasrHiI`^%O zd% zRNhD(Ml)RIr+cbAa!+mKfQm{eU=A_gBD5?P{ro3xtrjm$vVxNwVX@rvS$7|8wy9*Id_pGF^nHQF1$6OCz@UE{@)a-HLs!3X%|a$_cF(E=RqUyi1rK91KW?*px1<^NQ8228%}GT_-(Ig znwFg{rLDx0T)@P#yhRSucq2WDCb~^+)J+yhA1SRDnJ-4tGql*Q1>8mF-G&5^F`BXB zyP2(H(!?^fVq4`$+SuH?7~uPGDJ*19wMFv}yehwsAIY%OZc4F8FVaWA&IZ;W%!&j@ zH1{xDtF6&euT`r)-EXxMl$mY0-FCOaM3dX>0kEzEShidJ$A`O}PiZ=s{ z(!611tLiOdZ#0|o;afP}v+o+~w>S5j^R8OM!rC^Qt4XO^*~b;sncX6j->wF5GhIbf z5v3$)Din1Lr>G+!)j4#FIWe{*i`^q$e=$WBC}ff8wvlby1bdhG_AY-xRc7$^mo_nH z1Z)m>Bq@{mgGCfD9|nh}z*U8u4I;MS?Go+>>IGYCSeiCP8#i}QGes1DWOgv3ZDHo~ zoVtR4l~%pbtS_5%&_}x)N%D}AKNU1lOb9>W7t*%__Gq{{{{T}TKltrgy3uTI+`GXT zbB)X+LR%d&NIfW`lH6>pJTaxJke2bTNASZK{{Y!*P76pa;Sj?bq+|G1Kp&8ziU49- z6qG;OXCwQBjDISUTX^PNI)lamARJLeF%3<~u0OD%gcCo~pP+;Q)fB>J$ ziYOTnu^|5d$KAp9%Krd`J#5p-c%*lkSo7v!{v9Zyiy;-4Mz>u2?}9Onss8}P)8Nvr zpp?o6=gvW7Uzh2CD58NAJ*jyMi;0vFK2;+F^rt~?VFz^5OP=y$kH`urpkmDrogTp> z$L^TV>L`K_A~2p~zB|4%`U)tZ1(Gi@NZ!ky>x|TP_KR)^S!7^)HaPr%qKZIPy3n-S z&Bkju3$e*_3Kl2VBpQW0H>M*Y`!q}wv4%V!{C0{dp&BV`6fpTvsEIoWzyMTfX3FZu z?%{wr!xh2(D58OuNXW)bR{IkIzv*`zcFObmQAGe*0|OOWOA#c(Lvb*{$WT`U^rDIY znoD@4O_CzI`Y;4lC($h(i3_tQqN0ECG*LhkL#A262G}Dk0UL~Klb@idn?|&0H>&LP zR7C)Olu<|tVAHMb`3M=a!!oW%qgh3}J->9*4hY*_u#7eA;pQ9ua}mu(Sjk2gJ?SC6IzK5G<}vLuo}WDEU4qKW`$ z(~(95(*1s3e@c>DC?-M~0650rqa*VaQ9$~Y;?gbRaNC)=zb)t#@Y_J(IiDZu-+#{dNRE2Uell7vC0Hpe2 z$^e$;X$PZ(BmV$FsUfoR92K~jpcoOIyMIAN6agf#848g!vE%!MbNPxOw=nL^^2?w3 z=tm!r6j4A4HL}bA-pT&%IsC;(1%z-&TuAMaj4)h&fKf#NO)NfQZ4x|u^-O2=6vVY< zkU6=XfH~U5ar~&FfFzbLH8$^b_grW7s709@f6}Mm=c@DiQAGe)CK2xA|+iZRwKSw`8u!q}IxXAdD;h^%7|k&+mhVQ6G05fMW6 zwX%zneH&@6e!u(N`#krrd++-^=ltu0E0MPjW$g_Yp zKub+cLrq0XLqkJHM@!ELxxmQ4z{thM&J5w@=I7((hQS2jm&F8xq=jKHaYYGf8CfI} z$uFj?ri4(vEQdt=y#z!@N5{y(c=5u8iwHrOAmaZVoM#sh{Ca0!n78aM5S60`)uW#(^?(H8O9vz>M{<=T__}{Ga z_201n;bJ{^QBYEXDQW(?KolY86U<6UB`8bHrfEXs=+7>Mh@pjOWfl`V=!E4=x1mk} z!}J_*j|Cj8)VE^r!0T{uc^Unjb0vf=6HTQA-QRn30#9`GGclc3cuRRMPlBr6p&-B;D|zsM1DWK z`BTF~eYNFBLkdHqV-DAMjrX{o#gx?`vo|zTXjA67W)iOQ@rfO`LmzMKjB(RDjif8i zw;{Npa_E)Av$edIfHk*Yv7_r&ir3@h9hrRg$bjL{?yK|XdG(H@6w~EV-rfm+PRhS6 zQsdusFrV$&jTQY+DtGx?AlB4=aMpOI9vjRP)OPU!796(Sa&zI&uF=cdb7`?l~Z7nZ`$tbr@9r#cPU39-awdHXTJfr<=mtL$}kqvf$`UeF_ksAQKsFM zEUdfAI}T2(jgPx^5F}p>iYdS8i%E`bTk+=BzC;G-__}LAhlEnQ2?cdGnmVvf9*t~q zg-p@DR@C~_C`p{eS>Jdf!u#7<_a8Kgc4_YSYDw1fvp|y@Mxr*zds?dN*GKAp9cx<6 zaFGH1jIeS-cz0j2mJ4yagfHioQCmaX9=tBo#oXx?Z>oL*>(ZEIw(Sqj2b??5p^RzN z!y+Tub%~ST`ZiV=~f9qW%B%*Dh9phsqicT#pgdl#73vo zcL-42|8P&?g)&ef#ojJl=k`U1gJjg%gdHHnK-@5)+7y*GJ%8fX_}_o)dzqhT%EjZf z?`LMik(vJb-QOIZ#DHrz;4{)1o_8kLWXNPJN3_*3Mz6)_y+Nmhu|<`=^UJE+5~c!y zxzxPmG*e64B3yjzb`A(X#?C&TT;nCqlM1Sv@b95I}VxIsg4R4v${Ou665vC4U#7g(|QDmWEQ^{y}Qg7g;lqt zYVTJ5f&M+({1kZthA9%94&3^+KS#*pw8MjZiP_EKhcfCfo6lQDwi>=&_QE^pm@8@b zOT7|lvVucdev-}}1}*Wh`+QuHzUjqJnvdESHu~&j_DT!i&Vq=Xf%?6 z=~R>46>p^YL(ypwyL-#@_su{3bP$}fcRO%#^ka-E(VT8BXwO<1i+!h58J-kkXy~gs zfP(&xk(22Q53cHMx#LXb60G8v1aIgH1WViZ`0vX%L@i!I3>kXWHCiskeb+pKuBAx! zJ+juOcO}`t)iSk}S2$ zS%?Vya=R+J%qi!IHx>WLvnh0r=ZM{NvHene5lp40Y-S-k>Ql#~ z?ay8bSD?@Sw!UuqHmgr}C!yQ=5nREIWw{wE0Es=;i3V>4(?=f}eYOn9vIBAhNm*R%8mW_|*eX{@Pl;~8T4K>T%xqzmXDWxGN% z!f$Krlx=ABy?*dVXKOB6DxyR1^nm*Y+w~Cddi_snX|@@6jlA*lywM`#$tQnF%d-1Yrb4?w<&pg@!W1@Q~;>+LYVMyVN z$PhEn+)UeISDBS2GH_>-32CII_f@yvP5U$V#PPI$c_+cc>AhuoP>b)3|8MhET={Wk zm?ztnC6vmB>>J&zS0VR;YmS3&f@aV~3j!GPCS?B*92)nA6+^OQY)(i_JgKw9o?6s? zi!vP=v#5Y&jfmbdZ+Sa}dOgzCOUs7v9n_On8RH!8aszrVIi=0N~!Ws{qiA_)NBxlKUGO$;MHf*>*X3yE@0(}4TYo*n{#HH*sd!SL*rgX zkO3aWTkoqc=Tl%_z7Q^xa#?nF4HPm=jwXsh!&cdOSi;R&-l@4Fv!wi6A}gp@ZOi?5 zAL>J`qJs-9jkxpjolW{-mbN@o*6%s=7Q3l8KPkV2lvt>=a5YyI)>^7!<_5A1(~AP_ zd0Q>27nVbBC)Q6~O~$MhY>JE_8OHClbzYdgz29%xV>X;R?-t{-Ia#C zCrq)DzB?hh+xRkqh}+#ua_%^OeZMC?%3?dn;p<_e`T5ffJy^)YjDpJ_9QV7C6P~F> z1x6T4pjAkH#+lszR)pYc0mqqjtfQdc&*B2FBt3VTyBcXEV}c5lGwCD}roK?LN0ga}A)A|2^1 zbOb_?B25J89YR9#a?XA0y|vz_ckiAx^JVtT?BC3uy*Fu^vL-CIQ|1A`hRMa%Ibo302Ok{6%{2l6%7qFHCa2HJP%N_(OeO{ zrB2Ik>P#mTz#;u2DVttcqq3dTY;;dV#w9S4f$=IAHxIAq9}qEd30XOL1w|#8rk1vj zt{y_)+``hz+Q!z-)y>_*)63iE(c>pU!B0a%U;Y^t9rNmS?7QTY)c0wBrDx>ie#*-) zC@d=eQdM13TUX!E*wNY5-Gk}v>mM7(PE1Zsf1g=eUirEDYi<4a#{R+K(ecUY8SeaV zE((C^Us&YvUts@-i;c`hNli^fP4_n!1!XX~Q?XIg2;QQ-qHap(9KbFl{eqrDBPqMG zok3W}Y>(3=aFp??i0qQ+{@-Z-kp2I_BL5?@{{j17t{H%tih}%jRBQkoIDvBc%6Qo^ zE!@Sqs0)?5@dx_88*K_M-cY!~5xwwW?oHf~SC}JB3)ueo!?5+%BZ-5E7)BB>Gj)Af z&!!ESvSjqp-&!kE042KnvpFVg;uyKQ6h5s;;#l`elZRlph8{l?3sSl$#eeWEn@2B$ z-;w~v1V7|~T0HzRyrT$8Sv(i$6P}{XIy2Ic(<0KMv>G93T({4%RoWiAzfR+V!D8RX zuuG3_Wltl=!iOd%cfMEFBQ*&UFQ4XLIK=d|W7G{ztBG>Su;pM9@O13NWJ`dohYQmU zO`4W#*KIc-o;RAoYI|(8Ie;_04 zK{9bmJNb<~ZcbH6_28cDj*K~@lDkz8z?@FKv_Y(GCSm4Fhi_@7oRRr&4j!6>mO zc&lLMa34Q|UdIJ|&md6>E2KyX@Ahfful;xlh)19L@|j5Rt=>aM7j;phFg)89sbC?8 zU_#F-Ch>`lzw@6WqV`vij@q34KcW@(6dYNNvvZOyD!w_{L<|NmT;-^Hz5|eetkH;d z%hex1eA#yUY)OCbTf8J}>o)^h43fSu=% ze82pt>Nu@h2-ow_BBEvEwYV+4A9gzu42tE*UUV&Fg;0`1cR_oL*jCP)vZQjyVhp*t zFu|WH$gs>M-I92yc`x#h+jo2-7-M-#J+XP!i38p)Jx3V*$_LZ&Y>wW`&px2IdRRmJ z1QFZ}6rc1K+1fdoEjJ#AFyProHOTWv$FKd{|(2{7@;{^i|OOARm?(8dwk=YuFy%dzCR0mBKZbkb|3%e05udcKG%= zuS8~ANjMFv;}Qs=hRGbY*qaV6j>IP=cbXc_#qH=8h3UQ3UShC6Vuu`7!1hpm83(4{ zt(qm5nZ2P=Znj*3A;o$l`Gz4C(PQ!%PG5de>Oyus56f<1$L)vmPDw!0rp3p+OCg3j z=f23S*Qd%H^YOQ+$C8pQskvg5r3L_L;}pe0tpq1$?8ES7&hNdg=}7p`N7q&1I6Ro} zfxKSsCQztf>6+xnm~A9iAH3y zshKs7Y@)?pWhPeGN+C47f~chl-uVz4gaQu}Le0?+x#{&k_o=128knq3Q@u=>!$Puq zPnAFp@MZS^BCd2?fW~*zr_-{&Z6##LqkcswIh0rMeMww3AtLp$&x`7_VWnxMa^!@8 zL(byb>~9APaT0K)zQvSS8gagv)OwOi^(XEM)PruqkTsSPzOF5qvJE!3 z>S&Ucg#AIq2)BOb1iej5Uh~Fxw!}LE%_XSXANhSY8;`?vuGS?Cvae|k8fQ}n+=)2j;}h)b?R5{qH5AdL;L15LAXBch z+cp_g5_}RDla#spz|b>qZok(8)epDZO8c6{&7SIdt6;(R2CH``(ge zV-A#2pwOM$F+P>+tQ91{vE(ecn=2Hw!s#UaLqnx^`e(RqvrV{+R)hI<|7VFi`6Fq2 z#n7u!KC66q^WU}b{c@f$6fRk(gqx!<(6A*p6 zHX61*EiL^}ce1HMeX_dj(l1+b5U;ks6UF1?0YEC~5>n^vH%#Zwf`qBRsa8o%%zXFj zI=B=v^mxf(qi5T3xnrVMBEuxLAtQ0WW_v)W4(_%k(VT@x{v3jn#OW{gLXHs_060ov!6+xhEc? z7oSOv3Uc`%m7>-LU zq2E>>N`Dv8l`5-ARHUpsN|GC@S*gbn``3Z>5dH~B?V;!!dDP&uP^*g+0WAfUd*P_o zJZ14CrTrTnDW!O%*RLzqwMmr+pSQFf^==%yC-L1kE8e;BmFL_%9_*DUEwzPCQw_Ld z2p{-MTJ81*YhFOH6zs8T$kZD{$M`b;oSR9J^Saf6I?9%tAf+2nE3GDosw>q?jjnVT zy%A>vpnJ>7&cd0~QQpC{_Nwu$dFvA#3WbTgZYQT?Ijqjp$H`gn`tiC7KB_+Homgjo z5csVpIMHqqtVHlB8yXB~SFO5gg}OZ_^IXh4z^*Dfz${d_dg9o?_e19+P#4mE;Fmar zhdPrNGf7bQ^es2`6^U}t>%)L(+#Szc{48a0RvYxrvHh|hNuP~xe6xbbd)S28Mnv(| zLE(FBw)f;Z%0h{yBYgSJ;JpD7a35w~Hlwv)msx5XS!j;UC|hQ+zKa!wg*_a8pQP5! zmg|gn%UF!xIuxXN!A!4~+U#Apfq-V5Ca1Pj*jXu^_V~>6CKMOx5BJYxPAr+vh%!!!Fk+VA1?wpw0qT6Mc;do}O++ugiM><0ArDvbLe3&_jhk=Lcn_Zwl>g(`NU(W= z!(>9PkE^HfFSX9|u{U<_vIcf_xCg%=+doYiRqiwmF>H_K%|VyqF99qwQ3s~a!!>A4 zry#uCDACa8-8SeG`I_MbL30R_o+R&Vl$@pPe4_jlkdlLBChgY%G2*at zI-)W@D`42Jk*{|#&2A!W-8Brx-1w3}*=P$rdOXWH$TnB?rnkk$#d^~QcAPA_J-44d zD@e2V=o>}GYKK`s#}BD<67bi)97h#>k>s;4gvF4o0-^lehm|aa|%*na2t8L&eyp_o9 z&9;dIzw5VU_Hxs?P8-U*Xufoj!A49MKamj2Y1m3{9uQyjQw?qKFDA!CdNox2&7Mai zI~ZA0LoceS_F6HcvFG4(4|vaO6(v89dL3|!knjP;bFZA5Jjh_6Yt6^qRa|!2;#tt! zGz!Fy5O~FQWDzu!RNM>SP9x{nSP2J2Figp$bMN9qc_s9MjJZ`5v-BMTpOq#73ut10 ztVBq<)orA}EpFoXG$MfnoLI7*d6ny2xR}8=Uy%S_u_6s`-k={mSF~JX_quM3yk&jQ zaIg6g9U8FLhJD?rP{IB7qAk5g@%SVs)&};@#uqY07z{8;oLVLUU!lWv@WQZbUm6;` zb0$P%fO{bb$lmP1dd@-)DbH6(Kx|IzI4a}d8VQJ!@hAVxhKN1>h2Pn_I0KVvhUj%CCWOeB-?1?7cos3>-bqpC1fmPT5xWg``J zZ5@4meZ)c3A!A)rO+9_x-$Ou9D3qU%Us6CoQul!D0p0&O>?Q!l1Dpe_!612n0|o-a zKff#mTwX9liGsaKgBR59pXf zL{47dmJbruy_R0cgS4n>7IPY8D(HC;;-I|Z5|UEVib@BSRaEs242_IYCP$B1T3OrJ z+MRMf?c(a@?&1B%MIYZwe*VEBp<&?>#K`M6ZpPnANK7K%$;iyg&dJSt@UW=3g!-to z>{)dUt@ingy81UQt!?ccon75S!y}_(#{yTif4%{C0r= z@ZYSx`)}BPxL|uO4o*%mC--j`h$DROfMJ|m2Xr99<|nx?1c}J&UgHt9NH45vh9dQx zm||XpL0)kM{b|M3-?YDE{~avu|C0R+_HP#*;0J^DJ`W57%z&@2wIX`lnLf8Jgg!`c zXlonDI9BzovpoV%n=@m++?D^zw~0H`mIT^FZrwUL{;2ZjW>S`osa6q2L^NG1Tut!) z%}?Vu{eGk-6s2Jh^eY%WyIaKR4{p*;1hQ8H{80AA6Xq>Xj4L}Niz(_Tv z1zK71I#gcs9%`H>Mjcg^>8zojcpcLov&e{K72TOFXjMKs8)OZ_T!CE@T|LlLs^vV+q|Xujh-HJ(`PoUuRRwy@yf%`Z#VjdRlQ>k=IW@P0Ro zT3s=xzS_6NtHSt|z?Zgcz{2#(($p(rXz|DS+cx-%Sx8mMDPNh44Q@Kk zI*}%P(@hToi|=Xd75TbA8JRL*2?#RIr8Mq4Z?V?qlOS2B;d5jH9s0$RNZWl98v{q@ z&gqzzc+2P@i45X5OM^fCk&Cf(vt3IKt=&eU-8fuR4dx}z%l_GCi@;KaKeN9lyS`-u z2=6$&h(*#ID&FA;lBZLkdp3DIg?epSQ!&einyE2leH+(#x66IVbhc8vbpf84>5-N~ zR#kbGG>Nau?wh)NN-9)`^bjm(M8M^aR$h3~5dNIteZ}r3{r*q>ymt_jK z`K<3n+mjzVXF<1Q$%iv*DI8CAJfxKBll!7HQgfUQ`HQt(9~nyx*bUH!ONPCxeUx&p zq)^DsvFhGmpOO-O_==Gt81+@Y8NFpK>Qv5D|0R((V*ylu>8ld*GT4k{d7DZp-NJsR zPC^yQ+|G3mv4}mF%1=-$06_TuBi6DYQ|+}^z9FwsA}(*f&KavnE!Lt zApr88j^?Y=6v~nylmc@D^FJbk-G6!@Yf@aDUSYqat3C!qh3>`0np+0==ZUA|i_iFr z5BHniedsrb;(C5P|9yu=-HxT$D^#ggNLEGN%2bUfN#Z6eao_aN7NNjG{B=}nYhF1) zN_;V1F%fO;61czy;3sfPh*W8;|7+Rd^RB!ubG5Ob4Az$^qIr-WlNZg#qK+qC6*(nS z9`)}|HA)wboSwnPJ5|&w@-}UB&qAJHx3jSgjV`-yNeV)r)*JkTcxqn@Lr;7 zxBt+#tvgsfceBrD9#bZLqYGL8`h}2OJCzd?@06C|e4vVA2Ks(ltA-Kj(AQroSa@k2 z)0(3?|IB>RdZ@za2ILbs&?<6xhM9cvX={O}G-}CX3{jn%HmNe2C!{<9UaKeGh3udd zh*W8F&mQr-IJ`_X=`r?5!Pdej*Hb!=@Od%;P_YtHJlYuym z2h72SI5FEJ3fe1idOM4si}K#06o@Q(<_HtRPyU+X-moRf*lv&YP=$Kw*qlhSVLf>6 zf|Wv?+c}G0i=*I;2%gYtL zTTY-FHXu>qQ72(Uj~a5L6W8$arfDGKnP(2~26Nef_lp#RbZ3(k!V9xvVF&fF<#{LN zLB0kupv_Gs_*XCH6^Xxb+v%G;8>Z4>W_Q<~`F#{0JL4TH)jS`xv*aFnA*PLlz^q#^ z%9HGui=E+aD+gjh_r_hLiv~)UO4Le`w5xH{D%9qV%BG89t=hgn_>;7uvj}sy$wRCB zhSJJOA2ig5$|Sha5~zb~l7&Vp3W<@(PE*|UcYOg*Z1y@bM3A&%86IOiG`nh9L7Tr( zuq3`()&=uq1|?HE=t(R;XGfE-gnrE*@XXa`r^GWKipl@WH%f>+mncR%)aU7kV00^p146H{5o+6{Ltsu42_s4dft%?AxF3S0Uvlr~uZ05$d!I)E18jj{j4_biwVfSxEa zqvNwLMh=~^3p^r;G+?EYOi{QQEDnpK`6L0>{BZm4naoPA>wbm+wU4>h~okQ)7j<@=liW%MpBGh^<1_h#y`0bG}=Ol;wl!i+h& zj?}RdICQXCT|GnC_GibAKMLmA+qKSOQg@YVXBSHzJA_h85B_u!VFQ=Dc4;r!0FNX2 zqefEd79(Ukx~Mbvz{sy)>;1ucy$z?rFCM<5uQL_k;~74Q zY7xu#^zg)QUhotwWIp$x^;pk$noWU!3@2Zm1!^g{XYN<))_7U literal 0 HcmV?d00001 diff --git a/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-9-Newspaper .jpg b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/images/image-9-Newspaper .jpg new file mode 100644 index 0000000000000000000000000000000000000000..3055a361538b4f7ee5d08d6cfd0bb2ffc4d7e306 GIT binary patch literal 2763 zcmbW&c{tSH9tZI6m@y2Z8D#lc!eoX}V~I*8C2J@O$x=i`lkDp>S+a$(ge;9D%aEBi zOBBTzp_IgAWG9SY7};XRT>XCcx%YYQU-#bkc|PZ#^LfrWujl-6cwc#=z)nk33sV3D z0szqV0C`Yc63IupvtkQZzb(|7Lht00WmP4T~g9AvU^lfXf<_Ry#x9N2M--L zIbmvMZeeME=B$IG(>Z6)OJ3fWuUz#B3FQ>w&%ZL|KWmfyZ9gwFhtvi}Yi^MA?y1^c&)35b9}+m8o^16W}7M?g4kIehAN zrvVx)TB5=4PaIb(xl7w14FDbXXvoc)PwNY3h$j}^N-Dbku#m4eMc=nwknndmtRueY z*pq+ME1^3i;Mwuqw?8s$&fZ5yKD7%~j~6AWR;|{!Ae_%v9m$BY*juH|iq3{QCB*}4 z-*dMRYPIoOy<>Nu7~Pl@Ew2vC<)|_Zm46 zB@2_b7nNF>J+!BING_NEU`4yQ>TuW8OazvL4EQm`t-z$~OpBJ6N1R|Isr8f!(YHCY zRw7leZD@Aqpq2Q2n?Zyib2scTqu5YX^lHuqrZMkA^6EfoR#>Y2;d|31{%y9ax2*;T zXbi4<%TgfMBjgtk$mQt2o3~9w$#j?Y=obwW8|N6MN^FTPq4!LZ%5`vF7&bM$C}}~H zbGWjt%*7hX!0}yA+gDgDPpp&QOHwfCop^m^$m?ia`UeF^6t#=_BCNBc=Ff~Sxe~}{ zf^+V2bG7X?9uSvlyCFo} zz{5AyoI4^HCmd@MqmeD>FC`eqtRhpuy5T1SW>ifr!+ZnO0R2o@95%94y zOJ#k^{zxs%em}D??i*U@BL$a6roMVm2{vgpDugE|R}V?|IG$umriv! z;FuUL0ySRvg;)2ze}DjWnlnNcy~}a>p|h)}r6c7sTX(QvU5HB>y0oNER~cw|IO?eo zn_S=!vrjTih($;3mHR5U7~rq?{ly2iFpoe~A1 zP%;KaeV>Ni=W2%)lSEE;k*!cUrHO!ucoyphdU~xegQ#L*vcv<{?K646$Ce(~eXF_2 z&C}^Tpvj(drTm9_sN%D7PI+;Zt4EQr9>YH7DcyLPJMJj?Fzg}Tv*m!J`p@w0M)|R$ zn>PCcN!iROcY}3%YSG4}q7BU6mYaz4zEk_=UkqTb)w^qXsp=DKJ(HQ0pQp8C9kTOg z^ZPc50z<;3VMo!rTj7H=-xyv5xSL_#$( zdYyWyD}w~>mfcg*ae;K`{`M4iT?TR^+pgUhAr7#q!J zY;hl-E)Sdv?eJLS%M*QVPk|U}jWL7=qy~{-?3`O!-2CzRadaIb1Ly5VV9uJ12*{#h zEnPis#Kk0fC~KsLk8a|_DJZ%dsDJs0`MU*KS|xe%>* z!K;%wi|X(8;9AVuyB|kT#4N|7Y}t-ZTyEy<4W`Qla=JV%GP$kS#|AyZRJ}VHLEs>3 zw~9(7f4Xy!vQ!Qa`0%U8mbEm0c_klDq%@(=S&^gk`>&0)7icR}@KJ4##p3gY*8N)4 zN?jXtyXlt)do{n}F|N%?<3C?v{iW9OF+3n+*OUG!jq3fiA3lNm1&Of50Hrv2Dl4{@4JSLU3b*rI>yWzL^ zuqk2FYmLKx!ULK|xIXIm#y_!DTGPJNcgj6wAv4u7*fl0Y!tX(SwAsB^Q)yykN>+QF z;-y8?awGqzr+C1`oSRUSYHEtV)b4b!$V75j;!v4}>id;QuHXLS!NLidACW2CKX0_M zB}+?KA&LVq|I#eI#)De=lQbKv>&tE7lJSkg1u>sFnJ2%xUt{Q3_&7EknQ)ttBI0z+ zyEkoz72+b!+GRPLpRFjPf|;fTtMb^E-rZ*#6HbYLXskFFD!{d5*pKh)Ct7@83oN-yY(1ams#*VdOj!<@hTLH{!c##f~KK05EoA*)jgK6W$p* z;*z)O-d2a!D)i!se%9mutZ$!h5mr%wRCi@+QB2jiLiving area (feet $^{2}$ )Price (1000$s)21044001600330240036914162323000540⋮⋮ \ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-1 House Price Data.html b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-1 House Price Data.html new file mode 100644 index 000000000..8a393d6f8 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-1 House Price Data.html @@ -0,0 +1,94 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
square feetprice (in $1000)
1000200
1200250
1400300
1600350
1800400
2000450
2200500
2400550
2600600
2800650
3000700
3200650
3400600
3600550
3800500
4000450
4200400
4400350
4600300
4800250
5000200
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-2 Loss Functions.html b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-2 Loss Functions.html new file mode 100644 index 000000000..85df8f661 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-2 Loss Functions.html @@ -0,0 +1,42 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
xSquare LossAbsoluteε - insensitive
1.01.01.00.7
0.50.20.40.2
0.00.00.00.0
0.50.20.40.2
1.01.01.00.7
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-3 Data Table.html b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-3 Data Table.html new file mode 100644 index 000000000..6726fb879 --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-introduction-statistical-learning-theory/tables/table-3 Data Table.html @@ -0,0 +1,34 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
xLine 1Line 2Line 3Line 4
01.01.01.01.0
10.00.00.00.0
21.00.50.20.3
\ No newline at end of file diff --git a/apps/api/app/data/demo_documents/stem-jurafsky-transformers/chunks.json b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/chunks.json new file mode 100644 index 000000000..834ad3e7b --- /dev/null +++ b/apps/api/app/data/demo_documents/stem-jurafsky-transformers/chunks.json @@ -0,0 +1,4362 @@ +{ + "chunks": [ + { + "chunk_id": "babc058d-28ee-535b-9ffe-892d38458e93", + "type": "image", + "content": "\n8